HAL Id: hal-02189035
https://hal.archives-ouvertes.fr/hal-02189035
Submitted on 18 Jul 2019
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of
sci-entific research documents, whether they are
pub-lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Distributed under a Creative Commons Attribution| 4.0 International License
genes shapes plasticity in plant metabolism
Zhenhua Liu, R Tavares, E S Forsythe, F André, R Lugan, G Jonasson, S
Boutet-Mercey, T Tohge, M A Beilstein, Daniele Werck Reichhart, et al.
To cite this version:
Zhenhua Liu, R Tavares, E S Forsythe, F André, R Lugan, et al.. Evolutionary interplay between
sister cytochrome P450 genes shapes plasticity in plant metabolism. Nature Communications, Nature
Publishing Group, 2016, 7. �hal-02189035�
ARTICLE
Received 8 Mar 2016
|
Accepted 26 Aug 2016
|
Published 7 Oct 2016
Evolutionary interplay between sister cytochrome
P450 genes shapes plasticity in plant metabolism
Zhenhua Liu
1,w
, Raquel Tavares
2
, Evan S. Forsythe
3
, Franc¸ois Andre
´
4
, Raphae
¨l Lugan
1,w
, Gabriella Jonasson
4
,
Ste
´phanie Boutet-Mercey
5
, Takayuki Tohge
6
, Mark A. Beilstein
3
, Danie
`le Werck-Reichhart
1,7,8
& Hugues Renault
1,7,8
Expansion of the cytochrome P450 gene family is often proposed to have a critical role in the
evolution of metabolic complexity, in particular in microorganisms, insects and plants.
However, the molecular mechanisms underlying the evolution of this complexity are poorly
understood. Here we describe the evolutionary history of a plant P450 retrogene, which
emerged and underwent fixation in the common ancestor of Brassicales, before undergoing
tandem duplication in the ancestor of Brassicaceae. Duplication leads first to gain of dual
functions in one of the copies. Both sister genes are retained through subsequent speciation
but eventually return to a single copy in two of three diverging lineages. In the lineage in
which both copies are maintained, the ancestral functions are split between paralogs
and a novel function arises in the copy under relaxed selection. Our work illustrates how
retrotransposition and gene duplication can favour the emergence of novel metabolic
functions.
DOI: 10.1038/ncomms13026
OPEN
1Institute of Plant Molecular Biology, CNRS, University of Strasbourg, 12 rue du Ge´ne´ral Zimmer, Strasbourg 67084 France.2Universite´ Lyon 1, CNRS, UMR 5558, Laboratoire de Biome´trie et Biologie E´volutive, 16 rue Raphael Dubois, 69622 Villeurbanne Cedex, France.3School of Plant Sciences, University of Arizona, Tucson, Arizona 85721, USA.4iBiTec-S/SB2SM, UMR 9198 CNRS, University Paris Sud, CEA Saclay, 91191 Gif-sur-Yvette, France.5Institut Jean-Pierre Bourgin, UMR 1318 INRA-AgroParisTech, Saclay Plant Sciences RD10, 78026 Versailles, France.6Max-Planck-Institute of Molecular Plant Physiology, Department of Molecular Physiology, 14476 Potsdam-Golm, Germany.7University of Strasbourg Institute for Advanced Study, 67000 Strasbourg, France.8Freiburg Institute for Advanced Studies, University of Freiburg, 79104 Freiburg, Germany. w Present addresses: Department of Metabolic Biology, John Innes Centre, Norwich Research Park, Norwich NR4 7UH, UK (Z.L.); UMR Qualisud, University of Avignon, 301 rue Baruch de Spinoza, BP21239, 84916 Avignon Cedex 9, France (R.L.). Correspondence and requests for materials should be addressed to D.W.-R. (email: werck@unistra.fr).
E
volution of new protein functions is often initiated by gene
duplication, but most duplicates are discarded by purifying
selection
1–5. The retention of new copies may be explained
by
either
neofunctionalization
or
subfunctionalization
1,6,7,
with several models for subfunctionalization proposed
5,7–11.
Functional innovation may result from an expansion of a
previous biochemical activity or from a change in the
biochemical properties of the protein
12,13. Acquisition of an
additional biological activity can then lead to a pleiotropic
protein. Multiple analyses have shown that the evolutionary
scenario after gene duplication may be more complex, implicating
several of these mechanisms, acting one after another
14–16, as has
been suggested by theory
17. However, experimental evidence of
the models and their interplay is often difficult to obtain. We
describe here the evolutionary history of a plant retrogene
18that
stemmed at an early step of Brassicales evolution.
The parent gene, represented by CYP98A3 in Arabidopsis
thaliana (family Brassicaceae, order Brassicales), controls a major
branch point in the plant phenolic pathway
19,20. It is broadly
expressed in plants, especially in vascular tissues, and encodes the
phenolic ring meta-hydroxylase of lignin precursors. This enzyme
evolved in land plants around 400 million years ago (Ma) to
provide metabolites essential for protection against desiccation
and ultraviolet radiation, water conduction and erect growth
21.
Due to its gate-keeping role and the high flux in the lignin
pathway, CYP98A3 and its orthologs are under strong purifying
selection
18. Only a few lineage-specific duplications are inferred
in phylogenies of CYP98A3 across land plants. One such
duplication occurred via transposition within Brassicales and
was followed by a tandem duplication giving rise to CYP98A8
and
CYP98A9
genes
leading
to
the
formation
of
N
1,N
5-di(hydroxyferuloyl)-N
10-sinapoyl-spermidine,
a
major
pollen coat and wall constituent in A. thaliana
18,22. Available
data indicate that both CYP98A8 and CYP98A9 catalysed the first
meta-hydroxylation of the phenolic ring of pollen constituents
(3
0-hydroxylation of N
1,N
5,N
10-tri-coumaroyl-spermidine) and
that CYP98A8 also catalysed the second (5
0-hydroxylation of
N
1,N
5,N
10-tri-feruloyl-spermidine). The recent emergence of this
pathway and the sequencing of a representative set of Brassicales
genomes offers the opportunity to determine the evolutionary
process leading to these new P450 functions.
We show here that the founding retroposition leading to the
CYP98A8 and CYP98A9 common ancestor occurred before the
divergence of Cleomaceae and was followed by a duplication
before the divergence of the core Brassicaceae lineages. We
demonstrate that the fate and function of the resulting paralogs
differ between the Brassicaceae lineages, with either loss of one of
the paralogs, the other maintaining a dual function in lineage II
and III, or conservation of the two gene copies in lineage I, with
subfunctionalization and gain of an additional activity in
flavonoid metabolism by one of the duplicates.
Results
Evolutionary history of the CYP98 family in Brassicales.
A large set of genomes in Brassicales has now been sequenced,
allowing the reconstruction of CYP98A8 and CYP98A9
evolutionary history (Fig. 1a; Supplementary Fig. 1 and
Supplementary Table 1). The reconstructed CYP98 phylogeny
first shows that a CYP98A8/9 homolog is found in some, but
not all Brassicales. For example, Tarenaya hassleriana, belonging
to Cleomaceae, the sister family that split from Brassicaceae
around 70 Ma
23,24, contains a CYP98A8/9 copy, but the more
early diverging Carica papaya (Caricaceae; 100 MY divergence
with Brassicaceae) does not. This dates the CYP98A8/9 founding
retroposition between 70 and 100 Ma
25. The tandem duplication
generating the CYP98A8 and CYP98A9 paralogs occurred after
the divergence of the genus Aethionema, predating the divergence
of the major lineages I, II and III of core Brassicaceae (Fig. 1a;
ref. 23). The duplication pattern indicates that lineages II and III
subsequently lost CYP98A9, while lineage I retained it. A
CYP98A8 ortholog is present in all available Brassicaceae
genomes, which include lineage I and lineage II species
23, but
CYP98A9 is found only in lineage I (Fig. 1a).
In the absence of sequenced lineage III species, we designed
degenerate primers to amplify CYP98A8/9 homologs from the
lineage III species Chorispora tenella and Euclidium syriacum.
A CYP98A8 orthologous sequence could be retrieved from both
species (Supplementary Fig. 2), but no CYP98A9. However,
BLAST search of the CYP98A8 flanking regions, respectively,
found one and four correctly ordered remnant CYP98A9
fragments
in
C.
tenella
and
E.
syriacum,
respectively
(Supplementary Fig. 2), indicating that CYP98A9 had been
present and was subsequently lost from the genome. The presence
of a fragmented CYP98A9 confirms the duplication history
inferred from the gene tree. Thus, the tandem duplication event
that gave birth to CYP98A9 predates the split between sister
lineages I, II and III, around 45 Ma, and CYP98A9 is retained
only in lineage I.
Activity of the CYP98A8/9 proteins in vivo and in vitro. We
then assessed CYP98 enzyme activity in each lineage using two
complementary approaches. One was the yeast-expression of
representative CYP98A8/A9 genes for in vitro assay of catalytic
activity (Fig. 1b). The second employed metabolic profiling of
flower buds from members of each Brassicaceae lineage and a
representative of the earliest diverging branch in the family (that
is, the genus Aethionema) (Fig. 1c). In vitro assays revealed that:
(1) the CYP98A8/9 protein from Adenium arabicum and
T. hassleriana catalysed only 3
0-hydroxylation of
tri-coumaroyl-spermidine; (2) the radiation of core Brassicaceae was preceded
by acquisition of the 5
0-hydroxylase activity, shown by the
ability of CYP98A8 from Eutrema salsugineum (lineage II) and
A. thaliana (lineage I) to catalyse both 3
0-hydroxylation of
tri-coumaroyl-spermidine and 5
0-hydroxylation of
tri-feruloyl-spermidine; (3) A. thaliana CYP98A9 has 3
0- but not
5
0-hydroxylase activity, similar to A. arabicum and T. hassleriana.
Taken together, our in vitro data suggest that 5
0-hydroxylase
activity evolved in CYP98A8, post duplication.
Phenolic profiling supported the results from the enzyme
assays. The 5
0-hydroxylated phenolamide
di-(hydroxyferuloyl)-sinapoyl-spermidine was absent from the flowers of Aethionema
grandiflorum, but present as at least two different isomers in core
Brassicaceae (Fig. 1c; Supplementary Fig. 3). Hence, we inferred
that the single CYP98A8/9 copy in the ancestor of both
Brassicaceae and Cleomaceae lacked 5
0-hydroxylase activity.
Moreover, the single CYP98A8 copy present in lineages II and
III performed both 3
0- and 5
0-hydroxylations, in vitro and in vivo.
This finding is also supported for lineage I species by the ability of
A. thaliana CYP98A8 to catalyse both reactions in vitro. Both sets
of analyses thus concurred to indicate that the acquisition of
5
0-hydroxylation occurred in the ancestor of core Brassicaceae.
However, whether this acquisition predated, or was subsequent
to, the gene duplication that gave rise to CYP98A9 remained an
open question.
To address this question we took advantage of a recently
released A. thaliana cyp98a9 insertion mutant (Supplementary
Fig. 4) and compared its young flower bud phenolic profile to
those of the previously characterized cyp98a8 insertion line and
wild-type plants. Ultra-performance liquid
chromatography-tandem mass spectrometry (UPLC-MS/MS) profiles of the flower
buds revealed a dramatic decrease in the major phenolamide
compound N
1,N
5-di-(hydroxyferuloyl)-N
10-sinapoyl-spermidine
in both the cyp98a8 and cyp98a9 single mutants, with a
concomitant accumulation of N
1,N
5,N
10-tri-feruloyl-spermidine
in cyp98a8. The latter compound was completely absent in
cyp98a9 buds (Fig. 2). This confirmed that subfunctionalization
of CYP98A8 and CYP98A9 in the phenolamide pathway
occurred in lineage I. The complete partitioning of functions
in vivo, when 3
0-hydroxylase activity is shared by both enzymes
tested in vitro, implied a modification in protein–protein
interaction resulting in spatial reorganization of partner enzymes
in the tapetal cells. We then hypothesized that the split of
function we detected in lineage I most likely contributed to the
retention of CYP98A9.
Molecular evolution of the CYP98 duplicates in Brassicales. To
gain some insight into the molecular evolutionary mechanism
that led to functional shifts in CYP98A8 and CYP98A9,
we compared rates of non-synonymous to synonymous codon
substitutions (omega, o ¼ d
N/d
S) along several branches of the
Brassicales CYP98 phylogeny. Both the branch model and the
clade model of Codeml analyses indicated that all CYP98
sequences were under purifying selection (all oo0.4, and even
0.2 for the branch model) and that the lineage I CYP98A8
sequences were significantly more constrained than those of
CYP98A9 and CYP98A8 from lineages II and II (Fig. 1a;
Supplementary Note 1; Supplementary Table 2). Interestingly,
when applying the free-ratio model of Codeml (independent o
ratio for each branch), we detected a relaxation of selection
occurring along the branch leading to the CYP98A8 clade
(o ¼ 0.63; Supplementary Fig. 5), which could reflect the
acquisition of the 5
0-hydroxylase activity in CYP98A8. With a
single exception, o values were below 1; o for lineage I CYP98A3
and CYP98A8 sequences was lower than for CYP98A8s from
lineages II and III. The highest o was associated to the basal
internal branch of the CYP98A9 clade that showed evidence
of relaxed selection (o ¼ 1.01; Supplementary Fig. 5), which
suggested the possibility of a change of function emerging
post-duplication in CYP98A9s. A McDonald–Kreitman test
comparing within-population polymorphism with
between-species divergence
26for A. thaliana and L. alabamica CYP98A9s
showed a significant signature of purifying selection acting on
A. thaliana CYP98A9 proteins (Supplementary Table 3). All
models, including Codeml branch-site and Fitmodel analysis
(Supplementary Note 1; Supplementary Table 2), failed to detect a
significant signature of positive selection along any branch of the
tree and/or sequence sites.
CYP98A8 and CYP98A9 wire to different regulatory networks.
To examine the possibility of a change of function emerging in
0.1 substitutions/site Shrenkiella parvula A3 Brassica rapa A8 Capsella grandiflora A3 Shrenkiella parvula A8 Boechera stricta A9 Capsella rubella A8 Boechera stricta A8 Eutrema salsugineum A8 Capsella rubella A3 Arabidopsis thaliana A3 Arabidopsis thaliana A8
Tarenaya hassleriana A8/9
Brassica rapa A3–2 Capsella rubella A9 Capsella grandiflora A8 Chorispora tenella A8 Leavenworthia alabamica A8 Boechera stricta A3 Sisymbrium irio A8 Euclidium syriacum A8 Aethionema arabicum A3 Brassica rapa A3–1 Arabidopsis lyrata A3 Arabidopsis lyrata A9 Leavenworthia alabamica A9 Capsella grandiflora A9 Arabidopsis lyrata A8 Tarenaya hassleriana A3 Arabidopsis thaliana A9
Brassica rapa A3–3 Aethionema arabicum A8/9
Sinapis alba A8 Eutrema salsugineum A3 100 92 100 100 100 100 93 100 100 100 100 100 100 100 100 Lineage I Lineage II Lineage III Retroposition Loss of CYP98A9 in lineages II and III
= 0.22 Tandem duplication Retention time Relative absorbance at 320 nm AtCYP98A9 AtCYP98A9 AtCYP98A8 AtCYP98A8 EsCYP98A8 EsCYP98A8 AaCYP98A8/9 AaCYP98A8/94 ThCYP98A8/9 ThCYP98A8/9 Tri-coumaroyl-spermidine Tri-feruloyl-spermidine – NADPH + NADPH CYP98A3 clade
3′-hydroxylation activity 5′-hydroxylation activity
100 100 70 93 98 57 78 73 98 98 38 35 47 74 = 0.11 = 0.20 = 0.24 = 0.08 100 100 Arabidopsis thaliana (lineage I) Eutrema salsugineum (lineage II) Bunias orientalis (lineage III) Aethionema grandiflorum Parents of 207 ES+ m/z = 736 2.85e5
Relative detetctor response
4 5 6
Time (min) Core Brassicaceae
a
b
c
Figure 1 | Evolutionary history of the CYP98A8/9 clade in Brassicales/Brassicaceae. (a) Annotated protein maximum likelihood tree illustrating the phylogenetic relationships in the CYP98 family in Brassicales. The CYP98A3 clade is the parent to CYP98A8/9s. Evolution regimes (o¼ dN/dS) computed
with codeml Branch model and acting on different phylogenetic groups are indicated. Branch support in the tree was calculated from 1,000 bootstraps replicates. (b) In vitro assay of CYP98A8/A9 enzymes activities. Tri-coumaroyl-spermidine and tri-feruloyl-spermidine were used to assess the 30- and
50-hydroxylase activities of yeast-expressed proteins. CYP98A8/A9 from all species have 30-hydroxylase activity. Only CYP98A8 from Brassicaceae have
50-hydroxylase activity. (c) Search for sinapoylated (m/z 207) phenolamides by parent ion scan analysis of flower buds demonstrates the absence in vivo of
the tri-hydroxylated phenolamide di-(hydroxyferuloyl)-sinapoyl-spermidine (m/z 736) in A. grandiflorum, while it is present in Brassicaceae from lineages I, II and III. Colour of each profile is according to the lineage colour code ina.
CYP98A9 after duplication, we first analysed in more detail the
CYP98A9 expression pattern in A. thaliana. CYP98A9 was
pre-viously shown to share CYP98A8 expression during pollen
development, but its promoter was also active in root tips and
revealed weak expression in vascular tissues
18. Functional
divergence
following
the
tandem
duplication
is
further
suggested by a co-expression analysis indicating a strong
correlation of CYP98A8 expression with a major network of
genes involved in flower development, whereas CYP98A9
expands to an additional network related to seeds (Fig. 3a–c;
Supplementary Table 4). This observation was confirmed by
b-glucuronidase (GUS) staining, which showed CYP98A9
promoter activity in mature, green seed endosperm (Fig. 3d),
consistent with a recent microarray analysis
27. In sum, both our
own and published expression data indicate that the role of
CYP98A9 reached beyond pollen metabolism.
N H N H N H N H N H N H N H N H N H N H N O O O OH OH HO N O O O OH OH HO HO OH OH N O O O OH OH HO MeO OMe OMe N O O O OH OH HO MeO OMe OMe OH OH OH N O O O OH OH HO MeO OMe OMe OH OH OMe 90 120 0 20,000 25,000 0 150 300 450 600 0 2,500 5,000 7,500 10,000 0 5 10 Col-0 cyp98a8 Col-4 cyp98a9 CYP98A9 AtSM1 CYP98A8 AtSM1 400 800 Col-0
cyp98a8 Col-4 cyp98a9
Col-0
cyp98a8 Col-4 cyp98a9
Col-0
cyp98a8 Col-4 cyp98a9
Relative abundance Relative abundance Relative abundance Relative abundance N1,N5,N10-tri-coumaroyl-spermidine N1,N5,N10-tri-caffeoyl-spermidine N1,N5,N10-tri-feruloyl-spermidine N1,N5,N10-tri-(hydroxyferuloyl)-spermidine N1,N5-di-(hydroxyferuloyl)-N10-sinapoyl-spermidine
a
b
c
d
e
Figure 2 | Functional specialization of CYP98A8 and CYP98A9 in vivo. (a) The anther-specific phenolamide pathway as previously delineated in A. thaliana. (b–e) Targeted metabolic profiling in A. thaliana flower buds shows that phenolamide intermediates downstream of N1,N5,N10 -tri-coumaroyl-spermidine are absent from the cyp98a9 mutant, indicating that CYP98A8 does not perform the 30-hydroxylation in vivo: (b) N1,N5,N10 -tri-coumaroyl-spermidine, (c) N1,N5,N10-tri-feruloyl-spermidine, (d) N1,N5,N10-tri-(hydroxyferuloyl)-spermidine, (e) N1,N5-di-(hydroxyferuloyl)-N10-sinapoyl-spermidine. Results are the mean±s.e. of three independent determinations. Reversible activity of the spermidine:hydroxycinnamoyl transferase acting upstream of CYP98A9 presumably accounts for the lower accumulation of the N1,N5,N10-tri-coumaroyl-spermidine in the cyp98a9 mutant compared with the accumulation of N1,N5,N10-tri-feruloyl-spermidine in cyp98a8.
CYP98A9 gained an additional activity in flavonoid metabolism.
To reveal potential additional functions and overlooked activities
of CYP98A9, we used CYP98A9 loss and gain of function lines.
The latter were generated using the promoter of the upstream
P450 gene in the phenolic pathway, encoding the cinnamic acid
4-hydroxylase (C4H)
28that provides precursors for the lignin,
phenolamide and flavonoid pathways. A similar CYP98A8
overexpression construct was used as control. We observed no
differences in development, seed set or colour, and fertility
between wild-type and cyp98a9 null mutants grown under
controlled conditions. No obvious alteration in plant growth,
seed
set
and
phenotype
was
likewise
detected
for
pC4H:AtCYP98A8. Conversely, seeds from
pC4H:AtCYP98A9-transformed plants showed a typical transparent testa (tt)
phenotype (yellow instead of brown seeds) (Fig. 4a), typical of
seed coats deficient in flavonoids and tannins
29,30. Vanillin seed
staining of epicatechins and proanthocyanidins
31confirmed that
only CYP98A9 expression led to a seed coat tannins defect
(Supplementary Fig. 6a). Moreover, UPLC-MS/MS profiling of
soluble
flavonoids
indicated
that
AtCYP98A9
expression
resulted in a complete depletion in quercetin-3-O-rhamnoside
(Supplementary Table 5), the major soluble seed flavonol in
A. thaliana
32. CYP98A9 expression thus impacts soluble
flavonoid
metabolism
and
not
just
condensed
tannin
composition
or
polymerization.
However,
no
significant
alteration in whole seed, flower buds or developing siliques
flavonoids was detected in the insertion mutant cyp98a9 (Fig. 4a;
Supplementary Figs 6b and 7), but time- or cell-specific alteration
in flavonoid content in this mutant cannot be excluded.
To determine if this flavonoid-impacting function was specific
to A. thaliana CYP98A9, overexpression constructs of another
lineage I CYP98A9 (CrCYP98A9, Capsella rubella), and of
representative members of lineage II CYP98A8 (EsCYP98A8,
E. salsugineum) and of the ancestral AaCYP98A8/9 (A. arabicum)
were tested for their impact on seeds. Only expression of
C. rubella CYP98A9 prevented seed flavonoid accumulation
(Fig. 4a; Supplementary Fig. 6b). The capacity to impair flavonoid
accumulation was thus likely acquired by CYP98A9 before the
3220 3220220222022202222222222200000 3 3 3 3 3 at3g3g33gg at3 at 32323233220323333322220222222220 at3t3ggggggg13g131333333333333333333333333330000 at3g45480g45g45 2320 23 0 232 2320000 at5 aggg a at at at5g at5g at5gt5g a a a at at at a a a at5g at55gg a a at at a agggggggggg62gg62gg622222 80 2888088808088888888880 28 2 28 2888 28 28888888888 at1g at1 at at a a at1 at1 a att11111ggggggggggggg at1t11111111111ggggggggggggggggggggggggggggggggggggggggggggggggggg0gg0gggggggggggggggggg0g0100010010001112 3770 377 37 3770 3770 37 3 377 at3g3g33gggggg232333333333 7990 7990999999 g g g g g at1t111111gggggg676767677779907990799077999 980 980 98 98080 98080000000 at3g at3g at3g33gggggg1111111 aatatataaaaaaaaataaaatatat3t3gt3gt3gt3gt3gtttt3tt3gtttttttttttt3g33ggggggggggggggggggg52g2228102811000 4815 48815 48151 4815555 48 4881588818181815 t4g t4g t4g t4gggg at a a at4ttt at at at at4t at atgggggggggggg1g1g111144444444 18088080 1 18888 1 2181 2 211 2 21 2 2 2 2 2g 2g 2g 2g t2g2 a a a a a a a a a a a a a a at2 at at a a a a a a a a a a a a a2 at2tt2t2gt2t222g22222g att a a a a a a a a att a a a a a a a a at a a at a a atg4gggggg4g4g4gggggg424244442222222222222222222222211118aaaaaaaa 5000 50 50 500 32 32 32 32 1ggggggg 1g 1g at1 at1 att1t1 at a1t1t11111 attttt at1tt1 000000000 a a a a a at a a a a at at1g111g1111g1g111ggggg2ggggggggggggggggggggg2gggg223232323333325252500000000000 30 30 30 3 3 300000000 30 3 843 84343444343333 8 8 843 8444 84 84 84 8444 84444444 843434343444 84 844 1g 1g 1g at at at at a a a at111111 a a a ag28ggg28g28g2888888888888888843434333333333033330000000000 0 0 0 0 0 0 0 7 7 7 7 7 357575 35555555555555555 355555555555555555 3 3575757575557077000 3575 35757 3 35 35 35555555757555 355555557555555777070777707770000000000000 35755 35575555555555557555555555555555 3 3 35 35 355570777707700000000 g g g g g g g 1 t11 t111 a a a at1t1t1tttttt11111gg at a a att1t1g1 a a a at1tt1tttttt111111gg at attgg2gggggg2gggggg2gg2gg2232323222323232222333333333333333333333333333 3240 3 g 1g at a at1 at1t11111111g1g1g1g11ggg a a a at1 at1g1g11 a ag23gggggggggg2333 8375 8375 t1g1gg t1 at at1 at1 at 500 50 50 50 50505005005000050005555555 at3 at at at3tt at at at at3t3tt3tt3tt3t3t3tt3t3tt33333333333333333333333333333333333333333333g33g3g3333333333333333333333333333333333g3g3g3g3g3g3g3g3g333333gggggggggggg a3ggggggggggggggggggggggggggggggggggggggggggggggggg2gg2gggg2g2g2g2g2g2ggggggggg222525225222525252525222222255 at1g06990 at1g at1g at1g at1t g1gggg at11g111g1g1g1g1g1g at1t1g1g1g1g11g1ggggggggggggggggggggggggggggggggggg06g06ggggggggggggggggggg 9780 978 at1gt1g11g at1g1gg79gggg7999 6170 617 6 at1gg06g666 no_ no n n no n n no__match528_m ch_match528____mmmmch528 605 605 at5gt5g at5gt5gt55g5g5g55g5ggggggggggg6g61g6111 7330 7330 7330 73 7330330 at4g4g at4g27gg77777 0 0 0 43030 4300 4 4 4330000 43 43000 430 43300000 430 430303000000 34 34 34 3 344 3 34 34 34 34 34 34 34 34 3444 3444444430000000000000 at1gt1 ag at1gtg3gggggggg33gg33gg3g3333333333333333333 80656 8 65 at1g at1g ag08gg08080888 90700700000 9070070000 9070077070 90 900 900 907007000 9070 at2g a at2 a2g2g2gggggggg1919199999999999 at2g03850g03g03 625050 625 62 62 6225555505555555 6250 62250252522 622 625 62 622 62 622 622 62 6255555000000000000 625025 62 62525 6505555550505000000 at1gg a ag0g06gg0ggg0g060060060006000066666666666666666666666666666aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa 610000 at3gt3g3gg06gg06gggg at4g28090g28g28 100 1 10 1 10000 82 8 8 8222 8 82 82222222222 82 82 82122222122222222122212221212222121222222222221111111111111111111111110000000000000000000 t5g5ggg t5555g5g5g55g5 t5g5g555g555g555g55 t5g5g t5g5 t55 at5t5 at5 at at5 a at a a a a a at a a a5 attt5 a a a a a a a at5t55555 at5t5t5t5555 a a a a a a a a a a a a a agggggggggggggggggggg4gg4g4gg4ggggggggggggggggggggggggggg4gggg4g4g4g4448444848484844484848444848448448484444444444448444444444444444444448888888888888888888 710 71 71111 711 7 71 71 711011110 710 711111 7 7 7 711111 7 7 7 67710 67 67 67710000 at1gt1t1 a1g11g1 att1gttttttg a a a a aggg at1t1 a1111111111ggggg a a a at1t1 at1t at1t1 at att1g1g11111g1ggggg a a a a a at1g a a ag at a a a at1g26780g26g26 070 070 0 07 0 0 at1gg61g11111111 at1g20150 at1g20g20 4550 4550 455 45 4 455 4550 at1gg74g744444444 50 5 50 50 50 5 5 5 50 5 5 5 5 5 5 5 5 5 5 5 5 5 5 311 31 311 31111111111111111111111111 31111111 3 3 31 3111 g g at1 at1 at11 at at1 at1 a111111 a a agggg a a aggggggggggggg a1gggggggggggggggggggggggggggggggggggggggggggggggggg13gggggggg1gg13g13ggggggg1gggg1gggg1g1gg1311133333333333 8395 83555 839 8 8395395 83 839539 8339 833 83 8395395 83939959 839 8339555 83 5 8395 8395 at4g a at gg at4 at4g28ggggg28282888888888888888888 2960 296000 29 29969696969699696999699999699996666660606066606666000 29 2960 2960 2960999999966666666600 29 299 29 2 2 2999999 29 29 2 2 2 299 at3g3 at at3 a3 at3gt3g at3t3 at3g3g3g3gg4ggggggg4g42424242222222222222222222222222222222222222222222222222222 9070 9070 at5gg49g99 0000000 000000 9500 95 950 95 9000 at1gt1g1g1gg69g69gggg999999 0 0 0 0 350 350500 3500000 3 3 03 03 03 033 0 03 033 0333333 g g at1gt1g1g1g a at g1g1gggggggggggg30g30g330303000000000000000 at3g63100g63g63 at2g18420g18g18 4850 4 485 48 485 at4g at4g34g44444 9250 925 9250 at4gg29g999 800 800 800 8 38 3 380 3880 38 38 380 380 38 38008 38 38 380 3 3800 3 3 3 g g at2g a at2gt2ggggggg23g232333333333333333333333333333333 at1g56360 at1gg56g56g 0 0 0 0 0 0 0 0 0 0 0 0 0 0 20 20000 20000 2 2 2 2 2000000000 200 2 92 92 92 92 92222 9 92 9 92 92 92 9 9 9 92 92 9 9 9 599 59 59 59 599 5 5 5 5 5 5 5 5 5 5 5 5 5 59999999999999999992999999999999999999992222222222222222022202222222222200000 5 5 5 5 5 5 g at a a at1t1g at1 at1g at1gt1tt1gt1t1t1t111111111111111 a attttttt1ttt1t11111 at1 a at1 a at1gggggg5555555555555555555555555 at5g07230t55g77727230233030 at5t555 a5g at5gt55g 8505 85 85550 85 850505555 85 85 8 8 850550 85 85 85 8 8 8 85 8505505555500 8 68 68 68 68 6 68 688 68 68 688580000 6 6 6 6 6 68888888888888888888888 at1g1g at1 at1g at at1t11gg at1tt attt at a a a a a a at1g1111 at1gt1ttt11g1g1gggg attt1g1g11111 at a a a a a a a a a a a a a a a a a a a a a a a a a a a aggggggggggggggggggggggggggggg6g6g6g6gg6gg6gg6g6g6g6ggg66666666666666666666666666666666666666666666666666666666 6150 6 at1gg36g66 0 0 10000 511000000 751 751 75 751 751 7515151011000000000 at5g at5g at55 at5ggg07ggggg07077777000000000000000 75205252202020000 at5g5555g555 at a at at5 a at at a55 at attt55555555g555555g55555g555g555g5555gg at at at at at a at at attttt5t5t5t5t555g55533gggggggggggggggggggggggggg0707007077777 75555555555 7 75535533333 7535555555535535333 755535553333333333gggggg0000000000 at5 at5t5t55 a a at att5t5t5t555 a a a att at attt5tttttttttt5t555g5555555555555555555g555555g55g5555ggggggggggg a a a a a at a att at at a at a a a a a att at at a a a a a a at att555555g55g5g5555g5g55555ggg at5 at5t5t a a a a a a 5 a atttt at attt at atgggggggggggggggggggggggggggggggggggggggggggggggggggggg0ggggggggggggggggggggggg000 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 75404040404444 754040440 7540400 75 754 75 754 g at5g a at5 at5 075 a 55gg7575000000000000 at5 a at5g at5gggggggggggggg0ggg0g07777777777 5 at55 at5t5t55 att5tt5555555g555555555g55555555gggggggg7757577577757757757560560565656565605606060000 att5tt5555g55555ggg attttt5tt5t5ttt555555555555g07555555gggg0g0g0gg07ggggggggggg07g0gggggggggggggggg07ggg070707000707007777775757577777777755555555555 a a at1t1t1111 at1t1 at1 at1 at at11 at1 at1 at4g30040g30g30 at1g235801gg23g23g35835803333330 at1g23600 at1g at1g1g1gg23g23gggg 20 352 g at attt1gtt1gt1g 352t11g1g1g23521g111gggggggg 3g23gggggg2g2223233333333333333333333523333352353352352522202202202202222222202020 at1g23590g at1g at1g1gg at1gt1gt1gg23g23gggggggg 734034 at1g ag07ggg07gg 67070 67700 36 36 366 g g6767 a at1t1 a att ag at1ttt111g363666 a a a a 67677 att a a a agg3633 at1t1tg a a atttt a11g at1t1 att a at a att1ttt1t111gggggggggggggg 3ggg23ggggggg23gg33333333 t3g51590 at3 at at3g3 att att3g 59050 at3gt3 a 3gg5g51ggg5151511159055900 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0020 0020 at1g at1g1g0200 at1g30020020 at1t1 at1gg a1g1ggg3gg30gg0000000000000000002000000000200020 at5g62080g62g622222222222220802080208202082080208020208020208000000 at1g15460 at1g at1g at1gg15g15ggg 480 48 at3gg at3gg45g4ggg5454 87575 8 875 875 87 875 87 87 87 87 87 8 8 8 at1g at1 a ag68g68888888888888888888888 280808 at1g ag01g0gg282282802822880 160 160 160 1 160 160 1 160600 1 1 at3ggg at3g at3g3g at3g a3gggggggggggg52gg52g52212212122121212121 at3gat3gatag23ggg237g22gg237707707070770 7404400 at2gt2g at2g2g2g2g2gg22gg at2g2gggggggg037g030333737377474077474 at1g23250 at11t1gggggg222505000000 at1111tt1111 430 4 430 4 4 4 4 at1gg at1g att1 ag28ggggg28888448484884 570 5 5 5 5 5 5 5 at1g at1g at1g at1g at1g a atttt attggggggggggg23gg23gg2335233535353535353353335aaaaataatatatatatatat1gatata 1at1gaaat1gat1g1111111t11t111111111g1g11g1111111gg 240 240 240 2 2 240 20 240 24 2 2 2 at1gg2332g323323232323332323233332323322 990 990900 99 9 9 9 990 999000000000000000000000 99 99 9 9 99 9 at1g at1g at1tg at1t1g06g06g00ggggg69696969699699 at5g52160 atg52g521 910 9 at2gt2t2gt2g16692ggggggg9 940 94 9 94 at3gg2669g696969 780 7 780 780 7 780 780 780 7 at1ggg at1gg79gggg7997997997979797979777000 at1g06170 at1gt1gg06gg061ggg at4g10850gg10g108g h528 h528 h5 no_matcmatcccc 4300 43 4 at1ggg at1ggg at att11111 at1g at1 a at ag3334gggggggg34 at5g53510g53g535 at2g19070ggg19g190gg 260 2600 at1g at1 at1g1gg at1g11111gg1ggggggggg06gggggggg6262 250 2 2 25 2 2505 250 25 2505 at1g at1gg0662gg62662626262 160 1 160 g at1g at1g a g1gg71gggggggg1111 600 6 6 600 6 600 600 at1g at1gg23g23gg36336363636363636 at3g23840g23g238 140 140 140 140 1 1 1 1 140444 at1gg at1gt1t1g13ggggggg3133333311111 070 070 070 07 at1gg610g000 540 540 at1ggg74gg4545 at3g42850g42g428 810 81 8 81 81 810 81 at3gg52g28282828282828atattt 960 960 at3g at3gg42gg2929 700 700 7 at4gg28g8787 at5g60500;at5g60510600t5510 a ag6050005555555t55g5t5t5t5g600 a ag605gggg606060 at gggg 420 at2gg18g8484 at1g08065g08g080 at5g07230g at5g at5gt5gg07g072gggg 5900 59 at5g at5g55g555 530 53 530 530 5 530 5 0 at5gg at5g at at5g5gg at5gg at5g at5gt5g at5gg07ggggggggggggg07g07gg070707775757575757575 605 60505 605 605 605 60 605 605 6 6 6 6 6 at5gg616g66666666666 58080 5 5 58 58 58080 5 58 58080 5880 5 5 58080 58 58080 5 5 5 5 5 58080 58 58080 5 58 588 5 58 5 5 5 5 5 5 5 5 at1g1gggggg235ggg23g335535353533335335533335335353353335335555 520 5 520 520 520 at1g a at1g a a at at a a at1g1g at1 a a a a att1 a a a a a a at1t11g a a a a1g1ggg at1g1g a a a a a a a a at1gt1g a a a a a a11gggggggggggggggg2335g23gg353535355 pCYP98A8:uidA pCYP98A9:uidA CYP98A8 CYP98A9Flower Mature green seed Flower Mature green seed
Cell culture/primary cell Sperm cell Protoplast Root protoplast Columella protoplast Lateral root cap protoplast Root epidermis and lateral root cap protoplast Root cortex protoplast Root endodermis and quiescent center protoplast Root stele protoplast Root phloem protoplast Root protophloem and metaphloem protoplast Root phloem companion cell protoplast Root xylem protoplast Root phloem pole pericycle protoplast Root xylem pole pericycle protoplast Root cortex, endodermis and quiescent center protoplast Root epidermal atrichoblast protoplast Root hair cell protoplast Lateral root primordium protoplast Leaf protoplast Guard cell protoplast Mesophyll cell protoplast Root cell Root cortex cell Root epid. atrichoblast and phloem companion cell Root endodermis and quiescent center cell Root stele cell Root vascular tissue cell Root phloem companion cell Root culture Seedling culture Giant cell Shoot cell Cotyledon and leaf pavement cell Trichome and leaf petiole epidermis cell Cotyledon and leaf guard cell Shoot phloem companion cell Shoot vascular tissue and bundle sheath cell Rosette cell Internode cell Starch sheath (endodermis) cell Interfascicular cambium cell Guard cell Seedling Cotyledon Shoot apex Shoot apical meristem Hypocotyl Radicle Inflorescence Flower Stamen Pollen Abscission zone Pistil Carpel Stigma Ovary Ovule Petal Sepal Pedicel Silique Seed Embryo Suspensor Endosperm Micropylar endosperm Peripheral endosperm Chalazal endosperm Testa General seed coat Chalazal seed coat Replum Shoot Inflorescence stem Node Phloem Developing meristemoid zone Rosette Stem Leaf Petiole Juvenile leaf Adult leaf Senescent leaf Axillary bud Cauline leaf Shoot apex Shoot apical meristem Leaf primordia Axillary shoot Hypocotyl Xylem Cork Roots Primary root Root tip Root apical meristem Elongation zone Maturation zone Pericycle Lateral root Callus
CYP98A8 co-expression network CYP98A9 co-expression network
Cluster A Cluster B 900 90 9 900000 9 9 9 9 90 9000 9 9 9 900 90 9 9 9 900000 at3gt3gt3g33gggggggggggggggg08g08888989888998989999999 740 740 74 740 740 at3g a at3g at3g3g at3gg a 3g at3g a a a ag58g58gggggg58585887878787 at3g29300303000 at3g 240 24 2 at3gg46g62626262 970 970 90 at3g3gg3gg at3ggg52g52ggggggg29292292 220 2 220 at5gg at5g at5g at5 at5ttt5555 at5t5t55ggg10g102gggggg0202 250 25 250 2 2 2 at1g at1gg at1gg at1g at1g a ag71g712ggggggg2222 at5g579205 at5ggg57g579ggg 955 9 955 955 at1gg54g54gg4949494499 860 860 8 at1gg6998g9898 at5g27200 at5gg27g272g t5g03820 t5 0 t5g0 t at5g0381 at5g03 at5g03815g0381g0g038g0g00338181100000;a;aatatatat at4g10490 at4gg10g104g at4g10640g10g10606640400404040 at4g28365;at4g32490836
at4g28363665;at6565655;a
650 650 6 60 650 60 g g g g g 1 1 1 1 1 at at att1t1t1t1t11g1gggggggg50gggg066060606 at1g64800g64g648 at1g14750g14g147 at4g33330g33g333 180 1 1 18 1 180 1 1 18 1 1 18 18080 180 1 18808 1 3g 3g at3
at3tt3g11t3g33g11ggg at3gat3gaaat3gt3gg27t3g3ggggggggggg77777857 at5g59590 at5gg59g595g 050 05 0 050 0 050 at3gg1990g9090909090 00 00 600 60 600 6 60 6 at4g at4ggg3336ggg3636363636600000 at5g13600 at5g at5ggg13g136ggg 550 at1g at1gt1g at1ggg74gggggggg45 180 180 18080 at3gg18g81811 h481 h481 h481 h4 h h h no_matcmatcccccc at5g07260 at5g at5g5g5g5gggggggg072gg07g0g07gg0 510 5 510 at5g at5g at5g5g5g at5g at5g at5gg07gggggggggg757575 380 380 at1ggg04ggg4343 at5g12460446046046600000000 at5g38180t5g at5gt5g55gg38g381ggggggg 590 5 at3gg02g2525 900 9 90 90 9 9 9 at5 a555g at5g5g at5g55g55g5ggggggg99999999atatatat Cluster A Cluster B Genes Anatomical parts 0 100% Expression potential
a
c
b
d
Figure 3 | Expansion of CYP98A9 expression network in A. thaliana. (a) AtCYP98A8 and (b) AtCYP98A9 co-expression networks. AtCYP98A9 is associated with two distinct co-expression clusters, labelled A and B. (c) Hierarchical clustering of genes involved in the different co-expression networks. Each element of the matrix represents the expression potential of a given gene (column) in a defined anatomic compartment (line). Blue arrowheads indicate flower- and male reproductive tissue-related samples. Red arrowheads indicate seed-related samples. (d) Analysis of pCYP98A8:uidA and pCYP98A9:uidA expression patterns in opening flowers and green seeds.
radiation of lineage I. We then set out to identify the flavonoid
substrate of CYP98A9s likely to explain the transparent testa
phenotype. In vitro enzymatic screens of flavonoid candidates
spotted the AtCYP98A9-dependent conversion of naringenin into
eriodictyol (that is, 3
0-hydroxy-naringenin), a reaction very poorly
catalysed by AtCYP98A8, EsCYP98A8 and AaCYP98A8/9
(Fig. 4b,c). AtCYP98A8 and AtCYP98A9 homology modelling
and docking experiments in addition confirmed that only
AtCYP98A9 efficiently anchored S-naringenin in its active site,
with B-ring oriented for 3
0-hydroxylation (Fig. 5a; Supplementary
Fig. 8). From the three residues critical for naringenin
stabilization, only Lys
209was found discriminant throughout
the CYP98A8s/CYP98A9s alignment (Supplementary Fig. 9). The
involvement of Lys
209in AtCYP98A9 naringenin hydroxylase
activity was supported by Lys
209substitution with uncharged
amino acids (Ala and Ile) leading to up to 50% selective decrease
in flavonoid hydroxylase activity (Fig. 5b). According to Fitmodel
results (MX þ S1 model, Fig. 5c; Supplementary Note 1;
Supplementary Table 2) used to estimate different d
N/d
Sratios
to each sequence site and branch on the phylogenetic tree
(see Methods), this site was constrained in CYP98A9 sequences
(o ¼ 0.038—codon AAG in the six species analysed;
Supple-mentary Fig. 9), but evolved neutrally (o ¼ 1.000) at the basal
branch of the A9 clade and in most of the other branches of the
A8A9 clade (Fig. 5c). Lys
209and the surrounding residues that
also vary between CYP98A8 and CYP98A9 might thus be
instrumental in the further evolution of CYP98A9 function.
It is possible that the novel CYP98A9 function is still latent, yet
flavonoids are regarded as important determinants of plant fitness
as ultraviolet protectors, antimicrobials and antifeedants
33. They
are reported to impact seed germination and growth
34. It thus can
also be postulated that the new function of CYP98A9 contributed
to its retention in lineage I.
Discussion
In summary, our work provides an unprecedented example of
multistep and divergent evolution of a retrogene expressed in
male reproductive tissues, where successive duplication, along
with acquisition and partitioning of a dual function, led to
progressive, lineage-specific expansion of phenolic metabolism,
and emergence of new and diverse metabolic capacities in the
plant. Our experimental data are fully supported by molecular
evolutionary results and highlight the rise of an extended gene
activity simultaneously rewiring gene expression networks and
leading to an enzyme with pleiotropic activity. Metabolic profiling
of plant tissues and ectopic gene expression provide a practical
demonstration that in vitro enzyme activity and gene expression
in a single context are not sufficient to describe gene function.
Our data, in addition, reveal a striking case of divergent evolution
of paralogs in sister lineages (Fig. 6), illustrating how relaxation of
purifying selection following gene duplication can allow for the
exploration of novel metabolic pathways.
Methods
Plant materials and growth conditions
.
Unless otherwise stated, plants were grown on soil under a 16h/8h (day/night) regime with 70% humidity. The cyp98a8 insertion mutant (SALK_131366) was isolated and characterized previously18. The cyp98a9 insertion mutant (#SK27279) was obtained from the Saskatoon transfer DNA collection35. The correct transfer DNA insertion in theCYP98A9 gene was confirmed by PCR-based genotyping and further validated by the drastic decrease in CYP98A9 expression observed in the mutant by quantitative reverse transcription PCR (qRT–PCR) (Supplementary Fig. 4). Primers used for the mutant validation are listed in Supplementary Table 6. For parent ion scan analysis, flower buds from A. thaliana and E. salsugineum grown under controlled conditions were harvested. Flower buds from T. hassleriana were a kind gift from Pr. Eric Schranz (Wageningen University). Bunias orientalis and A. grandiflorum were sampled in the botanical garden of the University of Strasbourg.
Cloning procedures and production of transgenic lines
.
The intronless coding sequences of AtCYP98A8, AtCYP98A9, CrCYP98A9, EsCYP98A8 and AaCYP98A8/9 were PCR-amplified from genomic DNA using Gateway compatible primers and introduced into pDONR207 vector by BP reaction to generate pENTRY clones. Expression constructs pC4H:AtCYP98A8, pC4H:AtCYP98A9, pC4H:CrCYP98A9, pC4H:EsCYP98A8 and pC4H:AaCYP98A8/9 were obtained by recombining the corresponding pENTRY with the pCC0996 Gateway destination vector that contains a 3 kb promoter fragment of A. thaliana Cinnamate 4-Hydroxylase (pC4H)36using LR clonase (Invitrogen). Primers used for molecularcloning are listed in Supplementary Table 6. Agrobacterium tumefaciens GV3101 carrying the expression vector was used to transform Arabidopsis wild-type plants (Ws background) by dipping immature inflorescences in a bacterial
suspension37. Transgenic plants were selected on soil via Basta treatment (250 mg l 1; Agro Evo). T2 seeds were used for subsequent analyses.
Amplification of CYP98A region from lineage III species
.
To obtain CYP98A8 and CYP98A9 sequence data from species in Brassicaceae lineage III, we performed Ler tt4-1 Ws pC4H: AtCYP98A9 pC4H: CrCYP98A9 pC4H: AtCYP98A8 pC4H: EsCYP98A8 pC4H: AaCYP98A8/9 Col-4 cyp98a9 MRM ES+: 289.4 > 153 3.92e4 Standard AtCYP98A9 + AtCYP98A8 + AtCYP98A9 – AtCYP98A8 – 8.0 9.0 10.0 Relative signal Time (min) O OH OH O OH HO Eriodictyol 0 25 50 75 100 125 AtCYP98A9 AtCYP98A8 EsCYP98A8 AaCYP98A8/9Naringenin hydroxylase activity (%)
a
b
c
Figure 4 | CYP98A9 acquired more pleiotropic activity favouring flavonoid metabolism. (a) Colour of the seeds from A. thaliana transgenic lines expressing CYP98A8/A9 genes under control of the A. thaliana C4H promoter. Only expression of lineage I CYP98A9 leads to a transparent testa (tt) phenotype. Col-4: wild-type Arabidopsis Columbia-4. Ler: wild-type Arabidopsis Landsberg erecta. Ws: wild-type Arabidopsis Wassilewskija. tt4-1: transparent testa 4 mutant for the chalcone synthase gene. (b) In vitro enzyme assays showing the CYP98A9-dependent conversion of naringenin into eriodictyol. Assays with (þ ) and controls without ( ) NADPH were performed. Eriodictyol (grayed elution time) was detected using the m/z 289.44153.0 transition in multiple reactions monitoring (MRM) positive mode. (c) Relative naringenin hydroxylase activity (% relative to highest activity set to 100). Error bars represent the s.e. from three independent assays. Aa, Aethionema arabicum; At, Arabidospsis thaliana; Es, Eutrema salsugineum.
multiple rounds of high-efficiency thermal asymmetric interlaced PCR (hiTAIL-PCR)38. Initial nested primer sets were designed to anneal to regions
highly conserved in known Brassicaceae CYP98A8 and CYP98A9 sequences. Polymorphic sites were incorporated into primers, and thus the resulting primers were degenerate at several nucleotide positions. Amplicons were sequenced and used to inform sequence-specific primer design to walk upstream and downstream
along the chromosome using additional rounds of hiTAIL-pPCR38. Primers are available in Supplementary Table 6.
E. syriacum and C. tenella DNA samples were those used in Beilstein et al.39.
Nested amplification was performed in a two-step process according to the hiTAIL-PCR protocol38. The process was repeated until the assembled reads reached flanking genes. The software CoGe40was used to compare assembled
Shrenkiella parvula A3 Brassica rapa A8 Capsella grandiflora A3 Shrenkiella parvula A8 Boechera stricta A9 Capsella rubella A8 Boechera stricta A8 Eutrema salsugineum A8 Capsella rubella A3 Arabidopsis thaliana A3 Arabidopsis thaliana A8
Tarenaya hassleriana A8/9
Brassica rapa A3-2 Capsella rubella A9 Capsella grandiflora A8 Chorispora tenella A8 Leavenworthia alabamica A8 Boechera stricta A3 Sisymbrium irio A8 Euclidium syriacum A8 Aethionema arabicum A3 Brassica rapa A3-1 Arabidopsis lyrata A3 Arabidopsis lyrata A9 Leavenworthia alabamica A9 Capsella grandiflora A9 Arabidopsis lyrata A8 Tarenaya hassleriana A3 Arabidopsis thaliana A9
Brassica rapa A3-3 Aethionema arabicum A8/9 Sinapis alba A8 Eutrema salsugineum A3 = 1 TCG (Ser) ACG (Thr) AAG (Lys) = 0.038 0 25 50 75 100 125 AtCYP98A9_WT AtCYP98A9_Lys209Ala AtCYP98A9_Lys209Ile
Naringenin hydroxylase activity (%)
Lys209 Arg96 Trp109 S-naringenin Heme
a
c
b
Figure 5 | Lys209is a determinant for CYP98A9 flavonoid hydroxylase activity. (a) Docking of S-naringenin in the AtCYP98A9 active site. Arg96, Trp109 and Lys209anchor naringenin for B-ring 30-hydroxylation. (b) Lys209substitution with alanine or isoleucine reduces AtCYP98A9 naringenin hydroxylase activity in vitro. Error bars represent the s.e. from three independent assays. (c) Fitmodel reconstitution of the history of the selection pressure on the AtCYP98A9 Lys209codon. The most frequent codon and amino acid are indicated on branches.
CYP98A8/9 CYP98A9 CYP98A8 CYP98A8 Lineage I CYP98A8
Lineages II and III
CYP98A8
CYP98A9 Phenolamide
3′-hydroxylase Phenolamide5′-hydroxylase Naringenin3′-hydroxylase
5′ 3′ Duplication – CYP98A8 CYP98A9 5′ 3′ Core evolutionary history
+
a
b
c
RetropositionFigure 6 | Model of the CYP98A8/CYP98A9 evolution in Brassicaceae. (a) Initial steps of the CYP98A8/9 evolution shared by lineages I, II and III. The initial CYP98A3 retroposition event in Brassicales (100-70 Ma) led to the CYP98A8/9 ancestral gene, which became a phenolamide 30-hydroxylase. The
retroposition was followed by a tandem gene duplication in Brassicaceae around 45 Ma leading to the CYP98A8 and CYP98A9 sister genes. CYP98A8 then acquired an additional function, the phenolamide 50-hydrolase activity. (b) Further evolution of CYP98A8 and CYP98A9 in the lineage I of the Brassicaceae,
CYP98A8 lost the phenolamide 30-hydroxylase activity in vivo, and maintained only the 50-hydroxylase activity. Meanwhile, CYP98A9 maintained the
30-hydroxylase activity, and simultaneously gained naringenin 30-hydroxylase activity. (c) Further evolution of CYP98A8 and CYP98A9 in the lineages II and
III of the Brassicaceae, in lineages II and III, the CYP98A9 copy was lost, and CYP98A8 maintained both 30- and 50-hydroxylase activities.
sequence reads for E. syriacum and C. tenella with published whole-genome sequences from Brassicaceae lineage I and II species.
PCR products were separated on a 1% agarose gel with low EDTA before excision, filtration (Millapore Centrifugal Filter Units), and ethanol precipitation. Purified PCR products were ligated into pGEM-T Easy vector (Promega) and transformed into TOP10 competent cells and plated on selective media for screening. Positive transformants were cultured overnight in liquid media, and plasmid DNA harvested using an alkaline lysis kit (Zymo Research). Cloned fragments were sequenced in both directions to achieve a minimum of 85% overlap. Overlapping sequence reads were assembled into a consensus sequence using the software Geneious v. 6.6 (Biomatters).
Recombinant CYP98 protein production in yeast
.
Intronless coding sequences of EsCYP98A8, AaCYP98A8/9 and ThCYP98A8/9 genes were PCR-amplified from genomic DNA using USER-compatible primers41. Amplicons were subsequentlyintroduced in the pYeDP60u2 vector42and verified by sequencing. For AtCYP98A8
and AtCYP98A9 pYeDP60 plasmids construction, the corresponding coding sequences were PCR-amplified from genomic DNA and introduced between the BamHI and KpnI sites and BamHI and EcoRI sites, respectively, as described previously18. Primers used for molecular cloning are visible in Supplementary
Table 6.
The Saccharomyces cerevisiae WAT11 strain, expressing the ATR1 cytochrome P450 reductase from A. thaliana43, was transformed with pYeDP60 vectors and
selected on minimum SGI medium (1 g l 1bactocasamino acids, 7 g l 1yeast nitrogen base, 20 g l 1glucose and 40 mg l 1L-tryptophan). Liquid cultures were initiated from selected colonies grown on SGI for 18 h at 28 °C under agitation (160 r.p.m.). Ten milliliters of SGI culture were used to inoculate 200 ml of YPGE (10 g l 1bactopeptone, 10 g l 1yeast extract, 5 g l 1glucose, 3% ethanol by volume). After 30 h growth at 28 °C at 160 r.p.m., recombinant proteins production was induced by supplementing the medium with 10 ml of 200 g l 1galactose and further incubation at 20 °C for 16 h.
Yeast cells were harvested by centrifugation at 7,500g for 10 min at 4 °C, washed with TEK buffer (50 mM Tris-HCl pH 7.5, 1 mM EDTA, 100 mM KCl) and resuspended in 2 ml of TES (50 mM Tris-HCl pH 7.5, 1 mM EDTA, 600 mM sorbitol) supplemented with 5 mM 2-mercaptoethanol and 10 g l 1bovine serum albumin, fraction V (buffer A). Cell suspensions were transferred to 50 ml conical tubes and homogenized with 0.5 mm glass beads by handshaking five times for 1 min each. Beads were washed twice with 30 ml of cold buffer A. Cell debris and remaining glass beads were removed from the pooled lysates by 20 min centrifugation at 7,500g and 4 °C. Supernatant was filtrated on Miracloth (22–25 mm pore size, Calbiochem, Maryland) and microsomal fraction pelleted by centrifugation at 100,000g and 4 °C for 1 h. Pelleted membranes were resuspended in TEG buffer (50 mM Tris-HCl pH 7.5, 0.5 mM EDTA, 30% glycerol by volume) with a Potter-Elvehjem homogenizer. Microsomal membranes were stored at
20 °C until processing.
Enzyme assays
.
Enzyme assays were conducted in 100 ml final volume, containing 20 mM potassium phosphate buffer (pH 7.4), 200 mg microsomal proteins, 100 mM substrate and 500 mM NADPH. Reactions were initiated with NADPH addition, incubated for 30 min at 28 °C in the dark and terminated with 10 ml of 50% acetic acid. Samples were centrifuged at 13,000g for 5 min to pellet microsomal membranes and supernatants were used for product analysis. Hydroxycinnamoyl-spermidines used as substrates and chromatographic references were chemically synthesized18. For naringenin 30-hydroxylation, activity was normalized to theactivity recorded for the same enzyme with tri-p-coumaroyl-spermidine and expressed as a percentage of the maximal activity.
Extraction and analysis of phenolamides and flavonoids
.
For flavonoid analysis, mature dry seeds were ground in a 2 ml safe-locker tube with 1 ml of 75% acetonitrile for 5 min using a TissueLyser II (Qiagen) at 4 °C. The extracts were sonicated for 20 min on ice. After centrifugation (14,000g, 15 min), the pellet was further extracted with 1 ml of 75% acetonitrile by overnight shaking at 4 °C. The two extracts were pooled, dried under vacuum and resuspended in 200 ml of 75% acetonitrile before UPLC-MS/MS analysis44.To release flavonoid aglycones, crude extracts were dried in vacuo. Dry residues were re-suspended in a 50% methanol solution containing 2 N HCl and 2.5 mg ml 1ascorbic acid as an antioxidant. Acid hydrolysis was performed at
80 °C for 2 h. Samples were cooled down before UPLC-MS/MS analysis. For the analysis of phenolamides, samples were first extracted with 1 ml of ice-cold 80% methanol. After centrifugation, the supernatant was recovered and the pellet further extracted with 1 ml of ice-cold 80% methanol overnight at 4 °C. Extracts were pooled, dried under vacuum and resuspended in 200 ml of 80% methanol before UPLC-MS/MS analysis. The mobile phase consisted of water (A) and acetonitrile (B), both containing 0.1% formic acid. Run started with 1 min of 95% A, followed by linear gradients to reach 53% B at 7 min and 100% B at 8 min, then isocratic conditions using B for 2.5 min. Return to initial conditions was achieved in 3.5 min, with a total run time of 15 min. The column was maintained at 28 °C with a flow rate of 0.4 ml min 1, injecting 3 or 5 ml sample. Ultraviolet absorbance was recorded from 200 to 600 nm. MS scans in positive and negative
mode were performed, and multiple reactions monitoring were developed for every detected peak. Compounds were identified by comparison of two orthogonal data (RT and MS/MS) relative to authentic compounds analysed under identical experimental conditions, or putatively annotated by comparison of MS/MS spectra with published data45,46.
Vanillin assay
.
Mature dry seeds were incubated in a solution of 1% (w/v) vanillin (Fluka #350346) prepared in 6 N HCl at room temperature for 1 h. Vanillin turns red upon binding to flavan-3,4-diols (leucoanthocyanidins) and flavan-4-ols (catechins), which are present either as monomers or as terminal subunits of proanthocyanidins31.Determination of gene expression divergence
.
pCYP98A8:uidA and pCYP98A9:uidA plants18were incubated in 90% acetone for 20 min on ice, rinsed with water, and transferred to a GUS solution containing 1 mM 5-bromo-4-chloro-3-indolyl-b-D-glucuronide (X-Gluc), 100 mM sodium phosphate (pH 7.0), 10 mMEDTA, 0.5 mM potassium ferricyanide, 0.5 mM potassium ferrocyanide and 0.1% (v/v) Triton X-100. Samples were incubated at 37 °C in the dark overnight. Tissues were cleared three times in 75% ethanol before imaging with a Nikon (ECLIPSE, E800) microscope. For GUS staining of seeds, additional clearing steps were added. Opened siliques were first treated with ethanol:acetic acid (v/v) for 4 h, and then restored in Hoyer’s medium before microsopy analysis47.
Gene co-expression network analysis was performed with the PlaNet on line tool (http://aranet.mpimp-golm.mpg.de/) using At1g74540 (CYP98A8) and At1g74550 (CYP98A9) as gene id inputs. The gene list corresponding to the different co-expression networks (see Supplementary Table 4) were then subjected to hierarchical clustering based on Pearson correlation coefficient and with optimal leaf-ordering using the Genevestigator embedded tool48.
RNA extraction and gene expression analysis by qRT–PCR
.
Samples were harvested and snap-frozen in liquid nitrogen. Total RNA was isolated from pooled tissues using NucleoSpin RNA Plant kit (Macherey-Nagel) and subsequently treated with DNase I (Fermentas). In all, 1.5 mg of total RNA was reverse-transcribed with SuperScript III Reverse Transcriptase (Invitrogen) in the presence of 200 ng random primers. qRT–PCR plates were prepared with a Biomek 3,000 pipetting system (Beckman Coulter, Villepinte, France) and run on a LightCycler 480 II device (Roche). Each reaction was prepared using 2 ml of tenfold diluted complementary DNA, 5 ml of LightCycler 480 SYBR Green I Master (Roche) and 250 nM of forward and reverse primers in a total volume of 10 ml. The amplification program was 95 °C for 10 min and 40 cycles (95 °C denaturation for 10 s, annealing at 60 °C for 15 s, extension at 72 °C for 15 s), followed by a melting curve analysis from 55 to 95 °C to check for transcripts specificity. All reactions were performed in triplicate. The SAND gene expression was stable among flower samples and thus chosen as internal standard for normalization. Primers used for qRT–PCR detection were previously described18and their sequences are provided in Supplementary Table 6.Phylogenetic dN/dSanalysis
.
Coding sequences were aligned with MUSCLE49and phylogenies were built using PhyML50using the generalized time-reversible (GTR) model with gamma distribution (four rate classes of sites with optimized alpha) via the Seaview software (http://pbil.univ-lyon1.fr/)51. PAML (codeml) (http://abacus.gene.ucl.ac.uk/software/paml.html)52and Fitmodel53were applied to the coding-sequence alignments and phylogenetic trees. All the analyses were run both with a trimmed alignment (404 over 560 codons) where the most variable sites, mainly in the 50and 30regions of the coding sequences, were suppressed by
GBlocks54,55and with a longer alignment where all but the gap-containing sites
were conserved (457 over 560 codons). We ran ‘Branch model’, ‘Site model’, ‘Clade model’, ‘Branch-site’ and ‘Free-ratio’ analyses following codeml standard procedures. We compared nested models using the likelihood ratio test framework to test statistical differences in dN/dS. We ran Fitmodel M2a þ S1
(with o1oo2oo3 and switch between selection regimes allowed) and MX þ S1 (with o0r1; 1oo1r1.5; o241.5 and switch between selection regimes allowed) as an alternative ‘Branch-site’ analysis and M2a (no switch between selection regimes allowed) as a null model to perform a likelihood ratio test53. All results
and associated statistical tests are available in Supplementary Note 1 and Supplementary Table 2.
Homology modelling
.
All Modeller and Autodock calculations were performed using the computing facilities of the CEA-DSV (cluster Gabriel) at Saclay.Three-dimensional models of A. thaliana CYP98A8/9 isoforms deprived of their membrane spanning domain (residues 24 to 491 for 98A8, and residues 24 to 485 for 98A9) were built using Modeller9v12 (refs 56,57) and the crystal structures of CYP2C9 (PDB 1OG2), CYP2A6 (PDB 2PG6), CYP2D6 (PDB 2F9Q), CYP2B4 (PDB 3MVR), and CYP2C5 (PDB 1DT6) as templates. The selection of the template structures for CYP98A8/9 model rebuilding followed four criteria: the best similarity according to PDB-Blast, the highest alignment length, the atomic resolution and ligand-free structures for getting undeformed states. These mammalian CYPs exhibited rather low sequence identities with AtCYP98A8
isoforms (24 to 25%), but the three-dimensional structures are well-aligned by MUSTANG algorithm58, and the resulting alignment was consistent with that obtained with MAFFT-L-INS-I algorithm59. The validated sequence alignment
used as input for Modeller is available in Supplementary Fig. 10.
For each isoform, two runs of 100 models were performed with a further loop refinement protocol, and the generated models sorted by the MODELLER objective function were evaluated by their discrete optimized protein energy and GA341 scores calculated by Modeller. For each isoform, the five best models
(corresponding to the lowest discrete optimized protein energy scores) issued from each run were pooled and submitted to the online metaserver SAVES (structural analysis and verification server (http://services.mbi.ucla.edu/SAVES), and finally to the QMEAN scoring function60server61for model quality assessment. The final model was the best one according to a good compromise between the scores calculated by the SAVES server scoring programs and the QMEAN value. As a result, the selected AtCYP98A8 and AtCYP98A9 models had a QMEAN score equal to 0.701 and 0.689, respectively, which are good scores when compared to the QMEAN scores of individual PDB templates.
Docking experiments
.
The protein structures generated by Modeller were first stripped of all hydrogen atoms, and then atom charges and hydrogen atoms were added to the protein with the UCSF Chimera package (www.cgl.ucsf.edu/chimera)62using AMBER ff99SB parameters. The first parameters applied for the heme (defined as FeIIIprotoporphyrin IX) were obtained from Oda et al.63, and in the docking experiments with S-Naringenin (Nar) the heme cofactor was defined as the highly reactive intermediate compound I (FeIV¼ O)þby using the parameters (geometry
and atom charges) of an AMBER-compatible heme model developed by Shahrokh et al.64. The atom charges of the proximal thiolate were taken from the same work.
Molecular docking experiments with the flavonoid S-naringenin (Nar) at the active site were performed using AutoDock 4 (release 4.2.6) in the semi-flexible mode, and prepared with AutoDock Tools65. The ligand molecule, Nar, was parametrized under Maestro 9.2 (www.schro¨dinger.com) molecular modelling suite and the structure saved under MOL2 format as input files for AutoDock. Partial charges were assigned using OPLS 2005 force field. There is only one rotatable bond in Nar, but several conformations of the non-aromatic ring have been generated as starting point for docking to allow better conformational sampling under Autodock 4, since the algorithm cannot handle ring flexibility. The MOL2 format files, created in Chimera for the receptor and Maestro for the ligand, were converted into the PDBQT format file by AutoDockTools, which merges all nonpolar hydrogen atoms to the carbon atoms they are bonded to. The receptor was kept rigid.
The docking box, in which grid maps were computed using program AutoGrid65, included the active site with the iron-protoporphyrin group on one edge, and the whole distal moiety of the enzyme, including access channels and protein surface, to allow a large sampling of potential poses. The grid built by AutoGrid included 100, 70 and 86 points in x, y and z directions, with a grid spacing of 0.37Å to allow a good compromise between resolution of the explored volume and the size of the binding area. For each Nar conformer, 500 independent runs were performed using the Lamarckian genetic algorithm66. The default
settings were used for all other parameters.
The poses were assigned a score calculated by Autodock that can be considered as an estimated free energy of ligand binding (indicative of binding affinity), then clustered as a function of the closeness of their positions and conformations with root mean square deviation set at 2.0 Å, and finally ranked by their binding score (for the best pose in the cluster). The resulting histograms of 500 poses of Nar docked into AtCYP98A8 and AtCYP98A9 are visible in the Supplementary Fig. 11. Remarkably, whereas the docking computational conditions were identical for both isoforms, the solutions found for AtCYP98A9 are gathered in the same cluster, within 2 Å of root mean square deviation of clustering, while the histogram of AtCYP98A8 displays a more scattered profile. The best clusters, corresponding to the lowest (that is, the most negative) energetic binding scores, were well-resolved in the histogram of scores for AtCYP98A8 but corresponded to non-productive metabolism positions. All clusters under 4 kcal mol 1were independently investigated for their consistency with productive position for 30-hydroxylation
reaction.
Data availability
.
The data that support the findings of this study are available from the corresponding author upon request. CYP98A8 sequence data from E. syriacum and C. tenella have been deposited in GenBank with accession codes KX754460 and KX754461.References
1. Ohno, S. Evolution by Gene Duplication (Springer Verlag, 1970).
2. Lynch, M. & Conery, J. S. The evolutionary fate and consequences of duplicate genes. Science 290, 1151–1155 (2000).
3. Davison, A. J, Benko, M. & Harrach, B. Genetic content and evolution of adenoviruses. J. Gen. Virol. 84, 2895–2908 (2003).
4. Roth, C. et al. Evolution after gene duplication: models, mechanisms, sequences, systems, and organisms. J. Exp. Zool. B Mol. Dev. Evol. 308, 58–73 (2007).
5. Innan, H. & Kondrashov, F. The evolution of gene duplications: classifying and distinguishing between models. Nat. Rev. Genet. 11, 97–108 (2010). 6. Hughes, A. L. The evolution of functionally novel proteins after gene
duplication. Proc. Biol. Sci. 256, 119–124 (1994).
7. Force, A. et al. Preservation of duplicate genes by complementary, degenerative mutations. Genetics 151, 1531–1545 (1999).
8. Stoltzfus, A. On the possibility of constructive neutral evolution. J. Mol. Evol. 49,169–181 (1999).
9. Des Marais, D. L. & Rausher, M. D. Escape from adaptive conflict after duplication in an anthocyanin pathway gene. Nature 454, 762–765 (2008). 10. Hittinger, C. T. & Carroll, S. B. Gene duplication and the adaptive evolution of
a classic genetic switch. Nature 449, 677–681 (2007).
11. Deng, C., Cheng, C. H., Ye, H., He, X. & Chen, L. Evolution of an antifreeze protein by neofunctionalization under escape from adaptive conflict. Proc. Natl Acad. Sci. USA 107, 21593–21598 (2010).
12. Henderson, B. & Martin, A. C. Protein moonlighting: a new factor in biology and medicine. Biochem. Soc. Trans. 42, 1671–1678 (2014).
13. Copley, S. D. An evolutionary perspective on protein moonlighting. Biochem. Soc. Trans. 42, 1684–1691 (2014).
14. Aury, J. M. et al. Global trends of whole-genome duplications revealed by the ciliate Paramecium tetraurelia. Nature 444, 171–178 (2006).
15. He, X. & Zhang, J. Rapid subfunctionalization accompanied by prolonged and substantial neofunctionalization in duplicate gene evolution. Genetics 169, 1157–1164 (2005).
16. Beilstein, M. A. et al. Evolution of the telomere-associated protein POT1a in Arabidopsis thaliana is characterized by positive selection to reinforce protein-protein interaction. Mol. Biol. Evo. 32, 1329–1341 (2015).
17. Lynch, M. & Walsh, B. The Origins of Genome Architecture (Sinauer Associates Sunderland, 2007).
18. Matsuno, M. et al. Evolution of a novel phenolic pathway for pollen development. Science 325, 1688–1692 (2009).
19. Schoch, G. et al. CYP98A3 from Arabidopsis thaliana is a 3’-hydroxylase of phenolic esters, a missing link in the phenylpropanoid pathway. J. Biol. Chem. 276,36566–36574 (2001).
20. Franke, R. et al. The Arabidopsis REF8 gene encodes the 3-hydroxylase of phenylpropanoid metabolism. Plant J. 30, 33–45 (2002).
21. Weng, J. K. & Chapple, C. The origin and evolution of lignin biosynthesis. New Phytol. 187, 273–285 (2010).
22. Xu, J. et al. ABORTED MICROSPORES acts as a master regulator of pollen wall formation in Arabidopsis. Plant Cell 26, 1544–1556 (2014).
23. Beilstein, M. A., Nagalingum, N. S., Clements, M. D., Manchester, S. R. & Mathews, S. Dated molecular phylogenies indicate a Miocene origin for Arabidopsis thaliana. Proc. Natl Acad. Sci. USA 107, 18724–18728 (2010). 24. Cheng, S. et al. The Tarenaya hassleriana genome provides insight into
reproductive trait and genome evolution of crucifers. Plant Cell 25, 2813–2830 (2013).
25. Ming, R. et al. The draft genome of the transgenic tropical fruit tree papaya (Carica papaya Linnaeus). Nature 452, 991–996 (2008).
26. McDonald, J. H. & Kreitman, M. Adaptive protein evolution at the Adh locus in Drosophila. Nature 351, 652–654 (1991).
27. Belmonte, M. F. et al. Comprehensive developmental profiles of gene activity in regions and subregions of the Arabidopsis seed. Proc. Natl Acad. Sci. USA 110, E435–E444 (2013).
28. Bell-Lelong, D. A., Cusumano, J. C., Meyer, K. & Chapple, C. Cinnamate-4-hydroxylase expression in Arabidopsis. Regulation in response to development and the environment. Plant Physiol. 113, 729–738 (1997).
29. Buer, C. S. & Muday, G. K. The transparent testa 4 mutation prevents flavonoid synthesis and alters auxin transport and the response of Arabidopsis roots to gravity and light. Plant Cell 16, 1191–1205 (2004).
30. Buer, C. S., Imin, N. & Djordjevic, M. A. Flavonoids: new roles for old molecules. J. Integr. Plant Biol. 52, 98–111 (2010).
31. Deshpande, S. S., Cheryan, M. & Salunkhe, D. K. Tannin analysis of food products. Crit. Rev. Food Sci. Nutr. 24, 401–449 (1986).
32. Lepiniec, L. et al. Genetics and biochemistry of seed flavonoids. Annu. Rev. Plant Biol. 57, 405–430 (2006).
33. Mierziak, J., Kostyn, K. & Kulma, A. Flavonoids as important molecules of plant interactions with the environment. Molecules 19, 16240–16265 (2014).
34. Doughty, J., Aljabri, M. & Scott, R. J. Flavonoids and the regulation of seed size in Arabidopsis. Biochem. Soc. Trans. 42, 364–369 (2014).
35. Robinson, S. J. et al. An archived activation tagged population of Arabidopsis thaliana to facilitate forward genetics approaches. BMC Plant Biol. 9, 101 (2009).
36. Weng, J. K., Mo, H. & Chapple, C. Over-expression of F5H in COMT-deficient Arabidopsis leads to enrichment of an unusual lignin and disruption of pollen wall formation. Plant J. 64, 898–911 (2010).
37. Clough, S. J. & Bent, A. F. Floral dip: a simplified method for Agrobacterium-mediated transformation of Arabidopsis thaliana. Plant J. 16, 735–743 (1998).
