A hydroxycinnamoyltransferase responsible forsynthesizing suberin aromatics in ArabidopsisJin-Ying Gou, Xiao-Hong Yu, and Chang-Jun Liu1

Biology Department, Brookhaven National Laboratory, Upton, NY 11973

Edited by Richard A. Dixon, The Samuel Roberts Noble Foundation, Ardmore, OK, and approved September 16, 2009 (received for review May 19, 2009)

Suberin, a polyester polymer in the cell wall of terrestrial plants,controls the transport of water and nutrients and protects plant frompathogenic infections and environmental stresses. Structurally,suberin consists of aliphatic and aromatic domains; p-hydroxycin-namates, such as ferulate, p-coumarate, and/or sinapate, are themajor phenolic constituents of the latter. By analyzing the ‘‘wall-bound’’ phenolics of mutant lines of Arabidopsis deficient in a familyof acyl-CoA dependent acyltransferase (BAHD) genes, we discoveredthat the formation of aromatic suberin in Arabidopsis, primarily inseed and root tissues, depends on a member of the BAHD superfamilyof enzymes encoded by At5g41040. This enzyme exhibits an �-hydroxyacid hydroxycinnamoyltransferase activity with an in vitrokinetic preference for feruloyl-CoA and 16-hydroxypalmitic acid.Knocking down or knocking out the At5g41040 gene in Arabidopsisreduces specifically the quantity of ferulate in suberin, but does notaffect the accumulation of p-coumarate or sinapate. The loss of thesuberin phenolic differentially affects the aliphatic monomer loadsand alters the permeability and sensitivity of seeds and roots to saltstress. This highlights the importance of suberin aromatics in thepolymer’s function.

BAHD superfamily � wall-bound phenolics

Land plants have evolved different mechanical/defensive machin-eries to reinforce their cell wall integrity and rigidity to protect

themselves from various environmental stresses (1). Along withwell recognized lignification, cell wall suberization is anotherphysiologically important strategy to regulate the apoplastic trans-port of water and solutes and to protect the plant from the invasionof pathogens (1–3).

Suberin occurs in the cell walls of external and internal planttissues. Suberized cells primarily are present in underground tissues(e.g., epidermis, endodermis, exodermis, root and tube phellem), inthe coats of mature seeds, in the bundle-sheath cells, and in thephellem of aerial tissues that undergo secondary thickening (2, 4).

Structurally, suberin is a complex lipophilic polymer, containinga fatty acid-derived domain (aliphatic suberin) and a (poly)hydroxy-cinnamate domain (aromatic suberin). The aliphatic suberin is a3D, glycerol-bridged polyester network, comprised primarily of�-hydroxyacids and �,�-dicarboxylic acids, with chain lengths rang-ing from C-16 to C-32 (mainly C-18). They exist as discretecomponents between the plasmalemma and the primary cell wallmatrix (2, 3, 5). The aromatic domain is principally composed ofp-hydroxycinnamates (e.g., ferulate, p-coumarate, and sinapate)and their derivatives and possibly a low level of monolignols (4, 6).The abundance of those phenolics detected in the suberized tissuesvaries in different species and ranges as high as approximately 10%of total suberin content (3). The aromatic units of suberin arecovalently linked with the aliphatic domain through ester bonds.These aromatic units are then presumably polymerized via radicalcoupling reactions to form an aromatic domain, which is incorpo-rated within the matrix of the primary cell wall (4, 7, 8). A buildingblock, i.e., the trimer of �-feruloyl acylglycerol diester, was char-acterized from cork suberin, implicating the role of ferulate incross-linking the aliphatic suberin polymer to the adjacent polyaro-matic domain (9).

Despite intensive analyses of the compositions of suberin indifferent species (10–13), details about the biosynthesis and dep-osition of the monomeric precursors of suberin, and the macro-molecular assembly of its components, are not entirely understood.Some progress recently was made toward clarifying the biosynthesisof aliphatic suberin and of cutin, a related lipid polymer. Geneticand biochemical studies unveiled the catalytic steps for fatty acidelongation, �-oxidation, and (poly)glycerol acylation, the crucialreactions for the formation of the structural elements of lipidpolymers (14–17). Among these studies, Beisson et al. (17) linkedan acyltransferase of GPAT family (acyl-CoA, glycerol-3-phosphate acyltransferase), namely GPAT5, to the biogenesis ofsuberin polyester. The enzyme catalyzes the formation of acylglyc-erol in vitro; and the mutant plant gpat5 exhibited a 50% decreasein the content of aliphatic suberin in roots and seed coats. Thereduction primarily occurred in the very-long-chain dicarboxylicacids and hydroxyacids.

Compared with our knowledge of the biosynthesis of aliphaticsuberin, the molecular mechanisms underlying the formation of thearomatics of suberin are less well established. A series of hydroxy-cinnamate esters of the long chain alcohols/acids (i.e., C-16 to C-32)were found in the suberizing tissues of different species (8, 11, 18,19), suggesting that the phenolic ester conjugates may function assuberin precursors (4). The activity of a hydroxycinnamoyl-CoA:�-hydroxyacid O-hydroxycinnamoyltransferase (HHT) was detectedin the wound-healing potato-tuber discs; the enzyme then waspurified from suspension cultures of tobacco cells (20, 21). HHTcatalyzes the formation of feruloylpalmitic acid and alkyl ferulateesters in vitro. The time course of its response to wounding, and itsdistribution in the tissues of potato tubers coincided with thedeposition of suberin (21). The response of HHT induced by theplant hormone abscisic acid is similar to those of several otherenzymes that might be involved in synthesizing suberin in the tubers(22, 23). HHT activity was detected in root extracts of many higherplants (24), pointing to its general role in forming the hydroxycin-namate esters of fatty acids in suberizing tissues. However, the geneencoding HHT has not been identified, nor has the role of HHT inthe biosynthesis of suberin aromatic been affirmed convincingly.

A growing number of plant-specific acyl-CoA dependent acyl-transferases, the so-called BAHD superfamily enzymes, recentlywere identified and characterized through genetic and biochemicalstudies (25, 26). The name ‘‘BAHD family’’ originated from its first4 characterized enzymes: Clarkia breweri benzyl alcohol O-acetyltransferase (BEAT); Gentiana triflora anthocyanin O-hydroxycinnamoyltransferase (AHCT); Dianthus caryophyllus an-thranilate N-hydroxycinnamoyl/benzoyl transferase (HCBT); andCatharanthus roseus deacetylvindoline 4-O-acetyltransferase

Author contributions: C.-J.L. designed research; J.-Y.G. and X.-H.Y. performed research;J.-Y.G. and C.-J.L. analyzed data; and C.-J.L. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Data deposition: The sequence reported in this paper has been deposited in the GenBankdatabase (accession no. GQ176867).

1To whom correspondence should be addressed. E-mail: cliu@bnl.gov.

This article contains supporting information online at www.pnas.org/cgi/content/full/0905555106/DCSupplemental.

www.pnas.org�cgi�doi�10.1073�pnas.0905555106 PNAS � November 3, 2009 � vol. 106 � no. 44 � 18855–18860

PLA

NT

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

May

17,

202

0

(DAT). The characterized BAHD enzymes use a range of CoA-thioester donors, including the aromatic CoAs, to modify a varietyof plant metabolites, such as shikimate-phenylpropanoid deriva-tives, alkaloids, terpenoids, polyamines, and the short- or middlechain aliphatic alcohols (27–31). The BAHD enzymes’ broadspectrum of substrate specificity and diverse biological functionslead us to assume their potential involvement in the acylation of cellwall associated biopolymer components, such as polysaccharides,lignin, and suberin (32).

To systematically characterize the functions of the BAHD familyof acyl-CoA-dependent acyltransferases and, primarily, to ascertainwhich enzymes potentially are involved in modifying the cell wallcomponents, we identified 61 and 94 putative BAHD family genes,respectively, from the Arabidopsis and Populus genomes (32).Subsequently, we selected and screened a batch of correspondingT-DNA insertion homozygous mutant lines of Arabidopsis. Byanalyzing and quantifying the ‘‘wall-bound’’ acyl esters, we distin-guished a few mutant lines with reduced levels of wall-boundphenolics, designated RWP mutants. One of them, RWP1, exhibiteda decrease in a specific ferulate constituent in the cell wall suberinfraction of the root, stem, and mature seed of Arabidopsis. Theprotein encoded by the corresponding RWP1 gene displayed atransacylation activity that conjugates hydroxycinnamate (via itsCoA thioester) onto the �-hydroxyacids, so demonstrating a typicalHHT activity. Knocking down or knocking out this �-hydroxy-acid:hydroxycinnamoyltransferase specifically lowered the ferulatecontent of suberin, but did not affect the accumulation of p-coumarate or sinapate in roots or seeds. The loads of the lipidmonomers of aliphatic suberin changed differentially; in particular,dicarboxylic acids increased. Nevertheless, the knockout plantsdisplayed an obviously enhanced permeability and developmentalsensitivity to ion stresses; their stress-response behaviors resemblethose observed in Arabidopsis lines that were severely deficient inaliphatic suberin. These in vitro and in vivo results indicate that theRWP1 enzyme functions specifically as a �-hydroxyfatty acid:feruloyltransferase for the synthesis of aromatics of the suberinpolymer. The aromatic components, similar to the aliphatic poly-esters, are critical for the proper structural organization andfunction of suberin.

ResultsReduced Levels of Wall-Bound Ferulate Accumulate in RWP1 Mutants.We identified 61 putative genes from Arabidopsis genome se-quences by using the conserved sequence motifs of BAHD super-family members (32). After excluding some functionally character-ized members and their close homologues, we collected the T-DNAinsertion mutant lines for the remaining putative BAHD membersand screened them by genomic PCR for their homozygosity [ex-emplified in supporting information (SI) Fig. S1]. Subsequently, weprepared extractable-free cell wall fractions from stems of thosehomozygous mutant lines and hydrolyzed them with NaOH. Wethen examined the changes of the wall-bound phenolics of thehydrolysates via HPLC/diode array detection/MS.

A homozygous mutant line with a T-DNA insertion in theAt5g41040 gene from 2 independent alleles, SALK�101708(rwp1–1) and Salk�048898 (rwp1–2) (Fig. 1A) exhibited approxi-mately 48% to 66% decline in specific ferulate content in thehydrolysate from old stems, compared with that of the WT plants(Fig. S2). The gene At5g41040 encodes a deduced polypeptide of457 aa residues, encompassing the typical diagnostic sequencesignatures of ‘‘HXXXD’’ and ‘‘DFGWG’’ of BAHD family mem-bers (Fig. S3). In the mutant line rwp1–1 (SALK�101708), theT-DNA was inserted into the first intron of the gene, whereas in theallelic rwp1–2 (SALK�048898), the insertion of T-DNA disruptedthe gene at the third exon (Fig. 1A). The transcript of the At5g41040gene was detected readily in the WT seedlings, but was barelynoticeable in both homozygous insertion alleles (Fig. 1B), confirm-

ing that the insertion eliminated the production of the At5g41040transcript.

Examining At5g41040 gene expression in different tissues ofArabidopsis by in silico microarray analysis and quantitative RT-PCR (Fig. 1C) revealed the presence of this transcript in above- andunder-ground tissues. A high RNA abundance was detected in thesiliques, roots, and seedlings, with less in the stem, implicating theprimary functional sites of At5g41040 may be in roots and seeds(within the silique).

Wall-Bound Ferulate Is Reduced in the Suberin Polymers. To deter-mine which cell wall component experiences a reduction in ferulate,we treated extract-free cell wall materials of the roots from WTArabidopsis and from 2 At5g41040 allelic mutant lines with ammo-nium oxalate and hot acid to release pectin, and then digested theresidues sequentially with endo-�-xylanase and Driselase to release,respectively, the oligomers of xylan and the other sugars. Oursubsequent saponification analyses of these different oligosaccha-ride preparations did not reveal any wall-bound phenolics or anychange of the phenolics. However, after treating the enzyme-digested insoluble residues of the roots and mature seeds withBF3/methanol solution—i.e., the routine method of trans-esterification used to harshly depolymerize cell wall suberin—wenoted a marked change in the HPLC-UV profiles of the suberindepolymerized hydrolysates from At5g41040 mutants, comparedwith those from WT plants (Fig. 2 A-F). The phenolic profiles fromthe WT roots and seeds displayed a predominant ferulate peak, theelectrospray mass spectrum of which showed the molecular ion(M�H�) of 209 (m/z) as the methyl ester derivative from thedepolymerization (Fig. 2A). Its UV spectrum (Fig. 2D) and theretention time were identical to those of authentic ferulic acid afterthe same treatment. In addition, a few minor peaks in the phenolicextract from the WT root and seed suberin depolymerizationexhibited a similar UV spectrum to that of ferulate, but a largermolecular mass, suggesting that they might be different feruloylatedderivatives from the suberin polymer (Fig. 2 A and D). In contrast,ferulate and its derivatives were greatly reduced or essentiallylacking, respectively, from the root or seed suberins of both allelicmutant lines (Fig. 2 B, C, E, and F). The ferulate level in 8-week-oldroots of mutant lines was only approximately 20% of WT levels(Fig. 2G), with almost negligible amounts in their mature seeds(Fig. 2H). Besides ferulate, we also detected the other phenolics,e.g., p-coumarate, or sinapate, from root or seed suberin depoly-

SALK_101708

SALK_048898

A

rwp1_1

rwp1_2 Rela

tive

exp

ress

ion

leve

l (%

)

020406080

100120

WT rwp1_1 rwp1_2

B

C

Rela

tive

exp

ress

ion

le

vel (

%)

Root Seedling StemLeaf Flower Silique0

20

40

60

80

100

120

Fig. 1. At5g41040 gene structure, T-DNA insertions, and gene expressionpattern. (A) Schematic models of T-DNA insertions in lines of SALK�101708(rwp1–1) and SALK�048898 (rwp1–2) alleles. The 5� and 3�-UTR, exon, andintron are represented, respectively, by the dark box, open box, and line. Thetriangles show the position of the T-DNA insertion. (B) Quantitative RT-PCRanalysis of At5g41040 gene expression in seedlings of WT and 2 homozygousmutant lines, rwp1�1 and rwp1�2. (C) Pattern of tissue specific expression ofAt5g41040 gene in WT Arabidopsis.

18856 � www.pnas.org�cgi�doi�10.1073�pnas.0905555106 Gou et al.

Dow

nloa

ded

by g

uest

on

May

17,

202

0

merization. However, in contrast to the ferulate, the levels ofp-coumarate in root or sinapate in seed suberin were essentiallyunchanged or even slightly increased in the mutant plants (Fig. 2 Gand H). To confirm the BF3/methanol trans-esterification results,we further conducted thioacidolytic analyses on preparations ofseed suberin and resolved the released phenolics by liquid chro-matography (LC)/MS. Many phenolic derivatives were detected, 2of which proved identical to those derived from authentic ferulicacid treated with thioacidolysis (Fig. S4 A–D); consistently, in the2 allelic mutant lines, they were reduced to only approximately 37%and 50% of WT levels, respectively (Fig. S4G). We also usedGC-MS to assess the lipid monomers from seed and root suberinBF3/methanol depolymerization (Fig. S5A). We calculated the levelof a particular monomer based on the ion-peak area per unit sample

(Fig. 2I). The lipid compositions in WT and mutant root or seedsuberin showed differential changes. The �-hydroxy acids remainedalmost unchanged or were only slightly reduced in both mutantlines; but almost all of the �,�-dicarboxylic acids and a few alkanoicacid detected exhibited increases (Fig. 2I). After staining the seedswith the lipophilic suberin dye Sudan red, we observed no colora-tion difference between the seed surfaces of the mutant and WTlines, further confirming that the overall load of aliphatic suberin inseed coats of the mutant lines had not been lowered (Fig. S5B).

Together, these data suggest that the protein encoded byAt5g41040 specifically and substantially affects the accumulation ofthe ferulate constituent of suberin in roots and seeds.

The At5g41040 Gene Encodes a Hydroxycinnamoyl-CoA:�-Hydroxy-acid O-Hydroxycinnamoyltransferase. To determine whether theAt5g41040 gene encodes a functional hydroxycinnamoyltransferase,we sub-cloned the cDNA of the At5g41040 encoding the ORF intothe pHIS9 gateway protein-expression vector (36), and expressed itin Escherichia coli. The purified recombinant enzyme from E. coliwas incubated with feruloyl-CoA and 16-hydroxyhexadecanoic acid(16-hydroxypalmitic acid), the common type of �-hydroxyacid insuberin. The in vitro assay was analyzed by LC-MS. A uniqueproduct in the reaction of the recombinant enzyme from E. colicontaining pHIS9-At5g41040 was detected (Fig. 3A), which has aUV spectrum similar to ferulate, but its positive atmosphericpressure chemical ionization (APCI)-ion mass spectrum gives anmolecular ion [M�H]� at an m/z of 449, indicating that the ferulateis conjugated with 16-hydroxypalmitic acid (Fig. 3C). Besides themajor molecular ion, the mass spectrum also contains fragment ionsof 177 and 163, further suggesting the presence of a feruloyl residuein the product (Fig. 3C). No such activity was apparent in theboiled-inactive recombinant enzyme (Fig. 3B), or in the emptyvector control extract.

The activity of the recombinant enzyme was maximal in a pH 7.5phosphate buffer. The optimal temperature for maximum activitywas 10 °C. Although the enzyme showed activity with both feruloyl-CoA and p-coumaroyl-CoA, it demonstrated approximately four-fold higher binding affinity and catalytic efficiency with the formerthan with the latter (Table 1). No activity was detected withbenzoyl-CoA and caffeoyl-CoA. Among the aliphatic alcohols, fattyacids, and monolignols we tested, the enzyme showed a highestactivity for the feruloylation of the �-hydroxyacid 16-hydroxyhexa-decanoic acid, with a catalytic ratio at 3,430 M�1s�1. Only mildactivity was detected with a few medium-chain �-alkanols oralkenols. No activity was apparent with fatty acids lacking an

209.1

195.1

177.1

m/z100 200 nm200 300 400

Ints.×10

mAU

0

20

40

12

AA

bsor

banc

e at

330

nm

(mA

U)

0

100

200

300

0

100

200

300

0

100

200

300

do

cosa

no

l

C24

:0d

C22

:0d

C18

:2d

C18

:1d

C16

:0d

C16

:0o

C16

:1o

C18

:2o

C18

:1o

C22

:0o

C24

:0o

C24

:0

C22

:0

C18

:0

C18

:1

C16

:0

×1010

0

2

4

6

8

Pea

k ar

ea p

er g

ram

see

d

p-Coumarate

5

0

[M+H]+

B

C

Ferulate

Ferulate

Sinapate

WT rwp1-1 rwp1-2

2015 16 17 18 19 (min)0

20

40

60

0

20

40

60

0

20

40

60

nm200 300 400

mAU

0

20

40

11 12 13 14 15

D

E

F147.1

165.1

179.1Ints.X10

4

0

2

m/z100 200

[M+H]+

1

2

Ferulate

4

SinapateFerulate

Alkanoic acids Hydroxy acids Dicarboxylic acids

μmole

/g D

W ro

ot C

W

0

0.3

0.6

0.9

1.2

Ferulate Drv1 Drv2 Drv3 p-Coumarate

WTrwp1-1rwp1-2

G H

IFerulate Drv4 Sinapate

nmol

e/g

DW

see

d

WTrwp1-1

rwp1-2

0

40

80

120

160

200

Fig. 2. Compositional analysis of suberin from the WT Arabidopsis and rwp1lines. (A-C) Portion of HPLC profiles of the phenolics from root suberin of theWT (A) and T-DNA insertion mutant line rwp1�1 (B) and rwp1–2 (C). (D-F)Portion of HPLC profiles of the phenolics from seed suberin of the WT (D) andrwp1–1 and 1–2 (E and F). Insets: Mass and/or UV spectra of the detectedmethyl esters of phenolics. (G and H) Quantification of the phenolic contentsdetected in the suberins of the WT and rwp1 mutant roots (G) and seeds (H);Drv 1–4 represent the identified ferulate derivates from the root and seedsuberin extract, Drv 3 resolved in the wash phase of root HPLC profile is notshown in A. (DW, dry weight; CW, cell wall.) Data are mean � SD of triplicatemeasurements. (I) Aliphatic compositions of seed suberin of the WT and rwpmutant lines determined by GC-MS. Data shown are mean � SD of triplicates.Polyol fatty acid 10, 16-hydroxy 16:0, and 9,10,18-hydroxy 18:1 are present asC16:1 and C18:2, respectively (o, omega-hydroxy acid; d, dicarboxylic acid).

Retention time (min)

×10 449.3 [M+H]

431.2391.4

362.4209.2177.1

163.299.3

m/z500400300200100

Intensity4

0123

C

431HO O

O CH2 ]CH2 C OHC CH2

391209177OHO

H3CO

O CH2 [ CH2 13OHC CH2

+

B

16 18 201412108

Feruloyl-CoA

A

0

10

20

30

Ab

sorb

ance

at

330

nm

16 18 201412108

16-feruloylpalmitic acid

mAu

Fig. 3. In vitro activity of RWP1 expressed in E. coli. (A and B). The HPLCchromatograms of enzymatic assay of the Arabidopsis recombinant RWP1 (A)and the heat-inactivated enzyme (B) incubated with feruloyl-CoA and 16-hydroxypalmitic acid shows production of 16-feruloylpalmitic acid. (C) Themass spectrum and structure of enzymatic product, and (D) phylogenetic treeof RWP1 homologues in different plants.

Gou et al. PNAS � November 3, 2009 � vol. 106 � no. 44 � 18857

PLA

NT

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

May

17,

202

0

�-hydroxyl, or with the aromatic monolignols (Table 1). These datasuggest that the At5g41040 gene encodes a typical hydroxycin-namoyl-CoA:�-hydroxyacid O-hydroxycinnamoyltransferase,henceforth referred to as AtHHT, in vitro.

HHT activity has been detected in the crude extracts of roots ofmany terrestrial plants (24). Searching the gene database with theAt5g41040 sequence uncovered many close homologue sequencesfrom the gymnosperms, and the dicot and monocot angiosperms.These include Brassica rapa, Populus trichocarpa, Solanum lycop-ersicum, Picea sitchensis, Oryza sativa, Zea mays, Glycine max, andMedicago truncatula (Fig. S6). These data imply that HHT genesand their activity might be conserved in different species.

Root Tips and Seed Coats of AtHHT Mutant Line Show High Perme-ability and Sensitivity to Salts. Tetrazolium red is a cationic dye thatwas previously used in monitoring permeability of seed coats of thealiphatic suberin-deficient Arabidopsis (17). We used it to incubatethe 5-d after germination (DAG) roots of AtHHT mutant lines.Within 4 h, the meristem of the mutant roots quickly adsorbed thered dye, whereas the roots of WT were less stained (Fig. 4A).Similarly, when incubating mature seeds from the mutant lines withtetrazolium red, the staining on seed coats appeared first near thehilum, and then diffused outward on the surface. After staining for8 h, an intense color was apparent on the surface of the mutantseeds’ coats, whereas the WT seeds exhibited only faint colorationprimarily restricted to the hilum region (Fig. 4B). These resultssuggest that the reduction of aromatic suberin in the mutant linesincreases the permeability to salt of the roots and seed coats.

When seeds from AtHHT mutant and WT lines were germinatedon Murashige and Skoog agar medium with increasing salt con-centrations, the percentage of germinated seeds from mutant lineswas approximately 40% to 80% less than that of the WT seeds atconcentrations of 100 to 200 mM of NaCl, KCl, or Na2SO4 (Fig.

4C). Moreover, the mutant seeds that germinated at 100 mM ofNaCl or KCl showed a much more severe developmental arrest,before establishing green cotyledons (Fig. 4D). The enhancedpermeability of the mutant seeds, and their delayed germinationand seedling development in different salts, suggest that aromaticsuberin plays a critical role in controlling the tissues’ ion uptake.

DiscussionFerulate has been characterized as a major aromatic constituent ofsuberin in Arabidopsis, potato, and many other plants (8, 33).Ferulate esters of long-chain fatty acids and alkan-1-ols, and a smallquantity of a mono-feruloylglycerol were isolated from the suberinof potato periderm; those ferulate esters were postulated as themonomeric precursors of suberin (11). By systematically analyzingthe wall-bound phenolics of BAHD acyltransferase knock-down/KO mutants, followed by biochemical characterization, wedemonstrate that the formation of long-chain fatty acid-ferulateester requires a novel BAHD family member, i.e., AtHHT that isencoded by At5g41040. We analyzed 2 independent AtHHT ho-mozygous allelic mutant lines that showed the specific disruption ofthe At5g41040 transcripts (Fig. 1B and Fig. S1). Consequent to genedisruption, the mutant alleles displayed a severe reduction of theaccumulation of suberin ferulate in roots and seeds, respectively.The recombinant protein of At5g41040 displayed a hydroxycin-namoyltransferase activity and a kinetic preference for feruloyl-CoA and acceptor 16-hydroxypalimtic acid. It is possible thatAtHHT also accepts other very long chain �-hydroxyacid sub-strates, e.g., C-18 or C-24, but these were not examined in the assaybecause of commercial unavailability of these chemicals. Thecatalytic properties of AtHHT are similar to the reported HHTactivities from potato tubers, tobacco cells, and many other species(20, 21), except that AtHHT requires a optimal temperature lowerthan the normal physiological condition for its maximum activity.

Table 1. Activity and kinetics of recombinant At5 g41040 on different acyldonors and acceptors

Chemicals Relative activity, %* Km, �M Kcat, � 10�3 (s�1) Kcat/Km, (M�1*s�1)

Acyl donor†

Feruloyl-CoA 100 9.7 � 3 17.3 � 1.5 1787p-Coumaroyl-CoA 28.5 36.7 � 15.9 13.6 � 0.9 372Caffeoyl-CoA ND – – –Benzoyl-CoA ND – – –

Acyl acceptor‡

16-OH-palmitic acid 100 5.1 � 1.5 17.2 � 4.4 3,430Palmitic acid ND – – –Nonadecanoic acid ND – – –12-OH-octadecanoic acid ND – – –12-OH-octadecanol ND – – –1-Octadecanol ND – – –1-Hexadecanol ND – – –1-Tetradecanol ND – – –1-Dodecanol ND – – –1-Decanol ND – – –1-Octanol 24.5 – – –2-Octanol ND – – –1-Heptanol 18.9 – – –cis-2-hexen-1-ol 26.6 – – –trans-2-hexen-1-ol ND – – –1-Hexanol ND – – –1-Propanol ND – – –Sinapyl alcohol ND – – –Coniferyl alcohol ND – – –

*Specific activity with feruloyl-CoA and 16-hydroxypalmitic acid (1.68 nmol mg�1 min�1) was taken to be 100%.ND, not detected.

†Reactions were performed using 16-hydroxypalmitic acid as the acyl acceptor.‡Reactions were performed using feruloyl-CoA as the acyl donor.

18858 � www.pnas.org�cgi�doi�10.1073�pnas.0905555106 Gou et al.

Dow

nloa

ded

by g

uest

on

May

17,

202

0

This unique biochemical property may infer a potential biologicalfunction of this enzyme and the related suberin synthesis in thetolerance of cold environment; alternatively, the recombinantprotein may be thermosensitive in vitro.

The aromatic constituents of suberin, p-coumarate, ferulate,sinapate, and their derivatives differentially accumulate in thetissues of Arabidopsis. We found that ferulate, p-coumarate, andtheir derivatives were the predominant phenolic constituents ofroot suberin (Fig. 2A), whereas in mature seeds we mainly detectedferulate and sinapate (Fig. 2D). In AtHHT-KO plants, the ferulate(and its derivatives) of suberin were reduced specifically, whereasthe content of p-coumarate and sinapate were not lowered (Fig. 2Gand H). These data indicate that AtHHT encoded by At5g41400biologically functions as a specific feruloyltransferase for modifyingthe long-chain fatty acid constituents of suberin. Further, the dataimply that different acyltransferases might exist for forming otheraromatic esters of suberin in Arabidopsis. Examining the phylogenyof BAHD family members of Arabidopsis (32), we found 2 addi-tional homologous genes that share approximately 74% to 76%amino acid identity with At5g41040 (Fig. S6). It will be interestingto examine their functions next. Recently, 2 polyamine hydroxy-cinnamoyltransferases, which are involved in the formation ofspermidine conjugates in seeds, were characterized from Arabidop-sis (31). Both in vitro and in vivo analyses demonstrated the strict

thioester-donor discrimination of both enzymes in the transacyla-tion of polyamines. AtHHT recognized p-coumaroyl-CoA andferuloyl-CoA in vitro, but it kinetically prefers the latter, which mayexplain its high specificity for feruloylation in vivo.

The suberin phenolics presumably cross-link together, forming apolymeric domain that associates with the cell wall’s polysaccha-rides. They may be also esterified to long-chain hydroxyl fattyacids/alcohols within the poly(aliphatic) domain (3, 6). However, itis challenging to quantitatively determine all those phenolics re-leased from different sub-fractions of suberized tissues because nodepolymerization method can selectively isolate those phenolicsand because of the structural complexity of the putative poly(phe-nolic) domain, the chemical complexity of their derivatization, andthe potential degradation of suberin aromatics under rigorousdepolymerization. In our BF3/methanol trans-esterification andthioacidolytic analyses, a small portion of suberin ferulate remainedin the root or seed suberins of both allelic mutant lines (Fig. 2 A-Hand Fig. S4). Although this might signify that the deposition of somesuberin ferulates in a presumptive subfraction is independent of theactivity of the characterized AtHHT, it is more likely that theremaining suberin ferulates in both mutant lines were generated viaexisting AtHHT activity derived from residual gene expression inboth mutant lines (Fig. 1B).

A reduction/loss of suberin ferulate coincided with an increase in�,�-dicarboxylic acids, the major lipid component of Arabidopsissuberin (Fig. 2I and Fig. S5). This increase in dicarboxylic acids mayresult from the increased availability of the precursor, the �-hy-droxy acids, as a result of the deficiency of AtHHT in the mutantlines. Alternatively, the reciprocal accumulation of suberin aromat-ics and the major lipid components may implicate a potentialmetabolic compensatory mechanism that might reinforce theplant’s mechanical/defensive machinery. Despite the finding thatthe load of total aliphatic suberin in the KO allelic mutantsexhibited no reduction or even a slight increase, the mutant plantsdisplayed an obvious alteration of the permeability to ionic dye andthe sensitivity to salt stresses. Indeed, their responsive behaviorsand phenotype are comparable to those of the AtGAPT5 mutantline, in which the aliphatic compositions were greatly reduced as aresult of the disruption of the gene for aliphatic suberin biosynthesis(17). These data further suggest that a deficiency of AtHHT mostlikely disturbs both the compositional and structural properties ofsuberin, and that the aromatic domain of suberin has as significanta role in conferring suberin’s physiological functions as does thealiphatic domain.

Materials and MethodsPlant Materials. The seeds of A. thaliana (Col0) T-DNA insertion mutants of theputative BAHD family of acyltransferase genes, including SALK�101708 andSALK�048898, were purchased from Arabidopsis Research Center. The seed ger-mination and plant growth in plate or in soil were kept in a growth chamberunder a 16/8 h light/dark regimen at 22 °C.

Chemicals. p-Coumaroyl and feruloyl-CoA were synthesized following themethod described by Stockigt and Zenk (34). Caffeoyl-CoA was kindly providedby Fang Chen of the Samuel Roberts Noble Foundation (Ardmore, OK). Aceto-nitrile was purchased from Fisher. All other solvents and chemicals were pur-chased from Sigma-Aldrich unless otherwise stated.

Homozygous Mutant Isolation and Gene Expression Analysis. T-DNA insertioninformation for each mutant line was obtained from the Arabidopsis ResearchCenter and the Salk Institute. To select homozygous lines, genomic DNA wasextracted from young leaves and PCR was performed with T-DNA-specific primerLBb1.3 and gene-specific primers (Table S1). To confirm the gene knockout, weused quantitative RT-PCR to detect the transcript in 5-DAG seedlings with genespecific primers (Table S1). To analyze gene-expression pattern, we extractedRNA with TRIzol solution (Invitrogen) from the following tissues according tomanufacturer’s instructions: root and seedling at 14 DAG in 1⁄2 MS plate, leaf,stem,flower,andsiliqueof8-weekplantsgrowninsoil.Reverse transcriptionwascarried out with SuperScript III First Strand Synthesis System (Invitrogen) accord-ing to the user’s manual. Real-time PCR was carried out with iQ SYBR Green

rwp1_2 rwp1_1 WT A

B rwp1_2 rwp1_1 WT

Na2SO4

KCl

NaCl

0

40

80

120

0 100 200 300 (mM) 0 100 200 300

0

40

80

120

0

40

80

120

C D

WT rwp1-1 rwp1-2

Ger

nim

atio

n (

%)

Seed

ling

est

ablis

hm

ent

(%)

Na2SO4

KCl

NaCl

Fig. 4. Permeability to dye and seed germination of WT and rwp1 mutantsunder various salt stress conditions. (A) Permeability of 5-DAG roots to tetrazo-lium red after 4 h incubation. The area showing the difference of the staining inroot tip is indicated by the arrow. The more intensive the color, the higherpermeability to the dye. (B) Permeability of mature seeds to tetrazolium red after8 h incubation. (C and D) Percentage of seed germination (C) and seedlingestablishment (D) of WT and rwp1 seeds under increasing salt medium. Seedswere germinated for 14 d. The ratio of root punching out seedlings (i.e., germi-nation) and ratio of seedlings with 2 green cotyledons (i.e., seedling establish-ment) were scored. Values are means of triplicates involving approximately 150seeds for each replicate. Seeds from mutant lines show a reduced germinationrate at approximately 100 to 200 mM of NaCl, KCl, and Na2SO4, and a decreasedestablishment rate at 100 mM of NaCl and KCl.

Gou et al. PNAS � November 3, 2009 � vol. 106 � no. 44 � 18859

PLA

NT

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

May

17,

202

0

Supermix (Bio-Rad), with gene specific primers in an iCycler real-time PCR ma-chine (Bio-Rad) in triplicates using AtTub4 as internal standard (Table S1).

Wall-Bound Phenolic Analysis. We carried out the wall-bound phenolic extrac-tion and analysis as previously described (35). Stems from 8-week-old plantswere collected and used. The detailed methods are described in the SI Text.

Preparation and Depolymerizing Analysis of Suberin. We extracted suberin fromroots and seeds via the method of Franke et al. (33) and the procedure of Molinaet al. (12), respectively, but with minor modifications. Roots of 8-week-old plantsand the mature seeds, harvested and dried for 2 weeks, were used. Before thesuberin depolymerization, the cell wall pectin and xylan were sequentially sep-arated from the extract-free residuals and saponified. The remaining suberinpreparations were depolymerized with BF3/methanol trans-esterification and/orthioacidolysis method. The detailed procedures are described in the SI Text.

To quantify the content of the suberin phenolics, the UV-absorptive area ofparticular peak from each sample first was normalized with that of the internalstandard, thencalibratedwiththestandardcurvesforferulate,p-coumarate,andsinapate established in the same HPLC running using a series of concentrations ofauthentic chemicals.

Gene Isolation and Phylogenetic Analysis. Full-length cDNA of At5g41040 genewas amplified from the total RNAs of A. thaliana (Col0) by RT-PCR. Sequencealignment and phylogenetic analyses were performed as described (32).

Protein Purification, Enzymatic Assay, and Kinetic Analysis. The open-readingframe of the Arabidopsis gene was sub-cloned onto the pHis9 protein expressionvector by gateway cloning, and the recombinant protein was produced inBL21(DE3) and purified as described (36). The enzyme activity was first examinedby incubating the recombinant protein with 50 �M feruloyl-CoA and 40 �M16-hydroxypalmitic acid in 0.1 M MES buffer (pH 6.0) for 30 min, using the boiledprotein or the crude extract of E. coli with empty vector as the control.

To determine the optimal temperature and pH for enzyme activity ofAt5g41040, the protein was incubated with 50 �M feruloyl-CoA and 20 �m16-hydroxypalmitic acid in 0.1 M MES buffer (pH 6.0) at different temperatures;or, in 0.1 M MES, NaPi and CAPS buffers with different pH at 10 °C for 30 min.

To determine its substrate specificity, 50 �M feruloyl-CoA and 20 �M of

acceptor substrates were mixed in 0.1 M Pi buffer (pH 7.5) and kept at 10 °C for5 min before adding 5 �g purified protein to initiate the reaction. After 30 minincubation at 10 °C, 50 �L 50% acetonitrile was added to stop the reaction and20 �L reaction mixtures were injected for HPLC analysis. To monitor thepreference for thioesters, we used 16-hydroxypalmitic acid at a fixed concen-tration of 40 �M.

To measure the steady-state kinetics for the acceptor16-hydroxypalmitic acid,we fixed feruloyl-CoA at 100 �M and used 1 �g protein with a series of concen-trations of acceptor substrate. To measure the kinetics for thioesters, 16-hydroxypalmitic acid was fixed at 200 �M, and a series of concentrations offeruloyl-CoA or p-coumaroyl-CoA was used with 1 �g purified protein. Theincubation proceeded for 5 min before adding 50 �L of 50% hot acetonitrile(60 °C), and the mixture was heated at 99 °C for 1 min to fully stop the reactionand to prevent the potential reverse conversion of product by residual enzymeactivity.

Salt Treatment and Dye Staining. We added a series of concentration of saltsto the media and autoclaved them. After spreading the Arabidopsis seeds overthe plates, they were kept in a growth chamber at 22 °C for 14 d, after whichwe measured the percentage of seed germination (i.e., the root punching outseedlings) and the percentage of seedling establishment (i.e., seedlings with2 green cotyledons). For dye staining, we stained both 5-DAG roots fromseedlings on a 1⁄2 MS plate, and mature seeds in 1% tetrazolium red in H2O for4 or 8 h at 30 °C, and then rinsed the material with water before imagingunder a dissection microscope.

For Sudan red staining, seeds were bleached as described by Beisson et al.(17). The dried seeds were stained in a saturated Sudan red solution (in 92%ethanol) for 30 s at 70 °C, rinsed with water, and imaged under a dissectionmicroscope (33).

ACKNOWLEDGMENTS. We thank Drs. John Ralph and Benjamin Burr for theinsightful comments and careful edits on our manuscript. We thank Drs. FangChen and Richard Dixon at the Samuel Roberts Noble Foundation for provid-ing caffeoyl-CoA substrate. This work was initiated by the Department ofEnergy (DOE)/US Department of Agriculture joint Plant Feedstock Genomicsprogram (project Bo-135). The chemical analysis was partially supported byOffice of Basic Energy Science (DOE) project DEAC0298CH10886 (to C.J.L.).

1. Boerjan W, Ralph J, Baucher M (2003) Lignin biosynthesis. Annu Rev Plant Biol54:519–546.

2. Kolattukudy PE (1981) Structure, biosynthesis, and biodegradation of cutin andsuberin. Annual Rev Plant Physiol 32:539–567.

3. Franke R, Schreiber L (2007) Suberin–a biopolyester forming apoplastic plant inter-faces. Curr Opin Plant Biol 10:252–259.

4. Pollard M, Beisson F, Li Y, Ohlrogge JB (2008) Building lipid barriers: Biosynthesis ofcutin and suberin. Trends Plants Sci 13:236–246.

5. Bernards MA (2002) Demystifying suberin. Can J Bot 80:227–240.6. Bernards MA, Lewis NG (1998) The macromolecular aromatic domain in suberized

tissue: A changing paradigm. Phytochem 47:915–933.7. Riley RG, Kolattukudy PE (1975) Evidence for covalently attached p-coumaric acid and

ferulic acid in cutins and suberins. Plant Physiol 56:650–654.8. Bernards MA, Lopez ML, Zajicek J, Lewis NG (1995) Hydroxycinnamic acid-derived

polymers constitute the polyaromatic domain of suberin. J Biol Chem 270:7382–7386.9. Graca J, Santos S (2006) Glycerol-derived ester oligomers from cork suberin. Chem Phys

Lipids 144:96–107.10. Lopes MH, Neto CP, Barros AS, Rutledge D, Delgadillo I, Gil AM (2000) Quantitation of

aliphatic suberin in Quercus suber L. cork by FTIR spectroscopy and solid-state (13)C-NMR spectroscopy. Biopolymers 57:344–351.

11. Graca J, Pereira H (2000) Suberin structure in potato periderm: Glycerol, long-chainmonomers, and glyceryl and feruloyl dimers. J Agric Food Chem 48:5476–5483.

12. Molina I, Bonaventure G, Ohlrogge J, Pollard M (2006) The lipid polyester compositionof Arabidopsis thaliana and Brassica napus seeds. Phytochem 67:2597–2610.

13. Schreiber L, Franke R, Hartmann KD, Ranathunge K, Steudle E (2005) The chemicalcomposition of suberin in apoplastic barriers affects radial hydraulic conductivitydifferently in the roots of rice (Oryza sativa L. cv. IR64) and corn (Zea mays L. cv. Helix).J Exp Bot 56:1427–1436.

14. Franke R, et al. (2009) The DAISY gene from Arabidopsis encodes a fatty acid elongasecondensing enzyme involved in the biosynthesis of aliphatic suberin in roots and thechalaza-micropyle region of seeds. Plant J 57:80–95.

15. Xiao F, et al. (2004) Arabidopsis CYP86A2 represses Pseudomonas syringae type IIIgenes and is required for cuticle development. EMBO J 26:2158–2168.

16. Li Y, Beisson F, Koo AJ, Molina I, Pollard M, Ohlrogge J (2007) Identification ofacyltransferases required for cutin biosynthesis and production of cutin with suberin-like monomers. Proc Natl Acad Sci USA 104:18339–18344.

17. Beisson F, Li Y, Bonaventure G, Pollard M, Ohlrogge JB (2007) The acyltransferase GPAT5is required for the synthesis of suberin in seed coat and root of Arabidopsis. Plant Cell19:351–368.

18. Cottle W, Kolattukudy PE (1982) Biosynthesis, deposition, and partial characterizationof potato suberin phenolics. Plant Physiol 69:393–399.

19. Bernards MA, Razem FA (2001) The poly(phenolic) domain of potato suberin: Anon-lignin cell wall bio-polymer. Phytochem 57:1115–1122.

20. Lotfy S, Javelle F, Negrel J (1996) Purification and characterization of hydroxycin-namoyl-coenzyme A: �-hydroxypalmitic acid O-hydroxycinnamoyltransferase fromtobacco (Nicotiana tabacum L.) cell-suspension cultures. Planta 199:475–480.

21. Lotfy S, Negrel J, Javelle F (1994) Formation of �-feruloyloxypalmitic acid by an enzymefrom wound-healing potato tuber discs. Phytochem 35:1419–1424.

22. Negrel J, Lotfy S, Javelle F (1995) Modulation of the activity of two hydroxycinnamoyl-transferases in wound-healing potato tuber discs in response to pectinase or abscisicacid. J Plant Physiol 146:318–322.

23. Cottle W, Kolattukudy PE (1982) Abscisic acid stimulation of suberization : Induction ofenzymes and deposition of polymeric components and associated waxes in tissuecultures of potato tuber. Plant Physiol 70:775–780.

24. Lotfy S, Javelle F, Negrel J (1995) Distribution of hydroxycinnamoyl-CoA: �-hydroxyp-almitic acid O-hydroxycinnamoyltransferase in higher plants. Phytochem 40:389–391.

25. St-Pierre B, De Luca V (2000) Evolution of acyltransferase genes: Origin and diversifi-cation of the BAHD superfamily of acyltransferases involved in secondary metabolism.Recent Advances in Phytochemistry: Evolution of metabolic pathways, eds Romeo JT,Ibrahim RK, Varin L, De Luca V (Elsevier, Amsterdam), Vol 34, pp 285–315.

26. D’Auria JC (2006) Acyltransferases in plants: A good time to be BAHD. Curr Opin PlantBiol 9:331–340.

27. St-Pierre B, Laflamme P, Alarco A-M, De Luca V (1998) The terminal O-acyltransferaseinvolved in vindoline biosynthesis defines a new class of proteins responsible forcoenzymes A-dependent acyltransfer. Plant J 14:703–713.

28. Yang Q, et al. (2004) Analysis of the involvement of hydroxyanthranilate hydroxycin-namoyltransferase and caffeoyl-CoA 3-O-methyltransferase in phytoalexin biosynthe-sis in oat. Mol Plant Microbe Interact 17:81–89.

29. Walker K, Long R, Croteau R (2002) The final acylation step in taxol biosynthesis:Cloning of the taxoid C13-side-chain N-benzoyltransferase from Taxus. Proc Natl AcadSci USA 99:9166–9171.

30. Dudareva N, D’Auria JC, Nam KH, Raguso RA, Pichersky E (1998) Acetyl-CoA:benzylal-cohol acetyltransferase–an enzyme involved in floral scent production in Clarkiabreweri. Plant J 14:297–304.

31. Luo J, et al. (2009) A novel polyamine acyltransferase responsible for the accumulationof spermidine conjugates in Arabidopsis seed. Plant Cell 21:318–333.

32. Yu X-H, Gou J-Y, Liu CJ (2009) BAHD superfamily of acyl-CoA dependent acyltrans-ferases in Populus and Arabidopsis: Bioinformatics and gene expression. Plant Mol BiolDOI 10.1007/s11103–009-9482–1.

33. Franke R, et al. (2005) Apoplastic polyesters in Arabidopsis surface tissues–a typicalsuberin and a particular cutin. Phytochem 66:2643–2658.

34. Stockigt J, Zenk MH (1975) Chemical syntheses and properties of hydroxycinnamoyl-coenzyme A derivatives. Z Naturforschung 30C:352–358.

35. Gou JY, Park S, Yu XH, Miller LM, Liu CJ (2008) Compositional characterization andimaging of ‘‘wall-bound’’ acylesters of Populus trichocarpa reveal differential accu-mulation of acyl molecules in normal and reactive woods. Planta 229:15–24.

36. Yu XH, Liu CJ (2006) Development of an analytical method for genome-wide functionalidentification of plant acyl-CoA dependent acyltransferases. Anal Biochem 358:146–148.

18860 � www.pnas.org�cgi�doi�10.1073�pnas.0905555106 Gou et al.

Dow

nloa

ded

by g

uest

on

May

17,

202

0