Ptr-miR397a is a negative regulator of laccase genesaffecting lignin content in Populus trichocarpaShanfa Lua,1, Quanzi Lib,c,1, Hairong Weid, Mao-Ju Change, Sermsawat Tunlaya-Anukitb, Hoon Kimf, Jie Liub,Jingyuan Songa, Ying-Hsuan Sung, Lichai Yuana, Ting-Feng Yehe, Ilona Peszlenh, John Ralphf, Ronald R. Sederoffb,2,and Vincent L. Chiangb,2

aInstitute of Medicinal Plant Development, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing 100193, China; bForestBiotechnology Group, Department of Forestry and Environmental Resources, North Carolina State University, Raleigh, NC 27695; cCollege of Forestry,Shandong Agricultural University, Taian, Shandong 271018, China; dSchool of Forest Resources and Environmental Science, Michigan Technological University,Houghton, MI 49931; eSchool of Forestry and Resource Conservation, National Taiwan University, Taipei 10617, Taiwan; fDepartment of Biochemistry,Wisconsin Energy Institute and Department of Energy Great Lakes Bioenergy Research Center, University of Wisconsin, Madison, WI 53726; gDepartmentof Forestry, National Chung Hsing University, Taichung 40227, Taiwan; and hDepartment of Forest Biomaterials, North Carolina State University, Raleigh,NC 27695

Contributed by Ronald R. Sederoff, May 10, 2013 (sent for review April 2, 2013)

Laccases, as early as 1959, were proposed to catalyze the oxidativepolymerization of monolignols. Genetic evidence in support of thishypothesis has been elusive due to functional redundancy of lac-case genes. An Arabidopsis double mutant demonstrated theinvolvement of laccases in lignin biosynthesis. We previously iden-tified a subset of laccase genes to be targets of a microRNA(miRNA) ptr-miR397a in Populus trichocarpa. To elucidate the rolesof ptr-miR397a and its targets, we characterized the laccase genefamily and identified 49 laccase gene models, of which 29 werepredicted to be targets of ptr-miR397a. We overexpressed Ptr-MIR397a in transgenic P. trichocarpa. In each of all nine transgeniclines tested, 17 PtrLACs were down-regulated as analyzed by RNA-seq. Transgenic lines with severe reduction in the expression ofthese laccase genes resulted in an !40% decrease in the total lac-case activity. Overexpression of Ptr-MIR397a in these transgeniclines also reduced lignin content, whereas levels of all monolignolbiosynthetic gene transcripts remained unchanged. A hierarchicalgenetic regulatory network (GRN) built by a bottom-up graphicGaussian model algorithm provides additional support for a roleof ptr-miR397a as a negative regulator of laccases for lignin bio-synthesis. Full transcriptome–based differential gene expression inthe overexpressed transgenics and protein domain analyses impli-cate previously unidentified transcription factors and their targets inan extended hierarchical GRN including ptr-miR397a and laccasesthat coregulate lignin biosynthesis in wood formation. Ptr-miR397a,laccases, and other regulatory components of this network may pro-vide additional strategies for genetic manipulation of lignin content.

Lignin, an abundant biological polymer affecting the ecologyof the terrestrial biosphere, is vital for the integrity of plant

cell walls, the strength of stems, and resistance against pests andpathogens (1). Lignin is also a major barrier in the pulping andbiomass-to-ethanol processes (2–4). For extracting cellulose(pulping) or for enzymatic degradation of cellulose for bio-ethanol, harsh chemical or physical treatments are used to re-duce interactions with lignin or other cell wall components (2–4).Reducing lignin content or altering lignin structure to reduce itsrecalcitrance are major goals for more efficient processing.Lignin is polymerized primarily from three monolignol pre-

cursors, p-coumaryl alcohol, coniferyl alcohol, and sinapyl alcohol(1, 5). Over five decades, efforts have beenmade to understand thebiosynthesis of the primary monolignols and to modify the quan-tity or composition of lignin. The polymerization of monolignolsinto a lignin polymer has long been thought to occur through ox-idative polymerization catalyzed by either laccases or peroxidases(6). The mechanisms and specificity of the roles of the oxidativeenzymes in lignin polymerization have been controversial (7).Laccases (EC. 1.10.3.2) are multicopper oxidoreductases.

Plant laccase was the first enzyme shown to be able to polymerizelignin monomers in vitro (6), and the expression of some laccasegenes is closely correlated with lignin deposition in xylem (8–10).

Very few genetic experiments have been published supportingthe involvement of laccases in lignin polymerization. Ranochaet al. (11) isolated five laccase cDNAs (lac1, lac2, lac3, lac90, andlac110) from the stem-differentiating xylem of Populus tricho-carpa. Suppression of the lac3 gene resulted in no significantalteration in lignin content and composition, although an alter-ation of the xylem fiber cell walls was observed in antisensesuppressed lac3, lac90, or lac110 (12). Only recently has geneticdata been obtained implicating laccases in lignin polymerizationin Arabidopsis thaliana. Simultaneous disruption of LAC4 andthe other laccase, LAC17, resulted in a reduction of lignincontent in Arabidopsis stems, whereas single-gene mutations inLAC4 or LAC17 caused only a modest reduction in lignin con-tent (13). Zhou et al. (14) identified MYB58 and MYB63 to betranscriptional activators of lignin biosynthesis and MYB58 di-rectly activates LAC4. In Arabidopsis, there are 17 laccase genes,and 8 are expressed in stems (13). These results suggest func-tional redundancy of laccases in lignin polymerization, requiringmultiple mutations to observe significant effects.Trees are particularly suitable for studying lignin biosynthesis

because of the abundance of lignin in wood and the role of ligninin the structure and physiology of the plant.We used P. trichocarpaas a model woody plant to investigate the laccase gene family andits regulators related to lignin biosynthesis. Previously, we iden-tified a subset of 28 laccase genes to be targets of miR397 (15).MIR397 is a family of small and noncoding microRNAs (miRNAs)conserved in dicots, monocots, and gymnosperms (16). We foundthree MIR397 gene models in the P. trichocarpa genome: Ptr-MIR397a, b, and c (17). Only ptr-miR397a was sufficiently abun-dant to be detected by high throughput sequencing and real-timequantitative reverse transcription PCR (qRT-PCR) (17, 18). All ofthe tissues analyzed had detectable levels of mature ptr-miR397a.The levels in phloem, mature leaves, and stem differentiating xylem(SDX) were higher than those in young leaves, young stems, androots (17). MiR397-directed cleavage of laccase transcripts hasbeen observed in Arabidopsis (19) and P. trichocarpa (15). MiR397could be a negative regulator of lignin content.We characterized the P. trichocarpa laccase gene family and

created transgenics overexpressing in Ptr-MIR397a. Twenty-three SDX-expressed laccases are targets of ptr-miR397a in P.trichocarpa. Overexpression of Ptr-MIR397a in P. trichocarpa

Author contributions: S.L., Q.L., R.R.S., and V.L.C. designed research; S.L., Q.L., H.W., M.-J.C.,S.T.-A., H.K., J.L., J.S., L.Y., and I.P. performed research; S.L., Q.L., H.W., S.T.-A., H.K., Y.-H.S.,J.R., R.R.S., and V.L.C. analyzed data; and S.L., Q.L., H.W., H.K., T.-F.Y., R.R.S., and V.L.C.wrote the paper.

The authors declare no conflict of interest.1S.L. and Q.L. contributed equally to this work.2To whom correspondence may be addressed. E-mail: vchiang@ncsu.edu or ron_sederoff@ncsu.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1308936110/-/DCSupplemental.

10848–10853 | PNAS | June 25, 2013 | vol. 110 | no. 26 www.pnas.org/cgi/doi/10.1073/pnas.1308936110

resulted in a reduction in lignin content, whereas levels of mono-lignol biosynthetic gene transcripts remained unchanged, verify-ing the involvement of laccases in activation of monolignolsfor polymerization. We conclude that ptr-miR397a is a masterregulator of polymerization in lignin biosynthesis. Using geneexpression data and bioinformatic analysis of transcription fac-tors (TFs), we constructed a hierarchical genetic regulatorynetwork including ptr-miR397a, laccases, and associated TFsand their targets.

ResultsGenomewide Characterization of Laccase Genes in P. trichocarpa.Plant laccases, which are members of a large family of multi-copper oxidases, consist of three blue copper protein domainswith signature sequences (20). We used 17 Arabidopsis laccaseprotein sequences to blast the P. trichocarpa genome (v2.2) andfound 49 laccase gene models (PtrLACs) with Cu-oxidasedomains (Table S1) that fall into six clades (Fig. S1). Suppressionof Arabidopsis AtLAC4 and AtLAC17 resulted in reduction oflignin content (13). Seven PtrLACs are closely related toAtLAC4 in clade 1, and 11 are clustered with AtLAC17 in clade 2(Fig. S1). Analysis of laccase mRNA abundance using qRT-PCRrevealed transcripts of many laccase genes in each tissue exam-ined and some show specificity for SDX (Fig. 1). Of a total of 49gene models, transcripts for 30 PtrLACs were detected in SDX,of which 17 are abundant. These 17 belong to clades 1, 2, and 5,where the two lignin-related Arabidopsis laccases (AtLAC4 andAtLAC17) are also clustered. The abundance of multiple lac-cases within SDX suggests functional redundancy of laccases forlignin biosynthesis in wood formation.

Twenty-Nine PtrLACs Are Targets of ptr-miR397a. To determinewhether any of these 49 laccases are regulated by ptr-miR397a,we screened for ptr-miR397a target sequences in the laccasetranscripts using psRNATarget (21). Twenty-nine PtrLACs were

predicted targets (Fig. S2A). The targets of ptr-miR397a are ina sequence encoding a conserved Cu-oxidase domain (Fig. S2A).We randomly selected 7 PtrLACs (1, 2, 13, 18, 21, 26, and 30)from the 29 predicted targets and characterized them by the

Fig. 1. Expression of PtrLAC genes in tissues of P. trichocarpa. Fold changes of transcript levels in young leaves (L1), mature leaves (L2), young stems (S), SDX(X), phloem (P), and young roots (R) are shown. Transcript levels in young leaves were arbitrarily set to 1, except PtrLAC13, PtrLAC18, PtrLAC26, and PtrLAC37,whose levels in young roots were set to 1.

Fig. 2. Ptr-miR397a targets 29 PtrLACs for cleavage and overexpression inP. trichocarpa. (A) Experimentally validated targets of ptr-miR397a. Thecleavage sites were determined by the modified 5! RNA ligase-mediatedRACE. PtrLAC sequence of each complementary site from 5! to 3! and ptr-miR397a sequence from 3! to 5! are shown. Watson–Crick pairing (verticaldashes) andG:Twobble pairing (circles) are indicated. Vertical arrows indicatethe 5! termini of miRNA-guided cleavage products, as identified by 5! RACE,with the frequency of clones shown. (B) The 2XCaMV35Sp::MIR397a plasmidused for overexpression of Ptr-MIR397a in P. trichocarpa. (C) Ptr-MIR397aexpression in the three WT plants (W1–W3) and nine transgenic lines.

Lu et al. PNAS | June 25, 2013 | vol. 110 | no. 26 | 10849

PLANTBIOLO

GY

modified 5!-rapid amplification of cDNA ends (RACE) (22). Allseven were authentic targets of ptr-miR397a (Fig. 2A). This ex-periment verified the computational prediction and supporteda regulatory role of ptr-miR397a in suppressing these laccases.The 29 PtrLACs targeted by ptr-miR397a belong to clades 1, 2, 4,and 5 (Fig. S1). No predicted targets were found in clades 3 and6. The seven P. trichocarpa homologs of AtLAC4 in clade 1, andthe 11 P. trichocarpa homologs of AtLAC17 in clade 2 are allpredicted targets of ptr-miR397a. Among the 17 PtrLACsabundantly expressed in SDX, 13 are targets of ptr-miR397a(Fig. S1), suggesting a role of ptr-miR397a in secondary cellwall biosynthesis.

Cloning and Overexpression of the Full-Length Ptr-MIR397a. To fur-ther elucidate the roles of ptr-miR397a and its target PtrLACs,we cloned the full-length Ptr-MIR397a cDNA using 5! and 3!RACE PCR, followed by full cDNA amplification. The primaryPtr-MIR397a transcript is 1,387 bases. The mature ptr-miR397aand ptr-miR397a* are close to the 5!-end of Ptr-MIR397a (Fig.S2 B and C). We prepared a pBI121-based construct to over-express Ptr-MIR397a under the control of a double CauliflowerMosaic Virus (CaMV) 35S promoter (2!CaMV35Sp::MIR397a;Fig. 2B). The construct was introduced into P. trichocarpa usingAgrobacterium (23). Thirty-two transgenic lines were obtained.Significant overexpression of Ptr-MIR397a was found in all ninerandomly selected lines (Fig. 2C).

Overexpression of Ptr-MIR397a Reduced Lignin Content. All plantsfrom the nine overexpression transgenic lines were maintained

in a greenhouse and harvested at the age of 6 mo. No growth dif-ferences were observed compared withWT plants, and SDX cells didnot show any anatomical difference compared with WT (Fig. S3).Based on the highest level of ptr-miR397a, four lines were selectedfor analysis of lignin content. Lignin content of the four lines,397a-1, 397a-6, 397a-9, and 397a-10, showed reduction in Klasonlignin content ranging from 12% to 22% (P < 0.005; Table 1).Significant increases (P < 0.05) in xylan content and decreases ofmannan and uronic acid content were also found (Table 1). Otherhemicelluloses, including rhamnan, arabinan, and galactan, as well ascellulose, showed no significant changes (Table 1).

Lignin Composition and Structure. To examine whether lignincomposition was altered in plants overexpressing Ptr-MIR397a,woody stem tissues were analyzed using nitrobenzene oxidation(Table 2). P. trichocarpa lignin is rich in syringyl (S) subunits, repre-sented by the oxidation products syringaldehyde and syringic acid.The guaiacyl (G) lignin, represented by vanillin and vanillic acid, wasabout half of the S content, whereas the content of p-hydroxyphenyl(H) subunits in lignin, represented by p-hydroxybenzaldehyde,was low (0.2%), as expected. Overexpression transgenics showedan S/G ratio of 2.2 compared with 2.1 in WT. The difference wasnot significant.Lignin structural components and linkages were determined

using NMR spectroscopy on lignins remaining after cellulolyticenzyme digestion (24). Three transgenic lines, 397a-1, 397a-9,and 397a-10, covering the range of the observed lignin reductionwere analyzed and compared with WT. Consistent with theresults from nitrobenzene oxidation, the stem wood was rich in S,

Table 1. Composition of WT and transgenic plants

WT1 WT2 397-1 397-6 397-9 397-10 Prob > jtj

Lignin*Klason 20.5 ± 0.0 20.0 ± 1.2 17.8 ± 0.4 16.2 ± 0.0 15.8 ± 0.1 16.2 ± 0.7 0.005Acid-soluble 3.3 ± 0.1 3.9 ± 0.2 4.7 ± 0.4 5.0 ± 0.0 4.4 ± 0.2 4.6 ± 0.4 0.015Total lignin 23.7 ± 0.1 23.9 ± 1.4 22.5 ± 0.0 21.2 ± 0.1 20.1 ± 0.1 20.8 ± 1.0 0.022

Polysaccharide*Rhamnan 0.4 ± 0.0 0.5 ± 0.2 0.6 ± 0.1 0.5 ± 0.0 0.4 ± 0.0 0.6 ± 0.0 0.391Arabinan 0.5 ± 0.0 0.4 ± 0.0 0.6 ± 0.1 0.4 ± 0.0 0.4 ± 0.0 0.5 ± 0.0 0.765Xylan 16.8 ± 0.1 15.8 ± 0.0 17.9 ± 0.9 18.1 ± 0.0 17.2 ± 0.2 17.3 ± 0.3 0.043Mannan 2.9 ± 0.2 3.1 ± 0.3 2.8 ± 0.3 2.6 ± 0.0 2.6 ± 0.0 2.6 ± 0.1 0.023Galactan 1.5 ± 0.3 1.7 ± 0.3 1.2 ± 0.2 1.4 ± 0.0 0.8 ± 0.0 1.0 ± 0.0 0.070Glucan 38.3 ± 0.9 42.2 ± 1.5 43.2 ± 1.7 41.1 ± 0.6 43.8 ± 0.1 44.7 ± 1.2 0.149Uronic acid 10.9 ± 0.1 11.2 ± 0.7 8.2 ± 0.7 8.6 ± 0.9 8.7 ± 0.4 9.2 ± 0.8 0.002

Values are means ± SE (n = 2 for lignin and polysaccharide analysis). Prob > jtj value in bold shows significantchanges using JMP analysis (P < 0.05).*Values are expressed as weight percent based on vacuum-dried extractive free wood weight.

Table 2. Lignin composition by nitrobenzene oxidation

Mol %* WT1 WT2 397-1 397-6 397-9 397-10

p-Hydroxybenzaldehyde 0.5 ± 0.0 0.4 ± 0.1 0.9 ± 0.0 0.9 ± 0.0 0.9 ± 0.0 0.9 ± 0.0p-Hydroxybenzoic acid — — — — — —

Vanillin 17.3 ± 0.9 16.5 ± 2.2 16.7 ± 0.2 17.0 ± 0.9 16.3 ± 0.2 17.2 ± 0.0Vanillic acid 0.5 ± 0.0 0.3 ± 0.1 0.2 ± 0.2 — 0.6 ± 0.0 0.6 ± 0.0Syringaldehyde 34.7 ± 0.5 34.1 ± 3.0 36.5 ± 0.6 37.2 ± 2.2 34.9 ± 1.1 36.8 ± 0.4Syringic acid 2.0 ± 0.2 1.5 ± 0.1 1.4 ± 0.6 1.1 ± 0.1 2.3 ± 0.4 2.7 ± 0.1H† 0.5 ± 0.0 0.4 ± 0.1 0.9 ± 0.0 0.9 ± 0.0 0.9 ± 0.0 0.9 ± 0.0V‡ 17.8 ± 0.9 16.9 ± 2.2 16.9 ± 0.1 17.0 ± 0.9 16.9 ± 0.2 17.8 ± 0.0S§ 36.7 ± 0.3 35.6 ± 3.1 37.8 ± 1.2 38.3 ± 2.3 37.3 ± 1.5 39.4 ± 0.3S/V ratio 2.1 ± 0.1 2.1 ± 0.1 2.2 ± 0.1 2.2 ± 0.0 2.2 ± 0.1 2.2 ± 0.0H/V ratio 0.03 ± 0.00 0.02 ± 0.00 0.05 ± 0.00 0.05 ± 0.00 0.05 ± 0.00 0.05 ± 0.00

—, below detection limits.*Results are means ± SE (n = 2), assuming that lignin’s C9 molecular mass is 210 g/mol.†Sum of p-hydroxybenzaldehyde and p-hydroxybenzoic acid.‡Sum of vanillin and vanillic acid.§Sum of syringaldehyde and syringic acid.

10850 | www.pnas.org/cgi/doi/10.1073/pnas.1308936110 Lu et al.

lower in G, and very low in H (Fig. 3 A and B; Table S2). S/Gratios in the aromatic region of 397a-1, 397a-9, and 397a-10lignin were 2.19, 3.03, and 2.56, respectively, compared with 2.09in WT (Table S2). We examined the side chain region (Fig. 4 Aand B), which generally substantiates the H/G/S ratio changes inthe aromatic region and provides details regarding the bondingtypes and distribution of interunit linkages in the lignin structure(25, 26). Most of the lignin linkage types were nicely resolvedin the 2D heteronuclear single-quantum correlation (HSQC)spectrum for all samples (Table S2). The compositional ratios ofthe side chains of lignin in the transgenics were almost identicalto those in WT plants. Only minor structural changes were noticedfrom the relative intensity estimation (Table S2). The !-aryl etherunits A in transgenics were slightly increased 3.4% (Table S2),consistent with the marginally increased S levels revealed by thecorrelations in the aromatic region. Syringyl units have a higherprobability of !-aryl ether coupling in the growing lignin polymer(7, 26). Other structures, such as phenylcoumarans (B), resinols(C), and spirodienones (SD), were relatively decreased in thetransgenics; phenylcoumarans require participation by coniferylalcohol and are therefore likely lower due to higher S/G. Over-expression of PtrMIR397a caused only slight changes in lignincomposition and structure.

Down-Regulation of PtrLAC Transcript Abundance by Overexpressionof Ptr-MIR397a. RNA-seq of the SDX from nine transgenic linesand three WT trees was carried out to examine the changes in thetranscriptome caused by Ptr-MIR397a overexpression. Differentially

expressed genes (DEGs) were identified using edgeR (27) bycomparing the transcript abundance of each gene in the transgeniclines and theWT trees. We identified a total of 459 DEGs, with 289down-regulated and 170 up-regulated. Thirty-four laccase genes ofthe 49 laccase gene models in P. trichocarpa were expressed in theWT SDX (Table S3), consistent with our qRT-PCR analysis (Fig.1). The expression of 17 of the 34 SDX-expressed laccase genes wasdown-regulated by 32–86% compared with WT (Table S3), ofwhich 5 are homologs of AtLAC4 and 6 are homologs of AtLAC17.Consistent with the reduction of laccase gene expression, the totallaccase activity of SDX proteins in 397-3, 397-6, and 397-26 wasreduced by !40% compared with WT (Fig. 5). No change intranscript abundance was observed for all 20 SDX monolignolgenes (28), indicating that the lignin content reduction is specificand caused by down-regulation of laccases.

Specific Domain Families in the Xylem Transcriptome Respond to Ptr-MIR397a Overexpression. To identify overrepresented gene familiesamong the DEGs, we performed a domain enrichment analysis(29). Several protein families were overrepresented, including 20laccases (types 1, 2, and 3 multicopper oxidases), 7 peroxidases,and 4 chalcone/stilbene synthases (Table S4).

Overexpression of ptr-miR397a Affects Expression of Network Genes.Ptr-MIR397a is part of a hierarchical network controlling woodformation. Overexpression of Ptr-MIR397a combined with RNA-seq may identify downstream components of a network. Wedeveloped and used a unique bottom-up graphic Gaussian model(GGM) algorithm to construct a causal hierarchical genetic reg-ulatory network (GRN) from DEGs. The bottom-up algorithm

ppm 8.0 7.5 6.5 6.0

130

125

120

115

105

100

7.0

110

ppm 8.0 7.5 6.5 6.0

130

125

120

115

105

100

7.0

110

OHO O

OHOMe

OMe

1

O SDspirodienone

OH

1

X1cinnamyl alcohol

(endgroup)

OH

1

X2cinnamaldehyde

(endgroup)

Pyridine(solvent)

O R

OMeMeO

26S S´

SSyringyl

O R

OMe

2

5

6G

GGuaiacyl

O R

OMeMeO

O

26

S´Syringyl

26

OR

2635 H

Hp-Hydroxyphenyl

O

OH

O

26

35PB

PBp-Hydroxybenzoate

B 397-1

A WT1

G2

PB2/6

H2/6 H2/6

G5 + G6

(H3/5 & PB3/5)

S2/6

S´2/6

Pyridine

X1 X1

X2

X26 X26

S: 67.5%G: 32.3%H: 0.2%PB: 2.5%S/G: 2.09

SD2´

SD6´

G2

PB2/6

G5 + G6

(H3/5 & PB3/5)

S2/6

S´2/6

Pyridine

X1 X1

X2

S: 68.5%G: 31.3%H: 0.2%PB: 2.1%S/G: 2.19

SD2´

SD6´

Fig. 3. HSQC NMR spectra of CELs from Ptr-MIR397a–overexpressingtransgenic wood. Lignin aromatic region is shown for (A) WT1 and (B) 397-1.Correlations from the various aromatic ring unit types are well dispersed andcan be categorized as the core lignin units (S, syringyl; G, guaiacyl; H,p-hydroxyphenyl), as well as from the p-hydroxybenzoate PB units known toacylate lignin side chains.

Lignin sidechain region

OHO

HO

OMe4

1

A-aryl ether ( -O-4)

Polysaccharide, etc.

Methoxyl

A-G-aryl ether ( -O-4-H/G)

OHO

HO

OMe

G4

1

A-S-aryl ether ( -O-4-S)

OHO

HO

OMe

MeOS

41

X1cinnamyl alcohol

(endgroup)

OH

1

O

HO OMe5

4

1

Bphenylcoumaran ( -5)

O

O

1

1

Cresinol ( - )

O

OHO O

OHOMe

OMe

1

SDspirodienone ( -1)

26

5.56.0 5.0 4.5 4.0 3.5 3.0 2.5

80

90

70

60

50

5.56.0 5.0 4.5 4.0 3.5 3.0 2.5

80

90

70

60

50

SD

SD

SD SD

Methoxyl

A

A

B

B

B

X1

A-G

A-SC

C

C

SD

SD

Methoxyl

A

A

B

B

B

X1

A-G

A-SC

C

C

A: 86.5%B: 3.7%C: 7.3%SD: 2.4%X1: 5.3%

A: 88.0%B: 3.1%C: 6.6%SD: 2.3%X1: 4.8%

B 397-1

A WT1

Fig. 4. HSQC NMR spectra of CELs from Ptr-MIR397a–overexpressingtransgenic wood. The lignin side chain region is shown for WT1 (A) and 397-1(B). The major interunit structural units A, B, C, SD, as well as end-units X1and X2, are also shown, color coded by their indicated structures.

Lu et al. PNAS | June 25, 2013 | vol. 110 | no. 26 | 10851

PLANTBIOLO

GY

constructs a multilayer GRN starting from the enzyme level andbuilding up. Using these 19 significantly down- and up-regulatedlaccase genes, identified by DEG, domain-enrichment, and alltranscription factors including ptr-miR397a as input for thebottom up algorithm, a GRN of three layers was obtained (Fig.6). Ptr-miR397 is at the second layer and directly controlsthe laccase (bottom). The specific causal relationships of ptr-miR397a and their target laccase genes were accurately pre-dicted when 19 laccase genes and 15 others (encoding chalcone/stilbene synthase and peroxidase protein domains) were used asthe bottom/first layer in an implementation of the GGM. Theregulatory genes in the second and third layers were recognizedby the GGM bottom-up algorithm from 1,208 TFs and ptr-miR397a. The GRN indicates the control of 13 laccases and 4peroxidases by ptr-miR397a and TFs. No direct regulatory rela-tionships between ptr-miR397a and the other genes used as theimplementation in the bottom layer were identified. Among the 13PtrLACs identified in the bottom layer, 12 PtrLACs are directlyregulated by ptr-miR397a, and these 12 PtrLACs belong toclades 1, 2, and 5 (Fig. S1). PtrLAC43, belonging to clade 6 (Fig.S1), is not directly regulated by ptr-miR397a but is regulated byother TFs in the second layer. These results are consistent withthe computational inference that no predicted targets were foundin clades 3 and 6. Adding 44 genes involved in lignocellulosicbiosynthesis in the input failed to identify any causal relation-ships between ptr-miR397 and these genes, indicating that theregulatory relationships between ptr-miR397 and the 12 laccasegenes are specific.In the GRN, we identified several wood formation–related

TFs present in the upper layers. Two MYB genes, PtrMYB021

(30) and PtrMYB52 (Table S5), are present in second layer, andtwo NAC genes, PtrSND1-A2 and PtrVND6-C2 (30), are in thethird layer. PtrMYB021 is the poplar ortholog of ArabidopsisAtMYB46, known to up-regulate AtLAC10 and other laccasegenes through an eight-nucleotide core motif (RKTWGGTR)(31, 32), further validating the GRN. We also found some geneswith unknown functions in the GRN. For example, PtrUNE12(Table S5), encoding a basic helix–loop–helix (bHLH) TF, hasan inferred interaction with PtrSND1-A2, 10 laccase genes, and 4peroxidase genes (Table S5), suggesting it is an important reg-ulator in lignin biosynthesis. All of the proposed interactions(Fig. 6) need experimental validation.

DiscussionThe involvement of laccases in lignin polymerization was pro-posed more than five decades ago based on the ability ofa cambial extract containing laccase activities from spruce toproduce a lignin-like polymer (6). Peroxidase activities were alsofound in the cambial sap, and therefore peroxidases were alsoimplicated in lignin polymerization (6). Since 1959, evidence hasimplicated both laccases and peroxidases in lignin polymeriza-tion. Genetic data to demonstrate an essential role for eitherlaccases or peroxidases in lignin biosynthesis have been difficultto obtain due to the large number of functionally redundantgenes in these gene families (13, 33). Transgenic suppression ofperoxidases in tobacco and aspen has shown effects on lignincontent (34, 35). Transgenic suppression of laccases in poplar(12) showed effects on phenolic metabolites but no effect onlignin content. In a recent publication, a double mutant of twolaccase genes affected lignin content. Each single mutant hadslightly reduced levels of lignin, but the double mutant reducedlignin content as much as 40%, providing clear genetic evidenceto support the role of laccases (13).Transgenic P. trichocarpa plants overexpressing Ptr-MIR397a

resulted in a reduction in Klason lignin content up to 22% (Table1), consistent with the results from the Arabidopsis mutants (13).Nitrobenzene oxidation and NMR showed few if any changes inlignin composition and structure. The transcript levels of all ofthe monolignol pathway genes (not predicted targets of ptr-miR397a) were not significantly affected, providing evidence forthe specificity of the laccase effect. The supply of monolignols inthe Ptr-MIR397a transgenics can be expected to be essentially thesame as WT. These results verify the involvement of laccases inlignin polymerization and suggest ptr-miR397a to be a masterregulator in the process of lignin polymerization. Ptr-miR397a isthe first example of a miRNA regulating lignin biosynthesis forwood formation.

Fig. 5. Quantification of laccase activity of purified SDX proteins from oneWT (WT1) and three transgenic lines (397-3, 397-6, and 397-26). Data rep-resent means ± SD (n = 5). **P < 0.01.

PtrSND1-A2

PtrVND6-C2

PtrMYBCDC5

PtrARF8

PtrMYB42

PtrMYB48

PtrARF7

PtrHB-1-1

PtrHB-1Ptr

ABF3

PtrLAC43

PtrPO6

PtrLAC41

PtrPO2

PtrPO3

PtrLAC49

PtrPO34

PtrNF-YC9Ptr

MYB021ptr-

miR397aPtr

MYB73Ptr

MYB50Ptr

UNE12PtrHSFB4

PtrARF2

PtrDAG1

PtrMYB52

PtrLAC26

PtrLAC23

PtrLAC2

PtrLAC24

PtrLAC40

PtrLAC30

PtrLAC15

PtrLAC14

PtrLAC18

PtrLAC1

Fig. 6. A three-layer genetic regulatory network(GRN) indicating the control of 13 laccase genes and4 peroxidase genes. On the bottom layer, 12 laccasegenes are shown as direct targets (red lines) of ptr-miR397a (yellow). Five of the laccases genes (green)have been validated as targets for ptr-miR397a by5! RACE. One laccase PtrLAC43 is implicated in thenetwork by putative interactions with several TFs inlayer 2. Four peroxidases are included in the GRNalso because of proposed interaction with TFs inlayer 2. The thickness of the connecting linesreflects the calculated relative strength of the pro-posed interaction. The remainder of layer 2 iscomposed of TFs selected by the GGM algorithmbased on putative interactions with genes in thebottom layer. Similarly, layer 3 (top layer) representsTFs derived by GGM based on interactions with theTFs in layer 2 as described in SIMaterials andMethods.See Tables S1 and S5 for gene models.

10852 | www.pnas.org/cgi/doi/10.1073/pnas.1308936110 Lu et al.

Laccase is a blue copper oxidase, a metalloenzyme with fourcopper ions required for utilization of dioxygen as a substrate(36). Cu+ is highly toxic to living cells (37). Organisms need tomaintain copper homeostasis for many processes such as energytransduction, iron mobilization, and oxidative stress responses(38). High copper levels cause lignin accumulation in plants (39,40), and copper deficiency decreases plant lignin content (41).The underlying mechanisms are not known. If copper-bindingprotein laccases were reduced in Ptr-MIR397a overexpressiontransgenics, then copper could accumulate. To alleviate the ef-fect of higher copper concentration, plant cells accumulate hy-drogen peroxide and increase activity of antioxidant enzymes,including superoxide dismutase, catalase, ascorbate peroxidase,glutathione reductase, and particularly peroxidase (39, 40). BothDEG and domain enrichment analysis of the Ptr-MIR397a–overexpressing transgenics showed seven peroxidase genes weresignificantly up-regulated (Table S4). Alternatively, the relativeabundance of laccases and peroxidases may be reciprocally reg-ulated for lignin polymerization, further contributing to func-tional redundancy of enzymes for oxidative polymerization.Redundancy of oxidative enzymes is likely the reason for themaximum of 22% lignin content reduction when 17 laccasesgenes are down-regulated by 32–86%.GRNs resulting from the bottom-up GGM algorithm con-

firmed that ptr-miR397a is a negative regulator of manymembers of the laccase gene family. The GRN indicates thatoverexpression of Ptr-MIR397a leads to a coordinated repression of

laccases and up-regulation of peroxidases in the bottom layer ofthe network (Fig. 6). The TFs in the network regulating lac-cases and peroxidases indicate that multiple regulators are in-volved in overlapping control of specific gene regulation.Future studies should investigate the consequences of thisregulation and the extent of its indirect effects. The networksuggests several additional targets including ptr-miR397a andlaccases for genetic manipulation of lignin formation, poten-tially leading to improved materials for pulp/paper and biofuelproduction.

Materials and MethodsPlant materials, qRT-PCR, computational prediction and experimental vali-dation of ptr-miR397a targets followed those of previous publications (22,23). Ptr-MIR397a cDNA cloning, vector construction and transformation,wood chemical composition and lignin composition analysis, NMR samplepreparation and analysis, and RNA-seq analysis are described in detail in SIMaterials and Methods. Primers used for qRT-PCR and validation of ptr-miR397a targets are listed in Tables S6 and S7.

ACKNOWLEDGMENTS. This work was supported by grants from NationalScience Foundation Plant Genome Research Program Grant (DBI-0922391) toV.L.C.; the National Key Basic Research Program of China (973 program)(2012CB114502) to S.L.; the National Natural Science Foundation of China(31070534) to L.Y.; the DOE Great Lakes Bioenergy Research Center (DOEOffice of Science BER DE-FC02-07ER64494) to J.R. and H.K., and the NorthCarolina State University Forest Biotechnology Industrial Research Consor-tium (grant no. 556051) to Q.L. and J.L.

1. Sarkanen K (1971) Lignin, Occurrence, Formation, Structure and Reactions, edsSarkanen KV, Kudwig CH (Wiley-Interscience, New York), pp 95–163.

2. Chen F, Dixon RA (2007) Lignin modification improves fermentable sugar yields forbiofuel production. Nat Biotechnol 25(7):759–761.

3. Chiang VL (2002) From rags to riches. Nat Biotechnol 20(6):557–558.4. Sarkanen KV (1976) Renewable resources for the production of fuels and chemicals.

Science 191(4228):773–776.5. Higuchi T (2003) Pathways for monolignol biosynthesis via metabolic grids: Coniferyl

aldehyde 5-hydroxylase, a possible key enzyme in angiosperm syringyl lignin bio-synthesis. Proc Jpn Acad Ser B 79(8):227–236.

6. Freudenberg K (1959) Biosynthesis and constitution of lignin. Nature 183(4669):1152–1155.

7. Ralph J, et al. (2004) Lignins: Natural polymers from oxidative coupling of 4-hydrox-yphenyl-propanoids. Phytochem Rev 3(1–2):29–60.

8. Dean J, Eriksson K (1994) Laccase and the deposition of lignin in vascular plants.Holzforschung 48(Suppl):21–33.

9. Sato Y, Wuli B, Sederoff Rs Whetten R (2001) Molecular cloning and expression ofeight laccase cDNAs in loblolly pine (pinus taeda). J Plant Res 114(2):147–155.

10. Bao W, O’malley DM, Whetten R, Sederoff RR (1993) A laccase associated with lig-nification in loblolly pine xylem. Science 260(5108):672–674.

11. Ranocha P, et al. (1999) Biochemical characterization, molecular cloning and ex-pression of laccases—a divergent gene family—in poplar. Eur J Biochem 259(1–2):485–495.

12. Ranocha P, et al. (2002) Laccase down-regulation causes alterations in phenolic me-tabolism and cell wall structure in poplar. Plant Physiol 129(1):145–155.

13. Berthet S, et al. (2011) Disruption of LACCASE4 and 17 results in tissue-specific al-terations to lignification of Arabidopsis thaliana stems. Plant Cell 23(3):1124–1137.

14. Zhou J, Lee C, Zhong R, Ye ZH (2009) MYB58 and MYB63 are transcriptional activatorsof the lignin biosynthetic pathway during secondary cell wall formation in Arabi-dopsis. Plant Cell 21(1):248–266.

15. Lu S, Sun YH, Chiang VL (2008) Stress-responsive microRNAs in Populus. Plant J 55(1):131–151.

16. Kozomara A, Griffiths-Jones S (2011) miRBase: Integrating microRNA annotation anddeep-sequencing data. Nucleic Acids Res 39(Database issue):D152–D157.

17. Lu S, Sun YH, Chiang VL (2009) Adenylation of plant miRNAs. Nucleic Acids Res 37(6):1878–1885.

18. Lu S, Yang C, Chiang VL (2011) Conservation and diversity of microRNA-associatedcopper-regulatory networks in Populus trichocarpa. J Integr Plant Biol 53(11):879–891.

19. Jones-Rhoades MW, Bartel DP (2004) Computational identification of plant micro-RNAs and their targets, including a stress-induced miRNA. Mol Cell 14(6):787–799.

20. Zhukhlistova NE, Zhukova YN, Lyashenko AV, Zaitsev VN, Mikhailov AM (2008) Three-dimensional organization of three-domain copper oxidases: A review. Crystallogr Rep53(1):92–109.

21. Dai X, Zhuang Z, Zhao PX (2011) Computational analysis of miRNA targets in plants:Current status and challenges. Brief Bioinform 12(2):115–121.

22. Lu S, et al. (2005) Novel and mechanical stress-responsive MicroRNAs in Populus tri-chocarpa that are absent from Arabidopsis. Plant Cell 17(8):2186–2203.

23. Song J, Lu S, Chen ZZ, Lourenco R, Chiang VL (2006) Genetic transformation ofPopulus trichocarpa genotype Nisqually-1: A functional genomic tool for woodyplants. Plant Cell Physiol 47(11):1582–1589.

24. Balakshin M, Capanema E, Gracz H, Chang HM, Jameel H (2011) Quantification oflignin-carbohydrate linkages with high-resolution NMR spectroscopy. Planta 233(6):1097–1110.

25. Kim H, Ralph J (2010) Solution-state 2D NMR of ball-milled plant cell wall gels inDMSO-d(6)/pyridine-d(5). Org Biomol Chem 8(3):576–591.

26. Ralph J, Landucci LL (2010) Lignin and Lignans: Advances in Chemistry, eds Heitner C,Dimmel DR, Schmidt JA (CRC Press, Boca Raton, FL), pp 137–234.

27. Robinson MD, McCarthy DJ, Smyth GK (2010) edgeR: A Bioconductor package fordifferential expression analysis of digital gene expression data. Bioinformatics 26(1):139–140.

28. Shi R, et al. (2010) Towards a systems approach for lignin biosynthesis in Populustrichocarpa: transcript abundance and specificity of the monolignol biosyntheticgenes. Plant Cell Physiol 51(1):144–163.

29. Wei H, et al. (2013) Global transcriptomic profiling of aspen trees under elevated[CO2] to identify potential molecular mechanisms responsible for enhanced radialgrowth. J Plant Res 126(2):305–320.

30. Li Q, et al. (2012) Splice variant of the SND1 transcription factor is a dominant neg-ative of SND1 members and their regulation in Populus trichocarpa. Proc Natl Acad SciUSA 109(36):14699–14704.

31. Ko JH, Kim WC, Han KH (2009) Ectopic expression of MYB46 identifies transcriptionalregulatory genes involved in secondary wall biosynthesis in Arabidopsis. Plant J 60(4):649–665.

32. Kim WC, Ko JH, Han KH (2012) Identification of a cis-acting regulatory motif recog-nized by MYB46, a master transcriptional regulator of secondary wall biosynthesis.Plant Mol Biol 78(4-5):489–501.

33. Hiraga S, Sasaki K, Ito H, Ohashi Y, Matsui H (2001) A large family of class III plantperoxidases. Plant Cell Physiol 42(5):462–468.

34. Blee KA, et al. (2003) A lignin-specific peroxidase in tobacco whose antisense sup-pression leads to vascular tissue modification. Phytochemistry 64(1):163–176.

35. Li Y, Kajita S, Kawai S, Katayama Y, Morohoshi N (2003) Down-regulation of an an-ionic peroxidase in transgenic aspen and its effect on lignin characteristics. J Plant Res116(3):175–182.

36. Solomon EI, Sundaram UM, Machonkin TE (1996) Multicopper oxidases and oxy-genases. Chem Rev 96(7):2563–2606.

37. Wong M, Bradshaw A (1982) A comparison of the toxicity of heavy metals, using rootelongation of rye grass, lolium perenne. New Phytol 91(2):255–261.

38. Arredondo M, Núñez MT (2005) Iron and copper metabolism. Mol Aspects Med 26(4-5):313–327.

39. Zhang H, Xia Y, Wang G, Shen Z (2008) Excess copper induces accumulation of hy-drogen peroxide and increases lipid peroxidation and total activity of copper-zincsuperoxide dismutase in roots of Elsholtzia haichowensis. Planta 227(2):465–475.

40. Chen E, Chen Y, Chen L, Liu Z (2002) Effect of copper on peroxidase activity and lignincontent in raphanus sativus. Plant Physiol Biochem 40(5):439–444.

41. Robson A, Hartley R, Jarvis S (1981) Effect of copper deficiency on phenolic and otherconstituents of wheat cell walls. New Phytol 89(3):361–371.

Lu et al. PNAS | June 25, 2013 | vol. 110 | no. 26 | 10853

PLANTBIOLO

GY

Supporting InformationLu et al. 10.1073/pnas.1308936110SI Materials and MethodsPlant Materials. Samples were collected from 6-mo-old Populustrichocarpa (Nisqually-1) WT and transgenic trees maintained ina greenhouse as described (1).

PtrLAC Gene Identification and Phylogenetic Analysis. PtrLACs wereidentified by comparison of 17 Arabidopsis laccase protein se-quences (TAIR, www.arabidopsis.org/) against the P. trichocarpagenome v2.2 using the tBLASTn and BLASTp algorithms (www.ncbi.nlm.nih.gov/BLAST) (2, 3), applying an e-value cutoff of 1e–10

to screen for functional homologs (4, 5). Cu-oxidase domains weredetected by BLAST of the deduced PtrLAC sequences againstthe National Center for Biotechnology Information conserveddomain database (www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi). A phylogenetic tree was constructed using MEGA 5.1 (6).The reliability of branching was assessed by bootstrap resamplingusing 1,000 replicates. Nodes supported by bootstrap probabilities>50% are shown.

Quantitative Real-Time PCR of Laccase Transcript Abundance. TotalRNAs from young leaves, mature leaves, young stems, young roots,stem differentiating xylem (SDX), and phloem were isolated usingcetyltrimethylammonium bromide (CTAB) (7). Gene-specificforward and reverse primers (Table S6) were used for quantitativeRT-PCR (qRT-PCR) to estimate gene-specific laccase transcriptabundance in different tissues as described (8, 9).

Computational Prediction and Experimental Validation of ptr-miR397aTargets. Ptr-miR397a targets were predicted from the P. tricho-carpa genome v2.2 by the psRNATarget server (2, 10). A penaltyscore cutoff of 2.5 calculated as previously described (11) wasapplied to potential target sites in the microRNA (miRNA):mRNA duplexes. Experimental validation of predicted targetswas carried out using a modified RNA ligase-mediated 5!-rapidamplification of cDNA ends (RACE) using the GeneRacer kit(Invitrogen) as described (11). PCRs were performed on cDNAfrom SDX and young roots of 6-mo-old P. trichocarpa plantsusing the GeneRacer 5! primer and gene-specific primers (TableS7). Nested PCRs were performed using the GeneRacer 5!nested primer and the nested gene-specific primers (Table S7).

Isolation of PtrMIR397a cDNA. The transcriptional start site and thepolyadenylation site of PtrMIR397a were determined by 5!- and3!-RACE using GeneRacer. The gene-specific and nested gene-specific primers are 5!-GTGAGCTCCACAGACCACGCAAAT-3! and 5!-CGCAAATGCCTGTTTCGCACCACAA-3! for 5!RACE and 5!-GCATCTTGAAGCATTGGAATCAGAGA-3!and 5!-CAGAGATACGAGGGCCCAATGGTA-3! for 3!RACE,respectively. The PtrMIR397a cDNA was amplified by PCR usingthe gene-specific forward primer (5!-GCGGATCCATTGCA-TTTGAAAGTGAGCTAGCTA-3!; BamHI site is underlined)and the gene-specific reverse primer (5!-GTCCCGGGCTTAG-AACTTACACGGGACCTA-3!; SmaI site is underlined). PCRproducts were cloned into pCR4-TOPO vectors (Invitrogen) andsequence validated.

Vector Construction of CaMV35Sp::MIR397a and Transformation. ThePtrMIR397a cDNA was excised from the pCR4-TOPO vectorwith BamHI/SmaI and then inserted into the modified pBI121with a double CaMV 35S promoter to replace the GUS gene.The construct was mobilized into Agrobacterium strain C58 using

the freeze/thaw method for P. trichocarpa transformation asdescribed (12).

Light Microscopy. The stem transverse section preparation andimage taken were followed as described by Li et al. (1).

qRT-PCR Analysis of Ptr-miR397a. SDX total RNA from three WTtrees and nine transgenic lines of overexpressing PtrMIR397a wasisolated using CTAB (7). Quantitation of ptr-miR397a by qRT-PCR was conducted as described (13).

Wood Chemical Composition. Debarked increment cores (12 cm)were cut into two sections (6 cm), placed in acetone, and held for2 d at room temperature. Acetone was drained and replaced fourtimes at 48-h intervals. After drying, the wood was ground inaWiley mill with a 40-mesh screen. The resulting woodmeals (40–60 mesh) was used for lignin content determination (14) to es-timate both Klason lignin and acid-soluble lignin for total lignincontent using an extinction coefficient of 110 g!1·cm!1 at 205 nm(15). Neutral sugars in the acid-soluble fraction were derivatizedto alditol acetates for quantitation by gas chromatography usinga flame ionization detector (GC-FID; Agilent 7890A) (16). Theuronic acid content was determined by the sulfuric acid-carba-zole method (17). Glucuronolactone was used as the standard,and the optical density was measured at 530 nm.

Nitrobenzene Oxidation Analysis. Nitrobenzene oxidation analysiswas conducted following Katahira and Nakatsubo (18). Extrac-tive-free wood meal was oxidized with nitrobenzene, extractedwith CH2Cl2, trimethylsilylated with N-methyl-N-(trimethylsily1)-trifluoroacetamide (Sigma), and quantified by GC-FID using aDB-5 GC-column.

NMR Sample Preparation and NMR Experiments. Two grams of 40- to60-mesh wood meal samples were ball-milled using a Planetarymicromill Pulverisette 7 (Fritsch, Germany) in a 50-mL ZrO2 bowlwith 18 ZrO2 ball bearings (1 cm diameter) at 600 rpm, with 15 minof rest after every 30 min of milling. The total actual milling time foreach sample was 3 h. Cellulolytic enzyme lignin (CEL) was isolatedaccording to Chang et al. (19). One gram of the ball-milled materialwas suspended in 20 mL 20 mM NaOAc buffer (pH 4.5). Crudecellulase (50 mg) was added, and the reaction was incubated at48 °C for 48 h. The solution was centrifuged, and the residues werewashed with 20 mL distilled water three times and freeze-dried.The CELs were collected directly into NMR tubes (40 mg for

each sample) and dissolved using DMSO-d6/pyridine-d5 (4:1) (20,21). NMR spectra were acquired on a Bruker Biospin Avance700-MHz spectrometer equipped with a cryogenically cooled5-mm TXI 1H/13C/15N gradient probe with inverse geometry(proton coils closest to the sample). The central DMSO solventpeak was used as an internal reference (δC, 39.5; δH, 2.49 ppm).The 13C–1H correlation experiment was an adiabatic hetero-nuclear single-quantum correlation (HSQC) experiment (Brukerstandard pulse sequence “hsqcetgpsisp.2”; phase-sensitive gra-dient-edited-2D HSQC using adiabatic pulses for inversion andrefocusing) (22). HSQC experiments were carried out using thefollowing parameters: acquired from 11.5 to !0.5 ppm in F2 (1H)with 3,366 data points (acquisition time, 200 ms) and 215 to !5ppm in F1 (13C) with 560 increments (F1 acquisition time, 7.2ms) of 32 scans with a 1-s interscan delay; the d24 delay was set to0.86 ms (1/8J, J = 145 Hz). The total acquisition time fora sample was 6 h. Processing used typical matched Gaussianapodization (GB = 0.001, LB = !0.5) in F2 and squared cosine-

Lu et al. www.pnas.org/cgi/content/short/1308936110 1 of 9

bell and one level of linear prediction (32 coefficients) in F1.Volume integration of contours in HSQC plots used Bruker’sTopSpin 3 (Mac version) software and used spectra reprocessedwithout linear prediction.

RNA-seq Data Processing and Differential Expressed Gene Analysis.One microgram SDX total RNA from three WT and ninetransgenic lines was used for the RNA-seq library constructionfollowing the TruSeq RNA Sample Prep v2 LS protocol (Illu-mina). Six RNA-seq libraries were pooled at equal concentrationsfor multiplex sequencing. Libraries were assayed for quality andquantity (NCSU Genomic Sciences Laboratory) and sent forsequencing by BGI on an Illumina HiSEq. 2000. The resultingsequences ("100b) were mapped to the P. trichocarpa genomev2.0, gene annotation v2.2 (www.phytozome.org) using TOPHAT(23). The frequency of raw counts was determined by BEDtools(24) and normalized using the Trimmed mean of M value (TMM)(25). The genes with a count lower than 15 per million per li-brary were filtered out, and differential expressed genes (DEGs)were obtained by pairwise comparison of transgenic and WTlibraries using edgeR/Bioconductor (26). The significance ofDEGs is based on a false discovery rate (FDR) of 0.05. Weinferred the protein domains in the full P. trichocarpa proteomeand in the DEGs using InterproScan (27) and estimated theextent of enrichment of specific domains in the DEGs usinghypergeometric probability (28).

Laccase Enzyme Assays. Laccase purification and enzyme assayfollowed Berth et al. (29) with modificaitons. Three grams of SDXwas homogenized in 15 mL extraction buffer [25 mM Tris, pH 7.0,200 mM CaCl2, 10% (vol/vol) glycerol, 4 μM sodium cacodylate, 1mM PMSF, 1 mg/mL leupeptin, and 1 mg/mL pepstatin A].Proteins were eluted from Concanavalin-A Sepharose with 10 mLelution buffer (29), and 50 μL was used in the enzyme assays (29).

Construction of a Genetic Regulatory Network using a Graphic Gauss-ian Model Algorithm. First, we developed and used a graphicGaussian model (GGM) bottom-up algorithm to calculate rela-tionships of selected genes with specific transcription factors

(TFs) selected from a full genomewide pool of TFs. We con-structed a three-layer genetic regulatory network (GRN) begin-ning with those genes that encode enriched protein domains ofeither laccases (also classified as types 1, 2, and 3 multicopperoxidases), chalcone/stilbene synthases, or peroxidases. Beginningwith these genes as the bottom (first layer) of a hierarchicalnetwork, we used the GGM to select TFs that represent the nextlevel of regulators in the network (second layer). The selected TFswere then used in the same way to select a higher level of reg-ulators, providing a three-layered GRN. The GGM bottom-upalgorithm works as follows. First, it selects a pair of genes, x and yfrom the bottom layer, and then examines their relationshipswith each TF from a genomewide TF pool that includes 1,208TFs and ptr-miR397a. Transcript abundance data were obtainedfrom RNA-seq. Relative abundance of ptr-miR397a was ob-tained by qRT-PCR. We tested if the interaction with any TF (orptr-miR397a) denoted as z, increases the correlation of x and y.We calculated Spearman partial correlations between x and y inthe absence of z, rxyjz = pcor(x, yjz) and compared this with thex and y correlation, rxy, where rxy is the Spearman coefficient (ρ)of variables x and y. If rxy is significant and rxyjz = 0 (insignificant),then z interferes with the correlation of x and y. If rxy is significantand rxyjz is significant, we tested the significance of differenced following (30), where

d= rxy !rxy ! rzxrzy!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!"

1! r2zx#$

1! r2zy%r :

If d is significant, we infer that z interferes (regulates) x and y.After all combinations of x and y in the bottom layer were testedwith all TFs, we sorted all TF genes based on the total numberof times each of them interfered with all pairwise combinationsof genes in the bottom layer. TF genes, with high interferencefrequencies (including ptr-miR397a), were selected to be thesecond-layer regulators. The same procedure was then appliedto the second layer of regulators to obtain the third layersof regulators.

1. Li Q, et al. (2011) Down-regulation of glycosyltransferase 8D genes in Populustrichocarpa caused reduced mechanical strength and xylan content in wood. TreePhysiol 31(2):226–236.

2. Tuskan GA, et al. (2006) The genome of black cottonwood, Populus trichocarpa (Torr.& Gray). Science 313(5793):1596–1604.

3. Altschul SF, et al. (1997) Gapped BLAST and PSI-BLAST: A new generation of proteindatabase search programs. Nucleic Acids Res 25(17):3389–3402.

4. Ma Y, et al. (2012) Genome-wide identification and characterization of novel genesinvolved in terpenoid biosynthesis in Salvia miltiorrhiza. J Exp Bot 63(7):2809–2823.

5. Wu B, et al. (2012) Comprehensive transcriptome analysis reveals novel genesinvolved in cardiac glycoside biosynthesis and mlncRNAs associated with secondarymetabolism and stress response in Digitalis purpurea. BMC Genomics 13(1):15.

6. Tamura K, et al. (2011) MEGA5: Molecular evolutionary genetics analysis usingmaximum likelihood, evolutionary distance, and maximum parsimony methods. MolBiol Evol 28(10):2731–2739.

7. Chang S, Puryear J, Cairney J (1993) A simple and efficient method for isolating RNAfrom pine trees. Plant Mol Biol Rep 11(2):113–116.

8. Lu S, Zhou Y, Li L, Chiang VL (2006) Distinct roles of cinnamate 4-hydroxylase genes inPopulus. Plant Cell Physiol 47(7):905–914.

9. Lu S, Sun YH, Amerson H, Chiang VL (2007) MicroRNAs in loblolly pine (Pinus taeda L.)and their association with fusiform rust gall development. Plant J 51(6):1077–1098.

10. Dai X, Zhuang Z, Zhao PX (2011) Computational analysis of miRNA targets in plants:Current status and challenges. Brief Bioinform 12(2):115–121.

11. Lu S, et al. (2005) Novel and mechanical stress-responsive MicroRNAs in Populustrichocarpa that are absent from Arabidopsis. Plant Cell 17(8):2186–2203.

12. Song J, Lu S, Chen ZZ, Lourenco R, Chiang VL (2006) Genetic transformation ofPopulus trichocarpa genotype Nisqually-1: A functional genomic tool for woodyplants. Plant Cell Physiol 47(11):1582–1589.

13. Shi R, Sun YH, Zhang XH, Chiang VL (2012) Poly(T) adaptor RT-PCR. Methods Mol Biol822:53–66.

14. Yeh TF, et al. (2005) Rapid screening of wood chemical component variations usingtransmittance near-infrared spectroscopy. J Agric Food Chem 53(9):3328–3332.

15. Dence CW (1992) The determination of lignin. Methods in Lignin Chemistry edsLin SY, Dence CW (Springer, Berlin Heidelberg), pp 33–61.

16. Blakeney AB, Harris PJ, Henry RJ, Stone BA (1983) A simple and rapid preparation ofalditol acetates for monosaccharide analysis. Carbohydr Res 113(2):291–299.

17. Bitter T, Muir HM (1962) A modified uronic acid carbazole reaction. Anal Biochem4(4):330–334.

18. Katahira R, Nakatsubo F (2001) Determination of nitrobenzene oxidation products byGC and H-1-NMR spectroscopy using 5-iodovanillin as a new internal standard.J Wood Sci 47(5):378–382.

19. Chang H, Cowling EB, Brown W (1975) Comparative studies on cellulolytic enzymelignin and milled wood lignin of sweetgum and spruce. Holzforschung 29(5):153–159.

20. Kim H, Ralph J, Akiyama T (2008) Solution-state 2D NMR of ball-milled plant cell wallgels in DMSO-d(6). Bioenergy Res 1(1):56–66.

21. Kim H, Ralph J (2010) Solution-state 2D NMR of ball-milled plant cell wall gels inDMSO-d(6)/pyridine-d(5). Org Biomol Chem 8(3):576–591.

22. Kup!ce E, Freeman R (2007) Compensated adiabatic inversion pulses: broadband INEPTand HSQC. J Magn Reson 187(2):258–265.

23. Trapnell C, Pachter L, Salzberg SL (2009) TopHat: Discovering splice junctions withRNA-Seq. Bioinformatics 25(9):1105–1111.

24. Quinlan AR, Hall IM (2010) BEDTools: A flexible suite of utilities for comparinggenomic features. Bioinformatics 26(6):841–842.

25. Robinson MD, Oshlack A (2010) A scaling normalization method for differentialexpression analysis of RNA-seq data. Genome Biol 11(3):R25.

26. Robinson MD, McCarthy DJ, Smyth GK (2010) edgeR: A Bioconductor package fordifferential expression analysis of digital gene expression data. Bioinformatics 26(1):139–140.

27. Zdobnov EM, Apweiler R (2001) InterProScan: An integration platform for thesignature-recognition methods in InterPro. Bioinf 17(9):847–848.

28. Wei H, et al. (2013) Global transcriptomic profiling of aspen trees under elevated[CO2] to identify potential molecular mechanisms responsible for enhanced radialgrowth. J Plant Res 126(2):305–320.

29. Berthet S, et al. (2011) Disruption of LACCASE4 and 17 results in tissue-specificalterations to lignification of Arabidopsis thaliana stems. Plant Cell 23(3):1124–1137.

30. MacKinnon DP, Lockwood CM, Hoffman JM, West SG, Sheets V (2002) A comparisonof methods to test mediation and other intervening variable effects. Psychol Methods7(1):83–104.

Lu et al. www.pnas.org/cgi/content/short/1308936110 2 of 9

Fig. S1. Phylogenetic relationships of 49 P. trichocarpa laccases and 17 Arabidopsis laccases. Six clades are indicated by C1–C6. Penalty scores for mismatchedpatterns in ptr-miR397a:PtrLAC duplexes within a 20-base sequence window are shown in parentheses. Among the 17 PtrLACs abundantly expressed in SDX, 13(PtrLAC2, 7, 11, 12, 14, 15, 18, 23, 24, 26, 30, 40, and 49) are targets of ptr-miR397a, whereas the other 4 (PtrLAC 17, 19, 25, and 27) are not targets.

Lu et al. www.pnas.org/cgi/content/short/1308936110 3 of 9

Fig. S2. Ptr-miR397a and its targets. (A) Ptr-miR397a targets 29 PtrLACs for cleavage. (B) Ptr-MIR397a gene structure with the location of ptr-miR397a and ptr-miR397a* shown. Black boxes represent exons. A 116-bp intron was found in Ptr-MIR397a by aligning the cDNA sequence against the genome sequence (v2.2).(C) The predicted secondary structure of the first 800 nucleotides of Ptr-MIR397a cDNA. Arrows indicate the orientation of ptr-miR397a and ptr-miR397a*.

Fig. S3. Comparison of wood anatomical structures between ptr-miR397a transgenic and WT plants. Digital images of transverse stem sections from (A) 397-WT1, (B) 397-WT2, (C) 397-6, and (D) 397-10 were taken under a light microscope at !1,000 magnification. (Scale bars, 10 μm.)

Lu et al. www.pnas.org/cgi/content/short/1308936110 4 of 9

Table S1. Identification of 49 laccase gene models in theP. trichocarpa genome

Laccase name Gene model

PtrLAC1 POPTR_0001s14010PtrLAC2 POPTR_0001s18500PtrLAC3 POPTR_0001s21380PtrLAC4 POPTR_0001s25580PtrLAC5 POPTR_0001s35740PtrLAC6 POPTR_0001s41160PtrLAC7 POPTR_0001s41170PtrLAC8 POPTR_0004s16370PtrLAC9 POPTR_0005s22240PtrLAC10 POPTR_0005s22250PtrLAC11 POPTR_0006s08740PtrLAC12 POPTR_0006s08780PtrLAC13 POPTR_0006s09520PtrLAC14 POPTR_0006s09830PtrLAC15 POPTR_0006s09840PtrLAC16 POPTR_0007s13050PtrLAC17 POPTR_0008s06430PtrLAC18 POPTR_0008s07370PtrLAC19 POPTR_0008s07380PtrLAC20 POPTR_0009s03940PtrLAC21 POPTR_0009s04720PtrLAC22 POPTR_0009s10550PtrLAC23 POPTR_0009s15840PtrLAC24 POPTR_0009s15860PtrLAC25 POPTR_0010s19080PtrLAC26 POPTR_0010s19090PtrLAC27 POPTR_0010s20050PtrLAC28 POPTR_0011s06880PtrLAC29 POPTR_0011s12090PtrLAC30 POPTR_0011s12100PtrLAC31 POPTR_0012s04620PtrLAC32 POPTR_0013s14890PtrLAC33 POPTR_0014s09610PtrLAC34 POPTR_0015s04340PtrLAC35 POPTR_0015s04350PtrLAC36 POPTR_0015s04370PtrLAC37 POPTR_0016s11500PtrLAC38 POPTR_0016s11520PtrLAC39 POPTR_0016s11540PtrLAC40 POPTR_0016s11950PtrLAC41 POPTR_0016s11960PtrLAC42 POPTR_0019s11810PtrLAC43 POPTR_0019s11820PtrLAC44 POPTR_0019s11830PtrLAC45 POPTR_0019s11850PtrLAC46 POPTR_0019s11860PtrLAC47 POPTR_0019s14530PtrLAC48 POPTR_0091s00270PtrLAC49 POPTR_0958s00200

Table S2. NMR-derived H:G:S and interunit linkage data for WTand transgenic lignin samples

Sample %H %G %S %PB S/G %A %B %C %SD

WT1 0.2 32.3 67.5 2.5 2.09 86.5 3.7 7.3 2.4397-1 0.2 31.3 68.5 2.1 2.19 88.0 3.1 6.6 2.3397-9 0.3 24.7 74.9 2.9 3.03 89.9 2.3 6.0 1.8397-10 0.3 28.0 71.8 2.4 2.56 88.5 2.9 6.1 2.6

%PB is based on the total lignin and was not included in the total lignin.A, β–O–4 (β-aryl ether); B, β–5 (phenylcoumaran); C, β–β (resinol); SD, β-1(spirodienone) (see Figs. 4 and 5 for structures).

Lu et al. www.pnas.org/cgi/content/short/1308936110 5 of 9

Table S3. Effects on transcript abundance of 34 SDX-expressed laccase genes by Ptr-MIR397a overexpression

Gene Read count average in WT SDXRead count average in

transgenic SDX logFC logCPM P value FDR

PtrLAC1 21.49 3.04 !2.85 2.92 <0.001 <0.001PtrLAC2 85.23 14.61 !2.38 5.00 <0.001 <0.001PtrLAC24 243.72 55.57 !1.97 6.68 <0.001 <0.001PtrLAC18 408.64 106.52 !1.91 7.51 <0.001 <0.001PtrLAC23 204.93 61.55 !1.58 6.60 <0.001 <0.001PtrLAC40 148.70 46.44 !1.73 6.17 <0.001 <0.001PtrLAC15 193.96 74.23 !1.40 6.70 <0.001 <0.001PtrLAC26 107.71 42.92 !1.28 5.88 <0.001 <0.001PtrLAC14 159.20 65.55 !1.30 6.47 <0.001 <0.001PtrLAC41 23.78 9.43 !1.42 3.70 <0.001 <0.001PtrLAC49 130.69 54.82 !1.27 6.20 <0.001 <0.001PtrLAC20 30.16 58.18 1.10 5.68 <0.001 <0.001PtrLAC30 51.49 32.79 !0.63 5.23 <0.001 0.007PtrLAC44 26.40 17.20 !0.70 4.29 <0.001 0.016PtrLAC45 22.39 14.56 !0.72 4.05 <0.001 0.017PtrLAC8 5.16 3.25 !0.73 1.92 0.001 0.030PtrLAC25 118.27 179.58 0.69 7.36 0.001 0.035PtrLAC46 17.28 11.54 !0.64 3.70 0.001 0.039PtrLAC43 46.95 32.12 !0.54 5.16 0.001 0.049PtrLAC6 7.17 4.68 !0.56 2.40 0.001 0.053PtrLAC29 19.64 13.86 !0.54 3.93 0.003 0.113PtrLAC21 69.36 50.62 !0.43 5.79 0.006 0.164PtrLAC42 23.67 17.14 !0.46 4.23 0.006 0.168PtrLAC7 109.19 82.45 !0.38 6.48 0.014 0.261PtrLAC12 133.44 106.64 !0.33 6.82 0.050 0.515PtrLAC33 11.33 9.23 !0.25 3.29 0.100 0.707PtrLAC11 313.90 262.03 !0.31 8.10 0.113 0.735PtrLAC17 201.96 236.67 0.25 7.83 0.188 0.865PtrLAC5 8.12 7.01 !0.33 2.87 0.224 0.911PtrLAC19 185.25 163.46 !0.16 7.40 0.273 0.946PtrLAC16 21.93 20.51 !0.06 4.38 0.573 1.000PtrLAC27 266.91 284.15 0.08 8.13 0.609 1.000PtrLAC4 2.93 2.64 !0.17 1.44 0.483 1.000PtrLAC47 3.80 3.87 0.08 1.94 0.894 1.000

cpm, read counts per million reads; FC, fold change. FDR value <0.05 is bold.

Lu et al. www.pnas.org/cgi/content/short/1308936110 6 of 9

Table S4. Fifty enriched protein domains identified from 459 DEGS in the context of genomic background

Interpro ID P value DescriptionGene number in DEGs vs.

that in the genome

IPR001154 0 DNA topoisomerase II, eukaryotic-type 2:2IPR017761 5.2375E!26 Laccase 20:34IPR002355 1.51259E!24 Multicopper oxidase, copper-binding 20:38IPR001117 9.8465E!22 Multicopper oxidase, type 1 20:48IPR011706 9.8465E!22 Multicopper oxidase, type 2 20:48IPR011707 9.8465E!22 Multicopper oxidase, type 3 20:48IPR008972 4.2319E!17 Cupredoxin 20:74IPR001099 8.0448E!07 Chalcone/stilbene synthase, N-terminal 4:5IPR011141 8.0448E!07 Polyketide synthase, type III 4:5IPR018088 8.0448E!07 Chalcone/stilbene synthase, active site 4:5IPR001404 2.8492E!06 Heat shock protein Hsp90 5:10IPR004022 6.1593E!06 DDT family 5:11IPR018500 6.1593E!06 DDT subgroup 5:11IPR018501 6.1593E!06 DDT superfamily 5:11IPR000823 1.03912E!05 Plant peroxidase 7:28IPR001371 0.000012151 Glycoside hydrolase, family 14B, plant 5:12IPR011684 0.000012151 KIP1-like 5:12IPR018238 0.000012151 Glycoside hydrolase, family 14 5:12IPR001752 2.02377E-05 Kinesin, motor region 11:86IPR001554 0.00006285 Glycoside hydrolase, family 14 5:15IPR006766 0.00006285 Phosphate-induced protein 1 5:15IPR002068 0.000068297 Heat shock protein Hsp20 6:24IPR010255 0.000211595 Haem peroxidase 7:40IPR002483 0.000220813 Splicing factor PWI 3:5IPR016140 0.00111454 Bifunctional inhibitor/plant lipid transfer 6:35IPR008978 0.00136175 HSP20-like chaperone 6:36IPR008263 0.00207824 Glycoside hydrolase, family 16 4:15IPR000167 0.00248467 Dehydrin 2:3IPR002160 0.00248467 Proteinase inhibitor I3, Kunitz legume 2:3IPR002205 0.00248467 DNA topoisomerase, type IIA 2:3IPR011065 0.00248467 Kunitz inhibitor ST1-like 2:3IPR013757 0.00248467 DNA topoisomerase, type IIA 2:3IPR013758 0.00248467 DNA topoisomerase, type IIA 2:3IPR002016 0.00262713 Haem peroxidase, plant/fungal/bacterial 7:55IPR012328 0.0029749 Chalcone and stilbene synthases 4:16IPR003594 0.0092599 ATP-binding region, ATPase-like 7:65IPR001241 0.0098046 DNA topoisomerase, type IIA 2:4IPR013506 0.0098046 DNA topoisomerase, type IIA 2:4IPR013759 0.0098046 DNA topoisomerase, type IIA 2:4IPR015310 0.0098046 Activator of Hsp90 ATPase, N-terminal 2:4IPR013770 0.0099303 Plant lipid transfer protein 4:20IPR004648 0.0133661 Tetrapeptide transporter 3:11IPR016197 0.0197349 Chromo domain-like 3:12IPR000232 0.0206148 Heat shock factor (HSF)-type, DNA-binding 4:23IPR010544 0.0241763 Kinesin-related 2:5IPR013760 0.0241763 DNA topoisomerase, type IIA, central 2:5IPR018368 0.028073 Chaperonin ClpA/B, conserved site 3:13IPR013126 0.0315926 Heat shock protein 70 4:25IPR000109 0.043247923 TGF-beta receptor 5:42IPR003480 0.046569755 Transferase 4:27

Lu et al. www.pnas.org/cgi/content/short/1308936110 7 of 9

Table S5. Gene models of the genes listed in Fig. 6

Gene name Gene model

PtrPO2 POPTR_0003s21620PtrPO3 POPTR_0002s01960PtrPO6 POPTR_0007s13420PtrPO34 POPTR_0001s04840PtrARF2 POPTR_0015s11660PtrMYB52 POPTR_0007s01430PtrDAG1 POPTR_0014s09640PtrUNE12 POPTR_0002s04620PtrHSFB4 POPTR_0001s28040PtrMYB50 POPTR_0015s09430PtrMYB52 POPTR_0007s01430PtrMYB73 POPTR_0007s10490PtrNF-YC9 POPTR_0013s02580PtrABF3 POPTR_0009s10400PtrARF8 POPTR_0017s01870PtrARF7 POPTR_0006s07740PtrMYBCDC5 POPTR_0019s03520PtrMYB42 POPTR_0003s11360PtrHB-1–1 POPTR_0011s05660PtrHB-1 POPTR_0004s04840

Lu et al. www.pnas.org/cgi/content/short/1308936110 8 of 9

Table S6. Primers used for qRT-PCR analysis of laccase gene expression

Gene name Forward primer Reverse primer

PtrLAC1 ATTAATGGACTCCCAGGGCCATTA TTCACATACACAGCATCAGCTTCCPtrLAC2 AGTTTTGAGTTTCACCAATATTACCTG CCAGAGGCAGCTCATAGCAAATPtrLAC3 TCACCATCAATGGAGAACCAGGATA GCCAGGGGTGATCATTAGATAGTCTPtrLAC4 AGGGAGGATATAATTTACCAGTACGG CAATGGTGTCGATTTTGAAGGGTTTAPtrLAC5 GGCAAGACTTACATGCTCCGCTTA AAACAGCATCAACATCAACAATGGTAPtrLAC6 CTCTGATGCCTACACTATCAATGGAT TGGTTTCACATAAATAGCATCAACACPtrLAC7 CTCTGATGCCTACACTATCAATGGAT GTTTGCTATGCTGAAGAAGAGCTCGPtrLAC8 CATACCATTAATGGGAAGCCAGGGA CAATCAGTACTGCTTGAGTGGTAAAATPtrLAC9 CATTCACTGGCATGGAGTGAAACAA GCCACCATAATGTTCCTTCTTCGTTPtrLAC10 GAGACACAATCTATGTTACGGTCCAT CCAGGTTGGATTGGACACTGTGTAPtrLAC11 GGCCTTCCAGGGCCTTTATACAAT CTGGACTGATGAGGAGCGTGTCAPtrLAC12 CCAGGAAAAACCTATCTCCTTCGTT CGGGGTGATGAGTACAATGTGGGTPtrLAC13 CTACCCTTGCTCTCAAAACAGAATATT CGATGACAACAACGCCAGTGACGPtrLAC14 TGCCTCTTCCCTGCTTTGGTCC GCCATTGACGGTAACAATCGGCPtrLAC15 TCTTGCTGTCTGCCTCCTCCCC GCCATTGACGGTAACAATCGGCPtrLAC16 CATACTTTTGCAATGGAGATTGAATCA ATCTGCCTGGAACTTGGTTGGCTAPtrLAC17 ACCTCCCAGTGACCTCCCCAAC CTTTGTTTGAGCACTTGAGCCCAGPtrLAC18 GGTCAACCTGGTGATCTTTTTAACTGC AACTGTAAACTTGTGGTTGGCTATGPtrLAC19 CTTGGTAGAGAATGGAGTTGGAC TTGTCGGGCTTATGAATTCCAACPtrLAC20 GCTGATCCTGAGGCAGTGATTAGA AAGGTTGGTCTCAAAAGGCTTCACAPtrLAC21 CACGGAGGGTATAATTTATCAGTACAT CAACGGTGTCGATTTTGAAGGGTTTTPtrLAC22 ATACATTTACCTTGGAGGTTGAACAG CAATTAGTATGGTTTGAGTTGTAAAGGPtrLAC23 CGAGGCACTACAAGTTCAACATAG CCATTAACAGTCACCATGCTCTTCPtrLAC24 AGTTTTGAGTTTCACCAATAGTACCTA CCAGAGGCAGCTCATAGCAAAGPtrLAC25 CTTGGTCGAGAATGGAGTCGGAG GCTTATGAATTTCAACAAACTGGTACGPtrLAC26 GGTCAACCTGGTGATCTTTATAATTGT TAAGATAAGAGGCATCAGCACCAATAPtrLAC27 ATGAGTCAATACTACCACCTCCT CTATACTCCTCCCTCCCGCATCPtrLAC28 CACAAGACTCTGCAGCACCAAGAA GGCTGTTTCACTCCATGCCAGTGAPtrLAC29 AGGCCCAAATGTCTCTGATGCTTAT AAGAAGAGCTCGTCATTGAGTGCGPtrLAC30 GCAGATCCAGAGGCGATCATTAGT AAGAAGAGCTCGTCATTGAGGGCAPtrLAC31 ATGCCGTATCCATTCTCAGCTCAG GGCATTCCATTTATGGTGTAAGCGTPtrLAC32 GTAATCTTCAGTTTCCAAGCTCATTT TTTGATCACAAGAGTATCGCCATCGPtrLAC33 GGCACCTTCTTCTGGCATGCTCAT AGATGTACAATATCCTGAAGCCAGTAPtrLAC34 GGCGTTACCGTGTTGTTCTTCAC CAATCCGATTTTTGACCTTAATATGAATACPtrLAC35 ACGAAGCAGCTTCTAATGGTGAAC TCTGAGCAATCCGATTTTTGACCPtrLAC36 GGGATCCAGATGAAGTTGAAAACAG GCTATCGCAAAGAAGAGCTCATCCPtrLAC37 TGTTGATGTCGAAAACCAGGCAGAA GATTGGCTATCTTGAAAAAGAGCTGGPtrLAC38 GCTTCTGCTGCAATTGTGGAACG AAGCGTTGGGCCTGGCAAGCTGPtrLAC39 GATCCTGTCAACGACCATAAGAAT CCAGATGACAGTGAAAGAACCACPtrLAC40 AAGGCGGTTTCACTTTGCCAGTCC TAGGTGGCATCAACTTCCACGACAPtrLAC41 GGGCGGTTTCACATTGCCAGTCG GTAGGTGGCATCAACTTCCACAAGPtrLAC42 ACAATTATTATTGGATCTTGGTTTAAGG CATTGATGGTAAGACTGTTGGATATATTPtrLAC43 GAAATCAACACTGTTGCACTTCG GGCGCATACTTTGTGCCGTTGTATPtrLAC44 ACAATTATTATTGGATCTTGGTTTAAGG CATTGATGGTAAGACTGTTGGATATATTPtrLAC45 ACAATTATTATTGGATCTTGGTTTAAGC CCATTGATGGTAAGACTATTGGATGGPtrLAC46 GCAACTGGTGCTGGACCTGCAAT TTAACCTTCAAACGGTATGTGTTTTCTPtrLAC47 GTAATCATCAGTTTCCAAGCATCTTC CTTGATCACAAGAGTATCACCGTCAPtrLAC48 GCTCATTCTATGTGAAAAACCTTACAG AGCGTTGGGCCTGGCAAGCTTPtrLAC49 CAGCAAATATGGAGTACTCCAATTG TGCCGAATCCTGCACTCGACCPtr18SrRNA CGAAGACGATCAGATACCGTCCTA TTTCTCATAAGGTGCTGGCGGAGT

Table S7. Primers used for validation of ptr-miR397a targets

Gene name Gene-specific primer Nested gene-specific primer

PtrLAC1 ACGTAAGGTCTAGCAGTCATGAAAAATT GTGGTGGGGTTTGGTCTTGAGAAGGPtrLAC2 GGCAATGGGTTTGGTCTTCAGAAGAAC GTGGTCTGTCCTGCTGTTATAACCAAGPtrLAC13 TGTAATAAGATCCCACTTCCTGGTCT CGATGACAACAACGCCAGTGACGPtrLAC18 CTGCCATGTAATATCGGCCAGGTAGT CTGTGGTTTGGCCAGGTCCTAGCATGPtrLAC21 CCACGGTTTGCCGTCACAAGAACGTTT CAACGGTGTCGATTTTGAAGGGTTTTPtrLAC26 CTGCCATGTAATATCGGCCAGGTAGT CAGTGGTTTGGCCTGGTCCTAGCATTPtrLAC30 AGTCGGGTTTGGTTTTCAGAAGAACA GGTTTGTCCAGGGGTAATGAGAAGT

Lu et al. www.pnas.org/cgi/content/short/1308936110 9 of 9