UBC Faculty Research and Publications

Defense mechanisms against herbivory in Picea: sequence evolution and expression regulation of gene family… Porth, Ilga; Hamberger, Björn; White, Richard; Ritland, Kermit Dec 16, 2011

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-12864_2011_Article_3868.pdf [ 618.67kB ]
JSON: 52383-1.0221292.json
JSON-LD: 52383-1.0221292-ld.json
RDF/XML (Pretty): 52383-1.0221292-rdf.xml
RDF/JSON: 52383-1.0221292-rdf.json
Turtle: 52383-1.0221292-turtle.txt
N-Triples: 52383-1.0221292-rdf-ntriples.txt
Original Record: 52383-1.0221292-source.json
Full Text

Full Text

RESEARCH ARTICLE Open AccessDefense mechanisms against herbivory in Picea:sequence evolution and expression regulation ofgene family members in the phenylpropanoidpathwayIlga Porth1,2, Björn Hamberger2, Richard White3 and Kermit Ritland1*AbstractBackground: In trees, a substantial amount of carbon is directed towards production of phenolics fordevelopment and defense. This metabolic pathway is also a major factor in resistance to insect pathogens inspruce. In such gene families, environmental stimuli may have an important effect on the evolutionary fate ofduplicated genes, and different expression patterns may indicate functional diversification.Results: Gene families in spruce (Picea) have expanded to superfamilies, including O-methyltransferases,cytochrome-P450, and dirigents/classIII-peroxidases. Neo-functionalization of superfamily members from differentclades is reflected in expression diversification. Genetical genomics can provide new insights into the genetic basisand evolution of insect resistance in plants. Adopting this approach, we merged genotype data (252 SNPs in asegregating pedigree), gene expression levels (for 428 phenylpropanoid-related genes) and measures ofsusceptibility to Pissodes stobi, using a partial-diallel crossing-design with white spruce (Picea glauca). Thirty-eightexpressed phenylpropanoid-related genes co-segregated with weevil susceptibility, indicating either causative orreactive effects of these genes to weevil resistance. We identified eight regulatory genomic regions with extensiveoverlap of quantitative trait loci from susceptibility and growth phenotypes (pQTLs) and expression QTL (eQTL)hotspots. In particular, SNPs within two different CCoAOMT loci regulate phenotypic variation from a common setof 24 genes and three resistance traits.Conclusions: Pest resistance was associated with individual candidate genes as well as with trans-regulatoryhotspots along the spruce genome. Our results showed that specific genes within the phenylpropanoid pathwayhave been duplicated and diversified in the conifer in a process fundamentally different from short-livedangiosperm species. These findings add to the information about the role of the phenylpropanoid pathway in theevolution of plant defense mechanisms against insect pests and provide substantial potential for the functionalcharacterization of several not yet resolved alternative pathways in plant defenses.BackgroundLignin, the structural component of land plants, is a het-eropolymer of coupled phenylpropanoid monomersderived from hydroxycinnamyl alcohols; it is also thesecond most abundant biopolymer after cellulose onearth. In gymnosperms, 30% of wood dry weight is lig-nin, in angiosperms this value is up to 25% ([1,2]).Lignocellulosic polymers are the dominant carbon sinkin forest ecosystems and account for approximately 20%of the terrestrial carbon storage [3]. In addition, thephenylpropanoid pathway synthesizes a plethora of spe-cialized plant products (anthocyanins, flavonoids, con-densed tannins, stilbenes, soluble and cell wall-boundphenolics, and other polyphenols) with protective func-tions (antioxidant [4], radical scavenging [5], stressinduction [6], UV [7], flavonoids and UV tolerance[8,9]). The biosynthesis of those phenylpropanoidsinvolves intricate networks [10] as shown by the KEGG* Correspondence: kermit.ritland@ubc.ca1Department of Forest Sciences, University of British Columbia, 2424 MainMall, Vancouver, BC V6T1Z4, CanadaFull list of author information is available at the end of the articlePorth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608© 2011 Porth et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative CommonsAttribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction inany medium, provided the original work is properly cited.pathways http://www.genome.jp/kegg/pathway/map/map00940.html.The importance of specialized phenolic compoundsfor plant chemical defense has been the subject of inten-sive study in angiosperms ([11,12]), however, knowledgein gymnosperm lineages is sparse. In general, phenolicsparticipate in pre-formed defenses (toxins, cell wall asso-ciated compounds) as well as in active defense reactions(antimicrobial phytoalexins). Biosynthesis occursthrough a limited number of metabolic pathways (Figure1A), however, the extreme structural diversity of com-pounds found so far (Figure 1B) is due to biosyntheticenzymes that belong to multi-member gene families.Phenylpropanoid biosynthesis and its genetic basis hasbeen recently reviewed [13]. For poplar, gene familiesinvolved in phenylpropanoid biosynthesis exhibit anexpansion and diversification in comparison to short-lived annuals [14], suggesting diversification of chemicaldefense strategies. The limited number of phenylpropa-noid pathway genes functionally characterized in gym-nosperm species include those involved in angiospermlignin biosynthesis (peroxidases: Picea abies [15], O-methyltransferases: Pinus taeda [16], P. sylvestris [17], P.radiata [18], hydroxycinnamoyl-CoA:shikimate hydroxy-cinnamoyltransferase: P. radiata [19], and MYB: P.taeda: [20]).Constitutively present phenolics have high priority asdefenses against bark-boring insects in long-lived coni-fers, but pre-formed defenses are regarded costly for theplant since “assimilates” are diverted. This impacts4-coumaroyl-CoAphenylalanineHGferuloyl-CoAanthocyaninserythose-4-phosphate + phosphoenol pyruvateDHSPALC4H4CLchalconesstilbenes STSCHSSHCTC3HHCTcaffeoyl-CoACCoAOMTOMT CCR CADCCR CADF5HCOMT CADconiferaldehydemonolignolsCHIflavanonesF3H/F3’H/F3’5’HflavonolsLARshikimatearogenateADT4-coumarylaldehydeANSADIR/LAC/PRXBstilbenesR2R1R3R4monolignolsHOO HR1R2anthocyanidinsOHHO OHOOHR2R1myb/NACSAMSFigure 1 Schematic overview of the phenylpropanoid pathway. A Gene families of the main branches for which spruce gene memberswere profiled by microarray analysis are indicated in red. Given in grey are transcription factors, S-adenosylmethionine synthetase and genefamilies not part of the core phenylpropanoid pathway in gymnosperms; B examples of metabolites representing the structural variety generatedwithin the core phenylpropanoid pathway, stilbenes: R1-R2-R3-R4: (OH, OCH3)-(OH, O-glucopyranose, OCH3)- (OH, OCH3, H)-(H, OH), [105];anthocyanidins: R1: (OH, H), R2: H, [106]; monolignols: R1-R2: H-H (p-coumarylalcohol), R1-R2: H3CO-H (coniferyl alcohol), R1-R2: H3CO- H3CO(sinapyl alcohol), [10].Porth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608Page 2 of 26growth rate and reproduction [21]. Our previous find-ings (unpublished and [22]) already reflected the bio-chemical link between growth and established(constitutive) defenses in this important metabolicpathway.Conifers have extremely large genomes (ca. 20 billionbases), which largely consist of repetitive elements([23,24]), but the enormous size of conifer genomes isalso likely due to complex gene families that have beensignificantly enlarged from angiosperm ancestors. Thesefamilies have structural and regulatory functions relatedto defense/resistance mechanisms against pests andpathogens (including terpene synthases, cytochromesP450, TIR-NBS-LRR genes, pathogen-resistance genesand dirigents [25-29]). Those multigenic familiesresulted from duplications of genes that furtherdiverged.In the plant kingdom, angiosperms are well studiedcompared to ancient lineages such as mosses, ferns, orgymnosperms [30]. Therefore, we must rely uponknown and characterized candidates from angiospermsto gain insight into the evolution of sequence orthologsof conifer phenylpropanoid-like genes. This involves evi-dence for neofunctionalization, subfunctionalization orneo-subfunctionalization within gene families. Geneduplication is recognized as an important mechanismfor adaptive evolution in plants [31]. Duplication allows“neofunctionalization” or the retention of ancestral func-tion by one gene and the origin of new functions by thesecond gene. Likely, the novel gene function adopted byone copy of the duplicated genes, which now determinestheir functional differences, is co-opted from a second-ary ancestral property already present before the dupli-cation event [32]. The fate of specialized terpenesynthases in spruce is one well-studied example thatexemplifies how neo-functionalizations account forfunctional plasticity in a multi-gene family [33].Another mechanism for maintenance of gene dupli-cates involves gene conservation using gene-dosageeffects and functional redundancy [34]. Moreover, lossof genes and regulation of gene activity through tran-script accumulation are other important aspects to con-sider in gene evolution. Phylogenetic trees are robust forprediction of orthologous genes in the presence of genelosses and varying rates of evolution among the sampledtaxa [35]. Hence, in our study of multi-gene families inspruce we employed phylogenetic reconstruction tofacilitate gene annotations.In the present work, we determined the extent of geneco-expression by correlation analysis for 428 EST micro-array elements (Additional File 1). The analysis includeda number of multi-gene families originating from 26biosynthetic genes, three structural genes related to thephenylpropanoid pathway, transcription factors NACand myb (of which few have been characterized as keyregulators of secondary wall biosynthesis and phenylpro-panoid metabolism [36,37]), and SAMS (a key enzymefor methylation reactions which are also important inlignin formation [38]) (Additional File 1).The same ESTs were studied by means of “geneticalgenomics” to investigate the genetics of defensemechanisms of white spruce against the white pine wee-vil (Pissodes strobi). Genetical genomics [39] combinesgene expression with genotyping to map quantitativetrait loci (QTL) for gene expression. This new approachto the study of quantitative genetics allows rapid anddirect discovery of genes underlying a trait of interest("positional candidate genes”) or genes closely regulatedwith the QTL by “genetic co-localization” (co-segrega-tion of transcript variation of genes with the phenotypictrait of interest). For example, due to the genetic as wellas environmental and developmental sources of varia-tion, transcript level variation at a gene locus underlyinga QTL is highly associated with the total phenotypicvariation in the actual quantitative trait [38]. Hence, insystems where the environment shows a strong effecton the phenotype, this approach has advantages overconventional QTL analysis with low resolution thatfurther necessitates positional gene cloning [38].ResultsGenetical Genomics Reveals Candidate Genes for WeevilResistance and Genomic Regions with RegulatoryHotspotsIn the present study we merged genetics and transcrip-tomics [40,41] to better understand weevil resistance inspruce, an important phenotypic trait that defines thelife history of this economically valuable perennial plantorganism. Gene expression variation represents the phe-notype most directly related to DNA sequence poly-morphism, such that in principle each transcript has acorresponding gene with known position in the genome[42]. Genetical genomics allows assaying thousands ofthese gene expression traits simultaneously and thusprovides data on a large and unbiased set of traits (ibi-dem); these ‘expression phenotypes’ are then accessibleto standard QTL analysis. Using the gene expressiontraits obtained from our transcriptome study weachieved fine scale phenotyping that was merged withphenotyping of the conventional traits. This allowed usto identify positional candidate genes for any phenotypicvariation by testing for co-segregation of markers withgene expression, individual resistance traits, or the com-posite trait represented by all sampled defense traits(see below). This way, we found certain members ofcomplex gene families within the phenylpropanoidmetabolic pathway individually associated with weevilresistance.Porth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608Page 3 of 26We used the experimental setup as presented in Fig-ure 2. Plant material was harvested - as outlined in theMethods section - from the progeny of resistant-female-by-susceptible-male crosses, which showed wide segre-gation for weevil resistance and had shared parentage([43]). Thus, cross 26 of ♀PG87*♂PG165, cross 27 of♀PG87*♂PG117, cross 29 of ♀PG21*♂PG165 and cross32 of ♀PG21*♂PG117 forming the partial diallel werechosen for further analysis. The plant material used forexpression profiling (bark tissue from tree leaders) washarvested in a randomized fashion to minimize bias dueto the sampling procedure. DNA from an expandedmapping population (417 individuals in total) was geno-typed using a custom-built 384 multiplexed SNP chip.This information was used to estimate pairwise recom-bination rates between SNP loci and subsequently con-struct the framework genetic linkage map for localizingthe QTLs (Additional Files 2 and 3). Phenotypic infor-mation for QTL analysis was obtained from (a) treeheight, (b) weevil attack, (c) oviposition, and (d) tran-scriptomics data. Measures were taken for the initialtree height in 1995, and heights in years three and fiveas well as leader length in year five preceding the artifi-cial augmentation of the local weevil population inOctober of the same year. Attack rates in 2000 and2001 were classified as successful top kills, failure to killthe leader, and no attack. For the same years, egg countsalong the leaders were summarized into five discreteclasses. Comprehensive information about these mea-sures can be found in Additional File 4. In the transcrip-tomics experiment a two color 21.8 k spruce EST arraywas employed for gene expression profiling in the partialdiallel progeny. A distant pair design that maximizeddirect comparisons between different alleles at eachlocus [41] was modified for outbred individuals (Addi-tional File 5). Signal intensities can be found under theGEO accession number GSE22116. We used the signalintensity at each EST spot for the following analyses: (a)generation of expression QTLs (eQTLs) and (b) estab-lishing a co-expression network. In the latter the gene-gene interactions were assessed. Specifically, a Gaussianmodel and a shrinkage method were employed to evalu-ate direct gene connections (see Methods). QTLs weremapped in the diallel progeny using a likelihood func-tion to assess the phenotype effect conditional on geno-typic variation. A QTL was significant at LOD≥3.84(Additional Files 6 and 7). A goodness-of-fit test assum-ing a uniform distribution was performed to test35 21 5 9 A1 8 15 23 1019 40 13 4 2 6 3020 34 7 32 18 27 3741 42 31 39 22 25 16 2928 26 3 36 14 11 33 1726 31 11 39 36 38 12 2423 34 25 17 6 7 32 19 235 20 41 42 27 16 3 18 1   B21 28 14 4 38 29 24 13 379 15 8 40 5 12 30 10 22 3327 11 39 32 16 29 1 33 36 21     C19 15 37 2 13 12 38 17 34 269 22 30 28 4 23 20 31 25 35 614 3 41 40 5 8 42 24 10 18 75. Gene expression prof iling (21840 ESTs)QTL analysis2. (randomized) sampling among and within plots3. SNP genotyping (384 SNPs)x PG165 PG117PG87 72 65PG21 68 73x PG165 PG117PG87 48 36PG21 50 54Microarray experimental designGenetic linkage map4. Identif ication of  the bestindividuals within a family as Cy5/Cy3 probe pairsPhenotype data1. Collecting of  resistance, height dataRandomized plots within the three blocksFigure 2 Outline of the experimental setup for QTL identification. Layout of study site showing the randomized location of plots (QTLmapping families PG87*PG165 (cross 26), PG87*PG117 (cross 27), PG21*PG165 (cross29) and PG21*PG117 (cross32), in yellow) within blocks A, B,C. The numbers of individuals per family for genotyping, phenotyping and microarray analysis, respectively, are given. For details about whichindividuals were directly compared using the distant pair design, see Additional File 5.Porth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608Page 4 of 26whether the observed frequencies of eQTLs along thelinkage map differed significantly from the expectedvalue. Following the rejection of this null hypothesis“eQTL hotspots” signified eQTL clusters with ≥38eQTLs at a given locus. Positional candidate genes wereidentified by collocation of at least 40% of their eQTLswith phenotypic trait QTLs based on the criteria foridentifying significant QTLs (10,000 randomizations, p ≤0.05). The positional candidate genes for the general‘resistance’ trait were identified based on strong colloca-tions of gene expression variation with all six studiedresistance traits (see Material and Methods).Expression variation from 428 gene spots generated4,221 significant eQTLs (LOD ≥ 3.84). Figure 1A showsthe phenylpropanoid pathway, gene families from thecore branch as well as related families; data relevant forthis study are found in Additional File 1. Phenylpropa-noid pathway-linked genes that showed strong associa-tion with the general ‘resistance’ phenotype were:SAMS, one ADT-like gene (PicglADTL8, Figure 3), anannotated spruce CYP750 (C24, Additional File 8), anda member with similarity to LAR. An additional 34 phe-nylpropanoid-related candidate genes were identifiedbased on significant co-segregation with individual wee-vil resistance traits such as attack rates and ovipositionestimates: transcription factors (seven putative mybs andone putative NAC), putative members of the upstreamshikimate pathway (DAHP and DHQD-SD), one puta-tive PAL, two annotated CHS as well as one STS, twoOMTs from group C and one from group D (OMTL),two P450s (CYP75/F3’H and CYP750, respectively),three different representatives of phenylpropanoidreductases (PCBER, PLR and IFR), one DIR (f-family),six PRXR (including the stress inducible PicabPRX2orthologue), two putative LACs, one putative catechol-O-methyltransferase and two genes with similarity toLAR. Most associations were found for weevil suscept-ibility measured in the year 2000 and for both yearsPethyADT3PethyADT1PethyADT2* *ArathADT2ArathADT1******PicglADT5ArathADT4ArathADT5ArathADT3ArathADT6PicglADT2PicglADTL6PicglADTL8PgeADTL7PicglADT4PicglADT3PicglADTL2PicglADT1PicglADTL1PicglADTL3PicglADTL4PicglADTL50.1Figure 3 Arogenate dehydratase (ADT) and related families. The flash highlights functionally characterized ADT. The bar represents 0.1amino acid changes. Asterisks indicate 80% and above confidence through bootstrap values. The dashed circles indicate the distinct core cladeof the bona fide ADT and two related spruce ADT-like (ADTL) families. Arath, Arabidopsis thaliana; Picgl, white spruce (Picea glauca); Pge, interiorspruce (Picea glauca x engelmannii); Pethy, petunia (Petunia x hybrida); the red font indicates representation on the microarray. GenBankaccession numbers are given in Additional File 14.Porth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608Page 5 of 262000 and 2001. No dedicated monolignol biosyntheticgene co-segregated extensively with any resistance trait.However, 29 genes from our gene set were associatedwith individual height growth traits (predominantly treeheight measures taken in 1999 preceding the artificiallyenforced weevil attacks), Additional File 1. As many as11 peroxidases co-segregated with growth traits, amongthem several genes which are implied in lignin polymer-ization by radical coupling of the monolignols. Their co-expression pattern with other phenylpropanoid-relatedgenes is shown in Figure 4.We superimposed at a given SNP marker pQTL maps ofindividual phenotypic traits and counts of significanteQTLs from the studied gene set (Figure 5). We foundregulatory hotspots comprising multiple pQTLs andaccumulated eQTLs. Along this QTL density map, eightloci represented hubs of trans-eQTLs, which also corre-sponded with at least three pQTLs: four loci were asso-ciated exclusively with resistance QTLs, and each twoloci with growth QTLs and QTLs from individual traitsof both growth and resistance, respectively (Figure 5).The composition of those eQTL hotspots with extensivepQTL overlap is given in Additional File 9.Since the phenylpropanoid pathway proved to beimportant in insect resistance, we looked at the repre-sentation of individual branches within the pathway aswell as individual gene families among those regulatoryhotspots that were associated with pQTLs. We havesummarized the average number of detected eQTLs pergene family as well as gene families whose membershave eQTLs at a minimum of two regulatory hotspotswith pQTL association (Additional File 10 and Figure6). Specifically, we compared members from transcrip-tion factors myb and NAC, the shikimate pathway, PAL,C4H and 4CL (main branch of the phenylpropanoidpathway), and gene families that are involved in the bio-synthesis of various secondary compounds as well as lig-nin and lignans, respectively (Additional File 10).Among all pathway branches, the branch of the phenyl-propanoid pathway that is directly involved in the bio-synthesis of the secondary compounds generated onaverage the highest number of eQTLs, whereas the shi-kimate pathway generated the lowest number of eQTLson average. Also, the transcription factors followed bythe pathway branch that is directly related to lignin/lig-nan biosynthesis contributed on average the most1 23456789101112Figure 4 Co-expression network for the phenylpropanoid pathway. See also Additional File 1 for annotations (probability cutoff for “trueconnections” between gene pairs was 0.80; connections are not drawn to scale), vertices marked in the network: red: positional candidates forresistance trait, green: for growth trait, magenta: for both resistance and growth trait (see text); wide range subgraphs can be followed by bluevertices, important connections are framed and are numbered by order of appearance in the Results and in the Discussion, red: focus on O-methyltransferases, blue: peroxidases; green: dirigent-likes.Porth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608Page 6 of 26members to resistance associated hotspots (AdditionalFile 10). In 20 out of 29 studied gene families, individualfamily members contributed eQTLs to at least two dif-ferent phenotype-associated eQTL hotspots. Forexample, members of CCoAOMT, CCR, ADT, as well asmyb, PRXR, OMT, P450 families had eQTLs preferen-tially associated with resistance hotspots (Figure 6). Spe-cifically, we found ADTs (and one ADT-like gene),Figure 5 Representation of the eQTL density map with overlapping positions of pQTLs at individual marker positions. Linkage groups(LG1-13) are displayed horizontally, black bars indicate SNP marker positions in linkage groups (see also Additional File 2); arrows mark positionswith at least three pQTLs (LOD > 3.84, i.e., values above horizontal line) and eQTL numbers ≥ 38.02468101214161820mybNACSAMSDAHPDHQSDHQD-SDSKEPSPSCSCMADT_aPAL4CL_aCHS_aCHIFLSP450_aDFR_aPCBER_aPLRANS3GTCCoAOMT_aOMT _aCCRCADLACPRXR_aDIR_agrowth phenotyperesistance phenotypeall combinationsFigure 6 Representation of 29 studied gene families at eight pQTL associated eQTL hotspots. Gene families are presented in thefollowing order: transcription factors, early shikimate pathway, the core, and the late phenylpropanoid pathway. Vertical bars show numbers ofmembers of gene families with members having eQTLs at a minimum of two different phenotype-associated eQTL hotspots (all combinations),eQTLs among pure resistance hotspot(s) and other hotspots with resistance association (resistance phenotype associated) as well as eQTLsamong pure growth hotspot(s) and other hotspots with growth association (growth phenotype associated), see also Additional Files 9 and 10;“a”... spruce gene family with known/determined phylogeny.Porth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608Page 7 of 26OMTs from group C as well as both AEOMT genes, thelignin-forming genes PiglCCoAOMT3, PicabPRX1 andPicabPRX18 orthologs and the stress inducible Picab-PIPRX gene, and the P450s that are involved in a widerange of derivatization reactions (CYP75/F3’H, CYP750,F3’5’H/F3H and C4H class, respectively), Additional File9. The DIR gene family members had eQTLs predomi-nantly associated with growth trait QTLs (Figure 6) thatwere mainly representatives of the f-family (AdditionalFile 9).Although most spruce gene markers used to build ourframework linkage map lack consistent annotations(67% using TAIR7, 54% using Viridiplantae databasesgave no hits), Additional File 2, for the few loci that arepotential gene expression regulators located withineQTL hotspots and affect highly complex phenotypes,we further infer their putative gene function in spruce.For example, the growth trait associated locus on LG13with accumulated eQTLs from 49 array elements (Fig-ure 5) represents a glutamate decarboxylase (GAD)gene. GAD catalyzes gamma-aminobutyric acid (GABA)synthesis via decarboxylation from the amino acid gluta-mate. The regulation of GAD enzyme activity is vital fornormal plant development, and is accomplished by cal-cium/calmodulin binding to the specific CaM domain ofGAD allowing the plant to respond to various externalstimuli [44]. Recently, the locally enhanced productionof GABA in the plant was shown to be connected witha deterrence reaction of the host responding to herbi-vore attack [45]. Two loci that encode bona fideCCoAOMT genes (CCoAOMT-1, CCoAOMT-2) onLG6 represented eQTL hotspots that overlapped withQTL regions for various resistance traits (Figure 5). Atthe two different CCoAOMT loci a common set ofpQTLs (i.e., atk_2000, egg_2000 and sum_egg) as wellas eQTLs (generated by 24 members of 11 differentgene families) clustered. Most prominently the followinggene families were involved: PRXR (PicabPRX18, e.g.),DIR (PgeDIR13, a-family; PicsiDIR31, PicsiDIR27, f-family), OMT (AEOMTs PicglOMT-17 and PicglOMT-18; PgOMT-35, clade II), myb (TT2 orthologue, e.g.)and ADT (PicglADT3, PicglADT4), Additional File 9.Both CCoAOMT genes have also cis-eQTLs (AdditionalFile 9) that likely represent promoter polymorphismsinvolved in the differential expression of the genes. Ourfinding of cis-eQTLs connected by extensive trans-regu-latory interactions describes an example of epistasis thatis commonly observed in the regulation of pathway bio-synthetic genes [46].Arogenate Dehydratase and Related SequencesWhile arogenate dehydratase activity was first detectedin Nicotiana sylvestris [47], gene cloning and functionalcharacterization was only recently reported in Petuniahybrida [48]. Five putative white spruce arogenate dehy-dratases (ADT) were identified with similarity to charac-terized bacterial and fungal prephenate dehydratase(PD). They group together with A. thaliana and P.hybrida (Figure 3) implied in the catalysis of the laststep of the shikimate pathway specific for the biosynth-esis of phenylalanine (the entry molecule for the corephenylpropanoid pathway) (Figure 1). In addition, eightADT-like white spruce genes fall into two more remoteclusters of the small family (Figure 3).Four ADT genes as well as six ADT-like genes wererepresented as elements on our microarray. The ADT-like gene PicglADTL8 was identified as positional candi-date for the general resistance phenotype per se, Addi-tional File 1. In the co-expression network three ADTs(PicglADT2 WS00810_A15, PicglADT3 WS00921_B18,and PicglADT4 WS0261_E23) as well as one ADT-like(PicglADTL7 WS00937_M15) were present ("PD”,PD_4, PD_6, PD_12 and PD_9, respectively; Figure 4).Despite the central importance of ADT in generatingthe precursor molecule phenylalanine, ADTs were posi-tioned at the edges of long-range subgraphs in gene-gene interactions, Figure 4 (8), (12). PicglADT3 was sig-nificantly co-expressed with a class II 4CL geneWS01010_M10 (4CL_11), see below (Figure 4 (8), sup-porting its function as ADT. Expression of PicglADT2 isnegatively correlated with the cluster of constitutive diri-gents of the f and b/d class (DIR_2, DIR_32, DIR_31,DIR_20, DIR_18), Figure 4 (12). The ADT-like Pic-glADTL7 was co-expressed with OPCL PicglACL10(Figure 4, 4CL_3 WS00729_F23) suggesting both genesact in a related pathway.4-Coumarate:CoA-Ligase (4CL) and Related Acyl CoALigases (ACL)In the biosynthesis of lignin, soluble and wall boundphenolics and flavonids, 4-coumarate:CoA ligase (4CL;EC plays a gate keeper role as the enzyme gen-erates not only the CoA ester of coumaric acid in thelast step of the core phenylpropanoid pathway (Figure1), but the higher substituted coumarate derivatives,such as caffeic acid, ferulic acid and 5-hydroxyferulicacid, and sinapic acid in some angiosperms as well. 4CLis part of a family of adenylate-forming enzymes presentin all organisms, including an Arabidopsis gene recentlyshown to encode a functional 3-oxo-2(2’-[Z]-pentenyl)-cyclopentane-1-octanoic acid (OPC-8) CoA ligase(OPCL) catalyzing an essential step in jasmonic acidbiosynthesis [49,50]. The 4CL gene family is moderatelysized [14] and shows four bona fide 4CL genes forspruce (Figure 7). Three white spruce 4CL candidatesclustered closely with Arabidopsis 4CL3 (involved in fla-vonoid biosynthesis) and one spruce gene was foundorthologous to loblolly pine 4CL1 (for which a role inPorth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608Page 8 of 26the biosynthesis of lignin in compression wood wasimplied) (Figure 7) [51,52]. In the context of adenylate-forming acyl-CoA ligases, 11 white spruce candidatesfell into four divergent clusters, out of which four clo-sely grouped with the Arabidopsis OPCL, indicating asimilar function (Figure 7). While this represents a largesubfamily, our knowledge about the function of themembers is limited.Of 13 array elements with similarity to acyl-CoAligase genes, three represent sequence orthologs ofbona fide 4CL genes, while ten elements have highersimilarity to oxo-pentenyl-cyclopentane (OPCL), Addi-tional File 1 and Figure 7. Two spruce 4CL genes(Picgl4CL3 WS00112_J15 and Picsi4CL2WS01010_M10) were present in our co-expressionnetwork, Figure 4. They represent spruce classII 4CLgenes, which are involved in the formation of flavo-noids and soluble phenolics. Picgl4CL3 (4CL_1) wasdirectly co-expressed with two positional candidatesfor resistance traits (Picsi-PKS22 WS0014_M21 andLAR WS00929_C24), Figure 4 (1). Picsi4CL2 (4CL_11)is co-expressed with PicglADT3, Figure 4 (8). Theabsence of the classI representative Picgl4CL4 (Addi-tional File 1, Figure 7) and other lignin forming genes(CCoAOMT, and laccases implicated in constitutivelignifications [53]) along with the presence of othergene family members such as those involved in defensemechanisms like dirigents (a-family), P450s, polyketidesynthases and phenylpropanoid reductases (see alsoAdditional File 1) suggests higher importance of the0.1ArathOPCL1PicglACL1PicglACL4PicglACL6PgeACL7PicglACL8PgeACL9PicglACL10PicglACL12PicglACL13PicglACL14PgeACL15PicglACL16PicglACL17PgeACL18Arath4CL4Arath4CL2Arath4CL1Picgl4CL1Picsi4CL2Arath4CL3Picgl4CL3Pt4CL1Picgl4CL4************Figure 7 Acyl-CoA ligase family with 4-coumarate:CoA ligases (4CL) and oxo-pentenyl-cyclopentane (OPC) ligases, and related acyl-CoA ligases (ACL). The flash highlights functionally characterized polyketide synthases. The bar represents 0.1 amino acid changes. Asterisksindicate 80% and above confidence through bootstrap values. The dashed circle indicates the distinct clade of the bona fide 4CL. Arath,Arabidopsis thaliana; Picgl, white spruce (Picea glauca); Pge interior spruce (Picea glauca x engelmanii); Picsi, Sitka spruce (P. sitchensis); Pinta,loblolly pine (Pinus taeda); the red font indicates representation on the microarray. GenBank accession numbers are given in Additional File 14.Porth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608Page 9 of 26recovered gene-gene interactions for defense mechan-isms than for normal plant development.CCoAOMT Superfamily, Related to Caffeoyl-CoA O-MethyltransferaseCaffeoyl-CoA O-methyltransferase (CCoAOMT, EC2.1.1.104) catalyses O-methylation of the hydroxyl groupat the C3 position of the phenolic ring in conversion ofcaffeoyl-CoA to feruloyl-CoA (Figure 1) and has beencharacterized in many plants, including loblolly pine,where it is associated with developmental lignification[54]. Three white spruce genes are closely related with aloblolly pine CCoAOMT in the clade of bona fideCCoAOMT (Figure 8). The diversity observed in thewhite spruce CCoAOMT clade is the result of lineage-specific gene duplications that were retained throughoutevolution.All three genes involved in monolignol biosynthesiswere absent from the co-expression network. However,PicglOMT1, the spruce specific remote OMT(CCOMTL) (Additional File 11) that is putativelyinvolved in phenylpropanoid metabolism and a posi-tional candidate for Ht_1997 growth trait (AdditionalFile 1), was represented by two array spots(IS0014_O19, WS0022_P17) in the network. These twogenes termed CCoAOMT_1 and CCoAOMT_2 weredirectly connected to transcription factors (NAC_10WS00911_P24 and myb_5 WS00716_G01, respectively)as well as to a putative shikimate EPSPS synthaseWS0085_B05, Figure 4 (1). Furthermore, two catechol-O-methyltransferase-like genes WS0107_O19 andWS01013_K15 (Additional File 1) resided at the edgesof the network (CCoAOMT_9, CCoAOMT_8, Figure 4):The first clustered with various annotated genes0.1***StelpCCoAOMT*AtCCoAOMT2PoptrCAML1**ArathCOMTL4ArathCOMTL3ArathCOMTL7*PoptrCAML4PoptrCAML3***OrysaCCOMTL2OrysaCCOMTL3OrysaCCOMTL4OrysaCCOMTL5PhypaCCOMTL1PicglCCOMTL1OrysaCAML1ArathCOMTL6PoptrCAML2pArathCOMTL5**MescrCCoAOMTbona fide CCoAOMTmultifunctional CCoAOMTCCoAOMTL ICCoAOMTL II******0.1PhypaCCOMTL2MescrCCoAOMT1AtCCoAOMTPoptrCAM2PoptrCAM1OrysaCCoAOMT1PicglCCoAOMT1PicglCCoAOMT2PicsiCCoAOMT1PicsiCCoAOMT2PicglCCoAOMT3PintaCCoAOMTFigure 8 The caffeoyl-CoA O-methyltransferase family. Unrooted maximum likelihood phylogenetic tree. Highlighted are clusters of themultifunctional CCoAOMT, the subfamily of CCoAOMT-likes I and II, a rice specific subfamily without dicotyledon representatives and in the insetthe expanded bona fide CCoAOMT; Picgl, white spruce (P. glauca); Picsi, Sitka spruce (P. sitchensis); Pinta, loblolly pine (Pinus taeda); Arath,Arabidopsis thaliana; Poptr, poplar (Populus tremuloides); Orysa, rice (Oryza sativa); Phypa, Physcomitrella patens; Mescr, ice plant(Mesembryanthemum crystallinum); Stelp, Stellaria longipes. The scale bar represents 0.1 amino acid substitutions per site; asterisks indicate cladeswith bootstrap confidence greater than 80% and the flash highlights functionally characterized genes. The red font indicates representation onthe microarray. GenBank accession numbers are given in Additional File 14.Porth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608Page 10 of 26(PicglOMT-13 WS0261_A24, OPCL WS00729_F23,ADT-like WS00937_M15, PgeDIR5 WS00924_E04, seebelow and Additional File 1), while the second, a posi-tional candidate for the egg_2001 resistance trait, con-nected to genes with unknown functions. Thephylogeny for said catechol-O-methyltransferase (-like)sequences suggests they are ubiquitous genes that lack acounterpart in modern/angiosperm plants (AdditionalFile 12) and are part of a yet to be elucidated pathwayin conifer defenses.Polyketide Synthases Related to Chalcone and StilbeneSynthases (CHS/STS)Stilbene synthase (STS, EC and chalconesynthase (CHS, EC are plant-specific polyketidesynthases at the entry of stilbenoid and flavonoid bio-syntheses, respectively (Figure 1). Both types of enzymesperform a sequential condensation of three acetate unitsto a CoA-ester to form an intermediate that is foldedinto the aromatic ring systems of naringenin chalconeor the stilbene backbone [55,56]. While flavonoids,which play a vital role as UV protective pigments inplants, are ubiquitous in the land plant kingdom andwere reported in the basal lineages of liverworts andmosses [57], stilbenes have so far been detected in rela-tively few plant families where they contribute to theresistance of woody tissues to degradation and act asphytoalexins. The main constitutive stilbene glycosidesin Picea species are astringin and isorhapontin [58].Fourteen spruce CHS-like sequences form a tight clusterwith the Japanese red pine CHS, indicating severalduplications of functionally related polyketide synthases(Figure 9). While the ancestor of modern polyketidesynthases predates the evolution of gymnosperms andangiosperms, the close relationship of STS and CHSwithin the angiosperms and gymnosperms indicates thatthese functions evolved independently several times. Ingymnosperms distinct clades of both pine and spruceorthologues are found for STS and for CHS, consistentwith the presence of both functions in a common ances-tor of these lineages.Strong collocation of gene expression and trait varia-tion identified two CHS genes Picsi-PKS21 and Picsi-PKS22 as well as the STS gene Picgl-PKS6 as potentialcandidates for weevil resistance traits (sum_egg,egg_2001 and egg_2001, respectively), Additional File 1.Picsi-PKS22 WS0014_M21 is present within the net-work (CHS_1) and co-expressed with Picgl4CL3 and aperoxidase WS0063_P06 that co-segregated with thesum_atk resistance trait (Figure 4 (1)). Picgl-PKS6WS00925_G22, (CHS_11, Figure 4 (5)) was negativelyco-expressed with peroxidase PicsiPRX2 WS0074_A10that responded to wounding ([59]). The CHS geneWS0024_K18 (termed CHS_3) from the Picsi-PKS14cluster is associated with a myb transcription factor(myb_32, Figure 4 (3)).Picgl-PKS2 WS00731_E22 (see also Figure 9) wasnegatively co-expressed with a-family dirigent PgeDIR2WS00911_I09 (Figure 4 (11), CHS_8 and DIR_10). Theexpression of PgeDIR2 was significantly triggered byinsect feeding [29], see below. Interestingly, Picgl-PKS2was co-expressed with an ANAC057 transcription factorWS00929_E05 (NAC_14, Figure 4 (11)) that accumu-lated with increasing growth rate (unpublished). Thecontrasting expression pattern found for diverse mem-bers of the polyketide synthase gene family reflects thedifferential gene regulation for individual members.O-Methyltransferase Superfamily, Related to Caffeic Acid/Coniferaldehyde O-Methyltransferase (COMT)A total 37 of OMT/COMTL candidates were identifiedin white spruce through mining of the Treenomix ESTdatabase (Spruce V8 database will be published else-where, K. Ritland, pers. com.), relative to 17 genes inArabidopsis, eight in poplar, and six in rice (Figure 10).The founding members of the O-methyl transferasesuperfamily (OMT, EC involved in the synthesisof methylated plant phenolics are related to the forma-tion of the angiosperm monolignol precursor of S-lignin.However, the S-lignin pathway and sinapic acid-derivedmetabolites, such as syringyl lignins, are absent in gym-nosperms (Figure 1). Even though they are broadlyrepresented in the OMT/COMTL superfamily, evidencefor dedicated bona fide COMT orthologues in whitespruce is missing.OMTs were among phenylpropanoid gene familiesmost completely represented in the co-expression net-work (Figure 4). Three COMTLs (PicglOMT-11WS00715_G04/WS0261_F17, PicglOMT-10WS0023_M11, PicglOMT-7 WS02611_F21) similar tothe stress inducible O-methyltransferase from scots pine([17]; clade I, Figure 10 and 11) were highly co-expressed, OMT_5, OMT_21, OMT_22, OMT_1, Figure4 (2). Their expression was negatively correlated withthe expression of a sequence (WS00716_K11) thatshares similarity to AtPER16/AtPER45 and is a posi-tional candidate for the Hgt1999 trait (Additional File1). Five spruce OMTs (WS0071_H13, WS0078_K09,WS0104_J07, WS0046_C03, and WS01013_K11) thatare part of the lineage specific clade in the plant OMT/COMTL superfamily (Figure 10) cluster within the net-work (OMT_4, OMT_10, OMT_19, OMT_3, andOMT_18, Figure 4 (3)) and were co-expressed with atranscription factor that weakly resembles the secondarywall-associated AtMYB83 (WS0079_B12). PicglOMT-1WS0099_A18, with strong similarity to a beta-alaninebetaine synthase (clade III, Figure 10 and 11), representsa positional candidate for the egg_2000 trait and is co-Porth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608Page 11 of 26expressed with PgeDIR5, a dirigent of the a-subfamily,DIR_15, Figure 4 (4). Although their classification hereis based on phylogenetic analysis alone, and biochemicalcharacterization needs to support the annotation, sprucegenes of the OMT/COMTL superfamily possibly extentthe repertoire of spruce defense mechanisms.Cytochrome P450s Involved in Oxygenation ofPhenylpropanoidsIn addition to a major plant carbon sink - the ligninbiosynthesis - cytochromes P450 contribute to the bio-synthesis of many bioactive phenolic derivatives (for areview see [60]). In the families of the cinnamte hydro-xylase (C4H, CYP73) and coumaroyl-shikimate 3’-hydro-xylase (C3’H, CYP98) two and one white sprucecandidate were identified, corresponding to single genesin Arabidopsis, while no white spruce coniferylaldehyde5-hydroxylase (CA5H, CYP84) sequence orthologueswere found. In the CYP75 family, seven diverse whitespruce candidates form two distinct clusters with theArabidopsis CYP75B1 encoding a functional F3’H [61].Phylogenetically related, but without biochemical func-tion assigned, 45 white spruce candidates constitute theunusual large and diverse conifer specific CYP750PindePSS30.1PicglPKS15PicglPKS14PicglPKS18PicglPKS17PicsiPKS12PicsiPKS35PicglPKS33PicglPKS13**PicabPKS11PindeCHSPicglPKS16PicsiPKS22PicglPKS19PicglPKS20PicsiPKS21***PicsiPKS8PicglPKS10PicglPKS7PicglPKS5PicglPKS9PicglPKS6*PindePSS1PintaPKSPindePSS2**PicsiPKS1PicglPKS2***ArathCHSVitviPKS42VitviRVS*SorbiSTSSorbiCHS*PoptrPKS36VitviPKS41*HypanCHSPoptrPKS38PoptrPKS37*ArahyPKS43ArahySTS**CHS and STSCHSSTS?AngiospermGymnospermFigure 9 Phylogeny of higher land plant polyketide synthases with chalcone synthases (CHS), and stilbene synthases (STS). Arath,Arabidopsis thaliana; Poptr, poplar (Populus tremuloides); Vitvi, grape vine (Vitis vinifera); Pinde, Japanese red pine (Pinus densiflora); Arahy, peanut(Arachis hypogaea); Hypan, Hypericum androsaemum; Marpo, liverwort (Marchantia polymorpha); Sorbi, sorghum (Sorghum bicolor); Picab, Norwayspruce (Picea abies); Picgl, white spruce (P. glauca); Picsi, Sitka spruce (P. sitchensis); Pinta, Loblolly pine (Pinus taeda); The flash highlightsfunctionally characterized polyketide synthases. The bar represents 0.1 amino acid changes. Asterisks indicate 80% and above confidencethrough bootstrap values and the flash highlights functionally characterized genes. Dashed boxes indicate distinct clades with sub-families. Thetree was rooted using the distantly related liverwort stilbenecarboxylate synthase and moss chalcone synthase-like sequences as outgroup (notshown). The red font indicates representation on the microarray. GenBank accession numbers are given in Additional File 14.Porth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608Page 12 of 26family, an outgroup to angiosperm CYP84. Anothergroup of P450-dependent monooxygenases forms thespruce-specific outgroup of nine CYP76-likes (Addi-tional File 8).Within CYP75/F3’H sequences, family member C59represents a positional candidate for the sum_egg resis-tance trait. Another representative (C57, WS00931_D17)is present in our co-expression network (Figure 4 (5))and is negatively correlated with NAC_1(WS00713_M11), a NAC transcription factor that is acandidate for the atk_2001 resistance trait. The functionof the diverse CYP750 family (Additional File 8) isunknown, however, the sequence relationship withCYP84 (F5H), CYP75(F3’5’H/F3’H), CYP98 (C3H), andCYP73 (C4H) which are involved in oxidation reactionsof the phenylpropanoid metabolism indicates that these0.1PinsyPMTPicglOMT3-12PicglOMT2PicglOMT10PicglOMT7OrysaCOMTPoptrCOMT2PoptrCOMT1ArathCOMTArathCCOMTL5PoptrCOMTL3PoptrCOMTL1PoptrCOMTL2pArathCOMTL16ArathCOMTL11ArathCOMTL10ArathCOMTL15ArathCOMTL8ArathCOMTL7ArathCOMTL9ArathCOMTL1ArathCOMTL4ArathCOMTL3ArathCOMTL2ArathCOMTL6AmmmaBOMTOrysaOMTL3OrysaOMTL2OrysaOMTL1PicglOMT1OrysaOMTL4LimlaBANMTArathCOMTL12PoptrCOMTL4PicglOMT19-24PicglOMT19PicglOMT15PicglOMT13PintaAEOMTPicglOMT18PicglOMT17PapsoROMTRoschEOMTPoptrCOMTL6PoptrCOMTL5ArathCOMTL13ArathCOMTL14IIIAIII******************* ****** *******PicglOMT14PicglOMT16PicglOMT28PicglOMT27PicglOMT26PicglOMT25 PicglOMT30PicglOMT32-37PicglOMT32PicglOMT35PicglOMT36PicglOMT37PicglOMT31PicglOMT29Figure 10 The O-methyltransferase superfamily family with caffeic acid 3-O-methyltransferase (COMT), pinosylvin O-methyltransferase(PMT), eugenol O-methyltransferase (EOMT), reticulin O-methyltransferase (ROMT), beta alanine N-methyl transferase (BANMT), andbergaptol O-methyltransferase (BOMT). A Unrooted maximum likelihood phylogenetic tree; Picgl, white spruce (P. glauca); Pinta, loblolly pine(Pinus taeda); Arath, Arabidopsis thaliana; Poptr, poplar (Populus tremuloides); Orysa, rice (Oryza sativa); Ammma, bullwort (Ammi majus); Limla,Limonium latifolium; Rosch, china rose (Rosa chinensis); Pinsy, scots pine (Pinus sylvestris); Papso, poppy (Papaver somniferum); Highlighted arefunctionally distinct clusters of COMT/OMT, with (I), (II), and (III) expanded in Figure 11, B; the scale bar represents 0.1 amino acid substitutionsper site; asterisks indicate clades with bootstap confidence greater than 80% and the flash highlights functionally characterized genes. The redfont indicates representation on the microarray. GenBank accession numbers are given in Additional File 14.Porth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608Page 13 of 26families and CYP750s may share a common progenitorand possibly a structurally similar substrate of the phe-nylpropanoid class. On our array 11 spruce specificCYP750 genes were represented (Additional File 1), oneCYP750 (C24) was identified in our study as a positionalcandidate for the weevil resistance phenotype per se, andthree CYP750s (C18, C12 and C56) are found in the co-expression network. C18 (F5H_5, WS00934_G23) wasco-expressed with a candidate (LAR similarity, DFR_3)for the general resistance phenotype (Figure 4 (7)). C12(F5H_3, WS00923_E07) was negatively correlated withmyb myb_1 (WS00917_H19) that is loosely related toTT2, a putative regulator of proanthocyanidin biosynth-esis [62], and in our study a positional candidate forboth atk_2000 and sum_atk resistance traits (AdditionalFile 1, Figure 4 (8)). Expression patterns for bothCYP750s suggest diverged, though unknown, functionsconsistent with their position in different subclades(Additional File 8). The third CYP750 C56 (F3’H_8,WS00922_H05) was negatively correlated to a cluster ofco-expressed peroxidase genes of which some areimplied in monolignol polymerization (PicabPRX18IS0011_F24 (PRXR_1), spruce ArathPER42WS0031_G05 (PRXR_13), and ArathPER64WS01016_D12 (PRXR_58) orthologues, see elsewhere)(Figure 4 (8)). Future characterization of spruce CYP750candidates will need to uncover their enzymatic functionin planta.0.1PinsyPMTPicglOMT3PicglOMT2PicglOMT8PicglOMT5PicglOMT4PicglOMT10PicglOMT6PicglOMT12PicglOMT7PicglOMT9PicglOMT11OrysaCOMTPoptrCOMT2PoptrCOMT1ArathCOMTArathCOMTL5PoptrOMTL3PoptrOMTL1PoptrOMTL2pIPintaAEOMTPicglOMT18PicglOMT17PapsoROMTRoschEOMTPoptrOMTL6PoptrOMTL5ArathOMTL13ArathOMTL14IIPicglOMT1OrysaCOMTL4LimlaBANMTArathCOMTL12PoptrCOMTL4IIIBFigure 11 The O-methyltransferase superfamily family with caffeic acid 3-O-methyltransferase (COMT), pinosylvin O-methyltransferase(PMT), eugenol O-methyltransferase (EOMT), reticulin O-methyltransferase (ROMT), beta alanine N-methyl transferase (BANMT), andbergaptol O-methyltransferase (BOMT). B expanded excerpts of functionally distinct clusters (I), (II), and (III) from Figure 10, A; Picgl, whitespruce (P. glauca); Pinta, loblolly pine (Pinus taeda); Arath, Arabidopsis thaliana; Poptr, poplar (Populus tremuloides); Orysa, rice (Oryza sativa);Ammma, bullwort (Ammi majus); Limla, Limonium latifolium; Rosch, china rose (Rosa chinensis); Pinsy, scots pine (Pinus sylvestris); Papso, poppy(Papaver somniferum); the scale bar represents 0.1 amino acid substitutions per site; the flash highlights functionally characterized genes.GenBank accession numbers are given in Additional File 14.Porth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608Page 14 of 26Phenylpropanoid ReductasesPhenylpropanoid reductases are a class of closely relatedenzymes that include leucoanthocyanidin and isoflavonereductases (LAR/IFR), pinoresinol and lariciresinolreductasess (PLR), phenylcoumaran benzylic etherreductase (PBR), and isoeugenol synthase (IES) thathave been shown to participate in the biosynthesis of aplethora of constitutive and induced defense relatedphenylpropanoids and phytoalexins, such as lignans andisoflavans [63-67]. With the exception of the IFR family,for which no white spruce homologues were identified,at least two white spruce candidates were detected inthe other families (Figure 12A). A total of 12 whitespruce homologs cluster closely with the loblolly pinePBR1 (Figure 12B) which reduces the benzylic etherfunctionalities of both dehydrodiconiferyl alcohol anddihydrodehydrodiconiferyl alcohol in the formation of 8-5’-linked lignans [66]. However, the phylogenetic dis-tance of the white spruce candidates to the functionallycharacterized angiosperm and gymnosperm phenylpro-panoid reductases does not allow functional predictionsextending beyond similarity of the mechanism of thecatalyzed reaction.Out of 21 array elements annotated as phenylpropa-noid reductases 11 were present within the co-expres-sion network ("PCBER”, Figure 4). Gene expressionvariation of PicglPPR05 (represented by three array ele-ments: WS00723_J06, WS0048_K01, WS00920_D17(Figure 4 (9)): PCBER_6, PCBER_3, PCBER_16) was co-expressed with the secondary wall-associated AtMYB20(myb_19; WS0071_H14), which was a candidate for theldr99 growth trait. PicglPPR21 is a positional candidatefor the sum_egg resistance trait; its transcript abundancewas lowest in the spruce mapping family with the mostvigorous growth and highest weevil attack rate (unpub-lished). In the network this PCBER gene (WS0086_D12;PCBER_11) was connected to PicglDIR10 (WS0261_J16;DIR_30), a member of the b/d-subfamily and the clusterof peroxidases involving monolignol forming enzymes(AtPER64, AtPER42, PicabPRX18, see above, andPopalPRX WS0099_B17 (PRXR_51)), Figure 4 (8). Tran-scripts of PicglPPR13 (WS01011_J14) accumulated withincreasing growth rate, and PicglPPR13 (PCBER_19) wasnegatively co-expressed with the resistance phenotypecandidate LAR WS00725_B17, DFR_3, Figure 4 (7). Thelatter is correlated with another (remote) LAR represen-tative WS00926_B24 (DFR_6) that was lowest expressedin the spruce mapping family 26, the stress induciblePicabPRX2 WS00928_G19 (PRXR_45), and the clusterof dirigents involved in constitutive defenses, DIR_2,DIR_32, DIR_31, DIR_18, DIR_20, Figure 4 (12), seealso below. The expression pattern for PicglPPR13 sug-gests that this PCBER could be involved either in sus-ceptibility or tolerance to insect feeding. However, forother representatives (PicglPPR21) their co-expressionpattern suggests involvement in constitutive defenses.Dirigent ProteinsA recent study on Sitka spruce dirigent-like proteins(DIR) [29] reported 35 unique DIR or DIR-like genes.For our survey, we were able to use expression datafrom 23 elements annotated as unique DIRs and repre-sented on the microarray covering distinct a-, b/d-, andf-subfamilies (Additional File 1). The DIR f-subfamily isspruce specific, while angiosperm members can befound for both a- and b/d- subfamilies, [29] showed thatDIRs from different clades have distinct gene expressionpatterns with a suggested role in induced defense toweevil feeding for members of the a- subfamily. Repre-sentatives of b/d- and f-subfamilies are more likelyinvolved in primary processes or constitutive defense(ibidem).Within the network PicglDIR12 WS00815_A07 (b/d-subfamily, DIR_7) was co-expressed with a growth traitassociated transcription factor resembling secondarywall-associated AtMYB20 (myb_19) and with a clusterof six putative laccases, Figure 4 (9). A cluster of consti-tutive dirigents involving b/d-subfamily members (Picsi-DIR17 WS01012_J06, PicsiDIR11 WS01012_K18,PicsiDIR21 WS02610_M19, PicglDIR7 WS0262_G08(Additional File 1)), and PicsiDIR26 WS0058_H22 fromf-subfamily were co-expressed with remote LAR(DFR_3), one of the identified positional candidates forweevil resistance, Figure 4 (12).PgeDIR13 (WS00914_H24) was highly up-regulated inbark tissue following weevil feeding and for PgeDIR2(WS00911_I09), another a-subfamily member, the high-est induction was registered, [29]. In the network, Pge-DIR13 (DIR_11) was co-expressed with two other a-subfamily members (PicsiDIR19 WS01011_J07, Picsi-DIR16 WS01032_M02), DIR_17, DIR_25, Figure 4 (10),and for PgeDIR2 gene expression was negatively regu-lated with the expression of the polyketide synthasePicgl-PKS2, CHS_8, Figure 4 (11). Thus, our experimentconfirms previous observations that dirigent subfamiliesin constitutive defenses are differentially regulated thandirigents in induced defenses. Similar to the cluster ofspruce/conifer-specific members from the OMT super-family (OMTL), see above, dirigents from the same sub-family are also strongly co-expressed. This providesevidence for the presence of a multitude of gene copieswith the same or highly similar protein functions thatact in related pathways.Class III PeroxidasesPlant class III peroxidases represent a key multifunc-tional enzyme family that is involved in such diverseprocesses as auxin catabolism, defenses, general stressPorth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608Page 15 of 26AB0.1PicglPPR04PicglPPR13PintaPBR1*PicglPPR15PicglPPR05LotjaPTRArathIFRH1*ArathIFRH3ArathIFRH2*PicglPPR09PicglPPR18ThupPLRPicglPPR19ArathPPR2ArathPPR1**PethyIES1PicglPPR11PicglPPR03**PicglPPR06VitviLAR1PicglPPR02PgePPR20***0.1PicglPPR04PicglPPR13PicglPPR12PintaPBR1PicglPPR14*PicglPPR10PicglPPR01PicglPPR08*PicglPPR16PicglPPR17PicglPPR07**PicglPPR15PicglPPR05*PicglPPR1/7/8/10/12/14/16/17PicglPPR21Figure 12 The phenylpropanoid reductase family with leucoanthocyanidin reductase (LAR), pinoresinol-lariciresinol reductase (PLR),phenylcoumaran benzylic ether reductase (PBR), and isoeugenol synthase (IES). A Unrooted maximum likelyhood phylogenetic tree(isoflavonone reductase homolog, IFRH); B Expanded conifer phenylcoumaran benzylic ether reductase family, rooted with Lotus japonicuspterocarpan reductase. Thupl, western red cedar (Thuja plicata); Lotja, Lotus japonicus; Pinta, loblolly pine (Pinus taeda); Vitvi, grape vine (Vitisvinifera); Arath, Arabidopsis thaliana; Picgl, white spruce (Picea glauca); Pge interior spruce (Picea glauca x engelmanii); the flash highlightsfunctionally characterized genes; asterisks indicate 80% and higher bootstrap values; the scale represents 0.1 amino acid changes. The red fontindicates representation on the microarray. GenBank accession numbers are given in Additional File 14.Porth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608Page 16 of 26responses, and lignification [68]. Peroxidases were sug-gested to function in the radical reaction of monolignolsto the polymer lignin [69]. Misregulation of peroxidasesprovided indications for altered lignin composition ([70],review of xylem class III peroxidases in lignification:[71] and [72]). Recently, two lignin-forming peroxidaseshave been identified and characterized from Norwayspruce suspension culture showing substrate preferencefor H- or G-lignin precursor molecules, respectively[15]. To date, 17 peroxidase unigenes for conifer ligninpolymerization have been identified and their expressionabundance in different tissues and after biotic and abio-tic challenges studied [73].For eight genes (PicabPRX1, PicabPRX2, PicabPRX3,PicabPRX4, PicabPRX6/PicabPRX7 cluster, PicabPRX16/17, PicabPRX18, PicabPIPRX) the respective whitespruce orthologues were represented on the microarray(Additional File 1, Additional File 13). While some per-oxidases with proposed roles in lignin polymerizationare related to developmental lignification (PicabPRX1,PicabPRX4, PicabPRX16/17 cluster, PicabPRX18, arrayelements: see Additional File 1), others were stress indu-cible (PicabPRX2, PicabPRX3, PicabPIPRX, elements:Additional File 1) and reflect the specific stressresponses of peroxidases ([73,15]). In addition, geneexpression of two other inducible peroxidases (Pic-siPRX1 and PicsiPRX2, [59]) were surveyed (AdditionalFile 1). PicsiPRX1 (WS0012_J09, PRXR_4) was co-expressed with several other peroxidases, Figure 4 (6).Constitutive expression levels of PicsiPRX2(WS0074_A10) increased with higher inherent growthrate of individuals (unpublished). PicsiPRX2 expression(PRXR_26) was negatively correlated with expression ofstilbene synthase Picgl-PKS6 (CHS_11), a positional can-didate for the egg_2001 resistance trait (Figure 4 (5)).We suggest that PicsiPRX2 and Picgl-PKS6 are involvedin different defense strategies. The white sprucePicabPRX2 orthologue WS00928_G19 is significantlyassociated with the sum_egg resistance trait, and is alsoco-expressed with the remote leucoanthocyanin reduc-tase, one positional candidate for the “resistance pheno-type” per se, (PRXR_45, DFR_3, Figure 4 (7), AdditionalFile 1).The ability to bind lignin was demonstrated in vitrofor PicabPRX18 [73]. The white spruce orthologue ofPicabPRX18 is a positional candidate for Hgt1999growth trait. PicabPRX16/17 is a positional candidatefor Hgt1997 trait. A co-expression cluster involving thewhite spruce orthologue of PopalPRX WS0099_B17, aputative spruce homologue of the secondary wall-asso-ciated transcription factor AtMYB83 WS0051_I02 andS-adenosylmethionine synthetase (SAMS) WS0261_C18were positional candidates for Hgt_1995 trait (PRXR_51,myb_47, SAMS_6, Figure 4 (8)), Additional File 1. Also,the PicabPRX4 orthologue WS00912_J06 (PRXR_37)was co-expressed with a cluster of peroxidases withgrowth associations (Figure 4 (5), Additional File 1),among them PicabPRX6/PicabPRX7 (WS00910_L21/WS01025_F04: PRXR_36, PRXR_69). Interestingly, tran-script abundance for all genes in this cluster decreasedwith growth rate. This suggests that a number of peroxi-dases are important in reinforcement of anatomicalstructures (via lignification, e.g.). Similarly, lignin biosyn-thetic genes are down-regulated in fast growing indivi-duals [38].DiscussionThe stem-borer Pissodes strobi damages the host tree byattacking the shoot apical leader of the previous year’sgrowth. The feeding larvae typically consume phloemtissue. In the spruce bark, the secondary phloem andthe cambium are very active chemical defense produc-tion zones that include secondary metabolites such asthe polyphenolics that are abundantly present in theparenchyma. Pre-formed defenses are established co-ordinatedly during the development of secondary xylemin the apical shoot [22].We studied the complexity of phenylpropanoid-relatedgene families and gene regulation of individual familymembers. We found evidence for new pathways in coni-fer defenses that involve genes without counterparts inmodern plants that need to be further investigated.Employing genetical genomics, we showed that thisapproach is valuable for elucidating interactions betweengenes that originate from complex families with multiplerelated sequences. By including weevil resistance andheight growth traits in the quantitative analysis, weidentified eight genomic regions with extensive accumu-lation of phenotypic variation from multiple traits coin-ciding with hotspots of transcript abundance variation.The identified master regulons ranked hierarchicallyhighest among all genotyped loci, since they are likelyresponsible for massive gene expression variation in thepathway and were linked to changes in the studiedphenotypes.Phenylpropanoid and Related Gene Families in SprucePhenylpropanoids fulfill important functions as defensecompounds, e.g. simple hydroxycinnamic acids andmonolignols or the structurally complex flavonoids, iso-flavonoids, and stilbenes. They have overlapping roles assignaling molecules in plant development and plantdefense [74]. The conifer phenylpropanoid pathway ischaracterized by a higher level of complexity comparedto angiosperms ([14] and this study) that might includeseveral not yet resolved alternative pathways. Theexpansion of phenylpropanoid and related genes intomulti-gene families serves either the independentPorth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608Page 17 of 26regulation of the biosynthesis of different classes ofcompounds or the gene dosage through massive accu-mulation of these compounds [74]. For example, weidentified several associations with weevil resistancetraits for individual members of the NADPH reductasegene family with a range of product specificity of relatedphenylpropanoid-derived plant defense compounds (Fig-ure 12). We also identified associations for memberswithin two spruce- (or generally conifer-) specificgroups massively represented in the OMT superfamily(Figure 10) and the phenylpropanoid P450s (clade ofCYP750, Additional File 8), respectively, both withentirely unknown functions (Additional File 1).Generally, plant specialized metabolism is character-ized by gene duplication events and gene diversificationsleading to modified/optimized product specificity andmodified tissue specific expression [75]. However, it iscurrently unclear whether adaptive or non-adaptiveforces are more important for the maintenance of geneduplicates [34]. In brief, new pathways were establishedamong different plant species by independent recruit-ment and inactivation of biosynthetic enzymes throughdownregulation or loss of certain genes of an ancestralpathway [75]. Interestingly, it has been shown thatCCoAOMT-suppression in pine, unlike in angiosperms,generates only moderate reduction in lignin due to com-pensatory reactions that lead to the incorporation ofunusual monolignols (catechyl units) [76]. In the O-methyltransferase family (CCoAOMT sequence similar-ity) we also found one gene with high similarity to func-tionally characterized catechol-OMT conserved acrossspecies of the eukaryotic and prokaryotic kingdoms, butwas lacking in the angiosperms (Additional File 12).In total, 44 positional candidate genes were directlyassociated with resistance traits (Additional File 1). Posi-tional candidate genes were prominently identified inexpanded gene families such as dirigents and peroxi-dases, for which some members are implicated in down-stream reactions of lignan/lignin formation. Plant classIII peroxidases are a key multifunctional enzyme familythat is involved in diverse developmental and defenseprocesses ([68,71]). Evidence that certain peroxidasesmay function in lignin polymerization came from genemisexpression experiments that altered lignin composi-tion [70]. In several gene families of the core ligninpathway, members exist that can function in bothstress/elicitor response and developmental lignification[77]. However, given the individual families sizes, phe-nylpropanoid reductase, CYP750, CHS/STS, and OMTgene families showed the highest proportion of posi-tional candidate genes for resistance. Spruce genes withsequence similarities to COMT (caffeic acid O-methyl-transferase) and F5H (ferulate-5-hydroxylase), twoenzymes involved in the syringyl (S-) lignin formation inangiosperms, were identified. Due to the lack of thesepathways in gymnosperms and to their low sequencesimilarity, these sequences were annotated as OMTsputatively involved in stilbenoid synthesis and P450sputatively acting in flavonoid synthesis, respectively (seeabove). It is likely that enzymatic reactions in mono-lignol/lignin biosynthesis are adopted from conservedancestral pathways that originally functioned in protec-tion against microbial infection or UV radiation. Inter-estingly, those reactions emerged in different plantlineages by convergent evolution [78].Gene Expression Regulation of Gene Family Members inthe Phenylpropanoid PathwayMicroarray experiments that aim to elucidate geneexpression differences between treatment and controlgroups usually involve reference or related designs in asmall sample set and cross-hybridizations between genespots with high nucleotide similarity complicate datainterpretation. However, the specific statistical proce-dure employed in genetical genomics that studies largeexperimental populations of randomized genetic back-ground reduces this effect. Positions of eQTLs locategenomic regions that harbor regulatory elements thatcontrol the expression of a single gene or a subset ofgenes acting in the same pathway. In the case of cis-reg-ulation, the genomic location of the eQTL coincideswith the physical location of the regulated gene, whiletrans-acting eQTLs identify regulatory elements for thegene elsewhere in the genome. Distribution of eQTLsmay spread evenly on the genome or appear in clustersor in “hotspots” depending on the genetic architectureof gene interactions. In this way, master regulonsinvolved in multiple traits (i.e., pleiotropic genes) can beidentified, and the amount of epistasis from interactingloci can be uncovered [39]. We used this approach toidentify spruce genes that are potentially important forconstitutive defense mechanisms and to establish resis-tance against Pissodes strobi.Apical leaders of 188 Picea glauca individuals werestudied for gene expression changes in phenylpropanoidgenes prioritized from our 21.8 k spruce EST chip (seeMaterial and Methods for details). Genomic regionswith clustered expression and phenotypic trait variationmight contain master regulons ([46,79]). Regions under-lying trans-eQTL hotspots have pleiotropic geneticbackground (Figure 5). Loci that accumulate pQTLs ofboth growth and resistance traits may be involved in fit-ness trade-offs. We identified two such eQTL hotspotson the spruce genome (LG 2 and LG 3). The thoroughfunctional analysis of such largely pleiotropic genes ishampered because their knock-out mutants likely exhi-bit lethal or highly deleterious phenotypes [46]. How-ever, recently, a locus was characterized which conveysPorth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608Page 18 of 26plant resistance to a wide spectrum of pathogens andherbivores and at the same time reduces vegetativegrowth [80]. The dirigents represent a candidate genefamily for which some members were associated withboth growth and resistance trait variation, e.g., expres-sion variation for PicsiDIR17 (b/d-family) as well as Pic-siDIR31 (f-family) mapped onto eQTL hotspotsassociated with growth traits and resistance traits (Addi-tional File 9).Phylogeny and the analysis of the underlying geneexpression pattern identified functional divergenceamong members of gene superfamilies. In several casesgene expression of family members was co-regulatedwith variation in weevil resistance. eQTLs that accumu-lated at resistance hotspots were generated fromdiverged gene family members of CCoAOMT (PicglC-CoAOMT1, PicglCCoAOMT2, PicglCCoAOMT3, seealso below), OMT (AEOMT and group C genes), PRXR(PicabPIPRX, PicabPRX1, PicabPRX2, PicabPRX18, e.g.),CYP450 (CYP75/F3’H, F3’5’H/F3H, CYP750), DIR (a-and f-family, e.g.), the PCBER (LAR, PCBER), and 4CL(acyl-CoA ligase, OPCL, 4CL) (Additional File 9). It hasrecently been shown that tandem gene duplication andsubsequent gene retention is particularly common forbiotic stress genes [81]. This finding indicates that theincreased rate of gene duplication and diversificationoffers the potential for adaptive evolution and diver-gence. Thus, the neo-and subfunctionalizations found inthe multi-gene families of the spruce phenylpropanoidpathway are possibly highly important for beneficialphenotypic innovations.Linkage groups on our spruce QTL map wereassigned following [82] and the spruce-loblolly pinecomparative mapping project (C. Liewlaksaneeyanawin,pers. comm; Additional File 2). Therefore, valuablecross-species comparisons for QTL positions are avail-able for Pinaceae, e.g., the two bona fide CCoAOMTgenes CCoAOMT-1 and CCoAOMT-2 mapped to thespruce LG6 (Additional Files 2 and 9, Figure 5), in syn-teny with the recently published loblolly pine [83].Not only did the two bona fide CCoAOMT genes har-bor eQTL hotspots but also overlapped with QTLregions for the traits atk_2000, egg_2000, and sum_egg(Figure 5). This suggests a close association of mono-lignol formation with defense against the stem borer Pis-sodes strobi. It is known that cell wall-integral ligninprovides a physical barrier for invading pests. Specifi-cally, the lignified parenchyma cells are the importantpre-formed anatomical structures that are involved insuch constitutive defenses [21]. In addition, localized denovo synthesis of lignin with structural similarity toearly developmental lignins occurs in response to stressand is associated with wound healing ([84,85,6,86]). Atthe two CCoAOMT loci, eQTLs were mostly generatedfrom genes of the shikimate pathway, monolignol bio-synthesis and downstream condensation reactions, lig-nan formation, flavonoid biosynthesis, andmultifunctional OMT activity. The genomic region har-boring the cluster of CCoAOMT genes flags its generalimportance. We found in our large-scale transcriptomicsstudy that 36% and 53% of the mapped trans-eQTLsgenerated from 1307 and 992 genes, respectively, werecommonly shared between both CCoAOMT loci. Cer-tain gene ontology categories were overrepresented andcommonly shared between the two eQTL cluster (sec-ondary metabolic process, response to biotic stimulus, e.g.). This suggests interactions between both CCoAOMTloci and other genes that extend beyond the phenylpro-panoid metabolism. Further investigations will substanti-ate this in detail.In contrast to CCoAOMT-1 and CCoAOMT-2, asso-ciation with extensive phenotypic variation is currentlyunknown for CCoAOMT-3. Yet, the three CCoAOMTgenes represent our most comprehensive example of dif-ferential gene regulation for duplicated genes (the phylo-geny of the family is given in Figure 8). No trans-eQTLfrom CCoAOMT-1 mapped onto CCoAOMT-2 andvice versa; however, the third CCoAOMT gene exhibitedtrans-eQTLs at both loci, suggesting a lower regulatoryhierarchy for CCoAOMT-3 among all three genes (Fig-ure 8). CCoAOMT-1 expression variation also mappedonto the growth trait associated eQTL hotspot onLG13, and, generally, the transcript abundance ofCCoAOMT-1 was increased with higher inherentgrowth rate in the experimental population (not shown).CCoAOMT-3 is a positional candidate for hgt1999 trait.At locus CCoAOMT-2 eQTLs from ERFs linked todefensive gene expression (ERFs from group IX [87],specifically, transcription repressors (ERF3, ERF4, ERF7[88])) accumulated (unpublished results). Consequently,CCoAOMT-2 could indeed function in constitutiveresistance. The identified regulators (ERFs) were shownto affect important constitutively expressed defensegenes such as basic chitinase and glycosyl hydrolasefamily 17 protein (beta-1,3-glucanase, [88]). The threemonolignol biosynthesis genes were likely retainedthroughout evolution because of their individual impor-tance and their temporally and spatially dependentrecruitment for lignin formation: lignin that is constitu-tively expressed and related to primary processes, ligninthat is wound inducible and participates in activedefense, or lignin that is deposited for wound healing.ConclusionsThis study describes the genetics of pest resistance andutilizes genetical genomics to elucidate the genetic basisand evolution of phenolics based insect resistance in acommercially valuable conifer tree species. The resultsPorth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608Page 19 of 26add to recent QTL studies that have dissected mechan-isms underlying the genetic control of wood traits andvolume growth in eucalyptus but were completely lack-ing testing of biotic stress resistance, and in particular,for conifers [89]. We utilized eQTL mapping to conducta pre-selection of the candidate genes by testing for cor-relations with resistance traits. The findings can facili-tate research efforts on targeted gene association studiesspecific to nonstructured conifer populations aimed atresolving the genetic basis of host resistance toherbivory.Our work also provides a substantial account and per-spective of the functional characterization of severalunresolved alternative pathways in plant defenses. Thesubfunctionalization of individual gene family membersof the phenylpropanoid pathway was evident at the phy-logenetic and at the gene expression level. Using geneti-cal genomics, we confirmed that some candidate genes inthis pathway were indeed genetically associated withQTLs for white pine weevil resistance. Weevil resistancewas also associated with trans-regulatory hotspots. Atsuch genomic regions, eQTLs were generated by severalsubfamilies from multimember families like 4CL,CCoAOMT, OMT, CYP450, PPR, DIR, and PRXR. Tan-dem gene duplication and subsequent gene retention iscommon for biotic stress genes [81]. Phenylpropanoidmetabolism has been identified as a source of tandemlyrepeated orthologous groups [81]. Based on their tightgenetic linkage [83], bona fide CCoAOMT genes in coni-fers might be tandemly repeated in the genome. In thisstudy we found that all three genes present in the sprucegenome are functionally diverged. Thus, in multi-genefamilies of the phenylpropanoid/lignin pathway indivi-dual members can be recruited for plant developmentand pest resistance, respectively. Neo- and subfunctiona-lizations in such gene families can be importantresources for genetic variation in the perennial. Recently,it was shown that members of the small PAL gene familyin the herbaceous annual Arabidopsis can have both dis-tinct and overlapping functions in processes related togrowth, development, and responses to environmentalstresses [90]. Our results add to the existing literature onthe role of the phenylpropanoid pathway in the evolutionof conifer defense mechanism against insect pests. Ourfindings highlight that specific genes within the phenyl-propanoid pathway can be duplicated and diversified inlong-lived conifers in a process that is fundamentally dif-ferent from shorter lived angiosperm species.MethodsPicea Glauca (Moench) Voss × Picea Engelmanii Parry exEngelm. PedigreeAn outline of the experimental setup is given in Figure2. Experimental spruce populations were chosen from acontrolled-cross progeny trial established in 1995 atKalamalka Research Station in Vernon, BC, Canada [43].The respective males and females originated from indi-viduals previously ranked for weevil-resistance in open-pollinated progeny tests [91]. Out of twenty crosses i.e.,resistant-female-by-susceptible-male crosses, fourcrosses with markedly intermediate weevil resistance[43] and forming a partial diallel were studied in depth(cross 26 from ♀PG87*♂PG165, cross 27 from♀PG87*♂PG117, cross 29 from ♀PG21*♂PG165 andcross 32 from ♀PG21*♂PG117, respectively and Figure2), see also below.In-silico SNPs were identified from the TreenomixEST database (K. Ritland pers. comm.). From theseSNPs, a mapping population of 417 individuals involvinga factorial cross design of 3 × 2 crosses was screenedusing a 384-plex GoldenGate Genotyping BeadArrayIllumina platform ([92], Illumina Inc., San Diego, CA,USA) at the CMMT Genotyping and Gene ExpressionCore Facility, Centre for Molecular Medicine and Ther-apeutics, Vancouver, BC. Total genomic DNA had beenisolated from flushing bud/needle tissue of those indivi-duals according to the cetyltrimethylammonium bro-mide (CTAB) method established by [93]. Genotypeswere scored using the BeadStudio software.Of all putative SNP loci in the genotyping assay, 73.4-76.0% high quality SNPs were considered in the analysis;Individuals in the crosses that could not be confirmed asfull-sibs were removed from subsequent phenotyping(cross 26: 7%, cross 27: 10%, cross 29: 4%, cross 32: 1%of the trees alive in 2006). Using a joint likelihoodmethod [94] we determined recombination ratesbetween each pair of loci, and relative genetic distances(centiMorgans) using JoinMap 3.0 software [95] (originalgenotype data are provided in the Additional File 3). Wefollowed the approach by [96] for grouping markers andmarker ordering. We included two additional crosses (C.Liewlaksaneeyanawin, pers. comm.,) in the analysis tomaximize the number of mapped loci. The establishedframework genetic map is based on 252 SNPs and ispresented in Additional File 2.Tree Height, Weevil Attack, and Oviposition DataMeasures were taken for the initial tree height in 1995(year one), and heights in years three and five as well asleader length in year five preceding the artificial aug-mentation of the local weevil population in October ofthe same year (hgt1995, hgt1997, hgt1999, and ldr1999,respectively). Attack rates in 2000 and 2001 (atk2000,atk2001) were classified as successful ‘top kills’, ‘failure’to kill the leader and ‘no attack’ [43], in addition, for thesame years, egg counts along the leaders (egg2000,egg2001) were summarized into five discrete classes(successful egg laying was recognized as feedingPorth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608Page 20 of 26punctures that contain egg covering fecal plugs), R.Alfaro, pers. Communication. For details see AdditionalFile 4. In addition, the sum of weevil attacks and thesum of oviposition for 2000 and 2001 were used as traits(sum_atk, and sum_egg).Tissue Collection, RNA Preparation, and MicroarrayTree material within a replicate block was sampled in arandomized fashion among the plots (i.e., crosses, Figure2). Terminal leaders from trees in a block were collectedin the mornings of May 16, 17, and 18, 2006, respec-tively. At the location bark/phloem tissue was immedi-ately harvested from cut leaders as described previously([97,59]), flash frozen in liquid nitrogen, and stored at-80°C until processed. Total RNA from unattacked indi-viduals was isolated following the protocol of [98] andquantified using NanoDrop® ND-1000 Spectrophot-ometer; RNA integrity was evaluated using the Agilent2100 Bioanalyzer. The 21,840 spruce ESTs on the arrayinvolved elements from 12 different cDNA libraries,built from different tissues (bark, phloem, xylem), whichwere under different developmental stages, as well aswound/methyljasmonate treated (ca. 6,500 elements)and untreated (ca. 15,400). Complete details of cDNAmicroarray fabrication and quality control are describedelsewhere (S. Ralph and co-workers, Gene ExpressionOmnibus database GEO: GPL5423 and http://www.tree-nomix.ca/).Microarray Experimental Design, Gene ExpressionProfiling, and Pre-processing of Expression DataAfter we tested six genotyped crosses using the pre-viously collected phenotypic data (see above), we deter-mined that genotype differences between most and leastresistant progeny were highest in crosses 26, 27, 29, and32. A distant pair design for microarray analysis thatmaximized direct comparisons between different allelesat each locus was originally introduced by [41] and wasmodified here for outbred individuals. We estimated thegenetic distance for possible probe-pairs genome-wideby using all segregating SNP loci (122 on average). Sucha procedure maximized the number of distant pairs in agiven cross. A 25% improvement over random pairingwas achieved. We also balanced the two dyes across thethree replicate blocks (i.e., sampling on three differentdays), the different batches of microarray fabrication,and different experimenters (see below). Our designresulted in 94 hybridizations profiling 48 individuals incross 26, 36 in cross 27, and 50 in cross 29, as well as54 individuals in cross 32 (Additional File 5).Hybridizations were performed using the GenisphereArray350 kit (Genisphere, Hatfield, USA) followingmanufacturer’s instructions. Forty micrograms totalRNA was reverse transcribed using Superscript IIreverse transcriptase (Invitrogen) and oligo d(T18) pri-mers with a 5’ unique sequence overhang specific toeither the Cy3 or Cy5 labeling reactions. The RNAstrand of the resulting cDNA:RNA hybrid was hydro-lyzed in 0.075 M NaOH/0.0075 M EDTA at 65°C for 15min followed by neutralization in 0.175 M Tris-HCl (pH8.0).Following pooling of the appropriate Cy3 and Cy5cDNAs, samples were precipitated with linear acryla-mide and resuspended in 17.7 μL nuclease-free water. A27.3 μL hybridization mixture containing 22.5 μL 2 ×SDS buffer, 4 μL LNA d(T) blocker, 2 μg sheared sal-mon testes DNA (Invitrogen) and 0.3 μL of Cy5-labeledGFP cDNA (Cy5-dUTP and Ready-To-Go labelingbeads, Amersham Pharmacia Biotech) was added to thecDNA probe. Immediately prior to use, arrays were pre-washed 2× in 0.1% SDS at room temperature for 5 mineach, followed by two washes in MilliQ-H2O for 2 mineach, 3 min at 95°C in MilliQ-H2O, and dried by centri-fugation (3 min at 2000 rpm in an IEC Centra CL2 cen-trifuge with rotor IEC 2367-00 in 50 mL conical tube).The cDNA probe was heat denatured at 80°C for 10min, then maintained at 65°C prior to adding to amicroarray slide heated to 55°C, covered with a 22 × 60× 1.5 mm glass coverslip (Fisher Scientific), and incu-bated for 16 h at 60°C. Arrays were washed in 2× SSC,0.2% SDS at room temperature for 5 min to remove thecoverslip, followed by 15 min at 65°C in the same solu-tion, then three washes of 5 min in 2× SSC at roomtemperature, and three washes of 5 min in 0.2× SSC atroom temperature, and dried by centrifugation. The Cy3and Cy5 3DNA capture reagent (Genisphere) were thenhybridized to the bound cDNA on the microarray in a45 μL volume consisting of 22.5 μL 2× SDS buffer, 17.5μL nuclease-free water, 2.5 μL Cy3 capture reagent, and2.5 μL Cy5 capture reagent. The 3DNA capture reagentis bound to its complementary cDNA capture sequenceon the Cy3 or Cy5 oligo d(T) primers. The secondhybridization was performed for 3 h at 60°C and wasthen washed and dried as before.Fluorescent images of hybridized arrays were acquiredby using ScanArray Express (PerkinElmer, Foster City,USA). The Cy3 and Cy5 cyanine fluors were excited at543 nm and 633 nm, respectively. All scans were per-formed at the same laser power (90%), but with thephotomultiplier tube settings for the two channelsadjusted such that the ratio of the mean signal intensi-ties was ~1, and the percentage of saturated array ele-ments was < 0.5% but > 0%, while minimizingbackground fluorescence. Fluorescent intensity datawere extracted by using the ImaGene 6.0 software (Bio-discovery, El Segundo, USA). Signal intensity measure-ments were deposited in GEO under the accessionnumber GSE22116.Porth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608Page 21 of 26After quantification of the signal intensities in eacharray were completed, the local background was sub-tracted for each subgrid. Data were normalized to com-pensate for non-linearity of intensity distributions usingthe variance stabilizing normalization method [99].Finally, all 94 slides underwent simultaneous normaliza-tion to achieve a similar and array-independent overallexpression level and variance for every channel. The lin-ear modelhi = μ + dye + block + batch + person + εi, (1)with μ as the overall mean, was then fit to the normal-ized intensities of each gene i (hi) in the Cy3 and Cy5channels to account for technical effects within the experi-ment (gene-specific ‘dye’ effect, replicate ‘block’, microar-ray fabrication ‘batch’, experimenter ‘person’ are all fixedeffects). The residuals were used in the subsequent eQTLanalysis. The above statistics were carried out using the Rstatistical package http://www.r-project.org.QTL AnalysisQTLs were mapped in the diallel progeny by employinga likelihood function to assess the phenotype effect(gene expression, other traits) conditional on genotypicvariation. We used R http://www.r-project.org to displayQTL density maps. A QTL was significant at LOD≥3.84(defined as the log likelihood ratio for the statistical sup-port of the presence of a QTL within a 10 cM markerinterval) and had to be detected for at least one parentin the diallel, however we did not show whether thatspecific QTL was also detected in any other parent (Fig-ure 5). A goodness-of-fit test assuming a uniform distri-bution was performed to test whether the observedfrequencies of eQTLs along the linkage map differedsignificantly from the expected value. Following therejection of the null hypothesis (c2 = 3,551, df = 251, p-value < 2.2e-16), we declared “eQTL hotspots” if thenumber of eQTLs at a given locus was equal or abovethe maximum value (i.e., 38) for assessed eQTL clustersfrom a randomly generated data set using 4,221 eQTLs,252 markers, and running 1,000 replicates. Positionalcandidate genes were identified by collocation of at least40% of their eQTLs with phenotypic trait QTLs basedon the criteria for identifying significant QTLs (seeabove) and running 10,000 randomizations (p ≤ 0.05).Co-Expression Network AnalysisWe used the software package GeneNet 1.2.3 http://strimmerlab.org/software/genenet/ to assess gene-geneinteractions. First, the ‘family’ effect was removed fromthe data using the ANOVA modelhi = μ + dye + block + family + batch + person + εi. (2)The residuals were retained and utilized in the analy-sis. Gene co-expression based on partial correlationswas assessed using a graphical Gaussian model (GGM)and a shrinkage method implemented in the GeneNetsoftware package [100]. Here, neighbors have directdependencies, since indirect gene connections areremoved by this method. As a significance cut-off weused the 80% probability for presence of gene-pair inter-connection. For visual representation of the network weused Pajek 1.24 software [101].Phylogenetic AnalysesContigs for candidate genes (Figure 1) from white spruce(Picea glauca) were created in silico from EST sequencesretrieved through reciprocal BLAST searches as describedin [25-29]. An arbitrary cut-off of sequence identityexceeding 99% was used to identify possible allelic varia-tion. Phylogenetic analyses on aligned amino acidsequences (dialign2; http://bioweb2.pasteur.fr/; manuallycurated; Bioedit v7.0.9 http://www.mbio.ncsu.edu/BioEdit/bioedit.html) were tested for consistency and the recon-structed maximum likelihood tree was bootstrapped[PhyML; http://www.atgc-montpellier.fr/phyml/binaries.php; four rate substitution categories, g shape parameteroptimized, JTT (Jones-Taylor-Thornton) substitutionmodel, BioNJ starting tree and 100 bootstrap repetitions[102]] and displayed as phylogram using treeview32 1.6.6(http://taxonomy.zoology.gla.ac.uk/rod/treeview.html[103]). Tree topologies were supported using the indepen-dent maximum likelihood algorithm TREE-PUZZLE 5.2(http://www.tree-puzzle.de[104]). Array elements weremapped onto the phylogenetic trees by homology. Infor-mation about sequences from Arabidopsis and function-ally characterized orthologues that were retrieved frompublic databases is given in Additional File 14. In decreas-ing order of priority, we used: (i) model plants withsequenced genomes (Arabidopsis, poplar and rice), (ii)functionally characterized members from non-modelplants (gymnosperms, finally illustrative members fromother angiosperms). Fully sequenced genomes allow asses-sing possible species- or lineage-specific expansions whichare indicative of the mode of evolution (for example in therice and spruce COMT superfamily). Functionally charac-terized individual members demonstrate diversification(neofunctionalization) in subfamilies.Additional materialAdditional file 1: List of 428 spruce candidate genes of thephenylpropanoid biosynthesis and related pathway from ourmicroarray experiment, with AGI annotations, and trait associations.Additional information is provided regarding the code of each individualtranscript (microarray element) within the gene co-expression network(Figure 4) as well as about amino acid length and isoelectric point (pI)for the classification of the classIII peroxidases (prxr).Porth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608Page 22 of 26Additional file 2: Framework linkage map as displayed in Figure 5,252 SNPs, with annotations of contig sequences that were used forSNP detection. Information about SNP loci annotation (Viridiplantae andTAIR7) and their positions along the linkage groups (LG) in centiMorganmap distance.Additional file 3: Genotype data for individuals of the QTL mappingpopulation with resolved sibship (see Material and Methods). SNPsthat are part of the framework linkage map are displayed regarding theirassignment to linkage groups (LG) and their positions on LGs in mapdistances (centiMorgan, cM).Additional file 4: The phenotypic data for the four crosses 26, 27,29, and 32 used in the eQTL study contain information about (a)tree height, (b) weevil attack, and (c) oviposition. For explanations ofcolumn headers and codes, see end of table.Additional file 5: Design and hybridizations as indicated in Methodsand from which signal intensity measurements were deposited inGEO. Information is provided for each hybridization (slide number asdeposited in GEO) about the origin of the slide (batch identifiers A, B, C),the experimenter, i.e., the person identifier (1, 2, 3) for probehybridization, the dye label of each RNA sample (Cy3 or Cy5) and thesample identifier (26A04, e.g., indicates an individual from cross26,replicate block A, plot tree 4).Additional file 6: QTLs for (a) tree height, (b) weevil attack, and (c)oviposition. A QTL was significant at LOD≥3.84; allele effect and %phenotypic variation explained by QTL is given. We summarizedresults for an expanded mapping population (five parents, a progeny of369 individuals) as will be described elsewhere (Porth et al., in prep.) andfor the following traits: tree heights (initial height in 1995, height in 1997,height in 1999), leader length in 1999, weevil attacks (assessed in 2000,in 2001 and the total attacks summed up for both years) and oviposition(egg plugs in 2000, in 2001, and total egg plugs summed up for bothyears).Additional file 7: Significant QTLs for gene expression; allele effectand % phenotypic variation explained by the QTL is given.Information about the detected eQTLs which were significant atLOD≥3.84 is provided including the information if a QTL was detected inmore than one parent (these details are not shown in the density mapof Figure 5), see Methods for further explanation.Additional file 8: Phenylpropanoid P450 (F5H-F3H), phylogenetictree, array elements (in red).Additional file 9: Eight phenotype associated expression hotspotswith listings of gene family members that contributed eQTLs (geneannotations where applicable). See also the information given inAdditional File 1 and Figure 5.Additional file 10: Summary on gene families with individualmembers having eQTLs at a minimum of two phenotype associatedexpression hotspots. This table summarizes characteristics of theinvestigated gene families associated with the phenylpropanoid pathwayand shows the total number of gene-family members, the detectedaverage number of eQTLs per family, and the number of the identifiedgene-family members with eQTLs at a minimum of two out of the eightidentified regulatory hotspots as shown in Figure 6.Additional file 11: The O-methyltransferase superfamily familyincluding outgroup PicglOMT1, array element (in red).Additional file 12: Catechol-OMTs including ancient white spruceCatechol-O-Methyltransferase (PicglOMT2), array element (in red).Additional file 13: Class III Peroxidases, phylogenetic tree, arrayelements (in red) and non-treenomix, phyml 10× tree, rooted withSpruce67/86.Additional file 14: Accessions and identifiers of non-sprucemembers of the phenylpropanoid pathway families. Accessions andidentifiers are provided for the following gene families: OMT/OMTL,CCoAOMT, PCBER, PKS, 4CL/ACL, and ADT.List of abbreviationsPicgl: Picea glauca; Picsi: Picea sitchensis; Pge: Picea engelmanii; 4CL: 4-coumarate-CoA ligase; ADT/PD: arogenate dehydratase/prephenatedehydratase; C4H: cinnamate-4-hydroxylase; CAD: cinnamyl-alcoholdehydrogenase; CCoAOMT: caffeoyl-CoA 3-O-methyltransferase; CCR:cinnamoyl-CoA reductase; CHI: chalcone (flavanone) isomerase; CHS:chalcone synthase; CM: chorismate mutase; CS: chorismate synthase; DAHP:3-deoxy-D-arabino-heptulosonate 7-phosphate synthase; DFR:dihydroflavonol 4-reductase; DHQD-SD: 3-dehydroquinate dehydratase/shikimate 5-dehydrogenase; DHQS: 3-dehydroquinate synthase; DIR: diseaseresistance-responsive/dirigent; EPSPS: 5-enolpyruvylshikimate-3-phosphate/EPSP synthase; F3’H: flavonoid 3’-monooxygenase; F5H: ferulate 5-hydroxylase; FLS: flavonol synthase; LAC: laccase; OMT: O-methyltransferase;OPCL: oxo-pentenyl-cyclopentane ligase, PAL: phenylalanine ammonia-lyase;PCBER: phenylcoumaran benzylic ether reductase; PLR: pinoresinol-lariciresinol reductase; PRXR: peroxidase; SK: shikimate kinase; SAMS: S-adenosyl-methionine synthase; MYB: myeloblastosis domain; NAC: nascentpolypeptide-associated complex; QTL: quantitative trait locus; eQTL: geneexpression QTL; pQTL: phenotypic trait QTL; SNP: single nucleotidepolymorphism; GGM: graphical Gaussian model; LOD: logarithm of the oddsgives the likelihood ratio between two hypotheses.Acknowledgements and fundingWe thank Barry Jaquish (BC Ministry of Forests) as well as Rene Alfaro(Canadian Forest Service) for providing us with original data and previouslyunpublished data, and Carol Ritland for support throughout the study,Charles Chen for help with R script as well as Joerg Bohlmann for providingresources. We acknowledge Gillian Leung, and Michelle Tang for technicalsupport, Susan Findlay, Tristan Gillan, Jun Zhang, CherdsakLiewlaksanyannawin, and Claire Cullis for help with sample collection. Weacknowledge two anonymous reviewers for very helpful comments. Wethank Jessica P. Flores for proofreading. This work was carried out withfinancial support from Genome British Columbia and Genome Canada.Author details1Department of Forest Sciences, University of British Columbia, 2424 MainMall, Vancouver, BC V6T1Z4, Canada. 2Michael Smith Laboratories, Universityof British Columbia, 2185 East Mall, Vancouver, BC V6T1Z4, Canada.3Department of Statistics, University of British Columbia, 6356 AgriculturalRoad, Vancouver, BC V6T1Z2, Canada.Authors’ contributionsIP, RW, and KR designed experiments, conducted the data analysis andinterpretation of data and results. IP carried out experiments. BH contributedto phylogenetic analysis. KR conceived of the overall study. IP and BH wrotethe manuscript. All authors read and approved the final manuscript.Received: 22 July 2011 Accepted: 16 December 2011Published: 16 December 2011References1. Boerjan W, Ralph J, Baucher M: Lignin biosynthesis. Annual Review of PlantBiology 2003, 54:519-546.2. Campbell MM, Sederoff RR: Variation in lignin content and composition -Mechanism of control and implications for the genetic improvement ofplants. Plant Physiology 1996, 110(1):3-13.3. Schlesinger WH, Lichter J: Limited carbon storage in soil and litter ofexperimental forest plots under increased atmospheric CO2. Nature 2001,411(6836):466-469.4. Hagerman AE, Riedl KM, Jones GA, Sovik KN, Ritchard NT, Hartzfeld PW,Riechel TL: High Molecular Weight Plant Polyphenolics (Tannins) asBiological Antioxidants. Journal of Agricultural and Food Chemistry 1998,46(5):1887-1892.5. Grace SC, Logan BA: Energy dissipation and radical scavenging by theplant phenylpropanoid pathway. Philosophical Transactions of the RoyalSociety of London Series B-Biological Sciences 2000, 355(1402):1499-1510.6. Dixon RA, Paiva NL: Stress-Induced Phenylpropanoid Metabolism. PlantCell 1995, 7(7):1085-1097.Porth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608Page 23 of 267. Bharti AK, Khurana JP: Mutants of Arabidopsis as tools to understand theregulation of phenylpropanoid pathway and UVB protectionmechanisms. Photochem Photobiol 1997, 65(5):765-776.8. Bieza K, Lois R: An Arabidopsis mutant tolerant to lethal ultraviolet-Blevels shows constitutively elevated accumulation of flavonoids andother phenolics. Plant Physiol 2001, 126(3):1105-1115.9. Bashandy T, Taconnat L, Renou J-P, Meyer Y, Reichheld J-P: Accumulationof Flavonoids in an ntra ntrb Mutant Leads to Tolerance to UV-C. MolPlant 2009, 2(2):249-258.10. Ferrer JL, Austin MB, Stewart C, Noe JP: Structure and function ofenzymes involved in the biosynthesis of phenylpropanoids. PlantPhysiology and Biochemistry 2008, 46(3):356-370.11. Haviola S, Kapari L, Ossipov V, Rantala MJ, Ruuhola T, Haukioja E: Foliarphenolics are differently associated with Epirrita autumnata growth andimmunocompetence. J Chem Ecol 2007, 33(5):1013-1023.12. Simmonds MS: Flavonoid-insect interactions: recent advances in ourknowledge. Phytochemistry 2003, 64(1):21-30.13. Vogt T: Phenylpropanoid Biosynthesis. Molecular Plant 2010, 3(1):2-20.14. Hamberger B, Ellis M, Friedmann M, Souza CDA, Barbazuk B, Douglas CJ:Genome-wide analyses of phenylpropanoid-related genes in Populustrichocarpa, Arabidopsis thaliana, and Oryza sativa: the Populus lignintoolbox and conservation and diversification of angiosperm genefamilies. Canadian Journal of Botany-Revue Canadienne De Botanique 2007,85(12):1182-1201.15. Koutaniemi S, Toikka MM, Karkonen A, Mustonen M, Lundell T, Simola LK,Kilpelainen IA, Teeri TH: Characterization of basic p-coumaryl andconiferyl alcohol oxidizing peroxidases from a lignin-forming Picea abiessuspension culture. Plant Mol Biol 2005, 58(2):141-157.16. Li LG, Popko JL, Zhang XH, Osakabe K, Tsai CJ, Joshi CP, Chiang VL: A novelmultifunctional O-methyltransferase implicated in a dual methylationpathway associated with lignin biosynthesis in loblolly pine. Proceedingsof the National Academy of Sciences of the United States of America 1997,94(10):5461-5466.17. Chiron H, Drouet A, Claudot AC, Eckerskorn C, Trost M, Heller W, Ernst D,Sandermann H Jr: Molecular cloning and functional expression of astress-induced multifunctional O-methyltransferase with pinosylvinmethyltransferase activity from Scots pine (Pinus sylvestris L.). Plant MolBiol 2000, 44(6):733-745.18. Moyle R, Moody J, Phillips L, Walter C, Wagner A: Isolation andcharacterization of a Pinus radiata lignin biosynthesis-related O-methyltransferase promoter. Plant Cell Reports 2002, 20(11):1052-1060.19. Wagner A, Ralph J, Akiyama T, Flint H, Phillips L, Torr K, Nanayakkara B,Kiri LT: Exploring lignification in conifers by silencing hydroxycinnamoyl-CoA: shikimate hydroxycinnamoyltransferase in Pinus radiata. Proceedingsof the National Academy of Sciences of the United States of America 2007,104(28):11856-11861.20. Bomal C, Bedon F, Caron S, Mansfield SD, Levasseur C, Cooke JEK, Blais S,Tremblay L, Morency MJ, Pavy N, et al: Involvement of Pinus taeda MYB1and MYB8 in phenylpropanoid metabolism and secondary cell wallbiogenesis: a comparative in planta analysis. Journal of ExperimentalBotany 2008, 59(14):3925-3939.21. Franceschi VR, Krokene P, Christiansen E, Krekling T: Anatomical andchemical defenses of conifer bark against bark beetles and other pests.New Phytologist 2005, 167(2):353-375.22. Friedmann M, Ralph SG, Aeschliman D, Zhuang J, Ritland K, Ellis BE,Bohlmann J, Douglas CJ: Microarray gene expression profiling ofdevelopmental transitions in Sitka spruce (Picea sitchensis) apicalshoots. Journal of Experimental Botany 2007, 58(3):593-614.23. Ahuja MR, Neale DB: Evolution of genome size in conifers. Silvae Genetica2005, 54(3):126-137.24. Kinlaw CS, Neale DB: Complex gene families in pine genomes. Trends inPlant Science 1997, 2(9):356-359.25. Hamberger B, Bohlmann J: Cytochrome P450 mono-oxygenases in conifergenomes: discovery of members of the terpenoid oxygenasesuperfamily in spruce and pine. Biochemical Society transactions 2006,34(Pt 6):1209-1214.26. Liu JJ, Ekramoddoullah AK: Isolation, genetic variation and expression ofTIR-NBS-LRR resistance gene analogs from western white pine (Pinusmonticola Dougl. ex. D. Don.). Mol Genet Genomics 2003, 270(5):432-441.27. Liu JJ, Ekramoddoullah AK: Characterization, expression and evolution oftwo novel subfamilies of Pinus monticola cDNAs encoding pathogenesis-related (PR)-10 proteins. Tree physiology 2004, 24(12):1377-1385.28. Martin DM, Fäldt J, Bohlmann J: Functional Characterization of NineNorway Spruce TPS Genes and Evolution of Gymnosperm TerpeneSynthases of the TPS-d Subfamily. Plant Physiol 2004, 135(4):1908-1927.29. Ralph S, Park JY, Bohlmann J, Mansfield SD: Dirigent proteins in coniferdefense: gene discovery, phylogeny, and differential wound- and insect-induced expression of a family of DIR and DIR-like genes in spruce(Picea spp.). Plant molecular biology 2006, 60(1):21-40.30. Passardi F, Longet D, Penel C, Dunand C: The class III peroxidasemultigenic in land plants family in rice and its evolution. Phytochemistry2004, 65(13):1879-1893.31. Flagel LE, Wendel JF: Gene duplication and evolutionary novelty inplants. New Phytologist 2009, 183(3):557-564.32. Conant GC, Wolfe KH: Turning a hobby into a job: How duplicated genesfind new functions. Nature Reviews Genetics 2008, 9(12):938-950.33. Keeling CI, Weisshaar S, Lin RPC, Bohlmann J: Functional plasticity ofparalogous diterpene synthases involved in conifer defense. Proceedingsof the National Academy of Sciences of the United States of America 2008,105(3):1085-1090.34. Hahn MW: Distinguishing Among Evolutionary Models for theMaintenance of Gene Duplicates. Journal of Heredity 2009, 100(5):605-617.35. Kuzniar A, van Ham R, Pongor S, Leunissen JAM: The quest for orthologs:finding the corresponding gene across genomes. Trends in Genetics 2008,24(11):539-551.36. Zhong RQ, Lee CH, Ye ZH: Functional Characterization of Poplar Wood-Associated NAC Domain Transcription Factors. Plant Physiology 2010,152(2):1044-1055.37. Zhong RQ, Ye ZH: Regulation of cell wall biosynthesis. Current Opinion inPlant Biology 2007, 10(6):564-572.38. Kirst M, Myburg AA, De Leon JPG, Kirst ME, Scott J, Sederoff R: Coordinatedgenetic regulation of growth and lignin revealed by quantitative traitlocus analysis of cDNA microarray data in an interspecific backcross ofeucalyptus. Plant Physiology 2004, 135(4):2368-2378.39. Farrall M: Quantitative genetic variation: a post-modern view. HumanMolecular Genetics 2004, 13:R1-R7.40. Jansen RC, Nap JP: Genetical genomics: the added value fromsegregation. Trends in Genetics 2001, 17(7):388-391.41. Fu JY, Jansen RC: Optimal design and analysis of genetic studies on geneexpression. Genetics 2006, 172(3):1993-1999.42. Rockman MV, Kruglyak L: Genetics of global gene expression. NatureReviews Genetics 2006, 7(11):862-872.43. Alfaro RI, VanAkker L, Jaquish B, King J: Weevil resistance of progenyderived from putatively resistant and susceptible interior spruce parents.Forest Ecology and Management 2004, 202(1-3):369-377.44. Baum G, LevYadun S, Fridmann Y, Arazi T, Katsnelson H, Zik M, Fromm H:Calmodulin binding to glutamate decarboxylase is required forregulation of glutamate and GABA metabolism and normaldevelopment in plants. Embo Journal 1996, 15(12):2988-2996.45. Bown AW, MacGregor KB, Shelp BJ: Gamma-aminobutyrate: defenseagainst invertebrate pests? Trends in Plant Science 2006, 11(9):424-427.46. Kliebenstein D: Quantitative Genomics: Analyzing Intraspecific VariationUsing Global Gene Expression Polymorphisms or eQTLs. Annual Review ofPlant Biology 2009, 60:93-114.47. Jung E, Zamir LO, Jensen RA: Chloroplasts of higher plants synthesize L-phenylalanine via L-arogenate. Proc Natl Acad Sci USA 1986,83(19):7231-7235.48. Maeda H, Shasany AK, Schnepp J, Orlova I, Taguchi G, Cooper BR,Rhodes D, Pichersky E, Dudareva N: RNAi Suppression of ArogenateDehydratase1 Reveals That Phenylalanine Is Synthesized Predominantlyvia the Arogenate Pathway in Petunia Petals. Plant Cell 2010,22(3):832-849.49. de Azevedo Souza C, Barbazuk B, Ralph SG, Bohlmann J, Hamberger B,Douglas CJ: Genome-wide analysis of a land plant-specific acyl:coenzymeA synthetase (ACS) gene family in Arabidopsis, poplar, rice andPhyscomitrella. New Phytol 2008, 179(4):987-1003.50. Kienow L, Schneider K, Bartsch M, Stuible HP, Weng H, Miersch O,Wasternack C, Kombrink E: Jasmonates meet fatty acids: functionalPorth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608Page 24 of 26analysis of a new acyl-coenzyme A synthetase family from Arabidopsisthaliana. J Exp Bot 2008, 59(2):403-419.51. Ehlting J, Buttner D, Wang Q, Douglas CJ, Somssich IE, Kombrink E: Three4-coumarate:coenzyme A ligases in Arabidopsis thaliana represent twoevolutionarily divergent classes in angiosperms. Plant J 1999, 19(1):9-20.52. Zhang XH, Chiang VL: Molecular cloning of 4-coumarate:coenzyme Aligase in loblolly pine and the roles of this enzyme in the biosynthesisof lignin in compression wood. Plant Physiol 1997, 113(1):65-74.53. Berthet S, Demont-Caulet N, Pollet B, Bidzinski P, Cézard L, Le Bris P,Borrega N, Hervé J, Blondet E, Balzergue S, et al: Disruption of LACCASE4and 17 Results in Tissue-Specific Alterations to Lignification ofArabidopsis thaliana Stems. The Plant Cell Online 2011, 23(3):1124-1137.54. Li L, Osakabe Y, Joshi CP, Chiang VL: Secondary xylem-specific expressionof caffeoyl-coenzyme A 3-O-methyltransferase plays an important role inthe methylation pathway associated with lignin biosynthesis in loblollypine. Plant Mol Biol 1999, 40(4):555-565.55. Jiang C, Schommer CK, Kim SY, Suh DY: Cloning and characterization ofchalcone synthase from the moss, Physcomitrella patens. Phytochemistry2006, 67(23):2531-2540.56. Kreuzaler F, Hahlbrock K: Enzymic synthesis of an aromatic ring fromacetate units. Partial purification and some properties of flavanonesynthase from cell-suspension cultures of Petroselinum hortense. Eur JBiochem 1975, 56(1):205-213.57. Schroder G, Brown JW, Schroder J: Molecular analysis of resveratrolsynthase. cDNA, genomic clones and relationship with chalconesynthase. Eur J Biochem 1988, 172(1):161-169.58. Lindberg M, Lundgren L, Gref R, Johansson M: Stilbenes and resin acids inrelation to the penetration of Heterobasidion annosum through the barkof Picea abies. European Journal of Forest Pathology 1992, 22(2):95-106.59. Ralph SG, Yueh H, Friedmann M, Aeschliman D, Zeznik JA, Nelson CC,Butterfield YSN, Kirkpatrick R, Liu J, Jones SJM, et al: Conifer defenceagainst insects: microarray gene expression profiling of Sitka spruce(Picea sitchensis) induced by mechanical wounding or feeding byspruce budworms (Choristoneura occidentalis) or white pine weevils(Pissodes strobi) reveals large-scale changes of the host transcriptome.Plant Cell and Environment 2006, 29(8):1545-1570.60. Ehlting J, Hamberger B, Million-Rousseau R, Werck-Reichhart D:Cytochromes P450 in phenolic metabolism. Phytochemistry Reviews 2006,5(2):239-270.61. Schoenbohm C, Martens S, Eder C, Forkmann G, Weisshaar B: Identificationof the Arabidopsis thaliana flavonoid 3’-hydroxylase gene and functionalexpression of the encoded P450 enzyme. Biol Chem 2000, 381(8):749-753.62. Nesi N, Jond C, Debeaujon I, Caboche M, Lepiniec L: The Arabidopsis TT2gene encodes an R2R3 MYB domain protein that acts as a keydeterminant for proanthocyanidin accumulation in developing seed.Plant Cell 2001, 13(9):2099-2114.63. Akashi T, Koshimizu S, Aoki T, Ayabe S: Identification of cDNAs encodingpterocarpan reductase involved in isoflavan phytoalexin biosynthesis inLotus japonicus by EST mining. FEBS Lett 2006, 580(24):5666-5670.64. Bogs J, Downey MO, Harvey JS, Ashton AR, Tanner GJ, Robinson SP:Proanthocyanidin synthesis and expression of genes encodingleucoanthocyanidin reductase and anthocyanidin reductase indeveloping grape berries and grapevine leaves. Plant Physiol 2005,139(2):652-663.65. Fujita M, Gang DR, Davin LB, Lewis NG: Recombinant pinoresinol-lariciresinol reductases from western red cedar (Thuja plicata) catalyzeopposite enantiospecific conversions. J Biol Chem 1999, 274(2):618-627.66. Gang DR, Kasahara H, Xia ZQ, Vander Mijnsbrugge K, Bauw G, Boerjan W,Van Montagu M, Davin LB, Lewis NG: Evolution of plant defensemechanisms. Relationships of phenylcoumaran benzylic etherreductases to pinoresinol-lariciresinol and isoflavone reductases. J BiolChem 1999, 274(11):7516-7527.67. Koeduka T, Fridman E, Gang DR, Vassao DG, Jackson BL, Kish CM, Orlova I,Spassova SM, Lewis NG, Noel JP, et al: Eugenol and isoeugenol,characteristic aromatic constituents of spices, are biosynthesized viareduction of a coniferyl alcohol ester. Proc Natl Acad Sci USA 2006,103(26):10128-10133.68. Cosio C, Dunand C: Specific functions of individual class III peroxidasegenes. Journal of Experimental Botany 2009, 60(2):391-408.69. Tokunaga N, Kaneta T, Sato S, Sato Y: Analysis of expression profiles ofthree peroxidase genes associated with lignification in Arabidopsisthaliana. Physiol Plant 2009, 136(2):237-249.70. Li YH, Kajita S, Kawai S, Katayama Y, Morohoshi N: Down-regulation of ananionic peroxidase in transgenic aspen and its effect on lignincharacteristics. Journal of Plant Research 2003, 116(3):175-182.71. Marjamaa K, Kukkola EM, Fagerstedt KV: The role of xylem class IIIperoxidases in lignification. Journal of Experimental Botany 2009,60(2):367-376.72. Fagerstedt KV, Kukkola EM, Koistinen VVT, Takahashi J, Marjamaa K: Cell WallLignin is Polymerised by Class III Secretable Plant Peroxidases in NorwaySpruce. Journal of Integrative Plant Biology 2010, 52(2):186-194.73. Koutaniemi S, Warinowski T, Karkonen A, Alatalo E, Fossdal CG, Saranpaa P,Laakso T, Fagerstedt KV, Simola LK, Paulin L, et al: Expression profiling ofthe lignin biosynthetic pathway in Norway spruce using EST sequencingand real-time RT-PCR. Plant Molecular Biology 2007, 65(3):311-328.74. Dixon RA, Achnine L, Kota P, Liu CJ, Reddy MSS, Wang LJ: Thephenylpropanoid pathway and plant defence - a genomics perspective.Molecular Plant Pathology 2002, 3(5):371-390.75. Ober D: Seeing double: gene duplication and diversification in plantsecondary metabolism. Trends in Plant Science 2005, 10(9):444-449.76. Wagner A, Tobimatsu Y, Phillips L, Flint H, Torr K, Donaldson L, Pears L,Ralph J: CCoAOMT suppression modifies lignin composition in Pinusradiata. The Plant Journal 2011, 67(1):11.77. Raes J, Rohde A, Christensen JH, Van de Peer Y, Boerjan W: Genome-widecharacterization of the lignification toolbox in Arabidopsis. PlantPhysiology 2003, 133(3):1051-1071.78. Martone PT, Estevez JM, Lu FC, Ruel K, Denny MW, Somerville C, Ralph J:Discovery of Lignin in Seaweed Reveals Convergent Evolution of Cell-Wall Architecture. Current Biology 2009, 19(2):169-175.79. Potokina E, Druka A, Luo ZW, Wise R, Waugh R, Kearsey M: Geneexpression quantitative trait locus analysis of 16,000 barley genesreveals a complex pattern of genome-wide transcriptional regulation.Plant Journal 2008, 53(1):90-101.80. Todesco M, Balasubramanian S, Hu TT, Traw MB, Horton M, Epple P,Kuhns C, Sureshkumar S, Schwartz C, Lanz C, et al: Natural allelic variationunderlying a major fitness trade-off in Arabidopsis thaliana. Nature 2010,465(7298):632-U129.81. Hanada K, Zou C, Lehti-Shiu MD, Shinozaki K, Shiu SH: Importance oflineage-specific expansion of plant tandem duplicates in the adaptiveresponse to environmental stimuli. Plant Physiology 2008, 148(2):993-1003.82. Sewell MM, Sherman BK, Neale DB: A consensus map for loblolly pine(Pinus taeda L.). I. Construction and integration of individual linkagemaps from two outbred three-generation pedigrees. Genetics 1999,151(1):321-330.83. Eckert AJ, Pande B, Ersoz ES, Wright MH, Rashbrook VK, Nicolet CM,Neale DB: High-throughput genotyping and mapping of singlenucleotide polymorphisms in loblolly pine (Pinus taeda L.). Tree Genetics& Genomes 2009, 5(1):225-234.84. Wallis C, Eyles A, Chorbadjian R, Gardener BM, Hansen R, Cipollini D,Herms DA, Bonello P: Systemic induction of phloem secondarymetabolism and its relationship to resistance to a canker pathogen inAustrian pine. New Phytologist 2008, 177(3):767-778.85. Lange BM, Lapierre C, Sandermann H: Elicitor-induced spruce stress lignin- structural similarity to early developmental lignins. Plant Physiology1995, 108(3):1277-1287.86. Kim YJ, Kim DG, Lee SH, Lee I: Wound-induced expression of the ferulate5-hydroxylase gene in Camptotheca acuminata. Biochimica Et BiophysicaActa-General Subjects 2006, 1760(2):182-190.87. Nakano T, Suzuki K, Fujimura T, Shinshi H: Genome-wide analysis of theERF gene family in Arabidopsis and rice. Plant Physiology 2006,140(2):411-432.88. Yang Z, Tian LN, Latoszek-Green M, Brown D, Wu KQ: Arabidopsis ERF4 isa transcriptional repressor capable of modulating ethylene and abscisicacid responses. Plant Molecular Biology 2005, 58(4):585-596.89. Grattapaglia D, Plomion C, Kirst M, Sederoff RR: Genomics of growth traitsin forest trees. Current Opinion in Plant Biology 2009, 12(2):148-156.90. Huang J, Gu M, Lai Z, Fan B, Shi K, Zhou Y-H, Yu J-Q, Chen Z: FunctionalAnalysis of the Arabidopsis PAL Gene Family in Plant Growth,Porth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608Page 25 of 26Development and Response to Environmental Stress. Plant Physiol 2010,pp.110.157370.91. Kiss GK, Yanchuk AD: Preliminary evaluation of genetic-variation of weevilresistance in interior spruce in british-columbia. Canadian Journal ofForest Research-Revue Canadienne De Recherche Forestiere 1991,21(2):230-234.92. Fan JB, Chee MS, Gunderson KL: Highly parallel genomic assays. NatureReviews Genetics 2006, 7(8):632-644.93. Doyle JJ, Doyle JL: Isolation of plant DNA from fresh tissue. Focus 1990,12:13-15.94. Hu XS, Goodwillie C, Ritland KM: Joining genetic linkage maps using ajoint likelihood function. Theoretical and Applied Genetics 2004,109(5):996-1004.95. Stam P: Construction of integrated genetic-linkage maps by means of anew computer package - joinmap. Plant Journal 1993, 3(5):739-744.96. Jermstad KD, Bassoni DL, Jech KS, Ritchie GA, Wheeler NC, Neale DB:Mapping of quantitative trait loci controlling adaptive traits in coastalDouglas fir. III. Quantitative trait loci-by-environment interactions.Genetics 2003, 165(3):1489-1506.97. Ralph S, Park JY, Bohlmann J, Mansfield SD: Dirigent proteins in coniferdefense: gene discovery, phylogeny, and differential wound- and insect-induced expression of a family of DIR and DIR-like genes in spruce(Picea spp.). Plant Molecular Biology 2006, 60(1):21-40.98. Kolosova N, Miller B, Ralph S, Ellis BE, Douglas C, Ritland K, Bohlmann J:Isolation of high-quality RNA from gymnosperm and angiosperm trees.Biotechniques 2004, 36(5):821-824.99. Huber W, von Heydebreck A, Sultmann H, Poustka A, Vingron M: Variancestabilization applied to microarray data calibration and to thequantification of differential expression. Bioinformatics 2002, 18:S96-S104.100. Schaefer J, Strimmer K: A shrinkage approach to large-scale covariancematrix estimation and implications for functional genomics. StatisticalApplications in Genetics and Molecular Biology 2005, 4:ISSN 1544-6115(print)|1544-6115(electronic).101. De Nooy W, Mrvar A, Batagelj V: Exploratory Social Network Analysis withPajek (Structural Analysis in the Social Sciences). Cambridge UniversityPress; 2005.102. Guindon S, Gascuel O: A simple, fast, and accurate algorithm to estimatelarge phylogenies by maximum likelihood. Syst Biol 2003, 52(5):696-704.103. Page RD: Visualizing phylogenetic trees using TreeView. Curr ProtocBioinformatics 2002, Chapter 6(Unit 6 2).104. Schmidt HA, von Haeseler A: Maximum-likelihood analysis using TREE-PUZZLE. Curr Protoc Bioinformatics 2007, Chapter 6(Unit 6 6).105. Chong J, Poutaraud A, Hugueney P: Metabolism and roles of stilbenes inplants. Plant Science (Oxford) 2009, 177(3):143-155.106. Turnbull JJ, Sobey WJ, Aplin RT, Hassan A, Firmin JL, Schofield CJ,Prescott AG: Are anthocyanidins the immediate products ofanthocyanidin synthase? Chemical Communications 2000, , 24: 2473-2474.doi:10.1186/1471-2164-12-608Cite this article as: Porth et al.: Defense mechanisms against herbivoryin Picea: sequence evolution and expression regulation of gene familymembers in the phenylpropanoid pathway. BMC Genomics 2011 12:608.Submit your next manuscript to BioMed Centraland take full advantage of: • Convenient online submission• Thorough peer review• No space constraints or color figure charges• Immediate publication on acceptance• Inclusion in PubMed, CAS, Scopus and Google Scholar• Research which is freely available for redistributionSubmit your manuscript at www.biomedcentral.com/submitPorth et al. BMC Genomics 2011, 12:608http://www.biomedcentral.com/1471-2164/12/608Page 26 of 26


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items