UBC Faculty Research and Publications

High-resolution genetic mapping of allelic variants associated with cell wall chemistry in Populus Muchero, Wellington; Guo, Jianjun; DiFazio, Stephen P; Chen, Jin-Gui; Ranjan, Priya; Slavov, Gancho T; Gunter, Lee E; Jawdy, Sara; Bryan, Anthony C; Sykes, Robert; Ziebell, Angela; Klápště, Jaroslav; Porth, Ilga; Skyba, Oleksandr; Unda, Faride; El-Kassaby, Yousry A; Douglas, Carl J; Mansfield, Shawn D; Martin, Joel; Schackwitz, Wendy; Evans, Luke M; Czarnecki, Olaf; Tuskan, Gerald A Jan 23, 2015

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

Item Metadata


52383-12864_2015_Article_1215.pdf [ 1.46MB ]
JSON: 52383-1.0221345.json
JSON-LD: 52383-1.0221345-ld.json
RDF/XML (Pretty): 52383-1.0221345-rdf.xml
RDF/JSON: 52383-1.0221345-rdf.json
Turtle: 52383-1.0221345-turtle.txt
N-Triples: 52383-1.0221345-rdf-ntriples.txt
Original Record: 52383-1.0221345-source.json
Full Text

Full Text

RESEARCH ARTICLE Open AccessHigh-resolution genetic mapping of allelicell wall chemistry inKANADI transcription factor as well as a protein kinase. Results from protoplast transient expression assays suggestedMuchero et al. BMC Genomics  (2015) 16:24 DOI 10.1186/s12864-015-1215-z37831, USAFull list of author information is available at the end of the articlethat each of the polymorphisms conferred allelic differences in the activation of cellulose, hemicelluloses, and ligninpathway marker genes.Conclusion: This study illustrates the utility of complementary QTL and association mapping as tools for genediscovery with no a priori candidate gene selection. This proof of concept in a perennial organism opens upopportunities for discovery of novel genetic determinants of economically important but complex traits in plants.Keywords: QTL cloning, Association genetics, Cell wall recalcitrance, Lignin, Cellulose, Hemicellulose* Correspondence: mucherow@ornl.gov1BioEnergy Science Center, Oak Ridge National Laboratory, Oak Ridge, TNtranscription factor annotated as an ANGUSTIFOLIA C-termPopulusWellington Muchero1*, Jianjun Guo1,9, Stephen P DiFazio2, Jin-Gui Chen1, Priya Ranjan1, Gancho T Slavov3,Lee E Gunter1, Sara Jawdy1, Anthony C Bryan1, Robert Sykes4, Angela Ziebell4, Jaroslav Klápště5,10, Ilga Porth5,Oleksandr Skyba6, Faride Unda6, Yousry A El-Kassaby5, Carl J Douglas7, Shawn D Mansfield6, Joel Martin8,Wendy Schackwitz8, Luke M Evans2, Olaf Czarnecki1 and Gerald A Tuskan1AbstractBackground: QTL cloning for the discovery of genes underlying polygenic traits has historically been cumbersome inlong-lived perennial plants like Populus. Linkage disequilibrium-based association mapping has been proposed as acloning tool, and recent advances in high-throughput genotyping and whole-genome resequencing enable markersaturation to levels sufficient for association mapping with no a priori candidate gene selection. Here, multiyear andmultienvironment evaluation of cell wall phenotypes was conducted in an interspecific P. trichocarpa x P. deltoidespseudo-backcross mapping pedigree and two partially overlapping populations of unrelated P. trichocarpa genotypesusing pyrolysis molecular beam mass spectrometry, saccharification, and/ or traditional wet chemistry. QTL mappingwas conducted using a high-density genetic map with 3,568 SNP markers. As a fine-mapping approach, chromosome-wide association mapping targeting a QTL hot-spot on linkage group XIV was performed in the two P. trichocarpapopulations. Both populations were genotyped using the 34 K Populus Infinium SNP array and whole-genomeresequencing of one of the populations facilitated marker-saturation of candidate intervals for gene identification.Results: Five QTLs ranging in size from 0.6 to 1.8 Mb were mapped on linkage group XIV for lignin content, syringyl toguaiacyl (S/G) ratio, 5- and 6-carbon sugars using the mapping pedigree. Six candidate loci exhibiting significantassociations with phenotypes were identified within QTL intervals. These associations were reproducible across multipleenvironments, two independent genotyping platforms, and different plant growth stages. cDNA sequencing for allelicvariants of three of the six loci identified polymorphisms leading to variable length poly glutamine (PolyQ) stretch in ainus Binding Protein (CtBP) and premature stop codons in avariants associated with c© 2015 Muchero et al.; licensee Biomed Central. This is an Open Access article distributed under the terms of the CreativeCommons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, andreproduction in any medium, provided the original work is properly credited. The Creative Commons Public DomainDedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article,unless otherwise stated.Muchero et al. BMC Genomics  (2015) 16:24 Page 2 of 14BackgroundThe genus Populus represents an economically and eco-logically important set of species whose uses include pulpand paper, carbon sequestration, phytoremediation, and,more recently, feedstock for the lignocellulosic biofuels in-dustry [1,2]. However, a barrier to widespread adoption ofPopulus and other plants as a biofuels feedstock is the in-herent recalcitrance of cell walls to enzymatic digestionand microbial deconstruction into simple sugars. This re-calcitrance leads to increased processing costs for conver-sion of biomass into liquid fuels [3]. Genetic manipulationof biomass to enhance sugar release from the harvestedaerial portion of plants has been a key approach in effortsto establish an economically sustainable lignocellulosicbiofuels industry [4,5].Quantitative trait locus (QTL) studies have been suc-cessfully used in Populus to identify genomic regions asso-ciated with cell wall recalcitrance phenotypes includinglignin content, syringyl-to-guaiacyl (S/G) ratio, and 5- and6-carbon sugar content [6,7]. However, QTL mapping ap-proaches do not have sufficient mapping resolution toidentify the causal genes due to extensive linkage and thelarge concomitant genomic intervals that typically encom-pass dozens to hundreds of genes. Regardless of this limita-tion, QTL analysis provides a useful approach to mitigatefalse positives in genetic mapping studies. As a comple-mentary approach, linkage disequilibrium (LD)-based asso-ciation mapping has been proposed as a fine mappingapproach with potential for single-gene resolution [8].Recent studies in Populus have demonstrated the feasi-bility of association mapping by a priori targeting geneswith known function in cell wall biosynthesis [9-12]. Thesestudies have characterized the extent of LD over physicaldistance and reported r2 dropping to 0.1 within 200 to700 bp. However, Guerra et al. observed distances up to2.5 kb for a single candidate gene [10]. The latter estimatewas in close agreement with findings reported by Slavovet al. in a population of unrelated P. trichocarpa genotypes,where r2 dropped below 0.2 within 3–6 kb on a whole-genome scale, as well as in the vicinity of coding regions[13]. Given that the average gene size for P. trichocarpa isapproximately 5 kb, these results suggest that fine-scalemapping using association approaches in Populus shouldbe possible.In this study, we sought to identify loci underlying cellwall recalcitrance phenotypes using a sequential combin-ation of P. trichocarpa x P. deltoides pseudo-backcross-based QTL mapping and then association mapping amongtwo partially overlapping populations of unrelated P. tri-chocarpa genotypes to assess co-location of significantSNP trait associations within QTL intervals. We also re-port on the reproducibility of associations across alternatephenotyping and genotyping platforms, as well as acrossmultiple sampling environments and varying plant ages.Finally, using the Populus protoplast transient expressionassay, we demonstrate the effects of allelic variants of can-didate genes on reporter gene expression for cellulose,hemicellulose, and lignin biosynthetic pathways.ResultsPhenotyping for cell wall traitsThere was significant phenotypic variation in all popula-tions for cell wall traits analyzed in this study. In thepseudo-backcross pedigree, which was characterized usingpyrolysis molecular beam mass spectrometry (pyMBMS),phenotypic values for each trait were significantly corre-lated within and between age 2 and age 4 plant materials(Additional file 1A). Additionally, we observed correla-tions between different phenotypes which were generallyhigher within the same environment. For example, ligninand S/G ratio were significantly correlated in age 2 plants(r = 0.37, p ≤ 0.00001) and age 3 plants (r = 0.36, p ≤0.00001), and 5- and 6-carbon sugars were negatively cor-related with lignin content in the age 2 plants (r = −0.65,p ≤ 0.00001 and r = −0.77, p ≤ 0.00001, respectively).The BESC association mapping population was charac-terized using pyMBMS and saccharification assays usingwood samples collected from the native, Corvallis, andClatskanie environments; the Surrey population was char-acterized using traditional wet chemistry using samplescollected from the Surrey field site. Phenotypic correlationswere generally higher within the same environment withvarying levels of significance across different environments(Additional files 1B and C). The S/G ratio exhibited thehighest correlation across different environments, achiev-ing r = 0.43, p ≤ 0.00001 (n = 258) between the Corvallisand Clatskanie common gardens and r = 0.31, p ≤ 0.00001(n = 795) between the Clatskanie and native environments.Similarly, the S/G ratio had the highest correlation be-tween different phenotyping platforms, reaching r = 0.61,p ≤ 0.00001 (n = 146) between the pyMBMS-characterizednative and the wet chemistry-characterized Surrey envi-ronments (Additional file 1D). Glucose release was nega-tively correlated with lignin content in both native andClatskanie environments as well as between the native en-vironments and the Surrey populations phenotyped usingdifferent platforms (Additional file 1D). However, the mar-ginal to low levels of correlation suggest significant envir-onmental and developmental effects on trait expression.Details for phenotyping characteristics are provided inAdditional file 2.SNP genotyping in pseudo-backcross pedigree and geneticmap constructionWe incorporated 3,568 of the 3,751 segregating SNPmarkers into 19 linkage groups (LGs) corresponding tothe 19 Populus chromosomes. The map was 3,053.9 cM inlength, with the longest linkage group being 3791.2 cM forFrom the Populus 34 K Infinium array-based associationAn amino acid transporter, Potri.014G036500, harbored aMuchero et al. BMC Genomics  (2015) 16:24 Page 3 of 14LG I and the shortest being 98.7 cM for LG XIX. Thenumber of markers in a single linkage group ranged from93 for LG XII to 458 for LG I. The average marker dis-tance was 0.8 cM, and the map covered 90% of the P. tri-chocarpa reference genome. The target LG XIV had 180SNP markers, with a median marker distance of 0.5 cMand an average of 0.8 cM. The largest marker distance was5.8 cM, and only 15 intervals had distances greater than2.0 cM.SNP genotyping in P. trichocarpa populationsArray performance data for the 34 K Illumina Infinium®SNP array are described in detail by Geraldes et al. [14],and SNP genotyping results for the Surrey population aredescribed by Porth et al. [12,15]. For the BESC population,of the 34 K SNPs, 27,940 had <10% missing data, withMAF across all loci ranging from 0.044 to 0.500. On thetarget chromosome XIV, 1,439 SNPs met the minimumcriteria (i.e., <10% missing data and MAF ≥0.05) for use inassociation mapping.QTL mappingFive-hundred-and-fifteen genotypes from the pseudo-backcross population, with both phenotypic and genotypicdata, were used in QTL mapping. A broad QTL hotspotfor lignin content, S/G ratio, and 5- and 6-carbon sugarswas identified on linkage group XIV corresponding tochromosome 14 of the Populus genome (Figure 1). Using adrop in LOD score of 1 between peaks to distinguishneighboring QTLs, we identified five putative QTLs eachfor S/G ratio and lignin content and three for 5- and 6-carbon sugars (Additional file 3). All QTLs exceeded thegenome-wise LOD significance thresholds for each pheno-type in each experiment with percentage phenotypic vari-ance explained (% PVE) ranging from 1.9 to 7.5%. QTLprofiles were reproducible between phenotypic data setscollected in two different years on 2- and 4-year-old pro-geny, respectively (Figure 1). Individual SNPs co-locatingwith QTL peaks were highly consistent between age 2 andage 4 plant materials. Lignin content and 5- and 6-carbonsugar contents had the most robust co-location of QTLpeaks. In each case, the same SNP markers had the highestLOD scores for each phenotype in each experiment/year(Additional file 3). The two largest QTLs occurred adjacentto each other and were both associated with S/G ratio,achieving LOD scores above 8 in each case. SNP markersscaffold_14_7068969 and scaffold_14_7559196 were asso-ciated with the QTL peaks in both the 2008 and 2010phenotypic datasets, both exceeding 7% PVE. QTL inter-vals ranged from 0.559 to 1.766 Mb (Additional file 3).Population structure within the association populationAfter excluding genotypes exhibiting evidence of clonalityand high levels of relatedness for the BESC population, weSNP, scaffold_14_2979511, which had a LOD score of 5.72and was at the peak position of the QTL identified for S/Gratio in the pseudo-backcross population (Additional file 3,Figure 1A). In Clatskanie, the same gene harbored a SNP,scaffold_14_2980220, which was significantly associatedwith pyMBMS peak m/z = 135 (p = 1.27E−05). pyMBMSpeak 135 is derived from 4-vinyl-guaiacol, a component ofG lignin [16]. Four additional SNPs mapping within a6.5 kb interval about this same gene had significant associa-tions with MBMS-estimated 6-carbon sugars (p = 1.14E−07)and glucose/xylose release (p = 3.86E−06) in Corvallis, glu-mapping effort, we identified seven SNPs within six candi-date genes which exceeded the chromosome-wide –log10(P) = 4.46 [p = 3.47E−05 (p ≤ 0.05)] Bonferroni-adjusted sig-nificance threshold (Additional file 4). Each of the sixgenes also harbored SNPs exhibiting significance whencorrection for false positives was conducted at the QTLinterval level. Altogether, twelve SNPs from six candidategenes were ranked first in 14 unique marker-trait associa-tions across the four sampling environments (Additionalfile 4). Reanalysis of candidate gene intervals saturatedusing whole-genome resequencing data identified 21 SNPsfrom five of the six intervals with significant trait associa-tions (Table 1 and Additional file 5). SNPs from the rese-quencing data set with the lowest p-values mapped within10.0 kb or less across multiple environments for five of thesix intervals identified using the Infinium array data(Table 1). For the single remaining interval which en-compassed a 17.9 kb-candidate gene, SNPs mapped within1.5 kb across all three environments for the Infiniumarray and within 30.7 kb across two environments forresequencing-based associations (Table 1). Significant as-sociations were detected within QTL intervals for all phe-notypes noted above (i.e., S/G ratio, lignin content, and5- and 6-carbon sugars).Candidate genesanalyzed a set of 886 genotypes for population structure.There was a substantial increase in probability In P(D) asa function of number of subpopulations from K = 1 up toK = 6. The smallest differences among In P(D) values wereobserved from K = 7 up to K = 10, after which the valuesexhibited substantial decrease between K = 11 and K = 15.We selected K = 10, which had the highest ln P(D), as thenumber of subpopulations in the Q matrix generated as acovariate in association analysis. Attributes of the Surreypopulation are described by Porth et al. [12].Association mappingcose release (p = 6.54E−05) in Clatskanie, and arabinose con-tent (p = 5.63E−04) in Surrey (Additional file 4, Table 1).Muchero et al. BMC Genomics  (2015) 16:24 Page 4 of 14A KANADI transcription factor, Potri.014G037200,harbored SNP markers scaffold_14_3027959 and scaf-fold_14_3028120 that showed significant associations withglucose/xylose (p = 6.57E−06) release and glucose release(p = 1.38E−05) based on a saccharification assay from woodsamples collected from the parental trees in their nativeenvironments (Additional file 4, Table 1). SNPs foundFigure 1 LOD score profiles (solid lines) showing QTLs on Populus chr6-carbon (C6) sugars and co-location of QTL peaks with SNP-trait asso(B) copper transport ATOX1-like, and (C) Ca2+ transporting ATPase gewithin a 10 kb region encompassing the transcription fac-tor were also significantly associated with glucose releasefrom wood samples collected in Clatskanie (p = 8.89E−05)and Corvallis (p = 8.64E−05) (Table 1).An ANGUSTIFOLIA CtBP transcription factor, Potri.014G089400, harbored SNPs from the Infinium arraythat were significantly associated with xylose releaseomosome XIV for lignin content, S/G ratio, 5-carbon (C5) andciations (closed circles) for the (A) amino acid transporter,nes.aitBMbinBMcosasesecosaseBMragsitBMMuchero et al. BMC Genomics  (2015) 16:24 Page 5 of 14Table 1 SNP trait associations with the highest significanceNearest gene Environment Infinium arraySNP marker p-value TraPotri.014G036500 Clatskanie scaffold_14_2980220 1.27E−05 pyM(Amino acidtransporter)Corvallis - - -Surrey scaffold_14_2977633 5.63E−04 AraPotri.014G037200 Clatskanie scaffold_14_3028570 8.55E−04 pyM(KANADI TF) Corvallis - - -Native scaffold_14_3028120 6.57E−06 GlurelePotri.014G089400 Corvallis scaffold_14_7043301 1.06E−05 Xylo(ANGUSTIFOLIA TF) Native scaffold_14_7044284 6.84E−04 GlurelePotri.014G089700 Clatskanie scaffold_14_7055338 1.32E−05 pyM(Coppertransporter)Corvallis - - -Native - - -Surrey scaffold_14_7053760 9.18E−05 AvedenPotri.014G101900 Clatskanie scaffold_14_7971054 8.31E−05 pyM(Ca2+−transporting Corvallis - - -(p = 1.06E−05) at the chromosome-wise threshold fromwood samples collected from Corvallis and glucose/xyloserelease (p = 6.84E−04) at the QTL-wise threshold fromwood samples collected from their native environment.There were no significant associations when we reanalyzedthe same interval using SNPs from the resequencing effort.However, three SNPs spanning a 2.7 kb region had suggest-ive associations with glucose/xylose release (p = 6.84E−04)and 5-carbon sugars (p = 4.63E-04) in the native environ-ments and 6-carbon sugars (p = 5.36E−04) in Corvallis.A copper transport protein ATOX1-related gene,Potri.014G089700, was the nearest gene to the peak ofa major QTL for S/G ratio which peaked on SNP scaf-fold_14_7068969 (Additional file 3). The gene itselfcontained two SNPs, scaffold_14_7053760 and scaf-fold_14_7054863, which ranked first out of 1,439 SNPs fortwo traits in the Clatskanie and Surrey sites (Additionalfile 4). Three significant and two suggestive associationswere observed for SNPs within a 9.5 kb interval encom-passing Potri.014G089700. These were associated withpyMBMS peak mz = 120 (p = 1.32E−05), 6-carbon sugars(p = 1.96E−05), average wood density (p = 9.18E−05), xyloserelease (p = 3.20E−04), and S/G ratio (p = 3.36E−04) acrossall four test environments (Additional file 4, Table 1).ATPase)Native scaffold_14_7969314 1.55E−06 5-carbPotri.014G142700 Clatskanie scaffold_14_10865898 3.00E−05 pyMBM(Protein kinase) Corvallis - - -Native scaffold_14_10867394 5.59E−04 GlucosSurrey scaffold_14_10865898 7.99E−05 Syring*Suggestive associations not significant at the Bonferroni-adjusted p-value.cross different environments for six candidate intervalsRe-sequencingSNP marker p-value TraitS m/z 135 scaffold_14_2986800 6.54E−05 Glucose releasescaffold_14_2983445 1.14E−07 6-carbon sugarsose - - -S m/z 125 scaffold_14_3025294 8.89E−05 Glucose releasescaffold_14_3035343 8.64E−05 Glucose releasee/xylose scaffold_14_3029359 4.43E-04* Glucose/xylosereleaserelease scaffold_14_7041563 4.63E-04* 5-carbon sugarse/xylose scaffold_14_7044259 5.36E-04* 6-carbon sugarsS m/z 120 scaffold_14_7053121 1.96E−05 6-carbon sugarsscaffold_14_7062644 3.20E-04* Xylose releasescaffold_14_7054885 3.36E-04* S/G ratioe woody- - -S m/z 102 - - -scaffold_14_7970015 9.03E−06 Xylose releaseTwo SNPs mapping within a Ca2+ transporting ATPase,Potri.014G101900, had trait associations with p-values ex-ceeding the chromosome-wise Bonferroni-adjusted signifi-cance threshold in two of the four environments. SNPmarker scaffold_14_7969314 was associated with pyMBMS-estimated 5-carbon sugars (p = 1.55E−06) in the native envi-ronments and SNP scaffold_14_7970015 was associatedwith xylose release (p = 9.03E−06) in Corvallis (Additionalfile 4, Table 1). Two additional SNPs encompassing a4.5 kb region surrounding Potri.014G101900 had significantassociations with pyMBMS peak m/z = 102 (p = 8.31E−05)in Clatskanie and pyMBMS-estimated 5-carbon sugars(p = 5.61E−05) in the native environments (Additionalfiles 4 and 5).A protein kinase Potri.014G142700 located 517 basesupstream of the QTL peak (SNP scaffold_14_10864383)for lignin content identified in the pseudo-backcrosspedigree using wood samples from 2-year-old plants(Additional file 3). A SNP marker, scaffold_14_10865898,mapping within Potri.014G142700 ranked first in Clats-kanie (p = 3.00E−05) for association with pyMBMS peakm/z = 235, which is derived from methoxyglycolic acid[17]. The same SNP was associated with wet-chemistry es-timates for syringyl and guaiacyl monomers (p = 7.99E−05on sugars scaffold_14_7966546 5.61E−05 5-carbon sugarsS m/z 235 scaffold_14_10886410 2.42E-04* S/G ratioscaffold_14_10855744 5.49E−07 Xylose releasee release - - -yl monomers - - -and p = 9.49E−05) in the Surrey environment (Additionalfile 4, Table 1). A second SNP mapping within the samegene ranked third in the native environments for associ-ation with glucose release.Protoplast assaysUsing a protoplast transient expression assay to acti-vate reporter genes, we tested alternate alleles forthree of the six candidate genes—KANADI transcrip-tion factor, Potri.014G037200; ANGUSTIFOLIA CtBP,Potri.014G089400; and the protein kinase, Potri.014G142700.The remaining three candidate genes—the amino acidtransporter, Potri.014G036500; the copper transport pro-tein ATOX1-related gene, Potri.014G089700; and the Ca2+transporting ATPase, Potri.014G101900, are not suitablefor short-lived transient expression assays because they arenot known to function at the regulatory level required toevaluate pathway reporter gene activation.Among the alternate alleles for the KANADI transcrip-tion factor, Potri.014G037200, there was a single T → Gnucleotide substitution at position 496 of the cDNA(Figure 2A) which changed the codon TTA to a TGA pre-mature stop codon leading to a truncated protein of 164amino acids (Figure 2A). Based on the reference genomeprotein structure prediction, the MYB-like DNA-bindingdomain for this gene spans amino acid positions 135 to186. As such, the allele carrying the premature stopcodon was missing 22 amino acids at the C terminusend of the DNA binding domain. The apparent effectof the premature stop codon was supported by thetransient expression assay where the truncated proteinhad markedly lower activation of the lignin biosyntheticpathway reporter gene (CCoAOMT1) compared to thefull-length allele (Figure 3A).Among the alternate alleles for ANGUSTIFOLIA CtBP,Potri.014G089400, there was a tri-nucleotide repeat poly-morphism which resulted in the addition of CAGCAG athoMuchero et al. BMC Genomics  (2015) 16:24 Page 6 of 14Figure 2 Partial protein and cDNA alignments of alternate alleles stranscription factor, Potri.014G037200; (B) Angustifolia CtBP transcripkinase, Potri.014G142700.wing positions and effects of polymorphisms in the (A) KANADItion factor, Potri.014G089400; and (C) proteinMuchero et al. BMC Genomics  (2015) 16:24 Page 7 of 14position 96 from the start codon in one of the alleles.These polymorphisms resulted in two additional glu-tamine residues in the mature protein. As such, the allelederived from genotype BESC-293 had a longer PolyQ se-quence compared to the allele derived from BESC-470(Figure 2B). In the same allelic variants, there was also aSNP (A/G) polymorphism which resulted in a Thr/Alaamino acid substitution at positions 650 of the longer vari-ant and 648 of the shorter variant (Additional file 6). Re-sults of the transient expression assay suggested that thelonger PolyQ/Thr allele had significantly more activationof the CCoAOMT1 reporter gene compared to the shorterPolyQ/Ala allele. The opposite was true when weFigure 3 Differences in activation of reporter genes CesA8 (cellulose)of the (A) KANADI transcription factor, (B) Angustifolia CtBP transcripdeviations based on three replicates.evaluated activation of the CesA8 reporter gene wherethe shorter PolyQ allele showed significantly higher ac-tivation of the lignin pathway reporter gene (Figure 3B),suggesting that this gene might be involved in concurrentactivation/repression of the cellulose and lignin biosyn-thetic pathway.Among the alternate alleles for the protein kinase,Potri.014G142700, there was a SNP polymorphism result-ing from a single nucleotide indel at position 1,469 of thecDNA which resulted in a premature stop codon at aminoacid 355, resulting in a truncated protein (Figure 2C). Thepredicted functional domain for Potri.014G142700, theprotein tyrosine kinase domain, occurring at positions 568, GT43B (hemicellulose) and CCoAOMT1 (lignin) by allelic variantstion factor, and (C) protein kinase. Error bars indicate standardMuchero et al. BMC Genomics  (2015) 16:24 Page 8 of 14to 813, indicates that the entire kinase domain was absentin the truncated protein. The truncated protein had sig-nificantly lower activation of all three reporter genes com-pared to the allele encoding a full protein (Figure 3C).Putative loss-of-function mutations in pseudo-backcrossparental genotypesA key objective of this study was to evaluate co-locationof 1) QTLs detected in the pseudo-backcross populationand 2) significant association-based SNPs derived fromeither the Illumina array or resequence-based genotyp-ing data. Within the pseudo-backcross population, we resequenced the P. trichocarpa grandparent ‵93-968′ andthe two P. deltoides ‘ILL-101′ and ‘D124′ parents andlooked for evidence of high-impact mutations such asframeshifts and premature stop codons within the sixcandidate genes segregating in the BC1 population.Based on this analysis, the ‵93-968′ P. trichocarpa geno-type carried mutations leading to stop-gained andframe-shifts in three of the six candidate genes. For theKANADI transcription factor, an A → C nucleotide sub-stitution on position 3,030,533 resulted in a prematurestop codon. A single nucleotide insertion on position7,972,460 resulted in a frame-shift mutation within theCa2+ transporting ATPase. In the protein kinase, a singlenucleotide insertion on position 10,865,690 resulted in aframe-shift mutation. Mutations in the KANADI tran-scription factor and protein kinase occurred in the het-erozygous state, whereas the mutation in the Ca2+transporting ATPase was homozygous for the allele lead-ing to the frame-shift in the ‵93-968′ genotype. In allthree instances, the mutations fell within the intervalsdelimited by SNPs with the strongest associations acrossmultiple environments described above.DiscussionIn this study, three different phenotyping platforms wereemployed to assess variation in cell wall chemistry withinPopulus mapping populations. Of the three, the saccharifi-cation assay is markedly different from pyMBMS and trad-itional cell wall chemistry. This assay measures cell wallrecalcitrance by quantifying the amount of monomericsugars released during cell wall digestion [1-5]. Since thesemeasures tend to vary depending on pretreatment sever-ity, estimates of cell wall composition cannot be confi-dently inferred from this analysis alone. Alternatively,pyMBMS provides a rapid and high-throughput platformfor estimating cell wall composition [18-20]. However, thisanalysis provides estimates for a limited number of gross-scale measurements, including total lignin, S/G ratio, and5-carbon and 6-carbon sugars; concerns are often raisedabout the accuracy of estimates that are based on inten-sities of a small subset of the hundreds of peaks generatedduring the analysis [20]. As such, the molecular resolutionprovided by this assay is limited. On the other hand, trad-itional cell wall chemistry provides unambiguous quantifi-cation of cell wall components and provided quantitativedata on up to seventeen different cell wall components ina study by Porth et al. [15]. This method has the disadvan-tage of being laborious and expensive when it comes toscreening large numbers of samples.Ranges in pyMBMS phenotypes were generally repro-ducible among alternate sampling years for the pseudo-backcross pedigree and among environments for theassociation population. Correlations between phenotypesand between sampling years ranged from moderate to verystrong. On the other hand, correlations between the vari-ous phenotypes and between phenotypes from differentenvironments ranged from not significant to moderatelystrong. S/G ratio registered the highest correlationbetween environments with r = 0.31 between plants ofvarying ages sampled from the native environments and4-year-old plants in the Clatskanie common garden. Ingeneral, glucose release was negatively correlated with lig-nin content within and between different environments,as well as between phenotypes generated using alternateassays. Significant correlations between different pheno-types corroborate findings from previous studies that canbe attributed to two distinct phenomena. First, cell wallrecalcitrance is a physical phenomenon resulting fromthe physical protection of cellulose and hemicellulose fibersby lignin polymers which results in negative correlation be-tween percent lignin content and sugar release during thesaccharification process [1-5]. Second, pleiotropy is a mo-lecular genetics phenomenon in which the same geneticloci mediate expression of two different traits. In this re-gard phenotypes such as lignin content and S/G ratio ex-hibit high correlations and co-location of QTLs, suggestingthe existence of a common set of genes that mediate ex-pression of both traits [6,7].Our results indicate a genetic basis of cell wall recalci-trance phenotypes in Populus and support previouswork, including studies employing QTL analysis in inter-specific pedigrees [6,7] and LD-based association map-ping in diverse P. trichocarpa [9,12] and in P. nigra [10].To date, detection of genetic loci based on associationmapping has been done mostly in single environments.Our results are the first to show correspondence be-tween loci identified using the two complementary gen-etic mapping methods across multiple 1) phenotypingand genotyping platforms, 2) environments and growthstages, and 3) years for cell wall traits. This ability toevaluate complex traits across multiple variables pro-vides the much-needed validation of bona fide marker-trait associations in light of often-cited risk of false posi-tives in association mapping studies [8].There was remarkable overlap between QTLs detectedin the pseudobackcross population using samples fromMuchero et al. BMC Genomics  (2015) 16:24 Page 9 of 14age 2 and age 3 wood samples. Specifically, five putativeQTLs for S/G ratio and lignin content and three for 5-and 6-carbon sugars were reproducibly identified in thisstudy on LG XIV in the pseudobackcross population. AllQTLs exceeded the genome-wise LOD significance ineach instance. QTLs for S/G ratio exhibited the highestsignificance, exceeding LOD scores of 8 in the two inde-pendent datasets. In previous studies, Yin et al. reportedsignificant over-representation of cell wall chemistry QTLson three Populus chromosomes, including chromosomeXIV, in an F2 P. trichocarpa x P. deltoides pedigree, Family331 [7]. Subsequently, Ranjan et al. reported the genome-anchoring of a subset of these QTLs onto the Populus ref-erence genome [21]. Interestingly, a QTL identified andanchored on chromosome XIV in that study co-locatedwith a QTL we identified in the current study using a dif-ferent mapping population. Ranjan et al. anchored a QTLfor xylem S/G ratio encompassing the interval 2,282,351to 3,052,827 of the v2.2 assembly [21], which correspondswith a QTL peaking at position 2,979,511 (v3.0) for S/Gratio identified in this study. Based on the co-location ofQTLs in independent datasets, as well as separate pe-digrees, we explored co-location of loci identified usingassociation mapping in two populations of diverse P.trichocarpa genotypes.To this end, we identified 37 significant SNPs mappingwithin or adjacent to the six candidate genes on chromo-some XIV exhibiting reproducible association with cell wallchemistry phenotypes. Remarkably, we detected significantassociations for the same candidate genes for related woodchemistry traits measured in drastically different environ-ments and for samples collected from the original sourcegenotype growing in natural stands across a broad latitu-dinal range. All six genes mapped within the QTL intervalsdescribed above. In general, SNPs with significant associa-tions mapped in tight intervals for the six candidate genesacross genotyping and platforms, as well as across environ-ments and plant ages. For example, the protein kinase con-tains a SNP which ranked first in two environments forassociations with cell wall chemistry phenotypes character-ized on 4- and 9-year-old plants using pyMBMS and wetchemistry, respectively.Among the six candidate gene loci, an amino acid trans-porter gene, Potri.014G036500, exhibited the strongestevidence of co-location in both the QTL and associationanalyses. Two SNPs within 0.71 kb of each other were in-dependently associated with a QTL peak and an associ-ation mapping peak. Interestingly, the same QTL intervalwas also identified in a different mapping population,Family 331, by Ranjan et al. [21]. Amino acid transportersare recognized as key players in lignin biosynthesis sinceamino acids represent an important form of transportableorganic nitrogen [22] and the amino acid phenylalanine isa precursor for lignin biosynthesis [23].SNPs mapping within or close to a copper transportprotein ATOX1-related gene, Potri.014G089700, exhibitedreproducible associations across native and common gar-den environments as well as between phenotyping plat-forms, ranking in the top 2 associations in the Clatskanieand Surrey environments. SNPs mapping within a 1.6 kbinterval ranked first or second for associations with per-cent lignin, average wood density, and pyMBMS m/z =120 in the two environments and across the two pheno-typing platforms. Potri.014G089700 also mapped close tothe QTL peak for S/G ratio, which peaked at position7,068,969. In previous publications, copper has been im-plicated in lignin biosynthesis in other plant species. Fore-most, copper is a well-recognized enzyme cofactor forlaccase enzymes whose function has been linked to lignincontent in P trichocarpa [24]. Other examples implicatingcopper in lignin biosynthesis include increased shi-kimate dehydrogenase and peroxidase activity observed inresponse to elevated copper concentrations leading toenhanced accumulations of phenolics and lignin inCapsicum annuum seedling hypocotyls [25]. Kováčik andKlejdus also showed enhanced accumulation of phenolicacids and lignin in copper-treated Matricaria chamomilla[26]. In Pinus radiata, copper deficiency was associatedwith poorly lignified wood, leading to deformed trees [27].Based on these observations, we hypothesize that sub-cellular copper concentration may play an importantrole in regulating S/G ratio as well as lignin content. Inthat regard, cells with reduced copper concentrationsmay exhibit higher rates of lignin biosynthesis. The pre-dicted function of Potri.014G089700 in copper transporta-tion is supported by multiple repeats of the endoplasmicreticulum-homing dilysine motif (KKEE) [28] found in thepredicted protein.The ANGUSTIFOLIA CtBP gene, Potri.014G089400,which mapped within the same QTL interval and adjacentto the copper transport protein described above, was sig-nificantly associated with glucose and xylose release fromwood samples collected from both the native environmentand the Corvallis common garden. Based on transcript andproteome profiling of developing xylem in Populus, Kalluriet al. reported that Potri.014G089400 had enhanced ESTexpression and protein abundance in xylem tissue [29].Subsequent cDNA cloning and sequencing using geno-types in our study carrying alternate alleles of the twoSNPs revealed a tri-nucleotide CAGCAG repeat poly-morphism leading to a PolyQ length polymorphism, aswell as a single amino acid substitution between the twoalternate alleles. Protoplast assays using the alternate allelesindicated that the allele with the longer PolyQ sequencedisplayed significantly higher activation of the cellulosepathway reporter gene CesA8, but it had lower activationof the lignin pathway reporter gene CCoAOMT1 comparedto the shorter allele. Although the accompanying aminoMuchero et al. BMC Genomics  (2015) 16:24 Page 10 of 14acid substitution cannot be ruled out as causal, effects ofvariable-length PolyQ stretches on transcription factor ac-tivity have been well documented in diverse organisms[30]. In addition, activator/repressor activity of ANGUSTI-FOLIA CtBP transcription factor has also been reported inArabidopsis, where the Arabidopsis ortholog was shown toregulate leaf-cell expansion, arrangement of cortical micro-tubules, and the expression of genes involved in cell wallformation [31,32].cDNA sequencing of the KANADI transcription factor,Potri.014G037200, and the protein kinase, Potri.014G142700,encoding genes, each having support from multiple en-vironments and alternate genetic mapping experiments,revealed a single nucleotide substitution and a single nu-cleotide deletion, respectively, that resulted in prematurestop codons in both transcripts. In each case, the func-tional domain for each gene was partially or fully lost inthe truncated proteins. Transient expression assays for theprotein kinase suggested reduced activation of both thecellulose and the hemicelluloses pathways by the alleleharboring the premature stop codon. Similarly, the alleleharboring the premature stop codon in the KANADI tran-scription factor had markedly reduced activation of thelignin biosynthetic pathway. Based on gene family classi-fication of the Arabidopsis ortholog At5g42630, thistranscription factor belongs to the KANADI family oftranscription factors, which has been implicated in spatialarrangement of phloem, cambium, and xylem [33].Although the specific influence of the Ca2+ transport-ing ATPase gene, Potri.014G101900, on cell wall biosyn-thesis will require further validation, there were robustassociations between SNPs within Potri.014G101900 and5- and 6-carbon sugars across different environments.The importance of calcium in modulating cell wall prop-erties is well recognized [34,35]. Specifically, Lautneret al. demonstrated the favorable effect of calcium nutri-tion on wood formation in Populus using P. tremula x P.tremuloides [35]. In this study, Lautner et al. reported asignificant reduction of the syringyl units of lignin undercalcium starvation as well as an increase in fiber lengthwith increasing Ca2+ nutrition [35].ConclusionsGiven that map-based cloning of candidate genes is ex-tremely time and resource intensive, even for model spe-cies with abundant resources, the possibility of usingassociation mapping as a tool for cloning genes of func-tional relevance has the potential to accelerate genetic im-provement of the complex perennial Populus. In thisstudy, we have demonstrated the power of complementarygenetic mapping approaches (i.e., QTL- and association-based) to identify genes affecting cell wall recalcitranceacross diverse genetic backgrounds, sampling environ-ments, and phenotyping and genotyping platforms. Usingthis combinatorial approach, we identified genes that havenot been previously linked to cell wall recalcitrance inPopulus. Although the physiological impact of calcium,copper, and amino acid availability on cell wall structurehas been previously described, the genes involved in trans-portation (and thus the putative availability of these metalions and lignin precursors to developing cell walls) providehitherto unexplored targets for modulating cell wall bio-synthesis. Using the Populus protoplast transient expres-sion assay, we obtained supportive evidence for threegenes identified by our genetic approach as potential novelregulators of the expression of major cell wall biosynthesisenzymes. Using this in vivo system to confirm function ofthe putative causal alleles was instructive in dissecting thepotential biological roles of beneficial alleles.We have also demonstrated the potential value of nat-ural variants as a source and platform for molecularstudies in plants. Five of the six genes identified in thisstudy had sequence-based evidence of naturally occur-ring loss-of-function mutations. Natural variants couldserve as an important resource for studying of genefunction in systems such as Populus in which geneticmanipulation can be cumbersome. In breeding pro-grams, genotypes carrying loss-of-function and enhancermutations could be strategically used in marker-assistedbreeding schemes to pyramid complementary mutationsthat might result in superior phenotypes, similar to the‘breeding with rare defective alleles’ (BRDA) approach il-lustrated by Vanholme et al. [36]. Finally, the apparentenhancement of transcription factor activity by thePolyQ stretch suggests the possibility of using naturallyheightened activity as well as opportunities to engineermultiple tandem PolyQ segments for enhanced versionsof transcription factors regulating the expression ofgenes affecting economically important traits.MethodsPlant materialsQTL mapping pedigreeA pseudo-backcross population with 712 individuals wasestablished in a replicated field trial in Morgantown, WV(39°39′32″N 79°54′19″W). The trial consisted of two clonalreplicates in an interlocking block design [37]. Genotypeswere initially planted at 2 m × 2 m spacing using rooted cut-tings in June 2007. The trial was thinned by removing 50%of clones in a diamond fashion in December 2008 (age 2),leaving plants at 2.83 m × 4 m spacing. The remaining ge-notypes were harvested in January 2010 (age 4).Association mapping populationsA population of 1,089 black cottonwood genotypes (P. tri-chocarpa) was collected from native stands to encompassthe central portion of the natural range of the species,stretching from 38.8° to 54.3° N latitude from California toMuchero et al. BMC Genomics  (2015) 16:24 Page 11 of 14British Columbia [13], hereafter referred to as the BESCpopulation. Wood samples for this study were collectedfrom the native ortets in addition to clonally replicatedwood samples from field sites in Corvallis, OR (44°34′14.81″N 123°16′33.59″W) (age 2) and Clatskanie, OR(46°6′11″N 123°12′13″W) (age 4). A partially overlappingand independently phenotyped population of 499 P. tri-chocarpa genotypes was assembled from 44° to 58.6° Nand established in Surrey, British Colombia, as describedby Porth et al. [15]. The population shared 146 genotypeswith the BESC population. Wood samples for phenotypingwere collected from the Surrey plantation from 9-year-oldtrees. Details about the establishment, sampling and phe-notyping of the mapping pedigree and association popula-tions are given in Additional file 2.Although overlap in genotypes was observed across mul-tiple locations, differences in age at sampling representedan additional variable with significant contribution inphenotypic expression. As such, each defined environmentwas taken to represent the physical and climatic conditionsas well as the developmental stage of the plants at each site.PhenotypingWood disks cut from each stem 1.2 m off the ground foreach genotype in the pseudo-backcross mapping pedigreewere collected in December 2008 and February 2010 from2- and 3-year-old plants, respectively. In January 2008,4.3 mm increment cores were collected from 570 of the1,100 ortet P. trichocarpa genotypes in their native envi-ronments in the core of the range (from the ColumbiaRiver in northern Oregon to the Skagit River in northernWashington). In December 2010, 300 single-replicate stemdisks were harvested from 2-year-old plants in Corvallis,OR. In June 2012, 4.3 mm increment cores were collectedfrom 932 4-year-old plants in Clatskanie, OR. Of these932 genotypes, 235 had two clonal replicates. Debarkedand air-dried increment cores and stem disks were groundusing a Wiley Mini-Mill (Swedesboro, NJ) with a 20-meshscreen. Lignin content, S/G ratio, and 5- and 6-carbonsugar content were determined by pyMBMS analysis, andglucose and xylose release were characterized based onthe saccharification assay. These assays were conducted atthe National Renewable Energy Laboratory (Golden, CO)as described below. The Surrey population was character-ized for 17 cell wall traits using traditional wet chemistryassays at the University of British Colombia, Vancouver,BC, Canada as described in Porth et al. [15].For the BESC population, 300, 797, and 926 genotypeswere available for analyses in Corvallis, native, andClatskanie environments, respectively. After eliminatinggenotypes with evidence of sibship [12] and missing SNPdata >10%, the BESC and Surrey populations shared 146genotypes for phenotypic correlation analysis and 123genotypes for association mapping analysis.SaccharificationWood samples were treated with α-amylase (spirizymeUltra—0.25%, Novozymes, North America, Inc., Franklinton,NC) and β-glucosidase (Liquozyme SC DS—1.5%,Novozymes) in 0.1 M sodium acetate (24 h, 55°C, pH 5.0)to remove starch (16 ml enzyme solution per 1 g biomass).This was followed by Soxhlet extraction in ethanol (95%v/v) for 24 h to remove extractives. After overnight drying,5 mg (±0.5 mg) of extractives-free biomass was weighed intriplicate into a solid Hastelloy 96-well microtiter plate.250 μl H2O were added, and samples were sealed with sili-cone adhesive and Teflon tape and heated at 180°C for40 min. Once cooled, 40 μl of buffer-enzyme stock wasadded. The buffer-enzyme stock consisted of 8% CTec2(Novozymes) in 1 M sodium citrate buffer. The sampleswere then gently mixed and left to statically incubate at50°C for 70 h. After the 70 h incubation, an aliquot of thesaccharified hydrolysate was diluted and evaluated usingthe glucose oxidase/peroxidase and xylose dehydrogenaseassays (Megazyme International Ireland, Wicklow, Ireland).Results were calculated using calibration curves con-structed from standard mixtures of glucose and xylose.Pyrolysis MBMSA commercially available MBMS designed specificallyfor plant biomass analysis was used for pyrolysis vaporanalysis [18-20]. Briefly, approximately 4 mg of air-dried20 mesh biomass was introduced into a quartz pyrolysisreactor via 80 μl stainless steel Eco-Cups provided withthe autosampler. Mass spectral data from 30–450m/zwere acquired on a Merlin Automation data system ver-sion 3.0 using 17 eV electron impact ionization.Lignin estimates were determined by summing the inten-sities of peaks assigned to lignin compounds as describedby Sykes et al. [20]. The lignin intensities were then cor-rected to a standard with a known Klason lignin contentusing a single point correction technique. S/G ratios weredetermined by summing the peaks ascribed to syringylmoieties (namely, m/z 154, 167, 168, 182, 194, 208, and210) and dividing by the sum of peaks ascribed to guaiacyl-derived units (m/z 124, 137, 138, 150, 164, and 178).Sugar estimates were determined by summing the inten-sities of peaks assigned to 5- and 6- carbon sugars frommodel compound experiments [18]. 5-carbon sugars weredetermined by summing peak intensities of m/z 57, 73, 85,96, and 114, while 6-carbon sugars were estimated bysumming peak intensities m/z 57, 60, 73, 98, 126, and 144.SNP genotyping of the ‵52-124′ pedigree and mapconstruction712 pseudo-backcross progeny and parental lines weregenotyped using a 5 K Illumina Infinium SNP array(Illumina, San Diego, CA) containing 5,390 probes.Details of array design, target SNP selection, andpression assay as described in Additional file 2. cDNA se-1991], Potri.014G089400_BESC-293 [GenBank:KP271992],Additional file 6: Protein alignment showing allelic variants of theAngustifolia CtBP transcription factor Potri.014G089400.Muchero et al. BMC Genomics  (2015) 16:24 Page 12 of 14SNP analysis and genetic map construction are givenin Additional file 2.Illumina Infinium SNP genotypingThe 34 K Illumina Infinium® SNP array described byGeraldes et al. [14] was used to genotype 991 and 334 indi-viduals of the BESC and Surrey populations, respectively.Whole-genome resequencingA second genotyping platform based on whole-genomeresequencing was used to characterize SNP and indelpolymorphisms in 673 of the 1,089 genotypes. Details ofthis analysis are provided in Additional file 2.QTL mappingA maximum likelihood inference method implemented inthe Multiple-QTL Mapping (MQM) package of MapQTL6.0 [38] was used to identify QTLs within the pseudo-backcross population. One thousand permutations wereconducted separately for each trait and experiment to deter-mine genome-wise LOD significance threshold at p ≤ 0.05[39]. QTLs were declared significant when identified (i.e.,having LOD scores above the significance threshold) in atleast two independent experiments or between two differentphenotypes in the same experiment. A drop in LOD scoreof 1.0 was used to declare adjacent QTLs as separate loci.Association mapping based on Infinium array dataBased on evidence of a major QTL hotspot for cell wallphenotypes, SNPs distributed across chromosome XIV ofthe assembly were specifically evaluated for associationwith recalcitrance phenotypes. SNPs with a MAF ≥0.05from the Infinium array and resequencing data were usedin this part of the study. Firstly, SNP trait associations wereevaluated for the Infinium array on a whole chromosomescale by including all SNPs regardless of QTL information.Secondly, we extracted SNPs lying within the ±1 LOD in-tervals delimiting individual QTLs and performed a QTLscale analysis. TASSEL 3.0 (http://www.maizegenetics.net)was used to identify marker-trait associations based on themixed linear model analysis using kinship (K) to definegenetic covariance among genotypes and population struc-ture (Q) as a covariate [40]. Cell wall chemistry pheno-types, as well as individual m/z peak intensities from thepyMBMS analysis, were analyzed. Descriptions of cell wallphenotypes can be found in Additional file 2.Association mapping based on resequencing dataCandidate gene intervals identified based on the Infiniumarray data were saturated with SNPs from the resequencingeffort and reanalyzed for associations using phenotypic datafrom Corvallis, Clatskanie, and native environments. Candi-date intervals were saturated by selecting SNPs within eachcandidate gene plus 10 kb flanking regions.Additional file 7: Validation of the constructed transient over-expression vector using the GUS reporter assay.Additional file 8: Validation of the feasibility of using Populusprotoplasts as a tool to study regulation of reporter genes for thecellulose, hemicellulose and lignin biosynthesis pathways.Additional file 9: Primer list.Additional file 10: Phenotypic variation in mapping populations.Potri.014G142700_BESC-16 [GenBank:KP271993], Potri.014G142700_BESC-120 [GenBank:KP271994]. Additionalfiles 7, 8, 9 and 10 provide supporting details for materialsand methods and results described in Additional file 2.Availability of supporting dataSNP data is available from the Joint Genome Institute’sPhytozome database (http://genome.jgi-psf.org/pages/dyna-micOrganismDownload.jsf?organism=Ptrichocarpa).Additional filesAdditional file 1: Phenotypic correlations.Additional file 2: Material and methods.Additional file 3: QTL mapping results.Additional file 4: 34 K Infinium array-based association mappingresults.Additional file 5: Whole-genome resequencing-based associationmapping results.quences from alternate alleles were deposited underaccession numbers Potri.014G037200_BESC-125 [Gen-Bank:KP271989], Potri.014G037200_BESC-293 [GenBank:KP271990], Potri.014G089400_BESC-470 [GenBank:KP27Statistical analysisCorrection for multiple testing was conducted using the un-adjusted Bonferroni correction [41] on a chromosome-wiselevel using all SNP markers and on a QTL-interval-wiselevel employing SNPs falling within QTL and candidategene intervals. Spearman’s rank correlation analyses wereperformed using the Statistix 8 software [42].cDNA cloning and Populus protoplast transientexpression assayRegulatory genes including two transcription factors and aprotein kinase, whose activity could be measured relativeto activation of reporter genes, were selected for cloningand use in a transient expression assays. Genotypes carry-ing alternate alleles were used to clone two allelic variantsof each gene for side-by-side comparison relative to anempty vector control using the protoplast transient ex-Competing interestsThe authors declare that they have no competing interests.manipulation of lignin reduces recalcitrance and improves ethanolMuchero et al. BMC Genomics  (2015) 16:24 Page 13 of 14production from switchgrass. Proc Natl Acad Sci U S A. 2011;108:3803–8.6. Novaes E, Osorio L, Drost DR, Miles BL, Boaventura-Novaes CRD, Benedict C,et al. Quantitative genetic analysis of biomass and wood chemistry ofPopulus under different nitrogen levels. New Phytol. 2009;182:878–90.7. Yin T, Zhang X, Gunter L, Priya R, Sykes R, Davis M, et al. Differentialdetection of genetic loci underlying stem and root lignin content inPopulus. PLoS ONE. 2010;5:e14021.8. Rafalski JA. Association genetics in crop improvement. Curr Opin Plant Biol.Authors’ contributionsWM, SPD, J-GC, YAE, CJD, SDM, GAT conceived of the study. JG, SJ, ACB, OCperformed molecular experiments. LEG, RS, AZ, OS, FU performed phenotypingexperiments. WM, PR, GTS, LEG, JK, IP analyzed genotypic data. WM, SPD, JM,WS, LME, GAT analyzed whole-genome resequencing data. WM, SPD, GAT, JK,IP performed statistical analyses. WM, SPD, J-GC, JG, RS, AZ, JM, WS, GAT wrotemanuscript. All authors read and approved the final manuscript.AcknowledgementsThis research was supported, in part, by funding from the BioEnergy ScienceCenter (Oak Ridge National Laboratory), a US Department of Energy (DOE)Bioenergy Research Center supported by the Office of Biological andEnvironmental Research in the DOE Office of Science. Oak Ridge NationalLaboratory is managed by UT-Battelle, LLC, for the U.S. Dept. of Energy undercontract DE-AC05-00OR22725. The work conducted by the U.S. Department ofEnergy Joint Genome Institute was supported by the Office of Science of the U.S.Department of Energy under Contract No. DE-AC02-05CH11231. We alsoacknowledge financial support from the Genome British Columbia AppliedGenomics Innovation Program (Project 103BIO) and the Genome Canada LargeScale Applied Research Program (Project 168BIO). Jianjun Guo was supported by apost-doctoral fellowship from the Natural Sciences and Engineering ResearchCouncil of Canada. The funding agencies had no role in study design, datacollection and analysis, decision to publish, or preparation of the manuscript.Author details1BioEnergy Science Center, Oak Ridge National Laboratory, Oak Ridge, TN37831, USA. 2Department of Biology, West Virginia University, Morgantown,WV 26506, USA. 3Institute of Biological, Environmental and Rural Sciences,Aberystwyth University, Aberystwyth SY23 3EB, UK. 4Bioscience Center,National Renewable Energy Laboratory, 15013 Denver West Parkway, GoldenCO 80401, USA. 5Department of Forest and Conservation Sciences, Faculty ofForestry, University of British Columbia, Forest Sciences Centre, 2424 MainMall, Vancouver, BC V6T 1Z4, Canada. 6Department of Wood Science, Facultyof Forestry, University of British Columbia, Forest Sciences Centre, 2424 MainMall, Vancouver, BC V6T 1Z4, Canada. 7Department of Botany, University ofBritish Columbia, Vancouver, BC V6T 1Z4, Canada. 8U.S. Department ofEnergy Joint Genome Institute, Walnut Creek, CA 94598, USA. 9Currentaddress: Department of Plant Biology, Carnegie Institute for Science,Stanford, CA 94305, USA. 10Department of Genetics and Physiology of ForestTrees, Faculty of Forestry and Wood Sciences, Czech University of LifeSciences in Prague, Kamýcká 129, 165 21 Praha 6, Czech Republic.Received: 27 April 2014 Accepted: 2 January 2015References1. Dinus RJ, Payne P, Sewell MM, Chiang VL, Tuskan GA. Genetic modificationof short rotation poplar wood: Properties for ethanol fuel and fiberproduction. Crit Rev Plant Sci. 2001;20:51–69.2. Studer MH, DeMartini JD, Davis MF, Sykes RW, Davison B, Keller M, et al.Lignin content in natural Populus variants affects sugar release. Proc NatlAcad Sci U S A. 2011;108:6300–5.3. Lynd LR, Cushman JH, Nichols RJ, Wyman CE. Fuel ethanol from cellulosicbiomass. Science. 1991;251:1318–23.4. Vermerris W, Saballos A, Ejeta G, Mosier NS, Ladisch MR. Molecular breedingto enhance ethanol production from corn and sorghum stover. Crop Sci.2007;47:S142–53.5. Fu C, Mielenz J, Xiao X, Ge Y, Hamilton CY, Rodriguez Jr M, et al. Genetic2010;13:174–80.9. Wegrzyn JL, Eckert AJ, Choi M, Lee JM, Stanton BJ, Sykes R, et al. Associationgenetics of traits controlling lignin and cellulose biosynthesis in blackcottonwood (Populus trichocarpa, Salicaceae) secondary xylem. New Phytol.2010;188:515–32.10. Guerra FP, Wegrzyn JL, Sykes R, Davis MF, Stanton BJ, Neale DB. Associationgenetics of chemical wood properties in black poplar (Populus nigra). NewPhytol. 2013;197:162–76.11. Du Q, Pan W, Tian J, Li B, Zhang D. The UDP-Glucuronate decarboxylasegene family in Populus: Structure, expression and association genetics.PLoS ONE. 2013;8:e60880.12. Porth I, Klápště J, Skyba O, Hannemann J, Mckown AD, Guy R, et al.Genome-wide association mapping for wood characteristics in Populusidentifies an array of candidate single nucleotide polymorphisms.New Phytol. 2013;200:710–26.13. Slavov GT, DiFazio SP, Martin J, Schackwitz W, Muchero W, et al. Genomeresequencing reveals multiscale geographic structure and extensive linkagedisequilibrium in the forest tree Populus trichocarpa. New Phytol.2012;196:713–25.14. Geraldes A, DiFazio S, Slavov GT, Priya R, Muchero W, et al. A 34 K SNPgenotyping array for Populus trichocarpa: Design, application to the study ofnatural populations and transferability to other Populus species. Mol EcolResour. 2013;13:306–23.15. Porth I, Klápště J, Skyba O, Lai BSK, Geraldes A, Muchero W, et al. Populustrichocarpa cell wall chemistry and ultrastructure trait variation, geneticcontrol and genetic correlations. New Phytol. 2013;197:777–90.16. Faix O, Meier D, Fortmann. Thermal degradation products of wood: Gaschromatographic separation and mass spectrometric characterization ofmonomeric lignin derived products. Holz als Rohund Werkstoff.1990;48:281–5.17. Sun Y-P, Nguyen KL, Wallis AFA. Ring-opened products from reaction of ligninmodel compounds with UV-assisted peroxide. Holzforschung. 1998;52:61–6.18. Evans RJ, Milne TA. Molecular characterization of the pyrolysis of biomass. 1.Fundamentals. Energy Fuels. 1987;1:123–37.19. Tuskan G, West D, Bradshaw HD, Neale D, Sewell M, et al. Two high-throughput techniques for determining wood properties as part of amolecular genetics analysis of hybrid poplar and loblolly pine. ApplBiochem Biotechnol. 1999;77:55–65.20. Sykes R, Yung M, Novaes E, Kirst M, Peter G, Davis M. High-throughputscreening of plant cell-wall composition using pyrolysis molecular beammass spectroscopy. In: Mielenz JR, editor. Biofuels: Methods and Protocols.New York: Humana Press; 2009. p. 169–83.21. Ranjan P, Yin T, Zhang X, Kalluri UC, Yang X, Jawdy S, et al. Bioinformatics-based identification of candidate genes from QTLs associated with cell walltraits in Populus. Bioenergy Res. 2010;3:172–82.22. Plavcová L, Hacke UG, Almeida-Rodriguez AM, Li E, Douglas CJ. Gene expressionpatterns underlying changes in xylem structure and function in response toincreased nitrogen availability in hybrid poplar. Plant Cell Environ.2013;36:186–99.23. Chen L, Ortiz-Lopez A, Jung A, Bush DR. ANT1, an aromatic and neutralamino acid transporter in Arabidopsis. Plant Physiol. 2001;125:1813–20.24. Lu S, Li Q, Wei H, Chang M-J, Tunlaya-Anukit S, Kim H, et al: Ptr-miR397a is anegative regulator of laccase genes affecting lignin content in Populustrichocarpa. Proc Natl Acad Sci USA 2013, doi:10.1073/pnas.1308936110.25. Dı́az J, Bernal A, Pomar F, Merino F. Induction of shikimate dehydrogenaseand peroxidase in pepper (Capsicum annuum L.) seedlings in response tocopper stress and its relation to lignifications. Plant Sci. 2001;161:179–88.26. Kováčik J, Klejdus B. Dynamics of phenolic acids and lignin accumulation inmetal-treated Matricaria chamomilla roots. Plant Cell Rep. 2008;27:605–15.27. Downes GM, Ward JV, Turvey ND. Lignin distribution across tracheid cellwalls of poorly lignified wood from deformed copper deficient Pinus radiata(D. Don). Wood Sci Technol. 1991;25:7–14.28. Manley HA, Lennon VA. Endoplasmic reticulum membrane-sorting proteinof lymphocytes (BAP31) is highly expressed in neurons and discreteendocrine cells. J Histochem Cytochem. 2001;49:1235–43.29. Kalluri UC, Hurst GB, Lankford PK, Priya R, Pelletier DA. Shotgun proteomeprofile of Populus developing xylem. Proteomics. 2009;9:4871–80.30. Atanesyan L, Günther V, Dichtl B, Georgiev O, Schaffner W. Polyglutaminetracts as modulators of transcriptional activation from yeast to mammals.Biol Chem. 2012;393:63–70.31. Chinnadurai G. CtBP family proteins: More than transcriptional corepressors.Bioessays. 2002;25:9–12.32. Kim G-T, Shoda K, Tsuge T, Cho K-H, Uchimiya H, et al. The ANGUSTIFOLIAgene of Arabidopsis, a plant CtBP gene, regulates leaf-cell expansion, thearrangement of cortical microtubules in leaf cells and expression of a geneinvolved in cell-wall formation. EMBO J. 2002;21:1267–79.33. IIegems M, Douet V, Meylan-Bettex M, Uyttewaal M, Brand L, Bowman JL,et al. Interplay of auxin, KANADI and Class III HD-ZIP transcription factors invascular tissue formation. Development. 2010;137:975–84.34. Hepler PK. Calcium: A central regulator of plant growth and development.Plant Cell. 2005;17:2142–55.35. Lautner S, Ehlting B, Windeisen E, Rennenberg H, Matyssek R, et al. Calciumnutrition has a significant influence on wood formation in poplar. NewPhytol. 2007;173:743–52.36. Vanholme B, Cesarino I, Goeminne G, Kim H, Marroni F, Van Acker R, et al.Breeding with rare defective alleles (BRDA): a natural Populus nigra HCTmutant with modified lignin as a case study. New Phytol. 2013;198:765–76.37. Libby WJ, Cockerham CC. Random non-contiguous plots in interlocking fieldlayouts. Silvae Genet. 1980;29:183–90.38. Van Ooijen JW. MapQTL®6, Software for the mapping of quantitative traitloci in experimental populations of diploid species Kyazma BV. Netherlands:Wageningen; 2009.39. Churchill GA, Doerge RW. Empirical threshold values for quantitative traitmapping. Genetics. 1994;138:963–71.40. Yu J, Pressoir G, Briggs WH, Bi IV, Yamasaki M, et al. A unified mixed-modelmethod for association mapping accounting for multiple levels ofrelatedness. Nat Genet. 2005;38:203–8.41. Bonferroni CE. II calcolo delle assicurazioni su gruppi di teste chapter“Studi in Onore del Professore Salvatore ortu Carboni” Rome 13–60. 1935.42. Analytical software. Statistix 8 User’s manual: Analytical software Tallahassee,FL. 2003.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 redistributionMuchero et al. BMC Genomics  (2015) 16:24 Page 14 of 14Submit your manuscript at www.biomedcentral.com/submit


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