UBC Faculty Research and Publications

Transcriptional programming during cell wall maturation in the expanding Arabidopsis stem Hall, Hardy; Ellis, Brian Jan 25, 2013

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

Item Metadata


52383-12870_2012_Article_1264.pdf [ 2.32MB ]
JSON: 52383-1.0223420.json
JSON-LD: 52383-1.0223420-ld.json
RDF/XML (Pretty): 52383-1.0223420-rdf.xml
RDF/JSON: 52383-1.0223420-rdf.json
Turtle: 52383-1.0223420-turtle.txt
N-Triples: 52383-1.0223420-rdf-ntriples.txt
Original Record: 52383-1.0223420-source.json
Full Text

Full Text

Transcriptional programming during cell wallmaturation in the expanding Arabidopsis stemHall and EllisHall and Ellis BMC Plant Biology 2013, 13:14http://www.biomedcentral.com/1471-2229/13/14RESEARCH ARTICLE Open AccessTranscriptional programming during cell wallmaturation in the expanding Arabidopsis stemHardy Hall1,2 and Brian Ellis1*AbstractBackground: Plant cell walls are complex dynamic structures that play a vital role in coordinating the directionalgrowth of plant tissues. The rapid elongation of the inflorescence stem in the model plant Arabidopsis thaliana isaccompanied by radical changes in cell wall structure and chemistry, but analysis of the underlying mechanismsand identification of the genes that are involved has been hampered by difficulties in accurately sampling discretedevelopmental states along the developing stem.Results: By creating stem growth kinematic profiles for individual expanding Arabidopsis stems we have been ableto harvest and pool developmentally-matched tissue samples, and to use these for comparative analysis of globaltranscript profiles at four distinct phases of stem growth: the period of elongation rate increase, the point ofmaximum growth rate, the point of stem growth cessation and the fully matured stem. The resulting profilesidentify numerous genes whose expression is affected as the stem tissues pass through these defined growthtransitions, including both novel loci and genes identified in earlier studies. Of particular note is the preponderanceof highly active genes associated with secondary cell wall deposition in the region of stem growth cessation, andof genes associated with defence and stress responses in the fully mature stem.Conclusions: The use of growth kinematic profiling to create tissue samples that are accurately positioned alongthe expansion growth continuum of Arabidopsis inflorescence stems establishes a new standard for transcriptprofiling analyses of such tissues. The resulting expression profiles identify a substantial number of genes whoseexpression is correlated for the first time with rapid cell wall extension and subsequent fortification, and thusprovide an important new resource for plant biologists interested in gene discovery related to plant biomassaccumulation.Keywords: Cell wall, Anisotropy, Growth kinematic profiling, Transcriptome, Microarray, Arabidopsis, Inflorescence stemBackgroundDirectional (anisotropic) cell wall expansion is an integralpart of most plant developmental processes, where itfacilitates the structural changes necessary for proper celland organ morphogenesis. The initial expansive growthphase, which requires both addition of new extracellularpolymers and remodeling of existing components in theprimary cell walls, is often succeeded by cell wall thickeningand rigidification processes to create secondary cell wallsthat enhance the structural integrity of the organ, but alsocurtail further wall extension. These sequential processesrequire a high degree of dynamic, context-specificcoordination of cell wall building, reconstruction and forti-fication in order to harness the underlying driving force ofturgor pressure in a spatially-defined manner (reviewed in[1,2]). Consistent with such developmental complexity, atleast one thousand genes in Arabidopsis have been shownto have some association with cell wall synthesis andremodeling [3].The gene expression patterns associated with cell wallexpansion and/or secondary cell wall formation inArabidopsis have been analyzed in several studies in effortsto identify participating genes and understand the biologicalroles of their products [4-8]. To specifically address cellexpansion processes, for example, transcript profiles havebeen compared with protein accumulation profiles inArabidopsis seedling hypocotyls that were undergoing rapidcell elongation without significant cell division [9].* Correspondence: bee@msl.ubc.ca1Michael Smith Laboratories, University of British Columbia, Vancouver BCV6T 1Z4, CanadaFull list of author information is available at the end of the article© 2013 Hall and Ellis; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the CreativeCommons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, andreproduction in any medium, provided the original work is properly cited.Hall and Ellis BMC Plant Biology 2013, 13:14http://www.biomedcentral.com/1471-2229/13/14Transcript profiling has also been conducted with in vitrocultured Zinnia mesophyll cells [10,11] and Arabidopsissubcultured cells [12] that were induced to trans-differentiate into tracheary element-like cells, a process thatis accompanied by deposition of distinctive secondary cellwall thickenings on top of the original primary cell walls. Inanother approach, large-scale correlation analysis of publicmicroarray data enabled the in silico identification of geneswhose expression is strongly aligned with expression of spe-cific members of the Arabidopsis cellulose synthase (CesA)gene family that are believed to be predominantly involvedin either primary or secondary cell wall biogenesis [13].The Arabidopsis inflorescence stem represents an attract-ive experimental system for such gene discovery studiessince it provides more substantial amounts of tissue foranalyses, and its tissue architecture is largely establishedprior to bolting, which means that stem expansion is pri-marily the product of cell elongation, rather than division.However, the growing stem also presents a continuum ofdevelopmental states along the organ as its component cellstransition from early anisotropic expansion growth togrowth cessation, and finally to cell wall fortification.Integrated with these changes in cell expansion activity areadditional changes associated with the differentiation andmaturation of the discrete tissues that comprise the stem.In order to accurately monitor the gene expressionchanges, or transcriptional programming, that accompanythese various stem growth transitions, it is essential tosample and pool stem tissues that are verifiably associatedwith specific stages of development. A number of inflores-cence stem profiling studies have attempted to comparethe global transcriptional changes occurring betweenspecific developmental stages [4,5,7,14], but the experi-mental strategies employed have typically comparedtissues from visually selected regions of multiple plants,and have operated under two untested assumptions: 1)that the pooled plants whose stems are being sampled allhave similar developmental proportioning, and 2) that thesampling guidelines for the harvested plants, derived fromdestructive analysis of a different set of plants, accuratelyassociate features such as appearance of lignification inthe interfascicular fibres of the stem [4] with specificdevelopmental growth stages. Contrary to these assum-ptions, growth kinematic profiling of expanding inflores-cence stems of individual Arabidopsis plants has recentlydemonstrated that stem growth profiles actually varywidely from plant to plant, even within genetically homo-geneous populations [15]. As a consequence, dataobtained from indirectly selected and harvested stemregions are likely to be relatively poorly correlated withonset of processes involved in cell wall extension or modi-fication events associated with specific growth stages.In this study, we have applied growth kinematic profiling(GKP) to a series of individual Arabidopsis inflorescencestems as described in Hall and Ellis (2012) [15], and usedthe resulting growth rate profiles to generate pooledsamples of stem tissue that accurately represent fourdiscrete growth stages along the cell wall expansion devel-opmental continuum. Our use of GKP-based sampling wasexpected to reduce the biological noise associated withindirect sampling strategies used in previous studies, andthus increase the sensitivity (power to detect actual differ-ential expression) and accuracy of transcript profiling.Microarray-based assessment of global transcript abun-dance in these GKP-matched stem samples then enabledus to generate transcriptomic datasets that can bepositioned with confidence within a validated developmen-tal context of cell wall expansion performance (relativeelemental growth rates). The resulting gene expressionprofiles demonstrate the participation of many genes thathad earlier been linked to primary or secondary cell wallsynthesis, but they also highlight expression changes in arange of unique genes whose role(s) in cell wall maturationor stem expansion have yet to be assessed.ResultsStage-specific transcriptome analysisIn order to position transcript profiles accurately withinthe cell wall expansion continuum, we employed growthkinematic profiling to establish relative elemental growthrates (REGRs) for contiguous stem segments harvestedfrom a series of individual plants [15]. These profilesallowed us to identify three developmental stages foreach plant being sampled: 1) an apical region wheretissues begin to differentiate, and directional cell growthis initializing (termed ‘young’, or YNG), 2) a regionwhere directional growth of the stem is most rapid(termed 'maximum growth-rate’, or MGR), and 3) aregion where elongative growth is finishing (termed'cessation', or CSS). These stages were therefore eachrepresented by samples consisting of multiple pooledsegments, each of which had been harvested from astem location centered upon a specific GKP-identifiedgrowth phase. The tissue selection protocol is outlinedin Figure 1A-C, while the growth kinematic profiles forall the plants used in this study are provided inAdditional file 1: Figure S1. To facilitate comparison ofour data with the results of other transcriptome studiesin inflorescence stems, we also harvested a segment(termed 'mature', or OLD) from the base of each of thestems and pooled these for inclusion in our transcriptprofiling analysis.To maximize both statistical power and flexibility inanalysis, our experiment directly co-hybridized all pair-wisecombinations of developmental stage samples (‘completefactorial’ experimental design) and utilized a ‘mixed effectsmodel’ analysis [16] to compare the four developmentalHall and Ellis BMC Plant Biology 2013, 13:14 Page 2 of 17http://www.biomedcentral.com/1471-2229/13/14stages on the basis of six biological replicates, each pooledfrom a common set of thirty-four randomly-assignedplants.Examination of differential expression between stagesThe goal of this study was to identify genes whose expres-sion in the inflorescence stem differs most strongly betweendifferent growth stages, since these are expected to repre-sent the loci most actively involved in the accompanyingtranscriptional reprogramming. The mixed effects model-based analysis of this experiment generates six possiblepair-wise comparisons between the four stages, for whichthe complete statistical analysis is presented in Additionalfiles 2, 3 and 4: Table S1.For detailed analysis, we focus here on three growth stagecomparisons (YNG-MGR, MGR-CSS, and CSS-OLD), andexamine the arithmetic differences between the mean (log2)signal intensities for each gene. Although as many as 4635genes are differentially expressed (q-value<0.05) betweenstages in these comparisons, we have restricted discussionto the forty largest gene expression differences, which arepresented as conventional fold-change ratios, together withthe log2 ratios from which they were derived, and measuresof false-discovery-corrected statistical significance of two-sample t-test scores (q-values), for the YNG/MGR (Tables 1and 2), MGR/CSS (Tables 3 and 4), and CSS/OLD (Tables 5and 6) comparisons. Since we have applied stringentfiltering criteria (see Table legends), these lists should bepredominantly populated by genes whose transcriptabundances are being modulated in a radical fashion duringeach associated growth transition. To look for potentialfunctional relationships among these short-listed, up- anddown-regulated genes, their annotations, gene ontology(GO) assignments and possible promoter motif enrichmentwere examined. GO term enrichment was determinedrelative to whole-genome averages using the ‘AtCoeCis’web-tool [17], which reports enrichment (fold-change),statistical significance (p-value), and the proportion of thegenes in each short-list that have been assigned thatspecific GO term ('score') (Additional file 5: Table S2,Additional file 6: Table S3, Additional file 7: Table S4).To establish the status of the stem transcriptome prior tothe period at which the maximal elongation rate has beenachieved, gene expression profiling was conducted on thetop 1 cm segment of the stem (with flowers removed), andthis YNG transcript profile was then compared to theprofile generated from segments representing the GKP-identified maximum growth rate (MGR) phase. The fortygenes exhibiting either high transcript copy number in theYNG stage sample relative to the MGR stage (positive fold-change values), or in the MGR stage sample relative to theYNG stage (negative fold-change values) are listed inTables 1 and 2, respectively Among the most up-regulatedgenes in the YNG-dominant set are two members of thefour-member ELONGATION FACTOR 1-ALPHA genefamily [18] (EF-1-α A2 (At1g07930) and EF-1-α A3Growth rate (%change/hr)Segment number from apex1412108642Time after tagging (minutes)10305070901101301500. 2 4 6051015B62R2 Plant  2Relative Growth Rate (%change/hr)Distance from stem base (cm)ApexrebmuntnemgeS123456789101112131415MGRYNGCSSOLDBA C16Figure 1 Representative growth profiling and harvesting. A) Representative surface plot of relative elongation growth rates (% change perhour, vertical axis) plotted against the number of segments (defined by optical marker tags) from the apex downwards, over the duration of theimaging period in 10 minutes intervals. The darker grey-shaded, nearest profile denotes the last 10-minute interval before harvest, depicted in thegreater detail in the right-hand scatterplot. B) Corresponding scatter plot of growth rates (% change in length per hour) against distance fromthe stem base for specific segments. Segments are numbered from the top of the plant downwards in the right-hand margin. The LOWESSregression curve follows the best fit through the growth rate data for this plant over a given 10' interval. Green dotted lines represented 65%confidence intervals for the LOWESS regression curve. Closed-box/arrow indicates the stem position that matches the maximum growth rate ofthe regression curve (segment 5), plotted as the right-most vertical dotted line, while the open-box/arrow indicates the first position below thetop of the stem where the growth rate falls to zero (segment 10). C) Harvesting zones for young (YNG), maximum growth-rate (MGR), cessation(CSS), and stem base (OLD) zones based upon LOWESS curve. See methods for description of zone establishment. See Additional file 1: Figure S1for complete set of 34 growth kinematic profiles.Hall and Ellis BMC Plant Biology 2013, 13:14 Page 3 of 17http://www.biomedcentral.com/1471-2229/13/14(At1g07940)). This association with active protein synthesisis also reflected in the GO term enrichment analysis of thisset (Additional file 5: Table S2), which indicates >80-foldenrichment in GO terms containing ‘translation’. Alsofound within this ‘YNG-up-regulated’ list are genes relatedto signaling (RLK902 [19]; CLE16(CLAVATA3 homologue)[20]; LOX2 [21]; At1g62950, a LRR protein kinase), as wellas transcription factors (ZF-HD class AtHB33; NAC063).Cell-cell communication mediated by peptides derivedfrom CLE gene products, acting together with cognatereceptor kinases, represents part of the elaborate signalingnetwork that helps guide plant development [22]. Whileknown cell wall-associated genes are not notablyover-represented within the ‘YNG-up-regulated’ list,one gene encoding a putative glucan endo-β(1→3)-glucosidase (At4g14080) [23] is up-regulated 15-foldover the MGR stage.Although cell wall expansion is expected to be takingplace in both the YNG and the MGR stage tissues, geneswhose products are uniquely required for rapid expansionshould be relatively more highly expressed in the latter.The most strongly differentially up-regulated (~40-fold)gene in the MGR tissues relative to YNG is a peroxidase(PER64) that has been previously reported to be up-regulated in stems in response to mechanical load [24].The peroxidase gene family in Arabidopsis is large, and itsmembers play a number of roles in cellular metabolism,including modulation of reactive oxygen species accu-mulation [25] and the oxidative coupling of aromaticmetabolites such as the monolignols that serve asprecursors for the lignin polymer [26,27]. The expres-sion of PER64 in Arabidopsis has been shown to beconcentrated in the protoxylem [28], where lignifica-tion of patterned secondary cell wall thickeningscontributes to cell wall stabilization during vascularelongation, a spatial specificity that is consistent withthe strong PER64 expression in MGR tissues.The MGR up-regulated list also contains several genesmore directly related to primary cell wall formation andre-modeling, including a xyloglucan endotransglycosylase/hydrolase MERISTEM-5 (MERI5B/XTH24), a putativepectinase (At1g80170), two arabinogalactan proteins(AGP12, AGP13) and a MYB transcription factor(MYB61) that has recently been shown to contributeto both cell wall synthesis and regulation of plantcarbon allocation [29-32]. In addition, several genesencoding proteins associated with phytohormone sig-nalling are more highly expressed in the MGR tissues,Table 1 Twenty most differentially expressed genes with higher expression in YNG stage relative to MGR stageAccession Gene annotation1 YNG/MGR Fold-change2 q-value3AT1G07930 elongation factor 1-alpha / EF-1-alpha 22.5 4.1E-02AT5G25754 unknown protein 19.2 4.7E-02AT1G11520 spliceosome associated protein-related 18.9 3.4E-02AT1G07940 elongation factor 1-alpha / EF-1-alpha 18.9 4.1E-02AT5G22430 unknown protein 17.6 2.2E-02AT1G75240 ARABIDOPSIS THALIANA HOMEOBOX PROTEIN 33 (AtHB33) 16.2 1.4E-02AT4G14080 MATERNAL EFFECT EMBRYO ARREST 48 (MEE48) 15.4 2.2E-02AT1G01300 aspartyl protease family protein located in membrane, plant-type cell wall 14.6 4.0E-02AT2G07739 unknown protein 14.4 4.1E-02AT3G13470 chaperonin, putative with domain Cpn60/TCP-1 (InterPro:IPR002423) 13.1 2.3E-02AT4G34850 chalcone and stilbene synthase family protein involved in phenylpropanoid biosynthetic process 13.0 2.5E-02AT3G17840 RECEPTOR-LIKE KINASE 902 (RLK902) 12.6 3.0E-02AT1G52030 myrosinase binding protein, putative 12.2 2.2E-02AT5G15720 GDSL-MOTIF LIPASE 7 (GLIP7) 11.6 3.7E-02AT2G18020 EMBRYO DEFECTIVE 2296 (EMB2296) 11.5 3.6E-02AT5G62080 protease inhibitor/seed storage/lipid transfer protein (LTP) family protein 11.4 2.9E-02AT2G01505 CLAVATA3/ESR-RELATED 16 (CLE16) 10.8 4.3E-02AT3G45140 LIPOXYGENASE 2 (LOX2) 10.6 1.5E-02AT1G62950 leucine-rich repeat transmembrane protein kinase 10.4 2.7E-02AT3G55210 ARABIDOPSIS NAC DOMAIN CONTAINING PROTEIN 63 (anac063) 10.0 3.0E-02AT4G04720 CPK21 9.9 4.2E-021Gene descriptions are abbreviated from TAIR10 genome release. Putative functions are stated in lowercase.2Genes ranked according to fold-change values derived from log2 ratios of YNG (numerator) and MGR (denominator).3 Derived from false discovery-rate correction of p-values.Hall and Ellis BMC Plant Biology 2013, 13:14 Page 4 of 17http://www.biomedcentral.com/1471-2229/13/14including a putative ACC oxidase, the GA-responsiveMINI ZINC FINGER 1 (MIF1) [33] and another geneGASA6 (At1g74670) reported to be GA-responsive [34].Comparison of the MGR stage gene expression patternsto those observed at the more mature CSS stage providesanother view of those genes that are most relevant to activestem expansion, by contrasting their performance in therapidly expanding MGR tissues with that seen in the CSStissues where cell wall expansion has ceased. Interestingly,the list of twenty genes whose expression is ‘Higher inMGR relative to CSS’ (Table 3) is led, not by genes knownto be associated with cell wall synthesis or modification,but byMAJOR LATEX PROTEIN 423 (MLP423), a memberof the BET V1 class of allergens that exhibits sequencehomology to ABA- and stress-responsive proteins fromvarious plant species (EMBL-EBI database information).MLP423 is accompanied by two members of the large(108-member) GDSL-type lipase homologue gene family,and by other genes associated with lipid metabolism/transport, but few, if any, genes known to be directlyinvolved in cell wall synthesis are included. This profileimplies that the genes populating the ‘Higher in MGRrelative to CSS’ list are primarily those whose expression isrelatively strongly reduced as the cells make their transitionfrom rapid anisotropic expansion to maturation.The ‘Higher in CSS relative to MGR’ gene list (Table 4),on the other hand, would be expected to capturethose genes that make a major contribution to there-programming associated with transition to a phase ofcell wall stabilization and rigidification. Consistent with thisprediction, this list is dominated by genes associated withformation of non-expanding walls, including all three of thecellulose synthase genes believed to be involved in cellulosemicrofibril deposition during secondary cell wall biosyn-thesis (CESA4/IRX5, CESA7/IRX3, and CESA8/IRX1)[14,35-37], and CHITINASE-LIKE PROTEIN 2 (CTL2) [38]whose loss-of-function mutant displays cellulose biosyn-thesis defects [39]. Also strongly represented are genesrequired for xylan biosynthesis/modification, including aUDP-GLUCURONIC ACID DECARBOXYLASE 3/UXS3[40] that provides UDP-xylose for xylan backbone synthe-sis, IRREGULAR XYLEM 9 (IRX9) [41] and FRAGILEFIBER 8 (FRA8) [42] whose encoded proteins build and ex-tend the glucuronosylxylan polymer, and two xylan modifi-cation genes: a xyloglucan-specific endotransglycosylase/hydrolase 19 (XTH19) [43], and REDUCED WALLACETYLATION 1 (RWA1) [44]. Other cell wall modifica-tion genes are present, including two pectinesterases(At2g43050, At2g45220), one of which is the most stronglydifferentially-expressed gene in the list. The prominence ofTable 2 Twenty most differentially expressed genes with higher expression in MGR stage relative to YNG stageAccession Gene annotation1 YNG/MGR Fold-change2 q-value3AT3G13520 ARABINOGALACTAN PROTEIN 12 (AGP12) −6.8 3.1E-02AT1G80170 putative polygalacturonase (pectinase) −7.3 1.4E-02AT1G09540 MYB DOMAIN PROTEIN 61 (MYB61) −7.4 2.6E-02AT3G05880 RARE-COLD-INDUCIBLE 2A (RCI2A) −7.5 1.3E-02AT1G77330 similar to 1-aminocyclopropane-1-carboxylate oxidase −7.5 1.9E-02AT1G72430 Auxin responsive SAUR protein −8.0 3.2E-02AT4G23496 SPIRAL1-LIKE5 (SP1L5) −8.1 4.7E-02AT4G03205 SOUL heme-binding family protein −8.1 1.9E-02AT1G67865 unknown protein −8.2 3.4E-02AT4G26320 ARABINOGALACTAN PROTEIN 13 (AGP13) −8.8 1.9E-02AT3G19710 BRANCHED-CHAIN AMINOTRANSFERASE4 (BCAT4) −9.6 1.5E-02AT5G48560 basic helix-loop-helix (bHLH) family protein −9.9 4.5E-02AT3G55240 Overexpression leads to PEL (Pseudo-Etiolation in Light) phenotype −10.0 1.5E-02AT1G74670 putative gibberellin-responsive protein (GASA6) −10.2 3.3E-02AT4G30270 MERISTEM-5 (MERI5B) −11.5 1.6E-02AT1G74660 MINI ZINC FINGER 1 (MIF1) −12.0 4.9E-03AT4G29905 unknown protein −12.9 4.3E-02AT3G45160 unknown protein −15.3 1.1E-02AT5G05960 protease inhibitor/seed storage/lipid transfer protein (LTP) family protein −24.2 1.5E-02AT5G42180 peroxidase 64 (PER64) located in plant-type cell wall −40.0 9.4E-041Gene descriptions are abbreviated from TAIR10 genome release. Putative functions are stated in lowercase.2Genes ranked according to fold-change values derived from log2 ratios of YNG (numerator) and MGR (denominator).3 Derived from false discovery-rate correction of p-values.Hall and Ellis BMC Plant Biology 2013, 13:14 Page 5 of 17http://www.biomedcentral.com/1471-2229/13/14these pectin de-methylating enzymes in the MGR→CSStransition list is consistent with a current model for plantcell wall rigidification in which a reduction in the levels ofpectin methylesterification leads to enhanced calcium ioncross-linking and wall stiffening [45-48].In addition to genes whose encoded products affectcell wall polysaccharide biosynthesis, the list includesIRREGULAR XYLEM 12 (IRX12/LAC4). Laccases arethought to contribute to polymerization of lignin insecondary walls, and LAC4 expression has previously beenshown to be specific to xylary and interfascicular fibres inthe Arabidopsis stem. Lignin deposition is largelyunaffected in the lac4 loss-of-function mutant, but isstrongly reduced in the lac4/lac17 double loss-of-functionmutant [49]. It is noteworthy that we observed no signifi-cant difference in expression of LAC17 between the CSSand MGR stages (1.3-fold differential, CSS/MGR). Overall,nine of the twenty genes featured in this list also occuramong a set of ‘xylem-specific’ Arabidopsis genes identi-fied through analysis of public datasets [50], consistentwith a metabolic commitment in CSS tissues to cell wallrigidification in xylem fibres and tracheary elements oncestem expansion ceases.While growth kinematic data cannot precisely positionthe base of the stem along the developmental continuum(growth kinematic profiling can only distinguish stemregions on the basis of their rates of expansion), it is clearfrom previous microscopic analysis [4,51] that the OLDstage tissue displays an advanced phase of organ growthand cell wall maturation in the 10-15 cm tall Columbiaplants examined in this study. Based on our presentunderstanding of the stem maturation process, the CSS andOLD samples are expected to contain tissues activelyengaged in earlier and later stages of secondary cell wallformation and reinforcement, respectively. Tables 5 and 6present the twenty genes whose expression is ‘Higher inCSS relative to OLD’ and the twenty genes whoseexpression is ‘Higher in OLD relative to CSS’, respectively.Displaying high expression in the CSS relative to OLDsamples are GERMIN-LIKE PROTEIN 3 (GER3/GLP3)(At5g20630) and GERMIN-LIKE PROTEIN 1 (GER1/GLP1) (At1g72610). GER3 also appeared in the list ofgenes more highly expressed in MGR tissues than in CSS(Table 3), indicating that expression of this member of theGER gene family follows a steeply declining trajectoryduring the stem maturation process. While specificdevelopmental roles for GLP1 and 3 have yet to be iden-tified, GER proteins are apoplastic glycoproteins thathave been widely associated with plant disease resistanceand ROS modulation, particularly in the cereals [52].Table 3 Twenty most differentially expressed genes with higher expression in MGR stage relative to CSS stageAccession Gene annotation1 MGR/CSS Fold-change2 q-value3AT1G24020 MLP-LIKE PROTEIN 423 (MLP423) 16.7 2.9E-03AT5G33370 GDSL-like lipase 5.7 2.3E-02AT2G02320 PHLOEM PROTEIN 2-B7 (AtPP2-B7) 5.2 6.6E-03AT2G38540 LIPID TRANSFER PROTEIN 1 (LP1) 5.2 5.0E-05AT5G24780 VEGETATIVE STORAGE PROTEIN 1 (VSP1) 4.8 3.6E-03AT2G02850 PLANTACYANIN (ARPN) 3.9 2.4E-03AT2G33810 SQUAMOSA PROMOTER BINDING PROTEIN-LIKE 3 (SPL3) 3.8 4.2E-02AT3G04290 LI-TOLERANT LIPASE 1 (LTL1) 3.7 2.4E-03AT1G55490 chloroplast 60 kDa chaperonin beta subunit 3.6 9.1E-04AT5G20630 GERMIN 3 (GER3) 3.5 3.8E-04AT2G39670 radical SAM domain-containing protein 3.5 2.1E-02AT3G47650 bundle-sheath defective protein 2 family / bsd2 family 3.4 3.2E-02AT5G15230 GAST1 PROTEIN HOMOLOG 4 (GASA4) 3.4 3.1E-04AT3G47340 GLUTAMINE-DEPENDENT ASPARAGINE SYNTHETASE (ASN1) 3.2 1.3E-02AT5G20720 CHAPERONIN 20 (CPN20) 3.2 7.6E-03AT5G55450 protease inhibitor/lipid transfer protein (LTP) family protein 3.2 3.6E-03AT5G61170 40S ribosomal protein S19 (RPS19C) 3.1 4.3E-02AT3G08740 elongation factor P (EF-P) family protein 3.1 2.0E-02AT3G21410 F-box family protein (FBW1) 3.1 2.6E-02AT2G02130 LOW-MW CYSTEINE-RICH 68 (LCR68)(PDF2.3) 3.0 3.5E-021Gene descriptions are abbreviated from TAIR10 genome release. Putative functions are stated in lowercase.2Genes ranked according to fold-change values derived from log2 ratios of MGR (numerator) and CSS (denominator).3 Derived from false discovery-rate correction of p-values.Hall and Ellis BMC Plant Biology 2013, 13:14 Page 6 of 17http://www.biomedcentral.com/1471-2229/13/14Interestingly, another Arabidopsis GER homologue(GLP10, At3G62020) whose expression was previouslyfound to be highly correlated with secondary cell wall-associated CESAs (CesA4, 7 and 8) in regression analysisof public microarray datasets [13], also displayed elevatedexpression at both the CSS and OLD stages in our study(Additional files 2, 3 and 4: Table S1).Also more highly expressed at this earlier stage ofcell wall maturation are two pectate lyases (polyga-lacturonases), At3g07010 and At3g15720, previouslyassociated with cell separation [53], and ALPHA-XYLOSIDASE 1/AXY3 (At1g68560), an exoglycosylasethat acts specifically on non-fucosylated xyloglucans [54]and is essential for apoplastic xyloglucan modification[55]. Several other up-regulated genes are less clearlylinked to cell wall processes, but the functions oftheir encoded proteins may be related to the over-representation of ‘turgor pressure’ in the GO termenrichment analysis for this gene set (Additional file 7:Table S4).The list of genes most highly expressed in OLD tissuesrelative to CSS tissues (Table 6) is particularly striking: sixof the eight most highly up-regulated genes encode PLANTDEFENSIN (PDF) proteins, small cysteine-rich peptideshomologous to anti-microbial peptides that are widelydistributed within the eukaryotes [56]. Since both CSS andOLD tissues were harvested only seconds apart, anartifactual pattern of wounding-induced gene induction isnot likely. Instead, it appears that accumulation of theproducts of such classical “defense” genes may form anintegral part of the normal maturation of the inflorescencestem, perhaps reflecting a commitment to protection ofthese tissues until fertilization and seed dispersal aresuccessfully completed.Relatively few cell wall-specific genes appear in the‘higher in OLD than in CSS’ short list, with the exception ofEXTENSIN 3/RSH and another proline-rich extensin-likefamily protein. EXT3/RSH plays an essential role in cellwall deposition through formation of EXTENSIN proteinscaffolds that cross-link other cell wall constituents, therebycontributing to cell wall rigidification [57,58]. The mostup-regulated of all the genes at the OLD stage relative tothe CSS stage is the chloroplast-localized FATTY ACIDREDUCTASE 6 (FAR6). A similar pattern of elevated FAR6expression was earlier observed in microarray analysis ofepidermal peels from the stem base [59] as well as in stemsections harvested from the base of mature ArabidopsisCol-0 plants [6]. Accompanying FAR6 in this list of mosthighly expressed genes is a wax synthase homologue(At5g22490), a co-occurrence pattern consistent withTable 4 Twenty most differentially expressed genes with higher expression in CSS stage relative to MGR stageAccession Gene annotation1 MGR/CSS Fold-change2 q-value3AT5G25110 CBL-INTERACTING PROTEIN KINASE 25 (CIPK25)(SnRK3.25) −5.1 1.8E-02AT2G43050 pectin methylesterase −5.5 2.8E-03AT4G30290 XYLOGLUCAN ENDOTRANSGLUCOSYLASE (XTH19) −5.5 2.8E-04AT5G59290 UDP-glucuronic acid decarboxylase 3 (UXS3) −5.6 5.8E-04AT2G38080 LACCASE 4 (IRX12) −6.0 4.1E-04AT2G37090 IRREGULAR XYLEM 9 (IRX9) −6.1 4.3E-04AT5G46340 REDUCED WALL ACETYLATION 1 (RWA1) −6.1 3.0E-03AT1G03740 S/T protein kinase −6.1 7.8E-06AT5G01360 TRICHOME BIREFRINGENCE-LIKE 3 (TBL3) −6.9 2.3E-03AT2G28315 Nucleotide/sugar transporter family protein −7.3 4.0E-03AT1G22480 plastocyanin-like domain-containing protein −7.8 3.6E-04AT5G17420 CESA7(IRX3) −8.8 1.6E-04AT3G18660 glucuronic acid substitution of xylan1 (GUX1) −9.0 5.7E-04AT4G18780 CESA8 (IRX1) −9.5 9.4E-04AT3G16920 CHITINASE-LIKE PROTEIN 2 (CTL2) −10.1 1.3E-05AT2G28110 FRAGILE FIBER 8 (FRA8) −10.3 4.8E-04AT2G03200 aspartyl protease family protein −10.4 9.6E-05AT1G63910 AtMYB103 −11.5 2.3E-03AT5G44030 CESA4 (IRX5) −11.5 5.6E-05AT2G45220 pectin methylesterase −38.4 1.8E-051Gene descriptions are abbreviated from TAIR10 genome release. Putative functions are stated in lowercase.2Genes ranked according to fold-change values derived from log2 ratios of MGR (numerator) and CSS (denominator).3 Derived from false discovery-rate correction of p-values.Hall and Ellis BMC Plant Biology 2013, 13:14 Page 7 of 17http://www.biomedcentral.com/1471-2229/13/14epidermal cells in fully mature stems actively synthesizingboth their cuticle polyester network and the associated waxmatrix. The modest representation of explicitly cell wall-associated genes in this CSS-to-OLD transition list impliesthat the CSS and OLD stage tissues share quite similar tran-scriptional profiles in terms of secondary cell wall formationprocesses, and that the metabolic commitment to cell wallfortification in stem tissues does not change dramaticallyafter cessation of active elongation.Stage-specific, whole-genome co-expression analysisWhile differential gene expression datasets contrastingdiscrete growth stages provide initial insights into thebiology underlying specific developmental transitions,potential functional relationships between gene productscan also be revealed by considering transcript abundancesacross all the sampled developmental stages. The under-lying rationale is that genes co-expressed at one stage andexhibiting similar association patterns across a broaderdevelopmental range may represent a subset of genesinvolved in specific biological processes.The ‘mixed effects model’ approach used in this studyallowed us to generate developmental stage 'estimates' fromtwo-channel arrays, which can be expressed as mean fold-change values (biological replicates = 6) of transcriptabundance at one stage relative to a hypothetical meanvalue of zero across the entire experiment. It should benoted that these 'estimates' can be computed with the samestatistical power as applies to the log2 differential expressionratios reported in Tables 1, 2, 3, 4, 5, 6. This treatmentprovides a more intuitive means of visualizing gene expres-sion trajectories, and provides the basis for formalco-expression analysis. The genes associated with each co-expression set (cluster) are identified in a filterable columnwithin the full-genome dataset (Additional files 2, 3 and 4:Table S1), and their AGI codes are also listed separately inAdditional file 8: Table S5 for easier access.Hierarchical divisive clustering was performed on the4635 genes whose means were most significantly differentfrom the gene-wise mean of all stages (q-value <0.05)(Figure 2), thereby filtering out the great majority (22 294)of genes whose expression displayed little perceptiblechange during the course of stem elongation. The 4635genes fall within eight major co-expression clusters thatexhibit distinct developmental trajectories. With the excep-tion of cluster 2 (Figure 2), which exhibits elevated expres-sion at only the YNG and OLD stages, genes within themajor clusters exhibit a single peak in transcript abundanceassociated with a discrete developmental stage, accompan-ied by lower expression in all the sets before and/or afterTable 5 Twenty most differentially expressed genes with higher expression in CSS stage relative to OLD stageAccession Gene annotation1 CSS/OLD Fold-change2 q-value3AT1G12845 unknown protein 10.1 4.1E-02AT5G20630 GERMIN 3 (GER3) 6.5 1.6E-02AT3G07010 pectate lyase family protein 6.3 7.8E-03AT1G72610 GERMIN-LIKE PROTEIN 1 (GER1) 6.2 4.0E-02AT1G64660 ARABIDOPSIS THALIANA METHIONINE GAMMA-LYASE (ATMGL) 5.7 9.6E-03AT3G15720 glycoside hydrolase family 28 protein / polygalacturonase (pectinase) family protein 5.3 3.5E-02AT1G80280 hydrolase, alpha/beta-fold family protein 5.3 3.4E-02AT1G68600 unknown protein 5.0 4.3E-02AT5G38430 ribulose bisphosphate carboxylase small chain 1B / RuBisCO small subunit 1B (RBCS-1B) (ATS1B) 4.8 4.3E-02AT2G39010 PLASMA MEMBRANE INTRINSIC PROTEIN 2E (PIP2E) 4.8 2.2E-02AT2G38540 LIPID TRANSFER PROTEIN 1 (LTP1) 4.7 2.8E-02AT3G16240 DELTA TONOPLAST INTEGRAL PROTEIN (DELTA-TIP) 4.7 2.2E-02AT4G03205 coproporphyrinogen oxidase activity in porphyrin biosynthetic process within chloroplast 4.4 3.9E-02AT3G48970 copper-binding family protein in metal ion transport 4.3 4.9E-02AT1G75900 family II extracellular lipase 3 (EXL3), carboxylesterase activity, acyltransferase activity 4.3 1.5E-02AT2G05790 glycosyl hydrolase family 17 protein 4.2 7.8E-03AT5G38420 ribulose bisphosphate carboxylase small chain 2B / RuBisCO small subunit 2B (RBCS-2B) (ATS2B) 4.2 4.1E-02AT1G68560 ALPHA-XYLOSIDASE 1 (XYL1) 4.2 4.1E-02AT5G22580 unknown protein 4.2 2.8E-02AT3G12610 DNA-DAMAGE REPAIR/TOLERATION 100 (DRT100) 4.1 2.2E-021Gene descriptions are abbreviated from TAIR10 genome release. Putative functions are stated in lowercase.2Genes ranked according to fold-change values derived from log2 ratios of CSS (numerator) and OLD (denominator).3 Derived from false discovery-rate correction of p-values.Hall and Ellis BMC Plant Biology 2013, 13:14 Page 8 of 17http://www.biomedcentral.com/1471-2229/13/14this peak. Each co-expression set (cluster) thus appearsto have a uniquely defined developmental window inwhich the associated genes act, and the clusters areengaged sequentially during elongation and maturationof the inflorescence stem. It is interesting that whilemost clusters (1–7) contain at least 400+ genes, Cluster8 (out-group of Clusters 5–7), which contains genesthat are up-regulated only at the base of the stem(OLD), is limited to 16 genes.To test for functional relatedness of the genes popu-lating these clusters, we examined gene ontology en-richment within several sub-clusters that exhibitedmarked patterns of coordinated up- or down-regulationspecific to single developmental stages (Figure 2; sub-clusters 1.1, 2.1, 5.1, 5.2, and Cluster 8). Although box-plotting of the mean 'estimates' for these sub-clustersclearly demonstrates the extent to which these co-expression sets are synchronously up- or down-regulatedwith respect to the other stages (Additional file 9: FigureS2), gene ontology analysis revealed only modest GO termenrichment within these subsets of co-expressed genes(Additional file 10: Figure S3, Additional file 11: Table S6),indicating that, despite their shared expression pattern,the genes in each sub-cluster do not display obvious func-tional relatedness.DiscussionWhile gene expression profiling has been applied previouslyto the expanding inflorescence stem in Arabidopsis, thosestudies have all suffered from various limitations thatimpede our ability to accurately align the resulting expres-sion profiles with the developmental state of the tissuebeing sampled. To address this issue, we have made use ofgrowth kinematic profiling (GKP) to establish, for eachplant being sampled, the precise state of growth extensionthat each stem section represents. Collectively, thesesections span the growth and maturation states of the stem,and they also represent a cell wall development continuum.Thus, GKP-guided pooling of sections associated withdiscrete zones along that continuum (e.g. the pointat which extension growth ceases) makes it possible togenerate tissue samples whose gene expression profiles canbe confidently aligned with specific developmental states.While the growing stem is a complex organ consisting ofmultiple tissues, the common denominator across all ofthese cell types is the coordinated and initially rapid aniso-tropic expansion of their cell walls along the axis of growth.This expansion ultimately comes to a halt as the walls ofsome tissues, most notably the vascular tissues andsupporting fibres, are reinforced with non-extensiblesecondary wall layers. Thus, while the biological processesTable 6 Twenty most differentially expressed genes with higher expression in OLD stage relative to CSS stageAccession Gene annotation1 CSS/OLD Fold-change2 q-value3AT1G21310 EXTENSIN 3 (ATEXT3) −3.9 4.8E-02AT5G09840 unknown protein −4.1 3.7E-02AT3G54580 proline-rich extensin-like family protein, structural constituent of cell wall −5.0 4.8E-02AT5G54230 MYB DOMAIN PROTEIN 49 (MYB49) −5.0 4.1E-02AT2G28780 unknown protein −5.0 2.9E-02AT1G70830 Bet v I allergen family −5.2 3.9E-02AT2G02930 GLUTATHIONE S-TRANSFERASE F3 (ATGSTF3) −5.4 3.0E-02AT4G15390 acyl-transferase family protein −6.1 1.6E-02AT4G08780 peroxidase, putative −8.2 4.5E-02AT1G02930 GLUTATHIONE S-TRANSFERASE 6 (GSTF6) −8.6 4.3E-02AT2G36120 DEFECTIVELY ORGANIZED TRIBUTARIES 1 (DOT1) −8.8 3.5E-02AT1G19530 unknown protein −9.7 3.1E-02AT1G75830 LOW-MOLECULAR-WEIGHT CYSTEINE-RICH 67 (LCR67) −9.9 1.5E-02AT2G26020 PLANT DEFENSIN 1.2B (PDF1.2b) −10.7 4.3E-02AT5G22490 Wax ester synthase homologue −11.5 4.5E-02AT5G44420 PLANT DEFENSIN 1.2 (PDF1.2) −13.1 3.9E-02AT2G26010 PLANT DEFENSIN 1.3 (PDF1.3) −16.2 4.4E-02AT4G16260 O-glycosyl hydrolase −17.9 4.9E-02AT5G44430 PLANT DEFENSIN 1.2C (PDF1.2c) −19.9 3.6E-02AT3G56700 FATTY ACID REDUCTASE 6 (FAR6) −55.4 7.8E-031Gene descriptions are abbreviated from TAIR10 genome release. Putative functions are stated in lowercase.2Genes ranked according to fold-change values derived from log2 ratios of CSS (numerator) and OLD (denominator).3 Derived from false discovery-rate correction of p-values.Hall and Ellis BMC Plant Biology 2013, 13:14 Page 9 of 17http://www.biomedcentral.com/1471-2229/13/14being probed in these samples are not restricted to cellwall expansion/modification, the latter processes can beexpected to dominate the broader landscape of transcrip-tional changes that accompany the maturation of the stemand its component tissues.‘Young’ stage tissue displays a complex transcriptionalprofileThe YNG stage sampled in this study captures the top1 cm of the Arabidopsis stem and so encompasses a devel-opmentally complex region containing the shoot apicalmeristem, and up to twenty short internodes which boreflowers and/or siliques prior to harvest. Since multipletissue development trajectories are being initiated withinYNG samples, it is not surprising that cell wall-forming/modifying processes do not dominate the gene expressionprofile of YNG stage tissue in comparison with the MGRstage. It is interesting that the up-regulated MEE48 endo-β(1→3)-glucosidase (Figure 2) was previously characterizedas an anther-specific gene whose proposed functioninvolved callose degradation during pollen exine formation[23]. Since all floral tissue had been deliberately removedfrom the YNG tissue at the time of sampling, MEE48 mustplay additional roles in development. The prominence ofMEE48 expression in the YNG transcriptome may berelated to the importance of callose hydrolysis for the devel-opment of new cell plate structures during cytokinesis [60],a process that is actively underway in the apical meristem.Rapid tissue extension is associated with a uniquetranscriptional signatureBoth the YNG/MGR comparison and MGR/CSS compari-son gene lists provide a perspective on the genes up-regulated in cells undergoing extension growth at theirmaximum rate (MGR). Genes associated with gibberellicacid (GA)-mediated elongation are prominently expressedat the MGR stage, consistent with the known role ofgibberellic acid as an effector of directional cell growth [61].The redox-associated cysteine-rich signal peptide, GA-STIMULATED ARABIDOPSIS 4 (GASA4) [62], is 3.4-foldup-regulated in MGR relative to CSS (Table 3), while theGA-responsive transcription factor, MINI ZINC FINGER 1(MIF1) [33], is up-regulated 12-fold in the MGR stagerelative to the YNG stage. Loss-of-function at theMIF1 locus results in unresponsiveness to GA and inflore-scence stem dwarfism [33]. MERISTEM-5 (MERI5B,XTH24; At4g30270) encodes a Group 2 xyloglucanendotransglycosylase/hydrolase [63], a group of proteinsthat facilitate the remodelling of hemicellulose to allowcellulose microfibril separation and ‘creep’ during aniso-tropic cell wall expansion [1]. MERI5B expression is alsoelevated in MGR relative to YNG tissues, and clusters withMIF1 across all the developmental stages studied (Figure 2).In addition to displaying co-expression, MIF1 and MERI5Bco-cluster with the arabinogalactan protein, AGP13, andthe MYB61 transcription factor in a set that exhibits peakexpression somewhat later, at the onset of cessation(Figure 2, cluster 7).In contrast to XTH24, expression of another XTH,ENDOXYLOGLUCAN TRANSFERASE A1 (EXGT-A1),required for normal cell wall expansion [64] is restricted tothe MGR stage (Figure 2, cluster 5), where it clusters withanother member of the AGP family, AGP12, whose expres-sion is >6-fold higher in the MGR stage relative to theYNG stage. An additional five AGPs (AGP14, 21, 22, 24and FLA13) were found to be significantly up-regulated(q-value<0.05) in the MGR stage relative to YNG stage(Additional files 2, 3 and 4: Table S1), suggesting that asuite of AGPs may be contributing to the uniquestructural dynamics of rapidly expanding cells at theMGR stage.‘Cessation’ stage gene expression is dominated bysecondary cell wall processesThe composition of the shortlist of twenty genes most up-regulated at the CSS stage relative to the MGR stage(Table 4) is particularly striking, since thirteen appear to befunctionally related to secondary cell wall biosynthesis. It isalso notable that the population of this list is completelydistinct from those genes whose expression is dominant inCSS tissues relative to OLD tissues, suggesting that theseare genes whose expression becomes elevated as cell expan-sion slows and then remains elevated through ensuing stemmaturation.Central among these cell wall-associated genes are thethree cellulose synthases that are essential for secondarycell wall formation [13,35]. CESA8 is thought to belong tothe same multi-protein biosynthetic complex as CESA4and CESA7 [65], which have similar contributions tosecondary cell wall synthesis [35], although their relativeproportions in the CESA complex remain unknown. In ourco-expression analysis, CESA8 clusters differently fromCESA4 and CESA7, primarily due to increasingly elevatedexpression of CESA8 in the OLD tissue sample (Figure 2,cluster 8). In contrast, the expression of CESA4 and CESA7(Figure 2, cluster 6) does not change significantly from theCSS to OLD stages. Since the relative stability and turnoverrates for the three CESA proteins are unknown, thesedifferences in gene expression do not necessarily conflictwith the predicted abundance of their cognate proteins inthe plasma membrane. It is also possible that the relativeproportions of secondary cell wall-associated CESAs withincellulose synthase complexes do not remain fixed through-out the period of secondary cell wall formation.By definition, only primary cell walls are capable ofexpanding [66], and the great majority of this expansionwould be occurring above the point of cessation beingsampled in this study. We therefore anticipated thatHall and Ellis BMC Plant Biology 2013, 13:14 Page 10 of 17http://www.biomedcentral.com/1471-2229/13/14cellulose synthase genes associated with primary cell wallformation (notably, CESA1, 3 and 6) would be up-regulatedin the YNG and MGR stages relative to their expression inthe CSS and OLD stages. Instead, CESA1 and 6 are not sig-nificantly differentially expressed between pre- and post-cessation stages, and CESA3 is actually more highlyexpressed in the CSS and OLD stages than in the YNG andMGR stages (Figure 2, Additional files 2, 3 and 4: Table S1).This is not an isolated example of the behavior of theCESA3 gene; Ko and Han (2004) [6] had earlier observedelevated CESA3 expression in the base of fully matureArabidopsis Col-0 stems (>25 cm in height) relative to thebases of less mature stems (5 and 10 cm in height), whileCESA1 and CESA6 expression declined with stem matur-ation. Ehlting et al. (2005) [4] also detected higher levels ofCESA3 expression at the mid-point of 10 cm stems ofLandsberg erecta (Ler) plants than in the top 3 cm stemregion of those plants. Another study found that, whileAtCESA4, 7 and 8 were up-regulated in mature Col-0stems, none of the canonical ‘primary cell wall’ CESA genes(CESA1, 3, 6) were present in the list of CESAs significantlyup-regulated in actively elongating tissues [5], likely due tothe persistent expression of CESA1, 3 and 6 at later, non-elongating stages as well. In the present study, CESA6,which is generally considered to be important for cellulosedeposition in primary cell walls [13], is most highly up-regulated in the MGR and CSS stage tissues (Figure 2,Cluster 5), where it is co-expressed with CESA2 (previouslyassociated with radial cell wall reinforcement [65]) and withCESA10.Collectively, these data suggest that deployment ofparticular cellulose synthases in plant cells does not followCesA3KNAT7, ATPMEPCRDCOMTCesA7MIF1, MER15BPE, AGP13, MYB61CesA4, FLA12, IRX8,9 FLA11, IRX7/FRA8CesA6CesA8GATL-1FAR6AGP2EXT3, HRGP1CesA10MLP423PER64DHS3PAL1 PAL4, IAA12, ARF4DHS2ARF8MEE48EF-1-αARF12CesA2EXGT-A1AT1G12845 AGP12ARF1811.15.15.282345672.1Fold-change−4 −2 0 2 4YNG MGR CSS OLDFigure 2 Hierarchical clustering of 4635 differentially expressed genes (q-value<0.05). On the basis of relative expression between cell wallexpansion stages as outlined in Figure 1; top 1 cm of plant (YNG), maximum growth-rate (MGR), cessation of elongation (CSS) and base ofprimary stem at rosette (OLD). Inset; 11-level colourimetric fold-change scale. Clusters (1–8) and stage-specific sub-clusters (1.1,2.1,5.1,5.2) arenumbered for subsequent examination. The positions of representative genes associated with cell wall processes have been indicated along theright margin (described in ‘Results’ and/or ‘Discussion’). PE=pectin esterase, At2g45220.Hall and Ellis BMC Plant Biology 2013, 13:14 Page 11 of 17http://www.biomedcentral.com/1471-2229/13/14a pattern of exclusive association with either actively elong-ating tissues (i.e. with primary cell wall synthesis) orpost-elongation tissues (i.e. secondary cell wall synthesis).Instead, a more diverse co-occurring set of cell wall-forming/modifying processes may recruit distinct sub-setsof CESA and CESA-LIKE family members for specificdevelopmental programming (e.g. intrusive growth ofinterfascicular fibres).Another strong indication that the CSS tissue sample ac-curately captures the transition from primary to secondarycell wall formation is the presence in the ‘higher-in-CSS’gene lists of a suite of genes specifically associated withaccumulation of glucuronylarabinoxylans (‘xylans’), includ-ing XYLOSE SYNTHASE 3, FRA8, IRX9, RWA1, XTH19and GUX1 [67] (Figure 2, cluster 7). Other xylan-relatedgenes (XYLOSE SYNTHASE 6; XTR4; BXL1; BXL2; EXGT-A1; XTH18) are also up-regulated in the CSS tissue(Additional files 2, 3 and 4: Table S1), but failed to qualifyfor the “top twenty” short-list of genes more highlyexpressed in CSS than in OLD.A possible positive regulator of secondary cell wall devel-opment, the MYB61 transcription factor gene, is alsostrongly up-regulated in the CSS stage (Figures 2, S7).MYB61 has been proposed to promote cell wall lignifica-tion [31], and more specifically to regulate three cellwall–associated genes encoding the KNAT7 transcriptionfactor, the lignin biosynthesis enzyme CAFFEOYL-COAO-METHYLTRANSFERASE 7, and a pectin methylesterase(At2g45220) [29]. While our expression data confirm anassociation of MYB61 with secondary cell wall formation,such involvement is likely conditioned by other factorssince MYB61 activity has also been associated with a widerange of biological processes in plants, including seedcoat mucilage production [32], stomatal closure [30], andpleiotropic control of photosynthate partitioning [29].Arabinogalactan proteins (AGP) form a largelyuncharacterized class of proteoglycans that likely play struc-tural and/or signaling roles in cell wall development.Indeed, a number of AGP family members exhibit signifi-cant modulation of expression at the onset of secondarycell wall formation (Additional files 2, 3 and 4: Table S1).For example, expression of AGP18, a member of a lysine-rich, GPI-anchored sub-family that includes AGP17 andAGP19, is down-regulated at the OLD stage relative to theMGR stage, and the loss-of-function agp18 mutantpossesses shortened inflorescence stems [68], indicating apossible role for AGP18 in promoting cell wall expansion.AGP12 also shows a significant drop in expression coinci-dent with the onset of cessation (Additional files 2, 3 and 4:Tables S1), consistent with a similar functional association.FASCICLIN-LIKE 8 (FLA8), on other hand, is up-regulated at the OLD stage relative to the MGR stage(Additional files 2, 3 and 4: Table S1). FLA8/AGP8 belongsto sub-family of AGPs that contain a fasciclin domain andoften possess a glycosyl phosphatidyl inositol (GPI) anchor[69]. A poplar homologue of AtFLA8 was observed to beup-regulated significantly in tension wood, but not inopposite wood, in poplar stems, when compared to its ex-pression in differentiating xylem [70]. AGP21 also appearssignificantly up-regulated in the OLD stage relative to theMGR stage (Additional files 2, 3 and 4: Table S1). Interest-ingly, AGP21, similar in sequence to AGP12 and AGP14[71], is down-regulated ~4-fold upon silencing of thetranscription factor PRODUCTION OF ANTHOCYANINPIGMENT 1 (PAP1/MYB75), coincident with increased cellwall thickness in xylary and interfascicular fibres [72]. Themembers of the large AGP family thus appear to have func-tionally diverged, as revealed in part through differences inspatiotemporal regulation of their expression [63].Protection and fortification are hallmarks of OLD stemtissueThe base of the stem of 10-15 cm Columbia plants (OLDtissue samples) contains highly contrasting tissues, includ-ing live, photosynthetically active cells located adjacent tothick-walled, highly lignified fibres of the interfascicularregion that are presumably in the advanced stages ofprogrammed cell death.In general, however, genes related to secondary cell wallsynthesis are most active in this region of the lower stem(Figure 2; clusters 2, 6, 7 and 8). Interestingly, CESA8,which appears in cluster 8, exhibits its highest levelof expression at this stage, as does XYLOGLUCANENDOTRANSGLYCOSYLASE/HYDROLASE 18 (XTH18).This expression data is consistent with other results linkingxyloglucan deposition with late stages of secondary cell wallsynthesis. For instance, incorporation of xyloglucan hasbeen observed to continue in cotton fibres after cessationof wall extension [73], and PttXET16 activity was associatedwith secondary vasculature of poplar [74].In the final stages of fibre secondary cell wall maturation,the polysaccharide matrix is typically impregnated with thephenylpropanoid polymer, lignin (reviewed in [75]). Severalgenes whose products are associated with the shikimic acidand phenylpropanoid pathways, and lignification exhibitcorresponding expression patterns within this dataset,although differences in their clustering suggest subtledistinctions in the timing of their expression (Figure 2). Forinstance, PAL1, a member of the PHENYLALANINEAMMONIA- LYASE (PAL) gene family whose activity isrequired for phenylalanine allocation to phenylpropanoidmetabolism, appears highly expressed in both the CSS andOLD stages, coincident with up-regulation of 3-DEOXY-D-ARABINO-HEPTULOSONATE 7-PHOSPHATE (DAHP)SYNTHASE 3 (DHS3), which regulates the intake of carboninto the shikimate pathway. Maximum expression of PAL4,on the other hand, occurs in the OLD stage, suggestingHall and Ellis BMC Plant Biology 2013, 13:14 Page 12 of 17http://www.biomedcentral.com/1471-2229/13/14that different PAL family members may be playingdistinct roles.ConclusionThe large plant-to-plant variation in stem growth kinematicprofiles that we identified earlier [15] makes it clear thatearlier studies in which stems from multiple plants havebeen pooled to create biological replicate samples are inev-itably compromised in their ability to accurately place cellu-lar changes in an elongative development context. Bycontrast, the concordance of our GKP-guided gene expres-sion data sets with current knowledge of cell wall biologyprovides strong evidence of the ability of this approach tocapture development stage-specific information. At thesame time, those known players in our data sets are accom-panied by numerous genes of currently unknown biologicalfunction, which makes them high priority candidates forfurther research into the processes underpinning both plantcell expansion and deposition of the cellulose-rich cell wallsthat comprise plant biomass.MethodsPlant growth, growth kinematic profiling and samplingPlant growth and imaging was conducted as described inHall & Ellis (2012) [15], using applied paper tags assynthetic optical markers for growth kinematic profiling.Tagged and imaged plants were harvested sequentiallybetween 1 and 3 pm (mid-day where daylight cycle occursbetween 6 am and 10 pm on a 16hL:8hD regime) at20-minute intervals. Stem segments (~1 cm) were immedi-ately snap-frozen in liquid nitrogen and deposited into0.2 mL PCR tubes for −80°C storage. Segments were subse-quently pooled on the basis of growth kinematic profilingdata (shown in Additional file 1: Figure S1) and experimen-tal design objectives, as outlined in Figure 1.RNA processingWhole stem segments (pooled according to growthkinematic profile equivalence) were homogenized in liquidnitrogen with a pre-cooled mortar and pestle. The frozenpowder was then transferred to 1.5 mL microcentrifugetubes, weighed and combined with TRIzol™ reagent(cat#15596-026, Invitrogen)(1 mL TRIzol per 100 mgtissue), vortexed and incubated at room temperature for5 minutes. Chloroform (0.2 mL for each 1 mL of TRIzol)was added, vortexed for 15 seconds, incubated for1 minute at room temperature, and centrifuged at 15000 gfor 10 minutes at 4°C. The aqueous phase was transferredto fresh RNAse-free tubes and then combined with anequal volume of isopropanol and incubated 20 minutes onice. RNAtotal was pelleted by centrifugation at 15000 g for10 minutes, and pellets were washed with 1 mL 75%ethanol in RNAse-free water. Following a 5-minute pelletdrying phase, pellets were resuspended in 25 μL RNAse-free water and incubated on ice for 1 hour. Each resuspen-sion was treated with 1/10th volume (~2 μL) 10X DNAseI buffer and 1 μL 10X DNAse I (from RNAqueousWMicro kit; cat#AM1931, Ambion) for 20 minutes at37°C followed by addition of 2 μL DNAse inactivationreagent (also from RNAqueousWMicro kit) and incubatedat room temperature for 2 minutes. Samples werecentrifuged at 13000 g for 1.5 minutes and thesupernatant transferred to RNAse-free tubes and storedat −80°C.Reverse transcription and labellingFor each biological replicate, approximately 20 μg RNAtotalwas incubated with 2 ug oligo(dT) primer (cat#18418-012,0.5 ug/ul, Invitrogen) in a 22.5 μL volume of RNA-primermix and denatured at 70°C for 10 minutes. Reaction mixwas prepared such that each sample contained 9 μL 5XFirst Strand buffer (supplied with Superscript II, Invitrogen,cat#18064-014), 0.23 μL each of 0.1 mM dATP (cat#10216-108, Invitrogen), dCTP (cat#10217-016, Invitrogen), anddGTP (cat#10218-014,Invitrogen), as well as 0.045 μLdTTP (cat#10219-012, Invitrogen) for a total reaction mixvolume of 18.5 μL. This reaction mix was combined with18.5 μL RNA-primer mix along with 1.5 μL (1.5 moles/μL)of the appropriate Cyanine dye; Cy5-dUTP (cat#45-000-740, Fisher) or Cy3-dUTP (cat#45-000-738, Fisher). Afterincubation at 42°C for 2 minutes, 1 uL 40U/uL RNAaseInhibitor (cat#10777-019, 40U/ul, Invitrogen) and 1.2 ul200 U/μL Superscript II (cat#18064-014,200 U/ul,Invitrogen) were added for a final volume of 45 μL whichwas incubated at 42°C for 2.5 hours, then deactivated with0.5 M NaH/50 mM EDTA at 65°C for 15 minutes. Thereaction was neutralized with 7.1 uL 1 M Tris–HCl(pH7.5). Samples were cleaned of unlabeled probe viacentrifugal filtration using Amicon 0.5-Ultra 30 kDa filters(cat#UFC503096, Millipore) prior to array hybridization.Array hybridizationFor transcript profiling, we employed custom two-channelmicroarrays spotted with 26 929 70-mer oligonucleotidesoriginally synthesized on the basis of ‘The ArabidopsisInformation Resource’ (TAIR) ‘5’ release of the Arabidopsisgenome (www.Arabidopsis.org), with gene annotationssubsequently updated to the current genome release(TAIR10) [76]. The microarray slides were first pre-conditioned by incubating them in Coplin jars with 50°C2X SSC for 20 minutes, followed by room temperaturewashes with 0.2X SSC and ddH2O for 5 and 3 minutes,respectively using an Advawash AV400 machine(Advalytix/Beckman-Coulter). Pre-hybridization solutionof 1X formamide-based hybridization buffer (pre-warmedto 80°C) from Vial 7 of the 3DNA Array 350 kit(cat#W300130, Genisphere) was then added to the gapHall and Ellis BMC Plant Biology 2013, 13:14 Page 13 of 17http://www.biomedcentral.com/1471-2229/13/14between each slide and a pre-placed m-Series lifterslip(cat#48382-251, VWR) within the Slidebooster (Advalytix/Beckman-Coulter) hybridization chamber and subse-quently incubated for 1–1.5 hours at 50°C with sonication(power=15, pulse=1 second 'on', 9 seconds 'off'). Slideswere then washed in 2X SSC (0.2% SDS) for 15 minutes at65°C followed by room temperature washes in 2X SSCand 0.2X SSC for 10 minutes each, and centrifuged at700 rpm until dry in Advatubes (cat# OAX05216,Advalytix/Beckman-Coulter). Equal volumes of labeledCy3 and Cy5 mixes (12.5 μL each) were combined with25 μL 2X formamide buffer (Vial 7, 3DNA Array350 kit)and the 50 μL hybridization mix added to the gap betweenthe 42°C pre-warmed slides and the pre-placed m-Serieslifter-slips. Slides were then incubated at 42°C for 16–18 hours with sonication (power=15, pulse=1 second 'on',9 seconds 'off'). Post-hybridization washing was carriedout in reduced lighting with 42°C 2X SSC (0.2% SDS) for15 minutes followed by room temperature washes with2X SSC and 0.2X SSC for 15 minutes each, thencentrifuged until dry at 700 rpm. Slides were stored ina light-proof desiccating chamber until fluorescencescanning.Microarray scanning and spot quantificationHybridized arrays were scanned with a ScanArray ExpressHT (Perkin-Elmer) scanner and associated software, using543 nm laser irradiation for Cy3, and 633 m laser for Cy5fluorescence. Laser power was adjusted for each slide indi-vidually within the range of 95-100% such that ~1-2% ofspotted probes (presumed positive controls) yieldedsaturated signals. PMT gain ranged from 60-95%, set foreach slide such that fluorescence intensity of sub-gridregions surrounding spots did not exceed 400 (16-bit gray-scale). TIFF images of array scans were imported intoImagene (Biodiscovery Software) and grid templates wereroughly placed before applying the 'auto-adjust' function tobest fit the subgrids on a per-spot basis, allowing spot sizevariation from 15-20 μm. Median pixel intensitiescomputed from spot regions were used to represent spotintensity in subsequent analyses.Microarray data analysisData analysis was carried out in the statistical programmingenvironment R (cran.r-project.org/) using custom scriptsand contributed packages. To remove local backgroundnoise, the mean signal intensity of the dimmest five percentof spots within each of 48 subgrids was subtracted fromeach array element using a custom script, then variancestabilization normalization (VSN) was applied to eachchannel to normalize for non-linearity in variance acrossspot intensities [77] using the function ‘vsn’ (‘vsn’ package,Bioconductor). Normalized intensities were then fit to themixed effects model [16] using the ‘lme’ function (‘nlme’package), and all pairwise differential expressions for arrayelements were computed as the log2 intensity differencevalues between treatment class intensities. Associatedmeasures of significance (p-values relative to null hypoth-esis, log2 difference equals zero) were corrected for false-discovery rate using a custom script based upon standardq-value calculation [78], and ‘estimates’ were computed asthe log2 intensity difference of each treatment class to themean of all treatment classes (normalized to zero).Associated measures of significance (p-values relative tonull hypothesis of log2 difference = 0) were also correctedfor false-discovery rate as described above. Raw and outputdata were exported along with TAIR10 annotations in thesupplemental data (Additional files 2, 3 and 4: Table S1).For hierarchical clustering, dissimilarity matrices werecomputed from filtered datasets using the ‘diana’ function(‘cluster’ package) and rendered as dendrograms using the‘dendro’ function (‘cluster’ package). Heatmaps weregenerated using the 'heatmap.2' function ('gplots' package).Availability of supporting dataGene annotations, raw expression data, statistical analysis,mean differentials, mean estimates, and gene categorizationfor the full genome are provided in Additional files 2,3 and 4: Table S1, and have been deposited withArrayExpress following the MIAME conventions [79], asaccession E-MEXP-3525.Additional filesAdditional file 1: Figure S1. Surface plots of relative elongationgrowth rates and LOWESS-predicted growth kinematic profiles (n=34).Plotting as described in Hall & Ellis (2012) except that segments arenumbered from bottom upwards for surface plots. Plants arranged bycolumn according to independently grown and observed batches.Additional file 2: Table S1 part 1. Raw, processed data, TAIR10annotations and clustering information for all gene-specific arrayelements. See first tab of file for column header descriptions.Additional file 3: Table S1 part 2. Raw, processed data, TAIR10annotations and clustering information for all gene-specific arrayelements. See first tab of file for column header descriptions.Additional file 4: Table S1 part 3. Raw, processed data, TAIR10annotations and clustering information for all gene-specific arrayelements. See first tab of file for column header descriptions.Additional file 5: Table S2. ATCOECIS reports for enrichment of geneontology (GO) terms for genes most significantly different (q-value<4.7E-02) between YNG and MGR stages. GO terms appearing more than oncein the gene list are shown ranked according to significance ofenrichment p-value (<0.05). Only GO terms with two or more genes inthe input set and showing enrichment compared to the backgroundfrequency (in the full genome) are reported (number of genes indicatedin brackets). Score indicates the fraction of input genes annotated withthe GO term. 'Term occurrences' column indicates the number of co-occurrences of each GO term in the AtCoeCis results among allexpression categories (left-most column).Additional file 6: Table S3. ATCOECIS reports for enrichment of geneontology (GO) terms for genes most significantly different (q-value<6.72E-02) between MGR and CSS stages. GO terms appearing more than oncein the gene list are shown ranked according to significance ofHall and Ellis BMC Plant Biology 2013, 13:14 Page 14 of 17http://www.biomedcentral.com/1471-2229/13/14enrichment p-value (<0.05). Only GO terms with two or more genes inthe input set and showing enrichment compared to the backgroundfrequency (in the full genome) are reported (number of genes indicatedin brackets).[move to methods; P-values are calculated using thehypergeometric distribution [17]. Score indicates the fraction of inputgenes annotated with the GO term. 'Term occurrences' column indicatesthe number of co-occurences of each GO term in the AtCoeCis resultsamong all expression categories (left-most column).Additional file 7: Table S4. ATCOECIS reports for enrichment of geneontology (GO) terms for genes most significantly different (q-value<4.9E-02) between CSS and OLD stages. GO terms appearing more than oncein the gene list are shown ranked according to significance ofenrichment p-value (<0.3). Only GO terms with two or more genes in theinput set and showing enrichment compared to the backgroundfrequency (in the full genome) are reported (number of genes indicatedin brackets). Score indicates the fraction of input genes annotated withthe GO term. 'Term occurrences' column indicates the number of co-occurences of each GO term in the AtCoeCis results among allexpression categories (left-most column).Additional file 8: Table S5. List of Arabidopsis gene index (AGI) codesfor each of the genes in the clusters specified in Figure 2.Additional file 9: Figure S2. Boxplots depicting distribution ofestimates of relative gene expression (fold-change) of eachdevelopmental stage for the sub-clusters identified in Figure 2. Boxesbound upper and lower quartiles, dark horizontal bars denote medianvalues, whiskers represent 95% confidence intervals, circles representoutliers occuring in upper and lower 2.5 percentiles. Cluster 8 is alsodepicted in Figure 2.Additional file 10: Figure S3. Gene ontology (GO) SLIM termenrichment analysis for clusters depicted in Figure 2. A) Boxplotsdepicting distribution of term enrichment across all clusters, expressed asfold-change relative to abundance in the full genome, for each of thethree GO SLIM categories; 'cellular component', 'molecular function', and'biological process'. Boxes bound upper and lower quartiles, darkhorizontal bars denote median values, whiskers represent 95%confidence intervals, circles represent outliers occuring in upper andlower 2.5 percentiles. B) Barplots exhibiting term enrichment for eachcluster in each of the three GO SLIM categories; colour assignment forbars is indicated in Figure 'A'. The number of genes (accessions) includedin each cluster is indicated at the base of the 'biological process' barplot.Additional file 11: Table S6. ATCOECIS reports for enrichment of geneontology (GO) terms for [sub] clusters of genes displaying stage-specificexpression. Top 10 most significant over-represented GO terms appearingmore than once in the gene list are shown ranked according to p-value(<0.05). Only GO terms with two or more genes in the input set andshowing enrichment compared to the background frequency (in the fullgenome) are reported (number of genes indicated in brackets). Scoreindicates the fraction of input genes annotated with the GO term. 'Termoccurrences' column indicates the number of co-occurrences of each GOterm in the AtCoeCis results among all expression categories (left-mostcolumn).Competing interestsThe authors declare no competing interests.Authors’ contributionsHardy Hall carried out the microarray experiments and data analyses. HardyHall and Brian Ellis both participated in the conception and design ofexperiments, as well as drafting the manuscript. Both authors have read andapproved the final manuscript.AcknowledgementsWe would like to acknowledge invaluable discussions with Dr. Jürgen Ehlting(University of Victoria) regarding microarray design and application, andinflorescence stem sampling challenges. We are greatly indebted to AnneHaegert and Dr. Stéphane Le Bihan at The Jack Bell Prostate Centre,Vancouver, for microarray printing and quality control. This project wouldnot have been possible without the resources of the Michael SmithLaboratories, UBC, and financial support from Genome Canada, and theNatural Sciences and Engineering Research Council of Canada. Finally, we aregrateful to Rick White (SCARL, University of British Columbia) for statisticalconsultation.Author details1Michael Smith Laboratories, University of British Columbia, Vancouver BCV6T 1Z4, Canada. 2Currently: Swedish University of Agricultural Sciences(SLU), Umeå 901 83, Sweden.Received: 20 November 2012 Accepted: 21 January 2013Published: 25 January 2013References1. Cosgrove DJ: Assembly and enlargement of the primary cell wall inplants. Annu Rev Cell Dev Biol 1997, 13:171–201.2. Szymanski DB, Cosgrove DJ: Dynamic coordination of cytoskeletal and cellwall systems during plant cell morphogenesis. Curr Biol 2009, 19:R800–11.3. Somerville CR, Bauer S, Brininstool G, Facette M, Hamann T, Milne J,Osborne E, Paredez A, Persson S, Raab T, Vorwerk S, Youngs H: Toward asystems approach to understanding plant cell walls. Science 2004,306(5705):2206–2211.4. Ehlting J, Mattheus N, Aeschliman DS, Li E, Hamberger B, Cullis IF, Zhuang J,Kaneda M, Mansfield SD, Samuels L, Ritland K, Ellis BE, Bohlmann J, DouglasCJ: Global transcript profiling of primary stems from Arabidopsis thalianaidentifies candidate genes for missing links in lignin biosynthesis andtranscriptional regulators of fiber differentiation. Plant J 2005, 42:618–640.5. Imoto K, Yokoyama R, Nishitani K: Comprehensive approach to genesinvolved in cell wall modifications in Arabidopsis thaliana. Plant Mol Biol2005, 58(2):177–192.6. Ko J, Han KH: Arabidopsis whole-transcriptome profiling defines thefeatures of coordinated regulations that occur during secondary growth.Plant Mol Biol 2004, 55(3):433–453.7. Minic Z, Jamet E, San-Clemente H, Pelletier S, Renou JP, Rihouey C, Okinyo DP,Proux C, Lerouge P, Jouanin L: Transcriptomic analysis of Arabidopsisdeveloping stems: a close-up on cell wall genes. BMC Plant Biol 2009, 9(1):6.8. Oh S, Park S, Han KH: Transcriptional regulation of secondary growth inArabidopsis thaliana. J Exp Bot 2003, 54(393):2709–2722.9. Jamet E, Roujol D, San-Clemente H, Irshad M, Soubigou-Taconnat L, Renou J,Pont-Lezica R: Cell wall biogenesis of Arabidopsis thaliana elongating cells:transcriptomics complements proteomics. BMC Genomics 2009, 10(1):505.10. Milioni D, Sado P, Stacey N, Domingo C, Roberts K, McCann M: Differentialexpression of cell-wall-related genes during the formation of trachearyelements in the Zinnia mesophyll cell system. Plant Mol Biol 2001,47(1–2):221–238.11. Demura T, Tashiro G, Horiguchi G, Kishimoto N, Kubo M, Matsuoka N,Minami A, Nagata-Hiwatashi M, Nakamura K, Okamura Y, Sassa N, Suzuki S,Yazaki J, Kikuchi S, Fukuda H: Visualization by comprehensive microarrayanalysis of gene expression programs during transdifferentiation ofmesophyll cells into xylem cells. Proc Natl Acad Sci USA 2002,99(24):15794–15799.12. Ohashi-Ito K, Oda Y, Fukuda H: Arabidopsis VASCULAR-RELATED NAC-DOMAIN6 directly regulates the genes that govern programmed celldeath and secondary wall formation during xylem differentiation.Plant Cell 2010, 22(10):3461–3473.13. Persson S, Wei H, Milne J, Page GP, Somerville CR: Identification of genesrequired for cellulose synthesis by regression analysis of publicmicroarray data sets. Proc Natl Acad Sci USA 2005, 102:8633–8638.14. Brown DM, Zeef LAH, Ellis J, Goodacre R, Turner SR: Identification of novelgenes in Arabidopsis involved in secondary cell wall formation usingexpression profiling and reverse genetics. Plant Cell 2005,17(8):2281–2295.15. Hall HC, Ellis BE: Developmentally equivalent tissue sampling based ongrowth kinematic profiling of Arabidopsis inflorescence stems. NewPhytol 2012, 194(1):287–296.16. Tempelman RJ: Assessing statistical precision, power, and robustness ofalternative experimental designs for two color microarray platformsbased on mixed effects models. Vet Immunol Immunopathol 2005,105(3–4):175–186.17. Vandepoele K, Quimbaya M, Casneuf T, De Veylder L, Van de Peer Y:Unraveling transcriptional control in Arabidopsis using cis-regulatoryelements and coexpression networks. Plant Physiol 2009, 150(2):535–546.Hall and Ellis BMC Plant Biology 2013, 13:14 Page 15 of 17http://www.biomedcentral.com/1471-2229/13/1418. Axelos M, Bardet C, Liboz T, Le Van Thai A, Curie C, Lescure B: The genefamily encoding the Arabidopsis thaliana translation elongation factorEF-1 alpha: molecular cloning, characterization and expression. Mol GenGenet 1989, 219:106–112.19. Tarutani Y, Morimoto T, Sasaki A, Yasuda M, Nakashita H, Yoshida S,Yamaguchi I, Suzuki Y: Molecular characterization of two highlyhomologous receptor-like kinase genes, RLK902 and RKL1, in Arabidopsisthaliana. Biosci Biotechnol Biochem 2004, 68:1935–1941.20. Cock JM, McCormick S: A large family of genes that share homology withCLAVATA3. Plant Physiol 2001, 126:939–942.21. Bell E, Creelman RA, Mullet JE: A chloroplast lipoxygenase is required forwound-induced jasmonic acid accumulation in Arabidopsis. Proc NatlAcad Sci 1995, 92(19):8675–8679.22. Jun J, Fiume E, Roeder AHK, Meng L, Sharma VK, Osmont KS, Baker C, HaCM, Meyerowitz EM, Feldman LJ, Fletcher JC: Comprehensive analysis ofCLE polypeptide signaling gene expression and overexpression activityin Arabidopsis. Plant Physiol 2010, 154(4):1721–1736.23. Dobritsa AA, Geanconteri A, Shrestha J, Carlson A, Kooyers N,Coerper D, Urbanczyk-Wochniak E, Bench BJ, Sumner LW, Swanson R,Preuss D: A large-scale genetic screen in Arabidopsis to identifygenes involved in pollen exine production. Plant Physiol 2011,157(2):947–970.24. Koizumi K, Yokoyama R, Nishitani K: Mechanical load inducesupregulation of transcripts for a set of genes implicated insecondary wall formation in the supporting tissue of Arabidopsisthaliana. J Plant Res 2009, 122:651–659.25. Apel K, Hirt H: Reactive oxygen species: Metabolism, oxidative stress, andsignal transduction. Annu Rev Plant Biol 2004, 55:373–399.26. Boerjan W, Ralph J, Baucher M: Lignin biosynthesis. Annu Rev Plant Biol2003, 54:519–546.27. Vanholme R, Demedts B, Morreel K, Ralph J, Boerjan W: Lignin biosynthesisand structure. Plant Physiol 2010, 153(3):895–905.28. Wenzel CL, Hester Q, Mattsson J: Identification of genes expressedin vascular tissues using NPA-induced vascular overgrowth inArabidopsis. Plant Cell Physiol 2008, 49(3):457–468.29. Romano JM, Dubos C, Prouse MB, Wilkins O, Hong H, Poole M, KangK, Li E, Douglas CJ, Western TL, Mansfield SD, Campbell MM:AtMYB61, an R2R3-MYB transcription factor, functions as apleiotropic regulator via a small gene network. New Phytol 2012,195(4):774–786.30. Liang Y, Dubos C, Dodd IC, Holroyd GH, Hetherington AM, Campbell MM:AtMYB61, an R2R3-MYB transcription factor controlling stomatalaperture in Arabidopsis thaliana. Curr Biol 2005, 15(13):1201–1206.31. Newman LJ, Perazza DE, Juda L, Campbell MM: Involvement of the R2R3-MYB, AtMYB61, in the ectopic lignification and dark-photomorphogeniccomponents of the det3 mutant phenotype. Plant J 2004, 37(2):239–250.32. Penfield S, Meissner RC, Shoue DA, Carpita NC, Bevan MW: MYB61 isrequired for mucilage deposition and extrusion in the Arabidopsis seedcoat. Plant Cell 2001, 13(12):2777–2791.33. Hu W, Ma H: Characterization of a novel putative zinc finger geneMIF1: involvement in multiple hormonal regulation of Arabidopsisdevelopment. Plant J 2006, 45(3):399–422.34. Lin P, Pomeranz MC, Jikumaru Y, Kang SG, Hah C, Fujioka S, KamiyaY, Jang J: The Arabidopsis tandem zinc finger protein AtTZF1affects ABA- and GA-mediated growth, stress and gene expressionresponses. Plant J 2011, 65(2):253–268.35. Gardiner JC, Taylor NG, Turner SR: Control of cellulose synthase complexlocalization in developing xylem. Plant Cell 2003, 15(8):1740–1748.36. Hamann T, Osborne E, Young HL, Misson J, Nussaum L, SomervilleCR: Global expression analysis of CESA and CSL genes inArabidopsis. Cellulose 2004, 11:279–286.37. Holland N, Holland D, Helentjaris T, Dhugga KS, Xoconostle-Cazares B,Delmer DP: A comparative analysis of the plant cellulose synthase (CesA)gene family. Plant Physiol 2000, 123(4):1313–1324.38. Hossain MA, Noh H, Kim K, Koh E, Wi S, Bae H, Lee H, Hong S: Mutation ofthe chitinase-like protein-encoding AtCTL2 gene enhances ligninaccumulation in dark-grown Arabidopsis seedlings. J Plant Physiol 2010,167(8):650–658.39. Sánchez-Rodríguez C, Bauer S, Hématy K, Saxe F, Ibanez AB, Vodermaier V,Konlechner C, Sampathkumar A, Rüggeberg M, Aichinger E, Neumetzler L,Burgert I, Somerville C, Hauser M, Persson S: CHITINASE-LIKE1/POM-POM1and its homolog CTL2 Are glucan-interacting proteins important forcellulose biosynthesis in Arabidopsis. Plant Cell 2012, 24(2):589–607.40. Pattathil S, Harper AD, Bar-Peled M: Biosynthesis of UDP-xylose:characterization of membrane-bound AtUxs2. Planta 2005,221(4):538–548.41. Wu A, Hörnblad E, Voxeur A, Gerber L, Rihouey C, Lerouge P, Marchant A:Analysis of the Arabidopsis IRX9/IRX9-L and IRX14/IRX14-L pairs ofglycosyltransferase genes reveals critical contributions to biosynthesis ofthe hemicellulose glucuronoxylan. Plant Physiol 2010, 153(2):542–554.42. Lee C, Teng Q, Huang W, Zhong R, Ye Z: The F8H glycosyltransferase is afunctional paralog of FRA8 involved in glucuronoxylan biosynthesis inArabidopsis. Plant Cell Physiol 2009, 50(4):812–827.43. Maris A, Kaewthai N, Eklöf JM, Miller JG, Brumer H, Fry SC, Verbelen J,Vissenberg K: Differences in enzymic properties of five recombinantxyloglucan endotransglucosylase/hydrolase (XTH) proteins of Arabidopsisthaliana. J Exp Bot 2011, 62(1):261–271.44. Lee C, Teng Q, Zhong R, Ye Z: The four Arabidopsis REDUCED WALLACETYLATION genes are expressed in secondary wall-containing cellsand required for the acetylation of xylan. Plant Cell Physiol 2011,52:1289–1301.45. Derbyshire P, McCann MC, Roberts K: Restricted cell elongation inArabidopsis hypocotyls is associated with a reduced average pectinesterification level. BMC Plant Biol 2007, 7:31.46. Hongo S, Sato K, Yokoyama R, Nishitani K: Demethylesterification of theprimary wall by PECTIN METHYLESTERASE35 provides mechanicalsupport to the Arabidopsis stem. Plant Cell 2012, 24(6):2624–34.47. Kim JB, Carpita NC: Changes in esterification of the uronic acid groups ofcell wall polysaccharides during elongation of maize coleoptiles.Plant Physiol 1992, 98:646–653.48. Siedlecka A, Wiklund S, Peronne M, Micheli F, Lesniewska J, Sethson I,Edlund U, Richard L, Sundberg B, Mellerowicz EJ: Pectin methyl esteraseinhibits intrusive and symplastic cell growth in developing wood cells ofPopulus. Plant Physiol 2008, 146:554–565.49. Berthet S, Demont-Caulet N, Pollet B, Bidzinski P, Cézard L, Le Bris P, BorregaN, Hervé J, Blondet E, Balzergue S, Lapierre C, Jouanin L: Disruption ofLACCASE4 and 17 results in tissue-specific alterations to lignification ofArabidopsis thaliana stems. Plant Cell 2011, 23(3):1124–1137.50. Ko J, Beers E, Han K: Global comparative transcriptome analysis identifiesgene network regulating secondary xylem development in Arabidopsisthaliana. Mol Genet Genomics 2006, 276(6):517–531.51. Ko J, Han K, Park S, Yang J: Plant body weight-induced secondary growthin Arabidopsis and its transcription phenotype revealed by whole-transcriptome profiling. Plant Physiol 2004, 135(2):1069–1083.52. Manosalva PM, Bruce M, Leach JE: Rice 14-3-3 protein (GF14e) negativelyaffects cell death and disease resistance. Plant J 2011, 68(5):777–787.53. Sun L, van Nocker S: Analysis of promoter activity of members of thePECTATE LYASE-LIKE (PLL) gene family in cell separation in Arabidopsis.BMC Plant Biol 2010, 10(1):152.54. Sampedro J, Pardo B, Gianzo C, Guitián E, Revilla G, Zarra I: Lack of α-xylosidase activity in Arabidopsis alters xyloglucan composition andresults in growth defects. Plant Physiol 2010, 154(3):1105–1115.55. Günl M, Pauly M: AXY3 encodes a α-xylosidase that impacts the structureand accessibility of the hemicellulose xyloglucan in Arabidopsis plantcell walls. Planta 2011, 233(4):707–719.56. Thomma B, Cammue B, Thevissen K: Plant defensins. Planta 2002,216:193–202.57. Rapaport H: Ordered peptide assemblies at interfaces. Supramol Chem2006, 18(5):445–454.58. Cannon MC, Terneus K, Hall Q, Tan L, Wang Y, Wegenhart BL, Chen L,Lamport DTA, Chen Y, Kieliszewski MJ: Self-assembly of the plant cell wallrequires an extensin scaffold. Proc Natl Acad Sci 2008, 105(6):2226–2231.59. Suh MC, Samuels AL, Jetter R, Kunst L, Pollard M, Ohlrogge J, Beisson F:Cuticular lipid composition, surface structure, and gene expression inArabidopsis stem epidermis. Plant Physiol 2005, 139(4):1649–1665.60. Chen X, Kim J: Callose synthesis in higher plants. Plant Signal Behav 2009,4(6):489–492.61. Schwechheimer C: Understanding gibberellic acid signaling - are wethere yet? Curr Opin Plant Biol 2008, 11(1):9–15.62. Rubinovich L, Weiss D: The Arabidopsis cysteine-rich protein GASA4promotes GA responses and exhibits redox activity in bacteria and inplanta. Plant J 2010, 64(6):1018–1027.Hall and Ellis BMC Plant Biology 2013, 13:14 Page 16 of 17http://www.biomedcentral.com/1471-2229/13/1463. Rose JKC, Braam J, Fry SC, Nishitani K: The XTH family of enzymes involvedin xyloglucan endotransglucosylation and endohydrolysis: Currentperspectives and a new unifying nomenclature. Plant Cell Physiol 2002,43(12):1421–1435.64. Akamatsu T, Hanzawa Y, Ohtake Y, Takahashi T, Nishitani K, Komeda Y:Expression of endoxyloglucan transferase genes inacaulis mutants ofArabidopsis. Plant Physiol 1999, 121(3):715–722.65. Endler A, Persson S: Cellulose Synthases and synthesis in Arabidopsis.Mol Plant 2011, 4(2):199–211.66. McNeil M, Darvill AG, Fry SC, Albersheim P: Structure and function of theprimary cell walls of plants. Annu Rev Biochem 1984, 53(1):625–663.67. Doering A, Lathe R, Persson S: An update on xylan synthesis. Mol Plant2012, 5(4):769–771.68. Yang J, Sardar HS, McGovern KR, Zhang Y, Showalter AM: A lysine-richarabinogalactan protein in Arabidopsis is essential for plant growth anddevelopment, including cell division and expansion. Plant J 2007,49(4):629–640.69. MacMillan CP, Mansfield SD, Stachurski ZH, Evans R, Southerton SG:Fasciclin-like arabinogalactan proteins: specialization for stembiomechanics and cell wall architecture in Arabidopsis and eucalyptus.Plant J 2010, 62:689–703.70. Lafarguette F, Leplé J, Déjardin A, Laurans F, Costa G, Lesage-Descauses M,Pilate G: Poplar genes encoding fasciclin-like arabinogalactan proteinsare highly expressed in tension wood. New Phytol 2004, 164(1):107–121.71. Ito S, Suzuki Y, Miyamoto K, Ueda J, Yamaguchi I: AtFLA11, a fasciclin-likearabinogalactan-protein, specifically localized in sclerenchyma cells.Biosci Biotechnol Biochem 2005, 69(10):1963–1969.72. Bhargava A, Mansfield SD, Hall HC, Douglas CJ, Ellis B: MYB75 functions inregulation of secondary cell wall formation in the Arabidopsisinflorescence stem. Plant Physiol 2010, 154(3):1428–1438.73. Tokumoto H, Wakabayashi K, Kamisaka S, Hoson T: Changes in the sugarcomposition and molecular mass distribution of matrix polysaccharidesduring cotton fiber development. Plant Cell Physiol 2002, 43(4):411–418.74. Bourquin V, Nishikubo N, Abe H, Brumer H, Denman S, Eklund M,Christiernin M, Teeri TT, Sundberg B, Mellerowicz EJ: Xyloglucanendotransglycosylases have a function during the formation ofsecondary cell walls of vascular tissues. Plant Cell 2002,14(12):3073–3088.75. Humphreys JM, Chapple C: Rewriting the lignin roadmap. Curr Opin PlantBiol 2002, 5(3):224–229.76. Douglas CJ, Ehlting J: Arabidopsis thaliana full genome longmermicroarrays: a powerful gene discovery tool for agriculture and forestry.Transgenic Res 2005, 14:551–561.77. Huber W, von Heydebreck A, Sultmann H, Poustka A, Vingron M:Variance stabilization applied to microarray data calibration andto the quantification of differential expression. Bioinformatics 2002,18(suppl 1):S96–S104.78. Storey J: A direct approach to false discovery rates. Journal of the RoyalStatistical Society: Series B (Statistical Methodology) 2002, 64:479.79. Brazma A, Hingamp P, Quackenbush J, Sherlock G, Spellman P,Stoeckert C, Aach J, Ansorge W, Ball CA, Causton HC, Gaasterland T,Glenisson P, Holstege FC, Kim IF, Markowitz V, Matese JC, ParkinsonH, Robinson A, Sarkans U, Schulze-Kremer S, Stewart J, Taylor R, ViloJ, Vingron M: Minimum information about a microarrayexperiment (MIAME)-toward standards for microarray data.Nat Genet 2001, 29:365–371.doi:10.1186/1471-2229-13-14Cite this article as: Hall and Ellis: Transcriptional programming duringcell wall maturation in the expanding Arabidopsis stem. BMC PlantBiology 2013 13:14.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/submitHall and Ellis BMC Plant Biology 2013, 13:14 Page 17 of 17http://www.biomedcentral.com/1471-2229/13/14


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