REGULAR ARTICLE The proteome of maritime pine wood forming tissue Jean-Marc Gion 1 *, Céline Lalanne 1 *, Grégoire Le Provost 1 *, Hélène Ferry-Dumazet 2 , Jorge Paiva 1, 3 , Phillipe Chaumeil 1 , Jean-Marc Frigerio 1 , Jean Brach 1 , Aurélien Barré 2 , Antoine de Daruvar 2, 4 , Stéphane Claverol 5 , Marc Bonneu 5 , Nicolas Sommerer 6 , Luc Negroni 7 and Christophe Plomion 1 1 UMR 1202 BIOGECO, INRA, Equipe de Génétique, Cestas, France 2 Centre de Bioinformatique de Bordeaux, Université V. Segalen Bordeaux 2, Bordeaux, France 3 Plant Cell Biotechnology Lab. IBET/ITQB, Oeiras, Portugal 4 UMR 5162, Génomique Fonctionnelle des Trypanosomatides, CNRS – Université Bordeaux 2, Bordeaux, France 5 Pôle Protéomique, Plateforme Génomique Fonctionnelle Bordeaux, Université V. Segalen Bordeaux 2, Bordeaux, France 6 Unité de Recherches Protéomique, UR 1199, INRA, Montpellier, France 7 UMR de Génétique Végétale, INRA/UPS/CNRS/INA-PG, Gif-sur-Yvette, France Wood is one of our most important natural resources. Surprisingly, we know hardly anything about the details of the process of wood formation. The aim of this work was to describe the main proteins expressed in wood forming tissue of a conifer species (Pinus pinaster Ait.). Using high resolution 2-DE with linear pH gradient ranging from 4 to 7, a total of 1039 spots were detected. Out of the 240 spots analyzed by MS/MS, 67.9% were identified, 16.7% presented no homology in the data- bases, and 15.4% corresponded to protein mixtures. Out of the 57 spots analyzed by MALDI-MS, only 15.8% were identified. Most of the 175 identified proteins play a role in either defense (19.4%), carbohydrates (16.6%) and amino acid (14.9%) metabolisms, genes and proteins expression (13.1%), cytoskeleton (8%), cell wall biosynthesis (5.7%), secondary (5.1%) and primary (4%) metabolisms. A summary of the identified proteins, their putative functions, and behavior in different types of wood are presented. This information was introduced into the PROTICdb database and is accessible at http://cbib1.cbib.u-bordeaux2.fr/Protic/Protic/home/index.php. Finally, the average protein amount was compared with their respective transcript abundance as quantified through EST counting in a cDNA-library constructed with mRNA extracted from wood forming tissue. Received: September 6, 2004 Revised: November 17, 2004 Accepted: December 14, 2004 Keywords: Mass spectrometry / Pinus pinaster Ait. / Proteome analysis / Wood Proteomics 2005, 5, 3731–3751 3731 1 Introduction In perennial plants, the successive addition of secondary xylem tissue differentiated from the vascular cambium gives rise to a unique tissue called wood. Wood is composed of non-conducting and conducting elements implicated in the long distance transport of water and nutriments in trees. In conifers, wood is comprised of two main cell types: tracheids and ray parenchyma. This simplicity hides the fact that it is also a highly variable raw material. Field experiments have shown genetic factors can influence the activity of the vas- cular cambium and the differentiation of newly divided cells, ultimately influencing wood and end-use properties [1–3]. The ageing process constitutes another important source of variation affecting the characteristics of secondary xylem (reviewed in Zobel and Sprague [4]). Wood derived from a young cambium is referred as juvenile wood (JW), while Correspondence: Dr. Christophe Plomion, UMR 1202 BIOGECO, INRA, Equipe de Génétique, 69 route d’Arcachon, F-33610 Cestas Cédex, France E-mail: [email protected]Fax: 133-5-5712-2881 Abbreviations: SAM-S, S-adenosylmethionine synthetase; JW, juvenile wood; MW, mature wood; EW, early wood; LW, late wood; OW, opposite wood; CW, compression wood * These authors contributed equally. 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.de DOI 10.1002/pmic.200401197
21
Embed
REGULAR ARTICLE The proteome of maritime pine wood forming ... et al Proteomics 2005.pdf · REGULAR ARTICLE The proteome of maritime pine wood forming tissue Jean-Marc Gion 1*, Céline
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
REGULAR ARTICLE
The proteome of maritime pine wood forming tissue
Jean-Marc Gion1*, Céline Lalanne1*, Grégoire Le Provost1*, Hélène Ferry-Dumazet2,Jorge Paiva1, 3, Phillipe Chaumeil1, Jean-Marc Frigerio1, Jean Brach1, Aurélien Barré2,Antoine de Daruvar2, 4, Stéphane Claverol5, Marc Bonneu5, Nicolas Sommerer6,Luc Negroni7 and Christophe Plomion1
1 UMR 1202 BIOGECO, INRA, Equipe de Génétique, Cestas, France2 Centre de Bioinformatique de Bordeaux, Université V. Segalen Bordeaux 2, Bordeaux, France3 Plant Cell Biotechnology Lab. IBET/ITQB, Oeiras, Portugal4 UMR 5162, Génomique Fonctionnelle des Trypanosomatides, CNRS – Université Bordeaux 2, Bordeaux, France5 Pôle Protéomique, Plateforme Génomique Fonctionnelle Bordeaux, Université V. Segalen Bordeaux 2,
Bordeaux, France6 Unité de Recherches Protéomique, UR 1199, INRA, Montpellier, France7 UMR de Génétique Végétale, INRA/UPS/CNRS/INA-PG, Gif-sur-Yvette, France
Wood is one of our most important natural resources. Surprisingly, we know hardly anything aboutthe details of the process of wood formation. The aim of this work was to describe the main proteinsexpressed in wood forming tissue of a conifer species (Pinus pinaster Ait.). Using high resolution2-DE with linear pH gradient ranging from 4 to 7, a total of 1039 spots were detected. Out of the240 spots analyzed by MS/MS, 67.9% were identified, 16.7% presented no homology in the data-bases, and 15.4% corresponded to protein mixtures. Out of the 57 spots analyzed by MALDI-MS,only 15.8% were identified. Most of the 175 identified proteins play a role in either defense (19.4%),carbohydrates (16.6%) and amino acid (14.9%) metabolisms, genes and proteins expression (13.1%),cytoskeleton (8%), cell wall biosynthesis (5.7%), secondary (5.1%) and primary (4%) metabolisms. Asummary of the identified proteins, their putative functions, and behavior in different types of woodare presented. This information was introduced into the PROTICdb database and is accessible athttp://cbib1.cbib.u-bordeaux2.fr/Protic/Protic/home/index.php. Finally, the average protein amountwas compared with their respective transcript abundance as quantified through ESTcounting in acDNA-library constructed with mRNA extracted from wood forming tissue.
Received: September 6, 2004Revised: November 17, 2004
In perennial plants, the successive addition of secondaryxylem tissue differentiated from the vascular cambium givesrise to a unique tissue called wood. Wood is composed of
non-conducting and conducting elements implicated in thelong distance transport of water and nutriments in trees. Inconifers, wood is comprised of two main cell types: tracheidsand ray parenchyma. This simplicity hides the fact that it isalso a highly variable raw material. Field experiments haveshown genetic factors can influence the activity of the vas-cular cambium and the differentiation of newly divided cells,ultimately influencing wood and end-use properties [1–3].The ageing process constitutes another important source ofvariation affecting the characteristics of secondary xylem(reviewed in Zobel and Sprague [4]). Wood derived from ayoung cambium is referred as juvenile wood (JW), while
Correspondence: Dr. Christophe Plomion, UMR 1202 BIOGECO,INRA, Equipe de Génétique, 69 route d’Arcachon, F-33610 CestasCédex, FranceE-mail: [email protected]: 133-5-5712-2881
3732 J.-M. Gion et al. Proteomics 2005, 5, 3731–3751
wood formed by an older cambium is referred as maturewood (MW). In fast growing pine trees, the transition be-tween JW and MW occurs around 10 years of age (Fig. 1A)and is accompanied by drastic changes in many wood prop-erties. Seasonal and gravitational effects are, among envi-ronmental factors, the most significant external sources ofvariation affecting wood characteristic. In temperate zones,climatic variation during the annual course of the vascularcambium give rise to early wood (EW) formed early duringthe growing season, and late wood (LW) formed in late sum-mer (Fig. 1B). The major changes are in the structure of thetracheids, which affect their ability to transport water underwet and dry conditions. A change in the orientation of aconifer tree stem stimulates the formation of a specializedtype of wood at the underside of a bent tree, termed com-pression wood (CW) (Fig. 1C). It serves to reorient the stemto a vertical position. CW differs anatomically in its chemicalcomposition, compared to opposite wood (OW) formed atthe other side of the leaning stem (reviewed in Timell [5]).The formation of these six types of wood is the result of pro-found molecular changes during xylogenesis, triggered byexternal (e.g., temperature, photoperiod [6, 7]) and/or endog-enous factors (e.g., phytohormones, sugars [8]). The con-siderable plasticity in anatomical, chemical and physicalwood properties provides a unique opportunity to dissect themolecular and biochemical mechanisms responsible forsuch differences.
Wood formation (xylogenesis) includes four major steps:cell division, cell expansion, secondary cell wall thickeningand programmed cell death (reviewed by Lachaud et al. [7]
and Plomion et al. [9]). It is a complex phenomenon driven bythe coordinate expression of numerous genes especiallyinvolved in the biosynthesis and the assembly of poly-saccharides, lignins, and cell wall proteins [8, 10]. Up to now,the study of molecular mechanisms involved in the develop-ment of wood has mainly taken a transcriptomic approach,combining expressed sequence tag (EST) sequencing andtranscript profiling [11–18]. Comparatively, there has beenno large-scale project to identify proteins from differentiat-ing secondary xylem, and only few studies have reported on ahandful of proteins in wood forming tissue [19–24]. Theobjective of the present work was to partially fill this gap andprovide for the first time in a forest tree species an overviewof the proteome expressed in this highly specialized tissue,and to serve as a basis for future proteome comparisons ofenvironmentally challenged trees, and in the course of theirdevelopment.
Maritime pine (Pinus pinaster Ait.), a conifer of greateconomic and ecological interest in Southwestern Europe(where it covers 4 millionhectares), was chosen as a modelspecies. A reference map was first obtained using high reso-lution 2-DE with proteins extracted from differentiatingxylem associated to the six types of wood mentioned above. Atotal of 300 spots were then excised from the gels and ana-lyzed by LC ESI-MS/MS, MALDI-TOF MS or internal se-quencing. The identified proteins are discussed and classi-fied based on their putative function and their behavior inthe six types of wood. Finally, the expression levels of 95 pro-teins quantified by 2-DE was compared with mRNA levelsquantified by EST counting.
Figure 1. The six types of wood typically found in a conifer tree. (A) Juvenile wood (JW) vs. mature wood (MW),(B) early wood (EW) vs. late wood (LW), and (C) compression wood (CW) vs. opposite wood (OW). (D) Upper andlower part of a leaning stem of a 4 months bent tree showing the red wood phenotype of CW immediately afterdebarking, 14-year-old tree.
To take into account the natural variability found in the woodof an adult conifer tree, differentiating xylem tissues weresampled: (i) at the base (breast height) and at the top of thestem of a 30-year-old maritime pine tree, corresponding toxylem associated to MW (formed by a 25-year-old cambium)and JW (formed by a 3-year-old cambium), respectively.Samples were taken in April (27.04.01). (ii) in April(25.04.00) and August (23.08.00) of two 14-year-old maritimepine trees belonging of the same clone (accession #4015),corresponding to xylem associated to EW and LW, respec-tively. (iii) in the upper and lower side of a 14-year-old mar-itime pine genotype (accession #105), whose grafted copieswere bent to a 157 angle by tying their trunk to neighbortrees, and sampled after 2 years of mechanical bending, cor-responding to xylem associated to OW and CW, respectively.Samples were taken in August (23.08.00).
After that, bark, phloem and cambium were peeled fromthe stem, scrapings were taken from exposed differentiatingxylem, immediately frozen in liquid N2 and stored at 2807Cuntil used for protein extraction.
2.2 Protein extraction and quantification
Starting from 500 mg fresh tissue, total protein of each of thesix samples described above was extracted following the pro-cedure described by Damerval et al. [25], with the followingmodifications: (i) for protein resolubilization, the “UKS”buffer was replaced by “TCT” buffer (urea 7 M, thiourea 2 M,Triton X-100 0.4%, CHAPS 4%, DTT 10 mM, IPG buffer 1%),(ii) samples were then centrifuged (4 min, 2000 rpm, 207C)and the supernatant was transferred to a new Eppendorftube. This step was added in order to insure that all cellularfragments were removed from the extract. Proteins werestored at 2807C. Three extractions were completed for eachsample and pooled for protein quantification. The resultingmix was quantified over six replicated assays, using the pro-tocol described by Ramagli et al. [26]. The mean concentra-tion was then calculated and used to load 300 mg of proteinson each IPG strip.
2.3 2-DE
2-DE [27] was used to analyze total protein from the xylemsamples following the procedure of Bahrman et al. [28]adapted for the IPGphor system (Amersham Biosciences,Uppsala, Sweden). For the IEF, 24 cm strips were used with alinear pH gradient ranging from 4 to 7. Proteins were mixedwith a strip rehydration solution (urea 7 M, thiourea 2 M, Tri-ton X-100 0.4%, CHAPS 4%, DTT 10 mM, IPG buffer 1%).The IPGphor system was then programmed for 12 h at 30 V(active rehydration), 1 h at 200 V, 1 h at 500 V, 1 h at 1000 V,
30 min from 1000 V to 8000 V and finally 8000 V per hour toachieve a total of 74 000 Vh. After approximately 15 h of IEF,strips were equilibrated (SDS saturation) with a 10 mL ofequilibration solution (Tris-HCl pH 8.8 50 mM, urea 6 M,glycerol 30%, SDS 2%, bromophenol blue). Equilibrationwas performed in two steps, with DTT (65 mM) in the firstequilibration, and iodoacetamide (135 mM) in the secondequilibration (without DTT). SDS-PAGE was performed bybatches of 15 gels run in a buffer (Tris 25 mM, glycine 0.2 M
and SDS 0.1 M) at 110 V for 17 h. To ensure gel reproduci-bility, five replicates were performed for each sample, result-ing in a total of 30 gels from which the four best were select-ed with the help of the image analysis software.
2.4 Gel staining
CBB G-250 (Bio-Rad, Hercules, CA, USA) was used for gelstaining. Gels were fixed for 2 h in a solution containing2% phosphoric acid and 50% ethanol. After three waterwashings of 30 min each, the gels were placed in an incuba-tion solution (methanol 34%, ammonium sulfate 17%,phosphoric acid 2%) for 1 h, and then immersed in a stain-ing solution (methanol 34%, ammonium sulfate 17%, phos-phoric acid 2%, Coomassie blue 0.05%) for 5 days. Finally,the gels were stored in a 5% acid acetic solution before scan-ning and spot picking after several days.
2.5 Image acquisition and spot detection
Stained gels were digitalized using the M141 image scannerand the LabScan software (Amersham Biosciences). First, acalibration with a grey scale was necessary to transform greylevels into OD values for each pixel of the gel picture. Thecalibration method used was the colloidal blue methoddescribed in the LabScan manual. All the gel pictures weresaved as tiff files. Image analysis was performed using theImage Master 2D-Elite software (IM2D; Amersham Bio-sciences). The 30 gel images were placed in one folder. Thewizard detection method proposed by the software was usedto detect the spots. Then, automatically detected spots weremanually checked, and some of them manually added orremoved. Following the detection procedure, the volume foreach spot corresponded to a gross value. In order to eliminatethe background from this gross value, the mode of non spotof IM2D was used. Finally, all the gels were matched in orderto attribute a common spot identity for the same spotsderived from different images. For this, we used the auto-matically matching options of IM2D. After visual checking ofthe matching, the IM2D software was used to construct amaster gel (reference gel, Fig. 2). For each sample, when aprotein was detected in all of the four replicates, this proteinwas automatically added to the master gel, thus creating areference map of wood forming tissue. Normalized volumeswere finally obtained using the total spot volume normal-ization procedure of IM2D.
3734 J.-M. Gion et al. Proteomics 2005, 5, 3731–3751
Figure 2. Reference 2-DE map for maritime pine wood forming tissue (4–7 linear gradient). Proteins that were identified are marked witharrows and numbers following Table 1. Unknown function proteins are squared.
2.6 Characterization by MS
2.6.1 In-gel protein digestion
CBB-stained protein spots were manually excised from thegels and washed twice with ultra-pure water. Spots weresubsequently washed in H2O/MeOH/acetic acid (47.5:47.5:5)until destaining. The solvent mixture was removed and re-placed by ACN. After shrinking of the gel pieces, ACN wasremoved and gel pieces were dried in a vacuum centrifuge.Gel pieces were rehydrated in 10 ng/mL trypsin (Sigma-Aldrich, St. Louis, MO, USA) in 50 mM NH4HCO3 andincubated overnight at 377C. The supernatant was removedand stored at 2207C, and the gel pieces were incubated15 min in 50 mM NH4HCO3 at room temperature underrotary shaking. This second supernatant was pooled with theprevious one, and a H2O/ACN/HCOOH (47.5:47.5:5) solu-tion was added to the gel pieces for 15 min. This step was
repeated once. Supernatants were pooled and concentratedin a vacuum centrifuge to a final volume of 30 mL. Digestswere finally acidified by addition of 1.8 mL of acetic acid andstored at 2207C.
2.6.2 On-line capillary HPLC nanospray ion trap
MS/MS analysis
Peptide mixtures were analyzed by on-line capillary HPLC(LC Packings, Amsterdam, The Netherlands) coupled to ananospray LCQ ion trap mass spectrometer (Thermo-Finnigan, San Jose, CA, USA). Peptides were separated on a75 mm id615 cm C18 PepMapTM column (LC Packings).The flow rate was set at 200 nL/min. Peptides were elutedusing a 5–50% linear gradient of solvent B in 30 min (sol-vent A was 0.1% formic acid in 5% ACN, and solvent B was0.1% formic acid in 80% ACN). The mass spectrometer wasoperated in positive ion mode at a 2.5 kV needle voltage and a
44 V capillary voltage. Data acquisition was performed in adata-dependent mode consisting of alternatively in a singlerun, a full scan MS over the range m/z 50–2000 and a fullscan MS/MS in an exclusion dynamic mode. MS/MS datawere acquired using a 2 m/z units ion isolation window, a35% relative collision energy, and a 5 min dynamic exclusionduration. Peptides were identified with SEQUEST (Thermo-Finnigan, Torrence, CA, USA) using the 18 254 Pinus pinasterEST (http://cbi.labri.fr/outils/SAM/COMPLETE/index.php),and the 59 447 Pinus taeda xylem EST comprising 8046 con-tigs and 12 437 singletons (http://pinetree.ccgb.umn.edu/).The contig names in Table 1 correspond to the Novem-ber 2002 assembly [29]. The Swiss-Prot database (http://us.expasy.org/sprot/) was also used to evaluate the rate of proteinidentification using nonconiferous nucleotide sequences.
2.6.3 PMF by MALDI-TOF MS
For 57 spots (six of which being also sampled for LC ESI-MS/MS analysis), we used MALDI-TOF MS following the proto-col and analysis procedure described by Sarry et al. [30]. TheMASCOT search engine software (Matrix Science, London,UK) was used to search the NCBI nonredundant and specificpine EST databases on a local server.
2.7 Differentiating xylem cDNA library construction
and EST sequencing
A composite cDNA library was obtained using equal amountsof total RNA extracted from the same samples as describedfor the protein analysis, but for JW and MW. Total RNA weremixed and poly A(1) RNA isolated from this bulked sample.The cDNA library was made using the l-ZAP-cDNA synthe-sis kit (Stratagene, La Jolla, CA, USA). Approximately10 000 l clones were excised to generate plasmid clones. The10 000 plasmid clones were sequenced using the Templifi kit(Amersham Biosciences), by single pass from the 5’-end togenerate the EST collection. Only sequences longer than60 nucleotides were kept for further analysis. EST annotationwas based on a search for homology with public protein andnucleic acid sequence databases using the BLAST software[31]. Homologs were sequentially searched in Swiss-Prot(BLASTX), TrEMBL (BLASTX), EMBL (BLASTN), and lastlyin dbEST database (BLASTN). At each step, the process wasstopped if a gene with similar sequence was found (definedby an expected value lower than 1025 for BLASTX and 10210
for BLASTN searches). A total of 8429 EST were finally sub-mitted to dbEST (http://www.ncbi.nlm.nih.gov/dbEST/), andcan be retrieved using the search fields organism [Pinuspinaster] and tissue_type[xylem].
2.8 Statistical analyses
To appreciate the relatedness between the six types of wood(i.e., JW, MW, EW, LW, OW, CW), based on a proteomic dis-tance obtained from either the 1039 detected spots, or a
restricted dataset of 215 spots (the 175 known and 40 un-known function proteins), we used the hierarchical cluster-ing software EPCLUST available at URL: http://ep.ebi.ac.uk/EP/. The Euclidian distance and UPGMA algorithm wereused for the analysis. The same software and options wereused to cluster the 215 spots according their log2 trans-formed expression profiles along the six samples. Simplet-tests were performed for each pair-wise comparison,namely JW vs. MW, EW vs. LW, and OW vs. CW, to detectthose proteins showing significant (p-value , 0.01) onto-genic, seasonal, and gravitational effect.
3 Results and discussion
3.1 2-DE reference map of maritime pine wood
forming tissue
The 2-DE reference map of maritime pine wood formingtissue was established using proteins extracted from differ-entiating xylem associated to JW, MW, EW, LW, CW and OW,and separated by 2-DE. For each samples five replicated gelswere performed. After colloidal blue staining, they werescanned with the LabScan software and analyzed using theIM2D software. The image obtained for differentiating xylemassociated to OW after 2 years of bending (replicate #1) wasrandomly chosen to build the reference gel (master gel) onwhich spots specifically detected on other samples (through-out the four best replicates) were added. Overall, 445, 468,506, 581, 552, 570 spots were detected in MW, JW, EW, LW,OW, and CW, respectively. A total of 1039 spots were finallyplaced on the reference map (Fig. 2), among which 300 pro-teins (29%) were excised from the polyacrylamide gel andanalyzed by mass spectrometry or internal microsequencing.As shown in Fig. 2, spots were picked randomly to ensure agood representation in terms of pH, molecular weight andprotein abundance.
3.2 Protein identification success rate
LC ESI-MS/MS analysis of the 240 spots was used for proteinidentification using protein (Swiss-Prot) and nucleotide(Pinus pinaster and Pinus taeda EST and contigs) databases.The overall identification success rate was 67.9%, corre-sponding to 163 spots identified. It should be noted thatsearching both databases appears quite redundant, since inonly five cases protein identification was achieved usingSwiss-Prot. It should also be noted that for 71 of the identifiedspots, the same hit was obtained in both nucleotide and pro-tein databases. In other words, the overlap in the number ofproteins identified in both databases was 43.6%. Thus, thepine EST allowed the identification of an additional 87 spots.This result clearly indicates the utility of pine EST to achievea high rate of protein identification. As for the remainingspots, 40 (16.7%) presented no homology in the query data-bases. Given the amount of pine xylem EST in public data-
bases (,90 000), these proteins might correspond to very raretranscript that have not be sequenced in EST projects to date,and may indicate a specific role of these proteins in wood for-mation. Finally, 37 spots (15.4%) corresponded to mixtures ofproteins. Comparatively, only 9 (15.8%) of the 57 spots analyzedby MALDI-TOF MS were identified using the same databases,from which six were also analyzed and identified by MS/MS(#83, #212, #217, #226, #242, and #282), and three (#215, #216,and #223) corresponded to tubulin b chain isoforms. Suchdiscrepancy in identification rates between MS/MS and MSalone has recently been reported in Panax ginseng [32], and canlargely be attributed to the lack of genome database or extensiveEST datasets for most plant species, gymnosperms in particu-lar. To these 166 proteins identified by mass spectrometry, oneshould add another nine proteins corresponding to very abun-dant proteins in wood forming tissues, and whose function waspreviously determined by internal microsequencing [19].These proteins have been also localized on the reference gel(Fig. 2) and correspond to the following functions: #80, #81,and #88: HSP 70 kDa, #113: protein disulfide isomerase (EC5.3.4.1), #266: SAM-S (S-adenosylmethionine synthetase) (EC2.5.1.6), #297: actin, the most intense spot on our gels, #430:isoflavone reductase (EC 1.3.1.) probably the second mostintense spot not only in this experiment, but also reported asthe most highly expressed protein in poplar wood formingtissue [20], #439: caffeoyl-CoA-O-methyltransferase (EC2.1.1.104), and #482: ascorbate peroxidase (EC 1.11.1.11).Overall, 175 proteins were thus identified in this study. Theseproteins are marked with arrows and numbers in Fig. 2. Over-all, membrane proteins were under-represented, with theexception of a vacuolar ATPase subunit (#340). This observa-tion is not specific to the wood proteome, and can be attributedto the general poor solubilization of such proteins.
3.3 Protein identification of wood forming tissue and
functional classification
A summary of protein functions is given in Table 1. The func-tional distribution of the known function proteins is reportedin Fig. 3. Proteins were classified into 15 groups based on
functional categories using the DBGET system (http://www.genome.ad.jp/dbget/). Interestingly, 87% of the pro-teins fell into eight major groups, while 13% were classifiedin seven other minor groups. Major groups were for “defense”(34 spots), “carbohydrate metabolism” (29 spots), “amino acidmetabolism” (26 spots), “genes and proteins expression”(23 spots) (including signal transduction, transcription,translation, protein assembly, modification and degradation),“cytoskeleton” (14 spots), “cell wall” (10 spots), “secondarymetabolism” (9 spots), and “primary metabolism” (7 spots).Overall, 39.4% of the proteins appeared as multiple spots andaccounted for most of the proteins found in the groups. Sucha high number of spots attributed to one protein has recentlybeen reported in Medicago truncatula [33]. This observationmay reflect post-translational modification, allelic variation ofthe same protein (e.g., position shift variants), isozyme varia-tion (proteins encoded by paralogs), alternative splicingevents, but also protein degradation.
SAM-S for example was represented by 14 spots (3.36%of the 300 studied proteins), accounting for 53.8% of theproteins of the amino acid metabolism category. SAM servesas a universal methyl group donor in numerous trans-methylation reactions that involve many types of acceptormolecules (proteins, nucleic acids, polysaccharides, fattyacids). It is also the substrate of many reactions, such asvitamins, polyamines, the gaseous phytohormone ethylene,and nucleotide biosynthesis. SAM is believed to be next toATP for the number of reactions in which a biological com-pound is used [34]. The transcript of SAM-S was found at ahigh level in cDNA libraries constructed from pine andpoplar differentiating xylem [11, 12, 35]. Among others,SAM-S plays a role in the methylation of monolignol pre-cursors during lignin biosynthesis [36].
The carbohydrate metabolism category presented aredundancy of 51.7%, with most proteins present two to fourtimes, e.g., pyrophosphate fructose 6-phosphate 1-phospho-transferase and ascorbate peroxidase (four spots), 2,3-bisphos-phoglycerate-independent phosphoglycerate mutase andphosphoglucomutase (three spots), fructokinase, transketo-lase, alcohol dehydrogenase, enolase, aconitase (two spots).
Figure 3. Functional distributionof the major proteins in mar-itime pine wood forming tissue,as separated by 2-DE.
3744 J.-M. Gion et al. Proteomics 2005, 5, 3731–3751
Carbohydrates (cellulose, hemicelluloses) are essential com-ponents of the cell wall. The abundance of these proteinsclearly shows the importance of carbohydrate biosyntheticpathways during xylogenesis [37].
Arabinogalactan proteins (AGP) represented 40% of thecell wall category. EST sequencing in trees [12, 35, 38] hasrevealed a high level of transcript accumulation of these pro-teins in wood forming tissue. In the maritime pine xylemEST database, this protein ranked first among the mosthighly expressed genes. Moreover, Loopstra and Sederoff [39]reported that some AGP were preferentially expressed indifferentiating xylem compared to other tissue, suggestingan important role of these proteins in vascular development.Enzymes involved in the second most abundant compoundof the cell wall, lignins, were also well represented in the setof studied proteins, with four proteins identified, corre-sponding to C-CoA-OMT (caffeoyl-CoA-O-methyltransfer-ase), COMT (caffeic acid 3-O methyltransferase), CAD (cin-namyl-alcohol dehydrogenase) and a peroxidase.
The 10 spots classified in the cytoskeleton category cor-responded to only two proteins, namely tubulin (a and bsubunits) and actin. These proteins were also found to behighly expressed at the transcriptome level in loblolly pine[14] and poplar [35] wood forming tissues. Actin and tubulinconstitutes essential component of the cytoskeleton. Corticalmicrotubules are mainly composed of a and b tubulin. Cor-tical microtubules could determine the cell wall pattern bydefining the position and the orientation of cellulose micro-fibrils during the differentiation of tracheid elements [40],probably by guiding the movement of the cellulose-synthe-sizing complex in the plasma membrane.
The defense category mainly comprised heat shock pro-teins (HSP): 20 spots in total of low and high molecularweight, representing 11.4% of the identified proteins in thisstudy. Canton et al. [16] showed that HSP were much moreexpressed in differentiating xylem of maritime pine com-pared to other tissues (pollen, bud, phloem, cambium, nee-dles, and root). HSP are well known to be produced in re-sponse to various stresses [41–43]. Synthesis also occursduring developmental processes such as pollen or seedmaturation [44], and early seedling growth [45]. Recently, LeProvost et al. [46] hypothesized that some HSPs could havespecific role during wood formation, showing that theseproteins are important proteins for the normal developmentof secondary wood. The presence of multiple spots corre-sponding to LEA (late embryogenesis abundant) like pro-teins, abscisic stress ripening-like protein and stress inducedproteins is also worth noting and could be related to thepresence of drought stressed tissues, namely LW, OW andCW, sampled in summer.
The gene and protein expression category was repre-sented by 23 spots involved in signal transduction, tran-scription, translation, protein assembly, modification anddegradation. Most of these proteins (11 spots) correspondedeither to ribosomal protein or initiation and elongation fac-tors. It has been reported that the expression of initiation
factor could be related to GTP-binding protein. Moreover,Schultheiss et al. [47] have shown that GTP-binding proteinsare potentially involved in cellular shape and cell wall for-mation.
Subunits of the ATP-synthase complex were the mostabundant proteins of the primary metabolism category. Thisobservation is certainly related to the high energy demandfor tracheids elongation and growth.
The secondary metabolism was represented by eightproteins, two of which being similar to protein disulfideisomerase (PDI). In endoplasmic reticulum of eukaryotes,PDI catalyzes the formation, isomerization and reduction ofdisulfide bonds to ensure the correct folding of secretoryproteins prior to their further modification and transport[48]. High expression of PDI during wood formation isprobably related to the high metabolic activity existing invascular cambium.
3.4 Correlation between protein and mRNA
abundance
The relationship between mRNA and protein abundances isneeded to elucidate the processes and regulation of tran-scription and translation. For this comparison, mRNAabundance was estimated for unique functional annotationsby counting the number of ESTs corresponding to thesefunctions among the 8429 xylem ESTs. Protein amount wasestimated by determining the volume of each spot averagedacross the four samples used to generate the cDNA library(i.e., OW, CW, EW, and LW). Average values for a given func-tional annotation (e.g., SAM-S) were then summed to obtainthe global volume of that function. Raw data for proteinamount and number of EST are provided in Table 2. Thecorrelation between mRNA and protein level is shown inFig. 4. There was a general trend of increased protein level,resulting from increase in mRNA level. The Pearson corre-lation coefficient for the whole dataset was 0.46. However,when highly expressed genes at the transcript (AGP) andprotein (SAM-S, actin, a and b tubulin, ATP synthase bchain, isoflavone reductase and HSP70 kDa) levels wereremoved from the dataset, the correlation coefficient was stillpositive but dropped down to 0.31. Although our analysiswas restricted to a limited number of highly abundant pro-teins (i.e., revealed by 2-DE), this result indicates a weaklypositive correlation between mRNA and protein abundance.A similar result has also been reported in yeast by Gygi et al.[49] and Washburn et al. [50]. In a recent report on the pro-teome of Medicago truncatula, Watson et al. [51] reported on amoderate level (r = 0.50) of correlation between mRNAabundance (estimated by EST counting) and protein amount(estimated by 2-DE). Given the biased representation of thepresent proteome (poor representation of membrane, highmolecular weight and basic proteins), we believe that 0.31may represent a lower limit of the wood proteome.
The clustering of the 215 proteins (175 known and40 unknown function proteins) analyzed in this study andquantified over the six types of wood, clearly showed thatseasonal effect was the main factor controlling protein accu-mulation in wood forming tissue. Indeed, the six samplesclustered together into two distinct sub-trees (Fig. 5), withthe three samples collected in summer (LW, CW and OW)forming a first branch, and the three samples collected inspring (EW, JW and MW) forming a second branch. Then,ontogenic (JW vs. MW) and gravitational (OW vs. CW) effectsranked second and third, respectively. The same conclusionscould be drawn from the analysis of the whole dataset(1039 spots, data not shown). A simple pair-wise comparison(t-test), confirmed that the seasonal factor presented themost important effect on protein accumulation in secondaryxylem during wood formation. Indeed, 39.5, 30.7, and 20% ofthe identified proteins exhibited distinctive expression pat-terns between EW vs. LW, JW vs. MW, and OW vs. CW,respectively (Table 1). Examples of differentially expressed
proteins are given in Fig. 6. Given that the quality of woodand derived products largely depends on the physical andchemical properties of xylem secondary cell wall, and giventhe phenotypic differences in terms of wood quality betweenthese six samples, it can be hypothesized that some of theseproteins could be related to the changes in secondary cellwall structure and composition, and therefore representinteresting targets potentially controlling wood and end-useproperties.
To cluster the proteins showing similar expressionprofiles in the six types of wood, hierarchical clusteringwas applied to the 215 proteins (Fig. 7A). Interestingly,while among the most abundant proteins, actin, tubulins,AGP, 14-3-3 tended to be clustered and expressed con-stitutively across the six types of wood, SAM-S spots werefound to be distributed throughout the dendrogram,showing that members of this multigene family are eitherexpressed constitutively (#279, #282, #275, #321), or speci-fically regulated by environmental and/or ontogenic factors(Fig. 7B, cluster C1). In the following paragraphs we willonly discuss three clusters characteristic of protein over-
Figure 4. Correlation betweenprotein (x axis, normalized vol-ume) and mRNA (y axis, num-ber of EST) abundance. Plainline: correlation for the wholedataset (88 functions listed inTable 2, r = 0.46). Inset: correla-tion for a restricted datasetwhere AGP, SAM-S, actin, a andb tubulin, ATP synthase b chain,isoflavone reductase andHSP70 kDa were removed (r =0.31).
Figure 5. Clustering result between JW, MW, EW, LW, OW, andCW based on the proteins listed in Tab. 1.
expressed in either LW, MW, or CW, as well as one clusterand one protein overexpressed in normal wood and EW,respectively.
The LW cluster contained 21 proteins (Fig. 7B, clus-ter C2), most of which already reported as drought respon-sive in plants, such as: (i) the abscisic stress ripening protein,a plant gene with unknown biological role that becomesoverexpressed under water- and salt-stress conditions [52];(ii) low and high molecular weight HSP, including theendoplasmin precursor 94 kDa glucose-regulated protein (aholomog of HSP90) and stress-induced proteins (sti1-likeprotein) (reviewed by [53]); and (iii) isoflavone reductase [54].As for the two SAM-S accumulating in LW tissue (#260,#213), it should be reminded that besides their central role inplant growth and development [55], these proteins have alsobeen reported as drought stress regulated [56, 57]. Con-versely, one spot (#586) corresponding to an arabinoga-lactan/proline-rich protein (AGP) was identified as an EWprotein. AGP are abundant in the plant cell wall. They havebeen reported as among the most expressed genes in differ-entiating xylem of poplar and pine stems, undergoing radialexpansion by secondary growth [11, 12, 14, 16, 18]. Althoughtheir exact functions are unclear, they are implicated indiverse processes associated with plant growth and develop-ment, including cell proliferation (reviewed in [58]). Its over-expression in EW could be related to the higher rate of celldivision occurring in spring time.
The MW cluster contained 23 co-regulated proteins(Fig. 7B, cluster C3), most of which were also significantlydifferentially expressed between JW and MW as revealed byt-tests. As opposed to differentiating xylem formed by young
cambium, MW differentiating xylem is characterized bylarge cells with thick cell wall and lower microfibril angle,high cellulose content, and lower lignin content [4]. MW isalso characterized by higher LW and lower CW content per-centage. We hypothesized that the molecular mechanisms,as revealed by the protein overexpressed in MW, contributedto the delay of programmed cell death (PCD), therefore pro-longating cell wall deposition, resulting in the higher wooddensity characteristic of MW. Our results suggest that fourmechanisms, DNA reparation, cell detoxication, proteolysisregulation, and reduction of cytoplasm acidosis contribute toprolonged cell life.
In plants, PCD is involved in the terminal differentiationof xylem vessels [59]. DNA degradation has been reported asone of the key events associated with tracheary element dif-ferentiation [60]. In this study, a low abundant protein(spot #425) corresponding to a DNA-damage repair/tolera-tion protein DTR 102 was found to be overexpressed in MW,supporting our hypothesis.
In addition, several proteins of the MWcluster belonged tothe defense category, particularly involved in detoxicationmechanisms, also likely to contribute to the delay of PCD.These proteins included a superoxide dismutase [Cu-Zn](#545), one glutahione S-transferase (#499), and one glutathi-one peroxidase (#519). One germin-like protein (GLP, #497)categorized in the “unclassified” category was also present inthis cluster. Germins/GLP have been reported as proteinsinvolved in defense mechanisms (with either oxalate oxidaseor extracellular [Mn] superoxide dismutase activities [61, 62]).Interestingly, spot #497 was tightly linked with spot #545,reinforcing the putative role of this GLP in oxidative stressdefense. Kim and Tripplett [63] recently reported that a cottonfiber germin-like protein (GhGLP1) exhibited a maximalexpression with stages of maximal cotton fiber elongation,suggesting that some GLP may be important for cell wallexpansion. All together, these results suggest an importantrole of GLP in MW differentiation. The accumulation of ade-nine phosphoribosyl transferase (APT) (spot #503), a proteinimplicated in salvage of adenine to AMP, in MW differentiat-ing cells could also be interpreted as a defense mechanismagainst adenine, a toxic compound for the cell.
3748 J.-M. Gion et al. Proteomics 2005, 5, 3731–3751
Figure 6. Cuttings from 2-D gels showing different types of behavior in protein accumulation between JW vs. MW, EW vs. LW, and OW vs.CW.
Ubiquitin-dependent proteolysis plays a crucial role dur-ing the development in all organisms. Especially in plants, ithas been shown that phytohormone action depended on theubiquitin-proteasome pathway [64]. Recently, Paux et al. [17]showed the importance of auxin signaling through ubiqui-tin-dependent proteolysis, during wood formation. Ourfindings, i.e., up-regulation of a GTP-binding protein (Ras-related protein ARA-3, #517) and a proteasome protein(#491) in MW forming tissue, highlight the importance ofprotein degradation in the regulation of MW differentiation.
According to Drew [65], metabolic changes under anoxiahelp maintain cell survival by generating ATP anaerobicallyand minimizing the cytoplasmic acidosis associated with celldeath. In anaerobic treatment of maize seedlings, 20 anaero-bic proteins (ANP), which accounted for more then 70% oftotal translation, were selectively synthesized [66]. Most ofthese ANP corresponded to enzymes of glycolysis or sugar-phosphate metabolism [67]. In this study, proteins of thecarbohydrate and primary metabolisms category were alsofound to be up-regulated in MW forming tissue, namely afructokinase (#355), an alcohol dehydrogenase (#305), aphosphoglucomutase (#84), and NADP-dependent malicenzyme (spot #109). Besides these mechanisms, the accu-mulation of a vacuolar ATP synthase (spot #340; H1-ATPase(V-ATPase)) in MW corroborates the hypothesis of cyto-plasmic acidosis reduction during MW formation.
The CW cluster contained four proteins (Fig. 7B, clus-ter C4), including a structural enzyme of the flavonoid bio-synthetic pathway, leucoanthocyanidin dioxygenase (antho-cyanidin synthase). This enzyme catalyzes the reaction from
the colorless leucoanthocyanidin compound to the antho-cyanidin pigment responsible of the dark red or purple colorin plant tissues [68]. CW is also characterized by a reddishcolor (Fig. 1D). Although the molecular basis for this colorhas not been established, it has been suggested that it couldbe attributed to the polymerization of coniferaldehyde [69].Our finding suggests that the typical color of CW formingtissue also results from the biosynthesis of flavonoids.
The OW or “normal” wood (NW) cluster contained eightproteins (Fig. 7B, cluster C5) down-regulated in CW. CW ishighly lignified with more p-hydroxyphenyl subunits, andcontains less cellulose than NW. Microfibril angle of the cel-lulose fibers in the S2 layer of the cell wall is high, tracheidlength is reduced, the cross-sectional profile becomes iso-diametric, and intercellular spaces become larger comparedto NW [5]. We hypothesized that the molecular mechanismsdetermining cell shape and cell size [70], as revealed by someof the proteins underexpressed in CW (actin, profilin,nucleoside diphosphate kinase), are disturbed in gravi-stimulated tissue, leading to the typical cell phenotypeobserved in CW. Actin (spot #300) filaments are responsiblefor many aspects of cell behavior, including cell division,movement, and expansion. Profilin (PFN, #549) is a ubiqui-tous actin monomer-binding protein involved in the organi-zation of the cytoskeleton of eukaryotes, including higherplants. It is thought to regulate actin polymerization in re-sponse to extracellular signals. In cotton, it was observed thatone PFN-like protein was activated during the fiber elonga-tion period [71]. In Arabidopsis, it was observed that PFNplays an important role in cell elongation, cell shape main-
Figure 7. Hierarchical clustering analysis of the 175 identified and the 40 unknown proteins and example of clus-ters. Samples correspond to mature (M), juvenile (J), early (E), late (L), compression (C) and opposite (O) wood.Left panel (A): clustering of the whole dataset, right panels (B): C1/SAM-S cluster, C2/LW cluster, C3/MW cluster,C4/CW cluster, C5/normal wood cluster.
tenance, and polarized growth of root hair [72]. Cells of Ara-bidopsis plants expressing antisense PFN were shorter, andmore isodiametric, compared to wild-type. These resultsstrongly suggest that the specific shape of CW tracheids canresult (at least in part) from the down-regulation of actin andPFN in wood forming tissue. By comparing coleoptilelengths, nucleoside diphosphate kinase (NDK, #554) en-zyme activities, and cell size in non-transformants and anti-NDP kinase rice plants, Pan et al. [73] found that the cellelongation process was predominantly inhibited in epi-dermal cells of coleoptiles in antisense plants. This resultsuggests that the reduction of NDK accumulation couldcontribute to the shorter cells characteristic of CW. The threeother known function proteins of the normal wood cluster(glycine-rich RNA-binding protein #551, proteasome sub-
unit alpha type 4 #477, 40S ribosomal protein S12, #552)indicate that proteins of the genes and proteins expressioncategory were underexpressed in CW forming tissue.
3.6 A database to store and query the maritime pine
wood proteome
We recently described a complete web-based application“PROTICdb” (http://moulon.inra.fr/,bioinfo/PROTICdb,[74]), dedicated to the storage, query, and analysis of plantproteomic data. Maritime pine proteomes of differentiatingxylem, corresponding to different developmental stages andtreatments, have been stored with this application. We havedeveloped a new website to make these data publicly avail-able (http://cbib1.cbib.u-bordeaux2.fr/Protic/Protic/home/
3750 J.-M. Gion et al. Proteomics 2005, 5, 3731–3751
index.php). As a first level of information, the ‘Plants’, ‘Pro-tocols’, and ‘Protein identification’ hyperlinks providerespectively details about (i) the plant material and theexperimental conditions used, the organs sampled, the pro-tein quantity loaded on IPG strips; (ii) the protocols used forprotein extraction, first and second dimension electrophore-sis, staining, and image digitalization; and (iii) identifiedspots, including the query databases, number of matchingpeptides, protein coverage by the matching peptides, acces-sion number, matching species, assignment, theoretical andobserved pI and Mr (Da), MS techniques, and the MS plat-form where the spots have been processed.
A Java applet is used to access diverse information di-rectly on 2-DE images. 2-DE images can be downloaded byselecting their own ID and press the ‘load’ button. Up to fourimages can be visualized at the same time. By selecting the‘detected’ or ‘identification’ mode, all detected spots or allidentified spots are displayed with blue or red crosses,respectively. By Moving the mouse over a cross, a first level ofinformation is displayed, i.e., spot ID for any gel or referencespot ID for the master gel, Mr, pI and annotation (as inTable 1). More information (spot relationships, identificationand quantification) can be retrieved by left clicking on themouse. In respect to the identified spots, SEQUEST data(including peptide sequences) and links to nucleotide orprotein databases are also provided.
4 Concluding remarks
In this report, we have identified for the first time a highnumber (175) of known function proteins expressed in thewood forming tissue in a forest tree species. Identificationsuccess rate of MS/MS was high, over 70%, resulting mainlyfrom the use of pine EST. This is to be compared to the 16%success rate of MADLI-TOF MS. It is concluded that thecombined analysis of MS/MS spectra and EST sequences,provides an efficient and accurate protein identificationmethod for pine proteome analysis. A comparison betweenprotein and mRNA levels showed that at least 30% of theproteins were correlated with their corresponding mRNAlevels. This also means that for the majority of the proteinstheir level could not be predicted from transcript accumula-tion. This result demonstrates that a proteomic approach iscertainly a relevant approach for tracking genes involved inwood formation and determining wood quality.
The reference map represents the most comprehensivewood proteome projects to date and provides a good basis forfuture proteomic comparisons. Approximately 20 samples ofdifferentiating xylem taken along, (1) a gradient of gravi-stimulated xylem tissue, with samples collected on treesbended for few hours to few months, (2) a seasonal gradient,with samples taken every 15 days during the growing season(i.e., from April to August), and (3) a cambial age gradientwith samples taken every 4 m along the bole of an adult tree,have been collected, and are being analyzed by 2-DE com-
bined with LC ESI-MS/MS. This new dataset should shednew light onto the protein machinery involved in wood for-mation.
We thank anonymous reviewers for helpful comments on themanuscript. This research was supported by grants from the Eu-ropean Union (GEMINI project no. QLK-5-CT-1999-00942)and France (DERF no. 01.40.40/99; Région Aquitaineno. 20000307007, and INRA “Lignome”). JP was supported byfellowship SFRH/BD/3129/2000 from FCT/MCT Portugal.
5 References
[1] Zobel, B. J., Van Buijtenen, J. P., Wood Variation: Its Causesand Control, Springer-Verlag Berlin 1989.
[2] Cornelius, J., Can. J. For. Res. 1984, 24, 372–379.
[3] Pot, D., Chantre, G., Rozenberg, P., Rodrigues, J. C., Lloyd, J.G. et al., Ann. For. Sci. 2002, 59, 563–575.
[4] Zobel, B. J., Sprague, J. R., Juvenile Wood in Forest Trees,Springer-Verlag Berlin Heidelberg New York 1998.
[5] Timell, T. E., Compression Wood in Gymnosperms,Springer-Verlag Heidelberg 1986.