Salt stress induces the formation of a novel type of ‘pressure wood’ in two Populus species


Author for correspondence:
Andrea Polle
Tel: +49 551393480


  • Salinity causes osmotic stress and limits biomass production of plants. The goal of this study was to investigate mechanisms underlying hydraulic adaptation to salinity.
  • Anatomical, ecophysiological and transcriptional responses to salinity were investigated in the xylem of a salt-sensitive (Populus × canescens) and a salt-tolerant species (Populus euphratica).
  • Moderate salt stress, which suppressed but did not abolish photosynthesis and radial growth in P. × canescens, resulted in hydraulic adaptation by increased vessel frequencies and decreased vessel lumina. Transcript abundances of a suite of genes (FLA, COB-like, BAM, XET, etc.) previously shown to be activated during tension wood formation, were collectively suppressed in developing xylem, whereas those for stress and defense-related genes increased. A subset of cell wall-related genes was also suppressed in salt-exposed P. euphratica, although this species largely excluded sodium and showed no anatomical alterations. Salt exposure influenced cell wall composition involving increases in the lignin : carbohydrate ratio in both species.
  • In conclusion, hydraulic stress adaptation involves cell wall modifications reciprocal to tension wood formation that result in the formation of a novel type of reaction wood in upright stems named ‘pressure wood’. Our data suggest that transcriptional co-regulation of a core set of genes determines reaction wood composition.


Soil salinity is a worldwide problem. Salt-affected soils cover a total area of 397 million ha, nearly 3% of the total land mass (FAO, 2006). Of this area, 77 million ha comprise agricultural land. The land fraction lost for farming because of soil salinity will increase in the coming years as a result of inappropriate irrigation and effects of global warming (Kundzewicz et al., 2007). Research on salt-tolerance mechanisms in plants has been intensive, but mainly focused on agricultural crops or the model plant Arabidopsis (Amtmann & Sanders, 1999; Munns & Tester, 2008). To date, wood is gaining importance as a renewable resource. Sustainable woody biomass production on marginal or degraded soils in areas afflicted by salt or drought is desirable because this may contribute to solving the conflict of land use for either bioenergy or food production. For tree improvement toward increased stress tolerance, in-depth understanding of tree adaptation to salinity is required. Basic cellular defense mechanisms such as activation of the salt overly sensitive (SOS) pathway and antioxidative systems to prevent cellular injury are similar in tree and herbaceous species, but trees, which persist for many years on salt-affected sites, need additional protective measures to maintain growth and wood production (Chen & Polle, 2010).

A hallmark of woody lifestyle is the formation of a tall stem that supplies the crown with water and mineral nutrients. Osmotic stresses such as salinity and drought diminish radial growth and change xylem anatomy (Chen & Polle, 2010; Fischer & Polle, 2010). Ecophysiological studies in woody species along a gradient from humid to arid conditions showed that vessel diameters are inversely related to vessel numbers (Sperry et al., 2008; Teichmann et al., 2008). Woody species share a basic construction principle for drought adaptation that results in cell wall reinforcement by increasing wall thickness relative to lumen area while avoiding excess loss in water conductance by increasing vessel numbers. Recent studies with P. × canescens showed that this relationship also holds for the composition of the hydraulic system within one genotype (Arend & Fromm, 2007; Kundzewicz et al., 2007; Beniwal et al., 2010) and points to environmental sensing that enables flexible adjustment of xylem anatomy to osmotic conditions.

The molecular mechanisms leading to adaptive modifications in xylem anatomy of trees exposed to high salinity are not yet clear. Growth reductions may be the result of decreases in photosynthetic activity, leading to a lack of photosynthates (Kozlowski, 1997; Escalante-Perez et al., 2009) or decreases in the K : Na ratio, leading to a lack of potassium (Langer et al., 2002; Escalante-Perez et al., 2009) or changes in hormone concentrations, especially reductions in auxin and increases in ethylene concentrations affecting xylem anatomy (Junghans et al., 2004, 2006; Teichmann et al., 2008). It is unknown if adaptation of the hydraulic system to salt stress results in modification of chemical wood composition.

The aim of the present study was to link transcriptional responses to salt stress with anatomical and chemical changes in the xylem of the secondary growth zone. We compared nonstressed and salt-stressed plants of the salt-sensitive P. × canescens with the salt-tolerant Populus euphratica. We measured transcriptional profiles of developing xylem tissue, wood anatomy and composition, element content, osmotic potential, net CO2 assimilation rates and phloem sugar contents to disentangle the influence of carbohydrate supply, ion imbalances and osmotic disturbance on the hydraulic system.

Materials and Methods

Plant material, growth conditions and stress treatment

Plantlets from P. × canescens (P. alba × P. tremula) clone INRA717 1-B4 (Lepléet al., 1992) and P. euphratica clone B2 from the Ein Avdat valley in Israel (Broschéet al., 2005) were multiplied by in vitro micropropagation (Rutledge & Douglas, 1988) and kept in aerated hydroponic culture using Long Ashton nutrient solution (Hewitt & Smith, 1975), which was changed weekly. Plants were grown in a climatized chamber at 26°C, with a relative air humidity of 60% and a 16 : 8 h, light : dark cycle with photosynthetic active radiation (PAR) of 150 μmol m−2 s−1 (fluorescent lamps: L58W/25 and 58W/840, Osram, Munich, Germany; and TLD 58W/840 Philips, Amsterdam, the Netherlands). After 8 wk plants of each species were divided into three groups, of which one received 25 mM NaCl in the nutrient solution for 2 wk. Subsequently the salt-acclimated group was exposed for 2 wk to 100 mM NaCl and the second group to 25 mM NaCl. The third group served as an unstressed control. Each group consisted of 24 individual plants, and all 144 plants were harvested at the end of the 12 wk period. Plant height and stem diameter 2 cm above the root neck (digital caliper) were recorded regularly.

Harvest of plant material

Mature leaves were sampled from the middle third of the plant, that is, from 0.2 to 0.4 m for P. × canescens and from 0.3 to 0.6 m for P. euphratica. Stem samples were taken from a height of 0.05 m above the root–stem junction. Whole roots were also harvested. The developing xylem was defined as all extraxylary tissue obtained by scraping the surface of the xylem with a razor after removal of the bark (as described by Teichmann et al., 2008). Samples were directly frozen in liquid nitrogen and kept at −80°C. For anatomical analysis, samples were fixed in FAE (2% formaldehyde, 5% acetic acid, 63% ethanol). For DW, osmolyte content, Fourier transform infrared (FTIR) spectroscopy and element analysis, samples were dried in an oven (Memmert, Schwabach, Germany) at 70°C for 7 d, or at 105°C for 5 d (osmolyte content).

Anatomy and histochemistry

Cross-sections (20 μm thick) were obtained using a sliding microtome. Cuttings were stained for 6 min with astra blue (0.1% astra blue, w/v), washed and stained for 2 min with safranine (0.1% safranine, w/v), then washed and mounted on glass slides. Sections were viewed under a light microscope (Axioskop, Zeiss) using ×250 magnification. Photographs were taken with a digital camera (Axiocam, Zeiss). Morphometric measurements (cell lumina and cell number) were carried out in newly formed wood tissue in the outer 100 μm of the xylem of stem cross-sections using ImageJ (Abramoff et al., 2004). The predicted hydraulic conductivity was calculated as ∑r4 mm−2.

The pH in the developing xylem was determined by soaking fresh debarked stem segments for 2 min in 0.02% toluylene red solution. The stem segments were dried at room temperature and photographed with a DSLR camera (Pentax *ist DL2). CMYK color values were digitally measured in ImageJ (Abramoff et al., 2004) and calibrated with a pH series. The pH series was prepared by adding 0.2 ml 0.2% toluylene red solution to 1.8 ml of water whose pH had been adjusted by NaOH and HCl. pH values were determined with a pH meter (pH meter pH526, WTW, Weilheim, Germany).

Element distribution by EDX-SEM

After freeze-drying, stem cross-sections were coated with chromium and examined on a scanning electron microscope (AMR 1200, Leitz), equipped with an energy dispersive x-ray (EDX) microanalysis system (KEVEX 4000). Element-specific X-ray spectra were obtained from the developing xylem using a reduced scan raster area at ×500 magnification. Relative element concentrations were expressed as peak : background values from 24 recorded spectra.

Osmolyte content and element analysis

Osmolyte content was measured in samples of leaves, stem, developing xylem and roots. Dried samples were milled (Retsch, Haan, Germany). Fifty micrograms of powder were incubated in 2 ml reaction tubes with 1.5 ml H2O at 55°C overnight. Samples were centrifuged at 100 g for 25 min. Fifty microliters of the supernatant were analyzed in a cryoscopic osmometer (Osmomat 030, Gonotec, Berlin, Germany). Using dry and fresh mass measurements, osmolyte concentrations were calculated. Osmotic potential was estimated using the van’t Hoff law Ψ = −RTc, where Ψ is the osmotic potential in Pa, R = 8.314 JK−1 mol−1, T the absolute temperature in K, and c the solute concentration in mol l−1.

For element analysis, tissue powder was pressure-extracted in HNO3 (Heinrichs et al., 1986). Elements were measured by inductively coupled plasma-optical emission spectrometry (ICP-OES; Spectroflame, SPECTRO Analytical Instruments, Kleve, Germany).

Photosynthetic gas exchange

Gas exchange rates were measured using a portable gas exchange fluorescence system (GFS-3000 Walz, Effeltrich, Germany). Leaves (numbers 9 or 10 below the apex) were dark-adapted for 30 min, subsequently exposed for 30 min to light (1000 μmol PAR m−2 s−1 and 30°C leaf temperature) and then gas exchange parameters were averaged for 5 min.

Carbohydrate content of phloem exudates by HPLC

Phloem exudates of bark pieces were collected as described in Rennenberg et al. (1996). Phloem exudate (1 ml) was mixed with 20 mg of polyvinylpolypyrrolidone (PVPP), shaken continuously for 1 h at 4°C, and centrifuged at 12 000 g for 10 min at 4°C. The supernatant was diluted with one volume of deionized water. An aliquot of 100 μl of the diluted supernatant was injected into a high-performance liquid chromatography (HPLC) system (Dionex DX 500, Dionex, Idstein, Germany). Carbohydrates were separated on a CarboPac PA1 separation column (250 × 4 mm; Dionex, Idstein, Germany) with 56 mM NaOH as an eluent (flow rate = 1 ml min−1). Eluted carbohydrates were detected in an amperometric cell equipped with a gold electrode (Dionex DX 500). Identification and quantification of individual carbohydrates which eluted 8–16 min after injection was done with external standards of sucrose, fructose, and glucose. Peak areas were determined with the PeakNet software (version 5.1, Dionex, Idstein, Germany).

Analysis of wood composition by FTIR-ATR spectroscopy

Fourier transform infrared spectroscopy–attenuated total reflection (FTIR–ATR) spectra of developing xylem were recorded with an FTIR spectrometer (Equinox 55, Bruker Optics, Ettlingen, Germany) with a deuterium trigylcine sulfate detector and an attached ATR unit (DuraSamplIR, SensIR Europe, Warrington, UK) at a resolution of 4 cm−1 in the range from 600 to 4000 cm−1.

The bark was peeled off of wood samples before drying so that the uppermost layer of the dried wood consisted of developing xylem. After drying (at 70°C for 7 d) and acclimatization to the measuring room for 24 h, the wood samples were pressed against the diamond crystal of the ATR device; uniform pressure application was ensured using a torque knob. Individual analyses consisted of 32 scans which were averaged to give one spectrum. Each sample was analysed five times in different places, and the five spectra were averaged again, resulting in one mean spectrum per sample. Background scanning and correction were carried out regularly after 10–15 min.

Mean spectra for individual plants were processed using spectroscopy software (OPUS version 6.5, Bruker). A cluster analysis was conducted for the range 1750–1200 cm−1 after calculation of first derivatives with nine smoothing points and vector normalization. Compilation of a dendrogram was done by implementing Ward’s algorithm (Ward, 1963).

RNA extraction

Frozen tissue of developing xylem was ground in a ball mill in liquid nitrogen (Retsch). Tissue powder (500 mg) was used for RNA extraction after Chang et al. (1993) with minor modifications: the extraction buffer contained 2%β-mercaptoethanol and no spermindine. RNA was purified (RNeasy Mini Kit, Qiagen). Total RNA yield and purity were determined spectrophotometrically (BioPhotometer, Eppendorf, Hamburg, Germany) at A260 and A280. RNA integrity was assessed on an Agilent 2100 Bioanalyzer (Agilent, Santa Clara, CA, USA) at the Microarray Facility (Tübingen, Germany).

Microarray analysis and functional annotation

For the control and 100 mM NaCl treatment of both poplar species, three biological replicates were analyzed on a GeneChip® Poplar Genome Array (Affymetrix, Santa Clara, CA, USA). Synthesis of one-cycle cDNA and biotin-labeled cRNA, fragmenting of cRNA, hybridization to the Poplar Genome Array, washing, staining and scanning were performed as stated by Affymetrix (GeneChip® Expression Analysis Technical Manual) at the Microarray Facility, Tübingen. Raw and normalized data are available at the ArrayExpress-database (EMBL:E-MEXP-2031).

To obtain an up-to-date annotation of the microarray probe sets for further analysis, including current Gene Ontology (GO) terms, we performed several steps of BLAST searches (Altschul et al., 1990). The BLAST+ executable from the National Center for Biotechnology Information (NCBI; was used to perform a nucleotide BLAST (BLASTn) of the Affymetrix GeneChip Poplar Genome Array target sequences against the Phytozome Populus trichocarpa v2.0 transcript database. Gene identifiers (GIs) of the of best hits were retrieved by applying best-hit overhang and best-hit score parameters of 0.1 and 0.05, respectively, as proposed by NCBI in the BLAST Command Line Applications User Manual, and an e-value cutoff of 1E–5. AGI IDs of the closest Arabidopsis homologs were determined by a translated nucleotide BLAST (BLASTx) of the coding sequences of the best P. trichocarpa hits against the Arabidopsis thaliana protein sequence dataset of the representative gene models, applying the same best-hit parameters but a slightly more stringent e-value cutoff of 1E–10 to account for cross-species dissimilarities. Annotation details were taken from the latest release of The Arabidopsis Information Resource (TAIR) genome, TAIR10 (November 2010, and the GO TAIR gene association file (release May 2011,

Statistical analysis of the raw signal intensity data of the Affymetrix GeneChips was conducted using the following functions from packages released by the bioconductor project (Gentleman et al., 2004), implemented in R ( Background correction, quantile normalization and summarization of the Affymetrix CEL output files were computed using the rma algorithm from the affy package (Irizarry et al., 2003), resulting in a raw list of normalized probe set values. Probe sets referring to genes that were not expressed in the developing xylem were removed by calculating Affimetrix’s MAS 5.0 change calls using the mas5calls function from the affy package; only probe sets with three ‘present’ calls in one of the two treatments were regarded. Also, probe sets for which either no closest Arabidopsis homolog was found or no functional annotation was available in the TAIR10 genome release were removed.

Statistical testing for differentially expressed genes and control of the false discovery rate (FDR) was performed on this filtered list of normalized probe set values with a significance analysis of microarrays (SAM) using the ‘sam’ function from the ‘siggenes’ package (Tusher et al., 2001). In cases where multiple probe sets corresponded to one gene, duplicates were removed from the list of significant genes by using the Phytozome P. trichocarpa GI as a unique identifier; probe sets with the most significant P-value were kept for further analysis.

GO term enrichment analysis and digital northern analysis

For statistical analysis of overrepresented GO terms, a GO term enrichment analysis was conducted using ‘The Ontologizer’ (Bauer et al., 2008). A GO file for Populus was adapted by linking the Phytozome P. trichocarpa v2.0 GIs of gene models represented on the Affymetrix GeneChip to the GO identifier of their closest Arabidopsis matches as listed in the GO TAIR gene association file. The gene ‘population’ was defined as the list of filtered probe sets used for the SAM with the Phytozome GI as an unique identifier. ‘Study sets’ were lists of genes that showed either significantly increased or decreased transcript abundances. As parameter settings, parent–child intersection analysis with Bonferroni correction was used. Graphical outputs were also produced with ‘The Ontologizer’ using the term-for-term approach with Bonferroni correction. The lists of genes in the ‘study sets’ were analyzed in PopGenie ( using the tool DigitalNorthern (Sjödin et al., 2009).

Phylogenetic analysis

For the phylogenetic analysis of the fasciclin-like arabinogalactan-protein (FLA) genes, sequences for open reading frames for Populus and Arabidopsis were obtained from the Phytozome P. trichocarpa v2.0 database and from TAIR. Open reading frames were translated into protein code by GeneDoc (Nicholas et al., 1997) and the resulting amino acid sequences were aligned using Clustal W version 2.0 (Larkin et al., 2007). The unrooted tree was generated using the TreeView program (Page, 1996).

Validation of microarrays by quantitative real-time polymerase chain reaction (qRT-PCR)

Total RNA was treated with DNAse (Turbo DNA-free kit, Ambion, Austin, TX) and transcribed to cDNA with a RevertAid™ First Strand cDNA Synthesis Kit (MBI Fermentas, St Leon-Rot, Germany). qRT-PCR was performed on an iCycler (Bio-Rad, Hercules, CA, USA) using ABsolute qPCR SYBR Green Fluorescein Mix (ABgene, Surrey, UK). Primer design for the qRT-PCR was performed with the Oligo Explorer and suitable primers were tested for similar melting temperature (Tm), primer dimers and primer loops by Oligo Analyzer (both Gene Link, Hawthorne, NY, USA, qRT-PCR output was analyzed using the MyiQ software (Bio-Rad). Statistical analysis was conducted using the Pair Wise Fixed Reallocation Randomisation Test (Pfaffl, 2001) implemented in Excel with the Relative Expression Software Tool (REST) – 384 (Pfaffl et al., 2002). qRT-PCR confirmed microarray data (for tested genes, primers and results, see Supporting Information Fig. S1).


Osmotic gradient in and ecophysiological responses of P. × canescens and P. euphratica under salt stress

In nonstressed P. × canescens and P. euphratica the difference in the osmotic potential between roots and leaves was approx. Δ = 0.75 MPa (Fig. 1a,b). When the plants were stepwise acclimated to initially 25 mM NaCl and then to 100 mM NaCl, the osmotic potential of all tissues decreased in P. × canescens, whereas no significant changes were found in P. euphratica (Fig. 1a,b). The strongest decrease in osmotic potential occurred in developing xylem of P. × canescens and was the result of massive Na accumulation (Fig. 1c). Na accumulation was moderate in P. euphratica (Fig. 1d). Previous studies have shown that P. euphratica readjusts its internal water balance after exposure to excess salinity within 48 h by a combination of Na accumulation and decreases in K, Ca and sugar concentrations (Ottow et al., 2005; Brinker et al., 2010). Closer inspection of the ion distribution in developing xylem by EDX-SEM revealed that Na accumulation resulted in an approx. fourfold K decrease in P. × canescens, whereas the reduction in P. euphratica was less pronounced (1.5-fold, see Fig. S2).

Figure 1.

Osmotic potentials and sodium concentrations in Populus × canescens and Populus euphratica tissues in response to increasing NaCl concentrations in the nutrient solution. Plants were harvested under control conditions (white), after 2 wk of salt stress with 25 mM NaCl (hatched), and after an additional 2 wk of salt stress with 100 mM NaCl (black). *, osmotic potential of the nutrient solution with 25 mM NaCl; **, osmotic potential of the nutrient solution with 100 mM NaCl. The osmotic potential of the nutrient solution without added NaCl was −0.01 MPa. Osmotic potentials and salt concentrations were calculated based on the water content of the respective tissues. Bars indicate means ± SE (= 6 (osmolytes), = 3 (Na), = 5 (Na in developing (dev.) xylem). Different lower-case letters above bars indicate significant differences at P ≤ 0.05 obtained by ANOVA followed by a multiple range test (Tukey’s honestly significant difference (HSD)).

Salt exposure led to reduced radial growth in P. × canescens, but not in P. euphratica (Table 1). Likewise, only P. × canescens showed stress symptoms such as leaf senescence (not shown) as well as reductions but not complete abolishment of net CO2 assimilation (Table 1). Despite the decrease in net CO2 assimilation, sugar concentrations in phloem sap of P. × canescens increased under salt stress (Table 1). In P. euphratica neither net CO2 assimilation nor phloem sap carbohydrate concentrations were affected by salt exposure (Table 1). In the developing xylem of 100 mM NaCl-stressed P. × canescens or P. euphratica, no changes in these carbohydrates were found in comparison with controls (not shown).

Table 1.   Plant performance under salt stress
SpeciesSalt treatmentNet CO2 assimilation (μmol m−2 s−1)Phloem soluble sugars
NaCl (mM)Radial growth* (μm d−1)Glucose (μmol g−1 FW)Fructose (μmol g−1 FW)Sucrose (μmol g−1 FW)
  1. n.a., not available.

  2. Data are shown for poplars acclimated for 2 wk to 25 mM NaCl or additionally exposed for A further 2 wk to 100 mM NaCl in hydroponic solution and unstressed controls. Data are means of = 6 (± SE). Different letters indicate significant differences at < 0.05. *Radial growth was determined during the last 2 wk before harvest.

Populus × canescens047.6 (± 7.1) c4.58 (± 0.56) b9.08 (± 1.45) a2.80 (± 0.48) a7.07 (± 1.35) a
2537.9 (± 5.9) bn.a.21.17 (± 4.38) b4.70 (± 0.61) b18.39 (± 2.25) b
10013.1 (± 7.1) a1.46 (± 0.28) a18.43 (± 1.30) b3.36 (± 0.26) ab21.96 (± 0.87) b
Populus euphratica015.8 (± 3.1) a6.22 (± 1.20) b13.22 (± 2.23) a2.96 (± 0.31) a9.24 (± 0.74) a
2510.9 (± 3.9) an.a.14.00 (± 2.15) a3.45 (± 0.55) a9.96 (± 1.63) a
10010.5 (± 4.9) a5.50 (± 0.23) b18.92 (± 5.47) a4.72 (± 1.76) a12.61 (± 2.07) a

Salt stress affects wood anatomy and composition of P. × canescens and P. euphratica

Because salt stress caused a decrease in the osmotic pressure in the xylem of P. × canescens but not in P. euphratica, we anticipated changes in wood anatomy only in P. × canescens. As expected, the number of vessels per area increased and the mean vessel diameter decreased from c. 31 to 23 μm, resulting in decreasing vessel lumina with increasing salt stress in P. × canescens, whereas none of these traits was affected in P. euphratica (Fig. 2a,b). The predicted hydraulic conductivity computed from vessel frequency and individual sizes of vessel lumina remained constant in both species (Fig. 2c). The changes in frequency and lumen area were limited to the vessels; numbers of fibers and ray cells did not vary in response to salinity (P. × canescens/P. euphratica: number of fibers, 5090/5270 mm−2; number of ray cells, 585/ 580 mm−2)

Figure 2.

Changes in xylem anatomy of Populus × canescens and Populus euphratica in response to salt stress. Plants were harvested under control conditions (white), after 2 wk of salt stress with 25 mM NaCl (hatched), and after an additional 2 wk of salt stress with 100 mM NaCl (black). (a) Number of vessels per cross-sectional area in mm2; (b) mean lumen area of individual vessels in μm2; (c) predicted conductivity, given as the sum of the fourth power radii per area of xylem (∑rmm−2). Bars indicate means ± SE (= 5). Different lower-case letters above bars indicate significant differences at  0.05 obtained by ANOVA followed by a multiple range test (Tukey’s honestly significant difference (HSD)).

According to the acid-growth theory, cell expansion requires the acidification of the cell walls; protons displace Ca+ ions from their linkage positions between pectin molecules, thereby loosening the cell wall (Maggio et al., 2006). We therefore expected that the observed increases in osmotic pressure and growth reductions in salt-stressed P. × canescens would be accompanied by alkalinization of the cell wall. However, the opposite was observed (Fig. 3). Calibration of the observed changes in color intensity indicated decreases by 0.9 and 0.6 pH units in the developing xylem of salt-exposed P. × canescens and P. euphratica, respectively (Figs S3, S4).

Figure 3.

pH changes in developing xylem of Populus × canescens and Populus euphratica in response to salt treatment. Stems were debarked and immediately soaked in toluylene red (0.02%). For estimation of the changes in pH value, a calibration series of differing pH values was stained with toluylene red (0.02%). For details see Figs S2, S3.

To obtain a chemical fingerprint of the molecular wood composition, FTIR–ATR spectra were measured directly on the surface of the newly formed xylem of P. × canescens and P. euphratica under control conditions and salt stress, respectively (Fig. 4). The spectra showed pronounced differences for the developing xylem of both species and revealed salinity-induced changes in major wood compounds, that is, hemicelluloses (peak 1) and protein (peak 2), lignin (peak 8) and cellulose (peak 6, Fig. 4a). Notably, wood composition of P. euphratica, whose anatomy and osmotic potential were not significantly affected, was also influenced by exposure to excess salinity (Fig. 4a). Undirected multivariate analysis showed that the spectra of individual plants formed two main clusters according to species, which were divided into subclusters according to salt treatment (Fig. 4b). The ratios of characteristic wavenumbers showed changes in the lignin : hemicellulose ratio in both poplar species and a significant increase in the lignin : cellulose ratio in P. × canescens (Table S1). These results indicate that the chemical wood composition, in particular cell wall carbohydrate composition, is responsive to salt exposure and that these changes are uncoupled from or at least precede anatomical changes.

Figure 4.

Chemical fingerprints of developing xylem from Populus. (a) Mean baseline-corrected Fourier transform infrared (FTIR) spectra of developing xylem in the region of wavenumbers from 1200 to 1750 cm−1 of Populus × canescens (red control, orange salt stressed) and Populus euphratica (blue, control; cyan, salt-stressed). (b) Cluster analysis of the fingerprint regions of the FTIR spectra. Salt stress was induced by 2 wk of acclimation to 25 mM NaCl and an additional 2 wk of exposure to 100 mM NaCl in the nutrient solution. Spectra are means of five biological replicates. First derivates of spectra after baseline correction were used for cluster analysis using Ward’s algorithm and correlation coefficients as distance metrics. Peak numbers in (a) refer to the following molecular components: 1, 1738cm−1 = C=O stretch in unconjugated ketones, carbonyls and ester groups in xylans (hemicellulose); 2, 1650cm−1 = absorbed O–H and conjugated C=O of proteins; 3, 1596cm−1 = aromatic skeletal vibrations in lignin plus C=O stretch; 4, 1505 cm−1 = aromatic skeletal vibrations in lignin plus C=O stretch; 5, 1462 cm−1 = C–H deformation; asymmetric in -CH3 and -CH2-; lignin and carbohydrates; 6, 1425 cm−1 = aromatic skeletal vibrations combined with C–H plane deformation; lignin and carbohydrates; 7, 1375 cm−1 = C–H deformation in cellulose and hemicellulose; 8, 1330cm−1 = syringyl ring plus guaiacyl ring condensed; 9, 1235 cm−1 = syringyl nuclei deformation combined with deformation of cellulose. Wavenumber assignments were taken from Rana et al. (2008) and Zhou et al. (2011).

Transcriptome analysis of developing xylem in P. × canescens and P. euphratica in response to salt stress reveals distinct wood and stress-related regulons

To gain insight into the molecular events associated with the anatomical and compositional changes in poplar wood under salt stress, transcriptional profiles of developing xylem were analyzed on whole-genome microarrays. Genes with significantly changed transcript abundances (differentially expressed genes, DEGs) in response to salt in comparison with untreated controls were determined for each species. The response of the two species differed strongly. In P. × canescens, a total of 382 genes showed significantly different transcript abundances, but only 39 DEGs were detected in P. euphratica. A total of 210 DEGs in P. × canescens showed increased transcript abundance and 172 showed decreased transcript abundance, while in P. euphratica only one gene was up-regulated (Tables S2–S4). Nine DEGs were common to both species, all involved in cell wall metabolism (Table 2).

Table 2.   Salt-responsive genes that are common to Populus × canescens (P × c) and Populus euphratica (P. eu) developing xylem and involved in cell wall metabolism
Affymetrix probe set IDPhytozome v2.0 GIAGIDescriptionFold change in
P × cPeu
  1. Xylem was harvested from unstressed controls and from trees acclimated for 2 wk to 25 mM and exposed for a further 2 wk to 100 mM NaCl. Complete lists of genes with significantly changed transcript abundances are shown in Tables S2–S4.

  2. nsc, not significantly changed.

PtpAffx.4337.2.S1_atPOPTR_0011s04960At4g27450Aluminum-induced protein with YGL and LRDR motifs0.110.15
PtpAffx.3761.4.S1_atPOPTR_0008s20870At5g18670Beta-amylase 30.440.21
Ptp.3978.1.S1_atPOPTR_0006s13190At2g37130Peroxidase superfamily protein0.210.15
Ptp.6380.1.S1_atPOPTR_0004s11620At5g15630COBRA-like extracellular glycosyl-phosphatidyl inositol-anchored protein family0.050.14
Ptp.3083.1.S1_s_atPOPTR_0009s01730At5g60490FASCICLIN-like arabinogalactan-protein 120.040.20
Ptp.371.2.S1_x_atPOPTR_0013s14780At5g60490FASCICLIN-like arabinogalactan-protein 120.140.19
PtpAffx.142854.1.S1_s_atPOPTR_0012s02220At5g60490FASCICLIN-like arabinogalactan-protein 120.230.21
PtpAffx.162047.1.S1_s_atPOPTR_0009s01740At5g60490FASCICLIN-like arabinogalactan-protein 120.070.18
PtpAffx.33081.1.S1_atPOPTR_0004s22030At5g60490FASCICLIN-like arabinogalactan-protein 120.060.23
Ptp.3500.1.S1_atPOPTR_0012s02210At5g60490FASCICLIN-like arabinogalactan-protein 120.27nsc
Ptp.371.1.S1_atPOPTR_0013s14760At5g60490FASCICLIN-like arabinogalactan-protein 120.18nsc
PtpAffx.249.461.S1_s_atPOPTR_0001s32800At5g60490FASCICLIN-like arabinogalactan-protein 120.22nsc
Ptp.2503.1.S1_atPOPTR_0001s09330At5g12250Beta-6 tubulin0.41nsc
PtpAffx.202369.1.S1_atPOPTR_0002s19830At1g02640Beta-xylosidase 20.27nsc
Ptp.617.1.S1_atPOPTR_0004s18840At4g38400Expansin-like A20.37nsc
Ptp.4742.2.S1_a_atPOPTR_0019s04700At2g39050Hydroxyproline-rich glycoprotein family protein0.55nsc
Ptp.4810.1.A1_s_atPOPTR_0003s17450At1g04680Pectin lyase-like superfamily protein0.12nsc
PtpAffx.211973.1.S1_atPOPTR_0014s17620At5g48900Pectin lyase-like superfamily protein0.30nsc
PtpAffx.76062.1.A1_s_atPOPTR_0010s16180At1g48100Pectin lyase-like superfamily protein0.17nsc
Ptp.7635.1.S1_atPOPTR_0014s12180At2g47550Plant invertase/pectin methylesterase inhibitor superfamily0.16nsc
PtpAffx.140560.1.A1_atPOPTR_0005s06270At5g09760Plant invertase/pectin methylesterase inhibitor superfamily0.30nsc
PtpAffx.112384.1.S1_atPOPTR_0018s10300At4g25810Xyloglucan endotransglycosylase 615.1nsc
Ptp.5229.1.S1_s_atPOPTR_0007s14570At5g65730Xyloglucan endotransglucosylase/hydrolase 6nsc0.07
PtpAffx.107571.1.A1_s_atPOPTR_0001s16270At3g14310Pectin methylesterase 3nsc0.08

To detect fundamental patterns underlying the transcriptional changes in P. × canescens, overrepresented GO categories were determined for DEGs with increased and decreased transcript abundances, respectively (Table 3). GO terms overrepresented among up-regulated genes in P. × canescens belonged to the following main categories: ‘oxidation-reduction process’, ‘multi-organism process’, ‘secondary metabolism’, and ‘antioxidant activity’. A closer inspection of the enriched GO term hierarchy revealed many defense-related categories (immune response, response to fungus, bacterium, toxin, etc.) and, in addition, also GO terms for ‘secondary metabolism’ with genes involved in flavonoid and phenylpropanoid metabolism as well as developmental GO terms (e.g. pollen tube guidance) with transcription factors known to regulate cell division (Fig. S5). Overall, these observations support the idea that cell division was still active and radial growth not abolished in response to salt (Table 1).

Table 3.   Enriched Gene Ontology (GO) terms in the lists of differentially expressed genes in developing xylem of Populus × canescens and Populus euphratica under salt stress
GO IDGene ontologyNumber of terms inAdjusted P-valueDirection of change and GO term name
Total populationStudy set
  1. GO terms after Bonferroni correction in mode ‘parent–child intersection’ are shown.

  2. BP, biological process; MF, molecular function; CC, cellular component.

P. × canescens 12604210 Genes with increased transcript abundance
GO:0055114BP900410.000 213Oxidation-reduction process
GO:0050896BP3034860.000 241Response to stimulus
GO:0051704BP606290.000 582Multi-organism process
GO:0019825MF5070.009 975Oxygen binding
GO:0040011BP1140.023 101Locomotion
GO:0019748BP229150.045 592Secondary metabolic process
GO:0016209MF6370.075 101Antioxidant activity
P. × canescens 12604171 Genes with decreased transcript abundance
GO:0071944CC1941510.000 286Cell periphery
GO:0071554BP194120.004 142Cell wall organization or biogenesis
GO:0042044BP3050.055 002Fluid transport
P. euphratica 11 74738 Genes with decreased transcript abundance
GO:0071554BP17860.034 34Cell wall organization or biogenesis

Among genes with decreased transcript abundance in P. × canescens, the GO terms ‘cell periphery’, ‘cell wall organization or biogenesis’ and ‘fluid transport’ were overrepresented. Remarkably, the only GO term overrepresented in P. euphratica was also ‘cell wall organization or biogenesis’ (Table 3). In P. × canescens, the genes annotated by this GO term coded for two plant invertase/pectin methylesterase inhibitors, an expansin-like protein A2, a COBRA-like protein and eight FLAs, and showed the largest overlap (six of 12) with P. euphratica (Table 2). Phylogenetic analysis of the related FLA genes from P. trichocarpa showed that salt-responsive FLAs in P. × canescens and P. euphratica formed a distinct group within subgroup A (subgroup with glycosylphosphatidylinositol (GPI) membrane anchor) with highest similarity with AtFLA12 (locus At5g60490) (Fig. 5). The overabundance of FLAs was curious because members of this subfamily were massively induced during tension wood formation in poplar (Andersson-Gunnerås et al., 2006). Digital northern analysis employing the DEGs with decreased transcript abundance with poplar libraries available in POPGenie (Sjödin et al., 2009) revealed clustering of 24 and 36% of the salt-repressed DEGs in the cambium and the tension wood libraries, respectively (Fig. 6). By contrast, digital northern analysis of DEGs with increased transcript abundance did not reveal a clear pattern (not shown). This indicates that a suite of genes activated during tension wood formation is collectively suppressed during wood formation under salt stress.

Figure 5.

Phylogenetic tree of fasciclin-like arabinogalactan proteins (FLAs). Eight closest homologs of Populus trichocarpa FLAs to the differentially expressed genes in salt-stressed poplars and the 21 FLA genes in Arabidopsis are shown. Phytozome gene identifiers and fold-changes under salt stress are listed in Table 3. AGI locus identifiers for Arabidopsis FLAs were obtained from Johnson et al. (2003).

Figure 6.

Digital northern analysis of genes with decreased transcript abundance after salt stress using expressed sequence tag (EST) libraries for different tissues. The heat map shows the frequencies of genes within EST libraries available at PopGenIE ( that are coregulated with genes with decreased transcript abundance in developing xylem of Populus × canescens under salt stress.


Salt-induced changes in cell wall composition precede stress responses in the xylem

A remarkable result of this study was the clear distinction of salt responses related to xylem modification and stress adaptation. Transporters putatively involved in salt exclusion were constituively up-regulated in P. euphratica compared with salt-sensitive poplars (Ding et al., 2010; Janz et al., 2010). Here we show that this trait led to exclusion of salt from the above-ground transport path, in particular from the developing xylem. By contrast, highest Na accumulation occurred in the developing xylem of P. × canescens, consequently requiring strong activation of defenses to prevent injury. In the overrepresented GO term categories for genes with increased transcript abundance typical representatives for osmotic stress as well as for antioxidative defenses were found, for example, genes coding for proteins involved in osmotic adjustment (LEA protein, osmotin precurser, myo-inositol-1-phosphate synthase, trehalose phosphatase), detoxification (glutathione-S-transferases, superoxide dismutase, alcohol dehydrogenase) and in signal perception and transduction (ABA-induced protein phosphatase 2C (PP2C), WRKY and MYB transcription factors, the salt responsive homeobox leucin zipper transcription factor HB-7, various serine/threonine protein kinases, 9-cis-epoxycarotenoid dioxygenase involved in ABA biosynthesis) (Table S2). The activation of these genes has also been reported in previous salt screens in roots or leaves of poplar as well as in other plant species (Broschéet al., 2005; Gong et al., 2005; Ottow et al., 2005; Ma et al., 2006; Teichmann et al., 2008; Brinker et al., 2010; Ding et al., 2010; Qiu et al., 2011). Since our plants were acclimated to high salinity and still showed wood formation, this suite of genes must be important to sustain a new degree of cellular homeostasis under stress.

Two GO categories enriched under salt stress in the developing xylem of P. × canescens deserve specific attention: ‘multicellular process’ and ‘secondary metabolism’. The first category contains a collection of transcription factors involved in cell division and vascular development (e.g. homologs to the Arabidopsis genes KNAT-3 (Truernit et al., 2006) recently identified in Juglans nigra during heartwood formation (Huang et al., 2009), MONOPTEROS (Ohashi-Ito & Fukuda, 2010), ANAC087 and ANAC082 (Guan & Nothnagel, 2004), and ARF16 (Ding et al., 2010) required for meristem differentiation), thus supporting our cytological and morphometric analyses, which revealed ongoing growth. Significant accumulation of carbohydrates along the transport route (this study) as well as in roots of osmotically stressed poplars (Luo et al., 2009a,b; Galvez et al., 2011) may, however, point to disturbed phloem unloading. Because massive carbohydrate depletion in the developing xylem was not observed, the reasons for the growth reductions remain unclear. Decreased carbon flux to fuel cell wall formation and inhibited vessel expansion by hydraulic stress may have diminished radial growth. It is possible that decreased potassium availability plays a role in this respect (Langer et al., 2002; Wind et al., 2004), but this requires further analyses because it is not clear if elevated concentrations of Na can replace K.

The second category ‘secondary metabolism’ was not identified in P. euphratica. However, this species exhibits constitutive activation of secondary metabolism, even in the absence of salt stress (Janz et al., 2010). Quantitative trait locus (QTL) analyses in poplar progenies revealed that genes for secondary metabolites correlate with drought tolerance (Street et al., 2006). Increased concentrations of flavonoids and phenolics may protect plants against environmental cues because they act as antioxidants preventing cellular injury (Rice-Evans et al., 1997). The pathways for flavonoid production and precursors for lignin (flavonoid 3 hydroxylase, flavonol synthase, flavonol-O-methyl transferase, cytochrome P 450 family protein, cinnamoyl CoA reductase, peroxidases) were stimulated in response to salt stress in the developing xylem of P. × canescens. Substantial increases in lignin relative to cellulose were not detected, suggesting that the production of phenol-based compounds served to increase the antioxidant capacity. It should also be recalled that the FTIR analyses were conducted on the surface of young, not fully differentiated xylem, where lignification was not yet accomplished and therefore the question of whether salt stress led to increased lignin concentrations cannot be conclusively answered yet. However, both species showed changes in cell wall carbohydrates. P. euphratica displayed only moderate salt accumulation, which resulted neither in detectable osmotic stress nor in anatomical changes. Our results indicate that processes involved in cell wall modification are responsive to subtle alterations in hydraulic signals or the ion balance and can clearly be distinguished from cellular defense reactions.

Changes in xylem anatomy and cell wall composition adapt poplar to hydraulic stress through generation of ‘pressure wood’

Osmotic stress causes strong negative xylem pressures, which can lead to cavitation and subsequently to conduit collapse in the xylem (Hacke & Sperry, 2001). Because the integrity of the water transport system is essential for survival under stressful conditions, plants across a vast range of biomes increase their wall strength by decreasing the ratio of vessel lumen to cell wall thickness in response to water limitation (Hacke et al., 2001). This phenomenon is also known as the hydraulic safety principle because single cavitation incidents pose a lower threat when a high number of conduits are available (Zimmermann, 1983; Tyree & Ewers, 1991). Notably, this hydraulic adaptation strategy can also be found when comparing salt-adapted mangrove species (many small vessels) with nonmangrove species (fewer, larger vessels) of the same genus (Janssonius, 1950). Flexible adjustment of vessel lumina to changes in osmotic conditions by salt or drought has also been reported for poplars (Junghans et al., 2006; Arend & Fromm, 2007; Beniwal et al., 2010; Schreiber et al., 2011; this study) and results in formation of false year rings under field conditions (Liphschitz & Waisel, 1970).

A surprising result of our study was that developing xylem undergoing hydraulic adaptation exhibited a coordinated reduction of gene transcripts identified in tension wood formation of poplar (Andersson-Gunnerås et al., 2006). Tension wood is formed on the upper side of the stem in response to gravitational stimuli and is characterized by low lignin content and strong accumulation of cellulose in the fiber lumina (Timell, 1986). Our results show that hydraulic adaptation apparently involves the opposite regulation of a suite of genes required for tension wood formation, namely FLAs, COBRA-like (homolog to Arabidopsis COBL4), and genes encoding xyloglucan endo transglycolyase, pectin methylesterase, pectin lyase, expansin, xylosidase, and amylase. Most of these genes are members of large gene families in poplars. With the exception of the xylosidase gene, members of these gene families were up-regulated in tension wood (Andersson-Gunnerås et al., 2006) and down-regulated here. Conversely, gene expression of the phenylpropanoid pathway was increased here and significantly decreased in tension wood (Andersson-Gunnerås et al., 2006). In tension wood, cellulose biosynthetic genes (CES) were increased but suppression of cellulose synthase genes was not found here, although cellulose formation was impaired. Genetic analysis showed that cellulose deposition is under the control of AtCOB4 in Arabidopsis and the ortholog BC1 in rice (Brown et al., 2005; Sato et al., 2010). The ortholog of these genes, COBRA-like, was suppressed in salt-stressed poplar, pointing to suppression of cellulose biosynthesis by transcriptional regulation. Furthermore, the β-amylase 3 gene identified here is a homolog of ArabidopsisCYT1, whose failure in knockout mutants resulted in significant decreases in cellulose and accumulation of defense compounds (Lukowitz et al., 2001). Therefore, transcriptional repression of poplar COBRA-like and β-amylase 3 may have caused alterations in cell wall carbohydrate composition.

The only gene with increased transcript abundance in the young xylem of salt-exposed P. euphratica encoded a 1-aminocyclopropane-1-carboxylic acid oxidase (ACC oxidase), a gene required for ethylene production. In poplar, ethylene affects the fiber : vessel ratio (Junghans et al., 2004) and fiber extensibility (Qin et al., 2007). Recently, it was shown that ethylene controls tension wood formation (Love et al., 2009). In compression wood of conifers, a strong correlation of increasing lignin : cellulose ratios and ACC oxidase was reported (Plomion et al., 2000), which might point to a role of ethylene in wood formation under pressure stress.

Fasciclin-like arabinogalactan protein genes (or FLA genes or FLAs) are an extremely expanded gene family with several hundred members in poplar (Johnson et al., 2003; Lafarguette et al., 2004; Andersson-Gunnerås et al., 2006). The FLAs identified as down-regulated in this study form a group of paralogs which have no true orthologs in Arabidopsis; this observation has also been made for the FLAs massively up-regulated during the formation of tension wood (Andersson-Gunnerås et al., 2006). FLAs, which form a distinct subgroup of arabinogalactan-proteins (AGPs), are thought to be involved in various processes of xylem differentiation such as cell–cell signaling, cell division, adhesion, and microfibril orientation and may function as wall integrity sensors affecting downstream signal transduction (Humphrey et al., 2007; Seifert & Blaukopf, 2010). In compression wood of loblolly pine, an AGP with GPI anchor (AGP5) showed decreased expression (Zhang et al., 2000). The Arabidopsis Salt Overly Sensitive 5 (SOS5) protein, a synonym for FLA4 (Swiss-Prot:Q9SNC3), is required for controlled cell expansion (Shi et al., 2003). Phylogenetic comparisons showed that the group A of FLA genes was conserved in monocots and dicots and was specific to stems (MacMillan et al., 2010). Double mutants of Atfla12/fla11 showed reduced tensile strength and stiffness, thus rendering cell walls more brittle (MacMillan et al., 2010). Atfla11 and Atfla12/fla11 mutants also contained lower amounts of cellulose and higher lignin concentrations (Persson et al., 2005; MacMillan et al., 2010). In this study, P. × canescens and P. euphratica showed only a few common reactions towards salt stress; one striking commonality was the suppression of five specific FLAs and changes in chemical wood composition resembling those of Arabidopsis fla mutants.

We conclude that salinity caused formation of a novel type of ‘pressure’ wood, presumably as a result of transcriptional coregulation of the same set of genes involved in tension wood biosynthesis, but in the opposite direction. At present, it is mostly unknown how trees sense and signal environmental cues to modify cell walls (Seifert & Blaukopf, 2010). The identification of a putative master relay leading to coordinated stimulation or inhibition of genes for reaction wood formation will be an important research target for future studies.

With regard to practical implications, higher wood density as the result of hydraulic adaptation affects wood biomechanical properties and its technological utilization (Irvine & Grace, 1997; Holtta et al., 2002). Increased lignin : cellulose ratios, as found in compression wood of conifers, have negative effects on pulpability and render such trees less suitable for papermaking (Yeh et al., 2006). However, this wood may be useful as fuelwood for local populations and, therefore, knowledge of the mechanistic basis of wood formed under environmental constraints will help to accelerate the selection of trees that can grow on sodic soils.


We are grateful to S. Wolfart and C. Kettner for excellent technical assistance, to the German Science foundation (DFG) for funding Poplar Research Group Germany (FOR496, Po362/12, Po362/13, Re515/20, Schn653/4) and to the European Commission for supporting parts of this study within the Seventh Framework Program for Research, Project Energypoplar (FP7-211917).