Differential gene expression in senescing leaves of two silver birch genotypes in response to elevated CO2 and tropospheric ozone

Authors


S. Kontunen-Soppela. Fax: +358 13 2513590; e-mail: sari.kontunen-soppela@joensuu.fi

ABSTRACT

Long-term effects of elevated CO2 and O3 concentrations on gene expression in silver birch (Betula pendula Roth) leaves were studied during the end of the growing season. Two birch genotypes, clones 4 and 80, with different ozone growth responses, were exposed to 2× ambient CO2 and/or O3 in open-top chambers (OTCs). Microarray analyses were performed after 2 years of exposure, and the transcriptional profiles were compared to key physiological characteristics during leaf senescence. There were genotypic differences in the responses to CO2 and O3. Clone 80 exhibited greater transcriptional response and capacity to alter metabolism, resulting in better stress tolerance. The gene expression patterns of birch leaves indicated contrasting responses of senescence-related genes to elevated CO2 and O3. Elevated CO2 delayed leaf senescence and reduced associated transcriptional changes, whereas elevated O3 advanced leaf senescence because of increased oxidative stress. The combined treatment demonstrated that elevated CO2 only temporarily alleviated the negative effects of O3. Gene expression data alone were insufficient to explain the O3 response in birch, and additional physiological and biochemical data were required to understand the true O3 sensitivity of these clones.

Abbreviations
CC

chamber control

DHPPG

3,4′-dihydroxypropiophenone 3-β-d-glucoside

EC

elevated CO2

EC + EO

combined elevated CO2 + elevated O3

EO

elevated O3

EST

expressed sequence tag

FDR

false discovery rate

NPQ

non-photochemical quenching of chlorophyll fluorescence

PSII

photosystem II

RbcS

Rubisco small subunit gene

INTRODUCTION

Over the last few decades, human activity has accelerated atmospheric changes. CO2 emissions have increased by about 80% between 1970 and 2004 (IPCC 2007). In parallel, increasing NOx and hydrocarbon emissions cause higher tropospheric ozone (O3) concentrations (Percy & Ferretti 2004). For plants, these changes have contrasting effects. Rising CO2 concentration increases photosynthesis, which in C3 plants is limited by atmospheric [CO2] (e.g. Drake, Jacob & Gonzalez-Meler 2000; Woodward 2002). However, the CO2 responses in plants show great variation because of differences between species and in other environmental factors (Drake et al. 2000; Woodward 2002; Nowak, Ellsworth & Smith 2004). Forest trees are generally the most responsive to elevated [CO2] (Ainsworth & Long 2005). O3 is considered the most harmful tropospheric pollutant to plants (Matyssek & Sandermann 2003) because it affects nearly all photosynthetic processes (Long & Naidu 2002) and causes damage to cell structures (Oksanen et al. 2004). Therefore, O3 accelerates climate change by reducing the ability of plants to accumulate carbon via photosynthesis to biomass production. It has recently been suggested that the role of O3 in decreasing the CO2 uptake is underestimated in the current carbon sequestration models (Sitch et al. 2007; Wittig, Ainsworth & Long 2007).

The impact of O3 on plants is determined by [O3], O3 stomatal flux, exposure length and the plant defence capabilities such as the antioxidative capacity of the plant (Baier et al. 2005). O3 sensitivity is also dependent on developmental factors such as the age of plant (Nunn et al. 2005; Karnosky et al. 2007; Wittig et al. 2007), leaf ontogeny (Oksanen et al. 2005) and the timing of exposure and tree size (Oksanen 2003). Chronic O3 stress during several growing seasons with fluctuating [O3] and adaptation to the stress requires the adjustment of metabolism. The adjustment results in changes in resource allocation, in favour of repair, defence and compensations mechanisms (Dizengremel 2001). The metabolic modifications in chronic O3 are seen as altered contents of phenolic compounds (Yamaji et al. 2003; Peltonen, Vapaavuori & Julkunen-Tiitto 2005; Kontunen-Soppela et al. 2007) and chemical changes in tree leaves (Oksanen et al. 2005; Valkama, Koricheva & Oksanen 2007).

The simultaneous exposure of plants to elevated [CO2] and [O3] has resulted in both decreased growth and compensation for the O3-induced growth losses by CO2 (Rebbeck & Scherzer 2002; Kubiske et al. 2006). In silver birch, only small growth losses were found in elevated [O3], and the growth responses in a combined elevated [CO2] and [O3] treatment were similar to responses in elevated [CO2] (Riikonen et al. 2004). These two greenhouse gases also have opposite effects on leaf senescence. While the onset of leaf abscission (Riikonen et al. 2004) and leaf yellowing (Syrjäläet al. unpublished) are delayed in elevated [CO2], they are accelerated by O3 promoting early leaf abscission (Karnosky et al. 1996; Nowak et al. 2004; Riikonen et al. 2004).

The effects of elevated [CO2] on the transcriptome have been studied, for example, in soy bean (Ainsworth et al. 2006; Casteel et al. 2008), Arabidopsis (Li et al. 2006b) and in the tree Populus (Gupta et al. 2005; Taylor et al. 2005; Druart et al. 2006; Cseke et al. 2009). These studies show that responses to elevated [CO2] vary depending on species, genotypes and leaf ages. Both decreased (Li et al. 2006b) and increased (Gupta et al. 2005) accumulation of photosynthesis light-harvesting transcripts have been observed. Elevated [CO2] also alters the flux through various biosynthetic routes, and changes secondary metabolite composition. In Populus, the flavonoid biosynthesis route was activated (Druart et al. 2006). Expression of phenylpropanoid pathway genes decreased (Gupta et al. 2005), or alternatively decreased or increased depending on the Populus genotype (Cseke et al. 2009). Elevated [CO2] enhances cell expansion-related genes (Gupta et al. 2005; Taylor et al. 2005; Druart et al. 2006) and mediates changes in metabolism by inducing genes related to signalling and regulation (Ainsworth et al. 2006; Li et al. 2006b). In addition, candidate genes for determining plant adaptation to elevated [CO2] have been studied with Populus (Rae et al. 2006, 2007).

Acute O3 exposure of tree leaves elicits defence gene expression commonly induced by all kinds of oxidative stress (Olbrich et al. 2005; Rizzo et al. 2007), correspondingly to data on herbaceous plants (e.g. Tamaoki et al. 2003; Li et al. 2006a; Tosti et al. 2006). The changes in gene expression observed for chronic O3 exposure in mature trees are similar to, but often of smaller magnitude than in acute stress, and are influenced by other stresses in field experiments (Gupta et al. 2005; Jehnes et al. 2007; Paolacci et al. 2007). Furthermore, different O3-responsive mechanisms may function, depending on the level of O3 and on acute or chronic stress (Vahala et al. 2003).

Changes in gene expression caused by the combined treatment with elevated [CO2] and [O3] have so far been reported in only two studies. In soy bean, the combined treatment produced relatively few alterations in gene expression (Casteel et al. 2008). The combined treatment in Populus led to gene expression patterns different from either gas alone (Gupta et al. 2005).

In the current study, as a part of a larger climate change experiment (see e.g. Riikonen et al. 2004; Vapaavuori et al. 2009), we report genotypic differences in the gene expression patterns of birch (Betula pendula Roth) leaves caused by long-term exposure to elevated [CO2] and [O3]. Two birch clones, showing different sensitivity to O3 in terms of growth (Riikonen et al. 2004), were exposed to elevated [CO2] and [O3] alone and in combination, in open-top chambers (OTCs). This study focuses on leaf senescence phenomenon, because our previous phenological data showed differences in leaf abscission between genotypes, delayed abscission in elevated CO2 and a trend for accelerated senescence in elevated O3 (Riikonen et al. 2004). Leaf samples were harvested twice at the end of the second exposure summer, and gene expression patterns were analysed with DNA microarrays. The gene expression data are compared to key physiological parameters in order to understand mechanistic links.

MATERIALS AND METHODS

Experimental design

An OTC system in Suonenjoki Research Station, Finland (62°05′N, 27°00′E) was run during 1999–2001 in order to study the responses of silver birch (B. pendula Roth) trees to elevated [CO2] and [O3] alone and in combination. The experiment is described in detail in Vapaavuori et al. (2002). Two birch genotypes, clone 4 and clone 80 with different responses to O3, were included in the study. Clone 4 was more sensitive to O3 than clone 80 according to growth responses (Riikonen 2004). In addition, clone 80 was characterized as a physiologically more active genotype, which appeared, for example, as higher gas exchange and growth rates. The clones have a different origin: clone 4 was from Valkeakoski (61°08′N, 28°49′E), while clone 80 was from Eno (62°48′N, 30°05′E) (Mutikainen et al. 2000). Both clones were represented as four replicate trees in each treatment. The experiment included altogether 32 tree individuals, each growing in a separate OTC, assigned to the following treatments: CC, 2× background CO2 (EC), 2× background O3 (EO) and combined 2× background CO2 + 2× background O3 (EC + EO). The exposure and meteorological data are presented in more detail in Riikonen et al. (2005).

Samplings for the microarray analyses were done on 1 August and 6 September 2000 (day of the year 214 and 250). Ten short shoot leaves of each tree were collected in each sampling, and leaf discs (diameter: 2 cm) were punched, pooled, frozen immediately in liquid N2 and stored at −80 °C. To obtain a common reference sample for each clone (4 and 80) and time-point (August and September), the leaf discs of the four CC trees were pooled.

Photosynthesis, chlorophyll fluorescence, ribulose 1·5-bisphosphate carboxylase/oxygenase (Rubisco), soluble sugars and starch and leaf abscission

Gas exchange measurements were carried out in late July and late August on short shoot leaves from three different tree heights, using a portable gas exchange apparatus Li-6400 (Li-Cor Inc., Lincoln, NE, USA). The measurements were done at 20 °C, the growth conditions of CO2, saturating light intensity and at ambient relative humidity of the air. Chlorophyll fluorescence measurements were made with a portable pulse amplitude-modulated fluorometer (MINI-PAM, Walz, Effeltrich, Germany). More details from the measurements are described in Riikonen et al. (2004). Immediately after gas exchange measurements, the leaves were detached, and a 2 cm2 disc was frozen in liquid N2 for the determination of Rubisco, chlorophyll, soluble sugars and starch (Hansen & Møller 1975). In data analysis, the samples from the different tree heights were pooled and analysed as described in Riikonen et al. (2004). Leaf abscission was monitored by collecting the fallen leaves within the chambers weekly as described in Riikonen et al. (2004).

RNA extraction and amplification

RNA was isolated from birch leaves according to Chang, Puryear & Cairney (1993) with the following modifications: frozen leaf discs (200–300 mg) were homogenated first in liquid N2 with sand, and then in prewarmed (65 °C) extraction buffer where 1% Tween 80 (Fluka, Buchs, Switzerland) was added. The homogenates were incubated at 65 °C for 50 min to 2 h, shaken for 15 min and incubated at 65 °C for 15 min before the extraction was continued with chloroform:IAA and RNA precipitation with LiCl as in Chang et al. (1993). The isolated RNA (1 µg) was amplified with AminoAllyl MessageAmp aRNA Kit (Ambion, Austin, TX, USA) according to the manufacturer's instructions.

Hybridization on microarrays

A common reference design was used for the microarray analyses (Churchill 2002; Smyth 2005). All treatments within each clone and time-point were hybridized against the common reference obtained from the respective ambient control chambers. All hybridizations were done with dye swaps. The hybridization design of the experiment is shown in Fig. 1.

Figure 1.

Hybridization design of the experiment. Common reference for both time-points (August, September) was obtained by pooling the leaves of four individual trees in ambient control chambers. The common reference was hybridized against samples from four replicate trees in each treatment at each time-point. The treatments were CC, EC, EO and EC + EO. The design shown in the figure was repeated for both clones 4 and 80. All hybridizations were done with dye swaps.

The cDNA microarrays used in this study consisted of 8143 Populus euphratica ESTs representing approximately 6340 distinct genes, and an undetermined number of paralogues (Broschéet al. 2005). The ESTs were obtained from P. euphratica leaf, shoot and root control library, and several stress-related cDNA libraries of plants subjected to the following treatments: elevated CO2, different irradiance levels, drought stress, flooding stress, O3, cold and freezing, salt stress and cadmium stress. The array has previously been successfully applied for birch (Ruonala et al. 2006). The cDNA was spotted in triplicate on epoxy silane-coated Nexterion E-borosilicate glass (SCHOTT AG, Mainz, Germany), with the size of 75.6 × 25.0 mm, at the Finnish DNA Microarray Centre in Turku. The array design is available in ArrayExpress with the accession number A-MEXP-1042 (http://www.ebi.ac.uk/microarray-as/aer/entry).

The aRNA was labelled with Cy3 or Cy5 (Amersham Biosciences, Buckingshire, England) before the hybridization according to the instructions in the AminoAllyl MessageAmp aRNA Kit (Ambion). The slides were pre-hybridized in pre-hybridization buffer (2% BSA, 5× SSC, 0.1% SDS) for 30 min to 3 h at 65 °C. The arrays were hybridized in 50% formamide, 5×SSC, 0.1% SDS, 5×Denhardt's solution (Sigma, St Louis, MO, USA) and 10% Herring sperm (1 mg mL−1) (Sigma) for 16–18 h at 42 °C. The slides were then washed and scanned immediately after the hybridization with GenePix 4200AL scanner (Axon Instruments, Union City, CA, USA) at 635 and 532 nm.

Microarray data analysis

Images were analysed in GenePixPro 5.0 (Axon Instruments). Visually bad spots or areas on the array, small spots (diameter <50 µm) and low-intensity spots were marked with negative flags. Low-intensity spots were determined as spots where less than 55% of the pixels had intensity above the background +1SD in either channel.

The data from GenePixPro were analysed with the limma-package (linear models for microarray data) (Smyth 2004) in R (http://www.r-project.org/). The spot quality weights were set at 0.1 for all negative flags, obtained from the GenePixPro program, and at 1 for the rest of the spots. The background correction for the data was done with normexp-function (Ritchie et al. 2007), and the background corrected mean intensities were used for calculations. The average of triplicate spots on each slide was used for analyses. The data were normalized with the loess method. Linear models with moderated t-statistic (set at P < 0.005) were used to find genes that were differentially expressed (Smyth 2004, 2005).

Hierarchical clustering of the gene expression changes between control and treatments was done in R. Euclidean clustering with complete linkage method was used to cluster the gene expression data by treatment/sampling time/clone that was significantly changed in at least one of the clones or treatments or at a sampling time.

Gene annotation and gene ontology (GO) analysis

Annotations, as well as GO term annotation and the function-based analysis of genes that were differentially expressed, were performed using the software Blast2GO (Conesa et al. 2005; Götz et al. 2008). GO terms for each of the three main categories (biological process, molecular function and cellular component) were obtained from sequence similarity (E value 1e−5) of the EST sequences on the array. The application default parameters were used for the other analyses.

Statistical analysis of physiological data

The main effects and interactions of clone, CO2, O3 and sampling time on leaf physiological parameters, and concentrations of phenolic compounds were analysed by means of the linear mixed models analysis of variance (anova), using the replicate tree number as a random factor (SPSS 14.0 for Windows, SPPS Inc. 2005, Chicago, IL, USA). Where necessary, the data were ln transformed to achieve the normal distribution of residuals. The P values <0.05 are reported as significant.

RESULTS

Natural leaf senescence

Leaf senescence, monitored as the time-course of leaf abscission, started earlier in the elevated O3 treatment (EO), and was significantly delayed by the elevated [CO2] (P < 0.001) (Fig. 2). The difference between clones in leaf abscission was significant (P < 0.001). In September, there was a significant decrease in chlorophyll fluorescence parameters, the maximal photochemical efficiency of PSII (Fv/Fm), the actual photochemical efficiency of PSII (Fv′/Fm′) and the NPQ of leaves when compared with the August samples (Table 1). The starch concentration of leaves increased in September, but no changes were seen in the amount of chlorophyll, concentration of soluble sugars or Rubisco amount or activity (Fig. 3; Table 1). Leaf age also affected the concentrations of many phenolic compound groups (Table 2; Supporting Information Table S2).

Figure 2.

Effects of treatments on the timing of leaf abscission. Leaf abscission is presented as the percentage of total leaf area in clones 4 (a) and 80 (b) during the autumn 2000. The treatments were as in Fig. 1. Data show the mean values ± SE for four replicate samples. The samples for the microarray and phenolic compound analyses were taken at days 214 and 250. The measurements for physiological parameters were made at days 210 and 237.

Table 1.  The P values for the main effects of the time of sampling, elevated CO2 and O3 concentrations, clone and their interactions on physiological parameters of silver birch short shoot leaves
 TimeCloneCO2O3CO2 × O3Clone × O3Time × CO2Clone × CO2Clone × CO2 × O3Time × CO2 × O3Time × clone × O3
  1. Net photosynthesis (Pn, µmol m−2 s−1), ribulose 1·5-bisphosphate carboxylase/oxygenase (Rubisco) amount (Rub, g m−2), total activity of Rubisco (Rub tot act, µmol m−2 s−1), amount of chlorophyll (Chl, mg m−2), chlorophyll a/b ratio (Chl a/b), actual photochemical efficiency of PSII (Fv′/Fm′), the NPQ (NPQ (Fm – Fm′)/Fm′), maximal photochemical efficiency of PSII (Fv/Fm), concentrations of soluble sugars and starch (mg g−1 DW). Significant changes (P < 0.05) are shown in bold, and P values >0.05 and <0.1 are regarded as a trend; non-significant changes are left open. Arrows indicate increase (↑) or decrease (↓) compared to ambient control or August sampling (time).

Pn<0.001 <0.001   0.005    
Rub  <0.0010.004   0.011   
Rub tot act 0.085<0.001    0.058   
Chl 0.003 0.093↓0.059   0.039  
Chl a/b 0.0190.019     0.094  
Fv′/Fm<0.001 0.009        
NPQ<0.001 <0.0010.007 0.0680.04   0.040
Fv/Fm<0.001          
Soluble sugars   0.054↓     0.018 
Starch0.004 <0.001        
Figure 3.

Physiological parameters of clones 4 and 80 in the different treatments ± SD. Ribulose 1·5-bisphosphate carboxylase/oxygenase (Rubisco) amount, Rubisco activity, NPQ, chlorophyll amount and the soluble sugars and starch presented for clones 4 (left panel) and 80 (right panel) at days 210 and 237. Data are the means ± SE from four replicate trees. The treatments were as in Fig. 1.

Table 2.  The P values for the main effects of the time of sampling, clone, elevated O3 and CO2 concentrations, and their interactions on concentrations of phenolic compounds and compound groups in silver birch short shoot leaves
 TimeCloneCO2O3Time × cloneCO2 × O3Clone× CO2 × O3Clone × CO2Clone × O3Time× clone × CO2 × O3
  1. Data show the statistics for concentrations of phenolic compounds, except for total measured phenolics where total amount of phenolics were tested. LMWP: low-molecular-weight phenolics. Values shown in bold are significant at P < 0.05 level, P values >0.05 and <0.1 are regarded as a trend and the non-significant changes are left open. Arrows indicate increase (↑) or decrease(↓) compared to ambient control or August sampling (time).

DHPPG <0.000<0.0010.007 0.0170.067   
Pentagalloyl glucose<0.001     0.0440.0320.0490.057
Phenolic acids<0.001<0.001<0.001  0.070 0.036  
Flavone aglycons0.0130.0010.0190.089↑0.0590.073 0.017  
Flavonol glycosides0.002         
Catechin derivatives<0.001         
LMWP0.006         
Condensed tannins0.0030.009        
Total measured phenolics0.0020.006        

Responses to treatments

Elevated [CO2]

Gene expression changes caused by elevated CO2 (EC) were similar in both clones according to the hierarchical clustering (Fig. 4), and the number of genes (18) showing statistically significant changes in expression was the same in the September samples. In contrast, in August only, one EST was similar for both clones (Table 3; Supporting Information Table S1). In September, genes with decreased gene expression were mainly related to photosynthesis light reactions and protein synthesis (five ESTs of ribosomal genes). In clone 4 in September, decreased expression of enzymes related to secondary metabolism (flavonoid 3-hydroxylase, UDP-glucosyl-transferase, sinapyl alcohol dehydrogenase) was observed (Table 3; Supporting Information Table S1).

Figure 4.

Dendrogram showing dissimilarities in GE between treatments and the sampling times. Euclidean clustering with complete linkage method was used to cluster the treatment/sampling time/clone with gene expression data that were significantly changed in at least one of the clones (4 and 80) in one of the treatments at a sampling time (August and September). The treatments were as in Fig. 1. Height shows the distance between clusters in the dendrogram.

Table 3.  Selected genes showing significant differences in expression in leaves of silver birch grown under elevated CO2 , (EC) and O3 (EO) concentrations alone or in combination (EC+EO) compared with control ambient conditions
GenBank EST accession no.Gene functionECEOEC+EO
AugSepAugSepAugSep
804804804804804804
  1. Results are shown separately for the two genotypes (clones 80 and 4) and sampling times (August and September). Log fold changes (log2 value of fold change to ambient) shown in bold are significant (P < 0.005).

AJ776591Xylosidase−0.40−0.15−0.46−1.01−0.44−0.67−0.74−0.25−0.07−0.220.00−0.34
AJ775080Cytochrome b6 f complex subunit IV0.030.81−0.40−0.93−0.550.08−0.72−0.51−0.07−0.770.080.10
AJ777882ATP-citrate synthase−0.11−0.410.020.11−0.660.070.110.220.080.11−0.080.01
AJ780854Sinapyl alcohol dehydrogenase−0.28−0.62−0.11−0.36−0.20−0.27−0.02−0.09−0.280.00−0.28−0.16
AJ768767Photosystem II protein D1−0.100.101.03−0.12−0.57−0.01−0.040.330.170.050.030.39
AJ769778Beta-amylase0.02−0.03−0.56−0.51−0.220.02−0.17−0.160.04−0.39−0.06−0.19
AJ770490Photosystem I assembly protein ycf3−0.100.06−0.57−0.560.140.33−0.550.00−0.26−0.71−0.13−0.12
AJ769407phytocyanin−0.370.38−1.34−1.12−0.810.18−0.85−0.43−0.59−1.36−0.28−0.79
AJ771235senescence-associated protein−0.030.25−0.94−0.940.11−0.11−0.510.05−0.57−1.160.040.62
AJ768695early light-inducible protein−0.300.08−0.41−0.04−0.790.05−0.080.140.050.42−0.160.20
AJ769270NADH-plastoquinone oxidoreductase subunit 1−0.41−0.18−0.80−0.820.160.80−0.57−0.20−0.64−0.94−0.07−0.21
AJ769365NADH -plastoquinone oxidoreductase subunit j0.160.45−0.46−0.80−0.040.12−0.62−0.47−0.63−1.380.010.00
AJ767424NADH dehydrogenase subunit k0.010.35−0.81−1.04−0.050.43−0.82−0.38−1.26−1.82−0.17−0.03
AJ768613Cytochrome oxidase subunit I0.40−0.100.07−1.00−0.04−0.41−0.14−1.270.040.260.00−0.40
AJ767435Serine carboxypeptidase−0.040.20−0.56−0.600.060.540.140.12−0.75−1.09−0.080.01
AJ772359Thioredoxin-like protein−0.36−0.21−0.34−0.44−0.59−0.240.060.00−0.06−0.42−0.03−0.17
AJ76939430S ribosomal protein−0.01−0.040.020.02−0.50−0.11−0.19−0.22−0.14−0.06−0.140.08
AJ77602440s ribosomal protein s20−0.31−0.20−0.04−0.01−0.50−0.240.01−0.15−0.02−0.17−0.040.07
AJ77969360s ribosomal protein−0.60−0.21−0.45−0.49−0.71−0.16−0.100.040.02−0.08−0.16−0.33
AJ778942caffeoyl-CoA 3-o-methyltransferase0.03−0.25−0.240.03−0.54−0.020.010.160.010.23−0.160.01
AJ768301Carbonic anhydrase−0.42−0.120.04−0.07−0.74−0.09−0.07−0.21−0.13−0.140.22−0.03
AJ770887Chlorophyll a-b binding protein 6a−0.18−0.11−0.040.11−0.67−0.240.06−0.290.09−0.040.02−0.04
AJ772950Photosystem I protein−0.29−0.01−0.09−0.14−0.65−0.32−0.17−0.35−0.10−0.02−0.020.21
AJ778941Photosystem II subunit x−0.69−0.01−0.30−0.27−0.92−0.37−0.28−0.11−0.150.000.03−0.25
AJ770833Photosystem I reaction centre subunit psaK−0.63−0.17−0.25−0.33−0.81−0.47−0.24−0.04−0.07−0.270.14−0.15
AJ778183Photosystem II oxygen-evolving complex protein 3-like−0.230.02−0.26−0.17−0.60−0.51−0.31−0.240.04−0.070.38−0.26
AJ767466Chloroplast protease−0.19−0.07−0.27−0.18−0.54−0.03−0.120.07−0.04−0.25−0.150.09
AJ768869Mg protoporphyrin IX chelatase−0.15−0.04−0.070.04−0.53−0.16−0.14−0.02−0.030.050.16−0.01
AJ773162Chloroplast protein CP12−0.460.04−0.13−0.08−0.94−0.31−0.17−0.070.110.18−0.060.00
AJ774698Light-harvesting complex I protein lhca2−0.26−0.20−0.220.02−0.59−0.08−0.27−0.23−0.01−0.060.030.12
AJ779264Fructose-1,6-bisphosphatase−0.58−0.31−0.17−0.19−0.86−0.25−0.29−0.14−0.10−0.29−0.08−0.09
AJ779831Glutaredoxin−0.350.00−0.35−0.09−0.72−0.130.020.11−0.030.03−0.10−0.10
AJ780048Glyceraldehyde-3-phosphate dehydrogenase−0.30−0.31−0.09−0.30−0.69−0.37−0.08−0.030.210.13−0.07−0.02
AJ779238Leucoanthocyanidin reductase−0.17−0.070.00−0.08−0.62−0.06−0.110.060.130.03−0.08−0.03
AJ769142Lil3 protein−0.39−0.26−0.080.05−1.04−0.14−0.17−0.190.08−0.200.050.16
AJ773781Quinone oxidoreductase-like protein (AT1G23740)−0.26−0.200.040.11−0.66−0.07−0.040.13−0.030.000.010.02
AJ773103Threonine endopeptidase−0.20−0.16−0.090.28−0.55−0.150.02−0.04−0.030.19−0.060.06
AJ774130Ubiquitin ligase−0.04−0.21−0.22−0.08−0.50−0.030.110.030.140.310.010.14
AJ771714Alternative oxidase−0.030.11−0.06−0.100.00−0.57−0.23−0.16−0.040.01−0.02−0.04
AJ772258Protein phosphatase pp2a regulatory subunit0.120.130.010.000.05−0.510.07−0.16−0.05−0.36−0.130.11
AJ769602Cryptochrome 20.010.110.20−0.020.14−0.50−0.130.20−0.040.040.04−0.08
AJ773315Ubiquitin-like protein 5−0.26−0.13−0.29−0.20−0.15−0.010.240.55−0.09−0.04−0.04−0.02
AJ770033Ferritin, chloroplastic0.01−0.28−0.14−0.18−0.36−0.060.350.59−0.09−0.10−0.120.03
AJ77201812-Oxophytodienoate reductase0.00−0.13−0.05−0.050.100.130.210.600.050.06−0.150.00
AJ773599S-adenosyl-l-methionine synthetase−0.25−0.26−0.13−0.05−0.170.190.100.41−0.04−0.11−0.030.20
AJ780058Bark storage protein0.25−0.280.070.220.110.10−0.07−0.24−0.080.590.01−0.12
AJ773972Cinnamoyl reductase-like protein0.010.020.070.13−0.02−0.010.090.070.080.72−0.19−0.03
AJ780816Ribulose-phosphate 3-epimerase−0.74−0.33−0.37−0.37−0.350.06−0.190.16−0.81−0.89−0.160.31
AJ770203Rubisco activase−0.64−0.33−0.24−0.30−0.100.16−0.150.12−0.33−0.74−0.050.22

Photosynthetic activity (Pn, measured at growth [CO2]) and the actual photochemical efficiency of PSII (Fv′/Fm′) increased in EC (Table 1), and the NPQ decreased (Fig. 3; Table 1). Total activity and amount of Rubisco protein decreased in EC (Table 1; Fig. 3). The down-regulation of Rubisco in EC was greater in clone 80 than in clone 4. There were a few CO2 × clone and CO2 × sampling time interactions, indicating clonal and time-dependant variation in response to EC.

Elevated [CO2] had a significant impact on the concentration of phenolic acids that increased, and the concentrations of DHPPG and flavone aglycons that decreased (Table 2; Supporting Information Table S2). There were many clone × CO2 interactions in the concentrations of phenolics, showing a differential response of the studied clones to EC (Table 2).

Elevated [O3]

According to hierarchical clustering, the gene expression profile in the EO treatment in August was different from the other treatments (Fig. 4). Elevated [O3] had a greater effect on the gene expression of clone 80 than on clone 4 especially in August, when the number of genes with decreased expression was markedly higher in clone 80 (256 ESTs in clone 80, 81 ESTs in clone 4) (Supporting Information Table S1). Many genes down-regulated in EO were connected to general metabolism, including translation-related (24 ESTs) genes associated with photosynthesis (30 ESTs) and glycolysis (six ESTs). In August, increased expression in clone 80 included enzymes related to secondary metabolism (cinnamyl alcohol dehydrogenase), cell wall biosynthesis-related enzymes (e.g. rhamnose synthase) and many unknown ESTs (Supporting Information Table S1). In clone 4, three ESTs increasing in expression coded for NADH plastoquinone oxidoreductase. In September, only one EST was up-regulated in clone 80 in EO, while in clone 4 the ESTs with increased expression were related to stress.

EO also decreased the amounts of Rubisco, chlorophyll and concentrations of soluble sugars, and increased NPQ (Fig. 3; Table 1). Furthermore, elevated O3 increased the concentrations of DHPPG and flavone aglycons in the leaves (Table 2; Supporting Information Table S2).

Combined elevated [CO2] and [O3] treatment

In contrast to single EO and EC treatments, combined EC + EO exposure induced more gene changes in expression in clone 4 than in clone 80 in August (Supporting Information Table S1). The expression pattern of EC + EO in August resembled that of EC treatment in September especially in clone 4 (Fig. 4; Table 3). Expression of ESTs representing NADH plastoquinone oxidoreductase and ribosomal proteins, as well as senescence-associated protein and serine carboxypeptidase, was decreased in both clones in EC + EO in August, whereas in EC these changes were seen in September (Table 3). In September, the clustering analysis showed the similarity of EC + EO treatment to EO treatment (Fig. 4), although significant changes in gene expression were not similar in these treatments (Table 3).

Elevated [CO2] affected leaf abscission and the parameters related to photosynthesis similarly at both O3 levels. However, a significant CO2 × O3 × clone interaction revealed that the chlorophyll concentration was reduced under elevated [O3] only in ambient [CO2] in clone 4, while in clone 80 it was reduced also in combined elevated [CO2] and [O3] (Fig. 3; Table 1). Interaction effects of [CO2] and [O3] on phenolic concentrations were observed in DHPPG, phenolic acids and flavone aglycons (Table 2; Supporting Information Table S2).

DISCUSSION

Senescence phenomena differed between clones

Leaf senescence is an active process during which some metabolic pathways are turned on, whereas others are deactivated. Previous studies with birch in greenhouse conditions have shown that leaf senescence is associated with decreases in expression of photosynthesis and nutrient remobilization-related genes (Valjakka et al. 1999; Sillanpääet al. 2005). Growth cessation, bud set and leaf senescence are under photoperiodic control in birch (Li et al. 2003), and therefore the sampling in this study was timed to the period at the end of the growing season when leaf senescence normally starts. Decreased chlorophyll fluorescence indicated the beginning of leaf senescence at the time of sampling (Fig. 3).

The two genotypes studied differed in phenology. Clone 80, originating from a slightly northern location in Finland, is adapted to shorter growing period than clone 4 (Mutikainen et al. 2000), and is expected to senesce earlier. Our data show that senescence started at around the same time in the two clones, but proceeded faster in clone 80 (Fig. 2 and Riikonen et al. 2003). Riikonen et al. (2003) also suggested that clone 80 is more effective in the translocation of leaf nutrients than clone 4.

Elevated [CO2] decreases the expression of C metabolism and senescence-related genes

Samples for gene expression analysis were harvested towards the end of the growing season, and the leaves have thus been exposed to the treatments continuously since bud burst in early May. Accordingly, the leaves had reached a new state of homeostasis, and only a few genes showed altered expression in elevated [CO2]. In agreement with previous studies on mature aspen leaves, most changes were seen as decreased transcript levels (Gupta et al. 2005; Taylor et al. 2005; Cseke et al. 2009).

In the September samples, expression of a phosphatidylinositol/phosphatidylcholine transfer protein was strongly decreased in both clones. This protein is related to phospholipid-mediated signal transduction processes, and it may regulate lipid biosynthetic processes in plants (Li, Xie & Bankaitis 2000). Lower levels of transcripts related to senescence of birch leaves, such as β-amylase, metallothioneins (Bhalerao et al. 2003) and a senescence-related protein, indicate a slow-down in senescence processes. The delayed leaf senescence of trees in elevated [CO2], based on phenological and physiological measurements, has been reported earlier (Riikonen et al. 2004; Taylor et al. 2008), but this is the first study indicating alterations also in senescence-related gene expression.

The changes in gene expression under elevated [CO2] were supported by the physiological, chemical and structural changes in leaves, and the growth responses of the trees. Despite down-regulation of net photosynthesis when measured at ambient [CO2], the net photosynthesis increased at 720 ppm CO2 (Table 1). This caused accumulation of starch in the leaves of both clones (Table 1; see also Oksanen et al. 2005). The concentration of hemicellulose was higher under EC (Oksanen et al. 2005), which may be related to the decreased expression of xylosidase gene potentially involved in the hemicellulose metabolism of secondary cell wall (Goujon et al. 2003). The observed surplus of carbohydrates in EC was shunted to non-structural compounds, seen as higher concentration of starch (Riikonen et al. 2005); larger size of chloroplasts and starch grains (Oksanen et al. 2005); and accumulation of compounds of the phenylpropanoid pathway, such as phenolic acids (Oksanen et al. 2005; Peltonen et al. 2005).

There were some clonal differences in gene expression in response to EC. Genes related to secondary metabolism were decreasing in expression in clone 4, and accumulation of phenolic compounds was different between clones (Peltonen et al. 2005). Clonal variation in carbon partitioning to growth and secondary metabolism under elevated [CO2] were also observed in Populus (Cseke et al. 2009).

Ozone-induced changes in gene expression are dependent on clone

Many studies show that O3 sensitivity and gene expression profiles vary both between species and within ecotypes of the same species (Li et al. 2006a; Rizzo et al. 2007; Puckette, Tang & Mahalingam 2008). This study illustrates the same phenomenon with large differences in gene expression between the clones 4 and 80.

In early August, the majority of genes with altered expression had decreased transcript levels. Similar results have been previously reported in Arabidopsis where five times more genes were down- than up-regulated under chronic O3 exposure (Miyazaki et al. 2004). In clone 80, the EO treatment decreased the expression of many ribosomal genes which may be related to decreased protein synthesis in leaves. Many photosynthesis-related (Chl a/b binding, and PSI and II proteins, GAPDH, CP12) and carbon assimilation-related transcripts (RbcS, carbonic anhydrase) had reduced expression as previously shown in Populus (Gupta et al. 2005). The observed changes in gene expression were in line with the decrease in physiological parameters. The decline in photosynthetic light reactions was also seen as an increase in the NPQ, indicating increased energy dissipation as heat when the capacity to electron transfer decreases (Muller, Li & Niyogi 2001; Niyogi et al. 2005). The increased NPQ may be associated with the increased NADH plastoquinone oxidoreductase expression in EO in August. NADH plastoquinone oxidoreductase (ndh) is induced by natural senescence (Catal, Sabater & Guéra 1997) and O3 (Guéra et al. 2005). NDH is related to the acclimation of photosynthesis in changing environmental conditions, and suggested to protect PSI from photoinhibition (Rumeau, Peltier & Cournac 2007). However, the decreased expression of photosynthesis-related genes and biochemistry of photosynthesis could not be recorded as decreasing net photosynthesis (Riikonen et al. 2005), nor in growth or total biomass, which was very little affected (Riikonen et al. 2004).

In response to EO in September, clone 80 had few alterations in gene expression which included decreased expression of NADH plastoquinone oxidoreductase that has been associated with oxidative stress (Rumeau et al. 2007). In contrast, clone 4 had increased expression of genes associated with leaf ageing and/or oxidative stress including: ferritin (Murgia et al. 2007), 12-oxophytodienoate reductase (He et al. 2002) and ubiquitin (Wegener et al. 1997). The increased expression of 12-oxophytodienoate reductase points to the synthesis of jasmonic acid (Schaller 2001), one regulator of O3 responses in plants (see e.g. Baier et al. 2005). These findings are in accordance with the long-term O3 exposures of trembling aspen (Gupta et al. 2005) or beech (Jehnes et al. 2007), where the expression of oxidative stress-induced signalling and defence-related genes increased in leaves. Although oxidative stress responses in September were more marked in clone 4 in terms of gene expression, the physiological parameters measured indicated stronger oxidative stress in clone 80 under EO. The stomatal conductance and thus O3 uptake were greater in clone 80, but also the ascorbate concentration was higher, indicating superior scavenging of reactive oxygen species in clone 80 (Padu et al. 2005). The number of mitochondria and peroxisomes increased, and there was accumulation of H2O2 (Oksanen et al. 2005) and some phenolic compounds, such as chlorogenic acid, myricetin glycosides and flavone aglycons in clone 80 (see also Peltonen et al. 2005). Recently, Kontunen-Soppela et al. (2007) reported that a more O3-tolerant birch genotype showed a stronger shift in the leaf metabolome towards defence-related compounds, such as phenolics, than a sensitive one. The clonal differences in gene expression and other factors in response to O3 suggest that a wide set of changes are needed for a better O3 tolerance, and that increased expression of oxidative stress-related genes does not automatically signify a better tolerance of the stress. Accordingly, Calfapietra et al. (2008) and Ryan et al. (2009) stated that good O3 tolerance in Populus is a combination of different factors, such as reduced O3 uptake through stomata, higher isoprene emissions and carotenoid concentrations.

CO2 alleviates the harmful effects of ozone in combined CO2 and O3 treatment

Currently, [CO2] and [O3] in the atmosphere increase in parallel, and thus it is important to understand the combined effects of these gases on plant performance. Elevated [CO2] mitigates the harmful effects of O3 on plants (Riikonen et al. 2004, 2005; Oksanen et al. 2005; Peltonen et al. 2005), but more importantly O3 may offset the beneficial effects of CO2 on plant photosynthesis and growth (Wustman et al. 2001; Karnosky et al. 2003). These observations are supported by the gene expression data from the current study. In clone 4, the gene expression changes in EC + EO treatment in August were similar to the EC treatment in September. The results are in contrast to data from experiment with aspen, where the effects of combined elevated [CO2] and [O3] to gene expression showed more resemblance to O3 treatment (Gupta et al. 2005). Our phenological and physiological data support the similarity between EC + EO and EC treatments. The timetable of leaf abscission was very similar in the EC and EC + EO treatments, and most of the physiological measurements show that in the EC + EO treatment, CO2 was able to alleviate the negative effect of O3 (see Oksanen et al. 2005).

In contrast to the August sampling, the gene expression profile in the EC + EO treatment in September was more similar to the EO than to the EC treatment. This suggests that the CO2 effect on alleviating EO effects on gene expression was temporary. This was also seen in the concentrations of most phenolic compounds and physiological parameters where the EC + EO treatment resembled the EC treatment in August, but not in September. Accordingly, Riikonen et al. (2008) reported the disappearance of beneficial CO2 effects on paper birch photosynthesis in combined elevated [CO2] and [O3] from mid-August onwards in senescing leaves.

CONCLUSIONS

In this birch O3 and CO2 enrichment study, we combine information on responses derived from gene expression, phenological and physiological data. By using two genotypes with differential O3 responses, we show that the better adaptation capacity of one birch clone to increasing greenhouse gases is linked to a greater capability to alter its metabolism. Furthermore, increased expression of oxidative stress-related genes alone cannot predict the sensitivity to O3, but other data (e.g. growth and physiological parameters) are needed to assess the O3 tolerance in birch. The main CO2-induced changes were seen as a general down-regulation of carbohydrate metabolism and delayed senescence of leaves. In elevated O3, many senescence-associated genes were up-regulated, indicating earlier leaf senescence caused by increased oxidative stress. Analysis of gene expression does not necessarily reflect the activity of the corresponding proteins or biochemical processes, because several other processes, including post-translational modifications and protein stability, provide additional regulatory steps. This can be seen in our data where elevated O3 decreased expression of photosynthesis-related genes without affecting net photosynthesis. Thus, gene expression data need to be interpreted carefully and should, if possible, be complemented with physiological data. In an ecological perspective, the deleterious impact of increasing tropospheric [O3] in leaves may be diminished by elevated [CO2], but the beneficial effect decreases towards the end of the growing season. This study increases the understanding of the mechanisms behind the O3 tolerance, the responsiveness of birch to elevated [CO2] and the interactive effects of these two greenhouse gases in nature.

ACKNOWLEDGMENTS

The research was supported by the Academy of Finland (projects 40924, 047074, 240016) and the European Commission (project ERBIC15CT-980102). We thank Marja-Leena Jalkanen, Pia Lappalainen and Elina Välimäki for their skillful technical assistance, and Dr Kirk Overmyer for revising the English language. Turku Centre for Biotechnology, University of Turku and Åbo Akademi University are acknowledged for providing the arrays.

Ancillary