Cell wall composition contributes to the control of transpiration efficiency in Arabidopsis thaliana


For correspondence (fax +44 117 331 7984; e-mail alistair.hetherington@bristol.ac.uk).


To identify loci in Arabidopsis involved in the control of transpirational water loss and transpiration efficiency (TE) we carried out an infrared thermal imaging-based screen. We report the identification of a new allele of the Arabidopsis CesA7 cellulose synthase locus designated AtCesA7irx3-5 involved in the control of TE. Leaves of the AtCesA7irx3-5 mutant are warmer than the wild type (WT). This is due to reduced stomatal pore widths brought about by guard cells that are significantly smaller than the WT. The xylem of the AtCesA7irx3-5 mutant is also partially collapsed, and we suggest that the small guard cells in the mutant result from decreased water supply to the developing leaf. We used carbon isotope discrimination to show that TE is increased in AtCesA7irx3-5 when compared with the WT. Our work identifies a new class of genes that affects TE and raises the possibility that other genes involved in cell wall biosynthesis will have an impact on water use efficiency.


In most parts of the world water is a major limitation to crop yield and food production; the world’s population is increasing at such a rate that this limitation is deemed to be one of the most crucial problems facing society (Eckardt et al., 2009). It is against the backdrop of increasing pressures on the world’s water resources that attempts to produce crops that use water more efficiently (i.e. use less water while maintaining or improving yield) assume great importance. Although plants take up significant amounts of water, only a small percentage of the water absorbed by the roots is directly used during growth and development. The bulk of it is lost to the atmosphere through leaf transpiration (Hales, 1727).

There are a number of strategies that are likely to lead to improved water usage in the agricultural environment. There has already been considerable success in the sustainable use of water through the development of novel irrigation strategies such as deficit irrigation and partial root drying (Morison et al., 2008). The transpirational efficiency (TE) of leaves (the amount of carbon fixed per unit water transpired) can also be improved (Morison et al., 2008; Reynolds and Tuberosa, 2008; Yoo et al., 2009). For example, Richards and colleagues used leaf carbon isotope discrimination to screen wheat for genotypes exhibiting increased TE. This resulted in the isolation of the improved lines ‘Drysdale’ and ‘Reece’ that are now in commercial production (Rebetzke et al., 2002). Several individual processes could be manipulated in order to improve TE, such as, by definition, the efficiency of photosynthetic carbon assimilation or the mechanisms responsible for controlling water uptake and loss by the plant. Within this latter category are mechanisms modifying the sites of water uptake (modification of root architecture and/or the density of root hairs), water transport (modification of uptake mechanisms or of the xylem network) and the sites of water loss to the atmosphere (modification of stomatal density and/or function).

The objective of our work was to identify genes involved in the control of transpirational water loss. To do this we carried out a genetic screen in Arabidopsis to identify lines that exhibited lesions in the control of transpirational water loss using infrared thermal imaging as a proxy (Wang et al., 2004). We reasoned that such an approach would be likely to identify genes involved in the control of stomatal conductance and water movement throughout the plant. In this paper we describe the isolation of a new allele of the Arabidopsis CesA7 locus. We demonstrate that this mutant exhibits increased TE and provide evidence that this is due to smaller stomatal pores and reduced water transport due to collapsed xylem elements.


In order to identify genes involved in the control of transpiration we carried out a screen of M2 plants from an ethyl methanesulphonate (EMS)-mutagenised Arabidopsis (background Columbia-2) population using infrared thermography exactly as described previously (Xie et al., 2006). Figure 1 shows an image of a mutant line named 34A identified in this screen compared with an image of wild-type (WT) Columbia (Col-2). At 6 weeks 34A is markedly smaller than WT (Figure 1a,b) and when exposed to a sudden drop in atmospheric relative humidity the leaf temperature of 34A is higher than that of WT (Figure 1c,d). This result suggests that, in comparison with WT, 34A exhibits a reduced transpiration rate. Genetic analysis demonstrated that this phenotype was caused by a single, recessive Mendelian mutation. To determine the identity of the gene responsible for the 34A phenotype we adopted a map-based cloning strategy. The 34A mutation was mapped to a region on chromosome 5 between the Col/Ld insertion/deletion (indel) marker CER456149 and a single nucleotide polymorphism at position 4896 on the bacterial artificial chromosome (BAC) clone K3M16. The region includes 19 annotated genes, with no obvious candidates for stomatal function. We therefore obtained as many insertional knockout mutants as were available for the predicted genes from the stock centres, and examined them for similar phenotypes to 34A. In addition to the thermal imaging phenotype, 34A has a compact rosette and dark green appearance. A SALK line (029940) with an insertion in At5g17420 closely resembled 34A. At5g17420 encodes a cellulose synthase subunit, AtCesA7, which is essential for cellulose synthesis during secondary cell wall formation (Turner and Somerville, 1997). Mutations in AtCesA7 have previously been identified in screens for cell wall properties and hence the locus is also known as IRREGULAR XYLEM 3 (IRX3), FRAGILE FIBER 5 (FRA5) and MURUS 10 (MUR10) (Taylor et al., 1999; Zhong et al., 2003; Bosca et al., 2006). To determine whether 34A is a new allele at this locus, we performed F1 complementation tests with the irx3-1 allele. All the F1 plants had the mutant phenotype. For further confirmation, we sequenced At5g17420 from 34A and identified a single D524N missense mutation, which affects a highly conserved residue in the protein. This is different from the previously identified mutations at this locus, as in mur10-2 His-734 is replaced with a Tyr residue. fra5 causes a missense amino acid substitution (Pro557 to Thr), while irx3-4 is a T-DNA insertion mutant (most likely a complete loss of CesA7 function; Taylor et al., 1999; Bosca et al., 2006).

Figure 1.

 34A plants are smaller and hotter than the wild type (WT).
Six-week-old WT (a) and 34A (AtCesA7irx3-5) (b) plants grown in long-day conditions (16/8 h). 34A (AtCesA7irx3-5) is smaller in stature, has a reduced rosette and fewer inflorescences than WT.
Infrared thermal imaging of 5-week-old WT following a drop in atmospheric relative humidity (see Experimental Procedures for full details) (c) and 34A (AtCesA7irx3-5) (d) plants showing that 34A (AtCesA7irx3-5) leaves are warmer than WT.

To confirm that a mutation at the AtCesA7 locus was responsible for the ‘hot’ phenotype we observed in line 34A we examined leaf thermal profiles of irx3-4 and mur10-2, two additional independent alleles of AtCesA7 (Turner and Somerville, 1997; Brown et al., 2005; Bosca et al., 2006). Figure 2 shows that, like the 34A line, both independent alleles of AtCesA7 display a warmer leaf temperature in comparison with their respective WT. In the light of our sequencing data and our comparative infrared thermographic analysis we refer to the 34A line henceforth as AtCesA7irx3-5, in line with the nomenclature proposed by Taylor (2008).

Figure 2.

 34A exhibits a similar leaf thermal profile to two independent alleles of AtCesA7.
Leaf temperatures of 34A (AtCesA7irx3-5), irx3-4 and mur10-2 (filled bars) are shown paired with their corresponding wild type (WT; open bars) leaf temperature. Thermal imaging was conducted under steady-state conditions. Values are means ± SE, = 20 (for each genotype a spot on four leaves from five plants was measured).

The thermal imaging data (Figures 1c,d and 2) indicate that transpiration rates were reduced in AtCesA7irx3-5 compared with WT, and this was confirmed experimentally as shown in Figure 3. In order to understand the origin of the reduced water loss in this mutant we first investigated whether stomatal function was compromised in AtCesA7irx3-5. Figure 4(a,b) shows, in comparison with their respective WTs, that neither AtCesA7irx3-5 nor irx3-4 were impaired in their ability to respond to ABA in terms of either ABA-inhibited stomatal opening or ABA-promoted stomatal closure. Figure 4c shows that both alleles also exhibit WT behaviour when exposed to the stomatal closure-inducing signal of 700 p.p.m. CO2. However, what was obvious from these data was that the maximum apertures attained by both irx3-4 and AtCesA7irx3-5 during light-stimulated stomatal opening were consistently smaller than in the WT. This prompted us to investigate whether the guard cells of AtCesA7irx3-5 were smaller than in the WT. The results presented in Figure 5a show that this was indeed the case. We also compared the density of stomata and epidermal pavement cells in AtCesA7irx3-5 and irx3-4. From the results presented in Figure 5b it is apparent that pavement cell densities, and to a lesser extent stomatal densities, are greater in AtCesA7irx3-5 and irx3-4 than in the WT, consistent with the smaller cells in these mutants (Figure 5a). Due to the much reduced size of the pavement cells, the stomatal index, i.e. the contribution of stomata to the total number of epidermal cells, has declined significantly in the two mutants compared with the WT.

Figure 3.

AtCesA7irx3-5 and irx3-4 exhibit a decreased rate of water loss compared with their respective wild types (WTs).
Weight loss of detached aerial parts was monitored over a 2-h period for AtCesA7irx3-5 and its WT, Col-2 and irx3-4 and its WT Col-0. Values are means ± SE, = 2 for each individual experimental repeat. The experiment was repeated three times with comparable results recorded on each occasion.

Figure 4.

AtCesA7 alleles exhibit wild-type responses to ABA and CO2.
Epidermal peels were treated to monitor stomatal response to ABA-induced closure (a), ABA inhibition of opening (b) or 700 p.p.m. CO2 (c). After a 2.5-h pre-incubation in which stomata were opened in CO2 free air (a and c) or closed (b), the appropriate treatment was given and peels were incubated for a further 2–2.5 h after which stomatal apertures were measured (a, = 80; b, = 80; c, = 40; bars = means ± SE).

Figure 5.

AtCesA7irx3-5 epidermis has smaller stomatal complexes and pavement cells, increased stomatal density and reduced stomatal index when compared with the wild type (Col-2, Col-0).
(a) Guard and epidermal cell area (= 200, **< 0.001).
(b) Stomatal and pavement cell density; the stomatal index is expressed as the percentage of total epidermal cells (= 30, **< 0.001). Bars = means ± SE.

The reduced transpiration rate which manifests itself in the increased leaf temperature observed in the AtCesA7irx3-5 mutant can therefore be ascribed to a reduction in the size of stomatal complexes and the maximum size of the stomatal pores. This may not be the complete answer, however, and other factors may contribute to the reduced transpiration rate in the mutants. In this context it is interesting to note that irx3-4 is known to exhibit a collapsed xylem phenotype (Turner and Somerville, 1997). This might be expected to drop the water potential of the leaves and thereby probably to decrease cellular turgor and stomatal aperture during gas exchange. To investigate whether AtCesA7irx3-5 displays a collapsed xylem phenotype we carried out an anatomical comparison between the mutant and the WT. The results presented in Figure 6 reveal that AtCesA7irx3-5 shows clear evidence of xylem collapse in stems. However, when we measured the water potential of mature leaves under well-watered conditions these were similar, within measurement error, in both mutant and WT (data not shown).

Figure 6.

AtCesA7irx3-5 exhibits a collapsed xylem phenotype.
(a) Cross-section of basal stem internode in wild-type plants.
(b) Cross-section of basal stem internode in AtCesA7irx3-5 plants.
mx, metaxylem; px, protoxylem. Scale bar = 25 μm.

Given our finding that the AtCesA7irx3-5 mutant exhibits significantly reduced transpirational water loss we decided to investigate how this might impact on transpiration efficiency (TE). The data in Figure 7 show that carbon isotope discrimination in both mature leaves and whole rosettes of AtCesA7irx3-5 is much lower than in the WT, indicative of a significantly greater TE in AtCesA7irx3-5 compared with the WT.

Figure 7.

AtCesA7irx3-5 exhibits much reduced carbon isotopic discrimination compared with wild type (Col-2) in either mature leaves or whole rosettes comprising leaves at different stages of development.
Carbon isotope discrimination in organic matter was used as a quantitative reporter for the integrated transpiration efficiency (TE) over the 5-week growth period between sowing and sampling. The lower its value, the greater TE. Bars = means ± SE.


34A exhibits a ‘hot’ phenotype brought about in part by a reduction in the size of stomatal complexes and pore aperture

Our screen for mutants with aberrant leaf thermal profiles has the potential to detect at least three classes of mutations: (i) mutations affecting water uptake by the root, (ii) mutations affecting water transport through the vascular system, and (iii) mutations directly affecting stomatal function or development. Previously, the use of infrared thermal imaging led to the identification of Arabidopsis mutants carrying lesions affecting stomatal responses to ABA (Merlot et al., 2002, 2007; Wang et al., 2004), relative humidity (Xie et al., 2006) and CO2 (Hashimoto et al., 2006). All these mutations affect stomatal function – the ability of stomata to open or close in response to endogenous and exogenous signals. In contrast to these investigations, our phenotypic characterization reveals that stomatal function of the 34A mutant is apparently unimpaired (Figure 4). Instead, our results show that in 34A the hot phenotype we observe (Figures 1 and 2) is in part brought about by the smaller stomatal complexes with reduced pore aperture (Figure 5a) that more than compensated for the slightly increased stomatal density.

The 34A phenotype is caused by a mutation at the AtCesA7 locus

We used map-based cloning to characterize the locus responsible for the 34A phenotype, and this turned out to be AtCesA7. Arabidopsis has 10 CesA genes which encode plasma membrane localised complexes that form hexamers involved in the synthesis of cellulose (Somerville et al., 2004; Persson et al., 2005). Different AtCesA genes are responsible for cellulose synthesis in primary and secondary cell walls. AtCesA1, -3 and -6 are involved in synthesis of primary cell wall cellulose, while AtCesA4, -7 and -8 are involved in the synthesis of secondary cell walls. The remaining CesA genes do not have an established phenotype (Persson et al., 2005).

It is likely that the small guard cells seen in the AtCesA7irx3-5 mutant result from impeded water supply to the developing leaf brought about by partially collapsed xylem elements

Previous work (Jones et al., 2003, 2005) has shown that maintaining the correct spectrum of carbohydrates in the cell wall is important for maintaining stomatal function. These investigations, however, focused on the non-cellulosic fraction, specifically arabinans and pectin (Jones et al., 2003, 2005). In contrast, we found that a gene involved in the synthesis of the cellulosic component of secondary cell walls affected stomatal development (size and proportion). We feel it is unlikely that the reduced cell size seen in the 34A mutant is caused directly by the AtCesA7irx3-5 lesion. This is because AtCESA7 is involved in cellulose synthesis during secondary cell wall synthesis. As the secondary cell wall is only laid down in cells that have ceased growth (Brett and Waldron, 1990) this suggests that the reduced guard cell size was already set before the predicted activity of AtCESA7. So what might be the explanation for the small cells observed in AtCesA7irx3-5?

Cell expansion is driven by turgor pressure (Boyer, 1985). In this context it is important to remember that AtCesA7 was first isolated as irregular xylem3 (irx3) on the basis of a collapsed stem xylem phenotype (Turner and Somerville, 1997; Taylor et al., 1999). Our mutant allele, AtCesA7irx3-5, also exhibits this phenotype (Figure 6) and also shows a marked reduction in the number of functional conducting xylem elements in the leaf vasculature (data not shown). We propose that the small size of the cells in AtCesA7irx3-5 is due to the lower turgor pressure in the developing leaf cells caused by the reduced and partially collapsed xylem. This interpretation is supported by the lower rate of water loss we measured from the mutant leaves compared with the WT (Figure 3). Interestingly, AtCesA8irx1, which encodes a cellulose synthase subunit involved in secondary cell wall formation, also exhibits a collapsed xylem phenotype and displays reduced water loss as a result of reduced water flow through the xylem (Chen et al., 2005). On the basis of our results obtained with AtCesA7irx3-5 we would predict that AtCesA8irx1 should display a ‘hot’ leaf phenotype. We investigated this possibility using infrared thermography, and found that this was indeed the case (data not shown). Another possibility would be that AtCesA7irx3-5 exhibits increased levels of ABA relative to the WT and that this contributes to the smaller guard cells. Although we have not investigated this suggestion, it receives some support from the observation that ABA-treated Tradescantia virginiana displays reduced stomatal apertures and smaller pore areas than untreated plants (Franks and Farquhar, 2001), and stomatal apertures in ABA-hypersensitive Arabidopsis mutants are lower than in WT (Rubio et al., 2009). If this turns out to be the case it would be consistent with the proposal that transpiration rate and ABA control stomatal development (Lake and Woodward, 2008).

Improving TE

The primary aim of our work was to identify mutants with improved TE. Our measurements of a greatly decreased carbon isotopic discrimination in AtCesA7irx3-5 (Figure 7) provide strong evidence for a much greater TE than in the WT (Farquhar et al., 1982; Masle et al., 1993, 2005). On the basis of these results we would predict that AtCesA8irx1 and other mutations that impair cell wall biosynthesis would also be likely to increase TE. For example the leaf wilting 3 mutant (lew3), which encodes a putative α-1,2-mannosyltransferase (ALG11), causes disruption of cellulose synthesis, partially collapsed xylem and reduced water loss from detached leaves (Zhang et al., 2009) as we see in AtCesA7irx3-5. As orthologous genes of AtCesA7 are known to be present in maize, rice and barley (ZmCesA12, OsCesA9 and HvCesA8, respectively; Appenzeller et al., 2004; Burton et al., 2004) it would be interesting to find out if mutations in these genes cause similar developmental and TE phenotypes to those in Arabidopsis.

The results of the work described in this paper identify a new class of genes with a significant impact on leaf transpirational water loss and thereby TE. Previous loci implicated in this process have been involved in the control of guard cell function (MRP5, Klein et al., 2003; ABP9, Zhang et al., 2008; NCED1, Thompson et al., 2007), stomatal development (GPA1, Nilson and Assmann, 2010; HDG11, Yu et al., 2008; ERECTA, Masle et al., 2005; GTL1, Yoo et al., 2008) and root and leaf mesophyll development (HRD, Karaba et al., 2007; esb1, Baxter et al., 2009; Masle et al., 2005). AtCesA7irx3-5 is involved in secondary cell wall deposition. We propose that the phenotype observed in the AtCesA7irx3-5 mutant is probably the result of decreased cell turgor due to reduced xylem water transport capacity. Interestingly, the esb1 (enhanced suberin) mutant which exhibits increased root suberin displays reduced stomatal aperture. The authors suggest that the increased root suberin may cause enhanced hydraulic resistance to water transport from the root, resulting in lower water potential (Baxter et al., 2009).

It is perhaps rather obvious from Figure 1 that although AtCesA7irx3-5 exhibits increased TE its dwarf phenotype is likely to be too severe to be attractive to plant breeders. Our work does highlight, however, the potential of making use of genes affecting water transport and stomatal development to produce crops better equipped to contribute to sustainable agriculture and the challenges associated with environment change.

Experimental Procedures

Plant material

Except where noted, Arabidopsis (Arabidopsis thaliana) ecotype Columbia and its descended seeds were grown essentially as described previously (Webb and Hetherington, 1997), except that illumination was 100 ± 20 μmol m−2 sec−1 provided by metal halide lamps with a 10-h photoperiod and temperature of 22°C. For the experiments we used 5–6-week-old plants.

Screening for mutants with altered transpirational water loss and identification of a mutation in the AtCesA7 locus

This was carried out as described previously (Xie et al., 2006). Briefly, 20000 seeds from an A. thaliana ecotype (Col-2) EMS M2 population representing 40 independent pools (each pool corresponding to approximately 1000 M1 plants) (Stimberg et al., 2002) were germinated and grown in a 3:1 mix of peat-based compost (SHL, Multipurpose; William Sinclair Horticulture, http://www.william-sinclair.co.uk/) and washed horticultural silver sand (Gem Horticulture, http://www.gemgardening.co.uk/) in Perspex horticultural propagators (Henleaze Garden Shop, http://www.henleazegardenshop.co.uk/). The propagators were placed in a growth room [10.5 h photoperiod, 150 μmol m−2 sec−1; air temperature 23 ± 2°C, relative humidity (RH) 25 ± 5%] and the plants were grown until they were 3 weeks old. The relative humidity inside the propagators was 65%. Forty minutes prior to infra-red thermography the acrylic lid was removed from the propagator. This caused the plants to experience a drop in RH from 65 to 25%. Plants displaying aberrant leaf temperatures were detected using infrared thermography conducted using an Inframetrics ThermaCam SC1000 focal plane array (256 × 256 pixel platinum silicide) imaging radiometer (3.4–5 μm) fitted with a 16° lens (Flir Systems Inc., http://www.flir.com/) and the data analysed using ThermaGRAM 95 Pro image analysis software (Thermoteknix Systems Ltd, http://www.thermoteknix.com/). Mutants exhibiting altered leaf surface temperature compared with WT were selected and self-pollinated, and seed (M3) was collected for further investigation. Backcross seed (F1s) were obtained by using the mutant lines as female and WT (Col-2) as male, and the F2 was used for segregation analysis. Mutants segregating in F2 were backcrossed to Col-2 for another two generations before being used for fine mapping and phenotypic analysis. One mutant, named 34A, was selected for further analysis. The backcrossed 34A line was out-crossed to Landsberg erecta (Ler) plants, and the resulting progeny were self-fertilised. Mutant individuals were selected from the F2 and DNA was prepared from leaf samples using DNeasy 96 plant kits according to the manufacturer’s instructions (Qiagen, http://www.qiagen.com/). The DNA samples were used to genotype the F2s with respect to known cleaved amplified polymorphic sequence (CAPS) and simple sequence length polymorphism (SSLP) markers (http://www.arabidopsis.org/), allowing co-segregation and relative position to be assessed for each marker.

Measurements of stomatal apertures

Plants were grown, and the inhibition of stomatal opening or the promotion of stomatal closure by ABA were investigated on isolated epidermal strips from 5–6-week-old plants using procedures described previously (Webb and Hetherington, 1997; Worrall et al., 2008). The promotion of stomatal closure by elevated CO2 was performed in essentially the same way as the promotion of stomatal closure by ABA, except that instead of adding external ABA, 700 p.p.m. CO2 (at 800 ml min−1) was perfused into the incubation buffer after the epidermal peels were incubated for 2.5 h under stomatal opening conditions.

Measurements of stomatal density, index and cell size

Stomatal density determinations were carried out using the dental impression procedure described in Gray et al. (2000). Measurements of guard cell and epidermal areas were captured during stomatal density determinations and were further processed and analysed with Imagetool software (http://ddsdx.uthscsa.edu/dig/itdesc.html).

Stem and leaf anatomy

Fresh cross-sections of stems (second oldest internode) and of leaf petioles were sampled on 10-week-old fully mature plants grown under higher light (250 μE m−2 sec−1). After staining for 30 sec in 0.05% toluidine blue these sections were mounted in water and observed by light microscopy (Zeiss Axioplan microscope; http://www.zeiss.com/) for examination of the vascular anatomy and staining of cellulose and lignin.

Water loss measurements

For water loss measurements, rosettes of AtCesA7 mutants and WT were detached from the soil surface and weighed immediately in foil boats. The boats with the plants were then placed on the laboratory bench (temperature 23–24°C, RH 50–53%) and weighed at the indicated time intervals. Two replicates were performed for each line. The percentage loss of fresh weight was calculated based on the initial weight of the plants. The experiment was repeated three times with similar results obtained in all three experiments.

Leaf carbon isotopic discrimination

The whole rosette or three mature leaves of at least eight WT and mutant plants grown under the same conditions as for observations of stem and leaf anatomy were sampled for measurement of carbon isotope discrimination at the very beginning of bolting. These samples were oven dried and ground to a fine powder. Isotope analyses were conducted with an Isochrom mass spectrometer (Micromass, http://www.waters.com/) coupled to a Carlo Erba elemental analyser (CE Instruments, http://www.ceinstruments.co.uk/) operating in continuous flow mode (Australian National University (ANU), Canberra, Australia). Carbon isotope ratios were obtained in δ-notation, where δ = R/Rstandard– 1 and R and Rstandard are the isotope ratios of the plant sample and the PDB standard, respectively. The δ13C values were then converted to carbon isotopic discrimination values, Δ13C, using the equation: inline image (Farquhar et al., 1982), where δa is the δ13C of atmospheric CO2 and δp is the δ13C of the plant material. δ13C of atmospheric CO2 was assumed to be −8%.


The authors acknowledge the assistance of the UK BBSRC (grant number BB/D0100201/1) for providing research funding. The authors are also grateful to Professor Simon Turner (University of Manchester, UK) for the gift of the irx3-4 seed and Dr Neil Taylor (University of York, UK) for the gift of the irx3-1 seed.