The Arabidopsis thaliana aquaporin AtPIP1;2 is a physiologically relevant CO2 transport facilitator

Authors


(fax +49 6151 163836; e-mail kaldenhoff@bio.tu-darmstadt.de).

Summary

Cellular exchange of carbon dioxide (CO2) is of extraordinary importance for life. Despite this significance, its molecular mechanisms are still unclear and a matter of controversy. In contrast to other living organisms, plants are physiologically limited by the availability of CO2. In most plants, net photosynthesis is directly dependent on CO2 diffusion from the atmosphere to the chloroplast. Thus, it is important to analyze CO2 transport with regards to its effect on photosynthesis. A mutation of the Arabidopsis thaliana AtPIP1;2 gene, which was characterized as a non-water transporting but CO2 transport-facilitating aquaporin in heterologous expression systems, correlated with a reduction in photosynthesis under a wide range of atmospheric CO2 concentrations. Here, we could demonstrate that the effect was caused by reduced CO2 conductivity in leaf tissue. It is concluded that the AtPIP1;2 gene product limits CO2 diffusion and photosynthesis in leaves.

Introduction

The exchange of water and CO2 play a crucial role in living organisms. Accordingly, the respective transport mechanisms for water or CO2 are of general interest. Although the existence of protein-facilitated membrane water transport has been challenged for some time, it is now well established that aquaporins function as water pores in biomembranes and decrease membrane resistance for water (Agre et al., 1998). In contrast to the wide acceptance of the water transport function, aquaporin-facilitated CO2 transport is under debate (Verkman, 2002). This is because of contradicting results from studies using artificial bilayers (Missner et al., 2008), molecular dynamics simulations (Hub and de Groot, 2006) and theoretical considerations (Gutknecht et al., 1988) on the one hand, versus cell-based CO2 transport assays (Cooper et al., 2002; Blank and Ehmke, 2003; Uehlein et al., 2003; Musa-Aziz et al., 2009), and physiological analysis in animals (Endeward and Gros, 2005) and plants (Javot and Maurel, 2002; Siefritz et al., 2002; Uehlein et al., 2003; Hanba et al., 2004), on the other. The common limitation of photosynthesis by CO2 availability makes plants an ideal system for testing the physiological relevance of cellular features assumed to alter resistance to CO2 diffusion. CO2 enters the plant via leaf stomata and crosses cell membranes as well as chloroplast membranes to be incorporated into sugar compounds (Evans et al., 2009). However, for a long time the CO2 diffusion resistance of mesophyll cells within a leaf, or its inverse CO2 conductance (gm), was considered to be constant. Only recently has increasing evidence for variable gm, and environmental factors controlling gm, emerged (Flexas et al., 2007a; Monti et al., 2009; Montpied et al., 2009; Perez-Martin et al., 2009). Using well-established physiological techniques for the determination of mesophyll CO2 conductance in leaves, data indicating the significance of aquaporins in these processes have been obtained. For example, Oryza sativa (rice) overexpressing aquaporin HsPIP2;1 or Nicotiana tabacum (tobacco) overexpressing aquaporin NtAQP1 showed increased gm (Hanba et al., 2004; Flexas et al., 2006). Furthermore, the inner chloroplast membrane, which has low intrinsic CO2 permeability without aquaporins, appeared to show increased CO2 transport rates when the aquaporin NtAQP1 was present (Uehlein et al., 2008).

However, it has not been demonstrated that an aquaporin limits CO2 transport, and consequently restricts photosynthetic efficiency, without the possibility of unpredictable side effects caused by RNA-silencing techniques, for example. In the case of aquaporins in plants it is exceedingly difficult to achieve gene-specific effects by RNA silencing, because either the number of aquaporin genes is unknown or/and the genome contains numerous closely related genes.

In addition to characterizing the molecular transport properties of aquaporins in the Saccharomyces cerevisiae (yeast) heterologous expression system, we analyzed the effect of disruption to the aquaporin gene AtPIP1;2 by a T-DNA insertion on CO2 transport and photosynthesis using carbon isotope discrimination determination in plants, as well as by measuring CO2 exchange during photosynthesis.

Results

Functional characterization of aquaporins in the heterologous expression system of yeast

In order to characterize AtPIP1;2 and AtPIP2;3 function, the respective proteins were expressed in yeast and the cells were examined with regard to water or CO2 transport rates in a stopped-flow spectrophotometer (Otto et al., 2010). For the determination of CO2 membrane transport-driven intracellular acidification, a carbonic anhydrase from tobacco was co-expressed with the aquaporin. Western blot analysis with a carbonic anhydrase-specific antibody (Ludwig et al., 1998) revealed that independent of the aquaporin co-expressed in yeast, the carbonic anhydrase content remained unchanged (Figure S1, Appendix S1). Furthermore, carbonic anhydrase reaction activity in yeast was estimated by reaction product analysis using a mass spectrometer (Endeward et al., 2008). It was found to be of adequate enzymatic activity in the yeast strains under investigation to ensure that the conversion reaction to carbonic acid was not rate limiting (activity > 800 s−1), and was similar to values obtained in human erythrocytes (Geers and Gros, 2000). The abundance of Arabidopsis aquaporins was analyzed by Western blot with specific antibodies against AtPIP1;2 and AtPIP2;3, respectively (Figure S2, Appendix S1). In order to assay aquaporin function at the time point of highest activity, a time course was performed first. Activity peaks were obtained 12 h after induction for water transport analysis, and at 16 h for CO2 transport assays (Figures S3 and S4). Analysis of NtPIP2;1, a tobacco PIP2 member, served as an internal control. NtPIP2;1 was already shown to be highly water permeable (Fischer and Kaldenhoff, 2008), which was confirmed in our experiments (Pf = 0.242 ± 0.014 cm s−1). In accordance with prior reports (Biela et al., 1999; Martre et al., 2002; Fetter et al., 2004), the functional studies revealed that AtPIP1;2 did not increase the membrane water permeability (control and AtPIP1;2 Pf = 0.003 ± 0.001 cm s−1), whereas the PIP2 aquaporin considerably raised water transport rates of yeast cells subjected to hyperosmotic conditions (Pf = 0.078 ± 0.005 cm s−1; Figure 1a). The reverse situation for aquaporin function was observed for CO2-dependent intracellular acidification, as determined by fluoresceine fluorescence quenching rates (Figure 1b). Here, the rates in AtPIP2;3-expressing yeast were similar to those of the controls, and the rates in AtPIP1;2-expressing yeasts were increased (AtPIP1;2 = 0.018 ± 0.0004 cm s−1; control or AtPIP2;3 = 0.005 ± 0.0014 cm s−1). Yeast cells expressing NtAQP1 showed an increased acidification rate (NtAQP1 = 0.021 ± 0.0006 cm s−1), verifying data obtained by expression in Xenopus oocytes (Uehlein et al., 2003). These results lead to the conclusion that both Arabidopsis aquaporins were functionally expressed in yeasts, although with different specificity regarding water or CO2 transport facilitation.

Figure 1.

 Water permeability or CO2 permeability of intact yeast protoplasts expressing AtPIP1;2 or AtPIP2;3.
(a) Kinetics of yeast protoplasts swelling was recorded as time courses of decreased scattered light intensity in a stopped-flow spectrophotometer for non-induced cells or those expressing NtPIP2;1, AtPIP1;2 or AtPIP2;3, respectively, as indicated (n = 30). The osmotic permeability coefficient (Pf) for AtPIP1;2 indicates no function as a water-transport facilitator, compared with controls. Expression of AtPIP2;3 increased the water permeability of yeast membrane by about 30-fold.
(b) Kinetics of decreasing fluorescence intensity of intact yeast controls or cells expressing NtAQP1, AtPIP1;2 or AtPIP2;3 and the tobacco chloroplast carbonic anhydrase as a result of cell acidification in a stopped-flow spectrophotometer after the addition of CO2-enriched buffer, as indicated (n = 30). Values are means ± SE. a,b,cDifferent statistical significance groups.

Function in plants

For the analysis of aquaporin function in plants, respective T-DNA insertion lines were analyzed and compared with controls. The localization of T-DNA in insertion lines atpip1;2-1 and atpip2;3-1 were verified by PCR with T-DNA (LBb1) and gene-specific primers (P1_for/_rev and P2_for/rev, respectively; Figure 2b) using genomic DNA as the template. The two lines were found to be homozygous for the aquaporin insertion because gene-specific primers failed to provide an amplification product (Figure 2c). Southern hybridization of fragmented genomic DNA was probed with a T-DNA specific probe. It indicated a single T-DNA insertion in atpip1;2-1 and an additional T-DNA insertion, besides that in the PIP2 aquaporin gene, in atpip2;3-1 (Figure S5, Appendix S1). This was confirmed by thermal asymmetric interlaced (TAIL)-PCR, which revealed an additional T-DNA located in a non-coding region of chromosome 3 (Table S1, Appendix S1). The T-DNA localization in the respective aquaporin gene is depicted in Figure 2a. Both lines show a complete absence of the corresponding aquaporin mRNA, as demonstrated by northern blotting (Figure 2d). The T-DNA insertion lines as well as the wild-type control plants did not display significant differences in root length or growth, root surface, leaf number or leaf area under different growth conditions. Leaf morphology with regard to leaf thickness and tissue texture was indistinguishable. Also, stomata density or stomata size resembled that of the wild type, and parameters important for photosynthetic efficiency, including chlorophyll content or Rubisco content, were like those from the wild type. Moreover, the relative water content (RWC) was insignificantly different among all three lines, as well as leaf dry mass per area (LMA; Table 1). Both measures, RWC and leaf area, display plant water status (Hsiao, 1973; Jones, 2007). Furthermore, the velocity of stomata closure after treatment with abscisic acid was not modified, which also indicates that cellular water transport in leaves and guard cells was not significantly affected by the T-DNA insertions (Figure S6, Appendix S1). With this regard, and somewhat distinct to the results presented by Postaire et al. (2010), who showed a moderate difference in leaf hydraulic conductivity under certain conditions, we couldn’t find significant differences in leaf hydraulic conductance in green house-grown plants using a modified high pressure flow meter (Tyree et al., 1995). In conclusion, the two lines provided us with comparable plant material to study the physiological consequences of AtPIP1;2 loss of function, and evaluate it against that of AtPIP2;3 as well as in the wild type. By relating data from AtPIP2;3 and the wild type, we were able to detect and exclude a possible general or unspecific effect of the T-DNA insertion.

Figure 2.

 T-DNA insertion position in atpip1;2-1 and atpip2;3-1, and its effect on gene expression level.
(a) Scheme shows that both lines carry a T-DNA insertion in exon 2 and the primer positions used in this study.
(b) The T-DNA flanking sequence could be amplified in a PCR using primer pairs of either P1_for or P2_for and LBb1.
(c) The amplification of the full-length gene was not possible in both T-DNA insertion lines because of an insertion in both alleles of the respective gene. In all PCR experiments a general positive and negative control was included.
(d) Northern blot hybridization with aquaporin-specific probes showed only signals for AtPIP1;2 and AtPIP2;3 aquaporins in wild-type leaf tissue. Signals were absent in RNA from both T-DNA insertion mutants (upper panel). An ethidium bromide stain was used as the equal loading control (lower panel). The template of the internal PCR positive control was AtPIP1;2 cDNA amplified with primers P1_for and P1_rev.

Table 1.   Morphological and physiological plant characteristics, determined in Darmstadt, Germany
Plant characteristicsWild typeatpip1;2-1atpip2;3-1
  1. LMA, leaf dry mass per area; RWC, relative water content.

  2. Values are means ± SEs (n = 5; except n = 10 for Fv/Fm). *Statistically significant differences (< 0.05).

Rubisco content (% of wild type)100116.35 ± 9.5696.74 ± 6.75
Total Chl. content (μg cm−2)18.60 ± 0.7219.12 ± 0.5218.83 ± 0.56
Chlorophyll a (μg cm−2)14.95 ± 0.5315.39 ± 0.4215.32 ± 0.71
Chlorophyll b (μg cm−2)3.62 ± 0.213.73 ± 0.423.64 ± 0.18
Chlorophyll a/b ratio4.16 ± 0.154.13 ± 0.114.2 ± 0.07
Fv/Fm0.813 ± 0.0160.826 ± 0.0040.821 ± 0.012
Number of leaves per plant16.4 ± 0.2416.6 ± 0.417 ± 0.45
Leaf rosette diameter (cm)8.71 ± 0.248.74 ± 0.288.79 ± 0.3
Average leaf area (cm2)5.32 ± 0.35.12 ± 0.155.22 ± 0.39
Stomatal density (no. mm−2)16 ± 214 ± 116 ± 2
RWC (%)81.6 ± 2.282.0 ± 0.883.8 ± 0.7
LMA (g m−2)13.41 ± 1.5612.67 ± 1.2013.08 ± 1.31

Assessment of the CO2 assimilation rate (AN) under different irradiation intensities (photon flux density, PFD) revealed that the atpip1;2-1 response under low irradiation (<100 μmol m−2 s−1) resembled that of the wild type and the atpip2;3-1 line (Figure 3a). Under these conditions photosynthesis is limited by light and by the light-driven electron transport. The data indicate that quantum yield was similar in all lines. The situation changed under higher irradiance (>300 μmol m−2 s−1), where photosynthesis becomes light saturated and limited by the carboxylation rate, respectively, by CO2 diffusion (Sharkey et al., 2007). Here, the atpip1;2-1 line shows significantly reduced values (Table 2). As no variation in Rubisco content compared with the control lines was observed, it is possible that the disparity arose from a reduced CO2 diffusion rate.

Figure 3.

 Photosynthesis response curves of Arabidopsis thaliana plants.
(a) The response of net photosynthesis (AN) to light intensity (PFD) of wild-type, atpip1;2-1 and atpip2;3-1 plants under 400 ppm CO2. atpip1;2-1 plants showed reduced AN compared with wild-type plants, whereas atpip2;3-1 plants were able to use light as efficiently as the wild type. Values are means ± SEs (n = 5).
(b) The response of net photosynthesis to substomata CO2 concentration (Ci) of wild-type, atpip1;2-1 and atpip2;3-1 plants detected under a saturating light intensity (750 μmol m−2 s−1). Whereas atpip2;3-1 plants were able to keep AN in the same range as the wild type, atpip1;2-1 plants were impaired. Values are means ± SEs (n = 5).

Table 2.   Photosynthetic parameters determined in Darmstadt, Germany
 AN (μmol CO2 m−2 s−1)Ci (μmol mol−1)gm (mol CO2 m−2 s−1 bar−1)Vcmax (μmol m−2 s_1)Jmax (μmol m−2 s−1)Slope AN/Ci curve initial phase
  1. AN, net photosynthesis; Ci, substomata CO2 concentration; gm, mesophyll conductance to CO2, estimated as described by Harley et al. (1992); Jmax, maximum capacity of electron transport; Vcmax, maximum velocity of carboxylation.

  2. Values are means ± SEs (n = 5; except n = 4 for gm). *Statistically significant differences (P < 0.05).

Wild type13.303 ± 0.85315.63 ± 7.930.1608 ± 0.02396.3 ± 4.94111.17 ± 5.670.058 ± 0.003
atpip1;2-110.18 ± 0.91*304.46 ± 11.420.0989 ± 0.012*76.6 ± 5.35*84.4 ± 5.91*0.046 ± 0.002*
atpip2;3-113.43 ± 0.61328.23 ± 4.920.1477 ± 0.01489.2 ± 9.46101 ± 5.320.057 ± 0.002

The relation between photosynthesis and leaf internal CO2 concentration (AN/Ci curves; Figure 3b) provides information of how internal and external factors affect the components of photosynthesis (Sharkey et al., 2007). When Ci ranges are comparable, the response of AN to Ci eliminates the effect of stomata and boundary layer on CO2 supply for photosynthesis. For the measurements depicted in Figure 3b, Ci values did not differ significantly between the lines under investigation. The mathematical prediction (Farquhar et al., 1980), and its modification (von Caemmerer et al., 2009), suggests three AN limiting phases. The first is restricted by: (i) Rubisco activity, which depends on CO2 availability at the enzymes active site; followed by (ii) a ribulosebisphosphate; and (iii) a triose-phosphate limitation phase. AN/Ci curves revealed that the response to increasing substomatal CO2 concentrations (Ci) was reduced for the first phase in atpip1;2-1 lines (Figure 3b, inset). Although wild-type and atpip2;3-1 plants reached a maximum value of about 17.5 μmol m−2 s−1, atpip1;2-1 plants could only achieve a net photosynthesis rate of about 14 μmol m−2 s−1 under conditions of high CO2 concentrations. These data allow the interpretation that in atpip1;2-1, CO2 supply for photosynthetic CO2 fixation was lower, and thereby reduced the availability of substrates synthesized by the Calvin–Benson cycle. Data calculation revealed a significant reduction of the maximum ribulosebisphosphate saturated rate of carboxylation (Vcmax), and the maximum of electron transport (Jmax) in atpip1;2-1 (Table 2), which would argue for a reduction in Rubisco content. However, this is not the case in atpip1;2-1 plants.

Under ambient atmospheric CO2 concentrations (400 ppm) and moderate light intensities (300 μmol m−2 s−1), stomata aperture was reduced about 25% in atpip1;2-1 in comparison with controls (see Figure S6, before the application of ABA). Stomata closure should lead to reduced Ci levels under comparable AN values, because atmospheric CO2 supply would be reduced while the CO2 sink remains constant. Conversely, in atpip1;2-1 plants Ci appeared to be increased, not decreased, for the same AN as the wild type (Figure 3b). Reduced mesophyll conductance could be an explanation for this observation. Based on the combined chlorophyll fluorescence and gas exchange data, an estimation of leaf tissue CO2 conductance (gm) indicated a reduction of approximately 40% in plants without AtPIP1;2, using the variable J method (Harley et al., 1992) (Table 2).

This result was confirmed in a separate experiment where photosynthetic isotope discrimination (C13/C12) was determined in New Mexico (USA). We combined leaf gas exchange with tunable diode laser spectroscopy (TDL) to perform online measurements of ΔC13/C12 (Flexas et al., 2006; Barbour et al., 2007; Uehlein and Kaldenhoff, 2008; Bickford et al., 2009). At the assay light intensity of 600 μmol photons m−2 s−1, AN,gs and CO2 concentration in the chloroplast (Cc) were not significantly different between the wild type and atpip1;2-1 (< 0.21, 0.11 and 0.19, respectively; Table 3). Ci was slightly higher in the atpip1;2-1 line (< 0.02). However, gm was significantly lower in atpip1;2-1 when a point-based calculation was employed (21% lower, with < 0.02). A non-significant trend of lower AN and higher gs in atpip1;2-1 plants favored higher Ci values, and their combined effect on the calculation of Ci generated a significant difference for this parameter. Furthermore, Cc is calculated from AN, gm and Ci, and with AN being similar between the two lines, the effects of a high Ci in the mutant offsets the effects of a low gm.

Table 3.   Canopy isotopic gas exchange determined in New Mexico, USA
 AN (μmol CO2 m−2 s−1)gs (mmol m−2 s−1)Ci (μmol mol−1)13C gm (mol CO2 m−2 s−1 bar−1)
  1. Photosynthetic parameters and 13gm mesophyll conductance to CO2, measured using online photosynthetic CO2 discrimination. Values are means ± SEs (n = 3). *Statistically significant differences (< 0.05).

Wild type11.5 ± 0.60.19 ± 0.02231 ± 80.23 ± 0.02
atpip1;2-111.0 ± 0.30.21 ± 0.01259 ± 5*0.19 ± 0.01*

Transformation for complementation of the mutant line atpip1;2-1 with the cDNA of the corresponding aquaporin provided three independently transformed lines (atpip1;2-1/PRO35S:AtPIP1;2#1, #2 and #5), where AtPIP1;2 expression was restored, as indicated by semi-quantitative RT-PCR (Figure 4a). Gas exchange experiments with these plants assessing the response of net photosynthesis rate to changing light intensities, as well as to changing internal CO2 concentrations, revealed a significant difference compared with the response of atpip1;2-1 plants (Figure 4b,c). In both cases complemented plants showed light-response and AN/Ci curves like those of the wild-type controls. Based on the data of AN/Ci curves, mesophyll conductance to CO2 was calculated (Table 4). It was found to be not significantly different from wild-type gm as well as the values for AN and Ci at ambient CO2 concentrations (Table 4). We conclude that by transformation of a construct with the original, unmutated AtPIP1;2 cDNA we could obtain full complementation of the mutated phenotype. Consequently, the differences to controls obtained in atpip1;2-1 rely on the mutation in AtPIP1;2 only.

Figure 4.

 Analysis of atpip1;2-1/PRO35S:AtPIP1;2 complementation line (AtPIP1;2).
(a) Semi-quantitative RT-PCR with PIP1;2 specific primer (downstream of T-DNA) and UBQ10 primer (internal standard) showed no expression in atpip1;2-1, but did show expression in the wildtype and the aquaporin complementation line AtPIP1;2.
(b) The response of net photosynthesis (AN) to light intensity (PFD) of wild-type, atpip1;2-1 and atpip1;2-1/PRO35S:AtPIP1;2 plants under 400 ppm CO2. atpip1;2-1 complementation plants showed a similar reaction to wild-type plants. Values are means ± SEs (n = 3).
(c) The response of net photosynthesis to substomata CO2 concentration (Ci) of wild-type, atpip1;2-1 and atpip1;2-1/PRO35S:AtPIP1;2 plants detected under saturating light intensity (750 μmol m−2 s−1). AtPIP1;2 complementation plants were able to keep AN in the same range as the wild type. Values are means ± SE (n = 3).

Table 4.   Photosynthetic parameters of atpip1;2-1/PRO35S:AtPIP1;2 (AtPIP1;2)
 AN (μmol CO2 m−2 s−1)Ci (μmol mol−1)gm (mol CO2 m−2 s−1 bar−1)
  1. AN, net photosynthesis; Ci, substomata CO2 concentration; gm, mesophyll conductance to CO2, estimated as described by Harley et al. (1992).

  2. Values are means ± SEs (n = 3) determined in Darmstadt, Germany, and also given as % of wild type.

AtPIP1;212.94 ± 0.50334.88 ± 40.1418 ± 0.017
% of wildtype97.24106.1088.24

Discussion

Physiological data from tobacco aquaporin RNAi or antisense lines that indicated a function of a PIP1 aquaporin in roots or leaves were questioned, because a side effect of the transformed construct on the expression of other related aquaporin genes could not be completely excluded (Siefritz et al., 2001; Uehlein et al., 2003). With regard to a function in plant water transport in roots, the function of PIP1 aquaporins was recently confirmed using the same T-DNA insertion line (Postaire et al., 2010). In that paper the presented physiological data and those reported here rely on the characterization of monogenic homozygous aquaporin gene knock-out lines. Direct side effects on other related genes can therefore be excluded. However, indirect effects on other processes are possible, and must be considered as consequences of altered aquaporin function. Such an indirect effect could be the reduced Vcmax and Jmax in atpip1;2-1, as well as the reduced CO2 assimilation rate under CO2 saturation. A possible reason could be an altered feedback or sugar export limitation, which could be an acclimation of the photosynthetic machinery to the changed mesophyll conductance. A limitation of AN under CO2 saturation has been observed in several studies where the influence of inhibited aquaporin function on gm has been examined (Terashima and Ono, 2002, Flexas et al., 2006).

The fact that gm, the conductivity of leaf tissue for CO2, decreases by about 40% without aquaporin AtPIP1;2, one of the highest expressed PIP1 members in leaves (Jang et al., 2004), is in agreement with the data from yeast where PIP1 expression increases CO2-triggered intracellular acidification by about fourfold. This is in contrast to theoretical considerations and biophysical studies on the lipid bilayer (Hub and de Groot, 2006; Missner et al., 2008). These indicate that membrane CO2 transport rates were so high that a role of gas transport-facilitating proteins is unlikely, and the process is merely limited by unstirred layers on both sides of the membrane. Our studies show that CO2 transport rates in yeasts or plants were much lower than those for pure lipid bilayers, and were in a range where protein-facilitated transport could be relevant. One reason for this observation could be an unexpected high surface coverage of membrane-integrated and membrane-associated proteins reducing the free bilayer surface for CO2 diffusion. Membrane protein crowding and ectodomains might also limit exposure of lipid surfaces to the adjacent aqueous regions (Engelman, 2005). As a consequence, the actual CO2 transport rates would be reduced. Unstirred water layers would not be rate limiting under these conditions, and a significant role of certain aquaporins in the biological CO2 transport mechanism would fulfill the requirements of theoretical considerations. On the other hand, the estimation of CO2-triggered intracellular acidification rates with regard to CO2 transport rates in yeasts was based on determining the intracellular pH via fluorescence quenching, which is a rather indirect procedure. Therefore, some inaccuracy for the absolute values of CO2 transport rates must be conceded. But, the in vivo (yeast) assessed CO2 transport rates were found to be rate limited by the function of AtPIP1;2, but not by the function of AtPIP2;3. This suggests that AtPIP1;2 functions as a CO2 transport facilitator in living cells.

As indicated by our in planta data from independent experiments – gas exchange or isotope discrimination – the function of AtPIP1;2 has an effect on mesophyll CO2 conductance. The impact of AtPIP1;2 is variable, comparing gm obtained by leaf gas exchange/chlorophyll fluorescence and online photosynthetic CO2 discrimination. This is probably a result of the different growth conditions in New Mexico, where plants grow under 20% lower CO2 partial pressure (high elevation) and in drier conditions. Both CO2 partial pressure and lower humidity would favour the plant minimizing mesophyll resistance via anatomical means.

As already mentioned, Postaire et al. (2010) pointed out that AtPIP1;2 is involved in plant water transport, and therefore the absence of aquaporins leads to reduced hydraulic conductivity. A deficit in root water uptake could bring on reduced gs as well as reduced relative water content in leaves and leaf area (resulting from stomata closure, deficit in water supply and limited leaf expansion growth) (Bouchabke et al., 2008; Centritto et al., 2009). In turn, these drought-related changes could affect AN and gm. However, our data do not support any symptoms of an altered water supply, as leaf area and RWC were found to be insignificantly different. Besides, there is evidence that gs and gm can act independently, as demonstrated by Vrabl et al. (2009). In this study Helianthus plants were treated with ABA and, among other measures, values for gs and gm were determined. Whereas gs was decreased by about 30%, gm remained unaffected. In addition, we could not observe significant differences in the velocity of stomata closure when controls and T-DNA tagged lines were compared (Figure S6). This is also an indication that on the cellular level water transport rates were unaltered, although aquaporins were expressed in these cells (Kaldenhoff et al., 1995).

The function of AtPIP1;2 and homologous proteins as physiological active CO2 transport facilitators provides a molecular mechanism and an explanation for the observed adaptation of gm to changing growth conditions (Flexas et al., 2006, 2007a; Evans et al., 2009; Galle et al., 2009). In summary, our data show that CO2 availability for photosynthesis, and consequently net photosynthesis, were modified by AtPIP1;2 function.

Experimental procedures

Plant material and growth conditions

The atpip1;2-1 and atpip2;3-1 lines were created by T-DNA insertion mutagenesis, and were obtained from the Nottingham Arabidopsis Stock Centre (Alonso et al., 2003; atpip1;2-1 = N519794, atpip2;3-1 = N617876). Plants of the T3 generation were used for genotyping, and derived T4 plants were anatomically and physiologically characterized. The presence of a T-DNA insertion in the respective target genes was verified by PCR using primer pair P1 and LBb1 for atpip1;2-1, and P2 and LBb1 for atpip2;3-1. Plants were grown on soil under standard glasshouse conditions with a 12-h photoperiod (24°C day/20°C night), 60% relative humidity, a photon flux density at plant height of about 70–80 μmol m−2 s−1 and an ambient CO2 concentration of 370 μmol mol−1.

Growth conditions at the University of New Mexico

Seeds were combined with a 0.1% agarose solution, sown on pre-wetted soil in custom containers consisting of a bottom-less 40-ml tube embedded in a 150-ml pot and cold-stratified in the dark at 4°C for 3 days. Plants were then moved to a growth chamber equipped with a mixture of metal halide and high-pressure sodium lamps (E8 model; Conviron, http://www.conviron.com) with a 16-h photoperiod, day/night temperatures of 22/18°C, relative humidity of 30–60% and a PFD of 150 μmol photons m−2 s−1. Plants were watered daily and fertilized every other day with a commercial fertilizer, and grown for 4–5 weeks (until they reached a projected leaf area of approximately 20 cm2) prior to measurement.

Gas exchange and chlorophyll fluorescence measurements

Gas exchange measurements were performed using the GFS-3000 photosynthesis system (Walz, http://www.walz.com). Water and CO2 concentrations at the inlet and outlet of the cuvette were measured using a differential infrared gas analyzer (IRGA). The cuvette flow was adjusted to 750 μmol s−1, and its area was 3 cm2. Once plant leaves were light adapted at a saturating PFD of 1000 and 400 μmol mol−1 CO2 (Ca), light response curves were recorded at nine different light intensities (0–1000 μmol m−2 s−1) by decreasing the applied PFD in a stepwise fashion. CO2 response curves were obtained by measuring the net photosynthesis rate depending on varying CO2 concentrations in the cuvette. Leaves were adjusted to 750 μmol m−2 s−1 PFD and 400 μmol mol−1 CO2. Measurements were started after leaves showed a constant photosynthetic rate. First, the CO2 concentration was stepwise reduced to a Ca of 50 μmol mol−1 CO2, followed by 400 μmol mol−1 CO2, to regain initial CO2 assimilation rates. Subsequently Ca was increased stepwise to 1000 μmol mol−1. Chlorophyll fluorescence was measured in parallel using the LED-Array/PAM-Fluorometer 3055-FL (Walz). According to (Genty et al., 1989) PSII quantum yield (φPSII) was obtained and used to calculate the electron transport rate (Jflu) as described by the manufacturer. The leaf temperature was kept at 25°C, and the leaf-to-air vapor pressure deficit was maintained at 1.2 kPa. Recordings were made on the seventh or eighth pair of leaves of 6-week-old plants.

Estimation of gm by gas exchange and chlorophyll fluorescence

Mesophyll conductance for CO2 (gm) was calculated according to Harley et al. (1992), as described by (Flexas et al., 2007b). A curve-fitting method (Ethier and Livingston, 2004; Ethier et al., 2006) was used to estimate Vcmax and Jmax, applying the online tool described in Sharkey et al. (2007).

Measurement methods of online photosynthetic CO2 discrimination

Online measurements of photosynthetic 13CO2 discrimination from a combined IRGA-Tunable Diode Laser system (Flexas et al., 2006; Barbour et al., 2007; Uehlein et al., 2008; Bickford et al., 2009) were used to calculate gm and Cc (concentration of CO2 at the site of carboxylation) using the equations of (Evans et al., 1986). The TGA-100 (Campbell Scientific, http://www.campbellsci.com) measures absolute concentrations of 13CO2 and 12CO2 at a frequency of 10 Hz from dry air before and after exposure to a photosynthesizing shoot. The 10-Hz data were averaged over 10 s to calculate the isotopic composition (δ13C) of sampled air with a precision of 0.05–0.09‰. Measurements of photosynthesis were made using IRGAs in a LI-6400 portable gas exchange system with a custom whole plant shoot chamber and a 20 × 20-cm white light emitting diode array (Photon Systems Instruments, http://www.psi.cz). All data are reported for a light intensity of 600 μmol m−2 s−1. CO2 concentrations were converted to partial pressures in calculations (atmospheric pressure was approximately 0.845 bars). Isotopic point-based calculations of gm (Bickford et al., 2009) at a PFD of 600 μmol m−2 s−1 are included, as gm may change with light intensity. The point-based calculations assume no significant effects of fractionation by photorespiration and respiration. Leaf temperature was kept at 25°C and leaves were provided with 380 μmol mol−1 CO2 and a δ13C of approximately −4‰.

Determination of Rubisco and chlorophyll content, morphological characteristics, LMA and RWC

Rubisco content was determined by extracting soluble proteins from leaf samples and performing western blot analysis as described by Uehlein et al. (2008) using Rubisco-LSU antibody (Agrisera, http://www.agrisera.com). Protein content was quantified with Quantity One® (Bio-Rad, http://www.bio-rad.com). Chlorophyll was extracted from leaf samples and determined as described using acetone as the solvent (Porra et al., 1989). Leaf anatomical parameters were examined of 6-week-old plants. Leaf number was counted and the rosette diameter measured at three different points per plant. Leaf area was determined after scanning with imagej. Stomata length and density was assessed by making imprints of the leaf abaxial side with clear nail polish. After an incubation of 3–5 min at 20°C, light microscope pictures were taken and analyzed with imagej. In order to obtain RWC, leaves were harvested and their fresh weight (FW) was determined. After a 16-h incubation in H2Odest. at 4°C, their turgid weight (TW) was obtained. Afterwards, leaves were dried at 70°C for 48 h and their dry weight (DW) was measured. RWC was then calculated as follows: RWC = [(FW − DW)/(TW − DW)] × 100%. LMA was obtained applying following formula: LMA = DW/area.

Plant transformation and selection

Complementation lines were generated by floral-dip transformation of atpip1;2-1 plants via Agrobacterium tumefaciens (GV3101), as described by Clough and Bent (1998). cDNA of AtPIP1;2 was cloned into pMDC32 (Curtis and Grossniklaus, 2003) applying the GatewayTM Technique (Invitrogen, http://www.invitrogen.com). Selection was performed based on a protocol given by Harrison et al. (2006).

Semi-quantitative RT-PCR

RNA was isolated (RNeasy plant mini kit; Qiagen, http://www.qiagen.com) and transcribed into first-strand cDNA using an oligo(dT) Primer and a M-MULV reverse transcriptase (NEB, http://www.neb.com) as described in the manufacturer’s manual. For a 50-μl PCR reaction, 2 μl of first-strand cDNA from each plant were used as a template. The reaction was completed, divided into two halves and primer pairs for UBQ10 (for, 5′-GATCTTTGCCGGAAAACAATTGGAG-3′; rev, 5′-CTTGTCATTAGAAAGAAAGAGATAACAGG-3′) were added to one reaction mix. For the PIP1 aquaporin primer pair for, 5′-ATAGCGGCCGCATGGAAGGTAAAGAAGAAGATG-3′, and rev, 5′-ATAGTCGACTTAGCTTCTGGACTTGAATG-3′, was added. The optimal number of PCR cycles was determined in preceding experiments to ensure that the PCR was in the linear phase of amplification (30 and 27 cycles for AtPIP1;2 and Ubiquitin, respectively). Aliquots of each reaction mix were separated in a 1.2% agarose gel, documented, analyzed by imagej for EtBr intensity and quantified.

Northern blot analysis

Total RNA was extracted from Arabidopsis leaves using the Qiagen RNeasy Plant Mini kit. A 10-μg portion of total RNA of different preparations were separated according to size on a 1% denaturing agarose-formaldehyde gel, and transferred to a nylon membrane by capillary blotting. The nucleic acids were UV cross-linked and prehybridized in DIG easy hyb (Roche, http://www.roche.com). The hybridization with DIG-labeled probes and washing was performed at 42°C according to the manufacturer’s manual. The DIG system includes an anti-DIG antibody, which is linked to an alkaline phosphatase. Target aquaporin mRNA was detected via this enzymatic activity. Specific aquaporin DIG-labeled probes were obtained by using the respective coding sequence of the AQP genes as the template in a PCR. These probes were specific because in both cases no signal could be detected for the T-DNA insertion line.

Protein expression in yeast

Yeast cells (W303) were transfected with an expression plasmid (pYES-DEST52; Invitrogen) carrying the CDS of the respective aqua-porin under an inducible GAL1 promotor using the lithium acetate method. Yeast cells used for CO2-permeability measurements contained an additional plasmid (pGREG505-2μORI, Euroscarf, modified; Otto et al., 2010) carrying a tobacco chloroplast carbonic anhydrase, also under the GAL1 promotor. Selection was based on ura3 and ura3/leu2 complementation. Yeast transformants were cultured in synthetic complete medium for 18 h (the optimal induction time) at 30°C. Cultures were diluted to OD 0.6 and heterologous protein expression was induced by changing the carbon source of the medium from glucose to galactose.

Analysis of water permeability

Water permeability of intact yeast protoplasts was assessed by stopped-flow spectrophotometry (SFM300; BioLogic, http://www.bio-logic.info) as described by Fischer and Kaldenhoff (2008) and Otto et al. (2010). The S. cerevisiae cells were first protoplasted by cell wall degradation using Zymolyase. After an incubation of 45 min in 50 mm KH2PO4 (pH 7.2), 0.2%β-mercaptoethanol, 2.4 m Sorbit, 1 × 10−3% Zymolyase 20T and 0.1% BSA cells were suspended in 1.8 m Sorbit, 50 mm NaCl, 5 mm CaCl2 and 10 mm Tris/HCl (pH 8.0). The protoplasts were exposed to a 300 mosmol outwardly directed osmotic gradient to induce protoplast swelling by mixing cells with 1.2 m Sorbit, 50 mm NaCl, 5 mm CaCl2 and 10 mm Tris/HCl (pH 8.0). Volume change was followed by the decrease of scattered light intensity in the spectrophotometer. Quantification of water conductivity was achieved by fitting a single exponential function on the initial 100 ms on protoplast swelling kinetics using biokine (BioLogic). The osmotic water permeability coefficients (Pf) were calculated using the rate constant of the exponential decay, the partial volume of water, the external osmolarity after the mixing event, and the initial mean protoplast volume and surface (van Heeswijk and van Os, 1986). The initial protoplast size was determined by light microscopy.

Analysis of CO2 permeability

The CO2 membrane transport rates were determined by loading whole yeast cells with the pH-sensitive fluoresceinediacetate, as described by Bertl and Kaldenhoff (2007). Cells were suspended in 25 mm HEPES/NaOH, 75 mm NaCl (pH 6) and transferred into the stopped-flow device. Kinetics were detected at 490 nm. A 515-nm cut-off filter was used to detect fluorescence emission. Settings were the same as for water. Cells were rapidly mixed with 25 mm HEPES/NaOH, 75 mm NaHCO3 at pH 6, and cytosolic acidification can be followed via the decrease of fluorescence emission. To calculate membrane CO2 permeability, the intravesicular pH was assessed prior to CO2 uptake (Slavik, 1982). The rate constant of the acidification was determined over the initial 100 ms. CO2 permeability was calculated using the method of Yang et al. (2000). To make sure that carbonic anhydrase activity was not limiting the conversion reaction, its activity was assayed via mass spectrometry as described by Endeward et al. (2008).

Accession numbers

Sequence data from this article can be found in GenBank/EMBL data libraries under accession numbers: At2g45960 (AtPIP1;2) and At2g37180 (AtPIP2;3), as well as M94135 (tobacco carbonic anhydrase). Accession numbers of sequence data: At2g45960 (AtPIP1;2); At2g37180 (AtPIP2;3); and M94135 (Nicotiana tabacum). Arabidopsis seed stock number: N519794 (atpip1;2-1); N617876 (atpip2;3-1).

Acknowledgements

RK and MH thank the German Israel Foundation (GIF 904-4.12/2006) for financial support. This work was supported by a grant from the United States National Science Foundation (IOS-0719118) to DTH, and by funding from the American Society of Plant Biologists (SURF 2009) and the United States National Institutes for Health (NIH-NIGMS GM060201-09) for DP. We are indebted to N. Pede and V. Endeward for carbonic anhydrase (CA) activity determination in yeast. We also thank B. Otto and M. Urban for CA western blots.

Ancillary