Tropospheric ozone (O3) is a phytotoxic air pollutant whose current background concentrations in parts of East Asia have caused estimated rice yield losses of up to 20%; currently, however, little is known about the mechanisms of O3 tolerance in rice. We previously identified a quantitative trait locus (QTL) in rice called OzT8, which was associated with relative dry weight under ozone stress. The photosynthetic response in SL46, a Nipponbare (NB)–Kasalath chromosome segment substitution line (SL) containing the OzT8 locus, was compared to the parent NB in multiple ozone fumigation experiments (100 ppb, 8 h d–1, 23 d). By day 23, SL46 showed significantly less reduction of photosynthetic capacity compared to NB; the maximum carboxylation rate of ribulose 1·5-bisphosphate carboxylase/oxygenase (Rubisco) decreased by 24% in SL46 compared to 49% in NB, and the maximum electron transport rate decreased by 16 and 39%, respectively. The midday carbon assimilation rates also showed a similar trend, but there was no genotypic difference in stomatal conductance. These results indicate that the OzT8 locus confers ozone tolerance via biochemical acclimation, not avoidance, making it a potentially valuable target for breeding of ozone tolerance into future rice lines. The sequence of photosynthetic response of rice to ozone stress and related tolerance factors are also discussed.
Tropospheric ozone (O3) is a phytotoxic air pollutant whose global background concentrations have been rising since the Industrial Revolution, and has been widely recognized as a problem affecting both natural plant communities and agricultural systems (Ashmore & Marshall 1999; Fowler 2008; Booker et al. 2009). Ozone enters the plant through the stomata and quickly reacts with the apoplast of nearby mesophyll cells, producing reactive oxygen species (ROS) and triggering a series of signalling cascades and plant defence responses that ultimately result in a range of effects, including visible foliar damage, decreased photosynthetic capacity and accelerated senescence (Omasa et al. 2000; Leipner, Oxborough & Baker 2001; Long & Naidu 2002; Kangasjarvi, Jaspers & Kollist 2005). All or some of these factors contribute to a decrease in crop yield under ozone stress (Heagle 1989; Ashmore 2005).
Nitrogen oxides are the main precursors to O3 in the troposphere, and their increased emissions in East and South Asia have led to rapidly rising surface [O3] in the region (Wang & Mauzerall 2006). At current [O3], crop yield losses have been estimated to be upwards of 30% in parts of eastern China (Wang & Mauzerall 2004). In the coming decades, large increases in O3 are predicted for parts of East, Southeast and South Asia, where the majority of worldwide rice is grown (Aunan, Berntsen & Seip 2000; Wang & Mauzerall 2004; Sitch et al. 2007). This raises serious concerns about the medium- and long-term food security in East and South Asia, because the current balance between crop production and demand can only be maintained by further increases in food production (Ainsworth 2008; Feng & Kobayashi 2009). In addition, because previous estimates of future yield losses under ozone stress in Asia are taken from models that are based on North American and European dose–response relationships, the actual yield losses expected in the future may currently be underestimated (Emberson et al. 2009).
The negative impact of O3 on plant growth and crop yields has been well documented in the United States and in Europe for their locally important crops, wheat and soybean (Ashmore & Marshall 1999). However, comparatively little is known about the effect of O3 on rice (Orzya sativa L.), the main Asian crop, and the mechanisms of O3 tolerance in rice are still poorly understood (Chen et al. in press). Almost all genetic and physiological studies conducted thus far have looked at genes or biochemical pathways triggered by O3 exposure without distinguishing between simple stress responses (e.g. closing of stomata) and genetically driven biochemical stress tolerance mechanisms. Previous studies on ozone tolerance in rice have contrasted the response of different rice cultivars (Inada et al. 2008; Pang, Kobayashi & Zhu 2009), but the disparate genetic background of the cultivars makes it difficult to identify any particular genetic basis of ozone tolerance and breeds this into existing lines.
As an alternative approach, a quantitative trait locus (QTL) mapping of ozone tolerance in rice was conducted by Frei, Tanaka & Wissuwa (2008) using 98 BC1F5 backcross inbred lines of Nipponbare (NB) (a moderately ozone-sensitive, modern cultivar from Japan) and Kasalath (an ozone-tolerant, traditional cultivar from India). The study identified four QTLs associated with leaf bronzing, one QTL for stomatal conductance and one QTL for relative shoot dry weight. The effect of two of these QTLs was previously tested by comparing the ozone response of NB to selected chromosome segment substitution lines (SLs) carrying introgressions from Kasalath in the genetic background of NB at the respective QTL positions. This led to the confirmation of two QTLs, termed as OzT3 and OzT9, which are associated with an increase and decrease, respectively, in the level of leaf bronzing in response to ozone stress (Frei et al. 2008).
Another QTL identified by Frei et al. (2008), OzT8, was associated with higher relative dry weight (i.e. less biomass loss) under ozone-stressed conditions. Because higher biomass at the seedling stage may be caused by higher photosynthetic rates and tillering, OzT8 may also be possibly associated with higher yield at the end of the season. Therefore, this QTL could potentially be very valuable as we attempt to maintain or increase crop yields in the face of global change and increasingly frequent environmental stress episodes.
The objectives of this study were to further investigate the phenotypic response of OzT8 to ozone stress, with a particular focus on photosynthesis, and to characterize the underlying physiological mechanism(s) of ozone stress tolerance associated with OzT8. Three hypotheses were tested: (1) OzT8 shows a higher relative biomass under ozone stress because of higher photosynthetic carbon assimilation rate; (2) OzT8 confers ozone tolerance via stress avoidance (i.e. exclusion of ozone from the leaf intercellular space via stomatal closure); and (3) OzT8 confers ozone tolerance via maintenance of carbon assimilation rate and biochemical photosynthetic capacity. To test these hypotheses, the photosynthetic response to ozone of the parent line NB was contrasted to that of SL46, a derived line that is genetically 92% identical to NB which contains introgressions from the donor parent Kasalath in a region that includes the OzT8 QTL. In addition, the sequence of photosynthetic response of the lines to ozone stress was also examined under varying timescales (from hourly to weekly) to identify when genotypic differences occurred.
MATERIALS AND METHODS
Plant material and growth conditions
Two rice (O. sativa L.) genotypes were used throughout this study: NB, a modern variety of the japonica subspecies, and SL46, one of the NB–Kasalath chromosome segment SLs containing the OzT8 locus. The boundaries of the QTL are defined by the nearest restriction fragment length polymorphism (RFLP) genetic markers, R202 and R2676, located on chromosome 8 (Frei et al. 2008). Further details regarding the genetic make-up of SL46, as well as other NB/Kasalath chromosome segment SLs, can be accessed at the website of the Rice Genome Resource Center (http://www.rgrc.dna.affrc.go.jp/ineNKCSSL54.html).
Seeds were surface-sterilized via washing in 0.5 m potassium hypochlorite solution (KOCl) for 10 min, rinsed in water and then placed in Petri dishes in the dark at 27 °C for germination. Upon germination, the seeds were placed on a metal mesh above a solution containing 0.5 mm CaCl2 and grown under low light in a growth cabinet (Conviron, Winnipeg, Canada). Approximately 7 d after germination, the seedlings were transferred to 75 L hydroponic tanks containing 0.5× Yoshida nutrient solution (Yoshida et al. 1976), and grown under greenhouse conditions at a temperature of 28–32 °C during the day and 22–25 °C during the night. After 2 weeks, the nutrient solutions were exchanged for full-strength solutions. The pH of the nutrient solution in each tank was maintained near 5.5 over the duration of the experiments.
Ozone fumigation was conducted using a UV light-based O3 generator (model OES-10A-S; Dylec Inc., Inashiki-gun, Japan) linked to an O3 analyser (model 1150; Dylec Inc.). The concentration of O3 was controlled to within ±15% of the target concentration for the duration of each fumigation.
Two experiments were conducted using this fumigation system.
Experiment 1 was conducted in open-top chambers (0.9 × 1.2 × 1.2 m) consisting of metal pipe frames wrapped in soft transparent plastic sheeting. Ozone was distributed evenly over the span of the open-top chambers using a leaf blower that pushed air through plastic tubing hung above each chamber. The plastic tubing contained small holes set at various angles, which ensured even distribution of ozone over the span of the chambers (specific details regarding the spatial distribution of ozone within the chambers can be found in Frei et al. 2008). These chambers were placed in a climate-controlled greenhouse set to 30 °C during the day and 25 °C at night. Lighting was ambient, typical of springtime in the region of Japan near Tokyo, and day length was approximately 12 h.
At 28 d after germination, the plants were subject to O3 fumigation (100 ppb from 0900 to 1600 h, 23 d) in the open-top chambers (N = 4). Midday photosynthetic gas exchange and chlorophyll fluorescence measurements were taken on the second most recently expanded leaves at days 5 and 23 of the fumigation. In addition, CO2 response curves were taken at day 23. At the end of the experiment, the plants were harvested and placed in a drying oven at 70 °C for at least 3 d, after which whole-plant dry weight was measured.
Experiment 2 was conducted in controlled-environment growth chambers (TGE-2H-2S; Tabai Espec Corp., Osaka, Japan). At 28 d after germination, the plants were moved into the growth chambers and subjected to O3 fumigation (100 ppb from 0900 to 1600 h) for 3 d (N = 3). The growth chambers were set at a temperature of 28 °C during the day and 25 °C during the night, and photosynthetic photon flux density (PPFD) was approximately 400 µmol m−2 s−1 at plant height. Day length was set at 12 h. Steady-state photosynthesis and stomatal conductance were measured on the second most recently expanded leaves of each plant at 0, 3, 6, 30 and 54 h after the start of O3 fumigation.
Midday photosynthetic measurements
Midday ambient carbon assimilation rate (A), stomatal conductance (gs) and chlorophyll fluorescence parameters (ΦPSII, Fv′/Fm′ and Fq′/Fv′, as described in greater detail below) were measured on the second most recently expanded leaf using a portable photosynthetic gas exchange system (LI-6400; Li-Cor Inc., Lincoln, USA) equipped with a leaf fluorescence chamber head (LI-6400-40, Licor Inc., Lincoln, NE, USA). Chlorophyll fluorescence parameters were calculated from saturating flashes that were taken once the raw fluorescence value had stabilized (typically 4–6 min after beginning gas exchange measurements). Leaves were measured under a PPFD of 700 µmol m−2 s−1 in experiment 1, and 400 µmol m−2 s−1 in experiment 2 (based on the ambient light conditions in each growth environment). Other environmental conditions were as follows in both experiments: [CO2] of 400 µmol mol−1, relative humidity of approximately 60–70% and leaf temperature of 28 °C.
Chlorophyll fluorescence terminology
The chlorophyll fluorescence terminology used here follows the convention of Baker & Oxborough (2004). ΦPSII is the operating quantum efficiency of photosystem II (PSII) photochemistry under actinic light. ΦPSII is also known as Fq′/Fm′, which can be broken into two component fluorescence ratios as follows:
where Fm′ is the maximum fluorescence under actinic light; Fv′ is the difference between Fm′ and the minimum fluorescence under actinic light (Fo′); and Fq′ is the difference between Fm′ and F′, the level of fluorescence prior to the saturating flash. Further explanation of these terms can be found in Baker & Oxborough (2004).
Fv′/Fm′ gives an estimate of the potential PSII operating efficiency under the given photon flux density, while Fq′/Fv′ (also commonly referred to as qP) is a non-linear estimate of the fraction of open PSII centres, that is, the oxidation state of QA. The product of these two fluorescence ratios, Fq′/Fm′, is the realized operating efficiency of PSII photochemistry in the light.
Estimation of biochemical photosynthetic capacity
CO2 response curves were measured in experiment 1 using the mentioned portable photosynthesis system. All measurements were taken under a PPFD of 1200 µmol m−2 s−1 (deemed to be saturating for plants grown in the greenhouse), and A was measured at various [CO2] in the following order: 400, 200, 100, 50, 400, 500, 600, 700, 800, 1000 µmol mol−1.
Plots of A versus leaf intercellular [CO2] (Ci) were used to solve for the maximum carboxylation rate of ribulose 1·5-bisphosphate carboxylase/oxygenase (Rubisco) (Vc,max) and the maximum electron transport rate (Jmax) using the equations of Farquhar, Caemmerer & Berry (1980). When necessary, measurements were corrected to 25 °C using temperature response functions from Bernacchi et al. (2001) and Bernacchi, Pimentel & Long (2003) for the Rubisco- and ribulose-1,5-bisphosphate (RuBP)-limited portions of the A–Ci curves, respectively.
Modelled A–Ci curves and stomatal limitation
Theoretical CO2 response curves for the control and ozone treatment of both genotypes were generated using the mean values of Vc,max and Jmax as estimated from the CO2 response curves, according to the equations of the Farquhar et al. (1980) model (as described in Long & Bernacchi 2003). The respective CO2 supply functions were determined by the combined stomatal and boundary layer conductance as measured under ambient conditions near midday at day 23.
Stomatal limitation (l) was estimated for each treatment–genotype combination according to Long & Bernacchi (2003):
where A″ is the CO2 assimilation rate where Ci = Ca = 400 µmol mol−1 (i.e. assuming an infinite stomatal and boundary layer conductance to CO2), and A′ is the CO2 assimilation rate at which the CO2 supply curve and the A–Ci curve intersect.
Data from experiment 1 were analysed as a randomized complete block design using PROC MIXED in SAS 9.1 (Cary, NC, USA). Data from experiment 2 were analysed as a mixed model, repeated measures randomized complete block design. Mean comparisons were done using linear contrasts in a two-way analysis of variance (ANOVA). The best-fit variance/covariance matrices for each variable were chosen using Akaike's information criterion (Littell, Henry & Ammerman 1998; Littell, Pendergast & Natarajan 2000).
At the end of experiment 1, the dry weight of NB in the control was similar to that of SL46 (Fig. 1a). However, there was a significant decrease in dry weight under ozone in both genotypes (34% for NB and 15% for SL46), and the ozone-by-genotype effect was significant (P = 0.03). In the ozone treatment, the dry weight of SL46 was significantly higher than that of NB (Fig. 1a, P = 0.013). Similar trends were observed in both the aboveground and belowground biomass (Fig. 1b,c); as a result, the root-to-shoot ratio was unchanged by ozone treatment in either genotype (Fig. 1d).
Daily midday photosynthetic rates, stomatal conductance and chlorophyll fluorescence
By day 5 of the ozone fumigation in experiment 1, there was a significant decrease in midday carbon assimilation rate, as well as stomatal conductance in both genotypes (Tables 1 & 2). However, by the third week of ozone fumigation, SL46 showed significantly less reduction of carbon assimilation rate under ozone stress compared to NB (25 and 41%, respectively; P = 0.015; Table 1), but there was no genotypic difference in stomatal conductance (Table 2).
Table 1. Midday net photosynthetic carbon assimilation rates (A) and stomatal conductance (gs) from experiment 1
Table 2. P values from the analysis of variance (anova) of midday photosynthetic data from experiment 1
Values less than 0.05 are indicated in bold.
Ozone × genotype
Ozone × genotype
At day 5 of the fumigation, there was no significant difference between NB and SL46 in any of the chlorophyll fluorescence parameters (Tables 2 & 3). However, by day 23, the ΦPSII of NB was significantly decreased by ozone (−23%), whereas SL46 showed no significant difference between control and ozone treatments. Fv′/Fm′ showed a statistically significant decrease in SL46 under ozone, but the magnitude of the response was very similar between genotypes (−6.1 and −8.7% for NB and SL46, respectively). However, the response of Fq′/Fv′ to ozone differed between the genotypes (Table 2). Fq′/Fv′ did not significantly change in SL46, but dropped by 17.6% in NB.
Table 3. Chlorophyll fluorescence data from days 5 and 23 of experiment 1
Photosynthetic biochemical capacity
At the third week of fumigation, SL46 showed significantly less reduction of biochemical photosynthetic capacity under ozone stress compared to NB (Fig. 2). The maximum carboxylation capacity of Rubisco (Vc,max) decreased by 24% in SL46 compared to 49% in NB (P < 0.001). The maximum electron transport rate (Jmax) decreased by 16 and 39%, respectively (P = 0.004). The large decrease in Vc,max in NB under ozone stress resulted in a significant decrease of the Vc,max:Jmax ratio in NB, but the ratio was unchanged in SL46 (Fig. 2c).
Modelled A–Ci curves and stomatal limitation
In order to examine the combined effects of decreased stomatal conductance and decreased photosynthetic capacity under ozone stress, theoretical CO2 response curves for control and ozone-treated NB and SL46 plants were modelled using the mean Vc,max and Jmax values shown in Fig. 2. CO2 supply functions were drawn based on the mean stomatal conductance of leaves measured near midday under ambient conditions at day 23 (Table 1). Stomatal limitation was estimated at ambient [CO2] by determining the intersection of the CO2 response curve with its corresponding CO2 supply function (Fig. 3). Under control conditions, stomatal limitation was 0.083 for NB and 0.126 for SL46 (Table 4). In the ozone treatment, stomatal limitation increased in both genotypes and was similar in both genotypes (0.212 for NB, 0.218 for SL46).
Table 4. Predicted values of net photosynthesis (A) at infinite stomatal conductance (i.e. when Ca = Ci = 380 µmol mol−1), at the actual mean stomatal conductance measured during midday at day 23 of experiment 1, and the estimated stomatal limitation to photosynthesis
At infinite gs
At measured gs
Short-term photosynthetic response to ozone
A second ozone fumigation (experiment 2) was conducted in growth chambers to investigate the short-term response of both NB and SL46 upon initial exposure to ozone. Measurements of photosynthesis and stomatal conductance were taken at several time-points over the first 3 d of fumigation (Fig. 4). There were no significant genotype-by-ozone interaction effects for photosynthesis; however, there was a significant ozone effect (P < 0.05) within 6 h of ozone exposure (Fig. 4a; Table 5). This difference persisted over the next 2 d. At 54 h after the beginning of ozone exposure (midday of the third day), the net carbon assimilation rate for NB showed a decrease of 23.4% under ozone stress, whereas the photosynthetic rate of SL46 was decreased by only 14.0%.
Table 5. P values from the analysis of variance (anova) of the short-term response to ozone in experiment 2
Values less than 0.05 are indicated in bold.
Ozone × genotype
Ozone × time
Genotype × time
Ozone × genotype × time
For stomatal conductance, significant treatment effects were seen in both genotypes as soon as 3 h into the exposure, and persisted thereafter (Fig. 4b). In contrast with photosynthesis, there was a statistically significant ozone-by-genotype effect over the overall course of the fumigation (Table 5). By midday of the third day (54 h after the start of ozone fumigation), SL46 in the ozone treatment maintained 71.8% of its non-stress stomatal conductance, compared to 43.9% for NB.
The effect of OzT8 on photosynthesis in rice under ozone stress
A previous ozone tolerance screening of various rice cultivars and subsequent QTL mapping study identified several QTLs associated with higher or lower ozone tolerance (Frei et al. 2008). A subsequent work has confirmed and characterized two of these QTLs (OzT9 and OzT3), which are associated with higher and lower leaf bronzing under ozone stress (Frei et al. 2010). However, this study is the first to confirm the dry weight effect of the QTL OzT8 through the SL SL46, and found that this effect is accompanied by a higher relative midday photosynthetic carbon assimilation rate under ozone stress (Fig. 1; Table 1). Although the photosynthetic rate in ozone was higher in SL46 than in NB, stomatal conductance showed no significant difference between the two genotypes (Table 1). In fact, in some instances, it appeared that stomatal conductance under ozone stress was higher in SL46 than in NB (Fig. 4b). Therefore, we can confidently reject the hypothesis that OzT8 confers ozone tolerance via stomatal closure and avoidance of ozone.
However, the hypothesis that OzT8 is associated with higher biochemical photosynthetic capacity under ozone stress was confirmed (Fig. 2). Therefore, the results here show that the underlying mechanism of ozone tolerance associated with the OzT8 locus is relative maintenance of leaf photosynthetic capacity and daily carbon assimilation rate while under ozone stress.
The comparison of midday photosynthetic rate, stomatal conductance and chlorophyll fluorescence at days 5 and 23 of the ozone fumigation (Tables 1–3) showed a notable difference between leaves that had developed prior to onset of fumigation, and leaves that had developed during the treatment. Although there was a significant decrease in photosynthetic rate and stomatal conductance under ozone in both genotypes at day 5, there was no treatment effect for ΦPSII or its component fluorescence ratios (Table 2). However, for leaves that developed and grew under constant ozone stress, the ΦPSII was significantly decreased in NB, but not in SL46. This was a direct result of a significant decrease in qP in NB, which is non-proportionally related to the oxidation state of QA, that is, processes downstream of the PSII reaction centre. This result points towards the presence of ozone effects on Rubisco carboxylation capacity and the Calvin cycle, and is confirmed by a large observed decrease in Vc,max in NB (Fig. 2). This is consistent with the hypothesis in the literature that one of the major impacts of chronic [O3] is its negative effect on Rubisco (Farage & Long 1999; Goumenaki et al. 2010).
Leaves that developed under control conditions and were exposed to ozone responded differently than leaves that had developed under ozone-stressed conditions. This suggests that care should be taken when conducting experiments using only short-term ozone fumigations (e.g. several hours for a single day), because some effects may not appear within such a short treatment period. Other studies have also found that different physiological mechanisms occur in response to ozone at different timescales (Chen, Frank & Long 2009).
The analysis of the theoretical CO2 response curves indicated that stomatal limitation for both genotypes was similar under ozone stress (Table 4). This further makes the case that the higher midday photosynthetic rates in SL46 over NB under ozone stress could be attributed primarily to its higher photosynthetic capacity (Vc,max and Jmax) under stress conditions, and not caused by any differences in stomatal response.
If resistance to ozone stress in SL46 cannot be attributed to differences in ozone uptake into the plant, what are other possible reasons? Previous research has suggested that differences in antioxidant and radical-scavenging capability, especially in the apoplastic space, may confer resistance to ozone effects (Conklin & Barth 2004). Another possibility is that the defence response and signalling pathways that typically accompany ozone stress are altered, resulting in lower levels of key stress hormones such as ethylene, which would then limit the defence response-induced oxidative stress and foliar injury, as well as acceleration of senescence and down-regulation of Rubisco (Vahala et al. 2003; Kangasjarvi et al. 2005).
Under control conditions, the Vc,max and Jmax of SL46 were slightly lower than those of NB (Fig. 2); stomatal conductance was also lower in SL46 at some time-points (Fig. 4). This could potentially be attributed to a trade-off between stress tolerance and maximum achievable productivity, but could also be caused by other Kasalath chromosome inserts in SL46 that are not associated with ozone stress tolerance. However, a comparison of the grain yield between NB and SL46 under non-stressed conditions showed no significant difference at the end of the growing season (Chen et al. in press). Future fine-mapping of the Kasalath inserts in SL46 via additional SL46/NB crosses will be able to clarify this issue. Regardless, the results of this current study indicate a strong potential for identification of novel genes associated with increased ozone tolerance in rice within the OzT8 locus.
The sequence of ozone effects on photosynthesis in rice
Some previous literature has suggested that stomatal response to ozone is mediated via changes in the mesophyll, that is, the decrease of stomatal conductance under ozone stress follows the decrease in photosynthetic biochemical capacity. This hypothesis was confirmed in oak under acute ozone stress conditions, and in wheat and pea plants exposed to either acute or chronic ozone conditions (Farage & Long 1995, 1999). In particular, it was found that decreases in Rubisco carboxylation capacity preceded any changes in stomatal conductance or other photosynthetic components.
However, the results of this study indicate that the short-term response to ozone in the case of rice is fundamentally different. Stomatal conductance in rice plants responded within 3 h of exposure to a moderate concentration of ozone and precluded the decrease in steady-state photosynthetic rate (Fig. 4). This result implies that the plant was able to sense the presence of ozone and respond by closing its stomata even before the photosynthetic capacity in the mesophyll had been affected. Similar results have been found in the rice cultivar Koshihikari, whose stomatal conductance showed a significant decrease within even the first hour of exposure to ozone (Imai & Kobori 2008). This rapid closure of stomata upon exposure to even moderate levels of ozone may prevent a significant amount of ozone from entering into the leaf intercellular space in the field, and may provide a partial explanation for why rice has been found to be relatively tolerant to ozone compared to other agricultural crop species. However, this tolerance mechanism is only viable in the short term. If the plants are exposed to high ozone concentrations on a daily basis and react by consistently decreasing their stomatal conductance, the photosynthetic rate and long-term growth or biomass accumulation will suffer. Therefore, they must develop some mechanism of acclimation or biochemical tolerance to allow the stomata to remain open enough to conduct the gas exchange necessary for photosynthesis without accumulating additional damage from ozone. The exact mechanism for this long-term tolerance is not yet clear; however, the results of this study suggest that the QTL OzT8 may confer this type of ozone tolerance.
The authors thank Dr. Juan Pariasca Tanaka for his logistical assistance and constructive comments, and Dr. Hidemitsu Sakai for the use of his gas exchange equipment. This work was supported by the Japan Society for the Promotion of Science Postdoctoral Fellowship.