Rubisco activity in Mediterranean species is regulated by the chloroplastic CO2 concentration under water stress.

Water stress decreases the availability of the gaseous substrate for ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) by decreasing leaf conductance to CO(2). In spite of limiting photosynthetic carbon assimilation, especially in those environments where drought is the predominant factor affecting plant growth and yield, the effects of water deprivation on the mechanisms that control Rubisco activity are unclear. In the present study, 11 Mediterranean species, representing different growth forms, were subject to increasing levels of drought stress, the most severe one followed by rewatering. The results confirmed species-specific patterns in the decrease in the initial activity and activation state of Rubisco as drought stress and leaf dehydration intensified. Nevertheless, all species followed roughly the same trend when Rubisco activity was related to stomatal conductance (g(s)) and chloroplastic CO(2) concentration (C(c)), suggesting that deactivation of Rubisco sites could be induced by low C(c), as a result of water stress. The threshold level of C(c) that triggered Rubisco deactivation was dependent on leaf characteristics and was related to the maximum attained for each species under non-stressing conditions. Those species adapted to low C(c) were more capable of maintaining active Rubisco as drought stress intensified.

Introduction

Worldwide, water availability is considered the environmental factor that most strongly influences plant growth and yield (Boyer, 1982), and global climate change is expected to exacerbate water limitations in semi-arid ecosystems like the Mediterranean (IPCC 2007). Although photosynthesis has frequently been documented among the primary physiological processes affected by water shortage, a controversy still exists on whether drought effects on photosynthesis are due to ‘stomatal’ or ‘non-stomatal’ limitations (Flexas and Medrano, 2002; Lawlor and Cornic, 2002). Stomatal limitations result from the resistance to CO2 diffusion from the atmosphere to intercellular leaf spaces, while non-stomatal limitations are often assumed to be metabolic constraints. However, there is increasing evidence that the conductance determining transfer of CO2 from the intercellular leaf spaces to the vicinity of ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco), i.e. the mesophyll conductance, gm, has a finite value and frequently decreases under drought (Flexas ; Warren, 2008). Therefore, the controversy of ‘stomatal’ versus ‘non-stomatal’ limitations of photosynthesis has been replaced by ‘diffusive’ versus ‘metabolic’ (or biochemical) limitations (Grassi and Magnani, 2005; Galmés ). While it is well established that under mild to moderate stress, drought-induced reductions in photosynthesis are mainly due to diffusive limitations—i.e. decreased stomatal and mesophyll conductance—(Chaves ; Flexas and Medrano, 2002), there is less of a consensus about the limiting process when drought intensity is increased. Similarly, there is much less information on the limitations to photosynthesis recovery by severely stressed plants when water is restored (Galmés ; Gallé ).

Current atmospheric CO2 concentration is well below saturating for photosynthesis, so the metabolic rate that limits photosynthesis in the majority of plants is carboxylation by Rubisco rather than electron transport or RuBP regeneration, although sometimes a co-limitation of both metabolic components has been suggested (Lambers ). Under drought, CO2 availability in leaves is further reduced because of stomatal closure, hence leading plants to operate in a CO2 range where Rubisco carboxylation is the sole metabolic rate limiting photosynthesis. Therefore, despite claims that drought impairs chloroplast ATP synthesis and RuBP regeneration (Tezara ), if any metabolic limitation occurs under drought this is very likely to be mediated by Rubisco activity (Flexas ; Grassi and Magnani, 2005). Rubisco activity can be estimated by calculations from in vitro and/or in vivo measurements (von Caemmerer, 2000). Estimates from in vivo measurements are based on the assumptions of the Farquhar model adapted to finite gm—i.e. calculation of the initial slope of the AN–Cc curve or Vc,max—(Farquhar ; Niinemets ), and usually show that the carboxylation capacity is preserved until water stress becomes severe (Bota ; Grassi and Magnani, 2005; Galmés ; Gallé ). On the contrary, inconsistencies arise from in vitro measurements concerning the metabolic impairment of photosynthesis, and particularly of Rubisco activity, under drought. The reported effects of water stress on the activity of Rubisco differ between reports, precluding a definitive conclusion. Hence, observed effects range from major reductions of Rubisco activity (Majumdar ; Maroco ; Parry ), to minor decreases (Flexas ), and even no effect at all (Giménez ; Gunasekera and Berkowitz, 1993; Lal ; Pankovic ; Delfine ; Pelloux ). This apparent controversy may be attributed to differences in the intensity of drought (Flexas ). Alternatively, it may arise from interspecific differences, but there are studies reporting contradictory effects of water stress on Rubisco activity for the same species. For instance, Parry showed, in tobacco plants, that down-regulation of Rubisco activity occurred concomitantly with decreases in the leaf relative water content (RWC), while Flexas found no correlation between these parameters also in tobacco. In an attempt to explain such discrepancies, Flexas demonstrated that differences between species disappeared when Rubisco activity was plotted against stomatal conductance (gs) rather than RWC. However, a direct relationship between gs and Rubisco activity is unlikely. It is possible, though, that this observation might be due to an indirect effect, related to the influence of gs on the CO2 concentration in leaves which, in turn, also depends on the rate of assimilation, controlled by Rubisco activity (Flexas ).

Rubisco activity depends on (i) its activation by reaction with CO2, resulting in the carbamylation of a lysyl residue in its active site, (ii) the concentration of Rubisco, and (iii) the concentration of tight-binding inhibitors (Roy and Andrews, 2000). An inverse correlation between the carbamylation state and Rubisco concentration has been reported (Quick ; Ruuska ; Eichelmann ). Changes in the concentration of Rubisco have been observed under water stress (e.g. Holaday ), but even in this situation Rubisco concentration is considered to be present sufficiently in excess not to limit photosynthesis (Pickersgill, 1986; Warren and Adams, 2004). Increases in Rubisco tight-binding inhibitors have also been reported under conditions of water shortage, although this has not been firmly established (Parry ). Therefore, adjustments of Rubisco activity under water stress are mainly attributed to changes in the carbamylation and activation state. In situations where light is saturating and leaf conductance does not limit CO2 supply (i.e. in the absence of stress), the activation state of Rubisco is maximum (Sage ). However, water stress, through changes in gs and gm, induces a decrease in the concentration of CO2 in leaves and therefore in the amount of activator CO2 bound by carbamylation to Rubisco. Deactivation of Rubisco by loss of the carbamate complex in the active site has been observed at substomatal CO2 concentrations (Ci) of <100 μmol mol−1 in some species, but not in others (Perchorowicz and Jensen, 1983; von Caemmerer and Edmondson, 1986; Sage ). This threshold of internal CO2 concentration coincides roughly with the threshold of chloroplastic CO2 concentration (Cc) inducing decreases in in vivo Vc,max in several species (Galmés ) and in vitro Rubisco activity in tobacco and soybean (Flexas ). Overall, this evidence suggests that drought-driven effects on Rubisco activity are caused by CO2 starvation and its consequent de-carbamylation of the Rubisco catalytic sites. Therefore, a survey relating Rubisco activity and Cc for a larger number of species subjected to gradual imposition of water stress and recovery is justified and would help in defining the causes of metabolic limitation of photosynthesis under drought.

Problems exist when directly relating in vitro measurements of Rubisco activity to estimates of Rubisco activity derived from gas exchange measurements in leaves. Incomplete extraction of Rubisco, prior to activity analysis, has been shown to be responsible for inadequate activity to support photosynthetic rates measured by gas exchange in loblolly pine (Rogers ). Rubisco extractions from leaves are often incomplete even when care is taken to prevent inactivation and precipitation by high contents of phenolic or other secondary compounds (Beadle ; Keys and Parry, 1990). There are also inaccuracies in estimating Rubisco activity from in vivo measurements of gas exchange in leaves and especially in the determination of gm needed for the model used. The combined gas exchange and chlorophyll fluorescence methods for calculating gm rely on the estimation of parameters requiring a number of assumptions and subject to technical limitations, which become relatively more important in certain groups of species, like sclerophylls, with leaves covered by hairs, and in heterobaric leaves (Flexas ; Pons ). Despite these difficulties, qualitative relationships of photosynthetic activity can be seen between in vitro and in vivo estimates.

Here, a set of data showing the effects of water stress on Rubisco activity for 11 Mediterranean species from contrasting environments grown under common conditions is presented. The objectives of the present work were: (i) to test the relationship between in vitro measurements and in vivo estimations of Rubisco activity; (ii) to check whether there is a general pattern in the response of Rubisco activity to drought stress in different species, and if so, (iii) to discern whether drought-driven changes in Rubisco activity are related to Cc.

Material and methods

Species selection and treatments

Eleven Mediterranean species naturally occurring in the Balearic Islands, six of them endemic to these islands, were selected for this study (see Galmés for detailed description of the species). The selection of the species included taxons representing different growth forms and leaf habits: two evergreen sclerophyll shrubs (Pistacia lentiscus and Hypericum balearicum), two evergreen sclerophyll semi-shrubs (Limonium gibertii and Limonium magallufianum), three summer semi-deciduous shrubs (Lavatera maritima, Phlomis italica, and Cistus albidus), and four herbaceous (Beta maritima ssp. maritima, Beta maritima ssp. marcosii, Diplotaxis ibicensis, and Lysimachia minoricensis). D. ibicensis, B. maritima ssp. maritima, B. maritima ssp. marcosii, L. gibertii, L. magallufianum, and L. maritima are species inhabiting the coastal, driest, and hottest areas with annual precipitation typically <400 l m−2. P. lentiscus, H. balearicum, P. italica, and C. albidus are species typical of Mediterranean macchia with annual precipitation between 400 and 800 l m−2. Finally, L. minoricensis lives near open water sources, and probably suffers less from water stress episodes.

The ages of the plants differed because of the different phenology of the species selected. Plants of P. lentiscus, H. balearicum, C. albidus, P. italica, and L. maritima were 3 years old, plants of L. magallufianum, and L. gibertii were 18 months old, and plants of D. ibicensis, B. maritima ssp. marcosii, and B. maritima ssp. maritima were 6 months old at the onset of the experiments. Similarly, the different phenology also influenced the age of leaves used for measurements, which was ∼1 month for those species with the highest rate of leaf emergence (i.e. herbaceous species), and ∼2–4 months for the remaining species.

Water treatments are explained in detail in Galmés . In brief: 10 adult plants per species were placed in a controlled growth chamber with a 12 h photoperiod (26° C day, 20° C night) and a photon flux density at the top of the leaves of ∼700 μmol m−2 s−1. Plants were irrigated daily with 50% Hoagland's solution for 1 month. Measurements corresponding to well-watered treatment (WW) were made during the first day of the experiment and corresponded to plants watered at soil saturation. Thereafter, irrigation was stopped in five plants for each species. Pots were weighed daily to determine the amount of water loss. Plant water availability was referred to the control after measurement of soil dry weight in four samples representative of the substrate mixture used in the experiment. Measurements were performed on days 4, 8, and 13–17 after the last irrigation, when plants were regarded as affected by mild (MiWS), moderate (MoWS), and severe water stress (SeWS), respectively. Plants were considered to be under SeWS when gs was close to zero, which was achieved between 13 and 17 d after withholding water, depending on the species. At this time, pots were rewatered to field capacity, and recovery treatment (RW) was determined on the next day. Control plants were watered daily throughout the experiment and measured every 5–6 d to ensure they had maintained constant values.

Plant water status

Parameters related to plant water status, leaf relative water content (RWCPD) and water potential (ΨPD) were measured pre-dawn as indicated in Galmés .

Photosynthetic measurements

Chlorophyll fluorescence parameters were measured on attached leaves using a portable pulse amplitude modulation fluorometer (PAM-2000; Walz, Effeltrich, Germany). At mid-morning, the PSII photochemical efficiency (ΔF/Fm′) was determined by measuring the steady state fluorescence (Fs) and the maximum fluorescence (Fm′) during a light-saturating pulse of ∼8000 μmol m−2 s−1, following Genty :

The electron transport rate (ETR) was then calculated as:

where PPFD is the photosynthetically active photon flux density, α is the leaf absorptance, and β is the distribution of absorbed energy between the two photosystems. β was assumed to be 0.5. Leaf absorptances were determined for all 11 species in 10 replicates on leaves of well-irrigated plants with a spectroradiometer coupled to an integration sphere (UniSpec; PP-Systems, Amesbury, MA, USA). A value of 0.84 was obtained for all species, except for C. albidus and P. italica (0.74 and 0.77, respectively). Potential changes in leaf absorptance with water stress were not assessed and assumed to be small and to induce no important biases in the calculations of ETR.

Light-saturated net CO2 assimilation rates (AN) and gs were measured at mid-morning on attached, fully developed young leaves of four to five plants per species and treatment, using a gas exchange system (Li-6400; Li-Cor Inc., Lincoln, NE, USA). Environmental conditions in the leaf chamber consisted of a PPFD of 1500 μmol m−2 s−1, a vapour pressure deficit of 1.0–1.5 kPa, an air temperature of 25º C, and an ambient CO2 concentration (Ca) of 400 μmol mol air−1. After inducing steady-state photosynthesis, the photosynthesis response to varying substomatal CO2 concentration (Ci) was measured as explained in Galmés . To transform AN–Ci curves to AN–Cc curves, the CO2 concentration in the chloroplasts (Cc) was calculated from combined gas exchange and chlorophyll fluorescence measurements according to Epron and using the in vitro Rubisco specificity factor values from Galmés .

Biochemical measurements

For measurements of Rubisco activity, two leaf discs (1 cm2 each) were punched, frozen immediately in liquid N2, and kept at –80° C until assay. Four to five replicates per species and treatment were taken at midday immediately after steady-state gas exchange measurements. Environmental conditions in the leaf chamber were identical to those explained above in the photosynthetic measurements section.

For Rubisco activity analysis, samples were ground to a fine powder in a mortar, previously chilled with liquid nitrogen, and homogenized in 1 ml of an ice-cold extraction medium. Preliminary tests determined that the most appropriate protein extraction media for Rubisco (i.e. that yielded the maximum activity for each species) were: (A) 0.1 M Bicine, 50 mM β-mercaptoethanol, 11 mM sodium diethyldithiocarbamate (Na-DIECA), 6% (w/v) polyethyleneglycol (PEG) 4000, 1 mM benzamidine, 1 mM e-amino-n-caproic acid, and 1 mM phenylmethylsulfonylfluoride (PMSF), at pH 8, and (B) containing 0.1 M HEPES, 3% (w/v) polyvinylpyrrolidone (PVP) 25, 6% (w/v) PEG 4000, 50 mM β-mercaptoethanol, 2 mM dithiothreitol (DTT), 10% glycerol, 5 mM MgCl2, 5 mM ethyleneglycol-bis(b-aminoethylether)-N,N,N'N'-tetraacetic acid (EGTA), and 2 mM PMSF, at pH 8.0. Buffer A was used to extract Rubisco from most of the species. Buffer B was used with P. lentiscus and C. albidus. Extracts were clarified by centrifugation (12 000 rpm at 4° C for 1 min) and the supernatant immediately assayed at 25° C for Rubisco activity.

Initial and total activities were determined according to Keys and Parry (1990). Initial activity was determined by adding 25 μl of supernatant to 475 ml of a CO2-free assay buffer containing 100 mM Bicine, pH 8.2, and 20 mM MgCl2, to which NaH14CO3 (7.4 kBq μmol−1) and RuBP had been added, to concentrations of 10 and 0.4 mM, respectively, immediately prior to adding the extract. Total activity was determined by incubating 20 ml of extract for 3 min in 980 ml of the same assay buffer without RuBP and to allow for the carbamylation of all available active sites. The assay was started by adding RuBP to 0.4 mM as above. Rubisco activation state was obtained from the ratio of initial to total activity.

Total soluble protein was determined on the same samples according to the method of Bradford (1976).

Statistical analyses

Regression coefficients and bivariate correlations were calculated with the 11.0 Sigma Plot software package (SPSS, Chicago, IL, USA). A set of simple ANOVAs were performed to compare different treatments. Differences between means were revealed by Duncan analyses (P<0.05) performed with the SPSS 17.0 software package.

Results and discussion

Progressive effects of water stress on Rubisco parameters

The response of pre-dawn leaf water potential (ΨPD) and relative water content (RWCPD) to water stress and recovery during this experiment has previously been reported (Galmés ). Leaf RWCPD and ΨPD decreased progressively with soil water content (SWC) in all species (Table 1). The largest decreases in RWCPD were observed at SeWS, and ranged from 38% for P. italica to 70% for L. magallufianum. Similarly, at SeWS ΨPD varied from less than –5 MPa for P. italica to –1 MPa for D. ibicensis. The extent of recovery 24 h after rewatering severely stressed plants depended on the species: all species except P. lentiscus and P. italica recovered RWCPD and ΨPD to values similar to those measured under WW conditions.

Table 1.

SWC (expressed as a percentage of well-watered treatments), leaf RWCPD, and leaf ΨPD for the 11 species under the different treatments

Species		WW	MiWS	MoWS	SeWS	RW
D. ibicensis	SWC (%)	100.0±0.0	68.6±3.4	37.5±5.3	26.7±5.8	89.7±4.1
	RWCPD	80.2±1.3	70.4±1.4	67.5±3.2	62.3±7.2	79.9±2.8
	ΨPD	–0.43±–0.03	–0.69±–0.02	–0.94±–0.06	–1.00±–0.11	–0.98±–0.03
L. minoricensis	SWC (%)	100.0±0.0	65.6±2.6	58.4±3.6	49.5±4.6	90.6±5.7
	RWCPD	76.1±0.9	70.6±6.2	73.5±4.2	44.7±3.8	71.8±3.6
	ΨPD	–0.57±–0.08	–1.19±–0.31	–1.31±–0.39	–2.79±–0.26	–0.64±–0.07
B. maritima ssp. marcosii	SWC (%)	100.0±0.0	62.8±1.5	19.5±5.0	7.4±5.0	90.8±2.1
	RWCPD	80.1±3.8	87.3±2.1	78.0±1.7	51.1±5.7	86.7±2.0
	ΨPD	–0.33±–0.01	–0.45±–0.03	–0.89±–0.04	–3.34±–0.10	–0.52±–0.06
B. maritima ssp. maritima	SWC (%)	100.0±0.0	56.9±5.0	26.8±1.2	12.5±5.3	91.4±3.2
	RWCPD	78.0±0.9	80.2±4.1	79.0±2.5	51.0±3.5	84.7±1.9
	ΨPD	–0.33±–0.01	–0.44±–0.05	–0.80±–0.05	–3.74±–0.09	–0.43±–0.01
L. magallufianum	SWC (%)	100.0±0.0	54.2±2.3	46.0±1.5	40.1±1.3	98.7±2.8
	RWCPD	88.9±0.5	85.8±2.0	80.3±2.3	69.5±1.6	87.0±2.7
	ΨPD	–0.53±–0.03	–0.55±–0.07	–0.65±–0.07	–1.38±–0.14	–1.02±–0.03
L. gibertii	SWC (%)	100.0±0.0	56.8±3.4	49.1±3.4	44.6±3.4	92.3±2.5
	RWCPD	90.6±0.9	88.3±1.9	77.6±0.8	66.8±5.1	87.7±4.2
	ΨPD	–0.55±–0.03	–0.49±–0.04	–0.61±–0.03	–1.05±–0.09	–0.93±–0.07
H. balearicum	SWC (%)	100.0±0.0	85.9±7.3	52.7±7.0	30.7±5.9	85.4±4.4
	RWCPD	91.2±1.3	93.1±1.1	89.9±1.3	48.7±5.2	85.5±2.4
	ΨPD	–0.38±–0.00	–0.43±–0.02	–0.66±–0.05	–2.96±–0.16	–0.94±–0.14
P. lentiscus	SWC (%)	100.0±0.0	61.4±7.6	25.5±8.2	14.6±7.1	88.6±5.5
	RWCPD	94.8±0.5	87.5±3.3	86.1±3.8	53.7±4.9	81.2±5.4
	ΨPD	–0.30±–0.03	–0.45±–0.05	–1.49±–0.58	–4.55±–0.45	–2.68±–0.28
P. italica	SWC (%)	100.0±0.0	74.8±4.6	57.0±1.8	47.7±1.8	114.7±5.4
	RWCPD	83.6±1.5	80.0±0.8	75.6±0.8	37.9±3.0	74.4±1.6
	ΨPD	–0.33±–0.01	–0.42±–0.03	–1.98±–0.11	< –5.00	–0.65±–0.05
C. albidus	SWC (%)	100.0±0.0	72.7±4.9	53.4±1.8	42.7±2.9	104.4±1.6
	RWCPD	75.7±4.3	88.3±4.3	79.5±4.0	46.0±6.1	70.2±1.5
	ΨPD	–0.51±–0.04	–0.68±–0.06	–1.14±–0.14	–4.12±–0.56	–2.10±–0.10
L. maritima	SWC (%)	100.0±0.0	57.7±6.1	35.9±6.8	27.0±6.0	91.1±3.4
	RWCPD	86.6±2.3	82.2±2.5	73.4±3.5	54.8±5.5	80.6±3.2
	ΨPD	–0.41±–0.05	–0.89±–0.09	–1.53±–0.13	–3.54±–0.25	–0.73±–0.21

Values are means±standard errors of four to five replicates. Data taken from Galmés .

All species showed a progressive decline in the Rubisco initial activity as water stress intensified, starting at MiWS, except for the two Beta spp. (Fig. 1). Rubisco activation state followed a slightly different pattern, maintaining values similar to those in irrigated plants under MiWS to MoWS, depending on the species, and declining under SeWS (Fig. 1). This trend is illustrated by the biphasic relationship between activation state and initial activity (Fig. 2), with an initial phase where activation state remains constant while initial activity decreases followed by a second phase where decreases in both parameters occurred concurrently. This fact suggests that under MiWS to MoWS, decreases in Rubisco activity are mainly due to decreases in the concentration of Rubisco reaction sites (i.e. initial and total Rubisco activity decrease concomitantly), either due to decreases in the enzyme concentration or to increases in the inhibitors bound to the sites of reaction. Early drought-induced losses of Rubisco have been reported by both transcriptomic (Kawaguchi ) and proteomic approaches (Holaday ), although increases have also been reported in some species (Cramer ; Vincent ). The species selected in this study, although inhabiting contrasting environments, are typically Mediterranean, and therefore adapted to frequent but unpredictable drought events (Galmés ). Moreover, due to their relatively low gs and gm (Galmés ), they often present Cc values close to or below the threshold of 100 μmol mol−1, i.e. they operate normally close to conditions favourable for Rubisco inactivation (Perchorowicz and Jensen, 1983). Among potentially advantageous adaptations to such stress conditions, interactions of Rubisco with tight-binding inhibitors would prevent Rubisco that is not being used for catalysis from being degraded by proteases (Parry ). Thereafter, at MoWS to SeWS, deactivation of Rubisco becomes the predominant cause for decreased Rubisco activity, in agreement with previous reports (Sharkey and Seemann, 1989). When exposed to severe water stress, many plants accumulate osmolytes, such as proline, glycine betaine, and sugar alcohols (Yoshiba ). In vitro analyses demonstrate that high concentrations of these osmolytes have potential to curtail the activity of Rubisco (Sivakumar ; Sivakumar ). Therefore, the observed decrease in Rubisco activity at SeWS could be, at least partially, due to the accumulation of osmolytes.

Fig. 1.

Rubisco initial activity (filled symbols) and activation state (open symbols) under different treatments: WW, MiWS, MoWS, SeWS, and RW. Values are means±SE of four to five replicates per species and treatment. Different letters denote statistical differences by Duncan test (P<0.05) between treatments for each parameter.

Fig. 2.

Relationship between Rubisco initial activity and activation state. Values are means±SE of four to five replicates per species and treatment. Values from the RW treatment were excluded from this analysis. Symbols and species as follows: black circles, D. ibicensis; black squares, L. minoricensis; black triangles, B. maritima ssp. marcosii; upside down black triangles, B. maritima ssp. maritima; light-grey squares L. magallufianum; light-grey circles, L. gibertii, dark-grey circles, C. albidus; dark-grey, L. maritima; dark-grey triangles, P. italica; open circles, H. balearicum; and open squares, P. lentiscus.

The recovery of Rubisco initial activity and activation state after SeWS was also species dependent. In general, activation state recovered better than initial activity, although with some exceptions, such as the two Limonium species (Fig. 1). This pattern is indicative of higher recovery of total activity as compared with initial activity, and consequently that an increase in the concentration of Rubisco active sites occurred after rewatering (Parry ).

Relationship between in vitro and in vivo estimates of Rubisco activity

Rapid, one-step homogenization procedures used for Rubisco extraction are required to preserve enzyme activation state (Keys and Parry, 1990). However, Rogers demonstrated that these procedures fail to efficiently extract Rubisco from leaves, and therefore lead to underestimation of Rubisco activity. In the present study, Rubisco activity was ∼20% of the estimated maximum velocity of carboxylation (Vc,max) in all 11 species (data not shown). Fig. 3 shows the relationship between Rubisco initial activity and Vc,max, both parameters expressed as a percentage of the values at WW, to unify axis scales. Despite such bias, the relationship between Rubisco initial activity and Vc,max was linear and significant for each species (i.e. the lowest R2 was for H. balearicum with 0.522), indicating that in vitro measurements of Rubisco activity were qualitatively consistent with in vivo values, as previously reported (e.g. Myers ; Rogers ). On the other hand, it is also documented that the effectiveness of Rubisco extraction and assay procedures is dependent on the presence of phenols and other secondary metabolites (Beadle ; Keys and Parry, 1990), which varies depending on the species and stress conditions (Turtola ; André ). This fact could explain differences in the slope of the relationship between Rubisco initial activity and Vc,max between species (Fig. 3). In this sense, D. ibicensis, B. maritima ssp. maritima, P. italica, and L. maritima were the species presenting the highest slope, while L. magallufianum, L. gibertii, and B. maritima ssp. marcosii presented the lowest slope.

Fig. 3.

Relationship between Rubisco initial activity and maximum velocity of carboxylation (Vc,max) for the 11 species. To unify scales between different species, values are expressed in percentage with respect to WW plants, and represent means±SE of four to five replicates per species and treatment. Empty symbols represent values from RW treatments. Regression coefficients for each relationship are shown.

Alternatively, it may be argued that the estimation of Vc,max might be biased, due to the multiple assumptions included in its calculation. To test this possibility, it may be necessary to compare Rubisco activity values with some in vivo parameter, such as the ETR, which is more straightforward to determine. A ‘photosynthetic control’ exists in leaves where the balance between ΔpH, ATP synthesis, and redox state adjust supply to demand so that the [ATP]/[ADP] and [NADPH]/[NADP+] ratios are remarkably constant in steady-state conditions (Foyer ). Therefore, it is likely that the rate of electron transport in the thylakoids and Rubisco activity in the stroma are co-regulated, and therefore ETR would indirectly reflect variations in Rubisco activity. ETR is indeed co-regulated with Vc,max in Mediterranean species (Galmés ), and its use in place of Vc,max as a proxy for metabolic limitation results in very similar estimations of the latter under water stress (Gallé ). In the present study, initial activity of Rubisco was highly correlated with ETR (Fig. 4A), as it was with the CO2 net assimilation rate (AN) (Fig. 4B). These results confirm that despite potential activity losses during extraction, in vitro measurements of Rubisco activity were proportional to in vivo measurements and qualitatively representative of what occurs in vivo. Although the relationship between Rubisco initial activity and AN was linear when plotting all species together (Fig. 4B), it was logarithmic for 6 of the 11 species when considered individually (data not shown). In particular, some of the species with larger maximum AN and gs display a logarithmic response, while evergreen sclerophylls spanning a shorter range of values display a linear response. This could be related to the fact that the latter, due to lower maximum gs and gm, operate always at Cc below the general inflexion point for Rubisco inactivation (see next section).

Fig. 4.

Relationship between Rubisco initial activity and (A) ETR, and (B) AN. Values are means±standard error of four to five replicates per species and treatment. Values from rewatering treatments were excluded from this analysis. Regression coefficients and significance for each relationship are shown. Symbols and species are as in Fig. 2.

Causes for reduced Rubisco activity under water stress

Both Rubisco initial activity and its activation state presented a positive linear relationship when plotted against RWCPD (Fig. 5A, C); however, it turned into a logarithmic biphasic relationship when these two parameters were plotted against gs (Fig. 5B, D). Although all four correlations were highly significant (P<0.0001), by using gs the regression coefficients became higher, in agreement with previous surveys suggesting gs as a reference parameter (Medrano ; Flexas ). Moreover, when analysed for each individual species, the relationship between Rubisco parameters and RWCPD was significant for only three of the species (Table 2), while that with gs was generally significant (not shown)

Fig. 5.

Relationship between Rubisco initial activity and activation state with the leaf RWCPD and gs. Values are means±SE of four to five replicates per species and treatment. Values from rewatering treatments were excluded from this analysis. Regression coefficients and significance for each relationship are shown. Symbols and species are as in Fig. 2.

Table 2.

Bivariate correlations between Rubisco initial activity, Rubisco activation state, leaf RWCPD, and Cc for the 11 species included in the study

	Initial activity	Activation state
D. ibicensis
RWCPD	0.314	0.388
Cc	0.700	0.840*
L. minoricensis
RWCPD	0.725	0.629
Cc	0.862*	0.888**
B. maritima ssp. marcosii
RWCPD	0.821*	0.673
Cc	0.874**	0.862*
B. maritima ssp. maritima
RWCPD	0.724	0.653
Cc	0.927**	0.754*
L. magallufianum
RWCPD	0.785*	0.527
Cc	0.954**	0.890**
L. gibertii
RWCPD	0.725	0.192
Cc	0.758*	0.992**
H. balearicum
RWCPD	0.728	0.818*
Cc	0.080	0.008
P. lentiscus
RWCPD	0.687	0.967**
Cc	0.811*	0.149
C. albidus
RWCPD	0.651	0.325
Cc	0.800*	0.375
P. italica
RWCPD	0.334	0.763*
Cc	0.774*	0.661
L. maritime
RWCPD	0.958**	0.957**
Cc	0.822*	0.780*

Rewatering data were included in the analysis. Linear and logarithmic correlations were considered and values correspond to the best fit.

P<0.05; **P<0.01.

Rubisco initial activity and activation state did not correlate with the substomatal CO2 concentration (Ci) (Fig. 6A, C). In fact, maximum values for initial activity and activation state were observed at intermediate Ci values, perhaps reflecting the typical response of Ci to drought consisting in decreases at mild to moderate stress followed by increases at more severe stress (Medrano ). By contrast, when plotting Rubisco initial activity and activation state against Cc, significant correlations were observed (Fig. 6B, D). As occurred with RWCPD, the relationships between Rubisco initial activity and Cc and between activation state and Cc, presented an important scattering when plotting all species together. However, the regression coefficients generally improved when considering each species individually, being always higher than those with RWCPD (Table 2). By breaking the covariation between gs and RWC, Flexas observed that down-regulation of Rubisco activity was not dependent on RWC, but occurred at a certain threshold of gs (<0.05 mol m−2 s−1) and Cc (<100 μmol mol−1), although the relationship between Rubisco activity and Cc was less clear. In the present study, Rubisco initial activity correlated, at P < 0.05, with RWC for three species, and with Cc for nine species (Table 2). A similar trend was observed for the activation state (Table 2). It is worth noting that RW values were not consistent with the relationship in several species. Remarkably, the poorest correlations were found in shrubs (H. balearicum, P. lentiscus, C. albidus, and P. italica), with higher leaf mass per area (LMA), while in all herbaceous (D. ibicensis, L. minoricensis, B. maritima ssp. marcosii, and B. maritima ssp. maritima), and semi-shrub (L. magallufianum and L. gibertii) species both Rubisco initial activity and its activation state significantly correlated with Cc. In general, these data demonstrate that regulation of Rubisco activity is more related to CO2 availability in the vicinity of the enzyme than to leaf hydration status. Species-dependent patterns of Rubisco activity regulation under drought stress have previously been documented (Parry ; Bota ; Flexas ).

Fig. 6.

Relationship between Rubisco initial activity and activation state with substomatal (Ci) and chloroplastic (Cc) CO2 concentration. Values are means±SE of four to five replicates per species and treatment. Values from rewatering treatments were excluded from this analysis. Regression coefficients and significance for each relationship are shown. Symbols and species are as in Fig. 2.

The present results suggest that low Cc induced by the imposition of water stress could induce deactivation of Rubisco sites. Previously, in vivo Rubisco deactivation was observed at Ci<100 μmol mol−1 in several species, e.g. Chenopodium albus, Raphanus sativus, Triticum aestivum, and Phaseolus vulgaris (Perchorowicz and Jensen, 1983; von Caemmerer and Edmondson, 1986; Sage ; Sage ). The Kact[CO2], i.e. the CO2 concentration that supports half of the maximal activation has been reported to be 120 μmol mol−1 for in vitro systems with activase (Portis ). Sage suggested that in vivo Kact[CO2] is probably <60 μmol mol−1. In agreement with this suggestion, in the present study Rubisco activation started to decline at Cc below ∼120 μmol mol−1 (Fig. 6D). At this low Cc, the equilibrium between carbamylated and decarbamylated Rubisco will shift towards the latter form. The major aim of the present study was to find global relationships between Rubisco activity and physiology under drought using several species pooled. However, it is clear that species belonging to very different growth forms and leaf habitats, even if they all often experience drought in their natural environments, undergo different average operational and maximum Cc. For instance, evergreen sclerophylls show the lowest stomatal and mesophyll conductance even under irrigation (Galmés et al., 2007b), and therefore the lowest maximum Cc (Fig. 7, light-grey and empty symbols). Perennial and annual herbs, as well as semi-deciduous species, present much larger gs and gm when irrigated, and therefore experience larger Cc (Fig. 7, black and dark-grey symbols). These differential features may induce differences in Rubisco between species. Indeed, Galmés et al. (2005) showed that Rubisco specificity factor positively scaled with sclerophylly, i.e. it was larger in species with lower gs and gm. It is shown here (Fig. 6D), as in Perchorowitz and Jensen (1983) and Flexas that, globally, inactivation of Rubisco is induced by a Cc of <100 μmol mol−1. However, due to their intrinsic leaf morphology and physiology, evergreen species always operate at Cc <100 μmol mol−1, even with stomata fully open (Fig. 7), while herbs and semi-deciduous plants present maximum Cc values of between 130 and 200 μmol mol−1, operating at lower Cc only when subjected to drought. This seems to have induced evolutionary differences at the level of Rubisco activation sensitivity to Cc. Hence, even when in all species the inflexion point for Rubisco inactivation is at a Cc of ∼100±40 μmol mol−1 (Fig. 6), there is a linear correlation between the maximum operational Cc and the actual value of Cc at the inflexion point for Rubisco inactivation (Fig. 7).

Fig. 7.

Relationship between the chloroplastic CO2 concentration at its maximum (Cc,max) and at the inflexion point (Cc,ip) in the relationship between Rubisco activation state and Cc. For all species, Cc,max was Cc at WW conditions. For the calculation of Cc,ip, RW data were not considered. For all species, except L. minoricensis, a logarithmic function was applied in the relationship Rubisco activation state versus Cc, and the lowest R2 was 0.883. L. minoricensis is not included in the analysis because its Cc,ip could not be obtained (Cc,ip>100 μmol mol−1) due to the linear relationship between Rubisco activation state and Cc. Regression coefficient and significance of the relationship are shown. Symbols and species are as in Fig. 2.

These results suggest that those species adapted to low Cc,max are able to maintain active Rubisco at lower Cc. One possible explanation could be that Rubisco from these species has an intrinsically lower Kact[CO2]. Another possibility is that Rubisco activase is more active or present in higher concentrations in these species. Higher concentration and/or activation of Rubisco activase would improve removal of RuBP and other tightly bound sugar phosphates from the sites of reaction (Portis ). Finally, formation of side-products, such as xylulose-1,5-bisphosphate and pentodiulose-1,5-bisphosphate, from the enediol intermediate during Rubisco catalysis can progressively inhibit Rubisco turnover. Pearce (2006) demonstrated that the rate of production of these side-products increased when the [CO2]/[O2] ratio decreased, and such decreases occur under water stress (Galmés ). In consequence, differences in Cc,ip may be caused by a lower rate of formation of side-products at a given Cc for species with lower Cc,ip. In addition, Pearce (2006) also suggested that species with a higher specificity factor produce relatively fewer side-products, and Galmés observed, by studying these same species included in the present work, that Rubisco specificity factor tended to be higher in species with sclerophyll leaves (and therefore lower Cc,ip). Differences among species and groups in Rubisco responses to drought will deserve better attention in future studies.

Concluding remarks

The present study shows that decreases in the activity of Rubisco in response to drought occur in well-adapted Mediterranean species belonging to different growth forms and leaf habits. Decreases in both Rubisco initial activity and its activation correlated better with changes in stomatal conductance than with changes in relative water content. Since a direct effect of stomatal conductance on Rubisco is unlikely, we suggest that the effect is mediated by decreases in CO2 availability induced by lower stomatal and mesophyll conductance, rather than by increases in leaf dehydration. In fact, both Rubisco initial activity, and especially, Rubisco activation, showed a threshold relationship with in vivo estimated Cc that matches theoretical values obtained from in vitro studies. While there are some differences between species in the precise Cc threshold value so that species with low gs and thick leaves are able to maintain carbamylated Rubisco at lower values of Cc, evergreen sclerophyll species operate most of their life below that threshold, due to low maximum gs and gm even when water is available. In herbs and semi-deciduous species, the threshold is often reached during the frequent droughts occurring in the Mediterranean.

In conclusion, the results of the present study suggest that drought-driven decreases in Cc may induce deactivation of Rubisco and that photosynthesis in Mediterranean species may be constrained by lower Rubisco activation as a consequence of the need to keep stomata substantially closed in a water-scarce environment.