Ribulose-1,5-Bisphosphate Carboxylase/Oxygenase Content, Assimilatory Charge, and Mesophyll Conductance in Leaves

DISCUSSION

By varying growth conditions and leaf age we obtained leaves that had a wide range of E _t (0.14–5 g protein m⁻², 2–75 μmol active sites m⁻², or 0.5–5 mm active sites in the chloroplast stroma) to find relationships betweenE _t and photosynthetic parameters of the leaf. In contrast to most previous researchers, we did not compare theE _t with the photosynthetic rate at atmospheric CO₂ and O₂concentrations but, instead, with μ under nonphotorespiratory conditions, which is the closest parameter to the carboxylation conductance measurable from leaf gas exchange. μ underestimates the carboxylation conductance by the proportion ofr _md (since the totalr _m = r _md +r _c). r _mdusually makes up 20% to 30% of the totalr _m (Laisk and Loreto, 1996), and the rest of the resistance is due to the limited speed of carboxylation.

Carboxylation conductance is proportional to theE _t or V _Mprovided that K _m(CO₂) is constant. We found that μ was linearly related toE _t at low E _tvalues but was saturated at higher E _tvalues. This saturation phenomenon caused a 2- to 3-fold decrease in the apparent k _cat calculated from μ andE _t (Fig. 2). This saturation is at variance with what was found in other studies, in which the photosynthetic rate at atmospheric levels of CO₂ and O₂ was linearly correlated withE _t (Makino et al., 1983; Jacob and Lawlor, 1992; Jiang and Rodermel, 1995). A linear relationship betweenE _t and V _M of Rubisco, calculated from measured μ, was also obtained by von Caemmerer et al. (1994). However, in these investigations theE _t did not exceed 35 μmol m⁻², whereas the calculatedk _cat was 3.5 s⁻¹. With this average k _cat, the data in Figure1 would also approximately satisfy a linear regression below 35 μmol active sites m⁻². Saturating relationships between E _t and photosynthesis have been found in other studies. For example, whenE _t exceeded 4 g m⁻² in rice (Makino et al., 1994, 1997), wheat (Lawlor et al., 1989), or leaves of a willow canopy (Vapaavuori and Vuorinen, 1989), the dependence between the rate of CO₂ assimilation andE _t declined from linearity. InChlorella pyrenoidosa cultures grown under high CO₂, E _t was higher but specific activity was lower than in cultures grown under low CO₂ (Yokota and Canvin, 1986). These results agree with those in Figure 1, which shows a decrease ofk _cat with increasingE _t. In our experiments the linearity between μ and E _t broke down above 2 g protein m⁻² or 25 μmol active sites m⁻².

The saturating dependence between the E _tand μ may be caused by physical and/or chemical properties. Physically, as r _cdecreases, ther _md eventually begins to determine μ. Calculations based on leaf anatomy have shown that anr _md of 0.01 s mm⁻¹ corresponds to a diffusion distance of 1 μm in chloroplasts (Laisk et al., 1970).r _md for CO₂ has been estimated to be sufficient to cause depletion of CO₂ at the carboxylation sites and to introduce curvilinearity into the relationship between photosynthesis andE _t (Evans, 1983). Other calculations have shown that a decrease in C _i cannot be large enough to cause curvilinearity in the relationship (Makino et al., 1985a; Lawlor et al., 1989). Calculations from fluorescence-based measurements of the electron-transport rate and the net CO₂-exchange rate (Laisk and Loreto, 1996) showed that r _md was usually between 0.02 and 0.04 s mm⁻¹ in herbaceous plants and did not exceed 20% to 30% of the total r _m. Measurements of sunflower in our laboratory have given similar results, estimating r _md from 0.02 to 0.06 s mm⁻¹ for full-grown leaves (Laisk and Sumberg, 1994; V. Oja and E. Eichelmann, unpublished results). In this work the minimum values of r _m were about 0.1 s mm⁻¹, which means that diffusionally limited μ is about 2 to 5 times greater than the actual maximum μ. This estimate shows that μ could still be carboxylase limited, even at the maximum E _t. This was clear in low-light-grown plants.

r _c is determined by the content and activity of Rubisco on one hand and by inhibiting stroma metabolites, e.g. PGA and free Pi, on the other (Badger and Lorimer, 1981; Foyer et al., 1987). Because our measurements of μ were carried out at limiting CO₂ and low O₂concentrations, it is likely that Pi and PGA accumulated at very low levels and RuBP was at a maximum (Badger et al., 1984; Dietz and Heber, 1986; Seemann and Sharkey, 1986). Therefore, we initially ignore the metabolite effects and discuss the μ as a function of the content and activity of Rubisco. The data in Figure 1B most clearly show that, in high-light-grown leaves at low E _t, μ increases proportionally with E _t. The calculated k _cat was 8.1 s⁻¹, slightly higher than the highest values obtained in Rubisco assays in vitro (5–6 s⁻¹;Evans and Seemann, 1984; Flachmann et al., 1997), whereas the modeledk _cat values in mature leaves are 3 to 3.6 s⁻¹ (von Caemmerer et al., 1994; Mate et al., 1996).

In leaves that had the highest k _cat values we assume that all of the Rubisco was completely active. When theE _t exceeded 20 μmol active sites m⁻² and μ approached 11 mm s⁻¹, the relationship was saturated, althoughE _t further increased to 65 μmol m⁻². The abrupt saturation of the relationship (Fig. 1, B and C) supports the case of chemical limitation, because the limitation by r _d would have resulted in a slowly saturating rectangular hyperbola. However, the abrupt saturation might be induced by changes in leaf mesostructure that took place in parallel with the changes in E _t, because the relationship shown in Figure 2 is close to a rectangular hyperbola. Proceeding from the model of chemical limitation, we see that there was a ceiling that limited the amount of active Rubisco to about 20 μmol reaction sites m⁻² in high-light-grown leaves. In low-light-grown leaves the relationship tended toward saturation at slightly higher E _t than in the high-light-grown leaves, but the maximum μ was considerably lower.

It appears that in low-light-grown leaves the same or only a slightly greater maximum number of reaction sites could be activated as in high-light leaves, but each site apparently had a slowerk _cat. Proceeding from this, a corresponding model would involve two processes that determine the Rubisco activity in vivo: one that determines the specific activity (turnover rate) of each active site and another that limits the maximum number of active sites. The first process is sensitively regulated by growth light, whereas the second is relatively insensitive to this parameter. For example, there may be a dynamic equilibrium between activation by the Rubisco activase system and nonenzymatic deactivation of Rubisco by a large molar excess of RuBP and other metabolites (Portis et al., 1986; Zhu and Jensen, 1991).

In leaf extracts containing a constant concentration and specific activity of Rubisco activase (about 0.45 μmol min⁻¹ mg⁻¹), the specific activity of Rubisco declined as the amount of Rubisco increased (Portis et al., 1986). This result is similar to what we found in intact leaves: an increase in the E _t resulted in a decrease of k _cat (Fig. 1). In experiments by Portis et al. (1986) the optimum concentration was 100 μg mL⁻¹ Rubisco and 300 μg mL⁻¹ Rubisco activase. In intact leaves this ratio seems to be shifted toward a higherE _t. In transgenic leaves with limited expression of Rubisco activase, photosynthesis was not influenced before activase content decreased more than 10 times (Mate et al., 1996). This result seemingly contradicts our above-described model, which emphasizes the limiting role of activase; however, several cofactors are needed to activate the activase itself.

ATP is an important cofactor of Rubisco activase in isolated chloroplasts (Robinson and Portis, 1988) but not in intact leaves (Brooks et al., 1988). In leaves light-induced electron transport through the PSI region of the electron-transport chain and the establishment of a transthylakoid pH gradient are obligatory for full activation of Rubisco (Campbell and Ogren, 1990, 1992). A good correlation between the dark-light pH difference in chloroplast stroma and μ has been observed (Eichelmann and Laisk, 1990). These results suggest that its binding sites at the PSI region of thylakoids may govern the activation process of Rubisco, not the content and activity of the Rubisco activase. Another important factor limiting the activation process may be the limited diffusibility of activase and Rubisco proteins, which could limit the protein-to-protein interaction required for activation. In our experimentsE _t increased to 350 mg mL⁻¹ in the chloroplast stroma, or to about 700 mg mL⁻¹ of total soluble protein. At such a concentration the state of the protein is close to crystalline, strongly limiting the movement of the activase and the carboxylation substrates. In crystalline Rubisco, the water content is 200 mg g⁻¹ protein (Paulsen and Lane, 1966). In protein-bound water, diffusion of small molecules is from 2 to 5 times slower than in free water (Lawlor et al., 1989), and the Michaelis-Menten kinetics of Rubisco catalysis may no longer apply because of limited diffusibility of RuBP and PGA.

Whatever the mechanistic reasons, our results confirm the notion that a part of Rubisco is not available for carboxylation and plays the role of storage protein. This idea was suggested as early as the 1960s, when Rubisco was discovered and found to make up as much as 60% of soluble protein in the chloroplast (Paulsen and Lane, 1966). Alteration of the source-to-sink ratio by removing the fruit in soybean resulted in the formation of an insoluble form of Rubisco in leaf extracts, with a specific activity 5 times less than that of the soluble form (Crafts-Brandner et al., 1991).

In transgenic rice leaves it was necessary to reduce Rubisco levels by about 55% to achieve 100% Rubisco activation (Makino et al., 1997). Using radioactive label Mae et al. (1983) demonstrated that Rubisco is a major N depot to support the growth of young tissues. This is in good accordance with our measurements of E _tduring the life span of sunflower leaves (Figs. 1D and 2). During leaf expansion E _t increased, and during senescence it decreased again (e.g. leaf nos. 3 and 9 in Fig. 2). In a 2-week-old plant, the two leaves had similarE _t (3.2 mm) and μ (570 s⁻¹) values. Later (1.5 weeks), leaf no. 9 was fully expanded and E _t increased to 4.2 mm, whereas μ (620 s⁻¹) increased only 8%. Leaf no. 3 had begun to senesce and theE _t (1.6 mm) had decreased by 50%, whereas μ (330 s⁻¹) had decreased by 40%. The process of remobilizing the N depot from old leaves to support the growth of new leaves explains why, during the adaptation of high-light-grown leaves to low light, the data points of the μ versusE _t graph are above the data points for normal low-light-grown leaves (Fig. 1B): the readapted leaf still has a larger N content than the low-light-grown leaf.

As defined above, the second component of carboxylation resistance is related to inhibiting stroma metabolites. μ correlated surprisingly well with assimilatory charge (Figs. 3 and 4), which is a gas-exchange measure of the RuBP pool (Laisk et al., 1984, 1987; Eichelmann and Laisk, 1990; Ruuska et al., 1998). It is not easy to explain why the μ versus AC relationships were so linear up to RuBP concentrations of 10 to 20 mm, considering thatK _m(RuBP) for Rubisco is about 40 μm in vitro (Servaites et al., 1991). Farquhar (1979) andvon Caemmerer and Farquhar (1985) attempted to resolve this problem by assuming that not all RuBP is available for carboxylation, only its nonchelated part, whereas the RuBP-Mg²⁺ complex cannot be a substrate for carboxylation. In the presence of ample Mg²⁺ only a minor fraction of RuBP remains free to be the carboxylation substrate. When the RuBP concentration is less than the concentration of active sites (2 mm in the activated state in our experiment) the reaction kinetics remain proportional to the RuBP concentration (Farquhar, 1979). This is a plausible explanation, except that it does not explain why the concentration of substrate RuBP, a small residual of the budget, is so constant and well reproducible from experiment to experiment, as is the measured maximum μ.

According to Michaelis-Menten kinetics, the reaction rate is expected to have little dependence on the RuBP concentration until it decreases to the concentration of active sites. The measured proportionality between the RuBP pool (AC) and reaction rate (μ) suggests that Rubisco must become progressively inhibited while the RuBP concentration decreases. In experiments with extracted Rubisco having a concentration up to 1.6 mm active sites, Paech (1986) found that an initial, fast PGA formation soon faded, indicating decreasing Rubisco activity in time. Eichelmann (1985) measured kinetic curves of RuBP utilization in extracts from sunflower leaves and in solutions of partially purified enzyme with E _t values from 5 to 30 mg mL⁻¹ protein (0.07–0.4 mm active sites). No RuBP saturation plateau was observed, and the kinetic curves of RuBP utilization were very similar to postillumination uptake curves in leaves (Laisk et al., 1987).

The continuous decrease in Rubisco activity during the RuBP utilization reaction was explained by competitive inhibition of the enzyme by the product, PGA. The K _I(PGA) calculated from the in vitro curves was 0.54 mm. However, when external PGA was added or when RuBP was repeatedly injected, leaving previously generated PGA in the reaction mixture, itsK _I(PGA) was much higher (3.7 mm). It was concluded that enzyme-bound, freshly generated PGA has a higher efficiency for inhibition than the free PGA. A similar result was reported later (Fong and Butcher, 1988), confirming that enzyme-bound PGA molecules formed by RuBP carboxylation are not equivalent to free PGA. The explanation is that PGA formed from C-3, C-4, and C-5 of the six-carbon intermediate (2-carboxy-3-keto-arabinitol-1,5-bisphosphate) is rapidly released, but PGA formed from C-1, C-2, and C-2′ is tightly bound to the Mg²⁺ of the enzyme and occupies the active site for a relatively long time.

Reversible binding of external PGA to Mg²⁺ of the active site is less likely. PGA and free Pi are both known to be competitive inhibitors for Rubisco: K _I(PGA) is 0.85 mm (Badger and Lorimer, 1981) andK _I(Pi) is 0.65 mm (Jordan et al., 1983). It is a general rule that the RuBP pool is complementary to PGA and Pi pools (Badger et al., 1984), because the total Pi pool is a constant and other carbon reduction cycle pools (except hexose phosphates under some conditions) are smaller than those directly involved in RuBP carboxylation. In the initial state of our postillumination experiments, under saturating light and limiting CO₂ concentrations, most of the Pi was probably bound in RuBP, leaving little of it in PGA or in a free state. Beginning from this state, decreases of AC were created by decreasing the light intensity for 10 s. This time was sufficient to convert a part of the RuBP into PGA, increasing the PGA-to-RuBP ratio. The relationship between μ and the RuBP pool (measured asAC) in the presence of a varying PGA-to-RuBP ratio was strictly linear, with the same proportionality constant for different leaves grown under similar conditions (Table I; Fig. 3) but also when Pi and Man were fed to influence the cytosolic and chloroplast Pi pools (Fig. 4). Linear kinetics have been modeled to describe the postillumination CO₂-fixation process, during which the RuBP pool decreases and the PGA pool increases, using the above inhibition coefficient by PGA (Laisk et al., 1987), showing that product inhibition can explain the close-to-linear kinetics of Rubisco.

The proportionality constant relating μ to AC, SCE, was 0.04 to 0.065 mm s⁻¹μmol⁻¹ m². SCE varies when external conditions lead to variations in Rubisco activation state (e.g. waiting until Rubisco activity stabilizes at each limiting PAD;Laisk et al., 1984), but its maximum value is usually relatively constant in similarly grown leaves. It is a rule of thumb that SCE is higher in leaves that have lower maximum AC pools (Laisk et al., 1984, 1987) and decreases with senescence (Eichelmann and Laisk, 1990). This is consistent with the above product-inhibition model, which predicts that the true uninhibited μ can be measured only under conditions in which PGA and Pi pools are minimal, i.e. under saturating light and limiting CO₂ concentrations and low O₂ concentrations.

Under low light and in the presence of saturating CO₂ or high O₂concentrations, the PGA pool is not minimized and it has an inhibitory effect on Rubisco. Evidently, this is one reason why μ decreases with increasing O₂ concentration (Laisk and Loreto, 1996), despite the fact that RuBP pool may still be large enough to saturate the enzyme. During postillumination CO₂ fixation, the initially maximum μ decreases continuously during the whole period of RuBP consumption, which lasts longer the greater the RuBP pool. However, the larger the chloroplast Pi pool, the larger the RuBP pool. Thus, SCE is steeper the lower the initial RuBP concentration, i.e. the lower the chloroplast Pi content. An important conclusion is that SCE is not a true characteristic of Rubisco but is influenced by the chloroplast Pi pool. To the extent the latter stays constant, relative changes in SCE may be interpreted to indicate changes in Rubisco-specific activity, but one should be cautious when comparing SCEs from different species or leaves. It is also important that only μ measured under conditions in which PGA and Pi are minimized can be safely interpreted as Rubisco activity and related to its content.

We also made some observations that are more complicated to explain by the competitive inhibition model. Incubation of a leaf for 10 min at only a 4°C higher temperature (27°C) caused a 40% increase inAC, whereas the μ versus AC relationship remained constant. Staying with the model, we conclude that a part of the Pi was converted from an inhibitory state (PGA, Pi) into the substrate state (RuBP). Thus, one must be careful in interpreting the maximum AC _M pool and the corresponding μ obtained under the light-saturating and CO₂-limiting conditions as a true maximum. Some Pi may still reside in an inhibitory form, e.g. as hexose phosphate.

Another way to change μ and AC while maintaining SCE constant was to feed compounds through the phloem (Fig. 4). Our treatments were expected to increase (feeding Pi) or decrease (feeding Man) Pi concentration in the cytosol and to influence the Pi concentration and Pi distribution in stroma sugar phosphates via the Pi translocator. Since these treatments were not expected to change the total Pi in the stroma but only its distribution between free and organic forms, the correlated changes between μ and ACwere expected. Either increasing or decreasing cytosolic Pi resulted in a decrease in the RuBP pool. It is possible that the formation of hexose phosphates traps Pi and thereby decreases the amount available for RuBP formation, as suggested previously (Eichelmann and Laisk, 1994).

Aside from the proportionality between μ and AC,A _M was also proportional to μ (Fig. 6). This proportionality was noted earlier when photosynthesis was influenced by salinity (Seemann and Sharkey, 1986), in ozonation of bean (Moldau et al., 1991; Moldau and Kull, 1993) and aspen (Kull et al., 1996), by varying growth light in sunflower (Eichelmann and Laisk, 1990), and in a Cytb ₆/f-deficient antisense mutant of tobacco (H. Eichelmann, unpublished data). The clearest demonstration so far of the proportionality between μ and the rate of CO₂ assimilation at 500 μbar CO₂ in spinach leaves grown at different nitrate levels was given by Evans and Terashima (1988). The novelty of our present work is that the slope of μ versusA _M was dependent on the growth light (Fig.6).

The relationship between μ and A _Mdeserves to be discussed because it was theoretically unexpected.A _M is usually determined by end product (Suc and starch) synthesis rate and Pi turnover, which limits RuBP regeneration (Laisk and Laarin, 1983), whereas μ is determined by Rubisco and carbon reduction cycle pools. Such good correlative relationships between different photosynthetic parameters show that genes controlling different subsystems of leaf photosynthetic metabolism are proportionally expressed and controlled during development and adaptation.

It is possible that the total pool of Pi in chloroplasts is an important parameter controlling the development of the photosynthetic apparatus, or at least proportionally related to its amount. Presently we know very little about factors that determine the total Pi pool in chloroplasts. The good proportionality between different photosynthetic parameters may be a result of spatial compartmentation of the photosynthetic machinery into small units that can perform almost completely independently, or of association of enzymes into supercomplexes that channel intermediates (Gontero et al., 1988;Süss et al., 1995; Echeverria et al., 1997).

We suggest that saturation of the relationship betweenE _t and μ is caused by three limiting components. Physical diffusion resistance in the liquid phase would limit the μ at about 2 to 5 times higher values than measured. An essential limitation of μ is caused by factors that allow only partial activation of Rubisco. We suggest that these are related to Rubisco activase, which itself is activated by thylakoid membrane-bound complexes such as PSI, ATPase, and the transthylakoid pH gradient. The slower diffusibility of Rubisco at high protein concentrations in stroma is a factor that may limit the activation process of Rubisco, breaking the balance between activation and deactivation processes. Finally, chloroplast metabolites, especially PGA and free Pi, control the reaction kinetics of RuBP carboxylation by competitively binding to active sites.