Irradiance and phenotype: comparative eco-development of sun and shade leaves in relation to photosynthetic CO₂ diffusion

Abstract

The subject of this paper, sun leaves are thicker and show higher photosynthetic rates than the shade leaves, is approached in two ways. The first seeks to answer the question: why are sun leaves thicker than shade leaves? To do this, CO₂ diffusion within a leaf is examined first. Because affinity of Rubisco for CO₂ is low, the carboxylation of ribulose 1,5-bisphosphate is competitively inhibited by O₂, and the oxygenation of ribulose 1,5-bisphosphate leads to energy-consuming photorespiration, it is essential for C₃ plants to maintain the CO₂ concentration in the chloroplast as high as possible. Since the internal conductance for CO₂ diffusion from the intercellular space to the chloroplast stroma is finite and relatively small, C₃ leaves should have sufficient mesophyll surfaces occupied by chloroplasts to secure the area for CO₂ dissolution and transport. This explains why sun leaves are thicker. The second approach is mechanistic or ‘how-oriented’. Mechanisms are discussed as to how sun leaves become thicker than shade leaves, in particular, the long-distance signal transduction from mature leaves to leaf primordia inducing the periclinal division of the palisade tissue cells. To increase the mesophyll surface area, the leaf can either be thicker or have smaller cells. Issues of cell size are discussed to understand plasticity in leaf thickness.

Aquaporin, cell wall, chloroplasts, conductance, diffusion, intercellular spaces, mechanical strength, photosynthesis, resistance to CO₂ diffusion, stomata

Why are sun leaves thicker than shade leaves?

The rate of photosynthesis of C₃ leaves strongly depends on the CO₂ concentration in the chloroplast, because affinity for CO₂ of the primary CO₂-fixing enzyme, ribulose 1,5-bisphosphate carboxylase/oxygenase (Rubisco), is low (von Caemmerer and Quick, 2000). Moreover, the carboxylation of ribulose 1,5-bisphosphate (RuBP) is competitively inhibited by O₂, and the oxygenation leads to the energy-consuming photorespiration processes. For C₃ plants to perform efficient CO₂ fixation in terms of economical use of energy and resources it is essential, therefore, to increase the conductance for CO₂ diffusion from the ambient air to the chloroplasts. Since the thickness of C₃ leaves is one of the important determinants in conductance for CO₂ diffusion and is a key factor to answering the question why sun leaves are thicker than shade leaves, the path of CO₂ diffusion is first traced from the ambient air to the chloroplast stroma.

Rubisco

Rubisco is a hexadecameric enzyme having eight large subunits encoded in the chloroplast DNA and eight small subunits encoded in the nuclear DNA in higher plants and green algae. Each large subunit has one active site that catalyses the following reaction:

\[\mathrm{RuBP}{+}\mathrm{CO}_{2}{+}\mathrm{H}_{2}0{\rightarrow}2{\,}\mathrm{phosphoglycerate}{\,}(\mathrm{PGA})\]

The maximum rate (k_cat) at CO₂ saturation at 25 °C is very low and typically 3 mol CO₂ mol active site⁻¹ s⁻¹ (24 mol CO₂ mol enzyme⁻¹ s⁻¹), which is lower than those of other Calvin–Benson cycle enzymes by two to three orders of magnitude (Roy and Andrews, 2000). The affinity for CO₂ is low and the Michaelis constant for CO₂, K_c, in the absence of O₂ is about 12 μM. This concentration is near the CO₂ concentration in the water equilibrated with the atmosphere having a CO₂ partial pressure at 37 Pa. Moreover, Rubisco shows high oxygenation activity. Molecular oxygen, O₂, competes with CO₂ for the same substrate, RuBP, in the active site:

\[\mathrm{RuBP}{+}\mathrm{O}_{2}{\rightarrow}\mathrm{PGA}{+}\mathrm{phosphoglycolate}\]

In the presence of 21% oxygen, the apparent K_c increases to about 20 μM due to substantial oxygenation. Because phosphoglycolate is an inhibitor of the triose phosphate isomerase of the Calvin–Benson cycle (Leegood, 1990), plants should detoxify this compound and salvage as much carbon as possible. The photorespiration cycle plays these two roles. By this cycle 1.5 C is salvaged and 0.5 C is lost as 0.5 CO₂ per oxygenation. The photorespiration pathway cycles back 0.5 PGA (1.5 C) to the Calvin–Benson cycle at the expense of 1 ATP and 0.5 NADPH (=1 ferredoxin). Thus, 1.5 PGA (4.5 C) is produced per oxygenation. For the regeneration of RuBP (5 C), 0.5 C is lacking. If 1/6 triose-phosphate is used for the 0.5 C, then 5 ATP and 3 NADPH are used altogether because the Calvin–Benson cycle requires 3 ATP and 2 NADPH for 1 carboxylation, RuBP regeneration, and 1/3 triose phosphate production. Thus, the energy requirement and carbon loss by photorespiration are huge (for stoichiometry, see Heldt, 1999; von Caemmerer, 2000). Thus, it is of supreme importance for plants to maintain the CO₂ concentration at the carboxylation site as high as possible for efficient carboxylation, and suppression of photorespiration as has been repeatedly pointed out (for an early review, see Raven, 1970).

Diffusion of CO₂ from the ambient air to the intercellular spaces

In photosynthesizing C₃ leaves, which have no biochemical CO₂ concentrating mechanisms, the CO₂-concentration at the carboxylation site, C_c, is lower than that in the ambient air, C_a, and CO₂ diffuses to the chloroplast stroma along the gradient of CO₂ concentration (Fig. 1).

Fig. 1.

Diffusion of CO₂ from the ambient air to the chloroplast stroma. C_a, CO₂ concentration in the air; C_s, substomatal CO₂ concentration; C_i, intercellular CO₂ concentration; C_c, CO₂ concentration in the chloroplast stroma. CO₂ in the intercellular spaces is dissolved in the water at the cell wall surface and diffuses to the chloroplast stroma through the cell wall, cell membrane, cytosol, and the chloroplast envelope. Because of the large resistance to CO₂ diffusion in the liquid phase, CO₂ flux via pathway B (dotted line in the right panel) relative to that via pathway A (continuous line) is negligible.

The CO₂ concentration in the substomatal cavity, C_s, is lower than C_a. C_a can be estimated by the gas exchange technique (Farquhar and Sharkey, 1982):

\[C_{\mathrm{s}}{=}C_{\mathrm{a}}{-}1.6P/g_{\mathrm{leaf,w}}\]

(1)

where P is the rate of net photosynthetic CO₂ fixation per unit leaf area, g_leaf,w is the leaf conductance for water vapour, and 1.6 is a physical conversion factor from the conductance for water vapour to that for CO₂. g_leaf,w is expressed as:

\[g_{\mathrm{leaf,w}}{=}g_{\mathrm{s,w}}{+}g_{\mathrm{c,w}}\]

(2)

where g_s,w is the stomatal conductance for water vapour and g_c,w is cuticular conductance for water vapour. Because CO₂ diffuses almost exclusively through stomata and not across the epidermis, the use of g_s,w instead of g_leaf,w is more appropriate. However, due to technical difficulty, g_leaf,w is routinely used for g_s,w. When g_c,w is much smaller than g_s,w, g_leaf,w can be used for g_s,w. In vigorously photosynthesizing C₃ leaves with widely open stomata, C_s/C_a would range from 0.6 to 0.9.

The value of g_c,w is relatively constant, irrespective of g_s,w values, and thereby the contribution of g_c would be significant when g_s,w is small. Thus, when some stress factor induces the closure of stomata, calculated C_s using g_leaf,w tends to be overestimated (for a review, see Evans et al., 2004). When stomata tend to close, photosynthesis can be non-uniform over the leaf. This also often causes overestimation of C_s (for reviews, see Terashima, 1992; Evans et al., 2004).

When the C₃ leaf is vigorously photosynthesizing, the bulk CO₂ concentration in the intercellular spaces, C_i, is lower than C_s due to the resistance to CO₂ diffusion in the intercellular spaces, r_ias (Parkhurst, 1994). However, this resistance is usually much smaller than the stomatal resistance, r_s. Except for very thick hypostomatous leaves (Parkhurst and Mott, 1990) or succulent leaves with small intercellular spaces (Maxwell et al., 1997), C_i/C_s is >0.9 for hypostomatous leaves (Terashima et al., 2001). The resistance to CO₂ diffusion in the intercellular spaces in the amphistomatous leaves is one-third to one-quarter of that in the hypostomatous leaves having the same thickness (Parkhurst et al., 1988; Terashima et al, 2001). Thus, it is unlikely that r_ias is a major limiting factor of leaf photosynthesis, particularly in the amphistomatous leaves. From here, therefore, C_s will not be distinguished from C_i and only C_i will be used.

CO₂ concentration at the carboxylation site in the chloroplast stroma, C_c, in C₃ plants is even lower than C_i (Evans and von Caemmerer, 1996; Evans and Loreto, 2000). Two different methods have mainly been used to estimate C_c. One method is based on the comparison of the electron transport rate estimated by the fluorescence method and the gas exchange rate measured simultaneously. This method is simpler but relies on several assumptions. The other one, the concurrent measurement of gas exchange and carbon isotope discrimination, is more complex but gives a more accurate estimation. C_c/C_i, thus obtained for vigorously photosynthesizing C₃ leaves, ranges from 0.5 to 0.75.

From the intercellular spaces to the chloroplast stroma

The data of conductance for CO₂ diffusion from the substomatal cavity to the chloroplast stroma, g_i, are plotted against the cumulated chloroplast surface area that faces the intercellular spaces on a leaf area basis, S_c (Fig. 2). The resistance to CO₂ diffusion in the liquid phase for a given distance is 10⁴ of that in the gas phase for the same distance. When pH is greater than 6, the ratio of the cumulated concentration of the inorganic carbon species relative to that of CO₂ (ΣC/CO₂) is >1 (Nobel, 1999). Then, the diffusion of CO₂ will be faster with the increase in ΣC/CO₂ if the interconversion among these inorganic carbon species is fast enough. Because cytoplasmic carbonic anhydrase activity would be sufficient (for a review, see Coleman 2000), ΣC/CO₂ in the cytosol at pH 7 may be >7. Still, the flux of inorganic carbon species via the pathway like B (dotted line) in the right panel of Fig. 1 should be negligible compared with that of pathway A (continuous line) because of the large liquid phase resistance. Thus, S_c is important as the active area for CO₂ diffusion to the chloroplast stroma (Laisk et al., 1970; Raven, 1970, 1977; Nobel, 1977; Evans and Loreto, 2000; Evans et al., 2004).

Fig. 2.

Internal conductance (g_i) plotted against chloroplast surface area, S_c. Open circles, annual herbaceous plants; open triangles, deciduous broad-leaved trees; solid triangles, evergreen broad-leaved trees; an open square, Kalanchoë daigremontiana. Faint lines denote wall conductance per unit leaf area. g_w is drawn assuming that p/τ of the cell wall is 0.1 for wall thicknesses of 0.1, 0.2, 0.3, 0.4, and 0.5 μm, respectively. See Table 1 for details of the samples.

The data in Fig. 2 that include plants of various functional types (Table 1) are, however, scattered. Among them, g_i values for annual herbs such as crop species are greatest when compared at a given S_c and range from 0.2 to 0.5 mol CO₂ m⁻² s⁻¹. On the other hand, those of evergreen broad-leaved trees are much lower and range from 0.03 to 0.2 mol CO₂ m⁻² s⁻¹.

Among the evergreen broad-leaved trees, Citrus spp. showed greater g_i than Macadamia integrifolia and the evergreen broad-leaved trees native to Japanese laurel forests (for scientific names and their authorship, see Table 1 and legend to Fig. 3). In mesic deciduous broad-leaved trees, g_i values are intermediate between those of annual herbs and evergreen trees. Among them, g_i for the tree species that develop leaves successively (Alnus japonica, Populus maximowiczii, and Prunus persica) tends to be greater than those for Acer spp. that flush their leaves in the spring. When the data points are grouped into annuals, including wheat and rice, deciduous broad-leaved trees, and evergreen broad-leaved trees, a tendency that g_i increases with the increase in S_c may be apparent.

Fig. 3.

Effects of environmental manipulations, genetic manipulations, and senescence on the S_c versus g_i relationship. Open circles, tobacco plants expressing antisense Rubisco small subunits (Evans et al., 1994); solid triangles, rice plants over-expressing barley aquaporin (Hanba et al., 2004); open triangles, senescing wheat leaves (Evans and Vellen, 1996); solid circles, diploid and artificially induced autotetraploid of Phlox drummondii Hook. (P Vyas et al., unpublished observation); open squares, alpine and low land Polygonum cuspidatum Sieb. et Zucc. (Kogami et al., 2001).

Table 1.

Mesophyll surface area, chloroplast surface area and internal conductance of species of various functional types

Species	Smes (m2 m−2)	Sc (m2 m−2)	gi (mol Co2 m−2 s−1)	Source
Annuals
Nicotiana tabacum L. (sun)	23.7	21.8	0.470	Evans et al., 1994
N. tabacum (shade)	17.0	14.5	0.370	Evans et al., 1994
Triticum aestivum L.	23.5	18.8	0.430	Evans and Vallen, 1996
Oryza sativa L.	19.0	14.0	0.200	Hanba et al., 2004
Phaseolus vulgaris L.	14.3	13.0	0.290	Hanba et al., 2003
P. vulgaris	16.0	14.5	0.250	Hanba et al., 2003
Phlox drummondii Hook.	14.6	8.5	0.160	P Vyas et al., unpublished
Deciduous broadleaved trees
Acer mono Maxim. (17% sun)	14.0	12.0	0.125	Hanba et al., 2001
Ac. mono (full sun)	12.0	8.5	0.065	Hanba et al., 2001
Acer palmatum Thunb.(17% sun)	8.0	5.5	0.030	Hanba et al., 2001
Ac. palmatum (full sun)	10.0	7.5	0.052	Hanba et al., 2001
Acer rufinerve Sieb. et Zucc. (17% sun)	8.5	5	0.047	Hanba et al., 2001
Ac. rufinerve (full sun)	18.0	13.5	0.047	Hanba et al., 2001
Alnus japonica (Thunb.) Steud.	20.0	18.0	0.110	Hanba et al., 2001
Populus maximowiczii A. Henry	78.0	15.5	0.100	Hanba et al., 2001
Prunus persica L. Batsch(sun)	30.1	21.0	0.370	Syvetsen et al., 1995
Pr. persica (shade)	26.6	14.0	0.291	Syvetsen et al., 1995
Evergreen trees
Quercus glauca Thunb. ex Murray	17.0	10.8	0.066	Hanba et al., 1999
Q. glauca	11.2	7.8	0.076	Hanba et al., 1999
Q. phillyraeoides A. Gray	31.4	15.4	0.143	Hanba et al., 1999
Cinnamomum camphora (L.) J. Presl	19.6	14.5	0.061	Hanba et al., 1999
Castanopsis sieboldii (Makino) Hatus.	20.7	16.7	0.044	Hanba et al., 1999
Cas. sieboldii	22.1	17.8	0.082	Hanba et al., 1999
Cas. sieboldii	25.0	25.0	0.100	Miyazawa and Terashima, 2001
Ligustrum lucidum Aiton	20.3	10.5	0.067	Hanba et al., 1999
L. lucidum	28.0	14.6	0.113	Hanba et al., 1999
Camellia japonica L.	41.6	20.5	0.119	Hanba et al., 1999
Cam. japonica	32.9	16.2	0.069	Hanba et al., 1999
Citrus paradisi Macfad	26.2	11.0	0.208	Syvetsen et al., 1995
Cit. paradisi	24.7	11.0	0.157	Syvetsen et al., 1995
Citrus limon (L.) Burm.f.	26.4	10.0	0.149	Syvetsen et al., 1995
Macadamia integrifolia Maiden et Betche	24.4	15.0	0.114	Syvetsen et al., 1995
CAM
Kalanchoë daigremontiana Hamet et Perr.	25.4	9.8	0.050	Maxwell et al., 1997

The main difference among these groups is the thickness of the mesophyll cell wall (δ_w). The δ_w in typical annuals, deciduous broad-leaved species, and evergreen broad-leaved species range from 0.1 to 0.2, 0.2 to 0.3, and 0.3 to 0.5 μm, respectively. Within the deciduous tree species, the flush-type species tend to show greater δ_w than in the successive-type species (Hanba et al., 2001, 2002). Acer rufinerve (100% light) showed a large S_c but a low g_i, which is probably explained by its very large δ_w for a deciduous tree species, 0.25 μm for the palisade tissue cells and 0.46 μm for the spongy tissue cells.

A clonal herbaceous perennial, Polygonum cuspidatum (synonymous to Reynoutria japonica Houttuyn), is a well-known pioneer plant as the first colonizer of volcanic deserts (Adachi et al., 1996). The plants grown at 2500 m a.s.l. in a volcanic desert on Mt Fuji-san, Japan, had a thick δ_w of 0.35–0.42 μm, while the same species at 10 m a.s.l. had a δ_w of 0.22–0.29 μm. In agreement with the difference in wall thickness, g_i for the plants from 2500 m was 0.076 mol m⁻² s⁻¹ while that for the plants from 10 m was 0.2 mol m⁻² s⁻¹ (Kogami et al., 2001; see Fig. 3). When autotetraploidy was artificially induced in Phlox drummondii, the leaves became thicker and S_c increased. However, at the same time, cell wall thickness increased considerably (from 0.12 to 0.24 μm), which would partly explain the absence of the increase in g_i with the increase in S_c (P Vyas, unpublished observation; see Fig. 3).

S_c represents the area for CO₂ dissolution and δ_w represents the path length for CO₂ diffusion in the liquid phase. Thus, it is understandable, that g_i is proportional to S_c and inversely related to δ_w.

The role of aquaporins

In addition to the two physical factors dealt with above, the role of aquaporins in CO₂ diffusion will be highlighted. Changes in g_i without marked changes in S_c and/or cell δ_w have been reported. For example, Evans and Vellen (1996) followed senescing wheat leaves and observed the drastic decrease in g_i without marked changes in S_c (Fig. 3). Perhaps δ_w did not increase greatly either. Decreases in g_i have been reported for plants under water stress (Flexas et al., 2002) and salt stress (Delfine et al., 1998, 1999). These studies indicate that CO₂ permeability of the plasma membrane or chloroplast envelope might have changed.

Aquaporins, the most abundant proteins in plant plasma membranes, transport water molecules according to the gradient of the water potential (Maurel, 1997; Kjellbom et al., 1999). However, there are reports indicating that some animal aquaporins transport CO₂ (Cooper and Boron, 1998; Nakhoul et al., 1998; Yang et al., 2000).

Terashima and Ono (2002) estimated g_i using the concurrent measurements of gas exchange and fluorescence in the leaves of Vicia faba L. and Phaseolus vulgaris L. before and after the application of HgCl₂, an inhibitor of most of the aquaporins, to the petiole. Because g_i and hydraulic conductivity of the mesophyll cells decreased at the same concentration range of HgCl₂, it is proposed that aquaporins are involved in diffusion of CO₂ across the plasma membrane. Application of chloromercuribenzene sulphonate that would not permeate membranes easily gave similar results (I Terashima, Y Tazoe, V Oja, A Laisk, unpublished results). Bernacchi et al. (2002) observed a clear peak at 35–37 °C in the temperature dependence of g_i measured by the fluorescence method in tobacco, which suggests the involvement of protein(s) in CO₂ diffusion across the plasma membrane.

Recently, it was shown that the aquaporin 1 of Nicotiana tabacum L., expressed in Xenopus oocytes, transfers CO₂ (Uehlein et al., 2003). In this study, tobacco aquaporin 1 (NtAQP1) was expressed in Xenopus oocytes and the CO₂ permeability was monitored as the decrease in pH with a pH micro-electrode. Hanba et al. (2004) attempted to over-express an aquaporin of Hordeum vulgare L. (HvPIP2; 1) in rice (Oryza sativa L.). By contrast to their original expectation, they obtained plants with varying aquaporin contents. The concurrent measurement of the gas exchange and carbon isotope discrimination of these leaves revealed that g_i clearly increases with the increase in aquaporin abundance.

Besides the abundance of aquaporins, conductance of CO₂ through aquaporins could also be regulated, for example, by pH and/or protein phosphorylation, because water permeability has been reported to be regulated by pH (Tournaire-Roux et al., 2003), and by protein phosphorylation (Maurel, 1997; Kjellbom et al., 1999). These possibilities are to be studied.

Components of the internal conductance

The internal conductance (or resistance) can be dissected further into several components (Fig. 4). The resistance to CO₂ diffusion from the cell wall surface to the carboxylation site per unit chloroplast surface area (R_i) may be expressed as:

\[R_{\mathrm{i}}{=}R_{\mathrm{w}}{+}\frac{1}{\frac{1}{R_{\mathrm{aq}}}{+}\frac{1}{R_{\mathrm{bm}}}}{+}R_{\mathrm{cytosol}}{+}R_{\mathrm{env}}{+}R_{\mathrm{stroma}}\]

(3)

where R_w, R_aq, R_bm, R_cytosol, R_env, and R_stroma are resistances to CO₂ diffusion across the cell wall, aquaporin, bulk plasma membrane, cytosol, chloroplast envelope, and chloroplast stroma to the carboxylation site expressed on a unit chloroplast surface area, respectively. The corresponding conductances are G_aq, G_bm, G_cytosol, G_env, and G_stroma. Note that lower-case letters, r and g, are used for the resistance and conductance, respectively, on a leaf area basis.

Fig. 4.

Components of the internal resistance.

For the moment, let us neglect R_cytosol, because the distance between the plasma membrane and chloroplast envelope is very small in many wild plants, and the abundance of inorganic carbon species, mostly

\(\mathrm{HCO}_{3}^{{-}}\)

and CO₂, relative to CO₂ (ΣC/CO₂) may be >7 at pH 7 if the carbonic anhydrase activity in the cytosolic is sufficient (Evans et al., 1994; Nobel, 1999; Coleman, 2000). R_env is also neglected. In the light, pH in the chloroplast stroma increases and there is carbonic anhydrase (Coleman, 2000). Thus the abundance of

\(\mathrm{HCO}_{3}^{{-}}\)

relative to CO₂ would be >50 (Nobel, 1999). Because

\(\mathrm{HCO}_{3}^{{-}}\)

also diffuses together with CO₂, R_stroma would be very small (Raven and Glidewell, 1981; Cowan, 1986).

Wall

The wall conductance for the unit chloroplast surface area can be expressed as:

\[G_{\mathrm{w}}{=}\frac{1}{R_{\mathrm{w}}}{=}\frac{p{\cdot}D_{\mathrm{C}}{\cdot}K_{\mathrm{CO}_{2}}}{\mathrm{{\tau}}{\cdot}\mathrm{{\delta}}_{\mathrm{w}}}\]

(4)

where p is porosity, D_C is the weighed diffusion coefficient of inorganic carbon species in water,

\(K_{\mathrm{CO}_{2}}\)

is the partitioning coefficient, the ratio of the concentrations of the inorganic carbon species in the liquid phase to that of CO₂ in gas phase (Nobel, 1999), τ is tortuosity, and δ_w is cell wall thickness.

Nobel (1999) assumed that the p/τ of the mesophyll cell wall is around 0.3. Because D_C at 20 °C is not very different from the diffusion coefficient of CO₂ (see below), 1.7×10⁻⁹ m² s⁻¹, he used pD_C/τ of 5×10⁻¹⁰ m² s⁻¹.

\(K_{\mathrm{CO}_{2}}\)

used was 1 because apoplast pH is somewhat acidic (Nobel, 1999; Stahlberg and Van Volkenburgh, 1999; but see Raven and Farquhar, 1989) and the inorganic carbon species is almost exclusively CO₂ and δ_w was 0.3×10⁻⁶ m, then G_w was calculated to be 1.7×10⁻³ m s⁻¹. Assuming S_c is 20 m² m⁻², the wall conductance on the leaf area basis, g_w, is 0.03 m s⁻¹, which corresponds to 1.2 mol m⁻² s⁻¹. Based on such calculations, Nobel claimed that wall conductance is very large and thus the cell wall is not limiting photosynthesis. However, the actual p/τ is not known for mesophyll cell walls of higher plants.

Although permeability of CO₂ across the cell wall has not been measured, permeability of water across the cell wall was measured. Kamiya et al. (1962) measured water permeability of the cell wall in Nitella flexilis (L.) Ag., a characean plant having giant internodal cells, using the pressurizing method. They found that the cell wall, which is10 μm thick, showed water permeability of 5×10⁻⁷ m s⁻¹ atm⁻¹. They also found that the permeability is inversely proportional to the cell wall thickness. Because 1 atm corresponds to the concentration difference of 40.4 mol m⁻³ at 25 °C, p/τ of the cell wall can be calculated from:

\[\frac{5{\times}10^{{-}7}}{V_{\mathrm{w}}}{=}\frac{p{\cdot}\mathrm{D}_{\mathrm{H}_{2}\mathrm{O}}}{\mathrm{{\tau}}{\cdot}\mathrm{{\delta}}_{\mathrm{w}}}{\cdot}40.4\]

(5)

where V_w is the molar volume of water (18×10⁻⁶ m³) and

\(D_{\mathrm{H}_{2}\mathrm{O}}\)

is the diffusion coefficient of water in water (2.3×10⁻⁹ m² s⁻¹ at 25 °C). When δ_w is 10×10⁻⁶ m, p/τ is calculated to be 3.

It is known that the permeability obtained by the pressurizing technique is greater than that by the diffusion method if the material has bulky holes or pipes (Gutknecht, 1967). The p/τ value of 3, therefore, indicates that there are holes or pipes in the cell wall.

Gutknecht (1967) measured the wall conductance (P_d) in the cell wall of a giant green alga Valonia ventricosa J. Agardh by the diffusion method using tritiated water and obtained 2.5×10⁻⁶ m s⁻¹. Because δ_w of these algae would be around 10 μm (not given by Gutknecht, 1967; but see Okuda et al., 1997),

\[2.5{\times}10^{{-}6}{=}(p{\cdot}D_{\mathrm{H}_{2}\mathrm{O}})/(\mathrm{{\tau}}{\cdot}\mathrm{{\delta}}_{\mathrm{w}})\]

(6)

p/τ thus obtained is 0.011. If this p/τ value, δ_w of 0.3 μm, and S_c of 20 m² m⁻² are used, then a very low g_w of 0.045 mol m⁻² s⁻¹ is obtained. This value is too low, even when the lowest g_i values in Fig. 2 are considered.

Accurate measurements of wall resistance to diffusion of water and CO₂ with higher plants are needed. It should be noted that one cannot use the pressure probe method in the pressurizing mode to determine wall resistance.

Assuming p/τ =0.1, g_w (see the grey lines in Fig. 2) was calculated. If p/τ is less than about 0.1, then wall conductance is an important factor determining internal conductance. Of course, p/τ values would vary considerably, because of the variation in cell wall constituents across species. However, it is proposed that the p/τ value is around 0.1 or lower, because the difference in g_i among leaves having a given S_c, but from different functional groups, is roughly explained by δ_w.

The resistance to CO₂ diffusion in the liquid phase is 10⁴ of that in the gas phase. If p/τ is 0.1, then the wall thickness of 0.3 μm will correspond to an air layer 30 mm thick. Thus, wall resistance would be an important limiting factor of photosynthesis.

Plasma membrane

When HgCl₂ was applied to the leaflets of Phaseolus vulgaris L., g_i decreased by up to 60% depending on the concentration of HgCl₂ (Terashima and Ono, 2002). Some aquaporins are known to be insensitive to HgCl₂ (Kjellbom et al., 1999) and, therefore, a part of the residual conductance could be attributed to such insensitive aquaporins. As already mentioned above, the tobacco aquaporin (NtAQP1) has been shown to transport CO₂ across the plasma membrane of Xenopus oocytes (Uehlein et al., 2003)

Hanba et al. (2004) have reported that rice leaves expressing various amounts of barley aquaporin (HvPIP2,1) showed differences in g_i, the range of which corresponds to 50% of the g_i in the control plants. g_i in tobacco plants expressing antisense aquaporin 1 was lower than that of the control plants by 40%, while the plants over-expressing aquaporin 1 showed g_i greater than that in the control plants by 30% (Flexas et al., 2004; International Photosynthesis Congress, 2004, Montreal; J Flexas, personal communication).

There are >10 different aquaporin species that are expressed in the plasma membrane of plants (Kjellbom et al., 1999). It is not known how many of these are responsible for the transport of CO₂. Thus, CO₂ transport activity in these aquaporins should be examined on a one by one basis. Then, the contribution of aquaporins to the overall conductance of the plasma membrane should be studied.

At this stage, only a rough estimation can be made. Let us assume that the CO₂ diffusion through aquaporins in a P. vulgaris leaf is completely suppressed by HgCl₂ and g_i decreased from 0.3 mol m⁻² s⁻¹ to 0.12 mol m⁻² s⁻¹. If p/τ, cell wall thickness, and S_c are 0.1, 0.1 μm, and 13 m² m⁻², respectively, then g_w will be 0.9 mol m⁻² s⁻¹. If it is further assumed that internal resistance is the sum of wall resistance and membrane resistance only, the membrane conductance values before and after the HgCl₂ treatment are 0.45 and 0.138 mol m⁻² s⁻¹. The corresponding conductance for the unit chloroplast surface area, G_aq and G_bm is calculated to be 0.024 and 0.011 mol m⁻² s⁻¹, or 5.86×10⁻⁴ and 2.69×10⁻⁴ m s⁻¹, respectively. These calculations may indicate that more than two-thirds of CO₂ molecules transported across the plasma membrane are via aquaporins.

It should also be pointed out that the effects of changes in abundance or conductivity of aquaporins on total internal conductance is largest in annual plants with thin cell walls but smallest in the evergreen broad-leaved tree species.

In summary, g_i is determined by S_c, wall thickness, and by abundance and/or conductivity of aquaporins. These aquaporins that transport CO₂ may be called ‘cooporins’ to highlight CO₂-porins that are co-operating with other photosynthetic components such as carbonic anhydrase.

Why are sun leaves thicker than shade leaves?

The light-saturated rate of leaf photosynthesis per unit area (P_max) in C₃ plants strongly depends on leaf nitrogen content and photosynthetic components such as Rubisco, cytochrome f, H⁺-ATPase, and reaction centres. P_max is also strongly correlated with structural parameters such as leaf thickness, leaf mass per area, mesophyll surface area (S_mes), and chloroplast surface area (S_c). Importance of S_c as an area for CO₂ dissolution and for the CO₂ diffusion pathway has been already discussed.

Let us consider the drawdown of CO₂ concentration from the intercellular spaces to the stroma (ΔC=C_i–C_c) for a unit chloroplast surface area (Fig. 5). The drawdown is proportional to the flux of CO₂ across the cell wall, plasma membrane, cytosol, and chloroplast envelope per unit chloroplast surface area and to the internal resistance, R_i. With the increase in the amount of Rubisco per unit chloroplast surface area, the CO₂ flux increases. However, the photosynthetic rate per Rubisco decreases because C_c decreases. From the viewpoint of efficiency of Rubisco use (or nitrogen use), thicker leaves with greater S_c would be advantageous because the amount of Rubisco per S_c becomes smaller and thereby C_c would increase. On the other hand, the construction and maintenance costs of thick leaves are expensive. Also, the drawdown of CO₂ concentration during the diffusion in the intercellular spaces, C_s–C_i, becomes greater with the increase in leaf thickness, which thus causes a decrease in the bulk C_i. In the latter two aspects, thick leaves are not at all advantageous.

Fig. 5.

Consequences of the difference in Rubisco content per unit chloroplast surface area. Given that the Rubisco content per leaf area is identical, thinner leaves having a smaller S_c should have more Rubisco per S_c. Then Rubisco in the thinner leaves would operate at a lower CO₂ concentration.

Effects of various aspects of mesophyll structure, in particular mesophyll thickness on photosynthesis, were evaluated using a one-dimensional model of CO₂ diffusion in the C₃ leaf (Terashima et al., 2001). In this model, mesophyll is composed of columnar cells, the lateral surfaces of which are fully occupied by chloroplasts. When mesophyll thickness was increased, keeping the Rubisco content per leaf area constant, S_c increases and Rubisco content per S_c decreases. The model leaves are either hypostomatous or amphistomatous. The main results can be summarized as follows:Obviously, neither (i) nor (ii) explains the strong relationship between P_max and S_c. Thus (iii) or constraints of this kind explain the strong relationships between P_max and leaf morphological parameters such as mesophyll thickness and S_c. In other words, leaf thickness is determined as a compromise between the increase in chloroplast surface area for CO₂ dissolution and the decrease in the construction and maintenance costs of the leaf. If such the economical optimum is strongly favoured, S_c/S_mes can be expected to be very high. Moreover, thickness of chloroplasts or Rubisco/S_c would not vary much. Both are roughly true in nature (see below). Thus, this would be the answer to the question: why are sun leaves thicker than shade leaves?

• (i) When mesophyll thickness was increased keeping the Rubisco content per leaf area constant, the rate of photosynthesis per leaf area increased at first due to the increase in C_c, attained the peak value, and then gradually decreased. The gradual decrease was due to the gradual decrease in C_i due to the increased r_ias. The thickness that gives the peak value was identical for the leaves with various Rubisco contents per leaf area.

• (ii) The thickness that gives the maximum photosynthetic rate for the amphistomatous leaves was greater than that for the hypostomatous leaves because r_ias in the former were one-quarter to one-third of that in the latter leaves.

• (iii) The mesophyll thickness that realizes a given photosynthetic rate per unit mesophyll thickness increased with the increase in the Rubisco content per leaf area.

Because C₄ plants have CO₂ concentration mechanisms, and the CO₂ fixation enzyme is located in the cytosol of mesophyll cells, the arguments above are exclusively for C₃ plants. The changes in leaf thickness or mesophyll cell surface area have not been intensively studied for C₄ plants. It was found recently that the thickness of the leaves of Amaranthus cruentus L., an NAD-ME-type C₄ dicotyledonous plant, did not respond to growth in the light (Fig. 6; Tazoe et al., 2005). For responses of C₄ photosynthesis to growth photosynthetic photon flux densities (PPFD), see the review by Sage (2006).

Fig. 6.

Sun and shade leaves of Amaranthus cruentus L. The plants were grown in a glasshouse. The midday PPFD for high-light plants was 1600 μmol m⁻² s⁻¹, while that for low-light plants was around 500 μmol m⁻² s⁻¹. High-light and low-light leaves showed a P_max of about 30 and 15 μmol CO₂ m⁻² s⁻¹, respectively (Y Tazoe, unpublished micrographs).

How do C₃ sun leaves have greater S_c than shade leaves?

In many deciduous broad-leaved tree species, sun leaves have a greater S_c than shade leaves. In sun leaves, the height of the palisade tissue is greater than in shade leaves. In some cases, thickening of the palisade tissue is accompanied by periclinal cell division as well as cell elongation of the palisade tissue cells.

Some ecotypes of Japanese beech (Fagus crenata Blume), including those on the Pacific side, differentiate sun leaves with thick palisade tissue comprising two cell layers (T Koike, personal communication; Fig. 7). In such ecotypes, not only the number of leaves (Kozlowski and Clausen, 1966) but also the number of cell layers in the palisade tissue of these leaves is determined in the winter buds, by early winter of the year prior to leaf unfolding (Eschrich et al., 1989). When the sun-exposed branches with young expanding leaves of F. crenata were shaded by shade cloths, the resultant leaves showed intermediate characteristics: they had the palisade tissue that comprised two cell layers but the height of the palisade tissue and P_max were lower than those in the fully exposed sun leaves (Uemura et al., 2000; A Uemura and A Ishida, personal communication). These strongly indicate that several different signals are used for the determination of characteristics of sun leaves, including multi-layered palisade tissue, greater cell height, larger S_c, higher contents of photosynthetic enzymes, and higher stomatal frequency and conductance. At least, the current-year PPFD and previous-year PPFD play different roles (Kimura et al., 1998; Uemura et al., 2000).

Fig. 7.

Sun and shade leaves of Fagus crenata Blume (A–C) and Fagus japonica Maxim. (D–F). (A) and (D) were from fully exposed shoots; (B) and (E) were from the shoots that had been exposed but were shaded from early spring until sampling in mid-summer; (C) and (F) were from the lower shaded shoots. Note that the palisade tissues of F. japonica had only one cell layer irrespective the light conditions (Uemura et al., 2000). Original micrographs of I Terashima.

Using Chenopodium album L., an annual herb, Yano and Terashima (2001) established a more sophisticated experimental system and examined the differentiation processes of sun and shade leaves. The plants were shaded in various ways and the effects of these shade treatments on the properties of the developing leaves were examined. When mature leaves were exposed to high light, the developing leaves, irrespective of their light environments, formed palisade tissue with two cell layers. On the other hand, when mature leaves were shaded, palisade tissue with one cell layer was formed. These results clearly showed that the light environment of mature leaves determined the number of cell layers in the palisade tissue of new leaves. There must be a signal transduction system that conveys the signal(s) from the mature leaves to the developing leaves (Yano and Terashima, 2001).

Yano and Terashima (2004) also conducted a detailed developmental study of sun and shade leaves of C. album. Whether the plants were grown under typical sun or shade conditions, the number of cells in the palisade tissue per leaf was almost identical. Moreover, in sun leaves, anticlinal and periclinal divisions occur almost synchronously. Thus, the signal from the mature leaves regulates the direction of cell division. In the future sun leaves, the signal probably induces periclinal division in addition to anticlinal division, while the signal from the shaded mature leaves only allows the cells to divide anticlinally (Yano and Terashima, 2004).

Yano and Terashima (2001, 2004) hypothesized that the signal is the abundance of photosynthates; when the photosynthates from mature leaves are abundant, leaves would develop into sun leaves. Currently, the effects of the sucrose content of the agar on leaf development in Arabidopsis thaliana (L.) Heynh. are being examined. There is a clear indication that the number of cell layers in the palisade tissue increases with increasing sucrose content (S Yano, H Tsukaya, personal communication).

Stomatal frequency and morphology of epidermal cells in young leaves are also regulated by the environment of the mature leaves (Lake et al., 2001; Thomas et al., 2003; Coupe et al., 2006), although the signal transduction mechanisms involved may be different from those that regulate the direction of cell division. For the palisade tissue cells to elongate, the local light environment of the developing leaves appears to be essential, although detailed studies have not been conducted.

On the other hand, chloroplast properties are mostly determined by the local light environment of the developing leaves (Terashima and Hikosaka, 1995; Yano and Terashima, 2001). It is known that there is a gradient in light environment within a leaf (Terashima and Saeki, 1983, 1985) and that chloroplasts within the leaf acclimate to their respective light environments (Terashima and Inoue, 1985a, b). When the leaves were inverted or irradiated from the bottom after their unfolding, the properties of chloroplasts changed and acclimated to their new light environments (Terashima and Takenaka, 1986; Terashima et al., 1986). Thus, acclimation of chloroplast properties to their local light environments is very plastic.

After maturation, leaves of some species respond to changes in the light environment. When grown with sufficient nutrients, most of the mesophyll cell surfaces facing the intercellular spaces are occupied by chloroplasts. However, it was shown that, for the shade leaves of Chenopodium album, unoccupied spaces are indispensable for re-acclimating to brighter light environments (Oguchi et al., 2003). When the shade leaves of C. album were exposed to a higher light for growth, an increase in S_c was observed. This increase in S_c was not accompanied by an increase in mesophyll surface area. Oguchi et al. (2005) examined shade leaves of three deciduous tree species, Betula ermanii Cam., Fagus crenata Blume, and Acer rufinerve Sieb. et Zucc. In F. crenata mesophyll surfaces were fully occupied by chloroplasts and did not increase P_max when the plants were exposed to a higher growth PPFD. On the other hand, shade leaves of B. ermanii that had unoccupied mesophyll surfaces increased P_max in response to the exposure. Shade leaves of A. rufinerve elongated palisade tissue cells and increased P_max in response to the increase in growth PPFD.

In Hedera helix L., the mature palisade tissue cells divided periclinally in response to the increase in growth PPFD (Bauer and Thoni, 1988). Such division of the palisade tissue cells has not been described for other species. However, for the evergreen leaves of extended longevity, such plastic adjustment may be important. Recently, mechanisms of the sun and shade leaf differentiation were reviewed elsewhere (Terashima et al., 2005).

Plasticity of leaf structure: importance of cell size

In this review, attention was first drawn to the importance of having sufficient S_c for efficient photosynthesis. Mechanisms responsible for the regulation of S_c are also discussed. These arguments are based on an assumption that cell diameter would not change. However, in nature, the size of mesophyll cells varies greatly and the leaf can increase S_c and decrease resistance to CO₂ diffusion in the intercellular spaces by decreasing cell size (Terashima et al., 2001, 2005; Miyazawa et al., 2003). The leaf with smaller cells is also mechanically tougher (Terashima et al., 2001). Actually, mesophyll cell size differs considerably across species (Terashima et al., 2001). However, leaves do not have very small cells. This could be because leaves exhibiting considerable rates of leaf area expansion, adequate heat capacitance, high efficiency of resource use, etc. have been favoured by natural selection (Terashima et al., 2001). The merits or demerits of having large and small mesophyll cells are summarized in Table 2. If the leaves have large S_c with large cells, as is found in annual herbs, the leaves should be thick. Such leaves would expand quicker, keep their shape by turgor, and have stomata on both leaf surfaces. The leaves with extended longevity tend to have small cells (Terashima et al., 2001). We mentioned previously that alpine plants tend to have smaller mesophyll cells (Terashima et al., 2005). However, their cells are not necessarily smaller than those of related lowland species (Körner, 1999). It is also worth mentioning that the mesophyll cell sizes of bonsai plants are comparable with those of plants grown normally (Körner et al., 1989). Although such leaves with small cells are mechanically tougher, area expansion would be slower. Moreover, the volume ratio of chloroplasts/nucleus in small cells might be smaller. Because N and P are important constituents of nucleic acids or proteins in nuclei as well, the cost, which may be called ‘information cost,’ would be much greater in leaves of small cells. If this informational cost is large, the nitrogen use efficiency of photosynthetic production would decrease. If this is very important, cell size would be greater in leaves under low availability of P and/or N,. There are other features such as water relationships with respect to cell size. Comprehensive studies approaching the diversity in leaf thickness based on causes and consequences of cell size should be made.

Table 2.

Possible effects of cell size on leaf characteristics

	Small cells	Large cells
Leaf thickness for the same Sc	Thin	Thick
Mechanical strength	Strong	Weak
Maintenance of leaf morphology	Cell walls (armour)	Turgor (balloon)
Longevity	Long	Short
Heat capacity	Small	Large
Intercellular resistance to CO2 diffusion	Small	Large
Stomatal occurrence	Hypostomatous	Amphistomatous
Area expansion	Slow	Fast
Informational cost (nuclear N and P/chloroplast N and P)	High	Low

†

Present address: Department of Botany, University of Delhi, Delhi 110 007, India

‡

Present address: National Institute for Basic Biology, Myodaiji-cho, Okazaki 444-8585, Japan

We thank the Society for Experimental Biology for the symposium on plant plasticity. We also thank the organizer of the symposium, Dr Owen Atkin, for his kind invitation to the symposium. Drs K Noguchi and A Antonio Suzuki, and Mr K Yoshida read an early draft and made insightful comments. IT would like to dedicate this review paper to late Dr Hiroshi Terashima, his father and a gynecologist, who passed away on 22 February 2005, to thank him for his continual encouragement.