Faster Rubisco Is the Key to Superior Nitrogen-Use Efficiency in NADP-Malic Enzyme Relative to NAD-Malic Enzyme C₄ Grasses

Abstract

In 27 C₄ grasses grown under adequate or deficient nitrogen (N) supplies, N-use efficiency at the photosynthetic (assimilation rate per unit leaf N) and whole-plant (dry mass per total leaf N) level was greater in NADP-malic enzyme (ME) than NAD-ME species. This was due to lower N content in NADP-ME than NAD-ME leaves because neither assimilation rates nor plant dry mass differed significantly between the two C₄ subtypes. Relative to NAD-ME, NADP-ME leaves had greater in vivo (assimilation rate per Rubisco catalytic sites) and in vitro Rubisco turnover rates (k_cat; 3.8 versus 5.7 s⁻¹ at 25°C). The two parameters were linearly related. In 2 NAD-ME (Panicum miliaceum and Panicum coloratum) and 2 NADP-ME (Sorghum bicolor and Cenchrus ciliaris) grasses, 30% of leaf N was allocated to thylakoids and 5% to 9% to amino acids and nitrate. Soluble protein represented a smaller fraction of leaf N in NADP-ME (41%) than in NAD-ME (53%) leaves, of which Rubisco accounted for one-seventh. Soluble protein averaged 7 and 10 g (mmol chlorophyll)⁻¹ in NADP-ME and NAD-ME leaves, respectively. The majority (65%) of leaf N and chlorophyll was found in the mesophyll of NADP-ME and bundle sheath of NAD-ME leaves. The mesophyll-bundle sheath distribution of functional thylakoid complexes (photosystems I and II and cytochrome f) varied among species, with a tendency to be mostly located in the mesophyll. In conclusion, superior N-use efficiency of NADP-ME relative to NAD-ME grasses was achieved with less leaf N, soluble protein, and Rubisco having a faster k_cat.

C₄ photosynthesis involves the close collaboration of two photosynthetic cell types, the mesophyll (M) and bundle sheath (BS). A key characteristic of the C₄ syndrome is the operation of a CO₂ concentrating mechanism, which serves to raise the CO₂ concentration in the BS around Rubisco to levels high enough to suppress photorespiration and almost saturate photosynthesis in air (Hatch, 1987). This explains the commonly observed high photosynthetic rates of C₄ relative to C₃ leaves, when comparisons are made under high light and temperature. C₄ plants also have greater photosynthetic rates and accumulate more biomass than C₃ plants for less leaf nitrogen (N) and Rubisco (Bolton and Brown, 1978; Brown, 1978; Schmitt and Edwards, 1981; Ghannoum et al., 1997; Ghannoum and Conroy, 1998; Makino et al., 2003). The C₄ photosynthetic pathway is divided into three biochemical subtypes following the major C₄ acid decarboxylation enzyme (NAD-malic enzyme [ME], NADP-ME, and phosphoenolpyruvate carboxykinase; Hatch, 1987). C₄ grasses with different biochemical subtypes have characteristic leaf anatomy (Hattersley, 1992) and different geographic distribution according to rainfall, such as seen in Australia (Hattersley, 1992) and South Africa (Ellis et al., 1980). With increasing rainfall, NADP-ME grasses increase in abundance, whereas NAD-ME grasses become less abundant. The aforementioned observations triggered our interest in the comparative physiology of the C₄ subtypes, especially because little is understood about the physiological significance of the biochemical diversification of C₄ photosynthesis. Consequently, we undertook a number of studies comparing NAD-ME and NADP-ME grasses because they represent the two most contrasting and floristically abundant subtypes (Hattersley, 1992). Restricting our study to two subtypes also allowed us to include more species in our comparisons. Previously, we showed that water stress enhances water use efficiency to a greater extent in NAD-ME than NADP-ME grasses (Ghannoum et al., 2002). While carrying out these experiments, we observed that NADP-ME grasses tended to have a lower leaf N content and greater whole-plant (NUE) and photosynthetic (PNUE) N-use efficiencies than their NAD-ME counterparts.

Differences in PNUE are mainly brought about by differences in photosynthetic capacity or foliar N allocation either within the photosynthetic apparatus or to nonphotosynthetic pools (e.g. cell walls, nitrate; Field and Mooney, 1986; Poorter and Evans, 1998). In C₃ plants, photosynthetic N allocation is well established (Evans and Seemann, 1989). Recently, Makino et al. (2003) constructed a leaf N budget for maize (Zea mays), a NADP-ME monocot. However, foliar N allocation in C₄ leaves is complicated by the presence of three biochemical subtypes and two photosynthetic cell types. There is indirect evidence that the C₄ cycle has a greater N requirement in the NAD-ME relative to the NADP-ME subtype. For example, the C₄ cycle of the NAD-ME relative to the NADP-ME pathway involves more enzymatic steps and amino acids (Kanai and Edwards, 1999). Also, it is well known that the BS chloroplasts of NADP-ME grasses lack PSII activity (Hatch, 1987), which may translate into reduced thylakoid N cost of NADP-ME leaves. In addition, differences in PNUE may be due to variations in the efficiency of the photosynthetic apparatus, such as enzyme kinetics. The superior PNUE of C₄ relative to C₃ leaves is not only due to higher photosynthetic rates per leaf N and Rubisco content but also to a Rubisco with faster turnover rate (k_cat; von Caemmerer and Quick, 2000; Sage, 2002). Which of these factors best explain our observed differences in leaf N and PNUE between NAD-ME and NADP-ME grasses? To answer this question, we carried out a series of experiments addressing three specific questions. (1) How widespread are the differences in leaf N, PNUE, and NUE between NAD-ME and NADP-ME grasses? Do they persist under low N supply? (See experiments 1–3.) (2) Are these differences due to Rubisco amount and/or specific activity? (See experiment 3.) (3) Are these differences related to N partitioning in the M and BS tissues? (See experiment 4.)

RESULTS

Leaf A, N, PNUE, and NUE (Experiments 1–3)

In three separate experiments, different combinations of 13 NAD-ME and 14 NADP-ME C₄ grasses were grown under adequate (experiments 1 and 2) and/or deficient (experiment 3) N supplies (Table I). In each case, average PNUE and NUE were significantly greater in NADP-ME than in NAD-ME grasses (Fig. 1, C and D; Table II). This difference was associated with lower leaf N in NADP-ME grasses, whereas average CO₂ assimilation rates (A) were similar for the two subtypes (Fig. 1, A and B; Table II). NADP-ME species accumulated more biomass than NAD-ME counterparts in only one out of three experiments (Table II). Low N supply reduced leaf A and N, biomass accumulation, and tillering, whereas it increased PNUE, NUE, and biomass allocation to roots in all species (Fig. 1; Table II). N deficiency affected measured parameters equally in both subtypes (Tables II and III). Whole-plant N concentration per dry mass was not significantly different between the two subtypes. Lower shoot (leaf and stem) N in NADP-ME relative to NAD-ME grasses was counterbalanced by higher root N (Table II). Of all measured parameters, leaf N and PNUE had the most consistent and significant difference, and species values showed the greatest range separation between the two subtypes (Fig. 1; Table II).

Table I.

List of 13 NAD-ME and 15 NADP-ME C₄ grass species used in 4 experiments

Experiments 1 to 3 were used for growth, gas exchange, and biochemical analysis. Experiment 4 was used for BS and M N budget analysis.

Species	Experiment
	1	2	3	4
NAD-ME	n = 10	n = 9	n = 7	n = 2
Astrebla lappacea (Lindl.) Domin	*	*	*
Astrebla pectinata (Lindl.) F. Muell. ex Benth.	*	*	*
Astrebla squarrosa C.E. Hubbard	*	*
Cynodon dactylon	*	*
Enteropogon acicularis Nees	*
Eleusine coracana Gaertn		*
Eragrostis setifolia N.M. Wolf	*
Eragrostis superba N.M. Wolf	*	*	*
Leptochloa dubia P. Beauv.	*	*
Panicum coloratum	*	*	*	*
Panicum decompositum R.Br.	*	*	*
Panicum miliaceum			*	*
Panicum virgatum			*
NADP-ME	n = 14	n = 9	n = 7	n = 2
Bothriochloa biloba Kuntze	*	*	*
Bothriochloa bladhii Kuntze	*	*	*
Bothriochloa pertusa	*
Cenchrus ciliaris	*	*	*	*
Cymbopogon ambiguus Spreng.	*	*
Cymbopogon bombycinus Spreng.	*
Dichanthium aristatum (R.Br.) A. Camus	*
Dichanthium sericeum (R.Br.) A. Camus	*	*	*
Digitaria brownii (R&S) Hughes	*	*
Digitaria smutsi Haller	*
Paspalum dilatatum	*	*	*
Paspalum notatum	*	*	*
Pennisetum alopecuroides Rich.	*
Pennisetum clandestinum Hochst. ex Chiov.	*	*	*
Sorghum bicolor cv pop sorghum				*

Figure 1.

Box-whisker plots of assimilation rates (A), N content (B), PNUE (C), and NUE (D) in 13 NAD-ME and 14 NADP-ME C₄ grasses (experiments 1–3; Table I). The box and whisker represent the 25 to 75 percentile and minimum-maximum distributions of the data, respectively. Means (•) and significance levels for the subtype effect calculated as described in “Materials and Methods” are shown and are: *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ns, not significant (P ≥ 0.05). HN, High N soil supply; LN, low N soil supply.

Table II.

Summary of average steady-state gas exchange parameters, leaf, stem (including sheath), root, and whole-plant N, leaf mass per area, PNUE, NUE, and some growth parameters of NAD-ME and NADP-ME C₄ grasses from three independent experiments (experiments 1–3; Table I)

Maximal phosphoenolpyruvate carboxylase (V_{p max}) and Rubisco (V_{c max}) activities were estimated from A/C_i curves (experiment 2) using the C₄ photosynthesis model of von Caemmerer (2000). Within the same experiment, values followed by the same letter are not significantly different (P > 0.05). −, Not determined. HN, High N soil supply; LN, low N soil supply.

	Experiment 1		Experiment 2		Experiment 3
Parameter	HN		HN		HN		LN
	NAD-ME	NADP-ME	NAD-ME	NADP-ME	NAD-ME	NADP-ME	NAD-ME	NADP-ME
A (μmol m−2 s−1)	30aa	33aa	39ab	40ab	35b	35b	25a	24a
g (mol m−2 s−1)	0.25aa	0.28aa	0.26ab	0.28ab	0.22b	0.23b	0.16a	0.16a
Ci/Ca	0.37aa	0.38aa	0.31ab	0.33ab	0.26a	0.26a	0.29a	0.33b
Vp max (μmol m−2 s−1)	–	–	81a	82a	–	–	–	–
Vc max (μmol m−2 s−1)	–	–	47a	46a	–	–	–	–
Vp max/Vc max	–	–	1.72a	1.78a	–	–	–	–
Leaf N (mmol m−2)	119b	70a	115b	88a	108d	81c	47b	37a
Leaf N (mg g−1)	39b	32a	34b	31a	41c	37b	18a	18a
Stem N (mg g−1)	–	–	21b	19a	–	–	–	–
Root N (mg g−1)	–	–	16a	19b	–	–	–	–
Plant N (mg g−1)	–	–	24a	23a	–	–	–	–
Leaf mass per area (g m−2)	44ba	31aa	42b	37a	36c	31b	35c	28a
A/N (μmol CO2 [mol N]−1 s−1)	313a	525b	351a	478b	299a	388b	445c	611d
NUE (g dry mass [g leaf N]−1)	67a	111b	95a	116b	65a	88b	191c	218d
Tiller no. (plant−1)	–	–	–	–	10b	16c	4a	5a
Plant dry mass (g plant−1)	9.2aa	14.6ba	9.9ab	10.7ab	4.4b	4.5b	1.5a	1.3a
Leaf mass fraction (% plant dry mass)	39ba	33aa	35bb	31ab	41d	35c	32b	28a
Stem mass fraction (% plant dry mass)	50aa	50aa	47ab	54bb	35c	44d	26a	28a
Root mass fraction (% plant dry mass)	11aa	17ba	18bb	15ab	24b	21a	42c	44c

^aData taken from Ghannoum et al. (2001).

^bData taken from Ghannoum et al. (2002).

Table III.

Leaf N, Chl, Rubisco sites, and the fraction of leaf N invested in Rubisco of seven NAD-ME and seven NADP-ME grasses grown under high (HN) and low (LN) soil N supply (experiment 3; Table I)

Values are means ± se. Significance levels are as described in the legend of Figure 1.

Species	Leaf N		Chl		Rubisco		Rubisco N
	HN	LN	HN	LN	HN	LN	HN	LN
	mmol m−2		μmol m−2		μmol sites m−2		% leaf N
NAD-ME
Astrebla lappacea	103 ± 6	30 ± 1	389 ± 27	215 ± 0	11.1 ± 0.6	4.9 ± 0.3	8.4 ± 0.0	12.8 ± 1.4
Astrebla pectinata	126 ± 0	90 ± 6	453 ± 19	277 ± 0	6.8 ± 0.7	4.6 ± 0.3	4.2 ± 0.4	4.0 ± 0.0
Eragrostis superba	118 ± 1	48 ± 3	423 ± 11	263 ± 11	8.5 ± 0.2	4.7 ± 0.1	5.6 ± 0.2	7.7 ± 0.3
Panicum coloratum	119 ± 16	53 ± 2	473 ± 21	296 ± 27	11.0 ± 1.1	4.8 ± 0.3	7.3 ± 0.2	7.2 ± 0.3
Panicum decompositum	128 ± 5	60 ± 10	605 ± 60	383 ± 45	10.4 ± 1.2	6.9 ± 1.9	6.4 ± 0.5	8.8 ± 1.1
Panicum miliaceum	143 ± 7	47 ± 9	452 ± 25	224 ± 8	9.9 ± 1.9	3.6 ± 0.1	5.5 ± 1.3	6.1 ± 1.1
Panicum virgatum	132 ± 6	73 ± 2	646 ± 6	402 ± 10	8.8 ± 0.4	5.0 ± 0.3	5.3 ± 0.5	5.3 ± 0.2
NADP-ME
Bothriochloa biloba	66 ± 4	44 ± 0	503 ± 69	326 ± 23	4.0 ± 0.6	2.4 ± 0.4	4.7 ± 0.4	4.3 ± 0.7
Bothriochloa bladhi	81 ± 3	60 ± 3	488 ± 25	512 ± 81	4.8 ± 0.1	4.6 ± 0.8	4.6 ± 0.3	6.0 ± 0.9
Cenchrus ciliaris	98 ± 1.4	27 ± 3	488 ± 16	203 ± 7	6.4 ± 0.4	2.0 ± 0.2	5.1 ± 0.2	5.7 ± 0.1
Dichanthium sericeum	83 ± 7	38 ± 1	586 ± 80	281 ± 9	7.4 ± 0.3	2.5 ± 0.2	7.0 ± 0.4	5.2 ± 0.5
Paspalum dilatatum	100 ± 12	40 ± 3	442 ± 71	268 ± 22	7.2 ± 0.6	4.5 ± 0.3	5.8 ± 1.2	8.8 ± 0.1
Paspalum notatum	87 ± 1	37 ± 1	536 ± 62	223 ± 10	8.2 ± 0.1	3.9 ± 0.4	7.4 ± 0.2	8.3 ± 0.5
Pennisetum clandestinum	108 ± 1	37 ± 1	462 ± 10	241 ± 14	6.0 ± 0.8	2.6 ± 0.0	4.4 ± 0.5	5.7 ± 0.1
Average (NAD-ME)	124 ± 5	57 ± 7	492 ± 36	294 ± 28	9.5 ± 0.6	4.9 ± 0.4	6.1 ± 0.5	7.4 ± 1.1
Average (NADP-ME)	89 ± 5	41 ± 4	501 ± 18	293 ± 39	6.3 ± 0.6	3.2 ± 0.4	5.6 ± 0.5	6.3 ± 0.6
P (subtype)	***	***	ns	ns	***	***	**	**
P (N)	***	***	***	***	***	***	**	**
P (N × subtype)	ns	ns	ns	ns	ns	ns	ns	ns

Rubisco Amount and Activity (Experiment 3)

N deficiency reduced leaf A, chlorophyll (Chl), N, and Rubisco content in seven NAD-ME and seven NADP-ME C₄ grasses, and this effect was not significantly different between the two subtypes (Fig. 2; Table III). Leaf N and A as well as leaf N and Rubisco sites were well correlated (r² between 0.62–0.74) across species and treatments (Fig. 2, A and B). The slopes of the former (A versus N; Fig. 2A) but not the latter (Rubisco versus N; Fig. 2B) relationships were significantly different between the two subtypes (P < 0.05). In vitro Rubisco catalytic turnover rate (k_cat) was assayed at 25°C, for better comparison with the literature, and 10 mm NaHCO₃ by measuring Rubisco activity and content of catalytic sites on the same extract of leaves harvested under high light. In vitro k_cat, thus measured, was well correlated with calculated in vivo k_cat at 30°C (A/Rubisco sites; Fig. 2C). k_cat values were significantly greater in NADP-ME than in NAD-ME species (Fig. 3; Table IV).

Figure 2.

Relationships between assimilation rates (A) or Rubisco sites (B) and N content and between assimilation rates per Rubisco sites (in vivo k_cat at 30°C) and Rubisco in vitro k_cat at 25°C (C) for 7 NAD-ME (•, ▴) and 7 NADP-ME (○, ▵) C₄ grasses grown under high (•, ○; HN) and low (▴, ▵; LN) soil N supplies (experiment 3; Table I). Lines represent linear fits for NAD-ME (solid line, A: y = 15 + 0.16x, r² = 0.70; B: y = 2.18+ 0.05x, r² = 0.62), NADP-ME (dotted line, A: y = 14 + 0.24x, r² = 0.72; B: y = 0.74+ 0.06x, r² = 0.72), and all data (solid line, C: y = 1.14+ 0.79x, r² = 0.62).

Figure 3.

Box-whisker plots of assimilation rates per Rubisco sites (in vivo k_cat at 30°C; A) and Rubisco in vitro k_cat at 25°C (B) for 7 NAD-ME and 7 NADP-ME C₄ grasses grown under high (HN) and low (LN) soil N supplies (experiment 3; Table I). Other details are similar to Figure 1.

Table IV.

In vivo (assimilation rate [30°C] per Rubisco sites) and in vitro (25°C) k_cat (mol CO₂ [mol sites]⁻¹ s⁻¹) of 7 NAD-ME and 7 NADP-ME grasses grown under high (HN) and low (LN) soil N supply (experiment 3; Table I)

Other details are as described for Table III.

Species	A/Rubisco (30°C)		In Vitro kcat (25°C)
	HN	LN	HN
NAD-ME
Astrebla lappacea	3.7 ± 0.3	4.2 ± 0.1	3.9 ± 0.1
Astrebla pectinata	5.1 ± 0.5	4.7 ± 0.2	5.3 ± 0.3
Eragrostis superba	4.0 ± 0.3	4.5 ± 0.2	4.9 ± 0.3
Panicum coloratum	2.8 ± 0.2	4.9 ± 0.3	2.8 ± 0.2
Panicum decompositum	3.5 ± 0.3	4.6 ± 0.7	3.4 ± 0.0
Panicum miliaceum	4.0 ± 0.9	5.9 ± 0.1	1.9 ± 0.0
Panicum virgatum	4.3 ± 0.2	5.5 ± 0.1	4.7 ± 0.6
NADP-ME
Bothriochloa biloba	7.1 ± 0.5	11.5 ± 1.6	6.0 ± 0.3
Bothriochloa bladhi	6.8 ± 0.5	5.4 ± 0.1	6.2 ± 0.3
Cenchrus ciliaris	6.5 ± 0.4	11.4 ± 1.1	6.1 ± 0.3
Dichanthium sericeum	4.4 ± 0.0	9.6 ± 0.3	6.1 ± 0.1
Paspalum dilatatum	4.5 ± 0.9	5.6 ± 0.5	4.6 ± 0.1
Paspalum notatum	5.6 ± 0.0	5.8 ± 0.0	4.9 ± 0.1
Pennisetum clandestinum	6.7 ± 0.7	7.8 ± 0.1	6.4 ± 0.2
Average (NAD-ME)	3.9 ± 0.3	4.9 ± 0.2	3.8 ± 0.5
Average (NADP-ME)	5.9 ± 0.4	8.2 ± 1.0	5.7 ± 0.3
P (subtype)	***	***	***
P (N)	*	*
P (N × subtype)	ns	ns

Leaf N Partitioning in the BS and M (Experiment 4)

The two NAD-ME species (Panicum miliaceum and Panicum coloratum) allocated relatively more of their leaf N to Rubisco (8%) and other soluble proteins (45%) compared with the two NADP-ME species (Sorghum bicolor and Cenchrus ciliaris; 5% and 36%, respectively; Fig. 4). In the larger survey of 14 C₄ grasses, Rubisco N fraction also tended to be slightly larger in the NAD-ME than NADP-ME grasses (Table III). For the 4 grasses, 30% of leaf N was recovered in the thylakoids, 8% in amino acids, and 5% as nitrate. Up to 4% and 17% of total leaf N were unaccounted for in the NAD-ME and NADP-ME leaves, respectively (Fig. 4). About 60% to 67% of total leaf N and Chl were allocated to the BS tissue of NAD-ME species compared with 33% to 38% for NADP-ME species (Table V). This uneven and opposite N and Chl distribution between BS and M tissues was corroborated by the pattern of Chl a autofluorescence (Fig. 5). Most of the Chl a autofluorescence signal came from the BS of the two NAD-ME species (Fig. 5, A and B), whereas the strongest signal emanated from the surrounding M tissue in the two NADP-ME species (Fig. 5, C and D). The distribution of thylakoid N followed that of Chl. About 48% to 60% and 32% to 37% of total thylakoid N were found in the BS of NAD-ME and NADP-ME leaves, respectively (Table V). About 62% and 55% of soluble protein were found in the BS of NAD-ME and NADP-ME leaves, respectively (Fig. 4; Table V). More than 70% of amino acids were found in the M of all 4 species. Nitrate was evenly distributed between the two tissues except in S. bicolor, in which most of it was in the M (Fig. 4; Table V).

Figure 4.

Total leaf N budget for P. miliaceum (A), P. coloratum (B), S. bicolor (C), and C. ciliaris (D; experiment 4; Table I). The contributions of soluble proteins without Rubisco, Rubisco, thylakoids, amino acids, nitrate, and others are calculated as described in “Materials and Methods.” The popped-out slices represent the BS fractions, whereas the remaining slices represent the M fractions. The percentage of total leaf N is shown next to each slice.

Table V.

Total leaf and BS Chl and N, thylakoid N, soluble protein, amino acids, nitrate, Rubisco, PSII, PSI, and Cyt f concentrations of two NAD-ME (P. miliaceum and P. coloratum) and two NADP-ME (S. bicolor and C. ciliaris) C₄ grasses (experiment 4; Table I)

Dark respiration (R_d) and dark-adapted F_v/F_m are also shown. Values are means ± se. Significance levels are as described in the legend of Figure 1. Within the same row, values followed by the same letter are not significantly different (P > 0.05).

Parameter	Fraction	P. miliaceum	P. coloratum	S. bicolor	C. ciliaris
Chl a/b	Leaf (n = 12)	4.2a ± 0.1	4.3ab ± 0.1	4.4b ± 0.1	4.4ab ± 0.1
	BS (n = 4)	4.1a ± 0.1	4.1a ± 0.2	6.3b ± 0.3	6.4b ± 0.5
	PFraction	ns	ns	**	**
Chl (μmol m−2)	Leaf (n = 12)	424a ± 26	425a ± 24	579b ± 36	464a ± 37
	BS (n = 4)	284b ± 18	256b ± 26	206a ± 15	164a ± 21
	BS/Leaf (PFraction)	0.60 (*)	0.59 (*)	0.33 (**)	0.38 (**)
Total N (mmol m−2)	Leaf (n = 6)	78a ± 10	87a ± 8	116b ± 3	87a ± 3
	BS (n = 3)	47b ± 5	51b ± 3	39ab ± 4	33a ± 2
	BS/Leaf (PFraction)	0.67 (*)	0.60 (***)	0.36 (***)	0.35 (**)
Thylakoid N (mmol N m−2)	Leaf (n = 3)	23.5a ± 1.7	25.2ab ± 0.4	30.3c ± 0.5	28.6bc ± 0.5
	BS (n = 2)	14.0c ± 0.3	12.0b ± 0.7	11.2b ± 0.3	9.2a ± 0.3
	BS/Leaf (PFraction)	0.60 (*)	0.48 (**)	0.37 (***)	0.32 (**)
Soluble protein (g m−2)	Leaf (n = 3)	3.8a ± 0.4	4.6a ± 0.3	3.7a ± 0.4	3.4a ± 0.2
	BS (n = 3)	2.3a ± 0.3	3.2a ± 1.1	2.0a ± 0.2	1.9a ± 0.1
	BS/Leaf (PFraction)	0.61 (*)	0.64 (*)	0.55 (*)	0.55 (*)
Amino acids (mmol m−2)	Leaf (n = 3)	6.6a ± 0.6	6.1a ± 0.9	10.0a ± 1.7	5.7a ± 0.8
	BS (n = 2)	1.3a ± 0.1	1.8a ± 0.5	1.3a ± 0.1	0.90a ± 0.1
	BS/Leaf (PFraction)	0.20 (**)	0.30 (*)	0.13 (*)	0.16 (*)
Nitrate (mmol m−2)	Leaf (n = 3)	3.7a ± 0.1	4.6b ± 0.4	7.5bb ± 1.2	2.7a ± 0.4
	BS (n = 2)	1.7a ± 0.3	2.1a ± 0.5	1.9a ± 0.1	1.3a ± 0.1
	BS/Leaf (PFraction)	0.46 (**)	0.45 (*)	0.25 (*)	0.49 (*)
Rubisco (μmol m−2)	Leaf (n = 12)	0.97b ± 0.06	1.25c ± 0.08	0.97b ± 0.06	0.73a ± 0.04
PSII (μmol m−2)	Leaf (n = 6)	1.03a ± 0.06	0.99a ± 0.08	1.35b ± 0.06	1.15a ± 0.06
	BS (n = 2)	0.18b ± 0.02	0.35c ± 0.01	0.02a ± 0.02	0.05a ± 0.01
	BS/Leaf (PFraction)	0.17 (***)	0.35 (***)	0.01 (***)	0.04 (***)
PSI (μmol m−2)	Leaf (n = 2)	1.06a ± 0.07	0.95a ± 0.05	1.01a ± 0.01	0.93a ± 0.13
	BS (n = 2)	0.26a ± 0.07	0.44b ± 0.02	0.37ab ± 0.01	0.36ab ± 0.03
	BS/Leaf (PFraction)	0.24 (*)	0.46 (*)	0.37 (***)	0.39 (*)
Cyt f (μmol m−2)	Leaf (n = 2)	1.03a ± 0.02	1.00a ± 0.05	1.00a ± 0.04	1.05a ± 0.16
	BS (n = 2)	0.39c ± 0.02	0.41c ± 0.03	0.20a ± 0.01	0.30b ± 0.01
	BS/Leaf (PFraction)	0.37 (**)	0.38 (*)	0.20 (**)	0.34 (*)
Rd (μmol m−2 s−1)	Leaf (n = 3)	−1.7a ± 0.12	−1.9a ± 0.08	−1.9a ± 0.10	−1.8a ± 0.40
Fv/Fm	Leaf (n = 3)	0.75a ± 0.01	0.81a ± 0.01	0.80a ± 0.01	0.80a ± 0.01
	BS (n = 3)	0.75b ± 0.01	0.81b ± 0.01	0.40a ± 0.01	0.45a ± 0.02
	PFraction	ns	ns	***	***

Figure 5.

Chl a autofluorescence of leaf cross sections of two NAD-ME (P. miliaceum [A] and P. coloratum [B]) and two NADP-ME (S. bicolor [C] and C. ciliaris [D]) C₄ grasses (experiment 4; Table I). The images were obtained using confocal microscopy. Cell walls are shown in green and Chl a autofluorescence in red. The BS/leaf ratios of Chl and total N are shown for each species. Note the change of scale between the NAD-ME and NADP-ME images.

Thylakoid N Partitioning in the BS and M (Experiment 4)

The content of functional PSII, PSI, and cytochrome (Cyt) f complexes approached 1 μmol m⁻² in all 4 C₄ grasses, except for a significantly higher PSII content in S. bicolor (Table V). Although thylakoids accounted for a similar fraction of leaf N (approximately 30%) in the four species (P > 0.05), vast differences were found in thylakoid composition between the BS and M (Fig. 6). Relative to M, BS tissues of NAD-ME species had lower functional PSI and PSII reaction centers and Cyt f, all expressed per unit Chl (Fig. 6). Per unit Chl, functional PSI did not differ between M and BS in NADP-ME species, whereas both PSII and Cyt f were lower in the BS (Fig. 6). Diminished PSII activity in BS of NADP-ME leaves resulted in reduced photochemical efficiency (F_v/F_m; Table V). Thylakoid N can be divided into N pools associated either with pigment-protein complexes or with ATP synthesis and electron transport complexes. When expressed on a Chl basis, measured thylakoid N increased linearly with Cyt f content (Fig. 7). Values measured on the C₄ grasses here scattered around those reported for C₃ leaves. Calculated values were generally less than measured values for the BS tissues (Fig. 7).

Figure 6.

Concentration of Rubisco (A), PSII (B), Cyt f (C), and PSI (D) proteins expressed per leaf (black bars), M (white bars), or BS (hatched bars) Chl in four C₄ grasses (experiment 4; Table I) and spinach (Evans and Terashima, 1987). Values are means ± se. See Table V for number of replications.

Figure 7.

Thylakoid N cost as a function of Cyt f content per unit Chl. Closed symbols are values measured either on whole-leaf extracts (▪) or BS (•) thylakoids, or by difference for the M thylakoids (♦). Half-filled symbols represent thylakoid N cost calculated from the contents of functional PSII, PSI, and Cyt f for each species and fraction. Open symbols denote thylakoids prepared from C₃ leaves (spinach [Terashima and Evans, 1988] and pea [Evans, 1987]). The solid line shows the calculated N cost for thylakoids with 2 mmol PSI (mol Chl)⁻¹, 2.5 mmol PSII (mol Chl)⁻¹, and varying Cyt f content, which represents the average complex concentrations observed for the C₄ leaves.

DISCUSSION

N-Use Efficiency and Rubisco in NAD-ME and NADP-ME C₄ Grasses

This study was sparked by our observation that C₄ grasses belonging to the NADP-ME subtype have higher NUE and PNUE and lower leaf N content than NAD-ME grasses. Since then, we consolidated this finding in 3 separate experiments using various combinations of 27 NAD-ME and NADP-ME grasses grown under adequate and deficient soil N supplies (Fig. 1; Table II). Along the same line, Bowman (1991) found that two NADP-ME Panicum species accumulated more biomass per total shoot N than four NAD-ME species. When Taub and Lerdau (2000) compared the response of A to leaf N in three NAD-ME and three NADP-ME grasses, they concluded that species variations were greater than differences in PNUE between the two C₄ subtypes. It is likely that the small number of species used in the latter study hindered the emergence of a clear trend. Here, we report a consistent and convincing difference (P < 0.05) in NUE, PNUE, and leaf N for a large number of NAD-ME and NADP-ME C₄ grasses (Fig. 1). Differences in leaf N concentration were counterbalanced by changes in stem and root N, such that the same amount of total N was required to produce a gram of whole-plant dry mass in both types of C₄ grasses (Table II). Our subsequent investigation focused on dissecting leaf N in order to identify the N pool(s), which may contribute to the difference between the two subtypes in leaf N and PNUE.

Using four C₄ grasses, we found that NAD-ME species invest a greater proportion of leaf N in soluble protein and Rubisco relative to NADP-ME species. In a larger survey, NADP-ME grasses achieved the same photosynthetic rates as NAD-ME counterparts with less leaf N and Rubisco contents. Greater A per Rubisco sites (in vivo k_cat) in NADP-ME than in NAD-ME species suggests, among other things, that there is an intrinsic difference in Rubisco k_cat between the two subtypes. This was confirmed by in vitro k_cat measurements. Under the high light and temperature conditions used during our gas exchange measurements, C₄ photosynthesis operates near maximal Rubisco activity. Therefore, the close correlation between Rubisco's in vivo and in vitro k_cat indicates that most variations in photosynthetic efficiency per Rubisco and, hence, per leaf N between grasses of the NAD-ME and NADP-ME subtypes are related to differences in Rubisco k_cat (Fig. 2C). In accordance with our results, Seemann et al. (1984) measured Rubisco k_cat (at 30°C) for a number of C₃ and C₄ species, obtaining values of 4.8 (s⁻¹) for 1 NAD-ME and 6.4 to 7.6 (s⁻¹) for 4 NADP-ME grasses. Comparisons of Rubisco kinetics reveal that the higher k_cat of C₄ relative to C₃ Rubisco entails a greater Michaelis-Menten constant for CO₂ (K_{m [CO2]}; Yeoh et al., 1980, 1981; Seemann et al., 1984; Wessinger et al., 1989; Sage, 2002). Given that Rubisco operates near CO₂ saturation in C₄ plants, C₄ photosynthesis is more sensitive to increases in Rubisco k_cat than K_{m (CO2)} (von Caemmerer, 2000). In contrast with the C₃ versus C₄ comparison, differences in k_cat between the two C₄ subtypes do not appear to have had implications on K_{m (CO2)}. In an extensive survey, Yeoh et al. (1980) found no significant difference in K_{m (CO2)} between NAD-ME and NADP-ME grasses. Although a link between Rubisco k_cat and K_{m (CO2)} is theoretically justified (Morell et al., 1992), the two parameters have been shown to vary independently in Rubisco from different origins or when subjected to site-directed mutagenesis (Whitney et al., 1999, 2001). There is suggestion that changes in Rubisco k_cat are accompanied by a trade off in specificity for CO₂ relative to O₂ (S_c/o; Zhu et al., 2004). However, S_c/o does not appear to differ much among higher plants (Kane et al., 1994). This argument remains constrained by the scarcity of kinetic data for C₄ Rubisco (von Caemmerer and Quick, 2000). In brief, NADP-ME grasses achieved similar photosynthetic capacity to NAD-ME counterparts with less leaf N and Rubisco having higher k_cat, mirroring the well known differences between C₃ and C₄ plants.

N Partitioning in the Leaves of NAD-ME and NADP-ME C₄ Grasses

When grown under high light, C₃ leaves typically allocate 22% of leaf N to thylakoids and 60% to soluble protein, one-third of which is Rubisco (Evans and Poorter, 2001). In our study, C₄ leaves allocated 30% of leaf N to thylakoids, and 41% (NADP-ME) and 53% (NAD-ME) to soluble protein, one-seventh of which is Rubisco. This distribution is close to what Makino et al. (2003) reported for maize (NADP-ME), in which 34% and 33% of leaf N were found in thylakoids and soluble protein, respectively. The lower soluble protein fraction of C₄ relative to C₃ leaves increases the apparent allocation of N to thylakoids in C₄ leaves. Hence, the amount of soluble protein per Chl is greater for C₃ (10–20 g [mmol Chl]⁻¹) than C₄ leaves (6.3–7.4 NADP-ME, 8.9–10.8 NAD-ME). Using available literature values, Evans and von Caemmerer (2000) calculated the soluble protein cost of the C₃ and C₄ cycles. They estimated similar costs for C₃ and NAD-ME C₄ leaves (6.4 g soluble protein [mmol Chl]⁻¹) and a slightly lower value for NADP-ME leaves (6.1 g soluble protein [mmol Chl]⁻¹). The C₃ monocots wheat (Triticum aestivum; Evans, 1983) and rice (Oryza sativa; Makino et al., 1997) have about 10 g of soluble protein (mmol Chl)⁻¹, which is equivalent to the NAD-ME grasses. The slightly higher protein cost of the NAD-ME than the NADP-ME C₄ subtype may reflect the high activity of Asp and Ala aminotransferases in NAD-ME leaves. The calculation of Evans and von Caemmerer (2000) was based on measured Rubisco content of Amaranthus edulis (C₄ dicot), in which Rubisco accounted for 8% of leaf N (Sage et al., 1987). This proportion is similar to what Makino et al. (2003) reported for maize and is close to the upper bound of values observed in our study (Figs. 2B and 4). Our reported values for the soluble protein cost per Chl is 12% (NADP-ME) and 55% (NAD-ME) above that estimated for just the C₃ and C₄ cycle enzymes by Evans and von Caemmerer (2000). In brief, C₄ (relative to C₃) and NADP-ME (relative to NAD-ME) leaves allocate a lower proportion of N to soluble protein relative to thylakoids.

N Partitioning between the BS and M of NAD-ME and NADP-ME C₄ Grasses

The pattern of Chl distribution found in our study agrees with what has been reported for C₄ species (Mayne et al., 1975; Jenkins and Boag, 1985). The distribution of total leaf N between the M and BS mirrors that of Chl, indicating that most leaf N is linked to photosynthesis in the C₄ grasses. The intercellular compartmentation of the photochemical reactions of C₄ photosynthesis is less well documented than that of the CO₂ fixation reactions (Edwards et al., 1976; Hatch and Osmond, 1976; Hatch, 1987; Edwards and Krall, 1992). BS deficiency in PSII activity, as reported here and by other workers, represents the best documented example of the unequal composition of BS and M thylakoids in C₄ leaves (Mayne et al., 1975; Edwards et al., 1976; Ghirardi and Melis, 1984; Meierhoff and Westhoff, 1993). Suppressing PSII activity in the gas-tight BS cells serves to prevent the accumulation of detrimentally high O₂ concentrations. Our data indicate that the 2 NAD-ME grasses have also gone a long way in locating 65% to 83% of PSII activity in the M, away from the BS. This is higher than what has been reported for the NAD-ME grasses P. miliaceum and Panicum capillare: 46% based on the Hill reaction and 13% based on delayed light emission (Mayne et al., 1975; Edwards et al., 1976).

This discrepancy may be reconciled by recent evidence suggesting that BS thylakoids of NADP-ME species (e.g. maize and S. bicolor) contain incomplete and inactive PSII centers, devoid of the oxygen evolving complex (Meierhoff and Westhoff, 1993; Bassi et al., 1995). Data presented in Table V are consistent with there being inactive PSII complexes in the BS for two reasons. First, low PSII activity was observed in the BS of the NADP-ME species, whereas PSI content per unit Chl was similar for both BS and M thylakoids (Fig. 6B) and the Chl a/b ratio was higher in the BS, meaning that there could be no increase in the proportion of Chl associated with light-harvesting complexes (LHCs). Secondly, the observed amount of N per unit Chl in BS thylakoids exceeded that predicted on the basis of active PSII, PSI, and Cyt f complexes for all four species (Fig. 7). The N cost per Chl (mol N [mol Chl]⁻¹) of PSII is 83, whereas it is only 33 and 26 for PSI and LHCs, respectively. Any underestimate of PSII content reduces calculated thylakoid N content. Thus, it is likely that BS thylakoids of NAD-ME and NADP-ME species contain varying degrees of functional and nonfunctional PSII centers. PSII distribution between M and BS and active and inactive complexes varies among species and may be influenced by growth conditions.

The uneven PSII distribution necessitates a substantial 3-phosphoglycerate/triose phosphate (PGA/TP) shuttle service between the M and BS of NADP-ME and NAD-ME leaves. This is possible because the enzymes for PGA reduction are present in both cell types and subtypes (Hatch and Osmond, 1976). In addition, NADP-ME and NAD-ME species possess sufficient PGA and TP concentration gradients between the M and BS, such that up to 50% of PGA produced in the BS can be reduced in the M (Hatch and Osmond, 1976; Leegood and von Caemmerer, 1988, 1989). Interestingly, Rubisco content per Chl in the BS is high and approximates those found in C₃ leaves (Fig. 4A). Thus, the presence of functional and nonfunctional PSII centers in BS tissue may enable the thylakoid membrane to better compact and make more room for Rubisco in the stroma (Chow, 1999). Although PSI was more equally distributed between the M and BS in 3 species, 76% of leaf PSI activity was found in the M of P. miliaceum. This departs from early reports (Mayne et al., 1975) and suggests that antenna size of the photosystems is smaller in the M than BS of P. miliaceum. The distribution of thylakoid complexes between the M and BS of C₄ grasses does not mirror closely the distribution of Chl and N.

The close match between measured and calculated thylakoid N contents (Fig. 7) suggests that N costs derived from C₃ plants are appropriate for C₄ plants. This is surprising given the differences in energetic requirements between the M and BS and the C₃ and C₄ cycles. The N cost of 8.85 mol N (mol Cyt f)⁻¹ suggested by Evans and Seemann (1989) and used here is considerably greater than that suggested by Hikosaka and Terashima (1995; 5.43 mol N [mol Cyt f]⁻¹). The uncertainty comes largely from there being few quantitative measurements of ATP synthase. Makino et al. (2003) compared leaf N budgets from rice with maize and measured both Cyt f and ATP synthase. The amount of N in ATP synthase relative to Cyt f was 4.7 and 5.3 for rice and maize, respectively, close to the value of 4.4 assumed here for the calculations (Fig. 7). The fraction of leaf N associated with ATP synthase and electron transport components was 5.7% and 9.4% for rice and maize, respectively, compared to 7.6% to 11.7% calculated for the 4 C₄ species here. The proportion calculated for 10 dicotyledonous C₃ species is similar, averaging 8.2% (Evans and Poorter, 2001).

CONCLUSION

In summary, NADP-ME grasses achieved similar photosynthetic capacity to NAD-ME counterparts with less leaf N and Rubisco having higher k_cat. Compared to NADP-ME, NAD-ME leaves allocate a greater fraction of leaf N to soluble protein and a similar fraction to thylakoids. Although the majority of leaf Chl and N were found in the BS of NAD-ME and M of NADP-ME grasses, the distribution of thylakoid complexes was species specific.

MATERIALS AND METHODS

This study reports on four experiments (Table I). Experiments 1, 2 (related to Ghannoum et al., 2001, 2002), and 3 were used for NUE and A/N calculations. Experiment 3 was also used for Rubisco activity assays. Experiment 4 was used for N budget analysis.

Plant Culture (Experiments 1–3)

Soil was supplemented with basal nutrients (excluding N) and added to 5- to 10-L pots (Ghannoum and Conroy, 1998; Ghannoum et al., 2001). Briefly, the soil was leached of readily available N by repeatedly washing it with water. Pots were transferred to a temperature-controlled glasshouse, where air temperatures ranged between 28°C to 30°C /22°C to 24°C day/night. Midday photosynthetic active radiation averaged 1,000 μmol m⁻² s⁻¹. Seeds for 13 NAD-ME and 14 NADP-ME C₄ grasses (Table I) were obtained from local commercial suppliers or CSIRO Tropical Agriculture, St Lucia, Australia, and sown directly into the potted soil. Pots were watered to full capacity once daily with either water (low N treatment) or 60 mg N kg⁻¹ soil (high N treatment) supplied as NH₄NO₃. After germination, plants were thinned to two to four seedlings per pot. There were three to four pots per species and N treatment.

Plant Culture (Experiment 4)

Seeds for two NAD-ME (Panicum miliaceum and Panicum coloratum) and 2 NADP-ME (Sorghum bicolor and Cenchrus ciliaris) C₄ grasses were sown directly in 4-L pots containing sterilized garden soil supplemented with 4 g of a controlled release fertilizer (15/4.8/10.8/1.2 nitrogen/phosphorus/potassium/magnesium plus trace elements [boron, copper, iron, manganese, molybdenum, and zinc]; Osmocot Plus; Scotts, Baulkam Hills, Australia). Plants were grown in controlled environment chambers (Phoenix, Adelaide, Australia) lit for 9 h with metal halide lamps supplying 800 μmol quanta m⁻² s⁻¹ followed by 1 h of incandescent lighting. Air temperature and relative humidity were maintained at 28/24°C and 60/80% day/night, respectively. Plants were watered daily and used for analysis 4 weeks after germination.

Growth and NUE Measurements (Experiments 1–3)

Plants were harvested 5 to 7 weeks after germination and separated into leaves, stems (including sheaths), and roots. Leaf area was determined by a digital image analyzer (Delta-T, Cambridge, UK). Roots were washed free of soil. Harvested samples were oven-dried at 80°C, weighed, and then ground to powder. Percentage of N was determined on the ground tissues using a flash combustion CNS analyzer (Fison NA1500; Fison Instruments, Milan, Italy). NUE was calculated as the ratio of plant dry mass to total leaf N content at harvest.

Gas Exchange Measurements (Experiments 1–4)

Measurements were made on the youngest fully expanded leaves 1 week before harvest using a portable photosynthesis system (LI-6400; LI-COR, Lincoln, NE), at a photosynthetic photon flux density of 1,500 (experiments 1 and 2) and 1,800 (experiments 3 and 4) μmol m⁻² s⁻¹ supplied by an in-built red/blue LED light source, a CO₂ partial pressure of 380 μbar, leaf temperature of 30°C, and leaf-to-air vapor pressure difference between 1.5 to 2.0 kPa. Dark respiration (R_d) was measured after 30 min of dark adaptation, before the light was turned on. In experiment 2, the response of A to step increases of intercellular CO₂ (C_i) was measured. A/C_i curves were fitted using the C₄ photosynthesis model of von Caemmerer (2000) to estimate maximal phosphoenolpyruvate carboxylase (V_{p max}) and Rubisco (V_{c max}) activities. At the end of gas exchange measurements, leaves were cut under high light (>1,000 μmol m⁻² s⁻¹) and either immersed in liquid N₂ then stored at −80°C for biochemical analysis or their area determined before being oven-dried, weighed, and their percentage of N measured as described above.

Preparation of BS Strands (Experiment 4)

The four grass species (Table I) were chosen based on their ability to produce highly pure BS strands by mechanical blending as checked under a light microscope (Agostino et al., 1989). The midrib was removed from the youngest fully expanded leaves, which were cut into 10- × 2-mm sections using a razor blade, soaked in distilled water, then blotted dry. About 7 g of fresh leaf sections were mixed with 100 mL of ice-cold blending buffer (50 mm Na-PO₄, 0.33 m sorbitol, 4 mm dithiothreitol (DTT), 5 mm MgCl₂, 2 mm EDTA, pH 7.5) in a 250-mL cup of a Sorvall (Newton, CT) Omnimixer, and blended for 5 × 10 s at 60% speed. The homogenate was filtered through 1-mm, 0.5-mm, then 88-μm nylon mesh. The BS strands collected on the 88-μm mesh were washed copiously with blending buffer, then stored on ice.

Rubisco, Chl, Soluble Protein, Amino Acids, and Nitrate (Experiments 3 and 4)

Leaf sections (1–2 cm²) or BS preparations (100 nmol Chl) were extracted in 1 mL of ice-cold buffer (50 mm EPPS-NaOH, 5 mm DTT, 15 mm NaHCO₃, 20 mm MgCl₂, 2 mm EDTA, 4% protease inhibitor cocktail [Sigma, St. Louis], 0.1% [w/v] polyvinylpolypyrrolidone, pH 8.0) and with or without 0.05% Triton X-100 using a 2-mL Potter-Elvehjem glass homogenizer kept on ice. Subsamples were taken from the crude homogenate for Chl determination in 80% acetone (Porra et al., 1989). The remaining homogenate was centrifuged at 16,100g for 0.5 min and the supernatant used for the following analysis. Total Rubisco content was estimated by the irreversible binding of [¹⁴C]2-carboxy-d-arabinitol 1,5-bisphosphate to the fully carbamylated enzyme (Ruuska et al., 1998). Maximal Rubisco activity was assayed within 5 min of extraction, by incubating 20 μL of supernatant in 500 μL of reaction buffer (50 mm EPPS-NaOH, 20 mm MgCl₂, 1 mm EDTA, 10 mm NaH¹⁴CO₃, pH 8.0) for 5 min at 25°C. Reaction was initiated by adding 0.5 mm ribulose 1,5-bisphosphate and stopped after 1 min with 7% formic acid. Acid-stable ¹⁴C and specific activity of NaH¹⁴CO₃ were then determined (Ruuska et al. 2000). In vitro k_cat was calculated as the ratio of Rubisco activity to content of [¹⁴C]2-carboxy-d-arabinitol 1,5-bisphosphate binding sites, determined for the same leaf extract. Tobacco leaves were used as a reference with each run of activity assays, yielding k_cat between 3.2 to 3.4 s⁻¹. Soluble proteins were measured using a Coomassie Plus kit (Pierce). Free amino acids and nitrate were determined according to Moore (1968) and Cataldo et al. (1975), respectively.

Thylakoid Preparation and PSI and Cyt f Measurements (Experiment 4)

Leaf sections or BS preparations were extracted in a 7-mL Potter-Elvehjem glass homogenizer in ice-cold extraction buffer (50 mm Na-PO₄, 0.33 m sorbitol, 2 mm DTT, 5 mm MgCl₂, 2 mm EDTA, pH 6.5). Several extractions were carried out and pooled. The extracts were filtered as described above. The filtrate was centrifuged for 1 min at 1,000g and 4°C. The pellet, enriched in tissue debris and starch, was discarded. The supernatant was centrifuged for 4 min at 2,000g. The pellet was gently suspended and centrifuged twice in 40 mL of extraction buffer before being incubated for 5 min on ice in lysis buffer (extraction buffer devoid of sorbitol), then centrifuged. The pellet was finally washed, suspended in extraction buffer, snap frozen in liquid N₂, then stored at −80°C until used for PSI, Cyt f, or total N analysis. PSI was calculated as the amount of P₇₀₀ using the absorbance change at 702 nm induced by flashes of blue-green light, after correcting for fluorescence (Evans, 1987). Cyt f complex was estimated from hydroquinol-reduced minus ferricyanide-oxidized difference spectra using a dual beam spectrophotometer (model 557; Perkin Elmer, Foster City, CA; Evans, 1987).

Functional Centers and Photochemical Efficiency of PSII (Experiment 4)

Functional PSII centers of leaf and BS tissue were quantified by the O₂ yield at 1% CO₂ and repetitive flashes at 10 Hz, with continuous background far-red light (Chow et al., 1989). BS preparations were layered on glass filter paper (Whatman GF/C; W&R Balston, Maidstone, England), and a gentle suction was applied to remove excess buffer. Photochemical efficiency of PSII was measured following 15 min of dark adaptation of leaf sections or BS preparations, using a portable fluorometer (Plant Efficiency Analyzer; Hansatech Instruments, Norfolk, UK).

Total Leaf, BS, and Thylakoid N (Experiment 4)

Total leaf N was determined on oven-dried, ground leaf sections taken from the same or matching leaves used for various analyses. BS fractions taken after Rubisco determination were oven-dried in preweighed tubes, weighed, then a subsample analyzed for total percentage of N. Similar results were obtained whether fractions were washed in buffer or distilled water before drying. To obtain pure thylakoids for N analysis, crude thylakoid preparations (0.5–1.0 μmol Chl) were layered over a centrifuge tube containing 30% Percol in washing buffer (25 mm Na-PO₄, 5 mm MgCl₂, pH 6.5) and centrifuged at 2,000g for 4 min and 4°C. The pure thylakoid bands situated about 1 cm from the top were sucked out and washed twice with the same washing buffer, then suspended in distilled water. Subsamples with known Chl content were dried in tin cups at 80°C then analyzed for percentage of N.

Chl a Autofluorescence (Experiment 4)

Freshly hand cut, leaf cross sections were stained with 0.01 mg mL⁻¹ propidium iodide to highlight cell walls, then mounted on a scanning confocal microscope (Leica SP2 LSCM; Leica Microsystems, Wetzlar, Germany). Sections were excited with weak laser beams at 488 nm for autofluorescence and 543 nm for the stain. Emission was collected between 690 to 740 nm and 550 to 620 nm for Chl a autofluorescence and the stain, respectively.

N Budget Calculation (Experiment 4) and Data Analysis

Leaf parameters were measured on area, Chl, and/or Rubisco basis. BS parameters were measured on a Chl and/or Rubisco basis. BS parameters (P) were converted to leaf area basis using the following formula (Rubisco exclusively located in BS):

M proportion was calculated as the difference between leaf and BS fractions. Total N budgets for the BS and M were determined assuming that soluble protein (including Rubisco) contains 16% N by mass, molecular mass of 550,000g mol⁻¹ for Rubisco, and that amino acids and nitrate contain 1 mol N per mol. Thylakoid N cost, T (mol N [mol Chl]⁻¹), was calculated according to Evans and Seemann (1989) and Hikosaka and Terashima (1995) from the following equation: T = [PSII] × 83.3 × 0.06 + [PSI] × 32.8 × 0.184 + [LHC] × 26 × 0.013 + [Cyt f] × 8.85, where [LHC] = (1,000 − [PSII] × 60 − [PSI] × 184)/13. [PSII], [PSI], [LHC], and [Cyt f] have the units mmol (mol Chl)⁻¹.

The pigment protein complexes have N costs of 83.3, 32.8, and 26 mol N (mol Chl)⁻¹ and bind 60, 184, and 13 mol Chl (mol complex)⁻¹ for PSII, PSI, and LHC, respectively. The N cost associated with Cyt f was 8.85 mol N (mol Cyt f)⁻¹.

The subtype effect (experiments 1 and 2) was calculated by two-way ANOVA (nested design, species nested in subtype) using a general linear model. In experiment 3, a three-way, nested ANOVA was used. Species and fraction effects (experiment 4) were analyzed by one-way ANOVA. For data presented in Tables II and V, a posthoc, Tukey HSD test was carried out on the grouped means.