| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Anatomy and Morphology |
Department of Ecology and Evolutionary Biology, University of Toronto, Toronto, Ontario, M5S 1A1, Canada; Department of Biological Sciences, Yarmouk University, Irbid, Hashemite Kingdom of Jordan
Received for publication July 13, 2006. Accepted for publication December 28, 2006.
ABSTRACT
C4 photosynthesis and Kranz anatomy occur in 16 eudicot families, a striking example of convergent evolution. Biochemical subtyping for 13 previously undiagnosed C4 eudicot species indicated that 10 were NADP-malic enzyme (ME) and three were NAD-ME. A total of 33 C4 species, encompassing four Kranz anatomical types (atriplicoid, kochioid, salsoloid, and suaedioid), and 21 closely related C3 species were included in a quantitative anatomical study in which we found that, unlike similar studies in grasses and sedges, anatomical type had no predictive value for the biochemical subtype. In a multivariate canonical discriminant analysis, C4 species were distinguished from C3 species by the mesophyll to bundle sheath ratio and exposure of the bundle sheath surface to intercellular space. Discrimination between NADP-ME and NAD-ME was not significant, although in a Mantel test grouping by biochemical subtype was significant, while grouping by family was not. This comprehensive survey of C4 anatomy and biochemistry unequivocally demonstrated that atriplicoid anatomy and NADP-ME biochemistry predominate in many evolutionary lineages. In addition to a main decarboxylating enzyme, high activity of a second decarboxylating enzyme was often observed. Notably, PEP-carboxykinase activity was significant in a number of species, demonstrating that this enzyme could also serve as a secondary pathway for C4 metabolism in eudicots.
Key Words: C4 photosynthesis • Kranz anatomy • mesophyll to bundle sheath ratio • NAD-malic enzyme • NADP-malic enzyme • PEP-carboxykinase enzyme • vein density
In the majority of C4 plants, functioning of the C4 pathway requires metabolic cooperation of two adjoining and distinct chlorenchymatous tissue types: outer mesophyll (or photosynthetic carbon assimilative [PCA]) and inner bundle sheath (or photosynthetic carbon reductive [PCR]) tissues. These tissues are arranged concentrically with respect to vascular tissues in C4 shoots, a structural pattern known as Kranz anatomy. Kranz anatomy provides one of the best examples of the intimate connection between plant form and function and represents a suite of structural characters that have evolved repeatedly from C3 ancestors (Dengler and Nelson, 1999
; Kellogg, 1999
; Sage 2001
, 2004
). This internal architecture physically partitions the biochemical events of the C4 pathway into two main phases. In the first phase, atmospheric CO2 is initially assimilated into C4 acids by PCA-tissue-specific phosphoenolpyruvate carboxylase. In the second phase, these acids diffuse into the PCR compartment, where they are decarboxylated, and the released CO2 is re-fixed by PCR-tissue-specific Rubisco. This biphasic C4 system enhances CO2 levels around Rubisco, suppressing photorespiration and improving plant carbon balance (Kanai and Edwards, 1999
).
Three subtypes of the C4 biochemical pathway are defined, based on the leading C4 acid decarboxylase that liberates CO2 from C4 acids in the PCR compartment: NADP-malic enzyme (NADP-ME), NAD-malic enzyme (NAD-ME), and phosphoenolpyruvate carboxykinase (PEP-CK) (Kanai and Edwards, 1999
). These three subtypes occur in grasses (Poaceae), and each is well characterized by a "classical" suite of anatomical and ultrastructural features, including the number of bundle sheath layers, the position of chloroplasts within the PCR cells, the development of PCR chloroplast grana, the number and size of PCR mitochondria, and the occurrence of a suberin lamella within the PCR cell walls (Gutierrez et al., 1974
; Hatch et al., 1975
; Hattersley and Browning, 1981
; Yoshimura et al., 2004
; Ueno et al., 2005
). More extensive surveys led to the identification of 14 distinct anatomical–biochemical combinations within C4 grasses (Prendergast and Hattersley, 1987
; Prendergast et al., 1987
). Despite this phylogenetic and anatomical diversity, certain quantitative anatomical characters were strongly correlated with biochemical subtype (Hattersley, 1984
; Dengler et al., 1994
). For instance, "classical" NADP-ME grasses have centrifugally placed PCR chloroplasts with reduced PCR granal development, significantly closer vein spacing, more PCA tissue, and less PCR tissue than the NAD-ME and PCK biochemical types, regardless of lineage (Dengler et al., 1994
). The striking anatomical feature shared by the three C4 biochemical subtypes in grasses is the reduced volume of PCA relative to PCR tissues, a trait that shortens the path length for the rapid bidirectional flow of C4 metabolites between PCA and PCR tissues (Hattersley, 1984
; Dengler et al., 1994
). Largely because PCR sheaths are positioned adjacent to vascular tissue (with a few exceptions), the reduction of PCA and enhancement of PCR tissue is associated with a decrease in interveinal distance in C4 grasses (Crookston and Moss, 1974
; Kawamitsu et al., 1985
; Ohsugi and Murata, 1986
; Dengler et al., 1994
; Ueno et al., 2006
).
In contrast to C4 grasses, only NADP-ME and NAD-ME subtypes are present in eudicots; there are no known PEP-CK type species (Sage, 2004
). Also unlike the grasses, these two subtypes are not obviously distinguishable on the basis of anatomical characters; for example, PCR cell chloroplasts adopt a centripetal position in both NADP-ME and NAD-ME subtypes (Carolin et al., 1978
; Dengler and Nelson, 1999
), with infrequent exceptions where chloroplasts are centrifugally placed (Carolin et al., 1978
; Freitag and Stichler, 2000
; Edwards et al., 2004
). As in C4 grasses, however, the two biochemical subtypes in C4 eudicots are distinguished by pronounced cytological dimorphism between PCA and PCR cells (Olesen, 1974
; Edwards and Walker, 1983
; Voznesenskaya et al., 1999
). For instance, PCR chloroplasts in NADP-ME subtype C4 eudicots have greatly reduced grana (associated with a high ratio of photosystem [PS] I to PS II), while those of PCA cells possess well-developed grana (abundant levels of PSI and PSII). The higher PSI : PSII ratio in PCR chloroplasts of NADP-ME species reflects a higher proportion of cyclic electron flow relative to linear electron flow. The reverse is true for PCR chloroplasts of NAD-ME-subtype C4 plants (Kanai and Edwards, 1999
; Voznesenskaya et al., 1999
; Takabayashi et al., 2005
). Although photosynthetic subtypes in C4 eudicots can be distinguished at the ultrastructural level, diagnostic features at the level of light microscopy remain to be identified.
Advances in phylogenetic analyses still allow for an approximation of the distribution of C4 photosynthesis in the eudicots. In an initial survey, Sage (2004)
listed approximately 30 C4 lineages within 16 eudicot families, though more recent phylogenies indicate that this is an underestimate (McKown et al., 2005
; Müller and Börsch, 2005
). Sage (2004)
included the known anatomical types and biochemical subtype(s) of many of the C4 eudicot lineages; however, gaps are apparent in the record. Moreover, in most groups, the anatomical patterns of closely related C3 species are unknown, making it difficult to relate pre-existing patterns in C3 species with the more derived C4 Kranz anatomy. Among the eudicot families in which the C4 syndrome has arisen, Chenopodiaceae contains the greatest number of C4 species and has the most biochemical, anatomical, and ecological diversity of all C4 eudicot lineages (Pyankov et al., 2001b
; Sage, 2001
; Kadereit et al., 2003
; Edwards et al., 2004
). Variation in Kranz anatomy in C4 chenopods has led to the recognition of five distinct forms: atriplicoid, kochioid, salsoloid, Kranz-suaedoid, and conospermoid (Carolin et al., 1975
; Freitag and Stichler, 2000
). Many of the other C4 eudicot evolutionary lineages are not as well characterized. Specifically, C4 members of the genera Aerva (Amaranthaceae); Boerhavia (Nyctaginaceae); Trianthema, Cypselea, and Zaleya (Aizoaceae); Calligonum (Polygonaceae); Heliotropium (Boraginaceae); Pectis (Asteraceae); Zygophyllum and Kallstroemia (Zygophyllaceae); Blepharis (Acanthaceae); Polycarpaea (Caryophyllaceae); and Anticharis (Scrophulariaceae) have not yet been characterized either biochemically and/or anatomically.
The main objectives of the present work were to determine the biochemical subtype for C4 eudicots that had not yet been characterized and to identify quantitative anatomical characteristics that might discriminate between C3 and C4 groups as a whole as well as between the NADP-ME and NAD-ME biochemical subtypes. Because quantitative anatomical traits have high predictive value for biochemical subtype in the grasses and sedges, we hypothesized that similar correlations would be observed for the eudicots. To test this, we analyzed mature photosynthetic and nonphotosynthetic tissues from a broad spectrum of C4 eudicots and related C3 species (but excluding the conospermoid and the single-cell C4 chenopods; Edwards et al., 2004
). Using quantitative characteristics that have been shown to distinguish between biochemical subtypes in the grasses and sedges (Hattersley, 1984
; Dengler et al., 1994
; Soros and Dengler, 1998
), we carried out a discriminate analysis function to determine if defined biochemical subtypes could be distinguished anatomically and, if so, which traits contributed the most to discrimination.
MATERIALS AND METHODS
Plant material and growth conditions
A total of 54 species representing 13 of the 16 eudicot families known to contain C4 taxa were examined (Table 1). The species sampling included 21 C3 and 33 C4 eudicots. Of the C4 species, nine were already known to be NADP-ME, and 11 were known to be NAD-ME. Representative C3 (Phaseolus acutifolius) and C4 NADP-ME (Flaveria trinervia), NAD-ME (Amaranthus edulis), and PEP-CK (Melinis minutiflora) plants were included in the enzymic analyses to provide a clear reference for the comparisons. Taxa from three families known to possess C4 species (Scrophulariaceae, Molluginaceae, Gisekiaceae) were unavailable. Species names were confirmed by identification using appropriate regional floristic works.
|
Extraction and assay of decarboxylating enzymes
For the assays of NADP-ME, NAD-ME, and PEP-CK, crude leaf extracts were prepared using the extraction buffer of Ueno (1992)
. Leaf tissue (0.1 g fresh mass) was harvested from fully illuminated leaves, frozen in liquid N2, and rapidly ground to a fine powder using a pre-chilled mortar and pestle with the aid of acid-washed sand. One mL of ice-cold extraction medium was added to the powder, and grinding was continued at 4°C for about 30 s. The grinding solution contained 50 mM HEPES-KOH (pH 7.5), 10 mM MgCl2, 2.5 mM MnCl2, 5 mM diothiothreitol (DTT), 0.2 mM Na4-EDTA, 0.5% (w/v) BSA, and 2.5% (w/v) insoluble polyvinyl pyrrolidone (PVP). Tissues from Calligonum caput-medusae had very low extractability for each of these enzymes using the extraction buffer described. Instead, the grinding medium used contained 50 mM HEPES-KOH (pH 7.5), 2.5 mM MgCl2, 2.5 mM MnCl2, 5 mM DTT, 0.2 mM Na4-EDTA, 0.2% (w/v) Triton X-100, 0.7% (w/v) BSA, and 25 mg PVP (Ueno, 1998a). After removal of samples for chlorophyll estimation, crude extracts were immediately centrifuged in an Eppendorf microcentrifuge (Centrifuge 5415, Brinkmann Instruments, Westbury, New York, USA) at 10 000 x g for 45 s, and the supernatants were assayed at 30°C for enzymatic activity using a Diode array spectrophotometer set at 340 nm (model 8452A, Hewlett, Packard, Palo Alto, California, USA).
The activities of NADP-ME and NAD-ME were determined by monitoring the formation of NADPH and NADH, respectively. The reaction mixture for assay of NADP-ME contained 50 mM Tris-HCl (pH 8.2), 1 mM Na4-EDTA, 20 mM MgCl2, 0.5 mM NADP+, 5 mM Na-malate, and enzyme extract (Ku et al., 1991). The reaction was initiated by addition of malate. The assay medium for NAD-ME assay contained 25 mM HEPES-KOH (pH 7.2), 5 mM DTT, 0.2 mM Na4-EDTA, 2.5 mM NAD+, 5 mM Na-malate, 8 mM (NH4)2SO4, 75 µM coenzyme A or 25 µM acetyl coenzyme A, 8 mM MnCl2, 25 µM NADH, and enzyme extract (modified from Hatch and Kagawa, 1974
; Hatch et al., 1982
). The reaction was started by addition of MnCl2.
PEP-CK was assayed in the carboxylase direction (Reiskind and Bowes, 1991
; Chen et al., 2002
; Walker et al., 2002
) following NADH depletion in a reaction mixture containing 100 mM HEPES-KOH (pH 7.0), 4% (v/v) 2-mercaptoethanol, 100 mM KCl, 90 mM NaHCO3, 5 mM PEP, 1 mM ADP, 10 µM MnCl2, 4 mM MgCl2, 0.14 mM NADH, 6 units malate dehydrogenase (MDH) and enzyme extract. All reaction rates were recorded within a range where the increase in A340 was linear. Chlorophyll content in the extracts was determined spectrophotometrically in 96% (v/v) ethanol according to Wintermans and De Mots (1965)
.
Leaf anatomy
Small pieces of leaf tissue (approx. 1–2 mm2, five leaves per species) containing only high-order veins were excised from the midportion of laminate leaves. For stems and cylindrical leaves, tissue pieces, 1–2 mm3, were sampled. Cut tissues were fixed in FAA (70% ethanol : glacial acetic acid : formalin [18 : 1 : 1, v/v]) overnight at room temperature. Following a standard dehydration in a graded ethanol–acetone series, samples were infiltrated through acetone–Spurr's epoxy resin mixtures and cured in pure Spurr's resin (Spurr, 1969
). Transverse sections, 2 µm thick, were cut with a glass knife on a Porter Blum MT-2 ultramicrotome (Ivan Sorvall Inc., Norwalk, Connecticut, USA), dried onto poly-l-lysine (100 µg·ml–1, MW 560 000)-coated slides, and stained with 0.5% (w/v) toluidine blue O in 0.1% (w/v) Na2CO3. Sections were observed and viewed with a Reichert-Jung Polyvar microscope (Reichert-Jung, Vienna, Austria), and images were obtained using a Nikon DXM-1200 digital camera and ACT-1 software (Nikon, Tokyo, Japan). Images of 14 species not illustrated are available on request.
Captured images containing two to four veins for laminate leaves and four to seven veins for semicylindrical and cylindrical leaves and stems were used for quantitative analyses (Dengler et al., 1994
). Images were digitized using PCI Image-Pro Plus software (Media Cybernetics, Silver Spring, Maryland, USA). Measured variables included cross-sectional areas (as a proxy for volume) of (1) PCA (or C3 mesophyll [M], including all parenchymatous ground tissues, intercellular space, and substomatal cavities), (2) PCR (or C3 bundle sheath [BS]), (3) intercellular space (described next), and (4) epidermis (sum of upper and lower epidermal layers). Ratios of PCA to PCR (or C3 M to BS) were calculated, and areas were also expressed as percentages of total cross-sectional leaf area.
The area of intercellular space (ICS) was estimated stereometrically using a grid superimposed on a cross section and counting the proportion of dots falling within the spaces and cells (Parkhurst, 1982
). Leaf thickness at the thickest part of each vein sector was determined for laminate leaves but not for cylindrical leaves and stems. PCR tissue perimeter and length of PCR (or C3 BS) outer tangential walls exposed to ICS were also measured using Image Pro Plus software. Ratios of PCR (or C3 BS) surface area and PCR perimeter exposed to ICS to PCR area were also determined. Means of quantitative data for each species are available in supplemental data accompanying online version of this article (Appendix S1).
Vein pattern
Vein density was measured as total vein length per unit leaf surface area (Roth-Nebelsick et al., 2001
). The middle third of each replicate leaf was washed thoroughly with 70% (v/v) ethanol, bleached with 5% (w/v) NaOH, cleared in saturated chloral hydrate, and mounted in the same solution. Slides were examined under bright field and differential image contrast optics on a Reichert-Jung Polyvar microscope. Two representative images containing only higher order veins were taken for each cleared leaf. A total of 10 images per species (two images per leaf) were digitized as described for leaf sections.
Data analysis
Data were analyzed using nested analysis of variance on raw and transformed data sets using Proc GLM in the SAS program (SAS Institute, Cary, North Carolina, USA). The model used was: variable = meanoverall + photosynthetic type + species (type) + error. To meet the assumptions of ANOVA for normality and homoscedasticity of variance, we employed square-root or ln transformations for non-ratio variables, and an arcsine (square root) transformation for ratio and percentage-based variables, when needed. P values with a probability level 
0.05 are reported and regarded as indicators of significant differences among means of the tested groups. ANOVA was used to compare biochemical types (Table 4, Appendix S1, N = 14–32 species per type), but not to compare anatomical types within each biochemical type because n ranged from 1 (constrained by occurrence of types) to 16 species (Table 5).
|
|
). The included variables were leaf thickness, PCA (C3 M) and PCR (C3 BS) tissue areas, PCA to PCR area ratio, proportion of ICS, proportion of PCR (C3 BS) perimeter exposed to ICS, PCR area to volume ratio, proportion of epidermis tissue, and vein density. Multivariate CDA rotates a set of nv variables to maximize differences among ng groups chosen a priori. Specifically, it derives ng – 1 or nv (whichever is smaller) orthogonal canonical roots, which are linear combinations of the chosen variables. The first canonical root represents the combination of variables that describes the greatest amount of discrimination among groups. The second root defines the next largest amount of discrimination and is independent of the first root, and so on. Canonical coefficients define the linear projection of each variable onto each root. Raw canonical coefficients were standardized to a mean of zero and variance of one to simplify comparison between variables.
Both the univariate and multivariate analyses raise the question as to whether discrimination among the three photosynthetic types in eudicots is confounded by evolutionary history. To resolve this, we used a Mantel test (Sokal and Rohlf, 1995
after Mantel, 1967
) to evaluate the independence of two matrices. A correlation coefficient is calculated for the pair of matrices, and a randomization test is used to estimate the significance of the correlation. The first matrix in our analysis was a data matrix of pairwise distances among species, calculated from the first canonical root of the canonical discriminant analysis. This matrix was compared to a matrix of (1) family membership or (2) photosynthetic type group membership (coded as 1 for the same family or group and 0 when different).
RESULTS
Identification of C4 biochemical subtype
In the survey of the activities of the C4 decarboxylating enzymes, most (10 of 13) of the previously uncharacterized C4 eudicot species examined here belonged to the NADP-ME subtype (Table 2). Blepharis ciliaris (Acanthaceae), Trianthema portulacastrum and Cypselea humifusa (Aizoaceae), Aerva persica (Amaranthaceae), Pectis glaucescens (Asteraceae), Polycarpaea longiflora (Caryophyllaceae), Heliotropium polyphyllum (Heliotropiaceae), Boerhavia coccinea and B. domnii (Nyctaginaceae), and Kallstroemia grandiflora (Zygophyllaceae) had high NADP-ME activity, compared with the C3 control species Phaseolus acutifolius (Table 2). These values are of the magnitude expected for NADP-ME C4 species (as compared with the NADP-ME control species, Flaveria trinervia, Table 2). NADP-ME activity for these different species ranged between 455 and 960 µmol·mg chlorophyll–1·h–1, and was 4 to 33 times higher than the corresponding NAD-ME activity. NADP-ME activity ranges from 1.5 to 35 times higher than PEP-CK activity. In four species (T. portulacastrum, C. humifusa, and the two Boerhavia species), PEP-CK activity was substantial, indicating that it acts as a secondary decarboxylase behind NADP-ME in these plants (Table 2).
|
Leaf and photosynthetic stem anatomy
Four of the five basic types of C4 Kranz anatomy (e.g., Carolin et al., 1975
) were present in our surveyed species. In all C4 types, the chlorenchyma is organized into prominent and adjoining concentric layers. An inner layer composed of enlarged, cuboidal, tightly packed and thick-walled PCR cells (equivalent to BS in C3 species) is surrounded by an outer layer of radiate, thin-walled PCA (equivalent to M in C3 species) cells exposed to ICS. PCR cells universally contain numerous large and centripetally arranged chloroplasts, while PCA cells have relatively fewer and smaller chloroplasts, which are evenly distributed along peripheral walls. The four anatomical types are readily discerned on the basis of photosynthetic tissue arrangement with respect to the vascular and other tissues. In the atriplicoid type, the PCR tissue forms a complete (e.g., Fig. 1) or nearly complete (e.g., Fig. 3) sheath around vascular bundles. In the kochioid type, PCR cells are confined to the exterior of peripherally positioned veins and do not form a continuous layer (Fig. 16). Both the salsoloid and the suaedoid types are characterized by having a continuous stratum of PCR tissue at the periphery of leaves and photosynthetic stems. These two types are distinguished by the position of the vascular bundles in relation to the chlorenchyma. In the salsoloid type, minor vascular bundles are located adjacent to the PCR layer, while larger vascular bundles are more deeply embedded in water storage parenchyma (Fig. 17). In the suaedoid type, all vascular bundles are centrally placed in water storage tissue, and none contact PCR tissue directly (Fig. 18).
|
|
], Pectis glaucescens [Fig. 9]). Pectis glaucescens (Helenieae II; Fig. 9) has a large volume of ICS occupying the abaxial portion of the PCA, but the PCR sheath is isolated from the ICS by the close packing of the PCA tissue (Fig. 9). In the Cleomaceae, the Cleome lineage is characterized by atriplicoid anatomy with extensive ICS (Fig. 11); although PCA cells are not conspicuously radial, they still surround the PCR layer, separating it from the ICS. In the Cleomaceae C3 representative, the isolateral M is composed of palisade parenchyma cells, and BS cells lack conspicuous chloroplasts (Fig. 12). In the Caryophyllaceae, the representative species of the C4 Polycarpaea lineage also has atriplicoid anatomy (Fig. 13).
|
; Sage, 2004
) were investigated in this study. Each is distinguished by a distinctive C4 structural type (Table 3). All representative species of the Atripliceae lineage demonstrate atriplicoid anatomy (e.g., Figs. 14, 15). Kochia scoparia of the Camphorosmeae lineage displays the kochioid type (Fig. 16). Both Salsola komarovii (Fig. 17) and Haloxylon persicum (not illustrated, see also Pyankov et al., 1999b
) of the Salsoleae II lineage possess the salsoloid anatomy. Suaeda vermiculata of the Suadeoideae I lineage has suaedoid anatomy (Fig. 18). C3 representatives of the Chenopodiaceae included species having laminate leaves with either bifacial (e.g., Fig. 19) or unifacial (A. littoralis, not illustrated) M and species having succulent semicylindrical leaves (Fig. 21) or photosynthetic stems (Fig. 20) with unifacial M. In the C3 species Suaeda vera (Fig. 21), two layers of chlorenchyma are positioned at the exterior of nonchlorenchymatous parenchyma and the vascular bundles; however, a compact chlorenchymatous layer comparable to the PCR layer of the related C4Suaeda vermiculata is absent (compare Figs. 18, 21).
|
), Boraginaceae (Fig. 22), and Nyctaginaceae (Fig. 23) are characterized by atriplicoid anatomy (Table 6). The C3 representatives of these three families all have laminate leaves with bifacial M (not illustrated, not illustrated, Fig. 24, respectively). In the Polygonaceae, the Calligonum lineage is characterized by salsoloid anatomy (not illustrated; Winter et al., 1977
; Voznesenskaya and Gamaley, 1986
; Muhaidat et al., unpublished results). Representative species of the two identified C4 lineages of the Portulacaceae, Portulaca I and II, have atriplicoid anatomy, despite having succulent leaves that are either laminate (Fig. 25) or cylindrical (Fig. 26). In the Zygophyllaceae, representative species of the Tribulus lineage exhibit atriplicoid anatomy (Fig. 27), while Zygophyllum simplex of the Zygophyllum lineage is distinguished by the kochioid type (Fig. 28). C3 representatives of this family with either laminate leaves (Fig. 29) or succulent cylindrical leaves (Fig. 30) have unifacial type M.
Leaf venation
Despite the diversity in leaf size and shape, minor venation in all the investigated species with laminate leaves is essentially reticulate with the highest order veinlets ending freely in areoles (Figs. 31–45). Variation in vein pattern, density, and regularity of branching occurs among the representative C4 and related C3 species. For instance, in the Aizoaceae, vein density is lower in Trianthema portulacastrum (NADP-ME, Sesuvioideae II lineage (Fig. 31) than in Zaleya pentandra (NAD-ME, Sesuvioideae II lineage, Fig. 32). In the Amaranthaceae, vein density is higher in both C4 representative species, Aerva persica (NADP-ME, Amarantheae II lineage, Fig. 33) and Amaranthus retroflexus (NAD-ME, Amarantheae I lineage, Fig. 34) than in the C3 representative Aerva lanata (Fig. 35) and is highest in the NAD-ME A. retroflexus. In the Asteraceae, vein pattern appears equally dense, but is less branched in the C4 representative Pectis glaucescens (NADP-ME, Helenieae II lineage, Fig. 36) than in the C3 representative Tagetes erecta (Fig. 37). In the Cleomaceae, vein density is comparable in Cleome gynandra (NAD-ME, Cleome lineage, Fig. 38) and in the C3 representative C. trinervia (Fig. 39). In the Chenopodiaceae, vein density is higher in the C4 species Atriplex rosea (Fig. 40) and A. texana (not illustrated, see Appendix S1, see Supplementary Data with online version of article) (both NAD-ME, Atripliceae lineage) than in the C3 representative A. hortensis (Fig. 42), but not greater than in the C3 species Chenopodium botrys (Fig. 43) and C. album (not illustrated, see Appendix S1). Vein density in C4 species Atriplex hymenelytra (Fig. 41), A. polycarpa (not illustrated, see Appendix S1) and A. acanthocarpa (not illustrated, see Appendix S1) (all NAD-ME, Atripliceae lineage) is greater not only than in C3 species, but also than in C4Atriplex rosea (Fig. 40) and A. texana (not illustrated, see Appendix S1). In the Nyctaginaceae, vein density in the C4 representative species Boerhavia domnii (NADP-ME, Boerhavia lineage, Fig. 44) and in the C3 representative Commicarpus africanus (Fig. 45) is similar, although the distinct PCR cells of the C4 species gives the appearance of denser veins.
|
Multivariate canonical discriminant analysis of quantitative anatomy
CDA using nine quantitative variables strongly discriminated between C3 and C4 eudicots, but weakly discriminated between the two C4 eudicot biochemical subtypes (Table 3, Fig. 46). The first canonical root represents the combination of variables that describes the greatest amount of discrimination among groups and has an eigenvalue of 5.7, representing 96.2% of the variation in the data (Table 3). The canonical coefficients that made the greatest contribution to the first canonical root, and therefore to discriminating C4 from C3 eudicots, include PCA (M) to PCR (BS) ratio and % PCR (BS) perimeter exposed to ICS (Table 3). In contrast, the second canonical root has an eigenvalue of 0.23 and accounts for an insignificant 3.8% of the variation in the data (Table 3). Although the clouds of points representing NADP-ME and NAD-ME species overlap considerably, they show a tendency for different scores on the second canonical root, largely due to the contributions of leaf thickness and PCR (C3 BS) SA to V to area ratio (Fig. 46, Table 3).
|
Univariate analysis of quantitative anatomy
Vein density
There is a wide range of vein densities in C3 species, NADP-ME species, and NAD-ME species such that no statistical difference is observed among the three types (Fig. 47A).
|
Intercellular space
ICS volume expressed as a percentage of the total leaf/stem cross sectional area is significantly lower in C4 species than in the C3 species but not significantly different between the two C4 subtypes (Fig. 47F).
PCR surface: volume
The ratio of PCR (BS) external perimeter to tissue area was found to be markedly lower for both C4 subtypes than for C3 species (Fig. 48A). Mean values for the two C4 subtypes do not differ significantly. Similarly, the proportion of total PCR (BS) perimeter that is exposed to ICS is strikingly lower for the C4 subtypes than for C3 species (Fig. 48B). Between the two C4 subtypes, NAD-ME demonstrates a lower mean value than NADP-ME, but the difference is not significant.
|
DISCUSSION
In this study, we investigated leaf anatomy and biochemistry for a group of previously uncharacterized C4 species. In the process, we expanded knowledge of C4 photosynthesis for most of the postulated 32 eudicot lineages in which it is known to occur. Integration of our results with previously published work provides some striking examples of divergence and convergence in the evolution of C4 biochemistry and Kranz anatomy. Next we discuss each of the anatomical variants of Kranz anatomy in relation to biochemical subtype and phylogenetic occurrence and identify the quantitative anatomical variables that differentiate C4 from C3 eudicots. We also discuss the evidence for the presence of multiple decarboxylating pathways in C4 eudicots and the lack of any clear association between biochemical subtypes and leaf anatomy. By contrast, in grasses, leaf anatomy and biochemical subtype have strong characteristic associations (Hattersley, 1984
; Dengler et al., 1994
).
Diversification of leaf anatomy in C4 eudicot lineages
Atriplicoid
Atriplicoid anatomy is the most common and has arisen independently in 21 (of a total of 32) known C4 eudicot lineages, providing a striking example of extensive evolutionary convergence. Our findings showed for the first time that the atriplicoid leaf type is present in Aerva persica (Amarantheae II lineage, Amaranthaceae), Cypselea humifusa (Sesuvioideae I lineage, Aizoaceae), Pectis glaucescens (Helenieae II lineage, Asteraceae), Blepharis ciliaris (Blepharis lineage, Acanthaceae), and Kallstroemia grandiflora (Tribulus lineage, Zygophyllaceae; Table 6). We have also confirmed that this type of Kranz anatomy is present in 12 other lineages for which it has previously been described (Table 6). In addition to materials examined in this study, the atriplicoid Kranz type has also been reported in Salsola cotyledons (Salsoleae I lineage; Chenopodiaceae; Pyankov et al., 2001b
); Froelichia, Alternanthera and Tidestromia (Amarantheae III; Amaranthaceae); Flaveria, Parthenium, Chrysanthellum, and Isostigma (Helenieae I and Heliantheae lineages; Asteraceae); Gisekia (Gisekia lineage; Gisekiaceae); Mollugo (Mollugo lineage; Molluginaceae); Allionia and Okenia (Boerhavia lineage; Nyctaginanceae); and Tribulus (Tribulus lineage; Zygophyllaceae) (see references, Table 6). C4 eudicots lineages with atriplicoid Kranz anatomy are either NADP-ME or NAD-ME biochemical subtypes; of these, 14 lineages are NADP-ME, and six are NAD-ME, while two are unknown (Table 6). The appearance of atriplicoid type anatomy in numerous lineages of C4 eudicots, irrespective of the biochemical type, reflects, in part, the prevalence of laminate leaves in C3 ancestral taxa. Development of atriplicoid anatomy requires enlargement of BS tissue through enhanced cell expansion and/or proliferation and reduction of M tissue through suppression of the same processes. The localized M cell expansion and separation leading to ICS formation is also reduced, specifically adjacent to BS tissue (Dengler et al., 1995
; Dengler and Taylor, 2000
).
Salsoloid
The second most common C4 anatomical type in the eudicots is the salsoloid (Table 6). Outside the Chenopodiaceae (C4 species of Salsola and close relatives, Pyankov et al., 2001b
), this type is found in Asteraceae (Chrysanthellum) (Brown, 1975
; Peter and Katinas, 2003
; based on our criteria for anatomical type delineation) and Polygonaceae (Calligonum) (Winter et al., 1977
; Voznesenskaya and Gamaley, 1986
; Pyankov et al., 1994
, 2000a
). These studies and our own observations confirmed the occurrence of the salsoloid type amongst five distinct eudicot lineages; of these, two lineages are NADP-ME, while two lineages are NAD-ME (Table 6). The biochemical subtype of C4Chrysanthellum (Heliantheae lineage, Asteraceae) species is unknown.
Salsoloid anatomy, regardless of C4 biochemical subtype, has evolved exclusively from C3 ancestors having either thick fleshy leaves or stems as the principal assimilatory organs. Evolution of the salsoloid anatomy requires dramatic alteration in the photosynthetic tissue development because C3 ancestors have extensive mesophyll tissue and lack a distinctive cell layer that is anatomically comparable to the PCR layer in C4 derivatives (Dengler and Nelson, 1999
). For instance, in C3Salicornia californica (Fig. 20), individual veins are surrounded by nonchlorenchymatous BS, and an M tissue of subelongate cells does not form a distinct layer. In contrast, in the C4 relative Salsola komarovii (Fig. 17), ground tissue exterior to veins is reduced to two prominent layers specialized as PCA and PCR layers. Veins adjoin the inner side of the PCR layer and are embedded in enlarged nonchlorenchymatous succulent cells not obviously organized into definite BS layers toward the interior. Thus, evolution of salsoloid anatomy evidently requires reduction in ground tissue at the periphery of the photosynthetic organ, organization into distinct layers, specialization of PCR and PCA cells, and enhancement of water storage tissue (Dengler and Taylor, 2000
).
Kochioid
The kochioid type is the third most common pattern of Kranz anatomy in the eudicots. In addition to its occurrence in Kochia, Bassia, and Camphorosoma (all Camphorosmeae lineage, Chenopodiaceae) (Pyankov et al., 2001b
; Kadereit et al., 2003
), the kochioid type occurs in Zygophyllum simplex (Zygophyllum lineage, Zygophyllaceae; Crookston, 1972
; Voznesenskaya and Gamaley, 1986
; Table 6), an annual C4 halophyte with cylindrical succulent leaves (Khan and Ungar, 1997
). Interestingly, Zygophyllum simplex is biochemically C4 NAD-ME (Table 2), while the C4 chenopods with the kochioid type are specific to the NADP-ME subtype (Pyankov et al., 1999a
). Thus, C4 eudicots from two distantly related families have converged on this unique form of Kranz anatomy, yet have evolved different biochemical paths of C4 photosynthesis. Interestingly, both Kochia scoparia (Pyankov et al., 1999a
) and Zygophyllum simplex (R. Muhaidat, unpublished observations) have kochioid anatomy in the foliage leaves, but atriplicoid anatomy in the cotyledons. Mapping of biochemical and anatomical traits on the phylogeny of the tribe Salsoleae (Chenopodiaceae) indicates that atriplicoid type anatomy is basal, while the kochioid and salsoloid types are derived (Pyankov et al., 2001b
); thus, the laminar cotyledons may retain a basal character, while the foliage leaves display the derived conditions. During the evolution of the kochioid type, developmental events similar to those described for salsoloid anatomy were required for modification of ancestral C3 anatomy, with the distinction that the strictly vein-associated PCR tissue is likely derived from bundle sheath, not other ground tissue (compare C4Zygophyllum simplex [Fig. 28] with C3Z. coccineum [Fig. 30]).
Suaedoid
The suaedoid type of Kranz anatomy (salsina type according to Kadereit et al., 2003
) is restricted to C4 species of the genus Suaeda (Suaedoideae I lineage, Chenopodiaceae, Fig. 18), which is exclusively NAD-ME (Fisher et al., 1997
), and is the least common of the Kranz anatomies in our sample. Similar to the salsoloid and kochioid types, evolution of the suaedoid type requires an altered patterning of photosynthetic tissues, indicating it is evolutionarily complex, compared with the atriplicoid type.
The other anatomical categories among the C4 eudicots and that not examined in our study are: the "Isostigma type" demonstrated by certain species of the genus Isostigma (Heliantheae lineage, Asteraceae) (Peter and Katinas, 2003
), the conospermoid type (Freitag and Stichler, 2000
; the schoberia type according to Kadereit et al., 2003
) found in Suaeda acuminata (Suaedoideae II lineage, Chenopodiaceae), and the two versions of single-celled-C4 photosynthesis exhibited by two distinct lineages (Borszczowia and Bienertia) in the Suaedoideae tribe of the Chenopodiaceae (see Edwards et al., 2004
for a detailed review).
Quantitative anatomical differences between C3 and C4 species
A general discrimination between C3 and C4 species was detected by the multivariate canonical discriminant analysis, which considered the variations caused by the nine quantitative variables (Fig. 46). The analysis showed that C4 species are conspicuously displaced from C3 relatives along the axis of the first canonical root. For instance, C3 representatives of Chenopodiaceae (represented in Fig. 46 by blue inverted triangles) have a distinct distribution from C4 representatives of the same family (red and green inverted triangles). An analogous pattern is demonstrated by Aizoaceae (squares) and Amaranthaceae (diamonds) (Fig. 46). Using the Mantel test, we have also shown that grouping of species by biochemical subtype is significantly stronger than would be expected from chance alone, while grouping by family is not. Using CDA, we found that C4 are isolated from C3 eudicot species primarily by differences in PCA (M) to PCR (BS) ratio and % PCR (BS) perimeter exposed to intercellular airspace (Table 3). Next we discuss the differences in individual anatomical traits, based on the univariate analysis.
Vein density
In grasses, the distances between adjacent parallel veins are consistently shorter in C4 species than in C3 relatives, and their ranges rarely overlap (Kawamitsu et al., 1985
; Ohsugi and Murata, 1986
; Dengler et al., 1994
; Ueno et al., 2006
). Based on a small sample of dicots, Rao and Rajendrudu (1989)
reported significantly shorter interveinal distance in C4 species compared to C3 species. Vein spacing is inversely correlated with vein density such that species with shorter interveinal distances have generally higher vein density (Roth-Nebelsick et al., 2001
). Our analysis of vein density contrasts with the pattern found in C4 grasses, in showing no significant differences between C3 and C4 species (Fig. 47A). In addition, vein density made a very minor contribution in the CDA (Table 3). Some of the interspecific variability in vein density may be correlated with the species life form; for instance, the perennial C4 species of the cosmopolitan genus Atriplex (Atripliceae lineage, Chenopodiaceae), such as A. hymenelytra (Fig. 41), demonstrate denser venation compared to C4 annuals such as A. rosea (Fig. 40). In the genus Flaveria (Helenieae lineage, Asteraceae), vein density is greater in the C3 species (F. robusta) ancestral to the clades containing C4 species than in other, more basal C3 species, suggesting C4 species have evolved from C3 lineages with high vein density already (McKown and Dengler, 2007
, in this issue).
In contrast to grasses, our data indicate that vein spacing is not a reliable criterion for distinguishing the photosynthetic pathway across eudicots as a whole. High vein density is a common feature of xeromorphic leaves, regardless of the photosynthetic type (Esau, 1977
). Enhancement of vein density is favored in dry environments with high solar irradiance and low humidity (Zalenski, 1902
; Roth-Nebelsick et al., 2001
). Most C3 ancestors of the C4 lineages are present in hot, drought-prone regions of the subtropics, tropics, and warm temperate zones. For example, C3 species in the Flaveria, Chaemacyce, Heliotropium, Aizoaceae, Amaranthaceae, and Cleomaceae lineages occur in regions with high potential evapotranspiration (Table 1, Sage, 2004
and references therein). High evapotranspiration may select for reduced vein spacing (shorter IVD) to better supply mesophyll cells with water under high rates of water efflux. This could facilitate a reduction in the diffusion distance between M and BS cells in C3 plants, facilitating the beginning of carbon scavenging mechanisms such as refixation of photorespiratory CO2 (Sage, 2001
, 2004
). Early phases of C4 evolution are thought to involve shuttling of photorespiratory metabolites to the BS, where glycine decarboxylase becomes localized, and the CO2 it releases is re-fixed by BS Rubisco (reviewed in Monson and Rawsthorne, 2000
; Bauwe and Kolukisaoglu, 2003
; Sage, 2004
). Given this scenario, the lack of consistent difference in vein density between related C3 and C4 eudicots may reflect an adaptation to the transpirational environments that predisposes these taxa to evolve C4 photosynthesis.
Additionally, the evolution of dense vein spacing in C3 leaves could serve to repel herbivores and to reduce the risk of mechanical damage by wind. For example, Scheirs et al. (2001)
showed that interveinal distance influences not only herbivore host range, preference, and performance, but also oviposition patterns and/or larval survival. Overall, our findings highlight that evolution of Kranz anatomy in eudicots involves uniform enlargement of BS tissues and a reduction in M tissue (discussed next), but not through a histological alteration in vein spacing.
Photosynthetic tissue volumes
Several earlier comparisons of C4 monocots (reviewed in Dengler and Nelson, 1999
) noted restrictions on the volume of PCA tissue that can be functionally associated with PCR tissue to allow for rapid and coordinated flux of C4 photosynthetic metabolites. Our analysis supports the generality of this conclusion and for the first time demonstrates that the proportion of the PCA tissue is significantly lower in C4 eudicots than in C3 relatives. The reduction in PCA tissue volume is associated partly with a reduction in leaf thickness (discussed next) and/or number of M cell layers (McKown and Dengler, 2007
). Interestingly, PCA tissue area is smaller in species displaying kochioid, suaedoid, or salsoloid Kranz anatomy compared to those with the atriplicoid pattern, regardless of the biochemical subtype (Table 5). The proportion of PCR (or C3 BS) tissue area is significantly higher in C4 eudicots than in C3 relatives. The enhanced PCR tissue volume of C4 species reflects the physiological requirement for accommodating the numerous and large organelles that are involved in the C4 cycle and are required for generating high levels of CO2 around Rubisco (Hattersley, 1984
; Dengler and Nelson, 1999
; Kanai and Edwards, 1999
).
The relative proportions of PCA and PCR tissues have been an important diagnostic indicator of photosynthetic type in monocots (Hattersley, 1984
; Dengler et al., 1994
; Soros and Dengler, 1998
) because they reflect both the requirement for short diffusion distance for C4 metabolites between the two tissue types and the enhanced volume of BS tissue. Despite the diversity of C4 lineages and of Kranz anatomy types in our species sample, the ratios of PCA to PCR were consistently lower in C4 than in C3 eudicots and contributed strongly to observed differences between the types in the CDA (Table 3). Species with suaedoid and salsoloid Kranz types have the lowest ratios due to the greatly reduced PCA tissue area (Table 5, see Appendix S1, see Supplemental Data with online version of this article), while ratios for species with the kochioid type approach those of the atriplicoid Kranz type due to the diminished PCR tissue area (Table 5, see Appendix S1). C4 grasses have lower PCA to PCR tissue volume ratios [mean = 3.9 and 2.0 for NADP-ME and NAD-ME grasses, respectively (Hattersley, 1984
; Dengler et al., 1994
)] than do C4 eudicots (mean = 4.1 and 2.9 for NADP-ME and NAD-ME eudicots, respectively).
PCR tissue surface area
An essential feature of the CO2 concentrating system in C4 plants is the capacity to retain high CO2 partial pressure around Rubisco in PCR cells while reducing CO2 leakage. In C4 grasses, the low surface to volume relationships of PCR cells and reduced exposure of these cells to intercellular space, achieved by both the shape and radial orientation of PCA with respect to the PCR tissues, contribute to reduced leakage (Hattersley and Browning, 1981
; Dengler et al., 1994
). Similarly, we found strikingly lower values for surface area exposure of PCR (BS) to intercellular space in C4 compared to C3 eudicots.
Epidermis tissue
The proportion of epidermis tissue is significantly greater in C4 species on average than in C3 species, confirming the findings of Rao and Rajendrudu (1989)
in an earlier comparative study of 13 C3, C3-C4 intermediates, and C4 eudicots from eight genera. Enhanced epidermal volume may simply maintain leaf thickness required for physical strength (when M tissue is reduced) or provide succulent water storage tissue, but this property also has been suggested to be associated with improving light focusing and amplification within leaves of a variety of plants, especially obligate shade plants (Haberlandt, 1914
; Gabrys-Mizera, 1976
; Bone et al., 1985
; Vogelmann et al., 1996
). While the significance of epidermal volume for C4 plants is still unknown, photosynthesis in C4 plants does become saturated at very high light intensities compared to C3 plants (Long, 1999
; Sage and Pearcy, 2000
), and the additional higher ATP requirement to run the C4 cycle increases the photon requirement in C4 plants, particularly in dim light (Long, 1999
). Furthermore, differential photon requirements of PCR and PCA tissues have to be balanced to ensure metabolic coordination and focusing of incident PPFD by epidermal tissues may meet this requirement. Developing a large volume of epidermis tissue that enhances internal light levels in C4 eudicots may thus be an anatomical adaptation to enhance photon flow to PCR and PCA tissues.
Quantitative anatomy is similar between C4 NADP-ME and NAD-ME subtypes
In the C4 grasses, the three biochemical subtypes (NADP-ME, NAD-ME, PEP-CK) are distinguished by the quantitative anatomical variables measured in this study. For instance, the NAD-ME grasses differ significantly from the NADP-ME by having greater interveinal distance, smaller PCA tissue volume, larger PCR tissue volume, lower PCA to PCR tissue volume ratio, lower PCR surface to volume ratio (Hattersley, 1984
; Dengler et al., 1994
). Similarly, in the sedges, despite striking qualitative differences in organ anatomy, NAD-ME species consistently have greater interveinal distance, larger PCA tissue volume, larger PCR tissue volume, higher PCA to PCR tissue volume ratio and larger ICS volume than NADP-ME species (Soros and Dengler, 1998
). In contrast, no significant differences in these quantitative variables between NADP-ME and NAD-ME biochemical subtypes were found in our univariate analysis, and discrimination between these two types was insignificant in our CDA. Even within a single family, such as the Chenopodiaceae, NADP-ME and NAD-ME species remain clustered in the multivariate space (Fig. 46). Likely contributing factors to the lack of significant discrimination between types are the large number of independent lineages sampled and uneven sampling among the lineages (constrained by availability and accessibility to resources). The pronounced co-occurrence of multiple decarboxylating pathways in C4 eudicots could be a contributing factor in the lack of discrimination between the two C4 photosynthetic subtypes in eudicots (discussed next).
C4 biochemical subtypes in eudicots
The measured range of activities within our species sample (Table 2) generally is comparable to that recorded for other C4 species (Gutierrez et al., 1974
; Hatch et al., 1975
, 1982
). For the newly typed species, our subtype determinations are consistent with the ultrastructural investigations of Carolin et al. (1978)
and Voznesenskaya and Gamaley (1986)
. Carolin et al. (1978)
reported that chloroplasts of PCR cells in Boerhavia diffusa (Nyctaginaceae), Polycarpaea corymbosa (Caryophyllaceae), Trianthema pilosa, and T. compacta (Aizoaceae) have non-appressed thylakoids and rudimentary grana, characteristic features of NADP-ME subtype C4 plants (Edwards and Walker, 1983
). Voznesenskaya and Gamaley (1986)
also reported that PCR chloroplasts of Zygophyllum simplex (Zygophyllaceae) and Calligonum caput-medusae (Polygonaceae) have extensive granal development, consistent with their classification here as the NAD-ME subtype.
Leaves of Kallstroemia grandiflora (Zygophyllaceae), Heliotropium polyphyllum (Heliotropiaceae), and Pectis glaucescens (Asteraceae), typified as NADP-ME C4 plants (Table 2), had substantial NAD-ME activity, whereas activities similar to those of C3 plants characterized all other NADP-ME species examined in this study. Gutierrez et al. (1974)
, Hatch et al. (1975)
, Hatch and Mau (1977)
, and Rathnam (1978)
also measured substantial NAD-ME activities in a number of NADP-ME-type C4 species. Moreover, our enzyme activity data showed that a number of NADP-ME species (Trianthema portulacastrum, Aizoaceae; Boerhavia coccinea and B. domnii, Nyctaginaceae) and two NAD-ME species (Zaleya pentandra, Aizoaceae, and Cleome gynandra, Cleomaceae; Table 2) have significant activities of PEP-CK, often in the range reported for PEP-CK grasses (e.g., Gutierrez et al., 1974
; Hatch et al., 1
