Localization of carbonic anhydrase in legume nodules


Correspondence: CraigAtkins. Fax: +61 89 380 1001; e-mail: catkins@cyllene.uwa.edu.au


Extracts of the central infected zone and the surrounding cortex of nodules from Lupinus angustifolius L., Vigna unguiculata L. (Walp), Pisum sativum L., Phaseolus vulgaris L., Vicia faba L. and Medicago sativa L. contained significant activities of carbonic anhydrase (CA). Immunoassay of extracts using antisera to a putative nodule CA (Msca1) cloned from M. sativa also indicated expression in both tissue types. Quantitative confocal microscopy using laser scanning imaging and a fluorescent CA-specific probe (5-dimethylaminonaphthalene-1-sulfonamide [DNSA]) localized expression to the infected cells in the central zone tissue and a narrow band of 2–3 files of cells in the cortical tissue that corresponded to the inner cortex. In the infected cells, the enzyme activity was distributed evenly in the cytosol, but in the inner cortical cells, it was restricted to the periphery – possibly to the plasma membrane or cell wall. The functions of CA in these two tissues are considered in relation to the carbon metabolism of nodules and the participation of the inner cortex in the regulation of gaseous diffusive resistance.


The plant tissue fraction of nodules from a wide range of legume species shows significant rates of carbonic anhydrase (CA) activity compared to the supporting root (Atkins 1974). The partially purified native CA protein from bean (Phaseolus vulgaris L.) nodules differed from the leaf isozyme in size and in sensitivity to inhibition by azide and sulphonamides. This evidence led to the suggestion that the smaller isozyme in legumes was a nodule-specific form, associated in a functional way with nitrogen fixation (Atkins 1974). More complete knowledge of the C metabolism that supports nitrogen fixation (Rosendahl, Vance & Pedersen 1990), together with a new conceptual basis for understanding the regulation of nitrogenase activity by O2 availability (Hunt & Layzell 1993), offers a new framework in which to assess the significance of enhanced CA expression in nodules.

Recent studies of gene expression in developing alfalfa (Medicago sativa L.) nodules have demonstrated the transcription of a gene that, on the basis of sequence homology, is believed to encode CA (Msca1) (Coba de la Pena et al. 1997). Msca1 expression appears as a significant early event in the establishment and development of the nodule primordium, but subsequently, as nodules mature and become functional, transcription of this gene is restricted to the inner cells of the nodule cortex. The authors speculated that Msca1 expression, as an early nodulin in meristematic cells, is in some way associated with C metabolism, coinciding with an increase in amyloplast development in these cells. In more mature nodules, expression in the cortex was speculated as being related to control of O2 diffusion, through localized malate synthesis and osmoregulation. This latter interpretation was based on their finding that transcription of phosphoenolpyrnvate carboxylase (PEPC) was also localized in the inner cortical cells (Coba de la Pena et al. 1997). Coordinate expression of these two enzymes occurs in mesophyll cells of the C4-leaf where CA catalyses CO2 hydration to supply HCO3 to the carboxylase (Burnell & Hatch 1988). While significant PEPC activity has been recorded for extracts of nodule cortex (Atkins, Rainbird & Pate 1980; Shelp et al. 1983), the majority of activity in nodules is in fact recovered in extracts of the infected cell tissue (Atkins et al. 1980), where the enzyme is believed to participate in the synthesis of the C4-dicarboxylic acids supplied to bacteroids as respiratory substrates (succinate and malate) (Rosendahl et al. 1990). Thus it might be reasonably expected that CA would also be expressed in the infected tissue. However, the data of Coba de la Pena et al. (1997) indicate that compared to the cortex, Msca1 is not transcribed in the infected cell zone of alfalfa nodules. If there are different CA genes expressed in the two types of tissue, perhaps the Msca1 probe does not detect transcripts in the infected cell zone. In any case, it is yet to be shown that the protein encoded by Msca1 results in a functional carbonic anhydrase enzyme in nodule tissues.

This study of the localization of CA to different tissue types in developing legume nodules relies on direct assays of enzyme activity, detection of the protein using anti-bodies raised to a recombinant Msca1 antigen and quantitative confocal microscopy, using a fluorescent sulphonamide that is a binding site-specific probe for functional CA. Nodules from a range of species have been used to include those with determinate as well as indeterminate forms of development.


Plant material

Effectively nodulated cowpea (Vigna unguiculata L. [Walp] cv Vita 3), narrow-leafed lupin (Lupinus angustifolius L. cv Merrit), pea (Pisum sativum L. cv Greenfeast), broad bean (Vicia faba L. cv Aqua Dulce), common bean (Phaseolus vulgaris L. cv Sabo) and alfalfa (Medicago sativa L. cv Hunter River) plants were grown in sand culture with a nutrient solution containing no combined nitrogen under greenhouse conditions (Peoples et al. 1985). Cowpea and lupin were inoculated with Bradyrhizobium strains CB756 and WU425, respectively, at sowing. Pea, broad bean and alfalfa were inoculated with commercially prepared peat suspensions of currently recommended strains of Rhizobium (Bio-Care Technology Ltd., Somersby, NSW, Australia). Common bean was inoculated with Rhizobium phaseoli Strain 3605, obtained from the Rothamsted culture collection.

Nodules were harvested 21–28 d after sowing (DAS) and held on ice (for up to 10 min) before use in either microscopic studies or tissue extraction. For studies of development, lupin or cowpea nodules were harvested from as early as 7 DAS and periodically up to 58 DAS.

Dissection of nodules

Nodules were separated into central infected tissue (termed central zone tissue, CZ) and outer, cortical tissue under a microscope equipped with a cooled stage, and immediately frozen in liquid N2. The ease of dissection and yield of separated tissue types varied between species. The relatively large nodules from cowpea, bean and lupin were readily separated into cortex and CZ, while dissection of the cortex from the indeterminate nodules of pea, and especially alfalfa, was slow and more difficult. The tissues were collected from at least 20 nodules, sampled from four to six plants, and pooled for extraction. The fresh weight of whole nodules used for extraction was in the range 130–220 mg; for the cortex, the range was 54–76 mg; and for the central infected tissue, 17–31 mg.

Extraction and assay of CA activity

The methods for tissue extraction and assay of activity based on a change in pH have been described previously (Atkins 1974). However, the breaking medium was altered to contain 100 mM Tris-SO4 (pH 8·3), 10 mM EGTA, 1 mM PMSF and 10 mM dithiothreitol (added freshly). Three replicate assays from three separate nodule extracts were used and corrected in each case for a boiled extract blank. Activities were expressed as Wilbur-Anderson Units (Rickli et al. 1964) on a tissue fresh weight basis. Protein content was determined by the method of Lowry et al. (1951) following TCA precipitation and resolubilization, using crystalline bovine serum albumin as standard.

Preparation of Msca1 recombinant protein and polyclonal antibodies

The cDNA Msca1 clone of alfalfa (gift from Dr M. Crespi, CNRS, Gif-sur-Yvette, France) was ligated in the 5′ (EcoR1) to 3′ (XhoI) direction in pSK (isolated from a lambda-zap library). The pMsca1 cDNA insert was amplified with a CAN1 primer (GCGATCCCTAGCCATTAACAACTGAA, containing a BamHI site) and the T7 promoter, then digested with BamHI/KpnI to release a ca. 1200 bp insert. The insert was ligated into BamHI/KpnI-digested pQE30 (Qiagen, Clifton Hill, Victoria, Australia) to preserve the reading frame. The pMsca1Q DNA was used to transform competent E. coli JM109 cells. Expression of the recombinant protein was induced by the addition of 1 mM IPTG and the cells were harvested after overnight culture by centrifugation, lysed, and the lysate applied to a Ni-NTA resin column (Qiagen). After washing, the recombinant protein (carrying a terminal hexa-histidine extension that bound to the immobilized Ni) was eluted from the column under denaturing conditions with 8 M urea in 100 mM NaH2PO4 and 10 mM Tris at pH 4·5. Eluted fractions were collected and concentrated.

Antibodies to the recombinant Msca1 protein were raised in New Zealand White rabbits after injection with 0·1 mg protein diluted 1 : 1 with Titre Max adjuvent. A booster of a further 0·05 mg Msca1 protein was injected without adjuvent after 1 month and serum harvested after a further month.

Assay of nitrogenase activity

Detached nodulated segments of root from lupin or cowpea were assayed for rates of acetylene reduction in an open gas exchange system, as described previously (Dakora & Atkins 1990).


The presence of antigenic protein in extracts prepared from dissected tissues for enzyme activity assay was measured using Msca1 antiserum with an enzyme-linked immunosorbent assay (ELISA) technique. The secondary antiserum was goat/anti rabbit IgG coupled to alkaline phosphatase (Sigma, Castle Hill, NSW, Australia), the reaction was initiated in 96-well immunotitre plates with p-nitrophenyl phosphate as substrate, and the rate of colour development recorded at 405 nm in an automated plate reader. The data were expressed as relative optical density (OD) units. Using polyclonal rabbit antisera prepared against purified leghaemoglobin (gift from Dr C.A. Appleby, CSIRO, Canberra, Australia), the degree to which extracts from cortical tissue were contaminated with proteins from the central tissue zone was assessed.

Detection of CA in sectioned material by confocal microscopy

The method used was based on the observation by Chen & Kernohan (1967) that bovine erythrocyte CA formed a highly fluorescent complex with 5-dimethylaminonaphthalene-1-sulfonamide (DNSA). Subsequently, Newman & Raven (1993) successfully used DNSA as a fluorescent probe to localize CA in Ranunculus penicillatus spp. pseudofluitans leaves and the technique described below was a modification of the fluorescence microscopic methods used in that study.

Thin sections (≈0·1 mm) from freshly harvested nodules were cut by hand under a drop of 20 mM K phosphate buffer (pH 7·4) and, after rinsing, mounted in the same buffer on a 180-μm coverslip on the stage of a confocal microscope equipped with a 250 mW UV Coherent Enterprise Argon Ion Laser, a 100 mW Multiline Argon Ion Laser and a MRC-1000 UV confocal laser scanning imaging system. The sample was excited at 346 nm and fluorescence analysed at 460 nm from intact cells beneath the cut surface. After the initial scan of the tissue (at T = 0 min), 2 μL 10 μM DNSA solution (made up in 10% [v/v] dimethyl formamide in 20 mM K phosphate buffer, pH 7·4), was added to the preparation under the microscope (by displacement of liquid under the coverslip) and the scans collected every 2 min for periods up to 30 or 40 min. A second section from the same nodule was similarly mounted and scanned but 2 μL 10% (v/v) dimethyl formamide in K phosphate (pH 7·4) containing no DNSA was added after the initial scan. A third section was also mounted and scanned, but in this case the DNSA solution also contained 100 μM ethoxyzolamide. Ethoxyzolamide is a potent inhibitor of CA activity in plant extracts (including nodule extracts; Atkins 1974), and of the enzyme purified to homogeneity from a number of plant sources (Atkins et al. 1974). It is around 100 times more effective than acetazolamide (Atkins et al. 1974), 5–25 μM inhibiting CA activity by 50% or more.

When evenly cut sections about 0·1 mm thick were used, a linear rate of increase in fluorescence was recorded for at least 30 min after the initial scan (2 min). However, if the sections used were too thick, there was a lag for as much as 5–10 min before the fluorescence increased due to DNSA, but the increase was then linear. Because the distribution of tissues, especially the outer edge of the nodules in sections, was not clear before the onset of fluorescence, the section was also viewed under normal transmission microscopy with white light. In some cases, the weakly fluorescent image of cells in the outer cortex was assigned a false ‘blue’ colour to define the edge of the nodule in the field of view. To produce clearer images, the fluorescence was assigned a false colour range (‘autumn’) that changed from red to bright yellow and almost white as the intensity increased with time. Fluorescence due to DNSA addition was quantified by expressing the yield as the relative intensity collected from the whole field of view (comprising both infected zone and cortical tissue) or that collected from the same area of a representative sample of cells shown for each of the tissue types on the section. Autofluorescence due to the plant material at time zero, i.e. after the initial scan, was subtracted from the yield in each case and the DNSA-dependent fluorescence expressed in relative intensity units.


Activity of CA in different tissue types

Expression of carbonic anhydrase (CA) is not restricted to the infected tissue zone of mature and functional nodules (Table 1). In each of the symbioses studied, significant activity was also recovered in extracts that primarily contained the nodule cortex. The proportional distribution between the two tissues varied: in some cases, the cortex contained a greater activity on a fresh weight basis; in others, the central zone tissue (CZ) was more active. Immunoassay of the extracts from 28-day-old cowpea with anti Lhb serum indicated that those of the CZ contained 27 times the level of Lhb in extracts of cortex (27 ± 2 OD units compared to 1 ± 1 in the ELISA plate assays). The same extracts assayed with Msca1 antisera indicated values for the CZ that were only eight times those of the cortex (33 ± 4 OD units compared to 4 ± 1; data are means ± SE [n = 4]). Thus, while there was some contamination of the extracts of cortex with CA from the CZ tissue, clearly active enzyme is expressed in both tissue types.

Table 1.  Activity of carbonic anhydrase in extracts of whole nodules or cortical and central infected zone tissues dissected from nodules. Data are means ± SE (n = 3)
Legume speciesWhole nodule
(units g−1 fresh weight)
(units g−1 fresh weight)
Central zone
(units g−1 fresh weight)
  • a

    Assays from a single extraction of whole nodules and dissected tissues.

  • b

    Data collected from a single extraction of whole nodules and dissected tissues made and assayed in 1972 at the Seibersdorf Laboratory of the International Atomic Energy Agency, Vienna, Austria.

Vigna unguiculata1266 ± 148301 ± 18666 ± 17
Vicia faba1181 ± 67301 ± 661012 ± 97
Lupinus angustifolus1863 ± 1371765 ± 1681440 ± 72
Pisum sativum1516 ± 1551555 ± 651695 ± 24
Medicago sativa334a272375
Phaseolus vulgaris1702b1165630

Carbonic anhydrase activity was expressed in lupin nodules prior to the onset of N2 fixation and increased more or less coincidently with increasing soluble protein as nodules developed. However, while the protein content of nodules remained fairly constant 21 d after sowing (DAS), CA activity fell steadily and by 56 DAS, it had declined to about a quarter of the maximum (Fig. 1). In lupin, around 80% of CA activity was recovered in extracts of the cortex at 7 DAS; for cowpea, around 60% (Fig. 2). The relatively high proportion in cortex was maintained up to 21 DAS but by 33 DAS, this proportion declined significantly in extracts from both symbioses.

Figure 1.

Activity of carbonic anhydrase (●) and content of total soluble protein (▪) in extracts of lupin nodules during development. The arrow indicates the time when nitrogenase activity was first detected by acetylene reduction assay.

Figure 2.

Proportion of carbonic anhydrase activity in extracts of whole nodules of lupin (▪) and cowpea (●) recovered in extracts of dissected cortex in each case. The arrows indicate the times when nitrogenase activity was first detected by acetylene reduction assay. Data are means ± SE (n = 4).

Localization of CA activity by confocal microscopy

Sections of nodules from all species studied showed marked 5-dimethylaminonaphthalene-1-sulfonamide (DNSA)-dependent fluorescence at 460 nm with excitation at 346 nm. The fluorescence yield from the whole field of view generally increased in a roughly linear fashion with time, but in the absence of DNSA, it increased very slowly (Fig. 3), while concomitant addition of ethoxyzolamide essentially abolished fluorescence due to DNSA (Fig. 3). Restricting the fluorescence yield measurement to a defined group or area of cells increased the relative intensity measured after DNSA addition and reduced the background due to autofluorescence (Fig. 4). The two tissue types (CZ and outer cortex) shown in Fig. 4 represent the two extremes of fluorescence yield in sections of lupin nodules treated with DNSA. In the CZ tissue, the background without DNSA addition was less than 10% of the yield with DNSA added (Fig. 4a), while in the outer cortex there was a very small yield due to DNSA, and background accounted for about 50% (Fig. 4b).

Figure 3.

Time course for the development of relative fluorescence yield at 460 nm from whole fields of view of sections of lupin nodules excited at 346 nm using a confocal microscope equipped with a scanning imaging system. Sections were treated with a fluorescent CA-specific probe (▪), or not (bsl00066), or concomitantly with the probe and the CA inhibitor, 100 μM ethoxyzolamide (○).

Figure 4.

Time course for the development of relative fluorescence yield at 460 nm from equal areas of selected tissues of sections of lupin nodules excited at 346 nm using a confocal microscope equipped with a scanning imaging system. Sections were treated with a fluorescent CA-specific probe (▪), or not (●). (a): The central infected tissue zone. (b): The outer cortex tissue zone. The spatial distribution of the tissue zones is indicated in Figure 6.

More detailed studies of the tissue localization of fluorescence were restricted to nodules of lupin, cowpea and pea. Figure 5 summarizes data from a number of experiments with cowpea in which fluorescence was quantified for equal areas of cells in the outer cortex, 3–4 files of cells at the inner cortex, 2–3 files of cells (designated as the distribution zone) between the inner cortex and the first file of infected cells, and the CZ (shown in Figs 6, 7 and 8). In each case, the background fluorescence was subtracted so the yield was due to DNSA binding. Similar data were collected for the other symbioses and, although the proportion of fluorescence varied between tissue types and with age, in all cases the two major zones fluorescing were the CZ and the inner cortex.

Figure 5.

Time course for the development of relative fluorescence yield at 460 nm from equal areas of selected tissues of sections of cowpea nodules excited at 346 nm using a confocal microscope equipped with a scanning imaging system. Fluorescence due to the tissues without DNSA addition (autofluorescence) were subtracted from the fluorescence yields with the CA-specific probe. bsl00046 = outer cortex; ● = inner cortex; bsl00066 = distributing zone; ▪ = central infected tissue zone (the spatial distribution of the tissue zones is indicated in Figure 7). Data are means ± SE (n = 3).

Figure 6.

Confocal images of the development of fluorescence in the tissues of a section of a 21-day-old lupin nodule exposed to the CA-specific fluorescent probe DNSA and scanned 1 min (b), 20 min (c) and 30 min (d) following exposure. The micrograph in panel (a) is the section viewed by transmission of white light prior to excitation at 346 nm. The fluorescence at 460 nm was recorded in the scans and its relative intensity converted to a false colour scale, where the least intense was dull red and the most intense was bright yellow. In these scans, the edge of the nodule was defined by using a false blue colour. CZ = central infected zone; IC = inner cortex; OC = outer cortex. The size bar indicates 100 μm.

Figure 7.

Confocal images of the development of fluorescence in the tissues of a section of a 28-day-old cowpea nodule exposed to the CA-specific fluorescent probe DNSA and scanned 1 min (b), 10 min (c) and 20 min (d) following exposure. The micrograph in panel (a) is the section viewed by transmission of white light prior to excitation at 346 nm. The fluorescence at 460 nm was recorded in the scans and its relative intensity converted to a false colour scale, where the least intense was dull red and the most intense was bright yellow. CZ = central infected zone; IC = inner cortex; OC = outer cortex; DZ = distributing zone; uc = uninfected cell; ic = infected cell. The size bar indicates 50 μm.

Figure 8.

Confocal images of the development of fluorescence in the tissues of a section of a 21-day-old pea nodule exposed to the CA-specific fluorescent probe DNSA and scanned 0, 12 and 24 min following exposure. In the left hand panels, the section was concomitantly treated with the CA-specific inhibitor ethoxyzolamide (100 μM) together with DNSA at 0 min while in the panels on the right, the inhibitor was omitted. The fluorescence at 460 nm was recorded in the scans and its relative intensity converted to a false colour scale, where the least intense was dull red and the most intense was bright yellow. CZ = central infected zone; IC = inner cortex; OC = outer cortex; uc = uninfected cell; ic = infected cell; v = vacuole. The size bar indicates 50 μm.

Confocal images of the development of DNSA-dependent fluorescence in sections of nodules of similar ages are shown for lupin (Fig. 6), cowpea (Fig. 7) and pea (Fig. 8). In each symbiosis, time-dependent fluorescence due to DNSA developed in a few files of elongated cells in the inner cortex and in the cells of the CZ of the nodules. The first sign of fluorescence appeared in the inner cortex, where binding was localized to the periphery of the cell. This is shown for pea (Fig. 9) at higher magnification. Fluorescence in the CZ cells (Figs 6, 7 and 9) developed over the whole cell, except in pea nodules (Figs 8 and 9), where a clear ‘vacuole’ was seen. In both cowpea and pea nodules (Figs 7 and 8), the enlarged infected cells appeared to be more highly fluorescent than the smaller uninfected cells, while in lupin most cells in the CZ were highly fluorescent (Fig. 6). The time-dependent development of DNSA fluorescence was inhibited almost completely in both the inner cortex and CZ cells of each symbiosis (shown here for pea, Fig. 8) by the addition of ethoxyzolamide.

Figure 9.

Confocal images of fluorescence in the tissues of a section of a 21-day-old pea nodule exposed to the CA-specific fluorescent probe DNSA and scanned 30 min following exposure to 346 nm illumination. The fluorescence at 460 nm was recorded in the scans and its relative intensity converted to a false colour scale, where the least intense was dull red and the most intense was bright yellow. (a), Cells in the inner cortex; (b), infected cells in the central zone of the nodule. CZ = central infected zone; IC = inner cortex; ic = infected cell. The size bars indicate 10 μm.


The symbioses studied represented both determinate (cowpea and common bean) and indeterminate (alfalfa, pea, broad bean and lupin) forms of nodule development, with those included in the latter group ranging from the small and relatively simple forms of the Viceae to the large, collar-type nodules unique to Lupinus (Corby, Polhill & Sprent 1983). In all cases, carbonic anhydrase (CA) activity was recovered in extracts of both the central zone tissue (CZ) and the cortex. Expression occurred in both tissues before the onset of N2 fixation, increasing in activity more or less as the nodules developed and became functional. Like many other enzyme activities associated with the C and N metabolism of nodules (Atkins et al. 1980), CA activities declined after a maximum, corresponding roughly to maximum nitrogenase activity. Coba de la Pena et al. (1997) showed Msca1 expression in the nodule primordium before expression in the peripheral tissue of the developing alfalfa nodule. However, in the present study, it was not possible to determine if CA activity was expressed during the establishment of the primordium, or where it was localized in nodules less than 7 d after sowing (DAS). From 7 DAS onwards, CA was expressed in both the CZ and the cortex, indicating a functional role for the enzyme in both locations. Msca1 antiserum detected antigen in the extracts of both the central zone and the cortex, which, together with the demonstration of functional enzyme in these tissues, lends some support to the idea that the equivalent of the Msca1 gene might be expressed in both locations.

Although there is no direct experimental demonstration for the binding of 5-dimethylaminonaphthalene-1-sulfonamide (DNSA) to purified plant CA, the fact that the CA inhibitor, ethoxyzolamide, abolished DNSA-dependent fluorescence is consistent with a specific interaction. Newman & Raven (1993) presumed that DNSA was a specific probe for leaf CA on the basis of ethoxyzolamide inhibition and, while the leaf and nodule enzymes may differ, an earlier study showed that the CO2 hydration activity of both was highly susceptible to inhibition by sulphonamides (Atkins 1974). In fact, the nodule CA activity in crude extracts was more sensitive to ethoxyzolamide than the leaf enzyme from the same species (Atkins 1974). The localization data based on confocal fluorescent microscopy provide strong supporting evidence for CA expression both in the CZ and the cortex. However, while the CA in the infected cells appears to be more or less evenly distributed in the cell, cytosol expression in the cortex is confined to the inner cortex. The fact that CA was strongly expressed in just two or three files of cells, but was essentially absent from adjacent cells of the outer cortex and those of the distributing zone, suggests an important functional role for the enzyme in these cells. In all symbioses, the fluorescence due to CA activity in the inner cortical cells was peripheral. Cells of the inner cortex are highly vacuolate (Serraj et al. 1998), and although the micrographs might be interpreted to indicate an extracellular CA, there is not sufficient resolution to distinguish between the peripheral cytosol, the plasma membrane or the wall.

There is good genetic and biochemical evidence that C4 dicarboxylic acids (succinate and malate) are the principal substrates used in bacteroid respiration. The high level of PEPC in the cytosol of the central infected zone of nodules (Atkins et al. 1980; Shelp et al. 1983; Gordon 1991), together with 14CO2 labelling data (King et al. 1986), indicates that C4 acids arise as a consequence of sugar fermentation. Just as in mesophyll cells of the C4-leaf (Burnell & Hatch 1988), it is reasonable to suppose that in the infected cells, CA will also be co-ordinately expressed in the cytosol, to ensure an adequate rate of CO2 hydration in a neutral or slightly acidic environment. The fact that DNSA binding occurred throughout the infected cells, though far from definitive, is consistent with a cytosolic location for CA together with PEPC. Despite this apparently obvious role for CA, Raven & Newman (1994) have calculated that the uncatalysed rate of CO2 hydration was adequate for organic acid synthesis in soybean nodules, and they suggest that other possibilities, such as the short-term (transient) buffering of intracellular pH change or ion transport, should also be considered.

The functional significance of CA expressed in the cortical cells – and particularly in the inner cortex – is not so obvious. With the realization that nitrogenase activity in nodules is regulated by the flux of O2, and that this may be varied as a result of reversible changes to diffusive resistance (Hunt & Layzell 1993), there has been renewed interest in the structure of the cortex, and particularly in measurements of the size and distribution of intercellular gas spaces. Studies of cowpea and other determinate nodules have shown that the innermost few cell files (distributing zone) – that is, those immediately adjacent to the infected central zone – have prominent spaces, while the next four to eight files of cells (inner cortex) outside these are more compact, and have much less obvious spaces (Dakora & Atkins 1990; Parsons & Day 1990). The larger cells of the outer cortex also show significant gas spaces. The indeterminate nodules used in the present study were similar, except that the distributing zone was less obvious, and in those of lupin, it was absent or restricted to a single layer of cells. Direct measurements of dissolved O2 using microelectrodes (Witty et al. 1986) indicated a substantial drop in concentration across the inner cortex of soybean nodules, leading to the idea that these cells are the main sites of diffusive resistance. While this may be the case for the few symbioses that have been studied in detail, Brown & Walsh (1994, 1996), in a much wider survey of species, have described considerable variation in cortex structure. They suggest that other features might also be important in regulating gaseous permeability in particular symbioses, and it might be valuable to examine the pattern of CA expression in more diverse types of nodule. The cells of the inner cortex of cowpea and soybean nodules grown with supra- or sub-ambient pO2 show changes in development, especially in the size and distribution of gas spaces, that are consistent with adaptations to further restrict or enhance, respectively, gaseous permeability (Dakora & Atkins 1990, 1991). These morphological changes are more or less fixed and it is not clear if the resistance they impart can be reversibly regulated.

A number of hypotheses have been advanced to link the inner cortex with the reversible changes in diffusion resistance that regulate nitrogenase activity. While there is general acceptance that the mechanism probably involves a variable ‘aqueous barrier’ to the diffusion of gases, especially O2, the underlying metabolic events and processes that cause reversible changes to the water relations of these cells are yet to be described. Whether or not water is simply moved in and out of the gas spaces or reversibly hydrates an extracellular glycoprotein, which is specifically localized to the inner cortical cells (James et al. 1991) remains to be determined. Serraj et al. (1995) have provided some anatomical evidence that in soybean, these cells are osmocontractile and measurements of shifts in anion concentrations (Minchin et al. 1995) support this idea. Antibodies to a tonoplast intrinsic protein (γ-TIP) that has been identified as a putative aquaporin result in abundant labelling of the tonoplast membranes in the inner cortical cells of soybean (Serraj et al. 1998). The authors speculate that this protein is involved in reversible exchange of intercellular water, and regulates diffusion resistance. Taken together, these observations lead to the idea that the underlying mechanisms are closely related to those that occur in stomatal guard cells and that, like stomata, they rely on the synthesis of malate through PEPC. Given this scenario, a role for CA in accelerating the hydration of CO2 to ensure HCO3 supply is possible.

Although PEPC activity has been recorded in extracts of cortex from nodules of a number of species (Atkins et al. 1980; Shelp et al. 1983; Gordon 1991), there is some conflict about the location within this tissue zone. Coba de la Pena et al. (1997) showed that CA and PEPC transcripts were present in the inner cortex, while immunogold localization studies show that the PEPC protein is associated with pericycle transfer cells adjacent to vascular bundles (Robinson et al. 1996). Both of these studies used alfalfa nodules and it is difficult to reconcile their differences.

An alternative role for CA expressed in the inner cortex, which is not related to malate synthesis, can be envisaged. A number of authors (reviewed in Beckett et al. 1988) have drawn attention to the fact that convection, as well as diffusion, may contribute to the aeration of plant tissues faced with a significant aqueous barrier (e.g. submerged organs of water lilies or reeds; Armstrong & Beckett 1987). One possible convective mechanism termed ‘CO2 solubility-driven nonthroughflow convective gas transport’ has been suggested as a means of enhancing O2 flux to the leaves of submerged rice (Raskin & Kende 1983). Beckett et al. (1988) modelled this possibility, but although their mathematical analysis established that such a mechanism requires a substantial diffusive component, and that, of itself, it would not greatly enhance the flux of O2 to submerged rice, it might apply in the case of legume nodules. In simple terms, solubilization of CO2 in the aqueous phase of the nodule would result in a pressure gradient from the outside towards the CZ, creating a partial vacuum inside the diffusion barrier and causing a mass flow of gas (containing O2). Carbon dioxide dissolves to form carbonic acid in equilibrium with HCO3 and CA catalyses this equilibrium. However, if under steady rate conditions of respiration the HCO3 pool in the inner cortex is large, then CA would accelerate the conversion of HCO3 to CO2, to literally ‘exhale’ CO2 on the outside of the aqueous barrier, driving solubilization on the inside. Furthermore, if the peripheral location of CA in this tissue turns out to be extracellular, then the enzyme would exert its influence at the liquid–gas interface. Factors that could conceivably impact on this system to vary the proportion of convective versus diffusive flux would be the surface area of the gas–liquid interface, the level of CA and the pH (as it affects the position of the CO2/HCO3 equilibrium in solution). A unique CA isozyme that is external to the plasmalemma has been described in a range of microalgal cells and inferred in some aquatic macrophytes (reviewed in Badger & Price 1994). However, there is no evidence for a similar CA gene in angiosperms and, in any case, more definitive studies using precise localization tools at higher resolution are needed to establish the cellular location of ‘peripheral’ CA in the nodule inner cortex.

The idea that a negative pressure might be generated in the CZ was indicated previously in simulation models for the gas exchange of nodules (Sheehy et al. 1987; Hunt, Gaito & Layzell 1988). The conclusions were based on the existence of a closed ‘aqueous layer’, within which incomplete replacement of O2 consumption by respired CO2 (or H2 in hup+ symbioses) would lead to a lower than atmospheric pressure. However, for a mass flow of air to occur along this pressure gradient, there must be some open pores that connect the outside atmosphere and the central zone of the nodule. Thus, whether the intercellular spaces in the inner cortex are all ‘closed’ to gaseous flux or there is a mixture of ‘open and closed’ pores is an important consideration. Recent experiments by Witty & Minchin 1994) have addressed the question of open and/or closed pores in two symbioses: white lupin (Lupinus albus) and soybean (Glycine max). In soybean, half of the resistance to O2 influx was gaseous and half was consistent with diffusion through a water-filled barrier (closed gas spaces, water filled spaces or across cells). In lupin, the major diffusive resistance appeared to be due entirely to closed pores. The present study showed that in both of these types of symbiosis (soybean and cowpea are very similar in nodule structure and physiological responses to O2 supply; Dakora & Atkins 1990, 1991), CA expression and localization were very similar. Thus, a role for CA in the inner cortex enhancing convective flux of gas to the CZ would need to accommodate these observations.

A simulation model for nodule gas exchange that takes into account the solubilization of respired CO2 and the activity of CA at the inner cortex has been developed for nodules with a largely closed or a mixed open and closed aqueous barrier to gaseous diffusion (P. P. Thumfort & C. A. Atkins unpublished). The simulation used the parameters for CZ gas exchange and the modelling tools developed earlier (Thumfort et al. 1994). Whether the aqueous barrier contained numerous or few gas-filled pores, the model predicted that a lower pressure was generated in the CZ. The lower pressures were stable across a wide range of respiration rates and respiratory quotient values and occurred whether the symbiosis was hup+ or hup. The driving force for the predicted resultant convective flow of gas into the CZ was the solubilization of CO2 at the aqueous barrier. Full equilibration of HCO3/CO2 by CA across the barrier increased the negative pressure compared to a partial equilibration or the absence of CA at the inner cortex.

It is interesting that Coba de la Pena et al. (1997) reported Msca1 expression in root meristems but not at all in more mature root tissue. This would be consistent with a role for the enzyme in driving CO2-solubility-driven convective gas exchange to ensure a high inward flux of O2 into a densely packed tissue with a high respiratory demand. Perhaps this is a feature of all tissues of this type and a more thorough analysis of CA expression in a wider range of plant tissues seems justified.


The authors are grateful for the skilled technical assistance provided by Mr P. Storer throughout the study, assistance with confocal microscopy by Dr P. Rigby and use of the confocal microscope facility in the Department of Pharmacology, University of Western Australia by Prof. R. Goldie. The gifts of Msca1 clone by Dr M. Crespi, Institut des Sciences Vegetales, CNRS, Gif-sur-Yvette, France and of Lhb antiserum by Dr C. A. Appleby, CSIRO Plant Industry, Australia, are also acknowledged.