Introduction

Sucrose (Suc) occupies a unique position in plant life as it is a major product of photosynthesis with a central role as transport sugar, in growth, development, storage, signal transduction and stress (Levitt 1980; Avigad and Dey 1997; Smeekens 2000). It is also found in cyanobacteria, prokaryotes that perform oxygenic photosynthesis as plants do (Reed et al. 1984; Page-Sharp et al. 1999), and in a few species of purple bacteria (Khmelenina et al. 1999). In prokaryotes, although Suc accumulation is generally associated with environmental stress responses (Reed et al. 1984; Hagemann and Erdmann 1997; Khmelenina et al. 1999; Potts 2000), its function has not been fully elucidated.

Suc metabolism in plants has been extensively studied. Suc is synthesized in the cytoplasm from the sequential actions of sucrose-phosphate synthase (SPS; EC 2.4.1.14) and sucrose-phosphate phosphatase (SPP; EC 3.1.3.24), and it can be reversibly cleaved by sucrose synthase (SuS; EC 2.4.1.13), or irreversibly hydrolyzed by invertases (Inv; EC 3.2.1.26) when there is a high demand for hexoses (Winter and Huber 2000). Inv exist in several isoforms that differ in pH optimum for activity (acidic, neutral and alkaline; Winter and Huber 2000). The specific functions of the different Inv isoforms are not fully elucidated, but they appear to regulate the entry of Suc into different utilization pathways (Sturm 1999). Acidic pH optimum Inv, extensively studied, are cell-wall or vacuolar enzymes and are evolutionarily related to Inv from yeast and bacteria (Sturm and Chrispeels 1990). In contrast, our knowledge about alkaline/neutral Inv (A/N-Inv), which are thought to accumulate in the cytosol (Chen and Black 1992; Van den Ende and Van Laere 1995), is still very limited (Sturm 1999). They have been purified and characterized from soybean hypocotyl (Chen and Black 1992), Vicia faba cotyledons (Ross et al. 1996), Cichorium intybus roots (Van den Ende and Van Laere 1995), carrot (Lee and Sturm 1996) and sugarcane (Vorster and Botha 1998). Recently, a cDNA from Lolium temulentum was characterized as an A/N-Inv by expression in Escherichia coli (Gallagher and Pollock 1998), and a carrot cDNA was isolated by using an internal tryptic peptide sequence of the purified neutral Inv (N-Inv) (Sturm et al. 1999). The c-DNA-deduced amino-acid sequences of A/N-Inv show no similarity to sequences of acid Inv.

Much less is known about Suc metabolism in prokaryotes. SPS and SPP have recently been identified in Anabaena sp. (Porchia and Salerno 1996; Cumino et al. 2001, 2002) and in Synechocystis sp. PCC 6803 (Curatti et al. 1998; Lunn 2002). In contrast, SuS has been reported only in filamentous-nitrogen fixing cyanobacteria (Porchia et al. 1999; Curatti et al. 2000), in which it was shown to be involved in the cleavage of Suc in vivo and in the control of carbon flux into the N2-fixing filament (Curatti et al. 2002). On the other hand, the hydrolysis of Suc by A/N-Inv activity was reported in heterocysts of Anabaena variabilis (Schilling and Ehrnsperger 1985) associated with nitrogen fixation and in Scytonema where it appeared not to respond to salt stress (Page-Sharp et al. 1999).

The present study describes the first isolation and functional characterization of prokaryotic alkaline Inv (A-Inv) and N-Inv genes from Anabaena sp. PCC 7120 (invA and invB). Moreover, Anabaena InvA is the only A-Inv gene reported to date. Searches for the occurrence of these genes in complete genomes showed that A/N-Inv homologues are present in all cyanobacterial strains examined and in plants. Phylogenetic analysis suggests that modern plant A/N-Inv might have originated from an orthologous ancestral gene after the endosymbiotic origin of chloroplasts.

Materials and methods

Culture conditions

Cells of the cyanobacteria Anabaena sp. PCC 7120, A. variabilis PCC 7937, Fischerella sp. PCC 7520, Gloeothece sp. PCC 6909, Nostoc sp. PCC 7107, Nostoc sp. PCC 7413, N. ellisosporum, Oscillatoria sp. PCC 6304, Synechococcus sp. PCC 6301, Synechococcus sp. PCC 7942, Synechococcus sp. PCC 6714 and Synechocystis sp. PCC 6803, were cultivated in BG-11 liquid medium as described by Porchia et al. (1999). Anabaena sp. strain PCC 7120 was also grown in BG-11 (Curatti et al. 2002). E. coli DH5α and BL21(λDE3):pLysS (Novagen) strains were used in this study. E. coli was routinely grown in Luria-Bertani medium supplemented with 50 µg ml-1 carbenicillin.

Isolation and purification of A/N-Inv from Anabaena sp. PCC 7120

Packed cells (6–8 g FW), harvested from Anabaena cultures at exponential-phase, were resuspended, disrupted and centrifuged as described by Porchia and Salerno (1996). The supernatant fraction, referred to as crude extract, was loaded onto a column of DEAE-Sephacel (Amersham–Pharmacia) pre-equilibrated with 20 mM Hepes–NaOH (pH 8.0), containing 1 mM EDTA, 5 mM 2-mercaptoethanol (ME) and 20% (v/v) glycerol. This step was repeated using the same elution buffer at different pH values between 6.0 and 8.0. Inv activities were always detected in the non-bound proteins, which were pooled and immediately applied to a column of CM-cellulose (Bio-Rad) equilibrated with 10 mM potassium phosphate buffer (pH 6.0), containing 1 mM EDTA, 5 mM ME and 10% (v/v) glycerol. Proteins were eluted with a linear potassium phosphate (pH 6.0) gradient from 0.01 to 0.5 M in equilibration buffer. Fractions containing N-Inv activity were pooled, concentrated in an Amicon ultrafiltration cell and further purified by gel filtration through a Sephadex G-100 (Amersham–Pharmacia) column equilibrated with 50 mM Hepes–NaOH (pH 7) containing 100 mM KCl, 5 mM ME and 10% (v/v) glycerol. For relative molecular mass (M r) determinations, a Sephadex G-100 column was calibrated with standard molecular markers: blue dextran (2×103 kDa), alcohol dehydrogenase (150 kDa), bovine albumin (66 kDa) and carbonic anhydrase (29 kDa). The M r values of Anabaena Inv were estimated by chromatographing an aliquot of either the concentrated enzyme from the DEAE-Sephacel step (for the A-Inv enzyme) or from the CM-cellulose step (for the N-Inv enzyme), or of the recombinant enzymes produced in E. coli.

Enzyme assays

The pH optima were determined using 200 mM potassium phosphate buffer over a pH range from 5.0 to 9.0. Inv activity was determined in reaction mixtures containing 100 mM potassium phosphate buffer, 100 mM Suc and an appropriate volume of enzyme in a final volume of 50 µl. The assay was carried out at pH 6.8 to detect N-Inv, or at pH 8.0 in the presence of 20 mM MgCl2 to detect only A-Inv in preparations that contained both A/N-Inv activities. The mixture was incubated at 30 °C for different times. The amount of reducing sugar liberated was measured using Somogyi–Nelson reagent or by coupling hexokinase, phosphoglucose isomerase and glucose-6-phosphate dehydrogenase, and following the appearance of NADPH spectrophotometrically (Porchia et al. 1999). For K m determinations, different Suc concentrations (between 5 and 400 mM) were added to the incubation mixture (Lee and Sturm 1996). Inv substrate specificity was tested with Suc, raffinose, stachyose, melezitose, 1-kestose, trehalose and maltose over a concentration range of about 5–100 mM. Proteins were quantified according to Bradford (1976) or by absorbance at 280 nm.

Amplification, cloning and identification of Anabaena Inv genes

Standard procedures were followed for the isolation of Anabaena genomic DNA (Cai and Wolk 1990) and plasmids (Sambrook and Russell 2001). Two DNA sequences of 1,407 bp and 1,452 bp (named 1407 orf and 1452 orf), homologous to Daucus carota neutral Inv (Dc-N-Inv) and Lolium temulentum N-Inv (Lt-N-Inv) were obtained by BLAST searches against the genome of the Anabaena sp. strain PCC 7120 (http//www.kazusa.or.jp/cyano/). Both DNA fragments were PCR-amplified with the following primer pairs: 5′CCGGATCCATGAAAACCCCTCCAATTAATCAGAAGTCTCTT3′ and 5′CGGTCGACATTAAAGACTACAACCAGGGCCGATAAAA3′; and 5′CGGGATCCATGCAAAAGCTAAACGGACTGCTAACAAATGACAT3′ and 5′CGGTCGACATTAACGAGATATAGAAGCATATACACTGCCAATTTCAAA3′, respectively. The amplification products were ligated between the BamHI and XhoI sites of the expression vector pRSET-A (Invitrogen), and recombinant plasmids were named pR1407 and pR1452, respectively. The resulting constructs were introduced into E. coli strain BL21(λDE3):pLysS (Novargen) to produce recombinant proteins (named His6::InvA and His6::InvB, respectively), which were purified by Co2+ affinity chromatography (TALON resin; Clontech) and concentrated for further characterization studies.

Western immunoblots and immunotitration of Inv activities

Proteins were separated by SDS–PAGE on 12% polyacrylamide gels (Laemmli 1970) and visualized with Coomassie blue or electroblotted onto a nitrocellulose membrane (HyBond C; Amersham) as described by Renart and Sandoval (1984). The membranes were then probed with antibodies raised in rabbits against His6::InvB (anti-An-InvB) according to Salerno et al. (1998). The concentration dependence of Inv activity inhibition by anti-An-InvB was determined by adding different volumes of antibodies to aliquots of recombinant purified proteins as previously described (Salerno et al. 1991). Parallel incubations with pre-immune serum were carried out as a control. After incubation, Inv activity was determined in the supernatant fluids.

Isolation of RNA and RT–PCR assay

RNA from Anabaena was isolated using the TRIZOL reagent (Gibco–BRL). RNA quality and PCR products were analyzed by electrophoresis in 1% agarose gels. For RT–PCR analysis, total RNA (3 μg) treated with DNase (RQ1 Rnase-free Dnase; Promega) was reverse-transcribed using MMLV (Moloney murine leukemia virus) reverse transcriptase (Promega) and specific primers described above. PCR reactions were run on a PTC-100 thermal cycler (model-96 V; MJ Research, Mass., USA) for 20 cycles of 94 °C (1 min), 55 °C (1 min), and 72 °C (1 min), and a single step at 72 °C for 5 min. Under these conditions, PCR amplification occurs in the linear range. A constitutively expressed gene from Anabaena PCC 7120 (rnpB) was used as a loading control as described by Zhu et al. (2001).

Sequence analysis

The sequences reported in this paper have been deposited in the EMBL database (accession numbers AJ311089 and AJ491788 to AJ491793). Other sequences were obtained from the non-redundant protein databases National Center for Biotechnology (http://www.ncbi.nlm.nih.gov), the Kazusa DNA Research Institute, and the Department of Energy Joint Genome Institute (http://www.jgi.doe.gov) by BLAST searches (Altschul et al. 1990) using as query characterized A/N-Inv genes of plants. All orfs scored as A/N-Inv homologues had E-values ranged from 10−150 to 10−78 when compared with genes of established biochemical function (An-InvA, An-InvB). No sequences could be retrieved with E-values higher than 10−78 up to 10−1. Sequences used are from: Arabidopsis thaliana [At-A/N-InvA (AY120777), At-A/N-InvB (AC009323), At-A/N-InvC (AC020580), At-A/N-InvD (AY088388), At-A/N-InvE (AB012244), At-A/N-InvF (AC069273), At-A/N-InvG (AY065247), At-A/N-InvH (AC012393), At-A/N-InvI (NM_117019), At-A/N-InvJ (AL161515) and At-A/N-InvK (AL161586)], Beta vulgaris [Bv-N-Inv (AJ422050)], Daucus carota [Dc-N-Inv (Y16262)], Lolium temulentum [Lt-N-Inv (AP003143)], Oryza sativa (Os-A/N-Inv (AJ003114)], Anabaena sp. PCC 7120 [An-InvA (AJ491788) and An-InvB (AJ311089)], Nostoc punctiforme [Np-InvA (AJ491790) and Np-InvB (AJ491789)], Prochlorococcus marinus MIT9313 [PmMIT-Inv (AJ491792)], Prochlorococcus marinus MED4 [SmMED-Inv (AJ491791)], Synechococcus marinus WH8102 [Sm-Inv (AJ491793)], and Synechocystis sp. PCC 6803 [Sy-Inv (sll0626)]. Sequence alignments were generated with the CLUSTAL X (Thompson et al. 1997) software program (version 1.8), ProDom program, using the BLOSUM 62 matrix and Pfam database (Bateman et al. 1999). Dendrograms were compiled by using the neighbor-joining method (computed from 1,000 independent trials) of CLUSTAL X and the maximum parsimony algorithm of the PHYLIP package (Felsenstein 1993). Searches for sequences homologous to functionally characterized acid Inv from D. carota (CAA77267) and from Saccharomyces cerevisiae (NP_012104), Bacillus subtilis levanase (NP_390581) and E. coli sucrose hydrolase (NP_288933) were performed in all the cyanobacterium genomes sequenced to date.

Results

Presence of A/N-Inv in Anabaena sp. cells

Two An-A/N-Inv activity peaks were shown in Anabaena crude extracts in a pH-dependent activity assay, with maximum pHs in the ranges 6.5–6.9 and 7.6–8.0, respectively (Fig. 1a). This suggests the simultaneous presence of two Inv activities that may be ascribed to N-Inv and A-Inv. The two activities could not be separated when crude extracts were chromatographed on DEAE-Sephacel columns using elution buffers at different pHs. In all cases, A/N-Inv activities eluted with the non-bound protein fractions and showed a pH-dependent activity pattern similar to the crude extract (data not shown). These fractions were pooled and chromatographed through a CM-cellulose column, recovering the Inv activity in a single peak at 0.25 M potassium phosphate buffer. The pH optimum of this Inv activity was found to be in the range 6.4–6.8 (Fig. 1b), and the enzyme was therefore identified as an N-Inv. As no other fractions (either of the non-bound or bound protein) showed Inv activity (data not shown), we assumed that the A-Inv was very labile and susceptible to inactivation. N-Inv was further purified by gel filtration in a Sephadex G-100 column, reaching about 300-fold purification with a 0.1% recovery (data not shown). It should be noted that both Anabaena Inv showed specific activities about one order of magnitude lower than their plant counterparts.

Fig. 1a–d.
figure 1

Effect of pH on Inv activity in Anabaena sp. cells. Enzyme activity was measured in: a Anabaena crude extract, b partially purified N-Inv at the CM-cellulose purification step, c recombinant protein (His6::InvB) and d recombinant protein (His6::InvA)

Full size image

Functional identification of Anabaena A/N-Inv genes

BLAST searches using as query Dc-N-Inv and Lt-N-Inv revealed two homologous orfs (named 1407 orf and 1452 orf) in the Anabaena sp. PCC 7120 genome, 54% identical to the plant N-Inv, that encode proteins of 53,158 Da and 54,720 Da, respectively. Both 1407 orf and 1452 orf were identified as functional Inv genes (named invA and invB, respectively) because extracts of E. coli harboring pR1407 and pR1452 produced recombinant proteins (named His6::InvA and His6::InvB; Fig. 2a) that exhibited A-Inv and N-Inv activity, respectively (Fig. 1c, d). Recombinant proteins and the enzymes partially purified from Anabaena also showed similar apparent Kms for Suc, and inhibition by fructose and Tris–HCl (Table 1). Also, 1 mM p-hydroxymercuribenzoate strongly inhibited both Anabaena Inv (data not shown). The substrate specificities of the authentic Anabaena Inv and of the invA and invB products were determined by incubation with Suc, raffinose, stachyose, melezitose, 1-kestose, trehalose and maltose (all at 100 mM). Enzyme activity was observed only when Suc was used as a substrate. Anabaena A-Inv and N-Inv eluted from gel-filtration columns as polypeptides with Mrs of approximately 51 and 54 kDa, similar values to those of the recombinant proteins (Fig. 2b, c; Table 1). The purified Inv migrated as a single band in a denaturing SDS–PAGE gel with an estimated M r coincidental with the native values (Figs. 2a, 3a; Table 1). Additional immunological studies showed that anti-An-InvB revealed and immunotitrated both the recombinant and authentic A/N-An-Inv (Fig. 3 and data not shown). Remarkably, MgCl2 (between 5 and 40 mM) only inhibited the neutral An-InvB (about 10–75% at pH 6.8 and 100% at pH 8.0). Taken together, these results lead us to conclude that invA and invB correspond to the Anabaena A-Inv and N-Inv genes, respectively.

Fig. 2a–c.
figure 2

Molecular mass determination of Anabaena Inv. a SDS–PAGE (12%) of recombinant proteins stained with Coomassie blue. Lanes: 1 crude extract from non-induced E. coli (pR1452); 2 crude extract from E. coli (pR1452), isopropyl β-d-thiogalactpyranoside (IPTG)-induced; 3 purified His6::InvB; 4 crude extract from non-induced E. coli (pR1407); 5 crude extract from E. coli (pR1407) IPTG-induced; 6 purified His6::InvA. b Sephadex G-100 chromatography of authentic Anabaena N-Inv (filled squares) or recombinant His6::InvB (open squares). c Sephadex G-100 chromatography of authentic Anabaena A-Inv (filled triangles) or recombinant His6::InvA (open triangles). The arrows in b and c indicate the position of standard markers: BD blue dextran (2×103 kDa), AD alcohol dehydrogenase (150 kDa), AB bovine albumin (66 kDa), CA carbonic anhydrase (29 kDa)

Full size image
Table 1. Biochemical characterization of Anabaena invertases
Full size table
Fig. 3a, b.
figure 3

Immunoanalysis of Anabaena Inv. a Immunoblot analysis using anti-An-InvB. Lanes: 1 purified His6::InvB, 2 Anabaena Inv DEAE-Sephacel step, 3 Anabaena N-Inv Sephadex G-100 step. b Neutralization of Inv activity by anti-An-InvB. Purified His6::InvB (filled circles), purified His6::InvA (filled triangles), control incubation of His6::InvB with pre-immune serum (open circles), and control incubation of His6::InvA with pre-immune serum (filled squares)

Full size image

Sequence analysis of Anabaena Inv protein

The Anabaena invA and invB genes encode 468- and 483-amino-acid polypeptides respectively. A comparison of the deduced amino-acid sequences of both genes with Lt-N-Inv and Dc-N-Inv (Gallagher and Pollock 1998; Sturm et al. 1999), the only plant A/N-Inv sequences functionally characterized, showed a high amino-acid identity (54–56%) throughout all the sequences (Fig. 4). A differential feature of plant N-Inv is an amino-terminal extension of about 78 and 178 amino acids for the L. temulentum and D. carota proteins, respectively.

Fig. 4.
figure 4

Multiple sequence alignment of amino-acid sequences of alkaline/neutral invertase (A/N-Inv). Anabaena sp. PCC 7120 Inv (An-InvA, An-InvB), D. carota N-Inv (Dc-N-Inv) and L. temulentum N-Inv (Lt-N-Inv). Conserved motifs among members of the different families are indicated from I to IV [I β-glucosidase hydrolase glycosidase precursor family (–––); II trehalase sequences (----); III glycogen debranching enzyme family (–··–··), and IV sucrose-6-glycosyltransferase family (–·–·–)], which were calculated using the Pfam database and ProDom program. Identical residues are black and indicated with asterisks; conservative amino-acid substitutions are gray-dashed

Full size image

Occurrence of Inv in cyanobacteria, and phylogenetic analysis

We have retrieved only A/N-Inv homologues from all the available sequence databases. BLAST searches (including all the microbial genomes sequenced at the time of writing), using as query An-InvA, An-InvB, Dc-N-Inv and Lt-N-Inv, and multiple sequence alignments, showed that A/N-Inv are restricted to cyanobacterial species and plants. A dendrogram generated by multiple alignment, using the Clustal method, showed a coincidental and unique A/N-Inv in the picoplanktonic open-ocean cyanobacteria (PmMIT-Inv, PmMED-Inv, SmWH-Inv) regardless of the query sequence used (about 33–36% identical to both Anabaena Inv; Fig. 5). In N. punctiforme, another filamentous nitrogen-fixing cyanobacterium, there are also two separate A/N-Inv, named InvA and InvB from their homology to the characterized Anabaena gene products. Cyanobacterial A/N-Inv clearly diverge from the plant homologues (Fig. 5). Qualitatively similar tree topologies were observed when using a maximum parsimony algorithm (not shown). Interestingly, no sequence with similarity to acid Inv, levanase, or sucrose hydrolase could be found in the cyanobacterial genomes sequenced to date.

Fig. 5.
figure 5

Phylogenetic analysis of alkaline/neutral invertase (A/N-Inv). Neighbor-joining phylograms were constructed after sequence alignments of plant and cyanobacterial A/N-Inv sequences with the CLUSTAL X program using a BLOSSUM matrix and a bootstrap trial of 1,000. The graphical representation of the tree was generated using the Treeview16 program. Pm-MIT-Inv was employed as an outgroup

Full size image

The occurrence of A/N-Inv activity was also shown in crude extracts of cyanobacteria from different taxonomic groups (PCC sections I, III–V), either unicellular or filamentous, including nitrogen-fixing and non-fixing strains (Fig. 6).

Fig. 6.
figure 6

Invertase (Inv) activity in cyanobacteria from different taxonomic groups. Crude extracts were prepared from cells of Anabaena sp. PCC 7120, A. variabilis PCC 7937, Fischerella sp. PCC 7520, Gloeothece sp. PCC 6909, Nostoc sp. PCC 7107, Nostoc sp. PCC 7413, N. ellisosporum (Ne), Oscillatoria sp. PCC 6304, Synechococcus sp. PCC 6301, Synechococcus sp. PCC 7942, Synechococcus sp. PCC 6714 and Synechocystis sp. PCC 6803, at exponential phase. Enzyme activity was measured at pH 7.0

Full size image

Effect of the nitrogen source on the expression of Anabaena invA and invB

It has been reported that Inv activity increases in filaments of cyanobacteria grown diazotrophically and also that it is present in nitrate- and ammonium-grown cells (Schilling and Ehrnsperger 1985; Curatti et al. 2002). To determine if there is a differential expression of invA and invB in cells grown in different nitrogen sources, we measured A/N-Inv activities and analyzed the steady-state levels of their mRNAs (Fig. 7). An-InvA and An-Inv-B showed 3- and 6-fold higher activities, respectively, in cells under nitrogen-fixation conditions than in cells cultured with nitrate. Similar expression patterns were found when invA and invB mRNA contents were determined by RT–PCR.

Fig. 7a–c.
figure 7

Expression of invA and invB of Anabaena sp. PCC 7120 grown with different nitrogen sources. a Enzyme activities in crude extracts of Anabaena cells grown under N2-fixing and nitrate-assimilation conditions. Data are the mean ± SE of three independent experiments. b RT–PCR from total RNA extracted from Anabaena cells grown under N2-fixing or nitrate-assimilation conditions. c Amplification of Anabaena rnpB, used as a loading control

Full size image

Discussion

In this study, we report the first biochemical characterization of prokaryotic A/N Inv. The functional identification of two genes (invA and invB) in the Anabaena sp. PCC 7120 genome encoding An-InvA (An-A-Inv) and An-InvB (An-N-Inv) not only completes our knowledge of the enzymes involved in Suc metabolism in modern cyanobacteria but also gives a better insight into the origin and evolution of this central pathway in plant life.

A comparison of a number of biochemical properties of A/N-Inv showed that An-A-Inv and An-N-Inv display many of the characteristics common to the orthologous plant enzymes, regarding the range of pH optimum, substrate specificity, K m for Suc, K i for Tris, product inhibition and polypeptide M r (with the exception of Dc-A-Inv; Table 1). Even though cyanobacteria can be considered alkalophile prokaryotes, it has been shown that they have a limited ability to regulate their internal pH, and there is still no satisfactory understanding of the mechanisms involved in pH homeostasis (Giraldez-Ruiz et al. 1997). Thus, the discrimination between A-Inv and N-Inv is unlikely to be due to different physiological functions of the two isoforms. Also, a single amino-acid substitution can significantly change the pH optimum of Inv, suggesting that the presence of A-Inv and N-Inv in Anabaena cells might be ascribed to such event (Goetz and Roitsch 1999).

Unlike most plant A/N-Inv that were reported to be oligomers (Schaffer 1986; Masuda et al. 1987; Chen and Black 1992; Van den Ende and Van Laere 1995; Lee and Sturm 1996; Ross et al. 1996; Gallagher and Pollok 1998; Vorster and Botha 1998; Sturm et al. 1999), Anabaena native A-Inv and N-Inv are each composed of a single polypeptide chain with an M r of about 53 and 55 kDa (Figs. 2, 3a; Table 1), being the smallest polypeptide with that enzyme activity. In plants, the sugar cane N-Inv is an exception that was shown to be also active as a 60-kDa monomer (Vorster and Botha 1998). This structural difference between plant and cyanobacterial enzymes has also been reported for other Suc-metabolism-related proteins such as SPS and SPP involved in Anabaena and Synechocystis Suc biosynthesis (Porchia and Salerno 1996; Curatti et al. 1998; Cumino et al. 2001, 2002). Whether this feature, in addition to the presence of extra N-terminal (in plant SPS and Inv) or C-terminal (in plant SPP) domains in comparison with the cyanobacterial protein orthologues, might contribute to the higher specific activity of the plant enzymes is still not understood. The fact that An-A/N-Inv activities are highly specific for Suc suggests that they are not β-fructofuranosidases, consistent with reported substrate specificities for plant A/N-Inv, and unlike all known acid Inv (Sturm 1999; Sturm et al. 1999). Both the K m for Suc and the K i for fructose are in the range of the concentrations of these sugars occurring in Anabaena cells (data not shown). Inhibition by fructose and, to a minor extent, by glucose (not shown) suggests that Inv activity could be regulated in vivo by the reaction products, as reported in plants (Masuda et al. 1987; Van den Ende and Van Laere 1995; Lee and Sturm 1996; Ross et al. 1996; Vorster and Botha 1998). The differential inhibition of An-N-Inv by Mg2+ was not reported for plant A/N-Inv, for which the Mg2+ effect does not follow a general pattern (Morell and Copeland 1984; Van den Ende and Van Laere 1995; Lee and Sturm 1996). This result was useful because An-A-Inv activity, like that of some plant A/N-Inv (Sturm 1999), was rapidly lost during the purification from Anabaena cells and, consequently, its activity had to be assayed in enzyme preparations containing both An-A-Inv and An-N-Inv.

Two orfs of 1,407 bp and 1,452 bp present in the Anabaena sp. PCC 7120 genome were identified as the coding regions of Inv genes (invA and invB), as their expression in E. coli produced fully active A-Inv and N-Inv with biochemical and molecular characteristics similar to those of Inv purified from Anabaena cells (Table 1; Figs. 1, 2, 3). The deduced amino-acid sequence of invA is about 60% identical to that of invB, and they share a considerable homology (50–56%) with plant A/N-Inv (Fig. 4). Also, the amino-terminal extension (between 50 and 180 amino acids), which differs markedly in plant N-Inv (Sturm 1999; this paper), is absent in cyanobacterial Inv, indicating that it is not involved in the catalysis. On the other hand, with the exception of Lt-N-Inv, cyanobacterial and plant A/N-Inv sequences (including all the available putative sequences) do not conserve the four amino-acid residues (NDPN) of the pentapeptide NDPNG, a hallmark of the acid β-fructofuranosidases (Sturm and Chrispeels 1990; Sturm 1999), supporting the conclusion that this is not an essential sequence for A/N-Inv activity. In contrast, as in most plant A/N-Inv, the presence of sulfhydryl groups might be essential for the activity of Anabaena Inv because they are strongly inhibited by Hg2+ (Morell and Copeland 1984; Van den Ende and Van Laere 1995; Vorster and Botha 1998; Fig. 4). An-A-Inv and An-N-Inv share five conserved Cys and three conserved Met residues; two of the Cys and two of the Met residues are in similar positions within the polypeptide chains of Dc-N-Inv and Lt-N-Inv. Interestingly, conserved sequence motifs of members of the β-glucosidase hydrolase glycosidase precursor family, of three trehalase sequences, of members of the glycogen-debranching enzyme family, and of members of the sucrose-6-glycosyltransferase family, were found either in An-A-Inv and An-N-Inv or in plant A/N-Inv amino-acid deduced sequences (Fig. 4, motifs I–IV, respectively).

The analysis of the occurrence of A/N-Inv in complete genomes has been useful in providing a more comprehensive view of the origin and evolution of the Suc metabolic pathway. A/N-Inv homologues are present in all cyanobacterial genomes and in plant genomes, but not in non-cyanobacterial prokaryote genomes, examined to date. The presence of An-A-Inv homologues in open-ocean species (Fig. 5), phylogenetically located at the base of the cyanobacterial radiation (Palenik and Haselkorn 1992; Urbach et al. 1992) led us to conclude that A/N-Inv originated from an ancestral A-Inv like gene about 2–3.5 billion years ago, supporting the proposed origin of Suc metabolism based on the appearance of an ancestral SPS-like gene (Cumino et al. 2002). Also, the absence of SuS homologues in open-ocean strains (Cumino et al. 2002) led us to suggest that Suc hydrolysis only by A/N-Inv like proteins might be associated with the origin of Suc metabolism. Further gene duplications during cyanobacterial diversification might have originated two different A/N-Inv in filamentous nitrogen-fixing strains and plant A/N-Inv. In the Arabidopsis genome there are 11 putative A/N-Inv. All plant A/N-Inv cluster together in two sister clades, indicating that they might have originated by recent gene duplications of a common ancestor within the plant lineage (Fig. 5). However, molecular and functional characterization of plant A/N-Inv is only now beginning and conclusions are still limited by the relatively small number of sequences available (no plant A-Inv sequence is known to date).

In contrast to the knowledge of the physiological role of Suc in plants (Winter and Huber 2000), in cyanobacteria its role has not been elucidated yet. It has been suggested that Suc may act as a carbon carrier molecule during nitrogen fixation in filamentous heterocyst-forming cyanobacteria (Wolk et al. 1994), where A/N-Inv activity was shown to be predominantly located in the heterocysts (Schilling and Ehrnsperger 1985). Here, we have shown in Anabaena sp. PCC 7120 cells that both An-A-Inv and An-N-Inv are increased during culture under nitrogen-fixing conditions. The importance of alkaline Inv activity in nitrogen fixation has already been shown in developing soybean nodules where it appears to be the main enzyme of Suc breakdown (Morell and Copeland 1984). But the specific functions of the two Anabaena Inv isoforms await to be elucidated. Additionally, although with a lower expression level, An-A/N-Inv may account for the differences in Suc contents found in Anabaena (Fig. 7; Curatti et al. 2002) and in other filamentous heterocyst-forming or unicellular non-nitrogen-fixing cyanobacteria, grown with nitrate as nitrogen source (Fig. 7). A comparison of K m values for Suc of Anabaena SuS (about 300 mM; Porchia et. al. 1999) and An-A-Inv and An-N-Inv (10–30 mM; Table 1) seems to indicate that in Anabaena cells Suc utilization is preferentially mediated by Inv when there is a high demand for hexoses. Thus, in addition to its role in nitrogen fixation, cyanobacterial A/N-Inv are proposed, as was reported for plant cytosolic Inv (ap Rees 1974), to be "maintenance" enzymes involved in providing Suc degradation substrates when the other Suc hydrolytic activities (acid Inv and SuS in the case of plants, or only SuS in the case of cyanobacteria) are low (Sturm 1999; Curatti et al. 2002). The fact that A/N-Inv activity was measured either in unicellular or filamentous, nitrogen- and non-nitrogen-fixing strains (Fig. 6) also strengthens this conclusion.

In view of the substantial evidence that plant chloroplasts have a cyanobacterial origin and that genes could be transferred from the chloroplast to the nuclear genome (Douglas 1994; Martin and Herrmann 1998), it seems likely plants acquired A/N-Inv genes as part of Suc metabolism during the endosymbiotic origin of the chloroplast. The plastids did not retain the Suc biochemistry of the free-living ancestor and, instead of re-importing the gene products, they gave rise to a novel cytoplasmic pathway in the plant lineage (Cumino et al. 2002). Further comprehensive studies on A/N-Inv in other model organisms will help to support the phylogenetic conclusions as well as to promote understanding of the involvement of A/N-Inv in the regulation of Suc levels in relation to other physiological process.