Cloning, Sequencing, and Regulation of the Glutathione Reductase Gene from the Cyanobacterium Anabaena PCC 7120*

Glutathione reductase (GR) was purified from the cya- nobacterium Anabaena PCC 7120. A 3-kilobase genomic DNA fragment containing the coding sequence for the GR gene ( gor ) was identified and cloned by polymerase chain reaction based on sequences of selected peptides isolated from proteolyzed GR. The coding sequence encompassing 458 amino acid residues, as well as 360 base pairs of the 5 (cid:42) -flanking region and 430 base pairs of the 3 (cid:42) -flanking region, were determined. Genomic Southern analysis indicates that gor is a single-copy gene. A gor antisense RNA probe hybridized with a 1.4-kilobase transcript, suggesting that the gene is not part of an operon including additional genes. The deduced GR amino acid sequence shows 41 to 48% identity with those of human, Escherichia coli, Pseudomonas aeruginosa , pea, and Arabidopsis thaliana GR. The coding sequence of GR was overexpressed in a GR-deficient E. coli strain, SG5, and the recombinant protein was purified. Anabaena GR is NADPH-linked, but a Lys residue replaces an Arg residue involved in NADPH binding in GR from other species. In addition, Anabaena GR carries the G X G XX G “fingerprint” motif which otherwise char- acterizes NAD(H)-dependent enzymes. These differ-ences may contribute to the lack of affinity for 2 (cid:42)

Glutathione reductase (GR), 1 which is a widespread enzyme catalyzing the reduction of GSSG to GSH with NADPH as the reducing cofactor, is necessary for maintaining high GSH/ GSSG ratios in cells (1). GSH plays an important role in many cellular functions, including protection against oxidative stress (2,3). In particular, it is a key enzyme in the glutathioneascorbate cycle, which functions in peroxide scavenging and protection against other oxidative processes (4). A major source of active oxygen species in green, chlorophyllous tissues is derived from the photosynthetic machinery. Green tissues are therefore particularly dependent on efficient scavenging mechanisms, since active oxygen species are not only produced under stress but also under most growth conditions. GR activities in leaves are higher than that in non-photosynthetic tissues and increase with elevated concentrations of oxygen (5). Enzymes of the GSH-ascorbate pathway may also serve an essential protective role in relation to nitrogen fixation, a process catalyzed by the extremely oxygen-sensitive enzyme nitrogenase. For instance, in nitrogen-fixing soybean root nodules, the activities of all enzymes in the GSH-ascorbate pathway are elevated as compared to those in non-fixing nodules; e.g. the GR activity is increased about 4-fold (6). Since diazotrophic cyanobacteria rely on a plant-type oxygenic photosynthesis as well as nitrogen fixation for survival, the risk of oxidative damage is particularly pronounced. However, protective mechanisms operative in cyanobacteria have not been fully elucidated.
GR has been purified from a few cyanobacterial strains (7,8). It shows similar kinetic properties to that of the chloroplast enzyme (9). Furthermore, it has been suggested that in the cyanobacterium Gloeocapsa sp. LB795, GR together with other enzymes of the GSH-ascorbate pathway may serve to protect nitrogenase from being damaged during oxidative stress (10).
The enzyme has been characterized from a large number of sources, e.g. eubacteria, fungi, plants, and human (11). All the GRs isolated show remarkable similarity in molecular and kinetic properties, indicating high evolutionary conservation of the protein. X-ray crystallographic analysis of human GR at 1.54-Å resolution (12) and of Escherichia coli GR at 1.8-Å resolution (13) have been published.
In contrast to the large number of studies available on the enzymology of GR, the gene encoding this key enzyme (gor) has only been isolated from two prokaryotes, E. coli (14) and Pseudomonas aeruginosa (15). GR cDNA has been obtained from two plants, pea (16) and Arabidopsis thaliana (17), as well as from mouse and human cells (18). However, the regulation of gor in response to oxidative stress has been reported only for E. coli and Salmonella typhimurium (19,20), in which OxyR (a transcriptional activator) regulates the overexpression of nine proteins, including GR. However, no evidence that OxyR interacts directly with the gor promoter region has been presented. Here, we report the isolation and characterization of the GR gene from a filamentous nitrogen-fixing cyanobacterium, Anabaena PCC 7120. Furthermore, we present data on the influence of the nitrogen source in the growth medium on the regulation of GR gene expression and propose a potential regulatory mechanism for GR.

EXPERIMENTAL PROCEDURES
Cyanobacterial Culture Conditions-Anabaena sp. strain PCC 7120 was grown photo-autotrophically at 25°C with an illumination of approximately 100 E s Ϫ1 m Ϫ2 in BG11 medium (21), typically containing 20 mM NaNO 3 as the nitrogen source (nitrate-grown). Alternatively, the nitrate was removed from the medium to induce nitrogen-fixing conditions (N 2 -grown), or substituted with 3 mM (NH 4 ) 2 SO 4 (ammonium-grown). All cultures were bubbled with air.
GR Purification and Sequence Analysis of Peptides from Proteolyzed GR-Anabaena PCC 7120 (N 2 -grown) was harvested at late exponential growth phase. The GR protein was purified to homogeneity by a five-step procedure (7). Enzyme activity assays were performed as described (22). The purified protein was reduced with dithiothreitol and alkylated with 4-vinylpyridine, followed by desalting on a Fast Desalt-FIG. 1. Map of restriction endonuclease recognition sites of the gor region of Anabaena PCC 7120 genomic DNA and strategy for isolation, cloning, and sequencing of the gor gene. The arrows at the top show the direction and location of oligonucleotide primers in relation to the gor gene below. In the restriction map, the heavy black line represents the the region coding for GR. The HindIII linker sequence is indicated by (t). Only restriction sites for HindIII, TaqI, XbaI, and RsaI are shown. Genomic DNA and size-fractionated DNA prepared from Anabaena PCC 7120 were used as templates for PCR, as described under "Experimental Procedures," to amplify the gor gene region as three fragments: A (254 bp; degenerate primers P 1 , P 2 and P 3 ), B (532 bp; anchor PCR primer L H and specific primers P 4 and P 5 ), and C (ϳ2.3 kb; specific primers P 6 , P 7 and anchor PCR primer L H ). Some of the primers used had a recognition sequence for restriction enzymes at their 5Ј ends to facilitate subsequent cloning of the PCR-generated fragments into pGEM 3Zf(ϩ) for sequencing. In the sequencing strategy scheme, the arrows show the direction and approximate extent of each sequencing reaction. Arrows originating from vertical bars indicate sequence information obtained from DNA fragments subcloned into pGEM 3Zf(ϩ) vector and primed with the universal or reverse vector primers. Arrows originating from a dot indicate sequence information obtained by the use of oligonucleotide primers complementary to cloned fragment sequences.
Isolation of the GR Genomic DNA Fragments by PCR-Three degenerate oligonucleotide primers P 1 , P 2 , and P 3 (cf. Table I) were synthesized based on the amino acid sequences of the internal peptides AIAEN, FDEDI, and ISGRAT, respectively. Initially, 500 ng of Anabaena PCC 7120 genomic DNA was amplifed for eight cycles at an annealing temperature of 37°C by using primers P 1 and P 2 . The product obtained was then reamplified for 30 cycles at an annealing temperature of 55°C by using primers P 1 and P 3 . The second PCR product of about 250 bp was purifed from an agarose gel and cloned into the vector pGEM 3Zf(ϩ) (Promega). The identity of the second product, denoted fragment A, was determined by sequencing. Based on the nucleotide sequence of fragment A, four specific primers were synthesized (primers P 4 , P 5 , P 6 , and P 7 ; cf. Table I). A HindIII adaptor comprised of two complementary oligonucleotides with HindIII protruding ends (L H , L HЈ ; cf. Table I) was designed and synthesized for attachment to HindIII restriction fragments. These were used to isolate the up and downstream regions of the gor gene. Following Southern blot analysis, a portion of 3-kb HindIII fragments supposed to contain the GR coding region was purified from an agarose gel, ligated with HindIII adaptors, and used as a template for PCR. Both the upstream fragment B and the downstream fragment C (cf. Fig. 1) were obtained by a two-step PCR amplification using the following cycle parameters: 94°C for 1.5 min, 55°C for 2 min, and 72°C for 2 min, the procedure being repeated for 30 cycles. For isolation of fragment B, primer P 4 and adaptor primer L H were used for the first-step PCR, and primers P 5 and L H for amplification of the first-step PCR products. For isolation of fragment C, primers P 6 and L H were used for the first-step PCR, and primers P 7 and L H were used for amplification of first-step PCR products. All PCR amplifications were performed in 50-l reaction volumes containing template DNA (about 500 ng), primers (20 pmol of each), dNTPs (200 M each), and Taq DNA polymerase (2.5 units, Boehringer Mannheim Biochemica) in the buffer supplied with the enzyme.
Preparations of DNA, RNA, and Blotting Analysis-DNA and RNA from Anabaena PCC 7120 were prepared as described (23). For Southern blot analysis 15 g of genomic DNA was used for each restriction digestion. For Northern blot analysis 8 g of RNA was used. The DNA and RNA fragments were separated electrophoretically and transferred to nylon sheets (Hybond N; Amersham Corp.). Both hybridizations were carried out at 58°C for 15 h in 50% formamide, 5 ϫ SSC, 0.5% SDS, 1 ϫ Denhardt, 1 mM EDTA, using 32 P-labeled antisense RNA as a probe (24).
DNA Sequence Analysis-PCR-amplified fragments and their subfragments generated with restriction enzymes were cloned into pGEM 3Zf(ϩ) and sequenced with Sequenase version 2.0 (United States Biochemical Corp.) by the dideoxynucleotide termination method (26). Reactions were primed with the universal primer, the reverse primer, and oligonucleotides complementary to insert sequences, respectively, as described in Fig. 1.
Primer Extension Mapping of the Transcription Start Site of the gor Gene-RNA was isolated from Anabaena PCC 7120 late exponential phase cultures grown under various nitrogen conditions (NO 3 Ϫ , NH 4 ϩ , N 2 ) as described above. In each reaction, the 5Ј end-labeled oligonucleotide primer P 10 (ϩ22 to ϩ39, cf. Table I) was mixed with 30 g of total RNA. The mixture was hybridized and extended (24). Extension products were loaded onto a 6% (w/v) polyacrylamide sequencing gel along with a sequencing ladder generated with the same primer.
Expression of GR in E. coli-Using PCR and appropriate primer pairs (primers P 11 and P 12 , cf. Table I), the coding region of GR was subcloned into the expression vector pGEM-Taq: the Taq promoter was inserted in front of the polylinker region of pGEM 3Zf(ϩ). The recombinant protein was obtained by expression of the construct in a GRdeficient E. coli strain SG5 (data not shown).
Western Blot Analysis-A sample (100 g) of Anabaena PCC 7120 cell lysate was electrophoresed on a 14% (w/v) SDS-polyacrylamide gel and then transferred to 0.45 micron Hybond C Extra supported nitrocellulose (Amersham International, Buckinghamshire, United Kingdom). Nonspecific binding sites were blocked with phosphate-buffered saline, 3% (w/v) bovine serum albumin. An antiserum obtained from rabbits immunized with the recombinant GR protein was used at a 1:200 dilution. Incubation with the primary antibodies was for 4 h at 4°C in the same solution. Following five 5-min washes in phosphatebuffered saline, 0.05% (w/v) Tween 20, the membrane was incubated for 60 min with a 1:1000 dilution of a blotting grade affinity-purified goat anti-rabbit IgG antibody-alkaline phosphatase conjugates. The membrane was then washed six times, 5 min each, in phosphate-buffered saline, 0.1% Tween 20. Antigen-antibody complexes were detected using a chemiluminescence system (24).
Materials-All enzymes and chemicals used were of the highest quality available and were obtained from commercial sources. Oligonucleotides were synthesized by Operon Technologies Inc. (Alameda, CA).

Purification and Sequence Analysis of Peptides of GR from
Anabaena PCC 7120 -Anabaena PCC 7120 cells (30 g wet weight) N 2 -grown, were used for purification of GR. The enzyme was purified 6000-fold to electrophoretic homogeneity by a five-step procedure, involving ammonium sulfate fractionation, chromatography on DEAE-Sepharose CL-6B, Red Sepharose CL-6B, chromatofocusing, and gel filtration (7). The purified enzyme exhibited a specific activity of 213 units/mg, and its homogeneity was checked by electrophoresis under denaturing conditions. Fig. 2 shows a single protein band with an apparent subunit molecular mass of about 50 kDa after the last purification step. Unexpectedly, the enzyme failed to bind to 2Ј,5Ј-ADP Sepharose, commonly used for affinity chromatography purification of the enzyme from other organisms, such as human erythocytes (27), E. coli (18), and pea (28).
About 50 g of the purified GR protein was digested with a lysine-specific protease, and the resulting peptides were separated and isolated by reversed phase high performance liquid chromatography. Several of the peptides obtained in homogeneous form were subjected to sequence analysis. Based on sequence similarities to GRs from other organisms, the relative positions of three internal peptides was determined. These peptides were suitable for the design of degenerate primers used in the isolation of the gor gene.
Construction of a Nucleotide Probe for the gor Gene and Southern Blot Analysis-Three degenerate oligonucleotide primers synthesized on the basis of the amino acid sequences of three internal peptides are shown in Fig. 1 and Table I. Several different DNA fragments were produced by PCR amplification using primers P 1 and P 2 . Therefore, primer P 3 was used inter-nally for reamplification by nested PCR. By using primers P 1 and P 3 , a major PCR fragment of 254 bp (fragment A) was isolated. The deduced amino acid sequence was highly similar to other known GR sequences. Hence, an RNA probe generated from fragment A was used to analyze Anabaena PCC 7120 genomic DNA. The probe hybridized to a single band in all restriction digests (Fig. 3), indicating that the gor gene exists as a single copy gene in the Anabaena PCC 7120 genome. A similar result was obtained from low-stringency hybridization. The weak band (ϳ6 kb) in the high M r region seen using a DraIϩHindIII digestion (Fig. 3) may be due to the incomplete digestion by DraI. The nucleotide sequence of fragment A also provided the information needed for synthesis of the specific primers used to isolate the entire gor gene.
Isolation and Sequence Analysis of the gor Gene-Southern blot analysis of Anabaena PCC 7120 genomic DNA showed that the gor gene was contained within a 3-kb HindIII fragment. HindIII adapters were added to size-fractionated 3-kb HindIII fragments. Hence, fragments B and C were isolated following the PCR amplification procedures described under "Experimental Procedures." The PCR fragments and subfragments generated from the restriction enzyme digestion were cloned and sequenced. pUC/M13 forward primer (5Ј-CGC- FIG. 4. Nucleotide and deduced amino acid sequence of the Anabaena PCC 7120 gor gene. The deduced amino acid sequence in single-letter code is shown below the nucleotide sequence; the amino acid sequences determined directly by analysis of proteolytic peptides are underlined. A potential ribosome-binding site and three putative Ϫ10 and Ϫ35 sites are indicated by underlining of the nucleotide sequence. A potential binding site for BifA/NtcA, TGT(N 9or10 )ACA, is boxed, and its highly conserved sequences TGT and ACA are marked by asterisks ( * ) above. The vertical arrows indicate transcription start sites. The potential transcription terminators (two inverted repeat structures) are indicated by facing half-arrows beneath. CAGGGTTTTCCCAGTCACGAC-3Ј) and reverse primer (5Ј-AACAGCTATGAC-3Ј) as well as additional synthesized oligonucleotide primers (Table I) were used for sequencing. A detailed restriction endonuclease site map and the sequencing strategy used are depicted in Fig. 1. Sequencing of DNA in this region revealed an open reading frame of 458 codons with a putative ribosome-binding site 4 bp upstream of the ATG start codon (Fig. 4). The codon usage is typical of Anabaena PCC 7120 coding sequences (29), e.g. codons which are low in GϩC content in the third position are favored (Table II). In addition, 360 bp of the 5Ј-and 430 bp of the 3Ј-flanking sequences were determined.
Three putative canonical E. coli-type promoters were found within 100 nucleotides upstream of the translation start codon (Fig. 5A). At least two of them can be used for transcriptional initiation as demonstrated below. Furthermore, a putative BifA/NtcA binding site with the consensus sequence motif TGT(N 9or10 )ACA (30), was found upstream of the proximal promoter and overlapping with the middle promoter. The two DNA binding factors BifA and NtcA have been identified in Anabaena PCC 7120 and in the unicellular strain Synechococcus PCC 7942, respectively (31,32). Both belong to the cyclic AMP receptor protein family of prokaryotic regulatory proteins (33). NtcA apparently regulates nitrogen assimilation, while the function of BifA is not clear. The binding site sequence noted in front of the gor gene, TGTTGACAACTGACA (Ϫ70 to Ϫ56), is comparable to sequences identified as BifA binding sequences upstream of the glnA, xisA, and rbcL genes in Anabaena PCC 7120 (Fig. 5B) (30). It shares particularly high sequence similarity with the proximal BifA-binding site up- Both nucleotide and amino acid sequences of GR from Anabaena PCC 7120 showed high similarity to GR from other sources (Fig. 6, Table III). As expected, the GR family signature at amino acid residues 55-67, which is responsible for forming the redox-active disulfide bridge between Cys 58 and Cys 63 (numbering of human GR, Ref. 12, omitting Met 1 ) is highly conserved. Two arginine residues (Arg 218 and Arg 224 ) required for binding of the 2Ј-phosphate group of NADPH are also conserved in all five proteins, the only exception being in Anabaena GR, in which Arg 224 is replaced by lysine. This replacement may be contributing to the lack of binding to the 2Ј,5Ј-ADP Sepharose 4B affinity chromatography matrix noted in the attempts to purify both native and recombinant Anabaena GR.
Origins of gor Transcription-The existence of three putative E. coli-type promoters and one putative BifA/NtcA-binding site upstream of the coding sequence indicated a potential differential regulatory mechanism at the transcriptional level. In order to determine whether gor transcription was initiated from more than one promoter, the 5Ј initiation sites of the gor transcripts were mapped by primer extension of RNA obtained from Anabaena PCC 7120 grown on different nitrogen sources. Two sizes of RNA were observed after primer extension by use of the synthetic oligonucleotide primer P 10 , covering positions  ϩ22 to ϩ39. A shorter transcript (RNA I ) started at Ϫ23 nucleotides upstream of the translation initiation site and apparently derived from the proximal promoter. This transcript was detected under all three culture conditions used. A longer tran-script (RNA II ) originated from the middle promoter at Ϫ36. However, this latter transcript was observed only in ammonium grown cultures (Fig. 7). The extension product of RNA II was about 4-fold less abundant than that of RNA I , indicating FIG. 6. Amino acid sequence alignment of the Anabaena PCC 7120 GR with sequences of GR from other species. Residues that are identical or similar in all six sequences are marked by open squares beneath. Gaps have been introduced to give better alignments (indicated by dots). Double dots below the sequences indicate regions of residues important for GSSG binding. The region surrounding the redox-active disulfide bridge is boxed. The fingerprint motif of NADPH binding is doubly underlined. The Arg residues involved in NADPH binding are indicated by *, but the second of these Arg residues is replaced by Lys in Anabaena GR. Numbers of Anabaena and human GR residues are given on the right side of the sequences. Abbreviations and sources of sequences: A. thaliana, Arabidopsis thaliana (17); pea, Pisum sativum L. (16); Ps. aeruginosa, Pseudomonas aeruginosa (15); E. coli, Escherichia coli (14); human (18). that the promoter for RNA II is used to a smaller extent.
RNA preparations from nitrate-grown and from N 2 -grown cultures were also used for Northern blot analysis. In both cases, a single RNA species of 1.4 kb corresponding roughly to the coding length required for the gor gene (Fig. 8) was detected. This result together with the sequencing data suggest that the gor transcripts are not linked to transcripts of other genes and that the gene is not part of an operon.
GR Expression under Various Nitrogen Regimes-The GR expression construct was transfected into the GR-deficient E. coli strain SG5, and clones with high specific GR activity were selected. The recombinant protein synthesized in E. coli was purified and used for immunization of rabbits. Western blot analysis (Fig. 9) showed that the rabbit-anti GR antibody specifically recognized a polypeptide of about 50 kDa in crude extracts of Anabaena PCC 7120 and that the GR protein level was higher in ammonium-grown cultures than in cultures grown on the other nitrogen sources (NO 3 Ϫ , N 2 ). This result is consistent with our observation that cellular GR activity in-creases when Anabaena PCC 7120 is transferred from nitrate to ammonium medium (data not shown). DISCUSSION We here present the first complete DNA sequence determined for a gene encoding glutathione reductase from a photosynthetic prokaryote, the cyanobacterium Anabaena PCC 7120. The enzyme is of pivotal importance in the scavenging of reactive oxygen species. The significance of the cyanobacterial enzyme is particularly obvious, since cyanobacteria suffer electron leakage and consequent oxygen radical production not only during photosynthesis but also during nitrogen fixation.
Having obtained the nucleotide sequence, the amino acid sequence of the Anabaena enzyme could also be deduced. Both at the amino acid level and at nucleic acid level, GR of Anabaena PCC 7120 exhibits higher similarity to GR from plants than to those from bacteria (Table III). In addition, codons which are low in GϩC content are preferentially used (Table II). This is in contrast to the situation in Pseudomonas and E. coli, where a marked bias in favor of either G or C residues at the third position of each codon has been noted (34). These results support the endosymbiont theory, stating that cyanobacteria were forerunners of higher plant chloroplasts (35,36).
Catalytically important residues responsible for the redox  7. Origins of gor transcription. Lanes 1-3 represent the reverse transcriptase products of GR mRNA of Anabaena PCC 7120 cells grown on ammonium or nitrate or cultured under nitrogen fixing conditions, respectively. Lanes C, A, G, and T represent the results of sequence reactions in the region encompassing the promoter. The transcriptional start sites are indicated by arrows. Ϫ ) or ammonium (NH 4 ϩ ), i.e. non-nitrogenfixing conditions; were electrophoresed on a SDS-polyacrylamide gel. Proteins were blotted onto nitrocellulose, and GR was detected as an immunocomplex with anti-GR antibodies and visualized by chemiluminescence. reaction or involved in the binding of GSSG and NADPH are highly conserved in all six GR sequences examined. These include the redox active cysteines (Cys 58 , Cys 63 ; Fig. 6: numbering based on the human sequence) and flanking amino acids. Also, Arg 218 and Arg 224 of human GR, involved in NADPH binding, are conserved in five out of six GRs, the exception being Anabaena GR, in which Lys replaces the second Arg 224 . Computer modeling indicates that such a replacement gives rise to a longer distance between NADPH and the NADPH-binding residues of the enzyme. Thus, the poor affinity of GR from Anabaena for 2Ј,5Ј-ADP Sepharose may partly be caused by this replacement. Furthermore, most redox enzymes using NADP(H) contain a highly conserved GXGXXA "fingerprint" motif in the NADP(H)-binding domain. However, in NAD(H)-dependent enzymes the alanine residue is almost universally replaced by glycine (37). Noticeably, the Anabaena GR carries the GXGXXG sequence motif similar to that in NAD(H)-dependent enzymes. However, kinetic studies revealed that the enzyme still has about a 40-fold higher catalytic efficiency with NADPH than with NADH (data not shown). Therefore, our result suggests that the difference Ala/Gly 179 in the fingerprint motif is not the sole determinant of the GR coenzyme specificity (cf. 38).
Although the GR sequence is highly conserved in all six species examined, the regulation of the gene expression may be different. For instance, GR isoenzymes seem to be products of a multigene family in soybean root nodule (39) and in red spruce (40). In contrast, the multiple forms of GR identified in other photosynthetic organisms such as pea (16,41) and Arabidopsis thaliana (17) indicate the existence of post-translational regulatory mechanisms, as only a single-copy gor gene has been detected. The Anabaena PCC 7120 gor gene examined here, as well as those from E. coli and P. aeroginosa, are also likely to be single-copy genes.
Comparisons of the promoter regions of gor from E. coli and P. aeroginosa reveals one Ϫ35 and one Ϫ10 element upstream of the E. coli GR coding sequence and one element similar to the E. coli Ϫ10 consensus region upstream of P. aeroginosa gor. In contrast, three E. coli-type promoters were detected upstream the Anabaena PCC 7120 gor gene. Moreover, we could demonstrate that two of the promoters, the middle and the proximal promoters, can be used alternatively or in combination during growth, depending on the nitrogen source used. The proximal promoter is used under all growth conditions, while in the ammonium-grown culture, the middle promoter is also used. Therefore, the high GR expression seen in ammoniumgrown cultures probably reflects the dual transcriptional initiation in cultures using ammonium as the sole nitrogen source.
The putative BifA/NtcA-binding site detected upstream of the proximal promoter is partly overlapping with the middle promoter. Previous studies have shown that BifA may bind to upstream sequences of genes which have diverse functions in Anabaena PCC 7120 (31,32). Depending on the position of its binding site with respect to different promoters, it has been proposed that BifA may act as an activator or a repressor. Its regulatory role may be related to nitrate assimilation, as well as to other unknown functions. In Anabaena PCC 7120, the location of the binding position suggests that BifA probably is capable of repressing gor gene expression from the middle promoter, i.e. when grown on NO 3 Ϫ and N 2 , since BifA binding would interfere with the binding of RNA polymerase. However, since the function of BifA is not completely known, it is difficult to predict how and why such a binding factor would be involved in transcriptional regulation of a gene whose main function is to produce an enzyme contributing to a system involved in scavenging reactive oxygen species. In general, the mecha-nisms by which multiple transcription start sites of gene promoters are controlled remain poorly understood. It cannot be excluded that other factors are responsible for the differential transcription noted, e.g. high concentrations of ammonium may be toxic to cyanobacteria (42). Therefore, the up-regulation of GR through multiple transcription start sites may be a response to stress rather than to the nitrogen status. In order to clarify this point, the regulation of GR under various external conditions will be examined in nitrogen-fixing as well as nonnitrogen-fixing cyanobacteria.