The gene hetN encodes a putative oxidoreductase that is known to suppress heterocyst differentiation when present on a multicopy plasmid in Anabaena sp. PCC 7120. To mimic the hetN null phenotype and to examine where HetN acts in the regulatory cascade that controls heterocyst differentiation, we replaced the native chromosomal hetN promoter with the copper-inducible petE promoter. In the presence of copper, heterocyst formation was suppressed in undifferentiated filaments. When hetN expression was turned off by transferring cells to media lacking copper, the filaments initially displayed the wild-type pattern of single heterocysts but, 48 h after the induction of heterocyst formation, a pattern of multiple contiguous heterocysts predominated. Suppression of heterocyst formation by HetN appears to occur both upstream and downstream of the positive regulator HetR: overexpression of hetN in undifferentiated filaments prevents the wild-type pattern of hetR expression as well as the multiheterocyst phenotype normally observed when hetR is expressed from an inducible promoter. Green fluorescent protein fusions show that the expression of hetN in wild-type filaments normally occurs primarily in heterocysts. We propose that HetN is normally involved in the maintenance of heterocyst spacing after the initial heterocyst pattern has been established, but ectopic expression of hetN can also block the initial establishment of the pattern.
Anabaena 7120 is a filamentous cyanobacterium that is capable of both photosynthesis and dinitrogen fixation under aerobic conditions. Filaments of 100 cells or more grow as undifferentiated chains of vegetative cells in the presence of a source of combined nitrogen, such as nitrate or ammonia. In contrast, when Anabaena 7120 is deprived of combined nitrogen, approximately every tenth cell of the filament develops into a thick-walled heterocyst that is capable of fixing atmospheric dinitrogen ≈ 24 h after the onset of nitrogen starvation. The differentiated heterocysts are evenly spaced along the filament to form a semi-regular pattern. As the filament grows and the number of vegetative cells between heterocysts increases, single vegetative cells midway between two fixing cells develop into heterocysts, thus preserving the pattern of differentiated cells.
Heterocysts differ from vegetative cells in several ways. As the site of dinitrogen fixation, they protect nitrogenase by providing a barrier to O2 penetration from outside the cell and by spatially separating the O2-labile enzyme complex from oxygen produced by vegetative cells in the course of photosynthesis. An envelope composed of a laminated glycolipid layer covered by an external layer of polysaccharide surrounds the heterocyst cell wall and acts as a barrier to O2 penetration; oxidative enzymes are thought to scavenge oxygen that may enter the heterocyst; and the components of photosystem II are degraded to prevent oxygen production, while ATP continues to be generated by cyclic photophosphorylation around photosystem I (Haselkorn and Buikema, 1992; Wolk, 2000). It has been estimated that 600–1000 genes are specifically expressed in recently differentiated heterocysts (Lynn et al., 1986).
The gene hetR is essential for heterocyst differentiation and appears to control it: (i) an increase in hetR transcription can be detected 30 min after nitrogen deprivation, making it the first known heterocyst-specific gene to be expressed (Buikema and Haselkorn, 2001); (ii) a HetR mutant fails to show any signs of heterocyst differentiation; (iii) extra copies of hetR on a plasmid result in inappropriate heterocyst formation in filaments grown on nitrogen-containing media, and multiple contiguous heterocysts (Mch) differentiate on nitrogen-deficient media (Buikema and Haselkorn, 1991); and (iv) hetR is autoregulatory (Black et al., 1993). The sequence of HetR shows no similarities to proteins in GenBank and contains no identifiable structural motifs. An essential serine residue at position 179 led to speculation that it is an unusual serine protease (Buikema and Haselkorn, 1991) and, recently, it has been shown that recombinant HetR purified from Escherichia coli is, indeed, autodegradative (Zhou et al., 1998a). HetR in Anabaena filaments differs from recombinant HetR in two significant ways: HetR from filaments is present only as the full-length 33 kDa peptide, and the pI is about 2.8 pH units lower than that of recombinant HetR when cells are grown without combined nitrogen (Zhou et al., 1998b). HetR is apparently covalently modified in response to growth conditions.
In contrast to hetR, two other genes, patS and hetN, have been identified that suppress heterocyst differentiation when present on a multicopy plasmid. A patS null mutant exhibits a Mch phenotype on nitrogen-deficient media and inappropriate differentiation in the presence of nitrogen (Yoon and Golden, 1998). The gene encodes a 17-amino-acid peptide; the exogenous addition of its C-terminal pentapeptide to a culture of Anabaena prevents the induction of heterocysts. As in the case of hetR, expression of patS is induced soon after nitrogen deprivation but, after differentiation is complete, expression of patS returns to preinduction levels. It has been proposed that the PatS peptide diffuses away from differentiating proheterocysts through a contiguous periplasmic space to create a gradient of inhibitory signal that governs pattern formation in Anabaena 7120 (Yoon and Golden, 1998).
A second gene whose product can suppress heterocyst differentiation, hetN, was discovered independently by Black and Wolk (1994) and Bauer et al. (1995). The gene is in the same region as several operons containing fatty acid elongases required to make the glycolipids specific to the heterocyst envelope (C. C. Bauer, unpublished), and the original insertion mutant used to isolate hetN, N10, does not make heterocyst glycolipids (Ernst et al., 1992). The product of hetN is a putative ketoacyl reductase. Northern analysis revealed that hetN is not expressed until 12 h after induction by nitrogen step-down, at which time the cells that will become heterocysts have already differentiated irreversibly (Bauer et al., 1995). These results pose the following dilemma: how could a putative oxidoreductase control heterocyst formation when it is not made until after the initial pattern of heterocysts has already been determined? When Black and Wolk (1994) recreated insertion mutants of hetN, they recovered filaments that displayed one of three phenotypes: wild type, Het− or Mch. They speculated that second-site mutations or cis effects of the insertions on the expression of neighbouring genes may have been responsible for these multiple phenotypes. The phenotype resulting from a hetN mutation alone has remained unknown.
To avoid the complication of second-site mutations that may mask a lethal phenotype, we replaced the hetN promoter in the chromosome with the copper-inducible petE promoter. This allowed us to grow the cells with copper so that hetN was expressed and subsequently to remove copper from the medium to shut off hetN expression, mimicking the hetN null phenotype. It also allowed us to examine where hetN acts in the differentiation regulatory cascade by expressing hetN in vegetative cells at a time earlier in the differentiation process than is normal.
Overexpression of hetN suppresses heterocyst formation
The normal promoter region of the hetN gene was replaced with the copper-inducible petE promoter (PpetE) to allow regulation of hetN by controlling the level of copper in the medium. In wild-type Anabaena 7120, PpetE normally regulates transcription of the gene encoding plastocyanin, which serves as an electron carrier for photosystem I under copper-replete conditions (Ghassemian et al., 1994). When copper is present in the medium, petE is transcribed in abundance, whereas in the absence of copper, petE is severely downregulated. Plasmid pSMC113 was used to replace the chromosomal hetN promoter region with the petE promoter from Anabaena 7120, without the incorporation of additional plasmid DNA, to generate strain 7120PN. In strain 7120PN, PpetE and the coding region of hetN are fused at their respective ATG translational start sites and are located immediately downstream of the proposed transcriptional terminator of hetM, the gene upstream of hetN on the chromosome (Fig. 1). The region between hetN and hetI remains the same as in wild-type Anabaena 7120.
To test the PpetE::hetN fusion strain, we incubated it with N2 as a nitrogen source and found that heterocyst formation is suppressed when a normal level of CuSO4 is included in the medium. Wild-type Anabaena 7120 forms a regular pattern of 10% heterocysts when grown without a source of combined nitrogen in BG-11 media, which contains CuSO4 at a concentration of 0.3 µM (Fig. 2A). Conversely, heterocyst formation in strain 7120PN is completely suppressed under the same conditions (Fig. 2B). Earlier studies had shown that suppression by hetN on a multicopy plasmid required an intact hetN gene (Black and Wolk, 1994; Bauer et al., 1995) and, here, we reinforce the idea that expression of the hetN gene is responsible for this suppression. More importantly, this result shows that expression of hetN in strain 7120PN is controlled by PpetE, allowing us to turn hetN expression on or off.
It was possible that the suppression of heterocyst formation by overexpression of hetN was the result of cis effects on a neighbouring gene (Black and Wolk, 1994), in particular hetI(Fig. 1). HetI appears to be a member of a recently discovered superfamily of phosphopantetheinyl transferases (Lambelot et al., 1996; Wolk, 2000), which post-translationally modify the acyl carrier protein domains of polyketide and fatty acid synthases to make them active. As such, hetI may be necessary for heterocyst glycolipid synthesis. Because there is no apparent terminator sequence present in the 42 nucleotides that separate the 3′ ends of hetN and hetI, overexpression of hetN could affect the levels of transcription or translation of hetI (such as the concomitant overproduction of an antisense mRNA that limits the production of HetI). This seems unlikely, as the hetN transcript is not large enough to encompass both hetI and hetN (Bauer et al., 1995), but the possibility remained. To address this issue, a PpetE::hetN fusion on replicating plasmid pSMC115 was introduced into wild-type Anabaena 7120. As with strain 7120PN, complete inhibition of heterocyst formation was observed in media containing CuSO4. Because plasmid pSMC115 contains no hetI DNA, suppression must result from the expression of the hetN gene alone.
Multiple contiguous heterocysts result when Anabaena 7120 is grown diazotrophically without HetN
To mimic a hetN null mutant, strain 7120PN was transferred from normal BG-11 containing both nitrate (or ammonia) and CuSO4 to medium without CuSO4 and combined nitrogen. After 24 h, the pattern of heterocysts that developed appeared to be the same as in the wild-type strain; single heterocysts appeared at approximately every tenth cell along the filament (Fig. 2C). Double heterocysts were occasionally observed in strain 7120PN but, for the most part, the pattern of heterocysts appeared to be the same as in the wild type. But at 48 h and thereafter, a Mch pattern was observed in strain 7120PN, with strings of three, four and occasionally five heterocysts in a row (Fig. 2D). Continued growth of the culture without combined nitrogen and the presence of polar cyanophycin granules suggest that the heterocysts are functional. Like other strains that have Mch (Buikema and Haselkorn, 1991), the growth rate of 7120PN under these conditions was less than that of the wild type. 7120PN had a doubling time of 23 h versus 14 h for the wild type. In contrast, removing CuSO4 from the medium did not affect the pattern of heterocysts formed by the wild-type strain.
We were again concerned that the hetN phenotype might result from cis effects on the transcription or translation of the convergently transcribed hetI gene. In this case, however, underexpression of hetN would result in excess HetI. To ensure that the delayed Mch phenotype was caused by lack of HetN and not increased production of HetI, hetI was overexpressed in a similar manner from the petE promoter on plasmid pSMC119 in wild-type Anabaena 7120; no unusual spacing of heterocysts was observed. In addition, the presence of hetN as the only open reading frame (ORF) on the multicopy plasmid pSMC131 prevented heterocyst formation by strain 7120PN grown without CuSO4 or combined nitrogen. Plasmid pSMC131 contains no hetI sequence. Therefore, we conclude that the Mch phenotype results from a lack of HetN.
It was possible that the delayed onset of the Mch phenotype resulted from the time necessary for breakdown of HetN and/or its putative product after the expression of hetN had been turned off. To address this possibility, the filaments were grown without copper for 48 h before the removal of combined nitrogen. The same delay in the Mch phenotype was observed after nitrogen step-down from either nitrate or ammonia. Therefore, the delayed onset of multiple heterocysts appears to be an inherent feature of the hetN null phenotype, and not an artifact of experimental design.
HetN prevents the patterned expression of hetR
In wild-type cells, an increase in hetR expression can be seen as early as 30 min after nitrogen step-down, and expression of hetR is both necessary and sufficient to promote heterocyst differentiation (Buikema and Haselkorn, 1991; 2001). A hetR::gfp fusion was constructed and introduced into both wild-type Anabaena 7120 and strain 7120PN to determine where in the regulatory cascade, with respect to hetR, HetN suppresses heterocyst formation. When wild-type Anabaena 7120 harbouring a translational fusion between hetR and gfp on plasmid pSMC127 was induced for heterocyst formation in media containing CuSO4, a pattern of proheterocysts fluoresced brightly against a background of lesser vegetative cell fluorescence after 8 h (Fig. 3A). In contrast, strain 7120PN showed no pattern of brightly fluorescent cells in the same medium; all the cells fluoresced at the low level of wild-type vegetative cells (Fig. 3B). Promoterless gfp in wild-type Anabaena 7120 and strain 7120PN showed no green fluorescent protein (GFP) fluorescence. So, overexpression of hetN does not completely block the expression of hetR, but it does prevent the patterned, high level of expression found in cells destined to become heterocysts. When expression of hetN is prevented by withholding CuSO4 from the medium, groups of cells express hetR, as indicated by GFP fluorescence starting at ≈ 20 h after induction (Fig. 3C). This pattern of expression is consistent with the Mch phenotype that becomes evident at 48 h. Within the groups of fluorescent cells, often one or more bright cells are flanked by cells of intermediate fluorescence, brighter than the majority of cells, but less intense than the adjacent cell. This pattern probably reflects the asynchronous expression of hetR that is responsible for the delay observed in the Mch phenotype.
HetN also acts downstream of hetR transcription to prevent heterocyst development
When hetR is overexpressed ectopically, 20–30% of the cells become heterocysts, even when the cells are grown with combined nitrogen (Buikema and Haselkorn, 2001). Figure 4A shows wild-type Anabaena 7120 harbouring plasmid pPetHetR, which contains a PpetE::hetR translational fusion (Buikema and Haselkorn, 2001). In this strain, hetR expression from the plasmid is controlled by copper levels in the same way that expression of hetN is controlled in strain 7120PN. In addition to having a Mch phenotype, the filaments also show decreased spacing between groups of heterocysts. But when pPetHetR is put into strain 7120PN and hetN is overexpressed along with hetR from the same type of promoter, only 1% of the cells are heterocysts in nitrogen-deficient media, and heterocysts are absent in the presence of combined nitrogen (Fig. 4B). Therefore, in addition to affecting the pattern of expression of hetR, overexpression of hetN appears to act downstream of hetR transcription to prevent heterocyst development.
HetN is not required for heterocyst glycolipid synthesis
Nitrogenase is an O2-labile enzyme complex, and glycolipids produced specifically by heterocysts are believed to protect the enzyme complex from O2 diffusion into the cell. Before the hetN gene was characterized, the initial transposon insertion mutant that was used to isolate hetN (designated N10) was shown to be incapable of making these heterocyst glycolipids (Ernst et al., 1992). However, strain N10 had a Het− phenotype, meaning that no signs of heterocyst differentiation were evident. Black and Wolk (1994) later speculated, and as our results above support, that this phenotype was most probably caused by a second-site mutation that prevents an otherwise Mch phenotype. The lack of heterocyst glycolipid production by strain N10 may also have been the result of this second-site mutation, and not the insertion in hetN. To determine whether expression of hetN is necessary for glycolipid production, we used thin-layer chromatography to visualize the glycolipids produced by strain 7120PN grown without CuSO4 or combined nitrogen (Fig. 5). Despite the absence of hetN expression, strain 7120PN had wild-type levels of the two heterocyst-specific glycolipids, which differ only by the reduction of a keto group to a hydroxyl at position C-3 of the lipid moiety (Gambacorta et al., 1996). Therefore, we believe that hetN is not necessary for production of the two glycolipids that have been shown to be specific to heterocysts. In addition, when hetN was overexpressed, no additional glycolipids were seen (Fig. 5).
HetN is made primarily in heterocysts
To determine which cell type normally makes HetN protein, a translational fusion between hetN and gfp was introduced into wild-type Anabaena 7120 on plasmid pSMC126, and the filaments were induced to differentiate by nitrogen step-down. A faint, but clearly discernible, GFP fluorescence signal was observed primarily in developing heterocysts at 17 h after induction (Fig. 6). A lower level of chlorophyll fluorescence is seen in vegetative cells, which is also seen in the wild-type strain without the plasmid. Chlorophyll fluorescence is normally lower in heterocysts than in vegetative cells of wild-type filaments. The weak signal from heterocysts seen here is consistent with earlier Northern results that indicated a low level of hetN expression 12 h after nitrogen step-down (Bauer et al., 1995). Localization of hetN expression in heterocysts is consistent with the idea that production of HetN is required for the generation of a heterocyst suppression signal that originates in mature, or nearly mature, heterocysts.
Initial characterization of hetN by Black and Wolk (1994) suggested that either HetN prevents heterocyst formation directly or that suppression of heterocyst formation by multicopy hetN is the result of its effects on hetI. If HetN is directly responsible for suppression, then the various insertion mutant phenotypes they observed could be the result of second-site mutations that masked an otherwise Mch phenotype. If effects on hetI are responsible, then the various mutant phenotypes could be attributed to the type of insertion and individual cis effects on hetI. Work by Bauer et al. (1995), who isolated hetN independently, was also indecisive on this point because, as in the previous study, hetN was accompanied by at least a portion of the hetI gene in each experiment. Here, we overexpress hetN alone, without any hetI DNA downstream, and show that HetN is itself responsible for the suppression of heterocyst formation. We also find that the hetN null phenotype is a Mch pattern that is delayed ≈ 24 h.
When an Anabaena hetR point mutant was initially complemented to isolate the gene from a cosmid library, many of the plasmids recovered from the complemented strain contained the mutated copy of the gene. Presumably, there was selection for genetic exchanges between the chromosome and the plasmid to alleviate the Mch phenotype that results from supernumerary copies of an intact hetR gene (Buikema and Haselkorn, 1991). Mch strains grow more slowly than wild type and, in extreme cases, may be non-viable. To avoid selecting second-site mutations that would mask the true hetN phenotype, we expressed hetN from the chromosome using an inducible promoter that could be subsequently downregulated to mimic the hetN null phenotype. When expression of hetN is turned off, the resulting phenotype is a Mch pattern that is visible 48 h after induction, 24 h after the appearance of the wild-type pattern. The initial wild-type pattern of heterocysts followed by the Mch phenotype indicates that hetN is necessary for the maintenance, and not the establishment, of heterocyst spacing.
Like hetN, the gene patS suppresses heterocyst formation when overexpressed, and a patS null mutant exhibits a Mch phenotype. Unlike hetN, however, patS is normally expressed early by cells that will become heterocysts, and the absence of patS results in a Mch phenotype during initial pattern formation (Yoon and Golden, 1998). Heterocyst patterning in Anabaena 7120 can be regarded as having two temporally distinct stages. De novo generation of heterocysts in a filament of vegetative cells occurs when cells are switched from nitrogen-replete to nitrogen-deficient conditions. Thereafter, the existing pattern is maintained during prolonged diazotrophic growth. PatS is clearly involved in de novo pattern formation but, after the first round of differentiation is complete, the expression of patS returns to preinduction levels (Yoon and Golden, 1998). On the other hand, the delay in the Mch phenotype when hetN is turned off and the fact that hetN is normally not expressed until 12 h after nitrogen step-down suggest that HetN does not normally play a role in de novo heterocyst pattern formation. Instead, it appears to be necessary for maintenance of the pattern as filaments lengthen by cell growth and division, and new heterocysts form between existing ones.
The Mch phenotype observed when hetN is turned off is not as extreme as that of a patS null mutant. In the patS mutant, heterocysts differentiate inappropriately on media containing nitrate, up to 10 contiguous heterocysts can be found in media lacking a nitrogen source, and the number of vegetative cells between groups of contiguous heterocysts is markedly reduced (Yoon and Golden, 1998). Filaments with hetN turned off have none of these properties. We propose that HetN signals to vegetative cells that a nearby cell has differentiated into a heterocyst, which suppresses differentiation by these vegetative cells. Without confirmation of heterocyst differentiation by HetN, the filament mistakenly initiates a second round of de novo differentiation, which results in the differentiation of a second heterocyst adjacent to the first. The timing of hetN expression, which begins after the initial pattern of heterocysts has been determined (Bauer et al., 1995), is consistent with this role for HetN in differentiation.
Heterocyst differentiation involves a regulatory cascade that culminates in the formation of a functional heterocyst (for a review, see Wolk, 2000). Mutations in genes such as ntcA (Frías et al., 1994; Wei et al., 1994), hetR (Buikema and Haselkorn, 1991), hetC (Khudyakov and Wolk, 1997) and hetP (Fernandez-Pinas et al., 1994) that act early in this cascade prevent the expression of downstream genes and lead to a Het− phenotype. We believe, as Black and Wolk (1994) speculated, that mutant strain N10 has a second-site mutation in a gene early in the regulatory cascade. We also suspect that this second-site mutation accounts for the absence of heterocyst glycolipids in strain N10, because we found that cells with hetN turned off contained wild-type levels of heterocyst glycolipids, indicating that HetN is not necessary for their production. Instead, we suspect that hetN is required for the production of an inhibitory signal that suppresses heterocyst formation. A GFP fusion indicated that HetN is made primarily in heterocysts, so such an inhibitory signal would have to travel along the filament to suppress the development of intervening vegetative cells.
By expressing hetN unusually early in the differentiation process, we were able to identify where in the regulatory cascade hetN inhibits the formation of heterocysts. Presumably, there are factors that are specific to the maintenance of the pattern, but major factors such as HetR are undoubtedly involved in both de novo patterning and its maintenance. Overexpression of hetN prevented the appearance of the normal pattern of hetR expression, which suggests that HetN can block heterocyst formation upstream of hetR in the proposed regulatory cascade. On the other hand, when the normal transcriptional controls are bypassed by replacing the hetR promoter with an inducible promoter, overexpression of hetN can still prevent heterocyst formation, suggesting that either HetN also acts downstream of hetR, or it blocks a process distinct from the regulatory cascade controlled by hetR that is also necessary for heterocyst differentiation. The simplest explanation of these results, given what is known about hetR, is that hetN prevents the positive autoregulation of hetR(Fig. 7). Inhibition of any step in the positive feedback of hetR by HetN would be consistent with our results. For instance, inhibition could be mediated by interference with the modification of HetR that is believed to occur in nitrogen-starved filaments (Zhou et al., 1998b). If modification of HetR is necessary for both autoregulation and differentiation, then blockage of this modification would explain the apparently enigmatic relationship between hetR and hetN. Alternatively, the apparent activity of HetN both upstream and downstream of hetR could be explained by multiple points of suppression by HetN in the cascade of regulators leading to heterocyst formation. The earliest known factor specific to heterocyst differentiation, HetR, may be responsible for the integration of a variety of signals that indicate the nutritional and developmental state of the cell and its neighbours. To date, the transcriptional factor NtcA, which regulates nitrogen metabolism (Frías et al., 1994), hanA, which encodes the histone-like protein HU (Khudyakov and Wolk, 1996), and HetN have been shown to influence the regulation of hetR. Identification of the status of HetR in filaments overexpressing hetN and isolation of suppressors of the hetN overexpression Het− phenotype should clarify the mechanism of heterocyst suppression by HetN.
Anabaena 7120 and its derivatives were grown in BG-11 medium as described previously (Buikema and Haselkorn, 1991), supplemented with neomycin at a concentration of 45 µg ml−1 for the selection of replicating plasmids or 90 µg ml−1 for non-replicating plasmids that had integrated into the chromosome. Escherichia coli strains were grown in Luria–Bertani (LB) broth for liquid culture and LB solidified with 1.5% agar for plate culture. For selective growth, media were supplemented with 100 µg ml−1 ampicillin, 50 µg ml−1 kanamycin or 10 µg ml−1 chloramphenicol.
To induce heterocyst formation, exponentially growing cells at a concentration of ≈ 107 cells ml−1 were washed three times with 10 mM HEPES, pH 7.5, resuspended in BG-11 without combined nitrogen or antibiotics and incubated under growth conditions. The optical density at 750 nm (OD750) of cells resuspended in 30% sucrose (w/v) after centrifugation was used to measure the growth of cultures (Sigalat and Kouchkowski, 1975; Tandeau de Marsac and Houmard, 1988). Growth rates are the average of rates from duplicate cultures, which differed by less than 10%. For copper-replete conditions, standard BG-11 medium, which contains ≈ 0.3 µM CuSO4, was used. For copper-deficient conditions, CuSO4 was omitted from BG-11 medium, sterile plasticware was used where possible and solutions were filter sterilized. When glass containers were necessary, they were washed with 100 mM HCl after autoclaving and rinsed with filter-sterilized water before use. These conditions are critical because the petE promoter can be activated by very low levels of Cu2+.
Each gene fusion was generated in three sequential polymerase chain reactions (PCRs) using a total of four primers. The two primers at the fusion junction each contained a 13 bp 5′ tail with sequence complementary to the corresponding junction primer for a total of 26 bp overlap between the two. In the first step, the individual gene fragments to be joined were generated in a standard PCR reaction using Anabaena 7120 DNA, unless otherwise mentioned, as template. One microlitre of the gel-purified DNA diluted 1:5 in water was then cycled five times in a standard PCR reaction without additional primers. This step allowed each of the two initial products to serve as template for the other via the 26 bp overlap junction and, consequently, extend to full-length fusion product. In the third step, one microlitre of the previous reaction served as template in a standard PCR reaction using the two primers flanking the desired product. PCR fusion fragments were gel purified, cloned and sequenced to verify the integrity of the final products. After subcloning the fusion products into the appropriate vectors, plasmids were transferred by conjugation from E. coli strain UC585 (Liang et al., 1993) into Anabaena as described previously (Elhai and Wolk, 1988). To create pSMC113, the 3′ end of hetM was generated by PCR using primers hetM1142-F (5′-TTTAGATCTGCTGGTAAGGTAGTCAAGGAAAGTG-3′), which contains a BglII site (underlined), and hetM-R (5′-GTTCAATGAGGTTTAACCCGAATTCAAC-3′) and cloned into pGEM-T (Promega) to generate pSMC110. Primers PpetE-F (5′-GCTGACGTACTGAGTACACAGCTAAT-3′), PpetE-R (5′-CCTGTAAGAGTTGTCATGGCGTTCTCCTAACCTGTAG-3′), hetN-F (5′-TAGGAGAACGCCATGACAACTCTTACAGGTAAGACAG-3′) and hetN-R containing a 5′NdeI site (5′-TTTCATATGCATGAGCGATGAGACTCAACAGCTA-3′) were used to fuse the petE promoter to hetN at their respective ATG start codons. This fragment was cloned into pGEM-T Easy (Promega) and subsequently moved into pSMC110 as an NdeI–NotI fragment to create pSMC112. The PpetE::hetN fusion with the 3′ end of hetM located upstream was then moved to pRL278 (Cai and Wolk, 1990) as a BglII–SacI fragment to create pSMC113, which cannot replicate in Anabaena.
Plasmid pSMC115 was created by moving the PpetE::hetN fusion fragment referred to above from pGem-T Easy to pBluescript SK+ (Stratagene) as an EcoRI fragment to acquire the BamHI and KpnI sites that were subsequently used to move it to the BamHI–KpnI sites of pAM504 (Wei et al., 1994) to create pSMC115.
For pSMC119, the petE promoter and hetI were fused at their respective ATG start codons as described above using primers PpetE-F, petEI-R (5′-CAAGTATGCTGCAACATGGCGTTCTCCTAACCTGTAG-3′), hetI-F (5′-GTTAGGAGAACGCCATGTTGCAGCATACTTGGCTACC-3′) and hetI-R containing an NdeI site (5′-TTTCATATGCTCCGCAGTTGCTTAGGGAATGAGGT-3′). The resulting PCR product was cloned into pGEM-T Easy, transferred to pBluescript SK+ as an EcoRI fragment, excised with BamHI–KpnI and ligated into the BamHI–KpnI sites of pAM504 to create pSMC119.
For pSMC126, the hetN promoter and gfp were fused at their respective ATG start codons as described above using primers PhetN-BamF containing a BamHI site (5′-GGATCCGTTCTTAACCTTGGCGTGAGGAG-3′), PhetN-gfpR2 (5′-TTCTCCTTTACTCATTGTAACCTGCTAGTCTCCAAATTC-3′), gfp-hetNF2 (5′-TAGCAGGTTACAATGAGTAAAGGAGAAGAACTTTTCACTG -3′) and gfp-KpnR2, which contains a KpnI site (5′-GGTACCTTATTTGTATAGTTCATCCATGCC-3′). Plasmid pAM1951 (Yoon and Golden, 1998) was used as the gfp template DNA. The resulting PCR product was cloned into pGEM-T and subsequently moved into pAM504 as a BglII–KpnI fragment to create pSMC126.
For pSMC127, the hetR promoter region and gfp were fused at their respective ATG start codons as described above using primers PhetR-BamF containing a 5′BamHI site (5′-GGATCCCCTGCCAATGCAGAAGGTTAAAC-3′), hetR-gfpR2 (5′-TTCTCCTTTACTCATATTACAAATAGTTGAATAGCACGC-3′), gfp-hetRF2 (5′-ACTATTTGTAATATGAGTAAAGGAGAAGAACTTTTCACTG -3′) and gfp-KpnR2. Plasmid pAM1951 was used as the gfp template DNA. The resulting PCR product was cloned into pGEM-T and subsequently moved into pAM504 as a BglII–KpnI fragment to create pSMC127.
Construction of strain 7120PN
Plasmid pSMC113 was transferred by conjugation into wild-type Anabaena 7120, and cells in which the plasmid had integrated into the chromosome were selected for growth on neomycin. Two such plasmid integrates were grown in liquid culture without selection before the filaments were fragmented to mostly single cells using a Branson 1200 ultrasonic cleaner. The cells were subsequently plated on BG-11 containing 5% sucrose to select for cells in which resolution of the plasmid integrate resulted in loss of the sacB gene, which was part of the plasmid (Cai and Wolk, 1990). Eight of the 22 colonies tested were unable to grow on nitrogen-deficient medium containing copper, indicating that the hetN promoter had been replaced by the petE promoter. Replacement of the promoter was confirmed by PCR using the following combinations of primers: PpetE-F and hetI-F; hetM1142-F and PpetE-R; PhetN-BamF and PhetN-gfpR2. With the first two combinations of primers, appropriately sized fragments were observed for the eight strains unable to make heterocysts when grown with CuSO4, and no product was present for two strains tested that were able to grow on the same media. The results were reversed when the third combination of primers was used, indicating that, in the eight strains that could not grow on nitrogen-deficient medium containing CuSO4, the normal hetN promoter is absent. Each of these eight strains also exhibited a Mch phenotype when grown in nitrogen-deficient medium lacking copper, as described in the Results.
Photomicroscopy of Anabaena filaments on agar slabs was performed with a Zeiss Axiophot microscope equipped with fluorescence optics.
Wild-type Anabaena 7120 and strain 7120PN were grown in 200 ml cultures to mid-log phase, washed three times with 10 mM HEPES, pH 7.5, and each was split into two 200 ml cultures containing appropriate media. After 24 h, the cultures were extracted twice with 2:1 chloroform–methanol. The organic phases were separated on a silica gel 60 plate measuring 10 cm by 20 cm (EM Science) using a mobile phase of chloroform, methanol, acetic acid and water in a ratio of 85:15:10:3.7 (Nichols and Wood, 1968). Plates were developed by charring with 25% sulphuric acid at 240°C for 5 min to visualize glycolipids.
We thank Jeffrey Elhai and James Golden for supplying plasmids used in this study. We are grateful to Robert Haselkorn for his intellectual support, critical reading of the manuscript and stimulating discussions. This work was supported by NIH research grant GM21823 to Robert Haselkorn. S.M.C. is supported by Public Health Service grant GM20164.