DNA binding properties of the HrmR protein of Nostoc punctiforme responsible for transcriptional regulation of genes involved in the differentiation of hormogonia

Authors


Summary

Nostoc punctiforme is an example of a filamentous cyanobacterium that is capable of differentiating non-growing cells that constitute gliding filaments termed hormogonia. These gliding filaments serve in short distance dispersal and as infective units in establishing a symbiosis with plants, such as the bryophyte Anthoceros punctatus . Mutants of N . punctiforme exist which show elevated levels of initial infection of A . punctatus as a consequence of repeated cycles of hormogonium differentiation. Such mutations occur within the hrmA and hrmU genes. Further characterization of the hrm locus revealed several genes with an organizational and predicted protein sequence similarity to genes of heterotrophic bacteria that are involved in hexuronic acid metabolism. Genes in the N. punctiforme locus are transcribed in response to the presence of a plant extract containing hormogonium-repressing factors. A predicted transcriptional repressor encoded in the locus, HrmR , was shown herein to be a specific DNA binding protein that regulates the transcription of its own gene and that of hrmE , a nearby gene. The ability of HrmR to bind DNA was abolished upon addition of either galacturonate or lysate from specifically induced N . punctiforme cells, implying that the in vivo HrmR binding activity is modulated via an internal compound, most likely a sugar molecule.

Introduction

The oxygenic photosynthetic cyanobacterium Nostoc punctiforme strain ATCC 29133 (PCC 73102) is a unique bacterial representative because of its ability to develop into four distinct cell types: vegetative cells, heterocysts (specialized for nitrogen fixation), akinetes (spore-like cells for survival under cold desiccation conditions), and hormogonia (gliding filaments specialized for dispersal) (Meeks et al., 2002). Each cellular differentiation alternative occurs in response to specific environmental conditions. Nostoc punctiforme is also capable of entering into symbiosis with a wide range of plants (Meeks, 1998) and fungi (Mollenhauer et al., 1996). Symbiotic partners include bryophytes such as Anthoceros punctatus (Enderlin and Meeks, 1983), cycads represented by Macrozamia sp., from which it was originally isolated (Rippka et al., 1979), and the angiosperm Gunnera spp. (Johansson and Bergman, 1994). Two of the developmental alternatives, hormogonia and heterocysts, are of particular interest because they are under plant influence in the establishment of a nitrogen-fixing symbiosis. First, hormogonium formation is essential, but not wholly sufficient, for establishment of the A. punctatus association (Campbell and Meeks, 1989). The plant partner negatively and positively influences hormogonium formation and the behaviour of the formed hormogonia as they actively colonize existing cavities in the A. punctatus gametophyte tissue (Meeks, 1998). Second, once within the plant tissue, using unknown mechanisms, A. punctatus regulates growth and metabolic activities of the cyanobacterium and enhances the frequency of heterocyst differentiation by about threefold, relative to free-living cultures (Meeks and Elhai, 2002).

Similar to the Rhizobium-legume symbiosis (van Rhijn and Vanderleyden, 1995), chemical signalling is essential in formation of cyanobacterial symbioses. There is experimental evidence for at least three chemical signalling steps in establishment of the Nostoc spp.–A. punctatus association. First, nitrogen limited A. punctatus (and other symbiotic plant partners) excretes a hormogonium-inducing factor (HIF), which serves to ensure a high frequency of hormogonia in the cyanobacterial population (Campbell and Meeks, 1989). Second, to further increase the probability of an infection, the motile hormogonia are influenced by plant-produced chemoattractants to glide onto the gametophyte tissue and colonize the symbiotic cavities (Knight and Adams, 1996). Third, following colonization, a hormogonium repressing factor (HRF) is produced into the cavity; HRF is dominant over HIF and is thought to shift the developmental cascade away from formation of hormogonia, thereby allowing heterocyst differentiation and expression of nitrogenase (Cohen and Meeks, 1997). The chemical identities of the chemoattractants, HIF and HRF are unknown. The specific signals that result in the integrated physiology of slow growth, altered metabolism and enhanced heterocyst frequency of N. punctiforme in association with A. punctatus, or any other plant, are also unknown. Moreover, relatively little is known about the genetic targets in N. punctiforme that respond to the chemical signals and nothing at all about how these signals are transduced within the cells to affect transcriptional regulation of genes involved in the physiological adaptation.

One gene that has been shown to positively respond to the plant produced HIF encodes an alternative sigma subunit, SigH (Campbell et al., 1998). The genes that depend on SigH for transcription and the factors that lead to the induction of sigH transcription are unknown. The sigH mutant showed an elevated level of initial infection of A. punctatus (Campbell et al., 1998). Another N. punctiforme strain with a similar high infection phenotype is associated with a mutation in a gene designated hrmA (Cohen and Meeks, 1997). Mutations in hrmA and its upstream gene, hrmU, result in elevated levels of hormogonium formation, relative to the wild type, in the presence of HIF and the mutants appear to repeatedly produce hormogonia during co-culture with A. punctatus. SigH does not appear to be involved in transcription of hrmUA (Campbell et al., 1998).

Whereas the repeated hormogonium differentiation response lengthens the window of infection leading to high levels of initial colonization of the plant tissue by the hrmA and hrmU mutants, it is ultimately lethal by extinction for the following reason. Upon entry into the hormogonium cycle, vegetative cells divide uncoupled from biomass increase; there is a cessation of DNA replication (Herdman and Rippka, 1988) and no net increase in protein or chlorophyll (Campbell and Meeks, 1989). Cells of hormogonia are consequently smaller than cells of vegetative filaments. A partial recovery of macromolecular components occurs as the cells of hormogonia exit the hormogonium cycle and return to the vegetative cell state. Wild-type N. punctiforme normally exhibits a period of ‘immunity’ after the initial round of HIF-dependent hormogonium differentiation. The ‘immunity’ period is thought to provide the necessary recovery time for vegetative cells to completely regain biomass and accumulate the multiple chromosome copies characteristic of cyanobacteria (Herdman et al., 1979). The hrmA and hrmU genes are hypothesized to be involved in an ‘immunity’ period because the respective mutants express the predicted phenotype of repeated hormogonium formation to extinction (Cohen and Meeks, 1997). The hrmU and hrmA transcripts are induced by HRF, consistent with the idea that, once inside the symbiotic cavity, hormogonium formation is repressed in favour of heterocyst differentiation. In this paper, we present further sequence analysis 5′ of hrmUA, identifying four additional genes that have similarity to genes encoding enzymes of carbon metabolism. We specifically focus on characterization of the activity of a putative transcriptional repressor encoded by the hrmR gene in this locus and propose a model for the regulation of genes in the hrm locus by plant signals.

Results

Sequence analysis revealed the presence of two genes (hrmR and hrmI) 5′ of hrmU that are transcribed in the same direction, and a third gene (hrmK) transcribed divergently from the hrmRIUA cluster (Fig. 1). A fourth gene (hrmE ), is present 3′ from hrmK and separated from it by two divergently transcribed open reading frames encoding unknown proteins. The translated sequence of hrmE has 55% similarity to an aldehyde reductase; hrmK, 49% similarity to a gluconate kinase; hrmR, 39% similarity to transcriptional repressors in the LacI/GalR family, which includes UxuR, a transcriptional repressor of genes involved in hexuronic acid metabolism (Shulami et al., 1999); hrmI, 36% similarity to uronate isomerase; and hrmU, 57% and 55% similarities to 2-keto-3-deoxygluconate dehydrogenase and mannonate oxidoreductase respectively. The hrmA product has no significant sequence similarity in the entire database as of October, 2002. Except for HrmE, the known proteins are characteristic of gluconate, glucuronate and galacturonate catabolism in Escherichia coli (Lin, 1996; Peekhaus and Conway, 1998) and Bacillus species (Mekjian et al., 1999; Shulami et al., 1999). The genes in this locus of the N. punctiforme chromosome are absent in the sequenced genomes of closely related heterocyst-forming Anabaena sp. strain PCC 7120 (Kaneko et al., 2001) and the unicellular cyanobacterium Synechocystis sp. strain PCC 6803 (Kaneko et al., 1996), except for the unknown protein immediately 3′ from hrmK, which is also found in Synechocystis PCC 6803.

Figure 1.

A. Includes a map of open reading frames in the hrm locus of N. punctiforme . HrmA has no significant similarity in the database. HrmU shows similarity to 2-keto-3-deoxygluconate dehydrogenases and mannonate oxidoreductases. HrmI shows similarity to uronate isomerase. HrmR has similarity to the LacI/GalR family of transcriptional repressors. HrmK has similarity to gluconate kinases. Unk are unknown proteins. HrmE has similarity to aldehyde reductases. Arrows indicate the direction of transcription. Also shown is the location and direction of transcription of the cassette insertion into the hrmR gene. * indicates location of HrmR binding sites. Bar and number 1 indicates the location of the RS2 fragment, bar and number 2 is fragment RB2 and bar and number 3 is fragment RE2 used in EMSA experiments. B and C include identification of the transcript 5′ ends of the hrmR and hrmE genes and locations of the putative promoter regions with HrmR binding sites. Primer extension gels are shown on the left for hrmR in B and hrmE in C. The HrmR binding sites deduced from EMSA are shown relative to the putative transcriptional start points. C. The * shows the one base deviation from the other two identified binding sites. The − 10 and − 35 boxes drawn were drawn based solely on the location of each putative transcript start site and not on any specific knowledge of sequences recognized by a N . punctiforme sigma factor; in cyanobacteria, the − 35 box in particular could have various hypothetical designations.

Because the proteins homologous to HrmR are documented to function as transcriptional repressors, and UxuR binds to its own gene promoter, we asked whether the N. punctiforme HrmR protein could bind to specific DNA sequences and regulate transcription in the hrm locus. We first analysed binding 5′ of hrmR and as a consequence of those studies identified binding sites 5′ of hrmE. Primer extension experiments show a single 5′ end for both hrmR and hrmE only when cells were induced with HRF (Fig. 1B and C). The binding sites of HrmR, as determined in the experiments described below, relative to the putative transcript start points identified by primer extension, are depicted in Fig. 1B and C. The binding of HrmR in both genes is near the predicted site of either RNA polymerase binding and/or open complex formation (Record et al., 1994). Cyanobacterial promoters are characterized by general analogy to the E. coliσ70 promoter sequence; the consensus − 10 sequence is TANNNT and the − 35 sequence is highly variable (Curtis and Martin, 1994). The hrmR putative promoter sequence is consistent with the cyanobacterial consensus − 10 sequence, while the hrmE putative promoter lacks the 5′ T in the − 10 region.

The recombinant His6-HrmR (r-HrmR) protein was produced as an inclusion body in E. coli; it was denatured and refolded (see Experimental procedures) before gel shift experiments. The refolded r-HrmR bound to and retarded the migration of a 32P-labelled 79 bp (RS2 fragment; depicted by the bar labelled 1 in Fig. 1A) DNA fragment containing sequence near to the putative promoter region of hrmR (Fig. 2A). The RS2 gel shift was inhibited by the addition of antibody specific for the T7 portion of the recombinant r-HrmR protein. When the RS2 template was divided into two smaller fragments, RSL (bp + 1 to + 34, relative to the putative transcriptional start point) and RSR (bp + 28 to + 60), only RSL bound r-HrmR (Fig. 3; RSR data not shown). In the presence of increasing amounts of unlabelled either RSL or RSR fragments, only increasing amounts of RSL diminished binding to the radiolabelled RSL fragment (Fig. 3). Thus, r-HrmR, and we infer thereby HrmR, is a DNA binding protein that recognizes specific DNA sequences and not DNA in general.

Figure 2.

Electrophoretic mobility shift assays (EMSA) using r-HrmR. The presence or absence of the r-HrmR protein and/or antibody (Ab) to the T7 portion of the recombinant protein in each of the EMSA reactions is indicated by + or – above each of the lanes.

A. The labelled DNA template is fragment RS2.

B. The labelled DNA fragment is RB2.

C. The labelled DNA fragment is RE2.

Figure 3.

Competitive electrophoretic mobility shift assays using r-HrmR. The radiolabelled DNA template in all lanes is RSL.

A. Unlabelled RSL was added at the relative concentrations indicated across the top of the lanes.

B. Unlabelled RSR was added.

Upon close examination of the sequence of the RSL fragment, a 14 bp palindromic sequence with an axis of symmetry (5′-TGCACACGTGTGCA-3′) can be identified. A search of the unfinished N. punctiforme genome sequence using blast (Altschul et al., 1990) at http:www.jgi.doe.gov, with the 14 bp palindromic sequence as query, revealed only one other site containing exactly the same 14 bp sequence. This site is present 5′ of the gene we have termed hrmE, located approximately 2.3 kb 5′ of the hrmR gene in the chromosome (see Fig. 1A). An 89 bp DNA fragment, 5′ from the hrmE putative transcriptional start and containing the 14 bp palindromic sequence (fragment RB2; depicted by the bar labelled 2 in Fig. 1A), was specifically recognized by the r-HrmR protein in the gel shift assay (Fig. 2B). The r-HrmR binding to RB2 was also inhibited by antibody to the T7 portion of the recombinant protein (Fig. 2B). The only sequence similarity between fragments RS2 and RB2 is the 14 bp palindromic sequence. At 29 bp 3′ of the HrmR binding site near hrmE, a sequence is present that differs in 1 bp from the 14 bp palindromic sequence recognized by r-HrmR (5′-TGCACACCTGTGCA-3′). A gel shift assay using a fragment containing both the 14/14 and the 13/14 matches (fragment RE2; depicted by the bar and number 3 in Fig. 1A) indicated that indeed r-HrmR bound to both of these sequences (Fig. 2C). Thus, there are apparently two r-HrmR binding sites near the hrmE gene (Fig. 1C). In addition to genes of the hrm locus, the 13–14 bp palindromic sequence is absent in the genome of Anabaena PCC 7120 (Kaneko et al., 2001).

To test other similar sequences for specificity of r-HrmR binding, the N. punctiforme genome was searched for near matches of the 14 bp palindromic sequence that were also located 5′ of putative open reading frames. While no 12/14 sequence matches were detected, seven 11/14 matches were identified. Three of the 11/14 sequences were analysed by in vitro binding assays; they are identified in Table 1 as the annealed products of primers P183 and P184, P185 and P186, and P187 and P188. All three fragments contained the putative binding sequence in the middle of the fragment with surrounding genomic sequences on either side. Gel shift assays were conducted together with fragment RSL, which is similar in size, as the control. None of the 11/14 sequences showed a gel shift, while fragment RSL displayed a typical shift (data not shown).

Table 1. .  Primers used in the study, including those that upon annealing gave rise to templates for EMSA.
Primer5′ to 3′ sequence
P108ATGAATAAACGAAGAATTTC
P109GGGGAAGCTTCAAGATAGACACGATTTCAA
P117AATACATACAACTGCACACG
P120CTTACACCTGCTCTCCGAGC
P124ATCTTCAATTGAAATTCTTCG
P153AACTGCACACGTGTGCATTATTTTAAAATC
P154TTTAGATTTTAAAATAATGCACACGTGTGC
P155ATCTAAACTAAAAAATGAATAAACGAAG
P156AAATTCTTCGTTTATTCATTTTTTAGTTT
P164TGTTTTGTATACAAATGTAC
P165AGGTGTGCAAAACCTGTTTT
P166AACAAGCTCAATACAATCCG
P167GTCTTAACTTTTGCCAATCG
P168TGTAACGGATTGTATTGAGC
P183GTCAGGTAGCACACCTGTGCTACCTAATTT
P184AAAAAATTAGGTAGCACAGGTGTGCTACCT
P185CAGGGAAGACTGCACAGGTGTACCTTGTGC
P186CATGCACAAGGTACACCTGTGCAGTCTTCC
P187TTCACGAGTTTGCACACCTGTTTTGGAAT
P188TGAATTCCAAAACAGGTGTGCAAACTGC

The binding of r-HrmR to the RS2 (hrmR) template could be abolished by the addition of 80 mM galacturonate (Fig. 4A, lane 5) and not by any of the other sugars tested (Fig. 4), including those that support dark heterotrophic growth (fructose, glucose and sucrose) of N. punctiforme. Lower concentrations of galacturonate did not have the same inhibitory affect on DNA binding (data not shown). Because HRF induces transcription of hrmUA (Cohen and Meeks, 1997), lysate of HRF exposed cells could contained a ligand that binds to HrmR. Extract from hrmR mutant strain UCD 472, incubated with HRF before lysis, contained factor(s) which inhibited binding of r-HrmR to the RS2 fragment (Fig. 4B, lane 4). However, HrmR binding to the RS2 fragment was not affected by extracts of strain UCD 472 cells preincubated with naringin (Fig. 4C, lane 4), which induces hrmA expression (Cohen and Yamasaki, 2000), nor by extracts of strain UCD 472 cells not exposed to either HRF (Fig. 4B, lane 3) or naringin (Fig. 4C, lane 3). These results indicate that only HRF-induced cells contain the putative HrmR-binding ligand. In parallel experiments N. punctiforme strain UCD 328 (hrmA-luxAB; Cohen and Meeks, 1997) incubated with HRF for 6 h exhibited elevated levels of Lux activity (2.7 × 104 c.p.m. of light emission) as measured by a luminometer versus a control culture of strain UCD 328 which showed only basal levels (3.9 × 103 c.p.m.). Similarly, naringin incubated strain UCD 328 exhibited elevated levels of Lux activity at 1.3 × 104 c.p.m. as compared with the control flask at 4 × 103 c.p.m. HRF is a crude extract that apparently contains more than one factor that influences transcriptional activity in N. punctiforme.

Figure 4.

Survey of the effect of sugar and cell extract on binding activity of r-HrmR to the RS2 fragment. Lanes 1 of each of the three panels are negative control lanes with no added r-HrmR. Lanes 2 of each panel are positive control lanes containing r-HrmR.

A. 80 mM of each of the following sugars was added in the binding reaction: lane, 3 glucose; lane 4, sucrose; lane 5, galacturonate; lane 6, gluconate; lane 7, glucuronate; lane 8, fructose.

B. Lane 3 contains extract from N. punctiforme hrmR mutant strain UCD 472 and lane 4 contains extract from strain UCD 472 preincubated with HRF.

C. Lane 3 contains extract from strain UCD 472 and lane 4 contains extract from strain UCD 472 preincubated with naringin.

To determine whether HrmR regulates transcription of the hrmR and hrmE genes in vivo, we examined the pattern of transcription by Northern hybridization following exposure to HRF for varying times (Fig. 5). hrmR appeared as a 1.6 knt and hrmE as a 1.3 knt transcript. Both transcripts were unstable as evidenced by the low molecular mass smears in the blots. Approximately four times as much total RNA was required to detect these transcripts than is typically used in analysis of other N. punctiforme genes (Summers and Meeks, 1996; Campbell et al., 1998). The departure of the hrmE promoter from the cyanobacterial consensus − 10 sequence (Fig. 1C) is probably not the primary explanation for its low level of transcription, because hrmR has a similar low level of transcription, but a consensus − 10 sequence (Fig. 1B). In both cases, the hrmR and hrmE genes were induced in a burst pattern starting within 1.5 h exposure of cells to HRF and the transcripts decreased by 6 h, to no detectable RNA by 24 h. Transcripts were not detected from either gene under vegetative growth conditions in wild-type cultures (Fig. 5, lane 1 in all panels). However, in the hrmR mutant (PpsbA-npt cassette-insertion) strain UCD 472, hrmE (Fig. 5A) was transcribed even under normal vegetative growth conditions. hrmR was also transcribed in vegetative cells of strain UCD 472, but the data were not shown due to the large amount of transcription from the strong psbA promoter used in the insertion. These results are consistent with HrmR repression, in vivo, of hrmR and hrmE transcription in wild-type vegetative cells.

Figure 5.

Northern hybridizations using RNA from cells exposed to HRF for the amount of time indicated across the top of each lane. Strain UCD 472 is RNA from the hrmR mutant grown under normal vegetative growth conditions (not exposed to HRF) and in A was hybridized with the hrmE probe and in B was hybridized with the hrmR probe. Transcript sizes indicated were determined relative to standards electrophoresed on the same gel.

The phenotype of hrmR mutant strain UCD 472 is identical to the wild type in free-living culture, but it repeatedly failed to infect A. punctatus. Co-cultures of A. punctatus with wild-type N. punctiforme, within the same time frame and under the same experimental conditions, yielded typical infection frequencies. The addition of gluconate, glucuronate, or galacturonate did not increase or decrease the infection frequency of the wild-type or mutant strains during co-culture experiments and none supported dark heterotrophic growth of N. punctiforme.

Discussion

Based on sequence similarity, HrmR is predicted to be a DNA binding protein that also binds sugar ligands. The results presented here are consistent with those predictions. r-HrmR, and from which we infer that HrmR, is a DNA binding protein, with a distinct recognition sequence, that represses transcription of the hrmR and hrmE genes in vivo. The steric hindrance of the attached antibody prevented in vitro binding of the r-HrmR protein to its DNA target and shows that it was specifically the r-HrmR protein that bound to DNA and not another component from the protein preparation. This specificity is true whether the DNA template contains the hrmR promoter region (RS2 fragment) or the hrmE promoter region (RB2 fragment). Because the two fragments, RS2 and RB2, as well as the 33-mer RSL, have only the 14 bp palindromic sequence (5′-TGCACACGTGTGCA-3′) in common, we propose that this is the recognition site for the HrmR protein. Although we are interested in whether HrmR binds as a monomer, or as a higher order associated protein as with LacI, we currently do not know if cooperative binding or DNA looping plays a role in regulation of either or both promoters. The second binding site (5′-TGCACACCTGTGCA-3′) in the hrmE operator, which matches in 13 out of 14 bases, may indicate that hrmE is under more stringent transcriptional regulation than is hrmR. Less stringent repression of hrmR is consistent with the requirement for HrmR in vegetative cells to provide repressor function. The in vivo transcription of both the hrmE and hrmR genes was no longer repressed in the hrmR mutant, as shown by Northern hybridization. We conclude that the HrmR protein is responsible for repression of their transcription under normal vegetative growth conditions. The lesser conserved sequences, with 11 out of 14 matches, that were examined failed to bind r-HrmR. However, the refolding process may not have completely restored the full binding activity of the protein. Thus, we are currently unaware of any other HrmR binding sites in the genome, although it is possible that other sites exist and must be elucidated using other methods.

Because HrmR is predicted to be a member of the sugar-binding LacI/GalR family of transcriptional repressors, the possible affects of various sugars on the binding of HrmR to DNA were examined. Of the sugars tested, only galacturonate abolished binding activity. However, the concentration of sugar required was not stoichiometric with the amount of HrmR protein added in the assay. This may indicate that the actual in vivo inducer is not galacturonate, but a similar molecule, or that the renaturation process used to recover r-HrmR did not properly fold the sugar-binding portion of the protein; consequently, the binding capacity for its sugar effector was greatly diminished. Regardless of the binding affinity, the result is consistent with a predicted sugar-binding activity by HrmR. DNA binding of the global nitrogen regulator, NtcA, from Synechococcus sp. strain PCC 7942 is enhanced by the presence of 2-oxoglutarate (Vazquez-Bermudez et al., 2002). We are aware of no other experimental reports of sugar-binding transcriptional repressors in cyanobacteria. We hypothesize that the pathway of transducing the HRF signal into the cell in order to affect the transcription of at least the hrmR and hrmE genes involves a sugar molecule like, or containing, galacturonate (summarized in Fig. 6). This is our first demonstration of how N. punctiforme may sense an A. punctatus signal and regulate transcription of genes in response to that signal. Because extracts of cells induced in the presence of naringin do not inhibit HrmR binding to DNA, in contrast to HRF induced cell extract, we conclude that the in vivo transcriptional regulation of at least the hrmA gene is not the same mechanism as that for hrmR and hrmE. This observation is also consistent with results indicating that HrmR does not regulate the transcription of the hrmI, hrmU, or hrmA genes in the context of regulon or operon structure; the HrmR binding sequence is absent in the 5′ regions of those genes and Northern hybridization bands are consistent with primarily co-transcription of hrmUA and monocistronic hrmI transcription (data not shown). In this respect, transcription in the hrm locus of N. punctiforme differs from the regulon structure and polycistronic expression of hexuronic acid catabolic genes in E. coli (Peekhaus and Conway, 1998) and Bacillus species (Mekjian et al., 1999; Shulami et al., 1999). Transcriptional regulation in the hrm locus is complex and appears to involve different signal molecules and mechanisms, one of which is sugar-binding dependent derepression.

Figure 6.

Hypothetical model of HRF-dependent modulation of HrmR transcriptional regulation. In this model HrmR binds to the operator region of the hrmR and HrmE genes, repressing transcription. HRF, or a metabolic derivative, is transported into the cells; HRF, or a derivative, binds to HrmR and the ligand-HrmR molecule does not bind to the hrmE and hrmR operator regions, resulting in derepression of transcription. The concentration of the ligand decreases, as a result of instability or metabolism, relative to the increasing synthesis of HrmR, thereby allowing HrmR to again bind to the operator regions and repress transcription, especially of hrmE .

Expression of structural genes of the hrm locus is hypothesized to contribute to an ‘immunity’ period reflected in repression of hormogonia differentiation immediately following exit from the hormogonium cycle. Therefore, when these genes are constitutively expressed, we predict the N. punctiforme strain should no longer form hormogonia during co-culture with A. punctatus, leading to a non-infective phenotype. The data are consistent with this model. Mutation of the hrmR gene in strain UCD 472 resulted in a non-infective phenotype, presumably because of unregulated expression of hrmE. Conversely, when hrmR is overexpressed in a multicopy plasmid, as in strain UCD 328 carrying pSCR7, the resultant strain is highly infective and susceptible to repeated rounds of hormogonia differentiation (Cohen and Meeks, 1997), presumably as a result of the complete repression of hrmE transcription and possibly other genes. Under these conditions, an excess of HrmR could exceed the concentration of the in vivo activator ligand. Based on this model, a hrmE mutant should also be highly infective and lack the immunity period of hormogonium differentiation. We are currently testing this prediction.

Since the initial isolation of a high infection mutant within the hrmA gene (Cohen and Meeks, 1997) and our subsequent sequencing of that region of the chromosome, it was perplexing to discover that the hrm locus consists of genes organized in an operon-like topology and with sequence similarly to those of hexuronic acid metabolism. When these sugars are added during co-culture experiments with N. punctiforme and A. punctatus, they neither increase nor decrease the frequency of infection by wild-type or mutant strains, indicating the lack of a direct regulatory role in development and behaviour. Moreover, externally supplied hexuronic acids do not support dark heterotrophic growth of N. punctiforme, indicating that it is not capable of catabolising these compounds for carbon and energy. More subtly, the amounts of mRNA that accumulate from the entire hrm locus are extremely low, based on how much total RNA and exposure time is required to detect a band on Northern hybridizations and primer extensions. This low level of expression is not consistent with a major catabolic role and is in contrast to high level expression of genes known to encode catabolic enzymes, such as those of the opc operon involved in glucose-dependent dark growth of N. punctiforme (Summers et al., 1995).

What then is the function of these gene products? We hypothesize that they are part of a complex metabolic pathway that generates a signal metabolite involved in repression of hormogonium differentiation. The substrates, intermediates and end-product of the metabolic pathway need not be identical to those predicted based on the sequence similarities. There is precedence for an autogenic factor that inhibits hormogonium differentiation (Herdman and Rippka, 1988). The identity of this compound and its relationship to the hrm locus have yet to be determined. It is also intriguing to contemplate when and how N. punctiforme allowed conversion of these genes from a catabolic to a regulatory role. In conclusion, we reinforce this cautionary observation in the current age of genome sequencing and prediction of metabolic role from comparative sequence analysis: gene sequence homology alone is not sufficient to deduce gene product function.

Experimental procedures

Cultures and culture conditions

Nostoc punctiforme strain ATCC 29133 (PCC 73102) and all of the mutants derived from it were grown in standard minimal medium as described previously ( Enderlin and Meeks, 1983 ). Antibiotic supplementations when necessary were at the following final concentrations: Ampicillin (Ap) at 10 µg ml −1 , neomycin (Nm) at 10 µg ml −1 , erythromycin (Em) at 15 µg ml −1 . All E. coli cultures were grown in Luria–Bertani (LB) broth or plates with antibiotics at the following concentrations: Ap at 100 µg ml −1 , kanamycin (Km) at 50 µg ml −1 , chloramphenicol (Cm) at 30 µg ml −1 . The hornwort A. punctatus was cultured and the symbiosis reconstituted as previously described ( Enderlin and Meeks, 1983 ). The number of symbiotic colonies in the gametophyte tissue was scored by macroscopic examination after 2 weeks of co-culture ( Cohen and Meeks, 1997 ).

Construction, purification, and renaturation of r-HrmR

Primers P108 and P109 (all primers used in this study are listed in Table 1) were used to PCR amplify a 1.06 kb fragment containing the hrmR gene with a HindIII site on one end to facilitate its cloning into the protein expression vector pET28a (Novagen), digested with Ecl136II and HindIII and ligated into pET28a, thereby creating pSCR221. Automated sequencing was used to verify sequence fidelity of the PCR fragment and the recombinant protein (r-HrmR) was produced in E. coli strain BL21-DE3 (Novagen), following electroporation of pSCR221 to create strain UCD 497. r-HrmR contains a T7-tag as well as a His6-tag at the N-terminus of the protein near the predicted helix–turn–helix DNA binding motif of the protein. The majority of r-HrmR was present in inclusion bodies and was purified and refolded as follows. Escherichia coli strain UCD 497 was inoculated into Km-supplemented LB broth and grown exponentially at 37°C to an OD600 of 0.6. One millimolar IPTG was added for 2 h to induce production of the r-HrmR protein. Cells were then harvested by centrifugation at 3500 g. The pellets were stored at − 20°C. Two ml of break buffer (50 mM Tris pH 7.5, 0.5 M NaCl, 5 mM DTT) was added to thawing pellets and the cells were broken in a French press at 15 000 p.s.i. The inclusion bodies were collected by centrifugation at 16 000 g for 15 min at 4°C and washed twice with break buffer followed by solubilization in 100 µl of guanidine buffer (50 mM Tris pH 7.5, 6 M guanidine hydrochloride, 25 mM DTT) on ice for 1 h. The insoluble material was removed by centrifugation at 16 000 g for 15 min at 4°C. Bradford protein assay was used to determine the protein concentration of the supernatant and the concentration was then adjusted to 1.85 mg ml−1 using guanidine buffer. To renature, 10 µl of the protein solution in guanidine buffer was quickly diluted into 100 µl of cold buffer (12 mM Hepes, 4 mM Tris, 70 mM KCl, 1 mM DTT, 1 mM EDTA, 30 mM MgCl2, 12% glycerol, pH 7.9) and vortex mixed vigorously for 30 s. The renatured proteins were left on ice for 1 h and then either used immediately or stored at − 80°C for later use.

Electrophoretic mobility shift assays (EMSA)

All target DNA molecules were end-labelled using Klenow fill-in reactions according to manufacturer's instructions and an appropriate 32P-labelled nucleotide for the end filling reaction. Thus, all DNA templates used contained 5′ overhangs for this purpose. The end-labelled fragments were processed through Sephadex G-50 columns (spun columns) to remove unincorporated nucleotides. Radioactivity of an aliquot from each reaction was determined to calculate the amount of template to be added in each EMSA reaction (see below).

Fragment RS2 (containing the HrmR binding site near the hrmR gene) used for EMSA was generated by PCR using primers P117 and P124, yielding a 79 bp fragment which was then digested with MfeI to leave a 5′ overhang on one end of the fragment. Fragment RB2 (containing one of the HrmR binding sites near the hrmE gene) was also generated by PCR using primers P164 and P165 and the 89 bp fragment was digested with AccI to generate a 5′ overhang. Fragment RE2 (containing both binding sites near hrmE) was generated by PCR using P164 and P168 to obtain a 218 bp fragment, which was then incubated with AccI. Smaller DNA templates for EMSA were made by annealing two complementary DNA oligomers. Fragment RSL was generated by heating equal amounts of P153 and P154 at 65°C for 5 min and cooled at room temperature to allow annealing. Similarly, fragment RSR was generated by annealing P155 and P156. Fragment RSL encompasses + 1 to + 34 relative to the hrmR putative transcriptional start point. Fragment RSR extends from + 28 to + 60. The two fragments together encompass the same region as fragment RS2.

EMSA binding reactions contained 100 ng µl−1 sheared salmon sperm DNA, 100 µg ml−1 BSA, 10% glycerol, EMSA buffer (12 mM Hepes, 4 mM Tris, 700 mM KCl, 1 mM DTT, 1 mM EDTA, 30 mM MgCl2, pH 7.9), and 4 µl of refolded protein (estimated to be at approximately 85% r-HrmR by SDS-PAGE; thus, 4 µl equals 673 ng of total protein of which 572 ng is r-HrmR protein, equivalent to 13.5 pmoles r-HrmR per binding reaction). The above mixture was incubated on ice for 10 min, then 40 000–60 000 c.p.m. of end-labelled DNA template was added and allowed to bind at room temperature for 25 min. In DNA competition experiments unlabelled DNA was added at the same time as the 32P-labelled DNA at the concentrations indicated on the figures. To show that it was specifically r-HrmR which was binding to the DNA template, 1 µl of T7-tag antibody (Novagen) was added to the binding reaction when indicated. After the binding reaction, samples were quickly loaded onto a 5% non-denaturing polyacrylamide gel with 0.25% glycerol tolerant buffer (Unites States Biochemical, 216 g Tris, 72 g taurine and 4 g Na2EDTA.2H2O per litre at 20×) for separation. Gels were then dried and exposed to phosphorimager screens and images were analysed using Image Quant software (Molecular Dynamics) and then manipulated using Canvas 7 (Deneba) for publication.

Cell extracts used in EMSA experiments were prepared as follows. Ten ml of N. punctiforme strain UCD 472 (hrmR) culture, approximately 1 month old, were incubated overnight (approximately 16 h) with either 0.4 mM naringin or HRF (Cohen and Meeks, 1997) or with no additions as controls. Cells were then harvested by centrifugation at 5 000 g for 5 min and the pellets suspended in 1.0 ml of EMSA buffer and French pressed as above to break the cells. The cell extract was clarified by centrifugation at 16 000 g for 15 min at 4°C. A volume of 6 µl of each of the supernatants was used per binding reaction. To verify inducing activity of naringin and HRF preparations, a parallel culture of strain UCD 328 (containing luxAB fused with the hrmA gene) was incubated with naringin or HRF and inducing activity was monitored using a luminometer (Cohen and Meeks, 1997). The effect of various sugars on the binding ability of HrmR, using the RS2 DNA template, was tested by including 80 mM of each of the following sugars in separate binding reactions: glucose, sucrose, galacturonate, gluconate, glucuronate and fructose.

RNA hybridizations and primer extensions

Northern hybridizations were prepared as described previously (Summers et al., 1995) except using 40 µg RNA per lane. The hrmE probe was generated by PCR using primers P166 and P167 to produce a 815 bp internal fragment, which was then random primed (kit purchased from Gibco Life Sciences) in the presence of 32P-dCTP. The hrmR probe was generated by PCR using P108 and P109 and labelling as with the hrmE probe. Hybridization was under standard stringent conditions as recommended by the manufacturer of Gene Screen Plus (DuPont NEN).

Primer extensions were performed as previously described (Summers and Meeks, 1996). The hrmR primer extension was performed using primer P120 and 100 µg each of T0 RNA and T6 h + HRF RNA. The hrmE primer extension was initiated using primer P168 (defined earlier) and 100 µg of RNA from each of T0 and T1.5 h + HRF. The sequencing ladders for each respective primer extension was generated using the same primer as in the primer extension reaction, the T7 Sequenase V.2.0 sequencing kit (USB) and 35S-sequetide from DuPont NEN.

Construction of the hrmR mutant

A 4.6 kb XbaI/BamHI fragment containing part of the hrm locus from pSCR18 (Cohen and Meeks, 1997) was cloned into pBluescript KS+ (Stratagene) to create pSCR35. pSCR35 was then partially digested with MfeI and a 1.15 kb EcoRI fragment from pRL448, containing the PpsbA-npt gene (Elhai and Wolk, 1988), was ligated into the compatible site. Plasmids were then screened for insertion into the MfeI site that interrupts the hrmR gene. The fragment containing the hrm locus with the insertionally inactivated hrmR gene was excised with SpeI and XhoI and ligated into the SacB recombination vector pRL271 (Black et al., 1993). The resulting construct was checked for sucrose sensitivity in E. coli prior to triparental mating into N. punctiforme using a standard protocol (Cohen et al., 1994). Single and double recombinants in N. punctiforme were screened initially using antibiotic resistance as previously described (Cohen et al., 1994); their genotypes were verified by Southern hybridization of chromosomal DNA from the mutants.

Sequence and acquisition numbers

Double stranded DNA sequencing was performed by contract using ABI Prism Sequencers and the dye termination method. The sequence of hrmUA was previously deposited in GenBank (L37087). The GenBank accession number for hrmR, hrmK, and hrmI is AF266466. hrmE is gene 9 of Contig 338 in the May 31, 2001 assembly of the N. punctiforme genome (http:www.jgi.doe.gov/).

Acknowledgements

This work was supported by grant IBN 0080805 from the US National Science Foundation. We thank Nathan Heitzmann and Ju Guan for assistance in cloning, Michael Cohen for a gift of naringin, Dr Martin Privalsky for helpful advice on EMSA experiments and Ian Campbell for computer assistance in finding the hrmE binding sites.

Ancillary