Functional dissection and evidence for intercellular transfer of the heterocyst-differentiation PatS morphogen


  • Laura Corrales-Guerrero,

    1. Instituto de Bioquímica Vegetal y Fotosíntesis, Consejo Superior de Investigaciones Científicas and Universidad de Sevilla, Seville, Spain
    Search for more papers by this author
  • Vicente Mariscal,

    1. Instituto de Bioquímica Vegetal y Fotosíntesis, Consejo Superior de Investigaciones Científicas and Universidad de Sevilla, Seville, Spain
    Search for more papers by this author
  • Enrique Flores,

    1. Instituto de Bioquímica Vegetal y Fotosíntesis, Consejo Superior de Investigaciones Científicas and Universidad de Sevilla, Seville, Spain
    Search for more papers by this author
  • Antonia Herrero

    Corresponding author
    • Instituto de Bioquímica Vegetal y Fotosíntesis, Consejo Superior de Investigaciones Científicas and Universidad de Sevilla, Seville, Spain
    Search for more papers by this author

For correspondence. E-mail Tel. (+34) 95 448 9522; Fax (+34) 95 446 0065.


The formation of a diazotrophic cyanobacterial filament represents a simple example of biological development. In Anabaena, a non-random pattern of one nitrogen-fixing heterocyst separated by about 10 photosynthetic vegetative cells results from lateral inhibition elicited by the cells differentiating into heterocysts. Key to this process is the patS gene, which has been shown to produce an inhibitor of heterocyst differentiation that involves the C-terminal RGSGR pentapeptide. Complementation of a ΔpatSAnabaena mutant with different versions of PatS, including point mutations or tag fusions, showed that patS is translated into a 17-amino acid polypeptide. Alterations in the N-terminal part of PatS produced inhibition of heterocyst differentiation, thus this part of the peptide appears necessary for proper processing and self-immunity in the producing cells. Alterations in the C-terminal part of PatS led to over-differentiation, thus supporting its role in inhibition of heterocyst differentiation. A polypeptide, produced in proheterocysts, consisting of a methionine followed by the eight, but not the five, terminal amino acids of PatS recreated the full activity of the native peptide. Immunofluorescence detection showed that an RGSGR-containing peptide accumulated in the cells adjacent to the producing proheterocysts, illustrating intercellular transfer of a morphogen in the cyanobacterial filaments.


In the living world, multicellularity appears to have arisen several times during the course of evolution (Bonner, 1998). Communal behaviours that provide benefit to a group involving communication, cooperative behaviour and molecular exchange between cells are frequent in bacteria (Dunny, 2008; Manteca and Sanchez, 2009). Nevertheless, filamentous heterocyst-forming cyanobacteria represent a further step in multicellularity based on a structured supracellular organization including different interdependent cell types, each specialized in a different nutritional task, which contribute to the growth of the filament as the organismic unit (Flores and Herrero, 2010; Kumar et al., 2010). This organization can be traced back at least 2 billion years, before multicellularity appeared in eukaryotes (see Tomitani et al., 2006; Schirrmeister et al., 2011).

Heterocyst differentiation represents a remarkable solution for the co-occurrence of the incompatible processes of oxygenic photosynthesis and N2 fixation. Cyanobacteria ancestors developed oxygenic photosynthesis and played a crucial role in the evolution of the biosphere, including the oxidation of the Earth's atmosphere 2.22–2.45 Ga and the establishment of symbiotic associations leading to the development of algal and plant plastids (Knoll, 2008). While endorsing the cyanobacterial ancestors with a unique trait of using sunlight and water to generate energy for living, the development of oxygenic photosynthesis represented a burden for the use of atmospheric N2 as a source of nitrogen for growth due to the general sensitivity of the nitrogenase system to oxygen. In some filamentous cyanobacteria some of the cells in the filament differentiate into specialized cells, called heterocysts, where the N2-fixation system is confined, so that the vegetative cells perform oxygenic photosynthesis driving photosynthetic CO2 fixation, whereas heterocysts rely on a mainly heterotrophic metabolism to carry out the fixation of atmospheric N2. Vegetative cells provide heterocysts with reduced C to serve as source of energy and reductant, whereas heterocysts transfer combined nitrogen to the vegetative cells (Wolk et al., 1994; Flores and Herrero, 2010).

Heterocyst differentiation requires the action of a number of regulators, being pivotal the N-control transcription factor NtcA, a member of the CRP-family of regulators, and the differentiation-specific protein HetR, which is also a transcription factor (Flores and Herrero, 2010; Kumar et al., 2010; Kim et al., 2011; Herrero et al., 2013). Heterocysts are found in a non-random distribution along the filament (Wolk et al., 1994). In the case of strains of the genera Anabaena or Nostoc, heterocysts are distributed at semi-regular intervals separated by c. 10–15 vegetative cells. Nutritionally, this pattern appears to respond to the constraints imposed by the limited number of vegetative cells that can be fed by each heterocyst and the limited number of cells that can be dispensed of contributing to CO2 fixation. When filaments growing with combined nitrogen are subjected to nitrogen step-down, the pattern is established early, much before heterocyst maturation has been completed (Yoon and Golden, 2001). Thus, during this developmental process, positional information is generated before N2 fixation has started.

A product of the patS gene has been implicated in establishing the spatial pattern of heterocyst distribution after nitrogen deprivation (Yoon and Golden, 1998; 2001). The gene is induced early in the differentiating cells and its inactivation leads to a Mch (Multiple contiguous heterocysts) phenotype, whereas its overexpression abolishes differentiation. In Anabaena sp. strain PCC 7120 the patS gene would encode a 17- or 13-amino acid peptide of which a subset consisting of the 5 C-terminal amino acids (RGSGR, a peptide called PatS-5) inhibits differentiation when added to the external medium, although it fails to restore a wild-type heterocyst pattern on a patS mutant (Yoon and Golden, 1998). In vitro, PatS-5 has been shown to interact with HetR and to inhibit binding of HetR to DNA upstream from several regulated genes (Huang et al., 2004; Risser and Callahan, 2007; Feldmann et al., 2011; Du et al., 2012). It is considered that PatS, or a processed derivative of it, is exported from the differentiating cells to inhibit the differentiation of its neighbours likely by acting on the transcription factor HetR (Yoon and Golden, 1998; Khudyakov and Golden, 2004; Risser and Callahan, 2009). However, information about the identity of the primary gene product and of the molecule actually acting as a trans-cellular inhibitor of differentiation, thus behaving as a morphogen, as well as about the in vivo processing, is lacking.

In this work we present evidence for the synthesis of a 17-amino acid PatS in (pro)heterocysts. Adhering to definitions by C. P. Wolk, in Anabaena, a proheterocyst is an intermediate, between a vegetative cell and a heterocyst, which differs in shape and granulity from vegetative cells, and a mature heterocyst is a cell capable of aerobic fixation of N2 (see Wolk et al., 1994). We also show that the C-terminal octapeptide, but not the C-terminal heptapeptide, hexapeptide or pentapeptide, of PatS is sufficient for regulation of heterocyst patterning along the filament. Specific residues in the N-terminal part of PatS are crucial for self-immunity in the producing cells, likely reflecting their role in PatS processing and/or export, whereas specific C-terminal amino acids determine the heterocyst differentiation inhibitory activity of the peptide. Additionally, using an immunofluorescence approach, we have detected the in vivo location of PatS.


To better mimic the physiological conditions, we have chosen to express a number of versions of the patS gene from the native patS promoter in the chromosome (Fig. 1). In support of this approach, a recent report documents an extensive copy-number variability of heterologous replicons commonly used in strain PCC 7120 (Yang et al., 2012). The engineered patS gene was transferred by conjugation from Escherichia coli to Anabaena sp. strain CSVT20, a derivative of strain PCC 7120 in which the wild-type patS gene has been deleted. In this way, strains in which only the altered version of patS is present were generated (Table 1). Because strain CSVT20, like other patS deletion strains (Yoon and Golden, 1998), produces a Mch phenotype, which correlates with a disinhibition of heterocyst formation, this approach permitted a simple test of phenotypic complementation. As shown below, three basic phenotypes were observed with the altered patS genes: lack of complementation retaining the ΔpatS Mch phenotype, complementation resulting in a wild-type phenotype, and complete or partial inhibition of heterocyst differentiation.

Figure 1.

Scheme of the genetic constructs used in this work. A 180 bp DNA fragment containing the patS gene was deleted from the chromosome of Anabaena sp. strain PCC 7120 to produce strain CSVT20. The latter was used as parental strain to introduce a wild-type or an altered patS gene (patS*). A DNA fragment containing 672 bp of the region 5′ from patS and the patS* sequence was transferred, cloned in plasmid pCSV3 that lacks an origin of replication for Anabaena sp., to strain CSVT20. Single recombinants that had inserted the construct in the chromosome by a single recombination event were selected as SmR/SpR clones, which represented the different CSL strains.

Table 1. Phenotype of Anabaena sp. PCC 7120 mutants producing different PatS peptides
StrainMutant version of PatSHeterocysts (%)Mean interval sizeContiguous heterocysts (%)
  1. The amino acid sequence of the PatS peptide produced by the altered patS gene present in each strain is shown (the wild-type peptide sequence is MKAIMLVNFCDERGSGR). When appropriate, the location of the inserted GFP protein or 6-His tag is indicated between brackets. In the case of substitution mutations, the amino acid introduced is written in bold. All the mutant strains were generated using strain CSVT20 as parental. Heterocyst frequency (as percentage of total cells) is indicated for each strain. The mean size of vegetative cell intervals between heterocysts and the percentage of contiguous heterocysts (interval size = 0, percentage of total intervals) have been calculated from the data in Fig. 3.
PCC 7120 10.31 ± 0.649.67 ± 0.264.36 ± 0.58
CSVT20 Δ [180 bp]19.00 ± 2.125.69 ± 0.3820.81 ± 5.13
CSL22MKAIM[H6]LVNFCDERGSGR5.31 ± 0.7613.18 ± 0.592.27 ± 2.27
CSL45MRGSGR20.29 ± 1.585.59 ± 0.4711.30 ± 0.96
CSL46MERGSGR13.79 ± 1.127.40 ± 0.3621.11 ± 2.92
CSL94MDERGSGR10.01 ± 0.6111.12 ± 0.5512.46 ± 3.99
CSL47MCDERGSGR9.02 ± 0.8510.83 ± 0.514.59 ± 0.09
CSL48MKAIMLVNFCDERGSGR12.39 ± 0.7010.40 ± 0.386.01 ± 0.66
CSL44AKAIMLVNFCDERGSGR18.67 ± 0.695.55 ± 0.0125.55 ± 6.09
CSL49MKAIALVNFCDERGSGR10.56 ± 1.3010.69 ± 0.786.73 ± 3.05
CSL50MKAIMQVNFCDERGSGR10.09 ± 0.7610.83 ± 0.293.75 ± 2.15
CSL62MKAIVLQNFCDERGSGR6.01 ± 1.8217.63 ± 1.680
CSL52MKAIMLVEFCDERGSGR8.35 ± 1.0810.32 ± 0.531.33 ± 1.33
CSL53MKAIMLVNQCDERGSGR7.45 ± 1.6514.01 ± 0.970
CSL54MKAIMLVNFADERGSGR10.24 ± 0.719.85 ± 0.2411.15 ± 3.08
CSL55MKAIMLVNFCQERGSGR12.02 ± 2.059.68 ± 0.774.76 ± 2.10
CSL56MKAIMLVNFCDARGSGR11.46 ± 0.928.12 ± 0.219.08 ± 4.88
CSL57MKAIMLVNFCDEAGSGR22.75 ± 2.334.67 ± 0.1225.16 ± 5.24
CSL59MKAIMLVNFCDERASGR15.30 ± 2.387.47 ± 0.468.08 ± 1.82
CSL58MKAIMLVNFCDERGAGR24.62 ± 3.794.85 ± 0.4624.27 ± 4.90
CSL60MKAIMLVNFCDERGSAR16.92 ± 1.524.98 ± 0.1726.17 ± 4.03
CSL61MKAIMLVNFCDERGSGA16.03 ± 1.568.14 ± 0.219.75 ± 1.98

Expression of PatS-17

As mentioned above, the patS ORF could encode a 17- or 13-amino acid peptide, which depends on whether translation starts at the first or the second possible Met-encoding codon. We have studied translation initiation of the patS gene monitoring fluorescence emission from strains bearing the gfp-mut2 gene (lacking its start and stop codons) inserted in phase after the 1st (strain CSL21) or the 5th (strain CSL19) triplet of the patS ORF. As a reference of the time-course activation of the patS promoter, we monitored GFP fluorescence from strain CSVM17 (Mariscal et al., 2007), a derivative of strain PCC 7120 bearing a transcriptional PpatS-gfp fusion inserted in a neutral region of the Anabaena α megaplasmid, which has the same copy number as the chromosome (Lee et al., 2003). The GFP fluorescence was monitored in filaments grown with ammonium and subjected to combined-nitrogen deprivation. Strains CSL21 and CSL19 both produced GFP fluorescence, localized to specific cells (Fig. 2), indicating that the first Met codon can be used as a translation start of the patS gene.

Figure 2.

Expression of PatS–GFP translational fusions. Anabaena producing a polypeptide with the GFP inserted after the first (strain CSL21) or the second (strain CSL19) M residue of PatS or, as a reference, a PpatS-gfp transcriptional fusion in the strain PCC 7120 α megaplasmid (CSVM17) were grown in ammonium-containing medium and incubated in the absence of combined nitrogen for the indicated times. The filaments were visualized by confocal microscopy. Merged bright field/green GFP fluorescence micrographs are shown.

Phenotype of strains bearing GFP- and 6His-tag-fusions

In addition to the strains producing PatS–GFP fusions, we studied a strain derived from the patS mutant CSVT20 that bears a modified patS gene encoding a PatS version with a 6His sequence inserted between M5 and L6 of the polypeptide (strain CSL22). Phenotypic characteristics of strains CSL19, CSL21, and CSL22 with regard to the frequency and distribution of proheterocysts plus heterocysts, expression of nitrogenase activity and ability to grow diazotrophically were studied using strains PCC 7120 and CSVT20 as controls (Tables 2 and 2). In all the mutant strains generated in this work, proheterocysts and heterocysts were visualized by Alcian Blue staining that colours the Hep layer of the heterocyst envelope (see López-Igual et al., 2012b). The data of percentage of total heterocysts, with regard to total cells, and of contiguous heterocysts, as well as the distribution of sizes of vegetative cell intervals between heterocysts, are shown in Fig. 3 and summarized in Table 1.

Figure 3.

Heterocyst distribution in Anabaena sp. strain PCC 7120 and mutant strains. Except for strains PCC 7120 and CSVT20 (ΔpatS), the PatS polypeptide produced in each strain is indicated. In the PatS polypeptides containing point amino acid substitutions, the introduced amino acid(s) are highlighted. In strain CSL22, the diamond denotes the place of insertion of a 6-His sequence in the PatS polypeptide. The percentage of heterocysts formed by each strain is also indicated. Heterocyst distribution could be analysed accurately only in strains showing more than 5% heterocysts. The micrograph shows filaments of strain PCC 7120 in which proheterocysts and heterocysts appear blue after Alcian Blue staining.

Table 2. Phenotypic characteristics of Anabaena sp. PCC 7120 wild-type strain (PCC 7120), a patS deletion strain (CSVT20) and mutants producing PatS-tagged peptides
StrainNitrogenase activity nmol (μg chlorophyll a)−1 h−1Growth rate (day−1)
  1. Data are the mean and the standard deviation of the mean of the results of two to four experiments performed with independent cultures (nitrogenase) or of four independent cultures (growth rates). Genotypes are indicated in Table 1.
PCC 712019.69 ± 0.680.53 ± 0.020.45 ± 0.01
CSVT20 12.02 ± 0.860.49 ± 0.040.36 ± 0.03
CSL190.02 ± 0.010.58 ± 0.040.09 ± 0.03
CSL210.63 ± 0.100.53 ± 0.040.16 ± 0.05
CSL222.46 ± 1.130.64 ± 0.020.28 ± 0.03

Strain CSL21 (GFP inserted after M1) exhibited a (pro)heterocyst frequency of 2.94% of the total cells (the wild type has 10.31% and the control strain CSL48, bearing the wild-type patS sequence construct, has 12.39%) (Table 1), very low nitrogenase activity and impaired diazotrophic growth (Table 2). Strain CSL19 (GFP inserted after M5) exhibited a very low (pro)heterocyst frequency (0.54% of the total cells) and, consistent with a very low nitrogenase activity, a very low diazotrophic growth rate. Strain CSL22 (6His sequence inserted after M5) exhibited a (pro)heterocyst frequency of 5.31% of total cells. In this strain, nitrogenase activity and diazotrophic growth rate were also lower than in the wild type, although the impairment in diazotrophic growth was less pronounced than in CSL21. Thus, insertion of GFP after M1 (strain CSL21) or of 6His (strain CSL22) or GFP (strain CSL19) after M5 leads to impairment in heterocyst differentiation, which is most severe in the latter case. These results are consistent with the idea that the modified forms of PatS expressed in strains CSL19, CSL21 and CSL22 inhibit differentiation of the producing cells.

Point amino acid mutations of patS

Strains expressing versions of PatS with point amino acid substitutions in each position of the peptide were also investigated (Table 1 and Fig. 3). Whereas substitution M5A (present in strain CSL49) had little effect on (pro)heterocyst frequency or distribution with respect to the wild type or the CSL48 control strain, substitution M1A (in strain CSL44) produced a patS phenotype. These results indicate that M1 is necessary and sufficient as a translation start of patS.

In strain CSL90 (K2A) (pro)heterocyst frequency was 4.47% of total cells, whereas in strains CSL91 (A3Q) and CSL92 (I4Q) it was 2.10 and 1.97% respectively. Substitution L6Q (in strain CSL50) had almost no effect on (pro)heterocyst frequency or interval-size distribution. Substitution V7Q (in strain CSL51) abolished differentiation whereas V7A (strain CSL93) led to a (pro)heterocyst frequency of 3.43%. These results can be interpreted in terms of inhibition of differentiation in the cells producing PatS with the indicated changes, except for the one in L6. Additionally, results with strain CSL51 together with the behaviour of strain CSL62 (M5V, V7Q), which showed a frequency of (pro)heterocysts of 6.01%, indicated that V7 has a critical role to prevent an inhibitory activity of PatS in the producing cells, and that this role could be partially served by the presence of a V residue in the 5th position of PatS.

Substitutions N8E (strain CSL52) and F9Q (strain CSL53) also lowered the frequency of (pro)heterocysts (8.35% and 7.45% of total cells respectively) and the frequency of (pro)heterocysts consecutive to another (pro)heterocyst (interval size = 0) (1.33% of the total intervals in strain CSL52 and no doublets detected in strain CSL53; 6.01% in the control strain CSL48). Moreover, in strain CSL53 the mean interval size between heterocysts (14.01 cells) was somewhat higher than in the control strain (10.40 cells). Thus, N8 and F9 appear to influence prevention of inhibition by PatS in the producing cells.

Some amino acid substitutions in the C-terminal part of PatS, including PatS-5, blocked the PatS inhibitory activity. The R13A (in strain CSL57) and S15A (in strain CSL58) changes increased (pro)heterocyst frequency and shortened vegetative cell intervals between (pro)heterocysts to levels comparable to those of strain CSVT20; the G16A change (in strain CSL60) considerably increased (pro)heterocyst frequency (although to a lesser extent than in strains CSL57 or CSL58) and shortened intervals; and the G14A (in strain CSL59) and R17A (in strain CSL61) changes produced increases in pro(heterocyst) frequency similar to that in CSL60 and an alteration in the distribution of (pro)heterocysts less drastic, specially concerning the frequency of contiguous heterocysts, than that in strain CSL60. Substitutions C10A (strain CSL54) and E12A (strain CSL56) increased the frequency of doublets (frequency about 11.15% and 9.08% respectively), whereas substitution D11Q (strain CSL55) had very little effect on the studied parameters. Thus, residues R13, S15 and G16 appear to be essential and G14 and R17, and to a lesser extent C10 and E12, important for PatS activity (i.e. inhibition of heterocyst differentiation).

patS minigenes

To compare the activity of the PatS-5 peptide and that of peptides including the residues that we had found to influence the activity of PatS (i.e. whose mutation led to over-differentiation), strains that produce short versions of PatS consisting of a Met residue plus the C-terminal RGSGR pentapeptide (strain CSL45), ERGSGR hexapeptide (strain CSL46), DERGSGR heptapeptide (strain CSL94) or CDERGSGR octapeptide (strain CSL47) were constructed. As a control, strain CSL48 produces the full 17-amino acid PatS. In contrast to patS minigenes that have been investigated previously, which were provided in a multicopy plasmid (Wu et al., 2004), our constructs were present in the native patS locus in the chromosome (Fig. 1). (Pro)heterocyst frequency and distribution in strain CSL47 were similar to those in CSL48 and the wild-type strains. This observation indicates that the C-terminal octapeptide of PatS has an effect close to that of the full PatS gene product. In contrast, strain CSL45 showed a frequency of (pro)heterocysts and mean size of vegetative cell intervals between (pro)heterocysts similar to those of the patS mutant strain CSVT20, and a frequency of contiguous heterocysts intermediate between those of the control CSL48 and strain CSVT20. In strain CSL46, the frequency of (pro)heterocysts and the mean interval size are between the values of the wild type and the patS mutant, and the frequency of contiguous heterocysts is similar to that of this mutant. Finally, in strain CSL94 the frequency of (pro)heterocysts and the mean interval size are similar to the wild-type values, and the frequency of contiguous heterocysts is between the wild-type and the patS mutant values. Thus, strain CSL45 exhibits low, if any, PatS inhibitory activity, whereas strains CSL46 and CSL94 appear to exhibit partial PatS activity.

Detection of PatS by immunofluorescence

The detection of PatS or PatS-derived peptides in Anabaena sp. strain PCC 7120 was undertaken by immunofluorescence using antibodies raised against synthetic PatS-5 (see Experimental procedures). Ammonium-grown filaments of the wild type and strains CSVT20 and CSL45 were incubated 8 h in the absence of combined nitrogen and treated to detect fluorescence from PatS-5 antibodies. As revealed by Alcian Blue staining (not shown), (pro)heterocysts could already be detected in the culture. The fluorescence obtained from strain CSVT20, which lacks a patS gene, is presented in Fig. 4 as a reference background. Although this strain differentiates heterocysts, no cell exhibiting high fluorescence was observed. In filaments of wild-type strain PCC 7120, fluorescence was detected in a number of cells at levels above the background levels observed in the cells of strain CSVT20. As a representative example, the filament of strain PCC 7120 shown in Fig. 4A bears three cells with the morphology of proheterocysts, in which the fluorescence detected was lower than in their neighbouring vegetative cells. The distribution of fluorescence in proheterocysts and the cells occupying neighbouring positions 1–7 with regard to proheterocysts is shown in Fig. 4B. The data indicates that, on average, in strain PCC 7120 fluorescence in prohetrocysts is similar to the background levels, and it is higher in the cells contiguous to proheterocysts. Indeed, fluorescence appears to decrease gradually from the first to, at least, the fifth neighbour. Because the patS gene is expressed in proheterocysts (Yoon and Golden, 1998; 2001), and indeed in our experiments patS is expressed mainly in single cells at the 8 h time point after N step-down (see strain CSVM17 in Fig. 2), even in the event that (pro)heterocyst labelling with the antibodies were less effective than labelling of vegetative cells, our results show accumulation of a PatS peptide in non-producing cells adjacent to the producing proheterocysts. The observed spatial pattern of fluorescence is consistent with an export of PatS from the producing proheterocyst and further vegetative cell-to-vegetative cell transference. Also, the fact that no cell with fluorescence above the background was observed in strain CSVT20 speaks against detection of other PatS-5-containing proteins (e.g. HetN) under the used conditions.

Figure 4.

Immunofluorescence detection of PatS-5. Filaments of the indicated strains (for genotypes, see Table 1) grown in bubbled ammonium-containing cultures were incubated for 8 h in bubbled medium lacking combined nitrogen and used in an immunofluorescence test with antibodies raised against the RGSGR pentapeptide.

A. Visualization of filaments subjected to immunodetection of PatS-5. Bright field, fluorescence and bright field/fluorescence merge micrographs are shown from left to right. Arrowheads point to proheterocysts.

B. Quantification of immunofluorescence from PatS in filaments of strains PCC 7120 and CSL45. The fluorescence detected (arbitrary units) in proheterocysts (h) and their neighbouring cells (n1 to n7, the cell in position n1 is adjacent to the proheterocyst) in each strain is represented. The values represented are the mean of the fluorescence recorded in each position (301 cells from three different cultures were counted for PCC 7120, and 185 cells for CSL45) less the mean of the fluorescence measured in strain CSVT20 (96 cells), which was used as background level. Bars indicate the standard deviations of the mean. t-tests indicate that in strain PCC 7120 the fluorescence value for the phoheterocysts is different (P < 0.01) than those from their neighbours.

As described above, strain CSL45 produces a version of PatS (Met-PatS-5) that apparently lacks PatS function, since heterocyst differentiation is disinhibited in this strain. Figure 4A shows two short filaments of CSL45 that contain proheterocysts, which in the immunofluorescence analysis showed higher fluorescence than their neighbouring vegetative cells. Thus, in contrast to what happened in the wild type, the Met-PatS-5 peptide produced in strain CSL45 was apparently retained in the producing cells. Quantification of fluorescence in strain CSL45 showed that, on average, the PatS immunofluorescence signal is high in cells that could be identified as proheterocysts and in cells occupying the 1 to 3 neighbouring positions, and considerably lower beyond that position (Fig. 4B). In this strain, the high PatS signal in cell clusters may reflect the pattern of the patS gene expression, consistent with the observed high frequency of contiguous heterocysts (see Table 1). (Results obtained with strain CSL45 also shows that PatS could be already detected in 8 h proheterocysts.)


The patS gene is a key player in the establishment of the spatial pattern of development that takes place in response to nitrogen deprivation in the filaments of heterocyst-forming cyanobacteria, such as the model organism Anabaena sp. strain PCC 7120. Despite a considerable amount of recent work dealing with the effects of, mainly, the pentapeptide corresponding to the 5-residue C-terminal fragment of PatS, PatS-5, little is known about how PatS is processed and which is the actual molecule active as a morphogen in vivo. In this work, we have constructed a number of Anabaena strains expressing, as the only version and from the native patS promoter in the chromosome, variants of the patS gene. These include genes encoding PatS with a 6His-tag or GFP insertions or with amino-acid substitutions and minigenes encoding different extents of the C-terminal part of PatS.

The first and fifth codons of the strain PCC 7120 patS ORF (Kaneko et al., 2001) are Met-encoding codons, which until now made it uncertain the identity of the translation start of the patS gene product. Our results indicate that the first of these codons is the principal patS translational start. First, when gfp is inserted in frame after the first Met-encoding codon (in strain CSL21), GFP fluorescence is observed in cells with a specific distribution along the filament, indicating that M1 is readily used as a translational start. Second, the change M1A (in strain CSL44) results in a phenotype of percentage and distribution of (pro)heterocysts similar to that of the patS null mutant (strain CSVT20), whereas substitution M5A (in strain CSL49) produces little changes compared with the wild type, indicating that M1, but not M5, is required for PatS function. Finally, strain CSL21 shows a considerable decrease in heterocyst frequency and nitrogenase activity as compared with the wild type, indicating that alteration of the amino acid stretch preceding M5 by introduction of the GFP interferes with PatS function. Due to its small size, the presence of patS genes in genomic sequences is difficult to predict in the absence of experimental analysis. However, searching for RGSGR followed by a stop codon in the chromosomes of 13 heterocyst-forming cyanobacteria other than PCC 7120 translated in all six reading frames (J. Elhai, pers. comm.) identified putative PatS peptides in 10 of them (annotated only in one case, that of Nostoc punctiforme, although the peptide of Anabaena variabilis would be identical to that of PCC 7120). Besides PCC 7120 and A. variabilis, in two other strains PatS would be 17-amino acid peptides with M1 and M5; three strains could produce 15-amino acid peptides with only one M start codon, and four other could produce longer PatS peptides, two of them with one Met aligning to M5 of PCC 7120. Thus, a considerable variability in the N-terminal part of putative PatS peptides among heterocyst-forming cyanobacteria is apparent, with a number of strains putatively producing PatS peptides with extended N-terminal fragments.

Results from the study of strains producing PatS with insertions or amino acid substitutions lead us to consider two different regions of the 17-residue peptide in Anabaena sp. strain PCC 7120. In the N-terminal region, substitutions in the second (in strain CSL90), third (in CSL91) or fourth (in CSL92) positions of PatS lead to a decrease in (pro)heterocyst frequency, which is more severe in the two latter cases (Fig. 5). The changes in V7 have strong effects on differentiation: its substitution by A (in strain CSL93) impairs differentiation, and its substitution by Q (in CSL51) abolishes differentiation, although this can be compensated in part by introducing a V residue in the 5th position of the peptide (in CSL62). These results imply that the modified versions of PatS produced in strains CSL90, CSL91, CSL92, CSL93 and CSL51, as is also the case for those produced in CSL19, CSL21 and CSL22, inhibit the differentiation of the producing cells into heterocysts. Although to a lesser extent, also the changes in N8 (strain CSL52) and F9 (strain CSL53) inhibit differentiation. Whatever the mechanism the wild type uses for self-immunity of the producing cells against PatS, it appears impaired in these mutant strains. In contrast, the PatS inhibitory activity is retained in them. Thus, the 9-, and specially the 7-residue N-terminal part of PatS is required for self-immunity of the producing cells but not for PatS activity.

Figure 5.

The patS gene is expressed as a 17-residue polypeptide including an N-terminal part involved in PatS processing and a C-terminal part required for PatS regulatory function. An Anabaena strain carrying a mutation of a residue labelled black expresses a phenotype similar to the wild-type phenotype. A strain carrying a mutation of a residue labelled red exhibits less heterocyst differentiation than the wild type. A strain carrying a mutation of a residue labelled green exhibits more heterocyst differentiation than the wild type.

In the C-terminal region of PatS, whereas C10, E12, G14 and R17 appear important, R13, S15 and G16 are essential for the inhibitory function of PatS in cells adjacent to the producing cells (Fig. 5). The hetN gene is also needed for wild-type heterocyst patterning in Anabaena sp. strain PCC 7120, although its function is evident in filaments growing steadily on N2 rather than in filaments subjected to nitrogen deprivation (Callahan and Buikema, 2001). A HetN-related product would be produced by the mature heterocysts to inhibit neighbouring cells from differentiating. The 287-residue HetN polypeptide bears an internal ERGSGR sequence, of which RGSGR has been shown to be essential for heterocyst patterning, and in which the two R residues are essential to suppress heterocyst differentiation (Higa et al., 2012). Thus, some similarities may exist between the mechanism of action of PatS and that of HetN (Khudyakov and Golden, 2004).

As mentioned above, it has been proposed that a PatS-related product moves from the producing proheterocyst to the adjacent cells. We have detected by immunofluorescence PatS-5 showing an accumulation of a PatS-5 containing product in the cells adjacent to proheterocysts (Fig. 4). This represents a direct evidence for intercellular transfer of a PatS peptide. In other bacteria, peptides that have a role in the control of developmental processes have been identified, and in some cases the biological function depends on events of peptide transport in and out of the cells. A number of examples have been reported for Bacillus peptides involved in the regulation of sporulation, quorum sensing and virulence, such as PhrA and CSF pentapeptides that are produced by processing of precursor molecules and secretion (Lazazzera et al., 1997). The processed forms of both peptides are transported into the cell by the oligopeptide permease SpoOK (OPP) (Perego and Hoch, 1996; Solomon et al., 2011). PhrA is synthesized as a 44-amino acid molecule with an N-terminal hydrophobic domain and a C-terminal hydrophilic region separated by a potential signal peptidase I cleavage site (which would correspond to a V, H or A residue), implying that the carboxyl half of the protein is secreted from the cell (Perego and Hoch, 1996). In the case of CSF, an important role of the region comprising the 5 amino acids preceding the active pentapeptide in directing cleavage of the precursor protein has been reported (Laningan-Gerdes et al., 2008). In the case of PatS, consistent with its small size, no signal peptide could be recognized using different prediction programs (e.g. SignalP 4.0; Petersen et al., 2011), and no obvious matches to the pattern of signal peptidase I (Paetzel et al., 2002) could be recognized either. Thus, PatS would be exported from the producing proheterocysts by a novel mechanism, which may involve direction to the periplasm from which it could be imported by neighbouring vegetative cells, or direct cell-to-cell transfer through channels located in the intercellular septa (Flores and Herrero, 2010).

The possibility that PatS were cleaved to generate an active smaller peptide was consistent with all available data, but not finally proved. Previously, patS4 to patS8 minigenes were expressed in Anabaena from promoters with different cell specificity (Wu et al., 2004). When patS5 was expressed in proheterocysts in the patS background, from either the patS promoter or the (pro)heterocyst-specific hepA promoter, no or only a weak inhibition could be detected. These results are consistent with those obtained with strain CSL45 described above, confirming that PatS-5 expressed in proheterocysts is able to produce neither inhibition in the producing cells nor cell–cell signalling. The fact that similar results were obtained when the PpatS-patS5 construct was placed either in the chromosome (this work) or in a multicopy plasmid (Wu et al., 2004) makes a low peptide dosage an unlikely reason to explain lack of activity of the produced PatS-5. Moreover, we have shown that in strain CSL45 PatS-5 is accumulated in the producing proheterocysts (Fig. 4). In contrast, PatS-8 (in strain CSL47), which includes all the residues that our results have implicated in the inhibitory activity of PatS, appears to produce a regulation similar to that produced by wild-type PatS. When expressed from the copper-inducible petE promoter, which is active mainly in vegetative cells, patS8, patS7, patS6 and patS5, but not patS4, produced inhibition of heterocyst differentiation both in the wild type and a patS-null mutant (Wu et al., 2004). Expression of patS or patS5 from the rbcL promoter, which directs strong expression only in vegetative cells, completely inhibits heterocyst formation both in the wild type and the patS mutant (Wu et al., 2004). Additionally, PatS-5 added to the culture medium inhibits heterocyst differentiation (Yoon and Golden, 1998; Wu et al., 2004). Thus PatS-5, which is not active in the proheterocysts and would not be exported from these cells, inhibits heterocyst differentiation when placed in the vegetative cells. It appears that, in contrast to the receptor vegetative cells, the producing proheterocysts are insensitive to the small processed peptides whereas, as deduced from the phenotype of strains expressing peptides with point substitutions in the N-terminal 9-amino acid stretch, they are sensitive to the unprocessed PatS. In this regard, whether the processed peptides are accumulated at different steps of the differentiation process in the producing and receptor cells, and whether the mechanism of inhibition by the unprocessed and processed peptides is different will merit future investigation.

Several reports have shown in vitro activity of synthetic PatS-5 at binding to HetR and inhibition of HetR binding to DNA (Huang et al., 2004; Risser and Callahan, 2007; Feldmann et al., 2011; Du et al., 2012). Binding to HetR was tighter for PatS-6 (ERGSGR) than for PatS-5, and binding was tighter for PatS-5 than for PatS-7 (DERGSGR), whereas no binding was detected for PatS-8 (CDERGSGR) (Feldmann et al., 2012). Thus, some differences could be found between the relative affinities of the different peptides for in vitro binding to HetR and for their activity of inhibition of heterocyst differentiation. The most striking of these differences is that found with PatS-8, which does not appreciably bind to HetR in vitro, but appears to be fully active in strain CSL47. It is possible that: (i) the affinity for a molecular target (HetR) is influenced by additional conditions or factors in vivo; (ii) the molecule exhibiting a tighter binding, or even a higher in vivo activity when over-produced in the cyanobacterium (e.g. specifically in vegetative cells), is not the best suited for a regulatory effect in vivo; (iii) the presence of a Met residue in the in vivo-produced, but not in the in vitro used, peptides influences the studied parameters; and (iv) in addition to HetR, other molecular targets of PatS influence its role as inhibitor of heterocyst differentiation. Although with the present data it is not possible to exclude that the exported peptide could be further cleaved during the transfer or in the receptor cells to render PatS-5, it is also possible that this is not the physiologically significant inhibitory peptide.

In conclusion, our results are consistent with the model illustrated in Fig. 5, in which PatS-17 is the product of the patS gene in the proheterocysts. This polypeptide, which if accumulated appears to inhibit differentiation of the producing cells, is not exported as such, but it is susceptible of cleavage to render a smaller peptide derived from its C-terminal part and likely consisting of 8 amino acids, which would be the morphogen transferred to the neighbouring cells. In PatS, K2, A3, and I4 appear to be very important and V7 essential for PatS processing/export, but the fact that expression of M-PatS-8 in proheterocysts (in strain CSL47) leads to a wild-type heterocyst distribution implies that export from proheterocysts to vegetative cells could also take place unlinked from processing. Additionally, C10, E12, G14 and R17 appear important and R13, S15 and G16 essential for PatS activity of inhibition of differentiation.

Experimental procedures

Strains and growth conditions

Growth conditions of Anabaena sp. (also known as Nostoc sp.) strain PCC 7120 were as described (López-Igual et al., 2012a). Except for strain CSVT20, which was grown without antibiotics, all the mutants listed in Table 1 were grown in medium supplemented with streptomycin (Sm) and spectinomycin (Sp) at 5 μg ml−1 each in solid media and 2 μg ml−1 each in liquid media. Growth rate constant determinations were performed as in Valladares et al. (2011). Nitrogenase activity was determined under oxic conditions in filaments incubated for 24 h in bubbled cultures without nitrogen and without antibiotics (López-Igual et al., 2012b).

Construction of mutants

To delete the patS gene, two DNA fragments, one encompassing sequences upstream of the gene and the other including sequences downstream of the gene, were amplified by PCR using DNA from Anabaena sp. strain PCC 7120 as template and the primer pairs asl2301-8/asl2301-9 and asl2301-10/asl2301-11 respectively (all oligodeoxynucleotide primers are listed in Table 3). The upstream and downstream DNA fragments were cloned in pMBL-T (Dominion MBL, Spain), sequenced and inserted together, in direct orientation, into pCSRO. This plasmid, which is derived from pRL500 (Elhai and Wolk, 1988), can be mobilized by conjugation and bears an SmR/SpR determinant and the sacB gene for positive selection. The plasmid produced, pCSVT51, bears cloned Anabaena DNA from the asl2301 locus with a deletion of 180 bp, which includes the whole patS gene. Plasmid pCSVT51 was transferred to Anabaena sp. strain PCC 7120 by conjugation, which was performed as described (Elhai et al., 1997). Exconjugants were selected by their resistance to Sm and Sp and double recombinants were then selected by their resistance to sucrose. The clone isolated, CSVT20, lacked those 180 bp including the patS gene.

Table 3. Oligodeoxynucleotide primers used in this work
PrimerSequence (5′ → 3′)

For construction of fusions of the gfp-mut2 gene to patS, two DNA fragments were amplified, one using Anabaena DNA as template and primer pairs asl2301-9/als2301-15 (for strain CSL19) and asl2301-9/asl2301-17 (for strain CSL21), and the other using pCSAM135 (Flores et al., 2007) as template and primer pairs gfp-asl2301-1/gfp-asl2301-2 (for strain CSL19) and gfp-asl2301-5/gfp-asl2301-6 (for strain CSL21). PCR products were cloned together in pMBL-T producing pCSL29 and pCSL33 respectively. KpnI fragments from these plasmids bearing the gfp-mut2 gene were inserted into mobilizable vector pCSV3 (Valladares et al., 2011) digested with the same enzyme, producing pCSL30 and pCSL34 respectively.

For construction of a PatS-6His fusion (strain CSL22), a DNA fragment was amplified using Anabaena DNA as template and primer pair asl2301-9/als2301-18, and cloned in pMBL-T producing pCSL3. A KpnI fragment from this plasmid carrying the patS-6His fusion was inserted into pCSV3 digested with the same enzyme, producing pCSL36.

For construction of PatS point mutants and minigenes, a DNA fragment was amplified using Anabaena DNA as template and primer pairs asl2301-9/asl2301-32 to asl2301-49 (for strains CSL44 to CSL62), asl2301-9/asl2301-77 (for strain CSL94) or asl2301-69/asl2301-71 or asl2301-72 (for strains CSL91 and CSL92 respectively) and finally cloned into pCSV3 producing pCSL82 to pCSL95, pCSL110, pCSL104, and pCSL105 respectively. For construction of strains CSL90 and CSL93, DNA fragments, from 2771000 to 2771734 (coordinates of the strain PCC 7120 genomic sequence; Kaneko et al., 2001), including nucleotide changes to produce K2A and V7A substitutions, respectively, were synthesized (GeneArt®, Invitrogen) and cloned into pCSV3 producing plasmids pCSL103 and pCSL106 respectively.

Plasmids pCSL30, pCSL34, pCSL36, pCSL38, pCSL82 to pCSL95, pCSL110, pCSL104, pCSL105, pCSL103 and pCSL106 were transferred to the patS mutant, strain CSVT20, by conjugation (Elhai et al., 1997). Clones resistant to Sm and Sp were selected, and their genomic structure in the patS region was tested by PCR with primers asl2301-74 (which lies outside of the construct, 5′ from it) and asl2301-75 (which lies inside the region deleted in strain CSVT20) or asl2301-76 (which lies outside of the construct, 3′ from it) to check recombination in the correct place and segregation of the patS-mutant chromosomes. Additionally, except for CSL19 and CSL21, the patS ORF was sequenced in all the Anabaena strains generated in this work.


For light microscopy, filaments grown in BG110 + ammonium medium (in the presence of Sm and Sp for the CSL mutants) were harvested, washed with nitrogen-free (BG110) medium and incubated for 24 h in bubbled cultures at 30°C in the light. At least 300 cells or 100 intervals were counted for each strain in each of three independent experiments. Dividing cells were counted as two cells. For staining (pro)heterocysts, cell suspensions were mixed (1:10) with a 1% Alcian Blue (Sigma) solution (López-Igual et al., 2012b). For confocal microscopy, samples from cultures of Anabaena sp. set atop solidified medium (BG110 + ammonium or BG110) were visualized using a Leica HCX PLAN-APO 63X 1.4 NA oil immersion objective attached to a Leica TCS SP2 confocal laser-scanning microscope. GFP was excited using 488 nm irradiation from an argon ion laser. Fluorescence from GFP was monitored by collection across windows of 500–540 nm.

For immunofluorescence detection of the PatS-5 sequence, 300 μl of liquid cultures (containing 2–3 μg of chlorophyll a ml−1) were placed atop poly-l-lysine precoated microscope slides and covered with a 45 μm pore-size Millipore filter. After that, the filter was removed and the slide was let to dry at room temperature and, then, immersed in 70% ethanol during 30 min at −20°C and dried 10 min in an oven at 80°C. The cells were then washed twice by covering the slide with PBS-T buffer [140 mM NaCl, 1.5 mM KH2PO4, 2.7 mM KCl (pH 7.4), Tween-20 (0.05 p/v)] (2 min each time, room temperature) and treated with blocking buffer (5% milk powder in PBS-T, 15 min). The cells on the slides were then incubated with a primary anti-PatS-5 antibody solution (rabbit polyclonal antibody raised against synthetic RGSGR, Cambridge Research Biochemicals, purified by affinity chromatography and 1:100 diluted in blocking buffer) for 90 min, washed three times with PBS-T, incubated 45 min in the dark with secondary anti-rabbit antibody conjugated to fluorescein isithiocyanate (FITC) (Sigma, 1:500 dilution in PBS-T) and washed three times with PBS-T. After dried, several drops of FluorSave (Carl Biochem) were added atop, covered with a coverslip, sealed with nail lack and let dry overnight at 4°C. Fluorescence was imaged using a Leica DM6000B fluorescence microscope and an ORCA-ER camera (Hamamatsu). Fluorescence was monitored using a FITC L5 filter [excitation, band-pass (BP) 480/40 filter; emission, BP 527/30 filter] and quantified analysing in each cell ‘the mean gray value’ of the ImageJ software (


We thank Victoria Merino-Puerto (Instituto de Bioquímica Vegetal y Fotosíntesis, CSIC) for making strain CSVT20, and Jeff Elhai (Virginia Commonwealth University) for a bioinformatic search of PatS. Work was supported by grants BFU2007-60457 and BFU2010-17980 from the Spanish Government, co-financed by FEDER. L.C.-G. was the recipient of a JAE-predoc fellowship from CSIC.