Report

Heterocyst Pattern Formation Controlled by a Diffusible Peptide

See allHide authors and affiliations

Science  30 Oct 1998:
Vol. 282, Issue 5390, pp. 935-938
DOI: 10.1126/science.282.5390.935

Abstract

Many filamentous cyanobacteria grow as multicellular organisms that show a developmental pattern of single nitrogen-fixing heterocysts separated by approximately 10 vegetative cells. Overexpression of a 54–base-pair gene, patS, blocked heterocyst differentiation in Anabaena sp. strain PCC 7120. A patS null mutant showed an increased frequency of heterocysts and an abnormal pattern. Expression of a patS-gfp reporter was localized in developing proheterocysts. The addition of a synthetic peptide corresponding to the last five amino acids of PatS inhibited heterocyst development. PatS appears to control heterocyst pattern formation through intercellular signaling mechanisms.

The regulation of cell fate and pattern formation is a fundamental problem in developmental biology. Cell-cell communication often plays a key role in controlling development. Diffusible molecules that directly influence cell fate determination have been found in several eukaryotic organisms (1). Prokaryotic development in Bacillus,Streptomyces, and Myxococcus is also controlled by intercellular signaling (2, 3). We have investigated the regulation of cell fate determination and pattern formation in a prokaryote that grows as a simple multicellular organism.

When the filamentous cyanobacterium Anabaena sp. strain PCC 7120 grows diazotrophically, approximately every tenth vegetative cell terminally differentiates into a heterocyst (4) (Fig. 1, A and B). This simple, one-dimensional developmental pattern spatially separates two incompatible processes: oxygen-evolving photosynthesis in vegetative cells and oxygen-sensitive nitrogen fixation in heterocysts. We have found that a small gene,patS, is crucial for the formation and maintenance of a normal heterocyst pattern.

Figure 1

patS controls heterocyst development in Anabaena PCC 7120. Wild-type filaments (A) grown in BG-11 medium and (B) after the nitrogen step-down in BG-110 to induce heterocysts (arrowheads) are shown. (C) Overexpression ofpatS from pAM1691 (Fig. 2A) prevented heterocyst formation in BG-110, and (D) deletion ofpatS (AMC451) resulted in supernumerary heterocysts with an abnormal pattern in BG-110. Brackets indicate chains of contiguous heterocysts.Anabaena PCC 7120 and derived strains were grown as previously described (21). Differential interference contrast micrographs were taken before (A) and 24 hours after (B through D) heterocyst induction. Scale bars, 10 μm.

The patS gene was identified on the conjugal cosmid 8E11 (5), which suppressed heterocyst development (Fig. 2A). A 3.3-kb subclone (pAM1035) was shown to confer the heterocyst-suppression phenotype (Hets), and its sequence was determined (GenBank accession number AF046871). The same fragment isolated from an independent cosmid, 13C12, produced the same phenotype, indicating that the dominant Hets phenotype is a property of wild-type sequences. An analysis of subcloned fragments in shuttle vectors (Fig. 2A) prompted us to investigate a small, 51–base pair (bp), open reading frame (ORF) named patS (Fig. 2B).

Figure 2

(A) Identification of thepatS gene on cosmid 8E11. Subclones were generated by cloning the indicated DNA fragments into shuttle vectors (22). The Bam HI–Cla I fragment in pAM1035 is indicated as a thick black bar on the 8E11 map. Cloning and molecular techniques were performed as previously described (23). Plasmids were transferred into Anabaena PCC 7120 by conjugation fromEscherichia coli (19, 24). WT, wild type; Hets, heterocyst suppression; Hetc, heterocyst formation on a nitrate-containing medium; Pat, abnormal heterocyst pattern; Mch, multiple contiguous heterocysts; inverted triangle, transcription terminator; P with arrow, external promoter. (B) Nucleotide sequence of the smallest tested DNA fragment that is sufficient to suppress heterocyst development and the deduced amino acid sequence of PatS (14). Potential start codons are underlined. The amino acids encoded by four missense mutants are shown below the wild-type sequence. Nucleotide numbering begins with the first nucleotide in pAM1882. (C) Sequences of in-frame and out-of-frame patS-lacZ translational fusions (9). The Sma I site used for the constructions is shown in bold.

A 140-bp polymerase chain reaction (PCR) fragment containing thepatS ORF (pAM1686) conferred the Hets phenotype (Fig. 2A). Overexpression of patS by the AnabaenaPCC 7120 glnA promoter (pAM1691) completely blocked heterocyst formation (Figs. 1C and 2A). The antisense orientation (pAM1695) produced no noticeable phenotype. patS blocks development at an early stage because even the cryptic pattern of nonfluorescent cells, which is produced by some developmental mutants (6), was not observed.

To test whether different levels of transcription correlate with the degree of heterocyst inhibition, we placed patS under the control of the copper-inducible petE promoter (PpetE) (7) (Fig. 3A). Without the addition of CuSO4, the strain containing PpetE-patS was wild type. AspatS transcription was increased by the addition of CuSO4, there was a corresponding decrease in the frequency of heterocysts. We observed no influence of CuSO4 on heterocyst development when patS was cloned in the reverse orientation (Fig. 3A).

Figure 3

(A) patStranscription suppressed heterocyst formation. Strains containing pAM1714 (squares), which carries PpetE-patS, or the control plasmid pAM1716 (circles) (7) were washed and grown for several generations in copper-free BG-11 liquid medium to deplete the residual copper. At an optical density of 750 nm (OD750) = 0.4, cultures were induced by transferring the filaments to 2 ml of BG-110 that contained different concentrations of CuSO4. The percentage of heterocysts was determined microscopically 24 hours after induction. At least 1000 cells were scored for each sample. Values are shown for one representative of four independent experiments. (B andC) Inhibition of heterocyst differentiation by synthetic peptides. Peptides (Genosys Biotechnologies, The Woodlands, Texas) were added to wild-type (B) and ΔpatSstrain AMC451 (C) immediately after transfer to 2 ml of BG-110. One representative of three independent experiments is shown. PatS-5 (RGSGR) (squares); PatS-4 (GSGR) (triangles); PatS-G4S (RGSSR) (circles); and mixture of amino acids (R, G, and S) (diamonds). (D through H) Abnormal heterocyst pattern inΔpatS strain AMC451 and complementation by heterocyst-specific expression of patS. The following filaments were grown in liquid BG-11 to OD750 = 0.2, induced in BG-110 for 24 hours, and scored microscopically: wild type (D), AMC451 (E), AMC451 induced in the presence of 60 nM PatS-5 peptide (F), AMC483 (AMC451 containing pAM1715, which carriespatS that is driven by the hepA promoter) (G), and AMC493 [AMC451 containing pAM1685 as a negative control (Fig. 2A)] (H).

Mutations in patS resulted in a loss of the ability to suppress heterocysts. pAM1882 (Fig. 2A) was mutagenized and screened for plasmids that failed to suppress heterocysts (8). Four plasmids were identified, each with a missense mutation withinpatS (Fig. 2B).

patS potentially encodes a 17–amino acid peptide, starting at the first available ATG codon; however, other in-frame ATG and GTG codons are present. PatS has no homologs or sequence motifs in the databases. It contains a stretch of five hydrophobic amino acids in its NH2-terminal half, and its COOH-terminal half is mostly hydrophilic.

A patS-lacZ translational fusion showed thatpatS is translated and developmentally regulated (9). β-galactosidase (β-Gal)–specific activity of the in-frame fusion (Fig. 2C) increased about threefold (to 6000 units) during the 6 hours after the heterocyst induction and then decreased to the preinduction level after 27 hours. A direct analysis of thepatS message on RNA blots yielded similar results (10). The out-of-frame construct produced background levels of β-Gal.

Because many cell-cell signaling molecules in Gram-positive bacteria are peptides (11, 12) and because a long-standing model for the control of the heterocyst pattern involves a diffusible inhibitor that is produced by proheterocysts (4), we suspected that patS might encode an exported signaling molecule. We were intrigued with the phosphate regulator (phr) genes from Bacillus subtilis(13) and with phrC in particular, which encodes the quorum-sensing pheromone competence and sporulation stimulating factor (CSF) (3). The CSF is an unmodified exported pentapeptide that is processed from the COOH-terminal end of a 40–amino acid precursor. These precedents, and the fact that the fourpatS missense mutations happened to be in the last five codons (Fig. 2B), led us to test a synthetic pentapeptide corresponding to the COOH-terminal end of PatS.

This pentapeptide, PatS-5 (RGSGR), inhibited heterocyst formation at submicromolar concentrations (Fig. 3B) (14). Two altered peptides had reduced activity: PatS-4 (GSGR), which contains the last four amino acids, and PatS-G4S (RGSSR), which contains a G to S substitution corresponding to a mutation that reduced patSactivity (Fig. 2B) (14). A mixture of each amino acid (R, G, and S) present in PatS-5 had no effect on heterocyst formation (Fig. 3B) (14).

patS is required to inhibit heterocyst formation. ApatS deletion strain, AMC451 (15), formed heterocysts on a nitrate-containing medium; the wild type does not produce heterocysts in this medium. Twenty-four hours after the nitrogen step-down, wild-type filaments formed a pattern of single heterocysts that were separated by 8 to 14 vegetative cells (Figs. 1B and 3D). AMC451 formed multiple contiguous heterocysts and short vegetative-cell intervals (Figs. 1D and 3E). In the experiment shown inFig. 3, D and E, the wild type formed single heterocysts (100%), whereas AMC451 formed single (39%), double (55%), quadruple (3%), and sextuple (3%) heterocysts. Longer chains of up to 10 contiguous heterocysts were occasionally formed (Fig. 1D). The phenotype of theΔpatS mutant AMC451 was due to the loss of patSalone because AMC451 was complemented by patS that was introduced on plasmids pAM1882, pAM1835, and pAM1686.

Heterocyst formation of AMC451 was inhibited by exogenously added PatS-5, indicating a normal response to the pentapeptide (Fig. 3C); however, the pattern was still abnormal. The induction of AMC451 in the presence of 60 nM PatS-5 reduced the number of heterocysts to about wild-type levels (11%) but failed to restore a normal pattern (Fig. 3F). This is consistent with a model that requires a gradient of the PatS signal originating from proheterocysts.

Expressing patS in only proheterocysts suppressed the pattern defects of the ΔpatS mutant AMC451 and indicated that PatS functions in a manner that is nonautonomous to the cell. ThehepA promoter is induced in proheterocysts between 4.5 and 7 hours after the nitrogen step-down (16). AMC451 containing a plasmid-borne PhepA-patSfusion had a nearly wild-type pattern (Fig. 3G). PromoterlesspatS failed to complement the ΔpatS phenotype (Fig. 3H).

As predicted for a gene encoding a diffusible inhibitor that controls pattern, patS expression was highest in differentiating cells. A patS transcriptional fusion with green fluorescent protein (GFPmut2) (17) on a low copy number plasmid (pAM1951) was used as a reporter. Weak GFP fluorescence with no obvious pattern was found in nitrate-grown filaments (Fig. 4A). A distinct pattern of brightly fluorescent cells was formed within the 12 hours after induction, which is before proheterocysts could be identified. After 18 hours, strong GFP fluorescence was localized to differentiating proheterocysts. After filaments began diazotrophic growth, GFP fluorescence was strongest from single cells that were midway between two heterocysts, where a new heterocyst will form (Fig. 4B). A strain containing a vegetative cell–specific promoter fused to gfpshowed fluorescence from only vegetative cells (Fig. 4C). A strain containing pAM1956, which carries promoterless gfp, had no detectable GFP fluorescence.

Figure 4

Temporal and spatial expressions of apatS-gfp reporter (25). GFP fluorescence (left) and the corresponding differential interference contrast photomicrographs (right) (A) after induction for 0, 12, and 18 hours and (B) after prolonged diazotrophic growth. (C) A control strain containing gfpdriven by the vegetative cell–specific rbcL promoter (25). Triangles, proheterocysts; arrowheads, mature heterocysts; scale bars, 10 μm.

PatS seems to play a key role in heterocyst pattern formation by inhibiting the formation of adjacent heterocysts and by maintaining a minimum number of vegetative cells between heterocysts. The inhibition of neighboring cells by select differentiating cells (lateral inhibition) is an important mechanism of pattern formation in eukaryotic organisms (18). Because it takes ∼20 hours for heterocysts to mature and begin supplying fixed nitrogen to the filament, a specialized early inhibitory signal is required to allow only a fraction of starving cells to terminally differentiate. The first cells to differentiate increase the production of PatS to inhibit neighboring cells from forming heterocysts. PatS-producing cells must themselves be refractory to the PatS signal. The mechanism of immunity is unknown, but would be expected to involve a cell-autonomous inhibition of self-signaling such as that proposed for PrgY in enterococcal conjugation (11).

The PatS signal is likely to be a processed COOH-terminal peptide that is confined to the periplasm of this Gram-negative cyanobacterium. There is no evidence of heterocyst inhibition between filaments in mixed cultures or by a conditioned medium from a strain overexpressingpatS (10).

We propose a model in which a processed PatS peptide, originating from differentiating proheterocysts, diffuses along the filament's contiguous periplasmic space and is taken up by neighboring cells, creating a gradient of inhibitory signal. The intracellular target of PatS signaling is unknown, but components of a phosphorelay such as that found in Bacillus (13) are likely candidates.

REFERENCES AND NOTES

View Abstract

Navigate This Article