Many filamentous cyanobacteria are capable of gliding motility by an undefined mechanism. Within the heterocyst-forming clades, some strains, such as the Nostoc spp. and Fisherella spp., are motile only as specialized filaments termed hormogonia. Here we report on the phenotype of inactivation of a methyl-accepting chemotaxis-like protein in Nostoc punctiforme, designated HmpD. The gene hmpD was found to be essential for hormogonium development, motility and polysaccharide secretion. Comparative global transcriptional profiling of the ΔhmpD strain demonstrated that HmpD has a profound effect on the transcriptional programme of hormogonium development, influencing approximately half of the genes differentially transcribed during differentiation. Utilizing this transcriptomic data, we identified a gene locus, designated here as hps, that appears to encode for a novel polysaccharide secretion system. Transcripts for the genes in the hps locus are upregulated in two steps, with the second step dependent on HmpD. Deletion of hpsA, hpsBCD or hpsEFG resulted in the complete loss of motility and polysaccharide secretion, similar to deletion of hmpD. Genes in the hps locus are highly conserved in the filamentous cyanobacteria, but generally absent in unicellular strains, implying a common mechanism of motility unique to the filamentous cyanobacteria.
Many cyanobacteria are capable of motility. This characteristic has been generally overlooked because cyanobacteria universally lack a flagellar apparatus. While only one marine unicellular cyanobacterium is known to be capable of swimming in liquid suspension by an unknown mechanism (Brahamsha, 1999), multiple species of both unicellular and filamentous cyanobacteria are capable of movement across solid surfaces by processes termed gliding or twitching motility. To date, surface motility in cyanobacteria has only been well characterized by genetic analyses in the unicellular cyanobacterium Synechocystis sp. strain PCC 6803 (hereafter Synechocystis 6803). Motility in this organism is analogous to the social twitching motility seen in proteobacteria, dependent on extrusion and retraction of type IV pili (Bhaya et al., 2000).
Observationally, the gliding of filamentous cyanobacteria differs from that of the twitching motility seen in unicellular cyanobacteria. For example, filamentous cyanobacteria glide at rates of 1–10 μm per second, which is about 30–300 times faster than of unicellular cyanobacteria, and in a solitary non-social manner, implying that a different mechanism for motility may be employed (Castenholz, 1982; Adams, 2001). The actual mechanism of filamentous gliding is unknown, but polysaccharide secretion, type IV pili and contractile fibrils have all been implicated (Castenholz, 1982; Hoiczyk and Baumeister, 1997, 1998; Adams et al., 1999; Duggan et al., 2007). However, the motive force driving motility has not been conclusively determined, due in a large part to a lack of genetic studies characterizing the genes and gene products essential for motility in these organisms.
Propulsive polysaccharide secretion is the oldest and best supported descriptive model of gliding motility in filamentous cyanobacteria. Gliding filaments leave a trail of polysaccharide that can be visualized by staining with India ink and the rate of secretion in immobilized filaments correlates well with the rate of gliding (Walsby, 1968; Hoiczyk and Baumeister, 1998). Secreted polysaccharide is observed to originate from an area adjacent to the cell septa. In several different strains, ultrastructural studies have identified a ring of junctional pores on either side of the cell septa that are the putative sites of this polysaccharide secretion (Hoiczyk and Baumeister, 1995, 1998). One study calculated the force generated by an individual junctional pore in hormogonia of the thermophile Mastigocladus laminosus to be in the range of 100 pN, considerably stronger than most molecular motors studied (Robinson et al., 2007). The identity of the components comprising these junctional pores has remained elusive.
Filamentous cyanobacteria capable of gliding motility can be broadly separated into two classes based on the requirement, or not, to differentiate specialized filaments called hormogonia. The vegetative filaments of non-differentiating (taxonomic subsection III, Oscillatoriales) (Castenholz, 2001) and some heterocyst-forming (cells specialized for nitrogen fixation; subsections IV and V, Nostocales and Stigonematales respectively) (Castenholz, 2001) filamentous cyanobacteria are motile. However, in other heterocyst-formers, such as Nostoc spp., motility requires the transient differentiation of hormogonia. Hormogonia are non-growing, non-nitrogen-fixing filaments that function in dispersal and return to the vegetative growth state after a set period in laboratory culture; they respond to environmental stimuli such as light and the presence of symbiotic plant partners (Campbell and Meeks, 1989).
Here we report on the inactivation of a methyl-accepting chemotaxis-like protein (Mcp) in Nostoc punctiforme strain ATCC 29133 and establish its role in regulating hormogonium development, motility and polysaccharide secretion. Through the use of comparative transcriptomics, we defined a subset of genes involved in the hormogonium differentiation process that are under the control of this Mcp. This information was then used to identify a gene locus that mutational analysis indicates may encode for a novel polysaccharide secretion system unique to filamentous cyanobacteria.
Inactivation of hmpD
The genome of N. punctiforme harbours five loci of genes that encode chemotaxis-like proteins. The proteins encoded by each locus appear to constitute an individual signal transduction complex. We designate one of these loci the hmp (for hormogonium motility and polysaccharide) locus based on findings presented below. The hmp locus contains five genes, their names, locus tags and homologous proteins are as follows respectively: hmpA (NpF5960) – PatA family of CheY-like proteins; hmpB (NpF5961) – CheY; hmpC (NpF5962) – CheW; hmpD (NpF5963) – Mcp; and hmpE (NpF5964) – CheA. This locus, designated operon 1 in a bioinformatic study by Wuichet and Zhulin (2003), is highly conserved in cyanobacteria and similar to the pilG gene cluster in Synechocystis 6803, in which it has been reported to regulate synthesis of type IV pili (Yoshihara et al., 2002). The Mcp from the hmp locus, HmpD, consists of an N-terminal cytoplasmic sensing domain, two transmembrane domains separated by a 15-amino-acid region, presumably periplasmic, and standard Mcp hamp and signalling domains in the C-terminus. Although highly conserved in all cyanobacteria, the sensing domain is much larger in filamentous cyanobacteria, particularly those in taxonomic subsections IV and V, implying additional or alternative functions of this domain in these organisms.
Insertional inactivation of hmpD resulted in a hormogonium non-motile phenotype. To insure that the phenotype of the hmpD mutant strain, generated by insertion of an Ω-npt cassette, was responsible for the observed phenotype, an in-frame deletion of hmpD was generated. A basal level of approximately 10% of the wild-type vegetative filaments randomly differentiate into hormogonia at all times during culture in the presence of a combined nitrogen source (NH4+ or NO3−) or when dependent on N2 as the sole nitrogen source (Fig. 1A). Unlike the wild-type strain, the ΔhmpD strain (UCD 543) failed to differentiate a basal level of hormogonia under standard growth conditions (Fig. 1A) and growth in liquid cultures was more dispersed. The addition of Anthoceros punctatus plant exudate, which contains a potent hormogonium-inducing factor (HIF) (Campbell and Meeks, 1989), to cultures of the ΔhmpD strain resulted in the differentiation of filaments with the morphological characteristics of hormogonia, including a reduction in cell size, and often, but not exclusively, tapered cells at the ends of filaments; these filaments lacked any signs of motility as demonstrated by both plate assays (Fig. 1A–C) and time-lapse microscopy of individual filaments (Movies S1 and S2).
One descriptive model posits that secretion of polysaccharides provides the motive force for gliding in filamentous cyanobacteria. We first confirmed that N. punctiforme hormogonia secreted polysaccharide while gliding by staining of the slime trails with India ink particles during time-lapse microscopy (Fig. 1D, Movie S3). Because hormogonia of the hmpD-deletion strain were non-motile, attempts to stain this strain with India ink were inconclusive, leading us to employ a lectin-based fluorescent stain (RCA120 conjugated to fluorescein), previously demonstrated to show specificity to the hormogonium polysaccharide (Schüßler et al., 1997). Wild-type cultures with hormogonia, stained with RCA120-fluorescein, accumulated extracellular material that fluoresced cyan in colour (Fig. 2A). Conversely, the ΔhmpD strain failed to accumulate fluorescence material, consistent with hormogonia of this mutant strain failing to secrete polysaccharide (Fig. 2A).
Microarray comparison of the wild-type and ΔhmpD strain
Although HIF could induce differentiation of filaments in the ΔhmpD strain with morphological characteristics similar to wild-type hormogonia, their lack of motility and polysaccharide secretion implies that their differentiation may have been incomplete. The hormogonia in the mutant strain return to the vegetative growth state slightly more rapidly than do the wild type. To assess the effect of hmpD deletion on the transcriptional programme leading to hormogonium formation, a DNA microarray comparison of the transcriptomes for the wild-type and ΔhmpD strain over the time-course of hormogonium differentiation was performed. The transcriptomes of both strains were defined at 0, 1, 3, 6, 12, 18 and 24 h after hormogonium induction with HIF. The fact that the wild-type strain continually forms a basal level of hormogonia complicates the interpretation of a direct comparison between the mutant and wild-type strains at each time point. Therefore, comparisons were made between each time point and t = 0 for the same strain; the data were normalized in the R statistical program (Campbell et al., 2007) and subsequently analysed via a Bayesian analysis of time series (BATS) statistical approach (Angelini et al., 2008) to define the set of genes that show a statistically significant change in transcription over the course of hormogonium differentiation (Christman et al., 2011). This analysis yielded a total of 1188 genes with differential expression in the wild-type strain; 622 were upregulated and 566 were downregulated. Of these 1188 genes, 611 were also differentially expressed in the ΔhmpD mutant and showed a similar pattern of regulation, with 257 genes upregulated and 354 genes downregulated (Fig. 3A, Dataset S1). The remaining 577 genes with differential transcription in the wild type were not differentially transcribed in the ΔhmpD strain. Of the 577 genes, 365 were upregulated and 212 downregulated in wild-type hormogonia (Fig. 3A, Dataset S1). Deletion of hmpD, thus, influences the expression of slightly less than half of the genes differentially transcribed during the programme of hormogonium differentiation, with the majority (63%) of these genes failing to show increased transcription in the ΔhmpD strain. Included in this group are a large number of histidine kinases likely involved in environmental sensing, as well as genes annotated as gas vesicle proteins and type IV pilus components (Dataset S1). Additionally, the hmp locus appears to be autoregulatory, as hmpABC, genes upstream of hmpD, fail to show a sustained increase in transcription in the ΔhmpD strain (Fig. 3B).
Because the ΔhmpD strain fails to secrete hormogonium polysaccharide, we reasoned that hmpD may control transcription of the genes required for this process and, therefore, searched for genes that would encode for polysaccharide synthesis and secretion upregulated only in the wild type. We began by looking for glycosyl transferases, hallmarks of polysaccharide synthesis, and found a total of 15. Of these, five were found in a large contiguous cluster of chromosomal genes (NpF0066 to NpF0078) that we have provisionally designated the hps locus (hormogonium polysaccharide) (Fig. 3C, Table 1). The first four genes in this cluster show increased transcription in both the wild-type and ΔhmpD strain during hormogonium differentiation, while the downstream genes, including all five glycosyl transferases, were not differentially transcribed in the ΔhmpD strain.
Table 1. hps locus homologues in filamentous cyanobacteria
The hps locus is unique to, and highly conserved in, filamentous cyanobacteria
The hps locus in N. punctiforme encompasses 13 genes: two encoding conserved hypothetical proteins containing multiple transmembrane domains, hpsA and hpsJ; four encoding pseudopilins [proteins containing a leader sequence and prepilin peptidase (PilD) cleavage site with high similarity to type IV pilins, but lacking significant sequence similarity to the remaining pilin (PilA) sequence], hpsBCD and hpsH; and five glycosyl transferases, hpsEFG, hpsI and hpsK; as well as two transposases. Results from blastp searches indicate that the majority of the genes in this cluster are highly conserved in all three taxonomic subsections of filamentous cyanobacteria, but absent in unicellular strains (Table 1, Fig. 4). In subsections IV and V, the heterocyst-forming cyanobacteria, there is substantial synteny along this gene cluster, but with the transposases present in N. punctiforme absent in other strains, implying that these transposon insertions were a recent event. In subsection III filamentous cyanobacteria, these genes tend to be dispersed as subclusters among different sites in the genome. The presence of multiple glycosyl transferases, hallmarks of polysaccharide synthesis; pseudopilins, commonly found in type II secretion systems, type IV pili and archeal flagella (Peabody et al., 2003); and proteins of unknown function containing multiple transmembrane domains led us to the hypothesis that the hps locus encodes for a novel polysaccharide synthesis and secretion system essential for motility in filamentous cyanobacteria.
Inactivation of genes in the hps locus affects motility and polysaccharide secretion
To test the hypothesis that the hps locus encodes for a novel polysaccharide secretion system, strains harbouring in-frame deletions of hpsA (strain UCD 572), hpsBCD (strain UCD 573), hpsEFG (strain UCD 574) and hpsJ (strain UCD 600) were constructed. Deletions of contiguous hpsBCD and hpsEFG were constructed based on the rationale that inactivation of an individual pseudopilin or glycosyl transferase may not yield a detectable phenotypic effect due to potential redundancy in function. Three of these strains, the ΔhpsA, ΔhpsBCD and ΔhpsEFG strains failed to show either motility or polysaccharide secretion as evidenced by plate and time-lapse motility assays (Fig. 1B and C, Movie S4 with ΔhpsA depicted), and staining with RCA120-fluorescein (Fig. 2, only ΔhpsEFG strain depicted), respectively, although morphologically distinct hormogonium-like filaments were present (Fig. S1). In contrast, deletion of hpsJ resulted in a strain with decreased motility in plate and time-lapse assays (Fig. 1B and C, Movie S5). Unexpectedly, staining with RCA120-fluorescein was brighter in the ΔhpsJ strain than the wild type (Fig. 2B). Occasionally, in filaments dissociated from the accumulated polysaccharide, fluorescence was observed accumulating at the junction between cells and in trails left behind (Fig. 2B). These results support the hypothesis that the hps locus encodes for a novel polysaccharide secretion system.
The motor of gliding motility in filamentous cyanobacteria has remained a mystery in large part due to the lack of genetic and molecular approaches to understanding this phenomenon. By employing a reverse genetic approach, combined with global transcriptional profiling, we have provided two key findings that further the understanding of rapid gliding motility.
The first finding is the multiple roles that the hmp locus plays in the processes of hormogonium development, motility and polysaccharide secretion. Deletion of the hmpD gene results in strains that no longer undergo spontaneous rounds of a low level of hormogonium differentiation, indicating this locus is a key player in regulating hormogonium differentiation. The block in differentiation of hormogonia can be at least partially bypassed by addition of the plant-derived HIF, resulting in filaments with the morphological characteristics of hormogonia. However, these filaments lack any signs of motility and fail to secrete polysaccharide, indicating that HmpD is essential for these processes. Comparative transcriptomics show that approximately half of the genes differentially transcribed over the time-course of hormogonium differentiation are under the control of hmpD, reinforcing the idea that the hmp locus has a role in regulating the differentiation process. Positive feedback loops are common in bi-stable switches, such as those governing the development of differentiated cell types (Brandman et al., 2005). If, as we suggest, the hmp locus is positively autoregulated, that locus may play a central role in the regulation of the development of hormogonia. Major questions regarding the hmp locus still remain. For instance, this locus lacks any obvious response regulators containing DNA-binding domains indicative of transcriptional regulation. Thus, the mechanism by which the hmp locus regulates transcription of the hmp regulon is unknown; it likely involves a phosphorelay to other as yet to be determined proteins.
The results from the global transcriptional profiling of the ΔhmpD strain then led us to a second major finding, the identification of the hps locus. The hps locus may encode at least part of a polysaccharide synthesis and secretion system that includes glycosyltransferases for synthesis and pseudopilins for secretion. The latter, typically found in type II secretion systems, might conceivably be that part of the junctional pores that spans the periplasm. Deletions of four different regions within this gene cluster resulted in two distinct phenotypes. Deletion of hpsA, hpsBCD and hpsEFG yielded strains that both were non-motile and failed to secrete hormogonium polysaccharide. Specifically, the results with ΔhpsEFG are consistent with a requirement for polysaccharide secretion in gliding motility, as substratum or propulsion. Conversely, deletion of hpsJ resulted in a decrease in the rate of gliding, with a concomitant change in the nature or amount of polysaccharide produced, also consistent with the hypothesis that the hps locus encodes for the polysaccharide synthesis and secretion system. Additionally, in the hpsJ mutant, polysaccharide could be seen accumulating at the junctions between cells along the filament, implying that the cell junctions are the site at which this polysaccharide is secreted via junctional pores.
Transcription in the hps locus is indicative of a two-stage process. In the first stage, immediately upon initiation of hormogonium differentiation, the first four genes in the hps locus are upregulated. This includes hpsA, encoding the large, conserved-hypothetical membrane protein, and the pseudopilins hpsBCD. The second stage is dependent on the hmp locus, resulting in the subsequent upregulation of the rest of the genes in the hps locus, including all five of the glycosyl transferases, hpsEFG, hpsI and hpsK. The delay in synthesis of the secreted polysaccharide would allow for assembly of the structural components of the secretion apparatus (HpsABCD and, perhaps, HpsJ), resulting in subsequent co-ordination of synthesis, secretion and movement.
The mechanism of motility has been suggested to vary among different genera of filamentous cyanobacteria (Adams, 2001). This suggestion was based on the fact that some strains, such as N. punctiforme, express type IV pili on the surface of motile filaments, whereas type IV pili have not been widely observed in other motile filamentous cyanobacteria. A conclusion, therefore, is that type IV pili mediated gliding is used by some strains, while a second mechanism, such as polysaccharide secretion, may be employed by others. Putative polysaccharide-secreting junctional pores have been identified in numerous filamentous cyanobacteria (Hoiczyk and Baumeister, 1998) and extracellular polysaccharide is often present around motile strains as capsules, sheath, slime or trails (Schüßler et al., 1997; Hoiczyk and Baumeister, 1998). It is possible that hormogonia could utilize both mechanisms for different types of movement, such as occurs in adventurous (solitary) and social (aggregate or population) gliding in Myxobacteria (Zhang et al., 2011). While hormogonia can glide in the same direction in aggregates, there is no evidence for or against that movement as being behaviourally different from solitary filaments; individual filaments also glide into and away from aggregates. We suggest a third possibility. The fact that the hps locus is highly conserved among, and apparently unique to, filamentous cyanobacteria favours a common polysaccharide secretion mechanism of motility while type IV pili have another physiological role. We are reexamining the contributions of type IV to motility in hormogonia of N. punctiforme. Since type IV pili, type II secretion systems and the archael flagella share homologous components (Peabody et al., 2003), the polysaccharide secretion apparatus may have a common evolutionary history with the other secretion systems containing pseudopilins.
Strains and culture conditions
For a detailed description of the plasmids, strains and oligonucleotides used in this study refer to Tables S1 and S2 in the supplementary materials and methods. N. punctiforme strain ATCC 29133 and its derivatives were cultured in Allen and Arnon medium, diluted fourfold (AA/4) as previously described (Campbell et al., 2007). All mutant strains grew similar to the wild type with N2 as the sole nitrogen source. HIF from A. punctatus was generated as previously described (Campbell and Meeks, 1989). For hormogonium induction, cultures at a chlorophyll a (Chl a) concentration of 2–3 μg ml−1 were harvested at 1000 g for 5 min, washed three times with AA/4 and suspended in AA/4 + HIF. For selective growth, the medium was supplemented with 25 μg ml−1 neomycin. Escherichia coli cultures were grown in Luria–Bertani (LB) broth for liquid cultures or LB supplemented with 1.5% (w/v) agar for plates. Selective growth media were supplemented with 50 μg ml−1 kanamycin, 50 μg ml−1 ampicillin and 10 μg ml−1 chloramphenicol.
Plasmid and strain construction
All plasmid inserts created via PCR were sequenced to insure fidelity. Gene deletions, allelic replacements and introduction of mobilizable shuttle vectors were performed as previously described (Wong and Meeks, 2001) with the exception that vectors were transformed into E. coli strain UC585 (Liang et al., 1993) and introduced into appropriate strains of N. punctiforme via bi-parental conjugation. The details are presented in the supplementary information.
Plate motility assays
For plate motility assays, cultures of the appropriate strain were induced to differentiate hormogonia with the A. punctatus generated HIF as previously described (Campbell and Meeks, 1989). Hormogonium suspensions were concentrated by centrifugation and a 5 μl portion was spotted onto AA/4 supplemented with 0.5% (w/v) Noble agar. The spots were imaged at both t = 0 h and t = 24 h using a Wild M7 S dissecting microscope (Heerbrugg) set at 6× magnification and equipped with a Micropublisher 3.3 RTV camera using Q-Capture pro 5.0 software (QImaging). To determine the motility index, images were converted to threshold and the area of the image covered by the colony was quantified using ImageJ software (NIH). The area covered at t = 24 h was divided by the area at t = 0 h to define the motility index.
Time-lapse microscopy assays
For time-lapse microscopy assays of motility, a thin microchamber was first constructed by using heat to adhere a two-ply thick piece of parafilm to a glass slide. A square section of the parafilm slightly smaller than the size of a glass coverslip was then excised from the slide to create a thin microchamber. Fifty microlitres of each strain, induced for hormogonia as described above, was added to the microchamber slide and sealed with a glass coverslip. Cells were imaged using a 10× objective lens on a Nikon Eclipse 80i microscope mounted with a QImaging Micropublisher 3.3 RTV camera using Q-Capture pro 5.0 software (QImaging). Images were taken at 10 s intervals for 10 min.
Polysaccharide staining assays
For India ink staining of hormogonium polysaccharides, time-lapse microscopy was performed as described above, with Super Black India ink particles (Speedball Art Products), diluted 50-fold in medium containing 4 μM CaCl2 and 0.05% (v/v) Triton X-100 (Sigma). Fresh medium containing RCA120-fluorescein (Vector Laboratories) at 20 μg ml−1 was used to suspend hormogonia that had been collected by centrifugation. After 30 min of incubation, hormogonia were again harvested by centrifugation, washed once in fresh medium to remove residual dye, and visualized via fluorescence microscopy using a 40× objective lens on a Nikon Eclipse 80i microscope mounted with a QImaging Micropublisher 3.3 RTV camera using Q-Capture pro 5.0 software (QImaging) and equipped with an X-cite 120 Fluorescence Illumination System (Lumen Dynamics). The filter sets G-2A (all terminology as the manufacturer's designation: Ex 465–495 nm BP, Dichromatic mirror 565 nm LP, Em 590 nm LP) and B-2E/C (Ex 510–560 nm BP, Dichromatic mirror 505 nm LP, Em 515–555 nm BP) (Nikon) were used for autofluorescence and RCA120-fluorescein respectively.
DNA microarray experiments were performed as previously described (Risser et al., 2012) with the following exceptions. Two 500 ml cultures of N. punctiforme wild-type or the ΔhmpD strain (UCD 543) [supplemented with 2.5 mM NH4Cl and 5 mM 3-(N-morpholino)propanesulphonic acid (MOPS), pH 7.8] at 2–3 μg Chl a ml−1 were harvested by centrifugation at 1000 g for 5 min and washed three times with AA/4 (supplemented with MOPS-buffered 2.5 mM NH4Cl). A volume containing 100 μg of Chl a was harvested for the t = 0 h sample and the remainder was suspended in 600 ml of A. punctatus HIF and divided evenly among 12 flasks (supplemented with MOPS-buffered 2.5 mM NH4Cl immediately prior to use). Two 50 ml culture volume flasks, containing a total of 100 μg of Chl a, were harvested at t = 1, 3, 6, 12, 18 and 24 h after induction of hormogonia (n = 3). Buffered ammonium was included in the experiment in order to exclude the nitrogen starvation response so that only genes specific to hormogonium differentiation would be differentially expressed (Christman et al., 2011). Data analysis in the R statistical platform was performed as previously described (Campbell et al., 2007), with the exception that Bayesian analysis of time series (BATS) was performed on the normalized data to generate a list of genes with statistically significant changes in expression over the entire time-course (Angelini et al., 2008; Christman et al., 2011). The array data reported in this article have been deposited in the Gene Expression Omnibus (GEO) database, http://www.ncbi.nim.nih.gov/geo (Accession No. GSE42859).
This work was supported by Grant IOS 0822008 from the US National Science Foundation. We thank Elsie Campbell and Emiko Sano for comments during preparation of the manuscript. Kari Hagen constructed and characterized the initial insertion mutation in hmpD (NpF5963).