pilG Gene Cluster and Split pilL Genes Involved in Pilus Biogenesis, Motility and Genetic Transformation in the Cyanobacterium Synechocystis sp. PCC 6803

Abstract

The unicellular motile cyanobacterium Synechocystis sp. PCC 6803 exhibits phototactic motility that depends on the type IV-like thick pilus structure. By gene disruption analysis, we showed that a gene cluster of slr1041, slr1042, slr1043 and slr1044, whose predicted products are homologous to PatA, CheY, CheW and MCP, respectively, was more or less required for pilus assembly, motility and natural transformation competency with extraneous DNA. By sequence homology, the missing cheA-like gene in this cluster was identified as novel split genes, slr0073 and slr0322, at separate loci on the genome. This was confirmed by non-motile phenotype of their disruptants. Unique hyperpiliation was observed in the slr1042 and slr0073 disruptants, suggestive of their specific interaction with pilT1. The genes, thus identified as pil genes in this study, were designated pilG (slr1041), pilH (slr1042), pilI (slr1043), pilJ (slr1044), pilL-N (slr0073) and pilL-C (slr0322).

. (Received January 7, 2002; Accepted February 28, 2002).

Introduction

A number of cyanobacteria show cellular motility of gliding, twitching or swimming without flagella. Some filamentous cells such as hormogonia show conspicuous gliding motility, while single cells often show twitching or swimming motility. In many cases, they respond to light for the purpose of their localization under conditions suitable for photosynthetic production or for avoidance from photodamage.

In 1996, the entire genome sequence of a unicellular motile cyanobacterium Synechocystis sp. PCC 6803 was determined (Kaneko et al. 1996). Since then, quite a few genes have been identified to be involved in motility and phototaxis by mutational analysis of genes. Although single cell motility of this cyanobacterium must be evaluated with fine microscopic measurements, overall motility can be readily recognized as colony morphology on soft agar plates. Genes thus identified have been categorized into three types: (1) motility genes primarily involved in pilus assembly (pil genes), (2) secondary genes involved in motility, and (3) genes involved in positive phototaxis (pix genes).

The pil genes that are essential for motility in Synechocystis are pilA1 (sll1694), pilB1 (slr0063), pilC (slr0162+slr0163), pilD (slr1120), pilM (slr1274), pilN (slr1275), pilO (slr1276), pilQ (slr1277) and pilT1 (slr0161) (Bhaya et al. 2000, Yoshihara et al. 2001). In Synechocystis, these genes are also essential for transformation competency (Yoshihara et al. 2001). Electron microscopy showed that the thick pilus structure (diameter 5 nm) was completely lost from the cell surface of all these mutants except pilT1. By contrast, the pilT1 mutant showed enhanced assembly of the thick pili (Okamoto and Ohmori 1999, Bhaya et al. 2000). These phenotypes are practically the same as those of pil mutants defective in assembly of type IV pili in some Gram-negative bacteria. It has been established that type IV pilus structures are involved in twitching motility of Pseudomonas aeruginosa, social gliding motility in Myxococcus xanthus and transformation competency in Neisseria gonorrhoeae (Darzins and Russell 1997, Spormann 1999, Fussenegger et al. 1997). Thus, the type IV-like thick pili are the machinery responsible for both cell motility and natural transformation in Synechocystis (Yoshihara et al. 2001). Recent genome analysis also revealed the presence of those pil genes in many other cyanobacteria (Kaneko et al. 2001), although experimental evidence has not yet been available.

The second category includes a number of genes such as putative Ser/Thr protein phosphatase (slr2031) (Kamei et al. 1998) and Ser/Thr protein kinase (spkA) (Kamei et al. 2001). They do not appear to be linked directly to the pilus assembly. The third one includes a gene cluster of pixGHIJ1J2L (positive phototaxis, formerly pisGHIJ1J2L) (Yoshihara et al. 2000, Bhaya et al. 2001). The predicted products of pixG, pixH, pixI, pixJ1, pixJ2 and pixL show similarities to PatA, CheY, CheW, MCP (methyl-accepting chemotaxis protein), MCP and CheA, respectively, which take parts in a signal transduction system to switch flagellar rotation in bacterial chemotaxis (Aizawa et al. 2000). Furthermore, pixJ1 gene product bears a putative chromophore-binding region, which belongs to bacteriophytochrome (Yoshihara et al. 2000, Vierstra and Davis 2000). Thus, it was proposed that predicted products of pixGHIJ1J2L form a regulatory machinery of light perception and signal transduction pathway for the positive phototactic movement toward a light source in Synechocystis.

In the entire genome of Synechocystis, there are two more gene clusters that are homologous to the pixG cluster. Gene arrangement of these clusters is similar to that of the pixG cluster except the absence of a cheA-like ORF in one cluster. In this communication, we show that the cluster lacking a cheA-like ORF is involved in a regulatory process for assembly of the thick pilus structure, motility and natural transformation competency. We also identify novel split genes as the missing cheA. They are located far apart not only from the gene cluster but also from each other on the Synechocystis genome.

Results and Discussion

slr1041 gene cluster is essential for motility

In the entire genome of Synechocystis sp. PCC 6803, there are three homologous gene clusters that consist of a patA-like ORF, a cheY-like ORF, a cheW-like ORF, MCP-encoding ORF(s) and a cheA-like ORF (Fig. 1). However, there are some differences in gene arrangement among them. In the pixG cluster, which is involved in positive phototaxis, a tandem repeat of MCP-encoding genes (pixJ1 and pixJ2) was found (Yoshihara et al. 2000). On the other hand, a cheA-like ORF was missing in the slr1041 cluster, while it is present in the other clusters. We disrupted all the ORFs in the slr1041 and sll1291 clusters and found that the only slr1041 cluster was essential for motility.

Phototactic motility of the mutants was evaluated as movement of spotted cells on a soft agar plate under lateral illumination at 10 µE m^–2 s^–1 for 40 h (Fig. 2). Parent strain (PCC-P) used for gene disruption reproducibly moved to the light source, indicative of positive phototaxis (Fig. 2P). Mutant cells of slr1041 showed positive phototaxis like the wild type but their motility seemed to decline in some degree (Fig. 2A). Mutant cells of slr1042, slr1043 or slr1044 stayed at original spots with little spreading, indicative of non-motile phenotype (Fig. 2B, C, D). On the other hand, disruption of any genes in the sll1291 cluster (sll1291, sll1292, sll1293, sll1294 and sll1296) did not affect the phototactic motility (data not shown). The gene arrangement of the slr1041 cluster is similar to the pilGHIJKL cluster of P. aeruginosa that is required for twitching motility and assembly of type IV pili (Darzins 1993, Darzins 1994). Thus, we designated hereafter slr1041, slr1042, slr1043 and slr1044 as pilG, pilH, pilI, and pilJ, respectively, although pilK homolog could not be recognized in the Synechocystis genome.

cheA-like gene is split into two ORFs

By an homology search in the Synechocystis genome, we found slr0322 that is significantly homologous to pixL and another cheA-like ORF sll1296 (Fig. 3). Gene disruption showed that the slr0322 mutant was unambiguously non-motile (Fig. 2E). It is of note that orthologous genes including the cheA-like ORF are arranged similarly as a single cluster in the complete genome of Anabaena sp. PCC 7120; all0930 (pilG), all0929 (pilH), all0928 (pilI), all0927 (pilJ) and all0926 (pilL) (Kaneko et al. 2001). Domain analysis showed that Slr0322 had the dimerization domain, the CA (Catalytic and ATP-binding) domain, the regulation domain and two sets of receiver domains, while PixL and All0926 contained a single receiver domain. Strangely, Slr0322 lacked the N-terminal HPt (histidine-containing phosphotransfer) domain that is prerequisite to CheA-like proteins (Fig. 3). We searched again HPt-encoding ORF in the Synechocystis genome with the N-terminal HPt domain of All0926. It was found that a single-domain protein encoded by slr0073 is markedly homologous to the HPt of All0926 (40% identical). Sequence alignment of Slr0073 and Slr0322 with All0926 revealed that the autophosphorylatable His residue, the phospho-accepting Asp residues and other domains are highly conserved (Fig. 4). We disrupted slr0073 and found that it was also essential for motility (Fig. 5). It is, thus, concluded that a cheA-like pilL gene in Synechocystis was excised from the original pilG cluster and further split into at least two parts (slr0073 and slr0322) that are found at positions far apart from each other in the Synechocystis genome. We designate slr0073 and slr0322 as pilL-N and pilL-C, respectively. The sequence alignment also showed that a region between the HPt domain and the dimerization domain of All0926 did not correspond to Slr0073 or Slr0322. We searched the Synechocystis genome with this region of All0926 but could not find any clear homolog. Moreover, All0926 bears in this region a unique segment of eight tandem repeat sequences, each consisting of 36 amino acids (Fig. 3, 4). This repeat was not found in other predicted proteins of the Anabaena genome or even in the current non-redundant database.

Pilus structure of pil mutants

Pilus structure on the cell surface of each mutant was examined by electron microscopy after negative staining with phosphotungstic acid. As shown in Fig. 6A, on the surface of the parent PCC-P, two types of pili, the thick pili (long arrows) and the thin pili (short arrows) were observed in accordance with our earlier report (Yoshihara et al. 2001). The motile pilG mutant showed both thick pili and thin pili on the cell surface, which were indistinguishable from those of wild type (Fig. 6B). The non-motile pilH and pilL-N mutants showed the presence of the thick pili that were larger in length and more in number than the wild type, while the thin pili were seemingly unaffected (Fig. 6C, F). The non-motile pilI mutants retained the thick pili but in much reduced numbers, while the thin pili remained unchanged (Fig. 6D). The non-motile pilJ and pilL-C mutants almost lost the thick pili, while the thin pili were again unaffected (Fig. 6E, G). In some cases (e.g. Fig. 6 B, D, E), the remaining thin pili were attached together to form bundles as already described (Yoshihara et al. 2001). These results strongly supported the idea that accumulation of the thick pili, which are essential for phototactic motility, was more or less affected in the pil mutants of this study.

Transformation competency of mutants

It is well known that Synechocystis cells are naturally transformable to uptake extracellular DNA (Grigorieva and Shestakov 1982). We previously described that the typical pil gene mutants such as pilA1 and pilM, which were defective in motility and in assembly of the thick pilus structure, concomitantly lost ability of natural transformation. These results strongly suggested that the thick pili are essential for both motility and transformation (Yoshihara et al. 2001). Here we examined transformation competency of the mutants of the pilG cluster (Table 1). Under the experimental conditions we employed, wild-type cells gave the efficiency of about 0.42×10^–4 with our standard plasmid. The pilG mutant, which was motile and normally piliated, showed normal transformation competency. Competency of the pilH mutant, which lost motility but accumulated a greater amount of the thick pili, was about one-third of wild type. The pilJ and pilL-C mutants retained very slight competency (1.5% of wild type), although they showed non-motile phenotype and practically non-piliation in terms of the thick pilus structure. We could not estimate the competency of the pilI mutant, as it carried the same screening marker as the standard plasmid for the transformation experiments. These results showed that the transformation competency is well correlated with the degree of piliation as revealed by electron microscopy and motility as judged by colony morphology. At this moment, the competency is a more sensitive indicator than the other two for evaluation of the correlated mutant phenotype. It is obvious that these pil mutants more or less retained competency, in contrast with the complete absence in the pilA1 or pilM mutants (Yoshihara et al. 2001).

Signal transduction systems for motility in Synechocystis

Generally, MCP/CheA/CheY system is a major regulatory pathway of signal transduction for bacterial chemotaxis. In the unicellular cyanobacterium Synechocystis sp. PCC 6803, there are three sets of MCP/CheA/CheY systems (Fig. 1). Similar three sets of genes are also found in the genome of Anabaena sp. PCC 7120 (Kaneko et al. 2001), Nostocpunctiforme ATCC 29133 (http://www.jgi.doe.gov/JGI_microbial/html) and Thermosynechococcuselongatus strain BP-1 (Kaneko, T. and Tabata, S., personal communication) but not in Prochlorococcus spp. or Synechococcus sp. WH 8102. Based on hydoropathy profiles, putative transmembrane regions were predicted. A large hydrophilic region of MCPs can be assigned to be a sensory domain. The membrane topology and structural features of the sensory domains allow us to categorize cyanobacterial MCPs into three types (Fig. 7). PixJ1 and its homologs have multiple GAF domains that are predicted to bind a phytochrome-like chromophore (Yoshihara et al. 2000, Ohmori et al. 2001). The presumptive sensory domains of PilJ and Sll1294 differed in membrane topology, although they did not show any homology to known structures. Provided that the nascent N-terminus of a membrane-spanning protein resides on the cytoplasmic side of the plasma membrane, the presumptive sensory domain of PilJ was assumed to be on the cytoplasmic side like PixJ1, while that of Sll1294 was in the periplasmic space like the chemotaxis sensor MCPs (Moual and Koshland Jr. 1996).

PilT is a unique component of pilus assembly in cyanobacteria and heterotrophic bacteria, but its precise role or regulation has not yet been elucidated (Wall and Kaiser 1999). The pilT disruptant (pilT1 in Synechocystis) shows hyperpiliation (Whitchurch et al. 1991, Okamoto and Ohmori 1999, Bhaya et al. 2000), while the other pil disruptants show non-piliation (Yoshihara et al. 2001, Bhaya et al. 2000). Interestingly, the pilH and pilL-N mutants showed hyperpiliation, while the others in the pilG cluster showed more or less non-piliated phenotype (Fig. 6). The unique phenotype of hyperpiliation may suggest that both pilH and pilL-N regulate PilT1 activity directly or indirectly.

It is established that the CA domain and the HPt domain form a single His kinase of CheA (Bilwes et al. 1999, Mourey et al. 2001). However, mutants of PilL-N (HPt domain) and PilL-C (CA domain) showed the opposite phenotype in piliation. This may suggest that either PilL-N or PilL-C has unknown function in addition to the His kinase activity. It is also very likely that PilL-N/C His kinase phosphorylates CheY-like PilH, as judged from the conserved gene arrangement. In typical two-component regulatory systems, the phosphorylated form of response regulator proteins is active in their output domains and, therefore, the His kinase mutants show a phenotype practically identical to the mutants of cognate response regulators. By contrast, the opposite phenotype in piliation of our pilH and pilL-N/C mutants does not fit with this general idea and suggest that the non-phosphorylated forms of PilH and PilL-N may be active in positive regulation of PilT1. Further biochemical studies are needed to clarify such unusual features of PilL-N/C and PilH.

Materials and Methods

Culture and growth conditions

The motile strain of the unicellular cyanobacterium Synechocystis sp. PCC 6803 was obtained from Pasteur Culture Collection and a clone showing vigorous motility with positive phototaxis (PCC-P) was selected as a parent strain for gene disruption (Yoshihara et al. 2000). Cells were grown in liquid BG11 medium (Stanier et al. 1971) with air-bubbling containing 1% (v/v) CO₂ at 31°C at light intensity of 50 µE m^–2 s^–1. Kanamycin or spectinomycin was included at 20 µg ml^–1 when mutants were screened and maintained.

Insertional mutagenesis

The protein database derived from the Synechocystis genome was searched for homology using the BLAST program (Altschul et al. 1997). Each target ORF was amplified with a set of primers by polymerase chain reaction (Table 2). Amplified DNA of slr1041, slr1042, slr1044, slr0322 or slr0073 was cloned into the pT7Blue-T vector (Novagen, U.S.A.), and then interrupted at a unique restriction site by insertion of Tn5-derived kanamycin-resistant cassette in the same direction as the ORF. These constructs were designed not only to inactivate the target ORF but also to allow transcriptional expression of the downstream gene(s) from the promoter of aminoglycoside 3′-phosphotransferase in the cassette. A DNA fragment of slr1043 was amplified, cloned into pCR2.1 vector (Invitrogen, U.S.A.) and then interrupted at SmaI site with a spectinomycin-resistant cassette from the omega element (Prentki et al. 1991) in the same direction as slr1043. The spectinomycin cassette was also inserted to interrupt slr0073. Positive phototactic cells of Synechocystis were transformed with these DNAs as described previously (Hihara and Ikeuchi 1997). At least four independent transformants were selected and subjected to full segregation, which was confirmed by polymerase chain reaction (not shown).

Phototaxis and motility assay

Phototactic motility was examined by colony morphology on 0.8% (w/v) agar (Bacto-Agar, Difco, U.S.A.) as previously described with slight modifications (Yoshihara et al. 2001). For phototaxis and motility assay, cells in liquid culture at late log-phase (A₇₃₀=0.8–1.0) were concentrated to a cell density of 2×10⁹ cells ml^–1 by centrifugation and 1 µl suspension was spotted on the agar plate. Then, cells were cultivated at 31°C under lateral illumination with white fluorescent lamps of light intensity at 10 µE m^–2 s^–1 for 40 h. For the motility assay, single cells were spread on the soft agar plate and cultivated under non-directional illumination at 20–30 µE m^–2 s^–1 for 4 d.

Detection of pili by electron microscopy

Cells of each mutant cultivated on agar plates were collected by gentle suspension with BG11 medium. Pilus structures on the cell surface were examined after staining with 0.8% (w/v) phosphotungstic acid (pH 7.0) with transmission electron microscope (model 1200EX, JEOL, Tokyo, Japan) basically according to Yoshihara et al. (2001).

Transformation competency assay

Transformation competency was measured as described previously (Yoshihara et al. 2001). A test DNA carrying a genomic DNA of 3.1 kbp with insertion of the spectinomycin-resistant cassette (Hihara and Ikeuchi 1997) was introduced and transformants were selected on agar plates containing 20 µg spectinomycin ml^–1 after 1 d culture without antibiotics.

Acknowledgments

This work was supported by a Research Fellowship for Young Scientists from the Japan Society for the Promotion of Science (to S.Y.), by Grants-in-Aid for Scientific Research (to M.I.), by the Program for Promotion of Basic Research Activities for Innovative Biosciences of Japan (to M.I.) and by a Grant for Scientific Research from the Human Frontier Science program (to M.I.).

1

Corresponding author: E-mail, [e-mail]; Fax, [phone].

Fig. 1 Gene organization of the SynechocystispixG (formerly pisG) cluster, slr1041 cluster and sll1291 cluster. Homologous regulatory components for the chemotactic flagellar switching in enteric bacteria are shown as category. The direction of arrows indicates relative transcriptional orientation.

Fig. 2 Phototactic movements of colonies of the gene-disrupted mutants and their parent strain PCC-P. Each cell suspension of 1 µl was spotted and grown for 40 h under lateral illumination (arrowhead). P, wild type of positive phototaxis (PCC-P); A, Δslr1041; B, Δslr1042; C, Δslr1043; D, Δslr1044; E, Δslr0322. Dotted line shows the initial position before the illumination. Bar indicates 5 mm.

Fig. 3 Domain organization of Synechocystis PixL, PilL-N, PilL-C and Anabaena All0926. The conserved His residue (H) in the HPt domain and the conserved Asp residue (D) in the receiver domain are shown. HPt, histidine-containing phosphotransfer; CA, catalytic and ATP-binding.

Fig. 4 Sequence alignment of All0926 from Anabaena sp. strain PCC 7120 with Slr0073 (PilL-N) and Slr0322 (PilL-C) of Synechocystis. Black boxes indicate identical residues at the same position, in the receiver domain of PilL-C; residues which are identical to at least two sequences are emphasized. * and # indicate a specific His residue in the HPt domain and a specific Asp residue in the receiver domain, respectively. Note that only PilL-C harbors two receiver domains, while All0926 has only one domain (see Fig. 3). Domains defined in Fig. 3 are emphasized with thick underlines. Residues with thin underlines or dashed lines represent the repeat sequences of All0926.

Fig. 5 Colony morphology of the slr0073 mutant. Wild-type and slr0073 mutant cells were grown for 4 d on 0.8% agar plate. Panel A, wild type (PCC-P); B, Δslr0073. Bars show 0.2 mm.

Fig. 6 Electron micrographs of negatively stained cells of wild type and mutants. Panel A, wild type (PCC-P); B, Δslr1041; C, Δslr1042; D, Δslr1043; E, Δslr1044; F, Δslr0073; G, Δslr0322. Long and short arrows indicate the thick and the thin pili, respectively. Bars indicate 1 µm.

Fig. 7 Domain organization of PixJ1, PilJ and Sll1294 of Synechocystis. Key residues (Cys and His) for potential chromophore binding are shown in the GAF domains. Note that the putative N-terminus of PixJ1 (Sll0041) is longer than the originally annotated one (GenBank: AB077950).

Table 1

Transformation efficiencies of wild type and mutants

Strain	Transformation efficiency (%) a
PCC-P	100
ΔpilG	97.6
ΔpilH	28.6
ΔpilI	ND
ΔpilJ	1.5
ΔpilL-C	1.5

^a Value (%) represents the frequency of transformation relative to wild type. ND, not determined.

Table 2

Primers and positions of the insertional disruption for pil genes

Disrupted gene	Forward primer	Reverse primer	Position of insertional disruption / amplified DNA (bp)
slr1041	5-ACCATATGCAGGGAACCCTGAAC-3	5-GGGATGGAGATAATGATC-3	693 / 1196
slr1042	5-GAAATTTGTCGAATGGTCC-3	5-CTGGATATATCAAGTTGCC-3	432 / 1366
slr1043	5-GAAATTTGTCGAATGGTCC-3	5-CTGGATATATCAAGTTGCC-3	924 / 1366
slr1044	5-GCCATATGGCAACAGAAACCAATT-3	5-TGGCTTGGCTAAAGAGGG-3	444 / 913
slr0322	5-GGATGCAGGATTTATACG-3	5-TCTTTGGGCAAATAGTCC-3	492 / 989
slr0073	5-TTTAATTCGGGCCGTGGG-3	5-AACTAACGGGCGGCAGAA-3	501 / 1435

Note that the same sets of primers were used for disruption of slr1042/slr1043. This was because the neighboring ORFs in the same DNA fragment were insertionally interrupted at different positions.

Abbreviations

 
• ORF

  open reading frame.