Introduction

Bacterial cells may be surrounded by a capsule composed primarily of carbohydrate polymers or by a protein surface layer (S-layer), or both (Sleytr and Messner 2000). The crystalline S-layer is the outermost cell envelope component in many bacteria and archaea. Generally, the S-layer comprises a single protein or glycoprotein and completely covers the cell surface at all stages of bacterial growth. Ranging in thickness from 5 to 25 nm, S-layers have a lattice structure exhibiting identically sized pores with diameters of 2–8 nm. S-layer proteins, ranging in apparent molecular mass from 40 to 200 kDa, are among the most abundant cellular proteins (Sleytr and Messner 2000). The middle and C-terminal regions of S-layer proteins show low sequence identity. S-layers are reported to provide prokaryotic cells with a selective advantage by functioning as a protective coating, in cell adhesion, for surface recognition, and as ion traps.

Repeat-in-toxin (RTX) proteins are exotoxins produced by Gram-negative bacteria (Ludwig 1996). RTX proteins are considered to form pores in the cytoplasmic membranes of erythrocytes, leukocytes, and other cells, leading to the modification of cellular functions and/or lysis of host cells. RTX proteins are characterized by the GGXGXDXUX nonapeptide motif (X, any amino acid; U, an amino acid with a large hydrophobic side chain), which serves as two half-sites for Ca2+ binding. An array of the sequences forms a parallel β-roll structure (Baumann et al. 1993). Biochemical and molecular biological studies have best characterized HlyA of Escherichia coli (Wiles et al. 2008).

Cyanobacteria are ubiquitous microorganisms and conduct oxygenic photosynthesis in various environments. S-layers have been observed in 60 strains of cyanobacteria (Šmarda et al. 2002). SwmA, an S-layer protein of Synechococcus sp. WH8102, is involved in cell movement (swimming motility), and the S-layer of Synechococcus sp. GL24 has been reported to act as a template for fine-grain gypsum and calcite formation (Schultze-Lam et al. 1992; Brahamsha 1996). The functions of S-layers in other cyanobacteria have not been clarified.

We have reported that a glucose-tolerant (GT) strain of the cyanobacterium Synechocystis sp. PCC 6803, which is commonly used in studies, overproduces hemolysin-like protein (HLP) Sll1951, an RTX protein localizing in the S-layer (Sakiyama et al. 2006). However, its function is unclear. In the present study, we constructed two sll1951 (hlp) mutants, namely GDkF1 and GD, and used these to analyze the function of HLP. We demonstrate that HLP protects Synechocystis cells from growth inhibition by heavy metals, antibiotics, and osmotic stress.

Materials and methods

Synechocystis strains and culture conditions

A GT strain of Synechocystis sp. PCC 6803 was used as the wild-type strain throughout this study (refering to Williams 1988). The construction of the two hlp mutant strains, GDkF1 and GD, is described below. These cells were grown at 30°C in BG11 medium containing 20 mM HEPES–NaOH (pH 7.5) with continuous shaking at 90 rpm on a rotary shaker and under continuous illumination provided by incandescent lamps at an intensity of 70 μmol photons m−2 s−1 (Sakiyama et al. 2006). The GDkF1 and GD cells were grown in the presence of 25 μg ml−1 kanamycin and spectinomycin, respectively, for the selection of mutants. Escherichia coli strain JM109 (TaKaRa Bio, Ohtsu, Japan) was grown in Luria–Bertani (LB) medium containing the appropriate antibiotics at 37°C and used as a host for genetic manipulation.

To measure growth rates under various conditions, cultures of wild type and mutant cells were initiated at OD750 0.04–0.05, and the growth rate was calculated from the change in OD750 per hour. To test the effect of heavy metal stress, wild type and GD mutant cells were grown separately in medium containing 0–5 μM CuSO4, 0–10 μM CdCl2, or 0–10 μM ZnCl2. As BG11 medium usually contains 250 μM CaCl2, CaCl2 solution was added to Ca2+-free BG11 medium to prepare the medium containing various concentrations of Ca2+. To test the effect of antibiotic stress, 0–5 μg ml−1 ampicillin or kanamycin was added to the medium. To test the effect of osmotic stress, 0–1 M (final concentration) sorbitol was added to the medium.

Disruption and deletion of the hlp gene

The hlp gene was disrupted by insertional mutagenesis as described previously (Williams 1988). Briefly, DNA fragments containing part of the hlp gene were amplified from the chromosomal DNA of wild-type cells by polymerase chain reaction (PCR) using the primers GTCGACTTTGGGACGTTTCTGAGCCC and ACTAGTTCAGAGAGTTTAGGCGTAGA for the construction of GDkF1 and CAACCTCCAAACTGCTTTGGAAACCG and ACTAGTTCAGAGAGTTTAGGCGTAGA for the construction of GD.

The PCR fragment for GDkF1 was cloned into pT7Blue (Invitrogen, Carlsbad, CA), and a kanamycin-resistant gene cassette was introduced with the EZ::TN<KAN-2> system (Epicentre, Madison, WI). The PCR fragment for the GD mutant was cloned into pT7Blue (Invitrogen), which was cut at two HapI sites in the insertion sequence to substitute 2,390 bp of the coding region of sll1951 with a spectinomycin-resistance cassette. The resultant plasmids were propagated respectively in E. coli JM109, and the sites and direction of the inserted cassettes were confirmed by sequencing. Eight nucleotides in the coding region of sll1951 cloned in GDkF1 were substituted from the published sequence (Kaneko et al. 1996). The E. coli transformants were grown at 37°C in LB medium supplemented with 25 μg ml−1 kanamycin (for GDkF1) or 25 μg ml−1 spectinomycin (for GD). The purified plasmids were used, respectively, to transform wild-type cells and produce the hlp-disrupted mutant GDkF1 and the hlp-deficient mutant GD (Online Resource 1) by homologous recombination (Williams 1988). The presence of the inserted cassettes in the coding region of the hlp gene was confirmed by PCR.

Western blot analysis

The expression of HLP in the hlp-mutants was evaluated by Western blot analysis. Cell proteins were separated by SDS–PAGE and electro-transferred onto a polyvinylidene fluoride membrane. Rabbit anti-HLP serum (Sakiyama et al. 2006) and peroxidase-conjugated anti-rabbit IgG (Bio-Rad, Hercules, CA) were used as primary and secondary antibodies, respectively. After blocking with 5% (w/v) bovine serum albumin in TBST buffer [50 mM Tris–HCl (pH 8.0), 150 mM NaCl, 0.05% (w/v) Tween 20], the membrane was incubated with primary antibody at 1:1,000 dilution in buffer and then with secondary antibody at 1:1,600 dilution in the same buffer. Immunoreactive HLP was detected using dianisidine solution (10 mM Tris–HCl, pH 7.5, 0.57 mM o-dianisidine, 1% methanol, and 1.6% H2O2).

Measurement of survival rate under high temperature and desiccation stresses

Wild type and mutant cells at late-logarithmic phase (OD750 = 1) in liquid culture were used for determining the survival rate under high-temperature and desiccation stresses. To test tolerance to high-temperature stress, cells in 0.5 ml of liquid culture in 1.5-ml centrifuge tubes were incubated at 40.5, 50, 55.5, 60.5, and 69.5°C in a water bath for 5 min. The initial cell numbers were 3.0 × 107 and 2.8 × 107 cells ml−1 for wild type and GDkF1 cells, respectively. After incubation, the cells were treated with reagent from the Live/Dead BacLight Bacterial Viability kit (Molecular Probes, Eugene, OR). The living (green fluorescence) and dead cells (red fluorescence) were counted using a hemocytometer under an optical microscope (Eclipse E600; Nikon, Tokyo, Japan) with epifluorescence. The assays were carried out three times with the same cultures. The survival rates were calculated as the ratio of the number of living to total cells. To test tolerance to desiccation stress, cells in liquid culture at late-logarithmic phase (5 μl) were incubated and air-dried at 30°C under illumination (10 μmol photons m−2 s−1) in a 1.5-ml capless tube covered with a Petri dish for 0, 1, 2, and 3 h. The initial cell numbers were 3.0 × 107 and 3.2 × 107 cells ml−1 for wild type and GD mutant cells, respectively. After incubation, the numbers of living and dead cells were counted as described above. The assays were performed twice, and the survival rate was calculated as described above.

109Cd uptake assay

The 109Cd uptake activity was determined using the silicone–oil-layer centrifugation technique (Obata et al. 2004). First, the silicone–oil layer was prepared as a mixture of SH 550 and SH 556 at a ratio of 2:3 (Dow Corning Toray Silicone, Tokyo, Japan) in a 0.4-ml microcentrifuge tube (30-mm diameter and 115-mm length; no. 72.700; Assist, Tokyo, Japan). Then, 80 μl of the silicone–oil layer were placed at the bottom of the tube. For the 109Cd uptake experiment, 300 μl of cell suspension of wild type and GD cells were transferred from a pre-culture into a 1.5-ml centrifuge tube containing Tween 20 (final concentration, 0.05%), and the reaction was initiated by addition of 10 μl of 109Cd-chloride (126.17 MBq mg−1; PerkinElmer, Waltham, MA). The numbers of wild type and GD cells were adjusted to be the same with BG11 medium. At appropriate intervals, 200 μl of the each cell suspension were removed and layered onto the silicone–oil layer in separate microcentrifuge tubes. Then, the reaction was terminated by immediate centrifugation of the microcentrifuge tubes at 10,000×g for 1 min; the cells were passed through the silicone–oil layer and were separated from the reaction mixture. The tube was quickly frozen in liquid nitrogen. To determine the radioactivity in the culture medium and cell pellet, the tube was cut with a razor at the position of the silicone oil layer, and the radioactivity in each fraction was determined using a gamma counter (AccuFLEX γ7000; ALOKA, Japan). These experiments were performed three times under each condition.

Determination of cell size by flow cytometry

The sizes of the wild type and GD cells in liquid culture at late-logarithmic phase (OD750 = 1) were measured by flow cytometry (FACSCalibur, Becton–Dickinson, Franklin Lakes, NJ). The cells (1 ml) were diluted with 1 ml of sheath fluid solution (Becton–Dickinson) and analyzed. Beads of 2 μm and 6 μm (Polysciences, Warrington, PA) were analyzed first as controls. The results were analyzed using Cell Quest software (Becton–Dickinson). The FSC values (relative cell sizes) of the cells and beads were determined from histograms, and the sizes of both cell types were estimated by comparison with the FSC values of the beads.

Electron microscopy

The method for electron microscopy was according to Yubuki et al. (2007) with slight modification. Cell suspensions were mixed with an equal volume of 1% glutaraldehyde in sodium cacodylate buffer (0.2 M (CH3)2AsO(OH)-NaOH, pH 7.2) and incubated at 4°C for 2 h. After washing with cacodylate buffer, the cells were fixed by incubation in 1% osmium tetroxide at 4°C for 2 h. After washing with the buffer, the cells were dehydrated through a series of ethanol concentrations: 50% ethanol for 1 h, followed by 75, 90, 95, and 99.5% ethanol for 30 min each. For freeze-substitution fixation, the cells were fixed by incubation in anhydrous acetone containing 2% osmium tetroxide at −80°C for 48 h, after fixation in 1% glutaraldehyde at 4°C. Then, these cells were warmed to room temperature at a rate of 1°C min−1 with a pause at −20°C for 2 h and at 4°C for 1 h, washed, and dehydrated in 99.5% anhydrous acetone for 1 h. After dehydration with ethanol or anhydrous acetone, the cells were treated with a 1:1 mixture of 99.5% ethanol or anhydrous acetone and propylene oxide twice for 10 min each, and then with 99.5% propylene oxide twice for 10 min each. The propylene oxide was replaced with Spurr’s resin (Spurr 1969) by treatment with a 1:1 mixture of propylene oxide and resin, followed by treatment with the resin. The resin was polymerized at 70°C for 12 h. Thin sections (50-nm thickness) were cut with an ultramicrotome (EM-Ultracuts, Reichert, Germany) and stained with 2% uranyl acetate and lead citrate (Reynolds 1963). The samples were observed under a transmission electron microscope (JEM1010; JEOL, Tokyo, Japan).

Results and discussion

Influence of hlp mutation on the response to heat and desiccation stresses

We first speculated that HLP protects cells against high-temperature stress, in analogy to HecA and SigB in Synechocystis sp. PCC 6803 (Singh et al. 2006), because we had previously found that heat treatment above 60°C changes the conformation of HLP and releases the bound Ca2+ (Sakiyama et al. 2006). The survival rate of wild-type cells at high temperatures of 40.5–69.5°C was almost the same as that of GDkF1 cells (Online Resource 2A and B), suggesting that HLP does not protect wild-type cells against heat stress. Furthermore, the survival rate under desiccation stress was similar between wild-type cells and GD mutant cells (Online Resource 2C and D), indicating that HLP does not function as a protectant against desiccation stress.

Influence of hlp mutation on the response to heavy metal stress

In a previous study, some HLP was released from the surface of wild-type cells into the culture medium when the cells were exposed to more than 3 μM CuSO4 (Sakiyama et al. 2006); hence, we speculated that HLP may be involved in tolerance to metal ion stress. Retardation of the growth rate was more obvious in GD cells than in wild-type cells when the CuSO4 concentration in the medium was increased (Online Resource 3A and E). In addition, the concentration of CuSO4 required for 50% growth inhibition (ID50) was higher for wild-type cells than for GD cells in two experiments (Table 1).

Table 1 Comparison of ID50 values of heavy metals for growth inhibition in wild type and HLP-deficient mutant (GD) cells of Synechocyatis sp. PCC 6803
Full size table

The inhibitory effect of CdCl2 (above 1 μM) on the growth rate was more obvious in GD cells than in wild-type cells, and the difference became greater with time (Online Resource 3B and F). A similar inhibitory effect was observed with ZnCl2 at concentrations above 0.5 μM (Online Resource 3C, G). Furthermore, the ID50 values of CdCl2 and ZnCl2 were higher for wild-type cells than for GD cells (Table 1). These results suggest that HLP may function to protect cells from heavy metal stress.

HLP has been shown to bind Ca2+ at a ratio of 100 Ca2+ per HLP molecule, with 29 Ca2+ binding to one Ca2+-binding motif of HLP (Sakiyama et al. 2006). Assuming that HLP serves as a Ca2+ reservoir, we examined the effect of Ca2+ on the growth of wild-type and GD cells. However, no effect was observed in either cell line at Ca2+ concentrations up to 2 mM, under any conditions examined (Table 1 and Online Resource 3D). These results demonstrate that HLP has no function associated with calcium utilization.

HLP may reduce the ability of toxic metals such as Cu2+, Cd2+, and Zn2+ to enter cells, given that carboxyl and phosphate groups on cell surface molecules have been shown to absorb toxic metals in the cyanobacterium Spirulina (Chojnacka et al. 2005). HLP could bind with metals via free carboxyl groups of a polypeptide or attached polysaccharide. HLP does not appear to present a physical barrier against non-ionic molecules, as even myoglobin can pass through the S-layer (Sára et al. 1992). We speculated that in the wild-type cells of Synechocystis sp. PCC 6803 (GT strain), the toxic influence of heavy metals is initially reduced by HLP acting a diffusion barrier, and then enzymes are induced for more effective detoxifying mechanisms using the two-component system of the environmental stress response (Murata and Suzuki 2006). Further studies are necessary to elucidate the defense system against heavy metal stress in Synechocystis sp. PCC 6803.

Influence of hlp mutation on the response to antibiotics and osmotic stresses

The inhibitory effect of ampicillin or kanamycin on cell growth was more obvious in mutant cells than in wild-type cells at ampicillin or kanamycin concentrations higher than 0.1 μg ml−1, and the inhibitory effect increased with time (Online Resource 4A, B, D, and E). Moreover, the ID50 values of ampicillin and kanamycin were higher in the wild-type cells than in the mutant cells at all times in Experiment 1 and Experiment 2 (Table 2). These results suggest that HLP functions to reduce the inhibitory effect of ampicillin and kanamycin. Polysaccharides attached to HLP could form a biofilm matrix and thereby suppress the diffusion of antibiotics into the cell (Silverstein and Donatucci 2003).

Table 2 Comparison of ID50 values of antibiotics or sorbitol for growth inhibition in wild type and HLP-deficient mutant (GD) cells of Synechocyatis sp. PCC 6803
Full size table

Compared with GD cell growth, the growth of wild-type cells was more resistant to osmotic stress created by sorbitol (Online Resource 4C, F). The ID50 for sorbitol was higher in wild-type cells than in mutant cells at all times in Experiment 1, Experiment 2 (Table 2). The S-layer possesses characteristics of an exoskeleton and may be important for maintaining shape (Engelhardt 2007). Furthermore, liposomes harboring the S-layer protein of Bacillus stearothermophilus had an enhanced ability to maintain their shape against mechanical stress (Mader et al. 1999). Thus, HLP may function through an unknown mechanism to maintain cellular structure despite low turgor pressure in Synechocystis sp. PCC 6803 cells, allowing time for the induction of enzymes via the two-component system (Murata and Suzuki 2006). The defense system of Synechocystis sp. PCC 6803 cells under osmotic stress needs to be clarified.

109Cd uptake by wild-type and GD mutant cells

To investigate the resistance of wild-type cells to the effects of heavy metals, a 109Cd uptake assay was performed in wild type and GD cells. The radioactivity absorbed by wild type and GD cells reached a plateau within about 1 min in each experiment (Fig. 1a). After 1 min, 90% of the radioactivity was easily removed from both wild type and GD cells by treatment with 5 mM EDTA (Experiment 2 in Fig. 1b), suggesting that the 109Cd2+ was attached mainly on the cell surface. The ~10% of the 109Cd2+ that remained with the cells after EDTA treatment might have been incorporated into the cells. The radioactivity level associated with wild-type cells was 1.3- to 2.0-fold that associated with GD cells for the first 10 min in each experiment. However, after 1,500 min, the level of radioactivity was the same for both wild type and GD cells. Furthermore, the radioactivity that was removed by washing with EDTA decreased less in wild-type cells than in GD cells (Fig. 1b). These results indicate that while wild-type cells absorbed more Cd2+ than GD cells, more Cd2+ was incorporated into GD cells than into wild-type cells. Using an in vivo atomic absorption technique, we have previously demonstrated the binding of HLP with heavy metals (Sakiyama et al. 2006). Taken together, these data strongly support our proposal that HLP functions as a barrier against heavy metals and other toxic compounds.

Fig. 1
figure 1

109Cd incorporation and absorption by wild type and hlp-deleted mutant (GD) cells of Synechocystis sp. PCC 6803 during growth in the presence of Cd. a The amount of 109Cd radioactivity (cpm) associated with wild type and GD cells after incubation for the indicated times. b Change in 109Cd radioactivity expressed as percentage of the maximum. The maximum value in each experiment is 100%. Cells at logarithmic phase were harvested and used for a 109Cd uptake assay. Arrows indicate the times at which cells were treated with 5 mM EDTA to wash off the Cd bound to the outside of the cells. Cell numbers were adjusted (Experiment 1, 1.3 × 107 cells; Experiment 2, 3.8 × 106 cells; Experiment 3, 3.6 × 106 cells) with BG11 medium. The assays were carried out three times in three cultures grown independently. The means with standard deviations are indicated in each column

Full size image

The sizes of the wild type and GD cells were almost equal (2.33 ± 0.33 and 2.25 ± 0.62 μm, respectively), suggesting that the amount of extracellular material did not make a difference in the binding to HLP. Liu and Curtiss (2009) showed that 7 μM Ni2+ did not influence cell size in Synechocystis sp. PCC 6803. These results suggest that the difference in 109Cd binding between the wild type and GD cells may be explained by the presence or absence of HLP.

Electron micrographs showed that the wild-type cells had cell surface structures and an S-layer containing a high amount of electron-dense material (Fig. 2a). Unfortunately, GD mutant cells aggregated during fixation with osmium tetroxide at 4°C, preventing electron microscopic observation of the cell surface. With the freeze-substitution method of fixation, the S-layer of wild-type cells was ambiguous, and that of mutant cells was not observed; no clear photographs could be taken (data not shown). McCarren et al. (2005) showed cell surface structures (cell membrane, peptidoglycan, outer membrane, and S-layer) of the cyanobacterium Synecococcus sp. WH8102, but no S-layer in mutant cells. Several studies of the bacterial cell surface have reported that heavy metals bind to anions of uronic acid residues of S-layer material (De Philippis et al. 2001) or anionic amino acid residues of S-layer proteins (Merroun et al. 2005; Tang et al. 2009). The evidence suggests that HLP functions in defending cells from heavy metal toxicity (Fig. 2b).

Fig. 2
figure 2

Schematic model of the function of HLP in Synechocystis sp. PCC 6803. a An electron micrograph of a wild-type cell. CM cytoplasmic membrane, PP periplasm, PG peptidoglycan, OM outer membrane. b and c A schematic presentation of the function of HLP in wild type (b) and GD cells (c) under heavy metal stress. Minus (−) indicates negative charges of the carboxyl and sulfate groups in HLP and the outer membrane, as demonstrated by Panoff et al. (1988). Note that the S-layer protein of Synechocystis sp. PCC 6803 is suggested to be a hexamer in vivo (Vaara 1982; Šmarda et al. 2002), although the soluble form of HLP is a trimer (Sakiyama et al. 2006). Small black and gray circles represent Ca2+ and heavy metal ions, respectively. Heavy metal ions more easily penetrated GD cells than wild-type cells because of the absence of HLP in the former

Full size image

This is the first report showing that HLP functions as a barrier against various environmental and chemical stresses in vivo, although various functions have been reported for the S-layer in different organisms (Engelhardt 2007). Exopolysaccharides (EPS) of cells have been shown to prevent cellular access of certain antibiotics (Gilbert et al. 1997) and to sequester metals and toxins (Decho 1990; Flemming 1993). HLP in the S-layer may also function as a barrier against predators such as cyanophages and virulent bacteria. The morphological properties of the S-layer in vivo and the molecular nature of HLP are closely associated with HLP function. Further molecular, physiological, and morphological studies will be necessary to elucidate the functions of HLP, the S-layer, and HLP-binding polysaccharides.