The (PATAN)-CheY-Like Response Regulator PixE Interacts with the Motor ATPase PilB1 to Control Negative Phototaxis in the Cyanobacterium Synechocystis sp. PCC 6803

Abstract

The cyanobacterium Synechocystis sp. PCC 6803 can move directionally on a moist surface toward or away from a light source to reach optimal light conditions for its photosynthetic lifestyle. This behavior, called phototaxis, is mediated by type IV pili (T4P), which can pull a single cell into a certain direction. Several photoreceptors and their downstream signal transduction elements are involved in the control of phototaxis. However, the critical steps of local pilus assembly in positive and negative phototaxis remain elusive. One of the photoreceptors controlling negative phototaxis in Synechocystis is the blue-light sensor PixD. PixD forms a complex with the CheY-like response regulator PixE that dissociates upon illumination with blue light. In this study, we investigate the phototactic behavior of pixE deletion and overexpression mutants in response to unidirectional red light with or without additional blue-light irradiation. Furthermore, we show that PixD and PixE partly localize in spots close to the cytoplasmic membrane. Interaction studies of PixE with the motor ATPase PilB1, demonstrated by in vivo colocalization, yeast two-hybrid and coimmunoprecipitation analysis, suggest that the PixD–PixE signal transduction system targets the T4P directly, thereby controlling blue-light-dependent negative phototaxis. An intriguing feature of PixE is its distinctive structure with a PATAN (PatA N-terminus) domain. This domain is found in several other regulators, which are known to control directional phototaxis. As our PilB1 coimmunoprecipitation analysis revealed an enrichment of PATAN domain response regulators in the eluate, we suggest that multiple environmental signals can be integrated via these regulators to control pilus function.

Introduction

Cyanobacteria are widely distributed in many ecosystems spanning a large genetic and morphologic diversity. As ancestors of chloroplasts, they are to date the only known prokaryotes to gain their entire energy through oxygenic photosynthesis (Berman-Frank et al. 2003, Nowicka and Kruk 2016). To compensate for light fluctuations during the course of the day, certain cyanobacterial species have evolved the ability to move in a light-dependent manner, a phenomenon called phototaxis. The unicellular cyanobacterium Synechocystis sp. PCC 6803 (hereafter Synechocystis) exhibits directional surface-dependent motility using retractile type IV pili (T4P) (Bhaya et al. 2000, Yoshihara et al. 2001). Single cells detect the position of a light source by a focusing effect (Schuergers et al. 2016), leading to an asymmetric distribution of the T4P filaments (Nakane and Nishizaka 2017). Synechocystis harbors a variety of photoreceptors that are either directly or indirectly involved in the control of phototaxis and cover the entire visible spectrum (Masuda 2013, Wiltbank and Kehoe 2019). However, downstream signaling pathways and the specific targets of the respective regulators for the directional control of phototaxis remain largely elusive (Wilde and Mullineaux 2017).

To react to an environmental stimulus, such as light, or to perceive and transmit intracellular signals, bacteria have evolved complex regulatory networks facilitating a coordinated cell response. Two-component systems belong to the most important and widespread signal transduction mechanisms in bacteria. They typically comprise a histidine kinase and a cognate response regulator, which regulates gene expression or other cellular functions (Stock et al. 2000). The chemosensory system that controls flagellar motility is an atypical two-component pathway and very well studied in enteric bacteria, such as Escherichia coli (Eisenbach 1996). The CheY-like response regulator carrying an N-terminal phosphoacceptor or receiver domain (REC) accepts a signal from the histidine kinase CheA, leading to a phosphorylation-dependent conformational change in the response regulator. CheA indirectly senses input signals through the association with a methyl-accepting chemotaxis protein (MCP) and the adapter protein CheW (Sourjik and Wingreen 2012). The downstream function of the response regulator is in many cases determined by the output domains fused to REC in a modular fashion and ranges from the regulation of transcription, (de-)phosphorylation, biosynthesis or degradation of cyclic dimeric guanosine monophosphate and ligand binding to direct interaction with a target protein (Capra and Laub 2012). Originally, response regulators have been thought to be unique to bacteria and archaea. Subsequent studies, however, showed, that Arabidopsis thaliana harbors a membrane-bound ethylene receptor containing a domain, which in its monomeric state resembles the structure of the bacterial REC domain (Müller-Dieckmann et al. 1999). Up to now, response regulators have also been found in many eukaryotes such as Saccharomyces cerevisiae, Candida albicans and Dictyostelium discoideum (Wadhams and Armitage 2004).

T4P-based motility, including cyanobacterial phototaxis, is controlled by signal transduction pathways, which comprise proteins with homology to flagellar-mediated chemotaxis systems. Most importantly, Synechocystis phototaxis signal transduction pathways comprise a number of CheY-like response regulators that are genetically associated with photoreceptors (Schuergers et al. 2017). Six of these CheY-like response regulators belong to a family of response regulators of the PatA type that combine the C-terminal REC domain with a specific PATAN (PatA N-terminus) domain. Although the PATAN domain itself is found in many different bacteria, the PATAN-REC combination is specific for cyanobacteria (different species of Nostoc, Synechococcus, Thermosynechococcus, Anabaena and Synechocystis) and deltaproteobacteria such as Myxococcus xanthus and Geobacter sulfurreducens (Galperin 2006, Makarova et al. 2006). (PATAN)-CheY-like response regulators have been shown to be involved in chromatic adaptation (Chiang et al. 1992), interaction with circadian clock output proteins (Köbler et al. 2018), control of heterocyst pattern formation (Liang et al. 1992, Young-Robbins et al. 2010) and regulation of phototaxis (Yoshihara et al. 2000). In Synechocystis, three of the (PATAN)-CheY-like response regulators (PixG, LsiR and PixE) are involved in controlling the direction of phototaxis (Yoshihara and Ikeuchi 2004, Song et al. 2011, Sugimoto et al. 2017).

Previously it has been shown that PixE interacts in vitro in a blue-light intensity-dependent manner with the sensory input BLUF (blue light using flavin adenine dinucleotide) photoreceptor protein PixD (Tanaka et al. 2012), which is encoded approximately 150 bp downstream on the chromosome. Upon blue-light illumination, monomeric PixE is released from the dark-stable PixD₁₀–PixE₅ or PixD₁₀–PixE₄ complex (Yuan and Bauer 2008, Ren et al. 2013). A mutant lacking the photoreceptor PixD and hence exclusively harboring unbound PixE shows negative phototaxis under conditions where the wild type (WT) moves toward a light source, such as white illumination or light of wavelengths between 500 and 700 nm (Masuda and Ono 2004, Okajima et al. 2005). However, ΔpixE and ΔpixED mutants show positive phototaxis comparable to the WT, suggesting that free monomeric PixE reverses the direction of phototaxis (Sugimoto et al. 2017). PixE is a protein consisting of 380 amino acid residues. The C-terminus (residues 258–373) is similar to bacterial response regulators of the CheY type, whereas the PATAN domain is found at the N-terminus. Typical bacterial response regulators are phosphorylated at a conserved aspartate residue. In addition, a second aspartate and a lysine residue are important for the correct phosphorylation of enterobacterial CheY (Lukat et al. 1991). However, none of these three residues are conserved in PixE, suggesting that this response regulator is not phosphorylated by a histidine kinase as in a typical bacterial two-component system (Ren et al. 2013). Moreover, PixE is lacking a DNA-binding domain. Hence, downstream signaling upon light-induced dissociation of the PixD–PixE complex is most likely accomplished via direct protein–protein interaction of PixE with other proteins.

In rod-shaped bacteria such as M. xanthus, relocation of the pilus motor ATPases PilB (extension) and PilT (retraction) between the cell poles is a well-established model for direction switching (Bulyha et al. 2009). The spherical cells of Synechocystis, however, relocate PilB1 in a distinct crescent-like pattern to the front side into the direction of movement without limitation to a pole, thus enabling the bacterium to move in any direction (Schuergers et al. 2015). PilB1 binds the RNA chaperone Hfq, which localizes to the plasma membrane in a PilB1-dependent manner and is crucial for the function of T4P (Schuergers et al. 2014). The complete lack of motility of PilB1 deletion mutants underlines its prominent role as a mediator of directional phototaxis in Synechocystis. We, therefore, hypothesized that PilB1 might execute PixD–PixE signaling in Synechocystis by directly interacting with monomeric PixE.

To obtain a better understanding of the underlying mechanisms, we created complementation mutants of ΔpixD and ΔpixE strains with fluorescently labeled variants of PixD and PixE and studied the phototactic behavior as well as the localization pattern by fluorescence microscopy. Mutants expressing PixE showed negative phototaxis under any tested light conditions, whereas PixE deletion mutants showed a robust positive phototactic phenotype with the movement being faster and more directed toward the red-light source as compared with the WT. PixD and PixE localized in spots or crescents at the cytoplasmic membrane. Moreover, simultaneous visualization of PixE and PilB1 under phototaxis-inducing directional medium red-light conditions revealed partial colocalization of the two proteins. Direct interaction was confirmed by yeast two-hybrid and coimmunoprecipitation assays, thus corroborating the role of PixE as a mediator of negative phototaxis by binding to PilB1.

Results

PixE controls negative phototaxis in Synechocystis

To assess the influence of the (PATAN)-CheY-like response regulator PixE on the phototactic behavior of Synechocystis and to visualize its localization, we constructed plasmid-based complementation mutants of a previously described ΔpixE strain (Sugimoto et al. 2017). The upstream promoter sequence together with the coding region of pixE was cloned into the self-replicating vector pJRD215 bearing a kanamycin resistance cassette [ΔpixE (+pixE)]. In addition, a similar construct with a C-terminal mvenus sequence was created [ΔpixE (+pixE-mvenus)]. Fusion of the pixE promoter with the mvenus tag sequence alone served as a control [ΔpixE (+mvenus)]. Successful conjugation of ΔpixE was confirmed by PCR (data not shown).

We analyzed the phototactic behavior of the WT, ΔpixE and its complementation mutants in response to unidirectional weak red light in the absence or presence of vertical blue-light illumination. As previously reported (Sugimoto et al. 2017), the WT switched from positive to negative phototaxis when additional blue light was applied (Fig. 1). ΔpixE and the control strain, which also lacks PixE but expresses a fluorescent tag from the pixE promoter, showed enhanced positive phototaxis toward red light with respect to the WT (Fig. 1). In addition, switching from positive to negative phototaxis upon addition of blue light was suppressed in both strains. Hence, expression of the mVenus fluorescent tag from the plasmid pJRD215 does not alter the observed phenotype of the deletion mutant. Interestingly, both strains ΔpixE (+pixE) and ΔpixE (+pixE-mvenus) showed a robust negative phototactic phenotype under any light conditions tested. Since both mutants exhibited a similar behavior with comparable migration distances under both light conditions, we conclude that the mVenus protein tag is not altering the function of PixE. Negative phototaxis might be due to the overexpression of PixE from the introduced vector resulting in free PixE monomers under normal conditions.

Fig. 1

Phototaxis assay of Synechocystis WT and mutants in response to red and blue light. The cells were spotted onto BG-11 plates (0.3 µM Cu²⁺) solidified with 0.5% (w/v) agar. The vertical lines indicate the start position of the spotted cells. The plates were incubated under unidirectional red light (λ_max = 625 nm; fluence rate = 30 µmol photons m⁻² s⁻¹) displayed by the red arrow. As illustrated, the phototactic response was recorded after 7 d in the presence or absence of vertically illuminated blue light (λ_max = 475 nm; fluence rate = 150 µmol photons m⁻² s⁻¹).

We next wanted to elucidate the phototactic response on a single-cell level. Suspensions of the abovementioned strains were spotted onto agarose plates, and phototactic movement in response to lateral red-light illumination with or without vertical blue light was observed under an upright microscope. We recorded 3-min time-lapse videos and performed tracking measurements and quantification of the cell trajectories (Table 1). For that, we calculated the total velocity and the linear velocity of the cells. The former is derived from the total distance traveled (the sum of each single step), and the latter is derived from the linear distance traveled (the beeline between start and end points of the complete track). We next tested the directionality of the phototactic behavior of the different strains. Therefore, we calculated the final angle between each cell trajectory end point and the light source. Next, we defined 12 angular sections of 30° each and assigned the individual cells to their respective segment. The radial percentage distributions of the cells with respect to the incident light were visualized as polar plots (Fig. 2). We conducted a Rayleigh test of uniformity for each dataset and compared the mean resultant path lengths r, with r = 1 (perfect directional movement) and r = 0 (random movement). Without blue-light illumination, the r values of ΔpixE (0.95) and ΔpixE (+mvenus) (0.96) were significantly higher than those of the WT (0.82), with the movement being more directed toward the red-light source. In addition, the total velocity and the linear velocity of both mutants were higher than those of the WT and less cells were immotile. In concurrence with the results of the macroscopic phototaxis assay, the WT switched from positive to negative phototaxis upon additional vertical blue-light irradiation. Cells moved in a less directed fashion (r = 0.63) and with reduced velocity, but a larger proportion of cells were motile. Addition of blue light did not have any effect on the pronounced positive phototactic phenotype of ΔpixE or the control strain ΔpixE (+mvenus). The two strains expressing PixE [ΔpixE (+pixE) and ΔpixE (+pixE-mvenus)] on the other hand exhibited negative phototaxis under both light conditions, with comparable r values and velocities to that of the WT. Similar to the WT, the percentage of immotile cells was reduced upon blue-light illumination. In summary, overaccumulation of PixE always leads to negative phototaxis, whereas cells lacking PixE show a more directed and faster positive phototaxis toward a light source. Thus, PixE is a mediator of negative phototaxis and the abundance of PixE in the cell seems to be important for this response.

Fig. 2

Single-cell angular distribution of Synechocystis WT and mutants in response to red and blue light. Cells from phototactically trained cultures were diluted with BG-11 medium to an OD₇₅₀ of approximately 1.0 and spotted onto BG-11 µ-dishes (0.3 µM Cu²⁺) solidified with 0.3% (w/v) agarose. Phototactic movement in response to lateral red-light illumination (λ_max=625 nm; fluence rate=30 µmol photons m⁻² s⁻¹) in the presence (+BL) or absence (−BL) of vertically irradiated blue light (λ_max=475 nm; fluence rate=150 µmol photons m⁻² s⁻¹) was observed under an upright microscope. Three-minute time-lapse videos were acquired after 15 min of illumination. The angular percentage distribution of the cells was plotted for −BL (grey) and +BL (light red) conditions, with the overlap depicted in dark red. n, number of cells; r, mean resultant path length from a Rayleigh test.

Table 1

Comparison of motility characteristics in a microscopic red-light phototaxis assay

Strain	Immotile cells (%)		Total velocity (µm s−1)		Linear velocity (µm s−1)
	−BL	+BL	−BL	+BL	−BL	+BL
WT	5.5	2.5	+0.14	−0.09	+0.08	−0.03
ΔpixE	0.8	1.5	+0.20	+0.16	+0.14	+0.10
ΔpixE (+mvenus)	0.9	0.2	+0.21	+0.19	+0.15	+0.13
ΔpixE (+pixE)	13.6	2.3	−0.10	−0.10	−0.05	−0.04
ΔpixE (+pixE-mvenus)	6.6	0.9	−0.13	−0.13	−0.06	−0.05

Listed are the percentage of immotile cells and the total velocity and the linear velocity of different strains depending on the illumination conditions.

+, positive phototaxis; −, negative phototaxis; BL, blue light.

Spot-like localization of PixD and PixE at the cell membrane

To further dissect the cellular function of PixE in Synechocystis, we investigated the localization of PixE and its sensory input BLUF photoreceptor protein PixD by fluorescence microscopy and immunoelectron microscopy, respectively. Therefore, we generated a complementation mutant of a ΔpixD strain (Masuda and Ono 2004) expressing a C-terminally tagged PixD-eYFP from a neutral site in the chromosome under control of the promoter PpetJ, which is induced under copper limitation (Kuchmina et al. 2012, Schuergers et al. 2014). Inducibility and functionality of the fusion protein were confirmed by Western blot analysis with an α-PixD antibody and phototaxis assays (Fig. 3). Under inducing conditions, ΔpixD (+pixD-eyfp) exhibited a similar PixD expression level as the WT, whereas no PixD was detected at higher copper concentrations (Fig. 3A). In accordance, the phototactic phenotype of the complementation mutant changed gradually in a copper-dependent manner from negative under repressing conditions (resembling ΔpixD) to positive WT-like phototactic response under inducing conditions (Fig. 3B).

Fig. 3

Inducible rescue of a pixD deletion mutant with a functional eYFP-tagged PixD. WT and ΔpixD (+pixD-eyfp) were cultured at Cu²⁺ concentrations 0, 0.3 and 2.5 µM. ΔpixD cultured at (A) 0.3 µM Cu²⁺ or (B) 0, 0.3 and 2.5 µM Cu²⁺ was used as a negative control. Expression of PixD-eYFP is controlled by the PpetJ promoter, which is inducible by the depletion of copper in the medium. (A) Analysis of expression levels of PixD and PixD-eYFP by Western blotting with α-PixD antibody. (B) Phototaxis assay on BG-11 plates solidified with 0.5% (w/v) agar under different copper concentrations. The plates were incubated under unidirectional white light provided by a fluorescence lamp at a fluence rate of approximately 80 µmol photons m⁻² s⁻¹ for 7 d. The light direction is indicated by the arrow. The circles indicate the start position of the spotted cells.

Then, we studied the expression pattern of the PixD-eYFP fusion protein on a single-cell level by fluorescence microscopy (Fig. 4). Under strongly inducing copper depletion conditions, bright PixD-eYFP signals were detected as spots close to the cell membrane (Fig. 4B, C). Under mild inducing conditions of 0.3 µM Cu²⁺, some of the cells showed more than one PixD-eYFP spot (Fig. 4D). The localization pattern was the same as under strong inducing conditions, with the spots localized close to the cell membrane (Fig. 4E). PixD-eYFP expression was inhibited almost completely at 2.5 µM Cu²⁺ (Fig. 4F). When we analyzed the fluorescence signals from the tagged PixE protein in the ΔpixE (+pixE-mvenus) mutant, we found an approximately congruent localization pattern with spots or crescents localized along the cell membrane (Fig. 4H), resembling the expression pattern of PixD.

Fig. 4

Spot-like localization of PixD-eYFP and PixE-mVenus at the cell membrane detected by fluorescence microscopy. (A) ΔpixD cultured at 0.3 µM Cu²⁺ was used as a negative control, with red autofluorescence originating from chlorophyll. (B–F) ΔpixD (+pixD-eyfp) was cultured at Cu²⁺ concentrations 0, 0.3 and 2.5 µM, whereas (G) ΔpixE (+mvenus) and (H) ΔpixE (+pixE-mvenus) were cultivated under standard conditions (0.3 µM Cu²⁺). (C) Same as (B) but with a lower detection level. (E) Magnified picture section of (D). Scale bars as indicated. Detection level: +=low; ++=medium; +++=high. All images are merged red and yellow channels.

To confirm the localization results of PixE with another approach, we performed immunoelectron microscopy of a strain expressing a C-terminally FLAG-tagged version of PixE. We quantified the signals of the detected gold particles and found two clusters, with the larger one being again close to the cell membrane (Supplementary Fig. S1).

Colocalization and direct interaction of PixE and PilB1 in vivo

To test a hypothesized interaction of PixE with the pilus base and especially with PilB1, we simultaneously visualized both proteins by expressing PixE-mVenus and PilB1-GFP in a ΔpilB1 mutant. The cellular abundance of free monomeric PixE in this strain is presumably higher than in the WT, due to expression from a plasmid. We studied the localization patterns of the proteins under directional weak red-light conditions (fluence rate = 1 µmol photons m⁻² s⁻¹) as well as phototaxis-inducing directional medium red-light conditions (fluence rate = 50 µmol photons m⁻² s⁻¹). After light exposure, cells were fixed with glutaraldehyde and investigated by fluorescence microscopy (full-size images see Supplementary Fig. S2). The colocalization of PilB1-GFP and PixE-mVenus was analyzed using the Coloc 2 plugin in ImageJ. Quantification of the signal overlap by calculating the Pearson’s r value confirmed the mostly exclusive localization to opposite sides of the cell of PilB1-GFP and PixE-mVenus under weak light conditions (Fig. 5A) and the partial spot-like colocalization under phototaxis-inducing conditions, respectively (Fig. 5B).

Fig. 5

Quantification of colocalization of PilB1-GFP and PixE-mVenus under different light conditions. ΔpilB1 (+pilB1-gfp+pixE-mvenus) was cultured at 0 µM Cu²⁺ to induce the expression of PilB1-GFP. Localization of PilB1-GFP (green) and PixE-mVenus (magenta) was observed after (A) directional illumination of weak red light (fluence rate=1 µmol photons m⁻² s⁻¹) and after (B) phototaxis-inducing directional medium red-light conditions (fluence rate=50 µmol photons m⁻² s⁻¹). (A, B) Scale bars=1 µm. All images are merged green and yellow channels depicted as green and magenta, respectively. (C) Quantification of the Pearson’s correlation coefficient of all images with ImageJ reveals the mean r values of −0.30±0.04 for weak light conditions and 0.27±0.06 for phototaxis-inducing conditions (P = 0.0004).

We next wanted to verify the possible PixE–PilB1 interaction in a yeast two-hybrid assay. Interaction was assessed through growth on selective medium (Fig. 6). We found self-interaction of PixE and PilB1 and interaction between PixE and PilB1. However, PixE did not interact with the pilus retraction motor ATPase PilT1 in the yeast two-hybrid analysis (Fig. 6). Interaction was independent of the C-terminal or N-terminal position of the tag in all cases.

Fig. 6

Interaction study of PixE, PilB1 and PilT1 in a yeast two-hybrid assay. Yeast strain AH109 was transformed with two plasmids encoding PixE, PilB1 or PilT1 C- or N-terminally fused to either BD (GAL4 DNA-binding domain) or AD (GAL4 activation domain). Selection of transformants carrying both plasmids was achieved by plating on complete supplement medium lacking leucine and tryptophan (−Leu −Trp). BD and AD alone and KaiA dimer interaction were used as negative and positive controls, respectively. Protein–protein interaction was investigated by growth on leucine-, tryptophan- and histidine-depleted (−Leu −Trp −His) complete supplement selection medium. Sensitivity of the selection-gene-dependency was controlled by the addition of 12.5 mM 3-AT, which functions as a competitive inhibitor of the HIS3-gene product. Direct interaction of PixE–PixE (blue rectangle), PilB1–PilB1 (yellow rectangle), PilT1–PilT1 (red rectangle) and PixE–PilB1 (green rectangle) was identified.

To obtain a better understanding of the regulatory network of PilB1, we carried out a coimmunoprecipitation assay and screened for further potential interaction partners by mass spectrometry. Expression of PilB1-GFP in the ΔpilB1 (+pilB1-gfp) mutant was confirmed by fluorescence microscopy (Fig. 7A). A Synechocystis strain expressing FLAG-GFP was used as a control (Fig. 7B). Solubilized cell extracts were purified and examined by Western blot analysis with an α-GFP antibody (Fig. 7C). The elution fractions were then analyzed by liquid chromatography tandem mass spectrometry (LC–MS/MS). The normalized datasets (Supplementary Table S1) were used to calculate the enrichment factor for each detected protein (Supplementary Fig. S3). We detected a large amount of enriched proteins known to be either directly or indirectly involved in light signaling and motility, such as photoreceptors PixJ1, Cph1 and Cph2, pilus motor ATPases PilT1, PilT2 and PilB2, different histidine kinases and several CheY-like and (PATAN)-CheY-like response regulators. The motility-essential Hfq protein (Dienst et al. 2008), which has been shown to directly interact with PilB1 (Schuergers et al. 2014), was enriched 27-fold, thus validating the significance of our findings. PixE scored highest among the identified response regulators with a 12-fold enrichment with respect to the control. These results confirm the PixE–PilB1 interaction detected by fluorescence microscopy and yeast two-hybrid assays, implying the role of PixE as a mediator of negative phototaxis in Synechocystis. To achieve a more quantitative screening of the copurifying proteins, we carried out a protein function classification using DAVID 6.7 (Huang et al. 2009a, Huang et al. 2009b). We found eight functional clusters that were significantly enriched (Supplementary Table S2). The cluster with the highest enrichment score consisted of various response regulators and transcription factors, with a high abundance of PATAN-type proteins such as PixE, PixG, TaxP2, PilG and Slr1594 (Fig. 7D and Supplementary Fig. S4).

Fig. 7

Screening for PilB1 interaction partners by coimmunoprecipitation and LC–MS/MS. ΔpilB1 (+pilB1-gfp) and WT (+FLAG-gfp) as control strain were cultured in copper-depleted BG-11 medium to an OD₇₅₀ of approximately 1. Robust expression of (A) crescent-like PilB1-GFP and (B) cytoplasmic FLAG-GFP was observed under a fluorescence microscope. (A, B) Scale bars=5 µm. (B) Solubilized cell extract ‘sCE’ was purified with µMACS™ Anti-GFP MicroBeads, and a Western blot was performed with α-GFP antibody. The elution fraction ‘E’ was analyzed by LC–MS/MS. (D) Protein function classification of the copurifying proteins using the complete Co-IP dataset as background revealed eight significantly enriched clusters. The cluster with the highest enrichment score consisted of various response regulators and transcription factors, with a high abundance of PATAN-type proteins such as PixE, PixG, TaxP2, PilG and Slr1594.

Discussion

In this study, we provide in vitro and in vivo evidence for the interaction of the (PATAN)-CheY-like response regulator PixE with the pilus assembly motor ATPase PilB1. Based on colocalization studies under different light regimes, we suggest that the interaction in Synechocystis is to some extent light dependent, as colocalization was observed exclusively under phototaxis-inducing conditions, but not under weak light. In vivo, the interaction between both proteins depends on the availability of free monomeric PixE resulting from blue-light-dependent dissociation of the PixD–PixE complex. Conclusively, abundance of released PixE in the presence of strong blue light changes the direction of phototaxis from positive to negative. Directional switching of the WT in response to vertically illuminated intense blue light has been attributed to an excess of active monomeric PixE in an earlier study (Sugimoto et al. 2017). This is in agreement with the observations in our phototaxis assays.

In this study, we used two different assays to assess the phototactic response of Synechocystis and different mutants. Macroscopic phototaxis assays are long-term experiments. Thus, cell behavior is not only dependent on the incident light but also on secondary effects such as alterations of expression levels of proteins, nutrient status and photosynthetic performance. Quantification and comparison of macroscopic to microscopic phototactic responses can be challenging and ambiguous. Therefore, in this study, macroscopic assays served primarily to assess the general phenotype of a mutant, especially for the verification of successful mutant complementation. By contrast, microscopic phototaxis assays were used for the quantification of velocity (linear and total), directionality and percentage of moving cells excluding additional long-term effects.

We quantified the phototactic response of PixE deletion and complementation mutants to unidirectional red light with or without additional vertical blue light. The absence of PixE led to an enhanced positive phototactic response with respect to the WT under all tested light conditions, with cells moving faster and more directed toward the red-light source. In addition, switching to negative phototaxis upon blue-light irradiation was completely abolished in the mutants, leading to the conclusion that PixE is essential for the promotion of negative phototaxis under the tested conditions. As previously reported (Sugimoto et al. 2017), a PixE–PixD double mutant exhibited the same phenotype as the ΔpixE strain (data not shown), confirming that PixE is sufficient to account for the described phenotype. The reduced speed and directionality of the WT in comparison to the ΔpixE mutant are most likely due to the presence of some monomeric PixE even in the absence of blue light, indicating the role of PixE as an inducer of negative phototaxis by the inhibition of a basal positive phototactic response. In previous studies, similar phototactic phenotypes related to (PATAN)-CheY-like response regulators have been described. Downregulation of LsiR in response to ethylene also leads to enhanced and more directed positive phototaxis (Lacey and Binder 2016;Kuchmina et al. 2017), whereas induction of lsiR gene expression upon UV-A sensing by the same receptor UirS/ETR induces negative phototaxis (Song et al. 2011). Deletion of a response regulator involved in positive phototaxis PixG leads to a negative phototactic phenotype (Yoshihara et al. 2000). Thus, the abundance and ratio of specific (PATAN)-CheY-like response regulators in the cell seem to control the direction of movement.

In our study, mutant strains expressing native or tagged PixE [ΔpixE (+pixE) and ΔpixE (+pixE-mvenus)] moved away from red light, irrespective of additional blue-light illumination. Speed and directionality were comparable to the WT upon additional blue-light illumination (Fig. 2). This suggests that, in the created complementation mutants, there is an overexpression of PixE that cannot be compensated by PixD availability. Although pixE gene expression was under the control of its native promoter, a higher copy number of the expression plasmid could be an explanation for the accumulation of PixE. Since we are lacking an α-PixE antibody, this remains speculative. Nonetheless, the more than compensated phenotypes of the complementation strains strongly imply the functionality of PixE in these mutants. Importantly, a PixE–PixD double mutant is still able to move directionally, proving that PixE is not required for the directional response itself. A ΔpixD mutant has been reported to exhibit positive phototaxis under very weak light conditions below 15 µmol photons m⁻²s⁻¹ (Sugimoto et al. 2017). This effect cannot be attributed to the amount of free PixE in ΔpixD, which in this mutant should be light independent. Apparently, under weak light, the PixD–PixE response is overruled by another system competing with PixE for PilB1 binding.

When we tested the localization of PixD and PixE by fluorescence microscopy, we detected very similar localization patterns with the proteins clustering spot-like or crescent-like along the cytoplasmic membrane. Even though we did not assess colocalization, it can be speculated that, under dark or weak light conditions, both proteins occupy the same space in vivo and form a complex that is located close to the plasma membrane. Similar localization patterns of photoreceptors have been found in rod-shaped bacteria such as Synechococcus elongatus (Kondou et al. 2002, Yang et al. 2018), with clear localization to the cell poles. However, immunolocalization of FLAG-tagged PixE using electron microscopy revealed multiple spots of PixE in one cell (Supplementary Fig. S1). Multiple PixE signals are also seen in the colocalization analysis with GFP-tagged PilB1 (Fig. 5). At this stage, we cannot exclude that different tags and the level of overexpression have an effect on the specific localization or clustering of PixD and PixE. Nonetheless, light-dependent (co)localization of PixD/PixE and PixE/PilB1 at the plasma membrane could provide a reasonable explanation for the directional switching of phototaxis (Fig. 8).

Fig. 8

Proposed model for the directional switching of phototaxis in Synechocystis. (A) Under diffuse low light, the BLUF photoreceptor protein PixD forms a stable complex with the (PATAN)-CheY-like response regulator PixE at the cell membrane. Hence, PixE is unable to interact with the motor ATPase PilB1. Hexameric PilB1 is distributed evenly at the pilus base along the cell membrane. This results in a symmetric assembly and disassembly of T4P, leading to a random movement of the cell. (B) Upon lateral red illumination, Synechocystis focusses the incident light into a sharp focal point at the back side of the cell. By a yet unidentified mechanism, PilB1 relocalizes to the front side of the cell facing the light source. This leads to an asymmetric distribution of T4P and a movement toward the red-light source. (C) When in addition to lateral red illumination cells are subjected to strong diffuse blue illumination, the PixD–PixE complex dissociates, leading to free monomeric PixE in the cell. The same effect is achieved by the overexpression of PixE or disruption of pixD. PixE can now interact with PilB1. We hypothesize that this interaction leads to an activation of T4P at the back side of the cell and/or an inactivation of T4P at the front side of the cell, resulting in a directional switching of phototaxis away from the red-light source.

Surface motility using T4P needs the coordination of multiple pili to move into a certain direction. To change the direction, rod-shaped bacteria, such as M. xanthus, use a molecular clock mechanism to switch the location of T4P between the two cell poles. This is mainly based on alternate binding of PilB and PilT ATPases to one cell pole (Bulyha et al. 2009). The coccoid bacterium Neisseria gonorrhoeae is peritrichously piliated and performs a random movement with directional persistence. Coordination is explained by a tug-of-war principle including a directional memory (Marathe et al. 2014). By contrast, we have shown that Synechocystis moves in a very directional way toward a light source by sensing a focal point, which is formed at the distal side of the cell due to the refraction of light by the cell itself (Schuergers et al. 2016). So far, it is unclear how dynamic localization of pilus proteins and assembly and disassembly of the complex are connected with light sensing in Synechocystis. In the motile cyanobacterium Nostoc punctiforme, specific localization of a PilG homolog to the membrane has been observed, correlating with the localization of pilus proteins (Risser et al. 2014, Cho et al. 2017). In Pseudomonas, it is suggested that the phosphorylated CheY-like response regulator homolog PilG binds to PilB (and PilZ and FimX, which are not conserved in Synechocystis), whereas phosphorylated PilH interacts with PilT/PilU, thereby regulating pilus extension and retraction using a chemotaxis signal transduction system (Sampedro et al. 2015).

In this study, we could show that the (PATAN)-CheY-like response regulator PixE, which is controlled by the BLUF photoreceptor protein PixD, interacts with the main motor ATPase PilB1 and switches the direction of movement. Furthermore, in mass spectrometry analysis of the PilB1-enriched eluate, a variety of proteins were identified as additional putative PilB1 interaction partners, many of which playing a role in light-dependent signaling or motility. It should be noted that many of the detected proteins might turn out to be false-positive hits or were detected due to indirect interaction with other proteins. For example, the predicted interaction of PilB1 with the retraction ATPase PilT1 could not be verified in a yeast two-hybrid assay. Nevertheless, confirmation of the published Hfq–PilB1 interaction (Schuergers et al. 2014) validates the significance of our results. Interestingly, many of the known CheY-like response regulators were identified in this analysis, suggesting that they can also interact with the pilus base to control motility. In general, control of directional motility seems to be more complex in Synechocystis than in the newly isolated motile cyanobacterium S. elongatus (Yang et al. 2018). This organism contains two chemotaxis operons, but only tax1, encoding an MCP-like photoreceptor and homologs of CheY, CheA and CheW, controls directional motility as several mutants in this operon move in a random way. This is very different from Synechocystis where all known phototaxis mutants either change the direction of movement or inhibit motility completely, and no randomly moving mutant has been described so far. In contrast to S. elongatus, multiple photoreceptors (including non-MCP proteins) regulate the phototaxis of Synechocystis cells, thereby integrating different light (and ethylene) signals. In conclusion, phototaxis in Synechocystis is a tightly regulated light-dependent process likely relying on abundance as well as phosphorylation state of the involved CheY-like response regulators, which seemingly represent the key elements that transduce the multiple signals to control T4P motor function.

Materials and Methods

Culture conditions and strains

WT Synechocystis sp. PCC 6803 substrain PCC-P (Yoshihara et al. 2000) and mutants were cultivated in BG-11 medium or on BG-11 agar plates (Rippka et al. 1979) supplemented with 10 mM TES buffer (pH 8.0) and 1 mM Na₂S₂O₃ (Thiel et al. 1989) at 30°C under continuous white-light illumination of 50 µmol photons m⁻² s⁻¹ provided by a fluorescent lamp (Philips TLD Super 80/840). Escherichia coli was grown in lysogeny broth medium at 37°C. When appropriate, Synechocystis mutants were cultured at a final concentration of 40 µg ml⁻¹ kanamycin, 14 µg ml⁻¹ chloramphenicol, 5 µg ml⁻¹ streptomycin or 5 µg ml⁻¹ zeocin and E. coli was cultured at a final concentration of 100 µg ml⁻¹ ampicillin, 40 µg ml⁻¹ kanamycin, 24 µg ml⁻¹ chloramphenicol, 25 µg ml⁻¹ streptomycin or 25 µg ml⁻¹ zeocin.

Primers for mutant construction and constructed plasmids are listed in Supplementary Tables S3 and S4, respectively. The construction of Synechocystis ΔpixD and ΔpixE has been previously described (Masuda and Ono 2004, Sugimoto et al. 2017). For complementation, a C-terminally tagged pixD-eyfp was introduced in a neutral locus of the ΔpixD mutant by transformation with pSK9-PpetJ-pixD-eyfp. PixD was amplified from genomic Synechocystis WT DNA using the primer pair NdeI-pixD-fw and XhoI-pixD-rev introducing NdeI and XhoI restriction sites. The fragment was ligated into the NdeI and XhoI restriction sites of the pSK9-based expression vector (Kuchmina et al. 2012) bearing a C-terminal eyfp fragment flanked by XhoI and BglII restriction sites, respectively (Schuergers et al. 2014). Gene expression in the mutant ΔpixD (+pixD-eyfp) is under the control of the copper repressible promoter PpetJ.

Complementation of ΔpixE was achieved by triparental conjugation (Zinchenko et al. 1999) with pJRD215-PpixE-pixE or pJRD215-PpixE-pixE-mvenus. As a control, ΔpixE was conjugated with pJRD215-PpixE-mvenus. At first, the promoter region of pixE (PpixE) was amplified with the primer pair promoPixE-F-EcoRI and promoPixE-R-NdeI and the fragment was cloned into the EcoRI-NdeI-cut pJRD215 (Davison et al. 1987) creating the plasmid pJRD215-PpixE. The pixE-coding region was amplified with the primer pair PixE-F-NdeI and PixE-BamHI-R, and the fragment was cloned into the NdeI-BamHI-cut pJRD215-PpixE to obtain pJRD215-PpixE-pixE. For the construction of pJRD215-PpixE-pixE-mvenus, coding regions for pixE and mvenus were separately amplified by two primer pairs, BamH1-PixE-F and PixE-R and PixE-LINK-mV-F and mV-sph1-R, respectively. The two fragments were mixed and cloned together into the BamHI-SphI-cut pHSG299 (TaKaRa, Japan). The resultant plasmid was then used as a PCR template to amplify the fragment encoding pixE-mvenus with the primer pair PixE-F-NdeI and mVenus-R-BamHI. The amplified fragment was cloned into the NdeI-BamHI-cut pJRD215-PpixE to obtain pJRD215-PpixE-pixE-mvenus. For the construction of pJRD215-PpixE-mvenus, the fragment containing mvenus and pJRD215-PpixE was amplified by the primer pair promoPixE-mV-F and promoPixE-R, followed by circularization by In-Fusion Cloning kit reagents (Clontech, Germany) to obtain pJRD215-PpixE-mvenus. Sequences of all inserted DNA fragments and correct assembly of all plasmids were checked by sequencing.

The Synechocystis strain expressing PixE-FLAG was constructed as below. We previously constructed a plasmid containing full-length pixE and pixD, the regions 1-kbp upstream of pixE and 1-kbp downstream of pixD and a SmaI restriction site just after the pixD stop codon (Sugimoto et al. 2017). Using this plasmid as a template, the 1-kbp upstream of pixE- and pixE-coding regions and the 1-kbp downstream of pixD- and pixD-coding regions were separately amplified by PCR with two primer pairs, PixE-UP-EcoRI-Fsion2 and PixE-R-FLAG and PixED-F-FLAG and PixD-Down-R-SphI-Fsion2, respectively. The two fragments were mixed and then cloned into EcoRI-SphI-cut pUC18 using In-Fusion Cloning kit reagents. The resultant plasmid was digested with SmaI and ligated with a SmaI-cut spectinomycin/streptomycin resistance gene cassette. The resultant plasmid pUC18-PpixE-pixE-FLAG was transferred into a ΔpixDE strain (Sugimoto et al. 2017), and mutants were selected on BG-11 plates containing spectinomycin.

ΔpilB1 (+pilB1-gfp) (Linhartová et al. 2014, Schuergers et al. 2014) served as genetic background for the construction of ΔpilB1 (+pilB1-gfp+pixE-mvenus) strain obtained by conjugation with pJRD215-PpixE-pixE-mvenus.

Immunoblot analysis

Synechocystis WT and ΔpixD (+pixD-eyfp) were cultured in BG-11 medium at different copper concentrations (0, 0.3 and 2.5 µM). As a control, ΔpixD was grown under standard conditions (0.3 µM CuSo₄). Cells were harvested by centrifugation and washed with phosphate-buffered saline (PBS). The pellet was resuspended in PBS, and cells were disrupted in a mixer mill. The cell lysate was used for immunoblot analysis following standard procedures using α-PixD antibody.

Macroscopic phototaxis assays

Macroscopic phototaxis assays were performed as previously described (Jakob et al. 2017) with the following exceptions. Cells from freshly grown plates were resuspended in BG-11 and spotted on solidified 0.5% (w/v) agar plates infused with BG-11 medium with 0, 0.3 or 2.5 µM CuSO₄ lacking glucose. The phototactic behavior of the cells was examined under lateral white-light illumination from a halogen lamp (∼80 µmol photons m⁻² s⁻¹) or lateral red-light illumination (λ_max = 625 nm; fluence rate = 30 µmol photons m⁻² s⁻¹) from a light-emitting diode (LED) with or without additional vertically illuminated blue light (λ_max = 475 nm; fluence rate = 150 µmol photons m⁻² s⁻¹).

Microscopic phototaxis assays and tracking measurements

Single-cell motility experiments were performed as previously described (Jakob et al. 2017) with the following exceptions and additions. Cells from actively moving cultures were diluted with BG-11 medium to an OD₇₅₀ of approximately 1.0 and spotted onto BG-11 µ-dishes (0.3 µM Cu²⁺) solidified with 0.3% (w/v) agarose. Phototactic movement in response to lateral red-light illumination (λ_max = 625 nm; fluence rate = 30 µmol photons m⁻² s⁻¹) in the presence or absence of vertically irradiated blue light (λ_max = 475 nm; fluence rate = 150 µmol photons m⁻² s⁻¹) was observed under an upright microscope (Nikon Instruments, Japan) equipped with a 40× objective lens. Three-minute time-lapse videos with one frame every 3 s were acquired after 15 min of illumination. Cell tracking was performed using BacteriaMobilityQuant (Schuergers et al. 2016) that runs on MATLAB Runtime. To determine the directionality of the phototactic response, the final angular distribution of the raw tracks was analyzed using Wolfram Mathematica.

Fluorescence microscopy and quantification of colocalization

ΔpixD (+pixD-eyfp) and ΔpixE (+pixE-mvenus) cells were spotted onto BG-11 medium containing 0.8% (w/v) agar and different Cu²⁺ concentrations as indicated, and illuminated laterally with white light provided by a fluorescent lamp (∼50 µmol photons m⁻² s⁻¹) for 2 d. The 0.25% glutaraldehyde was directly spotted onto cells showing phototaxis to fix protein localization. Cells were then harvested and suspended in a buffer containing 137 mM NaCl, 2.7 mM KCl, 8.1 mM Na₂HPO₄ and 1.5 mM KH₂PO₄. Localization of PixD-eYFP and PixE-mVenus fluorescence signals in the cells was analyzed under a confocal laser-scanning microscope LSM780 (Zeiss, Germany). Emission from 490 to 508 nm by 488 nm excitation was observed.

To quantify the colocalization of green and yellow fluorescence emitted from PilB1-GFP and PixE-mVenus under weak (∼1 µmol photons m⁻² s⁻¹) or medium red-light conditions (∼50 µmol photons m⁻² s⁻¹) provided by an LED (Sanyo, Japan), we performed a pixel intensity spatial correlation analysis using the Coloc 2 plugin in ImageJ. For each condition, the Pearson’s correlation coefficient was calculated and averaged over three images, with +1 for perfect correlation, 0 for no correlation and −1 for perfect anticorrelation. The P value was calculated using Student’s t-test.

Immunogold labeling and transmission electron microscopy

We collected cells expressing PixE-FLAG, which moved toward lateral white-light illumination on an agar surface. WT cells were used as control. Cells were washed in 0.1 M PBS (pH 7.4). After fixing with 0.5% glutaraldehyde and 2% paraformaldehyde in 0.1 M PBS (pH 7.4), cells were embedded in LR White at 52°C for 3 d followed by standard dehydration procedure. For both WT and PixE-FLAG-expressing mutant, ultrathin sections (70-nm thickness) were collected on a formvar-coated transmission electron microscopy (TEM) nickel grid. For WT, the sections were post-stained with 2% uranyl acetate and Reynolds’ lead citrate prior to TEM examination. We used a Tecnai T12 operating at 120 kV, equipped with a Gatan Ultrascan 2k × 2k CCD camera for TEM imaging. For cells expressing PixE-FLAG, we conducted post-embedding immunogold labeling. Sections were first incubated with aldehyde blocking buffer for 5 min followed by incubation with blocking buffer for 30 min. Then, sections were incubated with monoclonal α-FLAG (mouse) antibody overnight at 4°C in the dark. After rinsing with PBS and washing buffer, sections were incubated with secondary α-IgG + IgM (mouse) antibody labeled with 10-nm goat-polyclonal gold particles overnight at 4°C in the dark. After washing with PBS and distilled water, sections were fixed with 1% glutaraldehyde for 15 min and contrasted with 2% uranyl acetate and Reynolds’ lead citrate prior to TEM examination.

Yeast two-hybrid assay

Primers for the construction of yeast expression plasmids and vectors used for the yeast two-hybrid assay are listed in Supplementary Tables S3 and S4, respectively. PixE, pilB1 and pilT1 sequences were amplified from genomic Synechocystis WT DNA introducing restriction sites via primers as indicated in Supplementary Table S3 and subsequently cut with the appropriate restriction enzymes. PCR fragments were ligated into the plasmids pCGADT7ah (Rausenberger et al. 2011), pGADT7ah (Hiltbrunner et al. 2005), pD153 (Shimizu-Sato et al. 2002) and pGBKT7 (Clontech) previously cut with the appropriate restriction enzymes, creating either N- or C-terminally tagged GAL4 activation domain (AD/prey) or GAL4 DNA-binding domain (BD/bait) fusion proteins. The plasmids pCGAD-kaiA-AD and pD153-kaiA-BD were used as positive control (Köbler et al. 2018).

Cotransformation of yeast strain AH109 (Clontech) was achieved using the Frozen-EZ Yeast Transformation II Kit (Zymo, Germany) following the manufacturer’s guidelines. Transformants containing bait and prey plasmids were selected on complete supplement mixture (CSM) dropout medium (MP Biomedicals, Germany) without leucine and tryptophan (−Leu −Trp). After incubation at 30°C for 4 d, cells were screened for interaction by streaking on CSM dropout medium lacking leucine, tryptophan and histidine (−Leu −Trp −His) supplemented with 12.5 mM 3-amino-1,2,4-triazole (3-AT; Roth, Germany) and incubated at 30°C for 6 d.

Coimmunoprecipitation assay and mass spectrometry

ΔpilB1 (+pilB1-gfp) and WT (+FLAG-gfp) as control strain (Schuergers et al. 2015) were grown in 100 ml of copper-depleted BG-11 medium and harvested at an OD₇₅₀ of approximately 1 by centrifugation. The pellets were washed and resuspended in buffer [50 mM HEPES (pH 7.0), 5 mM MgCl₂, 25 mM CaCl₂, 150 mM NaCl, 10% (v/v) glycerol and 0.1% (w/v) Tween20] supplemented with protease inhibitors [250 µg ml⁻¹ 4-(2-aminoethyl)benzenesulfonyl fluoride hydrochloride, 625 µg ml⁻¹p-aminobenzamidine and 5 mg ml⁻¹ 6-aminohexaonic acid]. Cell disruption was performed in a mixer mill with a 1:1 mixture of glass beads (0.1–0.11 and 0.25–0.5 mm), and crude cell extracts were solubilized with n-dodecyl-β-d-maltoside for 1 h. The supernatant was purified with µMACS™ Anti-GFP MicroBeads (Miltenyi Biotec, Germany) following the manufacturer’s guidelines, and immunoblot analysis with α-GFP antibody was performed following standard procedures. The elution fraction was analyzed by LC–MS/MS. Datasets (Supplementary Table S1) were normalized using MaxLFQ intensity in MaxQuant (Cox et al. 2014). Enrichment factors were calculated as label-free quantification ratios of both samples. Proteins with an enrichment factor of <5 were classified as noninteracting background. We grouped proteins that were detected in both samples and proteins that were exclusively detected in the PilB1-GFP-expressing mutant into two different categories. These enriched proteins (enrichment factor ≥ 5) were subjected to a protein function analysis using DAVID 6.7 set to the highest classification stringency (Huang et al. 2009a, Huang et al. 2009b). For each of the eight detected clusters, the enrichment score was retrieved using either the complete dataset of proteins detected by LC–MS/MS (Co-IP background) or the entire proteome of Synechocystis (Synechocystis background) as background (Supplementary Table S2).

Funding

Excellence Initiative of the German Research Foundation [Spemann Graduate School (GSC-4, 517) to A.J.], [(WI2014/8-1) to A.W] and by the Japan Society for the Promotion of Science [KAKENHI (19H04719) to S.M.].

Acknowledgments

We thank Dennis Zimmer and Christin Köbler for expression plasmids and Nibedita Priyadarshini for help with the Co-IP experiments.

Disclosures

The authors have no conflicts of interest to declare.