Tlr1612 is the major repressor of cell aggregation in the light-color-dependent c-di-GMP signaling network of Thermosynechococcus vulcanus

Cyclic diguanylate (c-di-GMP) is a bacterial second messenger involved in sessile/motile lifestyle transitions. We previously reported that c-di-GMP is a crucial inducer of cell aggregation of the cyanobacterium Thermosynechococcus vulcanus. The three cooperating cyanobacteriochrome photoreceptors (SesA/B/C) regulate cell aggregation in a light color–dependent manner by synthesizing/degrading c-di-GMP. Although a variety of c-di-GMP signaling proteins are encoded in cyanobacterial genomes, how c-di-GMP signaling networks are organized remains elusive. Here we experimentally demonstrate that the cellulose synthase Tll0007, which is essential for cell aggregation, binds c-di-GMP although the affinity is low (Kd = 63.9 ± 5.1 µM). We also show that SesA—the main trigger of cell aggregation—is subject to strict product feedback inhibition (IC50 = 1.07 ± 0.13 µM). These results suggest that SesA-produced c-di-GMP may not directly bind to Tll0007. We therefore systematically analyzed all 10 of the genes encoding proteins containing a c-di-GMP synthesis/degradation domain. We identified Tlr1612, harboring both domains, as the major repressor of cell aggregation under the repressing teal-green light irradiation. tlr1612 acts downstream of sesA and is not regulated transcriptionally by light color, suggesting that Tlr1612 may be involved in c-di-GMP amplification in the signaling cascade. Post-transcriptional control is likely crucial for the light-regulated c-di-GMP signaling.


Results
The cellulose synthase Tll0007 binds c-di-GMP via its PilZ domain. We previously reported that both c-di-GMP and the PilZ domain of Tll0007 are essential for inducing cell aggregation of T. vulcanus at 31 °C 21,22 . However, experimental verification of c-di-GMP binding by the PilZ domain of Tll0007 was lacking. To assess c-di-GMP binding to Tll0007, in the present study we designed c-di-GMP biosensor proteins using a fluorescence resonance energy transfer (FRET)-based system by concatenating two fluorescent proteins (mCyPet and mYPet, which are monomerized variants of CFP and YFP, respectively) with the PilZ domain of Tll0007 in between (Fig. 1A), according to 28,29 . For each purified biosensor protein, the addition of c-di-GMP resulted in a corresponding increase in FRET efficiency, whereas addition of GTP did not (Fig. 1B). The c-di-GMP concentration-responsive curve revealed a dissociation constant (K d ) of 63.9 ± 5.1 µM at 31 °C (Fig. 1C). Curve fitting revealed a Hill coefficient of 0.90 ± 0.14, indicating non-cooperative c-di-GMP binding.
The 601 RXXXR 605 motif in PilZ domains is necessary for binding c-di-GMP [30][31][32] . We thus created the mutant R605A and assessed the binding of c-di-GMP. R605A lacked the corresponding increase in FRET efficiency upon c-di-GMP addition (Fig. 1C), indicating that c-di-GMP binding depends on Arg605 in the RXXXR motif of Tll0007. R605A had somewhat higher FRET efficiency than the unmodified wild-type (WT) PilZ domain in the absence of added nucleotides, suggesting that R605A exists in a locked conformation that is slightly activated. These results suggested that Tll0007 binds c-di-GMP via its PilZ domain.
The DGC activity of SesA exhibits product feedback inhibition. SesA is the major trigger of cell aggregation in T. vulcanus by producing c-di-GMP under blue-light irradiation 21,22 . The DGC activities of many GGDEF proteins are negatively regulated by their product, i.e., c-di-GMP 3,33 . To assess any potential product feedback inhibition of SesA, we measured the initial rate of the DGC reaction in the presence of various concentrations of c-di-GMP using a pyrophosphate assay kit, which monitors production of pyrophosphate-the byproduct of the DGC reaction. The purified protein of SesA was pre-irradiated with blue light to retain it in its active form during the assay. In the absence of c-di-GMP, WT SesA showed high initial pyrophosphate production activity, followed by a gradual decrease in production with increasing c-di-GMP concentration, which eventually stopped after ~100 s (at approximately 5 µM c-di-GMP) ( Fig. 2A). Conversely, in the presence of 10 µM c-di-GMP, the DGC activity of WT SesA was strongly inhibited. These results indicated that the DGC activity of SesA is subject to product feedback inhibition.
The concentration-responsive curve for the DGC activity in the presence of c-di-GMP indicated an IC50 value (concentration for half-maximal inhibition) of 1.07 ± 0.13 µM at 31 °C (Fig. 2B). This affinity of c-di-GMP for the product inhibition site of SesA is comparable to values reported for other DGCs such as DgcA (IC50 = 0.96 µM) and PleD (IC50 = 5.8 µM) 33 . The DGC activity of SesA in the absence of c-di-GMP was 20.5 ± 2.1 µmol pyrophosphate µmol protein −1 min −1 ; i.e., a turnover rate (k cat ) ~10 min −1 . Curve fitting also indicated a Hill coefficient of 3.0 ± 0.3, implying high positive cooperativity of c-di-GMP binding to SesA.
Upon product feedback inhibition of DGCs, c-di-GMP binds to the allosteric product inhibition site (I-site), namely the RXXD motif 33 , which is located near the active-site within the GGDEF domain [34][35][36] . We used the SesA mutant R676A (within 676 RXXD) to assess product inhibition. This mutant showed high pyrophosphate production activity regardless of the presence of 10 µM c-di-GMP, with the activity being sustained during the assay for 600 s ( Fig. 2A). In fact, R676A maintained its high DGC activity even in the presence of 100 µM c-di-GMP   Fig. 2B), indicating that the RXXD motif is necessary for product inhibition of SesA. R676A had higher DGC activity than WT SesA even in the absence of c-di-GMP, suggesting that the R676A mutation alone stabilized the active enzyme conformation. The protein oligomerization of SesA may have been altered by the R676A mutation, as reported for the GGDEF domain of Thermotoga maritima TM1788 37,38 . Taken together, these results demonstrate that SesA is subject to product feedback inhibition by c-di-GMP with relatively high affinity.
Transcriptional regulation of the genes encoding DGC and PDEs. Given the above results, the amount of c-di-GMP produced by SesA (IC50 ~ 1 µM) may be insufficient to activate Tll0007 (K d ~ 64 µM). The gap between these values might imply that another step is necessary to amplify the c-di-GMP signal from SesA toward Tll0007, such as c-di-GMP-dependent transcriptional or enzymatic regulation of DGC and PDEs. In addition to SesA/B/C, seven other genes encode GGDEF/EAL/HD-GYP-domain proteins in Thermosynechococcus (Fig. S1). An amino acid sequence alignment of the GGDEF domains revealed that the residues necessary for DGC activity 39 are conserved in all GGDEF domains except for that in SesB (Fig. S2), which lacks DGC activity 21 . The RXXD motif of the GGDEF domain is conserved in SesA, SesC, Tlr1210, Tlr0627, Tll1049, and Tll1859, suggesting that some of these proteins may be subject to product inhibition. All the EAL domains in Thermosynechococcus appeared to be active based on an amino acid sequence alignment ( Fig. S3) 39,40 . Thus, the in silico analyses did not identify any genes encoding enzymatically inactive GGDEF/EAL/HD-GYP domains.
We assessed the transcriptional regulation of the 10 genes by real-time qPCR analysis. Total RNA was prepared from (1) cells irradiated with white light at 45 °C (standard condition), (2) cells irradiated with blue light at 31 °C (cell aggregation-ON condition), and (3) cells irradiated with teal-green light at 31 °C (cell aggregation-OFF condition). The relative expression of each gene was evaluated compared with the standard condition (Table 1). All the 10 genes were transcribed to a greater degree at 31 °C than at 45 °C, suggesting that a relatively low temperature leads to activation of the whole c-di-GMP signaling network. None of the genes were differentially expressed significantly between the cells under blue light or teal-green light. We previously reported that blue light-and green light-regulated production and degradation of c-di-GMP control the cell-aggregation-ON and -OFF regulation, respectively 21 . These results suggested that the c-di-GMP levels set by SesA/B/C in response to light color did not induce transcriptional change of genes encoding GGDEF/EAL/HD-GYP domain proteins; thus, post-transcriptional control of the GGDEF/EAL/HD-GYP domain proteins is crucial for determining whether cell aggregation is induced or repressed at relatively low temperatures.

Involvement of other DGC and PDEs in the regulation of cell aggregation.
We created gene-disruption mutants of the remaining seven genes (tlr1210, tlr1158, tll1049, tll1859, tll0627, tlr1612, and tlr0485) of T. vulcanus and analyzed the aggregation of cells irradiated with blue or teal-green light irradiation at 31 °C (Fig. 3). The phenotypes of the other three mutants (sesA, sesB, and sesC) have already been reported 21 . Under blue light, WT cells underwent strong cell aggregation whereas ∆tll0007 mutant cells did not aggregate, in agreement with our previous reports 22,24 . None of the seven gene-disruption mutants displayed altered cell aggregation, indicating that no other DGC besides SesA is essential for cell aggregation. Under teal-green light, ∆tlr1612 cells underwent strong aggregation that was comparable with the aggregation observed under blue light, whereas the WT and other mutants did not aggregate. ∆sesB cells also underwent some aggregation under teal-green light, although the aggregation index value was much lower (~20%) 21 . These results suggested that Tlr1612 is the primary protein that represses the aggregation of cells irradiated with teal-green-light.
Tlr1612 is epistatic to SesA for the regulation of cell aggregation. ∆tlr1612 cells underwent strong aggregation under blue or teal-green light (Fig. 3), whereas ∆sesA cells showed little aggregation under either condition 21,22 . We therefore examined the possible epistasis of sesA and tlr1612 by creating the ∆sesA/∆tlr1612 double mutant. ∆sesA/∆tlr1612 cells underwent strong aggregation under both light conditions, similar to ∆tlr1612 (Fig. 4), indicating that tlr1612 is epistatic to SesA for the regulation of cell aggregation.

Discussion
We investigated the c-di-GMP signaling network that regulates light-dependent cell aggregation in T. vulcanus. The cooperating cyanobacteriochrome photoreceptors SesA/B/C perceive blue-to-green light as an input to regulate c-di-GMP signaling 21 . However, it has remained elusive how other c-di-GMP signaling proteins function. The current study has refined our previous work and deepened our understanding of the sophisticated light-dependent c-di-GMP signaling network. Tll0007 is the first cyanobacterial c-di-GMP receptor to be experimentally validated, corroborating the existence of c-di-GMP signaling in cyanobacteria. The affinity of the PilZ domain of Tll0007 for c-di-GMP (K d = 63.9 ± 5.1 µM) (Fig. 1) was relatively low compared with other c-di-GMP receptors such as the Vibrio cholerae PlzD (K d = 0.2 µM) and VpsT (K d = 3.2 µM) and Pseudomonas aeruginosa Alg44 (K d = 12.7 µM) 12,28,41 . Using our FRET biosensor assay, the full-length YcgR of Escherichia coli yielded an affinity of 0.42 ± 0.08 µM (Fig. S4), which is consistent with a previously reported value (K d = 0.84 ± 0.16 µM) 9 , underscoring the reliability of our assay. We cannot exclude the possibility that the affinity of full-length Tll0007 for c-di-GMP is higher in vivo. However, Salmonella Typhimurium BcsA, a homolog of Tll0007, was found to have higher affinity (K d = 8.4 µM) than Tll0007 even with a similar construct and assay 28 . These results strongly suggest that a relatively high cellular concentration of c-di-GMP may be necessary to activate Tll0007.
The Hill coefficient for Tll0007 was 0.90 ± 0.14 ( Fig. 1), which contrasts with other PilZ domains that generally show positive cooperativity for c-di-GMP binding 28,42 . PilZ domains usually bind the c-di-GMP dimer 2,31,32,43 , although binding of the c-di-GMP monomer has also been reported 42 . The lack of cooperativity of c-di-GMP binding to Tll0007 suggests that Tll0007 can bind the c-di-GMP monomer and does not specifically recognize the c-di-GMP dimer. Notably, the binding of the c-di-GMP dimer to the PilZ domain protein Alg44 could induce functional output, whereas binding of the c-di-GMP monomer could not, indicating that a specific form of c-di-GMP is necessary for protein activation 31 . The high cooperativity that has been demonstrated for most PilZ domains appears to reflect the preference for binding the c-di-GMP dimer. Conversely, Tll0007 might be activated by the c-di-GMP monomer.
SesA is the major trigger of cell aggregation for T. vulcanus 21 , and we found that SesA is subject to the product feedback inhibition (Fig. 2). The high cooperativity of product inhibition permits SesA to respond in an all-or-none fashion over a narrow range of c-di-GMP concentration 44 . The DGC activity of SesA (k cat ~10 min −1 ) is relatively high compared with known DGCs such as TM1788(R158A) (k cat ~2.6 min −1 ) 37 and YdeH/DgcZ (k cat ~1.6 min −1 ) 45 , and comparable with Cph2 (k cat ~10.6 min −1 ) 18 . Such high DGC activity is likely important for SesA to rapidly produce sufficient c-di-GMP for the induction of cell aggregation upon blue-light exposure. Notably, Synechocystis cells over-expressing SesA (R676A) were inviable under white-light irradiation, whereas these cells grew normally when irradiated with both green and red light. This finding suggests that the high and unlimited DGC activity of SesA (R676A), when activated by blue light, is toxic to the Synechocystis cells, probably owing to exhaustion of GTP substrate and/or overactivation of c-di-GMP signaling. Thus, it is logical that product feedback inhibition is necessary to control the high DGC activity of SesA. Product feedback inhibition is assumed to be common for GGDEF domain proteins and has been intensively investigated in vitro [33][34][35]46 . However, the physiological importance of DGC product inhibition remains an unresolved issue.
Our study identified Tlr1612 as the crucial protein to repress cell aggregation under teal-green light (Fig. 3B). The expression of tlr1612 did not differ between the blue-light and teal-green-light conditions (Table 1), excluding the possibility that transcriptional regulation underlies the activation of tlr1612. However, tlr1612 was found to be epistatic to sesA (Fig. 4), indicating that Tlr1612 acts downstream of SesA in the c-di-GMP signaling network. Furthermore, our biochemical studies demonstrated that SesA-produced c-di-GMP levels barely exceeded ~5 µM owing to product inhibition (Fig. 2), with this level being far lower than the K d value for binding to Tll0007 (Fig. 1). However, no other single DGC was responsible for cell aggregation under blue light, whereas tlr1612 was essential for suppression of cell aggregation under teal-green light (Fig. 3, and see 21 for ∆sesB and ∆sesC). Although the predicted Tlr1612 protein harbors both GGDEF and EAL domains, it probably serves as the major PDE that suppresses cell aggregation under teal-green light and likely under blue light as well, according to our gene expression analysis. Nevertheless, SesA was able to trigger cell aggregation under blue light, indicative of a certain mechanism(s) that leads to an upshift of c-di-GMP level sufficient for the low-affinity site of Tll0007. At present, it is not known whether such an upshift could be achieved by SesA-dependent downregulation of Tlr1612 PDE, SesA-dependent upregulation of Tlr1612 DGC, or SesA-dependent upregulation of two or more yet unidentified DGCs including SesC (as upregulation of any single DGC except SesA was excluded by our gene disruption experiments). These scenarios are summarized in Fig. 5.
Notably, disruption of all the three photoreceptors (SesA/B/C) still allowed some cell aggregation (the aggregation index was 40-70% for all the tested wavelengths 21 ). This observation implies that certain levels of c-di-GMP are maintained during normal phototrophic growth owing to production by one or more DGC(s), even though the possibly strong PDE activity of Tlr1612 is also functioning. There are many indications that multiple DGCs and PDEs are often spatially differentiated in prokaryotic cells, in addition to the well-known temporal regulations 4,47 . The possible PDE activity of Tlr1612 may also serve as an insulator for other, potentially unrelated, effects of c-di-GMP signaling, which are not shown in Fig. 5. To reconcile the ambiguity of our model, the biochemical properties of Tlr1612 and of other DGC proteins must at least be ascertained. A somewhat similar two-step c-di-GMP signaling system has been reported in Salmonella, wherein the "first" c-di-GMP pool activates the transcription of the gene encoding the DGC protein AdrA, which produces higher levels of the "second" c-di-GMP pool that directly activates the cellulose synthase activity 48,49 . Such transcriptional regulation is unlikely in Thermosynechococcus as no gene exhibited a difference in transcript levels between the cell aggregation-ON and -OFF conditions (Table 1). Thus, post-transcriptional control will surely be of importance in the c-di-GMP signaling network of Thermosynechococcus, as also proposed for E. coli 50 .
Besides SesA/B/C and Tlr1612, six other genes encode GGDEF/EAL/HD-GYP domain proteins in Thermosynechococcus, although their functions remain completely unknown. Disruption of each of these six genes did not affect cell aggregation at 31 °C (Fig. 3), suggesting that these genes may have redundant roles in cell aggregation or be involved in other cellular responses-potentially regulated by c-di-GMP. All of the c-di-GMP signaling genes, especially tlr1210, were expressed to a greater degree at 31 °C compared with 45 °C (Table 1), suggesting that the overall c-di-GMP signaling governs various cellular responses especially at relatively low temperatures. A candidate for a c-di-GMP-regulated cellular responses is phototactic motility; indeed, c-di-GMP regulates phototaxis in the cyanobacterium Synechocystis 18,19 . The PilB ATPase, which drives the extension of type IV pili 51,52 , is a likely universal c-di-GMP-binding protein 10 . Future work will address how each GGDEF/EAL/ HD-GYP protein employs c-di-GMP to orchestrate cellular responses in cyanobacteria. The protein thermostability and the relatively small number of c-di-GMP genes of Thermosynechococcus suggest the potential for further detailed characterization of every protein in the sophisticated c-di-GMP signaling network. Table S1. Plasmids were constructed using the In-Fusion System (TaKaRa). For constructing FRET biosensors, fluorescent proteins were cloned from pCyPet-His and pYPet-His (Addgene). We created monomerized variants of CyPet (mCyPet) and YPet (mYPet) by introducing A206K 29,53 . The DNA encoding the chimeric protein consisting of mCyPet, Tll0007-PilZ (or the full-length YcgR of E. coli), and mYPet was cloned into pET28V, in which the protease recognition site in the original pET28a plasmid (Novagen), for removal of the N-terminal His tag, was replaced with one for tobacco etch virus (TEV) protease. Site-directed mutagenesis was performed using PrimeSTAR Max Basal Mutagenesis kit reagents (TaKaRa). The vector pTCH2031V was used to overexpress SesA in Synechocystis sp. PCC 6803 21 . The protease recognition site for removal of the N-terminal His tag was replaced with one for TEV protease, which could be used for future purification. The E. coli rrnB terminator sequence was inserted after the C-terminus of the protein-coding region incorporated into pTCH2031V.

Construction of plasmids and mutants. Primers used are listed in
For disruption of tlr1210, tlr1158, tll1049, tll1859, tll0627, and tlr1612 in T. vulcanus, a spectinomycin/ streptomycin-resistance cassette was inserted into the start codon of each gene. For disruption of tlr0485, the majority of the open reading frame was replaced by a chloramphenicol resistance-cassette.
Fluorescence spectroscopy. Fluorescence spectra were recorded at 31 °C using a RF-5300PC spectrofluorometer (Shimadzu, excitation at 425 nm, 5 nm excitation and 3 nm emission slit widths, emission scan from 450 to 600 nm) using 50 nM of an individual protein in 20 mM Hepes·NaOH (pH 7.5), 100 mM NaCl, and 10% (wt/ vol) glycerol. Binding affinity was determined by adding increasing concentrations of c-di-GMP (10 nM to 1 mM) or GTP (1 µM to 10 mM) and measuring the resulting YFP/CFP emission ratio (emission at 527 nm/emission at 475 nm). Curve fitting was performed using the program KaleidaGraph (Synergy Software).
Kinetic measurements of DGC activity. The amount of inorganic pyrophosphate, a by-product of c-di-GMP synthesis, was monitored at 360 nm with a UV-2600PC spectrophotometer using EnzChek pyrophosphate assay kit reagents (Invitrogen) 22,54 . The kit includes inorganic pyrophosphatase, which catalyzes conversion of inorganic pyrophosphate into 2 equivalents of inorganic phosphate. In the presence of inorganic phosphate, the substrate 2-amino-6-mercapto-7-methylpurine ribonucleoside was enzymatically converted to ribose 1-phosphate and 2-amino-6-mercapto-7-methylpurine, which absorbs 360-nm light. Each reaction contained 50 mM Tris-HCl, pH 7.5, 20 mM MgCl 2 , and 0.2 µM SesA in the presence of a varying amount of c-di-GMP. The reaction was initiated by the addition of 100 µM GTP, incubated at 31 °C, and monitored for 600 s.
Cyanobacterial strains and culture conditions. The T. vulcanus strain RKN (equivalent to National Institute for Environmental Studies 2134) that shows positive phototaxis was cultured at 45 °C in BG11 medium 55 . Culture density was monitored at 730 nm. Transformations of T. vulcanus were performed according to 56 . The antibiotic concentration in BG11 medium for selection of transformants was 5 µg·mL −1 chloramphenicol, 80 µg·mL −1 kanamycin, or 10 µg·mL −1 spectinomycin plus 5 µg·mL −1 streptomycin. PCR was used to confirm the complete segregation of the mutant allele (i.e., the WT loci were replaced with the mutant loci in all of the multiple copies of the cyanobacterial chromosome).  (Table S2), and 10 μL THUNDERBIRD SYBR qPCR Mix (Toyobo). Samples were denatured initially by heating at 95 °C for 1 min, followed by a 40-cycle amplification and quantification program (95 °C for 15 s and 60 °C for 30 s). Melting-curve analysis was conducted to ensure amplification of a single product. The amplification efficiency of each primer pair was determined by running three-fold serial dilutions (seven dilution series). A standard curve was generated by plotting the cycle threshold (C t ) value determined using the second derivative maximum method against the log of the dilution factor. Gene expression was normalized to that of rnpB 57 .

Real
Cell aggregation assay. Cultures of T. vulcanus WT and its disrupted mutants grown at 45 °C (OD 730 0.5-2) were diluted to yield a culture OD 730 of 0.2. These samples were then incubated at 31 °C for 48 h under photosynthetic red light along with blue or teal-green light. The results of the cellulose-dependent cell-aggregation assay, described as follows, were reported as an aggregation index (%) 24 . Briefly, after the irradiation period, cell suspensions were thoroughly mixed and aliquots transferred to cuvettes. The samples were held at room temperature in the dark for 30 min, during which time most of the aggregated cells precipitated. Then, the OD 730 of each sample was measured (denoted OD NA ; i.e., OD 730 of non-aggregated cells remaining in the culture medium). Next, cellulase (12.5 U·mL −1 ; Worthington Biochemical) was added to each cuvette sample, which was then incubated for 30 min at 37 °C to completely disperse the aggregated cells. The OD 730 of each sample (denoted OD total ) was then measured. The aggregation index (%) was defined as ((OD total − OD NA )/ OD total ) × 100.