Blue Light Regulated Two-Component Systems: Enzymatic and Functional Analyses of Light-Oxygen-Voltage(LOV)-Histidine Kinases and Downstream Response Regulators*

Abstract

Light is an essential environmental cue for diverse organisms. Many prokaryotic blue light photoreceptors use Light, Oxygen, Voltage (LOV) sensory domains to control the activities of diverse output domains, including histidine kinases (HK). Upon activation, these proteins autophosphorylate a histidine residue before subsequently transferring the phosphate to an aspartate residue in the receiver domain of a cognate response regulator (RR). Such phosphorylation, activates the output domain of the RR, leading to changes in gene expression, protein-protein interactions or enzymatic activities. Here we focus on one such light sensing LOV-HK from the marine bacterium Erythrobacter litoralis HTCC2594 (EL368), seeking to understand how kinase activity and subsequent downstream effects are regulated by light. We found that photoactivation of EL368 led to a significant enhancement in the incorporation of phosphate within the HK domain. Further enzymatic studies showed that the LOV domain affected both the LOV-HK turnover rate (k_cat) and K_m in a light-dependent manner. Using in vitro phosphotransfer profiling, we identified two target RRs for EL368 and two additional LOV-HKs (EL346 and EL362) encoded within the host genome. The two RRs include a PhyR-type transcriptional regulator (EL_PhyR) and a receiver-only protein (EL_LovR), reminiscent of stress-triggered systems in other bacteria. Taken together, our data provide a biochemical foundation for this light-regulated signaling module of sensors, effectors and regulators that control bacterial responses to environmental conditions.

Organisms have developed stimulus-response coupled mechanisms to sense and to adapt to the myriad of physical and chemical changes in their surroundings. A common example of such signaling in prokaryotes is provided by phosphorelay systems based on two types of proteins: a sensory histidine kinase (HK) and a downstream response regulator (RR); activation of the former leads to RR phosphorylation that induces cellular responses via changes in downstream gene expression or interactions within a signaling pathway^{1, 2}. Light is one of the prevalent stimuli sensed by bacteria, because of its dual and opposing roles as vital source of energy and as a potentially damaging factor to intracellular components. As such, a diverse set of photosensory proteins has evolved to allow these organisms to detect and cope with light throughout the UV and visible spectrum.

A prevalent type of blue light photosensor is provided by the Light-Oxygen-Voltage (LOV) domains^{3, 4}, a subset of the broader family of Per-ARNT-Sim (PAS) group of environmental sensors⁵. These proteins non-covalently bind FMN or FAD in the dark; upon blue light exposure, a covalent adduct is formed between the flavin isoalloxazine C4a position and a nearby cysteine residue. This bond is stable under illumination, but spontaneously breaks in the dark with a timescale of seconds to hours^{6, 7}. Notably, formation of this bond triggers conformational changes at the central β-sheet of the protein that are propagated to a wide array of downstream domains^8–10, including HKs, HTH DNA-binding domains, phosphodiesterases, sigma factor regulators (STAS) domains^{3, 11}. In LOV-containing HKs (LOV-HKs), light exposure typically leads to an enhancement in the overall kinase activity of the protein^12–14, implicating an important role of these proteins in sensing and responding to light stimulus. While this has been validated for several LOV-HKs^{3, 11, 12, 14–19}, the mechanisms that allow LOV domains to control HK activity remain an active topic of investigation. Such information would benefit both basic knowledge of signal transduction in an essential class of sensory proteins while also facilitating the development of optogenetic tools to control complex signaling pathways.

To this end, we used a series of biochemical tools to characterize the enzymatic mechanisms and signaling pathways of the three predicted LOV-HK proteins from the marine bacterium Erythrobacter litoralis HTCC2594¹⁴: EL368, EL346 and EL362. Both EL368 and EL346 undergo light-dependent activation of autophosphorylation kinase activity; in contrast, EL362 contains a point mutation within the critical LOV domain, rendering it insensitive to light. Mechanistically, steady state enzymatic analysis of EL368 showed that the observed photoactivation stems from changes in both k_cat and K_m parameters upon illumination. In vitro phosphotransfer profiling of EL368 and the two other E. litoralis LOV-HKs (EL346 and EL362) reveal two downstream targets: a PhyR-type anti-, anti-sigma RR and a single domain RR. Examination of their biochemical properties and comparison with the genetically-identified Caulobacter crescentus LovK/LovR stress signaling system¹⁶ suggests that these substrates are homologs of the C. crescentus PhyR and LovR proteins, leading us to designate them as EL_PhyR and EL_LovR, respectively. Intriguingly, all three LOV-HKs phosphorylate EL_LovR while EL_PhyR is targeted more specifically by only EL368 and EL346. This type of signaling network is reminiscent of general stress responses in several alphaproteobacteria^{20, 21}, particularly via parallels with the Caulobacter crescentus pathway where PhyR regulates an operon containing an EL368-homolog and its cognate RR¹⁶. Altogether, our findings suggest that LOV-HKs may participate in conserved branched transduction pathways in bacteria essential to biological responses to stress.

EXPERIMENTAL PROCEDURES

Cloning, expression and purification of LOV-HK and response regulator proteins

DNA-encoding sequences of the LOV-HK proteins EL368, EL346 and EL362 (NCBI Gene locus tags ELI_02980, ELI_04860, and ELI_07650, respectively) and anti-σ factor (locus tag ELI_10225) were amplified from Erythrobacter litoralis HTCC2594 genomic DNA²², and cloned into the pHis-Gβ1-parallel expression vector²³. Response regulators (locus tags in Table S1) were amplified from Erythrobacter litoralis HTCC2594 genomic DNA and cloned into pHis-parallel expression vector²⁴. Mutants EL368 (C93A) and EL362 (R150G) were generated with QuickChange II XL site-directed mutagenesis kit (Stratagene). Escherichia coli BL21(DE3) cells transformed with those vectors were grown at 37 °C in LB media containing 100 μg/mL of ampicillin to an A₆₀₀ of 0.8, then induced overnight at 18 °C by the addition of 0.4 mM isopropyl β-D-thiogalactoside. Cells were harvested, resuspended into 50 mM Tris pH 8.0, 100 mM NaCl buffer, lysed by extrusion and centrifuged at 48,000 g for 45 min. The supernatant fraction was initially loaded onto a Ni²⁺ affinity column (Chelating Sepharose^™ Fast Flow, GE Healthcare) pre-equilibrated with the above buffer plus 25 mM imidazole. Proteins were eluted using a 25 to 500 mM imidazole gradient, and then digested with His₆-TEV protease²⁵. Cleaved target proteins were separated from their tags and TEV protease using a Ni²⁺ affinity column; flowthroughs were collected and concentrated (Amicon Ultra, Millipore). A final gel filtration step used HiLoad 16/60 Superdex 75 or HiLoad 26/60 Superdex 200 columns (GE Healthcare), with both columns pre-equilibrated in 50 mM Tris pH 8.0, 100 mM NaCl buffer. All protein-containing fractions were analyzed by SDS-PAGE and stored at −20 °C with 50% glycerol (v/v). For the three LOV-HK proteins, all purification steps were performed under dim red light.

Visible absorption spectroscopy

UV-vis absorbance spectra of EL368, EL362 and EL362 (R150G) were collected on a Varian Cary 500 spectrophotometer using a quartz cuvette with 1 cm path length. Dark state samples were kept in the dark until data collection; light state samples were generated via flash illumination. Dark state recovery time constants were obtained by monitoring the increase in absorbance at 446 nm in 20 min intervals for 10 hr at room temperature. Data points were fitted to a two-phase exponential decay function using the software Prism (GraphPad Software Inc.). All proteins were in 50 mM Tris pH 8.0, 100 mM NaCl, 5 mM DTT solutions.

Autophosphorylation Assays

EL368 autophosphorylation reactions were performed at room temperature in 50 mM Tris pH 8.0, 100 mM NaCl, 5 mM MgCl₂ and 5 mM DTT. Reactions were initiated by adding mixtures of radiolabeled [γ–³²P]ATP (10 – 50 μCi, 6000 Ci/mmol, Perkin Elmer) and cold ATP (0.05 – 2 mM) in buffer with EL368 (5 μM final concentration) in dark (dim red light) or light (protein sample previously illuminated with photographic flash) conditions. At the appropriate time points, aliquots were removed and quenched into 4× SDS-PAGE sample buffer (50 mM Tris pH 6.8, 200 mM NaCl, 40 mM EDTA, 0.2% bromophenol blue, 10% (v/v) β-mercaptoethanol, 4% (w/w) SDS and 20% (v/v) glycerol. Samples were immediately immersed in liquid N₂ and stored at −80 °C, before SDS-PAGE analysis. After electrophoresis, the dye front and unincorporated ATP was removed with a razor blade. Subsequently, gels were dried under vacuum at 80 °C for 45 min and exposed for 30 to 60 min to a phosphoimager screen that was subsequently scanned with a Fujifilm FLA-5100 phosphoimager. Band intensity measurements were performed with MultiGauge software (Fujifilm). Initial rates of ³²P incorporation were calculated by fitting the first time points to a linear equation; the [ATP] dependence of these rates were fit to a Michaelis-Menten equation to extract K_m and k_cat parameters using Prism. EL362, EL362 (R150G) dark/lit states (5 μM) were autophosphorylated as mentioned above, but with 500 μM unlabeled ATP and 50 μCi of [γ–³²P]ATP (6000 Ci/mmol, Perkin Elmer).

Phosphotransfer profiling

Initially, EL368 or EL346 LOV-HK proteins (10 μM) were autophosphorylated in 50 mM Tris pH 8.0, 100 mM NaCl, 5 mM MgCl₂, 5 mM DTT, 4.5 μCi [γ–³²P]ATP (6000 Ci/mmol) and 1 mM ATP for 10 min at room temperature under white light conditions. Response regulators were diluted to 10 μM in the same buffer without nucleotides. Equimolar mixtures of phosphorylated EL368 or EL346 and response regulators (5 μM each) were made and incubated at room temperature for 30 s and 10 min intervals. Samples were treated and imaged for ³²P incorporation as described above.

Phosphotransfer kinetic measurements

To obtain the phosphotransfer rate of EL368 to the cognate response regulator EL_PhyR (gene locus ELI_10215), 5 μM EL368 in nucleotide-free buffer was mixed with 100 μM EL_PhyR containing 25 μCi [γ–³²P]ATP (6000 Ci/mmol) and 1.5 mM ATP. Aliquots were removed at appropriate time points, and treated as described above with the exception that the band corresponding to the phosphorylated response regulator was measured. Initial rates for the dark state were obtained with reactions performed under dim red light; lit state EL368 samples were generated by flash illumination prior to the beginning of the reaction and kept under regular white light room illumination. Phosphotransfer of EL362 (5 μM) was done as previously stated for EL368, but with 500 μM of ATP, and 5 μM of the cognate response regulator EL_LovR (gene locus ELI_07655).

Phosphatase assay

EL_PhyR phosphatase assays were performed at room temperature in 50 mM Tris, 100 mM NaCl, 5 mM DTT, 5 mM MgCl₂ at pH 8.0. 10 μM of EL368 was pre-incubated for 10 min with an equimolar concentration of EL_PhyR and 30 μCi of [γ-³²P]ATP. After pre-incubation 5 mM AMPPNP was added to the reaction to stop further phosphorylation from ATP. For reactions with the EL_LovR (wild type or D52A) proteins, 10 μM of these proteins were added simultaneously with 5 mM AMPPNP. Aliquots were removed at 0.25, 0.5, 1, 2, 5, 10 and 15 min. Samples were treated as described above.

Size exclusion chromatography and multiple angle laser light scattering

The oligomerization states of EL368 and EL362 were determined with integrated size exclusion chromatography and light scattering (SEC-MALLS). 500 μL of 20 μM EL368, EL368(56-368), EL362 and EL362(R150G) dark/lit samples were injected at 0.5 mL/min flow onto Superdex 75 10/300 or Superdex 200 10/300 analytical columns (GE Biosciences), pre-equilibrated with 50 mM Tris pH 8.0, 100 mM NaCl, 5 mM MgCl₂ and 5 mM DTT buffer. Using a 0.5 mL/min flow rate, these samples were detected post-elution using inline miniDAWN TREOS light scattering and Optilab rEX refractive index detectors (Wyatt Technology). All the procedures were performed at 5 °C. Data analyses and molecular weight calculations were carried out using the ASTRA V software (Wyatt Technology). The same approach was also used to study the complex of the EL_PhyR response regulator and its anti-σ factor. 500 μL samples containing EL_PhyR (30 μM), anti-σ factor (40 μM) individually or mixed were applied onto Superdex 75 10/300 in line with MALLS pre-equilibrated with the above-mentioned buffer. To study the interaction with the phosphorylated state of response regulator, EL_PhyR by itself or mixed with anti-σ factor was incubated with 0.5 μM EL368 and 2 mM ATP for 30 min prior to injection.

RESULTS

Characterization of light-dependent enzymatic activity

In addition to a conventional HWE-type HK domain²⁶, EL368 contains an N-terminal LOV domain, conferring kinase activation upon blue light exposure14. Measurements of the autophosphorylation activity of EL368 using [γ-³²P]ATP as substrate in the dark and light show that the protein is approximately four times more active in the lit state (v: 0.007±0.001 s⁻¹) than in the dark (v: 0.002±0.001 s⁻¹) (Figure 1a, b). Changes in kinase activity upon light illumination have also been reported for other natural LOV-HKs^12–14 along with an engineered variant of FixL²⁷, clearly indicating that light-induced conformational changes in the LOV domain can allosterically affect the overall activity of the HK domain. However, the undergoing mechanisms of such regulation remain unclear. In a very simplistic enzymatic model, the sensor domain could influence overall enzyme activity through changes in the affinity for various substrates (K_m), V_max or a combination of both. To address which of these enzymatic parameters are affected by illumination, we determined the net kinase activity of EL368 at varying concentrations of ATP. Figure 1b shows that the autophosphorylation reaction catalyzed by the enzyme supports a classical Michaelis-Menten kinetics mechanism in both conformational states. Upon activation the enzyme presents a four times overall increase in its catalytic rate (V_max(dark): 0.002±0.0001 s⁻¹, V_max(lit): 0.009±0.0005 s⁻¹). On the other hand, there is a slight, but measurable decrease of affinity for ATP in the lit state (385±52 μM) in comparison to the dark state (131±31 μM). Similar ATP affinities have been reported for other HKs^28–31.

Figure 1. Light enhances the phosphorylation activity of EL368.

a) Time course of phosphate incorporation into EL368 either in dark or white light conditions. Upon light exposure, the enzyme has a higher kinase activity and achieves steady state equilibrium more quickly than the dark state. b) Initial velocities obtained at different ATP concentrations were fit to a Michaelis-Menten equation (black line). c) EL368 (C93A) demonstrates that kinase activity observed in the dark state does not arise from residual cysteinyl-flavin adduct formation.

One practical issue for measurements of dark state kinase activity is the very slow photorecovery to the dark state observed for this enzyme (~2 hr, Figure S1), which would lead to the accumulation of lit state conformers upon incidental exposure to light. While we took reasonable steps to avoid such exposure, such artifacts could lead to aberrantly high background levels of activity in such phosphorylation assays. To rule this out, we engineered a mutant version (C93A) that binds to flavin but cannot form the cysteinyl adduct. As expected, the mutant displayed a dark state activity (ν: 0.002±0.0005 s⁻¹) comparable to wild-type (ν: 0.002±0.001 s⁻¹), and only minimally affected by light (ν: 0.002±0.001 s⁻¹) at 1 mM ATP (Figure 1c). Low levels of dark state phosphorylation have also been reported in other LOV-HK systems^12–14.

LOV domain stabilizes dimeric conformation

Most HKs studied to date function as dimers, with cis or trans-autophosphorylation between subunits of the oligomer^{28, 32–34}. To examine whether EL368 also operates as a dimer, we took advantage of the long-lived lit state (~2 hr) used SEC-MALLS to examine the solution oligomerization state of the protein in the dark and light. Results demonstrated that EL368 is primarily a very stable and light-independent dimer in solution (Figure 2). In addition, the minor changes in the elution profile between light and dark conditions suggests that there are no drastic changes in the shape of the enzyme upon activation, in agreement with data reported for the C. crescentus homolog LovK³⁵.

Figure 2. The dimeric solution structure of EL368 is stabilized by elements located N-terminal of the LOV domain.

SEC-MALLS traces correspond to full length EL368 (black: dark state, red: light) and EL368(56-368) (blue: dark, magenta: light); dRI traces from the Optilab rEX are shown as solid lines, plotted on the same relative scale, while MALLS-derived mass distributions are plotted as individual points. Full length EL368 is dimeric regardless of illumination, as SEC elution profiles and mass distributions are unaffected by light (apparent MW ~90 kDa at the center of the peak, compared to monomer MW: 41.4 kDa). Removal of 55 residues prior to the predicted LOV domain generates EL368(56-368) and destabilizes the dimer (apparent MW ~50 kDa at the center of the peak, compared to monomer MW: 35.2 kDa).

Secondary structure predictions of EL368 by Jpred³⁶ indicates that the N-terminal LOV domain is preceded by an additional 56 residue-long region (Figure S2). Previous studies of other LOV domains have demonstrated that such additional regions can affect photocycle lifetime and oligomerization state^{35, 37}, leading us to examine their influence in EL368. To do so, we constructed a truncated form of the protein, EL368(56-368). While removal of that segment did not affect the binding to flavin or solubility of the molecule, SEC-MALLS showed that this deletion affected the protein oligomerization state by biasing it towards lower molecular weight (Figure 2). The molecular weight distributions obtained for dark and lit samples suggest that protein eluted as a mixture of monomers and dimers, with minimal light-dependent effects.

Light regulated two-component signaling: Identification of target response regulators and influence of light stimulus on phosphotransfer

Two component systems are often composed of HKs and RRs found in operons, where they usually form an exclusive phosphotransfer pair. Nevertheless, HKs can be found as “orphans” without candidate RRs nearby in the genome, complicating identification of cognate pairs from sequence alone^{38, 39}. Analysis of the E. litoralis genome²² shows that EL368 is encoded as an orphan HK, with the nearest predicted RR over 80 kb away. To further characterize the influence of light stimulation at the phosphotransfer step and suggest a functional role for EL368, we proceeded to identify RR substrates using phosphotransfer profiling. This technique, based on comparisons of the ability of HKs to phosphorylate candidate RRs, takes advantage of the in vitro kinetic preference HKs exhibit for their in vivo cognate RR⁴⁰.

The E. litoralis genome is predicted to encode 23 RRs⁴¹ (Table S1), 21 of which we successfully expressed in E. coli. 12 of these proteins expressed as full-length constructs; the other 9 were expressed as isolated receiver domains. This approach, which considerably increased the total yield of soluble protein compared to the full-length forms, takes in consideration the fact that receiver domain defines the complex specificity and phosphotransfer catalysis⁴⁰. We were unable to obtain sufficient quantities of two proteins for further analysis, as they either went into inclusion bodies (ELI_11255) or failed to express (ELI_11920). With these reagents in hand, in vitro profiling of EL368 in a very short timescale (30 s) shows that the enzyme efficiently phosphorylated EL_LovR and EL_PhyR (Figure 3a). Extension of the reaction to 10 min allowed the protein to phosphorylate 4 additional response regulators at lower levels, although these may reflect non-specific crosstalk given the prolonged incubation time⁴⁰ (Figure 3b).

Figure 3. Phosphotransfer profiling of EL368.

HK was profiled against twenty-one RR of E. litoralis with phosphotransfer reaction times of thirty seconds (a) and ten minutes (b). In the longer time reaction several RR are phosphorylated by the enzyme; however in a much shorter time point only two kinetically preferred protein substrates (EL_PhyR and EL_LovR) are phosphorylated. Asterisks (*) indicate RRs fragments containing only the receiver domain as described in Table S1. For optimum resolution, this SDS-PAGE gel was run longer than comparable gels shown in Fig. 4 and 6; EL368 appears as a doublet here as a result. c) Schematic summary of phosphorylation patterns among the EL368 and EL346 LOV-HK proteins and their RR substrates.

In addition to EL368, E. litoralis also contains two other LOV-HKs: EL346¹⁴ and EL362 (vide infra). The former one is also predicted to be an orphan HK, therefore we performed the same biochemical-mapping scheme. Interestingly, EL346 demonstrated a kinetic preference for the same two RRs initially recognized by EL368 (Figure S3a; summarized in Figure 3c). Nevertheless, in a longer reaction time point, it phosphorylated only one additional substrate (ELI_09195) (Figure S3b). EL362 exhibited a different specificity and several other unique properties, discussed below.

To address whether a light stimulus would affect phosphotransfer as well as autophosphorylation, we conducted kinase assays in the dark and light with mixtures of EL368 and its substrates (1:20 ratio). These experiments utilized EL_PhyR as a substrate, as EL_LovR exhibited substantial phosphatase activity (vide infra). Following the same trends observed for the autophosphorylation, the amount of phosphorylated RR in the light was higher than in the dark state (ν (dark): 0.001±7×10⁻⁵ mol RR (mol kinase) ⁻¹ s^−1, ν (lit): 0.0080±3×10⁻⁴ mol RR (mol kinase) ⁻¹ s⁻¹), suggesting that the control exerted by photoactivation primarily affects the levels of phosphorylated HKs that in turn lead to an increase in the population of phosphorylated RRs (Figure 4).

Figure 4. Influence of light on the EL368 phosphorylation of the EL_PhyR substrate.

a) Time course of EL_PhyR by EL368 under dark and lit conditions. b) Quantitation of phosphorylation levels observed in panel a. As observed in the autophosphorylation reaction, light stimulates the phosphorylation activity of the enzyme. The linear phase of the curves (up to 55 s) was used to calculate the initial rates of phosphotransfer.

EL368 controls a transduction cascade involving an σ/anti-σ interaction system

To initially suggest potential functions of the two LOV-HK substrates, we used a bioinformatics approach with Hidden Markov Model domain identification implemented by Pfam⁴². This revealed that EL_PhyR combines a C-terminal REC domain with a N-terminal σ factor like domain, analogous to the PhyR response regulator. PhyR itself is found in an operon with genes for σ and anti-σ factors and a very conserved upstream promoter consensus sequence with -35 (GGAACC) and -10 (CGTT) elements^{20, 21, 43–45}. Under stress conditions, PhyR is activated and binds to the anti-σ factor, increasing the number of free σ subunits that may bind to DNA leading to the transcription of target genes²⁰. Notably, the chromosomal region of EL_PhyR shows that it has a similar arrangement as the PhyR regulon (Figure 5a), reinforcing the analogy between these proteins.

Figure 5. EL_PhyR binds to an anti-σ factor upon activation.

a) Top: Representation of the chromosomal region of EL_PhyR gene. Arrows indicate open reading frames for response regulator EL_PhyR (phyR-like), anti-σ and σ factors. The coordinates to the initiation codon of each gene are indicates by curved arrows. Bottom: DNA sequence of 5′ region of EL_PhyR gene, the putative promoter sequence (red) and the initial residues of the protein are displayed. b) SEC-MALLS results of the isolated versions of the inactive and active forms of EL_PhyR and the anti-σ factor. c) SEC-MALLS data of combinations of inactive EL_PhyR and anti-σ compared with a mix of the activated EL_PhyR/anti-σ. Changes in the elution volumes and MW distributions of mixtures of EL_PhyR and anti-σ factor show binding of the two proteins when response regulator is phosphorylated. Phosphorylated state of response regulator was generated by adding it to EL368 and ATP and incubating by 30 min at room temperature before injecting in a Superdex 75 10/300GL column. dRI and MW data are shown as in Figure 2.

To test the hypothesis that EL_PhyR may serve as a PhyR homolog, we examined a critical function: binding of phospho-EL_PhyR to the predicted anti-σ (ELI_10225) by using SEC-MALLS experiments with mixtures of both proteins. To generate active RR, we added very small amounts of EL368 (0.5 μM) and ATP (1 mM) prior injection onto the system. Phosphorylation did not trigger any drastic change in oligomerization state or shape of ELI_10225, which remains monomeric (Figure 5b); the anti-σ was also monomeric in solution. When both proteins were incubated together, we observed an interaction between the two only when the RR was activated, as demonstrated by a shift in elution volume to 10.6 mL and increase in molecular weight corresponding to formation of a 1:1 heterodimer of EL_PhyR (MW ~28 kDa) and ELI_10225 (MW ~6.4 kDa). We note that a minor population of the inactive RR is also capable of forming some complex with the anti-σ (grey line, 9.8 mL), but differences in both the efficiency of forming this complex and its SEC elution volume suggest that this minor complex substantially differs from the dimer formed by activated protein.

The second substrate, EL_LovR, is annotated by Pfam analysis⁴² as a single domain response regulator containing only a REC domain (SDRR). These molecules serve several roles in signaling pathways, from forming protein complexes with downstream elements to serving as phosphate sinks that modulate pathway activity^{41, 46}. The combination of EL_LovR and the PhyR homolog EL_PhyR is reminiscent of an analogous stress response system identified in C. crescentus¹⁶. In this system, the SDRR is proposed to act as a phosphate dump based solely on genetic approaches; to provide a biochemical validation for this model and demonstrate its function, we assayed the effects of EL_LovR on the stability of the EL_PhyR~P complex. To do so, we preincubated EL368, [γ–³²P]ATP and the EL_PhyR substrate under illumination, efficiently generating the EL_PhyR~P complex, before stopping further kinase activity by adding an excess of non-hydrolyzable AMPPNP. Aliquots taken from this reaction afterwards shows that the EL_PhyR~P complex is normally stable over the 15 min duration of the assay (Figure 6a). However, when the SDRR is added simultaneously with AMPPNP, we instead see a rapid decrease in EL_PhyR~P accompanied by a transient accumulation of EL_LovR~P (Figure 6b), followed by its rapid dephosphorylation. These data are consistent with EL_LovR serving as a phosphate sink that can rapidly deactivate the EL_PhyR by promoting phosphate turnover. This activity requires the phosphoaccepting D52 residue within the SDRR; mutation to alanine greatly slows these effects while dramatically lowers phosphorylation levels (with low residual phosphorylation likely occurring on nearby residues⁴⁷) (Figure 6c). Taken together our data suggests that the HK-PhyR-SDRR signaling module reported by genetics in C. crescentus can be reconstituted in vitro with these E. litoralis proteins.

Figure 6. Presence of EL_LovR induces phosphate loss of EL_PhyR.

a) Phosphatase assay of EL_PhyR after phosphotransfer by EL368, demonstrating that phosphorylated EL_PhyR is stable up to 15 min. b) EL_PhyR rapidly loses phosphate when incubated with unphosphorylated EL_LovR. c) The phosphoaccepting Asp52 residue is involved in the phosphate turnover of EL_PhyR, as shown by the reduced efficiency of EL_LovR D52A in this process.

EL_LovR forms a two-component system with a “blind” LOV histidine kinase: Non-flavin binding protein has impaired kinase activity

Analysis of chromosomal region near EL_LovR revealed another LOV-HK, (ELI_07650, named here as EL362) located 18 bp upstream of the RR, suggesting that they might both be located in a shared operon. Phosphotransfer assays confirmed that the enzyme could indeed use EL_LovR as a substrate (Figure S4); comparable assays with EL_PhyR and an unrelated response regulator showed no phosphorylation (Figure S4). Interestingly, EL362 also appears to contain a LOV domain based on the presence of the highly conserved GRNCRFLQ motif characteristic of the LOV family¹¹ (Figure S5). Nonetheless, we were unable to demonstrate flavin binding to purified recombinant EL362 protein. This was explained by the presence of an unusual Arg residue at the position of a normally conserved glycine within the Iβ strand of the putative LOV domain (Figure S6). We have confirmed that the mutation which generates the Arg 150 residue was found within DNA isolated from strains regrown in our laboratory, as well as being present in the original genomic sequencing data. A homology model of this domain clearly shows that the large Arg sidechain would protrude directly into the flavin binding site, blocking chromophore binding (Figure 7a). To confirm this hypothesis, we generated a EL362(R150G) mutant which reverts this position to the most abundant residue among LOV domains. This change restores both flavin binding and LOV photochemistry to EL362 (Figure 7b), as exhibited by illumination triggering the expected loss of absorbance near 450 nm and subsequent dark state recovery over approximately 3 hr. Reestablishment of the flavin binding restored photosensitivity to the kinase activity: dark and lit state samples of EL362(R150G) showed autophosphorylation rates of 0.008±7×10⁻⁴ s⁻¹ vs. 0.010±7×10⁻⁴ s⁻¹ (Figure 7c). Notably, both these levels are higher than the wild type enzyme (0.001±1×10⁻³ s⁻¹_, unaffected by illumination).

Figure 7. Biochemical characterization of EL362: A blind LOV-HK protein.

a) Predicted effects of R150 residues within EL362 LOV domain. Bottom center: Structure of a typical LOV domain (Arabidopsis thaliana phototropin 2 LOV2 = Atphot2 LOV2, PDB: 4EEP⁶²) showing the location of the internally-bound flavin chromophore. Top left: Expansion of the Atphot2 FMN binding site. Top right: Homology model of predicted site of FMN binding from EL362, superimposing the predicted location of Arg 150 with the FMN bound as seen in other LOV domains. b) UV-visible absorption spectra of EL362 and EL362 (R150G). The EL362 (R150G) spectrum displays the characteristic vibronic structure of a bound flavin. Upon white light illumination the flavin undergoes covalent adduct formation. EL362 does not present any spectrum signature indicating the presence of an attached flavin co-factor. EL362 (R150G) photorecovery monitored at 446 nm shows that it undergoes a photocycle after illumination with a dark state recovery that takes approximately 3 hr. c) Autophosphorylation kinetics. EL362 (R150G) (filled circles: dark state, open squares: lit state) has overall higher activity than EL362 (filled triangles). d) SEC-MALLS shows that wildtype EL362 (monomer MW: 40.7 kDa) and is predominantly monomeric in solution (MALLS trace ~50 kDa at center of peak). Upon restoration of cofactor binding by the R150G point mutation, we observed oligomerization (MALLS MW ~120 kDa in dark, ~100 kDa in the light) with some light-dependent effects on elution peak shape and molecular weight distribution (wild-type vs mutant). dRI and MW data are shown as in Figure 2.

We anticipate that these changes in the LOV domain substantially change the domain structure; given their corresponding perturbations of kinase activity, we asked if they affected the EL362 oligomerization state as seen for EL368 (Figure 2). SEC-MALLS experiments showed that wild-type EL362 eluted as mixture of monomers and higher order oligomers, while the mutant eluted primarily in larger complexes. However, different from EL368, photoexcitation of EL362(R150G) shifted the elution profile in comparison to its dark state (Figure 7d), suggesting a relatively large conformational change. Taken together EL368(56-368) and EL362 show different ways that LOV structure is essential to the activity of the HK domain.

DISCUSSION

LOV domains are a widespread group of photoswitches that regulate diverse effectors using light-driven allosteric changes^{3, 11, 15}. These processes have been studied in several systems, particularly the phototropin class of light-regulated serine/threonine kinases in plants. In these proteins, illumination triggers LOV domain conformational changes that disrupt and unfold a C-terminal “Jα helix”, subsequently stimulating kinase activity^{8, 9}. Some LOV-containing proteins appear to work through comparable light-induced release of helical elements, often leading to dimerization^{48, 49}; others alter the conformations of preformed dimers⁵⁰. Given the diversity of possible mechanisms and targets, it is important to understand how LOV domains control a variety of effectors, particularly given open questions about the natural regulation of several of these groups (e.g. histidine kinases).

Here we provide some biochemical insight into the mechanism of this regulation, showing that illumination activates EL368 via a dual effect, both enhancing the turnover rate for the kinase activity while slightly decreasing the affinity of the protein to ATP (Fig 1b). The best-developed mechanistic model for how such regulation might be achieved is based on data from YF1, an engineered LOV-HK protein containing a LOV domain fused onto a HK that is normally controlled by a gas-responsive PAS domain^51–53. This model suggests that light-induced conformational changes at the sensory domain are transmitted to the HK domain through movements in the intervening coiled-coil linker between them, similarly to HAMP domains that often couple sensory and catalytic domains in HKs. Recent crystal structures of both YF1⁵³ and a natural PAS-HK protein⁵⁴ bolster support for this regulatory model.

We believe that EL368 supports the above-mentioned model for two reasons: i) the presence of coiled coils predicted by the COILS server⁵⁵ to be located between V163-R223 to form a DHp domain (Figure S2). ii) the presence of a stable LOV dimer interface, which may provide the anchor point needed to trigger the rotational change. Experimental support for this is provided by EL368(56-368) (Figure 2) and EL362 (Figure 7), which have LOV domains affected by truncation or lack of chromophore binding and correspondingly lower amounts of dimer in solution.

Turning from HK regulation to the subsequent downstream targets, we note that the exciting discovery of prokaryotic LOV photosensory proteins with bioinformatic and biochemical methods^{3, 11, 15, 16, 56} has significantly preceded their functional characterization. To date, only few of these have had assessed biological roles: several histidine kinases (LovK¹², BM-LOV-HK¹⁴, R-LOV-HK¹⁹, Xac-LOV-HK¹⁷) along with the EL222 DNA binding protein⁵⁷ and the anti-sigma antagonist YtvA⁵⁸. Here, our investigations of EL368 and two other LOV-HK proteins add to this list, through the use of unbiased phosphotransfer profiling to identify the kinetically-preferred downstream targets of these enzymes. Our data reveal that the three enzymes present a remarkable kinetic preference for the same RR (EL_LovR), which is a potential cognate pair with EL362 based on sequence proximity (Figures 3, S3, S4). EL368 and EL346 also had a kinetic preference for one additional substrate, EL_PhyR (Figures 3, S3).

Such a “branched” signaling pathway, achieved by multiple enzyme/substrate pairings at the same step, facilitates more complex network behavior than otherwise possible from exclusive links. The biochemical⁵⁹ and structural^{34, 60, 61} bases of HK/RR specificity in a two-component pair is defined by residues located near the phosphorylated histidine within the HK DHp domain and mostly by residues in the RR α1 helix. Inspection of EL_LovR and EL_PhyR sequences show that they share similar α1 helices compared to the other E. litoralis RRs, perhaps specifying their selection by the LOV-HKs (Figure S7A). On the complementary HK surface, we observe sequence similarities among the DHp regions (Figure S7B) as well as differences that may contribute to the differential phosphorylation levels of RR substrates.

Our findings suggest the existence of a complex light-regulated branched pathway in E. litoralis involving “many-to-one” (EL346, EL362, EL368 with EL_LovR) and “one-to-many” (EL368, EL346 with EL_LovR and EL_PhyR) HK and RR relationships. “Many to one” relationships allow the cell to integrate multiple inputs into the phosphorylation levels of a specific RR and consequently undergo appropriate physiological answers for each level³⁸. EL368 and EL346 both feed into the same EL_LovR and EL_PhyR pathway; this apparent redundancy may be explained by either functional importance (e.g. different sensors have differential sensitivity to light) or simply due to a lack of selective pressure against keeping multiple LOV-HK proteins in the genome. We also note that this redundancy provides some robustness in the event of inactivation of one sensor, as perhaps seen by EL368 and EL346 compensating for the impaired function of the non-inducible EL362 kinase (Figure 7). We speculate that perhaps EL362 may actually represent a genetic relic of sorts by having served as the ancestral light sensor in this organism (in a conventional operon arrangement with its cognate RR) prior to having suffered an inactivating mutation in its LOV domain. While rigorous testing of the potential in vivo roles of multiple LOV-HK proteins will require future experimental validation, we note this redundancy is not restricted solely to E. litoralis: multiple eubacteria (including several Methylobacterium species) contain two or more LOV-HK genes.

An additional degree of complexity (“one to many” relationship) is added to the transduction pathway by the ability of EL368/EL346 to phosphorylate EL_PhyR in addition to EL_LovR. Our biochemical analyses of EL_PhyR allow us to classify it as a PhyR-like response regulator; an analogous protein in Methylobacterium extorquens AM1 (which contains multiple LOV-HK proteins, see above) participates in regulation of σ anti-σ interaction system, which is involved in resistance to multiple environmental stresses through activation of several target genes^{20, 21}. Interestingly, Foreman et al. demonstrated that transcription of the LovK-LovR operon in C. crescentus is upregulated by the same general stress σ factor¹⁶. In E. litoralis there is no direct evidence that EL368 and EL346 could be regulated by the σ factor. However, analysis of the intergenic region prior to the potential EL362-EL_LovR operon indicates the presence of a putative operator (GGAACT CCG AGC CGC TCA GTC GGTT) for the σ binding located 83 nucleotides prior to the start codon of EL362 gene (designated by the EL_sigT promoter in Figure S8), suggesting that E. litoralis may adopt a similar system of regulation as observed in the related alphaproteobacterium C. crescentus. Nevertheless, we do observe some differences between C. crescentus and E. litoralis, particularly in the latter use of “orphan” HKs to phosphorylate RR proteins distant in the genome (i.e. EL368, EL346 with EL_PhyR, EL_LovR). Notably, this allows E. litoralis to have a similar light regulated network architecture as observed in C. crescentus¹⁶ despite the inability of native EL362 to be controlled by illumination.

In conclusion, our findings indicate the discovery of a light regulated branched transduction pathway involving a set of several LOV-HKs and stress-related RRs. Additional biochemical and structural studies of these proteins would not only help to understand the molecular mechanisms of activation of HKs, but also enhance our knowledge about the complexity and diversity of two-component signaling systems.

ABBREVIATIONS

LOV domain

Light Oxygen Voltage domain

HK

histidine kinase

RR

response regulator

dRI

differential refractive index

MALLS

multiangle laser light scattering