Natural and Engineered Photoactivated Nucleotidyl Cyclases for Optogenetic Applications*

Cyclic nucleotides, cAMP and cGMP, are ubiquitous second messengers that regulate metabolic and behavioral responses in diverse organisms. We describe purification, engineering, and characterization of photoactivated nucleotidyl cyclases that can be used to manipulate cAMP and cGMP levels in vivo. We identified the blaC gene encoding a putative photoactivated adenylyl cyclase in the Beggiatoa sp. PS genome. BlaC contains a BLUF domain involved in blue-light sensing using FAD and a nucleotidyl cyclase domain. The blaC gene was overexpressed in Escherichia coli, and its product was purified. Irradiation of BlaC in vitro resulted in a small red shift in flavin absorbance, typical of BLUF photoreceptors. BlaC had adenylyl cyclase activity that was negligible in the dark and up-regulated by light by 2 orders of magnitude. To convert BlaC into a guanylyl cyclase, we constructed a model of the nucleotidyl cyclase domain and mutagenized several residues predicted to be involved in substrate binding. One triple mutant, designated BlgC, was found to have photoactivated guanylyl cyclase in vitro. Irradiation with blue light of the E. coli cya mutant expressing BlaC or BlgC resulted in the significant increases in cAMP or cGMP synthesis, respectively. BlaC, but not BlgC, restored cAMP-dependent growth of the mutant in the presence of light. Small protein sizes, negligible activities in the dark, high light-to-dark activation ratios, functionality at broad temperature range and physiological pH, as well as utilization of the naturally occurring flavins as chromophores make BlaC and BlgC attractive for optogenetic applications in various animal and microbial models.

The ability to activate or inactivate signal transduction pathways in vivo, during normal or disease conditions, in specific tissues and at desired times could provide unprecedented insights into cellular regulatory networks. However, the number of pharmaceutical activators or inhibitors of signaling proteins is limited, their target specificities are imperfect, and the spatiotemporal resolution of their action is low. A recently emerged optogenetic method can supplement pharmaceutical approaches (1,2). Optogenetics involves introduction of genes encoding natural or engineered photoactivated proteins with desired activities into cells and tissues of model organisms.
In naturally photoactivated proteins, photons absorbed by photoreceptor protein domains induce conformational changes that are propagated to change activities of the output domains (3). Light is unique in that it can control protein activities in a reversible manner and with high spatiotemporal resolution. Visible light, particularly at low intensities, is practically harmless. The spatiotemporal resolution that can be achieved by using photoregulated proteins is limited only by the width of a laser beam, which can be focused with a subcellular precision. We are at the dawn of an exciting era when optogenetic tools will become common in biomedical applications (4).
The flavin-containing photoreceptors of LOV 4 (17,18) and BLUF (19,20) families are also attractive for optogenetic applications because of their relatively small size (100 -140 amino acids), solubility, and spontaneous incorporation of flavin chromophores into apoproteins. Because flavins are present in all cell types, there is no need for chromophore delivery to target tissues. The biggest limitation of LOV-and BLUF-based photoactivated proteins is that blue light, which they absorb, has intrinsically low tissue penetration capacity. Therefore, external light sources can be used only for small model animals (e.g. Drosophila) (8), whereas surgically implanted photoemitting devices have to be used for larger animals (e.g. mice) (16).
The BLUF-domain containing photoactivated adenylyl cyclase from Euglena gracilis, PAC (21), proved to be a powerful analytical tool to control cAMP levels in Xenopus oocytes and neurons of Drosophila and the mollusk Aplysia (22,23). PAC proved useful despite the large size of its subunits (Ͼ1000 amino acids), difficulties in heterologous expression (24), and high background activity in the dark when expressed in vivo (22).
Here, we describe a small, BLUF domain containing bacterial light-activated adenylyl cyclase, designated BlaC. This enzyme belongs to class III nucleotidyl cyclases (for reviews, see Refs. 25,26). We also describe engineering and characterization of a bacterial light-activated guanylyl cyclase, BlgC. The products of adenylyl and guanylyl cyclases, cAMP and cGMP, are universal second messengers that control a variety of processes ranging from gene expression to ion transport to metabolism. In metazoans, these second messengers affect cell growth and differentiation, blood glucose levels, cardiac contractile function, learning and memory, intestinal fluid and electrolyte homeostasis, retinal phototransduction, among other things (25)(26)(27). We anticipate that the ability to turn on and off cAMP and cGMP synthesis using light, in desired tissues, at desired times during development or disease, will lead to new functional and mechanistic insights into cyclic nucleotide dependent pathways.

EXPERIMENTAL PROCEDURES
Microbiological Methods-Escherichia coli BL21(DE3) and DH5␣ and their derivatives were routinely grown in LB medium (28). For light-dependent experiments, cells were grown at 30°C on MacConkey agar (28) supplemented with 1% lactose. Irradiation was provided by light-emitting diode panels, either the All-blue (emission 465 nm) or All-red (635 nm) LED Grow Light panel 225 (30.5 ϫ 30.5-cm square; LED Wholesalers, CA). Light was administered at an irradiance of ϳ1 W m Ϫ2 for 48 h using the following regimen: 5-s light, 120-s dark.
Recombinant DNA Techniques-The mutation in the adenylyl cyclase cya gene of E. coli BL21(DE3) was constructed by a one-step gene inactivation method described by Datsenko and Wanner (29). The Beggiatoa sp. PS blaC gene (locus_tag BGP_1043; GI:153870309) (30) was synthesized by BioBasics, Inc. with the codon usage optimized for E. coli. The gene was cloned into pBAD/Myc-HisB (Invitrogen) to generate plasmid pBAD-blaC for arabinose-inducible expression in E. coli. For overexpression and protein purification purposes, blaC was cloned into the modified in-house vector pMal-c2x (NEB Biolabs) to generate a maltose-binding protein (MBP)-His 6 fusion (plasmid pMal-blaC). Site-directed mutagenesis using the QuikChange kit (Stratagene) was performed on blaC-containing plasmids to generate a guanylyl cyclase.
Protein Overexpression and Purification-The BlaC protein and its derivatives were purified as C-terminal fusions to the MBP-His 6 tag using amylose affinity chromatography according to specifications of the manufacturer (NEB Biolabs). Protein purification was performed under red light to avoid flavin photobleaching. The overnight cultures of E. coli DH5␣ [pMal-blaC]-expressing MBP-His 6 -BlaC or DH5␣ [pMal-blgC]-expressing MBP-His 6 -BlgC were grown to A 600 0.6 in LB supplemented with 100 g of ampicillin ml Ϫ1 at 30°C. The cultures were transferred to 18°C, and isopropyl 1-thio-␤-D-galactopyranoside was added to a final concentration of 0.5 mM to induce gene expression. Following a 16-h incuba-tion, bacteria were collected by centrifugation at 5,000 ϫ g for 15 min, washed, and resuspended in the amylose column binding buffer (50 mM Tris-HCl, pH 8.0, 350 mM NaCl, 10 mM MgCl 2 , 0.5 mM EDTA, 10% glycerol). Cells were disrupted using a French pressure cell, and cell debris was removed by centrifugation at 35,000 ϫ g for 45 min at 4°C. Two milliliters (bed volume) of amylose resin (NEB Biolabs) preequilibrated with the binding buffer was added to the soluble cell extract derived from a 1.5-liter culture and agitated for 1 h at 4°C. The mix was loaded onto a column, and the resin was washed with 200 ml of column binding buffer. Fractions were eluted with 12 ml of binding buffer containing 10 mM maltose. The protein was either used immediately or stored at Ϫ80°C in 20% v/v glycerol (final concentration). Protein concentrations were measured using a Bradford protein assay kit (Bio-Rad) with bovine serum albumin as the protein standard. Proteins were analyzed using SDS-PAGE.
Enzymatic Assays-Enzymatic assays were performed at room temperature unless specified otherwise. A standard reaction mixture (300 l) contained 5 M enzyme in the assay buffer (50 mM Tris-HCl, pH 8.0, 10% glycerol, 10 mM MgCl 2 , 0.5 mM EDTA). The protein was either kept in red light (Allred LED Grow Light panel) or was irradiated with blue light using the All-blue LED Grow Light panel at the approximate irradiance of 10 W m Ϫ2 for the duration of the assay. The reaction was started by the addition of ATP or GTP. Aliquots (50 l) were withdrawn at different time points and boiled for 5 min. The precipitated protein was removed by centrifugation at 15,000 ϫ g for 5 min. The supernatant was filtered through a 0.22-m pore size filter (MicroSolv) and analyzed by reversed-phase HPLC.
Nucleotide Detection-Nucleotide mixtures from enzymatic assays were separated and analyzed by the reversed-phase HPLC (Summit HPLC System, Dionex, Sunnyvale, CA), 15 ϫ 4.6-cm Supelcosil LC-18-T column (Sigma) using a gradient of the phosphate-methanol buffer system that was slightly modified from what was described earlier (31).
Intra-and extracellular cAMP and cGMP levels in E. coli cells that expressed BlaC or BlgC were analyzed by using an ELISA kit (Biomol) according to the instructions of the manufacturer. For these assays, exponentially (A 600 0.6) grown cultures were induced by the addition of 0 -1% (pBAD-blaC) or 1% (pBAD-blgC) arabinose. After induction, cultures were grown in the dark for an additional 16 h. Cells were subsequently collected by centrifugation, resuspended in 200 l of phosphate-buffered saline (28), placed on the top of a plastic surface, and either irradiated with blue light (10 W m Ϫ2 ) for 10 min or kept in the dark at room temperature. Cells were collected by centrifugation. Cell-free supernatant was used for analysis of extracellular cyclic nucleotides, whereas whole cell extracts obtained by sonication of precipitated cells were used for analysis of intracellular cyclic nucleotides.
Spectroscopy-Electronic absorption spectra were recorded with a UV-1601 PC UV-visible spectrophotometer (Shimadzu) at room temperature. Protein solutions (100 l) in 10-mm light path quartz cuvette were irradiated directly in the spectrophotometer from the top of the cuvette. The light originated from a halogen lamp (EKE 21V150 W, General Electric) with a flexible light guide.
Bioinformatics-To identify residues important for substrate specificity, a combination of sequence and structure analysis was utilized. A multiple sequence alignment of class III nucleotidyl cyclases was generated using MUSCLE (32). A phylogenetic tree was constructed using a maximum likelihood algorithm, PhyML 3.0 (33), with a model proposed by ProtTest (34). A homology model of BlaC was constructed using the Swiss Model server in project mode (35,36). The models were based upon Protein Data Bank (PDB) code 1WC5 and the multiple sequence alignment from the sequence analysis step. Protein structures and models were analyzed using Discovery Studio 2.5 (Accelrys) and Swiss PDB viewer (37).

RESULTS
The BlaC Protein from Beggiatoa sp. PS Is a Blue Light-activated Adenylyl Cyclase-In the genome of the uncultured Gammaproteobacterium Beggiatoa sp. PS we found a gene (locus_tag BGP_1043) that could encode BlaC. The BlaC protein contains two domains: BLUF (PF04940) (38), representing sensors of blue light using FAD, and guanylate_cyc (PF00211) representing an adenylyl or guanylyl cyclase catalytic domain (Fig. 1A). The latter, catalytic domain is also known as cyclase homology domain (25,26). Because the ability of bacteria to produce cGMP has never been convincingly demonstrated, we presumed that BlaC possesses adenylyl cyclase activity. The protein size, 350 amino acids, is much smaller than the sizes of ␣ or ␤ subunits of PAC from E. gracilis, which are approximately 1000 amino acids. This makes BlaC an attractive alternative to PAC, which is currently used in optogenetic applications involving manipulations of the cAMP levels (22,23).
To test whether BlaC had the predicted function, we constructed the BL21(DE3) cya mutant, which lacks a native adenylyl cyclase, and expressed the blaC gene from an arabinose-inducible promoter (plasmid pBAD-blaC). To improve protein expression, the blaC gene was synthesized based on the optimum E. coli codon usage. The lack of cAMP results in the inability of the BL21(DE3) cya mutant to up-regulate expression of the cAMP-CRP-dependent operons and prevents this strain from growing on minimal medium containing lactose. Expression of blaC restored growth of BL21(DE3) cya on MacConkey agar containing lactose and the inducer, 0.2% L-arabinose, only if the strain was grown in the light but not in the dark (Fig. 1B). The growth of BL21(DE3) cya expressing blaC could also be restored upon intermittent illumination with blue light emitted by a LED panel. Exposure to red light did not restore growth (not shown). This test confirmed the prediction that BlaC functions as a blue light-activated adenylyl cyclase.
To demonstrate that BlaC can be used for high precision regulation of cAMP synthesis in vivo, we spread a lawn of BL21(DE3) cya [pBAD-blaC] cells on a Petri dish containing MacConkey agar plus lactose and covered the plate with thick black paper, leaving an image in the center of the plate exposed to intermittent irradiation with blue light. As expected, only cells exposed to blue light were able to grow (Fig. 1C). Thus, BlaC can be used to control cell behavior (E. coli growth in this instance) by light with high spatial resolution.
Purification and Biochemical Characterization of BlaC-We overexpressed BlaC in E. coli as an N-terminal His 6 fusion and as a C-terminal fusion to the MBP-His 6 tag. The His 6 -BlaC fusion proved to be insoluble, regardless of overexpression conditions, and its purification was not pursued further. However, the MBP-His 6 -BlaC was partially soluble, particularly when induced at low temperature (18°C). We have purified this protein to Ͼ95% purity ( Fig. 2A). Because removal of the MBP-His 6 tag resulted in protein aggregation, all subsequent characterization was done using the MBP-His 6 -BlaC fusion (subsequently called BlaC for simplicity).
The protein, as purified, contained a mixture of flavins, FAD and FMN, as observed earlier (39). By extracting the cofactors and comparing protein and flavin amounts, we estimated that ϳ90% of the protein was in the holoform (not shown). Subsequent protein characterization was performed using proteins "as purified." The adenylyl cyclase activity was assayed using 5 M protein at pH 8.0 at room temperature. Under these conditions, adenylyl cyclase activity in the dark was undetectable for ϳ30 min. Irradiation with blue light (10 W m Ϫ2 ) increased activity by Ͼ100-fold (Fig. 2B). Precise calculation of the photoactivation fold was impossible because of the sensitivity of cAMP detection. The activity assays at different concentrations of Shown is growth on lactose of the E. coli BL21(DE3) cya mutant containing an empty vector, pBAD, or pBAD-blaC. Plates containing MacConkey agar plus lactose were grown at 30°C for 48 h in the dark or upon intermittent irradiation (5-s light, 120-s dark) with blue light (1 W m Ϫ2 ). C, high precision control of BL21(DE3) cya [pBAD-blaC] growth with light. The cells were spread evenly over the surface of the plate, which was covered with black paper except for the bucking bronco image. The plate was incubated as described in B. DECEMBER 31, 2010 • VOLUME 285 • NUMBER 53 the substrate, ATP, yielded K m 0.5 mM and V max 57 nmol (mg of protein) Ϫ1 min Ϫ1 (Fig. 2C). Adenylyl cyclase activity was dependent on the intensity of blue light irradiation. At the highest irradiation levels used here, 10 W m Ϫ2 , the maximum enzymatic activity was not reached (Fig. 2D). The pH optimum for BlaC was found to be 9.5 (Fig. 2E). The temperature optimum of BlaC was found to be 45°C (Fig. 2F). Such a high temperature optimum was somewhat unexpected for a bacterium isolated from the 4-m depth in the Baltic Sea (30). Under room temperature, BlaC had approximately one-forth of its maximal activity. The small size, lack of activity in the dark, high photoactivation fold, relatively high specific enzymatic activity, and functionality in the range of physiological pH and temperatures compatible with various microbial and animal models make BlaC particularly attractive for optogenetic applications.

Photoactivated Adenylyl and Guanylyl Cyclases
Photochemical Properties of BlaC-The electronic absorption spectrum of BlaC had a typical flavin spectrum. Upon irradiation, the spectrum underwent a 9-nm red shift, which is within the 5-10-nm range observed in BLUF domain photoreceptors characterized earlier (Fig. 3A).
After the light switch-off, flavin absorption recovered completely to the preirradiation dark state following single-exponential kinetics. At room temperature, half of the protein recovered back to the dark state by ϳ16 s (Fig. 3B). The fast return to the dark, inactive state is another important factor for an optogenetic probe that allows one to induce short, physiologically relevant cAMP impulses.
Importantly, prolonged exposure of BlaC to blue light at the highest intensity used here, 10 W m Ϫ2 , resulted in only a minor loss of the adenylyl cyclase activity (Fig. 3C). A high level of tolerance to continuous irradiation is yet another attractive feature of BlaC.
Conversion of BlaC into a Guanylyl Cyclase-The use of photoactivated guanylyl cyclases could greatly improve our understanding of cGMP signaling pathways. As no such enzymes currently exist, we set out to convert BlaC to a photoactivated guanylyl cyclase by changing its substrate specificity from ATP to GTP. A similar conversion, albeit in the opposite direction, from guanylyl to adenylyl cyclase, has been achieved earlier (40) by replacing as few as two amino acid residues.
To identify residues in BlaC involved in substrate binding, we performed a bioinformatics analysis of sequences and structures of adenylyl and guanylyl cyclase domains. An emphasis was placed on adenylyl or guanylyl cyclase domains from multidomain proteins of various architectures. This approach diminished conservation in the residues involved in inter-or intramolecular interactions while retaining the conserved residues essential for enzymatic activity.
We constructed a dataset of class III nucleotidyl cyclases consisting of three subsets: eukaryotic guanylyl cyclases (6 sequences), bacterial adenylyl cyclases (9 sequences), and a single putative bacterial guanylyl cyclase domain (Cya2 from Synechocystis sp. PCC 6803) (41). Each group has a representative three-dimensional structure, i.e. PDB 1WC5, 2W01, and 3ET6, respectively. The sequences of these proteins were aligned using MUSCLE (Fig. 4A and supplemental Fig. S1). A phylogenetic tree was constructed to confirm the subsets using PhyML (supplemental Fig. S2).
The multiple sequence alignment was analyzed to identify residues potentially involved in a functional shift between the  different subsets, e.g. residues that change substitution rate between the eukaryotic guanylyl cyclases and the bacterial adenylyl cyclases, or residues that change their physicochemical character between subsets. This analysis led us to three such residues in BlaC, i.e. Lys 197 , Asp 265 , and Thr 267 (Fig. 4A). All adenylyl cyclases in the selected subset have Lys at position equivalent to position 197 in BlaC, whereas all guanylyl cyclases sequences have Glu. At position 265, all adenylyl cyclases sequences have Asp or Glu, whereas all eukaryotic guanylyl cyclases have Arg, and the putative bacterial guanylyl cyclase has Lys. At position 267, all adenylyl cyclases have Thr, all eukaryotic guanylyl cyclases have a Cys, and the putative bacterial guanylyl cyclase has Gly.
We verified the significance of sequence analysis by performing structural analysis. The structure of a bacterial adenylyl cyclase with a bound ATP analog (1WC5) provided us with information about the residues located in the adenosinebinding pocket. A homology model of BlaC was built based upon that structure ( Fig. 4B and supplemental Fig. S3). Importantly, the residues, whose importance was suggested by sequence analysis, were found to be located in the adenosinebinding pocket of BlaC (Fig. 4, C and D).
In addition to these three residues, position Ala 277 (Fig.  4A), which is either a Ser or a Tyr in guanylyl cyclases, was added to the set of potential sites for site-directed mutagenesis. Based upon this analysis, the following individual mutations and combinations of mutations were engineered into the BlaC protein: K197E, D265K, D265R, T267G, T267C, and A277Y. All mutant proteins were overexpressed and purified, and their adenylyl and guanylyl cyclase activities were tested in vitro.
The single mutations, K197E or D265K, abolished adenylyl cyclase catalytic activity and did not result in any guanylyl cyclase activity (Fig. 5). Inspection of the guanylyl cyclase structures revealed that the original residues at these positions interact at the dimer interface. Hence, mutation of single residues might have resulted in electrostatic repulsion (Glu 197 and Asp 265 or Lys 197 and Lys 265 ) between monomers. A double mutant, K197E/D265K, on the other hand, gained some guanylyl cyclase activity while retaining significant adenylyl cyclase activity (Fig. 5). Simultaneous mutations of both K197E and D265K were necessary for guanylyl cyclase activity as evident from analysis of the following mutants: K197E/ T267G, K197E/A277Y, D265K/T267G, and D265K/A277Y (Fig. 5). The single or double mutations, T267G or A277Y, resulted in lower adenylyl cyclase activity but no gain in guanylyl cyclase activity (Fig. 5).

Photoactivated Adenylyl and Guanylyl Cyclases
Biochemical Characterization of BlgC-We purified MBP-His 6 -BlgC (subsequently referred to as BlgC) to Ͼ95% purity (Fig. 6A) and assayed its enzymatic activity in a manner similar to that of BlaC. Guanylyl cyclase activity of BlgC was undetectable in the dark and increased upon irradiation by approximately 2 orders of magnitude, i.e. the introduced mutations did not affect photoactivation properties of the protein (Fig.  6B).
The three introduced mutations drastically increased affinities of BlgC for ATP and GTP, compared with BlaC, i.e. K m(GTP) 25 M and K m(ATP) 10 M. At relatively low substrate concentrations, V max for the guanylyl compared with adenylyl cyclase reaction was ϳ7-fold higher, i.e. 19.5 versus 2.6 nmol (mg of protein) Ϫ1 min Ϫ1 (Fig. 6, C and D). Therefore, the predominantly guanylyl cyclase activity of BlgC is due to the higher turnover rate of GTP compared with ATP and despite the lower affinity for GTP compared with ATP.
We found that high, millimolar concentrations, GTP and ATP strongly inhibited guanylyl cyclase and residual adenylyl cyclase activities of BlgC (Fig. 6, C and D). Under our experimental conditions, adenylyl cyclase activity of BlgC in vitro was no longer detectable at 4 mM ATP (Fig. 6C). The significance of this observation for the performance of BlgC in vivo is discussed in the following section. The pH and temperature dependence of the guanylyl cyclase activity of BlgC were similar to those of BlaC (Fig. 6, E and F).

BlaC Functions as a Photoactivated Adenylyl Cyclase in
E. coli-We tested performance of BlaC and BlgC in vivo using E. coli BL21(DE3) cya as a model. We induced BlaC expression in BL21(DE3) cya [pBAD-blaC] for 16 h in the dark with 0 -1% arabinose, collected cells, resuspended them in a small volume, and irradiated them with blue light at 10 W m Ϫ2 (or kept them in the dark) for 10 min at room temperature. We measured cAMP and cGMP levels directly using a sensitive ELISA. E. coli has been reported to excrete most cAMP into the medium (42). We confirmed that, following irradiation of BL21(DE3) cya [pBAD-blaC], the extracellular cAMP concentration was 10-fold higher than the intracellular concentration (data not shown). Interestingly, extracellular and intracellular cGMP concentrations were similar, which likely reflects less efficient cGMP efflux in E. coli, compared with the cAMP efflux. Because measuring extracellular cyclic nucleotide levels proved more sensitive, we focused on these measurements.
The readings of the BL21(DE3) cya strain containing an empty vector corresponded to ϳ0.30 pmol of cAMP or cGMP/10 9 cells (Table 1). Because BL21(DE3) cya produces neither cAMP nor cGMP, these values represent nonspecific background. Expression of BlaC at low levels (induction with 0 -0.1% arabinose) in the dark did not affect the background levels (Table 1), which is consistent with the lack of adenylyl cyclase activity of BlaC in the dark in vitro (Fig. 2B). Irradiation of cells with blue light increased extracellular cAMP levels to 2.9, 13.0, or 36.3 pmol cAMP/10 9 cells at 0, 0.05, or 0.1% arabinose, respectively (Table 1). Therefore, one can photoactivate cAMP levels in E. coli by at least 2 orders of magnitude without increasing the background (dark) cAMP concentrations.  When BlaC was expressed at higher levels (induction with 0.2 and 1% arabinose), the undesirable cAMP synthesis in the dark became measurable (Table 1). We detected no contaminating guanylyl cyclase activity of BlaC even at the highest expression levels tested (Table 1).
BlgC Functions as a Photoactivated Guanylyl Cyclase in E. coli-We tested the performance of BlgC in E. coli in a manner similar to that of BlaC. First, we investigated whether the residual adenylyl cyclase activity of BlgC observed in vitro (Fig. 5) was sufficient to sustain growth of BL21(DE3) cya [pBAD-blgC] on MacConkey agar plus lactose. We found that it was not sufficient, whether the plates were grown in the dark or light, even when pBAD-blgC was induced with 1% arabinose (not shown).
We then proceeded to measure cAMP and cGMP synthesis in BL21(DE3) cya [pBAD-blgC] in the dark and light. A 10min irradiation with blue light of BL21(DE3) cya [pBAD-blgC] resulted in the extracellular cGMP levels of 4 pmol of cGMP/ 10 9 cells, whereas cGMP levels in the dark remained undetectable (Table 1). Therefore, BlgC can be photoactivated in vivo by at least 12-fold.
We detected no cAMP despite the relatively high expression of BlgC (1% arabinose) and much more sensitive cAMP detection compared with the detection of cGMP. Therefore, BlgC functions as a specific, photoactivated guanylyl cyclase in E. coli. The lack of cGMP signaling pathways in E. coli prevented us from exploring how BlgC might affect cell behavior.

DISCUSSION
In this study, we characterized a novel, bacterial photoactivated adenylyl cyclase, BlaC, and engineered a photoactivated guanylyl cyclase, BlgC. BlaC originates from a filamentous sulfur proteobacterium, Beggiatoa sp. PS, where its function is unknown. Our analysis revealed that BlaC possesses several properties that make it highly desirable for optogenetic applications. BlaC has a relatively small size, 350 amino acids, and uses flavin chromophores (FAD and/or FMN) present in all cell types. The adenylyl cyclase activity of BlaC in the dark is practically nonexistent in vitro (Fig. 2B) and in vivo (E. coli) when the protein is expressed at relatively low levels ( Fig. 1B and Table 1). The extent of photoactivation of BlaC can be readily adjusted by manipulating the expression level of BlaC (Table 1) as well as the intensity and duration of irradiation ( Fig. 2D). Importantly, BlaC is very tolerant to prolonged exposure to light (Fig. 3C).
Our photoactivation experiments in E. coli showed that cAMP levels can be increased by Ͼ2 orders of magnitude without undesirable dark activity (Table 1). Another advantageous property of BlaC is that the half-life of its lit state is short, i.e. 16 s at room temperature (Fig. 3B). This means that one can induce short spikes in cAMP synthesis, which is important for studying cAMP-dependent signal transduction pathways in vivo. It is also noteworthy that BlaC can operate in a range of temperatures and at physiological pH (Fig. 2, E and F), which makes it applicable for a variety of microbial and animal models.
Currently, no photoactivated guanylyl cyclases exist. To convert BlaC to a guanylyl cyclase, we reengineered its substrate-binding site. Our engineering efforts were facilitated by the available x-ray structures of eukaryotic and bacterial adenylyl and guanylyl cyclases. A combination of three mutations at the substrate-binding pocket (K197E/D265R/T267G) was sufficient to turn BlaC into BlgC, which has predominantly guanylyl cyclase activity. Interestingly, the affinity of BlgC for GTP is somewhat lower than that for ATP, whereas the rate of the guanylyl cyclase reaction is severalfold higher than that of the adenylyl cyclase reaction. Therefore, it is the kinetic factor, not substrate specificity, that determines the prevailing enzymatic activity of BlgC. Affinities comparable with those of GTP and ATP have been reported in native guanylyl cyclases, e.g. mammalian soluble guanylyl cyclase (43). The predominantly guanylyl cyclase activity of the catalytic domain from the putative bacterial guanylyl cyclase Cya2 also stemmed from more efficient GTP turnover (41).
Both guanylyl and residual adenylyl cyclase activities of BlgC are strongly inhibited by millimolar concentrations of ATP and GTP. The reason for this inhibition remains to be explored. In E. coli grown exponentially in a glucose-containing medium, intracellular ATP and GTP concentrations are ϳ10 mM ATP and 5 mM GTP (44). In cells grown at less than maximum growth rates, intracellular ATP and GTP levels are in the lower millimolar range (11). The inhibition by millimolar substrate concentrations explains why, even at relatively high expression of BlgC in the irradiated E. coli cells, cGMP a Proteins were expressed in BL21(DE3) cya containing vector pBAD/Myc-HisB (none), pBAD-blaC (BlaC), or pBAD-blgC (BlgC). b Induction with arabinose was carried for 16 h in the dark. c Cells resuspended in phosphate-buffered saline were kept in the dark or irradiated with blue light for 10 min. Extracellular cyclic nucleotides were measured using ELISA and normalized to 10 9 cells, which were assumed to yield A 600 1.0. d -, not done. DECEMBER 31, 2010 • VOLUME 285 • NUMBER 53 was accumulated to modest levels (compared with the levels of cAMP produced by BlaC) ( Table 1).

Photoactivated Adenylyl and Guanylyl Cyclases
Photoactivation of the guanylyl cyclase activity of BlgC in E. coli was Ͼ12-fold. Given that in vitro BlgC is photoactivated by 2 orders of magnitude (Fig. 6B), similar to the photoactivation fold of BlaC (Fig. 2B), and that photoactivation of BlaC in vitro and in vivo was similar, it is likely that BlgC is also photoactivated in vivo by 2 orders of magnitude.
BlgC possesses ϳ10% residual adenylyl cyclase activity in vitro (Fig. 5). This contaminating activity is undesirable for specific activation of cGMP-signaling pathways in vivo. Because intracellular concentrations of ATP are 2-to severalfold higher than those of GTP and because BlgC is inhibited by the physiological ATP levels stronger than by the physiological GTP levels (Fig. 6, C and D), the 10:1 guanylyl:adenylyl cyclase ratio displayed by BlgC in vitro is likely to be much higher in vivo. Consistent with this expectation, cAMP synthesis in E. coli is undetectable at the highest expression levels of BlgC tested here (Table 1). Therefore, BlgC can be used as a specific photoactivated guanylyl cyclase in vivo. In conclusion, we suggest that both native photoactivated adenylyl cyclase, BlaC, and engineered guanylyl cyclase, BlgC, can be used as optogenetic tools to manipulate cAMP and cGMP signaling pathways or to generate synthetic cAMP-or cGMP-dependent signaling cascades in various model systems.