Luminescence-activated nucleotide cyclase regulates spatial and temporal cAMP synthesis

cAMP is a ubiquitous second messenger that regulates cellular proliferation, differentiation, attachment, migration, and several other processes. It has become increasingly evident that tight regulation of cAMP accumulation and localization confers divergent yet specific signaling to downstream pathways. Currently, few tools are available that have sufficient spatial and temporal resolution to study location-biased cAMP signaling. Here, we introduce a new fusion protein consisting of a light-activated adenylyl cyclase (bPAC) and luciferase (nLuc). This construct allows dual activation of cAMP production through temporally precise photostimulation or chronic chemical stimulation that can be fine-tuned to mimic physiological levels and duration of cAMP synthesis to trigger downstream events. By targeting this construct to different compartments, we show that cAMP produced in the cytosol and nucleus stimulates proliferation in thyroid cells. The bPAC–nLuc fusion construct adds a new reagent to the available toolkit to study cAMP-regulated processes in living cells.

Thus, using a similar strategy as the luminopsin design (54), we demonstrate that the fusion of photoactivated AC from Beggiatoa (bPAC) and nanoluciferase (nLuc) allows dual photo and chemical regulation of cAMP production for live-cell applications. Additionally, we have added a layer of spatial regulation using various targeting motifs to different subcellular compartments. We found cytosolic and nuclear cAMP is sufficient to trigger cell mitogenesis. Thus, the temporal and spatial flexibility of this construct makes it widely adaptable for in vitro and potentially in vivo applications.

Luminescence-activated cyclase allows photo and chemical activation of cAMP production
To develop a tool with broad tunability, we prepared a fusion protein consisting of bPAC and nLuc. bPAC is a bluelight-sensitive cyclase with fast on/off kinetics and a high rate of synthesis (55). nLuc is a small (19 kDa) luciferase that emits blue-shifted luminescence in response to the luciferins furimazine (Fz) and h-coelenterazine (h-CTZ) (56). A Myctag was included to aid identification and visualization (Fig.  1A). In transfected HEK293 cells, h-CTZ elicits luminescence in the blue light range (maximum 455 nm) in a dosedependent manner (Fig. 1, B and C), demonstrating nLuc retains its luminescent properties in the fusion protein.
Luminescence was visible to the naked eye and could be imaged using long exposure times (Fig. S1A). Fz stimulation also generated luminescence with a maximum of 455 nm in a dose-dependent manner (Fig. S1, B and C). We used h-CTZ primarily for these experiments due to its commercial availability and sufficient activation of bPAC-nLuc. Chemical stimulation of nLuc activates light-sensitive bPAC. Using HC1 cells that have low basal levels of cAMP (57,58), h-CTZ stimulation in the presence of phosphodiesterase inhibitor IBMX induces up to Ͼ100-fold increase in cAMP accumulation within 10 min, as shown by cAMP ELISA analysis (Fig.   Figure 1. Luminescence-activated cyclase characterization. A, design of dual photo and chemical activator of cAMP production by fusion of bPAC and nLuc. B, transiently transfected HEK293 cells treated with 100 M h-CTZ luminesce at a maximum of 455 nm (mean Ϯ S.D. of n ϭ 4). C, this effect occurs in a dose-dependent manner (mean Ϯ S.D. of n ϭ 4). D, h-CTZ increases cAMP in HC1 cells expressing bPAC-nLuc in a dose-dependent manner as shown by ELISA analysis (mean Ϯ S.D. of n ϭ 4). E, accumulation of cytosolic cAMP in HEK293 co-expressing H208 and MYR-bPAC-nLuc stimulated with pulses of 4.4 W/mm 2 light (mean Ϯ S.D. of n ϭ 8 cells). F, stimulation with 1:1000 Fz promotes sustained cAMP production well over 30 min (mean Ϯ S.D. of n ϭ 4 cells). RLU, relative luminescence units.
The dual stimulation of cAMP production by bPAC-nLuc represents two signaling modalities: transient pulses of cAMP by light, and sustained cAMP production by luciferin stimulation. It is now well established that many GPCRs stimulate a transient peak of cAMP production from the plasma membrane and a following sustained phase of cAMP production during receptor endocytosis and redistribution in endosomes (31-33, 35, 59). Furthermore, multiple studies found oscillating patterns of cAMP regulate downstream functions (41)(42)(43)(44)(45)(46). To test whether bPAC-nLuc could mimic these patterns of cAMP signaling, we co-expressed membrane-localized bPAC-nLuc (MYR-bPAC-nLuc) and the cAMP FRET-based sensor, YFP-EPAC-Q270E-mScarletI (H208), in HEK293 cells. MYR-bPAC-nLuc generated transient spikes of cAMP accumulation in response to pulses of blue light between 1 and 100 s at 4.4 W/mm 2 (Fig. 1E), demonstrating bPAC retains its optogenetic properties in the fusion protein. cAMP levels returned to baseline values quickly, indicating tight temporal regulation with light. Fz induced a significantly longer signal, lasting wellbeyond 30 min (Fig. 1F). Fz itself did not affect FRET ratio recordings (Fig. S4B). We found the duration and amplitude of cAMP levels were easily tuned by altering light intensity and duration or luciferin concentration and duration of exposure. Together, these two methods of stimulation can potentially tease out the individual effects of the transient, sustained, and oscillating temporal patterns of cAMP signaling.

bPAC-nLuc stimulates cAMP-dependent thyroid cell proliferation
Sustained cAMP production is beneficial for studying longterm processes such as cell proliferation. Using the PCCL3 rat thyroid cell model (60), which requires continuous thyroidstimulating hormone (TSH)/cAMP signaling for S-phase entry (61-63), we examined the effects of light and chemical bPAC-nLuc activation. To find suitable conditions for light activation, we co-expressed bPAC-nLuc with a red cAMP dimerizationdependent sensor that was available at the time, prior to the H208 sensor. Activation using milliseconds to seconds-long pulses of 450 nm light triggered high cAMP production with levels returning to baseline values quickly ( Fig. 2A). We selected a 500-ms pulse at 4.4 W/mm 2 , which produced similar amplitudes of cAMP production compared with TSH-mediated cAMP (Fig. S2). PCCL3 transiently expressing bPAC-nLuc or bPAC-mCherry showed increased S-phase entry compared with controls when stimulated every 15 min over 24 h (Fig. 2B). Under these conditions where bPAC was overexpressed, there was a significant increase in basal proliferation, most likely due to residual dark activity (55, 64). This high basal activity potentially masked the full effects of light-stimulated proliferation; yet a modest but consistent increase was found. While optimizing conditions to stimulate proliferation, we suspected both the high basal activity and repeated pulses of light required for further stimulation suggested that continuous cAMP production was necessary.

Sustained cAMP signaling from localized bPAC-nLuc stimulates proliferation
To investigate sustained, localized cAMP signaling, we incorporated different subcellular targeting motifs to bPAC-nLuc. bPAC and nLuc were initially selected for their small size, resulting in a final ϳ60-kDa fusion protein. Because it is just at the upper limit for passively diffusing through nuclear pores (65), we observed the unmodified construct evenly distributed throughout the cytosol and nucleus. By modifying the N and/or C termini with targeting motifs, bPAC-nLuc localization can be re-distributed to specific subcellular regions. In PCCL3, constructs localized to the nucleus (NLS), cytosol (NES), endoplasmic reticulum (ER) lumen, and membrane (CAAX and MYR, where AA is aliphatic amino acid), as expected ( Fig. 3A and Fig.  S2D). All constructs luminesce immediately upon Fz stimulation, indicating effective drug access to the different compartments (Fig. S3A).
To monitor the behavior and efficacy of the targeted constructs, we modified the cytosolic H208 FRET sensor with an NLS sequence. The nuclear-targeted sensor (NLS-H208) exhibited similar cAMP affinity (EC 50 ), Hill coefficient, and range of activation in cell lysates (Fig. 4C). In transfected PCCL3 cells with similar levels of bPAC-nLuc expression, Fz-stimulated ER-bPAC-nLuc produced a small, transient For sustained, chemical activation of bPAC-nLuc, Fz was specifically designed to activate nLuc providing a high signal and long duration (56). We found Fz-stimulated luminescence was still detectable after ϳ2 h (Fig. S3A). However, Fz was unable to induce cell proliferation in PCCL3 cells expressing bPAC-nLuc (data not shown). New Fz derivatives, such as endurazine and its prototype Fz-4377, have been developed to provide long-lasting luminescence and lower toxicity. Fz-4377 contains a protecting group that requires esterase hydrolysis for activity, resulting in low but stable luminescence for Ͼ6 h (Fig.  S3B). PCCL3 cells stably expressing bPAC-nLuc triggered G 1 /S phase transition upon Fz-4377 treatment, indicating longlasting luminescence-stimulated cAMP production is sufficient for a full proliferative response (Fig. 3C) at levels comparable (ϳ30%) to 10 M forskolin or 1 international milliunit/ml TSH stimulation (62,66). NLS and NES bPAC-nLuc showed a similar increase; however, bPAC-nLuc localized to the ER lumen was unable to stimulate cell proliferation, despite its higher expression levels compared with the other constructs (Fig.  S3C). Unlike the overexpression model, basal activity was not significantly increased. Thus, the combined results indicate that sustained cAMP synthesized in the cytosol or nucleus can stimulate thyroid cell proliferation.

Discussion
The cAMP field evolved from the original free diffusion model to a more complex model where cAMP amplitudes, kinetics, and subcellular location became critical elements to generate signaling specificity. Accordingly, sensors with greater sensitivity and dynamic range and actuators able to mimic different cAMP kinetic profiles were developed to test the new hypotheses. Optogenetic tools offer precise spatio-temporal resolution, and several new constructs developed were able to modulate cAMP levels in cells and animals. In this study, we report the characterization of bPAC-nLuc, a new reagent with dual photo and chemical stimulation, to study cAMP regulated processes in living cells.
Most studies approach the identification of cAMP compartments by measuring local cAMP levels with targeted FRET sensors or signalosome disruption (67,68). However, causal association between a specific cAMP pool with a biological function requires reagents for localized cAMP synthesis. The soluble catalytic loops from G-protein-regulated cyclases, like the sACI/II chimera, were used as a forskolin-sensitive tool to increase cAMP (69). Truncated soluble AC (sAC t ), activated by bicarbonate, has also been used to study the effects of membrane, cytosolic, and nuclear cAMP signaling (23). However, in both scenarios, activation also stimulates endogenous ACs. pH changes from bicarbonate can also interfere with fluorescent sensors and other cyclases (70,71). Tsvetanova and von Zastrow (72) used targeted bPAC to show that cAMP produced at the plasma membrane or from endosomes yielded different transcriptional profiles. The temporal precision of bPAC is useful for transient activation of high cAMP concentrations; however, long-term stimulation can be limited by the hardware used to regulate light exposure (i.e. maintaining high voltage levels and overheating of LED) (73). To circumvent these current obstacles, the luminescence-activated cyclase presented

ACCELERATED COMMUNICATION: bPAC-nLuc regulates cAMP dynamics
here offers the ease of chemical activation without activating endogenous cyclases and allows dual-light stimulation within the same construct to compare the effects of transient with sustained activation.
The bPAC-nLuc fusion protein is widely adaptable for experimental use. Both light and luciferin concentrations modulate cAMP synthesis in a predictable, tunable manner. bPAC-nLuc is also small in size and can be targeted to different compartments. Using a custom-built Arduino-controlled LED system (Fig. S6 -S9 and Table S1), we can stimulate spikes of cAMP accumulation for seconds to minutes. With the advent of long-lasting Fz derivatives, chemical stimulation can promote luminescence for Ͻ1 h (h-CTZ) or beyond 6 h (Fz-4377). This combination presents a broad range of kinetic signatures to mimic different physiological cAMP signaling patterns within the same construct. A few disadvantages remain, including the need to minimize light exposure while working and high dark activity. In our thyroid cell model, overexpression leads to increased proliferation; fortunately, stable cell lines with lower expression exhibited lower basal proliferation. Hence, titrating expression with an inducible system may be recommended. The residual dark activity of bPAC has been noted by others (64) and must be considered when interpreting results.
Here, we demonstrate long-term luminescent stimulation of cAMP production can induce cAMP-dependent events, such as cell proliferation in thyroid epithelial cells. The thyroid represents a prototypical cAMP-responsive tissue where thyroid-stimulating hormone receptor (TSHR) signaling through cAMP is responsible for a majority of thyroid function (74). The combination bPAC-nLuc and long-lasting Fz-4377 stimulates proliferation similar to TSH-stimulated cAMP signaling (62,66). bPAC-nLuc has the potential to test both the spatial (e.g. compartments involved) and temporal (e.g. along G 0 -G 1 /S) requirements of cAMP, which remain undefined in the field. cAMP synthesized in the ER lumen exhibited low cAMP levels reaching the nucleus and cytosol and was unable to stimulate proliferation. This may be a result of lower construct activity or the ER membrane serving as a physical barrier to cAMP diffusion (75,76). Sustained cytosolic and nuclear cAMP accumulation sufficiently triggered proliferation. We found no distinct compartmentalization between the two regions with Fz stimulation, i.e. locally synthesized cAMP did not remain local. Sustained cAMP production or high concentrations may overcome the endogenous machinery maintaining cAMP gradients and mask potential compartmentalization. Hence, careful calibration is needed to compare regions. However, it was interesting to find nuclear-generated cAMP was sufficient to trigger proliferation; whether nuclear cAMP action is sufficient in the absence of a cytosolic component deserves further investigation. Considering that TSHR signaling activates membrane ACs, this finding suggests either cAMP diffusion into the nuclear compartment or, more likely, additional sources of cAMP, e.g. nuclear sAC (77), may play an important role in TSH action.
In summary, we developed a new tool that will aid in mechanistic studies of cAMP biology. bPAC-nLuc is one of the few available chemically-activated tools that does not stimulate endogenous ACs. Furthermore, the Arduino-controlled system presented supports adjustable timing and intensity in range with bPAC's optical properties. Dual photo and chemical activation allows flexibility in multiple models and experimental setups. Although it is not explored here, bPAC-nLuc has many potential applications in vivo for target organs where LED implants would be too invasive. Transgenic mice expressing bPAC have been reported (78), and Fz has been used with success in mice to activate nLuc (79). Careful dosing is necessary to prevent toxicity (80), although new analogs for in vivo studies are being developed. The ability to synthetically stimulate localized cAMP signaling will play an important role, as the field of cAMP compartmentalization moves toward studying disease models and potential therapeutic interventions.
The cAMP FRET sensor H208 (YFP-EPAC-Q270E-mScarletI), kindly provided before publication by Dr. Jalink, consists of a YFP and mScarletI (81) fluorescent pair flanking catalytically inactive Epac1 with a mutation rendering highaffinity cAMP binding (82). It exhibits a loss in FRET signal upon binding cAMP. Note that the Jalink lab does not recommend this sensor for general experimentation due to punctate speckles forming a few days after transfection. This was confirmed in our hands but did not hamper our experiments, which were performed 2 days after transfection. Cells exhibited a relatively even cytosolic distribution of the sensor in the experiments shown here (Fig. S3A). For further details on the construct, please contact the Jalink laboratory. NLS-H208 was constructed using PCR to subclone an NLS motif to the N terminus by using HindIII restriction enzymes.

Cell culture
HC1 and HEK293 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (FBS), penicillin (100 IU/liter), and streptomycin (100 mg/liter). PCCL3 cells were cultured as described previously (83). Stable cell lines were generated by lentiviral infection with a multiplicity of infection of 80 and 5 g/ml Polybrene for 24 h and selected with puromycin. Transiently transfected cells were treated 24 h after plating with X-tremeGENE TM HP ACCELERATED COMMUNICATION: bPAC-nLuc regulates cAMP dynamics (Roche Applied Science) for HC1 cells or Lipofectamine 3000 (ThermoFisher Scientific) for PCCL3 and HEK293 cells.

Luminescence assays
3 ϫ 10 4 HEK293 were seeded in opaque 96-well dishes (Corning) coated with poly-D-lysine. 24 h after transfection, cells were washed with PBS, incubated with OptiMEM lacking phenol red, and stimulated with h-CTZ (NanoLight Technology) or NanoGlo luciferase substrate (furimazine, Promega). The concentrations of Fz and its analogs are not provided; thus, we report treatments as dilutions from the stock solution. Luminescence was quantified using a Tecan Spark 20 M plate reader (SparkControl version 1.2) or SpectraMax Paradigm plate reader (SoftMax Pro 6.2.2) to measure specific wavelengths.

ELISA
1.5 ϫ 10 5 HC1 cells were seeded in a 6-well dish, transfected, and stimulated for 10 min with 100 M IBMX and h-CTZ. Cells were lysed and analyzed following the monoclonal anti-cAMP antibody-based direct ELISA kit (Neweast Biosciences) instructions and normalized by protein concentration using the Pierce TM BCA protein assay kit.

Optogenetic stimulation
Cells were kept in a dark environment using a red safelight lamp (Kodak GBX-2 Safelight Filter) with a 13-W amber compact fluorescence bulb (Low Blue Lights, Photonic Developments LLC) to prevent light exposure from wavelengths Ͻ500 nm. Light activation was achieved using a custom-built, Arduino-controlled system capable of regulating the duration, frequency, and intensity of light exposure. The illuminating high-power LED (royal blue CREE XTE Tri-Star LED, LED Supply) was mounted on a stage (Fig. S5). The intensity range spanned from 4.41 Ϯ 0.30 to 14.85 Ϯ 1.00 W/mm 2 as measured by a laser power meter (ThorLabs PM1100D, detector S130C) (Fig. 3B). For more details on the system, see Figs. S5-S8 and Table S1.

Real-time imaging
1.5 ϫ 10 6 PCCL3 cells seeded on 0.1% gelatin-coated 25-mm glass coverslips were transfected with a red dimerization-dependent sensor (Montana Molecular), H208, or NLS-H208. Cells were given fresh media for 24 h and hormone-starved for 3 h in Starvation Coon's media (lacking TSH, insulin, and hydrocortisone) containing 5% FBS. Cells were washed in PBS and imaged in OptiMEM lacking phenol red on an Olympus IX70 microscope equipped with a Till Polychrome V monochromator. Images were acquired every 10 s with a ϫ60/1.4 NA oil objective, 8 ϫ 8 binning, and a Hamamatsu CCD camera (Photonics Model C4742-80-12AG) using Slidebook software (Intelligent Imaging Innovations Inc.). The red cAMP dimerization-dependent sensor was excited at 570 nm (20% intensity, 15 nm bandpass) with emission filter FF01-620/52 (Semrock) and 560-nm long-pass dichroic filter. Cells were stimulated with 4.4 W/mm 2 LED (450 nm) light and 100 M IBMX at the end. H208 and NLS-208 were excited at 500 nm with emission filters 510/20 and 620/52 (Semrock) and a dichroic beam splitter 468/526/596 (BrightLine Semrock). Cells were stimulated with 1:3000 Fz or 1 unit/liter TSH and 100 M IBMX at the end. To confirm bPAC-nLuc expression at the end, a 10-s exposure at 480 nm was taken after 1:1000 Fz stimulation. The ratio of Scarlet/YFP (donor over acceptor) fluorescence was normalized to maximum FRET response as determined by IBMX stimulation.
HEK293 cells seeded on poly-D-lysinecoated glass coverslips were transfected using a low amount of MYR-bPAC-nLuc DNA (100 ng) to avoid sensor saturation. Cells were washed and imaged 48 h later in a 10 mM HEPES imaging buffer with 137 mM NaCl, 5.4 mM KCl, 2 mM CaCl 2 , 1 mM MgCl 2 , pH 7.3. Images were acquired every 30 s with a ϫ60/1.4 NA oil objective, 4 ϫ 4 binning. Cells were stimulated with 4.4 W/mm 2 LED light or 1:1000 Fz and normalized to maximum FRET response using 10 M forskolin and 100 M IBMX.

BrdU and EdU incorporation assays
BrdU incorporation in transfected cells was performed as described previously (66) but using 5% FBS for starvation and stimulation. Cells were kept in darkness or stimulated with 4.4 W/mm 2 450 nm light for 500 ms every 15 min over 24 h. Images were acquired using a ϫ20/0.70 NA HC PLAN APO objective on a Leica DM500B microscope and analyzed in ImageJ, counting % BrdU-positive nuclei/total DAPI-stained nuclei. For Edu incorporation, PCCL3 stable cell lines expressing pCDH-bPAC-nLuc constructs were seeded 2 ϫ 10 4 cells/ well in a 96-well dish (PerkinElmer Life Sciences, black-wall/ clear bottom). 24 h later, cells were starved (Starvation Coon's media, 5% FBS) for 16 h and stimulated with Nano-Glo Endurazine Live Cell Substrate prototype (Fz-4377) or DMSO. 8 h after stimulation, 10 M EdU was added and incubated an additional 16 h. Plates were imaged using a Nikon Eclipse Ti with a ϫ10 objective and with help from Dr. Watkins (Center for Biologic Imaging, University of Pittsburgh). Images were analyzed as % EdU-positive nuclei/total DAPI-stained nuclei using NIS Elements (Nikon Instruments). Immunofluorescent staining is included in the Supporting Methods.