Robust and sensitive GFP-based cGMP sensor for real time imaging in intact Caenorhabditis elegans

cGMP is a ubiquitous second messenger that plays a role in sensory signaling and plasticity through its regulation of ion channels and kinases. Previous studies that primarily used genetic and biochemical tools suggest that cGMP is spatiotemporally regulated in multiple sensory modalities, including light, heat, gases, salt and odor. FRET- and GFP-based cGMP sensors were developed to visualize cGMP in primary cell culture and Caenorhabditis elegans to corroborate these findings. While a FRET-based sensor has been used in an intact animal to visualize cGMP, the requirement of a multiple emission system limits its ability to be used on its own as well as with other sensors and fluorescent markers. Here, we demonstrate that WincG2, a codon-optimized version of the cpEGFP-based cGMP sensor FlincG3, can be used in C. elegans to visualize rapidly changing cGMP levels in living, behaving animals using a single fluorophore. We coexpressed the sensor with the blue light-activated guanylyl cyclases BeCyclOp and bPGC in body wall muscles and found that the rate of WincG2 fluorescence correlated with the rate of cGMP production by each cyclase. Furthermore, we show that WincG2 responds linearly upon NaCl concentration changes and SDS presentation in the cell bodies of the gustatory neuron ASER and the nociceptive phasmid neuron PHB, respectively. Intriguingly, WincG2 fluorescence in the ASER cell body decreased in response to a NaCl concentration downstep and either stopped decreasing or increased in response to a NaCl concentration upstep, which is opposite in sign to previously published calcium recordings. These results illustrate that WincG2 can be used to report rapidly changing cGMP levels in an intact animal and that the reporter can potentially reveal unexpected spatiotemporal landscapes of cGMP in response to stimuli. Author Summary cGMP is a second messenger that plays an important role in sensory signaling and neural plasticity. Previous genetic and biochemical studies indirectly suggest that cGMP is spatiotemporally regulated in neurons to modulate neural activity. While a FRET-based sensor for cGMP has been used in intact Caenorhabditis elegans to examine its spatiotemporal regulation in neurobiological processes, its use has been limited due to the complicated setup required to image this type of sensor. Here, we describe a GFP-based cGMP sensor that has been codon optimized for use in C. elegans and demonstrate that it responds robustly and reliably to endogenously changing cGMP levels. We show that the sensor responds to cGMP production by coexpressing it with blue light-activated guanylyl cyclases, and we show that it responds to NaCl and sodium dodecyl sulfate when expressed in a gustatory and nociceptive neuron, respectively. We think that this sensor can be used to investigate the spatiotemporal regulation of cGMP in neurons and its relationship to neural activity.

elegans' sensitivity to quinine through the flow of cGMP from sensory neurons to the 116 nociceptive neuron ASH through gap junctions [17,18], a visual tool could greatly enhance 117 our understanding of the process and its dynamics. In another nociceptive neuron, PHB, 118 genetic evidence suggests that the G protein coupled receptor srb-6 is required to sense 119 noxious liquids including sodium dodecyl sulfate (SDS) and dodecanoic acid [19], and 120 calcium recordings suggest that avoidance of isoamyl alcohol is at least partially mediated 121 by the cGMP-gated cation channel TAX-2/TAX-4, both suggesting that cGMP flux is 122 required for sensing some nociceptive cues [20]. Thus, genetic tools demonstrate the 123 importance of cGMP signaling in distinct sensory modalities, and a tool to visualize cGMP 124 fluxes with precise temporal and spatial fidelity could deepen our understanding of these 125 important processes, offering a more complete picture of the cGMP landscape and 126 dynamics in cells. Such a tool will provide an essential complementary approach to the 127 primarily genetic approaches that have been used to examine cGMP dynamics in this 128 be used to visualize the cGMP landscape in neurons, the complexity of a multiple 139 emission system might raise a barrier to its use. Thus, having a robust, single fluorophore 140 sensor for cGMP will complement the use of calcium sensors in this animal, providing a 141 way to investigate how the spatiotemporal regulation of cGMP influences neural activity 142 in vivo. Such a tool would be maximally efficient for probing the interplay between cGMP 143 and calcium dynamics within sensory compartments in this transparent organism. 144 Here, we show that WincG2 (Worm Indicator of cGMP -2), the C. elegans codon-145 optimized version of the GFP-based circularly permuted cGMP sensor FlincG3, reports 146 cGMP dynamics in vivo in C. elegans [23]. We characterize the biochemical and 147 biophysical properties of WincG2 upon cGMP binding in vivo by ectopically expressing 148 the blue light-activated guanylyl cyclases BeCyclOp and bPGC in muscle cells [24]. 149 Using the WincG2 reporter, we show for the first time that sensory stimulation of either 150 gustatory or nociceptive neurons triggers cGMP changes. We demonstrate that cGMP 151 reliably decreases in response to NaCl concentration downsteps and increases in 152 WincG2 is a codon-optimized circularly permuted cGMP sensor derived from the 161 mammalian sensor, FlincG3 162 WincG2, a genetically-encoded, circularly permuted EGFP-based cGMP sensor, is the 163 C. elegans codon-optimized version of FlincG3, which was initially characterized both in 164 vitro and in cell lines [23]. FlincG3 is based on FlincG, an earlier version of the sensor. to a 230-fold lower concentration of cGMP than cAMP, suggesting that it preferentially 179 binds to cGMP [23]. For our study we codon optimized FlincG3 for C. elegans and 180 inserted it into a standard C. elegans expression vector (Fig 1).

Stimulation of blue light-activated guanylyl cyclases increases WincG2 183 fluorescence 184
To test whether WincG2 can detect rapid changes in cGMP levels in an intact 185 animal, we utilized the C. elegans body wall muscle cells, which lack most endogenous 186 GCs. We coexpressed the reporter along with heterologous light-inducible GCs that have 187 different cGMP production rates [24,28]. BeCyclOp is a microbial rhodopsin from 188 Blastocladiella emersonii that is linked to a cytosolic GC domain (Fig 2A) BeCyclOp. This was followed by a slight decay over the duration of the recording (Fig 2B:  201 top green trace). By contrast, animals grown without ATR and thus having no BeCyclOp 202 activation exhibited an apparent decrease in WincG2 fluorescence (F-F0/F0 plateaued at 203 approximately -0.310 to -0.312 beginning at 8.98s) when exposed to blue light (Fig 2B:  204 bottom blue trace). The initial signal decayed; F-F0 became negative, then plateaued and 205 remained steady for the duration of the recording. Notably, the initial fluorescence 206 intensity F0 (as measured in the absence of the rhodopsin cofactor ATR; bottom blue trace 207 in Fig 2B)  WincG2 response to BeCyclOp, albeit with slightly slower kinetics ( Fig 2F). Thus, WincG2 237 appears to respond to cAMP. Indeed, bPAC is an efficient adenylyl cyclase that produces 238 cAMP at the rate of 10 ± 2 nmol per minute per mg. Thus, it is not surprising that WincG2 239 responds to the high production of cAMP by bPAC [31]. Since there are no amino acid 240 changes between WincG2 and FlincG3, it is expected that WincG2 is also activated more 241 effectively by cGMP relative to cAMP. However, these results indicate that it is important 242 [25]. Additionally, a study indicating that cGMP could be a putative second messenger in 252 ASER revealed that loss of the rGC GCY-22 blunts chemotaxis to Cl - [4]. This suggests 253 that cGMP levels could be modulated by changes in NaCl concentration [3]. 254 To explore this hypothesis, we expressed WincG2 in the ASE neuron pair and 255 monitored the sensor's response in the ASER cell body to ten 10-second steps between 256 50 and 0 mM NaCl. concentration steps and not to fluctuations in fluid pressure on the exposed nose of the 271 animal ( Fig 3B). 272 WincG2 has a cGMP-binding motif that could also potentially accommodate cAMP, C. elegans requires ASER activity to adjust their preferred NaCl concentration to 290 the concentration at which they were last fed; if ASER is killed, the animal's movement is 291 less directed in response to a linear NaCl gradient [26]. Plasticity requires NaCl sensation, 292 which in turn requires cGMP signaling; thus it is not surprising that gcy-22(tm2364) 293 animals which we showed to not respond to NaCl concentration changes (Fig 3) do not 294 exhibit a preference for the concentration of NaCl at which they were cultivated [16]. To 295 assess whether WincG2 expression affected an animal's ability to exhibit a preference for were compared to their nontransgenic siblings that did not express the WincG2 array and 298 wild-type animals. Animals were cultivated for approximately six hours in the presence of 299 OP50 E. coli bacteria on an NGM plate containing 25 mM, 50 mM, or 100 mM NaCl, then 300 placed onto a chemotaxis assay plate containing a NaCl gradient from approximately 40 301 to 90 mM NaCl ( concentration preference at each cultivation NaCl concentration was higher, though not 308 significantly different from wild-type animals ( Fig 4B: third set of three bars); however, 309 their preference for a higher NaCl concentration was significantly different from their 310 nontransgenic siblings when they were cultivated at 100mM NaCl (p<0.05). Additionally, 311 the animals' preference for a higher NaCl concentration seemed different from their 312 nontransgenic siblings when they were cultivated at 50mM, though this was not significant 313 (p=0.15), presumably due to variability of the transgenic animals' chemotaxis to NaCl 314 when cultivated at 50mM. Other lines injected at the same concentration exhibited NaCl 315 seeking behavior that was significantly different from both wild-type animals and their 316 nontransgenic siblings (S3 Fig). This may indicate that WincG2 expression lowers free 317 cGMP levels and therefore interferes with an aspect of cGMP dynamics in ASER that is 318 required for food to reset the animals' preference to their cultivation NaCl concentration.
WincG2 fluorescence was recorded for the line that exhibited behavior that was 320 not significantly different from wild-type animals. Importantly, in these animals, WincG2 To examine the ability of WincG2 to report cGMP changes in a neuron with a 337 different modality, we expressed WincG2 in the nociceptive PHB neurons, which had 338 been predicted to use cGMP as a second messenger. The PHBs are a pair of bilaterally 339 symmetric sensory neurons located in the lumbar ganglia that extend ciliated dendrites 340 into the phasmid structures within the tail of C. elegans. PHB neurons are chemosensory 341 cells that are required for the avoidance of noxious chemicals such as SDS, dodecanoic acid and other cues ( Fig 5A) [19, 20,32,33]. TAX-4 is required for PHB-mediated SDS 343 avoidance (Fig 5C), and calcium imaging has shown that PHB responds to SDS [20]. This 344 suggests that PHB may exhibit changes in cGMP levels in response to SDS that could 345 be monitored by recording changes in WincG2 fluorescence. 346 To test whether WincG2 affects the function of the PHB circuit, SDS response 347 assays ( Fig 5B) were performed on wild-type animals and animals expressing WincG2 in 348 PHB neurons. On average, wild-type animals halt movement into a drop of 1 mM SDS in 349 less than a second. If PHB function is impaired, as in tax-4 mutants, animals continue 350 moving into a drop of SDS as if it were a control buffer (M13). This increases the relative 351 response index to approximately 300% ( Fig 5C). We found that behavioral responses to 352 SDS were unaffected by WincG2 expression, indicating that WincG2 does not affect PHB 353 function ( Fig 5C). 354 To determine if cGMP changes in PHB neurons could be detected using WincG2, 355 the sensor's fluorescence in the cell body was measured in animals that were first 356 exposed to control buffer (M13), then to 1 mM SDS in M13 buffer. WincG2 fluorescence 357 remained largely steady in the absence of 1 mM SDS, but began to increase linearly 358 (R 2 =0.9633) upon exposure to 1 mM SDS (Fig 5D). The area of the curve for the traces 359 before and after SDS presentation is different (permutation test, p<0.00001), which 360 suggests that cGMP increases in response to SDS and that WincG2 responds acutely to 361 endogenously produced cGMP that is induced by an external stimulus (Fig 5E). 362 The cGMP sensor WincG2 was successfully used to monitor the dynamics of this 366 second messenger in a number of cells in C. elegans. First, WincG2 was used to monitor 367 the kinetics of cGMP production in body wall muscles that lack most endogenous GCs. 368 The rate of increase in WincG2 fluorescence corresponded with the rate of cGMP 369 produced by coexpressed blue light-activated GCs. WincG2 fluorescence increased 370 within less than a second of activation of BeCyclOp, which produces 17 cGMP molecules 371 per second. In contrast, WincG2 fluorescence increased in the order of minutes upon 372 activation of bPGC, which produces cGMP at a 50-fold lower rate relative to BeCyclOp 373 [24]. WincG2 fluorescence slightly increased in the presence of cAMP in C. elegans in 374 response to activation of bPAC. Thus, care must be taken to control for fluctuations in 375 cAMP by imaging in backgrounds that lack cGMP production. The high rate of cAMP 376 production due to bPAC, however, is likely exceeding any intrinsic cAMP production by 377 several fold, thus side effects from intrinsic cAMP fluctuation may affect cGMP imaging 378 only to a minor extent. 379 380 WincG2 reveals cGMP dynamics in sensory neurons that use cGMP as a second 381 messenger for sensory stimuli 382 We expressed WincG2 in sensory neurons that use cGMP as a second messenger 383 and found that the sensor responds robustly to changing stimulus presentation. In the 384 gustatory neuron ASER, changes in WincG2 fluorescence in response to repeated NaCl 385 concentration steps suggest that the sensor can respond reliably to changing cGMP 386 levels, providing, for the first time, visual evidence that cGMP levels in ASER are 387 modulated by NaCl concentration changes. Importantly, we demonstrated that WincG2 fluorescence changes require the rGC GCY-22. This has two implications: 1) the sensor 389 responds to cGMP in a physiologically relevant setting and 2) GCY-22 seems to be the 390 primary rGC that produces cGMP for the NaCl response. Interestingly, the changes we 391 observe in WincG2 fluorescence in ASER suggest that cGMP levels may be inversely The unexpected result that WincG2 fluorescence decreases in response to NaCl 400 downsteps could be due to the tight regulation of cGMP levels by phosphodiesterases. 401 Resolution of this apparent paradox, however, awaits further investigation. 402 In the nociceptive phasmid neuron PHB, genetic evidence suggesting that it uses 403 cGMP to signal the presence of SDS was corroborated by changes in WincG2 404 fluorescence. This is the first visual evidence for a cGMP-based signal in PHB, showing 405 that it increases in response to an environmental cue. Importantly, expression of WincG2 406 did not perturb the function of the phasmid neurons, as SDS repulsion was as robust in 407 the WincG2-expressing line as in wild-type animals. This is in contrast to WincG2 408 expression in ASER, which caused a slight preference for higher NaCl concentrations but 409 minimally affected the plasticity of the NaCl concentration cultivation preference. it will be necessary to select for lines that express WincG2 at the lowest levels that allow 424 for imaging thereby minimizing the potential for behavioral effects. Addition of a 425 subcellular localization signal may also mitigate off-target effects. 426 We think WincG2 could be acting as a cGMP sponge due to its effects on NaCl-427 seeking behavior when expressed in ASE (Fig 5B). These behavioral results suggest that 428 WincG2 could be altering free cGMP levels in ASER, which could lead to tuning the NaCl 429 concentration cultivation preference to be higher relative to nontransgenic siblings and 430 wild-type animals. If WincG2 can be shown to act as a cGMP sponge, this could also be 431 exploited to specifically and locally perturb cGMP levels. For example, if one could 432 localize a non-fluorescent form of WincG2 at the cilia, this may perturb function in a different way from when it is localized to the cell body. This would reveal specific functions 434 for cGMP signals at the sensory dendrites that are different from those in the cell body. 435 Additionally, the subcellular landscape of cGMP can also be probed using WincG2. 436 For instance, adding a small tag that localizes WincG2 to specific regions of the cell along 437 with a red protein for ratiometric imaging may reveal important aspects of the subcellular 438 landscape of cGMP.

Imaging WincG2 in ASER and PHB 565
WincG2 imaging was performed essentially as described for calcium imaging [17]. 566 Briefly, for imaging ASER, day 1 adults grown at 20C were transferred from NGM plates 567 containing OP50 to a 35 mm x 10 mm petri dish containing chemotaxis buffer with 50 mM  2 mOsm with sorbitol). The worms were then placed in a microfluidic device that can 570 expose the animal to stimulus [45]. A Zeiss 40x air objective on an inverted microscope 571 (Zeiss Axiovert 200) was used for imaging, and images were taken at rate of 1Hz with a 572 blue light exposure time of 30 ms using an ORCA-Flash 2.8 camera (Hamamatsu) for a 573 total of 103 frames. Recordings were taken within eight minutes of the animal's exposure 574 to chemotaxis buffer with 50 mM NaCl, and the animals were subjected to either ten 575 second steps between chemotaxis buffer with 50 and 0 mM NaCl (each containing 1 mM 576 levamisole) or switches between chemotaxis buffer with 50 mM NaCl (each containing 1 577 mM levamisole (Sigma-Aldrich), and one containing fluorescein). Images were obtained 578 using µManager (Version 1.4.22). Fluorescence intensity was measured using ImageJ. 579 To calculate the fluorescence intensity at a given time point (F), the fluorescence intensity 580 from the region of interest (ROI) encompassing the neuron was subtracted from the 581 background ROI. The fluorescence intensity F of the first three frames was averaged to 582 calculate F0. We used ∆F/F0 (F-F0/F0) to calculate the change in fluorescence intensity 583 at a given time point. 584 To image WincG2 in PHB neurons, animals were picked from NGM plates 585 containing OP50 onto a petri dish containing M13 control buffer, then placed tail-first into 586 the microfluidic device. They were exposed to control buffer for 15 seconds, and then to 587 1 mM SDS in M13 for 15 seconds. They were imaged at a rate of 2 Hz. To calculate the 588 fluorescence intensity of PHB at each frame, Image J was used to measure the total intensity 589 of the cell body. Background fluorescence was calculated by using ImageJ to measure the 590 minimum pixel value in the area surrounding the cell body, and this pixel value was multiplied 591 by the area of the cell body to get the total background. The total background was subtracted 592 from the total intensity of the cell body to calculate the fluorescence intensity. The PHB 593 WincG2 fluorescence intensities were adjusted for photobleaching using the following 594 method. The decrease in fluorescence during the first 29 frames when the animal was 595 exposed to control buffer was presumed to be due to photobleaching. Therefore, the average 596 difference between the values for the nth frame and the n+1th frame (up to the 29 th frame) 597 were calculated, and this average photobleaching value was then added back to each value 598 in the series. F0 was the average of the response to buffer adjusted for photobleaching 599 over the first 15 seconds, rather than only the first 3 data points. 600 601

NaCl cultivation assay 602
The NaCl cultivation assay was essentially performed as described [16]. Briefly, 603 day 1 animals grown at 20C were transferred from OP50-containing NGM plates 604 containing 50 mM NaCl to OP50-containing NGM plates containing 25 mM, 50 mM, or 605 100 mM NaCl for approximately six hours before being placed on a chemotaxis assay 606 plate containing regions of higher and lower NaCl for 45 minutes. Afterward, worms were 607 stored in 4C for at least 16 hours before calculating the chemotaxis index. Chemotaxis 608 index = [# animals at higher NaCl region -# animals at lower NaCl region]/ [# animals at 609 higher NaCl region + # animals at lower NaCl region + # animals outside origin]. The muscle contraction assay was essentially performed as described [24]. L4 628 animals were exposed to 0.9 mW/mm 2   GFP halves. This linker, along with the 6xHis-tag region (H6) and the Tag Region, were 834 retained from FlincG3. This C. elegans codon-optomized sensor, prepared by Genscript, 835 was inserted into a worm-specific Fire vector, pPD95.75, which contains synthetic introns