Functional basis for calmodulation of the TRPV5 calcium channel

Within the transient receptor potential (TRP) superfamily of ion channels, TRPV5 is a highly Ca2+-selective channel important for active reabsorption of Ca2+ in the kidney. Its channel activity is controlled by a negative feedback mechanism involving calmodulin (CaM) binding. Combining advanced microscopy techniques and biochemical assays, this study characterized the dynamic bilobal CaM regulation and binding stoichiometry. We demonstrate for the first time that functional (full-length) TRPV5 interacts with CaM in the absence of Ca2+, and this interaction is intensified at increasing Ca2+ concentrations sensed by the CaM C-lobe that achieves channel pore blocking. Channel inactivation occurs without CaM N-lobe calcification. Moreover, we reveal a 1:2 stoichiometry of TRPV5:CaM binding by implementing single molecule photobleaching counting, a technique with great potential for studying TRP channel regulation. In conclusion, our study proposes a new model for CaM- dependent regulation – calmodulation – of the Ca2+-selective TRPV5 that involves apoCaM interaction and lobe-specific actions.


Introduction 51
Transient receptor potential (TRP) channels are one of the largest classes of ion channels 52 and are widely expressed throughout the animal kingdom (1). Since their discovery, they 53 have emerged as key players in human physiology and were found to be associated with 54 various diseases, such as cancer, skeletal abnormalities, skin disorders, and chronic pain (2, 55 10 observed between the CaM C-lobe and the distal TRPV5 peptide. Notably, there was also 217 weak but consistent interaction of the N-lobe with both peptides ( Figure 3D). Next, we 218 addressed the potential independent effect on TRPV5 function. Fura-2 Ca 2+ imaging 219 demonstrated that co-expression of TRPV5 with either CaM N-or C-lobe increases the 220 intracellular Ca 2+ peak compared to TRPV5 with endogenous or overexpressed wildtype 221 CaM (Figure 3E/F). This indicates that, while both lobes can independently interact with 222 TRPV5, full length CaM is needed for TRPV5 inhibition. 223 We further explored the interaction of each CaM lobe in Ca 2+ -free conditions using FLIM-224 FRET analysis of PMLs from cells co-expressing TRPV5 and CaM N-or C-lobe. While CaM and 225 apoCaM interact with TRPV5, the N-and C-lobe interact in Ca 2+ -containing (2 mM), but not 226 in Ca 2+ -free (2 mM EGTA, 2 mM EDTA) conditions (Figure 3G-J). Summing up, the C-lobe is 227 the major entity for sensing Ca 2+ changes and closing TRPV5. The N-lobe (and the linker) act 228 as a support to the structural rearrangement required for the C-lobe to execute its function. 229 230

Stoichiometry of CaM binding to TRPV5 231
In addition to the dynamics of TRPV5-CaM binding, the stoichiometry of the TRPV5-CaM 232 complex is still under debate. Recent structural work by our group provided evidence for 233 binding of either 1 or 2 CaM molecules (16) to the tetrameric channel, while other groups 234 have described a 1:1 stoichiometry for TRPV5:CaM as well as for TRPV6:CaM (9, 10, 17,18). 235 To this end, we implemented a method to reveal the in vivo stoichiometry of TRPV5-CaM, 236 which is single molecule photobleaching counting (smPB) (24). Following the procedure 237 described in Methods, single molecule complexes at the cell surface were imaged using a 238 TIRF microscope for long acquisition lapses. To optimize the conditions of this technique for 239 our study, we started with HEK293 cells only expressing eGFP-TRPV5 and observed 240 individual molecules at the TIRF plane ( Figure 4A). Analysis of their traces was performed by 241 filtering them with a tailored Chung-Kennedy Filter Plugin as described previously (25). 242 Some obtained traces matched the expected number of 4 bleaching steps for a tetramer like 243 TRPV5 (Figure 4B), while there were also molecules with bleaching steps ranging from 1 to 244 3. By fitting the observed values to a binomial distribution, we could infer an eGFP folding 245 probability of 72% ( Figure 4C). Next, HEK293 cells co-expressing mCherry-TRPV5 and eGFP-246 CaM were subjected to smPB to analyse the stoichiometry of binding. Of note, a slightly 247 different distribution was found for mCherry-TRPV5 compared to eGFP, since the probability 248 of folding was lower (p=0.62, data not shown). In order to assess stoichiometry of binding, 249 low mobile mCherry-TRPV5 spots were chosen that were also positive for eGFP ( Figure 4D). 250 These were analysed in the presence and absence of extracellular Ca 2+ . The latter condition 251 was taken along to reduce potential changes in intracellular Ca 2+ levels as a result of Ca 2+ 252 influx. CaM analysis revealed mainly two type of traces: one or two bleaching steps ( Figure  253 4E-F). A highly reduced amount of three steps traces could be found (2% and 5% for 254 presence and absence of Ca 2+ , respectively). Fitting our results to a binomial distribution 255 with eGFP folding probability of 72% showed that the distributions fit with a preferential 256 stoichiometry of 2 CaM molecules per TRPV5 tetramer, either in the presence or absence of 257 extracellular Ca 2+ (Figure 4G-H). 258 259

Discussion 260
The present study sheds new light on the dynamic interaction between TRPV5 and CaM and 261 its stoichiometry. It shows a constitutive weak interaction of TRPV5 with apoCaM, and a 262 solid interaction with CaM C-lobe in the presence of 10nM of Ca 2+ , mimicking basal 263 physiological conditions. Importantly, our data provided evidence that a missing CaM N-lobe 264 impacts mildly on the TRPV5 inhibition, and most likely positions the CaM C-lobe. we observed a significant TRPV5-CaM interaction that is likely engaged by a calcified C-lobe, 302 while the N-lobe will remains empty (20). Concomitant with our results, Bokhovchuk and 303 colleagues also described a tight interaction between the CaM C-lobe and the TRPV5 C-304 terminus at basal Ca 2+ concentrations (10-100nM) (10). Specifically, we demonstrated that 305 the full-length CaM interaction mainly relies on C-lobe binding to the distal TRPV5 C-306 terminal helix, as no competition on TRPV5-CaM interaction was observed by the proximal 307 TRPV5 C-terminus peptide. Of note, these distal and proximal peptides are based on helices 308 defined in the cryo-EM TRPV5-CaM structures that were shown to interact with C-lobe and 309 N-lobe, respectively (16, 17). Together, this would be in line with the previously proposed 310 model of CaM-dependent inhibition of TRPV5 (and TRPV6) involving the N-lobe as Ca 2+ 311 sensor that merely triggers channel closure upon increasing Ca 2+ concentrations (9, 10, 16, 312 14 18, 39). However, our findings suggest that the C-lobe acts as both the sensing and 313 executing unit of CaM. Mutation on the N-lobe (CaM12) did result in CaM-dependent 314 channel inactivation, while an opposite effect was seen for the CaM34 and CaM1234 315 mutants. The relevance of the C-lobe modulating ion channel activity has been described for 316 other channels, such as of the Kv7 family (Kv7.1, Kv7.4 and Kv7.5) (36). 317

318
In addition to these CaM lobe-specific effects, it has been also described for several ion 319 channel families that interaction with both lobes is required to trigger complete Ca 2+ -320 dependent inhibition (15,40). This is likely due to a structural rearrangement of the full Next to the lobe-specific effects, this study sheds new light on the stoichiometry of binding. 333 We implemented the smPB technique to decipher the amount of CaM molecules per TRPV5 334 tetramer (41), and identified a preferable 1:2 TRPV5-CaM composition. So far, there has 335 been debate about the stoichiometry. Initial NMR studies with peptides of TRPV5 C-336 terminus suggested a 1:2 stoichiometry for CaM:TRPV5 696-729 (12). Later studies of the same 337 group, using longer TRPV5 C-terminus peptides, confirmed these results (10). Yet, recent 338 cryo-EM complex structures of TRPV5 with CaM suggest on one hand a 1:1 stoichiometry of 339 1 tetrameric TRPV5 channel binding to 1 CaM molecule (17), and on the other hand provide 340 evidence for a variable stoichiometry of either 1:1 or 1:2 (2 CaM molecules per TRPV5 341 tetramer) (16). Through the smPB method, we were the first to reveal in intact cells that 342 TRPV5 preferably binds 2 CaM molecules. 343 Since some ion channels undergo changes in stoichiometry depending on the Ca 2+ 344 concentration, experiments were performed in the presence and absence of Ca 2+ (42). For 345 example, the stoichiometry of L-type voltage-gated Ca 2+ channels was found to be 1:1 in the 346 absence of intracellular Ca 2+ , but increased to two CaM peptides at higher Ca 2+ 347 concentrations (43). We observed the same stoichiometry in the presence and absence of 348 extracellular Ca 2+ , suggesting that basal interaction with CaM is not dependent on Ca 2+ 349 influx. Analyzing our cryo-EM TRPV5-CaM complex structure (16), it is clear that only 1 CaM 350 molecule can occupy the lower cavity of the tetrameric channel pore. Therefore, we would 351 speculate that the CaM-mediated TRPV5 inactivation mimics the ball-and-chain mechanism 352 observed between Kv channels and Kv subunits (44). 353

354
In conclusion, our study demonstrates that the CaM-mediated TRPV5 channel inactivation is 355 a dynamic process involving 2 CaM molecules in close proximity to the channel pore that 356 exhibit lobe-specific regulation. The results established a weak, but persistent, interaction 357 between TRPV5 and apoCaM that may assist fast channel inhibition through calcification of concentration. The peak response is calculated as the difference in ratio upon Fura-2 Ca 2+ 539 buffer addition versus basal levels in Fura-2 EDTA buffer (t Ca -t EDTA ). All measurements were 540 performed at room temperature. 541 542

Statistical analysis 543
The immunoblot data were analyzed by comparing integrated optical densities of bands 544 using Fiji (49). The semi-quantification is shown as mean ± SEM and plotted against the log 545 free Ca 2+ concentration ( Figure 1E