In order to evaluate how CMK-1 subcellular localization is regulated prior to and during adaptation, we focused on the FLP thermonociceptor neurons, in which a robust change in localization is easily monitored with a CMK-1::mNeonGreen (mNG) protein fusion (Schild et al., 2014; Hostettler et al., 2017). In naïve animals (grown at 20°C), CMK-1 is mostly localized in the cytoplasm of FLP (Figure 1B). Exposure to moderately noxious temperature (28°C) causes CMK-1 to progressively translocate into the nucleus, being slightly enriched in the nucleus after 90 min (Figure 1B and C). These observations recapitulate previous findings obtained with different fluorescent tags (Schild et al., 2014; Hostettler et al., 2017).

Since CMK-1 subcellular localization varies according to cell types and conditions, and since putative NLS and NES are predicted from its primary sequence (Schild et al., 2014), CMK-1 localization may result from a dynamic equilibrium between nuclear import and export drives via canonical export/import pathways (Sorokin et al., 2007). To test this general working model, we started by addressing the mechanisms that promote the cytosolic accumulation of CMK-1 in FLP at 20°C. Two candidates NES were identified thanks to the NESSential prediction tool (Fu et al., 2013): NES^288-294 and NES^315-323 (superscript indexes indicate the residue positions, Figure 1A). We expressed CMK-1 mutant proteins affecting each of these NES candidates and examined the impact on CMK-1 localization in FLP at 20°C. Disrupting the NES^288-294 candidate sequence with a V292A/V294A double mutation or a Δ288–294 deletion prevented the cytoplasmic expression of CMK-1 and even yielded a nuclear accumulation (Figure 1D). In contrast, NES^315-323 candidate mutants displayed a cytoplasmic enrichment similar to that of wild-type CMK-1. These included I315A single mutant, V321A/V323A double mutant, I315A/V321A/V323A triple mutant, as well as Δ318–324 and Δ315–323 mutants, in which the NES candidate region is deleted in part or in full (Figure 1D). Collectively, these data indicate that the NES^288-294, but not the NES^314-323, is a key determinant of CMK-1 cytoplasmic localization in FLP at 20°C.

Because NES^288-294 sequence matches the binding consensus for exportins, we further tested whether CMK-1 is actively transported out of the nucleus at 20°C in FLP via this canonical pathway. Inhibiting exportins with the leptomycin B pharmacological inhibitor (Kudo et al., 1999) caused CMK-1::mNG nuclear translocation in FLP neurons at 20°C (Figure 1E).

Collectively, these results suggest that the predominant cytoplasmic expression of CMK-1 at 20°C in FLP requires NES^288-294 and relies on an exportin-dependent active nuclear export mechanism.

Our next goal was to clarify the mechanisms causing CMK-1 to accumulate in the nucleus upon exposure to moderately noxious temperatures (28°C, 90 min). Hypothesizing that the importin pathway could be involved, we started by examining the role of two candidate NLS, defined with an in silico prediction tool (Nguyen Ba et al., 2009): NLS^71-78 and NLS^297-308. Mutations affecting the NLS^71-78 candidate region all significantly impaired the nuclear relocalization of CMK-1 after prolonged heat exposure (Figure 1F, right). The inhibitory effect was partial in CMK-1(K71A) and CMK-1(R74Q) single mutants and complete in the CMK-1(R77S) single mutant, as well as in the CMK-1(K71A/R74Q/R77S) triple mutant. Introducing mutations in the NLS^297-307 candidate region (K307Q single mutant or a R302S/K307Q double mutant) also produced a partial reduction in CMK-1 nuclear accumulation. In contrast, none of the NLS-affecting mutations modulated the baseline cytoplasmic expression at 20°C (Figure 1F, left), suggesting that these elements might not be active in the absence of prolonged cell stimulation.

Taken together, our results suggest that (i) the canonical importin pathway may be involved, (ii) NLS^71-78 and NLS^297-308 are only engaged upon prolonged thermal stimulation, and (iii) NLS^71-78 plays a major role and NLS^297-308 a more minor role in driving CMK-1 nuclear accumulation.

Since FLP activity is controlled by temperature (Saro et al., 2020), we hypothesized that the temperature-evoked CMK-1 relocalization could reflect the prolonged cell-autonomous activity of FLP, independently from the inputs of additional neurons. Consistent with this model, we found that a prolonged heat treatment could still cause CMK-1 nuclear relocalization in unc-13 mutants with impaired synaptic transmission (Richmond et al., 1999) and in unc-31 mutants with impaired neuropeptide signaling (Speese et al., 2007; Figure 2A). These results suggest that the mechanisms leading to temperature-dependent CMK-1 nuclear relocalization could primarily depend on cell-autonomous heat-evoked FLP activity. However, we cannot rule out a redundant function of unc-13 and unc-31 gene products, nor the implication of electrical synpases.

Long-lasting thermal changes are mirrored by fluctuations in steady-state intracellular Ca²⁺ levels in FLP (Saro et al., 2020). In order to evaluate the impact of intracellular Ca²⁺ on CMK-1 subcellular localization, we first tested if an intracellular Ca²⁺ elevation is sufficient to trigger CMK-1 nuclear accumulation in the absence of heat stimuli. We used unc-68(dom13) gain-of-function mutants in which cytoplasmic Ca²⁺ levels are chronically elevated in FLP due to the expression of a deregulated ryanodine receptor/UNC-68 (Marques et al., 2019). In comparison to wild-type, CMK-1 nuclear expression was stronger at 20°C in unc-68(dom13) mutants (Figure 2B, top). These data suggest that a chronic elevation in intracellular Ca²⁺ levels in the unc-68(dom13) mutants is sufficient to favor CMK-1 accumulation in the nucleus.

Next, we evaluated the impact of impairing the elevation of free Ca²⁺ intracellular concentration in transgenic animals overexpressing the Ca²⁺ buffer protein Calbindin (Schumacher et al., 2012). We generated a [mec-3p::Calbindin] transgene to express Calbindin in FLP and confirmed that it could reduce intracellular Ca²⁺ levels, as evidenced by a blunted heat-evoked YFP/CFP ratio increase in transgenic animals coexpressing the Ca²⁺ sensor cameleon YC2.3 (Figure 2—figure supplement 2A). Then, we evaluated CMK-1 localization and found that, whereas the localization of CMK-1 at 20°C was indistinguishable from that in wild-type, Calbindin overexpression in FLP significantly impaired its accumulation in the nucleus at 28°C (Figure 2C). Overexpressing another Ca²⁺ buffer, Parvalbumin, also significantly reduced the nuclear accumulation of CMK-1 at 28°C (Figure 2—figure supplement 3). In contrast, a mutant Parvalbumin with mutations affecting all its calcium binding sites did not impair CMK-1 localization (Figure 2—figure supplement 3), indicating that the effect of Parvalbumin involves its Ca²⁺ binding activity. Collectively, the results of Ca²⁺ buffering experiments suggest that an elevation of intracellular Ca²⁺ in FLP is required for CMK-1 nuclear translocation upon prolonged thermal stimulation.

In an attempt to identify specific Ca²⁺-permeable channels involved, we evaluated the impact of reduction-of-function mutations in the genes coding for the L- and N-type voltage-gated calcium channels (VGCC; egl-19(n582), unc-2(gk355), unc-2(ra612)), for transient receptor potential channels (ocr-2(ak47); osm-9(ky5) double mutants) and for the ryanodine receptor (unc-68(r1161)). These mutations had previously been shown to affect FLP Ca²⁺ response to short stimuli lasting between 30 s and 5 min (Saro et al., 2020). Contrary to our expectations, none of these mutations caused a significant alteration of CMK-1 localization (Figure 2D). These observations prompted us to examine if these mutations could actually impair steady-state Ca²⁺ concentrations over a longer time frame matching that of CMK-1 translocation kinetics. Interestingly, none of the mutations significantly affected the Ca²⁺elevation caused by a 60 min incubation at 28°C (Figure 2—figure supplement 2B). These results suggest either that other channels/transporters are involved or that at least some of these channels work in a redundant manner in regulating Ca²⁺ steady-state levels and temperature-dependent CMK-1 nuclear accumulation.

Finally, reasoning that the binding of Ca²⁺/CaM to CMK-1 could be directly involved in the relocalization mechanism, we examined the impact of a CMK-1(W305S) mutation, affecting a key residue necessary for the initiation of the CaM binding to CaM Kinases (Matsushita and Nairn, 1998). This CMK-1 mutation had a similar effect to the buffering of calcium with exogenous Ca²⁺buffers, preventing CMK-1 nuclear translocation into the nucleus upon prolonged heat stimulation (28°C, Figure 2B, bottom). Furthermore, the W305S mutation was also able to counteract the CMK-1 nuclear accumulation observed in unc-68(dom13) mutant with permanently elevated calcium levels (Figure 2B).

Collectively, these results are consistent with a model in which prolonged thermal stimulation causes prolonged cell-autonomous FLP thermosensory activity, long-lasting changes in intracellular calcium concentration and binding of Ca²⁺/CaM on CMK-1 to promote CMK-1 nuclear translocation.

Our next goal was to understand how the binding of Ca²⁺/CaM on CMK-1 can favor CMK-1 nuclear accumulation. One possible mechanism would consist in the masking of the NES^288-294 element by the binding of Ca²⁺/CaM. Indeed, the NES^288-294 element is adjacent to the CaM binding domain and components of the export machinery might potentially compete with CaM for CMK-1 binding (Figure 3A, NES masking model). If this model is true, one would expect that the NES^288-294 element would be required to prevent CMK-1 nuclear accumulation at 28°C when the CaM-binding is disrupted by the W305S mutation. We therefore examined the localization of CMK-1(V292A/V294A/W305S) mutant proteins, in which both CaM binding and the NES^288-294 element are disrupted. Contrary to the predication made according to the NES masking model, we found that the triple mutant protein expression was cytoplasmic at 28°C, indistinguishable from that of the single W305S mutant (Figure 3B). Therefore, a functional NES^288-294 element is dispensable for the cytosolic retention at 28°C of CaM binding-disruption mutant CMK-1. These data suggest that the masking of the NES^288-294 element upon CaM binding to CMK-1(wt) is unlikely to be a major mechanism promoting CMK-1 nuclear accumulation at 28°C.

An alternative mechanism through which Ca²⁺/CaM binding could favor CMK-1 nuclear accumulation would be to trigger a conformational change that unmasks a functional NLS element (Figure 3C). Binding of importins would thereby be favored under high calcium conditions at 28°C, but not at 20°C when calcium levels are lower. The effect of mutations in the CMK-1 NLS^71-78 element is consistent with this model as they selectively impact CMK-1 localization at 28°C, but not at 20°C (Figure 1F). As presented in the next paragraphs, we further confirmed this NLS unmasking model by testing additional predictions regarding the localization of CMK-1 in vivo and directly quantifying protein affinities in vitro.

A first prediction of the NLS unmasking model is that the increased nuclear expression of CMK-1, triggered by unc-68(dom13) mutation (Figure 3D), will require a functional NLS^71-78 element. This was indeed the case as the expression of the NLS^71-78 mutant CMK-1(K71A/R74Q/R77S) was significantly more cytoplasmic than CMK-1(wt) in the unc-68(dom13), high-calcium background at both 20 and 28°C (Figure 3D). We note that the nuclear accumulation is not fully blocked, which might suggest the existence of a NLS^71-78-independent pathway. A second prediction of the model is that, because the NLS^71-78 element is masked at 20°C, the cytoplasmic exclusion observed in the NES^288-294 mutant at this temperature will not require the NLS^71-78 element. We therefore evaluated the localization of a CMK-1(K71A/R74Q/R77S/V292A/V294) mutant combining NLS^71-78- and NES^288-294-disrupting mutations (Figure 3E). According to our prediction, this mutant protein was excluded from the cytoplasm similarly to CMK-1(V292A/V294) mutant protein lacking only the functional NES^288-294 element. Therefore, at 20°C, when calcium levels are low and there is little CaM binding to CMK-1, the NLS^71-78 element is not engaged to regulate CMK-1 localization.

Next, we sought to more directly test the NLS unmasking model by quantifying the affinity between purified recombinant CMK-1 and importin-α in vitro. Indeed, importin-α are known to make direct interactions with NLS elements and CaM kinase IV was previously shown to interact with importin-α, even if the relevant NLS remained undefined in that case (Kotera et al., 2005). Three importin-α are encoded in the C. elegans genome (IMA-1, -2, and -3). Previous data suggest that IMA-3 is the only isoform expressed in somatic tissues (Geles and Adam, 2001). We confirmed the broad somatic expression of ima-3 using an [ima-3p::NLS::wrmScarlet] reporter. Expressing cells included FLP neurons, as judged with the green/red co-labeling with the mec-3p::CMK-1::mNG reporter (Figure 4A). We therefore decided to conduct the in vitro protein-protein interaction analysis with IMA-3. HisTag::IMA-3, CMK-1(wt), and CMK-1(K71A/R74Q/R77S) were then produced in Escherichia coli, purified and protein-protein interactions quantified using the MicroScale Thermophoresis (MST) (Jerabek-Willemsen et al., 2011). Interaction plots are presented in Figure 4B and C, dissociation constants (Kd) summarized in Table 1, and a schematic interpretation is illustrated in Figure 4D and E. In control experiments, we could not detect any affinity between CaM and IMA-3, nor between CaM and CMK-1 in the absence of Ca²⁺. In the absence of Ca²⁺/CaM, CMK-1 had a relatively low but detectable affinity for IMA-3 (Kd ≈ 1.54 μM). In the presence of Ca²⁺/CaM (2 mM Ca²⁺ and 8 µM CaM), the affinity for IMA-3 raised by about approximately threefold (Kd ≈ 0.45 μM). No significant increase in affinity was seen with Ca²⁺ only (2 mM Ca²⁺, no CaM). These data suggest that the binding of Ca²⁺/CaM to CMK-1 is sufficient to increase its affinity to IMA-3. This increased affinity effect was significantly reduced in CMK-1(K71A/R74Q/R77S) triple mutant, furthermore suggesting that the NLS^71-78 of CMK-1 mediates a major part of the Ca²⁺/CaM-binding-dependent IMA-3 affinity increase.

Collectively, the results of in vivo protein localization analysis and in vitro protein affinity quantification are consistent with the NLS^71-78 unmasking model (schematically depicted in Figure 3C) to explain the increased nuclear accumulation in FLP neurons upon prolonged activation.

In order to demonstrate the physiological relevance of the NLS^71-78 and NES^288-294-dependent control of CMK-1 subcellular localization, we engineered mutants with corresponding point mutations with CRISPR/Cas9 genome editing: cmk-1(syb1435), encoding an R77S mutation that disrupts the NLS^71-78 element, and cmk-1(syb1375), encoding for the V292A/V294A mutations that disrupt the NES^288-294 element (Figure 5A). While the NLS mutant appeared to be superficially wild-type as regards morphology and gross locomotion phenotypes, the NES mutants had partially impaired locomotion with a tendency to coil, which prevented their characterization in some behavioral assays.

First, we evaluate heat-evoked reversals in wild-type and cmk-1 mutants. Similar to previous findings (Lia and Glauser, 2020), loss of cmk-1 in cmk-1(ok287) mutants increased the animal responsiveness to acute stimuli (Figure 5B). Whereas disrupting the NLS^71-78 in cmk-1(syb1435) R77S mutants had no impact on the heat-evoked response, disrupting the NES^288-294 in cmk-1(syb1375) V292A/V294A double mutant significantly reduced animal responsiveness. In this latter case, the constitutive nuclear expression of the NES^288-294 mutant is thus linked to reduced aversive behaviors.

Second, we assessed heat-evoked avoidance using noxious heat thermogradient assays in which we scored how far animals spread in a noxious heat spatial thermogradient. We compared naïve animals (maintained at 20°C) and heat-adapted animals (maintained at 28°C for 1 hr prior to the assay). The adaptation treatment at 28°C caused a desensitization-like effect, shifting the distribution of wild-type animals toward higher temperatures (third quartile of their thermal distribution rising from 30.9 to 31.7°C, p>0.001, Figure 5C). Consistent with previous results (Schild et al., 2014), the loss of CMK-1 in cmk-1(ok287) mutants inhibited this adaptation effect (Figure 5C). We could not determine the impact of the NES^288-294-affecting mutations because cmk-1(syb1375) mutants tended to coil and reverse more frequently, which limited their dispersal on the assay plate. The NLS^71-78-affecting mutation in cmk-1(syb1435) R77S mutants did not affect the response of naïve animals (20°C), but impaired the desensitization effect upon adaptation at 28°C. These results highlight the importance of NLS^71-78-dependent CMK-1 nuclear localization in the process mediating noxious heat desensitization.

Thermal avoidance behaviors in C. elegans are mediated by the FLP pathway, but also additional thermosensory pathways (Liu et al., 2012). In order to selectively address the role of CMK-1 NES^288-294 and NLS^71-78 in the FLP pathway, we analyzed FLP-specific optogenetic-evoked reversals. We previously showed that cell-autonomous CMK-1-dependent thermal avoidance plasticity in the FLP pathway was not linked to the modulation of FLP thermal sensitivity, but to the modulation of FLP neurotransmission (Schild et al., 2014). We therefore expected that CMK-1-dependent FLP output modulation would be detectable with an optogenetic activation, even if this stimulation method bypasses FLP thermosensation.

First, we assessed the impact of NES and NLS mutations in naïve animals grown at 20°C. As compared to reversals in wild-type, the reversals caused by FLP optogenetic activation were significantly reduced in cmk-1(syb1375) V292A/V294A double mutants with altered NES^288-294 (Figure 5D). In contrast, cmk-1(syb1435) R77S mutants with altered NLS^71-78 responded like wild-type. To determine if the NES mutation acts cell-autonomously in FLP, we conducted cell-specific rescue experiments with the mec-3 promoter driving expression in FLP, but not in any of the downstream neurons mediating reversals. We observed a marked rescue effect, which was similar to that obtained with the endogenous cmk-1 promoter (Figure 5E). Whereas we cannot exclude the implication of other neurons as additional sites of action for CMK-1, our data suggest that NES^288-294-controlled CMK-1 cytoplasmic retention in FLP promotes reversals by upregulating FLP outputs.

Second, we analyzed the impact of thermal adaptation treatments on FLP-evoked reversals. A prolonged heat treatment (28°C for 90 min) reduced the reversal responses evoked by the optogenetic activation of FLP in wild-type animals (Figure 5F), confirming that heat-evoked plasticity entails a responsiveness reduction taking place downstream of FLP thermal sensitivity. This adaptation effect was blocked by the NLS^71-78-affecting mutation in cmk-1(syb1435) mutants.

Collectively, these results indicate that the NLS^71-78 and NES^288-294-dependent control of CMK-1 subcellular localization regulates FLP outputs and that NLS^71-78 -dependent nuclear accumulation is essential to dampen the FLP pathway response after persistent stimulations.

We next sought to test if the CMK-1 localization control mechanism identified in FLP was also engaged in other neurons. We focused on the AFD thermosensory neurons, in which thermal shifts in the 15–25°C range were previously shown to affect CMK-1 localization (Yu et al., 2014). Animals were grown at 20°C and then shifted to 15 or 25°C for an overnight incubation. In animals incubated at 15°C, we observed that a CMK-1(wt)::mNG fusion was enriched in the nucleus of AFD (Figure 6A and B). In contrast, CMK-1 localization was cytoplasmic at 25°C (Figure 6A and B). Hence, the impact of long-term thermal history on CMK-1 localization in AFD is the opposite to that in FLP.

Could CMK-1 nuclear accumulation at 15°C reflect a long-term intracellular calcium elevation in AFD, like it does in FLP at high temperature? At first glance, this model looks counterintuitive because heat stimuli trigger intracellular calcium elevations in AFD (Kimura et al., 2004). However, AFD is fast-adapting and these elevations are transients, lasting at most for a few minutes when the stimulus is held constant (Kimura et al., 2004; Clark et al., 2006). Calcium levels over a longer time frame are more likely to be relevant for CMK-1 localization. We therefore compared AFD intracellular calcium levels between animals grown overnight at 15 and 25°C, respectively. We expressed the ratiometric YC2.3 cameleon sensor in AFD and compared the baseline YFP/CFP ratio across unstimulated animals. Remarkably, the baseline YFP/CFP ratio in worms grown and recorded at 15°C (mean = 1.09; sem = 0.03) was significantly higher than that measured in worms grown and recorded at 25°C (mean = 0.93; sem = 0.03; Figure 6D). The magnitude of the YFP/CFP ratio difference caused by differences in the overnight growth temperature (ΔR ≈ 0.16, Figure 6E) was of the same order of magnitude as those caused by acute thermal up-steps (15°C to 25°C, ΔR ≈ 0.13) or down-steps (25°C to 15°C, ΔR ≈ 0.13, Figure 6E and F). We conclude that baseline intracellular calcium level in AFD inversely correlates with growth temperature, a situation opposite to that in FLP (Figure 6—figure supplement 2).

Our findings about AFD indicate that an intracellular calcium elevation (at 15°C) is linked to an accumulation of CMK-1 in the nucleus, whereas an intracellular calcium reduction (at 25°C) is linked to a cytoplasmic expression. It seemed therefore plausible that CMK-1 subcellular localization could be controlled in AFD via the same calcium-dependent mechanism as in FLP. We made a series of observations confirming this hypothesis. First, the nuclear accumulation at 15°C was significantly impaired in the Ca²⁺/CaM binding mutant CMK-1(W305S) and the NLS^71-78 mutant CMK-1(K71A/R74Q/R77S) (Figure 6B). Second, the UNC-68/ryanodine receptor unc-68(dom13) gain-of-function mutation caused a nuclear accumulation at 25°C (Figure 6C, bottom), suggesting that an elevation in calcium is sufficient to promote CMK-1 nuclear expression in AFD. Third, the nuclear accumulation in unc-68(dom13) mutant was completely abolished for CMK-1(W305S) mutant protein (Figure 6C, bottom), further pointing to the importance of Ca²⁺/CaM binding. Fourth, disrupting the NES^288-294 in CMK-1(V292A/V294A) mutants leads to a constitutive nuclear expression in AFD irrespective of the temperature (Figure 6B). Finally, we also confirmed that ima-3 is expressed in AFD (Figure 6—figure supplement 3). Collectively, these findings suggest that a similar mechanism control CMK-1 localization in FLP and AFD.

To further substantiate the physiological relevance of the NLS^71-78 and NES^288-294 elements in vivo in AFD, we evaluated the impact of engineered NLS and NES mutations on thermotaxis behavior and gcy-8 gene expression, which are known to be regulated by CMK-1.

First, we tested the thermotactic response of animals grown at 20°C and deposited at 23°C in a linear temperature spatial gradient. Under these conditions, wild-type animals produced a robust negative thermotaxis response, moving toward cooler temperature (Figure 6G). Whereas cmk-1(syb1435) R77S mutants behaved like wild-type, the response was significantly reduced in cmk-1(ok287) loss-of-function and cmk-1(syb1375) V292A/V294A mutants. These data suggest that an intact NES^288-294 is essential for normal thermotaxis under our assay conditions.

Second, we addressed the impact of NLS and NES mutations on the expression of a transcriptional reporter for the AFD-specific gene gcy-8, whose transcription level was previously reported to rely on CMK-1 (Satterlee et al., 2004). In comparison to growth conditions at 15°C, cmk-1(wt) animals grown at 25°C had a stronger gcy-8p::gfp reporter activity (Figure 6H and I). The loss of CMK-1 in cmk-1(ok287) null mutants significantly reduced gcy-8 transcription at either temperature and reverted the temperature impact. The growth-temperature effect was abolished in both cmk-1(syb1435) and cmk-1(syb1375) mutants, leaving expression levels closer to the situation in cmk-1(wt) animals grown at 25°C. The fact that both mutations caused the same effect despite having opposing effects on CMK-1 localization is in disagreement with a simple model where CMK-1 nuclear accumulation would suffice to regulate gcy-8 transcription. Instead, the ability of CMK-1 to operate an active nucleocytoplasmic shuttling or to be localized in the right compartment at specific times might be essential for gcy-8 gene transcription regulation.

Collectively, these results suggest that normal NES^288-294- and NLS^71-78-dependent CMK-1 subcellular localization control is essential to maintain the function of AFD.

C. elegans strains used in this study are reported in the Key resources table. All strains were grown as previously described Stiernagle, 2006 on nematode growth media (NGM) plates with OP50 E. coli, at 20°C (unless otherwise stated). For optogenetic experiments, we used NGM plates containing all-trans-retinal (ATR), as well as regular NGM plates as control. ATR plates were prepared by adding 0.1% (v/v) of ATR stock (100 mM, in ethanol) to the OP50 bacteria suspension prior to seeding. 300 μl of this mix was used to seed 6 cm plates, containing 8 ml of NGM.

Plasmid DNA was purified with the GenElute HP Plasmid miniprep kit (Sigma) and microinjected in the worm gonad according to a standard protocol (Evans, 2006). We used unc-122p::GFP/RFP as co-injection markers to identify transgenic animals.

Entry plasmids containing specific promoters were constructed by PCR from N2 genomic DNA, with primers flanked with attB4 and attB1r recombination sites and cloned into pDONR-P4-P1R vector (Invitrogen) by BP recombination. Plasmids and primer sequences are reported in the supplementary information, Appendix 1—table 1.

Entry plasmids containing specific coding DNA sequences were constructed by PCR from N2 cDNA with primers flanked with attB1 and attB2 recombination sites and cloned into pDONR_221 vector (Invitrogen) by BP recombination. Plasmids and primer sequences are listed in the supplementary information, Appendix 1—table 1.

All the deletions and point mutations were generated by inverse PCR-based site-directed mutagenesis (Hemsley et al., 1989). In brief, whole plasmids (entry plasmids containing cmk-1 coding DNA sequences) were amplified with the KOD Hot Start DNA Polymerase (Novagen; Merck). Primers were phosphorylated in 5′ and were designed to contain the desired point mutation(s) and to hybridize in a divergent and back-to-back manner on the plasmid. After electrophoresis, linear PCR products were purified from agarose gel (1%) with a Zymoclean-Gel DNA Recovery kit (Zymo Research) and circularized with DNA Ligation Kit <Mighty Mix> (Takara). The plasmid templates, primer sequences, and names of resulting mutation-carrying plasmids are reported in Appendix 1—table 1.

For CMK-1 subcellular localization analysis, the intensity of fluorescence was first measured for each neuron in three regions of interest (ROIs): Nucleus, Cytoplasm, Background. The nuclear/cytoplasmic ratio was calculated as (Nucleus – Background)/(Cytoplasm – Background). A ratio >1 indicates a nuclear accumulation of CMK-1, while a value <1 a cytoplasmic biased ratio. All three ROIs were ellipses of the same area. The Background ROI was defined in a worm region close to the neuron to take autofluorescence into account. Nuclear and Cytoplasmic ROIs were defined based on the mNG green signal and the shape of the neuron, via a procedure that was first validated with the use of a second fluorescent nuclear marker (nuclear ceBFP::NLS). We noticed that the nucleus was always laying very centrally, occupying a large part of the cell body (Figure 1—figure supplement 3). There is actually no possibility for the nucleus to fit in any narrower region of the cell. So even in situations where the green fluorescence signal is diffuse and does not itself reveal the localization of the nucleus, it is possible to make reliable predictions. The nuclear ROI was defined in the middle of the cell body (where it is the widest) and the cytoplasmic ROI decentered to a narrower region of the cell (Figure 1—figure supplement 3). We validated this approach by a blind test, in which we acquired a series of images with diffuse mNG (homogenously distributed between the nucleus and the cytoplasm) and nuclear BFP, and defined the ROI solely based on the green channel. After an a posteriori verification with the blue channel, we found that >98% of the ROI pairs (69/70) were correctly defined. Of note, when working with CMK-1::mNG, a nuclear accumulation or depletion of signal is often seen and further ascertains the localization of the nucleus. Therefore, the very rare errors caused by the mis-definition of ROIs will occur only when the nuclear and the cytoplasmic signals are similar; a situation where the ratio value will anyway be very close to 1, regardless of where the ROIs are defined. Based on this assessment, we concluded that our method is valid to score CMK-1:.mNG subcellular localization, without the need for a systematic use of a nuclear co-marker.

AFD images to measure gcy-8p::GFP transcriptional reporter expression were acquired and processed like described above for FLP, except that only two ROIs were defined (Cell and Background). The expression value was calculated as Cell – Background.

For ima-3p reporter imaging, a Leica TCS SPE-II confocal microscope equipped with 488 nm and 532 nm wavelength diode lasers was used with an ACS APO 40× (oil, NA = 1.15) objective.

DNA encoding proteins of interest were PCR amplified and cloned into NdeI and BamHI restriction sites of pDK2409 for the GST-TEV-tagged protein and of pDK2832 for the His6-tagged proteins. Plasmids were transformed into E. coli BL21 (DE3) (NEB). The proteins were purified after 5 hr of induction with 0.5 mM IPTG at 24–25°C. Cells were collected and lysed in a Microfluidizer Processor M-110L. The cell lysate clean supernatant was incubated for 2 hr with nickel-nitrilotriacetic acid beads (Ni-NTA-Qiagen, Hilden, Germany) as per the manufacturer’s instructions for IMA-3-His6 and with Glutathione superflow beads (Qiagen) for GST-tagged CMK-1 variants.

Ni-NTA beads binding IMA-3-His6 were washed three times in imidazole gradient and eluted in 1 mM PMSF, 0.1% NP-40, 500 mM imidazole, pH 8.0. The elution of CMK-1 was done by incubating the beads with 2% TEV enzyme in lysis buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1.5 mM MgCl₂, 5% glycerol) with 1 mM PMSF, 0.1% NP-40, 1 mM DTT. This step cleaves the GST tag to recover untagged CMK-1. For MST experiments, IMA-3-His6 was applied to Zeba desalting spin columns (Thermo Scientific) to remove the excess of imidazole and the buffer was exchanged to the same buffer of CMK-1-GST with 1 mM PMSF and 0.1% NP-40. Protein concentration was determined by Pierce Microplate BCA protein assay Kit-Reducing Agent Compatible (Thermo Scientific) using BSA as protein standard.

MST experiments were performed using a Monolith NT.115 from NanoTemper Technologies to assess the affinity between IMA-3 and CMK-1(wt) or CMK-1(K71A/R74Q/R77S). IMA-3-HIS6 was labeled using the RED-tris-NTA His tag protein labeling kit, resuspended in HEPES 15 mM, 0.05% Tween 20. Labeled IMA-3-HIS6 protein (at a concentration of 100 nM) was added to a serial dilution of unlabeled CMK-1 prepared in binding buffer (1× PBS with 0.05% Tween 20). Samples were loaded into MST standard capillaries Monolith NT.115 MO-k022 (NanoTemper) and MST measurements were performed using 40% laser power setting. For the assays in the presence of Ca²⁺/CaM, we added 2 mM CaCl₂ and 8 µM CaM (from bovine brain; high purity; Sigma).

The dissociation constants Kd were obtained by plotting the normalized fluorescence (Fnorm) against the logarithm of ligand concentration and fitting using the Kd model with the MO-Affinity Analysis software (NanoTemper Technologies). Experiments were performed in triplicates. Statistical comparisons between Kd values were made using the method described by Paternoster et al., 1998.

All experimental replicates were obtained over at least three different days. Adult worms were either synchronized by standard hypochlorite treatment or picked as L4 larvae on NGM plates 1 day before the experiments.

Optogenetic analysis was carried out in a previously described [FLP::CoChR] genetic background (Schild and Glauser, 2015), in which blue light stimuli of low intensities can activate FLP and trigger reversal responses.

Comparisons were made with one-way and two-way ANOVAs using Jamovi (The jamovi project (2021), jamovi (Version 1.6) [Computer Software]; retrieved from https://www.jamovi.org). A visual inspection of Q-Q plots and nonsignificant Kolmogorov–Smirnov tests (p>0.01) suggested that all datasets could be considered to follow a normal distribution. However, some datasets returned significant results with the Shapiro–Wilk test (p<0.01). For that reason, we conducted and reported the results of both regular ANOVAs and robust ANOVAs. Of note, the two methods gave similar results. Comparisons giving significant effects (p<0.01) with ANOVAs were followed by Bonferroni post-hoc tests.