Chloride oscillation in pacemaker neurons regulates circadian rhythms through a chloride-sensing WNK kinase signaling cascade

SUMMARY

Central pacemaker neurons regulate circadian rhythms and undergo diurnal variation in electrical activity in mammals and flies^1,2. Circadian variation in the intracellular chloride concentration of mammalian pacemaker neurons has been proposed to influence the response to GABAergic neurotransmission through GABA_A receptor chloride channels³. However, results have been contradictory^4–9, and a recent study demonstrated circadian variation in pacemaker neuron chloride without an effect on GABA response¹⁰. Therefore, whether and how intracellular chloride regulates circadian rhythms remains controversial. Here, we demonstrate a signaling role for intracellular chloride in the Drosophila small ventral lateral (sLN_v) pacemaker neurons. In control flies, intracellular chloride increases in sLN_vs over the course of the morning. Chloride transport through sodium-potassium-2-chloride (NKCC) and potassium-chloride (KCC) cotransporters is a major determinant of intracellular chloride concentrations¹¹. Drosophila melanogaster with loss-of-function mutations in the NKCC encoded by Ncc69 have abnormally low intracellular chloride six hours after lights on, loss of morning anticipation, and a prolonged circadian period. Loss of kcc, which is expected to increase intracellular chloride, suppresses the long-period phenotype of Ncc69 mutant flies. Activation of a chloride-inhibited kinase cascade, consisting of WNK (With No Lysine (K)) kinase and its downstream substrate, Fray, is necessary and sufficient to prolong period length. Fray activation of an inwardly rectifying potassium channel, Irk1, is also required for the long-period phenotype. These results indicate that the NKCC-dependent rise in intracellular chloride in Drosophila sLN_v pacemakers restrains WNK-Fray signaling and overactivation of an inwardly rectifying potassium channel to maintain normal circadian period length.

eTOC

Schellinger et al. demonstrate intracellular chloride oscillations in the Drosophila sLN_v central pacemaker neurons regulate morning anticipation and circadian period. Chloride has a signaling role via the chloride-inhibited WNK kinase, which controls activity of an inwardly rectifying potassium channel via the intermediary kinase, Fray.

Graphical Abstract

RESULTS

Intracellular chloride increases in sLN_v pacemakers during the morning in an NKCC-dependent manner

Na⁺/K⁺-ATPase activity results in low intracellular sodium and high intracellular potassium in most cells, generating an inward driving force for chloride through NKCCs and an outward driving force through KCCs¹¹. These transporters are expressed in mammalian suprachiasmatic nucleus (SCN) clock neurons, where they determine intracellular chloride and the GABA reversal potential^{4,5,7,9,10,12}. We used the transgenic sensor, ClopHensor, to measure intracellular chloride ex vivo in the small LN_v neurons (sLN_vs), which are Drosophila central brain pacemaker cells¹³. sLN_vs from controls and from flies with strong loss-of-function mutations in Ncc69, which encodes an NKCC^14–16, were tested in parallel every 4 hours after lights on (ZT0). ClopHensor couples chloride-sensitive GFP to chloride-insensitive dsRed for ratiometric chloride measurement^17–20, with the cyan/red ratio inverse to chloride (Figure S1). Intracellular pH, which influences chloride measurement¹⁷, was measured simultaneously, and was similar over time and between genotypes (Table S1). In controls, chloride rose during daytime, fell, and then rose again during nighttime (Figures 1A, S1D). Brains were exposed to light during dissection and imaging, which could confound nighttime measurements. However, these results suggest that intracellular chloride varies in sLN_v pacemakers over the day/night cycle, as in the mammalian SCN^3,4,9,10.

Figure 1. Intracellular chloride increases in sLN_v pacemakers during the morning in an NKCC-dependent manner.

A) Cyan/red ratios, which are the inverse of intracellular chloride concentrations (Figure S1), were measured using the transgenic sensor ClopHensor. The cyan/red ratios vary over time in sLN_v pacemaker neurons. Mean ± SEM shown. ZT0, lights on. p<0.0001 for time, p=0.8495 for genotype, two-way ANOVA. Intracellular pH from the same neurons and numbers of neurons analyzed is shown in Table S1. Control and Ncc69 mutant flies were analyzed in parallel, with different timepoints measured on different days. B) Cyan/red ratios, measured in the same sLN_vs over time, 2 and 6 hours after lights-on. Mean ± SEM shown. Table S2 lists genotypes and numbers of flies analyzed here and subsequently. Significant effects of time (p=0.0004) and genotype (p=0.0042), two-way repeated measures ANOVA. ***, p=0.0010 for cyan/red ratio in control vs. Ncc69^r2 mutant, Šidák’s multiple comparison test. ZT6 was significantly different from ZT2 in control (p=0.0007) but not Ncc69^r2 mutants (p=0.2577), Šidák’s multiple comparison test. NS, not significant.

See also Figure S1 and Tables S1–S2.

Intracellular chloride appeared to increase from ZT2 to ZT6 in controls, but not in Ncc69 mutants (Figures 1A, S1D). Since the ZT2 and ZT6 measurements were made in different neurons, we next performed repeated measurements in the same neurons at ZT2 and ZT6. Although the cyan/red ratios were slightly lower in this experiment, this experiment also showed that intracellular chloride rose from ZT2 to ZT6 in controls, but remained constant in Ncc69 mutants, with lower chloride concentrations in Ncc69 mutants compared to controls at ZT6 (Figures 1B, S1E). This indicates that intracellular chloride increases in sLN_vs over the course of the morning in an NKCC-dependent manner.

The Ncc69 NKCC is required in sLN_v pacemakers for normal circadian behavior

To test the functional consequences of loss of NKCC activity on circadian rhythms, we examined locomotor activity in Drosophila activity monitors. sLN_vs regulate morning anticipation, the increase in locomotor activity that occurs prior to the onset of daylight in light-dark conditions^21–23. Morning anticipation was abolished in Ncc69 mutants, whereas evening anticipation, which is regulated by other clock neurons^21–23, was not affected (Figures 2A, B).

Figure 2. The Ncc69 NKCC is required in sLN_v pacemakers for normal circadian behavior.

A) Actograms showing locomotor activity during lights off (grey bars) and lights on (white bars) in control and Ncc69^r2 mutants. The arrow indicates increasing locomotor activity (morning anticipation) prior to lights on. Morning anticipation is abolished in Ncc69^r2 mutants. Mean + SEM (dots) shown. B) Morning and evening anticipation index. An anticipation index of 0.5 (see Methods) indicates no anticipation, whereas a value of greater than 0.5 indicates anticipation. Median and 95^th percentile confidence intervals graphed for behavioral data in this and subsequent figures, with individual data points shown. ****, p<0.0001; NS, not significant, unpaired t-test. C) Circadian period is prolonged in Ncc69 mutants. Locomotor activity was measured in constant darkness for 7 days here and subsequently. ****, p<0.0001. Statistical tests used in C and D are listed in Table S3. D) Expression of wild-type Ncc69 in the LN_v pacemakers, under the control of pdf-GAL4, rescues the long-period phenotype of Ncc69 mutant flies. ****, p<0.0001. Right, representative actograms showing average activity across 7 days of subjective day (grey bars) and night (dark bars) in constant darkness. E) LN_v morphology is intact in Ncc69 mutants. PDF neuropeptide-expressing pacemaker neurons were visualized using α-PDF antibodies. Immunostaining in Ncc69 mutants and controls was performed in parallel. Scale bar, 100 μm.

See also Tables S2–S3.

sLN_vs also regulate the free-running clock^24,25. Ncc69^r2 mutants had a prolonged period in constant darkness that was rescued by LN_v-specific expression of wild-type Ncc69 (Figure 2C, D). Although the pdf-GAL4 driver is expressed in both small and large LN_v neurons (lLN_vs)²⁴, the latter have been implicated in sleep and arousal and likely do not contribute to the period-length phenotype^26–28.

Prolonged period was observed in Ncc69^r2 homozygous mutants and in flies with Ncc69^r2 over a deficiency deleting Ncc69, but not in heterozygotes (Figure 2C). LN_v morphology^24,29 was intact in Ncc69^r2 mutants (Figure 2E). We observed weaker rhythmicity in Ncc69^r2 homozygotes. This phenotype was not rescued by LN_v Ncc69 expression, and was not observed in Ncc69^r2/Df (Table S3). Therefore, the reduced rhythm strength may be due to mutation in a different gene, or Ncc69 activity outside the LN_vs, and was not pursued further. Roles for Ncc69 have been demonstrated in glia, including decreased rhythmicity with glial Ncc69 knockdown^14,30–32. However, glial expression of wild-type Ncc69 did not rescue the long-period phenotype or decreased rhythmicity of mutants (Table S3). Thus, Ncc69 is specifically required in sLN_vs for the maintenance of normal circadian period.

Period lengthening in Ncc69 mutants is due to low chloride activation of the WNK-Fray kinase cascade

We reasoned that decreasing KCC activity, which is expected to increase intracellular chloride, might reverse the Ncc69 mutant phenotype. kcc heterozygous mutation or LN_v knockdown resulted in period lengths similar to, or slightly longer than, controls (Table S3). However, both maneuvers suppressed the long-period phenotype of Ncc69 mutants (Figure 3A, B), further indicating that dysregulated intracellular chloride in sLN_vs drives the long-period phenotype of Ncc69 mutants.

Figure 3. Period lengthening in Ncc69 mutants is due to low chloride activation of the WNK-Fray kinase cascade.

Heterozygous loss of kcc (A) or knockdown of kcc in LN_v pacemakers (B) suppresses the long-period phenotype of Ncc69 mutants. LN_v knockdown of the chloride-inhibited WNK kinase (C) or its downstream target, Fray (D), suppresses the long-period phenotype of Ncc69 mutants. LN_v overexpression of Cl⁻-insensitive, but not wild-type, Drosophila WNK (E) or human WNK3 (F) results in lengthening of circadian period. G) Overexpression of activated Fray^T206E, but not kinase-dead Fray^D185A,T206E, increases period length. **, p=0.0026; ***, p=0.0007; ****, p<0.0001. See Table S3 for statistical tests used.

See also Tables S2–S4.

The Ncc69 NKCC is activated by the Drosophila WNK-Fray kinase cascade, homologs of the WNK-SPAK/OSR1 (Ste20-related proline alanine rich kinase/oxidative stress response) kinases that activate mammalian NKCCs^33,34. In addition, chloride inhibits WNK signaling through direct effects on mammalian and Drosophila WNKs^19,35,36. We first tested the hypothesis that LN_v loss of WNK or Fray would phenocopy the Ncc69 loss-of-function phenotype, i.e. circadian period lengthening, due to loss of positive regulation by the kinases. We observed only a slight increase (9–36 minutes) in period length in WNK knockdown flies, and there was no effect of Fray knockdown (Table S4), suggesting that WNK and Fray do not act upstream of Ncc69 in circadian period regulation.

The prolonged circadian period of Ncc69 mutants could also be due to decreased late-morning sLN_v intracellular chloride and excess WNK-Fray activation. Consistent with this hypothesis, LN_v knockdown of either WNK (Figure 3C) or Fray (Figure 3D) suppressed the Ncc69^r2 long-period phenotype. Conversely, LN_v overexpression of the Cl⁻-insensitive WNK^L421F mutant^19,35 caused period lengthening, phenocopying Ncc69 mutants (Figure 3E). Overexpressing wild-type WNK had no effect, emphasizing the importance of chloride regulation of WNK in circadian period regulation (Figure 3E). Mammalian WNK3 is expressed in SCN pacemaker neurons¹². LN_v overexpression of chloride-insensitive, but not wild-type, human WNK3^L295F also resulted in circadian period lengthening, indicating phylogenetically conserved effects of this signaling pathway (Figure 3F and Table S3). Thus, sLN_v loss of WNK chloride inhibition is sufficient to prolong period length.

WNK kinases phosphorylate the T-loop threonine of SPAK/OSR1 kinases to activate them^37,38. Mutation of the Fray T-loop threonine, T206, to a phosphomimicking glutamate results in WNK-independent Fray activation, and rescues the ion transport defect in renal tubules with Drosophila WNK knockdown³⁹. The Fray D185A mutation abolishes kinase activity³⁹. LN_v overexpression of Fray^T206E increased period length in a kinase activity-dependent manner, mirroring the phenotype of loss of Ncc69 or expression of chloride-insensitive WNK (Figure 3G, Table S3). Thus, loss of Ncc69 results in failure of intracellular chloride to rise during the morning hours, and, as a consequence, WNK and Fray overactivation, initiating a signaling cascade that prolongs circadian period.

Fray activates the inwardly rectifying potassium channel Irk1 to prolong circadian period

To identify possible Fray targets required for circadian period prolongation, we queried the Drosophila proteome for putative Fray RFXV/I binding motifs⁴⁰ and identified 127 candidates using an optimized motif (see Methods) (Data S1A). We pursued the inwardly rectifying potassium channel, Irk1 (also known as Ir), for three reasons. First, LN_v Irk1 knockdown affects circadian period⁴¹. Second, Irk1 transcript is highly enriched in LN_vs, and exhibits transcriptional and translational cycling^42,43. Third, the mammalian Fray homolog, OSR1, regulates mammalian inwardly rectifying potassium channels containing RFXV-related motifs (i.e., RXFXV)⁴⁴.

To examine whether Fray regulates Irk1 channel activity, we performed whole-cell patch clamp recordings of S2-R+ cultured Drosophila cells transfected with Irk1, with or without Fray^T206E co-expression from a multi-cistronic plasmid. Irk1 activity was determined from barium-sensitive currents (Figure S2). Fray^T206E expression increased Irk1 channel activity. Mutation of the putative Irk1 Fray-binding RFXV motif to RFXA (Irk1^V306A) decreased channel activity relative to wild-type Irk1 in the absence of co-transfected Fray^T206E. Fray is endogenously expressed in S2-R+ cells⁴⁵, so this could represent decreased stimulation by endogenous Fray, although we cannot rule out direct effects of the mutation on the channel. The Irk1 RFXA mutation also blunted stimulation by Fray^T206E, with no significant difference in Irk1 current density with or without Fray^T206E co-expression at a holding potential of −150 mV (Figure 4A, B). Together, these results indicate a stimulatory effect of Fray^T206E on wild-type, but not RFXA mutant Irk1.

Figure 4. Fray activates the inwardly rectifying potassium channel Irk1 to prolong circadian period.

Fray^T206E expression increases activity of wild-type Irk1, but not Irk1^V306A, carrying a mutation in the predicted Fray-binding RFXV motif. A) Current-voltage curves from S2-R+ Drosophila cultured cells transfected with wild-type Irk1 or Irk1^V306A, with or without co-expression of constitutively active Fray^T206E. B) Current density at −150 mV. **, p<0.01, ****, p<0.0001, two-way ANOVA with Tukey’s multiple comparisons test. n=12–16 cells analyzed/genotype. C) Irk1^V306A suppresses the long-period phenotype of Fray^T206E. Irk1 was knocked down in the LN_v pacemakers using RNAi, and replaced with Irk1^WT or Irk1^V306A transgenes with silent mutations in the RNAi target sites. ****, p<0.0001. See Table S3 for statistical tests used.

See also Table S2, S3, Data S1A and Figure S2.

We next assessed the in vivo role of Fray stimulation of Irk1. We designed Irk1^WT and Irk1^V306A transgenes to be resistant to a previously-validated short hairpin Irk1 RNAi⁴⁶, allowing replacement of endogenous pacemaker neuron Irk1 with either the wild-type or V306A mutant. LN_v Irk1 knockdown and replacement with either Irk1^WT or Irk1^V306A had no effect on period length (Figure 4C), consistent with the lack of phenotype with Fray knockdown (Table S4). Expressing Fray^T206E in the pacemaker neurons increased period length in wild-type flies and in flies with endogenous Irk1 replaced with Irk1^WT, while replacement of Irk1 with Irk1^V306A suppressed the long-period phenotype (Figure 4C). This suggests that Fray stimulation of Irk1 is required for the long-period phenotype observed with loss of Ncc69 and activation of the WNK-Fray kinase cascade.

DISCUSSION

Intracellular chloride oscillates in central pacemaker neurons of the mammalian SCN^3,4,9,10, but the functional significance of this oscillation has remained unclear. Here, we demonstrate an NKCC-dependent increase in intracellular chloride in Drosophila sLN_v pacemaker neurons over the course of the morning, which constrains activity of the chloride-sensitive WNK kinase, its downstream substrate, Fray, and an inwardly rectifying potassium channel, Irk1, to maintain normal circadian periodicity. Loss of the Ncc69 NKCC in LN_v pacemaker neurons results in loss of morning anticipation and lengthening of the circadian period in free-running conditions. Our observations are consistent with studies in the SCN implicating NKCC in the determination of intracellular chloride in mammalian pacemaker neurons^{4,5,7,9,10,12}, and connects intracellular chloride to behavioral circadian phenotypes.

Diurnal variations in intracellular chloride have been proposed to influence the effect of GABAergic neurotransmission on clock neurons in mammals^3–9, but a recent study challenged this idea¹⁰. Existing data indicate a minor role for sLN_v ligand-gated chloride channels in the regulation of circadian period^{27,28,47–49}. Rather, we demonstrate a signaling role for chloride in sLN_vs, via inhibition of WNK-Fray signaling. As chloride-sensitive kinases, WNKs are poised to interpret changes in intracellular chloride and initiate downstream signal transduction cascades^35,50,51. This has been studied in transepithelial ion transport in Drosophila and mammalian renal epithelia^19,52,53, as well as in the clearance of apoptotic corpses⁵⁴. Our findings further extend this concept to circadian pacemaker neurons.

KCC and another cation-chloride cotransporter, encoded by the NKCC gene, have been linked to the electrophysiological response to GABA in lLN_vs, which express GABA_A receptor chloride channels^27,28,55 (Eick et al, this issue). The transport activity of the NKCC-encoded transporter has not been characterized, but may differ from the Ncc69 NKCC, as suggested by sequence differences and the transport activity of an Aedes aegypti ortholog^56,57. Expression of Ncc69 and NKCC may also differ. Consistent with this idea, broad clock neuron knockout of NKCC has no effect on morning anticipation or period length, but leads to abnormal rhythmicity in constant light, as does NKCC overexpression or loss or gain of kcc⁵⁵ (Eick et al, this issue). Thus, intracellular chloride likely affects clock neurons both by affecting the driving force for chloride through neurotransmitter-gated chloride channels, and via inhibition of WNK-Fray signaling. Interestingly, knocking down WNK and Fray broadly in clock neurons phenocopies loss of kcc and NKCC in the same neurons (Eick et al., this issue), suggesting that WNK and Fray act in their usual regulatory roles upstream of the transporters in this case, and further highlighting distinct actions of this pathway in different subpopulations of Drosophila clock neurons.

Pacemaker neurons in flies and mammals undergo cell-autonomous, molecular clock-controlled circadian variation in electrical activity^{1,2,13,58–60}, and altering the excitability of the LN_v pacemakers disrupts circadian rhythms^41,61–65. Two voltage-gated potassium channels have been implicated in LN_v neuron electrical oscillations⁶⁶, and a sodium leak current and potassium channels contribute to the day/night cycling of resting membrane potential in Drosophila dorsal clock neurons⁶⁷. Because inwardly rectifying potassium channels play an important role in determining cellular membrane potential⁶⁸, which is thought to be a determinant of circadian variation in electrical activity^1,2, chloride regulation of Irk1 activity could also contribute to the diurnal variation in sLN_v excitability. Specifically, Irk1 activation at ZT6 due to low intracellular chloride and activation of the WNK-Fray pathway in Ncc69 mutants may disrupt the usual circadian pattern of membrane depolarization and hyperpolarization in sLN_vs⁶⁹, thereby altering period length.

Could intracellular chloride play a signaling role in SCN pacemaker neurons? NKCC1, KCCs and WNK3 are expressed in the rat SCN¹². The repertoire of ion channels regulating pacemaker neuron excitability is complex and incompletely understood, but large-conductance Ca²⁺-activated potassium channels have been implicated, and are regulated by mammalian WNKs^2,70–75. Whether oscillating intracellular chloride observed in SCN neurons regulates these or other ion channels modulating pacemaker neuron electrical properties will be of interest for future investigation.

STAR METHODS

RESOURCE AVAILABILITY

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Aylin Rodan (aylin.rodan@hsc.utah.edu).

Materials Availability

All materials generated for this study, such as plasmids and fly lines, are available upon request from the Lead Contact.

Data and Code Availability

All data reported in this paper will be shared by the lead contact upon request. This paper does not report original code. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Drosophila Strains and Fly Husbandry

Drosophila melanogaster strains used are shown in the Key Resources Table and the genotypes used for each experiment are shown in Table S2. w; pdf-GAL4, w; UAS-WNK^RNAi, w; UAS-Fray^RNAi, w; UAS-Irk1^RNAi, w; UAS-kcc^RNAi, w; UAS-Fray^T206E, w; UAS-WNK^WT, w; UAS-WNK^L421F, w; UAS-WNK3^WT, w; UAS-WNK3^L295F, w; UAS-Fray^D185A,T206E and w; UAS-dcr-2 were outcrossed for 5 generations to wBerlin, which was also used for generating heterozygous controls (e.g., w; pdf-GAL4/+). Except for w; UAS-Fray^RNAi (Vienna), and w; UAS-kcc^RNAi, knockdown of the targeted genes has previously been validated by qRT-PCR, as referenced in the Key Resources Table. Recombinant chromosomes and combinations of transgenes were generated by standard genetic techniques. Generation of new transgenic strains is described below. Flies were reared on a standard cornmeal-yeast-molasses medium prepared in a central kitchen at UT Southwestern or the University of Utah. Flies were reared at room temperature (22–23 °C) or at 25 °C. Young adult (<2 week old) male flies were used in all assays. Males were used to avoid female egg-laying and the emergence of larvae in Drosophila activity monitors.

Key resources table.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse monoclonal anti-PDF C7 antibody	Developmental Studies Hybridoma Bank	Cat#PDF C7; RRID: AB_760350
Goat anti-mouse polyclonal IgG (H+L) cross-adsorbed secondary antibody, Alexa Fluor 488	Thermo Fisher	Cat#A-11001
Chemicals, peptides, and recombinant proteins
CellFectin	Thermo Fisher	Cat#10362100
tributyltinchloride	Sigma	Cat#T50202
nigericin	Thermo Fisher	Cat#N1495
carbonyl cyanide 3-chlorophenylhydrazone	Sigma	Cat#C2759
valinomycin	Sigma	Cat#V0627
TransIT-Insect transfection reagent	Mirus Bio	Cat#6104
Critical commercial assays
LR Clonase II	Thermo Fisher	Cat#11791020
Phusion high-fidelity DNA polymerase	New England Biolabs	Cat#M0530
QIAquick	Qiagen	Cat#28104
pENTR/D-TOPO cloning kit	Thermo Fisher	Cat#K240020
QuikChange II XL	Agilent	Cat#200521
NEBuilder HiFi DNA Assembly	New England Biolabs	Cat#E2621
Experimental models: Cell lines
D. melanogaster: Cell line S2-R+	Krämer; Drosophila Genomics Resource Center	Cat#150; RRID:CVCL_Z831
Experimental models: Organisms/strains
D. melanogaster: control: wBerlin	Rothenfluh	N/A
D. melanogaster: LNv driver: w; pdf-GAL4	Rosbash24	N/A
D. melanogaster: Ncc69 mutant: w; Ncc69r2	Leiserson14	N/A
D. melanogaster: Ncc69 rescue: w; Ncc69r2 UAS-HA-Ncc692–3	Leiserson14	N/A
D. melanogaster: Glial Ncc69 rescue: w; gli-GAL4; Ncc69r2 UAS-HA-Ncc692–3	Leiserson14	N/A
D. melanogaster: kcc mutant: w; kccDHS1	Hekmat-Scafe86	N/A
D. melanogaster: RNAi targeting kcc: w; UAS-kccRNAi	Vienna Drosophila Resource Center	ID 101742; RRID: FlyBase_FBst0473615
D. melanogaster: Ncc69 deficiency: w1118;Df(3L)ED4475, p{3’.RS5+3.3’}ED4475/TM6C, cu1 Sb1	Bloomington Drosophila Stock Center	RRID: BDSC_8069
D. melanogaster: Dicer transgene: w; UAS-dcr-2	Bloomington Drosophila Stock Center	RRID: BDSC_24651
D. melanogaster: RNAi targeting WNK: w; UAS-WNKRNAi	Bloomington Drosophila Stock Center; validated in [19]	RRID: BDSC_42521
D. melanogaster: RNAi targeting WNK: w; UAS-WNKRNAi (Vienna)	Vienna Drosophila Resource Center; validated in [39]	ID 106928; RRID: FlyBase_FBst0478751
D. melanogaster: Wild-type WNK: w; UAS-WNKWT	Rodan19	N/A
D. melanogaster: Chloride-insensitive WNK: w; UAS-WNKL421F	Rodan19	N/A
D. melanogaster: Human WNK3: w; UAS-WNK3WT	Rodan31	N/A
D. melanogaster: Chloride-insensitive human WNK3: w; UAS-WNK3L295F	Rodan20	N/A
D. melanogaster: RNAi targeting Fray: w; UAS-frayRNAi	Rodan39	N/A
D. melanogaster: RNAi targeting Fray: w; UAS-frayRNAi (Vienna)	Vienna Drosophila Resource Center	ID 106919; RRID: FlyBase_FBst0478742
D. melanogaster: Activated Fray: w; UAS-FrayT206E	Rodan39	N/A
D. melanogaster: Kinase-dead activated Fray: w; UAS-FrayD185A,T206E	This paper	N/A
D. melanogaster: RNAi targeting Irk1: w; UAS-Irk1RNAi	Bloomington Drosophila Stock Center; validated in [46]	RRID: BDSC_25823
D. melanogaster: WT Irk1 (RNAi-resistant): w; UAS-Irk1WT-RR	This paper	N/A
D. melanogaster: V306A mutant Irk1 (RNAi-resistant): w; UAS-Irk1V306A-RR	This paper	N/A
D. melanogaster: pH/chloride sensor: w; UAS-ClopHensor c202	Krämer19	N/A
D. melanogaster: injection strain for UAS-FrayD185A,T206E: y1M{vas-int.Dm}ZH-2A w*; M{3xP3-RFP.attP′}ZH-22A	Bloomington Drosophila Stock Center	RRID: BDSC_24481
D. melanogaster: injection strain for UAS-Irk1WT-RR and UAS-Irk1V306A-RR: M{vas-int.Dm}ZH-2A, M{3xP3-RFP.attP}ZH- 51D	Bloomington Drosophila Stock Center	RRID: BDSC_24483
Oligonucleotides
See Table S6	N/A	N/A
Recombinant DNA
FrayD185A,T206E ORF in Gateway donor plasmid: pENTR-FrayD185A,T206E	Rodan39	N/A
Gateway destination vector: pUASg.attB	Bischof and Basler77	N/A
Irk1 ORF: FI16807	Drosophila Genomics Resource Center	Cat#1644763; Flybase_FBcl0742744
Gateway donor plasmid: pENTR	Thermo Fisher	Cat#K240020
Irk1WT ORF in Gateway donor plasmid: pENTR-Irk1WT	This paper	N/A
Irk1V306A ORF in Gateway donor plasmid: pENTR-Irk1V306A	This paper	N/A
RNAi-resistant Irk1WT ORF in Gateway donor plasmid: pENTR-Irk1WT-RR	This paper	N/A
RNAi-resistant Irk1V306A ORF in Gateway donor plasmid: pENTR-Irk1V306A-RR	This paper	N/A
RNAi-resistant Irk1WT in pUAS plasmid: pUASg.attB-Irk1WT-RR	This paper	N/A
RNAi-resistant Irk1V306A in pUAS plasmid: pUASg.attB-Irk1V306A-RR	This paper	N/A
Source for ClopHensor cDNA: pcDNA3-ClopHensor	Addgene17	Cat#25938; RRID: Addgene_25938
pIB plasmid	Thermo Fisher	Cat#V802001
pH/chloride sensor in S2 cell expression plasmid: pIB-ClopHensor	This paper	N/A
pAc5 source plasmid: pAc5 STABLE2 Neo	Addgene85	Cat#32426; RRID: Addgene_32426
Irk1WT in S2 cell expression plasmid: pAc5-Irk1WT	This paper	N/A
Irk1V306A in S2 cell expression plasmid: pAc5-Irk1V306A	This paper	N/A
FrayT206E ORF in Gateway donor plasmid: pENTR-FrayT206E	Rodan39	N/A
Irk1WT and FrayT206E in multi-cistronic S2 cell expression plasmid: pAc5-Irk1WT-FrayT206E	This paper	N/A
Irk1V306A and FrayT206E in multi-cistronic S2 cell expression plasmid: pAc5-Irk1V306A-FrayT206E	This paper	N/A
Software and algorithms
Software: Fly activity analysis suite (FAASx)	Boudinot and Rouyer, https://neuropsi.cnrs.fr/en/departments/cnn/group-leader-francois-rouyer/	N/A
Software: ClockLab, version 6	Actimetrics	RRID: SCR_014309
Software: FIJI	Image J, NIH, https://fiji.sc/	RRID: SCR_002285
Software: GraphPad Prism, version 9	GraphPad	RRID: SCR_002798
Software: Pulse	Molecular Devices	N/A
Software: pClamp, version 9.2	Molecular Devices	RRID: SCR_011323
Other
DAM2 Drosophila activity monitors	Trikinetics	N/A
35 mm glass bottom dishes with 14 mm microwell/#1.5 cover glass	CellVis	Cat#D35-14-1.5-N
NCBI protein database	https://www.ncbi.nlm.nih.gov/protein/	RRID:SCR_003257

Cell lines

S2-R+ cultured Drosophila cells were obtained from Helmut Krämer (UT Southwestern) or the Drosophila Genomics Resource Center (stock #150) and cultured at 25 °C in Schneider’s medium (Thermo Fisher Cat #21720001) with 10% FBS (Thermo Fisher Cat #10082139). S2-R+ cells are male⁷⁶. Cells were not authenticated prior to use.

METHOD DETAILS

Generation of Drosophila transgenics

The open reading frame encoding Fray^D185A,T206E was recombined from pENTR-Fray^{D185A,T206E 39} into the pUASg.attB Gateway-compatible destination vector, obtained from Johannes Bischof and Konrad Basler (Zürich, Switzerland)⁷⁷, using LR Clonase II (Thermo Fisher Cat #11791020). After sequence confirmation, midiprep DNA was injected into stock #24481 (y¹M{vas-int.Dm}ZH-2A w*; M{3xP3-RFP.attP′}ZH-22A) by Rainbow Transgenic Flies (Camarillo, CA). Single male transformants were isolated by the presence of ‘mini white’ and confirmation of the UAS-transgene was performed by PCR with sequence-specific primers.

A plasmid (FI16807, stock #1644763) containing the open reading frame of Irk1 was obtained from the Drosophila Genomics Resource Center (Indiana University, Bloomington, IN). The open reading frame was PCR-amplified (Phusion high-fidelity DNA polymerase, New England Biolabs Cat #M0530) using primers Irk1-F and Irk1-R (primers and plasmids are listed in the Key Resources Table and Table S6). After gel extraction (Qiagen QIAquick Cat #28104), the PCR product was cloned into pENTR using the pENTR/D-TOPO cloning kit (Thermo Fisher Cat #K240020) and the sequence confirmed by Sanger sequencing. Next, a mutation was introduced into pENTR-Irk1^WT to generate a mutation in the putative Fray-binding motif, in which Val 306 in the “RFXV” motif is mutated to Ala. The corresponding “GTG” was mutated to “GCG” using QuikChange II XL (Agilent Cat #200521) and primers Irk1-V306A-F and Irk1-V306A-R to generate pENTR-Irk1^V306A. Next, mutations were introduced into pENTR-Irk1^WT and pENTR-Irk1^V306A to render the transgenes resistant to knockdown by co-expression of the Irk1 RNAi. Every third nucleotide in the twenty base pairs targeted by the RNAi was mutated in order to introduce five silent mutations (i.e., CTAAAGGAACGCTTC was mutated to CTGAAAGAGCGTTTT), using QuikChange II XL (Agilent Cat #200521) and primers Irk1-RR-F and Irk1-RR-R. The resulting plasmids were called pENTR-Irk1^WT-RR and pENTR-Irk1^V306A-RR. Irk1 sequences in all plasmids were confirmed by Sanger sequencing. The open reading frames of Irk1^WT-RR and Irk1^V306A-RR were then recombined into pUASg.attB using LR Clonase II (Thermo Fisher Cat #11791020) to generate pUASg.attB-Irk1^WT-RR and pUASg.attB-Irk1^V306A-RR. Midiprep DNA was injected into stock #24483 (M{vas-int.Dm}ZH-2A, M{3xP3-RFP.attP}ZH- 51D), transformants isolated, and UAS-transgenes confirmed as above.

Circadian rhythm analysis

Male flies were collected within 72 hours of eclosion and kept on standard food for 3–5 days in a 12-hour oscillating light/dark incubator. After the entrainment period, individual flies were placed into 5 mm diameter glass cuvettes with standard medium at one end and a tissue plug at the other, allowing the flies free movement throughout the cuvette. Beam breaking by locomotor activity was recorded in 30 minute increments (bins) using Drosophila Activity Monitors (TriKinetics) in constant darkness over a period of 7 days. Activity data were then analyzed using FAASx software (Paris-Saclay Institute of Neuroscience). Cycle p analyses provided period length (tau) and rhythmic strength (power) for individual flies. Settings used were: Minimum period peak power 20, Minimum period peak width 0200. Output was restricted to flies with period lengths of 26 hours ± 10 hours (minimum tau 16, maximum 36). Flies that did not survive the full 7 days (168 hours) or were arrhythmic (power less than 20) were not included in tau analysis.

For experiments in light:dark (LD) conditions, individual flies were loaded into cuvettes 24–72 hours after eclosion and remained in the Drosophila Activity Monitors in LD at room temperature (~22–23 °C) for 6 days. Data from the final 3 days, after 3 days of entrainment, was used for analysis. Morning and evening anticipation index was calculated according to the methods of Schlichting et al.⁷⁸: morning anticipation index = sum of activity ZT21-ZT0/sum of activity ZT18-ZT0, and evening anticipation index = sum of activity ZT9-ZT12/sum of activity ZT6 – ZT12.

For the Irk1 experiment, analysis was performed in ClockLab version 6 (Actimetrics, Wilmette, IL), using the chi squared periodogram function. Settings used were: Start Day 1, End Day 9, type chi squared, start 16 hours (minimum tau), end 36 hours (maximum tau), significance 0.01. In order to classify rhythmic strength, cutoffs were determined by assaying wBerlin control flies. Flies that did not survive the full 8 days or were arrhythmic (power less than 2500) were not included in tau analysis.

PDF neuron immunohistochemistry

Brains were dissected from adult flies in PBS (phosphate-buffered saline, in mM: 137 NaCl, 2.7 KCl, 8.1 Na₂HPO₄, 1.5 KH₂PO₄, pH 7.3–7.4) and fixed for 20–40 minutes in 4% formaldehyde. Brains were then rinsed in PBS, followed by PBT (PBS+0.3% Triton X-100) 4–5 times. Brains were incubated in mouse anti-PDF (PDF C7 from Developmental Studies Hybridoma Bank, Iowa City, IA⁷⁹), 1:800 in 10% normal goat serum in PBT, overnight at 4°C, then rinsed 4–5 times in PBT and 2–3 times in PBS. Brains were then incubated in 1:800 goat anti-mouse Alexa Fluor 488 secondary antibody (Thermo Fisher Scientific Cat #A-11001) overnight, rinsed in PBT and PBS, and then mounted in 80% glycerol in PBS. Imaging was performed using a Zeiss LSM510 confocal microscope.

Measurement of intracellular Cl⁻

The transgenic sensor ClopHensor, which contains a chloride-sensitive enhanced GFP variant (E²GFP, carrying a T203Y mutation) coupled to a chloride-insensitive monomeric dsRed, allows ratiometric measurement of chloride based on emission after excitation at 458 nm (chloride-sensitive) and 543 nm (chloride-insensitive). ClopHensor is less prone to bleaching compared to other chloride sensors, and provides stable and long-lasting readouts of intracellular chloride, making it suitable for the measurements over time described below. Simultaneous measurement of pH can be accomplished due to the pH-dependent E²GFP signal after excitation at 488 nm and the pH-independent E²GFP signal after excitation at 458 nm, providing a ratiometric measurement of pH^17,18.

pH calibration

pcDNA3-ClopHensor (Addgene plasmid #25938) was cut with HindIII (New England Biolabs Cat #R3104) and NotI (New England Biolabs Cat #R3189) and the resulting product ligated into pIB (Thermo Fisher Cat #V802001) to generate pIB-ClopHensor. pIB-ClopHensor was transiently transfected into S2-R+ cultured Drosophila cells obtained from Helmut Krämer (UT Southwestern) using CellFectin reagent (Thermo Fisher Cat #10362100). 48 hours after transfection, cells were bathed in pH-varied solution containing: 38 mM Na gluconate, 100 mM K gluconate, 0.6 mM MgSO₄, 20 mM HEPES (varied pH), 10 μM tributyltinchloride (Sigma Cat #T50202), 5 μM nigericin (Thermo Fisher Cat #N1495), 5 μM carbonyl cyanide 3-chlorophenylhydrazone (Sigma Cat # C2759) and 5 μM valinomycin (Sigma Cat # V0627). After equilibration for at least 1 hour, cells were imaged using a Zeiss LSM510 confocal microscope, with excitation at 488 nm (green emission) and 458 nm (cyan emission). Individual cells (19–25 cells for each pH) were then outlined and pixel intensity measured using ImageJ without image manipulation. The ratios of green/cyan vs. pH were entered into GraphPad Prism, and a sigmoidal curve interpolated using the function “sigmoidal, 4PL, X is log(concentration).” This provided the values for the following equation, used to calculate intracellular pH (pH_i) in the pacemaker neurons¹⁸:

$$p{H}_i=p{K}_a-\frac{1}{p}\ast \log \left(\frac{B2-B1}{R_{pH}-B1}-1\right)$$

where R_pH is the experimentally derived green/cyan ratio, pK_a = 7.254, p = power (Hill slope, 1.668), and B1 (0.2603) and B2 (1.915) are the minimum and maximum asymptotic values of R_pH.

Cl⁻ calibration

Brains expressing ClopHensor in the pacemaker neurons (w/Y; pdf-GAL4 UAS-ClopHensor c202) were dissected from 3–5 day old flies in Drosophila saline, consisting of (in mM): NaCl 117.5, KCl 20, CaCl₂ 2, MgCl₂ 8.5, NaHCO₃ 10.2, NaH₂PO₄ 4.3, HEPES 15, and glucose 20, pH 7.0. Brains were attached to the bottom of 35 mm glass bottom dishes with 14 mm microwell/#1.5 cover glass (Cellvis) coated with poly-lysine, and the solution exchanged to the chloride calibration solution, consisting of (in mM): 100 mM Na-Cl/gluconate, 50 mM K-Cl/gluconate, 2 mM Ca-Cl/gluconate, 8.5 mM Mg-Cl/gluconate, 20 mM glucose, 15 mM HEPES pH 7.1, 10 μM tributyltinchloride, 5 μM nigericin, 5 μM carbonyl cyanide 3-chlorophenylhydrazone and 5 μM valinomycin. Cl/gluconate anions were adjusted to achieve varying chloride concentrations. After 1 hour equilibration, brains were imaged using a Zeiss LSM510 confocal microscope, with excitation at 543 nm (red emission) and 458 nm (cyan emission). Individual neuron cell bodies (10 per Cl⁻ concentration, from multiple brains) were outlined and pixel intensity measured in ImageJ without image manipulation. The ratios of cyan/red vs Cl⁻ were entered into GraphPad Prism, and a sigmoidal curve interpolated using the function “sigmoidal, 4PL, X is log(concentration).” This provided the values for the following equation, used to calculate intracellular Cl⁻ ([Cl⁻]_i)¹⁸:

$${\left[C{l}^{-}\right]}_i=K_d\ast {\left(\frac{A1-A2}{R_{Cl}-A2}-1\right)}^{\frac{1}{p}}$$

where R_Cl is the experimentally derived cyan/red ratio, K_d = 53.49, p = power (set to 1 per methods of¹⁸), and A1 (1.538) and A2 (0.597) are the maximum and minimum asymptotic values of R_Cl.

Measurement of R_pH and R_Cl

Male flies were entrained in 12:12 LD conditions at room temperature (~22–23 °C) for 4 days. Brains expressing ClopHensor in the pacemaker neurons were removed from the incubator at specific ZT times (time after lights on), after which they were exposed to ambient daytime light, and dissected in the following solution, adapted from solutions used for two-photon calcium imaging of fly brain neuronal activity⁸⁰: in mM, NaCl 108, KCl 5, CaCl₂ 2, MgCl₂ 8.2, NaHCO₃ 4, NaH₂PO₄ 1, trehalose 5, sucrose 10, HEPES 5, pH 7.5. Dissected brains were attached to the bottom of 35 mm glass bottom dishes with 14 mm microwell/#1.5 cover glass (Cellvis) coated with poly-lysine, and then bathed in the above solution for about 60 minutes prior to imaging using a Zeiss LSM510 confocal microscope, with excitation at 543 nm (red emission), 488 nm (green emission), and 458 nm (cyan emission). Individual sLN_v neuronal cell bodies (distinguished from lLN_v neurons based on position and size) were outlined in ImageJ and pixel intensity captured for each emission channel. The ratios of green/cyan and cyan/red were used to calculate pH and Cl⁻ as described above. pH and Cl⁻ were measured for each individual neuron. For practical reasons, measurements on brains removed at different ZT times across 24 hours were performed on separate days, but control and Ncc69 mutant brains were always tested in parallel. In order to specifically examine the effect of time of day in each genotype, we performed paired measurements in the same brains, as we performed previously in Malpighian tubule epithelial cells¹⁹. Flies were removed from the incubator at ZT2, and brains were dissected and imaged as above. The dishes were then sealed with parafilm to prevent evaporation and returned to the incubator (in which lights were still on) for four hours. The same brains were re-imaged 4 hours later.

Proteome-wide search for Fray binding motifs

The Drosophila proteome was searched for putative Fray binding motifs, using the methods of Delpire and Gagnon⁴⁰. First, the NCBI protein database was searched for Drosophila melanogaster and results were saved to a FASTA text file. Duplicate results and results from organisms other than D. melanogaster were eliminated. This list was then searched for two motifs. One motif, [S/G/V]RFx[V/I]xx[I/V/T/S], was derived from Delpire and Gagnon⁴⁰, and is called the “Gagnon motif” in (Data S1B). However, this screen failed to identify Ncc69, which we have previously validated as a Fray target³⁹. It also failed to identify KCC, which is a validated mammalian SPAK/OSR1 target^81–84. We therefore performed sequence alignment of the Drosophila homologs of three of the best-known families of mammalian SPAK/OSR-interacting proteins: WNKs, NKCCs/NCC, and KCCs (Drosophila WNK, Ncc69, and Drosophila KCC, respectively). From this we derived a second motif, referred to as the “Drosophila motif” (Data S1A): [D/E/N/Q/S/T/Y]RFx[V/I]xxxx[D/E/G/P]. A second list of proteins was generated using this motif.

Patch clamp analysis of Irk1

Generation of plasmids for S2 cell expression

To generate plasmids for cellular expression of Irk1^WT and Irk1^V306A, with or without Fray^T206E co-expression, the Irk1^WT and Irk1^V306A open reading frames were amplified from pENTR-Irk1^WT or pENTR-Irk1^V306A by PCR (Phusion high-fidelity DNA polymerase, New England Biolabs Cat #M0530). Primers pAc5-Irk1-F and pAc5-Irk1-R included additional sequence for subsequent Gibson assembly cloning into the multi-cistronic vector, pAc5 STABLE2 Neo (Addgene Cat #32426⁸⁵), which uses T2A sequences to generate multiple polypeptides off of a single transcript, and also contains a GFP cassette. The pAc5 plasmid backbone was PCR-amplified using primers pAc5-F and pAc5-R. The products were then assembled using NEBuilder HiFi DNA Assembly (New England Biolabs Cat #E2621) to generate plasmids pAc5-Irk1^WT and pAc5-Irk1^V306A. Irk1 inserts were confirmed by Sanger sequencing. Then, the open reading frame encoding Fray^T206E was amplified from an existing pENTR-Fray^T206E plasmid³⁹, using primers pAc5-Fray-F and pAc5-Fray-R, and the pAc5-Irk1^WT or pAc5-Irk1^V306A plasmid backbones were amplified using primers pAc5-Irk1-Fray-F and pAc5-Irk1-Fray-R. Using this strategy, the Neo^R cassette was replaced by the open reading frame encoding Fray^T206E, while the GFP cassette was retained to allow for identification of successfully transfected cells. After Gibson assembly, inserts were confirmed by Sanger sequencing. However, the Irk1 ORF contained a stop codon at the end, which would prevent expression of the downstream Fray^T206E. The stop codon was removed using QuikChange II XL (Agilent Cat #200521) and primers Irk1-TGA-F and Irk1-TGA-R.

Patch clamp of Irk1-expressing S2-R+ cells

S2-R+ Drosophila cultured cells were obtained from the Drosophila Genomics Resource Center (stock #150) and cultured at 25 °C in Schneider’s medium (Thermo Fisher Cat #21720001) with 10% FBS (Thermo Fisher Cat #10082139). Cells were seeded at a density of 1.7 × 10⁶ cells/ml in 12 well dishes for 24 hours and resuspended in serum-free medium prior to transfection. Cells were transfected with 1 μg plasmid DNA (pAc5-Irk1^WT, pAc5-Irk1^V306A, pAc5-Irk1^WT-Fray^T206E, or pAc5-Irk1^V306A-Fray^T206E) in 100 μL Opti-MEM (Thermo Fisher Cat #31985088) using 2 μL of TransIT-Insect transfection reagent (Mirus Bio Cat #6104). Medium was replaced with serum-containing medium 5 hours after transfection. Irk1 activity was measured in GFP+ cells 48–72 hours after transfection. Cells were harvested by centrifugation and resuspended in fresh medium. They were then plated on cover slips coated with poly-L-lysine (Sigma P8920). Ruptured whole cell recordings were performed at room temperature in a bath solution containing (in mM) 135 KCl, 1 MgCl₂, 2 CaCl₂, 15 glucose, 10 HEPES, 15 sucrose, pH 7.4 with Tris. Patch pipettes were pulled from borosilicate glass capillaries (Sutter Instruments) and heat-polished to give input resistances of 2–3 megaohms. The pipette recording solution contained (in mM) 135 KCl, 1 MgCl₂, 2 ATP-Mg, 0.1 GTP-Na, 5 EGTA, 10 HEPES, pH 7.2 with Tris. Cells were held at 0 mV and stimulated for 400 ms with step pulses from −150 mV to +90 mV with 20 mV steps. Currents were recorded with an Axopatch 200B patch-clamp amplifier and Pulse software (Molecular Devices, Sunnyvale, CA). 0.5 mM Ba²⁺ inhibited inward currents with 135 mM K⁺, and washing with Ba²⁺-free 135 mM K⁺ recovered the currents. Ba²⁺-sensitive current was therefore analyzed. Data acquisition and analysis were performed using pClamp v.9.2 (Molecular Devices).

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical testing was performed using GraphPad Prism, version 9. Data sets were analyzed for normality using the D’Agostino & Pearson normality test. Normally distributed data were compared using t-test or ANOVA and non-normally distributed data were compared using Mann-Whitney or Kruskal-Wallis test. Multiple comparisons testing was performed as indicated in the figure legends or Table S3. p<0.05 was considered statistically significant. Number of flies or cells examined and statistical parameters are indicated in the figure legends or Tables S1–S4.

Supplementary Material

Data S1

Data S1. List of proteins with putative Fray binding sites. Related to STAR Methods and Figure 4. A) Drosophila proteins with binding sites identified using the Drosophila motif, [D/E/N/Q/S/T/Y]RFx[V/I]xxxx[D/E/G/P]. B) Drosophila proteins with binding sites identified using the Gagnon motif, [S/G/V]RFx[V/I]xx[I/V/T/S].

Highlights.

• Intracellular chloride oscillates in Drosophila sLN_v pacemaker neurons.

• The morning increase in chloride restrains activity of the WNK-Fray kinase cascade

• Fray activates an inwardly rectifying potassium channel, Irk1

• Chloride inhibition of the WNK-Fray-Irk1 pathway maintains normal circadian period