Molecular Tuning of Electroreception in Sharks and Skates Nicholas W

Letter https://doi.org/10.1038/s41586-018-0160-9

Molecular tuning of electroreception in sharks and skates Nicholas W. Bellono1,2,3*, Duncan B. Leitch1,3 & David Julius1*

Ancient cartilaginous vertebrates, such as sharks, skates and rays, antagonist (Fig. 1b, Extended Data Fig. 2d). Instead, the voltage-gated possess specialized electrosensory organs that detect weak electric KV channel inhibitor 4-aminopyridine (4-AP) blocked outward 1–4 fields and relay this information to the central nervous system . currents from shark electrosensory cells (IKV, Fig. 1b). IKV was selective Sharks exploit this sensory modality for predation, whereas skates for K+ and exhibited a relatively high voltage-activation threshold may also use it to detect signals from conspecifics5. Here we analyse (Fig. 1c, Extended Data Fig. 2e, f). Furthermore, we observed fast shark and skate electrosensory cells to determine whether discrete activation and deactivation kinetics, whereas voltage-dependent inac- physiological properties could contribute to behaviourally relevant tivation (similar to desensitization) in response to prolonged voltage sensory tuning. We show that sharks and skates use a similar low pulses was weak and slow, which results in a K+ conductance of undi- threshold voltage-gated calcium channel to initiate cellular activity minished current amplitude even after repeated activation–deactivation but use distinct potassium channels to modulate this activity. cycles (Fig. 1d, e, Extended Data Fig. 2g). Among voltage-gated K+ Electrosensory cells from sharks express specially adapted voltage- channels, transcripts that encode the pore-forming subunit of KV1.3 gated potassium channels that support large, repetitive membrane predominated in shark ampullary organs (together with several KV voltage spikes capable of driving near-maximal vesicular release auxiliary subunits) and co-localized with CaV1.3 in electrosensory cells from elaborate ribbon synapses. By contrast, skates use a calcium- (Fig. 1f, g, Extended Data Fig. 2h). Outside of ampullary organs, only a activated potassium channel to produce small, tunable membrane truncated form of KV1.3 that lacks an essential N-terminal tetrameriza- voltage oscillations that elicit stimulus-dependent vesicular tion domain was observed in the brain (Extended Data Fig. 2i). KV1.3 release. We propose that these sensory adaptations support expression was not detected in skate ampullary organs, and only shark amplified indiscriminate signal detection in sharks compared with electrosensory cells exhibited 4-AP-sensitive voltage-gated K+ currents selective frequency detection in skates, potentially reflecting the (Extended Data Fig. 3a, b). Furthermore, both shark and skate elec- electroreceptive requirements of these elasmobranch species. Our trosensory cells expressed BK transcripts that, when heterologously findings demonstrate how sensory systems adapt to suit the lifestyle expressed, produced channels with similar properties; however, func- or environmental niche of an animal through discrete molecular and tional BK currents were observed only in the latter (Extended Data biophysical modifications. Fig. 3). As such, BK channels do not contribute appreciably to the major Electrosensory cells from the little skate (Leucoraja erinacea) express K+ conductance in shark electrosensory cells, at least not under the 2+ specialized low threshold CaV1.3 voltage-gated calcium (Ca ) channels developmental or physiological conditions examined here. In summary, + and big-conductance potassium (K , BK) channels that functionally shark electrosensory cells express a specific IKV with voltage-dependent couple to produce cellular membrane voltage oscillations6,7. In elec- properties that support repetitive stimulation. trosensory cells from the chain catshark (Scyliorhinus retifer, Fig. 1a), Shark KV1.3 is 80% identical to the human orthologue, but its voltage we similarly observed voltage-activated inward calcium currents threshold for activation was shifted to more depolarized values 2+ (ICaV) that were sensitive to L-type voltage-gated Ca -channel (CaV) compared with human KV1.3 (Fig. 2a, Extended Data Fig. 4a). modulators, had a low voltage threshold for activation, had steep voltage- Furthermore, shark KV1.3 was activated at slightly slower rates and dependence, and had a slow inactivation profile that contributed to deactivated with rapid kinetics, requiring substantially less negative a large window-current across physiological membrane voltages voltage to return to the resting state compared with the human channel (Extended Data Fig. 1a–d, f). As with skates, the pore-forming α (Fig. 2b, c, Extended Data Fig. 4b, c). Shark KV1.3 inactivation was subunit of CaV1.3 was the predominant CaV channel subtype expressed slow and only weakly voltage-dependent compared to human KV1.3 in shark electrosensory (ampullary) organs (Extended Data Fig. 1e). (Fig. 2d, Extended Data Fig. 4d). Consequently, shark KV1.3 produced Furthermore, both skate and shark orthologues contain a charged motif a conductance that could be repetitively stimulated with undiminished within the S2–S3 region of the IV repeat domain that confers low volt- amplitude, whereas human KV1.3 quickly inactivated with repetitive 6 age threshold to skate CaV1.3 (Extended Data Fig. 1g). Shark ampul- voltage pulses (Fig. 2e). These biophysical properties resemble those of lary organs also expressed several CaV auxiliary subunits, and the ICaV native shark IKV, which also exhibited a comparable pharmacological current density and activation threshold exhibited in shark elec- profile (Extended Data Fig. 4e, f). One notable difference is that the trosensory cells was similar to that of skates (Extended Data Fig. 1h, i). deactivation kinetics of native IKV were faster than those of the cloned Our results therefore suggest that CaV1.3 mediates the major depolar- channel, particularly at positive voltages. As such, although KV1.3 izing current in both skate and shark electrosensory cells. forms the predominant K+ conductance in shark electrosensory cells, In skate electrosensory cells, Ca2+ influx activated outward K+ cur- additional regulatory mechanisms may be provided by auxiliary subu- rents at relatively negative potentials to occlude inward Ca2+ currents6 nits, signalling cascades or structural proteins to further enhance rapid (Extended Data Fig. 2a–c). In shark cells, however, we observed large K+ deactivation. currents that were activated at more positive voltages and did not affect The voltage-dependent properties of shark KV1.3 probably derive the amplitude of inward current, which suggests reduced functional from altered voltage-sensor domain movements, which we verified by interaction between Ca2+ and K+ currents (Extended Data Fig. 2a– comparing gating currents from modified non-conductive shark and + c). Indeed, K currents were not affected by an ICaV blocker or a BK human KV1.3. For these experiments, we analysed a human isoform that

1Department of Physiology, University of California, San Francisco, San Francisco, CA, USA. 2Present address: Department of Molecular and Cellular Biology, Harvard University, Cambridge, MA, 3 USA. These authors contributed equally: Nicholas W. Bellono, Duncan B. Leitch. *e-mail: [email protected]; [email protected]

a b c a Activation b Deactivation c K+ 200 pA Control 1,000 1.0 1.0 100 100 ms Cd2+ 600 IbTx 10 max 0.5

4-AP I/I 0.5 Shark (ms) pA/pF

200 W Human 1

0.0 Normalized I –80 80 0.0 0.1 –200 –100 –50 0 50 Control Cd2+ 4-AP Voltage (mV) –100 –50 050 –100–50 050 Voltage (mV) 50 ms Voltage (mV) Voltage (mV) d Deact. Act. e f g 30 d 1.0 e 100 Inactivation Shark 20 0.8 10

Cav1.3 peak 0.6 I 10 s 10 /

(ms) 0.4 10s W 1 Cav1.3 I

pression (FPKM) pression 0 Kv1.3 0.2 Human 0.1 Ex 1 Kv1.3 DAPI 0.0 5 ms –100 0 100 5 ms kcna3akcnqkcnb1kcnh5akcnh2a n Voltage (mV) Shark 1 s Huma Fig. 1 | Major K+ current in shark electrosensory cells. a, Profile of chain Fig. 2 | Properties of shark KV. a, Average G–V relationships from catshark (Scyliorhinus retifer). b, Representative KV currents elicited by currents measured at −30 mV after activating pulses at the indicated increasing voltage pulses from −100 mV were inhibited by the KV blocker + = − ± = ± 4-AP but not by the I blocker Cd2 . Average I–V relationship from voltage. For shark KV1.3: Va1/2 5.4 0.4 mV, Ka 7.6 0.4 mV; CaV = − ± = ± = peak currents. n = 5, P < 0.0001 for outward control currents versus 4-AP, for human KV1.3: Va1/2 30.7 0.5 mV, Ka 4.7 0.5 mV. n 10, < two-way analysis of variance (ANOVA) with post hoc Tukey test. The P 0.0001 for Va1/2, two-tailed Student’s t-test. b, Deactivation kinetics − BK channel antagonist IbTx did not affect currents. c, G–V relationships of KV currents at 30 mV normalized to maximal amplitude elicited at 40 mV. Arrows indicate when deactivation rates were measured. exhibited a half-maximal activation voltage (Va1/2) of −4.1 ± 0.8 mV Right, expanded view. c, Deactivation properties of shark (red) and with a slope factor (Ka) of 9.5 ± 0.7 mV. n = 11. d, IKV activation and deactivation kinetics. Values of the time constant τ are obtained from human (black) KV1.3 channels. τ values from single exponential fits of single exponential fits of activation in response to the indicated voltage deactivation at the indicated voltages after an activating prepulse of 40 mV. = < pulses (10-mV increments from −100 mV) or deactivation at the indicated n 6, P 0.0001 for comparison of orthologues across voltage pulses, two-way ANOVA with post hoc Bonferroni test. d, Normalized currents voltages after an activating prepulse of 40 mV. n = 6. e, IKV inactivation over a 10-s, 40-mV pulse (top, 57 ± 4% current remained) and cumulative showing inactivation in response to a 10-s, 40-mV pulse. Average current = < inactivation properties in response to repetitive 40-mV pulses delivered remaining at the end of a 10-s step. n 8, P 0.0001, two-tailed Student’s in 5-ms intervals (bottom, 99 ± 2% current remained). Scale bars: Top, t-test. e, Currents demonstrating the differences in cumulative inactivation 500 pA, 1 s; bottom, 500 pA, 500 ms. n = 8. f, Voltage-gated K+-channel in shark (red) and human (black) KV1.3 channels in response to 40-mV − α-subunit mRNA expression in shark ampullary organs. Bars represent pulses from 100 mV delivered in 5-ms intervals. Representative of = ± fragments per kilobase of exon per million fragments mapped (FPKM). n 12. All data are represented as mean s.e.m, n denotes cells. g, Co-localization of CaV1.3 (red) and KV1.3 (green) transcripts within catshark ampullary organs. Nuclei were stained with 4ʹ,6-diamidino-2- substituted just the voltage-sensor-domain region (S1–S4) and found μ = phenylindole (DAPI) (blue). Scale bar, 100 m. Representative of n 4. that activation threshold and deactivation kinetics were greatly affected, All data are represented as mean ± s.e.m, n denotes cells or tissue sections. whereas inactivation remained similar (Extended Data Fig. 7a, b). By contrast, only inactivation was affected by replacing S5–S6 (Extended exhibits increased surface expression and therefore enhanced gating Data Fig. 7c–e), consistent with a role for the outer pore in C-type inac- 8 10 current amplitude (Extended Data Fig. 4g). Similar to ion (permeating) tivation . Therefore, specific structural adaptations in the KV1.3 trans- currents, upward (activating) voltage sensor movements (repre- membrane core specify physiologically relevant biophysical properties. sented as ON gating charge, QON) for shark KV1.3 exhibited a higher Membrane voltage (Vm) oscillations within ampullae control neuro voltage-activation threshold and slower kinetics compared with human transmitter release from electrosensory cells onto afferent nerves, 2,3 KV1.3 (Extended Data Fig. 5a–c). Moreover, shark KV1.3 gating-current thereby shaping signals to the central nervous system . In skate electro deactivation (QOFF; return of voltage sensors to a resting state after sensory cells, we found that functional coupling of CaV1.3 and BK depolarizing pulses) required less negative voltage and was accelerated mediates Vm oscillations that are tuned to low frequencies, such as those (Extended Data Fig. 5d–h). Consequently, ion tail-current deactiva- detected by behaving animals6. Consistent with our previous results, tion, which represents closure of the channel pore, was faster in shark skate electrosensory cells had a resting Vm of −54 mV, near the peak of KV1.3 than in human KV1.3 (Extended Data Fig. 5i). We next asked why the ICaV window current, at which cells exhibited spontaneous voltage shark KV1.3 appears to favour a resting state. When we recorded QOFF oscillations (Fig. 3a, b). Injecting current to bring the skate cell Vm to after a series of depolarizing voltage pulses of varying lengths, human various potentials modulated oscillatory behaviour, markedly changing KV1.3 deactivation kinetics markedly slowed after pulses longer than both frequency and amplitude across physiological membrane poten- 1 ms, whereas shark channel deactivation rates remained relatively fast tials (Fig. 3a, b). By contrast, shark electrosensory cells had an average and constant with increasing pulse lengths (Extended Data Fig. 6a–d). resting Vm of −66 mV, which is on the cusp of the ICaV window current This slowing in deactivation is characteristic of voltage-sensor- at which cells were relatively quiet (Fig. 3a, b). Injecting current to domain ‘relaxation’, which has been proposed to slow the closure of KV bring Vm to within the range of the ICaV threshold and window current 9 channels . We therefore propose that reduced voltage-sensor relaxa- elicited robust, repetitive Vm spiking that lasted for the duration tion in shark KV1.3 results in decreased stability of the open state; as of the recordings. Spiking only slightly decreased in amplitude and such, much less negative voltage is required to return channels to a frequency at more positive voltages (Fig. 3a, b), because spike ampli- resting state and mediate fast channel closure (depicted in our model, tude is probably determined by the voltage threshold of IKV, in addition Extended Data Fig. 6e). to ICaV window current. Indeed, the ICaV inhibitor nifedipine blocked We analysed shark–human KV1.3 chimaeras to see if specific evoked depolarization and 4-AP prevented or greatly slowed repolar- domains specify relevant biophysical attributes. Replacement of the ization (Fig. 3c). As such, in both shark and skate, activity is limited S1–S6 region of human KV1.3 with that of shark recapitulated the high to membrane voltages in which ICaV window current is observed, but voltage-activation threshold, rapid deactivation kinetics and weak inac- the dynamics in the two species are markedly different: the system tivation of wild-type shark KV1.3 channels, and the converse chimaera behaves as an all-or-none ON/OFF switch in sharks, but it is more also altered channel properties (Extended Data Fig. 7a, b). We next tunable in skates.

a b implicates IKV—while also revealing a resurgent ICaV upon hyperpolar- V SharkShark SSkatekate Vrest: Shark Skate m ization that could contribute to the initiation of the following voltage

–80 mV) 40 1.0 spike (Fig. 3d). Therefore, the low voltage threshold and inactivation ( de de properties of ICaV, coupled with the high voltage threshold, rapid deacti- u 20 0.5

–70 t

pli vation and weak inactivation of IKV, are suited to cooperatively mediate 0 0.0

Am Vm spiking in shark electrosensory cells. Notably, transcripts for 2+

–60 Ca -binding proteins and pumps were enriched in shark ampullary V Ca 2+ 1.2 1.0 I organs, which may help to facilitate repetitive Ca influx (Extended 0.6 0.5 Data Fig. 8a). Together, these cellular properties could support robust

–50 repetitive activity in shark electrosensory cells to amplify responses to Window Window

0.0 0.0 incoming electrical signals. By contrast, skate cells exhibit low-level Norm. amplitude Norm. –40 tonic activity at rest that could be retimed or modulated in response 14 1.0 to particular incoming electrical frequencies to alter neurotransmitter release. –30 7 0.5 To determine how synaptic vesicle release is affected by differences –20

15 mV 0 0.0 in Vm activity, we monitored membrane capacitance (Cm) to measure equency (Hz) equency Fr –70–50 –30 100 ms vesicle fusion in response to electrical stimuli. Depolarization of shark Voltage (mV) or skate electrosensory cells elicited inward currents and capacitance c 15 2+

V) 50 Control Nifedipine 4-AP changes that were blocked by Cd , which indicates that ICaV is required 40 (Hz) for vesicular release (Fig. 4a). Increasing the stimulus duration increased 30 10

changes in Cm (Fig. 4b, Extended Data Fig. 9a), but with distinct itude (m itude 15 mV l 20 5

500 ms 10 dynamics in shark compared with skate electrosensory cells: brief volt-

y Frequenc Amp 0 0 age stimuli induced larger changes in shark-cell Cm that saturated in response to longer stimuli, whereas skate cells exhibited Cm changes 4-AP 4-AP –68 mV Control Control that increased linearly with the duration of the stimulus (Fig. 4b). Nifedipine Nifedipine d V Differences in the number or distribution of synaptic vesicles could m Control4-AP account for these distinct vesicle release dynamics. Skate electrosensory cells contain large synaptic ribbons that tether numerous vesicles for 11,12 100 pA I release onto postsynaptic afferents . Indeed, shark and skate ampullae 250 ms expressed transcripts associated with ribbon synapses, and ultras-

25 pA tructural analysis of electrosensory cells revealed that both contain 50 ms remarkably long synaptic ribbons compared to those from mammalian hair cells13 (Fig. 4c, d, Extended Data Fig. 8b, c). Moreover, shark and Fig. 3 | Voltage dynamics in electrosensory cells. a, V -dependent m skate ribbons were similarly shaped and tethered an equivalent number spiking in shark electrosensory cells and smaller voltage oscillations from skate cells at the indicated membrane potentials (dotted lines) achieved of vesicles of comparable diameter. However, shark electrosensory cells by current injection. b, Average voltage oscillation amplitude, normalized had a larger ‘readily releasable’ vesicle pool (Extended Data Fig. 8d), 14,15 amplitude, and frequency at various membrane voltages for shark (red) which facilitates rapid and efficient exocytosis in other systems and and skate (blue) electrosensory cells. Values for oscillations from skate cells might account for our observation that short voltage stimuli induce were significantly different for voltages at which activity was observed, larger changes in Cm in electrosensory cells from the shark. Conversely, whereas shark-cell activity changed only slightly at more depolarized skate electrosensory cells contained more free cytosolic vesicles that, by voltages. n = 5, P < 0.01 for skate at −40, −50, −60 mV versus all other analogy with other systems14, may represent a larger ‘refilling pool’ of voltages, two-way ANOVA with post hoc Tukey test. The mean ± s.e.m. recently generated vesicles available for tethering and release (Extended − ± for resting membrane voltages (Vrest) for shark (red, 66.2 2.7 mV) Data Fig. 8d). Thus, skate cells may be better equipped to continuously and skate (blue, −54 ± 1.8 mV) are indicated as bars on each graph. n = 5 for skate and 10 for shark, P < 0.01 two-tailed Student’s t-test. supply vesicles for tonic release, reflected by the non-saturating, linear c, Brief current injection (2 pA, 5 ms, at arrow) from −68 mV elicited change in Cm in response to increasing the voltage of the stimulus. repetitive spiking that was inhibited by nifedipine. Brief injection elicited To determine the number of vesicles released in response to a single + sustained depolarization in the presence of 4-AP. n = 5, P < 0.001 for Vm spike or oscillation, we first integrated ICaV (QCa2 ) elicited by stimuli + amplitude of control versus nifedipine and for control versus nifedipine of varying duration, thereby establishing a relationship between QCa2 or 4-AP for frequency, one-way ANOVA. d, Left, simulated repetitive and Cm (Fig. 4e). Shark and skate cells responded equally to identical − − spiking voltage-clamp protocol (Vm, 65 mV to 15 mV) elicited voltage stimuli, further suggesting that ICaV is similar in both cell types currents that do not inactivate (I); right, depolarizing voltage increased and that K+-channel identity dictates the oscillation phenotype and inward currents until a voltage threshold was reached that activated a resulting amplitude of I (Extended Data Fig. 9b–d). We next fit Q 2+ large outward current (shaded grey area), which rapidly deactivated with CaV Ca induced by simulated V spikes or oscillations to 2+–C relation- hyperpolarizing voltage. The cycle repeats in the next simulated spike m QCa m protocol. The outward current was blocked by 4-AP, revealing a resurgent ships, and used the specific capacitance for a pure lipid membrane with inward current in the hyperpolarizing phase. Representative of n = 4. diameter equal to that of an electrosensory cell vesicle to estimate fusion 16 All data are represented as mean ± s.e.m, n denotes cells. events from measured changes in Cm . Notably, our calculations suggest that shark cells released at least ten times more vesicles in response to one Vm spike compared to skate cells subjected to a single To further assess how the kinetics and voltage dependence of ICaV oscillation event (Fig. 4f). Furthermore, as with other ribbon synapses, and IKV contribute to Vm spiking dynamics, we recorded currents in the large storage pool could facilitate a sustained release of vesicles in 14 response to simulated Vm spikes in the voltage-clamp mode. Repetitive response to repetitive Vm spiking, further amplifying the signals . simulated spikes induced inward and subsequent outward currents Taken together, these properties should render sharks acutely sensitive that rapidly activated and deactivated but did not inactivate (Fig. 3d). to incoming signals by greatly amplifying vesicular release to even very Inward ICaV was small at a resting potential of −65 mV and increased brief stimuli. By contrast, skate cells may better encode stimulus vari- with positive voltage until a high voltage threshold, which triggered an ation with tunable voltage oscillations and more graded vesicle release. outward current that rapidly deactivated when the voltage was returned We next asked how differences in cellular dynamics might contribute to resting levels (Fig. 3d). 4-AP blocked this outward current—which to sensation at the organismal level. Elasmobranchs preferentially

a Shark Skate b Shark Skate respond to low frequency electrical signals produced by their prey, but direct comparison of frequency selectivity between species is confounded Vm 200 5 2+ by behavioural and physiological differences . We therefore measured Cd (fF) m 100 changes in ventilatory rate as a basic, time-locked physiologic metric that

I Δ 25 pA C 0 is well-validated in electroreceptive elasmobranch species and readily 200 ms 0 1,000 2,000 observed in response to sensory cues, such as weak electrical stimuli or 60 odorants17,18 (Supplementary Videos 1 and 2). When presented with elec-

40 ) C 50 fF (fF trical stimuli of identical strength, the ventilation rates of sharks increased m m 20

C similarly at all stimulus frequencies (Fig. 4g). By contrast, the ventilation Δ 0 rate of skates maximally increased at low frequencies, resembling voltage Gm 1 nS 050 100 150 200 Time (ms) oscillations in their electrosensory cells and signals emitted by their electric organ (Fig. 4g). In both species, maximal electrically induced cdShark Skate 5 ventilatory responses were similar to those evoked by food odorants as 4 a comparative control (Fig. 4g). As such, shark electroreception may act 3 as a threshold detector for broad frequencies, potentially reflecting its 2 role in predation. By contrast, skate sensation appears more specifically 1 tuned to enable the detection of signals from prey as well as frequencies 5 SC 0 in the range of conspecific electric-organ discharges . Ribbon length ( μ m) Electroreception has independently evolved in many taxa to facilitate AF InnerOuter SharkSkate TypeType 1 2 particular behaviours ranging from predation to communication. By Mammalian hair cells analysing related species that use electroreception for distinct purposes, e Shark f g we found that subtle molecular variations considerably alter cellular Skate Shark Skate 75 1,500 0.6 properties that could ultimately mediate differences in behaviour. Our results suggest that molecular tuning of Vm oscillations in electrosen-

50 1,000 0.4 sory cells is important for the initial detection and discrimination

(fF) m

25 500 0.2 of salient electrical signals, although anatomical characteristics and

Δ C

Normalized processing by the central nervous system probably contribute to addi-

0 released Vesicles 0 0.0 19

ventilation response tional signal filtering (Extended Data Fig. 10). This observation is 010203040 Q 2+ (pC) 2 Hz5 Hz reminiscent of other sensory modalities in which sensory cells or their Ca SharkSkate Basal 10 Hz25 Hz50 Hz Food 100 Hz150 Hz200 Hz receptors are modified to mediate the detection of relevant stimuli. Fig. 4 | Tuning of electrosensory cell vesicular release and For example, expression and regulation of ion channels enable hair electrosensation. a, Representative capacitance measurements. A sine cells—which are developmentally related to electrosensory cells—to wave was applied before and after a 200-ms, −20-mV pulse to activate ICaV produce Vm oscillations of various amplitudes and frequencies to in shark (red) or skate (blue) electrosensory cells (Vm). ICaV (I) and 20,21 2+ mediate detection of particular auditory signals . Although it is not capacitance (Cm) changes were blocked by Cd (purple). Membrane known how broadly Vm oscillation tuning applies to electroreception, conductance (Gm) was constant in all quantified data. Representative of the electrosensory organs of paddlefish also express transcripts for Ca n = 4. b, Top, average capacitance changes in response to various durations V and K+ channels, which suggests that similar mechanisms may produce of voltage stimuli. The response relationship from shark electrosensory 22,23 cells (red) saturated in response to prolonged stimuli, whereas capacitance Vm oscillations in these systems . Furthermore, weakly electric fish changes in skate cells (blue) increased in a nearly linear fashion with use oscillating or spiking electrosensory organs to facilitate conspe- 24 increasing duration of the voltage stimulus. Bottom, brief stimuli elicited cific communication . Our results demonstrate one mechanism by larger capacitance changes in shark electrosensory cells compared with which K+-channel modification shapes electrosensory cell activity, but skate cells. The relationship between capacitance and duration of the differential regulation of K+ channels or other transduction compo- voltage stimulus in shark cells was best fit by an exponential relationship nents could provide alternative tuning mechanisms in distinct species that plateaued at 51fF with a time constant of 42 ms, whereas the −1 or under specific developmental or physiological states to facilitate relationship in skate cells was linear with a slope of 0.15 ± 0.01fF ms . dynamic electroreceptive behaviours5,25,26. n = 6. c, Micrographs showing shark and skate ribbon synapses. The electrosensory cell (SC), afferent nerve (AF), synapse (orange arrowhead) and synaptic ribbon (green arrowhead) are indicated. Scale bars, 500 nm. Online content d, Average ribbon length was similar in shark (red) and skate (blue) Any Methods, including any statements of data availability and Nature Research electrosensory cells, and is much longer than those from homologous reporting summaries, along with any additional references and Source Data files, mammalian hair cells13. n = 21. e, Average capacitance change elicited by are available in the online version of the paper at https://doi.org/10.1038/s41586- 018-0160-9. + + integrated ICaV (QCa2 ) revealed that less QCa2 is required to elicit large capacitance changes in shark electrosensory cells (red) compared with Received: 6 October 2017; Accepted: 4 April 2018; skate cells (blue). The relationship between capacitance and 2+ was QCa Published online xx xx xxxx. exponential for shark and linear for skate. n = 6. f, Average number of calculated vesicles released per spike or oscillation based on the 1. Josberger, E. E. et al. Proton conductivity in ampullae of Lorenzini jelly. Sci. Adv. + relationship between capacitance and QCa2 and simulated voltage 2, e1600112 (2016). + = < oscillation-induced QCa2 . n 6, P 0.0001, two-tailed Student’s t-test. 2. Clusin, W. T. & Bennett, M. V. The oscillatory responses of skate electroreceptors g, Normalized ventilatory responses elicited from live sharks or skates in to small voltage stimuli. J. Gen. Physiol. 73, 685–702 (1979). response to 50-µV electric stimuli at indicated frequencies. Food odorants 3. Teeter, J. H. & Bennett, M. V. L. Synaptic transmission in the ampullary were used as a positive control. Shark responses to electrical frequencies electroreceptor of the transparent catfsh, Kryptopterus. J. Comp. Physiol. 142, 371–377 (1981). were largely comparable, regardless of stimulus frequency. Skate peak 4. Kalmijn, A. J. The electric sense of sharks and rays. J. Exp. Biol. 55, 371–383 responses to low frequencies were significantly different from those at (1971). other frequencies but were comparable to odorant-elicited responses 5. Sisneros, J. A. & Tricas, T. C. Neuroethology and life history adaptations of the (n = 10, P < 0.0001 for 5, 10, 25 Hz versus other frequencies, one-way elasmobranch electric sense. J. Physiol. Paris 96, 379–389 (2002). ANOVA with multiple comparisons). All data are represented as 6. Bellono, N. W., Leitch, D. B. & Julius, D. Molecular basis of ancestral vertebrate ± electroreception. Nature 543, 391–396 (2017). mean s.e.m, n denotes cells, synaptic ribbons, or behavioural trials. 7. King, B. L., Shi, L. F., Kao, P. & Clusin, W. T. Calcium activated K+ channels in the electroreceptor of the skate confrmed by cloning. Details of subunits and splicing. Gene 578, 63–73 (2016).

8. Kubota, T., Correa, A. M. & Bezanilla, F. Mechanism of functional interaction 25. Meyer, J. H. & Zakon, H. H. Androgens alter the tuning of electroreceptors. between potassium channel Kv1.3 and sodium channel NavBeta1 subunit. Sci. Science 217, 635–637 (1982). Rep. 7, 45310 (2017). 26. Sisneros, J. A., Tricas, T. C. & Luer, C. A. Response properties and biological 9. Labro, A. J., Lacroix, J. J., Villalba-Galea, C. A., Snyders, D. J. & Bezanilla, F. function of the skate electrosensory system during ontogeny. J. Comp. Physiol. A Molecular mechanism for depolarization-induced modulation of Kv channel 183, 87–99 (1998). closure. J. Gen. Physiol. 140, 481–493 (2012). 10. Hoshi, T. & Armstrong, C. M. C-type inactivation of voltage-gated K+ channels: Acknowledgements We thank S. Bennett from the Marine Biological Laboratory pore constriction or dilation? J. Gen. Physiol. 141, 151–160 (2013). for supplying animals, J. Wong from the Gladstone/University of California, 11. Sejnowski, T. J. & Yodlowski, M. L. A freeze-fracture study of the skate San Francisco (UCSF) transmission electron microscopy core for performing electroreceptor. J. Neurocytol. 11, 897–912 (1982). electron microscopy, A. Zimmerman for help with capacitance measurements, 12. Fields, R. D. & Ellisman, M. H. Synaptic morphology and diferences in discussion and reading of the manuscript, R. Edwards for discussion, and R. sensitivity. Science 228, 197–199 (1985). Nicoll for input on the manuscript. This work was supported by a National 13. Nouvian, R., Beutner, D., Parsons, T. D. & Moser, T. Structure and function of the Institutes of Health (NIH) Institutional Research Service Award to the UCSF CVRI hair cell ribbon synapse. J. Membr. Biol. 209, 153–165 (2006). (T32HL007731 to N.W.B.), a Howard Hughes Medical Institute Fellowship of the 14. Matthews, G. & Fuchs, P. The diverse roles of ribbon synapses in sensory Life Sciences Research Foundation (N.W.B.), a Simons Foundation Postdoctoral neurotransmission. Nat. Rev. Neurosci. 11, 812–822 (2010). Fellowship to the Jane Coffin Childs Memorial Fund (D.B.L.) and grants from the 15. Graydon, C. W., Cho, S., Li, G. L., Kachar, B. & von Gersdorf, H. Sharp Ca2+ NIH (1K99DC016658 to D.B.L., K99DK115879 to N.W.B., and NS055299 and nanodomains beneath the ribbon promote highly synchronous multivesicular NS105038 to D.J.). release at hair cell synapses. J. Neurosci. 31, 16637–16650 (2011). 16. Hille, B. Ion channels of excitable membranes 3rd edn (Sinauer, Sunderland, Reviewer information Nature thanks B. Carlson, C. Lingle, H. von Gersdorff and 2001). the other anonymous reviewer(s) for their contribution to the peer review of this 17. Peters, R. C. & Evers, H. P. Frequency selectivity in the ampullary system of work. an elasmobranch fsh (Scyliorhinus canicula). J. Exp. Biol. 118, 99–109 (1985). Author contributions N.W.B. designed and performed electrophysiological 18. Kempster, R. M., Hart, N. S. & Collin, S. P. Survival of the stillest: predator studies, D.B.L. designed and performed gene expression, anatomical and avoidance in shark embryos. PLoS ONE 8, e52551 (2013). behavioural studies, and N.W.B., D.B.L. and D.J. wrote the manuscript. 19. Montgomery, J. C. & Bodznick, D. Signals and noise in the elasmobranch electrosensory system. J. Exp. Biol. 202, 1349–1355 (1999). Competing interests The authors declare no competing interests. 20. Fettiplace, R. & Fuchs, P. A. Mechanisms of hair cell tuning. Annu. Rev. Physiol. 61, 809–834 (1999). Additional information 21. Rutherford, M. A. & Roberts, W. M. Spikes and membrane potential oscillations Extended data is available for this paper at https://doi.org/10.1038/s41586- in hair cells generate periodic aferent activity in the frog sacculus. J. Neurosci. 018-0160-9. 29, 10025–10037 (2009). Supplementary information is available for this paper at https://doi. 22. Modrell, M. S. et al. Insights into electrosensory organ development, physiology org/10.1038/s41586-018-0160-9. and evolution from a lateral line-enriched transcriptome. eLife 6, e24197 Reprints and permissions information is available at http://www.nature.com/ (2017). reprints. 23. Neiman, A. B. & Russell, D. F. Two distinct types of noisy oscillators in Correspondence and requests for materials should be addressed to N.W.B. or electroreceptors of paddlefsh. J. Neurophysiol. 92, 492–509 (2004). D.J. 24. Baker, C. A., Huck, K. R. & Carlson, B. A. Peripheral sensory coding through Publisher’s note: Springer Nature remains neutral with regard to jurisdictional oscillatory synchrony in weakly electric fsh. eLife 4, e08163 (2015). claims in published maps and institutional affiliations.

Methods solution and diluted before use. Unless stated otherwise, the following concen- Animals and cells. Male and female chain catsharks (Scyliorhinus retifer) and little trations were used: 100 or 500 μM Cd2+, 1 μM Bay K, 10 μM nifedipine, 5 μM skates (Leucoraja erinacea) were provided by the Marine Biological Laboratory mibefradil, 1 mM 4-AP, 100 nM iberiotoxin, 10 mM TEA+, 10 μM NS11021, 100 (Woods Hole, MA, USA) and their use was approved by the UCSF Animal Care μM quinidine, 25 nM α-dendrotoxin, 10 nM margatoxin, 1 μM UK78282, 25 and Use Committee. Animals were euthanized with tricaine methanesulfonate nM Guangxitoxin-1E, 20 μM XE991. Pharmacological effects were quantified by (MS222, 1 g l−1). Ampullary organs of adult animals were removed from the hyoid differences in normalized peak current from the same cell after bath application cluster (skates) or buccal and supraorbital clusters (sharks) on ice and further of the drug (Itreatment/Icontrol). dissected by removing most of the canals and afferent nerve fibres. Ampullae Unless stated otherwise, ICaV was measured in response to 200-ms voltage pulses were treated with papain for less than 5 min and then electrosensory cells were in 10-mV increments from a −100 mV holding potential. G–V relationships were mechanically dissociated over the recording chamber. Isolated electrosensory derived from I–V curves by calculating G according to the following: G = ICa/(Vm − Erev), cells were identified by the presence of their single kinocilium. HEK293T cells and fit with a Boltzmann equation. Voltage-dependent inactivation was measured (American Type Culture Collection, ATCC) were grown in DMEM, 10% fetal calf during −20-mV voltage pulses after a series of 2-s prepulses ranging from −100 serum and 1% penicillin/streptomycin at 37 °C in 5% CO2. Cells were transfected to 60 mV in 10-mV increments. Voltage-dependent inactivation was quantified with Lipofectamine 2000 (Invitrogen) according to the manufacturer’s protocol. as I/Imax, with Imax occurring at the activating voltage pulse after a −100-mV Cell lines were verified from the ATCC but were not further tested for identity prepulse. IK was measured in response to 200-ms voltage pulses in 10-mV incre- or mycoplasma contamination during our studies. 100 ng of KV1.3 or 1 μg of ments from a −100-mV holding potential. G–V relationships were established non-conducting KV1.3 or BK constructs were co-expressed with 0.3 μg GFP. Mock from normalized tail currents measured at −30 mV after 50-ms voltage pulses transfection experiments (0.3 μg GFP) were performed as controls, in which min- in 10-mV increments from a −100-mV holding potential. Voltage-dependent imal voltage-activated outward current was observed. inactivation was measured during 40-mV voltage pulses after a series of 5-s Whole mount preparations. Juvenile fish were euthanized with an overdose of prepulses ranging from −100 to 60 mV in 10-mV increments. Cumulative inacti- MS-222 in artificial seawater and fixed in 4% paraformaldehyde for at least 24 h. vation was measured in response to 50-ms, 40-mV pulses every 5 ms. Activation The cartilage matrix and electroreceptor tubules were stained using Alcian Blue kinetics were determined by fitting the initial rising phase of currents activated by (20 mg Alcian Blue 8GX in 30 ml glacial acetic acid and 70 ml 100% ethanol) and various voltages with single exponentials. Deactivation kinetics were fit with a sin- bone was stained using Alizarin Red following previously published methods27. gle exponential upon repolarization to various voltages in 10-mV increments from Molecular biology. kcna3a and kcnma1a from shark ampullary organs were syn- 40 mV to −100 mV after 40-mV prepulses of the indicated durations. The rever- thesized by Genscript. Human KCNA3 was from Genscript, skate BK was from sal potential for IKV was measured by plotting tail current–voltage relationships the Julius laboratory, and mouse Kcnma1 was a gift from L. Salkoff (Addgene using a similar protocol that stepped in 10-mV increments from 40 mV to plasmid 16195). Chimaera synthesis and mutagenesis were carried out and veri- −120 mV. Single-channel currents were measured from the middle of the noise fied by Genscript or by the QuikChange Lightning site-directed mutagenesis kit band between closed and open states or calculated from the difference between (Agilent Genomics). Gaussian-fitted closed and open peaks on all-points amplitude histograms for Electrophysiology. Recordings were carried out at room temperature using a each excised patch record. Conductance was calculated from the linear slope of MultiClamp 700B amplifier (Axon Instruments) and digitized using a Digidata I–V relationships. In current-clamp mode, current injection was used to bring 1322A (Axon Instruments) interface and pClamp software (Axon Instruments). membrane potential to various values and then was fixed. In this case, membrane Capacitance and associated ion current measurements were amplified and digi- potential was defined as the base of the spiking or oscillating activity. Alternatively, tized with an EPC10 amplifier in lock-in mode (HEKA) and Patchmaster soft- brief current injection was delivered to determine the effects of pharmacological ware (HEKA). Unless stated otherwise, whole-cell data were filtered at 1 kHz and inhibitors on spiking activity. sampled at 10 kHz, and single-channel data were filtered at 5 kHz and sampled QON and QOFF represents the integral of non-linear gating current measured at 50 kHz. Data were leak subtracted online using a P/4 protocol, except for data during and after voltage pulses from holding potentials of −100 or 0 mV. QON was obtained using voltage ramp protocols, and membrane potentials were corrected quantified only from cells with no ionic current. ON gating-current kinetics were for liquid junction potentials. Electrosensory cell recordings were performed using quantified by single exponential fits of the slope of decreasing outward current borosilicate glass pipettes polished to 8–10 MΩ. For heterologous expression exper- elicited by voltage pulses in 10-mV increments from a −100 mV holding potential. iments in HEK293, recordings were performed using pipettes polished to 2–3 MΩ. OFF gating-current kinetics were calculated by single-exponential fits of the slope The extracellular solution was a modified ‘elasmobranch Ringer’s solution’ con- of increasing negative current elicited by voltage pulses in 10-mV increments from taining (in mM): 250 NaCl, 6 KCl, 4 CaCl2, 1 MgCl2, 10 glucose, 5 HEPES, 360 urea, a 0-mV holding potential. Voltage dependence of deactivation was also assessed pH 7.6. When analysing native K+ current properties, 500 μM Cd2+ was included by single-exponential fits of currents upon repolarization to various voltages in to block ICaV. Four intracellular solutions were used, as follows: for recording 10-mV increments from 20 mV to −100 mV after 40-mV prepulses of the indi- ICaV (in mM): 250 CsMeSO3, 1 MgCl2, 11 Cs-EGTA, 10 HEPES, 30 sucrose, 360 cated durations. Deactivation of gating currents was also measured at −100 mV urea, pH 7.6; for recording IK (in mM): 250 K-gluconate, 1 MgCl2, 11 K-EGTA, with exponential fits after a series of 40-mV voltage pulses of varying duration 10 HEPES, 30 sucrose, 360 urea, pH 7.6; for recording membrane potential from 0.5 ms to 30 ms. (in mM): 250 K-gluconate, 1 MgCl2, 1 K-EGTA, 10 HEPES, 20 sucrose, 2 MgATP, Capacitance measurements were performed using a 15-mV, 1.5-kHz sinusoidal 360 urea, pH 7.6; for recording capacitance changes (in mM): 250 CsMeSO3, stimulation protocol applied from −90 mV before and after depolarization pulses 1 MgCl2, 1 Cs-EGTA, 2 MgATP, 10 HEPES, 20 sucrose, 360 urea, pH 7.6. For of various lengths to acquire pre- and post-stimulus capacitance values. Cells were recording heterologous KV1.3 ionic current, the intracellular solution contained discarded when the series resistance (Rs) changed and exceeded the membrane (in mM): 145 K-gluconate, 5 KCl, 1 MgCl2, 5 K-EGTA, 10 HEPES, 10 sucrose, resistance (Rm) or if membrane conductance (Gm) varied greatly after depolarizing pH 7.2. The extracellular solution contained (in mM): 150 NaCl, 5 KCl, 2 CaCl2, voltage pulses. Whole-cell ion currents were filtered at 1 kHz and sampled at 2 MgCl2, 10 HEPES, 10 glucose, pH 7.4. For measuring KV1.3 gating currents, 10 kHz. Gating currents and capacitance measurements were filtered at 1 kHz and the intracellular solution contained (in mM): 150 NMDGMeSO3, 1 MgCl2, 10 sampled at 20 kHz. Capacitance records were filtered at 100 Hz during offline Cs-EGTA, 10 HEPES, 10 sucrose, pH 7.3. The extracellular solution contained analyses. Changes in capacitance were measured by averaging capacitance over (in mM): 150 TEACl, 1 CaCl2, 1 MgCl2, 10 HEPES, 10 glucose, pH 7.3. For gat- 200 ms after the depolarizing voltage step and subtracting the averaged capacitance ing current measurements, we used non-conducting shark (W407F) and human before depolarization. Intracellular ATP was included in all experiments and 28,29 + + (W436F) KV1.3 channels . A short isoform of human KV1.3, which exhibits 100 nM iberiotoxin, 1 mM 4-AP and intracellular Cs were used to block K 8 + identical gating properties but increased surface expression , was used to increase currents. The integral of ICaV (QCa2 ) was used to account for variability in the + the amplitude of the gating current, whereas the long isoform was used for most kinetics of ICaV. To calculate vesicle release, we fit QCa2 induced by simulated + ion current studies. For BK single-channel recordings, the intracellular solution spike- or oscillation-voltage protocols to QCa2 –Cm relationships. When identical + contained (in mM): 136 K-gluconate, 4 KCl, 1 K-EGTA, 1 HEDTA, 10 HEPES, 10 voltage protocols were used to elicit ICaV, QCa2 was the same from skate and shark glucose, pH 7.3. The extracellular solution contained (in mM): 136 K-gluconate, 4 electrosensory cells, consistent with the similar ICaV in these cells. + KCl, 1 MgCl2, 10 HEPES, 10 glucose, pH 7.3. Calculated concentrations of buffered Transcriptional sequencing and analysis. Poly A RNA was extracted from amp- Ca2+ added to intracellular solution were made using MaxChelator (C. Patton, ullary electrosensory cells, non-electroreceptor covered skin, muscle and fore- Stanford University). brain of an adult chain catshark then was reverse-transcribed using SuperScript The pharmacological inhibitors or agonists Bay K, nifedipine, mibefradil, III kit (Invitrogen). Sequencing libraries were prepared using the Illumina TruSeq 4-AP, XE991, NS11021 and quinidine were from Tocris. Iberiotoxin, α-dendro- Stranded mRNA Library Prep Kit according to the manufacturer’s instructions. toxin, margatoxin, UK78282 and Guangxitoxin-1E were from Alamone Labs. Libraries were sequenced on the Illumina Hi-Seq 4000 (V. C. Genomics Sequencing Compounds were dissolved in <1% vehicle (DMSO or water), which was used Laboratory, University of California, Berkeley) using 150 cycles of paired end reads, as a control. Ionic pore-blocker stocks were prepared in standard extracellular producing between 30 million and 40 million inserts for each sample.

Transcriptomes for each sample were assembled de novo using the Trinity suite Behavioural analysis. In an isolated location and under normal lighting condi- (version 2.1.0). Sequences were aligned to the zebrafish protein database (NCBI tions, individual juvenile skates (n = 6) and sharks (n = 5) of both sexes were assembly GRCz10) using the blastx tool from NCBI blast (version 2.2.31) using a allowed to freely move and habituate for 20 min in an ambient temperature, maximum E value of 1 × 10−5. Reciprocal blastx alignments (using zebrafish pro- seawater-filled cylindrical acrylic tank (diameter 28 cm). A sinusoidal electrical tein sequences that aligned to catshark sequences) were performed to the human stimulus (100 μA over 5 mm), generated by threading positive and negative ends protein database. Estimates of relative abundance for differential expression com- of tin-plated copper wire (300 VH, 22 gauge, NTE Electronics, Inc.) into seawater- parisons were performed using the RSEM software package within Trinity. These filled Tygon tubing powered by a function generator (Tone Generator Pro, values are reported as FPKM. Performance Audio) was randomly positioned and obscured by sand substrate In situ hybridization histochemistry. Adult chain catsharks were euthanized with in one of four circles (diameter 5.5 cm), all equally spaced from the centre of the an overdose of MS-222 in artificial seawater and transcardially perfused with PBS tank. After an initial recording of baseline ventilation frequency, individual fish followed by 4% paraformaldehyde. Ampullary organs were dissected from the buc- were stimulated at 2, 5, 10, 25, 50, 100, 150 and 200 Hz for 5 min, in randomized cal and supraorbital clusters and cryo-protected in 30% sucrose in PBS overnight. order. Each frequency was tested 10 times. All skates were exposed to a plume of Cryostat sections (15-μm thick) were probed with digoxigenin-labelled cRNA for Mysis shrimp odorant, and sharks were presented with squid odorant, to measure shark CaV1.3 and fluorescein-labelled cRNA for shark BK and KV1.3 receptors. responses to natural food stimuli. To prevent habituation to the stimuli, an interval Probes were generated by T7/T3 in vitro transcription reactions using a 500-nucle- of 20 min without electrical stimuli was used between each trial. A digital video otide fragment of CaV1.3 cDNA (nucleotides 3700 to 4200), a 325-nucleotide camera (Panasonic HC-V770) was positioned above the tank and used and measure fragment of BK cDNA (nucleotides 636 to 961) and a 470-nucleotide fragment ventilatory responses, as characterized by cyclical movement of spiracles or gills. of KV1.3 cDNA (nucleotides 670 to 1140). Hybridization was developed using Measurements were randomized and made blind to stimulation conditions. anti-digoxigenin and anti-fluorescein Fab fragments, followed by incubation with Statistical analysis. Data were analysed with Clampfit (Axon Instruments), Fast Red and streptavidin-conjugated Dylight 488 (to probe for BK) according to Patchmaster (HEKA) or Prism (Graphpad). Data are represented as mean ± s.e.m. published methods30. After hybridization and detection, sections were covered and n represents independent experiments for the number of cells in electro- with a coverslip and co-stained with DAPI as a nuclear marker (Prolong Gold physiology, quantified structures from histological analysis, or behavioural trials. Antifade Mountant with DAPI; Invitrogen). Data were considered significant if P < 0.05 using paired or unpaired two-tailed Transmission electron microscopy. Tissue samples from the catshark and skate Student’s t-tests or one- or two-way ANOVAs. All significance tests were justified hyoid capsule with electrosensory cells were fixed in 2% glutaraldehyde, 1% considering the experimental design and we assumed normal distribution and paraformaldehyde in 0.1 M sodium cacodylate buffer at pH 7.4, postfixed in 2% variance, as is common for similar experiments. Sample sizes were chosen on the osmium tetroxide in the same buffer, stained en bloc with 2% aqueous uranyl basis of the number of independent experiments required for statistical significance acetate, dehydrated in acetone, infiltrated and embedded in LX-112 resin (Ladd and technical feasibility. Research Industries). Semi-thin sections stained with toluidine blue were pre- Reporting summary. Further information on experimental design is available in pared to orient and locate the area of interest. Samples were ultrathin sectioned the Nature Research Reporting Summary linked to this paper. (typically 100 nm) on a Reichert Ultracut S ultramicrotome and counter-stained Data availability. Deep sequencing data are archived in the Gene Expression with 0.8% lead citrate. Grids were examined on a JEOL JEM-1230 transmission Omnibus under accession number GSE103977. The GenBank accession numbers electron microscope (JEOL USA, Inc.) and imaged with the Gatan Ultrascan 1000 for the α subunit are: CaV1.3 MF959522, KV1.3 MF959523 and BK MF959524. digital camera (Gatan Inc.). All measurements were performed in Image J (NIH) Other data are available from the corresponding author upon reasonable request. on electron micrographs adjusted for brightness and contrast (Photoshop CS6, Adobe Systems). Measurements of vesicles followed published methods31, with 27. Gillis, J. A., Dahn, R. D. & Shubin, N. H. Chondrogenesis and homology of the populations attached to the ribbon structure (‘attached’), in proximity to the syn- visceral skeleton in the little skate, Leucoraja erinacea (Chondrichthyes: Batoidea). J. Morphol. 270, 628–643 (2009). apse (‘readily releasable’) and freely filling cytosolic space (‘refilling’). 28. Sigg, D., Stefani, E. & Bezanilla, F. Gating current noise produced by To measure ribbon-shape variation, the difference between the traced distance elementary transitions in Shaker potassium channels. Science 264, 578–582 of the ribbon and the distance of a line drawn from the start to the end of the ribbon (1994). was divided by the distance of that line. All values represent a positive difference 29. Yang, Y., Yan, Y. & Sigworth, F. J. How does the W434F mutation block current in (increase in length) from this straight line. The angle of the ribbon to the surface Shaker potassium channels? J. Gen. Physiol. 109, 779–789 (1997). of the plasma membrane (‘angle from PM’) was measured using the angle tool in 30. Ishii, T., Omura, M. & Mombaerts, P. Protocols for two- and three-color fuorescent RNA in situ hybridization of the main and accessory olfactory ImageJ by selecting three points: a point on the ribbon about 150 nm from the syn- epithelia in mouse. J. Neurocytol. 33, 657–669 (2004). apse, a point at the juncture of the ribbon and the synaptic density on the plasma 31. Kantardzhieva, A., Liberman, M. C. & Sewell, W. F. Quantitative analysis of membrane, and a point located on the plasma membrane about 150 nm from the ribbons, vesicles, and cisterns at the cat inner hair cell synapse: correlations juncture with the ribbon, producing an acute angle. with spontaneous rate. J. Comp. Neurol. 521, 3260–3271 (2013).

a bc 2.5 + l

Cs 40 s a 2.0

/ 1.5

m 1.0

r -8080 t 0.5

I mV 0.0 + -40 ol 2 K r y pA/p F Cd 50pA cont Ba efradil 100ms nifedipinemib de I / Imax G / Gmax CaV expression 1.2 1.2 30

20 0.8 0.8 M

0.4 0.4 F 10

0.0 0.0 0 -80-40 .2 3 4 1 .2 .3 1 2 3 mV 1 1. 1. 2. 2 2 3. 3. 3.

f -30mV -90mV 2s

25pA % current remaining: 71.25 ± 7.47%

g IVS2 IVS3 shark MVLKLIAFKPMVCVPKKERWLGYFSDSWNVFDFLPKKER skate MVLKLIAFKPMVCVKKKERWLGYFSDSWNVFDFLKKKER human MVLKVIAFKPK------GYFSDAWNTFDSL

ICaV h CaV subunits i 20 0 0 15

M -20

K 10 (mV) P -30 1/ 2 (pA/pF ) F -40 Va 5 Ca V I 0 -60 -60 1 2 1 4 4 nb nb sharkskate c ac Ca acna2d C Cacna2dC Cacna2d

Extended Data Fig. 1 | Properties of shark ICaV. a, Top, isolated shark −70 mV and −40 mV. G–V relationships were established from current ampullary organs with attached canals and nerve fibres (scale bar, 100 μm); measurements during voltage pulses delivered in 10-mV increments bottom, a representative electrosensory cell patch-clamp experiment from −100 mV. Inactivation was measured at a −20 mV test pulse after (scale bar, 10 μm). b, Left, ICaV currents elicited by increasing voltage a series of voltage prepulses. n = 7. e, CaV α-subunit mRNA expression pulses from −100 mV; right, average current–voltage (I–V) relationship in shark ampullary organs. Bars represent FPKM. f, ICaV elicited by a (n = 7). c, ICaV exhibited an L-type CaV pharmacological profile: Peak 2-s depolarizing voltage step to −30 mV exhibits little inactivation. 2+ currents were regulated by Bay K (agonist), Cd (blocker) and nifedipine Representative of n = 5. g, CaV1.3 alignment revealed that the IVS2–S3 (antagonist), but not by mibefradil (T-type antagonist). n = 4. P < 0.0001 motif that confers low voltage threshold in skate CaV1.3 is conserved for control versus all treatments except mibefradil, one-way ANOVA with in the shark orthologue. h, Expression of CaV auxiliary subunits in post hoc Bonferroni test. d, ICaV conductance–voltage (G–V) relationship shark ampullary organs. Bars represent FPKM. i, Average ICaV current (black) with half-maximal activation voltage (Va1/2) of −54.6 ± 1.2 mV. density and voltage-activation threshold was similar in shark and skate Inactivation–voltage relationship (grey) with half-inactivation potential electrosensory cells. n = 6. All data are represented as mean ± s.e.m, (Vh1/2) of −62.9 ± 1.4 mV. A large window current was observed between n denotes cells.

a shark shark + skate + Cs K skate F F

p p

/ 100 /

F 100

F F F

A A

p p p

/ 100 100 1000 p 300 p

/ / /

A A A

p 50 p 750 p mV 200 p mV 50 -80 -80 500 -100 -100 100 -8080 250 -50 mV -8080 mV -8080 -8080 -100 -50 -250 mV -100 mV

bcd + K+ + Cs K 40 1000 100

60 60 )

)

F F

pA/p F

p /

/ control

/ A

A 2+

p (

p Cd (

p 50

( 30 30 500

( V

V mV

a V

a -80

K IbTx

C I

I 4-AP % 0 0 0 0 -40 k e rk te k te k e a a a har shar skat sh sk shar sk s skat

K+ Cd2+ e +40mV f g

1.0 -100mV 50

pA/p F

a m

E I

K / 0.5 I

-100 mV 50pA 0.0 -25 -100 0 100 3ms 200pA 500pA mV 50ms 1s

h i TM segments KV subunits 20 50 N123456C isoform 2: 40 15 4: M M 30

K K

P 10 P

F F 20 5 10 0 0 1 2 2 e le n g3 a c ki s s nab cng4 brain KcngKcn c KcngK mu K ampull

Extended Data Fig. 2 | Properties of shark IKV. a, Currents elicited by I–V plot). n = 5. Arrows indicate when currents were measured at the 500-ms voltage ramps in shark (red) or skate (blue) electrosensory cells indicated voltages after an activating prepulse of 40 mV (also shown + + 2+ in the presence of intracellular Cs (left) or K (right). The insets on the in expanded view). Extracellular Cd was included to block ICaV for + right show the inward ICaV in the presence of intracellular K . b, Average biophysical studies of IKV. f, IKV currents elicited by a voltage protocol ICaV current density elicited by voltage ramps was similar for both shark to obtain the G–V relationship. Arrows indicate when tail currents and skate cells in the presence of intracellular Cs+, but larger in shark were measured at −30 mV after voltage steps that increased in 10-mV cells in the presence of intracellular K+. Outward K+ current density increments from −100 mV (expanded view within inset). Representative was significantly larger in shark cells. n = 5, P < 0.0001 for shark versus of n = 11. g, Voltage-dependent inactivation properties of IKV. The arrow + skate ICaV or IK with intracellular K , two-tailed Student’s t-test. c, The indicates when inactivation was measured during 40-mV pulses after a + + percentage of ICaV remaining in the presence of K compared with Cs is series of prepulses that increased in 10-mV increments from −100 mV. markedly greater in shark cells compared with those from skate. d, Inward Vh1/2 was −5.5 ± 1.7 mV and inactivation was incomplete. n = 6. currents elicited by increasing voltage pulses from −100 mV were not h, mRNA expression of the KV channel auxiliary subunits in shark affected by IbTx or 4-AP, but were blocked by Cd2+. n = 5, P < 0.0001 ampullary organs. Bars represent FPKM. i, mRNA expression of kcna3 for inward control currents versus Cd2+, two-way ANOVA with post hoc isoforms. The major isoform studied is indicated in red. Other Tukey test. Peak inward currents were not affected by IbTx or 4-AP. low-expression ampullary isoforms were similar. The only isoform that e, Reversal potential for IKV in shark electrosensory cells is near the was appreciably expressed outside of ampullae was truncated and found in + reversal potential for selective K permeation (EK, blue arrow on the the brain (grey). All data are represented as mean ± s.e.m, n denotes cells.

a + b c K channel expression IK Cav1.3 100 BK shark 800 100 100 DAPI 80 skate

60 bloc k Cav1.3 400 50 50 FPKM 40 (pA/pF) K 20 I % 4-AP % IbTx bloc k 0 0 0 0 3 1 shark skate BK KV Kcna cnma K

d control e IbTx + K 4-AP NS11021 pore S6

p / 250 shark GYGDVYARTTLGRLFMVFFILGGLAMFARYVPEIAALR AA

control IbTx NS A p 200 skate GYGDVYARTTLGRLFMVFFILGGLAMFARYVPEIAALR AA 150 mouse GYGDVYAKTTLGRLFMVFFILGGLAMFASYVPEIIEL 100 human GYGDVYAKTTLGRLFMVFFILGGLAMFASYVPEIIELSY E 50 *******.******************** ***** * -8080 -50 mV 100pA 100ms

l 2500 a basal fgshark skate mouse sharkBK s

skate BK b

+ mouseBK 2 mV a

40 C 0

) 0 2+

2 2500

( Ca

30 t

20 c

a 0

1 2500

10 S NS

0 x

0 T 0

0 40 80 120 b

I 1 mV -5 0510 10pA 5pA i (pA) 20ms 100ms

Extended Data Fig. 3 | Properties of shark BK. a, mRNA expression the shark orthologue. f, Heterologously expressed shark and skate BK had + of major K -channel α-subunits (Kcna3, KV1.3) and (Kcnma1, BK) in relatively small single-channel conductance compared with mouse BK. shark and skate electrosensory cells. b, Average K+ current density, 4-AP- Shark BK = 109 ± 4 pS; skate BK = 105 ± 5 pS; mouse BK = 259 ± 12 pS. sensitive current (IKV), and IbTx-sensitive current (IBK) in shark and skate n = 5, P < 0.0001 for mouse versus shark or skate BK, two-way ANOVA electrosensory cells. n = 5, P < 0.0001 for shark versus skate cells for all with post hoc Tukey test. g, 1 μM Ca2+ increased open probability in shark comparisons, two-tailed Student’s t-test. c, Co-localization of CaV1.3 (red) BK expressed in excised inside-out patches, and 10 μM NS11021 increased and BK (green) transcripts within shark ampullary organs. Nuclei were open probability of channels in excised outside-out patches. NS11021 stained with DAPI (blue). Scale bar, 100 μm. Representative of n = 4. modulation was blocked by 100 nM IbTx. Holding voltage was 80 mV. 2+ d, In the presence of 4-AP, a relatively small outward current remained NPo: basal, 0.0061 ± 0.0014; Ca , 0.11 ± 0.011, NS11021, 0.24 ± 0.026. that was insensitive to IbTx and was slightly increased by NS11021 at very n = 5, P < 0.0001 for basal versus Ca2+ or NS11021, two-tailed Student’s positive voltages. n = 5, P < 0.05 for control versus NS11021 at 70 or 80 t-test. All data are represented as mean ± s.e.m, n denotes cells or tissue mV, two-way ANOVA with post hoc Tukey test. e, BK alignment revealed sections. that residues found to reduce conductance in skate BK are conserved in

a b c d shark human shark human shark human inactivation

100 1.2

0.8 )

s a

shark m

m 10 (

I/I human τ 0.4

0.0 1nA 1nA 1 1nA 5ms 050100 -1000100 1s 50ms 50ms mV 25ms mV e f g short long

control4-AP controlquinidine electrosensory sharkKv hKv1.3 isoform:

l l

e G-V inactivation deactivation

c y

r 1.0 1.2 100

o 1.0

l e

I I

s a

d d

b r

0.8 ) I

e e

s c

z z /

i i

l l

l m

500pA n

a 0.5 a ( 10 e

e 0.5

τ t

100ms m m a

r r e

r 0.4

o o

t I

N N

k r 0.0 0.0 0.0 1

h + -75-50 -250 25 -120-80 -400 40 -100 -50050 s P e n in 2 n 1 A n 8 mV mV mV TE 4-A toxi 1.5nA XE99 Quinidi UK782 100ms DendrotoxiMargatox Guagxi

Extended Data Fig. 4 | Properties of shark KV. a, Voltage-activated KV1.3 channels. The arrow indicates when inactivation was measured currents recorded in HEK293 cells expressing shark (red) or human during 40-mV pulses after a series of prepulses that increased in 10-mV (black) KV1.3. Arrows indicate when currents were measured at −30 mV increments from −100 mV. n = 9, P < 0.0001 for Vh1/2, two-tailed after voltage pulses that increased in 10-mV increments from −100 mV. Student’s t-test. e, IKV was reduced by 4-AP or quinidine in native shark The inset shows expansions of measured currents following arrows. electrosensory cells or heterologously expressed shark KV1.3. Currents b, Left, normalized currents elicited by a 40-mV voltage pulse were elicited by increasing voltage pulses from −90 mV (native) or −100 mV demonstrated that expressed shark KV channels open more slowly (heterologous). f, Pharmacological profile of shark electrosensory cell IKV compared with human orthologues. Right, average activation kinetics and heterologously expressed shark KV1.3. Currents measured at peak were slower for shark KV compared with human KV in response to voltages amplitude were reduced by 4-AP or quinidine, but not by other treatments. from 20 mV to 50 mV. n = 6, P < 0.0001 for contribution of orthologue Pharmacological modulation of native IKV and shark KV1.3 was similar. identity to series variance, two-way ANOVA with post hoc Bonferroni g, The short human isoform of KV1.3 (short N-terminal truncation) was test. c, Deactivation kinetics of normalized currents from shark and used to study gating currents because of enhanced expression, but channel 15 human KV channels at various repolarizing voltages after an activating properties are identical . Similarly, we found that activation threshold and prepulse of 40 mV. The arrow indicates when current properties were G–V relationship, voltage-dependent inactivation, and deactivation were measured during the voltage protocol. d, Inactivation properties (left) similar between long and short isoforms. n = 6. All data are represented as and average inactivation–voltage relationships (right) of shark (red, mean ± s.e.m, n denotes cells. Vh1/2 = 0.1 ± 2.8 mV) and human (black, Vh1/2 = −30.6 ± 0.9 mV)

a b c -100 mV HP shark human 1.0 100 shark

Q )

d Q 10 s

z m

i G ( l 0.5

τ N

m human O r 1

Q Q

N G 0.0 0.1 1nA -100 -500 50 0306090 mV mV 25ms

de100 f 0mV holding potential

shark human shark human

)

m (

τ 10

O Q

1 500pA 500pA 0 -8 500pA -120 -100 25ms 25ms 10ms mV

g 0mV holding h 0mV holding i ion current 1.0 100 shark human

OFF

) s

m (

0.5 τ 10

O Q

Normalized 0.0 1 1nA -160 -120 -80-40 0 40 -160 -120 -80 5ms mV mV

Extended Data Fig. 5 | Shark KV gating currents. a, Gating currents decreasing voltage pulses in increments of 10 mV from a holding potential recorded in HEK293 cells expressing non-conductive shark (red) or of 0 mV. Scale bars, 500 pA, 25 ms. g, Average charge (Q)–V relationships human (black) KV1.3 elicited by increasing voltage pulses in 10-mV of downward voltage sensor movement (QOFF) in response to decreasing increments from a −100-mV holding potential. b, Average voltage pulses from a holding potential of 0 mV. Shark KV1.3 Va1/2 was charge (Q)–V relationships. Shark KV1.3 Va1/2 was −33.4 ± 0.5 mV, −61.6 ± 1.9 mV and human KV1.3 Va1/2 was −110.9 ± 1.01 mV. n = 7, Ka = 6.1 ± 0.5 mV; and human KV1.3 Va1/2 was −54.25 ± 0.7 mV, P < 0.0001 for contribution of orthologue identity to series variance, Ka = 5.2 ± 0.6 mV. n = 9, P < 0.0001 for Va1/2 two-tailed Student’s t-test. two-way ANOVA with post hoc Bonferroni test. h, QOFF kinetics of voltage Dotted lines indicate associated G–V relationships. c, QON kinetics after sensor deactivation from a holding potential of 0 mV were significantly voltage sensor activation from a holding potential of −100 mV were faster in shark (red) compared with human (black). n = 7, P < 0.0001 for significantly slower in shark (red) compared with human (black). contribution of orthologue identity to series variance, two-way ANOVA n = 7, P < 0.0001 for contribution of orthologue identity to series with post hoc Bonferroni test. All data are represented as mean ± s.e.m. variance, two-way ANOVA with post hoc Bonferroni test. i, Ion tail currents (indicating channel closure) deactivated faster in shark d, Representative OFF gating current (QOFF) kinetics during repolarization (red) compared with human (black) KV1.3. Tail currents were measured at in 10-mV increments after a 40-mV prepulse. The arrow indicates when −100 mV after a series of activating voltage pulses that increased in deactivation rates were measured and purple traces show deactivation at 10-mV increments. Inset, arrows indicate when tail currents were measured −50 mV. e, QOFF kinetics were significantly faster in shark (red) compared after activating voltage pulses. Representative of n = 10. τ for deactivation with human (black). n = 11, P < 0.0001 for contribution of orthologue from 60-mV pulse: shark = 1.5 ± 0.1 ms; human = 5.5 ± 0.4 ms. identity to series variance, two-way ANOVA with post hoc Bonferroni P < 0.0001, two-tailed Student’s t-test. All data are represented as test. f, Gating currents from shark (red) or human (black) KV1.3 after mean ± s.e.m, n denotes cells.

ab10ms 1ms shark human shark human shark human

100 100 )

s ms m

( 25

τ 10 10 10

n o

i 5

t a

v 1 i

t 1 1

a e

1nA D 0.1 0.1 10ms -120 -100 -80 -120 -100 -80 mV mV

c shark human d 60

) 40

m ( τ 20

1nA 0 5ms 01530 ms e relaxed

restingactivated + + open +

+ + + stabilized open shark human

Extended Data Fig. 6 | Shark KV voltage sensor domain relaxation. in the slowing of QOFF with increasing pulse length. Deactivation was a, QOFF kinetics after either 10-ms or 1-ms activating prepulses of 40 mV. measured at −100 mV after a series of 40-mV voltage pulses of varying Purple traces indicate deactivation at −50 mV. Arrow indicates when duration from 0.5 ms to 30 ms. d, Average QOFF kinetics in response to current properties were measured during the voltage protocol. indicated voltage pulse lengths. n = 6, P < 0.0001 for comparison of b, Average QOFF kinetics were faster in shark compared with human orthologue QOFF kinetics after a 30-ms voltage pulse. e, Hypothetical during deactivation after 40-mV prepulses of 25, 10, or 5 ms duration, but model of shark and human KV1.3. Compared with its human orthologue, rates were similar after 1-ms activating prepulses. Kinetics were measured shark KV1.3 exhibits reduced voltage sensor domain relaxation, which at voltages that decreased in 10-mV increments from 40 mV. n = 9, stabilizes pore opening in human KV1.3. Reduced voltage sensor relaxation P < 0.0001 for contribution of orthologue identity to series variance at is indicated by dotted lines to suggest that this state (or states) may occur 25, 10, or 5 ms, but no significant difference was observed at 1 ms, two- to a lesser extent in the shark orthologue. Thus, compared with human way ANOVA with post hoc Bonferroni test. c, Shark KV1.3 QOFF kinetics KV1.3, the shark orthologue requires relatively less repolarizing voltage to were relatively unaffected by the duration of the activating voltage pulse, more quickly return to a resting/closed state. All data are represented as whereas human KV1.3 entered a proposed ‘relaxed’ state that resulted mean ± s.e.m, n denotes cells.

a shark human sS1-S6 sS1-S4 hS1-S6 hS1-S4

NS1S4S5S6C

shark human

2nA 50ms

1nA 2s

G-V deactivation inactivation b sS1S6 1.0 100 1.2

sS1S4 d

z 0.8 )

l s

hS1S6 a

m m

0.5 10 I ( m

r τ hS1S4 o 0.4

N human shark 0.0 1 0.0 -75-50 -250 25 -100 -500 50 -100 -500 50 mV mV mV

c d e shark S5-S6 human S5-S6 shark S5-S6 I 1.0 human S5-S6 d

1.2 e

shark a 0.5

human 0.8 o

a 0.0 m -100 -50050

I/I

0.4 100 )

s 10 m

0.0 (

-100 -500 50 τ 1nA1nA 1 mV 2nA 2s 2nA 2s -100 -50 050 50ms 50ms mV

Extended Data Fig. 7 | Shark–human KV chimaeric analyses. threshold and deactivation kinetics. Channels containing hS5–S6 exhibited a, Chimaeric shark–human KV1.3 channels reveal that shark KV S1–S6 the strongest voltage-dependent inactivation—nearly as efficient as confers differences in activation voltage threshold, deactivation kinetics wild-type human KV—whereas channels containing sS5–S6 displayed and inactivation. Top, the chimaera constructs analysed (shark in red, weaker inactivation. sS1–S4 had a smaller effect on inactivation. Vh1/2 for human in black). Middle, the arrow indicates when voltage-activated wild-type human = −34.6 ± 0.9 mV, wild-type shark = 0.1 ± 2.8 mV, currents were measured at −30 mV after a series of voltage pulses that hS1–S6 = −18.3 ± 0.4 mV, hS1–S4 = −12.3 ± 1.2 mV, sS1–S6 = increased in 10-mV increments from −100 mV. Bottom, the arrow −9.2 ± 0.8 mV, sS1–S4 = −16.0 ± 0.9 mV. n = 9 for each of wild-type indicates when inactivation was measured during 40-mV pulses after a shark, wild-type human, hS1–S6, hS1–S4 and sS1–S6, and 7 for sS1–S4. series of prepulses that increased in 10-mV increments from −100 mV. c, Currents elicited from the indicated S5–S6 chimaeric channels in b, Compared with wild-type human KV, average G–V relationships for response to voltage protocols to access voltage-dependence for activation wild-type shark, shark (s)S1–S6, and sS1–S4 channels were similarly and inactivation. d, sS5–S6 reduces voltage-dependent inactivation and shifted to positive voltages with more gradual slopes and deactivation hS5–S6 partially confers strong voltage-dependent inactivation in shark kinetics were faster. Va1/2 (mV) for wild-type human = −30.7 ± 0.5 mV, channels. n = 6. P < 0.0001 for hS5–S6 versus sS5–S6 or wild-type shark, slope factor (Ka) = 4.7 ± 0.5 mV; wild-type shark Va1/2 = −5.4 ± 0.5 mV, sS5–S6 versus hS5–S6 or wild-type human, two-way ANOVA with post Ka = 7.6 ± 0.4 mV; sS1–S6 Va1/2 = −9.7 ± 0.7 mV, Ka = 8.1 ± 0.6 mV; hoc Tukey test. e, Top, S5–S6 substitution did not greatly affect voltage- sS1–S4 Va1/2 = −6.5 ± 0.7 mV, Ka = 10.7 ± 1.2 mV. Average deactivation dependent activation. n = 6 for hS5–S6, 7 for sS5–S6. Bottom, S5–S6 kinetics of wild-type shark, sS1–S6, and sS1–S4 KV channels were faster substitution did not affect deactivation kinetics. n = 7 for hS5–S6, 9 for than those of wild-type human channels. Substitution of human (h)S1–S6 sS5–S6. All data are represented as mean ± s.e.m, n denotes cells. or hS1–S4 into the shark KV channel also partially shifted the activation

highest expressed highest expressed ribbon synapse a b ampullary transcripts ampullary ATPases transcripts

ampullae skin brain sharkskate

20000 500 250 200

M M M

K K K 150 P 10000 P 250 P F F F 100 50 0 0 0

+ + + ) + ) ) + ) ) ) ) ) 2 ) 1 e (H K 3 + /K (H a +/ olo) T a (C a c binding (N v0e2 (N (pic (EAAT 2+ oxidase) oxidase) oxidas p6v0c p (RIBEYE) VGLU a p2b1 ( (hemoglobin) At At (C At Pclo lc1a3 Atp1b2 Atp1a1 S Ctbp2a lc17a8 S Pvalb8 Hbbe3 (mitochondiral (mitochondiral(mitochondiral

t-co3 t-co1 Mt-co2 M M

c electrosensory cell

vesicle pools:

attached synaptic ribbon refilling

readily releasable

synapse / afferent nerve

d attached ribbon vesicle ribbon shape ribbon vesicle readily refilling vesicles density variation angle diameter releasable 600 100 r 150 150 60 20 60 m m 80 15 400 µ 100 100 40 40 60 10 40 nm vesicles 50 50 20 vesicles 20 200 5

20 vesicles / µ vesicles / angle from PM 0 0 0 0 0 0 0 % greater than linea k e k e k e k e k e k e ar t h har ka kate har har s skat sharskat s s sharks s skat s skat sharskat

Extended Data Fig. 8 | Electrosensory cell ribbon synapse bar, 500 nm. d, Quantification of attached vesicles, ribbon vesicle density, characteristics. a, Left, five highest expressed transcripts in shark ribbon shape variation and vesicle diameter was similar between shark ampullae. The Ca2+-binding protein parvalbumin 8 is the most highly and skate electrosensory cells. The readily releasable pool, quantified by expressed and is enriched in ampullae compared with other examined number of vesicles 150 nm from the synapse, was significantly larger in tissues. Right, five highest expressed ATPase transcripts in shark shark versus skate electrosensory cells. n = 18, P < 0.0001, two-tailed ampullae. A plasma membrane Ca2+ ATPase is highly expressed and Student’s t-test. The refilling pool density, quantified as detached cytosolic enriched in ampullae. Bars represent FPKM. b, Expression of transcripts vesicles, was significantly larger in skate electrosensory cells. n = 20 shark associated with ribbon synapses in shark and skate ampullae. Expression and 21 skate, P < 0.0001, two-tailed Mann–Whitney test. Shark ribbons of vGluT3 and EAAT1 suggests that the synapse could be glutamatergic. were more parallel to the plasma membrane in comparison to skate c, Transmission electron micrograph of skate ribbon synapse with arrows ribbons that were often more perpendicular. For angle quantification indicating electrosensory cell, synaptic ribbon and afferent nerve terminal. n = 20 shark and 21 skate ribbons, P < 0.0001, two-tailed Mann–Whitney Distinct vesicular pools are coloured: blue, attached to ribbon; green, test. All data are represented as mean ± s.e.m, n denotes counted refilling; yellow, readily releasable. An orange dotted line indicates the structures. 150-nm region in which the readily releasable pool was quantified. Scale

a shark skate b electrosensory cell: I sharkskate 10ms 150 C m 100 (pC) 50 Ca2+

2s Q 0 2s 0 1s 10 30 50 70 C 10 200 300 500 800 m 1s 10002000 .5s .5s ms .1s .1s

c d shark skate electrosensory cell: shark skate -15 -42 -55 8 V -65 m 6

I Ca2+ 4 Q 2 0 25pA oscillation k te a 50ms stimulus: har s sk

Extended Data Fig. 9 | Shark electrosensory cell vesicular release intracellular Cs+, extracellular 4-AP and IbTx. d, Voltage-clamp protocols characteristics. a, Top, currents and capacitance changes in response to a developed to simulate shark electrosensory cell spiking induced the same − + 10 ms, 20 mV voltage pulse in shark and skate electrosensory cells. Scale amount of QCa2 in shark or skate cells. Similarly, voltage protocols that bars, 50 pA, 200 ms. Bottom, representative capacitance changes in simulated smaller skate electrosensory cell voltage oscillations induced the − + + response to the indicated durations of a voltage stimulus of 20 mV. Scale same amount of QCa2 in shark or skate cells. QCa2 elicited by simulated − + bars, 25 fF, 200 ms. b, 20 mV voltage pulses of various durations induced voltage spikes was larger than QCa2 elicited by simulated oscillations. + = similar integrated ICaV (QCa2 ) in shark or skate electrosensory cells. n 6. n = 10 for shark and 5 for skate. All data are represented as mean ± s.e.m, c, ICaV elicited by simulated voltage spikes in shark electrosensory cells and n denotes cells. smaller voltage oscillations in skate cells.. K+ currents were blocked by

a b c C

6 cm 1 cm

def

3 cm 0.5 cm

Extended Data Fig. 10 | Schematic representation of ampullae of the catshark. c, Photograph of ampullary pores, visible on the ventral Lorenzini distribution in two elasmobranch species. a, The dorsal rostrum of an adult catshark. d, The dorsal surface of the little skate surface of the chain catshark (S. retifer), with black dots corresponding to (L. erinacea). The hyoid capsules from which electrosensory cells were individual ampullary pores and blue lines representing canal structures. obtained are indicated with arrowheads. This schematic was prepared The buccal and supraorbital clusters from which electrosensory cells were from photographs of four individual fish. e, The ventral surface of the obtained are indicated with arrowheads. This schematic was prepared skate. f, Photograph of a cleared Alcian blue-stained skate, revealing the from photographs of four individual fish. b, The ventral surface of ampullary canals from the ventral surface of the skate.

Corresponding author(s): David Julius

Reporting Summary Nature Research wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in reporting. For further information on Nature Research policies, see Authors & Referees and the Editorial Policy Checklist.

Statistical parameters When statistical analyses are reported, confirm that the following items are present in the relevant location (e.g. figure legend, table legend, main text, or Methods section). n/a Confirmed The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement An indication of whether measurements were taken from distinct samples or whether the same sample was measured repeatedly The statistical test(s) used AND whether they are one- or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section. A description of all covariates tested A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistics including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals)

For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted Give P values as exact values whenever suitable.

For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes Estimates of effect sizes (e.g. Cohen's d, Pearson's r), indicating how they were calculated

Clearly defined error bars State explicitly what error bars represent (e.g. SD, SE, CI)

Our web collection on statistics for biologists may be useful.

Software and code Policy information about availability of computer code Data collection Transcriptional profiling: Trinity suite for de novo assembly v2.1.0, NCBI Blast (v2.2.31) with a maximum E value of 1X10-5, RSEM v1.1.17 Cell physiology: pClamp v10, Clampfit, PatchMaster, Prism v7 Cellular structural measurements (ribbons): ImageJ v1.47 Behavioral experiments: iMovie 10.1.2

Data analysis Prism v7 was used for statistical analyses.

For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors/reviewers upon request. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Research guidelines for submitting code & software for further information. March 2018 Data

1 Policy information about availability of data nature research | reporting summary All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: - Accession codes, unique identifiers, or web links for publicly available datasets - A list of figures that have associated raw data - A description of any restrictions on data availability

Deep sequencing data are archived under GEO accession number GSE103977. The GenBank accession numbers for the α subunit are: CaV1.3 MF959522, KV1.3 MF959523, and BK MF959524. Other data are available from the authors upon request.

Field-specific reporting Please select the best fit for your research. If you are not sure, read the appropriate sections before making your selection. Life sciences Behavioural & social sciences For a reference copy of the document with all sections, see nature.com/authors/policies/ReportingSummary-flat.pdf

Life sciences

Study design All studies must disclose on these points even when the disclosure is negative. Sample size Sample sizes were chosen based on pilot experiments and prior experiences of the investigators in similar previously published experimental preparations. Behavioral experiments (n=10 under each stimulus condition) exceed previously published sample sizes for elasmobranch sensory behavioral experiments.

Data exclusions Data were excluded from behavioral experiments when basal ventilation frequency differed from the mean for each species by more than 2 standard deviations.

Replication All attempts at replication were successful.

Randomization Fish were chosen in random order for behavioral analyses and were placed back in their home aquaria for at least 24 hours between testing.

Blinding Investigators were blinded during behavioral data analysis phases.

Materials & experimental systems Policy information about availability of materials n/a Involved in the study Unique materials Antibodies Eukaryotic cell lines Research animals Human research participants

Eukaryotic cell lines Policy information about cell lines Cell line source(s) ATCC

Authentication Authenticated from vendor (ATCC).

Mycoplasma contamination Cell lines were negative for mycoplasma, as validated by ATCC. March 2018

Commonly misidentified lines HEK293T is listed among the ICLAC misidentified cell lines; however, these cells were used only as representative mammalian (See ICLAC register) expression systems, regardless of specific mammalian cell line identity. Cells were used solely in transient heterologous ion channel expression and patch clamp recordings.

2 Research animals nature research | reporting summary Policy information about studies involving animals; ARRIVE guidelines recommended for reporting animal research Animals/animal-derived materials Adult little skates (Leucoraja erinacea) and chain catsharks (Scyliorhinus retifer) of unknown age were used in cell physiology, transcriptional profiling, and histological analyses. Male and female fish were used, approximately 40 cm (skates) and 50 cm (catsharks) in total length. Adult skates weighed approximately 700 g, and catsharks weighed approximately 400 g.

For behavioral experiments, juvenile skates were at least 10 days post hatching, with a mean length of 10 cm and mass of 4.1 g of both sexes. Juvenile catsharks were at least 14 days post hatching, with a mean length of 12 cm and a mass of 3.8 g of both sexes. Method-specific reporting n/a Involved in the study ChIP-seq Flow cytometry Magnetic resonance imaging March 2018