Calcium activates the light-dependent conductance in melanopsin-expressing photoreceptors of amphioxus

Significance

“Circadian” photoreceptors of the mammalian retina use melanopsin, a photopigment that resembles invertebrate opsins and has widespread phylogenetic distribution, from prechordates to man. Although converging evidence in different species implicates the phospholipase C (PLC) pathway in melanopsin light-signaling, the downstream mechanism that controls the photoconductance has remained elusive. Here, we used melanopsin-expressing photoreceptors of amphioxus, the most basal extant chordate, to provide strong evidence implicating internally released calcium in the activation of the light-sensitive channels. For the first time, to our knowledge, exogenous application of a candidate messenger fully reproduces the effects of photoexcitation in a PLC-utilizing photoreceptor. Because of its unique phylogenetic position, amphioxus can also help shed light on the evolutionary history of diverse lineages of light-sensing cells and their transduction mechanisms.

Abstract

Melanopsin, the photopigment of the “circadian” receptors that regulate the biological clock and the pupillary reflex in mammals, is homologous to invertebrate rhodopsins. Evidence supporting the involvement of phosphoinositides in light-signaling has been garnered, but the downstream effectors that control the light-dependent conductance remain unknown. Microvillar photoreceptors of the primitive chordate amphioxus also express melanopsin and transduce light via phospholipase-C, apparently not acting through diacylglycerol. We therefore examined the role of calcium in activating the photoconductance, using simultaneous, high time-resolution measurements of membrane current and Ca²⁺ fluorescence. The light-induced calcium rise precedes the onset of the photocurrent, making it a candidate in the activation chain. Moreover, photolysis of caged Ca elicits an inward current of similar size, time course and pharmacology as the physiological photoresponse, but with a much shorter latency. Internally released calcium thus emerges as a key messenger to trigger the opening of light-dependent channels in melanopsin-expressing microvillar photoreceptors of early chordates.

The discovery of melanopsin-using photosensitive ganglion cells in the vertebrate retina (ipRGCs) (1, 2) not only clarified how various nonvisual light-dependent processes are controlled, but also incorporated novel contributions to visual function by nonconventional light sensors (3–5). In addition, the unexpected molecular kinship between melanopsin and the opsins of invertebrate microvillar photoreceptors (6, 7) questioned the long-held belief that this lineage of receptors was excluded from vertebrate phyla. Indeed, investigations of melanopsin signaling mechanisms in ipRGCs (8) implicated a G_qα protein tapping into the PLC pathway, the canonical scheme of invertebrate phototransduction (9). However, studies in native cells proved arduous because of the scarcity of ipRGCs, so a definitive picture of the downstream mechanisms of light transduction has yet to emerge. Some reports argue against both calcium and DAG as critical signaling elements, hinting instead at a potential role for PIP₂ in channel gating (8), but firm evidence has yet to be garnered. As for the nature of the light-sensitive conductance, ion channels of the TRP superfamily have been implicated (10–13).

Melanopsin has been documented in amphioxus (14), the most basal extant chordate (15); it expresses in two well-identified and relatively abundant cell types in the neural tube of this eyeless organism; these neurons, called Joseph and Hesse cells, have a microvillar architecture (16, 17) and are intrinsically photosensitive (18). These features make amphioxus an attractive model system to investigate melanopsin transduction and its evolutionary history (19); the availability of its sequenced genome (15) adds a powerful tool for the identification of key signaling elements. The photoconductance of Joseph and Hesse cells is permeable to Na and Ca, and is susceptible to blockers of TRP-class ion channels (20). In both cell types, PLC plays a critical role (21), as it has been proposed for ipRGCs; moreover, light mobilizes calcium from internal stores (18). Inhibitors of the IP₃ receptor and intracellular administration of high-affinity calcium chelators suppress photoresponsiveness (18, 21); by contrast, DAG analogs and PUFAs proved inert (21). These observations point to a critical role for the IP₃/Ca branch of the PLC cascade in photoexcitation. The present work scrutinized the proposition that internally released calcium may be a messenger that controls light-dependent ion channels.

Results

Illumination Causes a Rapid Increase in Intracellular Calcium.

Fig. 1A illustrates the response of a Joseph cell to a saturating flash of blue light (470 nm, close to the absorption peak of melanopsin). The photocurrent had a very brief latency (13.7 ms ± 3.5 SD, n = 19), reaching a peak in ∼20 ms (Fig. 1A, Inset); it was preceded by a minute outward hump representing the melanopsin photoisomerization current (ERC; ref. 22). The viability of calcium as a potential photoexcitation messenger requires that its elevation precede the opening of the light-dependent channels. Previous Ca measurements relied on fluorescence digital imaging (18) and did not afford sufficient temporal resolution to settle the issue. We turned to a much faster PMT-based system, and examined the light-induced Ca increase in cells loaded with Fluo-4-AM. As shown in Fig. 1B, upon turning on the epi-illumination beam (same intensity as in A) the PMT signal jumped to the basal fluorescence level (dotted line), and after a few milliseconds began a steep climb, reflecting the rising [Ca²⁺]_i. The mean latency of the Δ[Ca²⁺] was 10.3 ms ± 1.9 SD (n = 17 Joseph cells). Because the average duration of the latent period of the photocurrent and that of the Ca increase are within a similar, narrow range, comparisons across cells are insufficiently reliable to determine the order of events. We therefore measured simultaneously Δ[Ca²⁺] and membrane current (I_m) in the same cell; the electrical and the optical signals were processed through matched analog filters to avoid introducing any differential time shift. In Fig. 1C, the photocurrent and the Ca fluorescence signal appear to be nearly coincident; however, if the time scale is expanded (Fig. 1C, Inset) the fluorescence increase leads the onset of the photocurrent by a minute but distinct interval; to estimate the beginning of the rise in Ca fluorescence, which is noisier than the membrane current trace, we took the intersection of two lines, fitted to the baseline and to the early rising phase of the signal, respectively. The average lead time was 0.67 ms ± 0.39 ms SD (n = 9). The statistical significance of the effect was confirmed by a nonparametric test (Wilcoxon sign-rank test; P < 0.05). Population data are shown in Fig. 1D.

Fig. 1.

Light-triggered Δ[Ca]_i has a briefer latency than the electrical response. (A) Photocurrent in a Joseph cell voltage clamped at −50 mV and stimulated with a 100-ms saturating blue light (470 nm). (Inset) Time-expanded view of the initial portion of the recording. The initial outward transient is the melanopsin photo-isomerization current. (B) Light-elicited increase in cytosolic Ca measured with a photomultiplier in a Joseph cell preloaded with Fluo-4. The arrowheads mark the onset and termination of the epi-illumination light (same intensity as in A). The basal fluorescence level is indicated by the dotted line; a pronounced increase in fluorescence, reflecting the light-induced Ca release, occurred after a short delay (Inset). (C) Simultaneous measurement of membrane current and Ca fluorescence in a Joseph cell internally dialyzed with Fluo-4 through the recording micropipette. Upon expanding the time scale (Inset), it can be seen that onset of the rise in Ca fluorescence (first arrowhead, marking the intersection of the two fitted lines) precedes the activation of the photocurrent (second arrowhead). (D) Population data plot of the time lag between photocurrent and Ca rise onset.

Photo-Release of Caged Calcium Triggers a Large Inward Current.

We next ascertained whether an induced Ca raise can open the light-dependent channels. DM-nitrophen “caged calcium” was used for this purpose. A prime concern was to generate calcium changes that resemble those that occur physiologically upon light stimulation; these last for >100 ms (Fig. 1 B and C). By contrast, depending on the fraction of Ca-loaded vs. free DM-n, and UV light intensity, flash photolysis can produce brief Ca spikes (owing to the rapid rebinding to the free, unphotolyzed cage; refs. 23 and 24), or a more sustained elevation. After extensive in vitro tests (Fig. S1) we opted for more prolonged UV irradiation, where the balance between continuous Ca liberation and buffering results in a more controllable, longer-lasting Δ[Ca]_i. An obvious caveat with regard to the use of caged compounds in photoreceptors is that the photolysis light could also activate the light-transduction cascade, which was indeed the case (Fig. S2). To avoid confounding between direct and indirect effects of the uncaging light, we opted for a strategy in which the endogenous light response was disabled, to reveal the uncontaminated effect of photo-released Ca. Fig. 2 A and B illustrates a typical experiment, in which the solution in the pipette contained Ca-loaded DM-n and Fluo-4. Immediately upon attaining whole-cell clamp (Fig. 2A) a cell was exposed to the 470-nm epi-fluorescence illumination, followed 200 ms later by a step of UV light; at that point, dialysis was minimal (Fig. 2A, Inset), and little caged Ca and Ca indicator had penetrated the cell. As expected, the 470-nm light evoked a large photocurrent, but hardly any Ca-fluorescence. The subsequent UV light, on the other hand, produced a minute autofluorescence signal with the same rectangular time course as the stimulus, but failed to elicit any change in membrane current (the photoreceptor was completely desensitized by the blue epi-illumination beam). The procedure was repeated after 10 min of dialysis in the dark, at which point the outcome radically changed: as shown in Fig. 2B, when the 470-nm light was turned on, no photocurrent was elicited, indicating that the phototransducing machinery remained disabled (due to both the prior high-intensity UV and Ca-buffering by DM-n; Fig. S3). On the other hand, basal fluorescence had greatly increased, reflecting the dialysis of Fluo-4 (Fig. 2B, Inset); however, no delayed rise in Ca was observed (i.e., the phototransduction machinery was disabled upstream of the Ca-release step). The subsequent UV unmistakably produced Ca uncaging, indicated by the gradually rising fluorescence signal after an initial jump, and the larger post-UV level attained (see Fig. 2 for details). Concomitantly with the induced Δ[Ca²⁺], an inward current was triggered, with an amplitude and a time course resembling the native photocurrent of Fig. 2A (n = 12; 8 Joseph and 4 Hesse cells); on occasions, the photolysis-triggered current was not as large as the initial photocurrent, but it invariably attained an amplitude of several nA. The Ca-evoked current is accompanied by an increase in conductance (Fig. S4), like the physiological light response (ref. 18 and Fig. 2A). To further rule out any involvement of the melanopsin light-signaling pathway in the photolysis-triggered response, we determined that photo-released calcium acts downstream of PLC. Fig. 3A confirms that the PLC antagonist U73122 completely eliminates the photocurrent (21); Fig. 3B demonstrates that the drug inhibits light-induced internal Ca release: A robust Δ[Ca²⁺] was elicited twice in control conditions; subsequently, U73122 (10 μM) was applied by puffer pipette, causing a gradual disappearance of the calcium transient (n = 3). Having corroborated the efficacy of U73122 upstream of the photoconductance, we tested its effects in a Ca-photolysis experiment. Fig. 3C shows two examples using a procedure similar to that of Fig. 2, except that, after the initial trial, U73122 was constantly applied. In the presence of the antagonist, Ca uncaging triggered a sizable inward current, indicating that PLC activity is not required [the presence of a second hump of the photocurrent (e.g., Fig. 3C, Right) is occasionally observed; ref. 18]. A second argument against the involvement of upstream light-signaling elements is based on the kinetics of the UV-triggered current. Ca-release from DM-nitrophen is very fast (25), with rates in excess of 3 × 10⁴ s⁻¹; with sufficiently intense UV light, [Ca²⁺] can be increased significantly in the millisecond time scale (24). Direct effects of Ca on downstream elements ought to have a short latency, compared with the melanopsin-initiated photocurrent. Fig. 3 D and E illustrate that this is the case: In the traces displayed on an expanded time scale (E), the photolysis-triggered current is activated within 2.5 ms of stimulus application (average 3.63 ms ±1.68 SD, n = 12). By contrast the same UV light administered in the absence of caged Ca produces light-responses with a much longer latency (Fig. S5). The photolysis-elicited current obtains with a comparable average [Ca]_i elevation as that resulting from melanopsin photostimulation. Nonratiometric indicators, like Fluo-4, do not allow quantification of absolute [Ca] levels, but it is possible to compare the relative increase in Ca fluorescence (ΔF/F). Recordings of DM-n-loaded cells were selected in which the UV light induced a current akin to the saturating photocurrent, and ΔF/F was gauged in two ways: (i) the relative increase of fluorescence pre/post UV (1.63 ± 0.82 SEM, n = 12); and (ii) the relative increase during the photolysis light (1.42 ± 0.45 SEM, n = 12). In both cases the value obtained did not differ significantly (P > 0.10, Mann–Whitney test) with respect to the Δ[Ca]_i that accompanies the saturating light response (ΔF/F = 1.51 ± 0.22 SEM, n = 5).

Fig. 2.

Photo-release of caged calcium triggers a large inward current. (A) Photo-stimulation of a Joseph photoreceptor immediately upon attaining the whole-cell configuration with an electrode containing Fluo-4 and Ca-loaded DM-n. Application of the 470 nm epi-fluorescence beam for 450 ms elicited a saturating photocurrent; a 100-ms UV flash superimposed 200 ms later produced no effect. The photomultiplier trace shows virtually no Ca fluorescence, only a step-like UV auto-fluorescence signal. (B) Repetition of the same routine after 10 min of dialysis. No electrical response was produced by the blue light, but a large current was triggered by the UV photolysis flash. The PMT record shows a much increased basal fluorescence (dashed line) from the Fluo-4 that entered the cell; the superimposed UV flash caused an abrupt jump in fluorescence due to the added effective excitation light (dotted line), followed by a gradual further increase reflecting the rising [Ca] (arrow a). After termination of the uncaging light, the Ca fluorescence remained elevated compared with the initial level (arrow b).

Fig. 3.

Ca photorelease effects bypass melanopsin signaling. (A) Application of the PLC antagonist U73122 (10 μM) abolishes the photoresponse of a Joseph cell to a flash delivered every minute. (B) Elimination of the light-induced Ca elevation by U73122 in a Joseph cell loaded with Fluo-4. After two near-identical control responses, the drug (10 μM) was locally applied. Trials were spaced 5 min apart. (C) Ca-elicited currents persist in the presence of PLC blockade. In the two examples, just after attaining the whole-cell configuration only the photocurrent elicited by the blue light was observed (black traces). Subsequently, U73122 (10 μM) was steadily perfused, as DM-n and Fluo-4 were dialyzed into the cell. Ten minutes later, the UV photolysis flash triggered a substantial current concomitantly with a distinct increase in [Ca]_I (gray traces). (D) Speed of Ca photorelease effects on membrane current. Current elicited by a UV light in a Joseph cell after 9 min dialysis with Ca-loaded DM-n and Fluo-4. (E) Expansion of the initial portion of records in D. The dotted lines highlight the brief latency of the inward current (Δt), barely following the Δ[Ca] onset.

The Current Induced by Calcium Resembles the Native Photocurrent.

To assess the significance of the photolysis-triggered current, it is important to ascertain whether it is the same as the photocurrent, or a component thereof. In a study on Drosophila photoreceptors, Ca photorelease triggered a small (tens of pA) inward current that was attributed to the activation of an electrogenic Na/Ca exchanger by the imposed calcium load - rather than implicating the light-dependent conductance (26). In the present case, the sheer size of the Ca-evoked current (several nA) makes such a mechanism unlikely, as it would call for an implausibly high expression level of the exchanger; nonetheless, the issue was directly addressed by testing the effect of replacing extracellular Na with Li⁺, which has been reported not to support the forward Na/Ca exchanger cycle (27, 28). Because the photocurrent of Joseph and Hesse cells is largely carried by Na (18, 20), we first determined that Li⁺ permeates through the light-dependent channels, allowing us to monitor their activity (Fig. 4A); under these conditions, a small reduction in the late tail of the photocurrent was observed (Fig. 4A, Inset; n = 3), compatible with (but not proving) the presence of an exchanger current affected by Li⁺. Then, we tested UV-uncaging of DM-n in Li-substituted ASW, and obtained sizable Ca-triggered inward currents (Fig. 4B); this observation argues against a Na/Ca exchanger as the underlying mechanism. Subsequently, we examined the basic conduction and pharmacological properties of this current, with the constraint that only a limited number of uncaging trials could be administered before the cell deteriorated. Because of the transient nature of the current, one cannot use a ramp protocol to determine the shape of the full I–V curve, so steady holding potentials were used instead. We had previously shown that in ASW the photocurrent remains inwardly directed even at large positive voltages (20). In the same ionic conditions the Ca photolysis-triggered current is also inward with depolarizations to +50 mV or larger (n = 4); Fig. 5A shows an example in which the uncaging UV light was administered to a Joseph cell after stepping V_m to +80 mV. The physiological photocurrent can be reversed upon removal of extracellular Na (V_rev ∼−27 mV) (20). Fig. 5B shows that after substituting extracellular Na with NMDG, reversal of the Ca-elicited current obtains within a narrow voltage range very close to such value. In three instances, however, after multiple uncaging trials this current could be reversed in ASW (i.e., in the presence of external Na), in the range +12 to +15 mV; this likely reflects excessive intracellular Ca accumulation and the consequent shift in the equilibrium potential for this ion, a prime carrier of the photocurrent (18). Finally, a further parallelism between the currents evoked by melanopsin stimulation and by Ca uncaging was established with pharmacological blockers. The photocurrent of Joseph and Hesse cells is antagonized by ruthenium red and by lanthanum (20); as shown in Fig. 5C, bath application of these antagonists at the same concentrations also produced a quantitatively similar, reversible reduction in the current triggered by caged-Ca photolysis (n = 2).

Fig. 4.

Electrogenic Na/Ca exchanger does not account for the photolysis-evoked current. (A) Lithium permeates through the light-sensitive channels. A dim 470 nm, 100-ms flash was repetitively applied (1 per min) to a Hesse cell held at −50 mV. After three control responses in ASW, Na was replaced with Li, causing a ∼43% increase in photocurrent amplitude. (Inset) Normalized light responses measured in the same two conditions in a different cell: in the presence of lithium, the late tail of the photocurrent was reduced. (B) Effects of caged-Ca photolysis in Li-ASW. Superimposed traces with a protocol similar to that of Fig. 2 A and B, showing the initial photocurrent elicited by the epi-illumination beam just after membrane rupture, and the photolysis-triggered current after 10 min dialysis with caged calcium. (Inset) Blow-up of the fluorescence trace, showing the photolysis-induced Ca rise.

Fig. 5.

Identification of the Ca-evoked current with the native light response. (A) The Ca-evoked current remains inward with large depolarizations. A Joseph cell was superfused with ASW (containing 480 mM Na). After 10-min dialysis with caged calcium, the command voltage was stepped to +80 mV before delivering the UV photolysis flash. This triggered an inward current. (B) Reversal of the Ca-triggered current after replacement of external Na with NMDG. A Joseph cell was tested with caged-Ca photolysis, first near the resting potential and subsequently at voltages close to the value of the reversal potential of the photocurrent, previously measured under identical conditions; trials were spaced 3 min apart. (C) The Ca-evoked current is blocked by antagonists of TRP-type ion channels. A Hesse cell was subjected to photolysis of Ca-loaded DM-n in normal ASW and in the presence of 5 μM ruthenium red and 100 μM La³⁺. The treatment with the antagonists caused a fully reversible decrease in the amplitude of the Ca-triggered current.

The results strongly implicate light-evoked Ca release as a trigger for the photocurrent in amphioxus melanopsin-expressing photoreceptors. Nonetheless, discrepancies in the time course of the two events (e.g., Fig. 1) need to be addressed. In the first place, the photocurrent turns on abruptly, quickly attaining peak amplitude while [Ca] continues to grow. A plausible conjecture is that multiple Ca ions may be required for the reaction leading to channel opening, possibly in a cooperative manner, thus giving rise to a threshold-like phenomenon. Clues consistent with such a possibility were indeed obtained: we first documented a nonlinear dependency of the evoked current on Δ[Ca], by titrating the intensity of the UV light. In Fig. 6A, current amplitude is plotted as a function of the Ca fluorescence immediately after termination of the uncaging UV; the graph remains nearly flat over a large portion of the range, but then the current grows abruptly (n = 2). A plot of current vs. photolysis light intensity has a similar shape (Fig. 6A, Inset), indicating that it is not an artifact of Ca-fluorescence nonlinearity (e.g., dye saturation). This behavior can also be observed within a single trial (Fig. 6B) in which an attenuated UV step gives rise to a nearly linear rise in the Ca fluorescence; the concurrently recorded trace of I_m shows no hint of response during a significant interval, and then the inward current was suddenly activated (n = 2). Finally, the threshold behavior can also be corroborated with a double-pulse protocol: In Fig. 6C, a DM-n-loaded cell was stimulated with a UV light that elevated [Ca] insufficiently to elicit any current; this was followed by a second, identical flash, whose cumulative effect brought the [Ca] level only slightly higher: as a result a large current was triggered (n = 2). Nonlinearities of this sort likely participate in the initiation of the current triggered by melanopsin stimulation, and could help explain the diverging kinetics of rising phase of Δ[Ca²⁺] and I_m.

Fig. 6.

Nonlinear Ca-dependency of photolysis-triggered current. (A) Amplitude of photolysis-evoked current as a function of Ca fluorescence increase. 100-ms UV stimuli of increasing intensity were superimposed on 470 nm light steps to monitor Ca, in a Joseph cell loaded with Fluo-4 and DM-n/Ca. Current and Ca fluorescence were concomitantly measured immediately after the UV flash. (Inset) Plot of the photolysis-activated current vs. intensity of UV light. (B) Attenuated uncaging UV light, giving rise to a gradually increasing [Ca]; the simultaneously monitored current trace, by contrast, shows an abrupt onset. (C) Double uncaging flash experiment: a first stimulus, insufficient to elicit a current, was followed, 350 ms later, by a second UV flash which brought Ca concentration slightly higher, triggering a large inward current. (D) The persistence of the Ca fluorescence is not due to dye saturation. Ca fluorescence in Joseph cells preloaded with the low-affinity Ca indicators Fluo-5F and Calcium Green 5N (AM). In the case of Ca-Green 5N, the signal was very small, and ensemble averaging was used.

Another feature that needs explaining is the turn-off of the photocurrent, contrasted with the lingering fluorescence elevation (Fig. 1). A possible account would be saturation of the Ca indicator, producing a sustained fluorescence signal even if the real Ca change may have declined. A conductance activated by Ca with low affinity could be triggered only transiently, when the actual [Ca²⁺] is very high, and the inconsistency would be readily accounted for. To address this question, we tested the lower-affinity indicators Fluo-5F (K_D ∼ 2.3 μM; n = 6) and Ca-Green 5N (K_D ∼ 14 μM; n = 2), but the fluorescence time course (Fig. 6D) was similar as that recorded with Fluo-4 (K_D ∼ 0.34 μM). One can conclude that the light-induced Ca fluorescence is not distorted by saturation. A plausible alternative is that additional, separate processes are implicated in terminating the photocurrent, as occurs in other PLC-using photoreceptors (29, 30).

Discussion

Melanopsin-expressing light sensors are found throughout the chordate phylum, from amphioxus to man, and the early steps of their light-transduction cascade share key features with microvillar visual receptors of invertebrates: not only is there a significant homology in the photopigments (6, 7), but also a commonality in the type of heterotrimeric G protein (G_q) and the enzyme (PLC) recruited for light signaling (8, 21). In both classes of light sensors, however, the nature of the downstream effectors that control the photoconductance has remained elusive. In invertebrates the gamut of proposed candidates includes IP₃ (31, 32), Ca (33), DAG or metabolites thereof (34–36), PIP₂ (37), metabolic stress (38), capacitative Ca entry (39), protons (40), and mechanical forces (41). In fact, there are even indications that no universal mechanism may exists, because substances that exert a clear effect in one system, like PUFAs in Drosophila (35), proved inert in another, like Limulus (42), and vice versa (43). The situation is equally blurry for ipRGCs, where PUFAs and heparin proved ineffective, prompting the suggestion that neither downstream branch of the PLC cascade is involved; a possible role for PIP₂ was proposed instead (8). To date, neither in melanopsin-expressing cells nor in invertebrate microvillar visual receptors of any species had the application of a candidate messenger fully reproduced the effects of photostimulation. In amphioxus such a candidate has now emerged: Ca released from intracellular stores upon photostimulation is not only necessary for light responsiveness (18), but its rapid exogenous elevation, bypassing upstream steps, is also sufficient to activate a current whose amplitude, kinetics, ionic conduction, and pharmacology mimics the light-sensitive conductance.

Internally released calcium seems suitable for mediating the large, fast photoresponse of Joseph and Hesse cells. Amplification on such a short time scale poses a challenge to any signaling cascade. The conductance of the light-dependent channels in amphioxus is ∼32 pS, and unitary currents are ∼1–2 pA near resting V_m (22); considering p_o = 0.25, one would estimate that a saturating photocurrent must recruit >10⁴ channels within 20–30 ms, calling for swift, massive mobilization of an internal messenger. IP₃-gated Ca-release channels of the ER have a unitary conductance of ∼80 pS (44), and under physiological conditions they mediate a current of ∼0.2 pA (45), i.e., ∼0.6 × 10⁶ ions/s. This figure vastly exceeds the turnover number of a fast enzyme, so that a large local Ca elevation could be rapidly generated. In fact, our simultaneous measurements of photo-induced Δ[Ca²⁺] and membrane current confirm that the onset of the light-induced calcium rise is sufficiently swift. The biggest challenge currently resides in fully clarifying differences in kinetics between the Δ[Ca] and the photocurrent. These may in part arise from demonstrated nonlinearities and threshold effects of Ca elevation on the evoked current. Additional signaling molecules may be recruited to control photoresponse shut-off, as documented in Drosophila (29, 30); future exploration of such mechanisms will require surveying a multitude of potential modulators.

The amphioxus photocurrent is seemingly mediated by TRP-class channels (20), like in PLC-using visual receptors of invertebrates (9), and in ipRGCs (10–13). Interestingly, TRP channels gated by intracellular Ca have been described in various systems (46–48). It is tempting to propose that calcium may directly control the light-dependent channels in amphioxus, especially considering the brief delay between Ca-fluorescence changes and the onset of the photocurrent. However, the actual time lag is slightly underestimated due to the finite reaction time of the Ca indicator, and it would be premature to dismiss additional downstream steps. Measurements on perfused, inside-out patches proved technically formidable and would be unlikely to provide a definitive answer, because of documented instances of light responsiveness surviving patch excision (8, 49). Joseph and Hesse cells of amphioxus are ancestrally related to the ipRGCs of vertebrates (7) and the question arises on a possible kinship between the effectors of the two melanopsin signaling pathways. Murine ipRGCs are only minimally affected by heparin and by thapsigargin (8), which argues against such notion; however, their light response was gradually abolished by buffering [Ca]_i with BAPTA, although the onset of the effect was deemed too slow to represent direct interference with the messenger. Notwithstanding the likely ancestral relation between melanopsin-based photoreceptors of mammals and amphioxus, divergence of effector mechanisms is possible: there are precedents for photoreceptors of the same lineage in distantly related species (e.g., Drosophila vs. Limulus) seemingly recruiting different signaling elements to control the photoconductance, namely, the DAG (35, 36) vs. the IP₃ branches (31–33) of the PLC cascade. Further work will be required to determine whether an extrapolation of the present conclusions to vertebrates is warranted, or whether separate melanopsin effector mechanisms have evolved, perhaps related to the widely different kinetics of the photoresponses in question.

Materials and Methods

Cell Isolation.

Amphioxus (Branchiostoma floridae) were purchased from Gulf Specimens; enzymatically isolated Joseph and Hesse cells were obtained as described (18) and maintained in ASW (480 mM NaCl, 10 mM KCl, 10 mM CaCl₂, 49 mM MgCl₂, 10 mM Hepes, 5.4 mM glucose, pH 7.8).

Electrophysiological Recording.

Patch pipettes were filled with 100 mM KCl, 200 mM K-glutamate, 5 mM MgCl₂, 5 mM Na₂ATP, 20 mM NaCl, 1 mM EGTA, 300 mM Sucrose, 10 mM Hepes, 0.2 mM GTP (to which caged Ca and Ca indicator was added). Currents recorded with a Cairn Research Optopatch amplifier were low-pass filtered with a 4-pole Bessel filter; the cutoff frequency was usually set at 1 KHz, but was increased up to 5 KHz for measuring latencies. Data were digitized at 3–15 KHz (DT9834, Data Translation). The recording chamber was perfused via a multiport manifold to change the bath solution. “Puffer” pipettes were used for rapid, local application of chemicals.

Calcium Fluorescence.

The neural tube was incubated with Fluo-4 AM, Fluo-5F AM or Calcium Green-5N AM (5 μM + 0.05% Pluronic F-127, 2 h; Molecular Probes), before enzymatic dissociation. For simultaneous fluorescence and current measurements, the K salt of the dye (42 μM to 103 μM) was dialyzed via the patch pipette. Epi-illumination provided by a 470 nm LED source (Thorlabs) was further filtered by a 480–40 nm band-pass filter (Chroma). A Neofluar 100× oil-immersion objective was used throughout. An adjustable iris at a conjugated image plane restricted light collection. Cells were visualized with deep-red illumination by a CMOS camera (Thorlabs); shorter-wavelength light was diverted by a 585 nm dichroic mirror to a photomultiplier tube (Hammamatsu) connected to an I–V converter and a four-pole Bessel low-pass filter. A 535–50 nm barrier filter (Chroma) and an electromechanical shutter (Vincent Associates) were placed in front of the PMT housing.

Calcium Photorelease.

DM-nitrophen (Life Technologies; 5–10 mM) loaded with 0.6–0.7 equivalents of calcium (49) was mixed with the Ca indicator. pH measured with a minielectrode (Lazarus) was titrated to 7.3; aliquots were kept at −80 °C. A high-intensity, collimated 365 nm LED (OptoFlash, Cairn Research or Thorlabs) was used for photolysis; UV and epi-fluorescence beams were combined via a 440-nm cutoff dichroic reflector (Omega Optical). Preliminary assays in droplets embedded in a Sylgard matrix (50), tested various mixtures of DM-n, calcium, and Fluo-4, and combinations of flash intensity/duration to ensure photorelease (see Fig. S1 for details).