Osmoregulated Chloride Currents in Hemocytes from Mytilus galloprovincialis

We investigated the biophysical properties of the transport mediated by ion channels in hemocytes from the hemolymph of the bivalve Mytilus galloprovincialis. Besides other transporters, mytilus hemocytes possess a specialized channel sensitive to the osmotic pressure with functional properties similar to those of other transport proteins present in vertebrates. As chloride fluxes may play an important role in the regulation of cell volume in case of modifications of the ionic composition of the external medium, we focused our attention on an inwardly-rectifying voltage-dependent, chloride-selective channel activated by negative membrane potentials and potentiated by the low osmolality of the external solution. The chloride channel was slightly inhibited by micromolar concentrations of zinc chloride in the bath solution, while the antifouling agent zinc pyrithione did not affect the channel conductance at all. This is the first direct electrophysiological characterization of a functional ion channel in ancestral immunocytes of mytilus, which may bring a contribution to the understanding of the response of bivalves to salt and contaminant stresses.


Introduction
Marine bivalves are common components of the human diet and are also used as bioindicators in environmental monitoring systems. On the other hand, some invertebrates, such as Mediterranean mussels, constitute a major environmental problem for ship hull and industrial power plants that use marine water for cooling; for these reasons Mytilus galloprovincialis was inserted as an invasive species in the database of the "100 of the world's worst invasive species" [1].
Bivalves are able to filter large volumes of seawater, concentrating a series of contaminants within their tissues. They have an open circulatory system, the hemolymph, that is continuously exposed to fluctuating environmental factors, including mechanical and osmotic stresses as well as variable concentrations of salts and contaminants such as inorganic and organic metals. The hemocytes, the circulating cells, are responsible for the immune response of the mytilus to the environment and foreign materials; the roles played by these cells have been suggested to be equivalent to those of monocyte/macrophage lineages in vertebrates. Indeed in M.
galloprovincialis the hemocytes are involved in the phagocytic defence against foreign agents as well as in a series of other physiological functions such as wound and shell repair, nutrient digestion and excretion [2]. Interestingly, it has been reported that bivalves are subjected to disseminated neoplasia and the transmission of independent leukaemia-like diseases within individuals of the same species as well as among different mollusc species [3][4][5][6]. Indeed, recent results suggest that the transmission of tumour cells is more frequent in nature than previously thought and therefore studies on bivalves could be of interest also for a better understanding of cancer transmission in general and specifically of metastasization in humans.
The activity of ion channels has been reported in a few tissues of marine mussels; for example the patch-clamp technique was applied to cells from the ventricular myocytes of Mytilus edulis in order to characterize some voltage dependent channels [7]: namely two outward potassium currents ascribed to I k and I A channels, an inward L-type calcium channel and a tetrodotoxin sensitive Na-channel.
Despite the rapidly growing body of knowledge on ion transport in immune cells of vertebrates [8][9][10][11][12] as well as molluscs [13][14][15] little is known on channels in hemocytes of mussels. Only the effects of algal toxins on L-type calcium channels were investigated in the hemocytes from M. galloprovincialis by immunofluorescence experiments and confocal microscopy [16] or by analysis of cellular parameters and receptor recognition pattern in Mytilus chilensis [17].
The capability to respond to anisotonic conditions in these ancient and elementary organisms could be of primary importance to enlarge and improve the knowledge of similar processes in osmoregulated organisms. As it is well known that in mussels the internal medium follows the variations of the osmotic concentrations of the external medium with consequences on the increase/decrease of cellular volume, we adopted the patch-clamp technique in order to verify whether hemocytes under osmotic stress display any mechanism that may contribute to regulate their volume.
In this paper we were able to demonstrate that Mytilus galloprovincialis hemocytes possess a variety of specialized channels which appear to be qualitatively similar to other transport molecules previously identified in vertebrates. As chloride channels seem to play a potential role in the regulation of cell volume during transient modifications of the ionic composition of the external medium, we performed experiments in order to characterize the inwardly rectifying anionic current that was present and readily identified in several patch-clamp recordings in M. galloprovincialis hemocytes. Besides being selective for chloride these channels are voltagedependent and slowly activated by negative hyperpolarizing membrane potentials in moderate hyposmotic solutions that still allow the organism to activate a reasonable stress response [18]. On the contrary the chloride currents are reversibly inhibited by hyperosmotic conditions. In our working conditions, the currents were slightly inhibited by micromolar ZnCl 2 concentrations, while the organic compound zinc pyrithione (ZnPT 2 ) did not affect at all the current.

Materials and Methods
The ISMAR marine station was authorized by the Ministry of Infrastructure and Transport through the Genoa Port Authority. We confirm that the field studies did not involve endangered or protected species.

Hemolymph extraction
Adult specimens of Mytilus galloprovincialis were collected at the ISMAR marine station and transferred to the lab, cleaned of epibionts, allocated in tanks containing aerated filtered sea water and allowed to acclimate for a few days at 20˚C before experiments. The hemolymph was extracted from the posterior adductor muscle using a 2 ml syringe (Fig 1A), put in sterile tubes and kept in ice.

Microscopy
A home-made opto-electronic platform was used to acquire the optical images. The tool integrates standard modular components (Optem FUSION, Qioptiq Photonics GmbH & Co. KG) and is equipped with a Plan Fluor 40x objective (NIKON Instruments, Amsterdam, The  Netherlands) mounted on a motorized Z-axis with a 0.01 μm resolution step. The sample was mounted on a motorized X-Y stage with a 0.5 μm resolution step. A LED lamp and a Gig-E DMK 23G274 camera (The Imaging Source, Bremen, Germany) equipped with a CCD, Sony 1/1.8" (1600x1200 pixels), completed the equipment.
When needed, the bath solution was adjusted to increase the osmotic pressure to hyperosmotic values by the addition of sorbitol, namely ∏ = 1178±8 mosmol/kg. In the following we indicate this solution (which has the same salt concentrations of the hyposmotic solution) with the term hyperosmotic (bath) solution. To determine the ionic selectivity of the channel under study, when needed, potassium was substituted by N-methyl-D-Glucamine (NMDG) both in the bath and in the pipette solutions.
The osmolality of the solutions was determined with a vapor pressure osmometer 5100C (Wescor Inc., Logan, UT, USA) and the osmolality measurements represent mean values ±SEM (n = 20).
The patch-clamp technique [14] was applied to isolated hemocytes in the whole-cell mode. The ionic currents were recorded with an EPC7 (HEKA Instruments) current-voltage amplifier. Data were digitized using a 16 bit Instrutech A/D/A board (Instrutech, Elmont, N.Y.) interfaced to a computer, which generated the voltage stimulation protocol and stored the current responses on the computer hard disk. Current records were low-pass filtered with a 4-pole filter Kemo VBF8 (Kemo, Beckenham UK). When needed, two Ag-AgCl electrodes were supported by agar bridges and the applied voltages were corrected for the appropriate values of the Liquid Junction Potential measured according to [19]. Patch pipettes were pulled from thin-walled borosilicate glass tubing (Clark Electrochemical Instruments, Pangbourne, Reading, UK). The resistance of the patch pipettes in the bathing medium was in the order of 2-4 Mohm.
The ionic selectivity was monitored by the instantaneous values of the deactivating tail currents. A slow or a fast perfusion procedure was adopted to change the solution bathing the cell: in the first case the bath solution was changed by means of a peristaltic pump that was able to renew the entire volume of the bath in a few minutes. Instead in the fast procedure, the bathing medium surrounding the cell was changed using up to five large perfusion pipettes (with a tip in the order of 30 μm) each one filled with a different solution to be investigated [20][21][22]. Coarse movements were controlled by a hydraulic manipulator that allowed to switch within few seconds between the different perfusion pipettes bathing the cell.
The theoretical reversal potentials for various ions were calculated using the ionic activities coefficients derived from previous papers [23][24][25]. The activity coefficient of chloride in our experimental conditions (e.g. hyposmotic solution in the bath) was further checked by measuring the equilibrium potential by two Ag-AgCl electrodes (sensitive to chloride) which gave a value of 46.3 mV, in accordance with the theoretical value calculated using the activity coefficients for chloride provided by [25].
The cell capacitance was measured by means of the capacitance compensation circuitry of the voltage amplifier.
The relative open probability of the channel was obtained dividing the steady-state currents by (V-V Rev ), then the normalized conductance (G Norm ) was obtained by normalizing to the saturation value, plotted as a function of the driving voltage and best fitted by the Boltzmann distribution for a classical two state model, i.e.: where F, R and T have the usual meanings, z is the gating charge determining the steepness of the distribution, while V 1/2 is the half activation potential that depends both on z as well as on non-electrical work required to open the channel [13].

Results
Isolated hemocytes, displaying a large lysosomal compartment containing many granules, closely resemble (for dimensions and cell morphometric parameters, see Fig 1A-1C) granulocytes already investigated by other authors [2,[26][27][28]. After the transfer to the recording chamber, the cells initially displayed a rounded and ruffled shape, then, in few minutes (i.e. *<180 s), the hemocytes typically became very flat, firmly sticking to the bottom of the recording chamber ( Fig 1B and 1C). Moreover, minor modifications of the cell shape could be further observed with time, thus suggesting that other molecular mechanisms activate after the transfer to the bath solution.
We verified that the cell population did not show significant qualitative morphometric differences as well as adhesion properties on glass or plastic surface either in the hemolymph itself or in MASW: the increase of the two dimensional area of the cell was typically compensated by a decrease of the thickness which in many cases was reduced in the order of one micron or less (as qualitatively evaluated by our home-made opto-electronic platform). Despite the difficulties in performing patch-clamp experiments on these very flat cells, we were able to perform 94 electrophysiological recordings on mytilus hemocytes.

Basic electrophysiology
For electrophysiology experiments the hemocytes were typically first diluted in MASW then, when needed, the solution bathing the cell was changed by the slow or the fast perfusion procedure (see Materials and methods). In MASW we readily observed large time-dependent inward currents present mainly at negative membrane potentials (S1A Fig). These inward currents were frequently superimposed to other time-dependent components, such as the K + currents illustrated in S1B Fig This is not surprising as whole-cell currents typically comprise contributions mediated by different channels/transporters. Furthemore time-independent unspecific currents increasing linearly with the applied potential and overwhelming the signals of endogenous channels were occasionally observed.
Osmotic gradients may lead to alterations of the cell volume and to variations of water fluxes through the plasma membrane of hemocytes [27]. Furthermore the osmotic pressure of the ionic solutions and the transmembrane potential, two physical parameters important for cell survival, can also be used to separate the contributions to the ionic fluxes by different transporters.
Therefore, we decided to test whether we were able to isolate the hemocyte most significant current component by combining these two parameters, choosing two values of the osmotic pressure larger and smaller with respect to the osmotic pressure of the hemolymph (∏ = 1051, see Materials and methods). Furthemore, in order to acquire information on the ionic species carrying the charge, we further simplified our working conditions by adopting a solution with a low KCl concentration, a condition that favours larger inward currents. This choice also provides information on the selectivity of the channel(s) owing to the concentration gradient present between the bath and the internal pipette solution that, in the whole cell configuration, replaces the internal milieu of the cell.
Finally, on the basis of our previous experience [29,30], in order to reduce the leak-like timeindependent current, lanthanum chloride (LaCl 3 ) 0.5 mM was added to the bath solutions. This expedient allowed us to obtain high gigaseals, with minimal interference (I NoLa / I La = 0.9 ± 0.1, N = 7, data not shown), if any, on the properties of the endogenous time-dependent currents [29,30]. Therefore the majority of the experiments were done in the presence of 0.5 mM LaCl 3 in the bath solution.

Modulation of the inward current by hypotonicity
These conditions provided the typical current traces activated by V = -100 mV in hyposmotic (osmolality ∏ = 897 mosmol/kg) and hyperosmotic (∏ = 1178 mosmol/kg) conditions (see the profile in Fig 2A). The currents displayed in Fig 2C can be compared with a complete current family activated by hyposmotic conditions at different membrane potentials (see Fig 3B).
Indeed in the hyposmotic bath solution we could measure significantly larger time-dependent currents, with respect to the currents observed in the hyperosmotic solution. The current increase induced by the low osmotic pressure was reversible (as illustrated by the typical traces in Fig 2C) and by the plot of the steady-state current at V = -100 mV (Fig 2B) at the two different osmolalities. In order to verify whether the current increase in hyposmotic solution could be ascribed to any change in the membrane surface of the cells, we simultaneously monitored both the currents and the capacitances of the hemocytes in hyposmotic and hyperosmotic conditions ( Fig 2D). Interestingly while in hyposmotic solution the mean current was 3.5 times the value measured in hyperosmotic solution, the cell capacitance remained unaffected, thus indicating that the increase of the current could not be ascribed to an increase of the membrane surface and therefore to a recruitment of new channels by the fusion of endocytotic vesicles.

Voltage dependence of the inward current
In hyposmotic bath solution the slow inwardly-rectifying currents were typically elicited by hyperpolarizing pulses (see the voltage protocol in Fig 3A), activated slowly in a time-dependent manner (Fig 3B) and reached a steady-state plateau in a time lapse (dependent on the applied voltage) in the order of hundreds milliseconds. Finally these currents deactivated at repolarizing membrane potentials (see the tail currents in Figs 3B and 4A).
In order to quantify the dispersion of the current, the mean values of the steady-state currents were normalized to the steady-state current at V = -110 mV and the normalized current (I Norm ) was plotted as a function of the applied membrane potential (Fig 3C). From these data we could calculate the normalized conductance (G Norm in Fig 3D). In hyposmotic solutions, G Norm increased as a function of the membrane potential, with the tendency to saturate at large negative membrane potentials, as displayed in Fig 3D, where the Boltzmann distribution (continuous line) obtained from the best fit of the experimental data (filled symbols) is reported as a function of the applied potential. The gating charge z and the half activation potential (see Materials and methods) were: z(hypo) = 1.8± 0.1 and V 1/2 (hypo) = -37.2 ± 1.3 mV (n = 11).
The comparison of the Boltzmann distributions (Fig 3D) provides further support to the hypothesis that the activation of the inward current was favoured by the hypotonicity of the bath solution. Indeed the current decrease induced by the hyperosmotic solution is clearly due to a shift of the half activation potential of the current towards more negative membrane   Fig 3D) to hyperosmotic solution (empty symbols in Fig 3D).

The inward rectifying current is mediated by a chloride selective channel
Other evidences demonstrate that the inward currents were mediated by a chloride selective channel. Indeed, experiments performed in the absence of K + (i.e. substituting NMDG in the bath and in the pipette solutions, Fig 4A and 4B nor Na + could be responsible for the ionic currents mediated by the inwardly rectifying channel. Indeed, the reversal potentials (V Rev ) under diverse conditions, were always in accordance with a chloride selective channel. Furthermore, under a stimulation protocol applied to elicit a series of time-dependent currents (at V = -80 mV) followed by tail currents ranging from -80 mV up to +90 mV, in NMDG-Cl 50 mM in the bath and 530 mM NMDG-Cl in the pipette, V Rev was clearly comprised between +40 mV and +50 mV, as indicated by the two lines in Fig 4A and by the plot of the instantaneous tail currents vs. the tail potential in Fig 4B. Finally, in NMDG we were able to measure a mean reversal potential (after the correction for the Liquid Junction Potential) equal to V Rev = +40±2 mV (n = 8) at pH 7.6 as well as V Rev = +42±4 mV (n = 3) at pH 6.0: two values which are very close to what expected for a chloride-selective channel (V Nernst (Cl -) = +46 mV).
In , a value that also in this case is in accordance with the Nernst potential for chloride (V Nernst (Cl -) = +2 mV) and very far from the Nernst potentials for potassium (V Nernst (K + ) = -97 mV) and sodium which would be positive and indefinitely large.
Consistently with the value measured in the presence of internal and external NMDG, also in the hyposmotic KCl bath solution (S2C Fig), the reversal potential was %+ 50 mV a value which was again very close to the Nernst potential for chloride (V Nernst (Cl -) = +46 mV) and very distant from the Nernst potential for potassium (V Nernst (K + ) = -53 mV).
All these data consistently confirmed that potassium and sodium did not contribute to the time-dependent hemocyte currents. As some chloride transport proteins are chloride/proton antiporters, in order to verify the nature of the hemocyte chloride transporter, we also changed the pH of the hyposmotic bath solution: in the pH range from 6.0 to 8.0 (data not shown) we did not observe any variation of the reversal potential, thus providing a confirmation that the time-dependent currents were mediated by a chloride selective channel and not by a chloride/ proton antiporter [31,32].
Finally, when we replaced NaCl of the external MASW with an identical concentration of NaGluconate, a larger negative current was observed (S3 Fig). Consistently with other CLC-2 channels, this can be explained by the lower permeability of gluconate [33] with respect to chloride through the channel. The lower gluconate permeability shifts the reversal potential Boltzmann distributions was left-shifted towards more negative membrane potentials by as much as -44.9 ± 4.4 mV when the hyposmotic solution (filled circles) bathing the cell was replaced by an identical hyperosmotic solution (empty symbols at ∏ = 1178 mosmol/kg), thus implying that more negative membrane potentials are needed to activate the same current at higher osmolality. V 1/2 (hypo) = -37.2 ± 1.3 mV and V 1/2 (hyper) = -82.

Effects of zinc and zinc pyrithione on the inward rectifying channel
Looking for ions that may interact with the inwardly rectifying channel of mussel hemocytes, we verified that the addition of zinc to the bath solution determined a decrease of the current.  In order to verify whether the effects of zinc depend on the ionic form of the metal, we also tested zinc pyrithione, a well-known antifouling, antifungal and antibacterial agent where zinc is bound to two sulphur and two oxygen atoms. However, up to 30 μM ZnPT 2 did not affect the current appreciably: i.e. I ZnPT2 /I control = 1.1 ± 0.1 (n = 4, data not shown).

Discussion
Bivalve granulocytes are characterized by a large number of electron-dense internal granules: it has been reported that they represent the major population of cells present in the hemolymph of M. galloprovincialis [34]. In our experimental conditions, when the hemolymph was transferred to the petri dish chamber for the electrophysiological characterization, the hemocytes readily assumed a very flat configuration that made difficult to identify any distinct morphological characteristic. In addition, the mytilus hemocytes investigated by electrophysiological means displayed a series of different current components but no significant differences that might suggest the existence of different populations.

Regulatory volume decrease
Regulatory Volume Decrease (RVD) is generally achieved by the loss of ions and other osmolytes and the concomitant loss of water that is regulated by the transport of ions and/or osmolytes through the plasma membrane and the subsequent water efflux out of the cell. In many cell types, including bivalves [35], this behaviour is typically mediated by potassium and chloride electroneutral co-transport as well as VRAC (Volume-Regulated Anion Channels) which are typically inactive under resting conditions, but are able to contribute to a partial recovery of the cell size by a regulatory volume decrease mechanism in cells subjected to hypotonic stress [36][37][38].
It has been shown that mussels are sensitive to the chloride concentration of the bathing medium. In general, bivalves are able to survive to water chlorination by adopting defence strategies that induce the mussel to shut their valves as soon as they detect an anomalous chloride concentration [39,40]. It is well known that in molluscs the osmotic concentration of the internal medium follows the variations of the external environment and the hypotonic stress determines the swelling of diverse cell types, followed by the recovery of the original volume. For example, by using videometric methods it has been demonstrated that the cells from the digestive glands of M. galloprovincialis, exposed to rapid changes of the bathing solution (from 1100 to 800 mosmol/kg), undergo to a process of regulatory volume decrease [35]. Possibly the minor movements of M. galloprovincialis hemocytes (observed after their adhesion to the glass bottom) may depend on rearrangements due to a slow cell shrinkage (driven by the hypotonicity) that follows the faster reactions induced in the cells that perceive to be in a medium of different composition with respect to the hemolymph.

Ionic currents in mussel hemocytes
In mytilus hemocytes, beside the inward-rectifying channel, we also recognized a time-and a voltage-independent current component (increasing linearly with the voltage, data not shown) and occasional typical K + outward currents (S1B Fig) which could be ascribed to an n-type inactivating potassium channel [41,42] that looks very similar to animal potassium channel recorded, for example, in rat thymocytes [9].
A large number of different types of chloride channels such as cAMP-, calcium-, ligandand voltage-gated channels as well as volume regulated chloride channels are expressed almost ubiquitously in plant and animal tissues [43][44][45]. The kinetics and characteristics of the hemocyte inward channel are strongly reminiscent of the properties of the slow-activating volumeregulated chloride channel CLC-2 [38,[45][46][47][48]. This channel type is broadly expressed in a variety of vertebrate tissues-including brain, kidney, liver and heart-and cells-from epithelia to neurons. CLC-2 is inactive under basal physiological voltages and therefore it is supposed to be regulated by a series of parameters, such as the cell swelling [45]. Like in animal ClC-2 channel, in our working conditions, also the inward hemocyte channel displayed a strong dependence on the osmolality of the bathing solution.

Biophysical characteristics of the currents
The activation properties of the inward rectifying channel are well represented by the Boltzmann distribution that characterizes the normalized macroscopic conductance as a function of the applied potential and which provides information on the work to be done to open a voltage-dependent channel. Clearly hyposmotic bath solutions contribute to shift the range of activation (well represented by V 1/2 ) of the hemocyte inward-rectifying channel towards more positive membrane potentials. Interestingly, the steepness of the Boltzmann distribution does not change appreciably in Fig 3, thus indicating that the charges involved in the gating of the channel were not affected by the osmolality of the external solution. Consequently, the work required to open the channel seems to depend on additional non-electrical work that needs to be done in hyperosmotic conditions. Thus indicating a lower mobility of an uncharged segment of the protein which possibly plays a role in the channel opening.

Biocides
Copper and zinc are important contaminants of the marine environment owing to the large use of these metals as antifouling agents: zinc is frequently used as a weak primary biocidal pigment, but it is also adopted in combination with copper as a booster that increases hundreds fold the toxicity of Cu. As some organisms are resistant to inorganic metals, other agents, such as zinc pyrithione or copper pyrithione, are added as co-biocides to antifouling blends [49].
In addition, chemical modulators are useful tools to investigate the properties of ion channels. As it was demonstrated that ZnCl 2 and ZnPT 2 are able to modulate the activities of native and expressed ion channels [50][51][52], we verified whether the addition of micromolar concentrations of these two zinc compounds may have any effect on the osmoregulated chloride channel in hemocytes. On the addition of ZnCl 2 to the bath solution, in hyposmotic conditions one can observe a shift of the Boltzmann distribution towards negative membrane potentials. With respect to the control, this shift is definitely smaller but still appreciable compared to the shift observed on hyperosmotic conditions (see Fig 5). Interestingly, also in this case "z" remained almost unaltered (i.e. between 1.7 and 1.8 charge units, see the values reported in the legends of Figs 3 and 5). Incidentally, one can also observe that the decrease of the current induced by 30 μM Zn 2+ on the osmoregulated channel is in accordance with a comparable decrease of the CLC-2 chloride current which was reported to occur in native hyppocampal pyramidal cells that naturally express CLC-2 as well as in dorsal root ganglion cells overexpressing exogenous CLC-2 [53,54]. Interestingly, Zn 2+ also inhibits other chloride channels and transporters, such as ClC-1 and ClC-4 [45].
The observation that ZnPT 2 did not affect at all the chloride current of hemocytes possibly depends on the fact that pyrithione ligands are formally monoanions that chelate Zn 2+ via oxygen and sulfur centers. The pyrithione speciation with metals is relatively strong both in fresh and marine water and dissociation time is in the order of days (in the absence of light) while several hours are necessary under photolysis conditions [55][56][57]. Therefore we expect that during a typical patch-clamp experiment, ZnPT 2 remains almost unaltered. The fact that zinc pyrithione did not affect the conductance of the channel suggests that zinc must be in its divalent ionic form to be effective on the hypotonicity activated channel. However, the Boltzmann distribution (Fig 5) also suggests that the effect of Zn 2+ does not depend on a modification of the gating charge of the channel: possibly other mechanisms depending on the ionic charge of the metal could reduce the mobility of specific segments of the channel. In alternative Zn 2+ , but not ZnPT 2 may change some properties of the lipids surrounding the protein, that in turn might affect the channel properties [58].

Conclusions
Of the various systems that can contribute to RVD, swelling-activated chloride channels are present in a number of cell types. It has been shown that many cells and, among them, immunocells are able to slowly down regulate their volume after a rapid swelling under hyposmotic conditions [59]. It is well recognized that potassium and chloride transport [38] typically contribute to RVD in several cell types and specifically in immunocells, such as thymocytes from rats [59] and mice [60] as well as lymphoblastic leukemia cells [44]. Interestingly it has also been suggested that in Mytilus galloprovincialis digestive cells [35] and in Mytilus californianus gill cells [61] K + and Cl _ cooperate in RVD by the efflux of these two ions followed by an obliged efflux of water from the cell.
Owing to hypotonicity and voltage dependence of the hemocyte inwardly-rectifying current, we argue that, under basal physiological conditions, similarly to other channels involved in RVD [36][37][38] the inward channel has very small activity if any. Instead, it might be activated by a decrease of the osmolality of the external solution and/or by hyperpolarization: for example, hypotonic conditions could be determined by a dilution of sea water during heavy rainfall, river run off, climatic changes affecting the ocean conveyor belt. These processes may determine local and temporary decrease of water salts mainly at the sea surface. Since molluscs are osmoconformers [35], after a hypotonic water dilution the decrease of environmental [K + ] as well as other parameters and osmolytes [38] could determine a transient hyperpolarization of the cell and a simultaneous activation of inward chloride currents, i.e. outward chloride fluxes. In turn, this would induce a successive depolarisation and a parallel export of K + , Na + [62] and other organic compounds [38,61,63]. The net release of chloride and other ions as well as small organic molecules from the cell will contribute to partially counteract the osmotic stress avoiding a damage of the membrane due the fast cell swelling.
Furthermore, these mechanisms could allow the hemocytes to alert the organism that the external conditions are changing. Some compounds, such as ZnCl 2 but not ZnPt 2 , may interfere with these signals impairing the immunological response of the organism in critical conditions. s stimulation steps ranging from +50 mV up to -100 mV in -10 mV decrements. Holding and tail membrane potentials were +40 mV. B) Lower panel: macroscopic currents mediated by an outwardly rectifying channel in the hyperosmotic bath solution (i.e. 50 mM internal K + , ∏ = 1178 mosmol/kg) and standard pipette solution. Voltage pulses ranged from -50 mV to +100 mV in 10 mV increments. Holding and tail potentials were -80 mV and +20 mv, respectively. Currents were corrected for leakage. (TIF) S2 Fig. Selectivity properties of the inward rectifying current. A) Voltage protocol applied to reveal the tail currents in MASW. B) In MASW tail currents of the inward channel inverted at a potential (indicated by the arrow) comprised between 0 and 10 mV, a value compatible with the Nernst potential for chloride (V Nernst (Cl -) = +2mV) and very different from the Nernst potential for potassium (V Nernst (K + ) = -97mV). Standard pipette solution. C) Tail currents obtained in hyposmotic KCl standard solution. Tail voltages (indicated at the left side of the plot) ranged from 0 mV to + 70 mV. Also in this case a reversal potential of about +50 mV (indicated by the arrow) is in good agreement with the Nernst potential for chloride (V Nernst (Cl -) = +46 mV) and very different from (V Nernst (K + ) = -53 mV). (TIF)

S3 Fig. Gluconate is less permeable than chloride through the inward-rectifying channel.
Currents recorded in MASW and in an identical solution where 460 mM NaCl in the bath was substituted by Na-Gluconate. Currents were elicited by a main pulse to -100 mV from a holding and tail voltages at V = +40 mV. (TIF)