For the experiments chicken in the age from 5 to 21 days were used. After decapitation the eyes are removed out of the eye socket. Eyes are sectioned close to the equator and vitreous body is removed with a tweezers. The posterior eyecups are immersed in Ringer solution. The eyecups are then glued each in a Petri dish and put in the set up where they are perfused with Ringer solution. Before the measurements started the retinas are allowed to recover for 30 min.

The solution used to perfuse the isolated retinas has the following composition: 100 mM NaCl; 6 mM KCl; 1 mM MgSO₄; 1 mM CaCl₂ 2H₂O; 30 mM NaHCO₃; 30 mM glucose AND 10 mM TRIS with the pH adjusted to 7.4.

All chemicals used were obtained at least at p.a. grade at Sigma, Aldrich or Merck. Water was from a laboratory distillery and D₂O was acquired from Sigma.

The setup is enclosed in a Faraday cage in which a camera and a photomultiplier are aimed at the central retina region through a microscope. Electrical recordings were performed with extracellular glass microelectrodes (tip diameter around 10 mm, filled with potassium solution) inserted in the retina using a micromanipulator under optical control. The positioning was aimed to the inner plexiform layer. The electrical potential of the tissue relative to the bathing solution was measured then with a high impedance amplifier connected to an Ag/ AgCl-electrode in the glass micropippete versus another Ag/AgCl coil wire electrode immersed in the bath (reference electrode). The electrode amplifier output was digitized at 10 Hz and stored in a computer. The retina was maintained under perfusion at 1 ml/min rate and at 30˚C temperature. The digitized camera output had the electrode output superimposed before storage in a DVD recorder.

Figures 1 and 2 show the preparation and the recording procedure. In Figure 1 the panoramic view of a circular solitary wave of retinal spreading depression (RSD) is shown. The wavefront is seen approaching the tip of the pecten in central retina. The pecten is the black structure supposed to supply O₂ and nourishment to the avascular retina. This circular wavefront will invade the whole retina homogeneously.

Figure 2 shows the central retina region with a glass electrode inserted. Superimposed on the digitized camera output, a time series shows the temporal evolution of the field potential around the electrode tip. Upper row of photos: a RSD wave approaches the electrode and field potential begins to change; next the field potential reaches its peak value. Lower row: the field potential recovers to the baseline value; next the frame shows the moment when the second component of the optical profile of the RSD reaches peak. Note that this component has no clear field potential correspondence. The graphs on the side and below the photos detail the optical profile of circular RSDs recorded at 30˚C. On the side, the field potential and the optical changes in a matrix of 10 × 10 pixels (~50 µm × 50 µm) at the tissue over the tip of the electrode are given. The mean brightness of the pixel matrix is calculated and stored as time series.

The simultaneously recorded time series show the close relation between field-potential and IOS rising phase. Below the photos two complete profiles of RSDs are shown at two different space scales: the upper row

shows the mean brightness of a 10 × 10 matrix and the lower row the digitized output from the photomultiplier that samples a circular area with about 1mm diameter. Note the dissimilar rising phase of the first optical component and the similarity of the rising and shape of the second optical component at different space scales.

After successful insertion of the microelectrode in the innerplexiform layer, one control RSD was elicited mechanically and at the end of the optical profile record the normal solution was changed to the deuterium solution. The pumping rate was increased tenfold until the total volume of the perfused solution was three times the Petry dish volume. Then the pumping rate returned to the slow 1 ml/min rate. After one or two hours the solution was changed back to water solution.

The motivation for these experiments was the discussion in the literature about excitability in closed human cortex versus in vitro retina or open rodent cortices [26]: it was conceivable that rare spatiotemporal patterns in one experimental condition would be the most probable in another, hence the challenge of turning an in vitro retina model closer to the in situ cortex. We knew that in in-situ retinas the wave propagation pattern was more similar to closed circles compared to the calculated wave segments from patients with migraine [26,27]. We predicted that deuterium effects would shift the global excitability of the system towards a less excitable situation by disturbing the tight coupling between ionic and water flow in and out of cells. According to Pittendrigh et al. [18], the physicochemical properties of D₂O-(increase in viscosity) lower self-diffusion due to increased mass and different dissociation constant, would change ion mobility and stabilize the hydrophobic interaction in biopolymers, thus interfering in the tertiary and quaternary structure of protein complexes (and the electrochemical properties of the polyelectrolyte glycosaminoglycans— GAGs—of the extracellular matrix and basement membranes). Thus the solvent effects of deuterium would affect the system globally, disturbing macroscopic organized flows of ions and water and in consequence changing metabolism [22-24]. The D₂O effect on the Na⁺/K⁺-ATPase [18] predicted a long lasting recovery of the first peak and a longer interval between the first and second optical peaks of the IOS profile. The Q₁₀ of the IOS duration is 5—similar to the sodium pump Q₁₀— and the pumping rate by this protein complex is slowed down by D₂O in axons [18]. The foremost effect of the deuterium solutions in the two experimental models of excitable media we observed (retinas and B-Z) was a strong induction of non-stationarity and quick shift toward non-excitable state [27]. In our standard conditions with slow perfusion at 30˚C, the tissue transparency (measured as the baseline value of the photomultiplier output) increases only 2% to 3% in 8 to 10 hours, even with RSDs elicited each 20 minutes. Also in these conditions the retinas remain quiescent but readily responsive to mechanical, electrical or chemical stimuli and the shape and propagation pattern and velocity of RSDs remains stable, suggesting steady state conditions. By contrast, within seconds after the exchange to the deuterium solution and its contact with the basement membrane that covers the endfeet of the Müller cells, RSD waves were induced usually at the peripheral retina. Initially circular, they become self-organized spirals or segmentary and collapsing waves. Collapsing waves and

spontaneous standing patterns are rare events in our standard experimental situation, although standing patterns around lesions are frequent, and called “penumbra zones”. The same statement is true for the B-Z system: we only saw standing patterns with deuterium as solvent in the system [27].

The deuterium solution was applied to seven retinas and in five the pattern observed is the one shown in Figure 3.

The other two developed excitoxic responses and were dead within the first hour as shown in Figure 4.

The long time series shown in Figure 3 displays the RSDs IOS recorded in one experiment. The record begins at 40 minutes of the recording window with the optical profile of a control RSD and ends 90 minutes later after a series of RSDs invaded the central region sampled by the photomultiplier. At the end of the control wave profile optical artefacts show the record of the turbulence in the system due to the fast pumping rate of the D₂O solution. Within seconds of the contact between the D₂O solution and the basement membrane of the retinal vitreal interface, RSDs were induced. Initially circular, waves from the border could collide in the central region (third wave in the series is a collision of two fronts) then the wavefronts became irregular. Besides the irregular wavefronts, the profiles showed gradual increase of the delay between the first and second optical components, together with gradual disappearance of the first peak in the IOS. Discussions of the details of individual profiles are beyond the scope of this paper and will be discussed in another context about propagating waves and excitoxic responses in retinas.

The interpeak interval in the optical profiles showed gradual increase; usually it was doubled after 40 minutes of retinal exposition to deuterium solution Glycerol and a lowering step in temperature also increase the interpeak interval [9,21]. While the temperature effect is mainly attributable to the sodium pump (Q₁₀ = 5) and does not disturb coupling in the system, the glycerol effect more probably is caused by the AQPs channels activity. The AQPs linked to functional syndromes are AQP4 and AQP9, both expressed in glia [14-16] and in the Müller cells their expression appears to follow the surface to volume ratio of membranes, being heavy at the vitreous retina border and in the innerplexiform layer [15,16]. The AQP9 channel is permeant to water, glycerol and lactic acid. The later is very closely associated with the kinetics of the second peak of the RSD IOS and with neuronal-glial interactions [4,24]. The deuterium and glycerol solutions are likely to disturb the coupled flows of ions and water and lactic acid. For this reason, whereas low temperature increases the amplitude of the second optical peak of the IOS, deuterium and glycerol tend to depress the amplitude. At the limit of the system toward non-excitability, the second component can be absent.

The other two retinas of this experimental series developed excitotoxic responses to deuterium and were dead within one hour of exposition to the deuterium solution. The recordings of one of these experiments are summarized in Figure 4. The three time series were simultaneously recorded and the one labelled macroscopic IOS shows the photomultiplier outpu. The field potential is the microelectrode/amplifier system output. Additionally the microscopic IOS, the mean brightness of a 10 × 10 pixel matrix, overlaying the electrode tip (best experimenter guess) is given in arbitrary units. Note that in the first optical time series, the baseline transparency of the tissue changes very quickly compared to the pattern show in Figure 3, in which the loss of transparency shows slow steadily growth. The two artefacts labelled S1 and S2 show the mechanical stimulation followed by two RSDs. In the first of these waves the first optical peak is similar to controls (see Figure 2) at both spatial scales and the field potential also has the amplitude and shape of control waves. There was no discernible second optical component at either spatial scale. The second RSD shows an increase in the duration of the first optical peak with slow recovery phase and depressed amplitude of the field potential which shows slow and two phase recovery. A second optical peak can be seen after a 15 minutes interval at the microscopic scale IOS only. This interpeak interval is five fold the one measured in control conditions (see Figure 2), also the shape and amplitude of this component is altered and small compared to controls. The alterations in the shape of the IOS are similar but more marked in these two retinas (from the same animal) than in the other five, suggesting marked uncoupling of ionic and solvent flows followed by disruption of the homeostatic metabolic response to the dissipation of electrochemical gradients at wave onset

[6,24,28]. The fast linear rise in the photomultiplier time series marks the onset of the excitotoxic tissue response that culminated in tissue death. This linear growth looks the same irrespective of the origin of the excitotoxicity (v.g. glutamate or ouabain, data not shown). Note that at amicroscopic scale a step is observed and the photomultiplier output, then the macroscopic signal represents the integration of thousands of such steps arising in short sequence within central retina. At the beginning of the field potential time series, a linear fast shift of the baseline is observed. This shift follows a potential drop at the silver/silver-chloride coil reference electrode. Our recording arrangement makes the maintenance fluid to enter the chamber at the bottom and be sucked out at upper border of the Petry dish. Thus long before the solution touch retina surface, the reference electrode contacts it. This electrochemical shift at the electrode solvent interface has the amplitude of ~15 mV and shows the potential of the electrochemical effects of the deuterium solution at charged interfaces.