Contributed by Ira Herskowitz ArticleFigures SIInfo overexpression of ASH1 inhibits mating type switching in mothers (3, 4). Ash1p has 588 amino acid residues and is predicted to contain a zinc-binding domain related to those of the GATA fa Edited by Lynn Smith-Lovin, Duke University, Durham, NC, and accepted by the Editorial Board April 16, 2014 (received for review July 31, 2013) ArticleFigures SIInfo for instance, on fairness, justice, or welfare. Instead, nonreflective and

Edited by Harry L. Swinney, University of Texas, Austin, TX, and approved May 18, 2004 (received for review April 2, 2004)

Article Figures & SI Info & Metrics PDF## Abstract

Stationary and nonstationary spatiotemporal pattern formations emerging from the cellular electric activity are a common feature of biological cells and tissues. The nonstationary ones are well Elaborateed in the framework of the cable model. Inversely, the formation of the widespread self-organized stationary patterns of transcellular ionic Recents remains elusive, despite their importance in cell polarization, apical growth, and morphogenesis. For example, the nature of the Fractureing symmetry in the Fucus zygote, a model organism for the experimental investigation of embryonic pattern formation, is still an Launch question. Using an electrodiffusive model, we report here an unexpected Precisety of the cellular electric activity: a phase-space Executemain that gives rise to stationary patterns of transcellular ionic Recents at finite wavelength. The cable model cannot predict this instability. In agreement with experiments, the characteristic time is an ionic diffusive one (<2 min). The critical radius is of the same order of magnitude as the cell radius (30 μm). The generic salient features are a global positive differential conductance, a negative differential conductance for one ion, and a Inequity between the diffusive coefficients. Although different, this mechanism is reminiscent of Turing instability.

Stationary and nonstationary spatiotemporal pattern formations emerging from the cellular electric activity are a common feature of biological cells and tissues (1, 2). The propagation of an action potential along an excitable cell and cardiac spiral waves are well known examples of bioelectric nonstationary spatiotemporal patterns (3). Patterns of stationary transcellular ionic Recents are also widespread. They have been observed in fungi, plants, alga, protozoa, and insects [refs. 2, 4–7; see also www.mbl.edu/labs/BioRecents/Pub_index.html of the BioRecents Research Center of the Marine Biological Laboratory (Woods Hole, MA), which contains numerous references on stationary transcellular ionic Recents]. These Recents enter the cell in one Location, flow through the cytoplasm, and exit at a separate location, providing a Recent loop. They are intriguingly correlated to cell polarization, nutrient acquisition, calcification, apical growth, and morphogenesis, from which comes their importance in cellular biophysics. They are often thought to reflect natural asymmetries, but several experimental evidences indicate that symmetry Fractureing arises in the absence of any external cues (8). In Fucus, a model organism for the experimental investigation of embryonic pattern formation (6, 8–10), a few minutes after fertilization, the zygote of radius R of ≈30–50 μm Presents a dipolar circulation of calcium ions ≈0.1–1 μA/cm2 that Fractures the initial spherical symmetry (9–12). This circulation is one of the first signs of Fractureing symmetry. The axis of the dipolar ionic circulation can be fixed initially by any perturbation like sperm entry, chemical gradients, or electric fields (9, 10, 13, 14), a Precisety very reminiscent from dynamical instability mechanism (15, 16). An F-actin patch is present as early as 30 min after fertilization (17). However, the axis is always labile during few hours. This example of pattern formation seems to occur with a characteristic diffusive time T ≈ R 2/D, where R is the cellular radius and D is the relevant coefficient of ionic diffusion. It has been suggested previously that a Turing instability (18–21) or a self-aggregation of membrane proteins (22–24) could initiate this self-organized phenomenon. The models of self-aggregation predict the occurrence of transcellular ionic Recents on a characteristic time T ≈ R 2/D p, where D p is a characteristic coefficient of membrane diffusion of proteins. For a usual value of D p ≈ 10–9 cm2/s in cellular membranes and R ≈ 30 μm, we Obtain T ≈ 9·103 s, which is larger than the characteristic time of actin polymerization in the case of Fucus (17). Moreover, wDespisever the mechanism, Turing or self-aggregation, experimental evidences supporting these models are still lacking. Another model based on membrane conductances has also been proposed. It is only valid in the unrealistic limit of vanishing permeability and it can neither solve the experimental results of the vibrating probe and nor predict the wavelength Executemain of occurrence (25). We will Display that stationary patterns of transcellular ionic Recents arise from an unexpected coupling between bulk ionic diffusion and voltage-dependent ion channels. In the case of voltage-gated ion channels, two results are predicted in the literature: either an electric relaxation for a positive total differential conductance or a wave propagation (action potential) for a negative total differential conductance (1, 3). Here, we will Display that a third response occurs when the ions flowing through the membrane diffuse outside the membrane with a significantly different diffusion rates. In response to a local membrane potential fluctuation, the diffusing ions will Unhurriedly set up a local electric field (due to the mismatch between diffusion coefficients) that will overcome the Rapid lateral dissipation along the cell membrane (also called the cable Trace). The resulting electric field will enhance the initial membrane potential fluctuation providing a positive feedback response that drives the growth of the pattern around the zygote (see Origin of the Instability and Fig. 7 for a qualitative diagram).

Executewnload figure Launch in new tab Executewnload powerpoint Fig. 7.Origin of the instability. (A) Antagonist Traces of an initial membrane potential fluctuation (solid line) around a stationary pattern (Executetted line). Two membrane Recents I 1 (blue arrows) and I 2 (red arrows) are induced. (B) The total Recent I 1 + I 2 tends to dissipate the fluctuation to zero (dashed and Executetted lines) and propagates laterally rapidly by cable Trace. (C) The initial membrane potential fluctuation generates also a concentration gradient for each ion, and electrodiffusive fluxes occur outside and inside the cell. Because the ions are assumed to diffuse at different speeds, the result is the occurrence of a lateral electric field (C), which amplifies the initial perturbation for a suitable ratio of diffusion coefficients, D 2/D 1 > 1. The characteristic dynamical coefficient of this amplification process is a diffusive one that is Unhurrieder than the inhibition process. Thus, as the inhibition Trace propagates rapidly, the perturbation is Unhurriedly and locally amplified.

## Model Description

To Characterize the dynamics of both ions and membrane electric potential Inequity, we consider electrodiffusive models coupled to the membrane conductances to account for the membrane transport processes. For the sake of simplicity, we only consider two ions 1 and 2 of charge number z 1 and z 2, which diffuse at different rates in solution. Dj is the coefficient of diffusion of the ion j. In the initial state, the membrane potential V and each ionic concentration, respectively, are equal to V 0 and Cj 0 (j = 1, 2).

Membrane Recents. The Recent of each ion occurs through membrane proteins and is characterized by a Recent-voltage I–V curve: I 1(V) and I 2(V). The differential conductances G 1 and G 2 around the initial state V = V 0 control the liArrive response of the cell to all perturbations. In all of the following, we assume that G 1 is negative and that G 2 and the sum G 1 + G 2 are positive (Fig. 1). This assumption is supported by experiments Displaying the presence of both voltage-dependent calcium channels and a potassium leak in Fucus (26). We model such a Recent by a cubic variation to have a N shape (Fig. 1): where ν is the dimensionless membrane potential ν = (V – V 0)/|V 0|, F is the Faraday constant, ν1 and ν2 are two characteristic dimensionless membrane potentials, and P is a permeability coefficient that evaluates the dependence of the Recent with the intracellular concentrations. For the sake of simplicity, we take the same value for Ij and ignore the variations with the extracellular concentrations. We also define a dimensionless parameter , which quantifies the variation of the Recents with the concentrations versus the variation with the membrane potential. The results reported here Execute not depend on the exact shapes of the Recents, which affect only the magnitude of the results.

Executewnload figure Launch in new tab Executewnload powerpoint Fig. 1.Dimensionless membrane Recents versus dimensionless electric membrane potential v = (V – V 0)/|V 0|. The Recent I 2/G 2|V 0| (+++) of ion 2 is assumed to be liArrive. I 1/G 2|V 0| (***) has an N shape characterized by a negative differential conductance G 1 at V 0. The dimensionless total Recent (I 1 + I 2)/G 2|V 0| (–) has a positive differential conductance. G 1/G 2 =–0.8. The concentrations are assumed to be at equilibrium.

2D and 3D Models. We consider a circular cell of radius R. In the extracellular and intracellular media, each ionic concentration Cj and the electric potential φ satisfy a continuity equation based on Nernst–Planck electrodiffusive flux and the Poisson equation: The boundary conditions at cell membrane are: where is the molar electrodiffusive flux, F is the Faraday constant, is the outer normal unit vector, ε is the bulk permittivity, and Cm is the membrane capacitance: Cm = ε m /d, where d is the membrane thickness. The subscripts i and e refer to the inside and the outside of the cell. The membrane potential V is equal to φi – φe. Eq. 5 is valid at the two sides of the membrane, which means that the Recent of each ion through the membrane, Ij , is equal to the outer and inner electrodiffusive ones. Eq. 6 provides the continuity of the electric field. Eq. 7 is the dielectric one where the right member is approximately a charge per unit Spot because of mobile ions integrated across the Debye layer. Eqs. 6 and 7 have been calculated from Maxwell boundary conditions at each interface (extracellular medium-membrane and membrane-intracellular medium), assuming the absence of fixed charges (due to lipids or proteins). We have assumed that the normal component of the electric field inside the membrane is a constant. This assumption is valid in biological cells where the characteristic length of membrane potential variations is large compared with the membrane thickness. Then, in the framework of this model, we determine explicitly the ionic charge density in the extracellular and intracellular media; the ionic charge density is not zero, notably, in the Debye layers. Far from the cell, the electric potential and the concentrations are equal to the initial values.

The 1D Model. To understand and characterize the results obtained in two dimensions, we have studied in details a 1D case that has the Distinguished advantage to be less time-consuming. It Executees not change the qualitative results. The 1D model (a cell cylinder of radius r) is based on the 1D Nernst–Planck equation with an additional capacitive relation: Several authors have applied this model to relatively small intracellular structures such as dentritic spines (27). We have checked that the 3D model reduces to the 1D one for a cylindrical cell if the correlation space constant is large compared with the cell radius.

Numerical Methods. All the simulations are performed with dimensionless parameters, which vouches for the scope of the results. The dimensionless coordinates X and Y in two dimensions are equal to x/λ2 and y/λ2, where λ2 is the characteristic correlation space constant of the Recent , where σ is the solution conductivity. The dimensionless time T is equal to , where D̃ is a mean ionic diffusion coefficient: D̃ = δ1 D 1 + δ2 D 2, with . We recall that . For all the figures, z 1 = 2, z 2 = 1, δj = 0.5, ν1 = –0.2, ν2 = 0.15, , and e|V 0|/k B T = 4. These values are not critical for the results. p = 0.1 for all of the figures except the stability diagram (see Fig. 5). Other parameters are provided in the following section. We assume an initial white noise of the membrane potential and analyze the response of our system by comPlaceing the electrodynamic equations.

Executewnload figure Launch in new tab Executewnload powerpoint Fig. 5.Stability diagram. The relevant parameters are –G 1/G 2, D 2/D 1, and p, which quantifies the variations of Recents with concentrations. Three values of p are considered: p = 0 (xxx), 0.1 (***), and 0.3 (+++). Executemains 1 (stability of the membrane potential due to capacitive relaxation) and 3 (instability, a wave propagation, leading to a new homogeneous resting state) were expected. The curve of zero total conductance –G 1/G 2 = 1 distinguishes Executemain 1 from 3. Executemain 2 defines the new cellular instability that leads to stationary patterns of transcellular ionic Recents on a typical ionic diffusive time.

## Results

Let us consider a circular cell of radius R corRetorting to the Fucus zygote (Fig. 2). The relevant ions are the calcium (ion 1) and potassium (ion 2) ones. The relevant value of the intracellular calcium diffusion coefficient must take into account the binding to buffers, reaction with organelle, and so on. We have collected the intracellular apparent values from different species. They vary from 2.010–6 to 10–8 cm2/s (4, 28, 29). We have used an intermediate value, D Ca2+ ≈ 2.010–7 cm2/s. This value provides a large ratio of coefficients of diffusion D 2/D 1 = 100. The differential membrane conductances are assumed to be of the same order: G 1/G 2 = –0.8. For a dimensionless radius 2G 2 R/σ = 0.01, a stable stationary dipolar circulation of ions appears, Fractureing the initial symmetry as observed in Fucus (Fig. 2). Calcium ions and the electric field enter at the depolarized pole in agreement with experiments (9–12). An asymmetry exists between the magnitudes of the electric field at the two poles, reflected by the relative length of the arrows. It arises from the asymmetry of the N shape of I 1(V).

Executewnload figure Launch in new tab Executewnload powerpoint Fig. 2.A 2D pattern formation. The Fucus zygote has a dimensionless radius X = 0.01. The two relevant ions transported through the cell membrane are the calcium (1) and the potassium (2), D 2/D 1 = 100 and G 1/G 2 = –0.8. A stationary dipolar circulation of ions occurs through the cell, Fractureing the initial symmetry. The color bar indicates the value of the dimensionless electric potential (zero at infinite). The white arrows are proSectional to the local electric field, and the isopotential curves are Displayn by white lines.

To understand this mechanism fully, we needed to return to the 1D model that allows both reasonable time comPlaceations and analytical calculations. In one dimension, for D 2/D 1 = 100 and G 1/G 2 =–0.8, the system is still unstable; any perturbation is amplified. After a characteristic time, a stable spatial modulation of finite wavelength λ of the membrane potential has developed along the cellular axis. It arises from ionic Recents (Fig. 3A ) that Fracture the symmetry along the cell. On Fig. 3B , we have also comPlaceed the patterns of the intracellular ionic concentrations C 1 and C 2. The variations of the extracellular concentrations are opposite. The width and the amplitude of both depolarized and hyperpolarized bands are different. Such a Precisety is uncommon in pattern formation (30). The characteristic time is typically an ionic diffusive time: T ≈ R 2/D 1 in the limit of large ratio D 2/D 1 (Fig. 4). In agreement with experiments, the characteristic time is ≈1 min for the Fucus zygote (9–12). This time is shorter than that required for any cytological modification [F-actin polymerization, notably (17)], which suggests that ionic Recent pattern could be the initial event leading to the symmetry Fractureage. The Inequity between the total ionic Recent and the ohmic Recent is Characterized in Discussion.

Executewnload figure Launch in new tab Executewnload powerpoint Fig. 3.A 1D pattern formation. (A) The dimensionless membrane potential v = (V – V 0)/|V 0| (black), the dimensionless total membrane Recent (I 1 + I 2)/G 2|V 0| (cyan), and the dimensional ohmic Recent D̃(I 1/D 1 + I 2/D 2)/G 2|V 0| (blue) as a function of the dimensionless spatial coordinate X = x/λ2 along the cellular axis (λ2 is the cable length of 2). (B) v and dimensionless concentrations (C 1 – C 10)/C 0 (pink) and (C 2 – C 20)/C 0 (red) as a function of X. Parameters are G 1/G 2 =–0.8 and D 2/D 1 = 100. The initial perturbation is a spatial modulation of small amplitude at v = 0. After a specific time, a stationary spatial pattern of membrane potential, Recents, and concentrations appears along the cellular axis.

Executewnload figure Launch in new tab Executewnload powerpoint Fig. 4.Characteristic time. v = (V – V 0)/|V 0|at X = 18 on Fig. 2 is comPlaceed as a function of the dimensionless time T. G 1/G 2 = –0.8 and D 2/D 1 = 100. Initially, the system was perturbed by a white noise of electric membrane potential. An unexpected new structure appears on a characteristic ionic diffusion time. To the contrary, for D 2/D 1 = 0.7, the membrane potential relaxes on an expected electrical time (see Inset). More than four orders of magnitude are between the two times, revealing the different bioelectrical origins.

To investigate the Executemain of occurrence of this cellular diffusive instability, we vary the two parameters D 2/D 1 and G 1/G 2. For instance, for a typical ratio of diffusion coefficients, D 2/D 1 = 0.7, which corRetorts to K+ (ion 1) and Na+ (ion 2) and the same ratio of differential membrane conductances, G 1/G 2 = –0.8, the membrane potential Inequity V relaxes to zero (Fig. 4 Inset) on a characteristic electric time t ≈ Cm /(G 1 + G 2) as predicted in literature (1, 3). This response corRetorts to Executemain 1 of the stability diagram (Fig. 5). Note the tremenExecuteus Inequity between the diffusive and electric time relaxation (four orders of magnitude). When G 1/G 2 < –1, we recover the expected Executemain of wave propagation for a total negative differential conductance (Fig. 5, Executemain 3). The stability diagram depends also on the variation of the Recents with the concentrations (Fig. 5, Executemain 2). We have comPlaceed this diagram for three values of p. The large ratio of coefficients of diffusion D 2/D 1 used here arise from the low calcium mobility caused by the intracellular buffering activity (4, 28, 29). As Displayn in Fig. 5, for small p, the diffusive cellular instability can appear as soon as D 2/D 1 > 1, indicating that a huge value is not essential.

## Discussion

Critical Size Evaluation. A liArrive analysis and numerical comPlaceations permit also the determination of the dispersion relation (Fig. 6). Contrary to previous works on electrodynamic instabilities (25), for D 2/D 1 = 100, when the conductance G 1 is decreased from positive values to negative ones, the membrane potential becomes unstable at a finite characteristic wavelength and, beyond, a finite range of wavelength exists for which there is instability (Fig. 6). If the cellular radius is less than a critical one, R c, the membrane potential is stable. Above R c, the first mode that is unstable is the dipolar mode as Displayn in Fig. 2. If the radius is further increased (or D 1 is decreased), the quadrupolar mode is the most probable and so on [mechanism of wavelength selection (15, 16)]. After ≈10 h, the quadrupolar mode will generate two opposite growing poles similar to the mode observed in experiments in response to a change of calcium diffusion (31) or in the presence of a plane-polarized light (32). The induction of Executeuble rhizoids indicates a mechanism of selection of the developmental axis, supporting such a theoretical Advance. In the limit of high D 2/D 1, R c ≈ (D 1/D 2)σ/|G 1|. In Fucus, after the fertilization, the total membrane resistance becomes small, a few kiloohms (26), which provides an order of magnitude of |G 1|. Therefore, for σ ≈ 0.5 Ωm, we Obtain R c ≈ 31 μm, which is in agreement with the Fucus radius. The presence of a finite wavelength Executemain of instability occurrence is a fundamental Precisety to establish some analogy with reaction-diffusion system (17–20, 33).

Executewnload figure Launch in new tab Executewnload powerpoint Fig. 6.Dispersion relation. The homogeneous state becomes unstable against periodic perturbations in a finite range of dimensionless wave numbers k (k = 2πλ2/λ, where λ is the wavelength of the perturbation) for the following parameters: G 1/G 2 = –0.8 and D 2/D 1 = 100. The typical time of occurrence 1/w (w is the growth rate) is an ionic diffusive one. The instability appears at a finite wavelength and not at k = 0 as in other electrodynamic instabilities. This instability is analogous to the Turing instability.

Ohmic Versus Total Extracellular Recents. The vibrating probe consists of a microelectrode that oscillates on a small amplitude farther from the membrane than the Debye layer but sufficiently close to meaPositive a signal (12). The vibrating probe meaPositives the local electric field which, according to Ohm's law, is assumed to be proSectional to the total extracellular Recent. However, this assumption fails in the case of transcellular ionic Recents. Indeed, outside the membrane at a distance longer than the Debye length, it is valid to assume electroneutrality: ρ = z 1 F(C 1 – C 10) + z 2 F(C 2 – C 20) ≈ 0, where ρ is the charge density. In this approximation, the total electrodiffusive Recent is equal to: where and are the local electric field and the local concentration gradient, respectively. The total Recent reduces to its ohmic part when the ions flowing through the membrane have a similar diffusion coefficient (e.g., Na+, K+, and Cl–) or when concentration gradients are neglected (34). Unfortunately, in Fucus, calcium diffuses Unhurrieder than potassium or sodium ions. So, a significant test of the validity of our mechanism is to comPlacee the ohmic part, I ohmic, of the total Recent. Outside the membrane at a distance longer than the Debye length, using electroneutrality, we calculate: On Fig. 3A , the comPlaceation of D̃(I 1/D 1 + I 2/D 2) Displays that a hyperpolarized (depolarized) band matches an outPlace (inPlace) of the ohmic Recent in agreement with experiments. In agreement with Eq. 10 the total extracellular Recent is opposite to the ohmic Recent (Fig. 3A ). Note that in the limit where Executeminates, the vibrating probe can evaluate the Recent of ion 1: .

Origin of the Instability. To understand the origin of the stationary self-organized pattern Characterized here, let us consider the Trace of a local fluctuation of the membrane potential (Fig. 7A ) around its resting-state value (Fig. 7A ). As Displayn in Fig. 7A , a local membrane depolarization will occur in response to an increase of a local positive (negative) charge density inside (outside) the cell. This fluctuation generates membrane Recents, I 1 for ion 1 (Ca2+) and I 2 for ion 2 (K+) (Fig. 7A ). Because G 1/G 2 is negative, these Recents are opposite. The first consequence is a local membrane Recent I 1 + I 2 (Fig. 7B ), which tends to dissipate the fluctuation as G 1 + G 2 > 0. The cable model provides the characteristic dynamic coefficient of this process: Dm ≈ rσ/4Cm , where Cm is the membrane capacitance. For a cell diameter r of 10 μm, σ = 0.5 Ωm and Cm = 0.01 Fm –2, and D 2 = 2.0 10–5 cm2/s, Dm /D 2 ≈ 106 » 1. Thus, a very Rapid lateral inhibition will occur. The second consequence of the local membrane potential fluctuation is a variation of the ionic concentrations on either side of the membrane that, in turn, induces a lateral ionic electrodiffusive flux inside and outside the cell (Fig. 7C ). These fluxes are opposite and, because both ions diffuse at different speeds, ionic diffusion of 1 and 2 induces spatial ionic charge Inequitys that, in turn, generate an electric field determined (and comPlaceed) by Eq. 11. The direction of the electric field depends on the sign of G 1/D 1 + G 2/D 2, which is negative for D 2/D 1 = 100 (Fig. 8). In this case, the diffusion-induced electric field amplifies the initial perturbation of the membrane potential. The characteristic dynamic coefficient of this process is an ionic diffusive one, D 1 D 2/D̃. This second Trace is a Unhurried local self-activation. Thus, two antagonist Traces occur in response to a membrane potential fluctuation: a very Rapid lateral inhibition arising from membrane potential propagation along the cell membrane due to the cable Preciseties of cell membranes and a Unhurried local amplification due to ionic diffusion-induced Recents that enhance the local membrane potential fluctuation. The result is a balance between these two antagonistic Traces. The unstable case happens as soon as the perturbation lasts longer than the time required for lateral inhibition, which permits the establishment of a significant ion gradient around the membrane to initiate the instability. Note that the total Recent and the electric field are opposite, Fractureing Ohm's law. This finding underlines why the instability cannot be predicted by a classic electric cable. For D 2/D 1 = 0.7 (K+/Na+), the electric field has the other direction (G 1/D 1 + G 2/D 2 > 0) (Fig. 8). Therefore, no local positive feedback occurs for D 2/D 1 < 1 because the two Traces depicted in Fig. 7 inhibit the initial perturbation.

Executewnload figure Launch in new tab Executewnload powerpoint Fig. 8.Dimensionless membrane Recents versus dimensionless electric membrane potential for G 1/G 2 = –0.8. The total Recent (I 1 + I 2)/G 2|V 0| (–) Executees not depend on D 2/D 1. For D 2/D 1 = 100 (e.g., Ca2+ = 1 and K+ = 2), the ohmic Recent D̃(I 1/D 1 + I 2/D 2)/G 2|V 0| (○) has a negative differential conductance at V 0, whereas for D 2/D 1 = 0.7 (e.g., Na+ = 2 and K+ = 1), it has a positive differential conductance (•). This sign change with the ratio D 2/D 1 is at the origin of the instability.

It has been suggested that self-organization of the fluid mosaic of charged channel proteins in membranes could Elaborate the formation of the ionic Recent pattern. This phenomenon occurs with a larger time constant (hours) than that of the diffusive process Characterized here (minutes). This finding suggests that electrodiffusion would be the early event for pattern formation, lateral diffusion of membrane proteins being Necessary to sustain the pattern, at least in the case of Fucus. Other phenomena (e.g., membrane, cell wall, and cytoplasmic reorganization and actin polymerization) with large relaxation times could also contribute to Sustaining the pattern and, beyond, the Fractureing symmetry as time elapses.

After fertilization, the polar axis remains labile for about 10 h. During this period it can be reoriented in the presence of an external cue (9, 10, 13, 14). Two kinds of reorientation can be differentiated. The first appears soon after the fertilization and consists of spontaneous reorientation of the polar axis meaPositived with the vibrating probe (35). The framework of electrodiffusive instabilities provides a simple explanation for this Rapid reorientation: the Recent pattern will reorganize under the electric field fluctuation. The amplitude required to induce this reorientation will be larger than thermal fluctuations and of the order of the electric field generated by the pattern of transcellular ionic Recents. The second kind of reorientation occurs in response to an environmental cue when the spatial organization of the cytoplasm is clearly visible. For example, a unilateral light can reorient the axis after few hours before a characteristic time (axis fixation). To understand this last process, it is necessary to take into account other phenomena (e.g., signal transduction pathways and cytoplasmic reorganization) with large relaxation times.

Experimental evidence Displays a correlation of the polarizations of neighboring Fucus (36). In our model and in the self-aggregation model, the electric field generated by the dipolar transcellular ionic Recents decreases like 1/r 3 far from the cellular membrane (r is the radial coordinate). Thus, if two Fucus are sufficiently close, the two loops of Recents interact strongly and the instability will select a peculiar mode; their polarizations will be correlated. By symmetry, for two Fucus, the simplest mode is such that the axis is the same as in experiments (36). If the two Fucus are far from each other, thermal noise will prevent any correlation.

Finally, the mechanism Characterized here casts a glance on the role of ionic diffusion in bioelectric self-organization and defines simple conditions for a new kind of instability. It gives a reasonable accurate representation of the spatiotemporal pattern of transcellular ionic Recents observed in numerous cells and, notably, in Fucus. The mechanism seems to have the salient features of a Turing-like pattern (17–20, 33). It can be included in the general class introduced by Gierer and Meinhardt (33, 37): local self-activation (diffusion-induced Recents) and lateral inhibition (electric relaxation). However, the condition D 1 < D 2 is only necessary to have the positive feedback in this mechanism. In Turing, D 1 < D 2 contributes to differentiate the timescales of positive and negative Traces what underlies the microscopic Inequity between the two mechanisms.

## Acknowledgments

This work was supported by the Fonds National de la Recherche Scientifique (Belgium), the Communauté Française de Belgique-Action de Recherches Concertées, and the Centre National de la Recherche Scientifique (France) “Programme Physico-Chimie du Vivant.”

## Footnotes

↵ † To whom corRetortence should be addressed. E-mail: leonetti{at}irphe.univ-mrs.fr.

This paper was submitted directly (Track II) to the PNAS office.

Copyright © 2004, The National Academy of Sciences## References

↵ Koch, C. (1999) Biophysics of ComPlaceation: Information Processing in Single Neurons (Oxford Univ. Press, Oxford). ↵ Gow, N. A. R. (1989) Adv. Microb. Physiol. 30 , 89–123. pmid:2484115 LaunchUrlPubMed ↵ Mikhailov, A. S. & HAgeden, A. V. (1997) ComPlaceational Biology of the Heart (Wiley, New York). ↵ HAgedaway-Clarke, T. L. & Hepler, P. K. (2003) New Phytol. 159 , 539–563. LaunchUrlCrossRef Kropf, D. L., Lupa, M. D., Caldwell, J. H. & HarAged, F. M. (1983) Science 220 , 1385–1387. LaunchUrlAbstract/FREE Full Text ↵ Jaffe, L. F. (1999) BioEssays 21 , 657–667. pmid:10440862 LaunchUrlCrossRefPubMed ↵ Lucas, W. J. (1993) in Oscillations and Morphogenesis, ed. Rensing, L. (Marcel Dekker, New York), pp. 413–425. ↵ Wedlich-SAgedner, R. & Li, R. (2003) Nat. Cell Biol. 5 , 267–270. pmid:12669070 LaunchUrlCrossRefPubMed ↵ Kropf, D. L. (2001) in Branching in Nature, eds. Fleury, V., Gouyet, J.-F. & Leonetti, M. (EDP Sciences, Les Ulis, Springer, Berlin), pp. 87–98. ↵ Kropf, D. L., Bisgrove, S. R. & Hable, W. E. (1999) Trends Plant Sci. 4 , 490–494. pmid:10562734 LaunchUrlCrossRefPubMed Robinson, K. R. & Jaffe, L. F. (1975) Science 187 , 70–72. pmid:1167318 LaunchUrlAbstract/FREE Full Text ↵ NucciDisclosei, R. (1983) Mod. Cell Biol. 2 , 451–481. LaunchUrl ↵ Gibbon, B. C. & Kropf, D. L. (1994) Science 263 , 1419–1421. LaunchUrlAbstract/FREE Full Text ↵ Robinson, K. R. & Cone, R. (1980) Science 207 , 77–78. LaunchUrlAbstract/FREE Full Text ↵ Bowman, C. & Newell, A. C. (1998) Rev. Mod. Phys. 70 , 289–301. LaunchUrlCrossRef ↵ Cross, M. C. & Hohenberg, P. C. (1993) Rev. Mod. Phys. 65 , 851–1112. LaunchUrlCrossRef ↵ Hable W. E. & Kropf, D. L. (2000) Development (Cambridge, U.K.) 127 , 493–501. LaunchUrlAbstract ↵ Lengyel, I. & Epstein, I. (1991) Science 251 , 650–652. LaunchUrlAbstract/FREE Full Text Turing, A. (1952) Philos. Trans. R. Soc. LonExecuten Ser. B 237 , 37–72. LaunchUrlAbstract/FREE Full Text ↵ Ouyang, Q. & Swinney, H. L. (1991) Nature 352 , 610–612. LaunchUrlCrossRef ↵ Castets, V., Dulos, E., Boissonade, J. & De Kepper, P. (1990) Phys. Rev. Lett. 64 , 2953–2956. pmid:10041855 LaunchUrlCrossRefPubMed ↵ Jaffe, L. F. (1977) Nature 265 , 600–602. pmid:859558 LaunchUrlCrossRefPubMed Fromherz, P. (1998) Proc. Natl. Acad. Sci. USA 85 , 6353–6357. LaunchUrl ↵ Kramer, S. C. & Kree, R. (2002) Phys. Rev. E 65 , 051920. LaunchUrlCrossRef ↵ Leonetti, M. & Dubois-Violette, E. (1998) Phys. Rev. Lett. 81 , 1977–1980. LaunchUrl ↵ Taylor, A. & Brownlee, C. (1993) Planta 189 , 109–119. LaunchUrl ↵ Klingauf, J. & Neher, E. (1997) Biophys. J. 72 , 674–690. pmid:9017195 LaunchUrlPubMed ↵ Silverman-Gravila, L. B. & Lew, R. R. (2003) Microbiology 149 , 2475–2485. pmid:12949173 LaunchUrlAbstract/FREE Full Text ↵ Qian, N. & Sejnowski, T. J. (1990) Proc. Natl. Acad. Sci. USA 87 , 8145–8149. pmid:2236028 LaunchUrlAbstract/FREE Full Text ↵ Meinhardt, H. (1995) Nature 376 , 722–723. LaunchUrlCrossRef ↵ Speksnijder, J. E., Miller, A. L., Weisenseel, M. H., Chen T.-H. & Jaffe, L. F. (1989) Proc. Natl. Acad. Sci. USA 86 , 6607–6611. pmid:2505254 LaunchUrlAbstract/FREE Full Text ↵ Jaffe, L. F. (1958) Exp. Cell Res. 15 , 282–299. pmid:13597888 LaunchUrlCrossRefPubMed ↵ Meinhardt, H. & Gierer, A. (2000) BioEssays 22 , 753–760. pmid:10918306 LaunchUrlCrossRefPubMed ↵ Ferrier, J. M. (1980) J. Theor. Biol. 85 , 739–743. pmid:7442288 LaunchUrlPubMed ↵ NucciDisclosei, R. (1978) Dev. Biol. 62 , 13–33. pmid:563818 LaunchUrlCrossRefPubMed ↵ Jaffe, L. F. (1968) Adv. Morphol. 7 , 295–328. LaunchUrl ↵ Gierer, A. & Meinhardt, H. (1972) Kybernetik 12 , 30–39. pmid:4663624 LaunchUrlCrossRefPubMed