Top-down perspectives on cell membrane potential and protein transcription Authors: Javier Cervera, Michael Levin, Salvador Mafe Published: 2025 ABSTRACT We have explored a simple model for the multicellular interplay between bioelectricity and protein transcription using a top-down perspective that offers new insights complementary to the commonly used bottom-up descriptions. We model the non-excitable cell bioelectrical representation of the external environment, including the neighboring cells, using voltage-gated ion channels and intercellular junctions. The simulations make three predictions: (i) shifts in membrane potential allow transitions between gene expression states, (ii) in the case of different cell potential-gated transcriptions, depolarized cells cannot control the distinct gene expressions as effectively as polarized cells, and (iii) community effects should permit to extend the single-cell control to the multicellular level. Because the spatio-temporal distributions of instructive signaling ions and molecules depend on the local electric potentials, different multicellular potentials correlate with distinct downstream gene expression patterns. A central cell is able to measure the number of neighboring cells and learn their bioelectrical state from the downward-induced membrane potential changes, which suggests that multiscale bioelectricity can operate as a top-down mechanism in development and regeneration.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-31696-6. Introduction In biological phenomena, causation is commonly analyzed from the parts to the whole. However, in a complementary but rather controversial metaphor, top-down or downward causation (DC) has been cited when higher-level effects appear to influence lower-level cellular states and functions; see Refs1,2 and references therein. Here, we have considered a simple model for the interplay between bioelectricity and transcription in non-excitable cells that explores a top-down perspective. To this end, we invoke an experimentally based model of bioelectrical adaptive transcription that involves feedbacks at multiple size scales, emphasizing the central biological role of the cell3. The understanding of multicellular patterns and bio-fields4,5 can benefit from descriptions at scales higher than the underlying molecular events because a top-down control offers new opportunities for external actions. In particular, multiscale bioelectricity constitutes a high-level spatial scale control node that expands from the subcellular membrane potentials of organelles such as lysosomes and mitochondria to the spatio-temporal patterns of multicellular potentials characteristic of development and regeneration4. We consider here the coupling between bioelectricity and the transcription of two model proteins. To this end, we operationally describe the internal cell representation of the external environment, including the neighboring cells, using voltage-gated ion channels and intercellular junctions. This bioelectrically-centered model5 is complementary to traditional descriptions focused on biochemical networks and biomechanical fields. The membrane potential V, defined as the electrical potential difference between the cell inside and the external microenvironment, is central to metabolic, osmotic, developmental, and regeneration phenomena6–16. In particular, instructive polarized (high absolute value of the negative cell potential V) and depolarized (low absolute value of V) physiological windows are experimentally observed4,7,17–19. While crucial information of the body plan resides within the DNA as a basic single-cell language for making proteins, multicellular potentials can transduce short-term bioelectrical information into long-term transcriptional processes because of the coupling of the signaling ions and molecules concentrations with the local electric potentials4,5,7. In this way, the spatio-temporal patterns of cell potentials can provide downstream biochemical information for the establishment of different morphological regions4,5,15,16,20. Experimentally, voltage-sensitive dyes15,21 make evident that different anterior–posterior morphologies can result from changes in the axial electric potential4,15,20 regulated by ion channels and intercellular gap junctions. Moreover, controlled alterations of channels and junctions can result in phenotype changes4,7,20,22. Transcriptional control of mammalian cells by membrane depolarization showing the coupling between bioelectrical and biochemical signals has also been demonstrated in electrogenetic circuits11 and the role of membrane potential has been described in the onset of embryonic differentiation12. In this case, while gene expression is mediated by intracellular signal transducers and biochemical pathways, bioelectricity can also influence the timing of pluripotency genes. The above experimental facts suggest that top-down effects reflect not only the role of the whole cell state on subcellular basic processes but also the multicellular regulation of the cell state itself on the basis of the feedback between bioelectrical and transcriptional processes4,5. In these processes, top-down phenomena arises when a central cell identifies an environmental change and uses it to tune its individual state. Recently, we have described how a multicellular aggregate could adapt to environmental toxicity by means of the feedback between the cell membrane potential and downstream biochemical processes23. We consider now the effects that the upper-scale multicellular cell potentials may provoke on the lower-scale cellular transcription following the virtual processes of Fig. 1a and b. In our model a simple component, the initially depolarized or polarized central cell, can iterate its state with time in response to a collectively computed macroscopic property, the polarized or depolarized potentials of the neighboring cells forming the multicellular aggregate. Thus, the central cell can detect the states of neighboring cells from the induced membrane potential changes that eventually impact on the protein transcription rates. Our bioelectrical model involves top-down and bottom-up mechanisms schematically shown in Fig. 1c.Fig. 1(a) Scheme of the multicellular aggregate composed of a central cell and n neighboring cells. At time t = 0 the central cell is in a depolarized state (blue) characterized by a low absolute value \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left| V \right|$$\end{document} of the negative membrane potential V while the neighboring cells are in a polarized state (red) characterized by high \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left| V \right|$$\end{document}. At t > 0, the intercellular connectivity is established (arrow), allowing the polarized neighboring cells to repolarize the central cell. The expressions of two proteins A and B are coupled through the cell potential V. Because of the different regulations assumed for the V-dependent transcriptional rates of proteins A and B, the initially similar expression levels and concentrations pA and pB can become distinct after the central cell repolarization. (b) In the opposite situation, the central cell would be polarized (high \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left| V \right|$$\end{document}) while the neighboring cells are depolarized (low \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left| V \right|$$\end{document}) at time t = 0. Now, the initially distinct concentrations of proteins A and B can become similar after the central cell depolarization. (c) We attempt to frame the interplay between the bioelectrical and biochemical networks using the top-down and bottom-up mechanisms shown in the scheme. These mechanisms are fully described in the model section. (a) Scheme of the multicellular aggregate composed of a central cell and n neighboring cells. At time t = 0 the central cell is in a depolarized state (blue) characterized by a low absolute value \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left| V \right|$$\end{document} of the negative membrane potential V while the neighboring cells are in a polarized state (red) characterized by high \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left| V \right|$$\end{document}. At t > 0, the intercellular connectivity is established (arrow), allowing the polarized neighboring cells to repolarize the central cell. The expressions of two proteins A and B are coupled through the cell potential V. Because of the different regulations assumed for the V-dependent transcriptional rates of proteins A and B, the initially similar expression levels and concentrations pA and pB can become distinct after the central cell repolarization. (b) In the opposite situation, the central cell would be polarized (high \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left| V \right|$$\end{document}) while the neighboring cells are depolarized (low \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left| V \right|$$\end{document}) at time t = 0. Now, the initially distinct concentrations of proteins A and B can become similar after the central cell depolarization. (c) We attempt to frame the interplay between the bioelectrical and biochemical networks using the top-down and bottom-up mechanisms shown in the scheme. These mechanisms are fully described in the model section. We emphasize that the simulated conditions of Figs. 1a and b can be physiologically meaningful. In particular, a central cell surrounded by a finite number of normally polarized (depolarized) neighboring cells can still assume a depolarized (polarized) membrane potential when the voltage-gated intercellular connections are sufficiently weak. In particular, it is an experimental fact that relatively small depolarized patches can be surrounded by hyperpolarized extensive regions in development, regeneration, and initial tumorigenesis4,5. Bioelectrical model The model considers the case of a cell that expresses two generic proteins with opposite functional dependencies of their transcription rates \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r_{m,k} (V)$$\end{document} on the absolute cell potential \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left| V \right|$$\end{document} (Figs. 1 and 2), where k = protein A and protein B. In this cell, the voltage-dependent rate of protein A decreases with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left| V \right|$$\end{document} (negative regulation) while that of protein B increases with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left| V \right|$$\end{document} (positive regulation). Because of the negative values of the cell potential, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left| V \right|$$\end{document} is equivalent to –V here. In this bioelectrical model, the transcription rates \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r_{m,k} (V)$$\end{document} can be functions of the cell potential V because the concentrations Sk of the respective signaling ions or molecules depend on V, thus providing the feedback between the bioelectric and genetic networks of Fig. 2. Experimentally, the voltage-gated ion channels of Fig. 2 could be polarizing potassium channels and depolarizing sodium and calcium channels8,9,11,12, which are characterized by their rapid conductance responses to changes in the cell membrane potential, thus being central to the cell depolarization and repolarization phenomena that influence protein transcription4,5,8,11,12. As to the case of signaling ions and molecules, proteins, and transcription factors whose inside levels could be regulated by the cell potential, specific cases are given in References4–6,11,12,38. Note here that, while gene expression is also mediated by intracellular signal transducers and biochemical pathways essential to the expression of differentiation factors, bioelectricity may influence also the timing of pluripotency genes12.Fig. 2The bioelectrical and transcriptional feedback for the case of two generic proteins A and B whose transcriptions are modulated by the cell potential V that is regulated in turn by two ion channels of conductances \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{dep}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{pol}$$\end{document}. These additional channel proteins modulate the potential V by attempting to establish the equilibrium potentials potentials \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$E_{dep}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$E_{pol}$$\end{document} characteristic of the depolarized (dep) and polarized (pol) cell states5. Note that that proteins A (−, negative regulation) and B (+, positive regulation) are assumed to have opposite dependencies of their transcription rates on the absolute value \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left| V \right|$$\end{document} of the cell potential; see Fig. 1. In the model, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S_{{k}}$$\end{document} are the signaling ion or molecule concentrations and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S_{{{0,k}}}$$\end{document} are their reference values, with k = A and B. Because the cell potential V < 0, the absolute value |V| can be replaced by − V. The use of |V| should make it easier to identify the different cell potential regulations, negative or positive, of proteins A and B here. The bioelectrical and transcriptional feedback for the case of two generic proteins A and B whose transcriptions are modulated by the cell potential V that is regulated in turn by two ion channels of conductances \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{dep}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{pol}$$\end{document}. These additional channel proteins modulate the potential V by attempting to establish the equilibrium potentials potentials \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$E_{dep}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$E_{pol}$$\end{document} characteristic of the depolarized (dep) and polarized (pol) cell states5. Note that that proteins A (−, negative regulation) and B (+, positive regulation) are assumed to have opposite dependencies of their transcription rates on the absolute value \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left| V \right|$$\end{document} of the cell potential; see Fig. 1. In the model, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S_{{k}}$$\end{document} are the signaling ion or molecule concentrations and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S_{{{0,k}}}$$\end{document} are their reference values, with k = A and B. Because the cell potential V < 0, the absolute value |V| can be replaced by − V. The use of |V| should make it easier to identify the different cell potential regulations, negative or positive, of proteins A and B here. In Fig. 2, the transcription and translation of proteins A and B are described by a set of kinetic equations for the mRNA (mA and mB) and protein (pA and pB) concentrations, respectively. These concentrations are relative numbers defined in terms of absolute reference values that would depend on the system considered. The mRNA transcription (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r_{{{m,k}}}^{{}}$$\end{document}) and protein translation (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r_{{{p,k}}}$$\end{document}) rate constants, together with their respective degradation rates (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_{{{m,k}}}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_{{{p,k}}}$$\end{document}), k = A and B, are effective values for the multiple kinetic steps that may be involved in real cases. In the model, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r_{{{m,k}}}^{{o}}$$\end{document}, k = A and B, are reference values for the proteins transcriptional rates. Note that molecular diffusion within the cell is ignored because it should be relatively fast compared with the underlying genetic processes5. The central and neighboring cells in the aggregate do not constitute an isolated system but they are immersed in a microenvironment that fixes the external ionic concentrations and equilibrium potentials potentials \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$E_{{{dep}}}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$E_{{{pol}}}$$\end{document} characteristic of the dep and pol cell states. Also, all cells in the aggregate are ruled by the same equations. For every cell, the polarization state is modulated by the electrical currents \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$I_{{{dep}}}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$I_{{{pol}}}$$\end{document} flowing through the dep and pol channels, respectively, that connect the cell cytoplasm with the external microenvironment. In the balance equation for the cell potential V of Fig. 2, C is the cell capacitance, z is the number of effective charges for channel gating, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_{{{th,dep}}} = - V_{{T}}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_{{{pol}}} = - V_{{T}}$$\end{document} are the threshold potentials of the channels8. The thermal potential is defined as VT = RT/F = 27 mV, where R is the gas constant, T = 310 K is the temperature, and F is the Faraday constant8. We assume the channel equilibrium potentials take the limiting values \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$E_{{{pol}}}$$\end{document} = −60 mV and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$E_{{{dep}}}$$\end{document} = 0 mV, which should be approximately constant provided that the cell is in contact with a comparatively large microenvironment so that the extracellular ionic concentrations do not change with time5. Note also that the number of ions to be transported across the membrane in order to establish a given value of V is usually small compared with the total number of ions in the cell. This fact, together with the continuous action of the membrane ion pumps8, should result in approximately constant intracellular concentrations and equilibrium potentials. The intercellular connectivity allows to integrate the individual cell state into the multicellular aggregate. This higher spatial scale permits the enhancement of morphologically instructive transitions or the collective compensation of local disruptive perturbations in the cell polarization, as needed in particular biological stages24. Experimentally, specific bioelectric changes can contribute to tail regeneration, brain repair, and the induction of ectopic eyes4,7,25–28. Also, modifications of the intercellular connectivity may trigger transitions between instructive multicellular patterns5,7,20,27,29. Fig. 3a displays a model that qualitatively reproduces the typical experimental dependence of the intercellular conductance on the trans-junctional potential difference5,30,31. Figure 3b shows the balance equation for the cell potential V, which changes because of both the channel currents Idep and Ipol of Fig. 2 and the intercellular current Iij between the central cell i and all its neighboring cells j. The same equation applies to the neighboring cells j. The negative differential resistance region of Fig. 3a, which is observed at high enough potential differences between two neighboring cells, contributes to the regionalization of the multicellular aggregate in separated depolarized and polarized modules5,31. The integration of particular single-cell states into modules results in spatio-temporal patterns that are instructive for development and regeneration because the different bioelectrical modules eventually lead to distinct downstream gene expression regions4,5.Fig. 3(a) The conductance Gij of the intercellular junction between two neighboring cells i and j as a function of the cell potentials difference Vi – Vj. In this model, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{GJ}}}^{{o}} /G_{{{ref}}}$$\end{document} is the maximum conductance relative to a reference conductance \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{ref}}}$$\end{document} and V0 = 18 mV is a constant potential that controls the width of the bell-shaped function5. (b) Note that the cells are not isolated but in contact with their microenvironment. Thus, every cell potential depends on both the charge transfer between the cell and its microenvironment and the charge transfer between the cell and the neighboring cells. Consequently, the potential Vi of a given cell i can change because of the channel currents Idep,i and Ipol,i of Fig. 2 and the intercellular currents \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$I_{ij} = G_{ij} (V_{j} - V_{i} )$$\end{document}, which gives the sum of currents over the number of nearest neighbor (nn) cells j. The neighboring cells are also connected between them. Thus, the bottom-up effect of the central cell on neighboring cells is fully accounted for in the model: all neighboring cells j are bioelectrically connected to the central cell i through intercellular currents. In this way, the membrane potential of all cells in the aggregate can change because of the feedback mechanism. (a) The conductance Gij of the intercellular junction between two neighboring cells i and j as a function of the cell potentials difference Vi – Vj. In this model, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{GJ}}}^{{o}} /G_{{{ref}}}$$\end{document} is the maximum conductance relative to a reference conductance \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{ref}}}$$\end{document} and V0 = 18 mV is a constant potential that controls the width of the bell-shaped function5. (b) Note that the cells are not isolated but in contact with their microenvironment. Thus, every cell potential depends on both the charge transfer between the cell and its microenvironment and the charge transfer between the cell and the neighboring cells. Consequently, the potential Vi of a given cell i can change because of the channel currents Idep,i and Ipol,i of Fig. 2 and the intercellular currents \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$I_{ij} = G_{ij} (V_{j} - V_{i} )$$\end{document}, which gives the sum of currents over the number of nearest neighbor (nn) cells j. The neighboring cells are also connected between them. Thus, the bottom-up effect of the central cell on neighboring cells is fully accounted for in the model: all neighboring cells j are bioelectrically connected to the central cell i through intercellular currents. In this way, the membrane potential of all cells in the aggregate can change because of the feedback mechanism. In this multiscale model, the influence of the cell membrane potential on the subcellular proteins transcription (Fig. 2) is extended to a higher scale where the multicellular potentials influence the cell potential itself (Fig. 3). Also, because of the dependence of the cell potential V on the ionic concentrations and channels conductances (Fig. 2), V can contribute to the internal cell representation of the external environment, including the neighboring cells (Fig. 3). At the upper scale, the dynamics of multicellular potentials can reflect biologically-instructive external events4,32, similarly as neuronal spatio-temporal patterns constitute mental representations of the outside world33. Results and discussion We analyze now how the neighboring cell potentials influence the central cell potential and protein transcription in the cases of Figs. 1a and b. In this way, a simple component (the central cell) can modulate its subcellular transcriptional state in response to a collectively computed higher-scale property (the bioelectrical state of the multicellular aggregate). Figures 4a–d show the time-dependent cell potential and concentrations of proteins A and B in the central cell for different intercellular connectivity \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{GJ}}}^{{o}} /G_{{{ref}}}$$\end{document} (Fig. 3) and n = 7 neighboring cells. The dynamics of the cell potential are characterized by relatively fast bioelectric relaxation times C/Gdep of the order of 1 s, to be compared with protein transcription times of the order of 1 h5. Two opposite situations are simulated: initially, the central cell is in a depolarized (low \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left| V \right|$$\end{document}) state and the neighboring cells are in a polarized (high \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left| V \right|$$\end{document}) state in the case of Figs. 4a and b; on the contrary, the central cell is in a polarized (high \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left| V \right|$$\end{document}) state and the neighboring cells are in a depolarized (low \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left| V \right|$$\end{document}) state in Figs. 4c and 4d.Fig. 4(a) The time-dependent central cell potential V and protein concentrations pk (k = A, B) in the initially depolarized central cell for a number5,35 n = 7 of neighboring polarized cells and the intercellular conductance ratio \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{GJ}}}^{{o}} /G_{{{ref}}} = 0.1$$\end{document} (weak connectivity)5,31. (b) The case of (a) but now with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{GJ}}}^{{o}} /G_{{{ref}}} = 1.0$$\end{document} (strong connectivity)5,31. (c) The time-dependent cell potential and concentrations of proteins A and B in the initially polarized central cell for a number n = 7 neighboring depolarized cells and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{GJ}}}^{{o}} /G_{{{ref}}} = 0.1$$\end{document} (weak connectivity). (d) The case of (c) but now with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{GJ}}}^{{o}} /G_{{{ref}}} = 1.0$$\end{document} (strong connectivity). The scales (bottom) correspond to the central and neighboring cells potentials and the protein concentrations in the central cell. In all cases, the insets show the cell states initially and at times sufficiently large for the steady state values of potential and protein concentrations to have been achieved. These concentrations are relative numbers that depend on the system considered. The time units are those of the cell potential relaxation time5,39 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\normalsize C/G_{{{ref}}} = 1$$\end{document} s, so that the protein concentration relaxation times would be in the delayed range 1 h – 1 day4,5,25. In the simulations, the channel conductances \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{dep}}} /G_{{{ref}}} = 1.5$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{pol}}} /G_{{{ref}}} = 0.5$$\end{document} for the initially depolarized central cell and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{dep}}} /G_{{{ref}}} = 0.5$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{pol}}} /G_{{{ref}}} = 1.5$$\end{document} for the initially polarized central cell, are assumed5. For the neighboring cells, Gdep/Gref = Gpol/Gref = 1. We have also introduced the rate constants \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r_{{{m,k}}}^{{o}}$$\end{document} = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r_{{{p,k}}}$$\end{document} = 0.2 min−1 and the degradation \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_{{{m,k}}}$$\end{document} = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_{{{p,k}}}$$\end{document} = 0.005 min−1 for k = A and B proteins5. (a) The time-dependent central cell potential V and protein concentrations pk (k = A, B) in the initially depolarized central cell for a number5,35 n = 7 of neighboring polarized cells and the intercellular conductance ratio \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{GJ}}}^{{o}} /G_{{{ref}}} = 0.1$$\end{document} (weak connectivity)5,31. (b) The case of (a) but now with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{GJ}}}^{{o}} /G_{{{ref}}} = 1.0$$\end{document} (strong connectivity)5,31. (c) The time-dependent cell potential and concentrations of proteins A and B in the initially polarized central cell for a number n = 7 neighboring depolarized cells and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{GJ}}}^{{o}} /G_{{{ref}}} = 0.1$$\end{document} (weak connectivity). (d) The case of (c) but now with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{GJ}}}^{{o}} /G_{{{ref}}} = 1.0$$\end{document} (strong connectivity). The scales (bottom) correspond to the central and neighboring cells potentials and the protein concentrations in the central cell. In all cases, the insets show the cell states initially and at times sufficiently large for the steady state values of potential and protein concentrations to have been achieved. These concentrations are relative numbers that depend on the system considered. The time units are those of the cell potential relaxation time5,39 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\normalsize C/G_{{{ref}}} = 1$$\end{document} s, so that the protein concentration relaxation times would be in the delayed range 1 h – 1 day4,5,25. In the simulations, the channel conductances \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{dep}}} /G_{{{ref}}} = 1.5$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{pol}}} /G_{{{ref}}} = 0.5$$\end{document} for the initially depolarized central cell and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{dep}}} /G_{{{ref}}} = 0.5$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{pol}}} /G_{{{ref}}} = 1.5$$\end{document} for the initially polarized central cell, are assumed5. For the neighboring cells, Gdep/Gref = Gpol/Gref = 1. We have also introduced the rate constants \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r_{{{m,k}}}^{{o}}$$\end{document} = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r_{{{p,k}}}$$\end{document} = 0.2 min−1 and the degradation \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_{{{m,k}}}$$\end{document} = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_{{{p,k}}}$$\end{document} = 0.005 min−1 for k = A and B proteins5. The initial bioelectrical state of a weakly connected central cell cannot be changed by the neighboring cells; thus, the changes observed in the proteins transcription are moderate in this case (Figs. 4a and c). On the contrary, significant top-down effects, demonstrated by the bioelectrical and transcriptional changes, are feasible for strong intercellular connectivity (Figs. 4b and d). In Figs. 4a and b, the attempts to repolarize the central cell caused by the polarized neighboring cells induce a decrease in the expression of protein A and an increase in the case of protein B. On the contrary, in Figs. 4c and d, the attempts to depolarize the central cell caused by the depolarized neighboring cells induces just the opposite changes in the protein expressions: protein A concentration increases while protein B concentration decreases. Remarkably, the model simulations suggest that similar proteins expressions can become different by cell repolarization (Fig. 4b) while different proteins expressions can become similar by cell depolarization (Fig. 4d), as qualitatively shown in Fig. 1. Experimentally, changes in the cell polarization state can be induced by acting on the ion channel and intercellular junction conductances, which could be modified by the injection of specific mRNAs or the independent blocking of the dep and pol channels and the intercellular conductance4,24,31. These operational procedures should eventually result in downstream protein transcription changes4,23,27. Fig. 4b emphasizes the close correspondence of the two-state bioelectrical memory, characterized by the equilibrium potentials \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$E_{{{pol}}}$$\end{document} = −60 mV (polarized cell state) and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$E_{{{dep}}}$$\end{document} = 0 mV (depolarized cell state), with the two-state biochemical memory characterized by the respective protein concentrations \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$p_{{A}} \ne p_{{B}}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$p_{{A}} \simeq p_{{B}}$$\end{document}. In this way, the cell potential-dependent transcription of Fig. 2 establishes a multicellular interplay between bioelectrical and biochemical magnitudes of experimental relevance4,6,11,12. Figures 5a–c show the central and neighboring cells potentials, together with the concentrations of proteins A and B in the central cell, as a function of the number of neighboring cells in the surrounding patch at different intercellular connectivity. The initial conditions for this multicellular aggregate are similar to those of Figs. 4a and b. At t = 0, the central cell is characterized by a low value of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left| V \right|$$\end{document} while the neighboring cells show high values \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left| V \right|$$\end{document}, thus attempting to repolarize the initially depolarized central cell. The steady-state protein concentrations achieved at large enough times show significant changes of the proteins transcription in the central cell when the number of neighboring cells is high enough, which emphasizes again the scale-dependent nature of the top-down effects.Fig. 5(a) The steady-state concentrations pk of proteins A and B in the initially depolarized central cell as a function of the number of neighboring cells in the surrounding patch at fixed intercellular connectivity5,31 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{GJ}}}^{{o}} /G_{{{ref}}} = 0.1$$\end{document}. (b) The case \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{GJ}}}^{{o}} /G_{{{ref}}} = 0.2$$\end{document}. (c) The case \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{GJ}}}^{{o}} /G_{{{ref}}} = 1.0$$\end{document}. The scales show the central and neighboring cells potentials and the protein concentrations in the central cell. (a) The steady-state concentrations pk of proteins A and B in the initially depolarized central cell as a function of the number of neighboring cells in the surrounding patch at fixed intercellular connectivity5,31 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{GJ}}}^{{o}} /G_{{{ref}}} = 0.1$$\end{document}. (b) The case \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{GJ}}}^{{o}} /G_{{{ref}}} = 0.2$$\end{document}. (c) The case \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{GJ}}}^{{o}} /G_{{{ref}}} = 1.0$$\end{document}. The scales show the central and neighboring cells potentials and the protein concentrations in the central cell. Interestingly, the central cell can measure the number of neighboring cells and learn their bioelectrical state (Fig. 3) from the membrane potential-induced changes (Fig. 5). Also, the effects of increasing the connectivity (Figs. 5a–c) loosely resembles the synaptic reinforcement in neural learning. In our case, the central cell should have a sufficiently high intercellular conductance \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G_{{{GJ}}}^{{o}} /G_{{{ref}}}$$\end{document} with the neighboring cells to learn and establish the new transcriptional state, characterized by the protein concentrations \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$p_{{A}} \ne p_{{B}}$$\end{document} (Fig. 5c), from the initially different state with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$p_{{A}} \simeq p_{{B}}$$\end{document}(Fig. 5a). Experimentally, higher-scale effects on the central cell potential have been observed in the electrical activity of non-excitable cell aggregates34 as well as in astrocytes coupled through networks of gap junctions35. In the second case, the aggregate connectivity can compensate for local changes in the extracellular potassium concentration that would depolarize the central cell otherwise35. Taking together, Figs. 4 and 5 emphasize the role of intercellular communication in adaptive transcription and suggest the bioelectrical reprograming of cells via targeting the ion channel and junction proteins that regulate single-cell and multicellular potentials4,5,11,12,31. Note also that abnormally depolarized cell potentials and low intercellular coupling abolish the potential-dependent transcription because the central cell cannot keep different concentrations of proteins A and B (Figs. 4a and 5a) in the model of Figs. 1 and 2. This fact emphasizes the need of establishing the specific potential-dependent transcriptions characteristic of each biosystem. Here, possible applications are electrogenetic circuits where the transcriptional control of mammalian cells is regulated by the coupling between bioelectrical and biochemical signals11, embryonic differentiation processes where the cell potential influences the timing of crucial pluripotency genes12, and the changes in cell depolarization and intercellular connectivity characteristic of the initial stages of tumorigenesis; see Refs4,31. and references therein. A Methods section that makes reference to the software and input data options needed to reproduce the simulations of Figs. 4 and 5 with the equations of Figs. 2 and 3 is also included. In addition, the simulations data are available from the corresponding authors on reasonable request. Spatio-temporal patterns with different cell potentials and protein expressions, which are characteristic of developmental processes4, have been described in recent simulations; see Ref36 and references therein. Figure 6 shows the close correspondence between the different depolarized and polarized regions of a multicellular aggregate and the distinct transcriptions of proteins A and B. The changes in the protein concentrations that occur in the left and right regions of the multicellular aggregate can be understood from the results of Figs. 4 and 5. The oscillatory correlations of the bioelectrical cell states with the protein expression levels suggests an instructive role for membrane potentials reminiscent of biological symmetry breaking and segmentation phenomena5,37.Fig. 6Time evolution of cell potentials (top) and protein concentrations (bottom) for cells # 1 (left region) and 2 (right region) of a multicellular aggregate with N = 1165 cells. In this simulation, the channel proteins are also regulated by the cell potential V: the pol channel is negatively regulated by V while the dep channel is positively regulated by V, which results in oscillatory cell potentials and protein A and B concentrations. The dynamic depolarized (blue) and polarized (red) bands arise from an initial symmetry-breaking (dashed line): the mRNA transcription rate of the pol channel protein5 is \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r_{{{m,pol}}}^{{o}}$$\end{document} = 1.5 min−1 in the left 20% region of the aggregate while it decreases to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r_{{{m,pol}}}^{{o}}$$\end{document} = 0.75 min−1 in the right 80% region. On the contrary, the mRNA of the dep channel protein is \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r_{{{m,dep}}}^{{o}}$$\end{document} = 1.0 min−1 throughout the whole aggregate. The other rates5,36 are \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r_{{{p,k}}}$$\end{document} = 1.0 min−1 and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_{{{m,k}}}$$\end{document} = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_{{{p,k}}}$$\end{document} = 0.2 min−1, with the reference concentration5 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$p_{{{0,k}}}$$\end{document} = 80, for the channel proteins k = pol and dep. As to the rates of proteins A and B, they are \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r_{{{m,k}}}^{{o}}$$\end{document} = 2.0 min−1 and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r_{{{p,k}}}$$\end{document} = 2.0 min−1, with the degradation rates \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_{{{p,k}}}$$\end{document} = 0.4 min−1 and the reference concentrations \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$p_{{{0,k}}}$$\end{document} = 80, k = A and B. As in Fig. 4, t is in C/Gdep units, which gives cell potential times of the order of 1 s and protein transcription times of the order of 1 h5,36. Time evolution of cell potentials (top) and protein concentrations (bottom) for cells # 1 (left region) and 2 (right region) of a multicellular aggregate with N = 1165 cells. In this simulation, the channel proteins are also regulated by the cell potential V: the pol channel is negatively regulated by V while the dep channel is positively regulated by V, which results in oscillatory cell potentials and protein A and B concentrations. The dynamic depolarized (blue) and polarized (red) bands arise from an initial symmetry-breaking (dashed line): the mRNA transcription rate of the pol channel protein5 is \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r_{{{m,pol}}}^{{o}}$$\end{document} = 1.5 min−1 in the left 20% region of the aggregate while it decreases to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r_{{{m,pol}}}^{{o}}$$\end{document} = 0.75 min−1 in the right 80% region. On the contrary, the mRNA of the dep channel protein is \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r_{{{m,dep}}}^{{o}}$$\end{document} = 1.0 min−1 throughout the whole aggregate. The other rates5,36 are \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r_{{{p,k}}}$$\end{document} = 1.0 min−1 and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_{{{m,k}}}$$\end{document} = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_{{{p,k}}}$$\end{document} = 0.2 min−1, with the reference concentration5 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$p_{{{0,k}}}$$\end{document} = 80, for the channel proteins k = pol and dep. As to the rates of proteins A and B, they are \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r_{{{m,k}}}^{{o}}$$\end{document} = 2.0 min−1 and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r_{{{p,k}}}$$\end{document} = 2.0 min−1, with the degradation rates \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_{{{p,k}}}$$\end{document} = 0.4 min−1 and the reference concentrations \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$p_{{{0,k}}}$$\end{document} = 80, k = A and B. As in Fig. 4, t is in C/Gdep units, which gives cell potential times of the order of 1 s and protein transcription times of the order of 1 h5,36. Experimentally, changes in membrane potentials synchronously control the somite’s formation periodicity and growth for regular body pattern formation during somitogenesis; see Ref38. and references therein. Using chick embryos, it has been demonstrated that both temporal and spatial features can be simultaneously controlled by the cell potential of somite-forming cells, as reflected by the biochemical and structural changes38. Conclusions In Biology, feedback loops that couple different mechanisms occurring at distinct scales are commonplace. Here, we have considered a simple model for the interplay between bioelectricity and transcription using a top-down perspective that emphasizes intercellular communication (Fig. 1). We characterize the bioelectrical networks using voltage-gated single-cell channels and intercellular junctions (Figs. 2 and 3). Cell potentials are particularly suited to show top-down phenomena because of their sensitivity to different environmental chemical, mechanical, and electrical signals4,5,14,39. The resulting bioelectrical adaptive transcription offers new insights complementary to the common bottom-up description. While our feedback does not address the general philosophical issues of causation, other problems that can be related to ours involve coarse-graining mechanisms2, emergence theories40,41, the nature of evolutionary dynamics42, the physical influence of initial and boundary conditions on the differential equations that represent lower-level processes43, and the role of collective intelligence across the biological scales of morphogenesis, with an emphasis on top-down strategies44,45. One attractive feature of top-down control is the possibility to induce multi-modal downstream effects (whole organogenesis, appendage regeneration, complex organ repair) with relatively simple triggers44. The ability to induce modular, context-sensitive, and self-limiting cascades without micromanaging the outcomes46 should be of potential importance to biomedical applications47. Recent work has also examined different measures of causation, thus contributing to the understanding of the concept across disparate fields with different measures of it48. Interestingly, it was noted that macroscale effects could limit the expected uncertainties in causal relationships, leading to more robust causes at higher levels48, as loosely suggested by the community effects of Figs. 4 and 5. Note that the assumption that the update rules of Figs. 2 and 3 change with the time-dependent cell potentials immediately leads to dynamical equations and cell states that mutually influence each other41. Also, when the timescales of bioelectricty (membrane potential) and transcription (gene expression) are very different, the use of a steady-state approximation for the membrane potential changes would be possible49. The model simulations, which could be extended to the case of biomechanically-regulated and oscillatory processes5,50–53, make three predictions: (i) shifts in membrane potential may allow transitions between gene expression states, (ii) in the case of different cell potential-gated transcriptions, depolarized cells cannot control the distinct gene expressions as effectively as polarized cells, and (iii) community effects should permit to extend the single-cell control to the multicellular level. Indeed, the multicellular system can modulate the transcription of two proteins A and B characterizing the single-cell state (Figs. 4, 5, 6). Thus, the transcriptional changes associated with cell depolarization and repolarization phenomena can produce adaptive protein expression changes modulated by the intercellular connectivity23,36,54,55, with immediate consequences on the transcription of proteins A and B. This result, which emphasizes the role of both the single-cell and multicellular scales in Biology3–5,56,57, should be significant for bioelectrically-influenced developmental and regeneration processes4,5,13. Methods The algorithm employed in Figs. 4 and 5 has been implemented using html/javascript. The calculations take advantage of the vastly different characteristic times of bioelectricity (cell potential) and transcription (gene expression). Initially, the time evolution of the electrical potential is numerically calculated and the protein concentrations are assumed constant. After a time much longer than the bioelectrical time but shorter than the transcription time, when the potential no longer changes, the evolution of the protein concentrations is analytically calculated (see Supplemental Information). Figure 6 requires a full numerical integration of the evolution equations and the algorithm used in this case is described in Reference36. Supplementary Information Supplementary Information. Supplementary Information.