System-Size Biresonance For Intracellular Calcium Signaling

Jiqian Zhang,Zhonghuai Hou,Houwen Xin

DOI: https://doi.org/10.1002/cphc.200400089

IF: 3.52

2004-01-01

ChemPhysChem

Abstract:Using the chemical Langevin equation, the authors have addressed the effects of internal fluctuations on intracellular calcium signaling. It is found that the effect of the internal fluctuations depends on the volume of the cell. The signal-to-noise ratio (SNR; see plot) of the stochastic calcium oscillation shows two maxima, demonstrating the occurrence of system-size biresonance. Most importantly, the first resonance matches the size of real living cells. The study of noise-induced constructive effects in nonlinear dynamic systems has attracted considerable attention in the last two decades. It was demonstrated that there is a “resonant” noise intensity, at which the response of the system to a periodic force is maximally ordered, which is well-known as stochastic resonance (SR),1 and that the order of the noise-driven system itself can have a maximum in the absence of periodic forcing, which is called coherent resonance (CR).2 Stochastic resonance and coherent resonance have been observed in numerous experiments, and more and more attention has been paid to SR-like phenomena in biological systems, from ion-channel gating and neuron spiking, to life-supporting systems and human balance-control systems.3 As stated by Hänggi,3 it would indeed seem strange if nature would not have taken advantage of the benefits of ambient noises for nonlinear transmission and/or amplification of feeble information rather than ignoring it. Very recently, a new and quite interesting SR-like phenomenon, system-size resonance, has gained much attention.4–13 So far, mainly two types of “size resonance” behavior have been reported. On one hand, it was demonstrated that the collective behavior of an array of coupled noisy dynamical elements may be the most ordered when the system size (here the number of elements) has an optimal value.4–6 In such a case, the noise is external, and the system size plays a role in changing the noise strength that is subjected to the mean field. For example, system-size stochastic resonance was found in an ensemble of coupled noisy bistable elements subjected to a small periodic force,4 and system-size coherent resonance was demonstrated in a one-dimensional lattice of diffusively coupled excitable neurons in the absence of an external signal.5 On the other hand, for chemical oscillating reactions taking place in small systems, stochastic oscillations can be observed and there is an optimal system size at which such stochastic oscillations show the best performance.7–13 In such small systems, the molecule numbers of the reactants are often low and the internal noise resulting from the stochastic reaction events can not be ignored; it is generally accepted that the strength of the internal noise scales as 1/, where V is proportional to the system size. There have been a few quite interesting findings of this type. It was reported that ion-channel clusters of optimal sizes can enhance the encoding of a subthreshold stimulus.7–8 Shuai and Jung demonstrated that optimal intracellular calcium signaling appears at a certain size or distribution of the ion-channel clusters.9–11 In recent studies, using the Brusselator model, we have shown that the internal noise can induce stochastic oscillations in the region close to the deterministic oscillatory dynamics, and an optimal system size exists for such stochastic oscillations; this is characterized by a clear maximum in the signal-to-noise ratio (SNR) as a function of system size V.12 Similar results were also obtained in a circadian-clock system,13 which may have interesting implications for biological rhythms and signaling processes. Herein, we have investigated how the cell size would influence the intracellular calcium signaling process. We have used the model for intracellular calcium oscillation in hepatocytes, which was proposed by Höfer.14 To account for the internal noise, we have used the chemical Langevin equations proposed recently by Gillespie.15 Unlike previous studies,9–11 we do not consider the channel noise involving the calcium release from intracellular calcium stores, but those from the stochastic reaction events in the whole cell. In addition, we find that the SNR of the stochastic calcium oscillation, which is absent in the deterministic dynamics, shows double maxima as a function of the cell size, which may be called system-size biresonance. Interestingly, we find that one of the optimal sizes matches rather well with the real cell size, and such a match is rather robust to external stimulus. We show that such a biresonance phenomenon is quite relevant to the deterministic bifurcation features of the system. Calcium often acts as a second messenger in living cells, to regulate multiple cellular functions. There is a vast amount of literature devoted to the mathematical modeling of the intracellular calcium oscillations and waves observed in the experiments. Most of these models are deterministic and the stochastic effects due to the random channel dynamics or small cell size are neglected. However, the observation of localized stochastic Ca2+ puffs or sparks, and variations in the amplitudes and widths of the calcium oscillations suggest that internal noise must be taken into account.16 Actually, as mentioned above, the stochastic nature of the Ca2+ channel dynamics, involved in the release of Ca2+ from calcium stores to the cytosol, has already drawn much attention. But to our knowledge, the stochastic effects resulting from the small cell size has not been studied yet. According to the Höfer model, the calcium signaling dynamics in a single cell involve the interplay of calcium fluxes from and into the endoplasmic reticulum (ER) and across the plasma membrane (not considering the fluxes from and into other possible compartments such as mitochondria). By denoting the population numbers of free calcium ions in the cytosol as X and that in the whole cell by Z, the reactions in the cell can be grouped into four “elementary” processes involving the change of X or Z by 1.17 These processes and their reaction rates are defined in Table 1, where V is the volume of the cytosolic compartment of the cell, and x=X/V, z=Z/V denote the concentrations of the reactants. P is the concentration of inositol trisphosphate (IP3) in the cell, which denotes the level of the agonist simulation and is chosen to be the control parameter. Stochastic Processes Reaction Rates X→X+1 a1=Vρν0+νc+z, where kr(x,P)=k13+k2 X→X−1 a2=Vρν4+(1+β)x+αν3 Z→Z+1 a3=Vρν+νc Z→Z−1 a4=Vρνc With the variation of the control parameter P, this deterministic equation undergoes a Hopf bifurcation at P≈1.45 μM, above which calcium oscillations appear and below which only stable steady state can be observed. Figure 1 displays the bifurcation diagram for Equation (1) in the vicinity of P=1.45 μM, where the solid lines give the maxima and minima of the oscillations (left axis), and the dash line shows the corresponding frequency of the oscillation (right axis). One notes that the diagram can be divided into three regions. In region 1, no deterministic oscillation exists. In region 2, oscillation appears but the amplitude is quite small and increases gradually with the increment of P. In region 3, the oscillation is spike-like with large and nearly constant amplitude. Correspondingly, the oscillation frequency first decreases in region 2 and then keeps nearly constant in region 3. Notice there is a sharp inflexion at P≈1.47μM. One will see that such a bifurcation character maybe the very reason of the system-size biresonance phenomenon, as shown below. Bifurcation diagram for the deterministic model.The solid squares denote the maximum and minimum of the deterministic oscillation range (left axis), and the stars correspond to frequency (right axis). It is shown that the parameter space of P=IP3 concentration is divided into three distinct regions. We numerically integrate Equation (2) using the standard procedure for stochastic differential equations with a time step of 0.01 s.18 In regions 2 and 3, where deterministic oscillations exist, the effect of internal noise is destructive, that is, it makes the deterministic oscillation noisy, leads to phase diffusion, and reduces the correlation time of the oscillation.19 Therefore, we tune the control parameter P in region 1 but close to the Hopf bifurcation (HB) point, where the deterministic system does not sustain oscillation. Often, in such subthreshold cases, noise can play interesting and constructive roles. For a given subthreshold parameter P