1.Department of Basic Education, Fuyang Institute of Technology, Fuyang 236031, China 2.School of Aerospace Engineering and Applied Mechanics, Tongji University, Shanghai 200092, China 3.College of Mathematics and Computer Science, Chifeng University, Chifeng 024000, China
Fund Project:Project supported by the Fund for Less Developed Regions of the National Natural Science Foundation of China (Grant No. 11762001), the National Natural Science Foundation of China (Grant Nos. 12072236, 11872276), the Program for Excellent Young Talents in Colleges and Universities of Anhui Province of China (Grant No. gxyqZD2020077), and the Young Talents of Science and Technology in Universities of Inner Mongolia Autonomous Region, China (Grant No. NJYT-20-A09).
Received Date:14 May 2021
Accepted Date:18 June 2021
Available Online:15 August 2021
Published Online:05 November 2021
Abstract:Anticipated synchronization, the response of the driven subsystem which appears earlier than the stimulation from the driving subsystem, is a universally counterintuitive nonlinear behavior. This behavior is also observed in the experiment on the nervous system in different brain regions. In the present paper, the anticipated synchronization phenomenon evoked by the inhibitory autapse is simulated in the coupled system composed of Morris-Lecar model, and the condition of excitability of single neurons and parameter ranges for the anticipated synchronization is presented. For a coupled system composed of two neurons, whether both neurons are either type-I excitability/Hopf bifurcation or type-II excitability/saddle-node bifurcation on an invariant cycle, in a driven neuron unidirectionally receiving the excitatory synaptic current from a driving neuron the delayed synchronization (the response of the driven neuron appears after the drive of the driving neuron) instead of the anticipated synchronization is simulated. After the inhibitory autapse is introduced into the driven neuron, the anticipated synchronization can be simulated in the coupled neurons when both neurons are of type-II excitability. With the increase of the conductance of the inhibitory autapse, the transition from delayed synchronization to anticipated synchronization is simulated. The time interval between spike of the driving and driven neuron is acquired, and the parameter ranges of anticipated synchronization in the plane of conductance of the inhibitory autapse and excitatory synapse are obtained. However, if both neurons are of type-I excitability, only delayed synchronization is simulated for the driven neuron with inhibitory autapse. Furthermore, the appearance of anticipated synchronization for type-II neurons and no anticipated synchronization for type-I neurons are suggested to have a relationship between the different responses of firing to external inhibitory stimulation for neurons with type-II excitability and type-I excitability. For spiking of type-II neuron, when an inhibitory pulse stimulation is applied, the spike following the pulse appears earlier than the one in the absence of stimulation in a wide range of the stimulation phase. However, for spiking behavior of type-I excitability, the spike following an inhibitory pulse stimulation appears later than the spike in the absence of stimulation. The results present the condition of single neurons for the appearance of anticipated synchronization induced by the inhibitory self-feedback mediated by autapse, which is helpful for understanding the dynamics of the counter-intuitive behavior, anticipated synchronization, presenting possible measures to modulate the anticipated synchronization, and proving directions for further study of anticipated synchronization. Keywords:anticipated synchronization/ excitability/ bifurcation/ inhibitory autapse
其中, ${T_0}$为没受外激励时稳定的放电周期; ${t_{\text{d}}}$为外激励从峰值时刻到开始作用时刻的时间; ${T_1}$为受外激励后峰值出现的时刻(图1(a)). 例如, 在ML模型的II型兴奋性参数下, 无外激励刺激的放电(图1(a)黑线)周期${T_0} \approx $ 52.87 ms. 在${t_{\text{d}}}$ = 20 ms处, 施加负向方波脉冲刺激(幅值A = –7 μA/cm2, 宽度d = 4 ms), 扰动后的${T_1} \approx $ 51.69 ms (图1(a)红线). 由(16)式计算得Δ(20) $ \approx $ 0.0223 > 0, 说明扰动后的放电提前. 按此方法计算不同${t_{\text{d}}}$的Δ, 则可得PRC (图1(b)红色点线). 从图1(b)可知, 在相同的抑制性方波脉冲宽度(d = 4 ms)下, 不同的幅值(A = –3, –5, –6, –7 μA/cm2)下对应的PRC具有相同的性质. 当27.36 ms < ${t_{\text{d}}}$ < 52.73 ms时, Δ < 0, 扰动后的放电延后; 当0 < ${t_{\text{d}}}$ < 27.36 ms时, Δ > 0, 扰动后的放电提前. 图1(c)是图1(b)的局部放大图, 其中Δ > 0的最大幅值达到10 –2量级. 图 1 II型兴奋性ML模型在抑制性刺激下的放电 (a)与无刺激的放电(黑线)相比, 负向方波脉冲(虚线, 幅值A = –7 μA/cm2, 宽度d = 4 ms)诱发的放电(红线)提前; (b)不同的负向方波脉冲(宽度d = 4 ms)诱发的PRC; (c)图(b)的局部放大图; (d)黑线和红线分别对应图(a)的黑线和红线(V, dV/dt)的相轨迹, 从脉冲刺激结束到各自的动作电位峰值; (e)图(d)的局部放大图(从脉冲刺激开始到V = –20 mV, 箭头代表刺激结束); (f)图(d)的局部放大图( 动作电位峰值前) Figure1. Firing of ML model with type II excitability under the action of inhibitory stimulation: (a) Compared with no stimulations (black solid line), firing (red line) induced by negative square pulse (dashed line, amplitude A = –7 μA/cm2, width d = 4 ms) is earlier; (b) PRC induced by negative square pulses with different strengths (width d = 4 ms); (c) locally enlargement of panel (b); (d) black and red curve correspond to trajectory in (V, dV/dt) plane of black and red curve of panel (a), respectively (from begging time of the pulse stimulation to peak of the action potential); (e) enlargement of panel (d) (from begging time of the pulse stimulation to –20 mV); (f) enlargement of panel (d) (phase before the peak of action potential).
首先用ML模型II型兴奋性的参数(表1和表2的第1组参数), 探索DS, AS和PD的现象.抑制性自突触电导固定为${g_{\text{I}}}$ = 0.3 μS/cm2, 在不同的兴奋性电导值${g_{\text{E}}}$下, 模型可以表现出如图2所示的不同同步行为. 为了描述每个现象, 定义$ t_i^{\text{S}} $为驱动神经元的膜电位在第$i$个周期(即第$i$个峰时)的时间, 而$ t_i^{\text{R}} $为对应的被驱动神经元的峰值时间. S和R神经元之间的尖峰计时差$ \tau $定义为 图 2 抑制性自突触诱发II型兴奋性ML神经元模型产生的3种动力学行为(${g_{\text{I}}}$ = 0.3 μS/cm2). DS (${g_{\text{E}}}$ = 1.8 μS/cm2): (a1)驱动(黑)和被驱动(红)神经元的膜电位; (a2)两神经元放电时间间隔的变化. AS (${g_{\text{E}}}$ = 0.1 μS/cm2): (b1)驱动(黑)和被驱动(红)神经元的膜电位; (b2)两神经元放电时间间隔的变化. PD (${g_{\text{E}}}$ = 0.03 μS/cm2): (c1)驱动(黑)和被驱动(红)神经元的膜电位; (c2)两神经元放电时间间隔的变化 Figure2. Three dynamic behaviors induced by inhibitory autapse (${g_{\text{I}}}$ = 0.3 μS/cm2) in the ML neuron model with type II excitability. DS (${g_{\text{E}}}$ = 1.8 μS/cm2): (a1) Membrane potential of driving (black) and driven (red) neurons; (a2) change of time interval between spikes of two neurons. AS (${g_{\text{E}}}$ = 0.1 μS/cm2): (b1) Membrane potential of driving (black) and driven (red) neurons; (b2) change of time interval between spikes of two neurons. PD (${g_{\text{E}}}$ = 0.03 μS/cm2): (c1) Membrane potential of driving (black) and driven (red) neurons; (c2) change of time interval between spikes of two neurons.
ML模型在II型兴奋性的参数(表1和表2的第1组参数)下, 若抑制性自突触电导设为${g_{\text{I}}}$ = 0 μS/cm2, 则随着兴奋性突触电导值${g_{\text{E}}}$的增加, $\tau $始终大于0, 如图3所示. 说明在没有抑制性刺激耦合的神经元将不会产生AS. 图 3 单向耦合II型兴奋性ML模型在没有抑制性自突触(${g_{\text{I}}}$= 0)时产生DS (两神经元放电时差在不同兴奋性电导下大于0, 即$\tau $> 0) Figure3. DS of type II ML model with unidirectional excitatory coupling and without inhibitory autapse (${g_{\text{I}}}$ = 0) (time interval between spikes of the two neurons is larger than 0 at different values of conductance of excitatory synapse, i.e. $\tau $> 0).
33.1.4.引入抑制性自突触后耦合神经元产生AS -->
3.1.4.引入抑制性自突触后耦合神经元产生AS
对被驱动神经元引入抑制性自突触, 会引起AS ($\tau $< 0), 如图4所示. 红线和蓝线分别表示固定${g_{\text{I}}}$ = 0.8 μS/cm2和${g_{\text{I}}}$ = 1.5 μS/cm2时产生AS, 小于三角符号标记的${g_{\text{E}}}$范围代表PD. ${g_{\text{I}}}$越大, $\tau $的绝对值越大. 黑线代表${g_{\text{I}}}$ = 0.2 μS/cm2时, 随着${g_{\text{E}}}$的增大, AS可以变为DS ($\tau $> 0); 而随着${g_{\text{E}}}$的降低, AS变为PD (位于三角符号标记以左). 图 4 不同抑制性电导${g_{\text{I}}}$下兴奋性耦合的II型兴奋性ML模型的AS (两神经元放电时间间隔$\tau $随${g_{\text{E}}}$的变化).$\tau $> 0表示DS状态, $\tau $< 0代表AS状态. 三角符号标记以左出现PD Figure4. Anticipated synchronization of type II ML model with excitatory coupling at different values of the conductance of inhibitory autapse (changes of time interval $\tau $ between spikes of two neurons with respect to${g_{\text{E}}}$). $\tau $> 0 and $\tau $< 0 represent DS and AS states, respectively. PD locates left to the triangle.
为了全面地展示兴奋性单向耦合的II型兴奋性ML模型产生AS, DS和PD参数的范围, 图5给出了两神经元放电时间间隔$\tau $在参数空间(${g_{\text{I}}},{g_{\text{E}}}$)的分布. 图5(a)的红色区域表示DS ($\tau $> 0), 绿色区域表示AS ($\tau $< 0), 黑色区域表示PD ($\tau $的值不稳定). 从DS到AS转变的分界线可用线性关系${{{g_{\text{E}}}} \mathord{\left/ {\vphantom {{{g_{\text{E}}}} {{g_{\text{I}}}}}} \right. } {{g_{\text{I}}}}} \approx $3.444来近似, 而从AS到PD转变的分界线可用线性关系${{{g_{\text{E}}}} \mathord{\left/ {\vphantom {{{g_{\text{E}}}} {{g_{\text{I}}}}}} \right. } {{g_{\text{I}}}}} \approx $0.137来近似. 图 5 单向兴奋性耦合的II型兴奋性ML模型的3类行为在参数平面(${g_{\text{I}}}$, ${g_{\text{E}}}$)的分布 (a)两神经元放电间隔$\tau $的分布, 红、绿和黑色分别表示DS ($\tau $> 0), AS ($\tau $< 0)和PD ($\tau $的值不稳定); (b)两神经元放电间隔$\tau $的量值 Figure5. Distribution of three behavior in parameter plane (${g_{\text{I}}}$, ${g_{\text{E}}}$) of the type II ML model with unidirectional excitatory coupling: (a) Distribution of time interval $\tau $ between spikes of two neurons, red, green, and black indicate DS ($\tau $> 0), AS ($\tau $< 0), and PD (the value of $\tau $ is unstable), respectively; (b) values of time interval $\tau $.
在表1第2组参数下, ML模型在${I_{{\text{app}}}} \approx $39.96 μA/cm2处产生不变圆上的鞍结分岔, 对应I型兴奋性. 本文选取在${I_{{\text{app}}}}$ = 46 μA/cm2处的稳定放电, 对应放电周期${T_0} \approx $92.27 ms. 与II型兴奋性PRC的定义类似, 也可得到I型兴奋性的PRC, 如图6(a)所示. 负向脉冲刺激诱发I型兴奋性的PRC几乎全负, 如图6(a)所示, 与大家熟知的正向脉冲诱发的PRC几乎全正且关于Δ = 0是对称的. 在相同的抑制性方波脉冲宽度(d = 4 ms)下, 不同的幅值(A = –1, –3, –5, –7 μA/cm2)下对应的PRC具有相同的性质. Δ < 0, 意味着刺激后的放电延后. 图6(a)中出现了小部分Δ > 0的情况, 其局部放大图, 如图6(b)所示. 图6(a)与图1(b)中Δ > 0的幅值差别还是比较大的(见放大图), 图6(b)中Δ > 0的幅值的最大值大约是0.0038, 并且其参数区间较窄, 在当前的突触参数下, 抑制性电流不容易作用在这个窄的区间, 因而不易产生AS. 图 6 (a)不同抑制性方波脉冲(宽度d = 4 ms)刺激诱发I型兴奋性ML模型的PRC; (b)图(a)的局部放大图 Figure6. (a) PRC induced by inhibitory square pulses (width d = 4 ms) stimulation in ML model with type I excitability; (b) locally enlarged of panel (a).
33.2.2.没有抑制性自突触的耦合神经元系统不易产生AS, 只产生DS -->
3.2.2.没有抑制性自突触的耦合神经元系统不易产生AS, 只产生DS
ML模型在I型兴奋性的参数(表1和表2的第2组参数)下, 若抑制性自突触电导为 ${g_{\text{I}}}$ = 0 μS/cm2, 则随着兴奋性电导值${g_{\text{E}}}$的增大, $\tau $始终大于0, 如图7所示. 说明在没有抑制性自突触时, 耦合神经元系统将不会产生AS, 只产生DS. 图 7 单向耦合I型兴奋性ML模型在没有抑制性自突触电导(${g_{\text{I}}}$ = 0 μS/cm2)时产生DS (两神经元放电时差在不同兴奋性电导下大于0, 即$\tau $> 0) Figure7. DS of type I ML model with unidirectional excitatory coupling and without inhibitory autapse (${g_{\text{I}}}$ = 0) (time interval between spikes of the two neurons is larger than 0 at different values of conductance of excitatory autapse, i.e.$\tau $> 0).
33.2.3.引入抑制性自突触后耦合神经元系统不易产生AS -->
3.2.3.引入抑制性自突触后耦合神经元系统不易产生AS
由图7可知, 在没有抑制性自突触时, 耦合神经元将不会产生AS. 现引入抑制性自突触, 看是否会产生AS. 例如, 取抑制性自突触电导 ${g_{\text{I}}}$ = 0.2 μS/cm2和兴奋性耦合电导${g_{\text{E}}}$ = 0.02 μS/cm2, 如图8(a1)和图8(a2)所示, 其放电时间序列(驱动神经元放电(黑)在前、被驱动神经元放电(红)在后)和两神经元的放电时间间隔$ \tau \approx $4.25 ms, 均说明只产生了DS. 若兴奋性电导为${g_{\text{E}}}= 0$ μS/cm2, 只有抑制性自突触电流, 此时系统表现PD, 如图8(b1)和图8(b2)所示. $ {\tau _i} $的变化随着时间表现出周期性, 1个周期内的$ {\tau _i} $从约7.73 ms 延长到约106.83 ms. 延长是因为该结果是在没有抑制性自突触流, 只有兴奋性突触流的情况下产生的. 因此, 时间延长是由兴奋性突触流引起放电延后引起的. 无论如何引入抑制性自突触电流, 耦合神经元的行为均不易产生AS. 图 8 抑制性自突触诱发I型兴奋性ML神经元模型产生的2种动力学行为(${g_{\text{I}}}$ = 0.2 μS/cm2). DS (${g_{\text{E}}}$ = 0.02 μS/cm2): (a1)驱动(黑)和被驱动(红)神经元的膜电位; (a2)两神经元放电时间间隔的变化. PD (${g_{\text{E}}}$ = 0 μS/cm2): (b1)驱动(黑)和被驱动(红)神经元的膜电位; (b2)两神经元放电时间间隔的变化 Figure8. Two dynamical behaviors induced by inhibitory autapse in the ML neuron model with type I excitability (${g_{\text{I}}}$ = 0.2 μS/cm2). DS (${g_{\text{E}}}$ = 0.02 μS/cm2): (a1) Membrane potential of driving (black) and driven (red) neurons; (a2) change of time interval between spikes of two neurons. PD (${g_{\text{E}}}$ = 0.0 μS/cm2): (b1) Membrane potential of driving (black) and driven (red) neurons; (b2) change of time interval between spikes of two neurons.
图9是固定抑制性电导${g_{\text{I}}}$, 两神经元放电的时间间隔$\tau $随兴奋性电导${g_{\text{E}}}$的变化. 黑、红和蓝线分别表示固定${g_{\text{I}}}$ = 0.2 μS/cm2, ${g_{\text{I}}}$ = 0.8 μS/cm2和${g_{\text{I}}}$ = 1.5 μS/cm2时的$\tau $值. $\tau $> 0表示DS状态, 三角形标记以左代表PD. 从图9可知, 在3个抑制性电导下, 除在${g_{\text{E}}}$ = 0处产生了PD, 其余均产生了DS, 没有AS产生. 图 9 不同抑制性电导${g_{\text{I}}}$下兴奋性耦合I型兴奋性ML模型的DS (两神经元放电时间间隔$\tau $随${g_{\text{E}}}$的变化). $\tau $> 0表示DS状态, 三角符号标记处出现PD Figure9. DS of the type I ML model with excitatory coupling at different values of conductance of inhibitory autapse (changes of time interval $\tau $ between spikes of the two neurons was with respect to ${g_{\text{E}}}$). $\tau $> 0 represents DS state, and the phase-drift locates to the triangle.
为了更好地展示产生DS参数的范围, 图10给出了两神经元放电时差$\tau $在参数空间(${g_{\text{I}}}$, ${g_{\text{E}}}$)的分布. 如图10(a)所示, 除在${g_{\text{E}}}$ = 0处出现了PD(黑色区域), 其余存在DS的分布区域(红色区域, $\tau $> 0). 在DS区域中, 固定${g_{\text{I}}}$, $\tau $的值随着${g_{\text{E}}}$的增大而减小, 如图10(b)所示. 图 10 单向兴奋性耦合的I型兴奋性ML模型的DS在参数空间(${g_{\text{I}}}$, ${g_{\text{E}}}$)的分布 (a)两神经元放电间隔$\tau $的分布; 红和黑分别表示DS ($\tau $>0)和PD ($\tau $的值不稳定)区域; (b)两神经元放电间隔$\tau $的量值 Figure10. Distribution of DS in parameter plane (${g_{\text{I}}}$, ${g_{\text{E}}}$) of type I ML model with unidirectional excitatory coupling: (a) Distribution of time interval $\tau $ between spikes of two neurons, red and black indicate DS ($\tau $> 0) and PD (the value of $\tau $ is unstable), respectively; (b) values of time interval $\tau $.