Controlling bursting in cortical cultures with closed-loop multi-electrode stimulation

Abstract

One of the major modes of activity of high-density cultures of dissociated neurons is globally synchronized bursting. Unlike in vivo, neuronal ensembles in culture maintain activity patterns dominated by global bursts for the lifetime of the culture (up to 2 years). We hypothesize that persistence of bursting is caused by a lack of input from other brain areas. To study this hypothesis, we grew small but dense monolayer cultures of cortical neurons and glia from rat embryos on multi-electrode arrays and used electrical stimulation to substitute for afferents. We quantified the burstiness of the firing of the cultures in spontaneous activity and during several stimulation protocols. Although slow stimulation through individual electrodes increased burstiness as a result of burst entrainment, rapid stimulation reduced burstiness. Distributing stimuli across several electrodes, as well as continuously fine-tuning stimulus strength with closed-loop feedback, greatly enhanced burst control. We conclude that externally applied electrical stimulation can substitute for natural inputs to cortical neuronal ensembles in transforming burst-dominated activity to dispersed spiking, more reminiscent of the awake cortex in vivo. This nonpharmacological method of controlling bursts will be a critical tool for exploring the information processing capacities of neuronal ensembles in vitro and has potential applications for the treatment of epilepsy.

Figure 1.

Two-photon images of immunocytochemically stained neurons. A, MAP2. B, GABA. The two images show the same field of view. Negative controls showed no visible signals. Images were taken with a Carl Zeiss LSM510 multiphoton microscope. Arrows indicate GABA-positive neurons. Scale bar, 20 μm.

Figure 2.

Performance of feedback protocol. A, Median firing rate (dish-wide) achieved versus target. Any firing rate (FR) between the spontaneous rate and 500 SPSA could be stably maintained. Higher rates were not achievable in this culture without exceeding the safe voltage limit of 900 mV. The dotted line marks equality of achieved and target rates. Data are from five series on different sets of electrodes; 45 DIV. B, Stimulus voltage used to control firing rate at different levels.

Figure 3.

Examples of different spontaneous bursting patterns, with array-wide firing rates (line graphs) as well as per-electrode firing rates (grayscale plots). A, Chaotic bursting. Insets below show spike raster plots for a large global burst, a single-channel burst, and a small local burst, at 20× magnification (recorded at 25 DIV). B, Spontaneously regular bursting (recorded at 39 DIV). C, Superbursts (recorded at 34 DIV). The inset shows spikes at 10× magnification.

Figure 4.

Array-wide responses to stimulation. Each graph shows the responses on one electrode, represented according to the geometry of the array. The stimuli were delivered to the marked electrode. Vertical line indicates time of stimulation. Spikes were detected after artifact suppression (Wagenaar and Potter, 2002); TTX control confirmed the biological origin of all detected spikes.

Figure 5.

A, Burstiness during single-electrode stimulation (protocolS) and spontaneous activity (no stimulation). Each row shows the array-wide firing rate (coded by the grayscale at right) as a function of time during one 5 min experimental run. In the 10 examples of spontaneous activity shown (bottom), bursts occurred irregularly approximately once per minute. In the 10 examples of stimulation at 0.05 stim/sec, bursts were perfectly aligned with stimuli, except in a few cases during which a spontaneous burst just preceded the stimulus. (The stimulating electrode was different in each of the 10 rows.) At 0.1-0.2 stim/sec, bursts underwent period doubling. Bursts during stimulation at 1-5 stim/sec were less frequent, but still mostly stimulus locked. In the 10-50 stim/sec runs, burst control was perfect for the first 45 sec, after which a spontaneous-like pattern returned. Data are from a culture at 39 DIV. Note that experimental runs were executed in random order. B, Bursting during multi-electrode stimulation (protocol M), same culture. Perfect and sustained burst control is attained at the higher stimulation frequencies. Note the increase in tonic firing rate (background shading) as the stimulation frequency is increased. C, Burstiness index as a function of stimulation frequency, for single-electrode stimulation (□) and multi-electrode stimulation (▪). Slow single-electrode stimulation elevates the burstiness over spontaneous (unstimulated) levels (○), whereas rapid stimulation reduces it. Values are mean ± SEM from n = 100 runs on 10 cultures. The most effective protocol tested, 50 stim/sec distributed across 25 electrodes, suppressed bursts completely (n = 60 runs, 6 cultures).

Figure 6.

Stimuli presented to a single electrode (primary stimulation electrode; •) yielded much-reduced responses in the first 20 msec after stimulus when the stimulation rate was increased. (We focused on short-latency responses, because the majority of response spikes occurred at short latencies and because responses cannot be defined unambiguously beyond one interpulse interval, i.e., 25 msec for the highest stimulation frequency.) In fact, at a stimulation frequency of 40 stim/sec, the response was not much higher than the spontaneous firing rate (arrow at left). Each stimulation series lasted 5 min, and we discarded the responses recorded during the first 30 sec to measure the sustained response rate. Results are mean ± SEM from 53 electrode pairs in four cultures. During these experiments, we presented stimulus pulses to a second electrode every 5 sec. The responses in the first 20 msec after these latter stimuli (○) were not affected by the rate at which the first electrode was stimulated. Response strength in all cases was normalized to the results obtained from single-electrode stimulation at 0.2 stim/sec. The response strengths are plotted as a function of the frequency at which the primary electrode was stimulated. Inset and associated arrows indicate the stimulation protocol. Regardless of the frequency of the primary stimulation electrode, the secondary electrode was stimulated once every 5 sec.

Figure 7.

When switching between stimulation protocols, the activity pattern of a culture changed rapidly to match the new stimulation context. Here we show switches from rapid single-electrode stimulation to slow single-electrode stimulation, to rapid multi-electrode stimulation, and to no stimulation.

Figure 8.

Burstiness during closed-loop control of tonic firing rate. A, After an initial period of ∼15 sec during which the feedback algorithm settles, burst control was perfect at the higher target firing rates (f.r.). A culture at 43 DIV was used. (This culture was not tested at 800 SPSA.) B, Burstiness index decreased monotonically with the target rate and was always below the spontaneous level (○). Values are mean ± SEM from n = 85 runs using different sets of electrodes on 10 cultures.

Figure 9.

Assessment of the success rate of different burst-suppression protocols. Bars show the percentage of cultures in which each protocol successfully suppressed bursting, by random electrode selection (gray), by at least 1 of 10 selections tested (white), or by all of 10 selections tested (black). None of the cultures used in these experiments were burst-free in spontaneous activity. Protocols compared are as follows: protocol S at its optimal stimulation rate (10 stim/sec; S10); protocol M at the same rate (M10); protocol FB at its optimum (target 800 SPSA; FB); and protocol M at its optimum (50 stim/sec distributed across 25 electrodes; M50).