Experiment design and basic response properties.

A. (Top) Network schematic of hippocampal CA3-IN-CA1 network. (Bottom) A transverse hippocampal section showing channelrhodopsin expression in orange (tdTomato), with the stimulation grid drawn to scale overlaid on the CA3 network. An extracellular (field) electrode was used to record the total optogenetic excitation of the CA3 layer (black arrowhead), and CA1 recordings were made from individual pyramidal cells with a whole-cell patch clamp electrode (white arrowhead). (scale bar = 500µm) B. (Top) A DIC image of the CA3 cell layer with a few spots of a stimulation grid overlaid, drawn to scale. (Black dotted lines mark the outlines of a few CA3 cells). (Bottom) A schematic of the CA3-IN-CA1 layer showing the recruitment of CA3 pyramidal cells with three different patterns. The downstream CA1 pyramidal cell receives direct monosynaptic excitation from the activated population and disynaptic inhibition via a heterogeneous, shared population of CA1 interneurons. C. The main stimulation protocol consists of a pattern containing 1, 5, or 15 squares illuminated in the form of a pulse train delivered at frequencies 20, 30, 40, and 50 Hz, creating a 3-dimensional probing of the FFEI circuit. D. An example recording in the current clamp mode showing the postsynaptic potential of a patched CA1 cell (black) in response to a triggered 15-square pattern (red) used for pulsed optical stimulation in CA3 (cyan), estimated using field potential (orange). E. Extracellular recording of CA3 layer activation for 1,5,15 square patterns. F. Distribution of CA3 field response to first pulse for number of squares per pattern (Kruskal-Wallis test). G. Correlation of single neuron optical current recorded from patched CA3 cells with the corresponding field response for different pattern size values. H. Post synaptic potentials recorded in a CA1 pyramidal cell in current clamp for different pattern sizes. I. Distribution of PSPs across all recorded CA1 cells (n=16, p < 0.001, Kruskal-Wallis test). J. A slight decrease in the strength of field response was seen along the pulse train. (linear regression fit with slope 2%, r2=0.05, bars indicate 95% CI). K. Postsynaptic currents (PSC) recorded from CA1 cells in voltage clamp show the proportional relationship between excitatory and inhibitory PSCs for each pattern size. L. Similar to F and I, the distribution of PSC amplitude across the number of squares. M. A sample recording of excitatory (pink) and inhibitory (teal) currents recorded from a CA1 cell in the protocol described in panel D for a 15 square pattern. Blue traces in E,H,K represent the frame onset. Horizontal lines inside the violins represent quartiles. All violin plots show responses to probe pulse, thus without STP.

(A) Trial-normalised excitatory postsynaptic potentials across stimulation frequency and pulse index in the train for 5 square (Ai) and 15 square (Aii) patterns across all recorded cells (n=16). B. Spike probability as a function of the stimulus frequency, pulse index and number of squares in the current clamp cells showing higher likelihood in the first half of the train for five squares (top) and 15 square patterns (bottom). C. Post-synaptic responses were derived from kernel fits to valley to peak height for each response for both postsynaptic currents (voltage clamp, EPSC and IPSCs) and postsynaptic voltage (current clamp, not shown). D. Normalised EPSC and IPSC responses obtained for an example cell for a sample pattern showing consistent depression in inhibition and biphasic response in excitation. E. Both excitatory (Ei-Eii) and inhibitory (Fi-Fii) PSCs show short-term depression along the pulse train. However, this depression is much faster for inhibition than for excitation. Statistically, the short-term depression is significant across the number of squares, pulse index, and excitation vs inhibition (p<0.001, ANOVA)., and the input frequency (p<0.05, ANOVA).

A. Trend of normalised postsynaptic potentials in a 20 Hz pulse train received at CA1 pyramidal cells against patterns of size 5 and 15. B. Same as A for voltage clamp postsynaptic excitatory (red) and inhibitory (teal) currents. C. E-I balance, measured as the ratio of E to I, showing a rise along the pulse train that is significantly high in the later half of the train (p<0.001, ANOVA). D. The difference between the onset delays of excitation and inhibition. E. Scaling of observed responses in CA1 for patterns of different sizes (5, 7, 15 spots) as compared to the expected responses obtained by the sum of constituent single-spot responses, shown for a sample cell to illustrate the response normalisation that is fitted with a saturating (coefficient gamma, purple) and a linear function (coefficient slope, green). F. Gamma shows a leftward shift along the pulse train, here shown by comparing cumulative distribution of gamma for probe pulse vs pulse 8 of the train. (p<1e-3, Mann-Whitney test) G. Gamma depends both on pulse index and frequency. Gamma averaged over the first four pulses is larger than in the later four (p<0.001, ANOVA).

Multiscale model.

A: Model of reaction system in each bouton. Synaptic input triggers entry of extracellular Ca2+ into the bouton, and a pump removes the Ca. CaM buffers the Ca2+. Ca binds/unbinds from successive stages of the readily releasable (RR) pool of vesicles till they are docked, at which point the final Ca-binding step causes synaptic release. The released neurotransmitter (Glu for excitatory synapses, GABA for inhibitory) opens a ligand-gated receptor channel on the postsynaptic CA1 neuron, which is held voltage-clamped.(green arrow). The reaction scheme was identical for Glutamatergic and GABAergic synapses, but the rates were different to fit the respective voltage-clamp recordings. B: Close-up of dendrite of CA1 pyramidal neuron model. The cell has a compartment for soma, a 200-micron compartment for the dendrite, and 100 spines. Each spine is modelled as a head compartment, a neck compartment, and a presynaptic glutamatergic bouton. There are 200 GABAergic presynaptic boutons positioned directly on the dendrite. C: Network model. The CA3 has 16x16 neurons, projecting randomly to 16x16 interneurons. CA3 also projects directly to the Glu synapses on the spines. The interneurons project to GABA synapses.

Temporal summation and EI balance in burst stimuli.

A-D: EPSP measured at soma for representative cell 3531 (maroon) and simulation (yellow) for each frequency. E: Distribution and mean of EPSP over all recorded cells for each frequency for the experiment (maroon) and for a simulated neuron (yellow). There is no significant dependence of EPSP on frequency for the recordings (Spearman’s rank correlation p > 0.1). F-I: Simulated excitatory and inhibitory currents for each frequency. Note that the currents are measured close to resting potential, so the driving force (EGABA - Vm) is small compared to the voltage-clamp experiments in Figure 1K, where Vm was held at 0 mV. J: Simulated excitatory and inhibitory current peaks as a function of frequency and their ratio. K-N: Simulated excitatory and inhibitory conductances for each frequency. O: Simulated excitatory and inhibitory conductance peaks. The ratio is almost independent of frequency. Note that simulated synaptic conductances are reduced by ∼50x from nominal values to match the much smaller cell geometry and higher input impedance.

Experimental and simulated responses to Poisson spike train input, all patterns.

A: Experimental EPSP aligned with light trigger pulses. Blue trace is recorded data; red trace is the fit used to compute peaks and subsequent statistics. B: Same for simulated data. C-L: Comparisons for readouts of experiment (left column) and simulated (right column) data. Blue traces: 5-square stimulation, orange traces: 15-square stimulation. C, D: probability of trigger to generate a peak in the EPSP trace. E, F: Scatter plot of EPSP peaks as a function of time over the Poisson train. Note the initial decline due to STP and ChR2 desensitisation. Slope for 5 square: Experiment: −0.338±0.0727 mV/s, R^2=0.089; simulation: −1.09±0.14 mV/s, R^2=0.11. Slope for 15 square: Experiment 0.679±0.12 mV/s, R^2=0.022; simulation −1.11 ± 0.243 mV/s, R^2=0.072. Note that the R^2 values are small, due to large scatter in the data. G, H: Scatter plot of EPSP peaks vs. Inter-spike-interval. Note that the experimental data show an elevation of EPSP at short ISI. We fit this using a simple exponential decay function y=y0.exp(-t/tau)+y1. Tau for 5 square: Experiment: 13.5±0.54ms, R^2=0.152; Sim: 30.3±4.19ms, R^2=0.102. Tau for 15 square: Experiment: 18.3±1.93ms, R^2=0.038; Simulation:47.1±7.25ms, R^2=0.169. Note that the R^2 values are small, due to large scatter in the data. I, J: Distribution of EPSP peak amplitudes. The Kolmogorov-Smirnoff statistic comparing experiment and simulation was 0.176 for 5-square distributions, and 0.183 for 15-square, indicating that the histograms were similar, though the large number of observations meant that they could each be distinguished with p < 1e-50. K, L: Power spectral density of EPSP response over the entire dataset. Other than the 50Hz line noise spike in the experimental dataset in panel K, the distributions are very similar (Spearman’s coeff. = 0.92 and 0.94 for 5 and 15 square, p<1e-80 for both).

Mismatch responses to pattern changes.

A. Schematic of connectivity from pattern 1 (red) and pattern 2 (blue) to four unique and one overlapping synapse on the CA1 pyramidal cell. B. Schematic of changes in synaptic strength over a series of repeats of pattern 1, followed by pattern 2. Response amplitude is indicated by the height of EPSP in blue, near the soma, and synaptic strength by size of the synapse. The first cell is unstimulated, the second undergoes STP on the red synapses, and the third cell experiences depression of the synapses and hence lower amplitude response. The last cell receives pattern 2 (blue) which activates and strengthens two previously quiet synapses, and one overlapping one. This results in a larger response. C, D, E: Deterministic normalized EPSP responses to pattern sequence of prepulse followed by four patterns, each with eight repeats, in the order AAAAAAAABBBBBBBBCCCCCCCCDDDDDDDD. C. Deterministic model response where synapses lack STP. Due to small differences in the number of connecting synapses, the four patterns have slightly different EPSP amplitudes, but no transients. D. Deterministic model response when synapses have STP. There is a transient elevation in response every time the pattern changes, even if the new pattern elicited a weaker response in the no-STP case. E. Deterministic model response loses mismatch detection when pCA3_CA1 is raised to 0.1, that is, there is 10% connection probability between every presynaptic CA3 cell and the postsynaptic CA1 neuron. F, G, H: example traces and scatterplots for responses to the same pattern sequence. F: Experimental neuron 2821. G: reference model. H: Model with dense connections. Fi, Gi, Hi: sample EPSP traces. F,G,H remaining panels: Scatterplots for each EPSP peak over multiple repeats for the series of patterns. Cyan, yellow, pink and green indicate repeats for patterns A, B, C and D respectively. Blue line is mean. The strongest response occurs for the 50Hz repeats. There is a significant mismatch response going from pattern C to D (pink to green) for experimental data in Fiv (Wilcoxon test, p < 0.0001). Giv: In reference model (75% of spots set to zero), transitions one and two but not three show significance (p<0.0001, p<0.0001, p=0.674)p < 0.001). In Hiv (only 12.5% of spots set to zero) the first and third transitions are significant (Wilcoxon test, p=0.0003, p=0.0143). I: Methodology for estimation of mismatch response significance. Pulses 1 and 2 are before the transition, and 3 and 4 are after it. For clarity, data are offset to the mean of the four illustrated pulses. Dashed vertical line indicates transition. Peak EPSP is found for pulses 1 and 2, and compared with pulses 3 and 4 using the Wilcoxon test. In this example there is an elevation for only one of the transitions out of six. J: Distribution of mismatch responses in normalized EPSP. Note that the 8Hz case (blue) is narrow, so has fewer large transients. K: Heatmap of percentage of mismatch responses which were significant (Wilcoxon test, p < 0.05). Gamma frequency (20 and 50 Hz) and 15 square cases have more significant transitions. L-Q: Parameter dependence of mismatch response, computed as the ratio of the mean of the two samples before the pattern change, to the mean of the two samples after. Red triangles indicate the value of parameters in the reference model. L: Sparseness is defined as the number of stimulus points out of the total 16x16 stimulus which are set to zero. There is a steady rise in mismatch. M: mismatch vs. synaptic weight of glutamatergic synapses has a sharp peak at around 1 (corresponding to ∼0.158 nS per synapse). N. Mismatch against GABAergic synaptic weight peaks at around 20 (∼26 nS per synapse), and then almost plateaus. L, M, N: Mismatch vs. connectivity probability. O: Excitatory synapses from CA3 to the target CA1 pyramidal neuron. Note that this has a complementary slope to panel I. P: Excitatory synapses from CA3 to interneurons. Q: Inhibitory synapses from interneurons to postsynaptic CA1 pyramidal neuron. Both P and Q have a steady positive slope, which we interpret as due to increased saturation of connectivity of the inhibitory input to the CA1 cell, hence ensuring that depression in the GABAergic input is not pattern selective.

Network parameters and their definitions.

Spiking responses are selective for mismatch in patterned sequences.

All runs were performed using a spiking neuron model with stochastic synaptic chemical kinetics and each run consisted of 50 trials. Red triangles indicate transition times between patterns. For panels A to E and H to K, the stimulus consists of eight repeats of each four patterns as in AAAAAAAABBBBBBBBCCCCCCCCDDDDDDDD. Traces from C to M and P to S are spike rates with a moving window of 10 ms and 5 ms respectively. Mismatch statistics from C to M use the Mann-Whitney U test to compare the three pulses immediately before vs. the three immediately after a transition. P to us use the Mann-Whitney U test to compare 40 ms of response in the first theta cycle with the corresponding 40ms in the next three cycles. A: Spiking responses on three illustrative trials. B. Raster plot of spiking for reference stimulus. C: Reference model exhibits mismatch responses for the first two pattern transitions (p=0.002, 0.032, 0.434). D: Dense stimuli (25% spots nulled) gives a mismatch signal (p=0.014) only for the first transition. E: Sparse patterns (87.5% spots nulled) are mismatch selective but with less spiking (p=0.037, 0.407, 0.008). F: Oddball (deviant) stimulus presented every 8th pulse, against a uniform background: AAAAAAAABAAAAAAACAAAAAAADAAAAAAA. Though there is less spiking, all deviant stimuli trigger a response (p=0.044, 0.006, 0.035). G: Gap stimulus presented every 8th pulse, where instead of a deviant stimulus, no stimulus pattern was delivered. There is no tuning and activity rapidly drops to zero. H. Lack of STP in Glu removes mismatch detection, and spike rates are high. I. Lack of STP in GABA results in sustained strong inhibition and sparse firing, but some mismatch selectivity remains for the first two transitions (p=0.016, 0.003, 0.924). J. Complete removal of STP results in sustained brisk firing and no mismatch detection. K: Control with GABA synapses inactivated. There is elevated firing but the first pattern transition is still resolved (p = 0.006, 0.129, 0.851). L: Control with uniform pattern (A repeated 32 times). The spiking response rapidly drops to close to zero. M. Control with randomized pattern (one of pattern A to pattern E) on each pulse. Spiking response continues without any tuning, but with slow decay over the 32 pulses. N: Mismatch response (methods) is tuned to frequency of pattern repeats. The strongest responses are in the gamma range between 20 and 100 Hz. O: Schematic of gamma burst stimulus (green) repeated at theta frequency of 7.69 Hz (blue). Each burst consisted of a pattern repeated 5 times at 100 Hz. P: Theta modulated gamma burst responses showed strong spiking responses when each burst had a different pattern. Q: Spiking in second, third and fourth theta cycles was lower than the first when each burst had the same pattern (p=0.057, 0.029, 0.029). R, S: Continued strong firing when different patterns are presented in each of the five pulses of theta-modulated gamma. There is a small decline over successive theta cycles. R: No precession. The last burst is smaller than the first (p=0.042). S: Precession present. The last burst is again smaller than the first (p=0.029).