Spike Transmission through the Optic Tract-LGN Synapse
(A) Slice preparation showing the stimulating electrode in the optic tract and the recording electrode in the dorsal LGN.
(B) Circuitry of the LGN with functioning pathways in black (axons from the retina and to the visual cortex) and inactivated pathways in gray (cortex is removed to remove cortical input, and inhibition from interneurons is abolished with gabazine).
(C) Specimen trace showing that the stream of input action potentials (timing shown by stimulus trace) does not reliably generate postsynaptic action potentials.
(D) Spike frequency (mean ± SEM) in the input stimulus train (input) and evoked in 18 LGN cells (output). Points show data from individual cells.
(E) For cases where a postsynaptic action potential did occur (black distribution) or did not occur (red distribution), graphs show the probability of the preceding two presynaptic APs being separated by the interval shown on the abscissa (the area under each distribution is unity).
(F) Logical table stating possible input and output combinations, with specimen examples of each.
(G) Observed outcomes given an input AP (EPSPs had to be larger than 1 mV to be counted).
(H) Observed input APs given an output AP.
See also .

Conversion of Synaptic Conductance to Action Potentials in the LGN
(A) A section of the stimulus train applied to the optic tract axon.
(B) The EPSC train evoked in the LGN cell by the input train when voltage clamped at −55 mV.
(C) The AP sequence evoked in current clamp mode by the train in (A). Large vertical deflections in (B) (gray in inset) and small downward deflections in (C) are stimulus artifacts.
(D) The EPSC conductance time course derived from (B) for injection by dynamic clamp, with the same amplitude as evoked by synaptic input (g × 1) and scaled up and down in size (g × 3; g × ½).
(E) AP stream evoked by dynamic clamp injection at the soma of the g × 1 conductance trace in (D).
(F) AP stream evoked by dynamic clamp injection of the g × 3 and g × ½ conductance traces in (D).
All data are from the same cell.

Information Conveyed to the LGN by Natural Scenes
(A) Five-second segments of AP streams (1–5) recorded from ganglion cell axons in response to natural scenes [].
(B) AP responses (each line is one AP) of a specimen cell to five separate applications of trains 1–5 to the optic tract (real stimulation).
(C) Output information (mean ± SEM) in ten cells (shown as points) when the cell received AP-evoked synaptic currents (real stimulation) or had the measured conductance evoked by real stimulation injected at the soma with the same magnitude (dynamic clamp × 1).
See also .

Energy Use on Postsynaptic Currents in the LGN
(A and B) Two stimuli from one of the stimulus trains (A), chosen to evoke (B) an action potential or just an EPSC.
(C) The EPSCs evoked by the stimuli recorded at −55 mV in voltage clamp. The action potential in response to the first stimulus (as shown in B) was evoked by a larger EPSC (note that, because current was recorded in voltage-clamp mode, it does not reflect the sodium influx associated with the action potential itself). Integrating the current trace (area shaded in gray) gives the total postsynaptic charge entry. The actual Na⁺ entry is 1.42 times larger than this (see ). Na⁺ entry is then converted to ATP cost at a rate of one ATP molecule per three Na⁺ ions.
(D) The synaptic current, calculated from the conductance derived from (C), that is injected in dynamic clamp (with a conductance scaling factor of 1). Because the membrane potential is not voltage clamped, the current shows an outward deflection as the action potential depolarizes the cell positive to the reversal potential for the synapse.
(E) The Na⁺ current calculated to occur during dynamic clamping.
(F) The ATP used on extruding Na⁺ entering through the postsynaptic conductance, calculated under voltage-clamp conditions during stimulation of the optic tract, and when injecting the same conductance at the soma using dynamic clamp (ten cells; mean ± SEM shown as bar; individual cells shown as points).

Postsynaptic Conductance Magnitude Maximizes Information Transferred per Energy Used at the RGC-LGN Synapse
(A) Dependence of output information on synaptic conductance (g_syn) magnitude, when cells were stimulated with dynamic clamp (black points) with g_syn × 1 (applied to all ten cells) and other values (six cells for g_syn × 0.1, seven cells for ×0.3, nine cells for ×0.5, three cells for ×0.75, three cells for ×1.5, eight cells for ×3, seven cells for ×6, seven cells for ×9, and four cells for ×12) or with optic tract stimulation (white diamond; ten cells). Information is normalized to the value with g_syn × 1, for which the mean information rate was 20.6 ± 4.6 bits/s. Symbols and number of cells per condition are the same in (A) and (D)–(F).
(B) Relationship between firing frequency and g_syn for ten cells (fitted equation has the form F_max.g_synⁿ/(g_synⁿ + g_syn0.5ⁿ), where F_max = 9.9 Hz, n = 2, and g_syn0.5 = 1.4).
(C) Dependence of information rate on mean output firing frequency evoked by stimulus trains with different g_syn values in ten cells (fitted equation has the form I_max (1 − exp(−afⁿ)), where I_max = 83 bits/s, a = 0.1, n = 1.2, and f is frequency).
(D) Energy use on pumping out of postsynaptic ion influx as a function of g_syn multiplier used in dynamic clamp.
(E) Information divided by energy used on reversing the ion influx generating postsynaptic currents as a function of g_syn in ten cells (with, for each cell, the efficiency being normalized to the value at g_syn × 1; individual data for each cell are shown in ). The averaged data, shown in black, reveal a maximum very close to the physiological value of g_syn. 100% corresponds to 15.6 ± 2.7 bits per 10⁸ ATP molecules used. Equation fitted to the mean data has the form 100.g_syn/g_opt for g_syn < g_opt and 100.exp(−(g_syn − g_opt)/K) for g_syn > g_opt, where g_opt = 0.91 and K = 9.36.
(F) Information divided by energy use on reversing the ion influx generating postsynaptic currents and postsynaptic action potentials also shows a maximum near the physiological value of g_syn. Data are averaged over ten cells. 100% corresponds to 9.0 ± 1.2 bits per 10⁸ ATP molecules used. Fitted equation is as in (E) but with g_opt = 0.78 and K = 9.95.
Data are represented as mean ± SEM. See also .