Functional separation of long-range inputs by intrinsic dynamics of dorsal raphe 5-HT neurons

Abstract

Monoaminergic nuclei such as the serotonergic dorsal raphe nucleus (DRN) receive synaptic inputs containing functionally distinct streams of information, yet the dimensionality of the resulting output population code and its cellular underpinning are currently unknown. By combining electrophysiological and computational approaches, here we uncover separable neural encoding of two excitatory inputs conveying disjunct information to DRN 5-HT neurons - the lateral habenula (LHb) and medial prefrontal cortex (mPFC). Dual-color opsin strategies revealed that a population of 5-HT neurons receive inputs from both mPFC and LHb. Subthreshold excitatory postsynaptic potentials triggered by both inputs were largely indistinguishable, yet suprathreshold spiking behavior exhibited input-specific latencies and dispersion statistics. A support vector machine classifier demonstrated that input identity can be accurately decoded from spike timing, but not subthreshold events, of under ten 5-HT neurons. Upon examining the intrinsic cellular mechanisms in 5-HT neurons that couple EPSPs to spiking dynamics, we uncovered two likely candidate mechanisms: a low-threshold calcium conductance that selectively boosts slow excitatory inputs, and a subthreshold, voltage-dependent membrane noise that generates variation of spike latency and jitter. Stochastic simulations suggest that these two intrinsic properties of 5-HT neurons are sufficient to transform LHb and mPFC inputs into distinct output spiking patterns. These results reveal that hub-like networks like the DRN can segregate distinct informational streams by a cell-intrinsic mechanism. The resulting emergent population spike synchrony code provides a means for the DRN to widely broadcast these streams as a multiplexed signal. Significance statementPhylogenetically old neuromodulatory systems in the brain, such as the serotonergic dorsal raphe nucleus, are compact yet richly innervated structures. Here, we use the raphe as a testbed to ask how distinct informational sources to hub-like networks are processed or integrated into a coherent neural code. Using electrophysiological and computational methods, including biophysically grounded stochastic simulations, we find that intrinsic noise mechanisms in serotonergic neurons are critical to transform approximately matched subthreshold excitation into distinct spike timing profiles. Thus, cell-intrinsic noise mechanisms can effectively synthesize a spike synchrony code that, we hypothesize, multiplexes input information to hub-like networks at the population level even in the absence of strong local circuit interactions.