High-density recording of DRN neurons in-vivo

The midbrain dorsal raphe nucleus (DRN) contains the majority of the forebrain-projecting 5-
hydroxy-tryptamine (5-HT) neurons in the brain. These neurons are highly heterogenous in terms of
their molecular characteristics, and they interact with multiple types of neighbouring non-5-HT
neurons in ways that are not yet fully documented. One way to understand the functional
implications of this heterogeneity is to collect large-scale data-sets of DRN neural activity and to use
computational methods to help analyse the complex neural circuitry. Here, we commence to collect
such a data-set through high-density multi-site silicon electrode recordings in the DRN.
Recordings (Open Ephys) were made in urethane-anaesthetised mice using a silicon probe
(Cambridge NeuroTech, 32 channels) stereotaxically implanted into the DRN. EEG electrodes (3
channels) were placed bilaterally over the frontal cortex and right occipital cortex to record brain
state. After 1 h of baseline recording, neurons were screened for evidence of 5-HT1A receptormediated
autoinhibition by administration of the selective serotonin reuptake inhibitor citalopram
(10mg/kg i.p.). Recordings were continued for a further 1 h. Raw data from 32 channels were filtered
and single units were identified automatically using Kilosort and verified by manual clustering using
Phy. Spike trains were further analysed using a suite of custom-written Python scripts to reveal spike
waveform characteristics, firing rate and firing regularity. Spike-sorted neurons then underwent
clustering analysis to reveal groups of neurons with similar firing properties.
An initial analysis (~170 neurons) revealed multiple simultaneously recorded neurons (~35
neurons/mouse). Although much diversity in baseline firing properties was evident, clustering
analysis revealed 3 prominent groups of neurons; regular slow firing neurons previously identified as
putative 5-HT containing, irregular slow firing neurons, and fast firing neurons previously identified
as putative GABA containing. Citalopram inhibited all regular slow firing neurons and some irregular
slow firing neurons, while some of the latter were also excited.
Overall, the current high-density in vivo recordings show evidence of heterogeneity in the baseline
properties of DRN neurons as well as heterogeneity in their response to citalopram administration.
Ongoing experiments are expanding the size of the data-set to increase the power of the clustering
analysis and to commence computational analysis of DRN neuron interactions. Future experiments
will incorporate optotagging to aid chemical identification of the principal neuron clusters.