Final Report Summary - SENSMOD (Microcircuits for behavioral modulation of sensory cortex)
The cerebral cortex, the most superficial area of the mammalian brain, is involved in the most complex cognitive tasks. The total 4 million neurons in the mouse cortex (Roth and Dicke, 2005) can be classified in only tenths cell types based on their morphology, genetic expression, and physiological properties (Jiang et al., 2015). The first major cortical neuron subdivision separates excitatory (or pyramidal neurons) from inhibitory cells (or interneurons). Inhibitory neurons can be further divided into three main classes: Somatostatin- (Sst), Vasoactive intestinal peptide- (Vip), and Parvalbumin- (Pvalb) expressing interneurons; which in total account for approximately 83% of all inhibitory neurons (Rudy et al., 2011). The aim of the project is to understand the distinctive functional properties of the Sst, Vip and Pvalb expressing interneurons in the mouse primary visual cortex (V1).
V1 is the earliest visual information processing centre in the cortex, right after the subcortical brain areas (i.e. the retina and the thalamus). Visual information is processed differently in V1, depending on the properties of the stimuli. For example, the size of the visual stimuli affects how strong the neurons will respond. The majority of V1 neurons respond maximally to stimuli of a finite spatial size, a phenomenon called size tuning. This property implies that responses are suppressed when the stimulus is too large, and this suppression is thought to be mediated by Sst neurons (Adesnik et al., 2012). Visual stimuli responses are also strongly modulated by the context (for example locomotion in the mouse). It has been proposed that this effect is caused by a disinhibitory circuit, in which during locomotion, Vip neurons inhibit Sst neurons, thus disinhibiting pyramidal (Pyr) neurons (Fu et al., 2014). We asked how in different cell classes (Sst, Vip and Pvalb) modulation of responses by running interacts with size tuning during.
To answer this question, we recorded the neuron activities, which expressed by an activity-dependent green fluorescent protein (GCaMP6), using a two-photon microscope (Figure 1A,B). This technique allows simultaneous recording of hundreds of neurons. We used transgenic mice which expressed a red fluorescent protein (tdTomato) in only one of the major inhibitory neuron classes (either Pvalb, Vip or Sst). Therefore we can identify the specific class of an active neuron (Figure 1C). Furthermore we developed a new classification method of Pyr cells, among the unlabeled cells, based on the sparseness of their activity. The mouse was free to run on an air-suspended ball while we presented stimuli of different sizes (drifting gratings). Previously only one condition at a time (stimulus size or locomotion) was considered and few neurons were recorded simultaneously. The novelty of the approach is that we measured hundreds of classified neurons’ responses to a combination of stimulus size and locomotion in the same experiment.
When exploring combinations of stimulus size and locomotion we found inconsistencies with previous models. Sst cells, as opposed to the predictions of the disinhibitory circuit, increased their activity during running in all visual conditions except in darkenss. However, we found that running increased more prominently visual responses to large stimuli in Sst neurons and to small stimuli in Vip neurons. Meanwhile Pvalb and Pyr neurons did not interact with the size of the stimulus. We then measured correlations of neural activity across time between cell types, and found that the three interneuron types showed different relationships to Pyr activity. The Pvalb population had strong positive correlations with the Pyr population, in running or stationarity. The Vip correlations with Pyr cells were less prominent in stationarity, when their firing was overall weaker. The Sst population had low or negative correlations with Pyr cells during running, but these were also substantially weakened in stationarity. These results indicate that interneuron-Pyr correlations strongly depend on sensory and behavioral context. The predictions of the disynaptic inhibitory circuit involving Vip and Sst cells appear to be correct only for some of these sensory and behavioral contexts.
In conclusion, our data resolve apparent contradictions in the literature and suggest an alternative to the disinhibitory model, in which Vip and Sst cells inhibit each other in a stimulus-dependent winner-take-all circuit while modulating a sub-network of Pyr-Pvalb neurons that is tightly correlated (Figure 1D). These results can have a substantial impact on our understanding of the brain since cortical microcircuits and computations have strong similarities across cortical areas and mammals, including humans (Carandini and Heeger, 2012; Douglas and Martin, 2004). Furthermore the results could shed light on mental disorders such as schizophrenia. Schizophrenic patients suffer from a massive and unfiltered influx of sensory data, while data suggest that size tuning is reduced in schizophrenic patients (Dakin et al., 2005; Uhlhaas et al., 2004). A prediction of the model would be that abnormal behavior of a specific cell class, as it has been found in patients with mental disorders, could lead to a distortion of size tuning. Finally, this project has helped the advancement of the European Research Area in the application of two-photon microscope technique combined with calcium-fluorescent indicators through the understanding of key preprocessing of experimental data stages such as correction of out-of-focus fluorescence and cell classification.
Legend of Figure 1. Summary of the experiments and results. (A) Recordings of cortical neurons in awake mouse V1. Activity was recorded with two-photon microscopy while visual stimuli are presented to the mouse. The mouse was free to run on an air-suspended ball.
(B) Calcium-dependent green fluorescence from neurons expressing GCaMP6. (C) Red fluorescence from the recordings in A, indicating tdTomato expression in Sst neurons. (D) Schematic illustration of the interaction between running and size tuning in L2/3 of mouse V1. Vip and Sst cells are arranged in a mutual inhibition circuit and they mediate top-down modulation by running, however they need an additional drive from visual stimuli to be active enough to affect the Pyr-Pvalb circuit. When the stimulus is small, the Vip cells win over the Sst cells, that are inhibited and in turn disinhibit the Pyr cells. When the stimulus is large, the Sst cells win over the Vip cells and inhibit the Pyr cells.
Adesnik, H., Bruns, W., Taniguchi, H., Huang, Z.J. and Scanziani, M. (2012). Nature 490, 226-231.
Carandini, M., and Heeger, D.J. (2012). Nature reviews Neuroscience 13, 51-62.
Dakin, S., Carlin, P., and Hemsley, D. (2005). Current biology : CB 15, R822-824.
Douglas, R.J. and Martin, K.A. (2004). Annual review of neuroscience 27, 419-451.
Fu, Y., Tucciarone, J.M. Espinosa, J.S. Sheng, N., Darcy, D.P. Nicoll, R.A. Huang, Z.J. and Stryker, M.P. (2014). Cell 156, 1139-1152.
Jiang, X., Shen, S., Cadwell, C.R. Berens, P., Sinz, F., Ecker, A.S. Patel, S., and Tolias, A.S. (2015). Science 350, aac9462.
Roth, G., and Dicke, U. (2005). Trends in cognitive sciences 9, 250-257.
Rudy, B., Fishell, G., Lee, S., and Hjerling-Leffler, J. (2011). Developmental neurobiology 71, 45-61.
Uhlhaas, P.J. Silverstein, S.M. Phillips, W.A. and Lovell, P.G. (2004). Schizophrenia research 68, 249-260.
Cortical Processing Laboratory
The laboratory has two sites:
The biological research site is at the UCL Institute of Ophthalmology, University College London, 11-43 Bath Street, London EC1V 9EL.
The data processing site is at the Rockefeller Building, University College London, 21 University Street, London WC1E 6DE.
Dr. Mario Dipoppa
Prof. Kenneth Harris