Uncovering the effect of different brain regions on behavioral classification using recurrent neural networks

As our ability to record neural activity from a larger number of brain areas increases, we need to develop tools to understand how this activity is related to ongoing behavior. Recurrent neural networks (RNNs) have been shown to perform successful classification for sequence data. However, they are black box models: once trained, it is difficult to uncover the mechanisms that they are using to classify. In this study, we analyze the effect of RNNs on classifying behavior using a simulated dataset and a widefield neural activity dataset as mice perform a self-initiated behavior. We show that RNNs are comparable to, or outperform, traditional classification methods such as Support Vector Machine (SVM), and can also lead to accurate prediction of behavior. Using dimensionality reduction, we visualize the activity of the RNNs to better understand the classification mechanisms of the RNNs. Finally, we are able to accurately pinpoint the effect of different regions on behavioral classification. This study highlights the utility and interpretability of RNNs while classifying behavior using neural activity from different regions.

Automated task training and longitudinal monitoring of mouse mesoscale cortical circuits using home cages

We report improved automated open-source methodology for head-fixed mesoscale cortical imaging and/or behavioral training of home cage mice using Raspberry Pi-based hardware. Staged partial and probabilistic restraint allows mice to adjust to self-initiated headfixation over 3 weeks' time with ~50% participation rate. We support a cue-based behavioral licking task monitored by a capacitive touch-sensor water spout. While automatically head-fixed, we acquire spontaneous, movement-triggered, or licking task-evoked GCaMP6 cortical signals. An analysis pipeline marked both behavioral events, as well as analyzed brain fluorescence signals as they relate to spontaneous and/or task-evoked behavioral activity. Mice were trained to suppress licking and wait for cues that marked the delivery of water. Correct rewarded go-trials were associated with widespread activation of midline and lateral barrel cortex areas following a vibration cue and delayed frontal and lateral motor cortex activation. Cortical GCaMP signals predicted trial success and correlated strongly with trial-outcome dependent body movements.

Individualized tracking of self-directed motor learning in group-housed mice performing a skilled lever positioning task in the home cage

Skilled forelimb function in mice is traditionally studied through behavioral paradigms that require extensive training by investigators and are limited by the number of trials individual animals are able to perform within a supervised session. We developed a skilled lever positioning task that mice can perform within their home cage. The task requires mice to use their forelimb to precisely hold a lever mounted on a rotary encoder within a rewarded position to dispense a water reward. A Raspberry Pi microcomputer is used to record lever position during trials and to control task parameters, thus making this low-footprint apparatus ideal for use within animal housing facilities. Custom Python software automatically increments task difficulty by requiring a longer hold duration, or a more accurate hold position, to dispense a reward. The performance of individual animals within group-housed mice is tracked through radio-frequency identification implants, and data stored on the microcomputer may be accessed remotely through an active internet connection. Mice continuously engage in the task for over 2.5 mo and perform ~500 trials/24 h. Mice required ~15,000 trials to learn to hold the lever within a 10° range for 1.5 s and were able to further refine movement accuracy by limiting their error to a 5° range within each trial. These results demonstrate the feasibility of autonomously training group-housed mice on a forelimb motor task. This paradigm may be used in the future to assess functional recovery after injury or cortical reorganization induced by self-directed motor learning. NEW & NOTEWORTHY We developed a low-cost system for fully autonomous training of group-housed mice on a forelimb motor task. We demonstrate the feasibility of tracking both end-point, as well as kinematic performance of individual mice, with each performing thousands of trials over 2.5 mo. The task is run and controlled by a Raspberry Pi microcomputer, which allows for cages to be monitored remotely through an active internet connection.

Mesoscale brain explorer, a flexible python-based image analysis and visualization tool

Imaging of mesoscale brain activity is used to map interactions between brain regions. This work has benefited from the pioneering studies of Grinvald et al., who employed optical methods to image brain function by exploiting the properties of intrinsic optical signals and small molecule voltage-sensitive dyes. Mesoscale interareal brain imaging techniques have been advanced by cell targeted and selective recombinant indicators of neuronal activity. Spontaneous resting state activity is often collected during mesoscale imaging to provide the basis for mapping of connectivity relationships using correlation. However, the information content of mesoscale datasets is vast and is only superficially presented in manuscripts given the need to constrain measurements to a fixed set of frequencies, regions of interest, and other parameters. We describe a new open source tool written in python, termed mesoscale brain explorer (MBE), which provides an interface to process and explore these large datasets. The platform supports automated image processing pipelines with the ability to assess multiple trials and combine data from different animals. The tool provides functions for temporal filtering, averaging, and visualization of functional connectivity relations using time-dependent correlation. Here, we describe the tool and show applications, where previously published datasets were reanalyzed using MBE.

Determination of the molecular size of the 5-HT3 receptor binding site by radiation inactivation

The radiation inactivation technique has been used to estimate the molecular size of the 5-HT3 receptor binding site labelled by [3H]zacopride, in comparison with that of the 5-HT1A receptor binding site labelled by [3H]8-OH-DPAT, in rat cortical membranes. The calculated molecular weight of the 5-HT3 site: 35.4 +/- 2.2 kDa (mean +/- S.E.M., n = 4) was significantly less than that of the 5-HT1A site: 62.9 +/- 1.8 kDa (mean +/- S.E.M., n = 4) and of other 5-HT1 and 5-HT2 receptors of the G-protein coupled family. These data further support that the 5-HT3 receptor is not coupled to G-proteins in the rat brain.

Minaprine antagonises the serotonergic inhibitory effect of trifluoromethylphenylpiperazine (TFMPP) on acetylcholine release

The serotonin agonist, m-trifluoromethylphenylpiperazine (TFMPP), inhibited the K+-evoked release of [3H]acetylcholine ([3H]ACh) from rat hippocampal synaptosomes. The inhibitory effect of TFMPP was blocked by the non-selective 5-HT1 antagonist, methiothepin, but was not affected by ketanserin, mesulergine or spiperone. The 5-HT3 antagonist, MDL 72222, slightly reversed the inhibitory effect. The antidepressant, minaprine, did not modify the basal release of [3H]ACh but it antagonised the inhibitory effect of TFMPP on the K+-evoked release. The maximal reversal was found at 0.3 microM minaprine. These results suggest that minaprine interacts with heterologous presynaptic 5-HT1B receptors. A new approach is thus opened to the study of the mechanism of action of antidepressant drugs.