The overall goal of the FollowMe experiment is to record neural spiking activity from a large number of neurons. This is achieved by observing the increase in intracellular calcium concentration in a neuron soma associated with each action potential using fluorescence microscopy in the first step. Two photon excitation and a random access scanning principle are used to record calcium fluorescent signals from 40 neuron somato with a high spatial and temporal resolution.
In the second step, the precise spike timings are extracted from the fluorescent signals using a deconvolution algorithm. This technique can be used to record spiking activity from tens of neurons. The spiking data then can be used to measure mutual information, signal propagation between neuronal populations or neuro correlations.
The main advantage of this technique over existing methods like micro electrode recordings, is that one can record neuron spiking activity from a large number of neurons from a local population. This method can help answer key questions about cortical function on the level of local networks of neurons, such as the importance of correlations seeking of information and changes in networks dynamics with cortico plasticity For two photon excitation, an infrared pulse laser system with femtosecond pulses is used. A high laser output power is required to offset the large losses introduced by the optical components of the system.
A preacher per system consisting of two prisms imparts a negative group velocity dispersion onto the laser pulses prior to the custo optical deflectors to compensate for the temporal dispersion introduced by the AODs two AODs with large apertures deflect the laser beam in two dimensions. A reflective diffraction grading with 100 grooves per millimeter is placed 13 centimeters behind the AODs to compensate the spatial dispersion introduced by the AODs. When using short laser pulses, the laser beam is directed with two relay telescopes into the camera port of an upright microscope.
Irises are placed at regular intervals for alignment of the optical components. A dichroic beam splitter in front of the objective transmits the infrared excitation light to the specimen and reflects the fluorescence light from the specimen onto a detector. Epi and trans fluorescence detectors collect fluorescence signal through the objective and the condenser colored glass filters are placed in front of the detectors to prevent excitation light from reaching the detectors.
The A OD deflection angles are controlled by a computer equipped with a digital analog converter board, which in turn drives voltage controlled oscillators. The signal from the photo multipliers is relayed through a low pass Butterworth filter with a cutoff frequency of 100 kilohertz and is digitized by an analog digital converter at 156.25 kilohertz clock rate before being stored on a computer for analysis, alignment and electrical noise are tested by recording the distribution of fluorescent signals with and without laser light at low and high gain of the photo multipliers, as well as with and without indicator dithered random access scanning relies on the detection of intracellular increases in calcium. The first step in the protocol is to stain a large number of neurons with the Esther form of a calcium indicator by bolus injection.
Next, record the activity of the soma at each neuron at four locations for 6.4 microseconds at each location. To select the neurons of interest, acquire a full frame consisting of 256 by 256 pixels. Then manually select the centers of the neurons to be recorded within this image.
Three points at two micron distance around each center are automatically added by the control software in each cycle. Fluorescent signal is recorded from the four locations in all 40 neurons. Recording from all neurons in one cycle requires 1.536 milliseconds.
Repeat this sequential recording of all 40 neurons for five seconds. Spike detection from fluorescent somatic calcium signals relies on a high signal to noise ratio. A high signal to noise ratio can be achieved by increasing the excitation intensity.
However, the excitation intensity can only be increased to a certain limit because of photo damage. There is only a very small window of excitation intensity where photo damage is low and signal to noise ratio is sufficient to detect single spikes in order to ensure that the recorded signals are within the window of high spike detection. During an experiment, a number of online analysis softwares developed by our lab are used.
Next, calculate the approximate photon rate per neuron from a short time window of baseline noise of about 100 to 200 milliseconds. Number of photons and photon rate are calculated from the fluorescence values by fitting the distribution of relative fluorescence changes with this equation, then calculate the baseline fluorescence from the same time window and plot it as a function of time or trials. Keep the average decline of baseline below 0.0002 per second by adjusting the laser power.
Because spike detection rapidly declines when exceeding this limit. After that, verify the neuron somatic positions every 10 to 20 minutes by acquiring a full frame image. Adjust the recording locations if required.
The fluorescent signals resulting from neural activity often ate in time because of the long decay of calcium transient. In this procedure, a deconvolution method is used to spikes and spike timings from the fluorescent signals. Here we use a genetic algorithm to determine the most likely spike train model that fits the recorded fluorescent signal.
With each iteration of the algorithm, the maximum likelihood asymptotically approaches its maximum that is given by the noise of the fluorescence recordings. In the in homogeneous populations of neurons, the spike evoked calcium signal can vary between neurons. For unsupervised analysis of large data sets, we designed an algorithm that takes into account the variation of the spike evoked calcium signal from neuron to neuron to avoid a large number of false positive detections.
The joint distribution of the amplitude and decay time constant of single spike evoked calcium transient are recorded in a separate set of experiments from the same type of neurons under the same experimental conditions. To account for slow baseline changes and to reduce computational costs of decon, longer recordings are divided into several shorter traces of one to five seconds for each neuron in each recording, the deconvolution algorithm may test a large number of models up to 1 million different models or more to speed up deconvolution one experiment is de convoluted on up to 10 different computers. In parallel.
After deconvolution, the spike data are analyzed and inspected. The peri stimulus time histogram spike probability and the neuron firing rate are calculated in an automated manner. Here are the examples of a fluorescent signal recorded at very low excitation rate and the other at very high excitation rate within the detection window.
Note that each example shows the response to one spike. Shown here is a photo micrograph of an acute brain slice and two stimulation pipettes placed in the same cortical column. In layer four, these show neural responses for the repetitive stimuli, and this graph shows Shannon's mutual information signaled by the recorded population of neurons.
Here is the measurement of the propagation of super threshold spiking activity between different populations of cortical neurons. These are the detected spikes in cortical neuron's, response to electrical stimulation of thal cortical fibers. When recording neuronal activity using random access scanning, it is very important to record within the detection window.
Otherwise, it's very easy to end up with unusable data sets. Optical spike detection is still in its infancy. There's a lot of room for improvement.
With some improvements, we record from even more neurons. With other improvements, we will be even able to record in awake and behaving animals.
