Discussion

Simulation of interactions in a coupled circuit of two HCOs using varying input parameters allowed for preliminary exploration of the parameter space of the Izhikevich and Morris-Lecar neuron models. Under feedforward connection schemes, distinct spiking patterns were observed under different sets of input parameters, which will help facilitate understanding of how the modulation of different variables affects neuron behavior, specifically that in the context of half-center oscillator circuits. Our results will contribute to the creation of a database of simulations of coupled HCO motifs that can be used in the future to help implement such spiking neuron network models in neuromorphic hardware. Additional work should be done to examine the interplay of synaptic connections and model parameters at a finer resolution.

Though there were multiple issues with the design of the dynamic clamp, it overall worked and fulfilled its design of being able to artificially add a conductance to a cell in a single-cell electrophysiology setup. If we had more time, it would have been prudent to have a deeper understanding of how the Arduino IDE code actually works so that we could troubleshoot better. The breadboard design should be replaced with a PCB so that there are no loose resistors, the device is more robust, and does not have to be tested every time that it is used. Lastly, understanding why the dynamic clamp currently raises the resting potential of the cell is something we would have done with more time since it greatly impacts the health of the cell.

Our experiments showed that the addition of a higher positive shunt or lead conductance to a cell increased the overall frequency of spiking. Cell 1 displayed a significant drop in spike frequency at the addition of 3nS of conductance which has led us to believe that there is a breaking point where the resting potential of a cell reaches threshold and spiking decreases dramatically. We did not test the rest of the cells to this hypothetical “breaking point” so we do not know for sure that it or general cell fatigue is what caused the sudden drop in frequency in Cell 1.

Adding a sodium conductance between 100nS and 140nS in leech Retzius cells resulted in frequent bursting behavior. This conclusion was drawn from experiments on one cell so it is a preliminary result that we will solidify in the future.

We collected data for the shunt experiments on four cells with two trials each which led to an acceptable amount of data to draw conclusions on. The sodium conclusions are less concrete since they only come from one cell that we tested twice. In the future, we plan to do more experiments on adding a sodium conductance. In general, more iterations of any experiments are necessary in biology since there is inherent variation in that can dramatically vary results.

• Single-cell electrophysiology techniques

• MATLAB

• Leech dissection

• Circuits

• Large-scale neuronal simulation

• Small-team collaboration and communication

One of the relevant social and ethical considerations for our project is the use of animals. Since our project involved the use of leech neurons for measuring the action potential and modulating its cells with the dynamic clamp, we made sure to work with as few leeches as possible. It was also very important to make sure to anesthetize the leech before dissection and to dissect it in the most humane way possible. Other ethical and social issues involve transparency about our research findings and facilitating the sharing of data and material from experiments on leeches in order to avoid unnecessary repetition of experiments.

For developing a database for a computational model of HCO, there are health and safety risks for the system biologists, neuromorphic engineers, and researchers. Since half of the project involves computation and simulation, there are ergonomic risks to our team members such as keyboarding causing repetitive motion, forceful and static exertions, awkward posture and contact stress and glare are risk factors at a poorly designed computer workstation. To reduce these risk factors, the team members made sure to remember to stretch, fix their workstation, and improve and practice good neck and back posture.

1. Edit the sodium tab of the Arduino IDE code to make it so we do not have to change the position knob to elicit a reaction from changing the sodium conductance.

2. Make a written manual for the use of the dynamic clamp with this specific preamplifier and DAQ since one of our mentors wishes to use it educationally and in his own experiments. It will be written and be accompanied by videos.

3. Make a printed circuit board or PCB of the dynamic clamp since we have had multiple issues with loose resistors in our breadboard and a PCB would be more compact and robust. We then would encase it for shielding and ease of use.

4. Model HCOs in a leech ganglion to prove that it is in fact possible to model HCOs artificially. Doing so would allow researchers to experiment with HCOs without having to find them naturally in the nervous system.

5. Run MATLAB simulations for 2 coupled HCOs using finer division and variation of input parameters.

6. Run MATLAB simulations for 3 coupled HCOs and update metrics analysis code where necessary.

7. Incorporate the HCO simulation data into NeuroDyn, a neuromorphic chip with four fully programmable Hodgkin-Huxley neurons, and do novel integration with real neurons.