I'm curious to see what others say. I'm tempted to use what some in your part of the world would call the "blood obvious test". If you see an event, there is something there. If you can't eye ball it, it's probably not.

A more statistically minded approach is to take a 500 microsecond average (for example) 3 milliseconds before your presynaptic pulse (i.e. noise), and 3 milliseconds after your presynaptic pulse (i.e. signal). Then perform a reasonable statistical test on these two distributions. Look at the work by Somogyi (for instance) for examples, e.g. J Neurosci 1998 18(11): 4255-4270; or J Physiol 1998 506(3): 755–773. 30 events might not be enough, depending on how small the event is.

If you do it like Somogyi, you can't be far wrong.

Hi Simon,

if you have a monosypatic connection between your pair of neurons, than your PSCs will be almost time-locked with the action potential that you are inducing in the other neuron. From my previous recordings, the time-lapse between the AP and the recorded EPSPs were in the order of 0.4ms, which is pretty safe to say that it was a monosynaptic connection.

Another thing that you should keep in mind is if your postsynaptic currents were from a disynaptic connection, then that would mean that the AP that you are inducing in one neuron would be sufficient to elicit another AP in a second neuron before reaching your "recorded" neuron. But one knows that for a neuron to fire it requires more than one single input to reach AP threshold, so, if you see a postsynaptic current in your neurons by simple eliciting an AP in another neuron in your paired recordings, than, I would say that you are in presence of a monosynaptic connection.

Hope that this helps.

B.

Start by looking at the latency of the response in the 2nd cell. The delay will be short (a few ms at most) if it's monosynaptic. As the number of neurons and synapses between the two cells increases, so will the delay of the response in cell #2.

I think one of the classic references on synaptic delay is von Gersdorff et al, 1998: from the depolarization of voltage clamped presynaptic terminal to the EPSC in sniffer cell the delay was ~0.8 ms. In retina slice I measured ~1.5 ms for a reciprocal synapse (which is probably the lowest for a di-synaptic delay, from the depolarization of the presynaptic site to the IPCS at the same terminal).

As for detecting the existance of a synapse between the two cell , I agree with William Connelly above. With multiple stimulations you might have to consider the replenishment of presynaptic vesicles as well as other factors mediating short term plasticity.

Thank you very much everyone for this input

I am certain using the 'eyeball approach' that there is nothing there, but am looking for a way to statistically show that correctly.

I will read those reference suggested.

Thanks very much =)

By me, the usual way to do is to repeat stimulation serveral times and construct peristimulus time histogram (you place the zero point at the time of the stimulus an sum all the activity in bins before and after the srimulus). If you apply a narrow bin and calculate the Cumulative Sum, then you will get precise timing of the poststimulus event. It works even when your "eyeball approach' tells you there is nothing in the msecond cell. If you apply the frequencygram on top of it, you will be able to judge about the nature of interaction (inhibition vs. excitation). Reference: Piotrkiewicz M, Kudina L (2012). Analysis of motoneuron responses to composite synaptic volleys (computer simulation study). Exp. Brain Res 217(2): 209–221, DOI: 10.1007/s00221-011-2987-2

It all depends how much background synaptic activity you can see in your voltage-clamped cell. Spike triggered averaging is the most robust way: use the action potential of the current clamped cells to trigger an average the voltage clamped cell. Repeat this so that your average is compiled from as many spikes as you can record on the first channel - >100 if you can, more like 1000 if possible. You should have a smooth trace with a single synaptic current that occurs at a sensible latency (the latency will depend how far apart the two cells are, etc).

If you want to be really robust construct a 'dummy' channel of events that occur at the same average frequency as your I-clamped cell and use those to trigger an average of the same portion of data. It's the negative control - it should be perfectly flat. There are some nice examples of this technique from the respiratory literature -

Ono, K., Shiba, K., Nakazawa, K. & Shimoyama, I. Synaptic origin of the respiratory-modulated activity of laryngeal motoneurons. Neuroscience 140, 1079-108

What preparation are you using? Talk to Johanna Montgomery at U Auckland! (or look at one of her papers done in my lab ;-). If you are seeing nothing, you probably don't have a connection. If you are talking about a CNS synapse, the amplitude of a unitary synaptic current should be on the order of 5-10 pA or bigger. If you have good whole cell recording, your RMS noise should be on the order of 2 pA, so you should be able to see it without averaging. It should occur at a fixed 2 ish ms latency with regard to your presynaptic AP, and you should see it in at least 20-30 percent of your trials (i.e. at least 1/3rd ish of your APs, or more, should induce a synaptic current.

It seems like the critical measure is timing and its variation. A simple measure is to calculate the "jitter" (standard deviation of the latency) for a very low frequency series of shocks repeated 20-30 times. Since generation of an action potential will add to this variation between cells, it is likely best to use voltage clamp in the receiving neuron and measure the onset of the excitatory postsynaptic current or EPSC. If the focus is on whether the connection is monosynaptic, measuring under current clamp conditions and detecting the following action potential adds the step of generating the action potential in the following cell and adds time and variation to the measurement. This works well with axon activation of monosynaptic inputs. In our hands, monosynaptic connections have jitters of less than 200 microseconds for latencies that can be several milliseconds. See as an example M. W. Doyle and M. C. Andresen. Reliability of monosynaptic transmission in brain stem neurons in vitro. J.Neurophysiol. 85 (5):2213-2223, 2001.

In our pair recordings, the majority of the jitter is in the presynaptic action potential latency. The jitter in the postsynaptic response is pretty minimal. We just wrote our software to find the peak of the AP and the plot the postsynaptic response locked to the AP latency, which got rid of most of the jitter.