We often report “holding current” as the mean current over a range of time in a voltageclamp recording, but how do we cleanly measure holding current if fast synaptic neurotransmission (EPSCs and IPSCs) influence this mean? This page describes how to measure holding current (tonic current) in a way that is not influenced by fast (phasic) synaptic currents.
Experimental Data
Consider this voltageclamp experiment that adds a drug that changes holding current, but is confounded by the fact that it also changes sIPSC frequency. How do we measure the change in holding current (the slow upward shift of the trace) in a way that is not confounded by the fact that the drug also changes frequency and amplitude of synaptic events (fast downward deflections)? This is the problem that tonic/phasic analysis solves.
Raw Trace  Mean Current 

How to Isolate Tonic Current

Mean current (tonic current) in voltageclamp traces is “pulled” in the direction of spontaneous currents synaptic currents (phasic currents).

If a voltageclamp trace has no spontaneous synaptic currents, it is essentially a straight line. In practice the trace bobbles up and down because of noise.
Raw Trace With Synaptic Currents  Raw Trace Without Synaptic Currents 

 Noise is normally distributed, so if we represent all values from raw traces as a histogram of currents it will will appear as a Gaussian curve centered at the mean holding current.
Histogram With Synaptic Currents  Histogram Without Synaptic Currents 


The current value at the peak of the histogram represents tonic current and it is largely immune to phasic currents.

Revisiting the original data, using the peak of the histograms to measure holding current allows reporting of true tonic current even though spontaneous currents influence the mean current.
Raw Trace  Fitted Histograms 

Advanced Analyses
Curve Fitting

A Gaussian curve may be fitted to the clean portion of the histogram (uninfluenced by phasic currents), and the area of the histogram that falls outside the Gaussian curve serves as a quantitative measurement of phasic current.

Fitting a Gaussian curve to the histogram and reporting its center is slightly more accurate than simply reporting the value of the peak histogram bin.

The center of the Gaussian curve fitted to the center of the histogram is a good prediction of the tonic current which ignores the influence of phasic currents away from the center of the curve.

If phasic currents only extend in one direction (e.g., GABAergic IPSCs present when glutamatergic EPSCs are blocked with DNQX and AP5) you can fit the Gaussian curve to the data which is clean in the other direction.
Moving Baseline Subtraction
If recordings drift slowly up and down and/or the noise floor is extremely low, slow drift will influence the shape of the histogram more than the noise itself, complicating curve fitting and segregation of tonic and phasic currents.
One approach to overcome this issue is to measure the drift using a weighted moving window average (black line) then subtract it from the original trace to create a flatter trace.
Although this allows improved quantification of isolated phasic currents, this complicates the ability to determine tonic current since the trace is now centered at 0 pA.
Percentile Analysis
In conditions with extremely large and frequent spontaneous currents, an improved Gaussian curve may be fitted to only those values in the quietest portions of the trace. To achieve this, the trace can be broken into many segments a few milliseconds in length, then those segments are sorted by their standard deviation, and the lowest percentile segments can be included in the tonic/phasic analysis.
Additional Resources

Methods for recording and measuring tonic GABA_{A} receptormediated inhibition (Bright and Smart, 2013)

Python script used to generate these figures