Analysis of multichannel patch clamp recordings by hidden Markov models.

Conventional methods of analysis do not allow the kinetics of patch clamped ion channels to be completely determined if more than one channel is present in the patch. This hinders investigations on small ion channels as well as on channel cooperativity and the homogeneity of channel populations. We present a method to extract the rate constants and current amplitudes for each individual channel from multichannel patches by a one-step procedure. For this purpose, the current record is modeled by the superposed Markov processes of the opening and closing of each channel that is contaminated by noise (Hidden Markov Model). Channel parameters are obtained by maximum likelihood methods. Because the parameters can be calculated directly from the unfiltered record, the dwell time and missed event problems are widely diminished. Confidence bounds for the estimated parameters are given. Statistical tests to decide whether channels switch identically and/or independently are introduced. The application of the method is demonstrated with simulated data.