REAL-TIME MOLECULAR DETECTORS USING GRAMICIDIN ION CHANNEL NANO-BIOSENSORS Vikram Krishnamurthy, Kai Yiu Luk University of British Columbia Dept of Elec. and Comp. Engin. Vancouver, V6T 1Z4. Canada. Bruce Cornell AMBRI Pty Ltd. Chatswood, N.S.W. 2067. Australia. Don Martin University of Technology Sydney Sydney, N.S.W. 2007. Australia. ABSTRACT This paper deals with the experimental construction, stochastic mod- eling and statistical signal processing of a novel biosensor compris- ing of biological ion channels. Such nano-scale biosensors are built by incorporating dimeric Gramicidin ion channels into the bilayer membranes of giant unilamellar liposomes and then excising small patches of the membrane loaded with ion channels. We show that target molecules affect the statistics of the gating mechanism of the dimeric Gramicidin ion channels and present statistical verifications on the adequacy of a Hidden Markov Model for modeling of the biosensor. A likelihood ratio test is then devised to detect the pres- ence of target molecules. To test the sensitivity of this model we conducted patch-clamp experiments with and without the Methyl- benzthonium Chloride compound. The real-time detection algorithm was able to accurately detect the presence of the compound from al- terations in the patch-clamp recordings. This algorithm provides the sensitive detection system for ongoing development of lipid-based nano-sensors. Index Terms— Biomedical transducers, membrane ion chan- nels, Gramicidin, estimation, maximum likelihood detection 1. INTRODUCTION Ion channels are protein macromolecules commonly found in bio- logical cell membranes that form water filled nanotubes, typically a few Angstrom units in radius. In biological systems, ion channels selectively regulate the flow of ions into and out of a cell. By ex- ploiting the selective conductivity of ion channels in the presence of target molecules, biosensors are developed to detect molecular species of interest across a wide range of applications. These in- clude medical diagnostics, environmental monitoring and general bio-hazard detection. In particular, a novel biosenor, which incor- porated monomeric Gramicidin ion channels into a tethered lipid bi- layer membrane and exploited the changes in the association and disassociation probabilities of the Gramicidin dimers, was published by a coauthor of this paper in Nature. [1] This paper deals with the construction, modeling and statistical sig- nal processing associated with another ion channel based biosensor. In a giant lipid vesicle, covalent dimeric Gramicidin ion channels are incorporated by codispersion with the vesicle forming lipids. The gating mechanism in this biosensor is thought to arise from the ran- dom movement of excess lipid lenses in the liposome that diffuse over the membrane surface and block the conducting channels. We verify statistically that our Hidden Markov Model, which takes into account of the 1/f noise in the biosensor’s response, is an adequate model for the measured currents and can be used to derive impor- tant biological characteristics of these dimeric bis-gA ion channels. Using Bayesian signal processing methods on the output current of the biosensor, a likelihood ratio hypothesis test is devised to detect the presence of the target molecules. In the presence of the target molecules, the stochastic model of the output current of the biosen- sor changes. By detecting this model change in real time, we show that the biosensor can be used in real-time target molecule detection. 2. EXPERIMENTAL CONSTRUCTION OF THE MODEL BIOMIMETIC BIS-GA ION CHANNEL BIOSENSOR The biosensor considered in this paper was constructed by incorpo- rating bis-gA ion channels into the lipid bilayer membrane of giant unilamellar liposomes and then excising small patches (1 μm in di- ameter) of the lipid membrane using a patch-clamp micropipette. Figure 1 shows a fluorescence image of the optical section through the diameter of the biosensor. The solutions and chemicals used for the biosensor included bL-alpha-phosphatidylcholine (PC) from soybean, cholesterol, chloroform, sucrose, glucose, sodium chloride (NaCl), potassium chloride (KCl) and 4-(2-Hydroxyethyl)piperazine- 1-ethanesulfonic acid (HEPES). Fig. 1. Fluorescence image of biosensor’s horizontal optical section. The Gramicidin channels are labeled using FITC and identified by the green color. Giant unilamellar liposomes were prepared using our recent protocol [2] that was modified from a standard hydration procedure [3, 4, 5] with some modifications. In brief, a completely dried lipid film con- taining 100 μL of 10 mg·mL −1 PC with 10% (w/w) or 40%(w/w) cholesterol in chloroform was prepared in a glass test tube as de- scribed earlier. Rehydration of the lipid was later made at 45 ◦ C by the addition of a small amount of pure water (5 μL) to the tube for a few minutes (prehydration) followed by the addition of 5 mL of an aqueous solution of 0.1 M or 0.2 M sucrose. The tube was in- cubated at 45 ◦ C for 2-3 h. After gentle rocking overnight at room I  401 1424407281/07/$20.00 ©2007 IEEE ICASSP 2007