Differential Receptors Create Patterns Diagnostic for ATP and GTP Shawn C. McCleskey, Michael J. Griffin, Stephen E. Schneider, John T. McDevitt,* and Eric V. Anslyn* UniVersity of Texas at Austin, Department of Chemistry and Biochemistry, Austin, Texas 78712 Received September 30, 2002 ; E-mail: anslyn@ccwf.cc.utexas.edu Historically, chemical sensing has required the creation of a highly selective receptor for each component to be detected within a complex mixture. A more recent approach has been to use devices that rely on a series of chemo- or biosensors where analysis of complex mixtures arises from patterns produced by the combined response of all the sensors in the array. 1 This approach has been particularly successful for vapor-phase analysis. 2 Patterns can also be created that are diagnostic for single analytes. 3 We previously postulated that a series of random combinatorial receptors biased toward a class of analytes would be effective in an array setting. 4 To this end, we describe a sensing method utilizing a combinatorial library of receptors that can differentiate between highly structurally similar analytes, such as nucleotide phosphates, in water. The receptors consist of a rationally designed core with a binding cleft possessing guanidinium groups (Figure 1A). These guani- dinium groups impart an affinity for nucleotide triphosphates and are appended with tripeptides to incur differential binding properties. A previously reported screening of this same 4913-member library led to the identification a sensor (tripeptide Ser-Try-Ser) that was highly selective for adenosine 5′-triphosphate (ATP) over guanosine 5′-triphosphate (GTP). 5 The objective of this work is to determine whether the patterns generated by an unscreened library of receptors in an array can discriminate between structurally similar compounds, using ATP and GTP as test cases. Thirty beads from the library were randomly chosen, given an index number, and placed in a micromachined chip-based array platform that has been previously reported. 6 An electron micrograph of a representative 3 × 4 array platform is shown in Figure 1C. Sample delivery to the chemosensors occurs using a previously described flow cell. 6 A schematic of the flow cell is shown in Figure 1D. The sample is introduced over the array and passes around and through the beads to exit the bottom of the wells. Red, green, and blue (RGB) transmitted light intensity values were recorded for each bead in the array over the period of the assay via a charge- coupled device (CCD). The signaling protocol used with the array platform was an indicator-displacement assay similar to those exploited in many of our single-analyte sensing schemes. 7 A schematic of the indicator- displacement scheme for this system is shown in Figure 1B. To impart color to the library members, an anionic chromophore, fluorescein (2), was introduced into the array containing different members from the combinatorial library of receptors. The cationic receptors (1) associate with the indicator, bringing about a distinct orange color to each bead. Blank beads show no orange color, indicating little indicator uptake. Upon exposure to solutions of nucleotide phosphates, the analyte displaces the indicator, fluores- cein, at different rates (Figure 2A and 2B) and each bead loses color. The RGB intensity values for the 30 library beads in the array are recorded over time after a 2-mL injection of a 20 mM sample of ATP, GTP, or adenosine 5′-monophosphate (AMP) in 25 mM HEPES buffer (pH 7.5). Three trials were performed for ATP, GTP, and AMP for a total of nine trials, and absorbance values were calculated by taking the negative log of the ratio of the blue channel intensity over the red channel intensity for each bead. 6 In Figure 2, A and B show a representative normalized absorbance trace for two of the beads in the array after an injection of GTP and AMP, respectively. These traces reveal that each chemosensor responds differently to various nucleotide phosphate samples. The slope of the absorbance values from 210 to 435 s was calculated for each bead in each sample, and these values were used for analysis because the indicator displacement rates were found to be most reproducible in this region. The slope of bead 28 differs by 40% between the AMP and GTP trials, whereas the slope value for bead 23 in each plot differs by only 26%. Although these slopes are easily differentiated by qualitative visual inspection, the rates of displacement for several trials can be compared more quantitatively by using pattern recognition algorithms. Figure 1. A microscopic to macroscopic representation of the sensing protocol. (A) General molecular structure of resin bound library of receptors (1) and fluorescein (2). (B) Signal transduction scheme used to detect nucleotide phosphates within the resin bound sensor. AAn ) amino acid. (C) Scanning electron micrograph of a representative 3 × 4 micro-array containing glass beads. (D) Design of flow cell used in experiments. Published on Web 01/07/2003 1114 9 J. AM. CHEM. SOC. 2003, 125, 1114-1115 10.1021/ja021230b CCC: $25.00 © 2003 American Chemical Society