Inkjet Printing Methodologies for Drug Screening - Analytical

Mar 23, 2010 - This material is available free of charge via the Internet at http://pubs.acs.org. Reference .... Houston , J. G. and Banks , M. Curr. ...
0 downloads 0 Views 2MB Size
Anal. Chem. 2010, 82, 3104–3107

Letters to Analytical Chemistry Inkjet Printing Methodologies for Drug Screening Giuseppe Arrabito† and Bruno Pignataro*,‡ Scuola Superiore di Catania, Via San Nullo, 5/i, 95123, Catania, Italy, and Dipartimento di Chimica Fisica, “F. Accascina”, Universita` di Palermo, V. le Delle Scienze, Parco D’Orleans II, Ed. 17-90128, Palermo, Italy We show for the first time a contactless, low-cost, and rapid drug screening methodology by employing inkjet printing for molecular dispensing in a microarray format. Picoliter drops containing a model substrate (D-glucose)/ inhibitor (D-glucal) couple were accurately dispensed on a single layer consisting of the enzymatic target (glucose oxidase) covalently linked to a functionalized silicon oxide support. A simple colorimetric detection method allowed one to prove the screening capability of the microarray with the possibility to assay with high reproducibility at the single spot level. Measurements of the optical signal as a function of concentration and of time verified the occurrence at the solid-liquid interface of the competitive enzymatic inhibition with a similar behavior occurring for this system in a solution phase along with overcoming competition effects. We propose this methodology as a general application for drug screening purposes, since it may be extended to any kind of enzyme-substrate/ inhibitor or ligand-target biochemical system. The realization of high-speed, miniaturized, low-cost, and highthroughput biological devices in the microarray format is a very important issue in fields as biosensing, drug screening, environmental monitoring, forensic investigation, military defense, and so forth.1 The low-cost realization of these devices can be achieved by employing solution dispensing methodologies as microstamping, pin printing, and inkjet printing that do not require the use of masks, clean room facilities, or other costly equipments used in photolithography. Inkjet printing is a very promising dispensing methodology being a gentle and straightforward solution-drop deposition technique, enabling fragile biological materials to be printed on solid surfaces.2 In this respect, we already demonstrated its suitability for the fabrication of high-resolution microarrays with biological activity retention.3 Currently, as to the field of drug screening, conventional miniaturized screening technolo* Corresponding author. Phone: +39 091 645 98 46. Fax: +39 091 590 015. E-mail: [email protected]. † Scuola Superiore di Catania. ‡ Universita` di Palermo. (1) Pignataro, B. J. Mater. Chem. 2009, 19, 3338–3350. (2) Barbulovic-Nad, I.; Lucente, M.; Sun, Y.; Zhang, M.; Wheeler, A. R.; Bussmann, M. Crit. Rev. Biotechnol. 2006, 26, 237–259. (3) Arrabito, G.; Musumeci, C.; Aiello, V.; Libertino, S.; Compagnini, G.; Pignataro, B. Langmuir 2009, 25, 6312–6318.

3104

Analytical Chemistry, Vol. 82, No. 8, April 15, 2010

gies,4 like for instance those based on a robotic dispenser5 coupled with microwell arrays,6 do not satisfy ultrarapid and low-cost screening capabilities for the high-throughput screening of chemical libraries against hundreds of biological targets. These typically involve time and reagent consuming (micro-, nanoliter scale volumes) instrumental tools and facilities like clean rooms, photolithography, liquid handling robotics, complex detectors, and software for data processing.7 This work shows a low cost and high performance methodology based on an inkjet printing for delivering molecular systems in a picoliter drop microarray format coupled with an easily implemented detection tool that allows one to probe the target-drug interaction. In particular, we demonstrate the suitability of this approach for substrate/inhibitor molecular dispensing on their specific enzymatic target covalently linked to a solid support and quantifying the interaction by colorimetric detection. We optimized the dispensing parameters and demonstrated the effect of inhibition on the enzymatic activity for a model substrate/inhibitor/target (D-glucose/D-glucal/glucose-oxidase) system in a time and concentration dependent fashion. MATERIALS AND METHODS All reagents were purchased by Sigma Chemical Co., except for D-glucose and the below enzymatic kit (Megazyme, Ireland), and used without further purification. Target covered silicon oxide surfaces were prepared with a four-step immobilization procedure that involved plasma oxide activation, 3-aminopropyltriethoxysilane (APTES) silanization, glutharaldeyde (GA) derivatization, and finally Aspergillus niger glucose oxidase (GOx) soaking (see the Supporting Information). The immobilization of the protein was proved by X-ray photoelectron spectroscopy, and morphology of the active layer was investigated by scanning probe microscopy (see Figures S2 and S3 in the Supporting Information). Molecular inks containing either D-glucose or a mixture of D-glucose/D-glucal were dispensed on the GOx specific enzymatic target covalently linked to silicon oxide surfaces by using a Dimatix Materials Printer (DMP-2800, Fujifilm). Dispensed solutions (nominally of about 10 pL), containing a peroxidase-based kit for GOx (4) Houston, J. G.; Banks, M. Curr. Opin. Biotechnol. 1997, 8, 734–740. (5) Moore, K. W.; Newman, R.; Chan, G. K. Y.; Leech, C.; Allison, K.; Coulson, J.; Simpson, P. B. J. Assoc. Lab. Autom. 2007, 12, 115–123. (6) Khan, F.; Zhang, R.; Unciti-Broceta, A.; Dı´az-Mocho´n, J. J.; Bradley, M. Adv. Mater. 2007, 19, 3524–3528. (7) Hong, J.; Edel, J. B.; deMello, A. J. Drug Discovery Today 2009, 14, 134– 146. 10.1021/ac100169w  2010 American Chemical Society Published on Web 03/23/2010

activity assay (see the Supporting Information), D-glucose and D-glucal were spiked with glycerol at a relatively small percentage (30% v/v) for optimizing solution viscosity and surface tension without significantly affecting the biochemical reactions.3,8 D-glucose/D-glucal mixtures were employed in the whole compositional ratio range (from 0 to 1) as defined by χD-glucal ) nD-glucal/(nD-glucal + nD-glucose), where nD-glucal and nD-glucose are the moles of D-glucal and D-glucose, respectively (note that we maintained constant nD-glucose at 2.5 × 10-4 mol). Line spot microarrays were fabricated by a rectangle shape (1.375 mm × 0.975 mm). Typically 96 spots about 125 µm spaced were dispensed in sequential lines. The quality of the microarrays was represented by statistic parameters dealing with both dimensional and shape features of the spots (diameter, roundness, size regularity, alignment) measured on 40 spots (see the Supporting Information for the statistic parameters employed9). In order to print the microarrays, the voltage applied to the piezo for drop jetting was 20 V, while the pulse length of the waveform (the electrical signal applied to the piezo) was 11.52 µs (see Figure S4 in the Supporting Information). The colorimetric detection method consisted of the formation of a red colored quinoneimine dye, namely, 1,5dimethyl-4-(4-oxo-cyclohexa-2,5-dienyldeneamino)-2-phenyl-1,2dihydro-pyrazol-3-one, resulting from the formation of the product H2O2 from GOx mediated β-D-glucose oxidation.3 Optical images of the microarrays were obtained by a Nikon Eclipse ME600 microscope and managed by employing the software Lucia G on Mutech version 4.60 (Nikon, Italy). The spot signal detection was implemented by using Trace Histogram software. A rectangle that exactly circumscribed the spot region was defined. The signal consisted in the pixel distribution versus grayscale values ranging from 0 (black) to 250 (white). A double peak distribution was typically obtained, the peak at the highest grayscale values being associated with the spot signal (foreground) while the peak at the lowest grayscale values corresponds to the solidsupported enzymatic surface around the spot (background). RESULTS AND DISCUSSION In Figure 1, a sketch of the experimental procedure used for the screening methodology is reported. In part A, picoliter volume drops containing tens of picograms of molecules (either D-glucose or a mixture of D-glucose/D-glucal) are inkjet printed on a glucose oxidase monolayer immobilized on a silicon oxide surface. Upon hitting the solid surface, the dispensed drops form regular rounded spots (Figure 1B). The colorimetric detection based on the horseradish peroxidase method3 probes the interaction between the dispensed molecule and the enzymatic target at the single spot. Accordingly, whenever the enzymatic reaction occurs, a red dye complex forms. The more the spot is colored, the higher becomes the grayscale optical contrast of the spot with respect to the background. We verified that the solid-supported enzymatic surface can be reused for this screening purposes several times by simply rinsing with water. Moreover, the enzymatic activity of the array was only partially affected by aging in air for 3 months (see the Supporting Information). (8) Mugherli, L.; Burchak, O. N.; Balakireva, L. A.; Thomas, A.; Chatelain, F.; Balakirev, M. Y. Angew. Chem., Int. Ed. 2009, 48, 7639–7644. (9) Kim, P. G.; Park, K.; Cho, H. G. Int. J. Inf. Technol. 2005, 11, 117–124.

Figure 1. Pictorial sketch of the screening methodology here studied: (A) substrate and inhibitor containing drops dispensed by inkjet printing on the solid-supported enzymatic surface and (B) colorimetric detection for the enzymatic activity. No signal is obtained for a complete inhibition.

In parts A and B of Figure 2 are the optical images of alternated rich (absence of D-glucal) and D-glucose/D-glucal (χD-glucal ) 0.88) rich spotted lines, as freshly printed and after 90 min of incubation are shown, respectively. The spots had diameters of 38.8 ± 1.5 µm along with a coefficient of variation as low as 4% to be compared with those reachable by solid pin printing and by the widely used stealth pin printing techniques of about 2% and 12%, respectively.10 Also, the microarray spot definition was of high quality and comparable to that obtainable by pin printing methodologies,11 and a scale invariance of the spot roundness of 0.91 ± 0.04 along with a size regularity of 0.84 ± 0.09 were evaluated (see the Supporting Information). Moreover, the spot alignment error was 10.0 ± 4.8 µm. By comparing the two figures, one can easily observe that, in the absence of D-glucal, the brightness of the D-glucose rich spots is largely increased after 90 min of incubation. On this respect, Figure 2C reports grayscale color distributions (Gaussian fits) for two representative spots marked in Figure 2B as 1 and 2. Spot 1 is rich in D-glucose and shows a double-peak distribution with the foreground signal well distinguishable with respect to the background. As to Spot 2 (D-glucose/D-glucal rich) the foreground results in a shoulder of the background signal, i.e., only a scarce optical contrast is detected. For each spot we defined as optical intensity value (O.I.) the difference between the average foreground and the average background grayscale values. Note that the greatest optical intensity differences can be observed around the dots perimeter forming well contrasted halos. This can be likely explained by considering a solute transport to the ring periphery under drop evaporation12,13 that according to our D-glucose

(10) Mu ¨ ller, U. R.; Papen, R. In Microarray Technology and Its Applications; Mu ¨ ller, U. R., Nicolau, D. V., Eds.; Springer-Verlag: Berlin, Heidelberg, Germany, 2005; pp 73-88. (11) Dufva, M. Biomol. Eng. 2005, 22, 173–184. (12) Deegan, R. D.; Bakajin, O.; Dupont, T. F.; Huber, G.; Nagel, S. R.; Witten, T. A. Nature 1997, 389, 827–829.

Analytical Chemistry, Vol. 82, No. 8, April 15, 2010

3105

Figure 2. Optical images of a model screening array made of alternated D-glucose and D-glucose/D-glucal rich spotted lines (A) immediately after inkjet printing operation. The lines 1, 3, 5, and 7 marked by the arrows consist each of four D-glucose rich spots; (B) the same sampleregion after 90 min; (C) grayscale pixel intensity distribution with Gaussian fits for the regions marked by rectangles 1 (solid line) and 2 (dashed line) in part B enclosing representative D-glucose and D-glucose/D-glucal rich spots, respectively.

detection method results in light and dark perimeters for D-glucose rich and D-glucal rich drops, respectively. Unlikely, because of the low number of pixels, O.I. values cannot be calculated at the dot perimeter with significant statistics so that hereafter O.I. related to whole spots are reported. Figure 3A reports a graphic showing relative O.I. vs time for different χD-glucal values as measured at a single spot level. Relative O.I. values refer to the O.I. values at time t minus O.I. values at time zero. Note that reproducibility of measurements was quite good as demonstrated by the small error bars as well as by the coefficient of variation which was 11.4% as measured on about 40 different D-glucose rich spots. The signal variation is almost similar to that typically obtainable by fluorimetric detection methodologies.14,15 As expected, in the absence of D-glucal, O.I. increases hyperbolically (R2 ) 0.9939) with time due to the progress of the enzymatic reaction. The addition of D-glucal at χD-glucal ) 0.75 allows one to observe a different trend, which is well fitted by a sigmoidal curve (R2 ) 0.9959) matching the O.I. values of the above D-glucose spots after about 60 min of incubation. Such a behavior indicates that, in spite of possible problems occurring during heterophase assays,8 at our solid-supported enzymatic surface D-glucose outcompetes the D-glucal inhibition as it arises from a competitive inhibition mechanism.16 The inhibition occurs at similar concentration values already observed in the solution phase.17 The outcompeting behavior (13) Blossey, R.; Bosio, A. Langmuir 2002, 18, 2952–2954. (14) Jiang, L.; Yu, Z.; Du, W.; Tang, Z.; Jiang, T.; Zhang, C.; Lu, Z. Biosens. Bioelectron. 2008, 24, 376–382. (15) Litten, B.; Smith, R.; Banfield, E. J. Assoc. Lab. Autom. 2010, 15, 58–64.

3106

Analytical Chemistry, Vol. 82, No. 8, April 15, 2010

Figure 3. (A) Relative O.I. as a function of time for different χD-glucal [0 ) 0, O ) 0.75, ∆ ) 0.83, ] )0.88]. No significant O.I. increase has been found by employing χD-glucal g 0.88. (B) O.I. at 90 min vs χD-glucal in D-glucose:D-glucal rich spots. A spots sequence vs χD-glucal is inserted. Both in parts A and B, each value is averaged by five different spots.

is observed to be gradually lost by further increasing χD-glucal and seems to completely disappear at a molar fraction of 0.88 where no significant O.I. increase in time is observed. (16) Rowlands, A. R.; Panniers, R.; Henshaw, E. C. J. Biol. Chem. 1988, 263, 5526–5533. (17) Rogers, M. J.; Brandt, K. G. Biochemistry 1971, 10, 4624–4630.

In Figure 3B the effect of the inhibitor concentration on the GOx enzymatic activity at a fixed D-glucose concentration (0.071 M) is shown. In particular, O.I. values averaged on five different spots obtained after 90 min of incubation are reported as a function of the χD-glucal. A sigmoidal trend (R2 ) 0.9995) is observed with an inhibition threshold at a χD-glucal of 0.75. This behavior obtained at our solid/liquid interface well agrees with already reported literature dealing with such an enzyme-substrate/inhibitor system working in solution.17 Moreover, our single spot-based assay shows an acceptable quality factor, Z ) 0.65 (calculated on a 20 positive control assays, D-glucose rich spots, and 20 negative control assays, D-glucal rich spots), thus indicating the suitability of our method for high-throughput screening.18 The above findings clearly allow one to state that the proposed methodology could be used for drug-screening purposes in a microarray format. In addition, high-throughput drug screening would be easily reached by dispensing simultaneously, in a multichannel format, different molecular inks (see for instance ref 19 for oligonucleotide spotting) thus involving a large number of drug/target couples at the enzymatic solid-support surface. In order to generalize our methodology to every enzymesubstrate/inhibitor system and/or to every ligand-receptor biochemical couple, it is necessary to point out the following important issues: (1) The surface protein or receptor immobilization. Here we used a covalent cross-linking approach that is nonspecific and in principle could not preserve the function of any protein. If needed, affinity tag procedures could otherwise be used to guarantee the right molecule orientation and biological activity at the solid surface.20 (2) The signal transduction method. As a general colorimetric detection method via peroxidase (18) Zhang, J. H.; Chung, T. D.; Oldenburg, K. R. J. Biomol. Screen. 1999, 4, 67–73. (19) Cooley, P.; Hinson, D.; Trost, H.-J.; Antohe, B.; Wallace, D. In DNA Arrays: Methods and Protocols; Rampal, J. B. , Ed.; Humana Press: Totowa, NJ, 2001; Vol. 170, pp 117-129. (20) Zhu, H.; Snyder, M. Curr. Opin. Chem. Biol. 2003, 7, 55–63. (21) Vieira, A. Mol. Biotechnol. 1998, 10, 247–250. (22) Ulyashova, M. M.; Rubstova, M. Yu.; Batchmann, T.; Egorov, A. M. Moscow Univ. Chem. Bullettin 2008, 63, 75–79. (23) Eppinger, J.; Funeriu, D. P.; Miyake, M.; Denizot, L.; Miyake, J. Angew. Chem., Int. Ed. 2004, 43, 3806–3810.

oxidation one can for instance refer to ligand biotinylation and streptavidin-peroxidase conjugates21 as already shown for DNA hybridization.22 Moreover, if needed, sensitivity and signal/noise ratio may be obviously improved by using fluorophore tagging despite the low-cost colorimetric detection method used here.23 CONCLUSIONS Inkjet printing methodologies coupled with a simple and generalized detection method may satisfy rapid, low-cost, miniaturized (until single spot), and high-throughput screening requirements by easily spotting entire chemical libraries on solidsupported biological targets. In comparison to the conventionally used robotic systems coupled with photolithographic microwell arrays, our methodology allows one to realize at very low cost (no clean room facilities, prompt target-covered surface reusability, picoliter drop consumption) programmable microarrays in a up to 2 orders of magnitude faster (10 spots/s) fashion (e.g., refer to 0.1 spots/s of pin printing technique), a comparable positional accuracy and spot morphology by tuning only a few parameters like drop viscosity and surface tension, and the advantage to work not only on hard but also on soft surfaces without any risk of damage and with similar results of liquid phase assays. ACKNOWLEDGMENT The authors acknowledge Superlab (Consorzio Catania Ricerche) for his hospitality and Chiara Musumeci, Antonino Scandurra, and Giuseppe Francesco Indelli for their help. Italian MiUR (PRIN program) and University of Palermo are acknowledged for funding. SUPPORTING INFORMATION AVAILABLE Materials, solid-supported enzymatic surface preparation and characterization, jetting parameters for the microarray preparation, spot morphology and signal analysis of the microarray, and aging effects on the drug-screening device. This material is available free of charge via the Internet at http://pubs.acs.org.

Received for review January 20, 2010. Accepted March 15, 2010. AC100169W

Analytical Chemistry, Vol. 82, No. 8, April 15, 2010

3107