- Email: [email protected]
Monitoring “promiscuous” drug effects on single cells of multiple cell types
ANALYTICAL BIOCHEMISTRY Analytical Biochemistry 345 (2005) 320–325 www.elsevier.com/locate/yabio
Monitoring “promiscuous” drug eVects on single cells of multiple cell types Yina Kuang, David R. Walt ¤ Department of Chemistry, Tufts University, Medford, MA 02155, USA Received 24 May 2005 Available online 14 July 2005
Abstract Recent advances in genomics and molecular pathogenesis studies have determined that many diseases are caused by a multiplicity of factors. New drug regimens may consist of multiple biologically active agents designed to act synergistically on multiple biochemical targets. Live cell assays are becoming a standard for identifying new drug candidates with an emphasis on “homogeneous” living cell assays in which multiple cell lines are mixed and monitored simultaneously. In this study, we used a high-density single living cell array, based on an optical imaging Wber bundle microwell array, to simultaneously monitor “promiscuous” drug eVects on single cells of multiple cell types. Such a capability allows for a more comprehensive understanding of how cells dynamically respond to combinatorial drug libraries or how diVerent cellular pathways and regulation circuits respond cooperatively to drugs in individual cells. 2005 Elsevier Inc. All rights reserved. Keywords: Promiscuous drug eVects; Single cell measurement; Multiple cell types; Optical Wber cell array
As we enter the postgenome era, cells are becoming the new frontier for drug discovery as raw sequence information is translated into understanding how cells work and how disease processes operate. In living cells, the dynamic interactions of molecules give rise to a cell’s biological function. Such dynamic interactions enable living systems to respond and adapt when challenged with external stimuli such as drugs [1]. For this reason, living cell-based screens are becoming increasingly popular. Cell screens can be easily miniaturized to enable high-throughput and reduced costs while delivering high-content information, including speciWcity, target activity, toxicity, and bioavailability [2,3]. Living cells are organized at various levels of cellular networks and circuits [4]. The way in which a given cell reacts to a drug—as an agonist, as an antagonist, or with no response at all—depends on multiple factors inXu*
Corresponding author. Fax: +1 617 627 3443. E-mail address: [email protected] (D.R. Walt).
0003-2697/$ - see front matter 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.ab.2005.06.028
enced by a number of diVerent targets [5–7]. With the complexity of cell regulatory networks and pathways, the traditional single drug–single target approach to drug design is limited. Moreover, current knowledge obtained from both DNA or mRNA screens and molecular pathogenesis studies has determined that many diseases are caused by a multiplicity of factors [8,9]. To provide greater overall eYcacy, new drug design strategies may consist of multiple biologically active agents designed to act synergistically on multiple biochemical targets. To investigate multiple therapeutic agents targeted toward multiple cell targets, there is an increased emphasis on “homogeneous” assays in which multiple cell lines, each engineered to report one or more target activity, are mixed and monitored simultaneously [10,11]. Performing such assays requires technology platforms capable of dynamic measurements of “promiscuous” drug eVects on multiple cell lines simultaneously. Traditionally, most living cell assays have been conducted by collecting an averaged population response
Monitoring “promiscuous” drug eVects / Y. Kuang, D.R. Walt / Anal. Biochem. 345 (2005) 320–325
from cell cultures using standard microtiter plates or cuvettes. For drug discovery, it might be important to understand the diVerences between single cells because toxicity and/or eYcacy may be manifested by individual cell responses rather than by the ensemble response. Unlike the genome, there are potentially thousands of phenotypes in an isogenic cell population after drug application. Because each cell has a unique composition depending on its history and the unique microenvironment in which it resides, diVerent cells in an ostensibly identical cell population can give rise to diVerent responses [12]. Cell culture-based methods provide only a single averaged response, thereby eliminating potentially important information about cell-to-cell variations. To observe thousands of individual cells simultaneously and dynamically, we developed high-density living cell arrays based on ordered optical Wber microwell arrays. This platform consists of microwells etched into the end of optical Wber bundles, followed by randomly dispersing living cells into the microwells such that each microwell accommodates a single cell [13]. The diameter and depth of the microwells can be controlled and tailored to accommodate diVerent types and sizes of cells [14]. These arrays are well suited for analyzing individual cells repeatedly over time. Furthermore, cells can be exposed to various experimental conditions because the arrays have ready access to the solution in which they reside.
Materials and methods Strain and media Escherichia coli strains MG1655+pUA2699 and MG1655+pECFP contained the gene fusions recA::gfp and lacZ::ecfp, respectively. pUA2699 plasmid, originated from a low-copy number pSC101 plasmid, was provided by Uri Alon (Weizman Institute of Sciences, Israel). The plasmid pECFP(lacZ::ecfp) was purchased from Clontech (Palo Alto, CA, USA) and originated from a high-copy number pUC plasmid. Transformations were performed using the TransformAid Bacterial Transformation Kit (Fermentas, Hanover, MD, USA). The cells were cultured overnight in M9 minimal medium (Becton Dickinson, Le Pont de Claix, France) supplemented with 2 mM MgSO4, 0.1 mM CaCl2, 0.4% glycerol, 0.1% casamino acid (FisherBiotech, Fair Lawn, NJ, USA), and 50 g/ml kanamycin sulfate (Fisher ScientiWc, Fair Lawn, NJ, USA) or 100 g/ml ampicillin (Fisher ScientiWc) at 37 °C in an incubator shaker (New Brunswick ScientiWc, Edison, NJ). Fresh cultures were prepared by diluting the overnight culture 1:50 and incubating at 37 °C until the OD 600 reached 0.1 as measured using the Beckman DU 530 Life Science UV–Vis spectrophotometer (Beckman Coulter, Fullerton, CA, USA).
321
Imaging Wber-based living cell array fabrication The etched end of an imaging Wber bundle containing 3.1 m microwells (Illumina, San Diego, CA, USA) was covered with a thin layer of polyethylenimine (Sigma, St. Louis, MO, USA) by applying 1% polyethylenimine solution onto the etched Wber surface and allowing the solution to dry. A 10-mm long polyurethane tube with a 1-mm inner diameter (Small Parts, Miami Lakes, FL, USA) was attached to the etched end of the Wber to form an open chamber, with the etched end of the Wber forming the bottom of the chamber. Aliquots (10 l) of 1:100 cell dilutions from fresh cultures were loaded onto the chamber, and the Wber was centrifuged horizontally at 4000 rpm for 2 min. Measurements and data analysis The imaging Wber containing the single living cell array was mounted on an epiXuorescence microscope (model IX81, Olympus America, Melville, NY, USA). Fluorescence images were acquired from the proximal end of the Wber by a charge-coupled device (CCD)1 camera (Orca-ER, Hamamatsu, Japan). The cell array was Wrst exposed to fresh M9 medium supplemented with isopropyl--D-thiogalactoside (IPTG), mitomycin C (MMC), or both for a time period as speciWed in the text. The Xuorescence intensity was measured by IPlab software (Scanalytics, Fairfax, VA, USA). Fluorescence signals [480 nm excitation/520 nm emission for green Xuorescent protein (GFP) and 440 nm excitation/480 nm emission for enhanced cyan Xuorescent protein (ECFP)] from individual cells were measured with a 500-ms acquisition time immediately after exposing the cell array to M9 medium containing the compound(s). Fluorescence signals were measured every 5 min. All of the results were expressed as Xuorescence intensity percentage increases (It¡I0)/I0.
Results and discussion We previously demonstrated that individual living cells remain healthy and continue functioning after being inserted into microwells [15]. In the current work, we demonstrate the ability to simultaneously observe promiscuous drug eVects on single cells of multiple cell types in a single array format. The cell array is built on a 1-mm diameter optical Wber bundle that provides approximately 50,000 optically addressable microwells (2 £ 107 elements/cm2). A mixture of two E. coli cell types, MG1655+pUA2699(recA::gfp) [16] and MG1655 1 Abbreviations used: CCD, charge-coupled device; IPTG, isopropyl-D-thiogalactoside; MMC, mitomycin C; GFP, green Xuorescent protein; ECFP, enhanced cyan Xuorescent protein.
322
Monitoring “promiscuous” drug eVects / Y. Kuang, D.R. Walt / Anal. Biochem. 345 (2005) 320–325
+pECFP(lacZ::ecfp), both in the early exponential growth stage, were loaded onto an optical Wber well array. These two cell strains can report cell responses speciWcally to two diVerent drug stimuli based on two completely diVerent cellular systems. The Wrst cell strain produces GFP on DNA damage in response to exposure to generic genotoxins such as MMC. The second strain responds to nutrient starvation in which cellular glucose levels are low, and an alternative carbon source, lactose or its synthetic analogs (e.g., IPTG), is available. These cells are engineered to produce ECFP in response to switching the primary sugar metabolism from glucose to lactose. To observe multiple drug eVects on single cells of multiple cell types, a duplexed cell array was exposed to a deWned M9 medium supplemented with both 10 g/ml MMC and 5 mM IPTG. Fluorescence images (480 nm excitation/520 nm emission for GFP and 440 nm excita-
tion/480 nm emission for ECFP) were taken with 500-ms exposure times before drug exposure and every 5 min for 90 min after drug exposure. The locations of the two diVerent cell types in the array can be determined by observing their respective optical signatures. To account for the diVerences between the optical properties of GFP and ECFP, the Xuorescence signal from each cell was recorded as arbitrary units at any time t (It) but is reported as a percentage increase, that is, (It ¡ I0)/I0. Figs. 1A and B show individual cell response dynamics for both strains after exposure to the two compounds. Because both Xuorescent proteins have long half-lives [17], as expected, both strains exhibit steadily increasing Xuorescence levels over time in response to the two stimuli (MMC and IPTG). For each cell strain, although the cells are isogenic and exposed to the same stimuli, each cell exhibits its own individual response proWle. Such variation underscores how ensemble cell culture-based
Fig. 1. Monitoring individual cell responses from two bacterial cell types, MG1655+pECFP(lacZ::ecfp) and MG1655+pUA2699(recA::gfp) cells, to MMC and IPTG simultaneously on a single cell array format. The MG1655+pECFP(lacZ::ecfp) and MG1655+pUA2699(recA::gfp) cell responses were monitored by measuring the ECFP (A) and GFP (B), respectively. In a single exposure experiment, a cell array was exposed to medium supplemented with 5 mM IPTG and 10 g/ml MMC for 90 min. The insets show the Xuorescence level distribution across each cell population at 90 min. Error bars represent the standard deviation from three repeated assays.
Monitoring “promiscuous” drug eVects / Y. Kuang, D.R. Walt / Anal. Biochem. 345 (2005) 320–325
measurements can average out important information regarding cell-to-cell variability and may result in erroneous conclusions about the eVects of drugs on living systems. Initially, we expected dramatic diVerences in the Xuorescence signals between the two strains because there is a signiWcant diVerence in the plasmid copy numbers for the two strains; that is, the lacZ::ecfp fusion is located on a high-copy number plasmid (»500 copies), whereas the recA::gfp fusion is located on a low-copy number plasmid (»5 copies). For ease of comparison, the dynamics of the average responses of both cell strains are displayed in Figs. 2A and B, respectively. Even though MG1655+pECFP (lacZ:: ecfp) cells demonstrated a higher Xuorescence increase (0.42 § 0.13 at 90 min) as compared with MG1655+ pUA2699(recA::gfp) cells (0.17 § 0.04 at 90 min), the diVerence is not as dramatic as expected. For
323
MG1655+pECFP(lacZ::ecfp) cells, they respond not only to their own inducer IPTG but also to MMC. MMC is a strong genotoxin [18]. In our previous study, we showed that a concentration of 10 g/ml MMC is the threshold concentration for both genotoxicity and cytotoxicity [15]. We expected a possible cytotoxic eVect of MMC on MG1655+pECF P(lacZ::ecfp) cells. For MG1655+pUA2699(recA::gfp) cells, they respond not only to their own inducer MMC but also to IPTG. The activation of the recA gene is the Wrst step in activating bacterial SOS responses involved in DNA damage tolerance and error-prone replication. It has been reported that bacterial cells activate SOS responses under a variety of environmental stress conditions in addition to DNA damage [19–21], including high osmotic conditions [22]. In vitro studies have suggested that more than 1000 genes are involved in the DNA damage response in E. coli cells [23]. Therefore, even
Fig. 2. (A) Average responses of MG1655+pECFP(lacZ::ecfp) cells when incubated only with IPTG (solid circles) and when coincubated with IPTG and MMC (solid squares). (B) Average responses of MG1655+pUA2699(recA::gfp) cells when incubated with MMC and IPTG (solid squares), only with IPTG (solid circles), or with MMC after preincubation with IPTG for 115 min (solid diamonds). For ease of comparison, the IPTG-only response was moved up 0.14 U (dotted line) to compare its slope with that of the MMC after preincubation with IPTG for 115 min.
324
Monitoring “promiscuous” drug eVects / Y. Kuang, D.R. Walt / Anal. Biochem. 345 (2005) 320–325
though IPTG is neither a toxin nor the direct inducer for recA::gfp gene plasmid, MG1655+pUA2699 (recA::gfp) cells may respond to IPTG through the activation of the SOS response system. These results demonstrate the ability of the optical Wber array to simultaneously monitor multiple responses from two cell strains to two drugs. To test such possible promiscuous drug eVects, a second cell array was subjected to a sequential exposure experiment, Wrst to 5 mM IPTG for 115 min and then to 10 g/ml MMC for 65 min. Single cell response dynamics from both strains are displayed in Figs. 3A and B. Immediately after exposure to IPTG, a rapid increase in ECFP signals was obtained from MG1655+pECFP(lacZ:: ecfp) cells. The ECFP level continued to accumulate until 10 min after the medium was replaced with a medium containing MMC. Comparing the average responses (Fig. 2A) over the entire course (Wrst 90 min shown), the EGFP level is much higher in cells not coincubated with MMC than in cells coincubated with MMC, clearly indicating the toxic eVect of MMC on MG1655+pECFP(lacZ::ecfp) cells at a concentration of 10 g/ml. The termination of ECFP production in cells after the medium replacement was due to the toxic eVect of MMC on cells and the dilution of IPTG from cells given that IPTG is a small molecule that can freely diVuse through the cell membrane [24]. On the other hand, IPTG displayed an activation eVect on the GFP
level in MG1655+pUA2699(recA::gfp) cells, which show a low GFP signal increase when Wrst exposed to IPTG (Fig. 2B [dashed lines] and Fig. 3B), probably due to the response of the SOS gene networks to stress conditions caused by the very high concentration of IPTG (as we suspected). This initial low induction of GFP by IPTG allows a much faster cell response to MMC after the medium was replaced (Fig. 2B, solid diamonds), indicating that cell history (e.g., pretreatment with other drugs) has a signiWcant impact on the kinetic proWle of a speciWc drug to cells. These results underscore the synergistic eVects resulting from the simultaneous application of multiple active biomolecules. The genome-centric view of drug discovery is being replaced by a more comprehensive cell-centric view. The work reported in this article shows how optical microwell cell arrays can provide rich information in a single assay about the cross-activities of two “drugs” acting on two seemingly irrelevant cellular systems: a bacterial SOS response system and a sugar metabolic system. In conjunction with optical assays engineered to be expressed within living cells [25–27], such as other GFP variants with diVerent spectral properties, more diverse cell types can be accommodated on a single array. Such a capability will beneWt the drug screening process enormously by allowing for a more comprehensive understanding of how cells dynamically respond to combinatorial drug libraries or by examining how diVerent
Fig. 3. Monitoring individual cell responses from the two bacterial cell types sequentially exposed to 5 mM IPTG for 115 min and to 10 g/ml MMC for an additional 65 min. (A) Temporal proWle of individual MG1655+pECFP(lacZ::ecfp) cell responses. (B) Temporal proWle of individual MG1655+pUA2699(recA::gfp) cell responses. The change of medium at 115 min from IPTG to MMC caused a glitch in the Xuorescence intensity.
Monitoring “promiscuous” drug eVects / Y. Kuang, D.R. Walt / Anal. Biochem. 345 (2005) 320–325
cellular pathways and regulation circuits respond cooperatively to drugs in individual cells. This technique oVers a distinct advantage not only in high-content screening of combinatorial libraries of potential drug candidates but also in bioprocesses monitoring, environmental pollution monitoring, and diagnostic screening.
Acknowledgments We thank Uri Alon (Weizmann Institute of Science, Israel) for providing plasmid pUA2699 and Israel Biran for providing helpful discussions. References [1] A. Wagner, Robustness, evolvability, and neutrality, FEBS Lett. 579 (2005) 1772–1778. [2] J.J. Burbaum, Miniaturization technologies in HTS: how fast, how small, how soon? Drug Discov. Today 3 (1998) 313–322. [3] R. Kapur, K.A. Giuliano, M. Campana, T. Adams, K. Olson, D. Jung, M. Mrksich, C. Vasudevan, D.L. Taylor, Streamlining the drug discovery process by integrating miniaturization, high throughput screening, high content screening, and automation on the CellChip system, Biomed. Microdevices 2 (2000) 99–109. [4] M.P. Cary, G.D. Bader, C. Sander, Pathway information for systems biology, FEBS Lett. 579 (2005) 1815–1820. [5] S. Sen, M. Ferdig, QTL analysis for discovery of genes involved in drug responses, Curr. Drug Targets Infect. Disord. 4 (2004) 53–63. [6] M. Whitney, E. Rockenstein, G. Cantin, T. Knapp, G. Zlokarnik, P. Sanders, K. Durick, F.F. Craig, P.A. Negulescu, A genome-wide functional assay of signal transduction in living mammalian cells, Nat. Biotechnol. 16 (1998) 1329–1333. [7] T. Hampton, “Promiscuous” anticancer drugs that hit multiple targets may thwart resistance, J. Am. Med. Assoc. 292 (2004) 419–422. [8] V.C. Stephenson, R.A. Heyding, D.F. Weaver, The “promiscuous drug concept” with applications to Alzheimer’s disease, FEBS Lett. 579 (2005) 1338–1342. [9] M.B.H. Youdim, J.J. Buccafusco, Multi-functional drugs for various CNS targets in the treatment of neurodegenerative disorders, Trends Pharmacol. Sci. 26 (2005) 27–35. [10] K.L. Tucker, Y. Wang, J. Dausman, R. Jaenisch, A transgenic mouse strain expressing four drug-selectable marker genes, Nucleic Acids Res. 25 (1997) 3745–3746. [11] S.A. Sundberg, High-throughput and ultra-high-throughput screening: solution- and cell-based approaches, Curr. Opin. Biotechnol. 11 (2000) 47–53.
325
[12] M. May, G. Heebner, Laboratory technology trends: drug discovery (part 1), Science 295 (2002) 1349–1364. [13] I. Biran, D.R. Walt, Optical imaging Wber-based single live cell arrays: a high-density cell assay platform, Anal. Chem. 74 (2002) 3046–3054. [14] C. DiCesare, I. Biran, D.R. Walt, Individual cell migration analysis using Wber-optic bundles, Anal. Bioanal. Chem. 382 (2005) 37–43. [15] Y. Kuang, I. Biran, D.R. Walt, Living bacterial cell array for genotoxin monitoring, Anal. Chem. 76 (2004) 2902–2909. [16] M. Ronen, R. Rosenberg, B.I. Shraiman, U. Alon, Assigning numbers to the arrows: parameterizing a gene regulation network by using accurate expression kinetics, Proc. Natl. Acad. Sci. USA 99 (2002) 10555–10560. [17] R. Tombolini, A. Unge, M.E. Davey, F.J. de Bruijn, J.K. Jansson, Flow cytometric and microscopic analysis of GFP-tagged Pseudomonas Xuorescens bacteria, FEMS Microbiol. Ecol. 22 (1997) 17–28. [18] L.R. Ptitsyn, G. Horneck, O. Komova, S. Kozubek, E.A. Krasavin, M. Bonev, P. Rettberg, A biosensor for environmental genotoxin screening based on an SOS lux assay in recombinant Escherichia coli cells, Appl. Environ. Microbiol. 63 (1997) 4377–4384. [19] A. Aris, J.L. Corchero, A. Benito, X. Carbonell, E. Viaplana, A. Villaverde, The expression of recombinant genes from bacteriophage lambda strong promoters triggers the SOS response in Escherichia coli, Biotechnol. Bioeng. 60 (1998) 551–559. [20] C. Miller, L.E. Thomsen, C. Gaggero, R. Mosseri, H. Ingmer, S.N. Cohen, SOS response induction by -lactams and bacterial defense against antibiotic lethality, Science 305 (2004) 1629–1632. [21] J. Lee, H.C. Kim, S.W. Kim, S.I. Hong, Y.H. Park, Interplay of SOS induction, recombinant gene expression, and multimerization of plasmid vectors in Escherichia coli, Biotechnol. Bioeng. 80 (2002) 84–92. [22] S.M. Ruzal, C. Sanchez-Rivas, Physiological and genetic characterization of the osmotic stress response in Bacillus subtilis, Can. J. Microbiol. 40 (1994) 140–144. [23] P. Khil Pavel, R.D. Camerini-Otero, Over 1000 genes are involved in the DNA damage response of Escherichia coli, Mol. Microbiol. 44 (2002) 89–105. [24] A. Khlebnikov, J.D. Keasling, EVect of lacY expression on homogeneity of induction from the Ptac and Ptrc promoters by natural and synthetic inducers, Biotechnol. Prog. 18 (2002) 672–674. [25] J.E. Gonzalez, P.A. Negulescu, Intracellular detection assays for high-throughput screening, Curr. Opin. Biotechnol. 9 (1998) 624– 631. [26] D.L. Taylor, E.S. Woo, K.A. Giuliano, Real-time molecular and cellular analysis: the new frontier of drug discovery, Curr. Opin. Biotechnol. 12 (2001) 75–81. [27] J. Lippincott-Schwartz, G.H. Patterson, Development and use of Xuorescent protein markers in living cells, Science 300 (2003) 87–91.
Monitoring “promiscuous” drug eVects on single cells of multiple cell types Yina Kuang, David R. Walt ¤ Department of Chemistry, Tufts University, Medford, MA 02155, USA Received 24 May 2005 Available online 14 July 2005
Abstract Recent advances in genomics and molecular pathogenesis studies have determined that many diseases are caused by a multiplicity of factors. New drug regimens may consist of multiple biologically active agents designed to act synergistically on multiple biochemical targets. Live cell assays are becoming a standard for identifying new drug candidates with an emphasis on “homogeneous” living cell assays in which multiple cell lines are mixed and monitored simultaneously. In this study, we used a high-density single living cell array, based on an optical imaging Wber bundle microwell array, to simultaneously monitor “promiscuous” drug eVects on single cells of multiple cell types. Such a capability allows for a more comprehensive understanding of how cells dynamically respond to combinatorial drug libraries or how diVerent cellular pathways and regulation circuits respond cooperatively to drugs in individual cells. 2005 Elsevier Inc. All rights reserved. Keywords: Promiscuous drug eVects; Single cell measurement; Multiple cell types; Optical Wber cell array
As we enter the postgenome era, cells are becoming the new frontier for drug discovery as raw sequence information is translated into understanding how cells work and how disease processes operate. In living cells, the dynamic interactions of molecules give rise to a cell’s biological function. Such dynamic interactions enable living systems to respond and adapt when challenged with external stimuli such as drugs [1]. For this reason, living cell-based screens are becoming increasingly popular. Cell screens can be easily miniaturized to enable high-throughput and reduced costs while delivering high-content information, including speciWcity, target activity, toxicity, and bioavailability [2,3]. Living cells are organized at various levels of cellular networks and circuits [4]. The way in which a given cell reacts to a drug—as an agonist, as an antagonist, or with no response at all—depends on multiple factors inXu*
Corresponding author. Fax: +1 617 627 3443. E-mail address: [email protected] (D.R. Walt).
0003-2697/$ - see front matter 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.ab.2005.06.028
enced by a number of diVerent targets [5–7]. With the complexity of cell regulatory networks and pathways, the traditional single drug–single target approach to drug design is limited. Moreover, current knowledge obtained from both DNA or mRNA screens and molecular pathogenesis studies has determined that many diseases are caused by a multiplicity of factors [8,9]. To provide greater overall eYcacy, new drug design strategies may consist of multiple biologically active agents designed to act synergistically on multiple biochemical targets. To investigate multiple therapeutic agents targeted toward multiple cell targets, there is an increased emphasis on “homogeneous” assays in which multiple cell lines, each engineered to report one or more target activity, are mixed and monitored simultaneously [10,11]. Performing such assays requires technology platforms capable of dynamic measurements of “promiscuous” drug eVects on multiple cell lines simultaneously. Traditionally, most living cell assays have been conducted by collecting an averaged population response
Monitoring “promiscuous” drug eVects / Y. Kuang, D.R. Walt / Anal. Biochem. 345 (2005) 320–325
from cell cultures using standard microtiter plates or cuvettes. For drug discovery, it might be important to understand the diVerences between single cells because toxicity and/or eYcacy may be manifested by individual cell responses rather than by the ensemble response. Unlike the genome, there are potentially thousands of phenotypes in an isogenic cell population after drug application. Because each cell has a unique composition depending on its history and the unique microenvironment in which it resides, diVerent cells in an ostensibly identical cell population can give rise to diVerent responses [12]. Cell culture-based methods provide only a single averaged response, thereby eliminating potentially important information about cell-to-cell variations. To observe thousands of individual cells simultaneously and dynamically, we developed high-density living cell arrays based on ordered optical Wber microwell arrays. This platform consists of microwells etched into the end of optical Wber bundles, followed by randomly dispersing living cells into the microwells such that each microwell accommodates a single cell [13]. The diameter and depth of the microwells can be controlled and tailored to accommodate diVerent types and sizes of cells [14]. These arrays are well suited for analyzing individual cells repeatedly over time. Furthermore, cells can be exposed to various experimental conditions because the arrays have ready access to the solution in which they reside.
Materials and methods Strain and media Escherichia coli strains MG1655+pUA2699 and MG1655+pECFP contained the gene fusions recA::gfp and lacZ::ecfp, respectively. pUA2699 plasmid, originated from a low-copy number pSC101 plasmid, was provided by Uri Alon (Weizman Institute of Sciences, Israel). The plasmid pECFP(lacZ::ecfp) was purchased from Clontech (Palo Alto, CA, USA) and originated from a high-copy number pUC plasmid. Transformations were performed using the TransformAid Bacterial Transformation Kit (Fermentas, Hanover, MD, USA). The cells were cultured overnight in M9 minimal medium (Becton Dickinson, Le Pont de Claix, France) supplemented with 2 mM MgSO4, 0.1 mM CaCl2, 0.4% glycerol, 0.1% casamino acid (FisherBiotech, Fair Lawn, NJ, USA), and 50 g/ml kanamycin sulfate (Fisher ScientiWc, Fair Lawn, NJ, USA) or 100 g/ml ampicillin (Fisher ScientiWc) at 37 °C in an incubator shaker (New Brunswick ScientiWc, Edison, NJ). Fresh cultures were prepared by diluting the overnight culture 1:50 and incubating at 37 °C until the OD 600 reached 0.1 as measured using the Beckman DU 530 Life Science UV–Vis spectrophotometer (Beckman Coulter, Fullerton, CA, USA).
321
Imaging Wber-based living cell array fabrication The etched end of an imaging Wber bundle containing 3.1 m microwells (Illumina, San Diego, CA, USA) was covered with a thin layer of polyethylenimine (Sigma, St. Louis, MO, USA) by applying 1% polyethylenimine solution onto the etched Wber surface and allowing the solution to dry. A 10-mm long polyurethane tube with a 1-mm inner diameter (Small Parts, Miami Lakes, FL, USA) was attached to the etched end of the Wber to form an open chamber, with the etched end of the Wber forming the bottom of the chamber. Aliquots (10 l) of 1:100 cell dilutions from fresh cultures were loaded onto the chamber, and the Wber was centrifuged horizontally at 4000 rpm for 2 min. Measurements and data analysis The imaging Wber containing the single living cell array was mounted on an epiXuorescence microscope (model IX81, Olympus America, Melville, NY, USA). Fluorescence images were acquired from the proximal end of the Wber by a charge-coupled device (CCD)1 camera (Orca-ER, Hamamatsu, Japan). The cell array was Wrst exposed to fresh M9 medium supplemented with isopropyl--D-thiogalactoside (IPTG), mitomycin C (MMC), or both for a time period as speciWed in the text. The Xuorescence intensity was measured by IPlab software (Scanalytics, Fairfax, VA, USA). Fluorescence signals [480 nm excitation/520 nm emission for green Xuorescent protein (GFP) and 440 nm excitation/480 nm emission for enhanced cyan Xuorescent protein (ECFP)] from individual cells were measured with a 500-ms acquisition time immediately after exposing the cell array to M9 medium containing the compound(s). Fluorescence signals were measured every 5 min. All of the results were expressed as Xuorescence intensity percentage increases (It¡I0)/I0.
Results and discussion We previously demonstrated that individual living cells remain healthy and continue functioning after being inserted into microwells [15]. In the current work, we demonstrate the ability to simultaneously observe promiscuous drug eVects on single cells of multiple cell types in a single array format. The cell array is built on a 1-mm diameter optical Wber bundle that provides approximately 50,000 optically addressable microwells (2 £ 107 elements/cm2). A mixture of two E. coli cell types, MG1655+pUA2699(recA::gfp) [16] and MG1655 1 Abbreviations used: CCD, charge-coupled device; IPTG, isopropyl-D-thiogalactoside; MMC, mitomycin C; GFP, green Xuorescent protein; ECFP, enhanced cyan Xuorescent protein.
322
Monitoring “promiscuous” drug eVects / Y. Kuang, D.R. Walt / Anal. Biochem. 345 (2005) 320–325
+pECFP(lacZ::ecfp), both in the early exponential growth stage, were loaded onto an optical Wber well array. These two cell strains can report cell responses speciWcally to two diVerent drug stimuli based on two completely diVerent cellular systems. The Wrst cell strain produces GFP on DNA damage in response to exposure to generic genotoxins such as MMC. The second strain responds to nutrient starvation in which cellular glucose levels are low, and an alternative carbon source, lactose or its synthetic analogs (e.g., IPTG), is available. These cells are engineered to produce ECFP in response to switching the primary sugar metabolism from glucose to lactose. To observe multiple drug eVects on single cells of multiple cell types, a duplexed cell array was exposed to a deWned M9 medium supplemented with both 10 g/ml MMC and 5 mM IPTG. Fluorescence images (480 nm excitation/520 nm emission for GFP and 440 nm excita-
tion/480 nm emission for ECFP) were taken with 500-ms exposure times before drug exposure and every 5 min for 90 min after drug exposure. The locations of the two diVerent cell types in the array can be determined by observing their respective optical signatures. To account for the diVerences between the optical properties of GFP and ECFP, the Xuorescence signal from each cell was recorded as arbitrary units at any time t (It) but is reported as a percentage increase, that is, (It ¡ I0)/I0. Figs. 1A and B show individual cell response dynamics for both strains after exposure to the two compounds. Because both Xuorescent proteins have long half-lives [17], as expected, both strains exhibit steadily increasing Xuorescence levels over time in response to the two stimuli (MMC and IPTG). For each cell strain, although the cells are isogenic and exposed to the same stimuli, each cell exhibits its own individual response proWle. Such variation underscores how ensemble cell culture-based
Fig. 1. Monitoring individual cell responses from two bacterial cell types, MG1655+pECFP(lacZ::ecfp) and MG1655+pUA2699(recA::gfp) cells, to MMC and IPTG simultaneously on a single cell array format. The MG1655+pECFP(lacZ::ecfp) and MG1655+pUA2699(recA::gfp) cell responses were monitored by measuring the ECFP (A) and GFP (B), respectively. In a single exposure experiment, a cell array was exposed to medium supplemented with 5 mM IPTG and 10 g/ml MMC for 90 min. The insets show the Xuorescence level distribution across each cell population at 90 min. Error bars represent the standard deviation from three repeated assays.
Monitoring “promiscuous” drug eVects / Y. Kuang, D.R. Walt / Anal. Biochem. 345 (2005) 320–325
measurements can average out important information regarding cell-to-cell variability and may result in erroneous conclusions about the eVects of drugs on living systems. Initially, we expected dramatic diVerences in the Xuorescence signals between the two strains because there is a signiWcant diVerence in the plasmid copy numbers for the two strains; that is, the lacZ::ecfp fusion is located on a high-copy number plasmid (»500 copies), whereas the recA::gfp fusion is located on a low-copy number plasmid (»5 copies). For ease of comparison, the dynamics of the average responses of both cell strains are displayed in Figs. 2A and B, respectively. Even though MG1655+pECFP (lacZ:: ecfp) cells demonstrated a higher Xuorescence increase (0.42 § 0.13 at 90 min) as compared with MG1655+ pUA2699(recA::gfp) cells (0.17 § 0.04 at 90 min), the diVerence is not as dramatic as expected. For
323
MG1655+pECFP(lacZ::ecfp) cells, they respond not only to their own inducer IPTG but also to MMC. MMC is a strong genotoxin [18]. In our previous study, we showed that a concentration of 10 g/ml MMC is the threshold concentration for both genotoxicity and cytotoxicity [15]. We expected a possible cytotoxic eVect of MMC on MG1655+pECF P(lacZ::ecfp) cells. For MG1655+pUA2699(recA::gfp) cells, they respond not only to their own inducer MMC but also to IPTG. The activation of the recA gene is the Wrst step in activating bacterial SOS responses involved in DNA damage tolerance and error-prone replication. It has been reported that bacterial cells activate SOS responses under a variety of environmental stress conditions in addition to DNA damage [19–21], including high osmotic conditions [22]. In vitro studies have suggested that more than 1000 genes are involved in the DNA damage response in E. coli cells [23]. Therefore, even
Fig. 2. (A) Average responses of MG1655+pECFP(lacZ::ecfp) cells when incubated only with IPTG (solid circles) and when coincubated with IPTG and MMC (solid squares). (B) Average responses of MG1655+pUA2699(recA::gfp) cells when incubated with MMC and IPTG (solid squares), only with IPTG (solid circles), or with MMC after preincubation with IPTG for 115 min (solid diamonds). For ease of comparison, the IPTG-only response was moved up 0.14 U (dotted line) to compare its slope with that of the MMC after preincubation with IPTG for 115 min.
324
Monitoring “promiscuous” drug eVects / Y. Kuang, D.R. Walt / Anal. Biochem. 345 (2005) 320–325
though IPTG is neither a toxin nor the direct inducer for recA::gfp gene plasmid, MG1655+pUA2699 (recA::gfp) cells may respond to IPTG through the activation of the SOS response system. These results demonstrate the ability of the optical Wber array to simultaneously monitor multiple responses from two cell strains to two drugs. To test such possible promiscuous drug eVects, a second cell array was subjected to a sequential exposure experiment, Wrst to 5 mM IPTG for 115 min and then to 10 g/ml MMC for 65 min. Single cell response dynamics from both strains are displayed in Figs. 3A and B. Immediately after exposure to IPTG, a rapid increase in ECFP signals was obtained from MG1655+pECFP(lacZ:: ecfp) cells. The ECFP level continued to accumulate until 10 min after the medium was replaced with a medium containing MMC. Comparing the average responses (Fig. 2A) over the entire course (Wrst 90 min shown), the EGFP level is much higher in cells not coincubated with MMC than in cells coincubated with MMC, clearly indicating the toxic eVect of MMC on MG1655+pECFP(lacZ::ecfp) cells at a concentration of 10 g/ml. The termination of ECFP production in cells after the medium replacement was due to the toxic eVect of MMC on cells and the dilution of IPTG from cells given that IPTG is a small molecule that can freely diVuse through the cell membrane [24]. On the other hand, IPTG displayed an activation eVect on the GFP
level in MG1655+pUA2699(recA::gfp) cells, which show a low GFP signal increase when Wrst exposed to IPTG (Fig. 2B [dashed lines] and Fig. 3B), probably due to the response of the SOS gene networks to stress conditions caused by the very high concentration of IPTG (as we suspected). This initial low induction of GFP by IPTG allows a much faster cell response to MMC after the medium was replaced (Fig. 2B, solid diamonds), indicating that cell history (e.g., pretreatment with other drugs) has a signiWcant impact on the kinetic proWle of a speciWc drug to cells. These results underscore the synergistic eVects resulting from the simultaneous application of multiple active biomolecules. The genome-centric view of drug discovery is being replaced by a more comprehensive cell-centric view. The work reported in this article shows how optical microwell cell arrays can provide rich information in a single assay about the cross-activities of two “drugs” acting on two seemingly irrelevant cellular systems: a bacterial SOS response system and a sugar metabolic system. In conjunction with optical assays engineered to be expressed within living cells [25–27], such as other GFP variants with diVerent spectral properties, more diverse cell types can be accommodated on a single array. Such a capability will beneWt the drug screening process enormously by allowing for a more comprehensive understanding of how cells dynamically respond to combinatorial drug libraries or by examining how diVerent
Fig. 3. Monitoring individual cell responses from the two bacterial cell types sequentially exposed to 5 mM IPTG for 115 min and to 10 g/ml MMC for an additional 65 min. (A) Temporal proWle of individual MG1655+pECFP(lacZ::ecfp) cell responses. (B) Temporal proWle of individual MG1655+pUA2699(recA::gfp) cell responses. The change of medium at 115 min from IPTG to MMC caused a glitch in the Xuorescence intensity.
Monitoring “promiscuous” drug eVects / Y. Kuang, D.R. Walt / Anal. Biochem. 345 (2005) 320–325
cellular pathways and regulation circuits respond cooperatively to drugs in individual cells. This technique oVers a distinct advantage not only in high-content screening of combinatorial libraries of potential drug candidates but also in bioprocesses monitoring, environmental pollution monitoring, and diagnostic screening.
Acknowledgments We thank Uri Alon (Weizmann Institute of Science, Israel) for providing plasmid pUA2699 and Israel Biran for providing helpful discussions. References [1] A. Wagner, Robustness, evolvability, and neutrality, FEBS Lett. 579 (2005) 1772–1778. [2] J.J. Burbaum, Miniaturization technologies in HTS: how fast, how small, how soon? Drug Discov. Today 3 (1998) 313–322. [3] R. Kapur, K.A. Giuliano, M. Campana, T. Adams, K. Olson, D. Jung, M. Mrksich, C. Vasudevan, D.L. Taylor, Streamlining the drug discovery process by integrating miniaturization, high throughput screening, high content screening, and automation on the CellChip system, Biomed. Microdevices 2 (2000) 99–109. [4] M.P. Cary, G.D. Bader, C. Sander, Pathway information for systems biology, FEBS Lett. 579 (2005) 1815–1820. [5] S. Sen, M. Ferdig, QTL analysis for discovery of genes involved in drug responses, Curr. Drug Targets Infect. Disord. 4 (2004) 53–63. [6] M. Whitney, E. Rockenstein, G. Cantin, T. Knapp, G. Zlokarnik, P. Sanders, K. Durick, F.F. Craig, P.A. Negulescu, A genome-wide functional assay of signal transduction in living mammalian cells, Nat. Biotechnol. 16 (1998) 1329–1333. [7] T. Hampton, “Promiscuous” anticancer drugs that hit multiple targets may thwart resistance, J. Am. Med. Assoc. 292 (2004) 419–422. [8] V.C. Stephenson, R.A. Heyding, D.F. Weaver, The “promiscuous drug concept” with applications to Alzheimer’s disease, FEBS Lett. 579 (2005) 1338–1342. [9] M.B.H. Youdim, J.J. Buccafusco, Multi-functional drugs for various CNS targets in the treatment of neurodegenerative disorders, Trends Pharmacol. Sci. 26 (2005) 27–35. [10] K.L. Tucker, Y. Wang, J. Dausman, R. Jaenisch, A transgenic mouse strain expressing four drug-selectable marker genes, Nucleic Acids Res. 25 (1997) 3745–3746. [11] S.A. Sundberg, High-throughput and ultra-high-throughput screening: solution- and cell-based approaches, Curr. Opin. Biotechnol. 11 (2000) 47–53.
325
[12] M. May, G. Heebner, Laboratory technology trends: drug discovery (part 1), Science 295 (2002) 1349–1364. [13] I. Biran, D.R. Walt, Optical imaging Wber-based single live cell arrays: a high-density cell assay platform, Anal. Chem. 74 (2002) 3046–3054. [14] C. DiCesare, I. Biran, D.R. Walt, Individual cell migration analysis using Wber-optic bundles, Anal. Bioanal. Chem. 382 (2005) 37–43. [15] Y. Kuang, I. Biran, D.R. Walt, Living bacterial cell array for genotoxin monitoring, Anal. Chem. 76 (2004) 2902–2909. [16] M. Ronen, R. Rosenberg, B.I. Shraiman, U. Alon, Assigning numbers to the arrows: parameterizing a gene regulation network by using accurate expression kinetics, Proc. Natl. Acad. Sci. USA 99 (2002) 10555–10560. [17] R. Tombolini, A. Unge, M.E. Davey, F.J. de Bruijn, J.K. Jansson, Flow cytometric and microscopic analysis of GFP-tagged Pseudomonas Xuorescens bacteria, FEMS Microbiol. Ecol. 22 (1997) 17–28. [18] L.R. Ptitsyn, G. Horneck, O. Komova, S. Kozubek, E.A. Krasavin, M. Bonev, P. Rettberg, A biosensor for environmental genotoxin screening based on an SOS lux assay in recombinant Escherichia coli cells, Appl. Environ. Microbiol. 63 (1997) 4377–4384. [19] A. Aris, J.L. Corchero, A. Benito, X. Carbonell, E. Viaplana, A. Villaverde, The expression of recombinant genes from bacteriophage lambda strong promoters triggers the SOS response in Escherichia coli, Biotechnol. Bioeng. 60 (1998) 551–559. [20] C. Miller, L.E. Thomsen, C. Gaggero, R. Mosseri, H. Ingmer, S.N. Cohen, SOS response induction by -lactams and bacterial defense against antibiotic lethality, Science 305 (2004) 1629–1632. [21] J. Lee, H.C. Kim, S.W. Kim, S.I. Hong, Y.H. Park, Interplay of SOS induction, recombinant gene expression, and multimerization of plasmid vectors in Escherichia coli, Biotechnol. Bioeng. 80 (2002) 84–92. [22] S.M. Ruzal, C. Sanchez-Rivas, Physiological and genetic characterization of the osmotic stress response in Bacillus subtilis, Can. J. Microbiol. 40 (1994) 140–144. [23] P. Khil Pavel, R.D. Camerini-Otero, Over 1000 genes are involved in the DNA damage response of Escherichia coli, Mol. Microbiol. 44 (2002) 89–105. [24] A. Khlebnikov, J.D. Keasling, EVect of lacY expression on homogeneity of induction from the Ptac and Ptrc promoters by natural and synthetic inducers, Biotechnol. Prog. 18 (2002) 672–674. [25] J.E. Gonzalez, P.A. Negulescu, Intracellular detection assays for high-throughput screening, Curr. Opin. Biotechnol. 9 (1998) 624– 631. [26] D.L. Taylor, E.S. Woo, K.A. Giuliano, Real-time molecular and cellular analysis: the new frontier of drug discovery, Curr. Opin. Biotechnol. 12 (2001) 75–81. [27] J. Lippincott-Schwartz, G.H. Patterson, Development and use of Xuorescent protein markers in living cells, Science 300 (2003) 87–91.