Bioluminescent bioreporter integrated circuits: potentially small, rugged and inexpensive whole-cell biosensors for remote environmental monitoring.

Abstract

Summary, 33 Introduction, 33 Bioluminescent bioreporter technology, 34 Bioreporters, 34 Bioluminescence, 35 luxCDABE bioreporters, 36 Detecting bioluminescence, 39 Developing integrated circuits for bioluminescent bioreporter detection, 39 Interfacing bioluminescent bioreporters with integrated circuit detection, 40 Initial testing of a bioluminescent bioreporter with integrated circuit detection, 40 Off-chip wireless BBIC, 41 Future considerations, 42 Conclusions, 43 Acknowledgements, 43 References 43 A bioluminescent bioreporter integrated circuit (BBIC) is a novel whole-cell biosensor that combines the environmental monitoring capabilities of genetically engineered bioluminescent micro-organisms (bioreporters) with optical application-specific integrated circuits. A BBIC device consists of bioreporters sustained within a micro-environment, the integrated circuit microluminometer, and a light-tight enclosure. The bioreporter typically contains the luxCDABE reporter genes encoding the enzymes required for bioluminescence. In the presence of a targeted analyte, a gene (transcriptional) regulatory system induces the expression of the luxCDABE genes. Analytical benchmark data for exposure of the bioreporter Pseudomonas fluorescens 5RL to salicylate was determined using a flow-through test system. The detection limit (after a 45 min exposure) was ca 50 μg l−1 and response times decreased from ca 45 to 20 min as the concentration increased from 50 μg l−1 to 1 mg l−1. These results are currently being used to scrutinize enclosures and micro-environment configurations to develop a field deployable BBIC. A successful BBIC could provide a simple and inexpensive means of creating a ‘laboratory-on-a-chip’ and could be used in a network to protect valuable human and environmental resources. This article reviews the present state of luxCDABE-based bioreporter research, demonstrates the integration of the bioreporters with complementary metal oxide semiconductor photodiode-integrated circuits, and discusses future challenges for real-time in situ BBIC environmental monitoring. Safeguarding human and environmental resources against harmful agents requires the development of new in situ, real-time monitoring devices that can be easily deployed at multiple strategic locations. Ideally, a device would be able to detect a pollutant, toxic chemical, or warfare agent rapidly, at relevant concentrations, and in a cost-effective manner. Many conventional analytical technologies (e.g. spectrometry) with enough selectivity to identify trace amounts of specific analytes are too expensive to be deployed in a multi-sensor network and/or cannot be used in real-time monitoring because they require sample preparation steps prior to detection. Biosensor technology is an analytical option that can circumvent these problems by providing rapid, inexpensive, sensitive, and in some cases selective analysis of environmental samples (Collings and Caruso 1997; Matrubutham and Sayler 1998; Nistor and Emneus 1999; D'Souza 2001; Keane et al. 2002; Marazuela and Moreno-Bondi 2002). A biosensor is an analytical device that uses biological macromolecule(s) to recognize an analyte and subsequently activate a signal that is detected with a transducer. The transducer typically converts the biological response into an electrical signal. The biological recognition of the analyte affords biosensors excellent sensitivity (typically μg l−1 to mg l−1) and selectivity for the detection of an individual compound, a group of similar molecules (typically based on similar functionality and molecular shape) or a general toxic effect. Biological recognition elements are typically enzymes, antibodies, DNA, receptors, or regulator proteins and can be used either in whole-cells or isolated and used independently (D'Souza 2001). Biosensors using purified biomolecules can have rapid response times, but purification procedures can be time-consuming and expensive (Daunert et al. 2000). Permeabilized nonviable cells have been used as an inexpensive alternative to provide biological recognition elements (D'Souza 2001). However, both purified biological recognition elements and permeablized whole-cells mostly have been applied in laboratory settings or in point-of-use determinations and are probably less useful for long-term monitoring of environmental systems because of the degradation of the recognition element with time. Exploiting living micro-organisms in whole-cell biosensors (sometimes referred to as microbial biosensors) is beneficial as micro-organisms are continually synthesizing complex systems of biomolecules (including biological recognition elements) required for sensing and reacting to environmental changes. Micro-organisms are also ubiquitous and it is highly probable that a strain can be acquired to survive within the operating parameters of a given environment. Micro-organisms contain or can be genetically engineered to contain reporter gene(s) that expresses reporter protein(s) responsible for detectable signals. Furthermore, these reporter micro-organisms (referred to herein as bioreporters) can support multi-enzyme reactions, and allow more complicated biological recognition and signalling schemes for biosensors. Cofactors and cosubstrates for enzyme reactions can also be provided by bioreporters, thereby eliminating the need for adding exogenous reagents to produce signals. Bioreporters also have the ability to determine the bioavailability of an analyte instead of the total concentration, which can include chemical species of the analyte that are not physiologically relevant (e.g. insoluble) (Heitzer et al. 1992; Alexander 2000; Rasmussen et al. 2000). Limitations of the use of bioreporters include the facts that living bioreporters require nutrients to survive and can require minutes to produce a detectable signal because of the time needed for mass transfer processes and in some cases, the time required for the expression of the protein(s) responsible for the biological signal. A bioluminescent bioreporter integrated circuit (BBIC) is a unique whole-cell biosensor that uses an integrated circuit optical transducer to detect blue-green light (ca 490 nm) emitted by a genetically engineered bioreporter in response to a specific stimulus (Simpson et al. 1998; Simpson et al. 2001; Bolton et al. 2002). Conceptually, the BBIC device consists of a light-tight enclosure, a bioluminescent bioreporter strain, a micro-environment that sustains the living bioreporter cells, and the integrated circuit for light detection and signal processing (Fig. 1). During the detection process, a targeted analyte diffuses to the bioreporters within the micro-environment and interacts with the gene regulator protein of the bioreporter. This concentration-dependent interaction induces the expression of the luxCDABE genes that code for the proteins that produce light. The bioluminescence is then detected with an integrated circuit microluminometer and the signal relayed to a central database via wireless communication for potential mitigation. Conceptual diagram of the key components that make up a bioluminescent bioreporter integrated circuit (BBIC). The light-tight enclosure is used to block environmental light, provide diffusion-limited chemical contact with the outside environment, house the micro-environment and seal the electronics. The micro-environment supplies nutrients for the bioreporters, provides protection for the cells and in some cases may contain an encapsulation matrix. The integrated circuit detects the bioluminescence emitted by the bioreporters, processes the signal, and communicates the results to a central database. An application specific BBIC will also contain a battery that is not shown This article discusses and reviews the development of the bioluminescent bioreporter and light-detecting integrated circuit technologies, and the problems and current limitations associated with interfacing the bioreporter with microluminometers. The ultimate goal of the BBIC technology is the creation of a family of ‘laboratory on a chip’ biosensors for remote-sensing environmental applications. A bioreporter (sometimes referred to as a reporter cell) is a living organism that produces a measurable signal when it senses a particular chemical or a physical change in its environment. A bioreporter typically contains a gene (transcriptional)-regulatory system coupled to a reporter gene(s) that encodes for a reporter protein(s) responsible for the signal. These reporter and/or regulator genes can be native to the cell or acquired by genetic transformation. A common type of transcriptional regulatory system consists of a regulator gene(s), its expressed regulator protein(s) and the promoter/operator DNA sequence that controls the expression of downstream gene(s) (Fig. 2). These elements work in concert to sense environmental changes and control the expression of a gene or an operon to enhance the physiology of the cell. For example, the LysR-type transcriptional regulatory system uses positive control of induction to activate gene expression (Fig. 2). With this system, an analyte acting as the inducer forms a complex with the regulator protein. This complex interacts with the promoter to initiate the binding of RNA polymerase and the transcription of downstream genes (Fig. 2). By fusing a native promoter to the reporter gene(s) and inserting the fusion into a host cell, the resultant bioreporter is engineered to harness the native transcriptional regulatory system for the purpose of signalling that the native protein is being expressed and/or that the targeted environmental change has occurred. Conceptual diagram of a bioreporter with a LysR-type transcriptional regulatory system showing the steps involved in gene expression and signalling. The bioreporter cell contains a positively controlled transcriptional regulator system that induces both the reporter gene, and in this case, the genes encoding the protein that catabolize the analyte. A is the analyte, R designates the regulator protein, D is the degradative enzyme, and S is the signalling protein. The regulator gene is constitutively expressed and not shown. The detection process begins when the analyte, A, enters into the cytoplasm of the cell and forms a complex with the regulator protein, R (step 1). The complexes interact with the promoters to induce the expression of the genes (step 2). The genes are transcribed to produce mRNAs (step 3) and the mRNAs are translated to produce signalling proteins and enzymes to degrade the analyte (step 4). The signalling proteins can be directly detected or catalyze reactions to produce the detected signal (step 5). Pseudomonas fluorescens HK44 responds to salicylate in this manner. The bioreporter Ps. fluorescens 5RL contains a translational insertion of the reporter genes into the degradation gene, rendering the degradative pathway inactive and allowing the analyte to accumulate within the cell. This type of bioreporter is useful for an early-warning BBIC Several reporter genes have been isolated from a variety of naturally occurring organisms (Daunert et al. 2000) and have been categorized by the means of detection (Kohler et al. 2000; Keane et al. 2002). Reporter genes coding for proteins that produce colour changes, fluorescent molecules, electro-active species, or bioluminescence have been discovered and can be detected by conventional analytical devices. Colorimetric, fluorescent and electrochemical monitoring can require significant amounts of power, a limited resource in remotely deployed sensors. For example, colorimetric and fluorescent devices require power-draining sources of radiation (e.g. lasers, lamps or light emitting diodes) to excite the analyte for detection. Furthermore, environmental samples contain many compounds that can interfere with electrochemical, colorimetric and fluorescent detection. Some colorimetric and fluorescent signalling products also can remain active for long periods of time (even persisting after cell death), and thus are not conducive to dynamic real-time environmental monitoring. For example, the native green fluorescence protein (gfp) persists until cell lysis (Tombolini et al. 1997); however, variants have been made that have half-lives of 40 min to a few hours (Andersen et al. 1998). In contrast, detection of bioluminescence requires no external source of radiation, as light is produced by biochemical reactions. This simple analytical strategy also minimizes the background noise and requires no wavelength discrimination. As bioluminescence requires energy in the form of high-energy biomolecules, analytes that induce the lux system and provide energy can be used for dynamic monitoring applications. In a study that compared the responses of fluorescent (gfp and dsred proteins) and bioluminescent (luxCDABE and luc FF genes) Escherichia coli plasmid-based bioreporters, bioluminescence had faster response times and lower detection limits than the fluorescence signals (Hakkila et al. 2002). These substrates are typically not referred to as luciferins, as they are common cell metabolites or are derived from ordinary metabolites after minimal modification. The luxG gene is also found in luminescent species of bacteria and may play a role as a flavin reductase in the recycling of the FMNH2 (Zenno and Saigo 1994). The three most common variants of the luxAB gene isolated from Vibrio fischeri, V. harveyi, and Photorhabdus luminescens appear to have usable thermal stability to 30, 37 and 42°C, respectively (Meighen 1991), with P. luminescens having a half-life of 3 h at 45°C (Meighen 1991). Furthermore, the entire luxCDABE operons for V. harveyi and P. luminescens have strong signals when expressed in E. coli at 37°C (Szittner and Meighen 1990). The luc, ruc and lux genes have been used to create bioluminescent bioreporter strains (Daunert et al. 2000). The use of the eukaryotic reporter genes in foreign hosts typically requires the exogenous addition of the luciferin to generate light. A similar approach is sometimes used in bacterial systems using the lux system; the luxAB genes are expressed and an aldehyde is added exogenously. Use of the luxCDABE (the luxG is not included in most constructs) gene cluster offers advantages because the expressed proteins produce bioluminescence without the addition of exogenous substrates. Bioreporters using the luxCDABE cluster typically can be (i) naturally occurring environmental isolates, (ii) genetic constructs with constitutive promoters (always expressed), (iii) constructs with promoter-lux fusions for specific stress-related responses, (iv) constructs with gene regulatory systems that are specific for certain analytes or classes of analytes and (v) multiple constructs with different promoters functioning in a genomic-wide array. Naturally occurring bioluminescent bioreporter strains are used to assess the toxicity of a sample to cells (Bulich 1982; Kaiser and Palabrica 1991). Cultures of bioluminescent bacteria are mixed with various concentrations of a sample containing a potential toxic reagent and the bioluminescence is determined. In the presence of a toxin, the bioluminescence is attenuated either by loss of metabolic activity or cell death. By comparing the bioluminescence with a control sample that was not exposed to the toxin, the magnitude of the decrease is calculated and used to quantify toxicity. A number of different commercially available systems exist that provide the necessary reagents, freeze- or liquid-dried bioreporters and instrumentation to observe bioluminescence (Jennings et al. 2001). These systems are used to determine the toxicity of pure compounds and complex samples like wastewater effluents. Although a signal decrease is attributed to toxicity and provides evidence that a sample contains a toxic agent, the true source of a diminished signal is difficult to ascertain, as a number of mechanisms other than toxicity (including lack of energy or oxygen) can attenuate bioluminescence (Cronin and Schultz 1998). Bacterial strains other than naturally occurring hosts have been genetically engineered to act as bioluminescent bioreporters for the sensing of toxicity caused by pollutants or environmental stressors. For example, Kelly et al. (1999) constructed a bioluminescent bioreporter by insertion of the constitutively expressed luxCDABE genes (V. fischeri) into a Pseudomonas fluorescens strain isolated from a wastewater treatment facility. The bioluminescent bioreporter (designated Shk1) was used to estimate the toxicity of wastewater, the environment from which it was isolated. This overcomes a drawback of commercial assays using the marine micro-organism V. fischeri, which requires the addition of high salt medium to freshwater samples prior to testing (Lajoie et al. 2002). It was suggested that salt can alter the chemistry and thus the toxicity of a freshwater sample. Promoters for stress response genes have been fused to the luxCDABE cassette to produce bioreporters that respond to a number of different types of environmental and chemical stresses (Table 1). These types of bioreporters are not used to identify a particular stressor, but simply that the stress response has been activated. For example, heat shock promoters in bioreporters have been used to respond not just to conditions involving a change in temperature, but also to detect exposure to chemical and biochemical stressors such as antibiotics, organic compounds, oxidative reagents and heavy metals (Kregel 2002). For example, Belkin et al. (1997) constructed five plasmids containing different promoter–luxCDABE fusions and used the plasmids to transform an E. coli strain into a panel of bioluminescent bioreporters for monitoring stress responses. Promoters for katG and soxRS for oxidative stresses, recA for DNA damage, grpE for the heat shock/protein damage system and fabA involved in fatty-acid synthesis inhibition were used. The bioreporters reacted specifically to stressors within their respective groups, except for phenol, which nonspecifically induced all of the strains. Davidov et al. (2000) offered potential improvements in the SOS-type recA promoter-luxCDABE bioreporter response by constructing strains that (i) deactivated the expression of an efflux pump (tolC mutation) to improve sensitivity, (ii) provided a more stable construct by incorporating the promoter–lux fusion into the chromosome and (iii) used the luxCDABE genes from P. luminescens, allowing use at higher temperatures. A plasmid containing an SOS dependent promoter–luxCDABFE fusion (the lux genes were isolated from Photobacterium leiognathi) has been introduced into the Salmonella typhimurium TA1535 that was used in the Ames test for mutagenesis (Rabbow et al. 2002). The system gave comparable results to the MutatoxTM (Strategic Diagnostics Inc., Newark, NJ, USA), Ames and SOS-chemotests assays. Bioreporters containing a promoter–luxCDABE gene fusion have been used selectively to monitor for the presence of inorganic and organic analytes (Table 1). Genetic strategies to engineer this type of bioreporter involve (i) isolating micro-organisms that can express proteins that selectively degrade, bind, sequester, transport or transform the targeted analyte, (ii) determining the promoter sequence that regulates the selective function, (iii) fusing the promoter sequence to the luxCDABE cassette and (iv) inserting the promoter–luxCDABE fusion into a host strain containing the gene regulatory system. For example, Corbisier et al. (1999) developed a plasmid-based Alcaligenes eutrophus bioreporter containing the promoter for the pbr operon (efflux pump) fused to the luxCDABE genes from V. fischeri. The bioreporter was able to selectively detect lead, but not other potential interfering metals such as cadmium, copper, zinc and mercury. Analogous bioreporters using the luxCDABE genes have been developed to sense cadmium, chromate, cobalt, copper, iron, mercury, nickel and zinc (Table 1) (Rensing et al. 1999; Ramanathan et al. 1997). Bioreporters also have been developed to sense specific organic pollutants using promoter–luxCDABE fusions. Generally, the promoter that activates the genes coding for catabolism of the analyte is used to construct the bioreporter. A pioneering example of this type of bioreporter is Ps. fluorescens HK44 that was developed to monitor the biodegradation of napthalene (King et al. 1990). Genes that code for the enzymes catabolizing naphthalene are contained on a large plasmid, pUTK21, in two separate operons coding for the upper (nah) and lower (sal) pathways, respectively. The upper pathway converts naphthalene to salicylate and the lower pathway oxidizes salicylate. The nahR gene, located between the two operons, constitutively expresses the NahR protein (in the LysR family of transcriptional regulators) that diffuses into the cytoplasm and positively regulates both operons. Salicylate, the degradation product of naphthalene, forms a complex with the NahR protein and binds to the promoter, resulting in the induced expression of both sets of biodegradation genes (Schell 1985). The bioreporter Ps. fluorescens HK44 contains the luxCDABE genes from V. fischeri that are also under the control of the nahRGp promoter and are expressed in the presence of salicylate (e.g. Fig. 2). Bioluminescence of Ps. fluorescens HK44 in heavily contaminated soil slurries from a manufactured gas plant site was detected using light guides coupled to a photomultiplier tube (PMT) system (King et al. 1990). The strain was also used in field lysimeters to study the environmental monitoring capabilities of bioreporters and the environmental fate of genetically modified micro-organisms (Ripp et al. 2000). Bioreporters for other organic molecules have been developed and are listed in Table 1. In a large-scale and innovative effort, Van Dyk et al. (1998, 2001a,b) constructed 689 bioreporters with different promoter–luxCDABE gene fusions contained on plasmids creating a living array to study the genome-wide transcriptional responses of E. coli. Specifically, E. coli (DPD1675) was transformed with the plasmids containing different random E. coli DNA fragments that were ligated upstream of the promoterless luxCDABE genes. Successful bioreporter strains were selected from the transformants using sequencing data to ascertain the identity of unique promoters with the proper orientation. The array of bioreporters contained promoters for several global stress response regulons and covered ca 27% of the predicted transcriptional units of E. coli (Van Dyk 2002). The genome-wide promoter–luxCDABE gene fusions can be arrayed (e.g. microtiter plates) to demonstrate the overall transcriptional response of E. coli to physical and chemical perturbations and provided information similar to DNA arrays without the use of problematic nucleic acid isolation procedures (Van Dyk et al. 2001b). Bioluminescence of bioreporters has been detected with photodiodes, avalanche photodiodes and photomultiplier tubes and imaged with photographic film silicon intensified target cameras and charge-coupled device cameras, (Simpson et al. 1998; Contag et al. 2000; Greer and Szalay 2002). As with any other optical signal, the bioluminescence can be transferred to the detectors with lenses, fibre optics (Ripp et al. 2000; Marazuela and Moreno-Bondi 2002), liquid light guides or monitored directly by a detector. For example, a photon counting photomultiplier probe was inserted into the subsurface of lysimeters through 4′′ diameter bore holes to directly monitor the bioluminescence of Ps. fluorescens HK44 in polyaromatic hydrocarbon-contaminated soils (Ripp et al. 2000; Sayler et al. 2001). This detection system was portable, though it required an external power supply and an RS232 cable for data communication. The ideal detector for in situ real-time monitoring of remote environmental locations would be small and rugged, have good sensitivity, be inexpensive, and have both a global positioning system and wireless communication circuitry. Furthermore, the system should consume minimal power, as the size and mass of the battery has a great effect on the size and mass of the overall biosensor package. Complementary metal oxide semiconductor (CMOS) technology has the ability to make electronic devices that are small, inexpensive and have low power consumption. Using CMOS, a single integrated circuit can be manufactured to contain tens of millions of transistors. Thus, a single chip can integrate many different standard electronic functions. For example, integrated circuits containing photodiodes could be equipped with signal processing, radio frequency (RF) wireless telemetry and global positioning circuits (Simpson et al. 1998). Ancillary functions such as temperature sensing could also be incorporated onto the sensor. Thus, integrated circuits realized with CMOS are excellent platforms for the detection of bioluminescence from bioreporters. Initially, different types of photodiodes (p-diffusion/n-well and n-well/substrate photodiodes) and signal processing circuits were analysed with test chips to determine the optimal configuration for sensing low levels of bioluminescence (Simpson et al. 1998; Simpson et al. 2001). The n-well/substrate photodiode was determined to have greater quantum efficiency (66% at 490 nm) than the p-diffusion/n-well type and was selected for further development. Processing of the resultant photodiode signal was accomplished using a current-to-frequency converter circuit. Briefly, electrons from the photodiode are collected with a small on-chip capacitor. A gated integrator at the input allows the collection of the photocurrent (and dark current) until the output voltage reaches a threshold value. At the threshold voltage, a pulse is generated that resets the input gated integrator by closing a switch across the capacitor. The switch is then opened and the process continues to produce pulses that may be individually timed to obtain pulse interval measurements that are inversely proportional to current. Alternatively, the pulses can be counted for a fixed period of time to provide frequency data that is proportional to the current. Using the information gleaned from the test chips, a large area (1·47 mm2) photodiode integrated circuit was produced using the 0·5 μm bulk CMOS process. The chip demonstrated large dynamic ranges and was able to monitor bioluminescence produced by bioreporters (Simpson et al. 2001). Unlike PMT-based systems, the BBIC integrated circuit microluminometer can be exposed to room light without damage and can be exposed to alternating high and low light levels without a significant memory effect. A second version of the large area BBIC chip was produced to reduce the leakage current (Bolton et al. 2002). Using the n-well/substrate type photodiode and the optimized current-to-frequency converter, this version of the BBIC chip demonstrated a minimum detectable signal of 0·15 fA with a 25 min integration time at room temperature (Bolton et al. 2002). This minimum detectable signal corresponds to the bioluminescence emitted from ca 5000 fully induced Ps. fluorescens 5RL cells (see Section 5.1). Initially, silicon chips were wire bonded to standard 40-pin ceramic chip carriers for testing. To protect the wire bonds, the chips were encapsulated in silicone elastomer, polydimethyl siloxane (PDMS – Sylgard 184; Dow Corning, Midland, MI, USA). Enclosures have been and are being designed to (i) mount to the chip carriers, (ii) block room light, (iii) allow diffusion-limited contact with the outside environment, (iv) protect the electronics and/or (v) provide containment of the bioreporters. The ceramic carriers and enclosures were mounted onto sockets attached to a printed circuit test board. A power supply was connected to the test board to provide 3·3 V power and off-chip adjustable bias voltages. The output from the integrated circuit was transformed to a pulse period, converted to a value proportional to bioluminescence recorded and displayed with time using a personal computer, a digital interface board and software. Pseudomonas fluorescens 5RL was selected for initial analytical testing. This construct contains a salicylate-inducible regulatory system and the luxCDABE operon as a transpositional mutation of the sal operon (King et al. 1990). The lux genes were inserted into the operon that coded for the degradation of salicylate, thus preventing the expression of the degradation genes and allowing the intracellular accumulation of salicylate. The accumulation of the inducer molecule should provide a more rapid and sustained response and this genetic strategy may prove to be useful for the development of early warning type biosensors. Salicylate, an important plant biomolecule, is also a useful test analyte because it is water-soluble, nonvolatile, nontoxic and is not degraded by Ps. fluorescens 5RL. Initially, flow-through enclosures were developed to monitor real-time bioluminescence from cells in aerated liquid cultures. Cells in known physiological states were pumped through the flow-cell for on-line determinations of bioluminescence, while at the same time samples were removed for off-line analysis. These studies were designed to produce analytical benchmark data for testing other types of enclosures. The flow-through enclosures house glass tubes with a 9-mm outside diameter. Cultures were prepared by inoculating Ps. fluorescens 5RL cells into 1 l of sterile minimal salts medium supplemented with trace elements containing glucose as a carbon source, and rotating constantly at 170 rev min−1. All experiments were performed at 20·5 ± 0·5°C as temperature affects the dark signals and the growth of the bioreporters. When the cultures reached an O.D.546 of 0·02 they were pumped