Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Methods Enzymol. Author manuscript; available in PMC 2011 March 9.
Published in final edited form as:
PMCID: PMC3052260

Using Two-Component Systems and other Bacterial Regulatory Factors for the Fabrication of Synthetic Genetic Devices


Synthetic biology is an emerging field in which the procedures and methods of engineering are extended to living organisms, with the long-term goal of producing novel cell types that aid human society. For example, engineered cell types may sense a particular environment and express gene products that serve as an indicator of that environment, or effect a change in that environment. While we are still some way from producing cells with significant practical applications, the immediate goals of synthetic biology are to develop a quantitative understanding of genetic circuitry and its interactions with the environment and to develop modular genetic circuitry derived from standard, interoperable, parts, that can be introduced into cells and results in some desired input/output function. Using an engineering approach, the input/output function of each modular element is characterized independently, providing a toolkit of elements that can be linked in different ways to provide various circuit topologies. The principle of modularity, yet largely unproven for biological systems, suggests that modules will function appropriately based on their design characteristics when combined into larger synthetic genetic devices. This modularity concept is similar to that used to develop large computer programs, where inpendent software modules can be independently developed and later combined into the final program.

Here, we will start by pointing out the potential usefulness of two-component signal transduction systems for synthetic biology applications, and will describe our use of the E. coli NRI-NRII (NtrC-NtrB) two-component system for the construction of a synthetic genetic oscillator and toggle switch for E. coli. Procedures for conducting measurements of oscillatory behavior and toggle switch behavior of these synthetic genetic devices will be described. Next, we present a brief overview of device fabrication strategy and tactics, and will present a useful vector system for the construction of synthetic genetic modules and positioning these modules onto the bacterial chromosome in defined locations.

Using two-component signal transduction systems in synthetic biology approaches

Two-component signalling systems have been studied intensively for about 20 years, and provide numerous examples of systems where the cellular physiology of the regulatory phenomena and the activities of the signal transduction components are reasonably well understood (Hoch and Silhavy, 1995). Certain aspects of these signal transduction systems make them particularly useful for synthetic biology purposes. Foremost among these is that many two-component systems are not essential for viability under most growth conditions, and instead control fairly small numbers of genes, that are only required under some stress condition that need not be applied. Thus, “closed” systems can be envisioned, where all of the natural stress response genes and regulated promoters are deleted and the two-component system functions solely in the synthetic genetic device.

The mechanisms of two-component signal transduction are also useful from a synthetic biology perspective. These systems contain a transcriptional activator whose activity is controlled by reversible phosphorylation. This feature allows the activation activity to be “tuned” by experimental manipulation of the transmitter protein activities leading to receiver phosphorylation and dephosphorylation. This can be obtained, for example, by manipulating growth conditions, by using mutant forms of the transmitter protein, or by using mutant forms of the receiver protein (transcription factor) that have different levels of constitutive activity. A variety of transmitter and receiver protein mutants with fixed output activities are available for some two-component regulatory systems, and for the others it should be possible to introduce mutations based on available data from these characterized systems.

The sensory and signal-transduction properties of certain transmitter proteins, in particular EnvZ, are distinct functions, allowing facile re-engineering for development of new sensors. It has long been known that the transmitter module of EnvZ, when fused to the transmembrane portions of cellular chemotaxis receptors, allows OmpR-dependant gene transcription in response to the ligands of the chemotaxis receptors (Jin and Inouye, 1993). The generality of this phenomenon has been extended by linking the EnvZ transmitter module to a light-sensing receptor (Levskava et al, 2005), and may be extended further by using altered chemosensory apparatus to widen the range of environmental stimuli that can control expression from OmpR-dependent promoters. Presumably, all of the many transmitter proteins that are structurally related to EnvZ could be re-engineered for altered sensory activity using the same approaches, allowing coupling of a variety of stimuli to various receiver proteins.

The NRI/NRII (NtrC/NtrB) two-component system controlling nitrogen assimilation in many bacteria has a number of particularly appealing features. First, the transcriptional activator, NRI~P, acts from an enhancer that is relatively position independent (Ninfa et al, 1987). The main positioning limitation is that the enhancer must be a minimal distance of about 70 base pairs from the site of polymerase binding (promoter). This position independence of the enhancer is very useful in that it simplifies the construction of promoters that are combinatorially regulated by different transcription factors. Furthermore, the concentration of activator required to drive gene expression depends largely on the strength of the enhancer sequences, allowing experimental manipulation of promoter activation by using natural strong or weak enhancers, or by mutation of a natural enhancer (Feng et al, 1995). Thus, promoters in a synthetic genetic device can be activated sequentially, just as the natural Ntr promoters are sequentially activated by increasing NRI~P during the cellular response to nitrogen limitation (Atkinson et al, 2002). A second very useful property of NRI~P is that the activating species is either a hexamer or heptamer of NRI subunits, thus providing a very high kinetic order of transcriptional activation (Lee et al, 2003). A third useful property is that certain NRI~P dependent promoters are completely silent in the absence of activator. Specifically, the glnK promoter of E. coli exhibits essentially no basal expression in the absence of NRI~P, whereas it is a very strong promoter in the presence of a high concentration of NRI~P (Atkinson et al, 2002). Certain other Ntr promoters, such as the glnA promoter have a powerful and tight Ntr promoter (glnAp2) coupled with a weak non-Ntr promoter (glnAp1) that provides a basal level of expression in the absence of NRI~P, and is repressed by NRI~P (Reitzer and Magasanik, 1985). Thus, one has the choice of very tight or somewhat leaky transcription, and there is no reason to suspect that this property cannot be “tuned” by mutation to provide a desired ratio of basal/activated expression. A wide variety of mutant forms of the transmitter protein are available, that differ in their ability to bring about the phosphorylation and dephosphorylation of NRI and thus allow “tuning” of the phosphorylation state in the context of a synthetic genetic device (Atkinson and Ninfa, 1992; Atkinson and Ninfa, 1993, Pioszak and Ninfa, 2003). Finally, NRI-dependent promoters utilize RNA polymerase containing the minor sigma factor σ54, and are completely silent in the absence of this sigma factor, which is not essential for viability (Hunt and Magasanik, 1985). This opens the possibility for even more elaborate devices in the future, that use independent signalling pathways to control the presence of NRI~P and σ54, resulting in AND gate function from an Ntr promoter.

Using the NRI-NRII system to build a synthetic genetic clock

The basic circuit topology for the synthetic genetic clock is shown in Fig 1. The clock consists of two modules: the activator module and the repressor module. The activator module (left) consists of a promoter that drives the expression of activator, and is itself activated by activator. The activator also drives the expression of the repressor module (right), which produces repressor. The repressor protein blocks the expression of the activator module. Modeling of this circuit indicated that it had the potential to produce a variety of oscillatory outputs, ranging from sinusoidal oscillators to relaxation oscillators, as well as non-oscillatory steady states, depending on parameters (Atkinson et al, 2003, and unpublished data). Interestingly, modeling also indicated that for each set of parameters where the intact device produced oscillations, the isolated activator module would function as a toggle switch that displayed strong hysteresis of repression (Atkinson et al, 2003). The basic design of this clock is related to an earlier hypothetical clock (Barkai and Leibler, 2000), except that in the clock of Barkai and Leibler, the repressor protein antagonises the activity of the activator, as opposed to repressing activator expression.

Figure 1
Basic circuit topology of the synthetic genetic clock

The implementation of the design is depicted in Fig 2 (Atkinson et al, 2003). The activator module (left) consists of the E. coli glnA promoter region, driving the expression of NRI. The natural glnA promoter contains a strong enhancer (light blue boxes) and three low affinity activator binding sites that function as a band-limiter or governor (grey boxes, [Atkinson et al, 2002]). The region also contains two promoters (depicted as bent arrows in Fig 2): the upstream glnAp1 promoter is repressed by activator binding to the enhancer, and the downstream glnAp2 utilizes polymerase containing σ54 and is activated by NRI~P. To connect this module to the repressor module, “perfect” lac operators (dark blue boxes [Atkinson et al, 2003]), were added at two positions: immediately downstream from the glnAp2 transcriptional start site, imitating the position of the lacO1 operator in the lacZYA operon, and immediately upstream from position −162. The intention of this design was to permit Lac Repressor to form a DNA loop and silmutaneously contact both operators, as it does in repressing the lacZYA operon (Oehler et al, 1990). However, subsequent results have suggested that a repression DNA loop does not form, although the upstream operator is still required for good oscillatory function (unpublished data). The repressor module consists of the glnK promoter region and mRNA leader sequence, fused to the structural gene for Lac repressor, lacI. To facilitate the fusion, the initiation codon for lacI was converted to AUG. The glnK promoter has essentially no basal expression in the absence of activator, and requires a high concentration of activator for expression owing to its weak enhancer. The use of the Lac Repressor as the clock repressor permits easy synchronization of the cells with the lactose analogue IPTG. To bring about phosphorylation of NRI, the strain containing activator and repressor modules was transformed with the pBR322-derived p3Y15, which contains the mutant glnL2302 allele for the transmitter protein NRII, driven from its natural promoter. The product of the glnL2302 allele, NRII2302, is defective in bringing about the dephosphorylation of NRI~P, but is fully functional in autophosphorylation and phosphotransfer to NRI, ensuring the NRI is highly phosphorylated regardless of environmental conditions (Pioskaz and Ninfa, 2003).

Figure 2
Structure of the synthetic genetic clock

The expected function of the circuit is as follows: When repressed, neither activator nor repressor are synthesized, and their levels will decrease by dilution as cells grow and divide. Eventually, the concentration of repressor will decrease below the threshold for repression of the glnA control region, and the glnAp1 promoter will drive the expression of NRI. This NRI will be quickly phosphorylated, and will repress the glnAp1 promoter and activate the glnAp2 promoter, leading to a dramatic increase in the concentration of NRI~P. Eventually, the activator will reach a high enough concentration to activate the glnK promoter, leading to a burst of repressor synthesis. This will continue until the concentration of repressor is sufficient to repress expression of the activator module, and eventually the system will be returned to its repressed state. To monitor oscillations, the repression of the lacZYA operon and the activation of Ntr genes, such as the natural glnA gene can be monitored by measuring the enzymes β-galactosidase and glutamine synthetase as the clock-containing strain grows in continuous culture (12).

Fabrication of the synthetic genetic clock

A unique aspect of our synthetic genetic clock is that the activator and repressor modules are not contained on plasmids in the cell, but rather are integrated into defined chromosomal locations, referred to as “landing pads” (Atkinson et al, 2003). We expect that this should provide a stable copy number to the modules, and indeed allows subtle manipulation of the copy number by using different locations on the chromosome, as the copy number of genes in rapidly growing cells displays a gradient from the origin of replication to the terminus of replication. The landing pads were designed to have nearby selectable markers that facilitate transfer of the modules between strains using standard P1vir-mediated generalized transduction. The basic features of landing pads include bracketing the synthetic genetic module with transcriptional terminators, to prevent transcription from outside promoters. Two landing pads, located in the rbs and glnK regions of the chromosome, were used to integrate the activator and repressor modules into the chromosome. These landing pads were described previously and are available in the plasmids pRBS3 and pDK11, respectively (Table 1). Some of the features of these landing pads are provided in Table 1. Since the fabrication of our synthetic genetic clock, we have devised a convienient plasmid system for construction of new landing pads and for modular construction of synthetic genetic devices in these landing pads. This vector system will be described in a later section. Our original fabrication methods for building the synthetic genetic clock modules consisted of standard molecular biology methods including PCR, site-specific mutagenesis, etc, and were described previously (Atkinson et al, 2003). In a later section, we will describe improved methods and materials that should ease the construction of synthetic genetic devices.

Table 1
Useful strains, modules, and plasmids

Assembly of the clock strain was as follows (strains listed in Table 1): The starting strain 3.300 contains a lacI null mutation. The glnA::Tn5 mutation was then introduced by P1vir-mediated generalized transduction, with selection for kanamycin resistance and screening for glutamine auxotrophy, forming strain 3.300A. A deletion of glnL and glnG was then introduced by transducing 3.300A with phage grown on strain SN24 (glnLglnG), with selection for glutamine prototrophy (glnA+) and screening for kanamycin sensitivity, producing strain 3.300LG. Strain 3.300LG is thus lacI glnL glnG but glnA+ and lacZYA+, and serves as the host for the synthetic genetic clock.

To recombine the clock genetic modules onto their chromosomal landing pads, a recD mutant is transformed with linearized plasmid DNA, where the site of cleavage is within the plasmid vector sequences. We have had success using strain TE2680 (Elliott, 1992), and have observed that the recD::TN10 mutation (conferring tetracycline resistance) can be transduced from strain TE2680 into a variety of strain backgrounds, permitting transformation with linear DNA. We typically use electroporation of the linearized DNA to obtain a large number of transformants, and use up to 1 μg of DNA per transfection. The drug-resistance marker of the landing pad is selected (Table 1), and the transformants are screened for resistance to ampicillin, which is encoded by the plasmid vector sequences. Transfectants that have the landing pad drug resistance but lack resistance to ampicillin are those in which the landing pad has recombined onto the chromosome. Transducing phage are then grown on these recombinants, and the integrated landing pad is introduced into desired strains by selection for the associated drug resistance of the landing pad. Thus, to build the genetic clock, strain 3.300 LG was sequentially transduced with (introducing the activator module) and glnK: (the repressor module), with selection for gentamycin resistance and chloramphenicol resistance, respectively. This strain, referred to NC12 was then made competant by standard procedures and transformed with p3Y15 (encoding NRII2302), with selection for ampicillin resistance, forming strain NC12/p3Y15, which is then used for clock studies. We have observed that the final plasmid-containing strain can be stored as a freezer culture at −80 °C, and reproducible clock experiments can be performed by streaking a few ice crystals containing the frozen cells on LB medium containing ampicillin immediately prior to the experiment, and picking a single colony isolate for the experiment. This strain and the intermediate strains and modules (Table 1) can be obtained from the authors.

Functions of the individual clock modules

The activator module and repressor module functions can be independently measured in intact cells. The activator module, when present in cells containing wild-type NRII and wild-type lacI encoding Lac Repressor, is predicted to form an N-IMPLIES logic gate with respect to ammonia and IPTG (Fig 3). In the presence of wild-type NRII, the nitrogen-rich state brought about by the presence of ammonia causes the formation of the NRII-PII complex and the rapid dephosphorylation of NRI~P (18). Furthermore, in the absence of IPTG, the constitutively-present Lac Represor blocks the expression of the activator module. Expression of a reporter consisting of the glnK promoter to lacZ, or any other Ntr gene, thus requires the presence of IPTG and the absence of ammonia, providing the N-IMPLIES logic function. After construction of the bacterial strain, we observed that the cells indeed only expressed β-galactosidase in the absence of ammonia and the presence of IPTG.

Figure 3
The activator module is an N-IMPLIES gate with positive feedback

The activator module also displayed toggle-switch function in the absence of the repressor module, as predicted by modeling (Atkinson et al, 2003, Fig 4). For these experiments, where only repression of the activator module was examined, the activator module was integrated into a strain deleted for both glnL and glnG, and the transmitter protein function was complemented by the plasmid p3Y15, encoding NRII2302. The strain also contained a wild-type lacI and lacZ, but contained a mutation of lacY. The mutation of lacY is essential to allow control of the internal IPTG concentration by variation in the external IPTG concentration (Novice and Weiner, 1957). The assembly of the strain was as follows: strain M7044 (lacY) was transduced to glnA::Tn5 by selection for kanamycin resistance and checking for glutamine auxotrophy, forming strain M7044A. This was then transduced to glutamine prototrophy using phage grown on strain SN24 (glnLglnG), producing strain M7044LG. This strain was then transduced with the activator module, by selection for gentamycin resistance of the integrated rbs landing pad, producing strain TS1. Finally, strain TS1 was made competant by standard methods and transformed with plasmid p3Y15, with selection for ampicillin resistance.

Figure 4
The activator module can function as a toggle switch

Toggle switch experiments are performed by growing the strain overnight in the absence and presence of 0.1 mM IPTG, resulting in naive and induced cultures. The overnight cultures are then diluted one million-fold into media containing various concentrations of IPTG. Growth is continued for about 15–17 generations, until the cultures were in mid-log phase, and β-galactosidase and glutamine synthetase are measured. The glutamine synthetase measurement provides an indication of the level of activator, while the β-galactosidase measurement provides an indication of the level of repressor. It is expected that the naive and induced cultures should show equivalent levels of β-galactosidase expression, but, if there is hysteresis in the activator module control, should shown very different glutamine synthetase levels. As shown in Fig 4, this behavior was observed. It should be noted that since the glnL2302 mutation was present on p3Y15, a variety of growth media can be used for the experiment, including complex media such as LB or nutrient broth. To compensate for the loss of normal Ntr regulation, glutamine is included in all media at 0.2% (w/v). [Note that glutamine does not survive autoclaving and filter sterilized glutamine must be added to the medium after autoclaving.]

The repressor module, when present in cells containing a deletion of the natural lacI and wild-type lacZYA and Ntr system, is predicted to display OR gate function with regard to ammonia and IPTG control of lacZYA expression (Fig 5). This is because ammonia blocks the formation of the activator, while IPTG blocks the function of Lac Repressor produced from the repressor module. Consequently, either stimulus provides full expression of lacZYA (Fig 5). The designed strain (3.300 containing the repressor module) was observed to indeed show the OR gate function.

Figure 5
The repressor module functions as an OR logic gate

Procedures for clock experiments

We have used a variety of growth media for clock experiments; a very good medium providing rapid growth of the cells and good oscillatory function is W-salts based glucose-glutamine-caseamino acids, with the following formula (per liter): 10.5 g K2HPO4, 4.5 g KH2PO4, 0.65 mL of 1 M MgSO4, 0.04 g thiamine, 0.04 g tryptophan, 0.5 g glutamine, 10 g glucose 5 g Bacto caseamino acids, 0.1 g ampicillin, and 1 mL of 34 mg/mL chloramphenicol. The medium is filter sterilized, owing to its labile components. A single colony of the clock strain is picked from an LB + ampicillin plate and innoculated into 2 mL of medium, and allowed to grow to saturation. 1.2 mL of the overnight culture is used to innoculate 120 mL of medium containing 0.5 mM IPTG. This culture is incubated until the turbidity reaches the desired turbidity for the clock experiment; typically 10–12 hr incubation provides an OD600 of ~0.6, which gives good results. The cells are harvested by centrifugation, resuspended in 120 mL of fresh medium lacking IPTG, repelleted, again resuspended in 120 mL of fresh medium lacking IPTG, and introduced into the continuous culture device. The volumes stated above can be scaled as appropriate for a variety of working volumes required by different fermentors. The optical density of the culture can be corrected in the initial stages of the run by control of the fermentor nutrient pump. Best oscillatory behavior was observed when the continuous culture was run at OD600 of 0.5–0.6. Cells are grown in the continuous culture device at a constant optical density by controlling the nutrient flow. Samples are removed periodically (or obtained from the fermentor efflux) and assayed for β-galactosidase and glutamine synthetase.

We have used two methods for conducting the continuous culture experiments, one utilizing a standard laboratory fermentor that was not desiged to function as a turbidostat, and the other using a custom-built turbidostat. Both methods will be briefly described here. It is important to note that the clock strain does not grow at a constant rate in continuous culture; rather, the presence of the synthetic genetic clock causes growth to slow down as the activator module and lacZYA expression are derepressed and to speed up as the activator module and lacZYA are repressed. Thus, to maintain a constant culture optical density, the nutrient flow must be continuously adjusted. When using a standard laboratory fermentor not designed to function as a turbidostat, this is accomplished by periodically sampling the fermenter efflux, measuring OD600, and manually adjusting the nutrient flow rate so as to hold the culture at constant turbidity (12). Since clock experiments typically are run for 80 hrs or more, the experiments require lots of coffee and/or several people to take shifts minding the fermenter. Nevertheless, good oscillatory function can be easily observed in such experiments, despite the fairly crude control of the culture turbidity (Atkinson et al, 2003, Fig 6).

Figure 6
Results of clock experiments using a standard laboratory continuous culture device

For experiments with a typical laboratory fermentor, the starting culture at approximately the desired working OD600 is introduced directly into the reactor, and the optical density is corrected to the desired optical density by manipulation of the nutrient pump. Filtered air is pumped into the reactor and vigorous stirring is used to provide good aeration. Medium is pumped into the reactor from a 4L reservior, to which fresh sterile medium is added aseptically as needed. Samples are collected from the efflux line directly into a clean disposable cuvette; the OD600 of the culture is measured, and aliquots are used for the β-galactosidase and glutamine synthetase assay (12).

To automate the clock experiments, we developed a turbidostat as depicted in Fig 7. The reactor is a 500 mL Erlenmeyer flask that contains a 120 mL culture that is kept at constant optical density by varying the media flow appropriately. A close-up view of the reactor is shown in Fig 8; it has four connections to the outside: one line to pour media in, one line to suck culture out when the volume slightly exceeds 120 ml (waste line), one line to take samples, and one line to pump in filtered air. The air pump is a large fish tank aerator connected to a standard 0.22 μm filter to provide sterility. The reactor is contained within a standard small laboratory incubator, and good mixing of the culture is provided by placing the flask on a standard stir plate and including a magnetic stir bar in the vessle (not depicted). In addition to the stirring, the pumping in of air contributes to mixing within the reactor.

Figure 7
Schematic diagram for a home-made turbidostat
Figure 8
Close-up view of the reactor for a home-made turbidostat

The media is pumped in from a media reservoir at a variable speed. To maintain sterility and accomodate the large volumes of medium required, we use a home-made medium “tree” that consists of up to 6 one-gallon bottles connected in a daisy chain by sterilizable tygon tubing. This apparatus is sterilized dry under a tin-foil tent, placed into a bacteriological hood while still hot, and the filter sterilized medium is introduced after it cools sufficiently. The waste line sucks culture out and has twice the flow rate of incoming media, and it sucks culture out only when the volume goes higher than 120 mL. The waste line needs to have a stronger flow rate than the media line in order to keep the volume constant. The depicted design was chosen, as opposed to having the same flow rate of media in and culture out, because fluctuations in the flow rate of the lines, even if both lines are connected to the same pump, makes the control of the volume more difficult. Any slight difference in flowing rates may lead to volume accumulation or reduction, and since each experiment lasts several days, this constitutes a problem. The waste flow rate is made double to incoming medium flow rate by using tubing of twice the diameter and the same pump (Fig 7).

The sampling pump (Pump 2 in Fig 7) intermittently removes culture from the fermentor at a rate that is considerably less than the nutrient flow. Every five minutes a small aliquot of culture is pumped out through the sampling line to a flow cell (Fig 7). The flow rate of media is changed as a function of this OD600, in order to keep the optical density constant. After each OD600 reading, the sampling pump is shut off, and the flow cell is washed with sterile water, using water that is pumped into the sampling line from a water reservoir by the wash pump (Fig 7). This wash routine was found to be important, as continuous pumping of the culture through the flow cell and autosampler leads to the formation of a biofilm on the surfaces, fouling the reading of absorbance and contaminating the samples and instruments. Every 30 min, or six OD600 readings, a 2 mL sample of the culture is directed to waste to clear the lines and a 1 ml sample is collected in a well of a deep well 96-well microtitre plate. The automatic sampler allows the sampling line to point to any of the 96 wells in the microtitre plate, or to waste.

The system depicted in Fig 7 is fully computer-controlled. The media, sampling and wash are each pumped by a computer controlled peristaltic pump (media and sampling lines are pumped by VS-series Alitea peristaltic pumps, water is pumped by a S-series Alitea peristaltic pump), while the spectrometer (Ocean Optics SD2000) and autosampler (AIM 3200) also have an interface to the computer. The flow cell to measure absorbance, as well as the pumps, autosampler and spectrometer were purchased from FIAlab Instruments (Bellevue WA).

The software used to control the system was implemented in Labview (National Instruments, Austin TX). Labview is a graphical programming language that eases the interface of the computer with instruments and data acquisition devices. Some instruments, like the autosampler and valves, have an interface to the computer via the serial port and an ascii language, while others like the spectrometer or a data acquisition card, have some other way of interfacing with the computer, but they offer a basic library of functions in Labview to operate them. Essentially any instrument can be controlled from Labview. Besides its ease to interface with instruments, as a programming language Labview offers the capabilities of any other programming language (like C/C++ or Java). Using Labview, we designed the following algorithm to run the system (collect samples and keep the OD constant)

  • Set the target OD and the media flow rate (measured as the percentage of the top speed in the pump, initially 50%)
  • Every five minutes:
    • Pump culture from the flask to the flow cell, and make an OD measurement
    • Wash the flow cells with sterile water
    • Adjust the media flow rate according to the formula
  • Every half an hour (every 6 readings), pump 2 mL of culture to waste and collect 1 ml of culture in a deep well 96-well plate.

Conducting automated experiments requires that the samples be maintained at low temperature (4 °C) to prevent growth of the cultures and changes in the level of β-galactosidase and glutamine synthetase. We observed that samples held at 4 °C displayed stable levels of these two enzymes for several days. To provide uniformity, we routinely maintain all samples at 4 °C for at least 4 hrs before conducting the assays. The autosampler is contained within a standard small refrigerator, which we adapted to our purpose by drilling a small entry port into one side and a small “overflow” port into its base. The entire apparatus, including refrigerator, incubator, computer, pumps, and flow cell easily fits onto one standard laboratory bench.

The automated system utilized miniaturized assays, and uses samples that have been stored at 4 °C. We observe that the miniaturized assays are somewhat noiser than the standard assays, and that some of this may be due to inconsistent resuspension of the settled cells from the stored samples. Thus, we recommend special attention to resuspending the settled cells, by both extensively vortexing the collection plates and pipetting the samples into and out of the well several times before taking the aliquots for measurement. This noisiness notwithstanding, oscillatory behavior can be easily observed using an automated system with samples stored at 4 °C and miniaturized assays (Fig 9).

Figure 9
Results of a clock experiment employing an automated system and miniaturized β-galactosidase assay.

Procedures for the measurement of glutamine synthetase and β-galactosidase for clock and toggle-switch experiments

Glutamine synthetase microassay

  1. A reaction mix is assembled as follows:
    ReagentAmount (mL)
    0.45 M Imidazole pH 7.3335
    0.3 M NH2OH-HCl7
    0.01 M MnCl23.5
    0.06 M KAs2O4 pH 7.235
    6 mM ADP, pH 7.07
    1.5 mg/ml CTAB7
    2.9 % glutamine12
    Correct the pH to 7.27 with KOH.
  2. Transfer 400 μL of each sample to be assayed to a well in a deep well 96-well plate. Centrifuge to pellet the cells and carefully remove the supernatant and discard. Resuspend the cells in 40 μL of CTAB cell-wash buffer, and transfer to a standard flat-bottom 96 well microtitre plate. The CTAB cell-wash buffer is: 5 mM Imidazole pH 7.15, 0.1 mg/mL CTAB, 0.27 mM MnCl2.
  3. 100 μL aliquots of the reaction mix are pippetted into the wells; this starts the reaction. ypically, the mix aliquots are added at 20 sec intervals using an 8-channel micropipettor, and later the reactions will be stopped in the same sequence at 20 sec intervals. When all samples have been added, the plate is incubated at 37 °C in a standard microbiological incubator. In a separate plate, a sample from an early time point in the clock experiment (where the glutamine synthetase activity is expected to be high) are set up in triplicate in three separate wells of a microtitre plate. This plate will be used to determine the appropriate length of time to incubate the reactions. At 30 min intervals, stop solution is added to one of the three wells in the test plate, and the reactions are allowed to continue until a dark brown color is obtained upon stopping the test reaction. Typically, a one-hour incubation is sufficient for measurement of the activity. The stop solution is (per L): FeCl3·6H2O 55 g; TCA 20 g, HCl 21 mL.
  4. To stop the reactions, 100 μL of stop solution is added to each well, in the same sequence and time interval that the reactions were started. Upon addition of the stop solution, the reaction mixtures that contain high glutamine synthetase activity will turn dark brown. Since the cell samples should all have similar OD600, the oscillations in GS level should become immediately apparent as the reactions are stopped.
  5. The OD540 of the stopped reactions are directly measured in a plate reader. In a separate plate, the OD600 of the original cell samples are directly read. The glutamine synthetase activity, in arbitrary units, is simply the OD540 value divided by the OD600 value.
  6. For toggle-switch experiments, the cells from 1 mL of the cultures are pelleted in microcentrifuge tubes, and resuspended in 0.1 mL of CTAB-cell wash buffer. Aliquots of 20 μL and 50 μL are then added to the reaction mix in microtitre plates, and the procedure described above is followed.

β-galactosidase microassay

  1. The wells of a deep-well 96-well plate are loaded with 400 μL of Z-buffer, 10 μL of 0.1% SDS, and 10 μL of chloroform (add last). The formula for Z-buffer is (per liter): 16.1 g Na2HPO4.7H2O, 5.5 g NaH2PO4.H2O, 0.75 g KCl, 0.246 g MgSO4.7H2O. Immediately prior to the assay, add 0.27 mL of β-mercaptoethanol/100 mL of Z-buffer.
  2. A 100 μL of the culture sample is added progressively to each well and mixed thoroughly by pipetting (the chloroform must be well mixed into the suspension for uniform cell permeabilization). We typically use an 8-sample multichannel micropipettor to minimize the difference in exposure of samples to the chloroform. The reaction mixtures are then incubated 10 min at room temperature.
  3. To start the reactions, 100 μL of ONPG (4 mg/mL in 0.1 mM phosphate buffer) is added to the wells in sequence and at fixed intervals, and mixed thoroughly by pipetting.
  4. Allow the reaction to proceed until the appearance of yellow color in some of the wells. Note that oscillations in the level of β-galactosidase becomes readily apparent at this stage, and some of the wells should never turn bright yellow since they contain very low levels of β-galactosidase. Typically, a 5–10 min incubation is required.
  5. Reactions are stopped in the sequence they were started and at the same intervals, by adding 200 μL of 1M Na2CO3.
  6. The stopped reactions are allowed to sit for 10 min for full color development and settling of the cells, then 150μL aliquots (from the top) from each well is transfered to a 96 well flat bottom plate and the OD410 and OD560 are recorded. In a separate series of wells, the OD600 of the original culture samples is measured.
  7. The β-galactosidase activity is [OD410 - (1.75 · OD560)]/(OD600 · time).

Improved procedures for the fabrication of synthetic genetic modules and integration of these modules into chromosomal landing pads

During our synthetic biology studies, we have developed fabrication methods in concert with development of the clock, such that early versions of the clock do not incorporate the most useful aspects of fabrication methodologies developed later. Here, we will present our most recent vector system for the fabrication of synthetic genetic modules and incorporation of the fabricated modules into chromosomal landing pads. The vector system should allow the placement of any module in any position and on either strand of the chromosome, and permit the genetic modules to be easily constructed from diverse natural components.

To develop new landing pads, the vectors shown in Fig 10 can be used. A non-essential and easily scored gene, such as a sugar utilization gene, is chosen as the target for chromosomal integration. Two fragments of the gene of ~90 bp each are cloned between the EcoR1 and SacI sites, and between the XhoI and HinDIII sites. The landing pad will be localized by recombination of these sequences with their chromosomal counterparts; we typically choose “right” and “left” target sequences from within the same gene. Next, we introduce an antibiotic resistance gene flanked by strong transcriptional terminators between the SacI and NotI sites, as depicted in Fig 11 (Step1 plasmids). We have used kanamycin resistance and chloramphenicol resistance in Step 1 plasmids; we refer to this marker as “AB1”. Step 1 plasmids are functional landing pads that can be used to integrate modules into the target gene, those available and under construction are listed in Table 1. One clones the genetic module to be placed onto the chromosome between the remaining 5 unique sites that are flanked by the recombination target fragments. One could introduce the module as a unit after assembly in another plasmid, or sequentially add portions. Typically, we use the NdeI site for the begining of the structural gene, since NdeI sites contain an ATG initiation codon. If the structural gene is cloned as an NdeI-SalI fragment or as an NdeI-BamHI fragment, a strong terminator can be added between the SalI and XhoI sites, isolating the gene from extraneous transcription from both directions. Next, the promoter and mRNA leader sequence are added as a NotI-NdeI fragment, completing the module. Typically, this fragment is assembled in separate plasmids, or added as a synthesized DNA fragment by annealing long oligonucleotides that have been designed to have the appropriate overhangs. We have worked with synthesized fragments of length 100 bp, which permits the construction of many combinatorial promoters/leader sequences. All of our vectors contain a unique PstI site in the vector sequence that allows easy linerization of the plasmid for electroporation into a recD mutant. In the event that there is a PstI site in the module itself, several other unique sites are also present in the vector portion of the molecule, that might be used.

Figure 10
The pStep0 plasmids that can be used to create new landing pads
Figure 11
The pStep1-pStep4 plasmids, that can be used to place genetic modules onto the E. coli chromosome

Synthetic genetic devices can be built up by placing different modules into different chromosomal landing pads, as was done for our initial clock. The problem with this approach is that each module is linked to a distinct antibiotic resistance marker, and one quickly runs out of antibiotic resistance selections that offer very strong selection in the presence of each other. Tetracycline resistance provides a strong selective marker, but we exclude its use from our system since we use the recD::Tn10 (tet resistant) mutant to recombine modules onto the chromosome. Similarly, our plasmid vectors encode ampicillin resistance, and thus we do not use this drug resistance marker elsewhere in the system. We have found that kanamycin resistance and chloramphenicol resistance provide very strong selection (Table 1), and that gentamycin resistance, as found in our pRBS landing pad (Table 1), does not. Thus, integrating three separate modules onto the chromosome is a problem. To overcome this, two approaches are taken. First, we are currently investigating additional antibiotic resistance markers, such as spectinomycin resistance, for use as “AB1” in Step 1 plasmids. Second, we designed the vector system such that each landing pad can be used in a reiterative fashion by sequential integration of modules, which end up next to each other on the bacterial chromosome. The plasmids used for these sucessive integrations, pStep2, pStep3, and pStep4 are shown in Fig 11. As before, there are two versions of each of these to permit either orientation of the genetic module. The target sequences of the pStep2-pStep4 plasmids consist of internal segments of the drug resistance genes used in the prior step. Thus, if one uses a pStep1 plasmid with kanamycin as the selectable marker (AB1 in Fig 11), the corrsponding pStep2 plasmid will integrate into the AB1 gene and contain as a selectable marker chloramphenicol resistance (AB2). The pStep3 plasmid will contain AB1 as the selectable marker, and integrate into the AB2 gene, and the pStep4 plasmid will contain the AB2 gene as a selectable marker, and integrate into the AB1 gene. Thus, the drug resistance of the module is switched between AB1 and AB2 upon each sucessive integration, and the synthetic genetic constructs are each flanked by terminators and located next to one another on the chromosome. A simple PCR test of the final strain can then confirm the expected structure. The sequence of nesting is depited in Fig 12 as well as the predicted end product of an assembly where 4 promoter-gene fusions (G1-G4) are positioned in a single landing pad.

Figure 12
The sequence of nesting that occurs upon succesive use of the pStep1-pStep4 plasmids for integration of modules into the chromosome

Fabricating genetic modules

Most of the genetic modules that we fabricate consist of a regulated promoter that drives the expression of a regulatory gene. The minimal components therefore consist of a bacterial promoter, regulatory sites for control of the promoter, an mRNA leader sequence containing translational initiation sequences, the structural gene for the desired regulatory protein, and a transcriptional terminator sequence. For genetic isolation, it is advisable to also include transcriptional termination sequences upstream from the regulated promoter. Fabrication consists of choosing the sequences to be used and assembling the sequences. The modules pertenent to the clock and toggle switch experiments are presented in Table 1. A much more extensive registry of parts for synthetic biology constructions, including parts developed by over 40 laboratories (including our own) is the registry of BioBricks ( This registry contains a large variety of promoter, operator, mRNA leader, and terminator elements, as well as a variety of structural genes encoding transcription factors and reporters. Standard cloning methods can be used to introduce any element from the biobricks collection into our vectors, by either performing a blunt end ligation or using PCR to add appropriate sites to the biobrick. We are currently developing a modified series of landing pad vectors that are able to directly accept any Biobrick fragment or construct. Although cloning from BioBricks or other sources is an appropriate way to build modules, we have found that in many cases, simply cloning annealed long oligonucleotides of desired sequence is the most direct way to build modules.


  • Atkinson MR, Ninfa AJ. Characterization of Escherichia coli glnL mutations affecting nitrogen regulation. J Bacteriol. 1992;174:4538–4548. [PMC free article] [PubMed]
  • Atkinson MR, Ninfa AJ. Mutational analysis of the bacterial signal-transducing kinase/phosphatase nitrogen regulator II (NRII or NtrB) J Bacteriol. 1993;175:7016–7023. [PMC free article] [PubMed]
  • Atkinson MR, Blauwkamp TA, Bondarenko V, Studitsky V, Ninfa AJ. Activation of the glnA, glnK, and nac promoters as Escherichia coli undergoes the transition from nitrogen excess growth to nitrogen starvation. J Bacteriol. 2002;184:5358–5363. [PMC free article] [PubMed]
  • Atkinson MR, Pattaramanon N, Ninfa AJ. Governor of the glnAp2 promoter of Escherichia coli. Mol Microbiol. 2002;46:1247–1257. [PubMed]
  • Atkinson MR, Savageau MA, Meyers JT, Ninfa AJ. Development of genetic circuitry exhibiting toggle switch or oscillatory behavior in Escherichia coli. Cell. 2003;113:597–607. [PubMed]
  • Barkai N, Leibler S. Circadian clocks limited by noise. Nature. 2000;403:267–268. [PubMed]
  • Elliott T. A method for constructing single-copy lac fusions in Salmonella typhimurium and its application to the hemA-prfA operon. J Bacteriol. 1992;174:245–253. [PMC free article] [PubMed]
  • Feng J, Goss TJ, Bender RA, Ninfa AJ. Activation of transcription initiation from the nac promoter of Klebsiella aerogenes. J Bacteriol. 1995;177:5523–5534. [PMC free article] [PubMed]
  • Hoch JA, Silhavy TJ. Two-component signal transduction. American Society for Microbiology; Washington, D.C: 1995.
  • Hunt TP, Magasanik B. Transcription of glnA by purified Escherichia coli components: core RNA polymerase and the products of glnF, glnG, and glnL. Proc Natl Acad Sci USA. 1985;82:8453–8457. [PubMed]
  • Jin T, Inouye M. Ligand binding to the receptor domain regulates the ratio of kinase to phosphatase activities of the signaling domain of the hybrid Escherichia coli transmembrane receptor, Taz1. J Mol Biol. 1993;232:484–492. [PubMed]
  • Lee SY, De La Torre A, Yan D, Kustu S, Nixon BT, Wemmer DE. Regulation of the transcriptional activator NtrC1: structural studies of the regulatory and AAA+ ATPase domains. Genes Dev. 2003;17:2552–2563. [PubMed]
  • Levskava A, Chevalier AA, Tabor JJ, Simpson ZB, Lavery LA, Levy M, Davidson EA, Scouras A, Ellington AD, Marcotte EM, Voigt CA. Synthetic biology: engineering Escherichia coli to see light. Nature. 2005;438:441–442. [PubMed]
  • Ninfa AJ, Reitzer LJ, Magasanik B. Initiation of transcription at the bacterial glnAp2 promoter by purified E. coli components is facilitated by enhancers. Cell. 1987;50:1039–1046. [PubMed]
  • Novick A, Weiner M. Enzyme induction as an all-or-none phenomenon. Proc Natl Acad Sci USA. 1957;43:553–566. [PubMed]
  • Oehler S, Eismann ER, Kramer H, Muller-Hill B. The three operators of the lac operon cooperate in repression. EMBO J. 1990;9:973–979. [PubMed]
  • Pioszak AA, Ninfa AJ. Genetic and biochemical analysis of phosphatase activity of Escherichia coli NRII (NtrB) and its regulation by the PII signal-transduction protein. J Bacteriol. 2003;185:1299–1315. [PMC free article] [PubMed]
  • Pioszak AA, Ninfa AJ. Mechanism of the PII-activated phosphatase activity of Escherichia coli NRII (NtrB): how the diferent domains of NRII collaborate to act as a phosphatase. Biochemistry. 2003;42:8885–8899. [PubMed]
  • Reitzer LJ, Magasanik B. Expression of glnA in Escherichia coli is regulated at tandem promoters. Proc Natl Acad Sci USA. 1985;82:1979–1983. [PubMed]