A framework and model system to investigate linear system behavior in Escherichia coli
© Hajimorad et al; licensee BioMed Central Ltd. 2011
Received: 15 October 2010
Accepted: 22 April 2011
Published: 22 April 2011
The ability to compose biological systems from smaller elements that act independently of the other upon assembly may help make the forward engineering of biological systems practical. Engineering biology in this manner is made difficult by the inherent nonlinear response of organisms to genetic devices. Devices are inevitably coupled to one another in the cell because they share the same transcriptional machinery for expression. Thus, new properties can emerge when devices that had been characterized in isolation are expressed concurrently. We show in this report that, similar to physical systems, the Escherichia coli (E. coli) transcriptional system can exhibit linear behavior under "small" perturbation conditions. This, in turn, allows devices to be treated as independent modules.
We developed a framework and model system consisting of three devices to investigate linear system behavior in E. coli. Our framework employed the transfer curve concept to determine the amount of nonlinearity elicited by the E. coli transcriptional system in response to the devices. To this effect, the model system was quantitatively characterized using real-time quantitative PCR to produce device transfer curves (DTCs). Two of the devices encoded the bacterial neomycin phosphotransferase II (nptII) and chloramphenicol acetyl transferase (cat), while the third encoded the jellyfish-originating green fluorescent protein (gfp). The gfp device was the most nonlinear in our system, with nptII and cat devices eliciting linear responses. Superposition experiments verified these findings, with independence among the three devices having been lost when gfp was present at copy numbers above the lowest one used.
We show that linear system behavior is possible in E. coli. Elucidation of the mechanism underlying the nonlinearity observed in gfp may lead to design rules that ensure linear system behavior, enabling the accurate prediction of the quantitative behavior of a system assembled from individually characterized devices. Our work suggests that biological systems follow principles similar to physical ones, and that concepts borrowed from the latter (such as DTCs) may be of use in the characterization and design of biological systems.
Engineering biological systems with predictable, quantitative behavior is currently a challenging problem. Presently, this requires months (at times years) of trial-and-error type of experiments, with the engineering of functional systems being more akin to art than engineering . Synthetic biology aims to develop foundational principles and technologies that will enable the systematic forward engineering of biological systems [2–4]. In particular, synthetic biology aims to develop frameworks that apply the engineering principles of abstraction, modularity, and composition to biological engineering. Basic abstraction and physical composition frameworks have been applied to the engineering of biology through the use of BioBricks and the Registry of Standard Biological Parts [5–7]. A characteristic feature of other established engineering disciplines is the ability to design and construct systems by way of modularity. The concept of modularity allows engineers to design and build physical systems by bringing together modules that contribute independently to the whole, thereby giving rise to a system whose quantitative behavior can be predicted from its constituent modules [8–10]. A pressing research question is whether the complexity of living organisms allows engineers to design and construct biological systems from smaller elements characterized in isolation [11, 12]. The success of synthetic biology as an engineering discipline will depend, in part, on establishing the conditions necessary for this independence property to hold true in living systems . The research contribution of this study is the application of engineering principles towards realizing modularity and functional composition in biological systems. More specifically, we show that genetic devices (each consisting of a promoter, ribosome binding site, gene of interest, and transcription terminator) can behave in a standardized, quantitatively predictable manner. The ability to view devices as modules may be of benefit in such applications as metabolic pathway construction for the production of natural products and other chemicals (microbial chemical factories).
This study began with the hypothesis that the introduced synthetic devices can be viewed as perturbations to the E. coli system. That is, the amount of DNA acting as input to the transcriptional machinery of the host increases with their addition. So long as this increase (i.e., perturbation) is kept "small," the E. coli system may perhaps be approximated as a linear one with respect to the introduced synthetic devices, thereby enabling superposition and the decoupling of synthetic devices from one another. This is the small-signal approximation used in the field of electronic circuit design . There, it is used in the design of analog amplifiers, where voltage and current signals act as inputs to nonlinear, transistor-based systems. It should be noted that the copy number of synthetic devices may not be the only factor that perturbs microbial organisms. Promoter strength, ribosome binding site (RBS) strength, gene length, codon usage, and product function are perhaps important factors too. As the intent of our study was to investigate whether E. coli can accommodate linear system behavior with standard elements used to genetically modify the organism, we focused on copy number here. Our approach to investigate system nonlinearity involved varying the copy number of devices to generate transfer curves. Nonlinearity of physical systems is often investigated by using transfer curves, where the transfer curve of a system specifies how its output varies with respect to its input under steady-state conditions [16, 17]. We show with our approach that concepts applicable to physical systems also apply to biological ones, and that superposition is possible in E. coli under certain conditions.
Copy number of genetic device varied with plasmid origin of replication
Doubling time of cells grown in the indicated media harboring the plasmid backbone (Figure 3A)
45 ± 1
50 ± 2
54 ± 3
80 ± 3
83 ± 3
83 ± 5
84 ± 5
Linear device transfer curves obtained in E. coli
After verifying that the copy number of nptII can be varied in our system, we performed experiments to obtain its DTC. Cells that had been harvested above were used to quantify the transcript level of the nptII device. Total RNA was extracted from cells and transcript level quantified using real-time qPCR [28–30]. The results were plotted against the copy number values determined prior (Figure 4A), yielding the DTC (Figure 4B). RNA transcript level and copy number values have been normalized to that of the pSC101 construct, which were assigned a mean value of one in each case. The y-axis value for each data point indicates how that particular construct's steady-state transcript level compares relative to that of the pSC101 construct. Similarly, the x-axis value for each data point in the plot indicates how that particular construct's steady-state copy number compares relative to that of the pSC101 construct. Also shown in the plot are linear regressions fitted to the data. With R2 > 0.9, the data suggest that the E. coli system response to the nptII device perturbation can be considered linear. The 95% confidence interval for the y-intercepts were also -0.3 - 0.8 and -0.3 - 0.5 for LB and M9 media, respectively. These included the origin, further suggesting that a linear regression was an appropriate fit for the data. That is, one cannot have nptII RNA transcript produced when there is no corresponding DNA present in the cell. Our results also suggest that the DTC of nptII can be linear under different contexts. That is, the choice of growth medium does not appear to impact system linearity (Figure 4B).
We next introduced another device into the plasmid backbone to increase the perturbation level. The goal was to see whether the presence of an additional device would lead to nonlinear DTCs. This device encoded chloramphenicol acetyl transferase (cat), which confers resistance to the antibiotic chloramphenicol. The cat device was expressed from its native promoter (Figure 3B). E. coli DH1 cells harboring the constructs were grown in LB medium as described in the Methods. No chloramphenicol was added to the medium during growth, with only kanamycin having been used for selection purposes. The growth rate of cells was comparable among the constructs, with OD600 nm doubling every ~50 minutes in the log phase (data not shown). At mid-exponential growth, cells were harvested and total RNA and DNA extracted. Relative nptII and cat transcript levels and copy number were subsequently quantified using real-time qPCR (Figures 4C and 4D). As was done for the backbone (Figure 4B), linear regressions were fitted to the data. With a R2 > 0.96 for each device, the data suggest that the combined nptII and cat device perturbation level appears to be "small" enough to elicit a linear response from the E. coli system. The 95% confidence intervals were also -0.7 - 0.2 and -1.0 - 0.8 for nptII and cat y-intercepts, respectively, including the origin for each device once again. The fact that no chloramphenicol was present in the growth medium suggests linear transfer curve response is not due to antibiotic resistance mechanisms.
Introduction of gfp genetic device led to nonlinear device transfer curves
Doubling time of cells grown in LB medium harboring the different constructs in model system
45 ± 1
50 ± 2
54 ± 3
backbone with cat device
51 ± 2
58 ± 4
backbone with gfp device
46 ± 1
53 ± 2
56 ± 1
80 ± 2
backbone with cat and gfp devices
46 ± 1
62 ± 11
103 ± 2
As done for the other systems, a linear regression was fitted to the gfp data (Figure 5A). The results suggested that the E. coli system response to the gfp device was not linear (not shown). This was due to the 95% confidence interval for the y-intercept not including the origin (i.e. 1.2 - 3). The 95% confidence interval for the y-intercept still did not include the origin if only the first three constructs (i.e. the ones with the pSC101, p15A, and pMB1 replicons) were considered (data not shown). This suggested a piecewise-linear model for the data, with the first segment consisting of data points for the pSC101, p15A constructs (where the 95% interval for the y-intercept included the origin, data not shown) and the second segment data points for the pMB1, pUC constructs. The piecewise-linear approximation is used in electrical engineering to model nonlinear transfer curves . To arrive at a piecewise-linear model in a systematic manner, the problem was approached as a nonlinear least squares optimization . The NLIN Gauss-Newton procedure in SAS was used to fit the data to a piecewise-linear model consisting of two segments with unknown slopes and an unknown breakpoint (Figure 5A). The algorithm was not forced to go through the origin. This way, the appropriateness of the fit could later be verified by noting the y-intercept obtained from the slopes and breakpoint numerically computed by the NLIN procedure. No noticeable normality or variance issues were observed after analyzing the residuals in SAS (data not shown), strengthening the argument for the appropriateness of the model. A y-intercept of 0.07 was obtained, which is approximately equal to the origin. The change in slope between the two segments (~7X fold) was taken as a means to report the nonlinearity observed in the gfp DTC.
As was done for gfp, SAS was used to fit piecewise-linear models to the data for the nptII and cat devices (Figure 5B and 5C, respectively). Unlike the former, however, the fits that minimized the sum of squared residuals had the first segment consisting of data points for the pSC101, p15A, and pMB1 constructs (data not shown). The second segment could, thus, not be determined because the pUC construct remained as the only available point (i.e. one needs two points to fit a line). As an approximate solution to this problem, a piecewise-linear model was determined for each device by fitting a linear regression to the pSC101, p15A, pMB1 and pMB1, pUC constructs for the first and second segments, respectively (Figures 5B and 5C). The two segments for the nptII and cat devices had similar slopes. This was noticeably smaller than the ~7X fold change observed for gfp (Figure 5A).
Superposition lost at higher expression levels with the addition of gfp
The larger nonlinearity observed in the gfp DTC (Figure 5A) interestingly also manifested itself in superposition experiments involving device transcript levels. Cells containing the pSC101, p15A, and pUC origins that had been harvested above were used to quantify the transcript levels of the different devices. The latter two replicons were chosen so as to have data points on either side of the DTC breakpoints (Figure 5). The pSC101 origin was selected to investigate whether superposition observed at the plasmid copy level (Figure 6A) also applied to device transcript levels. Total RNA was extracted from cells and transcript levels quantified using real-time qPCR (Figures 6B, 6C, and 6D). As the plasmid copy varied among the constructs for a particular replicon (Figure 6A), transcript levels were not only normalized to the endogenous 16S but also to the plasmid copy number. That is, values reported are RNA produced per unit plasmid. Superposition would be in effect if the amount of RNA produced by the nptII device (Figure 6B) did not change after additional devices had been introduced into the plasmid. That is with superposition, if the plasmid backbone harboring genetic device nptII led to the production of a certain amount of that device's RNA, one would obtain the same amount upon addition of cat and/or gfp devices. Similar arguments apply to superposition for the cat and gfp devices (Figures 6C and 6D, respectively). As was the case for plasmid copy (Figure 6A), our results indicated that nptII transcript level is unaffected by the addition of cat and/or gfp genetic devices to a plasmid with a pSC101 replicon (Figure 6B). Device addition began to have an impact at higher copy numbers. The changes in nptII transcript level were, once again, the most pronounced by the addition of the gfp device, with those resulting from cat not being statistically significant even with the pUC origin. The data for cat and gfp RNA exhibited a similar pattern. Once again, cat or gfp transcript level was unaffected by the addition of the other device to a plasmid with a pSC101 replicon (Figures 6C and 6D). The addition of gfp, however, affected cat RNA at the higher copy numbers (Figure 6C). This was not the case in the reverse direction. That is, cat device addition did not impact gfp RNA at the higher copy numbers of p15A and pUC (Figure 6D). These results suggest that the extent of the changes brought about by gfp is large enough to mask those caused by the addition of cat.
Our results indicate that the E. coli biological system can exhibit linear system behavior (Figure 6). In the model system presented in this work, the necessary condition with all three genetic devices present was to use a plasmid backbone harboring the pSC101 replicon. That is, our experimental results showed superposition to be present at this copy number. The presence of superposition, however, was not only a consequence of having used the pSC101 origin. In the absence of the gfp device, superposition was found even with a pUC origin (Figure 6). The finding that superposition is possible under different contexts is important. It suggests that the nonlinearity in the E. coli system is not complex to the point of preventing design efforts to elicit a linear system response.
where RNA SS and DNA SS are the steady-state RNA and copy number of the encoding DNA, respectively. The nonlinearity observed in the piecewise-linear DTCs (Figure 5) may, thus, be modeled by a change in the slope term of equation (2). That is, the synthesis and/or degradation rate varies for the devices at higher copy numbers. Analysis of RNA degradation after cells had been treated with rifampicin did not reveal a noticeable change in the decay rates of gfp and cat transcripts at the higher copy number of pUC relative to p15A (data not shown). This suggests that the larger nonlinearity observed in gfp is due to a modulation in the synthesis rate. The fact that gfp and cat have identical PL promoters in our model system further suggests that the mechanism involved does not affect the initiation of transcription. Perhaps, the stringent response is implicated in this matter. Previous work has indicated that the E. coli stringent response can differentially impact the elongation rate of transcripts . We were unable to grow cells harboring the model system plasmids (Figure 3C) with the pMB1 and pUC replicons in M9 minimal medium (data not shown), suggesting that cells might be in a starvation like condition at the higher copy numbers. This is also supported by our growth rate data in LB medium, with the doubling time increasing at the higher copy numbers (Table 2). Gene sequence has been shown to influence transcriptional pausing of RNA polymerase in the presence of guanosine tetraphosphate (ppGpp) . Perhaps, the coupling present between transcription and translation in E. coli facilitates this effect, with ribosomal ppGpp synthesis affecting upstream RNA polymerase that is in the process of transcript elongation. Cooperative activity between RNA polymerase and ribosomes has been shown to modulate the elongation rate of transcripts in E. coli, with this linkage involving the NusE-NusG complex . It also needs to be acknowledged that copy number was the only perturbing factor considered in our study. The cat and gfp genes were expressed as done previously  (see below). As such, the devices have different RBS sequences/strengths. Promoter strength, RBS strength, and codon usage may be coupled perturbing factors because of the cooperative activity between RNA polymerase and ribosomes. Indeed, a range of promoters, RBS strengths, and codon usage need to be used to better elucidate the mechanism underlying the large observed nonlinearity in gfp's DTC (Figure 5).
Our results do, however, suggest that the transfer curve-based framework has application in the engineering of biological systems. We observed a correspondence between DTC nonlinearity and the break down of linear system behavior. That is, the gfp device was found to elicit a more nonlinear DTC response from the E. coli system than the other tested devices (~7X change in slope as compared to no change for nptII and cat devices, Figure 5), which was reflected in superposition being lost when gfp was present at copy numbers above the pSC101 level. While a change in growth rate offers an alternate gauge for nonlinearity, it does not appear to provide one with the same level of accuracy. The doubling time as monitored by OD600 nm only began to change noticeably with the pUC replicon (Table 2), failing to indicate changes to copy number (Figure 6A) and transcript (Figures 6B, C, and 6D) due to gfp at the other origins. This indicates the significance of quantitative techniques (such as DTCs) to synthetic biology characterization efforts because growth rate alone is unable to accurately capture changes that take place due to device addition. DTCs may have application in the general characterization of devices. A device could be characterized by cloning it into a standard plasmid and its copy number varied by way of different replicons. Based on the nonlinearity gauged from its resulting DTC, one may subsequently be able to determine whether the device is well suited for eliciting a predictable, linear response from E. coli when used in combination with other devices.
Determining factors that impact linear system behavior in E. coli would also be of benefit to synthetic biology. Such knowledge may enable the construction of biological systems using superposition because guidelines for the conditions necessary that ensure linear system behavior would be available. In this study, we focused on device copy number as the perturbing factor. Promoter strength is another important factor (as are RBS strength, gene length, codon usage, and product function). A library of constitutive promoters has been characterized using the cat and gfp genes . By expressing cat and gfp in the manner done in that study, the model system constructed in this work can be used to investigate the effect promoter strength has on linear system behavior. Our results appear to suggest that the identity of a device's promoter is not the only factor that impacts linearity in its RNA expression profile. While gfp had a promoter identical to that of cat, the former was the most nonlinear in our three-device model system (Figure 5). In fact, cat and nptII had similar DTCs in our three-device model system, but yet had different promoters. Comparison of the cat DTCs in our two-device (Figure 4D) and three-device (Figure 5C) constructs also supports this premise. The cat DTC was primarily linear for both cases. The cat device in one experiment, however, had its native promoter (Figure 4D), while the PL promoter was used in the other (Figure 5C). Comparison of the DTC results of our two-device (Figures 4C and 4D) and three-device (Figures 5B and 5C) constructs also suggest that the amount of nonlinearity in the E. coli system response to devices harboring nptII and cat genes is not impacted greatly by slow-growth conditions. Our results from the backbone, two-device, and three-device constructs also suggest that the DTC slope (which is the transfer curve gain) may act as a useful metric for characterizing promoter strength of a gene. The gain of the native nptII promoter was found to be ~1 in the various constructs tested (Figures 4B, 4C, and 5B), with similar values having been found irrespective of the choice of growth medium (Figures 4B) or a change in growth rate (Figures 4C and 5B).
Small-signal linearization techniques may also have application to other aspects of biological system behavior. Input-output relationships can be defined and experimentally measured to generate transfer curves, where piecewise linear models may subsequently be employed to determine the linear range of the system. Examples could include inducer concentration to activated transcription factor, activated transcription factor to RNA, and RNA to protein transfer curves. Measuring input-output characteristics and applying small-signal linearization techniques have the potential of reducing the complex mathematical equations used to model biological interactions to their simplest form; thereby, permitting predictable, quantitative behavior predictions. The limitation of small-signal linearization techniques is that the linearity property needs to be checked. As was observed in our model system, however, some systems can have a linear regime. So long as experiments are performed within this regime, one can avoid nonlinear effects and apply the simplifications associated with a small-signal linear model. And even if the system is to be operated in the nonlinear regime, it may be possible to introduce nonlinear correction factors to the obtained small-signal linear model. In Equation (2), for instance, this can be modeled by having the magnitudes of α and/or β dependent on the copy number (as opposed to constant values). With our results suggesting that the transfer curve and small-signal concepts used in electrical engineering can likewise be employed towards biological systems, the application of other concepts may also be of benefit to synthetic biology. The transfer curve concept is primarily of use in studying the steady-state behavior of a system. Linear systems can also be studied in the frequency domain by using Fourier techniques, which enable engineers to predict time-domain system response. These techniques have been used previously to study the yeast osmo-adaptation system . Indeed, the application of analysis and design techniques of other established engineering disciplines may enable the systematic forward engineering of biological systems for improved biotechnology applications.
We have presented a model system and framework to investigate linear system behavior in E. coli. With all three genetic devices present in the model system, we show the existence of superposition at the pSC101 copy number level. In the absence of the gfp device, linear system behavior was present even with a pUC replicon. The amount of nonlinearity in our model system appears to be biased towards the gfp device. This is in spite of the fact that the gfp and cat devices have identical constitutive promoters. Such a finding suggests additional factors besides promoter strength impact the amount of nonlinearity in a device's steady-state RNA expression profile. Our developed DTC method may have application in the systematic testing of device nonlinearity to determine whether a device will give a predictable output when used in combination with other devices. This, in turn, may enable the design and construction of biological systems with predictable, quantitative behavior from smaller elements characterized in isolation.
Bacterial strains, media, and enzymes
E. coli DH10B and DH5α were used for cloning. E. coli DH1 was used for expression work. Luria-Bertani (LB) media was made as described in . M9 minimal media + 0.5% glucose supplemented with micronutrients was made as described in . Restriction enzymes and T4 DNA ligase were purchased from New England Biolabs, with digestion and ligation reactions performed as recommended by the enzyme manufacturer. PCR reactions were performed with Phusion polymerase from Finnzymes, and the primers used were synthesized by Integrated DNA Technologies, Inc. The composition of the PCR reactions, cycle times, and temperatures followed those suggested by the enzyme manufacturer. PCR products were sequenced once cloned into the respective plasmids to ensure that no mutations had been introduced during the amplification process. In cases where single digest cloning was performed, sequencing was also used to select for constructs with inserts in the desired orientation.
The plasmid backbones with the replicons of pSC101, p15A, pMB1, and pUC (Figure 3A) were constructed using standard molecular biology techniques . These plasmids were named pAmin81 [GenBank:HQ283398], pAmin78 [GenBank:HQ283399], pAmin79 [GenBank:HQ283400], and pAmin80 [GenBank:HQ283401], respectively.
List of PCR primers used in the cloning of the plasmids constructed in this study
The construction of the model system series of plasmids (Figure 3C) proceeded by first creating a series of cat device (with the PL promoter) containing plasmids. PCR was used to obtain the cat open-reading frame as insert. The primer pairs used were cat_orf_F and cat_orf_R (Table 3), with pACYC184 having served as template. This insert was digested with KpnI, MluI, and ligated into pZE21  to create pAmin92. PCR was subsequently used to obtain the cat gene (with the PL promoter) as insert using the primer pairs PL_F and cat_wt_R (Table 3) and pAmin92 as template. This insert was digested with AatII, StuI, and ligated into the two-device (Figure 3B) series of plasmids to yield a series of cat device (with the PL promoter) containing constructs. Next, work began on constructing a gfp device (with the PL promoter) containing construct with pSC101 as the origin. This plasmid was called pAmin81+gfpPL. PCR was used to obtain a spacer (different in sequence from that above) and the cat gene as inserts. Primer pairs lacZ_2_F and lacZ_2_R (Table 3) were used for the spacer, and p50 gl served as the template. The cat gene was obtained by using the primers cat_wt_F and cat_wt_R (Table 3), with pACYC184 having served as template. The spacer here was to create spatial separation between the neighboring cat and gfp devices, and consisted of a ~600 bp fragment taken from within the coding region of the bacterial lacZ gene. The spacer and cat inserts were digested with AvrII, AatII and AatII, SacI, respectively, and ligated into a AvrII, SacI digested pAmin81 in a three-fragment ligation reaction to create pAmin93. The cat genetic device (complete with spacer and terminator) was subsequently transferred into pAmin81 to create pAmin99. PCR was next used to obtain the gfp open-reading frame as insert. Primer pairs gfp_F and gfp_R (Table 3) were used, with BBa_E0044  serving as the template. This insert was digested with KpnI, HindIII, and ligated into pZE21 to create pAmin100. PCR was then used to obtain the gfp gene (with PL promoter) using the primer pairs PL_F and gfp_2_R (Table 3) and pAmin100 as template. The creation of pAmin81+gfpPL subsequently proceeded by performing a three-fragment ligation reaction of this fragment digested with AatII, SacI, the ~2.5 kb fragment released from a AvrII, SacI digested pAmin93, and the ~2 kb fragment released from a AvrII, AatII digested pAmin99. The gfp genetic device (complete with spacer and terminator) was finally transferred from pAmin81+gfpPL into the cat device (with the PL promoter) containing series of plasmids described prior using BamHI, creating the desired series of plasmids (Figure 3C). Sub-cloning was used in order to arrive at the gfp device (with the PL promoter) containing constructs with the other three origins of replication. More specifically, the origins released from a AvrII, SmaI digested pAmin78, pAmin79, and pAmin80 were ligated into pAmin81+gfpPL to yield pAmin78+gfpPL, pAmin79+gfpPL, and pAmin80+gfpPL.
Bacterial growth conditions
E. coli DH1 cells were grown overnight at 30°C, 200 rpm shaking after inoculating 5 mL cultures of LB media (supplemented with 50 μg/mL kanamycin) with single colonies from freshly streaked plates. After sub-culturing (1:50) into shake flasks containing 50 mL of either M9 minimal or LB media (supplemented with 50 μg/mL kanamycin), cells were grown at 30°C, 200 rpm shaking until an OD600 nm of 0.3-0.4 was reached to approximate steady-state conditions. At this time, 1 mL of cells were added to ice chilled tubes with 100 μL of 10% phenol:90% EtOH stop solution , mixed, spun down, supernatant removed, and total RNA isolation proceeded immediately thereafter. Another 1 mL of cells were spun down, supernatant removed, and cell pellets subsequently frozen for total DNA isolation at a future date.
Bacterial total RNA isolation to quantify nptII, cat, and gfp expression levels
Bacterial cell pellets were resuspended in 700 μL buffer RLT (Qiagen), to which beta-mercaptoethanol had been added according to the manufacturer's instructions. Cells were subsequently lysed using 0.1 mm diameter glass beads in the Mini-Beadbeater-8 (Biospec). Following lysis, tube contents were spun down and 500 μL of lysate was transferred to new tubes. Total RNA extraction then proceeded by adding 500 μL of 25:24:1 phenol:chloroform:isoamyl alcohol, vortexing vigorously for ~1 min, allowing the tubes to sit at bench for a few minutes subsequent, and centrifugation for 15 min at 12000 × g, 4°C. Next, 300 μL of the upper aqueous phase was transferred to a new tube containing 300 μL nuclease free water. RNA extraction continued by adding 600 μL of chloroform to each tube, vigorous vortexing for ~1 min, allowing the tubes to sit at bench for a few minutes subsequent, and centrifugation for 15 min at 12000 × g, 4°C. Next, 300 μL of the upper aqueous phase was transferred to a new tube. Following chloroform extraction, total RNA was ethanol precipitated overnight, washed with 70% ethanol, and finally resuspended in 30 μL of nuclease free water. RNA concentration and purity were assayed using a Nanodrop spectrophotometer, and integrity examined on 2% agarose gels.
cDNA synthesis and real-time qPCR quantification of cellular nptII, cat, and gfp transcript levels
List of real-time qPCR primers used in this study
Bacterial total DNA isolation to quantify plasmid copy number
The DNA isolation method reported in the previous publications [24, 46] was adopted. Bacterial cell pellets were resuspended in 400 μL of 50 mM Tris/50 mM EDTA, pH 8, by vortex. Cell membranes were permealized by the addition of 8 μL of 50 mg/mL lysozyme (Sigma) in 10 mM Tris/1 mM EDTA, pH 8, followed by incubation at 37°C for 30 min. To complete cell lysis, 4 μL of 10% SDS and 8 μL of 20 mg/mL Proteinase K solution (Invitrogen) were added to each tube, mixed with a syringe with 21 gauge 1.5 inch needle, and incubated at 50°C for 30 min. Proteinase K was subsequently heat inactivated at 75°C for 10 min, and RNA was digested with the addition of 2 μL of 100 mg/mL RNase A solution (Qiagen) followed by incubation at 37°C for 30 min. Total DNA extraction then proceeded by adding 425 μL of 25:24:1 phenol:chloroform:isoamyl alcohol, vortexing vigorously for ~1 min, allowing the tubes to sit at bench for a few minutes subsequent, and centrifugation for 5 min at 12000 × g, 4°C. Next, 300 μL of the upper aqueous phase was transferred to a new tube using a wide-opening pipet tip. DNA extraction continued by adding 400 μL of chloroform to each tube, vigorous vortexing for ~1 min, allowing the tubes to sit at bench for a few minutes subsequent, and centrifugation for 5 min at 12000 × g, 4°C. Next, 200 μL of the upper aqueous phase was transferred to a new tube using a wide-opening pipet tip. Following chloroform extraction, total DNA was ethanol precipitated overnight, washed with 70% ethanol, and finally resuspended in 40 μL of nuclease free water. DNA concentration and purity were assayed using a Nanodrop spectrophotometer, and integrity examined on 1% agarose gels.
Real-time qPCR quantification of plasmid copy number
Primer sets specific to the nptII and 16S rDNA genes were used (Table 4). These primers amplified a single product of the expected size as confirmed by the melting temperatures of the amplicons. The nptII gene is a single-copy gene of the plasmids characterized in this study, while 16S is a multi-copy gene of E. coli chromosomal DNA  and was used for normalization purposes [24, 26]. Total DNA isolated from each strain was first digested overnight using EcoRI (New England Biolabs) at 37°C. Real-time qPCR was conducted on a BioRad iCycler with 96-well reaction blocks in the presence of SYBR Green under the following conditions: 1X iQ SYBR Green Supermix (BioRad), 150 nM nptII, or 500 nM 16S primers in a 25 μL reaction. Real-time qPCR cycling was 95°C for 3 min, followed by 40 cycles of 30 sec at 95°C, 30 sec at 60°C, and 30 sec at 72°C. Threshold cycles (Ct) were determined with iCycler (BioRad) software for all samples. A standard curve was prepared for quantification. For this purpose, a fourfold dilution series of a total of seven dilutions was prepared from a digested total DNA sample, and each dilution was subjected to qPCR analysis in triplicate with either the nptII- or 16S-specific primers. Obtained Ct values were used by the iCycler software package to plot a standard curve that allowed quantification of nptII or 16S in the digested total DNA samples (i.e. unknowns) relative to the DNA sample used to prepare the standard curve.
List of abbreviations used
- cat :
chloramphenicol acetyl transferase
device transfer curve
- gfp :
green fluorescent protein
- nptII :
neomycin phosphotransferase II
ribosome binding site.
The authors thank Murali Raghavendra Rao for helpful discussions concerning SAS and regression analysis. This research was supported by a National Science Foundation Graduate Research Fellowship (to MH); the Synthetic Biology Engineering Research Center (SynBERC), through a grant from the National Science Foundation; and the Joint BioEnergy Institute (JBEI), through contract DE-AC02-05CH11231 between Lawrence Berkeley National Laboratory and the U. S. Department of Energy, Office of Science, Office of Biological and Environmental Research.
- Keasling JD: Synthetic biology for synthetic chemistry. ACS Chem Biol 2008, 3: 64-76. 10.1021/cb7002434View ArticleGoogle Scholar
- Andrianantoandro E, Basu S, Karig DK, Weiss R: Synthetic biology: new engineering rules for an emerging discipline. Mol Syst Biol 2006, 2: E1-E14.View ArticleGoogle Scholar
- Endy D: Foundations for engineering biology. Nature 2005, 438: 449-453. 10.1038/nature04342View ArticleGoogle Scholar
- Hartwell LH, Hopfield JJ, Leibler S, Murray AW: From molecular to modular cell biology. Nature 1999, 402: C47-C52. 10.1038/35011540View ArticleGoogle Scholar
- Registry of Standard Biological Parts[http://partsregistry.org]
- Shetty RP, Endy D, Knight TF Jr: Engineering BioBrick vectors from BioBrick parts. J Biol Eng 2008, 2: 5. 10.1186/1754-1611-2-5View ArticleGoogle Scholar
- Anderson JC, Dueber JE, Leguia M, Wu GC, Goler JA, Arkin AP, Keasling JD: BglBricks: A flexible standard for biological part assembly. J Biol Eng 2010, 4: 1. 10.1186/1754-1611-4-1View ArticleGoogle Scholar
- Sauro HM: Modularity defined. Mol Syst Biol 2008, 4: 166.View ArticleGoogle Scholar
- Del Vecchio D, Ninfa AJ, Sontag ED: Modular cell biology: retroactivity and insulation. Mol Syst Biol 2008, 4: 161.View ArticleGoogle Scholar
- Agapakis CM, Silver PA: Modular electron transfer circuits for synthetic biology. Bioengineered Bugs 2010, 1: 6.View ArticleGoogle Scholar
- Arkin A: Setting the standard in synthetic biology. Nat Biotechnol 2008, 26: 771-774. 10.1038/nbt0708-771View ArticleGoogle Scholar
- Canton B, Labno A, Endy D: Refinement and standardization of synthetic biological parts and devices. Nat Biotechnol 2008, 26: 787-793. 10.1038/nbt1413View ArticleGoogle Scholar
- Benner SA, Sismour AM: Synthetic biology. Nat Genetics 2005, 6: 533-543.View ArticleGoogle Scholar
- Blattner FR, Plunkett G, Bloch CA, Perna NT, Burland V, Riley M, Collado-Vides J, Glasner JD, Rode CK, Mayhew GF, Gregor J, Davis NW, Kirkpatrick HA, Goeden MA, Rose DJ, Mau B, Shao Y: The complete genome sequence of Escherichia coli K-12. Science 1997, 277: 1453-1462. 10.1126/science.277.5331.1453View ArticleGoogle Scholar
- Lathi BP: Linear systems and signals. Carmichael: Berkeley-Cambridge Press; 1992.Google Scholar
- Gray PR, Hurst PJ, Lewis SH, Meyer RG: Analysis and design of analog integrated circuits. 4th edition. New York: John Wiley & Sons, Inc; 2001.Google Scholar
- Rabaey JM, Chandrakasan A, Nicolic B: Digital integrated circuits. 2nd edition. New Jersey: Prentice Hall; 2003.Google Scholar
- Kües U, Stahl U: Replication of plasmids in gram-negative bacteria. Micriobiol Rev 1989, 53: 491-516.Google Scholar
- Selzer G, Som T, Itoh T, Tomizawa J: The origin of replication of plasmid p15A and comparative studies on the nucleotide sequences around the origin of related plasmids. Cell 1983, 32: 119-129. 10.1016/0092-8674(83)90502-0View ArticleGoogle Scholar
- Scholtissek S, Grosse F: A cloning cartridge of lambda t(o) terminator. Nucleic Acids Res 1987, 15: 3185. 10.1093/nar/15.7.3185View ArticleGoogle Scholar
- Orosz A, Boros I, Venetianer P: Analysis of the complex transcription termination region of the Escherichia coli rrnB gene. Eur J Biochem 1991, 201: 653-659. 10.1111/j.1432-1033.1991.tb16326.xView ArticleGoogle Scholar
- Peterson J, Phillips GJ: New pSC101-derivative cloning vectors with elevated copy numbers. Plasmid 2008, 59: 193-201. 10.1016/j.plasmid.2008.01.004View ArticleGoogle Scholar
- Chang AC, Cohen SN: Construction and characterization of amplifiable multicopy DNA cloning vehicles derived from the P15A cryptic miniplasmid. J Bacteriol 1978, 134: 1141-1156.Google Scholar
- Lin-Chao S, Chen WT, Wong TT: High copy number of the pUC plasmid results from a Rom/Rop-suppressible point mutation in RNA II. Mol Microbiol 1992, 6: 3385-3393. 10.1111/j.1365-2958.1992.tb02206.xView ArticleGoogle Scholar
- Lee CL, Ow DS, Oh SK: Quantitative real-time polymerase chain reaction for determination of plasmid copy number in bacteria. J Microbiol Methods 2006, 65: 258-267. 10.1016/j.mimet.2005.07.019View ArticleGoogle Scholar
- Lee C, Kim J, Shin SG, Hwang S: Absolute and relative QPCR quantification of plasmid copy number in Escherichia coli. J Biotechnol 2006, 123: 273-280. 10.1016/j.jbiotec.2005.11.014View ArticleGoogle Scholar
- Tao L, Jackson RE, Cheng Q: Directed evolution of copy number of a broad host range plasmid for metabolic engineering. Metab Eng 2005, 7: 10-17. 10.1016/j.ymben.2004.05.006View ArticleGoogle Scholar
- Bustin SA: Absolute quantification of mRNA using real-time reverse transcription polymerase chain reaction assays. J Mol Endocrinol 2000, 25: 169-193. 10.1677/jme.0.0250169View ArticleGoogle Scholar
- Giulietti A, Overbergh L, Valckx D, Decallonne B, Bouillon R, Mathieu C: An overview of real-time quantitative PCR: applications to quantify cytokine gene expression. Methods 2001, 25: 386-401. 10.1006/meth.2001.1261View ArticleGoogle Scholar
- Wong ML, Medrano JF: Real-time PCR for mRNA quantification. Biotechniques 2005, 39: 75-85. 10.2144/05391RV01View ArticleGoogle Scholar
- Alper H, Fischer C, Nevoigt E, Stephanopoulos G: Tuning genetic control through promoter engineering. Proc Natl Acad Sci USA 2005, 102: 12678-12683. 10.1073/pnas.0504604102View ArticleGoogle Scholar
- Grabherr R, Bayer K: Impact of targeted vector design on ColE1 plasmid replication. Trends Biotechnol 2002, 20: 257-260. 10.1016/S0167-7799(02)01950-9View ArticleGoogle Scholar
- Weigandt TC: Low-phase-noise, low-timing-jitter design techniques for delay cell based VCOs and frequency synthesizers. In Ph.D. thesis. University of California, Berkeley, Electrical Engineering Department; 1998.Google Scholar
- Robbins KR, Saxton AM, Southern LL: Estimation of nutrient requirements using broken-line regression analysis. J Anim Sci 2006, 84: E155-E165.Google Scholar
- Bailey JE, Ollis DF: Biochemical engineering fundamentals. 2nd edition. New York: McGraw-Hill; 1986.Google Scholar
- Vogel U, Jensen KF: Effects of guanosine 3',5'-bisdiphosphate (ppGpp) on rate of transcription elongation in isoleucine-starved Escherichia coli. J Biol Chem 1994, 269: 16236-16241.Google Scholar
- Krohn M, Wagner R: Transcriptional pausing of RNA polymerase in the presence of guanosine tetraphosphate depends on the promoter and gene sequence. J Biol Chem 1996, 271: 23884-23894. 10.1074/jbc.271.39.23884View ArticleGoogle Scholar
- Proshkin S, Rahmouni AR, Mironov A, Nudler E: Cooperation between translating ribosomes and RNA polymerase in transcription elongation. Science 2010, 328: 504-508. 10.1126/science.1184939View ArticleGoogle Scholar
- Burmann BM, Schweimer K, Luo X, Wahl MC, Stitt BL, Gottesman ME, Rösch P: A NusE:NusG complex links transcription and translation. Science 2010, 328: 501-504. 10.1126/science.1184953View ArticleGoogle Scholar
- Mettetal JT, Muzzey K, Gómez-Uribe C, van Oudenaarden A: The frequency dependence of osmo-adaptation in Saccharomyces cerevisiae. Science 2008, 319: 482-484. 10.1126/science.1151582View ArticleGoogle Scholar
- Sambrook J, Fritsch EF, Maniatis T: Molecular cloning: a laboratory manual. 2nd edition. New York: Cold Spring Harbor Laboratory Press; 1989.Google Scholar
- Neidhardt FC, Bloch PL, Smith DF: Culture medium for enterobacteria. J Bacteriol 1974, 119: 736-747.Google Scholar
- Smolke CD, Carrier TA, Keasling JD: Coordinated, differential expression of two genes through directed mRNA cleavage and stabilization by secondary structures. Appl Environ Microbiol 2000, 66: 5399-5405. 10.1128/AEM.66.12.5399-5405.2000View ArticleGoogle Scholar
- Lutz R, Bujard H: Independent and tight regulation of transcriptional units in Escherichia coli via the LacR/O, the TetR/O and AraC/I1-I2 regulatory elements. Nucleic Acids Res 1997, 25: 1203-1210. 10.1093/nar/25.6.1203View ArticleGoogle Scholar
- Bernstein JA, Khodursky AB, Lin PH, Lin-Chao S, Cohen SN: Global analysis of mRNA decay and abundance in Escherichia coli at single-gene resolution using two-color fluorescent DNA microarrays. Proc Natl Acad Sci USA 2002, 99: 9697-9702. 10.1073/pnas.112318199View ArticleGoogle Scholar
- Pushnova EA, Geier M, Zhu YS: An easy and accurate agarose gel assay for quantification of bacterial plasmid copy numbers. Anal Biochem 2000, 284: 70-76. 10.1006/abio.2000.4668View ArticleGoogle Scholar
- Lee C, Lee S, Shin SG, Hwang S: Real-time PCR determination of rRNA gene copy number: absolute and relative quantification assays with Escherichia coli. Appl Microbiol Biotechnol 2008, 78: 371-376. 10.1007/s00253-007-1300-6View ArticleGoogle Scholar
- Mason G, Provero P, Vaira AM, Accotto GP: Estimating the number of integrations in transformed plants by quantitative real-time PCR. BMC Biotechnol 2002, 2: 20. 10.1186/1472-6750-2-20View ArticleGoogle Scholar
- Olivares EC, Hollis RP, Chalberg TW, Meuse L, Kay MA, Calos MP: Site-specific genomic integration produces therapeutic Factor IX levels in mice. Nat Biotechnol 2002, 20: 1124-1128. 10.1038/nbt753View ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.