A modular and extensible RNA-based gene-regulatory platform for engineering cellular function

Win et al. 10.1073/pnas.0703961104.

Supporting Information

Files in this Data Supplement:

SI Text
SI Figure 7
SI Figure 8
SI Figure 9
SI Figure 10
SI Table 1
SI Figure 11
SI Figure 12
SI Figure 13
SI Table 2
SI Figure 14
SI Figure 15
SI Figure 16
SI Table 3




SI Figure 7

Fig. 7. Control constructs supporting the design strategy for engineering ligand-regulated ribozyme switches. (A-C) Color schemes are as follows: catalytic core, purple; aptamer sequences, brown; loop sequences, blue; brown arrow, cleavage site. (A) Sequences of the ribozyme satellite RNA of tobacco ringspot virus control (sTRSV Contl) and stem integration control (hhRz I). (B) Sequences of the loop sequence controls in which the loop I and II sequences were replaced by the theophylline aptamer (L1R and L2R, respectively). (C) Sequences of the loop sequence controls in which the theophylline aptamer is connected directly to the loop I nucleotides through L1.3 and L1.4 (L1Theo) and the loop II nucleotides through L2.2 and L2.3 (L2Theo). (D) Gene expression levels (in fold) of the control constructs; 1-fold is defined as the reporter gene expression level of sTRSV relative to that of the background fluorescence level. The mean ± SD from at least three independent experiments is shown.

SI Figure 8

Fig. 8. Flow cytometry histograms of L2bulge1, L2bulgeOff1, and the ribozyme control cell populations grown in the presence (+) or absence (-) of 5 mM theophylline. Red line, cell populations grown in the absence of theophylline; green line, cell populations grown in 5 mM theophylline; shaded population, cell populations indicative of the noninduced cell population, shaded here to indicate the portion of cells in the population that have lost the plasmid and exhibit noninduced, or background, levels of autofluorescence. Histograms are representative of three independent experiments.

SI Figure 9

Fig. 9. Flow cytometry histograms of the helix-slipping-based ribozyme switch cell populations grown in the presence (+) or absence (-) of 5 mM theophylline. Population data were measured and are reported as described for SI Fig. 8. Histograms are representative of three independent experiments.

SI Figure 10

Fig. 10. Regulatory properties of the helix-slipping information transmission mechanism. The theophylline-dependent gene-regulatory behavior of L2cm4 and L1cm10. Gene expression levels are reported as described for Fig. 2, except that L1Theo was used as a nonswitch control for L1cm10.

SI Figure 11

Fig. 11. Temporal responses of L2bulge1, L1cm10, and L2cm4 in response to the addition of 5 mM theophylline (final concentration). The time point at which theophylline was added to the cultures is indicated by an arrow. Brown, 5 mM theophylline added to growing cultures; gray, no theophylline added to growing cultures. Gene expression levels are reported as relative fluorescence units (RFUs)/OD by dividing fluorescence units by the OD₆₀₀ of the cell sample and subtracting the background fluorescence level. L2bulge1 exhibits up-regulation of GFP levels in response to the addition of theophylline; L1cm10 and L2cm4 exhibit down-regulation of GFP levels in response to theophylline addition. The mean ± SD from at least three independent experiments is shown for all graphs.

SI Figure 12

Fig. 12. Sequences and structures of tuned ribozyme switches in the L2bulge and L2bulgeOff series. The nucleotides altered from the parent constructs, L2bulge1 and L2bulgeOff1, are highlighted. The two stable equilibrium conformations, ribozyme-active and -inactive conformations, are indicated for the parent ribozyme switches. The ribozyme-active conformations of L2bulge2-L2bulge5 are not shown because they are similar to that of L2bulge1. L2bulge6 and L2bulge7 each assume a single predominant conformation, ribozyme-inactive and ribozyme-active, respectively, and do not undergo theophylline-induced conformational switching. L2bulge8 and L2bulge9, modified from L2bulge7 by reducing the stability of the ribozyme-active conformation and the energy difference between the two conformations of L2bulge7, now become capable of switching. For these two modified switch constructs, only the ribozyme-active conformations are shown, because their ribozyme-inactive conformations are similar to those of the other switches illustrated. The ribozyme-inactive conformations of L2bulgeOff2 and L2bulgeOff3 are not shown because they are similar to that of L2bulgeOff1.

SI Figure 13

Fig. 13. Dynamic ranges of regulation of the ribozyme switches and controls engineered in this work. The regulatory effects at 5 mM theophylline are reported on a full transcriptional range spectrum scale without normalization to the corresponding base expression level of each switch in the absence of effector (0 mM). Little or no effector-mediated gene-regulatory effect was observed in the nonswitch control constructs. Gene expression fold is defined here as previously, where 1-fold is equivalent to the reporter gene expression level of sTRSV relative to the background fluorescence level. sTRSV is the most active ribozyme construct exhibiting the lowest gene expression level, and sTRSV Contl is the most inactive ribozyme construct exhibiting the highest gene expression level, providing a 50-fold range as the full spectrum. Arrows indicate the direction of regulation as a function of increasing concentration of theophylline. These switches offer diverse dynamic ranges of regulation and thus provide a broader utility to fit specific applications of interest. Data are reported from three independent experiments and the mean ±SD is the same as that reported in Figs. 1-6.

SI Figure 14

Fig. 14. Flow cytometry histograms of the tuned ribozyme switch series cell populations grown in the presence (+) or absence (-) of 5 mM theophylline. Population data were measured and are reported as described for SI Fig. 8. Histograms are representative of three independent experiments.

SI Figure 15

Fig. 15. Demonstration of theophylline-regulated cell growth by ribozyme switches through plate-based assays. Cells harboring ribozyme switches and control constructs were streaked on two plates containing the same medium except different effector concentrations (0 mM versus 5 mM theophylline). OFF switches (L1cm10, L2cm4, L2cm1, and L2bulgeOff1) exhibited suppressed cell growth on the plate containing 5 mM theophylline, whereas an ON switch (L2bulge8) exhibited a higher growth level on the plate containing 5 mM theophylline. The control constructs (L1Theo, L2Theo, sTRSV Contl, and sTRSV) exhibited similar growth levels on both plates. sTRSV exhibited no cell growth because of its efficient cleavage activity, and sTRSV Contl exhibited the highest levels of growth because of its lack of cleavage activity.

SI Figure 16

Fig. 16. Detection of intracellular accumulation of the substrate xanthosine and the product xanthine over three different time points. Accumulation of xanthosine was observed at earlier time points. Conversion of xanthosine to xanthine was detected at 24 h after substrate feeding, and a higher accumulation of xanthine was detected at 48 h after substrate feeding.

Table 1. Relative steady-state ribozyme switch and ribozyme control transcript levels in the presence or absence of theophylline

Quantitative RT-PCR (qRT-PCR) analysis was performed on L2bulge1, L1cm10, L2cm4, satellite RNA of tobacco ringspot virus (sTRSV), and sTRSV control (sTRSV Contl). Transcript levels in the presence or absence of theophylline are reported as fractions relative to those of sTRSV Contl. L2bulge1 exhibited a higher steady-state level of target transcript, whereas L1cm10 and L2cm4 exhibited lower steady-state target transcript levels in the presence of 5 mM theophylline than in the absence of theophylline. The ribozyme controls, sTRSV and sTRSV Contl, exhibited little effect on steady-state transcript levels due to the presence of theophylline. In addition, relative steady-state levels of these switches corresponded to the relative GFP expression levels as determined through the functional ribozyme switch assays. All data are reported from three independent experiments.

Table 2. Free energies (-ΔG, kcal/mol) of individual conformations (ribozyme-active and -inactive) and the energy difference (ΔΔG, kcal/mol) between the free energies of these two conformations predicted by RNAstructure4.2

SI Text

Glossary of Terms

Actuator domain: A switch domain that encodes the system control function. As used here, the actuator domain encodes the gene-regulatory function and is comprised of a hammerhead ribozyme sequence.

Communication module: A sequence element that typically forms an imperfectly paired double-stranded stem that can adopt different base pairs between nucleotides through a "slip-structure" mechanism. As used here, a communication module is a type of information transmission domain that transmits the binding state of the aptamer domain to the adjacent actuator domain through a helix-slipping mechanism. As demonstrated in this work, a communication module does not act in a modular fashion with other switch domains. The term is retained here from earlier work in the field of nucleic acid engineering.

Competing strand: The nucleic acid sequence within a strand-displacement domain that is bound to the general transmission region of the switch when the sensor domain is in the restored conformation (i.e., in the presence of ligand). The competing strand competes for binding with the switching strand, which is initially bound to this transmission region in the absence of ligand.

Component: A part of a system that encodes a distinct activity or function.

Composability: A property of a system that indicates its ability to be comprised of components that can be selected and assembled in a modular fashion to achieve a desired system performance. As used here, composability refers to the ability of the individual domains of the control system to be modularly linked without disrupting their activities.

Engineering design principle: A required property of a constructed system that enables use by others.

Framework: A basic conceptual structure that is used to solve a complex product design issue. As used here, the framework is used to reliably design and construct specific instances of RNA switches. The conceptual structure of our framework is comprised of specified engineering design principles and design strategies that enable extensible and reusable system design.

Helix-slipping domain: A subset of information transmission domains that act through a helix-slipping mechanism. The helix-slipping domain is also referred to as the communication module.

Helix-slipping mechanism: An information transmission mechanism that is based on an information transmission domain that functions through a helix-slipping event and does not allow for rational design. Such a helix-slipping event uses a communication module (or helix-slipping domain) within the general transmission region of the switch (the base stem of the aptamer) to result in disruption or restoration of the actuator domain in response to restoration of the sensor domain.

Information transmission domain: A switch domain that encodes the function of transmitting information between the sensor domain and the actuator domain.

Information transmission mechanism: A general mechanism for transmitting information between the sensor domain and the actuator domain of a switch. As used here, this mechanism regulates the activity of the actuator domain in response to the binding state of the sensor domain.

Modular: A property of a system comprised of modules that indicates whether the modules can by interchanged as parts without changing the interface between modules or the modules themselves.

Module: A self-contained system component that has a well defined interface with other system components.

Platform: A general framework on which specific applications can be implemented. As used here, the platform enables specific instances of switches to be built in a standardized manner.

Portability: A property of a system that indicates its ability to be implemented in environments different from that in which it was originally designed. As used here, portability refers to the ability of the control system to be implemented in different organisms.

Reliability: A property of a system that indicates its ability to perform and maintain its functions under a set of specified conditions. As used here, reliability refers to the ability of the information transmission domain to standardize the transmission of information between the sensor and actuator domains.

Scalability: A property of a system that indicates its ability to handle increasing work. As used here, scalability refers to the ability of the control system to be implemented across broad application space by being able to forward design its response to different molecular information.

Switch: A molecule that can adopt at least two different conformational states, where each state is associated with a different activity of the molecule. Often a ligand can bind to one or more conformations of the switch, such that the presence of the ligand shifts the equilibrium distribution across the adoptable conformations and therefore regulates the activity of the switch molecule. As used here, switch refers to an RNA molecule that can adopt different structures that correspond to different gene regulatory activities. An RNA switch is then a ligand-controlled gene-regulatory system.

Switch domain: A component of a switch that encodes a distinct activity or function.

Switching strand: The nucleic acid sequence within a strand displacement domain that is bound to the general transmission region of the switch when the sensor domain is in the disrupted conformation (i.e., in the absence of ligand). The switching strand is displaced by the competing strand in the presence of ligand.

Sensor domain: A switch domain that encodes a ligand-binding function. As used here, the sensor domain is comprised of an RNA aptamer sequence.

Strand-displacement domain: A subset of information transmission domains that act through a strand-displacement mechanism.

Strand-displacement mechanism: An information transmission mechanism that is based on the rational design of an information transmission domain that functions through a strand-displacement event. Such a strand-displacement event uses competitive binding of two nucleic acid sequences (the competing strand and the switching strand) to a general transmission region of the switch (the base stem of the aptamer) to result in disruption or restoration of the actuator domain in response to restoration of the sensor domain.

Universal: A system property that indicates its ability to maintain function across different applications, environments, and component interfaces. As used here, a universal system is composed of the five engineering design principles (scalability, portability, utility, composability, and reliability) and results in the specified extensible platform for RNA switch construction.

Utility: A property of a system that indicates its ability to be of practical use. As used here, utility refers to the ability of the control system to interface with different functional level components to enable forward design of the function that is being controlled by the system.

Ribozyme control constructs for loop sequence coupling and stem integration controls. To establish and make useful our design strategy we constructed and characterized a series of ribozyme controls. We characterized the regulatory activity of our ribozyme constructs within a modular ribozyme characterization system in the eukaryotic model organism Saccharomyces cerevisiae (Fig. 1A). First, an inactive ribozyme control, satellite RNA of tobacco ringspot virus control (sTRSV Contl), was constructed to adopt the same structural motif as satellite RNA of tobacco ringspot virus (sTRSV) (Fig. 1A) while carrying a scrambled catalytic core sequence [supporting information (SI) Fig. 7A]. Second, a synthetic sTRSV ribozyme (hhRz I) that contains closed loops in stems II and III and is embedded through stem I was constructed as a stem integration control (SI Fig. 7A). Finally, we constructed four loop sequence controls. In one set, stem loops I and II (L1R and L2R, respectively) were replaced by the theophylline aptamer TCT8-4 (1) (SI Fig. 7B); in another set, the theophylline aptamer was coupled directly to sequences in loops I and II (L1Theo and L2Theo, respectively) (SI Fig. 7C). sTRSV exhibited a 50-fold reduction in target expression levels relative to sTRSV Contl (SI Fig. 7D). hhRz I, L1R, and L2R exhibited target expression levels similar to that of sTRSV Contl, suggesting that ribozyme activity was abolished in these constructs. In contrast, L1Theo and L2Theo exhibited significantly lower target expression levels relative to sTRSV Contl. L1Theo and L2Theo were used as the primary base constructs in engineering our synthetic ribozyme switch platforms. In addition, scrambled core versions of L1Theo and L2Theo exhibited no theophylline-dependent shifts in gene expression (data not shown), indicating that theophylline binding in that region of the transcript alone was not responsible for the observed regulatory effects. Taken together, we find that our design strategy enables the construction of a universal ribozyme switch platform that satisfies the design principles of portability, utility, and composability.

Rational tuning strategies for strand-displacement-based switches. A series of nine tuned ON switches were constructed from L2bulge1 as a base structure by employing rational energetic tuning strategies developed in this work. This strategy was based on the effects of altering the predicted free energies of a particular conformation (-DG) and the predicted difference between the free energies of two conformations (DDG) on RNA conformational dynamics, or the ability of the RNA molecule to distribute between these two conformational states. SI Table 2 lists free energies (-DG) of ribozyme-active and -inactive conformations and the energy difference (DDG) between the free energies of these two conformations. Specifically, lowering values for either of these energetic measurements (-DG or DDG) is expected to make it easier for a particular RNA molecule to switch between the conformational states in question. Therefore, there is an anticipated optimum conformational energy and energetic difference between conformations to achieve the desired range of switching in response to effector concentration (i.e., energy measurements too high will result in stable nonswitch designs, and energy measurements or energy difference measurements too low will result in fairly equal distributions between the two conformational states and lower switching capabilities). It is also expected, then, that one can "push" switches into a nonswitch state by moving away from this energetic optimum. This strategy was examined in a series of tuning experiments, described below.

L2bulge2 and L2bulge3 (SI Fig. 12) replaced canonical base pairs in the aptamer base stem of the ribozyme-inactive conformation of L2bulge1 with U-G wobble pairs. As a result of these destabilizing alterations, both equilibrium conformations (ribozyme-active and -inactive) became less thermodynamically stable than those of L2bulge1, as estimated from their predicted free energies (-DG). In addition, the energy required to switch between the two equilibrium conformations was maintained similarly to that of L2bulge1, as estimated by the difference between the free energies of the two conformations (DDG). Ribozyme assays indicated that both L2bulge2 and L2bulge3 exhibited smaller dynamic ranges than that of L2bulge1 (Fig. 4B and SI Fig. 13). It is proposed that the lower stabilities of the conformational states enable more frequent dynamic switching between the two equilibrium conformations and therefore lower the difference in distribution, favoring one state over the other.

L2bulge4 (SI Fig. 12) incorporates an additional G-U wobble pair within the aptamer base stem of the ribozyme-inactive conformation of L2bulge1. However, this aptamer stem extension did not result in an appreciable, predicted change in the thermodynamic stabilities of the equilibrium conformations or the energy required to switch between the two equilibrium conformations when compared with L2bulge1. Ribozyme assays indicated that L2bulge4 exhibited a dynamic range in response to theophylline levels similar to that of L2bulge1 (Fig. 4B and SI Fig. 13).

L2bulge5 (SI Fig. 12) incorporates an additional canonical base pair (A-U) within the aptamer base stem of L2bulge1. As a result of this stabilizing alteration, the conformation of the ribozyme switch, in which the aptamer structure is formed and the catalytic core is disrupted (ribozyme-inactive), is increased in stability and is as stable as the conformation in which the catalytic core is not disrupted (ribozyme-active). The increased stability of the ribozyme-inactive conformation in L2bulge5, in comparison with L2bulge1 and L2bulge4, indicates that the equilibrium distribution between these two conformations will shift to favor the ribozyme-inactive conformation. Ribozyme assays indicated that L2bulge5 exhibited significantly higher GFP expression levels in the absence or presence of theophylline compared with those of L2bulge1 and L2bulge4, such that the theophylline-regulated increase in gene expression was similar to that of L2bulge3 but different in regulatory dynamic ranges (Fig. 4B and SI Fig. 13).

Two switches in this series, L2bulge6 and L2bulge7, were constructed to demonstrate the ability of this tuning strategy to push the ribozyme switch constructs out of a switchable energetic range and to approach nonswitching extremes. L2bulge6 (SI Fig. 12) was designed to energetically favor the conformation in which the aptamer structure is formed and the catalytic core is disrupted (ribozyme-inactive) in the absence of theophylline by introducing a stabilizing G-C base pair into the aptamer stem of this conformation. Because the aptamer conformation is expected to be favored in L2bulge6, the presence of theophylline is expected to have little or no effect on the conformational dynamics of this switch. L2bulge7 (SI Fig. 12) was designed to energetically favor the conformation in which the aptamer structure is not formed and the catalytic core is undisrupted (ribozyme-active) by introducing a stabilizing U-A base pair into the stem extending from loop II in this conformation. Because the stability of the ribozyme-active conformation is significantly higher than that of the ribozyme-inactive conformation for L2bulge7, the presence of theophylline is expected to have little effect on the conformational dynamics of this ribozyme switch. Ribozyme assays indicated that L2bulge7 exhibited very low GFP expression levels and L2bulge6 exhibited very high GFP expression levels in the presence or absence of theophylline (SI Fig. 13). As designed, both constructs exhibited little increase in target expression levels in response to theophylline by energetically favoring one of the two conformational states (Fig. 4B).

L2bulge 8 (SI Fig. 12) was modified from L2bulge7 by replacing the canonical base pair (U-A) with a wobble base pair (U-G), thereby reducing the stability of the ribozyme-active conformation of L2bulge7 and allowing it to adopt the ribozyme-inactive conformation. Similarly, L2bulge 9 (SI Fig. 12) was modified in such a way to reduce the energy difference between the two conformations of L2bulge7. Ribozyme assays indicated that L2bulge8 and L2bulge9 exhibited theophylline-dependent up-regulation of target gene expression in accordance with the reduced stabilities of the ribozyme-active conformations and energy differences between the two adoptable conformations for each of these switch constructs (Fig. 4B and SI Fig. 13).

In addition, a series of three tuned OFF switches were constructed by using rational energetic tuning strategies from L2bulgeOff1 as a base structure. L2bulgeOff2 and L2bulgeOff3 were constructed to demonstrate tunability of the OFF switch platform by using similar energetic design strategies (SI Fig. 12). These switch variants exhibited different theophylline-responsive dynamic ranges from that of L2bulgeOff1 (Fig. 4C and SI Fig. 13).

Flow cytometry analysis of the tuned ribozyme switch series demonstrated that the tuned switches exhibited corresponding shifts in the mean fluorescence of the cell populations in the presence or absence of theophylline (SI Fig. 14). The relative dynamic ranges of the switches across the full regulatory range bracketed by the ribozyme-active and -inactive controls, sTRSV and sTRSV Contl, respectively, are presented in SI Fig. 13.

Among the 12 tuned switches (both ON and OFF), the dynamic regulatory ranges of most of these switches are in agreement with our rational tuning strategies based on the -DG and DDG values predicted by RNAstructure 4.2. Two exceptions are noted: L2bulge9 and L2bulgeOff3. L2bulge9 exhibited a larger dynamic regulatory effect despite its higher DDG than L2bulge8. L2bulgeOff3 exhibited a smaller dynamic regulatory effect despite its smaller DDG than L2bulgeOff2. However, it is more difficult to make a direct comparison between L2bulgeOff2 and L2bulgeOff3, because both conformations of L2bulgeOff3 are significantly more stable than those of L2bulgeOff2, likely resulting in L2bulgeOff3 less frequently switching between its two conformations and thus enabling this molecule to get "trapped" in its lower free energy states. In addition, outliers may also have arisen because the RNAstructure program predicts these energy values based on the secondary structure of a particular conformation and does not take into consideration energy contributions from tertiary interactions [which have been observed previously (2)] in estimating these energies. Nevertheless, we demonstrate that energetic predictions based solely on secondary structure are useful for our rational tuning design strategies. The different dynamic regulatory ranges exhibited by our tuned switches in response to their specific effector (SI Fig. 13) validate that such response programming can be achieved by altering the nucleotide composition of the information transmission region within a switch, thereby demonstrating the interdependence among RNA sequence, structure, and function.

Materials and Methods

Plasmid and switch construction. The modular plasmid, pRzS, was constructed and used as a universal vector for the characterization of all ribozyme switches. The engineered ribozyme constructs were generated by PCR amplification using the appropriate oligonucleotide templates and primers. All oligonucleotides were synthesized by Integrated DNA Technologies (Coralville, IA). All engineered ribozyme constructs were cloned into two unique restriction sites, AvrII and XhoI, three nucleotides downstream of the yeast-enhanced GFP (yEGFP) stop codon and upstream of an ADH1 terminator. Cloned plasmids were transformed into an electrocompetent Escherichia coli strain, DH10B (Invitrogen, Carlsbad, CA), and all ribozyme constructs were confirmed by subsequent sequencing (Laragen, Los Angeles, CA). Confirmed plasmid constructs were transformed into S. cerevisiae strain W303 MATa his3-11,15 trp1-1 leu2-3 ura3-1 ade2-1 by using standard lithium acetate procedures (3).

Ribozyme characterization assays. S. cerevisiae cells harboring the appropriate plasmids were grown in synthetic complete medium supplemented with an appropriate dropout solution and sugar [2% (wt/vol) raffinose and 1% sucrose] overnight at 30°C. Overnight cultures were back-diluted into fresh medium to an optical density at 600 nm (OD₆₀₀) of »0.1 and grown at 30°C. An appropriate volume of concentrated effector stock (to the appropriate final concentration of theophylline or tetracycline) dissolved in medium or an equivalent volume of the medium (no effector control) was added to the cultures at the time of back-dilution. In addition, at this time an appropriate volume of galactose (2% final concentration) or an equivalent volume of water was added to the cultures for the induced and noninduced controls, respectively. For specificity assays, an appropriate volume of a concentrated caffeine or doxycycline stock (final concentrations of 1 mM and 250 mM, respectively) was added to a separate culture. Cells were grown to an OD₆₀₀ of 0.8-1.0 or for a period of »6 h before measuring GFP levels on a Safire fluorescent plate reader (Tecan, Männedorf, Switzerland) and/or on a Cell Lab Quanta SC flow cytometer (Beckman Coulter, Fullerton, CA). For temporal response assays, cultures were grown as described above in the absence of the appropriate effector, and fluorescence data were taken every 30 min. After 4 h of growth, appropriate volumes of the concentrated effector stock or plain medium were added to the cultures, and fluorescence was monitored for several hours thereafter.

Cell growth regulation assays. For liquid culture assays, S. cerevisiae cells carrying appropriate plasmids were back-diluted and grown according to procedures described above with minor modifications. A competitive inhibitor of the his5 gene product, 3-amino-triazole (3AT), was added to a final concentration of 5 mM to increase the threshold level of histidine required for cell growth. Cultures were grown in various theophylline concentrations, and the growth of each sample was monitored over a 24-h period. The theophylline-regulated growth at 24 h is reported in terms of OD₆₀₀ readings measured on the Safire plate reader. For plate-based assays, 10 ml of the back-diluted culture samples was streaked on plates containing 0 and 5 mM theophylline. A higher concentration of 3AT (25 mM) was used in the plate-based assays to optimize visual assessment of theophylline-regulated cell growth.

Metabolite sensing assays. S. cerevisiae cells carrying appropriate plasmids were back-diluted and grown according to procedures described above with minor modifications. Cultures were grown in the absence or presence of xanthosine (250 mM final concentration). To account for inducer depletion, galactose was added to the cultures at 8-h time intervals to a 2% final concentration. Fluorescence levels of the samples were monitored over a 48-h period according to procedures described above. For HPLC analysis, cell extracts were prepared after appropriate growth periods after xanthosine feeding by rapid freezing of cell cultures in liquid nitrogen in the form of beads. Frozen cell beads were subsequently lysed by grinding using a mortar and pestle followed by extraction with methanol. Intracellular metabolite levels were analyzed by using an HPLC system integrated with a mass spectrometer (HPLC-MS) (Agilent Technologies, Santa Clara, CA), which enables confirmation of metabolite peaks based on their corresponding molecular weights.

Fluorescence quantification. The population-averaged fluorescence of each sample was measured on the Safire fluorescence plate reader with the following settings: excitation wavelength of 485 nm, emission wavelength of 515 nm, and a gain of 100. Fluorescence readings were normalized to cell number by dividing fluorescence units by the OD₆₀₀ of the cell sample and subtracting the background fluorescence level to eliminate autofluorescence.

Fluorescence distributions within the cell populations were measured on a Quanta flow cytometer with the following settings: 488-nm laser line, 525-nm bandpass filter, and photomultiplier tube setting of 5.83. Fluorescence data were collected under low flow rates for »30,000 cells. Viable cells were selected and fluorescence levels were determined from 10,000 counts in this selected population. A noninduced cell population was used to set a "negative GFP" gate. Cells exhibiting fluorescence above this negative gate were defined as the "positive GFP" cell population.

Similar to previous reports (4, 5), we report gene expression levels as "fold," where 1-fold is defined as the reporter gene expression level of sTRSV relative to the background fluorescence level. Ligand-directed regulatory effects are reported as fold gene expression levels normalized to the levels in the absence of effector. All fluorescence data and mean ± SD are reported from at least three independent experiments.

Quantification of cellular transcript levels. cDNA was synthesized by using gene-specific primers (SI Table 3) and SuperScript III Reverse Transcriptase (Invitrogen) following the manufacturer's instructions. Relative transcript levels were quantified from the cDNA samples by using an appropriate primer set and iQ SYBR Green Supermix (Bio-Rad, Hercules, CA) according to the manufacturer's instructions on an iCycler iQ qRT-PCR machine (Bio-Rad). The resulting data were analyzed with the iCycler iQ software according to the manufacturer's instructions. Transcript levels of switch constructs were normalized to that of the endogenous actI gene (6) by using actI-specific primers. All data are reported from three independent experiments.

1. Jenison RD, Gill SC, Pardi A, Polisky B (1994) Science 263:1425-1429.

2. Khvorova A, Lescoute A, Westhof E, Jayasena SD (2003) Nat Struct Biol 10:708-712.

3. Gietz R, Woods R (2002) in Guide to Yeast Genetics and Molecular and Cell Biology, Part B, eds Guthrie C, Fink G (Academic, San Diego), Vol 350, pp 87-96.

4. Isaacs FJ, Dwyer DJ, Ding C, Pervouchine DD, Cantor CR, Collins JJ (2004) Nat Biotechnol 22:841-847.

5. Yen L, Svendsen J, Lee JS, Gray JT, Magnier M, Baba T, D'Amato RJ, Mulligan RC (2004) Nature 431:471-476.

6. Ng R, Abelson J (1980) Proc Natl Acad Sci USA 77:3912-3916.

• Early Edition
• Archives
• Online Submission

• Feature Articles

  (Expanded research articles of exceptional breadth)

• Commentaries

  (Expert commentary on leading research)

• Letters

  (Online-only letters to the editor)

• Inaugural Articles

  (Articles by recently elected NAS members)

• PNAS Profiles

  (Biographical profiles of Academy members)

• Sustainability Science

  (Interactions of human and environmental systems)

• Collected Papers

  (Editorials, Perspectives, Reviews, Colloquia, etc.)

• Special Features

  (Solicited articles on exceptional research topics)

• Sackler Colloquia

  (Scientific reports held under NAS auspices)