Rapidly Characterizing the Fast Dynamics of RNA Genetic Circuitry with Cell-Free Transcription–Translation (TX-TL) Systems

RNA regulators are emerging as powerful tools to engineer synthetic genetic networks or rewire existing ones. A potential strength of RNA networks is that they may be able to propagate signals on time scales that are set by the fast degradation rates of RNAs. However, a current bottleneck to verifying this potential is the slow design-build-test cycle of evaluating these networks in vivo. Here, we adapt an Escherichia coli-based cell-free transcription-translation (TX-TL) system for rapidly prototyping RNA networks. We used this system to measure the response time of an RNA transcription cascade to be approximately five minutes per step of the cascade. We also show that this response time can be adjusted with temperature and regulator threshold tuning. Finally, we use TX-TL to prototype a new RNA network, an RNA single input module, and show that this network temporally stages the expression of two genes in vivo.

The transcriptional attenuation mechanism from the Staphylococcus aureus plasmid pT181 Figure S2 Supporting plots for Plasmid architecture for attenuator and antisense plasmids Figure S4 GFP production rate plots used to calculate the average GFP production in Figure 3B TX-TL Figure S5 Total yeast RNA control experiment TX-TL Figure S6 Calculation of circuit response time, τ TX-TL Figure S7 Degradation of the malachite green aptamer in TX-TL TX-TL Table S1 Individual response times (τ) calculated for each experiment Table S2 P-values from Welch's t-test comparing response times from different experiments Figure S8 Magnified plots of Figure 6B TX-TL Figure S9 Theophylline toxicity experiments TX-TL Figure S10 Theophylline and aptamer-AS-2 co-spike experiment TX-TL Figure S11 Plasmid architecture for the 3-plasmid transcription cascade and steady state in vivo data In vivo Figure S12 In vivo SIM plasmid architecture In vivo  Table S3 Important DNA sequences Table S4 Plasmids used in this study References Appendix 1

Materials and Methods Steady state in vivo gene expression
Plasmid combinations were transformed into chemically competent E. coli TG1 cells, plated on Difco LB+Agar plates containing 100 µg/mL carbenicillin, 34 µg/mL chloramphenicol, and 100 µg/mL kanamycin and incubated overnight at 37°C. Plates were taken out of the incubator and left at room temperature for approximately 7 h. Four colonies were used to inoculate 300 µL of LB containing carbenicillin, chloramphenicol, and kanamycin at the concentrations above in a 2 mL 96-well block (Costar 3960), and grown approximately 17 h overnight at 37°C at 1,000 rpm in a Labnet Vortemp 56 bench top shaker. 4 µL of this overnight culture were then added to 196 µL (1:50 dilution) of M9 minimal media containing the selective antibiotics and grown for 4 h at the same conditions as the overnight culture. 100 µL of this culture were then transferred to a 96well plate (Costar 3631) containing 100 µL of PBS. SFGFP fluorescence (485 nm excitation, 520 nm emission) and optical density (OD, 600 nm) were then measured using a Biotek SynergyH1m plate reader.

RNA degradation in TX-TL
TX-TL buffer and extract tubes were thawed on ice for approximately 20 min. 2.2 µM of purified malachite green aptamer RNA was added to reaction tubes. Buffer and extract were mixed together with malachite green dye (Sigma M9015, final concentration 10 µM) and then added to the reaction tubes according to the previously published protocol 1 . 10 µL of each TX-TL reaction mixture was transferred to a 384-well plate (Nunc 142761), covered with a plate seal (Nunc 232701), and placed on a Biotek SynergyH1m plate reader. Temperature was controlled at either 29°C or 37°C. Malachite green fluorescence was measured (610 nm excitation, 650 emission) every 30 seconds. Fluorescence trajectories were fit to an exponential decay function in the form of y(t) = a*exp(-t/τ) + b. Fitted values of τ were used to calculate the half life (t 1/2 ) of the malachite green aptamer in TX-TL.  Figure S1. The transcriptional attenuation mechanism from the Staphylococcus aureus plasmid pT181 2,3 . The attenuator lies in the 5' untranslated region of the transcript and can fold into a structure that will allow transcription to continue if antisense RNA is not present (ON). Antisense RNA binding to the attenuator causes the formation of a terminator hairpin, stopping transcription before the gene of interest (OFF, indicated by x symbol).    Table S3 for sequence details of these plasmids.  Figure S4. GFP production rate plots used to calculate the average GFP production in Figure 3B. 0.5 nM of L1 and the indicated concentration of no-antisense control DNA (0-20 nM) was tested with three extract and buffer batches. Shaded regions represent standard deviations from at least 11 independent reactions calculated at each time point.

Supporting Figures and Tables
Boxes represent constant maximum SFGFP production regions used to calculate averages. Batch 2 reached constant GFP production faster than batches 1 and 3, therefore data was only collected for 2 h for this batch.    Welch's t-test was used to determine whether the L1+L2 and L1+L2+L3 distributions over the normalized experimental replicates were statistically different from each other. The plot shows a zoomed-in region from (B). The * on the plot indicate times at which the t-test p-value was less than 0.05, with values listed in the table. The difference threshold, Δ norm , in averaged normalized fluorescence was calculated at the earliest time where the two data sets were statistically different by this test. (D) For each experiment, the Δ norm was converted into an un-normalized scale, Δ, by multiplying by the appropriate normalization factor. Each independent L1+L2+L3 trajectory was then compared to the average L1+L2 trajectory for that experiment to find the specific time at which the L1+L2+L3 trajectory was consistently greater than the average L1+L2 curve by Δ. These times are defined as the response time for that spike replicate. Example trajectories and response time depiction are shown for experiment 1. (E) Response times were calculated for all replicates and averaged to give the final τ value. A similar procedure was used to measure response times from in vivo experiments, with variations in the procedure noted in "Response Time Calculation" in Methods.  Figure 3). 5.6 picomoles of aptamer was allowed to degrade in each reaction. Half lives (t 1/2 ) were calculated by fitting each trajectory to an exponential function of the form y(t) = a*exp(t/τ) + b. τ values were used to calculate t 1/2 . Error in t 1/2 represents standard deviations of six individual reactions. Inset plots magnify the initial fluorescence decay.    Table S3 for sequence details of these plasmids. (B) In vivo steady state expression data from cells cotransformed with L1 (blue bar), L1+L2 (red bar), or L1+L2+L3 (purple bar). Control plasmids lacking functional coding sequences were used in place of L2 and L3 for the (-) conditions. Error bars represent standard deviations of 4 independent transformants.  Figure S12. In vivo SIM plasmid architecture. (A) The bottom level (L1) contains both a single pT181 attenuator (Att-1) upstream of the RFP coding sequence and tandem pT181 attenuators (Att-1-Att-1) upstream of the SFGFP coding sequence on a pSC101 backbone with kanamycin resistance. L2 contains the pT181 antisense (AS-1) controlled by the pT181 mutant attenuator (Att-2) on a p15A backbone with chloramphenicol resistance. L3 contains the theophylline aptamer-pT181 mutant antisense fusion (aptamer-AS-2) on a ColE1 backbone with ampicillin resistance. (B) L2 and L3 are the same as in A. L1 contains Att-1 upstream of the SFGFP coding sequence and tandem Att-1-Att-1 upstream of the RFP coding sequence on a pSC101 backbone with kanamycin resistance. See Table S3 and S4 for sequence details of these plasmids.  Table S4 -Plasmids used in this study. Sequences in the plasmid architecture ( Figures  S3, S11, S12) can be found in Table S3.  We consider the double inversion RNA transcriptional cascade depicted in Figure  (1). In the simplest model, we can calculate the dynamical behavior of this network using ordinary di↵erential equations that capture the basic chemical reactions of gene expression at each level of the cascade ( [1]).
Here A i represents the concentration of the antisense signal species, and M and P denote the concentrations of mRNA and protein, respectively, of the experimentally observable fluorescent protein encoded in the first level of the cascade. We have also used the approximation that P does not degrade on the timescale of a TX-TL experiment and therefore has no degradation term. Since these equations represent number of molecules, k i and d i have units of 1/s. Note that since this is an RNA circuit, we only need to consider translation of the final reporter level -each of the intermediate levels of the cascade can be described by a single equation representing the transcription and degradation dynamics of the RNA species. We are also ignoring additional e↵ects due to the ribozyme in level 2 that is present in the real cascade ( Figure 2E of the main text.) Our goal is to estimate the response time of this network to a spike in the concentration of the level 3 DNA at time t = 0. To calculate this estimate, we make the simplifying assumption of a threshold function for f (A), following ( [1]). Under this assumption f (x ) = 0, and f (x < ) = 1 as depicted in Figure (2), for some threshold . Under this assumption, an antisense species will completely repress the transcription of its target when its concentration is above .  To model the spike experiment, we consider the initial condition We also assume that the reactions have been proceeding long enough before t = 0 for A 1 to have reached steady-state, i.e. (6) which uses eq. (5), and our threshold assumption. We now solve each equation in turn: Using the initial condition eq. (5), and an integration factor, we find (1 e d 2 t ).
Solving for A 1 (t) is made easier by considering the time, 2 , at which A 2 (t) reaches the threshold needed to attenuate the transcription of A 1 , which we label 2 (see Figure  5A of the main text). Solving A 2 ( 2 ) = 2 gives Using an integrating factor to solve for A 1 (t), we find Using the threshold function, the last integral can be taken from t = 0 to t = 2 , and plugging in the steady state condition for A 1 (0) from eq. (6), we find Similarly, to solve for M (t), we first find the time, 1 , at which A 1 (t) reaches the threshold needed to attenuate the transcription of M , which we label 1 . Solving A 1 ( 1 ) = 1 gives (11) Using an integrating factor to solve for M (t), we find We assume that initially A 1 (0) > 1 so that f (A 1 (t < 1 )) = 0 and M is initially not expressed. This also means M (0) = 0 by our threshold assumption. When t 1 , then A 1 (t) < 1 and M (t) can be expressed. Using this, we find  In a similar manner we can solve for P (t), using the fact that P (t < 1 ) = 0: which when we use eq. (13) we find (15) P (t) = ( 0, t < 1 , ⌘ , t 1 . Since the protein must go through a maturation step before it can be observed, characterized by a time ↵, we find that the circuit response time, ⌧ = 1 + ↵, to be  Figure 5A in the text shows a graphical representation of these results.