Quantitative real-time in vitro transcription assay (QRIVTA) for transcriptional regulation studies

Graphical abstract

Dear Editor,

The in vitro transcription (IVT) assay is a powerful tool frequently used in dissecting the molecular mechanism of transcriptional regulation and also plays an important role in the field of drug discovery and RNA-based therapeutics (Yang and Ma, 2016). Traditionally, the IVT assay adopted the incorporation of radioactive nucleotides for RNA detection, which requires time-consuming RNA isolation and PAGE analysis steps, as well as extra safety measures (Höfer et al., 2013; Yang and Ma, 2016). In addition, it could only be used to evaluate the transcription regulation qualitatively or semi-quantitatively. In recent years, several strategies have been proposed to develop real-time IVT (RT-IVT) assays based on fluorescence detection with fluorophore-labeled antisense probes such as the molecular beacon (MB) (Marras et al., 2004) or the fluorescent light-up RNA aptamer (Höfer et al., 2013; Huang et al., 2022; Jensen et al., 2023; Qin et al., 2022). The MB-based detection strategy is to use a single-stranded oligonucleotide with a stem-loop structure, which contains a fluorescent group in the 5ʹ end and a quencher group in the 3ʹ end. Upon hybridizing with the RNA target, the stem-loop structure of MB unfolds, and thus, the fluorophore is separated from the quencher, resulting in the activation of the fluorescence (Fig. S1). The RNA aptamer-based detection strategy is to introduce an RNA aptamer sequence into the target RNA, where the aptamer adopts a defined structure and can specifically bind to and activate the fluorescence of a fluorophore, such as the small molecule mimic of green fluorescent protein (DFHBI) (Fig. S1). These fluorescence detection strategies can be used for quantitative analysis and have a promising prospect in high-throughput and automatic detection. However, despite their great potential, the fluorescence-based RT-IVT assays are still not well established and rarely used in transcription-related studies nowadays. A major reason is the lack of a standard experimental workflow and data analysis method to ensure good data repeatability, reproducibility, and comparability.

In this study, we introduce a quantitative real-time in vitro transcription assay (QRIVTA) for transcription and transcriptional regulation studies, which can be easily carried out with a common real-time PCR thermocycler, along with guidelines for carrying out IVT experiments. It is observed that the fluorescence production rate is attenuated with time when using supercoiled plasmids as transcription templates, even in the presence of excessive rNTPs and MB. We show that the attenuation should reflect the topological characteristics of the supercoiled plasmid template. A new equation for quantitative analysis is proposed to precisely describe the entire time-dependent attenuating fluorescence data from QRIVTA using supercoiled plasmid templates. The transcription repression of the global transcription regulator H-NS on the promoter region of pathogenic E. coli LEE5 operon is used as a model system. Both the MB-based and RNA aptamer-based detection strategies are leveraged to optimize and develop the assay. We have systematically evaluated and optimized different experimental aspects of the fluorescence-based RT-IVT assay regarding plasmid template design, fluorescence detection strategy, sensitivity and specificity tests, experimental operation, data analysis, etc., which enable us to propose a standardized workflow and detailed practical guidelines for carrying out QRIVTA in transcriptional regulation studies.

We cloned the −304 to +171 sequences of the LEE5 promoter (defined as LEE5p gene) (Shin, 2017), followed by a molecular beacon (MB1) complementary sequence (Seq_MB1) (Dong et al., 2018) and a 47-bp E. coli rrnB T₁ terminator sequence (rrnBT₁_47), into a modified pUC19 plasmid (pUC19s) (Fig. S2A). Two plasmids were constructed, with the LEE5p oriented in a divergent (trans-pLEE5p) or tandem (cis-pLEE5p) direction relative to the AmpR gene (Fig. S2A).

Due to the weak fluorescence observed with MB1 during the RT-IVT assays, we designed a new molecular beacon (MB2) with its loop sequence complementary to the +129 to +146 region (Seq_MB2) of the LEE5p (Fig. S3), which exhibited much stronger fluorescence. For RT-IVT assays using the supercoiled cis-pLEE5p template, the fluorescence production rate is attenuated with time, regardless of the template concentration, and no increase in fluorescence without the DNA template, as expected (Fig. S2B). Since further increasing the template concentration didn't substantially enhance fluorescence intensity, 10 nmol/L was adopted for subsequent investigations. For the trans-pLEE5p template, the fluorescence production rate is also attenuated significantly. However, it slows down very quickly and plateaus after 30 min, while the cis-pLEE5p template shows slower attenuation and robust fluorescence increase over 120 min (Fig. S2C). We next examined the specificity of MB2 using modified cis-pLEE5p templates with Seq_MB2 mutated or deleted, detecting −8% nonspecific fluorescence (Fig. S2C), which may be resulted from the high GC content and the presence of 5-bp or more consecutive G/C stretches in the loop region of MB2 (Fig. S3).

We also carried out the RT-IVT assays using aptamer-based detection. Seq_MB1 in the cis-pLEE5p was substituted with either the Spinach (Jensen et al., 2023) or iSpinach (Qin et al., 2022) sequence ( Fig. S4A), and the fluorescence of aptamer-bound DFHBI was monitored. iSpinach exhibits higher fluorescence than Spinach at low DFHBI concentrations (Fig. S4B and S4C), and both aptamers display much weaker fluorescence compared with that of MB2. Notably, the iSpinach-based detection demonstrates high specificity without nonspecific fluorescence when using a cis-pLEE5p template lacking the iSpinach sequence (Fig. S4D). Then, the iSpinach-based detection was applied to evaluate the transcription termination efficiency of LEE5p by placing the iSpinach sequence downstream of the terminator (Fig. S5A). The rrnBT₁_47 terminator shows a termination efficiency of 93% (Figs. 1A and S5B). Five previously reported terminators with high efficiency were tested (Cambray et al., 2013; Chen et al., 2013), and only the M13D double terminator (M13 central and rrnD T₁) exhibits particularly high efficiency (~99%) (Figs. 1A and S5B). The rrnBT₁_91 terminator barely improves the efficiency to 95%, whereas all the others fall below 90%. As the tandem concatenation of terminators could enhance the termination efficiency (Mairhofer et al., 2015), we designed the RBT double terminator (rrnBT₁_47 and BBaU10), which also exhibits a superior termination efficiency (~99%) (Figs. 1A and S5B). The transcription of AmpR was also examined using the iSpinach-based detection ( Fig. S6A), which is less than one-tenth that of LEE5p (Fig. S6B). To avoid transcriptional read-through of AmpR, an rrnBT₁_47 terminator sequence was inserted between AmpR and iSpinach (Fig. S6A), which effectively terminates its transcription (Fig. S6B).

Due to the nonspecific issue with MB2, we refocused on MB1 after the optimization of terminators. The loop region of MB1, with a lower GC content (47.1%) and no long consecutive G/C stretch (Fig. S3), is less likely to cause nonspecific interactions. To optimize the fluorescence intensity of MB1, we modified its design and target RNA sequence by first removing one GC pair from the stem region of MB1 (MB1s) to facilitate its unwinding (Fig. 1B). Then, its target sequence Seq_MB1 was positioned in a loop region of a designed hairpin structure (Seq_MB1-hp) in the transcription template (Fig. 1B), to improve the accessibility of Seq_MB1 on the target RNA. As expected, the modifications result in significantly enhanced detection sensitivity and high specificity, with no nonspecific fluorescence from MB1s when using a template lacking Seq_MB1-hp (Fig. 1C).

With all the above optimizations, new plasmids cis-pLEE5p (cis-pLEE5pN) and trans-pLEE5p (trans-pLEE5pN) were constructed, with LEE5p followed by Seq_MB1-hp and the M13D terminator, positioned in a divergent or tandem orientation relative to the AmpR (Fig. 1D), respectively. Transcription of LEE5p in these plasmids was measured using either the MB1s- or iSpinach-based detection. It was found that fluorescence production rates are all attenuated over time with both detection strategies. Notably, different from the trans-pLEE5p, the fluorescence of the new trans-pLEE5pN template no longer plateaus quickly and maintains a continuous increase, although it is still weaker than that of cis-pLEE5pN (Fig. 1E). These results indicate that the transcription of AmpR in the old trans-pLEE5p template, which is not efficiently terminated, significantly impacts the LEE5p transcription.

In our RT-IVT assays, RNA polymerase (RNAP) substrates rNTPs (0.5 mmol/L) and MB (500 nmol/L) are used in vast excess (Fig. S7); their consumption should not be the reason for the attenuation of fluorescence production rate. Interestingly, both linearized and nicked cis-pLEE5pN templates resulted in a linear increase in fluorescence intensity (Figs. S8 and S9), indicating that the attenuation should also not be caused by the RNAP stability. Therefore, the attenuation of the fluorescence production rate should reflect some properties of the supercoiled plasmid template during transcription. It is well established that the transcription would generate positive supercoils in front of the RNAP elongation complex and negative supercoils behind on the DNA template, which was termed the "twin supercoiled domain model" (Chong et al., 2014). The positive supercoils accumulated in front of the RNAP elongation complex could slow down the transcription elongation, thus attenuating the RNA production rate over time. In addition, transcriptional pausing can cause RNAP stalling or being trapped on the supercoiled DNA template, which could also decelerate the transcription rate (Huang et al., 2022).

To evaluate the transcription efficiency in RT-IVT assays using supercoiled plasmids as transcription templates, it is more general to determine the initial fluorescence production rate (v₀) from the fluorescence curve, which is the maximal fluorescence production rate during the time course of the IVT assay, reflecting the maximal steady-state transcription rate. We first tested fitting the fluorescence curves using an exponential decay (increase form) function (Equation S1 in materials and methods) (Huang et al., 2022), but it did not produce very good curve fitting results for the data from the MB1s- or iSpinach-based detection (Fig. 1G). We then analyzed the first derivatives of the fluorescence data (Fig. S10), which reflect the fluorescence production rate at each time point. It appears that the production rate does not go to zero at the end of detection, even though it is attenuated with time. Therefore, the first derivative data can be well-fitted to an equation (Fig. 2A, Equation 1) with an exponentially decaying rate v_d(t) and a constant rate v_c (Fig. S10). Using the indefinite integral, we can derive the function of fluorescence intensity with respect to time (Fig. 2A; Equation 2). v₀ is thus the sum of v_d0 and v_c (v₀ =  v_d0 +  v_c) (Fig. 2A).

The transcription repression ability of H-NS on LEE5p was evaluated using both the MB1s- and iSpinach-based detection with the supercoiled cis-pLEE5pN template. A concentration-dependent decrease in fluorescence production rate is observed with the increase of H-NS concentration (Fig. 1F). Remarkably, all the fluorescence data of the LEE5p transcription obtained with or without H-NS, as well as the fluorescence data of the AmpR transcription, can be well fitted with Equation 2 (Figs. 1G, S6C and S11), demonstrating the robustness of this new equation.

Using Equation 2, we determined v₀ from both the MB1s- and iSpinach-based detection for different H-NS concentrations (Table S4). v₀ is used to reflect the transcription efficiency, and thus, the percentage of reduction in v₀ reflects the transcription repression ability of H-NS. The results show that the transcription repression ability of H-NS displays a steep increase rather than a linear increase above a certain concentration. For the MB1s-based detection, a slight transcription repression (9%) is observed with 1 μmol/L H-NS (Fig. 1H), which is markedly enhanced upon increasing the H-NS concentration from 1 to 3 μmol/L. Specifically, v₀ is repressed by 57% at 1.5 μmol/L, 74% at 2 μmol/L, and 94% at 3 μmol/L H-NS, respectively (Fig. 1H). Further increasing the concentration to 5 and 10 μmol/L results in nearly complete transcription repression. This is consistent with that H-NS oligomerizes in a concentration-dependent manner, which only forms higher-order oligomers at higher concentrations, and the DNA binding affinity of H-NS is intrinsically associated with its oligomerization level (Lukose et al., 2024). At the same concentration, the transcription repression ability of H-NS determined from the iSpinach-based detection is about the same as that of MB1s (Fig. 1H).

To investigate the potential impact of extra sequence downstream of the LEE5p on transcription repression of H-NS, we removed the Seq_MB1-hp from the cis-pLEE5pN template and used MB2-based detection to measure the LEE5p transcription (Figs. 1F and S12). Nonspecific fluorescence of MB2 was measured using a Seq_MB2-mutated plasmid template under identical conditions, and specific fluorescence was obtained by subtracting nonspecific fluorescence from the raw fluorescence (Fig. S12B). Notably, all the fluorescence data from the MB2-based detection can also be well-fitted with Equation 2 (Fig. 1G). The transcription repression ability of H-NS determined from the MB2-based detection is also essentially the same as those determined from the MB1s- and iSpinach-based detection (Fig. 1H). Collectively, we developed an efficient and robust QRIVTA for the IVT and transcriptional regulation studies.

The repression ability of H-NS on the linearized cis-pLEE5pN (L1) template was also evaluated using QRIVTA (Fig. S13), with v₀ obtained through linear regression. 2 μmol/L H-NS only represses v₀ by 10% for the linearized plasmid template (Fig. 1I), compared to 74% for the supercoiled plasmid template (Fig. 1H), indicating that H-NS has a much stronger transcription repression ability on the supercoiled plasmid template. This is different from a previous report, which claimed that the transcription repressions of H-NS on both the linear and supercoiled DNA are the same, based on the traditional IVT assay (Shin et al., 2012). As a nucleoid-associated protein, it is reasonable that H-NS bridges or condenses supercoiled DNA more effectively, thus enhancing its repression ability.

As it is well known that temperature affects the transcription repression ability of H-NS (Lukose et al., 2024), we measured the transcription repression ability of H-NS at 30°C and 25°C (Fig. S14). Again, all the fluorescence data at lower temperatures could also be well-fitted with Equation 2. H-NS exhibits stronger repression abilities at lower temperatures with QRIVTA (Fig. 1J), which should be attributed to that H-NS forms higher-order oligomers at lower temperatures (Lukose et al., 2024). 0.5 μmol/L H-NS can repress v₀ by 19% at 30°C and 34% at 25°C, compared to that 1 μmol/L H-NS only slightly represses v₀ by 9% at 37°C. The transcription repression ability of 1 μmol/L H-NS is significantly enhanced to 58% at 30°C and 73% at 25°C, respectively.

It is recommended to use supercoiled plasmids as the IVT templates for the transcription regulation studies to mimic the topological state of supercoiling DNA during transcription, as short linear fragment DNA templates are less physiologically representative (Will et al., 2014). For the first time, we proposed a more precise equation (Equation 2) to accurately extract the initial rate (v₀) of the fluorescence production and evaluate the transcription efficiency with supercoiled plasmid as the transcription template. Remarkably, all the fluorescence data from supercoiled plasmid templates in our study can be precisely fitted with this equation, irrespective of the detection strategy, the presence or absence of H-NS, or the experiment temperatures. More studies are required to illustrate the physical meanings of the other kinetic parameters (k, v_d0, v_c) determined from the equation.

Recently, Jensen et al. observed in their Spinach-based RT-IVT assay that the fluorescence curve shows an initial lag phase with a slow increase in fluorescence production rate, followed by a linear increase phase and then a time-dependent attenuation (Jensen et al., 2023). They used a custom MATLAB fitting program to extract the slope of the linear increase phase to determine the steady-state transcription rate. We found that the lag phase is primarily due to experimental procedure issues, which can be avoided by altering the procedure to initiate transcription with the addition of rNTPs to the premixed solution with all other components. When we initiated transcription by adding the DNA template, as Jensen et al. did, a lag phase appeared in the fluorescence curves for both MB1s- and iSpinach-based detection (Fig. S15A). We suspected that the lag phase observed is very likely due to that the template DNA plasmids, as macromolecules, take a longer time to diffuse and form the open complex for transcription initiation. On the contrary, rNTPs are small molecules with rapid diffusion rates, which could facilitate quick equilibrium in the samples. In addition, we found that the first derivatives of the fluorescence data do not reveal an obvious linear increase phase (Fig. S15B and S15C). Moreover, focusing on the presumed linear region of the fluorescence curves would omit the valuable transcriptional kinetics information of the entire fluorescence data. For comparison, the kinetic parameters of the IVT fluorescence data can be easily extracted with Equation 2 proposed in this study, which precisely describes the entire fluorescence data.

The QRIVTA enables the quantitative analysis of different intrinsic and extrinsic factors that influence transcription and transcriptional regulation, thereby enhancing our understanding of the underlying mechanisms. It is worth emphasizing that variations in the IVT experimental procedure can significantly affect the kinetics of the fluorescence production, which may result in poor data repeatability, reproducibility, and comparability. Consequently, it is crucial to standardize the fluorescence-based RT-IVT assay procedure. Detailed practical guidelines for carrying out QRIVTA, covering plasmid template design, fluorescence detection strategy, sensitivity and specificity tests, experimental operation, and data analysis, are provided in the Supplementary Material (Guidelines for QRIVTA).

In summary, a standardized and efficient workflow of QRIVTA is established, which can be easily applied in many aspects of studies related to transcriptional regulation (Fig. 2B). QRIVTA can be performed in 96- or 384-well plates using a common real-time PCR thermocycler widely available nowadays. We believe that the application of QRIVTA will not only advance the field of transcriptional regulation but also offer broad potential applications in areas such as synthetic biology and pharmaceutics.

References

[1] Cambray G, Guimaraes JC, Mutalik VK et al. Measurement and modeling of intrinsic transcription terminators. Nucleic Acids Res 2013;41:5139-5148.

[2] Chen YJ, Liu P, Nielsen AA et al. Characterization of 582 natural and synthetic terminators and quantification of their design constraints. Nat Methods 2013;10:659-664.

[3] Chong S, Chen C, Ge H et al. Mechanism of transcriptional bursting in bacteria. Cell 2014;158:314-326.

[4] Dong J, Wu T, Xiao Y et al. Target-triggered transcription machinery for ultra-selective and sensitive fluorescence detection of nucleoside triphosphates in one minute. Biosens Bioelectron 2018;100:333-340.

[5] Höfer K, Langejürgen LV, Jäschke A. Universal aptamer-based real-time monitoring of enzymatic RNA synthesis. J Am Chem Soc 2013;135:13692-13694.

[6] Huang YH, Trapp V, Puro O et al. Fluorogenic RNA aptamers to probe transcription initiation and co-transcriptional RNA folding by multi-subunit RNA polymerases. Methods Enzymol 2022;675:207-233.

[7] Jensen D, Ruiz Manzano A, Rector M et al. High-throughput, fluorescent-aptamer-based measurements of steady-state transcription rates for the Mycobacterium tuberculosis RNA polymerase. Nucleic Acids Res 2023;51:e99.

[8] Lukose B, Maruno T, Faidh MA et al. Molecular and thermodynamic determinants of self-assembly and hetero-oligomerization in the enterobacterial thermo-osmo-regulatory protein H-NS. Nucleic Acids Res 2024;52:2157-2173.

[9] Mairhofer J, Wittwer A, Cserjan-Puschmann M et al. Preventing T7 RNA polymerase read-through transcription-A synthetic termination signal capable of improving bioprocess stability. ACS Synth Biol 2015;4:265-273.

[10] Marras SA, Gold B, Kramer FR et al. Real-time measurement of in vitro transcription. Nucleic Acids Res 2004;32:e72.

[11] Qin W, Li L, Yang F et al. High-throughput iSpinach fluorescent aptamer-based real-time monitoring of in vitro transcription. Bioresour Bioprocess 2022;9:112.

[12] Shin M. The mechanism underlying Ler-mediated alleviation of gene repression by H-NS. Biochem Biophys Res Commun 2017;483:392-396.

[13] Shin M, Lagda AC, Lee JW et al. Gene silencing by H-NS from distal DNA site. Mol Microbiol 2012;86:707-719.

[14] Will WR, Bale DH, Reid PJ et al. Evolutionary expansion of a regulatory network by counter-silencing. Nat Commun 2014;5:5270.

[15] Yang X, Ma C. In vitro transcription assays and their application in drug discovery. J Vis Exp 2016;115:e54256.

RIGHTS & PERMISSIONS

The Author(s) 2024. Published by Oxford University Press on behalf of Higher Education Press.