When you choose to publish with PLOS, your research makes an impact. Make your work accessible to all, without restrictions, and accelerate scientific discovery with options like preprints and published peer review that make your work more Open.


iGEM REPORT: Application of RNA thermometers in gene circuits

Note for Reviewers

This work has been submitted for open review as an iGEM Report.

Reviewers, please consider the following questions when reviewing the report:

1. Does the submission accurately and clearly report the experiment carried out?

2. Does the report adhere to the PLOS Data Policy?

3. Is the report presented correctly and well written?

See iGEM Reviewer Instructions for more information.

Please leave your review as a comment below the Report.

Application of RNA thermometers in gene circuits

Yuwei Han, Jiongjian Li, Tiange Li*, Yakun Tang, Chenghao Shen, Gandi Yao, Wenbo Xu, Hao Wu, Xinhao Wang, Yizhou Wang, Zixu Wang, Tianhua Zhai, Zhe Yang, Guoxia Han, Dechang Xu, Yunpeng Zhong, Yacheng Guan

Department of Biological Sciences, Xi’an Jiaotong-Liverpool University



Author contributions

Conceptualization: J.J.L., G.D.Y. and Z.X.W.

Methodology: W.B.X., T.G.L. and X.H.W.

Investigation: Y.W.H., J.J.L., T.G.L., Y.K.T., C.H.S., G.D.Y., W.B.X., H.W., X.H.W., Y.Z.W., Z.X.W., T.H.Z. and Z.Y.

Writing: – Original Draft: H.W., W.B.X., Y.W.H. and Y.Z.W.

– Review & Editing: T.G.L and X.H.W.

Funding Acquisition: Y.P.Z.

Resources: W.B.X. and Y.Z.W.

Supervision: D.C.X., G.X.H., Y.P.Z. and Y.C.G.


RNA thermometers have potential to serve as effective synthetic biological parts as its simplicity and sensitivity to temperature changing. They are also indispensable components in the gene circuit designed by 2015 Xi’an Jiaotong-Liverpool University’s international genetically engineered machine program. For this reason, series of RNA thermometers in previous research were tested, using fluorescence density of green fluorescent protein to reflect the gene expression before and after the RNA thermometer switched on. It turned to be that the performance was not stable and accurate as expected. Possible reasons have been discussed as well as its limitations in application in a gene circuit context.


RNA thermometers (RNATs) are temperature-induced riboswitches. The secondary structure of RNA, usually a hairpin structure, senses the changing temperature in the environment and undergoes a conformation shift. The background principle of the switch is that the stability of the paring up sequence in the hairpin decreases as the temperature rising, while the specific effect of its conformation change on gene translation depends on the location of this RNAT.

The most common way to conduct a post transcription regulation is that the hairpin structure harbors the Shine-Dalgarno sequence (SD sequence), as shown in Figure 1. In this case, when the temperature is low and the double strand of the RNAT is stable, the SD sequence is involved so that it is inaccessible to the 30S unit of the bacterial ribosome, resulting in the repression of translation. (SD sequence is a polypurine sequence located in around 8 nt upstream from the start codon and responsible for anchoring ribosome in the RNA single strand). As the temperature rising up, the hairpin structure will gradually dissolves and finally vanishes. As a result, the SD sequence is exposed to ribosome and thus triggers the translation.


Figure 1

Fig. 1 Responsiveness of mRNA structures to environmental cues (1)


A conserved family of RNATs is the ROSE-like elements (ROSE stands for Repression Of heat Shock gene Expression), first discovered in the 5′ untranslated region of rhizobial heat shock genes (2). All the known RNAts presenting in ROSE family control expression of small heat shock genes (2). However, most naturally occurring ROSE elements are relatively large and fold into complex secondary structures that usually contain 2-4 hairpin structures (Fig. 2). Hence, for practical reason, we tested A1 RNAT that derived from ROSE family. Thank BIT-CHINA for provding DNA sequences of this part. The possible secondary structure of A1 was simulated by RNAstructure (Fig.3).


Figure 2

Fig. 2 Proposed RNA secondary structure of the 5’UTR from the prfA gene of the pathogenic bacterium Listeria monnocytogenes, which belongs to ROSE family (3)

Figure 3

Fig. 3 The possible secondary structure of A1.

Another extensively studied RNATs family in nature is FourU. They were firstly found in Salmonella, so called as four highly conserved uridine nucleotides are placed right opposite to the SD sequence. FourU elements usually have two hairpin structures, in the case of temperature rising, the one without SD sequence is relatively steady while the other contains hidden SD sequence would be melted. In 2008, TUDelft designed and submitted part K115002 to iGEM registry based on natural forms of FourU elements. The possible secondary structure of FourU was simulated by RNAstructure (Fig.4). This part was retested in this year to determination of the validity.


Figure 4

Fig. 4 The possible secondary structure of part K115002.

Inspired by nature, the possibility of designing synthetic RNA thermometers was explored based on the assumption that the RNATs do function by the proposed simple hairpin-melting mechanism that expose the ribosome-binding site. Therefore, Stem-loop structure with the SD sequence embedded in the stem would be the simplest RNAT. Juliane Neupert, Daniel Karcher and Ralph Bock (4) designed and tested a series of synthetic RNATs, among which RNAT U6 (Fig. 5) had the best performance. Synthetic RNAT U6 was chosen as one of our testing RNATs.


Figure 5

Fig. 5 The possible secondary structure of U6
The aim of this study is to test and modify the created RNATs in previous study and select the most effective one to function in the designed gene circuit.


First trail-qualitative test

In the first trail of experiments, a large scale of the RNA thermometers tested in previous studies was chosen. They were from iGEM registered parts (K115002, K115017), support of team BIT-China (A1, A2, A3), literature (U9, U10) and self-designed (U6-GC).

The constructed parts were unified as PT7-LacO- pET-28a random sequence-RNAT-eGFP and as a whole they were inserted in plasmids pET-28a with a selectable marker kanR. Then plasmids were transformed into BL21 (DE3) for expression. After transformation, samples were smeared on LB K+ plates and cultured separately at 30 degrees celcius, 37 degrees celcius and 42 degrees celcius for overnight culture. Next morning, monocolonies were selected and transferred to liquid LB K+ to grow for 4-8h, before induced by IPTG at respective temperature. Finally, pictures were taken under UV light to compare the light strength of the GFP.

Second trail-quantitative test

It was suspected that the poor performance of RNATs in first round of the test was due to the inappropriate secondary structure of T7-Lac O regulatory part. The transcription of T7 promoter starts at CACTATAG (transcription start site indicated in bold). Hence, lac operators may also be transcribed and therefore interfere with RNA thermometers by paring up. To determine whether this assumption was reasonable, the team redesigned temperature induced transcription cascades by introducing three different promoters: pBad, J23119, T7 and three RNA thermometers (A1, FourU, U6) to the project and eliminated any loop forming sequence between transcription initiation points and the start points of RNA thermometers in these nine constructs indicated showed below in Fig.6.

Figure 6

Fig. 6 Gene constructions in second trail

In the second trail, M1 to M9 plasmids were transformed to BL21 (DE3) competent cell and subsequently incubated overnight under 30°C and 37 °C respectively. The selected transformed monocolonies were then inoculated into 4ml LB tubes, which were shake at 220rpm at 30°C or 37 °C according to the plate where they were incubated. These tubes of strains were then induced using IPTG for W7, W8 and W9, and arabinose for W4, W5, and W6. J23119 is one of the strongest component promoters, therefore, no need to be induced. These tubes were then inoculated for another 8h before harvesting. The cell density were normalized to OD600 = 0.93, such that the centrifuged bacteria dot can be compared under UV irradiation. The best-performed strains were further tested for fluorescence of EGFP using plate reader. 

In order to avoid the noise of full cell caused by cell wall, the supernatant cell processing sonicate broken was used for the plate reader test. The supernatant were then serial diluted on a black 96-well plate where the fluorescents of EGFP were tested. (Fig 7) The emission ray of 480nm from the top the plate reader excited EGFP, which consequently attributed to the absorbance at about 580nm.

Figure 7

Fig. 7 Arrangement of the plate reader test


The first trail of the experiment aimed to qualitatively compare the RNATs from different families.  Fig 8 to 14 are illustrations of LacO-random sequence-RNAT secondary structure and a quick test of eGFP expression. Three rows (from above to below) indicated incubation temperature of 30oC, 37oC and 42oC. Six columns (from left to right) represented control and experimental groups alternating. As shown in Fig 8, synthetic RNAT A1, which was designed to have an annealing temperature of about 42oC, had obvious different fluorescence strength under the temperature 30oC, 37oC and 42oC.  The control groups without IPTG induction illustrated no fluorescence due to the tight restriction of Lac operator. However, RNAT A3, another synthetic ribothermometer similar to A1, showed no fluorescence from 30oC to 42oC (Fig 9). RNAT U6-GC (Fig 10), U9 (Fig 11) and U10 (Fig 12), whose expected annealing temperature is about 37oC, illustrated slight difference in eGFP expression between 30oC and 37oC. Among them U6-GC had the best performance, but the difference was smaller compared with A1 and K115017. Fig 13 and Fig 14 are the prediction of secondary structures of RNAT-K115002 and K115017 and the expression of eGFP following them. The predicted annealing time for them was about 32oC. For K115017, it was eas y to observe the expression of eGFP and the difference between 30oC and 37oC. However, the level of eGFP expression after K115002 was non-observable, similar to RNAT-A3.


Figure 8

Fig. 8 The predicted secondary structure and phenotype test result of RNAT-A1.


Figure 9

Fig. 9 The predicted secondary structure and phenotype test result of RNAT-A3.


Figure 10

Fig. 10 The predicted secondary structure and phenotype test result of RNAT-U6-GC.


Figure 11

Fig. 11 The predicted secondary structure and phenotype test result of RNAT-U9.


Figure 12

Fig. 12 The predicted secondary structure and phenotype test result of RNAT-U10.


Figure 13

Fig. 13 The predicted secondary structure and phenotype test result of RNAT-K115002.


Figure 14

Fig. 14 The predicted secondary structure and phenotype test result of RNAT-K115017.

The second trail of the experiment aimed to compare three different promoter and operator combination, the IPTG induced pT7 + LacO, arabinose induced pBAD and constitutively expressed J23119, linked with the three best performed RNATs A1, K115017 and U6-GC; it also quantitatively measured the level of eGFP expression to characterize the best performed RNAT A1. Fig 15-17 illustrated the expression of eGFP that follows different promoters; two rows indicated 30oC (above) and 37oC (below) incubation and three columns (from left to right) represented RNAT A1, K115017 and U6-GC respectively.  It could be seen that W1 (J23119+A1) which cultured at 37 is much brighter than at 30 and W7/8/9 (T7+A1/K115017/U6-GC), which cultured at 37 is a little brighter than at 30. W2/3/4/5/6 did not show conspicuous differences between 30 and 37. As Fig 18 shows, the fluorescence of the construction J23119-A1-eGFP at 37 was 12 times higher than the fluorescence of the sample at 30, which means the amount of protein expressed at 37 was 12 times bigger than at 30. The result of RNAT A1 with pBad promoter is demonstrated in Fig 19. As can be seen, the result is not optimistic. Before diluting the samples, all the fluorescence of the samples at 30 is higher than the fluorescence of the samples at 37. Figure 20 shows the result of RNAT A1 with T7 promoter. It seems that RNAT A1 also worked well with T7 promoter. When IPTG is added, the fluorescence of the samples at 37 was 6 times higher than the fluorescence of the samples at 30. On the contrary, in the absence of IPTG the fluorescence of each sample was really low, which means only small amount of protein expressed without induction.


Figure 15

Fig. 15 The performance of three RNAT following promoter J23119. Above row was incubated in 30oC, bottom row in 37oC. From left to right are RNAT A1, K115017 and U6-GC.


Figure 16

Fig. 16 The performance of three RNAT following promoter pBAD after induction. Above row was incubated in 30oC, bottom row in 37oC. From left to right are RNAT A1, K115017 and U6-GC.


Figure 17

Fig. 17 The performance of three RNAT following promoter pT7 and Lac operator after induction. Above row was incubated in 30oC, bottom row in 37oC. From left to right are RNAT A1, K115017 and U6-GC.


Figure 18

Fig. 18 Comparison of eGFP fluorescence in the construction of pJ23119-A1-eGFP incubated at 30oC and 37oC. Measured at OD 590.


Figure 19

Fig. 19 Comparison of eGFP fluorescence in the construction of pBAD-A1-eGFP incubated at 30oC and 37oC. Control group without arabinose induction. Measured at OD 590.


Figure 20

Fig. 20 Comparison of eGFP fluorescence in the construction of pT7-LacO-A1-eGFP incubated at 30oC and 37oC. Control group without IPTG induction. Measured at OD 590.


Generally, the performance of RNATs was not stable and accurate as expected. The difference in the brightness of GFP proteins under various temperatures is subtle and hardly can be observed by eyes. This may indicate that the secondary structure of the RNAT is vulnerable to the change of the intercellular and intracellular environment. This experiment has been rigorously conducted as the experimental protocol suggested in previous research (3), so the possibility of unaware changing happened in cells is under suspected. One hypothesis is that the gene context surrounding the RNAT may contain disturbing sequences which might pair up with RNAT sequence. Thus directly destroy the secondary structure of the RNAT. Otherwise it can also form hairpin structure themselves near the RNAT, which may be the case if the RNAT is closely followed an operator sequence. In this way, RNAT structure could also be affected. The second trial of this experiment partially proved this hypothesis as the performance of T7 promoter in the second trail was better. However, as there is still obvious difference in the performance of the RNAT after different promoters, providing in all cases the disturbing sequences have been eliminated, further causes needed to be identified. Additionally, it is suggested that magnesian ions also play a role in the stability of the RNAT, the relationship between the ion concentration and RNAT function is worthy to be characterized.

On the other hand, the non significant results are also partially due to the improperly chosen of the report protein, GFP. Its light density is susceptible to environments. Moreover, purely diluted the GFP solutions in a serial dilution factor of 10 resulted that there may be no obvious difference between resolutions with a difference in concentration of 103 to the observation of human eyes. Therefore, more effective in vivo detector or even detector system to precisely reflect the structure change in RNAT can be explored to monitor its performance, replacing the using of a report protein.

To improve the accuracy in RNAT test, strict control group should be included in each construction of the RNATs. Meanwhile, it is advocated to test the change in report protein expression for one strain under a gradually rising temperature, thus reflect the function of RNAT dynamically, rather than incubated under different temperature separately.



  1. TUDelft (2008) RNA thermometer [Online image]. Available from: http://2008.igem.org/File:Rna_thermometer.png (Accessed: 19th August 2015).
  2. Nocker A. A mRNA-based thermosensor controls expression of rhizobial heat shock genes. Nucleic Acids Research. 2001;29(23):4800-4807.
  3. Neupert JBock R. Designing and using synthetic RNA thermometers for temperature-controlled gene expression in bacteria. Nat Protoc. 2009;4(9):1262-1273.
  4. Neupert J, Karcher D, Bock R. Design of simple synthetic RNA thermometers for temperature-controlled gene expression in Escherichia coli. Nucleic Acids Research. 2008;36(19):e124-e124.


We acknowledge Mr Yunpeng Zhong for providing us instructions on biobrick construction and characterization, Mr Yacheng Guan for helping us improve the expression of proteins in the cell. We thank Dr Guoxia Han and Dr Dechang Xu for giving us guidance on doing research. We would like to thank Ziang Shi for his help in constructing our wiki webpages. This work was sponsored by Syn-bio Tech Ltd in Suzhou, and we acknowledge Snapgene that offered updated vision’s software to all iGEM teams.

Financial Disclosure

SynbioTech http://www.synbio-tech.com.cn/

The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing Interests

The authors have declared that no competing interests exist.

Ethics Statement


Data Availability

Yes – all data are fully available without restriction

The data can be found in notebook in iGEM wiki.

  1. Nice work guys, if you were to continue can you think of ways you might test your hypothesis that there were problems with the secondary structure of your RNAs?

    Perhaps one idea is that there are a number of techniques which can be used to test RNA secondary structure now such as SHAPE-SEQ (http://www.pnas.org/content/108/27/11063.long)

    Future suggestions when writing manuscripts:
    *It would be great to see error bars! letting readers know the number of times an experiment is repeated, and the variance.
    *It might help to provide a little more context in terms of references, what has been done before with these systems, how your results compare.
    *In the future you might try to write more generic methods sections, where certain procedures such as growth of e.coli are mentioned once.
    *Labeling all columns and rows in figures 8 to 14 would help interpretation.

    Anyway, well done!

  2. RNA thermometers have potential to serve as effective synthetic biological parts as its simplicity and sensitivity to temperature changing. It turned to be that the performance was not stable and accurate as expected. This has been always in the case in many studies and in our own research. After several failed attempts, we switch to temperature sensitive promoters, which seem to be more reliable.

    You can evolve the temperature sensitive promoters to generate temperature responsive promoters.

Leave a Reply

Your email address will not be published. Required fields are marked *

Add your ORCID here. (e.g. 0000-0002-7299-680X)

Back to top