This web page was produced as an assignment for an undergraduate course at Davidson College

A Critical Reading

Bayer T. S. and Smolke C. D. (2005) Programmalbe ligand-controlled riboregulators of eukaryotic gene expression. Nature Biotechnology 23:337-343.


Non-coding RNA has recently come into the limelight as an important regulator of gene expression at the translational level. Bayer and Smolke describe in their paper, an engineered trans-acting antisense RNA that binds to target mRNA sequences. The antiswitch, as they call it, is made of two parts. First there is an aptamer binding domain in the RNA that binds specific ligands. The ligands may be dyes, proteins, peptides, antibiotics, or other small molecules. In their investigation, Bayer and Smolke focus their work on an antiswitch with an aptamer binding region specific to theophylline. When the ligand, in this case, theophylline, binds to the aptamer region of the antiswitch RNA, the antiswitch changes shape. Here, the second part of the antiswitch, the stem, carries out the translational regulation. In the absence of the ligand, the antiswitch is in a shape in which the antisense RNA sequence (antisense to the target mRNA) of the switch has bound to itself, forming double stranded RNA. In this conformation, the antiswitch cannot bind to the target mRNA. In the presence of the ligand, however, the antiswitch changes shape and the once double stranded stem is now single stranded, and the antisense RNA can bind to the target mRNA and block translation (Fig. 1a and b). Translation is blocked because ribosomes cannot bind to the mRNA; thus, this is type of mRNA regulation is called a riboregulator. With this basic antiswitch riboregulator design, Bayer and Smolke engineer an antiswitch specific to GFP mRNA and measure the effectiveness of the antiswitch as a translational regulator by way of GFP production.

Data Results and Critique

The antiswitch first is shown to have a higher affinity for the antisense stem conformation, or the shape in which the antiswitch basepairs to itself (Fig. 1b). There must be a higher affinity for the antisense stem so that mRNA binding will happen only in the presence of the ligand and the conformation change of the antiswitch. This allows the antiswitch to be regulated by ligand presence.

The next figure, Fig. 1c, shows GFP expression levels as a product of ligand concentration. Three constructs and two ligands were used in the in vivo experiment. The first construct, represented in green is just the aptamer part of the antiswitch, which was not designed to interact directly with the target mRNA, and the ligand theophylline. It is used as the negative control and the baseline of the relative GFP expression in which the other constructs are compared to in this assay and all others testing relative GFP expression levels. As was expected, GFP expression remained relatively the same despite the increased concentration of theophylline present showing that the aptamer does not inhibit GFP mRNA translation. The second experiment tests the just the antisense part of the antiswitch, which is designed to bind to the GFP target mRNA and block translation, and the ligand theophylline. The positive control blocked nearly all GFP expression, as expected. The negative and positive results also show that theothylline does not inhibit GFP translation by itself, and that the antisense riboregulator sequence is necessary to block GFP expression.

The third test is also a labeled as a negative control. It is the complete antiswitch with both the aptamer and antisense constructs present (labeled s1.) GFP expression was measured in the presence of the ligand caffeine, which differs from theophylline by a single methyl group, but does not bind to the s1 aptamer. As the graph shows, the relative GFP expression level is decreased by 30% with s1 and caffeine. The authors do not attempt to explain the negative control’s decrease in GFP expression. Without controls for the caffeine ligand, the reader is left to assume that the 70% GFP expression is a product of antiswitches in the aptamer stem conformation – though the antisense stem conformation has a higher affinity, 100% of the antiswitches will not always be in the antisense state. Therefore, there will be some relative decreased GFP expression in the presence of the antiswitch. The authors, however, do not explain this concept in the paper. The s1 construct was used again, but this time in the presence of the activating ligand, theophylline. At low concentrations, GFP expression was similar to that of the s1/caffeine combination. However, between the concentration levels of 0.75mM and 0.8mM of theophylline, the GFP expression level dropped drastically to background noise levels.

This one figure illustrates the specificity of aptamer and ligand binding, the necessity of the ligand for sntiswitch inhibitory function, the on/off quality of the antiswitch in the presence of the ligand, and the appropriate concentration of theophylline to turn GFP expression off, relatively speaking.

Using a little over twice the predicted effective ligand concentration (2mM), Bayer and Smolke looked at in vivo GFP expression by time (Fig. 1d). Both experimental groups contained antiswitch s1. One group received a dose of theophylline treatment, and the other group did not. In the control group that did not receive the treatment, the GFP expression levels continued to increase. The experimental group that received a one-time dosage of the ligand showed a decrease in GFP expression. The expression levels, according to Bayer and Smolke, decreased at a rate corresponding to a half life of GFP protein, about 0.5 to 1 hour. This experiment shows the immediate effectiveness of the antiswitch at inhibiting GFP translation.

An in vitro gel-shift experiment was conducted to show the affinity of the antiswitch, the ligand, and the target mRNA (Fig. 1e). The band shift from the 2um to 10um concentration in the presence of equal amounts of antiswitch and target mRNA suggests that both the ligand and the target cooperate in conformational changes of the antiswitch. This assay would be more powerful if the investigators then tested the amount of target needed to make the same conformational changes of s1 when keeping the ligand concentration at a constant level, between the 2um and 10um they determined in the first gel-shift assay. Though the gel is very blurry, the reader can see that there is one band in each lane, and that there are two different distances in the bands – 10, 20, and 200 um are shorter than the 0.2 and 2um lanes. Interestingly, the band in the 10um lane is noticeable darker than the other bands, and the 200um band appears to be the second darkest. This is very contradictory as the experimental design called for equal amounts of labeled s1 in each experiment. The poor quality and questionable bands of this gel call the results of the assay into question.

Bayer and Smolke next reasoned that they should be able to manipulate the thermodynamics of the antiswitch by changing the antisense part of the construct. Antiswitches s2-s4 illustrate the changes (Fig. 2a) in the RNA sequence and the changed binding affinities. For each anitswitch s1-s4, relative GFP expression was measured over a range of theophylline concentrations. Antiswitches with higher antisense stem affinities required a higher concentration of ligand to make the conformational change. Antiswitches with lower antisense stem affinities required a lower ligand concentration to change shape (Fig. 2b).

Antiswitches s5 and s6 were manipulated by changing the aptamer binding domains. s5 and s6 was made with an aptamer with a lower affinity to theophylline than s1. s6, with the decreased affinity aptamer, also was manipulated to have a lower affinity antisense stem like that of s2. As shown in Fig. 2c, an increased concentration of theophylline was required for both s5 and s6 than s1. s6 required a lower concentration than s5 because of the decreased affinity of the antisense stem (supported by Fig. 2b).

Yet another antiswitch was made, s7, but this time a new aptamer construct was used, able to bind to tetracycline. This aptamer has the about the same affinity to tetracycline as the theophylline apatamer used in s1-4. Fig. 2d shows the similar GRP expression of s1 and s7 over a concentration of the appropriate ligand. This figure suggests that antiswitches made of different constructs but with similar affinity levels will respond in a similar way.

Figures 2b-d all show the modularity of the antiswitch. By altering the antisense or aptamer constructs of the antiswitch, the approximate concentration of ligand required to make the antiswitch conformational changes according to the law of thermodynamics. In other words, If the antisense stem has a higher affinity than the aptamer stem, more ligand is required to make the conformational change from antisense to aptamer, and vice versa. If the aptamer domain has a lower affinity for the ligand, than a higher concentration of the aptamer-binding ligand is required to change the structure of the antiswitch.

Bayer and Smolke go even further to make an antiswitch s8. This switch is different from the others, however, because it has a higher affinity for the antisense stem state than the aptamer stem state (Fig. 3a). This means that in the absence of the ligand, the antiswitch binds to the target mRNA to stop translation. In the presence of the ligand, the antiswitch basepairs to itself and does not inhibit GFP translation. This is, in essence, the opposite of s1 function. Yet another assay was conducted of relative GFP expression of s8 compared to s1 (Fig. 3b). The expression profiles for the two antiswitches are exactly opposite; when s1 inhibits GFP expression at high theophylline concentration, s8 inhibits translation at low concentrations. The simple reversal of function illustrates the flexibility of the antiswitch design.

The final figure illustrates the specificity of the antisense stem to the target mRNA and the versatile use of the antiswitch. Two antiswitches, s1 with theophylline ligand, specific to GFP mRNA, and an antiswitch with an aptamer specific to tetracycline and targeting yellow fluorescent protein (YFP) mRNA, were both expressed in cells (Fig. 4a). A series of assays measured relative expression dependent on ligand presence (Fig. 4b). In the absence of both ligands, GFP and YFP were expressed in high relative levels. Addition of only theophylline to the cells resulted in a nearly zero relative expression of GFP, while YFP expression was at the control level. When only tetracycline was present, YFP levels decreased to nearly zero, while GFP levels remained at the control level. In the presence of both theophylline and tetracycline, relative expression levels of GFP and YFP were nearly zero.

The functional assays measuring GFP expression in the cells were very effective, even though it was an non-direct way of measuring the effectiveness of mRNA translation. The most direct way to measure translational inhibition would be to measure amount of protein present in the cells. This would require at the least, a western blot and labeled GFP antibodies. It would also show if there was nonfunctional GFP protein present in the cells, possibly partially translated despite the presence of a riboregulator. This could not replace the GFP expression assays, as those are a more functional and probably more accurate measurement of functional GFP protein presence. However, an assay similar to the western blot outlined above would have been helpful in assuring the reader that a nearly zero relative expression level reading did in fact show that translation was inhibited.

A careful reader notices that relative expression levels for controls in the assays seldom equal a relative expression level of 1.0. This is never explicitly addressed in the paper. The reader is left to assume that only an antiswitch with extremely high antisense affinity levels would result in a relative expression rate of 1.0, and that the lower the antisense affinity level the lower the baseline GFP expression level is in the absence of the ligand. Comparison to the negative control in Fig. 1c was important to note in all the assays, but controls should have been used throughout the investigation to look at the GFP expression of only the antiswitch, in the absence of the ligand.

On the whole, the investigation was well organized. They addressed, and found results, for all their questions: quality of inhibition of translation by the antiswitch and appropriate ligand (Fig. 1c, d, & e), the thermodynamic properties of the antisense stem and aptamer binding domain (Fig 2b, c, & d), the flexible and reversible properties of the antiswitch (Fig. 3b) and the mRNA target, ligand, and antiswitch specificity and the complex system potential for the antiswitch (Fig. 4b). It was a successful investigation, but it now raises more questions.

Further Research

Bayer and Smolke suggest that antiswitches have a variety of practical application. The most obvious, and the one I will be focusing on in this section, is for gene therapy. In order to move toward gene therapy using antiswitches, researchers must start by looking at antiswitch technology in more complex organisms and in its effectiveness at inhibiting translation of overexpressed genes.

Antiswitches are based on thermodynamics, and therefore could be susceptible to changes in the cell environment, such as heat and pH, both of which, theoretically, would change the affinity of the antisense stem and ligand binding. The same assays used in Bayer’s and Smolke’s experiments could be done under varying heats and pH conditions. This would introduce some of the variables that would be seen in a complex organisms which, ideally, would be receiving the therapy. Next, the antiswitch should be used in a multicelled organism, such as C. elegans. Relative expression levels of GFP could be measured in a similar way as the levels in the yeast, to determine the antiswitch’s success. Cell localization assays for inhibited expression would be next.

Antiswitch gene therapy would have to be limited to diseases where a gene overexpresses a product because antiswitches do not induce expression, nor do they repair gene mutations. Identifying a testable disease – one with a single overexpressed gene product – in a simple organism such as a C. elegan or D. melanogaster would allow researchers to test this practical application of the antiswitch in a simple way. An antiswitch would have to be made to target the overexpressed protein’s mRNA, and an aptamer and ligand would have to be chosen that would not be toxic to the organism and that would easily reach target cells. GFP expression, measured by fluorescence obviously would not be appropriate, so either a quantitative measure of gene product or a relative level of phenotype expression would have to be done.

These are only the first steps in a process that will be long, but hopefully, promising.


Megan's Molecular Home Page

Molecular Biology Home

Davidson Home

Contact Megan at