This web page was produced as an assignment for an undergraduate course at Davidson College.
Review of:
Programmable ligand-controlled riboregulators of eukaryotic gene expression
Travis Bayer, Christina Smolke
Nature Biotechnology vol.23 n3 March 2005 337-343.
This study draws on the field of RNAi, and recent documentation of trans- acting RNAs. Using the knowledge of anti-sense RNA's inhibitory effects on translation, Bayer and Smolke have designed small, ligand-dependent RNAs that act as a switch that can turn on or off translation of an mRNA transcript; they can them "antiswitches." The design is elegant: a single strand of RNA that can base-pair with itself two form two different double-stranded conformations. In one of these conformations, there is a stem of the antisense sequence exposed; this stretch can base-pair with the mRNA transcript of the targeted gene and thus inactivate it. The antisense stem is sequestered in a double-stranded structure in the other conformation, which is more stable in the absence of its ligand. Switching between the two is regulated by binding of a ligand to a specific sequence on the antiswitch called an "aptamer," a sequence of nucleic acids that binds to a ligand with high specificity. The aptamer used in constructing the antiswitches in this paper were specific for theophylline.
Figure 1a gives a diagram of how this switching works, while 1b shows the equilibrium between the two forms, and how the antisense strand is thermodynamically favored to be exposed upon binding the ligand theophylline. 1c shows the relative GFP expression under four different conditions. The first, addition of the aptamer and theophylline, acts as a positive control. Since the aptamer only binds theophylline and does not interact with the mRNA transcript, there is no depression of GFP levels. Addition of the s1 antiswitch with caffeine serves two purposes. First, it shows the background decrease of GFP due to levels of the antisense strand conformation arising simply from the equilibrium. Second, because there is effect in increasing the concentration of the caffeine, it displays the specificity the aptamer has, as theophylline and caffeine differ by a single methyl group. Addition of the antisense strand alone with theophylline served as the negative control. With no other part of the construct to sequester its inhibitory effects, the antisense strand base-paired with the transcript readily and dramatically decreased the levels of GFP. The only treatment that varies with the effector concentration is s1 with theophylline. After a very slow decrease from 0.01 mM to about 0.9 mM, there is a sharp, almost asymptotic decline in GFP expression down to the levels of the antisense strand alone. In effect, this figure shows that the antiswitch works in turning off a gene. This line shows that the s1 antiswitch allows for some expression of the targeted gene when in its "off" position, while converts almost entirely to the antisense conformation in the presence of its ligand in sufficient concentration. 1d further supports this idea, showing the GFP expression of two cultures with the antiswitch. Both were grown to a steady level of both GFP and s1, at which point 2 mM theophylline was added to one culture. Before this addition, the two cultures showed almost identical expression levels of GFP. Afterwards, the culture with s1 only continued to increase before reaching a more or less steady state. The other culture initially had a lower rate of increase before GFP expression began to drop to background levels at a rate equivalent to the half life of GFP. This means that, while s1 does not destroy GFP, it does not allow for the production of any more. While not absolutely necessary, more of the treatments, such as an s1 culture with caffeine added, or a wild type culture with theophylline would have added strength to this graph. Figure 1e is a radiograph used to show differences in s1 mobility when it binds theophylline. All five lanes of increasing theophylline concentrations show smears with a single broad band. In the 0.2 and 2 µM lanes, the band has traveled slightly farther than the band in the 10, 20 and 200 µM lanes. Because there is a shift in mobility between 2 and 10 µM, it can be assumed that binding occurs at a concentration in between. What this figure is lacking, however, is a negative control lane of no theophylline and a positive control lane of about ten times the most theophylline used in the other experiments. Perhaps s1 has multiple binding sites and the first is bound at both 0.2 and 2 µM. Or perhaps there is a second binding site at a greater concentration. Indeed, Figure 2b-d seems to show two equivalence points for s1, one at 0.1 mM, and the other at 1 mM theophylline. While Bayer and Smolke state that the in vivo studies use the media concentration of the effector, which has been shown to be as many as a thousand times greater than the intracellular levels, it is still odd that they did use higher concentrations here. They would be able to make their statement about both the structure and function of s1 stronger by the addition of these control lanes.
Figure 2 examines the tunable nature of the antiswitch design. By changing the nucleotides involved in the antisense loop, the free energy associated with self-base pairing changes, making the construct more or less stable. 2a shows s1 again and three variants, s2-4. Although they do not all have the same aptamer stem, all four have the same free energy for it. 2b-d are graphs of more or less the same experiment separated for ease of reading; they all compare the variant constructs to s1, whose function was more rigorously examined in Figure 1. 2b shows that s2 and 4 have background repression of GFP similar to that of s1, though they react to shut off the gene at much lower concentrations, and with not as sharp a curve. s3, on the other hand, has a very similar range over which it reacts, and a very similar degree of steepness, but it has only 10% background repression of the gene, versus s1's 30%. In 2c, s5 and s6 likewise had decreased background repression of GFP, but both this time bound theophylline at higher concentrations and with only slightly less steepness. In 2d they show that different aptamers can be used for the same effect. The s7 antiswitch is responsive to tetracycline, not theophylline. Its curve is much like s1, showing slightly less initial repression, and no real change until 1 mM of effector, at which point GFP expression dropped almost to zero.
Figure 3 shows the flexibility of the antiswitch design. Whereas all the constructs before have been turning genes off, Bayer and Smolke show with their s8 construct that the same idea can be used to turn on repressed genes. Again they changed the nucleotides in the two stems so that the antisense stem was exposed in the more stable conformation, and hidden upon effector binding, as 3a diagrams. 3b is another graph showing expression of GFP relative to the control, comparing s1 and s8. While s1 starts high and ends low, s8 starts low and ends high, with both reacting at a little before 1 mM theophylline.
At last, Figure 4 shows that multiple antiswitches can be used in tandem when they react to different species. 4a shows a scheme of the combination, with both constructs serving as off switches, one for GFP under the control of theophylline, and the other for YFP (not Your Favorite Protein, mind you) under the control of tetracycline. In 4b they show that one gene can be turned off at a time, or both together depending on which effector is added. At no theophylline of tetracycline, both genes show maximum expression. When 5 mM theophylline is added, GFP expression drops to near zero, while YFP remains unchanged. The reverse is true when 5 mM tetracycline is added, with GFP at max levels and YFP very low. When 5 mM of both effectors was added, both genes were expressed at minimal levels.
Overall the paper gives very strong evidence for an exciting development in gene regulation. Its data show convincingly that their constructs turn off their target genes as Bayer and Smolke say they do. What they do not look at, however, is what other genes are turned off. After all, when the antisense stem is sequestered, the aptamer stem is exposed, and that that in itself could base-pair to a non-targeted transcript and effectively silence it. To look at this, they should perform a Northern blot using the aptamer stem as the probe to see if any genes would be at risk. Any banding would mean that they are potentially turning off genes that they do not want to silence. Before trying out the technique in clinical applications, as they suggest, they definitely would want to know if there would be such side effects. If such treatment would be safe, one immediate avenue of study should be microbiology and virology. Among the first genes transcribed upon pathogen invasion are its transcription factors. Since these naturally bind to DNA, it would not be difficult to imagine engineering an aptamer that used the TFs as its effector. Once activated, it would expose the antisense stem that would bind to the transcript for that very same TF, thereby inactivating it and preventing propagation. Before this method could work, though, naturally there would have to be an effective means to get the antiswitch into the target cells; cloning a vector into a yeast colony is rather different than trying to do the same to, say, hepatocytes in vivo . On the other hand, if this were to be used against viruses, one could engineer the viral shell to contain the DNA for the antiswitch and let it target the same cells that the pathogenic form would. Regardless, the new method shows great potential in clinical applications.