*This page was created as an assignment for an undergraduate course at Davidson College*


Review Paper

Programmable aligned riboregulators of eukaryotic gene expression

Travis S Bayer & Christina D Smolke

Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, CA 91125

Nature Biotechnology (2005) 23: 337-343


Summary and Critique

This paper investigates the role of noncoding RNA in regulation of gene expression. In nature many types of noncoding RNA can regulate expression through different pathways, but the authors for this paper focus on small trans-acting RNA (taRNA), which bind mRNA to regulate gene expression. The authors engineered antiswitches, synthetic constructs of this type, which are made of two parts: an aptamer stem and an antisense sequence. aptamers, sequences made of nucleic acid, bind specific ligands. When the aptamer binds its ligand, it causes conformational change in the antiswitch. Originally, the antisense sequence is double bound When the aptamer binds the ligand, the antisense sequence becomes single stranded, which frees it to regulate gene expression through mRNA binding. Potentially, scientists could regulate a wide range of genetic systems through specifically engineered antiswitches.

Figure 1

In part (a), the authors give a cartoon illustration of the antiswitch. First, they show the antiswitch in its original state, which cannot bind GFP mRNA (the target sequence for this experiment) in its double-stranded conformation. Consequently, GFP production is on. Then, they illustrate the conformational change that occurs after the ligand binds to the antiswitch. In this state, the antisense sequence becomes single stranded and binds to the GFP mRNA. As a result, GFP production is off. These cartoons provide a helpful visual image to make understanding of the system more intuitive. The aptamer stem and antisense sequence are color-coded blue and red, respectively, which helps clarify the interaction of the sequences. The only confusing part of this diagram is that the authors refer to the ligand as the "effector." Throughout the paper, they alternate between calling it the ligand and the effector. Sometimes, they simply refer to it by the specific molecule name (theophylline or tetracycline). This varying terminology, which they use without explanation, causes unnecessary confusion.

In part (b), the authors give the sequence, in its predicted structural conformation, of the first antiswitch (s1). They also indicate the stabilities of each stem of the switch. The stability of the antisense stem is slightly higher than that of the aptamer one, which implies that in the presence of the ligand (in this case, theophylline) and mRNA, the aptamer stem will form and free the antisense strand to bind the mRNA. They also show the sequence and conformation of the bound antiswitch and mRNA and highlight the mRNA start codon in green. Since this codon binds to the antiswitch, translation and gene expression will not occur. This figure clearly lays out the mechanics of the antiswitch system for s1. Again, it is helpful to see where the exact sequences interact.

In part (c), the authors graphically display the experimental results of in vivo (in yeast) GFP expression in the presence of s1, at different ligand concentration levels. They also graph the results for three control experiments. For the first control, a positive one, they simply inserted the aptamer and theophylline. The aptamer can never bind the mRNA so the cells express GFP (assigned a relative expression level of 1.0) at the same level for all concentrations of theophylline (from 0.01 to 10 mM). For the second control, a negative control, they inserted s1 and caffeine. Caffeine differs from theophylline by only one methyl group, yet does not work as a ligand for s1. Simply adding s1 lowered GFP expression levels to 0.8, but this level remained constant over all concentrations of caffeine. Thus, the aptamer requires a specific ligand to activate the complex. For the third control, a negative control, they inserted antisense and theophylline. As expected, the antisense bound the mRNA and repressed expression at all concentrations of theophylline. The ligand is only necessary for constructs including the aptamer Finally, for the experimental design, they inserted s1 and theophylline. Before any theophylline was added, GFP levels were equal to those of the second control (0.8). At a concentration of 0.8 mM of theophylline, GFP expression levels quickly dropped to effectively zero and remained there for all higher concentrations of the ligand. This quick change in expression levels indicates that the switch does work in an on/off manner rather than in linear continuum. Overall, the authors measurement of GFP expression levels as "relative" levels is somewhat confusing and vague. From the Methods, they quantified expression through fluorescence emission analysis so the measurement is more standardized than it first appears to be. All levels are in a ratio relative to the positive control level. Also, it could have been instructive if the authors had shown some of the actual data, such as glowing cells, rather than simply giving the graphical output.

In part (d), the authors measure the temporal response of s1 after addition of the ligand (theophylline). First, they let cells build up GFP expression levels to steady-state over about 3 hrs. They had two groups of cells: one with s1 and one without it. Then, they added theophylline to both groups and measured GFP expression over time. For the positive control cells lacking s1, GFP levels continued to climb and then leveled off. For cells with s1, GFP levels almost immediately began dropping at the same rate as the GFP half-life (0.5-1 h). Thus, antiswitch quickly activates in the presence of the ligand. Here they measure GFP expression by relative fluorescence united divided by the OD600 of the culture. The switch from measuring expression levels as ratios in part (c) to actual output in part (d) is initially confusing.

In part (e), the authors performed an in vitro experiment to determine antiswitch-ligand affinity. Through immunofluorescence, they assessed labeled s1 mobility in increasing concentrations of theophylline. The authors state that equal amounts of an mRNA target and the antiswitch were present in each lane. However, no loading control appears on the gel so it is impossible to assess if the bands are truly of equal size. The authors claim that s1 experiences a sharp shift in mobility between theophylline concentrations of 2 and 10 microM. If this statement is true, it indicates that somewhere between 2 and 10 microM of ligand is enough for the antiswitch to change conformation and bind the mRNA target. The bound complex would have a higher MW and move less. The gel has no MW control lane though, so again it is impossible to assess if this change in MW has truly occurred. Also, the bands are very smeared, and although they do appear to shift upward at 10 microM, the gel is difficult to read. Furthermore, if unequal amounts of s1 or the target transcript are present in any lane, that may cause more or less binding to occur and thus deceptive shifting. For this gel to be really clear, the authors also should have had a lane with just s1, a lane with just theophylline, a lane with just the mRNA target, and one to establish the standard band position of these molecules.

Figure 2

In part (a), the authors give the sequences and predicted conformations of s1, s2, s3, and s4. The last three antiswitches are constructs they engineered to either be more or less stable than s1. They designed these antiswitches to test the hypotheses that the relative stability of the molecule affects the ligand concentration range over which gene expression levels change. s2 is a destabilized structure that has a single nucleotide change that leads to mismatched bases. Since these bases cannot base pair, stability decreases. In s3, they increased the antisense stem by five nucleotides and formed three extra base pairs in the aptamer stem, which led to a construct with greater stability. In s4, they destabilized the antisense stem by altering the loop sequence. This figure gives the relevant information necessary to evaluate the next figures. It is extremely helpful that they show visual predictions of conformational differences.

In part (b), they measure in vivo levels of GFP expression at varying theophylline concentration for all four constructs. This experiment and figure is very similar to Figure 1 (c). This time s1 is the reference construct since its effect on GFP expression has already been measured relative to positive and negative controls. Initially,both s2 and s4 lead to GFP expression levels analogous to those under s1. However, at 0.1 mM for s4 and 0.2 mM for S2, GFP levels quickly decline. Cells with s1 do not experience drops in GFP levels until 0.8 mM theophylline. This result is expected since s2 and s4 are the destabilized constructs. Conversely, cells with s3 begin with levels of GFP expression higher than those in s1 cells (although still slightly lower than the normal expression level). The authors claim that GFP inhibition does not occur in s3 cells until concentrations of 1.25 mM theophylline. While s3 does appear to begin its most drastic decline in GFP expression slightly after s1, it does not appear to be that far afterwards. Part of the problem here is the scale of theophylline concentration is too large. On a smaller scale, the lines would be farther apart, and thus, it would be easier to interpret the concentrations. Overall, these results demonstrate the malleability of the antiswitch system; these constructs can be individualized to respond to higher or lower levels of ligand.

In part (c), the authors altered the ligand affinity of the antiswitch by changing the aptamer domain (but not stem) to one with a previously characterized, lower theophylline affinity. This new antiswitch is s5. Then, they created s6 by decreasing the stability of the antisense stem (analogous to changes for s2). They then experimentally tested in vivo GFP expression levels in cells with s1, s5, or s6, at increasing levels of theophylline. GFP expression was higher in both s5 and s6 than in s1. As expected, both constructs required almost ten fold higher ligand concentration levels than s1-s4. However, GFP expression in cells with s6 began decreasing a little bit before cells with s5. Since s6 is a less stable construct, it should require slightly lower ligand levels to start binding mRNA. Thus, this figure shows that antiswitch ligand concentration requirements can be altered through changing domains as well as changing stability. Clearly, it would be possible to engineer antiswitches very specific to a certain cell environment or requirements. As always, it would have been nice to have seen actual cell pictures for a visual comparison of GFP levels.

In part (d), the authors developed an antiswitch (s7) that required a different ligand, tetracycline. They then tested GFP expression levels in cells containing s7 as tetracycline concentration increased. They then compared these results to those of s1 cells treated with corresponding levels of theophylline. Cells with s7 started out with GFP levels higher than those in s1 cells. However, the GFP levels began to quickly drop at the same respective concentrations for each ligand (0.8 mM). Overall, the shape of the s7 curve closely mimics that of the s1 curve. These results indicate that this type of response curve is a standard feature of antiswitches and is not just a feature of antiswitches with theophylline for their ligand. Thus, antiswitches in general lead to on/off expression patterns (instead of continuous increasing or decreasing linear patterns). The one problem with this figure is that they only tested for one other ligand. Their results would seems more generalizable to antiswitches as a class if they tested several more ligands. With each ligand tested, the chance that all the systems just happen to behave the same decreases.

Figure 3

In part (a), the authors redesigned the antiswitch to turn GFP expression off in the absence of the ligand. When they added the ligand (theophylline), the antisense sequence became double strands and released the mRNA, so GFP expression turned on. This figure simply depicts the sequence and predicted conformation of the antiswitch (s8). In its default state, the aptamer end is less stable so it prefers to bind mRNA. Although the authors say that they switched the sequence and structure of s8 to make it an "on" switch instead of an "off" switch, it would be instructive if they described exactly what changes in sequence they made.

In part (b), the authors measured in vivo GFP expression levels in cells with s8 over increasing theophylline concentrations. They graphically compared this expression pattern to that of cells with s1. Initially, s8 cells produce no measurable GFP, but around 0.8 mM of ligand they quickly switch over to the "on" position and begin expressing relative GFP levels of 0.8. s1 cells display an expression pattern that is almost exactly a mirror image of s8. Initially, at zero ligand concentration, they express relative GFP levels of 0.7 and then quickly switch to the "off" position around 0.8 mM of theophylline. These results generally demonstrate the flexibility of the antiswitch system and show that it can be engineered to act in a wide range of ways.

Figure 4

In part (a), the authors provide a cartoon illustration of two antiswitch constructs. The first is s1, which binds theophylline as its ligand, after which it binds GFP mRNA and turns the gene off. The second is s9, a new antiswitch, similar in overall design and target. The only differences are that its ligand is tetracycline, and it binds YFP mRNA. Since these antiswitches have different ligands and different target mRNAs, they can be used simultaneously in the same cell. Although these cartoon illustrations make the set up of s1 and s9 very clear, they perhaps are not entirely necessary at this point in the paper. A reader who has been understanding most of the paper should be able to more intuitively understand making an antiswitch to target a different mRNA sequence.

In part (b), the authors tested in vivo the ability to use both s1 and s9 in the same cells to control expression of two different genes at once. At 0 mM concentrations of both ligands, the cells expressed both genes at close to full levels. These expression levels serve as controls for expression at varying ligand concentrations in the rest of the experiment. Since s9 is a new antiswitch that targets a different gene, it could be useful to see a YFG expression graph with positive and negative controls analogous to those in Figure 1 (c). These data would have supported the implicit conclusion that s9 regulates YFP in the same pattern that s1 regulates GFP. At 5 mM theophylline and 0 mM tetracycline, the cells do not express GFP and express YFP at essentially the same level as in the control. At 0 mM theophylline and 5 mM tetracycline, the cells express GFP at control levels and do not express YFP. At 5 mM theophylline and 5 mM tetracycline, the cells do not express either protein. Thus, these results show that two different antiswitch constructs can be used in the same cells to simultaneously, but independently, regulate gene expression for two different genes.

Future Experiments

Clearly, this technology has much room for research and expansion. Performing Western blots on the GFP or YFP protein expressed in all the cells with different antiswitches and varying ligand concentrations would be one immediate experiment that would reinforce the results of this paper. This method is another way to measure the same result (level of protein expression) that the fluorescence emissions tests measured. Quantifying the same result through different methods simply enhances the validity of the results.

Another interesting question is whether antiswitches can be used to modify, but not completely turn off or on, gene expression in cells. At first, it seems unlikely since this paper showed that the antiswitches work in an on or off fashion. However, cells with lower levels of the antiswitch might only have enough of the molecules to turn off some of the mRNA transcripts, for example. To test this idea, I would insert different levels of an antiswitch, through plasmid vectors, into yeast cells. All of the cells would receive the same concentration of ligand. That concentration would be the amount needed to flip the switch (0.8 mM for s1, for example). The outcome measurement would be GFP expression. I would measure the expression levels through cellular observation and pictures, through fluorescence emission, and through Western blotting as described above. If antiswitches can be used to modify but not completely eliminate gene expression, this function could have important implications for gene therapy. Many conditions exist in which patients need more or less of a protein, but not one of the extremes (full expression or no expression).

Also, how long does the ligand effectively work? Is there a point in time when it degrades or unbinds for some reason? To answer this question, I would do a temporal experiment. At time zero, I would insert enough s1 and theophylline into the cells to turn GFP expression off. Then, I would measure GFP expression over time through observation, fluorescence emission, and and protein quantification (blotting). If the ligand does degrade at some point, GFP should start being expressed again. Of course, the length of time the ligand lasts and if it degrades at all will probably be very dependent on the specific system. Thus, this test will probably have to be performed on each new antiswitch and its corresponding ligand. The results to this test have important implications to application of this technology. For some systems, perhaps the gene only needs to be off for a certain amount of time or causes detrimental effects in other parts of the cellular system if it is permanently off. For other genes (for people with an actively detrimental, over expressed protein), it could be best if the gene were permanently turned off. One must be careful about this line of reasoning, however. For example, it may seem desirable to permanently turn off the amyloid protein that causes Alzheimer's. However, this protein plays important functions in other parts of the body. In a situation like this, it may be best to target the transport aspects of the gene so that the protein can reach some areas of the body but will be inhibited from reaching others. This idea could be especially helpful if the target gene has different helper, transport proteins that take it to various locations. Then, one could target the specific transport protein.

This technology has many implications for all kinds of gene therapy and regulation of cellular systems in general. It will probably prove useful for genetic diseases, cancer therapy, other medical treatments, and agriculture. In each specific field, much research will have to go into the specific systems, antiswitches, ligands, and delivery techniques.


Paper Source

Bayer TS, Smolke TD. 2005. Programmable ligand-controlled riboregulators of eukaryotic gene expression. Nature Biotechnology 23 (3): 337-343.


My Home Page

Molecular Biology Main Page

Biology Main Page

© Copyright 2005 Department of Biology, Davidson College, Davidson, NC 28036
Send comments, questions, and suggestions to: lumarcil@davidson.edu