Design principles for ligand-sensing, conformation-switching ribozymes

Abstract

Nucleic acid sensor elements are proving increasingly useful in biotechnology and biomedical applications. A number of ligand-sensing, conformational-switching ribozymes (also known as allosteric ribozymes or aptazymes) have been generated by some combination of directed evolution or rational design. Such sensor elements typically fuse a molecular recognition domain (aptamer) with a catalytic signal generator (ribozyme). Although the rational design of aptazymes has begun to be explored, the relationships between the thermodynamics of aptazyme conformational changes and aptazyme performance in vitro and in vivo have not been examined in a quantitative framework. We have therefore developed a quantitative and predictive model for aptazymes as biosensors in vitro and as riboswitches in vivo. In the process, we have identified key relationships (or dimensionless parameters) that dictate aptazyme performance, and in consequence, established equations for precisely engineering aptazyme function. In particular, our analysis quantifies the intrinsic trade-off between ligand sensitivity and the dynamic range of activity. We were also able to determine how in vivo parameters, such as mRNA degradation rates, impact the design and function of aptazymes when used as riboswitches. Using this theoretical framework we were able to achieve quantitative agreement between our models and published data. In consequence, we are able to suggest experimental guidelines for quantitatively predicting the performance of aptazyme-based riboswitches. By identifying factors that limit the performance of previously published systems we were able to generate immediately testable hypotheses for their improvement. The robust theoretical framework and identified optimization parameters should now enable the precision design of aptazymes for biotechnological and clinical applications.

Figure 1. Schemas for aptazyme design.

(A) The general strategy for designing aptazymes, where the aptamer and the ribozyme are shown in blue and red, respectively. The stem used to connect the aptamer and the ribozyme (the communication module) is highlighted in a dotted green box. (B) Schema for ‘binding-assisted stem-formation.’ (C) Schema for a ‘slip structure.’ (D) Schema for ‘strand replacement’. In (B) to (D), the aptamer domain, the ribozyme domain and the communication module are shown in blue, red and green, respectively. The ligand for the aptamer domain is shown as a blue hexagon. Long gray lines indicate base-pairing; short grays lines (on the left) indicate un-paired bases; and dashed gray lines (on the right) indicate mis-paired bases or non-canonical base-pairs.

Figure 2. Kinetic model and performance of aptazymes as in vitro biosensors.

(A) Two-state model for aptazyme function. The aptazyme conformers with low and high affinities for ligand are shown as A and B, respectively. The ligand-bound states of these two conformers are shown as AL and BL. K _int is the equilibrium constant for the A-to-B transition. K _a(A) and K _a(B) are the association constants for the ligand (L, shown as a blue hexgon) with the A conformer and the B conformer, respectively. The first-order cleavage rate constants for conformer A and conformer B are defined as k _Cle(A) and k _Cle(B), respectively. Under certain conditions (see text), the AL conformer can be ignored and the model can be reduced to the enclosed green box. (B) The effect of cleavage tendency (ω) on the performance of ligand-activated aptazymes. (D) The effect of cleavage tendency (ω) on the performance of ligand-inhibited aptazymes. In (B) and (D) the relative ligand concentration ([L _tot]/K _d) is shown on the horizontal axis and the relative apparent cleavage rate constant (k _app/k _Cle) is shown on the vertical axis. The basal cleavage rate constants in the absence of ligand are shown by horizontal dashed lines. The values of are shown as a vertical dotted lines. (C) The relationship between the cleavage tendency (ω) of an aptazyme and the realistic ligand-dependent change in activity () is shown for ligand-activated aptazymes (left) and ligand-inhibited aptazymes (right). This relationship is shown for different maximum available ligand concentrations ().

Figure 3. Models of aptazyme-based riboswitches.

(A) Kinetic model for mRNA metabolism in the absence of ribozyme or aptazyme cleavage. (B) Kinetic model for mRNA metabolism when a constitutively active ribozyme is inserted into the 3′ UTR. (C) Kinetic model for mRNA metabolism when a ligand-activated aptazyme is inserted into the 3′ UTR. (D) Kinetic model for mRNA metabolism when a ligand-inhibited aptazyme is inserted into the 3′ UTR.

Figure 4. Performance of aptazymes as riboswitches.

(A) Ligand-activated aptazyme; (B) Ligand-inhibited aptazyme. The relative ligand concentration (ligand concentration divided by the of the aptamer) and the relative gene expression level (steady-state mRNA concentration of the aptazyme-harboring gene divided by that of the un-engineered gene) are shown on the horizontal and vertical axes, respectively.

Figure 5. Regulatory landscapes for aptazyme-based riboswitches.

(A) A guide to interpreting the regulatory landscape figures. When the type of aptazyme (ligand-activated or ligand-inhibited) and the values of D and are all known, the regulatory landscape of a given riboswitch can be determined. The left panel shows an example of a ligand-inhibited aptazyme with and . Different cleavage tendencies (ω, horizontal axis) and relative gene expression levels (vertical axis) are related by relative ligand concentrations (indicated by color mapping; the color scale is shown on the top of the right panel). The dynamic range of activities for a given ω can be determined by drawing a vertical line and looking at the relative gene expression levels at relative ligand concentrations 0 and 100. Such a vertical line is shown for a ω of 0.7, and the dynamic range (the ratio of the values at points and ) is ca. 6-fold. The relative concentration at point corresponds to the of this riboswitch. The dashed gray line represents the span of relative EC ₅₀ () values. (B to D) The regulatory landscapes of ligand-activated aptazymes with different D values ( = 10 and 100) but the same value ( = 100). Note that panel D is an expanded view of panel C where ω varies from 0 to 0.1 (instead of 0 to 1). (E to F) The regulatory landscapes of ligand-inhibited aptazymes with different D values ( = 10 and 100) but the same value ( = 100).

Figure 6. Quantitative relationships between and cleavage tendency ω.

For ligand-activated aptazymes (A) and ligand-inhibited aptazymes (B), is defined as fold-inhibition and fold-activation of gene expression across the realistic dynamic range of gene expression. The relationship between and cleavage tendency ω is shown for different, maximum possible relative ligand concentrations (; shown in different colors).

Figure 7. Analysis of an aptazyme-based riboswitch in vivo (Win and Smolke (2007) [14]).

The regulatory landscapes for ligand-activated (A) or ligand-inhibited (B) aptazyme-based riboswitches are shown. In these landscapes the value of D is 10, and the value of is 12. The published relative expression levels of the designed aptazymes in the absence of ligand (pink circles) and at saturating concentrations of ligand (pink triangles) are plotted versus the calculated cleavage tendency [using equations (25) and (35)]. (C) The discrepancy between designed cleavage tendencies (see Methods) and observed cleavage tendencies ( Table 1 ).