Nature-inspired engineering of an artificial ligase enzyme by domain fusion

Abstract

The function of most proteins is accomplished through the interplay of two or more protein domains and fine-tuned by natural evolution. In contrast, artificial enzymes have often been engineered from a single domain scaffold and frequently have lower catalytic activity than natural enzymes. We previously generated an artificial enzyme that catalyzed an RNA ligation by >2 million-fold but was likely limited in its activity by low substrate affinity. Inspired by nature's concept of domain fusion, we fused the artificial enzyme to a series of protein domains known to bind nucleic acids with the goal of improving its catalytic activity. The effect of the fused domains on catalytic activity varied greatly, yielding severalfold increases but also reductions caused by domains that previously enhanced nucleic acid binding in other protein engineering projects. The combination of the two better performing binding domains improved the activity of the parental ligase by more than an order of magnitude. These results demonstrate for the first time that nature's successful evolutionary mechanism of domain fusion can also improve an unevolved primordial-like protein whose structure and function had just been created in the test tube. The generation of multi-domain proteins might therefore be an ancient evolutionary process.

Figure 1.

Overview of increasing the catalytic activity of an artificial RNA ligase by fusing the enzyme to terminal substrate-binding domains (SBD). (A) The artificial ligase 10C catalyzes a splinted RNA ligation reaction. The 5′-triphosphorylated RNA is ligated to the 3′-hydroxyl of a second substrate in the presence of a complementary DNA or RNA splint. (B) A variety of nucleic acid binding domains (4–262 amino acids long) were fused to the N- and/or C-terminus of the artificial RNA ligase 10C (87 amino acids long). The fusion proteins were screened for increased ligation activity.

Figure 2.

Comparison of ligation activity of ligase 10C with substrate-binding domain fusions of ligase 10C. The fusion proteins are grouped and color-coded according to the type of their binding domain, similar to Table 1. Only fusions with a single substrate-binding domain are shown here. Ligation yields were determined after 1 h incubation. Assays contained 5 μM protein, 10 μM PPP-substrate, 15 μM DNA splint (DNA splint #1), and 20 μM OH-substrate (RNA-OH #1). The error bars represent the standard deviations of the means of four replicates. *P < 0.05 and ***P < 0.001 indicate statistically significant differences of ligation yield when compared to the ligation yield for unfused ligase 10C.

Figure 3.

Time course of ligation reaction for promising ligase 10C fusions identified in initial single time point screening (Figure 2). Observed rate constant of ligase 10C alone was determined for comparison (dotted line). The best performing individual N- and C-terminal binding domains were combined and assayed as ligase 10C fusions en-10C-R₄ and rgI-10C-R₄. Assays contained 1 μM protein, 10 μM PPP-substrate, 15 μM DNA splint (DNA splint #2) and 20 μM OH-substrate (RNA-OH #2).

Figure 4.

Time course of ligation reaction for the most active fusion enzymes. Observed rate constants were determined for ligase en-10C-R₄, ligase 10C-R₄ and the original ligase 10C without terminal substrate-binding domains. Assays contained 10 μM PPP-substrate, 15 μM DNA splint (DNA splint #2), 20 μM OH-substrate (RNA-OH #2) and either 5 μM ligase 10C, or 0.5 μM ligase 10C-R₄ or ligase en-10C-R₄. The error bars represent the standard error of the mean from two biological replicates with at least three technical replicates for each biological replicate.

Figure 5.

Initial reaction rate of reaction (v₀) versus substrate concentration for ligase en-10C-R₄. Assays contained 0.5 μM of enzyme together with PPP-substrate, DNA splint (DNA splint #2) and OH-substrate (RNA-OH #2) at a final molar ratio of 1:1.5:2. The error bars represent the standard error of the mean from two biological replicates with three technical replicates for each biological replicate. The data were fitted to the Michaelis–Menten kinetics model.

Figure 6.

Binding affinity of ligase variants investigated by fluorescence anisotropy (FA). (A) RNA and complementary fluorescein-labeled DNA oligonucleotide (RNA/DNA duplex) used in FA assay. (B) Comparison of binding affinities of ligase 10C and its domain fusions en-10C-R₄ and 10C-R₄ to the RNA/DNA duplex. Two-fold dilution series of the enzymes were prepared in ligation buffer with 100 μg/ml bovine serum albumin and 30 nM of the annealed RNA/DNA duplex in 1:1 molar ratio. The error bars represent the standard error of the mean from two biological replicates with three technical replicates for each biological replicate. (C) Saturation binding curve of ligase en-10C-R₄ calculated from FA data. The saturation curve was fitted to the direct binding model.