Design
Our objective is to produce protein variants that are more efficient and functional than their wildtype in a faster and more precise manner. We utilize the power of directed evolution, combining computational design with experimental validation to create superior protein systems.
Introduction
Our objective is to produce protein variants that are more efficient and functional than their wildtype in a faster and more precise manner. To reach the goal, we utilize the power of directed evolution, which is divided into two parts: mutation and screening. Our wet lab uses both in vivo and in vitro methods to mutate the Spycatcher, a protein that can form an irreversible covalent bond with its partner peptide, Spytag. We let the Spycatcher and Spytag bind together, and by evaluating the binding rate constant (k) with the luminescence from Luciferase, we can select the one with the highest k as the template for the next round of mutation. After repeating these steps for several generations, we can find the ultimate Spycatcher with the best k performance.
Mutation
In vivo
We designed a similar system inspired by and optimized from the EcORep, an orthogonal DNA replication system, to introduce mutations into the SpyCatcher sequence on the O-replicon (orthogonal replicon, independent from the host cell) inside bacteria.1. The system serves as a mutagenesis platform containing two operons separately, controlling the normal and error-prone replication of the O-replicon.
Figure 1. Overview of the orthogonal replication like system (EcORep-like system).
Schematic illustration of the orthogonal replicon containing the operon and ITR sequences for mutagenesis within bacteria.
Created with BioRender.
We flanked the O-replicon with inverted terminal repeats (ITRs), which can be recognized specifically by TP and ODNAP (orthogonal DNA polymerase), thereby restricting the mutational region to the sequence enclosed between them. The first operon contains the ODNAP sequence for normal replication, which is induced by IPTG, whereas the second operon carries the ODNAP sequence for the error-prone polymerase, which is induced by rhamnose. The inducers activate the promoters, which initiate transcription of sequences, followed by translation into polymerase, regulating the replicon.
Figure 2. Induction of normal and error-prone replication.
IPTG induces normal replication through the ODNAP operon, while rhamnose activates the error-prone polymerase, generating mutations in the target sequence.
Created with BioRender.
With this system, we can verify three aspects: the mutation is induced, the function of the switch is effective, and the orthogonality of the whole system. By controlling different time durations of mutation (on by adding rhamnose), we can determine the relationship between the number of mutated sites and time; comparing the all-on and all-off groups, we can verify that the switch is effective. Finally, orthogonality can be confirmed by showing that the mutation of SpyCatcher on the O-replicon is specifically dependent on the EcORep-like system.
In vitro
The core technique we utilize to mutate the Spycatcher sequence is error-prone PCR (epPCR). Error-prone PCR is a modified form of the polymerase chain reaction designed to deliberately increase the mutation rate during DNA amplification. By using low-fidelity DNA polymerases (such as Taq polymerase) and altering reaction conditions, such as raising Mg²⁺ concentration, adding Mn²⁺ ions, or using imbalanced dNTP pools, researchers can introduce random point mutations into a target gene2,3,4. It enables the generation of diverse mutant libraries that can be screened for variants with improved or novel properties.
We take six top-performing SpyCatcher002 mutant sequences produced by in silico simulations in the dry lab. To overcome potential biases of the deep learning model, we employed epPCR as a directed evolution strategy to explore and identify sequences that were missed by the model.
Figure 3. SpyCatcher002 mutant sequences provided by drylab.
The dry lab simulation system integrates data from mutation and screening experiments to refine model performance through iterative feedback loops.
Created with BioRender.
In light of previous studies, only specific regions of SpyCatcher002 are suitable for mutagenesis5. Therefore, we divide the gene into five fragments (so did the dry lab): two areas (118–180 bp and 253–318 bp) were subjected to epPCR, underwent a couple of iterations, while the remaining three were amplified using colony PCR. The five fragments were then assembled by fusion PCR. In order to combine the plasmid and the complete SpyCatcher, we chose two restriction enzymes, XbaI and BIpI, then conducted PCR for amplification.
Figure 4. Fragment amplification, restriction digestion, and generation of mutated SpyCatcher.
Illustration of the fragment amplification process, restriction enzyme digestion using XbaI and BlpI, and subsequent amplification of the mutated SpyCatcher (318 bp) generated via error-prone PCR with PrimeSTAR GXL DNA polymerase.
Created with BioRender.
Screening
We use Renilla luciferase (RLuc), an enzyme that catalyzes the chemical reaction of luciferin and other substances to produce luminescence, allowing us to evaluate the binding rate constant (k). Utilizing the split luciferase complementation assay (SLCA), RLuc was divided into two fragments: the N-terminal fragment (NLuc, residues 1–229 aa) and the C-terminal fragment (CLuc, 230–311 aa)8. We fuse these two residues to the SpyTag and the other half to the SpyCatcher, tested in different arrangements and combinations. When SpyTag and SpyCatcher are expressed after translation and covalently bind to each other, the two halves of luciferase are recombined into a functional protein, producing fluorescence that can be detected. We use the plate reader to monitor the relationship between time and fluorescence intensity. The instrument enables us to synchronize the starting point (t₀) of the SpyCatcher–SpyTag interaction, allowing for more convenient and accurate measurement of fluorescence kinetics. Then, we compared the rate of signal increase for each SpyCatcher–SpyTag complex. A faster increase in fluorescence (i.e., a steeper slope or sharper rise of the curve) indicates a higher reaction rate constant (k) and thus greater binding efficiency.
Figure 5. Translation of SpyCatcher/SpyTag.
Created with BioRender.Illustration of the fragment amplification process, restriction enzyme digestion using XbaI and BlpI, and subsequent amplification of the mutated SpyCatcher (318 bp) generated via error-prone PCR with PrimeSTAR GXL DNA polymerase.
Figure 6. SLCA and selection.
mRNA translation leads to the formation of the SpyCatcher protein, which folds into its functional structure. The interaction between SpyCatcher and SpyTag is evaluated through a split luciferase complementation assay, where fluorescence intensity reflects binding efficiency (k).
Created with BioRender.
Plus, a critical challenge in screening mutant libraries is distinguishing variants with similar k_on values—when binding kinetics are numerically close, different mutations may produce luminescence curves with similar RLU values that are difficult to differentiate experimentally. To address this, we computationally screened across five different initial concentrations (C₀: 0.01, 0.05, 0.1, 0.5, 1.0 µM) using our six predicted SpyCatcher variants to identify conditions that maximize curve separation. As shown in our simulation results, C₀ = 0.05 µM and 0.1 µM provide the highest discrimination power, allowing subtle differences in k_on to manifest as distinguishable luminescence trajectories.
Figure 7. Computational optimization of detection sensitivity showing simulated luminescence curves for SpyCatcher variants across different initial concentrations. The optimal concentrations (0.05 and 0.1 µM) provide maximum curve separation for distinguishing variants with subtle differences in k_on values.
The screening process was performed after in vitro epPCR. Each cycle, from epPCR followed by selection, was defined as one evolutionary loop. After several rounds of loops, we expect to identify SpyCatcher variants that are several times more efficient and functional than the original. Based on this principle, we can potentially apply this approach in the future to enhance the performance of various proteins or even discover novel, unique enzymes.6,7.