Circular permutation prediction reveals a viable backbone disconnection for split proteins: an approach in identifying a new functional split intein

Abstract

Split-protein systems have emerged as a powerful tool for detecting biomolecular interactions and reporting biological reactions. However, reliable methods for identifying viable split sites are still unavailable. In this study, we demonstrated the feasibility that valid circular permutation (CP) sites in proteins have the potential to act as split sites and that CP prediction can be used to search for internal permissive sites for creating new split proteins. Using a protein ligase, intein, as a model, CP predictor facilitated the creation of circular permutants in which backbone opening imposes the least detrimental effects on intein folding. We screened a series of predicted intein CPs and identified stable and native-fold CPs. When the valid CP sites were introduced as split sites, there was a reduction in folding enthalpy caused by the new backbone opening; however, the coincident loss in entropy was sufficient to be compensated, yielding a favorable free energy for self-association. Since split intein is exploited in protein semi-synthesis, we tested the related protein trans-splicing (PTS) activities of the corresponding split inteins. Notably, a novel functional split intein composed of the N-terminal 36 residues combined with the remaining C-terminal fragment was identified. Its PTS activity was shown to be better than current reported two-piece intein with a short N-terminal segment. Thus, the incorporation of in silico CP prediction facilitated the design of split intein as well as circular permutants.

Figure 1. Predicted CP probabilities (CPred scores) for Npu DnaE (NpuInt), Ssp DnaE and Ssp DnaB inteins (Npu: Nostoc punctiforme and Ssp: Synechchotcystis sp.) plotted versus residue number.

The coordinates of NpuInt (PDB code: 2KEQ) , Ssp DnaE (PDB code: 1ZD7) and Ssp DnaB (PDB code: 1MI8) were submitted to CPred to estimate the CP probability as a function of residue number. The locations of reported functional split sites (efficiency >50%) are indicated by closed black triangles, and asterisks indicate the naturally occurring split sites. The reported non-functional split sites in Ssp DnaE and Ssp DnaB are marked by closed red triangles. Two newly predicted Npu DnaE intein CP sites at residues 12 and 36 are indicated by open black triangles.

Figure 2. Secondary structure of the CP variants (A) CP36 and (B) CP102, evaluated according to the parameter Δδ_Cα−Δδ_Cβ.

The chemical shift values for ¹³C_α and ¹³C_β of CP36 and CP102 were obtained, and Δδ_Cα and Δδ_Cβ were calculated from the differences between the experimental values and random coil values. The value of Δδ_Cα−Δδ_Cβ for each residue represents the average of three consecutive residues, centered at the particular residue. The Δδ_Cα−Δδ_Cβ value derived from native NpuInt C1G (indicated by closed circles) is overlaid onto the CP results for comparison. The corresponding secondary structure of C1G is depicted at the top. The difference (ΔΔδ) in (Δδ_Cα−Δδ_Cβ) between each CP variant and C1G was calculated and is indicated at the bottom of the figure.

Figure 3. Tertiary structural comparison between CP variants and native NpuInt.

(A) Superimposition of the ¹H, ¹⁵N-HSQC spectra of native NpuInt C1G (blue) and CP36 (red). The spectra were taken at 25°C and 600 MHz. The residues with significant shifts are indicated. (B) Variation in the composite backbone NH and ¹⁵N chemical shift perturbation (Δδ_N+NH) obtained from the spectra of CP36 and C1G (upper panel) and CP102 and C1G (lower panel), where Δδ_N+NH = [(Δδ_NH ²+Δδ_N ²/25)/2]^1/2. Δδ_NH and Δδ_N were calculated from the differences for backbone NH and ¹⁵N, respectively. Spatial distribution of residues with significant Δδ_N+NH are mapped onto the NpuInt structure (PDB code: 2KEQ) , as indicated in different colors for (C) CP36 and (D) CP102. Dotted circles indicate the locations of the N- and C-terminal ends, and closed circles indicate the introduced CP sites.

Figure 4. Superposition of the ¹H-¹⁵N HSQC of NpuInt SP and the corresponding CP variant showing the overall correspondence between two spectra.

(A) NpuInt CP36 (blue) and NpuInt SP36 (red). (B) NpuInt CP102 (blue) and NpuInt SP102 (red).

Figure 5. Thermal stability of intein variants monitored by CD ellipticity at 224 nm.

(A) Far-UV CD spectra of NpuInt variants at 25°C. (B) The two-state thermal denaturation profiles of NpuInt variants. (C) Temperature dependence of the denaturation thermodynamics using van’t Hoff analysis.

Figure 6. In vitro protein trans-splicing (PTS) assay.

Time course of the protein ligation of GB1 and GB1 by (A) the naturally occurring split intein SP102 and (B) the engineered split intein SP36. (C) PTS kinetic analysis of the ligated product of GB1 duplication from SDS-PAGE after reaction using SP102 (blue line), SP36 (red line) and SP12 (green line). The schematic plot of the PTS reactions is depicted at the side.

Figure 7. ITC measurement representing the desired specificity of the following combinations of the N- and C-termini of inteins: SP102^N/SP102^C, SP36^N/SP36^C, SP102^N/SP36^C and SP36^N/SP102^C (from left to right).

Figure 8. Predicted CPred profiles of different split-protein systems: (A) green fluorescence protein (GFP), (B) β-lactamase, (C) dihydrofolate reductase (DHFR), (D) ubiquitin, (E) firefly luciferase and (F) ribonuclease A (RNase A).

The locations of reported split sites are indicated by red triangles.