Search
Close this search box.

Development of a new caged intein for multi-input conditional translation of synthetic mRNA – Scientific Reports

Comparison of normal and C-terminally truncated Rma N-inteins

Figure 1 shows a strategy for target protein-responsive translational repression. In the absence of the target protein, the nanobody does not direct the split CaVT protein in close proximity to each other and thus does not induce protein splicing. Consequently, the full-length CaVT is not reconstituted, and the translation of the target mRNA is not repressed. Conversely, in the presence of the target protein, the binding of the nanobody to the target protein brings the split CaVT proteins in close proximity to each other and reconstitutes the full-length CaVT through protein splicing of the caged intein. Then, the reconstituted full-length CaVT binds to the MS2 binding motif in the target mRNA through MS2CP and represses the translation of the target mRNA.

Figure 1
figure 1

Mechanism of protein-responsive translational regulation.

In order to establish a multi-input translational regulation system, the first step is to develop a caged intein that is orthogonal to the previously used caged Npu DnaE intein. Here, we used the Rma DnaB intein as the basis for the design of caged inteins. Because previous studies reported that the N-terminal 106 amino acids (aa) and the C-terminal 51 aa of Rma DnaB intein can be used as N- and C-inteins respectively17,18, we used these regions as the basis for the caged intein design. First, to check the protein trans-splicing efficiency of the split Rma DnaB intein, we used the cage-free N- and C-inteins which can induce spontaneous protein splicing regardless of the target protein. We fused N- and C-inteins to N- and C-terminal fragments of the C46-split CaVT (the CaVT split at the cysteine residue at position 46) respectively, because the C46-split CaVT showed translational repression only when it is reconstituted to the full-length CaVT4.

For this purpose, we designed three vectors. One is the vector to express the N-terminal fragment of C46-split CaVT fused with the normal Rma N-intein (RmaN). The second is the N-terminal fragment of C46-split CaVT fused with the variant of Rma N-intein with the C-terminal 4 amino acid residues removed (RmaN(-4)) whose high protein splicing efficiency was previously reported19. The third one is the C-terminal fragment of C46-split CaVT fused with the Rma C-intein (RmaC). To compare the protein splicing efficiency, the pDNA expressing C46-split CaVT was co-transfected with a firefly luciferase (Luc2) expression vector called pSV40-2xScMS2(C)-Luc2 which containing a strong MS2 binding motif. We also co-transfected control reporter pDNA called pNL1.1TK[Nluc/TK] into HeLa cells to express Oplophorus gracilirostris-derived NanoLuc (Nluc) as a transfection control (Fig. 2a,b). As shown in Fig. 2c, the combination of RmaN(-4) and RmaC caused slightly stronger translational repression than that of RmaN and RmaC. This result suggests a slightly higher protein splicing efficiency of RmaN(-4). So, we used RmaN(-4) and RmaC as a basis for the design of caged intein.

Figure 2
figure 2

Translational repression by the reconstituted CaVT in pDNA transfection. (a) Reporter pDNAs to evaluate translational repression. While the firefly luciferase (Luc2) expression vector called pSV40-2xScMS2(C)-Luc2 contains the target motif for translational repression, the Nluc expression vector called pNL1.1TK[Nluc/TK] lacks the motif and was used as a control. (b) pDNAs of split CaVT with uncaged inteins to check translational repression induced by spontaneous reconstitution. (c) HeLa cells were co-transfected with pSV40-2xScMS2(C)-Luc2 (40 ng/well), pNL1.1TK[Nluc/TK] (20 ng/well), and the pDNAs to express split CaVT (total 40 ng/well). Cells transfected with only one of pcDNA3.1-MS2CP(1–45)-RmaN, pcDNA3.1-MS2CP(1–45)-RmaN(-4), or pcDNA3.1-RmaC-MS2CP(46–116)-VPg(FCV) were used as negative controls for translational repression. The bar graph shows the Luc2/Nluc ratio (mean ± SD, n = 4). **P < 0.01, ***P < 0.001 compared to one of the negative controls (pcDNA3.1-RmaC-MS2CP(46–116)-VPg(FCV)) by Tukey’s multiple comparison test. (d) pDNAs to check the inhibitory effect of designed cages on unconditional reconstitution of split CaVT. (ef) HeLa cells were co-transfected with pSV40-2xScMS2(C)-Luc2 (40 ng/well), pNL1.1TK[Nluc/TK] (20 ng/well), and the pDNAs to express the caged split CaVT (total 40 ng/well). The positive control for translational repression is pcDNA3.1-MS2CP(1–45)-RmaN(-4) and pcDNA3.1-RmaC-MS2CP(46–116)-VPg(FCV) group. For the N-terminal fragment of C46-split CaVT, two types of caged RmaN(-4), named RmaN(-4)cage (e) and Rma(-4)Scage (f) respectively, were tested. The bar graph shows the Luc2/Nluc ratio (mean ± SD, n = 4). ***P < 0.001 compared with the positive control by Tukey’s multiple comparison test.

Prevention of unconditional protein splicing by caging Rma intein

Next, we designed the caged Rma DnaB intein based on the RmaN(-4) and RmaC, under the guidance of the amino acid sequence of other known caged inteins such as caged Npu DnaE intein16,20. A previous study using Npu DnaE intein reported that the protein splicing of split intein begins with the electrostatic interaction between the C-terminal anionic region of N-intein and the N-terminal cationic region of C-intein. This interaction triggers the formation of an intermediate structure, followed by the hydrophobic interaction between the N-terminal region of N-intein and the C-terminal region of C-intein to fold into a specific structure that is necessary to complete protein splicing21. Based on this folding mechanism, we designed the cage sequences for RmaN(-4) and RmaC using Clustal Omega for amino acid sequence alignment22 (Fig. S1). The caged RmaN(-4) (RmaN(-4)cage) was developed by adding amino acid residues 1–46 of RmaC to the C-terminal side of RmaN(-4). In addition, to investigate whether unconditional protein splicing of Rma DnaB intein can be inhibited by a shorter cage, we also added the amino acid residues 1–30 of RmaC to the C-terminal side of RmaN(-4) for developing the RmaN(-4) with short cage (RmaN(-4)Scage). The amino acid residues 1–30 of RmaC correspond to residues 1–13 of Npu C-intein (Fig. S1). This region of Npu C-intein was used as the first version of a cage for Npu N-intein, although it was insufficient to prevent unconditional protein splicing in the case of Npu DnaE intein16. Similarly, amino acid residues 51–102 of RmaN(-4) were added to the N-terminal side of RmaC to form the caged RmaC (RmaCcage) (Fig. 2d).

Then, pDNAs expressing the caged intein-fused C46-split CaVT was co-transfected into HeLa cells with pSV40-2xScMS2(C)-Luc2 and pNL1.1TK[Nluc/TK] to determine whether the designed cage could inhibit unconditional protein splicing of Rma DnaB intein. As expected, when the RmaN(-4)cage (or Rma(-4)Scage) and the RmaCcage were fused to the C46-split CaVT fragments as the vector for cell transfection, the Luc2 translation was not repressed, which means the caging successfully inhibits unconditional protein splicing of Rma DnaB intein (Figs. 2e,f, S2).

Caged Rma DnaB intein fused with nanobody for conditional translational repression

Next, to achieve target protein-responsive translational regulation, we constructed mRNAs to express fusion proteins of C46-split CaVT, the caged Rma DnaB intein, and nanobodies (Fig. S2). In the protein-responsive translational regulation system, two nanobodies must bind different epitopes of the same protein. Therefore, we selected two GFP-targeting nanobodies, GFP-enhancer nanobody23 and Lag1624, which were previously shown to bind different epitopes of GFP4,25. We additionally produced an artificial Luc2 mRNA harboring a strong MS2-binding motif at the 5′ UTR4 (Fig. 3a). The translational repression efficiency of these split CaVTs was analyzed based on the expression of Luc2 in mRNAs containing a strong MS2-binding motif.

Figure 3
figure 3

Conditional translational repression by the caged Rma DnaB intein-mediated reconstitution of EGFP-responsive C46-split CaVT. (a) Schematic depiction of mRNAs used in EGFP-responsive translational repression. (b-c) HeLa cells were co-transfected with 2xScMS2(C)-Luc2 (10 ng/well), Nluc (1 ng/well), and two types of mRNAs to express split CaVT per group (40 ng/well each), and EGFP or eDHFR (10 ng/well) mRNAs. The positive control for translational repression was MS2CP(1–45)-RmaN(-4) with RmaC-MS2CP(46–116)-VPg(FCV) group. The negative control for translational repression was MS2CP(1–45)-RmaN(-4)cage or MS2CP(1–45)-RmaN(-4)Scage with RmaCcage-MS2CP(46–116)-VPg(FCV) group. EGFP-responsive translational repression of Luc2 mRNA by RmaN(-4)cage combined with the cage-free RmaC (b). EGFP-responsive translational repression of Luc2 mRNA by RmaN(-4)Scage combined with the cage-free RmaC (c). The bar graph shows the Luc2/Nluc ratio (mean ± SD, n = 4). **P < 0.01, ***P < 0.001 compared with eDHFR-transfected (EGFP-untransfected) cells in each group by the unpaired two-sided Student’s t- test.

When RmaCcage was used as a component of EGFP-responsive C46-split CaVT, no conditional translational repression was observed. This may be due to the strong protein splicing inhibition by the designed cage, which prevents conditional protein splicing even in the presence of the target protein. In contrast, when RmaN(-4)cage was combined with the cage-free RmaC, EGFP-responsive translational repression of Luc2 mRNA was induced (Fig. 3b). We also tested that EGFP-responsive translational repression could also be induced when RmaN(-4)Scage was used instead of RmaN(-4)cage (Fig. 3c). Since RmaN(-4)cage shows stronger protein splicing inhibition than RmaN(-4)Scage in the absence of target protein, we selected RmaN(-4)cage for subsequent study.

Construct a target protein-responsive translational regulation system containing two different sets of caged intein

To achieve the multiple protein-responsive translational regulation, we planned to jointly use the caged Rma DnaB intein and the caged Npu DnaE intein in the same system. Prior to using these split caged intein pairs in the same application, we checked their orthogonality by transfecting EGFP-responsive C46-split CaVT containing either of these inteins (Figs. 3a, 4a). As shown in Fig. 4b, the combination of the caged Npu N-intein (eNpuNcage) and RmaC did not show EGFP-responsive protein splicing. Similar result was obtained by the combination of RmaN(-4)cage and the caged Npu C-intein (NpuCcage). In contrast, when the caged N-inteins were combined with their original counterparts, EGFP-responsive protein splicing was observed. These results suggest that the two pairs of split caged inteins, the caged Rma DnaB and Npu DnaE inteins, are orthogonal and can be used in the same system without cross-reaction.

Figure 4
figure 4

Orthogonality check of the caged Rma DnaB and Npu DnaE inteins by translational repression using EGFP-responsive C46-split CaVT. (a) Schematic depiction of mRNAs used in EGFP-responsive translational repression via caged Npu DnaE intein. (b) HeLa cells were co-transfected with 2xScMS2(C)-Luc2 (10 ng/well), Nluc (1 ng/well), and two types of mRNAs to express split CaVT per group, (40 ng/well each), and EGFP or eDHFR (10 ng/well) mRNAs. MS2CP(1–45)-eNpuNcage-Lag16 with GFPenhNb-NpuCcage-MS2CP(46–116)-VPg(FCV) group and MS2CP(1–45)-RmaN(-4)cage-Lag16 with GFPenhNb-RmaC-MS2CP(46–116)-VPg(FCV) group were used as positive controls for EGFP-responsive translational repression. The bar graph shows the Luc2/Nluc ratio (mean ± SD, n = 4). **P < 0.01, ***P < 0.001 compared with eDHFR-transfected (EGFP-untransfected) cells in each group by the unpaired two-sided Student’s t-test.

Construct a logic gate using orthogonal caged split inteins for multi-input conditional translation regulation

The utilization of split inteins has demonstrated their effectiveness in implementing cellular logic operations19,26,27,28,29,30. Therefore, we used two different sets of caged split inteins with C46-split CaVT to construct an OR gate (Fig. 5a) that enables simultaneous regulation of two different C46-split CaVT pairs to achieve translational regulation using two different intracellular proteins as inputs. In the OR gate, we used two sensor modules. One is the EGFP-responsive C46-split CaVT using the caged Rma DnaB intein, and the other is the Escherichia coli dihydrofolate reductase (eDHFR)-responsive C46-split CaVT using the caged Npu DnaE intein. The eDHFR-responsive C46-split CaVT uses eDHFR as a target protein. The N- and C-terminal fragments of it contain the eDHFR α epitope-targeting nanobody Nb113 and the β epitope-targeting nanobody CA169831 respectively, and induced the reconstitution of full-length CaVT in the presence of eDHFR4. Thus, like EGFP-responsive C46-split CaVT, eDHFR-responsive C46-split CaVT induces translational repression in an eDHFR-dependent manner. When at least one of EGFP and eDHFR was present, Luc2 translation was repressed (Fig. 5b). The result indicates that the reconstitution of CaVT can be induced by both target proteins and demonstrates the successful creation of the OR gate.

Figure 5
figure 5

Translational repression by the reconstituted EGFP- and eDHFR-responsive C46-split CaVT via orthogonal caged split inteins that enable OR gate. (a) Schematic and truth table of mRNAs used in target protein-responsive translational repression via orthogonal caged split intein. (b) Cell transfection using orthogonal caged split inteins that enable OR gate. HeLa cells were co-transfected with 2xScMS2(C)-Luc2 (10 ng/well), Nluc (1 ng/well), MS2CP(1–45)-eNpuNcage-Nb113 (17.5 ng/well), CA1698-NpuCcage-MS2CP(46–116)-VPg(FCV) (17.5 ng/well), MS2CP(1–45)-RmaN(-4)cage-Lag16 (17.5 ng/well), GFPenhNb-RmaC-MS2CP(46–116)-VPg(FCV) (17.5 ng/well), and EGFP or eDHFR (10 ng/well) mRNAs, also human codon-optimized Barstar mRNA (10 ng/well) as negative target mRNA. The bar graph shows the Luc2/Nluc ratio (mean ± SD, n = 4). *P < 0.05, **P < 0.01, ***P < 0.001 compared with the negative control by Tukey’s multiple comparison test.