Semisynthetic protein nanoreactor for single-molecule chemistry

Preparation of Polypeptide Fragments.

We designed polypeptide fragments that would provide for the efficient semisynthesis of a full-length αHL monomer by a three-way ligation strategy. In our design (Fig. 1A), a central segment [SP (synthetic peptide): SP[Thz¹¹⁴-Gly¹²⁶]-N-acyl-benzimidazolinone(Nbz)] was obtained by chemical synthesis. An N-terminal fragment (NTF: NTF[Ala¹-Met¹¹³]-^αthioester) and a C-terminal fragment (CTF: CTF[Cys¹²⁷-Asn²⁹³]-D₈H₆) were obtained after expression in Escherichia coli.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Preparation of an αHL polypeptide containing an unnatural amino acid by semisynthesis. (A) NTF and CTF were prepared by expression in E. coli. The central peptide, SP, was produced by SPPS. An unnatural amino acid residue (in this case l-propargylglycine, Prp) with a side chain containing a functional group of interest (yellow ball) was incorporated into the peptide. Two sequential NCLs splice the three segments into a full-length αHL polypeptide. (B) The synthetic αHL monomer (SM) containing an alkyne group was prepared by NCL, purified, and characterized by LC-MS: ([M+H]⁺ = 34,994 (obs), 34,994 Da (calcd)). (C) SDS/PAGE analysis of WT and synthetic αHL. Lane 1: molecular markers; lane 2: radiolabeled αHL monomer produced by IVTT; lane 3: radiolabeled WT₇ pores assembled in the presence of DPhPC liposomes (1 mg mL⁻¹); lane 4: purified synthetic αHL monomer (SM); lane 5: purified SM after treatment with DPhPC liposomes. SM₇ pores are formed, which comigrate with the WT₇ pore. An autoradiogram is superimposed on the Coomassie Blue stained gel.

Recombinant NTF[Ala¹-Met¹¹³]-^αthioester was produced by thiolysis of a fusion protein, NTF-GyrA intein-CBD (NTF[Ala¹-Met¹¹³]-Mycobacterium xenopi DNA gyrase A (Mxe GyrA) intein-chitin binding domain (CBD)) (SI Appendix, Fig. S1) derived from the plasmid pTXB3. The DNA encoding residues 1–113 of αHL were PCR-amplified from a plasmid bearing a full-length αHL gene and cloned into pTXB3. After expression at 37 °C, high levels of the fusion protein were found within inclusion bodies in the E. coli host. Following cell lysis and centrifugation, most of the pellet could be solubilized in denaturing buffer containing 8 M urea. To allow binding to chitin beads, the concentration of the urea was reduced to 1 M. Unbound proteins were washed off the column and the bound fusion protein was treated with 2-mercaptoethane sulfonate under gravity flow, yielding the NTF[Ala¹-Met¹¹³]-^αthioester in high purity as determined by SDS/PAGE (SI Appendix, Fig. S1B) and LC-MS (liquid chromatography–mass spectrometry) (SI Appendix, Fig. S2).

We prepared another plasmid (pT7-SC1-CDH, SI Appendix, Fig. S3) for the production of CTF[Cys¹²⁷-Asn²⁹³]-D₈H₆ under control of the T7 promoter. In this plasmid, the codons for residues 1–126 in the pT7-SC1 plasmid, which contains a full-length αHL gene, are replaced with codons for Met-Cys. CTF was again obtained from inclusion bodies, and then purified in 6 M guanidine hydrochloride by use of the His₆ tag at the C terminus. The N-terminal Met was found to be cleaved and the unmasked Cys residue had undergone condensation with pyruvic acid, an abundant metabolite (33). We regenerated free Cys at the N terminus by cleaving the thiazolidine (Thz) with methoxylamine (MeONH₂) at pH 4 (SI Appendix, Fig. S4B). The purified CTF (SI Appendix, Fig. S4A) was characterized by LC-MS (SI Appendix, Fig. S5).

The central synthetic peptide (SP: Thz¹¹⁴ThrLeuPrpTyrGlyPheAsnGlyAsnValThrGly¹²⁶-Nbz), in which Thr-117 is replaced with l-propargylglycine (Prp), was prepared by SPPS (34) with Fmoc-protected amino acids (SI Appendix, Fig. S6). The alkyne side chain of residue 117 was predicted to point into the transmembrane β-barrel of the αHL pore (35); residue 117 is a sensitive position for the detection of chemical reactions (6). The peptide as released from the resin contained a C-terminal Nbz group, which undergoes thiolysis, with (4-mercaptopropionic acetic acid, MPAA), to yield a peptide-^αarylthioester (34). The HPLC-purified peptide (SI Appendix, Fig. S7A) was characterized by LC-MS (SI Appendix, Fig. S7B).

Full-Length αHL Monomer Is Obtained by Native Chemical Ligation.

We then proceeded to assemble a full-length αHL polypeptide bearing the propargyl group (Fig. 1A) with two sequential NCL reactions each producing a native peptide bond: (i) ligation of the synthetic peptide to CTF (SI Appendix, Figs. S8 and S9), which yielded SP-CTF in 63% yield, and (ii) ligation of NTF to the first ligation product, which yielded the full-length synthetic monomer (SM) of αHL in 16% yield (SI Appendix, Fig. S10). Therefore, the overall yield of SM was 10%. The potential deletion product NTF-CTF was absent (Fig. 1B and SI Appendix, Fig. S8F). We purified SM by gel filtration under denaturing conditions (8 M urea) followed by ion-exchange chromatography (SI Appendix, Fig. S10 C and D), and characterized it by LC-MS, which yielded a single peak containing material of the theoretical mass (Fig. 1B). The purified SM was concentrated under denaturing conditions (8 M urea) to prevent premature assembly into a heptameric pore.

Semisynthetic αHL Monomers Correctly Refold to Heptameric Pores.

We next tested whether SM could be correctly folded into active monomers. WT αHL monomers prepared by in vitro transcription and translation (IVTT) and denatured in 8 M urea completely regained hemolytic activity after the urea concentration was reduced to 1 M (36). The urea concentration of the SM solution was reduced to ∼60 mM by repeated dilution and concentration in a centrifugal filter (Amicon, MWCO 3k). The SM was then examined for (i) hemolytic activity toward rabbit red blood cells (rRBCs); (ii) heptamer formation on liposomes and red blood cell membranes; and (iii) conformational integrity as reflected by limited proteolysis.

The specific hemolytic activity of SM [HC₅₀ = 39 ng mL⁻¹ (n = 3); HC₅₀, the concentration of protein giving 50% lysis at 100 min, Fig. 2A] was very similar to (SI Appendix, Fig. S11) that of recombinant WT αHL monomer [HC₅₀ = 35 ± 12 ng mL⁻¹ (n = 3), previously measured (37) as 31 ng mL⁻¹]. To visualize the formation of αHL heptamers, we incubated SM for 1 h at 37 °C in the presence of liposomes (1 mg mL⁻¹, diphytanoyl phosphatidylcholine, DPhPC) (38) and upon SDS-polyacrylamide gel electrophoresis, we observed a new band that migrated with a mobility corresponding to the heptameric state (Fig. 1C). In addition, we incubated SM in different ratios with radiolabeled WT αHL produced by IVTT in the presence of rabbit red blood cell membranes (rRBCm). SM oligomerized to form heteroheptamers with different stoichiometries (WT_7−nSM_n, n = 0–7) (Fig. 3A). Together, these data show that SM forms functional heptameric pores. A limited proteolysis experiment was carried out with proteinase K, to determine the conformational state of the αHL polypeptide in the monomer and the heptamer (39⇓–41). We found that the monomer was cleaved into two fragments of 20 and 15 kDa, indicating that it is correctly folded (Fig. 2B, lane 4). When the heptamer formed on liposomes was similarly treated, a large fraction of the polypeptide chains was protected from the protease (Fig. 2B, lane 5), indicating that the heptamer was correctly assembled (SI Appendix, Fig. S12).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Functional properties of the semisynthetic αHL polypeptide. (A) Hemolysis assays. Row 1: hemolysis by recombinant WT αHL; row 2: hemolysis by SM; row 3: rRBCs with no αHL addition. The decrease in light scattering over time was recorded in a microplate reader. WT (1 μL, 1.08 mg mL⁻¹) and SM αHL monomer (1 μL, 4.02 mg mL⁻¹) were added to the first column of row 1 and row 2 of the plate, respectively, and each sample was serially twofold diluted in a final volume of 100 μL per well. (B) Limited proteolysis with proteinase K of SM in solution and in the presence of DPhPC liposomes (1 mg mL⁻¹). Lane 1: molecular markers; lane 2: SM; lane 3: SM treated with DPhPC liposomes; lane 4: SM in solution digested by proteinase K. Two fragments (∼20 and 15 kDa) were generated (b and c); lane 5: SM treated with DPhPC liposomes was further treated with proteinase K and then heat-denatured at 95 °C to disassemble the heptamers. The central proteolytic site is inaccessible in SM₇, and most of the SM polypeptide is undigested. Fragment a is generated after cleavage near the N terminus in the prepore form of SM₇ (∼32 kDa) (41). Note that a single irrelevant lane was excised between lanes 4 and 5.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Structure and electrical properties of the WT₆SM₁ αHL pore. (A) Heteroheptameric pores. WT αHL (radiolabeled protein was obtained by IVTT) and SM were mixed in various ratios (increasing SM in the right-hand lanes) and incubated at 37 °C in the presence of rRBC membranes. The heptameric pores that formed were separated by SDS/PAGE based on the ability of the oligo-aspartate extension (D₈ tail) at the C terminus of SM to enhance electrophoretic mobility. WT₆SM₁ pores were eluted from the gel for single-channel recording. (B and C) Top and side view of the β-barrel (residues 109–150) of the WT₆SM₁ pore. The alkyne group of the Prp in SM is shown as a yellow stick structure, whereas the Thr-117 residues of the WT subunits are in green. The crystal structure of αHL (Protein Data Bank ID code 7AHL) was used as a template. The structure of WT₆SM₁ was generated by an energy minimization method by using AMBER (Version 12) (see SI Appendix, Materials and Methods). (D) Histogram of unitary conductance values of WT₆SM₁ pores (n = 55). (E) IV curves of WT₆SM₁ (mean values, n = 9) and WT αHL pores (mean values, n = 5). The latter were made from WT subunits prepared by IVTT.

The Electrical Properties of WT and Semisynthetic Pores Are Similar.

We examined the electrical properties of WT₆SM₁ pores by determining the mean conductance under defined conditions (Fig. 3D) and measuring IV curves (Fig. 3E). The conductance of WT₆SM₁ (0.79 ± 0.07 nS at +50 mV in 1 M KCl, 20 mM Bis-Tris propane, pH 8.5) was similar to that of the WT₇ pore (∼0.88 nS, same conditions). We also investigated the binding kinetics of β-cyclodextrin (βCD) with WT₆SM₁ to confirm that the structure of the transmembrane β-barrel was intact. βCD is a noncovalent molecular adaptor that acts as a transient channel blocker when it is lodged in the lumen of the αHL pore (42, 43). In single-channel recordings, the addition of βCD to the trans compartment produced reversible, transient, partial blockades of the ionic current. The mean dwell time of βCD in the pore (τ_off) did not vary with βCD concentration, whereas the mean lifetime of the unoccupied state (τ_on) was inversely proportional to the concentration, suggesting a bimolecular interaction (SI Appendix, Fig. S13). We determined the association rate constant (k_on) and dissociation rate constant (k_off) of βCD binding from the τ-values: k_on = 1.1 ± 0.3 × 10⁵ M⁻¹⋅s⁻¹; k_off = 1.4 ± 0.1 × 10³ s⁻¹; K_D = k_off/k_on = 1.3 ± 0.4 × 10⁻² M. Similar values were obtained previously for the WT αHL pore (43). Therefore, the alkyne residue presented by the SM subunit neither affects the electrical properties of the αHL pore nor its ability to bind the βCD adaptor.

The Semisynthetic Pore Reveals an Intermediate in the Copper(I)-Catalyzed Azide-Alkyne Cycloaddition Reaction.

We then used the protein nanoreactor approach to investigate click chemistry of the alkyne side chain in WT₆SM₁ at the single-molecule level. The copper(I)-catalyzed azide-alkyne cycloaddition (CuAAC), a click reaction (44), has been widely studied in materials science (45), bioconjugation (46), and drug discovery (47). We first examined the unfunctionalized WT₇ pore (SI Appendix, Fig. S15) in the presence of the reagents used for the click reaction, namely the Cu(I) catalyst and azide substrates (SI Appendix, Fig. S14, APB400: 0.4 kDa azide-PEG-biotin; APB700: 0.7 kDa azide-PEG-biotin). Cu(I) (20 mM) and APB400 (0.5–2.0 mM), neither separately nor together, affected the WT₇ pore (SI Appendix, Fig. S15 A–G), whereas APB700, either by itself (SI Appendix, Fig. S15 H–K) or with Cu(I) (SI Appendix, Fig. S15L), produced reversible events, which increased in frequency with APB700 concentration. These events most likely arise from entry of APB700 into the lumen of the pore, without covalent attachment, for periods sufficient to be detected (48). By examining the concentration dependence of the blockades (APB700, 0.5 to 2 mM) in the absence of Cu(I), we obtained k_on = 1.3 ± 0.2 × 10⁵ M⁻¹⋅s⁻¹; k_off = 8.7 ± 0.3 × 10³ s⁻¹; K_D = k_off/k_on = 6.7 ± 1.0 × 10⁻² M (SI Appendix, Fig. S17), similar to the values obtained by Movileanu et al. for a 940-Da PEG (48). The mean lifetime of the blockades was short (∼100 μs), resulting in a broad distribution of amplitudes under our data acquisition conditions. No permanent current decrease was observed with the WT₇ pore during 3 h of monitoring after the addition of Cu(I), with either of the azides present in both compartments.

We then carried out chemistry with single WT₆SM₁ pores. As in the case of the WT₇ pore, Cu(I) added to both the cis and trans compartments at the same time did not generate significant changes in the current (SI Appendix, Fig. S16 A and B), suggesting that if a Cu acetylide (Fig. 4 A, I) is formed it is short-lived. When APB700 was added to the trans compartment in the absence of Cu(I), short reversible current blockades characteristic of entry and exit from the pore were seen, which were similar to those observed with the WT₇ pore: k_on = 1.6 ± 0.2 × 10⁵ M⁻¹⋅s⁻¹; k_off = 5.6 ± 1.1 × 10³ s⁻¹; K_D = k_off/k_on = 3.5 ± 0.8 × 10⁻² M (SI Appendix, Fig. S16 C–L). In the case of APB400, the events were presumably too short to be registered at the data acquisition rate used (48). By contrast with WT₇, the addition of Cu(I) to both compartments in the presence of 2 mM APB400 or APB700 (trans) eventually led to an irreversible partial blockade of the WT₆SM₁ pore, suggesting that Cu(I) triggered a reaction between the substrate and the alkyne within the pore. Interestingly, the blockade of the pore proceeded in two steps (Fig. 4 B–D and SI Appendix, Table S1). An intermediate state (S_i) was formed first, characterized by a small current decrease (δI_b1 = Is_o − Is_i, where S_o is the unreacted state of the pore), which was subsequently converted to a second state (S_b) associated with a larger current decrease (δI_b2 = Is_i − Is_b). The percent decreases in the open pore current for APB400 (SI Appendix, Fig. S18A) were %δI_b1, _APB400 = 6.5 ± 3.4 and %δI_b2, _APB400 = 18.9 ± 3.2% (n = 11, where n represents the number of experiments). For the larger APB700, the percent decreases (SI Appendix, Fig. S18B) were %δI_b1, _APB700 = 9.5 ± 3.6 and %δI_b2, _APB700 = 31.1 ± 3.7% (n = 10). The first step (to form S_i) was irreversible, and the final step yielded a permanent partly blocked state, S_b. The S_i state is relatively noisy (Fig. 4 C and D), which suggests that the intermediate undergoes a reversible transformation, such as a conformational change or protonation and deprotonation [for APB400: I_rms(S_o) = 12.7 ± 1.6%, I_rms(S_i) = 16.3 ± 2.9%, I_rms(S_b) = 14.1 ± 2.3% (n = 11); for APB700: I_rms(S_o) = 17.4 ± 3.2%, I_rms(S_i) = 22.1 ± 2.8%, I_rms(S_b) = 14.7 ± 3.2% (n = 10), filtering: 5 kHz, sampling: 25 kHz, SI Appendix, Table S1]. The mean lifetime of S_i (SI Appendix, Figs. S19 and S20) was τ_i = 4.5 ± 0.6 s (k_i = 0.22 ± 0.03 s⁻¹) [n = 21, for APB400 (n = 11); for APB700 (n = 10)] and the mean overall reaction rate constant was 0.24 ± 0.01 M⁻¹⋅s⁻¹ (SI Appendix, Fig. S21). It is possible that the intermediate is one of the three sequential structures (II–IV, Fig. 4A) proposed by Worrell et al. (49), and based on its lifetime it is most likely the Cu triazolide (IV). However, the mechanism of the CuAAC reaction remains contentious, and our assignment is strictly hypothetical. More important is the fact that we observe a long-lived intermediate that must be incorporated into any viable reaction mechanism.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Click reaction in a WT₆SM₁ αHL nanoreactor. (A) A proposed mechanism (50) for the CuAAC reaction in the context of a WT₆SM₁ αHL pore. Parts of the β-strands have been removed to expose the reactive Prp residue. (A, I) formation of a Cu(I)-acetylide. (A, II) reversible binding of an azide substrate. (A, III) A dinuclear copper intermediate. (A, IV) formation of Cu triazolide intermediate. (A, V) The triazole product corresponding to the S_b state. The triazolide (A, IV) may correspond to the S_i intermediate in the current trace, but this assignment is speculative (see the text). (B) A schematic of the current traces corresponding to the CuAAC reaction. The unreacted state (S_o) is converted to an intermediate state (S_i) before the pore is permanently, partly blocked (S_b). The magnitudes of the current steps from S_o to S_i and S_i to S_b are denoted by δI_b1 and δI_b2, respectively (δI_b = δI_b1 + δI_b2). (C) Current recording with azide APB400 as the substrate (2 mM). (D) Current recording with azide APB700 as the substrate (2 mM). In both cases, the concentration of Cu(I) was 20 mM. Note that the time scales in C and D differ. The buffer was 1 M KCl, 20 mM Bis-Tris propane, 25 μM TCEP, pH 8.5. The recording was made at +50 mV. The current was filtered at 5 kHz and sampled at 25 kHz. For display, further digital filtering was carried out at 1 kHz.