Efficient, Traceless Semi-Synthesis of α-Synuclein Labeled with a Fluorophore/Thioamide FRET Pair

Abstract

We have shown that thioamides can be incorporated into proteins through semi-synthesis and used as probes to monitor structural changes. To date, our methods have required the presence of a cysteine at the peptide ligation site, which may not be present in the native peptide sequence. Here, we present a strategy for the semi-synthesis of thioproteins using homocysteine as a ligation point with subsequent masking as methionine, making the ligation “traceless.”

Fluorescence spectroscopy is a powerful technique for studying protein dynamics and stability. Distance-dependent chromophore interactions, such as Förster resonance energy transfer (FRET) and photoinduced electron transfer (PET) are routinely employed for monitoring protein conformation in real time.¹ We have recently shown that a peptide backbone thioamide, a single-atom substitution of the carbonyl oxygen with sulfur, can quench a variety of natural and unnatural amino acids.² Unlike commonly used fluorescence quenchers, thioamides are sufficiently small that they can be placed at nearly any position of the protein sequence without significantly altering the secondary structure.³ Our laboratory has developed semi-synthetic methods to incorporate these minimally perturbing fluorescence probe pairs into full-length proteins. Backbone thioamides cannot currently be installed into a protein by means of ribosomal expression. Therefore, we incorporate thioamides into full-length proteins using native chemical ligation (NCL), a fragment condensation reaction that typically takes place between peptides bearing a C-terminal thioester and an N-terminal cysteine, to form a native amide bond.⁴ Expressed protein ligation (EPL), a variant of NCL, allows us to express a large portion of the protein in E. coli cells, to which we can append a small synthetic thioamide-containing fragment with high efficiency.⁵

Misfolding of the abundant neuronal protein α-synuclein (αS) has been implicated in the pathogenesis of several debilitating neurodegenerative disorders.⁶ Under physiological conditions, αS is soluble, monomeric, and intrinsically disordered. In its pathological conformation, αS monomers associate with each other to form oligomers that eventually mature into β-sheet rich fibrils.⁷ Despite recent developments of new methods to study αS dynamics, the molecular details underlying its aberrant oligomerization remain poorly understood.⁸ Recently, we combined unnatural amino acid mutagenesis with native chemical ligation to install the FRET pair, p-cyanophenylalanine (Cnf) and a thioamide, into full-length αS for misfolding studies.⁹ This FRET system provides reliable distance measurements over a range of 8 to 30 Å within a protein’s three-dimensional structure. Using our double-labeled constructs, we were able to demonstrate that we can monitor conformational changes in monomeric αS using urea or trimethylamine oxide (TMAO) to denature or compact the protein, respectively.

Despite agreement of our data with other studies of αS in TMAO, we were somewhat concerned about the presence of the non-native Cys residues in our labeled αS constructs.¹⁰ Cys mutants of αS display enhanced aggregation kinetics and altered fibril morphology as a result of intermolecular disulfide bond formation.¹¹ Although we performed the previous FRET studies with our Cys-containing double-labeled constructs in the presence of a reducing agent, it is nonetheless possible that some amount of disulfide formation occurred during our experiments.

While many laboratories carry out protein synthesis with subsequent radical desulfurization of Cys to form Ala, this strategy is not viable for us as desulfurization of the thioamide would also occur.¹² Methods developed in our laboratory enable us to generate double-labeled αS for misfolding studies that does not contain cysteine at the ligation site. It has been shown that NCL can be performed with an N-terminal homocysteine (Hcs) in place of Cys. Hcs can then be selectively methylated to yield methionine at the ligation site.¹³ Hcs-mediated ligation was previously restricted to short C-terminal peptides in which Hcs could be installed using solid phase peptide synthesis (SPPS). We have recently established that we can use E. coli aminoacyl transferase (AaT) to deliver disulfide-protected Hcs (S-(thiomethyl)homocysteine, Hcm) to the N-terminus of a large expressed protein fragment.¹⁴ In E. coli cells, AaT transfers Phe, Leu, or Met from an aminoacyl tRNA to the N-terminus of a protein.¹⁵ By using a modified methionine aminoacyl tRNA synthetase enzyme (Met*RS), we can generate Hcm-tRNA in situ, which serves as a viable substrate for AaT.¹⁶ Once Hcm is transferred to the N-terminus of a protein, it is easily deprotected by a reducing agent to form Hcs for ligation. By combining unnatural amino acid mutagenesis with EPL at homocysteine (which is converted to methionine), we can incorporate our minimal FRET pair into αS in a traceless manner that minimizes unnecessary synthesis of peptide fragments composed of natural amino acids. A retrosynthetic analysis of our target protein, Ac-αS^FAsp'₂Cnf₃₉, is shown in Figure 1. The prime symbol is used to denote a backbone thioamide bond in an amino acid represented by the standard three-letter code, and αS^F represents a construct with all of the native Tyr residues (39, 125, 133, and 136) mutated to Phe to prevent chromophore cross talk.

Figure 1.

Retrosynthetic analysis of Ac-αS^FAsp'₂Cnf₃₉.

Previously, we have shown that fluorenylmethoxycarbonyl (Fmoc)-protected thiocarboxybenzotriazoles can be used with conventional Fmoc-protected amino acids to synthesize thioamide-containing peptides.^5a,9 The procedure for the preparation of Fmoc-Asp′(Ot-Bu)-nitrobenzotriazole 3 follows a previous report by Shallaby and is illustrated in Scheme 1.¹⁷ Coupling of 4-nitro-1,2-phenylenediamine to Fmoc-Asp(Ot-Bu)-OH was achieved using isobutyl chloroformate (IBCF) to generate 1 in 89% yield.²⁰ Amide 1 was then added to a solution of P₄S₁₀ and anhydrous Na₂CO₃ at 0 °C. Selective thionation of the Asp backbone carbonyl was completed within 1 h to afford compound 2 in 79% yield.²¹ Intramolecular diazonium cyclization of thioamide 2 gave the activated benzotriazole 3 in 67% yield.²² Compound 3 can be isolated from the final reaction by precipitation with ice-cold water in sufficient purity for peptide coupling reactions. Since the activated benzotriazoles are subject to degradation through hydrolysis and intramolecular nucleophilic attack by the carbonyl oxygens, it is best to minimize handling of compound 3 prior to peptide coupling.

Scheme 1.

Synthesis of aspartyl thioamide precursor 3.²⁰

Although, we have previously used on-resin methods to prepare thioesters for NCL, we find that off-resin activation is simplest for short peptides. The N-terminal thioamide-containing fragment of αS was synthesized using standard Fmoc-based SPPS procedures with the exception of direct acylation of Val₃ by benzotriazole 3 (i.e., no activating agents were added). Thiopeptide 4 was synthesized on 2-chlorotrityl resin, cleaved under mild conditions (10% AcOH) to retain the Asp side-chain protecting group, and then activated using benzotriazol-1-yloxytripyrrolidinophosphonium hexafluorophosphate (PyBOP) to form the mercaptopropionate thioester 5 (Scheme 2). Epimerization of the α-carbon of the C-terminal phenylalanine residue was eliminated by reducing the PyBOP activation time to 25 minutes.¹⁸ Following acidic deprotection of Asp′(Ot-Bu)₂, the thioester peptide 6 was isolated by HPLC (5.3 µmol, 5% yield) and its identity was confirmed by MALDI MS analysis.

Scheme 2.

Synthesis of thioamide-containing thioester peptide (6) for native chemical ligation.

The expressed C-terminal Cnf-containing αS fragment for NCL was generated as depicted in Figure 2. AaT selectively modifies the α-amine of proteins bearing a lysine or arginine as the N-terminal residue. Prior to this work, we demonstrated that Met₅Lys₆ could be used as a point of disconnection in the synthesis of αS.¹⁴ To minimize perturbation of native αS dynamics resulting from Cnf incorporation, we chose to replace the similarly sized Tyr or Phe at position 39. A plasmid containing the C-terminal protein fragment (αS^F_6–140) with a TAG codon installed at position 39 was prepared. An N-terminal His₁₁ tag and a Factor Xa proteolysis site precede the αS gene. E. coli cells were transformed with two plasmids, one containing the Cnf-selective mutant synthetase (CnfRS) with tRNA and the other containing His_Tag-αS^F_6–140TAG₃₉.¹⁹ Following an initial growth period, protein expression was induced by adding Cnf and isopropyl β-D-1-thiogalactopyranoside (IPTG) to the culture medium. The expressed protein, His_Tag-αS^F_6–140Cnf₃₉ (7), was purified by Ni-affinity chromatography and subsequently cleaved by Factor Xa to yield αS^F_6–140Cnf₃₉ (8). Following purification by ion-exchange chromatography, the truncated protein was incubated with AaT, Met*RS, E. coli total tRNA, Hcm, and ATP for 2 h resulting in complete formation of αS^F_5–140Hcm₅Cnf₃₉ (9). αS^F_5–140Hcm₅Cnf₃₉ was subjected to an additional round of ion-exchange chromatography prior to use in the NCL reaction. Typically, about 1 µmol of pure 9 is obtained from a 1 L protein expression.

Figure 2.

Ac-αS^FAsp′₂Cnf₃₉ synthesis. Left: Functionalization of αS^F_6–140Cnf₃₉ by a) cleavage of His₁₁ tag at Factor Xa site, b) attachment of Hcm by AaT-catalyzed modification, c) ligation of an N-terminal thioester peptide (6), and d) conversion of Hcs to Met by methylation to form Ac-αS^FAsp′₂Cnf₃₉ (11). Top Right: MALDI MS analysis of full length 11. Calcd m/z (M+H)⁺ 14480.1, found 14479.6. Middle Right: Trypsinized fragment corresponding to residues 1–6 of Ac-αS^FAsp′₂Cnf₃₉ confirms methylation. Calcd m/z (M+H)⁺ 828.34, found 828.46. The asterisk indicates the expected mass of the 1–6 fragment with unmethylated Hcs at position 5. Calcd m/z (M+H)⁺ 814.39. Bottom Right: Trypsinized fragment corresponding to residues 33–43 of Ac-αS^FAsp′₂Cnf₃₉ confirms Cnf incorporation. Calcd m/z (M+H)⁺ 1189.66, found 1189.85. The asterisk indicates the expected mass of the 33–43 fragment with Tyr at position 39. Calcd m/z (M+H)⁺ 1180.65.

The ligation reaction was initiated by incubation of 9 with 1.5 equivalents of Ac-MetAsp′ValPhe-SR (6) in degassed ligation buffer (6 M guanidinium hydrochloride, 0.2 M sodium phosphate, 20 mM tris(2-carboxyethyl)phosphine, 2% thiophenol, at pH 7.5). Disulfide deprotection of the Hcm residue was observed by MALDI MS within five min. The ligation reaction was allowed to proceed for 24 h prior to buffer exchange into methylation reaction buffer (20 mM Tris, pH 8.6). Ac-αS^FAsp′₂Hcs₅Cnf₃₉ (10) was converted to Ac-αS^FAsp′₂Cnf₃₉ (11), with native Met₅, by a short treatment (10 min) with 3000 equivalents of MeI. Allowing the reaction to proceed for longer periods of time led to S-alkylation of the thioamide (data not shown). Ac-αS^FAsp′₂Cnf₃₉ was isolated from the NCL reaction by HPLC and characterized by MALDI MS, PAGE gel, and UV-Vis and spectroscopy (Figure 2 and Supporting Information). The conversion over the two-step ligation/methylation sequence was 41%. MALDI MS analysis of trypsin-digested 11 confirmed selective alkylation of Hcs to form Met in the purified product.

Using a similar double-labeled construct containing cysteine at the ligation site (Ac-αS^FPhe′₄Cys₉Cnf₃₉), we observed thioamide quenching of Cnf after compaction of the protein in high concentrations of TMAO.⁹ Since the Ser-to-Cys mutation in this protein constituted a small deviation from the native αS protein sequence, we sought to investigate this phenomenon using our newly synthesized Ac-αS^FAsp′₂Cnf₃₉ construct (Asp′₂ used here because labeling at Phe′₄ would not be compatible with our ligation strategy using Hcs). To correct for any changes in Cnf fluorescence that are not due to the presence of the thioamide, an oxoamide control protein (αS^FCnf₃₉) was also prepared for fluorescence experiments. In buffer without TMAO, thioamide quenching of Cnf in Ac-αS^FAsp′₂Cnf₃₉ (i.e., FRET) was minimal. In accordance with our previous observations, thioamide quenching of Cnf increased as a function of TMAO concentration (Figure 3, left axis). To assign distance constraints based on these measurements, the measured quenching efficiency (E_Q) values were converted to distances using Förster theory. In 2 M and 4 M TMAO, R_FRET was calculated as 15.4 Å and 12.6 Å, respectively (Figure 3, right axis). Taken together with our previous results, these data indicate that the N-terminal region of αS undergoes significant compaction in high TMAO concentrations. Furthermore, our new results conclusively demonstrate that the observed quenching is not due to Cys-mediated dimerization of αS.

Figure 3.

Refolding assay. Left: Monomeric double-labeled αS (Ac-αS^FAsp′₂Cnf₃₉) is mixed with TMAO to induce compaction and an increase in quenching efficiency (E_Q). Right: E_Q (blue dashed line) of Ac-αS^FAsp′₂Cnf₃₉ determined at varying concentrations of TMAO. Interchromophore distance (R_FRET, green solid line) computed using Förster theory as described in Supporting Information.

By combining unnatural amino acid mutagenesis with EPL at Met, we can install our minimal Cnf/thioamide FRET probe pair into αS without introducing a mutation at the ligation site. We have demonstrated that selective alkylation of Hcs can be performed in the presence of a thioamide, rendering our new ligation strategy compatible with the minimal probe pairs that we have developed. The αS refolding studies presented here demonstrate how this methodology can be used to study protein folding dynamics, though these techniques are by no means limited to αS. Over 20,000 proteins in the PDB contain MetArg or MetLys motifs that can be used as points of disconnection in their retrosynthetic analyses.¹⁴ We are currently working to expand these methods even further by combining EPL at Met with multiple ligation strategies. This methodology will allow us to install thioamide-containing synthetic fragments in the middle of a large protein target in a traceless and efficient manner.

Biography

Rebecca F. Wissner (near right) earned a B.S. degree in Chemistry in 2009 from New York University, working with Prof. Paramjit Arora. She is a graduate student in Prof. Petersson’s group, studying protein folding using semi-synthetic proteins.

Anne M. Wagner (far left) graduated in 2005 with a B.S. in Chemistry from Mount Holyoke College, working with Prof. Megan Núñez. She received her Ph.D. from the University of Pennsylvania in 2013 for her work on protein functionalization.

John B. Warner (far right) received a B.S. in Chemistry from Clarkson University in 2008, working with Prof. Silvana Andreescu. He is a graduate student in Prof. Petersson’s group, developing methods for multi-functionalizing proteins.

E. James Petersson (near left) was educated at Dartmouth College, where he worked in the labs of David Lemal. He then studied under Dennis Dougherty at the California Institute of Technology as an NIH Predoctoral Fellow. After obtaining his Ph.D. in 2005, he worked as an NIH Postdoctoral Fellow at Yale University with Alanna Schepartz. He was appointed as Assistant Professor in the Department of Chemistry at the University of Pennsylvania in 2008 and in the Biochemistry and Molecular Biophysics group in the Perelman School of Medicine in 2013.