Efficient Total Chemical Synthesis of ¹³C=¹⁸O Isotopomers of Human Insulin for Isotope-Edited FTIR

Abstract

Isotope-edited two-dimensional Fourier transform infrared spectroscopy (2D FTIR) can potentially provide a unique probe of protein structure and dynamics. However, general methods for the site-specific incorporation of stable ¹³C=¹⁸O labels into the polypeptide backbone of the protein molecule have not yet been established. Here we describe, as a prototype for the incorporation of specific arrays of isotope labels, the total chemical synthesis – via a key ester insulin intermediate – of 97% enriched [(1-¹³C=¹⁸O)Phe^B24]human insulin, stable-isotope labeled at a single backbone amide carbonyl. The amino acid sequence as well as the positions of the disulfide bonds and the correctly folded structure were unambiguously confirmed by the X-ray crystal structure of the synthetic protein molecule. In vitro assays of the isotope labeled [(1-¹³C=¹⁸O)Phe^B24]human insulin showed that it had full insulin receptor binding activity. Linear and 2D IR spectra revealed a distinct red-shifted amide I carbonyl band peak at 1595 cm⁻¹ resulting from the (1-¹³C=¹⁸O)Phe^B24 backbone label. This work illustrates the utility of chemical synthesis to enable the application of advanced physical methods for the elucidation of the molecular basis of protein function.

Entry for the Table of Contents

Just one carbonyl, please. Human insulin was specifically labeled at the (1-¹³C=¹⁸O)Phe^B24 single backbone carbonyl, by total chemical synthesis via a key ester insulin intermediate. The synthetic protein was characterized by X-ray crystallography and had full insulin receptor-binding activity. Linear and 2D IR spectra of isotope-labeled human insulin revealed a distinct spectral peak at 1595 cm⁻¹, a red shifted amide I band from the (1-¹³C=¹⁸O)Phe^B24 label.

Two-dimensional Fourier transform infrared spectroscopy (2D FTIR) is emerging as a powerful technique for the study of the transient structures and dynamics of protein molecules.^[1] Most importantly, 2D FTIR makes use of backbone amide I vibrations to reveal protein secondary structure and to probe H-bonding interactions.^[2] Because even small protein molecules have dozens of backbone amides with similar vibrational properties, it is essential to make use of ¹³C=¹⁸O double isotope labeling of the carbonyls at specific peptide bonds, in order to isolate vibrational signals from the region(s) of interest in the protein molecule.^[2–3] Over the past decade, chemical synthesis of peptides with site-specific ¹³C=¹⁸O backbone amide labels as 2D FTIR probes has provided information ranging from structural studies of model peptides^[4] to the role of water in a homotetrameric proton channel.^[5] The research community seeks to extend such isotope-edited FTIR studies to more complex protein molecules.^{[1, 6]} However, with rare exceptions,^[7] residue-specific isotope labeling has been limited to small protein domains that are directly accessible by stepwise solid phase peptide synthesis (SPPS). There is a need for a general method for ¹³C=¹⁸O labeling peptide bond carbonyls at one or more specific sites in the backbone of a protein molecule. In the work reported here, we describe the application of total chemical synthesis to the site-specific stable isotope labeling of a protein molecule. The methods described can be used to specifically label any region of the molecule being studied, and will be applicable to a wide range of proteins.

Human insulin is a protein hormone that plays a key role in energy metabolism.^[8] The mature human insulin protein molecule consists of two polypeptide chains, the 21 amino acid residue A chain and the 30 amino acid residue B chain, with two inter-chain and one intra-chain covalent disulfide bonds (Figure 1).^[9] A key aspect of the biosynthesis and storage of human insulin is the formation of insulin dimers, which coordinate zinc ions to form zinc-containing hexamers that in turn form aggregates in the beta-cells of the pancreas.^[10] These secretory granules are the principal storage form of insulin, from which it is released into the circulation. As part of a research project to study the monomer-dimer equilibrium of the human insulin protein molecule using time-resolved 2D infrared spectroscopy,^[11] we set out to prepare human insulin labeled with stable-isotope reporter nuclei at a single backbone carbonyl in the insulin dimer interface. The carbonyl group of residue Phe^B24 is involved in critical H-bonding interactions across the Phe^B24-Tyr^B26 β-sheet interface in human insulin dimers.^[11] We wanted to ¹³C=¹⁸O label the backbone carbonyl of Phe^B24 alone in the 51 amino acid residue human insulin protein molecule, in order to be able to use 2D FTIR to probe key aspects of the monomer-dimer equilibrium of this important protein molecule.

Figure 1.

Covalent structure of the human insulin protein molecule. The amino acid sequences of the 21 residue A chain and 30 residue B chain are given using the single letter code. Stable-isotope labeled residue Phe^B24 is highlighted in red.

Human insulin is a notoriously difficult target for total chemical synthesis.^[12] Synthesis and handling of the individual 21 residue A and 30 residue B peptide chains is itself challenging,^[12b] and efficient formation of the folded insulin protein molecule can only be achieved by the use of orthogonally protected cysteine residue side chains and directed disulfide formation.^[13] We recently reported a simple and efficient route to the total chemical synthesis of DKP insulin, a form of insulin that is monomeric because of the replacement of His^B10 by Asp^B10 and that has substantially reduced aggregation properties because of the inversion of the –Pro^B28-Lys^B29– sequence.^[14] This efficient total synthesis proceeded via a 51 residue depsipeptide that folds efficiently to give ‘ester insulin’, which can be readily converted to fully active DKP insulin.^[15] The ester insulin approach has not been applied to the total synthesis of human insulin.

Here we describe the preparation of [(1-¹³C=¹⁸O)Phe^B24]human insulin by one-pot native chemical ligation^[16] of three unprotected synthetic peptide segments [Scheme 1]. This synthetic strategy was based on that previously used for the synthesis of monomeric DKP insulin, which protein was designed to have optimal handling characteristics.^[15b] In the case of human insulin, peptide chain aggregation and the formation of protein oligomers make this a substantially more challenging synthetic target and total synthesis by the ester insulin route required significant optimization.

Scheme 1.

Synthesis of [(1-¹³C=¹⁸O)Phe^B24]human insulin.

Details of the synthesis were as follows. The (1-¹³C=¹⁸O)Phe^B24 labeled depsipeptide Gly^A1-Glu^A4[OβThr^B30-Thz^B19]-Cys^A6-^αthioester 1 was synthesized at a 0.2 mmol scale on Boc-Cys-SCH₂CO-Ala-OCH₂-Pam-resin, using manual Boc chemistry in situ neutralization SPPS protocols (Scheme 2). The orthogonally protected ester-linked dipeptide Boc-Glu^A4[Oβ(Alloc-L-Thr^B30-α-OcHex)]-OH was used in the synthesis of the N-benzyloxycarbonyl-Gly^A1-Cys^A6 sequence, then the Alloc protecting group was removed and the Thz^B19-Thr^B30 sequence was assembled in stepwise fashion.

Scheme 2.

Preparation of the stable-isotope labeled segment Gly^A1-Glu^A4[OβThr^B30-(1-¹³C=¹⁸O)Phe^B24-Thz^B19]-Cys^A6-^αthioester 1 by solid phase peptide synthesis. Conditions: i) Boc-Ala-OCH₂Ph-CH₂COOH, DIC, DCM; ii) a) TFA, b) Trt-SCH₂COOH, HBTU, DIEA, DMF; iii) a) TFA/TIPS/water (95:2.5:2.5 v/v), b) BocAA-OH, HBTU, DIEA, DMF (SPPS, 2 cycles); iv) a) TFA, b) Boc-Glu[Oβ(Alloc-L-Thr-α-OcHex)]-OH, HBTU, DIEA; v) a) TFA, BocAA-OH, HBTU, DIEA, DMF (SPPS, 3 cycles); b) TFA followed by 25% DIEA/DMF, 2-Cl-ZOSu, DMF; vi) a) Pd(PPh₃)₄, PhSiH₃, DCM, b) BocAA-OH, HBTU, DIEA, DMF (SPPS, 11 cycles); HF-p-cresol.

Isotope labeled Boc-(1-¹³C=¹⁸O)Phe^B24 with 97.1% isotope enrichment was prepared as described in the Supporting Information, and efficiently incorporated under conditions that used a minimum amount of the isotope-labeled amino acid: 0.25 mmol, 0.4 M Boc-(1-¹³C=¹⁸O)Phe, 0.23 mmol HBTU, 0.55 mmol DIEA for 1 hr. The isotope-labeled depsipeptide-thioester was deprotected and simultaneously cleaved from the resin by treatment with anhydrous HF, purified by preparative reverse phase HPLC, and characterized by LCMS analysis to give Gly^A1-Glu^A4[OβThr^B30-(1-¹³C=¹⁸O)Phe^B24-Thz^B19]-Cys^A6-^αthioester 1 (Figure 2).

Figure 2.

[Upper panel] Stable-isotope labeled Gly^A1-Glu^A4[OβThr^B30-(1-¹³C=¹⁸O)Phe^B24-Thz^B19]-Cys^A6-^αthioester 1, R = CH₂CO-Ala; [Lower panel] LCMS characterization of 1 after prep-HPLC purification (inset showing the online ESI-MS spectrum taken across the whole UV peak).): observed mass 2195.0 ± 0.2 Da; calculated mass, 2195.0 Da (monoisotopic) and 2195.4 Da (average isotopes).The chromatographic separations were performed on a C4 (2.1×50 mm) column using a linear gradient (5–45%) of solvent B in solvent A over 40 min (solvent A = 0.1% TFA in water, solvent B = 0.08% TFA in acetonitrile).

The peptide Cys^A7-Asn^A21 2 and the peptide Phe^B1-Val^B18-^αthioester 3 were synthesized by manual Boc chemistry SPPS using in situ neutralization protocols on Boc-Asn-OCH₂-Pam-resin and Boc-Val-SCH₂CO-Ala-OCH₂-Pam-resin, respectively. Each peptide was cleaved from the resin and simultaneously deprotected by treatment with HF. Removal of the HF-stable DNP group from side chain-protected His residues in peptide 3 was achieved in solution using 50 mM MESNa at pH 7.0, to generate the deprotected peptide-SCH₂CH₂SO₃⁻ thioester. Both unprotected synthetic peptides were purified by preparative reverse phase HPLC and characterized by LCMS analysis (details in Supporting Information).

These three unprotected synthetic peptide segments were used in the preparation of the 51 residue ester insulin polypeptide chain.^[15b] The isotope-labeled peptide-thioester segment Gly^A1-Glu^A4[OβThr^B30-(1-¹³C=¹⁸O)Phe^B24-Thz^B19]-Cys^A6-^αthioester 1 was reacted with peptide Cys^A7-Asn^A21 2 by native chemical ligation at pH 7.0. Reaction progress was monitored by analytical LCMS. Once the ligation reaction was complete, the N-terminal Thz^B19 was converted to Cys^B19 by overnight treatment with MeONH₂.HCl at pH 4. To the same reaction mixture, after readjustment to pH 7.0, was added Phe^B1-Val^B18-^αthioester 3, 50 mM TCEP, and 200 mM MPAA. After the second native chemical ligation reaction was completed, the reaction mixture was diluted ten-fold with water, causing the product full length 51 residue isotope-labeled depsipeptide chain to precipitate. Crude full length polypeptide chain Gly^A1-Glu^A4[OβThr^B30-(1-¹³C=¹⁸O)Phe^B24-Phe^B1]-Gln^A5-Asn^A21 6 was isolated by centrifugation (see Supporting Information).

Without further purification, the lyophilized crude full length polypeptide 6 was folded at near-physiological pH 7.6 under the following conditions: 1.5 M GuHCl, 8 mM cysteine/1 mM cystine redox couple, 20 mM Tris at a polypeptide concentration of 0.05 mg/mL. Folding with concomitant formation of three disulfide bonds was essentially complete within 2 hours as indicated by the appearance of a sharper, earlier eluting peak in LCMS, and a decrease in the observed mass of 5.5 ± 0.8 Da (Figure 3).

Figure 3.

Folding stable-isotope labeled ester insulin, with concomitant formation of disulfide bonds. a) HPLC-traces showing the folding reaction at, t = 0 h and t = 2 h. Conditions: aqueous buffer containing 1.5 M GuHCl, 20 mM Tris, 8 mM Cysteine and 1 mM Cystine.HCl, pH 7.6, polypeptide concentration: 0.05 mg/mL; b) LC-MS data for the purified folded (1-¹³C=¹⁸O)Phe^B24 labeled human ester insulin (inset showing the online ESI-MS spectra taken across the whole of the main UV peak, and observed mass). The chromatographic separations were as described in Figure 2.

Folded isotope-labeled ester insulin was subjected to controlled saponification using 25 mM LiOH in water at pH 12.2 and 4 °C, at a protein concentration of 0.5 mg/mL. We saw improved reaction rate and yield when the reaction was carried out in water, rather than the mixed aqueous-organic solvent previously used.^[15] The reaction was essentially complete after 6 hrs (Figure 4 (a)). Trace amounts of oxidized B chain were formed under these experimental conditions, eluting just before the desired product. Then the reaction mixture was acidified to pH 3 and the isotope-labeled human insulin was purified by preparative reverse phase HPLC and characterized by LCMS (Figure 4 (b)): observed mass 5810.4 ± 0.6 Da; calculated mass for the isotope-labeled peptide 5810.6 Da (average isotope composition, + 3.0 Da for labeled ¹³C=¹⁸O atoms).

Figure 4.

Saponification of isotope-labeled ester insulin to give [(1-¹³C=¹⁸O)Phe^B24]human insulin. Reaction was carried out in 25 mM LiOH in water at pH 12.2 and 4 °C. a) HPLC profiles showing the progress of the reaction as a function of time; b) LC-MS data for the [(1-¹³C=¹⁸O)Phe^B24]human insulin after HPLC purification (inset showing the online ESI-MS spectra across the whole of the UV absorbing peak). Reverse phase HPLC under the same conditions as for Figure 3. (Bottom panel) High resolution direct infusion ESI-MS spectra of [(1-¹³C=¹⁸O)Phe^B24]human insulin M+4H⁺ m/z region. Inset showing the monoisotope regions from [(1-¹³C=¹⁸O)Phe^B24]human insulin and [(1-¹³C=¹⁶O)Phe^B24]human insulin. The ratio between those two peaks was 96.7% which corresponds closely to the (1-¹³C=¹⁸O)Phe^B24 isotope enrichment in the final protein molecule.

Key aspects of the synthesis of the human insulin isotopologue were as follows. Initial attempts to carry out isotope exchange reaction of Boc-(1-¹³C)Phe-OH with H₂¹⁸O were not fruitful when we followed a previously reported protocol for the efficient ¹⁸O isotope exchange of N^α-protected amino acids.^[17] We then carried out the Boc-protection/isotope exchange reaction in the reverse order. First, we conducted the exchange reaction of H-(1-¹³C)Phe-OH with a 1:1 mixture of H₂¹⁸O/dioxane at 100 °C;^[18] After a few cycles of the exchange reaction, satisfactory enrichment was obtained. Lyophilization followed by Boc-protection in dioxane/water and rapid acid-base extraction workup gave essentially pure Boc-(1-¹³C=¹⁸O)Phe.

Synthesis of the labeled peptide segment Gly^A1-Glu^A4[OβThr^B30-Phe^B24(1-¹³C=¹⁸O)-Thz^B19]-Cys^A6-^αthioester 1 was carried out using Boc chemistry in-situ neutralization SPPS protocols,^[19] in order to preserve the thioester moiety. After addition of the ester-linked dipeptide Boc-Glu^A4[Oβ(Alloc-L-Thr^B30-α-OcHex)]-OH with its three orthogonal protecting groups, each removable under distinct conditions (mild acidolysis; reductive cleavage; strong acidolysis), the Boc-group was removed and the remainder of the A-chain sequence was added through residue Gly^A1. The N-terminus of Gly^A1 was re-protected as 2-Cl-Z, then the Alloc protecting group was removed using Pd(PPh₃)₄/phenylsilane under argon atmosphere, and the sequence was assembled through Phe^B25. Amino acid Boc-(1-¹³C=¹⁸O)Phe^B24 was incorporated using modified coupling conditions to minimize the amount of valuable labeled amino acid used. Then the synthesis was continued through the coupling of Thz^B19, which was used as a cryptic form of N-terminal Cys^[20] to prevent undesired intramolecular cyclization. ^[21]

Covalent condensation of the three unprotected peptide segments Phe^B1-Val^B18-^αthioester, Gly^A1-Glu^A4[OβThr^B30-(1-¹³C=¹⁸O)Phe^B24-Thz^B19]-Cys^A6-^αthioester, and Cys^A7-Asn^A21 in a series of one-pot reactions was straightforward, and was carried out without isolation/purification of intermediate products. The linear 51-residue ester linked depsipeptide thus obtained was precipitated by addition of water and the partially-purified product was folded using a cysteine/cystine redox couple at pH 7.6 at a polypeptide concentration of 0.05 mg/mL. Attempts to carry out the folding at higher polypeptide concentrations of 0.3 mg/mL and 0.1 mg/mL gave poor yields of the folded ester insulin. A product eluting as a much earlier, sharper peak on analytical reverse phase HPLC together with a decrease in mass of 5.5 ± 0.8 Da was consistent with the formation of a folded ester insulin protein molecule containing three disulfide bonds. The overall yield of ester insulin based on limiting thioester peptide 1 was 25% (20 mg ester insulin was obtained from 80 mg of partially purified polypeptide). Saponification of ester insulin to form the native insulin protein molecule gave a somewhat later eluting peak that had an 18 Da mass increase, indicative of addition of the elements of water. The reaction proceeded smoothly and gave a 63% isolated yield (7.6 mg insulin from 12 mg of ester insulin). High resolution direct-infusion ESI MS of the purified isotope-labeled synthetic protein gave an observed monoisotopic mass of 5806.641 ± 0.003 Da (theoretical monoisotopic mass: 5806.645 Da).

High resolution direct infusion ESI mass spectrometry data of the purified synthetic [(1-¹³C=¹⁸O)Phe^B24]human insulin was also used to establish the ¹⁸O-labeling efficiency in the final insulin protein molecule (Figure 4, bottom panel). This was to see if there is any significant decrease in the degree of ¹⁸O-labeling in the final labeled protein synthetic product arising from chemical manipulations such as amino acid coupling conditions, acid treatment during Boc removals, and especially during final saponification with LiOH in water, which could lead to oxygen exchange and lower the enrichment. Analysis of the monoisotopic region revealed two distinct peaks at m/z 1452.1671 ± 0.0008 and 1452.6681 ± 0.0008 for the M+4H⁺ charge states of synthetic [(1-¹³C=¹⁶O)Phe^B24]human insulin (Theoretical monoisotopic M+4H⁺ m/z 1452.1681) and [(1-¹³C=¹⁸O)Phe^B24]human insulin (Theoretical monoisotopic M+4H⁺ m/z 1452.6692) respectively. Using the ratio between these two peaks the ¹⁸O enrichment was found to be 96.7%, which is in excellent agreement with the measured 97.1% enrichment of the starting (1-¹³C=¹⁸O)Phe amino acid (see Supporting Information), showing that within experimental error there was no loss of ¹⁸O during the chemical manipulations involved in the synthesis.

The crystal structure of purified [(1-¹³C=¹⁸O)Phe^B24]human insulin at a resolution of 1.25Å unambiguously confirmed the amino acid sequence, the covalent structure including disulfide bonds, and the correctly folded tertiary structure of the synthetic protein molecule (Figure 5).^[22].

Figure 5.

X-ray crystal structure of [(1-¹³C=¹⁸O)Phe^B24]human insulin PDB code 5ENA. Cartoon representations of the A chain (green) and B chain (cyan). Disulfide bonds are shown in yellow. The site of isotope labeling at (1-¹³C=¹⁸O)Phe^B24 is shown as stick representation (O, red).

Synthetic [(1-¹³C=¹⁸O)Phe^B24]human insulin was assayed for in vitro binding to the human insulin receptor, using wild-type biosynthetic human insulin for comparison (Figure S10). The receptor-binding activity of labeled human insulin was indistinguishable from that of wild-type human insulin within experimental uncertainty.

Finally, to demonstrate that the site-specific backbone bone carbonyl ¹³C=¹⁸O isotope label can be a useful probe of the dimer interface in the insulin self-association process, we performed linear and 2D IR spectroscopy of the [(1-¹³C=¹⁸O)Phe^B24]human insulin protein molecule (Figure 6). Both linear and 2D IR spectra reveal a distinct peak at 1595 cm⁻¹ resulting from the (1-¹³C=¹⁸O)Phe^B24 label, red shifted from the main amide I band. The peak is clearest for experimental conditions where the dimer population is known to dominate. This peak broadens and decreases in amplitude as the temperature is increased, consistent with solvent exposure of the label as the dimer population decreases.

Figure 6.

FTIR and 2D IR spectra of [(1-¹³C=¹⁸O)Phe^B24]human insulin in deuterated solvent at a concentration of 5 mg/mL in phosphate buffer and 20% ethanol at pD 2.0. (A) Linear FTIR spectra for the insulin amide I carbonyl stretch. [Inset: the FTIR difference spectrum of [(1-¹³C=¹⁸O)Phe^B24]human insulin from human insulin for the spectral region of the isotope label] (B) Temperature difference FTIR spectra of the (¹³C=¹⁸O)Phe^B24 isotope label relative to 55°C; 15° (blue) to 55°C (red). (C) 2D IR spectra at 15°C and (D) 55°C (waiting time of 150 fs).

In the work reported here, we have optimized a convergent route to the total synthesis of human insulin and applied it to the efficient site-specific labeling of the human insulin protein molecule. Ten milligram amounts of site-specifically labeled human insulin were obtained from a lab scale synthesis. The X-ray crystal structure of the synthetic protein molecule unambiguously confirmed its covalent structure, as well as the native disulfide bonds and correctly folded protein structure. The isotope-labeled [(1-¹³C=¹⁸O)Phe^B24]human insulin was fully active in insulin receptor binding. The single ¹³C=¹⁸O carbonyl was readily distinguished from the main amide I band in 2D FTIR spectra of the labeled protein molecule. In further work, we have prepared ten milligram amounts of the human insulin protein isotopomers individually backbone labeled with ¹³C=¹⁸O at Gly^B23 and Gly^B8, and a human insulin protein isotopologue double labeled with ¹³C=¹⁸O at both Gly^B20 and Gly^B23, for 2D FTIR studies of this important protein (data not shown).

Our results demonstrate the versatility and utility of total protein synthesis based on modern chemical ligation methods. The work described here overcomes the size limitations of stepwise SPPS, and will extend site-specific isotope labeling of backbone carbonyls to a wide range of target protein molecules, by total chemical^[23] or semi-synthesis.^[24] The synthetic approaches used in this work will be useful for the study of a variety of aspects of the molecular basis of function in human insulin and other protein molecules by 2D FTIR.