Selective Late-Stage Functionalization of Tryptophan-Containing Peptides To Facilitate Bioorthogonal Tetrazine Ligation

Abstract

Site-selective modification of complex peptides and the functionalization of their C–H bonds hold great promise for expanding their use in therapeutics and biomedical research. Herein, we leverage the power of late-stage chemoenzymatic catalysis using an indole prenyltransferase (IPT) enzyme and alkyl diphosphates to specifically modify the indole ring of tryptophan in clinically relevant peptides. Furthermore, the installed handle enables bioorthogonal click chemistry through an inverse electron-demand Diels–Alder (IEDDA) reaction with a biotin-conjugated tetrazine probe.

Graphical Abstract

Peptides are known to play significant roles in many physiological and biochemical processes.¹ They have wide applications in molecular imaging techniques and disease-specific probe development and serve as invaluable platforms for drug delivery and development. In this context, the tryptophan (Trp) residue contributes to the overall structure, functional stability, and activity of peptides and proteins.² The indole side chain of Trp possesses unique characters including electron-rich π system, hydrophobicity, poor nucleophilicity, and photoelectronic features.³ Trp has low natural abundance and is often involved in significant biological roles, such as ligand binding, enzyme catalysis, and signal transduction.⁴ Moreover, Trp exhibits strong absorption in the ultraviolet range, enabling its utility as a valuable tool in spectroscopic techniques for investigating peptide folding, conformational changes, and protein–protein interactions.⁵ Altering the structures of tryptophan-containing peptides (TCPs) can enhance their biological activities and improve their chemical and physical properties, which expands their utilization in different fields. Modified peptides can be achieved via de novo linear synthesis or by derivatization of the final product through late-stage functionalization (LSF). LSF offers a cost- and time-effective approach to functionalize C–H bonds without the need for numerous synthetic steps and labor-intensive workup and purification procedures.⁶ Additionally, LSF allows for the introduction of reactive handles amenable to bioorthogonal chemical reactions.^7,8 However, site-selective LSF of peptides is challenging as a result of the presence of several reactive groups, stereocenters that are prone to racemization, multiple C–H bonds, or the need for physiological conditions. While chemical,^9,10 enzymatic,⁸ and photocatalytic¹¹ methods have been developed for peptide LSF, they have faced challenges including focus on residues carrying nucleophilic functional groups, selectivity concerns, or the use of toxic metals or harsh conditions that affects the structural integrity of delicate macromolecules.

Indole prenyltransferases (IPTs) are dual substrate enzymes that catalyze the addition of isoprene moieties onto indolecontaining molecules in normal (C1′) and reverse (C3′) fashion (Figure 1).¹² IPTs can accommodate indole- and tryptophan-derived compounds as acceptors and native prenyl pyrophosphate substrates as donors. IPTs have shown flexibility in both their acceptor and donor substrate panel.¹³ Recently, we reported the selective modification of the Trp residue in daptomycin, a potent United States Food and Drug Administration (FDA)-approved antibiotic, and the synthesis of analogues, some with improved activity and different mechanisms.^14,15 This inspired us to determine if other complex peptides with different numbers and positions of Trp can be selectively targeted. In this work, we exploit the catalytic power of IPTs to achieve LSF of biologically active TCPs and selectively modify the benzene ring of Trp in complex peptides (Figure 2A). Conversion yields and kinetic data show selectivity toward Trp at the termini when more than one Trp is present. We also utilize the installed handle to enable bioorthogonal click chemistry through the inverse electron-demand Diels–Alder (IEDDA) reaction.

Figure 1.

Enzymatic activity of IPT enzymes in the presence of alkyl/aryl pyrophosphate donors and indole- and tryptophan-containing compounds.

Figure 2.

(A)Methodology, (B) tryptophan-containing peptide (TCP) acceptors (1a–1d), and (C) pyrophosphate donors (2a–2h) used in this study. Isolated yields after column chromatography are mentioned in parentheses.

Our selected peptides (Figure 2B) include clinically relevant or bioactive peptides: alarelin (H-Pyr-His-Trp-Ser-Tyr-dAla-Leu-Arg-Pro-NHEt, 1a), a nonapeptide agonist for the gonadotropin-releasing hormone receptor that stimulates the release of follicle-stimulating and luteinizing hormones (FSH and LH, respectively), and a hexapeptide (Trp-His-Trp-Leu-Gln-Leu, 1b), which is the N fragment of the α-mating factor that binds to a G-protein-coupled receptor in yeast to signal conjugation and mating. Additionally, we alkylate the decapeptides nafarelin (H-Pyr-His-Trp-Ser-Tyr-d2NaAla-Arg-Pro-Gly-NH₂, Synarel, 1c) and triptorelin (H-Pyr-His-Trp-Ser-Tyr-dTrp-Leu-Arg-Pro-Gly-NH₂, Trelstar/Triptodur, 1d) which are gonadotropin-releasing hormone receptor agonists (Figure 2B and Table S1 of the Supporting Information). Nafarelin is used to treat central precocious puberty and endometriosis, while triptorelin is used for the treatment of advanced prostate cancer. These TCPs contain different amino acid residues with variable reactive moieties, such as the guanidine group of Arg, imidazole ring of His, hydroxylic group of Ser or Tyr, aromatic group of Tyr, and naphthalene, pyroglutamine, proline, or ethylamide moieties. The selected peptides also vary in the number and position of Trp in the peptide chain (Figure 2B and Table S1 of the Supporting Information).

Our previous data showed CdpNPT to have broad promiscuous catalytic activity toward cyclic peptides.^14,15 Thus, the plasmid encoding cdpNPT and transformed into Escherichia coli BL21(DE3) cells was used to overproduce the corresponding protein (Figure S1 of the Supporting Information). N-His₆ recombinant CdpNPT was purified to homogeneity via nickel chelation to yield 11 mg mL⁻¹ (Figure S2 of the Supporting Information). Additionally, we synthesized eight diphosphate donors 2a–2h (Figure 2C) as previously described¹⁴ and purified using ion-exchange chromatography to provide overall yields of 14–39%. The selected synthesized donors had different functionalities and included the native substrate CdpNPT donor dimethylallylpyr-ophosphate (DMAPP, 2a) and the two natural IPT donors, geranyl and farnesyl pyrophosphates (GPP, 2b; FPP, 2c). We also synthesized two allyl analogues: compound 2d, the native compound 2a isomer, and compound 2e. Furthermore we synthesized the cycloalkene 2f, as well as the two aromatic analogues 2g and 2h. The structures of all pyrophosphate substrates were confirmed using high-resolution mass spectrometry (HRMS) and ¹H and ³¹P nuclear magnetic resonance (NMR) (Supporting MS and NMR Spectra of the Supporting Information). Analytical-scale reactions containing 0.8 mM of one of compounds 1a–1d, 1.6 mM compounds 2a–2h and 43 μM CdpNPT in 50 mM Tris supplemented with 10 mM CaCl₂ (pH 8.0), were incubated at 37 °C for 16 h (Figure 2A). Reactions under similar conditions lacking the enzyme were used as a control. High-performance liquid chromatography (HPLC) analyses (Figures S3–S6 of the Supporting Information) confirmed by mass spectrometry (MS) (Table S2 of the Supporting Information) showed reactions with compound 2a (CdpNPT native donor) having the highest conversion. Naturally occurring donors, compounds 2b and 2c, showed very low or no conversion with compounds 1a–1d, respectively, which is consistent with CdpNPT limitations with accommodating longer chain donors and, thus, hindering any reliable kinetic studies.^14,16 Moreover, no products were formed when using two Tyr-containing peptides lacking Trp as acceptor substrates showing the targeted specificity toward Trp (Figures S7 and S8 of the Supporting Information). Subsequently, Michaelis–Menten kinetics of CdpNPT with each of the four peptides 1a–1d and all donors that showed HPLC conversion higher than 5% were determined using GraphPad Prism 10.0.0 with nonlinear regression (Figures S9–S12 and Table S3 of the Supporting Information). In general, the enzyme total catalytic efficiency of compound 2 donors with compound 1b were 2–14-fold higher than that with compound 1a, 1c, or 1d. This disparity might suggest that the terminal Trp in compound 1b is more readily accessible for catalysis compared to other peptides where Trp is positioned in the middle of the peptide sequence. Our findings are further supported by analysis of CdpNPT kinetics with the cyclic lipopeptide daptomycin, which also features a terminal Trp that we reported earlier.¹⁴ The CdpNPT k_cat turnover numbers with daptomycin and compounds 2d, 2f, and 2h, while lower than that with compound 1b, exhibited a 2–11-fold increase compared to that with compounds 1a, 1c, and 1d. Furthermore, as expected and consistent with our HPLC data, CdpNPT k_cat turnover numbers of each of compounds 1a–1d (Figures S9–S12 and Table S3 of the Supporting Information) with native compound 2a (151.1 × 10⁻³–10.4 × 10⁻³ min⁻¹) was higher than any other donor tested. Other kinetically competitive donors include the compound 2a isomer, compound 2d (83.5 × 10⁻³–4.5 × 10⁻³ min¹), and cyclic alkene 2f (36.2 × 10⁻³–2 × 10^–3 min⁻¹). However, reactions with either compound 2a or 2f lead to appearance of multiple monoalkylated peaks compared to compound 2d, probably as a result of the formation of reverse (C3′) products^17–19 or modifications at different Trp positions. Thus, donor 2d was selected for larger scale reactions.

To identify the amino acid residue and position modified, 50 mM Tris/10 mMCaCl₂ (pH 8.0) containing 0.22–0.90 mM of each of compounds 1a–1d, 0.8–1.6 mM of compound 2d, and 22–43 μM of CdpNPT were prepared. HPLC–MS revealed the formation of one or major product 3 with each peptide with conversion yields of 21.9–90.8% (Figure 3 and Figures S13–S16 of the Supporting Information). Semi-preparative reverse-phase HPLC was used to purify the major products 3a–3d. The (+)-HR-ESI-MS [M + H]⁺ m/z 1235.6707, 950.5265, 1390.7060, and 1379.6988 indicated the molecular formulas of C₆₁H₈₆N₁₆O₁₂, C₅₀H₆₇N₁₁O₈, C₇₁H₉₁N₁₇O₁₃, and C₆₉H₉₀N₁₈O₁₃, respectively. The Δm/z of 68 between each peptide 1a–1d and the corresponding products 3a–3d was consistent with an additional –C₅H₈– group in each product (Tables S1 and S4 and Supporting Mass and NMR Spectra of the Supporting Information). Tandem MS fragmentation showed the modified amino acid to be Trp in all four peptides (Figure 3 and Figures S17–S28 of the Supporting Information). For peptides containing more than one Trp, modification occurred at the residue at one of the termini or closer to the end of the chain (Trp₁ for compound 1b and Trp₃ for compound 1d). This could be attributed to the way the peptide docks in the center of the IPT tunnel of the ABBA barrel, making it more accessible. Analysis of the 1D and 2D NMR spectroscopic data of each of compounds 3a–3d confirmed the presence of a pent-2-en-1-yl group indicating a normal (C-1′) and not reverse (C-3′) reaction (Figure 3 and Figures S29–S32 and Tables S5–S8 of the Supporting Information). The correlation spectroscopy (COSY) correlations between H-4/H-5 and H-1/H-2 as well as the heteronuclear multiple-bond correlations (HMBCs) between H-5 and C-1′/C-7, between H-7 and C-1′/C-5, between H-2′ and C-6, and between H-1′ and C-5/C-7 determined the regiospecificity to be at the C-6 of Trp in each of compounds 3a–3d (Figure 3 and Supporting Experimental Section and Figures S29–S32 of the Supporting Information). The configurational assignments of 2′-E were deduced by calculating the coupling constants of H-2′ (dt, J = 15.2 and 5.2 Hz) and the absence of nuclear Overhauser effect spectroscopy (NOESY) correlation.

Figure 3.

HPLC chromatograms, 1D and 2D NMR correlations, and MS/MS fragmentation data for (A) compound 3a (conversion of 27.6%) and (B) compound 3b (conversion of 90.8%). ¹H–¹H COSY correlations (—) and ¹H–¹³C HMBCs (→) are shown. All highlighted fragment ions were matched with a maximum deviation of 10 ppm from their theoretical masses.

Having shown that selective Trp C–C bond formation was achieved on TCPs, we next sought to establish if bioorthogonal click chemistry can be achieved on the enzymatically installed olefin handle. Tetrazine-containing tags react with electron-rich alkenes or alkynes via the IEDDA reaction to generate dihydropyridazine isomers that can be further oxidized to pyridazines.²⁰ These tetrazines can be bound to affinity tags or fluorescent labels to facilitate imaging, mapping, and labeling purposes.²¹ Hence, tetrazine-biotin dye substrate 4 was selected and incubated with equimolar concentrations of either compound 3a or 3b in aqueous solution of dimethyl sulfoxide (DMSO) for 16 h at room temperature (Figure 4A). The reaction was analyzed using HPLC–HRMS. The presence of [M + H]⁺ and [M + 2H]²⁺ m/z 1620.8193 and 810.9142 in compounds 3a/4 (C₈₀H₁₀₉N₂₁O₁₄S, compound 5a; Figure S33 and Table S9 of the Supporting Information) and 1335.6759 and 668.3423 in compounds 3b/4 (C₆₉H₉₀N₁₆O₁₀S, compound 5b; Figure S34 and Table S9 of the Supporting Information) reactions, respectively, were in agreement with the tetrazine ligation and subsequent formation of dihydropyridazine derivatives. Most importantly, the tandem MS pattern of each of compound 5a (Figure 4B and Figures S35–S37 of the Supporting Information) and compound 5b (Figure 4C and Figures S38–S40 of the Supporting Information) reaction was consistent with the biotin labeling on enzymatically modified Trp in each peptide. Noteworthy, the y6 (733.4337) and y7 (1372.7426) ions in compound 5a (Figures S35–S37 of the Supporting Information) and the b1-28 (loss of C=O, 612.3108) and y5 (696.3831) ions in compound 5b spectra (Figures S38–S40 of the Supporting Information) unambiguously demonstrated that enzymatically modified Trp in both compounds 5a and 5b underwent click chemistry. A reaction between the unmodified compound 1a or 1b and compound 4 was set up as a control under the same conditions, and no click products were detected. This shows the specificity of the IEDDA reaction to the installed handle.

Figure 4.

Bioorthogonal tetrazine ligation on the 2-pentenyl handle. (A) Reaction between Trp-modified peptides 3a/3b and tetrazine biotin 4 to generate compounds 5a/5b, respectively. Tandem MS fragmentation patterns of (B) compound 5a and (C) compound 5b. All highlighted fragment ions were matched with a maximum deviation of 10 ppm from their theoretical masses. Detailed tandem MS spectra are available in Figures S35–S40 of the Supporting Information.

The work described herein adds to the growing role of biocatalysis in the selective modification of peptides.^8,22,23 Our approach is distinctive compared to other reported methods used to chemoenzymatically label Trp in peptides. First, we demonstrate the ability to selectively target Trp in the presence of other residues and reactive functionalities on broad complex cyclic^14,15 and linear peptides with diverse handles. Second, we target the Trp indole C6 rather than N1, creating a stable C–C bond with the formation of a single or major product not liable for decomposition or reversible reaction.¹⁹ Remarkably, the challenging position to modify as a result of its lower nucleophilicity requires transition metals or directing groups in other studies.³ Third, the substrates used are clinically relevant hormones and peptides, demonstrating the biological significance. Fourth, the use of a 2-pentenyl handle reduces the possibilities of rearranged products and facilitates peptide tagging with biotin–dye substrates via click reaction. In contrast to methods targeting highly nucleophilic amino acids,¹⁰ such as Cys, or highly abundant residues, such as Lys, we selectively target Trp, an amino acid that plays a critical role in peptides, with low nucleophilicity and low natural abundance.

In short, we developed bioactive Trp-modified peptides under biological ambient temperature, pH, and aqueous conditions using a versatile late-stage chemoenzymatic functionalization method. We identified the structures using HRMS, tandem MS, and 1D and 2D NMR. Trp at or close to the termini of a peptide was more accessible and easier to modify even in the presence of other Trp residues, leading to higher conversion yields. In addition, the installed handle enabled downstream click chemistry via an interaction between the installed olefin handle and a tetrazine–biotin substrate. IPT engineering efforts to alter promiscuity,^24,25 donor substrate preference,¹⁶ or regiospecificity²⁶ will facilitate future improvement of desired activities. This approach opens the door for the capture and imaging of TCPs and other applications in chemical biology as well as the development of drug leads.

Data Availability Statement

The data underlying this study are available in the published article and its Supporting Information.