Facile de novo Sequencing of Tetrazine-Cyclized Peptides Through UV-Induced Ring-Opening and Cleavage from Solid Phase

Abstract

While most FDA-approved peptide drugs are cyclic, robust cyclization chemistry of peptides and the deconvolution of the cyclic peptide sequences using tandem mass spectrometry render cyclic peptide drug discovery difficult. Here we present a successful design of cyclic peptides on solid phase that addresses both these problems. We demonstrate that this peptide cyclization method using dichloro-s-tetrazine on solid-phase allows successful cyclization of a panel of random peptide sequences with varying charges and hydrophobicity. The cyclic peptides can be linearized and cleaved from the solid phase by simple UV light irradiation, and we demonstrate that accurate sequence information can be obtained for the UV-cleaved linearized peptides using tandem mass spectrometry. The tetrazine linker used in the cyclic peptides can further be explored for inverse electron-demand Diels Alder (IEDDA) reactions for screening or bioconjugation applications in the future.

Graphical Abstract

The successful design and synthesis of solid phase-bound monocyclic peptides with varying charges and hydrophobicity are presented. The design allows the bead-bound peptides to be linearized and cleaved from the solid phase by a single-step UV irradiation, allowing facile sequence identification using tandem mass spectrometry.

Introduction

Cyclic peptides are intermediate between small molecules and protein therapeutics in size and physicochemical properties. While linear ligands can adopt various conformations on binding proteins, cyclic peptides, on protein binding, are relatively restricted in their structural geometries^[1]. This results in cyclic ligands binding proteins with higher affinities than their linear analogs^[2], as they do not pay the entropic penalty on protein binding. In addition, cyclization can enhance resistance against proteolysis^[3], increasing the half-life of a ligand significantly^[4]. Cyclic peptides are, therefore, particularly attractive for targeting shallow extended protein-protein interactions (PPIs), often considered “undruggable” by small molecule drugs^[5]. The cyclic peptides’ enhanced stabilities and higher affinities for their targets have encouraged the development of cyclic peptide libraries and screening of the libraries for discovery of potential therapeutic candidates over the past several decades.

Cyclic peptide libraries can be synthesized biologically or chemically. They contain anywhere between thousands to 10¹⁸ unique peptide sequences and can be screened for binding to affect specific biological functions of the targets. Commonly used combinatorial peptide library technologies include phage display^[6], mRNA library^[7], One-bead-one-compound (OBOC) library^[8] and DNA encoded libraries^[9]. While disulfide-cyclized cyclic phage display libraries have been used for screening^[10], a significant drawback to this library is the low stability of the disulfide bond. Recently some phage libraries, such as cyclic or bicyclic phage libraries, have been cyclized through stable thioether^[11] and S-Aryl^[12] linkages. But even these libraries cannot undergo the harsh washes with organic solvents required during screening due to the phage’s biological nature. It is also non-trivial to incorporate multiple unnatural amino acids in these libraries. In contrast, chemically synthesized peptide libraries such as One-Bead-One-Compound (OBOC) peptide libraries allow easy incorporation of multiple unnatural amino acids, which may subsequently be used in cyclizing the peptide libraries^[13].

Chemically synthesized peptide libraries can be cyclized using reactions such as lactamization^[14], lactonization^[15], disulfide bond formation^[16], thioether formation^[17], and ring-closing metastasis^[18]. The different cyclization methods and position of the cyclization linker within the peptide can significantly impact the library’s homogeneous nature and isolated binders’ binding affinities and specificities. Therefore, choosing the appropriate ring size of the cyclic peptides and developing a robust cyclization procedure is essential for comprehensive library synthesis.

We previously reported on developing a seven-membered stapled library where peptides are cyclized through a triazole linker or an alkene staple, using copper-catalyzed Azide Alkyne cycloaddition (CuAAC) ‘click’ reaction or ruthenium-catalyzed Ring Closing Metathesis (RCM), respectively^[18]. For the click-cyclized library, five amino acids spaced between an alkyne-containing unnatural amino acid and an azide-containing unnatural amino acid were found to produce the ideal ring size to produce homogeneous monomeric cycles on resin^[19]. Screening this library yielded ligands that detected cancer biomarkers, malarial biomarkers^[18], ligands that inhibited kinase activity, inhibited heme sequestration^[20], inhibited botulinum neurotoxin uptake in cells^[21]; or modulated protein folding^[22]. However, sequencing of both the click-cyclized and the RCM-cyclized libraries required use of the time-consuming and arcane Edman peptide sequencing technology.

In this article, we report the utilization of a stapling reaction between cysteine (Cys) side chains by using a bis-functional reagent to develop cyclic peptides, and, eventually, OBOC cyclic peptide libraries. We use SnAr^[23], conveniently on-resin with the protected peptides, to get a series of (i, i+4) stapled peptides with a rigid aromatic linker. Unlike the previously reported libraries, this stapling scheme allows simplification of the peptide sequencing process through the use of Mass Spectrometry (MS).

For successful use of OBOC cyclic libraries for ligand discovery, it is critical to design the library so that cyclic peptides can be unambiguously sequenced. “While de novo sequencing has been successfully used to screen ligands from large linear peptide/hybrid libraries using technologies such as AS-MS^[24], interpreting de novo spectra are far more formidable for cyclic peptides, such as the click-cyclized peptides, due to non-specific peptide bond dissociation along the ring structure during MS. This can lead to a complex spectrum, each corresponding to a peptide fragment with starting position i, (1 ≤ i ≤ n)^[25]. The bioinformatics tools for high throughput de novo sequencing of cyclic peptides are still in infancy and constitute a significant limitation of this approach^[26].

To overcome the problematic MS/MS interpretation of cyclic peptide libraries, some strategies such as ladder sequencing^[27] and the one-bead-two-compound (OBTC) approach^{[14, 28]} have been developed. Spatially segregated OBTC libraries contain cyclic peptides on the bead surface and linear encoding peptides in their interior, but the difference in coupling and removal efficiencies of the orthogonal protection groups in the interior and exterior can lead to errors in synthesis and sequencing.

Another approach to overcoming cyclic peptide MS sequencing issues is to design the cyclic peptides such that the peptides can be linearized and cleaved from the resin by chemical treatment and then sequenced by MS/MS. Biron^[29] and Kodadek^[30] et al. developed similar technologies by adding methionine at two positions in a cyclic peptide/peptoid – one at the C-terminal and another inside the sequence – for simultaneous ring cleavage and cleavage from resin on treatment with cyanogen bromide. However, this strategy required positioning a methionine (Met) inside the cyclic ring at an invariable position, which might not be ideal for binding to a protein target and decreased library diversity. Removal of Met also used hazardous materials like cyanogen bromide. Recently Raj et al^[31] reported a cyclic peptide strategy, in which serine is incorporated instead of methionine inside and outside the cyclic peptide. Serine forms oxazolidinone by exposure to water under basic conditions and opens the cyclic peptide. This library has decreased library diversity, like the Met-incorporating libraries. Moreover, since most biological screens are done in aqueous buffer conditions where pH changes are part of the screening process^[18], such a library may undergo ring opening during screening. In this article, we report an alternative cyclic peptide design that allows the use of UV light, instead of hazardous reagents, for sequential or simultaneous linearization of peptide and its cleavage from the solid-phase resin beads. The tetrazine cyclic peptides reported here are stable in physiological conditions and highly acidic and basic conditions in time frames used for screening^[32].

Our approach relies on the selective incorporation of UV-cleavable cyclization linker, specifically S, S-dithiotetrazine inside the cyclic peptide, and the incorporation of a UV-cleavable group, specifically 3-amino-3-(2-nitrophenyl)propionic acid^[33] (ANP) outside the cyclic peptide, at the C-terminal. Brown et al described the successful synthesis of several tetrazine-cyclized peptides in solution and on solid support^{[32, 34]}. However, the solid-state cyclization was reported as low yielding (7–30%)^[34c] or unsuccessful^[34d]. The authors also demonstrated that the S, S-dithiotetrazine linker could be cleaved by UV light irradiation at 312 nm^[32]. We have optimized the on-bead cyclization reaction to obtain yields that were obtained previously during in-solution cyclization and have designed tetrazine-cyclized peptides that can be used as OBOC cyclic peptide libraries. Following the incorporation of ANP and a fixed glycine residue at the C-terminal of the peptide on the solid support (Tentagel S-NH₂ resin), two Cys residues with selectively removable sidechain protecting groups were incorporated in the peptide at fixed positions (Scheme 1). After chain assembly, the sidechain of the two Cys was selectively deprotected, reduced to thiol by treatment with a reductant, and activated by the electrophile dichloro-s-tetrazine resulting in the formation of a cyclic tetrazine linked peptide moiety on the resin (Scheme 1). Next, photochemical cleavage of the ANP linker by treatment with 350 nm UV light^[35] released the cyclic tetrazine-linked peptide from the solid support, which could then be deprotected in solution and quantified for the cyclization yield (Scheme 1A). Treatment of the on-bead peptide with UV light at 310 nm and 350 nm, on the other hand, released the dithiocyanate-containing linear peptide into solution, which was used for MS/MS sequencing (Scheme 1B). For the resulting dithiocyanate containing linear peptide, tandem or multistage (MSn) mass could easily differentiate the b-ion and y-ion series and, when used in conjunction with de novo sequencing algorithms, allowed for accurate interpretation and sequencing.

Scheme 1.

Overview of UV-cleavable cyclic peptide library synthesis and its utilization in A. quantifying cyclic peptide yield and B. de novo MS/MS sequencing.

Results and Discussion

To test the proposed methodology for synthesizing on-bead tetrazine-cyclized peptides, several cyclic peptides were synthesized on a solid support to confirm that the desired intramolecular cyclized monomers were formed as the major products without significant percentages of intermolecular reactions creating dimeric or polymeric cyclized products. Based on reports of previous successful tetrazine cyclization mainly in solution, we incorporated two Cys at (i, i+4) positions, with three amino acids between the two Cys in the model peptides^[32]. The linear peptides with 2 Cys at (i, i+4) groups with various protecting groups on the thiols were synthesized, the protecting groups were selectively removed, and cyclization of the peptide on-resin was done by treatment with 0.5 equivalents twice with dichloro-s-tetrazine. The peptides were treated with TFA for global deprotection, cleaved from the Rink Amide resin, and quantified using HPLC and LC-MS.

Two conditions had to be optimized for successfully synthesizing the (i, i+4) cyclized peptide –the efficient removal of the protecting group from Cys and the robust cyclization of the bead-bound peptide by treatment with dichloro-s-tetrazine. Selective removed protecting groups for Cys thiol that are conventionally used are Trityl (Trt) group^[36], 4-Monomethoxytrityl (Mmt)^[36] group, and Acetamidomethyl (Acm)^[36] group. Therefore, peptides were synthesized with these protecting groups.

Although the Trityl group has been successfully used for Cys side chain selective removal, linear peptide sequence Ac-NH-β-Ala-AFC(Trt)-CONH2, containing one Cys(Trt) group, could not be synthesized following literature protocol^[37]. The Trt group was not further pursued as a selective deprotection group as the cleavage of the peptide from the Rink amide resin was observed during the selective deprotection. Five peptides containing two Cys at (i, i+4) were synthesized, containing protective groups Acm or Mmt (Scheme 2, Table 1). The Acm or Mmt protecting groups were removed following literature protocols^[36], followed by cyclization with dichloro-s-tetrazine using the optimized cyclization protocol we developed and cleavage from solid support. The cyclic peptides cleaved from Rink amide resin (loading capacity 0.00051 moles/gm) were purified by reverse-phase HPLC (RP-HPLC), and the main fractions were subjected to LC-MS. The fraction yielding the correct mass was identified, and the area corresponding to the fraction on the RP-HPLC was calculated, as a percentage of the total area in the chromatogram, to quantify the cyclic peptide. Though tetrazine has a high absorbance at 280 nm, since for each protecting group, the same cyclic peptide sequence is synthesized, and their comparative yields are compared, any effect of tetrazine absorbance at 280 nm should be present to the same extent in the observed yields. The cyclization yields in table 1 are referred to as ‘crude’ yields as they are calculated using 280 nm absorbance. Comparing the crude yields of the cyclic peptides using the two different protecting groups allowed us to determine which protecting group (PG) would be optimal for the cyclization process. The removal efficiency of the Acm and Mmt groups was comparable for peptide 1 (Table 1, Figure S.1). but for peptide 2, Mmt group gave better yield (Table 1, Figure S.2). A third peptide 3, when synthesized using Mmt, also gave a good yield (Table 1, Figure S.3). Since the deprotection of Acm involved toxic heavy metal reagents, and Mmt appeared to function same or better as a selectively removable protecting group, Mmt was chosen as the best option for further use in peptide cyclization.

Scheme 2.

Estimation of PG removal efficiency from cyclization

Table 1.

Selection of the optimal side chain protecting group of cysteine from the comparison of cyclic peptide yield from two cysteine-containing peptides

Number	Peptide Sequence	PG	Cyclic peptideSI figure	Percentage crude yield of cyclic peptide[a]
1	Ac-NH-C(PG)-HWS-C(PG)G- CONH2	Acm Mmt	Ac-NH(CHWS C)TetrG-CONH2 1	49.78 47.19
2	Ac-NH-C(PG)-TYT-C(PG)- CONH2	Acm Mmt	Ac-NH(CTYT C)TetrG-CONH2 2	26.34 74.77
3	Ac-NH-C(PG)-VGA-C(PG)-G- CONH2	Mmt	Ac-NH(CVGA C)TetrG-CONH23	68.13

[a] yield percentage estimated from cyclic peptide 280 nm absorbance of HPLC chromatograms

We found that for successful cyclization, the amount of peptide being cyclized must be accurately determined. To ensure accurate determination of peptide equivalents, the resin should be washed with DCM and dried thoroughly in vacuo. This weight should be used for calculating the amount of tetrazine needed. Since in-solution tetrazine cyclization had used three equivalents of tetrazine^[34d], we initially used an excess of dichloro-s-tetrazine for on-bead cyclization. But when the cyclization of linear peptide 4, sequence GTC(Mmt)IEGC(Mmt)G, was set up on resin with excess dichloro-s-tetrazine, it was observed that two molecules of dichloro-s-tetrazine were added to peptide 4, due to an individual substitution of chloride in dichloro-s-tetrazine by each of the two Cys thiols (Figure 1A). When exactly 0.5 equivalents of tetrazine were used, and the reaction was repeated twice, a high yield of the (i, i+4) cyclic peptide 4 was obtained without forming linear peptides containing one or more appended tetrazines (Figure 1B). Similar results were also observed for the cyclization of linear peptide 5 GAC(Mmt)LTSC(Mmt)G on resin with more than 0.5 equivalents of dichloro-s-tetrazine (Figure S.4).

Figure 1. The number of tetrazine equivalents used during cyclization of the dicysteine-containing peptide on resin using the tetrazine linker dictates the linear or cyclic peptide formation.

A. Using more than 0.5 equivalent of the dichloro-s-tetrazine with the resin-bound peptide in the cyclization reaction causes the formation of a linear peptide containing two hydroxy-tetrazine moieties; B. Using 0.5 equivalents of dichloro-s-tetrazine for the resin-bound peptide in the cyclization reaction yields intramolecular monocyclic peptide as the main product. The MS shows the correct mass of monocyclic peptide.

Another critical parameter for successful cyclization was choosing the right kind of resin. Smith et al. previously reported a low yield of 10–15% of successful cyclization of peptides on-resin using dichloro-s-tetrazine^[32], using either Wang or N-methyl indole aminomethyl resin. Rink amide resin (ChemPep Inc, loading 0.3–0.6 mmol/gm, particle size 100–200 mesh, 1% crosslinking), used in table 1, is a polystyrene-based resin with a significant hydrophobic nature. To increase the percentage of cyclization, we switched to Tentagel-S-NH₂ resin (Rapp Polymere, loading 0.29 mmol/gm, 90 μm). The latter is a low crosslinked polystyrene matrix on which polyethylene glycol (PEG) spacer is attached to the matrix via an ethyl ether group which increases stability towards acid (Figure S.5). The Tentagel-S-NH₂ copolymer shows modified properties highly dominated by the PEG moiety^[38] and is much less hydrophobic than Rink amide or Wang resin. We believed that the lower hydrophobic nature of the resin would help in higher rates of cyclization. Therefore, the Tentagel-S-NH₂ resin was used for the rest of this work. When the same peptide is synthesized on Tentagel-S-NH₂ resin vs. on Rink amide resin, a much higher crude amount of peptide was obtained for the Tentagel-S-NH₂ resin.

Despite the many advantages of macrocyclic peptide libraries, intramolecular cyclization of peptides on the resin is synthetically challenging^[39]; unless the spacing between two reacting amino acids is optimal, there can be dimerized or polymeric peptides formed via intermolecular reactions^{[19, 40]}. Research from other groups suggests that cyclization efficiency is dependent on the tendency of the linear precursor peptide to form a preorganized cyclic conformation before cyclization^[41]. In a previous study, Brown et al. determined smaller ring sizes of (i, i+3) and (i, i+4) spacings between Cys residues to be optimal for tetrazine-linked peptide cyclization, as larger ring sizes resulted in higher rates of oligomerization^[32], and, in another study (i, i+2) tetrazine cyclized peptides were reported to give low yields of 7–30%^[34c]. We, therefore, chose to investigate the cyclization of (i, i+4) dicysteine-containing peptides with an s-tetrazine linker. A series of peptides, each consisting of eight amino acids, were synthesized with C-terminal Glycine and N-terminal Boc-amino acid. The C-terminal glycine was incorporated to ensure it acts as a spacer from the solid support and that the effect of the spacer on the tetrazine cyclization remains constant (Scheme 3, Table 2).

Scheme 3.

Estimation of on-bead cyclization of Cyclic (i, i+4) tetrazine-stapled peptides from HPLC chromatogram

Table 2.

Quantification of on-bead cyclization of Cyclic (i, i+4) tetrazine-stapled peptides.

Peptide	Linear peptide sequence	Percentage yield of linear peptide	Cyclic peptide sequence SI figure number	Percentage yield of cyclic peptide	Corrected percentage yield of cyclic peptide
4	GT[a]C(Mmt)IE[a]GC(Mmt)G	67.4	GT(CIEGC)TetrG6	61.6	91.4
5	GAC(Mmt)LT[a]SC(Mmt)G	54.3	GA(CLTSC)TetrG7	51.6	95.0
6	GS[a]C(Mmt)K[a]AK[a]C(Mmt)G	83.5	GS(CKAKC)TetrG8	29.6	35.4
7	GS[a]C(Mmt)GFAC(Mmt)G	67.6	GS(CGFAC) TetrG9	45.0	66.6
8	GK[a]C(Mmt)LVAC(Mmt)G	79.9	GK(CLVAC) TetrG10	44.2	55.3
9	GY[a]C(Mmt)T[a]Y[a]T[a]C(Mmt)G	54.8	GY(CTYTC) TetrG11	21.8	39.7
10	GGC(Mmt)LVH[a]C(Mmt)G	29.1	GG(CLVHC)TetrG12	28.9	99.3
11	GS[a]C(Mmt)D[a]AH[a]C(Mmt)G	77.1	GS(CDAHC) TetrG13	27.5	35.6
12	GW[a]C(Mmt)R[a]FAC(Mmt)G	43.9	GW(CRFAC) TetrG14	30.5	69.5
13	GY[a]C(Mmt)AK[a]VC(Mmt)G	67.8	GY(CAKVC) TetrG15	46.2	68.1
14	GW[a]C(Mmt)VGAC(Mmt)G	72.2	GW(CVGAC) TetrG16	59.2	82.0
15	GIC(Mmt)AW[a]AC(Mmt)G	57.4	GI(CAWAC) TetrG17	43.7	76.1
16	GE[a]C(Mmt)Y[a]N[a]S[a]C(Mmt)G	50.7	GE(CYNSC) TetrG18	29.3	57.8
17	GGC(Mmt)AY[a]VC(Mmt)G	44.1	GG(CAYVC)TetrG19	41.6	94.3
18	GK[a]C(Mmt)Y[a]K[a]Y[a]C(Mmt)G	76.2	GY(CYKYC) TetrG20	64.2	84.9
19	GA[a]C(Mmt)AK[a]GC(Mmt)G	31.0	GA(CAKGC) TetrG21	27.9	90.0

^[a]Indicated that the amino acid had a side chain protecting group (PG) present during the cyclization. For W, K, the PG was Boc; for S, T, Y, E, D, the PG was tBu; for H, N, the PG was Trt; for R, the PG was Pbf.

While any Boc-N-terminal protected amino acid can be used at the amino-terminal, in the peptide series in Table 2, we used Boc-glycine. Cys(Mmt) were incorporated in the 3^rd and 7^th position from the N-terminal, ensuring the (i, i+4) spacing. The remaining amino acids were varied randomly to explore if there was any effect of the variation on the efficacy of tetrazine cyclization. Amino acids with reactive side chains, such as His, Arg, Trp, and Lys, with bulky protecting groups Trt, Pbf, or Boc, were incorporated into the peptide series. After linear peptides were synthesized, a portion of each linear peptide was separated, the amino acid side chains deprotected by trifluoroacetic acid (TFA) treatment, and the peptide was cleaved from the resin by irradiation with 350 nm UV light^[33]. The peptides were subjected to RP-HPLC, and the main peaks were analyzed by LC-MS. The percentage yield of each linear peptide was calculated by dividing the integrated area of the peak corresponding to the correct mass by the total integrated peak area in the HPLC chromatogram^[19]. A part of the same batch of the peptide was cyclized with dichloro-s-tetrazine. After cyclization and TFA treatment, the cyclic peptides were irradiated with 350 nm UV light, analyzed by HPLC, and the main peaks from HPLC were analyzed by LC-MS. The cyclic peptide yield was calculated from the HPLC similarly to the linear peptide yield^[19] (Scheme 3). We observed, for most linear peptides, multiple major peaks corresponding to the correct mass of the peptide. Because of the high reactivity of the free thiol groups in the linear peptide, peptides show modifications such as disulfide bond formation, which elute at different time points on the HPLC. Since the amount of linear peptide formed significantly affected how much of the peptide could be cyclized later, we calculated a ‘corrected cyclic peptide yield’, which expressed the cyclic peptide yield as a percentage of the linear peptide yield. We believe this corrected value is a better indicator to estimate cyclization efficiency, as it takes into account the variability in the synthesis efficiency of different peptide sequences.

For most of the tetrazine-cyclized peptides in Table 2, the 280 nm chromatograms showed one major peak, because of the high absorbance of tetrazine at 280 nm. Multiple high peaks in the 280 nm HPLC chromatogram corresponding to hydrophobic dimeric and higher-order intermolecular cyclized peptides^[18–19] were not observed. This indicated that there were no significant levels of dimerization using S, S-dithiotetrazine as a linker for cyclization.

During on-resin tetrazine cyclization, the side chains of amino acids are protected. Thus, two categories of peptides were essentially cyclized – the first category with no side chain in the cyclic peptide and the second category with one or more side chains in the cyclic peptide. Although there was variation in cyclization yields, any systemic effect of the nature of the amino acid for the (i, i + 4) cyclization of sidechain-protected amino acid on-resin cyclization using dichloro-s-tetrazine was not observed. While peptides with no side chain protecting groups gave reasonably high yields, peptides containing one or more side chains containing amino acids also gave reasonably high yields, except for two peptides. An overall trend was a sufficiently high yield of (i, i + 4) tetrazine cyclized peptides.

A primary motivation for developing this cyclic library was that the cyclized peptide could be cleaved from resin on irradiation with UV light. The cyclic peptide was cleaved from the resin when shone with UV light at 350 nm due to the cleavage of the ANP linker, and then the cyclic peptide could be linearized on UV irradiation at 312 nm due to the cleavage of the tetrazine linker.

If one exposes a hit bead from a screen to 312 nM light, it can undergo simultaneous ring-opening and cleavage from the resin (Figure 2). The UV-cleavage method promises to be facile and environmentally friendly to sequence the peptide without treatment with harsh reagents like cyanogen bromide. Figure 2 showed that the cyclized peptide G(CVGAC)_TetzG, when irradiated with 312 nm UV light, yielded the correct linear mass of the peptide GC(SCN)VGAC(SCN)G-CONH₂. The HPLC of the UV-treated peptide showed that the dithiocyanate-containing linear peptide, with sequence GC(SCN)VGAC(SCN)G, was the main product (Figure S.22).

Figure 2. Simultaneous ring-opening and resin cleavage of tetrazine-cyclized peptide on-resin on irradiation with 312 nm UV light.

Peptide G(CVGAC)_TetzG on the resin was treated with a UV gel visualizer containing a 312 nm light lamp for 7 hours in methanol. The inset shows the MS of the cleaved peptide.

The synthesized peptides containing the tetrazine linkage in Table 1 were acetylated at the amino-terminal. However, it was observed that the acetylated peptides were hydrophobic, so they eluted at the end of the liquid chromatography gradient with other contaminants. This resulted in poor ionization and low-quality MS/MS spectra. It was also observed that some non-acetylated peptides appeared in the mass spectrometry, which was surprising. For the N-terminal acetylated peptides (Figure S.23), only a subset could be sequenced successfully. To reduce the hydrophobicity of these peptides and get better ionization, we added a Boc-glycine to the amino-terminal instead of an acetyl group. During the global deprotection with TFA to remove side chains, the Boc group also cleaved off, and therefore these peptides (Table 2, Table 3) contained glycine at the amino-terminal. The full panel of nine Boc-protected peptides was sequenced successfully by tandem mass spectrometry (Figure 3, Table 3). For MS/MS identification, it was expected that the modification of thiol to thiocyanate on the Cys (denoted as C*) would lead to the addition of 25 Da on each Cys after linearization by UV irradiation. MS2 spectra were manually annotated based on the fragments predicted in silico by the program Protein Prospector (San Francisco, CA)^[42].

Table 3.

MS/MS assignment of UV-cleaved linearized peptides

Peptide (S.I. fig)	Sequence	Monoisotopic MS1 Calculated: Observed	y ions	b ions	a ions
4*24	GTC*IEGC*G**	788.28;788.28	203.05(y2) )[a], 260.08 (y3),389.12 (y4), 502.21 (y5), 612.20(y6–18)	696.22(b7–18)[a], 568.21 (b6–18), 529.21 (b5), 511.19 (b5–18), 400.16 (b4), 287.08 (b3), 269.07 (b3–18), 159.08 (b2)	686.24 (a7), 558.23 (a6), 501.21 (a5), 372.17 (a4), 131.08 (a2)
6*	GSC*KAKC*G**	802.34; 802.34	203.06 (y2)[a],331.15 (y3), 402.19 (y4), 530.28 (y5),658.30 (y6)[a], 745.36 (y7)	728.29 (b7)[a], 600.29 (b6), 582.28 (b6–18), 472.19 (b5), 454.18 (b5–18), 401.16 (b4), 383.15 (b4–18), 273.06 (b3)[a], 255.05 (b3–18), 145.06 (b2)
7*25	GSC*GFAC*G**	750.24; 750.24	203.06 (y2)[a], 274.10 (y3), 421.16 (y4), 475.17 (y5)	548.19 (b6), 530.18 (b6–18), 477.15 (b5), 459.14 (b5–18), 330.08 (b4), 312.07 (b4–18), 273.06 (b3)[a], 255.05 (b3–18), 145.06 (b2)	520.19 (a6), 449.16 (a5)
8*26	GKC*LVAC*G**	799.37; 799.36	203.06 (y2)[a], 274.09 (y3), 373.16 (y4), 486.25 (y5),742.34 (y7)	597.31 (b6), 526.28 (b5), 427.21 (b4), 314.13 (b3)[a], 186.12 (b2)	569.32 (a6), 498.28 (a5),399.21 (a4)
9*27	GYC*TYTC*G**	916.30; 916.30	304.10 (y3), 467.17 (y4), 568.21 (y5), 696.22 (y6)[a], 859.28 (y7)	842.25 (b7)[a], 824.24 (b7–18), 714.25 (b6), 613.20 (b5), 595.19 (b5–18), 450.14 (b4), 432.13 (b4–18), 349.09 (b3)[a]	585.21 (a5)
10*28	GGC*LVHC*G**	794.31; 794.31	203.05 (y2)[a], 340.12 (y3), 439.18 (y4), 552.27 (y5)	720.27 (b7)[a], 592.26 (b6), 455.20 (b5), 356.14 (b4), 243.05 (b3)[a], 115.05 (b2)	692.27 (a7), 564.27 (a6), 427.21 (a5), 328.14 (a4)
11*29	GSC*DAHC*G**	399.62; 399.62 (diprotonated mass)	203.06 (y2)[a], 340.12 (y3), 411.15 (y4), 526.18 (y5), 654.18 (y6)*, 741.22 (y7)	724.19 (b7)[a], 578.17 (b6–18), 596.18 (b6), 370.08 (b4–18), 388.09 (b4), 255.05 (b3–18), 273.06 (b3)[a], 127.05 (b2–18), 145.06 (b2)	431.13 (a5), 245.07 (a3), 117.07 (a2)
12*30	GWC*RFAC*G**	473.6; 473.6 (diprotonated mass)	203.06 (y2)[a], 274.09 (y3), 421.16 (y4), 705.26 (y6)[a]	874.32 (b7)[a], 746.32 (b6), 675.28 (b5), 656.29 (b5–18), 528.21 (b4), 244.10 (b2)	718.32 (a6)
13*31	GYC*AKVC*G**	425.18; 425.18 (diprotonated mass)	203.06 (y2)[a], 302.13 (y3), 430.22 (y4), 501.26 (y5), 629.26 (y6)[a], 792.32 (y7)	647.30 (b6), 548.23 (b5), 420.13 (b4), 349.10 (b3)[a], 221.09 (b2)	193.10 (a2)

^[a]indicates that the ion contains a modified amino acid

Figure 3. MS/MS spectra of peptide 6* with sequence GSC*KAKC*G.

A. Deconvoluted isotopic envelope of the monoprotonated mass of peptide 6* at 3.72 min. Expected m/z: 802.34, observed m/z: 802.34. B. MS2 fragmentation of the 802.34 m/z precursor ion

As shown in Figure 3 and Table 3, we observed expected protonated intact masses for the new linearized peptide series corresponding to cyclic peptides 4 and 6-13 in Table 2. We used in silico generated fragmented mass of each peptide to interpret the experimental spectra of the UV linearized peptides, denoted as 4* and 6*−13*. We observed that the lower mass end of each series (b1, y1, and y2 ions) had low intensity or was not observed primarily due to the mass detection threshold of the instrument (Table 3). For all nine peptides, the fragmentation across the peptide backbone generated fragment ions to confirm the sequence of all the peptides with high confidence (Figure 3 and Figures S.24–S.31). We confirmed the low-intensity b-ions by the existence of a-ions and dehydrated b-ions (b-18 Da). For example, the MS2 spectrum of the peptide GSC*KAKC*G** (Figure 3) showed both strong b- and y-ion series. Dehydrated b-ions were observed more towards the higher mass end of the b-ion series.

A time course study of tetrazine cleavage was performed for peptides 4 and 8, which contained one negatively-charged and one positively-charged amino acid, respectively. Methanol solution of each peptide was purged with oxygen and irradiated in a photochemical chamber with 48 W 310 nm UV lamps for different time-periods^[32]. Since tetrazine has two absorption bands in the 350–600 nm wavelength, we monitored the UV-Visible spectra of the cyclized peptide from 350–600 nm before any UV irradiation and after irradiation for fifteen minutes, thirty minutes, and one hour (figure S.32). The bands corresponding to tetrazine vanished in 30 minutes for both peptides. Because of the elimination of nitrogen (mass difference 28) during linearization, the linearized and cyclic peptides could easily be distinguished via LC-MS. As seen in figures S.33 and S.34, when a cyclic peptide was irradiated with 310 nm UV light for fifteen minutes, there was a significant cleavage of the tetrazine linker (figure S.33A, figure S.34A); however, some percentage of cyclic peptide remained cyclized. When thirty minutes of UV irradiation was performed, under the same conditions, for peptide 4, the cleavage of the tetrazine was almost complete (Figure S.33D). But under the exact condition, peptide 8 still had insignificant amounts of cyclic peptide left (Figure S.34D). In either case, there was enough linearized peptide formed that the peptide sequences could be unambiguously assigned by de novo sequencing (Figure S.24, Figure S.26).

Conclusion

We designed (i, i+4) stapled cyclic peptides that can be linearized and cleaved from the resin by UV light treatment and then sequenced by MS/MS. The significance of this approach is that linearization of the peptide before sequencing sidesteps the cyclic MS/MS spectra deconvolution and does not require any truncation of the library or use of two compounds on the resin bead, which makes it distinct from the currently used methods. This peptide design also lowers the margin for error, as the peptide itself is sequenced rather than an analyzable counterpart anchored to the resin^{[14, 27]}. The accessible and robust Fmoc-based SPPS peptide synthesis followed by an optimized on-bead cyclization with high monomer yield represents a significant advance in developing MS/MS compatible cyclic peptide libraries, eliminating additional steps typically needed for the synthesis of a cyclic peptide library and its unambiguous sequencing.

Notably, we found this cyclization and linearization approach to be compatible with various amino acids and protecting groups commonly utilized in Fmoc SPPS. We found that the yield of the (i, i+4) cyclized peptides varied with sequence, but there was no trend in variation based on the nature of the amino acids. A reasonable cyclization yield of 50% or more was observed for twelve of the fifteen random peptide sequences, without significant dimerization.

The UV-induced linearization and cleavage strategy was critical in successfully sequencing the peptides by tandem mass spectrometry. We initially synthesized amino-terminal acetylated peptides as shown in table 1, but the high hydrophobicity and poor MS/MS results of the acetylated peptides led us to abandon the acetylation step. After that, peptides were synthesized with a free amino-terminal, as shown in table 2, and used for quantification of cyclization and MS/MS analysis. The mass spectra generated by the fragmentation of linear peptides were significantly less complicated than cyclic peptide MS/MS spectra.

An advantage of the developed peptide design is that the tetrazine linker can participate in the IEDDA reaction with target proteins with appropriate reacting partners such as an alkyne or alkene^[43]. Therefore, the covalent bond between the peptide and protein can help screen a protein target. The nature of the resultant OBOC library would ensure that even with the protein forming a covalent bond, there would be sufficient copies of the tetrazine-containing peptide that could then be irradiated by UV light and sequenced by MS/MS. While tetrazine reaction with trans-cyclooctyne has been used extensively for live cell imaging^[44], tumor imaging^[45], targeting molecular functionalities such as RNA^[46], and for biological tagging reactions^[43], very little work has been done with S, S-disubstituted tetrazine compounds, and the reported research creates opportunities for doing IEDDA reactions of alkenes or alkynes with the tetrazine-cyclized peptides to make new biological and biomimetic constructs.

Experimental Section

1. Solid Phase Peptide Synthesis.

Initiator+ (Biotage) was used to synthesize the peptides along with dimethylformamide (DMF), 0.5 M N, N′-Diisopropylcarbodiimide (DIC), 0.5 M Oxyma, and 5.0 equivalents of amino acids. Fmoc-protected amino acids were purchased from ChemImpex and ChemPep, while the other reagents and solvents were purchased from VWR. The SPE tube containing the Rink amide resin or the TentaGel-S-NH2 with the ANP linker already coupled was placed in the reactor chamber, where all reactions occurred. The resin was swelled in DMF at 70°C for 20 minutes while mixing. Fmoc deprotection was performed using 20% piperidine in dimethyl formamide (DMF) at room temperature for 3 minutes, and amino acid coupling for most amino acids took place at 75°C for 5 minutes. In contrast, the Fmoc-Cys(Mmt)-OH amino acid coupling took place at 50°C for 5 minutes. Peptides were also synthesized manually following the Solid Phase Peptide Synthesis (SPPS) Fmoc-based standard protocol. Resins (Rink amide or TentaGel-S-NH2) were swelled overnight at room temperature, followed by Fmoc deprotection using 20% piperidine in N-Methyl-2-pyrrolidone (NMP) 3× 10 minutes. Amino acid coupling was done using three equivalents of amino acid in NMP for 3–4 hours, while ANP (2 eq) linker was coupled overnight.

1.1. N-terminal protection.

The N-terminal of each peptide was protected by adding an acetyl group or coupling of Boc-Gly-OH to prevent side reactions. To the solid phase extraction (SPE) tube containing the peptide, 4ml of an acetylation solution containing 10% acetic anhydride, 12% 2,6-lutidine, and 78% DMF was added, and the peptide was shaken for 10 minutes, then the solution was drained. This process was repeated two more times, and lastly, the peptide was washed using DMF and dried using dichloromethane (DCM). Boc-Gly-OH was coupled manually following the standard SPPS protocol using three equivalents of Boc-Gly-OH, 2.98 equivalents of Hexafluorophosphate Azabenzotriazole Tetramethyl Uronium (HATU), and nine eq of full name (DIEA) in NMP overnight.

1.2. Mmt deprotection.

The 10mL deprotection solution was prepared using the following ratios: 3% trifluoroacetic acid (TFA), 8% triethylsilane (TES), and 89% DCM. Approximately 3 mL of the solution was added to the SPE tube containing the peptide and placed on the rotor for 10 minutes. After 10 minutes, the solution was drained and replaced with a new 3mL solution repeating the procedure twice, followed by 3 DMF washes.

1.3. Acm deprotection.

A 2M solution of mercury acetate in DMF was used to remove the Acm protecting group from Cys. The dried peptide was transferred to a round bottom flask covered with aluminum foil, and 4ml of the 2M solution of mercury acetate in DMF was added and stirred overnight. The next day, the peptide was transferred back to the SPE tube, the mercury acetate solution was drained, and the peptide was washed with DMF.

1.4. Thiol Reduction with B-mercaptoethanol.

Following Cys deprotection, the peptides were treated with a solution containing 20% β- mercaptoethanol, 1% DIEA, and 79% DMF to prevent disulfide bridge formation. The reduction solution was added to the SPE tube containing the peptide, and the tube was shaken for 1 hour. The solution was drained and repeated with fresh reduction solution, followed by DMF wash.

1.5. Peptide Cyclization.

Peptide cyclization was performed using 0.5 equivalents 3,6- dichloro-1,2,4,5-tetrazine and 2 equivalents 2,6-lutidine in 1 mL DMF in a round bottom flask. This protocol was repeated twice: first for 2 hours and again for 3.5 hours, draining and replacing the cyclization solution. In addition, the reaction was purged with argon atmosphere for the first 15 min of each reaction time.

1.6. TFA cleavage and side chain deprotection.

To remove any acid-labile side chain protecting groups, a TFA solution containing 88% TFA, 5% Phenol, 5% distilled water, and 2% TES was used. To a glass vial, dried peptide and 2 ml of cleavage solution were added, covered with aluminum foil, and stirred at room temperature for two hours. For peptides on Rink amide resin, this TFA solution removed the acid-labile side chain protecting groups and cleaved the peptide off the bead. However, for peptides on Tentagel-S-NH2 resin, this solution only removed the acid-labile side chain protecting groups.

For peptides on Rink amide resin, after two hours of treatment with the TFA cleavage cocktail, the solution was filtered using an acid-resistant disposable glass pipette containing glass wool into cold 45ml ether. After precipitation, the tube was vortexed and chilled in the −20°C freezer until white peptide precipitation formed. The ether precipitate was centrifuged for 20 minutes at 4000 RPM, then decanted, and the resulting peptide pellet was solubilized in a mixture of 50:50 acetonitrile and distilled water, followed by overnight lyophilization.

For peptides on Tentagel-S-NH2 resin, after the two-hour-long TFA treatment, the peptide solution was transferred back to the SPE tube, washed with water, followed by methanol, and dried with DCM.

2. 350 nm UV cleavage.

Peptides containing the ANP linker on TentaGel-S-NH2 resin were cleaved off the bead using a 350 nm photoreactor. After the TFA side chain deprotection, the dried peptide was added to a microcentrifuge tube with 1 ml of methanol and placed in Fotodyne FOTO/Phoresis UV Transilluminator for 6 hours. Subsequently, the microcentrifuge tube was briefly centrifuged, then the methanol solution was lyophilized overnight.

3. 312. nm UV cleavage.

The same procedure described in section 4 was used to linearize the peptide, except with a 312 nm FOTOMax UV transilluminator. This was a UV gel reading apparatus that had a low effective wattage and therefore needed a long irradiation time of 6 hours.

Later, a switch was made to a Rayonet photochemical chamber containing 48 watts of 310 nm lamps, the exposure time was reduced to 30 minutes, similar to the reported conditions^[32].

4. HPLC.

Peptides were analyzed using HPLC Agilent 1260 series and a 150mm Agilent Pursuit XRs analytical column, using a gradient of acetonitrile and double distilled water with 0.1% trifluoroacetic acid.

5. LC-MS.

HPLC eluted samples were further analyzed using LCMS Agilent 6120 Quadrupole, positive mode electrospray ionization (ESI), and LC Agilent 1260 Infinity 100mm Agilent Eclipse Plus column.

6. Mass Spectrometry for sequencing.

The linearized, lyophilized peptides were desalted by C18 spin columns (Thermo Fisher Scientific). If the expected mass was not observed, peptides were further purified by RP-HPLC. The mass and sequence of purified and desalted peptides were determined using LTQ orbitrap velos or Tribred Lumos mass spectrometer (Thermo Fisher Scientific). For the time-course study of tetrazine cleavage, the peptide solutions in methanol exposed to 310 nm UV light for different time-periods were directly analyzed by the Tribred Lumos mass spectrometer (Thermo Fisher Scientific). The sequence of the peptides was ascertained by manually annotating the spectrum. For manual annotation, in silco fragments of the linearized peptides were generated using the Protein Prospector MS-Product program and compared with the experimental spectra. The N-terminal acetylation, thiocyanate modification on cysteines, and C-terminal amidation as fixed modifications on linearized peptides were included in the generation of in silico fragment ions.