A Method for Rigorously Selective Capture and Simultaneous Fluorescent Labeling of N‑Terminal Glycine Peptides

Abstract

Although chemical methods for the selective derivatization of amino acid (AA) side chains in peptides and proteins are available, selective N-terminal labeling is challenging, especially for glycine, which has no side chain at the α-carbon position. We report here a double activation at glycine’s α-methylene group that allows this AA to be differentiated from the other 19 AAs. A condensation reaction of dibenzoylmethane with glycine results in the formation of an imine, and subsequent tautomerization is followed by intramolecular cyclization, leading to the formation of a fluorescent pyrrole ring. Additionally, the approach exhibits compatibility with AAs possessing reactive side chains. Further, the method allows for selective pull-down assays of N-terminal glycine peptides from mixtures without prior knowledge of the N-terminal peptide distribution.

Methods for labeling proteins or peptides at amino acid (AA) side chains,¹⁻⁴ as well as their N- or C-termini,⁵⁻⁸ are currently proliferating for the development of bioconjugates for use in therapeutics,^9,10 diagnostics,¹¹⁻¹³ and proteomics.^14,15 However, the amines of the 20 possible N-terminal peptides/proteins remain one of the most difficult functional groups to differentiate due to their similarity.^16,17 One successful example is the reaction between N-terminal cysteine and aldehydes, which incorporates the side-chain thiol in a thiazolidine ring.^18,19

One may anticipate that the labeling of glycine would be particularly challenging due to the lack of a side chain. However, when on the N-terminus, this AA has been identified as an important target. For example, the labeling of N-terminal glycine was used as a method to identify inhibitors of active and inactive kinases.²⁰ N-terminal glycine has also been identified as a target for proteasomal degradation, allowing for clearance of proteolytic fragments generated by caspase cleavage during apoptosis.²¹ Further, the targeting of N-terminal Gly-Gly residues using asparaginyl peptidase has led to designed protein−protein conjugates.²² An N-terminal GlyGly sequence is generated upon the treatment of ubiquitinated proteins with trypsin, allowing identification of the site of this posttranslational modification (PTM).²³ Apart from protein/peptide labeling, glycine is also used for modification on polymers.^24,25 The specific differentiation of glycine over other AAs would lead to distinct properties in their assemblies and biomedical functions.²⁶⁻²⁸ Therefore, developing selective methods for N-terminal glycine labeling is important for the functionalization of proteins, peptides, and glycine-modified materials.

Over the past decade, approaches have been developed for the labeling of glycine at the α-carbon.²⁹⁻³² These methods require the prefunctionalization of the amino groups which are not specific to glycine. However, in 2019, Rai reported an N-terminal glycine labeling method (Figure 1a).³³ This strategy involves a condensation of the N-terminal amine with a carboxyl-group-functionalized benzaldehyde, ultimately generating a vicinal amino alcohol via an aldol-like reaction involving an enol intermediate (Figure 1a). The approach was found to be inefficient with AAs other than glycine, postulated to be due to steric effects. The aldehyde reagent is created in multiple steps and used in excess, but the glycine amino group remains after labeling for further conjugation if desired. In 2024, Kanemoto reported an approach for tagging N-terminal glycine via a three-component [3,2]-cycloaddition with aldehydes and maleimides with Cu(MeCN)₄PF₆ catalysis (Figure 1b).³⁴ To achieve this selectivity (again based on steric size), the method uses a specific stoichiometry of base (Et₃N) and catalyst relative to peptide. Higher base concentration and catalyst loading, and sometimes different temperatures, lead to labeling other N-termini. Thus, the approach requires prior knowledge of the concentration of N-terminal glycine peptides in a mixture to achieve selectivity. However, the method is high yielding and particularly useful for broadly labeling all N-terminal amines over the ε-amine of lysine. To incorporate a fluorescent label, both the Rai and Kanemoto methods require appendage of the fluorophore to the conjugating agents.

Figure 1.

(a) Rai method.³³ (b) Kanemoto method.³⁴ (c) Glycine labeling via methylene “double activation” (this work).

We envisioned a different strategy to label an N-terminal glycine, one that would perform “double activation” on the α-methylene group, thereby not relying on a steric difference with the other 19 AAs’ methine groups, thus rigorously limiting the procedure to N-terminal glycine. Herein, we describe such an activation that involves a single step using a commercial reagent and imparting fluorescence without the use of an exogenously appended fluorophore (Figure 1c). Further, we demonstrate compatibility with AAs containing reactive side chains. Notably, our method selectively labels N-terminal glycine peptides in a mixture without any knowledge of the extent of such peptides, showcasing its utility in pull-down assays.

Our approach started with a mechanistic postulate of how to achieve “double activation”. We postulated that the reaction of a β-diketone with glycine would commence with a condensation to form an imine (Step A, Scheme 1), followed by enol tautomerization of the amide (Step B) (precedented from the Rai study). The enol could undergo a 5-exo-trig cyclization (Step C, activation 1). Elimination of water (Step D, activation 2) and tautomerization (Step E) would yield a pyrrole. Thus, the method is similar to the Knorr pyrrole synthesis.³⁵⁻³⁹ Further, because 2,5-diphenyl pyrroles are known fluorophores,⁴⁰⁻⁴² our diketone of choice was dibenzoylmethane (DBM-1), potentially offering an in situ way to place a fluorescent moiety on an N-terminal glycine.

Scheme 1.

Mechanism for the Formation of Pyrrole via the Condensation of N-Terminal Glycine and 1,3-Diketones

To validate our mechanistic postulate, we conducted a model reaction between glycine-N-methylamide hydrochloride (GMA) and DBM-1 using a variety of conditions, settling on DMF as solvent at 130 °C in a microwave reactor for 5 h (Figure 2a). Using optimal conditions, we isolated the desired molecule PY in 21% yield, which was characterized using NMR spectroscopy, high-resolution mass spectrometry (HRMS) (Figures S27, S28, and S31), and single-crystal X-ray diffraction (XRD) (Figure 2b). Further, as expected, PY showed a blue emission with a maximum wavelength of 367 nm (Figure 2c).

Figure 2.

(a) Reaction scheme. (b) Single-crystal structure of PY. (c) Absorbance spectrum of PY (30 μM in DMSO). (d) Fluorescence spectrum (excitation: 312 nm) of PY, and image of the solution under 254 nm UV irradiation.

According to the central postulate of Scheme 1, the derivatization was designed to be specific for glycine, as only this AA could lead to the formation of a pyrrole. To test this postulate, we performed a control experiment using alanine-N-methylamide hydrochloride and DBM-1. As shown in Figure S1, although the ion peaks of the corresponding imine were detected in mass spectra, no peak corresponding to the possibility of a water elimination analogous to Step D to generate a 2H-pyrrole (a 5-membered ring containing an α,β-unsaturated imine) was observed, validating the selectivity of ring formation for glycine.

Having success with the simple model glycine GMA, albeit in moderate yield, we moved to our goal of selectively labeling N-terminal glycine peptides. At first, we synthesized H₂N-GGAAAA as a model peptide via Solid Phase Peptide Synthesis (SPPS) with Rink Amide resin as the solid support. Delightfully, when H₂N-GGAAAA was allowed to react with DBM-1, the desired pyrrole Py-(G)GAAAA (Py stands for N-terminal pyrrole) was observed, and it was characterized using NMR spectroscopy and HRMS (Figures S2−S4). The product of the reaction is UV active and thus is easily detectable in LC-MS traces, but as the peptide itself (H₂N-GGAAAA) does not have any chromophore, this posed challenges in monitoring side products via LC-MS. To address this problem, we incorporated tyrosine as a chromophore and chose H₂N-GAAYAA as a second test peptide.

Again, when GAAYAA was allowed to react with DBM-1, the desired pyrrole Py-(G)AAYAA was observed (Figures 3a and S5; Table 1, entry 1). Due to the fluorescent nature of the small molecule pyrrole-based product (PY), we proceeded to conduct photophysical studies on the labeled peptides. As shown in Figure 3, Py-(G)AAYAA has a broader absorbance spectrum than the unlabeled peptide (H₂N-GAAYAA) and exhibits blue fluorescence with a maximum emission at 382 nm.

Figure 3.

(a) Reaction of H₂N-GAAYAA and DBM-1. (b) Absorbance spectrum of H₂N-GAAYAA. (c) Absorbance spectrum of Py-(G)AAYAA. (d) Emission spectrum of H₂N-GAAYAA (excitation: 275 nm) and Py-(G)AAYAA (excitation: 315 nm), all 30 μM in DMSO. (e) Image of solutions of H₂N-GAAYAA and Py-(G)AAYAA under 254 nm UV irradiation.

Table 1.

Percentage Conversion of Pyrrole Products Generated on Various Peptides

Entry	Peptide	Conversionb (%)
1	Py-(G)AAYAA	63
2	Py-(G)AKYAA	62
3	Py-(G)ASYAA	53
4	Py-(G)AQYAA	64
5	Py-(G)AEYAA	26
6	Py-(G)ADYAA	55
7	Py-(G)AMYAA	76
8	Py-(G)ACYAAa	30
9	Py-(G)YFAVY	74
10	Py-(G)WFAVY	58
11	Py-(G)SYWVY	45

^aCysteine was capped with iodoacetamide

^bConversions were calculated from UPLC data.

We next set out to explore the compatibility of the reaction of DBM-1 and N-terminal glycine peptides containing potentially reactive side chains such as lysine, serine, glutamine, glutamic acid, aspartic acid, methionine, and cysteine. Except cysteine, all other AAs are compatible, and percent conversion to the desired pyrrole products is given in Table 1 (entries 2−7) (Figures S6−S11). However, if we capped the cysteine with iodoacetamide, then the desired pyrrole product was observed (Table 1, entry 8; Figure S12). Further, the incorporation of lysine (H₂N-GAKYAA) does not affect the N-terminal labeling, given that it possesses a free amino group. However, the imine-based byproduct Py-(G)A(imine)KYAA was observed, but it could be converted to the desired product Py-(G)AKYAA after treatment in acidic media via hydrolysis (Figure S6b and S6c).

Next, we examined steric effects by placing bulky AAs, such as tyrosine and tryptophan, next to glycine. We were pleased to observe that the desired products formed in comparable yields (Table 1, entries 9 and 10; Figures S13 and S14). Additionally, to test the tolerance of this approach, we simultaneously incorporated multiple reactive AAs, including serine, tyrosine, and tryptophan, in the same peptide. Again, the approach was successful, yielding a 45% conversion to the desired product (Table 1, entry 11; Figure S15).

Regarding the specificity of labeling the glycine model compound GMA, we were interested in whether this specificity would be retained for peptide labeling. Thus, lysine- and valine-terminated peptides H₂N-KAAYAA and H₂N-VAAYAA, respectively, were allowed to react with DBM-1. However, only imine-based byproducts were detected by LC-MS for both peptides (Figures S16 and S17), which could be converted to reactants under acidic conditions. These results further validated the rigorous selectivity of our strategy.

The differentiation of N-terminal glycine peptides could be useful for the enrichment of such peptides from complex systems. Thus, we conducted a pull-down assay aimed at capturing these peptides from mixtures (Figure 4). To do so, we modified DBM-1 by introducing a TMS-protected single alkyne on one phenyl ring para to a ketone (DBM-2) (Figures S29, S30, and S32). Subsequently, we added DBM-2 to a peptide mixture having different AAs at the N-terminal (G, F, V, W, L, and Y) in a large enough excess to label all N-termini and performed the reaction under the optimal conditions given above. This process led to the successful labeling of only peptides containing N-terminal glycine (two regioisomers exist), while the N-termini of all other peptides formed an imine with DBM-2 (Figures S18−S25). The imine was hydrolyzed by treating the mixture with acidic water, followed by ether precipitation to remove DBM-2.

Figure 4.

Schematic diagram showing selective labeling and capturing of N-terminal glycine in the mixture of peptides using DBM-2.

To capture the labeled peptides, we used resin beads functionalized with azide groups, allowing them to undergo CuAAC chemistry with an alkyne handle. The resin was washed, and the labeled peptides were cleaved from the resin. Only the glycine-labeled product was observed (Figure S26), thereby highlighting the ability to selectively capture the target peptides without prior knowledge of the amount of N-terminal glycine peptides present.

In summary, we have developed a method for the selective labeling of N-terminal glycine peptides using a 1,3-diphenyl-1,3-diketone, affording fluorescent pyrrole-based products. By virtue of the “double activation” mechanism, it is rigorously selective for N-terminal glycine. The approach was successfully applied to the fluorescent labeling of several peptides, is compatible with potentially reactive amino acids, and can be used in pull-down assays. We are currently exploring the use of this chemistry to monitor the sites of ubiquitination due to the generation of Gly-Gly residues subsequent to trypsin cleavage.⁴³

Now that the chemical community has three different choices for labeling N-terminal glycine, it is worthwhile to compare each. The Rai and Kanemoto methods rely on the steric differentiation of glycine over the other AAs. The Rai method requires an excess of reagents, but it leaves the N-terminal free for further conjugation if desired. The Kanemoto method is the highest yielding and mildest of the three methods, but it requires precise stoichiometry of reagents. Each would require the appendage of a fluorophore for optical labeling. Our method has lower conversion (yet, typically over 50%) and requires heating, but it is rigorously selective and therefore can be used in mixtures and for pull-down assays, and it necessarily incorporates an optical label. Thus, each method has strengths and weaknesses that an investigator should consider when adapting the methods to their needs.