Dinitroimidazoles as bifunctional bioconjugation reagents for protein functionalization and peptide macrocyclization


1,4-DNIms are rapid and Cys-specific bioconjugation reagents

To initiate our investigation, we synthesized compound 1a and sought to evaluate its potential for cysteine bioconjugation (Fig. 2a). Incubation of compound 1a (1.0 mM) with N-acetyl-l-cysteine (1.0 mM) in 100 mM HEPES buffer at pH 8.0 leads to rapid and quantitative conversion to a cysteine adduct 1b in 15 s (Fig. 2a). Structural characterization of adduct 1b by nuclear magnetic resonance (NMR) and X-ray crystallographic analysis indicate that the cysteine thiol is directly conjugated to C-5 of the nitroimidazole (Fig. 2a, Supplementary Fig. 1). This result suggests that the reaction proceeds through a cine-substitution mechanism with N1-nitro as the leaving group upon nucleophilic addition of thiol to C-5 of compound 1a, which is followed by re-aromatization (Fig. 2a, Supplementary Fig. 2). With C-5 hindered by methyl substitution, 5-methyl-1,4-nitroimidazole 2c was unreactive with cysteine and remained intact under assay conditions, supporting this reaction mechanism (Supplementary Fig. 3).

Fig. 2
Fig. 2

Compound 1a modifies cysteine thiol with high efficiency and specificity by generating stable conjugates. a Reaction of compound 1a with cysteine and tripeptide 2a. The CCDC reference number of compound 1b is 1860451. b Reaction kinetics of compound 1a (0.2 mM) with tripeptide 2a (0.1 mM) at pH 7.0 and 8.0. c The reaction between of 1a (0.2 mM) and 2a (0.1 mM) under various pH for 2 min and 30 min. d Reactions of compound 1a (0.2 mM) with nucleophilic amino acids (0.1 mM) for 15 min under pH 7.0 and 8.0. e Chemical stability of the tripeptide-(4-nitroimidazole) conjugate 1c’ at pH 2.0, pH 10.0, and under oxidative conditions (10 mM H2O2 in Na2CO3 buffer at pH 8.0) at 37 °C for 10 h. The conversion values represent means ± SDs (standard deviations) of three independent experiments

The high reactivity of compound 1a prompted us to investigate its ability to modify cysteine residues in peptides. Tripeptide 2a was therefore synthesized and treated with compound 1a in HEPES buffer at pH 7.0. High-performance liquid chromatography (HPLC) analysis showed that 92% conversion of peptide 2a was achieved within 1 min, and full conversion was achieved after 5 min by generating peptide conjugate 1c (Fig. 2b, Supplementary Fig. 4). The reaction was significantly accelerated by elevating the pH of the buffer to 8.0, leading to full conversion in only 15 s (Fig. 2b, Supplementary Fig. 5). We next evaluated the pH dependence of this reaction. After incubating 1,4-DNIm 1a with tripeptide 2a at pH 7.0–9.0 for 2 min, near-quantitative modification was observed (Fig. 2c, Supplementary Figs. 46). Prolonging the incubation time to 30 min led to 28% and 74% conversion of tripeptide 2a even under acidic conditions at pH 4.0 and 5.0, respectively, demonstrating the wide applicable pH range of this reaction (Fig. 2c, Supplementary Figs. 710).

Next, we evaluated the chemoselectivity of compound 1a under aqueous conditions. Lysines are rich in natural proteins and usually solvent-exposed on protein surface, therefore representing a major competition. In a competitive assay, incubation of tripeptide 2a (0.1 mM) and a lysine-containing tripeptide 3a (1.0 mM) with compound 1a (0.12 mM) in HEPES buffer at pH 7.0 yielded conjugate 1c near quantitatively (Supplementary Fig. 11). No lysine adduct of tripeptide 3a was detected by liquid chromatography–mass spectrometry (LC-MS) analysis, demonstrating the dominating selectivity of 1,4-DNIms towards cysteine thiol over lysine ε-amine. Furthermore, compound 1a did not react with nucleophilic amino acids, including Tyr, Ser, Asp, Arg, Met, and His, under neutral aqueous conditions, further establishing the thiol specificity of 1,4-DNIms (Fig. 2d). Perfluoroaryl compounds have shown great potential in cysteine bioconjugation of peptides13. To compare their efficiency with 1,4-DNIms, we chose 6,6′-sulfonylbis(1,2,3,4,5-pentafluorobenzene) (pf1), the most reactive perfluoroaryl reagent reported so far30, as the model compound. First, we observed that compound pf1 exhibited a very low solubility (<0.05 mM) in aqueous solutions, whereas 1,4-DNIm 1a has a solubility higher than 10 mM. In HEPES buffer, pH 7.4, (pf1) at a concentration of 0.04 mM was unreactive with peptide 2a, whereas 1,4-DNIm 1a modified (2a) near quantitatively (Supplementary Fig. 12). This result shows that under neutral aqueous conditions, 1,4-DNIms are superior to perfluoroaryl compounds in terms of solubility and reactivity towards cysteine residues. Together, under acidic to neutral aqueous conditions, 1,4-DNIms are highly efficient and cysteine-specific bioconjugation reagents.

The stability of protein conjugates is a major concern for bioconjugation reagents. Therefore, we evaluated the stability of 1,4-DNIm 1a and the resulting (4-nitroimidazole)-thiol products. HPLC analysis showed that N-phenyl maleimide underwent complete hydrolysis in 2 h in 100 mM phosphate-buffered saline (PBS) buffer at pH 7.4, yielding the maleamic acid (Supplementary Fig. 13). In contrast, compound 1a displayed resistance to hydrolysis, remaining unchanged for up to 10 h (Supplementary Fig. 14). Next, we investigated the stability of the (4-nitroimidazole)-thiol linkage. Thiol-maleimide and thiol-arylation conjugates are reported to degrade under acidic, basic, or oxidative conditions25,30,31. We incubated conjugate 1c’ in aqueous buffer at pH 2.0, pH 10.0, or in the presence of 10 mM H2O2 for 10 h at 37 °C. HPLC analysis indicated that conjugate 1c’ did not decompose under tested conditions (Fig. 2e, Supplementary Fig. 15), whereas the N-ethyl maleimide conjugate of 2a′ underwent complete hydrolysis at pH 10.0 (Supplementary Fig. 16). Together, both 1,4-DNIm 1a and its cysteine conjugate exhibited enhanced stability compared with maleimides and maleimide-cysteine conjugates under various conditions, offering an attractive alternative for cysteine bioconjugation of proteins.

Protein conjugates bearing different classes of functionalities, such as fluorophores and biomolecules, are valuable in the fields of biomaterials and chemical biology. To test the applicability of 1,4-DNIms in protein modification and their compatibility with diverse functional components, we synthesized fluorescent 7-nitrobenzo-Z-oxa-1,3 diazole (NBD)-conjugate 4a and peptide arginine-glycine-aspartic acid (RGD)-conjugate 4b bearing a 1,4-DNIm warhead for cysteine bioconjugation (Fig. 3a). Sortase A (SrtA) with one free cysteine (Cys192) and 26 lysines was selected as a model protein. It was expected that 1,4-DNIms could achieve cysteine-specific labeling by overcoming the challenge of a large number of surface-exposed Lys under neutral aqueous conditions. Incubation of SrtA (20 µM) with compound 4a (200 µM) for 1 h in PBS buffer, pH 7.0, resulted in quantitative conversion to NBD-modified SrtA, as determined by LC-MS analysis (Fig. 3c). The NBD-modified SrtA was further digested by trypsin and subjected to tandem mass spectrometry (MS) analysis. Results showed that the SrtA186–205 segment carried the NBD modification at the Cys192 residue (Supplementary Fig. 17). To further confirm the Cys-specific labeling, SrtA was pre-treated with iodoacetamide (IAA) before incubation with NBD-(1,4-DNIm) 4a. No fluorescent labeling was observed for the IAA-treated SrtA, further indicating that 1,4-DNIms selectively modified cysteine residues (Fig. 3d). RGD peptides bind to the integrin receptors of cells and are widely used for targeted delivery of chemotherapeutics32,33. Compound 4b with a 1,4-DNIm moiety at the N terminus of RGD sequence was then incubated with SrtA under neutral buffer. To our delight, full conversion to RGD-SrtA conjugate was observed after 1 h (Fig. 3c). These results demonstrate that 1,4-DNIms are compatible with diverse functional components during protein modification.

Fig. 3
Fig. 3

Modification of SrtA with 1,4-DNIms bearing a fluorescent NBD moiety and an RGD peptide. a Chemical structures of 1,4-DNIm derivatives 4a and 4b. b Bioconjugation of SrtA by 1,4-DNIm derivatives. Conditions: SrtA (20 µM), compound 4a or 4b (200 µM), 100 mM PBS buffer, pH 7.0, 1 h. c LC-MS analysis of SrtA modified by compounds 4a and 4b. Molecular mass (average): SrtA calcd. 24,136.03 Da, observed 24,135.83 Da; (SrtA-4a) conjugate calcd. 24,596.44 Da, observed 24,596.77 Da; (SrtA-4b) conjugate calcd. 24,866.77 Da, observed 24,866.70 Da. d SDS-PAGE analysis of the fluorescent labeling of SrtA or iodoacetamide-treated SrtA by compound 4a

To make 1,4-DNIms as a general approach to prepare diverse protein conjugates, we synthesized compound 4c bearing an alkyne group for further functionalization through 1,3 dipolar azide–alkyne cycloaddition (AAC) (Fig. 4a). Bovine serum albumin (BSA), with a single free cysteine (Cys34) and 59 lysine residues, was chosen as a model protein. Treatment of BSA (100 µM) with compound 4c (500 µM) led to quantitative modification in 1 min, as determined by LC-MS analysis (Fig. 4b). To identify the modification site, the resulting BSA-4c conjugate was reduced, digested by trypsin, and subjected to LC-MS/MS analysis. Results showed that the modification occurs exclusively at the Cys34 residue (Fig. 4d), and no lysine modification is detected. Subsequently, fluorescent labeling of the BSA-4c conjugate was realized with a fluorogenic azide, sulfo-Cy3-azide, through CuAAC reaction. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis showed that BSA was successfully labeled with fluorophore Cy3 through the alkyne handle of compound 4c as control reactions without copper or Cy3-azide failed to produce any observable fluorescence (Fig. 4c). Therefore, the successful incorporation of an alkyne group as the click reaction handle through 1,4-DNIm bioconjugation provides a facile method for cysteine-selective addition of various payloads to proteins of interest.

Fig. 4
Fig. 4

Thiol-specific labeling of BSA protein and peptide stapling by 1,4-DNIm derivatives. a Bioconjugation of derivative 4c with BSA. The free thiol of Cys34 residue is highlighted in yellow. b ESI-MS spectra of BSA protein before and after treatment with compound 4c. Molecular mass (average): BSA calcd. 66,429.98 Da, observed 66,430.14 Da; BSA-4c calcd. 66,609.12 Da, observed 66,609.03 Da. c Gel image of the reaction between BSA-4c conjugate and sulfo-Cy3-azide by a CuAAC reaction. d LC-MS/MS analysis of a modified peptide fragment from trypsin digestion of BSA-4c conjugate. e Macrocyclization of cysteine-containing unprotected peptides by a bisfunctional 1,4-DNIm (5a). Conditions: peptides (1.0 mM) and compound 5a (1.0 mM) and sodium ascorbate (2.0 mM) were incubated in 100 mM HEPES buffer at pH 7.0 for 10 min before acidified with formic acid and subjected to HPLC analysis

The potential toxicity of the thiol-(4-nitroimidazole) linkage generated by bioconjugation is an important concern for the biological application of the 1,4-DNIm chemistry. To address this issue, we synthesized a fluorescein isothiocyanate (FITC)-labeled RGD peptide and an (1,4-DNIm)-modified cyclic RGD (cRGD) peptide (Supplementary Fig. 18). At a concentration of 10 µM, RGD peptides were efficiently uptaken by 293T cells, as demonstrated by the cell staining assay of FITC-RGD. Upon incubation with 10 µM cRGD-(4-nitroimidazole) conjugate, the viability of 293T cells was not affected (Supplementary Fig. 18), indicating that the thiol-(4-nitroimidazole) is not toxic under assay conditions.

Cys-specific peptide stapling by bisfunctional 1,4-DNIms

Cyclic peptides have great potential as therapeutic agents and biological tools, and thus the development of stapling reagents is of particular interest13,18,34,35,36,37. The capability of 1,4-DNIms in selective cysteine modification prompted us to explore their application in peptide stapling. Thus, a bisfunctional 1,4-DNIm (5a) was synthesized and incubated with unprotected peptides containing two cysteine residues at varying sites under neutral aqueous buffer for 10 min. Sodium ascorbate was added to the reaction to avoid disulfide formation caused by air oxidation. High conversions were observed for all peptides examined by yielding corresponding peptide macrocycles, as determined by HPLC analysis (Fig. 4e). In particular, Cys-specific crosslinking of peptide P3 with unprotected N terminus was achieved, demonstrating the chemoselectivity of 1,4-DNIms towards thiol groups over N-terminal amine.

1,4-DNIms modify lysine amine in organic solvents

Although lysines and N-terminal amine groups in proteins are inert to 1,4-DNIm treatment under neutral aqueous conditions, reactions between amines and 1,4-DNIms have been reported in the synthesis of nitroimidazole derivatives in organic solvents. Early evidence and recent computational studies suggest that the reaction is initiated by nucleophilic attack at C5 position of 1,4-DNIms and follows an ANRORC-like mechanism, consisting of the addition of a nucleophile followed by ring-opening and ring closure step (Fig. 1b)38,39,40. To provide mechanistic insight of this reaction, we utilized 15N-labeled aniline to react with 1,4-DNIm (1a), generating product 1e in 82% (Fig. 5a). High-resolution MS (HRMS) analysis showed that (1e) is isotopically labeled, indicating that the 15N atom from aniline is present in the product (Fig. 5a). NMR characterization of (1e) by analyzing 15N-C and 15N-H coupling signals further indicated that the 15N of analine was indeed incorporated into the newly formed imidazole ring by forming two C-15N bonds with C2 and C5 of (1a) (Supplementary Figs. 1924). Furthermore, X-ray crystallographic analysis unambiguously confirmed the structure of (1e) (Fig. 5a).

Fig. 5
Fig. 5

Mechanistic investigation of the reactions between 1,4-DNIms and amines. a 1,4-DNIm 1a reacts with 15N-labeled aniline to produce (1e). The CCDC reference number of compound 1e is 1876959. HRMS (ESI) analysis of product (1e): m/z calcd. for C9H8N215NO2 [M + H]+ 191.0587, found 191.0577; m/z calcd. for C9H7N215NNaO2 [M + Na]+ 213.0406, found 213.0400. b 2-isopropylaninline reacts with 2-methyl-1,4-DNIm 2d. HRMS (ESI) analysis of intermediate (1g): m/z calcd. for C13H17N5NaO4 [M + Na]+ 330.1178, found 330.1172; c UV–Vis characterization of products 1f and 1g

To examine whether the reaction is initiated by nucleophilic attack at the C5 positions of 1,4-DNIms, we employed 4-(2-aminoethyl)phenol to react with compound 1a, 5-methyl-1,4-DNIm 2c and 2-methyl-1,4-DNIm 2d (Supplementary Fig. 25). Results showed that compounds 1a and 2d reacted with similar efficiency, leading to 52% and 50% conversion, respectively. In contrast, 5-methyl-1,4-DNIm 2c was inert under assay conditions (2% conversion). No accumulation of ring-opening intermediates was observed in these reactions. Such substitution effect supports C5 position as the site for nucleophilic addition by amines.

To examine the proposed ring-opening step, we chose 2-isopropylaninline and 2-methyl-1,4-DNIm (2d) as reacting partners (Fig. 5b), in which the bulky substitutions are expected to slow down the ring closure and facilitate the capture of ring-opening intermediates. To our delight, in addition to the final product 1f, the accumulation of a labile intermediate 1g was observed by thin layer chromatography (Supplementary Fig. 26). HRMS analysis of the reaction mixture also revealed an intermediate with a mass matching (1g) (Fig. 5c). The ultraviolet–visible spectra of product 1g display a characteristic absorbance at 369 nm (Fig. 5c), suggesting the presence of an extended π-conjugation resulted from the opening of imidazole ring. Furthermore, gas chromatography-MS analysis of intermediate (1g) yielded molecular fragments that match the proposed structure of (1g) (Supplementary Fig. 27). Although we were not able to perform NMR analysis of (1g) due to its spontaneous conversion to product 1f, our data support the proposal that 1,4-DNIms react with amines through the ANRORC-like mechanism.

Next, we explored the application of 1,4-DNIms in Lys bioconjugation of peptides. After screening various reaction conditions, we found that in DMSO with DIEA (N,N-diisopropylethylamine) added as the base, compound 1a (0.1 mM) was able to modify a lysine-containing tripeptide 3a (0.1 mM) with near-quantitative conversion in 30 min (Fig. 6a). HRMS analysis and NMR analysis determined that product 1d is a ε-(4-nitroimidazole)-Lys derivative (Fig. 6a, Supplementary Figs. 2832). Next, we explored the chemoselectivity of compound 1a in DMSO via incubation with nucleophilic amino acids, including Tyr, Ser, Trp, His, Asp, and Arg. No reaction occurred with these amino acids, indicating that this reaction was lysine-specific. In comparison with perfluoroaryl reagent pf1, 1,4-DNIm 1a displayed similar reactivity when modifying a lysine-containing tripeptide in DMSO (Supplementary Fig. 33). The 4-nitroimidazole linkage resulted from Lys-(1,4-DNIm) bioconjugation was highly stable and remained intact after incubation under acidic (pH 2.0), basic (pH 10.0), and oxidative (10 mM H2O2) conditions at 37 °C for 10 h (Supplementary Fig. 34). To sum up, 1,4-DNIms react specifically with cysteine thiols under neutral aqueous conditions, and modify lysine ε-amines through a distinct mechanism with high selectivity in organic solvents in the absence of thiols by generating stable conjugates.

Fig. 6
Fig. 6

Macrocyclization of peptides by a bisfunctional 1,4-DNIm through lysine and cysteine crosslinking. a Lysine modification by 1,4-DNIm 1a. b Peptide macrocyclization by bisfunctional 1,4-DNIm 5a. c Preparation of a bicyclic peptide by controllable sequential macrocyclization by 1,4-DNIm 5a

Lys-specific peptide stapling by bisfunctional 1,4-DNIms

Next, we applied bisfunctional 1,4-DNIm 5a in peptide macrocyclization by crosslinking two lysine residues. By varying the site of lysine residues (i,i + 4 to i,i + 12), a series of peptide macrocycles were prepared in high yields (Fig. 6b). Diaminopropionic acid (Dap) in peptide P7-Dap was also efficiently crosslinked, demonstrating the versatility of this protocol. Peptide macrocyclization usually stabilizes the secondary structures of peptides and improves their proteolytic stability. Indeed, circular dichroism analysis showed that cyclization of peptide P8 increased the content of α-helical structure (Supplementary Fig. 35). Regarding the proteolytic stability of peptides, linear peptide P7 was completely hydrolyzed in the presence of proteinase K in 30 min, whereas cyclized P7 was significantly more stable and remained intact for up to 4 h (Supplementary Fig. 36).

Preparation of bicyclic peptides by bisfunctional 1,4-DNIms

The synthesis of peptide architectures with overlapping rings is particularly challenging because it often requires appropriate orthogonal ring-closing reactions, as well as the incorporation of functionalized unnatural amino acids41,42. The controllable chemoselectivity of 1,4-DNIms under aqueous and organic solvents provides a convenient method to integrate two consecutive stapling reactions and to enable regioselective construction of complex bicyclic peptides. For this purpose, we synthesized a model peptide P10 containing two cysteine and two lysine residues for i,i + 5-position macrocyclization (Fig. 6c). Incubation of stapling reagent 5a (0.1 mM) with peptide P10 (0.1 mM) in 100 mM HEPES buffer at pH 7.0 resulted in quantitative conversion to the cyclic product P11 in 30 s as determined by LC-MS (Supplementary Fig. 37). The Cys-specific crosslinking are confirmed by the lack of fragmentation between two cysteine residues during tandem MS analysis (Fig. 6c, Supplementary Fig. 38). The reaction mixture of 5a and P10 was then lyophilized without purification, redissovled in DMSO, and treated with another 2 equiv compound 5a. Results showed that a second crosslinking between lysine residues in the monocyclic peptide P11 was achieved in 30 min by yielding a bicyclic peptide P12 with overlapping ring structures, as indicated by tandem MS analysis (Fig. 6c, Supplementary Fig. 39). Thus, bisfunctional 1,4-DNIms are powerful peptide stapling reagents that allow controllable consecutive peptide stapling to generate peptide scaffolds with entangled macrocycles.

Source link


Please enter your comment!
Please enter your name here