S-Protected Cysteine Sulfoxide-Enabled Tryptophan-Selective Modification with Application to Peptide Lipidation

Abstract

Lipidation of peptides is a promising means of modification that can improve the therapeutic character of biologically active peptides. Here, a novel lipidation protocol for peptides is described. The C–H sulfenylation of indole in peptides using S-p-methoxybenzyl cysteine sulfoxide under acidic conditions in the presence of ammonium chloride, anisole, and triisopropylsilane enables late-stage tryptophan-selective peptide lipidation. This developed protocol has been used successfully for the lipidation of glucagon-like peptides. Oral glucose tolerance tests in wild-type mice indicated that the resulting lipidated peptides stimulate insulin secretion and exhibit a more long-lasting blood-glucose-lowering effect than a parent nonlipidated peptide.

Site-selective modification of peptides/proteins using reactions specific to particular residues allows modulation of a peptide’s chemical or biochemical character, and can improve their therapeutic effect.¹⁻⁴ Such modifications include lipidation which can improve the physical and pharmacological properties of peptides^5,6 through the binding of their lipid moieties with human serum albumin (HSA).⁷ Long-acting insulin (insulin degludec⁸) and the glucagon-like peptide-1 (GLP-1) (liraglutide⁹ and semaglutide¹⁰) are lipidated peptide therapeutic agents currently in clinical use. General access to peptide analogues has benefitted from the acylative coupling of a fatty acid-containing unit with the side chain of a lysine (Lys) attached on a solid support. This process requires only one Lys residue in the substrate or manipulation of amine protections for the selective acylation. In the reaction of peptides with lipids in solution, there is also concern that the reaction conditions suitable for hydrophilic peptides are inappropriate for coupling of the hydrophobic lipid molecules. Innovative protocols¹¹⁻¹⁵ have been explored to address these problems.

Recently, we reported the tryptophan (Trp) indole-selective C–H sulfenylation reaction using S-p-methoxybenzyl cysteine sulfoxide (Cys(MBzl)(O)) in a solution containing 1 M methanesulfonic acid (MSA) and 4 M guanidine hydrochloride (Gn·HCl) and trifluoroacetic acid (TFA) (Figure 1).¹⁶ The formation of S-chlorocysteine mediated by an ammonium chloride such as Gn·HCl under acidic conditions (Figure 1a) and subsequent electrophilic aromatic substitution (S_EAr) of the indole in Trp by the S-chlorocysteine leads to production of the tryptathionine moiety (Figure 1b).¹⁶⁻²⁰ The TFA employed for the reaction addresses the concern for the solubility of the peptides and lipids. Trp is an ideal modification residue because of its relatively low abundance in peptides,²¹⁻³³ and the applicability of Trp-selective sulfenylation as a residue selective lipidation reaction was evaluated.

Figure 1.

Formation of S-chlorocysteine from Cys(MBzl)(O) followed by C–H sulfenylation of Trp.

The lipidation of peptide substrates using a Cys(MBzl)(O)-incorporated lipid unit requires an intermolecular Trp-selective sulfenylation, which caused two concerns. One is the alkylation of the indole ring with MBzl cation resulting from the S-chlorocysteine-forming step.³⁴ The intramolecular C–H sulfenylation has no significant problem because the coexisting guanidine traps the transiently generated MBzl cation during the possible intermolecular indole alkylation. In contrast, the envisioned lipidation protocol includes the desired C–H sulfenylation and the undesired alkylation, both intermolecular reactions. Consequently, we surveyed various ammonium chlorides which are indispensable in the formation of S-chlorocysteine and evaluated their performance in trapping the MBzl cation (Table 1).

Table 1. Effects of Scavengers for the Suppression of Side Reactions Encountered in the Intermolecular Reaction.

entrya	1 (mM)	amine·HCl	additive	product conversion (%)b
1	1.2	Gn·HCl		3a (59), 4 (27), 5 (14)
2	1.2	DA·HCl		3a (86), 4 (7), 5 (7)
3	1.2	DA·HCl	anisole	3a (>95)
4	1.5	DA·HCl	anisole	3a (87), 5 (13)
5	2.0	DA·HCl	anisole	3a (62), 5 (38)
6c	1.2	DA·HCl	anisole, TIS	3a (>95, 94d)
7c	1.5	DA·HCl	anisole, TIS	3a (>95)
8c	2.0	DA·HCl	anisole, TIS	3a (>95)

^aEach reaction was conducted in 1 M MSA–4 M amine·HCl–(50 mM anisole)/TFA at 4 °C for 3 h. The reaction was diluted 5-fold with H₂O and directly analyzed by HPLC.

^bConversion (%) proportions were determined by HPLC analysis with UV detection at 220 nm and calculated using the equation: percent formation = 100 [(integ. 3a, 4, or 5)/(integ. 3a + 4 + 5)], where integ. = integration of peak area of the UV absorption.

^cAfter 3 h of reaction at 4 °C in the 1 M MSA system, 5% TIS was added to the reaction, which was then stirred at 37 °C for 30 min.

^dIsolated yield.

Initially, the effect of Gn·HCl on suppression of side reactions induced by the MBzl cation was reevaluated. The intermolecular reaction of Ac-GALFR-Trp-FG-NH₂ (2a) (1.0 mM) with Ac-L-Cys(MBzl)(O)-R-NH₂ (1) (1.2 mM) in 1 M MSA–4 M Gn·HCl/TFA at 4 °C for 3 h proceeded less efficiently than the intramolecular reaction and afforded the desired tryptathionine peptide 3a in 59% yield, but considerable amounts of the Trp-alkylated product 4 (27%) and the S-sulfenylated sulfonium peptide 5 (14%) were detected (Table 1, entry 1 and Figure 2). Among the ammonium chlorides that were examined, diisopropylammonium chloride (DA·HCl) most efficiently suppressed the alkylation with only 7% of the side product 4 remaining (Table 1, entry 2 and Figure S9). Although the amine/MBzl adducts resulting from the trapping of the cations retained the ability to transfer the MBzl cation to Trp under acidic conditions (Figure 2a), the performance of DA·HCl is superior to that of Gn·HCl, which can probably be attributed to the difference in the steric hindrance of the resulting amine/MBzl-adducts. The amine moiety of the DA/MBzl is more space-demanding than that of the Gn/MBzl adduct and this will hinder the amine/MBzl adducts from electrophilically attacking the Trp indole. The reaction in 1 M MSA–4 M DA·HCl/TFA in the presence of 50 mM anisole as a quencher of the amine/MBzl adduct led to the no formation of 4, yielding instead the desired 3a in a yield of more than 95% (entry 3 and Figure S10a). However, the increase in the concentration of 1 resulted in increasing formation of the S-sulfenylated material 5 (entries 4 and 5 and Figure S10b, c).

Figure 2.

Reactions involved in the lipidation of Trp using a Cys(MBzl)(O) derivative.

Finally, regeneration of the desired 3a from 5 was achieved by adding triisopropylsilane (TIS)³⁵ during the treatment of the mixture of 1 and 2a with 1 M MSA–4 M DA·HCl–50 mM anisole/TFA at 4 °C for 3 h, followed by an additional 30 min stirring in the presence of 5% TIS. The result was almost quantitative (94%) production of 3a (entries 6–8, Figure S10d–f, Figure S11). Consequently, the treatment with 1 M MSA–4 M DA·HCl–50 mM anisole/TFA and the subsequent reaction in the presence of TIS were adopted as the standard reaction conditions for the intermolecular lipidation. In addition to the tolerance of the formed Trp-Cys linkage under acidic conditions, the robustness of the linkage under basic conditions was confirmed by incubation of 3a in an aqueous buffer (pH 9.0) at 37 °C for 24 h (Figure S12).

A further concern is that the Trp-selective reaction for the lipidation requires stricter residue specificity than is required for its intramolecular use because of the expected excess use of the Cys(MBzl)(O) derivative for quantitative lipidation. The Trp selectivity in the intermolecular reaction was evaluated by the sulfenylation of the Trp-containing disulfide peptides 6 or the linear peptides 2 (Ac-G-Xaa-LFR-Trp-FG-NH₂: Xaa = Ala (2a), Ser (2b), His (2c), Lys (2d), Tyr (2e), Met (2f)). This sulfenylation uses from 1.2 to 2.0 equiv. of 1 under the optimized conditions at 4 °C and the results are summarized in Table 2 and Figure S13. With the exception of the Met-containing peptide 2f, almost quantitative conversion of the Trp-containing peptides (2b–2e and 6, both 1 mM) was achieved with excess sulfoxide 1 (Figure 3a for 2e and Figure S13a–e, i) as was the case for 2a (Table 1, entries 6–8). No significant reduction of the disulfide in peptide 7 was observed under the attempted conditions, even in the presence of TIS (Figures S14 and S15). The reaction of 2f with 1.2 or 1.5 equiv. of 1 at 4 °C remained incomplete even though the desired product 3f was obtained. The side product (2f + 35 Da (3f′) chlorination material) was observed as a main component (Figure 3b and Figure S13f, g). In contrast, the reaction at 37 °C dramatically improved the reaction profile, and the desired 3f was obtained with >88% conversion and 58% isolated yield (Figure S16) with 12% of the side product 3f′ remaining (Table 2, entry 8, Figure 3c, and Figure S13h). We hypothesized that the dramatic change that was observed could be attributed to the S-chlorocysteine species reacting preferentially with Met rather than Trp to afford the S-sulfenylated Met sulfonium cation. This cation then electrophilically attacks Trp at 37 °C to give the desired 3f.³⁶ Alternatively, the sulfonium cation is thought to participate in forming 3f′ at the low reaction temperature. The reactions of Ac-Nle-Cys(MBzl)(O)-NH₂ with Bz-Met-OMe in the presence or absence of Ac-Trp-OMe contributed to the identification of 3f′ (Figure S18). The formation of a chlorination material requires the presence of Ac-Trp-OMe, which is converted to the corresponding 2-chloro-Trp derivative (Figures S19–21). It was deduced from the observed results that the generated S-sulfenylated Met sulfonium cation should be converted to the corresponding S-chlorosulfonium cation, and this then reacts with the indole³⁷ or with a chloride anion to form chlorine (Cl₂)³⁸ enabling the chlorination of Trp (Figure S22). These findings indicate that the side product 3f′ is the 2-chloro-Trp congener of 2f.

Table 2. Trp-Selectivity in the Presence of Various Amino Acids.

entry	Trp peptide	1 (mM)	product conversion (%)a
1	2b	1.2	3b (>95)
2	2c	1.2	3c (>95)
3	2d	1.2	3d (>95)
4	2e	1.2	3e (>95)
5	2e	2.0	3e (>95)
6	2f	1.2	3f (30)
7	2f	1.5	3f (23)
8b	2f	1.2	3f (88, 58c)
9	6	1.2	7 (>95)

^aConversion [%] proportions were determined by HPLC analysis with UV detection at 220 nm and calculated from the equation: percent formation = 100[(integ. 3 or 7)/(integ. detected peptide materials)], where integ. = integration of peak area of the UV absorption.

^bReaction with MSA was conducted at 37 °C for 3 h.

^cIsolated yield.

Figure 3.

HPLC analyses of crude materials obtained from the reactions of 1 with (a) 2e (Table 2, entry 4) or (b, c) 2f (entries 6 and 8, respectively). Analytical HPLC conditions: linear gradient of 0.1% TFA/CH₃CN in 0.1% TFA/H₂O, 5 to 65% over 30 min. UV detection at 220 nm. *nonpeptidic materials. **disulfide peptide derived from 1 and ***sulfide form of 1. For a plausible mechanism for forming these peptidic materials, see Figure S22.

Having established the optimum conditions for the C–H sulfenylation of Trp residues, we next sought to incorporate a lipid unit into GLP-1 (7–37) (8) and the peptide segment 9 of semaglutide in clinical use (Figure 4).³⁹ Requirement of an appropriate choice of the suitable linker between the peptide and lipid molecule prompted us to incorporate two different lengths of 8-amino-3,6-dioxaoctanoic acid (miniPEG) linker.¹⁰ The necessary Cys(MBzl)(O)-incorporated lipid units (10 and 11) and the acceptor peptides (8 and 9) were prepared by 9-fluorenylmethyloxycarbonyl (Fmoc)-solid-phase peptide synthesis (SPPS) (Figures S23–26). Modification of peptides (1.0 mM) with the lipid units (1.2 mM) in 1 M MSA–4 M DA·HCl–50 mM anisole/TFA, with an additional 30 min treatment in the presence of TIS, resulted in the disappearance of parent peptides. HPLC analysis of the acidic quenched solution of the reaction showed that the crude reaction mixture has major and minor components with the same mass. However, dissolving the acid-treated samples in a buffer at pH 7.0 led to the disappearance of the minor components, indicating that the N–O acyl shift in the Gly-Thr sequence of the peptides occurs under acidic conditions (Figure S27).⁴⁰ All the attempted lipidation reactions proceeded efficiently, affording the corresponding lipidated peptides in good isolated yields after HPLC purification (12 (8 + 10), 64%; 13 (8 + 11), 67%; 14 (9 + 10), 59%; 15 (9 + 11): 53%) (Figures S28 and S29). Peptide mapping of the lipidated material 13 using chymotrypsin and trypsin showed that the Trp residue was selectively modified by the lipid unit (Figures S30 and S31). Furthermore, the lipidation on the indole 2-position was confirmed by NMR analysis of the lipidated peptide fragment obtained from the chymotrypsin digestion (Figures S32–S35).

Figure 4.

Lipidation of GLP-1 peptides (8 and 9) and HPLC analysis of crude 13. Analytical HPLC conditions: linear gradient of 0.1% TFA/CH₃CN in 0.1% TFA/H₂O, 5 to 95% over 30 min. UV detection at 220 nm. Aib = 2-aminoisobutyric acid.

Subsequently, we performed an oral glucose tolerance test (OGTT) in wild-type (WT) mice to evaluate the effect of the lipidated GLP-1 peptides (12, 13, 14, and 15) on glucose tolerance. After intraperitoneal administration of phosphate-buffered saline (PBS) or GLP-1 peptides, insulin concentrations and blood glucose were measured during the OGTT (Figure 5). After glucose ingestion and compared to PBS, the GLP-1 peptides including nonlipidated 8 were found to increase the insulin concentration (Figure 5a, b)) and reduce the glucose concentration (Figure 5c, d). Blood glucose levels at 120 and 150 min were significantly lower in 12, 13, 14, and 15 than in 8.

Figure 5.

(a, b) Blood insulin and (c, d) glucose concentration during an OGTT in 10-week-old male WT mice (n = 4–5 mice in each group). White circles and bars indicate PBS, black circles and bars indicate 8, red circles and bars indicate 12, orange circles and bars indicate 13, green circles and bars indicate 14, and blue circles and bars indicate 15. Statistical significance was calculated by one-way analysis of variance (ANOVA) with Tukey’s test using Statistical Package for the Social Sciences (SPSS) statistics. *p < 0.05 vs PBS, ^#p < 0.05 vs 8.

In conclusion, the C–H sulfenylation of indole in peptides using S-protected cysteine sulfoxide, Cys(MBzl)(O), under the acidic conditions allows highly efficient and selective lipidation of Trp residues. Attempted lipidation of GLP-1 peptides showcases the high performance of the developing toolbox for bioconjugation. Further applications of the protocol to other peptide substrates and in vivo and in vitro evaluation compared to a launched lipidated GLP-1 analogue are under review in our laboratories.

Glossary

Abbreviations

HAS

human serum albumin

GLP-1

glucagon-like peptide-1

MBzl

p-methoxybenzyl

Cys(MBzl)(O)

S-p-methoxybenzyl cysteine sulfoxide

MSA

methanesulfonic acid

Gn·HCl

guanidine hydrochloride

TFA

trifluoroacetic acid

S_EAr

electrophilic aromatic substitution

DA·HCl

diisopropylammonium chloride

TIS

triisopropylsilane

miniPEG

8-amino-3,6-dioxaoctanoic acid

Fmoc

9-fluorenylmethyloxycarbonyl

SPPS

solid-phase peptide synthesis

OGTT

oral glucose tolerance test

WT

wild-type

PBS

phosphate-buffered saline

ANOVA

analysis of variance

SPSS

statistical package for social science

Ac-L-Cys(MBzl)(O)-R-NH₂

Ac-Leu-Cys(MBzl)(O)-Arg-NH₂

Ac-GALFR-Trp-FG-NH₂

Ac-Gly-Ala-Leu-Phe-Arg-Trp-Phe-Gly-NH₂

Ac-G-Xaa-LFR-Trp-FG-NH₂,

Ac-Gly-Xaa-Leu-Phe-Arg-Trp-Phe-Gly-NH₂ (Xaa = Ala, Ser, His, Lys, Tyr, or Met)

Ac-L-Cys-G-Trp-RA-Cys-G-NH₂

Ac-Leu-Cys-Gly-Trp-Arg-Ala-Cys-Gly-NH₂

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.2c00161.

• Detailed experimental procedures and charts for the HPLC analyses of the attempted reactions or purified peptide samples (Figure S1 for 1, Figure S2 for 2a, Figure S3 for 2b, Figure S4 for 2c, Figure S5 for 2d, Figure S6 for 2e, Figure S7 for 2f, and Figure S8 for 6); ¹H NMR and HRMS spectra (Figure S17 for S2 and Figure S20 for S6); comparison of ¹H NMR (Figure S21 for S4 vs S6 and Figure S35 for frg. D vs S7) (PDF)

The authors declare no competing financial interest.