Bioconjugation of Green Fluorescent Protein via an Unexpectedly Stable Cyclic Sulfonium Intermediate

The development of methods for the modification of peptides and proteins under benign aqueous conditions has received a great deal of attention from both industrial and academic laboratories.¹ The ability to introduce a range of functional post-translational modifications, either natural or synthetic, to recombinantly expressed proteins, such as sugars, lipids, fluorophores, affinity tags or radionuclides, has been exploited to gain further insight into a plethora of biological process.² A number of strategies for protein modification through manipulation of the amino or carboxy terminus have been developed.³ However, the vast majority of approaches rely on the reactivity of naturally occurring amino acid side chains such as lysine, tyrosine or cysteine.⁴

One bioconjugation approach exploits the elimination of cysteine to dehydroalanine, which can then be modified by adding suitable nucleophiles. A number of reagents have been developed to convert both peptide- and protein-based cysteines to dehydroalanine, for example O-mesitylenesulfonylhydroxylamine and hexamethylphosphorous triamide.^{4b, 5} However, the majority of currently used reagents require harsh conditions that limit their application. Recently, Davis has described the elegant use of 1,4-bishaloalkanes for the β-elimination of cysteine to effect protein modification under mild conditions.⁶ An illustrative example is the treatment of a single cysteine mutant of subtilisin from Bacillus lentus SBL(S156C) (1) with 2,5-dibromohexanediamide (2) at pH 8, which afforded intermediate dehydroalanine 3, presumably via the transient formation of sulfonium 4 (Scheme 1). Dehydroalanine 3 could be subsequently functionalised with a nucleophile to give modified protein 5.

Scheme 1.

Modification of SBL(S156C) by cysteine β-elimination.

Intrigued by the potential application of the Davis double-alkylation methodology, we explored its applicability to a range of biological systems. Herein, we report preliminary studies on the modification of a cysteine mutant (S147C) of "superfolder" green fluorescent protein (GFP) bearing a single surface-accessible cysteine. We were able to synthesise unusually stable cyclic sulfonium species and to employ these in a novel bioconjugation that is complementary to current dehydroalanine-based technologies for substrates for which dehydroalanine formation is not facile. Although the presence of sulfonium ions in peptides and proteins derived from the alkylation of methionine is well established,⁷ to our knowledge this represents the first evidence for the formation of a stable cysteine-sulfonium on a protein.

We treated GFP(S147C) (6) with 2 under Davis' conditions (1500 equiv, 37 °C, 2 h, pH 8; Scheme 2).⁶ Unexpectedly the product was not the corresponding GFP dehydroalanine 7, rather electrospray (ES) MS indicated quantitative conversion to a species with a molecular weight equivalent to that of sulfonium 8 (29 486 calcd, 29 488 obs.). Indeed, throughout this study, we observed no evidence for the formation of dehydroalanine 7, or products derived therefrom, even upon incubation of sulfonium 8 for prolonged periods (vide infra). The isolation of this sulfonium species as a stable adduct and its reluctance to undergo elimination to give the corresponding dehydroalanine product 7 was noteworthy and we embarked upon further studies to understand and exploit this unexpected mode of reactivity.

Scheme 2.

Reaction of GFP(S-147C) (100 μL, 1 mg mL⁻¹ in H₂O, pH 8, 100 mm phosphate) with dibromodiamide (2). a) 2 (1500 equiv; 10 μL, 340 mm in DMF), 37 °C, 2 h.

In order to gain further evidence for the site of reaction between GFP(S147C) and 2,5-dibromohexanediamide, sulfonium 8 was treated with N-methylbromomaleimide (1 equiv).^4d No reaction was observed with 8, whereas 6 reacted cleanly, thus suggesting that 2 had reacted exclusively on cysteine to generate 8 (see the Supporting Information).

Optimisation of the formation of 8 was achieved by varying a number of parameters including the number of equivalents of 2, reaction temperature and reaction time (Table 1). Treatment of GFP(S147C) with 10 equivalents of 2 afforded no sulfonium 8 after 5 h; nevertheless, clear evidence for the formation of thioether 9 (29 566 calcd, 29 563 obs.) was apparent by LCMS (Table 1, entry 2). However, a prolonged incubation time, 20 h, afforded 8 as the only identifiable product in 53 % conversion (Table 1, entry 3). Increasing the concentration of 2,5-dibromohexanediamide afforded, as expected, more rapid formation of 8, with complete conversion of 6 at 21 °C with 25 or 50 equivalents of 2 within 20 h (Table 1, entries 4–9). At 4 °C, no sulfonium formation is apparent after 20 h even using 25 equivalents of 2,5-dibromohexanediamide (Table 1, entries 10 and 11). However, formation of sulfonium 8 at 4 °C is observed at higher concentrations of 2 (Table 1, entry 12). At 37 °C with 50 equivalents of 2 complete conversion of 6 to 8 could be rapidly achieved (Table 1, entry 13).

Table 1.

Optimisation of formation of sulfonium 8.^[a]

	2 [equiv]	T [°C]	t [h]	6 [%][b]	9 [%][b]	8 [%][b]
1	10	21	2	>95	0	0
2			5	91	9	0
3			20	47	0	53
4	25	21	2	81	19	0
5			5	50	17	33
6			20	0	0	>95
7	50	21	2	56	23	21
8			5	27	25	48
9			20	0	0	>95
10	10	4	20	>95	0	0
11	25		20	80	20	0
12	50		20	56	24	20
13	50	37	2	0	0	>95

[a] Conditions: 6 (100 μL, 1 mg mL⁻¹ in H₂O) at pH 8 (100 mM phosphate), 2 (10 μL in DMF). [b] Determined by ratio of peak heights in deconvoluted mass spectrum.

We then sought to establish the stability and reactivity profile of sulfonium 8. After preparation of sulfonium 8 under our optimised conditions (50 equiv 2, 37 °C, 2 h) excess 2 could be readily removed by repeated diafiltration into fresh buffer. Sulfonium 8 was found to be stable in the absence of nucleophiles at 21 °C for 24 h and at 4 °C for 1 month; although, significant decomposition was observed at 37 °C after 4 h (vide supra). Treatment of 8 with β-mercaptoethanol (1000 equiv, 37 °C, 2.5 h) afforded bisthioether 10 a (29 564 calcd, 29 566 obs.) in complete conversion, presumably through ring opening of the cyclic sulfonium (Scheme 3).

Scheme 3.

Reactivity of GFP(S147C) sulfonium 8 (100 μL, 0.9 mg mL⁻¹ in H₂O/DMF 10:1, pH 8, 100 mM phosphate) with β-mercaptoethanol (5 μL, 680 mM in H₂O). a) HSCH₂CH₂OH (1000 equiv), 37 °C, 2.5 h, DMF, H₂O, pH 8 (phosphate), >95 % conversion.

We then evaluated the reactivity of 8 with a range of simple nucleophiles. Treatment of 8 with the sodium salt of 1-thio-β-D-glucose (10 equiv) afforded the expected GFP–thioglucose conjugate 10 b as the sole product, with complete conversion after 5 h (Table 2, entry 1). Similarly, treatment of 8 with glutathione was also successful at both 21 and at 37 °C, generating the corresponding GFP–glutathione adduct 10 c in excellent conversion (entries 3–5). We also confirmed that these reactions could be carried out effectively at room temperature (21 °C) in 5 h, or more rapidly at 37 °C or with increased concentrations of the nucleophile (entries 2, 4 and 5). Treatment of sulfonium 8 with phenyl selenol (100 equiv) afforded the mixed thio/seleno bisether 10 d in excellent conversion (entry 6). Although the conversion from reaction between 8 and phthalimide to give 10 e was only modest (entry 7), we were delighted to observe clean reaction in high conversion upon treatment of 8 with sodium azide (1000 equiv) to generate azide-labelled GFP 10 f (entry 8).

Table 2.

Conjugation of nucleophiles with GFP sulfonium 8.^[a]

	R–H	Equiv	T [°C]	t [h]	Conv. 8 [%][b]	Product
1 2		10 100	37 21	5 5	>95 >95	10 b
3 4 5		10 100 200	37 21 37	5 5 2	>95 >95 >95	10 c
6		100	37	5	>95	10 d
7		1000	37	2.5	40	10 e
8		1000	37	2.5	>95	10 f

[a] Conditions: 8 (100 μL, 1 mg mL⁻¹ in H₂O) at pH 8 (100 mM phosphate), NuH (see the Supporting Information). [b] Determined by ratio of peak heights in deconvoluted mass spectrum.

In conclusion, we have shown that treatment of superfolder GFP with bis-alkylating agents can be used to prepare highly stable, yet synthetically useful cyclic protein sulfonium adducts. These species undergo clean addition reactions to allow the introduction of a variety of useful chemical motifs. We envisage that the stability of the generated cysteine-derived sulfonium is presumably strongly correlated to the chemical environment of the alkylated cysteine. This new bioconjugation technique could provide a useful entry into multiply functionalised proteins and further work will be directed toward extending these findings to other systems.