Fluorosulfate as a Latent Sulfate in Peptides and Proteins

Abstract

Sulfation widely exists in the eukaryotic proteome. However, understanding of the biological functions of sulfation in peptides and proteins has been hampered by the lack of methods to control its spatial or temporal distribution in the proteome. Herein, we report that fluorosulfate can serve as a latent precursor of sulfate in peptides and proteins, which can be efficiently converted into sulfate by hydroxamic acid reagents under physiologically relevant conditions. Photocaging the hydroxamic acid reagents further allowed for light-controlled activation of functional sulfopeptides. This work provides a valuable tool for probing functional roles of sulfation in peptides and proteins.

Graphical Abstract

O-Sulfation of the tyrosine residue is a post-translational modification (PTM) that widely exists in eukaryotic peptides and proteins (Figure 1a), and has been implicated to regulate a variety of biological functions such as immune response, hemostasis, and pathogen evasion.^1–2 However, only a small fraction of the sulfoproteome has been annotated.^3–4 A long-standing challenge for studying the sulfoproteome is that sulfation is highly heterogeneous, with various sulfopeptides and sulfoproteins exist in different sulfoforms.⁵ The seminal works of Schultz,⁶ Liu,⁷ Chatterjee,⁸ Niu,⁹ and Xiao¹⁰ that incorporate sulfotyrosine (sY) into proteins as a non-canonical amino acid (ncAA) represent notable examples to address this challenge. Expanding upon these advances, methods that allow researchers to spatiotemporally control sulfation in the proteomic context would be highly valuable for studying their functional roles in biology.¹¹

Figure 1. Background and our approach.

a) Sulfation widely exists in diverse bioactive peptides and proteins. b) In this work, fluorosulfate is incorporated in peptides and proteins as a latent sulfate and can be efficiently converted into sulfate by hydroxamic acid reagents under physiologically relevant conditions. c) Our approach mirrors the myrosinase-catalyzed Lossen-like rearrangement of glucosinolates in nature.

Caging strategies have been developed for various protein PTMs to probe how these PTMs regulate dynamic cellular events. Although a broad collection of caging groups are available for a variety of PTMs, a caging group that stably protects sulfotyrosine (sY) residues in peptides and proteins and can be efficiently removed under physiological conditions remains elusive.¹² The reasons for such a knowledge gap includes the high energy barrier for chemically activating the sulfate group for coupling chemistries, the lability of sY to acid, heat, and high-energy ionization, and the strong electron-withdrawing propensity of sulfate that renders commonly used benzylic ester caging groups unstable.^13–15 On the other hand, while multiple alkyl and aryl esters have been successfully used as protecting groups of sY in solid-phase peptide synthesis,^17–18 such as 2,2,2-trichloroethyl (TCE),^18–19 2,2-dichlorovinyl (DCV),^20–21 2,2,2-trifluoroethyl (TFE),²² neopentyl,^23–24 and phenyl²⁵ sulfate diesters, their deprotection conditions (e.g., hydrogenolysis,^{18–21, 25} strong base,²² heating,²³ high salt concentraton,²⁴ etc.) are incompatible with living systems.

In 2014, Sharpless et al. reported the reactivity of fluorosulfate in Sulfur(VI) Fluoride exchange (SuFEx) reaction.^26–29 Compared to other halogen-substituted sulfate derivatives, fluorosulfate not only has a size closest to that of sulfate, but is also far less electrophilic due to the π-donation from fluorine to sulfur.³⁰ As a result, fluorosulfate has demonstrated excellent metabolic stability in vivo.^31–32 The chemical inertness of fluorosulfate has allowed its tyrosine derivative, L-fluorosulfotyrosine (fsY), to be incorporated into peptides and proteins via solid-phase peptide synthesis^{12, 33} and ncAA mutagenesis.^{13, 32} Herein, we demonstrate that fluorosulfate can serve as a latent sulfate in sulfopeptides and sulfoproteins and can be efficiently converted into sulfate (hereafter denoted as “decaging”) by hydroxamic acid (HA) reagents under physiologically relevant conditions. Mechanistic studies revealed an unusual Lossen rearrangement pathway of fluorosulfate activation and decaging (Figure 1b) that is analogous to the myrosinase-mediated Lossen-like rearrangement of glucosinolate in nature (Figure 1c).³⁴

Our initial investigation confirmed that fluorosulfate is stable in various aqueous physiologically relevant conditions, such as buffer solution, cell lysate, and serum at neutral pH. Specifically, negligible (<5%) hydrolysis of fluorosulfate could be detected in aqueous buffer at neutral pH after 24 hours (Table S1, entries 1–2). Fluorosulfate also remained mostly intact after 12 hours in serum and after 48 hours in cell lysate (Table S2). Even tetramethylguanidine, a reagent previously reported to promote SuFEx reaction in aqueous solution,³⁵ showed no reactivity against fluorosulfate alone (Table S1). Interestingly, we found that the hydrolysis product of N-hydroxylsuccimide (Table S3 and Figure S2), N-hydroxylsuccinic acid monoamide (3), converted a fsY-containing hexapeptide 1 into the corresponding sulfopepitde 2 in 57% yield in one hour (Figure 2a). Encouraged by this finding, we examined other HA derivatives (Figure 2a, S4). Acetohydroxamic acid (4) promoted the reaction to 78% over one hour. Good yield (95%) of 2 was obtained using aromatic benzohydroxamic acid (5) under the same condition. The highest efficiency was observed when the cationic HA 6 and heteroaromatic HA 7 were used, achieving quantitative conversion in 30 minutes. Other non-HA α-nucleophile reagents such as oxime 8,³⁶ 2-aminoxime 9, and 1-hydroxybenzotriazole (10)³⁷ resulted in lower reaction efficiency. In contrast, triisopropylsilyl ether (TIPS)-masked HA 11 showed no reactivity until potassium fluoride (KF) was added to remove the TIPS protecting group (Table S4 and Figure S5), confirming that HA is the reactive center for fluorosulfate activation. It is also noteworthy that the decaging reaction mediated by 7 proceeded with no detectable side reaction in the presence of 20 equivalents of amino acids including lysine, histidine, tyrosine, and cysteine (Table S5).

Figure 2. Reagent screen and mechanistic investigation.

a) A variety of HA reagents were investigated for their ability to activate fluorosulfate in model peptide 1. Yields were determined by HPLC. b) Real-time LC-MS reaction monitoring identified two adducts of 7, 12 and 13, suggesting a Lossen rearrangement mechanism. c) No ¹⁸O-labeled products were found from the reaction in H₂¹⁸O buffer, suggesting that the sulfate product was not generated from direct hydrolysis.

To gain insight into the reaction mechanism, the reaction with 1 as the substrate and reagent 7 was monitored using liquid chromatography mass spectrometry (LC-MS) to capture the reaction intermediates (Figure 2b and Figure S6). An adduct (12) of 7 and 1 was detected, confirming the nucleophilic coupling between the HA reagent and the substrate. Surprisingly, an isocyanate adduct 13 was also detected within 10 minutes at 37 °C, suggesting an uncommon intramolecular Lossen rearrangement pathway. To further probe this possibility, we performed the decaging reaction of 1 by 7 in the buffer prepared exclusively using H₂¹⁸O. This reaction yielded 2 that contained no ¹⁸O isotope (Figure 2c and Figure S7), suggesting that the conversion of fluorosulfate into sulfate is not through direct hydrolysis. These results further support a Lossen rearrangement mechanism.^38–40 Such a pathway is similar to the myrosinase-catalyzed Lossen-like rearrangement of glucosinolate in Brassia plants, in which an inorganic sulfate and isothiocyante are generated from a thiohydroximate-O-sulfate intermediate (Figure 1c).³⁴

We then examined the decaging of various fluorosulfate-containing peptides mediated by 7 under physiological pH. Notably, peptides that contains multiple fsY residues, or nucleophilic residues (e.g., lysine or cysteine) were successfully decaged in high yields (Figure 3a). In addition, decaging can be achieved in a light-mediated fashion. The fsY residues in 16 was efficiently decaged after a 2-nitrobenzyl-caged reagent 20 was exposed to 370 nm UV light irradiation (Figure 3b).⁴¹ No conversion was observed in dark or without 20 (Figure S31).

Figure 3. Decaging of fluorosulfate-containing peptides and proteins under physiologically relevant conditions.

a) Fluorosulfate decaging in fsY-containing synthetic peptides. b) Light-mediated fluorosulfate decaging in C5aR1 22mer peptide using photocaged reagent 20. c) TTI peptide sequences and sulfation patterns. d) Thrombin inhibition assay of TTI peptides. Data were fitted to the Morrison inhibition model and error bars represent the standard deviation of three independent measurements. e) Light-mediated activation and decaging of fluorosulfate-containing TTI peptide TTI04(fsY) regulates its sulfation-dependent thrombin inhibitory activity. f) Fluorosulfate decaging in fsY-containing protein sfGFP-151-fsY and its corresponding high-resolution mass spectrometry.⁵⁰

We used tsetse thrombin inhibitor (TTI)⁴² peptides as a model system to probe the utility of the HA reagents in controlling the bioactivities associated with sulfation under physiologically relevant conditions. We used a standard human α-thrombin activity assay with Chromozym TH substrate to determine the inhibitory effects of the TTI peptides (latent) consisting of fsY residues at position 9 and 12: TTI02(fsY), TTI03(fsY), and TTI04(fsY), and the corresponding sY-containing TTI peptides (active): TTI02(sY), TTI03(sY), and TTI04(sY) (Figure 3c).^43–44 Although the latent TTI peptides still exhibited minor inhibitory effects compared to the non-sulfated control TTI01, the active TTI peptides demonstrated significantly higher potencies (Figure 3d and Figure S9).⁴⁵ The latent TTI peptides that were decaged in situ by reagent 7 all showed similar inhibitory effects as the purified active TTI peptides (Figure 3d). These results confirmed that fluorosulfate can serve as an effective latent sulfate in peptides, and can be facilely decaged in aqueous solution at neutral pH. Light-controlled decaging is also possible. For example, while the latent TTI04 (fsY) remains inactive for thrombin inhibition at 3.7 nM in the presence of 2-nitrobenzyl protected reagent 21 in dark, after irradiation thrombin activity was reduced to 21% (Figure 3e).

The small size of fluorine atom allows fsY to be facilely incorporated into proteins as a ncAA.^{32, 46} Following the procedure established by Wang et al.,³² we cloned the fsY-specific aminoacyl tRNA synthetase FsTyrRS and an optimal pyrrolysyl tRNA into plasmids for fsY incorporation into proteins. A sfGFP gene containing a TAG codon at position 151 was co-transformed along with the genes containing the FsTyrRA/tRNA pair into B95 E. coli cells.⁴⁷ The targeted sfGFP-151-fsY was successfully expressed in a 12 mg/L yield. Tandem MS results verified the incorporation of fsY at the TAG-specified position-151 (Figure S10).^{32, 48} Next, to confirm the conversion from fluorosulfate to sulfate in sfGFP-151-fsY by 7, as well as the integrity of the resulting sulfoprotein, we performed whole protein intact mass analyses of sfGFP-151-fsY before and after decaging using high-resolution Orbitrap mass spectrometry, which is capable of achieving sub-5 ppm mass accuracy⁴⁹ and can confidently resolve the 1.996 Da mass shift after decaging (Figure 3f and Figure S11).⁵⁰ Similarly, incorporation of fsY at position 3 of sfGFP and the subsequent decaging by 7 were also confirmed (Figure S12–S13). Furthermore, we showed the light-mediated decaging of sfGFP-151-fsY by the photocaged reagent 19 (Figure 3f), highlighting the potential of our approach for the spatiotemporal release of caged sulfoproteins.

Last, we tested the cytocompatibility of the fluorosuflate decaging reagents. Previously, cesium carbonate (Cs₂CO₃)/ethylene glycol³³ or 2 M ammonium acetate (NH₄OAc) aqueous solution^{24, 43} was used to remove the protecting groups for sulfate in peptides and small molecules. However, these conditions were found to be strongly denaturing to proteins (Figure S15) and highly toxic to live cells (Figure S16 and S17). In contrast, our reagents caused no protein denaturation and has low toxicity to cells at various concentrations. To mimic the cell membrane-bound sulfoproteins,⁵¹ we examined in situ fluorosulfate decaging on the surface of live Staphylococus aureus (S. aureus) cells (Figure 4a–d). S. aureus cells were chosen because there are no known endogenous sulfopeptides expressed on their surface, and the endogenous sortase A on their surface can be used to ligate peptides.^52–53 A fluorescently labeled peptide 22 consisting of a Tobacco Etch Virus (TEV) protease cleavage sequence⁵⁴ and a LPETG sortase A-recognition motif was ligated to the cell surface of S. aureus (Figure 4a, S18 and S19). The cell surface-ligated peptide was then decaged by reagent 7. Compared to the phosphate-buffered saline (PBS) buffer control, neither the cell surface ligation nor the fsY decaging experiments caused significant reduction of cell viability (Figure 4b). LC-MS analysis of the peptide residues cleaved after the decaging reaction (24) confirmed that the fsY were successfully converted into sY on live cell surface (Figure 4c). Finally, reagents 5-7 also exhibited low toxicity to mammalian cells even at millimolar concentrations based on the MTT assay (Figure 4d).⁵⁵

Figure 4. Cytocompatibility of the reagents.

a) Sortase A-mediated ligation of peptide 22 onto the S. aureus cell surface and its decaging followed by the TEV protease cleavage. b) Percent of S. aureus cell survived after sortase A-mediated ligation of 22 (Step I) and after fluorosulfate decaging by 7 (Step II) compared to the cells treated with PBS. The average data of two trials were plotted. c) LC-MS analysis of samples after the TEV cleavage identified the decaged peptide (24, bottom) compared to the cleaved peptide before decaging (23, top). d) MTT assay of the mammalian HEK-293T cells after incubation with various concentrations of reagents 5, 6, and 7. The average data of three trials were plotted.

In conclusion, we demonstrated that fluorosulfate is a physiologically compatible latent sulfate in peptides and proteins. Fluorosulfate is stable in neutral aqueous buffers, cell lysates, and serum, and can be efficiently converted into sulfate by easily modified and readily accessible HA reagents under physiologically relevant conditions via Lossen rearrangement. Leveraging the facile incorporation of fluorosulfate-containing amino acid fsY via solid-phase peptide synthesis and ncAA mutagenesis, our reported approach can be applied to studying a wide range of sulfopeptides and sulfoproteins in their physiological states. The excellent compatibility of our reagents with both bacterial and mammalian cells suggest that they are promising candidates for decaging fluorosulfate-containing peptides and proteins in experiments involving live systems.

ABBREVIATIONS

CCR5

C-C chemokine receptor type 5

PGSL-1

P-selectin glycoprotein ligand-1

HCII

heparin cofactor II

C5aR1

complement component 5a receptor 1