¹⁹F NMR-tags for peptidyl prolyl conformation analysis

Abstract

Proline cis/trans isomerism plays an important role in protein folding and mediating protein–protein interactions in short linear interacting motifs within intrinsically disordered protein regions. The slow exchange rate between cis and trans prolyl bonds provides distinct signals in ¹⁹F NMR analysis of fluorinated peptides, allowing for simple quantification of each population. However, fluorine is not naturally found in proteins but can be introduced using chemical tags. In this study, we evaluate a range of fluorinated cysteine-reactive ¹⁹F NMR tags to assess their ability to react with short, linear proline-containing peptides and accurately report on the equilibrium cis/trans-Pro populations. Several fluorinated electrophilic tags, including nitrobenzenes, sulfonylpyrimidines, and acrylamides, were found to react chemoselectively and reliably report on the %cis-Pro in the model peptide Ac-LPAAC. Other ¹⁹F NMR tags were found to be poor reporters of local proline conformation. Although pentafluoropyridine was non-chemoselective, it still reliably reported on %cis-Pro when conjugated via cysteine or tyrosine in Ac-LPAAX (X = Cys, Tyr, Lys) peptides. 3,4-Difluoronitrobenzene was found to be compatible with protein tagging, albeit it had modest reactivity and afforded a pair of regioisimeric tagging-products when reacted with a cysteine mutant of α-synuclein. These tools may be valuable for probing cis/trans-Pro populations in proteins.

---

Introduction

¹⁹F NMR is a valuable tool for the study of biological processes including protein folding, interactions and aggregation.¹¹⁹F NMR exhibits a wide chemical shift window (∼300 ppm) and is highly sensitive to its local chemical environment. ¹⁹F is a spin ½ nucleus with similar sensitivity to ¹H. It has 100% natural abundance and is almost entirely absent from nature in the organic form, including being absent in proteins. Therefore, ¹⁹F NMR studies of biological samples should exhibit no background signals – a limitation of ¹H NMR – allowing relatively straightforward analysis in native-like conditions.^2,3

Fluorinated unnatural amino acids can be readily incorporated into peptides and proteins for ¹⁹F NMR studies using solid-phase peptide synthesis or recombinant protein expression by genetic code expansion. However, the introduction of fluorine-tags by site-specific or chemoselective bioconjugation (fluorine-tagging) broadens the range of fluorine labels available, covers a wider range of chemical shifts, allows tuning of sensitivity and installation of multiple different reporters simultaneously. Fluorine-tagging can also, unlike recombinant expression or protein total synthesis methods, incorporate ¹⁹F NMR reporters into native proteins.

Cysteine is commonly targeted for protein fluorine-tagging, using warheads such as haloacetones,⁴N-aryl-2-haloacetamides,^5,6 acrylamides,⁷ maleimides,⁸ benzylbromides^9,10 and 2,2,2-trifluoroethanethiol.¹¹ For instance, the aliphatic CF₃ tag 3-bromo-1,1,1-trifluoroacetone (BTFA) was used to ¹⁹F label the Leucine transporter (LeuT) at a mutant cysteine residue, revealing four separate ¹⁹F NMR resonances due to specific conformational states.¹² The aromatic CF₃ tag 6-(trifluoromethyl)-2-pyridone (1) labelled cysteine in human serum albumin, enhancing ¹⁹F NMR chemical shift dispersion compared with BTFA due to the presence of solvent-dependent tautomers.¹³ Few fluorine tags directly arylate proteins for ¹⁹F NMR studies and this requires further evaluation. One such example is the aromatic bis-CF₃ tag bis(2,6-trifluoromethyl)pyridine (2), affording high ¹⁹F NMR signal-to-noise ratio and selective cysteine labelling of streptococcal protein G (GB1) to study conformational changes induced by dimerization and increasing [Ca²⁺].¹⁴ However, it is important to consider that introducing fluorine atoms or fluorine reporters via bioconjugation, can lead to perturbation of the native behaviour of a host protein.^15–20 Therefore, it is reasonable to be cautious of the possible effects of side chain tagging on both local and global conformational preferences. Therefore, tags should ideally be small in size and avoid introducing significant changes in polarity or charge.

Proteins are dynamic species that normally require the adoption of correctly folded states to perform their intended roles in biology. However, despite the long-established structure–function paradigm of proteins, up to 20% of eukaryotic proteins are intrinsically disordered (IDPs) or contain intrinsically disordered regions (IDRs),²¹ allowing them to adopt a variety of transient conformations to engage with different binding partners, expanding their range of roles.^22–24 Indeed, the majority of proteins exhibit distinct dynamic conformational changes involved in their normal functions. Prolyl bonds are a key facilitator of conformational change and the slow exchange between cis-Pro and trans-Pro isomers by rotation at the tertiary amide bond is often the rate-limiting step in protein folding.²⁵ Proline is also often enriched in the sequences of short linear interacting motifs (SLiMs) found in IDRs and plays a role in their flexibility and broad range of binding partners.^26,27 In some cases, one specific conformer has a significantly higher propensity to form an interaction than the other e.g. in the interaction between the prolactin receptor and 14–3–3 proteins a cis-Pro had affinity three orders of magnitude greater than trans-Pro.²⁸

Importantly, the cis/trans-Pro populations in SLiMs are not dictated by protein tertiary structures and are mostly defined by their local sequence.^29,30 We have previously shown that simple ‘conformational balance’ peptides, which include a mid-sequence proline and distal fluorinated amino acids (e.g., 4-fluorophenylalanine) introduced via SPPS, can effectively report on the influence of proximal amino acids on the populations of cis-Pro and trans-Pro.³¹ This was achieved through the integration of discrete ¹⁹F NMR signals owing to the slow rate of exchange between cis-Pro and trans-Pro conformations under ambient conditions (Fig. 1A). For translation to whole protein systems, new ¹⁹F NMR tags that report on proline cis–trans isomerisation are needed. In this work, we evaluate small, fluorinated ‘tags’ for their reactivity with a model thiol nucleophile (N-acetylcysteine) under biomimetic conditions and then apply selected examples to a proline-containing model peptide with a conjugatable cysteine, to reveal how these tags report on the cis/trans-Pro populations (Fig. 1B). The fluorine tags identified could in future be applied broadly to the fluorine-tagging of dynamic whole proteins, including SLiMs to study their folding and interactions.

Fig. 1 (A) Fluorinated amino acids can report on cis/trans-Pro populations in simple conformational balance model peptides by ¹⁹F NMR. (B) By adapting this model to include a conjugatable cysteine, new fluorine-tags can be evaluated as ¹⁹F NMR reporters of conformation.

Results and discussion

Reactivity screening of putative fluorinated protein tags

Initially, a small number of putative cysteine-reactive fluorinated reagents were identified as potential protein tags. This comprised a variety of different chemotypes and reactivities, including halo-aromatics (3–21; S_NAr), thiols (22; disulfide formation), benzyl bromides and α-halo ketones (23–25; S_N2) acrylates and maleimides (26, 27; conjugate-addition) (Fig. 2). These were mostly commercially available reagents with low molecular weight (<300 Da) and c LogP values (<3.0) to hopefully minimise their impact upon the tagged protein or peptide and to retain water-solubility. Many of the reagents contained aryl-CF₃ groups that afford greater signal-to-noise ratio than a single F and are reported to provide greater chemical shift dispersion compared to alkyl-CF₃ and reduced chemical shift anisotropy compared to aryl-F.⁹ Additionally, three fluorinated acrylamide-based tags (conjugate-acceptors) 28–30 were synthesised by reaction of a fluorinated amine with equimolar acrylic anhydride at room temperature for 0.5 h in acetonitrile without base (28, 29) or by slow addition of NaHCO₃ for synthesis of 30 (Scheme 1).

Fig. 2 Reactivity of fluorinated compounds with N-acetyl cysteine (NAC).

Scheme 1 Synthesis of acrylamides from amines and acrylic anhydride. Reagents and conditions for compounds 28 and 29: amine (1 eq.), acrylic anhydride (1 eq.), 23 °C, 0.5 h. Compound 30: amine·HCl (1 eq.), acrylic anhydride (1 eq.), water, increase pH from 2.0–7.0 with NaHCO₃.

To compare the reactivity of the fluorinated electrophilic reagents, compounds 3–30 (2 eq.) were treated with N-acetylcysteine (NAC) (1 eq.) in a solution of water–acetonitrile with DIPEA as a base at 23 °C (Scheme 2), and after 4 h, crude reactions were diluted 10-fold into water before analysis by analytical HPLC. The initial screening ruled out several of the fluorinated small molecules for further evaluation due to poor/no reactivity (summarised in Fig. 2), including difluorobenzamides 3 and 4, difluorobenzenesulfonamides 5 and 6, difluoro- and trifluoropyridines 7, 8, and 9, and halobenzotrifluorides 13–15. The halo(trifluoromethyl)pyridines 10–12 were also, somewhat surprisingly, unreactive despite their electron-deficient nature.

Scheme 2 Reaction screening of fluorine tags. Reagents and conditions: (a) fluorine tag (2 eq.), N-acetyl cysteine (1 eq.), DIPEA (5 eq.), water–acetonitrile (1 : 1), 23 °C, 4 h.

The remaining compounds 16–30 showed varying reactivity towards NAC and the desired conjugate could be identified by LCMS in most cases. Pentafluoropyridine 16 is known to be reactive towards most nucleophilic amino acid sidechains, giving N-, O- and S-centred conjugates,³² and was found to be highly reactive in our screen. The fluoro-nitrobenzenes 17–19 also showed good reactivity, although 17 and 19 both produced regioisomeric product mixtures. Fluoro-nitrobenzene 18 seemed to undergo selective substitution at the fluorine para to the nitro group judging by the single remaining signal in the ¹⁹F NMR. The difluorophenyl(methyl)sulfone 20 reacted cleanly with NAC, albeit relatively slowly (conversion 36% after 4 h) and was not explored further. The sulfonylpyrimidine 21 is reported to be reactive with cysteine,³³ and was found to react well in our model reaction, albeit with some formation of the unwanted fluorine-substitution product. 3-Chloro-trifluoroacetone 23, benzylbromides 24 and 25, N-(4-fluorophenyl)maleimide 27 and acrylamides 28–30 also mostly reacted cleanly under these conditions, although the N-(4-fluorophenyl)acrylamide 28 and N-(2-fluoroethyl)acrymalide 30 were slightly slow (77% and 76% conversion, respectively, after 4 h). The thiophenol 22 reacted with NAC forming a disulfide bond, which was likely aided by the DMSO used as solvent, therefore, this tag may require DMSO to be used as an additive in protein tagging applications. The acrylate 26 underwent the expected conjugate addition with NAC affording a mixture of diastereoisomers. Interestingly, the conjugation product was also observed to hydrolyse during the reaction according to LCMS analysis (see SI), releasing methanol (Scheme 3), unlike the related acrylamides 28–30. This may represent a useful reaction to conjugate a masked carboxylic acid to a cysteine thiol.

Scheme 3 Reagents and conditions: (a) DIPEA, water–acetonitrile (1 : 1), 23 °C, 4 h.

Probing prolyl conformation in two-state peptide models

Having identified several fluorinated electrophiles 16–30 that reacted with NAC under mild conditions, the next objectives were to (i) evaluate how well this translated to the conjugation with a cysteine-containing small peptide model and (ii) test whether each fluorinated tag could report on simple conformational differences by ¹⁹F NMR. The model peptide Ac-LPAA(4-FPhe)-NH₂ was previously characterised as a simple two-state ‘conformational balance’ containing a proline residue affording a mixture of cis-Pro and trans-Pro conformers (∼7–12% cis-Pro, Fig. 1A),^29,31 albeit with relatively poor chemical shift dispersion (0.02 ppm) between the cis-Pro and trans-Pro signals by ¹⁹F NMR. Here, the model peptide was synthesised by standard microwave-assisted Fmoc/^tBu-SPPS and a conjugatable cysteine residue replaced the 4-fluorophenylalanine (4F-Phe) from the earlier study. The cysteine-containing peptide was next treated with a selection of fluorinated electrophiles (Table 1) under the conditions used earlier (3.89 mM peptide, 7.78 mM tag, DIPEA, acetonitrile–water, 23 °C, 4 h) to obtain the conjugates (Scheme 4). The isolated yields of the tagged peptides were mostly moderate-to-low (below 40%), with some 16, 18, 23 and 29 affording fair yields (above 40%) (Table 1). The low yields were likely due to product loss during preparative HPLC purification by lyophilization, with reactions being conducted on a small scale (<5 mg crude peptide).

Table 1 Conjugation of fluorine-tags to conformational balance peptide and reporting of %cis-Pro by ¹⁹F NMR

Fluorine-tag		Ac-LPAAC conjugate	Conjugate isolated yield (%)	^19F NMR reported % cis-Pro	Dispersion^a (ppm)
a Chemical shift dispersion measured as the difference in chemical shift for cis-Pro and trans-Pro resonances. b Pair of signals observed. c After pure shift ^19F NMR to resolve multiplet. d Likely not reporting cis-Pro.
16		31	47	Multiplet (10%)^c	0.07^c
18		32	44	10%	0.06
19		33	29	16%	0.05
21		34	32	14%	0.05
22		35	37	10%	0.04
23		36	44	16%	0.05
24		37	24	Singlet	—
25		38	27	Singlet	—
26		39	21	12%^b	0.05^b
27		40	31	24%^d	0.07^d
28		41	39	25%^d	0.06^d
29		42	47	9%^c	0.06^c

---

Scheme 4 General scheme for the fluorine-tagging of a ‘conformational balance’ model peptide Ac-LPAAC-NH₂ for analysis by ¹⁹F NMR. Reagents and conditions: DIPEA, water–acetonitrile (1 : 1), 23 °C, 4 h.

As a benchmark, ¹H NMR analysis (pH 4.1) of the unmodified cysteine peptide showed the presence of two prolyl bond conformers by NOESY NMR with a relative population of ∼11.9% cisPro based on the averaged integration of amide NH resonances (see SI). Assigning ¹H NMR spectra of large peptides and proteins is challenging due to spectral crowding and signal overlap. Additionally, analysis is often conducted at acidic pH to prevent amide hydrogen exchange. In contrast, ¹⁹F NMR offers distinct advantages, including the ability to study prolyl bond conformations at neutral pH and a producing more easily interpretable spectrum.

To evaluate the ability of the tags to report on (or influence) prolyl bond cis–trans populations, the purified tagged-peptides were dissolved in PBS buffer (pH 7.4, 10% D₂O) at a concentration of 1.5 mM and analysed at 23 °C by standard 1D ¹⁹F NMR (32 scans, 375 MHz). Proton heteronuclear decoupling of ¹⁹F NMR spectroscopy was used to simplify spectra and improve signal quantification in cases where F–H coupling is possible and would result in multiplets. Tags 18–21, 26–28 all benefit from decoupling, whereas, in general the CF₃ containing tags e.g.22–25 and 29 do not require proton decoupling. Based on our previous study of the X-Pro peptide models,³¹ and the above ¹H NMR analysis, the Leu-Pro motif was expected to exhibit ∼7–12% cis-Pro. In most cases, the ¹⁹F NMR spectra of tagged peptides exhibited the expected pair of singlets representing the cis-Pro and trans-Pro (see SI).

3,4-Difluoro-nitrobenzene (18) again afforded a single regioselective substitution product 32. By integration of the resulting two NMR signals, the prolyl-bond status was estimated to be ∼10% cis-Pro (Fig. 3) in good agreement with the ¹H NMR analysis of the parent peptide. The thiophenol 22 disulfide conjugate 35 also reported 10% cis-Pro, however, 22 did not react without using DMSO as solvent.

Fig. 3 Example ¹⁹F NMR spectra of tagged Ac-LPAAC (not to scale).

The sulfonyl-pyrimidine 21 showed good reactivity towards the peptidyl cysteine and chemoselectivity for sulfone substitution over fluoride. The resulting peptide conjugate 34 displayed the characteristic cis–trans NMR signals, albeit the ¹⁹F NMR sensitivity was relatively poor compared with e.g.18 and displayed low signal to noise ratio due to lower solubility (see SI). Trifluoronitrobenzene 19, which had shown signs of poor regioselectivity with the initial NAC screen, this time conjugated to the peptide thiol relatively cleanly in the 4-position to give peptide 33. This afforded two equivalent fluorine environments with double the NMR signal integration compared to 32, but this tag reported slightly higher %cis-Pro content (16%) than expected. The 3-chloro-trifluoroacetone conjugate peptide 36 exhibited a stronger NMR signal with its three-equivalent fluorine atoms, however, 36 again reported slightly higher %cis-Pro content (16%) than expected. The fluoroacrylate conjugate 39 again produced a pair of diastereomers (Fig. 3), which was not separated during HPLC purification. The ¹⁹F NMR analysis revealed that the two diastereoisomers exhibited unique chemical shifts and each of these were split into the characteristic unequal pair of prolyl-isomers (Fig. 3). Despite this, both pairs of signals accurately reported ∼12% cis-Pro population based on relative peak area. Acrylamide conjugate 42 reported ∼9% cis-Pro, in line with the anticipated population.

The conjugation of pentafluoropyridine 16 to Ac-LPAAC-NH₂ initially afforded a complex ¹⁹F NMR spectrum (Fig. 4) owing to the F–F scalar coupling of the two different fluorine environments within the tetrafluoropyridyl-conjugate 31. Thus, it was impossible to directly quantify the populations of prolyl conformers from the NMR spectrum. A simple solution to this was to employ a pure shift broadband homonuclear decoupling ¹⁹F NMR pulse sequence³⁴ to remove the fluorine–fluorine scalar coupling and collapse the multiplets centred around approximately −93.5 and −137.8 ppm into pairs of singlets with a chemical shift dispersion of ∼0.07 ppm, revealing the presence of cis/trans-Pro conformers (Fig. 4). The pair of singlets observed for the 2-/5-position fluorine environment observed as the downfield resonance (−93.55 and −93.62 ppm) reported a cis-Pro population of 10%, which was consistent with the expected population.

Fig. 4 ¹⁹F NMR spectra of Ac-LPAAX-NH₂ peptide models tagged with 16 (not to scale). (A) with F–F coupling and (B) after pure shift F–F decoupling.

The highly electrophilic 16 is known to also react with the N- and O-nucleophiles of lysine and tyrosine, respectively.³² Therefore, it was of interest to probe how the ¹⁹F NMR spectrum (e.g. chemical shift and dispersion of the signals) is affected by changing the residue through which the tag was conjugated i.e. S, N or O nucleophiles and the nature of the sidechain (e.g. length, flexibility). Therefore, two new model peptides Ac-LPAAY and Ac-LPAAK were synthesised based on the same model peptide but replacing cysteine with lysine and tyrosine, respectively. The peptides were conjugated with 16 and analysed using ¹⁹F NMR. In each case, the ¹⁹F NMR spectrum displayed a similar pair of overlapping multiplets or broad unresolved resonances, and therefore, a pure shift pulse sequence was employed. The Tyr O-conjugate 43 exhibited a pair of downfield inequivalent singlets, that resembled those seen in the cysteine peptide (9% cis-Pro) but with different chemical shifts (−92.18 and −92.22 ppm) (Fig. 4). Therefore, 16 may have use in tagging and reporting simultaneously at different protein sites. However, for the Lys N-conjugate 44, the indistinguishable multiplet (−98.20 ppm) was replaced with an unresolved singlet after applying pure shift and it was not possible to estimate the %cis-Pro population. This suggests that fluorine-tagging of proteins with 16 through cysteine and tyrosine provides higher sensitivity to local structural changes. Whilst, tagging through the more flexible lysine side chain may be better suited for observing global protein conformational changes without being masked by local changes.

In general, there were only small differences in chemical shift dispersion observed between cis-Pro and trans-Pro resonances for each of the different fluorine tags despite their distinct chemotypes. Nonetheless, there was some evidence of a slight increase in chemical shift dispersion for N-(4-fluorophenyl)maleimide conjugate 40 (0.07 ppm) and tetrafluoropyridine cysteine conjugate 31 (0.07 ppm), which may be explained by the fluorine reporter being directly attached to the phenyl ring, making it more sensitive to the polarization of the aromatic electron cloud.⁹ Indeed, the disulfide aryl-CF₃ conjugate 35 afforded slightly smaller dispersion (0.04 ppm). Surprisingly, the trifluoromethylbenzyl-conjugates 37 and 38 – previously reported in protein studies^9,10 – afforded only singlets by ¹⁹F NMR for tagged peptides and did not report distinct signals for prolyl conformers. Conversely, the alkyl-CF₃ conjugates 36 (trifluoromethylketone – also used in protein studies;^4,12 0.05 ppm) and 42 (acrylamide; 0.06 ppm) afforded relatively good dispersion. This is despite contrary evidence that aryl-CF₃ tags exhibit improved chemical shift sensitivity over alkyl-CF₃ groups.⁹ It was also of note that the nature of the side chain conjugated affected the chemical shift dispersion. Tetrafluoropyridine tyrosine O-conjugate 43 exhibited smaller chemical shift dispersion (0.04 ppm) compared with the cysteine S-conjugate 31 (0.07 ppm), whilst there was no dispersion observed for the corresponding lysine N-conjugate. This may, however, be a function of distance from the proline or increased flexibility in the amino acid side chain.

Some tags proved to be of little value for probing proline bond conformations, for a variety of reasons (some noted above). For acrylamide peptide-conjugate 41 the %cis-Pro was abnormally high, around 25%, which was initially ascribed to the conjugate actually reporting the more proximal amide bond cis/trans populations.³⁵ However, this was unexpected due to the closely related acrylamide 42 displaying ∼9% cis-Pro and the reported fluorine tag, 2-bromo-N-(4-(trifluoromethyl)phenyl)acetamide (BTFMA) consistently proving to be useful for probing global protein conformational changes.^9,36 The reason for this discrepancy is unresolved. Maleimide-tagged 40 also reported an unlikely high %cis-Pro content (∼25%), perhaps due to the generation of two diastereomeric products from the conjugate addition. Therefore, these tags were deemed to be unreliable reporters of local %cis-Pro conformation.

Tagging reaction kinetics and chemoselectivity in aqueous buffer

To be of value to biological studies, the ¹⁹F NMR tags should also be reactive with specific protein side chains under mild physiological conditions. To compare different conjugation chemistries, nitrobenzene 18, sulfonylpyrimidine 21 chloroacetone 23 and acrylamide 29 – that were all reactive towards cysteine under semi-aqueous conditions – were mixed with equimolar NAC in Tris buffer (20 mM, pH 7.4) at 37 °C (0 min timepoint). The conjugation reaction was started by the addition of DIPEA (2 eq.) and the consumption of NAC was measured by analytical RP-HPLC peak area after quenching samples after 10, 20 and 30 minutes. Each of the compounds reacted rapidly with the thiol and achieved >50% conversion in <10 minutes (Fig. 5). Acrylamide 29, as well as chloroacetone 23 and sulfonylpyrimidine 21, all afforded complete consumption of NAC in this 10-minute window. The nitrobenzene 18 was somewhat slower, disagreeing with earlier experiments using 1 : 1 water-acetonitrile, affording ∼80% consumption after 30 minutes. Nevertheless, this could still be useful for protein bioconjugation.

Fig. 5 HPLC reaction kinetics for the potential fluorine-tags 18, 21, 23 and 29 with NAC in aqueous Tris buffer at 37 °C.

To rule out any potential off-target conjugation, 18, 21, 23 and 29 were treated with N-acetyl lysine (NAK) and N-acetyl tyrosine (NAY) under the same conditions as above for four hours with periodic monitoring by HPLC (Table 2). Compounds 18, 21 and 29 were unreactive towards either NAK or NAY and are so far unreported in the literature as cysteine-selective ¹⁹F NMR tags. However, chloroacetone 23, just like 16, reacted with both NAK and NAY and was non-selective for cysteine. This is despite the closely related bromoacetone being commonly used in the literature for cysteine-tagging.⁴

Table 2 Selectivity screening of compounds with alternative protein nucleophiles NAK and NAY

Tag		Consumption of nucleophile (%)
18		0	0
21		0	0
23		54	22
29		0	0

---

Fluorine-tagging an intrinsically-disordered protein

Finally, we aimed to evaluate whether the newly identified tags can also be used to label a protein. α-Synuclein is an intrinsically-disordered protein (IDP) that plays a role in Parkinson's disease by forming misfolded protein aggregates. A recombinant cysteine mutant of α-synuclein (A90C) has been used previously to study aggregate formation by tagging with fluorescent labels for FRET studies.^37,38 Despite its slower kinetics of conjugation with NAC compared with pyrimidine 21, 3,4-difluoro-nitrobenzene (18) was selected as an example tag for direct cysteine arylation at Cys-90 (Fig. 6) because it had afforded a relatively high conjugation yield (44%), reported the %cisPro accurately and exhibited greater chemical shift dispersion than 21 with baseline-resolution in the ¹⁹F NMR of the peptide model.

Fig. 6 α-synuclein tagging with 18. (A) ¹⁹F NMR of purified A90C α-synuclein tagged with 18. (B) MS of untagged purified A90C α-synuclein; (C) MS of purified A90C α-synuclein tagged with 18.

The tagging reaction of protein with 18 was found to be more challenging than expected. Several different reaction conditions were evaluated, including changing the tag concentration, buffer type, addition of organic base, temperature and reaction time (see SI). Tris buffer (adjusted to pH 8.6) was initially trialled based on earlier kinetics experiments, however, no tagged protein as observed by mass spectrometry after 18 h at 4 °C with 50 eq. of tag; whilst increasing the temperature to 23 °C was still unproductive. The buffer seemed to play a significant role and the reaction of protein (0.34 mM) with 18 (17 mM) in HEPES buffer adjusted to pH 8.6 afforded completely tagged protein after incubation for 18 h at 23 °C, as confirmed by mass spectrometry (MW = 14 630 Da, Fig. 6B and C). Given the earlier use of DIPEA in the kinetics experiments, we considered that adding the organic base (50 eq.) might allow us to obtain a faster tagging reaction by ensuring full ionisation of the Cys-90 thiolate. Unfortunately, despite also affording tagged product, this led to formation of an additional dehydroalanine product (MW = 14458) resulting from the elimination of the tag from cysteine (see SI). Despite this, it was found that we could obtain tagged protein with only 5 eq. of tag when adding DIPEA. ¹⁹F NMR analysis of the purified and buffer exchanged tagged protein (0.3 mM peptide, PBS buffer pH 7.4, 10% D₂O, 23 °C) exhibited a relatively sharp singlet at ∼ −107.29 ppm (Fig. 6). This indicated that the conjugation was regioselective and that a fluorinated protein signal could easily be observed, providing a tool for future studies of protein aggregation by ¹⁹F NMR.

Conclusions

This work has identified several previously unreported ¹⁹F NMR tags. Some of these exhibit well-dispersed cis/trans-Pro signals and allow accurate quantification of prolyl bond conformational status from at least three residues away in small model peptides. In most cases, the tags selectively conjugated to cysteine in unprotected peptides with rapid conjugation kinetics (<30 min) under biomimetic conditions. It was notable that when pentafluoropyridine (16) was used to tag cysteine, lysine and tyrosine, each of the S-, N- and O-conjugates afforded distinct chemical shifts and different chemical shift dispersions yet still reported the same %cis-Pro (where observable) after applying a broadband homonuclear decoupling pure shift sequence. Therefore, 16 may be useful for tagging different side chains simultaneously to study wider protein conformational changes. It was also evident that some tags either inaccurately reported or perhaps affected the prolyl bond conformational preferences.

In general, most of the tags afforded broadly similar peak widths and chemical shift dispersions between cis-Pro and trans-Pro resonances, ranging from 0.04 – 0.07 ppm. Tags 16 (after pure shift), 18, 22, 28 and 29 all afforded baseline-resolved signals and outperformed previously reported tags including trifluoromethylbenzyl groups 24 and 25 with respect to chemical shift dispersion in this prolyl cis/trans model. Cysteine-tagging also afforded greater dispersion than our earlier reported fluorinated amino acid 4-fluorophenylalanine (0.02 ppm) in the same position.³¹ However, in these peptides the chemical shift dispersion was more significantly affected by the nature of the amino acids proximal to proline (ranged from 0.01 to 0.17 ppm) than we have observed for the different tags reported here.³¹

3,4-Difluoronitrobenzene 18 was found to be compatible with protein tagging, requiring a slightly basic pH of 8.6 and room temperature to undergo nucleophilic aromatic substitution, whereas Michael acceptors such as maleimides are able to react rapidly at neutral pH and at low temperatures. Moreover, the addition of an organic base was detrimental, and led to the elimination of the tag, forming a dehydroalanine. Other tags studied may have been more reactive under milder conditions but were not tested. Future work will explore the specific structural features of ¹⁹F NMR tags that that affect chemical shift dispersion and accurate conformation reporting in SLiMs and IDPs.

Author contributions

GSMH conducted all synthetic and characterisation studies reported here. TM performed NMR analysis. FB performed protein tagging and characterisation work. KP and AL performed protein expression and purification. CRC and GSMH designed the experiments and wrote the manuscript together.

Conflicts of interest

There are no conflicts to declare.

Data availability

All data supporting this article have been included as part of the SI. This includes methods and full spectral characterisation. See DOI: https://doi.org/10.1039/d5cb00118h.

Acknowledgements

This research was funded in part, by the Biotechnology and Biological Sciences Research Council [BB/X019756/1] and The University of Edinburgh BBSRC Impact Acceleration Account [BBSRC IAA PIII141]. The University of Edinburgh provided PhD scholarship funding for GSMH, FB and TM. For the purpose of open access, the author has applied a creative commons attribution (CC BY) licence to any author accepted manuscript version arising.

Notes and references

1. G. S. Hanson and C. R. Coxon, ChemBioChem, 2024, 25, e202400195.
2. D. Gimenez, A. Phelan, C. D. Murphy and S. L. Cobb, Beilstein J. Org. Chem., 2021, 17, 293–318.
3. H. Chen, S. Viel, F. Ziarelli and L. Peng, Chem. Soc. Rev., 2013, 42, 7971–7982.
4. U. A. Hellmich, N. Pfleger and C. Glaubitz, Photochem. Photobiol., 2009, 85, 535–539.
5. A. Manglik, T. H. Kim, M. Masureel, C. Altenbach, Z. Yang, D. Hilger, M. T. Lerch, T. S. Kobilka, F. S. Thian, W. L. Hubbell, R. S. Prosser and B. K. Kobilka, Cell, 2015, 161, 1101–1111.
6. H. R. Kalbitzer, G. Rohr, E. Nowak, R. S. Goody, W. Kuhn and E. Zimmermann, NMR Biomed., 1992, 5, 347–350.
7. H. Hiscocks, A. Ung and G. Pascali, AppliedChem, 2023, 3, 256–278.
8. J. M. Edwards, J. P. Derrick, C. F. van der Walle and A. P. Golovanov, Mol. Pharm., 2018, 15, 2785–2796.
9. L. Ye, S. T. Larda, Y. F. Frank Li, A. Manglik and R. S. Prosser, J. Biomol. NMR, 2015, 62, 97–103.
10. M. Somlyay, K. Ledolter, M. Kitzler, G. Sandford, S. L. Cobb and R. Konrat, ChemBioChem, 2020, 21, 696–701.
11. R. Horst, J. J. Liu, R. C. Stevens and K. Wüthrich, Angew. Chem., Int. Ed., 2013, 52, 10762–10765.
12. A. Daminato, C. J. Loland and E. J. Cabrita, J. Neurochem., 2025, 169, e16278.
13. G. A. Frere, A. Hasabnis, C. B. Francisco, M. Suleiman, O. Alimowska, R. Rahmatullah, J. Gould, C. Y. C. Su, O. Voznyy, P. T. Gunning, E. A. Basso and R. S. Prosser, J. Am. Chem. Soc., 2024, 146, 3052–3064.
14. Z. Chai, Q. Wu, K. Cheng, X. Liu, Z. Zhang, L. Jiang, Z. Zhou, M. Liu and C. Li, Angew. Chem., Int. Ed., 2023, 62, e202300318.
15. A. A. Berger, J. S. Völler, N. Budisa and B. Koksch, Acc. Chem. Res., 2017, 50, 2093–2103.
16. S. A. Miles, J. A. Nillama and L. Hunter, Molecules, 2023, 28, 6192.
17. E. S. Eberhardt, N. Panasik and R. T. Raines, J. Am. Chem. Soc., 1996, 118, 12261–12266.
18. H. Iwai, A. Lingel and A. Pluckthun, J. Biol. Chem., 2001, 276, 16548–16554.
19. B. C. Buer, J. L. Meagher, J. A. Stuckey and E. N. G. Marsh, Proc. Natl. Acad. Sci. U. S. A., 2012, 109, 4810–4815.
20. M. A. Cejas, W. A. Kinney, C. Chen, J. G. Vinter, H. R. Almond, K. M. Balss, C. A. Maryanoff, U. Schmidt, M. Breslav, A. Mahan, E. Lacy and B. E. Maryanoff, Proc. Natl. Acad. Sci. U. S. A., 2008, 105, 8513–8518.
21. B. Xue, A. K. Dunker and V. N. Uversky, J. Biomol. Struct. Dyn., 2012, 30, 137–149.
22. C. J. Oldfield and A. K. Dunker, Annu. Rev. Biochem., 2014, 83, 553–584.
23. H. J. Dyson and P. E. Wright, Nat. Rev. Mol. Cell Biol., 2005, 6, 197–208.
24. M. M. Babu, Biochem. Soc. Trans., 2016, 44, 1185–1200.
25. C. Dugave and L. Demange, Chem. Rev., 2003, 103, 2475–2532.
26. V. N. Uversky, Intrinsically Disord. Proteins, 2013, 1, e24684.
27. B. Mateos, C. Conrad-Billroth, M. Schiavina, A. Beier, G. Kontaxis, R. Konrat, I. C. Felli and R. Pierattelli, J. Mol. Biol., 2020, 432, 3093–3111.
28. F. F. Theisen, A. Prestel, N. L. Jacobsen, O. K. Nyhegn-Eriksen, J. G. Olsen and B. B. Kragelund, J. Am. Chem. Soc., 2025, 147, 5714–5724.
29. U. Reimer, G. Scherer, M. Drewello, S. Kruber, M. Schutkowski and G. Fischer, J. Mol. Biol., 1998, 279, 449–460.
30. F. Sebák, J. Szolomájer, N. Papp, G. K. Tóth and A. Bodor, Front. Biosci.-Landmark, 2023, 28, 127.
31. P. M. Killoran, G. S. Hanson, S. J. Verhoork, M. Smith, D. Del Gobbo, L. Y. Lian and C. R. Coxon, Chem. – Eur. J., 2023, 29, e202203017.
32. D. Gimenez, C. A. Mooney, A. Dose, G. Sandford, C. R. Coxon and S. L. Cobb, Org. Biomol. Chem., 2017, 15, 4086–4095.
33. M. R. Bauer, A. C. Joerger and A. R. Fersht, Proc. Natl. Acad. Sci. U. S. A., 2016, 113, E5271–E5280.
34. L. Castañar, P. Nolis, A. Virgili and T. Parella, Chem. – Eur. J., 2013, 19, 17283–17286.
35. R. Quintanilla-Licea, J. F. Colunga-Valladares, A. Caballero-Quintero, C. Rodríguez-Padilla, R. Tamez-Guerra, R. Gómez-Flores and N. Waksman, Molecules, 2002, 7, 662–673.
36. A. Manglik, T. H. Kim, M. Masureel, C. Altenbach, Z. Yang, D. Hilger, M. T. Lerch, T. S. Kobilka, F. S. Thian, W. L. Hubbell and R. S. Prosser, Cell, 2015, 161, 1101–1111.
37. M. H. Horrocks, L. Tosatto, A. J. Dear, G. A. Garcia, M. Iljina, N. Cremades, M. Dalla Serra, T. P. Knowles, C. M. Dobson and D. Klenerman, Anal. Chem., 2015, 87, 8818–8826.
38. M. L. Choi, A. Chappard, B. P. Singh, C. Maclachlan, M. Rodrigues, E. I. Fedotova, A. V. Berezhnov, S. De, C. J. Peddie, D. Athauda, G. S. Virdi, W. Zhang, J. R. Evans, A. I. Wernick, Z. S. Zanjani, P. R. Angelova, N. Esteras, A. Y. Vinokurov, K. Morris, K. Jeacock, L. Tosatto, D. Little, P. Gissen, D. J. Clarke, T. Kunath, L. Collinson, D. Klenerman, A. Y. Abramov, M. H. Horrocks and S. Gandhi, Nat. Neurosci., 2022, 25, 1134–1148.

---