3S-fluoroproline as a probe to monitor proline isomerization during proein folding by ¹⁹F-NMR

Abstract

Variable-temperature inversion transfer NMR is used to determine the kinetic and thermodynamic parameters of cis-trans isomerization of N-Ac-(3R) and (3S)-fluoroproline-OMe.

Understanding how proteins change in structure from an unfolded to a folded state is complicated, in many cases, by the cis to trans isomerization of prolyl peptide bonds.¹ While dependent upon the residues surrounding a particular proline², it is assumed that in the unfolded state, proline will isomerize to equilibrium values approximated by model peptides (typically ~80% trans, 20% cis).³ Although this assumption is valid for some proteins, model peptides cannot a priori predict the population of cis-trans isomers in the unfolded state. The ability to directly measure this population in a protein would have distinct advantages, in particular in monitoring how this population changes during the folding process.⁴

The idea to monitor directly the population of proline cis-trans isomers in the unfolded state is not new, and was pioneered initially by Torchia and coworkers, through the biosynthetic incorporation of [4–¹³C] proline into collagen, and later applied to the study of staphylococcal nuclease.⁵ Although methods and instrumentation (cryogenic probes) have been developed to improve sensitivity, carbon remains one of the least sensitive nuclei for detection and thus has limited its use for monitoring processes which occur in real time.

As an alternative to carbon detection of proline cis-trans isomerization, we report here the kinetics and thermodynamics of isomerization of the simple compounds Ac-(3R) and (3S)-fluoroproline-OMe (compounds 1 and 2, respectively)⁶ in water-D₂O solution using ¹⁹F-NMR.⁷ Studies on these model compounds will provide a basis for future studies in which biosynthetic incorporation of 3S and 3R fluoroproline will be used for monitoring cis-trans isomerization of proline during protein folding

The use of ¹⁹F-NMR is unique for protein folding studies since a 1-D ¹⁹F spectrum provides residue-specific information on the folding of proteins, and the development of stopped-flow NMR methods has allowed time-dependent folding information to be obtained with dead times less than 2 seconds.⁸ The ¹⁹F signal to noise ratio is as good as hydrogen, which is approximately 8 times as sensitive as carbon.⁹ Fluorine is only slightly larger than hydrogen (0.15Å), and in most cases is non-structurally perturbing.¹⁰ Finally, London and coworkers have shown that the peptide [p-fluoro-Phe]bradykinin, which contains a -Pro-pFPhe- peptide bond, exhibited well resolved cis and trans resonances that could be used to monitor catalysis of cis-trans isomerization by the enzyme cyclophilin.¹¹

The Ac-(3S) and (3R)-fluoroproline-OMe analogs also exhibit clearly resolvable cis (E) and trans (Z) resonances – for 2 the resonances are separated by 0.8 ppm, while for 1 the resonances are separated by nearly 2 ppm. This large difference in chemical shift allowed us to monitor the kinetics of cis-trans isomerization using inversion-transfer NMR.¹² Eyring analysis of 1 and 2 is shown in Figure 1, and the activation parameters derived from this plot are summarized in Table I. The calculated equilibrium constant from the kinetic data is consistent with the observed ratio of integrated signal intensities of cis and trans isomers – at 37°C for 1, K_Z/E is 8.2 ± 0.2 (90% trans, 10% cis) and for 2 K_Z/E is 4.12 ± 0.04 (80% trans, 20% cis). In addition, the rate constants for 2 in water are similar to that reported for Ac-proline-OMe (37°C) in sodium phosphate buffer pH 7.4.^13a

Figure 1.

Eyring analysis of the temperature dependence of isomerization for 1 (closed symbols) and 2 (open symbols). Data were analyzed according to ref. 15. The symbols (●) (○) correspond to trans (Z) to cis (E) isomerization, whereas the symbols, (■) (□) correspond to cis to trans isomerization. Linear least squares fits of the data (−) are shown.

Table 1.

Eyring parameters for compounds 1 and 2.

		Eaa	ΔH‡a	ΔS‡a	k (s−1)b
1	E to Z	81.3 (2.1)c	78.8 (2.0)	−20.1 (.7)	.028 (.002)
	Z to E	73.5 (1.1)	70.9 (1.1)	−28.7 (.6)	.229 (.008)
2	E to Z	85.9 (2.7)	83.4 (2.6)	−10.8 (.5)	.016 (.001)
	Z to E	80.5 (2.9)	77.9 (2.8)	−16.5 (.8)	.065 (.005)

^aAll thermodynamic parameters are in kJ mol⁻¹ except entropy, which is in J mol⁻¹K⁻¹.

^bData recorded at 37°C.

^cError is in parentheses.

The data in Table 1 indicate that the transition state enthalpic differences between compounds 1 and 2 are small, suggesting that water-mediated hydrogen bonding to the prolyl peptide bond has not been perturbed. However, the entropic barrier of 1 is approximately two-fold greater than 2, with values that are negative, which is unusual for proline and other amides,¹³ and for proteins in general.¹⁴ One of the reviewers has aptly pointed out that a similar trend is also observed when comparing the Ac-4R and 4S fluoroproline-OMe derivatives^{13a, 15}, suggesting that a fluorine in the syn configuration of the pyrrolidine ring may sterically hinder the barrier to rotation around the C-N bond.

Raines and coworkers first showed that synthesis of a collagen-like peptide with the sequence H-(Pro-(4R)-fluoroproline-Gly)₁₀-OH resulted in a triple-helical collagen with greatly enhanced thermal stability, which was attributed to a stereoelectronic inductive effect of the fluorine.¹⁶ The inductive effect changed the kinetics and equilibrium values of cis and trans isomers of Ac-Pro-OMe, from ~80% trans, 20% cis to 90% trans, 10% cis. Independently, Renner and coworkers showed for the first time the ability to biosynthetically incorporate fluoroproline analogs into barstar C40A/C82A/P27A, which has only one cis proline (Pro 48), in E. coli. Incorporation of (4R)-fluoroproline decreased the stability of the protein, whereas incorporation of (4S)-fluoroproline increased the thermal stability, since (4S)-fluoroproline favors the cis isomer.¹⁵ Incorporation of the difluoro analog (4,4-F₂), which exhibits a cis/trans isomer ratio similar to proline, lead to a protein in which the stability was unchanged.

However, because of the overlap of the signals from the cis and trans isomers for the 4-fluoroproline derivatives, obtaining isomerization rates or equilibrium values from simple 1-D ¹⁹F-NMR experiments could not be achieved, and more complicated two-dimensional ¹⁹F-NMR methods were required.¹⁵ Based on our study using the small model compounds 1 and 2, the cis and trans isomers are easily distinguishable, and may provide a useful alternative for measuring the ratio of cis and trans isomers in the unfolded state and real-time kinetics of cis-trans proline isomerization by ¹⁹F-NMR.

Raines and coworkers have shown that hydroxylation at the 3-position of Ac-(3S)-hydroxyproline-OMe has little influence on the kinetics of cis-trans isomerization, and thus it is perhaps not surprising that Ac-(3S)-fluoroproline-OMe displays similar kinetic isomerization parameters as both natural and 3-hydroxy-substituted prolines.¹⁷ The study presented here indicates that biosynthetic incorporation of (3S)-fluoroproline into proteins would have little impact on the natural population of cis and trans isomers. Indeed, the biosynthetic incorporation of (3S)-fluoroproline, which was reported recently in an elegant study by Conticello and coworkers,¹⁷ and more recently into the protein ribonuclease T1 (Carl Frieden, personal communication), should display similar kinetics of cis-trans isomerization as that of the wild-type protein, and if there are effects on structure, activity or folding kinetics, it will be most likely due to other mechanisms.¹⁰