Inductive Effects on the Energetics of Prolyl Peptide Bond Isomerization: Implications for Collagen Folding and Stability

Abstract

The hydroxylation of proline residues in collagen enhances the stability of the collagen triple helix. Previous X-ray diffraction analyses had demonstrated that the presence of an electron-withdrawing substituent on the pyrrolidine ring of proline residues has significant structural consequences [Panasik, N., Jr.; Eberhardt, E. S.; Edison, A. S.; Powell, D. R.; Raines, R. T. Int. J. Pept. Protein Res. 1994, 44, 262–269]. Here, NMR and FTIR spectroscopy were used to ascertain kinetic and thermodynamic properties of N-acetyl-[β,γ-¹³C]D,L-proline methylester (1); N-acetyl-4(R)-hydroxy-L-proline [¹³C]methylester (2); and N-acetyl-4(R)-fluoro-L-proline methylester (3). The pK_a’s of the nitrogen atom in the parent amino acids decrease in the order: proline (10.8) > 4(R)-hydroxy-L-proline (9.68) > 4(R)-fluoro-L-proline (9.23). In water or dioxane, amide I vibrational modes decrease in the order: 1 > 2 > 3. At 37 °C in dioxane, the rate constants for amide bond isomerization are greater for 3 than 1. Each of these results is consistent with the traditional picture of amide resonance coupled with an inductive effect that results in a higher bond order in the amide C=O bond and a lower bond order in the amide C–N bond. Further, at 37 °C in water or dioxane equilibrium concentrations of the trans isomer increase in the order: 1 < 2 < 3. Inductive effects may therefore have a significant impact on the folding and stability of collagen, which has a preponderance of hydroxyproline residues, all with peptide bonds in the trans conformation.

Introduction

Collagen is the principle structural protein in vertebrates.^1,2 In vivo, structural collagen consists of three polypeptide chains that form an extended right-handed triple helix.³ Each polypeptide chain contains approximately three hundred repeats of the sequence: Gly–Xaa–Yaa, in which Xaa and Yaa are often L-proline (Pro) and 4(R)-hydroxy-L-proline (Hyp) residues, respectively. The hydroxylation of Pro residues is a post-translational modification catalyzed by the enzyme prolyl 4-hydroxylase. Defects in prolyl 4-hydroxylase activity have been associated with the aging process as well as a variety of diseases including arthritis and rheumatism.^4,5

The biosynthesis of collagen has been studied extensively.^6–8 Collagen strands are synthesized as prepropeptides in which the pre sequence targets the polypeptide to the Golgi complex and there is removed by a protease. Three procollagen polypeptides then become covalently crosslinked through interstrand disulfide bonds within the pro region. The crosslinked chains are subjected to a variety of post-translational modifications before cleavage of the propeptide region and secretion into the extracellular matrix, where the chains fold into a triple helix. Of these modifications, the hydroxylation of proline residues by prolyl 4-hydroxylase is the most prevalent, as Hyp constitutes approximately 10% of all collagen residues.

Numerous in vitro studies with procollagen and model peptides have explored the role of Hyp in the folding and stability of collagen.⁵ Procollagen polypeptides that are deficient in Hyp can form triple helices, but these triple helices are unstable at room temperature.^9,10 Thermal denaturation studies have demonstrated that triple helix stability correlates with both overall Hyp content and Hyp position within the polypeptide.¹¹ Further, studies on model peptides suggest that peptide conformation is significantly affected by the hydroxylation of proline residues.^12,13

Several models have been proposed in which Hyp mediates collagen stability by orienting water molecules to form interstrand hydrogen bonds.^14,15 In these models, no interstrand hydrogen bonds can be formed directly between the hydroxyl group of Hyp residues and any mainchain heteroatoms in the adjacent polypeptide chains of the triple helix.¹⁶ Recently, a high-resolution X–ray diffraction analysis has revealed that water molecules do indeed form bridges between hydroxyl groups of Hyp residues and mainchain carbonyl groups.16 These bridges consist of 1, 2, or 3 water molecules.¹⁷

Because 25% of the residues in a typical collagen molecule are Pro or Hyp, the properties of these residues are likely to contribute greatly to collagen stability. In most peptide bonds, the trans (Z) isomer is greatly favored over the cis (E) isomer. In contrast, the trans isomer of a prolyl peptide bond is only slightly favored over the cis isomer. The interconversion of cis and trans isomers about prolyl peptide bonds has been identified as the rate-limiting step in protein folding pathways,¹⁸ including that of collagen.^19–21 This attribute of collagen is not surprising because all peptide bonds in triple-helical collagen reside in the trans conformation. Moreover, the enzyme peptidyl–prolyl cis–trans isomerase (PPIase), which catalyzes the cis–trans interconversion of prolyl peptide bonds, accelerates the proper assembly of collagen molecules.²⁰

The objective of this study is to determine the energetic consequences for the prolyl peptide bond of having an electron-withdrawing group in the 4(R) position of the pyrrolidine ring. Accordingly, we synthesized derivatives of proline having a hydroxyl or fluoro group at this position. We used these derivatives to determine the effect of electron withdrawal on (i) the pK_a of the prolyl nitrogen, (ii) the amide I vibrational mode of a prolyl peptide bond, and (iii) the kinetics and thermodynamics of prolyl peptide bond isomerization. The results have significance for understanding the folding and stability of collagen.

Results

Three proline derivatives were synthesized for this study: N-acetyl-[β,γ-¹³C]D,L-proline methylester (1; Ac-Pro-OMe); N-acetyl-4(R)-hydroxy-L-proline [¹³C]methylester (2; Ac-Hyp-OMe); and N-acetyl-4(R)-fluoro-L-proline methylester (3; Ac-Flp-OMe) (Chart 1).²² The chirality of 2 is that found in natural collagen. The synthesis of 1–3 as the methylester avoids intramolecular hydrogen bonding, as had been observed in N-acetyl-L-proline²³ and N-acetyl-L-proline N-methylamide.^24,25 Compounds 1 and 2 were enriched with ¹³C to improve the precision of data from ¹³C NMR spectroscopy.^26,27 Compound 3 was synthesized because fluorine is more electronegative than is an oxygen and should therefore enhance any consequences of electron-withdrawal. Compound 3 was analyzed by ¹⁹F NMR spectroscopy.²⁸

Previously, we used X-ray diffraction analysis to determine the structure of unlabeled derivatives of 1, 2, and 3 (Table 1).²⁹ These structures indicated that the presence of electron-withdrawing substituents has significant structural consequences on the pyrrolidine ring. First, the peptide bond is cis and the pyrrolidine ring has C^γ endo pucker in 1; the peptide bond is trans and the pyrrolidine ring has C^γ exo pucker in 2 and 3. Second, the nitrogen becomes increasingly pyramidalized as the electron-withdrawing ability of the substituent is stronger. This increase in pyramidalization is consistent with an increase in the sp³ character of the prolyl nitrogen. Finally, although no difference was detected in bond lengths within the amide group, the C–C bonds adjacent to the electron-withdrawing substituent were shortened significantly.

Table 1.

Summary of X-Ray Diffraction Analyses ^a

compoundb	pyrrolidine ring pucker	peptide bond isomer	pyramidalization (δ;deg)c
1	Cγ-endo	cis	1.05
2a	Cγ-exo	trans	−9.24
2b	Cγ-exo	trans	−1.52
3a	Cγ-exo	trans	−16.31
3b	Cγ-exo	trans	−3.54

^aFrom ref. 29.

^bThe unit cells of crystalline 2 and 3 contain two molecules. In crystalline 2, the hydroxyl group of molecule 2a donates a hydrogen bond to the amide oxygen of molecule 2b.

^cThe parameter δ refers to the angle that the amide C–N bond makes with the plane defined by N, C^α, and C^δ of the pyrrolidine ring.

Inductive Effect on pK_a

Changes in pK_a are consistent with an inductive effect. Such effects on the nitrogen of a proline ring are evident in the previously determined pK_a’s of L-proline (10.64) and 4(R)-hydroxy-L-proline (9.66).³⁰ The lower pK_a of Hyp suggests that the hydroxyl group on the pyrrolidine ring is withdrawing electron density from the secondary amino group. Here, the pK_a’s of the parent amino acids of 1, 2, and 3 were determined by monitoring the pH-dependencies of ¹H chemical shifts. The pK_a’s of L-proline, 4(R)-hydroxy-L-proline, and 4(R)-fluoro-L-proline are 10.8, 9.68, and 9.23, respectively.³¹ This trend is similar to that observed for ethylamine (10.63), ethanolamine (9.50), and 2-fluoroethylamine (8.79),³² and is consistent with the manifestation of an inductive effect.

Inductive Effect on the Amide I Vibrational Mode

Changes in the frequency of vibrational modes can provide evidence for an inductive effect. The frequency of the amide I vibrational mode, which results primarily from the C=O stretching vibration, reports on the C=O bond order.33,34 In D₂O, the amide I vibrational modes of 1, 2, and 3 are maximal at 1608.10 cm⁻¹, 1613.08 cm⁻¹, and 1616.02 cm⁻¹, respectively (Figure 2). In dioxane, the amide I vibrational modes of 1, 2, and 3 are maximal at 1658.99 cm⁻¹, 1660.92 cm⁻¹, and 1664.78 cm⁻¹, respectively.³⁵ Thus in both solvents, the C=O bond order appears to increase in the order: 1 < 2 < 3.³⁶ This apparent inductive effect on the amide I vibrational mode is similar to that observed in the C=O stretching frequency of 4-substituted camphors.³⁷

Figure 2.

(a) Eyring plots of the cis-to-trans isomerization of 1 (○) and 3 (◇), and the trans-to-cis isomerization of 1 (●) and 3 (◆) in dioxane. (b) Eyring plots of the cis-to-trans isomerization of 1 (○) and 2 (□), and the trans-to-cis isomerization of 1 (●) and 2 (■) in 0.10 M sodium phosphate buffer, pH 7.2.

Inductive Effect on k_EZ and k_ZE. The traditional picture of amide resonance predicts that an increase in C=O bond order is accompanied by a decrease in C–N bond order.38 Such a decrease in C–N bond order would facilitate cis–trans isomerization of the amide bond. In contrast, ab initio calculations suggest that little change in C=O bond order accompanies isomerization in the gas phase.^39–41 These calculations do not yet lead to a consensus about the C=O bond order when molecules of solvent are included.^42–45 We had observed that for peptides similar to 1–3 the barrier to isomerization (ΔG^‡) correlates with the frequency of the amide I vibrational mode (υ),^26,46 as predicted from the traditional view. To search for an inductive effect on the rate of cis–trans prolyl peptide bond isomerization (eq 1), we measured these rates for 1 and 3 in dioxane⁴⁷ and 1 and 2 in water⁴⁸ by using inversion transfer NMR spectroscopy.

(1)

The effects of temperature on the cis-to-trans rate constant (k_ZE) and the trans-to-cis rate constant (k_ZE) are illustrated by Eyring plots in Figure 2. Values for ΔH^‡ and ΔS^‡ (±SE) were calculated from linear least-squares fits of the data in these plots to eq 2,⁴⁹

$$ln(k/T)=(-\mathrm{\Delta }{H}^{‡}/R)(1/T)+\mathrm{\Delta }{S}^{‡}/R+ln(k_{\text{B}}/h)$$ (2)

where R is the gas constant, k_B is the Boltzman constant, and h is Planck’s constant. The values of these activation parameters are listed in Table 2. Because k_EZ and k_ZE are lowered in protic solvents by the formation of hydrogen bonds to the oxygen of the prolyl peptide bond,^26,46,50 elevated temperatures were required to detect isomerization in water.

Table 2.

Activation Parameters for Isomerization of 1–3

compound	solvent	process	ΔH‡a (kcal/mol)	ΔS‡a [cal/(mol·K)]	k (37 °C) b (s−1)
1	dioxane	cis-to-trans	20.2 ± 1.0	6.2 ± 0.4	0.86 ± 0.44
		trans-to-cis	23.5 ± 0.3	14.4 ± 1.0	0.25 ± 0.05
	water	cis-to-trans	25.5 ± 1.4	16.2 ± 1.1	0.024 ± 0.17
		trans-to-cis	23.2 ± 1.3	7.1 ± 3.4	0.010 ± 0.008
2	dioxane	cis-to-trans		(not determined47)
		trans-to-cis		(not determined47)
	water	cis-to-trans	27.7 ± 1.2	22.5 ± 1.2	0.016 ± 0.10
		trans-to-cis	22.2 ± 0.8	3.8 ± 0.2	0.010 ± 0.004
3	dioxane	cis-to-trans	18.4 ± 0.4	1.2 ± 0.1	1.3 ± 0.3
		trans-to-cis	19.9 ± 0.4	3.2 ± 0.1	0.31 ± 0.06
	water	cis-to-trans		(not determined48)
		trans-to-cis		(not determined48)

^aValues ± SE were obtained by linear least-squares fitting of the data in Figure 2 to eq 2.

^bValues ± SE were calculated with eq 2.

The free energy barriers to isomerization of 1–3 are almost exclusively enthalpic in origin. No significant difference was detected in the values of ΔH^‡ for 1 and 2. A similar result had been obtained for Gly–Pro and Gly–Hyp in water.⁵¹ The ΔH^‡ values do, however, differ for 1 and 3. In considering proline derivatives 1 and 3 in dioxane, the enthalpic contribution to the barrier associated with k_EZ and k_ZE is reduced by 1.8 ± 1.1 and 3.6 ± 0.5 kcal/mol, respectively. If ΔH^‡ reflects the amount of bond breaking required for isomerization to occur, then these parameters suggest that the presence of fluorine on the pyrrolidine ring reduces the C–N bond order and hence the barrier to isomerization by withdrawing electron density towards the nitrogen.

Inductive Effect on K_Z/E

The corresponding cis-to-trans and trans-to-cis Eyring plots in Figure 2 are not parallel, indicating that the equilibrium constants are temperature dependent. The effects of temperature on the values of K_Z/E (= k_EZ/k_ZE) were measured directly by NMR spectroscopy, and the resulting data are illustrated by van’t Hoff plots in Figure 3. Values for ΔH° and ΔS° (±SE) were calculated from linear least-squares fits of the data in these plots to eq 3.

Figure 3.

(a) Van’t Hoff plots of the cis-to-trans isomerization of 1 (○), 2 (□), and 3 (◇) in dioxane. (b) Van’t Hoff plots of the cis-to-trans isomerization of 1 (○) and 2 (□) in 0.10 M sodium phosphate buffer, pH 7.2.

$$ln{K}_{Z/E}=(-\mathrm{\Delta }H°/R)(1/T)+\mathrm{\Delta }S°/R$$ (3)

These values are listed in Table 3. This analysis assumes that the enthalpic and entropic differences between the cis and trans isomers are independent of temperature, that is, that ΔC_p° = 0 for the reaction in eq 1.⁵² The linear van’t Hoff plots for the isomerization of 1–3 (Figure 3) and Ac-Gly-Pro-OMe²⁷ indicate that this assumption is likely to be valid for the reaction in eq 1.

Table 3.

Thermodynamic Parameters for Isomerization of 1–3

compound	solvent	ΔH° a (kcal/mol)	ΔS° a [cal/(mol·K)]	KZ/E (37 °C)b
1	dioxane	−1.04 ± 0.01	−0.84 ± 0.04	3.5 ± 0.1
	water	−1.04 ± 0.02	−0.46 ± 0.05	4.3 ± 0.2
2	dioxane	−1.65 ± 0.12	−2.58 ± 0.38	4.0 ± 1.1
	water	−1.87 ± 0.03	−2.55 ± 0.03	5.8 ± 0.3
3	dioxane	−2.29 ± 0.07	−4.60 ± 0.23	4.1 ± 0.7
	water	(not determined48)		6.2 ± 0.1 c

^aValues ± SE were obtained by linear least-squares fitting the data in Figure 3 to eq 3.

^bValues ± SE were calculated with eq 3.

^cMeasured directly.

In all conditions studied, the trans isomer of 1–3 is more stable than the cis isomer (Figure 3; Table 3). A similar result had been observed for Gly–Pro and Gly–Hyp in water at 25 °C.⁵¹ Moreover, the values of K_Z/E for 1–3 are dependent on temperature such that the trans isomer becomes increasingly favored as the temperature decreases. In other words, ΔH° < 0 for the reaction in eq 1, as had been observed for Ac-Gly-Pro-OMe²⁷ and calculated with the 6-31G** basis set of the Gaussian 82 ab initio program.⁵³ In addition, the relative stability of the trans isomer is greater in water than dioxane. A similar solvent effect was observed for Ac-Gly-Pro-OMe.²⁷ Finally and perhaps most significantly, the values of K_Z/E near room temperature increase in the order: 1 < 2 < 3.

Discussion

The presence of electron-withdrawing substituents on the pyrrolidine ring can influence the structure of proline residues.²⁹ Here, such substituents are shown to affect the kinetics and thermodynamics of prolyl peptide bond isomerization.

Kinetics

Changes in the free energy of activation (ΔG^‡ = ΔH^‡ − TΔS^‡) for prolyl peptide bond isomerization are proportional to changes in the frequency (n) of the amide I vibrational mode. In other words,

$$\mathrm{\Delta }\mathrm{\Delta }{G}^{‡}=a_{\text{I}}\mathrm{\Delta }\upsilon$$ (4)

where a_I is an empirical parameter that varies from −0.022 to −0.030 kcal·cm/mol, depending on the peptide and type of solvent.^26,46 In all comparisons of 1–3, Δυ < 8 cm⁻¹ (Figure 2). This small frequency difference corresponds to ΔΔG^‡ < 0.23 kcal/mol or only a <1.5-fold change in rate constant at 37 °C. For 1 and 3, this small difference is apparent in ΔH^‡, and in k_EZ and k_ZE calculated by extrapolation of the Eyring plots to 37 °C (Table 2), which is the approximate physiological temperature of vertebrates. For 1 and 2, these subtle differences are not detectable by our kinetic assay. Still, X-ray diffraction analyses (Table 1), pK_a determinations, and amide I vibrational modes (Figure 2) suggest that 2, like 3, has more electron density residing on the nitrogen of the prolyl peptide bond than does 1.

A typical collagen molecule has approximately thirty Hyp residues. The folding of the collagen triple helix is therefore likely to be accelerated by the cumulative effect of electron-withdrawing hydroxyl groups attached to C^γ of the pyrrolidine rings. Moreover, our results predict that the incorporation of 4(R)-fluoroproline into collagen or other proteins in which folding is limited by prolyl peptide bond isomerization would lead to a measurable increase in folding rate.

Thermodynamics

The value of K_Z/E for a prolyl peptide bond is mediated by contacts between C^α of the adjacent residue and C^α or C^δ of the proline residue.¹⁸ The value of K_Z/E for 1–3 increases as the electron-withdrawing ability of the C^δ substituent increases. The origin for this increase may be the shorter C^γ–C^δ bond. The C^γ–C^δ bond length is decreased by (0.013 ± 0.004) Å in 2 and (0.016 ± 0.003) Å in 3.²⁹ In effect, the hydroxyl group or fluorine atom attached to C^γ serves to pull C^δ away from C^α of the adjacent residue. This structural manifestation of the inductive effect should increase the stability of the trans isomer, thereby increasing K_Z/E.

An alternative explanation for the observed increase in K_Z/E is based on stereoelectronics and its effect on pyrrolidine ring pucker. Electron-withdrawing groups on C^γ can in theory alter the preferred conformation of the pyrrolidine ring. Specifically, the tendency of molecules to adopt the conformation that has the maximum number of gauche interactions between adjacent polar bonds has been termed the “gauche effect”.⁵⁴ The gauche effect has been invoked to explain the conformational preferences of double-helical nucleic acids,^55,56 but not that of Hyp residues. A gauche effect based on the nitrogen and the hydroxyl group of 2 or fluorine of 3 would impose C^γ-exo pucker on the pyrrolidine rings of 2 and 3. This pucker was indeed observed in crystalline 2 and 3 (but not 1; Table 1),²⁹ and in all of the Hyp residues (but only half of the Pro residues) in crystalline collagen.¹⁶ The role of the gauche effect in collagen structure and stability is the object of on-going work in our laboratory.

In water at 37 °C, K_Z/E is 1.5-fold larger for 2 than 1 (Table 3). Because triple-helical collagen typically contains approximately 3 × 30 Hyp residues, the effect of stabilizing the trans isomer of each Hyp residue therefore has a cumulative effect on collagen stability. This analysis does not exclude a role for water in stabilizing the collagen triple helix.¹⁷ It does, however, suggest that the inductive effect can contribute to this stability.

Conclusions

An electron-withdrawing substituent at the 4(R) position of a pyrrolidine ring has significant structural and energetic consequences. An apparent inductive effect increases pyramidalization of the prolyl nitrogen, lowers the nitrogen pK_a, shifts the amide I vibrational mode downfield, and reduces the energetic barrier to isomerization. Furthermore, such a substituent alters the prolyl peptide bond equilibrium constant. These effects have important implications for collagen folding and stability of triple-helical collagen, in which all peptide bonds are in the trans conformation.

Experimental Section

Materials

tert-Butoxycarbonyl-[β,γ-¹³C]-D,L-proline methylester,²² N-acetyl-4(R)-acetoxy-L-proline methylester,²⁹ and 3²⁹ were synthesized as described previously. L-Proline, 4(R)-hydroxyl-L-proline, and all other reagents were from Aldrich Chemical (Milwaukee, WI) and were used without further purification unless indicated otherwise. Solvents for NMR spectroscopy in pre-packaged ampules were from Aldrich Chemical and were used without further purification.

N-Acetyl-[β,γ-¹³C]-D,L-Proline Methylester (1)

tert-Butoxycarbonyl-[β,γ-¹³C]-D,L-proline methylester (0.50 g, 2.2 mmol) was added to a solution (10 mL) of dioxane containing HCl (4 N). The resulting solution was stirred at 25 °C for 30 min, then concentrated, dried, and stirred with acetic anhydride (25 mL) at 25 °C for 16 h. The reaction mixture was concentrated under reduced pressure, and the concentrate was dissolved in ethyl acetate. The resulting solution was washed with 1 N HCl (3 × 50 mL), 1 N NaOH (3 × 50 mL), and saturated aqueous NaCl. The organic extract was dried over MgSO₄, filtered, and concentrated to a yellow oil. The crude product was loaded on to a 10 g silica column in CHCl₃ and eluted with a gradient of methanol (1–5% v/v) in CHCl₃ to yield 1 as a white solid (0.37 g, 85%). R_f (ethylacetate) = 0.17. MH⁺(FAB) = 173.10. ¹H NMR (400 MHz, CDCl₃) δ 4.55 (d br, 1H, C₁^αH; minor isomer at 4.46), 3.76 (m, 1H, C₁^δH; minor isomer at 3.61), 3.62 (s, 3H, C₂H₃; minor isomer at 3.69), 3.47 (m, 1H, C₁^δH′; minor isomer at 3.33), 2.28 (m, 1H, C₁^βH; minor isomer at 2.09), 2.03 (s, 3H, C₀H₃; minor isomer at 1.95) 2.01 (m, 2H, C₁^γH; minor isomer at 1.82), 1.98 (m, 1H, C₁^βH′; minor isomer at 1.83). ¹³C NMR (CDCl₃) δ 47.52 (^δC₁; minor isomer 46.04), 29.19 (^βC₁; minor isomer 31.23).

N-Acetyl-4(R)-Hydroxy-L-Proline [¹³C]Methylester (2)

Anhydrous potassium carbonate (0.01g, 0.08 mmol) was added to a solution of N-acetyl-4(R)-acetoxy-L-proline methylester (0.15 g, 0.66 mmol) in 5 g of methanol enriched to 99.98% with ¹³C. The resulting slurry was stirred at 25 °C for 30 min. The mixture was filtered, and the filtrate was concentrated under reduced pressure. The crude product was loaded on to a 10 g silica column and eluted with ethyl acetate:hexanes (1:2 v/v) to yield 2 as a clear oil (0.079 g, 64%). R_f (ethyl acetate) = 0.17. MH⁺(FAB) = 189.14. ¹H NMR (CDCl₃) δ 4.62 (s, 1H, C₁^γH; minor isomer at 4.49), 4.33 (t, 1H; C₁^αH; minor isomer at 4.25), 3.80 (s, 1H, C₁^δ2H), 3.55 (s, 3H, OCH₃; minor isomer at 3.70), 3.35 (d, 1H, C₁^δ2H; minor isomer at 3.42), 2.91 (s, broad, 1H, OH), 2.12 (m, C₁^βH; minor isomer at 2.26), 1.90 (s, 3H, COCH₃; minor isomer at 1.78), 1.86 (m, 1H, C₁^βH); ¹³C NMR (CDCl₃) δ 58.40 (C₂; minor isomer at 57.24).

NMR Spectroscopy

Values of pK_a, k_EZ, k_ZE, and K_Z/E were determined by NMR spectroscopy.

pK_a Determinations

The secondary amine pK_a’s of the parent amino acids of 1, 2, and 3 were determined by pH-titration monitored by ¹H NMR spectroscopy. Experiments were performed on a Bruker AM500 instrument (498.68 MHz) using a 5 mm ¹H probe and ¹H bandpass filter. Solvent suppression was applied to the water signal, and a deuterium oxide insert was used to provide an external lock. Stock solutions (0.40 M; 50 mL) of L-proline, 4(R)-hydroxy-L-proline, and 4(R)-fluoro-L-proline (which was prepared by hydrolysis of tert-butoxycarbonyl-4(R)-fluoro-L-proline methylester) were prepared in 0.10 M sodium phosphate buffer, pH 7.0. An aliquot (0.40 mL) was removed from the stock solution, and the chemical shift difference (Δδ) between the α and β protons [for L-proline and 4(R)-hydroxy-L-proline], or the α and δ protons [for 4(R)-fluoro-L-proline], were measured at 25 °C. The aliquot was returned to the stock solution, and the pH of that solution was decreased by approximately 0.1 units by the addition of an aliquot (0.10 mL) of 1 N KOH. Values of pK_a were determined by nonlinear least-squares fits of the Δδ and pH data to the Henderson–Hasselbalch equation.

Kinetics

The cis–trans isomerization rate of 1–3 were determined by ¹³C (for 1 and 2) or ¹⁹F (for 3) NMR inversion transfer NMR spectroscopy.^57–59 Experiments were performed on a Bruker AM500 instrument (¹³C 125.68 MHz) using a 5 mm broadband probe, ¹H bandpass filter, and ¹³C lowpass filter; or a Bruker AM400 instrument (¹⁹F 376.48 MHz) using a 5 mm ¹⁹F probe, ¹H bandpass filter, ¹⁹F bandstop, and ¹⁹F bandpass filter.

Samples of 1, 2, and 3 were prepared at concentrations of 0.10 M and 1.0 mM in dioxane-d₈. Aqueous samples contained 20% (v/v) ²H₂O in 0.10 M sodium phosphate buffer, pH 7.2. The rate of prolyl peptide bond isomerization cannot be detected by this method at room temperature. Experiments were therefore conducted at elevated temperatures, 310–360 K. Temperature settings of the spectrometer were calibrated to within 1 °C by reference to a glycol standard.

Spectra were obtained using the inversion transfer pulse sequence in eq 5.

$$90{°}_x-\frac{1}{2}\mathrm{\Delta }\delta -90{°}_x-\tau -90{°}_x$$ (5)

Briefly, the signal for one isomer is placed on a carrier frequency, and the intensity change of the signal for the other isomer is monitored during its recovery from a selective 180° pulse. The time-dependence of the change in signal intensity allows for the determination of the isomerization rate.

The time-dependent peak height [M(τ)] for each resonance was fit by nonlinear least-squares to eq 6 and 7⁵⁷ with SIGMA PLOT 4.16 (Jandel Scientific; San Rafael, CA).

$$M_{\text{a}}(\tau )={C_{1e}}^{{\lambda }_1\tau }+{C_{2e}}^{{\lambda }_2\tau }+M_{\text{a}}^{\infty }$$ (6)

$$M_{\text{b}}(\tau )={C_{3e}}^{{\lambda }_1\tau }+{C_{4e}}^{{\lambda }_2\tau }+M_{\text{b}}^{\infty }$$ (7)

In eq 6 and 7, the “a” subscript refers to the cis resonance, the “b” subscript refers to the trans resonance, and M^∞ refers to the peak height at equilibrium. At each temperature, complementary experiments were performed in which the cis or trans peak was placed on the carrier frequency. Data from both experiments were fit to eq 6 and 7, with the cis constants denoted as in eq 5 and 6 and the trans constants denoted as C₁*, C₂*, C₃*, C₄*, λ_1a, and λ_2a. Values of k_ia and k_ib were then calculated by using eq 8 and 9.⁵⁷

$$k_{\text{ia}}=\frac{(C_1^{\text{*}}{\lambda }_{1\text{a}}+C_2^{\text{*}}{\lambda }_{2\text{a}})(C_3+C_4)-(C_3^{\text{*}}+C_4^{\text{*}})(C_1{\lambda }_1+C_2{\lambda }_2)}{(C_3^{\text{*}}+C_4^{\text{*}})(C_1+C_2)-(C_1^{\text{*}}+C_2^{\text{*}})(C_3+C_4)}$$ (8)

$$k_{\text{ib}}=\frac{(C_3^{\text{*}}{\lambda }_{1\text{a}}+C_4^{\text{*}}{\lambda }_{2\text{a}})(C_1+C_2)-(C_3{\lambda }_1+C_4{\lambda }_2)(C_1^{\text{*}}+C_2^{\text{*}})}{(C_3+C_4)(C_1^{\text{*}}+C_2^{\text{*}})-(C_3^{\text{*}}+C_4^{\text{*}})(C_1+C_2)}$$ (9)

The isomerization rate constants, k_EZ and k_ZE, were calculated by using eq 10 and 11, where α is the ratio of the cis line width to the trans line width.

$$k_{EZ}=\frac{(C_3{\lambda }_1+C_4{\lambda }_2)+k_{\text{ib}}(C_3+C_4)}{\alpha (C_1+C_2)}$$ (10)

$$k_{ZE}=\frac{\alpha (C_1{\lambda }_1+C_2{\lambda }_2)+k_{\text{ia}}(C_1+C_2)}{C_3+C_4}$$ (11)

Thermodynamics

The equilibrium constants for the interconversion of the cis and trans isomers of 1–3 were determined by measuring the peak areas of the ¹³C (for 1 and 2) or ¹⁹F (for 3) resonances for the two isomers. Peak areas were measured with the program FELIX 2.3 (Technologics; San Diego, CA). Experiments were conducted at 300–355 K. Equilibrium constants (K_Z/E = trans/cis) were calculated directly from the peak areas.

FTIR Spectroscopy

FTIR spectra were recorded on a Nicolet 5PC spectrometer. Experiments were performed at 25 °C using NaCl or CaF₂ plates, or a ZnSe crystal in a Spectra Tech circle cell. The frequency of amide I vibrational modes was determined to within 2 cm⁻¹.

Samples of 1, 2, and 3 were prepared at concentrations of 0.10 M and 1.0 mM in dioxane (which had been distilled from CaH₂) and in D₂O. No concentration effects were observed for 1 and 3 in either solvent. A second vibrational mode was present in the amide I region in a 0.10 M solution of 2. This mode was absent in a 1.0 mM solution of 2. Thus, all FTIR spectra in dioxane were recorded on 1.0 mM solutions of 1–3.

Figure 1.

(a) Amide I vibrational modes of 1 (υ_max = 1658.99 cm⁻¹), 2 (1660.92 cm⁻¹), and 3 (1664.78 cm⁻¹) in dioxane. (b) Amide I vibrational mode of 1 (υ_max = 1608.10 cm⁻¹), 2 (1613.08 cm⁻¹), and 3 (1616.02 cm⁻¹) in D₂O.

Chart 1.