An electronic effect on protein structure

Abstract

The well-known preference of the peptide bond for the trans conformation has been attributed to steric effects. Here, we show that a proline residue with an N-formyl group (H_i−1−C′_i−1=O_i−1), in which H_i−1 presents less steric hindrance than does O_i−1, likewise prefers a trans conformation. Thus, the preference of the peptide bond for the trans conformation cannot be explained by steric effects alone. Rather, an n → π* interaction between the oxygen of the peptide bond (O_i−1), and the subsequent carbonyl carbon in the polypeptide chain (C′_i) also contributes to this preference. The O_i−1 and C′_i distance and O_i−1···C′_i=O_i angle are especially favorable for such an n → π* interaction in a polyproline II helix. We propose that this electronic effect provides substantial stabilization to this and other elements of protein structure.

In folded proteins, only 0.03% of Xaa_i−1–nonPro_i peptide bonds are in the cis conformation (Stewart et al. 1990; Weiss et al. 1998; Jabs et al. 1999). This prevalence is 5.2% for Xaa_i−1–Pro_i peptide bonds. The greater stability of trans peptide bonds is assumed to arise solely from a steric effect (Schulz and Schirmer 1979; Creighton 1993; Kyte 1995; Fischer 2000). Simply put, the C^α substituents of each amino acid are forced into clashing proximity in the cis isomer, whereas this steric strain is relieved in the more pervasive trans isomer. The isomers of a prolyl peptide bond are more nearly isoenergetic, presumably because the C^δ protons of the pyrrolidine ring provide nearly as much steric encumbrance as do the C^α substituents.

Are steric effects alone responsible for the observed trans:cis ratio of prolyl peptide bonds? To answer this question, we synthesized N-formyl-l-proline methyl ester (FmProOMe; 1; Scheme 1 ▶). The amide bond in amide 1 is isologous to a prolyl peptide bond, except that the steric effects in amide 1 are reversed, favoring the cis rather than the trans isomer. This reversal occurs because the van der Waals radius of oxygen is greater than that of hydrogen and the C′_i−1=O_i−1 bond is longer than the C′_i−1−H_i−1 bond (Fig. 1A ▶). The synthesis of amide 1 as a methyl ester rather than a secondary amide avoids intramolecular hydrogen bonding to form a γ-turn, as has been observed in N-acetylproline (Madison and Schellman 1970; DeTar and Luthra 1977) and N-acetylproline N-methylamide (Matsuzaki and Iitaka 1971; Higashijima et al. 1977; Liang et al. 1992; Benzi et al. 2002).

Scheme 1.

Figure 1.

n → π* Interaction between O_i−1 and C′_i. (A) Structure of the trans isomer of N-formyl-l-proline methyl ester (1) in its C^γ-exo conformation. The C′_i−1–H and C′_i−1 =O_i−1 bond lengths are from the structure of crystalline dimethyl formamide (Borrmann et al. 2000). (B) Depiction of the n and π* natural bond orbitals of the trans isomer of N-acetyl-l-proline methyl ester (2) in its C^γ-exo conformation. The O_i−1· · · C′_i distance is δ_BD = 2.87 Å and the O_i−1· · · C′_i=O_i angle is τ_BD = 99.35° (DeRider et al. 2002). (C) Bürgi–Dunitz trajectory for the attack of a nucleophile on a carbonyl group to displace a leaving group (LG; Bürgi et al. 1973, ^1974a,^b; Bürgi and Dunitz 1983; Eliel and Wilen 1994). (D) Major and minor resonance forms of amide 1 (Pauling 1960) and a very minor structure that arises from the hyperconjugative delocalization of an n → π* interaction.

In addition to diminishing steric effects, we also sought to enhance those effects by increasing the steric bulk on either side of the amide bond of amide 1. Specifically, replacing the formyl proton with a methyl group yields N-acetyl-l-proline methyl ester (AcProOMe; 2; Scheme 1 ▶). Replacing the two C^δ protons with methyl groups yields N-formyl-5,5-dimethyl-l-proline methyl ester (Fm[dmPro]OMe; 3). We synthesized amides 2 and 3, and compared their values of K_trans/cis = [trans]/[cis] to that for amide 1. The resulting data reveal the manifestation of a previously unappreciated electronic effect on protein structure.

Results and Discussion

Existence of an electronic effect

Steric repulsion is perhaps the most well-known factor in molecular conformational stability. Based on steric considerations alone, the value of K_trans/cis for amide 1 should be <1. Yet, we find that the value of K_trans/cis, as determined by ¹H-NMR spectroscopy, is actually >1 in water, dioxane, and chloroform (Table 1, Fig. 2 ▶). This result provides the first experimental evidence that steric effects alone are not the sole determinant of the preference of a peptide bond for the trans conformation.

Table 1.

Values of Ktrans/cis for amides 1–3

	Solvent
Amide	D2O	Dioxane-d8	CDCl3
FmProOMe (1)	1.8	1.4	1.4
AcProOMe (2)	5.3	4.2	4.0
Fm(dmPro)OMe (3)	25	9	13

Values of K_trans/cis (±20%) were determined by integration of ¹H-NMR spectra obtained at 25°C. Resonances of the trans and cis isomers were assigned by using NOESY 1D spectroscopy with observed NOEs between the formyl (C′_i−1–H) or acetyl (C^α_i−1–H) and α (C^α_i–H) or δ (C^δ_i–H) protons.

Figure 2.

¹H NMR spectrum of N-formyl-l-proline methyl ester (1) in dioxane-d₈ at 25°C. Values of K_trans/cis were determined from integration of the indicated resonances.

What is the explanation for the value of K_trans/cis for amide 1 being >1? In a peptide bond, the oxygen (O_i−1) bears a partial negative charge, and the carbon (C′_i) bears a partial positive charge (Pauling 1960; Momany et al. 1975; Zimmerman and Scheraga 1976). The favorable Coulombic interaction between O_i−1 and C′_i could, of course, increase the value of K_trans/cis (Zimmerman and Scheraga 1976). The true picture is, however, more complex.

No less important than steric effects in dictating molecular conformation are the stabilizing effects of hyperconjugative delocalization (Cramer 1998; Weinhold 2001). A familiar manifestation of hyperconjugation in a biomolecule is the “anomeric effect,” which arises from the delocalization of a nonbonding pair of electrons (n) from the ring oxygen of pyranose sugars to the σ* orbital of the adjacent C–O bond (Deslongchamps 1983; Petillo and Lerner 1993). This n → σ* interaction stabilizes the α anomer.

We have analyzed in detail the conformational energetics of amides like 1 with density functional theory calculations using the natural bond orbital (NBO) paradigm (DeRider et al. 2002). These previous calculations reveal that the interaction between O_i−1 and C′_i has a large dependence on the O_i−1· · · C′_i=O_i angle as well as the O_i−1· · · C′_i distance, and is better described by quantum mechanics than by simple electrostatics. More precisely, the increase in the value of K_trans/cis arises from extensive hyperconjugative delocalization of a nonbonding pair of electrons (n) from the amide oxygen to the π* orbital of the ester carbon (Fig. 1B ▶).

The n → π* interaction in amide 1 is reminiscent of the Bürgi–Dunitz trajectory of organic chemistry, which describes the most favorable approach of a nucleophile (e.g., O_i−1) to the carbon of a carbonyl group (e.g., C′_i =O_i) during an acyl transfer reaction (Fig. 1C ▶; Bürgi et al. 1973^{, 1974a,b;} Bürgi and Dunitz 1983). Accordingly, the hyperconjugative delocalization that arises from an n → π* interaction between O_i−1 and C′_i can be depicted as a very minor resonance form of amide 1 (Fig. 1D ▶). A similar structure is a likely intermediate during the formation of a 2-oxazolin-5-one, or “azlactone,” which is a cleavage product that can arise during the chemical synthesis of peptides (Dakin and West 1928a,b,c,Kemp 1979).

Strength of n → π* interaction

The strength of the n → π* interaction in amides 1 and 2 can be estimated from K_trans/cis values for isologous amides that lack a C′_i=O_i group. The isopropyl and methyl groups of N-isopropyl-N-methylformamide present steric hindrance to O_i−1 and H_i−1 that are similar to those in amide 1 (Scheme 2 ▶). In contrast to amide 1, the trans isomer of N-isopropyl-N-methylformamide (4)

Scheme 2.

lacks any stabilization from an n → π* interaction, and its value of K_trans/cis is threefold lower than that of amide 1 (Laplanche and Rogers 1963; Stewart and Siddall III 1970). A comparison of N-isopropyl-N-methylacetamide and amide 2 yields a similar result (Scheme 3 ▶). From these data,

Scheme 3.

we estimate that the n → π* interaction contributes approximately 0.7 kcal/mole (= RTln3) at 25°C to the stability of the trans isomer of amides 1 and 2. This value is in gratifying agreement with the previous density functional theory calculations (DeRider et al. 2002), which indicate that n → π* delocalization provides 0.42 kcal/mole of stabilization energy when the pyrrolidine ring of amide 2 is in the C^γ-endo (major: 66%) pucker and 1.29 kcal/mole in the C^γ-exo (minor: 34%; Fig. 1B ▶) pucker. These calculations likewise assign a value of 0.7 kcal/mole (= 0.42 kcal/mole × 66% + 1.29 kcal/mole × 34%) to the strength of the n → π* interaction. It is also noteworthy that the preference for a trans isomer of amide 2 has an almost entirely enthalpic origin (Eberhardt et al. 1993), which is consistent with the enthalpic contribution expected from an n → π* interaction as well as steric effects. Finally, we recognize that an amide carbon is somewhat less electron-deficient than is an ester carbon. As a consequence, the conformational stability provided by an n → π* interaction will be lower, but still manifested, in an amide. That stability could be similar in magnitude to that from a cation–π interaction (Shi et al. 2002a), another nonclassical noncovalent interaction (Dougherty 1996).

Replacing the formyl or C^δ protons of amide 1 with bulkier methyl groups increases steric effects in a manner that should favor the trans isomer (Scheme 1 ▶). The value of K_trans/cis does indeed increase in the order 1 < 2 < 3 (Table 1). We conclude that steric effects are manifested in amides 1–3. Likewise, the identity of the residue preceding a proline residue is a primary determinant of the prevalence of a trans prolyl peptide bond (Grathwohl and Wüthrich 1981; Yao et al. 1994), as increasing steric hindrance tends to increase the relative stability of the trans isomer in peptides. For example, Gly_i−1–Pro_i and Ala_i−1–Pro_i peptide bonds have K_trans/cis values in water of 6.2 and 12, respectively (Reimer et al. 1998). Finally, 5,5-dimethylproline is known to provide a conformational lock for the cis conformation in peptides (An et al. 1999). Clearly, steric effects make a dominant contribution to the preference for the trans isomer of peptide bonds.

Implications of n → π* interaction

Proline has a unique influence on protein structure (MacArthur and Thornton 1991; Reiersen and Rees 2001). This influence arises largely from the constraint imposed by the pyrrolidine ring on its φ torsion angle, which tends to range from −75° in the C^γ-endo pucker to −60° in the C^γexo pucker (Vitagliano et al. 2001). This constraint preorganizes proline for a favorable n → π* interaction (DeRider et al. 2002). Moreover, the subtle modulation of this interaction, as occurs in diastereomers of 4-hydroxyproline and 4-fluoroproline, has been shown to affect the value of K_trans/cis (Bretscher et al. 2001; DeRider et al. 2002). An n → π* interaction is also likely to increase the value of K_trans/cis for Xaa_i−1–nonPro_i peptide bonds, albeit to a lesser extent.

The influence of an n → π* interaction extends beyond being a determinant of the value of K_trans/cis. A meaningful n → π* interaction can be defined as one in which the O_i−1· · · C′_i distance is δ_BD ≤ 3.2 Å and the O_i−1· · · C′_i=O_i angle is 99° ≤ τ_BD ≤ 119°, which is ±10° of the Bürgi–Dunitz trajectory (Fig. 1C ▶). Such an optimal n → π* interaction can exist in a polyproline II helix (P_II), right-handed α-helix (α_R), and left-handed α-helix (α_L; Fig. 3A ▶). Each of these secondary structures could gain stability from this electronic effect.

Figure 3.

Implications of the n → π* interaction between O_i−1 and C′_i. (A) Ramachandran plot (Ramachandran and Sasisekharan 1968; Richardson 1981) showing the two “n → π*” regions of the trans isomer of AcGlyNH₂. In these regions, the O_i−1· · · C′_i distance is δ_BD ≤ 3.2 Å and the O_i−1· · · C′_i=O_i angle is 99° ≤ τ_BD ≤ 119°. The white dot indicates the φ and ψ angles for an ideal polyproline II helix (B). (B) Energy-minimized structure of AcGly₃NH₂ in the conformation of a polyproline II helix with φ = −75° and ψ = +145°. The structure is depicted as a ball-and-stick (left) or space-filling (right) model. The O_i−1· · · C′_i distance (δ_BD = 3.2 Å) and O_i−1· · · C′_i=O_i angle (τ_BD = 103°) is indicated in the ball-and-stick model.

An n → π* interaction is likely to be of particular importance to the conformational stability of a P_II helix. An ideal P_II helix has main-chain dihedral angles (φ = −75°, ψ = +145°) that produce a favorable n → π* interaction, with δ_BD = 3.2 Å and τ_BD = 103° (Fig. 3B ▶). Unlike an α-helix, a P_II helix lacks intrastrand hydrogen bonds. Yet, the P_II helix is prevalent in polyproline (Isemura et al. 1968; Tiffany and Krimm 1968), has been observed in Lys₇ and Ala₇ peptides (Rucker and Creamer 2002; Shi et al. 2002b), and is a common motif in folded proteins (Siligardi and Drake 1995; Kleywegt and Jones 1996). An n → π* interaction between O_i−1 and C′_i does not rely on solvation. Likewise, specific hydration is not critical for the conformational stability of a P_II helix, which can form in methanol, trifluoroethanol, propionic acid, and benzyl alcohol (Knof and Engel 1974; Antonyraj et al. 1998; Nakamura et al. 2001). Analogous solvent-independent stereoelectronic effects enhance the conformational stability of the collagen triple helix (Engel and Prockop 1998; Holmgren et al. 1998; Bretscher et al. 2001; Jenkins and Raines 2002), in which each residue has φ and ψ torsion angles that are similar to those in a P_II helix (Bella et al. 1994).

In the analysis above, we have treated the C′_i(O_i)–N_i+1 amide as merely a ketone. Of course, the immonium resonance form C′_i(O_i⁻ ) = N_i+1⁺, which has been estimated to contribute as much as 40% to the structure of an amide (Pauling 1960), must also play a role in determining the most favorable position for O_i−1 (Baldwin 1976). Ideally, the O_i−1· · · C′_i=N_i+1⁺ angle, like the O_i−1· · · C′_i=O_i angle, would be near 109°. Remarkably, the O_i−1· · · C′_i=N_i+1⁺ angle in a P_II helix is identical to the O_i−1· · · C′_i=O_i angle (Fig. 3B ▶), as both are τ_BD = 103°. Thus, a P_II helix is stabilized by hyperconjugative delocalization of n from O_i−1 to the π* orbitals of both C′_i=O_i and C′_i=N_i+1⁺.

A favorable Coulombic interaction between O_i−1 and C′_i has been proposed to enhance the stability of the right-handed twist of β-strands (Maccallum et al. 1995). The O_i−1· · · C′_i distance (δ_BD = 3.45 Å) and O_i−1· · · C′_i=O_i angle (τ_BD = 125°) in a right-twisted β-strand (φ = −90°, ψ = +125°) are, however, inappropriate for a strong n → π* interaction (Fig. 3A ▶). We therefore believe that an interaction between O_i−1 and C′_i is unable to provide substantial stability to the right-handed twist of β-strands.

Finally, we note that the contribution of an n → π* interaction to conformational stability can be cooperative. Both the negative charge on O_i and the C′_i=O_i bond length increase as a result of an n → π* interaction between O_i−1 and C′_i (DeRider et al. 2002). These effects can in turn serve to increase the stabilization provided by an n → π* interaction between O_i and C′_i+1.

Materials and methods

General

Chemicals and solvents were from Aldrich. Reactions were monitored by thin-layer chromatography using TLC plates (AL SIL G/UV) from Whatman, with visualization by illumination with ultraviolet light or staining with I₂. NMR spectra were obtained with Bruker AC-250, Bruker AC-300, and Varian UNITY-500 spectrometers. Mass spectra were obtained with electrospray ionization (ESI) techniques at the University of Wisconsin Biotechnology Center.

Synthesis of N-formyl-l-proline methyl ester (1)

l-Proline (2.0 g, 17.4 mmole) was dissolved in formic acid (25 mL), and the resulting solution was cooled to 0°C. The cooled solution was added to a mixture of acetic anhydride (20 mL, 218 mmole) in formic acid (25 mL), which was also at 0°C. The resulting solution was allowed to warm to room temperature overnight. The solvent was then removed under reduced pressure. The residue was purified by flash chromatography (30 g silica gel, 5% v/v methanol in chloroform). Fractions containing N-formyl-l-proline were pooled, and the solvent was removed under reduced pressure to yield N-formyl-l-proline (1.73 g, 63%) as a clear oil. R_f = 0.6 (10% v/v methanol in chloroform). ¹H-NMR (CDCl₃, 300 MHz) δ 11.7 (s, 1H), 8.34 (s, 0.6H), 8.27 (s, 0.4H), 4.45–4.52 (m, 1H), 3.63–3.77 (m, 1H), 3.55 (apparent triplet, J = 7 Hz, 1H) 2.22–2.38 (m, 1H), 1.89–2.19 (m, 3H). ¹³C-NMR (CDCl₃, 75 MHz) δ 173.7, 173.6, 163.1, 162.2, 58.9, 56.6, 46.8, 44.1, 29.4, 29.0, 23.6, 22.5. MS (ESI) m/z 166.0 (M + Na⁺).

N-Formyl-l-proline was converted to amide 1 by esterification with diazomethane. Diazomethane was generated in situ with an Aldrich minidiazald reactor. (CAUTION! Accumulated diazomethane is highly explosive. See Aldrich Technical Information Bulletin No. AL-180.) Diazald (1.5 g, 7.0 mmole) was mostly dissolved in ether (10 mL). Residual solids were removed by decanting the yellow solution into a dropping funnel with clear-seal glass joints. At a rate of ∼1 drop every 2–3 sec, the diazald solution was added to the diazald reactor, which contained KOH (1.00 g, 17.9 mmole) dissolved in water/ethanol (40:60 v/v, 2.5 mL) heated to 70°C. The ethereal diazomethane immediately distilled over to the second chamber and condensed against a coldfinger at −78°C. The yellow diazomethane droplets then dripped into a solution of N-formyl-l-proline (0.50 g, 3.2 mmole) in ether/acetonitrile (50:50 v/v, 25 mL) cooled to 0°C. The reaction continued until the product solution turned light yellow. The reaction was then quenched with glacial acetic acid. The product was isolated by removal of the volatile components under reduced pressure. The residue was purified by flash chromatography (30 g silica gel, ethyl acetate/hexanes [50:50–100:0 v/v]). Fractions containing amide 1 were pooled, and the solvent was removed under reduced pressure to yield amide 1 as a pale yellow oil (0.53 g, 97%). R_f = 0.45 (ethyl acetate/hexane [80:20 v/v], I₂ staining). ¹H-NMR (D₂O, 500 MHz) δ 8.28 (s, 0.64H), 8.23 (s, 0.36H), 4.76 (dd, J = 8, 3 Hz, 0.36H) 4.52 (m, 0.64H), 3.70–3.85 (m, 4.3H). 3.55 (m, 0.7H), 2.22–2.46 (m, 1.3H), 1.98–2.12 (m, 2.6H). ¹³C-NMR (dioxane-d₈, 125 MHz) δ 172.9, 172.4, 161.3, 160.5, 58.7, 56.5, 52.2, 51.8, 46.1, 43.8, 29.8, 29.6, 24.3, 23.1. MS (ESI) m/z 180.0 (M + Na⁺). IR Data: (neat) v_max 1742.8 (s), 1672.1 (s).

Synthesis of acetyl-l-proline methyl ester (2)

Amide 2 was synthesized as described previously (Panasik Jr. et al. 1994).

Synthesis of N-formyl-5,5-dimethyl-l-proline methyl ester (3)

2-(3-Methyl-3-nitro-butyl)-[1,3]dioxolane was prepared as described previously (Bonnett et al. 1959). 2-(3-Methyl-3-nitro-butyl)-[1,3]dioxolane (68 g, 0.36 mole) was dissolved in methanol (0.25 L). Raney nickel was added to this solution, and the flask was filled with H₂(g) from a balloon. (CAUTION! Raney nickel is highly pyrophoric and will ignite methanol vapors if dry.) Periodically, the reaction was assayed for completion by ¹³C-NMR spectroscopy. Completion times ranged from overnight to 10 days, depending on the scale and catalyst loading. Upon completion, the catalyst was carefully filtered away. The solvent was removed under reduced pressure to yield 3-[1,3]dioxolan-2-yl-1,1-dimethyl-propylamine as a pale yellow oil (56.5 g, 99%). ¹H-NMR (CDCl₃, 300 MHz) δ 4.84 (t, J = 5 Hz, 1H), 3.82–3.98 (m, 4H), 1.65–1.75 (m, 2H), 1.42–1.51 (m, 2H), 1.10 (s, 6H). ¹³C-NMR (CDCl₃, 75.5 MHz) 104.0, 64.1, 48.3, 38.1, 29.67, 28.5. MS (ESI) m/z 160.2 (M + H⁺).

3-[1,3]Dioxolan-2-yl-1,1-dimethyl-propylamine (15.3g, 96 mmole) was dissolved in hot water (40 mL) and the pH lowered to ∼3 with 2 N HCl. The resulting solution was heated at reflux for 30 min, and then made basic by the addition of 6 N KOH and extracted (4×) with chloroform. The organic layer was dried over MgSO₄(s), and the solvent was removed under reduced pressure to yield a black oil. The black oil was purified by distillation under reduced pressure to yield 2,2-dimethyl-3,4-dihydro-2H-pyrrole as a clear pungent oil (7.5 g, 80%). ¹H-NMR (CDCl₃, 300 MHz) δ 7.33 (broad s, 1H), 2.52 (t, J = 7 Hz, 2H), 1.56 (t, J = 7 Hz, 2H), 1.2 (s, 6H). ¹³C-NMR (CDCl₃, 75.4 MHz) δ 163.1, 72.9, 36.7, 34.3, 28.5.

2,2-Dimethyl-3,4-dihydro-2H-pyrrole (12.4 g, 128 mmole) was dissolved in water (70 mL) at 0°C. KCN (16.3 g, 251 mmole) was added to the solution, and the pH was lowered from ∼13.4 to ∼6 over 2 h by the addition of 2 N HCl (128 mL, 256 mmole). The pH continued to rise, but was maintained near 6 by the addition of 2 N HCl. After 3 h at 0°C, the solution was made basic by the addition of 2 N NaOH, and extracted (4×) with chloroform. The combined organic extract was dried with MgSO₄(s), and the solvent removed under reduced pressure. The residue was purified by flash chromatography (0.4 kg silica gel, ethyl acetate/hexanes [65:35 v/v]). Fractions containing 5,5-dimethyl-pyrrolidine-2-carbonitrile were pooled, and the solvent was removed under reduced pressure to yield 5,5-dimethyl-pyrrolidine-2-carbonitrile as a pale yellow oil (7.47 g, 46%). ¹H-NMR (CDCl₃, 300 MHz) δ 4.07–4.12 (apparent dd, J = 8, 5 Hz, 1H), 2.12–2.36 (m, 2H), 1.79–1.90 (m, 2H), 1.61–1.72 (m, 1H), 1.30 (s, 3H), 1.17 (s, 3H). ¹³C-NMR (CDCl₃, 75.4 MHz) δ 122.0, 59.5, 46.5, 38.3, 30.9, 29.0, 28.8. ¹³C-DEPT-135 δ 46.5, 29.0, 28.8 positive (CH or CH₃), 38.3, 30.9 negative (CH₂). MS (ESI) m/z 125.2 (M + H⁺).

5,5-Dimethyl-pyrrolidine-2-carbonitrile was hydrolyzed to form 5,5-dimethylproline as described previously (Bonnett et al. 1959). Racemic 5,5-dimethylproline was resolved with d-tartrate as described previously (An et al. 1999) to yield 5,5-dimethyl-l-proline in greater than 97% ee. 5,5-Dimethyl-l-proline was converted to N-formyl-5,5-dimethyl-l-proline methyl ester by the procedure used to convert l-proline to N-formyl-l-proline methyl ester (vide supra). ¹H NMR (D₂O, 500 MHz) δ 8.25 (s, 0.96H), 8.10 (s, 0.04H), 4.48–4.58 (m, 1H), 3.76 (s, 3H), 2.2–2.45 (m, 1H), 1.85–2.15 (m, 3H), 1.47 (s, 3H), 1.42 (s, 3H). ¹H-NMR (dioxane-d₈, 500 MHz) δ 8.22 (s, 0.89H), 8.09 (s, 0.11H), 4.37–4.45 (m, 1H), 3.71 (s, 0.45H), 3.65 (s, 2.55H), 2.12–2.23 (m, 1H), 1.84–1.93 (m, 1.8H), 1.75–1.83 (m, 1.2H), 1.46–1.49 (2 singlets, 3H total), 1.35–1.38 (2 singlets, 3H total). ¹³C-NMR (dioxane-d₈, 125 MHz) δ 173.4, 172.5, 161.2, 159.8, 61.6, 61.0, 52.2, 51.8, 40.6, 40.0, 28.9, 28.8, 27.3, 26.6, 24.9. ¹H-NMR (CDCl₃, 250 MHz) δ 8.30 (s, 0.92H), 8.22 (s, 0.8H), 4.39–4.59 (m, 1H), 3.78 (s, 0.33H), 3.74 (s, 2.67H), 2.15–2.35 (m, 0.96H), 1.78–2.08 (m, 3.04H), 1.54–1.6 (pair of singlets, 3H total), 1.41–1.44 (pair of singlets, 3H total). ¹³C-NMR (CDCl₃, 62.9 MHz) δ 171.8, 161.2, 160.0, 60.5, 58.1, 52.0, 39.3, 29.1, 28.7, 26.3. ¹³C NMR DEPT-135 δ 160.0, 58.1, 52.0, 29.1, 28.7 are positive (CH or CH₃). δ 39.3, 26.3 are negative (CH₂). MS (ESI) m/z 208.1 (M + Na⁺), 160.1.

Determination of K_trans/cis values

cis and trans resonances of amides 1–3 were identified by using NOESY1D spectroscopy as described in the supporting information. Values of K_trans/cis were determined by integration of ¹H NMR spectra obtained at 25°C.

Depiction of natural bond orbitals

The n and π* NBOs of the trans isomer of AcProOMe (2) in its C^γ-exo conformation were depicted in Figure 1B ▶ with the program gOpenMol, version 2.2 (Laaksonen 1992; Bergman et al. 1997) by using NBO 4.0 output from the B3LYP/6-311+G(2d,p) level of theory (DeRider et al. 2002).

Calculation of O _i−1· · · C_i′ distances and O_i′· · · C_i′=O_i angles

The φ (C′_i−1–N_i–C^α_i–C′_i) and ψ (N_i–C^α_i–C′_i–N_i+1) torsion angles in the trans (ω = 180°) isomer of AcGlyNH₂ were varied in 5°-increments before energy minimization with the program MacSpartanPro, version 1.0.4 (Wavefunction). The O_i−1· · · C′_i distance (δ_BD) and O_i−1· · · C′_i=O_i angle (τ_BD) were recorded in each energy-minimized structure. Those structures with O_i−1· · · C′_i distances δ_BD ≤ 3.2 Å and O_i−1· · · C′_i=O_i angles 99° ≤ τ_BD ≤ 119° were depicted in Figure 3A ▶ on a Ramachandran plot (Ramachandran and Sasisekharan 1968; Richardson 1981).

Electronic supplemental material

NOESY 1D, ¹H-, and ¹³C-NMR spectra and IR spectra are available for amides 1–3.