Trans

J. Phys. Chem. B 2008, 112, 9975–9981

9975

NMR and Theoretical Calculations: A Unified View of the Cis/Trans Isomerization of 2-Substituted Thiazolidines Containing Peptides Helene Jamet,* Muriel Jourdan,* and Pascal Dumy De´partement de Chimie Mole´culaire, UMR-5250, ICMG FR-2607, CNRS, 301 rue de la Chimie, 38041 Grenoble Cedex 9, France ReceiVed: December 19, 2007; ReVised Manuscript ReceiVed: May 20, 2008

The cis/trans isomerization of peptides containing the pseudoproline (4R)-thiazolidine-4-carboxylic acid Cys(ΨR1,R2 pro) is investigated from both an experimental and a theoretical point of view by NMR and DFT calculations. A series of Ac-Cys(ΨR1,R2 pro)-OCH3 and Ac-Cys(ΨR1,R2 pro)-NHCH3 peptides were prepared to assess the influence of the substitution at the C2 position as well as of the amide following the thiazolidine residue. For each compound, the cis/trans ratio along with free energy, the puckering of the thiazolidine ring and the free rotational energy barrier are reported and discussed. We observe there is a pronounced effect of the C2 substituents and of the chirality upon the cis/trans ratio with the population of the cis content in the order (2R)-Cys(ΨCH3,H pro) < (2S)-Cys(ΨH,CH3 pro) < Cys(ΨCH3,CH3 pro). This is interpreted as a consequence of the steric effects imposed by the 2-methylation. For the dimethyl-substituted compounds our results revealed a destabilization of the trans conformers along with a lowering of the rotational trans to cis barrier. We also demonstrate that the introduction of methyl substituents on the thiazolidine cycle reduces the autocatalysis effect, and in particular that no autocatalysis occurs for the dimethyl derivative. A good agreement is obtained between all our experimental observations and our DFT calculations. It rationalizes the main effects which tailor the cis/trans isomerization. Introduction The structural peculiarity of the proline residue is recognized to have a profound impact in peptide and protein structure and its prevalence in protein sequences can be interpreted as an important means in nature to switch and direct peptide chains into favorable topologies.1 The small free energy difference (∆rG ≈ 2.1-10.5 kJ mol-1)2,3 between the two Xaa-Pro peptide bond isomers cis and trans combined with the high activation barrier for cis/trans isomerization4–6 (∆rG# ≈ 80-85 kJ mol-1) have provided some rational for the conformational control of the peptide backbone.7–11 For instance, this isomerization is thought to play a key role in the rate-determining step of protein folding7,10,12–17 and many other important biological processes.1,18–22 This central role is further emphasized by the existence of ubiquitous peptidylprolyl cis/trans isomerases (PPIase) that catalyze this isomerization in vitro and in vivo.15,16,18,23,24 Numerous studies on proline itself or analogues have been realized to obtain mechanistic insights for the cis/trans isomerization as well as to alter the cis/trans ratio (for review see refs 17 and 25). Experiments and theoretical calculations have been carried out on these models mainly to identify the conformational preferences, to measure the transition state free energy and kinetics parameters associated to the Xaa-Pro isomerization. Different strategies have been employed to lock the Xaa-Pro bond in cis conformation through directly modifying the proline residu, with alkyl substituents, heteroatomic substitution or even different ring size.17,25–27 In addition to facilitate the study of the cis/trans isomerization, these restrictions on the polypeptide backbone also provides a very useful tool for developing * To whom correspondence should be addressed. Telephone: (+33) 4-7651-43-74 or (+33) 4-76-51-44-30. Fax: (+33) 4-76-51-49-46. E-mail: [email protected] or [email protected].

bioactive peptides to target relevant biological process involving the cis/trans isomerization.19,25,28–31 Toward this end, Mutter et al. have developed thiazolidine compounds called pseudoproline, ΨPro Cys(ΨR1,R2 pro), and drawn the importance of the steric interaction that arises between the C2 position (which corresponds to Cδ in proline) with the preceding residue in controlling the cis/trans equilibrium.32–34 For instance, the cis/trans ratio can be tuned by simply changing the substituent nature and the conformation at position C2. Remarkably, a dimethyl substitution at C2 led to an unprecedent high cis content within the peptide chain along and this has been proposed to be associated with a large specific decrease of the trans to cis barrier.32–34 For this paper, we studied and rationalized the effects that tailor the cis/trans isomerization within thiazolidine containing compounds. We have studied experimentally by NMR and theoretically by DFT calculations, model peptide compounds Ac-Cys(ΨR1,R2 pro)-OCH3 (1a, (2S)2a, (2R)-2a) and Ac-Cys(ΨR1,R2 pro)-NHCH3 (1b, (2S)-2b, (2R)2b) to assess the influence of the substitution at C2 position as well as of the amide following the thiazolidine residue. For each compound, we focused on a complete description of the isomerization process in terms of energy with the aim to investigate the effect of the methylation on the thiazolidine ring. We reported and explained the cis/trans ratio along with the free energy, the puckering of the thiazolidine ring and the free rotational energy barrier. Our work shows a remarkable agreement between experimental and theoretical results. In particular we demonstrate why the Cys(ΨCH3,CH3 pro) is a powerful “tool”: as a weak effect of autocatalysis occurs for this compound, it can be used in any peptide sequence to exclusively constrain the Xaa-Cys(ΨCH3,CH3 pro) in cis conformation.

10.1021/jp7118982 CCC: $40.75  2008 American Chemical Society Published on Web 07/23/2008

9976 J. Phys. Chem. B, Vol. 112, No. 32, 2008

Jamet et al. TABLE 1: Experimental ∆rGtc° Measured from NMR Data in CDCl3 and Theoretical ∆rGtc Calculated in Chloroform (in kJ mol-1) and Percentage of the Cis and Trans Forms Measured at 298 K by NMR and Calculated on All Conformers substrate % cis/trans ∆rGtc° in CDCl3

Figure 1. 2-Substituted thiazolidines containing peptides investigated in this study.

1a (2S)-2a (2R)-2a 1b (2S)-2b (2R)-2b

88/11 45/55 35/65 97/3 75/25 20/80

-5.0 ( 0.1 0.4 ( 0.1 1.2 ( 0.1 -8.6 ( 0.1 -2.9 ( 0.1 3.3 ( 0.1

calculated % ∆rGtc in cis/trans chloroform 98.2/1.8 57.8/42.2 30.2/69.8 88.7/11.3 78.3/21.7 1.7/98.3

-15.3 0.7 4.1 -5.0 -2.8 12.4

Materials and Methods Peptide Synthesis. The monomethyl- or dimethyl-substituted thiazolidines containing compounds (2R)-2a-b, (2S)-2a-b and 1a-b (Figure 1) were synthesized as already described.33 NMR Spectroscopy. Samples of about 2-5 mg were dissolved in 600 µL of CDCl3, and spectra were recorded at 300 MHz and 500 MHz using an Avance Bruker or a Unity+ Varian spectrometer, respectively. Chemical shifts were calibrated with respect to the residual solvent signal. Data were processed using either XWINNMR or VNMR software. The resonances of the cis and trans isomer coexisting in solution were mainly identified in 1D spectra, on the basis of typical pattern expected for a proton involved in either a cis or trans amide bond. Stereospecific assignments for Hβ protons were done by analyzing their NOE cross peak intensities with the HR proton and their 3J(HR-Hβ) coupling constants. The coupling constant 3J(HR-Hβ) has been observed to be greater for the pro-R β than for the pro-S β proton.32 The puckering of the thiazolidine ring has been determined also on the basis of the 3J(HR-Hβ) coupling constant measurement: 3J(HR-Hβ)(pro S) + 3J(HR-Hβ)(pro R) < 9 Hz or >12 Hz is known to be characteristic of a βexo and γexo conformation, respectively.32 1D and NOESY spectra recorded at different fields, different mixing time τm and various temperatures allowed us to measure the cis and trans molar fraction (xc and xt) by integration of the characteristic proton on C2 carbon or HR proton and also the exchange (Ict ) Itc) and diagonal peak (Icc and Itt) intensities. The calculation of the kinetics and thermodynamics parameters was then performed as previously described33 using Eyring equations and equations below.

ktc )

kct )

(

2xc Ict arctanh τm xtIcc + xcItt

(

2xt Ict arctanh τm xtIcc + xcItt

)

)

Keq )

ktc kct

∆rG ° ) -RT ln(Keq)

( )

∆rGtc# ) RT ln

kbT hktc

( )

∆rGct# ) RT ln

kbT hkct

Computational Calculations. Geometry optimizations for the trans, cis conformers and the transition states in chloroform were performed with the B3LYP method implemented in the Gaussian-03 package35 using the CPCM method36 with a dielectric constant of 4.9 at 298.15K. The 6-31g** basis set was used for all atoms. To compare our theoretical and experimental data, βexo and γexo conformations were chosen as starting points for geometry optimization of the trans and cis conformers. The backbone torsion angles φ and ψ were -80° and -179° for all the compounds of the ester series and were

taken from the crystallographic structure of a dimethylsubstituted thiazolidine compound in the CSD.37 The compounds of the amide series were then constructed starting from the compounds of the ester series. The torsion angles φ and ψ were -80° and 0°, respectively, to allow a favorable interaction between N2H and N1 in the cis form or the presence of an hydrogen bond between N2H and C7dO2 in the trans forms. The cis and trans paths were analyzed for the twist around the imide bond, clockwise or counterclockwise. Vibrational frequencies were calculated for fully optimized conformations with the same functional and basis set than the geometry optimizations. They were used to compute the Gibbs free energy changes for the cis/trans isomerization with the scale factor of 0.89 at 298.15 K and 1 atm in chloroform. Each transition state was characterized by one imaginary frequency after frequency calculations. To compare our results to NMR data, the percentage and the puckering of the cis and the trans forms have been computed by taking into account all the conformers with their normalized Boltzmann weights obtained from their relative free energies. Results and Discussion The ΨPro containing peptides studied by NMR and calculations are displayed in Figure 1. To assess the influence of the substitution as well as the stereochemistry at position 2 of the thiazolidine ring to the cis/trans process, N-acetylated C2 diand monosubstituted thiazolidines 1a-b, (2R)-2a-b and (2S)-2a-b were prepared as model peptides. In addition, methyl ester or methyl amide (series a and b respectively) were used as C-terminal capping to investigate the influence of the peptide bond to the cis/trans equilibrium (autocatalysis effect). Stability of the Cis and the Trans Forms. We investigated in details the cis/trans equilibrium and the puckering of the different compounds. We report in Table 1, the free energies of the trans / cis process as determined by NMR measurements in CDCl3 and calculated in chloroform along with the proportion of the trans and cis forms. For the thiazolidine ring, two conformations βexo and γexo were considered for computational studies (βexo and γexo are also usually denoted “DOWN” and “UP” respectively). β and γ indicate the ring atom considered in the proline IUPAC numerotation. They correspond respectively to the C5 carbon atom and the sulfur atom of the thiazolidine (Figure 2). In an exo conformation, these atoms and the CO group (C6 and O1 atoms) are located on both sides of the plane defined by C2, N1 and C4 atoms (Figure 2). These two conformations were chosen as starting points for geometry optimizations of the cis and trans forms. The resulting minima given in Table 2 are called either minimum 1 for the optimization with the βexo or minimum 2 with the γexo puckering,

Cis/Trans Isomerization of 2-Substituted Thiazolidines

Figure 2. Definition of the torsions angles χ4 and χ5 and atoms labeling.

TABLE 2: Relative Gibbs Free Energies (in kJ mol-1), Normalized Boltzmann Weights (%) and Some Optimized Geometric Parameters (in Å) for the Different Conformers of the Cis and Trans Formsa cis conformer

trans conformer

minimum minimum minimum minimum 1 2 1 2 ∆ rG % d(C8-C4) d(C8-C6) d(C8-CR1) d(C8-CR2) (2S)-2a ∆rG % d(C8-C4) d(C8-C6) d(C8-CR2) (2R)-2a ∆rG % d(C8-C4) d(C8-C6) d(C8-CR1) 1b ∆rG % d(C8-C4) d(C8-C6) d(C8-CR1) d(C8-CR02) (2S)-2b ∆rG % d(C8-C4) d(C8-C6) d(C8-CR2) (2R)-2b ∆rG % d(C8-C4) d(C8-C6) d(C8-CR1)

1a

0.0 95 2.89 3.39 b b 0.0 57.7 2.93 3.30 b 2.4 15.1 2.88 3.58 b 0.0 86.9 2.89 3.50 b b 0.0 78.2 2.92 3.37 b 10.2 1.5 2.94 3.73 b

8.4 3.2 2.91 2.98 b b 13.8 0.1 2.95 2.95 b 2.4 15.1 2.92 3.18 b 9.6 1.8 2.89 3.09 b b 15.9 0.1 2.91 3.02 b 13.5 0.4 2.90 3.31 b

10.5 1.3 b b 3.41 3.23 0.8 41.8 b b 3.40 0.7 30.0 b b 3.52 5.6 9.0 b b 3.41 3.16 3.2 21.5 b b 3.27 12.1 0.7 b b 3.47

13.1 0.5 b b 3.62 3.06 12.3 0.4 b b 3.28 0.0 39.8 b b 3.63 9.0 2.3 b b 3.82 2.99 15.5 0.2 b b 3.02 0.0 97.4 b b 3.82

a Geometry optimizations starting with a βexo puckering give the minimum called minimum 1, those with a γexo puckering lead to minimum 2. b Not measured.

respectively. Their relative Gibbs free energies and their percentages calculated in chloroform are also reported in Table 2. Calculations were done by taking account the different conformers with their normalized Boltzmann weights obtained from their relative free energies to have a better comparison with NMR data for which we observe a mixture of the cis and trans forms. Overall, the experimental data are well corroborated with the theoretical results and point out comparable general trends up.

J. Phys. Chem. B, Vol. 112, No. 32, 2008 9977 We observed the highest cis content for all C2 disubstituted compounds 1a-b whereas the stereochemistry at position C2 modulated the cis/trans ratio of monosubstituted compounds 2ab. On the whole, the experimental ∆rGtc° and theoretical ∆rGtc increase in the order 1a < (2S)-2a< (2R)-2a and 1b < (2S)2b< (2R)-2b with a much more pronounced cis/trans ratio observed in the amide series compared with the ester series. This is emphasized by a net preference for the trans form for (2R)-2b whereas for (2S)-2b and 1b the cis form predominates experimentally and theoretically. In general, steric interactions between the CR pyrrolidine atom and the preceding residue provide some rational for the cis/ trans ratio found in proline-containing peptides and proteins.38 These interactions that are important in the cis conformation are released in the trans conformer and explained the high percentage of the trans amide bond. In ΨPro-containing model peptides, two steric interactions have to be considered. The side chain Ri-1 or the peptide backbone of the preceding residue still interacts with the C4 atom of the ring in the cis form whereas in the trans form the corresponding interactions stem from the C2 atom of the ring. In our study, the C2 atom of the cycle is substituted. Thus to point out the steric effects imposed by this substitution, the calculated distances d(C8-CR1) and d(C8-CR2) exhibited between acetyl and C2-methyl groups are reported for the different trans forms whereas for the cis forms it is the calculated distances d(C8-C4) (Table 2). We should find the shortest distances for the less abundant rotamer. Inspection of these distances revealed that in all cis forms d(C8-C4) is almost constant (around 2.91 Å). On the contrary, more variations are observed for d(C8-CR1) and d(C8-CR2) in the trans form, distances ranging from 2.99 up to 3.82 Å. This indicates that steric interactions arising from the trans form control the cis/trans equilibrium. In both series, the C2 disubstituted compounds 1a-b present two short distances d(C8-CR1) and d(C8-CR2). For example, for 1a, values are equal to respectively 3.41 and 3.23 Å for the minimum 1 and 3.62 and 3.06 Å for the minimum 2 (see Table 2). These interactions explain the most important destabilization of their trans forms compared to their monosubstituted analogues compounds for which only one interaction occurs. The smaller values found for d(C8-CR2) for the (2S) trans form compared to d(C8-CR1) for the (2R) trans form indicate a higher destabilization for the (2S) trans form than the (2R) trans form (see in Table 2 for example values of 3.40 and 3.28 Å for the trans conformers of (2S)-2a versus 3.52 and 3.63 Å for the trans conformers of (2R)2a). Thus steric effects imposed by the 2-methylation explain the cis/trans ratio observed in the two series. Similar conclusions have been reported by Kang on mono- and dimethylproline compounds.39 For the amide series either a favorable interaction between N2H and N1 in the cis form or the presence of an hydrogen bond between N2H and C7dO2 in the trans forms helps to understand the discrepancies observed between ester and amide. These intramolecular interactions have been already described by others4,40 and are illustrated in Figure 3 for the dimethyl compound 1b. From the fact that for the cis form of the two series, steric interactions are equivalent (d(C8-C4) almost constant), we can conclude that the attractive nonbond interaction between N2H and N1 induces a stabilization of the cis forms in the amide series. For the trans forms, the situation is more complex. For (2R)2b the hydrogen bond and steric interactions both stabilize the trans form (see the value of d(C8-CR1) ) 3.82 Å for the most

9978 J. Phys. Chem. B, Vol. 112, No. 32, 2008

Figure 3. (a) Trans conformer for 1b. (b) Cis conformer for 1b. Dashed lines represent the hydrogen bond in the trans form and the favorable interaction between N2H and N1 in the cis form.

stable minimum 1 of (2R)-2b compare to 3.63 and 3.52 Å for the two relatively close trans conformers of (2R)-2a). This is not the case for 1b and (2S)-2b for which we measure a decrease of d(C8-CR2) in the trans form compared to their analogues 1a and (2S)-2a. Value of d(C8-CR2) equals 3.27 Å for the most stable trans form of (2S)-2b whereas it is 3.40 Å for (2S)-2a. In fact, steric effects come into conflict with the stabilization induced by the hydrogen bond in the trans amide form. This suggests a more important stabilization of the trans form for (2R)-2b than 1b and (2S)-2b and explains the different cis/ trans ratio observed in the amide series compared with the ester series. Puckering. We report in Table 3 an average of the optimized torsion angles obtained for the different conformers and the puckering measured in solution. A βexo and γexo puckering correspond to a torsion angle χ4 (respectively χ5) (Figure 2) close to zero. Experimentally, the puckering of the thiazolidine ring was determined by measuring the sum 3J(HR-Hβ)(proS) + 3J(HR-Hβ)(proR), which has been proved to be a powerful tool to find out the puckering thiazolidine ring.32 For 1a and 1b, it was not possible to measure the coupling constants on the minor trans resonances due to the very broad and small signal. As NMR measures a mean puckering between the theoretical minimum 1 and 2 of the cis and trans forms, theoretical results are given in weighting the optimized torsion angle obtained for minimum 1 and minimum 2 in the cis and trans forms by their proportions for a better comparison. In the ester series, experimental results show a βexo puckering for all the cis conformers. Theoretical results lead to the same conclusion except for (2R)-2a for which the puckering is less pronounced. In the trans form, the βexo puckering of 1a is only determined by theoretical results. For (2S)-2a and (2R)-2a optimized values of χ4 and χ5 do not allow us to clearly define the puckering but NMR data indicate respectively a βexo and γexo puckering. For the cis conformation of the amide series, our experimental results show the same trend as above, all the compounds having a βexo conformation. Calculations are also in favor of a βexo conformation, except for (2S)-2b for which χ4 and χ5 are rather similar. In the trans conformation, 1b compound has a βexo conformation. For (2S)-2b it is still theoretically difficult to define the puckering based only on angle values but experimental data indicate a conserved βexo conformation. (2R)-2b has a γexo conformation. For this compound in both series a change of conformation from βexo to γexo occurs during the cis/trans isomerization process whereas for 1a-b and (2S)2a-b the puckering is conserved. To explain the puckering found for the cis and trans forms, we compared the stability of the different conformers. Their corresponding Gibbs free energies are reported in Table 2 with

Jamet et al. some geometric parameters that illustrate steric interactions. These are for the trans forms the distances d(C8-CR1) and d(C8-CR2) discussed before, and for the cis form the distance d(C8-C6), which corresponds to the distance between the acetyl and the carbonyl group (C6dO1 in Figure 2). In fact, the two conformations (βexo and γexo) are linked to the position of the CO group, which in the cis form is close to the acetyl group causing steric conflicts. For all the cis compounds the most stable minimum has the longer distance d(C8-C6) except for the cis conformers of (2R)-2a for which minima have the same energy (see Table 2). Analysis of d(C8-CR1) and d(C8-CR2) show that for all (2S)-2a-b (respectively (2R)-2a-b) trans forms, the most stable minimum has the longer distance d(C8-CR2) (respectively d(C8-CR1)) (see Table 2). For 1a-b trans forms, the distance d(C8-CR2) is shorter than d(C8-CR1). Thus d(C8-CR2) imposes the most stable minimum (minimum 1 of 1a with d(C8-CR2) equal to 3.23 Å versus 3.06 Å for minimum 2; minimum 1 of 1b 3.16 Å versus 2.99 Å for minimum 2). Thus, steric interactions can explain the βexo or γexo puckering obtained for the different compounds. Transition State. Experimental and calculated rotational free energies ∆rG#tc and ∆rG#ct have been obtained for all the compounds (Table 4–7). Theoretical calculations were done by considering only the most preferred cis and trans conformers. First, we consider 1a and (2S)-2a for which there is no change of puckering during the cis/trans isomerization. The puckering being fixed, two conformations for the transition states were possible. They correspond to a twist about the imide bond itself which can be clockwise or counterclockwise.4 Geometry optimizations were performed on the two conformers. For the two products, the clockwise path is found to be the more favorable, the transition state obtained with the counterclockwise path being destabilized in good agreement with recent studies of Kang et al.27,41,42 This can also be explained by steric interactions induced by the 2-methylation, i.e., by monitoring distances d(C8-CR1) and d(C8-CR2). For example, for the 1a compound, in the transition state resulting from the clockwise path, the distances d(C8-CR1) and d(C8-CR2) are computed to be 4.34 and 3.77 Å, respectively. In the counterclockwise path, the former distance does not change much and is 4.42 Å but the latter is shortened to 3.46 Å. This suggests a more important destabilization for this transition state. The difference between the two transition state free energies is around 8 kJ mol-1. Similar results are found for (2S)-2a in which only the distance d(C8-CR2) has to be considered. Table 4 gives the activation free energies calculated along the most favorable path and experimental values calculated using Gibbs and Eyring equations (see Materials and Methods). Experimental and calculations results are in good agreement and more particularly for (2S)-2a. We observed a large difference on the free energy barrier ∆rG#tc between the two compounds, which decreases as follows: ∆rG#tc((2S)-2a) > ∆rG#tc(1a). More discrepancies are obtained on ∆rG#ct: NMR data give similar values for the two compounds but more variations are obtained with calculations. To explain these results, we report in Figure 4A the free energy of each compound. For the two compounds the cis conformers are the most stable and thus are taken as the reference state in Figure 4A. The graph illustrates a more important destabilization of the trans form for 1a than for (2S)2a and less variations on the free energies of their transition states. These results can be explained by monitoring the distance d(C8-CR2) and d(C8-CR1). For the transition states, d(C8-CR2) varies from 3.77 Å for 1a to 3.71 Å for (2S)-2a. A larger

Cis/Trans Isomerization of 2-Substituted Thiazolidines

J. Phys. Chem. B, Vol. 112, No. 32, 2008 9979

TABLE 3: Theoretical and Experimental Puckering of the Thiazolidine Rings cis form

trans form puckering

1a (2S)-2a (2R)-2a 1b (2S)-2b (2R)-2b a

χ4a

χ5a

2.9 -7.0 -18.8 -8.7 -16.1 8.2

-28.9 -21.2 20.4 -19.5 -13.1 -21.9

puckering

calculated

experimental

βexo βexo b βexo b βexo

βexo βexo βexo βexo βexo βexo

χ4a

χ5a

9.9 -12.7 24.3 7.0 -16.5 36.7

-21.2 -15.2 -17.9 -24.2 -12.6 -16.7

calculated

experimental

βexo b b βexo b γexo

b βexo γexo b βexo γexo

χ4 and χ5 angles are given in degrees and are computational values. b The puckering could not be determined.

TABLE 4: Experimental and Theoretical Values ∆rG#tc and ∆rG#ct for 1a and (2S)-2a (in kJ mol-1) ∆rG#tc

∆rG#ct

substrate NMR data theoretical data NMR data theoretical data 1a (2S)-2a

68.4 ( 0.3 72.3 ( 0.3

66.9 71.3

73.4 ( 0.3 71.9 ( 0.3

77.4 72.0

deviation occurs for the trans form (3.23 Å for 1a compared with 3.40 Å for (2S)-2a). In parallel, for the 1a compound, d(C8-CR1) is computed to 3.41 Å for the trans form and 4.34 Å for the transition state. All these results highlight the destabilization of the trans form of 1a relatively to (2S)-2a, close free energies for the transition states of the two compounds and explain why the main variation is found on ∆rG#tc. In addition, experimental mean kinetic value of ktc around 7 and 1.3 s-1 for 1a and (2S)-2a, respectively, and kct around 0.9 and 1.6 s-1 indicate that the trans to cis interconversion is very fast for 1a compared to the other interconversions. All these results, linked to the high stability of its cis form, suggest that 1a can be used to successfully constrain an amide bond in the cis conformation. For (2R)-2a, experimental results indicate a change of puckering during the isomerization process (Table 3). Thus for this compound, the reaction path is more complex due to the puckering inversion between the trans and cis conformers. It is known that isomerization and inversion of puckering are dissociated.43 So we considered the two reactions separately and analyzed the paths: cis βexo f cis γexo f trans γexo or cis βexo f trans βexo f trans γexo. A clockwise path is considered for the isomerization step. The relative free energies of transition states are given in Table 5. In the two cases considered here, isomerization is found to be the limiting step. To compare these results with experimental values, we have to consider the cis γexo f trans γexo path of our calculation. Indeed, for this compound, experimentally, the most stable form is the trans conformation with a γexo puckering. Moreover, because the above-mentioned results show isomerization as the limiting step, it is thus justified to take only into account the cis γexo f trans γexo reaction. Experimental values of the global reaction are given in Table 5 and are similar to the theoretical values of the isomerization path. Finally, we considered the amide series. We limited our study to the two compounds (2S)-2b and 1b because they have a higher propensity to constrain the imidic bond in the cis form compared to the (2R)-2b compound. For this latter, we have demonstrated that the trans form is highly stabilized. For these two compounds, minimum 1 is the most stable in the cis and trans forms. Thus we looked only at the reaction cis βexo f trans βexo path. Table 6 gives theoretical and experimental values of the free activation energies.

Experimental and calculations results show similar trends. In this case, the larger variation is found on ∆rG#ct, which decreases as follow: ∆rG#ct(1b) > ∆rG#ct((2S)-2b). The free energy of each compound is reported in Figure 4B. The transition state is more destabilized for the disubstituted compound 1b than for (2S)-2b. The distance d(C8-CR2) (4.35 Å for 1b and 3.70 Å for (2S)-2b) cannot explain this result. This is rather the distance d(C8-C6), which makes a difference between the two transition states, the calculated value being 3.77 Å for 1b and 4.51 Å for (2S)-2b. The ∆rG#ct barrier is more important for 1b than for (2S)-2b, suggesting a slower cis to trans interconversion for the 1b compound than for (2S)-2b. This is confirmed by experimental kct values of 1 s-1 for 1b and 5 s-1 for (2S)-2b. Thus within the amide series, the dimethylthiazolidine 1b is also the best candidate to lock an amide bond in a cis conformation. Autocatalysis. The study of the two series of compounds with different C-terminal ends (methyl ester, series a, and methyl amide, series b) provides a mean to assess if the isomerization process is sensitive to the nature of the C-terminal residue (autocatalysis effect). As we have shown previously within the amide series, a favorable interaction occurs between the hydrogen of the N2H group and the lone pair of the imide nitrogen N1 in the cis form (Figure 3). Similar interaction exists in the transition state but in addition there is also a pyramidalization of the nitrogen. This deformation increases the interaction and an hydrogen bond is formed. This has been proposed to accelerate the cis to trans isomerization by stabilizing the transition state and lowering the rotational energy barriers.4,26,27,39,42,44,45 It is indeed what we observe in Figure 4. A good estimation of this effect is the free energy barrier difference ∆∆rG#cat ) ∆rG#(ester) - ∆rG#(amide) between the ester and amide compounds reported in Table 7. Experimentally, a weak autocatalysis effect of the interconversion cis to trans in CDCl3 occurs for the disubstituted compound (1a and 1b) (weak value of ∆∆rG#ct cat ) 0.4 kJ mol-1). This contrasts with the high ∆∆rG#ct cat values of 9.2 kJ mol-1 found for proline in the litterature.44 The value obtained for the monosubstituted 2S compound is intermediate. Thus, it can be conclude that the 2-methylation reduces the autocatalysis effect on the cis to trans process. This result is less visible on the interconversion trans to cis. Values are comparable to those found for the proline (5.8 kJ mol-1).44 However, in the case of 1a-b, the large value obtained for ∆∆rG#tc cat has to be considered with caution, the compound 1b being mainly in the cis conformation (97%). Thus, our value can be slightly inaccurate due to the difficulty in measuring broad and also very small NMR peaks for this compound. Theoretical calculations also show weaker values for the dimethyl compound (1a-b) compared with the S-monosubstituted ((2S)-2a-b). It illustrates the fact that the stabilization of the

9980 J. Phys. Chem. B, Vol. 112, No. 32, 2008

Jamet et al.

Figure 4. Theoretical free energies for 1a and (2S)-2a (A) and for 1b and (2S)-2b (B). Studies were done by considering only the most stable minimum for the cis and trans forms.

TABLE 5: Theoretical and Experimental Free Activation Energies of the Cis/Trans Isomerization for (2R)-2a (in kJ mol-1) isomerization ‡

∆rG

isomerization path cis βexo f trans βexo f trans γexo cis βexo f cis γexo f trans γexo cis f trans (CDCl3)

ct

puckering ‡

∆rG

tc

∆rG‡βγ

∆rG‡γβ

72.4

74.0

8.9

9.6

73.2

75.4

14.3

9.9

69.0 ( 0.3

70.2 ( 0.3

Conclusion

TABLE 6: Theoretical and Experimental Free Activation Energies of the Cis/Trans Isomerization for 1b and (2S)-2b (in kJ mol-1) ∆rG#tc

∆rG#ct

substrate NMR data theoretical data NMR data theoretical data 1b (2S)-2b

64.4 ( 0.3 66.0 ( 0.3

63.3 59.0

73.0 ( 0.3 69.0 ( 0.3

68.9 62.2

TABLE 7: Theoretical and Experimental Autocatalytic Effect ∆∆rG# cat (in kJ mol-1) ∆∆rG#tc cat substrate

∆∆rG#ct cat

NMR data theoretical data NMR data theoretical data

1a-b 4.0 ( 0.3 (2S)-2a-b 6.3 ( 0.3

3.6 12.2

This highlights the reduction of the autocatalytic effect by the disubstitution. This reduction proves that the efficiency of disubstituted thiazolidine is not “amino-acid” sequence dependent. Our experimental and theoretical results both demonstrate that 1b can be used as a “universal” mimetics of proline and introduced in any peptide sequence without affecting its powerful properties to lock the peptide bond in cis conformation.

0.4 ( 0.3 2.9 ( 0.3

8.5 9.8

transition state of 1b relatively to the transition state of 1a is less important than for (2S)-2b relatively to (2S)-2a. Steric interactions explain this result. Values of d(C8-C6) in 1a-b transition states are equal respectively to 3.77 Å for 1b and 4.30 Å for 1a. This steric interaction reduces the stabilization of the transition state of 1b induced by the hydrogen bond and consequently a decrease of the autocatalytic effect is observed. For (2S)-2a-b transition states these distances are similar and larger (4.51 Å for (2S)-2b and 4.47 Å for (2S)-2); as a consequence, the autocatalyic effect is more present. We also did similar studies on thiazolidines containing peptides without methyl substitution. Conformers and transition states were calculated for the ester and amide series. We found higher values of ∆∆rG#tc cat ) 14.5 kJ mol-1 and ∆∆rG#ct cat ) 17.8 kJ mol-1.

We reported a unified view (from an experimental and theoretical point of view) of the cis/trans isomerization process of 2-substituted thiazolidines containing peptides. To our knowledge, it is the first study combining experiments and theory in which a dimethyl-substituted thiazolidine compound is compared to its equivalent monomethyl-substituted. We focused on a complete description of the isomerization process in terms of energy with the aim to investigate the effect of the methylation on the thiazolidine ring. Our experimental and theoretical results are in good agreement. The higher cis content for a disubstituted compound (1) than for a monosubstituted one (molecules (2S)-2 and (2R)-2) can be interpreted as a consequence of the steric effects imposed by the 2-methylation, i.e., by monitoring the distances d(C8-CR1) and d(C8-CR2). These interactions, added to steric conflicts between acetyl and carbonyl groups, allowed us to explain the puckering measured for all the conformers and the change observed for some systems between the trans and cis conformers during the isomerization process. We performed a complete analysis of the reaction path of the cis/trans imide isomerization. A change of puckering is illustrated for the monosubstituted R compound (2R)-2a. For the dimethyl-substituted compounds 1a and 1b, our results revealed a destabilization of the trans conformer along with a lowering of the rotational trans to cis barrier. We also demonstrate that the introduction of methyl substituents on the thiazolidine cycle reduces the autocatalysis effect. All together, our results clearly show that the 1b compound can be use in any peptide sequence to exclusively lock the peptide bond in a cis conformation.

Cis/Trans Isomerization of 2-Substituted Thiazolidines Acknowledgment. We thank Roger Arnaud for calculations at the early stage of the study, the CECIC center for providing computer facilities, Didier Boturyn and Sabine Chierici for the synthesis of the compounds, Anne Milet for fruitful discussions and one referee for theoretical advice. Supporting Information Available: Cartesian coordinates of all optimized conformations. This information is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Andreotti, A. H. Biochemistry 2003, 42, 9515, and references herein. (2) Grathwohl, C.; Wuthrich, K. Biopolymers 1976, 15, 2025. (3) Maigret, B.; Perahia, D.; Pullman, B. J. Theor. Biol. 1970, 29, 275. (4) Fischer, S.; Dunbrack, R. L. J.; Karplus, M. J. Am. Chem. Soc. 1994, 116, 11931. (5) Drakenberg, T.; Forsen, S. Chem. Commun., J. Chem. Soc. D 1971, 21, 1404. (6) Kern, D.; Schutkowski, M.; Drakenberg, T. J. Am. Chem. Soc. 1997, 119, 8403. (7) Jabs, A.; Weiss, M. S.; Hilgenfeld, R. J. Mol. Biol. 1999, 286, 291. (8) Stewart, D. E.; Sarkar, A.; Wampler, J. E. J. Mol. Biol. 1990, 214, 253. (9) MacArthur, M. W.; Thornton, J. M. J. Mol. Biol. 1991, 218, 397. (10) Fischer, G. Chem. Soc. ReV. 2000, 29, 119. (11) Pal, D.; Chakrabarti, P. J. Mol. Biol. 1999, 294, 271. (12) Brandts, J. F.; Halvorson, H. R.; Brennan, M. Biochemistry 1975, 14, 4953. (13) Kim, P. S.; Baldwin, R. L. Annu. ReV. Biochem. 1982, 51, 459. (14) Wedemeyer, W. J.; Welker, E.; Scheraga, H. A. Biochemistry 2002, 41, 14637. (15) Schmid, F. X.; Mayr, L. M.; Mu¨cke, M.; Scho¨nbrunner, E. R. AdV. Protein Chem. 1993, 44, 25. (16) Stein, R. L. AdV. Protein Chem. 1993, 44, 1, and references therein. (17) Dugave, C.; Demange, L. Chem. ReV. 2003, 103, 2475, and references therein. (18) Fischer, G. Angew. Chem., Int. Ed. Engl. 1994, 33, 1415. (19) Tugarinov, V.; Zvi, A.; Levy, R.; Anglister, J. Nat. Struct. Biol. 1999, 6, 331. (20) Chen, J. K.; Schreiber, S. L. Angew. Chem., Int. Ed. Engl. 1995, 34, 953. (21) Lummis, S. C. R.; Beene, D. L.; Lee, L. W.; Lester, H. A.; Braodhurst, R. W.; Dougherty, D. A. Nature 2005, 438, 248.

J. Phys. Chem. B, Vol. 112, No. 32, 2008 9981 (22) Lu, K. P.; Finn, G.; Lee, T. H.; Nicholson, L. K. Nature Chem. Biol. 2007, 3, 619. (23) Schmid, F. X. Annu. ReV. Biophys. Biomol. Struct. 1993, 22, 123. (24) Fischer, G.; Bang, H.; Mech, C. Biomed. Biochim. Acta 1984, 43, 1101. (25) Che, Y.; Marshall, G. R. Biopolymers 2006, 81, 392. (26) Jhon, J. S.; Kang, Y. K. J. Phys. Chem. B 2007, 111, 3496. (27) Kang, Y. K.; Byun, B. J. J. Phys. Chem. B 2007, 111, 5377. (28) Chierici, S.; Jourdan, M.; Figuet, M.; Dumy, P. Org. Biomol. Chem. 2004, 2, 2437. (29) Garcia, J.; Dumy, P.; Rosen, O.; Anglister, J. Biochemistry 2006, 45, 4284. (30) Keller, M.; Boissard, C.; Patiny, L.; Chung, N. N.; Lemieux, C.; Mutter, M.; Schiller, P. W. J. Med. Chem. 2001, 44, 3896. (31) Wittelsberger, A.; Keller, M.; Scarpellino, L.; Patiny, L.; AchaOrbea, H.; Mutter, M. Angew. Chem., Int. Ed 2000, 39, 1111. (32) Dumy, P.; Keller, M.; Ryan, D. E.; Rohwedder, B.; Wo¨hr, T.; Mutter, M. J. Am. Chem. Soc. 1997, 119, 918. (33) Keller, M.; Sager, C.; Dumy, P.; Schutkowski, M.; Fischer, G. S.; Mutter, M. J. Am. Chem. Soc. 1998, 120, 2714. (34) Mutter, M.; Wo¨hr, T.; Gioria, S.; Keller, M. Biopolymers: Peptide Sci. 1999, 51, 121. (35) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A.; Stratmann, R. E., Jr.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.; Petersson, G.; Ayala, A. P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz, J. V.; Baboul, A. G.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Andres, J. L.; Gonzalez, C.; Head-Gordon, M.; Replogle, E. S.; Pople, J. A. Gaussian 03, revision A.1; Gaussian, Inc.: Pittsburgh, PA, 2003. (36) Barone, V.; Cossi, M. J. Phys. Chem. A 1998, 102, 1995. (37) Porter, N. A.; Rosenstein, I. J.; Breyer, R. A.; Bruhnke, J. D.; Wu, W. X.; McPhail, A. T. J. Am. Chem. Soc. 1992, 114, 7664. (38) Zimmerman, S. S.; Scheraga, H. A. Macromolecules 1976, 9, 408. (39) Kang, Y. K. J. Mol. Struct. (THEOCHEM) 2002, 585, 209. (40) Kang, Y. K. J. Phys. Chem. B 2002, 106, 2074. (41) Kang, Y. K. J. Phys. Chem. B 2006, 110, 21338. (42) Kang, Y. K.; Park, H. S. J. Phys. Chem. B 2007, 110, 12551. (43) Kang, Y. K.; Choi, H. Y. Biophys. Chem. 2004, 111, 135. (44) Cox, C.; Lectka, T. J. Am. Chem. Soc. 1998, 120, 10660. (45) Rankin, K. N.; Boyd, R. J. J. Phys. Chem. A 2002, 106, 11168.

JP7118982