Influence of Solvents upon Diketopiperazine Formation of FPG₈K

Abstract

Ion mobility spectrometry (IMS) and mass spectrometry (MS) techniques were used to monitor diketopiperazine (DKP) formation from the peptide FPG₈K at multiple defined temperatures in methanol, ethanol, propanol, and water, with the motivation to study the effect of solvent polarity on spontaneous solution dissociation. The reaction rate increases with decreasing solvent polarity. The observed rates of trans → cis isomerization of Phe¹–Pro² and the cis-Pro² isomer dissociation result in the cis isomer growing in abundance relative to the trans isomer throughout the reaction in all solvents. Analysis of rate constants derived from the data using a sequential unimolecular kinetics model that includes hidden intermediate states yields transition state thermodynamic values for both trans → cis isomerization of Phe¹–Pro² and dissociation. The measured thermochemistry appears to be closely correlated with these solvents’ dielectric constants: a lower solvent dielectric constant accelerates the reaction by reducing the enthalpic barrier, albeit with slight entropic restriction.

Graphical Abstract

INTRODUCTION

Proteins and peptides can decompose during synthesis and storage through several spontaneous chemical reactions.^1,2 One of these reactions is diketopiperazine (2,5-dioxopiperazine, DKP) formation, which involves an N-terminal amine nucleophilic attack on the carbonyl carbon between the second and third amino acid residues, breaking the peptide chain and forming a cyclic peptide composed of the first two residues from the N-terminus and a truncated sequence of the peptide.³⁻⁶ DKP formation happens especially when there is a proline residue at the second position from the N-terminus (i.e., penultimate proline). There are a few reports on this spontaneous degradation, including the dissociation of recombinant DNA-derived human growth hormone (rhGH) resulting in DKP and a truncated variant of rhGH;⁷ degradation of the eleven-residue neuropeptide substance P in solution and solid forms;⁸ and a recent report about self-cleavage of bradykinin at elevated temperature, which involves breaking the bond that is the most difficult to cleave enzymatically.⁹

Understanding the DKP formation reaction is of importance for other reasons as well. For example, it is a way of generating biologically active species throughout the body, such as cyclo(His–Pro).¹⁰ Also, preventing intramolecular aminolysis is important for the storage of pharmaceutically important proteins and peptides since there are reports of decomposition of such compounds.^11,12 In addition, a recent finding by the Hunt group shows that a variety of MHC class I-associated phosphorylated peptides cause immune response.¹³ These peptides have the potential to be used as novel vaccines for immunotherapy. However, many of these peptides contain a penultimate proline, which can lead to DKP formation and compromise the function of these peptides.

Several reports show that acids and bases, as well as buffer species, can influence the DKP formation rate in aqueous solution.^6,12,14 However, few reports analyzed the solvent effects on the DKP formation rate.^15,16 Also, the exact relationship between the rate of trans → cis isomerization of Xaa¹–Pro² and DKP formation has not been established. This is largely because almost all the previous research utilized techniques that were not able to separate the cis and trans isomers. The ion mobility spectrometry (IMS)–mass spectrometry (MS) technique has proven to be a useful tool to separate and analyze different conformations based on differences in molecules’ overall mobility due to different collision cross sections.^{17, 18} Recent work used IMS–MS to analyze the dissociation kinetics of bradykinin and substance P,^9,19 in which different isomers due to proline trans → cis isomerization of the Xaa¹–Pro² bond are monitored during the degradation process. Analyzing the isomers from each species during dissociation provided evidence that at least three and four intermediates are involved in spontaneous solution dissociation of bradykinin and substance P, respectively. In each case, a configurationally coupled protonation process is observed as the penultimate proline that undergoes trans → cis isomerization, similarly to a previous Pro7 study.²⁰ Here, the peptide FPG₈K is used as a model to analyze DKP formation kinetics in four solvents at multiple defined temperatures. FPG₈K was selected because the conformations resulting from penultimate proline trans → cis isomerization have sufficiently different collision cross sections to be separable by IMS, thus enabling easy monitoring of changes in the cis/trans ratio of proline during degradation. Transition state thermochemistry values are also calculated for isomerization in all four solvents and for dissociation in three of the four solvents. This thermochemistry appears to be closely correlated with these solvents’ dielectric constants. The implications of this are discussed below.

METHODS

Peptide Synthesis.

Peptides are synthesized according to a traditional Fmoc solid-phase peptide synthesis procedure, as detailed elsewhere.²¹ Briefly, 0.01 mmol Fmoc-Lys(Boc)-Wang resin is deprotected using piperidine/DMF (20% by volume). Then, the second amino acid from the C-terminus is activated using 3-(diethoxyphosphoryloxy)-1,2,3-benzotriazin-4(3H)-one (DEPBT) and N,N-diisopropylethylamine (DIEA) before mixing with the resin in 10 mL of DMF for ∼1.5 h. After mixing, the resulting product is washed with DMF, DCM, and MeOH three times each. This procedure is repeated until the peptide reaches the desired length. Cleavage from the resin is performed with TFA/H₂O/triisopropylsilane (95%/2.5%/2.5% V/V/V). The peptide is precipitated with diethyl ether and purified with HPLC.

Sample Preparation and Kinetics Experiment.

A detailed description can be found in a previous paper.⁹ Briefly, the peptide is dissolved into pure propanol, ethanol, methanol, or water to make 1 mM stock solutions which are stored at −20 °C. Stock solutions are diluted to 20 μM with 1% acetic acid (by volume) for use. The resulting alcohol solutions were then incubated at 65, 70, and 75 °C and tested periodically by IMS−MS. In water, we were able to monitor isomerization at 75 and 90 °C, but the dissociation process was too slow to be observed.

IMS Instrumentation.

A home-built instrument in which a 2 m drift tube is coupled with a time-of-flight (TOF) mass spectrometer (see Supporting Information, Figure S14) is used to monitor the dissociation. Detailed descriptions of the instrument and theory appear elsewhere.^18,22−29 Briefly, ions are produced by nanoelectrospray in a NanoMate autosampler (Advion, Ithica, NY) and then transferred into the source region. Ions that are stored in the source funnel are periodically pulsed into the drift tube for separation based on their overall structural differences. The drift tube is filled with ∼3.0 torr (∼4.0 mbar) He buffer gas and operated under a weak electric field of roughly 10 V·cm⁻¹. The ions are analyzed using a TOF mass spectrometer after migrating through the 2 m drift tube. It is useful to convert the ion drift time t_D into the collisional cross section (CCS) (Ω, in Ų) according to the following equation²⁵

$$\text{Ω}=\frac{{(18\pi )}^{1/2}}{16}\frac{\text{ ze }}{{\left(k_{\text{b}}T\right)}^{1/2}}{\left[\frac{1}{m_{\text{I}}}+\frac{1}{m_{\text{B}}}\right]}^{1/2}\frac{t_{\text{D}}E}{L}\frac{760}{P}\frac{T}{273.2}\frac{1}{N}$$ (1)

where the relevant terms are ion charge (ze), Boltzmann’s constant (k_b), mass of the ion (m_I), mass of the buffer gas (m_B), temperature of buffer gas (T), electric field value (E), drift tube length (L), pressure of buffer gas (P), and buffer gas neutral number density (N).

RESULTS AND DISCUSSION

Mass Spectral Data for [FPG₈K + 2H]²⁺ Dissociation in Different Solvents.

Figure 1 shows the mass spectra acquired after incubating FPG₈K in n-propanol, ethanol, and methanol (with 1% acetic acid) at 75 °C. In Figure 1a, peptide FPG₈K is incubated in n-propanol. At 0 min, there is one peak at m/z = 425.7 in the mass spectrum, corresponding to [FPG₈K + 2H]²⁺. At 21 min, a new peak appears at m/z = 301.9, representing [G₈K + 2H]²⁺. As time progresses, the [FPG₈K + 2H]²⁺ peak decreases in abundance, while the [G₈K + 2H]²⁺ peak grows in abundance. The degradation process is complete at around 125 min, when only [G₈K + 2H]²⁺ remains. The dissociation half-life is around 48 min. Figure 1b,c shows peptide FPG₈K dissociation mass spectra in ethanol and methanol, respectively. Similar dissociation patterns are observed: as incubation time increases, the initial [FPG₈K + 2H]²⁺ ion disappears, yielding the [G₈K + 2H]²⁺ ion. The dissociation half-lives in ethanol and methanol are 60 and 291 min, respectively. Peptide FPG₈K is also incubated in water. It takes 50 h for the dissociation to begin, which is much longer than in the other three solvents (see Supporting Information, Figure S1). It turns out that the FPG₈K dissociation rate increases with decreasing solvent polarity.

Figure 1.

Peptide FPG₈K dissociation mass spectra in n-propanol (a), ethanol (b), and methanol (c) with 1% acetic acid (by volume) at 75 °C. Self-cleavage of the Pro²−Gly³ bond happens during the incubation. The dissociation half-lives are 48, 60, and 291 min in n-propanol, ethanol, and methanol, respectively. The dissociation is assumed to follow the DKP formation mechanism, as illustrated in Scheme 1.

Collision Cross Section Distributions for [FPG₈K + 2H]²⁺.

The structural transition of FPG₈K can be directly monitored by examining IMS cross section distributions using the IMS–MS instrument. Pierson et al.³⁰ assigned proline isomers in bradykinin based on the fact that an alanine-substituted residue can only exist in the trans configuration. We took advantage of this approach here. Figure 2 shows the CCS distribution of the peptide FPG₈K and its alanine-substituted analogue FAG₈K. The ^DTCCS_He values for the two conformations are Ω = 197 ± 3 and 207 ± 2 Ų for the peptide FPG₈K. Only the trans isomer is present in FAG₈K with Ω = 206 ± 2 Ų, after correction for differences in the intrinsic size parameters between proline and alanine, as described previously.^31,32 Therefore, we assign the peak at Ω = 207 ± 2 Ų as the isomer with a trans penultimate proline, while the isomer at Ω= 197 ± 3 Ų has a cis penultimate proline.

Figure 2.

CCS distribution for Pro² → Ala of the [FPG₈K + 2H]²⁺ ion. Distributions are obtained in ethanol with 1% acetic acid (by volume) from FPG₈K and FAG₈K. Alanine-substituted distribution is shown on top of [FPG₈K + 2H]²⁺ as a comparison. Collision cross section distribution for the alanine-substituted peptide is corrected for the size difference between Ala and Pro.

Figure 3 shows the cross section distributions for FPG₈K in water, methanol, ethanol, and propanol. Both cis and trans isomers are observed, and the cis isomer is the major component. The percentages of the cis isomer are 88, 80, 75, and 60% in propanol, ethanol, methanol, and water, respectively. Although all these solvents favor the cis isomer overall, the amount of trans increases with solvent polarity.

Figure 3.

CCS distribution of the doubly charged peptide [FPG₈K + 2H]²⁺ ions in water, propanol, ethanol, and methanol. Distributions were collected from IMS–MS instruments by integrating the drift bins for the m/z range of [FPG₈K + 2H]²⁺ ions.

An interesting question arises: is there a relationship between dissociation kinetics and the isomerization ratio? Figure 4 (top) shows the normalized intensity plots of FPG₈K isomerization and dissociation in n-PrOH, EtOH, MeOH, and H₂O at 75 °C. As dissociation progresses, both the cis and trans isomers of [FPG₈K + 2H]²⁺ decrease in abundance, while the intensity of the fragment ion (i.e., [G₈K + 2H]²⁺) increases. A previous study shows that peptides with a penultimate proline in the trans configuration cannot undergo the DKP formation process;³³ therefore, the trans isomer of [FPG₈K + 2H]²⁺ must undergo trans → cis isomerization of the Phe¹–Pro² bond before dissociation. The plot here shows that both the isomerization and fragmentation rates increase with decreasing polarity of the solvent.

Figure 4.

Top: Normalized abundances of the doubly charged peptide [FPG₈K + 2H]²⁺ as incubation progresses in propanol (black), ethanol (red), methanol (blue), and water (cyan) at 75 °C. The hollow triangles and inverted triangles represent cis and trans [FPG₈K + 2H]²⁺ ions, respectively. The abundance of [FPG₈K + 2H]²⁺ ions’ cis and trans conformations is extracted from the CCS distribution obtained by IMS–MS. Kinetic fits from several models are shown, including model 1 (for methanol, green curves) and the best-fitting model for each process (see Tables S1–S6 for details). Bottom: Percentage of cis relative to trans [FPG₈K + 2H]²⁺ ions throughout the degradation process in propanol (rectangle), ethanol (circle), methanol (triangle), and water (diamond) at 75 °C. Cis % gradually increases as dissociation progresses to near 100% at the end of the dissociation. The rate of the cis % growth is proportional to the degradation rate. The entire dataset for the water system can be found in Supporting Information, Figure S16.

Figure 4 (bottom) shows the percent of the cis isomer in the total distribution of [FPG₈K + 2H]²⁺ during the degradation process in different solvents at 75 °C. The trans → cis isomerization of the Phe¹–Pro² bond continues even while the population of cis molecules is dissociating. The relative rates of the two steps result in the cis isomer growing in abundance relative to the trans isomer throughout the process. Unlike some protein folding processes in which the trans → cis isomerization of proline is the rate-limiting step^34–36, isomerization and cis isomer dissociation of FPG₈K have comparable rates, with isomerization being somewhat faster than dissociation. The cis isomer abundance of the plots in Figure 4 increases with decreasing solvent polarity, implying that the effect of the solvent on the rates is stronger for isomerization than for fragmentation.

Characterizing the Dynamics and Pathways of FPG₈K Dissociation.

In order to further understand the dynamics behind the dissociation, we modeled the experimental kinetics for several potential pathways and compared the sums of squares of the fitting residuals. This is conducted in a manner similar to that used in previous studies of dissociation of bradykinin and substance P.^9,19 Details of this type of fitting have been discussed previously.³⁷ We begin by considering the simplest pathway: FPG₈K with a trans penultimate proline undergoes trans → cis isomerization, while FPG₈K with a cis penultimate proline (both from the initial cis conformer population and the cis conformer converted from trans) dissociates to form cFP and G₈K, as shown by reactions 2 and 3

$${\left[{\text{trans- FPG}}_8\text{K}+2\text{H}\right]}^{2+}\to {\left[{\text{cis- FPG}}_8\text{K}+2\text{H}\right]}^{2+}$$ (2)

$${\left[{\text{cis- FPG}}_8\text{K}+2\text{H}\right]}^{2+}\to {\left[{\text{G}}_8\text{K}+2\text{H}\right]}^{2+}+\text{cFP}$$ (3)

In Figure 4 (top), the green lines show the best fits of these two models with the experimental data for methanol. These models fail to represent the general trend of the data. We compared 10 different model mechanisms for each process in each solvent. The sum of squares of the fitting residuals (ΣRSS) are listed with each candidate mechanism in Tables S1 (isomerization) and S2 (dissociation) for methanol. The single-step processes in 2 and 3 appear as model 1 in their respective tables. The sequential unimolecular models with three and four intermediates have the lowest ΣRSS and thus best reproduce the abundance profiles of trans → cis isomerization of the Phe¹–Pro² bond and cis isomer dissociation, respectively, as shown by the blue curves in Figure 4. The data for these two processes in ethanol and propanol are also best reproduced (ΣRSS at a minimum) by sequential unimolecular models, although with different numbers of unseen intermediates (see Supporting Information, Tables S3–S6). These intermediates are referred as unseen intermediates because they are not detected by IMS–MS but are derived from the best-fitting kinetics model.^9,37 FPG₈K trans → cis isomerization of the Phe¹–Pro² bond and cis-isomer dissociation require two and three intermediates in EtOH, while one and two intermediates yield the best fits to the 1-PrOH data. These fits for 75 °C are also shown in Figure 4, with data and fits for 70 and 65 °C found in Supporting Information, Figures S3–S7. The same type of figure and table for FPG₈K isomerization in water is shown in Table S7 and Figures S8 and S9, respectively.

The numbers of intermediates involved in dissociation in ethanol are similar to our previous study of bradykinin and substance P dissociation in the same solvent, in which, in total, at least three and four intermediates are involved in BK(2H⁺) → BK_(3–9)(2H⁺) + cRP(H⁺) and subP_(3–11)(2H⁺) → cKP(H⁺) + subP_(5–11), respectively. In addition, one and zero intermediates are derived from the dissociation curves for penultimate proline trans → cis isomerization in bradykinin and substance P, which are close to the two intermediates found in this study for the same process in ethanol. A previous Pro13 study shows that the all-cis-configured right-handed helical PPI conformer can undergo trans → cis isomerization of each peptide bond into an all-trans-configured PPII structure, during which six distinct long-lived intermediates are observed.³⁸ The proline chain in HisPro13 undergoes the same process via a cooperative two-state transition with ∼15 unseen intermediates.³⁷ These results suggest that intermediate conformations are likely inherent in proline trans → cis isomerization.

Transition State Thermodynamics.

After fitting the data at various temperatures, Arrhenius plots are generated for FPG₈K trans → cis isomerization of the Phe¹–Pro² bond and cis-FPG₈K dissociation in each solvent by plotting the natural log of the rate constant against inverse temperature according to eq 4.

$$\ln (k)=\frac{-E_a}{R}\cdot \frac{1}{T}+\ln (A)$$ (4)

(Supporting Information, Figures S10–S13) activation energy (E_a) and the pre-exponential factor (A) are determined from the slope and y-intercept, respectively, of the best-fitting line. Transition state theory then allows us to parse out the contributions of ΔH^⧧ and entropy (ΔS^⧧) to the transition barriers in each solvent based on eqs 5 and 6

$$\text{Δ}{H}^{‡}=E_{\text{a}}+RT$$ (5)

$$A=\frac{e{k}_{\text{B}}T}{h}{e}^{\text{Δ}{S}^{‡}/R}$$ (6)

$$\text{Δ}{G}^{‡}=\text{Δ}{H}^{‡}-T\text{Δ}S$$ (7)

where h is Planck’s constant, k_B is Boltzmann’s constant, R is the gas constant, T is temperature of the reaction, and e is Euler’s number. We calculated the Gibbs free energy according to eq 7 for each transition state (ΔG^⧧) from ΔH^⧧ and ΔS^⧧ determined here. The derived values are shown in Table 1. The consecutive-step analysis assumes identical rate constants for all steps through a group of unseen intermediates that lead to an observed change—either isomerization or dissociation—so these are single-step barriers.

Table 1.

Summary of Transition State Thermodynamic Values at 298 K for FPG₈K Trans → cis Isomerization and cis-FPG₈K Dissociation in Different Solvents

solvents	processa	ΔG⧧ (kJ·mol−1)	ΔH⧧ (kJ·mol−1)		ΔS⧧ (J·mol−1·K−1)	Ea (kJ·mol−1)
n-PrOH	trans → cis (1)	100.1 ± 2.1	104.3 ± 1.6	14.1 ± 4.7	101.8 ± 1.6
n-PrOH	dissociation (2)	99.4 ± 2.4	105.6 ± 2.2	20.9 ± 3.5	103.1 ± 2.2
EtOH	trans → cis (2)	100.5 ± 1.1	106.0 ± 0.5	18.4 ± 3.4	103.5 ± 0.5
EtOH	dissociation (3)	99.9 ± 3.7	109.5 ± 3.4	31.9 ± 5.5	107.0 ± 3.4
MeOH	trans → cis (3)	103.8 ± 1.7	110.0 ± 1.3	21.0 ± 3.7	107.6 ± 1.3
MeOH	dissociation (4)	104.1 ± 0.4	116.4 ± 0.2	41.1 ± 1.1	113.9 ± 0.2
H2O	trans → cis (7)	110.0 ± 3.2	120.4 ± 2.5	35.0 ± 6.7	117.9 ± 2.5

^aThe number in the parenthesis represents the number of unseen intermediates derived from the best-fitting model.

The derived enthalpy of activation for dissociation is higher in each solvent than that for the isomerization step, and the difference grows with solvent polarity. In propanol, the ΔH^⧧ values are essentially the same for the two steps in that the uncertainty ranges of the two values overlap. These differences easily exceed the uncertainty ranges for ethanol and methanol, so the solvent polarity effect is clearly real. Not only is each barrier higher for the dissociation step than for trans → cis isomerization of Phe¹–Pro² but also dissociation involves more consecutive barriers (due to the unseen intermediates) in all three solvents.

The entropy of activation for both steps depends similarly on solvent polarity. We find that the value of ΔS^⧧ for both the isomerization and dissociation steps increases with solvent polarity. In addition, the difference between ΔS^⧧ for isomerization and dissociation in a given solvent increases with solvent polarity. Unlike enthalpy, although, a higher entropy value eases the process. The higher the value of ΔS^⧧, the greater the loss of molecular order as the combined peptide–solvent system approaches the transition state. The dissociation transition state is looser than that of isomerization: the channel on the potential energy surface through which a cis conformer dissociates is less entropically restrictive than that through which trans to cis isomerization proceeds. In other words, in a potential energy diagram, a broader range of conformations is compatible with crossing the transition for cis conformer dissociation than for a trans isomer isomerization. In each solvent, the uncertainty range of ΔG^⧧ for trans → cis isomerization overlaps with that for dissociation, so the two values are statistically indistinguishable. The barriers for trans → cis isomerization at penultimate proline determined here are similar to previously reported values for other peptides.^9,39 While the ΔH^⧧ values account for the bulk of each free energy barrier, the entropic easing effect is the largest where the enthalpic barriers are the largest. This results in both the similarity of ΔG^⧧ for isomerization and dissociation in a given solvent and the fact that all seven ΔG^⧧ values across the three solvents are in the same neighborhood. There appears to be an enthalpy and entropy compensation that results in similar free energy barriers for each step, as observed in previous research.^40–44

Of these three solvents, FPG₈K degradation kinetics is the fastest in propanol. Propanol also yields the smallest number of intermediates involved in both FPG₈K trans → cis isomerization and cis-FPG₈K dissociation processes. We can make sense of this by noting that the strength of hydrogen bonding interactions between the peptide and solvent should increase with solvent polarity. Therefore, a more polar solvent requires more energy for the peptide to undergo structural change, whether trans → cis isomerization or cis conformation dissociation. This manifests as a larger ΔH^⧧ in a more polar solvent. The tighter solvent–peptide interactions in a more polar solvent also may play into the higher number of hidden intermediate states, as the peptide and surrounding solvent molecules both rearrange on the way to completing each of the two observable steps. The peptide–solvent and solvent–solvent interactions that must be overcome as the peptide rearranges have a greater initial ordering effect in a solvent that makes stronger hydrogen bonds. The observed trend of higher ΔS^⧧ in a more polar solvent thus seems more likely to grow out of greater order in the potential well before the transition barrier than from additional disorder in the transition state itself.

Transition State Thermochemistry and the Dielectric Constant of the Solvent.

Because of the changes in thermochemistry observed with variations in solvent polarity, it is interesting to consider the relationship of the measured thermochemistry and the dielectric constants of the solvents. Plots of transition state thermochemistry versus dielectric constant for both isomerization and dissociation of the peptide FPG₈K are shown in Figures 5 and 6, respectively. The corresponding regression values and uncertainties are listed in Table 2. Transition state thermochemistry appears to be correlated with the dielectric constant of the solvent. Both ΔH^⧧ and TΔS^⧧ increase with increases in the dielectric constant.

Figure 5.

ΔH^⧧ and TΔS^⧧ vs dielectric constant (values are used at 25 °C) diagram for the isomerization step of each solvent system investigated. The correlation equations for ΔH^⧧ vs dielectric constant and TΔS^⧧ vs dielectric constant are y = 0.26x + 99.96 and y = 0.99x + 2.79, respectively. Each value is the average of triplicate measurements, and error bars represent the standard deviation about the mean. T is 298.15 K.

Figure 6.

ΔH^⧧ and TΔS^⧧ vs dielectric constant (values are used at 25 °C) diagram for the dissociation step of each solvent system investigated. The correlation equation for ΔH^⧧ vs dielectric constant and TΔS^⧧ vs dielectric constant are y = 0.94x + 87.80 and y = 0.50x – 3.40, respectively. Each value is the average of triplicate measurements, and error bars represent the standard deviation about the mean. T is 298.15 K.

Table 2.

Regression Values and Uncertainties of Correlation of ΔH^⧧ and TΔS^⧧ Versus Dielectric Constant for Isomerization and Dissociation Processes

	process	equation	R-square
ΔH⧧ vs ε	isomerization	y = 0.26x + 99.96	0.975
ΔH⧧ vs ε	dissociation	y = 0.94x + 87.80	1.000
TΔS⧧ vs ε	isomerization	y = 0.99x + 2.79	0.974
TΔS⧧ vs ε	dissociation	y = 0.50x − 3.40	0.949

In Figures 5 and 6, trend lines are derived based on transition state enthalpy and entropy values. Extrapolation to ε = 1 yields transition state thermochemistry corresponding to vacuum: for the isomerization step, ΔH^⧧, ΔS^⧧, and ΔG^⧧ are 100.2 kJ·mol⁻¹, 9.7 J·mol⁻¹·K⁻¹, and 97.3 kJ·mol⁻¹, respectively, and for the dissociation step, ΔH^⧧, ΔS^⧧, and ΔG^⧧ are 88.7 kJ·mol⁻¹, −9.7 J·mol⁻¹·K⁻¹, and 91.6 kJ·mol⁻¹, respectively. These results represent the contributions from the peptide itself to the total transition state thermochemistry. The values here suggest that both steps are enthalpically disfavored. However, these extrapolations to vacuum predict ΔH^⧧ for isomerization to be higher than for dissociation, which is the opposite of our solution-phase results. In addition, while both steps are entropically favored in solution, in vacuum, the dissociation step is predicted to be restricted entropically. This comparison is also reversed from the solution-phase results. A negative ΔS⧧ makes sense for dissociation in that the reaction requires a precise attack of an amine nitrogen on a carbonyl carbon. These trend lines might also allow predictions about the reaction rate and transition state thermochemistry in biological environments with varying local dielectric characteristics. It is worth mentioning that the correlation curves are different with or without the derived transition state thermochemistry values of water for the isomerization step. The derived values for the peptide isomerization without including the data with water as a solvent are as follows: ΔH^⧧, ΔS^⧧, and ΔG^⧧ values are 94.4 kJ· mol⁻¹, 4.10 J·mol⁻¹·K⁻¹, and 93.3 kJ·mol⁻¹, respectively. Although these values vary once, we include the data associated with water as the solvent, the general thermochemical trends remain consistent and this does not change our overall conclusions.

Both ΔH^⧧ and TΔS^⧧ grow with increasing dielectric constant. This means the entropic barrier is eased as the enthalpic barrier grows. Conversely, the dissociation in the solvent with a lower dielectric constant is favored largely by lowering the enthalpic barrier, while slightly increasing the entropic barrier. This opposed trend of entropic and enthalpic barriers with a dielectric constant is reminiscent of enzyme behavior. The dielectric environment of the active pocket of an enzyme depends on which residues are present but will, in general, differ significantly from the dielectric environment of the solvent, which could contribute to high catalytic activity.⁴⁵⁻⁴⁷ This difference of permittivity leads to a decrease in activation energy at the expense of entropic restriction, which parallels the behavior we see here with changes in solvent dielectric constant.

The fact that the fragmentation rate increases with decreasing dielectric constant leads us to offer a final suggestion. In the nonpolar environment of a biological membrane, the DKP formation rate should be faster than the same reaction under aqueous conditions. This implies that as peptides are screened for possible drug or antigen use in vivo, susceptibility to DKP formation should be considered and, if possible, such studies should consider a variety of environments.

Proton Concentration from Acetic Acid Dissociation in Different Solvent.

Considering the fact that DKP formation reaction is influenced by the protonation of the N-terminus (a free amino group is necessary for the nucleophilic attack), the dissociation of acetic acid in different solvents, which contributes to the donation of protons to the solvent, could affect dissociation. The pK_a values of acetic acid in PrOH, EtOH, MeOH, and H₂O are 10.45, 10.32, 9.63, and 4.75, respectively.^48–50 Since 1% acetic acid (17.5 × 10⁴ μM) is added to each solution, the calculated [H⁺] in H₂O, MeOH, EtOH, and PrOH are 5.5 × 10⁻⁴, 2.0 × 10⁻⁶, 9.2 × 10⁻⁷, and 7.9 × 10⁻⁷ M, respectively. The correlation between log[H⁺] and the solvent dielectric constant appears to be linear (shown in Figure S15). The concentration of the peptide FPG₈K is 20 μM, ∼0.04, 10, 22, and 25 times the concentration of [H⁺] in H₂O, MeOH, EtOH, and PrOH, respectively. Overall, we see that this trend is opposed to experimental finding, suggesting that excess protons have little if any influence on this system.

CONCLUSIONS

IMS–MS techniques were used to investigate the dissociation kinetics of FPG₈K in different solvents at multiple defined temperatures. The CCS distribution shows that two conformations in solution result from the peptide’s trans → cis isomerization at proline. Using IMS–MS, we are able to monitor changes in abundance of the cis and trans isomers alongside the dissociation process. We found that there are two processes involved in the dissociation: trans → cis isomerization and cis isomer degradation. We found that DKP formation is more rapid in less polar solvents. Decreased solvent polarity also enhances the preference for the cis conformation over trans at the penultimate proline and speeds up the trans → cis isomerization process. By monitoring the dissociation process at multiple temperatures and modeling the dissociation pathway, we found that a number of unseen intermediates are involved in both isomerization and dissociation. This modeling also allowed us to determine transition state thermodynamic values. There appears to be a strong correlation between the dielectric constant and transition state enthalpy and entropy. The solvent with a lower dielectric constant speeds up the reaction by lowering the enthalpic barrier although slightly increasing the entropic barrier. The Gibbs free energy of activation in all the solvents are primarily controlled by transition state enthalpy.

Scheme 1.

Cleavage Mechanism of Pro²–Gly³ by DKP Formation with Phe¹–Pro² in the Cis configuration^a

^aThe cis and trans isomers mentioned in this paper are b and a, respectively, when R = Phe.