Monitoring the stabilities of a mixture of peptides by mass-spectrometry-based techniques

Abstract

Biomolecular degradation plays a key role in proteostasis. Typically, proteolytic enzymes degrade proteins into smaller peptides by breaking amino acid bonds between specific residues. Cleavage around proline residues is often missed and requires highly specific enzymes for peptide processing due to the cyclic proline side-chain. However, degradation can occur spontaneously (i.e. in the absence of enzymes). In this study, the influence of the first residue on the stability of a series of penultimate proline containing peptides, with the sequence Xaa–Pro–Gly–Gly (where Xaa is any amino acid), is investigated with mass spectrometry techniques. Peptides were incubated as mixtures at various solution temperatures (70°C to 90°C) and were periodically sampled over the duration of the experiment. At elevated temperatures, we observe dissociation after the Xaa–Pro motif for all sequences, but at different rates. Transition state thermochemistry was obtained by studying the temperature-dependent kinetics and although all peptides show relatively small differences in the transition state free energies (~95 kJ/mol), there is significant variability in the transition state entropy and enthalpy. This demonstrates that the side-chain of the first amino acid has a significant influence on the stability of the Xaa–Pro sequence. From these data, we demonstrate the ability to simultaneously measure the dissociation kinetics and relative transition state thermochemistries for a mixture of peptides, which vary only in the identity of the N-terminal amino acid.

Introduction

Biomolecular stability plays an important role in areas such as aging,^1,2 drug development and stability,^3–7 and even life at high temperatures.^8,9 Factors such as environment,^4,10 ligand-binding,^11,12 and all levels of biomolecule structure^13,14 can influence the lifetime of a biomolecule. For example, peptide therapeutics associated with the major histocompatibility complex (MHC) are being tailored to combat rheumatoid arthritis,¹⁵ diabetes,¹⁶ and various forms of cancer.^6,7 For these peptide-based therapies to be effective, they must exhibit significant in vivo stability to survive translocation to the cell surface. Thus, a detailed understanding of biomolecular stability, especially their resistance to degradation, is a key first-step in designing new biologics.

Unlike other amino acids, proline residues have the unique ability to exist in either the cis or trans forms;¹⁷ it has been shown that proline isomerization plays a key role in protein folding,^18–20 signaling,²¹ and regulation.²² The propensity to exist in either form has been investigated by spectroscopic,^23,24 molecular dynamics,²⁵ and ion mobility spectrometry-mass spectrometry (IMS-MS) based studies.^26,27 Glover et al.²⁸ showed the penultimate proline (i.e. proline in the second position) motif has a high propensity for presenting both cis and trans forms, and in follow-up studies, it was shown that this duality plays a key role in the dynamic and flexible nature of neuropeptides.²⁹ Penultimate prolines are prevalent in a number of systems, including signaling peptides,^30,31 neuropeptides,^29,32 and MHC-associated peptides,^33,34 and are suggested to prevent further enzymatic degradation.³¹ For neuropeptides specifically, proline was found to be most abundant in the penultimate position relative to all other amino acids, and furthermore, was most likely to occupy the second position relative to other positions in the sequence,²⁹ suggesting proline in the second position is important to the biological function of the molecule.

In the cell, biomolecular decomposition of proteins occurs through a vast number of enzymes, which are then broken down further to peptides and then to amino acids. However, a recent study by our group found that bradykinin (Arg–Pro–Pro–Gly–Phe–Ser–Pro–Phe–Arg) undergoes spontaneous cleavage of the first two residues (Arg–Pro), which is quite interesting considering that the Pro–Pro bond possesses significant resistance to any proteolytic enzyme.³⁵ The cleavage was preceded by a slow protonation event regulated by a conformational change. Interestingly, the BK_(3–9) product has been suggested to have almost an order of magnitude greater antigenicity,³⁶ suggesting the slow cleavage serves as a protection mechanism for aberrant bradykinin activity.

It has been proposed that the main cleavage product for these systems leads to fragmentation in which one of the products has a diketopiperazine (DKP) structure. This process is regulated by trans → cis isomerization of the proline. A cis-configured Xaa¹–Pro² bond is essential for DKP formation, as this isomerization event brings the N-terminus in close proximity to the carbonyl in the backbone.³⁷ This allows for nucleophilic attack of the N-terminus to the carbonyl carbon between the second and third residues to occur, causing cyclization to the DKP product.³⁷ This is illustrated in scheme 1. The resulting cyclic dipeptide (DKP) is spontaneously cleaved from the remaining C-terminal residues. We anticipate that these systems follow this mechanism, similar to what we observed for bradykinin.

Scheme 1.

Proposed solution mechanism of diketopiperazine formation for Gly–Pro–Gly–Gly.

In this study, we use electrospray ionization coupled to mass spectrometry to investigate the influence of the preceding N-terminal residue on the dissociation kinetics (i.e. DKP formation) of a series of model peptides with the sequence Xaa–Pro–Gly–Gly (XPGG). An energetic profile for each is determined by monitoring the dissociation at multiple temperatures, providing a detailed view of the forces driving dissociation. We find a strong correlation between transition state enthalpy and entropy, which balance such that the free energy barrier is similar for all peptides studied. As described below, the peptides were studied as a mixture of no less than five species per dissociation; these studies provide the first framework for simultaneous monitoring of the stabilities of a mixture of biomolecules, facilitating the development of “stability” databases.

Experimental section

Overview

Peptide fragmentation data were obtained by monitoring the mass spectra of peptide in solutions incubated at controlled temperatures at defined time points. Positively charged ions were generated via electrospray ionization using a Triversa Nanomate autosampler (Advion, Ithaca, NY) and detected with a Thermo LTQ velos mass spectrometer. Collision cross-section distributions (see supporting information, Figure S5) were collected on a home-built ion-mobility spectrometer coupled to a time-of-flight mass spectrometer.^38,39

Data analysis

The abundance of each peak in the mass spectrum was integrated using OriginPro 2016 (OriginLab Corporation, Northampton, MA), and relative abundances were determined relative to the internal standard. We note that while we do observe the products of dissociation for each peptide, the dipeptide products (DKP) do not appear to ionize well as compared to the intact peptide (especially for those lacking a basic residue at the first position). Because of this, we felt a more accurate representation of the relative stabilities was to compare abundances to an internal standard in solution rather than the abundances of the products. The abundances were then plotted as a function of time. Maple software (Waterloo Maple Inc., Waterloo, Canada) was used to simultaneously solve a set of first-order differential rate equations, which were written to describe all plausible mechanisms for the dissociation reaction. The equations generated could then be incorporated into OriginPro 2016 software using the Nonlinear Curve Fit tool and used to model the data.

Peptide synthesis

Peptides were synthesized by solid-phase Fmoc synthesis. Fmoc side-chain protected amino acids and Wang-type polystyrene resins were purchased from Midwest Biotech (Fishers, IN). A solution of 20% piperidine in dimethylformamide was used for the deprotection step and couplings were performed with a solution of 1 M 1,3-diisopropylcarboiimde and 1 M 6-chloro-1-hydroxybenzotriazole in N-methyl-2-pyrrolidone. A solution of trifluoroacetic acid: triisopropylsilane: methanol (18:1:1 v:v:v) was used to cleave peptides from the resin, which were then precipitated with diethyl ether and dried using a vacuum manifold. The peptides were then dissolved in pure 1-propanol (PrOH) at a concentration of 1 mM. Cysteine was not studied, as it is known to form disulfide bridges at elevated temperatures.

Sample preparation

Prior to use, peptides were stored at −20°C to prevent any premature degradation. The peptides were diluted from the 1 mM stock solution to a concentration of 5 μM in a PrOH solution with 0.5% acetic acid, which was preheated to the incubation temperature of the experiment. Each solution contained a mixture of 5–7 peptides each at 5 μM. Each peptide was incubated in a water bath at three different temperatures, 70°C, 80°C, and 90°C. The mixtures were varied for each run, and were sampled at specific times throughout the duration of reaction. An internal standard was spiked into the sample prior to being electrosprayed into the mass spectrometer. The internal standard was varied for each mixture and consisted of one of the sequences not present in the incubated mixture. All dissociation reactions were measured in triplicate. Also, a concentration study was performed (data not shown) to ensure no higher order structures formed (dimers, trimers, etc.), which would skew the rates of dissociation. At higher concentrations and/or with more peptides in the mixture, we observed nonspecific dimer formation, therefore, we optimized the size (number of peptides) and concentration of our studies to eliminate multimer formation and ensure accurate measurement of the rates of dissociation.

Determination of Arrhenius activation parameters

Arrhenius activation parameters were determined by plotting ln(k) as function of T⁻¹. The Arrhenius equation

$$k=A{e}^{\frac{-E_a}{RT}}$$ (1)

establishes the dependence of the rate constant (k) on temperature (T), which can be rearranged to determine the activation barrier (E_a) and pre-exponential factor (A) (R is the universal gas constant). Using transition state theory, transition state enthalpy (ΔH^‡), entropy (ΔS^‡), and Gibbs free energy (ΔG^‡) can be determined from equations (2), (3), and (4), respectively

$$\Delta H^{⧧}=E_a-RT$$ (2)

$$A=\frac{k_{B}T}{h}{e}^{\frac{\Delta S^{⧧}}{R}}$$ (3)

$$\Delta G^{⧧}=\Delta H^{‡}-T\Delta S^{⧧}$$ (4)

where k_b is Boltzmann’s constant and h is Planck’s constant.

Results and discussion

MS analysis of mixture containing multiple XPGG sequences

Figure 1 shows representative mass spectra for a mixture of XPGG sequences incubated at two different temperatures, 70°C and 90°C (for 80°C, see Figure S1). All peptides appear to ionize as singly-protonated ions, [XPGG + H]⁺, which is consistent across all temperatures for all peptides. For the plot shown at 70°C, we observe peaks at m/z=317.1, 331.2, 343.2, 345.1, 358.2, 386.2, and 393.2 assigned as the peptides, SPGG, TPGG, IPGG, DPGG, KPGG, RPGG, and YPGG, respectively. We monitor the ion intensities of the precursors relative to an internal standard (FPGG at m/z=377.2, indicated in red). As time progresses, the intensities of each peptide decrease relative to the internal standard. Interestingly, the decay of several species varies with respect to one another; an indication that subtle differences in their stabilities can be captured by analyzing their dissociation profiles. This is clear when comparing the ion intensities of IPGG and DPGG at 70°C, which at 0 min, are very similar in intensity. However, at 90 min ~74% of the initial IPGG signal is present, whereas only ~33% of DPGG remains. At the latest time point shown, (555 min) a small amount of IPGG remains and, DPGG has completely dissociated. In addition, at 90 min 58%, 59%, 39%, 29%, and 55% of SPGG, TPGG, KPGG, RPGG, and YPGG, respectively, remain, and all the peptides are almost entirely dissociated at the longest time point shown (555 min). Also shown in Figure 1 are the dipeptide products (DKP forms); RP and KP associated with RPGG and KPGG are labeled. While we do observe nearly all products of dissociation, we note that it appears the ionization efficiency of the dipeptides do not match that of their precursors. For DKP products with basic side-chains (RP, KP, HP) the dipeptide ionizes similarly to the precursor, but this does not seem to be the case for most of the peptides studied. Because of this, the data can be skewed when monitoring the reactants/product ratios as the dissociation progressed. Therefore, incorporation of an internal standard is imperative when studying the changes in MS ion intensities when monitoring dissociation profiles.

Figure 1.

Representative mass spectral distributions for a mixture of XPGG peptides at 70°C and 90°C. The peaks in the 0 min distribution are labeled by the first residue in the sequence with the red label indicating the internal standard used in the experiment. For 70°C, we additionally label the peaks associated with DKP dipeptides (R* and K*).

The right panel in Figure 1 shows that we observe similar results at 90°C (i.e. each peptide dissociates with unique behavior). In this dataset, we monitor the abundances of GPGG, SPGG, VPGG, LPGG, QPGG, and YPGG (at m/z=287.1, 317.1, 329.2, 343.2, 358.2, and 393.2, respectively) relative to the TPGG (m/z=331.2) internal standard. Again, the differences in dissociation behavior are clear when comparing pairs of XPGG peptides. At 0 min, the ion intensities of VPGG and QPGG are nearly equivalent. However, at 20 min, ~60%, of VPGG persists, whereas only ~2% of QPGG remains. An interesting case is the comparison of SPGG and YPGG, which are present at a ~60:40 ratio, respectively, at 0 min. At 20 min, their intensities have decreased, but the 60:40 ratio persists. While almost completely dissociated at 245 min, the abundance of YPGG overtakes that of SPGG. At the longest time point sampled (245 min), nearly all peptides have dissociated. Finally, as a check to ensure the rates were not being influenced by other species in the mixture, we followed the dissociation of APGG, EPGG, and KPGG (see supporting information) and observe almost identical half-lives (within experimental uncertainty, see Figure S2). These comparisons demonstrate how a single residue can influence the rate of Xaa–Pro cleavage (i.e. the overall stability of the peptide) and, more importantly, that these transitions can be followed as complex mixtures.

Kinetics analysis for the decay of XPGG abundance

Figure 2 shows the decay profiles for GPGG, TPGG, DPGG, HPGG, and LPGG at 70°C (expanded view of early time points for 70°C shown in Figure S3) and 90°C (for 80°C see Figure S4). A quantitative description of the dissociation can be captured by modeling the dissociation profile using traditional unimolecular kinetics. We first assume the simplest mechanism, GPGG(H⁺) → GP(H⁺)+GG; however, this typical unimolecular kinetics model does not capture the induction period observed in the experimental data. Furthermore, our studies show that this mechanism provides an inadequate fit for other peptides that present an induction period prior to dissociation (GPGG, LPGG, and TPGG). However, the lag-time prior to decay can be captured using a simple model that assumes a number of experimentally unobserved intermediates which precede isomerization⁴⁰ and dissociation.³⁵ A study of Bradykinin dissociation found that multiple steps (not captured by the experiment) are required to produce a good fit of the data upon dissociation at high temperature.³⁵ These previous studies rigorously sampled multiple approaches to fitting the data, including the use of nonsequential pathways and variation in the rate-limiting step(s). Because this framework has been sufficiently explored and well-established, we are confident in our use of these simple parameters for fitting these much simpler systems.

Figure 2.

Kinetics plots shown for five different XPGG peptides (DPGG, HPGG, TPGG, LPGG, and GPGG) at 70°C and 90°C. The relative intensities are plotted as a function of time.

After testing multiple models incorporating varying numbers of intermediates, we settle on the simple sequential model for all peptides which require up to one unobserved intermediate, as determined by evaluating the goodness of fit for each model (data not shown). The best models for GPGG, LPGG, and TPGG include one intermediate, whereas the DPGG and HPGG experimental decay profiles can be best-fit using a single-step unimolecular transition. The addition of an extra step in the dissociation of some sequences is likely due to cis/trans isomerization of the Xaa–Pro bond, suggesting the residue preceding Pro plays a key role in the dissociation kinetics.

Figure S5 shows the ion mobility–mass spectrometry data for several representative peptides. As shown, two peaks are observed (although one is often small) in the mobility distributions. The two peaks likely arise from cis/trans isomerization at the Xaa–Pro bond resulting in one peak represented by Xaa–trans–Pro and the other as Xaa–cis–Pro—this cis/trans isomerization is likely why some of the peptides proceed through an intermediate prior to dissociation. These results are consistent with those observed for BK, in which cis → trans isomerization of the Arg–Pro bond resulted in a lag time prior to dissociation – consistent with DKP formation, which required Pro to be in cis.³⁷ While we only provide a qualitative analysis of cis/trans isomerization in this work, further analysis will be provided in a forthcoming manuscript. However, these data support the idea that the rate and propensity for cis/trans isomerization may have a large influence on the delay prior to dissociation.

Grouping side-chains as a function of properties

It is interesting to consider the factors that influence the rates of peptide bond cleavage in these systems. By changing only the first amino acid, we see the influence of side-chain properties on the rates of decomposition. Variation of a single amino acid influences properties such as solubility, charge location, and overall conformation. Figure 3 shows the grouping of amino acids by side-chain properties with respect to their half-lives (t_1/2) at 70°C (details about the fitting of the kinetics data is presented later). With the exception of Pro and Gly (which make their own unique group), we group the nonpolar aliphatic, aromatic, polar aliphatic, nucleophilic, and basic side-chains. While detailed molecular modeling is necessary for an atomic description of the interaction networks, this approach allows us to visualize the influence that the physical properties of first residues have on the stability of the peptide. The nonpolar aliphatic residues (APGG, VPGG, IPGG, LPGG, MPGG) show significant variation in their respective t_1/2 values, but overall tend to dissociate relatively slow with an average t_1/2 of 148.8±41.3 min. If methionine is excluded from the calculation, which dissociates faster compared to the others in the group, the value becomes 165.1±28.4 min (~30% decrease in the standard deviation). The wide range of t_1/2 values suggests differences in the interactions between side-chains and various regions of the peptide, which may be why methionine dissociates much faster (as it can only form weak hydrogen bonds compared to other nonpolar aliphatic residues). The aromatic side-chains (FPGG, WPGG, and YPGG) also show slow dissociation and behave similarly, having average t_1/2 of 164.6±18.6 min. Additionally, the nucleophilic residues reside close to the middle in regard to half-life, having t_1/2 = 115.3±4.5 min. These species, along with the aromatics, can potentially partake in intramolecular interactions with the C-terminus and any charge carrying moiety, slowing any large structural changes that occur prior to dissociation. The two fastest groups to dissociate are polar aliphatic and basic; polar aliphatic side-chains (DPGG, EPGG, NPGG, QPGG) have an average t_1/2 of 52.1±19.5 min, while the basic (KPRR, RPGG, HPGG) groups have average t_1/2 of 62.6±8.4. PPGG and GPGG standout with t_1/2 values that are significantly slower than the others (t_1/2=518.6±37.8 and 253.8±21.9 min, respectively). This is an interesting finding, and it is not immediately clear as to why they would show the slowest dissociation. From these data, it is clear that amino acids with similar physical properties (e.g. nonpolar aliphatic, aromatic, polar aliphatic, nucleophilic, basic) provide similar t_1/2. While the specific intramolecular configurations may differ, amino acids with basic side-chains (Arg, Lys, His) likely possess the positive charge; therefore, the peptide can readily form DKP without needing to undergo charge-transfer from the N-terminus to some region along the peptide backbone. This leads to a wider range of structures that can form a DKP moiety and an overall faster transition relative to many of the other systems studied.

Figure 3.

t_1/2 for each XPGG sequence plotted at 70°C. Sequences are grouped based on the side-chain properties of the first residue (nonpolar aliphatic, aromatic, polar aliphatic, nucleophilic, basic, and P and G are grouped together). Amino acid backbone and side-chains are shown at top of plot.

Transition state thermochemistry

Once the dissociation pathway has been established from modeling of the kinetics, more insight of these systems can be gained by analyzing the transition state thermochemistry. We note this because it is important to stress that the transition state thermochemistry depends upon the best model used to fit the kinetics. Figure 4 shows representative Arrhenius plots for HPGG, DPGG, GPGG, TPGG, and LPGG, determined using their best-fit kinetics mode (five peptides shown in the Figure 2). The slope and intercept of each plot allows for calculation of activation barriers and pre-exponential factors, which can be converted into the corresponding transition state thermochemical values using equations (2) to (4). Table 1 lists all transition state thermochemistry values for each system studied. The magnitude of these values allows for inference on the role of intra- or inter-molecular forces in driving the dissociation, whereas the relative magnitudes are a powerful tool in developing a scale to predict the propensity for dissociation.

Figure 4.

Arrhenius plot generated from the kinetics at three temperatures for five different peptides (DPGG, HPGG, TPGG, LPGG, and GPGG). Error bars represent the standard deviation about the mean from triplicate measurements.

Table 1.

Transition state thermochemistry values for the dissociation of the Xaa–Pro–Gly–Gly peptides. Values for the entropy of activation and Gibbs free energy of activation are calculated at 298.15 K.

Sequence	A (s−1)	Ea (kJ/mol)	ΔH⧧ (kJ/mol)	ΔS⧧ (J/mol·K)	ΔG⧧ (kJ/mol)
APGG	10(10.3 ± 0.1)	82 ± 4	80 ± 4	−50 ± 3	94 ± 4
DPGG	10(8.9 ± 0.1)	71 ± 1	69 ± 1	−82 ± 1	93 ± 1
EPGG	10(8.1 ± 2.1)	71 ± 15	69 ± 15	−83 ± 21	94 ± 15
FPGG	10(11.2 ± 0.3)	89 ± 7	88 ± 8	−36 ± 3	95 ± 8
GPGG	10(9.4 ± 0.5)	78 ± 8	76 ± 8	−70 ± 9	97 ± 8
HPGG	10(11.2 ± 0.l)	86 ± 3	83 ± 3	−40 ± 1	95 ± 3
IPGG	10(9.4 ± 0.2)	78 ± 6	75 ± 6	−64 ± 6	95 ± 6
KPGG	10(13.1 ± 0.3)	99 ± 27	96 ± 26	−4 ± 1	98 ± 26
LPGG	10(13.9 ± 0.7)	104 ± 7	102 ± 6	14 ± 1	97 ± 6
MPGG	10(12.2 ± 0.3)	94 ± 14	92 ± 14	−19 ± 3	97 ± 14
NPGG	10(9.8 ± 0.3)	78 ± 2	75 ± 2	−64 ± 2	94 ± 2
PPGG	10(9.4 ± 0.1)	77 ± 1	75 ± 1	−69 ± 1	95 ± 1
QPGG	10(11.4 ± 0.6)	85 ± 12	83 ± 11	−31 ± 5	92 ± 11
RPGG	10(13.2 ± 0.2)	99 ± 8	96 ± 8	0 ± 1	97 ± 8
SPGG	10(9.2 ± 0.2)	74 ± 5	72 ± 5	−72 ± 6	93 ± 5
TPGG	10(9.2 ± 0.4)	74 ± 9	71 ± 8	−74 ± 8	93 ± 8
VPGG	10(10.8 ± 0.4)	85 ± 1	82 ± 1	−45 ± 1	96 ± 1
WPGG	10(13.2 ± 0.l)	103 ± 7	100 ± 7	1 ± 1	100 ± 7
YPGG	10(9.2 ± 0.6)	77 ± 16	75 ± 15	−74 ± 18	97 ± 15

While fragmentation of all peptides proceeds over a similar barrier, ΔG^⧧ barrier of ~95 kJ/mol, some differences are observed. For example, substantial variation is found in extracted values of ΔH^⧧ and ΔS^⧧. When acidic residues aspartic and glutamic acid (D or E) are in the first position, the barrier is entropically restricting compared to other amino acids, with ΔS^⧧=−82±1 and −83±21 J/mol·K, respectively. In contrast, basic residues lysine and arginine (K and R) have almost no entropic barrier with ΔS^⧧=−4±1 and 0±1 J/mol·K, respectively. Interestingly, the basic residue HPGG must overcome a substantially unfavorable entropic barrier (ΔS^⧧=−40±1 J/mol·K) before dissociation, likely due to the dual basic and aromatic nature of the histidine side-chain. Like D and E, serine, threonine, and tyrosine (S, T, and Y), which all bear a hydroxyl group in their side-chain, have significant unfavorable entropic barriers of ΔS^⧧=−72±6, −74±8, and −74±18 J/mol·K, respectively, suggesting the presence of the hydroxyl group is entropically restricting compared to other amino acid side-chains. Analysis of all the thermochemistry suggests that the enthalpy and entropy barriers are correlated; for example, sequences having unfavorable entropy barriers have enthalpy barriers which are more favorable. Acidic residues (D and E) have the most unfavorable entropic barrier; however, their enthalpic barriers are ΔH^⧧=69±1 and 69±15 kJ/mol, respectively, which is lower than all other sequences. Many of the peptides, which possess amino acids with similar physical properties display similar transition barriers. However, we note that some systems have values that do not correlate with other similar systems. For example, LPGG has barriers of ΔH^⧧=102±6 and ΔS^⧧=14±1; however, for IPGG, which one would assume would behave similarly, has barriers of ΔH^⧧=78±6 and ΔS^⧧=−64±6. This demonstrates the need for further exploration into the forces that guide their stability. In summary, the enthalpic and entropic barriers are different depending on the physical properties of the amino acid side-chain. However, the consistency in ΔG^⧧ suggests that enthalpy and entropy are correlated.

Interplay between entropy and enthalpy transition barriers

Figure 5 shows a plot of ΔH^⧧ against ΔS^⧧ for all systems studied. The interplay between ΔH^⧧ and ΔS^⧧ is clear; the values are almost perfectly correlated, which prevents variation in either ΔH^⧧ or ΔS^⧧ from significantly affecting ΔG^⧧. The occurrence of an enthalpy–entropy compensation has a rich history.^41–43 Numerous recent experimental findings of the enthalpy–entropy correlation plots for equilibrium and kinetic processes alike demonstrate that this phenomenon appears to be general—a balance that minimizes variation in free energy.⁴⁴ Additional evidence for this correlation, the enthalpy and entropy values for kinetic processes previously reported by our group for cis/trans isomerization of Pro13 and HisPro13 are also plotted in Figure 3. As with all the XPGG sequences, these other values also fall on or near the compensation line. This barrier height appears to be a kinetic speed limit—experiments that commence in a similar timeframe are limited by a similar barrier height. While there will be a small error associated with our measurements for ΔG^⧧, a simple calculation shows that an order of magnitude difference in rate results in ~7 kJ/mol change in ΔG^⧧. In other words, this demonstrates why all these reactions have similar transition state free energy barriers—slight changes in barrier height results in transitions that are either too fast to be observed or occur on slow timescales that are biologically improbable or experimentally intractable. Therefore, the enthalpy and entropy values work together to maintain a constant ΔG^⧧.

Figure 5.

Transition state enthalpy–entropy correlation plots for the dissociation of all the XPGG peptides (black squares) along with all the barriers associated with dissociation of bradykinin³⁵ (open triangles) and Substance P (open circles). In red are the values from nondissociative transitions. The open red triangles are the steps from the conversion of polyproline-I to polyproline-II transition (PPI→PPII) for Pro13,⁴⁵ and the red circle represents the values from the PPI→PPII transition His-Pro13.⁴⁰ Error bars represent the standard deviation about the mean from triplicate measurements.

Summary and conclusions

In summary, we studied the thermal stability of a library of peptides (Xaa–Pro–Gly–Gly, where Xaa is any amino acid other than cysteine), which vary only in the identity of the N-terminal amino acid. We find that while all sequences we measured dissociate at elevated temperatures, variation of the amino acid in the first position has a significant influence on the rate of degradation. For this library, the differences in the rates of dissociation are possibly associated with the rate of cis/trans isomerization of the Xaa–Pro bond along with the propensity of the peptide to sample both cis and trans configurations. A quantitative picture can be gleaned from these systems from fitting the kinetics data at multiple temperatures. We find that a one-step model for dissociation does not always capture the experimental data. In such cases, inclusion of a single unobserved state provides a better fit of the kinetics data. We observed clustering of dissociation rates by grouping the sequences with respect to their side-chain properties of the first amino acid. This begins to provide us with an understanding into how the properties of amino acids can influence the rate of dissociation. The Arrhenius analysis of the temperature-dependent rate constants show a correlation between enthalpy and entropy barriers, regardless of the sequence studied. The uniform ΔG^⧧ appears at a value, which is associated with reactions that occur on the timescales of our experiments. Overall, we envision these data can aid in rational design of stable biomolecules by providing a framework of amino acid contributions to the dissociation of penultimate proline containing systems. We note that one must use caution when developing assays for studying mixtures, as interactions between different systems can cause inaccuracies in the data collected. Furthermore, the ability to simultaneously monitor the stabilities of multiple peptides/biomolecules allows for rapid library building of systems with even small differences in sequence or side-chain modification. From this alone, one can envision designing a mass spectrometric framework which facilitates the rapid, rational design of exceptionally stable biomolecules.