Introduction

Predicting the higher order structure of oligopeptides and metallopeptides based on their primary structure is an important component for understanding protein folding and a key step in designing bioactive peptides or proteins. NMR spectroscopy and X-ray crystallography are the most commonly used techniques for providing atomic resolution of peptides or proteins’ secondary/tertiary structure. For NMR, the conformational interpretation is not straightforward for flexible oligopeptides and when multiple species are present [1]. Moreover, X-ray diffraction studies of metallopeptides are scarce due to problems with the crystallization of these complexes [2]. An alternative technique is electrospray ionization–ion mobility–mass spectrometry (ESI-IM-MS), which can quantify the populations of metallopeptides and infer intramolecular interactions after the evaporation of the solvent [3,4,5,6,7]. A recent study showed how Cu(I/II)-bound peptides that only differed by their number of inter- or intramolecular disulfide bridges, number of Cu(I) or Cu(II) ions, and their deprotonation sites could be resolved by ESI-IM-MS [8]. Native ESI-IM-MS uses soft ESI conditions and transmission parameters to try to preserve the solution-phase populations, conformations, and noncovalent interactions and was used to determine the populations and conformational states of Zn(II)-bound peptides derived from histidine-rich glycoprotein [9]. Over the last 7 years, our lab has undertaken ESI-IM-MS studies of the methanobactin [10, 11] from Methylosinus trichosporium (mb-OB3b) and a series of analog methanobactin (amb) oligopeptides [8, 12,13,14]. Methanobactins [15] are Cu(I)-binding oligopeptides secreted by methanotrophic bacteria in response to copper availability in their growth environment [16, 17]. The primary structure of mb-OB3b contains two oxazolone rings associated with enethiols and separated by 4 amino acids and pyrrolidine [18]. X-ray crystallography has shown that Cu(I) is coordinated by a distorted tetrahedral geometry by the two enethiol-oxazolones [19]. The amb oligopeptides were designed with sequences that mimicked the primary sequence of mb-OB3b but replaced the two enethiol-oxazolone ligands with two His-Cys ligands (2His-2Cys motif) used by zinc fingers to selectively coordinate Zn(II) [20]. The studies showed that mb-OB3b and the ambs exhibited pH-dependent reactivity that could be closely correlated to that of the solution phase and further showed ESI-IM-MS as a powerful analytical technique for characterizing peptide interactions with metal ions.

Here, we have undertaken a new series of ESI-IM-MS and DFT studies to interrogate the effect of the 2His-2Cys motif on the charge states, tertiary structure and Zn(II)-binding of six new amb oligopeptides (Scheme 1). [12, 13] The charge states of these ambs will be affected by the gas-phase basicities [21] of the 2His and the Pro residues [22, 23] and the gas-phase acidities [24] of the carboxyl group at the C-terminus and the substituent groups of the 2Cys and Tyr. Coulomb repulsion between two positive charge sites can also affect their apparent gas-phase basicities and the conformational structure of the oligopeptide [25,26,27]. Solvent composition [28] and droplet desolvation during ESI [29] also affects the charge of the peptide. Previous work has shown a disparity between the positive charge state distributions of oligopeptides produced from ESI to those of solution-phase equilibria [30]. Considering pKas and solution-phase chemistry, the hexaprotic ac-His1-Cys2-Gly3-Pro4-Tyr5-His6-Cys7 (Scheme 1A) would exhibit an overall charge state based on the protonation state of the carboxyl group at the C-terminus (pKa ≈ 3) [31], 2His (pKa ≈ 6.0) [32, 33], 2Cys (pKa < 8.3) [34], and Tyr (pKa ≈ 11.0), and generally form positively charged ions at pH < 6.0 and negatively charged ions at pH > 6.0.

Scheme 1
scheme 1

The primary sequence of A (MW = 857.96 Da, pI = 6.03) where ac = acetyl group and the color highlighting the regions modified with bold underscore for B (MW = 856.96, pI = 7.09); C (MW = 777.87, pI = 2.06); D (MW = 697.78, pI < 2.0); E (MW = 811.87, pI = 6.03); F (MW = 765.77, pI = 6.04); and G (MW = 788.99, pI < 8.3). The pI was calculated using https://pepcalc.com/ apart from D and G which have no basic groups and their pI is shown as an upper level based on the pKa of either the carboxylate C-terminus or Cys group, respectively

Full size image

The energy-resolved collision-induced dissociation (CID) of peptides is used for protein identification in proteomics [35] and more fundamentally has been applied with molecular modeling to understand the structure of ions and mechanisms of peptide fragmentation [36,37,38,39]. Studies of small peptides include the formation and fragmentation mechanisms of Zn(II) complexes of Gly [40,41,42]; transition metal-bound polyalanines [43]; metal-binding sites of angiotensin I [7]; Pd(II) sites of ubiquitin [44]; peptides that contain His [9, 45, 46], Cys [47,48,49], Pro [50, 51], Phe [52, 53], Met [49, 54], Arg [55], Lys [46], and Gly; [56, 57]; and the structures of a [58] and b ions. [59, 60] The salt-bridges in the tertiary structure of small proteins have also been investigated by CID in combination with electron transfer dissociation [61].

Previously, our study of amb5A (Scheme 1A) and its amidated C-terminus counterpart B indicated that these peptides can bind Zn(II) via different charge state conformers including coordination by the 2His-2Cys, backbone carbonyl groups, and C-terminus. [14] Zinc(II) is an essential cofactor involved in many metabolic enzymes and transcription factors, preferring “softer” ligands such as Cys and His [20]. For example, in metabolic enzymes, Zn(II) is partially exposed to the solvent and is found tetrahedrally coordinated via 3His and a water molecule in carbonic anhydrase, while other enzymes utilize Cys, Asp, or Glu. For transcription factors such as the Zn-finger family, Zn(II) is most commonly found tetrahedrally coordinated via the 2His-2Cys motif, although His-3Cys and 4Cys are also binding motifs [20].

For this endeavor, the series of heptapeptides A to G (Scheme 1) have been chosen to test whether their differences in overall charge state, tertiary structure, and Zn(II) chelation can be attributed to the His, Cys, or C-terminus sites. Scheme 1 shows the structure of the “parent” sequence A, and the modifications to that sequence, e.g., amidation of the C-terminus (B), replacement of His6 by Gly6 (C), replacement of His1 and His6 by Gly1 and Gly6 (D), replacement of Cys2 by Gly2 (E), replacement of Cys2 and Cys7 by Gly2 and Gly7 (F), and replacement of His1 and His6 by Cys1 and Cys6 and amidation of the C-terminus (G).

Experimental Section

Reagents and Sample Preparation

The design of the heptapeptides were based on incorporating features from zinc fingers, i.e., the 2His-2Cys motif, the symmetric N- and C-terminal arms connected via the tight turn of the Pro, with the Tyr residue providing a possible π-cation interaction between the aromatic ring and the zinc ion. The heptapeptides A and B were synthesized by Neo BioLab (https://neobiolab.com) and C, D, E, F, and G by PepmicCo (http://www.pepmic.com/). Ammonium hydroxide (trace metal grade) and glacial acetic acid (Optima grade) were both purchased from Fisher Scientific (http://www.Fishersci.com), and ZnCl2 (98 + %) and Zn(NO3)2·6H2O (99%) were purchased from Alfa Aesar (http://www.alfa.com). Solutions of 25.0 μM A to G or molar equivalents A to G and Zn(II) were prepared with > 17.8 MΩ cm deionized (DI) water (http://www.millipore.com), pH adjusted with glacial acetic acid or ammonium hydroxide and left to stand for 10 min before IM-MS analysis and pH measurement using a calibrated Orion 98 micro pH electrode. Experiments were repeated using ZnCl2 and Zn(NO3)2·6H2O to check for reproducibility and estimate the standard deviations. The collision cross section calibrant poly-DL-alanine was purchased from Sigma-Aldrich (http://www.sigmaaldrich.com) and dissolved in DI water/acetonitrile (50/50) to a final concentration of 10 ppm as previously described. [62, 63]

Waters Synapt High-Definition Mass Spectrometer (HDMS)

The samples were electrosprayed into a Waters Synapt High-Definition Mass Spectrometer (G1) [64] with a flow rate of 10 μL min−1 and 2.0-kV capillary voltage with the sampling cone 25 V and the extraction cone 3.0 V. Preliminary experiments showed that all ions appeared in the 50–2500 m/z region, and the quadrupole mass analyzer was used to guide all the ions to the three traveling wave (T-wave) ion guides or resolve the precursor ion for CID studies. The trap ion guide was operated with 4-V entrance voltage and a pressure of 2.25 × 10−2 mbar with an argon gas flow of 1.5 mL/min. The trap served as a storage cell for gating ions into the ion mobility (IM) T-wave ion guide. The IM ion guide was operated at a pressure of 0.518 mbar with a flow rate of 20.0 mL min−1 N2 buffer gas with ramped T-wave heights of 7.5–16.4 V (+ ion mode) or 13.0–22.0 V (−) and ramped T-wave velocities of 600–1400 m s−1, for every sweep of the IM T-wave ion guide. The transfer T-wave cell, with the same argon pressure as the trap, passed on the ion mobility separated ions to the orthogonal time-of-flight for m/z analysis, where the ion mobility–mass spectrum was acquired by synchronizing the gated release of ions into the mobility T-wave ion guide with the time-of-flight mass spectral acquisition.

Analysis of the amb5A-G IM-MS Spectra

The IM-MS analysis allowed for drift time separation and m/z identification of all the various amb5 species including coincidental m/z species. The positively and negatively charged species were identified based on their m/z and isotope patterns, and their drift time distributions were extracted using Driftscope 2.0. The centroid of each drift time distribution and the area under the drift time curve was determined using MassLynx 4.1. The centroids of the drift time distributions were used to calculate the collision cross sections of the various amb5 species, and the areas under the drift time curves were used to calculate their percent relative intensities at each pH step. The experiments were repeated 2–5 times to estimate the standard deviations.

Collision-Induced Dissociation and Breakdown Curves

To study the energy-dependent CID of A to G, the isotope distribution of the chosen precursor ions were resolved by the linear quadrupole, passed through the trap and the IM cells, and dissociated in the transfer T-wave cell using the lab-frame collision energy (CE) controlled by the entrance lens which resulted in reproducible breakdown curves [65]. The precursor and product ions were m/z analyzed by the TOF and identified by matching their isotope patterns to their theoretical isotope patterns from the MassLynx 4.1 software. The percent relative intensities of the precursor and product ions were calculated by normalizing their area under the drift time curve to the total area of all species produced at each CE.

Collision Cross Sections

Collision cross sections (CCSs) of the amb5 species were measured using a calibration method that provides CCS as measured in He gas (ΩHe) and allowed us to compare the ΩHe with those calculated by the Sigma program. In this method, previously measured CCS of poly-DL-alanine (PA) for negative [63, 66] and positive ions [62] measured in He gas (ΩHe) [67] were used to determine the ΩHe of the amb5 species. The PA ΩHe were converted to corrected CCS (Ωc) by Eq. (1), where z = charge of ion, ec = 1.602 × 10−19 C, mN2 = mass of N2 (Da), and mion = mass of the ion [68].

$$ \Omega \mathrm{c}=\frac{\Omega_{\mathrm{He}}}{z{e}_c\kern1em \sqrt{\left(\frac{1}{m_{N_2}}+\frac{1}{m_{\mathrm{ion}}}\right)}} $$
(1)

From the IM-MS analyses, the centroid of the measured arrival times (tA) for the calibrants and amb5 species were converted to centroid IM drift times (tD) using Eq. (2), where c = enhanced duty cycle delay coefficient (1.41).

$$ {t}_D={t}_A-\frac{c\kern0.5em \sqrt{m/z}}{1000} $$
(2)

The tD for the calibrants were plotted against their Ωc and using a least-squares regression fit the A′ and B constants were obtained using Eq. (3), where A′ is the correction for the temperature, pressure, and electric field parameters and B compensates for the nonlinear effect of the IM device. Four separate calibration curves were constructed for the singly and doubly charged species from the negative and positive ion analyses.

$$ {\Omega}_{\mathrm{c}}=A^{\prime }{t_D}^B $$
(3)

Using these A′ and B values the tD for the amb5 species were converted to Ωc using Eq. (3), and ΩHe using Eq. (1). The reported ΩHe are an average from 3 to 5 separate measurements and have estimated absolute errors of 2% [62, 63, 66].

Computational Methods

The B3LYP/LanL2DZ level of theory containing the Becke three parameter hybrid functionals with the Dunning basis set [69] and electron core potentials [70,71,72] from Gaussian 09 [73] was used to locate the geometry-optimized, gas-phase structures of species A to G. Optimization of gas-phase structures was chosen because our previous research on ac-His-Cys-Tyr-Pro-His-Cys (amb2) and A showed the inclusion of the SMD aqueous model produced similar results as the gas-phase method [13, 14]. For each calculation, initial structures were made in GaussView 5 based on various possible interactions between the substituent sites of His, Cys, Tyr, and the C-terminus which resulted in the overall charge of the conformer, while keeping all the amide bonds in the peptide backbone in their trans-configuration including Pro4. Optimized structures were modified using GaussView 5 to produce new starting structures for optimizations to further sample the conformational space of the conformers.

The CCSs were calculated for the geometry-optimized complexes using the ion size scaled Lennard-Jones (LJ) method from the Sigma program, developed and written by the Bowers group from the Department of Chemistry and Biochemistry at the University of California - Santa Barbara [74]. The LJ values are derived from the mean and standard deviation from 10 repeated calculations.

Results and Discussion

Positive Ion Charge Distributions of amb5A-G over pH 3 to pH 11

As discussed earlier, the hexaprotic amb5A (Scheme 1A) should exhibit an overall charge state based on the protonation state of the 2His, 2Cys, Tyr, Pro, and carboxyl group of the C-terminus. The relative populations of the two observed charge states [amb5A-G + H]+ and [amb5A-G + 2H]2+ electrosprayed from pH 2.9 to pH 10.9 solutions are shown in Figure 1 a and b, respectively. At pH 2.9, the species with both His1 and His6 imidazolium groups in the primary structure, A, B, E, and F, exhibited 75–82% intensity of the 2+ charge state, [amb5 + 2H]2+, which decreased to 56–61% as pH increased to pH 4.7. For A, E, and F, the deprotonation of the C-terminus carboxyl group, which favorably deprotonates over this pH range, could explain the overall charge state decrease from 2+ to 1+. For B, the deprotonation of the His imidazolium or the C-terminus amide group [75] was needed to decrease the charge, with B3LYP/LanL2DZ predicting the deprotonation of His1 was 29 kJ/mol more favorable than deprotonation of the C-terminus amide group (Figure S1). Above pH 4.7, the relative intensities of the 1+ and 2+ charge states did not change significantly, indicating that they were not influenced by solution-phase equilibria. For C, D, and G, the single-charged [amb5 + H]+ predominated over the whole pH 2.9 to 10.9 range because they either contained just one or zero imidazolium groups in their primary structure. For C, the remaining His1 imidazolium could hold the 1+ charge, but for D or G, the 1+ charge site was likely either Pro4, which exhibited the next highest gas-phase basicity after His [21,22,23] or solvation of the proton by carbonyl and/or thiol groups, with B3LYP predicting the solvation of the proton by two thiols was 14 kJ/mol less favorable in free energy than the protonation of Pro4 (Figure S2).

Figure 1
figure 1

Relative intensities of the positive charge states of species A to G, (a) [amb5 + H]+ and (b) [amb5 + 2H]2+over the pH range of pH 2.9–10.9. The lowest free energy conformers located using the B3LYP/LanL2DZ method (c) [amb5A + H]+ and (d) [amb5A + 2H]2+ with their theoretical collision cross sections (ΩHe) measured by the ion size scaled Lennard-Jones method [74]. The dashed lines represent the salt-bridges ( ) and hydrogen bonds ( ) that stabilize their tertiary structures with labels referred to in the text

Full size image

Collision Cross Sections and Tertiary Structures of Positive Ions

The CCS as measured in helium gas (ΩHe) of the different charge states were measured by IM-MS (Table 1) and compared to the ΩHe of the low energy conformers located from the B3LYP/LanL2DZ method as calculated by the LJ method. For [amb5A + H]+, the three lowest energy B3LYP conformers differed by < 1 kJ/mol in free energy and all had trans-Pro4 forming the hinge of the two arms, with the imidazolium group of His1 salt-bridged to the carboxylate group of the C-terminus (SB-1 in Figure 1c) and 4 hydrogen bonds, including His6 hydrogen bonded to backbone carbonyl groups and the hydroxyl group of Tyr5 hydrogen bonded to the carboxylate of the C-terminal (H-1) or the acetyl group of the N-terminal. These intramolecular interactions resulted in folded conformers with theoretical CCS (ΩHe) of 198 to 199 Å2, which all agreed with the IM-MS measured ΩHe = 199 ± 4 Å2 (Table 1). The lowest free energy conformer (ΩHe = 198 ± 2 Å2, Figure 1c) included three hydrogen bonds from the substituents of His6 (H-2, H-3) and Cys7 (H-4) to backbone amide groups. These interactions were exhibited in the tertiary structures of other [amb5 + H]+ species (Figures S1 and S3) with theoretical ΩHe that also agreed with the experimentally measured IM-MS ΩHe in Table 1.

Table 1 Collision Cross Sections (Å2) Measured in Helium (ΩHe) Buffer Gases for Positive and Negative Ions of amb5A-G (H Loss or Gain Not Shown) by Ion Mobility–Mass Spectrometry with Estimated Absolute Uncertainties of ± 4 Å2
Full size table

The two lowest free energy 2+ conformers of A located by B3LYP/LanL2DZ differed by only 14 kJ/mol, but they exhibited different CCS of ΩHe = 216 ± 4 Å2 (Figure 1d) and 247 ± 3 Å2 (Figure S4), with only the ΩHe of the former matching the IM-MS measured ΩHe = 216 ± 2 Å2 (Table 1). This 2+ conformer 1d exhibited four hydrogen bonds from the two imidazoliums to carbonyl or thiol groups and another hydrogen bond between Tyr5 and the C-terminus which stabilized its overall conformation. The more extended conformer (Figure S4) also exhibited four hydrogen bonds and was predicted by B3LYP to be the lowest in free energy. This indicates that the B3LYP calculations should not be used alone, and comparison with the experimental IM-MS measured ΩHe is necessary. This is especially relevant as the ESI process might produce kinetically trapped species rather than the lowest energy structures. Also, there maybe more than just one conformer present as indicated by the three lowest energy 1+ conformers that differed by < 1 kJ/mol and had similar sized CCS and hydrogen-bonded structures. The CID results discussed later also support the selected conformers because the removal of the imidazolium groups, which are integral to these salt-bridged and hydrogen-bonded conformers, result in a substantial decrease of 20 eV (lab-frame) in their breakdown energies.

Negative Ion Charge Distributions of amb5A-G over pH 3 to pH 11

Comparison of the relative intensities of the 1- and 2- charged species (Figure 2a, b) showed the principal species was [amb5 − H]− which decreased over pH 2.9–10.9 with [amb5 − 2H]2− exhibiting the greatest increase (10–28%) between pH 9.9 and 10.9. The potential negative charge sites are the Cys, Tyr, or C-terminus, but it was the substitution of the 2His by 2Gly or 2Cys that most adversely affected the relative intensity of the 2- charge state as observed for D and G. At first, this seems counterintuitive because the removal of a positively charged imidazolium group may be expected to make the 2- charge state more probable. However, the pKas of His and Cys bracket pH 7 by ± 1 unit and interactions between the weak imidazole bases and weak thiol acids can stabilize the negative charge of the deprotonated thiolates as demonstrated by the B3LYP modeling and CCS measurements described next.

Figure 2
figure 2

Relative intensities of the negative charge states of species A to G, (a) [amb5 − H]− and (b) [amb5 − 2H]2− over the pH range of pH 2.9–10.9. The lowest energy conformers located using the B3LYP/LanL2DZ method (c) [amb5A − H]− and (d) [amb5A − 2H]2− with theoretical collision cross sections (ΩHe) measured using the ion size scaled Lennard-Jones method [74]. The dashed lines represent the salt-bridges ( ) and hydrogen bonds ( ) that stabilize their tertiary structures

Full size image

Collision Cross Sections and Tertiary Structures of Negative Ions

The B3LYP conformer located for the 1- charge of A (Figure 2c) exhibited four salt-bridges between the imidazoliums of His1 and His6 and the thiolates of Cys2, Cys7, and carboxylate of the C-terminal. These salt-bridges were produced by the B3LYP geometry optimization even if the protons started on the thiol groups because the proton transfer to the imidazoles proceeded without an energy barrier. The stability contributed by these salt-bridges to the 1- conformer was also supported by the negative charge state distributions which showed the 1- charge state remained the predominant species over the pH 2.9–10.9 range. The lowest energy conformer 2c included hydrogen bonds between the C-terminus and hydroxyl group of Tyr5, and the backbone amide of Tyr5 and carbonyl group of Pro4 resulting in a folded structure with a theoretical ΩHe = 196 ± 2 Å2 which matched the IM-MS measured ΩHe = 196 ± 4 Å2 (Table 1). The next two lowest B3LYP conformers located had fewer salt-bridges of three and one, respectively, and higher free energies of 28 kJ/mol and 57 kJ/mol than 2c. Salt-bridges were present in the tertiary structures of the lowest free energy conformers of B and C (Figure S5), whose LJ ΩHe were also in agreement with the IM-MS ΩHe measurements (Table 1). Unlike with the positive ions, the negative ion conformers had larger free energy differences between conformers and our approach of using B3LYP to locate conformers and compare them with the IM-MS measured CCS and charge state distributions revealed the significant stability of the intramolecular salt-bridge network for conformer 2c. Moreover, the breakdown energies discussed later support the stability of the 2c conformer.

The lowest free energy 2- conformer of A (Figure S6) was folded by hydrogen bonds between the imidazoles, thiol(ate), and carboxylate in a similar way to 2c. However, this conformer’s theoretical ΩHe = 195 ± 2 Å2 was 23 Å2 smaller than the IM-MS measured ΩHe = 218 ± 2 Å2 (Table 1), which suggested a more open structure existed. A more extended structure 26 kJ/mol higher in free energy (Figure 2d) exhibited a tertiary structure with hydrogen bonding between the 2His, 2Cys, and acetylated N-terminus, with Tyr5 also hydrogen bonded to the C-terminus. This conformer’s LJ ΩHe = 222 ± 4 Å2 was in good agreement with the IM-MS ΩHe = 218 ± 2 Å2.

Collision-Induced Dissociation Studies and Breakdown Energies

The CID spectra of [amb5A-G]+ exhibited a series of a, b, and y-type product ions [76], and A, B, E, and F exhibited the most similar CID spectra because they all contained the His1 and His6 residues (Figure S7). The C and D precursors differed by not producing the y2 ion (C), or the y2, b2, b1, and a1 ions (D), which only formed when the His1 or His6 were present, consistent with previous studies that showed the b1 ion formed if the N-terminal substituent group can transfer a proton to the N-terminal amine N and/or cleave the first amide bond via nucleophilic attack [45, 55]. When both His residues were replaced by Gly (D), most product ions contained the Pro4 residue which supported that the enhanced gas-phase basicity of Pro4 made it a favorable proton site after His [22]. The only exception was the b3 ion, which was observed from species A to F, and was previously determined to form an oxazolone structure [58, 59]. The formation of the y4 ion which was also observed from A to F was consistent with the Pro effect [51] which enhances preference for cleavage of the amide bond on the N-terminus side of the Pro residue.

The breakdown curves (Figure 3) of the A to G precursor ions as a function of lab-frame CE show the primary products, which appeared at the lowest CE; dissociated into secondary products, which appeared at higher CE; and dissociated into the tertiary products, which appeared at the highest CE. The identity of the product ions in each group are shown in Tables S1 and S2. Comparison of the CE where the [amb5 + H]+ of A to G was 50% dissociated into products (Figure 3, Table 2) shows a general trend related to size and the number of vibrational modes available to absorb the CE. For example, A and B whose mass only differed by 1 Da exhibited the same 47.1 Vlab breakdown energy and a similar pattern for the formation of the primary, secondary, and tertiary products.

Figure 3
figure 3

Energy-resolved CID of the negatively and positively charged precursor A to G with products (Tables S1 and S2) grouped into the primary (Pri), secondary (Sec), and tertiary (Tert) products as described in the text. The lab-frame transfer collision energy where the A to G precursor was 50% dissociated into products is also shown

Full size image
Table 2 Breakdown Energies for Positive and Negative Ions of the amb5A-G Measured by Ion Mobility–Mass Spectrometry, Converted from the Lab-Frame (CE) to the Center-of-Mass Frame (Ecm) with the Number of Collisions Each Species Undergoes in the Transfer T-Wave Cell, and C-Terminus Interactions
Full size table

To correct for inelastic collisions, where the initial lab-frame energy is converted into translational, rotational, or vibrational energy, the CE was converted to center-of-mass CE, Ecm = CE MAr/(MAr + Mamb) where MAr is mass of the Ar collision gas and Mamb is mass of the amb5 peptide (Table 2). Collision theory predicted that the amb5A-G precursors undergo ~ 600–700 collisions as they pass through the transfer cell. The [amb5 + H]+ of A, B, E, and F exhibited the highest breakdown energies, Ecm = 2.09 to 2.10 eV, which coincided with their lowest energy B3LYP conformers (Figures 1c, S1, and S3) containing the imidazolium of His6 hydrogen bonded to two backbone carbonyl groups, and salt-bridges or hydrogen bonds from the His1 or Tyr5 substituent groups to the C-terminus (Table 2). Species C, D, and G exhibited the lowest breakdown energies, Ecm = 1.02 to 1.89 eV, because the replacement of the His residues with either Gly or Cys removed these salt-bridge or hydrogen-bonding interactions.

The principal CID products of the negatively charged [amb5 − H]− (Figure S7c, Table S2) for A, C, and D were the loss of H2S, 2H2S, 2H2S + CO2 and the formation of the z6 − 2H2S ion. The H2S + CO2 loss channel is favored when Cys is the C-terminal residue as previously observed from the Gly-Cys dipeptide [47, 49]. Species B did not exhibit 2H2S + CO2 loss, because CO2 loss could not form from the amidated C-terminus. Species E and F exhibited CO2 loss and formation of c4 and c1 ions that were only observed after the thiol of Cys2 was removed from the primary structure. Previous CID studies of negative ion peptides have also reported the common occurrence of c ions [37, 75].

The breakdown curves of the [amb5 − H]− (Figure 3) of A and C were similar and exhibited the same Ecm = 1.82 eV breakdown energy. However, species B and G exhibited much lower breakdown energies of Ecm = 1.26 eV and 1.23 eV, respectively, with B showing exclusive H2S and 2H2S loss, while G also showed these channels with a range of secondary and tertiary products (Table S2). These observations indicated the amidation of the C-terminus made H2S and 2H2S loss more labile. Previous work at the HF/6-31G(d)//AM1 level of theory showed H2S loss was initiated by formation of an enolate anion on the peptide backbone and passage over 0.87 eV transition state barrier followed by stepwise elimination of HS− and proton transfer [47]. However, when Cys was the C-terminus residue, H2S + CO2 loss was the prominent channel which occurred through a carboxylate charge-directed mechanism with a higher transition state barrier of 2.9 eV [47]. The amidation of the C-terminus, therefore, restricts the charge-directed mechanism and allows the H2S loss to proceed over the lower energy barrier. The amidation of the C-terminus also meant the salt-bridge between His1 and the carboxylate of the C-terminus, exhibited by the 1- conformers of A (Figure 2c) and C (Figure S5b) changed to a hydrogen bond for conformer B (Figure S5a). The His1 and the salt-bridge were also removed in D which exhibited a similarly low Ecm = 1.58 eV breakdown energy. However, the 1- conformers of E and F had the highest breakdown energies of Ecm = 1.87 eV and 2.47 eV, respectively, because E produced H2S + CO2 loss which had to pass over the high transition state barrier, and the thiol-free primary structure of F did not exhibit the labile H2S loss.

Zn(II)-Binding and Charge Distributions Between pH 5 and 10

How changes in the primary structure of A affected the Zn(II) chelation was tested by the IM-MS analyses of equimolar mixtures of either A to G with Zn(II) as a function of pH. Samples of the mass spectra for negative and positive ion species are shown in Figure 4. The main observed species included both singly and doubly charged amb5A-G and amb5A-G + Zn(II). The overall charge of the species was conserved by the loss or gain of protons, i.e., [amb5 − H]−, [amb5 − 2H]2−, [amb5 − 3H + Zn(II)]−, [amb5 + 2H]2+, [amb5 − H + Zn(II)]+, etc. Lesser signal for chloride adducts were observed, i.e., [amb5 − 2H + Zn(II) + Cl]−, and some dimers and trimers, i.e., [triamb5 − 7H + 3Zn(II)]−. A principal species was [amb5 − 3H + Zn(II)]− relating to the loss of 3 protons from the 2Cys and C-terminus which were the potentially important Zn(II)-chelating sites.

Figure 4
figure 4

Labeled mass spectra of the products observed from the reactions of (a) amb5A, (b) amb5B, (c) amb5C, (d) amb5D, (e) amb5E, (f) amb5F, or (g) amb5G with equimolar ZnCl2 at pH 9. Left panel: negative ions. Right panel: positive ions

Full size image

Distributions of Negatively Charged Species from amb5A-G + Zn(II) over pH 5 to pH 10

The analyses of negative ions (Figure 5) shows that all species A to G were principally in the [amb5 − H]− form at pH 5 or 6, but as pH increased to pH 7 and higher, the relative intensity of [amb5 − 3H + Zn(II)]− increased and became the principal species, apart from species F. Previous research of A and B indicated that the most favorable Zn(II) chelation was via the 2His-2Cys, C-terminus, and backbone carbonyl oxygen [14]. Figure 5 shows that species A, B, and C were predominantly in the [amb5 − 3H + Zn(II)]− form by pH 7, even though B had the blocked C-terminus which excluded Zn(II) chelation by the deprotonated carboxylate group, and C had His6 replaced by Gly6. Species D and E also exhibited formation of [amb5 − 3H + Zn(II)]− as the major species at pH 9–10 showing Zn(II) chelation was still favorable at this pH when neither imidazole or the thiol of Cys2 were present in the primary structure. Species D gave only trace formation of [amb5 − 4H + Zn(II)]2−, indicating that the basic properties of His which were present in A, B, C, and E enhanced these latter species deprotonation. Species F mainly formed [amb5 − H]− over pH 5–10 and exhibited < 12% of [amb5 − 3H + Zn(II)]− and [amb5 − 4H + Zn(II)]2− at any pH which showed that Zn(II) chelation was most dependent on the availability of at least one thiolate site. However, species G which had 4Cys in its primary structure was less effective at chelating Zn(II) than A, B, and C which all had His1 with the 2Cys in their primary structure over pH 7–8.

Figure 5
figure 5

Percent relative intensities of [amb5 − H]−, [amb5 − 3H + Zn(II)]−, and [amb5 − 4H + Zn(II)]2− measured from equimolar mixtures of Zn(II) and either (a) amb5A, (b) amb5B, (c) amb5C, (d) amb5D, (e) amb5E, (f) amb5F, or (g) amb5G over the range pH 5–10, and (h) the total percentage of all positively and negatively charged Zn(II)-bound species observed from these equimolar mixtures. The error bars are one standard deviation about the mean

Full size image

Distributions of Positively Charged Species from amb5A-G + Zn(II) over pH 5 to pH 10

A similar pH-dependent behavior to the negative ions was observed for positive ions (Figure S8) with A to G predominantly forming either [amb5 + 2H]2+ and/or [amb5 + H]+ at pH 5–6 and as pH increased the formation of [amb5 − H + Zn(II)]+ and [amb5 + Zn(II)]2+ became more significant. This indicated that the weakly basic and acidic substituent groups of His and Cys were playing a key role in chelation for the positive ions, because other possible chelating groups such as the backbone carbonyl oxygens or the carboxyl of the C-terminus would not exhibit this pH dependence. For the overall 2+ charge of A, [amb5A + Zn(II)]2+, the B3LYP modeling predicted the lowest energy Zn(II) coordination was via the two thiolates of the 2Cys, with 2His protonated as imidazoliums (Figure S9). Other options such as the imidazole coordination of Zn(II) and the Cys protonated as thiols were higher in energy and indicated that Cys could not chelate Zn(II) via the protonated thiol groups. Species A, B, C, D, E, and G exhibited increases of [amb5 − H + Zn(II)]+ and [amb5 + Zn(II)]2+ over pH 5–10, with the exceptions of D and G which did not form or formed < 10% of [amb5 + Zn(II)]2+, which coincided with their primary structures not containing His. Species F was the only species that exhibited a decrease in the formation of [amb5 − H + Zn(II)]+ and [amb5 + Zn(II)]2+ over pH 5–10, even though its sequence contained two imidazoles that were potential Zn(II) chelation sites.

Comparative Zn(II) Chelation by amb5A-G from Both Positive and Negative Ion Analyses

To determine the overall effectiveness of A to G for Zn(II) chelation, a plot that includes the relative percentages of all Zn(II)-bound species from the positive and negative charge analyses are shown in Figure 5h. At pH 5 and 6, all species exhibited ~15–40% Zn(II) chelation with the species with the amidated C-terminus, B and G, exhibiting the lowest binding, indicating the C-terminus carboxylate was involved in chelating Zn(II). As pH increased (pH ≥ 7) A, B, and C exhibited the greatest percent increase of Zn(II) chelation, with 65–92% of species chelating Zn(II). The modifications to the primary structure of A by substituting 2His with 2Gly (D) and Cys2 with Gly2 (E) decreased the percent of Zn(II) chelation to 38–62% over pH 7 to 10, whereas substituting 2Cys with 2Gly (F) further decreased Zn(II) chelation from 35 to 10% over pH 5 to 10. The substitution of 2His with 2Cys (G) also lowered Zn(II) chelation to ~ 45% over pH 7–8, compared to 70–90% for A, B, and C, indicating that the weak base property of His was required for efficient Zn(II) chelation over this pH range. Only at pH 9–10 was Zn(II) chelation exhibited by G equivalent to that exhibited by A, B, and C. This pH-dependent behavior showed that the amb5 peptides containing His1 and 2Cys exhibited the most effective Zn(II) chelation and provided evidence for the His role as proton acceptor for the deprotonation of Cys which acts as the most favorable Zn(II) anchoring site at pH 7–10. The two imidazoles did not effectively chelate Zn(II) at pH > 7 as evidenced by species F which exhibited the lowest Zn(II) chelation because its primary structure did not contain the two thiol groups. As pH becomes more basic, competition from the formation of Zn(OH)2 may necessitate that thiolate ligands are present in the amb5 sequence for effective Zn(II) chelation.

Collision Cross Sections and Zn(II) Coordination Exhibited by Positive and Negative Ions

A series of geometry-optimized conformers were located using the B3LYP/LanL2DZ method for [amb5 − 3H + Zn]− (Figure 6). The lowest free energy conformer of A contained cis-Pro (Figure 6a), which is the same cis-Pro conformer reported in our previous paper [14] but optimized with the acetylated N-terminal arm in the trans-configuration that lowered the free energy by 3.3 kJ/mol. The conformer exhibits Zn(II) chelation via 2Cys, carbonyl O of Cys2, and the carboxylate of the C-terminus. However, the LJ CCS of conformer 6a, ΩHe = 229 ± 2 Å2, is significantly larger than the IM-MS measured ΩHe = 200 ± 4 Å2 (Table 1) and this disparity was reported in our previous paper [14]. In our present research, a new series of trans-conformers were located and the lowest free energy trans-Pro conformer (Figure 6b) exhibited His1-Cys2-Cys7 and carboxylate C-terminus Zn(II) coordination with ΩHe = 205 ± 2 Å2, in much better agreement with the IM-MS ΩHe = 200 ± 4 Å2. Previous research [6, 77] has shown that IM-MS analyses can distinguish between the cis- and trans-Pro conformers of oligopeptides. The B3LYP/LanL2DZ method predicts that the trans-Pro conformer 6b is 125 kJ/mol higher in free energy than the cis-Pro conformer 6a. However, all the peptides A to G were originally synthesized with trans-peptide bonds and so the energy barrier for conversion of trans-Pro4 to cis-Pro may restrict its formation. This is further supported by the ΩHe of the trans-Pro conformers of B (Figure 6c) and C (Figure 6d) which also exhibit ΩHe in agreement with the IMMS measured ΩHe (Table 1), whereas their cis-Pro conformers were significantly larger (Figure S10). The trans-Pro conformers with their preference for His1-Cys2-Cys7 and C-terminus Zn(II) coordination also explains why A, B, and C are the most effective Zn(II) chelators because they contain all these chelating groups, whereas the Zn(II)-chelating ability of D, E, F, and G were all adversely affected when one or two of these chelating sites were modified.

Figure 6
figure 6

Comparison of the conformers located using the B3LYP/LanL2DZ method (a) cis-Pro [amb5A − 3H + Zn]− exhibiting 2Cys, backbone carbonyl and C-terminus coordination of Zn(II), (b) trans-Pro [amb5A − 3H + Zn]− exhibiting His1-2Cys and C-terminus coordination of Zn(II), (c) trans-Pro [amb5B − 3H + Zn]− exhibiting His1-2Cys and the amidated C-terminus coordination of Zn(II), and (d) trans-Pro [amb5C − 3H + Zn]− exhibiting His1-2Cys and C-terminus coordination of Zn(II), with the relative free energies (G), theoretical collision cross sections (ΩHe), [74] and hydrogen bonds () shown

Full size image

Conclusions

Using IM-MS and B3LYP/LanL2DZ computational studies, we have investigated the oligopeptide acetyl-His1-Cys2-Gly3-Pro4-Tyr5-His6-Cys7 which contained the 2His-2Cys motif and a series of oligopeptides that have modifications to the potential His, Cys, or C-terminus Zn(II)-binding sites. The results show that at pH ≥ 7, the weak basic properties of His are intrinsically linked to the weak acid properties of Cys through a proton transfer reaction which affects the charge states, tertiary structures, and Zn(II) coordination of the oligopeptides studied here. The positive and negative charge states were primarily influenced by the imidazolium groups and their interactions with the thiolates and carboxylate of the C-terminal. The lowest free energy B3LYP conformers predicted a salt-bridged interaction between the imidazolium and C-terminus carboxylate and the deprotonation of the thiols by the imidazoles acting as proton acceptors and providing electrostatic stabilization of the thiolates via salt-bridging or hydrogen bonding. These acid-base interactions of His and Cys also lowered the pH at which Zn(II) chelation effectively occurred, indicating that imidazoles act as proton acceptors for the deprotonation of the thiol groups that are the primary anchoring sites for Zn(II). This was supported by the DFT conformers which contained trans-Pro and His1-Cys2-Cys7 and C-terminus Zn(II) coordination whose theoretical ΩHe were in good agreement with the IM-MS measured ΩHe.