Unfolding and Folding of the Three-Helix Bundle Protein KIX in the Absence of Solvent
Electron capture dissociation was used to probe the structure, unfolding, and folding of KIX ions in the gas phase. At energies for vibrational activation that were sufficiently high to cause loss of small molecules such as NH3 and H2O by breaking of covalent bonds in about 5% of the KIX (M + nH)n+ ions with n = 7–9, only partial unfolding was observed, consistent with our previous hypothesis that salt bridges play an important role in stabilizing the native solution fold after transfer into the gas phase. Folding of the partially unfolded ions on a timescale of up to 10 s was observed only for (M + nH)n+ ions with n = 9, but not n = 7 and n = 8, which we attribute to differences in the distribution of charges within the (M + nH)n+ ions.
KeywordsElectron capture dissociation Gas phase Native mass spectrometry Protein Protein folding
Native mass spectrometry (MS) has, over the past 25 years, developed from interpreting mass spectra from electrospray ionization (ESI) of different solutions to approaches by which the dissociation of biomolecules such as proteins and nucleic acids and their noncovalent complexes is studied, or their rotationally averaged collision cross section is probed by ion mobility MS [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29]. Obtaining information of relevance to biological problems by native ESI MS relies, as a matter of course, on the preservation of solution structure after transfer into the gas phase, where it can be probed by a number of techniques, including electron capture dissociation (ECD) [30, 31]. However, the use of native ESI “is not without complications” , not least because the stability of a solution structure in the gas phase can presently not be reliably predicted. Nevertheless, some progress has been made in understanding the determinants of peptide and protein structure and stability in a gaseous environment [32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47].
We have previously postulated that salt bridges, either formed in the gas phase or already present in solution , can contribute substantially to the stabilization of the solution structure of a protein after transfer into the gas phase . In support of this hypothesis, calculations suggest that in the absence of solvent, the strength of an overall neutral salt bridge can be comparable to the strength of a covalent bond [49, 50]. Ionic hydrogen bonds  between protonated basic or deprotonated acidic sites and neutral molecules have strengths of typically 21–146 kJ/mol, which is up to a third of the strength of covalent bonds . Neutral hydrogen bonds between backbone amides constitute the basis of protein secondary structure (i.e., α-helices and β-sheets). The N-H⋅⋅⋅O=C hydrogen bond stability in the gas phase can be estimated to be close to the dimerization energy of two N-methylacetamide molecules of ~28 kJ/mol . Although in the gas phase neutral hydrogen bonds are substantially weaker than both ionic hydrogen bonds and salt bridges, they are typically more numerous in proteins with a high content of secondary structure. Finally, ion–dipole interactions, especially those that involve helix dipole moments, can stabilize protein ion gas phase structure . In the absence of sufficient stabilization by electrostatic interactions, desolvation can cause spontaneous unfolding of protein ions [53, 54, 55, 56, 57], which can subsequently fold into more stable gaseous ion structures . However, only a small number of experimental studies [34, 58, 59, 60, 61] have so far focused on peptide or protein folding in the gas phase, even though all possible structural transitions can affect data from native mass spectrometry experiments. We have recently reported that gaseous cytochrome c ions from horse and tuna heart, the fold of which is virtually identical in solution, show vastly dissimilar folding behavior, and found evidence that the formation of salt bridges is a major driving force for protein folding in the gas phase .
Here we investigate the unfolding and folding of the three-helix bundle protein KIX, for which ECD data indicated substantial preservation of the native solution structure in the (M + 7H)7+ ions, on a timescale of at least 4 s after transfer into the gas phase, even after vibrational ion activation by 28 eV collisions with argon gas . In this study, we subjected KIX (M + nH)n+ ions with n = 7–9 to higher energy collisions that were sufficiently high to break covalent bonds in ~5% of the ion populations, and discuss the observed partial unfolding in terms of electrostatic interactions. Folding data for the partially unfolded KIX ions is complemented with data for a KIX peptide, which are discussed in the context of Coulombic repulsion and possible interactions that drive the folding process.
Results and Discussion
The 20 KIX structures in pdb entry 2AGH, calculated on the basis of distance restraints from nuclear magnetic resonance (NMR) experiments in the Wright group , show a highly uniform backbone fold especially in the α-helical regions (α1: residues 16-30, α2: 42-61, α3: 65–88) but substantial variations in sidechain orientation of basic (H, K, R, N-terminus) and acidic (D, E, C-terminus) residues (Figure 1a). Other KIX structures (1KDX, 2LXS, 2LXT, 2LQH, 2LQI, 2KWF) exhibit essentially the same backbone fold and similarly large variations in sidechain orientation, notwithstanding the fact that all corresponding NMR experiments were conducted in the presence of different peptide ligands and under different solution conditions, i.e., different ionic strength (0–50 mM NaCl), buffers (Tris(hydroxymethyl)aminomethane acetate, potassium phosphate, 2-(N-morpholino)ethanesulfonic acid) at different concentrations (20–50 mM), and different pH values (5.5–6.0). Apparently, the conformational flexibility of KIX's basic and acidic sidechains is generally high in solution, although some electrostatic interactions such as the salt bridge between R19 and E55 (Figure 1a) were found in most of the above NMR structures.
We have previously proposed that the transfer of proteins into the gas phase by electrospray ionization causes the formation of salt bridges and ionic hydrogen bonds on the protein surface, by which a native fold can be stabilized during and after the phase transition [32, 48]. The high stability of KIX (M + 7H)7+ ions in the complete absence of solvent, on a timescale of up to 4 s , suggests that a sufficiently large number of electrostatic interactions have formed during ESI that, together with those already present in solution, prevent the native fold from disintegration in the gas phase. All salt bridges (SB, purple lines) and ionic hydrogen bonds (IHB, green lines) that are present in solution or can potentially form during ESI while retaining the backbone fold of the KIX structure 2AGH are illustrated in Figure 1b.
Comparing the possible salt bridges and ionic hydrogen bonds (Figure 1b) to the site-specific yields of c (black bars) and z• (open bars) fragments from ECD of KIX (M + 7H)7+ ions (Figure 1c, 0 eV) reveals that the possible interactions of R7, K8, H11, and H13 (SB: R7/E12, R7/E60, K8/E60, H11/E60, H13/E12, IHB: R7/N63, K8/W10) were, at least in a significant fraction of the ions, not present as this would have prevented separation of fragments from cleavage at sites 7, 8, and 12 . However, after vibrational activation for unfolding of the KIX (M + 7H)7+ ions (Figure 1c, 133 eV), the yield of separated c and z• fragments from cleavage at sites 7 and 12 increased substantially, and fragments from sites 10 and 11 appeared, consistent with an increase in the fraction of ions in which the interactions of R7, K8, H11, and H13 were broken. Moreover, separated fragments were observed from cleavage in the region of helix α1 (sites 15, 16, 20, 21, 25, 28), indicating loss of its secondary structure along with breaking of any interactions between helices α1 and α2 (SB: R19/E55, IHB: R19/Y59). The ECD patterns of the KIX (M + 8H)8+ ions with and without collisional activation are very similar (Figure 1d), although breaking of SB and IHB interactions between residues 42, 43, 45, and 46 is evident from the ~4-fold increase in yield of fragments from sites 42, 43, and 45. However, collisional activation had a far stronger effect on the structure of the KIX (M + 9H)9+ ions, the fragmentation pattern of which at 0 eV (Figure 1e) was very similar to that of the (M + 8H)8+ ions at 128 eV. Specifically, the data indicate nearly full unraveling of helix α2 along with breaking of the IHB between helices α2 and α3 (Y50/K75, E55/Y68), and significant loss of residual structure in the region comprising residues 6–57 (note that ECD does not produce c and z• ions from cleavage at the N-terminal side of proline residues, which applies to sites 31, 33, and 35).
In summary, the above data suggest that unfolding of the three-helix bundle structure of the KIX (M + 7H)7+ ions by collisional activation, at energies that are sufficiently high to cause loss of small molecules such as NH3 and H2O by breaking of covalent bonds in about 5% of the ions, is limited to the separation and disruption of helix α1 while retaining the higher order structure of helices α2 and α3. By contrast, helix α1 is already unraveled and separated from helices α2 and α3 in a significant fraction of the (M + 8H)8+ ions produced by ESI, along with uncoiling of the first turn of helix α2, and collisional activation merely increases this fraction without causing additional structural changes. The structure of the (M + 9H)9+ ions from ESI is very similar to that of the partially unfolded (M + 8H)8+ ions, but after collisional activation, substantial loss of tertiary and secondary structure is observed, even though the latter is largely retained in helix α3.
However, the data in Figure 3 show that charge transitions of the (M + 9H)9+ ions were different from those of the (M + 7H)7+ and (M + 8H)8+ ions. For example, the transition from one to two charges for c ions (left axis, the corresponding transition for complementary z• ions is from four to five charges on the right axis) is around site 16 for (M + 7H)7+ ions but around site 13 for (M + 9H)9+ ions, and that from three to four charges is around site 45 for (M + 8H)8+ ions but around site 38 for (M + 9H)9+ ions. It is generally difficult to pinpoint the exact location of all charged sites in protein (M + nH)n+ ions from ECD data [30, 71] as capture of an electron neutralizes a positive charge, and because the presence of zwitterionic motifs that comprise both positively and negatively charged sites cannot be excluded. Moreover, even in unfolded structures, protons can be shared between adjacent residues in homodimeric (e.g., K⋅⋅⋅H+⋅⋅⋅K) or heterodimeric (e.g., K⋅⋅⋅H+⋅⋅⋅Q) ionic hydrogen bonds . Nevertheless, the different charge transitions in Figure 3 clearly imply that charges are distributed differently in the 1–46 region for n = 7, 8 and n = 9. Because this is the only difference between the (M + 9H)9+ ions and the (M + 7H)7+ and (M + 8H)8+ ions in evidence, the propensity for folding must be related to the distribution of charges within the (M + nH)n+ ions.
Finally, we want to address the issue of disrupting and forming salt bridges in gaseous protein ions. Assuming, for example, that the native salt bridge between R19 and E55 is preserved in the (M + 7H)7+ ions of KIX, and broken by collisional activation, as indicated by the data in Figure 1c, does the latter involve charge separation? In other words, does breaking of the R19/E55 salt bridge produce protonated R19 and deprotonated E55 sidechains, or uncharged R19 and E55 sidechains? The proton affinity (PA) of guanidine as a sidechain model for arginine is 986 kJ/mol, and that of arginine is 1051 kJ/mol , with the 65 kJ/mol difference resulting from stabilization by intramolecular ionic hydrogen bonding in protonated arginine; the PA of propionate as a sidechain model for glutamate is 1454 kJ/mol . PA values of pentane-1-amine and 4-methyl-1H-imidazole as sidechain models for lysine and histidine are somewhat lower than that of guanidine, 953 and 924 kJ/mol, respectively, and that of acetate as a sidechain model for aspartate is 1453 kJ/mol . According to these PA values, proton transfer from a protonated basic sidechain to a deprotonated acidic sidechain is exothermic by 467–530 kJ/mol, which suggests that separation of residues in a salt bridge structure produces neutral sidechains unless the barrier between ionic (protonated basic sidechain and deprotonated acidic sidechain) and neutral forms (both sidechains uncharged) is sufficiently high to prevent proton transfer. Strittmatter and Williams have studied the energetics of heterodimers, AHB, consisting of trifluoroacetic acid, AH, and strong organic bases, B, and found by calculation that while the stability of the neutral (AH⋅B) and ion (A−⋅BH+) forms depends on the proton affinity of the base, barriers between the two forms were generally small . Moreover, separation of the neutral pair AH⋅B into AH and B required only 82–99 kJ/mol, whereas separation of the ion pair A−⋅BH+ into A− and BH+ required substantially more energy, 354–404 kJ/mol . Activation barriers for interconversion of zwitterionic and non-zwitterionic structures of sodiated octaglycine, (GGGGGGGG + Na)+, were also small, between –0.25 and 1.25 kJ/mol . It is thus reasonable to assume that by incrementally increasing an ion’s vibrational energy in low-energy collisional activation, a proton in a salt bridge structure will generally be transferred from the protonated basic sidechain to the deprotonated acidic sidechain before separation of the residues.
However, in compact protein ion structures like the KIX (M + nH)n+ ions investigated here, residues that form a salt bridge can at the same time be involved in additional electrostatic interactions with other residues that could substantially affect their proton affinity and the strength of a salt bridge. Previous studies have demonstrated the effect of inter- and intramolecular ionic and neutral hydrogen bonding on the proton affinity and related gas-phase basicity of neutral [74, 75, 76, 77] and deprotonated [78, 79, 80, 81] sites in amino acids and small peptides. In the native KIX structure, residue E55 not only forms a salt bridge with R19 but also an ionic hydrogen bond with Y68, and is in sufficiently close proximity to K52 to form yet another salt bridge (Figure 1b). If multiple electrostatic interactions can delocalize the negative charge of aspartate or glutamate residues, and thereby reduce their proton affinity to the extent that a positively charged, basic sidechain can be separated without causing proton transfer, remains an open question. Even so, the charge transition for the (M + 7H)7+ ions around site 16 (Figure 3a), near R19, is consistent with breaking of the R19/E55 salt bridge while retaining the positive and negative charge of R19 and E55, respectively; separation of protonated instead of uncharged R19 should also reduce the overall Coulombic repulsion in the partially unfolded ions. Regardless of whether or not proton transfer occurs upon breaking of salt bridges, the association of both neutral (e.g., R and E) and charged (e.g., protonated R and deprotonated E) pairs of basic and acidic sidechains can result in the formation of salt bridges as the barrier between zwitterionic (ion pair) and non-zwitterionic (neutral pair) structures is generally small [50, 73].
We have studied the unfolding and folding of KIX (M + nH)n+ ions from native ESI by electron capture dissociation. Vibrational ion activation at energies that were sufficiently high to cause loss of small molecules such as NH3 and H2O by breaking of covalent bonds in about 5% of the KIX ions with n = 7–9 was insufficient for full unfolding, but high enough to break the native R19/E55 salt bridge in the (M + 7H)7+ ions. Apparently, the strength of a salt bridge between a protonated basic and a deprotonated acidic sidechain in a gaseous protein ion is similar to the strength of a covalent bond. Specifically, the native R19/E55 salt bridge provides strong stabilization of KIX’s tertiary structure after transfer into the gas phase by conjoining helices α1 and α2. In all KIX ions studied here, helix stability against vibrational activation increased from α1 to α2 to α3, which is the same order of stability as that in solution  and that found for nonactivated (M + nH)n+ ions with n = 7–16 . This order of stability in the gas phase can be attributed to the number of stabilizing electrostatic interactions that increases from α1 to α2 to α3.
Folding of the KIX (M + nH)n+ ions on a 10 s timescale was observed only for n = 9, but not for n = 7 and n = 8, in contrast to what would be anticipated from Coulombic repulsion. Instead, the data for KIX and KIX(36–91) ions suggest that the propensity for protein folding in the gas phase is related to the distribution of charges within the (M + nH)n+ ions. Moreover, folding rates of different proteins showed a qualitative correlation with grand average of hydropathy (GRAVY) values, consistent with our previous hypothesis that the formation of electrostatic interactions, especially salt bridges, is a major driving force for protein folding in the gas phase . Our finding that the propensity for folding is determined by the intramolecular distribution of charges instead of ion net charge challenges the widespread assumption that protein unfolding in the gas phase is generally caused by Coulombic repulsion, in agreement with a previous study of multiply protonated polypropylenamine dendrimers .
Funding was provided by the Austrian Science Fund (FWF): Y372 and P27347 to K.B. The authors thank Heidelinde Glasner and Jovana Vusurovic for discussion, and Martin Tollinger and Sarina Grutsch for providing the KIX plasmid.
- 11.Heck, A.J.R., Jørgensen, T.J.D., O'Sullivan, M., Raumer, M.V., Derrick, P.J.: Gas-phase noncovalent interactions between vancomycin-group antibiotics and bacterial cell-wall precursor peptides probed by hydrogen/deuterium exchange. J. Am. Soc. Mass Spectrom. 9, 1255–1266 (1998)CrossRefGoogle Scholar
- 12.Williams, E.R., Schnier, P.D., Jockusch, R.: Blackbody infrared radiative dissociation of noncovalent biomolecule complexes. Abstr. Pap. Am. Chem. Soc. 215, U93 (1998)Google Scholar
- 28.Ma, X., Lai, L.B., Lai, S.M., Tanimoto, A., Foster, M.P., Wysocki, V.H., Gopalan, V.: Uncovering the stoichiometry of Pyrococcus furiosus RNAse P, a multi-subunit catalytic ribonucleoprotein complex, by surface-induced dissociation and ion mobility mass spectrometry. Angew. Chem. Int. Ed. 53, 11483–11487 (2014)CrossRefGoogle Scholar
- 45.Boyarkin, O.V., Kopysov, V.: Resonance energy transfer relates the gas-phase structure and pharmacological activity of opioid peptides. Angew. Chem. Int. Ed. 55, 689–692 (2016)Google Scholar
- 46.Kim, D., Pai, P.J., Creese, A.J., Jones, A.W., Russell, D.H., Cooper, H.J.: Probing the electron capture dissociation mass spectrometry of phosphopeptides with traveling wave ion mobility spectrometry and molecular dynamics simulations. J. Am. Soc. Mass Spectrom. 26, 1004–1013 (2015)CrossRefGoogle Scholar
- 47.Li, J.W., Taraszka, J.A., Counterman, A.E., Clemmer, D.E.: Influence of solvent composition and capillary temperature on the conformations of electrosprayed ions: unfolding of compact ubiquitin conformers from pseudonative and denatured solutions. Int. J. Mass Spectrom. 187, 37–47 (1999)CrossRefGoogle Scholar
- 48.Steinberg, M.Z., Elber, R., McLafferty, F.W., Gerber, R.B., Breuker, K.: Early structural evolution of native cytochrome c after solvent removal. Chem. BioChem. 9, 2417–2423 (2008)Google Scholar
- 66.Merenbloom, S.I., Flick, T.G., Daly, M.P., Williams, E.R.: Effects of select anions from the Hofmeister series on the gas-phase conformations of protein ions measured with traveling-wave ion mobility spectrometry/mass spectrometry. J. Am. Soc. Mass Spectrom. 22, 1978–1990 (2011)CrossRefGoogle Scholar
- 72.Available at: http://webbook.nist.gov. Accessed 21 Dec 2015
- 77.Meot-Ner, M.: The ionic hydrogen bond and ion solvation. 1. NH+⋅⋅⋅O, NH+⋅⋅⋅N, and OH+⋅⋅⋅O bonds. Correlations with proton affinity. Deviations due to structural effects. J. Am. Chem. Soc. 106, 1257–1264 (1984)Google Scholar
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.