PCCP View Article Online
Published on 12 December 2013. Downloaded by University of Western Ontario on 29/10/2014 04:33:36.
PAPER
Cite this: Phys. Chem. Chem. Phys., 2014, 16, 3007
View Journal | View Issue
Conformer-selective photoelectron spectroscopy of a-lactalbumin derived multianions in the gas phase† Matthias Vonderach,a Marc-Oliver Winghart,a Luke MacAleese,b Fabien Chirot,c Rodolphe Antoine,b Philippe Dugourd,b Patrick Weis,a Oliver Hampead and Manfred M. Kappes*ad We have recorded conformer-selective, gas-phase photoelectron spectra of a-lactalbumin derived multianions generated by electrospraying solutions of both the native protein and its denatured form (as prepared by breaking the sulfur–sulfur bonds by chemical reduction). Three different groups of gas-phase multianion conformers have been observed and characterized. Highly-folded and partiallyunfolded structures are obtained from solutions of the native protein. Only highly-elongated conformers are observed upon electrospraying the denatured protein. Adiabatic detachment energies were
Received 30th October 2013, Accepted 11th December 2013
determined at several negative charge states for each conformer group. In comparison to highly-
DOI: 10.1039/c3cp54596b
with increasing negative charge. By comparing experimental detachment energies for highly-elongated
www.rsc.org/pccp
structures with the predictions of a simple electrostatic model calculation, we have determined the effective dielectric shielding constant.
elongated conformations, highly-folded structures show a steeper decrease of electron binding energy
1. Introduction Electrospray ionization combined with ion mobility spectrometry provides a powerful way to probe the charge state dependence of biopolymer structures in the gas phase.1–13 Many studies on electrosprayed protein ions have shown that the underlying conformational distribution changes from ‘‘in vivo-like’’ to ‘‘more-extended’’ as the total charge increases (and repulsive Coulomb forces become stronger). An interesting variant of such studies is to compare the effect of increasing the charge in the case of two proteins differing only by their internal cohesion – as provided for by chemically cleaving cross-linking disulfide bonds prior to electrospraying. For example, Valentine et al. have demonstrated that breaking the disulfide bonds in the protein lysozyme leads to gas phase ion conformations differing significantly from those of the unmodified protein in the same a
Institute of Physical Chemistry, Karlsruhe Institute of Technology (KIT), 76128 Karlsruhe, Germany. E-mail: [email protected] b Institut Lumie`re Matie`re, UMR5306 Universite´ Lyon 1-CNRS, Universite´ de Lyon, 69622 Villeurbanne cedex, France c Institut des Sciences Analytiques, UMR5380 Universite´ Lyon 1-CNRS, Universite´ de Lyon, 69622 Villeurbanne cedex, France d Institute of Nanotechnology, Karlsruhe Institute of Technology (KIT), 76128 Karlsruhe, Germany † Electronic supplementary information (ESI) available. See DOI: 10.1039/ c3cp54596b
This journal is © the Owner Societies 2014
charge state.14 Furthermore, for comparable electrospray ionization conditions, significantly higher charge states of the reduced (i.e. cleaved) form could be observed. The issue of conformational stability versus charging-induced repulsive Coulomb forces in biopolymer ions is also closely connected to where the excess charges are localized and to how well they are dielectrically shielded from each other. A wellestablished method to estimate dielectric shielding constants in free protein ions is to compare experimentally determined basicities with those calculated for specific gas phase ion conformations – assuming that the gas phase basicity of a given group is reduced by its electrostatic interactions with neighbouring (partially shielded) charges. On the basis of such a model, Williams et al. have for example inferred dielectric constants of 2 for linear strands or 4 for the a-helix of cytochrome c cations.15–17 Photoelectron spectroscopy (PES) is a potentially useful method to probe corresponding repulsive electrostatic interactions between localized negative charges in multiply charged anions (MCAs). Specifically, PES can be used to determine electron binding energies (BE) and repulsive Coulomb barriers as influenced by these electrostatic interactions,18–25 thus allowing to evaluate the corresponding dielectric shielding constants independently. Recently, we have shown that it is possible to simultaneously acquire information about the electrostatic interactions and
Phys. Chem. Chem. Phys., 2014, 16, 3007--3013 | 3007
View Article Online
Published on 12 December 2013. Downloaded by University of Western Ontario on 29/10/2014 04:33:36.
Paper
the structural behavior of gas phase biopolymer MCAs by conformer resolved photoelectron spectroscopy – as a function of the excess charge state.26 An interesting question which can be addressed by such studies is how the electron binding energies and dielectric screening constants of ‘‘in vivo-like’’ protein conformations relate to those of the ‘‘more-extended’’ forms. In this paper we report conformer resolved photoelectron spectroscopy of bovine a-lactalbumin derived multianions over a wide range of excess negative charge states. The protein comprises a sequence of 123 amino acid residues, of which 20 acidic residues (aspartic and glutamic acids) appear to be the most likely sites for deprotonation towards forming multianions. Importantly, all 8 cysteines of the native protein are involved in cross-linking disulfide bonds. We present data for multianion conformers resulting from both the native protein and its disulfide-reduced solution form (having broken these sulfur– sulfur bonds) and use our measurements to assess dielectric screening by comparison to a simple structural model.
2. Experimental Experiments were performed using a home built instrument which couples electrospray ionization (ESI), ion mobility spectroscopy (IMS), mass spectrometry (MS) and photoelectron spectroscopy (PES). The setup has been described previously.26 Briefly, an electrospray ionization source is interfaced to a mobility drift cell by an hourglass ion funnel which is operated in pulsed mode to inject ion packets for fractionation according to their collision cross section (and charge) leading to a dispersion in drift time. To induce collision induced isomerization, the injection energy, corresponding to the difference between the last electrode of the ion funnel and the first electrode of the drift tube can be varied over a range of 10–150 V. The drift tube is usually operated at a dinitrogen pressure of 1.5 mbar and a voltage of 2 kV. After being separated according to their collision cross sections, ions are further discriminated by their mass-tocharge ratios in a quadrupole mass filter. From the acquired arrival time distribution (ATD), we can directly obtain the corresponding reduced ion mobility K0 (normalized to pressure and temperature). These mobilities can be converted into collision cross sections in dinitrogen, O, using the following equation. sffiffiffiffiffiffiffiffiffiffiffiffi 3q 2p 1 K0 ¼ (1) 16N0 mkB T O N0 is the number density, q is the charge of the ion, T is the temperature of the collision gas and m is the reduced mass. After passing the quadrupole mass filter, mass (and mobility) selected ions are electrostatically focussed into the detachment region of a photoelectron spectrometer. Here they are irradiated by a pulsed detachment laser (Nd:YAG) usually operated at 266 nm (4.66 eV). Some experiments were also performed at a detachment laser wavelength of 213 nm (fifth harmonic). The photoelectron spectrometer is of the magnetic bottle type which is based on the design of Kruit and Read.27,28 The kinetic energy
3008 | Phys. Chem. Chem. Phys., 2014, 16, 3007--3013
PCCP
KE of the photoelectrons is determined from their arrival time. KE is converted into electron binding energy, EBE, according to the relation (hn = EBE + KE). Photoelectron spectral intensities are plotted versus EBE also taking into account the appropriate Jacobi transformation (dt - dE). Background photoemission from metal surfaces leads to PE spectral noise problems above EBE = 3 eV (where direct detachment yields are low due to the repulsive Coulomb barriers pertaining). We therefore typically show PE spectra only up to this electron binding energy. Bovine a-lactalbumin (L6010) was purchased from Sigma Aldrich. The native form was sprayed from a 0.1 mM solution in a mixture of water–methanol (1 : 4/v:v) with sufficient ammonium hydroxide to reach a pH value of around 10. To break the sulfur– sulfur bonds of the protein by chemical reduction, DTT in excess (dithiothreitol, Sigma Aldrich) was added to an aqueous solution of the native protein (1 mM) and NaOH (100 mM). After reduction, the solution was diluted to a protein concentration of 0.1 mM in a mixture of water–methanol (1 : 4/v:v) before electrospraying. High resolution negative-ion mass spectra (see ESI,† Fig. S1 and S2) are fully consistent with the molecular weight of the primary sequence of a-lactalbumin (see e.g. protein identifier P00711 from the pdb database) subject to multiple deprotonations. Furthermore, a comparison of the isotopic distributions observed in high resolution scans of the peak 7 for native and reduced forms shows that 8 additional H atoms (due to 8 –SH groups formed upon cleavage of 4 –S–S– bonds) are present in the reduced form (Fig. S2, ESI†). Note that for electrospraying, the solution pH was raised so as to maximize the intensity of high negative charge states (in order to facilitate conformer-resolved PE spectroscopy). We did not systematically study the dependence of conformer resolved PE spectral intensities on solution pH and therefore make no claims concerning the conformer distributions present in solution.
3. Results and discussion Fig. 1 displays normalized collision cross sections of multianions resulting from solutions of both the native protein and its denatured (reduced) form as a function of the excess negative charge state. These were determined from the arrival time distributions shown in Fig. 2. For the native protein solutions, we observe up to two temporally resolved ion signals for each of the charge states from 5 to 9, indicating that there are up to two different types of conformers/components present. High resolution mass scans show the isotopic distribution of nominally 5 to be a superposition of roughly equal amounts of monomer 5 and dimer 10. We therefore assign the faster arrival time signal to the dimer 10. A similar situation may also pertain for ‘‘6’’: the faster arrival time feature has a cross section consistent with either a monomer or a dimer. Unfortunately, the mass resolution available was insufficient to decide this. All other ion mobility resolved signals observed for the native protein (in higher charge states) can be unequivocally assigned to monomers and can be classed into one of two different conformational groups
This journal is © the Owner Societies 2014
View Article Online
PCCP
Paper
repulsion between the extra charges. Then, the vertical detachment energy VDEn of an electron in a multiply charged anion can be described by: VDEn ¼ VDE
n1 X
Published on 12 December 2013. Downloaded by University of Western Ontario on 29/10/2014 04:33:36.
i¼1
Fig. 1 Normalized collision cross sections versus charge states for all monomeric species observed. Black symbols: multianion conformers resulting from native protein solutions. Green symbols: conformers electrosprayed from denatured (reduced) protein solutions.
according to their experimental collision cross sections (and charge state dependence thereof). In analogy to ref. 14 we call the group having small cross sections ‘‘highly-folded’’, whereas the more elongated structures are categorized as ‘‘partially-unfolded’’. Much higher average charge states are observed for the reduced protein under analogous electrospray conditions. Here, just one dominant conformation is resolved for each charge state. Where comparison at the same charge state is possible, the cross sections found for the reduced species are always significantly larger than those seen for ‘‘partially-unfolded’’ conformers resulting from native proteins. Consequently, we categorize the multianions generated by electrospraying reduced/denatured proteins as ‘‘highly-elongated’’. In addition to the arrival time distributions, Fig. 2 also shows the photoelectron spectra (266 nm detachment wavelength) corresponding to each resolved conformer. So as to allow for an easier overview, the figure has been plotted such that each column shows the PE spectra of a common conformational family. Generally, for a given conformer only one broad photoelectron spectral band is observed. Its position and shape depends on the charge state and the specific conformation. The highly-folded conformer group shows a strong spectral shift to lower binding energy as the charge is increased. For the highly-elongated group (derived from the reduced protein), the corresponding decrease in electron binding energies upon raising the charge is noticeably less. Interestingly, for the highest negative charge states of the highly-elongated conformers, the photoelectron bands become asymmetric and manifest a pronounced tail at their low energy side. The decrease in electron binding energy observed upon increasing the charge for the partially-unfolded group shows a behavior intermediate between these two cases. To first order the PE spectra of MCAs having localized excess charges can be interpreted based on the assumption that the electron binding energy is lowered by (additional) Coulomb
This journal is © the Owner Societies 2014
e2 4pe0 srij
(2)
Here, VDE is the detachment energy of a charged functional group (e.g. a carboxylate), s is a characteristic dielectric shielding constant (normally assumed to be around 2 to 5 (ref. 29–33)) and rij is the distance between the functional group and another (excess) charge residing on the molecule. Typically, the excess negative charges on multiply charged protein anions are likely to reside on deprotonated carboxylate groups, corresponding to a VDE of characteristically 3.5 to 4.5 eV, depending on the –COO local chemical environment.34,35 We assume that the dominant photodetachment processes occurring in our experiment correspond to detaching electrons directly from these charged carboxylates. Where are these deprotonated carboxylate groups actually located and what are the final states after photodetachment? Given a total number of m carboxylic acid groups in the protein, and a much smaller number k of deprotonated carboxylates (corresponding to the number of excess charges), the number of possible charge isomers initially present can be calculated as the permutation m!/[k!(m k)!]. Taking into account that each of the k electrons can be removed (in principle), the number of possible final ionized states is km!/[k!(m k)!]. a-Lactalbumin comprises 21 carboxylic acid groups. Therefore, as an example, there are 101 745 final state charge isomers possible after removing an electron from charge state 5. This ‘‘charge isomer’’ multiplicity is probably the main reason why our photoelectron spectra consist of rather broad bands (large initial state vibrational excitation levels likely also contribute). For a more quantitative analysis, the adiabatic electron affinities were extracted from the low energy flanks of the corresponding PE spectra. Most spectra show a more or less pronounced low energy tail so that a simple linearization is quite subjective. Instead, we determine upper and lower limiting values of the ADE by linearly extrapolating both the tail (lower limit) and the first linearly increasing part of the PE spectrum above this tail (upper limit). The results are displayed as a function of the normalized cross sections in Fig. 3. To exclude multiphoton detachment or delayed relaxation of electronically excited states as significantly contributing to the PE spectra, a number of measurements were also carried out at a detachment wavelength of 213 nm (ESI,† Fig. S3). ADE trends are consistent with the 266 nm measurements. As can already be inferred from the conformer resolved PE spectra (Fig. 2), the ‘‘highly-folded’’ conformation group shows a dramatic decrease in ADE upon deprotonation of additional carboxylic acid groups, whereas the corresponding collision cross sections remain approximately constant. On the other hand, the ‘‘highly-elongated’’ form shows a comparatively small decrease in ADE while the cross sections are seen to increase much more significantly upon raising the charge. For further analysis, repulsive Coulomb barriers have also been estimated.
Phys. Chem. Chem. Phys., 2014, 16, 3007--3013 | 3009
View Article Online
PCCP
Published on 12 December 2013. Downloaded by University of Western Ontario on 29/10/2014 04:33:36.
Paper
Fig. 2 Conformer resolved photoelectron spectra (266 nm) of multianions in various charge states obtained upon electrospraying native (left) and denatured a-lactalbumin (right) – in order of their arrival time. Also shown are the corresponding arrival time distributions (ATD) with resolved conformers indicated ((1)/(2)). Dotted curves (red) show arrival time distributions acquired under significantly higher injection energy conditions associated with isomerization.
We use the turning point of the high EBE tail in the PE spectra as a criterion for determining the RCB (as the difference between the photon energy and this energy). The resulting RCB estimates are shown in Fig. S4 (ESI†) as a function of charge state. ADE and RCB values show similar trends. How can these trends be rationalized? Typically, the collision cross section of an MCA having localized charges is directly correlated with the average distances between the charged functional groups. Then according to eqn (2), the detachment energy must be larger for elongated structures in comparison to
3010 | Phys. Chem. Chem. Phys., 2014, 16, 3007--3013
highly-folded structures of the same charge state because of the larger distances rij between the charged carboxylate groups. This is to first order the reason, why the experimentally determined ADE values of the octa- and nonaanions are larger for the highly-elongated conformation (denatured protein), smaller for the partially-unfolded group and even smaller in the case of the highly-folded species. Furthermore, the multianions derived from the reduced protein are apparently much better able to ‘‘compensate’’ additional negative charges than those resulting from the native protein,
This journal is © the Owner Societies 2014
View Article Online
PCCP
Paper
Published on 12 December 2013. Downloaded by University of Western Ontario on 29/10/2014 04:33:36.
We next introduce L, the length of one full helical turn. It is given by: pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi L ¼ h2 þ p2 d 2 (4) NL then defines the overall length of the coiled biopolymer chain which remains constant during the unfolding process. Then the charge-dependent diameter of the helix can be written as: v0 ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1ffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u 2ffi u 1u 1 1 4 Op A (5) d ¼ t@ L2 L p 2 4 Nc
Fig. 3 ADEs of each ion mobility resolved conformer group as a function of their normalized cross sections (and total negative charges, see also Fig. 1). Error bars correspond to upper and lower bounds resulting from the procedure used to assign ADE values to the spectra shown in Fig. 2 (see text).
as indicated by the slope of the curve in Fig. 3. This is much smaller for the highly-elongated form (implying that the additional repulsive Coulomb potential energy can be more easily dissipated by distortion). As mentioned before, the PE spectra of the highest charge states of the highly-elongated group show a low energy tail which becomes dominant upon increasing the charge. Correspondingly, the uncertainty range of the specified ADE value increases. As the unfolding process progresses, more and more hydrogen bonds will break. Generally, the VDE of a deprotonated carboxylic acid group is likely to be somewhat larger (around 4.5 eV or even higher) if this group is surrounded by positively polarized hydrogen atoms. On the other hand, the VDE of an electrostatically well isolated carboxylate group lies around 3.5 eV. We propose that the unfolding induced conversion of ‘‘coordinated carboxylates’’ to increasingly isolated carboxylates is the reason why the low energy flanks of the corresponding PE spectral bands appear more and more asymmetric at the highest charge states accessed here. In a simple molecular picture, charging induced distortion of the reduced protein can be thought of as corresponding to the stretching-transformation of a helix towards a linear strand.36 Raising the charge leads to an increase in the screw pitch while the inner diameter of the helix decreases correspondingly. We have developed a simple model to estimate how this should affect the ADE. For this, we first map the helix structure onto a cylinder having a diameter d. Assuming that d is much smaller than the length of the corresponding cylinder its collision cross section can be calculated according to: O = Nhdc
(3)
N is the number of helical coils, h is the pitch (i.e. the width of one complete helical turn as measured along the cylinder axis) and c is a constant to convert the longitudinal cross section of the cylinder into its absolute collision cross section (averaged over many projections).
This journal is © the Owner Societies 2014
Using this equation, it is possible to calculate the helix diameter based on the experimental cross section, if the pitch and the constant c are known. Typically, one coil of a perfect a-helix consists of 3.6 amino acids (corresponding to a starting pitch of 5.4 Å). The reduced a-lactalbumin then contains 34.17 such coils if we assume that it has a corresponding helical structure. Next we calibrate the parameter c (see also the ESI†). For this we use a known protein structure comprising many different helices each having a characteristic number of coils. Specifically, the pdb structure file of horse heart myoglobin,37 a protein comprising 75% a-helices, was first numerically divided into its individual helices. We then calculated collision cross sections for each of these component helices using the projection approximation (via the program sigma38,39). The resulting collision cross sections show a linear dependence on the number of coils with a slope of c = 2.76. Using the collision cross sections experimentally determined for the multianions resulting from reduced a-lactalbumin, we can now determine cylinder diameters d and total heights (Nh) for every charge state – as shown in the ESI.† This in turn allows us to model the ADEs based on the further assumption of a specific distribution of excess charges on the corresponding cylinder surfaces. For simplicity, we assume an alternating distribution of excess charges symmetrically localized on opposite sides of the cylinder in order to maximize charge separations. ADEs can then be estimated according to eqn (2) by inserting appropriate values for s. For the ADE of a deprotonated carboxylate we assume lower and upper bounds of 3.25 and 4.25 eV, respectively (to take into account different amounts of ‘‘coordination’’ by hydrogen bonding).‡ When assuming s = 1, i.e. no dielectric shielding, the slope of the theoretical line is noticeably steeper than experiment. Good agreement is obtained for s = 1.5 implying that even for the highly distorted structures studied here, there is still significant charge shielding by reason of polarization. If we further take into consideration that in reality carboxylates are not uniformly distributed but that they are instead localized at specific positions defined by the protein sequence, the correspondingly lowered distances between excess charges in specific regions would lead to an even larger effective shielding constant – consistent with literature assessments on the basis of other types of measurements. Note that an alternate way to estimate s – not ‡ To first order eqn (2) can also be analogously formulated in terms of ADEs.
Phys. Chem. Chem. Phys., 2014, 16, 3007--3013 | 3011
View Article Online
Published on 12 December 2013. Downloaded by University of Western Ontario on 29/10/2014 04:33:36.
Paper
PCCP
Fig. 4 Experimental ADEs of highly-elongated multianions (from the reduced protein) in comparison to a distorted helical model calculation plotting the average between upper/lower ADE bounds (indicated by shading – see text for details). Predictions are shown both for s = 1 and 1.5. The latter dielectric shielding constant provides the best agreement between the model and experiment.
pursued here – would be to calculate the repulsive Coulomb barrier for a specific value of s and to compare this to experimental RCB values. Finally, it is instructive to estimate the electrostatic forces acting on such a (helical) biopolymer and to compare these results to force-extension measurements on individual protein molecules as performed using atomic force microscopy or optical tweezer setups.40 In such experiments molecules are uniaxially stretched until at a certain force they start to unfold and/or bonds begin to rupture. For the comparison, we only need to consider axial tension. Therefore an electrostatic model even simpler than that of Fig. 4 suffices. We now assume that the negative charges (ze) are uniformly distributed over a chain of length l0 = Nh (i.e. number of coils times the pitch, see above). Neglecting shielding, the terminal charges experience an effective repulsive force comprising contributions from all remaining (z 1) charges: FðzÞ ¼
z1 e2 l0 2 X 1 2 4pe0 z 1 i i¼1
(6)
Fig. 5 displays the corresponding forces for a rigid chain of length l0 = 18.5 nm – as a function of the total number of localized excess charges z. Between z = 2 and 20, the electrostatic force increases by about three orders of magnitude (ca. 0.5–500 pN). By comparison, the nanomechanical measurements performed on various a-helices at room temperature show that the molecule begins to unwind its tertiary structure upon applying a stretching force in the range of o10–20 pN (typically described as the entropic regime). According to molecular dynamics simulations, higher forces of up to several nN then give rise to further transitions from a-helix to beta sheet secondary structures.41 Finally, forces significantly in excess of 1 nN are required in order to break a covalent bond.42
3012 | Phys. Chem. Chem. Phys., 2014, 16, 3007--3013
Fig. 5 Modelled electrostatic axial force of a uniformly charged chain (as described by eqn (6)) for a total chain length of l0 = 18.5 nm for different charge states.
Despite the obvious shortcomings of our crude electrostatic model, the order of magnitude agreement between the force required to transform a peptide chain from a-helix to beta sheet and that experienced by a terminal charge in a rigid chain of comparable z and l0 is intriguing. Conceivably, quantifying the structural transitions associated with electrostatically charging proteins in the gas-phase may offer a complementary approach to single biomolecule nanomechanics.
4. Conclusion In conclusion, we have recorded conformer resolved photoelectron spectra for all negative charge states observed upon electrospraying the native and the chemically reduced protein a-lactalbumin. By determining the collision cross sections of every resolvable conformer, we were able to divide the resulting gas phase multianion structures into three conformational groups: highlyfolded, partially-unfolded and highly-elongated structures. These conformational groups differ strongly in terms of the dependence of their experimental adiabatic detachment energies on the negative charge state. In particular, if the charge of the highlyfolded conformer group is increased (by further deprotonation of carboxylic acids functionalities) a large decrease in electron binding energy is observed. In contrast, the highly-elongated form shows a much smaller reduction in ADE upon further charging. This is due to much larger distances between excess charges in the reduced proteins. To qualitatively describe these observations, we have formulated a simple electrostatic model in which the reduced form is described as a helix which extends upon increasing charge. Within this model it was necessary to use an effective dielectric shielding constant of 1.5 in order to obtain agreement with experimental values, implying the presence of significant charge shielding due to polarization or dipole fluctuations. Even coarser estimations based on point-charge modelling indicate that the repulsive Coulomb forces prevailing for the higher multianionic charge states are
This journal is © the Owner Societies 2014
View Article Online
PCCP
Published on 12 December 2013. Downloaded by University of Western Ontario on 29/10/2014 04:33:36.
of the same order as the nanomechanical forces necessary to unfold individual protein and peptide molecules by uniaxial stretching. In future work, it will prove interesting to combine ion-mobility PES measurements with elasticity models in order to better understand the correlation between charge and structure of biopolymers.
Acknowledgements MK and PW acknowledge support by the Deutsche Forschungsgemeinschaft (DFG) towards developing the IMS-PES apparatus used in this study. MOW and MK also acknowledge support by the Fonds der Chemischen Industrie (FCI). RA and PD thank Marion Girod for recording some of the high resolution mass spectra.
References 1 D. E. Clemmer, R. R. Hudgins and M. F. Jarrold, J. Am. Chem. Soc., 1995, 117, 10141. 2 C. S. Hoaglund, S. J. Valentine, C. R. Sporleder, J. P. Reilly and D. E. Clemmer, Anal. Chem., 1998, 70, 2236. 3 D. Balbeur, J. Widart, B. Leyh, L. Cravello and E. De Pauw, J. Am. Soc. Mass Spectrom., 2008, 19, 938. 4 D. Clemmer and M. F. Jarrold, J. Mass Spectrom., 1997, 32, 577. 5 S. L. Bernstein, N. F. Dupuis, N. D. Lazo, T. Wyttenbach, M. M. Condron, G. Bitan, D. B. Teplow, J. E. Shea, B. T. Ruotolo, C. V. Robinson and M. T. Bowers, Nat. Chem., 2009, 1, 326. 6 C. S. Hoaglund-Hyzer, A. E. Counterman and D. E. Clemmer, Chem. Rev., 1999, 99, 3037. 7 S. I. Merenbloom, T. G. Flick, M. P. Daly and E. R. Williams, J. Am. Soc. Mass Spectrom., 2011, 22, 1978. 8 Y. Kang and D. J. Douglas, J. Am. Soc. Mass Spectrom., 2011, 22, 1187. 9 K. Thalassinos, S. E. Slade, K. R. Jennings, J. H. Scrivens, K. Giles, J. Wildgoose, J. Hoyes, R. H. Bateman and M. T. Bowers, Int. J. Mass Spectrom., 2004, 263, 55. 10 C. Bleiholder, N. F. Dupuis, T. Wyttenbach and M. T. Bowers, Nat. Chem., 2011, 3, 172. 11 D. E. Clemmer and S. J. Valentine, Nat. Chem., 2009, 1, 257. 12 H. Shi, N. A. Pierson, S. J. Valentine and D. E. Clemmer, J. Phys. Chem. B, 2012, 116, 3344. 13 T. Wyttenbach and M. T. Bowers, J. Phys. Chem. B, 2011, 115, 12266. 14 S. J. Valentine, J. G. Anderson, A. D. Ellington and D. E. Clemmer, J. Phys. Chem. B, 1997, 101, 3891. 15 D. S. Gross, P. D. Schnier, S. E. R. Cruz, C. K. Fagerquist and E. R. Williams, Proc. Natl. Acad. Sci. U. S. A., 1996, 93, 3143. 16 D. S. Gross and E. R. Williams, J. Am. Chem. Soc., 1995, 117, 883. 17 P. D. Schnier, D. S. Gross and E. R. Williams, J. Am. Chem. Soc., 1995, 117, 6747.
This journal is © the Owner Societies 2014
Paper
¨ffler, J. Friedrich, 18 J. M. Weber, I. N. Ioffe, K. M. Berndt, D. Lo O. T. Ehrler, A. S. Danell, J. H. Parks and M. M. Kappes, J. Am. Chem. Soc., 2004, 126, 8585. 19 L. S. Wang and X. B. Wang, J. Phys. Chem. A, 2000, 104, 1978. 20 L. S. Wang, C. F. Ding, X. B. Wang and S. E. Barlow, Rev. Sci. Instrum., 1999, 70, 1957. 21 K. Matheis, L. Joly, R. Antoine, F. Lepine, C. Bordas, O. T. Ehrler, A. R. Allouche, M. M. Kappes and P. Dugourd, J. Am. Chem. Soc., 2008, 130, 15903. 22 L. S. Wang, C. F. Ding, X. B. Wang, J. B. Nicholas and B. Nicholas, Phys. Rev. Lett., 1998, 81, 2667. 23 X. B. Wang and L. S. Wang, Nature, 1999, 400, 245. 24 M. Vonderach, O. T. Ehrler, K. Matheis, T. Karpuschkin, E. Papalazarou, C. Brunet, R. Antoine, P. Weis, O. Hampe, M. M. Kappes and P. Dugourd, Phys. Chem. Chem. Phys., 2011, 13, 15554. 25 M. Vonderach, O. T. Ehrler, K. Matheis, P. Weis and M. M. Kappes, J. Am. Chem. Soc., 2012, 134, 7830. 26 M. Vonderach, O. T. Ehrler, P. Weis and M. M. Kappes, Anal. Chem., 2011, 83, 1108. 27 P. Kruit and F. H. Read, J. Phys. E: Sci. Instrum., 1983, 16, 313. 28 O. Cheshnovsky, S. H. Yang, C. L. Pettiette, M. J. Craycraft and R. E. Smalley, Rev. Sci. Instrum., 1987, 56, 2131. 29 X. Song, J. Chem. Phys., 2002, 116, 9358. 30 G. N. Patargias, S. A. Harris and J. H. Harding, J. Chem. Phys., 2010, 132, 235103. 31 T. Simonson and C. L. Brooks, J. Am. Chem. Soc., 1996, 118, 8452. 32 W. C. Guest, N. R. Cashman and S. S. Plotkin, Phys. Chem. Chem. Phys., 2011, 13, 6286. 33 J. W. Pitera, M. Falta and W. F. van Gunsteren, Biophys. J., 2001, 80, 2546. 34 X. B. Wang, C. F. Ding and L. S. Wang, Phys. Rev. Lett., 1998, 81, 3351. 35 H. K. Woo, X. B. Wang, K. C. Lau and L. S. Wang, J. Phys. Chem. A, 2006, 110, 7801. 36 Note that denatured lactalbumin actually has low helix content in solution (R. Montserret and R. Antoine, private communication) such that our model has to be regarded as highly simplified. 37 D. M. Copeland, A. S. Soares, A. H. West and G. B. RichterAddo, J. Inorg. Biochem., 2006, 100, 1413. 38 T. Wyttenbach, G. von Helden, J. J. Batka, D. Carlat and M. T. Bowers, J. Am. Soc. Mass Spectrom., 1997, 8, 275. 39 T. Wyttenbach, M. Witt and M. T. Bowers, J. Am. Chem. Soc., 2000, 122, 3458. 40 M. Sotomayor and K. Schulten, Science, 2007, 316, 1144, and references therein. 41 Z. Qin and M. J. Buehler, Phys. Rev. Lett., 2010, 104, 198304. 42 M. Grandbois, M. Beyer, M. Rief, H. Clausen-Schaumann and H. E. Gaub, Science, 1999, 283, 1727.
Phys. Chem. Chem. Phys., 2014, 16, 3007--3013 | 3013
Published on 12 December 2013. Downloaded by University of Western Ontario on 29/10/2014 04:33:36.
PAPER
Cite this: Phys. Chem. Chem. Phys., 2014, 16, 3007
View Journal | View Issue
Conformer-selective photoelectron spectroscopy of a-lactalbumin derived multianions in the gas phase† Matthias Vonderach,a Marc-Oliver Winghart,a Luke MacAleese,b Fabien Chirot,c Rodolphe Antoine,b Philippe Dugourd,b Patrick Weis,a Oliver Hampead and Manfred M. Kappes*ad We have recorded conformer-selective, gas-phase photoelectron spectra of a-lactalbumin derived multianions generated by electrospraying solutions of both the native protein and its denatured form (as prepared by breaking the sulfur–sulfur bonds by chemical reduction). Three different groups of gas-phase multianion conformers have been observed and characterized. Highly-folded and partiallyunfolded structures are obtained from solutions of the native protein. Only highly-elongated conformers are observed upon electrospraying the denatured protein. Adiabatic detachment energies were
Received 30th October 2013, Accepted 11th December 2013
determined at several negative charge states for each conformer group. In comparison to highly-
DOI: 10.1039/c3cp54596b
with increasing negative charge. By comparing experimental detachment energies for highly-elongated
www.rsc.org/pccp
structures with the predictions of a simple electrostatic model calculation, we have determined the effective dielectric shielding constant.
elongated conformations, highly-folded structures show a steeper decrease of electron binding energy
1. Introduction Electrospray ionization combined with ion mobility spectrometry provides a powerful way to probe the charge state dependence of biopolymer structures in the gas phase.1–13 Many studies on electrosprayed protein ions have shown that the underlying conformational distribution changes from ‘‘in vivo-like’’ to ‘‘more-extended’’ as the total charge increases (and repulsive Coulomb forces become stronger). An interesting variant of such studies is to compare the effect of increasing the charge in the case of two proteins differing only by their internal cohesion – as provided for by chemically cleaving cross-linking disulfide bonds prior to electrospraying. For example, Valentine et al. have demonstrated that breaking the disulfide bonds in the protein lysozyme leads to gas phase ion conformations differing significantly from those of the unmodified protein in the same a
Institute of Physical Chemistry, Karlsruhe Institute of Technology (KIT), 76128 Karlsruhe, Germany. E-mail: [email protected] b Institut Lumie`re Matie`re, UMR5306 Universite´ Lyon 1-CNRS, Universite´ de Lyon, 69622 Villeurbanne cedex, France c Institut des Sciences Analytiques, UMR5380 Universite´ Lyon 1-CNRS, Universite´ de Lyon, 69622 Villeurbanne cedex, France d Institute of Nanotechnology, Karlsruhe Institute of Technology (KIT), 76128 Karlsruhe, Germany † Electronic supplementary information (ESI) available. See DOI: 10.1039/ c3cp54596b
This journal is © the Owner Societies 2014
charge state.14 Furthermore, for comparable electrospray ionization conditions, significantly higher charge states of the reduced (i.e. cleaved) form could be observed. The issue of conformational stability versus charging-induced repulsive Coulomb forces in biopolymer ions is also closely connected to where the excess charges are localized and to how well they are dielectrically shielded from each other. A wellestablished method to estimate dielectric shielding constants in free protein ions is to compare experimentally determined basicities with those calculated for specific gas phase ion conformations – assuming that the gas phase basicity of a given group is reduced by its electrostatic interactions with neighbouring (partially shielded) charges. On the basis of such a model, Williams et al. have for example inferred dielectric constants of 2 for linear strands or 4 for the a-helix of cytochrome c cations.15–17 Photoelectron spectroscopy (PES) is a potentially useful method to probe corresponding repulsive electrostatic interactions between localized negative charges in multiply charged anions (MCAs). Specifically, PES can be used to determine electron binding energies (BE) and repulsive Coulomb barriers as influenced by these electrostatic interactions,18–25 thus allowing to evaluate the corresponding dielectric shielding constants independently. Recently, we have shown that it is possible to simultaneously acquire information about the electrostatic interactions and
Phys. Chem. Chem. Phys., 2014, 16, 3007--3013 | 3007
View Article Online
Published on 12 December 2013. Downloaded by University of Western Ontario on 29/10/2014 04:33:36.
Paper
the structural behavior of gas phase biopolymer MCAs by conformer resolved photoelectron spectroscopy – as a function of the excess charge state.26 An interesting question which can be addressed by such studies is how the electron binding energies and dielectric screening constants of ‘‘in vivo-like’’ protein conformations relate to those of the ‘‘more-extended’’ forms. In this paper we report conformer resolved photoelectron spectroscopy of bovine a-lactalbumin derived multianions over a wide range of excess negative charge states. The protein comprises a sequence of 123 amino acid residues, of which 20 acidic residues (aspartic and glutamic acids) appear to be the most likely sites for deprotonation towards forming multianions. Importantly, all 8 cysteines of the native protein are involved in cross-linking disulfide bonds. We present data for multianion conformers resulting from both the native protein and its disulfide-reduced solution form (having broken these sulfur– sulfur bonds) and use our measurements to assess dielectric screening by comparison to a simple structural model.
2. Experimental Experiments were performed using a home built instrument which couples electrospray ionization (ESI), ion mobility spectroscopy (IMS), mass spectrometry (MS) and photoelectron spectroscopy (PES). The setup has been described previously.26 Briefly, an electrospray ionization source is interfaced to a mobility drift cell by an hourglass ion funnel which is operated in pulsed mode to inject ion packets for fractionation according to their collision cross section (and charge) leading to a dispersion in drift time. To induce collision induced isomerization, the injection energy, corresponding to the difference between the last electrode of the ion funnel and the first electrode of the drift tube can be varied over a range of 10–150 V. The drift tube is usually operated at a dinitrogen pressure of 1.5 mbar and a voltage of 2 kV. After being separated according to their collision cross sections, ions are further discriminated by their mass-tocharge ratios in a quadrupole mass filter. From the acquired arrival time distribution (ATD), we can directly obtain the corresponding reduced ion mobility K0 (normalized to pressure and temperature). These mobilities can be converted into collision cross sections in dinitrogen, O, using the following equation. sffiffiffiffiffiffiffiffiffiffiffiffi 3q 2p 1 K0 ¼ (1) 16N0 mkB T O N0 is the number density, q is the charge of the ion, T is the temperature of the collision gas and m is the reduced mass. After passing the quadrupole mass filter, mass (and mobility) selected ions are electrostatically focussed into the detachment region of a photoelectron spectrometer. Here they are irradiated by a pulsed detachment laser (Nd:YAG) usually operated at 266 nm (4.66 eV). Some experiments were also performed at a detachment laser wavelength of 213 nm (fifth harmonic). The photoelectron spectrometer is of the magnetic bottle type which is based on the design of Kruit and Read.27,28 The kinetic energy
3008 | Phys. Chem. Chem. Phys., 2014, 16, 3007--3013
PCCP
KE of the photoelectrons is determined from their arrival time. KE is converted into electron binding energy, EBE, according to the relation (hn = EBE + KE). Photoelectron spectral intensities are plotted versus EBE also taking into account the appropriate Jacobi transformation (dt - dE). Background photoemission from metal surfaces leads to PE spectral noise problems above EBE = 3 eV (where direct detachment yields are low due to the repulsive Coulomb barriers pertaining). We therefore typically show PE spectra only up to this electron binding energy. Bovine a-lactalbumin (L6010) was purchased from Sigma Aldrich. The native form was sprayed from a 0.1 mM solution in a mixture of water–methanol (1 : 4/v:v) with sufficient ammonium hydroxide to reach a pH value of around 10. To break the sulfur– sulfur bonds of the protein by chemical reduction, DTT in excess (dithiothreitol, Sigma Aldrich) was added to an aqueous solution of the native protein (1 mM) and NaOH (100 mM). After reduction, the solution was diluted to a protein concentration of 0.1 mM in a mixture of water–methanol (1 : 4/v:v) before electrospraying. High resolution negative-ion mass spectra (see ESI,† Fig. S1 and S2) are fully consistent with the molecular weight of the primary sequence of a-lactalbumin (see e.g. protein identifier P00711 from the pdb database) subject to multiple deprotonations. Furthermore, a comparison of the isotopic distributions observed in high resolution scans of the peak 7 for native and reduced forms shows that 8 additional H atoms (due to 8 –SH groups formed upon cleavage of 4 –S–S– bonds) are present in the reduced form (Fig. S2, ESI†). Note that for electrospraying, the solution pH was raised so as to maximize the intensity of high negative charge states (in order to facilitate conformer-resolved PE spectroscopy). We did not systematically study the dependence of conformer resolved PE spectral intensities on solution pH and therefore make no claims concerning the conformer distributions present in solution.
3. Results and discussion Fig. 1 displays normalized collision cross sections of multianions resulting from solutions of both the native protein and its denatured (reduced) form as a function of the excess negative charge state. These were determined from the arrival time distributions shown in Fig. 2. For the native protein solutions, we observe up to two temporally resolved ion signals for each of the charge states from 5 to 9, indicating that there are up to two different types of conformers/components present. High resolution mass scans show the isotopic distribution of nominally 5 to be a superposition of roughly equal amounts of monomer 5 and dimer 10. We therefore assign the faster arrival time signal to the dimer 10. A similar situation may also pertain for ‘‘6’’: the faster arrival time feature has a cross section consistent with either a monomer or a dimer. Unfortunately, the mass resolution available was insufficient to decide this. All other ion mobility resolved signals observed for the native protein (in higher charge states) can be unequivocally assigned to monomers and can be classed into one of two different conformational groups
This journal is © the Owner Societies 2014
View Article Online
PCCP
Paper
repulsion between the extra charges. Then, the vertical detachment energy VDEn of an electron in a multiply charged anion can be described by: VDEn ¼ VDE
n1 X
Published on 12 December 2013. Downloaded by University of Western Ontario on 29/10/2014 04:33:36.
i¼1
Fig. 1 Normalized collision cross sections versus charge states for all monomeric species observed. Black symbols: multianion conformers resulting from native protein solutions. Green symbols: conformers electrosprayed from denatured (reduced) protein solutions.
according to their experimental collision cross sections (and charge state dependence thereof). In analogy to ref. 14 we call the group having small cross sections ‘‘highly-folded’’, whereas the more elongated structures are categorized as ‘‘partially-unfolded’’. Much higher average charge states are observed for the reduced protein under analogous electrospray conditions. Here, just one dominant conformation is resolved for each charge state. Where comparison at the same charge state is possible, the cross sections found for the reduced species are always significantly larger than those seen for ‘‘partially-unfolded’’ conformers resulting from native proteins. Consequently, we categorize the multianions generated by electrospraying reduced/denatured proteins as ‘‘highly-elongated’’. In addition to the arrival time distributions, Fig. 2 also shows the photoelectron spectra (266 nm detachment wavelength) corresponding to each resolved conformer. So as to allow for an easier overview, the figure has been plotted such that each column shows the PE spectra of a common conformational family. Generally, for a given conformer only one broad photoelectron spectral band is observed. Its position and shape depends on the charge state and the specific conformation. The highly-folded conformer group shows a strong spectral shift to lower binding energy as the charge is increased. For the highly-elongated group (derived from the reduced protein), the corresponding decrease in electron binding energies upon raising the charge is noticeably less. Interestingly, for the highest negative charge states of the highly-elongated conformers, the photoelectron bands become asymmetric and manifest a pronounced tail at their low energy side. The decrease in electron binding energy observed upon increasing the charge for the partially-unfolded group shows a behavior intermediate between these two cases. To first order the PE spectra of MCAs having localized excess charges can be interpreted based on the assumption that the electron binding energy is lowered by (additional) Coulomb
This journal is © the Owner Societies 2014
e2 4pe0 srij
(2)
Here, VDE is the detachment energy of a charged functional group (e.g. a carboxylate), s is a characteristic dielectric shielding constant (normally assumed to be around 2 to 5 (ref. 29–33)) and rij is the distance between the functional group and another (excess) charge residing on the molecule. Typically, the excess negative charges on multiply charged protein anions are likely to reside on deprotonated carboxylate groups, corresponding to a VDE of characteristically 3.5 to 4.5 eV, depending on the –COO local chemical environment.34,35 We assume that the dominant photodetachment processes occurring in our experiment correspond to detaching electrons directly from these charged carboxylates. Where are these deprotonated carboxylate groups actually located and what are the final states after photodetachment? Given a total number of m carboxylic acid groups in the protein, and a much smaller number k of deprotonated carboxylates (corresponding to the number of excess charges), the number of possible charge isomers initially present can be calculated as the permutation m!/[k!(m k)!]. Taking into account that each of the k electrons can be removed (in principle), the number of possible final ionized states is km!/[k!(m k)!]. a-Lactalbumin comprises 21 carboxylic acid groups. Therefore, as an example, there are 101 745 final state charge isomers possible after removing an electron from charge state 5. This ‘‘charge isomer’’ multiplicity is probably the main reason why our photoelectron spectra consist of rather broad bands (large initial state vibrational excitation levels likely also contribute). For a more quantitative analysis, the adiabatic electron affinities were extracted from the low energy flanks of the corresponding PE spectra. Most spectra show a more or less pronounced low energy tail so that a simple linearization is quite subjective. Instead, we determine upper and lower limiting values of the ADE by linearly extrapolating both the tail (lower limit) and the first linearly increasing part of the PE spectrum above this tail (upper limit). The results are displayed as a function of the normalized cross sections in Fig. 3. To exclude multiphoton detachment or delayed relaxation of electronically excited states as significantly contributing to the PE spectra, a number of measurements were also carried out at a detachment wavelength of 213 nm (ESI,† Fig. S3). ADE trends are consistent with the 266 nm measurements. As can already be inferred from the conformer resolved PE spectra (Fig. 2), the ‘‘highly-folded’’ conformation group shows a dramatic decrease in ADE upon deprotonation of additional carboxylic acid groups, whereas the corresponding collision cross sections remain approximately constant. On the other hand, the ‘‘highly-elongated’’ form shows a comparatively small decrease in ADE while the cross sections are seen to increase much more significantly upon raising the charge. For further analysis, repulsive Coulomb barriers have also been estimated.
Phys. Chem. Chem. Phys., 2014, 16, 3007--3013 | 3009
View Article Online
PCCP
Published on 12 December 2013. Downloaded by University of Western Ontario on 29/10/2014 04:33:36.
Paper
Fig. 2 Conformer resolved photoelectron spectra (266 nm) of multianions in various charge states obtained upon electrospraying native (left) and denatured a-lactalbumin (right) – in order of their arrival time. Also shown are the corresponding arrival time distributions (ATD) with resolved conformers indicated ((1)/(2)). Dotted curves (red) show arrival time distributions acquired under significantly higher injection energy conditions associated with isomerization.
We use the turning point of the high EBE tail in the PE spectra as a criterion for determining the RCB (as the difference between the photon energy and this energy). The resulting RCB estimates are shown in Fig. S4 (ESI†) as a function of charge state. ADE and RCB values show similar trends. How can these trends be rationalized? Typically, the collision cross section of an MCA having localized charges is directly correlated with the average distances between the charged functional groups. Then according to eqn (2), the detachment energy must be larger for elongated structures in comparison to
3010 | Phys. Chem. Chem. Phys., 2014, 16, 3007--3013
highly-folded structures of the same charge state because of the larger distances rij between the charged carboxylate groups. This is to first order the reason, why the experimentally determined ADE values of the octa- and nonaanions are larger for the highly-elongated conformation (denatured protein), smaller for the partially-unfolded group and even smaller in the case of the highly-folded species. Furthermore, the multianions derived from the reduced protein are apparently much better able to ‘‘compensate’’ additional negative charges than those resulting from the native protein,
This journal is © the Owner Societies 2014
View Article Online
PCCP
Paper
Published on 12 December 2013. Downloaded by University of Western Ontario on 29/10/2014 04:33:36.
We next introduce L, the length of one full helical turn. It is given by: pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi L ¼ h2 þ p2 d 2 (4) NL then defines the overall length of the coiled biopolymer chain which remains constant during the unfolding process. Then the charge-dependent diameter of the helix can be written as: v0 ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1ffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u 2ffi u 1u 1 1 4 Op A (5) d ¼ t@ L2 L p 2 4 Nc
Fig. 3 ADEs of each ion mobility resolved conformer group as a function of their normalized cross sections (and total negative charges, see also Fig. 1). Error bars correspond to upper and lower bounds resulting from the procedure used to assign ADE values to the spectra shown in Fig. 2 (see text).
as indicated by the slope of the curve in Fig. 3. This is much smaller for the highly-elongated form (implying that the additional repulsive Coulomb potential energy can be more easily dissipated by distortion). As mentioned before, the PE spectra of the highest charge states of the highly-elongated group show a low energy tail which becomes dominant upon increasing the charge. Correspondingly, the uncertainty range of the specified ADE value increases. As the unfolding process progresses, more and more hydrogen bonds will break. Generally, the VDE of a deprotonated carboxylic acid group is likely to be somewhat larger (around 4.5 eV or even higher) if this group is surrounded by positively polarized hydrogen atoms. On the other hand, the VDE of an electrostatically well isolated carboxylate group lies around 3.5 eV. We propose that the unfolding induced conversion of ‘‘coordinated carboxylates’’ to increasingly isolated carboxylates is the reason why the low energy flanks of the corresponding PE spectral bands appear more and more asymmetric at the highest charge states accessed here. In a simple molecular picture, charging induced distortion of the reduced protein can be thought of as corresponding to the stretching-transformation of a helix towards a linear strand.36 Raising the charge leads to an increase in the screw pitch while the inner diameter of the helix decreases correspondingly. We have developed a simple model to estimate how this should affect the ADE. For this, we first map the helix structure onto a cylinder having a diameter d. Assuming that d is much smaller than the length of the corresponding cylinder its collision cross section can be calculated according to: O = Nhdc
(3)
N is the number of helical coils, h is the pitch (i.e. the width of one complete helical turn as measured along the cylinder axis) and c is a constant to convert the longitudinal cross section of the cylinder into its absolute collision cross section (averaged over many projections).
This journal is © the Owner Societies 2014
Using this equation, it is possible to calculate the helix diameter based on the experimental cross section, if the pitch and the constant c are known. Typically, one coil of a perfect a-helix consists of 3.6 amino acids (corresponding to a starting pitch of 5.4 Å). The reduced a-lactalbumin then contains 34.17 such coils if we assume that it has a corresponding helical structure. Next we calibrate the parameter c (see also the ESI†). For this we use a known protein structure comprising many different helices each having a characteristic number of coils. Specifically, the pdb structure file of horse heart myoglobin,37 a protein comprising 75% a-helices, was first numerically divided into its individual helices. We then calculated collision cross sections for each of these component helices using the projection approximation (via the program sigma38,39). The resulting collision cross sections show a linear dependence on the number of coils with a slope of c = 2.76. Using the collision cross sections experimentally determined for the multianions resulting from reduced a-lactalbumin, we can now determine cylinder diameters d and total heights (Nh) for every charge state – as shown in the ESI.† This in turn allows us to model the ADEs based on the further assumption of a specific distribution of excess charges on the corresponding cylinder surfaces. For simplicity, we assume an alternating distribution of excess charges symmetrically localized on opposite sides of the cylinder in order to maximize charge separations. ADEs can then be estimated according to eqn (2) by inserting appropriate values for s. For the ADE of a deprotonated carboxylate we assume lower and upper bounds of 3.25 and 4.25 eV, respectively (to take into account different amounts of ‘‘coordination’’ by hydrogen bonding).‡ When assuming s = 1, i.e. no dielectric shielding, the slope of the theoretical line is noticeably steeper than experiment. Good agreement is obtained for s = 1.5 implying that even for the highly distorted structures studied here, there is still significant charge shielding by reason of polarization. If we further take into consideration that in reality carboxylates are not uniformly distributed but that they are instead localized at specific positions defined by the protein sequence, the correspondingly lowered distances between excess charges in specific regions would lead to an even larger effective shielding constant – consistent with literature assessments on the basis of other types of measurements. Note that an alternate way to estimate s – not ‡ To first order eqn (2) can also be analogously formulated in terms of ADEs.
Phys. Chem. Chem. Phys., 2014, 16, 3007--3013 | 3011
View Article Online
Published on 12 December 2013. Downloaded by University of Western Ontario on 29/10/2014 04:33:36.
Paper
PCCP
Fig. 4 Experimental ADEs of highly-elongated multianions (from the reduced protein) in comparison to a distorted helical model calculation plotting the average between upper/lower ADE bounds (indicated by shading – see text for details). Predictions are shown both for s = 1 and 1.5. The latter dielectric shielding constant provides the best agreement between the model and experiment.
pursued here – would be to calculate the repulsive Coulomb barrier for a specific value of s and to compare this to experimental RCB values. Finally, it is instructive to estimate the electrostatic forces acting on such a (helical) biopolymer and to compare these results to force-extension measurements on individual protein molecules as performed using atomic force microscopy or optical tweezer setups.40 In such experiments molecules are uniaxially stretched until at a certain force they start to unfold and/or bonds begin to rupture. For the comparison, we only need to consider axial tension. Therefore an electrostatic model even simpler than that of Fig. 4 suffices. We now assume that the negative charges (ze) are uniformly distributed over a chain of length l0 = Nh (i.e. number of coils times the pitch, see above). Neglecting shielding, the terminal charges experience an effective repulsive force comprising contributions from all remaining (z 1) charges: FðzÞ ¼
z1 e2 l0 2 X 1 2 4pe0 z 1 i i¼1
(6)
Fig. 5 displays the corresponding forces for a rigid chain of length l0 = 18.5 nm – as a function of the total number of localized excess charges z. Between z = 2 and 20, the electrostatic force increases by about three orders of magnitude (ca. 0.5–500 pN). By comparison, the nanomechanical measurements performed on various a-helices at room temperature show that the molecule begins to unwind its tertiary structure upon applying a stretching force in the range of o10–20 pN (typically described as the entropic regime). According to molecular dynamics simulations, higher forces of up to several nN then give rise to further transitions from a-helix to beta sheet secondary structures.41 Finally, forces significantly in excess of 1 nN are required in order to break a covalent bond.42
3012 | Phys. Chem. Chem. Phys., 2014, 16, 3007--3013
Fig. 5 Modelled electrostatic axial force of a uniformly charged chain (as described by eqn (6)) for a total chain length of l0 = 18.5 nm for different charge states.
Despite the obvious shortcomings of our crude electrostatic model, the order of magnitude agreement between the force required to transform a peptide chain from a-helix to beta sheet and that experienced by a terminal charge in a rigid chain of comparable z and l0 is intriguing. Conceivably, quantifying the structural transitions associated with electrostatically charging proteins in the gas-phase may offer a complementary approach to single biomolecule nanomechanics.
4. Conclusion In conclusion, we have recorded conformer resolved photoelectron spectra for all negative charge states observed upon electrospraying the native and the chemically reduced protein a-lactalbumin. By determining the collision cross sections of every resolvable conformer, we were able to divide the resulting gas phase multianion structures into three conformational groups: highlyfolded, partially-unfolded and highly-elongated structures. These conformational groups differ strongly in terms of the dependence of their experimental adiabatic detachment energies on the negative charge state. In particular, if the charge of the highlyfolded conformer group is increased (by further deprotonation of carboxylic acids functionalities) a large decrease in electron binding energy is observed. In contrast, the highly-elongated form shows a much smaller reduction in ADE upon further charging. This is due to much larger distances between excess charges in the reduced proteins. To qualitatively describe these observations, we have formulated a simple electrostatic model in which the reduced form is described as a helix which extends upon increasing charge. Within this model it was necessary to use an effective dielectric shielding constant of 1.5 in order to obtain agreement with experimental values, implying the presence of significant charge shielding due to polarization or dipole fluctuations. Even coarser estimations based on point-charge modelling indicate that the repulsive Coulomb forces prevailing for the higher multianionic charge states are
This journal is © the Owner Societies 2014
View Article Online
PCCP
Published on 12 December 2013. Downloaded by University of Western Ontario on 29/10/2014 04:33:36.
of the same order as the nanomechanical forces necessary to unfold individual protein and peptide molecules by uniaxial stretching. In future work, it will prove interesting to combine ion-mobility PES measurements with elasticity models in order to better understand the correlation between charge and structure of biopolymers.
Acknowledgements MK and PW acknowledge support by the Deutsche Forschungsgemeinschaft (DFG) towards developing the IMS-PES apparatus used in this study. MOW and MK also acknowledge support by the Fonds der Chemischen Industrie (FCI). RA and PD thank Marion Girod for recording some of the high resolution mass spectra.
References 1 D. E. Clemmer, R. R. Hudgins and M. F. Jarrold, J. Am. Chem. Soc., 1995, 117, 10141. 2 C. S. Hoaglund, S. J. Valentine, C. R. Sporleder, J. P. Reilly and D. E. Clemmer, Anal. Chem., 1998, 70, 2236. 3 D. Balbeur, J. Widart, B. Leyh, L. Cravello and E. De Pauw, J. Am. Soc. Mass Spectrom., 2008, 19, 938. 4 D. Clemmer and M. F. Jarrold, J. Mass Spectrom., 1997, 32, 577. 5 S. L. Bernstein, N. F. Dupuis, N. D. Lazo, T. Wyttenbach, M. M. Condron, G. Bitan, D. B. Teplow, J. E. Shea, B. T. Ruotolo, C. V. Robinson and M. T. Bowers, Nat. Chem., 2009, 1, 326. 6 C. S. Hoaglund-Hyzer, A. E. Counterman and D. E. Clemmer, Chem. Rev., 1999, 99, 3037. 7 S. I. Merenbloom, T. G. Flick, M. P. Daly and E. R. Williams, J. Am. Soc. Mass Spectrom., 2011, 22, 1978. 8 Y. Kang and D. J. Douglas, J. Am. Soc. Mass Spectrom., 2011, 22, 1187. 9 K. Thalassinos, S. E. Slade, K. R. Jennings, J. H. Scrivens, K. Giles, J. Wildgoose, J. Hoyes, R. H. Bateman and M. T. Bowers, Int. J. Mass Spectrom., 2004, 263, 55. 10 C. Bleiholder, N. F. Dupuis, T. Wyttenbach and M. T. Bowers, Nat. Chem., 2011, 3, 172. 11 D. E. Clemmer and S. J. Valentine, Nat. Chem., 2009, 1, 257. 12 H. Shi, N. A. Pierson, S. J. Valentine and D. E. Clemmer, J. Phys. Chem. B, 2012, 116, 3344. 13 T. Wyttenbach and M. T. Bowers, J. Phys. Chem. B, 2011, 115, 12266. 14 S. J. Valentine, J. G. Anderson, A. D. Ellington and D. E. Clemmer, J. Phys. Chem. B, 1997, 101, 3891. 15 D. S. Gross, P. D. Schnier, S. E. R. Cruz, C. K. Fagerquist and E. R. Williams, Proc. Natl. Acad. Sci. U. S. A., 1996, 93, 3143. 16 D. S. Gross and E. R. Williams, J. Am. Chem. Soc., 1995, 117, 883. 17 P. D. Schnier, D. S. Gross and E. R. Williams, J. Am. Chem. Soc., 1995, 117, 6747.
This journal is © the Owner Societies 2014
Paper
¨ffler, J. Friedrich, 18 J. M. Weber, I. N. Ioffe, K. M. Berndt, D. Lo O. T. Ehrler, A. S. Danell, J. H. Parks and M. M. Kappes, J. Am. Chem. Soc., 2004, 126, 8585. 19 L. S. Wang and X. B. Wang, J. Phys. Chem. A, 2000, 104, 1978. 20 L. S. Wang, C. F. Ding, X. B. Wang and S. E. Barlow, Rev. Sci. Instrum., 1999, 70, 1957. 21 K. Matheis, L. Joly, R. Antoine, F. Lepine, C. Bordas, O. T. Ehrler, A. R. Allouche, M. M. Kappes and P. Dugourd, J. Am. Chem. Soc., 2008, 130, 15903. 22 L. S. Wang, C. F. Ding, X. B. Wang, J. B. Nicholas and B. Nicholas, Phys. Rev. Lett., 1998, 81, 2667. 23 X. B. Wang and L. S. Wang, Nature, 1999, 400, 245. 24 M. Vonderach, O. T. Ehrler, K. Matheis, T. Karpuschkin, E. Papalazarou, C. Brunet, R. Antoine, P. Weis, O. Hampe, M. M. Kappes and P. Dugourd, Phys. Chem. Chem. Phys., 2011, 13, 15554. 25 M. Vonderach, O. T. Ehrler, K. Matheis, P. Weis and M. M. Kappes, J. Am. Chem. Soc., 2012, 134, 7830. 26 M. Vonderach, O. T. Ehrler, P. Weis and M. M. Kappes, Anal. Chem., 2011, 83, 1108. 27 P. Kruit and F. H. Read, J. Phys. E: Sci. Instrum., 1983, 16, 313. 28 O. Cheshnovsky, S. H. Yang, C. L. Pettiette, M. J. Craycraft and R. E. Smalley, Rev. Sci. Instrum., 1987, 56, 2131. 29 X. Song, J. Chem. Phys., 2002, 116, 9358. 30 G. N. Patargias, S. A. Harris and J. H. Harding, J. Chem. Phys., 2010, 132, 235103. 31 T. Simonson and C. L. Brooks, J. Am. Chem. Soc., 1996, 118, 8452. 32 W. C. Guest, N. R. Cashman and S. S. Plotkin, Phys. Chem. Chem. Phys., 2011, 13, 6286. 33 J. W. Pitera, M. Falta and W. F. van Gunsteren, Biophys. J., 2001, 80, 2546. 34 X. B. Wang, C. F. Ding and L. S. Wang, Phys. Rev. Lett., 1998, 81, 3351. 35 H. K. Woo, X. B. Wang, K. C. Lau and L. S. Wang, J. Phys. Chem. A, 2006, 110, 7801. 36 Note that denatured lactalbumin actually has low helix content in solution (R. Montserret and R. Antoine, private communication) such that our model has to be regarded as highly simplified. 37 D. M. Copeland, A. S. Soares, A. H. West and G. B. RichterAddo, J. Inorg. Biochem., 2006, 100, 1413. 38 T. Wyttenbach, G. von Helden, J. J. Batka, D. Carlat and M. T. Bowers, J. Am. Soc. Mass Spectrom., 1997, 8, 275. 39 T. Wyttenbach, M. Witt and M. T. Bowers, J. Am. Chem. Soc., 2000, 122, 3458. 40 M. Sotomayor and K. Schulten, Science, 2007, 316, 1144, and references therein. 41 Z. Qin and M. J. Buehler, Phys. Rev. Lett., 2010, 104, 198304. 42 M. Grandbois, M. Beyer, M. Rief, H. Clausen-Schaumann and H. E. Gaub, Science, 1999, 283, 1727.
Phys. Chem. Chem. Phys., 2014, 16, 3007--3013 | 3013
