The link between structure and function of a given protein is a principal tenet of biology. The established approach to understand the function of a protein is to ‘solve’ its structure and subsequently investigate interactions between the protein and its binding partners. However, structure determination via crystallography or NMR is challenging for proteins where localized regions or even their entire structure fail to fold into a three-dimensional form. These so called IDPs (intrinsically disordered proteins) or intrinsically disordered regions constitute up to 40% of all expressed proteins, and a much higher percentage in proteins involved in the proliferation of cancer. For these proteins, there is a need to develop new methods for structural characterization which exploit their biophysical properties. IM (ion mobility)–MS is uniquely able to examine both absolute conformation(s), populations of conformation and also conformational change, and is therefore highly applicable to the study of IDPs. The present article details the technique of IM–MS and illustrates its use in assessing the relative disorder of the wild-type p53 DNA-core-binding domain of cellular tumour antigen p53. The IM data were acquired on a Waters Synapt HDMS instrument following nESI (nanoelectrospray ionization) from ‘native’ and low-pH solution conditions.
IM (ion mobility)–MS measurements on IDPs (intrinsically disordered proteins)
The technique of IM–MS has emerged as a powerful tool for the study of macromolecular structures [1–4]. It provides the rotationally averaged collision cross-section of gas-phase ions of proteins and nucleic acids along with mass-to-charge (m/z) information. Careful optimization of the nESI (nanoelectrospray ionization), source optic voltages and pressures allows for the non-covalent interactions to be preserved and a gentle transfer of intact macromolecules from solution into the solvent-free environment of a mass spectrometer. Following nESI, all proteins are present in a number of different charge states, commonly as (multiply) protonated species (for positive ionization) or (multiply) deprotonated (negative ionization); this is termed a CSD (charge-state distribution). Structured folded proteins with little conformational flexibility in solutions where the protein is buffered appropriate to its pI, give rise to gaseous ions carrying a relatively low number of charges, presenting a narrow CSD centred on high m/z values . In contrast, less structured or unfolded proteins can accommodate a larger variation in the number of protons related to the varied availability of ionizable sites on the protein surface, in turn a feature of the conformational flexibility in solution . Analysis of the relative intensities of ions in CSDs enables an indirect assessment of conformational families present in solution. Frimpong et al.  used CSD analysis following ESI (electrospray ionization)–MS to determine the conformational heterogeneity of monomeric α-synuclein, a neuronal protein which has a special relevance for understanding Parkinson's disease . Although this protein is highly disordered, it was found to populate four distinct conformational states, ranging from a compact highly structured one to a random coil .
IM–MS provides another dimension to gas-phase studies of biomolecules garnering more direct information on the molecular species ‘shape’ (collision cross-section, in Å2) along with the CSD, but does not provide atomic resolution compared with NMR and X-ray crystallography. IM–MS has been applied successfully to examine protein dynamics [5,9], protein–ligand and protein–protein interactions  and shows great promise for the structural characterization of IDPs. Bernstein et al.  performed experiments on α-synuclein which combine nESI–MS with IM measurements. Their findings indicate the presence of compact structures when spraying from low-pH solutions, whereas higher charge states occurring at physiological pH (6.8) have larger collision cross-sections and are relatively unfolded. The transition from compact to extended conformations occurs from charge state z=−8 to −9, where a 50% increase in cross-section is observed. IM–MS has also been utilized to study the structure of the DBD (DNA-binding domain) of the tumour-suppressor protein p53 with and without the functional zinc . Other research using IM–MS to study IDPs has examined changes in conformational equilibria of the intrinsically disordered HMGA (high-mobility group A) chromatin factors, owing to post-transitional modifications or sequence deletion , and the ability of IDPs to fold on binding to a partner using the salivary tannin-binding proteins . A variety of configurations of IM instrumentation have been developed: DT (drift tube) IM–MS, TW (travelling wave) IM–MS and FAIMS (high-field asymmetric-waveform ion-mobility spectrometry). These techniques are described briefly below.
Types of mobility separation
In a typical experiment, ion optics direct a pulse of ions into the drift cell, located within a vacuum system and across which is applied a weak electric field (5–100 V·cm−1). The ions are pulled through the drift cell, and their progress is impeded by collisions between the ions and a buffer gas, commonly helium. The movement of the ion through the gas depends on its mobility which is determined by the ratio of the electric field strength (E) to buffer gas number density (N). At low E/N or the low field limit, the velocity of the ions (υ) is low and independent of the strength of the electric field. In this case, the ions will drift with a constant velocity (υd), which is directly proportional to E and also depends on the mobility (K) as shown in eqn (1):
where K0 is the reduced mobility [the measured mobility K standardized for pressure (760 Torr) (1 Torr=133.322 Pa) and temperature (273.15 K)], z is the ion charge state, e is the elementary charge, μ is the reduced mass of the ion-neutral pair, kB is the Boltzmann constant, and T is the gas temperature. This experimental cross-section which is buffer-gas-dependent can be compared with collision cross-sections derived from co-ordinates obtained from other structural investigations (for example NMR structure co-ordinates or X-ray crystallography), or from computational predictions, to obtain atomically detailed conformational information. This feature is hugely beneficial as it allows for interpretation of protein folding/unfolding pathways to be addressed as well as clarifying the possible conformations adopted in the solvent-free environment.
Waters MS Technologies introduced the first commercially available IM–MS instrument, the Synapt HDMS in 2006 . This mobility separator uses a SRIG (stacked ring ion guide) as a mobility cell which consists of a series of ring electrodes with opposite phases of RF (radio frequency) voltage applied to consecutive electrodes. In order to propel ions through the SRIG, a transient DC (direct current) voltage (pulse) is sequentially superimposed on the top of the RF potential. This provides a continuous sequence of ‘travelling waves’ on which ions can ‘surf’, thereby converting the SRIG into a TWIG (travelling wave ion guide). At elevated gas pressure, ion species of high mobility roll over the wave less often than species of low mobility and so their transit time through the TW device is shorter, thus permitting the mobility-based separation of ions within the device. In the short time these instruments have been available, they have made a significant impact on the use of IM–MS for the study of macromolecular systems and in particular for the study of intact proteins and protein complexes [17–19]. The mobility of ion species obtained from these TW devices is not directly related to collision cross-section as in DT IM–MS instruments. In order to obtain the cross-sectional information, data obtained from Synapt instrumentation must be calibrated against known DT data, preferably with similar arrival times and masses. Using standards that have been previously measured in helium on research-grade DT instrumentation, it is possible to convert mobilities measured with nitrogen as a drift gas in Synapt into mass-selected ‘helium-based’ collision cross-sections [3,20,21].
An alternative method which separates ions on the basis of their mobility is FAIMS or differential mobility spectrometry [22,23]. In this technique, ions pass between two electrodes in the presence of a tangential gas flow. An alternating electric field, perpendicular to the gas flow, is applied such that the ions oscillate between the two electrodes and disperse according to their difference in mobilities. Unlike DT IM–MS, a high electric field is used and so mobility (and therefore collision cross-section) is not independent of electric field strength. In some cases, the heating of ions in high electric fields can induce protein unfolding. Nonetheless, FAIMS has proven useful in the separation of complex biological mixtures [24–26].
In this section, we illustrate the use of TW IM–MS to discern the relative disorder of the cellular tumour antigen p53. Known as ‘the guardian of the genome’, the transcription factor p53 has been the subject of a vast amount of research since its discovery in 1982 by Reich and Levine  and its analysis continues to be central to the fight against cancer. Upon a cellular stress, activation of p53 may trigger cell-cycle arrest, apoptosis and senescence, thereby destroying the DNA-damaged cells before they can proliferate. The protein consists of 393 amino acids and contains several distinct domains: the N-terminal transactivation domain, a proline-rich domain, the core DBD, the tetramerization domain and the C-terminal regulatory domain. Overall, p53 is highly disordered and has been shown to form both dimers and tetramers . The structure of the core DBD has been solved by X-ray crystallography and SAXS (small-angle X-ray scattering), both for wild-type p53 and several p53 core domain mutants . Although the core DBD forms a so-called ‘ordered’ region of p53, it possesses many disordered loops and it is in these that most p53 inactivation mutations are associated with nearly 50% of human cancer . Targeted studies into this core DBD of oncogenic p53 and its interaction with DNA are essential for anticancer drug-discovery efforts. We illustrate how IM–MS can be applied to the study of the conformation of p53.
MS of p53
Mass spectra were obtained on the TW IM–MS instrument, for p53 DBD from near-neutral [50 mM ammonium acetate, containing 10% (v/v) propan-2-ol (pH 6.8)] and from denaturing [50 mM ammonium acetate, containing 10% (v/v) propan-2-ol and 1% (v/v) formic acid (pH 1.5)] solution conditions (Figures 1A and 1B respectively). From the buffered solution, a CSD of the wild-type p53 monomer is observed with 8≤z≤18 and a few additional multimeric species (Figure 1A). Each of the monomer charge states is attributable to the zinc-bound form of the protein (calculated molecular mass, 24615.5 Da; observed molecular mass, 24612.4 Da), which suggests that all conformations present in solution retain the zinc-binding site, and that it is not perturbed by the nESI desolvation process or by any subsequent coulombically driven unfolding (see below). The most abundant species in Figure 1(A) correspond to monomeric [M+9H]9+ and [M+10H]10+ and are indicative of a compact form of the protein with a limited number of charge-accessible residues available for protonation. Two unique m/z values are present for a dimeric form of p53, [2M+zH]z+ where z=13 and 15, representing higher-order multimeric association in solution; again zinc is bound stoichiometrically. Higher charged species, where 11≤z≤18, are observed at low abundance, evidence of a small population of more unfolded states of p53. Even under ‘near-native’-like conditions, p53 displays a wide CSD (8≤z≤18), which we highlight as signature behaviour of IDPs in vacuo as reported by MS.
nESI spectra of p53 obtained on the TW IM–MS
At low pH (Figure 1B), the observed spectrum displays a wider charge state envelope over the acquired mass range (400–4000 m/z). Under these conditions, the most dominant species of p53 observed is a monomer. The deconvoluted molecular mass (24547.9 Da) is less than the molecular mass of the ‘native’ p53 monomer (24612.4 Da). This is due to the loss of the tetrahedrally co-ordinated zinc atom (molecular mass, 65.4 Da) which destabilizes the structure of the protein. The observed charge states in the range 9≤z≤34 attributable to facile protonation of solvent-exposed basic residues. These sites are proposed to be basic residues: there are 19 arginine, eight lysine and nine histidine residues in the p53 molecule, as well as at the N-terminus where proton addition could potentially occur. The spectrum displays a trimodal charge distribution with centres at z=27, 18 and 11, suggesting that at least three conformational families are present in the solution and transferred into the gas phase. The distribution in the range 22≤z≤34 is more populated than the range 12≤z≤22, and so we can infer that the denatured protein favours more extended conformations. In summary, as we compare Figures 1(A) and 1(B), addition of acid (lower pH) causes a shift in the most intense peaks to higher charge (lower m/z), and a decrease in intensity of lower charge (higher m/z) as the folded conformation is lost and the unfolded state is predominantly populated.
IM–MS of p53
Along with MS, IM measurements have been performed on the p53 DBD sprayed from two different solutions. The collision cross-sections obtained are in Supplementary Tables S1–S3 at http://www.biochemsoctrans.org/bst/040/bst0401021add.htm according to the procedure outlined in the Supplementary Online Data (Instrumental conditions section) at http://www.biochemsoctrans.org/bst/040/bst0401021add.htm and plotted as a function of charge state in Figure 2. The data taken at pH 1.5 (Figure 2A) shows a smooth near-linear increase in cross-section with respect to charge, which can be explained by the range of dynamic states of the protein in solution, resulting in differently populated protonated states, which have little structure, and are easily subject to coulombically driven unfolding once in the gas phase. Here, in the absence of solvent, each additional proton added to the flexible protein can cause local unfolding due to unfavourable interactions with other proximal proton(s). However, it also displays some regions of stability seen as ‘shelving’ in the cross-section values (z=8–10 and 12–14). For each of the three lowest charge states, (z=9, 10 and 11), a single conformation was resolved with collision cross-sections of 1671 Å2 (1 Å=0.1 nm), 1697 Å2 and 1805 Å2 respectively. At these low charge states, the structure is most likely to be compact, but loss of the zinc atom indicates that there is much less secondary structure than the conformers reported (see below) from solutions buffered at pH 6.8. For 12≤z≤24, multiple unfoldomers for each charge state were observed.
Collision cross-sections against charge obtained on the TW IM–MS of p53
Using methodologies established previously [31–33], we have calculated the collision cross-sections for co-ordinates expected from a fully extended form of the p53 DBD, built using xleap within the AMBER molecular modelling package (http://ambermd.org/) and find this value to be 7289 Å2 (Figure 2A). Even the highest charge state experimentally observable has a significantly smaller collision cross-section (by 22%), suggesting that the DBD of p53 retains some secondary structure, or possibly intermolecular interactions, albeit transient.
From buffered conditions (Figure 2B), there are two dominant charge states (z=9 and 10). The z=9 species provides a single resolvable conformer (1657 Å2), indicative of a highly compact form of the protein, whereas for z=10, two species are resolved, with the most populated being 1669 Å2 and a lower-intensity extended form (2157 Å2). These conformers are not baseline-resolved, which could be due to the resolution of the Synapt HDMS, or else that they are in conformational equilibrium on the timescale of these IM–MS experiments (5–10 ms). The low collision cross-sections for z=8–10 are smaller than those calculated from the co-ordinates obtained from the crystal structure deposited by Wang et al.  using the trajectory method . The trajectory method  provides the most accurate estimate of the collision cross-section for any given set of co-ordinates; we have not used it for the fully extended form of p53 since the size of this molecule is too large for our compilation of the MOBCAL software. The less accurate EHS (exact hard spheres) method gives qualitatively similar results  and so the two values are comparable. This follows the general trend for monomeric proteins that we have reported previously , and reveals conformers which are smaller than the equivalent charge state species obtained from the low-pH solution, suggesting that the bound zinc has a stabilizing effect. The second much lower populated conformational family which starts with the more extended conformer for z=10 does coincide with the value calculated from crystal structure co-ordinates. The dimers that are present following nESI have significantly lower (by 23%) collision cross-section than that obtained from the crystal structure; this form is clearly not stable in the gas phase under our experimental conditions.
For the monomeric p53 DBD, with increasing charge, a pathway of unfolding is clearly visible, and the collision cross-section increases. However, there is far more shelving than found at pH 1.5, suggesting a more concerted unfolding and again more stability to the protein. The calculated collision cross-sections for each charge state are higher than those of the equivalent charge state conformers obtained from the physiological pH solution. Figure 2(B) provides evidence of several stable states on the unfolding pathway. With increasing charge (z=10–13 and 15–18), at least two conformers can be observed. Although p53 has unfolded to a more extended structure, it still retains some non-covalent interactions that sustain the protein in these different conformations for the duration of the mobility experiment. Figure 3 presents the ATDs (arrival time distributions) for each charge state, showing the abundance of the conformation within the protein under buffered conditions. This shows clearly how the majority of the protein is located in the z=9 and 10 region, in a compact conformer, which also exists for z=11–13, although here it is gradually shifting to a more extended form (Figure 2B). There are other more extended conformations; however, the intensity of these is significantly lower. We have assigned collision cross-sections to these as resolvable conformers (Figure 2B and Supplementary Table S2); however, the lack of baseline resolution (Figure 3) suggests interconverting conformers and highlights the flexibility of this protein in these extended forms over the timescale of our experiment.
ATD plot of p53 under ‘native-like’ conditions
IM–MS provides attractive approaches to the study of these flexible disordered systems. These methods can test the (many) biophysical assertions and predictions regarding IDPs. The technique offers the capability to probe the conformational flexibility, binding and folding on binding and could also be adopted as a screening tool to assess the effects of interacting ligand molecules or post-translational modifications on IDP conformations. It must be noted, however, that IM is an inherently low-resolution technique. It relies on molecular dynamic simulations, high-resolution NMR or X-ray crystallographic structures to make inferences about the gas-phase structure of the ions being investigated; however, for proteins which possess much conformational flexibility and intrinsic disorder, we believe it has much to offer.
Intrinsically Disordered Proteins: A Biochemical Society Focused Meeting held at University of York, U.K., 26–27 March 2012. Organized and Edited by Jennifer Potts (York, U.K.) and Mike Williamson (Sheffield, U.K.).
arrival time distribution
exact hard spheres
high-field asymmetric-waveform ion-mobility spectrometry
intrinsically disordered protein
stacked ring ion guide
We thank Waters Micromass Technologies and the British Mass Spectrometry Society (BMSS) for financial support of our work. This work has been funded by the award of a Biotechnology and Biological Sciences Research Council Strategic Industrial Case studentship to E.J. in collaboration with Waters MS Technologies Centre.
Present address: Department of Molecular Structure, Amgen Inc., One Amgen Center Drive, Thousand Oaks, CA 91320, U.S.A.