Introduction

The conformation of a protein in solution is one of the main factors affecting the CSD it will generate in ESI-MS [1, 2]. However, the mechanism behind this phenomenon is still the object of debate [3]. In this work, we investigate the influence of experimental conditions, such as curtain-gas flow rate, pH, and solvent surface tension, by nano-ESI-MS [4] in order to aid interpretation of spectral changes in ESI-MS experiments and to better understand the mechanism behind the conformation dependence of protein CSD.

Application of ESI-MS to protein-folding studies gives useful information, which is complementary to that obtained by other biophysical methods and offers specific advantages in the analysis of poorly populated conformational states [5, 6]. Increasing compactness [7, 8, 9, 10] of the polypeptide chain generally results in lower charge states, providing a fast and sensitive tool for characterization of conformationally heterogeneous samples. Understanding and controlling the influence of instrumental settings on the CSD is of critical importance in interpreting conformational effects in ESI-MS [11]. The operating conditions can affect the stability of the protein during the electrospray, resulting in protein unfolding. Such effects can even be exploited for folding studies, but require at the same time accurately controlled conditions. The effects of temperature and cone voltage have been extensively documented [3, 11, 12, 13, 14, 15]. As reported previously, we note that the curtain-gas flow rate also has a critical effect on protein spectra obtained by nano-ESI-MS [16]. Other groups have reported effects of curtain gas on protein CSD [17] and on protein conformations in the gas phase [18].

Several lines of evidence indicate surprising lack of dependence of protein CSD on the pH of the original solution [19], as long as pH-dependent unfolding does not take place. The similar pH range of unfolding transitions monitored by ESI-MS and other spectroscopic methods under appropriately controlled conditions [2, 6, 9, 20, 21, 22] rules out the possibility that this phenomenon is a result of dramatic pH changes in the solution during the ESI process [23, 24, 25]. Thus, it seems that there might be a very limited influence of pH on the ESI process and vice versa. The strong pH dependence of protein stability is an obstacle to investigating the role of the pH on the ionization mechanism itself, uncoupled from conformational effects. The use of stabilizing agents, for example glycerol, has helped distinguish these two possible contributions to spectral changes in pH-variable experiments [26]. The results confirmed the predominance of conformational effects.

It has recently been suggested by several authors that the multiple charging of macromolecules in ESI-MS is limited by the Rayleigh-instability charge (q R ) of the droplets that generate the gas-phase ions [27, 28, 29]. Such behavior would be consistent with the charged-residue model for production of gas-phase protein ions [30]. In support of this hypothesis it has been shown that the maximum charge state of folded proteins fluctuates between 65% and 110% of the calculated number of elementary charges at the point of Rayleigh instability (z R ) for droplets of the same radius as the globular protein structure [29]. In this view, the ESI droplet generating the gas-phase ion carries a neutral protein molecule that acquires all the charges of the droplet when the solvent evaporates (except for some charge reduction that might occur by small-ion evaporation [31]). Charges carried by the protein would not affect the stability of the droplets as predicted by the Rayleigh equation, because they would be held together by the covalent bonds of the protein and not by the surface tension of the solvent. Thus, close correspondence between the observed maximum charge state of proteins in ESI-MS and z R would have two implications:

  1. 1.

    preexisting charges on the protein would not contribute to the final charge state in ESI-MS, and

  2. 2.

    all the charges carried by the precursor ESI droplet would be transferred to the protein molecule.

A testable prediction of this model is that the maximum charge state of macromolecules (at least those with globular structure) in ESI-MS should depend on the surface tension of the solvent according to the Rayleigh equation:

$$ q_R = z_R e = 8\pi \left( {\varepsilon _0 \gamma R^3 } \right)^{1/2} $$

where e is the elementary charge (1.6×10−19 Coulomb), ε o is the permittivity of vacuum (8.85×10−12 Coulomb2 N–1m−2), γ is the surface tension of the solvent, and R is the radius of the droplet. A recent paper [28] reports data for dendrimers, unfolded cyt c, poly(ethylene glycol)s, and 1,n-diaminoalkanes that suggest dependence of the maximum and average charge state in ESI-MS on the surface tension of the least volatile solvent component. It is generally argued that the droplets at later stages of the ESI process will contain almost exclusively the solvent component with the lowest vapor pressure. However, if the CSD of acid-unfolded cyt c in water–acetic acid were limited by the low surface tension of the organic cosolvent, employment of HCl instead of acetic acid to adjust the pH of the solution should result in significantly higher charge states. On the contrary, published spectra of HCl-unfolded cyt c [9] look quite similar to those typically observed in water–acetic acid. In order to rule out that differences between CSD under these two conditions might have been masked by different experimental conditions, we repeated such measurements obtaining directly comparable data for several different proteins. Furthermore, we tested conditions that enable direct comparison of CSD in water and acetic acid solutions.

Materials and methods

Bovin Ubq, chicken Lyz, horse cyt c, and horse Mb were purchased from Sigma. The protein powders were directly dissolved in water to 5 mg mL−1, and than diluted to final 5 μmol L−1 under the desired solvent conditions. The pH of the solutions was adjusted prior to protein addition, which was found not to change the pH of the sample. Mass spectra were recorded on a Mariner time-of-flight, ESI mass spectrometer (Applied Biosystems) equipped with a nano-electrospray sample source. Spectra were recorded in positive-ion mode at spray-tip potential 1600–1700 V, nozzle-to-skimmer potential 10–30 V, nozzle temperature 85 °C, and curtain-gas flow rate ~0.6 L min−1, unless otherwise stated. The integration time was 3 s per spectrum. The reported traces are averaged over at least ten spectra. The settings were optimized by tuning the instrument on the major peak of protein spectra under non-denaturing conditions. The reference spectrum for the folded protein was identified, as a general rule, using the protein dissolved in water and progressively milder instrumental conditions. Samples were sprayed at room temperature. The capillaries were purchased from Protana (inner diameter 0.69 mm, inner diameter of the emitter tip ~1 μm, and “medium” length of the emitter tip). The software for data recording (Mariner Workstation version 4.0) and data processing (Data Explorer version 4.0.0.1) was supplied by Applied Biosystems. The fraction of folded protein from cyt c bimodal spectra was calculated as the sum of the intensities of the peaks corresponding to the 8+, 9+ and 10+ ions, divided by the sum over all the peaks in the spectrum (Σ i=8–10 I i /Σ i=1-n I i ). The peaks of the 8+, 9+ and 10+ ions characterize the spectrum of the folded protein in water under our experimental conditions (Fig. 1) [3].

Fig. 1
figure 1

Effect of curtain-gas flow rate at pH~7. Nano-ESI-MS spectra of cyt c (a–d) or Mb (e–h) in water. The flow rate of the nitrogen curtain gas was 0.6 (a, e), 0.7 (b, f), 0.9 (c, g), or 1.2 (d, h) L min−1. “H” indicates the peak of free heme (m/z = 616.4). The charge state z corresponding to the main peak is indicated in each panel

Full size image

Results

Effect of curtain gas flow rate

The effect of curtain-gas flow rate on spectra of cyt c and Mb dissolved in pure water is shown in Fig. 1. Up to 0.7 L min−1, no shift in the main (i.e. most abundant) charge state is observed. A minor shift in the average charge state (Σ i z i I i /Σ i I i ) of Mb from 8.4 to 8.9 is caused by increasing the gas flow rate from 0.6 to 0.7 L min−1. Above 0.7 L min−1, a clear shift of the CSD takes place, bringing the main charge state of cyt c from 9+ to 11+ and that of Mb from 9+ to 13+. Asymmetric tailing of the cyt c distribution towards high charge states is also noticeable, as is an increasing relative intensity of the peak of free heme in the Mb spectra.

Figure 2 summarizes the results of similar experiments carried out at lower pH. We chose pH 2.7 for cyt c and pH 3.8 for Mb, as conditions under which the folded and the unfolded protein are concomitantly detectable. The bimodal spectra obtained with these samples reflect the dynamic equilibria between different conformational states in the original solutions. The peak envelope at lower m/z values corresponds to the unfolded protein [1, 2]. Under these conditions, increasing curtain-gas flow rates cause conversion between distinct species rather than progressive shift of a single bell-shaped CSD. Gas flow rates above 0.7 L min−1 result in increasing fractions of unfolded protein. A parallel increase in the relative intensity of the peak of free heme can also be observed in Mb spectra.

Fig. 2
figure 2

Effect of curtain-gas flow rate at low pH. Nano-ESI-MS spectra of cyt c at pH 2.7 (a–d) or Mb at pH 3.8 (e–h) in water–acetic acid solutions. The flow rate of the nitrogen curtain gas was 0.6 (a, e), 0.7 (b, f), 0.9 (c, g), or 1.2 (d, h) L min−1. “H” indicates the peak of free heme (m/z = 616.4). The labels “apo” and “holo” indicate, respectively, the CSD of the apoprotein and holoprotein. The charge state z corresponding to the main peak of each envelope is indicated in each panel

Full size image

The spectral changes illustrated in Fig. 2 clearly indicate that protein unfolding is promoted by high gas flow rates during the analysis. As previously reported [16], this effect is fully reversible (data not shown), indicating that the folding equilibrium in the bulk solution does not change during the experiment. It is relevant to note that the main charge state of the folded proteins in Fig. 2 does not shift with increasing gas flow rates, suggesting that the shift observed at higher pH (Fig. 1) is due to minor conformational changes rather than to the influence of instrumental settings on the ESI process. Indeed, such a mechanism would also be expected to operate at lower pH. The data presented here suggest that high gas flow rates perturb the folding equilibrium in the ESI droplets, resulting in partial unfolding at neutral pH and in global unfolding at acidic pH. The lack of resolution between different conformational states at pH 7 would then cause the apparent shift of CSD in the spectra in Fig. 1.

Monitoring equilibrium unfolding transitions

As documented in Figs. 1 and 2, it is possible to identify a range in which the curtain-gas flow rate does not affect the conformational stability of the protein during the analysis. Similar experiments show that the same is true for other critical conditions, like cone voltage and nozzle temperature (data not shown). Thus, provided that none of the instrumental settings becomes the limiting factor for protein stability during the electrospray, it is possible to employ ESI-MS for quantitative analysis of equilibrium folding transitions. As illustrated in Fig. 3, the relative amount of folded and unfolded protein determined by nano-ESI-MS under such conditions matches closely the values obtained by established spectroscopic methods. In this case, we compare our data for the acid-induced unfolding of oxidized cyt c with those obtained by Goto [32] by monitoring tryptophan fluorescence and Soret absorbance in solution at 20 °C. Several other studies show that the dependence of cyt c and Mb conformations on pH and methanol concentrations monitored by ESI-MS is in close agreement with previous characterization by fluorescence and nuclear magnetic resonance spectroscopy [1, 2, 6, 8, 9, 20, 21]. Divergence between ESI-MS and far-UV CD data has been described [9, 10, 33]. This fact reflects the different specificity of these two methods for tertiary and secondary protein structure, respectively. These studies showed that when α-helix formation is uncoupled from tertiary-structure formation, as in peptides or denatured proteins at extreme pH, ESI-MS spectra are not affected by the coil-to-helix transition.

Fig. 3
figure 3

Acid-induced unfolding of cyt c monitored by nano-ESI-MS (filled squares), compared with published results [32] obtained by means of tryptophan fluorescence (open circles) and Soret absorption (open triangles). The error bars for the ESI-MS data represent the standard deviation over five independent experiments

Full size image

Lack of dependence of protein CSD on pH

To test the effect of large pH changes in nano-ESI-MS uncoupled from conformational transitions we used the extremely stable proteins Ubq and Lyz. Figure 4 shows comparison of Ubq spectra in water and acidic solutions, and the effect of addition of 60% (v/v, here and below) methanol under each condition. The data at pH 2.2 were obtained using HCl, formic acid, or acetic acid to acidify the solution. The spectra of Ubq in water vary between the two CSD shown in panel (a), with that centered on the 8+ peak being most frequently observed. These spontaneous fluctuations are observed at any different tested condition. Predominance of the 6+ peak correlates with lower total ESI current and higher relative amount of peaks corresponding to the Ubq dimer. In agreement with previous reports [34], addition of 60% methanol to Ubq solutions at pH~7 does not affect the ESI mass spectrum, resulting in the same CSD typically observed in water.

Fig. 4
figure 4

Effect of different acids on the CSD of Ubq. Nano-ESI-MS spectra in the absence of methanol (a–d) or in the presence of 60% methanol (e–h). Other solvent components and final pH values: water pH~7 (a, e) (the inset shows a frequently observed alternative CSD); water–HCl pH 2.2 (b, f); water–formic acid pH 2.2 (c, g); water–acetic acid pH 2.2 (d, h). The charge state z corresponding to the main peak of each envelope is indicated in each panel

Full size image

In the absence of methanol, lowering the pH of Ubq solutions from ~7 to 2.2 by addition of HCl has almost no effect on the CSD. Ubq is known to be folded at pH 2 [35]. That this is also true under our experimental conditions is shown by the fact that we can still induce the typical shift towards higher charge states reflecting protein unfolding by addition of 60% methanol. Such a control is important in order to rule out the possibility that the apparent lack of pH effects results from opposite influences of solvent conditions and conformational changes. This result shows that, in the absence of acid-induced unfolding, even large pH changes do not affect the ESI process.

Interestingly, the qualitative features of the Ubq spectrum at pH 2.2 strongly depend on the acid employed to adjust the pH of the solution. With formic acid, a bimodal spectrum is obtained, consistent with an equilibrium between folded and unfolded states in the original protein sample. The addition of 60% methanol results in a higher fraction of unfolded protein. In the presence of acetic acid, a single peak envelope corresponding to the unfolded protein is observed. These results can be interpreted in terms of progressive destabilization of the folded protein structure at pH 2.2 by the increasing hydrophobicity of the additive in the order: HCl<formic acid<acetic acid. This sequence also reflects the different pK a of these acids and, therefore, the increasing total concentration required to bring the solution to pH 2.2. A pH value of 2.2 corresponds to an acetic acid content of ~10%. The far-UV circular dichroism (CD) spectra of Ubq at pH 2.2 in the presence of the different acids are very similar (data not shown), indicating that the denatured protein under these conditions retains substantial secondary structure. A surprising result is that addition of 60% methanol to the sample of unfolded protein in water–acetic acid, pH 2.2, apparently shifts the conformational equilibrium towards the folded state, giving rise to a bimodal spectrum with an increased contribution of a native-like component. More experiments will be needed in order to confidently ascribe this effect to increased stability of compact conformations.

Lyz is another protein that is stably folded in solution at pH 2 [36]. The data reported in Fig. 5 indicate that, for this protein too, the spectra from water (pH~7) or acidic solutions (pH 2.2) are very similar, ruling out direct effects of the acidity of the sample on the ionization mechanism in this range of pH. The main charge state is 10+ in either water, water–HCl or water–acetic acid, although a higher relative amount of the 9+ peak characterizes the latter condition. As in the case of Ubq, addition of 60% methanol to the protein at pH~7 does not affect the observed CSD. The partially unfolded state of oxidized Lyz at low pH in the presence of 60% methanol [37] results in a main charge state of 11+. Thus, we could not confirm the previously reported shift of Lyz CSD towards higher m/z values induced by high methanol concentrations at low pH [38]. However, the methanol concentration in that case was 80%. High curtain-gas flow rates in the presence of 60% methanol result in more extensive unfolding, as indicated by a main charge state of 12+.

Fig. 5
figure 5

Effect of different acids on the CSD of Lyz. Nano-ESI-MS spectra in the absence of methanol (a–c) or in the presence of 60% methanol (d–f). Other solvent components and final pH values: water pH~7 (a, d); water–HCl pH 2.2 (b); water–acetic acid pH 2.2 (c, e, f). The spectrum in panel f was recorded at high curtain-gas flow rate (1.2 L min−1). The charge state z corresponding to the main peak is indicated in each panel

Full size image

Comparison of acetic acid and HCl solutions

Unlike Ubq, the proteins Lyz, cyt c and Mb allow comparison of a single conformational state, either folded or unfolded, in water–HCl and water–acetic acid at low pH. Such a comparison is of interest, because the surface tension of the solution is different in these two cases, and surface tension affects the limit charge of ESI droplets according to the Rayleigh equation. HCl has very little influence on the surface tension of water (the value for a 2 mol L−1 HCl solution is 72.25×10−3 N m−1 at 20 °C) and it is present at much lower concentrations than acetic acid at same pH (~6 mmol L−1 HCl compared with ~1.7 mol L−1 acetic acid in the samples at pH 2.2). The surface tension of “mature” ESI droplets from water–acetic acid solutions is considered to be close to that of acetic acid (27.7×10−3 N m−1 at 20 °C) [28]. Taking the value of water for the HCl solutions (72.75×10−3 N m−1 at 20 °C), the Rayleigh-limit charge in droplets of equal radius is predicted to differ by approximately 50%. The ratio between the two values will be:

$$ z_{{R{\left( {{\text{water}}} \right)}}} /z_{{R{\left( {{\text{acetic}}\;{\text{acid}}} \right)}}} = {\left( {72.75/27.7} \right)}^{{1/2}} = 1.62 $$

Figure 5 shows that the spectra of Lyz at pH 2.2, in the presence of either HCl or acetic acid, are very similar. The main charge state is 10+ in both cases. The maximum charge state is 12+ in water–HCl and 13+ in water–acetic acid. Again, comparison with the spectra obtained with additional 60% methanol provides a control that the protein does not unfold during electrospray in water–acetic acid pH 2.2 under our experimental conditions. Thus, Lyz also offers the possibility of direct comparison of the acetic acid solution with pure water. The main charge state is 10+ in both cases, although there is a shift in the average charge state from 10.3 in water to 9.5 in water–acetic acid. Thus, the reduction in surface tension caused by the presence of acetic acid leaves the protein CSD almost unaffected.

Similar results are found also with cyt c. Figure 6(a–d) shows typical spectra at pH 2.8 and pH 2.2. The experiment at pH 2.8 is of interest because it generates bimodal spectra that enable monitoring of the CSD of folded and unfolded protein at the same time. A water–acetic acid solution at pH 2.8 contains ~1% acetic acid. The results show that the CSD at either pH 2.8 or pH 2.2 are not significantly affected by the nature of the acid employed. The main charge state of the unfolded protein fluctuates between 18+ and 17+ under all tested conditions, while its maximum charge state ranges between 20+ and 22+. The main charge state of the folded protein is 9+ in the presence of either HCl or acetic acid. It is relevant to note that the CSD corresponding to the folded protein in the presence of acetic acid at pH 2.8 is almost identical with that observed in pure water (Fig. 1), showing a narrow distribution with strong predominance of the 9+ ion.

Fig. 6
figure 6

Effect of different acids on the CSD of cyt c (a–d) or Mb (e–h). Nano-ESI-MS spectra in the solvents: water–HCl pH 2.8 (a, e); water–acetic acid pH 2.8 (b, f); water–HCl pH 2.2 (c, g); water–acetic acid pH 2.2 (d, h). The charge state z corresponding to the main peak of each envelope is indicated in each panel

Full size image

Horse Mb is completely unfolded in solution at pH<3 and low ionic strength [39]. As summarized in Fig. 6(e–h), its main charge state in nano-ESI-MS oscillates between 23+ and 24+ at pH 2.8, in the presence of either HCl or acetic acid, and in water–HCl at pH 2.2. The maximum charge state is 28+ in water–acetic acid pH 2.8, and 27+ in water–HCl. At pH 2.2 in the presence of acetic acid the main charge state is shifted to 20+, although oscillations between 22+ and 20+ can be observed. The maximum charge state is 27+, as in water–HCl. The peak of the free heme is readily detectable only in water–acetate pH 2.2. This effect could be due to the higher hydrophobicity of the solvent under this condition, resulting in increased solubility of free heme at pH values below the pK a of the propionic groups.

In conclusion, the results described in this section (summarized in Table 1) indicate that, in the absence of major conformational changes, the presence of acetic acid affects protein CSD from aqueous solutions only marginally, if at all. This is shown to be true for both folded and unfolded proteins. Minor effects are observed for some of the tested proteins. These effects are much smaller than predicted by the hypothesis that CSD in aqueous acetic acid are limited by the low surface tension of the organic cosolvent. These minor spectral changes can be explained by the charge-reducing effect of acetate ions due to involvement in proton-transfer reactions [40].

Table 1 Summary of the main (Main) and maximum (Max.) charge states observed for either folded (F) or unfolded (U) proteins under different solvent conditions. The maximum charge state of folded cyt c in bimodal spectra could not be determined (nd). The values of z R were calculated by use of the Rayleigh equation, using the radius of the folded protein R m employed by de la Mora [29] and a surface tension of 72.75×10−3 N m−1 for water and water–HCl, or 27.7×10−3 N m−1 for water–acetate
Full size table

Discussion

ESI-MS can be applied to protein conformational studies in two conceptually different ways. It can be used to analyze the mass changes produced by conformation-sensitive reactions or it can be employed to directly monitor conformational changes. In the first case, hydrogen–deuterium exchange, alkylation or other chemical-modification reactions can be used to probe protein conformations in solution, and mass spectrometry represents a powerful tool for analysis of the reaction products. In the second case the conformation-sensitive step is inherent the ESI-MS measurement itself. Protein ionization during the electrospray process provides information about the conformation that the protein held at the moment of transfer to the gas phase. This study is aimed at testing the influence of instrumental and solution conditions relevant to the latter approach. The results indicate that, under appropriately controlled conditions, the observed CSD reflect the species distributions of the original liquid sample. Furthermore, the data show that neither the pH nor the surface tension of the original solution seems to affect protein spectra obtained by nano-ESI-MS. The apparent lack of surface-tension effects is in conflict with the hypothesis that protein CSD are limited by the Rayleigh charge of the precursor ESI droplets.

The experiments on curtain-gas flow rate reported here provide further information on the previously reported effect of this parameter on protein CSD in nano-ESI-MS [16]. The results at pH~7 alone would be difficult to interpret, because a progressive shift of a single bell-shaped distribution could reflect either structural or instrumental effects. Nevertheless, the data at lower pH are clearly indicative of protein unfolding, showing interconversion of distinct peak envelopes and release of the cofactor in Mb. Under these conditions, no shift of the resolved CSD is observed at increasing gas flow rates. Thus, the shift observed at higher pH is most probably due to minor conformational changes. The perturbation of the folding equilibrium in the ESI droplets caused by the instrumental settings could result in global unfolding for samples in which compact conformations are already destabilized by the solvent conditions. Such an effect of curtain gas is most probably due to a higher total impact energy transferred, in average, to the droplets at higher gas flow rates. Unfolding of Mb ions in the gas phase by collision with nitrogen or argon gas has also been described [18]. It should be, nevertheless, mentioned in this context that a cooling effect of the curtain gas has been suggested to operate during electrospray of small ions [41].

A shift of cyt c CSD towards higher charge states in ESI-MS spectra acquired at increasing drying-gas flow rates had been already observed and interpreted in terms of mechanism of production of desolvated protein ions [17]. If the latter were formed by field evaporation from charged droplets, accelerating solvent evaporation by increasing gas flow rates would result in a higher charge density encountered, on average, by the departing ions at the droplet surface. Such an effect would explain the increasing average charge state observed at increasing gas flow rates and would, in turn, support the ion-evaporation mechanism [31] for protein electrospray. Although we cannot completely rule out that different mechanisms dominate at different pH, the results reported here suggest an alternative interpretation for the observed phenomenon. It should also be underscored that the threshold in surface charge density that controls the process according to Iribarne and Thomson’s theory [31] would imply a quite homogeneous population of droplets meeting the conditions for ion-evaporation.

In agreement with previous studies by different groups [1, 2, 9, 20], we observe that internal controls offered by the behavior of protein CSD enable identification of mild experimental conditions appropriate for folding studies. Under such conditions the species distributions observed in ESI-MS are found to reflect those in the original liquid sample. This opens the way to application of this method to quantitative studies. The possibility of temperature control inside the ESI capillary expands the potentiality of this experimental approach [13, 15, 42]. An interesting implication of the agreement between ESI-MS and other spectroscopic methods describing acid-induced unfolding of cyt c is to rule out major pH changes during the ESI process under the employed conditions. The same conclusion is supported by results obtained with other proteins at other pH values [9, 21, 43, 44].

The proteins Ubq and Lyz offer the possibility of uncoupling acidification from unfolding and, therefore, investigating the influence of pH on the ESI process itself. This is shown to be negligible in the pH range 2.2–7, as far as the qualitative features of protein CSD are concerned. This result is consistent with the notion that the charging behavior of analytes during electrospray is governed by their gas-phase basicity [27] and not by the ionization equilibria in solution. The results described here for Ubq at pH 2.2 also show that the nature of the acid employed to adjust the pH of the solution can significantly affect the stability of folded protein structures.

We have recently suggested [45], along the same line of arguments originally exposed by Katta and Chait [2], that one factor determining the different charge states of folded and unfolded proteins in ESI-MS could be the charges of opposite sign relative to the net charge of the macro ion. These charges (i.e. carboxylates in positive-ion mode) would be more effectively stabilized against neutralization by proton transfer reactions in folded than in unfolded conformations, because of engagement in specific intramolecular interactions [46] and stronger, attractive, long-range electrostatic interactions [47, 48, 49, 50, 51, 52]. The persistence of conformational effects at pH 2.2, after titration of all acidic groups of Ubq in solution [53], could be mistaken as evidence against a role of such groups in reducing the charge states of folded forms in ESI-MS relative to unfolded forms. On the contrary, it should be considered that the charge state of ionizable groups is reset according to their gas-phase reactivity during the ESI process. It has been shown that the net charge of a protein in solution is the main factor modulating the pK a values of its acidic residues in the folded state [47]. This effect will be amplified for media of lower dielectric constant. Thus, the gas-phase basicity of the –COO− groups will differ between folded and unfolded molecules even more than their pK a values in solution. Our hypothesis is supported by the observation that Mb variants with different amino acid composition give rise to different CSD in the folded state [54]. Thus, at least in this case, CSD seem to be limited by protein features and not by the availability of charges in the ESI droplets.

The hypothesis suggested by de la Mora [29], based on the Rayleigh-limit charge of ESI droplets, leads to an alternative interpretation of conformational effects in protein ESI mass spectra. According to that hypothesis, protein ionization is limited by the charges carried by the precursor droplet. Unfolded molecules would give rise to higher charge states than folded ones by forcing the droplet into a highly non-spherical shape that could accommodate more charges than predicted by the Rayleigh limit. This model leads to interesting questions that should be the subject of future work:

  • Is an elongated chain a good structural model for unfolded proteins inside ESI droplets?

  • Under which conditions does a non-spherical droplet hold more charges than a spherical one of equal volume?

  • Which mechanisms neutralize the protein during the electrospray before it acquires the surface charges of the final droplet?

Protein conformation in the denatured state is a very complex issue. In general, unfolded proteins in solution are more compact than the ideal random coil [55]. Measurements by nuclear magnetic resonance showed that polypeptide chains can maintain a native-like topology in 8 mol L−1 urea [56, 57]. Extended conformations might be favored for highly charged and poorly shielded molecules. In such circumstances predominance of long-range electrostatic repulsion could force the molecules into a stretched conformation. Nevertheless, in the suggested mechanism, the protein conformation has to affect the shape of the droplet before the last droplet fission, which would be the event determining the different charging of folded and unfolded proteins. At the same time, the protein must be neutral at this stage, if its final charge state corresponds to the number of charges on the surface of the daughter droplet. In this view, extended conformations of unfolded proteins would not be more favored inside ESI droplets than in normal solutions.

A straightforward implication of the model suggested by de la Mora is that the CSD of proteins in ESI-MS should be affected by the surface tension of the solvent according to the Rayleigh equation. This should be true at least for the maximum charge state and at least for folded proteins. In a recent paper [28], Iavarone and Williams conclude that the effects of droplet surface tension extend to the average charge state, and to unfolded proteins. Evidence supporting this conclusion is provided by the increased average charge state observed upon addition of low vapor-pressure compounds to unfolded cyt c in the presence of acetic acid, when the surface tension of the additive is higher than that of acetic acid. In this view, a shift of at least similar extent (average charge state increasing up to 22+) should be observed on comparing the CSD of unfolded cyt c in HCl and acetic acid solutions. As shown in the “Results” section, the prediction for folded proteins would be a ratio of ~1.6 between the maximum charge state in water–HCl and that in water–acetic acid. On the contrary, it is shown that these solvent conditions result in very similar CSD for several proteins in either folded or unfolded conformations. The minor differences observed in the CSD from HCl or acetic acid solutions could be explained by the charge-reducing effect of acetate ions in protein ESI-MS, because of involvement in proton-transfer reactions [40].

The data presented here also show that the spectra of Ubq and Lyz in 60% methanol at pH~7 are almost identical to those in pure water. Although this result cannot be confidently interpreted in terms of surface tension, because of the higher vapor pressure of methanol relative to water, it is still noteworthy that 60% methanol does not affect protein CSD, if unfolding does not occur. On the basis of this result, we can only conclude that changing the surface tension of the droplets from 72.75×10−3 N m−1 to ~33×10−3 N m−1 (value for 60% methanol at 20 °C [58]) in the first stages of the ESI process does not affect the final results. Nonetheless, it is shown here that also a low-surface tension compound with lower vapor pressure than water, for example acetic acid, does not affect the ionization process of either folded or unfolded proteins. In particular, the data show that CSD in the presence of acetic acid are not limited, as previously suggested [28], by the low surface tension of the cosolvent. These results could reflect general lack of dependence of protein CSD on the surface tension of the solution. Alternatively, the effect of acetic acid on the precursor ESI droplets might be lower than generally assumed, because of the peculiar non-equilibrium conditions of ESI-MS. However, recent studies with other low-vapor pressure, low-surface tension compounds further document the lack of the expected surface tension effects on protein spectra [59]. Thus, protein CSD in ESI-MS do not seem to be limited by the number of charges in the precursor droplets; instead they seem to be quite protein-specific.