The mechanisms whereby protein ions are released into the gas phase from charged droplets during electrospray ionization (ESI) continue to be controversial. Several pathways have been proposed. For native ESI the charged residue model (CRM) is favored; it entails the liberation of proteins via solvent evaporation to dryness. Unfolded proteins likely follow the chain ejection model (CEM), which involves the gradual expulsion of stretched-out chains from the droplet. According to the ion evaporation model (IEM) ions undergo electrostatically driven desorption from the droplet surface. The IEM is well supported for small precharged species such as Na+. However, it is unclear whether proteins can show IEM behavior as well. We examined this question using molecular dynamics (MD) simulations, mass spectrometry (MS), and ion mobility spectrometry (IMS) in positive ion mode. Ubiquitin was chosen as the model protein because of its structural stability which allows the protein charge in solution to be controlled via pH adjustment without changing the protein conformation. MD simulations on small ESI droplets (3 nm radius) showed CRM behavior regardless of the protein charge in solution. Surprisingly, many MD runs on larger droplets (5.5 nm radius) culminated in IEM ejection of ubiquitin, as long as the protein carried a sufficiently large positive solution charge. MD simulations predicted that nonspecific salt adducts are less prevalent for IEM-generated protein ions than for CRM products. This prediction was confirmed experimentally. Also, collision cross sections of MD structures were in good agreement with IMS data. Overall, this work reveals that the CRM, CEM, and IEM all represent viable pathways for generating gaseous protein ions during ESI. The IEM is favored for proteins that are tightly folded and highly charged in solution and for droplets in a suitable size regime.

Electrosprayed protein ions can retain native-like conformations. The intramolecular contacts that stabilize these compact gas-phase structures remain poorly understood. Recent work has uncovered abundant salt bridges in electrosprayed proteins. Salt bridges are zwitterionic BH + /A − contacts. The low dielectric constant in the vacuum strengthens electrostatic interactions, suggesting that salt bridges could be a key contributor to the retention of compact protein structures. A problem with this assertion is that H + are mobile, such that H + transfer can convert salt bridges into neutral B 0 /HA 0 contacts. This possible salt bridge annihilation puts into question the role of zwitterionic motifs in the gas phase, and it calls for a detailed analysis of BH + /A − versus B 0 /HA 0 interactions. Here, we investigate this issue using molecular dynamics (MD) simulations and electrospray experiments. MD data for short model peptides revealed that salt bridges with static H + have dissociation energies around 700 kJ mol −1 . The corresponding B 0 /HA 0 contacts are 1 order of magnitude weaker. When considering the effects of mobile H + , BH + /A − bond energies were found to be between these two extremes, confirming that H + migration can significantly weaken salt bridges. Next, we examined the protein ubiquitin under collision-induced unfolding (CIU) conditions. CIU simulations were conducted using three different MD models: (i) Positive-only runs with static H + did not allow for salt bridge formation and produced highly expanded CIU structures. (ii) Zwitterionic runs with static H + resulted in abundant salt bridges, culminating in much more compact CIU structures. (iii) Mobile H + simulations allowed for the dynamic formation/annihilation of salt bridges, generating CIU structures intermediate between scenarios (i) and (ii). Our results uncover that mobile H + limit the stabilizing effects of salt bridges in the gas phase. Failure to consider the effects of mobile H + in MD simulations will result in unrealistic outcomes under CIU conditions.

The transfer of peptide ions from solution into the gas phase by electrospray ionization (ESI) is an integral component of mass spectrometry (MS)-based proteomics. The mechanisms whereby gaseous peptide ions are released from charged ESI nanodroplets remain unclear. This is in contrast to intact protein ESI, which has been the focus of detailed investigations using molecular dynamics (MD) simulations and other methods. Under acidic liquid chromatography/MS conditions, many peptides carry a solution charge of 3+ or 2+. Because of this pre-existing charge and their relatively small size, prevailing views suggest that peptides follow the ion evaporation mechanism (IEM). The IEM entails analyte ejection from ESI droplets, driven by electrostatic repulsion between the analyte and droplet. Surprisingly, recent peptide MD investigations reported a different behavior, that is, the release of peptide ions via droplet evaporation to dryness which represents the hallmark of the charged residue mechanism (CRM). Here, we resolved this conundrum by performing MD simulations on a common model peptide (bradykinin) in Rayleigh-charged aqueous droplets. The primary focus was on pH 2 conditions (bradykinin solution charge = 3+), but we also verified that our MD strategy captured pH-dependent charge state shifts seen in ESI-MS experiments. In agreement with earlier simulations, we found that droplets with initial radii of 1.5–3 nm predominantly release peptide ions via the CRM. In contrast, somewhat larger radii (4–5 nm) favor IEM behavior. It appears that these are the first MD data to unequivocally demonstrate the viability of peptide IEM events. Electrostatic arguments can account for the observed droplet size dependence. In summary, both CRM and IEM can be operative in peptide ESI-MS. The prevalence of one over the other mechanism depends on the droplet size distribution in the ESI plume.

Native electrospray ionization (ESI)-mass spectrometry (MS) is widely used for the detection and characterization of multi-protein complexes. A well-known problem with this approach is the possible occurrence of nonspecific protein clustering in the ESI plume. This effect can distort the results of binding affinity measurements, and it can even generate gas-phase complexes from proteins that are strictly monomeric in bulk solution. By combining experiments and molecular dynamics (MD) simulations, the current work for the first time provides detailed insights into the ESI clustering of proteins. Using ubiquitin as a model system, we demonstrate how the entrapment of more than one protein molecule in an ESI droplet can generate nonspecific clusters (e.g., dimers or trimers) via solvent evaporation to dryness. These events are in line with earlier proposals, according to which protein clustering is associated with the charged residue model (CRM). MD simulations on cytochrome c (which carries a large intrinsic positive charge) confirmed the viability of this CRM avenue. In addition, the cytochrome c data uncovered an alternative mechanism where protein–protein contacts were formed early within ESI droplets, followed by cluster ejection from the droplet surface. This second pathway is consistent with the ion evaporation model (IEM). The observation of these IEM events for large protein clusters is unexpected because the IEM has been thought to be associated primarily with low-molecular-weight analytes. In all cases, our MD simulations produced protein clusters that were stabilized by intermolecular salt bridges. The MD-generated charge states agreed with experiments. Overall, this work reveals that ESI-induced protein clustering does not follow a tightly orchestrated pathway but can proceed along different avenues.