Heat capacities have been measured by ion calorimetry for size-selected aluminum cluster cations ranging in size from 150 to 342 atoms. All clusters show a sharp peak in their heat capacity versus temperature plots which is attributed to the melting transition. The large size dependent fluctuations in the melting temperatures found for smaller clusters in previous work have largely vanished. The melting temperatures for the 150–342 atom size range examined here are substantially below the bulk value and increase relatively smoothly with size. Above 180 atoms, they closely follow the 1/r dependence predicted by thermodynamic models. A notable exception occurs between Al283+ and Al284+, where the melting temperature suddenly jumps by 13 K. This jump represents a remarkably sudden change for such large clusters. The origin is probably structural, but the nature of the structural change remains unknown. The latent heat is still far from its bulk value (only 42%) even for the largest clusters studied here. When the data for aluminum clusters and particles are combined, melting temperatures are available over a size range spanning 6 decades, providing a complete picture of how the melting temperature changes with size.

The assembly of hundreds of identical proteins into an icosahedral virus capsid is a remarkable feat of molecular engineering. How this occurs is poorly understood. Key intermediates have been anticipated at the end of the assembly reaction, but it has not been possible to detect them. In this work we have used charge detection mass spectrometry to identify trapped intermediates from late in the assembly of the hepatitis B virus T = 4 capsid, a complex of 120 protein dimers. Prominent intermediates are found with 104/105, 110/111, and 117/118 dimers. Cryo-EM observations indicate the intermediates are incomplete capsids and, hence, on the assembly pathway. On the basis of their stability and kinetic accessibility we have proposed plausible structures. The prominent trapped intermediate with 104 dimers is attributed to an icosahedron missing two neighboring facets, the 111-dimer species is assigned to an icosahedron missing a single facet, and the intermediate with 117 dimers is assigned to a capsid missing a ring of three dimers in the center of a facet.

Recombinant adeno-associated viruses (AAVs) are promising vectors for human gene therapy. However, current methods for evaluating AAV particle populations and vector purity are inefficient and low resolution. Here, we show that charge detection mass spectrometry (CDMS) can resolve capsids that contain the entire vector genome from those that contain partial genomes and from empty capsids. Measurements were performed for both single-stranded and self-complementary genomes. The self-complementary AAV vector preparation appears to contain particles with partially truncated genomes averaging at half the genome length. Comparison to results from electron microscopy with manual particle counting shows that CDMS has no significant mass discrimination in the relevant mass range (after a correction for the ion velocity is taken into account). Empty AAV capsids are intrinsically heterogeneous, and capsids from different sources have slightly different masses. However, the average masses of both the empty and full capsids are in close agreement with expected values. Mass differences between the empty and full capsids for both single-stranded and self-complementary AAV vectors indicate that the genomes are largely packaged without counterions.

Charge detection mass spectrometry (CDMS) is a single-molecule technique particularly well-suited to measuring the mass and charge distributions of heterogeneous, MDa-sized ions. In this work, CDMS has been used to analyze the assembly products of two coat protein variants of bacteriophage P22. The assembly products show broad mass distributions extending from 5 to 15 MDa for A285Y and 5 to 25 MDa for A285T coat protein variants. Because the charge of large ions generated by electrospray ionization depends on their size, the charge can be used to distinguish hollow shells from more compact structures. A285T was found to form T = 4 and T = 7 procapsids, and A285Y makes a small number of T = 3 and T = 4 procapsids. Owing to the decreased stability of the A285Y and A285T particles, chemical cross-linking was required to stabilize them for electrospray CDMS.

Charge detection mass spectrometry (CDMS) is a single-particle technique where the masses of individual ions are determined from simultaneous measurement of each ion’s mass-to-charge ratio (m/z) and charge. CDMS has many desirable features: it has no upper mass limit, no mass discrimination, and it can analyze complex mixtures. However, the charge is measured directly, and the poor accuracy of the charge measurement has severely limited the mass resolution achievable with CDMS. Since the charge is quantized, it needs to be measured with sufficient accuracy to assign each ion to its correct charge state. This goal has now been largely achieved. By reducing the pressure to extend the trapping time and by implementing a novel analysis method that improves the signal-to-noise ratio and compensates for imperfections in the charge measurement, the uncertainty has been reduced to less than 0.20 e rmsd (root-mean-square deviation). With this unprecedented precision peaks due to different charge states are resolved in the charge spectrum. Further improvement can be achieved by quantizing the charge (rounding the measured charge to the nearest integer) and culling ions with measured charges midway between the integral values. After ions with charges more than one standard deviation from the mean are culled, the fraction of ions assigned to the wrong charge state is estimated to be 6.4 × 10–5 (i.e., less than 1 in 15 000). Since almost all remaining ions are assigned to their correct charge state, the uncertainty in the mass is now almost entirely limited by the uncertainty in the m/z measurement.

Charge detection mass spectrometry (CDMS) provides a direct measure of the mass of individual ions through nondestructive, simultaneous measurements of the mass to charge ratio and the charge. To improve the accuracy of the charge measurement, ions are trapped and recirculated through the charge detector. By substantially extending the trapping time, the uncertainty in the charge determination has been reduced by a factor of two, from 1.3 elementary charges (e) to 0.65 e. The limit of detection (the smallest charge that can be reliably measured) has been reduced by about the same proportion, from 13 to 7 e. The more precise charge measurements enable a substantial improvement in the mass resolution, which is critical for applications of CDMS to mixtures of high mass ions.

•A simple electrospray interface consisting of a long drift region followed by a DC ion carpet is described.•The performance is optimized by trajectory calculations and experiments.•Trajectory calculations indicate >90% transmission efficiency for thermalized ions.•m/z discrimination at the DC ion carpet is minimized.We describe a simple electrospray interface incorporating a long drift region and a DC ion carpet. The function of the drift region is to thermalize and desolvate the ions. It has a series of ring electrodes supplied with RF and DC voltages, which radially confine and transport ions towards the DC ion carpet. The DC ion carpet consists of a series of concentric rings on a planar printed circuit board, which provide a DC potential gradient that funnels ions through a central aperture. The design and operating parameters were optimized by trajectory calculations and experiments. According to calculations, the transmission efficiency for thermalized ions exceeds 90%, and there is little m/z discrimination at the exit aperture.

Heat capacities have been measured as a function of temperature for size-selected gallium cluster cations with between 60 and 183 atoms. Almost all clusters studied show a single peak in the heat capacity that is attributed to a melting transition. The peaks can be fit by a two-state model incorporating only fully solid-like and fully liquid-like species, and hence no partially melted intermediates. The exceptions are Ga90+, which does not show a peak, and Ga80+ and Ga81+, which show two peaks. For the clusters with two peaks, the lower temperature peak is attributed to a structural transition. The melting temperatures for clusters with less than 50 atoms have previously been shown to be hundreds of degrees above the bulk melting point. For clusters with more than 60 atoms the melting temperatures decrease, approaching the bulk value (303 K) at around 95 atoms, and then show several small upward excursions with increasing cluster size. A plot of the latent heat against the entropy change for melting reveals two groups of clusters: the latent heats and entropy changes for clusters with less than 94 atoms are distinct from those for clusters with more than 93 atoms. This observation suggests that a significant change in the nature of the bonding or the structure of the clusters occurs at 93–94 atoms. Even though the melting temperatures are close to the bulk value for the larger clusters studied here, the latent heats and entropies of melting are still far from the bulk values.

Pyruvate kinase multimers have been investigated by charge detection mass spectrometry (CDMS). In CDMS, the m/z and z are simultaneously measured for each ion, so the mass is determined directly. The measurements were made using a modified cone trap that incorporates an image charge detector with a cryogenically cooled preamplifier. With non-denaturing solution conditions, the tetrameric form of pyruvate kinase is observed along with aggregates of the tetramer. The time-of-flight m/z spectrum shows octamers and dodecamers. However, the lack of charge state resolution prevents identification of larger multimers. Multimers up to the 40-mer are revealed by CDMS. Their intensities fall-off exponentially with size. Evidence supporting a non-specific, solution-based aggregation mechanism is presented. The relationship between the m/z and mass of the multimers is consistent with the predictions of the charge residue model. Pyruvate kinase ions are held in the cone trap for up to 129 ms. With this long trapping time the root mean square deviation in the charge determination is reduced to 1.3 elementary charges.Graphical abstractHighlights► In charge detection mass spectrometry m/z and z are simultaneously measured to reveal mass of each ion directly. ► Multimers up to 40-mer are observed for pyruvate kinase. ► Results support non-specific solution-based aggregation. ► m/z values consistent with charge residue model. ► RMS deviation of charge determination reduced to 1.3 e.

A charge detection mass spectrometer (CDMS) with a limit of detection of 30 elementary charges (e) for a single ion is described. The new CDMS consists of an electrospray source coupled to a dual hemispherical deflection analyzer (HDA) followed by a modified cone trap incorporating an image charge detector. Ions are energy selected by the dual HDA prior to entering the trap. The fundamental oscillation frequency of the trapped ion is extracted by a fast Fourier transform (FFT). The oscillation frequency and kinetic energy provide the m/z. The magnitude of the FFT at the fundamental frequency is proportional to the charge. Simulations indicate that the charge is measured with an average uncertainty of 3.2 e. The mass of each ion is obtained from the m/z and the charge. Mass distributions have been measured for bovine serum albumin (BSA). The BSA ions were trapped for up to 1139 cycles. BSA monomer and multimer ions are evident in the measured mass distribution. The width of the monomer mass distribution (14 kDa) is consistent with the predicted uncertainty in the charge.Graphical abstractHighlights► A new charge detection mass spectrometer is described. ► Ions are energy selected prior to entering a cone trap with an image charge detector. ► A limit of detection of 30 elementary charges is achieved. ► The average uncertainty in the charge measurements is 3.2 elementary charges. ► Mass distributions measured for BSA (66.4 kDa) show monomer and multimer ions.

The reactions of CO2 on the Al100+ cluster have been investigated as a function of cluster temperature (300–1100 K) and relative kinetic energy (0.2–10 eV). Two main products are observed at low cluster temperature: Al100O+ (which is believed to result from a stripping reaction) and Al100CO2+ from complex formation. As the cluster temperature is raised, both products dissociate by loss of Al2O. Al100O+ forms Al98+, while Al100CO2+ forms Al98CO+ and Al96C+. In both cases, loss of Al2O turns-on above the melting temperature of Al100+. This presumably occurs because the overall reaction leading to the loss of Al2O is significantly less endothermic for the liquid cluster than for the solid.

The reactions of benzene on Al100+ have been investigated as a function of cluster temperature (300–1100 K) and relative kinetic energy (1–14 eV) by low-energy ion-beam methods and mass spectrometry. Benzene chemisorbs on both solid and liquid aluminum clusters to generate Al100C6D6+. A series of Al100–n+ (n = 1, 2, 3, ...) products was also observed. As the cluster temperature was raised above the melting temperature of Al100+, the Al100C6D6+ product dehydrogenates to form Al100C6D4+, Al100C6D2+, and Al100C6+. The degree of dehydrogenation was measured as a function of temperature. Very little Al100C6D2+ was observed, suggesting that the losses of the second and third D2 molecules are coordinated.

A novel image charge detection mass spectrometer (CDMS) with improved sensitivity and mass accuracy is described. The improved detector design and method of data analysis allow us to measure a reliable mass for a single macroion that is an order of magnitude smaller than previously achieved with CDMS. The apparatus employs an image charge detector array consisting of 22 detectors. The detectors are divided into two groups that can be floated at different potentials. The signals from the detector array are analyzed using a correlation approach to yield the velocities in the two groups of detectors and the charge. These quantities, together with the voltage difference between the two groups of detectors, provide a value for the mass. The mass, m/z, and charge distributions recorded for 300 kDa poly(ethylene oxide) (PEG) are presented. The mass distribution shows a peak at around 300 kDa with a width close to that expected from the polymer size distribution. In addition, there are broad peaks in the mass distribution at around 100 and 500 MDa. The 300 kDa ions have m/z ratios of ∼2 kDa/e, and the 100 and 500 MDa ions have m/z ratios of ∼40 kDa/e. The 100 and 500 MDa ions probably result from PEG aggregates that are either present in solution or the residue of large electrospray droplets.

Film droplets formed from the bursting of 2.4 mm diameter bubbles on the surface of pure water are predominantly negatively charged. The charge generated per bubble varies chaotically; a few bubbles generate more than −3 × 106 elementary charges (e) but the vast majority generate much less. The average is −5 × 104e/bubble, and it is not significantly affected by bubbling rate or temperature. The charge diminishes with increasing salt concentration and vanishes for concentrations above 10−3 M. We propose a mechanism consistent with the observed charge separation. The model relies on the assumption that the surface of pure water has a slight excess of hydroxide ions. The charge separation results when water with entrained counterions (H3O+) flows out of the thinning film of the bubble cap, leaving behind the excess OH− on the surface. Addition of salt reduces the Debye length, and the charge separation mechanism becomes less effective as the Debye length becomes small compared with the film thickness. The excess charge near the surface of pure water is very small, around −4 nC/m2.

Cross sections for chemisorption of N2 onto Al44+/− cluster ions have been measured as a function of relative kinetic energy and the temperature of the metal cluster. There is a kinetic energy threshold for chemisorption, indicating that it is an activated process. The threshold energies are around 3.5 eV when the clusters are in their solid phase and drop to around 2.5 eV when the clusters melt, indicating that the liquid clusters are much more reactive than the solid. Below the melting temperature the threshold for Al44− is smaller than for Al44+, but for the liquid clusters the anion and cation have similar thresholds. At high cluster temperatures and high collision energies the Al44N2+/− chemisorption product dissociates through several channels, including loss of Al, N2, and Al3N. Density functional calculations are employed to understand the thermodynamics and the dynamics of the reaction. The theoretical results suggest that the lowest energy pathway for activation of dinitrogen is not dynamically accessible under the experimental conditions, so that an explicit account of dynamical effects, via molecular dynamics simulations, is necessary in order to interpret the experimental measurements. The calculations reproduce all of the main features of the experimental results, including the kinetic energy thresholds of the anion and cation and the dissociation energies of the liquid Al44N2+/− product. The strong increase in reactivity on melting appears to be due to the volume change of melting and to atomic disorder.

Water droplets are generated by sonic spray, transferred into vacuum through a capillary interface, and then passed through two image charge detectors separated by a drift region. The image charge detectors measure the charge and velocity of each droplet. For around 1% of the droplets, the charge changes significantly between the detectors. In some cases it increases, in others it decreases, and for some droplets the charge changes polarity. We attribute the charge changing behavior to fragmentation caused by freezing. Simulations indicate that the time required for a droplet to cool and freeze in vacuum depends on its size, and that droplets with radii of 1–2 μm have the right size to freeze between the two detectors. These sizes correspond to the smaller end of the distribution present in the experiment. When the charge on a droplet increases or changes polarity, fragmentation must be accompanied by charge separation where fragments carry away opposite charges. In some cases, two fission fragments were observed in the second charge detector. We show examples where the droplet breaks apart to give fragments of the same charge and opposite charges. The fragmentation and charge changing behavior found here is consistent with what has been found in the freezing of larger suspended and supported droplets.Studies with two image charge detectors separated by a drift region show that micron sized water droplets fragment and change their charge while traveling through vacuum.

Charged water droplets generated by electrospray, sonic spray, and a vibrating orifice aerosol generator (VOAG) have been studied by digital macrophotography and image charge detection mass spectrometry. Image charge detection mass spectrometry provides information on the droplet size, charge, and velocity after transmission through a capillary interface. The digital images provide the droplet size distribution before they enter the capillary. Droplets with 10−100 μm radii generated by sonic spray and VOAG are reduced to 2−3 μm radii by transmission through the capillary interface. The droplets from sonic spray and VOAG are much more highly charged than expected for random charging, and positive droplets are much more prevalent than negative. For positive mode electrospray, >99% of the detected droplets carry a positive charge, whereas for negative mode electrospray, <30% of the detected droplets carry a negative charge (i.e., >70% carry a positive charge). These observation can all be accounted for by the aerodynamic breakup of the droplets in the capillary interface. This breakup reduces the droplets to a terminal size at which point further breakup does not occur. Charge separation during droplet breakup is responsible for the relatively high charges on the sonic spray and VOAG droplets and for the preference for positively charged droplets. The charge separation can be explained using the bag mechanism for droplet breakup and the electrical bilayer at the surface of water.

Calorimetry measurements have been used to probe the melting of aluminum cluster cations with 63 to 83 atoms. Heat capacities were determined as a function of temperature (from 150 to 1050 K) for size-selected cluster ions using an approach based on multicollision-induced dissociation. The experimental method is described in detail and the assumptions are critically evaluated. Most of the aluminum clusters in the size range examined here show a distinct peak in their heat capacities that is attributed to a melting transition (the peak is due to the latent heat). The melting temperatures are below the bulk melting point and show enormous fluctuations as a function of cluster size. Some clusters (for example, n=64, 68, and 69) do not show peaks in their heat capacities. This behavior is probably due to the clusters having a disordered solid-like phase, so that melting occurs without a latent heat.

Ion mobility measurements and molecular dynamic simulations have been performed for a series of peptides designed to have helix-turn-helix motifs. For peptides with two helical sections linked by a short loop region: AcA14KG3A14K+2H+, AcA14KG5A14K+2H+, AcA14KG7A14K+2H+, and AcA14KSar3A14K+2H+ (Ac = acetyl, A = alanine, G = glycine, Sar = sarcosine and K = lysine); a coiled-coil geometry with two anti-parallel helices is the lowest energy conformation. The helices uncouple and the coiled-coil unfolds as the temperature is raised. Equilibrium constants determined as a function of temperature yield enthalpy and entropy changes for the unfolding of the coiled-coil. The enthalpy and entropy changes depend on the length and nature of the loop region. For a peptide with three helical sections: protonated AcA14KG3A14KG3A14K; a coiled-coil bundle with three helices side-by-side is substantially less stable than a geometry with two helices in an antiparallel coiled-coil and the third helix collinear with one of the other two.

Ion mobility measurements and molecular dynamic simulations have been performed for a series of peptides designed to have helix-turn-helix motifs. For peptides with two helical sections linked by a short loop region: AcA14KG3A14K+2H+, AcA14KG5A14K+2H+, AcA14KG7A14K+2H+, and AcA14KSar3A14K+2H+ (Ac = acetyl, A = alanine, G = glycine, Sar = sarcosine and K = lysine); a coiled-coil geometry with two anti-parallel helices is the lowest energy conformation. The helices uncouple and the coiled-coil unfolds as the temperature is raised. Equilibrium constants determined as a function of temperature yield enthalpy and entropy changes for the unfolding of the coiled-coil. The enthalpy and entropy changes depend on the length and nature of the loop region. For a peptide with three helical sections: protonated AcA14KG3A14KG3A14K; a coiled-coil bundle with three helices side-by-side is substantially less stable than a geometry with two helices in an antiparallel coiled-coil and the third helix collinear with one of the other two.

Calorimetry measurements have been used to probe the melting of aluminum cluster cations with 63 to 83 atoms. Heat capacities were determined as a function of temperature (from 150 to 1050 K) for size-selected cluster ions using an approach based on multicollision-induced dissociation. The experimental method is described in detail and the assumptions are critically evaluated. Most of the aluminum clusters in the size range examined here show a distinct peak in their heat capacities that is attributed to a melting transition (the peak is due to the latent heat). The melting temperatures are below the bulk melting point and show enormous fluctuations as a function of cluster size. Some clusters (for example, n = 64, 68, and 69) do not show peaks in their heat capacities. This behavior is probably due to the clusters having a disordered solid-like phase, so that melting occurs without a latent heat.

•Mass spectrometry characterizes non-icosahedral assembly products of WHV.•A 150-dimer capsid is the most abundant non-icosahedral polymorph.•Cryo-electron microscopy provides evidence for ellipsoidal and spiral-like morphologies.•Most features cannot be explained by existing models for hexameric defects.Woodchuck hepatitis virus (WHV) is prone to aberrant assembly in vitro and can form a broad distribution of oversized particles. Characterizing aberrant assembly products is challenging because they are both large and heterogeneous. In this work, charge detection mass spectrometry (CDMS) is used to measure the distribution of WHV assembly products. CDMS is a single-particle technique where the masses of individual ions are determined from simultaneous measurement of each ion's charge and m/z (mass-to-charge) ratio. Under relatively aggressive, assembly promoting conditions, roughly half of the WHV assembly products are T = 4 capsids composed of exactly 120 dimers while the other half are a broad distribution of larger species that extends to beyond 210 dimers. There are prominent peaks at around 132 dimers and at 150 dimers. In part, the 150 dimer complex can be attributed to elongating a T = 4 capsid along its 5-fold axis by adding a ring of hexamers. However, most of the other features cannot be explained by existing models for hexameric defects. Cryo-electron microscopy provides evidence of elongated capsids. However, image analysis reveals that many of them are not closed but have “spiral-like” morphologies. The CDMS data indicate that oversized capsids have a preference for growth by addition of 3 or 4 dimers, probably by completion of hexameric vertices.Download high-res image (282KB)Download full-size image

The assembly of hundreds of identical proteins into an icosahedral virus capsid is a remarkable feat of molecular engineering. How this occurs is poorly understood. Key intermediates have been anticipated at the end of the assembly reaction, but it has not been possible to detect them. In this work we have used charge detection mass spectrometry to identify trapped intermediates from late in the assembly of the hepatitis B virus T = 4 capsid, a complex of 120 protein dimers. Prominent intermediates are found with 104/105, 110/111, and 117/118 dimers. Cryo-EM observations indicate the intermediates are incomplete capsids and, hence, on the assembly pathway. On the basis of their stability and kinetic accessibility we have proposed plausible structures. The prominent trapped intermediate with 104 dimers is attributed to an icosahedron missing two neighboring facets, the 111-dimer species is assigned to an icosahedron missing a single facet, and the intermediate with 117 dimers is assigned to a capsid missing a ring of three dimers in the center of a facet.