Software using maximum entropy (MaxEnt) analysis has been developed, and used to deconvolute complete electrospray spectra of protein mixtures. It automatically produces zero-charge mass spectra on a molecular mass scale, along with probabilistic quantification so that the reliability of features in the spectrum can be ascertained. Because maximum entropy is faithful to the experimental data, the results tend to have improved resolution and signal-to-noise ratio. This improved performance, particularly regarding resolution, is demonstrated on a haemoglobin containing two &globins separated by 12 Da at mlz 15 867 (0.08°/0). A separation of 12 Da was previously the closest at which mass measurement of two globins was practicable. Also, two hitherto unresolved P-globins from a second haemoglobin, separated by 9 Da (0.06%) were resolved by MaxEnt and their masses accurately measured. These are the first results using rigorous MaxEnt analysis in electrospray mass spectrometry.In the form initially produced by the mass spectrometer, the electrospray spectra of protein mixtures are complex, each protein in the mixture being represented by a series of multiply charged ions on a madcharge ratio scale.Generally these ions occur with mass-to-charge ratio ( M + z H ) l z , where M is the molecular mass of the protein, H' is the mass of the proton (1.00794 u) and z is the number of charges on an ion, a series of consecutive integers. A 15 kDa protein typically produces about 10 peaks with a range of z from 10-20. Larger proteins, e.g., albumins (66kDa) often have 20 or more peaks in the series.To aid interpretation, some means is required for generating a zero-charge mass spectrum, whereby each component in the mixture is transformed from the multiply charged ion series on a masslcharge ratio scale to a single peak on a molecular mass scale. Early methods] tended to produce artefacts and a baseline which increased with mass. A more recent approach2 has achieved an artefact-free zero-charge spectrum with an improved signal-to-noise ratio, but requires prior identification of the charge states in the multiply charged ion series. Current software allows semiautomatic identification of the charge states, but nevertheless a degree of operator intervention is usually required. Moreover, since the data in the original multiply charged spectrum are directly transformed onto a true molecular mass scale, components which are unresolved in the original data remain unresolved after transformation to the zero-charge spectrum. Yet the signals are known to be broadened, by both the isotopic distribution of the elements in the molecule (isotopic broadening) and the mass spectrometer. Hence the true underlying spectrum of masses will be sharper than the peaks in the original data. By incorAuthor to whom correspondence should be addressed. porating this broadening into the program, MaxEnt is able to deconvolute it from the data, thus enhancing the resolution which can be observed in the final MaxEnt mass spectrum. Peak heights grow proportiona...

A novel four- channel multiplexed electrospray liquid chromatography interface is described. This device has been used to analyse both single components and mixtures by liquid chromatography/mass spectrometry (LC/MS) as well as synthetic samples prepared by automated procedures. These data provided unambiguous molecular weight assignments to both major components and synthetic by-products in these samples. In this work particular attention has also been paid to the elimination of interchannel crosstalk. Copyright 1999 John Wiley & Sons, Ltd.

The maximum entropy (MaxEnt) technique has been well established in image processing for the last decade. In more recent times MaxEnt has been applied to spectroscopic techniques and shows considerable promise. As part of a much wider evaluation of the Cambridge software, specifically MemSys3, the MaxEnt technique has been successfully applied to electrospray mass spectra.

Mass measurement by electrospray mass spectrometry (ESMS) is used as a rapid preliminary verification of the identity of various recombinant proteins ranging from 7 to 44 kDa with an accuracy of 0.01-0.03%. ESMS not only improves the speed but also the reliability of the protein structure determination when used in conjunction with other methods of protein analysis. Modifications of these large molecules, for example the loss of C-terminal amino acids, N-terminal acetylation, 2-mercaptoethanol addition to a cysteine, and trace formation of a covalent dimer (3%), are easily detected individually or in mixtures by mass measurement using ESMS; feats which would be very difficult to achieve using classical biochemical methods. As little as 1% of several structurally related protein contaminants have been identified in a 15 kDa recombinant protein preparation.

No abstract