Foreword

Protein mass spectrometry: A personal historical view

In the early part of the 1990s the use of mass spectrometry for protein studies exploded due to the start of the proteomics era. Many young scientists in the field believe that protein mass spectrometry started at that time. This is not true. The first attempts to analyze peptides and proteins date back to the mid-1960s. A landmark publication for me personally was a paper by Mickey Barber and a number of French scientists in which they described the structure determination of a heavily and heterogeneously modified peptidolipid called Fortuitine with a molecular weight of 1,359 using electron impact (EI) mass spectrometry [1]. At that time all mass spectrometric analysis required volatile samples, i.e., samples which could be brought at the gas phase without thermal destruction by heating the sample. Fortuitine was a very fortuitous sample because it was N- and C-terminally blocked and in addition a number of the amide nitrogen atoms were methylated, all making the molecule more volatile. EI-ionization supplied sufficient energy to the molecular ion to cause fragmentation along the peptide backbone. The mass accuracy in the double focusing AEI MS9 electrostatic/magnetic sector instrument was so high that the mass of the molecular ions as well as the fragment ions could be determined with a mass accuracy of a few ppm. A few other groups worldwide were also trying to analyze peptides by mass spectrometry. Among the most prominent was Klaus Biemann at MIT, who combined the need for separation and mass spectrometry by producing derivatives of small peptides which were amenable for GCMS. He also developed computer programs to interpret the data. In a paper from 1966 [2] he has the following statement in a discussion: “In view of the small amount of material required (microgram quantities) and of the extreme speed (1–3 min of computer time) with which this objective and exhaustive interpretation is achieved, this approach shows considerable promise for the routine determination of the amino acid sequence of small peptides obtained upon partial hydrolysis of the oligopeptides resulting from enzymatic cleavage of large polypeptides. It should also be useful in synthetic work since the principle is independent of the end groups and additional protecting groups, the mass of which can be read in with the data”. This almost prophetical statement was really foreseeing the future of mass spectrometry in protein studies.

In the late 1960s and early1970s a number of groups including my own took up peptide mass spectrometry. We were still limited by the need of producing volatile derivatives. Two main lines were followed, the use of permethylated peptides developed by Das et al. [3] and further improved by Thomas et al. [4] using direct inlet probes, and the reduced and trimethyl silylated derivatives introduced by Biemann using separation and introduction of the samples by GCMS. At that time it gradually became clear that post-translational protein modifications were frequent and that mass spectrometry would be an important analytical method to identify and determine the nature and positions of these. My personal breakthrough, which changed the attitude of many of my colleagues from considering the combination of mass spectrometry and protein chemistry as a utopia to be potentially valuable for protein studies, was the discovery of γ-carboxy-glutamic acid residues in the blood clotting factors [5,6]. A number of other modifications such as methylations, acetylations and hydroxylations were also discovered by mass spectrometry in that period. Unfortunately, we could not prepare volatile derivatives of glycosylated and phosphorylated peptides and consequently not see these frequent modifications with our mass spectrometers. Another focus of peptide mass spectrometry was protein sequencing. There we were in competition with the stepwise chemical Edman degradation and the automated peptide sequencer. Due to the need for derivatization when using mass spectrometry and the limited peptide size we could analyze, we could not catch up with the sequencer. Whenever we made a little progress, for example, the ability to analyze peptide mixtures and automatically interpret the resulting complex spectra [7], we were outperformed by improvements of the sequencer. In the 1970s two new ionization methods, chemical ionization (CI) and field desorption (FD), were invented. Both were tested for peptide analysis, e.g. [8,9]. CI resulted in more uniform fragment peak intensity than EI throughout the mass range of the spectra but did not eliminate the need for volatile derivatives. FD was more promising because it allowed analysis of underivatized peptides. However, the spectra were extremely difficult to obtain because the desorption took place in a very short time window and with the scanning instruments used at that time, it was real luck to be at the right position of the scan when the peptide ion was generated. Consequently none of these new ionization techniques resulted in a breakthrough for protein mass spectrometry. In the 1970s mass spectrometry was recognized all over as a tool for discovery of certain types of protein modifications and for sequencing of N-terminally blocked peptides which were not amenable for Edman degradation. However, most of the time it was a hard uphill walk and only few scientists believed in us.

In the beginning of the 1980s the situation changed to excitement. The main reason was the developments in the nucleic acid field resulting in cDNA sequencing making de novo sequencing on the protein level partially redundant, and the advent of new ionization methods that allowed mass spectrometric analysis of underivatized peptides and intact proteins. Already in 1974 a new ionization method, plasma desorption mass spectrometry (PDMS), based on desorption of the sample directly from the solid state by bombardment with high energy (100 MeV) ions was described by McFarlanes group at Texas A&M [10]. The method remained largely unrecognized by protein mass spectrometrists until Bo Sundqvist and collaborators at the Uppsala University demonstrated that it was possible to obtain mass spectra of intact proteins using PDMS, first insulin [11] and later for larger proteins [12,13]. They also developed the first commercially available plasma desorption mass spectrometer through the spin-off company BioIon. It was a fully automated instrument based on the almost forgotten time-of-flight principle. I was lucky to collaborate with the group in Uppsala and to get the first of these new instruments in my laboratory and thereby to participate in this exciting new development. Independently Mickey Barber (the one who analyzed Fortuitine in 1964) published another method, fast atom bombardment (FAB) [14], that allowed desorption of underivatized peptides from a glycerol matrix. The method was quickly demonstrated also to allow mass spectrometry of insulin [15,16] and some larger intact proteins. FAB could readily be installed on existing sector field mass spectrometers (including our own) and therefore quickly became available in many laboratories, whereas acceptance of PDMS was slower because it required new instrumentation and also, due to the 252Cf source, extensive safety precautions. In the coming period most of the principles now used in proteomics were established. Thus, characterization of proteins by peptide mass mapping using FAB-MS was suggested by Howard Morris [17] and soon also taken up by PDMS, the later allowing digestion of the proteins after recording their mass spectra directly on the nitrocellulose-covered target. Thus a combined top-down and bottom-up strategy could be performed on a single sample preparation. Although FAB and PD being soft ionization methods, sufficient energy was supplied to the formed ions to cause some fragmentation. This was investigated for both methods and the now widely used nomenclature for mass spectrometric peptide fragmentation suggested [18]. To generate more fragment ions collision induced dissociation (CID) was introduced in mass spectrometric peptide analysis using either triple quadrupole or the giant four sector instruments. The latter were extensively used by Klaus Biemann, e.g. [19]. They allowed high-resolution mass spectra to be recorded from parent as well as fragment ions and he succeeded in de novo sequencing of several small- to medium-sized proteins using such instruments. The now key element in proteomics, i.e., identification of proteins and peptides based on comparison of their mass with cDNA sequence information was also initiated in that period [20,21].

The scientists who in the 1980s investigated the possibility for mass spectrometry of involatile molecules were a rather small group dominated by physicists who tried to understand the desorption phenomenon but included a few chemists and biologists who wanted to apply the techniques. The group, including among many others Franz Hillenkamp, Ken Standing, Michael Karas, Brian Chait and Marvin Vestal, met at regular symposia, first the Ion Formation of Organic Solids (IFOS) meetings organized by A. Benninghoven and later the Desorption meetings. The atmosphere was enthusiastic with open exchange of information. By that time it became clear to us that the FAB and PD were limited in terms of the molecular size of the proteins to approximately 25 kDa and 35 kDa, respectively. Although the methods...