Stephen Martin, Perseptive Biosystems, "Matrix Assisted Laser Desorption Ionization Time-of-Flight Mass Spectrometry: A Technique for the Masses". The general operational concepts of the LDMS method were reviewed, emphasizing results obtained from linear analysis, from reflectron analysis, and by using post source decay for detailed structural information. Three matrices are commonly used for peptide and protein analysis: 3,5-dimethoxy-4-hydroxy-cinnamic acid (sinapinic acid), a-cyano-4-hydroxy-cinnamic acid (ACHA), and 2,5-dihydroxy-benzoic acid (DHB). Samples may produce markedly different spectra when different matrices are used, e.g., the singly protonated ion is the prominent species when BSA is analyzed in sinapinic acid, but multiply protonated ions are more prominent when it is analyzed in ACHA. The polarity of the accelerating voltage (ion detection mode) also has an effect: when a glycoprotein was digested and analyzed as a peptide mixture, the three expected glycosylated peptides were seen only as negative ions (2). Fourteen of the fifteen peptides in the digest were seen, but this required detection of positive and negative ions and analysis in three different matrices. All six spectra showed a different peptide subset. Only one peptide could be seen in all six spectra, and five peptides were seen in only one of the six spectra.
Peak broadening during linear analysis limits the ability to resolve components with similar mass. In linear analysis, resolution-defined as mass divided by peak width at half-height-is typically less than 400. Broadening is caused partly by ion decomposition during flight, so that the parent and daughter ions arrive at the detector at almost, but not exactly, the same time. In reflectron analysis, an electrostatic mirror directs ions to a second detector, and during this process, parent and daughter ions are separated, providing resolution values up to about 2,000. Under favorable conditions, the isotopic species of bovine insulin (5,734 Da) can be partially resolved.
Reflectron analysis may also provide detailed structural information about samples if they undergo post source decay, i.e., decomposition after ionization. When properly adjusted, the reflectron allows measurement of daughter ion masses. Peptides fragment by the same mechanism in LDMS as they do in other mass analysis methods and, in fact, produce daughter ion spectra similar to those seen with tandem mass spectrometers using collision induced dissociation (3). From the fragment masses, the sequences of known and unknown peptides can be determined on much simpler instrumentation than used previously. Also, post-translational modifications often represent the most labile bonds in peptides, and they may be readily detected by post source decay, as already demonstrated by LDMS analysis of phosphorylated, sulfated, and glycosylated peptides.
Kathy Stone, Yale University School of Medicine, "LDMS as a Complement to Internal Protein Sequencing". Kathy Stone showed how peptide fractions can be screened by LDMS to improve the productivity of protein sequencing projects. Some peptides present in a mixture are not seen in the mixture's spectrum, perhaps because peptides do not ionize equally well, and so a study to define the peptide features that influence ionization was also described. In one study, LDMS was used to screen microbore HPLC fractions of peptides made by digesting Coomassie Blue-stained proteins in SDS-PAGE gel slices. Ninety-seven peptides from 64 submitted protein samples were included in this study. The first criterion used to select HPLC fractions for sequencing was absorbance peak symmetry; in an earlier study (4), this was shown to be a rather poor indicator of peptide purity-only about 65% of the samples meeting this criterion provided unambiguous, "useful" sequence data. Symmetrical peaks were then mass analyzed to provide a mass measurement before sequencing and to evaluate peak purity. Only 3% of the fraction was used for mass analysis, corresponding to an average of 700 fmol of peptide. The average mass accuracy was 0.25% (2.5 Da per 1,000) using external calibration standards. When these fractions contained more than one component by mass analysis with major-to-minor spectrum peak ratios of 2 to 10, about 65% of the samples provided "useful" sequence data, no better than when using only HPLC peak symmetry as the distinguishing criterion for suitability. However, when major-to-minor peak ratios were greater than 10, the success rate was about 90%. Furthermore, mass analysis identified digestion artifacts prior to sequencing, such as protease autolysis peptides and Coomassie Blue by-products. As expected, prior mass analysis increased the confidence of sequence assignments, occasionally allowed tentative assignments to be made with more certainty, and helped identify post-translational modifications. In short, incorporating LDMS increased the overall productivity of internal protein sequencing at the expense of only a negligible portion of each sample, even for samples present at low levels.
To evaluate peptide properties that influence ionization, a group of thirteen synthetic peptides was examined. The net charge of the peptides at pH 2 (the pH of the matrix solution) varied according to the number of basic residues and the state of the peptide termini, blocked or free. Each peptide was mass analyzed separately as serial dilutions to determine the lowest amount needed to provide a signal-to-noise ratio of at least 5. The limit of detection spanned a range of 10 fmol to 20 pmol and, in general, decreased as the charge at pH 2 increased. Two peptides-one with amino and carboxamide groups at the termini and no basic residues, the other with acetyl and carboxamide groups at the termini and no positively charged sidechains-did not provide spectra, even from 50 pmol of peptide. Although it is significant in other mass analysis methods, such as FAB (5), hydrophobicity appears not to be a prominent factor in LDMS ionization. Variations in peptide ionization are consistent with the simple spectra often produced from complex peptide mixtures, such as those resulting from protease digestion.
William J. Henzel, Genentech, Inc., "Identification of Proteins by Mass Searching of Peptide Fragments in a Protein Sequence Database". Proteins are normally identified in one- or two-dimensional gels on the basis of molecular weight and isoelectric point. Bill Henzel described an LDMS application that identifies proteins from gels with more certainty on the basis of their peptide mass fingerprints (6). In this application, proteins are separated by electrophoresis, transferred to PVDF membrane, then digested with a protease. Peptides are extracted from the membrane and mass analyzed as a mixture by LDMS. The observed masses are then searched against an entire protein sequence database that simulates the digestion conditions applied to the protein sample, e.g., Lys C digestion following carboxymethylation. The computer program permitting this search, Fragfit, allows for user-specified deviations in observed and expected masses and does not require that all the observed masses fit those predicted from one of the database sequences. All peptides expected from a digestion are not seen in the sample's spectrum, but unambiguous assignments can still be made. For example, E. coli uridine phosphorylase was identified from a two-dimensional gel of total E. coli proteins on the basis of only three peptides seen during LDMS, among 120,000 candidate proteins in the database. The Fragfit code has been refined to account for artifacts known to occur during digestion, e.g., a partial digest option can be used when needed. This application makes good use of the high sensitivity of LDMS. Twelve E. coli proteins from two-dimensional gels present at levels of 0.5 to 12 pmol were used to validate the method through both LDMS and protein sequence analysis.
This application provides a rapid alternative to Edman sequencing for identifying known proteins from gels. This is especially attractive for eukaryotic proteins, which often have blocked amino-termini. An example was given where a blocked protein from the serum of leukemia patients was identified as human prohibitin on the basis of six observed peptide masses.
Michael C. Fitzgerald, University of Wisconsin at Madison, "The Development of Matrix-Assisted Laser Desorption/Ionization Mass Spectrometry for DNA Sequencing". As the human genome project has developed, the need for methods that can analyze sequencing reaction products more rapidly than slab gel electrophoresis has received considerable attention. Mass spectrometry methods, especially LDMS, could provide an alternative for electrophoretic technology, permitting analysis of sequencing reaction products in minutes rather than hours. However, artifactual fragmentation of oligonucleotides has hampered early efforts to apply LDMS to DNA sequencing. Michael Fitzgerald described the progress made toward understanding the fragmentation process and how to avoid it. The principle of LDMS DNA sequencing is to carry out four separate extension reactions, as usual, one for each dideoxynucleoside triphosphate, and then analyze each reaction separately by LDMS, overlaying the spectra to obtain the DNA sequence. Matrices typically used in other LDMS applications, such as 2,5-dihydroxybenzoic acid, permit analysis of long homo-T polymers (up to 100 bases long) but do not produce high quality spectra for other oligonucleotides unless they are less than 15 bases long (7). These other oligonucleotides appear to fragment at the glycosidic bond or the 3' carbon-oxygen bond (especially pronounced in G homopolymers), interfering with quick interpretation of spectra. Fragmentation occurs when the sample-matrix mixture is irradiated and desorbed from the sample target, not through post source decay, and RNA is less susceptible to fragmentation than DNA.
A number of compounds were surveyed for use as matrices in this application, focusing on aromatic compounds containing basic amino groups and no acidic groups (8). Seven new matrices were found that permit sample preparation at pH 6-10, but the overall results were similar to those obtained with acidic matrices. The matrix 2-amino-5-nitropyridine produces useful spectra when the analyte-matrix mixture is treated with an ion exchange resin to minimize sample cationization, which would otherwise produce several, poorly resolved spectral peaks for each oligonucleotide species (e.g., [M-5H+3Na+K]- and [M-5H+2Na+2K]-, differing by only 15 Da). Another matrix, 3-hydroxypicolinic acid, found by Chris Becker's group in a separate survey for matrices (9), gives spectra without fragmentation for most oligonucleotides of mixed base composition up to 100 bases long. LDMS has improved to the point that it can provide rapid analysis of model DNA sequencing reaction product mixtures up to 41 bases long with peak resolutions of 100 to 200 from only 500 fmol of each oligonucleotide species. Enzymatic reactions typically produce about 1 fmol of each extended oligonucleotide species, so further work will focus on finding matrices or improved methods for analysis of longer oligonucleotides at higher sensitivity.
Robert T. McIver, Jr., University of California at Irvine, "High Resolution MALDI with an External Ion Source Fourier Transform Mass Spectrometer". LDMS is simple, sensitive, and suitable for a wide range of samples, but for now it receives low marks for mass accuracy and resolution. Robert McIver described an instrument developed in his laboratory that addresses these issues (10).
This instrument separates ion formation and ion detection into two discrete processes that can be optimized independently of each other. Ionization occurs in the normal manner, by irradiating samples embedded in matrices that act as proton sources, but the time-of-flight tube of a conventional LDMS instrument is replaced with an ion trap, a superconducting magnet operating at about 70,000 Gauss. Once ions enter the trap, each ion orbits within the magnetic field at a radial frequency determined by its mass, ranging from 100,000 Hz for peptides and 10,000 Hz for proteins. So, with this instrument mass is measured from coherent cyclotron resonance frequencies, rather than by the time taken to travel to an electron multiplier tube. The summation of all resonance frequencies generates an image current signal, and this composite signal is deconvoluted by Fourier transform on a computer to produce a spectrum.
In conventional LDMS, the highest resolution obtained in linear analysis is about 400, and in reflectron analysis about 2,000. This instrument has provided much higher resolution: 1,100,000 for [Arg8]-vasopressin (1,084 Da) and 90,000 for bovine insulin (5,734 Da). Isotopic species of both peptides are baseline resolved in the spectra. Resolution is so high that average molecular masses must be calculated from the isotopicly resolved spectral peaks, which are distributed as would be expected from theoretical calculations. Typical mass accuracies are about 0.002% or 2 Da per 100,000 (compared to 2 Da per 1,000 in linear analysis and 2 Da per 10,000 in reflectron analysis). Because the ion trap is based on superconducting magnet technology, which provides stable magnetic fields, the instrument needs to be calibrated only about once a month and internal calibration is unnecessary.
Five standard peptides spanning a mass range of about 1,000 to 6,000 Da provided spectra with signal-to-noise ratios of 50 to 100 when only 1 to 5 fmol was used for mass analysis. As the amount of sample was decreased further, into the amol range, unimolecular decay of the sample became more prominent, especially deamidation of peptides, to the extent that the singly protonated species was a minor component in a spectrum dominated by M-(NH3)n species. This suggests that the limit of detection is not based on detector technology but on other hardware issues relating to sample ionization.
References
Return to the The ABRF Home Page