CHAPTER 9

Laser Desorption Ionization Mass Spectrometry of Bioorganic Molecules

Michael Karas and Ute Bahr

1. Introduction Matrix-assisted laser desorption/ionization (LDI) mass spectrom-

etry (MS) is a very young method that has overcome the mass limita- tions for the mass spectrometry of biopolymers (1-4). Four years ago, a UV-absorbing matrix was used to extend the accessible mass range of UV-LDI of peptides, and a strong dependence on the UV absorption properties of the matrix was demonstrated. The mol-wt determination capability is now ca. 300,000, and since the method is still evolving, its potential is far from fully exploited. Two other mass spectrometric methods are currently applicable to high-molecular mass determina- tion; plasma desorption mass spectrometry (Chapter 10 in this vol.) enables the production and detection of molecular ions of proteins up to about 30 kDa. Another recent technique, electrospray MS (5), works by spraying a solution of sample into an electric field, producing highly charged droplets, from which molecular ions are desorbed. The char- acteristic feature of electrospray mass spectra is a distribution of multiple-charged molecular ions, allowing mass measurement of pro- teins up to 70 kDa (see Chapter 8).

UV-laser desorption of organic molecules without a matrix, mten- sively studied in the authors’ group for the past few years, shows systematic limitations. The ability to desorb intact molecular ions (and structure-specific fragment ions) was found to be related to a strong

From Methods in Molecular Biology, Vol 17’ Spectroscopfc Methods and Analyses NMR, Mass Spectrometry, and Metaltoprotem Techmques Edlted by. C Jones, B Mulloy,

and A H Thomas Copynght Q1993 Humana Press Inc , Totowa, NJ

215

216 Karas and Bahr

resonance absorption band of the analyte at the laser wavelength used and required careful maintenance of the applied laser power density (irradiance, W/cm2) at the lowest possible value (threshold). For larger molecules especially, intense fragmentation could not be avoided, imposing a mass limit of ca. 1500 dalton. These limitations were over- come by the introduction of the matrix desorption technique. The prm- ciple of this method, as we so far understand it, IS as follows: The analyte molecules are embedded m an excess matrix of small organic molecules that show a high resonant absorption at the laser wave- length used. The matrix absorbs the laser energy, thus inducing a soft disintegration of the sample-matrix mixture into free (gas phase) matrix and analyte molecules and molecular ions. A more detailed descrip- tion of the process is given in ref. 3. In general, only molecular ions of the analyte molecules are produced, and almost no fragmentation occurs. This makes the method well suited for mol-wt determinations and mixture analysis. The instructions given in the following are based on the experience of the authors’ group and that of Beavis and Chait at Rockefeller University (6-8); currently, more than 300 different proteins have been successfully analyzed, and no major limitations have yet emerged.

2. Materials A few microliters of a 1 Oe5- 1 Ow7M solution of the biopolymer sample,

typically 1 pmol or less, suffice for analysis. The practical limit is determined more by the volume of sample solution that can be handled easily, than by the amount consumed in the analysis. Since this is estimated to need only about lo-t7 mol, most of the sample material can be recovered after the analysis. The smallest amount of sample that has been used to date is about 50 fmol (5 x lo-” mol). Table 1 shows the substances usable as matrices, and their corresponding sol- vents and wavelengths.

The best results are obtained by using materials of the highest avail- able purity for analytes, matrix, and solvents. The degree of salt con- tamination, and the concentrations of additives, such as buffers, compatible with high-quality results depend both on their nature and on the matrix used. When using nicotinic acid, salt concentrations of up to about 10m3M cause only slight degradation of the signal quality, and the same is true for detergents, such as dodecylsulfate, Tween@, or

LDI-MS of Bioorganic Molecules 217

Table 1 Commonly Used Matrrces

Matrix

Nicotinic acid Ferulic acid*

Smapmrc acid* Caffetc actd* 2,5-dlhydroxybenzotc actd Vanillic acid 3-Nitrobenzylalcohol Pyrazine-carboxylic actd 2-Aminobenzoic acid 3-Aminopyrazine-Z

carboxyhc acid 7,8-Dihydroxycoumarin

Solvent

Water/lo% ethanol 1.1 Mixture of

ethanol 0.1% TFA II I9

Water/lo% ethanol Water/lo% ethanol (Liquid matrix) 1.1 Water/ethanol Water/20-30% ethanol 1: 1 Water/ethanol

Water/lo% ethanol

Wavelength

266,220-290 nm 266,337,355 nm

,I I,

266,337,355 nm 266 nm 266 nm 266 nm 266,331,355 nm 331 nm

266,337,355 nm

*Caffelc acid IS 3,4-dihydroxycmnamlc acid, ferulic acid IS the 3-methoxy denvatwe, and smapmic acid IS the 3,4-dlmethoxy denvatwe.

Triton 100”. In the case of cinnamic acid matrices, the protein may be loaded from a buffered solution (e.g., 50 mMNa-Citrate). In sinapinic acid, matrix salt concentrations of 1M are tolerable. This behavior is highly desirable for practical applications, because proteins can be examined under physiological conditions and problems owing to dena- turation and limited solubility can be avoided.

3. Methods

3.1. Sample Preparation Sample preparation is a critical step for successful matrix desorp-

tion. Even though matrix and sample will usually form a homogenous solution, separation of matrix and analyte may occur during drying, which is fatal to the laser desorption technique. The following very simple preparation technique is generally usable.

3.1.1. Proteins and Glycoproteins For protein and glycoprotein analysis, the solid sample is dissolved

in double-distilled water either alone or containing about 10% etha- nol. Depending on its solubility, the matrix IS dissolved m water, water/ ethanol, or water/acetomtrile (5-10 g/L). Trifluoroacetic acid (TFA)

218 Karas and Bahr

is usually added to a 0.1% concentration (see Table 1). The sample and matrix solutions (0.5-l & each) are mixed on an inert metallic (Ag, Pt) sample target, and are dried in a stream of (warm) air or in vacua inside the mass spectrometer. A greater excess of matrix is preferable when using cinnamic acid, and 1 lt,L of protein sample is therefore mixed with 10 ltL of matrix solution: 0.5-l l.tL of this final solution is used for analysis. The sample is then introduced into the vacuum chamber of the mass spectrometer for analysis.

3.1.2. Carbohydrates For carbohydrate analysis, about 1 g/L of aqueous solution is used.

This class of compounds works best in a 1:4 mixture with dihydroxy- benzoic acid at an excitation wavelength of 337 or 355 nm, or with nicotinic acid at 266 nm.

3.1.3. Nucleotides

Nucleotide samples are dissolved in double-distilled water at 0.1 to 1 g/L and mixed with 5 x 10m2M matrix in a ratio of 1: 1. The matrices are aminobenzoic acid or aminopyrazine carboxylic acid with a laser wavelength of 337 nm.

3.2. Apparatus Until now, laser ionization time-of-flight mass spectrometers were

only commercially available as microprobe instruments (LAMMA 1000, Leybold Heraeus, Koln, Germany; LIMA, Cambridge Instr./ Kratos, Cambridge, UK). For our experiments, a LAMMA 1000 laser microprobe prototype instrument was used, which is described in detail elsewhere (9). Figure 1 shows the schematic diagram of a laser mass spectrometer. Since then, the first instrumentation dedicated to the LDI of biopolymers has been introduced (Vestec Corp., Houston, TX), and further suppliers are expected to follow in the near future.

The sample is irradiated with short pulses of UV laser light, either with 5-10 ns laser pulses from a Q-switched Nd-Yag-laser at a wave- length of 266 nm (frequency-quadrupled) or at 355 nm (frequency- tripled). Pulses (3 ns) from a nitrogen laser at 337 nm may also be used. In each case, a matrix that absorbs at the appropriate wavelength is required. With microprobe instruments, the focus diameter is between 3-30 pm. Larger focal areas of ca. 100-500 l.un, attainable by single quartz lenses of 10-50 cm focal length, can be used and are advantageous,

LDI-MS of Bioorganic Molecules 219

I I

Soinple stage

Tronslentrec

Fig 1 Schematic diagram of a reflector-type time-of-fhght mass spectrometer with laser ion source.

because the microheterogeneity of the sample is then less important. Good results are only obtained over a narrow irradiance range, which (since it depends on the focal area) has to be determined for the instru- ment used. Typical values are in the 106-lo7 W/cm2 range. For a focal area of low3 cm2, an irradiance of lo7 W/cm2 corresponds to an energy of the laser pulse of only 100 J; thus, low-power, inexpensive lasers such as nitrogen lasers, can be used. Each pulse produces ions (positive as well as negative) that are accelerated by an appropriate voltage, focused through an ion-optical system, mass separated in a time-of- flight tube (with or without an ton reflector), and detected with a secondary ion multiplier. A digital oscilloscope providing a fast ana- log-to-digital converston (record length of 32K samples and vertical resolution of 8 bit) and a maximum time resolution of 10 ns (e.g., Le Croy 9400) is needed for signal recording. Although each laser pulse produces a complete mass spectrum, the signal-to-noise ratio is usu- ally improved by summing 1 O-30 single-shot spectra. Spectra are then

220 Karas and Bahr

Fig. 2 LDI mass spectrum of human albumm, obtained with 337 nm and feruhc acid as matrix Twenty single spectra have been accumulated

further processed (e.g., calibration and mass assignment) using per- sonal computer-based software. The time for a single analysis, including sample preparation, is typically cl5 min.

3.3. Analysis of Proteins and Glycoproteins

Figure 2 shows the mass spectrum of the protein human serum albumin (mol wt 66,437) obtained using a 337-nm wavelength laser, ferulic acid matrix and represents the accumulation of 20 single-laser shots. The molecular ion is the base peak, and is accompanied by doubly and triply charged molecular ions and cluster molecular ions [M] n+‘-. All of these signals can be used for mol-wt determination, Fro: results obtained for peptides and small proteins, it can be deduced that, whereas molecular ions are mainly (de)protonated species, the contribution of charged species formed by addition of alkali metal ions also has to be taken mto account. Low-mass signals can be attributed exclusively to the matrix or low-mol-wt contaminants. Fragment ions owing to the cleavage of covalent bonds (e.g., peptide bonds) are not observed even with increasing irradiance. Too high an irradiance will initially degrade the high-mass signal intensities and quality, until the whole process switches to a mode where only unspecific low-mol-wt

LDI-MS of Bioorganic Molecules 221

Fig. 3 LDI mass spectrum of a mixture of cyclodextrm and maltoheptaose

fragment ions are observed. Glycoprotems with a carbohydrate content of up to 80% have been analyzed and show the same characteristics.

3.4. Analysis of Carbohydrates

To date, only underivatized sugars and sugar mixtures, such as maltodextrins with a mol wt of 3500 and mannosides containing 1 N- acetylglucosamme up to a mol wt of 1700, have been analyzed. Quasi- molecular ions formed by sodium or potassium ion addition to a neutral molecule are the only species detected; neither multiply charged nor cluster ions are produced. Figure 3 shows the LDI spectrum of a mix- ture of maltoheptaose and cyclodextrin with mol wts of 1175 and 1134. The sodium adducted molecular ions of both substances and the potassium adduct of maltoheptaose are desorbed.

3.5. Analysis of Nucleotid-es Spectra from nucleic acids and oligonucleotides up to a mol wt of

39,000 have been obtained. Better signal intensities are given in the negative ion mode for oligonucleotides. When operating at 337 nm, several matrices can be used, but the best results were obtained with aminopyrazinecarboxylic acid and ammobenzoic acid. As the mol wt increases, the peaks become broader, because protons are progres- sively exchanged for alkali metal ions complicating mol-wt determina- tion. Figure 4 shows the negative ion spectrum of ohgodeoxythymidylic

222 Karas and Bahr

2000 5000 lb000 ii000

M,/z

Ag. 4 LDI mass spectrum of ohgodeoxythymidyhc acid (d[ptl15) with a mol wt of 458 1 The spectrum represents the sum of 25 single laser shots with ammopyrazine- carboxylic acid as matrix at 337-nm wavelength

acid d(pT)i5 with a mol wt of 458 1. Twenty-five single spectra were summed to produce the spectrum, which was obtained using an aminopyrazinecarbonic acid matrix.

3.6. Structural Information

Besides mol-wt information, matrix UV-LDI may also yield some information about subunit structure of proteins. Figure 5 shows the spectrum of lectin (from Cunavalia ensz’formis) obtained using a nico- tinic acid matrix. The matrix solution contained 10% ethanol. The molecular ion pattern observed differs from that of simple single- chain proteins. The intensity distribution cannot be interpreted as being the result of multiple charging of a larger protein or of the production of cluster ions from a smaller protein because of the absence of triply charged species. Thus, the spectrum represents the subunit structure of lectin with the stable tetra-, tri-, di-, and monomeric species, as well as their doubly charged species, as indicated m the assignment of Fig. 5. The dissociation into subunits is mainly the result of the conditions in the matrix-analyte mixture and is strongly promoted by the addition of alcohol. This was proven experimentally and also explains why matri-

LDI-MS of Bioorganic Molecules 223

Fig 5 LDI spectrum of lectm (from Cunaval~.~ enslfomzis) with a mol wt of 102,000.

ces that are less water-soluble and thus need a higher percentage oforganic solvent in the matrix solution (e.g., the cinnamic acid derivatives) show only signals corresponding to the subunits. All in all, there is strong evidence that matrix-LDI simply reflects the aggregation state m the initial matrix solution and may therefore be used to determine the state of association of proteins. Future work will show if enzyme/ substrate and/or antibody/antigen complexes can also be desorbed intact.

4. Notes 4.1. Mass Resolution

and Mass Determination Accuracy The mass resolution achieved in the microprobe instrument used by the

authors is typically 300-500 daltons in the low-mass range. For high- mass ions, peak widths at half-height corresponding to a mass resolution of 150 are the best observed, whereas Beavis and Chait report an upper limit of 700 with their experimental arrangement. Several factors contrib- uting to a reduced mass resolution have been defined so far; others need more careful examination. The different methods of ion detection (sepa- rate conversion dynode in front of a Venetian blind multiplier vs double- channel plate detector) are presumed to contribute.

224 Karas and Bahr

Another factor is the tendency of some matrices, e.g., nicotinic and vanillic acid, to form artifacts by the addition of matrix molecules, fragments, or photochemical products to the analyte ion. At low reso- lution, this may reduce the accuracy of mass determination, especially at high masses where shoulders arising from adduct signals are not resolved. A simple solution to this problem is the use of cinnamic acid matrices that do not promote adduct ion formation. Finally, mass reso- lution may only appear to be low because of a chemical heterogeneity in the molecules being analyzed.

Despite the rather low experimental mass resolution, the accuracy of the mol-wt determination may reach a relatively high level. This strictly relates to the calibration procedure. Calibration with low-mass signals will only give absolute accuracy values in the range of 0.2- 0.5% because of the necessarily imprecise time measurement and the need for extrapolation into the high-mass range. The highest possible accuracy is achieved by using a biopolymer of accurately known mass to provide the calibration masses. Masses are assigned by centroiding the upper symmetric part of all molecular ions having adequate signal- to-noise ratios (as in 252Cf plasma desorption mass spectrometry, Chapter 22). Figure 6 shows the spectrum of a mixture of trypsinogen from bovine pancreas and cytochrome c. The singly and doubly charged molecular ions of cytochrome c are used as calibration masses. The measured mol mass is 23,985 + 12 dalton, which corresponds to a mass accuracy of + 0.05%. The mol wt calculated from the sequence is 23,98 1. Better mass resolution, enabling accuracies of about lOA (up to 20 kDa), has been demonstrated by Beavis and Chait for well de- fined proteins (10).

4.2. Mass Range

The mass range accessible to the LDI technique reaches up to ca. 300 kDa. The trimer of urease subunits, which is known to be stable in solution, having a mass of 272,500 dalton, is the largest molecular ion detected. The heaviest ion measured so far is the dimer of glucose isomerase at a mass of 344,800 dalton.

4.3. Combination with Biochemical Methods

Matrix-LDI mass spectrometry, like other mass spectrometric tech- niques, shows its full potential in combination with other biochemical techniques. Only a few examples can be given here.

LDI-MS of Bioorganic Molecules 225

Fig. 6 LDI mass spectrum of trypsmogen with an admixture of cytochrome C for mass calibral ion

4.3.1. Combination with Enzymatic Methods This was exemplified by the authors using enzymatic cleavage of

the carbohydrate constituent of a glycoprotein. The sugar chain of beef spleen violet phosphatase (mol wt 35,050) was removed by cleavage with the enzyme endoglycosidase H. Comparison of the spectra from the intact and the treated sample thus gives an easy and accurate determi- nation of the mol wt of the sugar component (II). Conversely, if the amino acid composition is known, as in the case of Endoglucanase III (from the fungus Trichoderma reseii), the difference between the measured mol wt of 48,780 and the calculated value enabled determi- nation of a sugar content of 6620 dalton, corresponding to 15.7% (12).

Proteins composed of disulfide-linked subunits can be separated into their #constituent units by reduction of the disulfide bonds with dithiothreitol (DTT). This has been demonstrated by nicotinic acid matrix LDI of a monoclonal antibody (IgG from mouse against a human lymphokine). Before cleavage, the spectrum shows the intact protein with the molecular ion as base peak, and with multiply charged and cluster ions present. After addition of DTT, the spectrum gives infor- mation about the mol wt of the light and heavy chains. Chait and Field

226 Karas and Bahr

have proposed monitoring the time-course of enzymatic reactions by plasma desorption mass spectrometry of proteins bound to nitrocellu- lose membranes (13). Laser desorption mass spectrometry may equally well be used in place of PD-MS, e.g., for the (partial) sequencing of protems. Until this time, the strategy has only been applied to small peptides, for example, the removal of the C-terminal amino acid of porcine renin substrate by enzymatic cleavage with carboxypeptidase Y. With ferulic acid as the matrix, the enzyme can be used under normal buffering conditions. In practice, the enzyme solution is added to a small droplet of peptide solution. After a suitable interval, the matrix solution is added, which stops the enzymatic reaction by low- ering the pH. The spectra taken after solvent evaporation can thus be used to monitor the extent of enzymatic digestion. Future improve- ments in mass resolution will enable the technique to be applied m the lo-20 kDa range as an interesting tool when rapid determination of partial sequence is sufficient. Even in its existing form, matrix-LDI can also be used to monitor the time-course of other chemical or en- zymatic reactions, and for process control applications.

4.3.2. Combination with Blotting Techniques

Because lasers can easily be focused to a small, precisely located region of the target, the technique is capable of producing spatially resolved mass spectra. The combination of LDI-MS with one- or two- dimensional separation techniques, such as SDS-PAGE, is thus a chal- lenging prospect. In this way, correct mol wt could be assigned to the separated species, and incomplete separation detected. The most prom- ising approach is the use of blotting techniques, which simultaneously retain spatial resolution and effect clean-up of the separated sample molecules. The prerequisite for this is that common blotting mem- branes can be used as substrates for subsequent matrix LDI. Whereas nitrocellulose gave poor results, experiments using polyvinylidene difluoride (PVDF) were much more encouragmg. In the mitial inves- tigations, solutions of insulin and P-lactoglobulin were dripped onto a piece of PVDF and thoroughly washed with water and water/TFA solution. One microliter of a 30% formic acid solution of ferulic or sinapinic acid was then placed on the spot. These conditions are expected to at least partially overcome the strong protein-PVDF interaction. Experiments applying the technique to electroblotted proteins after

LDI-MS of Bioorganic Molecules 227

PAGE are currently in progress in the authors’ laboratory. An addi- tional, highly interesting aspect of the successful use of PVDF-sub- strates is that the strong binding of the proteins to the membranes allows easy micropurification of the sample. Matrix-LDI may also make use of the further advantages of such a technique as, e.g., per- forming microscale chemical reaction directly on the substrate.

Note Added in Proof Although this article covers the principal aspects, the MALDI tech-

nique has quickly developed and spread out since this chapter was written. Several dedicated commercial instruments are available today. For further reading the following article is recommended: Hillenkamp, F., Karas, M., Beavis, R. C., and Chait, B. T. (1991)Anal. Chem. 63,1193A.

References I Karas, M , Bahr, U , Ingendoh, A , and Hlllenkamp, F. (1989) Laser desorptlon/

lonizatlon mass spectrometry of proteins of mass 100,000 to 250,000 dalton Angew Chem Int Ed. Engl. 28,X0,761

2 Karas, M., Ingendoh, A , Bahr, U., and Hlllenkamp, F (1989) Ultraviolet-laser desorptlon/lonizatlon mass spectrometry of femtomolar amounts of large pro- teins. Blamed. Environ. Mass Spectrom 18,841-843

3 Karas, M , Bahr, U , and Hillenkamp, F. (1989) UV-laser matrix desorptlon/lon- lzation mass spectrometry of proteins in the 100,000 dalton range Int J. Mass Spectrom Ion Processes 92,23 l-242.

4. Karas, 111. and Hlllenkamp, F. (1988) Laser desorptlon Ionization of proteins with molecular masses exceeding 10,000 daltons Anal. Chem. 60,2299-2301.

5. Smith, R. D , Loo, J A, Edmonds, C G., Barmaga, C J., and Udseth, H R (1990) New developments m biochemical mass spectrometry: Electrospray Ion- ization. Anal. Chem 62,882-889.

6 Beavis, R. C and Chalt, B. T (1989) Factors affecting the ultraviolet laser desorptlonllonizatron mass spectrometry. Rapid Commun Mass Spectrom. 3, 233-23’7

7 Beavis, R. C and Chalt, B T. (1989) Matrix-asslsted laser desorptlon mass spec- trometry using 355 nm radiation. Rapid Commun. Mass Spectrom. 3,4361139.

8 Beavis, R. C and Chait, B T. (1989) Cinnamic acid derivatives as matrices for ultraviolet laser desorptlon mass spectrometry of proteins. Raprd Commun Mass Spectrom 3,432-435

9. Feigl, P , Schueler, B , and Hillenkamp, F. (1983) LAMMA 1000, a new mstru- ment for bulk microprobe mass analysis by pulsed laser irradiation. Int. J Mass Spectrom Ion Phys. 47, 15-18.

10. Beavis, R. C and Chalt, B T. (1990) High accuracy molecular mass determina- tion of proteins using matrix assisted laser desorptlon mass spectrometry. Anal. Chem 62, 1836-1840.

228 Karas and Bahr

11 Hillenkamp, F., Karas, M , Ingendoh, A., and Stahl, B (1990) Matrix-assisted UV-laser desorptton/ionlzation: A new approach to mass spectrometry of large biomolecules, m Biological Muss Spectrometry (Burlmgame, A and McClosky, J. A , eds.), Elsevier, Amsterdam, pp 49-60

12 Karas, M , Bahr, U., Ingendoh, A , Nordhoff, E , Stahl, B , Strupat, K , and Hillenkamp, F. (1990) Anal. Chim. Acta 241, 175-185.

13 Chait, B., Chaundhary, T , and Field, F H (1987) Mass spectrometrrc character- tzatlon of mtcroscale enzyme catalyzed reactions on surface-bound pepttdes and proteins, m Methods m Protean Sequence Analysis (Walsh, K A., ed ), Humana, Clifton, NJ, pp. 483492