Identification of furoyl-containing advanced glycation products in collagen samples from diabetic and healthy rats

~iochimica et Biophysica Acta,

1033 (1990) 13-18

13

Elsevier BBAGEN 23240

Identification of furoyl-containing advanced glycation products in collagen samples from diabetic and healthy rats A. L a p o l l a 1, C. G e r h a r d i n g e r 1, E. G h e z z o 2, R. S e r a g l i a 2, A. S t u r a r o D . F e d e l e 3 a n d P. T r a l d i 2

2, G .

C r e p a l d i 1,

lstituto di Medicina Interna, Patologia Medica 1, Policlinico Unioersitario, Padova, 2 CNR, Area di Ricerca di Padova, and ~ Cattedra di Endocrinologia, Universittl di Palermo, Palermo (Italy)

(Received4 April 1989) (Revised manuscriptreceived11 August 1989)

Key words: Diabetes;Advancedglycationproduct; Acid hydrolysis;Proteinasehydrolysis;Mass spectrometry;Insolublecollagen

The compounds resulting from the reaction of glucose with proteins (advanced glycation products) can be important markers of chronic diabetic complications. To test the possible diagnostic value of advanced glycation products containing the furoyl moiety, collagen samples from diabetic and healthy rats were analyzed by parent ion spectroscopy. In our study, we compared normal collagen, diabetic collagen and normal collagen incubated with different glucose concentrations and we employed different hydrolysis procedures (HC! and proteinase). Mass spectroscopic measurements performed on hydrolyzed samples showed that either different samples or different hydrolysis procedures produce a similar set of fmmyl-containing compounds. 2-(2-Furoyl)-4(5)-(2-furanyl)-lH-imidazole (FFI) which has been reported to be one of the advanced glycation products, was never found in any of the samples examined. Hence neither FFI nor furoyl-containing molecules can be considered markers of advanced glycation processes.

Introduction Diabetes is a well-known biochemical failure which affects about 3% of the world population [1]. The development of insulin therapy has overcome the acute complications of this disease, but a number of chronic complications related to it (neuropathy, nephropathy, retinopathy, macroangiopathy) have grown, due to the elongation of diabetic patients' mean life [2-4]. Long-term exposure to high glucose levels has been reported to be responsible for tissue damages leading to such chronic complications [5-8]. Recently, much research has been devoted to the identification of possible products of non-enzymatic interaction between glucose and proteins [9-11]. FFI, (2-(2-furoyl)-4(5)-(-2-furanyl)1H-imidazole), was proposed to belong to this class of compounds, as it was identified among the hydrolysis products of glycated proteins either in in vitro or in in vivo experiments [12]. Furthermore, from a biosynthetic

Abbrevations: FFI, 2-(2-furoyl)-4(5)-(2-furanyl)-lH-imidazde. Correspondence: A. Lapolla, Istituto di Medicina Interna, Patologia Medica 1, PoliclinicoUniversitario,Via Giustiniani2, 35128 Padova, Italy.

point of view, FFI was proposed to originate from a condensation reaction between two glucose moieties and two lysine-derived amino groups. Hence, FFI was considered a possible crosslink product of diagnostic value to quantify the primary site of tissue crosslinking. As a matter of fact, different levels of FFI (or related compounds) were detected in relation to different glucose concentrations [13] with fluorimetric and RIA methods. More recently, by means of mass spectrometric techniques (collisional spectroscopy, daughter and parent ion scans), we were able, on the one hand to confirm easily the presence of FFI in in vitro hydrolyzed glycated proteins (albumin and polylysine) [14], and on the other, to exclude such a presence in hydrolyzed collagen samples [15]. Such apparent contrast between fluorimetric and RIA data versus those of mass spectrometry can be explained only on account of the presence of FFI structurally related compounds in collagen samples. Moreover, Monnier and co-workers demonstrated that FFI cannot be considered a real advanced glycation product, but only a compound resulting from a reaction between furosine (a byproduct of acid-hydrolyzed-glycated proteins) and ammonia which was used as neutralizer after acid hydrolysis [16]. At this point an important question remains, i.e., which are the struct-

0304-4165/90/$03.50 © 1990 ElsevierSciencePublishers B.V. (BiomedicalDivision)

14 ural identities of the compounds responsible for the levels detected by fluorimetric and RIA data. We undertook the present study to investigate this aspect. Considering that the RIA antibody was mainly reactive against furoyl moiety, we chose to perform parent ion spectroscopy on samples of different origins, to find the molecular species containing furoyl cation which could be responsible for fluorimetric and RIA data.

Materials and Methods

COLLAGEN FIBRES i HONIOGENIZtNG

COLLAGE N I BUFFER (pH 7. 4)" SAMPLE I~/C~NTR'FUGAT'ON

~:'~'T"~ ~

r-------

| 4 " " ' " - -

COLLAGEN SAMPLE

~ -

S UI:~ERNATANT

SALT SOLUBLE FRACTION ~ . . . ~ . SUPERNATANT

COLLAGEN PRECm,T,T~,~ I SAMPLE J

ACID ~OLUSLE FRACTION i ar"*~'~ SUPERNATANT

Samples. Samples of tail collagen were obtained from five male Sprague-Dawley diabetic rats and from five healthy Sprague-Dawley rats matched for sex and age. The rats were made diabetic by subcutaneous injection of alloxan (100 mg/kg). Glycosuria and plasma glucose were determined after 1 week and only rats showing glycosuria and fasting plasma glucose higher than 300 m g / d l were selected for the study. The diabetic rats, not treated with insulin, were studied after 12 weeks; they showed a mean weight of 292 + 29 g (+S.E.); mean fasting plasma glucose [17] of 459 + 20 mg/dl; and mean HbAlc [18] of 3.36 + 0.24%. Healthy control rats showed on the day of the study a mean weight of 596 + 32 g, mean fasting plasma glucose of 112 + 11 m g / d l and a mean HbAlc value of 1.04 + 0.07% Preparation of collagen. The collagen treatment was the same as that proposed by Schnider [19] (Fig. 1). The samples obtained from rats were immediately frozen at - 7 0 °C until used. The frozen samples were homogenized in a Polytron in 1.0 mol/1 NaC1 0.05 mol/l Tris-HCl buffer (pH 7.4) in the proportion of 10 g samples/100 ml of buffer, until reaching a fine suspension. Suspensions were stirred for 18 h at + 4 ° C and then centrifuged at 11000 × g for 20 min at + 4 ° C . The precipitates were resuspended in the buffer, centrifuged twice more, and then extracted in a 0.05 mol/1 acetic acid solution (10 g/100 ml) for 18 h at + 4 ° C . The samples were centrifuged twice at 11000 x g for 20 min at +40C. The precipitate remaining from 0.75 g of samples was mixed with 20 ml of 0.05 mol/1 acetic acid and with pepsin (Sigma, St. Louis, MO, U.S.A.) at a concentration of 1.0 mg/ml. After an incubation of 18 h at + 4 ° C , NaC1 in Tris-HCl buffer (pH 7.4) was added to the solution to obtain a 0.17 mol/1 salt concentration. Samples were centrifuged three times at 50000 x g for 1 h at +40C. The insoluble collagen obtained in this way was washed twice with distilled water and centrifuged at 75 000 x g for 20 rain at + 4 ° C. The pellet was suspended in 30 ml NaCI 1 mol/1, and sonicated to obtain a fine suspension; finally, it was centrifuged at 20 000 × g for 20 rain at + 4 ° C . The pellet was mixed with 0.1 mol/1 CAC12/0.02 mol/1 Tris-CH1 (pH 7.55) containing 0.05% toluene in the proportion of 1 ml buffer/10 mg

COLLAGEN SAMPLE

PEPSIN SOLUBLE FRACTION

I

COLLAGENASE

COLLAGEN i

= I sPECTROSCOPv [

SAMPLE I SOLUBILIZE D I

I.VOROLVZml I SAMPLE I

I

1

.EO..A',ZAT,O.

i

t

s,.sl

,I .,o.ox, I PROLmE I

IANALYEI~I • 1.0 mol/I NoCt 0.05 mol/I

IMEAEUREMeNTI Yris HCt buffer

Fig. 1. Treatment of collagen samples.

sample, and then with a solution of purified collagenase (Sigma, St. Louis, MO, U.S.A.) in the proportion of 0.5 ml/10 nag sample. The collagenase solution was obtained by adding 1 nag of collagenase for each ml of 0.1 mol/1 CAC12/0.02 mol/l Tris-HC1 buffer (pH 7.55). The samples were incubated for 24 h at + 37 °C on a shaking platform and then centrifuged at 11000 × g for 30 rain. The supernatants were used to measure the amount of digested collagen and for spectroscopic and spectrometric determinations. To assess the effects of hydrolysis on the production of FFI and its related compounds, we performed parent ion spectroscopy measurements on diabetic and normal tail collagen samples either after HCI hydrolysis, as described below, or after proteinase digestion. The enzymatic digestion with proteinase K (Sigma, St. Louis, MO, U.S.A.) was obtained by dissolving the samples in 0.1 mol/l Tris-HCl buffer containing 0.1 mol/NaCl. The ratio of protein to enzyme was 10:1 (w/w). The samples were then incubated for 24 h at 37 °C and lyophilized. In vitro incubation of collagen. To carry out the glycation processes, we incubated healthy collagen samples with different amounts of glucose; more specifically, two samples of tail collagen were incubated with a quantity of glucose proportioned to 100 nag of collagen, 1 g of D-glucose (Sigma, St. Louis, MO, U.S.A.), and 1

15 TABLE I

sured. Samples were hydrolyzed in HC1 6 mol/1 for 18 h at + 95°C under vacuum. HC1 was then removed under reduced pressure and hydroxyproline was determined according to the procedures of Zender [20]. Spectroscopical measurements. Absorption measurements at 350 nm were performed using a VARIAN CARY 17D spectrometer against a blank collagenase. Mass spectrometric measurements. All mass spectrometric measurements were performed on a doublefocusing, reverse geometry VG ZAB2F instrument. The samples were introduced directly into the ion source operating under electron impact (EI) conditions (70 eV, 200/zA). Parent ion scans [21] on furoyl ions were performed by selecting the ionic species at m / z 95; its precursor ions were detected by means of B2/E = const linked scans [22]. Exact mass measurements were performed with the peak matching technique at 10000 resolution (10% valley definition). Daughter ion spectra were obtained by B / E = const linked scans [22]. Collisional experiments were performed by 8 keV ions colliding with air in the second field-free region. The pressure in the collision chamber was such as to reduce the main beam intensity to 40% of its usual value [21]. Statistical analysis. Student's t-test for unpaired data was used for statistical analysis. Our results are expressed as mean d=S,E.

Mean value ( + S.E.) of absorbance at 350 nm / m g hydroxyproline

Evaluation was of five diabetic collagen samples (1) five normal collagen samples (2), two normal collagen samples incubated with 10 g glucose (3), two normal collagen samples incubated with 0.5 glucose (4) and two normal collagen samples incubated with buffer only (5). Samples

Absorbance/mg hydroxyproline

1

0.63 + 0.06 0.305:0.04

2 3 4 5

1.4 5:0.01 0.85 5:0.01 0.35 5:0.01

Comparison 1 vs. 2: Comparison 3 vs. 5: Comparison 3 vs. 4: Comparison 4 vs. 5:

P P P P

< < < <

0.001 0.001 0.001 0.001

ml sodium phosphate buffer (pH 7.5, 0.5 mol/1 Na, 0.05% toluene). Two samples of tail collagen were incubated with glucose proportioned to 100 mg collagen, 0.05 g D-glucose, 1 ml phosphate buffer. Two samples of tail collagen were incubated only with buffer in the same proportion and used as control. After 28 days incubation at 37 ° C, the samples were extensively dialyzed against distilled water for 24 h and lyophilized. Then, the above-mentioned procedure was followed to prepare samples of insoluble collagen; every sample was subsequently divided into two aliquots, one for HC1 hydrolysis, the other for proteinase hydrolysis. Measurement of collagen. In order to assess the amount of digested collagen, hydroxyproline was mea-

Results The mean value of absorbance at 350 nm/mg hydroxyproline (+S.E.) was significantly higher in diabetic rats with respect to healthy rats (0.63 + 0.06 vs.

TABLE lI lonic species detected by parent ion scans (B2/E = cons 0 on furoyl cation (m / z 95) performed on the compounds under study

The ion abundance was calculated by comparing the absolute signal obtained introducing a comparable amount of samples and measuring the peak intensities with respect to the most abundant one, kept equal to 100. Relative abundance, + ffi ( 1 - 5 ) - 1 0 - s amp6res; + + ffi (5-10). 1 0 - s , amp6res. Compounds HC1 hydrolyzed normal collagen

Proteinase hydrolyzed normal collagen HCI hydrolyzed diabetic collagen Proteinase hydrolyzed diabetic collagen Collagen incubated with 0.5 g glucose and HCI hydrolyzed

Ionic species 96

110

124

150

+ + + + + +

++ + ++ +

+ + ++ +-4-

+ + + -4-

+

++

+

+

+

++

+

+

+

++

+

+

+

++

++

+

154

Collagen incubated with 10 g glucose and HCI hydrolyzed

Collagen incubated with 0.5 g glucose and proteinase hydrolyzed Collagen incubated with 10 g glucose and proteinas¢ hydrolyzed Structures assigned on the basis of accurate mass measurements and daughter ion spectroscopy

??

II

O

II

O

II

O

II

O

C H 3 ~ / ~ O

COOH

16

A

xlO all

xlO0

I I

II. II 11. II,

I, 1 il II

.q, w.

|

100

200

3 O0

xlO

400

r'ff 1iTJ 5 O0

6 O0

xlO0

x 1000

B fdl

100

200

300

400

500

600

Fig. 2. (A) 70 eV (El) mass spectrum of an HC1 hydrolyzed normal collagen sample. (B) 70 eV (EI) mass spectrum of an HCI hydrolyzcd collagen sample incubated with 10 g glucose.

0 . 3 0 + 0 . 0 4 ; p <0.001) (Table I). The value of abs o r b a n c e / m g hydroxyproline was 1.4 in collagen incubated with high glucose concentration, 0.85 in collagen incubated with more physiological glucose concentration, 0.35 in collagen incubated with buffer only (Table I). The results obtained by B 2 / E linked scans on furoyl cation ( m / z 95) performed on the samples under study are reported in Table II. The signal intensities of precursor ions are represented by + = (1-5)10 -8 amp6res and + + = ( 5 - 1 0 ) 1 0 -8 amp6res. In order to have quasi-qualitative data, comparable amounts from the samples were introduced. In Fig. 2, the 70 eV electron impact (El) mass spectra of HCI hydrolyzed normal collagen samples and HC1 hydrolyzed collagen samples incubated with 10 g glucose (A and B, respectively) are reported. They are quite complicated, showing ionic species up to m / z 600 for both the analyzed samples. In Fig. 3 the B 2 / E linked scans (parent ion spectra) of ionic species at r n / z 95 from the mass spectra in Fig. 2 are reported. By such approach all the precursor ions of furoyl moiety are easily detected, without any interference from the m a n y compounds present in the hydrolyzed mixture (Fig. 2). In all the eight samples analyzed, ionic species at m / z

A

124

110 L

124

110 ~

96

B

96

Fig. 3. (A) B2/E ~ const linked scan spectrum of ionic species at m/z 95 generated by E1 of an HCI hydrolyzed normal collagen sample. (B) B2/E = const linked scan spectrum of ionic species at m/z 95 generated by El of an HCI hydrolyzed sample incubated with 10 g glucose.

17 96, 110, 124 and 150 were detected, while ions at m/z 154 are present only in the parent ion scan of the 10 g glucose sample incubated with collagen and proteinase hydrolyzed. The abundance of ions appears to be increased in HC1 hydrolyzed compared with proteinase hydrolyzed samples; this proves that the former is more effective in the production of such compounds. Structure assignments for the detected ions were obtained by either exact mass measurements or collision spectroscopy data. Discussion

FFI was first proposed as marker of advanced glycation of proteins [12]. Interesting results were obtained in FFI identification in either in vitro or in vivo experiments by fluorimetric and RIA methods [12,13]. In the latter case, by developing antibodies against FFI and related compounds (all containing furoyl moiety), different levels were found between normal human globin and albumin and in vitro glycated albumin [13]. These data reasonably led to the conclusion that in in vitro glycated albumin, higher concentrations of FFI were present. Conversely, Monnier and co-workers [14] unequivocally proved that FFI cannot be considered a real marker of ~dvanced glycation. They showed that it results from the reaction between furosine and NH 3 during the HC1 hydrolysis-NH4C1 neutralization procedure [16]. This result was further confirmed by our study in which, instead of HC1 hydrolysis, protease digestion of proteins was used (Gerhardinger, C., Lapolla, A., Crepaldi, G., Fedele, D., Ghezzo, E. and Traldi, P., unpublished data). Many other different possible markers for the growing processes were proposed such as 5-hydroxyl-1neopentylpyrrole-2-carbaldehyde, 2-(2-hydroxyacetyl)1-neopentylpyrrole, and 2-acetyl-l-neopentylpyrrole [11]. Unfortunately, these cannot account for the RIA data. We undertook the present study with the aim of detecting possible FFI-related compounds containing furoyl moiety, which is responsible for the data obtained from RIA [12,13]. In order to reach this goal, we thought it interesting to employ a mass spectrometric technique, i.e., parent ion spectroscopy. This instrumental approach consists of the detection of all the precursor ions of a preselected ionic species. In a double-focus instrument, this is easily achieved by means of scanning both magnetic and electrostatic sectors (B and E, respectively) maintaining the ratio B2/E = const [21]. We chose the furoyl cation as preselected ionic species, because, due to its low ionization energy, it is easily formed. The fact that, in the FFI mass spectrum obtained in electron impact conditions, the furoyl cation relative abundance is about

85% is a significant example. Usually, E1 mass spectrometry fails to determine directly possible glycation products in collagen extracts. Fig. 2 shows the EI mass spectra of an HC1 hydrolyzed normal collagen sample (Fig. 2A) and of an HC1 hydrolyzed collagen sample incubated with 10 g glucose (Fig. 2B). These spectra appear to be rather complex due to the presence of peaks up to m/z 600 and do not allow any clear identification of molecular species of interest. In fact, these spectra show the molecular and fragment ions resulting from the many different species in the complex natural matrix, which, being volatile, become operative under EI conditions. This suggests that no diagnostic compound can be detected under El conditions. In recent years, most instrumental research in the field of mass spectrometry has been devoted to the direct analysis of trace compounds in complex matrices. One such technique was revealed to be a validated instrument in our study. B2/E linked scan, performed on the ionic species at m/z 95, led to the unequivocal results showed in Fig. 3A and B. By such methodology, all the possible precursors of the furoyl cation, at m/z 96, 110, 124 and 150, can be identified perfectly. It can be noted that usual E1 mass spectra fail to show these ionic species due to the high 'chemical noise' coming from the matrix. The data obtained by this approach can be found in Table II. There are no great differences between the various samples. Parent ion scans show the same molecular species with minor differences in relative abundance. The structure assigned to the molecular species detected in this way was determined by means of two different mass spectrometric techniques. By accurate mass measurements, the elemental formulae of the ionic species were easily obtained while collisional experiments (daughter scans) led to precious information about their structure. The power of this last technique in structure elucidation in the biomedical field has been reported by many authors [23]. 2-Furanaldehyde (m/z 96), 2-acetylfuran (m/z 110) and furanglyoxal (m/z 124) were previously detected in samples from in in vitro experiments of glycation of albumin and polylysine (Gerhardinger, C., Lapolla, A., Crepaldi, G., Fedele, D., Ghezzo, E. and Traldi, P., unpublished data). The presence of furoylglyoxal is significant: this compound was proposed by Monnier and co-workers as an intermediate product in the FFI synthesis, before reaction with ammonia [14]. These precursor ions were shown to be more abundant in samples originating from HC1 hydrolysis and this led us to conclude that acidification and further neutralization are very important in the production of these compounds. Ions at m/z 150 and 154 (the latter present in samples of collagen incubated with 10 g glucose and

18 proteinase hydrolyzed only) required further investigation. They were never detected in glyeated albumin and polylysine samples (Lapolla, A., Poli, T., Gerhardinger, C., Fedele, D., Crepaldi, G., Chiarello, D., Ghezzo, E. and Traldi, P., unpublished data). While both accurate mass measurements and daughter ion scans were in agreement with the structure of 2-carboxy-2-acetylfurane for the ions at m / z 154, the same analytical approaches did not lead to any clear identification of the ionic species at m / z 150, due to the multiple structure of the related peak. In conclusion the data discussed above prove that: (1) glycated and non-glycated collagen sampies contain the same furoyl-contalning compounds: (2) from a qualitative point of view, hydrolysis does not affect the production of such compounds and only slightly affects the compounds relative abundance. In other words, our data prove that normal collagen samples do not differ significantly from diabetic collagen samples and in vitro glycated collagen samples in the production of furoyl containing compounds. Considering that FFI was never found in any of the samples examined, it can be concluded that neither FFI nor furoyl-containing compounds can be considered markers of advanced glycation processes. Our data apparently disagree with the data obtained by RIA and those reported in the literature. This is probably due to the fact that different substrates were used in the experiments considered. In a previous study, we showed the differences between in vivo and in vitro experiments. By using the same technique, we were able to confirm the presence of some FFI-related imidazole derivatives among in vitro HC1 hydrolyzed glycated albumin and polylysine (Lapolla A., Poli T., Gerhardinger C. Fedele D., Chiarello D., Ghezzo E. and Traldi P., unpublished data). However, these compounds were not present in diabetic and healthy collagen samples. Furthermore, it is significant that no FFI or FFI-related compounds were produced even by incubating collagen samples with the same glucose amount employed for in vitro albumin and polylysine experiments. This does necessarily mean that the processes operating in vitro and in vivo are fundamentally different. Our efforts are now devoted to the structural identification of possible real makers of advanced glycation processes.

Almowledgments This work was supported by a Grant from Consiglio Nazionale delle Ricerche: Progetto finaliT~ato Medicina

Preventiva e Riabilitativa-Sottoprogetto Malattie Degenerative.

References 1 Krolewski, A. and Warram, J.H. (1985) in Joslin's Diabetes Mellitus (Marble, e,d) 12-42, Lea Febiger, Philadelphia. 2 Shade, D.S,, Santiago, J.V., Skyler, J.S. and Rizza, R.A. (1983) in Intensive insulin Therapy, Vol. 1, pp. 341-358, Medical Examination Publishing, Princeton. 3 Pirart, J. (1978) Diabetes Care 1, 168-188 and 252-263. 4 Cudworth, A.G., Bodensky, H.J. and West, K.M. (1982) in Complications of Diabetes (Keen, H. and Jarret, J., eds.) pp. 1-18, Edward Arnold, London. 5 Kennedy, L. and Bayness, J.W. (1984) Diabetologia 26, 93-98. 6 Brownlee, M., Vlassara, H. and Cerami, A. (1984) Ann. Int. Med. 101, 527-537. 7 Reynolds, T.M. (1963) Ist Adv. Food. Res. 12, 1-52. 8 Reynolds, T.M. (1965) lind Adv. Food. Res. 14, 167-283. 9 Clark, A.V. and Tannenbaum, S.R. (1970) J. AGO'.Food. Chem. 18, 891-894. 10 Clark, A.V. and Tannenbaum, S.R. (1974) J. Agr. Food. Chem. 22, 1083-1089. 11 Njoroge, G., Sayre, L. and Monnier, V.M. (1987) Carbohyd. Res. 167, 211-220. 12 Pongor, S, Ulrich, P., Bencsath, A. and Cerami A. (1984) Proc. Natl. Acad, Sci USA 81, 2684-2688. 13 Chang, J.C.F., Ulrich, P.C., Bucala, R. and Cerami A. (1985) J. Biol. Chem. 260, 7970-7974. 14 Pelli, B., Sturaro, A., Traldi, P., Lapolla, A., Poli, T., Fedele, D. and Crepaldi, G. (1988) Biomed. Env. Mass Spectr. 15, 7-11. 15 Poli, T., Lapolla, A., Pelli, B., Sturaro, A., Gerhardinger, C., Franchin, A., Barison, A., Traldi, P. and Fedele, D. (1988) G. Ital. Diab. 8, 209. 16 Njoroge, F.G., Fernandes, A.A. and Monnier, V.M. (1988) J. Biol. Chem. 263, 10646-10652. 17 Hugget, A. and Nixon, D.A. (1957) Lancet ii, 368-370. 18 Schifreen, R.S., Hickingbotham, M.G. and Bowers, G.N. (1980) Clin. Chem. 26, 466-470. 19 Schnider, S.L. and Kohn, R.R. (1981) J. Clin. Invest. 67, 1630-1635. 20 Zender, H. (1972) Clin, Chim. Acta 37, 263-269. 21 Jennings, KR and Mason, RS (1983) In Tandem Mass Spectrometry (McLafferty, F.W., ed.), pp. 197-222, J. Wiley and Sons, London. 22 Bruins, A.P., Jenning, KR and Evans, S. (1978) Int. J. Mass Spectrom. Ion. Phys. 26, 395-395. 23 Busch, K.L., Glish, G.L. and McLuckey, S.A. (1988) MS/MS, Techiques and Applications of Tandem Mass Spectroscopy VCH, pp. 217-266, New York.