000
Anal. Chem. I Q S 4 , 56,000-006
(7) Glass. R. S.;Faulkner, L. R. J . Phys. Chem. 1981, 8 5 , 1160. (8) Ghss, R. S.; Faulkner. L. R . J . Phys. Chem. 1982, 8 6 , 1652. (9) Lvttmer, J. D.; Bard, A. J. J . Phys. Chem. 1981, 85, 1155.
(23) Ikeda, T.; Schmehl, R.; Denisevich, P.; Willman, K.; Murray, R. W. J . Am. Chem. SOC.1982, 104, 2683. (24) Gileadi, E.; Fullenwider, M. A.; Bockris, J. O'M. J . Electrochem. SOC. 1966, 113. 926. (25) Sakamoto, Y.; Kamohara, H. Nippon Kinzoku Gakkaishi 1961, 4 5 , 797-802; Chem. Abstr. 1981, 95, 191049s. (26) Zak, J.; Kuwana. T. J . Am. Chem. SOC.1982, 104, 5514. (27) Cope, A. C.; Trumbull, E. R. Org. React ( N . Y . ) 1960, 1 1 , 317. (28) Schuldiner, S.; Warner, T. B. Anal. Chem. 1964, 3 6 , 2510. (29) Schuldiner, S.; Warner, T. 6.J . Nectrochem. SOC. 1965, 772, 212.
(10) Christie, J. H.; Osteryoung, R. A.; Anson, F. C. J . Nectroanal. Chem. 1967, 13. 236. (11) Anson, F. C. Acc. Chem. Res. 1975, 8 , 400. (12) Faulkner. L. R., unpublished results, University of Illinois, 1974-1983. (13) Pons, S.; Khoo, S. B. Nectrochim. Acta 1982, 27, 1161. (14) Pons, S.; Khoo, S. B. J . Am. Chem. SOC. 1982, 104, 3845. (15) Bockris, J. O'M.; Reddy, A. K. N. "Modern Electrochemistry";Plenum: New York, 1970; Vol. 2, p 1328 ff. (16) Berman, R.: Faulkner, L. R. J . Am. Chem. SOC. 1972, 9 4 , 6317. (17) Berman, R.; Faulkner, L. R. J . Am. Chem. SOC.1973, 95, 3083. (18) House, H. 0.; Feng, E.; Peet, N. P. J . OIg. Chem. 1971, 3 6 , 2371. (19) Emilsson, T., unpublished results, University of Illinois, 1983. (20) He, P.; Avery, J. P.; Faulkner L. R . Anal. Chem. 1982, 5 4 , 1313A. (21) Adams, R. N. "Electrochemistry at Solid Electrodes"; Marcel Dekker: New York, 1969; p 128.
RECEIVED for review June 29, 1983. Accepted January 17, 1984. This research has been generously supported by the National Science Foundation via Grant CHE-81-06026.
Fast Atom Bombardment Combined with Tandem Mass Spectrometry for the Determination of Cyclic Peptides Kenneth B. Tomer,* Frank W. Crow, and Michael L. Gross Midwest Center for Mass Spectrometry, Department of Chemistry, University of Nebraska, Lincoln, Nebraska 68588
Kenneth D. Kopple Department of Chemistry, Illinois Institute of Technology, I I T Center, Chicago, Illinois 60616
The mass spectra and colllslonal activated decomposition (CAD) spectra of a series of underhratlzed cydk pdypeptldes lonlzed by fast atom bombardment (FAB) are reported. All compounds ghre easlly identlflable protonated molecular Ions. The colllslonal actlvatlon decomposttlon spectra of products from the (M H)' can be understood In terms of protonation of an amide nitrogen, sclsslon of the N-acyl bond, and fragmentatbn by bss of successive amlno acld resldues from the resulting ring-opened acyllum Ion. Thus, mass spectrometry experlments permit the molecular weight and the amlno acld sequence to be asslgned.
+
One of the most promising new techniques in the field of biochemical mass spectrometry is fast atom bombardment (FAB) ( I ) . FAB is a "soft" ionization technique which is especially applicable to the analysis of nonvolatile, polar, or zwitterionic compounds. Protonated or cationized positive ions or [M - HI- negative ions are produced, often with little fragmentation. The combination of FAB and tandem mass spectrometry (often called mass spectrometry/mass spectrometry, MS/MS) can provide the structural information obtainable from fragment ions which may be lacking in the FAB spectrum. MS/MS may be viewed as an alternative to gas chromatography/mass spectrometry (GC/MS) for determination of targeted compounds which have low vapor pressure and/or are thermally unstable ( 2 ) . These are often the properties of substances of biological interest. The first mass spectrometer (MS-I) of the tandem instrument is used for separation, a feature MS-I shares with an HPLC. Additionally, however, the collisional activation following separation causes fragmentation to give product ions which are necessary for verification of the molecular ion assignment and for structural interpretation. As was mentioned above, these fragments are
often missing or of low intensity in the mass spectrum of FAB-produced ions. An additional advantage of the combination is that the inevitable matrix ions associated with FAB are also removed in the separation step of MS/MS analysis. If MS-I is a high-resolution mass spectrometer and if glycerol oligomer ions are superimposed on analyte ions a t the same nominal mass, use of the high-resolution capabilities of the instrument can separate the two ions. Thus, the analyte ion can be collisionally activated without interference from matrix ions of the same nominal mass (3, 4). The analysis of peptides by mass spectrometry has been an area which has long challenged the analytical biochemist. For the most part, peptide analysis by mass spectrometry has required the use of derivatization techniques or degradation to smaller peptides or amino acids to make the peptides more amenable for mass spectral analysis. Recently, however, successful analyses of underivatized peptides have been performed by the application of the techniques of field desorption and FAB (5-8). Recently, papers have appeared in which FAB and MS/MS have been described for the determination of biological materials which have molecular weights above a few hundred (9-18). In the specific area of the determination of cyclic peptides by mass spectrometry, the application involving electron ionization has been burdened by the additional problem of the extensive formation of rearrangement ions (19,20). Since collisionally activated decomposition (CAD)spectra are typically dominated by simple cleavage reactions rather than rearrangements (21),the CAD spectra of cyclic peptides may provide straightforward sequencing information. Here we report FAB/MS and FAB/ MS/MS spectra which lead us to postulate a mechanism for the gas-phase degradation of cyclic peptides which permits determination of the amino acid sequence of these peptides.
EXPERIMENTAL SECTION The synthesis of the cyclic peptides VI-IX has been reported elsewhere (22). Compounds I-V were prepared by A. Go at IIT.
62 1984 American Chemical Society 0003-2700/84/0356-0880$01.50/0
ANALYTICAL CHEMISTRY, VOL. 56, NO. 6, MAY 1984
881
Table I. Cyclic Polypeptides mol wt cyclo-(Lys-Pro-Ala), (I) cyclo-(Leu-Pro-Gly), (11) cyclo-(Phe-Lys-Ala-Pro),(111)
cyclo-[Phe-Lys(tB0C)-Pro-Ala], (IV) cyclo-[Phe-Lys(tB0C)-Pro-Ala-Lys(tB0C)Phe-Pro-Ala-)]( V ) cyclo-(Phe-Pro-Gln),(VI) cyclo-(Val-Pro-Gln), (VII) cyclo-(Val-Pro-Gln),(VIII) cyclo-(Leu-Pro-Gln), ( I X )
592 1068 886 1086 1086
MY'
5?3
742 648 1296 1352
All compounds were purified by HPLC before analysis. Mass spectra were obtained with a Kratos MS-50 triple analyzer mass spectrometer which has been recently described (23). Briefly, the instrument consista of a high-resolution MS-I of Nier-Johnson geometry and a standard Kratos MS-50 followed by an electrostatic analyzer used as MS-11. The FAB ion source was of the standard Kratos design and was equipped with an ION TECH atom gun. The peptides (ca. 0.1 pg) were dissolved in glycerol, and a drop of the solution was placed on the copper target end of a direct insertion probe. The bombardment was with 7-kV argon or xenon atoms, CAD spectra were taken by activatingthe ions via collisions with helium gas after the double focusing analysis and by scanning MS-11, and electrostatic analyzer. CAD spectra were acquired and signal averaged with a standard Kratos DS-55 computer system using software written in this laboratory. In a typical case, 1pL of a 1 mg/mL solution of the peptide in glycerol was sufficient to obtain a CAD spectrum by averaging 20 scans (20-s scans). The SIN ratio of 31 for minor CAD peaks and >100:1 for major CAD peaks was obtained. These values are for 60000 ions/s of main beam (M + H)+and are very compound dependent.
io
bl,,,
0
-
:: 20
,ii , ,
~
186
I50
,
,
, ,
, ,
,
,
,
,6
53
ZOO
-97
LIS-PRO-D-ALA
353
333
255
M/Z
Flgure 1. FAWMS spectrum of cyclo-(Lys-Pro-o-Ala),.
j
'iZ.3
RESULTS AND DISCUSSION We have analyzed the cyclic peptides listed in Table I by FAB mass spectrometry and FAB MS/MS. The discussion of our results will be illustrated by focusing in detail on a few examples. The remaining compounds investigated follow the same general pattern. cyclo -(Lys-Pro-D-Ala)2 (I). The positive ion FAB
103
230
330
400
560
600
Figure 2. FAB/MS/MS spectrum of the (M -I-H)' ion, m l z 593, of
cyclo -(Lys-Pro-D-Ala),.
4
I
I
0 0
I
I
0
4 CH I
CYCLO-(LY S-PRO-D-ALAI2 (underlined numbers i n d i c a t e s i t e s o f protonation)
spectrum, shown in Figure 1, shows that the protonated molecular ion can be observed a t m / z 593 with a relative abundance of 35%; it is the second most abundant ion in the spectrum. No fragment ions are observed between the protonated molecular ion and m/z 297 which corresponds to half of the molecule, [Pro-Ala-Lys]H+. Other fragment ions are observed m/z 200 [H-Lys-Pro]+, m / z 169 [H-Pro-Ala]+,mlz 141 [m/z 169 - CO]', and m/z 129 [H-Lys]+. Since no higher
Figure 3. Representation of fragmentation of cyclo-(Lys-Pro-Ala), arising from protonation on site 1 (proline).
mass fragmentions are observed, complete sequencing information is not unequivocal. The elemental composition of this peptide was confirmed by its exact mass which was measured by peak matching. The proton bound glycerol hexamer, mlz 553, and the ion due to loss of two water molecules from the proton bound glycerol heptamer, m / z 609, were used as mass standards. The elemental composition of the m / z 593 was found to be CzsH,,Ns06,with a relative error in the mass measurement of 2 ppm. The FAB MS/MS spectrum is shown in Figure 2. In contrast to Figure 1,numerous high mass frament ions are observed. The origins of the major fragment ions can be understood in terms of a mechanism in which protonation occurs on an amide nitrogen of a peptide bond. When this is a ring nitrogen, the N-acyl bond may then cleave leading to a ring-opened peptide acylium ion. Further cleavages then occur by loss of successive amino acid residues to give three series of ions al-el, a2-e2, and a3-e3 (see Figure 3 and Table 11). The first two series, which arise by protonation on proline or lysine nitrogens, respectively, dominate the spectrum. The
882
ANALYTICAL CHEMISTRY, VOL. 56, NO. 6, MAY 1984
Table 11. MS/MS Fragmentation of cyclo-(Lys-Pro-Ala), protonation on proline (I)=
protonation on lysine ( 2 ) a
protonation on alanine ( 3 ) a
protonation on lysine side chain ( 4)'7
ion
loss of b
ion
loss of b
ion
loss of b
ion
loss ofC
:
Lys Ala Pro LYS Ala
a2
Ala Pro LYS Ala Pro CO, NH,
a3 b3 c, d, e3
Pro LY s Ala Pro LYS
m n
NH, CH2NH, CH,CH,NH, (CH, ),", (CH, )",
Cl
dl e1
b2
c2 d2
;12
4
r t
a Site refers to sites designated in the structures. The designated loss is of the specified residue and all other residues listed above it. For example, ion c , refers to loss of Lys, Ala, and Pro although it has not been established that the losses are sequential. All ions arise from loss of the indicated fragment from the protonated molecular ion.
Table 111. MS/MS Fragmentation of cyclo-(Leu-Pro-Gly ), protonation on proline ( l ) = ion
protonation on leucine ( 2 ) O
loss o f b
ion
protonation on glycine ( 3 ) a
loss of
ion
protonation on side chains ion
loss o f b
Pro Leu GlY Pro Leu GlY Pro Leu G~Y Pro Leu
loss ofC
m n
NH
¶
C3H7 C4H9
r
cof
a Site refers to sites designated in the structures. The designated loss is of the specified residue and all other residues listed above it. For example, ion c, refers to loss of Leu, Gly, and Pro although it has not been established that the losses are sequential. Ion i , ' also loses CO and C3H, to yield ion z . e All ions Ion c, also loses CO and C,H, to yield ion y. arise from loss of the indicated fragment from the protonated molecular ions. This loss can arise from any of the ring opened ( M t H)' ions.
0
100
150
200
250
300
n/2
350
400
450
500
550
600
Figure 4. FAB/MS spectrum of cydo -(Leu-Pro-Gly),.
series a3-e3, arising from protonation on alanine, is quite weak. In addition, fragment ions corresponding to loss of CO from the acylium ions, denoted as "prime" (i.e., all, bl', ... a i , b i , ...),are alsoabserved. Additional ions due to lysine side chain protonation and to loss of CO and NH3 from protonated lysine acylium ions (e2/1 m / z 85) are also observed (Figure 3 and Table 11). Most of the major ions as well as many of the minor ions are readily interpreted on the basis of this mechanism. The sequence of amino acids can be deduced by correctly interpreting the CAD spectrum. cyclo-(Leu-Pro-Gly),(11). The mass spectrum of I1 ionized by FAB is shown in Figure 4. In contrast to that of I, several fragment ions of minor intensity are observed at
560
I ob0
Flgure 5. FABIMSIMS spectrum of the (M -t H)' ion, m / z 1069, of cyc/o -(Leu-Pro-Gly),.
60
2 50
L
40
IO
1-
,A,'
4, 400
,,i,,, n,",,I , , ..
452
I
502
1 . 1 11 1
550
t
,, ,, LO0
Flgure 8. FAB/MS spectrum of cyc/o -(Phe-Lys(t B0C)-Pro-Ala-Lys-
( t BOC)-Phe-Pro-Ala-).
are small and/or overlapping. The spectra do, however, distinctly differ from one another. These spectra are interpretable on the same basis as used for the previous examples. Protonation may occur on any one of the ring nitrogens (indicated as sites 1-8 in Table V and sites 1-4 in Tables IV and VI). This is followed by cleavage of the protonated peptide bond to yield linear acylium ions which undergo losses of amino acid residues in an apparent sequential fashion. The fragmentation gives rise to four series of fragment ions designated as al-gl, az-gz, a3-g3, and a4-g4 for I11 and IV (see Figures 9 and 10). However, there are eight
4 Id0
3
Figure 10. FAB/MS/MS spectrum of the (MH
m / z 887, of cyclo-(Phe-Lys(f BOC)-Pro-Ala-),.
- 2fBOC 4- 2H)'
ion,
series of fragment ions formed from V due to its unsymmetrical nature, and they are designated al-gl, ap-g2, ..., a8-ga (see Figure 11). The origins of the ions in the spectra of 111, IV, and V are given in Tables IV-VI. In addition, protonation on the amine group of the lysine side chain (site 5 for I11 and IV and site 9 for IV) gives rise to a series of side chain fragmentations as indicated in Tables IV-VI. For each MS/MS spectrum, the fragment ion series arising from protonation on proline tends to be more abundant than any other. For example, al, bl + bl/, cl, el, and fl fl' are
+
ANALYTICAL CHEMISTRY, VOL. 56, NO. 6, MAY 1984
885
7 x2
I 3
E
E
@LiY
+
Flgure 11. FAB/MS/MS spectrum of the (MH - 2tBOC 2H)+ ion, m / z 887, of cyclo-(Phe-Lys(tB0C)-Pro-Ala-Lys(t B0C)-Phe-Pro-Ala-).
r PHE-PRO-GLN-PHE-PRO-GLN
mll(l: I
x z
100
200
300
500
400
600
760
+
Flgure 12. FAB MSlMS spectrum of the (M H)' ion, m / z 743, of cyclo4Phe-hc-Gln), (VI). Series a,-el arises from protonation on Pro, a,-e, from Gln, and 8,-e, from Phe. Primes arise from CO loss from corresponding unprimed; Ions I, m, n, and q arise from loss of NH,, CO, CONH3, and C,H,CONH, from (M H)', respectively. Ions r and t arise from loss NH3 from ions a, and c,,,, respectively.
+
1
100
200
300
500
400
600
+
Flgure 13. FAB MS/MS spectrum of the (M H)+ ion, m / z 649, of cyclo-(Val-Pro-Gln), (VII). Series a,-e, arises from protonation on Pro, a,-e, from Gln, and a3-e, from Val. Primes arise from CO loss from corresponding unprimed; ions m, n, q, and r, arise from loss of NH, CO, CONH,, and C,H,CQNH3 from (M is due to loss of NH3 from ion a,.
+ H)',
respectively; ion t
the most abundant a, b, c, e, and f series, respectively, for 111. The d fragment is common to all series and g,' is the second most abundant g fragment. For IV (Figure lo), al, bl, el, and fl arise by protonation on proline (see Table VI), and they are the predominant ions in the a, b, e, and f series, respectively. The d fragment is again common to all series. The situation is more complex for cyclic peptide V because protonation on a given amino acid may lead to loss of two different residues. Nevertheless, a2,4, b1,2, d1,2,3,5,7,8', and fl,zt are abundant members of their series, and they arise, a t least in part, from proline protonation. Lysine protonation is competitive but not dominant. For example, g3, g5,6 + g"5,6, and c3 and g3 are all abundant series members for 111, IV, and V, respectively. Cyclic Peptides VI-IX. The CAD spectra of the (M H)+ions of the remaining four cyclic peptides studied, VI-IX, are presented in Figures 12-15. The sequence ions are clearly
+
886
ANALYTICAL CHEMISTRY, VOL. 56,NO. 6, MAY 1984
Table VI. MS/MS Fragmentation of cyclo-(Phe-Lys-Pro-Ala), protonation of proline ( ion
loss ofb
protonation of phenylalanine ( 2)' ion
protonation of lysine ( 3 ) a
loss ofb
ion
protonation of alanine ( 4 ) a
protonation of lysine side chain ( 5)'
loss of b
ion
loss o f b
ion
Phe Ala Pro LYS Phe Ala Pro CO, NH,
a4 b4
Pro
1 m n
c4
d, e4 f4
g4
LYS
Phe A1a Pro LYS Phe
loss ofC
CH,NH,
4 r
Site refers t o sites designated in the structure. The designated loss is of the specified residue and all other residues listed above it. For example, ion c, refers to loss of Lys, Phe, and Ala although it has not been established that the losses are sequential. All ions arise from loss of the indicated fragment from the protonated molecular ion. This loss can arise from any of the ring opened (M + H)' ions. amino acid residues which permits the determination of the amino acid sequence of these cyclic peptides. We are currently extending this work to similar compounds such as depsipeptides, cyclopeptides which have peptide side chains, and bicyclic peptides.
LITERATURE CITED 500
1000
+
Figure 14. FAB MS/MS spectrum of the (M H)' ion, m l r 1297, of cyclo-(Val-Pro-Gln), (VIII). Series a,-k, arises from protonatlon of Pro, a2-k, from Gln, and a,-k, from Val. Primes arise from CO loss from corresponding unprimed; ions n, q, r, and t arise from loss of NH,, CO, CONH,, and CH,CONH, from (M H)', respectively.
+
7
rLEU-PRO-GLN-LEU-PRO-GLN-IEU-PRO-GLN-LEU-PRO-GLN,
500
I 000
Figure 15. FAB MS/MS spectrum of the (M -tH)+ Ion, m / z 1353, of
cyclo-(Leu-Pro-Gln)4 (IX). Series a,-k, arises from protonation on Pro, a2-k2 from Leu, and a,-k, from Gln. Primes arlse from CO loss from corresponding unprimed; ions I, m, n, q, and r arise from loss of NH, CO, CONH,, C4H,, and C,H,CONH, from (M -t H)', respectively: ion t from loss of CO and CH,CONH3 from a,.
present in all cases, although a t low relative abundances in some cases, and readily interpretable on the same basis as above. In most cases, protonation on one of the possible amide nitrogens gives rise to a predominant series of ions. Furthermore, we note that almost all the peaks in the CAD spectra are identified as pertaining to the amino acid sequence. Thus, the spectra are encumbered with few extraneous peaks that do not bear on the amino acid sequence.
CONCLUSION The combination of FAB and MS/MS provides not only molecular weight information but also structural information for cyclic peptides. The fragmentation patterns for the cyclic peptides are dominated by ions arising from protonation on nitrogen, followed by cleavage of the bond between the protonated nitrogen and the adjacent carbonyl. The resulting linear acylium ion then fragments further by loss of successive
(1) Barber, M.; Bordoli, R. S.; Sedgwick, R. D.; Tyler, A. N. J. Chem. SOC.,Chem. Commun. 1981, 325. (2) Cooks, R. G.; Glish, G. L. Chem. Eng. News 1981, 4 8 , 54. (3) Crow, F. W.; Tomer, K. 6.; Gross, M. L. Mass Spectrom. Rev. 1983, 2,47-76. (4) Crow, F. W.; Tomer, K. 6.; Gross, M. L.; McCloskey, J. A.; Bergstrom, D. E. Anal. Biochem., in press. (5) Wood, 0. W. Mass Spectrom. Rev. 1982, 1 , 63-102,and references therein. (6) Morris, H. R.; Dell, A.; Judkins, M.; McDowell, R. A.; Panico, M.; Taylor, G. W. Peptides, Proc. Am. Pept. Symp., 7th 1981, 745. (7) Busch, K. L.; Cooks, R. G. Science 1982,278, 247. (8) Rinehart, K. L.,Jr. Science 1982, 218, 254. (9) Gross, M. L.; McCrery, D.; Crow, F.; Tomer, K. 6.; Pope, M. R.; Ciuffetti, L. M.; Knoche, H. W.; Daly, J. M.; Dunkle, L. D. Tetrahedron Lett. 1982, 23, 5381-5384. (IO) Tomer, K. 8.; Crow, F. W.; Gross, M. L. Int. J. Mass Spectrom. Ion Phys. 1983, 4 6 , 375-378. (11) Tomer, K . 6.; Crow, F. W.; Knoche, H. W.; Gross, M. L. Anal. Chem. 1983, 55, 1033-1036. (12) Barber, M.; Bordoli, R. S.; Sedgwick, R. D.; Tyler, A. N. Homed. Mass Spectrom. 1982,9,208-214. (13) Heerma, W.; Kamerling, J. P.; Slotboom, A. J.; van Scharrenburg, G. J. M.; Green, E. N.; Lewls, I. A. S. Blomed. Mass Spectrom. 1983, 10, 13-16. (14) Taylor, L. C. E.; Hazelby, D.; Wakefield, C. J. Int. J. Mass Spectrom. Ion Phys. 1983, 46, 407-410. (15) Desiderio, D. M.; Katakuse, I. Anal. Biochem. 1983, 129, 425-429. (16) Matsuo, T.; Matsudo, H.; Almoto, S.; Shimonlshi, Y.; Higuchi, T.; Maruyama, Y. Int. J. Mass Spectrom. Ion Physlcs 1983, 4 6 , 423-426. (17) Amster, I. J.; Baldwin, M. A,; Cheng, M. T.; Proctor, C. J.; McLafferty, F. W. J . Am. Chem. SOC. 1983, 105, 1654-1655. (18) Panico, M.; Sindona, G.; Uccella, N. J. Am. Chem. SOC.1983, 105, 5607-5810. (19) Wllllams, D. H., Ed. "Specialist Periodical Reports: Mass Spectrometry"; The Chemical Society: London, 1971;Vol. 1, Chapter 4. (20) Anderegg, R. J.; Blemann, K.; Buchi, G.; Cushman, M. J. Am. Chem. SOC. 1976, 98,3365. (21) McLafferty, F. W.; Kornfeld, R.; Haddon, W. F.; Levsen, K.; Sakal, I.; Bente, P. F., 111; Tsal, S. C.; Schuddemaje, H. D. R. J. Am. &ern. SOC. 1973,95,38a6. (22) Kopple, K. D.; Parameswaran Int. J. Pept. Proteln Res. 1983,27, 289-280. (23) Gross, M. L.; Chess, E. K.; Lyon, P. A.; Crow, F. W.; Evans, S.; Tudge, H. Int. J. Mass Spectrom. Ion Phys. 1982, 4 2 , 243.
RECEIVED for review September 19,1983. Accepted February 6, 1984. This work was supported by the Midwest Center for
Mass Spectrometry, a National Science Foundation Regional Instrumentation Facility (Grant No. CHE 82-11164), and NIGMS Grant No. GM 26071.