Anal. Chem. 1989,
416
(6) Ryssel, H.; Ruge. 1. Ionenlmphntation; B. G. Teubner: Stuttgatt, FRG, 1978. (7) Budll, M.; Goser, K.; Stlngeder, G.; Grasserbauer, M.; Potzl, H. Proceedlngs of the 75th Internatlonal Conference on Defects in Semiconductors; August 22-26. 1988, Budapest, Hungary, in press. , c. c. J. ~ l e ~ t r o c h s mSOC. . 1977, 124, 1107. (8) Falr, R. B.; T S ~ IJ. (9) Morehead, F. F.; Lever, R. F. Appl. Phys. Left. 1086, 48, 151. (10) Mukaney, B. J.; Rlchardson, W. B. Appl. Phys. Lett. 1987, 57, 1439. (11) Morehead, F. F.; Stolwijk, N. A,: Meyberg, w.; Gas&, u. Appl. phys. Len. 1983,42, 690.
61,416-422 (12)Jungling, W.; Pichler, P.;Selberherr, S.;Guerrero, E.; Patzl. H. I€€€ Trans. Nectron. Devices 1985,13-32, 156.
RECEIVED for review March 11, 1988. Accepted November 21, 1988. This work has been supported by the Fonds zur Forderung der wissenschaftlichen Forschung (Project No. s 43/ 10).
Plasma Desorption Mass Spectrometric Analysis of Mycobacterial Glycolipids Ian Jardine*#’and Gale Scanlan
Department of Pharmacology, Mayo Clinic, Rochester, Minnesota 55905 Michael McNeil and Patrick J. Brennan
Department of Microbiology, Colorado State University, Fort Collins, Colorado 80523
Mycobacteria are characterked by s p e c k or type-specific glycollprcl antigem. These are generally of the folkwkrg three types: the trehaloeacontahlng, acylated - & a r b (LOS), the C-mycoslde glycopeptlddlplds (GPL), and the phenoilc glycollpkls (POL). To date, convenient m a s spectrometrk analysis of the intact form of these complex glycolipids has proved to be difficult. The successful plasma desorption mass qmctrometrlc analyyslo of intact mycobacterlal glycdlplds of the LOS, QPL, and POL types is now reported, allowlng IocaUon of the acyl resldues and providlng ollgosaccharide sequence and molecular welght intormatlon.
INTRODUCTION Defining the molecular structure of the surface glycosyl appendages of individual mycobacteria allows precise identification of individual bacterial species and subspecies and provides a means of tracing the epidemiology of infections. These mycobacterial antigenic structures are commonly complex glycolipids. Many of the mycobacteria examined to date are characterized by species- and type-specific glycolipid antigens that fall into three broad categories: the trehalosecontaining, acylated lipooligosaccharides (LOS), the Cmycoside glycopeptidolipids (GPL), and the phenolic glycolipids (PGL) (Figure l) (1). It should be noted that the acyl trehalose-containing pyruvylated glycolipid described by Ballou and colleagues (2,3)does not conform to the empirical structure shown in Figure 1;in this case the fatty acyl residues are symmetrically distributed on both glucosyl residues of the terminal trehalose unit. Mass spectrometric analysis of the intact form of many of these glycolipids has proved to be difficult. Following a recent preliminary report ( 4 ) , the present paper describes the successful plasma desorption mass spectrometric (PDMS) analysis of intact mycobacterial antigenic glycolipids of the LOS, GPL, and PGL types. Analysis of the intact glycolipids
* To whom correspondence should be addressed.
ICurrent address: Finnigan MAT Corp., 355 River Oaks Pky., San Jose, CA 95134. 0003-2700/89/0361-0416$01.50/0
as well as their peracetylated derivatives was accomplished at the 1-10 pg level from nitrocellulose-coated PDMS targets. Both the underivatized and derivatized species provide molecular weight information in the positive ion PDMS spectrum. The peracetylated derivatives, in general, also provide considerable oligosaccharide sequence information. The negative ion PDMS spectra provide the fatty acyl composition of the glycolipids for ester-linked fatty acyl groups.
EXPERIMENTAL SECTION Glycolipids were isolated and purified as previously described (5-12). Glycolipids were peracetylated with a 2:l (v/v) mixture of trifluoroacetic anhydride/acetic acid by the method of Bourne et al. (13, 14) as modified for microscale derivatization of glycolipids by Dell and Tiller (15). Glycolipids were dissolved in chloroform/methanol/2-propanol(10:45;45, v/v) to a concentration of approximately 10 pg/hL. Up to 3 p L of solution was then placed on nitrocellulose-coated PDMS target foils prepared as previously described (16,17). After 3-5 min the sample foil was spin-dried at approximately 2000 rpm on a bench-top centrifuge fitted with a sample stage holder and loaded in the PDMS instrument. Plasma desorption mass spectra were obtained on a BIO-ION Nordic (Uppsala, Sweden), BIN-1OK californium-252 plasma desorption time-of-flight mass spectrometer (18, 19) using an accelerating voltage of +20 kV on the sample stage with respect to the transmission grid for positive ions (-16 kV for negative ions) and a flight tube length of 15 cm. The ion source of the mass spectrometer consists of a thin 1O-rc.Cisample of californium-252 emitting simultaneously two almost collinear fission fragments at the rate of approximately 2200 events/s. One of the fission fragments is used to desorb and ionize the glycolipid sample adsorbed on the nitrocellulose surface, while the associated fission fragment from the same fission event is detected, providing a “start” signal for the desorption event. The secondary ions desorbed from the sample target surface are accelerated and then drift in a field-freeregion to a “stop” detector. The signals from “start” and “stop” detectors are processed by fast electronics and then enter a time-to-digital converter (TDC), which has a maximum time resolution of 1 ns/channel and can store up to 64 secondary ions per primary ion. Spectra were generally acquired for approximately 1h. The recorded time-of-flight spectra were converted to mass spectra by using the time centroids for H+and Na+ as calibration peaks. The calculated and determined molecular weights of all ions are D 1989 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 61, NO. 5, MARCH 1, 1989 LIPOOLIGOSACCHARIDES (LOS)
OLIGOUCCHARIOES O +*R
GLYCOPEPTIDOLIPIDS (GPL)
FH' R
-NK-CH-CO-NH-CH-(CO-
NH-CHh
- C 4 4
H,
HC-CHI
OLIGOYCCWIOES-b PHENOLIC GLYCOLIPIDS OLICOUCCHARIDES
(PGL)
PR
PR
e ( C H 2 ) t o -CH-CY-CH-(CH2)4
-0
9""
-CH-C+C& AH3
Flgure 1. Generalized structures of llpooligosaccharides (LOS),glycopeptldolipids (GPL), and phenolic glycolipids (PGL). R represents
various long chain fatty acyl groups. the isotopically averaged molecular weights. All PD mass spectra shown in this paper have been automatically background-subtracted by the computer. Mass spectra of some samples were re-recorded after washing of the sample on the foil with water. Some samples were reanalyzed in similar fashion but at higher resolution by using a BIN-1OK PDMS instrument fitted with a Mamyrin ion reflector (20,21)with a total flight path of 2 m. This instrument is a prototype developed by Bio-Ion Nordic based on a recently reported design (22). For good signal to noise ratios with this less sensitive time-of-flight configuration, samples were analyzed with the reflector system for at least 4 h.
RESULTS AND DISCUSSION
Lipooligosaccharides (LOS). We have previously described the surface antigens of Mycobacterium kansasii as trehalose-containing lipooligosaccharides (LOS),which at the nonreducing end bear a unique amino sugar- and fucosecontaining diglycosyl "epitope", whereas the putative reducing end consists of an a,a-trehalose-containingtetraglucosyl "core" (23, 24). Although fast atom bombardment (FAB) mass spectra of the deacylated LOS produced important molecular weight and sequence information (24), we were unable to generate FAB mass spectra of the intact LOS species, even after trying a number of different FABMS sample matrices.
/OH
Positive ion PDMS analysis, however, of the intact LOS (4) revealed abundant molecular weight associated ions, generally MNa+ species, and it could be inferred that the intact LOSS of M . kansasii generally contain three 2,4-dimethyltetradecanoyl fatty acid esters. The negative ion PDMS spectrum produced an abundant carboxylate anion for the predominant 2,4-dimethyltetradecanoic(C16species) esters. Such analysis has now been carried out on a number of different LOSS of general structure shown in Figure 1 from other mycobacterial species. The molecular weight determination of underivatized M. malmoense LOS ( 5 ) shown in Figure 2 is illustrative. The abundant ion at m / z 2276 is assigned as the MNa+ species. This mass, in fact, was 126 daltons higher than expected from the other analytical data accumulated for this molecule. The negative ion spectrum of the same material revealed abundant ions at m l z 143,326, and 410 (inset in Figure 21, presumed to be the carboxylate anions of c8, C21,and C27fatty acyl lipids. Only the CB1and C27 fatty acyl groups had previously been detected in M. malmoense after alkalinolysis to release fatty acids, conversion to methyl esters, and GC/MS. The volatile c8 fatty acid had apparently been lost in workup. The combination of these three fatty acids with the deduced structure of the M. malmoense oligosaccharide now gave the correct molecular weight of 2253. The location of the fatty acid esters on the carbohydrate chain could not be unequivocally determined from these mass spectra or from other data such as NMR (5). Peracetylation of the M. malmoense LOS, followed by PDMS analysis from nitrocellulose targets, produced the spectrum shown in Figure 3. The molecular weight associated ion (MNa+) has shifted to m / z 3074, an increase of 798, or 19 acetyl groups, which is the expected addition. Extensive carbohydrate sequence information is now, however, revealed in this PDMS spectrum of the peracetylated LOS. The sequence ions at m / z 331,620, 850,1052,1254,1484,1715,and 2003, as well as 1033, unambiguously confirm the sugar sequence and, significantly locate all three fatty acid esters to the terminal glucosyl sugar of the trehalose unit. Similar PDMS analysis of the LOS from M. linda (6) and from M . szulgai (7) has also recently allowed us to complete the structures of these molecules and assign the fatty acids to analogous positions.
P"
1500
M. MALMOENSE LOS
2276
G
g!
417
750-
c
Do
m/z Flgure 2. Partial positive and negative (inset)ion plasma desorption mass spectrum of the
M . malmoense LOS.
418
ANALYTICAL CHEMISTRY, VOL. 61, NO. 5, MARCH 1, 1989
R=
-
-CO(CH2),sCH3
or -CO(CH,),,CH,
or
-CO(CHJ&H,
2003
M. MALMOENSE LOS PERACETYLATED
00
m/z Flgure 3. Positive ion plasma desorption mass spectrum of peracetylated M. malmoense LOS.
Flgure 4. Structure of the GPL from M . avium serovar 4 .
Glycopeptidolipids (GPL). The type-specific antigens of the Mycobacterium auium complex (consisting of a t least 28 serovars of M. avium and M. intracellulare) consist of surface-coated, polar "C-mycoside" glycopeptidolipids (GPL) (Figure 1). The GPLs consist of a tripeptide, H,N-D-Phe-Dallo-Thr-D-Ala-COOH, with an amide-linked fatty acyl function at the NH2 terminus and a carboxy-reduced L-alanine, L-alaninol, a t the COOH terminus, which in turn, is glycosidically linked to an invariant 3,4-di-O-methylrhamnose residue. A haptenic oligosaccharide is attached to the hydroxyl function of the internal ~ - a l l ~ - T hwhich r , in differing among serovars, is responsible for serological specificity ( I , 8, 9). These structures have been given the generic title of polar GPLs or polar C-mycosides since M . avium serovars also contain a set of apolar C-mycoside GPLs that are monoglycosylated at the other position and are not antigenic (1). Although the apolar C-mycosides have been examined by field desorption mass spectrometry (251, mass spectra of the antigenic, multiglycosylated polar GPLs have not to date been reported. Our own attempts at analysis of these materials by FABMS have so far been unsuccessful. In contrast, PDMS analysis of underivatized and peracetylated GPL from nitrocellulose-coated targets has, so far without exception in more than 10 different M . auium serovars, provided molecular weight and considerable structural information (26). The GPL example used to illustrate the application of PDMS to this molecular class is of M . auium serovar 4, a common infection of AIDS patients, and is shown in Figure 4. This specific GPL was isolated in both the native acetylated and in the deacetylated forms, and its oligosaccharide hapten was released as the oligosaccharide alditol by reductive elimination (9). The structure of the deacetylated triglycosyl alditol was established as Q-O-Me-~-Rhap-(al-4)-2-0-Me-
IO
m/z
1
1679
DEACYLATED
m/z Figure 5. (a) Partial positive Ion plasma desorption mass spectrum of the native GPL from M . avium serovar 4 GPL. (b) Partial positive ion plasma desorption mass spectrum of the deacylated GPL from M . avium serovar 4.
-
L-Fuc~-(cY~ 3)-~-Rhap-(crl- 2)-6-deoxytalitol,in which the nonreducing-end disaccharide unit is unique to serovar 4 (9). The only remaining structural question was the number and location of the acetyl groups on this oligosaccharide in the native acetylated molecule, since it was assumed that the
ANALYTICAL CHEMISTRY, VOL. 61, NO. 5, MARCH 1, 1989
MNa' Calculated 1762.03 [C89.H150.N4.029.Na]
-
1761.9
I
1762.8
mlz Flgure 6. Partial positive ion plasma desorption mass spectrum of the native GPL from M . avium.
peptidolipid core of this serovariant would be the same as previously reported for this class of molecule (25). PDMS analysis of the intact and deacetylated GPL of serovar 4 revealed MNa+ ions at m/z 1763 and 1679, respectively (Figure 5). Additional higher mass ions 14 and 28 mass units above the most abundant MNa+ ions in these spectra are presumed to be methylene heterogeneity on the fatty acyl side chain, including a possible methyl ether at the side-chain hydroxyl group (25). The mass difference between these two spectra of 84 Da was accounted for as two acetyl groups. The signal to noise ratio for the deacylated serovar 4 GPL spectrum is superior to that of the native GPL spectrum because more material was available and used for analysis. The molecular weight of 1740 calculated for native serovar 4 GPL from these spectra was 2 lower than that calculated by assuming a monounsaturated, monohydroxylated CaI&,(OH)CO fatty acyl group, as previously reported for these
molecules (25). Since the uncertainty in mass assignment, when the short flight path TOF-PDMS system is used, for glycolipid molecules in general is approximately f 2 Da, we could not be certain of the apparent molecular weight discrepancy of 2. The samples were, therefore, reexamined on a similar TOF-PDMS system fitted with an ion reflector. Figure 6 shows the MNa+ ion for the native serovar 4 GPL obtain on this higher resolution TOF system. The isotopes are now seen to be resolved by this system, but more importantly, the mass accuracy, as confirmed on this instrument by the analysis of a number of other standard materials, is superior to that obtained with the short flight path TOF system and is better than k0.5 Da (see also below). This analysis confirmed that the molecular weight of this GPL was indeed 1763 (average MW) or 1762 (12C isotope), indicating that there was probably another site of unsaturation on the fatty acyl side chain as indicated in Figure 4. Peracetylation of the native GPL on serovar 4 produced the PDMS spectrum shown in Figure 7. This analysis was also conducted on the reflector TOF system to confirm the mass assignments. The MNa+ ion has now shifted, as predicted, by 294 Da or seven acetyl groups to mlz 2057.1 (calculated 2057.4). In addition, a series of fragment ions are apparent as indicated in Figure 7. The interpretation of these fragments is shown in Figure 7, supported by deuterium labeling experiments as discussed below. Peracetylation of the deacetylated GPL and analysis of PDMS (short flight path) produced, as expected, a spectrum essentially identical with that of the native GPL shown in Figure 7. Both the native and the deacetylated GPL's of serovar 4 were then perdeuteroacetylated and analyzed by PDMS (short flight path). The MNa+ shifted appropriately to mlz 2077.6 (calculated 2078.6) or by 21 deuteriums (seven acetyls) for the native GPL and to m / z 2084.0 (calculated 2084.6) or by 27 deuteriums (nine acetyls) for the deacetylated GPL. In addition, the indicated fragment ions (Figure 7) shifted as reported in Table I. This data confirms that the additional unsaturation is, indeed, on the fatty acyl chain (mlz 1582 and its respective shifts by 18 Da or six acetyls and by 24 Da or eight acetyls for the perdeuteroacetylated native and de-
15n2
?17
c_c
c__c
---245
447
G7R
qon
3500
PERACETYLATED M. AVlUM SEROVAR 4 GPL
.-E v)
-
410
1750
c C
m/z Figure 7. Positive Ion plasma desorption mass spectrum of peracetylated GPL from M . avium serovar 4.
420
ANALYT‘ICAL CHEMISTRY, VOL. 61, NO. 5, MARCH 1, 1989
mlz
Figure 8.
Positive and partial negative ion (inset) plasma desorption mass spectrum of M . leprae PGL-I.
4n
Table I. Deuterium Shifts ion
MNa 1582 908 678 447 245 217
3 4
5
deacetylated native per-D-acetylated per-D-acetylated +21 (7 Ac) +18 (6 Ac) +15 (5 Ac) +15 (5 Ac) +9 (3 Ac) +6 (2 Ac) +3 (1 Ac)
+27 +24 +21 +15 (weak) +6 +3
(9 Ac) (8 Ac) (7 Ac) (5 Ac)
(2 Ac) (1 A d
acetylated serovar 4 GPL’s, respectively). Furthermore, the additional 6-Da shift of only the m / z 908 fragment in the perdeuteroacetylated-deacetylated GPL spectrum locates the original acetyl groups of the native GPL to the 6-deoxytalose sugar as shown in Figure 4. That the acetyl groups were probably somewhere on the rhamnose-6deoxytalose nonantigenic portion of the tetrasaccharide rather than on the 4-0-methylrhamnose-2-0-methylfucose disaccharide epitope had been suggested by the fact that both the native and deacetylated serovar 4 GPL’s had equivalent antigenicities (9). Phenolic Glycolipids (PGL). We have previously reported on the structure of the serologically active phenolic glycolipids (PGL) from M. leprae (10-12) and summarized the work of others on the products from M . bouis and M . kansasii (1, 27). The major M . leprae molecule, phenolic glycolipid I, for example, consists of a phenolic phthiocerol “core” to which is glycosidically attached at the phenol group a special specific trisaccharide composed of 3-0-methylrhamnose, 2,3-di-O-methylrhamnose, and 3,6-di-O-methylglucose (see structure in Figure 9). Esterified to the hydroxyl functions of the branched glycolic chain are tetramethyl branched “mycocerosic” acids, either C30,C32, or C34. A mass spectrum of the intact phenolic glycolipid I from M. leprae has not previously been reported, and attempts in our lab to obtain FAB mass spectra have so far met with little success. Positive and negative ion PDMS analysis of underivatized unfractionated PGL-I from M . leprae from nitrocellulose-coated targets gave the spectra shown in Figure 8. The negative ion spectrum clearly reveals the carboxylic anions of the expected fatty acyl components of the M. leprae PGL-I, C3,, at m / z 452, C32at m / z 480, and C,, a t m / z 508. The ratios of the ion intensities of these fatty acids are very similar to those determined by GC/MS of the methylated fatty acids (11). The positive ion PDMS spectrum reveals a complex cluster of ions at approximately m / z 20W-2200. The main ions of this group are designated as shown as the C-30,34, the C-32,34, and the C-34,34 species. The ratios of the in-
> TME
Figure 9. HPLC separation of M . leprae POL-I on a CIS column using a 25% CHCI,/CH,OH mobile phase and monnoring UV absorption at 280 nm.
tensities of these ions are approximately what is expected, given that the C30,C32, and C, fatty acids are in the ratios revealed in the negative ion spectrum and that only two fatty acids per PGL molecule are present. To confirm the mass assignment of these MNa+ ions, the sample was rerun on the PDMS reflector TOF system. The isotopically averaged masses determined for the three major ions were 2049.8, 2077.7, and 2106.3 (calculated 2050.1, 2078.2, and 2105.9), respectively, or an apparent mass accuracy of f0.5 Da. Additional clusters of ions are observed at approximately 1500-1600 daltons and at approximately 990-1050 daltons. The former higher mass cluster is assigned as resulting from loss of one fatty acid group from the intact PGL molecule, while the latter results from subsequent loss of the trisaccharide moiety. The pattern or ratio of the ions in these clusters now reflects only one fatty acyl appendage instead of two, and so conforms approximately to the inherent total ratio of the CW,C32, and C3, fatty acyl groups. Also present in the spectra are abundant lower mass ions at m / z 191,365, and 526, resulting from cleavage of the sugar chain as shown in Figure 9. After peracetylation and PDMS analysis, the molecular weight associated ions shift by the expected addition of 126 daltons or three acetyl groups. The sugar fragments also shift appropriately (mlz 191 275; m / z 365 447; m / z 526 652) (spectrum not shown). Clearly the M . leprae PGL is a complex mixture, much more complex than was once thought (11). In an attempt to clarify this complexity, fractionation of the intact PGL-I species has been attempted. The material separates into a number of components by reversed-phase HPLC as shown in Figure 9. All of these fractions have been analyzed by PDMS, and the spectrum of HPLC peak 3 is shown in Figure 10 for illustration. This HPLC peak clearly consists predominantly of the ((2-30, 30) species. Similarly, peaks 4 and 5 consist almost exclusively of the (C-32, 32) species and the (C-34, 34) species, respectively. This example illustrates a
-
-
-
ANALYTICAL CHEMISTRY, VOL. 61, NO. 5, MARCH 1, 1989 (995)
526
421
~
\
\
0
M LEPRAE PHENOLIC QLVCOLIPID I Y W 1971
m/z Flgure 10. Partial positive ion plasma desorptlon mass spectrum of HPLC fraction 3 of M. leprae POL-I.
major strength of this PDMS approach to the analysis of mycobacterial glycolipids: the ability to rapidly and conveniently monitor chromatographic fractionation of mixtures of these important and complex compounds. In addition to M . leprae PGL-I, we have also successfully analyzed by PDMS the PGL from M. bouis, known as mycoside B (28) or "M. bouis identifying lipid" (29),on account of its use in differentiating M . bouis from all other mycobacteria, and the PGL from M . kansasii, once known as mycoside A but now called Phe-G1-KI. The molecular weight and fragmentation determined for both of these molecules agree with the structures recently published (30,31). In the latter work, it was reported but not shown that a FAB mass spectrum of the PGL from M . kansasii was obtained by using the amphiphilic liquid triethylene glycol mono-n-butyl ether as the FABMS matrix. In conclusion, we have demonstrated that complex bacterial glycolipids can be readily and reliably mass-analyzed by plasma desorption mass spectrometry. Samples are conveniently prepared on nitrocellulose-coated PDMS target foils a t the 1-1O-lg level and can be washed if necessary, before analysis. In the positive ion PDMS mode, underivatized glycolipids produce molecular weight associated ions, usually MNa+, with occasionally some useful fragment ions. In the negative ion mode, the fatty acyl composition of the glycolipid is determined for ester-linked groups. Peracetylation with its associated molecular weight shift allows the determination of hydroxyl groups after positive ion PDMS analysis, and in addition, considerable oligosaccharide sequence information is usually obtained. The short flight path PDMS system suffers from low resolution and a resulting lack of mass accuracy (*2 daltons for the glycolipid molecules studied). This situation is improved substantially by the incorporation of a Mamyrin ion reflector in the PDMS-TOF system. Tandem mass spectrometry (MS/MS) experiments have been demonstrated using such a unit (21)and may confer additional advantages for complex glycolipid analyses. Recent advances in californium-252 PDMS using Fourier transform ion cyclotron resonance (FTMS) systems (32,33)portend additional advances in mass
accuracy and MS/MS capability.
ACKNOWLEDGMENT We are indebted to B. Sundqvist, W. Ens, and P. Hikansson of the Tandem Accelerator Laboratory of Uppsala University, and to I. Kamensky of Bio-Ion, Nordic, Sweden, for allowing us to use their newly developed PDMS reflector TOF system. LITERATURE CITED (1) Brennan, P. J. I n M/crob/alL@ids;Ratledge, C., Wllklnson, S. 6.. Eds.; Academic Press: London, 1988; Vol. I, pp 203-298. (2) Saadat, S.; Ballou, C. E. J. 8/01. Chem. 1983, 258, 1813-1818. (3) Kamisango, K. I.; Saadat. S.; Dell, A.; Balbu, C. E. 1985, 260,
4117-4121. (4) Jardlne. I.; Hunter, S. W.; Brennan, P. J.; McNeal, C. J.; Macfarlane, R. D. 8iomed. Envlron. Mass Spectrom. 1988, 13, 273-276. (5) McNeil, M.; Tsang, A. Y.; McClatchy, J. K.; Stewart, C.; Jardlne, I.; Brennan, P. J. J. Becterbl. 1987, 189, 3312-3320. (6) Camphausen. R. T.; McNell, M.; Jardlne. I.; Brennan, P. J. J. Becterio/. 1987, 169, 5473-5480. (7) Hunter, S. W.; Barr, V. L.; McNeil, M.; Jardlne, 1.; Brennan, P. J. B b chemistry 1988, 27. 1549-1556. (8) Camphausen, R. T.; Jones, R. L.; Brennan, P. J. J. Becterbl. 1988, 768, 660-667. (9) McNell, M.; Tsang, A. Y.; Brennan, P. J. J. 8/01. Chem. 1987, 262, 2630-2635. (IO) Hunter, S . W.: Brennan. P. J. J. Bacterbl. 1981, 147, 728-735. (11) Hunter, S. W.; Fugiwara, T.; Brennan, P. J. J. 8iol. Chem. 1882, 257, 15072-1 5078. (12) Gaylord, H.; Brennan, P. J. Annu. Rev. Mbobbl. 1987, 4 7 , 645-675. (13) Bourne. E. J.; Tatlow. J. C.; Stacev. M.: Tedder, J. M. J . Chem. Soc. 1949, 2676-2879. (14) Bourne, E. J.; Tatlow, J. C.; Worrall, R. J. Chem. SOC. 1957, 3 - 15-318 . - - . -. (15) Dell, A.; Tiller, P. R. Blochem. Siophys. Res. Commun. 1988, 135, 1126-1134. (16) Jonsson, G. P.; Hedln, A. B.; Hakansson, P. L.; Sundqvist, B. V. R.; Save, B. G. S.; Nlelson, P. F.; Roepstorff, P.; Johansson, E.-€.; Kamensky, I.; Llndberg. M. S. L. Anal. Chem. 1988, 5 8 , 1084-87. (17) Kamensky, I.: Craig, A. G. Anal. Instrum. 1987* 16, 71-91. (18) Macfarlane, R. D.; Torgerson, D. F. Science (Washlngton, D . C . ) 1978, 197, 920-925. (19) Sundqvist, 8.; Kamensky, I.; Hakansson, J.; Kjellberg, J.; Salehpour, M.; Wddlyasekera. S.; Fohlman, J.; Peterson, P. A,; Roepstorff, P. Anal. Chem. 1984, 1 7 , 242-257. (20) Mamyrin, 8. A.; Karataev, V. I.; Shmikk. D. V.; Zagulln, V. A. Sov. Phys .-JETP (Engl. Trans/.)1973, 3 7 , 45. (21) Della Negra, S.; LeBeyec, Y. Anal. Chem. 1985, 57. 2035-2040. (22) Tang, X.; Beavis, R.; Ens, W.; Lafortune, F.; Schueler, B.; Standlng. K. G. Int. J . Mass Spectrom. I o n Processes W88, 85, 43-67. (23) Hunter, S. W.; Murphy, R. C.; Clay, K.; Goren, M. 8.; Brennan, P. J. J. Bioi. Chem. 1983, 258, 10481-10487.
Anal. Chem. 1989, 61, 422-428
422
(24) Hunter, S. W.; Jardlne, I.; Yanagihara. D. Y.; Brennan, P. J. Biochemi S e 1981, 2 4 , 2798-2805. (25) Daffe, M.; Laneele, M. A.; Puzo, G. Biochim. Siophys. Acta 1983, 751, 439-443. (26) Bozlc, C.; McNell. M.; Chatterjee, D.; Jardlne. I.; Brennan, P. J. J . Bid. Chem. 19889263, 14984-14991. (27) Brennan. P. J. Rev. Infect. Dis., In press. (28) Demarteau-Glnsburg, H.; Lederer, E. Blochim. Biophys. Acta 1983, 70 442-45 1. (29) Jarnagin. J. L.; Brennan, P. J.; Harris, S. K. Am. J . Vet. Res. 1983, 4 4 , 1920-1922. (30) Fourni6, J.J.: Rlvlere, M.; Puzo, G. J . Bid. Chem. 1987, 262, 3174-3 179. (31) FournE, J.J.; Rlvlere, M.; Papa, F.; Puzo, G. J . Biol. Chem. 1987, 262, 3180-3184. (32) Tabet. J. C.: Rapln, J.; Poreti, M.; Ggumann, T. Chimia 1986, 4 0 ,
169-171.
(33) Loo, J. A.; Willmms, E. R.; Amster, I. J.; Furlong, J. J. P.; Wang. B. H.; McLafferty,F. W.; Chait, B. T.; Field, F. H. Anal. Chem. 1987, 59, 1880-1882.
RECEIVEDfor review July 29,1988. Accepted November 10, 1988. This work was presented at the 35th and 36th Annual Conferences of the American Society for Mass Spectrometry and Allied Topics, May 24-29,1987, Denver, CO, and June 5-10,1988, San Francisco, CA, respectively. The research was supported by the National Institutes of Health (Grants AI18357 and AI-52582 to P.J.B. and GM-32928 and RR-02682 to I.J.).
Thin-Layer Chromatography with Urea-Solubilized ,8-Cyclodextrin Mobile Phases Willie
L. Hinze,* Daniel Y.Pharr,’ Zheng Sheng FU? a n d Walter G. Burkert3
Department of Chemistry, Laboratory of Analytical Micellar Chemistry, Wake Forest University, P.O. Box 7486, Winston-Salem, North Carolina 27109
Aqueous solutlons of urea-solubilized &cyclodextrln ( 8 4 0 ) are characterized and evaluated as a novel planar chromatographlc moblle phase. The resolutlon of Isomeric ortho-, meta-, and para-disubstituted benzenes as well as that of pesticide, polycycllc aromatlc hydrocarbon, and drug test mlxes on a polyamide stationary phase demonstrates the viaWIty of thls urea-sdublllzed B-CD medium as a thln-layer chromatographlc (TLC) moblle phase. I n addltlon, some chromatographlc data was obtalned on reversed-phase C-12 and C-18 stationary phases. The pertinent chromatographic parameters were determined and contrasted to those prevlously obtalned with CD mobile phases. The Armstrong pseudophase theory was applied and found to adequately describe the retentlon behavior exhlbited by the test solutes as a function of &CD concentram In the mobile phase. The effect of urea concentration and presence of alcohol additlves on retention and separation was atso assessed. Lastly, a brlef prospectus on the usefulness of solubilized 8-CD mobile phases in liquid chromatographlc separatlons is presented.
INTRODUCTION Cyclodextrins (CD) are homologous series of cyclic oligosaccharides consisting of D-(+)-glucopyranose units. They are well-known for their ability to form inclusion complexes with a variety of solute molecules in both the liquid and solid state (1,2). Due to their selective complexing ability, there has been considerable interest in their utilization as the active stationary-phase component in gas and liquid chromatography and as mobilephase additives in LC. Several excellent reviews summarize their use in these and related separation science applications (3-8). Although CD mobile phases have been widely employed in high-performance liquid chromatography Present address: Department of Chemistry, Virginia Military Institute, Lexington, VA 24450. *Present address: Chemistry Department, Northwestern Teacher’s Universit , Lanzhou, Gansu, People’s Republic of China. Present aiddress: BASF Corp., 1701 Westinghouse Blvd., Charlotte. NC 28241.
(HPLC) separations (9-1 7),there have been relatively few applications involving thin-layer chromatography (TLC). In fact, reversed-phase (RP) TLC has been almost exclusively employed to study the CD inclusion complexes formed by derivatives of a variety of organic compounds (18-23). In all but one of those reports, polymeric 0-CD was utilized as the mobile-phase additive. Aside from our initial reports concerning the TLC separation of positional isomeric disubstituted benzenes with aqueous a-CD mobile and polyamide stationary phases (4,24, W), there is apparently only one other instance in which a TLC separation has been reported (26). This sparsity of TLC applications stems in large measure from the limited solubility of the CDs, particularly P-CD, in aqueous media. This solubility problem has also created difficulties with their utilization as HPLC mobile phases (14, 17). In addition, the water-soluble P-CD derivatives previously used in TLC are reportedly more costly compared to that of the parent P-CD itself (27). We have recently reported that appreciable quantities of CDs can be solubilized in aqueous media via use of urea or base as solubilization agents (28). In this paper, we report on the results of our systematic study of the use of ureasolubilized 0-CD mobile phases for the planar chromatographic separation of a variety of isomeric disubstituted benzenes on polyamide as well as C-12 and C-18 R P stationary phases. The dependence of the solute retardation factors upon the @-CD concentration is rationalized in terms of existing pseudophase theory. The chromatographic behavior exhibited by selected test solutes upon variation of the urea composition in the mobile phase was examined. In addition, secondary modification of the urea/@-CDmobile-phase system with alcohols was briefly studied. The separation of some pesticides and drugs is also reported. The results clearly demonstrate the viability of using such urea-solubilized p-CD mobile phases in TLC separations. EXPERIMENTAL SECTION Materials. fl-Cyclodextrin (Advanced Separation Technologies, Whippany, NJ, and Sigma Chemical Co., St. Louis, MO), urea (A Grade, CalBiochem, San Diego, CA), ethanol (USPgrade, AAPER Alcohol & Chemical Go., Shellyville, KY), tert-butyl alcohol (Aldrich Chemical Co., Milwaukee, WI), and acetone
0003-2700/89/0361-0422$01.50/00 1989 American Chemical Society
