Four-sector tandem mass spectrometric analysis of complex mixtures

Duncan K. Bryant, Ronald C. Orlando, Catherine. Fenselau, Raymond C. Sowder, .... Simon H.J. Brown , Todd W. Mitchell , Stephen J. Blanksby. Biochimic...
1 downloads 0 Views 623KB Size
1110

Anal. Chem. 1991, 63, 1110-1114

Four-Sector Tandem Mass Spectrometric Analysis of Complex Mixtures of Phosphatidylcholines Present in a Human Immunodeficiency Virus Preparation Duncan K. Bryant,* Ronald C. Orlando, and Catherine Fenselau Department of Chemistry and Biochemistry, University of Maryland Baltimore County, Baltimore, Maryland 21228 Raymond C. Sowder and Louis E. Henderson PRIIDyn Corp., Frederick Cancer Research and Development Center, Frederick, Maryland 21 701

A number of phosphat#ylcholkrer have been Wated from an HIV-l/MN preparation by reversed-phase high-petfmance liquid chromatography (HPLC) and analyzed by fast atom b o m b a r m mass q e c h m b y (FABMS), FABMSM, and FABMS/MS/MS In both paskive- and negative-lon modes. Nogath" FABMS/MS wtth colldons was used to identify the length of the acyl groups and the degree of saturation, as well as their podtion on the glyceride group. FABMS/MS in the podtlvdon mock was used to identity the polar head group. Negatfve-bn FABMS/MS/MS was used to locate posttlonr of double bonds In acyl groups. We flnd that four-sector tandem maw spectrometry with high-energy coliislonal activation provides qualitative analysis of viral phosphatidyl llpids in conrldarabk dotali, as well as semiquantitative information. Approximate quantitation of the phorphatldylchdlru content of the HIV-l/MN preparath by measwing rdatlve peak heights of molecular ions In FABMS reveals an array of phosphatldylchollnesconsistent wtth that found In human erythrocytes, indicatlng the likely source of lipids In the viral membrane to be the host ceH membrane.

INTRODUCTION Fast atom bombardment (FAB) ionization has proven to be the optimal technique for mass spectral analysis of choline-containingphospholipids (1-8). In the positiveion mode, molecular ion stability is enhanced by the presence of the tetraalkylated ammonium cation, while in the negative-ion mode formation of a phosphoryl anion contributes to the stability of the negatively charged fragments. When combined with tandem mass spectrometry, not only can complex mixtures be analyzed, but also much structural information can be obtained (8). Positive-ion FABMS/MS can be used to identify the polar head group. Negative-ion FABMS/MS is useful in elucidating the structure and position of the acyl groups. The four-sector instrument used here provides high-energy collisions, unit resolution in both MS-1 and MS-2, while maintaining good ion transmiasion (9),and provides the capability to perform MS/MS/MS experiments. The higher resolution of maw selection for a four-sector maw spectrometer compared to a two-sector instrument is essential for studying phospholipids, many of which differ by only two mass units. In this study we have examined the phosphatidylcholine content of a purified preparation of the human immunodeficiency virus type I (HIV-l), the etiologic agent of the acquired immune deficiency syndrome (AIDS) (10-12). The virus is an enveloped retrovirus and is composed of a nucleoprotein core surrounded by a membrane composed of lipid and protein. Retroviruses assemble at plasma membranes of the host cell and are then released by a budding process. There is much evidence to support that much if not all of the viral lipid is obtained directly from the host cell

membrane in the budding process (13). Glycero-type phospholipids, along with protein, are the main constituents of biomembranes. The most abundant glycero-type phospholipid in mammalian cell membranes is phosphatidylcholine (131,which can account for up to 49.5% of the total phospholipid content and up to 30% of the total lipid content in baby hamster cells, for example (14). Acyl chain length has been found to vary from 10 to 24 carbon atoms (15) and up to 6 unsaturation centers per acyl group have been reported (16).

EXPERIMENTAL SECTION Virus Growth and Purification. HIV-1 (MN strain) virus was grown in H9 cells. Harvested cell culture fluid was initially purified by continuous-flow ultracentrifugation in a Pharmacia EN1 Diagnostics, Inc. Mark I1 K ultracentrifuge at 35 000 rpm, utilizing a 2040% sucrose in TNE (10 mM Tris-HC1, pH 7.2; 100 mM NaCI; 1mM EDTA) density gradient. The gradient was fractionated and fractions in the 1.13-1.17 g cm4 range were pooled and isopycnidy banded in a Beckman Instruments Ti-15 rotor at 30000 rpm, utilizing a 2540% sucrose in TNE gradient. Fractions in the 1.13-1.17 g cm-s range were pooled and pelleted in a Beckman Instruments Type 45Ti rotor at 30000 rpm for 60 min. Isolation of Phosphatidylcholinesfrom HIV-l/MN Virus. HIV-1/MN (158 mg) was disrupted in guanidine hydrochloride (33 mL, 8 M) and trifluoroacetic acid (100 gL; 20%) and injected onto an LKB high-performance liquid chromatograph. Reversed-phase HPLC was performed on a Waters Associates g Bondapac C18 column (19 mm by 150 mm). Mobile phases consisted of A = 0.05% TFA in water, B = 0.05% TFA in acetonitrile, and C = 0.05% TFA in 1-propanol. A gradient of 0-30% B in A was then run over 20 h followed by 100% B for 3 h at a flow rate of 2.6 mL min-' at room temperature. A gradient of 0-100% C in A was then run over 3 h at 5 mL mi& and 50 "C. Phosphatidylcholines were collected between 135 and 143 min after commencement of the heated propanol gradient. Fast Atom Bombardment Mass Spectrometry. FABMS of viral lipids was performed on the first two sectors (EB) of a JEOL HX110/HX110 mass spectrometer (Tokyo,Japan) at an accelerating voltage of 10 kV, with 100-Hz filtering, and a resolution of 1:lOOO. The JEOL FAB gun was operated at 6 kV with xenon as the FAB gas. Spectra were recorded with a JEOL DA5000 data system. Tandem Mass Spectrometry. FABMSIMS was carried out by using all four sectors (EEEB) on the JEOL HXllO/HXl10 instrument. Precursor ions were mass-selected on MS1 at a resolution of 1:1000 and linked BIE scans were made on MS2. Collision-induced dissociations (CID) occurred in the third field-free region, with helium used as the collision gas at a pressure sufficient to attenuate the precursor ion beam by 80%. The collision cell was floated at 4 kV. The CID spectra were recorded at a resolution of 1:1000 and with 100-Hz filtering. For FABMS/MS/MS (17)the precursor ions were formed in the ion source and underwent CID in the first field-free region (between the ion source and the first ESA (ESA = electrostatic analyzer)). The first-generation fragment ions were then energyand momentum-selectedand passed into a collision cell located

0003-2700/B1/0363-1110$02.50/00 19Bl American Chemical Sockty

ANALYTICAL CHEMISTRY, VOL. 63,NO. 11, JUNE 1, ?99l

-1

1111

I

R0

I

mh

li

Figure 3. Posltive-bn FABMS of phosphawylchoHnes from HIV-l/MN

virus.

1. Reversedphase HPLC chromatogram of HIV-1IMN material eluted by heated propanol gradient. (See Experimental Section.) Wavelength = 206 nm. The region where phosphatidylcholineswere collected is Indicated.

R'yoL fl +

0-P-0-

N(CH3)3

0R" J O

Flgure 2. Structure of phosphatidylcholines. in the third field-free region (between the first magnetic sector and the second ESA). The second-generationproduct ions were formed by CID and detected by scanning both the second ESA and magnet in a linked scan with a constant B / E ratio. For positive-ion FABMS and FABMS/MS 0.1 c ~ gof lipid in DMF (1 pL), 0.1% trifluoroaceticacid (0.5 gL), and 3-nitrobenzyl alcohol (0.5 pL) was used for analysis. In negative-ion mode FABMS, FABMS/MS, and FABMS/MS/MS 100 ng of lipid in DMF (1 pL) and triethanolamine (1 pL) was used for analysis.

RESULTS AND DISCUSSION Reversed-Phase HPLC Extraction of HIV-1/MN Phosphatidylcholines. Reversed-phase HPLC separation of the components of the HIV-1/MN virus was carried out in two stages. An acetonitrile gradient was initially employed to elute the proteins. This was followed by a heated propanol gradient to elute the lipid material, the chromatogram of which can be seen in Figure 1. Preliminary mass spectral analysis of HPLC column fractions indicated that phosphatidylcholines were all coeluted between 135 and 143 min. The general structure of phosphatidylcholines can be seen in Figure 2. Positive-Ion FABMS of the Phosphatidylcholine Mixture. The positive-ion mode FAB mass spectrum of the protonated molecule region of the phosphatidylcholine mixture is shown in Figure 3. A large number of peaks representing protonated molecules ([M + HI+) of phosphatidylcholines with differing acyl groups can be seen. FABMS spectra of single phosphatidylcholine species can give useful information on the nature of the polar head group and of the range of acyl groups (I, 18). However, to characterize individual phosphatidylcholines in the mixture shown in Figure 3, tandem mass spectrometry of the protonated molecules is required (see below). Negative-Ion FABMS of the Phosphatidylcholine Mixture. The negative-ion mode FAB mass spectrum of the molecular ion region of the phosphatidylcholine mixture is shown in Figure 4. The spectrum does not contain molecular anions ([M - HI-) from phosphatidylcholines, but does reveal the trio of characteristic ions, [M - 15]-, [M - 60]-, and [M - MI-, which can be assigned to [M - CH3]-, [M - HN(CHd3]-, and [M - CH2=CHN(CH3),]-, respectively (8, 19). Nega-

4. m t i v e k n F A N S of phosphatklylchdines trOm HIV-i/MN

virus.

1

IIM

734

I

Flgm 5. Poslthrekn FABMSIMS of d i p a m - .

tive-ion FABMS spectra of commercial phosphatidylcholine samples in the literature show the [M - 151- ion to be the most intense of the three molecular ions and the only one to have an even mass number (1,19). Thus most of the abundant ions in Figure 4 in the mass range m / z 600-850 are probably of this type. Like positive-ion FAB mass spectra of phosphatidylcholines, negative-ion FAB mass spectra contain considerable structural information (I). They contain a deacylated ion region, a fatty acid residue region, and a polar head group region. However, these regions can only be correctly interpreted for single phosphatidylcholine species, so to identify the structure of the molecular ions seen in Figure 4, negative-ion mode FABMS/MS is required. PositiveJon FABMS/MS of the Phosphatidylcholines. A typical positive-ion mode FABMS/MS spectrum of a phosphatidylcholine protonated molecule is shown in Figure 5. The CID spectrum is of the protonated molecule [M + H]+ = m / z 734.9, which h a been assigned the structure 1,2-dipalmitoylglycerophosphocholine.The most abundant ions in the spectrum are from the polar head group with ions a t m / z 184 and 224, corresponding to [(HO),P(O)OCH2CH2N(CH3)3]+and [CH,=CHCH,OP(O)(OH)OCH2CH2N(CH3)3]+, respectively. The ion at m / z 550 corresponds to [M - phosphorylcholine]+. The deacylated ion region is comparatively weak and consists of two ion pairs, [M - RCH==C=O]+ and [MH - RCH,COOH]+ (20). We find that for the higher molecular weight phosphatidylcholines,

1112

ANALYTICAL CHEMISTRY, VOL. 63,NO. 11, JUNE 1, 1991 I84

810

I

Flgure 6. Positivaion FABMSIMS of l-eicosatrienoyC2-stearoyC glycerophosphocholine. I

255

I

I

(I

Figure 8. Negative-ion FABMS/MS of l-paimitoyi-2octadecadienoylgiycerophosphocholine and 1-hexadecenoyl-2-

octadecenoylglycerophosphocholine.

Table I. Distribution of Carboxylate Groups in Phosphatidylcholinesof HIV-I/MN by FABMS and FABMS/MS abundance by positive-ion [M + H]+ FABMS/ %

m/z

Flgure 7. Negative-ion FABMS/MS of 1-paimttoyC2-octadecenoyigiycerophosphocholine.

720.8 732.8 734.9 758.9 760.9

16.4 18.9 79.9 56.7 100

abundance of carboxylate anions by negative-ion FABMS / MS/ % 1-position 2-position 56 81 100 42 63 83 16 18 13 52 23 12 98 78 100 21 12 3 36 18 63 72 21 32 78

100 100 100 100 76 100 29

product ions in this region become increasingly weaker and, 772.8 13.7 in some cases, nonexistent. This is to be expected, as colli100 39 sionally induced fragmentation is slowed in larger protonated 774.8 21.8 100 molecules. Figure 6 shows a FABMS/MS spectrum of the 39 protonated molecule m/z 810.9, which has been assigned the 17 structure l-eicosatrienoyl-2-stearoylglycerophosphocholine, 784.9 27.9 100 and it can be seen that the deacylated ion region is devoid 75 of abundant ions. The fatty acids in this phosphatidylcholine 786.8 58.4 100 33 were assigned from its negative-ion FABMS/MS spectrum. 21 The high-energy CID spectrum of dipalmitoylglycero9 phosphocholine (Figure 5) also contains some charge-remote 788.9 10.7 100 fragmentation in the form of sequential methylene losses from 22 the hydrocarbon chains of the acyl groups. From the work 810.9 42.4 100 of Gross et al. on high-energy CID of carboxylate anions (211, 812.8 12.6 100 34 this information could be used to identify the positions of 834.9 17.9 100 double bonds in the fatty acid groups, though in practice it 836.9 13.1 100 would not be possible to distinguish on which fatty acid group the olefinic bond(s) reside. The high-energy CID spectrum OThe abundance of the most intense carboxylate anion in each FABMS/MS spectrum is listed as 100%. of dipalmitoylglycerophosphocholinehas been reported previously (221, but our spectrum shows much higher resolution and a greater abundance of low-mass ions due to the fact that (Clal). The intensity of the carboxylate anion originating from the collision cell in our experiment was floated at 4 kV. These the 2-position of the phosphatidylcholine has been shown to remote-site fragmentation ions become less abundant as the be greater than that from the 1-position (1,8,20,23). Hence acyl chain length increases, as shown in the positive-ion in Figure 7 the palmitoyl group is situated on the 1-position FABMS/MS spectrum of 1-eicosatrienoyl-2-stearoylglycero- and the octadecenoyl group is on the 2-position of the phosphocholine (Figure 6), which contains little or no rephosphatidylcholine. mote-site fragmentation. Some of the peaks in the molecular ion region observed in Negative-Ion FABMS/MS of the Phosphatidylnegative-ion FABMS of the phosphatidylcholine mixture (Figure 4) correspond to [M - 151- from more than one cholines. A typical negative-ion FABMS/MS spectrum of a phosphatidylcholine [M - 151- ion is shown in Figure 7. The phosphatidylcholine. FABMSIMS of one of these peaks reCID spectrum is of 1-palmitoyl-2-octadecenoylglycerovealed a more complex fatty acid residue region. Figure 8 phosphocholine. The polar head group is characterized by shows the fatty acid region of a CID spectrum of the [M - 151two abundant ions at m/z 168 [O=P(OH)(0)OCHzCHzN(Cion mlz 742. Four carboxylate anions are observed, m/z 253 H3)J and m / z 208 [CH2=CHCH20P(0)OCHzCHz(CH,)z]-,( C I m~/ z 255 ( C 1 d , mlz 279 (Cle:d, and mlz 281 ( C d . which correspond to phosphorylcholine and glyceroKnowing that the molecular weights for the phosphatidylphosphocholine, respectively. Low-abundance deacylation ions cholines at [M - 151- = m / z 742 are the same, the hexadea t m / z 432 and 488 can be seen in the spectrum (Figure 71, cenoyl group must be on the same molecule as the octadewhich correspond to the ions [M - 15 - RCOI-. The most cenoyl group and the palmitoyl on the same phosphatidylimportant ions in the spectrum are those due to the carboxcholine as the octadecadienoylgroup. One can go further and ylate anions, [RCOOI-, e.g., m / z 255 (Cls:o), and m / z 281 state that most of the ClS acyl groups are situated on the

ANALYTICAL CHEMISTRY, VOL. 63, NO. 11, JUNE 1, 1991

1113

Table 11. Fatty Acids (as % Total by Weight) in HIV-l/MNVirus Phosphatidylcholines fatty acids

150 160 161 17:O 171 180 181 182 183 191 192 200 201 202 203 204 205 22:5 22:6

tissue

HIV-1/MNvirus human erythrocytesa bovine plasmab human plasmac a Reference

26.

1.7 35.6 5.3 31.2 19 1 26.1

Reference

27.

1.7 0.5

e Reference

1.0 10.8 25.7 5.5 0.7 11.8 18.9 22.8 29 16 23 13.6 12.9 24.7

0.6

0.3

0.2 0.2

0.6 0.5

1.8 1.1

0.1

0.5 0.7

1

5.9 1.9 4 3.5

6.7 7

0.3

1.3 0.8

1.8 2.1

1.1

4.1

28.

251

I1 I

m:z

Flgue Q. Negathreion FABMSIMSIMSof the OCtedec-Q-enoate anion from 1-palmitoyl-2sctadec-9-enoylglycerophosphochollne.

1-position of the phosphatidylcholine. Table I shows the fatty acid groups as elucidated from negative-ion FABMS/MS of all the peaks contributed by phosphatidylcholine [M - 151anions. Also listed in Table I is the abundance of the respective [M + H]+ ions. Many of the [M - 151- ions are little more intense than the background chemical noise; hence the respective abundances of the much more intense [M + H]+ ions are listed. Table I also gives relative abundances of the carboxylate anions observed in the negative-ion mode CID spectra of the [M - 151- ions. The most abundant carboxylate anion in each spectrum is given as 100%. Negative-Ion FABMS/MS/MS of the Phosphatidylcholines. When CID is performed in the first field-free region of the four-sector mass spectrometer, MS-1 can be set to transmit product ions into the second collision cell in the third field-free region for further CID analysis (17,241. In this way phosphatidylcholine [M - HI-ions have been collisionally activated and individual carboxylate anions have been transmitted to the second collision chamber for further collisional activation. With this method high-energy CID spectra can be obtained of the carboxylate groups, which can be used to locate the position of carbon-carbon olefinic bonds. Lack of sample restricted us to the analysis of two unsaturated carboxylate groups from two phosphatidylcholines in the mixture. Figure 9 shows the high-energy CID spectrum of the octadecenoate anion obtained by CID of l-palmitoyl-2octadecenoylglycerophosphocholine. This spectrum is very similar to that obtained by negative-ion high-energy CID of oleic acid (211, showing that the olefinic bond is situated at the 9-position. The product ions observed in Figure 9 are all charge-remote fragments. Using this method, we find that the olefinic bond in the eicosenoate group of l-eicosenoyl-2hexadecenoylglycerophosphocholine is in the 11-position. Fatty Acid Distribution of HIV-1 Phosphatidylcholines. From the data listed in Table I an approximate distribution of the fatty acids of phosphatidylcholines in HIV-1 virus can be calculated. Table I1 shows the distribution of fatty acids as a percentage total by weight in HIV-l/MN viral phosphatidylcholines. It is assumed that all the viral phosphatidylcholine content is isolated by reversed-phase HPLC and that they all give similar abundances of molecular ions per mole, which is a reasonable assumption to make as they are all chemically very similar (2). Where the molecular ions

of phosphatidylcholines coincide, their relative ratios are calculated from the ratios of carboxylate anions present in each phosphatidylcholine, values obtained from negative-ion mode CID spectra of the relevant molecular ions. The positions of the acyl groups for the two phosphatidylcholines containing highly unsaturated acyl groups (Table I) should be regarded as somewhat tentative, as new data by Huang et al. suggest that the [M - 151- ion can give erroneous data regarding the position of highly unsaturated acyl groups (25). Also listed in Table I1 are the literature values for fatty acids, by weight, of phosphatidylcholines from a selection of mammalian tissues, namely human erythrocytes (26) and bovine (27) and human plasma (28). It can be seen that there is considerable similarity between the distribution of fatty acids in HIV-1/MN viral phosphatidylcholines and of the phosphatidylcholines of human blood cells and plasma. Abundant evidence supports the assumption that much if not all of the viral lipid is obtained directly from the cell membrane in the budding process (29-31). The HIV-1/MN virus was grown in white blood cells. There is no equivalent fatty acid analysis of phosphatidylcholines in human white blood cells, hence the distribution for human erythrocytes is listed in Table 11. An interesting observation shown in Table I1 is the relatively high number of odd-numbered carbon fatty acyl groups in the phosphatidylcholines present in the HIV-1/MN viral preparation. It can be seen that C15,C1,, and Clgacids are observed, whereas only CI7 acid has previously been observed in phosphatidylcholines present in human blood cells and plasma (26,28). It is possible that the virus is exhibiting a differential specificity of fatty acid chain length. It is also possible that mass spectrometry is a more sensitive analytical technique that is able to detect relatively small quantities of certain phosphatidylcholines that might be elusive to traditional methods of analysis such as hydrolysis, gas chromatography, and two-dimensional thin-layer chromatography (32). LITERATURE CITED (1) Hayashi, A.; Matsubara. T.; Morka, M.; Kinoshita, T.; Nakamura. T. J . B(ochem. 1989, 106. 264-269. (2) Ho, B. C.; Fenselau. C.; Hansen, G.; Larsen, J.; Daniel, A. Clin. Chem. 1908, 29, 1349-1353. (3) Fenselau, C.; Heller, D. N.; Olthoff, J. K.; Colter, R. J.; Kishimoto, Y.; Uy,0. M. B M . Envhon. Mess Spectrom. 1989, 18, 1037-1045. (4) Ross, M. M.; Nelhof, R. A.; Campana, J. E. Anal. Chim. Acta 1988, 181, 149-157. (5) Mattlna. M. J. I.; Richardson, S. D.; Wood, M., Jr.; Zhou, Q. 2.; Conta-

do, M. J.; Msnger, F. M.; Abbey, L. E. Org. Mess Spectrom. 1988, 23, 292-296. (6) FenwM, 0. R.; Eagles, J.; Self, R. Bbmed. Environ. Mess Spectrom. 1983, 10. 382-386. (7) Ayanoglu, E.; Wegmann, A.; Pilet, 0.; Marbury, 0. D.; Hass, J. R.; Djeressl, C. J . Am. Chem. Soc. 1884. 106, 5246-5251. (8) Jensen, N. J.; Tomer, K. B.; Gross, M. L. Llplds 1988. 21. 580-588. (9) Sato, S.; Asada, T.; Ishihara, M.; Kunihho, F.; Kammel. Y.; Kubota, E.; Costello. C. E.: Martin. S.A.: Scoble. H. A.: Blemann. K. Anal. Chem. 1907, 59, 1652-1659. (10) Qurgo, C.; OUO, H.O.; Franchlni, 0.; Aldovini, A,; Collalti, E.; Farreil, K.; Wong.Staal, F.; Gello, R. C.; R e k , M. S., Jr. Vkobgy 1888, 164.

531-536. (11) Mueslng, M. A.; SmiVI, D. H.; Cabradllla. C. D.; Benton, C. V.; Lasky, L. A.; Capon, D. J. Netwe 1986, 313, 450-458. (12) WaiwHobson, S.; Sonlgo, P.; Danos, 0.; Cole, S.;Ailron, M. Cell 1986, 40,9-17. (13) Knight, C. A. chemisby of Vkuses. 2nd ed.; Springer-Verlag: New Y&. 1975. (14) Klenk, H . 4 . ; Choppin, P. W. Vkdogy 1989. 24, 155-157.

1114

AMI. them. 1901, 63, 1114-1118 Ansell, (3. B.; Hawlhom, J. N.; Daw”, R. M. C. F m and Fulctkn of-, 2nd ed.;ElseLondon, 1073 Vd. 3. IkockWhOff, H.; Hoyb, R. J.; Hwang, P. C. Bkdkkn. 8bphys. Acta 1967, 141, 541-548. Bryant, D. K.; Orlando, R. C. RapH Commun. Mess Spec”.1991, 5, 124-127. Helbr, D. N.; Murphy, C. M.; Cotter, R. J.; Fenselau, C. Anal. chem. 1988, 60, 2787-2791. Munster, H.: Stein, J.; Budziklewicz. H. 8iomed. Envkon. Mess SpecIra" 1986, 13, 423-427. Ohashl, Y. Blomed. Envkon. Mess Specfrum. 1964, 1 1 , 383-385. Jensen. N. J.; Tomer, K. B.; Gross, M. L. Anal. Chem. 1985, 60, 2018-2021. [ktmL. , J.; Moseky, M. A.; T a r , K. B.; Jorgensen. J. W. Anal. chem.1888, 81, 2504-2511. Munster. H.; Budzlkiewlcr, H. R aw Commwr. Mess Specfrum. 1987. 1 , 128-128. Mattin, S. A.; Johnson. R. S.; htdlo, C. E.; Bbmann, K. of Ihs 4lh Texas SJ” on Mess speciromelry; ~ohnwiley 6. Sons: Chichester. New York, Brlsbene, Toronto, Singapore, 1088; pp 135-150. Huang, 2.4.;a g e , D. A.; Sweeby, C. C. J . Am. Soc. Mess Specfrom., in press.

(28) Dodge, J. T.; Phillips, 0. B. J . L@URes. 1987, 8 . 887-875. (27) Hamhen. D. J.; Brockerhoff, H.; Banon, J. J . Bkl. U”.1960, 235, 1917-1923. (28) Mlps, G. B.; J. T. J . yold Res. 1967. 8 . 676-881. 1957. 128, 208-210. (29) WedtW. E. 2. Ne(30) Frommhagen, L. H.; Knlght. C. A. V1.dogy 1050, 8, 108-208. (31) Kknk, H.D.; Choppln, P. W. Vkobgy 1970. 40, 030-047. (32) Krlt~bvdcy,D.; Shapko, 1. L. In Melhodp In Vh&y; Maramosch, K., Koprowdti, H., E&.: Accldemic Press: New York, 1087; Vd. 3, pp 77-08.

m.

RECEIVED for review December 12,1990. Accepted March 4, 1991. This work was supported by a grant (BBS 87-14238) from the National Science Foundation. Mass spectral measurements were performed at the Structural Biochemistry Center, an NSF-supported Biological Instrument Center at the University of Maryland Baltimore County. Viral isolation and purification was performed under Contract No. N01CO-74102.

Process Monitoring by Parallel Column Gradient Elution Chromatography G. Thevenon-Emeric’ and F. E. Regnier* Department of Chemistry, Purdue University, West Lafayette, Indiana 47907

A system was developed that generates two gradlents simultaneously, 180’ out of phase. Essential elmants of the system were a pump ddlverlng solvent A at a constant vokctly, a pump ckllvorhg solvent B at a condant vdodty, two mixing chambers, and a tengort valve. By use of this gra dknt device and two reverwbphaw columns packed with 10oo-A pore diameter, 8-pm partido dze macroporour poiy(styrene-dlvinylbanzene) resin, SIXprotelnr were separated In 40 s In a single gradlent cycle.

INTRODUCTION Federal regulatory agencies mandate that the production of therapeutic substances for humans be carefully monitored at each step of the process. Once it has been validated that a production or purification step produces a therapeutic substance of given structure and purity, regulators require in subsequent applications of the process that it continues to perform at this specified level. The time required to determine whether a process is meeting specifications is an important issue in manufacturing. When this time is short relative to processing time, the analytical data may be used in some type of feedback system to adjust process parameters controlling product quality. In contrast, the control function of process monitoring is very different in those cases where quality assurance data are generated more slowly than the rate of production. Learning that a product did not meet specification after the process is completed only allows one to discard or recycle unacceptable material. Rapid process monitoring, in this regard, becomes an enabling technology for feedback control of a process. Present address: Purdue University, Pharmacy Building, Department of Medicinal Chemistry, West Lafayette, IN 47907.

Although high-performance liquid chromatography (HPLC) could be a powerful technique for process monitoring, the 20-30 min required for analysis of a protein mixture is too slow for many applications. Recent advances in column technology now make it possible to carry out protein separations in 30-60 s by using kinetically enhanced packing materials and high mobile-phase velocities (I). Unfortunately, conventional HPLC instruments do not readily adapt to the preparation of several hundred 60-s gradients at 5 mL/min. This paper examines the use of parallel stirred tanks operating 180° out of phase for the repetitive formation of gradients.

EXPERIMENTAL SECTION Apparatus. The pumping system and detector used in any particular separation will be described in the Results and figure legends of the paper. Pumping systems from three different instruments were used in this work a Varian 5500 pump (Walnut Creek, Ca) designated V5500, a Perkin Elmer series 400 pump (Norwalk, CT) designated PE400, and an Altex 110 pump (Berkley, CA) designated ALl10. Separations were monitored with one of three detectors: a Kratos Spectroflow 757 detector (Ramsey, NJ), a Varian UV-50 detector (Walnut Creek, CA), or a Hewlett-Packard 1040A photodiode array detector equipped with an HP85B personal computer (Palo Alto, CA). Sample injections onto all analytical columns were made with Valco Model C6-U valves (Houston, TX). A ten-port Valco Model C10-U switching valve was used in the gradient generator. A Rheodyne Model 7125 valve (Cotati, CA) equipped with a 12-mL loop was used to load the preparative column. The dynamic mixers used in the gradient generator were obtained from LDC/Milton Roy (Riveria Beach, FL). Analytical reversed-phase separations were carried out on columns 5 mm long by 6 mm internal diameter that were packed with 8pm PLRP-S lo00 resin (Polymer Laboratories, Shropshire, U.K.). The columns were packed under constant pressure with a 2-propanol slurry of particles at 4000 psi. Preparative anion-exchange chromatography was carried out on a 2.54 X 30 cm column packed with 20-pm SynChroprep AX

0003-2700/91/0383-1114S02.50/0 0 I991 American Chemical Society