ANALYTICAL
BIOCHEMISTRY
89, 213-219 (1978)
Separation of Quinoxaline Antibiotics by Coil Planet Centrifugation IAN A. SUTHERLAND,* * National
Institute for Pharmacology,
JEREMY Medical Medical
S. LEE,?,’ AND DENIS J. GAUVREAU~
Research, Schools,
Mill Hill, London, and t Department Hills Road. Cambridge. England
of
Received January 3, 1978 Liquid/liquid chromatography has been used for the isolation and purification of quinoxaline-type compounds on a coil planet centrifuge. Three solvent systems have been developed which are capable of high resolution both analytically and preparatively. Separations are described involving naturally occurring antibiotics, biosynthetically produced analogues, and chemically modified derivatives.
The quinoxaline series of antibiotics are highly active against Grampositive bacteria and inhibit various experimental tumours. Echinomycin (quinomycin A) and triostin A, which are two representative members of this series, have recently attracted a great deal of interest because of their novel mode of bifunctional intercalative binding to DNA and because they show some specificity in their binding to DNAs of different base sequence (1,2) (Fig. 1). To date, some 10 members of the series have been identified which differ only in the nature of the sulphur-containing cross-bridge and/or in the amino acid residues in the L-N-methylvaline position. There are, however, also the possibilities of generating many more potentially interesting and useful antibiotics by biosynthetic replacement of the chromophores and by chemical modification of existing antibiotics. A serious problem in this approach to the development of new antibiotics appears in the isolation and purification of such closely related compounds. The producing organism frequently manufactures several members of the series simultaneously, and this number is likely to be increased severalfold when biosynthetic replacement is attempted (3,4). Similarly, for molecules of this size chemical modification invariably leads to many products (5). In this report we should like to demonstrate how this problem can be overcome using liquid/liquid chromatography on a coil planet centrifuge (CPC) (6). Liquid/liquid chromatography is familiar as the technique of countercurrent distribution (CCD) (7). Provided two compounds have slightly 1 Present address: Department Canada.
of Biochemistry,
213
University
of Alberta,
Edmonton,
0003-2697/78/0891-0213$02.00/O Copyright 0 1978 by Academic Press, Inc. All rights of reproduction in any form reserved.
214
SUTHERLAND,
LEE, AND GAUVREAU c--.
‘0 \ O-SW ‘\ , I-“.
A
I I
L-Ala ---
1’ .’
Echinomycin
Triostin
A
FIG. 1. Structures of echinomycin (quinomycin A) and triostin A, which only differ in the nature of the sulphur-containing cross-bridge.
different partition coefficients in a two-phase solvent system, the compounds can be separated by a repetitive process of mixing and settling and the transference of one of the phases relative to the other. However, compounds with very similar partition coefficients require the use of many hundreds of transfers, resulting in long operative times, the use of large volumes of solvents, and problems of sample stability. Coil plant centrifugation can be envisaged as a method for achieving continuous
SEPARATION
OF QUINOXALINES
215
CCD. One phase of the two-phase solvent system is held in a coil on the rotating centrifuge while the other phase is pumped through it (Fig. 2). This technique makes it possible to complete the equivalent of 1000 transfers within a few hours using only 200 ml of solvents. Consequently, CPC does not suffer from the disadvantages of CCD yet it retains the latter’s potential for high resolution. Although liquid/liquid paper chromatography is also capable of high resolution it cannot be used preparatively for quinoxaline antibiotics (38). METHODS
Echinomycin was a gift from Drs. H. Bickel and K. Scheibli, CibaGeigy Ltd., Basle, Switzerland. Triostin A, a product of Shionogi and Co. Ltd., Osaka, Japan, was a gift of Dr. H. Otsuka. Quinomycin C was a gift from Dr. D. G. Martin, The Upjohn Co., Kalamazoo, Michigan. A methyl iodide-treated derivative of echinomycin was prepared by dissolving 200 mg of echinomycin in 14 ml of 50% (v/v) methyl iodide/acetone and refluxing for 48 hr in the presence of a trace of water and 50 mg of calcium carbonate. The reaction mixture was then worked up in water and extracted with chloroform to yield 150 mg of products. In addition, analogues of echinomycin in which quinoline replaces either one or two of the quinoxaline chromophores of the original molecule were prepared by solvent extraction of a culture of Streptomyces echinatus to which quinoline-2-carboxylic acid had been added (4). The extracts were
FIG. 2. Diagrammatic view of the coil planet centrifuge. The stack is geared to make one counterrevolution for each revolution of the centrifuge. In this way the leads are not twisted. The “recycle” mode of operation is simply achieved by connecting the output lead to the pump.
216
SUTHERLAND,
LEE,
AND
GAUVREAU
decolorized by active charcoal and then used directly without further purification. Three solvent systems were found to be suitable for use with quinoxaline antibiotics: I, acetone:water:heptane:ethyl acetate (3:l: 1: 1); II, l,l,l-trichloroethane:methanol:water (7:3: 1); and III, l,l, l-trichloroethane:methanol:water (7:5:1). For solvent system I the lower phase was the mobile phase, whereas for II and III the upper was used as the mobile phase. It was not found necessary to buffer the solvents and the absence of salts greatly facilitated the recovery of the products by solvent evaporation. Many other solvent systems were rejected because of inconvenient partition coefficients, too low average solubility, or unsuitable interfacial tension between the two phases. These parameters and thus a solvent’s suitability were evaluated by simple test-tube experiments (9). Solvent I was used on a centrifuge operating at 150 rpm at a radius of 60 cm using a 480 coil column of 0.16-cm bore Teflon tubing with an overall capacity of 60 ml. Solvent II was used on a centrifuge operating at 300 rpm at a radius of 20 cm using a 1040 coil column of 0.16-cm bore Teflon tubing with an overall capacity of 116 ml. For solvent system III the same column was used as for solvent II but it was operated at 500 rpm at a radius of 9.5 cm. The output from the centrifuges was monitored spectrophotometrically in l- or IO-mm flowthrough quartz cuvettes at 330 nm. The samples were injected, dissolved in 1 or 2 ml of the mobile phase. Further details of the theory of CPC and of the design and operation of the centrifuges have been published elsewhere (9,lO). RESULTS
AND DISCUSSION
Figure 3 shows the separation of echinomycin (MW = 1100) and quinomycin C (MW = 1156) using solvent I. The latter antibiotic has N,y-dimethylalloisoleucine in both the L-N-methylvaline positions and thus in effect has two extra ethyl groups on the basic backbone. The small double peak which elutes first is caused by an impurity present in both the echinomycin and quinomycin C samples which had been undetected previously. These minor components are not formed by degradation in the centrifuge because they were not present when the main peaks were recycled. As in CCD the position and spread of the eluted peaks can be exactly related to the partition coefficients of the samples and operating parameters of the centrifuge (11). Surprisingly, triostin A (MW = 1086) could not be separated from echinomycin using this solvent system even when the material was recycled up to 20 times. When, however, solvent II was used, adequate resolution was achieved after only two cycles of centrifugation (Fig. 4A). The width of the triostin A peak compared with that of echinomycin is notable and is most probably due to a slow interchange between the two different conformers which have been observed for this
SEPARATION
OF QUINOXALINES
TIME FROM SAMPLE
INJECTION
217
lmmsl
FIG. 3. The separation of echinomycin (E) and quinomycin C (Q) using solvent I. One milligram of each antibiotic was injected at Time 0 with a flow rate of 60 ml/hr. Note the small impurities which elute first.
antibiotic (12). Presumably the two conformers have slightly different partition coefficients in this solvent system, and the fact that they are not resolved suggests that a continual interchange is taking place. Solvent II has also been used in the isolation of the quinoline analogues of echinomycin (Fig. 4B). The small peaks which elute first are
TIME FROM
SAMPLE
INJECTION
(hnl
4. (A) The separation of triostin A (TA) and echinomycin (E). (B) The separation of the analogues of echinomycin (E) in which quinoline replaces either one (Al) or two (A2) of the quinoxaline chromophores of the original molecule. In both cases the sample weight was 4 mg and solvent system II was used at a flow rate of 120 ml/hr. FIG.
218
SUTHERLAND,
LEE, AND GAUVREAU
nonantibiotic impurities similar to those described above. The analogues are well resolved after only one cycle but the material could be recycled to give any desired degree of purity of the products. Moreover, sample capacities of 0.5 g are quite feasible. In this way it is possible to prepare series of analogous compounds rapidly and efficiently and thus to facilitate greatly structure/activity studies on these antibiotics. The technique has also been used in the preparation of derivatives of echinomycin. By treating echinomycin with methyl iodide it was hoped to cleave the sulphur-containing cross-bridge which in some way is important in determining the nucleotide sequence specificity of the quinoxaline antibiotics (2,13). It was found impossible to purify the crude methyl iodide-treated product by conventional absorption chromatography. An initial separation of the products was achieved, however, using solvent I, and one of the peaks showing antibiotic activity against Staphylococcus
ECHY-SPLIT
I
TIME FROM SAMPLE
2ND CYCLE
INJECTION
thrsl
FIG. 5. (A) Initial separation of methyl iodide-treated echinomycin. (B) Separation of “Echy-Split” into two further components Echy-Split I and Echy-Split II after recycling using solvent III. The sample size was (A) 45 mg and (B) 4 mg with a solvent flow rate of 60 ml/hr.
SEPARATION
OF QUINOXALINES
219
uuwus (unpublished data of D. Gauvreau) was selected for further study. This was further resolved into two components using solvent III (Fig. 5). Details of the structure and antibiotic activity of these purified products (Echy-Split I and Echy-Split II) are now under investigation and will be published elsewhere (13). The separations so far achieved lead us to believe that it will be possible to separate future quinoxaline derivatives by similar methods, and it is hoped that some of these derivatives will show less toxicity while retaining the antitumour activity of the parent antibiotics. There is no doubt that there are many other applications not only in the field of peptide antibiotics but also whenever efficient high resolution of similar compounds is required. ACKNOWLEDGMENTS We would like to thank Miss P. Newland of the National Institute for Biological Standards & Control for valuable technical assistance and Dr. M. J. Waring for a critical evaluation of the manuscript. J.S.L. is currently in receipt of an M.R.C. studentship and D.J.G. of a grant from La Fondation J. L. Levesque.
REFERENCES 1. Wakelin, L. P. G., and Waring, M. J. (1976) Biochem. J. 157, 721-740. 2. Waring, M. J., Wakehn, L. P. G., and Lee, J. S. (1975) Biochim. Biophys. Acra 407, 200-212. 3. Yoshida, T., Katagiri, K., and Yokozawa, S. (1961) J. Antibiotics (Tokyo) Ser. A 14, 330. 4. Yoshida, T., Kimura, Y., and Katagiri, K. (1968) J. Antibiotics (Tokyo) 21, 465. 5. Dell, A., Williams, D. H., Morris, H. R., Smith, G. A., Feeney, J., and Roberts, G. C. K. (1975) .I. Amer. Chem. Sot. 97, 2497-2502. 6. Ito, Y., and Bowman, R. L. (1971) Science 173, 420-422. 7. Craig, L. C., and Sogn, J. (1975) in Methods of Enzymology, Vol. 43, pp. 320-346, Academic Press, New York. 8. Otsuka, H., and Shoji, J. (1967) Tetrahedron 23, 1535-1542. 9. Sutherland, I. A., and Sharpe, J. E. E. (1976) J. Chromatogr. 122, 333-344. 10. Ito, Y., and Bowman, R. L. (1973) J. Chromatogr. Sci. 11, 284-291. 11. Priore, R. L., and Kirdani, R. Y. (1968) Anal. Eiochem. 24, 360-376. 12. Blake, T. J., Kalman, J. R., and Williams, D. H. (1977) Tetrahedron Lerf. 30,2621-2624. 13. Lee, J. S., and Waring, M. J. (1978) Biochem. J. 173, 129-144.
BIOCHEMISTRY
89, 213-219 (1978)
Separation of Quinoxaline Antibiotics by Coil Planet Centrifugation IAN A. SUTHERLAND,* * National
Institute for Pharmacology,
JEREMY Medical Medical
S. LEE,?,’ AND DENIS J. GAUVREAU~
Research, Schools,
Mill Hill, London, and t Department Hills Road. Cambridge. England
of
Received January 3, 1978 Liquid/liquid chromatography has been used for the isolation and purification of quinoxaline-type compounds on a coil planet centrifuge. Three solvent systems have been developed which are capable of high resolution both analytically and preparatively. Separations are described involving naturally occurring antibiotics, biosynthetically produced analogues, and chemically modified derivatives.
The quinoxaline series of antibiotics are highly active against Grampositive bacteria and inhibit various experimental tumours. Echinomycin (quinomycin A) and triostin A, which are two representative members of this series, have recently attracted a great deal of interest because of their novel mode of bifunctional intercalative binding to DNA and because they show some specificity in their binding to DNAs of different base sequence (1,2) (Fig. 1). To date, some 10 members of the series have been identified which differ only in the nature of the sulphur-containing cross-bridge and/or in the amino acid residues in the L-N-methylvaline position. There are, however, also the possibilities of generating many more potentially interesting and useful antibiotics by biosynthetic replacement of the chromophores and by chemical modification of existing antibiotics. A serious problem in this approach to the development of new antibiotics appears in the isolation and purification of such closely related compounds. The producing organism frequently manufactures several members of the series simultaneously, and this number is likely to be increased severalfold when biosynthetic replacement is attempted (3,4). Similarly, for molecules of this size chemical modification invariably leads to many products (5). In this report we should like to demonstrate how this problem can be overcome using liquid/liquid chromatography on a coil planet centrifuge (CPC) (6). Liquid/liquid chromatography is familiar as the technique of countercurrent distribution (CCD) (7). Provided two compounds have slightly 1 Present address: Department Canada.
of Biochemistry,
213
University
of Alberta,
Edmonton,
0003-2697/78/0891-0213$02.00/O Copyright 0 1978 by Academic Press, Inc. All rights of reproduction in any form reserved.
214
SUTHERLAND,
LEE, AND GAUVREAU c--.
‘0 \ O-SW ‘\ , I-“.
A
I I
L-Ala ---
1’ .’
Echinomycin
Triostin
A
FIG. 1. Structures of echinomycin (quinomycin A) and triostin A, which only differ in the nature of the sulphur-containing cross-bridge.
different partition coefficients in a two-phase solvent system, the compounds can be separated by a repetitive process of mixing and settling and the transference of one of the phases relative to the other. However, compounds with very similar partition coefficients require the use of many hundreds of transfers, resulting in long operative times, the use of large volumes of solvents, and problems of sample stability. Coil plant centrifugation can be envisaged as a method for achieving continuous
SEPARATION
OF QUINOXALINES
215
CCD. One phase of the two-phase solvent system is held in a coil on the rotating centrifuge while the other phase is pumped through it (Fig. 2). This technique makes it possible to complete the equivalent of 1000 transfers within a few hours using only 200 ml of solvents. Consequently, CPC does not suffer from the disadvantages of CCD yet it retains the latter’s potential for high resolution. Although liquid/liquid paper chromatography is also capable of high resolution it cannot be used preparatively for quinoxaline antibiotics (38). METHODS
Echinomycin was a gift from Drs. H. Bickel and K. Scheibli, CibaGeigy Ltd., Basle, Switzerland. Triostin A, a product of Shionogi and Co. Ltd., Osaka, Japan, was a gift of Dr. H. Otsuka. Quinomycin C was a gift from Dr. D. G. Martin, The Upjohn Co., Kalamazoo, Michigan. A methyl iodide-treated derivative of echinomycin was prepared by dissolving 200 mg of echinomycin in 14 ml of 50% (v/v) methyl iodide/acetone and refluxing for 48 hr in the presence of a trace of water and 50 mg of calcium carbonate. The reaction mixture was then worked up in water and extracted with chloroform to yield 150 mg of products. In addition, analogues of echinomycin in which quinoline replaces either one or two of the quinoxaline chromophores of the original molecule were prepared by solvent extraction of a culture of Streptomyces echinatus to which quinoline-2-carboxylic acid had been added (4). The extracts were
FIG. 2. Diagrammatic view of the coil planet centrifuge. The stack is geared to make one counterrevolution for each revolution of the centrifuge. In this way the leads are not twisted. The “recycle” mode of operation is simply achieved by connecting the output lead to the pump.
216
SUTHERLAND,
LEE,
AND
GAUVREAU
decolorized by active charcoal and then used directly without further purification. Three solvent systems were found to be suitable for use with quinoxaline antibiotics: I, acetone:water:heptane:ethyl acetate (3:l: 1: 1); II, l,l,l-trichloroethane:methanol:water (7:3: 1); and III, l,l, l-trichloroethane:methanol:water (7:5:1). For solvent system I the lower phase was the mobile phase, whereas for II and III the upper was used as the mobile phase. It was not found necessary to buffer the solvents and the absence of salts greatly facilitated the recovery of the products by solvent evaporation. Many other solvent systems were rejected because of inconvenient partition coefficients, too low average solubility, or unsuitable interfacial tension between the two phases. These parameters and thus a solvent’s suitability were evaluated by simple test-tube experiments (9). Solvent I was used on a centrifuge operating at 150 rpm at a radius of 60 cm using a 480 coil column of 0.16-cm bore Teflon tubing with an overall capacity of 60 ml. Solvent II was used on a centrifuge operating at 300 rpm at a radius of 20 cm using a 1040 coil column of 0.16-cm bore Teflon tubing with an overall capacity of 116 ml. For solvent system III the same column was used as for solvent II but it was operated at 500 rpm at a radius of 9.5 cm. The output from the centrifuges was monitored spectrophotometrically in l- or IO-mm flowthrough quartz cuvettes at 330 nm. The samples were injected, dissolved in 1 or 2 ml of the mobile phase. Further details of the theory of CPC and of the design and operation of the centrifuges have been published elsewhere (9,lO). RESULTS
AND DISCUSSION
Figure 3 shows the separation of echinomycin (MW = 1100) and quinomycin C (MW = 1156) using solvent I. The latter antibiotic has N,y-dimethylalloisoleucine in both the L-N-methylvaline positions and thus in effect has two extra ethyl groups on the basic backbone. The small double peak which elutes first is caused by an impurity present in both the echinomycin and quinomycin C samples which had been undetected previously. These minor components are not formed by degradation in the centrifuge because they were not present when the main peaks were recycled. As in CCD the position and spread of the eluted peaks can be exactly related to the partition coefficients of the samples and operating parameters of the centrifuge (11). Surprisingly, triostin A (MW = 1086) could not be separated from echinomycin using this solvent system even when the material was recycled up to 20 times. When, however, solvent II was used, adequate resolution was achieved after only two cycles of centrifugation (Fig. 4A). The width of the triostin A peak compared with that of echinomycin is notable and is most probably due to a slow interchange between the two different conformers which have been observed for this
SEPARATION
OF QUINOXALINES
TIME FROM SAMPLE
INJECTION
217
lmmsl
FIG. 3. The separation of echinomycin (E) and quinomycin C (Q) using solvent I. One milligram of each antibiotic was injected at Time 0 with a flow rate of 60 ml/hr. Note the small impurities which elute first.
antibiotic (12). Presumably the two conformers have slightly different partition coefficients in this solvent system, and the fact that they are not resolved suggests that a continual interchange is taking place. Solvent II has also been used in the isolation of the quinoline analogues of echinomycin (Fig. 4B). The small peaks which elute first are
TIME FROM
SAMPLE
INJECTION
(hnl
4. (A) The separation of triostin A (TA) and echinomycin (E). (B) The separation of the analogues of echinomycin (E) in which quinoline replaces either one (Al) or two (A2) of the quinoxaline chromophores of the original molecule. In both cases the sample weight was 4 mg and solvent system II was used at a flow rate of 120 ml/hr. FIG.
218
SUTHERLAND,
LEE, AND GAUVREAU
nonantibiotic impurities similar to those described above. The analogues are well resolved after only one cycle but the material could be recycled to give any desired degree of purity of the products. Moreover, sample capacities of 0.5 g are quite feasible. In this way it is possible to prepare series of analogous compounds rapidly and efficiently and thus to facilitate greatly structure/activity studies on these antibiotics. The technique has also been used in the preparation of derivatives of echinomycin. By treating echinomycin with methyl iodide it was hoped to cleave the sulphur-containing cross-bridge which in some way is important in determining the nucleotide sequence specificity of the quinoxaline antibiotics (2,13). It was found impossible to purify the crude methyl iodide-treated product by conventional absorption chromatography. An initial separation of the products was achieved, however, using solvent I, and one of the peaks showing antibiotic activity against Staphylococcus
ECHY-SPLIT
I
TIME FROM SAMPLE
2ND CYCLE
INJECTION
thrsl
FIG. 5. (A) Initial separation of methyl iodide-treated echinomycin. (B) Separation of “Echy-Split” into two further components Echy-Split I and Echy-Split II after recycling using solvent III. The sample size was (A) 45 mg and (B) 4 mg with a solvent flow rate of 60 ml/hr.
SEPARATION
OF QUINOXALINES
219
uuwus (unpublished data of D. Gauvreau) was selected for further study. This was further resolved into two components using solvent III (Fig. 5). Details of the structure and antibiotic activity of these purified products (Echy-Split I and Echy-Split II) are now under investigation and will be published elsewhere (13). The separations so far achieved lead us to believe that it will be possible to separate future quinoxaline derivatives by similar methods, and it is hoped that some of these derivatives will show less toxicity while retaining the antitumour activity of the parent antibiotics. There is no doubt that there are many other applications not only in the field of peptide antibiotics but also whenever efficient high resolution of similar compounds is required. ACKNOWLEDGMENTS We would like to thank Miss P. Newland of the National Institute for Biological Standards & Control for valuable technical assistance and Dr. M. J. Waring for a critical evaluation of the manuscript. J.S.L. is currently in receipt of an M.R.C. studentship and D.J.G. of a grant from La Fondation J. L. Levesque.
REFERENCES 1. Wakelin, L. P. G., and Waring, M. J. (1976) Biochem. J. 157, 721-740. 2. Waring, M. J., Wakehn, L. P. G., and Lee, J. S. (1975) Biochim. Biophys. Acra 407, 200-212. 3. Yoshida, T., Katagiri, K., and Yokozawa, S. (1961) J. Antibiotics (Tokyo) Ser. A 14, 330. 4. Yoshida, T., Kimura, Y., and Katagiri, K. (1968) J. Antibiotics (Tokyo) 21, 465. 5. Dell, A., Williams, D. H., Morris, H. R., Smith, G. A., Feeney, J., and Roberts, G. C. K. (1975) .I. Amer. Chem. Sot. 97, 2497-2502. 6. Ito, Y., and Bowman, R. L. (1971) Science 173, 420-422. 7. Craig, L. C., and Sogn, J. (1975) in Methods of Enzymology, Vol. 43, pp. 320-346, Academic Press, New York. 8. Otsuka, H., and Shoji, J. (1967) Tetrahedron 23, 1535-1542. 9. Sutherland, I. A., and Sharpe, J. E. E. (1976) J. Chromatogr. 122, 333-344. 10. Ito, Y., and Bowman, R. L. (1973) J. Chromatogr. Sci. 11, 284-291. 11. Priore, R. L., and Kirdani, R. Y. (1968) Anal. Eiochem. 24, 360-376. 12. Blake, T. J., Kalman, J. R., and Williams, D. H. (1977) Tetrahedron Lerf. 30,2621-2624. 13. Lee, J. S., and Waring, M. J. (1978) Biochem. J. 173, 129-144.
