Dec., 19503 ALBANS AND BAKER 667 Paper Chromatography in Penicillin Production Control BY J. W. ALBANS AND P. B. BAKER (Read at the meeting of the Society on Wednesday, April 6th, 1949) SYNoPsIs-The application of a method for estimating penicillin species as described by Baker, Dobson and Martin in the previous paper, to culture filtrates and similar process samples, is discussed. The changes in com- position that take place during fermentation and the use of the technique for examining such fermentation variables as the composition of the medium, the rate of aeration, the effect of precursors and of the fermentation period are also discussed. Some confirmation has been obtained of the existence of substances previously indicated, but not yet characterised, by the micro- biological chromatographic technique.The construction of a large vessel for multiple analyses is described. Esamples of the degree of reproducibility of the results are shown. As knowledge of penicillin therapy has increased and as methods of fermentation and processing penicillin have developed, there has been a demand for a rapid and accurate method of estimating the different penicillin species in fermentation liquors and in the fractions obtained during extraction and purification. Of the several methods proposed, the differential assays of Schmidt, Ward and Coghilll and of Higuchi and Peterson2 are of limited application unless a wide range of suitable organisms is available for the tests. The inherent error in biological assays of this kind is then multiplied and the method becomes cumbersome.The Craig3 counter-current distribution technique has been found suitable for a small number of samples, but it is slow and laborious. A paper chromatographic method developed by Goodall and Levi4p5 gives a direct qualitative estimate of the various penicillins in mixtures, but the quantitative treatment proposed is open to criticism. Moreover, the delay in obtaining the results, 72 hours, renders the method of limited use for production control when the results are required with a minimum of delay. The foregoing paper by Baker, Dobson and Martin,6 which describes a method for the separation and estimation of the penicillin hydroxamic acids, refers mainly to work on the crystalline penicillins that were obtained from fermentation liquors by various methods of extraction and purification.The method is claimed to be suitable for the differential estima- tion of penicillin species in mixtures; and with those penicillins that have been isolated and for which a standard curve can be constructed it is claimed that the results for specific penicillins are reliable. The present paper deals with applications of this basic technique to the estimation of the penicillins in culture fluids and similar process samples obtained during the extraction of penicillin. Any method of analysis that is to be useful as a control test for a production plant must fulfil certain conditions. When the processing of a large quantity of material is dependent on the results of a control test, the test should be speedy.The method must be one that can easily be carried out by unqualified assistants and should require the minimum of special apparatus. The technique to be described has been employed successfully by two assistants for more than a year. Throughout this period, supervision has been reduced to a minimum. The results obtained have been satisfactory and periodical checks on known mixtures of penicillins have shown that accuracy has been maintained. The only special apparatus used is the tank made of Yerspex and this will last indefinitely. The diaphragm pump, mentioned in the previous paper,6 has been replaced by a cheap centrifugal pump readily available from stock. The other apparatus consists of items normally used in a biochemical laboratory. On only one point does the method fail to fulfil the conditions mentioned above, viz., that the minimum period for dealing with a sample of liquor is about 12 hours and for a solid, 8 hours, but this rather long delay has not caused any trouble.Moreover, it often happens that the constitution of a sample is known before the normal bio-assay is available. Further, the method of calculating the results is very simple.658 ALBANS AND BAKER : PAFER CHROMATOGRAPHY IN [Vol. 75 THE ESTIMATION OF PENICILLIN SPECIES IN FERMENTATION LIQUORS AND PROCESS SAMPLES- Owing to the low concentration of penicillin compared with other solids, the normal method of separating and estimating the hydroxamic acids of the penicillins6 cannot be applied directly to culture filtrates and similar process samples.Thus, for a culture fluid of 600 units per ml., in order to achieve the desired loading of 1 mg. of each penicillin species, at least 300mg. of inactive material would be applied at each spot. To overcome this difficulty the penicillins are extracted into ether at about pH 2 and then converted into the hydroxamic acids. Removal of ether and subsequent concentration of the aqueous residue are carried out under such conditions that inactivation is reduced to a minimum. METHOD Extract 200 ml. of culture filtrate (or the equivalent of any other sample) , previously clarified with the aid of kieselguhr, with two 100-ml. portions of ether at pH 1-5 to 2.5, adjusting with 20 per cent. phosphoric acid. Transfer the ether extract to a 6-in. crystallising dish containing 0.5 ml.of 4 M hydroxylamine hydro- chloride and 0-7 ml. of 3 M sodium hydroxide. Mix well and see that the aqueous layer does not become too acid; adjust it to pH 6-2 with N sodium hydroxide, if necessary. Remove the ether by means of a gentle stream of air over the surface. Concentrate the aqueous residue (4 to 6 ml.) to about 0.5 to 1.0 ml. by means of a blast of warm air from a hair-drier fixed about 12 in. above the dish. Remove any insoluble matter by means of a filter-stick and apply the clear filtrate to buffered paper as previously described.6 (A check on the concentration achieved can be obtained by spotting samples of the concentrate and a control of known strength on to a filter-paper soaked in 2 per cent. ferric chloride.) In the earlier stages of the work the results were rather erratic, owing in the main to insufficient control of the extraction and conversion stages.Attempts to avoid emulsifica- tion led to incomplete extraction, e.g., only 25 per cent. of added penicillin could be extracted from solution in a corn-steep-liquor medium, whereas with an aqueous solution, where no emulsification occurred, the recovery was more than 90 per cent. A centrifuge had been used quite successfully to break the emulsions, but it was thought that an alternative method that involved no fire risk was desirable. The use of surface-active compounds showed some promise, but to work out the exact conditions would probably have required a consider- able amount of study and this idea has not yet been pursued further.If the fermentation liquor, free from mycelium, is heated to 60' C. and rapidly cooled before clarification, emulsification is less troublesome. The precipitation of proteins by tannic acid is unsatis- factory as the blue stain produced when the papers are sprayed with ferric chloride masks the bands due to penicillin. The acid extraction should, of course, be carried out rapidly, and it is of more importance to separate the ether extract and add it to the hydroxylamine reagent quickly than to attempt to make fine adjustments in the degree of acidity. In experiments with solutions of penicillin salts, it has been shown that, under the conditions recommended, there is no differential extraction of any of the penicillin species. It is vital that traces of emulsion or of aqueous layer be not added to the dish or the mixture will be too acid for conversion to hydroxamic acids.For the concentration of the hydroxamic acid solutions, lyophilic drying has been used with success, but the stability of these compounds has made it possible to use the quicker method described earlier as a routine. Break the emulsion by centrifuging. LARGE APPARATUS FOR MULTIPLE ANALYSES- As the development of this method of analysis progressed it became necessary to construct a larger vessel. In view of the critical humidity conditions necessary for good separation, a prototype (10 x 16 x 24 in. high) , consisting essentially of three of the smaller vessels, was built. However, it was eventually found that spa-rgers and curtains round the outer walls of the vessel were sufficient to achieve the desired degree of humidification in this size of vessel.With the arrangement indicated in Fig. 1, movement of the zones on paper 3b was rather slower than on the other papers and, in fact, fairly wide variations in the distances travelled were shown on the various papers (see Table I). The spargers and curtains are shown in Fig. 1.Dec., 19503 PEXICILLIN PRODUCTION CONTROL TABLE I 659 DISTANCES OF TRAVEL, MM., OF ZONES IN 64 HOURS Papers 7 A -I Iu 1b 2a 2b 3a 3b G 66 55 69 64 74 49 F 7s 79 97 93 100 76 D 102 107 128 I24 132 99 K 840 240 300 289 285 264 This difference in the distance of travel is attributed to striation in the spargers. In order to obtain similar separation on the different papers in the tank, it is advisable to have b Box 2 I , 3 t Fig.1. Large apparatus for multiple analyses the spray evenly distributed round the walls of the tank. This refinement is not absolutely essential, however, for, although the absolute distances of travel vary, the relative movements are constant, provided that the conditions specified6 are satisfied. This constancy of relative movement is shown by Tables I1 and 111. TABLE 11 RATIOS OF DISTANCE OF TRAVEL G F D K Papers l a lb 2n 9h 3 n 3b 1.00 1.00 1.00 1-00 1-00 1.00 1-39 1-44 1.41 1 4.4 1.35 . 1.53 1.83 1.94 1-85 1 *92 1.78 2-05 4.28 1.36 4-34 4.46 3.85 5-17 A 7 7 TABLE 111 MEAN VALUES OF RATIOS UNDER VARYING HUMIDITY CONDITIONS G F D K Spargers All in B out B, C out B, C, D out 7 A \ 1.00 1-00 1.00 1 -00 1-43 1.40 1.46 1-42 1.89 1-91 1.96 1.91 4.41 4-90 5-30 4.56 Mean 1-00 1.43 1.89 4.41660 ALBANS AND BAKER : PAPER CHROMATOGRAPHY IN Wol.75 This constancy of relative movement for penicillins G, F and D is of very great importance in determining the identity of the bands, as will be seen later. The variation shown by penicillin K is due to the fact that this band is in the pH 6.2 region of the paper, and the relative movements of bands in the two regions are not comparable. The degree of reproducibility attained with this vessel is indicated by the two examples given in Table IV. TABLE IV Mean REPLICATE ANALYSES OF A MIXTURE Operator A G F D K f- 47.5 44.1 46.0 44.0 47.3 45.4 45.7 per cent. w/w 13-5 25.5 15.8 26.6 14.2 28.0 14.5 27-8 11.’7 27.8 14.13 27.4 14.1 2’7.16; --A G Operator B F I) per cent.w/w Mean 48.7 49.4 45.0 43-2 47.5 45.4 46.5 14.6 15.7 15.2 13.1 15.8; 15.1 14.9 23.6 22.7 25.3 23.9 24.8 26.2 24-4 7 13.5 13.5 11.8 14.0 13.2 12.4 13.07 K 7 13.1 12.2 14.5 19.8 11.9 13.4 14-1 For the vessel the following details are regarded a s essential- (i) The return line to the circulating pump should not be of greater diameter than Q in.; otherwise, with a centrifugal pump, there is a tendency for the lighter ether layer to be circulated in preference to the salt solution, and this will lead to streaky and indistinct zones. (ii) To prevent warping of the longer sides of the vessel it is advisable to fix metal strips along the top. In addition, to prevent warping and leakage of vapour, flanges should be screwed to the lid. (iii) The spraying system should form a continuous ring round the vessel, with the pump feeding at two diametrically opposite points.In this way the sprays deliver at ail even pressure and the two phases remain emulsified. (iv) If the pump or any part of the apparatus is made of brass or other material attacked by the phthalate solution, the latter can be replaced by a 12 per cent. w/v solution of anhydrous sodium sulphate, which has approximately the same vapour pressure as 0.6 M phthalate. (v) With the increased surface area of the lid condensation occasionally occurs. If the droplets fall on the paper strips the strips become waterlogged and separation does not take place. A piece of filter-paper fastened to the lid to absorb the moisture, or a gabled lid to induce the drops to run to the side walls, overcomes this difficulty.APPLICATIONS TO FERMENTATION STUDIES- It is possible to investigate how the production of the different penicillins is affected by such factors as precursors, composition of the medium, aeration and culture variation. For example, a high rate of aeration favours the production of penicillin K, as is shown in Table V, where the proportion of penicillin K increases as the volume of medium in “shaken flask” cultures decreases. The simplicity of this technique has led t o its use in problems of fermentation.Dec., 19501 PENICILLIN PRODUCTION CONTROL TABLE V 661 Volume . . .. 100ml. 80 ml. 60 ml. 40 ml. K, per cent. . . . . 14.4 49.3 63.3 90.2 G, per cent. . . .. 85.6 50.7 36.7 9.8 The changing picture during fermentation has been followed on a small scale by the use of cultures that produce a mixture of penicillins predominantly G, F, D and K.Under the conditions used, the G content rose rapidly to a maximum and then decreased. This latter fall in G content was associated with an increasing production of penicillin K and an increasing biological titre. In fermentations of this kind, in which a mixture of penicillins is produced, the biological assay is of little value and can, in fact, be misleading. Detection of “newJJ penicillins-During some work recently carried out on fermentation liquors, samples taken during the early stages of fermentation have shown a number of bands riot associated with any of the major penicillins, X, G, F, D, K, but corresponding in position on the chromatograin to the minor constituents detected by the Goodall and Levi technique.6 The entities responsible for these bands have not been isolated nor has their biological activity been determined; but, since they are decomposed by penicillinase one is led to suppose that they are similar in structure to the known penicillins. TABLE VI POSITIONS OF UNIDENTIFIED ZONES x ? G F D (4 0.08 1 -00 1-50 2.13 - ( b ) - 0.55 1 -00 1.49 2.05 One species has been detected between the bands due to X and G, but, as this was in a sample from a single shaken flask culture, it has not been possible to follow its production or ultimate fate.A second “unknown” species, lying between D and K on the chromatogram, has been produced in the early stages of some stirred fermentations.The proportion rises to a maximum and eventually the band disappears as the-fermentation is prolonged. Tables VI and VII indicate the positions of these two fractions, and the figures show the relative distances travelled by the various penicillins. In Table VI, the values in column (a) are the means of 64 determinations. In column (b) the existence of a compound producing a band between those due to penicillins X and G is indicated. In Table VII the presence of a penicillin that produces bands between those due to penicillins D and K is shown. As the fermentation proceeds this band disappears. TABLE VII 60 hours 72 hours 84 hours G 1 4 0 1.00 1.00 F 1.47 1.49 1.46 L) 1.92 1.99 1.97 ? 2.25 2.24 - K 5.50 5.61 5.94 Thirdly, in many instances the K zone has been separated into its three components. In this connection a recent paper7 on the estimation of the penicillins as their “R-group acids,” indicates the heterogeneity of penicillin K, but the side chains of the individual compounds have not been established.A fourth mauve band appearing at the lower end of the pH 4 paper has been identified as due to phenylacetic acid; it also disappears as fermentation is prolonged. These points indicate the complexity of the problems confronting those investigating fermentation. At this stage we offer no theories about the constitution of compounds that appear as minor constituents nor on their role or fate during fermentation, but it seems likely that further investigation of them would give useful information about the mode of production of the various penicillins.NICHOLAS AND POLLAK: THE ISOLATION OF THE LINES [Vol. 76 662 Our thanks are due to numerous colleagues for providing samples for this investigation and to Boots Pure Drug Company Limited for permission to publish this paper. RBFErm %CES 1 . 2 . 3. 4. 5 . 6. 7. Schmidt, \V. H., \Yard, C . J,:., and Coghill, 1i P., ./. JJucf., 1945, 49, 1 1 1 Higuchi, I<., and Petcrson, I!‘. H., A m l . Chei,i., 1947, 19, 68. Craig, L. C., J . Biol. Chem., 1944, 155, 519. Goodall, R. K., and Lei4, A. A., Xalzur, 1946, 158, 678. -,- , Analyst, 1947, 72, 277. Baker, P. B., Dobson, I?., and Martin, A. J. P., Ibid., 1950, 75, 6.>1. Higuchi, K., and Peterson, My. €I., ,3nal. Chuui., 1949, 21, 659. BOOTS PURE DRUG Com.my LI~IITEI) TECHNICAL DEVELOPMEXT DEPARTMEST .IND BIOCHEMISTRY DIVISIOX 017 THE KESEARCII I h i ) i~ri-vi<\ I ISLAM, STREET 1;ii y t bubmitted, May, 1949 h-OTTIKGHASI -hiended, June, 1960