Vol. 53

Some Properties of the Bacitracin Polypeptides

By G. G. F. NEWTON AND E. P. ABRAHAMSir William Dunn School of Pathology, University of Oxford

(Received 11 August 1952)

Newton & Abraham (1950) found that crudebacitracin could be resolved into a series of similarpolypeptides by counter-current distribution be-tween solvents. Better resolution and largeramounts ofthe products have now been obtained byusing a greater number of transfers and startingwith more concentrated solutions than were used inthe earlier experiments. It has been possible, bya single distribution, to prepare the various baci-tracins in quantities sufficient for some of theirchemical and biological properties to be compared.

Hausmann, Ahrens & Harfenist (1951) and con-taining 101 tubes. The solvent system consisted ofn-butanol (4 vol.), amyl alcohol (1 vol.), in equi-librium with 0-05M-potassium phosphate buffer(5 vol.), at pH 7-0 (Newton & Abraham, 1950).Analysis of the distribution curve (see Experi-mental) shows that the crude bacitracin wasresolved, or partly resolved, into at least ten differentcomponents. Table 1 shows the proportions in whichnine of these components were present, their parti-tion coefficients (K) in the solvent system used, and

E and D

D

20 40 60Fundamental series Tube no.

100 80 60Withdrawn series

1

Fig. 1. 221-transfer distribution of crude bacitracin in amyl alcohol-n-butanol (4:1); 0-05M-potassium phosphatebuffer, pH 7-0. *-*, total material in 0-1 ml. samples, taken from both phases in the fundamental series andfrom the single phase in the withdrawn series; - - -, calculated curves; , approximate interpretation.

RESULTS

Resolution of crude bacitracin bycounter-current distribution

Fig. 1 shows a distribution curve ofa sample ofcrudebacitracin that was previously describec, as ayfivin,batch 1 (Arriagada, Savage, Abraham, Heatley &Sharp, 1949). The distribution was carried out in anall-glass machine, similar to that designed by Craig,

the fractions from the machine which were pooledand concentrated at the end of the experiment.The component named bacitracin A' (Fig. 1) has

not so far been investigated. Evidence for theexistence of such a component was obtained fromearlier experiments (Newton & Abraham, 1950),and this evidence has been strengthened by theresults of the present counter-current distribution.It is possible to obtain a preparation of bacitracin A

597

G. G. F. NEWTON AND E. P. ABRAHAM

Table 1. Component8 of crude bacitracin obtained by counter-current distribution (see Fig. 1)(In the fourth column the numbers carrying a bar represent the numbers of fractions withdrawn from the machine.)

BacitracinE IDJBA'AC

F1F,F3F3

% of total K = (concn. in alcohol phase/material concn. in aqueous phase at pH 7-0)

17-5

PROPERTIES OF BACITRACIN POLYPEPTIDES

Amino-acid compo8ition. On two-dimensionalpaper chromatograms (see Experimental), acidhydrolysates ofbacitracinA showed apattern whichclosely resembled that given by a mixture of thefollowing amino-acids in simple proportions byweight: cystine, 0-5; ornithine, 1; lysine, 1;histidine, 1; aspartic acid, 2; glutamic acid, 1;phenylalanine, 1; leucine, 1; and isoleucine, 2. Thisbasic pattern also appeared on chromatograms ofhydrolysates of the other bacitracins. However, astrong spot in the position occupied by valine wasshown, in addition, by bacitracins B, D and E, anda much weaker one by bacitracins C, G, F1, F2and F3.

201-601F

A

101 501p

0 1 a II I -

230240 250260 27028290 300 320Wavelength (mp.)

Fig. 3. Ultraviolet absorption spectra of bacitracins E, D,B andA in aqueous solution atpH 3 0. , bacitracins;- - -, acid hydrolysate of bacitracin A.

10

o23

-... 1.'-'

I a I a a IA-- -? =b =

30 250 270 290 310 330240 260 280 300 320

Wavelength (m,u.)

Fig. 4. Ultraviolet absorption spectra of bacitracins C andG in aqueous solution at pH 3-0. , bacitracins;- - -, acid hydrolysate of bacitracin A.

401-

F'%

30

20F,

F" _~~~~~F".F-

230 250 270 290 310 330240 260 280 300 320

Wavelength (m,u.)

Fig. 5. Ultraviolet absorption spectra of bacitracins F1,FI and F3 in aqueous solution at pH 3-0. -, baci-tracins; - - -, acid hydrolysate of bacitracin A.

A strong spot in the position occupied by glycinewas shown by bacitracins C and G. This spot gavethe same reddish colour as that given by glycinewhen developed with ninhydrin solution to whicha smallamount ofcollidine had beenadded (Woiwod,1949).The amount of histidine in some of the baci-

tracins was determined colorimetrically by thePauly reaction (Jorpes, 1932). The followingquantities of histidine (in moles/100 g. peptidehydrochloride) were estimated to be present bycomparing the colours given by the unhydrolysectpeptides with those given by standard solutions ofhistidine: E, 0-037; D, 0-048; B, 0-057; A, 0-057;F2, 0-057; Fs, 0-056. The colour given by bacitracinA after hydrolysis with 6N-hydrochloric acid at

50'S

40 ,1%

30

Vol. 53 599

60t

1

_

G. G. F. NEWTON AND E. P. ABRAHAM1100 for 24 hr. indicated the presence of 0-066 moleof histidine/100 g. peptide hydrochloride.An attempt was made to estimate the amount of

ornithine present in acid hydrolysates of variousbacitracins by the Chinard method described byStein & Moore (19I51). However, it was found thatlysine gave a colour with the Chinard reagent whosedensity was about one-quarter of that given byornithine. Hydrolysates ofequal weights (estimatedby the photometric ninhydrin method) of baci-tracins E, D, B, A, C, and a mixture of F1, F2 andFs, gave very similar Chinard colours. The colourdensities were those that would be expected ifapproximately 0-64 mole of omithine and 0-64 moleof lysine were present in 100 g. of each peptidehydrochloride.Amide nitrogen. Hydrolysis of the bacitracin

polypeptides results in the liberation of volatilebase. This base, which is assumed to be amnoniaderived from an amide group, is completely freed bytreating the peptides for 20 min. at 1000 with N-hydrochloric acid, or for 3 hr. at 370 with O0-N-sodium hydroxide. The following amounts of suchamide nitrogen (in g.equiv./100 g. peptide hydro-chloride) are liberated from the various bacitracins:E, 0-006; D, 0-011; B, 0-064; A, 0-065; C, 0-054;G, 0-057; F1, 0-069; F2, 0-069; F3, 0-069. It isevident that bacitracins E and D differ from theother polypeptides in yielding a much smallerquantity of volatile base and it is probable that theydo not contain an amide group.

Groups reacting with fluoro-2:4-dinitrobenzene(FDNB). Bacitracins A, B and G were allowed toreact with FDNB in sodium bicarbonate solution(Sanger, 1945) and the dinitrophenyl (DNP)derivatives hydrolysed with acid. The ether-solublematerial from the hydrolysates was analysed onpaper chromatograms, using the procedure ofBlackburn & Lowther (1951). The free amino-acidsand DNP derivatives in the water-soluble materialwere analysed on single-dimensional paper chro-matograms developed with benzyl alcohol-hydrogencyanide (Consden, Gordon, Martin, Rosenheim &Synge, 1945) and on two-dimensional paper chro-matograms developed first with butanol acetic acidand then with 80% (w/w) phenol in the presence ofthe vapour of 50% (v/v) acetic acid.The ether-soluble material from bacitracin A

appeared to contain DNP-leucine or DNP-iso-leucine. The leucine-isoleucine spot obtained fromthe water-soluble material was found to be reducedin size, whereas the phenylalaniine spot appeared tobe unchanged. The ether-soluble material frombacitracin B appeared to contain DNP-leucine, orDNP-isoleucine, and also DNP-valine, while thevaline spot on the chromatogram of the water-soluble material was reduced in size. The water-soluble material from bacitracin G appeared to

contain no glycine. DNP-glycine was not detectedin the ether-soluble material, but about 80% ofanyDNP-glycine present in the DNP derivative of thepeptide would have been destroyed under the con-ditions that were used for hydrolysis (Porter &Sanger, 1948).Chromatograms of the water-soluble material

from bacitracins A, B and G showed yellow spots inthe positions occupied by gly-DNP-histidine and8-DNP-ornithine (gly-DNP-histidine is a histidinederivative containing a DNP residue as a sub-stituent on the glyoxaline ring.) They showed thepresence of free lysine, but very little ornithine orhistidine.

Reduction of ferricyanide. On treatment with0-5N-hydrochloric acid at 1000 for 20 min. baci-tracin A develops an ability to reduce potassiumferricyanide at room temperature, which is due, atleast in part, to the liberation of a thiol group(Newton & Abraham, 1953). After this treatmentthe various bacitracins showed a positive nitro-prusside reaction and reduced the followingamounts offerricyanide (in g.equiv./100 g. peptide):E, 0-027; D, 0-047; B, 0-093; A, 0-093; C, 0-076;G, 0-049; F1, 0-047; F2, 0-067; F3, 0-067.

Antibacterial activity. Activity against Coryne-bacterium xerosi was measured by the cylinder-plate method (Heatley, 1944), the unit of activitybeing that defined by Arriagada et al. (1949). Thevarious bacitracins showed the following activitiesin units/mg.: E, 0-3; D, 0-5; B, 13-5; A, 36-0; C,18-0; G, 5-0; F1, 2-0; F2, 1-0; F3, 0-5.

DISCUSSION

The solvent system used in the counter-currentdistribution described in this paper was triedoriginally because its pH was one at which theoverall partition coefficient of bacitracin activitychanged rapidly with hydrogen-ion concentration.It was thought that these were conditions underwhich the greatest differences in the partitioncoefficients of the constituents ofthe mixture mightbe found. This idea appeared to be borne out by thefact that very poor resolution was obtained in asystem at a lower pH, consisting of sec.-butanol andaqueous acetic acid (Newton & Abraham, 1950).However, direct measurement of the distributionof bacitracins A, B and C between amyl alcohol-butanol and buffer at various hydrogen-ion con-centrations has shown that the partition coefficientsofthese three substances, at least, show considerabledifferences over a fairly widepH range. Some factorother than pH must therefore be responsible for thepoor resolving power of sec.-butanol and aqueousacetic acid. Possibly the relative inefficiency of thelatter can be related to the fact that it is closer thanthe amyl alcohol butanol-buffer system to the

600 I953

PROPERTIES OF BACITRACIN POLYPEPTIDEScritical point at which a single phase is formed: themore similar the two phases in their properties theless selective is the system likely to be. The structuralfeatures of the various bacitracins which are re-sponsible for the wide differences in their partitioncoefficients in the amyl alcohol-butanol-buffersystem are not yet known.The counter-current distribution was designed to

obtain the various bacitracins as pure as possiblein a single experiment. After about 200 transfersthe distribution curve of bacitracin A resembledclosely the curve that would be expected for a singlesubstance over more than half of its range. Inweighing the evidence provided by the distributionfor the homogeneity of bacitracin A, however, itmust be remembered that nearly 3000 transfershave been used for the separation of leucine andisoleucine (Craig et al. 1951). If bacitracin A werereally a mixture of polypeptides which differed, forexample, only in the relative amounts of leucine andisoleucine that they contained, it would not besurprising to find that the mixture was not resolvedby the procedure used here.The purity of the bacitracins which are found at

either end of the distribution curve is even morequestionable. If two substances have partitioncoefficients K1 and K2, and if the volumes of theupper and lower phases are equal, the best separa-tion after a given number of transfers with a fixedratio of K1/K2 is obtained when \/K1K2 = 1 (Craig &Craig, 1950). In other words, substances near thecentre of a distribution curve will be more easilyresolved than substances at the two ends. Re-latively little information about the purity ofbacitracins E and F is provided by the work de-scribed here.

In spite of these uncertainties, it is clear thatcrude bacitracin has been resolved at least intogroups of very closely related polypeptides, andit appears from their amino-acid compositionsthat bacitracins E, D, B or C cannot have beenformed from bacitracin A during the process ofpurification.The similarities in the amino-acid patterns shown

by hydrolysates of the various bacitracins make itjustifiable to describe the latter as members of asingle family of polypeptides. Clear-cut differ-ences, however, have been demonstrated in theirchemical as well as in their physical and biologicalproperties.

Hydrolysates of bacitracins E, D and B differsharply from the hydrolysate ofA in showing a spoton paper chromatograms in the same position asthat occupied by valine. The same spot is alsovisible, though much less intense, in chromatogramsof hydrolysates of bacitracins C, G, F, F2and F.Craig, Gregory & Barry (1949) found a substance inhydrolysates of bacitracin which behaved like

valine on chromatography. They reported that thesubstance appeared to be absent from the purestmaterial that they obtained by counter-currentdistribution, so that this material presumablyconsisted of bacitracin A. Craig et al. (1949) isolatedthe valine-like substance, however, and stated thatit was not valine. In view of this it is surprising thata hydrolysate of the crude bacitracin used in thepresent experiments behaved as though it containedvaline on microbiologicaI assay (see Experimental).Further investigation of the valine-like substance isclearly necessary.

Bacitracins C and G differ from the other poly-peptides in yielding on hydrolysis a substancewhich behaves like glycine. They also differ in theirultraviolet absorption spectra, both substancesshowing a strong absorption peak at 258 m,. anda shoulder at 268 mp. The possibility must be con-sidered that a glycine residue is not present as such inthe polypeptides, but that it is formed from a morecomplicated structure during the hydrolysis withconcentrated hydrochloric acid. For example,certain purines are known to yield glycine whenhydrolysed with mineral acids at temperaturesabove 1000 (Smith & Markham, 1950; Frick, 1952).

Carboxyl groups in polypeptides are commonlytitrated between pH 2-5 and 5-5, histidine residuesbetween pH 5-5 and 7-0, a-amino groups betweenpH 7-0 and 8-5, and 8- or e-amino groups betweenpH 9-5 and 11-5. The close similarity between thetitration curves of bacitracins B, A, C and G overthe pH range 2-5-8-5 therefore suggests that allthree substances contain the same proportion offree carboxyl, a-amino, and glyoxaline groups.Reaction with FDNB indicates that bacitracin Acontains a free a-amino group which is part of aleucine or an isoleucine residue. But bacitracin Breacts with FDNB to give a DNP-leucine (or iso-leucine) and also a DNP-'valine' derivative, andbacitracin G can react to give a DNP-leucine (orisoleucine) and probably a DNP-glycine derivative.Moreover, bacitracins B, C and G show moretitratable groups than bacitracin A in the pH range8-5-11-5. This raises the question as to whether the'valine' in B, as well as the glycine in C and G, isreally present as a simple a-amino-acid residue withthe amino group free, or whether it is part of astructure whose nature has not yet been determined.

Irrespective of the nature of the glycine and'valine' residues, however, there is evidence that thebacitracins contain special structural features.A structure responsible for the liberation of a thiolgroup on mild treatment with acid appears to becommon to all the bacitracins, and at least two otherspecial structures, one present in bacitracins C andG, and the other in bacitracins F1, F2 and F3, seemto be necessary to account for the ultravioletabsorption spectra.

Vol. 53 601

G. G. F. NEWTON AND E. P. ABRAHAM

EXPERIMENTAL

Counter-current di8tributiownSolvents. n-Butanol and amyl alcohol were purified in the

manner described by Newton & Abraham (1950). A mixtureof 1 1. ofwet butanol and 4 1. ofwet amyl alcohol was shakenwith 51. of 0-05M-potassium phosphate buffer, pH 6-8,which had been made up in glass-distilled water. Themixture was allowed to stand in a constant-temperatureroom for 24 hr. before use.

Distribution procedure. In order to obtain enough of thevarious bacitracins for chemical and biological investigationthe first four tubes of the distribution machine were loadedwith the starting material at as high a concentration aspossible. This was done in the following manner: 2-9 g. ofa sample of crude bacitracin hydrochloride (previouslyreferred to as crude ayfivin hydrochloride batch 1 by Sharpet al. 1949) were dissolved in 60 ml. glass-distilled water andthe solution adjusted to pH 5-0 with 1ON-KOH. Thissolution was extracted twice with 60 ml. of a solution of8-hydroxyquinoline in ether, the first ethereal solution con-taining 50 mg. of 8-hydroxyquinoline and the second 20 mg.It was then extracted four times with 60 ml. quantities ofether alone. The first two extracts were dark green and thelast two were colourless. Ether was removed from theaqueous solution in vacuo and the latter equilibrated withthe solvent system and saturated with the crude bacitracinas follows. Sufficient potassium phosphate buffer (pH 5-0)was added to make 80 ml. of 0-05M buffer and the solutionjust saturated with both amyl and n-butyl alcohols. Thevolumewas then madeupto 80ml. with water just saturatedwith amyl and butyl alcohols, 80 ml. of the upper phase ofthe solvent system were added, and lON-KOH was added tothe lower layer until the 'pH' (glass electrode) after equili-bration was 7-0. A small amount of an oil came out of solu-tion and was removed by centrifuging. Portions (20 ml.) ofeach layer of the resulting clear solutions were loaded intotubes 0-3 of the distribution machine.The distribution was begun with 46 transfers, using the

'fundamental' procedure (Craig & Craig, 1950). Singlewithdrawal oflower layer from the trailing edge of the bandwas then carried out until the contents of tubes 0-4, whichwere forming stable emulsions, had been removed. Thematerial withdrawn in these fractions consisted of a mixtureof bacitracins E and D. The fundamental procedure wasthen continued until 97 transfers had been completed andthis was followed by 124 single withdrawals of upper layer,made from the leading edge of the band. All the materialleft in the machine had been subjected to 221 transfers.

Analysis ofdistribution curve. Thedistribution curve shownin Fig. 1 was obtained by measuring the total amounts ofpolypeptideinsamples of the contents ofthe even-numberedtubes, using the method described by Newton & Abraham(1950). Partition coefficients (K) were determined over thewhole curve from the ratios of the amounts of materialin successive even-numbered tubes (T71T7+2). This wasdone by using equation (1) for the fundamental series andequation (2) for the withdrawn series (Berridge, Newton &Abraham, 1952; Newton & Abraham, 1952).

Tr/Tr+s=F,' 1 (1r)

T'/T'-2 = Fw F, (K + 1)2' (2)

I953

F_(r+1 -j[x- 1])(n-r+i[x -1])'where

F (n-1)W(n-R - J[x+1]-1)'n =number of transfers, r =tube number, R+1 = totalnumber of tubes in the machine and x=number of tubesinitially filled with an equal amount of material.The values obtained for the partition coefficients of the

various polypeptides were: KB=0-316, KA =0-788,Kc = 1-37 and KF, = 15-0 (approx.). No precise data wereobtained for the other polypeptides by this analysis.Theoretical curves were calculated for bacitracins A, B, Cand F,, using the values of K found above. The calculatedcurves were then fitted to the maxima of the experimentalcurve for these compounds. Material from tubes adjacent tothese maxima had already been found to have practicallyconstant partition coefficients. The interpretation of theremainder of the curve is only approximate.

Concentration of solutions. The fractions selected from thecounter-current distribution (Table 1) were concentrated inthe manner described by Newton & Abraham (1950).

Acid hydroly8is. A solution containing 2-5 mg. of thebacitracin hydrochloride was evaporated to dryness in asmall Pyrex test tube; 1 ml. of 6N-HC1 was then added andthe tube was sealed and heated at 1100 for 24 hr. Afterhydrolysis the HCI was removed in a stream of air at 1000.

Ultraviolet absorption spectra. These were measured in aBeckman spectrophotometer in aqueous solution at pH 3.

Electrometric titrations. These were carried out at 200 asdescribed by Newton & Abraham (1953).Amide N. This was measured as described by Newton &

Abraham (1953).Reduction offerricyanide. This was measured as described

by Newton & Abraham (1953).

Paper chromatographyThe chromatograms were run on Whatman no. 1 paper

(18-5 x 22-25 in.) as described by Dent (1948). The hydro-lysate from 200 pg. ofpolypeptide was used when chromato-grams were run in a single dimension, from 400,ug. whenthey were run in two dimensions.

Single-dimension chromatogramswere developed with thebutanol-acetic acid mixture of Woiwod (1949). For two-dimensional chromatograms the second solvent was 80%(w/w) phenol, and the atmosphere was saturated with 50%(v/v) acetic acid (Dent, 1948). This solvent gave tolerablygood separations of ornithine, lysine and histidine. Thephenol-saturated papers were dried at 370 for 4 or 5 hr.(Brush, Boutwell, Barton & Heidelberger, 1951). Butanol-acetic acid-saturated papers were dried at 800 for 30 min.The papers were sprayed with 0-2% (w/v) ninhydrin dis-solved in wet butanol containing 0-2% (v/v) of collidine(Woiwod, 1949).

Microbiological assay of valineValine was estimated microbiologically with Leuconostoc

mesenteroides (American Type Culture Collection, no. 7881).The medium and procedure were based on those of Steele,Sauberlich, Reynolds & Baumann (1949). The tubes wereincubated for 44 hr. at 370 and growth was estimatedturbidimetrically. The response ofthe organism to DL-valineand to an acid hydrolysate of crude bacitracin coincided

602

Vol. 53 PROPERTIES OF BACI9over a wide range of concentrations. 4200 p.g. of crudebacitracin produced the same response as 117,ug. of DL-valine. Norvaline did not replace valine for growth underthe conditions used.

DNP-bacitracin derivativesBacitracin (10 mg.) was mixed with 6-5 mg. ofNaHCO3in

5 ml. water. A solution of 10 mg. of FDNB in aqueousethanol was added, and the mixture was shaken at roomtemperature for 3 hr. The contents of the flask were thenevaporated almost to dryness and the residue was suspendedin about 5 ml. of water. The suspension was extracted twicewith an equal volume of ether. The DNP-bacitracin wasprecipitated from the aqueous solution by excess acid afterfirst removing any residual ether. The yield of DNP-bacitracin was usually between 60 and 80 %, assuming thatit contained oneDNP residue for each 400-500 g. ofpeptide.Both DNP-bacitracin A and DNP-bacitracin B had sharpmelting points close to 1850.

Hydrolysis of DNP-bacitracins. About 5 mg. of DNPbacitracin were heated in a sealed tube with 1 ml. of amixture of equal volumes of glacial acetic acid and conc.HCI. After 12-14 hr. at 1050 the acid was evaporated andthe residue taken up in a small volume of water and ex-tracted several times with ether. The ether extracts weretransferred to a weighed tube and evaporated to dryness.The aqueous residue was also evaporated to dryness.

Paper chromatography of DNP-amino-acidsEther-solublefraction. The weighed residue remaining after

evaporation of the ether was dissolved in ethanol to give asolution of 75 pg./5 pl. This solution (5 pl.) was applied toWhatman no. 4 paper which had been previously soaked inpotassium phthalate buffer of pH 6-3 and dried at roomtemperature (Blackburn & Lowther, 1951). The chromato-grams were usually developed with tert.-amyl alcoholsaturated with the same phthalate buffer. Benzyl alcohol towhich 10% of ethanol (v/v) had been added before justsaturating with buffer was found particularly useful forresolving DNP-valine from a mixture ofthe DNP derivativesof leucine, isoleucine and phenylalanine. DNP-glycine,DNP-alanine, DNP-valine and DNP-leucine were easilydistinguished using these systems, but the systems did notresolve the DNP derivatives of leucine, isoleucine andphenylalanine.

Water-solublefraction. The aqueous residue from the etherextraction was evaporated to dryness and dissolved in waterto givea solution containing 150 ,ug./5 pl. This solution (5 PI.)was applied to Whatman no. 1 paper and the chromatogramwas developed with a butanol-acetic acid mixture (Woiwod,1949). Experiments with model mixtures of amino-acidsand DNP derivatives showed that in this system e-DNP-lysine and 8-DNP-ornithine could be seen as distinct yellowspots, and that gly-DNP-histidine and S-DNP-cysteinegave very pale yellow spots, which could be easily seen in

TIRACIN POLYPEPTIDES 603ultraviolet light. e-DNP-lysine occupied a position justabove the leucine-isoleucine-phenylalanine spot, while8-DNP-ornithine was above, and just resolved from, e-DNP-lysine. S-DNP-cysteine was not resolved from 8-DNP-ornithine; gly-DNP-histidine ran to a position justbelow glutamic acid.

After marking the yellow spots the paper was sprayedwith ninhydrin and the unchanged amino-acids were re-vealed in the normal way.

SUMMARY

1. A sample of crude bacitracin has been resolvedby counter-current distribution between solventsinto at least ten polypeptides. These substanceshave been named bacitracins E, D, B, A', A, C, G,F1, F2 and F3. Their partition coefficients (= conen.in alcohol phase/conen. in aqueous phase) at pH 7 0increase in this order.

2. The ultraviolet absorption spectra of the baci-tracins fall into three groups. Bacitracins E, D, Band A show broad maxima at 253 m,u. BacitracinsC and G absorb more strongly and show a sharpmaximum at 250 m,. and a shoulder at 268 m,u.Bacitracins F1, F2 and F3 all show a shoulder at253 m,. and a broad maximum at 288 mit.

3. All the bacitracins contain cysteine, ornithine,lysine, histidine, aspartic acid, glutamic acid,leucine and/or isoleucine and phenylalanine residues.However, bacitracins B, D and E differ from A inyielding on acid hydrolysis a substance whichbehaves like valine on paper chromatograms, andbacitracin C differs from A in yielding a substancewhich behaves like glycine. Smaller amounts of thevaline-like substance are obtained from bacitracinsC, G, F1, F2 and F3.

4. The bacitracins contain no free thiol group,but a thiol group appears to be liberated in all ofthem on mild acid hydrolysis. At the same timeammonia appears to be liberated from an amidegroup in bacitracins B, A, C, G, F1, F2 and F3,bacitracins E and D containing very little, if any,amide nitrogen.

We are indebted to Dr June Lascelles for a microbio-logical assay of valine, and to Mr Omar Boys for runningmany paper chromatograms. We are grateful to Mrs A. Gilesfor technical assistance.One of us (G. G. F. N.) is indebted to the Medical Research

Council for a part-time personal grant. The work has beensupported by a grant from the Medical Research Council fortechnical assistance and supply of materials.

604 G. G. F. NEWTON AND E. P. ABRAHAM I953

REFERENCES

Arriagada, A., Savage, M. C., Abraham, E. P., Heatley,N. G. & Sharp, A. E. (1949). Brit. J. exp. Path. 30, 425.

Berridge, N. J., Newton, G. G. F. & Abraham, E. P. (1952).Biochem. J. 52, 529.

Blackburn, S. & Lowther, A. G. (1951). Biochem. J. 48,126.

Brush, M. K., Boutwell, R. K., Barton, A. D. & Heidel-berger, C. (1951). Science, 113, 4.

Consden, R., Gordon, A. H., Martin, A. J. P., Rosenheim, 0.& Synge, R. L. M. (1945). Biochem. J. 39, 251.

Craig, L. C. & Craig, D. (1950). In Technique of OrganicChemi8try, vol. 3, ch. 4. Ed by Weissberger, A. NewYork: Interscience.

Craig, L. C., Gregory, J. D. & Barry, G. T. (1949). J. clin.Inve8t. 28, 1014.

Craig, L. C., Hausmann, W., Ahrens, E. H. & Harfenist,E. J. (1951). Analyt. Chem. 23, 1236.

Dent, C. E. (1948). Biochem. J. 43, 169.

Frick, G. (1952). Nature, Lond., 169, 758.Heatley, N. G. (1944). Biochem. J. 38, 61.Jorpes, E. (1932). Biochem. J. 26, 1507.Newton, G. G. F. & Abraham, E. P. (1950). Biochem. J. 47,

257.Newton, G. G. F. & Abraham, E. P. (1952). Nature, Lond.,

169, 69.Newton, G. G. F. & Abraham, E. P. (1953). Biochem. J.

53, 604.Porter, R. R. & Sanger, F. (1948). Biochem. J. 42, 287.Sanger, F. (1945). Biochem. J. 39, 507.Sharp, V. E., Arriagada, A., Newton, G. G. F. & Abraham,

E. P. (1949). Brit. J. exp. Path. 30, 444.Smith, J. D. & Markham, R. (1950). Biochem. J. 46, 509.Steele, B. F., Sauberlich, H. E., Reynolds, M. S. & Baumann,

C. A. (1949). J. biol. Chem. 177, 533.Stein, W. H. & Moore, S. (1951). J. biol. Chem. 1i0, 103.Woiwod, A. J. (1949). J. gen. Microbiol. 3, 312.

Observations on the Nature of Bacitracin A

BY G. G. F. NEWTON AND E. P. ABRAHAMSir William Dunn School of Pathology, University of Oxford

(Received 11 August 1952)

Bacitracin A, the main active member of the baci-tracin family of polypeptides, was obtained bycounter-current distribution between solvents inamounts sufficient for some of its properties to beinvestigated in detail (Newton & Abraham, 1953).Previous work with bacitracin (Sharp, Arriagada,Newton & Abraham, 1949; Craig, Gregory & Barry,1949; Newton & Abraham, 1950) indicated that thesubstance was a good deal less stable than would beexpected if it consisted only of amino-acids joinedby normal peptide linkages. The investigationsdescribed in this paper were designed largely tothrow light on the nature of the more labile portionsof the molecule of bacitracin A.

RESULTS

Composition and molecular weightof bacitracin A

Barry, Gregory & Craig (1948) reported the resultsof an analysis made by Moore & Stein, on a starchcolumn, of the amino-acids present in purifiedbacitracin. Using these results as a basis it wasfound that paper chromatograms of an acid hydro-lysate of bacitracin A appeared very similar tochromatograms of the following mixture of amino-acids in the molecular proportions shown (Newton &

Abraham, 1953): cystine, 0-5; ornithine, 1; lysine, 1;histidine, 1; aspartic acid, 2; glutamic acid, 1;phenylalanine, 1; leucine, 1; isoleucine, 2.The relative values obtained by Moore & Stein for

the amino-acids other than cystine and ornithine(which was not determined) are in agreement withthe molecular proportions of the amino-acids in thismixture. The value suggested here for cystine isbased on the sulphur content of bacitracin Ahydrochloride, which was found by elementaryanalysis to be 1 82%.

Sharp et al. (1949) concluded from experimentswith D-amino-acid oxidase that some of the amino-acids in crude bacitracin (ayfivin) had the D-con-figuration. Craig et al. (1949) isolated DL-phenyl-alanine, DL-aspartic acid and partially racemic D-isoleucine from purified bacitracin.

If the weights of the residues of the amino-acidsshown above are combined, and if the weight ofthree molecules of hydrochloric acid is added, afigure of approximately 1500 is arrived at for theweight of the minimum stoicheiometric unit ofbacitracin A hydrochloride. The presence of onlyone atom of sulphur in the molecule is possiblebecause, although cystine is finally obtained onhydrolysis, it is derived from a cysteine residue (seebelow).