54 J. T. B. SHAW 1964

2. If tryptic hydrolysis is carried out in amedium containing 6 % of lithium thiocyanate noprecipitation occurs, and the hydrolysate may beacidified and dialysed to give a solution of peptidesof high molecular weight.

3. Further hydrolysis of this fraction withchymotrypsin shows that it consists predominantlyof long sequences of amino acid residues throughoutwhich glycyl residues alternate with alanine, serine,valine and tyrosine residues. Such sequences makeup almost 90 % of the fibroin molecule.

4. Amino acid analyses of the fractions obtainedby enzymic hydrolysis are presented, and the distri-bution of the amino acids in fibroin is discussed.

The author thanks Dr F. Lucas for assistance withanalyses performed on the automatic amino acid analyser,and for a sample of Gly-Val-Gly-Tyr isolated from achymotryptic digest of fibroin. The skilled technicalassistance of Miss D. E. Pinion is also gratefully acknow-ledged.

REFERENCES

Drucker, B., Hainsworth, R. & Smith, S. G. (1953). J.Text. lnst. 44, T420.

Drucker, B. & Smith, S. G. (1950). Nature, Lond., 165, 196.

Harris, J. I. (1956). Nature, Lond., 177, 471.Kirimura, J. (1962). Bull. seric. Exp. Sta. Japan, 17,

447.Lucas, F., Shaw, J. T. B. & Smith, S. G. (1957). Biochem.

J. 66, 468.Lucas, F., Shaw, J. T. B. & Smith, S. G. (1958). Advanc.

Protein Chem. 13, 107.Lucas, F., Shaw, J. T. B. & Smith, S. G. (1962). Biochem.

J. 83, 164.Lucas, F., Shaw, J. T. B. & Smith, S. G. (1963). Analyt.

Biochem. 6, 335.Moore, S., Spackman, D. H. & Stein, W. H. (1958).

Analyt. Chem. 30, 1185.Narita, K. (1954). J. chem. Soc. Japan, 75, 1005.Rees, M. W. (1946). Biochem. J. 40, 632.Schroeder, W. A. & Kay, L. M. (1955). J. Amer. chem. Soc.

77, 3908.Spackman, D. H., Stein, W. H. & Moore, S. (1958).

Analyt. Chem. 30, 1190.Spies, J. R. & Chambers, D. C. (1949). Analyt. Chem. 21,

1249.Stokes, J. L., Gunness, M., Dwyer, I. M. & Caswell, M. C.

(1945). J. biol. Chem. 160, 35.Waldschmidt-Leitz, E. & Zeiss, 0. (1955). Hoppe-Seyl. Z.

300, 49.Zuber, H. (1961). Kolloidzschr. 179, 100.Zuber, H., Ziegler, K. & Zahn, H. (1957). Z. Naturf. 12b,

734.

Biochem. J. (1964), 93, 54

Fractionation of the Fibroin of Bombyx mori with Alkali

BY J. T. B. SHAWShirley Institute, Manchester 20

(Received 6 January 1964)

In the course of a series of experiments on thesilk fibroin ofBombyx mori, Narita (1954) describedthe use of rivanol (6,9-diamino-2-ethoxyacridinelactate) to fractionate the dissolved protein. Heused samples of fibroin that had been 'degummed'with hot aqueous ammonia solution; these weredissolved in copper ethylenediamine and theresulting solutions were dialysed against water andbrought to pH 6-2 with phosphate buffer. Byadding aqueous rivanol under controlled conditionsto such a solution Narita obtained two fractions:a precipitate which represented about 15 % of thetotal protein, and a soluble fraction which con-stituted the remainder. The precipitate, whichNarita called 'silk plastin', was amorphous byX-ray-crystallographic standards, and in composi-tion it was substantially poorer in glycine andalanine and richer in most other amino acids thanwhole fibroin. The soluble material he called 'silkfibrin'; this was 'crystalline' by the same standards

when isolated from solution by precipitation withacetone, and was composed almost entirely ofglycine, alanine and serine, with a little tyrosine andvaline.When a solution of B. mori fibroin prepared in

these laboratories was treated with rivanol underthe conditions used by Narita, a different resultwas obtained: no fractionation was effected and thewhole of the dissolved protein was precipitated.This fibroin preparation had been obtained fromraw silk by degumming with hot soap solution, andthe solution was prepared by the use of concen-trated aqueous lithium thiocyanate instead ofcopper ethylenediamine; thus the two preparationsdiffered in that Narita's had been exposed to moreprolonged and vigorous attack by alkaline re-agents. It appears that attack by alkali is capableof breaking down the fibroin molecule in such away that subsequent fractionation gives rise totwo or more grossly dissimilar fractions. There is

FRACTIONATION OF FIBROIN WITH ALKALInot at present much information concerning theeffect of limited attack by alkali upon fibroin, orindeed upon proteins in general, and it seemedworthwhile to investigate the situation more fully.

MATERIALS AND METHODS

Bombyx mori silk fibroin. Japanese B. mori silk waspurchased from H. T. Gaddum and Co. Ltd., Manchester,and was degummed as described by Drucker, Hainsworth& Smith (1953). Solutions were prepared by dissolving thefibroin in aq. 60% (w/v) LiSCN (1 g. in 5 ml.), dilutingwith water to 50 ml. and dialysing the solution in ViskingII in. cellophan dialysis tubing against a stream of wateruntil free from SCN- ions.

Chemicals. 6,9-Diamino-2-ethoxyacridine lactate was acommercial sample purporting to be of not less than 97 %purity, purchased from the Aldrich Chemical Co. Inc.,Milwaukee 10, Wis., U.S.A. I-Fluoro-2,4-dinitrobenzenewas also a commercial preparation, obtained from L. Lightand Co. Ltd., Colnbrook, Bucks. Other chemicals were ofanalytical reagent grade.

Alkali-treatment. Solutions of fibroin containing about2% (w/v) of protein were diluted with N-NaOH and withwater so that they were 1% in protein and 0-1 N in NaOH.The pH of such solutions was found to be 12-8. In most ofthe experiments described in this paper the solutions werekept at 30 in a refrigerator, and were then brought topH 6-2 by the addition of N-phosphoric acid. Such a pre-paration resulting from an alkali-treatment lasting 48 hr. issubsequently referred to as a solution of A-fibroin (Gk.Au'etv, to loose). Similar experiments were performed inwhich the solution at pH 12-8 was maintained at 250.

Solutions at pH 110, 11-5, 12-0 and 12-5 were also pre-pared from 2% solutions of fibroin in water by the additionof equal volumes of an appropriate sodium borate buffer,0-2M in Na+ ions: the pH values were finally adjusted bythe addition of a few drops ofN-NaOH or m-boric acid. Thesolution at pH 13-0 was obtained by the cautious additionof N-NaOH to the standard pH 12-8 solution describedabove.

Viscosity measurements. The flow times of solutions offibroin, 1% (w/v) in protein, at a series of controlled valuesbetween pH 11-0 and 13-0, were measured in an Ostwaldviscometer at 25', at intervals between 0 and 24 hr. Theresults were plotted in the form:

(flow time of solution\\flow time of solvent J

against time. It was not possible to make accuratemeasurements of the viscosity of fibroin solutions at pHvalues below 11-0 because of surface denaturation.

Fractionation with rivanol. Solutions of fibroin and ofalkali-treated fibroin, 1% in protein, were brought topH 6-2 with phosphoric acid, and to about 3°. Aq. 1%rivanol was then added dropwise whilst the solution wasmaintained at 3±10 in an ice-water bath. When furtheraddition of rivanol produced no more precipitate, themixture was allowed to stand for some minutes, and theprecipitate, 'silk plastin', was collected by filtrationthrough a sintered-glass crucible. It was washed with alittle ice-water, with methanol and with ether, and thendried and weighed. Addition of too much rivanol resulted

in dissolution of some of the precipitate; this redissolvedfraction could be recovered by dialysing the filtrate againstwater, when removal of the excess of rivanol resulted in itsprecipitation. The 'silk plastin' prepared in this way wasgranular in texture and pale yellow, owing to the presenceof traces of strongly bound rivanol, the last of which couldbe removed by including methanol-0-l -HCl (9: 1, v/v) inthe washing sequence. Isolation of the dissolved 'silkfibrin' fractions in a dry state was not usually necessary,as analyses and end-group assays could be performedsatisfactorily with the solutions remaining after removal ofthe excess of rivanol and the salts by dialysis; addition ofethanol or acetone to these solutions led to the productionof very stable gels, from which it proved difficult to isolatethe dry material. If a dry sample of the fraction wasrequired, the best method was to shake the dialysed solu-tion with a little activated charcoal to remove the remain-ing rivanol, and then to freeze-dry the resulting colourlesssolution: the preparation obtained in this way was onlypartially soluble in water.

Fractionation with ammonium sulphate. A-Fibroin solu-tions, about 1% (w/v) in protein and 0-1M in Na+ ions,were brought to 200 and pH 6-2, and were treated with asaturated solution of (NH4)2504, with stirring. The(NH4)2SO4 soln. was run in dropwise until 12 ml. had beenadded/100 ml. of protein solution. The mixture wasallowed to stand for 10 min. and was then centrifuged. Theprecipitate (AF-Ap) was washed once with (NH4)2SO4 soln.(0-1 saturated) and was then dissolved in 5M-urea anddialysed against very dilute sodium phosphate buffer,pH 7-8; the dialysed solution was later assayed for nitrogen.The supernatant from the centrifugation plus the washingswere dialysed against water to remove (NH4)2SO4 andbuffer salts. When necessary, the product (AF-As) wasfinally isolated by freeze-drying.

Nomenclature. It is proposed to adopt Narita's (1954)terminology in a general form and to call fractions ofalkali-treated fibroin that can be precipitated with rivanol'silk plastin' and those that cannot be so precipitated,'silk fibrin'. However, more precise terms are needed todefine the preparations of silk plastin and silk fibrin thatwe have investigated most fully, and the polypeptidesobtained from standard A-fibroin solutions by rivanol pre-cipitation are subsequently called AF-Rp (plastin) andAF-Rs (fibrin), whereas those obtained by (NH4)2SO4fractionation are called AF-Ap (precipitate) and AF-As(soluble fraction).Amino acid analyses. These were performed by the

chromatographic methods of Moore, Spackman & Stein(1958) with the minor variations described by Lucas, Shaw& Smith (1962). Some of the analyses were subsequentlyconfirmed on an automatic amino acid analyser. Thesamples were hydrolysed with boiling HCI (6N) for 24 hr.and corrections were made for losses of serine and threo-nine by hydrolytic decomposition. The results were relatedto the total nitrogen content of the solutions, assayed bythe micro-Kjeldahl technique. Tryptophan assays wereperformed on separate samples of unhydrolysed materialby the colorimetric method of Spies & Chambers (1949).

End-residue analysis. N-Terminal residues of the frac-tions, and of fibroin, were identified and assayed by theFDNB* method. The techniques are essentially those of

* Abbreviation: FDNB, 1-fluoro-2,4-dinitrobenzene.

Vol. 93 55

J. T. B. SHAWSanger (1945) but certain modifications were introduced.Treatment of the protein and peptide fractions with FDNBwas carried out in ethanol-water (8:5, v/v) in the presenceof an excess of NaHC03 and FDNB, and the reaction wasallowed to continue overnight (16 hr.) at room temperaturewith shaking, to ensure completion. The fibroin sampleswere solutions in water (1-2%, w/v) that had been pre-pared by the use of LiSCN; the fraction AF-Rs and thealkali-treated fibroins were also available as solutions, andto each of these the appropriate amount of NaHCO3,dissolved in a little water, was added, followed by ethanolicFDNB. The AF-Rp fraction, which was not water-soluble,was reduced to a fine powder, suspended in water and thensubmitted to the same procedure. The products werecentrifuged, thoroughly washed with dilute HCI, withwater and with ethanol, and dried. Hydrolysis of weighedsamples of the DNP derivatives was carried out in conc.(11-7N) HCI at 40°, and was considered to be substantiallycomplete after 21 days. Under these conditions losses ofDNP-amino acids, especially of DNP-glycine, are muchreduced and are normally less than 10%. The liberatedDNP-amino acids were isolated by column chromato-graphy on buffered Celite and estimated spectrophoto-metrically as described by Lucas, Shaw & Smith (1963).

RESULTS

Vico8ity changes. The change in specific vis-cosity of 1 % (w/v) solutions of fibroin with time atvalues between pH 11 and 13 is shown in Fig. 1.These changes were found to be irreversible whenthe pH was returned to 6. Fig. 2 illustrates the fallin specific viscosity shown by 1 % (w/v) solutionsof fibroin in 0-1N-NaOH at 250 and at 30: the pH ofsuch solutions was between 12-7 and 12-8 and didnot change perceptibly during the experiments.

Fractionation with rivanol. Solutions of fibroinwere kept at pH 12-8 at 30 and at 25° and portionswere removed at intervals and tested for precipit-ability with rivanol. The results are shown in Fig. 3.It is clear that a steady state is reached after about2 days at 30 and after about 10 hr. at 250. Afterthese times of treatment, 10 % by weight of thefibroin is precipitable with rivanol. A solution offibroin that had been treated with pH 11 buffer for3 hr. at 250 was completely precipitated withrivanol at pH 6-2; complete precipitation alsooccurred when a solution of fibroin prepared withcopper ethylenediamine as described by Druckeret al. (1953), instead of with aq. lithium thiocyanate,was tested without pretreatment.

1-

1-

1-0.,-

O 0-8

9 0-6

0-4

0-2

0

00-8

~.0-6

0-4

0-2

, I, ,, I

0 4 8 12 16 20 24Time of treatment (hr.)

Fig. 1. Action of alkaline buffers on solutions of fibroin:change of specific viscosity with time of treatment at 250.0, pH 11-0; A, pH 11-5; +, pH 12-0; 0, pH 12-5; A,pH 13-0.

Time of treatment (days)

1-2

1-0

*- 0-8m0

0-6

0-4

0-2

0 4 8 12 16 20 24Time of treatment (hr.)

Fig. 2. Action of 0-1 N-NaOH on solutions of fibroin(pH 12-8): changes in specific viscosity with time of treat-ment at (a) 30 and (b) 250.

(b)

I I I I I I

56 1964

FRACTIONATION OF FIBROIN WITH ALKALI

Fractionation with ammonium sulphate. Treat-ment of a solution of A-fibroin with ammoniumsulphate under the conditions described in theMaterials and Methods section resulted in theproduction of a precipitate (AF-Ap) representingabout 10 % by weight of the total protein. Theremaining dissolved polypeptide (AF-As) could bedialysed against water without significant loss.

Solubility properties. The AF-Rs and AF-Asfractions were soluble in water and in aqueousbuffers at all pH values. The solutions were stableand did not gel even on prolonged standing. Thedissolved peptides could be fractionally precipi-tated by the addition of trichloroacetic acid or ofethanol; this suggested the presence of a mixture ofpeptides, probably heterogeneous with respect tochain length. The AF-Ap fraction could be dis-solved in aqueous urea, and the resulting solutionsurvived dialysis. The fraction remained soluble inaqueous buffers at values above pH 5-8 and below3-5, but was insoluble between these values;between values about pH 5-8 and 7-0 the solutionwas opalescent. The AF-Rp fraction showed similarproperties after the absorbed rivanol had beenremoved.Amino acid composition. The amino acid analyses

of B. mori fibroin, and of the four principal frag-ments here described, AF-Rp, AF-Rs, AF-Ap andAF-As, are given in Table 1 together with Narita's(1954) analyses of 'silk fibrin' and 'silk plastin'.The values for serine are corrected by a factor of10 % for losses due to hydrolytic decompositionand those for threonine by 5 %. The tentative

-4

o-

Ps0

80

60

40

20

0 2 4 6 8 10 12Time of treatment (days)

Pa~

0-0

80

60

40

(b)

20 -

0 4 8 12 16 20 24

Time of treatment (hr.)

Fig. 3. Action of O lN-NaOH on solutions of fibroin(pH 12-8): precipitability with rivanol after treatment at(a) 30 and (b) 25°.

Table 1. Amino acid composition of fibroin and of 8ome fractions of fibroin

Results are given in two forms: (1) as amino acid nitrogen as a percentage of total nitrogen; (2) as amino acidresidues/100 residues.

Fibroin*,

(1) (2)43-7 44-628-8 29-42-16 2-200-52 0-530-65 0-66

11-9 12-10-89 0-911-28 1-301-00 1-020-35 0-365-07 5-170-62 0-630-63 0-320-53 0-141-83 0-470-33 0-11

AF-Rp

(1) (2)14-0 18-312-6 16-54-12 5-404-23 5-543-98 5-217-92 10-42-00 2-628-22 10-84-98 6-522-34 3-13-63 4-751-90 2-492-40 1-573-15 1-38

12-6 4-121-0 0-60-56 0-360-26 0-346-5 -

AF-Ap

(1) (2)16-3 21-214-0 18-23-84 5-003-99 5-193-86 5-028-3 10-81-77 2-307-5 9-84-39 5-712-0 2-63-66 4-782-02 2-622-21 1-432-53 1-08

11-9 3-881-05 0-690-43 0-280-26 0-307-14

AF-Rs

(1) (2)44-7 46-429-3 30-41-87 1-940-15 0-160-31 0-3210-6 11-00-78 0-811-02 1-060-79 0-820-4 0-44-77 4-950-53 0-550-44 0-220-34 0-110-56 0-140-30 0-10

1-0 -

AF-AsI

(1) (2)44-5 45-730-0 30-82-13 2-190-1 0-10-35 0-36

11-7 12-00-75 0-771-09 1-120-74 0-76

4-70 4-830-6 0-60-48 0-250-39 0-130-79 0-200-35 0-12

1-4* Results for untreated fibroin are from Lucas, Shaw & Smith (1958).t Values for 'silk plastin' and 'silk fibrin' are from Narita (1954).

'Silk 'Silkplastin't fibrin t

(1) (1)36-1 47-826-4 29-31-84 2-41

}2-75 0-4811-5 11-51-38 0-444-10 0-122-12 0-560-705-12 4-051-25 -1-68

10-7 1-29

(a)

GlycineAlanineValineLeucineIsoleucineSerineThreonineAspartic acidGlutamic acidProlineTyrosinePhenylalanineLysineHistidineArginineTryptophanHydroxylysineMethionineAmmonia

Vol. 93 57

Table 2. N-Terminal residues of fibroin and of some fractions of fibroinResults are given in two forms: (1) as moles/105 g. of protein or peptide; (2) as m-moleslmole of main-chain

nitrogen.

GlycineAlanineSerineAspartic acidOthers*

TotalMean chain length

Fibroin

(1) (2)0-19 0-150-13 0.100-18 0-140-17 0-130-14 0-110-81 0 63

1587

A-Fibroin,

(1) (2)3-80 2-961-93 1-501-16 090

} 0-35 } 0-287-24 5-64

178

AF-Rs

(1) (2)4-28 3-172-12 1-571-22 0-900-33 j0-247-95 5-88

170

AF-Rp

(1) (2)1-46 1-410-88 0-850-74 0-710-24 0-231-18 1-144-50 4-34

230

' Silkfibrin't0-930-560-66

2-15

'Silkplastin't

0-920-711-05

2-68

* This fraction includes small amounts of unidentified DNP-amino acid and some incompletely hydrolysed DNP-peptides.

t Results from Narita (1954) for 'silk-fibrin' and 'silk-plastin' fractions, in the form moles/105 g. of DNP-peptide.

280

X 240

° 200

o160

- 1 0

80

; 0 4 8 12 16 20 24Time of treatment (days)

Fig. 4. Liberation of N-terminal residues during treat-ment of dissolved fibroin with 0-1 N-NaOH at 30. O, Total;0, glycine; A, alanine; A, serine.

identification of a minor component in some hydro-lysates as hydroxylysine is based wholly on itschromatographic behaviour and needs confirma-tion. The evidence for the presence of methionineas a minor component of the fibroin molecule(Shaw, 1964) is further supported by the analysisof fractions AF-Ap and AF-Rp.

End-residue assay. The results of estimations ofN-terminal residues with dissolved fibroin, withA-fibroin and with the fractions AF-Rp and AF-Rsare presented in Table 2; the results of end-groupanalyses performed on fractions AF-Ap and AF-Aswere similar to those for the corresponding fractionsobtained with rivanol, and are not recorded.Narita's (1954) estimates of the N-terminal residuesin his 'fibrin' and 'plastin' preparations are alsoincluded for comparison. The present results are

not corrected for losses during hydrolysis and

chromatography, but such losses are likely to besmall. The rate of production of total N-terminalresidues, and of the three principal N-terminalresidues glycine, alanine and serine, when afibroin solution is maintained at pH 12-8 at 30, isillustrated in Fig. 4.

DISCUSSION

These results show Narita's (1954) originalpostulate that dissolved fibroin can be fractionatedinto two chemically distinct fragments by pre-cipitation with rivanol to be incorrect, and Naritahas subsequently withdrawn this opinion in apersonal communication; he suggested instead thatbreakdown of the fibroin molecules may result fromthe dissolution of the protein in copper ethylene-diamine. However, solutions of fibroin preparedwith copper ethylenediamine are here shown tobehave towards rivanol in exactly the same way asthose prepared with lithium thiocyanate, and itfollows that the fractionation must have been dueto some other cause. This cause is the breakdown ofthe fibroin molecule from the use of boilingammonia solution in the preparation of the fibre.Shimizu (1957) confirmed that, though dissolvedfibroin from the silk gland was completely pre-cipitable by rivanol, fibrous fibroin, boiled with aq.0 5 % ammonia solution and dissolved in copperethylenediamine, was not completely precipitable;the amount of precipitate decreased with increasingduration of the ammonia treatment.The action of alkalis on proteins has not been

very widely studied, largely because the hydrolyticprocesses are complicated by secondary effectsincluding the racemization and breakdown ofamino acids, especially serine, threonine, cystineand arginine (Sanger, 1952). However, the patternof the hydrolysis of a protein with alkali differsfrom that with acid owing to the reversal of thecharge effects and to the comparative resistance ofcertain bonds, for example those involving the

58 J. T. B. SHAW 1964

FRACTIONATION OF FIBROIN WITH ALKALIamino groups of the hydroxyamino acids. Conse-quently, hydrolysis with alkali may provide usefulresults if extreme conditions are avoided.

There is a marked contrast between the sensi-tivity of dissolved and fibrous fibroin towardsattack by alkali, and also in the nature of thisattack. The action of 2N-sodium hydroxide onfibrous fibroin at 400 has been described by Shaw &Smith (1961); the fibre was quite rapidly attacked,and after 24 hr., when 73 % of the material hadbeen dissolved, no evidence was found either fromamino acid analysis or by X-ray diffraction for anypronounced differentiation in composition orstructure between the dissolved and the undis-solved fragments. Treatment with cold 0-1 N-sodium hydroxide gives rise to considerable dis-ruption of the dissolved fibroin molecule, but hasno perceptible effect on fibrous fibroin; thisdifference is explicable in terms of accessibility.Fibroin molecules probably exist in solution asstructureless 'random-coils'; the rate and patternof attack by alkali is therefore governed by therelative labilities of the peptide bonds (or othercovalent linkages), since all are equally accessible.In these circumstances the principal factor govern-ing the rate of fission of a particular bond is thedistribution of charge in its environment, and thisin turn is determined by the sequence of the aminoacids adjacent to it.

Figs. 1-3 show that the dissolved fibroin mole-cules are attacked by alkali even at low tempera-tures and under comparatively mild conditions ofpH. The resulting molecular disruption leads to asubstantial and irreversible drop in the viscosity ofthe solutions. Figs. 1 and 2 show that the dis-ruptive effect increases markedly at pH valuesabove 12, and that in 0- N-sodium hydroxide at 30the specific viscosity of a 1 % solution of fibroin isreduced to 50 °' of its original value in about 6 hr.,and to about 15 % in 48 hr. These changes inviscosity are accompanied by changes in precipita-bility with rivanol; a comparison of the graphs ofthis property (Fig. 3) with those illustrating theviscosity change (Fig. 2) demonstrates the re-semblance very clearly.The results in Table 1 show that fractionation of

dissolved fibroin by protein precipitants after alimited and controlled attack by alkali effects theseparation of the molecule into two grossly dis-similar parts. The fractionation may be comparedwith that encountered in tryptic hydrolysis (see thepreceding paper). The amino acid composition ofthe Tp (trypsin-precipitated) fraction is strikinglysimilar to that of fractions AF-As and AF-Rs, and allthree are recovered in yields corresponding toabout 90 %O of the whole protein. This, the majorbuilding block of the fibroin molecules, has aglycine content approaching 50 %, and is composed

to a very large extent of sequences in which glycylresidues alternate with those of alanine, serine,valine and tyrosine. The remaining 10 % of thefibroin molecule is totally dissimilar, and it is theisolation of this fragment that these procedureshave made possible; it is, of course, also present inthe tryptic hydrolysate of fibroin, but as themixture of short soluble peptides that constitutefraction Ts. This minor component can be isolated,and therefore presumably occurs in the originalprotein, in long sequences of about 230 residues.

Narita's (1954) analysis and end-residue estima-tions of 'silk fibrin' and 'silk plastin' (Tables 1and 2) confirm that these fractions were derivedfrom fibroin that had been attacked by alkali, butless extensively than the A-fibroin preparationdescribed here. Narita's 'silk plastin' was com-posed of longer peptide chains than fraction AF-Rp,and its composition shows that it may be con-sidered as a preparation of AF-Rp with somesequences of AF-Rs. From the rivanol-precipit-ability values, it appears that Narita's fibroin hadbeen degraded to an extent corresponding to afibroin solution after treatment for 20-30 hr. with0-1 N-sodium hydroxide at 3°.The mechanism of the fragmentation of the

fibroin molecule by alkali is obscure. There is nowconsiderable evidence for the existence ofmore thanone species of peptide chain in the fibroin molecule;certainly end-residue assays invariably show threeor four different N-terminal residues. The chainsmay be joined together either by a severely limitednumber of cystine cross-links or by some othertype of covalent bond, perhaps ester links involvingthe plentiful seryl residues. Either of these types ofcross-link might be labile under alkaline conditionsand release of the individual peptide chains maywell be an important feature of the attack. How-ever, it is not possible to establish this unequivoc-ally, for end-group assay of the treated fibroinsdemonstrates progressive release of N-terminalresidues that can only be attributed to concomi-tant main-chain cleavage.

Fig. 3 shows that prolongation of the alkalitreatment at 30 beyond 48 hr. does not result ina further reduction of material precipitablewith rivanol; from this it follows that most ofthe peptide bonds in this fraction, derived fromphase III (for definition of terminology see Shaw,1964) of the fibroin molecule, are resistant to cleav-age by alkali. It is the sequences of alternatingglycyl residues (phases I and II) that are mostsensitive to alkaline hydrolysis, as is shown by theprogressive release of N-terminal glycyl residuesand, less rapidly, those of alanine and serine, whichcontinued throughout the 25-day period for which itwas studied. Comparison of the chain lengthsshows that fractions AF-Rs and AF-As (mean chain

Vol. 93 59

J. T. B. SHAW

length 170 residues) are somewhat degradedrelative to the corresponding Tp and Tnd fractions(mean chain length about 350 residues). Conse-quently, though alkali treatment is at present theonly method available for isolating phase III of thefibroin molecules, enzymic hydrolysis, being morespecific, is to be preferred for isolating phases Iand II.The use of rivanol as a protein precipitant was,

until recently, largely confined to the separation ofthe proteins of blood plasma and in particular tothe isolation of y-globulin (Horejsi & Smetena,1956). Subsequently, Steinbuch & Quentin (1959)isolated ceruloplasmin by stepwise precipitation ofplasma with rivanol under different conditions ofpH, and Kaldor, Saifer & Vecsler (1961) havestudied the interaction between rivanol and bovineserum albumin under different conditions of pH,ionic strength and concentration. They concludedthat, in general, the interaction of serum albuminwith rivanol resembled that with simple metalcations rather than with cationic detergents. How-ever, Narita's (1954) observations that silk plastinis redissolved by the addition of an excess ofrivanol, corroborated by our experience withfraction AF-Rp, suggests that this interpretationmay not be valid in all circumstances.

It appears that rivanol is capable of formingcomplexes with proteins that exhibit appropriatearrangements of accessible polar side chains, andthat the consequent obstruction of these polargroups renders the rivanol-protein complex in-soluble. Rivanol is a strongly cationic dye (pKa11-04) and it is not surprising that, of the plasmaproteins, y-globulin, which is not precipitated, isthe most basic. The fact that the precipitablefraction of alkali-treated fibroin (AF-Rp) has alarge excess of acidic (carboxyl) over basic sidechains, and is thus a predominantly anionicmolecule, corroborates the view that rivanolfunctions largely by combining with the acidic sidechains of electronegative proteins. It follows thatthe fractional precipitation of proteins with rivanolis likely to be critically dependent upon pH, andNarita (1954) observed that the precipitation of his'silk-plastin' fraction was most complete at valuesbetween pH 6-0 and 6-4, whereas at values belowpH 5-0 and above 8-0 the amount of precipitate wasmuch reduced. In the soluble fraction AF-Rs,which is very deficient in polar side chains, the fewacidic and basic groups are present in approxi-mately equivalent amounts and precipitation atpH 6-2 would not therefore be expected. Rivanolis not strongly bound to either fraction: it can beremoved from the dissolved AF-Rs by shakingwith charcoal and from the precipitated fractionAF-Rp by washing with methanol containing a

little dilute hydrochloric acid.

The relatively high chain length of the phase IIIfragments that have now been isolated (about 230residues), considered with the fact that this phaserepresents only about nine amino acid residues/100residues in fibroin, suggests that if the fibroin mole-cule is a unique entity it contains more than 2000residues and thus must have a molecular weight inexcess of 170 000. Recent evidence accords with theview that the molecular weight of fibroin is veryhigh; ultracentrifuge studies by Prati, Moruzzi &Centola (1958), with a solution of fibroin in aqueouslithium bromide, afforded a value 270000, andlight-scattering experiments gave values around300000 (Hyde & Wippler, 1962). Moreover theend-group estimations quoted in Table 2 of thispaper imply a mean chain weight of about 120000(taking 78 as the mean residue weight); conse-quently, if a four-chain molecule with terminalalanine, aspartic acid, glycine and serine residues ispostulated, a molecular weight of at least 400000must be envisaged for fibroin. Clearly some of themost vital questions about the fibroin moleculethat remain unanswered concern the existence andnature of interchain linkages.

It is encouraging to see that, in fractions AF-Apand AF-Rp, several amino acids occur in propor-tions that bear simple relations to one another. Forexample, leucine, isoleucine and valine occur inclosely similar molar amounts, whereas asparticacid occurs twice, and threonine about half, asfrequently. Other relations, more or less precise,can be found. Although this fraction is not a singlemolecular species, further work may show that it iscomposed of a fairly simple and tractable mixtureof components, capable of resolution into entitiessuitable for sequential studies. If this is so, it willbe possible to determine whether the uniqueregularity of structure now evident throughoutalmost 90 % of the fibroin molecule has its counter-part in the remaining fragment.

SUMMARY

1. Dissolved Bombyx mori fibroin was treatedwith 0-1 N-sodium hydroxide at 30, and the rate ofpeptide-bond cleavage was measured, togetherwith the fall in viscosity.

2. Solutions of fibroin that had been treatedwith 0-IN-sodium hydroxide at 30 for 48 hr. werefractionated by means of the cationic dye rivanol(6,9-diamino-2-ethoxyacridine lactate) and by theaddition of saturated ammonium sulphate.

3. The precipitated fractions, which repre-sented about 10 % of the total protein, werefound to be substantially different in compositionfrom the soluble fractions. They contained littleglycine and had mean chain lengths of about 230residues.

196460

Vol. 93 FRACTIONATION OF FIBROIN WITH ALKALI 61

4. The fractionation of fibroin by alkali is com-pared with that obtained by hydrolysis withtrypsin, and the relation of the fractions to thefibroin molecule is discussed.

The skilled technical assistance of Miss D. E. Pinion isgratefully acknowledged.

RBFERENCES

Drucker, B., Hainsworth, R. & Smith, S. G. (1953). J.Text. Inst. 44, T420.

Horejsi, J. & Smetena, R. (1956). Acta med. scand. 155,65.

Hyde, A. J. & Wippler, C. (1962). J. Polym. Sci. 58, 1083.Kaldor, G., Saifer, A. & Vecsler, F. (1961). Arch. Biochem.

Biophys. 94, 207.Lucas, F., Shaw, J. T. B. & Smith, S. G. (1958). Advanc.

Protein Chem. 13, 107.

Lucas, F., Shaw, J. T. B. & Smith, S. G. (1962). Biochem.J. 83, 164.

Lucas, F., Shaw, J. T. B. & Smith, S. G. (1963). Analyt.Biochem. 6, 335.

Moore, S., Spackman, D. H. & Stein, W. H. (1958).Analyt. Chem. 30, 1185.

Narita, K. (1954). J. chem. Soc. Japan, 75, 1005.Prati, G., Moruzzi, P. & Centola, G. (1958). Seta, 4, 37.Sanger, F. (1945). Biochem. J. 39, 507.Sanger, F. (1952). Advanc. Protein Chem. 7, 1.Shaw, J. T. B. (1964). Biochem. J. 93, 45.Shaw, J. T. B. & Smith, S. G. (1961). Biochim. biophys.

Acta, 52, 305.Shimizu, M. (1957). In Chemistry of Proteins, vol. 5, p. 362.

Ed. by Akabori, S. & Mizushima, S. Tokyo: KyoritsuPublishing Co.

Spies, J. R. & Chambers, D. C. (1949). Analyt. Chem. 2-1,1249.

Steinbuch, M. & Quentin, M. (1959). Nature, Lond.,183, 323.

Biochem. J. (1964), 93, 61

Acetoacetate Formation by Liver Slices from Adult and Infant Rats

BY Z. DRAHOTA, P. HAHN, A. KLEINZELLER AND A. KOSTOLANSKAIretitute of Phyaiology and In8titute of Microbiology, Czecho81ovak Academy of Sciences,

Prague 6, Czechoslovakia

(Received 13 January 1964)

It has become evident in recent years that fat isthe main energy source for most infant mammals,particularly the rat. In that animal milk representsin fact a high-fat diet (Glass, 1956), and fat is notonly rapidly utilized but is also laid down duringthe first few days after birth (Hahn & Koldovsky,1960). Reflexions of this state are high lipid con-centrations in the blood of fed infant rats (Graf-netter, Hahn, Koldovsky & Viktora, 1960) and arapid suppression of lipogenesis in the liver verysoon after birth (Villee, 1958).

Since the end product of fatty acid oxidation inthe liver is acetoacetate (Lehninger, 1945; Quastel& Wheatley, 1933; Edson, 1935), it might be expect-ed that more of this substance is formed in the liverduring the suckling period than later in life, whenthe fat content of the diet and of the liver is low.Surprisingly, this has not previously been examined.Even concentrations of ketone bodies in the bloodof young rats have been determined only afterweaning (Wick & McKay, 1940).

In the present work an attempt has been made tofollow developmental changes in acetoacetateformation by liver slices and to correlate these withchanges in the ketone-body concentration in theblood. A preliminary report ofpart of this work hasappeared (Drahota, Hahn, Kleinzeller & Kosto-lanska, 1963).

MATERIALS AND METHODS

Animals. Male white Wistar rats fed on a standardlaboratory diet (Fabry, 1959) were used throughout. Ratsyounger than 23 days were employed regardless of sex.Litters were always reduced to eight on the second dayafter birth. When animals were starved, rats younger than23 days were kept at 300 without the mother animal in anincubator. Adult animals were kept at room temperature(about 220) and were deprived of both food and water.

Tissue preparations. Rats were decapitated and thelivers were rapidly removed and chilled. Tissue slices werecut by hand. They were gently dried on filter paper andweighed on a torsion balance, and 50-80 mg. portions placedin chilled Warburg vessels containing 3 ml. of salinemedium (see below). The slices were incubated at 370 inoxygen in a Warburg apparatus. The central well of eachflask contained 0-2 ml. of 20% (w/v) KOH solution. Sincehigh concentrations of phosphate inhibit acetoacetate for-mation (Quastel & Wheatley, 1933), the composition of thesaline medium was adjusted to the following: NaCl,134 mM; KCI, 5-6 mM; CaCl2 2-25 mM; potassium phosphate,4-2 mM; MgSO4, 1-4 mM; the final pH was 7-2. Otheradditions were made as indicated in the text. The Qo2 andrespiratory quotient (R.Q.) values were determined by thedirect Warburg method (see Dixon, 1943).

Analytical methods. Acetoacetate was determined colori-metrically according to the method of Walker (1954).Sodium acetoacetate used for calibration was preparedaccording to the method of Ljungren (1924) and standard-ized according to the method of Edson (1935).