ce ae BGEat SS) aahs ah Pee seed bases
>
es pata 3 = ies
retarted
«aw oN pao Ep A
rie bag Beatle gets
ae see ie ee =
+ 4 Sore
=
"» a ay Sa
2 ‘petace oles Nie nee LN on Gy Le Sey = atng Wane TES she ced Aobwe tbat 2 tS
val i
vy [AF
OF own
? ve P)
Th 1
{I To Lt
| r
CAMBRIDGE UNIVERSITY PRESS C. F. CLAY, Manacer LONDON: FETTER LANE, E.C. 4
H. K. LEWIS & CO., LTD., 136, GOWER STREET, LONDON, W.C. I WILLIAM WESLEY & SON, 28, ESSEX STREET, LONDON, W.C. 2 CHICAGO: THE UNIVERSITY OF CHICAGO PRESS BOMBAY, CALCUTTA, MADRAS: MACMILLAN & CO., LTD. TORONTO: J. M. DENT & SONS, LTD TOKYO: THE MARUZEN-KABUSHIKI-KAISHA
All rights reserved
THE
BIOCHEMICAL JOURNAL
EDITED FOR THE BIOCHEMICAL SOCIETY
BY
W. M. BAYLISS, F.R&3.
AND
ARTHUR HARDEN, F.R-S.
EDITORIAL COMMITTEE
Dr G. BARGER Pror. F. KEEBLE Pror. V. H. BLACKMAN Pror. B. MOORE Mr J. A. GARDNER Pror. W. RAMSDEN Pror. F. G. HOPKINS Dr E. J. RUSSELL
VOLUME XIII 1919 6% Bit Wao ( et
CAMBRIDGE AT THE UNIVERSITY PRESS
bg rg
Digitized by, e{te te Archive f
in 2009 wi h-funding from
Ontario Council of | iversity Libraries
Ve Cof, &
http://www.archive.org/details/biochemicaljourn 13bioc
oor rh
CONTENTS
No. 1 (May)
I. Oxidising Enzymes. I. The Nature of the“ Peroxide” naturally associated with certain direct Oxidising Systems in Plants. By M. WHELDALE ONSLOW
II. The Effects of Acids, Alkalies, and Sugars on the Growth and Indole Formation of Bacillus coli. By F. J. 8. Wyetu. (With One figure)
III. Observations on the Albuminoid Ammonia Test. By E. A. Cooper and J. A. HEwarp
IV. The Composition of Starch. Part I. Precipitation by Colloidal Iron. Part II. Precipitation by Iodine and Electrolytes. By J. MELLANBY . : ; ;
V. Observations on the Accuracy of Different Methods of Measuring small Volumes of Fluid. By F. W. ANDREWES
VI. On the Separation of Antitoxin and its Associated Proteins from Heat-denaturated Sera. By A. HomMER
VII. On the Increased Precipitability of Pseudoglobulin and of its Associated Antitoxin from Heat-denaturated Solutions. By A. Homer. (With Three figures)
VIII. The Relation of Sugar Excretion to Diet in Glycosuria. By J. Metiansy and C. R. Box. (With Six figures)
~ IX. Note on the Réle of the Antiscorbutic Factor in Nutrition. By J. C. DRumMonpD
X. Researches on the Fat-soluble Accessory Substance. I. Ob- servations upon its Nature and Properties. By J.C. DRumMonp .
XI. Researches on the Fat-soluble Accessory Substance. II. Ob- servations on its Réle in Nutrition and Influence on Fat Metabolism. By J. C. Dkummonp
PAGE
10
bo Or
56
81
vi JONTENTS
No. 2 (Juuy)
XII. Note on Xerophthalmia in Rats. By EK. C, BuLLEY
XIII. The Preparation of Silica Jelly for use as a Bacteriological Medium. By A. T. Leaa
XIV. Electrical Conductivity as a Measure of the Content of
Electrolytes of Vegetable Saps. By D. Haynes
XV. Contributions to the Study of the Vegetable Proteases. By EK. R. Fisher
XVI. On the Estimation of Sugar in Blood. By H. MacLean. (With Two figures)
XVII. The Picric Acid Method for the Estimation of Sugar in Blood and a Comparison of this Method with that of MacLean. By O. L. VAUGHAN DE WEssELow. (With One figure)
XVIII. Enterointoxication—its Causes and Treatment. By A. Distaso and J. H. SuGpEN
XIX. The Action of Ultra-Violet Rays on the Accessory Food Factors. By 8. 8. Zrrva. (With Four figures)
XxX. The Influence of Deficient Nutrition on the Production of Agelutinins, Complement and Amboceptor. By 8. 8. Zinva. (With Hight figures) ,
XXI. The Nomenclature of Blood Pigment and its Derivatives. By W. D. Haturpurton and O. RosENHEIM
XXII. The Anti-Scorbutic Value of Dry and Germinated Seeds. By H. Cuick and E. M. Dexr. (With Five figures)
XXIII. The Effect of Alcohol on the Digestion of Fibrin and Caseinogen by Trypsin. By E. 8. Ente
PAGE
103
LO7
Li
148
195
199
219
CONTENTS
No. 3 (NOVEMBER)
XXIV. Studies on Coagulation. I. On the Velocity of Gelation and Hydrolysis of Gelatin Sol. By R. Sxosr. (With Three figures)
XXV. Nitrogen Partition in the Urine of the Races in Singapore. By J. A. CAMPBELL
XXVI. Chemical Structure and Antigenic Specificity. A Com- parison of the Crystalline Egg-albumins of the Hen and the Duck. By H. D. Dakin and H. H. Date. (With Four figures)
XXVII. The Role of the Plasma Proteins in Diffusion. By T. H. Mitroy and J. F. DoneGan. (With Four figures)
XXVIII. The Effect of Methods of Extraction on the Composition of Expressed Apple Juice, and a Determination of the Sampling Error of such Juices. By D. Haynes and H. M. Jupp : : :
XXIX. A Comparison between the Precipitation of Antitoxic Sera by Sodium Sulphate and by Ammonium Sulphate. By A. Homer. (With Hight figures) ‘ : :
XXX. The Direct Replacement of Glycerol in Fats by Higher Poly- hydric Alcohols. Part I. Interaction of Olein and Stearin with Mannitol. By A. Lapwortn and L. K. Pearson
XXXI. The Direct Replacement of Glycerol in Fats by Higher Polyhydric Alcohols. Part Il. The value of Synthetic Mannitol Olive Oil as a Food. By W. D. Hatirsurton, J. C. DRumMonp and R. K. CANNAN : Cer ; é : . :
XXXII. Relative Anti-Scorbutic Value of Fresh, Dried and Heated Cow’s Milk. By R. E. Barnes and EK. M. Hume. (With Three figures)
vii
248
278
301
306
viii CONTENTS
No. 4 (DrcempBerr) XXXII. On the Mechanism of Oxalie Acid Formation by Aspergillus niger. By H. Raisrrick and A, B. CLARK
XXXIV. On the Self-Purification of Rivers and Streams. By the late A. EK. Cooper; EK. A, Cooper and J. A. Hewarp
XXXV. On the Digestibility of Cocoa Butter. Part I. By J. A. GARDNER and F, W. Fox
XXXVI. A New Method for Preparing Esters of Amino Acids. Composition of Caseinogen. By F. W. ForEMAN
XXXVII. On Amino-Acids. Part Il. Hydroxyglutamic Acid. By H. D. Dakin : ; é : : : : :
XXXVIII. Studies in the Acetone Concentration in Blood, Urine, and Alveolar air: I. A Micromethod for the Estimation of Acetone in
Blood, based on the lodoform Method. By E. M. P. Wipmark. (With ~
Two figures)
XXXIX. Studies on the Cycloclastic Power of Bacteria. Part I. A Quantitative Study of the Aerobic Decomposition of Histidine by Bacteria. By H. Ratstrick. (With Seven figures)
XL. Nitrogen Metabolism in Saccharomyces cerevisiae. By L. H. Lampirt. (With Three figures)
XLI. The Relationship of Lecithin to the Growth Cycle in Crus- tacea. By J. H. Paut and J. 8. SHarpe. (With One figure)
INDEX
PAGE
329
368
378
398
430
446
487
49]
aad
aS eine a ~
ae Sa
ee
a al
rem ae
ip sOcLDISING ENZYMES... Lo THE NATURE OF THE “BEROXIDE” NATURALLY ASSOCIATED WITH CERTAIN DIRECT OXIDISING SYSTEMS IN PLANTS. ; !
By MURIEL WHELDALE ONSLOW. From the Biochemical Laboratory, Cambridge.
(Received December 15th, 1918.)
Tue following account records some experiments which throw additional light on the nature of the oxidising enzymes of plants. The dual nature of an oxidase, 7.e. peroxide-peroxidase, and the fact that one component, the peroxidase, is an enzyme, has been established by previous workers. The resolution of the system into its component parts has also been effected to some extent.
The present work shows that the peroxide may arise by oxidation of an aromatic compound of a particular type of structure, the oxidation being activated by the peroxidase itself. It is also shown that the aromatic com- pound can be separated from the peroxidase by a purely chemical method, and that the two components can afterwards be reunited again. The peroxide once formed cooperates, in the usual way, with the peroxidase in carrying out further oxidation reactions such as are characteristic of “oxidases.”
DIRECT AND INDIRECT OXIDATION AND THE GENERALLY ACCEPTED INTERPRETATION.
It has been known for a long time that the tissues of many plants dis- colour or turn brown on injury, such as bruising, e.g. Potato tuber, Apple fruit?. The same phenomenon may be brought about in many cases by ex- posure of the tissues to chloroform vapour. A certain number of plants, on the other hand, do not show discoloration under similar treatment.
Further, it has been shown that the expressed juices, or water extracts, of plants which exhibit discoloration on injury, when added to guaiacum tincture immediately produce a blue colour. The juices, or water extracts, of plants which do not discolour, do not give a blue colour with guaiacum tincture until hydrogen peroxide has been added.
1 The discoloration on injury has been found to be characteristic of many or all members of certain Natural Orders, notably the Umbelliferae, Labiatae, Compositae and others. In other cases, only a fraction of the genera of an order apparently exhibit the phenomenon. Finally, in some orders, e.g. Cruciferae, the phenomenon is rare or possibly entirely absent [Wheldale, 1911]
Bioch. xt l
2 M. W. ONSLOW
It has been assumed, on the evidence of many investigators, that the phenomenon of discoloration of tissues on injury is connected with the action of oxidising enzymes.
The hypothesis generally accepted in regard to oxidising enzymes is one which classifies these substances as direct and indirect. When either the juice, or water extract, of a tissue gives immediately a blue colour, ¢.e. an oxidation product, with guaiacum solution, the tissue is said to contain a direct oxidising enzyme, or oxidase. If, however, the blue colour only appears after addition of hydrogen peroxide, the tissue is said to contain an indirect oxidising enzyme, or peroxidase. It has been postulated that, in general, a system which turns guaiacum blue consists of a peroxide and a peroxidase: the peroxidase acts upon the peroxide and transfers oxygen in an active state to readily oxidisable substances, such as guaiacum. In the case of plant tissues which give the direct reaction, it has been suggested that some organic substance in the plant acts as a peroxide, while in tissues giving the indirect reaction, the peroxidase only is present, and the peroxide may be supplied artificially in the form of hydrogen peroxide.
The original conception of the hypothesis of oxidising enzymes, which we owe to Chodat and Bach, was rather more complex than that outlined in the foregoing paragraph. These authors regarded an oxidase as consisting of two components, a peroxidase (as described above) and an oxygenase. “Als Oxygenasen bezeichnen Chodat und Bach den supponierten eiweisshaltigen oder organischen Anteil der bisherigen Oxydasen, der als Peroxyd bildender Korper sich mit dem Luftsauerstoff addierend, sich mit ihm zu einem K6rper
: : a) : der allgemeinen Formel F< | verbindet. Mit anderen Worten, es sind fer-
mentartige Korper, die sich mit dem Sauerstoff der Luft zu einem Peroxyd verbinden kénnen. Sie werden, wie andere fermentartige Kérper, durch Hitze zerstort, durch starken Alkohol gefallt, konnen vergiftet und geschadigt werden. Sie unterscheiden sich von gewohnlichen Peroxyden nur dadurch, dass sie wahrscheinlich hochmolekulare Kérper sind.’ [Chodat, 1910.]
DEVELOPMENT OF THE ABOVE HYPOTHESIS SUGGESTED BY THE PRESENT WORK.
The following modification of the hypothesis of oxidising enzymes, or, at any rate, of a certain class of these substances, is suggested from experimental evidence to be described later.
In plants, which brown on injury and give the direct oxidase reaction, there is present a peroxidase and, in addition, some aromatic compound con- taining the dihydroxy grouping characteristic of catechol. On injury or autolysis, the peroxidase itself activates the oxidation of the aromatic com- pound, and the oxidised product constitutes a peroxide (of Chodat and Bach). The peroxide-peroxidase system so formed will then blue guaiacum tincture.
&
EE a a
OXIDISING ENZYMES 3
It would appear, moreover, that in the above-mentioned plants the peroxidase is always associated with compounds containing the “catechol” grouping.
It has been found possible to prevent the formation of the peroxide- peroxidase system by extracting the aromatic compound with alcohol, leaving the peroxidase in the tissue residue. The peroxidase can then only give the indirect reaction. On adding the extract of the aromatic substance to the peroxidase, the system can now be synthesised and will give the direct re- action.
The above provides an extension of the system of Chodat and Bach, in that the peroxidase activates the formation of its own peroxide.
In plants, on the other hand, which give the indirect reaction and do not brown on injury, it appears that the peroxidase is not associated with com- pounds containing the “catechol” grouping, nor will these enzymes activate the oxidation of such compounds.
Aromatic compounds, such as those suggested, containing the dihydroxy grouping of the phenol catechol
OH
are widely distributed in plants [corroborated by Wolff, 1917 and Wolff and Rouchelman, 1917]. Such substances may be, in some cases, protocatechuic acid, protocatechuic aldehyde, caffeic acid or derivatives of these:
COOH CH=CH .COOH ax aN } | SO se aoe OH OH
There is evidence (based on the characteristic ferric chloride reaction) that plants, which brown on injury and give the direct reaction, do contain sub- stances of this nature and that they are absent from plants giving only the indirect reaction.
It has been shown that the peroxidases of plants which give the direct reaction when added to solutions of catechol, protocatechuic and caffeic acids (on neutralisation), give brown oxidation products which are in effect per- oxides. On addition of guaiacum to the mixture of the oxidised product and peroxidase, a blueing takes place. The same result is also obtained using the aromatic substance extracted from the plant itself.
The peroxidases of plants which give only the indirect reaction, on the other hand, do not act upon catechol, protocatechuic acid, etc., and no system blueing guaiacum is formed in their presence.
Hence it is to be pictured that, on injury or autolysis of tissues which dis- colour, the peroxidase comes into contact with a “‘catechol-lke” substance whose oxidation it activates. Extracts or juices of the tissues will then blue guaiacum.
1—2
~-
a M. W. ONSLOW
The conception of peroxidases, which oxidise polyphenols in the absence of hydrogen peroxide or other peroxide, differs from that formulated by Chodat and Bach. It might be maintained that the peroxidase used in the present account had traces of peroxide as impurity: even if this were so, the enzyme must still be defined as a peroxidase, since it is without action on guaiacum except in the presence of added hydrogen peroxide.
PREPARATION OF THE PEROXIDASE.
The employment of either expressed juices or aqueous extracts of tissues is not recommended for use in experiments on oxidising enzymes, since the presence of numerous other substances (sugars, tannins) largely interferes with the action of the enzymes. Moreover, in such liquids, on standing, re- actions of many kinds may take place including those of reduction. Hence the following method of procedure has been employed, and either the fruit of the Pear or the tuber of the Potato may be used as material, though the method has been applied to other plants. The tissues are very rapidly pounded and extracted with cold alcohol, whereby the enzymes are precipitated and retained in the cell residue, while other substances, such as sugars and aro- matics, are more or less completely removed.
The tissue of the Pear fruit constitutes a simpler case than that of the Potato tuber, since the latter contains, in addition to a peroxidase, a tyrosinase and a substance which it oxidises, presumably tyrosine. It is to the latter reaction that most of the darkening of expressed Potato juice is due. Tyro- sinase, however, has no action on guaiacum [Chodat, 1910]. If Pear fruit is employed, it is advisable to select a variety which browns rapidly on injury, as the aromatic content, and possibly the activity of the peroxidase, may vary in different varieties.
For preparation of the peroxidase, a pear (or potato) is peeled, and fiat it are cut thin slices which are rapidly pounded in a mortar, after adding sufficient 96% alcohol to prevent, as far as possible, any exposure of the tissue to air. The alcohol is then quickly filtered off on a filter pump. The residue can then be ground with more alcohol and filtered. The process should be repeated two or three times, and the grinding should be very thorough. After the final filtering, a white powder is left containing the insoluble residue of the cells including the peroxidase.
If a cold water extract of the residue be made and filtered, the following points can be demonstrated:
(a) It does not darken on standing in air. (A very slight darkening may take place after standing a number of hours.) 2
(b) Added to guaiacum tincture, it produces no blue colour within the time usually allowed for a positive reaction.
(c) Added to guaiacum tincture and hydrogen peroxide, a deep blue colour is obtained.
OXIDISING ENZYMES 5
From the above results it appears that the alcohol has removed some part of the system responsible for darkening and for the direct reaction. This point will be considered again later (see section on extraction of aromatic sub- stances). |
Another point to be noted is that if the grinding is not sufficiently rapid, the residue may be discoloured, in which case the water extract of the residue may give some direct reaction. This will also be considered later (see section on laccase).
Similar results to the above were obtained with the tissues of fruits of the Apple and Greengage.
ACTION OF THE PEROXIDASE ON SOLUTIONS OF VARIOUS PHENOLS AND AROMATIC ACIDS.
_ If an aqueous solution of the Pear (or Potato) peroxidase (as prepared in the previous section) is added to a dilute solution of pure catechol, a yellowish tint is rapidly developed which eventually deepens to a yellowish-brown. It was also shown by performing the experiment in a closed vessel attached to a gas burette that there was an absorption of oxygen during the action of the peroxidase on catechol. If some of this oxidised solution, also containing peroxidase, is added to guaiacum tincture, the latter is immediately turned blue.
With a solution of protocatechuic acid, instead of catechol, the same result is obtained, but only if the protocatechuic acid is first neutralised.
Solutions of the peroxidase, together with guaiacum tincture, were also added to solutions of other phenols, 7.e. phenol, quinol, resorcinol, pyrogallol and phloroglucinol, and to various acids, 7.e. gallic, tannic, benzoic and sali- cylic, each acid being neutralised with sodium carbonate before the addition of the enzyme. The solution of pyrogallol turned brown, and there was some darkening with quinol, but when guaiacum was subsequently added, there was no appreciable blueing in any case.
It would thus appear that the Pear and Potato Pera dah are able to activate the oxidation of compounds containing the catechol nucleus, and that the oxidation products can act as peroxides as regards the oxidation of guaiacum.
The action of the Pear and Potato enzymes on a crude solution of caffeic acid was examined in the following way. Coffee berries contain cafleatannic acid which is said to be a glucoside of caffeic acid. The ground berries were extracted with hot water, filtered, and the filtrate precipitated with lead acetate. A yellow precipitate of the lead salt of caffeatannic acid separated out. This was decomposed with sulphuric acid, filtered from lead sulphate and neutralised. When the Pear peroxidase was added to this neutralised extract, followed by guaiacum tincture, a blue colour was obtained. It is to be supposed that the berry contains some free caffeic acid, since the caffeatannic acid has both hydroxyls replaced by sugar.
6 M. W. ONSLOW
A similar browning of catechol solution, followed by blueing when guaiacum tincture is added, was obtained with enzymes from the following tissues (prepared in the way above described by treating with cold alcohol, and extracting the tissue residue with water): fruit of Apple and Greengage: leaf of Pear: flowers of Horse Chestnut (sculus Hippocastanum), and of a white variety of Foxglove (Digitalis). There is reason to believe that the enzymes of all plants which brown on injury and give the direct oxidase reaction (such as those above mentioned) will behave in the same way towards catechol and subsequently guaiacum.
The peroxidase from the fruit of Apple behaves like that of the Pear and Potato towards protocatechuic acid and crude caffeic acid, and also subse- quently towards guaiacum.
On the other hand enzymes (by alcohol) from leaves of Yellow Alyssum, flowers of Mallow (Malva moschata var. alba), white Stock (Matthiola) and white Arabis were without action on catechol, and there was no blueing on subsequent addition of guaiacum tincture. The tissues of these plants give no browning on injury and no direct guaiacum reaction.
EXTRACTION OF AROMATIC SUBSTANCES,
This is carried out by cutting pears (or potatoes) into thin slices as rapidly as possible, and dropping the slices immediately into boiling 96 °% alcohol. After boiling for a time, the hot alcohol is filtered off. The filtrate is then distilled «x vacuo until all the alcohol is removed. The residue consists of a comparatively small bulk of a turbid aqueous solution of those substances which have been extracted by alcohol, the water being largely derived from the tissue of the fruit or tuber. To this extract, after filtering, a concentrated solution of lead acetate is added. A pale yellow precipitate is formed, and the acetate is added until no more precipitate is produced. The latter is filtered off and washed. It is then decomposed with the minimum amount of sulphuric acid, and the lead sulphate filtered off.
A portion of this filtrate is nearly neutralised (to litmus) with caustic potash solution, and with it the following tests are made:
(a) On addition of a few drops of 4% ferric chloride solution, a green colour is produced. On further addition of a little dilute sodium carbonate, the green colour changes to bluish-purple and finally to purplish-red.
(b) An aqueous extract of the Pear (or Potato) peroxidase (prepared by alcohol) is added. The solution of aromatics, which has only a yellowish- brown tint, rapidly darkens to a deeper brown on standing about an hour.
(c) On addition of guaiacum tincture to the solution in (6) at any time within an hour after the addition of the peroxidase, a blue colour is obtained.
On addition of guaiacum tincture to the solution of aromatics which has been allowed to stand for some time without the presence of peroxidase, no blue colour is obtained within the time usually allowed for its appearance.
OXIDISING ENZYMES 7
[The result in (a) is also given by extracts of the Apple and Greengage fruits and leaves of Pear, all of which brown on injury. There is no such result with extract of Alyssum leaves, which do not brown on injury. |
The remainder of the filtrate is then extracted with ether several times and the ether distilled off. The residue ig taken up in water and the same test with ferric chloride repeated. It will be found to give the same reaction. It is not advisable to use the ether extract for testing with the oxidising enzyme, since ether itself may contain peroxides which complete the system. |
From the above observations it is clear that hot alcohol will extract from the fruit (or tuber) a substance which is precipitated by lead acetate and is also soluble in ether. It gives a reaction with ferric chloride and sodium car- bonate which is characteristic of catechol, protocatechuic acid, protocatechuic aldehyde and caffeic acid, all of which substances contain the ortho-dihydroxy grouping. This reaction will be referred to as the “catechol” reaction. It is also clear that the oxidation of the substance extracted in this way can be activated by the peroxidase, and the product of oxidation can function as a peroxide thus completing the system which oxidises guaiacum.
If the solution is neutralised before extraction with ether, the latter does not appear to extract the aromatic substance.
The precipitation by lead acetate, and the solubility in alcohol and ether, point to a resemblance between the aromatic substance of the Pear (or Potato) and both catechol and protocatechuic acid.
A similar and probably identical substance giving the “catechol” reaction was obtained by Reinke [1882] by hydrolysing Potato juice with hydrochloric acid and extracting with ether. It had been previously shown by Preusse [1879] that from a neutral or alkaline aqueous solution catechol can be ex- tracted by ether, but not protocatechuic acid. Reinke obtained no substance giving the ‘“‘catechol” reaction on ether extraction of the hydrolysed potato juice after neutralisation, and hence he concluded that the substance he described was not catechol. This conclusion appears to be confirmed by the experiments described in the present account. Reinke was also unable to identify his product with protocatechuic acid on account of its ready solu- bility in water, protocatechuic acid being only slightly soluble. Reinke suggests caffeic acid as the only alternative. The fact that the substance described in the present account is not extracted by ether from neutral aqueous solution indicates that it is probably an acid, but for complete identifi- cation it would be necessary to prepare it on a larger scale.
RELATIONSHIP BETWEEN THE PEROXIDASES OF THE PEAR AND POTATO AND LACCASE.
The direct oxidases of plants have generally been termed laccases by previous investigators [Chodat, 1910]. They are soluble in water, from which solution they can be precipitated by alcohol. Hence the method of preparation is usually by precipitation of the expressed plant juices by alcohol.
8 M. W. ONSLOW
From results described in the present account it appears possible that many of the so-called laccases may be produced by the adsorption of a certain amount of the oxidised aromatic (produced during crushing and extraction) by the precipitate (by alcohol) containing the crude peroxidase. On redis- solving in water, both peroxide and peroxidase will be present. It has been noted already that if the Pear (or Potato) tissue is either allowed to brown while pounding with alcohol, or is insufficiently extracted, the residue may be coloured, and in this case the water extract of the residue will give a direct guaiacum reaction.
It has been found that in the case of some tissues giving the direct reaction, it is practically impossible to prevent browning during maceration with alcohol, so that the residue is always somewhat coloured, and consequently the water extract gives the direct reaction.
INHIBITION OF OXIDISING ENZYMES.
It has been observed by several investigators that the action of oxidising enzymes may be inhibited by the presence of tannin, and, possibly, sugars. There is reason to believe that many plants may contain an enzyme of the type of that in the Pear fruit and Potato tuber, and also an aromatic with the catechol grouping, and yet, on injury, no browning takes place owing to inhibition by tannin present in the tissues.
CONCLUSIONS.
1. The tissues of many plants (e.g. fruit of Pear, tuber of Potato), on injury or exposure to chloroform vapour, turn brown: in other plants no browning occurs. Extracts of tissues which brown give the direct oxidase reaction with guaiacum: of those which do not brown, the indirect reaction only.
2. It would appear that the direct oxidase system in the Pear fruit and Potato tuber is due to the presence of a peroxidase (which blues guaiacum only on addition of hydrogen peroxide) and an aromatic substance giving the reaction characteristic of the catechol grouping. On imjury, the peroxidase activates the oxidation of the aromatic with the formation of a peroxide. The peroxide-peroxidase system so formed will then blue guaiacum.
3. The aromatic substance giving rise to the peroxide can be extracted and separated from the peroxidase, thus preventing formation of the system. The system can be synthesised afterwards by combining the extracted aromatic and the enzyme.
4. There is evidence that in plants in general which brown on injury the peroxidase is associated with an aromatic substance giving the reaction characteristic of the catechol grouping. In such plants the peroxidases activate the oxidation of the aromatic substances (either obtained from their tissues or supplied from an artificial source) giving rise to peroxides, and the system
~ <——_—"
i il ta id
OXIDISING ENZYMES 9
peroxide-peroxidase will then blue guaiacum. (In some plants however of this type the action may be masked or inhibited on injury by presence of tannins or other substances.)
5. Plants which do not brown on injury (and in which there is no apparent inhibition) do not contain a substance with the catechol grouping, and their enzymes do not catalyse the oxidation of substances with such a grouping.
This work is part of an investigation of oxidase action and fruit discolora- tion carried out for the Food Investigation Board under the general direction of Dr F. F. Blackman at the Biochemical Laboratory, Cambridge.
REFERENCES.
Chodat (1910). Handbuch der biochemischen Arbeitsmethoden. E. Abderhalden, Berlin, 3 (1), 42. Preusse (1879). Zeitsch. physiol. Chem. 2, 324.
Reinke (1882). Zeitsch. physiol. Chem. 6, 263. *
Wheldale (1911). Proc. Roy. Soc. B. 84, 121.
Wolff (1917). Ann. Inst. Past. 31, 92.
Wolff and Rouchelman (1917). Ann. Inst. Past. 31, 96.
Il. THE EFFECTS OF “ACIDS, ALKALIES, AND SUGARS ON THE GROWTH AND INDOLE FORMATION OF BACILLUS COLI.
By FRANK JOHN SADLER, WYETH.
From the Institute for the Study of Animal Nutrition, School of Agriculture, Cambridge University.
A Report to the Medical Research Committee.
(Received January 16th, 1919.) THe EFrect oF ALKALIES ON THE GROWTH OF BaciLLuS COL.
In an earlier paper [Wyeth, 1918] it was shown that the final reaction pro- duced by B. coli grown in 2 % glucose peptone was dependent upon the initial reaction of the medium, and was not a “ physiological constant” as had been suggested by Michaelis and Marcora [1912].
The results obtained by the use of different media and acids were recorded for various initial reactions lying between Py, = 4:23 and absolute neutrality (Py = 7-00). Since the introduction of the hydrogen electrode as a means of measuring the reaction (P4,) of liquids few results of the investigations of the growth of B. coli in media initially more alkaline than Py = 7-00 have been recorded. It was thought desirable, therefore, to supplement the former research by making an examination of the behaviour of B. coli when grown in 2 % glucose peptone made alkaline by the addition of N sodium hydrate: an endeavour also being made to determine a “‘limiting value” of alkalinity, above which the growth of the organism is inhibited.
Material and Experimental Methods.
Pure cultures of B. coli obtained from human faeces were used. Experi- ments described in a former paper indicate that no differences in behaviour exist between strains of human and bovine origin. In this connection it may be noted that Murray [1916] made a comparative study of B. coli isolated from the faeces of man, horse, and cow. He agrees that all the strains prepared exhibited remarkable similarity of behaviour, and especially as regards their acid production.
A sterile 4°% glucose peptone medium was prepared by the method described in the former paper. It was then rendered alkaline by the addition
GROWTH AND INDOLE FORMATION OF B. COLI 11
of N sodium hydroxidet. If the alkali be added before sterilisation, caramelisa- tion,—with consequent production of acids,—occurs and the medium does not attain the desired degree of alkalinity. The following procedure was therefore adopted. A number of flasks each containing 125 cc. of 4 % glucose peptone were prepared. To the contents of each flask the required volume, viz. (125 — x) ece., of distilled water was added. The media were then sterilised by heating in an autoclave for 1 hour at 120°, allowed to cool, and finally the necessary volume (xz cc.) of sterile N NaOH was added to the contents of each flask, which finally contained 250 ce. of sterile 2 % glucose peptone rendered alkaline with x cc. of N NaOH. In the series of experiments per- formed, the value of z was varied from 0 to 7-0 cc. of N/10 NaOH per 10 ce. of medium, and the initial reactions of the media ranged from Py = 7:0 to Py = 11-0. Inoculation was performed by adding to each 250 cc. of medium 0:5 ec. of a pure culture of B. coli grown for 18 hours in 2 % glucose peptone. The inoculated media were then incubated at 37° for 9 days, previous experi- ments having shown that the fermentation, as measured by change of final reaction (P,,;) was practically complete at the end of the eighth day.
Results of Inoculation Experiments.
In the data given below H,, Hg, etc. refer to strains of B. coli obtained from human faeces. Final Py, is the lowest value recorded during a period of 216 hours, and the + sign shown in column 5 of the table indicates that a positive result was obtained as regards fermentation, etc., while “0” indicates that a negative result was obtained.
Table I. B. coli grown in 2 % glucose peptone rendered alkaline by the addition of N NaOH.
Strain of B. coli used, H,. Time of incubation at 37°, 216 hours. Temperature of experiment, 20°.
Ce. of N/10 Ce. of N/10 Ce. of N/10 NaOH* per Tnitial Final NH, produced volatile acids No. of 10 ce. of reaction reaction Fermen- per 10 ce. of | produced per flask medium Py : Py tation medium 10 ce. medium 1 7-0 10-80 ~- 0 os — 2 6-0 10-50 a= 0 -- _ 3 5-0 10-28 - 0 — — + 4-0 9-82 4-82 + 0-02 2-65 5 35 9°72 4-81 + 0-03 2-78 6 3-0 9-51 4-8] + nil 2-80 7 2:5 9-40 4-79 + nil 2-68 8 2-0 8-97 4-72 + nil 2-65 i) 1-0 8-51 4-64 + 0-02 2-75
* Equivalent volumes of N NaOH were used.
1 N NaOH was used since a higher Pg was required than could have been obtained by the use of V/10 NaOH, the approximate reaction of which is Py =10-00. For ease of comparison the equivalent amounts of V/10 NaOH are shown in the tables.
12 Kr. J. 5S. WYETH
Consideration of Baperimental Data,
The above results are typical of a large number obtained in several series of experiments, and of these the final P,, values are represented graphically in Fig. 1, Curve I. It is obvious that, as in the case of B. coli grown in a 2% glucose peptone medium possessing an initial reaction less than Py, = 7-0, the final reaction produced by the organism when grown in alkaline glucose peptone (i.e. Whose initial reaction is more alkaline than Pj, = 7-0) is dependent upon the initial reaction (Pj) of the medium, although it varies within narrow limits. [t was found that growth is permitted by a solution the initial reaction of which is Py, = 9°82, but that a medium of initial reaction Py, = 10-28 inhibits growth. There is therefore a ‘limiting value” for the initial reaction of alkaline glucose peptone which lies somewhere between these two values. The final acid reaction reached by a culture of B. coli in alkaline glucose peptone of the highest permissible degree of alkalinity (7.e. initial reaction Py = 9-82) is such that its Pj, = 4:82 which is identical with that determined by Clark [1915] as the final acid reaction of “human” B. coli grown in a medium containing | % Witte peptone + 1 % glucose.
In the ‘“‘reaction resultant” curve for B. coli in 2 % glucose peptone shown in Fig. 1 (1) the initial reactions plotted on the 120° axis are so chosen as to cover the whole range of acidity and alkalinity within which growth of the organism in this medium is permitted. The form of the curve representing the final reactions (P,,) recorded shows that they form an ordered series lying between the acid limit of (approximately) Py = 4-27, and the alkaline limit of (approximately) P}, = 4:82. Further, the connection existing between the initial reaction of the medium and the final reaction of the resulting culture is such that a change in the reaction of the former, whether it be in the acid or in the alkaline direction, produces a corresponding, but much smaller, change in the latter. The form of the reaction resultant curve in Fig. 1 (1) shows also that the resultant reactions obtained throughout the whole range of bacterial activity are due to the production of constant amounts of acid. In addition to measuring the final reactions of the cultures a quantitative determination of the principal products of fermentation was made, some of the results of which are shown in Table I, columns 6 and 7. It was found that practically no ammonia is formed during the fermentation, and that, although considerable volumes of volatile and fixed acids are formed, the actual amounts of each are approximately constant throughout the whole range of experiment, thus confirming the inference drawn from the form of the reaction resultant curve (Fig. 1 [I]). It is evident that the fermentation of B. coli in glucose peptone is to all intents and purposes exclusively saccharolytic. In order to find whether proteolysis supervened after saccharolytic fermentation was complete a number of cultures were allowed to ferment for more than 9 days. In a number of instances a rise of E.M.F. equal to 1-2 millivolts was observed, and this was sufficiently constant in occurrence to prohibit the assumption
13
GROWTH AND INDOLE FORMATION OF B. COLI
[poppe HOUN®
J UNIprp euipoH |
-->
<
‘euowed %Z II ‘Aude -asomnyy x%Z [I] 109° fo Jun} Nsay uorpvoy
Eeapre io
a4
Ol W
uy 0:0
Os Ee
6+ Ofb Ez
OF [MENT P2iAy Ploy ba
14 KJ. 5S. WYETH
that it was always due to experimental error, although it must be conceded that such a variation in any individual culture certainly does fall within the limits of experimental error. That so small a reversion in the alkaline direction is found only in a certain proportion of the cultures proves conclusively that no appreciable amount of proteolysis occurs.
By constructing a titration curve for acetic acid in 2% glucose peptone and using this in conjunction with the initial reactions of the cultures subjected to quantitative examination, and also with the amounts of acid formed therein, a curve was constructed showing what would be the reaction resultants theoretically produced by the formation of the volumes of acid actually found as a result of these experiments. This curve was of the same shape as that resulting from the final reactions (Pj) actually found, and given in Fig. | [1]. The conclusions arrived at after an examination of the initial and final re- actions (P,,) of the cultures, and after a quantitative determination of the products of fermentation are therefore found to be in absolute agreement. It must be concluded that neither in the acid, nor in the alkaline range of initial reactions (Pj) is the final acid reaction attained by cultures of B. cola in 2 % glucose peptone a physiological constant. It has already been shown [ Wyeth, 1918] that the degree of acidity necessary to inhibit the fermentation of B. coli in a given medium is subject to very shght variations when different acids are added to produce the initial acid reaction of the medium, and it is possible, therefore, that similar small variations of the alkaline inhibition- point may be caused by varying the alkali used to produce the initial alkaline reaction of the medium. Similar—but larger—variations may be expected to result from the use of different media, as has already been demonstrated for glucose peptone and glucose phthalate media whose initial reactions lie between P,, 4:0 and 7-0. It may be permissible, perhaps, to point out that in certain parts of the reaction resultant curve of B. coli grown in 2 % glucose peptone, the final reaction resultant values present so little variation that it is only by the comparison of a large number of cultures, the initial reactions (P,,) of all of which have been accurately measured by means of the H-elec- trode, that it is possible to detect this ordered relationship between the initial and final reactions of the cultures. As will be shown below, this applies to a less extent to the curve for cultures of the same organism in 2 % peptone. The very narrow limits between which the final reactions produced by B. cola grown in 2 % glucose peptone (or similar media) can vary, appears to have led a number of earlier investigators,—few or none of whom appear to have subjected the initial reactions of their media to small and accurately-measured variations,—to postulate a physiological constant for the final reaction of B. coli grown in the media with which they experimented.
The results recorded in an earlier paper, together with those now pre- sented, conclusively disprove the existence of such a constant for B. coli, and Wolf and Harris [1917] arrive at the same conclusion as regards B. sporogenes and B. perfringens.
™~
GROWTH AND INDOLE FORMATION OF B. COLI 15
INDOLE PRODUCTION BY B. COLI GROWN IN 2 °/, PEPTONE.
Since the indole test is so frequently employed for diagnostic purposes, and as a proof of the presence of faecal coli in water and other media, it was thought desirable to determine whether indcle production ran parallel with the growth of the organism, as measured by the change of Py occurring in the medium employed, or whether fermentation unaccompanied by indole formation could be demonstrated.
The influence exerted by acids and carbohydrates upon the production of indole in cultures of B. coli has already engaged the attention of a number of investigators, but few exact measurements of the true acidity of such cultures appear to have been made. Although the primary object of the experiments described below was to determine the effect, if any, of acids and alkalies upon indole production it was found impossible entirely to separate this question from the much wider one of the influence of carbohydrates upon indole forma- tion, since, as will be shown, the statements made by certain investigators regarding this latter. phenomenon are by no means concordant. It was there- fore considered desirable to undertake a number of experiments involving the use of peptone containing various carbohydrates.
Peckham [1897] found that the indole production of B. coli may be taken aS an approximate measure of the amount of protein digestion due to the organism.
Theobald-Smith [1897] found that on the third or fourth day of incubation B. coli cultures in bouillon gave the indole reaction when the acidity was equal to NV /100 (by phenolphthalein), or when litmus was coloured faintly blue by the liquid, while with Dunham’s solution a well-marked violet-red colour was developed at the end of three days. He suggests that B. coli and other faculta- tive anaerobes can produce indole only when they are in contact with oxygen.
Marshall [1907] showed that B. coli grown for five days at 37° in peptonised beef broth containing 2-0 °% lactose failed to produce indole.
Glenn [1911] added considerably to our knowledge of the influence of carbohydrates on the production of indole by B. coli. His chief results are
shown below. Indole production
% acid hours Gas produced —<—— Medium production in 4 days 24 48 72...216 1% peptone oe 0 - + “Hoh 1 % peptone + 1 % glucose ac 2-2 = = 5 1 % peptone + 1 % saccharose + 1-2 _ = Sie = 1 % peptone + 1 °% lactose + 2-0 = = Sg aes 1 % peptone + 1 % mannitol + 2-0 ~ - wa 1 % peptone + 1 % starch ~ 0 + + tenet
He found that, in the case of B. coli, both glucose and lactose inhibit the production of indole, but, with Proteus vulgaris, lactose does not exert an inhibitory effect. He considers that his results ‘do not disprove the conclusion that the production of acid inhibits the formation of indole,” and he concludes
16 F, J. S. WYETH
also that the presence of 1%, or more than 1% of glucose always inhibits the production of indole by B. coli grown in peptone!. With Proteus vulgaris he found that the addition of lactic acid sufficient to produce an acidity equal to that of 0-5 % lactic acid inhibits indole production but that less than 0-5 %, of the acid does not produce an appreciable retardation of the process.
Fischer [1915] found, however, that of the sugars, lactose, maltose, galactose, fructose, and glucose, only the last named completely retards the formation of indole. Total inhibition occurs after 43 hours in a medium con- taining from 1-80 to 2-25 °% of glucose. He considers that the acids formed play no role in causing the retardation and that neither the H-ion concentra- tion nor the concentration of the undissociated acids can be taken as the reason of the retardation. He concludes that the acid curves he constructed as well as the retardation experiments he performed with a mixture of galac- tose and glucose, make it appear possible that lactose is not first hydrolysed by B. coli but is fermented as such. The cause of the retardation, he suggests, depends upon a peculiar property of the glucose which enables it to inactivate the proteolytic enzyme produced by the B. colv.
Homer [1916] suggests that the lessened indole formation resulting from the activity of B. coli in glucose-containing media is due to the formation of a glucose-tryptophan complex which is less easily attacked than is tryptophan itself. This complex she regards as chemically unsuitable for bacterial de- composition and hence a lessened growth of B. coli ensues.
Zunz and Gyorgy [1916] found it possible to grow B. coli in “l'eau physio- logique” and in “l'eau physiologique + 1% peptone” and that in these media indole formation occurred in less than 36 hours and 16 hours respec- tively. They conclude that indole production by B. coli depends upon the medium being rich in tryptophan and that the presence of certain carbo- hydrates inhibits the reaction. Unhke Glenn, they found that a medium containing peptone water + 1 % saccharose permits growth of B. coli together with indole formation within 96 hours—but not within 24 hours—after inoculation. Fischer does not appear to have experimented with saccharose, but, like Zunz and Gyérgy—and unlike Glenn—he found that neither 1 % lactose nor 1 °%, maltose added to a peptone medium sufficed to inhibit the formation of indole by B. col.
Material and Experimental Methods.
A number of media containing 2 °%4 peptone were prepared and inoculated by methods similar to those previously described. The initial reactions of the tubed media were produced by the addition of either N/10 HCl or N NaOH in such quantities as to produce 20 different media whose initial reactions ranged from Py = 4-0 to Py = 10-0. In addition a number of tubes of 2 %
1 Wolf has recently examined thirteen strains of B. proteus isolated from wounds, and has
found that none of them produced indole. It is possible that Glenn was working with impure strains of this organism.
GROWTH AND INDOLE FORMATION OF B. COLI 17
peptone of approximate neutral initial reaction (Py = 7-0) were prepared each containing 2 % of one of the following carbohydrates: glucose, maltose, lactose, saccharose, starch and of mannitol.
In order to determine the influence, if any, exerted by acids or alkalies upon indole formation in 2 °% peptone a number of sets of the tubes of this medium first mentioned above were inoculated with B. coli and incubated at 37°. They were tested for the presence of indole and the Py, of the cultures recorded at the end of 0, 4, 8, 24 hours after inoculation and then at intervals of 24 hours to the end of the ninth day of incubation. The rosindole and vanillin-hydrochloric acid tests were employed in determining the presence or absence of indole in the cultures. A number of the results obtained are embodied in Tables II and III, and on p.19. The B. colv used were of faecal origin, and no differences were exhibited between those of human and those of bovine origin.
The results obtained with a number of cultures of B. coli grown in 2 % peptone, containing no glucose but having been made of different initial reactions (P;,) may be summarised as follows:
(i) Whether the medium be rendered acid (initial reaction P,, less than 7-0) by the addition of N/10 HCl, or alkaline (initial reaction Py greater than 7-0) by the addition of V NaOH, indole production always occurs if the initial reaction of the medium be such as permits fermentation to take place.
(ii) If the initial reaction (P,) of the medium approximate to either the acid (Py, = 4:30) or the alkaline (Py = 9-37) limiting value for the growth of B. coli in 2 % peptone, the indole formation is subject to well marked retarda- tion, in some cases being delayed for 144 hours, whereas over the greater part of the range of the initial reactions employed indole is formed within 4 to 8 hours after inoculation.
(ii) No matter what may have been the initial reaction (Py) of the medium, indole was never detected in the culture unless and until its reaction reached a value lying between Py = 4:70 and Py, = 9:20. Thus in aculture, the initial reaction of which was Py = 4:33 and of which the acidity ultimately fell to Py, = 5-92, the production of indole was not observed until the expira- tion of 144 hours after inoculation, the reaction of the culture then being in the neighbourhood of Py = 4:80. Similarly, in a culture of initial reaction Py = 9-37, the alkalinity of which eventually diminished to Pj, = 8-55, indole was first detected 96 hours after inoculation when the reaction of the culture was found to be Py = 8-80. On the other hand with cultures whose initial reactions lay between Py = 8:89 and P, = 5-20 indole invariably was found in less than 24 hours after inoculation.
In no case was the presence of skatole detected.
Consideration of Results of Experimental Data.
It would appear that the presence of indole in a culture may safely be taken as an indication of the presence of B. coli of faecal origin, but that its Bioch, xr 2
18 F. J. S. WYETH
production—like the growth of the organism itself—is retarded by the presence in the medium of an excess of free acid or alkali. The inhibition of indole formation may, however, be one result of a lowered vitality of the B. cola consequent upon the presence of this excessive amount of acid or alkali, and it may even be a precursor of the death of the organism, brought about by the action of these substances. The results of these experiments appear to confirm Fischer's statement that the acids and alkalies formed as a result of fermentation by B. coli do not cause an inhibition of indole production. On this point Glenn comes to no definite conclusion although he appears to lean towards the opinion that the inhibition of indole formation may be due to the presence of acids formed during fermentation.
In experiments with B. coli grown in 2 % peptone, had the indole formation occurred only in the most acid media of the series under examination it might be argued from a comparison of Curves | and II of Fig. | that the inhibitory effect produced by the addition of glucose to a peptone medium might be due solely to the production of acids as a result of saccharolytic fermentation. This would, at first sight, appear to be more probable since the final acid reactions produced by B. coli fermenting in 2 % glucose peptone lie between Py = 4:27 and Py = 4:82. Indole formation is completely inhibited in cultures of B. coli in 2 % glucose peptone throughout this range of reaction, and also in 2% peptone media so long as their reactions remain more acid than P,, = 4:70. Since in the latter cultures the inhibition is obviously the result of added acid, and the two ranges of reaction are practically identical, there is at hand an obvious explanation of the inhibitory action due to the presence of glucose in the former class of cultures. This explanation, however, cannot be correct for the following reasons:
(i) At no period during the fermentation of a 2 % glucose peptone medium does B. coli produce indole even when in the initial stages of the process the reaction of the culture is much less acid than Py = 4-82 or Py = 4-70.
(ii) The retardation of indole formation by B. coli fermenting in 2% peptone occurs not only when excess of acid is present but also when there is excess of alkali present in the medium. Inhibition persists so long as the reaction of the culture is more alkaline than P,, = 9-20. It appears probable therefore that the inhibition of indole formation in cultures of B. cola in peptone containing excess of either acid or alkali is due to a cause entirely different from that which operates when the inhibition results from the presence of glucose in the medium. In the former case the inhibition may be due to a lowered vitality of the organism, produced by the action of the free hydrogen or hydroxyl ions present, while in the latter instance the cause of the inhibition may be attributed to a peculiar property of the glucose, which enables it to inactivate the proteolytic enzyme of the B. colt.
GROWTH AND INDOLE FORMATION OF B. COLI 19
THE INFLUENCE OF CERTAIN CARBOHYDRATES AND ALLIED COMPOUNDS UPON THE PRODUCTION OF INDOLE IN PEPTONE MEDIA.
For these experiments the tubed media containing 2 °% peptone together with either 1 % or 2 % of added carbohydrate were used (p. 16). The media were inoculated with different strains of B. coli. The cultures were incubated at 37° and examined at the end of periods of x hours as recorded in the sub- joined tables.
Indole formation at the end of hours
Medium inoculated 52: BR ry Gas with B. coli i 8 24 48 96 production 2% peptone + 1 % glucose 0 ) 0) 0 0 ae 2% peptone + 1 % lactose 0 0 + a ae ae 2% peptone + 1 % maltose 0 0 ab ae iL ae 2% peptone +1 % mannitol 0 (+?) tL ate $e ve 2% peptone + 1 % saccharose 0 0 0 (+?) a ate 2% peptone + 1 % starch + + + + + 0
Indole formati t hrs after inoculation Medium inoculated ye ee
- — ———— tas with B. coli C= 4 8 24 48 96 production 2 % peptone st aL ate a ce 0 2% peptone + 2 % glucose 0 0 0 0 0 + 2% peptone +2 % lactose 0 0 0 0 (+?) a 2% peptone + 2 % maltose 0) 0 0 0 (+?) + 2% peptone +2 % mannitol 0 0 0 (Cen et a 2% peptone + 2 % saccharose 0 0 0 0 (+2) ° 2 % peptone +2 % starch + + af a aL 0
From the above results it appears that the retardation produced by either 1 % or 2 % glucose in 2 % peptone may be regarded as absolute. The same percentages of starch produce no retardation, while that produced by mannitol is comparatively slight. Lactose and maltose appear to possess less retarding power than does glucose since | °% of maltose or lactose produces but little effect while 2 % of either of these sugars produces a retardation approximating to the absolute inhibition resulting from the action of glucose. Since the initial reaction of all these cultures approximated to Py = 7-03 it may be concluded that where retardation was observed it could not be ascribed to the initial presence (or to the ultimate production) of alkali or of acid, but must be the result of some specific action of the added carbohydrate. It must also be noted that where partial retardation occurred, the production of indole was observed at precisely the time when,—if inhibition were due to resultant acidity,—it should begin to be inhibited.
Tar Errects or Actps AND ALKALIES ON THE GROWTH OF B. COLI tN 2 % PEPTONE.
Material and Experimental Methods.
A number of flasks and tubes of 2 °% peptone medium of various initial reactions were prepared as described above. As, in the case of the experi- ments with glucose peptone, the acid used was N/10 HCl and the alkali
peep}
“ =
20 F. J. S. WYETH
employed was VN NaOH. Some twenty sets of media whose initial reactions ranged from Py = 3:50 to Py, = 10-50 were prepared. Inoculation and ineu- bation were performed as in the preceding experiments.
The tubed cultures were examined for changes of reaction (Py) and for the production of indole, while the flask cultures were submitted to the same tests, and in addition, a quantitative determination of the products of fermentation was made.
The results obtained in the two series of experiments are shown in the subjoined Tables II and III and a reaction resultant curve for B. coli grown in 2 % peptone was constructed, and is shown in Fig. | (II).
Table II. B. coli grown in 2% peptone rendered acid by the addition of N/10 hydrochloric acid. Strain of B. coli used. Hg.
Time of Incubation at 37°, 216 hours. Temperature of Experiment, 20°.
Ce. of N/10 No. of ec. of Ce. of V/10 volatile N/10 HCl NH, pro- acids pro- added per Initial Final duced per __ duced per No. of 10 ce. of reaction reaction Fermen- 10 cc. of 10 ce. of Indole flask medium Py Pr tation medium medium formation 1 3-0 3-54 —— 0 = = 10 2 2-5 3-93 — 0 — — 0 3 2-0 4-33 5:92 + 0-64 0-49 + at 144hrs at 12 hrs 4. : 5:5 é : 2 a ee 1-0 9°20 8-00 + 1-84 1-1 | ot at 7 OTE (+ at 12hrs = A= AAG FD) A=) + 3) 0-5 6-46 8-27 + 2-52 1:37 |. Sab oeies 6 0-0 7°25 8°37 + 3-00 1-60 + +at 12 hrs
Table Hl. B. coli grown in 2% peptone rendered alkaline by the addition of N NaOH. Strain of B. coli used, H,.
Time of Incubation at 37°, 216 hours. Temperature of Experiment, 20°.
Ce. of V/10 Cc. of N/10
Ce. of N/10 3 NH, pro-_ volatile acids NaOH added Initial Final duced per produced per No. of | per 10 cc. of reaction reaction Fermen- 10 cc. of 10 ce. of Indole flask medium* Pr Pu tation medium medium formation 1 5:0 10-23 — 0 = — 0 2 4-0 9-87 — 0 _ = 0) P : (+ at 96hrs + 9: 55 . i J 3 3-0 9°37 = 8-55 + 4-5 3-95 | 4+ hat Tene 2-0 J < ‘ 2. ( + at 24hrs 4 8-89 8-51 + 4-] 3°34 | + vataomee i: (+ at 12hrs 5 1-0 8-51 . 8-4 3°5 2-75 + ae 5 in (crea oie 6 0-5 7:95 8-40 + 3-2 2-21 +-+at 12 hrs
* Equivalent volumes of N NaOH were used.
GROWTH AND INDOLE FORMATION OF B. COLI 21
Consideration of Experimental Data.
The salient fact revealed by a study of the values of the initial and final reactions (P;) recorded in Tables II and III (columns 3 and 4) is that there is for B. coli grown in 2 % peptone,—as for the same organism in 2 °%/, glucose peptone,—an obvious connexion between the initial reaction of the medium and the final reaction of the culture. This is represented graphically by the reaction resultant curve (II) in Fig. 1. Next it is observed that there is an “acid”’ and an “alkaline limiting value” of the initial reactions between which growth is permitted but beyond which the activity of the organism is in- hibited. The limiting value in the acid range of initial reactions lies between Py = 4:33, which permits, and P,, = 3-93 which inhibits the fermentation of B. coli in 2% peptone. The degree of alkalinity which serves to inhibit fermentation les between an initial reaction of Py = 9:37 which permits and Py = 9-87 which inhibits the process. The activity of B. coli in 2 % peptone is thus found to be determined by almost the same initia] conditions of acid and alkaline reaction as is the case with the same organism fermenting in 2 °%/, glucose peptone.
There are however striking differences between the behaviour of B. coli when grown in the two media, as may be seen by comparing Table I with Tables II and III, and also Fig. 1 (I) and Fig. 1 (II).
In both media the final reactions attained by a culture of the bacillus occupy positions on an unbroken curve, and in both media a diminution of the initial acidity (P,,) of the medium results in a diminution of the resultant acid reaction of the culture but beyond these general resemblances the similarity does not extend.
Whereas the range of initial reactions within which growth is possible has been shown to be practically identical for B. coli grown in the two media there is a remarkable difference between the range of final reactions attained in them. In 2% glucose peptone the final reactions vary between the very narrow limits of Py = 4:27 and Py = 4-82 but in 2% peptone the final reactions produced by the organism are as widely separated as P,, = 5-92 and Py = 8:55 while it is quite possible that further experiments may lower the former of these two values to Py = 5-70, as will be seen by an inspection of Curve IT, Fig. 1.
It is obvious that in the case of B. coli grown in 2 % peptone any suggestion of a “physiological constant” for the final reaction values cannot be enter- tained. It will be noted that the curve (Fig. 1 [IT]) representing the reaction resultant of B. coli grown in 2 % peptone cuts the axis on which the initial reactions of the medium are plotted at a point representing a reaction of Py = 8-48. It is evident that if it were possible to prepare a 2 °% peptone medium of which the initial reaction were exactly P,, = 8-48 the fermentation of B. coli in this medium should be unaccompanied by any change of reaction. Several attempts to prepare a medium of this initial reaction were made, but
22 F. J. S. WYETH
without success. The medium most closely approximating to that desired had an initial reaction of Py, = 851. In it, fermentation was very active, indole, ammonia, volatile and fixed acids being formed in large quantities while the final reaction became slightly less alkaline—the ultimate value being Py = 8-46.
Passing from this point to the acid “death-point” of the organism it 1s observed that for cultures in a medium of which the initial reaction is acid (Py, less than 7-0) or possessing an alkalinity of less than P,, = 8-48, the activity of the bacillus results in a change of the reaction of the culture in an alkaline direction. The inference which naturally would be drawn from an inspection of this region of the reaction resultant curve, viz. that the fermen- tation of B. coli in 2% peptone the initial reaction of which is less alkaline than Py = 8-48 results in the formation of an excess of alkaline products, the reaction of the culture thereby becoming more alkaline than Py = 8-48, is confirmed by the results of the quantitative examination of the products of bacterial activity (Tables II and ITI, cols. 6 and 7).
In the absence of sugar, the action of B. coli on peptone is frankly proteo- lytic, the higher nitrogenous complexes being decomposed, with the consequent formation of ammonia, acids and indole.
When, however, B. coli ferments an alkaline 2 % peptone, the initial reaction of which is more alkaline than P,, = 8-48 the activity of the organism results in a diminished alkaline reaction in the culture (e.g. a medium whose initial reaction is P}, = 8-89, reaches a final reaction of P;, = 8-51). Reference to the results of quantitative examination of the fermentation products shows that this diminished alkalinity is due to a rapid increase in the amount of acid produced. In general it may be observed that, as the initial reaction of the 2 % peptone in which B. coli ferments is varied from Py = 4:33 to Pj, = 9:37, the production of alkali is at first more marked than that of acid, while in the latest stages the increase of acid production is greater than that of alkali formation. This accounts for the “reversion” of the course of the reaction when the critical point Py = 8-48 is passed. It must be emphasised that the reversion does not proceed far enough to interrupt the regular path of the reaction resultant curve (Fig. 1 [II]) since any given culture eventually reaches a final reaction which is more alkaline than that attained by the next lower (more acid) member of the series.
A complete and accurate estimation of the fixed acids produced in the reactions was not performed, but sufficient data were obtained to show that in the neighbourhood of the “alkaline limiting value” (P,, = 9°37) the total acids formed were in excess of the alkaline products, whereas in cultures the initial reactions of which were acid, neutral, or slightly alkaline (initial Py, = 4:33 to 8-00) the ammonia production was greater than the formation of acids. It is evident that the theoretical yield of product of a culture whose initial reaction is P,, = 8-48 (and is therefore constant during fermentation), should consist of equivalent quantities of acid and alkali. With the aid of
~
GROWTH AND INDOLE FORMATION OF B. COLI 23
titration curves for acetic acid and ammonia added to alkaline and acid 2 % peptone respectively a curve was plotted which represented the effect pro- duced by the addition to 2 °%% peptone of the amounts of ammonia and acids found in the quantitative determinations performed.
This was done by taking in turn the initial reaction of each culture and finding from the acetic and titration curve what would be the effect upon the initial reaction if the amount of acid actually found by experiment were added to the medium. To the new reaction value thus obtained is then applied the correction representing the effect produced by the amount of ammonia the presence of which had been determined by the experiment, this correction having been found from the ammonia titration curve. The final resultant obtained in each case represents the theoretical “reaction resultant” of the fermentation. The whole series of values thus obtained was plotted as a reaction resultant curve, which was found to be similar to, but not quite coincident with, that shown in Fig. 1 (II). This indicates that the curve repre- senting the observed final reactions is such as would result had it been plotted from a series of final reaction resultant values produced by the formation of increasing amounts of acids and ammonia in a series of media of increasing alkaline reaction, the amounts of these acids and ammonia being of the same
order as those found in the foregoing experiments.
SUMMARY.
1. Growth of B. coli is possible in peptone media of certain well-defined initial reactions, the limiting values of which are but slightly, if at all, affected by the presence or absence of sugars.
2. The approximate limits of initial reaction are Py = 4:27 to 9-87.
3. A change of the initial reaction of the medium results in a change, similar in direction, but smaller in magnitude, in the final reaction of the culture.
4. In the case of B. coli grown in 2 % glucose peptone, while the initial reactions of the media vary from P,, = 4:30 to Py = 9-82 the final reactions attained vary only between the very narrow limits of Py = 4:27 and 4-82.
5. In the case of the organism grown in 2 % peptone, when the initial reaction of the medium is varied from Py = 4:30 to Py = 9-37, the final reactions of the cultures vary from Py = 5-92 (5-70?) to Py = 8-55.
6. The saccharolytic fermentation resulting from the growth of B. coli in 2 % glucose peptone renders the culture more acid than the original medium.
7. The proteolytic fermentation resulting from the growth of B. coli in 2% peptone causes an increase of final alkalinity in the resulting culture unless the initial reaction lies between the alkaline limiting value (Pj, = 9°37 and Py, = 8-48 in which case the final reaction of the culture is less alkaline than the initial reaction of the medium.
24 F. J. 8S. WYETH
8. The saccharolytic fermentation of B. coli in 2 % glucose peptone media of different initial reaction produces approximately constant amounts of acids and no appreciable amount of ammonia.
9. The proteolytic fermentation of B. coli in 2% peptone results in the formation of acids and ammonia, the amounts of both of which increase as the initial reaction of the medium is varied in the direction of increased alka- linity. Near the alkaline limiting value the production of acids is greater than that of ammonia.
10. The formation of indole is retarded by the presence of free alkali or acid in the medium.
ll. The presence of certain sugars causes inactivity of the proteolytic enzyme produced by the bacillus and thus inhibits the formation of indole.
12. Different carbohydrates exhibit different degrees of indole-inhibiting power. The addition of 2 % glucose to peptone media produces complete in- hibition of indole formation; 2 % lactose or 2 % maltose produce almost complete inhibition, while that produced by 2 % saccharose or 2 °% mannite is only partial. 2% starch possesses no inhibitory power.
REFERENCES.
Clark (1915). J. Biol. Chem. 22, 87.
Fischer (1915). Biochem. Zeitsch. 70, 105.
Glenn (1911). Centralbl. Bact. Par. 1 Abt. Originale, 58, 481. Homer (1916). J. Hygiene, 15, 401.
Marshall (1907). J. Hygiene, 7, 581.
Michaelis and Marcora (1912). Zeitsch. Immunititsforsch. exp. Ther. 14, 170. Murray (1916). J. Infect. Dis. 19, 161.
Peckham (1897). J. Exper. Med. 2, 549.
Theobald-Smith (1897). J. Haper. Med. 2, 543.
Wolf and Harris (1917). Biochem. J. 14, 213.
Wyeth (1918). Biochem. J. 12, 382.
Zunz and Gyorgy (1916). J. Bact. 1. No. 6.
III. OBSERVATIONS ON THE ALBUMINOID AMMONIA TEST.
By EVELYN ASHLEY COOPER anp JOSEPH ALAN HEWARD.
From the Command Hygienie Laboratory, School of Army Sanitation, 3 Aldershot.
(Received February 10th, 1919.)
SoME time ago in this laboratory it was found that effluents of fair and even very good quality were yielding albuminoid ammonia figures of 1-0 to 2-0 parts per 100,000, while pure drinking water gave figures varying from 0-02 to 0-1.
The albuminoid ammonia figures of good effluents generally approximate to 0-15, while in the case of pure waters the figures are generallv less than 0-005.
It was evident that the technique involved some very considerable error, which quite vitiated the value of the test, and it was necessary to make an investigation to ascertain the cause of these abnormally high figures.
The distilling apparatus seemed to be quite above suspicion, as the corks were securely covered with tin-foil and there was no difficulty in obtaining ammonia-free water. Furthermore, the free and saline ammonia figures were normal.
Some impurity in the chemicals used in the albuminoid ammonia deter- mination was therefore suspected. It had been observed that the pernian-- ganate-alkali mixture was very difficult to free from ammonia, but after two days’ boiling it was always apparently possible to free the mixture by a single distillation in the apparatus.
The great difficulty experienced suggested however that possibly the mixture was not after all free from nitrogenous material and that by diluting and redistilling with ammonia-free water further yields of ammonia might be obtained.
It was resolved to test this possibility by experiment. Potassium per- manganate and caustic soda in the usual proportion of }.¢. to 10 ¢. respec- tively were boiled together in a small flask with ammonia-free distilled water for two days. The mixture was then poured into a litre of ammonia-free water in the distilling apparatus and distillation was carried out. Large amounts of ammonia were generated, e.g. 0-0001 g. When the ammonia ceased to come over, the ammonia-free distillate was collected and the contents of the flask were concentrated to a small bulk. The ammonia-free distillate was next
26 K. A. COOPER AND J. A. HEWARD
poured back and the distillation recommenced. As much ammonia was again produced as in the first distillation. This could often be repeated for several days; sometimes however the generation of ammonia ceased after a few distillations. Taking 0-0001 ¢. as a typical yield from one distillation, as 500 ee. of a drinking water are used in the albuminoid ammonia determination, this would correspond to the enormous error of 0-02 parts per 100,000, Occasionally a single distillation yielded even larger amounts, and the error Was correspondingly increased,
It was evident that this was at any rate one source of error. The per- manganate-alkali mixture could apparently be cleared of ammonia by a single distillation, but when added to the water being analysed the dilution and boiling would lead to the production of more ammonia, sufficient in amount to give altogether inaccurate results.
Prolonged boiling rarely led to the purification of the mixture, and it was concluded that a stable nitrogenous impurity was present which only decom- posed with extreme slowness. The condition favourable to decomposition was great dilution, while concentration stopped the process entirely.
Through the kindness of Prof. Harden, a sample of permanganate-alkali mixture was obtained from a laboratory in which normal albuminoid figures were being obtained. When 50 cc. of this solution (containing } ¢. of per- manganate and 10 g. of alkali) was distilled with ammonia-free water, only 0-000005 g. ammonia was generated in each distillation. This corresponded to the small error in the albuminoid-ammonia figure of a water of 0-001 parts per 100,000.
It was thought that this mixture could be employed to ascertain whether the impurity existed in the permanganate or the soda in the contaminated mixture which was yielding the high results.
Accordingly 50 ce. of the pure mixture was boiled with 4 ¢. of the suspected permanganate for two days and the solution was then distilled with ammonia- free water in the ammonia apparatus. Large amounts of ammonia were generated, e.g. 0-O001 g. ammonia in each distillation. This pointed to the existence of a nitrogenous impurity in the permanganate.
The pure mixture was next separately heated with 10g. of the suspected soda and distilled as before. During each distillation only small amounts of ammonia were produced, e.g. 0-000005 g.
The impurity was thus chiefly concentrated in the permanganate.
Experiments were next carried out with the same brand of permanganate and soda as constituted the pure mixture. A solution was prepared in this laboratory, and after one day’s boiling it generally yielded an error of about 0-001 to 0-002 parts per 100,000 in an albuminoid ammonia determination for a water. On one occasion the error was 0-003.
Some control experiments were also carried out to ascertain the amount of albuminoid ammonia generated from strychnine when decomposed by the permanganate-alkali. The alkaloid yields half its nitrogen content as albu-
THE ALBUMINOID AMMONIA TEST 27
minoid ammonia. The results as would be expected were about 10 % high, since if 0-0001 g. were produced from the strychnine this would be increased to about 0-00011 owing to the trace of nitrogenous impurity in the perman- ganate.
For ordinary work this would be sufficiently accurate, but for very im- portant work it would be advisable to make a control estimation, under the same conditions of dilution and rate of boiling, of the ammonia yielded in one distillation of the permanganate-alkali mixture with ammonia-free water. This figure could then be subtracted from the figure obtained for the water or sewage effluent which was being analysed.
When fresh supplies of chemicals come into the laboratory, tests should be made before using them in routine work to ascertain whether the organic impurity is present in unusually large amount.
The presence of this oxidisable impurity may be one explanation of the observed decomposition of about 5% of the permanganate in the control distilled water test carried out in the determination of the Tidy figure or oxygen absorbed in 4 hours at 27° C.: e.g.
Initial titration 10 cc. N/80 permanganate, equals 14-9 cc. standard thiosul- phate. (10 ce. permanganate, NOVGe 25 7, Elo OO. 100 ee. distilled water.)
Titration after 4 hours’ incubation at 27° C. equals 14-2 cc. thiosulphate.
SUMMARY.
1. Potassium permanganate may contain a stable nitrogenous impurity which cannot as a rule be removed by prolonged boiling with alkali.
2. The impurity is not decomposed at all in concentrated alkaline solution, but gradually decomposes when the solution is considerably diluted. Conse-. quently permanganate-alkali mixture can apparently be freed from ammonia by boiling with water, but when the resulting concentrated solution is again diluted and distilled, ammonia may once more be liberated in large quantities. When the mixture apparently freed from ammonia is boiled with the water, the albuminoid ammonia figure of which is being determined, the yield of ammonia is thus greatly augmented by that liberated from the permanganate.
3. The error involved may be so great as to vitiate the value of the albuminoid ammonia test altogether.
4. In the case of a purer brand of permanganate the error is reduced to about 0-002 parts albuminoid ammonia per 100,000.
5. It is essential in routine work to test fresh supplies of chemicals to ensure that the impurity is not present in excessive amount and in very accurate work a control experiment should be made during each determination of albuminoid ammonia.
IV. THE COMPOSITION OF STARCH. PART I. PRECIPITATION BY COLLOIDAL IRON. PART II. PRECIPITATION BY IODINE AND ELECTROLYTES.
By JOHN MELLANBY. From the Physiological Laboratory, St Thomas's Hospital, London.
(Received February 10th, 1919.)
Tue experiments recorded in this paper deal with the effects produced by (a) colloidal iron and (0b) iodine on a solution of potato starch in water. The starch solution was obtained by adding starch suspended in a small volume of cold water to the required volume of boiling water and continuing the
boiling for one minute.
|. THE PRECIPITATION OF STARCH BY COLLOIDAL IRON.
There are considerable differences of opinion regarding the nature of starch granules, but, generally speaking, starch is now regarded as consisting of two substances. Nageli who investigated the action of hydrochloric acid and amylolytic enzymes on starch proposed the names amylogranulose and amylo- cellulose to designate them. Amylogranulose, the chief constituent of starch, goes into solution under the action of hydrolytic agents and gives a blue colour with iodine. Amylocellulose on the other hand does not go into solution under these conditions and is not coloured blue by iodine. This work has been - extended by Maquenne and Roux [1903, 1905] and Fernbach and Wolff [1904]. From their experimental results they deduce that starch granules consist of two substances, (4) amylose and (b) amylopecten. Amylose, the chief constituent of starch, is insoluble in cold water but completely soluble in water boiled under pressure. On cooling such a solution the soluble amylose tends to revert to the insoluble form. Amylopecten, on the other hand, is of an entirely different nature. It swells up without dissolving when heated with water and is not coloured blue with iodine.
The results recorded in the following pages indicate that the soluble portion of starch (amylogranulose) may be further differentiated into a series of fractions.
(i) Precipitation by colloidal iron only.
The investigations of Bottazzi and Victoroff [1910] on the electrical state of starch in solution indicate that starch is electrically neutral. This con- clusion is not supported by the effects observed when colloidal iron is added
COMPOSITION OF STARCH 29
to a solution of starch. Under these conditions three well marked phases may be recognised: (i) a portion of the starch is precipitated by the colloidal iron only, (ii) a second portion of the starch is carried down with the colloidal iron when an electrolyte is added to the solution, and (iii) the filtrate from (11) con- tains starch which is not precipitated by colloidal iron either in the presence or absence of electrolytes.
The amount of starch precipitated by colloidal iron only is shown in the following experiment.
Colloidal iron was added to 1 °% starch in the following proportions:
Colloidal iron Starch 1 % H,O 4 cc 20 ee. 16 ce. 8 cc. 20 cc. 12 ee.
A gelatinous precipitate was produced in each case. This precipitate was allowed to settle, after which the amount of starch remaining in the clear fluid, and the amount of starch in the original solution, were determined. In both ‘cases it was found that 80 °% of the original starch was precipitated by the colloidal iron. The amount of starch precipitated was independent of the amount of iron added. The lower limit, giving the minimal amount of iron necessary for precipitation, was not determined. Precipitation presumably depends upon the neutralisation of the negative charge on the starch by the positive charge on the colloidal iron. It became of interest therefore to deter- mine how far the precipitation limits of colloidal iron by electrolytes were influenced by the addition of a solution of starch to it.
The minimal quantity of potassium sulphate required to precipitate | ce. of colloidal iron is given in the following figures.
Colloidal iron H,O M/10 K,SO, 1 ce. 8-95 ce. 0-05 ce. No precipitation 8-9 0-1 = 8-8 0-2 me 8-6 0-4 as 8-5 0-5 Complete precipitation
With 0:4 ce. M/10 K,SO, there was no formation of distinct granules, but there was an obvious change in the appearance of the solution viewed by partially reflected light.
The same experiment was repeated except that 5 cc. of starch replaced 5 ce. of added water, thus:
Colloidal iron Starch 1% H,O M/10 K,SO, I ce: ce: 4 ce. 0-0 ee. No precipitation ' 3°95 0-05 oF 3°9 0-1 3°8 0-2 Ae ' 3-6 0-4 Partial precipitation 3°5 0-5 Complete a
In the first four tubes there was a precipitate of starch, but this precipitate was greatly augmented by precipitated iron in the fifth tube. In the sixth tube
30 J. MELLANBY
complete precipitation of the iron had been produced. The partial precipita- tion of the colloidal iron in the presence of starch by 0-4 cc. K,SO, (7/10) compared with the amount 0:5 ce, KySO, (JZ/10) required when no starch was present indicates that the starch had diminished to a small degree the quantity of electrolyte required for precipitation. The results show that the electric charge on starch in solution is negative, though small in comparison with the electropositive charge on colloidal iron.
(ii) The precipitation of starch by colloidal iron in the presence of electrolytes.
The amount of starch precipitated from solution by colloidal iron 1s in- creased by the presence of electrolytes. But however much colloidal iron may be used, and however much precipitating electrolyte may be added, some starch remains in the filtrate. The amount of starch not precipitated by colloidal iron in the presence of electrolytes was determined.
To 100 ce. of approximately 1 % starch, 20 ce. of colloidal iron were added. Five minutes later 1-2 ec. of K,SO, (17) were added as the precipitating electro- lyte. The resultant filtrate was water clear in appearance but gave a deep blue colour on the addition of iodine. ‘The quantities of starch in the original solution and filtrate were determined by hydrolysis to dextrose. The results showed that 89 °% of starch was precipitated whilst 11 °% remained in solution. Now a previous experiment had shown that on the addition of colloidal iron only to the starch solution 80% of the starch was precipitated. We must therefore conclude that 9 °% of the original starch is carried down by colloidal iron precipitated by electrolytes. The results show that three types of starch must be recognised as present in a solution of potato starch in water: (a) starch precipitated from solution by colloidal iron only; (8) starch taken down with colloidal iron when precipitated by a divalent negative ion; (y) starch not affected by colloidal iron in any way. The results given above show that a solution of potato starch contains 80 % of (a), 9% of (8), and 11 % of (y). It may be remarked that the first fraction (a) includes the insoluble constituent of starch (amylocellulose) since this insoluble constituent would be mechanic- ally carried down by the soluble starch precipitated by the colloidal iron. On this assumption the starch granule consists of amylocellulose and amylo- eranulose (a), (8) and (vy), the terms (a), (6) and (y) designating those varieties which are precipitated by (a) colloidal iron only, (f) colloidal iron in the presence of electrolytes, and (y) not precipitated by colloidal iron.
This, however, does not exhaust the possibilities of the substances con- tained in the starch granule, since precipitation by iodine in the presence of electrolytes, the amount of iodine added being just short of that required for complete precipitation of the starch, gives a filtrate which is coloured brown by iodine. This colour is due to the presence of dextrin in the starch solution. The presence of dextrin might be due to its production when the starch is dissolved in boiling water, but against this hypothesis is the fact that glycogen dissolved in cold water gives similar results. The complete precipitation of
COMPOSITION OF STARCH 31
starch by equivalent quantities of iodine in the presence of electrolytes shows that the granule contains no soluble substances which do not react with iodine. It appears reasonable to conclude that the starch granule contains a large number of polymers of the general formula (Cg,H,)0;)n increasing in complexity from dextrin to cellulose, the dextrin and cellulose being present in small quantities whilst amylogranulose (a), (8) and (y) form the bulk of the eranule. Probably other methods of precipitation would differentiate the amylogranulose (a) into a series of fractions.
Il. THE PRECIPITATION OF STARCH BY IODINE AND ELECTROLYTES.
When iodine dissolved in potassium iodide is added to a solution of starch in water two characteristic changes may be observed: (i) the solution assumes a deep blue colour, and (ii) the lyophilic emulsoid colloid of starch is changed into a suspensoid colloid of marked lyophobic character.
A considerable amount of experimental work has been done to determine the nature of the changes involved. Owing to the varying content of “starch iodide” the chemical aspect of the problem has given place to views in which the physical element predominates. Kiister [1894] put forward the hypothesis that iodide of starch consists of a solid solution of iodine in starch, and ex- plained in this way the varying content of the iodine in starch iodide with the iodine concentration of the solution with which it was in equilibrium. Harrison [1911] on the other hand considered that in the case of iodine and starch an adsorption compound was formed. The most recent work on this subject has been done by Barger and Field [1912] in their analyses of the phenomena observed in the formation of blue adsorption compounds of iodine with various substances. From their experimental work they conclude that on the addition of iodine to starch there is a considerable adsorption of iodine, that the pre- sence of electrolytes (potassium iodide) is necessary to bring about this adsorption, and that the blue substances formed behave as negative suspension colloids which have been rendered lyophobic by the iodine adsorbed.
The electronegative nature of starch iodide was shown by the experiments of Padoa and Savaré [1906] in which starch iodide moved to the anode when placed in an electric field. The experiments of Barger and Field on the pre- cipitation of starch iodide by cations of inorganic salts confirmed this deduc- tion as to the electronegative nature of starch iodide and demonstrated that the addition of iodine to starch changed a lyophilic emulsoid colloid (the starch solution) to a lyophobic suspensoid colloid (starch iodide). The facts that starch dissolved in water forms an emulsoid colloid and carries a small negative charge, that the reacting iodine dissolved in potassium iodide is presumably ionised and negatively charged, and that the resulting starch iodide forms a suspensoid colloid and is also negatively charged, militate against the accep- tance of a purely physical hypothesis for the explanation of those changes which occur when iodine is added to starch. The experiments recorded in the
32 J. MELLANBY
following pages on the effects observed when iodine dissolved in potassium iodide is added to a solution of starch indicate that the iodine first reacts quantitatively with the starch producing a definite chemical compound and that the starch iodide thus formed is a lyophobic suspensoid colloid which adsorbs iodine from solution according to the recognised laws of adsorption.
(A) The quantitative relation of iodine to starch.
The following experiments indicate that on the addition of small quantities of iodine to starch a reaction takes place in accordance with the recognised laws of chemical action.
(1) Varying vdine. Varying quantities of iodine were added to 10 cc. of a 1 % solution of starch contained in a series of tubes, and the resulting starch iodide was precipitated from solution by a small quantity of magnesium sulphate. After an interval of five minutes the contents of the tubes were filtered and the filtrates were tested for excess of iodine or starch. The follow- ing figures show the results obtained in a typical experiment:
Starch 1 °% Todine 1% H,O MgSO, (17) Filtrate
10 ce. 0:3 ec. 9-6 ce. 0-1 ce. Excess of starch 10 0-4 9-5 0-1 3 - 10 0-5 9-4 0-1 = = 10 0-6 9-3 0-1 i a. 10 0:7 9-2 0-1 Nil
10 0-8 9:] O-1 Excess of iodine 10 0-9 9-0 0-1 Pe 3
An experiment done under these conditions shows that when iodine is added to starch, and the lyophobic starch iodide is precipitated from solution by a divalent ion, a point is reached at which the filtrate contains neither starch nor iodine. Previous to this point the filtrate contains an excess of starch, and after this point the filtrate contains an excess of iodine. It is difficult to postulate an hypothesis which would explain these results on the basis of physical adsorption. The occurrence of a definite point at which the filtrate contains neither starch nor iodine suggests that iodine reacts chemically with the starch, and that at this point there is a definite chemical equivalence of the starch and iodine. This deduction is strengthened by the experimental observations that the equivalent point 1s not affected by dilution, temperature, or precipitating electrolyte.
(1) Varying concentration of starch. In the following experiment the quantity of starch, contained in a final volume of 20 cc., was varied from 10 ce. of a 1 % solution to 10 ce. of a 0-2 % solution. The same quantity of precipitating electrolyte (MgSO,) was added in each case and the quantity of iodine required to produce a filtrate containing neither starch nor iodine determined :
1 % starch H,0 101% MgS0O, (1) Filtrate 10 ce. 5:5 ce. 4-75 ce. 0:2 ce. Excess iodine
10 5:75 4-5 0-2 No iodine
COMPOSITION OF STARCH 33
0-8 % starch H,O 0-1%I1 MgSO, (1) Filtrate 10 ce. 6-2 ce, 3°8 cc. 0-2 ce. Excess iodine 10 6-4 3-6 0-2 No iodine
0-6 % starch H,O 10-1% Mgso, (1) Filtrate 10 ce. 7:2 ce. 2°8 cc. 0:2 ce. Excess iodine 10 7:4 2-6 0-2 No iodine
0-4 % starch H,O 101% MgSOoO, (M) Filtrate 10 ce, 8-2 cc. 1-8 ec. 0-2 cc. ixcess iodine 10 8-4 16 0-2 No iodine
0-2 % starch H,O iO 129%, MgsO, (JZ) - Filtrate 10 ce. 9-0 ce. 1-0 cc. 0-2 cc. Excess iodine 10 9-2 0:8 0:2 No iodine
The mean values obtained from the above results are as follows:
Ce. of iodine
% starch required Jodine/starch 10 ce. of 1 % 4-625 4:6 10 ce. of 0-8 % 3-7 * 4-6 10 ce. of 0-6 2:7 4:5 10 ce. of 0-4 1-7 4-25 10 ce. of 0-2 0-9 4:5
The direct relation between the amount of starch and the quantity of iodine required to precipitate it in the presence of an electrolyte again indicates the chemical equivalence of the starch and iodine.
Similar experiments were made in which the precipitating electrolyte was varied from 0-2 cc. MgSO, (17/10) to 0-2 ce. MgSO, (M). There was no change in the point at which the filtrate contained neither starch nor iodine. Similarly varying the temperature at which the experiment was done from 15° to 40° did not influence the quantitative values of the reaction.
(ii) The rcodine equivalent of starch. All these results indicate that when iodine is added to starch a quantitative reaction takes place between the starch and the ionised iodine. The iodine equivalent of starch was determined from an experiment in which the amounts of starch and iodine in the solutions used had been accurately determined.
It was found that when 10 cc. of a starch solution containing 0-087 g. of starch were treated with 0-0068 g. of iodine the filtrate obtained after precipita- tion by MgSO, contained neither starch nor iodine. Therefore, on the above hypothesis of the chemical interaction of starch and iodine,
0-0068 g. of I is equivalent to 0-087 g. of starch, or 127 g. of I is equivalent to 1635 g. of starch.
Now the sum of the atomic weights of CgH,90; is 162; and the empirical formula for starch has been assumed to be (CgH,)0;)n. Therefore the least value of n, assuming that one molecule of starch reacts with one atom of iodine, is 10.
The experimental results obtained on adding colloidal iron to starch show that starch contains a variety of polymers. The value n = 10 can therefore
Bioch, x11 :
34 J. MELLANBY
be taken only as a mean value for a number of starch complexes in which n is continually varying.
(iv) The codine equivalent of starch granulose (y). The iodine equivalent of starch granulose (y) was determined in the way described in the previous experiment. The following results give the details of an experiment:
100 ce. of starch were precipitated by 20 ce. of colloidal iron and 1-2 ce. K,SO, (MW). 5 ce. of the resulting filtrate containing starch granulose (y) were put into a series of tubes to which were added varying quantities of iodine and MgSO, (47/10) to make the volume 10 cc. The filtrates were tested with starch and iodine.
Starch granulose (y) H,O 01%1 MgSO, (27/10) Filtrate 5 ce. 1-9 ce. 3-0 ce. 0-1 ce. Excess of iodine 5 2:9 2-0 0-1 53 “A 5 3-9 1-0 0-1 + a 5 4°] 0-8 O-1 Nil 5 4:3 0-6 0-1 Excess of starch
The original starch solution gave as equivalent quantities 5 cc. starch and 4 ec. iodine. The amount of starch granulose (y) contained in the filtrate was 10-6 °% of the total starch contained in the original solution.
Therefore the iodine equivalent of starch granulose (y) compared with that of an equal quantity of the original starch was approximately as two to one. In other words if 1600 g. of the original starch were equivalent to a gram atom of iodine, approximately 800 ¢. of starch granulose (y) were equivalent to the same quantity of iodine.
In the case of starch (CgH, 05) 19 1s equivalent to I and for starch granulose (y) (CgH, 905); 18s equivalent to I.
(B) The adsorption of codine by starch iodide.
In the foregoing experiments the amount of iodine added to the starch solution before the addition of the precipitating electrolyte was relatively small, and under these circumstances it has been shown that the reaction passes through a definite stage previous to which there is an excess of starch, and after which there is an excess of iodine, in the solution. This result is not compatible with the hypothesis that starch adsorbs iodine ab initio. The question however arises whether starch iodide when formed further reacts with an excess of iodine, contained in the solution, in a chemical or physical way. Substances containing a variable quantity of iodine in relation to the starch have been isolated and this fact suggests that a physical process enters into the reaction after the formation of the starch iodide compound. The following experiment illustrates the relation of the iodine contained in the filtrate and that added to the original starch solution when the quantity of added iodine is continuously varied.
COMPOSITION OF STARCH 35
Excess of Iodine Starch 1% Iodine1% H,O MgSO, (JZ) I in filtrate precipitated
25 cc. 1-5 ce. 22-5 cc. ces (starch in filtrate) 1-5 ce. 25 2-0 22:0 1 0 2-0 (a) 25 2:5 21-5 ] 0-15 ee. 2°37 25 5-0 19-0 ] 0-89 4-11 25 7-5 16-5 ] 2-8 4-7 25 10-0 14:0 i 5-03 4:97 25 12-5 11-5 ] 6-9 5-6
The figures show that after the equivalent point (x) at which the filtrate contains neither starch nor iodine, the starch iodide precipitated contains a varying quantity of iodine, the amount of iodine in the “starch iodide” con- tinually increasing as the iodine content of the original mixture increases. In fact the figures show that after the point (z) the precipitated “starch iodide” takes down iodine with it from solution according to the recognised laws of adsorption. Therefore in the action of iodine on starch two separate and distinct phases may be recognised.
(1) The iodine reacts with the starch in a quantitative manner, approxi- mately 1 gram-atom of iodine being equivalent to 1600 g. of starch or 800 g. of starch granulose (y).
(2) The lyophobic starch iodide thus formed when precipitated by electro- lytes adsorbs iodine from solution according to the recognised laws of adsorp- tion provided an amount of iodine is present in excess of that required for the first reaction.
SUMMARY.
1. Precipitation of starch by colloidal iron shows that starch granulose can be separated into three fractions, (a), (8) and (y), forming 80%, 9 % and 11 % respectively of the starch granulose. (a) is precipitated by colloidal iron ° only, (8) by colloidal iron and electrolytes, and (y) is not precipitated by colloidal iron under any conditions.
2. Precipitation of starch by iodine and electrolytes shows (a) that starch contains an insoluble constituent which does not react with iodine (amylo- cellulose), (b) that all the soluble constituents of starch are precipitated by iodine in the presence of electrolytes, and (c) that the final fraction precipitated by iodine gives a brown colour with iodine.
3. The results (1) and (2) indicate that starch contains a variety of polymers varying in complexity from amylodextrin to amylocellulose, the relative quantities of dextrin and cellulose being small whilst the bulk of the eranule is composed of amylogranulose (a).
4. Todine reacts with starch in a quantitative manner forming starch iodide. Approximately 1600 g. of starch are equivalent to 127 g. of iodine, or (CgH,905)19 is equivalent to a gram-atom of iodine. In the case of starch eranulose (y) approximately 800 g. react with 127 g. of iodine; or (CgH 995); is equivalent to a gram-atom of iodine.
3—2
36 J. MELLANBY
5. Starch iodide adsorbs iodine from solution so that after the equivalent
point the amount of iodine contained in the precipitated starch iodide is i function of the amount of iodine contained in the original solution. tm]
REFERENCES.
Barger and Field (1912). J. Chem. Soc. 101, 1304.
Bottazzi and Victoroft (1910). Atte R. Accad. Lincer, |v], 19, i, 7. Fernbach and Wolff (1904). Compl. Rend. 138, 819.
Harrison (1911). Zettsch. Chem. Ind. Kolloide, 9, 5.
Kiister (1894). Annalen, 283, 360.
Maquenne and Roux (1903). Compt. Rend. 137, 88.
(1905). Compt. Rend. 140, 1303.
Padoa and Savaré (1906). Gazzetta, 36, i, 313.
V. OBSERVATIONS ON THE ACCURACY OF DIFFERENT METHODS OF MEASURING SMALL VOLUMES OF FLUID.
By FREDERICK WILLIAM ANDREWES. From the Pathological Laboratory, St Bartholomew’s Hospital, London.
(Received February 13th, 1919.)
Every serologist has his own favourite method of measuring the small volumes of different fluids which he needs in performing complement fixation reactions or in making a series of serum dilutions for agglutination. He usually regards the method to which he is accustomed as the best and may resent any imputa- tion on its accuracy, but few serologists take the trouble to check the method they employ by careful control observations. There is no convenient method for determining the percentage of serum in a dilution sufficiently closely, but there are plenty of chemical substances which can be substituted for serum. One cannot, indeed, carry out volumetric analyses on the small volumes involved in Wright’s method of preparing serum dilutions in a capillary pipette, but one can use larger volumes in the same way and analyse the results satisfactorily and can very safely reason that errors detected with the larger volumes are likely to be exaggerated with the smaller.
I have lately had occasion to enquire into possible errors in my own sero- logical technique and I have carried out a certain number of observations, as carefully as I was able, on the lines suggested above. Although there is nothing astonishing in the results it may be of service to put them on record. It has been a gain to me to ascertain the degree of my own error and it would probably be a useful exercise for any serologist to check his methods in similar fashion.
The fluids chosen for measurement were (1) a saturated solution of iodine in saturated potassium iodide solution, and (2) a not quite saturated solution of calcium chloride in water. The latter was on the whole the more satisfactory, being cleaner to work with. The dilutions were titrated in the ordinary way, the iodine with N/10 and N/100 thiosulphate and a starch indicator, and the calcium chloride with N/10 and N/100 silver nitrate and a potassium chromate indicator. My thanks are due to Dr W. H. Hurtley, Lecturer on Chemistry at St Bartholomew’s Hospital, for kind advice and help in the matter.
There are two ways of measuring fluids, apart from weighing, which is out of the question for serological work. One is measurement by pipette, and the
38 lk. W. ANDREWES
other measurement by drops, and I was anxious to compare the accuracy of these two methods.
In measuring by pipette the serologist is at a disadvantage as compared with the chemist. Volumetric analysis has been developed into a fine art by generations of chemists and the degree of accuracy obtainable is very high. The chemist has his standard burettes and pipettes and has been trained how to use them. Unless the serologist has been trained as a chemist he is apt to be less scrupulous in the use of the instruments at his command, and he lies under the further disadvantage that he commonly has to measure very small quantities of fluid by means of instruments much smaller and more liable to error than those used by the chemist. The first problem that arises is therefore the degree of error likely to be present in careful pipette measurements of small volumes of fluid, ranging from 1 ee. to 0-01 ec., such as the serologist often has to employ.
THE RANGE OF ERROR IN SMALL PIPETTE MEASUREMENTS.
Those who work with graduated | cc. and 0-1 ce. pipettes usually employ as motive force either a rubber teat or a mouthpiece connected with the pipette by a rubber tube. Some such method is necessary with a 0-1 cc. instrument, though with a 1 cc. pipette one can work by gravity, controlling the flow with the finger only.
To check the accuracy of delivery of small volumes from such pipettes the following procedure was adopted. A concentrated solution of calcium chloride was carefully titrated against N/10 AgNO,: 1 cc. of the CaCl, solution was found equivalent to 61-4 cc. V/10 AgNO; the measurement of the | cc. was performed with great care by Dr Hurtley. Two Lautenschlager pipettes were washed out with strong sulphuric acid and then repeatedly with distilled water: one was a | cc. pipette graduated in tenths and hundredths, the other a 0-1 ce. pipette graduated in hundredths and two-hundredths. A well-fitting rubber teat was attached to each. I then attempted to deliver into evaporating dishes amounts of the calcium chloride solution varying from 1 cc. to 0-01 ec., and titrated them against AgNOs, using a decinormal solution for the larger and a centinormal one for the smaller amounts. Even 0-01 cc. of the calcium chloride solution required 6 cc. of centinormal AgNO,, so that the titration could be carried out with considerable accuracy. At the same time my colleague, Dr R. G. Canti, who was provided with a similar pair of Lauten- schlager pipettes, which he was accustomed to use with a mouthpiece and rubber tube, performed similar measurements. There was thus an opportunity for comparing the teat with the mouthpiece as a motive force.
The results of this experiment are shown in the following table. In the second column is shown the expected value of each amount, assuming the correctness of the preliminary titration, in the third and fifth columns the value actually found, and in the fourth and sixth the percentage by which the value found differed from what it ought to have been.
MEASUREMENT OF SMALL VOLUMES 39
Table I. Errors in measurements with small graduated pipettes.
Measurements with 1-0 ce. pipette.
Correct value F. W. A. using teat R. G. C. using mouthpiece
Amount ince. V/10 ee eS EIN IEE measured AgNO, Value found °%% deviation Value found °% deviation
1-0 ce. 61-4 60-8 — 0-98 60-0 — 2-28
1-0 61-4 61-9 + 0-81 62-0 + 0:97
1-0 61-4 61-1 — 0-49 — a
0-5 30-7 30-5 — 0-66 — _—
0-1 6-14 5-7 —- 7:17 6-0 — 2-28 0-1 6-14 5:75 — 6:36 -— —
Measurements with 0-1 ce. pipette.
O-lce: 6-14 6-0 — 2-28 6-15 + 0-16 O-1 6-14 — == 6-5 + 5:86 0-05 3:07 2-95 — 3-91 3-015 — 1-79 0-01 0-614 0-605 — 1-47 0-68 + 10-74 0-01 0-614 0-75 + 22-15 0-775 + 26-22 0-01 0-614 0-635 + 3-42 0-74 + 20-52
The chief interest of these figures is that they represent the actual results obtained by two ordinary workers, reasonably practised in pipette measure- ments, and done with extra care because there was a certain emulation as to who could produce the best results. Doubtless there are many workers who could do better: we only regard ourselves as representing the averagel.
It is evidently possible to measure | cc. or 0-5 cc. with reasonable accuracy : my own four measurements of this order were all within | % of the correct amount. Attempts to deliver 0-1 cc. from a | cc. pipette were less successful. This quantity can be more accurately delivered from a 0-1 cc. instrument, and one of Dr Canti’s efforts was almost exact; but he failed to repeat it. The margin of error is clearly greater with the smaller instrument, and when it comes to delivering 0-01 cc. the error is apt to become a serious one. The grosser errors were always in the direction of delivering too much, probably owing to transfer of fluid on the exterior of the nozzle. It is clearly unwise for the average worker to rely on the measurement of one hundredth of a cc. from a pipette with any accuracy.
Two further points came out in the course of this experiment, though they are not indicated in the table. A shghtly larger volume is delivered when the pipette is held vertical and emptied slowly. In delivering one-tenth of the total graduated length, more accuracy is attained by measuring one of the middle tenths than the terminal one.
1 [ communicated my results to Dr Carl Browning of the Middlesex Hospital, who thereupon undertook some similar observations—using a narrow 0-1 cc. pipette, provided with a rubber tube and mouthpiece. The fluid used was hydrochloric acid, and the measured volumes were titrated against N/300 sodium hydrate. Twenty successive measurements of 0-01 cc. were made, and of these sixteen differed from the mean by less than 2 %, while the extreme variations from the mean were less than 5 %. Eighteen measurements of 0-03 cc. were also made, and fifteen of these were within 1-3 % of the mean, while the extreme variations from the mean were 2-9 %, These efforts are clearly better than those of Dr Canti and myself.
40 KF. W. ANDREWES
KRRORS INCIDENT TO THE PREPARATION OF A SERIES OF DOUBLING DILUTIONS,
Inasmuch as this is one of the commonest procedures in agglutination work, considerable interest attaches to the accuracy of the different methods by which it can be carried out, and especially to a comparison of the pipette and drop methods. I therefore performed a number of experiments with this object in view.
In order to have an idea of the standard of accuracy which might be attainable I invited Dr Hurtley to prepare me a progressive series of doubling dilutions of the same calcium chloride solution as was used in other experi- ments. The solution was at this time slightly stronger: (this was the first time it had been used); the preliminary titration showed that | cc. = 61-6 N/10 AgNO,. Dr Hurtley prepared the dilutions with great care, employing all the precautions incident to volumetric analysis, and even washing out and drying the pipette between successive measurements. The volume measured at each step was 2-5 cc. Five dilutions were prepared, from 1 in 2 to | in 32, each being made from the one preceding it, so that any error might be cumu- lative. The results are shown in Table II.
Table Il. Deviations from accuracy in a series of dilutions prepared by a chemist. Primary solution; 1 ce. =61-6 NV/10 AgNO,.
% deviation from
Dilution Correct figure Result on titration correct figure in’ (2 30:8 30-65 —0°5 lin 4 15-4 15:45 +0°3 lin 8 17 7:85 +1-9 lin 16 3°85 3°80 -1:3 1 in 32 1-92 2-00 +4-1
The results seen in this table show that a trained chemist, using adequate methods, can prepare a series of desired dilutions with a high degree of accuracy. The slight apparent errors are now +, now —, which suggests that they may depend upon errors in titration rather than in the preparation of the dilutions.
But no serologist can work in this laborious way. He usually employs his pipette in quite a different manner. It is usually not a “delivery” pipette, but one graduated to the point and he blows out the contents as completely as he can. This process inevitably leads to error unless the instrument is dried out at each operation, and what is more, in preparing a series of doubling dilutions in the ordinary way, the error is cumulative. In order to get some idea of how great it may become, I made the two following experiments, using a delivery pipette in the manner of a serologist.
I took an ordinary 4 cc. pipette and measured into each of eight test tubes 4 cc. of distilled water, as carefully as possible but expelling the whole amount of fluid by blowing. The pipette was then dried out, and 4 cc. of strong calcium
MEASUREMENT OF SMALL VOLUMES 41
chloride solution (1 cc. = 61:5 cc. N /10 AgNO.) added to the first tube. The mixture was taken up into the pipette and expelled six times, to ensure thorough mixing and to wash out the pipette, and then, without any other washing or drying, 4 cc. of the mixture were transferred to tube 2 and the process repeated, and so on to the eighth tube, the pipette being at each stage washed out six times with the mixture just made. The results on titration were as follows:
Table III. Showing cumulative error from incorrect use of pipette.
Primary solution; 1 cc. =61-5 cc. N/10 AgNQO3. % deviation from
Dilution Correct figure Result on titration correct figure lin 2 30-75 31-0 + 0-81 lin 4 15-375 15-825 + 2-92 lin 8 7-687 8-05 + 4:72 lin 16 3-843 4-0 + 4-08 lin 32 1-921 2-06 + 7-23 lin 64 0-960 1-04 + 8-33 1 in 128 0-48 0-535 + 11-45 1 in 256 0-24 0-275 + 14-58
This series illustrates very well the progressive nature of the error in a short series of dilutions when the pipette is wrongly used. If it can amount to 14 % using so large a pipette as one of 4 cc., how much greater must it be with a capillary pipette in which the surface wetted bears a far larger propor- tion to the volume of fluid measured.
A method of preparing serum dilutions sometimes employed is as follows: Supposing one has three rows of tubes to fill, as in an ordinary T.A.B. agglu- tination, a tenfold dilution of the serum is prepared, and one volume placed in each of the first three tubes. Three volumes of the tenfold diluted serum are then put in a watch-glass and three volumes of saline added and mixed. A volume of this mixture is placed in each of the second three tubes, leaving three in the watch-glass, to which a further three volumes of saline are added for the third dilution and so on. The pipette is thus alternately wetted with pure saline and with the dilution at which one has arrived. If, on reaching the end, the volume remaining in the watch-glass is measured, it is never three volumes, but always less, by a half or a third of a volume.
I have tested the degree of inaccuracy of this obviously inaccurate method in the following manner. I used a home-made pipette of narrow glass tubing, holding about 2 cc. to the top mark: it was actuated by a rubber bulb. The solution employed was a strong one of iodine in potassium iodide. I put two pipettesful of water into a glass dish, washed out the pipette several times with iodine solution, and then added two pipettesful of iodine solution to the dish and mixed well. Two pipettesful were transferred to tube 1, and then two more of water added to the dish and mixed. Hight tubes were thus filled in succession. The results on titration with sodium thiosulphate are seen in
Table IV.
42 KF. W. ANDREWES
Table IV. Showing cumulative error from erroneous technique. Primary solution; | ce. =61-14 cc, NV/10 thiosulphate.
©, deviation from
Dilution Correct figure Result on titration correct figure lin 2 30:57 30°7 + 0-42 Lins 4 15-28 15-1 1-8 lin 8 7-64 7:45 - 38 lin 16 3°82 3-49 - 8-7 Lin) 32 1-91 1-63 — 14-7 lin 64 0-955 0-795 — 16-8 Lin 128 0-477 0-36 — 24:6 L in 256 0-238 0-175 — 26:5
Whereas in the preceding experiment the error was a progressive one on the + side, this one shows a similar and worse error on the — side. The direction of the error depends on the precise departure from the proper use of the pipette. These are not the only experiments of the sort which I have made, but all have shown the same sort of thing. These are selected and set out here in detail to illustrate the fact that an error which appears negligible in the individual measurement becomes a serious one when it is cumulative. I know of no way in which a serologist is likely to use a pipette which will avoid such error: this is not said in blame, for he has not the time to employ the technique of the trained chemist. But he ought to know what his error is liable to be.
THE DROP METHOD OF MEASURING.
All the sources of error which attend measurement by pipette are entirely avoided in measurement by drops, and in their stead we encounter the new set of difficulties which attends the delivery of equal drops. These however are more easily overcome than in the case of the pipette. Donald has shown us a simple technique by which, with a little practice, it is easy to deliver sub- stantially equal drops of different fluids. [1915, 1916. ]
The three important factors governing the size of a drop issuing from a vertically held pipette are (1) the external diameter of the nozzle, (2) the surface tension of the fluid, and (3) the rate of dropping. By the aid of a wire and drill gauge, such as can be bought at any good tool shop, pipettes of constant size can be made as they are needed. The variation in surface tension of differ- ent fluids, such as normal saline, fresh human serum, or phenolated rabbit serum, can be compensated by determining which holes in the gauge will furnish pipettes delivering sensibly equal drops of these fluids: once determined this has not to be done again. With a little practice any one can keep his drop rate reasonably constant at about one per second. All this can be read in detail in Donald’s published papers.
There are two ways in which one may employ the method in measuring small volumes of fluids of different surface tension. With a single pipette one can determine the relative size of the drops furnished by each fluid, keeping the conditions constant—e.g. by ascertaining how many drops of each go to
MEASUREMENT OF SMALL VOLUMES 43
make up | cc. It is then easy to calculate how many drops of each must be mixed to form any desired dilution. The objection to this way of working lies in the impossibility of measuring fractions of a drop. The second plan, that of calibrating two pipettes to deliver equal drops, is therefore much to be preferred.
In endeavouring to test the accuracy of the drop method, I prepared a series of eight doubling dilutions of strong calcium chloride solution. The drops of this solution and of distilled water were found to be so nearly equal in size (256 to 245) that I did not attempt to make two pipettes to deliver equal drops, but adopted the first of the two plans just mentioned; I used one pipette, calculating the numbers of drops required and neglecting fractions of a drop. I did not therefore employ the method in the best way; nevertheless, as will be seen from Table V, the results were surprisingly good.
Table V. Showing smallness of error with drop method.
Primary solution of CaCl,; 1 cc. =61-5 ec. N/10 AgNO,. % deviation from
Dilution Correct figure Result on titration correct figure pine? 30-75 31-25 EG Ain 4: 15-375 14:8 —- 3-7
iin 1S 7-687 7-65 — 05
lin 16 3°84 3-85 + 0:3
Tim 32 1-92 1-95 + 1:5 lin 64 0-96 0-945 - 16
1 in 128 0-48 0-49 ae ll
1 in 256 0-24 0-265 + 10-4
It will be seen that the errors here are, with the exception of the eighth dilution, of the same small order as in the series of dilutions prepared by Dr Hurtley by “chemical” use of the pipette (see Table II). The result is very much better than anything I myself could achieve with a pipette. On the first two occasions on which I tried to make this series of dilutions the results had been wide of the mark, owing to miscalculation of the relative size of the drops of the two fluids. After Mr Donald had been so kind as to visit my laboratory and to point out certain mistakes in technique, the thing be- came easy, and the above figures represent my third attempt. The dropping was done at constant rate from a burette, provided with a Marriott’s tube to keep the pressure even, and not by hand.
The advantages claimed by Donald for his drop technique appear to me to be justified. Anyone can make and calibrate his own pipettes, easily and cheaply as they are required. There is no cumulative error in preparing an ascending series of dilutions. The drop method can be used with the same accuracy in measuring very small volumes as larger ones, whereas it is with the small measurements frequently required by the serologist that the pipette method shows up to such disadvantage. Lastly, provided that the conditions for its use are duly observed, the drop method seems intrinsically more accurate than measurement by pipette.
44 F. W. ANDREWES
CONCLUSIONS.
The figures which have been given above are doubtless less exact than could have been obtained by weighing. Any errors which may have been incident to the process of volumetric analysis are of course superadded to those depending on faulty technique in the preparation of the series of dilutions. The errors of titration are probably, however, less than 1 % and [ believe that the results obtained are of sufficient accuracy to allow of a fair estimate of the degrees of error which may attend the measurement of small volumes of fluid in serological work.
The chief conclusions which appear to follow from these few observations are as follows:
(1) Volumes of 1 ec. and 0-5 ce. can be measured by pipette with reason- able accuracy. The error attendant on the attempt to deliver 0-1 cc. may amount to 5%, even when a 0-1 cc. pipette is employed. The delivery of 0-01 ce. from a pipette may be exceedingly inaccurate.
(2) The only way in which really accurate results can be obtained with a pipette in preparing an ascending series of dilutions, is to use it as a delivery pipette and to wash and dry it out between successive measurements. The alternative method, that of using it as a delivery pipette, mixing the dilution by shaking, and then washing out two or three times with the mixture, rejecting the washings, is one hardly adapted for serological work. Unless such precautions are taken the small error introduced, perhaps only one of 0-5 °% in a single measurement, though probably much more with a capillary pipette, becomes magnified into one of 10 % or more at the eighth dilution.
(3) The drop method is greatly to be preferred, for serological work, to the pipette method, provided it is properly carried out with calibrated pipettes.
It may be urged that in a process such as the agglutination test, beset with liabilities to inaccuracy, an error of 10 % or so in the serum dilutions is not of much moment. For rough work perhaps it is not, but for accurate work it is. Improvement in agglutination technique can only be brought about by endeavouring to correct the individual sources of error which attend it. This is equally true of such methods as are involved in the technique of the Wasser- mann reaction. My excuse for publishing these few observations is that I have learned a good deal from them myself, and that it is possible that others may find them not without interest.
REFERENCES.
Donald (1915). Lancet, ii, 1245. —— (1916). Lancet, ii, 423.
VI. ON THE SEPARATION OF ANTITOXIN AND ITS ASSOCIATED PROTEINS FROM HEAT- DENATURATED SERA. |
By ANNIE HOMER. From the Lister Institute of Preventive Medicine. (Received February 7th, 1919.)
Ir is recognised that in unheated antidiphtheritic and antitetanic sera the antitoxins are associated with the “salt-soluble” globulins precipitated be- tween 30 and 50 % of saturation with ammonium sulphate. For this reason, in the routine concentration of such sera, it has been customary to isolate and dialyse the protein precipitated between these limits and, recently, observa- tions have been published in regard to the factors limiting the degree of con- centration thereby obtained {[Homer, 1917, 1].
Further work has shown that, in heated sera, the association of the anti- toxin with these so-called Second Fraction Precipitates is affected by the extent of the denaturation of the proteins induced during the preliminary heating of the serum. The heat-denaturation is evidenced by an increased precipitability of each of the individual serum proteins by ammonium sulphate, a phenomenon which is accompanied by an increase in the ratio of “salt- insoluble” to “‘salt-soluble” globulins and by an association of antitoxin with the “salt-insoluble” protein thus formed [Homer, 1917, 2, 1918, 1}.
In the light of these observations it was necessary to extend the scope of the previous investigations so as to ascertain:
I. To what extent the limits for the precipitation with ammonium sulphate can be narrowed so as to include, in the Second Fraction Precipitates from heated sera, only those proteins with which the antitoxin is definitely associated.
II. Whether or no the antitoxin is evenly distributed throughout the protein fractions precipitated at successive stages between the limits indicated in (I).
I. THE LIMITS FOR THE PRECIPITATION OF THE SECOND FRACTION PRECIPITATES.
The gradient for the curves representing the precipitation of the serum proteins at progressively increasing percentages of saturation with ammonium sulphate indicates that there are no critical points for the precipitation of the eu- or of the pseudo-globulin or of the serum albumin.
16 A. HOMER
The protein precipitated from unheated sera between 30 and 50%, of saturation with this salt consists mainly of pseudoglobulin admixed with small amounts of euglobulin and of albumin, whereas that similarly precipi- tated from heated sera contains a lower proportion of admixed euglobulin and a greater proportion of albumin. As was to be surmised, the relative propor- tion of the proteins precipitated from heated sera at this stage was influenced by the extent of the denaturation induced during the heating of the serum.
Kxperimental work was therefore undertaken to ascertain to what extent (a) the upper, and (6) the lower limit for the degree of saturation with am- monium sulphate could be changed so as to retain in the Second Fraction Precipitates only those proteins with which the antitoxin is definitely associated.
(a) The upper limit for the precipitation of the Second Fraction Precipitates.
The following observations show that the extent of the precipitation of albumin with the Second Fraction is a function of the heat-denaturation induced in the serum.
To separate volumes of plasma were added the requisite volumes of acid or of alkali to adjust the | H’] to suitable values. The separate sera, after being heated at a temperature of 57° for six hours, were made 50 % of saturation with ammonium sulphate and filtered. From the gravimetric determinations of the albumin content of the filtrates was calculated the percentage of albumin precipitated with the Second Fraction Precipitates. It was found that, where the heat-denaturation had been of the order of 60, 50, 45, 35 and 22 % re- spectively the Second Fractions contained 75, 55, 45, 40 and 22 % of the total albumin of the plasma.
As the antitoxin is attached to the pseudoglobulin and not to the albumin fraction of the serum proteins, the presence of albumin in the Second Fraction Precipitates must reduce the degree of concentration; its exclusion is, there- fore, desirable.
It was found that the precipitation of heat-denaturated albumin could be eliminated by a sufficient lowering of the degree of saturation with ammonium sulphate. But, in the concentration of sera, where the main object is to ensure the complete recovery of the antitoxin in association with a minimum per- centage of the proteins of the original plasma, such a procedure is not practic- able. For, as there is no sharp line of demarcation between the precipitation of pseudoglobulin and of albumin from heat-denaturated sera, the lowering of the upper precipitation limit to the extent required for the complete exclusion of the albumin leads to considerable losses of antitoxin; the con- centration of ammonium sulphate thereby adopted is insufficient to pre- cipitate the whole of the pseudoglobulin and its associated antitoxin.
Thus, from sera in which a heat-denaturation of 35 °% had been induced, there was a slight precipitation of denaturated albumin at 40 % of saturation
CONCENTRATION OF HEAT-DENATURATED SERA 47
with ammonium sulphate; the extent of the precipitation rapidly increased with the further addition of ammonium sulphate. In order, therefore, to avoid the precipitation of albumin in the Second Fraction Precipitates it was necessary to lower the upper precipitation limit from 50 to 38 % of saturation.
On the other hand the precipitation of the pseudoglobulin and antitoxin from the same serum was not complete until the concentration of the sulphate had reached 44 % of saturation. Thus, the precipitate isolated between 30 and 42 % of saturation contained only about 90% of the antitoxin, while rather less than 80 % of the antitoxin was associated with the protein pre- cipitated between 30 and 40 % of saturation with the salt.
This difficulty was even more pronounced in the fractional precipitation of the antitoxin and its associated protein from sera showing a more extensive heat-denaturation. Thus, from sera showing a heat-denaturation of 50 % or more there was a marked precipitation of albumin even at 26 % of saturation with ammonium sulphate. It is obvious that in such cases the elimination of albumin from the end products is impracticable.
In dealing with the routine sera in which the heat-denaturation was of the order of 35 % or less, it was decided that the upper limit should not be reduced to less than 44 °% of saturation. Even this reduction led to the exclusion of a considerable proportion of the denaturated albumin which would have been precipitated at 5C °%, and, as a result, the end products in the former case showed a greater degree of concentration than was obtained in the latter. Thus, from a given serum, while the fraction isolated between 30 and 50% of saturation with ammonium sulphate contained 33-3 % of the total serum proteins, that between 3C and 44 °% of saturation contained only 25 % of the proteins. The load of antitoxin carried by the protein fraction thus isolated between the wider limits was of the order of 16,000 units and that between the narrower limits of 17,500 units per g. of protein.
(b) The lower limit for the precipitation of the Second Fraction Precipitates.
Having demonstrated that the degree of concentration is improved by the lowering of the upper limit for the precipitation of the Second Fraction by ammonium sulphate, the next step was to ascertain to what extent the lower limit could be raised so as to ensure the exclusion of that portion of the heat- denaturated globulin to which no antitoxin is attached.
The preliminary precipitation and separation of the First Fraction Pre- cipitates (30 % of saturation with ammonium sulphate) from unheated sera remove most of the euglobulin together with a small amount of pseudo- globulin and antitoxin. The pseudoglobulin being “salt-solubie” can be recovered by the extraction of the precipitates with brine. On the other hand the corresponding precipitates from heated sera contain relatively greater amounts of heat-denaturated pseudoglobulin of which a certain proportion
18 A. HOMER
has been converted into a “salt-insoluble” condition; the extent of the con- version is a function of the denaturation.
In sera showing a heat-denaturation of 40° or less, the denaturated pseudoglobulin precipitated with the First Fraction Precipitates also carried down appreciable amounts of antitoxin but, as the latter was associated with salt-soluble protein, it could be recovered by an extraction of the precipitates with brine.
The precipitation of antitoxin from sera showing a more extensive heat- denaturation began at, and was completed at, a much lower concentration with ammonium sulphate than in the above. The proportion of antitoxin carried down with the First Fraction Precipitates increased progressively with the heat-denaturation. At the same time, owing to the increasing in- solubility of the denaturated protein, there was a corresponding decrease in the percentage recovery of antitoxin and its associated proteins in a form suitable for clinical use. In fact, under some conditions, e.g. in sera more acid than P,, 4-5 or in sera containing excessive amounts of phenol and its homo- logues, or of ether, chloroform, etc., while practically the whole of the antitoxin is precipitated with the First Fraction Precipitates, only a small percentage of it can be recovered in solution, as the protein to which it is attached has become almost entirely “salt-insoluble.”
From these observations it is clear that, in such sera, the lower limit for the precipitation of the Second Fraction with ammonium sulphate must be re- duced rather than raised.
The attempts to increase the degree of concentration by raising the limit for the precipitation of the First Fraction Precipitates were, therefore, con- fined to such sera as showed a denaturation of less than 40 °% and the problem was accordingly studied in respect of sera showing a denaturation (1) of from 25 to 40 %, and (2) of less than 25 %.
The end products from (1) were clear and readily filtrable. Those from (2) were cloudy and, owing to the suspension of heat-denaturated protein, could only be filtered with difficulty; they also showed a lower degree of concen- tration and a lower percentage removal of the serum proteins than was obtained from the sera in (1).
(1) Sera in which a heat-denaturation of 25 to 40.% had been induced.
It was found that in such sera the amount of protein precipitated with the Second Fractions isolated between 30 and 44 % of saturation with ammonium sulphate was not appreciably less than in those isolated between 33 and 44 % of saturation. On the other hand by raising the lower limit for the precipita- tion of the First Fraction Precipitates to 36 % of saturation, the amount of protein isolated between 36 and 44 °% of saturation was only 70 % of that obtained between the wider limits. At the same time, however, there was a corresponding proportional increase (30 %) in the percentage of the antitoxin precipitated with the protein.
CONCENTRATION OF HEAT-DENATURATED SERA 49
It is evident that no material gain resulted from the raising of the lower precipitation limit. On the contrary, the method of procedure was disadvan- tageous. It transpired that, while the antitoxin carried down with the First Fraction precipitated from the serum in question at 30 % of saturation with ammonium sulphate could be completely recovered by extraction of the precipitates with brine, that carried down with the precipitates at 33 and at 36 % of saturation with the sulphate could not be recovered to the same extent. In the latter cases it seems as though the greater concentration of sulphions, presumably by an electrical charging of the protein molecules, favours the conversion of the heat-denaturated pseudoglobulin and _ its associated antitoxin into a “salt-insoluble” condition; the antitoxin attached to the protein thus rendered “salt-insoluble” cannot be recovered in a form suitable for clinical use even though its activity has in no way been impaired.
(2) Sera in which a heat-denaturation of less than 25 % had been induced.
As was to be expected the raising of the lower limit for the isolation of the Second Fraction Precipitates from sera showing a heat-denaturation of 25 % or less led to a more marked diminution in the amount of protein precipitated than was shown in the sera discussed in (1).
Thus, in the cases investigated, the precipitate between 30 and 44 % of saturation with ammonium sulphate contained 15 and 48 % more protein than where the limits for precipitation had been 33-44 and 36-44 % of saturation respectively. Moreover, the products obtained from the dialysis of the protein precipitated between the widest limits were cloudy and un- satisfactory; those from the precipitates isolated between the narrower limits were clear, as the increased concentration of ammonium sulphate had given the particles of suspended heat-denaturated protein the necessary charge to ensure their aggregation, precipitation, and separation with the First Fraction Precipitates.
The determination of the antitoxin associated with the Second Fraction Precipitates thus isolated showed that while those precipitated between 30 and 44 % of saturation with ammonium sulphate contained the whole of the antitoxin of the original plasma, those between 33 and 44 % of saturation contained 95 % of the antitoxin, while only 65 °% was associated with the fraction isolated between 36 and 44 % of saturation with the salt.
Hence, notwithstanding the increased precipitation of antitoxin with the First Fraction Precipitates induced by the raising of the lower limit of satura- tion with ammonium sulphate to 33 and 36 % respectively, the proportionally greater removal of protein thereby effected led to an increase in the degree of concentration shown by the end products. But, in no case was the concentra- tion increased beyond that which would have been obtained between the wider limits of precipitation had the serum been subjected to the preliminary adjust- ment recently advocated {Homer, 1918, 2].
Bioch. xut 4
50 A. HOMER
These observations show that the cloudy end products of unadjusted sera can be avoided by precipitating the First Fraction at 33 or 36 °%, of saturation with ammonium sulphate. Such an expedient will probably appeal to those engaged in routine work as being a simpler process than the preliminary adjust- ment of the serum. However, for the reason given above (p. 49), those who adopt the procedure must be prepared to face losses considerably greater than were recorded in my previous communication | Homer, 1918, 2], where the Kirst Fraction was precipitated at 30% of saturation with the sulphate.
It is thus evident that, in the routine concentration of sera, in which a complete recovery of antitoxin is to be desired, the Second Fraction should not be precipitated within narrower limits than 30 to 44 °% of saturation with ammonium sulphate irrespective of whether the serum has been suitably adjusted previous to its being heated.
Il. Is THE ANTITOXIN EVENLY DISTRIBUTED THROUGHOUT THE PROTEIN FRACTIONS SUCCESSIVELY PRECIPITATED AT STAGES BETWEEN THE LIMITS FIXED IN (I)?
The Second Fraction Precipitates from heat-denaturated sera were further fractionated in successive stages between the limits suggested in (I) in order to ascertain whether the antitoxin was evenly distributed throughout the protein fractions respectively isolated, or whether it was mainly precipitated at a definite concentration with ammonium sulphate.
The consideration of these points was of importance for, in the former case, the evidence would indicate that, in order to isolate antitoxin as a separate entity, means other than the fractional precipitation of the serum with salts must be employed. In the latter case then, for special clinical purposes such as for the intravenous or intrathecal injection of antitoxin, where a high unitage of antitoxin per gram of protein would be advantageous, it might be advisable to precipitate the antitoxin and associated protein within the narrow limits thus indicated, even though by so doing the percentage recovery of the antitoxin were considerably reduced.
Experimental work was accordingly instituted in regard to the successive fractionation of the end products from the concentration of sera by the Homer (1916, 1918) methods. The investigation was confined to the examina- tion of:
(a) Clear End Products, i.e. those from sera in which the reaction had been adjusted so as to ensure a heat-denaturation of about 35 %.
(b) Cloudy End Products, i.e. those from sera in which the heat-denatura- tion was of the order of 25 % or less.
In each case the end product was diluted with normal saline so as to reduce the protein content of the liquid to the order of 5 %.
To a known volume of the diluted liquid was added a sufficient volume of a saturated solution of ammonium sulphate to bring the concentration of this
CONCENTRATION OF HEAT-DENATURATED SERA 51
salt in the mixture up to 30 % of saturation. If a precipitate were formed at this stage the liquids were filtered. To an aliquot part of the filtrate was added the volume of the saturated solution necessary to raise the concentration of the sulphate in the filtrate to 33 °4 of saturation. The liquids were again filtered and by a similar method of procedure the animonium sulphate content of the filtrates was successively raised to the order of 36, 40, 45 and 50 % of saturation.
The precipitates isolated at each of the successive stages were pressed and dialysed in the usual way. The protein content of the respective residues from dialysis was estimated with the aid of the Zeiss Refractometer; determinations were also made of the percentage of the total antitoxin units appearing in the respective end products.
Owing to the scarcity of experimental animals the research could not be carried out so extensively as originally planned. Nevertheless the results, so far as could be obtained, have provided sufficient data to elucidate the problem under discussion.
(a) The successive fractionation of the Clear End Products obtained from sera showing a heat-denaturation of about 35 %.
Antidiphtheritic plasma was concentrated by the Homer (1918) method, the heat-denaturation of the plasma being regulated by the preliminary adjust- ment of the reaction.
The end products, obtained from the dialvsis of the protein precipitated from the heated plasma between 30 and 45 % of saturation with ammonium sulphate, were clear and remained clear on dilution, for they were free from any opalescent suspension of heat-denaturated protein; subsequent tests showed that they were also free from “salt-insoluble” protein.
The end product, under consideration, contained 15,000 units of antitoxin per gram of protein. After the necessary dilution with saline it was precipi- tated in the manner described above; the results are embodied in Table I.
Table 1. The successive fractionation of the Clear End Products from the con- centration of antitoxic sera in which a heat-denaturation of 35%, had been induced.
(The end product taken contained 15,000 units of antitoxin per gram of protein.)
Percentages of satura-
tion with ammonium The residues from the dialysis of the respective fractions showed sulphate marking the —-——————_——_ A_____— — — ~ limits for the precipi- Percentage Percentage of Number of units of anti- tation of the various of the total the total anti- toxin associated with fractions Appearance proteins toxic units 1 g. of protein 0-30 a 0-0 0-0 _-
30-33 clear traces traces —
33-45 clear 80-0 73°5 13,800
33-40 clear 62-5 59-5 14,500
40-45 clear 18-5 16-5 13.800
45-50 clear 2-5 2:8 16,000
4—2
' © ~
A. HOMER
t
Krom a study of the table it will be seen that there was no appreciable precipitation of protein when the concentration of ammonium sulphate in the diluted end product was successively raised to 30 and to 33 °%, of saturation respectively,
Between 33 and 40 °
» of saturation with the sulphate there was precipi-
tated 62:5 % of the total pseudoglobulin of the end product, and with this protein there was associated 59-5 °%, of the total antitoxin.
The protein precipitated from the diluted end product between 33 and 45 % of saturation with ammonium sulphate comprised 80 % of the total protein and 73-5 % of the total antitoxic units. Further fractionation indicated that 62-5 % of the protein and 59-5 °% of the antitoxin were precipitated between 33 and 40 % of saturation with ammonium sulphate; 18-5 °% of the protein and 16:5 °% of the antitoxin were precipitated between 40 and 45% of saturation with the sulphate; 2-5 °% of the total protein and 2-8 % of the total antitoxin were precipitated between 45 and 50 °% of saturation with the sulphate.
Unfortunately owing to the shortage of experimental animals the fractiona- tion was not continued further.
An examination of these results shows that, within the limits of experi- mental error, the percentage precipitation of antitoxin in the respective sub- fractions is directly proportional to the percentage precipitation of protein. Moreover, it will be seen that, at each stage of the precipitation, the load of antitoxin per gram of protein in solution was practically the same as that exhibited in the original liquid taken for fractionation, viz. of the order of 15,000 units.
Similar results were obtained from the fractionation of other clear end products.
The data thus obtained have furnished sufficient evidence for the con- clusion that the antitoxin precipitated with its associated proteins from the heat-denaturated sera under discussion is proportionately distributed between the protein fractions successively precipitated at increasing concentrations of ammonium sulphate.
(b) The successive fractionation of the Cloudy End Products from sera showing a heat-denaturation of 25 % or less.
It has been demonstrated that the concentration of antitoxic sera which, prior to their being heated, contained an unsuitable percentage of phenol and its homologues or of which the reaction lay between P,, 5-5 and 7-4, led to the production of cloudy end products. The latter yield a lower degree of con- centration than that furnished by sera adjusted so as to ensure not only a more extensive denaturation but also better conditions for the precipitation of the heat-denaturated protein with the First Fraction Precipitates at 30% of saturation with ammonium sulphate.
CONCENTRATION OF HEAT-DENATURATED SERA 53
Some of these cloudy products were taken and fractionally precipitated in stages in the manner described above (p. 50).
The data with regard to the percentage precipitation of the total protein and of the total antitoxin with the fractions precipitated at the various stages have been included in Table II.
Table Il. The successive fractionation of the Cloudy End Products from the concentration of antitoxic plasma in which a heat-denaturation of 25 °/, had been induced.
(The end product under consideration contained 12,000 units of antitoxin per gram of protein in solution.)
Percentages of satura- tion with ammonium ‘The residues from the dialysis of the various protein fractions showed = A
sulphate marking the SSS = — limits for the precipita- Percentage Percentage Number of units of anti- tion of the respective of the total of the total toxin associated with fractions Appearance protein antitoxin 1 g. of protein 0-30 —- 0-0 0-0 -—
30-33 thick suspension 1G:O ss 6-0 4,550
30-36 markedly opalescent 48-8 35:0 8,600
33-36 clear 32-0 28-0 10,500
33-40 clear 57-0 58-0 12,300
40-45 clear 8-9 10-0 13,500
36-40 clear 24-5 30-0 14,500
In the case recorded, the addition of ammonium sulphate to the diluted end product to the extent of 30 °% of saturation did not cause a precipitation of protein.
Between 30 and 33 % of saturation there was precipitated 16 % of the total protein and with this was associated only 6 % of the total antitoxin. The fraction between 30 and 36 % of saturation contained 48-8 °% of the total protein and 35 % of the total antitoxin. The percentage of the total protein precipitated in both of these fractions is considerably greater than that of the antitoxin, the disparity being more marked in the former fraction.
An analysis of the respective results shows that, by difference, there was a direct proportionality between the percentage of protein and antitoxin in the fraction precipitated between 33 and 36 % of saturation with the sulphate. A separation of the protein precipitated at 33% of saturation with the sul- phate was made and the filtrate was sub-fractionated further, the details being given in Table II.
These results indicate that, after the preliminary separation of the protein precipitated at 33°% of saturation with the sulphate, the sub-fractions successively isolated show the same proportionality as regards the relative precipitation of antitoxin and protein as was found in the sera discussed in (1).
In another case, not recorded in the Table, in which the heat-denaturation was less than 20 %, the sub-fraction precipitated between 30 and 33 % of saturation with ammonium sulphate contained about 20 % of the total protein
D4 A. HOMER
and about 5 °% of the total antitoxin; that between 33 and 36 °%, of saturation
)
(, of the total protein and about 16 % of the
with the sulphate contained 36 ‘ total antitoxin. In such sera the sub-fractions isolated from the filtrates after the preliminary removal of the protein precipitated at 36%, of saturation showed the proportionality exhibited after the preliminary removal of protein precipitated at 33 °% in the previous case and at 30 %, of saturation in the sera discussed in (a).
A study of Table II also shows that in the First Sub-fraction (30-33 % of saturation) the load of antitoxin per gram of protein is considerably less than in the subsequent fractions and in the original solution. Moreover, as was to be expected, the sub-fractions subsequently precipitated after the preliminary removal of this First Sub-fraction contained a load of antitoxic units per gram of protein slightly greater than that shown by the original liquid. However, the load of antitoxin per gram of protein was in no case greater than it would have been had the reaction been suitably adjusted as in (a).
From these results it is obvious that in heat-denaturated sera showing a denaturation of 25 % the concentration of the end products and their degree of clarity can be increased by raising the precipitation limit for the First Fraction Precipitates to 33% of saturation with ammonium sulphate. In sera showing a lower degree of denaturation it is necessary to raise the limit to 36 °% of saturation in order to achieve the maximum advantage. In these cases the increased concentration of the sulphate ensures the precipitation and separation of heat-denaturated globulin to which a small portion only of the antitoxin is attached: this protein would have been precipitated at 30 % of saturation in sera suitably adjusted as in (a).
After the preliminary removal of the denaturated globulin to which a com- paratively low percentage of the antitoxin is attached, the remainder of the antitoxin was found to be evenly distributed throughout the successive sub- fractions precipitated by successively raising the concentration of ammonium sulphate.
It was also observed that, in the unadjusted sera under discussion, the protein fractions respectively precipitated between 30 and 33, and between 30 and 36 % of saturation with the sulphate did not dialyse completely. The increased concentration of sulphions had induced an apparent irreversible precipitation of heat-denaturated protein. The precipitate thus formed was only partially redissolved by long saturation in about ten times its bulk of a saturated solution of salt.
A study of the data given in Tables I and II shows that, while from the whole serum the proteins associated with the antitoxin were precipitated between 30 and 45 % of saturation with ammonium sulphate, the precipitation of the refined products within the same limits yielded only 80 % of the pro- tein-antitoxin combination. It has been my experience that the percentage addition of ammonium sulphate necessary for the complete precipitation of a particular fraction of an individual protein is influenced by the concentration
CONCENTRATION OF HEAT-DENATURATED SERA 55
of the latter and by the presence of other proteins in solution: it is also affected by the degree of purity of the protein and its rarity in solution. The relationships hereby involved are being more fully investigated.
SUMMARY.
1. For the complete recovery of antitoxin during the concentration of sera, showing a heat-denaturation of 35% or less, by fractional methods employing the use of ammonium sulphate it is advisable to precipitate the Second Fraction between 30 and 45 % of saturation with the sulphate.
If the upper limit be reduced there will be incomplete precipitation of pseudoglobulin and antitoxin and a certain percentage of the latter will be discarded with the “albumin” filtrates.
If the lower limit be raised then antitoxin will be precipitated with the First Fraction Precipitates in a form not readily soluble in brine, and, there- fore, to all intents and purposes lost.
2. In the precipitation of the Second Fraction from sera in which a denaturation of 25 % or less has been induced, the raising of the lower limit to 33 or to 36 % of saturation leads to the production of clearer and more concentrated end products than those obtained by the adoption of the lower limit of 30 % of saturation. ;
However, for the reasons given in (1) the benefit thus gained is minimised by the greater losses of antitoxin thereby incurred.
Moreover, the degree of concentration is not increased beyond that which would have been obtained had the serum been suitably adjusted and subse- quently treated as in (1).
3. In heated sera showing a denaturation of 35 % or less the bulk of the antitoxin is associated with the proteins precipitated between 36 and 45 % of saturation with ammonium sulphate.
4. The further fractionation of the protein isolated between the limits indicated in (3) showed that the percentage of the total antitoxin precipitated between progressively increasing percentages of saturation with the sulphate is directly proportional to the percentage precipitation of protein at the respective stages.
5. From these observations it is clear that, in order to isolate antitoxin, means other than the fractional precipitation of the serum proteins must be employed.
REFERENCES.
Homer (1917, 1). J. Hygiene, 15, 580. (1917, 2). Biochem..J. 11, 292.
— (1918, 1). Biochem. J. 12, 190. —— (1918, 2). J. Hygiene, 17, 51.
VII. ON THE INCREASED PRECIPITABILITY OF PSEUDOGLOBULIN AND OF ITS ASSOCIATED ANTITOXIN FROM HEAT-DENATURATED SO- LUTIONS.
By ANNIE HOMER. From the Lister Institute of Preventive Medicine. (Received February 18th, 1919.)
Some time ago an investigation was undertaken in order to furnish data as regards the increased precipitability of pseudoglobulin relatively to that of its associated antitoxin from solutions of which, prior to their being heated, the reaction had been adjusted to values between Pj, 4:5 and 9-5.
In view of the need for the immediate application of the results to routine work, the data obtained with regard to the precipitability of the protein from solutions of which the reaction had been adjusted to about Py 8-0 were included in a previous paper |Homer, 1917, 1]; the data furnished from the wider aspect of the problem form the subject matter of the present communi- cation.
The study of the changes in the precipitability of the pseudoglobulin pre- sented no difficulty, but, unfortunately, the scarcity of experimental animals has considerably limited the scope of the investigation as regards the fate of the antitoxin.
(a) THE INCREASED PRECIPITABILITY OF PSEUDOGLOBULIN FROM ITS HEAT-DENATURATED SOLUTION BY AMMONIUM SULPHATE.
The solution of pseudoglobulin and antitoxin used in the experiment. was prepared as follows from unheated oxalated antidiphtheritic plasma pooled from the bleedings from several horses.
To 4 litres of plasma was added an equal volume of a saturated solution of ammonium sulphate; the mixture was allowed to remain at room temperature for 4 hours before being filtered. The precipitate, after being well drained, was thrown into 8 litres of a half saturated solution of ammonium sulphate, care being taken both to break up the lumps of precipitate and to keep the mixture stirred from time to time. After two days of this treatment the mixture was filtered. The well drained precipitate was gently pressed between cloths and boards to express as much of the adhering fluid as possible; it was then dissolved in 4 litres of water and the solution was submitted to a repetition of the above described process.
PRECIPITABILITY OF PSEUDOGLOBULIN AND ANTITOXIN 57
To the solution of globulins finally obtained were added 4 litres of a satur- ated solution of sodium chloride together with sufficient solid salt to ensure complete saturation. The brined liquid was allowed to remain at room tem- perature for seven days and was then filtered. To the filtrate, containing the “salt-soluble” globulin in solution, was added 0-25 % of glacial acetic acid; the ensuing precipitate of pseudoglobulin was filtered, pressed and dialysed in the usual way. The residue from dialysis was diluted so as to reduce the protein content to the same order as that of the original plasma.
To twelve separate volumes of the experimental liquid were added varying amounts of acetic acid or of ammonia to adjust the reaction to values between P,, 4:5.and 9-5; the [H’] of each of the liquids was measured by the electrical method. The stoppered bottles containing the adjusted liquids were placed in a water bath at 57-5° and were kept at this temperature for a period of six hours.
Samples of the unheated solutions were made respectively 26, 30, 33, 36 38, 40, 42-5, 44-5 and 47 % of saturation with ammonium sulphate and were then filtered. The protein content of the filtrates was measured by means of the Zeiss Immersion Refractometer, and the data thus obtained were com- pared with those derived from similar determinations of the heated liquids.
Table I. Showing the influence of heat-denaturation on the precipitability of pseudoglobulin from its solutions at increasing concentrations of ammonium sulphate.
The solutions which contained 7-48 % of pseudoglobulin were heated at 57-5° for 6 hours.
Residual percentage of pseudoglobulin in the filtrates from the
Experi- Pu of the heated liquids respectively brought to the following degrees of mental liquids prior to saturation with ammonium sulphate liquid their being a Bos _ — No. heated 26 30 33 36 38 40 42-5 44:5 47 0 control unheated 740 7:33 5°16 2-48 1-88 1-48 1:04 0-40 0-20 1 9-5 3-47, 93:08) 1-37 0:60. 0-20) 0-00, 0,00) | 0:00" 0-00 2 9-3 4-38 3-61 1-74 0:64 0:34 0-14 0:00 0:00 0-00 3 8-7 704 4-31 294 1-40 0-53 036 0-00 0:00 0:00 4 8-3 736 468 2:58 1:64 1-26 105 0-21 ,_ 000 0:00 5 7:5 748 5:33 3°24 1:54 I-50 -— 0:15 0:02 0-00 6 6-4 (elon 10-0408) 9-DOln 24 ele On O21 10:02) 0-00 dl 59 Tce 5:46 3°18 2-28 1-26 0-80 0-20 0-00 0-00 8 5-1 5-05 4:62 2:95 1:79 095 0-75 O21 0-00 0-00 9 4-9 336 2:77 1-87 0:18 | 0:20. 0:00. ©6-00. 0:00 0-00 10 4-65 185 185 162 000 0-00 0-00 0:00 0:00 0:00 1] 4-5 1-05 1-05 0-46 0-00 0-00 0-00 0-00 0-00 0-00
The results embodied in Table I indicate the percentages of pseudoglobulin remaining in solution in the filtrates from the respective pseudoglobulin- ammonium-sulphate mixtures. From these data was calculated the per- centage precipitation of pseudoglobulin at the respective stages: the latter results have been depicted in Table IT.
58 A. HOMER
Table IL. Showing the relation between the extent of the heat-denaturation of
pseudoglobulin and its increased precipitability by ammonium sulphate.
The solutions which contained 7:48 °% of pseudoglobulin were heated at 57-5° for 6 hours. o hs
Pru of the Percentage Percentage of the total pseudoglobulin precipitated Experi- liquids denaturation from the solutions by the addition of ammonium sulphate mental — prior to of the to the following degrees of saturation liquid their being — pseudo- ; 4 = No. heated elobulin 26 30 33 36 38 40 42-5 44:5 47 0 control unheated 00 16 30:1 67:0 75:7 80:0 86:0 94:7 97:3 l 9-5 53-5 53-4 58-8 82-4 92:0 97:3 100-0 100-0 100-0 100-0 2 9-3 47°3 41-2 51-5 76:4 92-5 95-7 98-1 100-0 100-0 100-0 3 8-7 38:1 55 42-1 71:3 80-1 92-9 95-9 100-0 100-0 100-0 1 8:3 34-0 10 37-6 62-7 78:0 84:0 86:0 97:3 100-0 100-0 5 io 25-0 0:0 27:5 56:2 79:3 80-0 --- 98-0 100-0 100-0 6 6-4 21-8 3:0 243 51:0 67:0 840 90-0 98-0 99-0 100-0 7 5:9 22-7 3:1 26-4 55-6 69-4 84:2 89:0 97-5 99-0 100-0 8 5-1 33°3 32:2 38:0 60:5 76:0 87-2 89-5 97-5 100-0 100-0 9 4-9 60-0 55:0 63-0 75-7 98:0 98-0 100-0 100-0 100-0 100-0 10 4-65 70-0 75:5 75:5 78-0 100-0 100-0 100-0 100-0 100-0 100-0 ll 4-5 83-0 86:0 86:0 94-0 100-0 100-0 100-0 100-0 100-0 100-0
From a study of the tables it will be seen that, in the unheated liquid, the precipitation of pseudoglobulin begins at about 28 % and is not quite com- pleted at 47 % of saturation with the sulphate.
In the heated liquids there was complete precipitation of the protein at a much lower concentration of sulphate than that required in the unheated liquid. Thus, where a heat-denaturation of from 20 to 30 % had been induced, the precipitation of protein was complete at about 44 %, of saturation with the sulphate; where the denaturation was of the order of 50 %, there was complete precipitation of the protein between 38 and 40% of saturation with the sulphate; as the denaturation was further increased, so the concentration of ammonium sulphate necessary for the complete precipitation of the pseudo- globulin diminished.
It will also be seen that the precipitability of the protein, at each degree of saturation with the sulphate, was considerably greater in the heated than in the unheated liquids, the increase being most pronounced where the heat- denaturation was greatest. :
The latter point is more clearly brought out by the plotting of the results given in Table II in the form of curves. For the sake of clearness it was necessary to present the curves in two figures. In Fig. 1 have been included the curves representing the precipitation of protein by ammonium sulphate from the unheated liquid and from the heated acid liquids of which the reaction lay between the values P,, 7-4 and 4-5: the curves in Fig. 2 represent the precipitation of protein from the heated alkaline liquids of which the re- action lay between Py, 7-1 and 9-5.
From the relative positions of corresponding points on the curves it is evident that. at the respective concentrations of ammonium sulphate, the
PRECIPITABILITY OF PSEUDOGLOBULIN AND ANTITOXIN 59
precipitation of protein is greater from the heated than from the unheated liquids.
To take a case in point: in the unheated solution 40 °% of protein was con- verted from the emulsoid to the suspensoid condition at 34 % of saturation with ammonium sulphate: the same percentage conversion was obtained in the heated liquids at lower concentrations with ammonium sulphate. Thus, in the heated liquids Nos. 1, 2, 3, 4, 5, 6, 7 and 8 in which the denaturation had been of the order of 53-5, 47-3, 38-1, 34, 25, 21-8, 22-7 and 33-3 % the above precipitation of protein was respectively achieved at 25, 28-8, 30-5, 31, 31-5, 31-5, 31 and 30 % of saturation with the sulphate.
> emt / 37 s Gp es Y iS) 90 ay if 7 : os S Curve Tl Py 4 <a Xe e . 2 Sy —ex — eX —* / . Res 2 + ye aia es aaa rae 2 80 Py 4°65 # aS = Curve 10. Px L— 4 07. —_—A- ree . Cie ye Va = 70 bs fal Ss & = ae Lo > 60 qe: Pia, = eee ee 2 50 7 oe 3 Zee g 40 of, rel 5 \ : . S gure em a/ + ~~ _? = i=" 80 coe = ® af Ss A, 12 8 a 20 aes 2 st S Se & a ae Hi < ® 4. 0 7S 5 10 git Ay AO
Gqmea Ie eRe) BOO) ) 6Ss4i 86. 198. Qa aM ue} 48 Percentages of saturation with ammonium sulphate. Fig. 1. Curves representing the percentage precipitation of pseudoglobulin from unheated and from heated solutions by increasing concentrations of ammonium sulphate. Curve 0 for the unheated liquid; Curves 6, 7, 8, 9, 10 and 11 for the heated liquids of which the reaction, prior to their being heated, had been adjusted to Py 6-4; 5-9; 5-1; 4:9; 4-65 and 4-5 respectively.
It will be seen that the group of curves 0, 3, 4, 5, 6 and 7, representing the precipitation of protein from those liquids in which a heat-denaturation of 40 % or less had been induced, is marked by a similarity of configuration throughout the whole of the range investigated; the shift in the relative positions of corresponding points on the curves is constant and is a measure of the increased precipitation induced by the heat-denaturation of the protein.
On the other hand, while there is a marked similarity between the con-
60 A. HOMER
figuration of the individual curves in the group 1, 2, 8, 9, 10 and 11 the com- posite type of curve differs from that of the former group. The shift in the relative positions of corresponding points is not constant, the magnitude of the increased precipitation being relatively greater for the lower concentra- tions with the sulphate, a phenomenon which is accompanied by an increasing ‘salt-insolubility”’ of the denaturated protein precipitated at stages in the flat portion of the curves.
Percentage precipitation of pseudoglobulin from its solution.
26 28 30 8632 34 36 38 40 42 44 46 48 Percentages of saturation with ammonium sulphate.
Fig. 2. Curves representing the percentage precipitation of pseudoglobulin from unheated and from heated solutions by increasing concentrations of ammonium sulphate. Curve 0 for the unheated solution. Curves 1, 2, 3, 4 and 5 for the heated solutions of which, prior to their being heated, the reaction had been adjusted to Py 9-5; 9:3; 8-7; 8-3 and 7-5 respectively. ;
(b) THE PRECIPITATION OF ANTITOXIN WITH HEAT-DENATURATED PSEUDOGLOBULIN.
It had been my intention to ascertain the limits of saturation with am- monium sulphate between which the antitoxin was precipitated from each of the twelve solutions dealt with in (a). But, owing to the long continued scarcity of experimental animals, it was impossible to carry out the scheme on such an extensive and comprehensive scale, and, therefore, the scope of the work has been restricted to a determination of the extent of the association of antitoxin with the protein precipitated at a conveniently chosen concen- tration with ammonium sulphate, viz. at 30 % of saturation with the salt.
PRECIPITABILITY OF PSEUDOGLOBULIN AND ANTITOXIN 61
The method of procedure was as follows. To each of the twelve solutions dealt with in (a) was added the necessary volume of a saturated solution of ammonium sulphate to bring the concentration of the latter in the mixture up to 30% of saturation. The mixture was filtered; the protein and the anti- toxin content of the filtrates were estimated in the usual way; the values obtained were compared with those for the experimental solution. From these data was calculated the percentage of the total antitoxin precipitated with the protein from the respective solutions.
The data thus furnished were incorporated in Table III and in the curve in Fig. 3. They show that while in the unheated liquid the precipitation of antitoxin at 30 % of saturation with the sulphate is negligible, in the heated liquids the proportion precipitated at this stage is considerable and increases with the extent of the denaturation.
Percentage precipitation of antitoxin with the pseudoglobulin.
20 30 40 50 60 70 80 90 Percentage precipitation of the pseudoglobulin. Fig. 3.
Curve representing the relative precipitations of pseudoglobulin and antitoxin at 30% of saturation with ammonium sulphate from solutions of which, prior to their being heated, the reaction had been adjusted to values between Py 4-5 and 9-5.
G2 A. HOMER
Table ILL. Showing the relation between the extent of the heat-denaturation of pseudoglobulin and the precipitation of antitoxin with the denaturated
protein by 30%, of saturation with ammonium sulphate.
The solution of pseudoglobulin showed a potency of 450 units per ce.
Percentage of the Percentage of the total
total pseudoglo- — antitoxic units precipi- Percentage de- bulin precipitated tated with the pseudo- Ex peri- Py of the liquid naturation of — at 30 % of satura- clobulin at 30 % of mental prior to its the pseudo- tion with ammon- saturation with liquid No. being heated globulin ium sulphate ammonium sulphate 0 control 7-1 unheated 1-6 -- | 9-5 53°5 58°38 50 2 93 47°3 51-5 40 3 8-7 38:1 42-1 30 4 8:3 34-0 37-6 20 5 75 25-0 27:5 20 6 6-4 21:8 24-3 12 7 5-9 22-7 26-4 15 8 5-1 33:3 38-0 20 9 4-9 60-0 63-0 Hs) 10 4-65 70-0 75:5 > 66< 80 11 4-55 83-0 86-0 >80<90 12 4-4 too solid to measure > 90
The precipitation of antitoxin was least from those liquids showing the least extensive denaturation (Nos. 4, 5, 6, 7 and 8), and in these cases the pro- portion of the total protein precipitated by the sulphate was not a linear measure of the proportion of the total antitoxin associated with the precipi- tate. In the more acid and in the more alkaline liquids, there was a marked increase in the precipitation of protein and of antitoxin, but it was found that the increased precipitation of protein beyond that shown in Nos. 4, 5, 6, 7 and 8 was a measure of the accompanying increased precipitation of antitoxin at this stage.
It seems justifiable to conclude that in the unheated liquid there is a com- paratively small load of antitoxin attached to the fraction of pseudoglobulin which is precipitated at about 34 to 35% of saturation with ammonium sulphate, that is, with the pseudoglobulin which, in unheated sera, would be precipitated at concentrations of ammonium sulphate slightly greater than that required for the precipitation of the euglobulin. A comparatively low degree of heat-denaturation suffices to ensure the precipitation of this fraction of the pseudoglobulin at 30% of saturation with the sulphate. A more extensive denaturation leads to the precipitation at this stage of fractions of the pseudoglobulin which, in the unheated hquids, would require con- siderably greater concentrations of the sulphate. The antitoxin is evenly distributed throughout the range of the latter fractions, a deduction readily made from the study of the curve in Fig. 3 taken in conjunction with the results obtained in a recent investigation of the further fractionation of the proteins of heat-denaturated sera [Homer, 1919].
PRECIPITABILITY OF PSEUDOGLOBULIN AND ANTITOXIN 63
The data thus obtained also furnish additional evidence that, in the fractional precipitation of antitoxin and pseudoglobulin by ammonium sul- phate, an increase in the extent of the denaturation, up to a point, favours the isolation of the antitoxin with a correspondingly decreased percentage of the original protein. But beyond this previously described limit |Homer, 1917, 2,3; 1919] a more extensive denaturation ceases to be advantageous, for, with the further increased removal of protein there is a correspondingly in- creased proportional precipitation of antitoxin; moreover, much of the anti- toxin thus precipitated is lost, as the protein with which it is associated has been converted into a “salt-insoluble” condition.
The consideration of the above results taken in conjunction with those from my previous investigations leads me to the conclusion that, in order to isolate the bulk of the antitoxin from antitoxic sera in association with a minimum percentage of the total serum proteins, there is no need for the pre- liminary prolonged heating of the serum originally advocated by Banzhaf and Gibson [1907] and now generally adopted by those engaged in the con- centration of sera. For while I have shown that, if the preliminary heating be conducted to the best advantage, the antitoxin can be isolated from heated sera in association with a minimum amount of protein in the fraction pre- cipitated between 30 and 44 % of saturation with ammonium sulphate, my more recent work demonstrates that the same results would have been obtained by the isolation of the protein fraction precipitated from the unheated serum between (czrc.) 36 and 50 % of saturation with the sulphate.
The concentration of antitoxic sera by the fractional precipitation of the unheated serum by ammonium sulphate, as indicated above, is obviously a shorter process than those in which a preliminary heating of the serum is adopted. Furthermore, the fractional precipitation of the unheated plasma or serum between the higher limits for the concentration of ammonium sulphate is not attended with the filtration difficulties so often experienced with heated sera fractionated between the lower limits, for, in the former case not only is the complete separation of the euglobulin with the First Fraction assured, but the possibility of trouble arising from the incomplete agglutination of particles of heat-denaturated protein is avoided.
The above considerations seem to indicate that from the point of view of the isolation of antitoxin and its associated protein, there is no practical advantage to be gained by the preliminary heat-denaturation of the serum.
There is, however, some evidence to show that, from a clinical point of view, the heating of the serum is beneficial. For, it has been found that the sub- cutaneous injection into mice and guinea-pigs of a given amount of cresylic acid causes more pronounced toxic symptoms when administered in solution in presence of unheated serum proteins than when in the presence of serum proteins previously heated to 57-5° for four hours.
My observations with regard to the fractional precipitation and concen- tration of unheated sera and to the lessened toxicity of cresylic acid in presence of heat-denaturated proteins will be dealt with in later communications.
64 A. HOMER
SUMMARY.
|. The increased precipitability of pseudoglobulin from its heat-de- naturated solutions at concentrations of ammonium sulphate ranging from 26 to 47 &% of saturation is a function of the heat-denaturation.
2. The increased precipitation of pseudoglobulin thus induced in (1) at 30% of saturation with ammonium sulphate is accompanied by an increased precipitation of antitoxin. With the least extensive denaturation, the per- centage of the total proteins precipitated at this stage is greater than that of the antitoxin; as the denaturation increases so the further increased pre- cipitability of the protein becomes a linear measure of the increased precipi- tation of the antitoxin.
3. Inthe concentration of antitoxic sera by the fractional precipitation of the serum with ammonium sulphate there is no need for a preliminary pro- longed heating of the serum. The results that are now obtained by the isolation from the heated serum of the protein fraction precipitated between 30 and 44 © of saturation with ammonium sulphate could be obtained from the unheated serum between 36 and 50% of saturation with the sulphate.
4. The heating of the serum reduces the toxicity of the cresylic acid- protein complex.
5. The data furnished in this research also point to the conclusion that, in order to isolate antitoxin as a separate entity, means other than the frac- tional precipitation of pseudoglobulin solutions by salts must be employed.
REFERENCES. Banzhaf and Gibson (1907). Collected Studies Research Lab. Dep Health, New York, 8, 97. Homer (1917, 1). Biochem. J. 11, 292. — (1917, 2). J. Hygiene, 15, 580. —— (1917, 3). Biochem. J. 11, 21. —— (1919). Biochem. J. 18, 45.
VIII. THE RELATION OF SUGAR EXCRETION TO DIET IN: GLYCOSURIA.
By JOHN MELLANBY anp CHARLES R. BOX. From the Physiological Laboratory, St Thomas's Hospital, London.
(Received February 15th, 1919.)
THE modern treatment of diabetes is founded on the experimental work of Allen [1915] on the effects of starvation on the excretion of sugar in glycosuria. The observation that the urine of many people suffering from this disease may be freed from sugar by starvation forms the basis of the experimental work detailed in the following pages.
It is clear that if the urine of a man suffering from glycosuria is rendered sugar free by starvation, then, the effect of any diet on his sugar excretion may be determined. The accuracy of the results is limited by the fact that it is not desirable to keep a man in a constant state of starvation. But, by a careful choice of the time at which a desired food is given, and by making a graphic time record of the sugar subsequently excreted, a considerable amount of information may be obtained on the effect of that foodstuff on the excretion of sugar. From an analysis of these results, considered in conjunction with other observations in this disease, an hypothesis has been advanced on the causation of glycosuria in diabetes mellitus.
The main results recorded were obtained from a man, aet. 53, suffering from diabetes mellitus. When he came under observation he was excreting, per diem, about 250 g. of dextrose contained in four litres of urine.
The sugar determinations were made by a method originally described by Wood and Berry [1903]. In this method 25 cc. of a copper solution (suggested by Soldaini [1876]) were boiled for 15 minutes with | ce. of urine, and the precipitated cuprous oxide was filtered off by asbestos felt contained in a Gooch crucible. The cuprous oxide was dissolved in 25 ce. of a 2-5 % solution of ferric sulphate contained in 25 % H,SO,, and the equivalent amount of ferrous sulphate formed was estimated by a standard solution of perman- ganate. The strength of the permanganate solution was such that 1g. of dextrose was equivalent to 264 cc.. KMnQ,.
The man’s diet was carefully weighed and controlled during the experi- mental period. Whenever he emptied his bladder the amount of urine excretéd and the time of excretion were noted. The quantity of dextrose contained in l ce. of urine was determined. The man was encouraged to micturate at frequent intervals so that a time curve showing the rate of sugar excretion could be obtained with a fair degree of accuracy.
Bioch, xm 5
66 J. MELLANBY AND CG. R. BOX
(a) Starvation. When the man first came under observation a three days’ fast was required to free his urine from sugar. The following detailed experi- ment was made three weeks later and shows, inter alia, that a considerable improvement had taken place in his condition. His last meal was taken at 6.30 p.m. Within 24 hours his urine was free from sugar. The table shows the effect of deprivation of all food on the average sugar and urine excretion per hour.
Amount ce, KMnO, Total Sugar Urine Time of urine for 1 ce, sugar per hour per hour :
ce. g. g. ce.
4.30 p.m. 253 16-4 15-7 — = 8 SS 224 12-6 10-7 3:6 64. Midnight 337 8-4 10-8 2-68 84 2.30 a.m. 280 9-9 10-5 4-2 112 8 5 224 79 6-7 1-22 40 9 2 224 7:95 6-73 6-73 224. 11 x 450 1-35 2-3 1-15 225 1.30 p.m. 476 0-85 1-54. 0-62 190 3.30 ,, 253 0-25 0-24 0-12 226 5.30 ,, 364 0-05 0-068 0-034 182 8 cf 503 0-05 0-096 0-038 201 1 a.m. 224 0-05 0-042 0-008 45
Sugar in grams per hour
6pm 8 10 12 Qam 4 6 8 10 12 QPm
The results are shown graphically in Fig. 1. The rapid disappearance of sugar from the urime in starvation is shown in a clear way. From midnight to noon he excreted 37 g., from noon to midnight 2 g. of dextrose. The dis- appearance of sugar from the urine when no food was taken afforded a means
SUGAR EXCRETION AND DIET IN GLYCOSURIA 67
whereby the effect of diet on the excretion of dextrose in the urine could be determined. Incidentally it may be remarked that the curve shows the effect of taking caffeine (contained in tea without milk or sugar) at 8a.m. The amount of urine was greatly increased, and, although the percentage of sugar in the urine remained constant, yet the total quantity of sugar excreted showed a prominent rise. Again, it is well known that the amount of urine excreted in glycosuria is roughly proportional to the amount of sugar contained in the urine. It may be observed that whereas this generalisation held true for the first 12 hours when the urine contained a large amount of dextrose, the parallelism ceased when the urine became sugar free.
(b) Extractives. The object of this experiment was to ascertain whether the absorption of any substance from the alimentary canal, considered inde- pendently of its energy value, caused an excretion of sugar in the urine. Since only one-fifth of the total energy contained in the extractives of meat is utilised by the body, the man was given 840 cc. of beef tea to drink when his urine was free from sugar. The following table shows the results observed:
Amount cc. KMnO, Total Sugar Urine Time of urine for 1 ce. sugar per hour per hour CO. g. g. Ce. 9-10.15 a.m. 56 1-6 0-3 0-15 25 4.30 p.m. 140 6-6 3°58 0-7 25 8 3 84 4-6 1-45 0-41 24
The beef tea was taken between 2.30 p.m. and 4.30 p.m. Only a small quantity of sugar was excreted, about equal to the carbohydrate extracted from the beef. No definite evidence was obtained in favour of the assumption that the extractives contained in beef may give origin to sugar, nor that the act of absorption from the alimentary canal may cause the excretion of sugar in the urine.
(c) Alcohol. It was of interest to determine the effect of a relatively simple substance of a high calorific value on the excretion of sugar. Alcohol was chosen for this purpose since it has obtained a reputation as a foodstuff of considerable value in the dietetic treatment of diabetes mellitus. 112 cc. of whisky containing 30 % of alcohol were given in two portions at 10 p.m. and 4 a.m. respectively. In each case water was added to make the volume of the diluted fluid 250 cc. The following figures show the excretion of sugar and urine during the experimental period.
Amount ce. KMnO, Total Sugar Urine Time of urine for 1 ce. sugar per hour per hour cet g. g. ce. 8-11 p.m. 56 3°5 0-74. 0-245 19 1 a.m. 112 3-0 1:3 0-65 37 Dees: 140 0-8 0-42 0-42 140 Oy. 56 0-9 0-20 0-05 14
The figures show that after the ingestion of alcohol the excretion of sugar , increased a small amount only—from 0-245 ¢. to 0-65 g. of sugar per hour. Since alcohol is metabolised to a considerable extent by the body—only 2 %
5—2
68 J. MELLANBY AND C. R. BOX
being excreted—it is evident that the tissues can utilise a simple substance containing carbon, hydrogen and oxygen only without a concomitant excretion of sugar in the urine. This fact is of some practical value since the calorific value of alcohol is seven and the available energy obtained from 112 ce. of whisky containing 30 % alcohol amounts to 235 K.
(d) Protein. The effect of protein on the excretion of sugar was determined by giving a meal of 112 ¢. of caseinogen at 4.30 p.m. The urine was not free from sugar before giving the caseinogen since, four hours previously, the man had eaten a small quantity of beef and green vegetables. However, the com- parative results, and particularly the graphic record of these results, clearly show the effect of caseinogen on the excretion of sugar.
Amount ec. KMnO, Total Sugar Urine Time of urine for 1 ce. sugar per hour per hour 10.30 a.m. to ce. g. g. ce. 2.30 p.m. 196 2-1 1-56 0-39 49 AO OlEss 280 36 3:8 1-9 140 8 5; 168 4-5 3°12 0-89 45 10 of 336 4-9 6-25 3-12 168 5 a.m. 364 0-4 0-56 0-08 52
The results are shown graphically in Fig. 2. It is evident from the figure that the excretion of sugar between 8 p.m. and 10 p.m. was due to the ingested caseinogen. On this assumption the metabolism of 112 g. of caseinogen caused the excretion of 6 g. of dextrose in the urine. The caseinogen was sugar-free so that the dextrose must have originated from the amino-acids of the protein. It is of interest to compare this result with that obtained by Janney [1915] on a phlorhizinised dog. In this case 48 °, of the ingested protein appeared in the urine as dextrose. Since it is assumed that a dog treated with this drug is incapable of utilising dextrose, the deduction is made that caseinogen contains 48 °% of amino-acids which normally may give rise to dextrose. From these figures it may be calculated that the man was capable of utilising 89 % of the amino-acids of caseinogen which were available for conversion into dextrose.
(e) Carbohydrate. Whatever the carbohydrate eaten—starch, cane sugar, lactose or laevulose—dextrose only was excreted in the urine. The effect of carbohydrate on the excretion of sugar was determined in detail. Carbohy- drate in three forms was given: (i) laevulose, (ii) cane sugar, (11) oatmeal (starch).
(i) Laevulose. The view is often expressed that laevulose is utilised to a considerable degree by people suffering from glycosuria. The results recorded show that laevulose is utilised to a less degree than either cane sugar or starch.
56 g. of laevulose dissolved in 224 cc. of water were taken at 8 p.m. The comparative results depicted in Fig. 3 show the effect on the sugar excretion of this ingested laevulose.
SUGAR EXCRETION AND DIET IN GLYCOSURIA 69
=.
Le) [e2) eae
Sugar in grams per hour
_
112 Grams of Caseinogen 10am 12 Qpm 4 6 8 10 12 Gam 4
a ee
iS
[ee]
Sugar in grams per hour (o>) Si a Ss
2 «» |56 Grams of . Me ee
Laevulose
70 J. MELLANBY AND C. R. BOX
Amount ec, KMnO, Total Sugar Urine Time of urine for 1 ce. sugar per hour per hour
ec, g. g. ec,
3 to 5 p.m. 140 77 4-04 2:02 70 S . 56 9-2 1-93 0-64 19 S50.» ay 168 9-2 5:85 11:7 336 10 - 280 15-1 16-1 10-7 187 1.30 a.m. 280 12-0 12:7 3-65 80 3 * 70 10-5 2:8 1-86 46 6 - 112 5-0 2:1] 0-7 Bi,
The curve shows that the ingested laevulose caused a very rapid rise in the amount of sugar excreted in the urine. Within 30 minutes the rate of sugar excretion increased from 0-64 g. to 11-7 ¢. per hour. The results also show that the greater part of the effect passed off in two hours since after that time the average rate of sugar excretion fell from 10-7 g. to 3-65 g. per hour. The figures allow an approximate calculation to be made of the amount of sugar excreted after eating 56 ¢. of laevulose. The average sugar excretion before eating the laevulose at 8 p.m. was 0-64 &. per hour. It again fell to this level at 3a.m. It may therefore be concluded that the sugar excreted, due to the ingested laevulose, was confined to the urine formed between 8 p.m. and 3a.m. Also it may be assumed that during this period 0-6 g. of sugar per hour was due to the metabolism of foodstuffs, other than laevulose, previously eaten. On these assumptions it may be calculated that 33 ¢. of sugar were excreted as the result of eating 56 ¢. of laevulose—or of the laevulose ingested 41 % was utilised whilst 59 °4 was excreted in the urine. Comparative colour tests showed that practically the whole of the reducing sugar excreted was dextrose. The urine passed at 8.30 p.m. and 10 p.m. gave a characteristic Selivanoff reaction. The colour produced under standard conditions when compared with the colour produced under similar conditions in normal urine to which known amounts of laevulose had been added indicated that only 0-75 g. of laevulose was excreted as such. Therefore the following conclusions may be summarised from these observations: (a) nearly 40% of the ingested laevulose was excreted as dextrose, and (b) within two hours two-thirds of the excreted sugar had been so transformed and excreted.
(11) Cane sugar. In order to determine the capacity of the man to utilise this form of carbohydrate, 112 ¢. of cane sugar were dissolved in 450 cc. of water. The syrup was taken at 7.15 p.m. The following figures show the amount of sugar excreted during the experimental period:
Amount ce. KMnO, Total Sugar Urine Time of urine for 1 ce. sugar per hour per hour 11.15 a.m. to ce. g. g. ce:
4.30 p.m. 140 6-6 3°52 0-67 26-5 8 a5 84 4-6 1-52 0-435 24-0 ] a.m. * 392 17-4 25°8 5:2 78-0 4 95 140 21-3 11-4 3°8 47-0 U 3 56 4-0 0-85 0-28 8-0
SUGAR EXCRETION AND DIET IN GLYCOSURIA 7g
These results, shown graphically in Fig. 4, illustrate in a clear manner the effect of eating 112 g. of cane sugar on the amount of sugar excreted in the urine. At 8 p.m., immediately after eating the cane sugar, the average amount of sugar per hour in the urine was 0-435 g.; up to 4.a.m. there was a large rise in the excretion, and after that time the average per hour fell to 0-28 g. It may be concluded that the sugar excreted at 1 a.m. and 4 a.m. was mainly derived from the ingested cane sugar. On this assumption, and allowing 0-3 ¢. of sugar per hour as derived from some source other than the cane sugar, it may be calculated that the ingestion of 112 g. of cane sugar resulted in the
(Si)
p
je)
Sugar in grams per hour
i)
1
Foe h 12 Grams of Cane Sugar
4em 6 8 10 12 Qam 4 6 8 Fig. 4.
output of 36-5 2g. of sugar in the urine. There was no cane sugar contained in the urine, since its reducing power was unchanged after hydrolysing with | % H,SO, at 120° for 30 minutes. Also there was only a trace of laevulose in the urine—less than 0-1 g. Hence we may conclude that the 36-5 g. of reducing sugar contained in the urine consisted of dextrose only. Now 112g. of cane sugar when hydrolysed yield 124g. of invert sugar. Therefore after the absorption of 124 g. of invert sugar from the alimentary canal 87-5 g. were utilised by the tissues whilst 36-5 @. were excreted as dextrose. Despite the fact that the man excreted sugar not only on a carbohydrate diet but also on a protein diet, his tissues possessed a capacity to utilise carbohydrate derived from cane sugar to the extent of 71 %, only 29 % being excreted as dextrose in the urine.
72 J. MELLANBY AND C. R. BOX
(iii) Oatmeal. This foodstuff has been advocated by Van Noorden as a suitable diet for diabetics. The capacity of the man to utilise carbohydrate when presented in this form was therefore determined. On the first occasion the man ate 28 g. of oatmeal at midday. The following figures show the effect of this food on his excretion of sugar.
Amount ec. KMnO, Total Sugar Urine Time of urine for 1 ce. sugar per hour per hour 8.30 a.m. to ce. g. g. CC. 10.30 a.m. 84 2-05 0-65 0:33 42 3 p.m 84. 1:05 0-34 0-07 18 4 5 252 2-4 2-28 2-28 252 10 i 168 5D 3°5 0-87 42 2 a.m. 280 0-25 0-26 0-06 70
The results are shown graphically in Fig. 5. It may be seen from the diagram that the ingestion of 23 g. of oatmeal resulted in the excretion of 5:78 @. of dextrose. Now 23 e. of oatmeal contain 14-7 g. of starch, and this on hydrolysis yields 16-4 ¢. of dextrose. Therefore a diet containing carbo- hydrate equivalent to 16-4 @. of dextrose resulted in 10-6 g. being retained by the tissues and 5-8 @, being excreted in the urine. In other words 35 % of the available carbohydrate was excreted as dextrose whilst 65 °% was utilised by the tissues of the body.
Sugar in grams per hour
23 Grams
of Oatmeal
10am 12 Qpm 4 6 8 10 12 Qam Fig. 5.
A second experiment was made in which 184 g. of oatmeal were given in four equal portions, 7.e. 469. at 5a.m., 8.30a.m., 12.30 p.m. and 4.30 p.m.
SUGAR EXCRETION AND DIET IN GLYCOSURIA 73
The object of the experiment was to determine whether he could utilise carbo- hydrate to a greater or less extent when it was continuously supplied to the tissues of the body during the day, in contradistinction to the above experi- ment in which a small quantity of carbohydrate was given at one meal. The following results were obtained:
Amount ec. KMnO, Total Sugar Urine Time of urine for 1 ce. sugar per hour per hour
ce. g. g. ce.
1-5 a.m. 112 sel 3-9 0-97 28 9:30° 3; 168 14-0 8-9 2-0 37 Noon 168 13:3 8-5 3-4 67 3 p.m. 168 13-0 8-25 2-75 56 Be Ay 84 12-4 3-95 1-97 42 10 53 252 18-8 17-9 3-6 50 2.15 a.m. 112 9-5 4-03 0-95 26
The results are plotted in Fig. 6. The graphic record shows that the sugar excretion due to the oatmeal eaten at the four meals during the day extended from 9.30 a.m. to 10 p.m. During this time the amount of dextrose excreted (allowing | g. of sugar per hour as derived from sources other than the oatmeal, i.e. the quantity excreted per hour before and after the experimental period) amounted to 35-5¢g. The oatmeal eaten contained 118 g. of starch and this on hydrolysis would yield 130g. of dextrose. Therefore of the total carbo- hydrate (calculated as dextrose) available during the day approximately 94-5 o. were metabolised by the tissues of the body and 35-5 g. were excreted in the urine; or in other words, 73 °/, was metabolised and 27 % excreted.
3'5
3:0
ine) ol
a
Sugar in grams per hour
fe)
5 A BCD 46 Grams of Oatmeal | 4am 6 8. 10 12 Qpm 4 6 8 10 1.
74 J. MELLANBY AND C. R. BOX
K.XPERIMENTAL RESULTS.
The results detailed in the previous pages as to the relation of sugar excretion to diet in a man suffering from a moderate form of diabetes mellitus are summarised below:
(1) When first determined the daily output of sugar was about 250 g. contained in 4 litres of urine. Starvation freed the urine from sugar in 60 hours. Three weeks later the sugar excretion per diem was 120 ¢. contained in 3 litres of urine. A second period of starvation rendered the urine sugar-free in 24 hours.
(2) Whatever the carbohydrate eaten—starch, cane sugar, lactose or laevulose—dextrose only was excreted in the urine.
(3) Starch and cane sugar were utilised to the extent of 70 %; laevulose, on the other hand, was utilised only to the extent of 40 %.
(4) The capacity of the tissues to utilise sugar was not an absolute amount but was determined by the quantity of carbohydrate presented to them. Thus when 14:7 @. of starch were eaten 65 % was metabolised by the tissues and 35 °%% was excreted as dextrose. When 118 g. of starch were eaten 73 % was utilised by the tissues and 27 °% excreted as dextrose. In the latter case nine times as much carbohydrate was metabolised as in the former case.-
(5) Amino-acids, contained in caseinogen; available for conversion into dextrose, were utilised to a greater extent than starch. Thus only 6g. of dextrose were excreted in the urine after eating 112g. of caseinogen, 2.e. 89 °% of the amino-acids available for conversion into dextrose were meta- bolised.
(6) Within two hours of the ingestion of 56 g. of laevulose 30 e. of dextrose were excreted in the urine.
(7) The absorption of extractives of small energy value from the alimen- tary canal produced no increased excretion of sugar.
(8) Considerable quantities of alcohol were metabohsed without a con- comitant excretion of sugar in the urine.
The following additional facts have been observed in other cases of diabetes mellitus. :
(9) In diabetes mellitus there is a marked hyperglycaemia. The amount of dextrose in the blood may exceed 0-4 %. During starvation the amount of sugar in the blood diminishes and reaches a normal level when the excretion of sugar in the urine ceases.
(10) The liver of a child whose death was due to diabetes mellitus possessed as great an action on carbohydrate and fat as that of a child suddenly killed by accident.
SUGAR EXCRETION AND DIET IN GLYCOSURIA 75
DIscUSSION OF RESULTS—THE ABNORMAL FACTOR IN DIABETES MELLITUS.
Various hypotheses have been advanced as to the causation of this disease. Among such hypotheses may be mentioned those which assign the abnor- mality to (a) the alimentary canal, (b) the liver, (c) the muscles. The experi- mental results given in the previous pages offer some evidence in favour or against these hypotheses.
(a) The alimentary canal. The hypothesis that the essential cause of diabetes is an alimentary toxaemia received strong support from the results obtained by starvation on the excretion of sugar. But the observations that some cases of diabetes continue to excrete sugar even after prolonged periods of starvation, and that the absorption of water, alcohol or extractives from the alimentary canal does not increase the excretion of sugar, militate against this hypothesis.
(b) The liver. In many fatal cases of diabetes the liver is found enlarged. This fact, together with the knowledge of the part played by the liver in carbo- hydrate metabolism, forms the basis of the hypothesis that in diabetes mellitus the liver is at fault. The assumption that hypertrophy is due to an attempt on the part of the organism to compensate for a diminished functional activity receives no support from a consideration of the ferment activity of the organ. Within two hours of the ingestion of 56 g. of laevulose 30 g. of dextrose were excreted in the urine—a chemical change which may be assumed with some degree of certainty to have occurred in the liver. Further, preparations made from the liver of a child who had died from diabetes were no less active on various forms of carbohydrate and fat than similar preparations made from the liver of a child accidentally killed. Apart from the question of ferment activity, however, it might be assumed that the essential failure in the chain of carbohydrate metabolism in diabetes mellitus is the inability of the liver to form compounds, analogous to phosphatides in fat metabolism, for the trans- port of dextrose to the muscles. This assumption is negatived by the fact that in diabetes the body can utilise to an equal degree carbohydrates presented to it in large or small quantities. It is clear that any synthetic deficiency would be more marked the greater the quantity of substance to be elaborated.
(c) The muscles. Cohnheim’s [1903] hypothesis that there is a deficiency of pancreatic kinase required to activate the precursor of the glycolytic ferment in muscle—diabetes being due to a failure of the muscles to utilise dextrose, secondary to a disorder of the pancreas—was not supported by the experimental results of Patterson and Starling [1913] on the rate of utilisation of dextrose by the perfused hearts of depancreatised dogs. They demonstrated that pancreatic glycosuria in dogs does not result in the production of such a condition that cardiac muscle is unable to utilise dextrose presented to it in a perfusing fluid.
Again, there is no evidence that the muscles are unable to utilise dextrose in diabetes. Muscular weakness is never a prominent symptom except in advanced cases of the disease. Further, many observations have been made
76 J. MELLANBY AND C, R. BOX
showing that even with pronounced glycosuria no acetone or allied bodies may be present in the urine. Now an early sign of carbohydrate deficiency in the body is the exeretion of B-hydroxybutyrie acid and acetone. Therefore, in these cases, it is clear that the tissues are able to elaborate all the carbo- hydrate they require for their metabolic activities. In fact, if the quantity of dextrose present in the blood is such that the muscle cells can elaborate an amount of energy complex commensurate with their requirements, then the concomitant excretion of sugar by the kidneys does not involve any evidence of muscular weakness or the excretion of acetone bodies in the urine. The fundamental fact observed in the experiments recorded is that the capacity of the tissues to utilise carbohydrate is not an absolute amount but is deter- mined by the quantity of carbohydrate presented to them. Now an established law of ferment action is that the quantity of substrate changed by a ferment is proportional to its concentration. We must therefore consider the carbo- hydrate ferments of the muscle cells as the weak link in the chain of carbo- hydrate metabolism. The following hypothesis is put forward to elucidate the phenomena observed.
The muscles in diabeles mellitus are able, to a limited extent only, to utilise dextrose in the concentration at which it exists in normal blood, the degree of limitation being proportionate to the severity of the disorder. There is therefore a call for a higher concentration demanding more sugar from the sources of supply. Thus the percentage of sugar accumulates in the blood until it reaches a level at which it is excreted by the kidneys. On this hypothesis the accumulation of sugar in the blood and the resultant glycosuria are secondary to changes in the synthetic activity of the muscle ferments. The muscles are able to utilise dextrose but, compared with normal muscle, they are unable to synthesise the inogen complex at a rate commensurate with their expenditure of energy from dextrose present in the blood at a concen- tration below that at which it is excreted by the kidneys. Hyperglycaemia in diabetes is a compensatory mechanism. The resultant glycosuria is a secon- dary effect depending on the fact that so far as the tissues of the body generally are concerned the activities of the kidneys are so regulated that dextrose present in the blood in a percentage greater than a certain amount is excreted in the urine.
This hypothesis demands as a corollary that the muscles are the active agents which mobilise the carbohydrate stores of the body. We propose to consider the problem, whether the active agent is a chemical hormone elabor- ated by the muscles or whether the central nervous system is involved in the metabolic chain, in a subsequent communication.
‘
REFERENCES.
Allen (1915). Amer. J. Med. Sci. 150, 480.
Cohnheim (1903). Zeitsch. physiol. Chem. 39, 336. Janney (1915). J. Biol. Chem. 20, 321.
Patterson and Starling (1913). J. Physiol. 47, 137. Soldaini (1876). Gazzetta, 6, 322.
Wood and Berry (1903). Proc. Camb. Phil. Soc. 12, 97.
IX. NOTE ON THE ROLE OF THE ANTI- SCORBUTIC FACTOR IN NUTRITION.
By JACK CECIL DRUMMOND.
From the Biochemical Laboratory of the Research Institute, Cancer Hospital, London.
(Received February 24th, 1919.)
CONSIDERABLE interest is focussed at the present time upon the antiscorbutic substance found in many natural foodstuffs, and the relationship of this sub- stance to scurvy in man and to experimental scurvy in guinea-pigs and monkeys lias been the subject of many investigations during the past few years. These researches have demonstrated that if individuals of these species are fed upon a dietary deficient in the antiscorbutic substance they will sooner or later exhibit the classical symptoms of scurvy. Other species, however, appear to be able to thrive for long periods of time upon similar dietaries, without showing apparent symptoms of ill-health, much less definite pathological lesions. Such a species is the rat. The researches upon the growth problem have been carried out very largely with this animal as an experimental subject, and they have yielded results which indicate that the rat requires for a satisfactory completion of its life cycle a diet containing not only adequately adjusted supplies of protein, fat, carbohydrate and inorganic salts, but also a sufficient supply of two accessory factors, which have been provisionally termed “fat-soluble A,” and “water-soluble B.” The general opinion has up to the present held that if the diet is adequate in these respects the rat will show a normal rate of growth and a normal standard of nutrition throughout the usual span. This has led to the assumption that the rat is a species repre- sentative of a type showing no susceptibility to scurvy, since to all intents and purposes it has been possible to obtain a normal standard of nutrition through- out its life period, although the diet has been seriously, if not totally, deficient in the antiscorbutic factor or “ water-soluble C.” The only alternative to such an assumption, as was pointed out by McCollum and Pitz [1917] is to conclude that scurvy is in reality not a deficiency disease in the sense in which the term has been employed during the past ten years.
Since this alternative has been shown to be unsatisfactory, and since it is now established without any doubt that scurvy is indeed a typical deficiency disease [Harden and Zilva 1918, 1], we are faced with the need for an explana- tion of the ability of the rat to grow.and maintain apparently excellent health in the absence of the antiscorbutic substance.
78 J. C. DRUMMOND
Quite recently Harden and Zilva |1918, 2] have brought forward experi- mental evidence that the antiscorbutie factor does play a beneficial réle in the nutrition of the rat. They state that rats subsisting on a diet containing the antiscorbutic substance as well as the water-soluble and fat-soluble factors erow better than rats from the diet of which the antiscorbutic factor is absent.
This, they point out, indicates that rats are susceptible to an antiscorbutic deficiency although they do not develop definite lesions of the disease. Some time prior to the publication of this paper the same point was under investi- gation in this laboratory, because some rough experiments had indicated that recovery from the pathological condition induced by a deficiency of fat-soluble A was in certain cases more rapid when a diet containing butter fat plus orange juice was given than when the ration contained butter fat but no fruit juice.
Two standard dietaries were made up as follows:
Ration | Ration 2
Purified caseinogen 20 parts 20 parts
be. starch OY 3 35 Is{0) he Butter fat PAD) 5 20) es Salt mixture De bes Dyes Yeast extract by Sars by ys Orange juice Ove IK) 53 Water A0 io 30) a
Several healthy litters of young stock rats of approximately the same age were selected and each litter was halved. Two batches were then made up of the halves of the divided litters, so that half of each litter would receive each dietary. At the commencement of the experiment the animals were about six weeks old and were all of very similar body weight. These two batches were fed, the one on the orange juice diet and the other on what may be termed the scorbutic ration.
Both lots grew well, and after maturity was reached breeding was per- mitted, and the growth and development of the second generation were also watched.
Throughout the experiment, more satisfactory development was shown by the batches receiving the orange juice addition to their diet. The differ- ence was not marked until the animals approached maturity, and was not discernible to the eye, being only apparent in a study of the body weight.
Table I gives the average weights of male rats upon the two diets. All the animals in each batch showed body weights very close to the average, so that there was no case of one or two low weights causing an unfair reduction of the value.
3
THE ANTISCORBUTIC FACTOR IN NUTRITION 79
Table I. Average Weights of Male Rats in grams.
Ration 1 Ration 2 Donaldson
Days (6 animals) (6 animals) [1915] 0 62 64 -- 30 124 122 125 60 161 183 : 184 90 193 213 223 120 208 234. 244 150 220 248 258 180 233 256 268 210 241 263 274 240 245 278 280 270 249 287 296
A similar observation was made in the case of the females, but in this case it is difficult to illustrate the fact clearly in tabular manner owing to the dis- turbances in the weight increments caused by pregnancies. In these two batches there was just as little to choose between the appearance of the animals as was observed in the case of the males. After maturity was reached the breeding propensities of both batches were good, although on the whole a larger number of litters was obtained from the females receiving the orange juice diet.
The representatives of the second generation nourished upon the two rations were in every case of good appearance, but Table II demonstrates that those receiving the orange juice supplement were able to grow at the more rapid rate. All litters were reduced to a uniform size (4) immediately after birth, so as to equalise the labour of rearing the young animals imposed upon the females of the two sets.
Table IT. Males Females g — Ration | Ration 2. Standard value Ration 1 Ration 2. Standard value Days average of 9 average of 11 of Donaldson average of 10 average of 18 of Donaldson 0 (birth) 5-7 ¢. 6-02. as 5-7 g. 6-0 ¢. = 7 14:5 ,, 15-0 ,, = 14-0 ,, 14-6 ,, a 14 23°5 ,, 23°D ;, aan 22°6 ,, 22°6 ., — 21 34:2 ,, 36-0 ,, 30-0 g. 30-2 ,, 32:0), 28-0 ¢. 28 48-0 ,, 52:0 ,, 48-5 ,, 48-0 ,, 50:0 ,, 41-0 ,, 56 122-0 ,, 146-0 ,, 110-0 ,, 1020 ,, 105-0 ,, 100-0 ,, 84 165-0 ,, 204:3 ,, 173°0 ,, 132-0 ,, 146-0 ,. 143-0 ,, 112 TiO) 226-4 ,, 213-0 ,, 146-0 ,, 159-0 ,, 166-0 ,,
Although these figures represent the averages of somewhat small numbers of animals, they may be taken as indicating that the rat requires the anti- scorbutic factor in order to achieve a normal development, and that although the requirements of this species are of a very much smaller order than those exhibited by man, the monkey or the guinea-pig, they are sufficiently well- marked to dispel any idea that there exists a fundamental difference in the nutritive requirements of the two types of animal.
80 J. GC. DRUMMOND
These results are in agreement with those published recently by Harden and Zilva [1918, 2].
It may therefore be accepted as experimentally proven that the dietary requirements of the higher animals include in addition to a satisfactorily balanced ration of protein, fats, carbohydrate and mineral salts, an adequate supply of three accessory food factors:
1. Fat-soluble A. 2. Water-soluble B, or antineuritic factor. 3. Water-soluble C, or antiscorbutic factor.
REFERENCES.
Donaldson (1915). The Rat. Philadelphia. Harden and Zilva (1918, 1). Biochem. J. 12, 270. (1918, 2). Biochem. J. 12, 408.
McCollum and Pitz (1917). J. Biol. Chem. 3, 229.
X. RESEARCHES ON THE FAT-SOLUBLE AC- CESSORY SUBSTANCE. I. OBSERVATIONS UPON ITS: NATURE AND PROPERTIES‘
By JACK CECIL DRUMMOND.
From the Biochemical Laboratory of the Research Institute, Cancer Hospital, London.
(Received February 24th, 1919.)
OTHER than numerous observations upon the distribution of fat-soluble A in foodstuffs curiously little attention has been devoted to that important sub- stance. The present study was begun with the object of ascertaining its chemical nature.
Few definite statements regarding the properties of fat-soluble A are to be found in the literature. By many it has been considered to be a relatively thermostable substance, a view which has led to suggestions being advanced that it was a more or less clearly defined chemical unit such as a lipoid. McCollum and Davis [1914, 1] reported that the factor was present in an ether extract of boiled eggs, whilst Osborne and Mendel [1915] found the growth promoting properties of butter fat unaffected by treatment with steam for over two hours.
More recently, however, the thermostability of the accessory factor has been questioned by Steenbock, Boutwell and Kent [1918], who have demon- strated that exposure of butter fat to a temperature of 100° C. for four hours is sufficient to destroy most, if not all, of its growth stimulating powers.
Regarding the chemical nature of fat-soluble A practically no information has been derived from a study of the literature. McCollum and Davis [1914, 2] hydrolysed butter fat in a non-aqueous medium at room temperature and, without attempting any separation of the products of hydrolysis, believed they had obtained evidence that fat-soluble A had survived the process.
A very considerable amount of work has been carried out in this laboratory aiming at the isolation and identification of fat-soluble A. Much of the earlier work, however, was based on an assumption that the factor is relatively stable to temperatures below 120°.
This assumption was founded partly on the results obtained by previous workers which have already been referred to, and partly on the observation
1 The author wishes to express his indebtedness to the Medical Research Committee for a grant which defrayed part of the expenses of this investigation.
Bioch. x11 6
82 J. C. DRUMMOND
that many oils, in the preparation of which drastic processes employing high temperatures have been used, were found to be rich sources of fat-soluble A. Kxamples of oils of this type are many fish oils |Drummond, 1918].
Kew of the early experiments will be described because they have scarcely any interest in the light of the results yielded by the examination of the stability of the accessory factor A.
The thermostability of this factor was not doubted until an enquiry into the effects of hydrogenation upon the factor present in oils was undertaken in collaboration with Professor W. D. Halliburton, at the request of the Oil and Fats Committee of the Food Investigation Board.
This investigation yielded results which demonstrated beyond any doubt that fat-soluble A, in the form in which it oécurs in natural animal fats, 1s much less stable to high temperatures than has been previously assumed. Such an observation necessitated a complete revision of all the previous experiments to isolate and identify the factor, in many of which high tempera- tures had been employed.
EXPERIMENTAL METHOD.
At an early stage it became apparent that some more or less approximately standardised method of testing substances for the presence of fat-soluble A was necessary, if any results of a comparable nature were to be obtained. The elaboration of such a method presented very great difficulties, as may well be imagined by those who have had experience in this field of research.
Although the method to be described is far from perfect, it has nevertheless yielded somewhat more useful results than would have been obtained by direct feeding tests of the usual type. The method is at present being used as a routine in this laboratory and should be useful for comparative studies, when improved by certain modifications which are now under consideration.
The animals used in trials of this nature are selected from our own carefully bred stock, and are, therefore, of a higher standard than those which form the average stock supplied by outside breeders.
The beneficial results of using none but home-bred stock are quickly apparent in this type of research and should be borne in mind by anyone who attempts to undertake it. Young healthy rats selected from the main stock
and weighing about 50g. are fed upon an artificial ration constituted as given below:
Purified caseinogen re ine ss ... 20 parts sin Starch: |): Sas est ae ioe. OO) ee Salt mixture as as of Soh ees Ae con fees Yeast extract (source of water-soluble B) VO kes Butter fat (source of fat-soluble A)... seat eae
Filtered orange juice (source of water-solubleC) 5 ,, The last constituent is added because it has been recently shown that rats show a more satisfactory growth and development when their ration contains
a
é >
THE FAT-SOLUBLE ACCESSORY SUBSTANCE 83
the antiscorbutic factor in addition to the other requisites. [Harden and Zilva, 1918; Drummond, 1919. ]
Whilst nourished upon this ration the young rats should show a strictly normal rate of growth and maintain good health. Any which fail to do so in this preliminary period are regarded as unfit for experimental purposes and are rejected. The remainder, having given evidence of a normal power to erow, are removed from the complete ration when they have attained an average body weight of 70 to 80 g. and are given a dietary similar to the above but in which the butter fat has been replaced by an equivalent amount of a fat known to be deficient in fat-soluble A.
For this purpose considerable use has been made of hardened linseed oil, which contains no detectable traces of the accessory substance A. Occasionally when a very important experiment was undertaken this fat was further purified by one recrystallisation from alcohol.
Fed upon this diet deficient in fat-soluble A the experimental animals should quickly cease growing. Occasionally a very vigorous individual will continue to grow for some weeks after the deficiency has been introduced and such animals should not be employed in experiments of a comparative nature. The majority of the young animals show greatly retarded growth after a week upon the deficient dietary. When it is definitely established that growth is inhibited by the deficiency of fat-soluble A, the linseed oil is either wholly or partially replaced by the substance to be tested. The behaviour of the animal is then closely watched during a period of from four to six weeks. An absence of fat- soluble A in the added fraction is indicated by no growth on the part of the animal, followed by a decline in health accompanied by the characteristic eye- condition which has somewhat loosely been termed a xerophthalmia.
On the other hand, if the animal shows growth, the rate