198

INSULIN.1BY J. J. R. MACLEOD, F.R.S., M.B., CH.B. ABERD.,PROFESSOR OF PHYSIOLOGY IN THE UNIVERSITY OF TORONTO.

[Prof. Macleod gave an historical survey of the massof work published on the relationship between thepancreas and disturbances in carbohydrate meta-bolism, from that of Brunner in the seventeenthcentury up to the time when convincing evidencethat an antidiabetic hormone does actuallv exist inthe pancreas was furnished by Banting and Best, theirexperiments being of a different type from those oftheir predecessors. He continued :-]

It now became apparent that attempts should bemade to see whether the crude alcoholic extracts ofox pancreas could be sufficiently freed from thosesubstances which made them unsuitable for sub-cutaneous administration to diabetic patients, eitherbecause of local irritation or general toxic effects.At this stage of the investigation we were joined byJ. B. Collip, who, in a remarkably short period oftime, succeeded in isolating from the alcoholic extracts,by fractional precipitation with alcohol, a precipitatewhich contained the antidiabetic hormone in highconcentration, and which, in watery solution, couldbe injected subcutaneously in man without anydeleterious effects. At the same time it was alsodiscovered that the blood-sugar is lowered in normalrabbits by subcutaneous injection of extracts con-

taining the antidiabetic hormone, and to thisobservation was due in large part the rapid progresswhich it was possible to make in the isolation ofinsulin. It offered a ready means for testing therelative potency of the various precipitates andfiltrates, and of determining the most favourableconditions under which extraction of the gland shouldbe carried out.Provided with a comparatively simple method for

the preparation of insulin, it now became possible toundertake a systematic investigation of its physio-logical and chemical properties and its possibletherapeutic value, and I shall endeavour to describebriefly what has been found, and to indicate, in ageneral way, what appear to be the most hopefullines for further research. It seemed advisable atthis stage to group the problems demanding immediateinvestigation, and this was done as follows : 1. Theeffect of insulin on the respiratory exchange, thedistribution of glycogen, and the metabolism of fatin animals rendered diabetic by pancreatectomy.2. Its therapeutic effect in diabetes mellitus. 3. Itseffect on the blood-sugar of normal animals and thesymptoms which result from overdosage. 4. Its

pharmacological assay. 5. Its effect on the blood-sugar in the various forms of experimental hyper-glycaemia. 6. The physiological mechanism by whichit lowers the blood-sugar. 7. Its source. 8. Itschemical reactions and its preparation on a largescale. In planning for the investigations of theseproblems we were in the fortunate position of havingthe whole-hearted collaboration of several trainedworkers, not only in my own department (E. C. Noble,J. Hepburn, and J. K. Latchford), but also in that ofinternal medicine under the direction of Prof. DuncanGraham (W. R. Campbell and A. A. Fletcher).The Effect o/’ ZMSM/tM. o)t the Metabolism o/* CarbohydratesThe Effeet of Insulin on the Metabolism of Carbohydrates

and Fats in Experimental Pancreatic Diabetes. I.From the time carbohydrates are absorbed, mainly

as glucose, into the blood of the portal circulation untilthey are completely oxidised, changes in chemicalstructure are constantly occurring. These consist,partly, in a condensation of several glucose moleculesto form glycogen, and partly in a splitting of themolecule, proceeding through various intermediatestages. The intermediary substances formed at eachstage are doubtless in a certain state of equilibrium,

1 A lecture (abridged) delivered before the Eleventh Interna-tional Congress of Physiology at Edinburgh on July 24th.

one with another. Many of them do not accumulatesufficiently to be detectable by chemical means, beingchanged into the next stage almost as quickly as theyare produced, and at the present time we are limited,in our attempts to follow the various steps in theprocess, to observations on the concentration ofglucose in the blood, the amount of glycogen depositedin the tissues, and the type of combustion occurringin the organism as a whole. In diabetes marked andsignificant alterations occur in each of these ; theblood-sugar rises to a high level, glycogen practicallydisappears from the liver and becomes decreased inthe muscles-except the heart, in which it increases-and the nature of the combustion process, as revealedby the behaviour of the respiratory quotient, becomeschanged so as to indicate that no carbohydrate isbeing oxidised. It is safe to assume that the increasein blood-sugar is secondary to the other changes, butit is difficult in the present state of knowledge tounderstand the relationship. It has been a favouritehypothesis that the primary fault in diabetes is inthe oxidation of glucose, or in the chemical changeswhich precede this, and that this leads to the draftingto the tissues from the glycogen reserves of a plethoraof sugar; a sort of forced-feeding process, as it were.Another hypothesis is that the glucose molecule mustbecome altered in some way before it can be eitheroxidised or condensed into glycogen, and that thisalteration does not occur in diabetes. Through thework of Irvine and his school much light has beenthrown in recent years on the structure of the sugarmolecule, and it has been shown that besides thewell-known a and &bgr; varieties of glucose, in which theoxygen linkage is between the first and fourth carbonatoms, there is also a third variety, called 7 glucose,in which the oxygen ring is displaced from the normalstable position. The detection of the 7 sugars dependsmainly on a comparison of the reducing and polarisingpowers of the solutions, and, by using this method,Winter and Smith have recently published resultswhich they interpret as showing that the alteration inthe glucose molecule alluded to above is the productionof 7 glucose. y glucose, according to this view, is anecessary preliminary stage in the oxidation of glucose,and, if we suppose that it is also so for its condensationinto glycogen, the inter-relationship of the threechanges found in. diabetes becomes immediatelyintelligible.

I have dwelt on these theoretical considerationsbecause of their great interest in connexion with themechanism of the action of insulin. For it is a

significant fact that this hormone, when administeredto depancreated animals, completely restores tonormal each of the three functions. We haverepeatedly found, for example, when it is given alongwith carbohydrate to depancreated dogs, that verylarge quantities of glycogen become deposited in theliver and that the respiratory quotient becomes raisedin the same manner as it does in normal animals withcarbohydrate alone. We know also from the work ofBanting and Best on depancreated dogs that itreduces the blood-sugar to the physiological level,that the glycosuria disappears, and that the well-beingof the animal becomes normal in every regard.We must conclude, therefore, that insulin is neces-

sary for the metabolism of carbohydrate, and we maysuppose that its function is in some way related to thepreparation of the glucose molecule for combustionand condensation. But as to whether this preparationconsists in a change of a, &bgr; glucose into the -y varietycannot as yet be considered as settled. Dr. G. S.Eadie has repeated Winter and Smith’s observations,and by strictly following their instructions hasobtained similar results. Briefly stated, these resultsare that after removal of the proteins by phospho-tungstic acid and alcohol from normal blood, thedextrorotary power of the filtrate is lower than itshould be as calculated from the reducing power, andon standing 24 hours the differences become less, andlater they practically disappear. On the other handwhen blood from diabetic patients or hyperglycaemicanimals is used, the dextrorotary and reducing value

199

correspond from the start, or the former may exceedthe latter. Administration of insulin causes thEdiabetic blood to behave in these regards like normal.The question is with regard to the interpretation oi

the results, and, as Hewlett has pointed out, this neednot necessarily be that 7 glucose is present, but rathelthat chemical changes have occurred in the verycomplicated procedure which is involved in thepreparation of the solutions for the polariscope.Others have believed that the blood may contain,besides glucose, other carbohydrates or relatedsubstances having a relatively low dextrorotarypower. Since this question is not yet settled I neednot dwell on the further claim of Winter and Smiththat insulin is capable of causing -y glucose to beformed when added to a solution of ordinary glucosealong with an extract of liver. I may state, however,that we have so far been unable to confirm this result.The action of insulin on the metabolism of fat in

diabetic animals is equally as striking as it is on that ofcarbohydrates, and here again its effects are revealedby chemical changes occurring at both ends as it wereof the metabolic chain. In a few days after pancreat-ectomy in the dog, the blood and the liver containexcessive quantities of fat, although the extreme edegrees of lipaemia sometimes encountered in diabeticpatients have not been observed. There is also acertain degree of acetonuria, although again this ismuch less pronounced than it is in diabetes in man.Evidently the migration of fat in the body is abnormal,and its final oxidation seriously interfered with.When insulin is given along with carbohydrates tothese animals, the acetonuria very promptly dis-appears, the blood fat, apparently more slowly,returns to normal, and the liver in a few days containsno more fat than that of a normal, similarly fedanimal. As the glycogen increases in this viscus thefat correspondingly declines. We cannot state atwhich stage in the metabolism of fats insulin exercisesits corrective influence, but wherever this may be,there can be little doubt that it acts indirectly becauseof its effect on carbohydrates. It has been said thatfat burns in the fire of carbohydrates, and althoughthis generalisation may, in some respects, be misleadingit is, nevertheless, a useful one to guide us in inter-preting the results with insulin.

’L’he Therapeutic Effects of Insulin inDiabetic 3fellitus.

As would be expected, the effects of insulin onpatients suffering from diabetes mellitus, in so far asthese have been studied, are of the same generalnature as those observed on diabetic (depancreated)dogs. Until the laboratory investigations had fur-nished perfectly definite results no serious attemptswere made to carry the investigations into the clinic,nor would it have been permissible to do so in viewof the failures of previous workers to provide anextract that did not cause toxic-symptoms. Withoutanimal experimentation we would be no further to-dayin the treatment of diabetes, and in the undoubtedsaving of human life which has been the result, thanwe were a little over a year ago when the treatmentwas first applied on any considerable scale.

In studying the results obtained by administrationof insulin on patients, we must bear in mind, however,that two factors may come into play that are notinvolved in the experiments on completely depan-created animals. One of these is dependent on thefact that there probably still remains in the majorityof diabetic patients a certain remnant of pancreatictissue capable of secreting insulin ; and the other, oncertain differences between laboratory animals andman, related in part to peculiarities in the type ofmetabolism and in part to the greater development ofthe nervous system, and to which the frequentoccurrence of coma as a serious symptom of theclinical form of the disease is due. Moreover, wemust remember that subjective phenomena can bestudied only on man.The first case in which a complete study of the

therapeutic value of insulin was made was that of a

L boy of 14 years of age, exhibiting all the classical symptoms of the disease, and on whom treatment by. dietetic control had resulted in no improvement.: Daily injections of insulin between Jan. 23rd andl Feb. 4th, 1922, reduced the blood-sugar even to thenormal level, greatly diminished the glycosuria, and. often caused it to disappear, removed entirely the, ketonuria for a period of some days, and, perhapsmost striking of all,

" the boy became brighter, moreactive, looked better, and said he felt stronger."There could be no doubt as to the great therapeuticvalue of the treatment, when judged both by theimprovement in the objective and subjective symptoms sof the disease. This, along with other cases treated byinsulin, and reported by Banting, Best, Collip,Campbell, and Fletcher, is only the first of a largenumber, exhibiting the disease in all its forms, thathave been most carefully studied in numerous clinicsboth in Canada and the United States and in thiscountry, and I cannot refrain from expressing myadmiration for the excellent spirit of collaborationwhich has made this possible. I am not qualified,nor even if I were would it be in place for me here todiscuss further the clinical results, but I may say thatthere can be no doubt that insulin is capable, whenwisely used under controlled conditions, of savingmany cases of diabetes which would otherwiseinevitably have succumbed to coma, or to one of themany other complications of the disease. It is alsocertain that, as an adjunct to dietetic control, treat-ment with insulin greatly improves the nutritivecondition and increases the resistance towardsinfections, thereby prolonging the expectation of life.The subjective condition and the sense of well-beingof the patients are also vastly improved. As towhether continued treatment with insulin will enablethe diabetic patient to recover any of his lost power ofproducing this internal secretion in his own pancreas,nothing definite can be said at present. It is at leastsignificant in this connexion that the islet tissuedevelops considerable powers of regeneration aftermuch of it has been destroyed as a result of ligation ofthe secreting ducts of the pancreas, as has beenshown by R. R. Bensley to be the case, at least in therabbit and guinea-pig. Arising as it does in thesecases from outgrowths of the duct epithelium, may itnot be possible in man, in whom the pancreas has notbeen completely destroyed by disease, that new islettissue will become regenerated when, by administrationof insulin, the strain to produce this hormoneendogenously is removed ?

Returning now to the more purely physiologicalaspects of the subject, let me briefly lay before youwhat is known regarding the effects of insulin onnormal animals. In the consideration of theseresults we must bear in mind that the conditions in thenormal animal are fundamentally different from thoseobtaining in the diabetic. In the normal animalthere is probably available at all times a sufficientamount of insulin for every purpose for which it isrequired, so that the addition of more from withoutupsets the metabolic balance and thereby induces anabnormal state. In the diabetic, on the other hand,the injection of insulin only restores to the body ahitherto missing part of the metabolic machinery andenables it to run smoothly again.

The Effect of Insulin on the Blood-sugar ofNormal Animals.

Within a few minutes of the intravenous or sub-cutaneous injection of a moderate dose of insulin, thepercentage of sugar in the blood begins to fall, and itcontinues to do so more or less in a straight line for aperiod which varies from about one half-hour toseveral hours, according to the animal, after which thecurve may gradually become less steep or it may beginto rise again.The prompt onset and the steep nature of the fall

in blood-sugar are characteristic features of theaction of insulin, and are quite unlike the hypo-glycaemic effects that may be produced by adminis-tration of other substances, such as phosphorus or the

200

insulin-like materials that have been prepared byCollip and others from clams and from variousvegetable sources. Indeed, the character of initial fallis such as to give the impression that the processresponsible for it must be one operating within theblood itself-an intravascular glycolysis. But thiscannot be the case, since we have been unable to findthat the addition of insulin to defibrinated blood keptunder sterile conditions at body temperature outsidethe body has any influence on the rate at which sugardisappears ; neither is there any difference in thisregard between the blood of normal animals and that ofanimals injected with insulin at varying periodspreceding the withdrawal of the blood. Evidently,therefore, the locus of action of insulin is extra-vascular ; it is in the tissue cells. We must concludethat the insulin sets up some process by which, as itwere, a vacuum for sugar becomes established in thesecells, so that sugar is removed from the blood. Onemay imagine that in the tissues there exists a certainbalance between the concentration of free glucose,the complex carbohydrates from which it may bederived, and the various derivatives into whichit breaks down prior to its oxidation. In other words,one may speak of a certain tension of glucose as beingnecessary in the tissues and consider the action ofinsulin as having the effect of reducing this tension sothat glucose diffuses into them from the blood in orderto maintain it.

After about one half-hour the blood-sugar in therabbit usually begins to rise again, or at least the fall tobecome much less pronounced, indicating that theeffect of insulin is becoming neutralised by theliberation of sufficient glucose from the glycogen storesof the muscles and liver. This, of course, means thatthe hypoglycaemia following insulin will be much more t

prolonged and will ultimately become much morepronounced in animals provided with small reserves ofglycogen than in those that are rich in glycogen.Such can readily be shown to be the case, and itis significant that a well-fed animal may withstandmany times the dose of insulin that would provefatal to one that was starved.

It is customary to imagine that the glycogenicmechanism of the liver is so finely attuned to respondto the concentration of sugar in the blood that

glycogenolysis immediately sets in when this fallsbelow the normal level. That such is not the casefollowing insulin may depend on the fact that theglycogen, which is first of all called upon to maintainthe glucose tension in the tissues, is that of themuscles themselves ; but we will not here venturefurther into the various interesting consequences ofsuch an hypothesis. It is at least significant thatthe glycogen of the muscles, as well as of the liver, isusually reduced in animals that have been givenlarge doses of insulin (McCormack and O’Brien andDale and Dudley). On account of its practicalsignificance it should be pointed out that it is becauseof the poverty of the glycogen reserves that the blood-sugar of diabetic patients treated with insulin ismore apt to fall suddenly to a dangerously low levelthan is that of normal persons.

The Hypoglyece-m-ic S’ymptoms.When the percentage of blood-sugar has fallen to

about 0-045 curious symptoms supervene. Thesewere first observed in rabbits, and in a typical caseconsist in a violent convulsive seizure in which theanimal throws itself over sideways, usually first in onedirection, then in the other, with the head retractedand the hind limbs in an extended position. Thecondition is not unlike that caused by strychnine or byacute asphyxia, except that certain groups of musclesare less affected by the convulsions than others.After a period, usually of from 30 seconds to a minute,the convulsions cease and the animal lies on its side inan unconscious state, perhaps showing running-likemovements of the extremities, and with rapid, shallowbreathing. The exact condition of the animal at thisstage may, however, vary considerably, well-fed andtherefore glycogen-rich animals often sitting in

apparently normal fashion between the convulsiveseizures, which are often brought on by attempts tomove. After a varying period the comatose stage isfollowed by another convulsive seizure, and thesephases may continue alternately for an hour or more,the convulsions becoming feebler and feebler and therectal temperature falling, until at last the animal diesof respiratory failure. After death rigor mortis setsin at once. The arterial blood is venous in colour, andit clots very quickly. Preceding the onset of typicalconvulsions there are usually premonitory symptomsof hyperexcitability and evidence of hunger; some-

times, however, paralysis of the extremities is thefirst symptom.The symptoms vary somewhat in other animals.

In the dog the first signs are usually very rapidbreathing, restlessness, and general hypersensitivity,muscular twitching then becomes evident and -thesphincters may relax. Barking is often a prominentsymptom, and there may be frothing at the mouth.At this stage, or as the first symptom, convulsions notunlike those seen in the rabbit may supervene, andbetween them the animal lies on its side, showingviolent twitching of the musculature, almost a tetany.It is evidently unconscious. The rates of breathingand of the pulse increase. Inspiration is usuallyshort and jerky, and inspiratory tetanus not infrequent,so that artificial respiration may have to be applied.Attempts to get on its feet are often the cause forconvulsive seizures and, during recovery, the musclesof the extremities, particularly the anterior, are seento have entirely lost their power of coordinate action.In etherised dogs even massive doses of insulin haveno immediate effect on the blood pressure.

In the cat the symptoms are like those in the dog,profuse salivation, mewing, and relaxation of thesphincters being especially marked.

In the mouse kept at room temperature, A. Kroghobserved that insulin may only cause the animal tobecome weak, but its temperature falls to a very lowlevel. If it be kept in an incubator at about 28° C.,however, characteristic symptoms supervene-convul-sions and coma, preceded by paralysis of the legs-with doses of insulin, which, when compared on abody-weight basis, are only about one-fifth thatnecessary for rabbits.

This influence of body temperature is particularlyinteresting, since it probably explains the apparentimmunity of frogs even to massive doses of insulin.We failed to note any symptoms even in three daysafter injecting insulin. but A. Krogh has since foundthat they occur in a day or so later, a fact which wehave confirmed.

In man, in whom the subjective symptoms can beinterpreted and cardio-vascular changes more accu-rately observed, the picture is different. When theblood-sugar reaches about 0-075 per cent. the patientexperiences extreme hunger and a sense of fatigue.He usually becomes anxious and may lose hisemotional control. Actual tremor of the musculatureis rarely seen, but there is a definite sense of tremulous-ness and some incoordination for fine movements." Vasomotor phenomena are common ; pallor or

flushing, sometimes one after the other ; a sense ofheat, of chilliness, almost always a profuse sweat

"

(Banting, Campbell, and Fletcher). At lower levelsof blood-sugar, acute mental distress, mental disturb-ances, delirium, and finally coma, with loss of thedeep reflexes supervene.The correspondence between the onset of definite

objective symptoms and a blood-sugar percentage of0-045 in laboratory animals is remarkable. Sometimes,as in starved animals receiving a large dose of insulin,the blood-sugar may be found to be considerablybelow this level when convulsions first appear. Veryoccasionally it is decidedly higher, and immediatelyafter a convulsion the sugar may rise somewhat.These facts suggest that the symptoms are the resultof the lowering of the tension of glucose in the tissuecells below a certain critical level-a condition of"

glucatonia," we may call it-and to which the>hypoglycaemia runs parallel. This view is supported

201

by the striking manner in which the symptoms arealmost instantly removed by administration of glucose,and it is of great significance that other hexosemonosaccharides have only a slightly beneficial effect,although, of course, their injection causes a markedrise in the reducing power of the blood. Mannoseand laevulose seem to be somewhat more efficient thangalactose. None of the pentoses or disaccharides hasany effect. Injection of epinephrin, subcutaneously,also removes the symptoms, no doubt because itexcites the production of glucose out of the glycogenreserves. It is of use, therefore. only when thereserves are considerable, and it should not be dependedupon for antidoting the symptoms of glucatonia indiabetic patients, because in them the glycogen storesare apt to be low. A similar relationship between thepercentage of blood-sugar and convulsive symptomshad previously been described by Mann and Magathin dogs deprived of the liver ; administration of glucosealso removed the symptoms, other monosaccharideshaving only a slight effect.

There can be little doubt, then, that the symptomsare related to a lowering of the glucose tension in thetissues. But we cannot imagine that this, in itself, isresponsible for the violent stimulation of the respira-tory and muscular centres that evidently occurs.

Some unphysiological condition must become developedas a result of the glucatonia, and it has been suggestedthat anoxaemia may be an important factor. Olmstedand Logan point out that there is a remarkableresemblance between an insulin convulsion and anasphyxial convulsion, and that the arterial blood isinvariably intensely venous in colour when convulsionssupervene. " It may possibly be that through thelowering of blood-sugar certain oxidative processesbecome depressed to such a degree that the brain cellsare affected in much the same manner as in asphyxia."Whatever may be the nature of the toxic condition,

it is clearly on the cells of the pons and medulla thatit acts. It does not, like strychnine, act on the motorcentres of the spinal cord, since insulin causes noconvulsive symptoms in a decapitate cat, althoughit may lower the blood-sugar to well below theconvulsive level (Olmsted and Logan).On what particular detail in the chemical structure

of the molecule the remarkable specificity of glucosein antidoting the symptoms may depend, cannot asyet be said, but it is evident that, by studying therelative antidoting value of various sugars andsubstituted sugars, we are furnished with an interestingmethod for determining the biological significance ofthe different side chains and of the other moleculararrangements of the glucose molecule.

The Pharmacological Assay of InsuZ,in.The extent to which insulin lowers the blood-sugar

forms the basis for its pharmacological assay, one unitbeing originally defined by us as the amount which isrequired to lower it to 0-045 per cent. within fourhours in a rabbit weighing 2 kg. and after 24hours’ fasting. Since convulsions supervene at thislevel, their occurrence can be used as a check on theassay, and when small animals, such as mice, are used,so that large numbers can be injected with varyingamounts of insulin, it can be made by observing theonset of the symptoms alone. In this case one mouseunit is defined by A. Krogh as the quantity of insulinwhich can cause convulsions in 50 per cent. of theanimals within two hours, precautions being taken tomaintain the body temperature and to standardisethe diet. Dale and Burn have also tried this methodof assay.Even when rabbits are carefully selected so as to

correspond in weight and in every other particular,are similarly fed prior to the preliminary starvationperiod, and are injected intravenously so as to obviateany uncertainties in the rate of absorption, the assaysas judged either by the blood-sugar curve or theoccurrence of convulsions, do not always correspond.Differences in glycogen content, which can never beaccurately adjusted, by dieting are probably largelyresponsible. We have attempted to circumvent these (

difficulties by confining the observations on the bloodsugar to the first half hour after the insulin injections,hoping in this way to circumvent the influence of theglycogen reserves, but with no conspicuous success.Probably the most accurate method of assay is thatworked out in Toronto by Dr. F. N. Allen on depan-created dogs. The principle of this method, whichwas suggested by our clinical associates, is to determinethe number of grammes of glucose that a given quantityof insulin can cause to be metabolised while the animalis on a diet containing enough carbohydrate so thatthere is some glycosuria. He has found that thecarbohydrate equivalent of insulin, as it is called,varies somewhat with the weight of the animal andthe carbohydrate balance, but that is satisfactorilyconstant when these conditions are made uniform.It is not likely that this method will succeed in theclinic, because of the varying degree to which differentdiabetic patients retain the power to produceinsulin.

I have said enough about this problem of assay toindicate how difficult it is, and I will not detain youfurther with detail. It is probably advisable to pointout, however, that it has been found advantageousto adopt, for clinical purposes, temporarily at least,a unit which is one-third that of the standard unit asdefined above. This is done to obviate the necessityof prescribing fractions of a unit for the treatment ofmild cases of diabetes.

The Effect of Ins1.Ûin in Experimental Hyperglycaemia.Insulin also lowers the blood-sugar in every known

form of experimental hyperglycaemia, whether this beproduced by the addition of glucose exogenously orendogenously. With regard to the former type,Eadie has found that when insulin is injected intorabbits coinciding with, or at varying periods precedingthe injection of sugar, the curve of the resultinghyperglycsemia becomes greatly altered. When theinsulin injection coincides with that of sugar the curverises to the same height as with sugar alone, but itreturns much sooner to the normal level, andwhen the sugar is given some time after theinsulin the curve is not only of shorter durationbut is also much less in height. The maximal

influence of insulin in suppressing the curve is whenthe sugar is given at an interval of about 75-90minutes after the insulin. At longer intervals thecurves progressively rise again, the suppressinginfluence of insulin having, as a rule, disappeared inabout eight hours. We have attempted to use thesefacts as the basis for a method of assay, but with nogreat success.

Of the endogenous forms of experimental hyper-glycaemia, we have studied the effect of insulin on thosedue to piqure, to epinephrin, to asphyxial conditions,and to anaesthetics, and Stewart and Rogoff havestudied it on the peculiar form due to morphine.The hyperglycaemia which becomes rapidly developedin glycogen-rich rabbits in all of these conditions isgreatly depressed, if not entirely absent, in animalsinjected with insulin, this depending, obviously, onthe amount injected. Most attention has been paidto epinephrin hyperglycaemia, because in this case

alone is it possible to control the intensity of thehyperglycaemia by varying the dosage. This, it willbe seen, offers a method for the assay of insulin, andEadie has found, when equal quantities of epinephrinare injected into a series of uniformly fed rabbitsalong with varying quantities of insulin, that there isusually a definite mathematical relationship betweenthe dose of insulin and the extent to which the blood-sugar rises.

Further interest attaches to the antagonistic actionbetween insulin and epinephrin, with regard to theblood-sugar, because they are both hormones, althoughit is questionable whether the hormonic action of

epinephrin is of importance to the animal. It is

significant in this connexion that Stewart and Rogoffhave found that the hypoglycsemic effect of insulinis not perceptibly different in adrenalectomised as

compared with normal animals, thus showing that

202

the adrenals are in no way linked with the internal Isecretion of the pancreas in the control of sugar ’Irmetabolism.The observation that insulin can prevent the

hyperglyceemia due to ether is of clinical importance,because it is probably on account of this hyper-glycaemia that a large part of the risk of surgicaloperations on diabetic patients is due. It should bepointed out, however, that the hypoglycsemic actionof insulin is much more readily demonstrated by havingthe animal well under the effects of insulin before theether is administered, than by giving insulin toanimals that are already anaesthetised. This wouldseem to indicate that diabetic patients should betreated with insulin before being anaesthetised, and,indeed, that this preliminary treatment should extendover several days so that the reserves of glycogenmay be augmented. I am informed by my clinicalcolleagues that insulin is of little avail in combatingpost-operative coma if its administration is delayeduntil after the operation.An interesting antagonism between pituitrin and

insulin has been described by Burn. Given in largedoses, pituitrin like insulin may inhibit the hyper-glycaemia due to epinephrin or to ether, but wheninsulin and pituitrin are given together to normalanimals, instead of there being a greater fall in blood-sugar than insulin alone would produce, there is littlechange or, it may be, a rise. A dose of pituitrinwhich in itself has little effect on blood-sugar is thuscapable of neutralising the hypoglycaemic action ofinsulin. This observation has been confirmed byLogan and Olmsted, who have also found that it isnecessary to give enormous amounts of insulin tocause any decided decrease in the hyperglycsemiawhich is present in decerebrate cats still retaining thepituitary gland. When this gland is absent, on theother hand, not only is the hyperglycsemia much lesspronounced, as Bazett had suggested it would be,but insulin promptly lowers the blood-sugar andhypoglycaemic convulsions can readily be induced.

All of the forms of experimental hyperglycsemiajust alluded to are primarily due to a rapid breakdown Iof the glycogen of the liver, and that produced byepinephrin may therefore be taken as a type. Thisdoes not, of course, warrant the assumption, thatsome have made, that a hypersecretion of this hormoneis the underlying cause of the hyperglycsemia in theother forms. Indeed, Stewart and Rogoff have con-clusively shown that such is not the case, with thepossible exception of that caused by morphine.Taking epinephrin hyperglycaemia as a type, then, itseemed of interest to compare the amount of glycogenleft in the livers of uniformly well-fed rabbits afterthey had been given equal doses of epinephrin, someof them at the same time also receiving insulin.This observation, made by E. C. Noble, has shownvery definitely that the insulin protects the glycogenagainst breakdown. In one experiment, lastingeight hours, the livers of two animals receivingepinephrin alone contained 1-40 and 1-96 per cent.of glycogen, whereas those of three others whichwere given insulin along with the same amounts ofepinephrin contained 8-12, 6-4, and 3-64 per cent.of glycogen. It is of interest to note that it wasnecessary to give about 40 times the usual dose ofinsulin to cause convulsions under these conditions,and that sometimes the convulsions occurred whilethere was still over 10 per cent. of glycogen remain-ing in the liver. The results show quite clearlythat insulin antidotes the glycogenolytic action ofepinephrin.The Physiological Mechanism by which Insulin lowers

the Blood-Sugar.And now we are arrived at a stage where it will

be advantageous to consider briefly the possiblemechanism of the hypoglyceemic action of insulin.We have presented evidence that this is not becauseit stimulates increased glycolysis within the blooditself, and have concluded that insulin in some waylowers the tension of glucose in the tissue cells.

Further evidence that insulin causes the sugar topass from the circulating fluid into the tissues morequickly than normal has been furnished by perfusionexperiments on the isolated mammalian heart.Hepburn and Latchford, using Locke’s method, found,as previous workers have, that an average of 0-9 mg.of glucose per gramme of (rabbit) heart per hour dis-appears from the perfusion fluid. By adding insulin tothe perfusion fluid the average sugar consumptionrose to over 3-0 mg., and in some cases reached thevalue of over 4’0 mg., the highest without insulinbeing below 2 mg. They did not find that insulinaffected the heart-beat, nor could they detect anysignificant differences in the glycogen content of thehearts used in the two groups of experiments.Hoping to be able to demonstrate a similar effect of

insulin on the sugar content of fluid perfused through .

skeletal muscles, these workers attempted to perfusethe hind limbs of various laboratory mammals, butalthough every possible precaution was taken toavoid interruption of the circulation prior to startingthe perfusion, or a fall in temperature, they couldnever succeed in preventing either marked oedema orvaso-constriction. This failure led them to comparethe sugar concentration of blood removed at the sametime from the femoral artery and vein, and from theportal vein, of anaesthetised or decerebrated animals,and to see whether insulin would cause the differencesnormally existing between these bloods to becomechanged. It was found not to do so. McCormackhas since found that this method is also incapable ofrevealing an increased retention of glucose as a resultof insulin, either in the muscles or in the liver ofdepancreated animals, in which, as we have seen,insulin certainly causes glycogen to be formed. Thisshows that the method is not sufficiently sensitive forthe purpose for which it was employed.The evidence of the observations on glycolysis and

the perfused heart leave little doubt, however, thatthe glucose is lost in the tissue cells, and the problemnarrows itself down to the cause for its disappearance.The possibilities that may be considered are : itspolymerisation into glycogen, its oxidation, and itsreduction to substances similar to fatty acids.With regard to its possible conversion into glycogen,

the fact that insulin causes large amounts of this poly-saccharide to become deposited in the liver of depan-created animals when they are fed on carbohydrateswould seem at first sight to lend support. But theconditions, as already pointed out, are fundamentallydifferent in the normal organism. As a matter offact, it has been shown that insulin given to normalrabbits, instead of causing glycogen to be deposited,has the opposite effect, and either prevents thedeposition of glycogen when it is given alongwith carbohydrates to previously starved animals(McCormack and O’Brien), or causes the stores ofthis substance to become reduced when given towell-fed animals (Dale and Dudley).

Further evidence that insulin does not stimulateformation of glycogen in normal animals has beenobtained by Noble by perfusion experiments on theliver of the turtle. When each of the two lobes isperfused for several hours with Ringer’s solutioncontaining glucose, under exactly similar conditionsexcept that the fluid on one side contains insulin, therelative amounts of glycogen in the two lobes werefound to be the same as they usually are. Since thenormal variation may be considerable, the method is.not a very delicate one, and on this account Noblehas also used that of observing the sugar content ofthe perfusion fluid, as recommended by Snyder,Martin, and Levin. By comparing this for equalperiods of time, with and without the addition ofinsulin, no differences could be detected. The onlyobjection to the experiment is that insulin maynot act quickly enough in the case of cold-bloodedanimals to make it possible in the time of perfusion formeasurable quantities of glycogen to be deposited.

Clearly, therefore, none of the disappearance ofsugar from the blood of normal animals which follows.the injection of insulin can be accounted for by its.

203

polymerisation to glycogen. Why, then, does insulincause glycogen to be deposited in the diabetic animal,and why does it inhibit the breakdown of glycogen inexperimental hyperglycaemia ? The reply, based onthe considerations already set forth, is that an excessof insulin beyond that actually required for thecontrol of carbohydrate metabolism develops towardsglycogen, an action which is the reverse of that whichit has under normal conditions. When insulin is

given to the diabetic animal, on the other hand, itsupplies a previously missing agency necessary forthe oxidation and polymerisation of glucose. Toexplain the retarding effect which insulin has onglycogen breakdown in experimental hyperglycoemia,we may suppose that the sugar which is produced bythis process is a mixture of active and inactive forms,and that insulin, by tending to produce the former inrelative excess, disturbs the balance existing betweenthe sugars and glycogen, so that less glycogen isbroken down. This suggestion assumes some sort ofbalanced action between glycogen and its hydrolyticproducts, for which, however, there is no experimentalevidence.

-

The possibility that glucose disappears because I

excess of insulin causes its oxidation to be increasedhas been investigated by observing the respiratoryexchange. In the experiments of Dixon, Eadie, andPember, dogs and rabbits were used. In the former,the results show that insulin causes a very markedincrease in oxygen consumption and in respiratoryvolume and rate, but very little change in the respira-tory quotient. These changes do not, however, beginto be evident until from 30-60 minutes after givingthe insulin, by which time the blood-sugar hascompleted its initial fall and stands at about 0-060 percent., which indicates that the sudden call for sugaron the part of the tissue cells cannot be attributedto its increased oxidative breakdown. There can belittle doubt that the increased metabolism is dependentupon the hyperexcitability and the increased musculartone of the animal. It reaches its acme when con-vulsions supervene, and it develops either steadilyor intermittently. Evidently, as the tension of glucosedeclines, conditions become established within thenerve centres which excite them and which may bedependent upon an intracellular lack of oxygen.That the development of the symptoms depends onthe lowering of the glucose tension is supported bythe observation that their severity is lessened by theinjection of glucose.

In rabbits, in which it will be recalled increasedmuscular tone is not so marked a symptom as in dogs,the increase in oxygen consumption does not set inuntil the convulsions occur--indeed, there is often,if not usually, a drop both in this and in respiratoryvolume. Sometimes in rabbits the respiratoryquotient rises decidedly. It would be rash at this

-

stage to draw definite conclusions from these results,but it is quite plain that the initial fall in blood-sugar,which is primarily what we must attempt to explain,is not due to any increased combustion of glucose.The remaining possibility that glucose is reduced

to substances related to the fatty acids receives somesupport in certain of the observations from thebehaviour of the respiratory quotient. This may riseto over unity without there being any sufficient degreeof hyperpnoea, and of consequent " washing out " of002 from the blood to account for it. To be quitecertain of this latter possibility, Dixon and Pemberhave examined the blood removed from dogs atregular, frequent intervals after giving insulin withoutfinding that there is any significant change in the002-combining power. This observation, by theway, also shows that there can be no production ofacid substances as a result of insulin, a fact whichhas also been demonstrated by F. N. Allen fromobservations on the behaviour of the acid excretionby way of the urine.We are, therefore, at a loss to explain the cause for

the initial disappearance of sugar from the blood ; it is not converted into glycogen, its oxidation in thetissues is not augmented, at least until after increased

muscular activity sets in, and if some of it does becomereduced to fat-like substances this is not an invariableprocess.

The Source of Insulin in the Higher Animals.The evidence that insulin is derived in the higher

animals from the islets of Langerhans and not fromthe acinar (zymogenous) cells of the pancreas has,until recently, been of an indirect nature. Continuingthe work of Diamare and Laguesse on the comparative eanatomy of the islets of Langerhans, Rennie, in 1905,described in detail their relationship to that of thezymogenous tissue in numerous Teleostei. He found,particularly in Lophius piscatorius, Myoxocephalus,and the Pleuronectidse, that the islet cells exist asseparate encapsulated glands lying in the mesenteryquite distinct and apart from the bulk of the zymo-genous tissue, which is usually spread as narrow bandsalong the branches of the mesenteric blood-vessels.These glands he called principal islets, and be foundthe largest of them to be situated near the anteriorpole of the spleen.Although the more recently discovered methods of

differential staining for islet cells have not beensuccessfully applied to the principal islets, no one hasdoubted the conclusion of Diamare and Rennie thatthey are homologous with those of the islets ofLangerhans of the mammalian pancreas. The absenceof these structures in the zymogenous tissue in theTeleostei supports this view, as does also the fact thatin the Elasmobranchi, in which there are no principalislets, the pancreas itself, which is a compact gland,contains abundant islet tissue. In collaboration withFraser, Rennie investigated the effect of the principalislets, administered by mouth, on diabetic patients,but without results that could be considered to showdefinite amelioration of the symptoms. In one casean extract of the gland was injected subcutaneously,but the development of toxic symptoms made arepetition of the experiment inadvisable. In view ofthe now well-established fact that insulin is quiteinactive when given by mouth, it is clear why theseearlier observations should have failed to yield satis-factory results. Recently, we have reinvestigatedthe problem, and have found very definitely thatalcoholic extracts of the principal islets cause a

profound lowering of the blood-sugar of the normaland diabetic animals, including man, when they areinjected subcutaneously or intravenously, whereassimilarly prepared extracts of the zymogenous tissuehave no such effect-indeed, not infrequently, rathercause the percentage of blood-sugar to increase. Incontrast to these results, an extract of the pancreas ofthe skate, which contains islet tissue, has a markedinsulin effect,,2 and so also had one prepared from theintercsecal fat of the trout (in which the islets andzymogenous tissue lie separately embedded). Theislets are, therefore, the only source of insulin in thebony fishes, and we may conclude that they are alsoso in the mainmals, since in them none of this hormoneis produced after pancreatectomy.

This conclusion is supported by the experiments onduct ligation and on grafting already referred to, andby the experiments of Banting and Best, although inall of these there still remained the possibility thatpartially-degenerated or regenerated zymogenoustissue might be the source of the hormone.Much attention has been given in recent years to the

mechanism of the control of the secretion of glandsproducing internal secretions, and, since many ofthose concerned in the production of the externalsecretions are under nerve control, it has been supposedthat this must also be the case with the ductless glands.It cannot be said, however, that much success hasattended the various endeavours to demonstrate thisnerve control. Since the islets of Langerhans areprovided with a rich nerve supply, which is believedto come in large part from the right vagus, and sincean increased discharge of their internal secretion,insulin, would be revealed by the behaviour of the

2 These extracts have an unusually marked effect in causingviolent convulsions.

204

blood-sugar, McCormack and O’Brien have investi-gated the effect which is produced on the latter byelectrical stimulation of the right vagus, in decerebratecats and in etherised dogs and rabbits. To avoidcardio-inhibitory effects, the nerve has either beenstimulated below the origin of the cardiac fibres, or ithas been allowed to undergo partial degeneration.The fact that hyperglycaemia follows stimulation of thegreat splanchnic nerve further suggests the possibilitythat hypoglycsemia might result from stimulation ofthe vagus.

Similar observations were made some years ago byde Corral working in Asher’s laboratory, with resultswhich are at least suggestive. Those of the presentinvestigation are more definite, although not entirelyconvincing. In many cases no change can be detectedin the level of the blood-sugar following the stimulationbut in several others there is a prompt fall, which isusually of a temporary nature, in spite of continuedstimulation. In one or two cases, however, stimula-tion with slow, weak induced shocks for a long periodhas caused a gradual decline in blood-sugar, in onecase to 0-070 per cent., with a return towards thenormal level after removal of the stimulus. It is ofgreat interest in connexion with these experimentsthat Best and Scott have recently discovered thatinsulin can be prepared from blood by the usualmethods.

The Chemical Reactions and the Preparation of Insulin. ’

Before concluding, allow me to say a few wordsconcerning the methods used in the preparation ofinsulin and its chemical properties, in so far as theseare known at present. The method of preparationdeveloped by Collip depends on the solubility of insulinin concentrations of ethyl-alcohol up to about 90 percent., above which it becomes precipitated. Sincemost proteins are insoluble at concentrations ofalcohol that are considerably below this, they can beremoved from the original extract of pancreas (whichis made with 50-60 per cent. alcohol) by fractionalprecipitation after the extract has been concentratedby evaporation in vacuo. A watery solution of theprecipitate which settles out at a concentration of 90per cent. alcohol is insulin. It can be evaporated todryness in vacuo without loss of potency, and thepowder contains about 14 per cent. of nitrogen, is freefrom phosphorus (Doisy, Samogyi, and Shaffer), butcontains sulphur. A solution gives a distinct biurettest, but only faintly, those of Millon and Hopkins-Cole. The insulin is readily absorbed by kaolin andcharcoal, and considerable losses are apt to occur inpassing it through a porcelain filter. It does notdialyse through membranes or collodion sacs ; it is

precipitated by half-saturation with ammoniumsulphate, by barium salts of weak acids, and by picricacid. The solubility in water and alcohol is deter-mined by the H-ion concentration, the iso-electricpoint being at a pH between 5 and 6. Watery solu-tions at a pH below 5 can be boiled for several hourswithout any loss in potency, but this occurs morerapidly when the pH is about 7. The potency quicklyvanishes in the presence of pepsin or trypsin. In acid

I

solution insulin is relatively stable to mild oxidisingand reducing agents. Insulin is also soluble in phenoland the cresols, glacial acetic acid, and formamide. Itis insoluble in most other organic solvents (Malony).

These reactions, and its destruction by pepsin and itrypsin, suggest that insulin is of the nature of analbumose, but it is possible that it is really onlyadherent to this protein, and is itself of much simplerchemical structure. This view is supported by thefact that we have been unable on several occasions to ]detect any biuret reaction with insulin prepared from (the pancreas of the skate or to obtain from this source i

precipitates containing it by means of high concentra- <tions of alcohol, or by adjustment of the iso-electric ,point. Whether it is really a protein or not, it issignificant that all the methods now used for itspurification are based on reactions that are character- gistic of colloids. Thus precipitation at the iso-electric tpoint (Doisy, Samogyi, and Shaffer), or by picric acid t

followed by the regeneration from the picrate as ahydrochloride (Dudley), or by benzoic acid (Malony),or by half saturation with ammonium sulphate. Itis highly creditable to those who have been responsiblefor producing insulin on a large scale for clinical usethat they should have succeeded in so short a time intheir work.

It may be assumed that insulin is present in allanimals, and since structures similar to the islets’ofLangerhans have not been described in the inverte-brates, it is likely that in them its site of production isdelegated to other structures. What these may be inthe more highly-developed invertebrates is as yetundetermined, but it is of interest that Collip, shortlyafter isolating insulin from mammalian pancreas,prepared it from clams, and suggested that it must bepresent wherever glycogen is found. Later he prepareda substance like it from yeast, th as confirming Winterand Smith. The insulin prepared from these sourcesappears to differ from that prepared from pancreas,in that its action in lowering the blood-sugar isdelayed, sometimes for a day or so, which is also thecase with an insulin-like substance prepared by Collipfrom various vegetable sources and styled by him, inrecognition of these differences " glucokinin." Bestand Scott have also succeeded in preparing insulin, orinsulin-like substances, from various other sources.

It would appear, therefore, that the distribution ofhormones capable of accelerating the disappearanceof glucose from the blood is widespread, not only inthe animal kingdom but in the vegetable as well, butit is probable that these do not all have the samephysiological action. In the investigation of this wemust bear in mind that many substances have beenknown for long to be capable of lowering the blood-sugar ; for example, peptone, hydrazin, certainmineral salts, phlorhizin, &c. All of these also causeglycogen to disappear from the liver, and, indeed,certain of them cause structural changes in this organ.Although it would certainly be desirable to havesubstances like glucokinin having a more prolongedaction than insulin in depressing the blood-sugar, wemust exercise some caution in recommending themfor clinical use until it is certain that they have nodeleterious effects on the liver. It is possible alsothat these substances merely stimulate the secretionof insulin from the pancreas of the animal into whichthey are injected, or serve as precursors for its produc-tion. If such be the case, then they can have littlevalue in diabetes.

I have endeavoured in this review to adhere strictlyto ascertained facts, and have refrained from venturingto speculate with regard to the numerous problems inmetabolism, both of animals and plants, for theinvestigation of which insulin may prove a usefulinstrument. If I have succeeded in giving you abird’s-eye view, as it were, of the results that have beenobtained, I shall consider that my efforts have not beenin vain, and I shall feel repaid if, as an outcome, othersjoin us in the work on insulin.

KING’S COLLEGE HOSPITAL FESTIVAL DINNER.-The Festival Dinner of King’s College Hospital was held onJuly 17th at the Savoy Hotel, the Duke of Connaught in thechair. The Duke of Connaught, in proposing the toast ofthe hospital, spoke of the anxiety given by the presentposition of the London hospitals to those whose duty andprivilege it was to conduct their work, and referred to thefact that King’s College Hospital was the first that gaveaccommodation as a military hospital (No. 4 London GeneralHospital) during the war. During the time when the hospitalwas in military hands some 31,000 soldiers passed throughIts wards, together with 11,000 civilians ; over 370 past andpresent members of the medical school, and over 150 trainedaurses served with the forces. Unfortunately the waroccurred at the time when the appeal for the new hospitalwas being made, and the building debt at the end of 1922imounted to 673,000. He associated with the toast of the10spital the name of Viscount Hambleden, its chairman.Viscount Hambleden, replying, emphasised the value of thevoluntary hospital system, and stated that during the pastrear the 5600 in-patients and 42,000 out-patients had them-eives given .613,000 towards the cost of their maintenance. Atihe close of the proceedings the Duke of Connaught announcedhat 26000 had been subscribed during the evening.