Therole ofdiffusion barriers in determining the ... · in the median nerves during and after a 30 minute period of limb ischaemia, ... Therole ofdiffusion barriers in determining

J. Neurol. Neurosurg. Psychiat., 1970, 33, 310-318

The role of diffusion barriers in determiningthe excitability of peripheral nerve

K. N. SENEVIRATNE AND 0. A. PEIRIS

From the Departments of Physiology and Medicine, Faculty of Medicine, University of Ceylon, Colombo,Ceylon

SUMMARY The excitability changes occurring in normal isolated peripheral nerves of rats have beenstudied during exposure to hypoxic and anoxic conditions before and after the administration ofinsulin. The changes observed have been explained in terms of the dynamics of K' equilibrium in theperiaxonal spaces, and attention is drawn to the importance of the relative impermeability of theperiaxonal diffusion barrier in determining this equilibrium. Isolated peripheral nerves fromalloxan-diabetic rats, studied under similar conditions, show significant differences in the sequenceof their excitability changes. It has been shown that the rate of change of excitability in these nervesis slower than those of control nerves. These results have now been interpreted in terms of the K'changes in the periaxonal space. It is concluded that these slower excitability changes are due to anincrease in the permeability of the diffusion barrier of the diabetic nerve to potassium.

Recent studies have drawn attention to the re-sistance of the peripheral nerve of diabetic subjectsto inactivation by ischaemia (Steiness, 1959, 1961aand b; Castaigne, Cathala, Dry, and Mastropaolo,1966; Gregersen, 1968). Seneviratne and Peiris(1968a and b), measuring the excitability changesin the median nerves during and after a 30 minuteperiod of limb ischaemia, have demonstrated thatthe nerves of diabetic subjects do not show thephases of ischaemic and post-ischaemic hyper-excitability that are characteristic of the normalsubject. The absence of the sensations of para-esthesiae in the diabetic subjects was attributed tothe relatively small rate of change of excitabilityseen during the ischaemic and post-ischaemicphases. The most characteristic functional differencebetween the normal and diabetic nerves, however,was the very limited extent to which the diabeticnerve was inactivated by a 30 minute period ofcomplete vascular occlusion. Seneviratne and Peiris(1969) have also demonstrated that the isolatedperipheral nerves of alloxan-diabetic rats show asimilar resistance in vitro to the effects of hypoxia.Two alternative mechanisms could be responsible

for this phenomenon. It has been suggested thatdiabetic nerves can maintain their activity underanoxic conditions by utilizing non-oxidative meta-bolic pathways. The experiments described in thispaper were designed to test the alternative hypothesissuggested by Seneviratne and Peiris (1969) that

the sequence of excitability changes during anoxiawas determined by the dynamics of K' equilibriumin the periaxonal space.

METHODS

Experimental diabetes was produced in 6-month-oldrats weighing between 100 to 150 g. A single dose ofalloxan-monohydrate of 150 mg/kg body weight infreshly prepared citrate-phosphate buffer was injectedintraperitoneally at a concentration of 10 mg/ml. Thecriteria used to establish diabetes were a blood glucoselevel of over 200 mg/100 ml. and a glycosuria of 1% ormore. The animals were maintained in this conditionfor at least four weeks before being used for study.Litter mates of these rats and the alloxanized non-diabetic animals served as controls in the experimentsdescribed below. The rats were anaesthetized withintraperitoneal sodium pentobarbitone and the sciaticnerves dissected out rapidly from the level of the sciaticnotch to the gastrocnemius tendon. The nerves werecleaned of fat and blood vessels, but the nerve sheathwas left intact. The nerve was then immersed in mam-malian Tyrode's solution at 37°C and exposed to agas mixture containing 95% 02 + 5% CO2 for 15minutes to minimize any injury activity. At the end ofthis period the nerve was mounted in a small moistnerve chamber containing platinum recording and stimu-lating electrodes and a gas inlet tube. The chamberwas made air-tight and the nerve exposed to varyinggas mixtures which were admitted into the chamberat a constant flow rate. The gas mixtures used in theseexperiments were 95% 02 + 5% C02, 95% N2 + 5%

310

Protected by copyright.

on March 10, 2020 by guest.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.33.3.310 on 1 June 1970. Dow

nloaded from

The role of diffusion barriers in determining the excitability ofperipheral nerve

CO2 and 8% 03 + 5% CO2 in nitrogen. The nervewas stimulated using square-wave stimuli of 0-01 msecduration and variable voltage and the responses amplifiedand displayed on one beam of an oscilloscope, the sweepof which was triggered by the stimulator output. Singlesweeps of the trace were photographed on 35 mm film.In all the experiments described below a nerve was usedfor one experiment only. Details of the technique usedin producing the experimental diabetes, and of recordingthe evoked responses from the nerve in the gas chamberhave been described earlier (Seneviratne and Peiris,1969).

EXPERIMENT 1 The nerve mounted in the chamberwas exposed to the 95% 02 + 5% CO2 gas mixtureand the stimulus strength required to produce a responsenearly 50% (± 5%) of the maximum response amplitudewas determined. The nerve was then exposed to the8% 03 + 5% CO2 in N2 gas mixture and the responsesto this stimulus recorded at two minute intervals for30 minutes. One hour after the first sciatic nerve wasdissected out, the second nerve was removed, immersedfor 15 minutes in the oxygenated Tyrode solution, andmounted in the nerve chamber. The stimulus strengthrequired to produce a 50% response was determined,the nerve exposed to the 95% N2 + 5% CO2 mixture,and the evoked responses to this stimulus recorded attwo minute intervals for 30 minutes. Fifteen controlanimals were used in this series. The first sciatic nerveof each animal was exposed to the 8% 02 mixture orthe 95% N2 mixture alternately to ensure that the delayof one hour between the experiments was not a criticalfactor.

EXPERIMENT 2 One sciatic nerve was dissected out, asample of venous blood taken for estimation of glucosecontent, and 20 u. soluble insulin (Boots) injectedintraperitoneally. The isolated nerve was then exposedto the 8% °0 mixture and its responses to a stimulusproducing a 50% of maximum response recorded attwo minute intervals. One hour later the second nervewas dissected out, the blood glucose level determined,the nerve exposed to the 8% 02 mixture, and itsresponses to the sub-maximal stimulus recorded. Tenhealthy animals were studied in this series.

EXPERIMENT 3 Experiment 2 was repeated using 10diabetic rats.

RESULTS

EXPERIMENT 1 The results obtained from oneexperiment of this series is illustrated graphicallyin Fig. 1. It shows the response of a nerve in 8%02 increasing to a maximum amplitude of 170%of its original size in the 10th minute, after which itdiminishes in size, total inactivation being reachedin the 22nd minute. The parallel nerve exposed tothe 95% N2 + 5% CO2 mixture, however, reachesa peak amplitude of 200% in the sixth minute,and total inactivation in the 12th minute. Figure 2

0N

:1600

CD10

: 80

400)

0 10 20 30Minutes

FIG. 1. The effects of hypoxia (filled circles) and anoxia(open circles) on the amplitude of the compound actionpotential evoked by a sub-maximal stimulus of constantsize.

depicts the relationship between the maximumresponse amplitude and the time taken to reach itfor the 15 pairs of nerves in this series and from10 nerves in experiment 2. In 8% 02 the meanmaximum potential size was 141-8% (range 108%-175 %), this being reached in a mean time of 6-4minutes (range two to 10 minutes). In the 95% N2mixture the corresponding values for maximumsize were 165-0% (range 114-228%) and for time

aL)N

.V_

U'a1)

0C

QU'

a1)

a1)00)C

EEx

12Minutes

FIG. 2. Time to reach maximum increase of responseamplitude. Control nerves in hypoxia (filled circles) andanoxia (open circles). Straight lines fitted by the methodofleast squares.

311

Protected by copyright.

on March 10, 2020 by guest.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.33.3.310 on 1 June 1970. Dow

nloaded from

K. N. Seneviratne and 0. A. Peiris

41JAi..

AAt

JLJLJL

FIG. 3. Effects of hypoxia on the compoundaction potential evoked by a sub-maximalstimulus of constant strength. Peripheral nerveof a control rat. Upper row: nerve beforeinsulin. Lower row: nerve after administrationof insulin. In each record the lower tracemonitors the stimcllus. Figures indicate time inminutes.

0 5 10 15 20 30

200a)N

160

c

0

120

' 80

CDX 40 -

C.)

0 10 20 30Minutes

FIG. 4. Results obtained from experiment depicted inFig. 3 plotted graphically. Percentage changes in re-

sponse amplitude of nerve before insulin (filled circles)and after insulin (open circles).

4 7 minutes (range two to eight minutes). In 8 % 02all the nerves were inactivated to less than 10%of their original size in a mean time of 24-2 minutes(range 18 to 28 minutes), whereas in 95% N2 themean inactivation time was 14 4 minutes (range10 to 19 minutes).

EXPERIMENT 2 Ten experiments were done in thisseries, and in each animal the blood glucose levelhad fallen by at least 25 mg/100 ml. (range 26 to 44mg/100 ml.) at the time the second nerve was dis-sected out for experiment. The evoked responsesobtained from one such experiment are reproducedin Fig. 3, and the results of this experiment de-picted graphically in Fig. 4. The changes in thebehaviour of the nerves before and after insulinseen in these two curves were typical of this series.

Before insulin a maximum amplitude of 140%was reached in six minutes and near completeinactivation occurred at the 25th minute, whereasafter insulin a peak potential size of 160% wasreached in four minutes, the nerve being inactivatedto less than 10% of its original size by the 18thminute.The results of six similar experiments are repre-

sented in Fig. 5. Figure 4 shows that after insulinthe nerves attain a larger peak amplitude after ashorter time than do the corresponding nervesbefore insulin. Figure 6 and the Table, which sum-marizes the results obtained from the experimentsof series 1 and 2, illustrate the close similaritythat exists between the behaviour of nerves fromhealthy rats exposed to a 95% N2 + 5% CO2mixture and comparable nerves from insulinizedrats exposed to a gas mixture containing 8% 02+ 5% CO2 in nitrogen.

EXPERIMENT 3 The ten rats studied in this serieshad been diabetic for periods varying from 30 to50 days and had initial blood glucose levels varyingfrom 238 to 516 mg/100 ml. One hour after theinjection of insulin the blood glucose level of eachanimal had fallen by at least 75 mg/100 ml. (range76 to 210mg/100 ml.). Theevoked responses obtainedduring one experiment where the nerves were ex-posed to the 8 % 02 mixture before and after insulinare reproduced in Fig. 7, and expressed graphicallyin Fig. 8.The first nerve shows a peak response amplitude

of 115% at the sixth minute, after which the nerveis slowly inactivated, but even at the end of 30minutes the response retained 52% of its originalsize. In contrast, the parallel nerve, which wasdissected out one hour after the administrationof insulin, gave a response which was 152% of its

312

j - --

Protected by copyright.

on March 10, 2020 by guest.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.33.3.310 on 1 June 1970. Dow

nloaded from

The role of diffusion barriers in determining the excitability ofperipheral nerve

0 10 20 30

0 10 20 30

200

160

120

80

40 -

0 10 20 30

0 10 20 30

Minutes

FIG. 5. Results obtainedfromsix control nerves, showingeffect of hypoxia on the sizeof the evoked response beforeinsulin (filled circles) andafter insulin (open circles).

0 10 20 30

resting size at the end of six minutes, and it retainedthis size for a further 12 minutes. At the end ofthe 30 minute period of hypoxia the response wasstill 124% of its original amplitude. The results ofsix similar experiments, reproduced in Fig. 9,illustrate the characteristic changes that takeplace in diabetic nerves after insulin. In all of themthe rising phases of the curves have steeper gradi-ents, reach higher levels, maintain these higherlevels for longer periods, and even after 30 minutesretain potential sizes larger than those shown bythe corresponding pre-insulin nerves. There was no

obvious relationship between the extents of thesechanges and the duration of the diabetic states,their original glucose levels or the levels to whichthe glucose levels had fallen.

DISCUSSION

There is evidence that intracellular potassiumis released from the nerve during anoxia and thatpotassium is reabsorbed in the post-anoxic re-covery period (Fenn and Gerschman, 1950; Shanes,1950). Krnjevic (1955) has shown that there is a

TABLEEXCITABILITY CHANGES IN ANOXIC AND HYPOXIC GAS MIXTURES

Maximum potential Time to reach Time to reachsize (%) maximum size (min) 90% inactivation (min)

In 8% 02 108-175 2-0-10-0 18-0-28-0 )(N = 15) (Mean 141-8) (Mean 6-4) (Mean 24-2)

Exp. I -Pis <-001 Pis <-001In N, 114-228 2-0- 8-0 10-0-19-0(N = 15) (Mean 165) (Mean 4-7) (Mean 14-4) J

Before 140-176 6 0-9 0 18-0-26-0 )insulin (N = 10) (Mean 153-2) (Mean 7-4) (Mean 24-1)

Expt. 2 -Pis <-001 -P is <-001After 120-206 3-8-6-0 11-0-22-0insulin (N = 10) (Mean 167-4) (Mean 4-9) J Mean 16-20 J

Values of P determined by Student's t test.

a)N._

.um

01)cnco0C.

0)a)-cC.)0)

0

313

Protected by copyright.

on March 10, 2020 by guest.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.33.3.310 on 1 June 1970. Dow

nloaded from

K. N. Seneviratne and 0. A. Peiris

a).n

Q1)

a)

0-n

a).C_

xn

200 ra,NeU

0

U,Q)

5C:

160

120

80a)CYc- 40

0Minutes

FIG. 6. Time to reach maximum increase of responseamplitude. Ten pairs of control nerves in hypoxia beforeinsulin (filled circles) and after insulin (open circles).Continuous straight lines fitted by the method of leastsquares. Interrupted line indicates slope of line obtainedfrom anoxic nerves.

slow efflux of K' from cat nerve when it is suspendedin oxygenated isotonic sucrose, 10% of the totalintracellular K' diffusing out within 10 minutes.When the nerve was exposed to chloroform fumesat 37°C for 20 minutes beforehand, the K' effluxwas exponential with a very much shorter timeconstant, 750% of the total K' diffusing out witha half period of 3 5 minutes. Maruhashi and Wright(1967) have shown that oxidative metabolic energyis required to maintain the resynthesis of the ATPof the nerve membrane enzyme complex. Whendeprived of this metabolic energy Ca- is releasedfrom the membrane, increasing its permeability(Frankenhaeuser and Hodgkin, 1957; Kimizukaand Koketsu, 1963), and permitting the Na' K' tomigrate down their concentration gradients.

0

a

10 20 30Minutes

FIG. 8. Results obtained from experiment depicted inFig. 7 plotted graphically. Percentage changes in responseamplitude of diabetic nerve before insulin (filled circles)and after insulin (open circles).

The accumulation of K' in the periaxonal spaceswould reduce the resting membrane potential ofthe nerve fibres (Shanes, 1950; Huxley and Stampfli,1951; Adrian, 1956). The initial depolarizationproduced in this manner would, in effect, resultin an increase of the excitability of the nerve bylowering the threshold voltage required for stimu-lation. Continued increase of the periaxonal K'concentration would, however, cause more extensivedepolarization which results in failure of actionpotential generation and conduction block. Thissequence of change would account for the transientphase of hyperexcitability seen in peripheral nervesin vitro at a stage before they are inactivated bythe anoxia (Heinbecker, 1929; Lehmann, 1937;Seneviratne and Peiris, 1969). Similar changes have

I/1\_ FIG. 7. Effects of hypoxia on the responsesevoked by a sub-maximal stimulus of constantstrength. Peripheral nerve of an alloxan-diabetic rat. Upper row: nerve before insulin.Lower row: nerve after insulin. In eachrecord the lower trace monitors the stimulus.Figures indicate time in minutes.

314

Protected by copyright.

on March 10, 2020 by guest.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.33.3.310 on 1 June 1970. Dow

nloaded from

The role of diffusion barriers in determining the excitability of peripheral nerve

200r

160

120

80

200r

1E

1"

401

0 10 20 30

200

20

'orLI40

0 lo 20 30

160

140

80

401

0 10 20 30

FIG. 9. Results obtained fromsix diabetic nerves showingeffects of hypoxia on theresponse amplitude beforeinsulin (filled circles) and afterinsulin (open circles).

C) 200cmc

-s 160 -

120

80 -

40 -

0 10 20 30 0 10 20 30

Minutes

0 10 20.-30

been demonstrated in human peripheral nerves

during experimental limb ischaemia (Kugelberg,1946; Fullerton, 1963). Seneviratne and Peiris(1968a) measured excitability changes in the sensory

fibres of the median nerve during vascular occlusionby observing changes in the size of the responsesevoked by a constant sub-maximal stimulus ofcritical size. It was shown that an increase in theexcitability of fibres ordinarily just beyond therange of this stimulus would reduce their thresholdsand render them excitable. Thus increases inexcitability would result in the recruitment ofadditional fibres, this change being manifest as an

increase in the size of the recorded compoundaction potential. Conversely, a decrease of excita-bility would be indicated by a diminution in thesize of the recorded potential. That a high surfaceK' concentration is sufficient to produce depolariza-tion block is evident from the studies of Shanes(1951) who showed that washing of the nerve

with oxygen-free Ringer's solution alone was

sufficient to restore functional activity in anoxicfrog nerve fibres.

If surface K' concentrations are to play a criticalrole in the depolarization process, it would benecessary to postulate the existence of diffusionbarriers in the nerve which would restrict the free

diffusion of ions from the periaxonal to the extra-cellular fluid compartment. It is conceivable thatthe functional changes described above couldoccur either if individual axons were surrounded bydiffusion barriers or if groups of axons were enclosedwithin a common barrier. There is general agree-ment that nerves are surrounded by relativelyimpermeable external sheaths (Feng and Gerard,1930; Feng and Liu, 1949; Rashbass and Rushton,1949; Crescitelli, 1951; Dainty and Krnjevic,1955), and it has been suggested by Krnjevic(1954) that the function of such a sheath may beto preserve the specific character of the internalenvironment of nerve tissue. There is, however,no agreement as to the precise site of the barrier.Causey and Palmer (1953) viewed the epineuriumas the diffusion barrier, whereas Huxley and Stampfli(1951); Krnjevic (1955); Thomas (1963); andGamble (1964) favoured the perineurium. TheSchwann cells of myelinated fibres are ensheathedby a basement membrane which is continuous across

the nodes of Ranvier from one internode to another.Haftek and Thomas (1968) indicate that the neuril-emmal sheath, made up of the Schwann cell base-ment membrane and the inner endoneurial collagensheath of Plenk and Laidlaw, is a continuouselastic tube which remains intact even during

a)N

a)a)

Cecna)

. _

315

o0

Protected by copyright.

on March 10, 2020 by guest.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.33.3.310 on 1 June 1970. Dow

nloaded from

K. N. Seneviratne and 0. A. Peiris

nerve degeneration, serving to define the channelswithin which the Schwann cells proliferate duringnerve regeneration.The results obtained in experiment 1 provide

evidence in favour of the hypothesis outlined above.Theory predicts that the K' efflux would be largerwhen the nerve is in the anoxic gas mixture thanwhen it is in the 8% 02 mixture; it is thereforeanticipated that during anoxia the K' accumulatingwithin the diffusion barrier would attain the con-centrations required for critical depolarizationof the membrane and for inactivation of the mem-brane at time intervals which are shorter than forcomparable nerves during hypoxia. The results ofexperiment 1 indicate that, with the nerve in 8% 02,the mean time to reach maximum amplitude was6-4 minutes, and to reach inactivation 24-2 minutes,whereas in the 95% 02 + 5% CO2 mixture thecorresponding times were reduced to 4-7 and 14-4minutes.Another observation derived from the results

of experiments 1, 2, and 3 is that the maximumamplitudes reached tend to be greater as the timesrequired to reach this become shorter. It has beenargued above that increase in response size is dueto progressive recruitment of new fibres of higherthreshold, as continuing depolarization lowerstheir thresholds to levels which make them responsiveto the test stimulus. Continued accumulation ofK' would, however, also tend to depolarize towardsinactivation levels the low threshold fibres whichearlier contributed to the response. Since the sizeof the response at any given time represents thesummated contributions of the active fibres, itfollows that this potential would be increased bythe contributions being made by the freshly re-cruited fibres and diminished as a result of theprogressive withdrawal of the low threshold fibres.In this situation it is conceivable that a more rapidincrease of the periaxonal K' concentration would,during a transient period, permit of a greaterdegree of synchronization of the recruitment process.This would allow of occasions when nearly allthe fibres of the compound nerve are responsive,at a given moment, to the submaximal test stimulus.

Zierler (1959) showed that insulin producesan influx of K' into excised rat muscle, and that itincreases the resting membrane potential of muscleby about 5 mV. There is also evidence (Gamble,1962) that K' ions are required for oxidativephosphorylation in mitochondria, and that K'ions move into cells when glucose is taken upfrom the extracellular fluid into tissue cells andconverted into glycogen. Field and Adams (1964)have shown that insulin increases the in vitrouptake of glucose by alloxan-diabetic rabbit nerves

and suggest that insulin in vitro reverses the meta-bolic defect of diabetic nerves which is evidencedby its depleted resting glycogen levels. In experi-ments 2 and 3 insulin has been used to increase theinitial intraneuronal K' content, with the specificobject of increasing the K efflux from them duringhypoxia.

In experiment 2 the results obtained from suchnerves are compared with those from control nervesin 8% 02. It is expected that nerves with an in-creased intraneuronal K' content would, duringhypoxia, produce a more rapid build up of peri-axonal K' concentrations within their impermeablebarriers, leading to a more rapid developmentof the hyperexcitable and inexcitable states.This is borne out by the results of experiment 2.Control nerves in 8% 02 reached maximum excita-bility in a mean time of 7 4 minutes and wereinactivated by the 24th minute, whereas afterinsulin the corresponding times were reduced to4 9 and 16-2 minutes.

In experiment 3 the nerves from the diabeticanimals were first exposed to the 8% 02 mixture,and the results obtained are essentially similar tothose published earlier (Seneviratne and Peiris,1969). The characteristic differences between thebehaviour of these nerves and those from healthyrats relate to the extent of inactivation by thehypoxia and to the time relationships of the earlyphase of hyperexcitability. The results show thatthere is significant delay in the rate at which thenerve is subsequently inactivated. These changescould be related to a slower build up of K' con-centration in the periaxonal spaces of the diabeticnerve. Since the K' concentration in the space isdetermined by the relative rate of K' efflux from thehypoxic nerve and the rate at which it could diffuseout across the barrier, it follows that a slower K'build up could be accounted for by a changein either of these states. Our results show thatthe rate of increase of excitability in the post-insulin diabetic nerves is consistently greater thanin the pre-insulin nerves. This is evidence that therate of K' build up in the space is now faster thanbefore, a circumstance which should lead to quickerinactivation were the barrier an impermeableone. The results, however, show that this does notoccur, but that inactivation occurs at an even slowerrate than in the pre-insulin nerve. It is thereforenecessary to postulate that the diffusion barrierlimiting the periaxonal space in the diabetic nerveis a permeable one, permitting the K' to diffuseout more freely across it.

Thickening of the basement membrane of smallblood vessels has been consistently demonstratedin the retina, kidney, skin, stomach, skeletal muscle,

316

Protected by copyright.

on March 10, 2020 by guest.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.33.3.310 on 1 June 1970. Dow

nloaded from

The role of diffusion barriers in determining the excitability ofperipheral nerve

vasa vasorum of the aorta, and the vasa nervorumof peripheral nerve (Fagerberg, 1966). The character-istic lesions of diabetic microangiopathy withretinopathy, glomerulosclerosis, and diffuse base-ment membrane thickening have also been demon-strated in dogs made diabetic by bovine pituitarygrowth hormone or alloxan (Engerman andBloodworth, 1965). Spiro (1963) has suggested thatthe diversion of hexose along insulin independentpathways could result in the over-production of thebasement membrane substance and other glyco-proteins, which could account for the observedthickening. Blumenthal, Hirata, Owens, andBemns (1964) have suggested that the basementmembrane abnormality may be a manifestationof a more generalized antigen-antibody reaction.

Bischoff (1968) demonstrated a significant thicken-ing of the basement membrane surrounding theSchwann cells of the peripheral nerves of diabeticsubjects associated with the duplication and multi-plication of the membrane. He observes that inorder of frequency of occurrence of morphologicalchanges in these nerves, hyperplasia of the basementmembrane is the most prominent. This structuralchange is, however, not characteristic ofanyparticularaetiology because reduplication of the membranealso occurs in regenerating nerves (Nathanieland Pease, 1963). Simpson (1962) has suggestedthat the role of one or other of the axon coveringsis paramount in determining the velocity changesseen in disease and nerve compression experi-ments. He discussed the possibility that this couldbe due to physical distortion of the lipid-proteinmolecular arrangement in myelin or of the peri-neurium, or to a more direct effect on the excitablemembrane of the nerve fibre affecting the rate oftrans-membranal exchange of electrolytes. Bischoff(1968) suggests that the only early abnormalityof ultrastructure which could have some bearingon the functional disturbance in diabetic neuropathyis the thickening and reduplication of the basementmembrane surrounding the Schwann cell. He arguesthat hyperplasia of this exchange barrier would notbe without effect on metabolic exchange, and thatthis could be expected to cause a delay in thedevelopment of the action potential which wouldresult in a diminution in the overall speed of con-duction of the nerve impulse. More recent studies,however, indicate that the lowering of the conduc-tion velocity is due to the paranodal demyelinationwhich occurs in these nerves. Seneviratne andPeiris (1968b) have since shown that the resistanceof the nerves of diabetic subjects to ischaemicinactivation can be demonstrated even in thepresence of normal motor and sensory nerveconduction velocities. In this paper it is suggested

that the resistance is due primarily to an increasedpermeability of the periaxonal diffusion barriersto K'. Bischoff's (1968) observation of thickenedand reduplicated basement membranes in diabeticnerves is not necessarily incompatible with thisincreased permeability to the diffusion of electro-lytes, because the thickening of the basementmembrane of the glomerular capillaries seen in avariety of kidney lesions is invariably accompaniedby an increase of its permeability to the plasmaproteins.

We are grateful to Messrs. K. S. A. B. Fernando andS. Vairavanathan for their invaluable technical assistance.One of us (K.N.S.) is in receipt of research grants fromthe University of Ceylon, Colombo, and the Ministryof Scientific Research of the Government of Ceylon.

REFERENCES

Adrian, R. H. (1956). The effect of internal and externalpotassium concentration on the membrane potentialoffrog muscle. J. Physiol. (Lond.), 133, 631-658.

Bischoff, A. (1968). Diabetic neuropathy. Germ. med. Mth.,13,214-217.

Blumenthal, H. J., Hirata, Y., Owens, C. T., and Berns, A. W.(1964). In Small Blood Vessel Involvement in DiabetesMellitus. Edited by M. D. Siperstein, A. R. Colwell,and K. Meyer. Amer. Inst. Biol. Sci: Washington, D.C.

Castaigne, P., Cathala, H. P., Dry, J., and Mastropaolo, C.(1966). Les r6ponses des nerfs et des muscles a desstimulations electriques au cours d'une 6preuve degarrot ischemique chez l'homme normal et chez lediab6tique. Rev. neurol., 115, 61-66.

Causey, G., and Palmer, E. (1953). The epineurial sheathof a nerve as a barrier to the diffusion of phosphateions. J. Anat. (Lond.), 87, 30-36.

Crescitelli, F. (1951). Nerve sheath as a barrier to the actionof certain substances. Amer. J. Physiol., 166, 229-240.

Dainty, J., and Krnjevid, K. (1955). The rate of exchangeof "INa in cat nerves. J. Physiol. (Lond.), 128,489-503.

Engerman, R. L., and Bloodworth, J. M. B., Jr. (1965).Experimental diabetic retinopathy in dogs. Arch.Ophthal., 73, 205-210.

Fagerberg, S. E. (1966). The angiopathy of diabetes, inDiabetes Mellitus, pp. 53-62. Edited by L. J. P. Duncan,Pfizer Medical Monographs I, Edinburgh, EdinburghUniversity Press.

Feng, T. P., and Gerard, R. W. (1930). Mechanism of nerveasphyxiation: with a note on the nerve sheath as adiffusion barrier. Proc. soc. exp. Biol. (N. Y.), 27,1073-1076.

Feng, T. P., and Liu, Y. M. (1949). The connective tissuesheath of the nerve as an effective diffusion barrier.J. cell. comp. Physiol., 34, 1-16.

Fenn, W. O., and Gerschman, R. (1950). The loss of potas-sium from frog nerves in anoxia and other conditions.J. gen. Physiol., 33, 195-204.

Field, R. A., and Adams, L. C. (1964). Insulin responseof peripheral nerve 1. Effects on glucose metabolismand permeability. Medicine (Baltimore), 43, 275-279.

Frankenhaeuser, B., and Hodgkin, A. L. (1957). The actionof calcium on the electrical properties of squid axons.J. Physiol. (Lond.), 137, 218-244.

317

Protected by copyright.

on March 10, 2020 by guest.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.33.3.310 on 1 June 1970. Dow

nloaded from

K. N. Seneviratne and 0. A. Peiris

Fullerton, P. M. (1963). The effect of ischaemia on nerveconduction in the carpal tunnel syndrome. J. Neurol.Neurosurg. Psychiat., 26, 385-397.

Gamble, J. L. (1962). Retention of potassium by mito-chondria. Amer. J. Physiol., 203, 866-890.

Gamble, H. J. (1964). Comparative electron-microscopicobservations on the connective tissues of a peripheralnerve and a spinal nerve root in the rat. J. Anat. (Lond.),98, 17-25.

Gregersen, G. (1968). A study of peripheral nerves in diabeticsubjects during ischaemia. J. Neurol. Neurosurg.Psychiat., 31, 175-181.

Haftek, J., and Thomas, P. K. (1968). Electron-microscopeobservations on the effects of localized crush injurieson the connective tissues of peripheral nerve. J. Anat.(Lond.), 103,233-243.

Heinbecker, P. (1929). Effect of anoxemia, carbon dioxideand lactic acid on electrical phenomena of myelinatedfibers of the peripheral nervous system. Amer. J.Physiol., 89, 58-83.

Huxley, A. F., and Stampfli, R. (1951). Effect of potassiumand sodium on resting and action potentials of singlemyelinated nerve fibres. J. Physiol. (Lond.), 112, 496-508.

Kimizuka, H., and Koketsu, K. (1963). Changes in the mem-brane permeability of frog's sartorius muscle fibersin Ca-free EDTA solution. J. gen. Physiol., 47, 379-392.

Krnjevid, K. (1954). The connective tissue of the frogsciatic nerve. Quart. J. exp. Physiol., 39, 55-72.

Krnjevid, K. (1955). The distribution of Na and K in catnerves. J. Physiol. (Lond.), 128,473-488.

Kugelberg, E. (1946). Injury activity and trigger zones inhuman nerves. Brain, 69, 310-324.

Lehmann, J. E. (1937). The effect of asphyxia on mam-malian A nerve fibers. Amer. J. Physiol., 119, 111-120.

Maruhashi, J., and Wright, E. B. (1967). Effect of oxygenlack on the single isolated mammalian (rat) nervefiber. J. Neurophysiol., 30, 434-452.

Nathaniel, E. J. H., and Pease, D. C. (1963). Collagen and

basement membrane formation by Schwann cells duringnerve regeneration. J. Ultrastruct. Res., 9, 550-560.

Rashbass, C., and Rushton, W. A. H. (1949). The relationof structure to the spread of excitation in frog's sciatictrunk. J. Physiol. (Lond.), 110, 110-135.

Seneviratne, K. N., and Peiris, 0. A. (1968a). The effect ofischaemia on the excitability of human sensory nerve.J. Neurol. Neurosurg. Psychiat., 31, 338-347.

Seneviratne, K. N., and Peiris, 0. A., (1968b). The effect ofischaemia on the excitability of sensory nerves in diabetesmellitus. J. Neurol. Neurosurg. Psychiat., 31, 348-353.

Seneviratne, K. N., and Peiris, 0. A., (1969). The effects ofhypoxia on the excitability of the isolated peripheralnerves of alloxan-diabetic rats. J. Neurol. Neurosurg.Psychiat., 32,462469.

Shanes, A. M. (1950). Electrical phenomena in nerve. 1.Squid giant axon. J. gen. Physiol., 33, 57-74.

Shanes, A. M. (1951). Potassium movement in relationto nerve activity. J. gen. Physiol., 34, 795-807.

Simpson, J. A. (1962). Conduction velocity of peripheralnerves in human metabolic disorders. Electroenceph.clin. Neurophysiol, Suppl., 22, 36-43.

Spiro, R. G. (1963). Glycoproteins and diabetes. Diabetes,12,223-230.

Steiness, I. (1959). Vibratory perception in diabetics duringarrested blood flow to the limb. Acta med. scand., 163,195-205.

Steiness, I., (1961a). Vibratory perception in non-diabeticsubjects during ischaemia, with special reference to theconditions in hyperglycaemia, after carbohydratestarvation and after rcortisone administration. Actamed. scand., 169, 17-26.

Steiness, I. (1961b). Influence of diabetic status onvibratory perception during ischaemia. Acta med.scand., 170, 319-338.

Thomas, P. K. (1963). The connective tissue of peripheralnerve: an electron-microscope study. J. Anat. (Lond.),97, 35-44.

Zierler, K. L. (1959). Effect of insulin on membrane potentialand potassium content of rat muscle. Amer. J. Physiol.,197, 515-523.

318

Protected by copyright.

on March 10, 2020 by guest.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.33.3.310 on 1 June 1970. Dow

nloaded from