Correlation between electrophysiological effects of mexiletine and ischemic protection in central nervous system white matter

Neuroscience Vol. 71, No. 1, pp. 27-36, 1996

~

Pergamon

0306-4522(95)00430-0

Elsevier ScienceLtd Copyright © 1996 IBRO Printed in Great Britain. All rights reserved 0306-4522/96 $15.00 + 0.00

CORRELATION BETWEEN ELECTROPHYSIOLOGICAL EFFECTS OF MEXILETINE A N D ISCHEMIC PROTECTION IN CENTRAL NERVOUS SYSTEM WHITE MATTER P. K. STYS* and H. L E S I U K Loeb Medical Research Institute, Division of Neuroscience, Ottawa Civic Hospital, University of Ottawa, Ottawa, Canada Abstract--Protection of CNS white matter tracts in brain and spinal cord is essential for maximizing clinical recovery from disorders such as stroke or spinal cord injury. Central myelinated axons are damaged by anoxia/ischemia in a Ca2+-dependent manner. Leakage of Na + into the axoplasm through Na + channels causes Ca 2+ overload mainly by reverse Na+~Ca 2÷ exchange. Na ÷ channel blockers have thus been shown to be protective in an in vitro anoxic rat optic nerve model. Mexiletine (10 #M 1 mM), an antiarrhythmic and use-dependent Na ÷ channel blocker, was also significantly protective, as measured by recovery of the compound action potential after a 60 min anoxic exposure in vitro. More importantly, mexiletine (80 mg/kg, i.p.) also significantly protected optic nerves from injury in a model of in situ ischemia. This in situ model is more clinically relevant as it addresses drug pharmacokinetics, toxicity and CNS penetration. Optic nerve recovery cycles (defined as shifts in latency of compound action potentials with paired stimulation) were used to measure the concentration of mexiletine in optic nerves afler systemic administration, estimated at ~ 42 #M 1 h after a single dose of 80 mg/kg, i.p. These results indicate that mexiletine is able to penetrate into the CNS at concentrations sufficient to confer significant protection. Na ÷ channel blockers such as mexiletine may prove to be effective clinical therapeutic agents for protecting CNS white matter tracts against anoxic/ischemic injury. Key words: axon, anoxia, Na + channel, local anesthetic, supernormal period, recovery cycle.

Protection of functional integrity of myelinated axons within the white matter of brain and spinal cord after stroke or spinal cord injury is essential for maximizing clinical recovery from these disorders. Whereas numerous studies have been performed to explore mechanisms of anoxic/ischemic damage to neuronal cell bodies, only recently has detailed information become available regarding this type of injury in m a m m a l i a n CNS white matter. Using the in vitro rat optic nerve as a model of white matter anoxia, Stys et al. have shown that Ca 2+ influx into axons plays a key role in mediating irreversible injury, 25 and that the majority of the deleterious Ca 2+ influx occurs via reverse operation of the N a * - C a 2+ exchanger. 28 Central to this hypothesis is the requirement for an increase in axoplasmic [Na +] and membrane depolarization, factors that will drive the N a + - C a 2+ exchanger to operate in the Ca 2+ import mode. The presumed ionic shifts were recently confirmed directly, ~3 with the N a + influx mediated largely by Na ÷ channels. 28 Thus, blockers of Na ÷ channels such as tetrodotoxin (TTX), 2s local anesthetics, 24 antiarrhythmics 21 and certain anticonvulsants l° have

been shown to be highly protective against white matter anoxic injury in vitro. However, many of these agents are either very toxic (TTX), or it is not known whether the required concentrations, identified from in vitro studies, would be tolerated systemically. Moreover, an additional complication involves adequate CNS penetration even if the previous constraints are satisfied. Based on these earlier in vitro studies, we therefore undertook investigations to identify a compound, from the class of use-dependent N a + channel blockers, that would be well tolerated, and be able to penetrate the blood-brain barrier sufficiently to protect white matter from anoxic/ischemic injury.

EXPERIMENTALPROCEDURES In vitro optic nerve recordings Adult Long-Evans rats aged 50-70 days were

anesthetized with 20% 02/80% CO 2 and decapitated. Optic nerves were dissected free, placed in a modified interface recording chamber (Medical Systems Corp., Greenvale, NY) and perfused with artificial cerebrospinal fluid (aCSF) at 2ml/min (in mM: NaCI 126, KCI 3.0, MgSO4 2.0, NaHCO 3 26, NaH2PO 4 1.25, CaCI 2 2.0, dextrose 10, pH 7.45, 37.0___0.2°C). The tissue was aerated with a 95% 02/5% CO 2 gas mixture. Nerves were incubated for approximately 90 min before control measurements were begun to allow stabilization of axoplasmic electrolyte content t3 and

*To whom correspondence should be addressed. Abbreviations: aCSF, artificial cerebrospinal fluid; CAP,

compound action potential; TTX, tetrodotoxin. 27

28

P. K. Stys and H. Lesiuk

Recovery cycles

electrophysiological properties. 23 The functional integrity of the nerves was monitored electrophysiologically with suction electrodes. ]9,2~Square (100 # s) constant voltage stimulus pulses were applied and the 25% supramaximal intensity (i.e. stimulus intensity 25% greater than that just required to elicit a maximal response) determined individually for each nerve. The responses were amplified and filtered (DC to 10kHz) with a CyberAmp 380 signal conditioner (Axon Instruments, Foster City, CA), digitized at 500 kHz and 12 bits vertical resolution (MacADIOS II with high-speed A/D daughterboard, G W Instruments, Somerville, MA), stored and analysed using WaveTrak data acquisition software 2° on an Apple Macintosh Quadra 700.

Myelinated axons undergo a predictable oscillation of excitability and conduction velocity following an action potential, that outlasts the refractory periods by hundreds of milliseconds?7 Expressed as a function of the time following the conditioning volley, these changes are termed the recovery cycle of an axon. Optic nerve recovery cycles were determined by delivering pairs of stimuli, both at 25% supramaximal intensity (the same supramaximal intensity as used for single volleys, see above), at interstimulus intervals ('tab', see insets, Fig. 1) varying between 1.5 and 800 ms, and precisely determining the peak latencies of the first, conditioning response (a) and second, test response (b) at

A. 10 8 6

-

2

--~-~-~--

0

42_

.

,Cr

',,

tab

~,

_~ -10 ~

-12

~

-14

0-

-16 -18

-20

a

b

-22

o

B.

~;

~o

~o

,o

~o

o; ~o ~;

~o

,00.'0

~o~o~,;o~o

30 25 20

~ " 15 ,.-,t 10 "

~,

5

~

0

I II

. ~ ~---~..-=

"

•

.g -5 ~ -lo ~

g2b

-15

-20 -25 n -30

17

-35

--

/

a

b

-40

0

10

20

30

40

50

60

70

80

90

100 110

120

130

140

150

tab (ms) Fig. I. The characteristic rat optic nerve compound action potential has three peaks (insets). Pairs of stimuli were delivered to optic nerves at varying interstimulus intervals (tab, ms). The first response (a) of a pair conditioned the axons so that the second response (b) was conducted at a different velocity. Only the first two peaks were studied, and the difference between peak latencies (the latency shift) of response a and b (see insets) was calculated at each tab. A plot of the latency shifts vs t,b defines the recovery cycle, and is shown separately for peak I (panel A) and peak 2 (panel B). Positive latency shifts indicate that the peak in the second response (b) of a pair was conducted more slowly, and negative shifts imply supernormal conduction. At very short tab, the relative refractory period was seen, followed by a supernormal period extending to ~ 30 ms for peak 1 and ~ 40 ms for peak 2. Peak 1 exhibited a minimal subnormal phase at 40 ms, followed by a subtle and prolonged (to ~ 500 ms, not shown) second supernormal period. Peak 2 had subnormal conduction until ~ 800 ms. Solid squares represent the mean of 154 optic nerve recovery cycles, with one standard deviation about the mean shown by dashed lines.

Mexiletine and white matter ischemia various tab. Typically, eight pairs of compound action potentials (CAPs) were averaged at each tab. Averaged waves were then processed with an interpolating nonrecursive digital filter (pass-band: 0-12kHz, stop-band: > 50kHz, minimum stop-band attenuation: 150dB) to remove any residual noise that may result in detection of spurious peaks and to increase the sampling resolution eight-fold from 2 # s to 250 ns, allowing very precise determination of peak positions in time. Non-recursive digital filters were chosen because they exhibit linear phase characteristics that minimize changes to the shape of the waveforms that would artifactually shift the positions of the peaks. 8 Peak latencies were computed from stimulus onset and the difference in latency of corresponding peaks in response b compared to response a was expressed as a latency shift in microseconds. A plot of latency shifts vs tab is defined as a recovery cycle (see Fig. 1). In vitro anoxia Control CAPs were recorded after a 90 min equilibration period following dissection. Anoxia was induced for a standard 60rain period by switching to 95% N2/5% CO 2 ambient gas delivered at 4 l/rain. Perfusing solutions were continuously bubbled with 95% N2/5% CO 2 to minimize dissolved 02 during the anoxic period. Nerves were re-oxygenated for 60 rain after which post-anoxic readings were collected. In experiments with mexiletine5, the drug was applied for 60 rain before anoxia, and continued during the anoxic period. Wash began at the time of re-oxygenation. Longer periods of re-oxygenation were sometimes required to wash out residual mexiletine at higher concentrations. CAP area was used as the most reliable and reproducible indicator of nerve functionY '33 Ratios of CAP areas before and after anoxia, mathematically corrected for electrode resistance changes, ~9'23 were calculated to determine the degree of recovery of optic nerve function. CAP areas were also determined after 60 min of mexiletine application, before anoxia, to assess the effects of the drug alone on optic nerve excitability. In situ optic nerve ischemia To study any protective effects of Na ÷ channel blocking agents systemically administered in vivo, a model was developed where a controlled ischemic injury can be delivered to the optic nerve, while still allowing the quantitative and reliable recordings offered by suction electrodes in vitro. Rats were anesthetized with 20% 02/80% CO2 and decapitated. The head was completely immersed in a water bath maintained at 37.0 ___0.3°C for a set time (30-120 min). Optic nerves were then dissected free, and allowed to equilibrate for 90 min in the in vitro recording chamber as described above, after which responses were recorded. Mexiletine was dissolved in sterile water and administered by intraperitoneal injection (typically 1 ml of solution) 30-60 min before decapitation to allow drug absorption and distribution into the CNS. In some rats, recovery cycles were recorded after mexiletine administration without ischemia by injecting the drug i.p., killing the animal 1 h later to remove the optic nerves, and recording immediately without a 90 min equilibration period which would cause washout of any drug that was absorbed from the circulation. Errors are reported as standard deviations, and statistical significance was calculated using ANOVA with Dunnett's post hoc test. Materials" All compounds were from Sigma except mexiletine-HC1 which was a generous gift from Boehringer-Ingelheim.

29 RESULTS

Recovery cycles in normal optic nerves and effects o f mexiletine Intrinsic alterations of conduction velocity of rat optic nerve axons following a single conditioning action potential (the recovery cycle) were measured in vitro and are shown in Fig. 1. The differences in peak latency between the two responses of a pair (latency shift) are plotted against the interval between stimuli (tab) f o r peak 1 (faster axons; panel A) and peak 2 (slower axons; panel B). Positive latency shifts indicate that the corresponding peak was slower in the second response of the pair, whereas negative values reflect faster conduction. The third peak of the characteristic triphasic optic nerve response (see insets, Fig. 1) was often broad, making it difficult to accurately detect its postion, resulting in unreliable recovery cycles, and was therefore not routinely analysed. The graphs represent the mean recovery cycle ( _+ 1 S.D., dashed curves) of peaks 1 and 2 from 154 normal nerves. With closely spaced stimuli (tab < 1.5 ms for peak 1, and < 3 ms for peak 2) a relative refractory period was evident with the second response conducted more slowly than the conditioning volley (positive latency shifts). At longer interstimulus intervals (to ~ 30 ms for peak 1 and 40 ms for peak 2) axons entered a supernormal period where conduction velocity was greater in the second response of a pair (peak latency was shorter and therefore shifts were negative) than the first, control CAP. Peak 2 exhibited a subtle but prolonged phase of subnormality following its supernormal period which lasted up to 800 ms. Peak 1 exhibited a minimal subnormal phase at 40 ms, then entered into a second more subtle supernormal period lasting until 500 ms. tab greater than 800 ms were not explored therefore it is not known whether peak 2 also exhibited a second supernormality. The effects of various concentrations of mexiletine on optic nerve recovery cycles are shown in Fig. 2. Curves in the absence of drug (control) are taken from Fig. 1 and drawn for reference. Mexiletine induced a subnormal period in both peaks. The panels on the right are curves with the control recovery cycle subtracted, to show the net effect of paired stimulation in the presence of mexiletine. The subnormality induced by this agent decays in an exponential manner over 150-200 ms. This effect was concentration-dependent, and at 100 # M or greater, the entire curve was subnormal indicating that peaks in the second response of a pair propagated more slowly than in the first at all tab; although supernormality was abolished in absolute terms at higher concentrations, a relative supernormal period was still evident for both peaks (left panels), though the depth of the supernormality was somewhat reduced at higher concentrations. This suggests that the effect of mexiletine was additive to, and did not significantly interfere with, the intrinsic mechanisms of

P. K. Stys and H. Lesiuk

30 50

50

40

40

30

30

20

20

10

10

at=

-I0

-10

-20 [~cont_l 0

20

, 40

,

6b

80

,

1~0

120

-20

1~0

0

20

40

e0

a0

100

1~0

40

1~o

40

80 70 60 50 40 30

o~

20 10

(1) 13-

••/00

70 80 60

~M

50 40 30 20

0 -10

-10

-20

-20

-30 -40

control

2'0

-30 -40

&

8'0

e~

1;o

~o

1;~o

t a b (ms)

3

Lit-"

2'0

&

6'0

8'0 40 tab (ms)

Fig. 2. Recovery cycles for peak 1 (top graphs) and peak 2 (bottom graphs) in normal aCSF ("control") and in the presence of mexiletine. As the concentration of the drug was increased, a progressivelygreater subnormal period was seen for both peaks, although peak 2 was affected more in absolute terms (note difference in y-axis scales). The graphs on the right show the effects of mexiletine in isolation after subtraction of the control curves. The subnormal effect developed rapidly at very short interstimulus intervals, as evidenced by a rapid rise of the curves, indicating rapid binding of mexiletine to the Na ÷ channel. This was followed by a more gradual decay over 100-150ms, reflectinga much slower unbinding of the drug from the channel.

activity-dependent modulation of conduction velocity in these axons. The subnormality was more pronounced in absolute terms for peak 2 at any given concentration. These recovery cycles acquired in vitro in the presence of known concentrations of mexiletine can be used to estimate the tissue concentrations of drug after in vivo administration (see below). Optic nerve anoxia in vitro: protective effects o f mexiletine

Figure 3 (upper panels) shows representative CAPs following a single stimulus in normal aCSF, after 1 h of mexiletine application in vitro before anoxia, and 1 to 3 h after 60 min of anoxia (mexiletine was continued throughout the anoxic exposure, and the nerves were washed during the re-oxygenation period). The bar graph quantitatively shows the effects of the drug on CAP areas before anoxia, and the protective effects following the anoxic challenge, expressed as a percentage of control CAP area (before drug/anoxia). In normal aCSF without mexiletine, the mean CAP area recovered to 21.3 + 9.9% of control area. There was a dose-dependent reduction in the size of optic nerve CAPs at increasing concentrations of mexiletine: at 10/~ M the pre-anoxic

CAP area was virtually unaffected, although peak latencies were slightly prolonged. With higher concentrations, the size of the response was progressively reduced, and almost completely abolished at 1 mM. The degree of postanoxic recovery was also concentration dependent, and was greatest at 100/~M (52.9 4- 18% vs 21.3 + 9.9%, P < 0.0001). At higher concentrations (300 #M and 1 mM), CAP recovery was less than at 100 #M, in spite of 2 or 3 h of wash, probably due to residual anesthetic effects from incomplete wash-out of the drug. In situ optic nerve ischemia While the in vitro anoxic model is useful for quantitative measurement of optic nerve injury and protective pharmacology, the in situ optic nerve ischemia model (see Experimental Procedures) was developed to better approximate ischemic injury to white matter in the clinical setting, such as in stroke or spinal cord injury. Figure 4 shows representative CAPs from optic nerves maintained under ischemic conditions in situ for 30-120min at 37°C, then allowed to recover for 60-90rain in the in vitro chamber under normoxic conditions. Because control readings prior to injury cannot be obtained with this

31

Mexiletine and white matter ischemia mexiletine 1 mM

mexiletine 100 p,M

mexiletine 30 ~tM

normal CSF

control ash ~1 mV 1 ms

\ lh post-anoxia/wash

1h post-anoxiaJwash

post-anoxia

\ lh in mex

8

100

8

[] []

l h in mex / pre-anoxia 1-3h post-anoxia / wash

80 'i!!:'i! i!!:!': ~i!~!i!~I~: ~ ~!~

n < 0

60 !~!:i!:i!!i!i~:i~i~

iii!ii'ii~ii~i!~'i cO 0

iiiiiiiiiii!ili

40

ii:i!!~!i!ili!!iii!~iii!i i!!i!!~!iiii~!iiiii....

o~

iii~i~ii!iii!ii

20

0

control

10 I.tM

30 I.tM

100 t.tM

300 I~M

1 mM

mexiletine concentration

Fig. 3. Effects of mexiletine on pre-anoxic optic nerve CAPs, and protective effects after 60 min of anoxia. Top panels show representative CAPs in normal aCSF ("control") and in the presence of mexiletine. At increasing concentrations of the drug, peak latencies were increasingly prolonged compared with control (e.g. 30 and 100 #M). CAP area was reduced beginning at 100 #M, and the pre-anoxic response almost completely abolished at I mM. Higher concentrations were also more protective against a 60 min anoxic challenge. The bar graph shows these results quantitatively. Lightly stippled bars represent mean CAP area after 1 h of exposure to mexiletine expressed as a percentage of control CAP area. Dark bars show the percentage of CAP area recovery after 60 min of anoxia and 1-3 h of re-oxygenation/wash. As the mexiletine concentration was increased, pre-anoxic CAP magnitude was reduced and a parallel increase of post-anoxic recovery was observed (*P < 0.05, **P < 0.0005, ***P < 0.0001 compared to control CAP area recovery; n in each treatment group is shown above bars).

model, only raw C A P areas are shown in the graph (left axis), rather than the ratio of area compared to its own control as with the in vitro recordings. These values were compared to a set of control C A P areas ("0 m i n " ischemia) obtained from other experiments, and the percentages with respect to control are shown on the right axis. Thirty minutes of ischemia had virtually no effect on mean C A P area (reduced to 96% of control), but the shape of the waveform became broader. After 60 min of ischemia mean C A P area was reduced by over one half compared to control, and injury increased with the time of ischemic exposure in an exponential manner (time constant 35.7 min), so that after 120min very little activity remained (CAP area reduced to 7% of control).

Effects of mexiletine after in vivo admin&trat[on Before testing mexiletine's ability to protect white matter after systemic administration, it was necessary

to assess this agent's ability, if any, to penetrate into the CNS at doses tolerated by the animals. The effects on optic nerve recovery cycles provided a convenient and reliable measure of the tissue concentration. Rats were administered 80 mg/kg of mexiletine intraperitoneally, and left for 30-60 min to allow the drug to be absorbed and distributed. Nerves were dissected out, perfused with normal a C S F in the in vitro recording chamber, and recovery cycles acquired immediately (without the usual 60-90rain equilibration period) to minimize drug wash-out from the tissue. Out of four rats studied in this manner, three were sluggish and obtunded after 10-20 rain of injection, whereas one animal showed no behavioral effects. The optic nerve recovery cycles from the three obtunded rats were similar and the mean tracings from these six nerves are shown as solid squares in Fig. 5. A clear subnormal period was observed for peak 1 (arrow, panel A) and peak 2 (arrow, panel B). In contrast, the mean recovery cycle from two nerves taken from the rat that was unaffected by the drug

32

P. K. Stys and H. Lesiuk control

60 rain

120 min

90 min

~

1 mV

1 ms

8

60-

120

50- n i ~

...............................................

1O0

f) "o

tfi E

g t~ t~ o_ < (O

40-

80

30-

60 18

20-

18

20

40 20

100

o 8

0 I

I

I

I

I

I

I

0

20

40

60

80

100

120

Ischemia time (min)

Fig. 4. CAP area recovery as a function of time of/n situ ischemia. Optic nerves were subjected to various periods of ischemia in situ followed by recovery in vitro (see Experimental Procedures). Representative CAPs are shown in the upper panels, and the mean absolute CAP areas plotted against ischemia time in the graph. Although 30 min of ischemia did not depress the CAP area, the shape of the response was altered suggesting a disturbance of conduction velocity of constituent axons. Drop-out of conducting fibres was evident with periods of ischemia greater than 30 min as indicated by a progressively reduced mean CAP area compared with control. The degree of injury developed in an exponential manner after this initial 30 min latent period, with a time constant of approximately 35 min. The right axis represents percent CAP area recovery compared to control ("0 min of ischemia", dashed line; numbers above symbols represent number of nerves at each time).

injection appeared normal, with no evidence of a subnormal period. Therefore, although our sample size was small, there appears to be a correlation between clinical effects of mexiletine and this agent's modulation of optic nerve recovery cycles. Panels C and D show the mean recovery cycle from the six affected nerves superimposed on recovery cycles from nerves perfused with known concentrations of mexiletine in vitro (see Fig. 2). N o t e that the in vitro curves were obtained from nerves that were equilibrated in the recording chamber for 90 min, whereas the curves acquired after systemic mexiletine administration were measured immediately after dissection to minimize drug wash-out. Control recovery cycles obtained immediately after dissection without an equilibration period in the chamber revealed a narrower supernormal period, especially for peak 1, but no significant subnormality (results not shown). Therefore the clear subnormal phases seen with immediate recordings in Fig. 5 were due to mexiletine and were not an artefact of early recording after dissection. The in vitro data can then serve as a set of calibration curves to determine the tissue concentration of mexiletine after systemic administration, which was estimated at between 30 and 100/~M (see Discussion and Fig. 7).

Protective effect o f mexiletine after in vivo administration The previous results indicated that C N S concentrations of mexiletine after intraperitoneal administration may reach l e v e l s high enough to confer significant protection to optic nerves. This agent was therefore tested as a neuroprotectant in the in situ ischemia model. After intraperitoneal injection of 80 mg/kg of mexiletine, rats were allowed 60 min for drug absorption before being killed. As mentioned above most animals appeared sluggish and obtunded for 10-20 min after injection, with occasional motor spasms noted. One of 22 animals died 20 min after injection. Ischemia was maintained for 60 to 120 min, the nerves removed, and allowed to equilibrate in vitro before C A P measurements were taken. Means of post-ischemic C A P areas without ("control", taken from Fig. 4 for comparison) and with mexiletine injection are summarized in Fig. 6. Mexiletine significantly improved outcome for all ischemic periods; with 60 min of ischemia, mean C A P area increased from control of 20.9 to 30.9 with mexiletine, an improvement of 48% (P < 0.001). With longer ischemic exposure (75 min), C A P area doubled (control C A P area of 12.0 vs 23.5 after mexiletine, 96% increase, P < 0.0002). With still longer ischemia, the degree of protection diminished in absolute terms,

r

Mexiletine and white matter ischemia

Peak 1 latency shifts

33

Peak 2 latency shifts

A.

B. 10

40 t

.

~

1,1~ .10 ] 1 ~

-

7.

- O - 1 rat wd~out clinical effects - - I - 3 rats with clinical effects

l

S

0

40

--O- 1 rat without clinical effects - I - 3 rats with clinical effects

~ 80

' 120

160

I 40

200

i 80

•

i 120

tab (ms)

C.

i ' i ' i ' i 160 200 240 280 tab (ms)

D.

2O

6050403020-

5

10o

-

c ~

0

~

0 -10 -

-

0

-20 -

-10

-30 -

-15 I

0

•

l

I

I

I

I

40

80

120

160

200

-40 I

0

'

I

40

'

I

80

tab (ms)

'

I

'

120

I

160

I ' I * I 200 240 280

tab (ms)

Fig. 5. Effects of systemic mexiletine on recovery cycles of peaks 1 and 2 in optic nerve CAPs. Four rats were given 80 mg/kg of mexiletine i.p. and optic nerves removed 30-60 min later for electrophysiological study. Three of four animals displayed side-effects from the drug in the form of obtundation; optic nerve recovery cycles (mean of six nerves) from these three rats (black squares) exhibited a prominent subnormal period (arrows, panels A and B) induced by mexiletine, whereas the curves from the unaffected animal (mean of two nerves, open circles) appeared relatively normal. Curves from the top panels are shown superimposed on recovery cycles obtained in vitro in known concentrations of mexiletine in the perfusate (data from Fig. 2). From the size of the subnormal period, the tissue concentration of mexiletine after systemic administration was estimated at between 30 and 100/~M (see Discussion).

though relative improvement was robust, even with prolonged insults (e.g. 71% CAP area increase at 120 min ischemia). DISCUSSION

The in vitro rat optic nerve model 14'25 is a wellcharacterized system ideally suited for the study of white matter injury. Recent work has greatly increased our understanding of the sequence of events leading to anoxic/ischemic white matter damage: energy failure followed by axonal membrane depolarization and Na + influx causes reverse operation of the N a + - C a 2+ exchange system which results in axonal Ca 2+ overload and irreversible Ca2+-mediated structural injury. ~3'25'3° Blockade of axonal Na + conductance interrupts this cascade, and is highly protectivein vitro using a variety of "use-dependent" or "phasic" (i.e. whose channel blocking potency increases with the frequency of activation) Na + channel blockers such as local anesthetics, 24 anticonvulsants j° and an-

tiarrhythmic compounds. 21 However, these encouraging in vitro results cannot be directly extrapolated to the in vivo situation, where issues such as bioavailability, metabolism, toxicity and CNS penetration may conspire to limit the efficacy of promising agents. The in situ optic nerve ischemic model and measurement of axonal recovery cycles described here were developed to assess the CNS penetration and efficacy of potentially protective compounds in a model which more closely reflects the clinical situation. Optic nerve recovery cycle

Myelinated and unmyelinated axons undergo a predictable change in excitability and velocity of action potential propagation following even a single volley. In addition to the well-characterized refractory periods, many fibres exhibit a later increase in excitability and conduction velocity (the supernormal period), 2A5'32followed in some axons by a late subnormality. H,z2This activity-dependent oscillation of fibre excitability, termed the recovery cycle, can be mapped

34

P.K. Stys and H. Lesiuk 4

35 30 25 o5 E

,=80 mg/kg i.p.

20

E

v

15 cO ft.

< o

10 5 0 60

75

90

120

Ischemia time (min)

Fig. 6. Means of absolute CAP areas plotted against time of ischemic exposure, without (lightly stippled bars) and following i.p. administration of mexiletine 80 mg/kg (dark bars). Although the absolute size of the CAP diminished with longer ischemic exposure in both control and treatment groups, mexiletine significantly increased the degree of recovery of mean CAP area for all ischemic periods. At 75 min, for instance, CAP area was improved by 96% with mexiletine treatment (control CAP area 11.98 vs 23.52 after mexiletine). *P < 0.05, **P < 0.001, ***P < 0.0002 compared to control ischemia. Numbers above bars represent number of nerves in each group.

using pairs of stimuli delivered at varying intervals as shown in Figs I and 2, and represents a distinct 6050"~

40-

g --~ 30~

20-

~

10-

O-

:: 0

5

0

i

1O0

i 150

i 200

i

250

i 300

[mexiletine] (pM)

Fig. 7. The magnitude of the subnormal period increased with higher concentrations of mexiletine (see Fig. 2). Graph shows maximal subnormal latency shift plotted against mexiletine concentration for peak 1 (open circles) and peak 2 (black squares) recorded in vitro. There is a highly significant linear relationship (r 2 =0.994 and 0.998 for peaks 1 and 2, respectively) over the concentration range tested (0-300~M). The linear interpolation was used to estimate more accurately the tissue concentration of mexiletine from the maximal subnormal latency shift (mean maximal shift was 5.8#s and 25.7#s for peaks 1 and 2, obtained from data in Fig. 5, black squares) produced after systemic administration. Independent estimates from peak 1 and peak 2 subnormal periods agree closely at 41 and 44pM, respectively (dashed lines).

signature c o m m o n to axons of similar type. Although the precise mechanisms underlying the recovery cycle are not fully understood (see Stys and W a x m a n 2v for review), this technique can be used as a sensitive measure of pharmacological "use-dependent" Na + channel block; Fig. 2 illustrates how increasing concentrations of mexiletine produced a progressively greater subnormal period, reflecting activitydependent slowing of the second action potential of a pair. The panels on the right of Fig. 2 show the effects of mexiletine on recovery cycles in isolation (after subtraction of the native recovery cycle without drug), illustrating the activity-dependent inhibitory effects of this agent which develop very quickly following a single conditioning volley, and decay more slowly over 100-150ms. The slower second peak exhibits a greater effect in absolute terms than the faster fibres in peak 1. This induction of subnormality is shared by many local anesthetic-type compounds such as lidocaine ~6 and its quaternary analogue QX-314 (Stys, unpublished data), and was used as an electrophysiological marker to quantitate the concentration of drug present in the optic nerve after systemic administration. There is a highly linear relationship between the maximal subnormal shift (estimated individually for peaks 1 and 2 from the left graphs in Fig. 2) and mexiletine concentration in vitro as illustrated in Fig. 7. By measuring the degree of subnormality after systemic administration of mexiletine (Fig. 5), the

Mexiletine and white matter ischemia optic nerve concentration was estimated at 41/~M from peak 1 shift, and 4 4 # M from peak 2 shift (dashed lines, Fig. 7). Importantly, these calculations indicate that mexiletine was absorbed into the circulation and was able to penetrate the blood-brain barrier to reach concentrations at its target (i.e. the Na + channel) potentially sufficient to protect white matter from anoxia/ischemia. Other compounds with properties similar to mexiletine, such as QX-31424 and prajmaline 2~ have been shown to be highly protective against white matter anoxia in vitro, but failed to penetrate into the CNS as judged by a lack of effect on optic nerve recovery cycles (Stys, unpublished data), and would be likely to be ineffective against ischemic injury. Equally importantly, this technique may prove to be a powerful tool for assessing the pharmacokinetics of "use-dependent" Na + channel blockers at the membrane level, when assays of whole tissue or serum concentration may be misleading if drug distribution from the vascular compartment is impeded, as is often the case in the CNS. Optic nerve ischem&

Although 30 min of ischemia altered the shape of the CAP, the area remained very close to control (Fig. 4), suggesting that conduction velocities of constituent axons were altered, but that most fibres were able to conduct action potentials, at least at low frequencies. Slowing of conduction velocity could be due to early damage to the myelin sheath, such as separation of the paranodal junctions, 3~ changes in the properties of axolemmal ionic conductances, or the ionic composition of the axoplasm. ~3 After this 30 rain latent period, irreversible injury developed rapidly in an exponential manner, with a time constant of about 35 min. This time course may have important implications for the initiation of therapy in stroke and spinal cord injury (where white matter tracts may be damaged to a significant extent by ischemiag,34). Interestingly, optic nerves sustained greater damage by 60 min of in vitro anoxia alone than by 60 min of in situ ischemia (where presumably both oxygen and glucose were lacking). This is most likely to be due to an unlimited source of external Na ÷ and Ca z÷ in vitro, compared to a restricted amount available in a comparatively small extracellular space which is not replenished in the ischemic state in situ. Also, any 0 2 accumulated in trapped red cells in the in situ model could be utilized rather than being rapidly removed by a large flow of N 2 in an in vitro chamber. The important finding in this study is the ability of mexiletine to significantly protect optic nerves from

35

ischemic injury after systemic administration. Although the electrophysiological function [as judged by absolute CAP area (see Fig. 6) which closely reflects the number of viable axons capable of conduction 3'33] of drug-treated nerves was still below non-ischemic controls, the relationship between the number of remaining healthy axons in a tract and clinical function is highly non-linear. 1 Therefore, it is possible that rescuing even a modest absolute number of fibres with an agent such as mexiletine in spinal cord injured patients, for instance, could result in very significant clinical improvement. Moreover, the ability of mexiletine to protect optic nerves from in vitro anoxia is more modest when compared with other use-dependent Na ÷ channel blockers such as QX-31424 and prajmaline. 2~ While the detailed molecular mechanisms of channel block by charged and uncharged anesthetics is quite complex, 4 it has been suggested that the permanently charged quaternary amine structure of QX-314 and prajmaline might render these more selective at open Na ÷ channels 1237A8 and probably for this reason, result in a compound that is more protective with less depression of the pre-anoxic CAP. 2~'24 The permanently charged form probably also prevents these agents from penetrating into the CNS. Mexiletine is a primary amine with a p K a of 8.4, and exists in both neutral and protonated forms at physiological pH. We hypothesize that the former is able to cross the blood-brain barrier while the latter, charged form (which will increase in proportion under acidotic conditions during ischemia) is more potent at blocking open (possibly a non-inactivating subtype 26) Na + channels and interrupting the anoxic cascade. 28 CONCLUSIONS

While the present results demonstrate the feasibility of protecting CNS white matter by systemic administration of "use-dependent" Na + channel blockers, it is very likely that similar agents could be developed with the CNS bioavailability of mexiletine and the greater efficacy of a quaternary amine such as QX-314. Moreover, it is possible that Na ÷ channel blockers could prove useful adjuncts in treating gray matter ischemic injury, which may involve Na + loading and subsequent reverse N a + - C a 2÷ exchange, excitotoxicity from glutamate efflux mediated by reverse Na+-glutamate co-transport, 29 and/or Na ÷dependent cytotoxic edema. 6 Acknowledgements--The authors would like to thank James

Surowiak for his excellent technical assistance, and Andre Dabrowski, Ellen Lehning and Michael Todd for statistical advice.

REFERENCES

1. Blight A. R. (1983) Cellular morphology of chronic spinal cord injury in the cat: analysis of myelinated axons by line-sampling. Neuroscience 10, 521-543. 2. Bostock H. and Grafe P. (1985) Activity-dependent excitability changes in normal and demyelinated rat spinal root axons. J. Physiol., Lond. 365, 239-257.

36

P. K. Stys and H. Lesiuk

3. Buchthal F. and Rosenfalck A. (1966) Evoked action potentials and conduction velocity in human sensory nerves. Brain Res. 3, 1-122. 4. Butterworth J. F. IV and Strichartz G. R. (1990) Molecular mechanisms of local anesthesia: a review. Anesthesiology 72, 711-734. 5. Campbell T. J. (1989) Subclassification of class I antiarrhythmic drugs. In Antiarrhythmic Drugs (ed. Vaughn Williams E. M.), pp. 135-155. Springer, Berlin. 6. Choi D. W. (1987) Ionic dependence of glutamate neurotoxicity. J. Neurosci. 7, 369-379. 7. Cummins K. L., Perkel D. H. and Dorfman L. J. (1979) Nerve fiber conduction velocity distributions: I. Estimation based on the single-fiber and compound action potentials. Electroenceph. clin. NeurophysioL 46, 634-646. 8. Embree P. M. and Kimble B. (1991) C Language Algorithms for Digital Signal Processing, pp. 19-22. Prentice-Hall, Englewood Cliffs, NJ. 9. Fehlings M. G., Tator C. H. and Linden R. D. (1989) The relationships among the severity of spinal cord injury, motor and somatosensory evoked potentials and spinal cord blood flow. Electroenceph. clin. Neurophysiol. 74, 241-259. 10. Fern R., Ransom B., Stys P. K. and Waxman S. G. (1993) Pharmacological protection of CNS white matter during anoxia: actions of phenytoin, carbamazepine and diazepam. J. Pharmac. exp. Ther. 266, 1549-1555. 1 I. Gasser H. S. and Grundfest H. (1936) Action and excitability in mammalian A fibers. Am. J. Physiol. 117, 113. 12. Khodorov B. I. and Zaborovskaya L. D. (1983) Blockade of sodium and potassium channels in the node of Ranvier by ajmaline and N-propyl ajmaline. Gen. Physiol. Biophys. 2, 233-268. 13. LoPachin R. M. Jr and Stys P. K. (1995) Elemental composition and water content of rat optic nerve myelinated axons and glial cells: effects of in vitro anoxia and reoxygenation. J. Neurosci. 15, 6735-6746. 14. Ransom B. R., Stys P. K. and Waxman S. G. (1990) The pathophysiology of anoxia in mammalian white matter. Stroke 21, 52-57. 15. Raymond S. A. (1979) Effects of nerve impulses on threshold of frog sciatic nerve. J. Physiol., Lond. 290, 273-303. 16. Raymond S. A. (1992) Subblocking concentrations of local anesthetics: effects on impulse generation and conduction in single myelinated sciatic nerve axons in frog. Anesth. Analg. 75, 906-921. 17. Stafstrom C. E., Schwindt P. C., Chubb M. C. and Crill W. E. (1985) Properties of persistent sodium conductance and calcium conductance of layer V neurons from cat sensorimotor cortex. J. Neurophysiol. 53, 153-170. 18. Strichartz G. R. (1973) The inhibition of sodium currents in myelinated nerve by quaternary derivatives of lidocaine. J. Gen. Physiol. 62, 37 57. 19. Stys P. K. (1992) Suction electrode recording from nerves and fiber tracts. In Practical Electrophysiological Methods: A Guide for In Vitro Studies in Vertebrate Neurobiology (eds Kettenman H. and Grantyn R.), pp. 189 194. John Wiley & Sons, New York. 20. Stys P. K. (1994) WaveTrak: a data acquisition system and waveform database for the Macintosh. Sci. Comput. Automation 10, 19-24. 21. Stys P. K. (1995) Protective effects of antiarrhythmic agents against anoxic injury in CNS white matter. J. cerebr. Blood Flow Metab. 15, 425-432. 22. Stys P. K. and Ashby P. (1990) An automated technique for measuring the recovery cycle of human nerves. Muscle Nerve 13, 750-758. 23. Stys P. K., Ransom B. R. and Waxman S. G. (1991) Compound action potential of nerve recorded by suction electrode: a theoretical and experimental analysis. Brain Res. 546, 18--32. 24. Stys P. K., Ransom B. R. and Waxman S. G. (1992) Tertiary and quaternary local anesthetics protect CNS white matter from anoxic injury at concentrations that do not block excitability. J. Neurophysiol. 67, 236-240. 25. Stys P. K., Ransom B. R., Waxman S. G. and Davis P. K. (1990) Role of extracellular calcium in anoxic injury of mammalian central white matter. Proc. natn. Acad. Sci. U.S.A. 87, 4212-4216. 26. Stys P. K., Sontheimer H., Ransom B. R. and Waxman S. G. (1993) Non-inactivating, TTX-sensitive Na ÷ conductance in rat optic nerve axons. Proc. natn. Acad. Sci. U.S.A. 90, 697645980. 27. Stys P. K. and Waxman S. G. (1994) Activity-dependent modulation of excitability: implications for axonal physiology and pathophysiology. Muscle Nerve 17, 969-974. 28. Stys P. K., Waxman S. G. and Ransom B. R. (1992) Ionic mechanisms of anoxic injury in mammalian CNS white matter: role of Na ÷ channels and Na+-Ca 2÷ exchanger. J. Neurosci. 12, 430-439. 29. Taylor C. P., Geer J. J. and Burke S. P. (1992) Endogenous extracellular glutamate accumulation in rat neocortical cultures by reversal of the transmembrane sodium gradient. Neurosci. Lett. 145, 197-200. 30. Waxman S. G., Black J. A., Ransom B. R. and Stys P. K. (1993) Protection of the axonal cytoskeleton in anoxic optic nerve by decreased extracellular calcium. Brain Res. 614, 137-145. 31. Waxman S. G., Black J. A., Stys P. K. and Ransom B. R. (1992) Ultrastructural concomitants of anoxic injury and early post-anoxic recovery in rat optic nerve. Brain Res. 574, 105-119. 32. Waxman S. G. and Swadlow H. A. (1976) Morphology and physiology of visual callosal axons: evidence for a supernormal period in central myelinated axons. Brain Res. 113, 179-187. 33. Wijesinghe R. S., Gielen F. L. H. and Wikswo J. P. Jr (1991) A model for compound action potentials and currents in a nerve bundle I: the forward calculation. Ann. biomed. Engng 19, 43-72. 34. Young W. (1987) The post-injury responses in trauma and ischemia: secondary injury or protective mechanisms? Cent. Nerv. Syst. Trauma 4, 27-51. (Accepted 2 August 1995)