283

Brain Research, 592 (1992) 283-297 © 1992 Elsevier Science Publishers B.V. All rights reserved 0006-8993/92/$05,00

BRES 18162

Three types of sodium channels in adult rat dorsal root ganglion neurons J.M. Caffrey, D.L. Eng, J.A. Black, S.G. W a x m a n a n d J.D. Kocsis Department of Neurology, Yale University School of Medicine, New Haven, CT 06510 (USA) and Neuroscience Research Center, Department of Veteran'sAffairs Medical Center, West Haven, CT 06516 (USA) (Accepted 19 May 1992)

Key words: Sodium channel; Dorsal root ganglion; TI'X; Voltage clamp

Several types of Na + currents have previously been demonstrated in dorsal root ganglion (DRG) neurons isolated from neonatal rats, but their expression in adult neurons has not been studied. Na + current properties in adult dorsal root ganglion (DRG) neurons of defined size class were investigated in isolated neurons maintained in primary culture using a combination of microelectrode current clamp, patch voltage clamp and immunocytochemical techniques. Intracellular current clamp recordings identified differing relative contributions of TTX-sensitive and -resistant inward currents to action potential waveforms in DRG neuronal populations of defined size. Patch voltage clamp recordings identified three distinct kinetic types of Na + current differentially distributed among these size classes of DRG neurons. 'Small' DRG neurons co-express two types of Na + current: (i) a rapidly-inactivating, TTX-sensitive 'fast' current and (ii) a slowly-activating and -inactivating, TTX-resistant 'slow' current. The TrX-sensitive Ha + current in these cells was almost completely inactivated at typical resting potentials. 'Large' cells expressed a ~!ngle TrX-sensitive Na + current identified as 'intermediate' by its inactivation rate constants. 'Medium'-sized neurons either co-expressed 'fast' and 'slow' current or expressed only 'intermediate' current. Na + channel expression in these size classes was also measured by immunocytochcmical techniques. An antibody against brain-type Na + channels (Ab7493) I° labeled small and large neurons with similar intensity. These results demonstrate that three types of Na + currents can he detected which correlate with electrogenic properties of physiologically and anatomically distinct populations of adult rat DRG neurons.

INTRODUCTION

Dorsal root ganglion (DRG) neurons transmit sensory information from peripheral regions such as muscle and skin to the central nervous system. These primary afferent neurons form a heterogeneous population based on neuroanatomica121, electrophysiological t4,ts,26, and immunocytochemical7 criteria. Cytochemical data indicate three major classes of DRG neurons: types A, B, and C. The soma diameter of these different classes is well-correlated with axonal diameters and, to a lesser extent, with different sensory modalities is. For example, large-diameter neurons give rise to rapidly-conducting, large-diameter, myelinated axons which transmit sensory input from muscle spindles, Golgi tendon organs or touch and pressure receptors. Small type-C neurons give rise to small-diameter,

slowly-conducting, nonmyelinated axons fimctionally associated with pain and temperature sensation. A number of studies 1~'24'3° have indicated that at least two, perhaps three, kinetically and pharmacologically distinct sodium channels are present on the somata of neonatal rat DRG neurons. One of these channels is resistant to micromolar concentrations of TTX and exhibits slow activation and inactivation kinetics. The TTX-sensitive current(s) appear to be either expressed alone or in conjunction with TTX-resistant current. However, the relative distributions of these sodium currents among the three major size classy.r, of DRG neurons are not known. Moreover, most electrophysiological studies of DRG neurons (including the studies cited above) ha':e been carried out in tissue culture of cell lines or primary cultures derived from fetal or neonatal tissue. While this has

Correspondence: S.G. Waxman, Department of Neurology, Yale Medical School, LCI 707, 333 Cedar Street, New Haven, CT 06510, USA. Fax: (1) (203) 785-5694.

284 advantages in providing stable culture preparations, there is evidence indicating that channel distributions can change with development ~5; thus it is possible that distributions of channels a n d / o r receptors in neonatal D R G neurons might be markedly different from ~hose derived from adult tissue. In order to study distribution, density and kinetic properties of Na + channels, as well as their relative contribution to action potential waveforms in adult D R G neurons, we have used a combined approach of whole-cell and excised-patch voltage clamp, microelec, trode current clamp recording and immunocytoehemical staining for Na + channels to study dissociated neurons from D R G isolated from adult rats. O u r re-

sults indicate that three distinct types of Na + currents are expressed in adult neurons. Moreover, these currents are expressed in restricted, physiologically identifiable subpopulations of D R G neurons. MATERIALS AND METHODS Call isolation and culture methods

Lumbar ganglia (L4 and LS) from young Wistar rats were rapidly excised following exsanguination under pentobarbitai anaesthesia and freed of both nerve trunks and connective tissue sheath. Pooled ganglia were incubated with gentle agitation in HBSS containing (in raM): 136.9 NaCi, 4.16 NaHCO.~, 0.36 Na2HPO4, 5.36 KCI, 0.44 KH,PO.~, 5.5 glucose, I0 HEPES (pH 7.3) and I mg/ml eollagenase (Sigma type II or Boehringer types A or D) for 20 to 30 rain; followed by an additional incubation of 20-30 rain in HBSS+ I mg/ml collagenase + 20 U/ml papain (Worthington). Ganglia w e r e

Fig, !. ! h;ffmann contra~t micrographst~f DRG neurons I DIV (A) and 5 DIV (B) in culture, Bar = 50 ~m,

285 then transfered to culture media (1:1 Dulbecco's Modified Eagle's Medium (DMEM) and Ham's FI2 medium containing 10% fetal bovine serum (Gibeo) and 100 U / m l penicillin/streptomycin (Gibco)); and gently triturated using a siliconized Pasteur pipette. Liberated cells were then plated on a poly-ornithine/laminin substrafe coated on a glass coverslip and maintained in culture for 1-6 days. Efficiency of attachment was increased if neurons were not denuded of accessory cells by enzymatic digestion and subsequent isolation procedures. These ceils migrate from the neuron to the coverslip substrate within 6-12 h, leaving the upper surface of the neuron sufficiently clean for application of patch clamp techniques. Approximately 30% of cells in all size classes usually exhibited a lavish outgrowth of neurites within the first 24-48 h, particularly if a stub of the former axon remained.

Electrophysiologicai methods Coverslips of cultured neurons were mounted in a recording chamber and continuously superfused with oxyget~ated Krebs solution, containing (in raM): 124 NaCI, 3 KCI, 2 CaCI 2, 2 MgCI 2, 1.3 NaH2PO4, 26 NaHCO 2, 10 glucose (buffered to pH 7.4 with NuOH). Neurons chosen by smooth appearance of cell surface and cytoplasm were impaled with microelectrodes of 20-50 M,Q when filled with 2 M potassium acetate. In experiments where rates of action potential polarization were measured, electrode capacitance was reduced by a coating of nail polish to within 300 p.m of the tip. Data for analysis was selected from cells which maintained stable resting potential and input resistance as determined by active bridge balance for at least 30 rain. Data were digitized at 8 or 20/~s/pt (TL-! DMA interface, Axon Instruments) and rates of action potential polarization were automatically determined using commercial software (P-Clamp, Axon Instruments). Patch voltage clamp analysis of membrane properties was accomplished using slight modification of methods described in Sakmann and Neher "~9.Pipettes (7052 glass, A&M Systems) of 3-5/~m diameter and 1-5 MD resistance when filled with internal solutions (see below) were sealed to neuronal membranes with gentle suction. Patches with shunt resistance greater than 5 Gllwere ruptured by further suction and an 'outside-out' patch was formed by withdrawal of the pipette tip from the cell. In cells of 20 p,m or less diameter and neurite outgrowth of less than I cell diameter, analysis of mucroscopic ('whole-cell') current was attempted. Data were analyzed in cuses where capacity currents could be reduced by more than 75% and Na + current kinetics were unchanged by reduction of external Nit* or by steady state reduction of channel availability. Remaining capacity and linear leakage currents were subtractetl by an on.line ' P / 4 ' procedure using commercial software (P-Clamp, Axon Instruments). Current signals were pre-filtered ut a bandwidth of 5-10 kHz and digitized at 8 and 211 tts/pt (TL.I DMA interface, Axon Instruments). Solutions contained (m mM): externalfidl No: 1511No, 5 K, 2 Ca, 10 CI, 150 PIPES (buffered to pH 7.3 with NMDG); external ! / 4 No: 37.5 Na, 113.5 NMDG, 5 K, 2 Ca, 10 CI, 150 PIPES (buffered to pH 7.3 with NMDG); external 0 No~~20 Ba: 150 NMDG, 20 Ba, 10 CI, 150 PIPES (buffered to pH 7.3 with NMDG); internal: 150 NMDG, 20 EGTA, 10 CI, 150 HEPES (buffered to pH 7.3 with NMDG), 10 NaH2PO 4, 2 Mg, 2 Na2ATP, 200 mg/ml Na-heparin

(Sigma). Patch clamp studies of Na + current in some neuronal preparations such as GH3 cells T M have shown that voltage-dependence of gating parameters can shift by 10 to 20 mV during dialysis of intracellular contents. Such shifts have not been observed in all neuronal Na + current preparations (cf. refs. 13,171 and were not observed in the course of our experiments with excised patches from DRG: if they occurred, it must have been during the first few seconds of patch formation.

Cytochemical methods Culture preparations were washed several times in phosphatebuffered saline (PBS) at 4°C, fixed with 4% paraformaldehyde/0.01% glutaraldehyde in Sorensen's phosphate buffer for 5 rain at 4°C and then rinsed several times with PBS at room temperature. Fixed

preparations were then incubated with a I:II.~Ddilution of primary antibody (Ab74931 and a 1:2011 dilution of rhodamine-conjugated goat anti-mouse secondary antibody as previously described 5.22. Stained cells were examined using a Leitz Aristoplan microscope equipped with fluorescence optics. Control experiments with an absence of primary antibody or substitution of an irrelevent antibody exhibited a lack of immunostaining. Cultures fixed and stained for fluorescence microscopy were also quantitatively examined for staining intensity using confocal laser ~canning microscopy, using a Bio-Rad MRC-50D system with a 550 nm filter block for detection of rhodamine fluorescence. Gain of the detection system was set such that maximum signals from the brightest specimens (i.e. "large' neurons, at ! day in vitro (DIV) were just below threshold for detector saturation (white display). Noncellular background was adjusted to black display. This calibration range was recorded and utilized in all data collected from subsequent preparations.

RESULTS

Current damp analysis Previous studies ts have indicated that some aspects of Na + current TI'X-sensitivity in DRG neurons can be correlated with cell size. Action potentials from neurons > 50 mm in diameter (Fig. IA,B; hereafter defined as 'large' neurons) were completely but reversibly blocked by exposure to 100 nM TTX (Fig. 2A). Other properties of this class of DRG neuron such as resting potential, input resistance, action potential amplitude, duration and maximum rate of rise are listed in Table !. Approximately 48% of neurons of diameter between 30 and 50 tzm (Fig. IA,B; hereafter defined as 'medium' neurons) exhibited passive and regenerative properties indistinguishable from those of 'large' neurons. Action potentials in these cells were likewise completely and reversibly abolished by 100 nM TTX. The remainder of this cell population expressed action potentials ~,,hich exhibited varying degrees of inflection during repolarization. As shown in Fig. 2B, amplitude and maximum rate of rise of action potentials from this class of cells were reduced by 100 nM TTX but the capacity for regenerative response was rarely lost. Other passive and active electrical properties of these classes of cells are listed in Table I. in cells < 30 ~m in diameter (Fig. IA,B; hereafter defined as 'small' neurons), only strongly-inflected action potentials are observed. 100 nM TTX has virtually no effect on rates of rise of action potentials evoked from normal resting potential in these cells (Fig. 2C). However, anode-break responses in such cells are completely and reversibly abolished by these same levels of TTX (Fig. 3A,B). To determine whether active electrical properties of DRG neurons were influenced by short-term primary culture, recordings were made from cells maintained in vitro for 4 to 6 days, selected (i) by size class and

286 TABLE !

Electroph~,sologicc.! properties of DRG neurons measured after 1- 2 days in citro

Diameter(~m) Resting potential (mV) Input resistance ( M ~ ) dV/dtmu (V/s) Duration (ms at 50% AP height)

Large neurons

Medium neurons

Small neurons

58.77:1:3.0 (n = 13) -68.4 :!: 6.6 43.6 + 18.5 209.3 4.73.1 1.72 + 0.81

37.0 + 3.6 (n = 8) -57.6 + 5.2 39.0 + 10.6 120.7 ±62.9 2.48 + 1.00

25.6 + 2.4 (n = 15~ -51,3 +_13.6 96.1 ±49.0 53.4 ±23.2 4.95 + 2.04

appearance (cf. Fig. 1) and (ii) by stability of membrane potential and passive properties as indicated in Materials and Methods. In 'large' neurons, selected for resting potential and passive membrane properties identical to those of Day 1-2 cultures, action potential rate of rise is slightly (but not significantly) reduced; duration as also slightly increased. Action potential properties were similarly unchanged in 'medium' and 'small' neurons over this time span. These are compared in Table II. The scatter plot in Fig. 4 correlating action potential rate of rise with (duration) -I illustrates that properties of small and large neurons form distinct clusters whose centra do not significantly change with time in culture. Properties of medium-sized neurons fall into one or other of these clusters rather than forming a continuum between them. Voltage clamp analysis Voltage clamp studies of Na + current properties in isolated adult DRO neurons required application of two separate patch clamp techniques. Na + currents in cells < 20 ~m in diameter could usually be evaluated by 'whole-cell' voltage clamp using electrodes 5-10/~m in diameter. In larger cells, outside-out patches formed on electrodes 3-5 ~m in diameter were found to contain sufficient density of Na + channels that macroscopic currents of 100-800 pA were usually observed. Na + currents from 'large' neurons, in approximately 80% of patches from cells > 30 ~ m in diameter (n -13; including all cells > 50 ~ m in diameter, n =, 9), a single class of Na + current was identifiable (Fig. 5A).

Threshold for activation of current was - 50 mV with a peak at - 2 0 mV (Fig. 5B). The current availability relation (using data obtained at test potentials of - 10 and - 2 0 mV) as a function of holding potential could be fit by a single Boltzman distribution whose midpoint was - 87.5 mV and whose slope factor was 8 mV/e-fold potential change (Fig. 5C). Decay rates could be fitted to a single exponential process whose time constant exhibited marked voltage dependence, varying from 6 ms at a test potential of - 4 0 mV to an asymptote less than 0.9 ms at potentials > - 1 0 mV (Fig. 5D). Time course of current activation was insufficiently resolved for kinetic analysis. This Na + current could be completely and reversible blocked by 100 nM TTX (not shown). Na + and Ca 2+ currents in 'small' neurons. Because action potentials in 'small' (C-type) DRG neurons are influenced by both Na + and Ca 2+ and are TTX-resistanP 1'4°, it has been speculated that inward current is conducted by a single channel type permeant to both Na + and Ca 2+. Fig. 6A,B illustrates distinct Na + and Ca z+ currents measured in whole-cell voltage clamp in this class of cell. Na + currents (Fig. 6A) measured by whole.cell voltage clamp in this class of cells exhibited biphasic inactivation kinetics and pharmacological properties, discussed in more detail below, compatible with co-ex. press±on of two distinct Na ~" channel isotypes (which we refer to as 'fast' and 'slow') similar to those originally described by Kostyuk et a l l j. By contrast, wholecell currents evoked in 0 Na+/20 Ba 2+ external solu-

TABLE 11

Electrophysologicai properties of DRG neurons measured after 4 - 6 days in vitro

Diameter (jam) Resting potential (mV) Input resislanc¢ (MI?) dF/dtm~ (V/s) Duration (ms at 50% AP height)

Large neuron~

Medium neurons

56.0 ± 6.8 (n -- 21) -66.1 + 6,1 28,8 4.14.7 120.4 4-24.7 2.01 4- 1.00

36,0 4. 2,8

Small neurons 23.2 ± 1.8

(n = 5)

( n = 5)

-66.8 4- 8.4 34,6 4.12.7 96,9 4. 13,5 1.84 4. O.63

- 5 0 . 2 ±10.0 61.6 ± 12,7 43,1 4-13.3 5.76-',- 1.96

287

L control

~

L

Small

Fig, 2, Action potentials under control conditions representative of size classes illustrated in Fig. h A: large cells; B: medium cells; C: small cells. Vertical calibration: 20 mV. Horizontal calibration: 2 ms, The abcissa of calibration bars is aligned at zero membrane potential. Right column: action potentials from the same cells in the presenceof IUO nM TTX.

tion exhibit no inactivation over this time scale but activate and deactivate rapidly upon step polarization of membrane potential (Fig. 6B) as expected of neuronal Ca 2+ currents. Substitution of Ba 2+ for Ca 2+ in external solutions can affect current decay rates (i.e. currents exhibit Ca 2+-dependent inactivation s at longer time base as shown in Fig. 6C. These Ca 2+ currents exhibit pharmacological properties identifying them as 'L-type 'tg, or high-threshold Ca 2+ current. Negative shift of activation threshold and prolongation of tail current by the dihydropyridine agonist BAY K8644 (1 /~M) is illustrated in Fig. 6D. 'T-type' or low-threshold Ca 2+ currents were not detected (n = 23) in this sizeclass of D R G neuron. Outside-out patche~ from approximately 63% (n = 5) of 'medium'-sized neurons exhibited Na + currents with kinetic properties indistinguishable from those described above in patches from 'large' neurons. 37%

Fig, 3. 100 nM TTX preferentially blocks the anode-break action potential in small DRG neuron while sparing the action potential which may be directly evoked by depolarization from the resting potential, A: control; B: 100 nM TTX. Vertical calibration: 10 mV. Horizontal calibration: 10 ms,

(n = 3) of patches from this size class exhibited Na + currents which were identifiable as the 'fast', TTX-sensatire and 'slow', TTX-resistant currents in 'small' neurons. 1.0 • large cell 1.2 DIV O •rOe ceil, 4.6 ON • mediumcell, t .2 DIV v mediumcell, 4.0 OiV

1.4

~

i small cell 1.g DIV o s m e ce ,4.0D V

1,2

0

•

O

gO

I.0

"7, 0.a g: 0

¢ ~,

ee

o• ~70

o

o

"

0,4

oj~•n u

0.2

0

Oe

,

",

ant3 0,0 O.

n 60

a 100

i ,, J 150 200

I '260

t

I 4OO

Rate of Rise (Vlsec) Fig, 4. Correlation of action potential maximum rate of rise with inverse duration in I - 2 DIV neurons and in those of 4-6 DIV. Filled symbols: I - 2 DIV; open symbols: 4-6DIV. Circles: 'large' neurons; squares: 'small' neurons; inverted triangles: 'medium' neurons. Note that dV/dt and action potential duration of 'small' neurons are distinct from those of large neurons and that prop•ties of 'medium' neurons fall into either the 'small' or the 'large' distributions.

288 and 0 mV. 'Slow' current was resistant to 100 nM TTX whereas 'fast' current was completely blocked. Summation of data from patches excised from medium-sized neurons with 'fast' and 'slow' Na + currents (as illustrated in Fig. 7) can be shown to reproduce the kinetic features of whole-cell Na + currents from small neurons (Fig. 8A). The TTX-sensitive Na + currents in small and large DRG neurons have at least one major, distinguishing kinetic feature. This is illustrated in Fig. 8B by superposition of currents from panels 5A and 7A. Inactivation rates of 'fast' current (small cells) is substantially faster that that of 'intermediate' current (large cells) at test potentials just above activation threshold. At more positive test potentials decay rates of 'fast' and 'intermediate' current become similar.

'Fast' Na + current, Rapidly-activating and -inactivating

currents with kinetic properties similar to those of the initial component of whole-cell (i.e. 'fast' currents) current from 'small' cells could be detected in excised patches from some 'medium' cells (Fig. 7A; upper set of traces). These were activated at a threshold of - 50 mV and with a peak at - 2 0 mV (filled circles; Fig. 7B). Current availability as a function of holding potential could be described by a single Boltzman distribution with a half-maximal point of - 85 mV and a slope factor of 7 mV/e-foid potential change (filled circles; Fig. 7C). Current decay rates exhibited little change, varying between 0.9 and 0.4 ms over a range of test potentials between - 5 0 and 0 mV (filled circles; Fig. 7D). 'Slow' Na + current. Threshold for activation of 'slow' current was - 4 0 mV with a peak at - 10 inV. Steady state current availability could be fitted to a single Boltzman distribution which exhibited a i/i/2 of approximately - 45 mV and a slope factor of 9 mV/e-fold potential change (open circles; Fig. 7C). Decay rates could be fitted to a single exponential whose time constant was strongly voltage dependent, varying between 12 ms and 2.5 ms over potentials between - 4 0

A.

lmmunocytochemical analysis

It was of interest to determine whether the antibody Ab7493 ~°, which was generated against purified Na + channels from adult rat brain would yield information concerning Na + channel expression in dorsal root ganglion neurons, particularly when currents in small and large neurons possess distinguishing kinetic features.

B| -60

-40

-20

0

20 !

40 t

Vm(mV)

"0'2 /

I'0

L'f

_j,ooo ~msl¢

D.

(m.c)

C.

"6

['0"

0,8.

'4

0.6. 0'4'

"2

0.2' -tt

,

o,oo

.o

,o

vm(mv) .

~

.4'o

-2 o

-6'o

....-

o

I 20

Vh(mY) Fig. 5. A: ensemble Na* currents from an outside-out patch excised from a 52 p m diameter neuron, Holding potential: - 1 1 0 mV; test potentials are indicated at each trace. B: normalized peak current-voltage relation compiled from 5 similar data sets from 'large' cells ( > 50 ~.m diameter, as discussed in the text), C: steady state current availability relation plotted as a function of holding potential; maintained at indicated levels for 30-120 s prior to test depolarization to allow slow inactivation processes to reach a steady state. Graphed data were fitted to a single Boltzman distribution as described in the text. D: plot of current decay/inactivation time constant as a function of test potential, fitted to a single exponential decay process as indicated in text. Points show means and standard deviation of data from the five patches utilized in (B).

289 Representitive fields of dorsal root ganglion cells that have been labeled with sodium channel antibody Ab7493 are shown in Fig. 9. The cell bodies of D R G neurons cultured for 1 day exhibit intense immunostaining; in addition, the proximal portions of some neurites display robust immunoreactivity. The three size classes of D R G neurons appear to possess essentially equal intensities of immunoreactive staining (Fig. 9A,B,C). In contrast to the robust immunostaining at 1 DIV, D R G neurons at 4 DIV exhibit attenuated immunoreactivity (Fig. 9D,E,F), which is apparent on all size classes of neurons. T h e extensive neurite outgrowths from D R G neurons at this time in culture also exhibit readily detectable sodium channel immunosraining. Intensity measurements of Ab7493 staining obtained by confocal laser microscopy exhibited spatial heterogeneity in D R G neurons from all size classes at 1 DIV. This is illustrated in images of Fig. 10 (panels A - C ) and corresponding intensity scans (Fig. l l A - C ) in which number of pixels within the cell perimeter are binned according to intensity. Such plots from the majority of neurons in each size class exhibit two maxima whose means are separated by 3-4-fold differences in intensity. By contrast, fibroblasts and Schwann

cells invariably exhibited a single intensity distribution whose mean was significantly less than any of the peaks observed in neurons (Figs. 10D anc! I iD). Ab7493 staining in neuritic/axonal outgrowths from 4-6 DIV neurons could be detected by this technique. Mean intensities of independently measured axonal domains were similar to those of the lesser intensity peak observed in the cell s o m a (Figs. 10E and l l E ) , although small axonal 'hot-spots' could also be observed. These patches did not exhibit any spatial periodicity of the type suggested for 'phi nodes '32 observed in demyelinated axons, The mean intensities of the non-neural cell population did not change significantly between 1 and 4 - 6 DIV (data not shown). Staining intensity in these cells also did not appear to be influenced by contact with axon outgrowths over this time span. Intensity of immunostaining by Ab7493 appeared to be slightly reduced over time in culture and of similar intensity in neuronal somata and extended neurites. These cytochemical data were supported by two types of patch clamp experiment. Amplitude of patch current obtained from 'large' neurons at 1 and 6 DIV using size-matched pipettes appeared slightly reduced (by approximatel~ "5 + 10% n = 16), without other changes

.,,.'~-"-~ ....

C"

. . . . . . . . ,p. . . . . . . .

:\Y;7:~:yz:_"t)i~v

D,

-~mV

-20m OmV

~ : control o : B A Y 1