Brain Research, 585 (1992) 56-62 © 1992 Elsevier Science Publishers B.V. All rights reserved 0006-8993/92/$05.00
56
BRES 17811
Heterogeneity in ECs0 and n n of G A B A A receptors on dorsal root ganglion neurons freshly isolated from adult rats Geoffrey White Neurogen Corporation, Branford, CT 06405 (USA) (Accepted 21 January 1992)
Key words: y-Aminobutyric acid; Inhibition; Dorsal root ganglion; Pat-;h-clamp; Ion channel
GABA activates a CI- current through the GABA A receptor/ionophore complex that influences excitability of neurons. Studies using expression of cloned cDNAs coding for different GABAA receptor/ionophore subunits suggest that the EC.~o and Hill coefficient for GABA are influenced by subunit composition. However, no direct evidence for such heterogeneity has been reported for vertebrate neurons, i have investigated the heterogeneity of EC.~o and Hill coefficients (n H) of isolated dorsal root ganglion neurons using the whole-call patch clamp technique. The ECso for GABA varied from 26 to 107 ~M among neurons, n H calculated from the logistic equation varied from 1.18 to 2.0. A negative correlation was found between the ECs~ and n H (r ffi-0.81). Both n H and ECso differed between some cells. However, in some instances, n H differed between cells while ECso values were similar, and in other cells, ECs(~ values differed and n n was similar. In addition, when cells were categorized according to action potential shape, the ECso and Hill coefficients differed among cell types in some instances and were similar in other instances, These findings demonstrate that different pharmacological profiles for GABA can be observed in adult mammalian neurons. Selective distribution of such pharmacological subtypes of GABA A receptors may contribute to control of neuronal excitability.
INTRODUCTION GABA is the major inhibitow neurotransmitter in the mammalian brain. GABA can influence neuronal excitability by activating a Cl~current through the GABAA receptor/ionophore complex. This complex is a hetero-oligomeric protein for which a variety of candidate polypeptide subunits have been identified 4.~. Both the mRNAs coding for different subunits and their polypeptide products are differentially expressed within the brain and possibly within neurons them. selves ~'~. Recent studies using cloned mRNAs and cDNAs coding for different GABA^ receptor/ ionophore subunits suggest that subunit composition can influence the value of ECs0 and Hill coefficients (n~t) ''~x. Taken together, the above findings raise the possibility that expression of different subunit combinations within the nervous system can result in a tope. graphical pattern of pharmacological and physiological properties of GABAA receptor/ionophore complexes m'. Interestingly, no direct evidence from voltage-
clamp studies in vertebrate neurons has been reported that demonstrates such heterogeneity in EC~, and n~t for the GABA^ receptor. In fact, a high degree of homogeneity has been observed (EC~t}~ 10-50 ~M and n H ffi 1.8-2), based on averages of concentration. response curves generated from a number of neurons L~''~. Three possible ways to account for this lack of heterogeneity are: (i) ECso and n H are the same in all neurons, (ii) ECs0 and nit were the same in neurons in the cited studies but are different from other neurons, and (iii) EC~, and nit were heterogeneous among indvidual neurons and the apparent homogeneity resulted from averaging responses from different neurons, l have tested the first possibility by evaluating concentration-response curves from individual neurons isolated from dorsal root ganglia. MATERIALS AND METHODS
Cell isolation Dorsal root ganglia (DRG) were dissected from male SpragueDawley rats (200-500 g) and placed in oxygenated DMEM with
Correspondence: G. White, Neurogen Corporation, 35 Northeast Industrial Rd., Branford, CT 06405, USA. Fax: (1)(203)481-8683.
57 NaCI substituted for NaHCO 3 (340 mosm), 0.4 mg/ml type I!I trypsin, 0.3 m g / m l type VII collagenase, and 0.1 mg/mi type IV DNase I!. The DRG containing enzyme-rich solution was incubated at 35°C in a shaking water bath for 45-60 rain. Individual neurons were dissociated from the ganglia by vigorous shaking after which 0.4 mg trypsin inhibitor {Sigma type I! from soybeans) was added. Neurons were kept at room temperature (20-22°C) in the DMEM solution and were aliquoted as needed to a 35 mm petri dish. Neurons were devoid of processes as described previously 22.
Electrophysiological responses were obtained using the whole-cell patch-clamp technique with an Axopatch ID amplifier taxon Instruments) and pClamp software tAxon Instruments) as described previously 2t. Subsequent to gaining access to the cell interior, membrane potential with 0 current injection was negative to - 50 mV and input resistance was > 500 GO in cells included in this study. Membrane potential was subsequently voltage-clamped to - 60 mV. Series resistance (3-4 M,Q) was compensated by 90%. Current amplitude was < 1.5 hA. Therefore, errors due to series resistance were negligable. External solution contained (raM): 150 NaCI, 5 KCI, 2 CaCI 2, 1 MgCI 2, 10 HEPES, 10 glucose, osmolarity was adjusted to 320 using sucrose and pH was adjusted to 7.4 using NaOH. Recording electrode solution contained (raM): 130 KCI, ! CaCI,, 2 MgCI~, II EGTA, 10 HEPES, 5 mM MgATP, osmolarity was adjusted to 310 using sucrose and pH was adjusted to 7.4 using KOH.
Hill coefficient and can be interpreted as a measure of cooperativity and minimal number of binding sites when agonist/receptor/ response relationships are known to fit a specific set of constraints. However, such constraints are more often met in drug binding studies than in physiological studies7.12. For electrophysiological experiments in which agonist activated current flow is the measured endpoint, these constraints include that charge transfer per unit time be equivalent for each molecule of agonist bound. In the absence of this constraint, non-interacting binding sites could appear to be interacting in terms of binding affinity if total charge transfer per unit time increased or decreased disproportionately each time an additional agonist molecule bound to a given complex. Such confounds might be present in the case for the GABA A receptor/ ionophore complex because multiple receptors or channel types with different charge transfer characteristics may be recruited at different concentrations of GABA 23 and bursting substates may exist which depend upon the number of agonist molecules bound, as well as on the concentration of agonist presentt7. For this study, n H will therefore be interpreted as no more than a measure of steepness of the concentration-response relationship for a given neuron. To facilitate a statistical comparison of ECso and n H among neurons (using t-tests), amplitudes of current evoked by increasing concentrations of GABA were transformed to a linear relationship using the Hill plot. For Hill plots, data was selected to surround the estimated ECs0, thus data points at the extremes were discarded for this part of the analysis ~2.
Drug rt,sllon,~;e$
RESULTS
Recording
The concentration-response analysis was carried out on neurons ranging from 20 to 30 /~m in diameter. GABA was applied to neurons using a 17-barrel delivery system composed of a linear array of 200 ~ m (i.d.) fused silica tubing (Hewlett Packard). Delivery rate of drug containing external solution was ,,, 90 t~m/ms calculated from a flow rate of 175/~l/minute. Superfusion of neurons with drug containing external solution was initiated by opening a valve after placing the delivery barrel to within I(10/~m of the neuron. Drug delivery was terminated by closing the valve and removing, the delivery pipets from the bath. Sometimes cells were exposed to drug-containing solution by supeffusing a cell in non-drug-containing inflation and then switching directly to a barrel from which drug-containing solution was flowing, in this case, drug delivery was terminaled by moving back to the barrel from which non.drug.containing solution was flowing. Results t'rom either method were comparable within a given neuron as reporled by others "~. With repealed exposure to u given concentration of drug by either method or between methods within a given cell usually resulted in amplitude differences of less than 5%. Occassionally, a difference of less than 10% was observed. Concentrations of GABA were applied in ascending order. In 6 neurons the initial concentration of GABA (5 or 10/zM) was applied after the final application of the highest GABA concentration to verify stability of response amplitudes. The final response amplitude under these circumstances was within 10% of the response amplitude which preceded the series of drug applications. At least 2 rain were allowed to elapse between application of concentrations of GABA < 25/~M and at least 7 min elapsed between applications of > 25/~M GABA. Drug responses were collected on a Gould chart recorder model RS3200 and evaluated as described previously~t.
oan'e /itting Curve fitting was carried out using KaleidaGraph (Abelbeck Software) on a Mac II {Apple Computer). Concentration-response curves were fit to the logistic equation:
Heterogeneity in concentration response to GABA
Fig. 1 illustrates heterogeneity among neurons in parameters of their concentration-response relationship to GABA. A family of currents evoked by increasing concentrations of GABA is illustrated in Fig. I A. The response amplitudes from Fig. 1A are plotted in Fig. I B and were fitted to the logistic equation. When concentration-response curves of this type were compared among neurons a heterogeneity in EC.~0 and n~ was oberved (Fig. IC). From left to right in Fig. IC the calculated ECs0s for individual concentration-response curves were 26, 63, and 100/~M GABA and nil values were 2.0, 1.49, and 1.23 (r: > 0.99 in each case). A Hill plot was constructed from these data for the purpose of statistical comparison of concentration-response parameters among neurons (Fig. ID). EC.~0 values calculated in this manner were 24, 62, and 102 pM and n H values were 2.0, 1.45, and 1.24 (r2> 0.99 in each case). The three slopes differed from one another (P < 0.01, combined probability from multiple t-tests, d f = 8 for individual tests) as did the EC50 values (P < 0.01, combined probability from multiple t-tests, d f - 8 for individual tests).
!--/max X l / ( l + ( E C s 0 / [ d r u g ] ) " "
Con'elations among ECho, n n, and maximum response amplitude
where i is current amplitude, /max is the calculated maximum current amplitude, ECso is the concentration of agonist that produced the half maximum current amplitude, [drug] refers to a given drug concentation, and n H is a factor that describes the steepness of the concentration-response curve, n H is generally referred to as the
A significant degree of variability in the EC50 for GABA can be explained by the variability in n H (Fig. 2A). A correlation between specific membrane current and either EC50 or nrl was not statistically ~ignificant
58 A 2.2
1.8
SO
150
~
2SO:i 3
IV"
O
5
1.6
1.,
1
0
~
2 3 Seconds
4
r = O.81 p< O.O1
1.2
5
1
1.'2
350
3
1.'4
1'.6 l:a log EC50
c:o-2'0o
250 < Ir~
1.~0
l
1
5
4 2
50
r~ R"--'r",r~
0.5
........ • I o
IO0.
I000
GABA(pM)
I
o
E
1
-o.s
lOO
o
7s -1,5, o
5O
log GABA (pM}
2~
0 !0
i1 1. S. . . . . . os
1000
GABa {IJM)
o
-
100
o
---~o" 1 , s
tog GABA (IjM) Fig. !. DRG neurons respond to application of increasing concentra. tions of OABA in a heterogeneous manner. A: family of currents activated by application of increasing concentrations of GABA, Current records were digitized at 143 Hz for the purpose of presentation. Application lasted 15 s. Only the first 5 s are shown for clarity, v axis is in picoampures (pA), B: plot of current amplitudes ,ql.wn in A, The response to 400 /~M was not included in A for clarity. The curve was generated from the logistic equation, C: normali/.ed current amplitudes fit to the logistic equation show difference~ in EC.~. and n . . Data was normalized to lm,,x from the I.gistic equation when applied to data before normalization. Data in the right hand curve are from A. See text for further details. D: Hill plot of data from C. Of all the neurons evaluated, the greatest variance was observed for the curve illustrated on the left of Fig, IC and D. However, this variance is so small that the neuron is still considered to be representative. S.E.M.s for ECs, and n H values are shown in Fig. 2A.
Fig. 2. Diversity of EC~0 and n . values, A: ECs, for GABA shows a negative correlation with nip r - correlation coefficient, P is from a t.test, df - 15. Hill plots of cells marked I, 2, 3 are shown in Fig. IC and $.B,M,s are indicated for cells I, 2, 3, 4, and 5, B: Hill plot of c~lls marked 2, 4, 5 in A, Cell 2 is also shown in Fig, IC and D.
(r = 41 and 38, respectively, P > 0,05, t-test, df = 15). Because cell size of freshly isolated DRG neurons can be an indicator of pharmacological properties (e.g. refs. 10 and 21), the relationship between whole.ceU capacitance and ECho, nil, and current density were evaluated, r was between 0 and +0,10 in each instance. The plot of Fig. 2A suggested a number of combinations of EC,~, and nil values could be observed in mammalian neurons, as suggested from studies using expression of cloned mRNAs encoding various G A B A ^ receptor subunits t", As shown in Fig, 1, cells can differ from one another in both nit and ECso values. As shown in Fig, 2B, some cells (cells 2 and 4) can share ECs, values (62 vs, 59 ~M, P > 0,1, df = 8) and differ in nit values (1,45 vs. 1,'/2, P < 0,001, df = 8) while other cells (cells 2 and 5) can have similar n H values (1,45 vs. 1,54, P > 0.1, df = 8) and differ in ECso values (62 vs. 34 pM, P < 0.001, d f = 8). Cells 4 and 5 differed from one another in both parameters (P < 0.02, combined P values, df = 8 for individual tests). As can
59 be seen from inspection of Fig. 2A, some cells did not differ in either value.
Effect of averaging concentration-response curves across neurons" Concentration-response curves are generally represented in the literature as averages from many neurons. The normalized amplitudes of the concentrationresponse curves from this study were averaged, plotted, and fitted to the logistic equation (Fig. 3). The ECs0 was 51 /~M and n H was 1.48 (r2> 0.99). It is not obvious from the small error bars, and the large r 2 value for the regression that this curve is composed of a heterogeneous set of curves.
Comparison between methods for calculating slopes of concentration-response curves The slope of a concentration-response curve can be calculated in at least two ways: (1) from a range of agonist concentrations using the logistic equation (n H1) or (2) from the maximal slope of the concentration-response curve on a log-log plot (nil2). This can be achieved by dividing the ratio of response amplitudes evoked by doubling agonist concentrations by 2 (e.g. ref. 3). Naturally, agonist concentrations must be within the range of the maximal slope. The findings of Yasui etal. za and Twyman etal. t7 would predict that an n H value calculated from lower concentrations of GABA might be different from that calculated at higher concentrations of GABA. n , 2 was calculated from the slope of the concentration-response crave at 5 and l0 ~M OABA and divided by 2. hill and nil2 were different in most instances and no correlation was observed (Fig. 4), In addition, no correlation was observed between nlj2 and EC~0 ( r - 0.03). It was possi-
."2
E
Q 1.S , - -
o£ .
I
-r
I ~ I f
e-
I ~/~j'~6 ._,,I •
*~,. .., o
.
e.7
•
u -
• e
~
u
u
1
1,5
2
nH f r o m
--
r = 0.07 p> 0.2 '
' l
2.5
10 pM/5 pM
Fig. 4. n n l calculated from the logistic equation is different from nil2 calculated from low concentrations of agonist. The thick line shows actual linear regression. The thin line shows theoretical distribution if both methods resulted in similar n H values. See text for details.
ble that nH calculated from 5 and 10 ~M GABA was underestimating the true n H (see ref. 3). If this were the case then the difference between n H calculated in the different manners should be related to EC~0 values. Again, no significant correlation was observed (r ffi 0.38, P > 0.2), suggesting that a large systematic error due to variation in ECsos was not introduced. In addition, the curves fit by the logistic equation were extrapolated to regions of low concentrations of GABA (0,I ~M) and the experimental values for n , 2 were compared to the theoretical value of n~j2 for the slope between 5 and 10 ~M GABA. In each case, the difference between the idealized value and the empirical value for nil2 could not be accounted for by either: (1) variability in response measurements or (2) the deviation of agonist concentrations from the area of the maximal slope (not shown).
Expression of ECso and nn as a function of cell type
150ECso = 5'1 IJM 0 4.J
I1
.
o~
CO 100 o~
I=L
~g
C
5o
0
.
1
i
10
100
1000
GABA (pM)
Fig. 3. Concentration-response averaged from all neurons, n---14, 17, 17, 17, 17, 15, 8, and 7 for 5, 10, 25, 50, 100, 200, 400, and 800 ttM GABA respectively. Bars show standard errors. The apparant absence of error bars in some instances resulted from small standard errors.
It is of interest to determine whether different cell types can exhibit different pharmacological properties for the GABA A receptor ionophore complex. Different types of neurons can express different combinations of mRNA coding for subunit variants of the GABA A receptor/ionophore complex I'~'l'~. If GABA^ subunit composition can contribute to the expression of different ECs0 and n}~ values, then it might be expected that these values could be different in different cell types. Within the size range of neurons examined, different shapes of the action potential were observed. Action potential shape has been taken as an indicator of subtypes of DRG neurons s'21. The action potential waveform (top) and the derivatized waveform (bottom) in Fig. 5A are different from those in Fig. 5C
A
B
/
/~ EC50=34laM ~ H=1"87
c
A EC5o=37HM
D
A
ECso=1071JM
A ECbo-991JM
-5' rnV7~......._.._. -56 mV~~'-~-40mV ~oo v/sL5m$ Fig. 5. Different cell types can express different or similar EC~. and n n values. Upper waveforms are reconstructions of action potentials digitized at 50 kHz. Lower wavcforms are first derivatives of action potential waveforms. Action potentials evoked by 600 pA pulses of 2-3 ms duration, as indicated by the initial square deflection in the derivatized wavefi)rms. EC~. and n H were calculated from fittin~ the GABA concentration-.vcspons¢ data from each cell to the h)ljistic equation,
were
GABA, (2) averaging of concentration-response data across neurons can mask underlying heterogeneity, (3) a negative correlation exists between ECs0 and n n calculated from the logistic equation, (4) n n calculated from low concentrations of agonist can differ from n n calculated from a wide range of agonist concentrations in adult mammalian neurons, and (5) different populations of neurons can express similar or different ECso and n H values. Other pharmacological and physiological properties of the GABA A receptor/ionophore complex were not investigated in the present study, Such properties include rectification of GABA activated current and sensitivity to agents which alter receptor/ionophore function. These properties may also be sensitive to subunit composition ~t'. Thus it should be noted that neurons that express similar ECs0 and nit values may differ in other properties and neurons that have different ECs0 and n n values may be similar in other properties. Effect of enzyme treatment on GABA a responses One of the great disadvantages of studying freshly isolated neurons is the ever present possibility that one is studying the effects of enzyme treatment itself. The reason for using freshly isolated DRG neurons is that they are electrically compact, thus providing an ideal space clamp model. The present study could not be reliably carried out in cultured neurons due to problems controlling voltage and ionic gradients in distant processes. Arguing against effects of treatment are the observations that over the course of many studies using the present methods every freshly isolated DRG neuron has shown a GABA^ type response to application of GABA (n > 1000), conventional benzodiazepine pharmacology (n > 75), and conventional responses to pentobarbitai and various GABA antagonists (n > 25). Nonetheless, it is still possible that subtle effects of enzyme treatment gave rise to the observed differences in EC,~, and n , reported for tiffs study. The range of EC~. and nil values observed in this study can not be accounted for entirely by enzymatic differences between preparations, For example, in preparation 32991, the entire range of parameters was observed. The EC,~,s were 37, 82, and 107 #M and the respective n n values were 1.71, 1.18, and 1.24. Such variability could still be accounted for by enzymatic differences ~mong neurons within a preparation.
enzyme
and D as are their respective EC.~. and ntl values. This indicates that different neuronal types can display different pharmacological properties for GABA. O:~ the other hand, while the waveforms arc also different in Fig. 5A and B, their respective EC~0 and n , values indicate that different cell types can also express similar pharmacological properties. Whether individual types ot neurons as defined by their action potential waveforms can exhibit different pharmacological properties is an equally interesting question but was not addressed due to the difficulty in determining when action potential waveforms are the same. For example, while the waveform is different between Figs. 5A and B, it is not obvious whetller the differences between Figs. 5B and D are sufficient to constitute a separate populations of neurons. DISCUSSION This study has demonstrated that (1) adult rat neurons can display a variety of EC~0 and nit values for
Determbzation of response amplitudes In this study, two possible contaminants of the shape of the concentration/r~sponse curves are desensitiza-
61 tion and changes in intracellular CI- concentration. Because response amplitudes were measured at their peak, significant desensitization would result in an underestimation of peaks and a flattening of the concentration-response curve, A similar situation could result if Cl- redistributed significantly during application of GABA. Peak amplitudes were not measurabIy altered by desensitization. The rates of desensitizatien were biexponential and on the order of seconds and tens of seconds with no clear pattern among neurons that was attributable to ECs0 or n H (data not shown, 10 neurons evaluated). The rates were similar to those reported by Akaike et al. ~. At concentrations of GABA at which significant desensitization was observed, response amplitude calculated as described in Materials and Methods were indistinguishable from response amplitudes calculated by back extrapolating the response amplitude to time zero from responses digitized at 143 Hz. The empirically observed rise times of responses in this study were also independent of ECs0 and n H and were similar to those described by Akaike et al. ! (e.g. on the order of tens of milliseconds for concentrations of GABA ~ 100/~M). Given that the rise time of the response was several orders of magnitidue faster than the decay time, it is not surprising that no difference was detected between the two methods of measuring response amplitude. Under the conditions of this study (large somata, high i~ltracellular CI-, and small current amplitude), a contamination of response amplitude arising from changes in intracellular C l ' concentration would not be expected to measurably contribute to the response amplitude evoked by application of GABA 2.
On the interpretation of n u Two methods were used for calculating n H. The values of n H calc:llated by the two methods were different within neurons and apparently unrelated, however both methods resulted in heterogeneity in n H among neurons. One explanation for why nH differed within neurons is that concentration dependent events may occur which alter nil. In fact, such a finding has been reported on the single channel level insofar as n H is reduced considerably over the concentration range of 0.5 to 5/.tM 13. The lack of a correlation between n H when calculated by the two methods indicates that the factors responsible for expression of nHl and nil2 were not the same in every neuron. This suggests that the pharmacological properties of the GABAA recept o r / ionophore complex may be quite intricate.
On the relevence of the negative correlation between n u and ECho While cooperativity between binding sites on the GABA A receptor/ionophore complex is difficult to prove (as discussed in Materials and Methods), it is also difficult to rule out. If true cooperativity in binding does occur, then the negative correlation would be consistent with the increase in ECso resulting in part from an increase in cooperativity. The existence of the negative correlation between n H and ECso was not obvious from studies using expression of cDNA to evaluate differences in pharmacological properties of GABA A receptor subtypes ~6. The negative correlation was modeled using a logarithmic fit for simplicity. In fact, the fit to a logarithmic function was only marginally better than a linear fit (r = -0.81 vs. -0.76, respectively). The true form of the actual relationship remains to be deterrained from a larger sample of neurons. A larger sample number could, for example, highlight a step function resulting from discrete combinations of subunits. The continuous nature of the data could result from the presence of heterogeneous populations of independent binding sites within individual neurons. While the presence of such separate non-interacting binding sites generally results in an n H of < 1, if the ECs0 values are separated by less than ~ 10-fold, then the measured ECs0 will reflect the multiple ECs0 values and n H can be greater than 1 (e.g. Fig. 3). Such a phenomenon could also contribute to the differences detected in n H1 and n H2. This contribution of heterogeneous independent binding sites to GABA A receptor pharmacology is difficult to address in the absence of complete control over factors such as lipid environment, posttranslationai modifications, and suhunit stoichiometry. Effect of averaging concentration-response curves Averaging concentration-response curves from different neurons without first comparing their respective slopes and ECso values makes interpretation and comparison among cells from different regions of the nervous system or different animal species difficult because the average may be composed of several populations of neurons as defined by concentration-response to GABA. This situation may hold for other ligandactivated channels as well. For example, the ATP-gated inward current may be composed of a heterogeneous population of receptors 3 and capsaicin-evoked inward currents may also arise from a heterogeneous set of receptors because adult neurons exhibit different sensitivities to the drug, one in the tens of nM range and
62 the other in the hundreds of nM range -'l (unpublished data). Factors that may contribute to differences in pharmacological properties D i f f e r e n c e s in pharmacological properties described in this p a p e r could be a c c o u n t e d for by differing combinations o f variants of G A B A A r e c e p t o r / i o n o p h o r e subunits, posttranslational modifications, or o t h e r metabolic processes. It is likely that heterogeneity in subunit c o m b i n a tions contributed to the pharmacological heterogeneity observed in this study. Such an influence has b e e n reported for G A B A A responses in n o n - n e u r o n a l expression systems treated with various combinations of G A B A A subunit c D N A s 1~''!8. In addition, a recent report by Persohn et al. ~5 indicates that D R G n e u r o n s express m R N A for G A B A A receptor subunits in a h e t e r o g e n e o u s manner. Pharmacological properties were evaluated for this study in a select group of neurons, thus w h e t h e r heterogeneity of subunit composition exists in this particular group of neurons r e m a i n s to be d e t e r m i n e d . it is possible that posttranslational modifications and metabolic factors could also contribute to heterogeneity in EC5, and n~l values; however, little is yet known about such interactions.
Physiological significance of pharmacological heterogene. ity This study has demonstrated that pharmacological properties of the GABA^ receptor/ionophore complex differ among cell types. Small (10-20%) changes in the strength of inhibitory systems can significantly alter neuronal excitability s. Thus, subtle differences in GABAergic pharmacological properties may contribute to local control of neuronal excitability. To what extent differences in ECs() and Hill slopes contributes to inhibitory strength is unclear given that concentrations of GABA at the synapse may be quite high. it may be that differences betwe,:rt these two parameters simply accompany other parameters such as channel open time, susceptibility to metabolic control, or sensitivity to various pharmacological agents. Differences between neurons in these parameters could result in differences in inhibitory strength. REFERENCES I Akaike, N., indue, M. and Krishtal, O.A., 'Concentration.clamp' study of gamma-aminobutyric acid-induced chloride current kinetics in frog sensory neurones, J. Physiol., 379 (1986) 171-185. 2 Akaike, N., lnomata, N., and Tokutomi, N., Contribution of chloride shifts to the fade of gamma-aminobutyric acid-gated currents in frog dorsal root ganglion cells, J. Physiol., 391 (1987) 219-234.
3 Bean, B.P., ATP-activated channels in rat and bullfrog neurons: concentration dependence and kinetics, I. Neurosci., 10 (1990) 1-10. 4 Bormann, J., Hamill, O.P. and Sakmann, B., Mechanism of anion permeation through channels gated by glycine and gammaaminobulyric acid in mouse cultured spinal ncurones, I. Physiol., 385 (1987) 243-286. 5 Changnac-Amitai, Y. and Conners, B.W., Horizontal spread of synchronized activity in neocortex and its control by GABAmediated inhibition, J. Neurophysiol., 61 (1989) 747-758. 6 Choi, D.W. and Fischbach, G.D., GABA conductance of chick spinal cord and dorsal root ganglion neurons in cell culture, J. Neurophsyiol., 45 (1981)605-620. 7 DeLean, A.P., Munson, P.J. and Rodbard, D., Simultaneous analysis of families of sigmoidal curves: application to bioassay, radioligand assay, and physiological dose-response curves, Am. J. Physiol., 235 (1978) E97-EI02. 8 Harper, A.A. and Lawson, S.N., Electrical properties of rat dorsal root ganglion neurones with different peripheral nerve conduction velocities, J. Physiol., 359 (1985) 47-63. 9 Holland, K.D., Yoon, K.W., Ferrendelli, J.A., Covey, D.F. and Rothman, S.M., Gamma-butyrolactone antagonism of the picrotoxin receptor: comparison of a pure antagonist and a mixed antagonist/inverse agonist, Mol. Pharmacol., 39 (1991) 79-84. 10 Huettner, J.E., Glutamate receptor channels in rat DRG neurons: activation by kainate and quisqualate and blockade of desensitization by Con A, Neuron, 5 (1990) 255-266. I I Juiz, J.M., Helfert, RH., Wenthold. RJ., Bias, A.L. and AItschuler, R.A., Immunocytochemical localization of the GABA A/benzodiazepi,e receptor in the guinea pig cochlear nucleus: evidence for receptor localization heterogeneity, Brain Res., 504 (1989) 173-179. 12 Limbird, L.E., Cell Surface Receptors: A Short Course on Theory and Methods, Martinus Niijhoff, Boston, MA, 1986. 13 Macdonald, R.L., Rogers, C.J. and Twyman, R.E., Kinetic properties of the GABA A receptor main conductance state of mouse spinal cord neurones in culture, J. Phy.viol., 410 (1989) 479-499. 14 Pl~tneau, D.K. and Mayer, M,L., Structure-activity relationships ft~, amino acid transmitter condidates acting at N-methyl-t)~lsp,rtat¢ and qutsqualatc receptors. J. Neuroa'i.. !0 (1990)23852399. 15 Persohn, E., Malhcrbu, P., t, ld Richards, J.G., In situ hybridi~ation histochcmtstry reveals a diversity of GABA^ receptor subinit messenger RNAs in neurons of the rat spinal cord and dorsal root ganglia, Neuroscien('t,, 42 (1991) 497-507, 16 Sigel, E,, Baur, R,, Trube, G., Mohler, H, and Malherbe, P.0 The effect of subunit composition of rat brain GABA^ receptors on channel function, Neuron, 5 (1990) 703-711. 17 Twyman, R.E., Rogers, C.J. and Macdonald, R.L., Intraburst kinetic propertie~ of the GABA^ receptor main conductance state of mouse spinal cord neurones in culture, J. Physiol., 423 (1990) 193-220. 18 Verdoorn, T., Draguhn, A., Ymer, S., Seeburg, P.H. and Sakmann, B., Functional properties of recombinant rat GABA^ receptors depend upon subunit composition, Neuron, 4 (1990) t)!9-928. 19 Vicini, $., Pharmacologic significance of the structural heterogeneity of the GABA^ receptor-chloride ion channel complex, Neurol~ychopharmacolo~,, 4 (1991) 9- I 5, 20 Weiss, D.S. and Magleby, K.L., Gating scheme for single GABA-activated CI- channels determined from stability plots, dwell-time distributions, and adjacent interval durations, J. Neu. ros¢'i,, 9 (1989) 1314-1324. 21 White, G., GABAA-receptor-activated current in dorsal root ganglion neurons freshly isolated from adult rats, J. Neurophysiol., 64 (1990) 57-63. 22 White, G., Lovinger, D.M. and Weight, F.F., Ethanol inhibits NMDA-activated current but does not alter GABA-activated current in an isolated adult mammalian neuron, Brain Res., 507 (1990) 332-336. 23 Yasui, S., Ishizuka, S. and Akaike, N., GABA activates different types of chloride-conducting receptor-ionophore complexes in a dose-dependent manner, Brain Res., 344 (1985) 176-180.
56
BRES 17811
Heterogeneity in ECs0 and n n of G A B A A receptors on dorsal root ganglion neurons freshly isolated from adult rats Geoffrey White Neurogen Corporation, Branford, CT 06405 (USA) (Accepted 21 January 1992)
Key words: y-Aminobutyric acid; Inhibition; Dorsal root ganglion; Pat-;h-clamp; Ion channel
GABA activates a CI- current through the GABA A receptor/ionophore complex that influences excitability of neurons. Studies using expression of cloned cDNAs coding for different GABAA receptor/ionophore subunits suggest that the EC.~o and Hill coefficient for GABA are influenced by subunit composition. However, no direct evidence for such heterogeneity has been reported for vertebrate neurons, i have investigated the heterogeneity of EC.~o and Hill coefficients (n H) of isolated dorsal root ganglion neurons using the whole-call patch clamp technique. The ECso for GABA varied from 26 to 107 ~M among neurons, n H calculated from the logistic equation varied from 1.18 to 2.0. A negative correlation was found between the ECs~ and n H (r ffi-0.81). Both n H and ECso differed between some cells. However, in some instances, n H differed between cells while ECso values were similar, and in other cells, ECs(~ values differed and n n was similar. In addition, when cells were categorized according to action potential shape, the ECso and Hill coefficients differed among cell types in some instances and were similar in other instances, These findings demonstrate that different pharmacological profiles for GABA can be observed in adult mammalian neurons. Selective distribution of such pharmacological subtypes of GABA A receptors may contribute to control of neuronal excitability.
INTRODUCTION GABA is the major inhibitow neurotransmitter in the mammalian brain. GABA can influence neuronal excitability by activating a Cl~current through the GABAA receptor/ionophore complex. This complex is a hetero-oligomeric protein for which a variety of candidate polypeptide subunits have been identified 4.~. Both the mRNAs coding for different subunits and their polypeptide products are differentially expressed within the brain and possibly within neurons them. selves ~'~. Recent studies using cloned mRNAs and cDNAs coding for different GABA^ receptor/ ionophore subunits suggest that subunit composition can influence the value of ECs0 and Hill coefficients (n~t) ''~x. Taken together, the above findings raise the possibility that expression of different subunit combinations within the nervous system can result in a tope. graphical pattern of pharmacological and physiological properties of GABAA receptor/ionophore complexes m'. Interestingly, no direct evidence from voltage-
clamp studies in vertebrate neurons has been reported that demonstrates such heterogeneity in EC~, and n~t for the GABA^ receptor. In fact, a high degree of homogeneity has been observed (EC~t}~ 10-50 ~M and n H ffi 1.8-2), based on averages of concentration. response curves generated from a number of neurons L~''~. Three possible ways to account for this lack of heterogeneity are: (i) ECso and n H are the same in all neurons, (ii) ECs0 and nit were the same in neurons in the cited studies but are different from other neurons, and (iii) EC~, and nit were heterogeneous among indvidual neurons and the apparent homogeneity resulted from averaging responses from different neurons, l have tested the first possibility by evaluating concentration-response curves from individual neurons isolated from dorsal root ganglia. MATERIALS AND METHODS
Cell isolation Dorsal root ganglia (DRG) were dissected from male SpragueDawley rats (200-500 g) and placed in oxygenated DMEM with
Correspondence: G. White, Neurogen Corporation, 35 Northeast Industrial Rd., Branford, CT 06405, USA. Fax: (1)(203)481-8683.
57 NaCI substituted for NaHCO 3 (340 mosm), 0.4 mg/ml type I!I trypsin, 0.3 m g / m l type VII collagenase, and 0.1 mg/mi type IV DNase I!. The DRG containing enzyme-rich solution was incubated at 35°C in a shaking water bath for 45-60 rain. Individual neurons were dissociated from the ganglia by vigorous shaking after which 0.4 mg trypsin inhibitor {Sigma type I! from soybeans) was added. Neurons were kept at room temperature (20-22°C) in the DMEM solution and were aliquoted as needed to a 35 mm petri dish. Neurons were devoid of processes as described previously 22.
Electrophysiological responses were obtained using the whole-cell patch-clamp technique with an Axopatch ID amplifier taxon Instruments) and pClamp software tAxon Instruments) as described previously 2t. Subsequent to gaining access to the cell interior, membrane potential with 0 current injection was negative to - 50 mV and input resistance was > 500 GO in cells included in this study. Membrane potential was subsequently voltage-clamped to - 60 mV. Series resistance (3-4 M,Q) was compensated by 90%. Current amplitude was < 1.5 hA. Therefore, errors due to series resistance were negligable. External solution contained (raM): 150 NaCI, 5 KCI, 2 CaCI 2, 1 MgCI 2, 10 HEPES, 10 glucose, osmolarity was adjusted to 320 using sucrose and pH was adjusted to 7.4 using NaOH. Recording electrode solution contained (raM): 130 KCI, ! CaCI,, 2 MgCI~, II EGTA, 10 HEPES, 5 mM MgATP, osmolarity was adjusted to 310 using sucrose and pH was adjusted to 7.4 using KOH.
Hill coefficient and can be interpreted as a measure of cooperativity and minimal number of binding sites when agonist/receptor/ response relationships are known to fit a specific set of constraints. However, such constraints are more often met in drug binding studies than in physiological studies7.12. For electrophysiological experiments in which agonist activated current flow is the measured endpoint, these constraints include that charge transfer per unit time be equivalent for each molecule of agonist bound. In the absence of this constraint, non-interacting binding sites could appear to be interacting in terms of binding affinity if total charge transfer per unit time increased or decreased disproportionately each time an additional agonist molecule bound to a given complex. Such confounds might be present in the case for the GABA A receptor/ ionophore complex because multiple receptors or channel types with different charge transfer characteristics may be recruited at different concentrations of GABA 23 and bursting substates may exist which depend upon the number of agonist molecules bound, as well as on the concentration of agonist presentt7. For this study, n H will therefore be interpreted as no more than a measure of steepness of the concentration-response relationship for a given neuron. To facilitate a statistical comparison of ECso and n H among neurons (using t-tests), amplitudes of current evoked by increasing concentrations of GABA were transformed to a linear relationship using the Hill plot. For Hill plots, data was selected to surround the estimated ECs0, thus data points at the extremes were discarded for this part of the analysis ~2.
Drug rt,sllon,~;e$
RESULTS
Recording
The concentration-response analysis was carried out on neurons ranging from 20 to 30 /~m in diameter. GABA was applied to neurons using a 17-barrel delivery system composed of a linear array of 200 ~ m (i.d.) fused silica tubing (Hewlett Packard). Delivery rate of drug containing external solution was ,,, 90 t~m/ms calculated from a flow rate of 175/~l/minute. Superfusion of neurons with drug containing external solution was initiated by opening a valve after placing the delivery barrel to within I(10/~m of the neuron. Drug delivery was terminated by closing the valve and removing, the delivery pipets from the bath. Sometimes cells were exposed to drug-containing solution by supeffusing a cell in non-drug-containing inflation and then switching directly to a barrel from which drug-containing solution was flowing, in this case, drug delivery was terminaled by moving back to the barrel from which non.drug.containing solution was flowing. Results t'rom either method were comparable within a given neuron as reporled by others "~. With repealed exposure to u given concentration of drug by either method or between methods within a given cell usually resulted in amplitude differences of less than 5%. Occassionally, a difference of less than 10% was observed. Concentrations of GABA were applied in ascending order. In 6 neurons the initial concentration of GABA (5 or 10/zM) was applied after the final application of the highest GABA concentration to verify stability of response amplitudes. The final response amplitude under these circumstances was within 10% of the response amplitude which preceded the series of drug applications. At least 2 rain were allowed to elapse between application of concentrations of GABA < 25/~M and at least 7 min elapsed between applications of > 25/~M GABA. Drug responses were collected on a Gould chart recorder model RS3200 and evaluated as described previously~t.
oan'e /itting Curve fitting was carried out using KaleidaGraph (Abelbeck Software) on a Mac II {Apple Computer). Concentration-response curves were fit to the logistic equation:
Heterogeneity in concentration response to GABA
Fig. 1 illustrates heterogeneity among neurons in parameters of their concentration-response relationship to GABA. A family of currents evoked by increasing concentrations of GABA is illustrated in Fig. I A. The response amplitudes from Fig. 1A are plotted in Fig. I B and were fitted to the logistic equation. When concentration-response curves of this type were compared among neurons a heterogeneity in EC.~0 and n~ was oberved (Fig. IC). From left to right in Fig. IC the calculated ECs0s for individual concentration-response curves were 26, 63, and 100/~M GABA and nil values were 2.0, 1.49, and 1.23 (r: > 0.99 in each case). A Hill plot was constructed from these data for the purpose of statistical comparison of concentration-response parameters among neurons (Fig. ID). EC.~0 values calculated in this manner were 24, 62, and 102 pM and n H values were 2.0, 1.45, and 1.24 (r2> 0.99 in each case). The three slopes differed from one another (P < 0.01, combined probability from multiple t-tests, d f = 8 for individual tests) as did the EC50 values (P < 0.01, combined probability from multiple t-tests, d f - 8 for individual tests).
!--/max X l / ( l + ( E C s 0 / [ d r u g ] ) " "
Con'elations among ECho, n n, and maximum response amplitude
where i is current amplitude, /max is the calculated maximum current amplitude, ECso is the concentration of agonist that produced the half maximum current amplitude, [drug] refers to a given drug concentation, and n H is a factor that describes the steepness of the concentration-response curve, n H is generally referred to as the
A significant degree of variability in the EC50 for GABA can be explained by the variability in n H (Fig. 2A). A correlation between specific membrane current and either EC50 or nrl was not statistically ~ignificant
58 A 2.2
1.8
SO
150
~
2SO:i 3
IV"
O
5
1.6
1.,
1
0
~
2 3 Seconds
4
r = O.81 p< O.O1
1.2
5
1
1.'2
350
3
1.'4
1'.6 l:a log EC50
c:o-2'0o
250 < Ir~
1.~0
l
1
5
4 2
50
r~ R"--'r",r~
0.5
........ • I o
IO0.
I000
GABA(pM)
I
o
E
1
-o.s
lOO
o
7s -1,5, o
5O
log GABA (pM}
2~
0 !0
i1 1. S. . . . . . os
1000
GABa {IJM)
o
-
100
o
---~o" 1 , s
tog GABA (IjM) Fig. !. DRG neurons respond to application of increasing concentra. tions of OABA in a heterogeneous manner. A: family of currents activated by application of increasing concentrations of GABA, Current records were digitized at 143 Hz for the purpose of presentation. Application lasted 15 s. Only the first 5 s are shown for clarity, v axis is in picoampures (pA), B: plot of current amplitudes ,ql.wn in A, The response to 400 /~M was not included in A for clarity. The curve was generated from the logistic equation, C: normali/.ed current amplitudes fit to the logistic equation show difference~ in EC.~. and n . . Data was normalized to lm,,x from the I.gistic equation when applied to data before normalization. Data in the right hand curve are from A. See text for further details. D: Hill plot of data from C. Of all the neurons evaluated, the greatest variance was observed for the curve illustrated on the left of Fig, IC and D. However, this variance is so small that the neuron is still considered to be representative. S.E.M.s for ECs, and n H values are shown in Fig. 2A.
Fig. 2. Diversity of EC~0 and n . values, A: ECs, for GABA shows a negative correlation with nip r - correlation coefficient, P is from a t.test, df - 15. Hill plots of cells marked I, 2, 3 are shown in Fig. IC and $.B,M,s are indicated for cells I, 2, 3, 4, and 5, B: Hill plot of c~lls marked 2, 4, 5 in A, Cell 2 is also shown in Fig, IC and D.
(r = 41 and 38, respectively, P > 0,05, t-test, df = 15). Because cell size of freshly isolated DRG neurons can be an indicator of pharmacological properties (e.g. refs. 10 and 21), the relationship between whole.ceU capacitance and ECho, nil, and current density were evaluated, r was between 0 and +0,10 in each instance. The plot of Fig. 2A suggested a number of combinations of EC,~, and nil values could be observed in mammalian neurons, as suggested from studies using expression of cloned mRNAs encoding various G A B A ^ receptor subunits t", As shown in Fig, 1, cells can differ from one another in both nit and ECso values. As shown in Fig, 2B, some cells (cells 2 and 4) can share ECs, values (62 vs, 59 ~M, P > 0,1, df = 8) and differ in nit values (1,45 vs. 1,'/2, P < 0,001, df = 8) while other cells (cells 2 and 5) can have similar n H values (1,45 vs. 1,54, P > 0.1, df = 8) and differ in ECso values (62 vs. 34 pM, P < 0.001, d f = 8). Cells 4 and 5 differed from one another in both parameters (P < 0.02, combined P values, df = 8 for individual tests). As can
59 be seen from inspection of Fig. 2A, some cells did not differ in either value.
Effect of averaging concentration-response curves across neurons" Concentration-response curves are generally represented in the literature as averages from many neurons. The normalized amplitudes of the concentrationresponse curves from this study were averaged, plotted, and fitted to the logistic equation (Fig. 3). The ECs0 was 51 /~M and n H was 1.48 (r2> 0.99). It is not obvious from the small error bars, and the large r 2 value for the regression that this curve is composed of a heterogeneous set of curves.
Comparison between methods for calculating slopes of concentration-response curves The slope of a concentration-response curve can be calculated in at least two ways: (1) from a range of agonist concentrations using the logistic equation (n H1) or (2) from the maximal slope of the concentration-response curve on a log-log plot (nil2). This can be achieved by dividing the ratio of response amplitudes evoked by doubling agonist concentrations by 2 (e.g. ref. 3). Naturally, agonist concentrations must be within the range of the maximal slope. The findings of Yasui etal. za and Twyman etal. t7 would predict that an n H value calculated from lower concentrations of GABA might be different from that calculated at higher concentrations of GABA. n , 2 was calculated from the slope of the concentration-response crave at 5 and l0 ~M OABA and divided by 2. hill and nil2 were different in most instances and no correlation was observed (Fig. 4), In addition, no correlation was observed between nlj2 and EC~0 ( r - 0.03). It was possi-
."2
E
Q 1.S , - -
o£ .
I
-r
I ~ I f
e-
I ~/~j'~6 ._,,I •
*~,. .., o
.
e.7
•
u -
• e
~
u
u
1
1,5
2
nH f r o m
--
r = 0.07 p> 0.2 '
' l
2.5
10 pM/5 pM
Fig. 4. n n l calculated from the logistic equation is different from nil2 calculated from low concentrations of agonist. The thick line shows actual linear regression. The thin line shows theoretical distribution if both methods resulted in similar n H values. See text for details.
ble that nH calculated from 5 and 10 ~M GABA was underestimating the true n H (see ref. 3). If this were the case then the difference between n H calculated in the different manners should be related to EC~0 values. Again, no significant correlation was observed (r ffi 0.38, P > 0.2), suggesting that a large systematic error due to variation in ECsos was not introduced. In addition, the curves fit by the logistic equation were extrapolated to regions of low concentrations of GABA (0,I ~M) and the experimental values for n , 2 were compared to the theoretical value of n~j2 for the slope between 5 and 10 ~M GABA. In each case, the difference between the idealized value and the empirical value for nil2 could not be accounted for by either: (1) variability in response measurements or (2) the deviation of agonist concentrations from the area of the maximal slope (not shown).
Expression of ECso and nn as a function of cell type
150ECso = 5'1 IJM 0 4.J
I1
.
o~
CO 100 o~
I=L
~g
C
5o
0
.
1
i
10
100
1000
GABA (pM)
Fig. 3. Concentration-response averaged from all neurons, n---14, 17, 17, 17, 17, 15, 8, and 7 for 5, 10, 25, 50, 100, 200, 400, and 800 ttM GABA respectively. Bars show standard errors. The apparant absence of error bars in some instances resulted from small standard errors.
It is of interest to determine whether different cell types can exhibit different pharmacological properties for the GABA A receptor ionophore complex. Different types of neurons can express different combinations of mRNA coding for subunit variants of the GABA A receptor/ionophore complex I'~'l'~. If GABA^ subunit composition can contribute to the expression of different ECs0 and n}~ values, then it might be expected that these values could be different in different cell types. Within the size range of neurons examined, different shapes of the action potential were observed. Action potential shape has been taken as an indicator of subtypes of DRG neurons s'21. The action potential waveform (top) and the derivatized waveform (bottom) in Fig. 5A are different from those in Fig. 5C
A
B
/
/~ EC50=34laM ~ H=1"87
c
A EC5o=37HM
D
A
ECso=1071JM
A ECbo-991JM
-5' rnV7~......._.._. -56 mV~~'-~-40mV ~oo v/sL5m$ Fig. 5. Different cell types can express different or similar EC~. and n n values. Upper waveforms are reconstructions of action potentials digitized at 50 kHz. Lower wavcforms are first derivatives of action potential waveforms. Action potentials evoked by 600 pA pulses of 2-3 ms duration, as indicated by the initial square deflection in the derivatized wavefi)rms. EC~. and n H were calculated from fittin~ the GABA concentration-.vcspons¢ data from each cell to the h)ljistic equation,
were
GABA, (2) averaging of concentration-response data across neurons can mask underlying heterogeneity, (3) a negative correlation exists between ECs0 and n n calculated from the logistic equation, (4) n n calculated from low concentrations of agonist can differ from n n calculated from a wide range of agonist concentrations in adult mammalian neurons, and (5) different populations of neurons can express similar or different ECso and n H values. Other pharmacological and physiological properties of the GABA A receptor/ionophore complex were not investigated in the present study, Such properties include rectification of GABA activated current and sensitivity to agents which alter receptor/ionophore function. These properties may also be sensitive to subunit composition ~t'. Thus it should be noted that neurons that express similar ECs0 and nit values may differ in other properties and neurons that have different ECs0 and n n values may be similar in other properties. Effect of enzyme treatment on GABA a responses One of the great disadvantages of studying freshly isolated neurons is the ever present possibility that one is studying the effects of enzyme treatment itself. The reason for using freshly isolated DRG neurons is that they are electrically compact, thus providing an ideal space clamp model. The present study could not be reliably carried out in cultured neurons due to problems controlling voltage and ionic gradients in distant processes. Arguing against effects of treatment are the observations that over the course of many studies using the present methods every freshly isolated DRG neuron has shown a GABA^ type response to application of GABA (n > 1000), conventional benzodiazepine pharmacology (n > 75), and conventional responses to pentobarbitai and various GABA antagonists (n > 25). Nonetheless, it is still possible that subtle effects of enzyme treatment gave rise to the observed differences in EC,~, and n , reported for tiffs study. The range of EC~. and nil values observed in this study can not be accounted for entirely by enzymatic differences between preparations, For example, in preparation 32991, the entire range of parameters was observed. The EC,~,s were 37, 82, and 107 #M and the respective n n values were 1.71, 1.18, and 1.24. Such variability could still be accounted for by enzymatic differences ~mong neurons within a preparation.
enzyme
and D as are their respective EC.~. and ntl values. This indicates that different neuronal types can display different pharmacological properties for GABA. O:~ the other hand, while the waveforms arc also different in Fig. 5A and B, their respective EC~0 and n , values indicate that different cell types can also express similar pharmacological properties. Whether individual types ot neurons as defined by their action potential waveforms can exhibit different pharmacological properties is an equally interesting question but was not addressed due to the difficulty in determining when action potential waveforms are the same. For example, while the waveform is different between Figs. 5A and B, it is not obvious whetller the differences between Figs. 5B and D are sufficient to constitute a separate populations of neurons. DISCUSSION This study has demonstrated that (1) adult rat neurons can display a variety of EC~0 and nit values for
Determbzation of response amplitudes In this study, two possible contaminants of the shape of the concentration/r~sponse curves are desensitiza-
61 tion and changes in intracellular CI- concentration. Because response amplitudes were measured at their peak, significant desensitization would result in an underestimation of peaks and a flattening of the concentration-response curve, A similar situation could result if Cl- redistributed significantly during application of GABA. Peak amplitudes were not measurabIy altered by desensitization. The rates of desensitizatien were biexponential and on the order of seconds and tens of seconds with no clear pattern among neurons that was attributable to ECs0 or n H (data not shown, 10 neurons evaluated). The rates were similar to those reported by Akaike et al. ~. At concentrations of GABA at which significant desensitization was observed, response amplitude calculated as described in Materials and Methods were indistinguishable from response amplitudes calculated by back extrapolating the response amplitude to time zero from responses digitized at 143 Hz. The empirically observed rise times of responses in this study were also independent of ECs0 and n H and were similar to those described by Akaike et al. ! (e.g. on the order of tens of milliseconds for concentrations of GABA ~ 100/~M). Given that the rise time of the response was several orders of magnitidue faster than the decay time, it is not surprising that no difference was detected between the two methods of measuring response amplitude. Under the conditions of this study (large somata, high i~ltracellular CI-, and small current amplitude), a contamination of response amplitude arising from changes in intracellular C l ' concentration would not be expected to measurably contribute to the response amplitude evoked by application of GABA 2.
On the interpretation of n u Two methods were used for calculating n H. The values of n H calc:llated by the two methods were different within neurons and apparently unrelated, however both methods resulted in heterogeneity in n H among neurons. One explanation for why nH differed within neurons is that concentration dependent events may occur which alter nil. In fact, such a finding has been reported on the single channel level insofar as n H is reduced considerably over the concentration range of 0.5 to 5/.tM 13. The lack of a correlation between n H when calculated by the two methods indicates that the factors responsible for expression of nHl and nil2 were not the same in every neuron. This suggests that the pharmacological properties of the GABAA recept o r / ionophore complex may be quite intricate.
On the relevence of the negative correlation between n u and ECho While cooperativity between binding sites on the GABA A receptor/ionophore complex is difficult to prove (as discussed in Materials and Methods), it is also difficult to rule out. If true cooperativity in binding does occur, then the negative correlation would be consistent with the increase in ECso resulting in part from an increase in cooperativity. The existence of the negative correlation between n H and ECso was not obvious from studies using expression of cDNA to evaluate differences in pharmacological properties of GABA A receptor subtypes ~6. The negative correlation was modeled using a logarithmic fit for simplicity. In fact, the fit to a logarithmic function was only marginally better than a linear fit (r = -0.81 vs. -0.76, respectively). The true form of the actual relationship remains to be deterrained from a larger sample of neurons. A larger sample number could, for example, highlight a step function resulting from discrete combinations of subunits. The continuous nature of the data could result from the presence of heterogeneous populations of independent binding sites within individual neurons. While the presence of such separate non-interacting binding sites generally results in an n H of < 1, if the ECs0 values are separated by less than ~ 10-fold, then the measured ECs0 will reflect the multiple ECs0 values and n H can be greater than 1 (e.g. Fig. 3). Such a phenomenon could also contribute to the differences detected in n H1 and n H2. This contribution of heterogeneous independent binding sites to GABA A receptor pharmacology is difficult to address in the absence of complete control over factors such as lipid environment, posttranslationai modifications, and suhunit stoichiometry. Effect of averaging concentration-response curves Averaging concentration-response curves from different neurons without first comparing their respective slopes and ECso values makes interpretation and comparison among cells from different regions of the nervous system or different animal species difficult because the average may be composed of several populations of neurons as defined by concentration-response to GABA. This situation may hold for other ligandactivated channels as well. For example, the ATP-gated inward current may be composed of a heterogeneous population of receptors 3 and capsaicin-evoked inward currents may also arise from a heterogeneous set of receptors because adult neurons exhibit different sensitivities to the drug, one in the tens of nM range and
62 the other in the hundreds of nM range -'l (unpublished data). Factors that may contribute to differences in pharmacological properties D i f f e r e n c e s in pharmacological properties described in this p a p e r could be a c c o u n t e d for by differing combinations o f variants of G A B A A r e c e p t o r / i o n o p h o r e subunits, posttranslational modifications, or o t h e r metabolic processes. It is likely that heterogeneity in subunit c o m b i n a tions contributed to the pharmacological heterogeneity observed in this study. Such an influence has b e e n reported for G A B A A responses in n o n - n e u r o n a l expression systems treated with various combinations of G A B A A subunit c D N A s 1~''!8. In addition, a recent report by Persohn et al. ~5 indicates that D R G n e u r o n s express m R N A for G A B A A receptor subunits in a h e t e r o g e n e o u s manner. Pharmacological properties were evaluated for this study in a select group of neurons, thus w h e t h e r heterogeneity of subunit composition exists in this particular group of neurons r e m a i n s to be d e t e r m i n e d . it is possible that posttranslational modifications and metabolic factors could also contribute to heterogeneity in EC5, and n~l values; however, little is yet known about such interactions.
Physiological significance of pharmacological heterogene. ity This study has demonstrated that pharmacological properties of the GABA^ receptor/ionophore complex differ among cell types. Small (10-20%) changes in the strength of inhibitory systems can significantly alter neuronal excitability s. Thus, subtle differences in GABAergic pharmacological properties may contribute to local control of neuronal excitability. To what extent differences in ECs() and Hill slopes contributes to inhibitory strength is unclear given that concentrations of GABA at the synapse may be quite high. it may be that differences betwe,:rt these two parameters simply accompany other parameters such as channel open time, susceptibility to metabolic control, or sensitivity to various pharmacological agents. Differences between neurons in these parameters could result in differences in inhibitory strength. REFERENCES I Akaike, N., indue, M. and Krishtal, O.A., 'Concentration.clamp' study of gamma-aminobutyric acid-induced chloride current kinetics in frog sensory neurones, J. Physiol., 379 (1986) 171-185. 2 Akaike, N., lnomata, N., and Tokutomi, N., Contribution of chloride shifts to the fade of gamma-aminobutyric acid-gated currents in frog dorsal root ganglion cells, J. Physiol., 391 (1987) 219-234.
3 Bean, B.P., ATP-activated channels in rat and bullfrog neurons: concentration dependence and kinetics, I. Neurosci., 10 (1990) 1-10. 4 Bormann, J., Hamill, O.P. and Sakmann, B., Mechanism of anion permeation through channels gated by glycine and gammaaminobulyric acid in mouse cultured spinal ncurones, I. Physiol., 385 (1987) 243-286. 5 Changnac-Amitai, Y. and Conners, B.W., Horizontal spread of synchronized activity in neocortex and its control by GABAmediated inhibition, J. Neurophysiol., 61 (1989) 747-758. 6 Choi, D.W. and Fischbach, G.D., GABA conductance of chick spinal cord and dorsal root ganglion neurons in cell culture, J. Neurophsyiol., 45 (1981)605-620. 7 DeLean, A.P., Munson, P.J. and Rodbard, D., Simultaneous analysis of families of sigmoidal curves: application to bioassay, radioligand assay, and physiological dose-response curves, Am. J. Physiol., 235 (1978) E97-EI02. 8 Harper, A.A. and Lawson, S.N., Electrical properties of rat dorsal root ganglion neurones with different peripheral nerve conduction velocities, J. Physiol., 359 (1985) 47-63. 9 Holland, K.D., Yoon, K.W., Ferrendelli, J.A., Covey, D.F. and Rothman, S.M., Gamma-butyrolactone antagonism of the picrotoxin receptor: comparison of a pure antagonist and a mixed antagonist/inverse agonist, Mol. Pharmacol., 39 (1991) 79-84. 10 Huettner, J.E., Glutamate receptor channels in rat DRG neurons: activation by kainate and quisqualate and blockade of desensitization by Con A, Neuron, 5 (1990) 255-266. I I Juiz, J.M., Helfert, RH., Wenthold. RJ., Bias, A.L. and AItschuler, R.A., Immunocytochemical localization of the GABA A/benzodiazepi,e receptor in the guinea pig cochlear nucleus: evidence for receptor localization heterogeneity, Brain Res., 504 (1989) 173-179. 12 Limbird, L.E., Cell Surface Receptors: A Short Course on Theory and Methods, Martinus Niijhoff, Boston, MA, 1986. 13 Macdonald, R.L., Rogers, C.J. and Twyman, R.E., Kinetic properties of the GABA A receptor main conductance state of mouse spinal cord neurones in culture, J. Phy.viol., 410 (1989) 479-499. 14 Pl~tneau, D.K. and Mayer, M,L., Structure-activity relationships ft~, amino acid transmitter condidates acting at N-methyl-t)~lsp,rtat¢ and qutsqualatc receptors. J. Neuroa'i.. !0 (1990)23852399. 15 Persohn, E., Malhcrbu, P., t, ld Richards, J.G., In situ hybridi~ation histochcmtstry reveals a diversity of GABA^ receptor subinit messenger RNAs in neurons of the rat spinal cord and dorsal root ganglia, Neuroscien('t,, 42 (1991) 497-507, 16 Sigel, E,, Baur, R,, Trube, G., Mohler, H, and Malherbe, P.0 The effect of subunit composition of rat brain GABA^ receptors on channel function, Neuron, 5 (1990) 703-711. 17 Twyman, R.E., Rogers, C.J. and Macdonald, R.L., Intraburst kinetic propertie~ of the GABA^ receptor main conductance state of mouse spinal cord neurones in culture, J. Physiol., 423 (1990) 193-220. 18 Verdoorn, T., Draguhn, A., Ymer, S., Seeburg, P.H. and Sakmann, B., Functional properties of recombinant rat GABA^ receptors depend upon subunit composition, Neuron, 4 (1990) t)!9-928. 19 Vicini, $., Pharmacologic significance of the structural heterogeneity of the GABA^ receptor-chloride ion channel complex, Neurol~ychopharmacolo~,, 4 (1991) 9- I 5, 20 Weiss, D.S. and Magleby, K.L., Gating scheme for single GABA-activated CI- channels determined from stability plots, dwell-time distributions, and adjacent interval durations, J. Neu. ros¢'i,, 9 (1989) 1314-1324. 21 White, G., GABAA-receptor-activated current in dorsal root ganglion neurons freshly isolated from adult rats, J. Neurophysiol., 64 (1990) 57-63. 22 White, G., Lovinger, D.M. and Weight, F.F., Ethanol inhibits NMDA-activated current but does not alter GABA-activated current in an isolated adult mammalian neuron, Brain Res., 507 (1990) 332-336. 23 Yasui, S., Ishizuka, S. and Akaike, N., GABA activates different types of chloride-conducting receptor-ionophore complexes in a dose-dependent manner, Brain Res., 344 (1985) 176-180.
