Neuron,

Vol. 5, 703-711, November, 1990, Copyright

0 1990 by Cell Press

The Effect of Subunit Composition of Rat Brain GABA* Receptors on Channel Function Erwin Sigel: Roland Baur: Gerhard Trube,+ Hanns MGhler,* and Pari Malherbet *Department of Pharmacology University of Bern Bern Switzerland +Pharmaceutical Research F. Hoffmann-La Roche Ltd. Base1 Switzerland *Department of Pharmacology University of Zurich Zurich Switzerland

Summary Different combinations of cloned rat brain subunit isoforms of the GABAA receptor channel were expressed in Xenopus oocytes. The voltage-clamp technique was then used to measure properties of the CABA-induced membrane currents and to study the effects of various modulators of the CABAA receptor channel (diazepam, DMCM, pentobarbital, and picrotoxin). This approach was used to obtain information on the minimal structural requirements for several functional properties of the ion channel. The combination a5B2y2 was identified as the minimal requirement reproducing consensus properties of the vertebrate CABAA receptor channel, including cooperativity of CABAdependent channel gating with a K, in the range of 10 PM, modulation by various drugs acting at the benzodiazepine binding site, picrotoxin sensitivity, and barbiturate effects. Introduction Biochemical isolation of a GABAJbenzodiazepine receptor complex from bovine brain, followed by analysis using gel electrophoresis under denaturing conditions, indicated the existence of two subunits, which were named a and p (Sigel et al., 1983; Sigel and Barnard, 1984). Primary peptide sequences derived from the purified material enabled molecular cloning of two subunits from bovine brain (Schofield et al., 1987). Subsequent functional expression in Xenopus oocytes led to ion channels, claimed to have many of the properties known for the native GABAA receptor channel, in particular benzodiazepine sensitivity of the GABA-induced ion current (Schofield et al., 1987). It soon became clear that bovine brain contains additional subunit isoforms (Levitan et al., 1988a; Ymer et al., 1989a) and that expressed a-8 subunit combinations lack benzodiazepine responsiveness in the low micromolar concentration range and display noncooperative GABA gating of the channel (Levitan et al., 198813; Malherbe et al., 1989a). The functional defi-

ciencies are not simply due to the inability of the Xenopus oocyte to assemble GABA* receptor channels properly, since expression of mRNA derived from whole brain leads to expression of channels exhibiting the same pharmacological properties (Parker et al., 1986; Sigel and Baur, 1988) as measured in vertebrate neurons (Yakushiji et al., 1989; Akaike et al., 1990). Subsequent cDNA cloning revealed a bewildering number of distinct subunit isoforms (reviewed by Schofield, 1989; Olsen and Tobin, 1990). The molecular cloning of multiple subunit isoforms from bovine, human, and rat brain (Schofield et al., 1989; Pritchett et al., 1989b; Ymer et al., 1989a, 1989b; Lolait et al., 1989; Shivers et al., 1989; Khrestchatisky et al., 1989; Malherbe et al., 1990a, 1990b, 199Oc; Pritchett and Seeburg, 1990) strongly supports the notion of GABAA receptor heterogeneity (reviewed by Sieghart, 1989). Although it is clear from biochemical work that not more than four or five subunits may be accommodated to form a channel complex (Sigel et al., 1983), it is not clear, so far, which subunits form a functional ion channel in situ. A precise description of the functional properties of the channels expressed from different subunit combinations is clearly needed. In the a-j3 combinations, which lead to expression of large GABA-gated currents, the respective isoform of a subunit has been reported to affect the affinity for GABA gating (Levitan et al., 1988a; Malherbe et al., 1990a) and the pharmacological properties (Pritchett et al., 1989a) of the channel. In the case of human receptor subunits, combined transfection into 293 cells of the al and PI subunit together with the ~2 subunit led to full sensitivity to benzodiazepine receptor ligands (Pritchett et al., 1989b), whereas for the same rat subunit combination expressed in the Xenopus oocyte, only a weak sensitivity was found (Malherbe et al., 1990b). On the other hand, it has been impossible to express a channel displaying cooperativity to GABA gating. Furthermore, most functional studies performed so far with cloned subunit isoforms lack detailed analysis of the channel properties. Very often drugs were applied at a single concentration only. The results of these studies are very difficult to interpret, since, for example, the concentration dependence of drugs acting at the benzodiazepine binding site is biphasic (Sigel and Baur, 1988; Yakushiji et al., 1989) and may vary for channels formed by different subunit combinations. In this paperwe have focused on the functional role of six rat brain GABAA receptor subunit isoforms, al, a3, a5 (this subunit has also been named a4 [Khrestchatisky et al., 1989]), pl, 82, and ~2. These subunits were functionally expressed in different combinations in order to explore the contribution of the different subunit isoforms to the properties of GABA-gated receptor channel. The GABA concentration eliciting

NWVXl 704

half-maximal current (K,) and the Hill coefficient for GABA gating, along with the effect of drugs acting at the benzodiazepine binding site and at the barbiturate site, were investigated. The study led to the assignment of some of the functional properties to defined subunit isoforms. In addition, a minimal composition of a5B2y2 has been found to display the behavior known for the GABAA receptor channel in situ. Results Six cloned rat GABAA receptor subunit isoforms were available for this study. In principle, there exist 63 different subunit combinations, whose detailed functional analysis clearly exceeded our working capacity. Therefore, we decided first to analyze functional expression of the subunits individually and then to proceed with pairs and triples of subunit combinations, with the aim to define minimal structural requirements for the individual functional parameters. Homomeric Channels and Suppression of Their Formation As reported for some of the subunits earlier (Sigel et al., 1989), oocytes injected with RNA coding for a single subunit showed either little response to the application of 100 PM GABA (Table 1) or no response at all. This is in agreement with the report by Khrestchatisky et al. (1989), but in contrast to that by Blair et al. (1988), who reported single-channel recordings from expressed individual subunits. The Bl subunit resulted in the formation of an anion-selective channel that was open in the absence of GABA, but closed in response to the channel blocker picrotoxin (Sigel et al., 1989). It is interesting to note that coinjection of stoichiometric amounts of all the available a subunit RNAs, but not of the ~2 subunit RNA, together with the Bl RNA suppressed the formation of this nongated anion channel (Table 1). GABA-Dependent Gating of the Channel: Affinity and Cooperativity Subunit combinations consisting of two subunit isoforms displayed different affinity for GABA-dependent gatingof the channel. a5 in combination with Bl led to formation of channels displaying a higher affinity toward GABA compared with aIs1 and a3Bl and led to some cooperativity in the opening of the channel by GABA (Malherbe et al., 1990a). It is also interesting to note that a subunits are not strictly needed in order to observe GABA-activated currents, since B2y2 was sufficient to induce these currents (Table 1). However, from the combinations studied, it seems that the current amplitudes are much smaller in the absence of a subunits than in their presence. In the dual combinations studied, no functional channels were observed in the absence of a B subunit. Coexpression of y2 together with alB1 or alB2 induced in each case a Ffold increase in the K, for channel gating by GABA. Replacement of Bl by B2, in

the subunit combinations alB1 and alBly2, led in both cases to a 13-fold increase in K,. Thus, a, 8, and y subunits affect the affinity of GABA gating. None of the dual and triple subunit combinations mentioned so far displayed cooperativity for GABA-dependent gating of the channel, with the possible exception of a5Bl. Subsequently, all six subunit isoforms available to us were coexpressed, although biochemical size determinations exclude that hexameric subunit isoforms form a single-channel complex (Sigel et al., 1983). To our surprise, an apparently homogeneous population of ion channels was expressed and displayed cooperative gating by GABA. This encouraged us to try other combinations with a smaller number of subunits. Expression of five different subunits led to results that were difficult to interpret. This was due to K, shifts toward lower values, which were observed for some combinations during the expression period. The values shown in Table 1 were obtained at least 4 days after injection of the RNAs. Except for the combination that lacked a5, all combinations showed strong cooperativity for GABA gating, again pointing to a role of this subunit isoform in cooperativity. Omission of a5 also significantly increased the K, for GABA. In line with the postulated role of y2 in mediating the effect of diazepam (see below), the combination lacking y2 completely lost modulation by diazepam. Subsequently, we returned to combinations with a smaller number of subunits. We found that combining ~2, the subunit conferring benzodiazepine sensitivity, with a5, the subunit probably responsible for cooperativity and high affinity for GABA-dependent channel gating, and 82 led to the formation of an ion channel with the desired properties (Table 1; a5B2y2). Addition of either al or a3 to a5B2y2 had little effect on the propertiesof the resulting channel. We subsequently found that neither the ~2 nor a B subunit was required for cooperativity of GABA gating (Table 1, u3a5y2 and ala582) and that a B subunit was not required for diazepam sensitivity (a3a5y2). However, in the absence of a B subunit, the expressed current amplitudes were small. A typical example of the determination of the affinity and cooperativity of GABA-dependent gating is shown in Figure 1. An oocyte was exposed to increasing concentrations of GABA, and the log of the current amplitude elicited versus the log of the GABA concentration was plotted. The limiting slopes from at least three experiments performed with three different oocytes were averaged to obtain the Hill coefficient, n (Table 1). At the end of the experiment, the oocytes were exposed to 10 mM GABA in order to obtain an estimate of the maximal current amplitude, and the K, was estimated. In the above experiments emphasis was put on the lower GABA concentration range. Full dose-response curves (Figures 2a and 2b) were also evaluated in order to obtain K, and n for some subunit combinations. In the experiments shown in Figures 1 and 2, oocytes were injected with the

Functional Architecture of CABAA Receptor Channels 705

Table 1. Functional

Subunits

injected

Chick Braina

Properties

of Different

lsoform Combinations

after Expression

in the Xenopus

CABA

Picrotoxin

Pentobarbital

I,,,

I

I (%I,,,; 1 mM)

(nA)

- 2500

al a3

n n n n n - 3

(10/2) (8/2) (6/2) (8/2) (10/2) (6/2)

W/2 P2Y2

>3000 >3000 >3000 - 890 n n n - 16 - 155

(25/6) (25/4) (26/4) (7/2) (13/2) (7/2) (6/2) (9/l) (7/2)

aluSe ala5y2 a3a5y2 alply2e alpZy2 a3ply2 a3P2y2 a5ply2 a5P2y2

>2000 n - 470 >3000 >3000 >3000 >2000 >3000 >3000

(7/l) (7/2) (8/2) (34/6) (12/4) (3/l) (4/l) (3/l) (23/3)

ala3P2y2 ala5P2y2 a3a5!32y2

>3000 (8/l) >3000 (7/l) >3000 (7/l)

ala3a5plp2s ala3a5fily2s ala3a5P2y2s ala3plp2y2s ala5Blp2y2s a3a5blp2y2s

>3000 >3000 >3000 >3000 >3000 >3000

ala3a5plp2y2

>3000 (1012)

;;15b P2 Y2 alplc a3DlC a5PlC aID aly2 a3y2 a5y2

Subunit

K, (cLM)

H”

21

1.7

13 17 1.1 139

*7 * 11 * 0.7 * 18

(411) (3/W (5/l) (311)

64

f 11

(311)

18

*3

147 74 985 240 487

* + f f f

17 14

*7 *3

145

(3/l ) (4/l) (3/l) (3/l) (5/l) (4/l)

39 13 352 136 42

0.9 1.0 1.2 1.1

n

* f * *

0.1 0.0 0.1 0.1

(3/l) (4/Z) (5/2) (3/I) (3/l)’ (3/l)

1.7 1.0 1.0 1.2 1.3 1.6 1.7

(3/l) (7/2)

* 75

(3/l)

*I *5

(311) (311)

9 44 20

* 22

(l/V (3/l) (l/l)h

13 f 4

n n n 408 n n

(10/2) (3/I) (3/l) (4716) (6/I) (3/I)

28 5 10 n n

(712) (712) (912) (6/I) (13/2)

* f * * * + *

0.1 0.0 0.1 0.0 0.0 0.0 0.1

* f * * * *

88 6

+I

(l/W cwh

17

*7

(3/l)

62

f 28

(5/2)

1.6 f 0.2

-2+5

VI

-1

(6)

f IO

44 f 13

-2*4 15 * 6

30 f 10

198 * 49

(5)

4 * 16

(3)

176 * 61

(4) (13) (IO) (3) (3) (3) (4)

32 f 8

(7) (3)

20 (6/2) n (6/2)

98 f 15 65 f 16

n (l/l)

29 f 9

15 * 108 * 23 * 299 * 31 * 136*41

16 f a 34 f 14 12 f 3

355 * 163 (4) 191 * 26 (3) 16Oi26 (3)

1.9 * 0.1 1.7 f 0.0 2.0 f 0.0 1.7 1.4 1.4 1.2 1.5 1.5

Stimulation by Diazepam (o/o;1 PM) 130

132 f 13

196 (9/l) n (6/I) 1.6 f 0.0

21 34

(nA; 10 PM)

Oocyte

0.1 0.1 0.1 0.1 0.1 0.1

-5 57 150 113

f * _t *

7 37 25 50 14

5 13 25 21

(3) (3) (3) (3)

35 f 14

(3) (3)

36 f 18

(4)

a+3 n

6r2

In the body of the table, n represents