European Journal of Pharmacology - Molecular Pharma~v#Jgy Section, 225 (19921321-330

321

~:2 1992 Elsevier Science Publishers B.V. All rights resepeed 0922-4106/92/$05.00

EJPMOL 90287

Effects of s u b u n i t types of the r e c o m b i n a n t GABA A receptor on the response to a neurosteroid S h a h i d H. Z a m a n , R y u z o S h i n g a i ', R o b e r t J H a r v e y ', M a r k G. D a r l i s o n ' a n d E r i c A. B a r n a r d MRC Molecular Neurobiolo~, Unit, MRC Centre, Cambridge. UK

Received 11 September 1991, revised MS received 27 December 1991. accepted 7 Janua~, 1992

When vertebrate brain poly(A) + RNA is expressed in Xenopus oocytes the response of the GABA receptors formed is lound to be inhibited allosterically by a neurosteroid, pregnenolone sulphate (PS). This negative modulation was reproduced after expressing RNAs encoding bovine GABA a receptor subunits it_, the combinations ai +/3t, or ai + ~l + y2 (where i = 1.2 or 3). The characteristics of this inhibition vary significantly with the type of the a subunit ( a l . a2, or a3) used. When the bovine y2L alternate form of the 3,2 subunit was replaced by the human 72S subunit, the behaviour was unchanged: the human y2S subunit used is a newly-cloned form, which encodes a polypeptide with two amino acid differences from the human 3,2 subunit previously described. The results of co-application of PS and 3o-hydroxy-5a-pregnan-ol-20-one, a neurosteroid which is a positive modulator of the GABA a receptor, indicate that these act at different sites on the receptor. PS also increases the desensitisation of the receptor by GABA. This effect, also, is e~-subunit-type dependent and occurs by an acceleration of the fast phase of desensitisation. Neurosteroid; GABA A receptor; Cloned subtypes; Xenopus oocyte expression

1. Introduction In addition to the well-documented allosteric modulations of the activity of vertebrate brain G A B A A receptors by benzodiazepines, barbiturates, /3-carbolines and several convulsants such as picrotoxin and t e r t - b u t y l b i c y c l o p h o s p h o r o t h o t h i o n a t e ( T B P S ) (reviewed by Barnard, 1988), the modulation which is exerted by certain steroids has a particular interest, since these occur in the brain (Baulieu, 19811 and therefore may be natural regulators of this receptor. Thus, the gonadal and adrenal steroids such as progesterone and deoxycorticosterone respectively, and their A-ring reduced nmtabolites (e.g. 3a-hydroxy-5a-pregnan-ol-20-one (3a-OH-DT-IP) and 3 a - p r e g u a n - 5 a - 2 t diol-20-one ( 3 a - T H D O C ) ) are positive allosteric modulators of G A B A A receptors on neurones (Majewska et al., 1986; reviewed by Gee, 1988; Schumacher and

Correspondence to: Professor E.A. Barnard, MRC Molecular Neurobiology Unit, MRC Centre, Hills Road, Cambridge CB2 2QH, UK. Tel. 44 (223) 402400; Fax 44 (223) 402483. i Present addresses: Institut ftir Zellbiochemie und klinische Neurobiologic, UniversitSts-Krankenhaus Eppendorf, Universit{it Hamburg, Martinistrasse 52, 2000 Hamburg 20, Germany (M.G.D. and R.J.H.) mid Laboratory of Bioscience, Facully of Engineering, lwate University, 4 Ueda, Morioka 202, Japan (R.S.).

McEwen, 1989). Certain other steroids which have actually b e e n found to be produced in the brain (Corp~chot et al.. 1983; Hu et at., 1987; Majewska and Schwartz, 1987) also modulate the action of G A B A at the G A B A A receptor, i.e. pregnenolone, dehydroepiandrosterone and their sulphates (pregnenolone sulphate (PS) and dehydroepiandrosterone sulphate (DHS), respectively) (Carette and Poulain, t984; Majewska et al., 1988, 1990; Demirg~ren et at, 1991). T h e r e has been some indication from binding studies that the site of action on the G A B A ; , receptor structure of PS is different from that of the reduced A-ring metabolites of progesterone (MNewska et a l , 1986; M ~ e w s k a and Schwartz, 1987; G e e et al., 1989; L6pez-Cotom6 et al, 19901. D e p e n d i n g on the neurosteroid, the action of G A B A can be augmented or antagonised. Thus, 3ce-OH-DHP and 3 a - T H D O C potentiate the response to G A B A by prolonging the burst duration of the channels (Callachan et aL, 1987; Lambert et ak, 1990), whereas the sulphates PS and D H S have a suppressive effect on the G A B A response (Carette and Poulain, t984; Majewska and Schwartz. !987). In cultured neurones from several tissue sources PS at concentrations above t0 # M produces a non-competitive antagonism of the G A B A - i n duced current, but She degree of inhibition is independ e n t of the m e m b r a n e potential (Majewska et al.,

322

tively, were introduced into these primers to facilitate cloning. Reactions contained: 1 /~1 of each oligonucleotide (0.2 ~ g / ~ l ) , 5/~1 10 x Taq DNA polymerase buffer (500 mM KC1, 100 mM Tris-HC1, pH 8.3, 0.1% (w/v) gelatin), 2 ul 100 mM MgC12, 8 /~i 1.25 mM each dNTP, ~ 5 ng human fetal brain first-strand cDNA, 0.5 ~1 of Taq D N A polymerase (Perkin-Elmer, Cetus; 5 U / ~ I ) , and distilled water to a final volume of 50/~l. Amplification was for 40 cycles of 94°C for 1 min (denaturation) and 72°C for 3 rain (annealing and extension), with a final extension at 72°C for 7 rain. Analysis of the PCR by agarose gel electrophoresis revealed a single product of ~ 1.7 kb, which was cloned into PstI- and BamHI-restricted pBluescript II SK + (Stratagene) and sequenced. The plasmid used here, phGRy2sense, contains a eDNA that is seen to have two nucleotide changes in the coding region, resulting in two amino acid substitutions, when compared with the published sequence (Pritehett et al., 1989). Residue 81 in the mature polypeptide is a threonine instead of a methionine; since a threonine residue is found at this position in the sequences deduced from three human y2-subunit cDNAs (independently isolated) that were characterised during this study, and is also found in the rat (Shivers et al., 1989) and the chicken (Glencorse et al., 1990) 72-subunit polypeptides, this difference is unlikely to be due to PCR error. We also find residue 336, in the mature subunit, which occurs in the proposed intraceUular loop region, to be a serine rather than an asparagine. Since the encoded human y2 subunit described here lacks an 8 amino acid insertion in the intracetlular loop (Whiting et al., 1990), we refer to it as the 72S subunit.

1988). Moreover, PS decreases channel opening frequency without affecting channel burst duration (Mienville and Vicini, 1989). These effects show that the inhibition by PS is not due to open channel block, but rather to binding at a site where it allosterically opposes the opening of the channel induced by GABA. A variety of isoforms of the a,/3 and y subunits of the GABA A receptor (reviewed by Olsen and Tobin, 1990) have been deduced from complementary DNA (cDNA) cloning studies. Their various combinations, when expressed in vitro (in Xenopus oocytes or mammalian cells), can show differences in eleetrophysiological and pharmacological properties (Blair et al., 1988; Levitan et al., 1988a, b; Barnard et al., 1989; Pritchett et al., 1989; Ymer et al., 1989; Malherbe et al., 1990; Puia et al., 1990; Sigel et al., 1990; Verdoorn et al., 1990; Shingai et al., 1991). Particular hetero-oligomerie combinations of these subunit isoforms are believed to occur in different neurones or regions or developmental stages of the brain (Wisden et ai., 1988, 1989; Shivers et al., 1989; Olsen and Tobin, 1990). To understand their different functions, it will be necessary to recognise differences between particular subtypes of t h e GABA A receptor in the actions of its natural modulators. The neurosteroids, in particular, being established as such, are of much interest for this purpose. Studies on some subunit combinations in vitro with the positive modulators in this series have been made in this (Shingai et al., 1991) and another (Puia et al., 1990) laboratory, showing some specific subunit isoform influences. We now take this further by showing subunit requirements and isoform-dependent differences in the activity of the negative modulator, PS.

2.2. In vitro transcription and RNA extraction

2. Materials and methods

Capped transcripts were synthesised from the bovine G A B A A receptor a l - (Schofield et al., 1987), a2-, o.3(Levitan et al., 1988a) and/31-subunit (Schofield et al., 1987) cDNAs as described previously. The plasmid, phGRy2sense, containing the human 72S-subunit eDNA was linearised using BamHI; R N A was transcribed from this and from the bovine y2L-subunit cDNA using T7 R N A polymerase. The cDNA encoding the bovine 72L subunit was as described by Whiting et al. (1990); that eDNA was generously provided by Dr. P. Whiting (Merck Sharp & Dohme Research Laboratories, Harlow, UK). RNAs were polyadenylated with poly(A) polymerase (Drummond et al., 1985); this increases the expression of some subunit types. Each R N A was verified as giving a single band of the predicted size after eleetrophoresis in a 1.5% agarose/ formaldehyde gel. Chick whole brain pol~(A) + R N A was extracted by a standard guanidine isothiocyanate method (Barnard and Bilbe, 1987).

Z1. Isolation of human GABA A receptor y2-subunit cDNAs Oligo(dT)-p~imed first-strand cDNA was synthesised (Watson and Jackson, 1985) from 1/~g of 8-weekold human fetal brain poly(A) + RNA, a generous gift from Dr. M. Goedert (Laboratory of Molecular Biology, Cambridge, UK). The polymerase chain reaction (PCR) was performed using this cDNA as template and two 35-base oligonueleotide primers that were designed to correspond to portions of the 5'-(sense: 5 '-ATGGTATCTGCAG G C G ] T I T T G C T G A T C T F ATCTC-3') and 3'-(anti-sense: 5'-GTGAATAGGATCCATATCAGTAAAACCCATACCTC-3') untranslated regions of the published human GABA A re¢eptor y2-subunit eDNA sequence (Pritchett et al., t989). By incorporating base mismatches, either a PstI site (underlined) or a BamHI site (underlined), respec-

~:

'

~

.

.

.

.

.

.

/~

'

:

!:!~:

i¸

:::i

?

:

323

2.3. Oocyte injection and electrophysiology Recombinant RNAs (10 ng each) in the stated combinations were co-injected into Xenopus oocytes of stage 5 or 6, which had been previously defolliculated by incubation in modified Barth's medium (Barnard and Bilbe, 1987), omitting Ca 2+ but with the addition of 1 m g / m l collagenase (Sigma type II), at 20°C for 50 rain, followed by manual defolliculation. Chick brain poly(A) + RNA was used as in Van Renterghem et al. (1987). After 2 - I 0 days of culture at 18°C (with daily changes) in modified Barth's medium supplemented with 2.5 mM pyruvate, gentamicin (25 m g / l ) and benzyl penicillin (12 rag/i), the oocytes were placed in 120 p.l wells, and (except where noted) voltage-clamped at - 6 0 mV by two 1-5 MD electrodes filled with 3 M KCI. The oocytes were superfused with amphibian saline: (in raM) 100 NaC1, 2 KCI, 1.8 CaCI 2, plus 5 HEPES or 10 Tris-HCl. Drugs (freshly dissolved) were bath-applied in this saline, with a dead time of 5 s. The pH of the saline and of each drug-containing solution was adjusted to 7.4. Oocytes were washed with the saline for 15-25 rain between GABA applications, and for 15-40 rain between steroid applications depending on the concentrations. Reference recordings of the current evoked by GABA alone were ust~ally made before and after at least three successive recordings of (GABA + steroid); cases where significant drift therein was seen were discarded. For constructing the concentration-response plots, the magnitudes of the peak currents induced were employed, and (except in the special cases noted) at least three oocytes were used for each point. The overall mean value is reported for the responses to (usually) three applications of the same concentration of a compound to each of these ooc!,'tes.

3. Results

3.1. Responses to GABA without and with PS RNAs encoding a bovine a subunit ( a l or a 2 or a3) and the /31 subunit (listed below as /3), with or without a y2-subunit RNA, were co-injected into oocytes. When the y2 subunit was used this was the human form (T2S) in most of the experiments, but when it was replaced by the bovine y2L subunit no difference in the results was obtained. The bovine y2L subunit is 99% identical in sequence to the human y2S subunit, apart from the octapeptide insertion in the deduced intraceltular loop region which distinguishes y2L from the shorter y2S (Whiting et al., 1990; Kofuji et al., 1991). Dose-response curves with GABA over the concentration range 10 -7 M to 10 -3 M are illustrated in fig. 1 for two ternary, combinations. The

I00~

i

T

W

20 0

!0-8

, ---'-n 10-7 10 6 10-5 10-4 10-3 Concentration of GABA [M]

Fig. 1. Concentration-response relationships for the peak current ( m e a n ± S . E . M . ) induced by G A B A in oocytes expressing either a l / 3 y (closed circles) or a3/3y (open circles) receptors. The arrows show, for comparison, the midpoints for the corresponding binary. combinations (Levilan et al., 1988b). When, in a test made with al~gy, the h u m a n y 2 S subunit replaced the bovine y 2 L the same curve and ECru value were obtained.

corresponding curves of ai/3 binary combinations (midpoints shown here by arrows) were as reported pi,:viously (Levitan et al., 1988b). Both these terna~" combinations (whether the human or bovine 3,2) showed a weaker (by 2-fold) GABA potency compared to the corresponding bina~, expression with cd and a3. Some co-operativity of the GABA response was restored for a t f l by the presence of either y, with the Hill coefficient rising from < 1 to 1.2 in the ternary combination. From such curves, a concentration of GABA was chosen for each bina~" or ternary combination which would produce 40-60% of the maximum response, for use in comparing the effect of PS on the response to GABA. At 10 - ~ to 10 -~ M PS, about 50% of all oocytes expressing GABA a receptors tested (even within a given batch) gave a reproducible augmentation of the GABA-induced current. This effect was present with each a isoform and in both the binary and the ternary, combinations. At higher concentrations PS was inhibitory and that effect was manifested, in contrast, on all of the injected oocytes (figs. 2A, 3A, 3B). A comparison was made with the effect of PS co-applied with GABA on oocytes expressing a general population of GABA A receptor subunit RNAs, i.e. after injection with poly(A) + RNA t¥om whole brain (in this case chick, which expresses particularly strongly (Houamed et al., 19:34; Smart et al., 1987)). With this, PS likewise decreased the response to GABA; this effect was just detectable at 10 -7 M PS, but the effect was reproducible in alt oocytes tested. At higher PS concentrations there was a dose-dependent increase in the inhibition (fig. 2B). The effect was reversed upon washing, as was true, too, with all of the recombinant combinations tested. A plo~ of such data for many points over the range 10-t~ M ~o 10 -a M PS (not shown) gave the

324

A. 0~11~

n

u

n 1 rain

B. Chickwhole brain

case in fig. 4. Hence, it must be assumed that for full equilibration with PS, pre-application for at least 30 s is needed. Longer pre-incubations with PS than this are not advantageous, since they introduce the resistance to wash-out and are therefore leaving a pool of PS in the oocyte membrane. All further studies were, thus, performed with PS pre-equilibrated for 30 s and remaining in the bathing medium when the G A B A was added to this.

3.3. Binary subunit combinations

~--2~ i~ 4f--5~40rlA2rairl Fig. 2. Effects of PS on currents induced by GABA. (A) Oocytes injected with a l + f l RNAs, using (1) 5 /~M GABA, (2) 5 #M GABA+10 -9 M PS, and (3) 5 #M GABA+10 -5 M PS. (B) Oocytes injected with chick whole brain poly(A) + RNA using (1) 10 gM GABA, (2) 10 /zM GABA+10 -~ M PS, (3) l0 /tM GABA+ 10 -7 M PS, (4) 10 gM GABA+ 10 ~6 M PS, apd (5) (after a 15-min wash) 10 ~M GABA. Solid bars show the period of application of GABA (or GABA + PS).

The IC50 for inhibition by PS is 2 - 3 x 10 - 6 M for the three binary combinations studied (fig. 5A). There was no statistical difference between them at any concentration of PS. PS above 10 -7 M, with GABA, generated a biphasic current in some batches of injected ooc3"tes: after the normal inward current (at V h = - 6 0 mV) evoked by GABA, a small outward

A. 0~21} ~

n

~

~

IC50 of PS as 10 -5 M. In these latter experiments only slight augmentation by PS was seen at 10-to M, since the PS was not pre-applied (cf. Section 3.2).

3.2. Equilibration with PS The effect of PS was found to be greater when the oocytes were pre-incubated with PS, even for a few seconds, before the co-application of G A B A and PS (compare fig. 3A, recordi,qgs 2 and 3, or 4 and 5). The G A B A response was then potentiated at low concentrations (10 - t ° to 10 -8 M) of PS, in about 50% of all oocytes tested, with any of the combinations. This was, therefore, as before, but on average the effect was easier to detect (as in fig. 3B) than with co-application. In all cases, as before, higher PS levels inhibited the response to GABA. Fig. 3B illustrates this for one combination, with potentiation and inhibition appearing successively in the same oocyte with increasing PS concentration. We noticed that the G A B A responses frequently became larger in the later stages of successive applications of PS on a given oocyte, despite intervening extensive washes which were ample when G A B A alone was used. This is attributed to a small amount of PS persisting in the membrane, due to its lipophilicity, even after an extensive attempted wash-out of the drug. This parallels the potentiation seen with the pre-application of PS at very low concentrations. The inhibitory action of PS, with every combination, increased with the period of pre-application, but reached a plateau within 30 s, as is illustrated for the a313y

mm

.....

I 5o A I

5

2

rain

B. o~3py Ell 2

3

lOOnA

Fig. 3. Potentiation and inhibition of GABA-induced currents by PS: (A) The cx2+ 13 combination. Currents induced by (1) 3/xM G.ad3A, (2) 3 /.*M GABA+ 1 0 - 7 M P S , ( 3 ) pre-application of 10 -7 M PS (open bar) for 40 s followed by 3 /.tM GABA+10 -7 M PS (filled bar), (4) 3/xM GABA+ 10 -6 M PS, and (5) pre-application of 10 -6 M PS for 40 s (open bar) followed by 3 /.tM GABA+10 -6 M PS (filled bar). (B) The ~3+/3 + Y combination. (1) 10 btM GABA, (2) pre-application of 10 -9 M PS for 90 s (open bar) followed by 10 ~M GABA+ 10 -9 M PS (filled bar), and (3) pre-application of 10 -7 M PS for 60 s (open bar) followed by 10/xM GABA+ 10 -7 M PS (filled bar).

325

O

100~..~.~

..... @

4020-

o--2'o

,:0

8'0 1;0

Pro-application time of PS (s) Fig. 4. The dependence of the response to PS upon pro-application. The GABA-induced current was potentiated by 10 -9 M PS (open circles), but inhibited by l0 " M (filled circles) and 10 -6 M (triangles) PS, all illustrated in the same oocyte expressing the a3 +/3 + 3' combination. The latter two effects are increased by pro-application.

current occurred (fig. 2A, pane~ 3, and fig. 3A, panels 4 and 5). Such an outward current is also seen in these cases when PS is present alone, and is described further below. The reversal potentials of the currents induced by G A B A were (on different oocytes) - 2 0 , - 2 0 and - 2 5 mV for a l +/3, and - 2 0 and - 2 0 mV for a 2 +/3.

3.4. Ternary combinations The addition of the 3' subunit changed the sensitivity to PS inhibition in combinations containing either

the a 2 or a3 but not the a l subunit (fig. 5B). The IC~:j values for PS acting on the subunit combinations at/3y, a2/3y and a313y were approximately 4, 1, and 0.4 × 10 -6 M, respectively. Statistically, the 10-fold difference produced by replacing a l by a3 is significant at P < 0.001 (at 10 -6 M PS). The augmentation at 10 -~ M PS is not seen with every, combination in fig. 5A and B, because the results with all oocytes tested (n = 7 in each group, i.e. either with or without y) were included here. Given that about 50% of these oocytes gave the augmentation, random sampling of the oocytes would result in the overall mean values for the plots, in the potentiation range only, being very. subject to sampiing errors.

3.5. Currents induced by PS alone An application of PS alone at above 10 -~ M produced an ionic current in the binary combinations. PS generated negligible currents (less than 2 nA) in uninjected ooc3-tes. The magnitude of this current in injected oocytes (seen in fig. 4A, panels 3 and 5) was quite small and variable, depending on the batch of ooeytes (up to 20 nA at the peak for 10 -6 M PS). This current was outward at potentials more negative than its reversal potential, It was measured as - 2 5 to - 2 0 mV for c~l +/3 and - 2 0 mV for a 2 +/3; these values are close to the reversal potential of the GABA-evot-ed current (see Section 3.3). There was no change in this reversal potential when the extraceltular potassium concentration was increased to 100 mM for c~l +/3 (data not shown). The current increased gradually with

1.2

~6}

t

.

]

1,1

o

0.9

rr

~

O

~

rr c o

0

0.8

0.70.5-

rn