Neuron,
Vol. 2, 53-60,
January,
1990, Copyright
0 1990 hy Cell Prets
Apparent Desensitization of NMDA Responses in Xenopus Oocytes Involves Calcium-Dependent Chloride Current John l? Leonard and Stephen R. Kelso Department of Biological Sciences University of Illinois at Chicago Chicago, Illinois 60607
Summary N-Methyl-o-aspartate (NMDA) receptors were expressed and studied in Xenopus oocytes injected with rat brain RNA. NMDA application elicits a rapid inward current that decays in several seconds to a relatively stable level. This decay is reportedly due to desensitization. However, we found the early transient component could be evoked more than once during a single application of NMDA, suggesting that the receptor did not actually desensitize. Removal of external Ca*+, replacement of Ca2+ with Ba2+, or intracellular injection of ECTA abolished the transient component. Furthermore, a variety of Cl- channel blockers nearly eliminated the transient component and inhibited the plateau current as well. We propose that a significant portion of the NMDA current recorded in oocytes is carried by a transient inward Cl- current triggered by Ca2+ influx through the NMDA receptor/channel. Introduction The N-methyl-o-aspartate (NMDA) subtype of receptor for the neurotransmitter glutamate is receiving much attention because of its apparent role in a wide range of phenomena, including the triggering of longterm synaptic modification, developmental plasticity, susceptibility to hypoxic brain damage, and epilepsy. Several properties distinguish it pharmacologically and physiologically from two other subtypes of glutamate receptors, commonly referred to by their specific excitatory amino acid agonists, kainate and quisqualate. The gating of the NMDA receptor is unique in its dependence upon both ligand binding and membrane potential (via open channel block by Mg*+; Mayer and Westbrook, 1987; Mayer et al., 1984; Nowak et al., 1984). It also exhibits a regulatory site for glycine (Kushner et al., 1988; Verdoorn et al., 1987; Johnson and Ascher, 1987), an inhibitory modulatory site for Zn*+ (Westbrook and Mayer, 1987; Peters et al., 1987), open channel blockade by certain compounds (Mayer et al., 1988; Huettner and Bean, 1988; MacDonald et al., 1987), and a significant permeability to Ca*+ (Burgoyne et al., 1988; Hori et al., 1985; MacDermott et al., 1986; Murphy and Miller, 1988). In addition, the NMDA receptor can be reliably identified by the use of specific competitive antagonists (Watkins and Olverman, 1987). Recently, several investigators have exploited the capability of the Xenopus oocyte to express functional ionic channels after injection of mRNA to study
currents mediated by brain-derived excitatory amino acid receptors. These studies have generally established that all three subtypes of glutamate receptor can be expressed and that responses evoked by NMDA in RNA-injected oocytes exhibit most of the same pharmacological and physiological characteristics as neural preparations. These include blockade by specific antagonists (Verdoorn et al., 1987), a voltage-dependent block by Mg*+ (Kushner et al., 1988; Verdoorn et al., 1987), requirement for glycine (Kleckner and Dingledine, 1988; Watson et al., 1989), and open channel blockade by phencyclidine receptor agonists (Fang et al., 1988; Kushner et al., 1988). Although detailed studies of the ionic selectivity properties of NMDA receptors have not yet been completed, it is generally assumed that the receptor in oocytes is also permeable mainly to Na+, K+, and Ca*+, as is the case in cultured neurons. In the oocyte, one of the distinguishing characteristics of the NMDA response is a rapid initial current that decays in a few seconds to a relatively steady level; a characteristic that has been assumed to reflect a receptor desensitization (Kushner et al., 1988; Lerma et al., 1989). However, the native Xenopus oocyte expresses a Cl- channel that is activated by rises in intracellular Ca2+ concentration (Barish, 1983; Miledi, 1982). If the NMDA receptor in RNA-injected oocytes is permeable to Ca*+, as it is in neurons, then it is quite possible that Ca2+ influx could trigger the Ca*+dependent Cl- current, Ic,(ca), that is normally present. Since Ec, is approximately -20 to -30 mV in the oocyte (Barish, 1983), IcI(caI would be inward at the hyperpolarized potentials typically used in oocyte studies of NMDA responses. The fact that Ecr is so close to the NMDA reversal potential makes it difficult to separate the two currents electrophysiologically. In this study we have examined the NMDA response under a variety of conditions that either reduce Ca2+ influx or inhrbrt IcI(caI. We find that the major portion of the early phase and some of the late phase of the NMDA response in these cells is carried by Ca2+dependent Cl- current. Much of the apparent desensitization of the NMDA receptor observed during bath application to an oocyte is an artifact of the Xenopus expression system brought about by the activation of the endogenous, transient IcI,Ca,. Results In oocytes injected with rat brain RNA and voltageclamped at -80 mV, bath application of IO-100 PM NMDA in standard Mg*+-free saline typically resulted in a large inward current of 50-300 nA (n = 43 cells). All five different RNA preparations used reliably expressed NMDA receptors upon injection into oocytes. No NMDA responses were ever seen in noninjected cells (n = 10). As illustrated in Figure IA, the waveform
A
was observed in 6 of 6 oocytes. A second signature of NMDA channels is a blockade by Mg2+. Addition of 1 mM Mg 2+ to the nominally Mg2+-free standard saline reversibly prevented elicitation of an NMDA response at -80 mV (n = 7). As reported by others, the block by Mg2+ was less complete at more depolarized levels (-40 to -30 mV; data not shown). Having verified that our NMDA response in oocytes is mediated by classic APV- and Mg2+-sensitive NMDA receptor/channel complexes, we proceeded to examine the response in more detail.
NMDA
1 200 nA 10 set
NMDA
‘::: Figure 1. Initial Decay Receptor Desensitization
+ Ma’+ L
NMDA
of NMDA
Response
Is Inconsistent
with
(A) Top: response to 50 s application of 50 PM NMDA. Bottom: in the same oocyte, addition of 1 mM Mg*+ at 25 s after start of 50 PM NMDA application led to a blockade of current and was followed on wash-out by a second transient inward current. Horizontal bars indicate duration of NMDA application. In this and subsequent figures in which Vhold = -80 mV, a 10 mV step to -70 mV prior to NMDA application was used to monitor changes in input conductance. (B) Voltage-clamp protocol (top) and current response (bottom). NMDA (50 PM) was applied during a depolarization to -30 mV and during the return to -80 mV. The large transient occurs at a time when receptors have already been continuously exposed to NMDA for 25 s.
of the NMDA-induced current is distinct from that evoked by other excitatory amino acids; it has a rapid onset, reaching a peak in l-2 s, and then decays within 5-10 s to a relatively stable level that lasts for as long as the NMDA is present. In these experiments the longest period of application was 75 s. The responses were very repeatable upon successive applications and could in fact be repeated within 30 s, which was approximately the shortest time in which the NMDA wash-out resulted in a return to the initial level of holding current. To verify that our NMDA response was mediated by specific glutamate receptors of the NMDA subtype, we coapplied NMDA and the specific antagonist DL-2amino 5-phosphonovaleric acid (APV). At IO PM, APV produced a potent, reversible inhibition of the NMDA response. At 100 PM, a complete and reversible block
The Early Transient Component Is Not Due to “Desensitization” The transient nature of the early part of the NMDA response in Xenopus oocytes has been ascribed to a desensitization of the receptor (Kushner et al., 1988; Lerma et al., 1989). However, under certain conditions, we could induce a fast, transient inward current late in the period of NMDA exposure, at a time when the receptors should presumably have been “desensitized.” Due to the unique voltage-dependent open channel block of the NMDA receptor/channel by Mg2+, it is possible to block current flow through the channel independently of agonist binding at the NMDA binding site. Thus, as illustrated in Figure IA, current flow through the channel could be blocked in the middle of continuous application of NMDA, after the early transient phase and during the stable portion of the response. Under these conditions, wash-out of the Mg2+ led to a resumption of inward current flow. However, the inward current did not return to the stable level that was obtained before Mg2+ blockade, but instead exhibited a rapid transient phase similar to the response at the very beginning of NMDA application (n = 3 oocytes). This result suggests that the NMDA receptors were not really desensitized during the late plateau phase. A transient inward current late in the period of NMDA application could also be evoked by another method that greatly reduces the NMDA response. Thus, a similar result occurred when the current was decreased in amplitude not by Mg2+ blockade, but rather by a change in holding potential that decreased the driving force (Figure 1B). In this case, NMDA was applied during a command pulse to -30 mV and was still present on return to -80 mV holding level. Although the NMDA receptors had ample opportunity to desensitize in the presence of NMDA at -30 mV, a large, early transient peak was immediately elicited on return to -80 mV (n = 4 oocytes). The Early Component Is Ca*+-Dependent The Xenopus oocyte normally expresses a Cl- conductance that is triggered by a rise in intracellular Ca2+ and is inward at -80 mV. When this current, Icl(ca), is evoked at +I0 mV by Ca*+ entry through endogenous voltage-dependent Ca2+ channels, it peaks in l-2 s and decays over a period of 5-10 s (Barish, 1983; Miledi, 1982). Other ways of increasing the intra-
Cl Component 5.5
ol NMDA
I” Xenopus
Response\
Oocyte\
cellular Ca2+ concentration, such as direct injection of Ca2+ from a puffer pipette (Miledi and Parker, 1984) or use of Ca2+ ionophore (Boton et al., 1989) or any neurotransmitter that produces IP3 (Oosawa and Yamagishi, 1989; Sugiyama et al., 1989), will also reliably induce lcI,caI. Because the NMDA receptor has been shown in a variety of preparations to be permeable to Ca2+, in addition to Na+ and K+, we examined the possibility that part of the NMDA response in oo-
were readily reversible, we did not attempt to examine NMDA responses under these conditions. The decrease observed during the late phase in Ca2+-deficient saline again suggests that some IcI(caI flows throughout the NMDA application. In the final series of experiments examining the role of Ca’+ in the early inward component, we injected EGTA into oocytes to chelate any Ca2+ that might enter through NMDA-activated channels. The success of the injection was assessed by monitoring IcI(ca) elicited by activating voltage-dependent Ca2+ channels with a step to +I0 mV (Figure 3A). Pressure injection of 0.5 M ECXA (up to 5 nl) usually reduced this response by as much as 90% within 4 min of injection (n = 8). As noted by others (Miledi and Parker, 1984), the time course and completeness of the blockade depended upon the amount of ECXA injected. Figure 3A shows outward I,-I(~~) currents at voltages more positive than Ec,, and Figure 3B shows NMDA-activated currents in the same cell at -80 mV, at which any Cl- current would be inward. In 6 cells injected with ECXA, the NMDA respor,se was measured after lc~(~~)
cytes is carried secondarily by ICI(ca). Thus, the early transient component might be mainly IcI(ca) that is triggered as soon as Ca2+ enters through NMDA channels and relaxes as a result of known Cl- channel inactivation (Boton et al., 1989). In an attempt to block or reduce the early inward component of the response, several methods were used to reduce Ca2+ entry during NMDA application. In one series of experiments, Ca2+ was replaced by Ba2+. Ba2+ normally permeates the NMDA channel, although it is also a partial blocker (Mayer and Westbrook, 1987). Moreover, Ba2+ is a poor activator of I CNCa) in oocytes (Barish, 1983). In 15 of 17 oocytes examined, the early peak was completely eliminated in Ba2+-substituted saline (Figure 2A). In experiments in which Ca2+ was lowered or omitted entirely from the perfusion medium, the early transient current was greatly reduced or absent (Figure 2B). When extracellular Ca2+ was lowered to 0.1 mM, both early and late components were reduced (66% + 5%, [mean f SEMI and 48% + IO%, respectively, n = 4). Similar results were obtained in nominally Ca2+-free saline, as the early component was decreased by 69% + 3% and the late component by 38% + 7% (n = 8). We attempted to reduce Ca2+ further by including 0.5 mM EGTA in the divalent-free perfusion medium. However, in media containing ECXA and no divalent cations, the oocytes often developed a large leakage current within l-2 min of switching to the divalent-free solution. Although the effects
A
Ca++-saline
Ba+‘-saline
NMDA
NMDA
was blocked. In these cases the early component of the NMDA response was also absent or was greatly reduced (56% + 6% decrease; see Figure 38). This result indicates that the early component depends on internal Ca2+. The ECXA injection experiments also suggest that the effect of changing external Ca2+ levels either by reduction (Figure2B) or by replacement with Ba2+ (Figure 2A) is actually via the ultimate effect on internal Ca2+. The NMDA Response Is Partly Cl--Mediated If Ca2+ entering the NMDA-gated channel actually produces a major current by activation of endogenous ICI(ca) channels, then one should be able to inhibit NMDA responses directly with compounds shown to inhibit ICI(ca) in oocytes (Boton et al., 1989; Snutch et al., 1986). We found this to be true for a variety of
Figure 2. Elimination ponent of NMDA ficient Salines
Ca++-saline NMDA
(A) Inward currents recorded at -80 mV during application of 50 uM NMDA. Middle trace shows a reversible block of the transient current and a partial block of the late phase in saline containing 20 mM Ba*+ in place of Ca’+. (B) Reduction of extracellular Ca*+ from 2.0 mM too.1 mM caused a reversible block of the transient component of the response to 50 uM NMDA at -80 mV.
I+-
B
2 mM
0.1 mM
Ca++
NMDA
NMDA
Ca++
of Early Transient ComResponse in Ca2+-De-
2 mM
Ca++
NMDA
=-y200 IIA 1 25 sac
Nt?UKNl 56
A
1’ post
Figure 3. Inhibition ot l(l,(.,, and Responses by Intracellular ECXA
EGTA
r
+10
+10
I
I
-80
-100
2’ post
6
-80
-100
NMDA
(A) Traces show the transient outward current that occurs during the indicated depolarization to +I0 mV. Responses were taken before (left) and after (right) pressure injection of -1 nl of 0.5 M ECTA via an intracellular micropipette. (B) In the oocyte used in (A). Inward currents were elicited at -80 mV by bath application of 50 PM NMDA before (left) and after (right) EGTA injection. Note the loss of the transient component of the NMDA response.
EGTA NMDA
NMDA
L
200 nA
10 set
early inward current (52% k 5%) and some decrease in the later phase (33% + 5%, n = 3). Again, similar results were obtained in 5 experiments using 500 PM 9-anthracene carboxylate. The peak current and latephase current were decreased 62% 5 3% and 25% k 5%, respectively. Under certain conditions, the contribution of the additional Cl- conductance to the NMDA response was reflected in the current-voltage relation. In most experiments, the reversal potential of the NMDA response was measured by stepping the command voltage to different membrane potentials, holding for approximately 25 s to allow voltage-activated currents to reach steady state, and then applying NMDA to measure the current induced during the steady phase 20-30 s after the beginning of the response. The rever-
Cl- channel blockers, including 9-anthracene carboxylate, 4+dinitro-stilbene 2,2-disulfonic acid (DNDS), niflumic acid, and flufenamic acid. Flufenamic acid and niflumic acid were the more effective blocking agents (K, = 30 and 19 PM, respectively; M. M. White unpublished data). At 500 FM, flufenamic acid effectively blocked ICI(caj evoked by Ca’+ entry via voltagedependent Ca2+ channels (Figure 4A). As expected, flufenamic acid also reversibly inhibited the early peak of the NMDA response (Figure 4B; 79% + 2% decrease in 6oocytes). There was also a smaller decrease in the later phase of the NMDA response (56% + 3% decrease). Similar results were obtained using somewhat higher doses of other Cl- channel blockers. The addition of 1 mM DNDS to the normal perfusion medium caused a large reversible reduction in the
Control
Flu
Figure3. Responses
Wash
L
600 nA
-\
1%x
I‘\
\ -A.
-.-A +10 -100
+10 -60
-100
NMDA
-60
NMDA
7 I--
lnhibltlorr by a Cl
or I,,,(,,, Channel
and NMDA Blocker
iA) Trace\ show the transient outward current during the indicated step depolarization. Responses were taken before (Control), during (Flu), and after (Wash) bath perfusion of 500 PM flufenamic acid. (B) In the oocyte used in (A), responses to 50 PM NMDA were reversibly inhibited bv the flufenamic acid.
Cl- Component
of NMDA
Response\
rn Xrnopus
Oorytes
57
The contaminating Cl- conductance was more pronounced when the current-voltage relation was assessed using ramp voltage commands. The reversal potential was shifted even closer to EC,, and additional nonlinearities were observed between -15 mV and +I0 mV (Figure 5A, continuous curve). In 5 cells we used ramp commands to measure the NMDA current-voltage relation in the presence or absence of 300 PM niflumic acid, which blocks the Cl--mediated component of the response. In all cases, the conductance was decreased (from 2.2 + 0.4 uS to 0.9 i 0.3 PS) and the nonlinearities between -15 mV and +I0 mV were reduced. Figure 5B also illustrates a small shift in the reversal potential away from EcI for the current-voltage relation measured in the presence of niflumic acid. In spite of the fact that ramp measurements indicated that the reversal potential for the NMDA response is near EcI, we consistently recorded substantial inward currents at -20 mV during long depolarizing steps (39 + 6 nA, n = 9). It is possible that a time-dependent inactivation of lc:~(,-~, (Boton et al., 1989) during long depolarizing commands results in a more pure NMDA current. This would explain the discrepancy between the apparent reversal potential of lNMDA measured by ramps (approximately -20 mV) and the observation of inward NMDA currents at -20 mV during long depolarizing commands.
i
-100 nA
Figure 5. Dependence of NMDA on the State of IcIIcdj Activation
Current-Voltage
Relation
Up-
(A) Current-voltage curves were obtained by a slow ramp command (continuous curve) or by applying 50 FM NMDA at least 25 s after initiating a 2 rmrn long voltage step to different potentials. In the long pulse method, currents were measured at the early peak (squares) or after the response had obtained a steady inward current (25 s, circles). The insert shows waveforms of currents recorded at the indicated holding potentials. (B) Current-voltage curves obtained by the ramp method in normal saline and in saline containing the Cl- channel blocker niflumic acid (Nif) at 300 PM.
sal potential measured in this manner (Figure 5A, circles) was -3 + 1 mV (n = 7), which is similar to that seen in neurons (Mayer and Westbrook, 1987). However, when the NMDA response was measured at the early peak (Figure 5A, squares), the greater conductance during the early phase was associated with a more hyperpolarized reversal potential. This is to be expected if a large part of the early component is mediated by Cl-, which reverses at Ecr g -20 mV. The inset in Figure 5A shows the expected change in waveform when the early transient Cl- component was eliminated by holding the membrane potential at -20 mV (~Ecr). Elimination of the early transient mV was quantified in 10 component at E, = -20 cells by taking the ratio of the teau phase currents. Although was quite prominent at Vhold = tio of 2.6 + 0.3), at Vhold = -20 currents were nearly identical
early peak to late plathe early component -80 mV (early/late ramV, the early and late (1.2 + 0.1).
Discussion These results support the conclusion that a significant portion of the NMDA-evoked current in oocytes is carried by ICI(ca), which is activated secondarily by Ca2+ influx through the NMDA-activated channel. Reduction of Ca2+ influx by reducing extracellular Ca2+ or by replacing it with Ba’+ caused a general reduction of the NMDA response, and particularly the early transient component. Reduction of intracellular Ca2+ by injection of EGTA similarly reduced or eliminated the early component even though Ca2+ influx was presumably normal (since the driving force for Ca2+ was unchanged or possibly slightly increased). Finally, the Cl- mediation of the early component as well as a portion of the late component is indicated by the inhibitory effects of a variety of known Clchannel blockers and the elimination of the early component through abolishing the driving force on Cl- by holding the membrane potential at Ecr. The apparent desensitization observed in oocytes does not appear to depend on a rise in intracellular Ca2+ per se. All of the various Cl- channel blockers used would not be expected to block specifically the Ca2+ flux through the NMDA receptor/channel. In addition, the NMDA-mediated Cal+ influx should continue at EC,, even though the apparent desensitization is eliminated. Similarly, Ba2+ per se does not eliminate desensitization of neuronal NMDA receptors (Mayer and Westbrook, 1987). It must be emphasized that neuronal NMDA receptors are capable of desensitization (Mayer et al., 1989).
Recent
single-channel
recordings
of
mouse
cortical
neurons using isolated outside-out patches (Sather et al., 1989; Sot. Neurosci., abstract) demonstrate a desensitization, with a time constant of 100-400 ms, that persists in low Ca2+, high glycine salines. Perhaps similar experiments at the single-channel level on oocytes would also reveal a fast component of desensitization inherent to the NMDA receptor/channel complex. Although this contaminating Cl- current may explain several anomalies in previous oocyte studies, it also confounds the interpretation of several aspects of NMDA receptor/channel function, including the kinetics of channel gating, the current-voltage relation, and the modulation by second messengers or other biochemical agents. Previous studies of NMDA-evoked currents in oocytes illustrate a variety of waveforms showing pronounced (Kushner et al., 1988; Lerma et al., 1989) or absent (Kleckner and Dingledine, 1988; Fong et al., 1988; Verdoorn and Dingledine, 1988) early transients. Our results suggest that the variability in the NMDA-evoked response is due to the variability in the secondary activation of ICr(caj. In this scenario, any circumstance that produces larger currents will lead to greater changes in intracellular Ca2+ and will therefore result in greater activation of the transient IcI(caI. For example, the apparent desensitization of receptors at higher doses of NMDA (Kushner et al., 1988; Lerma et al., 1989) could instead be due to stimulation of a relatively large, inactivating IcIIc+ The time course of the early transient component of the NMDA response is strikingly similar to that of IcI(cdj produced when intracellular Ca2+ is elevated by a variety of other direct methods. The Icr(caj reaches a peak in l-2 s and decays substantially (but incompletely) over a period of 5-10 s regardless of whether Ca2+ is elevated by activation of endogenous voltage-gated Ca*+ channels (Barish, 1983; Miledi, 1982), by direct injection of Ca2+ from a puffer pipette (Miledi and Parker, 1984), or by application of Ca2+ ionophore (Boton et al., 1989). The time course of bath application of NMDA to the oocyte does not allow us to determine whether a more rapid desensitization occurs on a time scale of
Vol. 2, 53-60,
January,
1990, Copyright
0 1990 hy Cell Prets
Apparent Desensitization of NMDA Responses in Xenopus Oocytes Involves Calcium-Dependent Chloride Current John l? Leonard and Stephen R. Kelso Department of Biological Sciences University of Illinois at Chicago Chicago, Illinois 60607
Summary N-Methyl-o-aspartate (NMDA) receptors were expressed and studied in Xenopus oocytes injected with rat brain RNA. NMDA application elicits a rapid inward current that decays in several seconds to a relatively stable level. This decay is reportedly due to desensitization. However, we found the early transient component could be evoked more than once during a single application of NMDA, suggesting that the receptor did not actually desensitize. Removal of external Ca*+, replacement of Ca2+ with Ba2+, or intracellular injection of ECTA abolished the transient component. Furthermore, a variety of Cl- channel blockers nearly eliminated the transient component and inhibited the plateau current as well. We propose that a significant portion of the NMDA current recorded in oocytes is carried by a transient inward Cl- current triggered by Ca2+ influx through the NMDA receptor/channel. Introduction The N-methyl-o-aspartate (NMDA) subtype of receptor for the neurotransmitter glutamate is receiving much attention because of its apparent role in a wide range of phenomena, including the triggering of longterm synaptic modification, developmental plasticity, susceptibility to hypoxic brain damage, and epilepsy. Several properties distinguish it pharmacologically and physiologically from two other subtypes of glutamate receptors, commonly referred to by their specific excitatory amino acid agonists, kainate and quisqualate. The gating of the NMDA receptor is unique in its dependence upon both ligand binding and membrane potential (via open channel block by Mg*+; Mayer and Westbrook, 1987; Mayer et al., 1984; Nowak et al., 1984). It also exhibits a regulatory site for glycine (Kushner et al., 1988; Verdoorn et al., 1987; Johnson and Ascher, 1987), an inhibitory modulatory site for Zn*+ (Westbrook and Mayer, 1987; Peters et al., 1987), open channel blockade by certain compounds (Mayer et al., 1988; Huettner and Bean, 1988; MacDonald et al., 1987), and a significant permeability to Ca*+ (Burgoyne et al., 1988; Hori et al., 1985; MacDermott et al., 1986; Murphy and Miller, 1988). In addition, the NMDA receptor can be reliably identified by the use of specific competitive antagonists (Watkins and Olverman, 1987). Recently, several investigators have exploited the capability of the Xenopus oocyte to express functional ionic channels after injection of mRNA to study
currents mediated by brain-derived excitatory amino acid receptors. These studies have generally established that all three subtypes of glutamate receptor can be expressed and that responses evoked by NMDA in RNA-injected oocytes exhibit most of the same pharmacological and physiological characteristics as neural preparations. These include blockade by specific antagonists (Verdoorn et al., 1987), a voltage-dependent block by Mg*+ (Kushner et al., 1988; Verdoorn et al., 1987), requirement for glycine (Kleckner and Dingledine, 1988; Watson et al., 1989), and open channel blockade by phencyclidine receptor agonists (Fang et al., 1988; Kushner et al., 1988). Although detailed studies of the ionic selectivity properties of NMDA receptors have not yet been completed, it is generally assumed that the receptor in oocytes is also permeable mainly to Na+, K+, and Ca*+, as is the case in cultured neurons. In the oocyte, one of the distinguishing characteristics of the NMDA response is a rapid initial current that decays in a few seconds to a relatively steady level; a characteristic that has been assumed to reflect a receptor desensitization (Kushner et al., 1988; Lerma et al., 1989). However, the native Xenopus oocyte expresses a Cl- channel that is activated by rises in intracellular Ca2+ concentration (Barish, 1983; Miledi, 1982). If the NMDA receptor in RNA-injected oocytes is permeable to Ca*+, as it is in neurons, then it is quite possible that Ca2+ influx could trigger the Ca*+dependent Cl- current, Ic,(ca), that is normally present. Since Ec, is approximately -20 to -30 mV in the oocyte (Barish, 1983), IcI(caI would be inward at the hyperpolarized potentials typically used in oocyte studies of NMDA responses. The fact that Ecr is so close to the NMDA reversal potential makes it difficult to separate the two currents electrophysiologically. In this study we have examined the NMDA response under a variety of conditions that either reduce Ca2+ influx or inhrbrt IcI(caI. We find that the major portion of the early phase and some of the late phase of the NMDA response in these cells is carried by Ca2+dependent Cl- current. Much of the apparent desensitization of the NMDA receptor observed during bath application to an oocyte is an artifact of the Xenopus expression system brought about by the activation of the endogenous, transient IcI,Ca,. Results In oocytes injected with rat brain RNA and voltageclamped at -80 mV, bath application of IO-100 PM NMDA in standard Mg*+-free saline typically resulted in a large inward current of 50-300 nA (n = 43 cells). All five different RNA preparations used reliably expressed NMDA receptors upon injection into oocytes. No NMDA responses were ever seen in noninjected cells (n = 10). As illustrated in Figure IA, the waveform
A
was observed in 6 of 6 oocytes. A second signature of NMDA channels is a blockade by Mg2+. Addition of 1 mM Mg 2+ to the nominally Mg2+-free standard saline reversibly prevented elicitation of an NMDA response at -80 mV (n = 7). As reported by others, the block by Mg2+ was less complete at more depolarized levels (-40 to -30 mV; data not shown). Having verified that our NMDA response in oocytes is mediated by classic APV- and Mg2+-sensitive NMDA receptor/channel complexes, we proceeded to examine the response in more detail.
NMDA
1 200 nA 10 set
NMDA
‘::: Figure 1. Initial Decay Receptor Desensitization
+ Ma’+ L
NMDA
of NMDA
Response
Is Inconsistent
with
(A) Top: response to 50 s application of 50 PM NMDA. Bottom: in the same oocyte, addition of 1 mM Mg*+ at 25 s after start of 50 PM NMDA application led to a blockade of current and was followed on wash-out by a second transient inward current. Horizontal bars indicate duration of NMDA application. In this and subsequent figures in which Vhold = -80 mV, a 10 mV step to -70 mV prior to NMDA application was used to monitor changes in input conductance. (B) Voltage-clamp protocol (top) and current response (bottom). NMDA (50 PM) was applied during a depolarization to -30 mV and during the return to -80 mV. The large transient occurs at a time when receptors have already been continuously exposed to NMDA for 25 s.
of the NMDA-induced current is distinct from that evoked by other excitatory amino acids; it has a rapid onset, reaching a peak in l-2 s, and then decays within 5-10 s to a relatively stable level that lasts for as long as the NMDA is present. In these experiments the longest period of application was 75 s. The responses were very repeatable upon successive applications and could in fact be repeated within 30 s, which was approximately the shortest time in which the NMDA wash-out resulted in a return to the initial level of holding current. To verify that our NMDA response was mediated by specific glutamate receptors of the NMDA subtype, we coapplied NMDA and the specific antagonist DL-2amino 5-phosphonovaleric acid (APV). At IO PM, APV produced a potent, reversible inhibition of the NMDA response. At 100 PM, a complete and reversible block
The Early Transient Component Is Not Due to “Desensitization” The transient nature of the early part of the NMDA response in Xenopus oocytes has been ascribed to a desensitization of the receptor (Kushner et al., 1988; Lerma et al., 1989). However, under certain conditions, we could induce a fast, transient inward current late in the period of NMDA exposure, at a time when the receptors should presumably have been “desensitized.” Due to the unique voltage-dependent open channel block of the NMDA receptor/channel by Mg2+, it is possible to block current flow through the channel independently of agonist binding at the NMDA binding site. Thus, as illustrated in Figure IA, current flow through the channel could be blocked in the middle of continuous application of NMDA, after the early transient phase and during the stable portion of the response. Under these conditions, wash-out of the Mg2+ led to a resumption of inward current flow. However, the inward current did not return to the stable level that was obtained before Mg2+ blockade, but instead exhibited a rapid transient phase similar to the response at the very beginning of NMDA application (n = 3 oocytes). This result suggests that the NMDA receptors were not really desensitized during the late plateau phase. A transient inward current late in the period of NMDA application could also be evoked by another method that greatly reduces the NMDA response. Thus, a similar result occurred when the current was decreased in amplitude not by Mg2+ blockade, but rather by a change in holding potential that decreased the driving force (Figure 1B). In this case, NMDA was applied during a command pulse to -30 mV and was still present on return to -80 mV holding level. Although the NMDA receptors had ample opportunity to desensitize in the presence of NMDA at -30 mV, a large, early transient peak was immediately elicited on return to -80 mV (n = 4 oocytes). The Early Component Is Ca*+-Dependent The Xenopus oocyte normally expresses a Cl- conductance that is triggered by a rise in intracellular Ca2+ and is inward at -80 mV. When this current, Icl(ca), is evoked at +I0 mV by Ca*+ entry through endogenous voltage-dependent Ca2+ channels, it peaks in l-2 s and decays over a period of 5-10 s (Barish, 1983; Miledi, 1982). Other ways of increasing the intra-
Cl Component 5.5
ol NMDA
I” Xenopus
Response\
Oocyte\
cellular Ca2+ concentration, such as direct injection of Ca2+ from a puffer pipette (Miledi and Parker, 1984) or use of Ca2+ ionophore (Boton et al., 1989) or any neurotransmitter that produces IP3 (Oosawa and Yamagishi, 1989; Sugiyama et al., 1989), will also reliably induce lcI,caI. Because the NMDA receptor has been shown in a variety of preparations to be permeable to Ca2+, in addition to Na+ and K+, we examined the possibility that part of the NMDA response in oo-
were readily reversible, we did not attempt to examine NMDA responses under these conditions. The decrease observed during the late phase in Ca2+-deficient saline again suggests that some IcI(caI flows throughout the NMDA application. In the final series of experiments examining the role of Ca’+ in the early inward component, we injected EGTA into oocytes to chelate any Ca2+ that might enter through NMDA-activated channels. The success of the injection was assessed by monitoring IcI(ca) elicited by activating voltage-dependent Ca2+ channels with a step to +I0 mV (Figure 3A). Pressure injection of 0.5 M ECXA (up to 5 nl) usually reduced this response by as much as 90% within 4 min of injection (n = 8). As noted by others (Miledi and Parker, 1984), the time course and completeness of the blockade depended upon the amount of ECXA injected. Figure 3A shows outward I,-I(~~) currents at voltages more positive than Ec,, and Figure 3B shows NMDA-activated currents in the same cell at -80 mV, at which any Cl- current would be inward. In 6 cells injected with ECXA, the NMDA respor,se was measured after lc~(~~)
cytes is carried secondarily by ICI(ca). Thus, the early transient component might be mainly IcI(ca) that is triggered as soon as Ca2+ enters through NMDA channels and relaxes as a result of known Cl- channel inactivation (Boton et al., 1989). In an attempt to block or reduce the early inward component of the response, several methods were used to reduce Ca2+ entry during NMDA application. In one series of experiments, Ca2+ was replaced by Ba2+. Ba2+ normally permeates the NMDA channel, although it is also a partial blocker (Mayer and Westbrook, 1987). Moreover, Ba2+ is a poor activator of I CNCa) in oocytes (Barish, 1983). In 15 of 17 oocytes examined, the early peak was completely eliminated in Ba2+-substituted saline (Figure 2A). In experiments in which Ca2+ was lowered or omitted entirely from the perfusion medium, the early transient current was greatly reduced or absent (Figure 2B). When extracellular Ca2+ was lowered to 0.1 mM, both early and late components were reduced (66% + 5%, [mean f SEMI and 48% + IO%, respectively, n = 4). Similar results were obtained in nominally Ca2+-free saline, as the early component was decreased by 69% + 3% and the late component by 38% + 7% (n = 8). We attempted to reduce Ca2+ further by including 0.5 mM EGTA in the divalent-free perfusion medium. However, in media containing ECXA and no divalent cations, the oocytes often developed a large leakage current within l-2 min of switching to the divalent-free solution. Although the effects
A
Ca++-saline
Ba+‘-saline
NMDA
NMDA
was blocked. In these cases the early component of the NMDA response was also absent or was greatly reduced (56% + 6% decrease; see Figure 38). This result indicates that the early component depends on internal Ca2+. The ECXA injection experiments also suggest that the effect of changing external Ca2+ levels either by reduction (Figure2B) or by replacement with Ba2+ (Figure 2A) is actually via the ultimate effect on internal Ca2+. The NMDA Response Is Partly Cl--Mediated If Ca2+ entering the NMDA-gated channel actually produces a major current by activation of endogenous ICI(ca) channels, then one should be able to inhibit NMDA responses directly with compounds shown to inhibit ICI(ca) in oocytes (Boton et al., 1989; Snutch et al., 1986). We found this to be true for a variety of
Figure 2. Elimination ponent of NMDA ficient Salines
Ca++-saline NMDA
(A) Inward currents recorded at -80 mV during application of 50 uM NMDA. Middle trace shows a reversible block of the transient current and a partial block of the late phase in saline containing 20 mM Ba*+ in place of Ca’+. (B) Reduction of extracellular Ca*+ from 2.0 mM too.1 mM caused a reversible block of the transient component of the response to 50 uM NMDA at -80 mV.
I+-
B
2 mM
0.1 mM
Ca++
NMDA
NMDA
Ca++
of Early Transient ComResponse in Ca2+-De-
2 mM
Ca++
NMDA
=-y200 IIA 1 25 sac
Nt?UKNl 56
A
1’ post
Figure 3. Inhibition ot l(l,(.,, and Responses by Intracellular ECXA
EGTA
r
+10
+10
I
I
-80
-100
2’ post
6
-80
-100
NMDA
(A) Traces show the transient outward current that occurs during the indicated depolarization to +I0 mV. Responses were taken before (left) and after (right) pressure injection of -1 nl of 0.5 M ECTA via an intracellular micropipette. (B) In the oocyte used in (A). Inward currents were elicited at -80 mV by bath application of 50 PM NMDA before (left) and after (right) EGTA injection. Note the loss of the transient component of the NMDA response.
EGTA NMDA
NMDA
L
200 nA
10 set
early inward current (52% k 5%) and some decrease in the later phase (33% + 5%, n = 3). Again, similar results were obtained in 5 experiments using 500 PM 9-anthracene carboxylate. The peak current and latephase current were decreased 62% 5 3% and 25% k 5%, respectively. Under certain conditions, the contribution of the additional Cl- conductance to the NMDA response was reflected in the current-voltage relation. In most experiments, the reversal potential of the NMDA response was measured by stepping the command voltage to different membrane potentials, holding for approximately 25 s to allow voltage-activated currents to reach steady state, and then applying NMDA to measure the current induced during the steady phase 20-30 s after the beginning of the response. The rever-
Cl- channel blockers, including 9-anthracene carboxylate, 4+dinitro-stilbene 2,2-disulfonic acid (DNDS), niflumic acid, and flufenamic acid. Flufenamic acid and niflumic acid were the more effective blocking agents (K, = 30 and 19 PM, respectively; M. M. White unpublished data). At 500 FM, flufenamic acid effectively blocked ICI(caj evoked by Ca’+ entry via voltagedependent Ca2+ channels (Figure 4A). As expected, flufenamic acid also reversibly inhibited the early peak of the NMDA response (Figure 4B; 79% + 2% decrease in 6oocytes). There was also a smaller decrease in the later phase of the NMDA response (56% + 3% decrease). Similar results were obtained using somewhat higher doses of other Cl- channel blockers. The addition of 1 mM DNDS to the normal perfusion medium caused a large reversible reduction in the
Control
Flu
Figure3. Responses
Wash
L
600 nA
-\
1%x
I‘\
\ -A.
-.-A +10 -100
+10 -60
-100
NMDA
-60
NMDA
7 I--
lnhibltlorr by a Cl
or I,,,(,,, Channel
and NMDA Blocker
iA) Trace\ show the transient outward current during the indicated step depolarization. Responses were taken before (Control), during (Flu), and after (Wash) bath perfusion of 500 PM flufenamic acid. (B) In the oocyte used in (A), responses to 50 PM NMDA were reversibly inhibited bv the flufenamic acid.
Cl- Component
of NMDA
Response\
rn Xrnopus
Oorytes
57
The contaminating Cl- conductance was more pronounced when the current-voltage relation was assessed using ramp voltage commands. The reversal potential was shifted even closer to EC,, and additional nonlinearities were observed between -15 mV and +I0 mV (Figure 5A, continuous curve). In 5 cells we used ramp commands to measure the NMDA current-voltage relation in the presence or absence of 300 PM niflumic acid, which blocks the Cl--mediated component of the response. In all cases, the conductance was decreased (from 2.2 + 0.4 uS to 0.9 i 0.3 PS) and the nonlinearities between -15 mV and +I0 mV were reduced. Figure 5B also illustrates a small shift in the reversal potential away from EcI for the current-voltage relation measured in the presence of niflumic acid. In spite of the fact that ramp measurements indicated that the reversal potential for the NMDA response is near EcI, we consistently recorded substantial inward currents at -20 mV during long depolarizing steps (39 + 6 nA, n = 9). It is possible that a time-dependent inactivation of lc:~(,-~, (Boton et al., 1989) during long depolarizing commands results in a more pure NMDA current. This would explain the discrepancy between the apparent reversal potential of lNMDA measured by ramps (approximately -20 mV) and the observation of inward NMDA currents at -20 mV during long depolarizing commands.
i
-100 nA
Figure 5. Dependence of NMDA on the State of IcIIcdj Activation
Current-Voltage
Relation
Up-
(A) Current-voltage curves were obtained by a slow ramp command (continuous curve) or by applying 50 FM NMDA at least 25 s after initiating a 2 rmrn long voltage step to different potentials. In the long pulse method, currents were measured at the early peak (squares) or after the response had obtained a steady inward current (25 s, circles). The insert shows waveforms of currents recorded at the indicated holding potentials. (B) Current-voltage curves obtained by the ramp method in normal saline and in saline containing the Cl- channel blocker niflumic acid (Nif) at 300 PM.
sal potential measured in this manner (Figure 5A, circles) was -3 + 1 mV (n = 7), which is similar to that seen in neurons (Mayer and Westbrook, 1987). However, when the NMDA response was measured at the early peak (Figure 5A, squares), the greater conductance during the early phase was associated with a more hyperpolarized reversal potential. This is to be expected if a large part of the early component is mediated by Cl-, which reverses at Ecr g -20 mV. The inset in Figure 5A shows the expected change in waveform when the early transient Cl- component was eliminated by holding the membrane potential at -20 mV (~Ecr). Elimination of the early transient mV was quantified in 10 component at E, = -20 cells by taking the ratio of the teau phase currents. Although was quite prominent at Vhold = tio of 2.6 + 0.3), at Vhold = -20 currents were nearly identical
early peak to late plathe early component -80 mV (early/late ramV, the early and late (1.2 + 0.1).
Discussion These results support the conclusion that a significant portion of the NMDA-evoked current in oocytes is carried by ICI(ca), which is activated secondarily by Ca2+ influx through the NMDA-activated channel. Reduction of Ca2+ influx by reducing extracellular Ca2+ or by replacing it with Ba’+ caused a general reduction of the NMDA response, and particularly the early transient component. Reduction of intracellular Ca2+ by injection of EGTA similarly reduced or eliminated the early component even though Ca2+ influx was presumably normal (since the driving force for Ca2+ was unchanged or possibly slightly increased). Finally, the Cl- mediation of the early component as well as a portion of the late component is indicated by the inhibitory effects of a variety of known Clchannel blockers and the elimination of the early component through abolishing the driving force on Cl- by holding the membrane potential at Ecr. The apparent desensitization observed in oocytes does not appear to depend on a rise in intracellular Ca2+ per se. All of the various Cl- channel blockers used would not be expected to block specifically the Ca2+ flux through the NMDA receptor/channel. In addition, the NMDA-mediated Cal+ influx should continue at EC,, even though the apparent desensitization is eliminated. Similarly, Ba2+ per se does not eliminate desensitization of neuronal NMDA receptors (Mayer and Westbrook, 1987). It must be emphasized that neuronal NMDA receptors are capable of desensitization (Mayer et al., 1989).
Recent
single-channel
recordings
of
mouse
cortical
neurons using isolated outside-out patches (Sather et al., 1989; Sot. Neurosci., abstract) demonstrate a desensitization, with a time constant of 100-400 ms, that persists in low Ca2+, high glycine salines. Perhaps similar experiments at the single-channel level on oocytes would also reveal a fast component of desensitization inherent to the NMDA receptor/channel complex. Although this contaminating Cl- current may explain several anomalies in previous oocyte studies, it also confounds the interpretation of several aspects of NMDA receptor/channel function, including the kinetics of channel gating, the current-voltage relation, and the modulation by second messengers or other biochemical agents. Previous studies of NMDA-evoked currents in oocytes illustrate a variety of waveforms showing pronounced (Kushner et al., 1988; Lerma et al., 1989) or absent (Kleckner and Dingledine, 1988; Fong et al., 1988; Verdoorn and Dingledine, 1988) early transients. Our results suggest that the variability in the NMDA-evoked response is due to the variability in the secondary activation of ICr(caj. In this scenario, any circumstance that produces larger currents will lead to greater changes in intracellular Ca2+ and will therefore result in greater activation of the transient IcI(caI. For example, the apparent desensitization of receptors at higher doses of NMDA (Kushner et al., 1988; Lerma et al., 1989) could instead be due to stimulation of a relatively large, inactivating IcIIc+ The time course of the early transient component of the NMDA response is strikingly similar to that of IcI(cdj produced when intracellular Ca2+ is elevated by a variety of other direct methods. The Icr(caj reaches a peak in l-2 s and decays substantially (but incompletely) over a period of 5-10 s regardless of whether Ca2+ is elevated by activation of endogenous voltage-gated Ca*+ channels (Barish, 1983; Miledi, 1982), by direct injection of Ca2+ from a puffer pipette (Miledi and Parker, 1984), or by application of Ca2+ ionophore (Boton et al., 1989). The time course of bath application of NMDA to the oocyte does not allow us to determine whether a more rapid desensitization occurs on a time scale of
