- Email: [email protected]
Expression of glutamate receptor subunits in α-motoneurons
Molecular Brain Research 52 Ž1997. 38–45
Research report
Expression of glutamate receptor subunits in a-motoneurons Roselynn Temkin a , Deborah Lowe a , Penny Jensen b, Hanns Hatt c , Dean O. Smith a
d,)
Neuroscience Training Program, UniÕersity of Wisconsin, 1300 UniÕersity AÕenue, Madison, WI 53706, USA b Department of Physiology, UniÕersity of Wisconsin, 1300 UniÕersity AÕenue, Madison, WI 53706, USA c Ruhr-UniÕersitat 150, 44780 Bochum, Germany ¨ Bochum, Lehrstuhl fur ¨ Zellphysiologie, UniÕersitatstr. ¨ d Pacific Biomedical Research Center, UniÕersity of Hawaii, Honolulu, HI 96816, USA Accepted 15 July 1997
Abstract Whole-cell recordings from 6.5 day embryonic chick a-motoneurons indicated the presence of AMPA, kainate, and NMDA glutamate receptor subtypes in each motoneuron tested. AMPA consistently evoked a desensitizing response, while kainate could evoke either a desensitizing or non-desensitizing whole-cell response. In excised membrane patches, desensitizing AMPA responses appeared to be colocalized with non-desensitizing kainate responses. Desensitizing kainate responses were seen in some patches which were not responsive to AMPA, suggesting that kainate selective subunits and AMPA selective subunits localize separately on the motoneuron membrane. To determine which of the known glutamate receptor subunits might underlie these responses, we used RT-PCR amplification to detect subunits present in mRNA isolated from adult rat spinal cord and from a highly enriched motoneuron population from embryonic chick. Sequencing of the the amplified cDNA was used to verify the identity of the products and of the alternative splice variants of GluR1–4. In rat spinal cord, all subunits that we attempted to detect, including AMPA selective subunits GluR1–4, kainate selective subunits GluR5–7 and KA1–2, and NMDA subunit NR1 were present. The isolated motoneurons also contained AMPA subunits GluR1, 2, and 4, and kainate subunits GluR6 and 7. The GluR2 and 4 subunits were specifically processed by splicing, present primarily as the flip splice form. q 1997 Elsevier Science B.V. Keywords: Glutamate; AMPA; Kainate; N-Methyl-D-aspartate ŽNMDA.; Motoneuron; Spinal cord; Alternative splicing; RNA
1. Introduction Glutamate is a major excitatory neurotransmitter. In the spinal cord, for example, it depolarizes motoneurons in embryonic w25x and adult w9,4x animals. Furthermore, receptors sensitive to the three glutamate agonists AMPA, kainate, and N-methyl-D-aspartate ŽNMDA., have been identified in spinal cord cells w1,2,31x and, more specifically, in a-motoneurons dissociated from both embryonic chick and neonatal rat w32,35,37x. Within the past six years, the genetic basis of glutamate-mediated synaptic transmission has been resolved at the molecular level and has been found to be remarkably diverse. cDNA encoding at least 14 different glutamate receptor subtypes has been cloned from rodent brain and sequenced, and many different spliced or edited variants
)
Corresponding author. University of Hawaii, 2444 Dole Street, Bachman 204, Honolulu, HI 96822, USA. Fax: q1 Ž808. 956-8061. 0169-328Xr97r$17.00 q 1997 Elsevier Science B.V. All rights reserved. PII S 0 1 6 9 - 3 2 8 X Ž 9 7 . 0 0 2 4 9 - 0
have been identified w14x. Functional expression revealed four different cDNA molecules ŽGluR1–GluR4. that encode receptors activated by both AMPA and kainate w7,15,16,24x. The GluR1–4 subunits exist in alternately spliced variants that affect the desensitization of the receptors. These variants have been designated flip and flop w33x. In addition, three subunits ŽGluR5–7. that bind kainate but not AMPA have been cloned w5,6,12x. Agonist sensitivity and ionic permeability of GluR5–7 can also vary depending on the presence of associated kainate-binding proteins ŽKA1 and KA2. w13x. Lastly, five cDNA molecules ŽNR1 and NR2A-D. that encode NMDA-activated receptors have also been cloned w21,22x. With such diversity of genetic message, how is synaptic specificity achieved? In situ hybridization of AMPA-selective receptors ŽGluR1–4. w16,29x and of kainate-selective receptors ŽGluR5–7, KA1, and KA2. w3x revealed that the various subunits display different regional distribution. Each subunit was found to be localized to certain brain regions, and this localization was distinctly different for all
R. Temkin et al.r Molecular Brain Research 52 (1997) 38–45
subunits. Based on this result, a population of similar neurons might be hypothesized to express some subset of the glutamate receptor subunits that would account for cell type-specific responses. In this study, we tested this hypothesis by assaying expression of the specific glutamate receptor subtypes in a population of early embryonic Žstage 28. chick amotoneurons. At this developmental stage, differentiation into specific motor units has presumably not yet occurred, for individual motoneurons are just at the point of establishing initial synaptic connections w10x. Furthermore, at this stage, there is little variability in the currents activated by glutamate w32x. We show that mRNA encoding a subset of AMPA and kainate subtypes was present in this population of neurons. Specific pre-mRNA splicing had occurred, with the flip forms appearing to predominate. Genomic and cDNA sequence analysis of chick glutamate receptors also demonstrated avian editing of GluR2 but not GluR6 Žw19x; cf. w17x..
2. Materials and methods 2.1. Experimental preparation In an initial series of control experiments, glutamate receptor expression was tested in rat spinal cord. Adult rats were anesthetized by intraperitoneal injection of chloral hydrate Ž2.8 mmolrkg, i.p... Parallel incisions were made above spinal vertebra T12 and below L 2 , and the vertebral column and spinal cord were cut at these incisions. This segment of spinal cord was removed by forcing cold saline solution through the vertebral column with a syringe. Spinal roots were trimmed from the spinal cord segment. The cord was then frozen immediately in liquid nitrogen, where it was stored until mRNA extraction. An enriched fraction of chick motoneurons was prepared using techniques described in detail in Rosenheimer and Smith w28x. Lumbar spinal cords from 6.5-day Žstage 28. embryonic chicks were dissected, freed of their meninges and dorsal root ganglia, and collected in ice-cold buffer. Following incubation at 378 with 0.05% trypsin and 0.005% DNase I, cells were dissociated by gentle trituration. Motoneuron-enriched cell fractions were then generated on the basis of their buoyant density in a metrizamide solution w30x. The effectiveness of this procedure for enrichment of motoneurons was verified by labeling the motoneurons via cut ventral roots with the fluorescent dye rhodamine prior to dissociation. Approximately 92% of the recovered cells contained the fluorescent label and were identified specifically as motoneurons w28x. Their diameters were measured using a Coulter counter. The diameters were distributed in a unimodal ‘bell-shaped’ fashion, and the average value Ž"S.D.. was 10 " 2 m m Ž n s 4094.. The remaining 8% of the cells were not identified. Because of their size, we
39
suspect that they were motoneurons that had not been labeled sufficiently for fluorescence detection. After isolation, the cells were used either for electrophysiological recording or for mRNA extraction. Cells for recording were plated and incubated in Dulbecco’s modified Eagle medium plus 10% horse serum at 378 with 5% CO 2 . Prior to recording, the DME was replaced by HEPES buffered bath solution containing Žin mM.: NaCl 140, KCl 5.3, CaCl 2 2, Na 2 HPO4 0.67, KH 2 PO4 0.22, HEPES 15, and glucose 5.6. The pH was adjusted to 7.4 using either 10 N HCl or NaOH. Recordings were obtained from cells maintained between 1 and 2 days in culture. Similar results were obtained regardless of the time in culture. 2.2. Electrophysiological recording Membrane currents were recorded at room temperature Ž22 to 248C. in whole-cell or outside-out configuration. Throughout the recording experiments, the cells were bathed with continuously flowing bath solution. Glutamate was dissolved in this extracellular saline. The electrode solution contained Žin mM.: KCl 140, CaCl 2 1, MgCl 2 2, EGTA 11, glucose 10, and HEPES 10 ŽpH s 7.4.. For whole-cell recording, cAMP Ž100 m M. was included in the pipette except where noted. The currents were amplified ŽList EPC7 and Dagan 8900. and stored on video tape for subsequent evaluation. Resting membrane potentials were generally between y45 and y60 mV. Records were filtered at 1 to 4 kHz using an 8-pole Bessel filter ŽFrequency Devices, Haverhill, MA. and digitized at 10 to 50 kHz. They were analyzed using a PDP 11r73-based acquisition system ŽIndec Systems, CA.. Curve-fitting routines used the Marquardt-Levenberg techniques provided in MicroCal Origin software ŽMicroCal, Northampton, MA.. 2.3. Agonist application To achieve rapid agonist application, a ‘liquid-filament’ system was used to apply ligand-containing solution to the electrode tip w11x. Ligands were kept in separate pressurized Ž1 atu. chambers connected in parallel by a manual valve to the delivery system. The ligand-containing solution was ejected under pressure from the storage chamber into a polyethylene tube drawn to a 30-m m tip. This tube was fixed in the bath parallel to the direction of flow of the background bathing solution. The ligand solution exited the polyethylene tube as a thin Ž30 m m. stream flowing at the same rate as the surrounding background solution Ž6 mlrmin.. After traveling a distance of 5 mm, the agonist and the surrounding background solutions were removed from the chamber using a pulse-free pump. This ‘peristaltic-type’ pump Žmanufactured in the Physiology Institute, Technical University of Munich. utilized an eccentric bearing to bring 16 teeth onto the uptake tubing; one tooth contacted the tubing at all times, thus minimizing dead
40
R. Temkin et al.r Molecular Brain Research 52 (1997) 38–45
space in the tube, diffusion of agonist into the background solution, and deviations from laminar flow. The ligand-delivery tube was connected to a piezo crystal. On application of 100 V to the piezo, the tube shifted upward 20 m m within about 200 m s. With this system, ligand was introduced to the electrode tip at a constantly maintained concentration within - 1 ms under optimal conditions w11x; receptor activation times were somewhat slower, for they depended on patch characteristics and recording configuration Žactivation times were slower in whole-cell recordings.. On cessation of the voltage pulse, the ligand-containing solution was lowered rapidly, and the electrode tip was again exposed to normal background solution. Spurious results could arise from any inconsistencies in agonist delivery or recording conditions. For example, excessive diffusion from the agonist stream into the background solution could cause receptor desensitization, which might then be interpreted as failure to respond to subsequent agonist application. Therefore, to monitor these conditions, control applications of glutamate Žusually 1 mM. were delivered between sequential applications of the other specific agonists. Stable ligand-delivery and recording conditions could sometimes be obtained for times exceeding 15 min. Regression analyses indicated that neither the rise time Žfrom 20% to 80% of peak amplitude. nor the peak amplitude varied significantly Ž P ) 0.05. in these optimal recordings. Results from these stable experiments were then used for further study.
Table 1 PCR primers X
GluR-1
5 primer – TGGTGGTTCTTCACCCTGATCAT X 3 primer – TATGGCTTCATTGATGGATTGC
GluR-2
5 X 3 X 5 X 3
GluR-3
5 primer – GAGTCGACCATGAACGAGTA X 3 primer – AATTCTGGGTATTGGTGCTGG
GluR-4
5 primer – TGGTGGTTCTTCACCCTGATCAT X 3 primer – ACTCCCAGTGATGGATAACCTG
GluR-5
5 primer – TGGTGGTTCTTCACCCTGATCAT X 3 primer – CACTGCTCAACGTCATTGTTCT
GluR-6
5 X 3 X 5 X 3
GluR-7
5 primer – TGGTGGAAGACGGCAAGTAC X 3 primer – CCTTGTTGCATCAGGGAGCC
KA-1
5 primer – GGCCAGTGGCATGTGGCAGA X 3 primer – AAGACACAGCTGACAGCCAG
KA-2
5 primer – GGGGTGTGGTACTCTAACCG X 3 primer – AAGACACAGCTGACAGCCAG
NRa-1
5 primer – CTGTGGTTGCTAGTAGGACTGT X 3 primer – GCATTCCTGATACCGAACCCAT
X
primer primer primer primer
– – – –
TGGTGGTTCTTCACCCTGATCAT TGCAAAATTCTGGGAATTCTGC AGGAAATGACACGTCTGGGC GTGTAGGAGGAGATTATGAT
X
X
X
X
primer primer primer primer
– – – –
TGGTGGTTTTTCACACTTATCAT CGCTGGCACTTCAGGGACATTC GGACAATGGAATGGAATGGTTC GTATACGAAGAAATGATGAT
X
X
X
X
X
X
Oligonucleotide sequences are presented in the 5 to 3 direction.
2.4. mRNA isolation and GluR amplification Poly-A mRNA was isolated using a commercial kit ŽMicro-FastTrack, Invitrogen, San Diego, CA.. Chick amotoneurons were prepared for mRNA isolation by first being pelleted by centrifugation and then resuspended in 2% SDS, 1.5 mM MgCl 2 , 200 mM Tris-HCl pH 7.5, 200 mM NaCl lysis buffer containing RNase inhibitor ŽInvitrogen.. Frozen rat spinal cords were homogenized in identical lysis buffer. After oligo ŽdT. cellulose spin-column selection of poly-A mRNA and ethanol precipitation, the mRNA was resuspended in sterile distilled water and was ready for reverse transcription. The mRNA was reversed transcribed in a 20 m l reaction with 50 units of M-MLV reverse transcriptase and 1.5 m M of the 3X primer for a specific glutamate receptor subunit. The cDNA was then amplified by PCR in a 100 m l reaction with 2.5 units of AmpliTaq DNA polymerase. Primers specific for only a single glutamate receptor subunit were used for each amplification. Their nucleotide sequences are presented in Table 1. As a positive control for the appropriate size of product, glutamate receptor cDNA clones were amplified with the same primers. These clones were kindly provided by Stephen Heinemann ŽSalk Institute, La Jolla, CA, USA.. In rat, alternative splice forms can be distinguished by
restriction fragment length polymorphism. For GluR1, only the flip form of the amplified product can be cut with AluI, while only the flop form can be cut with HphI. For the GluR2 and GluR4 products, the flip forms can be cut with AvaI, and the flop forms with HpaI. For GluR3, the flip form can be cut with MspI, and flop with HpaI. Therefore, we initially tested for the alternative splice variants of the chick PCR products by restriction digest analysis. The identity and splice form of chick PCR products from spinal cord were tested by sequencing. Sequencing was performed by the University of Wisconsin Biotechnology Center. Sequencing of the TM2 region of GluR2 and the TM1 and TM2 regions of GluR6 was also used to determine the proportion of edited variants present in this cell population. 2.5. Chemicals Most biochemicals were obtained from Sigma ŽSt. Louis, MO.. DME was purchased from Gibco BRL ŽGaithersburg, MD.. Restriction enzymes were from Promega ŽMadison, WI., except for HphI which was from New England Biolabs ŽBeverly, MA.. All enzymes and reagents for RT-PCR were obtained from Perkin Elmer ŽNorwalk, CT.. Other sources are noted in context.
R. Temkin et al.r Molecular Brain Research 52 (1997) 38–45
3. Results
Table 2 Desensitization time constants
3.1. Responses to glutamate and its agonists
ŽA. Whole-cell kainate response With cAMP tdes s 58.1"10.3 ms, ns 28 Without cAMP tdes s66.4"12.6 ms, ns 22 ŽB. Outside-out patch AMPA tdes s 4.97"2.17 ms, ns129 Kainate tdes s 59.2"8.2 ms, ns110
In whole-cell configuration, currents were activated by the three glutamate agonists AMPA, kainate, and NMDA. Receptors for all three agonists were present on each motoneuron tested by sequential application of the different agonists at concentrations ranging from 10 m M to 10 mM ŽFig. 1.. In most cases, AMPA-activated currents decayed to a sustained steady-state level that was 23 " 8% of peak Ž n s 20.. In rare cases, the steady state current was negligible Ž5 " 1% of peak, n s 3.. Cyclothiazide Ž100 m M. caused a 62% Ž"12% S.D., n s 4. increase in steady-state amplitude of these AMPA-induced currents. Kainate Ž1 mM. evoked two disparate responses, either non-desensitizing ŽFig. 1C. or with a small desensitizing component ŽFig. 1D.. This is probably a function of intracellular cAMP levels. With 100 m M cAMP in the electrode solution, 71% of motoneurons exhibited the desensitizing response Ž n s 28., whereas without cAMP in the electrode, only 15% of responses were desensitizing following activation by kainate Ž n s 39.. The desensitization time constants were determined by fitting a single exponential curve to the decay phase of the current in the presence and absence of cAMP. They were not significantly different ŽTable 2A.. GluR6 expressed homomerically desensitizes in response to kainate w12x, while the AMPA preferring GluR1–4 do not. Cyclothiazide Ž100 m M. caused a 7% Ž"6% S.D., n s 3. increase in the non-desensitizing kainate-evoked responses, which is not significant statistically, but a significant Ž P - 0.05. 22% Ž"12% S.D., n s 5. potentiation of the desensitizing currents. Therefore, the desensitizing component seen in re-
Fig. 1. Motoneuron whole-cell recordings. In each case, the holding potential was y40 mV. The concentration of each agonist was 1 mM. The duration of agonist application is indicated by the bar above the records. ŽA,B. Glutamate ŽA. and AMPA ŽB. responses both exhibit rapidly decaying current. ŽC,D. Whole cell current elicited by kainate differs between cells. Each cell displays one of two alternate responses; ŽC. some cells exhibited a non-desensitizing response to kainate. ŽD. Other cells exhibited a slightly desensitizing response to kainate. ŽE. A single exponential is fit to the desensitizing component of whole-cell kainate response.
41
Means"S.E.
sponse to kainate in some whole-cell recordings may be a result of the activation of kainate selective subunits, while the steady state current of both the desensitizing and non-desensitizing whole-cell responses may be mediated by kainate activation of the AMPA selective subunits. Since subunit composition of glutamate receptors affects channel kinetics as well as agonist sensitivity, desensitization kinetics were measured in outside-out patches. In those patches which responded to AMPA, the receptors desensitized rapidly in response to AMPA but not in response to kainate ŽFig. 2A; Table 2., characteristic of the AMPA selective subunits GluR1–4. Other patches displayed rapidly decaying responses to kainate but did not respond to AMPA ŽFig. 2B; Table 2., characteristic of the kainate selective subunits. Thus it appears that each class of glutamate receptor subunit may localize separately on the motoneuron surface. 3.2. Glutamate receptor mRNA To determine which glutamate receptor subunits may contribute to these motoneuron responses, we used PCR to
Fig. 2. Responses to AMPA and kainate in outside-out patches from chick motoneurons. Holding potential was y60 mV. The agonists Ž1 mM. were applied sequentially to the same patches. The duration of agonist application is indicated by the bar above the records. ŽA. AMPA was applied initially, and then, after removal of the AMPA, kainate was applied to the same patch. ŽB. Kainate was applied initially, and then, after removal of the kainate, AMPA was applied to the same patch. The patch did not respond to the application of AMPA. To confirm that the lack of response was not due to desensitization, kainate was reapplied to the same patch, and it evoked a response.
42
R. Temkin et al.r Molecular Brain Research 52 (1997) 38–45
detect the presence of the various subunits. Chick glutamate receptor sequences are available only for GluRs 1–4 w26x, so some of our primers are based on rat mRNA sequence. Therefore, we tested the efficiency of our primers on rat spinal cord mRNA. In the rat spinal cord, every glutamate receptor subunit that we attempted to amplify was detected ŽFig. 3.. The detected subunits include GluR1–4, GluR5–7, KA1–2 and NR1. The products from spinal cord were verified by sequencing. We then used the same primers to determine which subunits were present in a homogeneous population of chick motoneurons. In the isolated chick motoneurons, primers for AMPA selective subunits GluR1, 2, and 4 amplified products of the correct size ŽFig. 4., as did kainate selective subunits GluR6 and 7 ŽFig. 4B.. All chick motoneuron products were confirmed by sequencing, and they showed at least 80% similarity to the homologous region in rat. We were unable to detect AMPA selective subunit GluR3 using primers based on rat or chick w26x sequences. Nor could we detect kainate selective subunits GluR5, KA1, and KA2, or NMDA subunit NR1. Fig. 4. Glutamate receptor expression in chick a-motoneurons. In each gel, lane 1 is HincII-digested f X174 DNA size marker. Lane 2 Žpc. is PCR product amplified from the cloned subunit cDNA. Lane 3 Žmn. is PCR product from the chick motoneurons. Lane 4 Žnt. is the no template control ŽNT.. AMPA selective subunits GluR1, 2, 3, and 4 were detected. Subunits GluR6 and 7 were the only kainate selective subunits detected in motoneurons.
3.3. AlternatiÕe splicing
Fig. 3. Glutamate receptor expression in rat spinal cord. In each gel, lane 1 is a HincII-digested f X174 DNA size marker ŽUnited States Biochemical, Cleveland, OH, USA.. Lane 2 Žpc. is PCR product amplified from the cloned subunit cDNA. Lane 3 Žsc. is PCR product from rat spinal cord. Lane 4 Žnt. is the no template control. AMPA selective subunits GluR1–4 were detected. Kainate selective subunits GluR5–7 and KA1–2 were also present. The primary NMDA receptor subunit, NR1, was present.
The AMPA selective subunits, GluR1–4, can exist in two alternatively spliced forms, designated flip and flop w33x. These splice forms apparently regulate the rate and amount of desensitization of glutamate activated responses. In rat, the splice forms can be distinguished by restriction fragment length polymorphism. Moreover, there is 99% homology between rat and chick sequence in the alternatively spliced exon region w26x. Therefore, we initially tested for the alternative splice variants of the chick PCR products by restriction digest analysis ŽFig. 5.. By this analysis, the chick motoneurons appeared to contain both the flip and flop splice variants of GluR1, the flip splice form of GluR2, and the flop splice form of GluR4. Subsequent sequencing verified the splice variants of GluR1 and GluR2. Sequencing of the GluR4 product showed higher homology to rat flip than to flop, despite the restriction analysis which suggested flop due to the product being cut by HpaI and not by AvaI. The sequence of the GluR4 product did not have a restriction site for either HpaI or for AvaI. Only a fraction of GluR4 product was ever cut by HpaI, which could have been due either to inefficient cutting or to the presence of both forms. The lack of an HpaI site in the sequenced product, as well as the higher homology to flip, suggests that flip was the predominant
R. Temkin et al.r Molecular Brain Research 52 (1997) 38–45
43
Fig. 5. Restriction fragment analysis of fliprflop splice variants. ŽA. The map shows the location of the alternative sequence region of the receptor. ŽB. The maps illustrate the restriction enzyme sites which distinguish the rat flip and flop sequence for each subunit. ŽC. Gels show results of PCR product restriction digest. GluR1 is cut by either AluI or HphI. GluR2 is cut by AvaI but not by HpaI. GluR4 is cut by HpaI but not by AvaI, however sequencing of the PCR products revealed the AvaI site to be missing in GluR4 flip.
form in the sample prepared for sequencing. The partial cutting with HpaI still suggests that some flop form is also present. The lack of cutting by AvaI, which had appeared to suggest the absence of the flip form, was due to only a single nucleotide mismatch in the recognition site. Therefore AvaI cannot be used for distinguishing GluR4 splice forms in chick.
4. Discussion Using PCR, we have detected the presence in amotoneurons of mRNA for glutamate receptor subunits of two of the agonist selective classes: AMPA selective GluR1, 2, and 4, and kainate selective GluR6 and 7. The AMPA subunits are modified by alternative splicing. Although GluR1 seems to exist in comparable amounts of both variants, GluR 2 is present only as flip, and GluR4 is present primarily as flip, with small amounts of flop present. Our detection of all AMPA selective subunits and NR1
in rat spinal cord is consistent with previous results which have detected by in situ hybridization the presence of GluR1–4 in rat spinal cord in both dorsal and ventral horn w29,34x and NR1 in spinal cord and in motoneurons w34x. Our inability to detect NR1 in the motoneurons with the primers used suggests that the chick NMDA receptor has sequence different from that of rat in the region of at least one of the primers. However, an NMDA receptor must be present, as it has been demonstrated electrophysiologically. We detected all of the kainate selective subunits in rat spinal cord. In adult rat spinal cord, in situ hybridization for kainate selective subunits detected GluR5, GluR7, KA1 and KA2 w34x, although, of these, only GluR5 and KA1 were localized to motoneurons. GluR6 was not detected by in situ hybridization in spinal cord. By PCR, we were able to detect GluR6 and GluR7 in both rat spinal cord and chick motoneurons. The greater sensitivity of PCR is possibly responsible for the detection of low abundance subunits missed by in situ, although actual differences in expression between species is feasible. We were unable to detect either GluR5 or KA1 in the chick motoneurons.
44
R. Temkin et al.r Molecular Brain Research 52 (1997) 38–45
This discrepancy between our results and those of Tolle ¨ et al. w34x may be due to difference in subunit expression between rat and chick or to differences in expression between adult and embryonic motoneurons. Alternatively, species sequence differences in some subunits may have led to mismatch between the rat based primers and the chick mRNA sequence and prevented amplification of these subunits. By comparing electrophysiological recordings of nonNMDA receptors in motoneurons to the known characteristics of each cloned subunit, we have attempted to account for the motoneuron responses by the presence of various subunits. In whole cell recordings, kainate can give rise to two responses, slightly desensitizing and non-desensitizing. The presence of GluR6 in the motoneurons is likely to account for the desensitizing component of the desensitizing kainate response, while GluRs 1, 2, and 4 are likely to account for the steady state of either response. Despite the fact that whole cell recordings imply the existence of AMPA selective and kainate selective subunits in the same cell, outside-out patch recordings, in which a subset of patches respond only to kainate and not to AMPA, suggest that each class of subunits assorts separately on the cell surface. Our results are consistent with experiments that demonstrated that subunits of each class co-precipitated independently w8,27,36x. This supports the concept that synaptic specificity can be achieved by independent localization of glutamate receptor subunits. Specificity of response is affected not only by which subunits are present in a receptor but also by modification of the individual subunits. Desensitization kinetics of AMPA responses are regulated by the alternatively spliced forms of GluR1–4. Motoneuron whole cell responses to glutamate or AMPA ŽFig. 1. typically display a steady state to peak ratio which is characteristic of the expressed flip forms of GluR1 and 2 w33x. As GluR2 has been shown to dominate regulation of the steady state amplitude w33x, the presence of only the flip form of GluR2 likely accounts for the sustained current usually seen in the motoneurons. Although both splice variants of GluR1 are present, the peak to steady state ratio appears to coincide with the flip form alone, as the flop form gives rise to an almost nonexistent desensitizing component w33x. Thus, the flip form of GluR1 appears to be dominant when both variants are present. The rate of desensitization has been shown to be controlled primarily by GluR3 and GluR4, as the flop forms of GluR3 and 4 desensitize faster than the flip forms when expressed in Xenopus oocytes w23x. In AMPA-responsive outside-out patches from motoneurons, the desensitization time constant Žtdes s 4.97 " 2.17 ms. more closely resembles that of the GluR4 fliprGluR2 flip combination in oocytes Žtdes s 6.1 " 1.5 ms. than that of GluR4 floprGluR2 flip Žtdes s 1.1 " 0.2 ms. w23x. Thus, it appears that GluR4 flip might dominate GluR4 flop in control of desensitization rate in heteromeric receptors. It is also possible that GluR4 flop is in low abundance
relative to GluR4 flip. This possibility is supported by the fact that the flop form is not apparent in the sequence of the chick spinal cord GluR4 PCR product, and that HpaI usually cuts only a small fraction of the chick motoneuron GluR4 product. Previously, the alternate splice forms were shown to be developmentally regulated w20x. Flip forms were shown to predominate early in development, while flop form levels begin to increase around postnatal day 8 and are the predominant form in the adult. In our embryonic day 6.5 chick motoneurons, both flip and flop forms of subunits were present. However electrophysiological recordings suggested that the flip forms dominated in control of the response to agonist. Therefore, our results are not inconsistent with flip being important early in development during synapse formation. GluR2 is edited in these chick motoneurons, but the edited form of GluR6 has been shown to be genomically encoded w19x. The absence of GluR6 editing in chick indicates that the aspects of development for which this switch is necessary in rat do not exist in chick or are performed by another mechanism. A similar absence of RNA editing has been demonstrated in another non-mammalian species, the fish species, Oreochromis sp. w18x. In this species, however, the lack of editing is seen for GluR2, which is edited in chick. The subunits which we have detected are concordant with the electrophysiological responses of the motoneurons to various agonists. As there are some minor differences in response between cells, it is possible that the subunits detected in the motoneuron population may or may not be present in each individual cell.
Acknowledgements We thank Dr. Stephen Heinemann at the Salk Institute for providing the glutamate receptor clones used in this study. Also, we thank Prof. J. Dudel for his support during experiments performed at the Physiology Institute, Technical University of Munich. This work was supported by NIH Grants NS13600 and GM07507 and the Deutsche Forschungsgemeinschaft ŽSFB 220..
References w1x P. Ascher, L. Nowak, Quisqualate- and kainate-activated channels in mouse central neurones in culture, J. Physiol. 399 Ž1988. 227–245. w2x P. Ascher, P. Bregestovski, L. Nowak, N-Methyl-D-aspartateactivated channels of mouse central neurones in magnesium-free solutions, J. Physiol. 399 Ž1988. 207–226. w3x S. Bahn, B. Volk, W. Wisden, Kainate receptor gene expression in the developing rat brain, J. Neurosci. 14 Ž1994. 5525–5547. w4x G. Bernardi, W. Ziegelgansberger, A. Herz, E.A. Puil, Intracellular studies on the action of L-glutamic acid on spinal neurons of the cat, Brain Res. 39 Ž1972. 523–525.
R. Temkin et al.r Molecular Brain Research 52 (1997) 38–45 w5x B. Bettler, J. Boulter, I. Hermans-Borgmeyer, A. O’Shea-Greenfield, E.S. Deneris, C. Moll, U. Borgmeyer, M. Hollmann, S.F. Heinemann, Cloning of a novel glutamate receptor subunit, GluR5: expression in the nervous system during development, Neuron 5 Ž1990. 583–595. w6x B. Bettler, J. Egebjerg, G. Sharma, G. Pecht, I. Hermans-Borgmeyer, C. Moll, C.F. Stevens, S. Heinemann, Cloning of a putative glutamate receptor: A low affinity kainate-binding subunit, Neuron 8 Ž1992. 257–265. w7x J. Boulter, M. Hollmann, A. O’Shea-Greenfield, M. Hartley, E. Deneris, C. Maron, S.F. Heinemann, Molecular cloning and functional expression of glutamate receptor subunit genes, Science 249 Ž1990. 1033–1037. w8x N. Brose, G.W. Huntley, Y. Stern-Bach, G. Sharma, J.H. Morrison, S.F. Heinemann, Differential assembly of coexpressed glutamate receptor subunits in neurons of rat cerebral cortex, J. Biol. Chem. 269 Ž1994. 16780–16784. w9x D.R. Curtis, J.C. Watkins, Acidic amino acids with strong excitatory actions on mammalian neurones, J. Physiol. 166 Ž1963. 1–14. w10x L.M. Dahm, L.T. Landmesser, The regulation of intramuscular nerve branching during normal development and following activity blockade, Dev. Biol. 130 Ž1988. 621–644. w11x J. Dudel, C. Franke, H. Hatt, Rapid activation, desensitization and resensitization of synaptic channels of crayfish after glutamate pulses, Biophys. J. 57 Ž1990. 535–545. w12x J. Egebjerg, Bettler, I. Hermans-Borgmeyer, S.F. Heinemann, Cloning of a cDNA for a glutamate receptor subunit activated by kainate but not AMPA, Nature 351 Ž1991. 745–748. w13x A. Herb, N. Burnashev, P. Werner, B. Sakmann, W. Wisden, P.H. Seeburg, The KA-2 subunit of excitatory amino acid receptors shows widespread expression in brain and forms ion channels with distantly related subunits, Neuron 8 Ž1992. 775–785. w14x M. Hollmann, S. Heinemann, Cloned glutamate receptors, Annu. Rev. Neurosci. 17 Ž1994. 31–108. w15x M. Hollmann, C. Maron, S. Heinemann, N-Glycosylation site tagging suggests a three transmembrane domain topology for the glutamate receptor GluR1, Neuron 13 Ž1994. 1331–1343. w16x K. Keinanen, W. Wisden, B. Sommer, P. Werner, A. Herb, T.A. ¨ Verdoorn, B. Sakmann, P.H. Seeburg, A family of AMPA-selective glutamate receptors, Science 249 Ž1990. 556–560. w17x M. Kohler, N. Burnashev, B. Sakmann, P.H. Seeburg, Determinants ¨ of Ca2q permeability in both TM1 and TM2 of high affinity kainate receptor channels: diversity by RNA editing, Neuron 10 Ž1993. 491–500. w18x S. Kung, Y. Wu, W. Chow, Characterization of two fish glutamate receptor cDNA molecules: absence of RNA editing at the QrR site, Mol. Brain. Res. 35 Ž1996. 119–130. w19x D.L. Lowe, K. Jahn, D.O. Smith, Glutamate receptor editing in the mammalian hippocampus and avian neurons, Mol. Brain Res. 48 Ž1997. 37–44. w20x H. Monyer, P.H. Seeburg, W. Wisden, Glutamate-operated channels: developmentally early and mature forms arise by alternative splicing, Neuron 6 Ž1991. 799–810. w21x H. Monyer, R. Sprengel, R. Schoepfer, A. Herb, M. Higuchi, H. Lomeli, N. Burnashev, B. Sakmann, P.H. Seeburg, Heteromeric NMDA receptors: molecular and functional distinction of subtypes, Science 256 Ž1992. 1217–1221.
45
w22x K. Moriyoshi, M. Masu, T. Ishii, R. Shigemoto, N. Mizuno, S. Nakanishi, Molecular cloning and characterization of the rat NMDA receptor, Nature 354 Ž1991. 31–37. w23x J. Mosbacher, R. Schoepfer, H. Monyer, N. Burnashev, P.H. Seeburg, J.P. Ruppersberg, A molecular determinant for submillisecond desensitization in glutamate receptors, Science 266 Ž1994. 1059– 1062. w24x N. Nakanishi, N.A. Shneider, R. Axel, A family of glutamate receptor genes: evidence for the formation of heteromultimeric receptors with distinct channel properties, Neuron 5 Ž1990. 569–581. w25x R.J. O’Brien, G.D. Fischbach, Characterization of excitatory amino acid receptors expressed by embryonic chick motoneurons in vitro, J. Neurosci. 6 Ž1986. 3275–3283. w26x T. Paperna, Y. Lamed, V.I. Teichberg, cDNA cloning of chick brain a-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors reveals conservation of structure, function and post-transcriptional process with mammalian receptors, Mol. Brain Res. Ž1996. 101–113. w27x R.B. Puchalski, J.C. Louis, N. Brose, S.F. Traynelis, J. Egebjerg, V. Kukekov, R.J. Wenthold, S.W. Rogers, F. Lin, T. Moran, J.H. Morrison, S.F. Heinemann, Selective RNA editing and subunit assembly of native glutamate receptors, Neuron 13 Ž1994. 131–147. w28x J.L. Rosenheimer, D.O. Smith, Age-related increase in soluble and cell surface-associated neurite-outgrowth factors from rat muscle, Brain Res. 509 Ž1990. 309–320. w29x K. Sato, H. Kiyama, M. Tohyama, The differential expression patterns of messenger RNAs encoding non-N-methyl-D-aspartate glutamate receptor subunits ŽGluR1–4. in the rat brain, Neuroscience 52 Ž1993. 515–539. w30x R.L. Schnarr, A.E. Schaffner, Separation of cell types from embryonic chicken and rat spinal cord: characterization of motoneuron enriched fractions, J. Neurosci. 1 Ž1981. 204–217. w31x D.O. Smith, C. Franke, J.L. Rosenheimer, F. Zufall, H. Hatt, Glutamate-activated channels in adult rat ventral spinal cord cells, J. Neurophysiol. 66 Ž1991. 369–378. w32x D.O. Smith, C. Franke, J.L. Rosenheimer, F. Zufall, H. Hatt, Desensitization and resensitization rates of glutamate-activated channels may regulate motoneuron excitability, J. Neurophysiol. 66 Ž1991. 1166–1175. w33x B. Sommer, K. Kainanen, T.A. Verdoorn, W. Wisden, N. Burna¨ shev, A. Herb, M. Kohler, T. Takagi, B. Sakmann, P.H. Seeburg, ¨ Flip and Flop: A cell-specific functional switch in glutamate-operated channels of the CNS, Science 249 Ž1990. 1580–1585. w34x T.R. Tolle, A. Berthele, W. Zieglgansberger, P.H. Seeburg, W. ¨ ¨ Wisden, The differential expression of 16 NMDA and non-NMDA receptor subunits in the rat spinal cord and in periaqueductal gray, J. Neurosci. 13 Ž1993. 5009–5028. w35x L.O. Trussell, G.D. Fischbach, Glutamate receptor desensitization and its role in synaptic transmission, Neuron 3 Ž1989. 209–218. w36x R.J. Wenthold, N. Yokotani, K. Doi, K. Wada, Immunochemical characterization of the non-NMDA glutamate receptor using subunit-specific antibodies, J. Biol. Chem. 267 Ž1992. 501–507. w37x C.F. Zorumski, J. Yang, AMPA, kainate, and quisqualate activate a common receptor-channel complex on embryonic chick motoneurons, J. Neurosci. 8 Ž1988. 4277–4286.
Research report
Expression of glutamate receptor subunits in a-motoneurons Roselynn Temkin a , Deborah Lowe a , Penny Jensen b, Hanns Hatt c , Dean O. Smith a
d,)
Neuroscience Training Program, UniÕersity of Wisconsin, 1300 UniÕersity AÕenue, Madison, WI 53706, USA b Department of Physiology, UniÕersity of Wisconsin, 1300 UniÕersity AÕenue, Madison, WI 53706, USA c Ruhr-UniÕersitat 150, 44780 Bochum, Germany ¨ Bochum, Lehrstuhl fur ¨ Zellphysiologie, UniÕersitatstr. ¨ d Pacific Biomedical Research Center, UniÕersity of Hawaii, Honolulu, HI 96816, USA Accepted 15 July 1997
Abstract Whole-cell recordings from 6.5 day embryonic chick a-motoneurons indicated the presence of AMPA, kainate, and NMDA glutamate receptor subtypes in each motoneuron tested. AMPA consistently evoked a desensitizing response, while kainate could evoke either a desensitizing or non-desensitizing whole-cell response. In excised membrane patches, desensitizing AMPA responses appeared to be colocalized with non-desensitizing kainate responses. Desensitizing kainate responses were seen in some patches which were not responsive to AMPA, suggesting that kainate selective subunits and AMPA selective subunits localize separately on the motoneuron membrane. To determine which of the known glutamate receptor subunits might underlie these responses, we used RT-PCR amplification to detect subunits present in mRNA isolated from adult rat spinal cord and from a highly enriched motoneuron population from embryonic chick. Sequencing of the the amplified cDNA was used to verify the identity of the products and of the alternative splice variants of GluR1–4. In rat spinal cord, all subunits that we attempted to detect, including AMPA selective subunits GluR1–4, kainate selective subunits GluR5–7 and KA1–2, and NMDA subunit NR1 were present. The isolated motoneurons also contained AMPA subunits GluR1, 2, and 4, and kainate subunits GluR6 and 7. The GluR2 and 4 subunits were specifically processed by splicing, present primarily as the flip splice form. q 1997 Elsevier Science B.V. Keywords: Glutamate; AMPA; Kainate; N-Methyl-D-aspartate ŽNMDA.; Motoneuron; Spinal cord; Alternative splicing; RNA
1. Introduction Glutamate is a major excitatory neurotransmitter. In the spinal cord, for example, it depolarizes motoneurons in embryonic w25x and adult w9,4x animals. Furthermore, receptors sensitive to the three glutamate agonists AMPA, kainate, and N-methyl-D-aspartate ŽNMDA., have been identified in spinal cord cells w1,2,31x and, more specifically, in a-motoneurons dissociated from both embryonic chick and neonatal rat w32,35,37x. Within the past six years, the genetic basis of glutamate-mediated synaptic transmission has been resolved at the molecular level and has been found to be remarkably diverse. cDNA encoding at least 14 different glutamate receptor subtypes has been cloned from rodent brain and sequenced, and many different spliced or edited variants
)
Corresponding author. University of Hawaii, 2444 Dole Street, Bachman 204, Honolulu, HI 96822, USA. Fax: q1 Ž808. 956-8061. 0169-328Xr97r$17.00 q 1997 Elsevier Science B.V. All rights reserved. PII S 0 1 6 9 - 3 2 8 X Ž 9 7 . 0 0 2 4 9 - 0
have been identified w14x. Functional expression revealed four different cDNA molecules ŽGluR1–GluR4. that encode receptors activated by both AMPA and kainate w7,15,16,24x. The GluR1–4 subunits exist in alternately spliced variants that affect the desensitization of the receptors. These variants have been designated flip and flop w33x. In addition, three subunits ŽGluR5–7. that bind kainate but not AMPA have been cloned w5,6,12x. Agonist sensitivity and ionic permeability of GluR5–7 can also vary depending on the presence of associated kainate-binding proteins ŽKA1 and KA2. w13x. Lastly, five cDNA molecules ŽNR1 and NR2A-D. that encode NMDA-activated receptors have also been cloned w21,22x. With such diversity of genetic message, how is synaptic specificity achieved? In situ hybridization of AMPA-selective receptors ŽGluR1–4. w16,29x and of kainate-selective receptors ŽGluR5–7, KA1, and KA2. w3x revealed that the various subunits display different regional distribution. Each subunit was found to be localized to certain brain regions, and this localization was distinctly different for all
R. Temkin et al.r Molecular Brain Research 52 (1997) 38–45
subunits. Based on this result, a population of similar neurons might be hypothesized to express some subset of the glutamate receptor subunits that would account for cell type-specific responses. In this study, we tested this hypothesis by assaying expression of the specific glutamate receptor subtypes in a population of early embryonic Žstage 28. chick amotoneurons. At this developmental stage, differentiation into specific motor units has presumably not yet occurred, for individual motoneurons are just at the point of establishing initial synaptic connections w10x. Furthermore, at this stage, there is little variability in the currents activated by glutamate w32x. We show that mRNA encoding a subset of AMPA and kainate subtypes was present in this population of neurons. Specific pre-mRNA splicing had occurred, with the flip forms appearing to predominate. Genomic and cDNA sequence analysis of chick glutamate receptors also demonstrated avian editing of GluR2 but not GluR6 Žw19x; cf. w17x..
2. Materials and methods 2.1. Experimental preparation In an initial series of control experiments, glutamate receptor expression was tested in rat spinal cord. Adult rats were anesthetized by intraperitoneal injection of chloral hydrate Ž2.8 mmolrkg, i.p... Parallel incisions were made above spinal vertebra T12 and below L 2 , and the vertebral column and spinal cord were cut at these incisions. This segment of spinal cord was removed by forcing cold saline solution through the vertebral column with a syringe. Spinal roots were trimmed from the spinal cord segment. The cord was then frozen immediately in liquid nitrogen, where it was stored until mRNA extraction. An enriched fraction of chick motoneurons was prepared using techniques described in detail in Rosenheimer and Smith w28x. Lumbar spinal cords from 6.5-day Žstage 28. embryonic chicks were dissected, freed of their meninges and dorsal root ganglia, and collected in ice-cold buffer. Following incubation at 378 with 0.05% trypsin and 0.005% DNase I, cells were dissociated by gentle trituration. Motoneuron-enriched cell fractions were then generated on the basis of their buoyant density in a metrizamide solution w30x. The effectiveness of this procedure for enrichment of motoneurons was verified by labeling the motoneurons via cut ventral roots with the fluorescent dye rhodamine prior to dissociation. Approximately 92% of the recovered cells contained the fluorescent label and were identified specifically as motoneurons w28x. Their diameters were measured using a Coulter counter. The diameters were distributed in a unimodal ‘bell-shaped’ fashion, and the average value Ž"S.D.. was 10 " 2 m m Ž n s 4094.. The remaining 8% of the cells were not identified. Because of their size, we
39
suspect that they were motoneurons that had not been labeled sufficiently for fluorescence detection. After isolation, the cells were used either for electrophysiological recording or for mRNA extraction. Cells for recording were plated and incubated in Dulbecco’s modified Eagle medium plus 10% horse serum at 378 with 5% CO 2 . Prior to recording, the DME was replaced by HEPES buffered bath solution containing Žin mM.: NaCl 140, KCl 5.3, CaCl 2 2, Na 2 HPO4 0.67, KH 2 PO4 0.22, HEPES 15, and glucose 5.6. The pH was adjusted to 7.4 using either 10 N HCl or NaOH. Recordings were obtained from cells maintained between 1 and 2 days in culture. Similar results were obtained regardless of the time in culture. 2.2. Electrophysiological recording Membrane currents were recorded at room temperature Ž22 to 248C. in whole-cell or outside-out configuration. Throughout the recording experiments, the cells were bathed with continuously flowing bath solution. Glutamate was dissolved in this extracellular saline. The electrode solution contained Žin mM.: KCl 140, CaCl 2 1, MgCl 2 2, EGTA 11, glucose 10, and HEPES 10 ŽpH s 7.4.. For whole-cell recording, cAMP Ž100 m M. was included in the pipette except where noted. The currents were amplified ŽList EPC7 and Dagan 8900. and stored on video tape for subsequent evaluation. Resting membrane potentials were generally between y45 and y60 mV. Records were filtered at 1 to 4 kHz using an 8-pole Bessel filter ŽFrequency Devices, Haverhill, MA. and digitized at 10 to 50 kHz. They were analyzed using a PDP 11r73-based acquisition system ŽIndec Systems, CA.. Curve-fitting routines used the Marquardt-Levenberg techniques provided in MicroCal Origin software ŽMicroCal, Northampton, MA.. 2.3. Agonist application To achieve rapid agonist application, a ‘liquid-filament’ system was used to apply ligand-containing solution to the electrode tip w11x. Ligands were kept in separate pressurized Ž1 atu. chambers connected in parallel by a manual valve to the delivery system. The ligand-containing solution was ejected under pressure from the storage chamber into a polyethylene tube drawn to a 30-m m tip. This tube was fixed in the bath parallel to the direction of flow of the background bathing solution. The ligand solution exited the polyethylene tube as a thin Ž30 m m. stream flowing at the same rate as the surrounding background solution Ž6 mlrmin.. After traveling a distance of 5 mm, the agonist and the surrounding background solutions were removed from the chamber using a pulse-free pump. This ‘peristaltic-type’ pump Žmanufactured in the Physiology Institute, Technical University of Munich. utilized an eccentric bearing to bring 16 teeth onto the uptake tubing; one tooth contacted the tubing at all times, thus minimizing dead
40
R. Temkin et al.r Molecular Brain Research 52 (1997) 38–45
space in the tube, diffusion of agonist into the background solution, and deviations from laminar flow. The ligand-delivery tube was connected to a piezo crystal. On application of 100 V to the piezo, the tube shifted upward 20 m m within about 200 m s. With this system, ligand was introduced to the electrode tip at a constantly maintained concentration within - 1 ms under optimal conditions w11x; receptor activation times were somewhat slower, for they depended on patch characteristics and recording configuration Žactivation times were slower in whole-cell recordings.. On cessation of the voltage pulse, the ligand-containing solution was lowered rapidly, and the electrode tip was again exposed to normal background solution. Spurious results could arise from any inconsistencies in agonist delivery or recording conditions. For example, excessive diffusion from the agonist stream into the background solution could cause receptor desensitization, which might then be interpreted as failure to respond to subsequent agonist application. Therefore, to monitor these conditions, control applications of glutamate Žusually 1 mM. were delivered between sequential applications of the other specific agonists. Stable ligand-delivery and recording conditions could sometimes be obtained for times exceeding 15 min. Regression analyses indicated that neither the rise time Žfrom 20% to 80% of peak amplitude. nor the peak amplitude varied significantly Ž P ) 0.05. in these optimal recordings. Results from these stable experiments were then used for further study.
Table 1 PCR primers X
GluR-1
5 primer – TGGTGGTTCTTCACCCTGATCAT X 3 primer – TATGGCTTCATTGATGGATTGC
GluR-2
5 X 3 X 5 X 3
GluR-3
5 primer – GAGTCGACCATGAACGAGTA X 3 primer – AATTCTGGGTATTGGTGCTGG
GluR-4
5 primer – TGGTGGTTCTTCACCCTGATCAT X 3 primer – ACTCCCAGTGATGGATAACCTG
GluR-5
5 primer – TGGTGGTTCTTCACCCTGATCAT X 3 primer – CACTGCTCAACGTCATTGTTCT
GluR-6
5 X 3 X 5 X 3
GluR-7
5 primer – TGGTGGAAGACGGCAAGTAC X 3 primer – CCTTGTTGCATCAGGGAGCC
KA-1
5 primer – GGCCAGTGGCATGTGGCAGA X 3 primer – AAGACACAGCTGACAGCCAG
KA-2
5 primer – GGGGTGTGGTACTCTAACCG X 3 primer – AAGACACAGCTGACAGCCAG
NRa-1
5 primer – CTGTGGTTGCTAGTAGGACTGT X 3 primer – GCATTCCTGATACCGAACCCAT
X
primer primer primer primer
– – – –
TGGTGGTTCTTCACCCTGATCAT TGCAAAATTCTGGGAATTCTGC AGGAAATGACACGTCTGGGC GTGTAGGAGGAGATTATGAT
X
X
X
X
primer primer primer primer
– – – –
TGGTGGTTTTTCACACTTATCAT CGCTGGCACTTCAGGGACATTC GGACAATGGAATGGAATGGTTC GTATACGAAGAAATGATGAT
X
X
X
X
X
X
Oligonucleotide sequences are presented in the 5 to 3 direction.
2.4. mRNA isolation and GluR amplification Poly-A mRNA was isolated using a commercial kit ŽMicro-FastTrack, Invitrogen, San Diego, CA.. Chick amotoneurons were prepared for mRNA isolation by first being pelleted by centrifugation and then resuspended in 2% SDS, 1.5 mM MgCl 2 , 200 mM Tris-HCl pH 7.5, 200 mM NaCl lysis buffer containing RNase inhibitor ŽInvitrogen.. Frozen rat spinal cords were homogenized in identical lysis buffer. After oligo ŽdT. cellulose spin-column selection of poly-A mRNA and ethanol precipitation, the mRNA was resuspended in sterile distilled water and was ready for reverse transcription. The mRNA was reversed transcribed in a 20 m l reaction with 50 units of M-MLV reverse transcriptase and 1.5 m M of the 3X primer for a specific glutamate receptor subunit. The cDNA was then amplified by PCR in a 100 m l reaction with 2.5 units of AmpliTaq DNA polymerase. Primers specific for only a single glutamate receptor subunit were used for each amplification. Their nucleotide sequences are presented in Table 1. As a positive control for the appropriate size of product, glutamate receptor cDNA clones were amplified with the same primers. These clones were kindly provided by Stephen Heinemann ŽSalk Institute, La Jolla, CA, USA.. In rat, alternative splice forms can be distinguished by
restriction fragment length polymorphism. For GluR1, only the flip form of the amplified product can be cut with AluI, while only the flop form can be cut with HphI. For the GluR2 and GluR4 products, the flip forms can be cut with AvaI, and the flop forms with HpaI. For GluR3, the flip form can be cut with MspI, and flop with HpaI. Therefore, we initially tested for the alternative splice variants of the chick PCR products by restriction digest analysis. The identity and splice form of chick PCR products from spinal cord were tested by sequencing. Sequencing was performed by the University of Wisconsin Biotechnology Center. Sequencing of the TM2 region of GluR2 and the TM1 and TM2 regions of GluR6 was also used to determine the proportion of edited variants present in this cell population. 2.5. Chemicals Most biochemicals were obtained from Sigma ŽSt. Louis, MO.. DME was purchased from Gibco BRL ŽGaithersburg, MD.. Restriction enzymes were from Promega ŽMadison, WI., except for HphI which was from New England Biolabs ŽBeverly, MA.. All enzymes and reagents for RT-PCR were obtained from Perkin Elmer ŽNorwalk, CT.. Other sources are noted in context.
R. Temkin et al.r Molecular Brain Research 52 (1997) 38–45
3. Results
Table 2 Desensitization time constants
3.1. Responses to glutamate and its agonists
ŽA. Whole-cell kainate response With cAMP tdes s 58.1"10.3 ms, ns 28 Without cAMP tdes s66.4"12.6 ms, ns 22 ŽB. Outside-out patch AMPA tdes s 4.97"2.17 ms, ns129 Kainate tdes s 59.2"8.2 ms, ns110
In whole-cell configuration, currents were activated by the three glutamate agonists AMPA, kainate, and NMDA. Receptors for all three agonists were present on each motoneuron tested by sequential application of the different agonists at concentrations ranging from 10 m M to 10 mM ŽFig. 1.. In most cases, AMPA-activated currents decayed to a sustained steady-state level that was 23 " 8% of peak Ž n s 20.. In rare cases, the steady state current was negligible Ž5 " 1% of peak, n s 3.. Cyclothiazide Ž100 m M. caused a 62% Ž"12% S.D., n s 4. increase in steady-state amplitude of these AMPA-induced currents. Kainate Ž1 mM. evoked two disparate responses, either non-desensitizing ŽFig. 1C. or with a small desensitizing component ŽFig. 1D.. This is probably a function of intracellular cAMP levels. With 100 m M cAMP in the electrode solution, 71% of motoneurons exhibited the desensitizing response Ž n s 28., whereas without cAMP in the electrode, only 15% of responses were desensitizing following activation by kainate Ž n s 39.. The desensitization time constants were determined by fitting a single exponential curve to the decay phase of the current in the presence and absence of cAMP. They were not significantly different ŽTable 2A.. GluR6 expressed homomerically desensitizes in response to kainate w12x, while the AMPA preferring GluR1–4 do not. Cyclothiazide Ž100 m M. caused a 7% Ž"6% S.D., n s 3. increase in the non-desensitizing kainate-evoked responses, which is not significant statistically, but a significant Ž P - 0.05. 22% Ž"12% S.D., n s 5. potentiation of the desensitizing currents. Therefore, the desensitizing component seen in re-
Fig. 1. Motoneuron whole-cell recordings. In each case, the holding potential was y40 mV. The concentration of each agonist was 1 mM. The duration of agonist application is indicated by the bar above the records. ŽA,B. Glutamate ŽA. and AMPA ŽB. responses both exhibit rapidly decaying current. ŽC,D. Whole cell current elicited by kainate differs between cells. Each cell displays one of two alternate responses; ŽC. some cells exhibited a non-desensitizing response to kainate. ŽD. Other cells exhibited a slightly desensitizing response to kainate. ŽE. A single exponential is fit to the desensitizing component of whole-cell kainate response.
41
Means"S.E.
sponse to kainate in some whole-cell recordings may be a result of the activation of kainate selective subunits, while the steady state current of both the desensitizing and non-desensitizing whole-cell responses may be mediated by kainate activation of the AMPA selective subunits. Since subunit composition of glutamate receptors affects channel kinetics as well as agonist sensitivity, desensitization kinetics were measured in outside-out patches. In those patches which responded to AMPA, the receptors desensitized rapidly in response to AMPA but not in response to kainate ŽFig. 2A; Table 2., characteristic of the AMPA selective subunits GluR1–4. Other patches displayed rapidly decaying responses to kainate but did not respond to AMPA ŽFig. 2B; Table 2., characteristic of the kainate selective subunits. Thus it appears that each class of glutamate receptor subunit may localize separately on the motoneuron surface. 3.2. Glutamate receptor mRNA To determine which glutamate receptor subunits may contribute to these motoneuron responses, we used PCR to
Fig. 2. Responses to AMPA and kainate in outside-out patches from chick motoneurons. Holding potential was y60 mV. The agonists Ž1 mM. were applied sequentially to the same patches. The duration of agonist application is indicated by the bar above the records. ŽA. AMPA was applied initially, and then, after removal of the AMPA, kainate was applied to the same patch. ŽB. Kainate was applied initially, and then, after removal of the kainate, AMPA was applied to the same patch. The patch did not respond to the application of AMPA. To confirm that the lack of response was not due to desensitization, kainate was reapplied to the same patch, and it evoked a response.
42
R. Temkin et al.r Molecular Brain Research 52 (1997) 38–45
detect the presence of the various subunits. Chick glutamate receptor sequences are available only for GluRs 1–4 w26x, so some of our primers are based on rat mRNA sequence. Therefore, we tested the efficiency of our primers on rat spinal cord mRNA. In the rat spinal cord, every glutamate receptor subunit that we attempted to amplify was detected ŽFig. 3.. The detected subunits include GluR1–4, GluR5–7, KA1–2 and NR1. The products from spinal cord were verified by sequencing. We then used the same primers to determine which subunits were present in a homogeneous population of chick motoneurons. In the isolated chick motoneurons, primers for AMPA selective subunits GluR1, 2, and 4 amplified products of the correct size ŽFig. 4., as did kainate selective subunits GluR6 and 7 ŽFig. 4B.. All chick motoneuron products were confirmed by sequencing, and they showed at least 80% similarity to the homologous region in rat. We were unable to detect AMPA selective subunit GluR3 using primers based on rat or chick w26x sequences. Nor could we detect kainate selective subunits GluR5, KA1, and KA2, or NMDA subunit NR1. Fig. 4. Glutamate receptor expression in chick a-motoneurons. In each gel, lane 1 is HincII-digested f X174 DNA size marker. Lane 2 Žpc. is PCR product amplified from the cloned subunit cDNA. Lane 3 Žmn. is PCR product from the chick motoneurons. Lane 4 Žnt. is the no template control ŽNT.. AMPA selective subunits GluR1, 2, 3, and 4 were detected. Subunits GluR6 and 7 were the only kainate selective subunits detected in motoneurons.
3.3. AlternatiÕe splicing
Fig. 3. Glutamate receptor expression in rat spinal cord. In each gel, lane 1 is a HincII-digested f X174 DNA size marker ŽUnited States Biochemical, Cleveland, OH, USA.. Lane 2 Žpc. is PCR product amplified from the cloned subunit cDNA. Lane 3 Žsc. is PCR product from rat spinal cord. Lane 4 Žnt. is the no template control. AMPA selective subunits GluR1–4 were detected. Kainate selective subunits GluR5–7 and KA1–2 were also present. The primary NMDA receptor subunit, NR1, was present.
The AMPA selective subunits, GluR1–4, can exist in two alternatively spliced forms, designated flip and flop w33x. These splice forms apparently regulate the rate and amount of desensitization of glutamate activated responses. In rat, the splice forms can be distinguished by restriction fragment length polymorphism. Moreover, there is 99% homology between rat and chick sequence in the alternatively spliced exon region w26x. Therefore, we initially tested for the alternative splice variants of the chick PCR products by restriction digest analysis ŽFig. 5.. By this analysis, the chick motoneurons appeared to contain both the flip and flop splice variants of GluR1, the flip splice form of GluR2, and the flop splice form of GluR4. Subsequent sequencing verified the splice variants of GluR1 and GluR2. Sequencing of the GluR4 product showed higher homology to rat flip than to flop, despite the restriction analysis which suggested flop due to the product being cut by HpaI and not by AvaI. The sequence of the GluR4 product did not have a restriction site for either HpaI or for AvaI. Only a fraction of GluR4 product was ever cut by HpaI, which could have been due either to inefficient cutting or to the presence of both forms. The lack of an HpaI site in the sequenced product, as well as the higher homology to flip, suggests that flip was the predominant
R. Temkin et al.r Molecular Brain Research 52 (1997) 38–45
43
Fig. 5. Restriction fragment analysis of fliprflop splice variants. ŽA. The map shows the location of the alternative sequence region of the receptor. ŽB. The maps illustrate the restriction enzyme sites which distinguish the rat flip and flop sequence for each subunit. ŽC. Gels show results of PCR product restriction digest. GluR1 is cut by either AluI or HphI. GluR2 is cut by AvaI but not by HpaI. GluR4 is cut by HpaI but not by AvaI, however sequencing of the PCR products revealed the AvaI site to be missing in GluR4 flip.
form in the sample prepared for sequencing. The partial cutting with HpaI still suggests that some flop form is also present. The lack of cutting by AvaI, which had appeared to suggest the absence of the flip form, was due to only a single nucleotide mismatch in the recognition site. Therefore AvaI cannot be used for distinguishing GluR4 splice forms in chick.
4. Discussion Using PCR, we have detected the presence in amotoneurons of mRNA for glutamate receptor subunits of two of the agonist selective classes: AMPA selective GluR1, 2, and 4, and kainate selective GluR6 and 7. The AMPA subunits are modified by alternative splicing. Although GluR1 seems to exist in comparable amounts of both variants, GluR 2 is present only as flip, and GluR4 is present primarily as flip, with small amounts of flop present. Our detection of all AMPA selective subunits and NR1
in rat spinal cord is consistent with previous results which have detected by in situ hybridization the presence of GluR1–4 in rat spinal cord in both dorsal and ventral horn w29,34x and NR1 in spinal cord and in motoneurons w34x. Our inability to detect NR1 in the motoneurons with the primers used suggests that the chick NMDA receptor has sequence different from that of rat in the region of at least one of the primers. However, an NMDA receptor must be present, as it has been demonstrated electrophysiologically. We detected all of the kainate selective subunits in rat spinal cord. In adult rat spinal cord, in situ hybridization for kainate selective subunits detected GluR5, GluR7, KA1 and KA2 w34x, although, of these, only GluR5 and KA1 were localized to motoneurons. GluR6 was not detected by in situ hybridization in spinal cord. By PCR, we were able to detect GluR6 and GluR7 in both rat spinal cord and chick motoneurons. The greater sensitivity of PCR is possibly responsible for the detection of low abundance subunits missed by in situ, although actual differences in expression between species is feasible. We were unable to detect either GluR5 or KA1 in the chick motoneurons.
44
R. Temkin et al.r Molecular Brain Research 52 (1997) 38–45
This discrepancy between our results and those of Tolle ¨ et al. w34x may be due to difference in subunit expression between rat and chick or to differences in expression between adult and embryonic motoneurons. Alternatively, species sequence differences in some subunits may have led to mismatch between the rat based primers and the chick mRNA sequence and prevented amplification of these subunits. By comparing electrophysiological recordings of nonNMDA receptors in motoneurons to the known characteristics of each cloned subunit, we have attempted to account for the motoneuron responses by the presence of various subunits. In whole cell recordings, kainate can give rise to two responses, slightly desensitizing and non-desensitizing. The presence of GluR6 in the motoneurons is likely to account for the desensitizing component of the desensitizing kainate response, while GluRs 1, 2, and 4 are likely to account for the steady state of either response. Despite the fact that whole cell recordings imply the existence of AMPA selective and kainate selective subunits in the same cell, outside-out patch recordings, in which a subset of patches respond only to kainate and not to AMPA, suggest that each class of subunits assorts separately on the cell surface. Our results are consistent with experiments that demonstrated that subunits of each class co-precipitated independently w8,27,36x. This supports the concept that synaptic specificity can be achieved by independent localization of glutamate receptor subunits. Specificity of response is affected not only by which subunits are present in a receptor but also by modification of the individual subunits. Desensitization kinetics of AMPA responses are regulated by the alternatively spliced forms of GluR1–4. Motoneuron whole cell responses to glutamate or AMPA ŽFig. 1. typically display a steady state to peak ratio which is characteristic of the expressed flip forms of GluR1 and 2 w33x. As GluR2 has been shown to dominate regulation of the steady state amplitude w33x, the presence of only the flip form of GluR2 likely accounts for the sustained current usually seen in the motoneurons. Although both splice variants of GluR1 are present, the peak to steady state ratio appears to coincide with the flip form alone, as the flop form gives rise to an almost nonexistent desensitizing component w33x. Thus, the flip form of GluR1 appears to be dominant when both variants are present. The rate of desensitization has been shown to be controlled primarily by GluR3 and GluR4, as the flop forms of GluR3 and 4 desensitize faster than the flip forms when expressed in Xenopus oocytes w23x. In AMPA-responsive outside-out patches from motoneurons, the desensitization time constant Žtdes s 4.97 " 2.17 ms. more closely resembles that of the GluR4 fliprGluR2 flip combination in oocytes Žtdes s 6.1 " 1.5 ms. than that of GluR4 floprGluR2 flip Žtdes s 1.1 " 0.2 ms. w23x. Thus, it appears that GluR4 flip might dominate GluR4 flop in control of desensitization rate in heteromeric receptors. It is also possible that GluR4 flop is in low abundance
relative to GluR4 flip. This possibility is supported by the fact that the flop form is not apparent in the sequence of the chick spinal cord GluR4 PCR product, and that HpaI usually cuts only a small fraction of the chick motoneuron GluR4 product. Previously, the alternate splice forms were shown to be developmentally regulated w20x. Flip forms were shown to predominate early in development, while flop form levels begin to increase around postnatal day 8 and are the predominant form in the adult. In our embryonic day 6.5 chick motoneurons, both flip and flop forms of subunits were present. However electrophysiological recordings suggested that the flip forms dominated in control of the response to agonist. Therefore, our results are not inconsistent with flip being important early in development during synapse formation. GluR2 is edited in these chick motoneurons, but the edited form of GluR6 has been shown to be genomically encoded w19x. The absence of GluR6 editing in chick indicates that the aspects of development for which this switch is necessary in rat do not exist in chick or are performed by another mechanism. A similar absence of RNA editing has been demonstrated in another non-mammalian species, the fish species, Oreochromis sp. w18x. In this species, however, the lack of editing is seen for GluR2, which is edited in chick. The subunits which we have detected are concordant with the electrophysiological responses of the motoneurons to various agonists. As there are some minor differences in response between cells, it is possible that the subunits detected in the motoneuron population may or may not be present in each individual cell.
Acknowledgements We thank Dr. Stephen Heinemann at the Salk Institute for providing the glutamate receptor clones used in this study. Also, we thank Prof. J. Dudel for his support during experiments performed at the Physiology Institute, Technical University of Munich. This work was supported by NIH Grants NS13600 and GM07507 and the Deutsche Forschungsgemeinschaft ŽSFB 220..
References w1x P. Ascher, L. Nowak, Quisqualate- and kainate-activated channels in mouse central neurones in culture, J. Physiol. 399 Ž1988. 227–245. w2x P. Ascher, P. Bregestovski, L. Nowak, N-Methyl-D-aspartateactivated channels of mouse central neurones in magnesium-free solutions, J. Physiol. 399 Ž1988. 207–226. w3x S. Bahn, B. Volk, W. Wisden, Kainate receptor gene expression in the developing rat brain, J. Neurosci. 14 Ž1994. 5525–5547. w4x G. Bernardi, W. Ziegelgansberger, A. Herz, E.A. Puil, Intracellular studies on the action of L-glutamic acid on spinal neurons of the cat, Brain Res. 39 Ž1972. 523–525.
R. Temkin et al.r Molecular Brain Research 52 (1997) 38–45 w5x B. Bettler, J. Boulter, I. Hermans-Borgmeyer, A. O’Shea-Greenfield, E.S. Deneris, C. Moll, U. Borgmeyer, M. Hollmann, S.F. Heinemann, Cloning of a novel glutamate receptor subunit, GluR5: expression in the nervous system during development, Neuron 5 Ž1990. 583–595. w6x B. Bettler, J. Egebjerg, G. Sharma, G. Pecht, I. Hermans-Borgmeyer, C. Moll, C.F. Stevens, S. Heinemann, Cloning of a putative glutamate receptor: A low affinity kainate-binding subunit, Neuron 8 Ž1992. 257–265. w7x J. Boulter, M. Hollmann, A. O’Shea-Greenfield, M. Hartley, E. Deneris, C. Maron, S.F. Heinemann, Molecular cloning and functional expression of glutamate receptor subunit genes, Science 249 Ž1990. 1033–1037. w8x N. Brose, G.W. Huntley, Y. Stern-Bach, G. Sharma, J.H. Morrison, S.F. Heinemann, Differential assembly of coexpressed glutamate receptor subunits in neurons of rat cerebral cortex, J. Biol. Chem. 269 Ž1994. 16780–16784. w9x D.R. Curtis, J.C. Watkins, Acidic amino acids with strong excitatory actions on mammalian neurones, J. Physiol. 166 Ž1963. 1–14. w10x L.M. Dahm, L.T. Landmesser, The regulation of intramuscular nerve branching during normal development and following activity blockade, Dev. Biol. 130 Ž1988. 621–644. w11x J. Dudel, C. Franke, H. Hatt, Rapid activation, desensitization and resensitization of synaptic channels of crayfish after glutamate pulses, Biophys. J. 57 Ž1990. 535–545. w12x J. Egebjerg, Bettler, I. Hermans-Borgmeyer, S.F. Heinemann, Cloning of a cDNA for a glutamate receptor subunit activated by kainate but not AMPA, Nature 351 Ž1991. 745–748. w13x A. Herb, N. Burnashev, P. Werner, B. Sakmann, W. Wisden, P.H. Seeburg, The KA-2 subunit of excitatory amino acid receptors shows widespread expression in brain and forms ion channels with distantly related subunits, Neuron 8 Ž1992. 775–785. w14x M. Hollmann, S. Heinemann, Cloned glutamate receptors, Annu. Rev. Neurosci. 17 Ž1994. 31–108. w15x M. Hollmann, C. Maron, S. Heinemann, N-Glycosylation site tagging suggests a three transmembrane domain topology for the glutamate receptor GluR1, Neuron 13 Ž1994. 1331–1343. w16x K. Keinanen, W. Wisden, B. Sommer, P. Werner, A. Herb, T.A. ¨ Verdoorn, B. Sakmann, P.H. Seeburg, A family of AMPA-selective glutamate receptors, Science 249 Ž1990. 556–560. w17x M. Kohler, N. Burnashev, B. Sakmann, P.H. Seeburg, Determinants ¨ of Ca2q permeability in both TM1 and TM2 of high affinity kainate receptor channels: diversity by RNA editing, Neuron 10 Ž1993. 491–500. w18x S. Kung, Y. Wu, W. Chow, Characterization of two fish glutamate receptor cDNA molecules: absence of RNA editing at the QrR site, Mol. Brain. Res. 35 Ž1996. 119–130. w19x D.L. Lowe, K. Jahn, D.O. Smith, Glutamate receptor editing in the mammalian hippocampus and avian neurons, Mol. Brain Res. 48 Ž1997. 37–44. w20x H. Monyer, P.H. Seeburg, W. Wisden, Glutamate-operated channels: developmentally early and mature forms arise by alternative splicing, Neuron 6 Ž1991. 799–810. w21x H. Monyer, R. Sprengel, R. Schoepfer, A. Herb, M. Higuchi, H. Lomeli, N. Burnashev, B. Sakmann, P.H. Seeburg, Heteromeric NMDA receptors: molecular and functional distinction of subtypes, Science 256 Ž1992. 1217–1221.
45
w22x K. Moriyoshi, M. Masu, T. Ishii, R. Shigemoto, N. Mizuno, S. Nakanishi, Molecular cloning and characterization of the rat NMDA receptor, Nature 354 Ž1991. 31–37. w23x J. Mosbacher, R. Schoepfer, H. Monyer, N. Burnashev, P.H. Seeburg, J.P. Ruppersberg, A molecular determinant for submillisecond desensitization in glutamate receptors, Science 266 Ž1994. 1059– 1062. w24x N. Nakanishi, N.A. Shneider, R. Axel, A family of glutamate receptor genes: evidence for the formation of heteromultimeric receptors with distinct channel properties, Neuron 5 Ž1990. 569–581. w25x R.J. O’Brien, G.D. Fischbach, Characterization of excitatory amino acid receptors expressed by embryonic chick motoneurons in vitro, J. Neurosci. 6 Ž1986. 3275–3283. w26x T. Paperna, Y. Lamed, V.I. Teichberg, cDNA cloning of chick brain a-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors reveals conservation of structure, function and post-transcriptional process with mammalian receptors, Mol. Brain Res. Ž1996. 101–113. w27x R.B. Puchalski, J.C. Louis, N. Brose, S.F. Traynelis, J. Egebjerg, V. Kukekov, R.J. Wenthold, S.W. Rogers, F. Lin, T. Moran, J.H. Morrison, S.F. Heinemann, Selective RNA editing and subunit assembly of native glutamate receptors, Neuron 13 Ž1994. 131–147. w28x J.L. Rosenheimer, D.O. Smith, Age-related increase in soluble and cell surface-associated neurite-outgrowth factors from rat muscle, Brain Res. 509 Ž1990. 309–320. w29x K. Sato, H. Kiyama, M. Tohyama, The differential expression patterns of messenger RNAs encoding non-N-methyl-D-aspartate glutamate receptor subunits ŽGluR1–4. in the rat brain, Neuroscience 52 Ž1993. 515–539. w30x R.L. Schnarr, A.E. Schaffner, Separation of cell types from embryonic chicken and rat spinal cord: characterization of motoneuron enriched fractions, J. Neurosci. 1 Ž1981. 204–217. w31x D.O. Smith, C. Franke, J.L. Rosenheimer, F. Zufall, H. Hatt, Glutamate-activated channels in adult rat ventral spinal cord cells, J. Neurophysiol. 66 Ž1991. 369–378. w32x D.O. Smith, C. Franke, J.L. Rosenheimer, F. Zufall, H. Hatt, Desensitization and resensitization rates of glutamate-activated channels may regulate motoneuron excitability, J. Neurophysiol. 66 Ž1991. 1166–1175. w33x B. Sommer, K. Kainanen, T.A. Verdoorn, W. Wisden, N. Burna¨ shev, A. Herb, M. Kohler, T. Takagi, B. Sakmann, P.H. Seeburg, ¨ Flip and Flop: A cell-specific functional switch in glutamate-operated channels of the CNS, Science 249 Ž1990. 1580–1585. w34x T.R. Tolle, A. Berthele, W. Zieglgansberger, P.H. Seeburg, W. ¨ ¨ Wisden, The differential expression of 16 NMDA and non-NMDA receptor subunits in the rat spinal cord and in periaqueductal gray, J. Neurosci. 13 Ž1993. 5009–5028. w35x L.O. Trussell, G.D. Fischbach, Glutamate receptor desensitization and its role in synaptic transmission, Neuron 3 Ž1989. 209–218. w36x R.J. Wenthold, N. Yokotani, K. Doi, K. Wada, Immunochemical characterization of the non-NMDA glutamate receptor using subunit-specific antibodies, J. Biol. Chem. 267 Ž1992. 501–507. w37x C.F. Zorumski, J. Yang, AMPA, kainate, and quisqualate activate a common receptor-channel complex on embryonic chick motoneurons, J. Neurosci. 8 Ž1988. 4277–4286.