SEARCH

SEARCH BY CITATION

Keywords:

  • competitive RT-PCR;
  • GluR1–4;
  • internal standard;
  • rat;
  • whole-cell patch-clamp

Abstract

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

α-Amino-3-hydroxy-5-methylisoxazole-4-propionate (AMPA) receptor subunit (GluR1–4) mRNAs expressed by single neurons in rat hippocampal cultures were quantified by single-cell RT-PCR using an internal standard RNA after whole-cell patch-clamp recording. The internal standard RNA, derived from GluR2 with a single nucleotide substitution, was reverse-transcribed and PCR-amplified with the same efficiency as GluR1–4 mRNAs. The mean mRNA numbers harvested in vitro from pyramidal-like neurons on day 9 were 1150 ± 324 molecules of GluR1, 1080 ± 273 molecules of GluR2, 100 ± 20 molecules of GluR3, and 50 ± 10 molecules of GluR4 (mean ± SEM, n = 12). In a non-pyramidal neuronal population that expresses AMPA receptors characterized by high Ca2+ permeability, the numbers of GluR1, GluR3 and GluR4 mRNA molecules harvested per cell were 354 ± 64, 25 ± 17 and 168 ± 36, respectively (n = 8). The GluR2 mRNA was not detected in this cell type. The calculated ratio of AMPAR mRNA molecules per total mRNA molecules was 1/240 in pyramidal-like neurons (1/500 for GluR2), being in the range obtained with total RNA from rat forebrain and cerebellum (1/170 and 1/380, respectively). Finally, our results indicated that the proportion of GluR1–4 mRNA located in neurites reached ∼60% in pyramidal-like neurons. However, we found no evidence of preferential subcellular distribution of a given subunit.

Abbreviations used
AMPA

α-amino-3-hydroxy-5-methylisoxazole-4-propionate

AMPAR

AMPA-type glutamate receptor channels

mRNA

messenger RNA

PSL

photo-stimulated luminescence

SAGE

serial analysis of gene expression.

α-Amino-3-hydroxy-5-methylisoxazole-4-propionate (AMPA)-type glutamate receptor channels (AMPARs) mediate fast neurotransmission at excitatory synapses in the mammalian CNS. They are assembled from four subunits, GluR1–4, either alone or in various combinations. The functional properties of AMPARs are dependent on their subunit composition (Wisden and Seeburg 1993; Hollmann and Heinemann 1994). For example, increasing the proportions of GluR2 in AMPARs determines a gradual switch from receptors with high calcium permeability and inwardly rectifying current/voltage (I/V) relation to receptors with little calcium permeability and outwardly rectifying I/V relation (Hollmann and Heinemann 1994; Jonas and Burnashev 1995). To correlate the presence of mRNA of each subunit with functional properties of native AMPARs at the single cell level, the mRNAs expressed in various types of neurons and glial cells have been analyzed using a combination of whole-cell patch-clamp recording and reverse transcription-PCR amplification (single-cell RT-PCR, Lambolez et al. 1992). Although the relative proportions of GluR1–4 mRNAs expressed by single cells were determined by single-cell RT-PCR in previous studies, none of these reports investigated the absolute number of AMPAR mRNA molecules expressed by single cells.

In this study, we established a protocol that enabled quantification of AMPAR subunit mRNA molecules expressed by single cells using an internal standard RNA. This protocol was applied in rat hippocampal cultures to the quantification of AMPARs mRNA molecules expressed either by pyramidal-like neurons expressing high levels of GluR2 or by a neuronal type that had been reported to lack GluR2 (type-II neurons; Ozawa et al. 1991; Bochet et al. 1994; Tsuzuki et al. 2000). Furthermore, we analyzed developmental changes of AMPAR subunit mRNA expression in culture and correlated quantitative information obtained from single cells to that obtained by analyzing total RNA purified from cultures after cell counting.

Materials and methods

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Construction of internal standard cRNA

The GluR2flip cDNA contained in the Bluescript plasmid, a kind gift of Dr Michael Hollmann (pRB14 from Boulter et al. 1990), was subjected to site-directed mutagenesis. The nucleotide at position 2098 (position 1 being the first nucleotide of the coding sequence) was substituted from C to G using the Sculptor in vitro mutagenesis system (Amersham-Pharmacia, Buckinghamshire, UK). This substitution abolished a Bsp1286I site and introduced an StuI restriction site (see Fig. 1a). Internal standard cRNA, designated as R2S (3246 bases, with a 3′-untranslated region of 543 bases including a 17mer poly-A tail), was generated using T3 RNA polymerase, treated by DNaseI and purified. The same protocol was used to generate GluR1–4 cRNAs from their cognate cDNAs contained in pBluescript (GluR1, 3 were kind gifts by Drs Jim Boulter, Michael Hollmann and Stephen Heinemann and GluR4 by Dr Peter H. Seeburg). The final products were quantified by both UV spectrometry and agarose gel electrophoresis against yeast ribosomal RNA. The cRNA was stored at 3 × 1011 molecules/µL (0.55 µg/µL) in 70% ethanol at − 20°C and diluted with water just before use.

image

Figure 1. Quantification of mRNAs of AMPAR subunits using an internal standard cRNA. (a) Upper panel: Two different PCRs were used; amplification of GluR1–4 with AMPA up and AMPA lo primers, and that with R2 up and AMPA lo primers. Arrows indicate positions of primers and their extensions by Taq polymerase. Left panel: Modification of GluR2 cDNA by site-directed mutagenesis. Bsp1286I cuts GluR2 cDNA at position 2099. Substitution of C at position 2098 to G abolished this recognition site and created an StuI restriction site. Right panel: Subunit-specific digestion by BglI, Bsp1286I, Eco47III and EcoRI for GluR1, GluR2, GluR3 and GluR4, respectively. (b) Known amounts of GluR2flip and R2S cRNA molecules were mixed at different ratios and subjected to RT-PCR specific for GluR2 with the lower primer labeled. GEL: After digestion with Bsp1286I (B lanes) or StuI (S lanes), the fragments were separated by agarose gel electrophoresis. Lane M shows DNA size marker with the dense 600 bp band. PSL: photo-stimulated luminescence image of the same gel. The bottom graph shows the normalized amounts of GluR2 and R2S RT-PCR products obtained by PSL quantification (n = 3 for each proportion). (c) Equal amounts of GluR1, 2, 3 and 4 cRNA molecules (indicated values are for individual subunits) were mixed and subjected to RT-PCR with AMPA up and AMPA lo primers with the lower primer labeled. The graph shows the proportions obtained after subunit-specific restriction analysis and PSL quantification (n = 4 for 107, 103 and 102, and n = 6 for 10). In the 10-molecule range, overall 33% drops-outs occurred regardless of the subunit. However, at least two subunits were detected in all trials. Error bars indicate SEM.

Download figure to PowerPoint

Preparation of total RNA

Total RNA from forebrain or cerebellum of 3-month-old rats was purified using Trizol (Life Technologies, Rockville, MD, USA), treated with DNaseI and after phenol-chloroform extraction, passed through Sephadex G50 and ethanol precipitated. Total RNA was prepared from cell cultures after cell counting using Trizol in the presence of acrylamide carrier (Ethachinmate, NipponGene, Toyama, Japan) without further treatment.

Hippocampal cell culture

Two types of hippocampal cell cultures were used: serum-free neuronal culture and mixed neuronal and glial culture. Serum-free neuronal cultures were prepared from the embryonic day 18 rat hippocampus (Wistar-ST, SLC, Hamamatsu, Japan) as previously described (Brewer et al. 1993). The hippocampi were dissected and individual cells were dissociated by trituration without enzymatic treatment. After allowing non-dispersed tissues to settle for 3 min, the supernatant was transferred to a 15-mL tube and centrifuged for 40 s at 100 g. The pellet was gently resuspended in Neurobasal medium (Life Technologies) supplemented with 25 µm glutamate, 0.5 mm glutamine and B27. The cells were plated at a density of 80–160 cells/mm2 in poly-d-lysine-coated 3.5 cm culture dishes and maintained at 37°C in a humidified incubator under an atmosphere of 95% air and 5% CO2. The medium was changed to Neurobasal medium containing B27 and glutamine on day 4 and exchanged once a week thereafter. Mixed neuronal and glial hippocampal cultures were prepared as described previously (Iino et al. 1990).

Patch-clamp recording

Whole-cell currents of hippocampal cells were recorded with patch-pipettes filled with 7.5 µL of internal solution containing 140 mm CsCl, 5 mm EGTA, 10 mm HEPES (adjusted to pH 7.2 with KOH). The resistance of the patch-pipette was 3–5 MΩ. The external solution contained 145 mm NaCl, 5 mm KCl, 2.4 mm CaCl2, 10 mm glucose and 10 mm HEPES (adjusted to pH 7.4 with NaOH). To activate AMPARs, 100 µm of the non-desensitizing agonist kainate was applied through a double-barreled theta tube with tip diameter of 200–300 µm as described (Koike et al. 2000).

Single-cell RT-PCR

Single-cell RT-PCR was performed essentially as described (Lambolez et al. 1992; Bochet et al. 1994; Tsuzuki et al. 2000). Following whole-cell recording, as much as possible of the cell's content was aspirated into the patch-pipette. The contents of the pipette were expelled into a test tube and immediately a known quantity (100–20 000 molecules in 0.5 µL) of internal standard R2S cRNA was added. Then, 2.75 µL of a solution containing random hexamers (Roche Diagnostics Corporation, Indianapolis, IN, final concentration 5 µm), a degenerate primer for GluR1–4 (GC(A/G) CT(C/G) GTC TT(T/G) T, final concentration 0.5 µm) and dATP, dCTP, dGTP, dTTP (Amersham-Pharmacia, final concentration 0.5 mm each was added). The tube was heated at 94°C for 1 min and placed on ice for 1 min. After addition of MgCl2 (final concentration 3 mm), dithiothreitol (final concentration 10 mm), 10 U of ribonuclease inhibitor (Promega, Madison, WI, USA) and 50 U of Superscript II reverse-transcriptase (Life Technologies) in a final volume of 10 µL, the reaction was performed at 37°C for 1 h. The RT products were amplified with the common GluR1–4 primers (upper primer, AMPA up: CCTTTGGCCTATGAGATCTGGATGTG; lower primer, AMPA lo: TCGTACCACCATTTGTTTTTCA) as described (Lambolez et al. 1992), yielding a 750-bp fragment. Ten microliters of the first PCR products were separated by agarose gel electrophoresis, and quantified against a known amount of molecular weight marker. The 750 bp fragment was gel-purified with QIAquick Gel Extraction Kit (Qiagen, Hilden, Germany).

Quantification of GluR1–4 subunits

Subsequent PCR amplifications were performed in 50 µL reaction mixtures containing 106−108 molecules of the purified fragment with one primer being 32P 5′-end labeled using [γ-32P]ATP (Bresatec, Adelaide, Australia) and T4 polynucleotide kinase (Takara, Ohtsu, Japan). PCR for quantification of GluR2 was performed with 20 cycles using a labeled GluR2-specific nested upper primer (R2 up: GAAGATGGAAGAGAAACACAAAGT) and the common lower primer. Another PCR for determination of GluR1–4 ratios was performed with 20 cycles with common upper and lower primers, the upper primer being labeled.

Aliquots of 2 µL of the final products were then digested with subunit-specific restriction enzymes as described (see Fig. 1a and Lambolez et al. 1992), the R2S internal standard products being cut by StuI, and separated by agarose gel electrophoresis. After electrophoresis, the gel was fixed in 7% trichloroacetic acid (TCA) for 1 h at 4°C and dried under vacuum. The radioactivity of each band was measured as photo-stimulated luminescence (PSL) with an image analyzer (BAS2000-MacBAS, Fuji Photo Film, Tokyo, Japan). The amounts of 32P in both cut and uncut fragments were measured in each lane and the relative ratio of each subunit to internal standard was determined.

Results are expressed as mean ± SEM.

Results

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Quantification of AMPAR subunits mRNAs using internal standard cRNA

GluR2 is the key subunit that determines channel properties of AMPARs. To prepare an internal standard for quantification of the amount of GluR2 mRNA, a mutant of GluR2 cDNA was produced by site-directed mutagenesis that exchanged C at position 2098 to G. This substitution abolished the Bsp1286I restriction site specific for GluR2, and created an StuI site (Fig. 1a). cRNA was transcribed in vitro from this mutant cDNA, and designated as R2S cRNA. The transcript was 3.2 kb in length, and comprised the 3′-untranslated region and poly A tail (543 bases). Samples containing either GluR2 or R2S cRNAs alone, or the following GluR2 : R2S ratios, 10 : 1, 10 : 3, 10 : 10, 3 : 10 and 1 : 10, were subjected to RT-PCR using R2 up and AMPA lo primers. The results of restriction analysis of the PCR products are shown in Fig. 1(b), GEL. The product derived from GluR2 cRNA alone was completely digested with Bsp1286I, and that from R2S cRNA alone was completely digested with StuI. The products from mixtures of GluR2 and R2S cRNAs were partially digested with both enzymes. The amounts of both fragments were quantified as the intensity of photo-stimulated luminescence (PSL) determined with an image analyzer (Fig. 1b, PSL). In five different ratios of GluR2 cRNA to R2S cRNA, the relative abundances of GluR2 cRNA were 90.7 ± 0.3%, 77.0 ± 1.5%, 48.8 ± 1.5%, 22.2 ± 1.1%, 9.6 ± 0.7% (mean ± SEM, n = 3), respectively, close to the predicted ratios of 90.9%, 76.9%, 50%, 23.1% and 9.1%. These results indicated that the GluR2-specific competitive RT-PCR maintained original proportions of the GluR2 and R2S cRNA mixture in this 0.1–10-fold range.

image

Figure 2. Quantification of GluR1–4 mRNAs in forebrain and cerebellar total RNA. (a) The internal standard R2S cRNA was added at various amounts to forebrain or cerebellar total RNA (two different RNA preparations were used for each brain region) prior to GluR2-specific RT-PCR with the upper primer labeled. PSL images are shown. The proportion of Bsp1286I-digested fragment (B lanes) decreased with increases in R2S cRNA input (S lanes: StuI digest, Left B and S lanes: no R2S cRNA added). (b) R2S/GluR2 ratios obtained by PSL were plotted against the number of R2S internal standard cRNA molecules added (circles: cerebellum, squares: forebrain). Each point represents the mean ± SEM (n = 5). The two lines through data plots were drawn according to the best least-squares fit. (c) Total RNA from forebrain or cerebellum was subjected to RT-PCR with AMPA up and AMPA lo primers with the lower primer labeled. PSL image of subunit-specific restriction analysis of PCR products is shown. (d) The graph shows the GluR1–4 proportions obtained after PSL quantification (n = 11 for both forebrain and cerebellum).

Download figure to PowerPoint

We next examined whether the original proportions of GluR1–4 cRNAs were maintained throughout RT-PCR using simultaneous amplification with AMPA up and AMPA lo primers (Fig. 1a and Lambolez et al. 1992). Equal amounts (108 molecules each) of GluR1–4 cRNAs were mixed and reverse-transcribed, and the RT product was serially diluted and subjected to PCR. PCR products were gel-purified and subjected to a second PCR using the same primers, the lower primer being labeled. The labeled products were digested with subunit-specific restriction enzymes (Fig. 1a) and quantified. As shown in Fig. 1(c), the amplified products contained approximately the same amounts of GluR1–4 even when cDNA was diluted to 10 copies for each subunit (expressed as original cRNA copy number). Upon reduction to 10 molecules, one or two subunits were occasionally undetectable (not shown). However, these drop-outs occurred randomly among GluR1–4 (overall 33% drops-outs regardless of the subunit). These results indicated that the competitiveness is retained even at a very small number of cDNA copies. These results are consistent with those reported previously (Lambolez et al. 1992; Lambolez et al. 1996).

Quantification of AMPAR subunits mRNAs in brain extracts

Using R2S cRNA as an internal standard, we estimated the number of GluR2 mRNA molecules in rat forebrain and cerebellum total RNA. With increases in amount of R2S cRNA added to RNA from either forebrain or cerebellum, the proportion of the fragment derived from R2S gradually increased (Fig. 2a). In Fig. 2(b), the ratios of R2S cRNA relative to GluR2 mRNA (R2S/GluR2) are plotted against the number of R2S cRNA molecules added to each sample. There was a roughly linear relationship between R2S/GluR2 ratio and the number of R2S cRNA molecules in both forebrain and cerebellum samples. From the slopes of the linear regression lines, the amounts of GluR2 mRNAs present in our forebrain and cerebellum RNA samples were estimated to be 9.95 ± 0.18 × 107 molecules/µg forebrain total RNA and 1.61 ± 0.13 × 107 molecules/µg cerebellum total RNA.

Next, the relative amounts of GluR1–4 in rat forebrain and cerebellar total RNA were examined (Fig. 2c,d). GluR1 and GluR2 were the two major components in the AMPAR subunits expressed in the forebrain. On the other hand, GluR1 and GluR4 were the major components found in the cerebellum. The relative abundances of GluR1, GluR2, GluR3 and GluR4 were 33.5 ± 1.4%, 37.9 ± 1.3%, 24.2 ± 1.1% and 6.5 ± 1.3%, respectively, in forebrain RNA (n = 11), and 50.2 ± 2.4%, 13.3 ± 0.6%, 8.4 ± 0.7%, 28.7 ± 1.1% in cerebellar RNA (n = 11). The absolute numbers of GluR1, GluR3 and GluR4 mRNAs were deduced from the above determination of GluR2 mRNA numbers. They were 8.8 × 107 (GluR1), 6.3 × 107 (GluR3) and 1.7 × 107 (GluR4) molecules per µg forebrain RNA, and 6.1 × 107 (GluR1), 1.0 × 107 (GluR3) and 3.5 × 107 (GluR4) molecules per µg cerebellar RNA.

mRNA is generally estimated to represent 3–5% of total RNA. Assuming that it represents 5% of total RNA and that their mean size is 2 kb, the number of mRNA molecules would be 4.6 × 1010 per µg. Therefore, GluR1–4 mRNA would be present at one copy per 170 mRNA molecules in our sample of forebrain RNA and one per 380 in cerebellum.

Quantification of GluR1–4 mRNAs in single hippocampal neurons

We next adapted the quantification of GluR1–4 mRNAs to single-cell mRNAs harvested with patch pipettes. In glia-free cultures (see Materials and methods), the overwhelming majority of cells displayed triangular-shaped somata bearing morphological resemblance to pyramidal neurons. The AMPARs in these cells displayed a linear or outwardly rectifying I–V relation (not shown) indicative of a high GluR2/GluR1–4 ratio.

Typical results of quantification of GluR1–4 mRNAs expressed by a single cultured neuron on day 9 in vitro are shown in Fig. 3(a). After whole-cell patch-clamp recordings, the cell contents were aspirated into the patch-pipette and expelled into a reaction tube. R2S internal standard cRNA (400 molecules) was then added and the mixture was subjected to RT-PCR with AMPA up and AMPA lo primers. The amplified product was gel-purified and second PCR was performed with 32P-labeled upper primers. Quantification of GluR2 mRNA was performed using GluR2-specific PCR (Fig. 3a, left panel). The amplified product was partially digested with either Bsp1286I (B lane) or StuI (S lane). Digestion with both enzymes left no undigested fragment (B + S lane), showing that the amplification was specific for GluR2 and R2S. The amount of Bsp1286I-digested fragment was 2.8-fold higher than that of StuI-digested fragment, indicating that the initial number of GluR2 mRNA molecules was 2.8-fold higher than that of R2S cRNA. Since 400 molecules of R2S cRNA were added as an internal standard, the number of GluR2 mRNAs harvested from the cell was estimated to be 1110.

image

Figure 3. Quantification of GluR1–4 in single hippocampal neurons in culture. (a) PSL image of restriction analysis of the RT-PCR product obtained from a single neuron after addition of 400 molecules of R2S internal standard cRNA. Left: after GluR2-specific amplification with labeled upper primer, the PCR product was digested with either Bsp1286I (B) or StuI (S). The product was digested to completion by treatment with both enzymes (B + S). NC: no cut, indicates no treatment with these enzymes. The initial amount of GluR2 mRNA was calculated to be 1100 molecules (see text). Right: after GluR1–4 amplification with labeled upper primer, the PCR product was subjected to restriction analysis with BglI (lane 1), Bsp1286I (lane 2), Eco47III (lane 3), EcoRI (lane 4) and StuI (lane 5), which selectively digest GluR1, GluR2, GluR3, GluR4 and R2S, respectively. All: restriction by the five enzymes. NC: no cut. (b) PSL images of RT-PCR and restriction analyses obtained from 3 pyramidal-like neurons from glia-free cultures (left panels) and 2 type-II neurons (right panels). The amounts of R2S internal standard cRNA input are indicated. Note in left panels that the estimate of the number of AMPAR mRNA was little affected by changing the number of R2S cRNA molecules from 400 to 2000 in this neuronal type. These 3 neurons from glia-free cultures predominantly expressed GluR1 and 2. For type-II neurons, the GluR2-specific PCR product was completely digested by StuI (right panels, B and S lanes) with either 400 or 100 R2S cRNA input molecules, indicating that less than 10 GluR2 mRNA molecules were harvested per cell. Type II neurons expressed GluR1 and 4. Lane 5 shows R2S specific digest by StuI. (c) Mean numbers of AMPAR subunit molecules harvested from single neurons (sum is total GluR1–4) in glial-free cultures at day 2 (n = 4, open bars), day 3 (n = 5, stippled bars), day 4–5 (n = 5, double hatched bars) and after day 9 (n = 12, black bars) and single type-II neurons (n = 8, hatched bars).

Download figure to PowerPoint

Another aliquot of the first PCR product was also subjected to second PCR with AMPA up and AMPA lo primers to determine GluR1–4 proportions. The amplified fragment was completely cut by simultaneous treatment with GluR1, 2, 3, 4 and R2S cDNA-specific enzymes, and proportions of GluR1, 2, 3, 4 and R2S fragments were determined (Fig. 3a, right, lanes 1–5). The ratios of GluR1, 2, 3 and 4 fragments relative to R2S fragment were 2.1, 2.6, 0.7 and 0.3, respectively. Since the number of R2S cRNA molecules was 400, the numbers of GluR1, GluR2, GluR3 and GluR4 mRNAs harvested from the cell were estimated to be 820, 1020, 250 and 130, respectively. The number of GluR2 mRNA molecules estimated here was close to the number obtained from the R2-specific PCR (1110, see above). Indeed, the ratio of the number of GluR2 mRNA molecules estimated using the R2 up and AMPA lo primer pair relative to that estimated using AMPA up and AMPA lo primer pair was 101 ± 8% (n = 15). Thus, the number of GluR2 mRNA molecules could be determined by either method.

As R2S/GluR2 ratios of 0.1–10 were found to be suitable for accurate quantification using PSL, the amount of internal standard should be determined empirically for each neuronal type. We initially added 50 molecules of R2S to harvested pyramidal-like neurons from day 9–16 cultures but failed to detect the fragment from the internal standard R2S cRNA. We next added 20 000 molecules and failed to detect the fragment from GluR2 mRNA. Finally, we found that addition of 400 molecules of R2S was the most suitable for quantification of GluR1–4 mRNAs in this cell type. The results obtained by adding 0, 400 or 2000 molecules of R2S to three different neurons are shown in Fig. 3(b), left panels. In cell 76K1 (upper left), the GluR2-specific PCR yielded a product totally digested by Bsp1286I (lane B) as expected from the absence of R2S cRNA input, and the GluR1–4 PCR showed expression of GluR1 and 2 in this cell (lanes 1,2). Cell 76R5 (middle left, 400 R2S copies), showed predominance of GluR2 over R2S (lanes B, S) and expressed GluR1 and 2. Quantification indicated that 1300 copies of GluR2 and 900 copies of GluR1 mRNA had been harvested from this cell. Material harvested from cell 76R2 (bottom left), which received input of 2000 R2S copies, contained 1400 copies of GluR2 and 470 copies of GluR1 mRNA. Quantitative results obtained in 12 pyramidal-like neurons analyzed after 9–16 days in culture are summarized in Fig. 3(c) (black bars), and showed that the major constituents of AMPARs were GluR1 with values ranging from 185 to 3433 (1150 ± 324 copies, 46% of GluR1–4) and GluR2 with values ranging from 221 to 3465 (1080 ± 273 copies, 43%) in this cell type. The high proportion of GluR2 in GluR1–4 mRNA was consistent with the linear or outwardly rectifying I/V curves of AMPARs expressed by these neurons.

In mixed neuronal and glial hippocampal cultures, a group of non-pyramidal neurons, designated as type-II neurons (Iino et al. 1990; Ozawa et al. 1991), expressed AMPARs characterized by strong inward rectification and high Ca2+ permeability. Consistent with this observation, Bochet et al. (1994) reported that type-II neurons expressed only GluR1 and GluR4, while GluR2 was not detected. However, since the lower limit for detection of AMPAR mRNA was not defined in previous reports, the presence of a small amount of GluR2 mRNA in type-II neurons could not be ruled out. This issue was addressed using the present method.

Hippocampal neurons in mixed neuronal and glial cultures (see Materials and methods) were first characterized electrophysiologically by the I/V relation of their responses to kainate application (see Materials and methods), and after harvesting, the cell's content was analyzed by single-cell RT-PCR with known amounts of internal standard. In eight type-I non-pyramidal neurons with linear or outwardly rectifying I/V relations, analyzed as controls with 400 molecules of R2S, the mean numbers of GluR1, 2, 3 and 4 mRNA molecules per cell were 433 ± 183, 1017 ± 336, 186 ± 81 and 236 ± 67, respectively. The mean total number of GluR1–4 mRNA molecules per cell was 1854 ± 594. The relative abundance of GluR2 to GluR1–4 was 55% (results not shown).

A minority of neurons exhibited kainate responses with strong inward rectification typical of type-II neurons. Quantification of GluR1–4 mRNA in type-II neurons was initially performed using 400 molecules of R2S. Under these conditions, the product of GluR2-specific amplification was completely digested by StuI (Fig. 3b, upper right, lanes B, S). Even when R2S cRNA input was further reduced to 100 molecules, no amplified products derived from GluR2 were detected in any of the eight type-II neurons tested (see example in Fig. 3b, lower right, lanes B, S). As shown in Fig. 3(b), right panels, we detected only GluR1 and GluR4 mRNAs in type-II neurons (except for two cells where the additional presence of GluR3 was detected at 133 and 48 copies per 789 and 494 copies of GluR1–4, respectively). The mean numbers of GluR1 and GluR4 mRNA molecules found in eight type-II neurons were 354 ± 64 with values ranging from 185 to 3433 and 168 ± 36 with values ranging from 39 to 298 (in one cell GluR4 was not detected), respectively (Fig. 3c, hatched bars). The mean number of GluR3 mRNA molecules per type-II neuron derived from the present study was 25 ± 17. If one assumes that type-II neurons are homogeneous in terms of expression of the GluR3 mRNA, our results indicate that below 25 copies per cell, the detection of a given transcript becomes stochastic and reaches the lower limit of the present quantification method (see also Fig. 1c). Therefore, under a similar assumption, the mean number of GluR2 mRNA would be below 25 copies per type-II cell (< 5% of the total AMPAR subunit mRNA, see also discussion).

Developmental changes of AMPAR subunits mRNAs expression in glia-free cultures

Between days 2 and 9, in vitro in glia-free cultures, the size of neurons increased with a marked extension of neurites (Fig. 4a). The membrane capacitance increased from 5.8 ± 0.7 pF (n = 4) on day 2 to 34.8 ± 2.2 pF (n = 12) on day 9 in vitro(Fig. 4b). AMPAR-mediated currents were not detected on day 2 in vitro. Although current responses to 100 µm kainate were scarcely detected in cells on day 3 in vitro, clear responses were seen in all neurons on day 4–5, and their mean amplitude reached 883 ± 114 pA at −60 mV on day 9 in vitro(Fig. 4b). Nevertheless, substantial amounts of AMPAR mRNAs could already be harvested on day 2 and reached maximal levels on day 4–5 in vitro(Fig. 4b). Thus, there was a lag between the expression of AMPAR mRNAs and functional receptors. The results of quantitative analyses of GluR1–4 mRNA harvested from single neurons on day 2 (n = 4, open bars), day 3 (n = 5, stippled bars) and day 4–5 (n = 5, double hatched bars), performed as described above, are shown in Fig. 3(c). The major subunits were GluR1 (46, 37 and 38% of GluR1–4 at day 2, 3 and 4–5, respectively) and GluR2 (25, 22 and 41% of GluR1–4 at day 2, 3 and 4–5, respectively). Both subunits reached their day 9 levels and ratios (see above and black bars in Fig. 3c) at day 4–5, whereas GluR3 and 4 expression decreased between day 4–5 and day 9. The sum of GluR1–4 mRNA harvested reached maximal level at day 4–5 (Fig. 3c).

image

Figure 4. Developmental changes of AMPAR subunit expression in glia-free cultures. (a) Reverse phase-contrast view of glia-free hippocampal cultures on days 3, 5 and 9 in vitro. Note prominent extension of neurites at day 9. (b) The amount of GluR1–4 mRNA molecules harvested from single cells, the membrane capacitance (Cm) and the amplitude of current responses to 100 µm kainate were normalized to day 9 values and plotted against days in vitro. (c) GluR1–4 were quantified in RNA purified from glia-free culture dishes after cell counting. The number of cells, the amount of total RNA/cell, the amount of GluR2/cell and the amount of GluR1–4/cell were normalized to day 2-values and plotted against days in vitro. (d) The number of GluR2 molecules per cell harvested from single cells (open bars, patch) and that estimated from total RNA in culture dishes (black bars, dish) were plotted against days in vitro. Note the match between the two values until day 4–5 and their separation after day 9.

Download figure to PowerPoint

AMPAR mRNA quantification results obtained from single cells were next compared with GluR1–4 quantification on RNA purified from glia-free cultures after cell counting (see Materials and methods), performed on 0.1 µg total RNA with 107 copies of R2S internal standard. Although we observed a small decrease in cell number over time, the average amounts of RNA per cell, estimated as the total RNA amount divided by cell number, were 8.3 (100%), 10.2 (123%), 23.2 (280%) and 46.1 pg (555%) per cell on days 2, 3, 4–5 and 9, respectively (Fig. 4c). A steeper increase was observed in the normalized number of GluR2 and GluR1–4 mRNA copies per cell over time. We found 241 (100%), 460 (191%), 1053 (437%) and 3000 copies (1245%) of GluR2 mRNA per cell and 752 (100%), 1288 (171%), 3190 (424%) and 9367 copies (1246%) of GluR1–4 mRNA per cell on day 2, 3, 4–5 and 9, respectively (Fig. 4c). Assuming that mRNA represents 5% of total RNA and that their mean size is 2 kb, the number of mRNA molecules would be 2.3 × 106 per cell at day 9, and GluR1–4 mRNA would be present at one copy per 240 mRNA molecules in our glia-free cultures.

Although GluR2 numbers per cell found here were consistent with numbers found in harvested material from single cells on day 2, 3 and 4–5 (respectively 351, 518 and 1087, see Fig. 4d), we found a discrepancy between these numbers on day 9 (see Fig. 4d, patch and dish bars). This discrepancy was attributed to neurite extension, which is most prominent after day 4–5 (Fig. 4a). Indeed, single cell analysis proceeds mostly with material harvested from somatic patch-clamp. This would explain why both GluR2 and GluR1–4 mRNA copies harvested on single cells reached their maximal numbers on day 4–5 (Fig. 4b), whereas both AMPAR currents and GluR1–4 copies per cell quantified from whole culture extract increased sharply between day 4–5 and day 9 (see Figs 4b and c, respectively). Although this is an indication that a substantial proportion of AMPAR mRNAs are located in neurites, we found no evidence of preferential localization of a given subunit. Indeed, the GluR2/GluR1–4 ratios obtained from single cells, roughly constant between day 4–5 and 9 (41 and 43%, respectively), were also constant in RNA purified from whole cultures (33% and 34%, respectively).

Discussion

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Quantitative RT-PCR using an internal standard cRNA

Quantification of mRNA by standard RT-PCR is hampered by the fact that both RT and PCR efficiencies show wide sample-to-sample variations depending on many factors including the initial copy number of target RNAs. These problems may be overcome by competitive RT-PCR using an internal standard RNA (Becker-Andre and Hahlbrock 1989; Kagami et al. 1996; Lambolez and Rossier 2000), provided that the target RNA and internal standard RNA are reverse-transcribed and amplified with the same efficiency, and that the relative proportions of the target and internal standard RT-PCR products can be measured accurately.

In this study, GluR1–4 cRNAs, GluR1–4 natural mRNAs and R2S cRNA were reverse-transcribed and PCR-amplified with the same efficiency. Indeed, initial proportions among GluR1–4 and R2S cRNAs were maintained throughout both GluR2-specific and GluR1–4 RT-PCR procedures. Similarly, increasing the input amount of R2S internal standard to forebrain or cerebellum total RNA resulted in a linear increase of its proportion in the RT-PCR products. These results validated the subsequent quantification procedure using restriction enzymes, agarose gel electrophoresis and PSL analysis for the determination of the number of GluR1–4 mRNA molecules.

GluR1–4 proportions in brain extracts and single hippocampal neurons

The relative abundances of GluR1–4 subunits mRNA in the forebrain (GluR2 > GluR1 > GluR3 > GluR4) and cerebellum (GluR1 > GluR4 > GluR2 > GluR3) were consistent with those reported previously (see review by (Wisden and Seeburg 1993). In single pyramidal-like neurons from glia-free cultures, we found proportions of GluR1–4 subunit mRNA (GluR1 ≃ GluR2 > GluR3 ≃ GluR4) consistent with both previous reports on GluR1–4 expression in pyramidal neurons (Jonas et al. 1994; Geiger et al. 1995; Lambolez et al. 1996) and I–V relation of AMPAR-mediated currents. The high GluR2 proportion found in type-I non-pyramidal neurons from mixed neuronal and glial cultures were also consistent with their I–V relations. As previously reported (Bochet et al. 1994; Tsuzuki et al. 2000), type-II neurons expressed GluR1 and GluR4 with GluR2 expression below the limit of detection.

Determination of the number of GluR1–4 mRNA molecules in brain or culture extracts and in single neurons

The calculated ratios of AMPAR mRNA molecules per total mRNA molecules were 1/380 in the cerebellum, 1/170 in the forebrain and 1/240 in pyramidal-like neurons from glia-free cultures. These values, calculated under the assumption that mRNA represents 5% of total RNA and that their mean size is 2 kb, all fell within the same range. They were obtained using three different RNA collection procedures and scales and three different amounts of R2S cRNA internal standard added: 108 R2S copies for RNA from cerebellum or forebrain (10–100 µg scale), 107 R2S copies for RNA from culture dishes (1–10 µg scale) and 102−103 R2S copies for single pyramidal-like neurons (10–100 pg scale). These results taken together support the accuracy of the present estimates of GluR1–4 mRNA copy numbers per total mRNA molecules and per cell. Indeed, although adsorption of mRNA onto the patch-pipette may result in underestimation of individual cell's contents, the absolute numbers of GluR2 molecules harvested per cell were consistent with those obtained with RNA purified from culture dishes after cell counting (Fig. 4b, days 2, 3 and 4–5).

According to our calculations, the GluR2 mRNA was present at roughly 1/500 mRNA molecules in pyramidal-like neurons. Serial analysis of gene expression (SAGE) studies (Velculescu et al. 1995), a method for global analyses of mRNA species expression, have enabled classes of mRNA abundance to be defined. In pyramidal-like neurons, GluR2 (1/500 mRNA molecules) falls into the medium to high abundance category of mRNAs, as defined by SAGE.

In this study, the absolute numbers of GluR2 mRNA and of GluR1–4 mRNA molecules expressed by single cultured pyramidal cell-like neurons on day 9 in vitro were estimated to be ∼1100 and ∼2400 molecules, respectively. The numbers of GluR1–4 mRNA molecules expressed by single type-I and type-II cultured hippocampal non-pyramidal cells were ∼1800 and ∼600, respectively. The number of β-actin mRNA copies expressed by single normal rat kidney cells was estimated to be ∼500 by fluorescent in situ hybridization (Femino et al. 1998). Quantitative RNase protection studies showed the presence of 1800 copies of α7 and 900 copies of α3 ACh receptor subunit mRNAs per neuron in chick ciliary ganglions at embryonic day 18 (Corriveau and Berg 1993). Thus, the absolute numbers of GluR1–4 mRNA molecules in single neurons obtained in this study were in a range similar to those of other mRNAs in other cell types estimated by different methods.

Quantification of low copy number mRNA from single cells

Although the present RT-PCR protocol maintained the original GluR1–4 cRNA proportions, random drop-outs occurred at an overall frequency of 33% when cDNA was diluted to 10 copies (expressed as original cRNA copy number, see Fig. 1c). This corresponds to the mean presence of one cDNA molecule for each subunit, as predicted by a Poisson distribution. Moreover, the RT efficiency, which is therefore 10% in these conditions (108 molecules of cRNA for each subunit), is known to decrease in samples containing low total RNA copy number (Gerard and D'Alessio 1993). It is thus expected that in single cells, stochastic variability combined with actual cell-to-cell variability will affect the precision of low copy number mRNAs quantification. In the present study, the GluR3 mRNA was detected in two out of eight type-II neurons and its quantification yielded a mean number of 25 molecules per type-II cell. In a total of 45 type-II cells analyzed so far (present study; Bochet et al. 1994; Tsuzuki et al. 2000; and unpublished results), four cells showed detectable levels of GluR3, whereas GluR2 was never detected, even when radioactive probing of Southern blots was used. These results indicate that the mean GluR2 mRNA number per type-II cell was below that estimated for GluR3 (25 copies < 5% of the total AMPAR subunit mRNA).

AMPAR expression and subcellular distribution during development in vitro

Two discrepancies were observed between the amounts of AMPAR subunit mRNAs harvested from single cells and the amplitude of AMPAR-mediated current in single hippocampal neurons during development.

First, almost no AMPAR-mediated current was detected earlier than day 3 in vitro, whereas substantial amounts of GluR1–4 mRNAs were harvested. A plausible explanation is that embryonic day 18 neurons used for plating do not express functional AMPARs (Hsia et al. 1998) and at least 2 days may be required for de novo expression and translocation of functional AMPARs after their mRNA expression.

A second discrepancy was that the amplitude of AMPAR-mediated current increased between day 4–5 and 9 in vitro, whereas the amounts of GluR1–4 mRNAs harvested from single neurons reached maximum level at day 4–5. In contrast, the current increase almost perfectly matched the GluR1–4 mRNA increase as measured from total RNA extracted from culture dishes (Figs 4b and c). Since single-cell analysis was performed mostly with material harvested by somatic patch-clamp, this is an indication that a substantial proportion of AMPAR mRNAs are located in neurites, as reported previously (Paquet et al. 1997). However, we found no evidence of preferential subcellular localization of a given subunit. This was consistent with the strong correlation between GluR2/GluR1–4 proportions and native AMPAR channel properties observed in numerous cell types tested by single-cell RT-PCR (Bochet et al. 1994; Jonas et al. 1994; Geiger et al. 1995; Angulo et al. 1997; Tsuzuki et al. 2000). This correlation would have been hampered by preferential distribution of a given subunit mRNA.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

We thank Dr Masae Iino for her excellent technical assistance with patch-clamp recordings from type-I and type-II non-pyramidal neurons in rat hippocampal cultures. We also thank Drs Jim Boulter, Stephen Heinemann and Michael Hollmann for their kind gifts of plasmids containing GluR1, GluR2 and GluR3, and Dr Peter H. Seeburg for his kind gift of the plasmid containing GluR-D (GluR4) cDNA. This work was supported by CREST (Core Research for Evolutional Science and Technology) of Japan Science and Technology Corporation (JST) and the Ministry of Education, Science, Sports and Culture of Japan.

References

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References