Comparison of present data with immunocytochemical, in situ hybridization and RT-PCR studies
The single cell PCR technique used in the present study allows the analysis of many different genes in parallel. Additionally, it has a high spatial resolution and thereby excludes false positive results as a result of the expression of glutamate receptor subunits in non-neuronal cells. In the present study, single neurons of the PBC, the NH and the NTS were analyzed for their NMDA receptor subunit composition. So far, no studies of the NMDA receptor subunit composition in identified single brainstem neurons have been published. Previous investigations concerning the expression of glutamate receptors in interneurons of the PBC neurons (Paarmann et al. 2000) neglected the differential expression of NMDAR1 splice variants and NR3A.
Various authors reported a strong expression of NMDAR1 in the PBC, the NH and the NTS in the brainstem of rat at the RNA level (Watanabe et al. 1994; Paarmann et al. 2000). The ratio between ‘a’ and ‘b’ splice variants of NMDAR1, however, has not yet been studied in brainstem, while in forebrain and cerebellum ratios of NMDAR1-a to NMDAR1-b of 5 : 1, and 1 : 5, respectively, were found (Anantharam et al. 1992; Sugihara et al. 1992). Laurie and Seeburg (1994) showed in an in situ hybridization study that both N-terminal splice variants are expressed in brainstem. However, the expression of these splice variants in identified brainstem nuclei or neurons was not investigated. The present study clearly demonstrates that NMDAR1-a splice variants are very prominent in all three cell types analyzed. In contrast, NMDAR1-b splice variants were only detected in the PBC and the NH, but hardly in the NTS (Table 3). Therefore, the detection of both NMDAR1-a and NMDAR1-b splice variants in brainstem neurons confirms the results of the in situ hybridization study of Laurie and Seeburg (1994).
The present study revealed that the C-terminal NMDAR1-1 splice variants are quite rare in brainstem neurons analyzed (Table 3). These results are consistent with the already-described weak expression of these splice variants in brainstem (Laurie and Seeburg 1994; Benke et al. 1995). In contrast, a strong expression of a different C-terminal splice variant, NMDAR1-2, has been previously demonstrated in brainstem (Laurie and Seeburg 1994), but not at the level of identified nuclei or neurons. So far, to study NMDAR1 expression in the PBC, the NH and the NTS, an antibody has been used which recognizes NMDAR1-1 as well as NMDAR1-2 splice variants (Petralia et al. 1994b; Ambalavanar et al. 1998). However, because NMDAR1-1 is quite rare in brainstem (Laurie and Seeburg 1994; Benke et al. 1995), the immunoreactivity detected by Petralia et al. (1994b) in rat and by Ambalavanar et al. (1998) in cat supposedly is almost exclusively because of the presence of NMDAR1-2. Ambalavanar et al. (1998) found a much weaker immunoreactivity in the NH compared with PBC and NTS. This finding fits exactly the results of the present single cell PCR study (Table 3). This study revealed that NMDAR1-2 splice variants are abundant in the PBC and NTS, but not in NH. In cat brainstem, however, with the same antibody used by Amabalavanar et al., in another immunocytochemical study no differences between the PBC, the NH and the NTS were observed (Petralia et al. 1994b), reflecting differences between organisms analyzed (cat vs. rat).
Much less prominent than NMDAR1-2 is a different C-terminal splice variant of NMDAR1, NMDAR1-3. So far, NMDAR1-3 expression has only been described in hippocampus and cortex, but not in brainstem. However, even in the hippocampus and the cortex, the other C-terminal splice variants are more prominent than NMDAR1-3 (Laurie and Seeburg 1994). This generally weak expression of NMDAR1-3 fits the very weak (PBC, NTS) or undetectable (NH) expression found in brainstem neurons in the present study (Table 3).
Conversely, NMDAR1-4 turned out to be very abundant in all three cell types analyzed in the present study (Table 3). This finding confirms the results of Laurie and Seeburg (1994) who described a strong expression of NMDAR1-4 in brainstem.
Compared with NMDAR1 splice variants, much more is known about NR2 subunit expression in PBC, NH and NTS. In mouse brainstem, expression of NR2A starts later than expression of NR2B. However, between P0 and P21, NR2B is largely replaced by NR2A (Watanabe et al. 1992). In P5 to P8 mice and in P21 mice, both NR2A and NR2B have been detected in PBC, NH and NTS, except for NR2B in PBC of P21 mice (Watanabe et al. 1994; Paarmann et al. 2000). These findings argue in favour of the immunoreactivity detected with an antibody specific for both NR2A and NR2B in all three nuclei in young rats (Petralia et al. 1994a) being caused by the presence of both subunits. Therefore, a strong expression of NR2A and NR2B demonstrated in the present study in P6 to P11 rats in all three cell types is similar to the findings of the previous studies.
In contrast to NR2A and NR2B, for NR2C slightly contradictory results have been reported so far. Watanabe et al. (1994) observed no expression of NR2C in these nuclei, whereas in a different study NR2C was detected in all three nuclei, although weaker in the PBC and NH (Paarmann et al. 2000). In the present study, which utilized P6-P11 rats, NR2C was found in PBC neurons at a similar level as described for P5 to P8 mice. However, differences between the two RT–PCR studies were apparent for the NTS. NR2C was reported to be strongly expressed in the NTS at the nuclei level (Paarmann et al. 2000), but was hardly detectable in single NTS neurons in the present study. These findings argue for a strong NR2C expression in non-neuronal cells in the NTS or for differences between P5 to P8 mice and P6 to P11 rats. In the NH, in single neurons a weak expression of NR2C was noted in this study, a finding that has been already described at the nuclei level (Paarmann et al. 2000).
The expression of NR2D in the brainstem has been previously demonstrated both at the RNA level (Watanabe et al. 1992; Ishii et al. 1993; Monyer et al. 1994), and at the protein level (Wenzel et al. 1995; Wenzel et al. 1996; Laurie et al. 1997). However, in an in situ hybridization study (Watanabe et al. 1994) detected no NR2D expression in the PBC, the NH and the NTS of P21 mice. In contrast, in a different study strong expression of NR2D was seen at the level of the nuclei in the PBC, the NH and the NTS, and even in single neurons of the PBC (Paarmann et al. 2000). In the present study NR2D turned out to be very abundant in all three analyzed cell types, a finding which is in line with this previous report. Therefore, the observed differences between the two RT–PCR studies and the in situ hybridization study might be explained by a higher sensitivity of RT–PCR, and a stronger NR2D expression as a result of a different developmental stage being analyzed (Monyer et al. 1994).
In contrast to the NR2 subunits, NR3A expression differed markedly between the cell types studied here. NR3A was found to be much more abundant in PBC neurons compared with NH and NTS neurons (Table 3). The expression of NR3A in the PBC, NH and NTS has been previously reported (Paarmann et al. 2000) and was further confirmed in the present study. As expression of NR3A is strongly dependent on the developmental stage (Sucher et al. 1995), the much stronger expression of NR3A in PBC neurons compared with NH and NTS neurons, which was not observed in a different RT–PCR study (Paarmann et al. 2000), is probably caused by differences in the developmental state of the experimental animals (6- to 11-day-old rats in this study compared with 5- to 8-day-old mice in the previous study).
Many previously analyzed neurons show relatively slow kinetics of the NMDA currents (Table 2). In comparison with the fast decay times found for hypoglossal neurons, which have also been described by other groups (O'Brien et al. 1997), and PBC neurons, decay times in the NTS were approximately 10 times slower. In some cases, the kinetics of NMDA currents could only be fitted adequately using two exponential functions, indicating a biphasic current profile.
The expression of different C-terminal NMDAR1 splice variants may affect binding of NMDAR1 to intracellular proteins. The C1-cassette, present in NMDAR1-1 and NMDAR1-3, has been reported to bind several proteins such as neurofilaments (Ehlers et al. 1998), calmodulin (Ehlers et al. 1996) and yotiao (Lin et al. 1998). Furthermore, it contains an ER retention signal (Standley et al. 2000) and a nuclear localization signal (Holmes et al. 2002). Ca2+-activated calmodulin can reduce the mean open time of NMDA receptors containing NMDAR1-1 and NMDAR1-3 (Ehlers et al. 1996). Therefore, as the C1 cassette was hardly detectable in this study, the NMDA receptors of PBC, NH and NH are probably largely resistant to this direct inhibitory effect of calmodulin and, because of the lack of the ER retention signal, can be easily transported to the cell surface. Additionally, the putative proteolytic cleavage and redirection of the C-terminal domain to the nucleus is very unlikely to occur in the NH and the NTS. In contrast to the C1-cassette, the function of the C2-cassette is not yet clear. NMDAR1-2 and NMDAR1-4 do not only differ in the presence or absence, respectively, of the C2-cassette, but also in their C-termini (Sugihara et al. 1992; Hollmann et al. 1993). Although literature about a direct interaction between the C-terminus of NMDAR1-4 and a PDZ domain of PSD95 is contradictory (Kornau et al. 1995; Bassand et al. 1999), the C-terminus is able to bind COPII (Mu et al. 2003) and is likely capable of binding to a PDZ domain protein (Standley et al. 2000). The presence of the COPII binding site in NMDAR1-4 further facilitates the transport of the NMDA receptors of the PBC, the NH, and the NTS to the cell surface. Because of the much stronger expression of NMDAR1-b splice variants in the PBC and the NH, their NMDA receptors might be regulated in a different manner compared with neurons showing slower deactivation time courses, like those in the NTS. The NMDA receptors of NH and PBC are supposedly more susceptible to protein kinase C-mediated potentiation (Durand et al. 1993), but less to potentiation by zinc (Hollmann et al. 1993), neurosteroids (Malayev et al. 1998; Ceccon et al. 2001) and polyamines (Zheng et al. 1994).
The present study clearly demonstrates that all three analyzed cell types have distinct expression patterns regarding their NMDAR1 splice variants, but not their NR2 subunits. Because NH and NTS cells differ only in the expression of one NMDAR1 splice variant, NMDAR1-4b in NH compared with NMDAR1-2a in NTS (Table 5), this difference might provide a possible molecular basis for the observed differences in the decay time of deactivation of NMDA EPSCs between NTS [+40 mV, 300 ms (Titz and Keller 1997)] and the much faster NH [+40 mV, 126 ms (O'Brien et al. 1997)]. Decay times of deactivation of NMDA receptors have been previously studied in recombinant expression systems (Monyer et al. 1992, 1994; Vicini et al. 1998; Rumbaugh et al. 2000). They have been shown to depend on both the NR2 subunits as well as on the NR1 splice variant. Co-expression of the five most prominent NMDAR1 splice variants with NR2A did not reveal any significant differences (Vicini et al. 1998). However, upon co-expression with NR2B, the presence of the NMDAR1-b splice variants leads to a faster deactivation compared with the corresponding NMDAR1-a splice variants (Rumbaugh et al. 2000). Our single cell PCR data are consistent with a simple model featuring a tetrameric NMDA receptor consisting of one NR2B, one NR2D, one NMDAR1-4a, and one additional NMDAR1 subunit depending on the specific cell type. In the NTS, this additional subunit would be NMDAR1-2a, in the NH NMDAR1-4b, and in the PBC, which like NH shows very fast decay times of deactivation (Table 2), NMDAR1-2b. However, in the PBC different NMDAR1 splice variants such as NMDAR1-4b and NMDAR1-2a cannot be excluded, because C- and N-terminal splice variants were analyzed independently. Additionally, this model does not take into account the strong NR3A expression in the PBC. Using co-immunoprecipitation of NR3A with NMDAR1 and NR2B, Das et al. (1998) showed that NR3A is in fact part of a functional NMDA receptor complex. However, so far no studies have been published which determined the ratio between NR3A and NMDAR1 and NR2 in a functional NMDA receptor complex.
Taken together, our data suggest that the fast kinetics of the NMDA EPSCs in the PBC and the NH compared with the NTS could be achieved by a much stronger expression of NMDAR1-b splice variants. Thus, the decay times of deactivation of the NMDA receptors might not, as previously believed, depend solely on the NR2 subunits, but also on the NMDAR1 splice variants.
The present study clearly showed that the frequency of breathing in young rats can exceed 5 Hz. Therefore, to fulfil a modulatory function in each cycle of breathing, NMDA receptors have to act within 200 ms. Out of all analyzed different types of neurons, only the NMDA receptors expressed in the PBC and the NH, as demonstrated in this study, are able to act within this short timescale, which might be achieved by expression of unique combinations of NMDA receptor subunits. The results presented here delineate the NMDA receptor subunits and splice variants expressed in respiratory-related and control cells; however, the assembly of the splice variants and the spatial distribution of receptors remain unknown. In addition to the differential pattern of NMDA receptor mRNA expression, other processes such as post-translational modification could potentially modulate the specific properties of NMDA receptors. Our conclusion from the data presented is that the observed differential expression of one NMDA receptor subunit might regulate fast kinetics of synaptic currents.