The principal findings of the present study are that, as judged by cGMP immunohistochemistry, the targets of NO in the hippocampus are much more widespread than previous evidence of this type had suggested and, in particular, that the principal neurones participating directly in NO-dependent LTP are NO-responders, rendering unnecessary more complicated hypotheses invoking the participation of other cells. Further, results aimed at determining the immunolocation of NO receptor protein raise doubts about the veracity of current antibodies used for this purpose, although our data obtained using cGMP immunohistochemistry are generally in accordance with predictions based on the distribution of NO receptor subunit mRNA, as studied by in situ hybridization.
No previous investigation of the location of cGMP by immunohistochemistry in hippocampal slices has succeeded in detecting the nucleotide in pyramidal neurones after exposure to NO in combination with various PDE inhibitors 
. We could broadly replicate the results of these past studies () but it transpired that the conditions used were substantially submaximal for cGMP accumulation (), suggesting that the differences in pyramidal cell staining are simply explicable on the basis of the size of the cGMP signal relative to the detection threshold of the immunohistochemical method.
Previously, the highest level of cGMP accumulation reported in hippocampal slices in response to NO in the presence of PDE inhibitors (typically IBMX) was around 60 pmol/mg protein 
with values in the range 20–50 pmol/mg protein being common in studies where cGMP immunohistochemistry was conducted in parallel 
. Assuming a protein concentration of 100 mg/ml in brain (taking total protein to be 10% of weight and a tissue density of 1 g/ml), 20–30 pmol/mg protein translates into an overall cGMP concentration of 2–3 µM. Given that the cGMP antibody used for immunofluorescence has a detection limit of about 10 µM 
, it is unsurprising that the extent of visible immunohistochemical labelling was limited. The predominance of staining in astrocytes observed in previous studies 
and in may reflect a low PDE/high guanylyl cyclase activity of these cells, reminiscent of astrocytes in the cerebellum where, even in the absence of a PDE inhibitor, cGMP can reach near-millimolar concentrations 
. From measurements in the hippocampal stratum radiatum 
, astrocytes take up only 4–8% of the volume, so these cells having cGMP concentrations above the threshold for detection would still be compatible with an overall cGMP concentration of 2–3 µM.
In marked contrast to these previous findings, the cGMP level we observed under optimal conditions was around 2 orders of magnitude higher, at around 1700 pmol/mg protein (). This value appears to be the highest recorded for brain, surpassing even maximal NO-evoked cGMP responses in the cerebellum, which are around 1000 pmol/mg protein 
and comparable to the response amplitude observed in rat platelets, a pure population of NO-responder cells, in the presence of a PDE inhibitor 
. With an overall corresponding tissue cGMP concentration of around 170 µM, it is predictable that the immunofluorescent signal would be more intense and widespread ().
Two methodological factors are likely to have favoured the accumulation of higher levels of cGMP in the present study. First, we used immature rat hippocampal slices, which have been shown to generate around 5-fold more NO-dependent cGMP in response to NMDA than those of the adult 
. With few exceptions 
, previous tests of the location of NO-evoked cGMP in the hippocampus have been made using slices prepared from adult rats 
and/or mice 
. The second factor is the particular combination of compounds used to promote cGMP accumulation: BAY 60-7550 to inhibit PDE-2, the major cGMP-degrading enzyme in hippocampus 
, BAY 41-2772 to increase the maximal NO-evoked guanylyl cyclase activity 
, and DEA/NO to deliver authentic NO. This combination had not been tested previously.
Even without supplementary NO, hippocampal cGMP was comparatively high (580 pmol/mg protein; ), reflecting endogenous NO acting under the potentiating influence of BAY 41-2272. From experiments on purified NO-activated guanylyl cyclase 
, 10 µM BAY 41-2272 shifts the NO concentration-response curve to the left by a factor of at least 20, enabling marked stimulation of guanylyl cyclase by the picomolar NO concentrations purported to exist in unstimulated brain slices 
. Much of this resting NO concentration appears to be derived from eNOS, which is capable of generating NO tonically over periods of hours as a result of phosphorylation of the enzyme by the kinase Akt 
. Immunohistochemically, the difference in cGMP labelling in response to endogenous and exogenous NO was mainly one of degree (i.e. staining more intense and/or encompassing more of the pyramidal cell layer; compare and ), suggesting that exogenous NO exposes targets that would be accessed by higher levels of endogenously-generated NO, which potentially rise to the low nanomolar range on nNOS stimulation 
. It is unclear why pyramidal cell somata (and granule cell somata in the dentate gyrus) were the most resistant to the accumulation of cGMP into the detectable range. High PDE activity and/or low NO-activated guanylyl cyclase activity and/or the signalling pathway being concentrated at sites distant from the cell bodies are possible reasons.
The widespread distribution of cGMP immunoreactivity in the hippocampus in response to NO implies that most of the constituent cellular elements are potential NO targets, including pyramidal neurones, interneurones, astrocytes and at least some dentate granule cells. Some cells, notably CNPase-positive oligodendrocytes () and some presumed interneurones (), remained cGMP-immunonegative but their failure to accumulate cGMP to detectable levels does not necessarily exclude them as targets of NO, although some interneurones genuinely appear to lack NO receptor subunits 
: for example, their dominant PDE may not be PDE-2.
Other approaches to the identification of NO targets in the hippocampus are in situ
hybridization and immunocytochemistry for the NO receptive guanylyl cyclase subunits. In the immature rat hippocampus 
, like in the adult 
, mRNA for the NO receptor is widespread. The common β1 subunit and the α2 subunit appear to be strongly expressed in the pyramidal cell layer and in the granule cells of the dentate gyrus 
. The results of Pifarre et al.
also indicate expression of α2 subunit mRNA in scattered cells, presumably interneurones and/or glial cells (their ) whereas others report that this subunit is restricted to pyramidal neurones 
. The α1-subunit mRNA shows a more diffuse distribution, reportedly being exclusively expressed in subpopulations of interneurones in one study 
but also detected in pyramidal cells in another 
. Irrespective of these discrepancies in detail, our results are consistent with the broad expression of NO-activated guanylyl cyclase indicated by in situ
There have been several descriptions of the distribution of NO-activated guanylyl cyclase subunits in the hippocampus using immunohistochemistry. Our results with two different antibodies against the common β1-subunit are in general agreement with each other ( and S3
) and with previous findings using this method 
. The β1-protein distribution is also compatible with the mRNA distribution (see above) and with our cGMP immunofluorescence results (), signifying that the protein detected by immunohistochemistry is genuine. However, this conclusion is weakened by the fact that a similar distribution obtained using an antibody against the α1-subunit proved non-specific (). This antibody has been the only one used to map the location of α1-protein beforehand and has led to the conclusion that the α1β1 isoform is only in interneurones 
. Whilst consistent with the in situ
hybridization result carried out by the same group 
, and with our results with the same antibody used at a similar dilution (1
10,000), the persistence of the staining in α1-knockout tissue raises concerns about the authenticity of the staining (). Nevertheless, populations of interneurones did display particularly intense cGMP immunoreactivity in our study ().
In the hippocampus, NO signalling has been most studied in the context of LTP in the CA1 subfield and current evidence suggests that after its formation by nNOS postsynaptically, NO acts both pre- and postsynaptically to bring about a coordinated and persistent increase in synaptic efficacy. Postsynaptically, functional evidence indicates that NO, through cGMP, triggers gene expression 
and the insertion of AMPA receptors into the cell membrane 
. By showing cGMP accumulation in at least some CA1 pyramidal neurones (), our results provide anatomical support for a postsynaptic site of action. Higher resolution methods will be needed to explore the subcellular locations of cGMP but there was obvious staining of the apical dendrites of CA3 neurones (, and ) and, to judge from the lack of empty dendritic profiles extending into stratum radiatum in the CA1 subfield (, , ), it is probable that pyramidal cell dendrites are NO-responsive. Immunohistochemistry for the β1-subunit of NO-activated guanylyl cyclase also produces staining in CA1 dendrites (
; Figure S3
). Presynaptically, NO reportedly increases neurotransmitter release through cGMP 
. Consistent with this site of action, cGMP labelling was seen in axons throughout the hippocampus () but, again, higher resolution would be needed to determine if cGMP accumulates in nerve terminals specifically. Also consistent with a presynaptic effect of NO, examination of the location of the β1 NO receptor subunit by immunohistochemistry reports presynaptic staining in some synapses 
, as do studies showing NO-evoked cGMP accumulation in some nerve terminals 
. Hence, an anatomical picture that coheres with functional evidence with respect to hippocampal NO-cGMP signalling and synaptic plasticity is beginning to taking shape.