Neurogranin is locally concentrated in rat cortical and hippocampal neurons
Mar t in N e u n e r - J e h l e a, J ean -P i e r r e D e n i z o t b, J acques Ma l l e t a,*
a Laboratoire de Gdndtique Mol~culaire de la Neurotransmission et des Processus Neurod~gdn~ratifs, C.N.R.S., Bat. CERVI, H@ital de la Piti~ Salp~tribre, 83, by. de l'H@ital, F-75613 Paris, France
b lnstitut A. Fessard, C.N.R.S., 91198 Gif-sur-Yvette Cedex, France
Accepted 25 June 1996
Abstract
The rat protein kinase C substrate neurogranin has a granular distribution in cortical and hippocampal neurons. We demonstrate that in these cells, granular labelling corresponds to a local concentration of neurogranin-immunoreactivity at both the membranes of mitochondria and trans-Golgi vesicles and soma-proximal dendritic shaft and spinal head structures. Our findings suggest that the function of neurogranin could be affected by protein assembly at these cellular sites.
Neurogranin (M r - -17 kDa; NG, RC3, BICKS) [7,20,28,29] is a neuron-specific rat protein, homologous to the bovine protein p 17 [2]. It has recently been recognized as an in vivo substrate for both the c~-, /3-, and y-isoforms of the serine/threonine protein kinase C (PKC) [4,12] and the synapse-associated Ca2+-dependent phosphorylase ki- nase [18]. These kinases phosphorylate the Ser-36 residue of NG in the IQ domain [5] which comprises a calmodulin (CAM) binding motif [3,18,27]. As shown in vitro, an addition of a negatively charged phosphate to Ser-36 re- duces the affinity of NG to CaZ+-free CaM [10]. This adds to the notion that NG is a cellular CaM regulator, linking protein kinase-dependent processes with CaM-mediated pathways [11,16]. Several potential cellular sites for the action of NG have been identified in rat striatal neurons [29]. NG is dispersed in cell bodies but associated with cell nuclei, dendritic spines, and some myelinated axons. Thus, in the striatum, NG may act in a variety of cellular compartments. In cortical and hippocampal cells, NG-im- munolabelling has been reported to be often granular and associated with yet unspecified cellular structures [20,28]. To identify these structures, we analyzed the subcellular distribution pattern of the protein in NG-enriched rat brain areas including layer II of the parietal cerebral cortex and the hippocampal area CA1 [17,20].
Three male Sprague-Dawley rats (200-300 g b.wt; kept under a 12 h light/12 h dark cycle) were anaes- thetized (0.4 g chloral hydrate/kg b.wt; i.p.) and transcar- dially perfused (fixative: 4% (w/v) paraformaldehyde, 15% (v /v ) picric acid, 0.15 M phosphate buffer pH 7.4 [25]) at light-onset. 60 /xm frontal sections were cut from their brains with a vibrotome and postfixed overnight at 4°C in the same fixative containing 1% (v /v ) glutaralde- hyde. Freely floating sections were then exposed to 0.1% (v /v ) Triton X-100 in phosphate-buffered saline (PBS) for 1 h at ambient temperature and incubated overnight at 4°C with affinity-purified anti-NG antibodies (10 /xg /ml PBS, 2% (v /v ) normal goat serum) [17]. Immunoreaction was carried out with the Vectastain Elite Kit (Vector Laborato- ries), 3,3'-diaminobenzidine (DAB, Sigma), and hydrogen peroxide according to the manufacturer's instructions. Control sections of the same brain regions were similarly treated but not incubated with anti-NG antibodies. After overnight fixation at 4°C with 0.25% (w/v) osmium te- traoxide and 0.75% (w/v) potassium-perchlorate in 0.1 M phosphate buffer (pH 7.8), DAB-stained and control sec- tions were embedded in Spurr's resin (TAAB). 800 slices were cut from their immunoreactive surface using an ultratome (LKB), mounted on copper grids, and inspected under an electron microscope (Philips CM10). Where indi- cated, ultrathin slices were counterstained at ambient tem- perature with saturated solutions of uranyl acetate (Pro- labo) for 30 rain and lead citrate (ICN Pharmaceutical) for 3 min [21 ].
150 M. Neuner-Jehle et a l . / Brain Research 733 (1996) 149-154
The specificity of immunostaining for NG protein was verified by light microscopy on sections of three rat brains. The distribution of NG-immunoreactivity (IR) is similar in
all rats examined. In particular, layer II neurons of the parietal cerebral cortex and pyramidal cells of the hip- pocampal area CA1 revealed intense NG-IR in both neu-
M. Neuner-Jehle et al. / Brain Research 733 (1996) 149-154 151
ronal perikarya and the soma-proximal region of their dendrites (Fig. 1A and Fig. 2A). This pattern is identical to the previously described distribution of both NG mRNA and NG protein [17,20,22,28,29], indicating that the affin- ity-purified anti-NG antibodies used specifically recognize NG protein in rat brain sections.
At the subcellular level, the NG-IR distribution patterns in both cortical layer II neurons and hippocampal CA1 pyramidal cells are similar in all rat brains examined. In these neurons, dispersed electron-dense NG-deposits are confined to peripheral cytoplasmic sites of cell somata whereas the perikaryal plasma membrane is devoid of immunolabelling (Fig. 1C and Fig. 2C). In close associa- tion with regions of dispersed staining, which may corre- spond to cytoplasmic sites of a non-aggregated NG protein form, a local concentration of electron-dense immunoreac- tion deposits was observed near and at the membrane of particular hippocampal mitochondria (Fig. 2C, arrow heads). Membranous labelling of mitochondria is apparent also in cortical cells (Fig. 1C, double arrow) but not in control sections of either brain region (Fig. 1B and Fig. 2D). Furthermore, membranous immunolabelling was fre- quently detected at both the Golgi apparatus (GA) and the endoplasmic reticulum (ER) of cortical and hippocampal neurons. At the GA, electron-dense immunoreaction is confined to the t rans -Golg i system where it is associated mainly with membranes of vesicular structures (Fig. 1C,D and Fig. 2C; arrows). This observation is consistent with a previous report of the presence of NG in microsomal rat brain preparations [30]. It also indicates that the protein may serve as a substrate for the Golgi-associated c~- and fllt-isoforms of PKC thought to be involved in vesicle trafficking [13,23]. At the ER, membranous labelling is less pronounced and was observed also in control sections, indicating that the ER membrane may be non-specifically stained. Electron-dense immunolabelling has been found neither within mitochondria nor in the lumen of both the GA and the ER. Similarly, the cell nuclei of cortical and hippocampal neurons and their nuclear membranes are devoid of NG-IR. The local concentration of NG in partic- ular perikaryal compartments in both brain regions con- trasts with the previously reported dispersed NG-im- munolabelling in neuronal somata of the rat striatum [29]. It further suggests that cellular protein accumulation is characteristic of cortical and hippocampal neurons. As it
has been proposed for NG, a glycine-rich sequence at the carboxy-terminus may mediate multimeric protein assem- bly [28]. Moreover, NG protein assembly could also de- pend on ADP-ribosylation which inhibits polymerization of other proteins such as tubulin [24] and non-muscle actin [1]. Indeed, NG has recently been found to be ADP- ribosylated near its PKC phosphorylation site [7]. Such reversible post-translational protein modification may ac- count for the observed co-localization of the aggregated and dissociated forms of NG in cortical and hippocampa! neurons. By introducing a net negative charge into the CaM binding region of NG, ADP-ribosylation may also influence the protein's binding affinity to Ca2+-free CaM. Possibly, ADP-ribosylation-dependent protein assembly regulates the availability of NG for protein kinase- and CaM-mediated cellular reaction pathways.
Similar to the pattern of a locally confined perikaryal immunolabelling, electron-dense NG-IR is also concen- trated within both dendritic shaft and spine compartments. In dendritic shafts of hippocampal pyramidal cells, NG-IR is apparent at and close to filamentous structures of the cytoskeleton (Fig. 2B; arrows). It is unclear whether NG expresses its function at these sites or is transported to other cellular compartments. Highest levels of dendritic NG-IR were observed in particular spine heads of cortical layer II (Fig. 1E) and apical hippocampal CA1 dendrites (Fig. 2E). The spinal immunoreactivity is consistent with a previous report of the presence of NG in synaptosomal rat brain preparations, as determined by immunoblotting [30]. Spinal NG-IR is associated mainly with soma-proximal axospinal symmetric synapses whereas axosomal junctions and shaft synapses are devoid of immunolabelling. In immuno-positive spines, NG-immunolabelling is particu- larly strong within the cytoplasm of their head compart- ments, in postsynaptic densities, and on the subsynaptic membrane (Fig. 1C,E and Fig. 2E). This post- and subsy- naptic staining pattern resembles that of NG-positive den- dritic spines of asymmetric synapses in rat striatum [29]. The presence of NG in both cytoplasmic and membranous cellular environments is consistent with both the detection of the protein in soluble and membranous protein fractions of the rat cerebral cortex and an association of NG with insoluble synaptosomal protein fractions [3,11,30]. In the cytoplasm of immuno-positive spine heads, NG-IR is often granular. As described for neuronal perikarya, an accumu-
Fig. 1. NG-immunostaining in cell bodies and soma-proximal dendritic synapses of cortical layer II neurons. A: under a light-microscope, NG-IR is visible mainly in cell somata and their soma-proximal dendritic regions (arrows). B: in a control section, neither the cell nucleus (nu) nor mitochondria (mi) are immunolabelled. The membrane of the endoplasmic reticulum (er) appears to be more electron-dense than other membranous structures. C: near the perikaryal plasma membrane (asterisk), immunolabelling is closely associated with membranes of both mitochondria (mi; double arrow) and vesicles of the trans-Golgi apparatus (ga; arrow) whereas the cell nucleus (nu) and the endoplasmic reticulum (er) are devoid of immunoreactivity. Near the NG-positive cell body, dendritic spine heads are depicted showing intense granular immunostaining (arrow heads). D: close to the perikaryal plasma membrane, granular NG-immunolabelling is associated with a Golgi-like tubular structure (arrow). A myelinated axon (ax) nearby is devoid of NG-IR. E: on a counterstained ultrathin slice, granular NG-IR is detected in a dendritic spine head, notably in the cytoplasm, the postsynaptic density (arrow head), and on the subsynaptic membrane (double arrows)'. Together with an unlabelled dendritic spine nearby (asterisk) both spines form synapses with an unlabelled axonal bouton (ab). Brain sections: 60 /zm (A), 800 A (B-E), Scale bars: 50 /~m (A), 600 nm (B-D), 180 nm (E).
~4
I
M. Neuner-Jehle et al. / Brain Research 733 (1996) 149-154 153
lation of NG-IR in dendritic spines could be caused by protein assembly which may affect the availability of NG for PKC-dependent and CaM-mediated second messenger pathways [8,10]. Alternatively, granular immunoreaction could represent sites of mRNA translation at dendritic or spinal polyribosomes [6,26]. Evidence for dendritic mRNA expression has previously been provided by in situ hy- bridization histochemistry in rats showing NG mRNA in soma-proximal dendrites of both hippocampal pyramidal and granule cells and cortical neurons [15]. Moreover, both variants of NG mRNA (1.0 kb, 1.4 kb [22,28]) were detected in mossy-fiber synaptosomes of the rat hippocam- pal area CA1 [6], suggesting that both variants can be expressed in spinal compartments. Unlike dendritic spines, presynaptic terminals and myelinated axons of cortical and hippocampal neurons are devoid of electron-dense NG-im- munoreaction deposits. This is in contrast to the rat stria- tum where NG-IR has been detected in both presynaptic boutons of striatal neurons and myelinated corticofugal axons [29].
The specificity of NG-IR for a subpopul~tion of soma- proximal spine heads indicates that NG acts at synapses with particular functional properties. Long-term potentia- tion (LTP) of synaptic activity, for instance, leads to an increase of NG phosphorylation in the CA1 region of rat hippocampal slices [14,19]. Moreover, an injection of anti- NG antibodies into cultured rat hippocampal CA1 pyrami- dal cells prevents the establishment of tetanus-evoked LTP [9]. The protein may therefore play a critical role in regulating postsynaptic processes of LTP. With respect to the observed specific spinal distribution pattern of NG in the rat cerebral cortex and hippocampus, it is possible that the protein accumulates in dendritic spines which are subjected to LTP-related changes in synaptic activity. This notion further implies that NG could serve as a neuronal marker for synaptic plasticity processes.
Acknowledgements
We thank Jean-Paul Bouillot and Denise Champion for photographic assistance. The work was supported by a grant from the Swiss National Research Foundation to MNJ.
References
[1] Aktories, K. and Wegner, A., Mechanisms of the cytopathic action of actin-ADP-ribosylating toxins, Mol. Microbiol., 6 (1992) 2905- 2908.
[2] Baudier, J., Bronner, C., Kligman, D. and Cole, R.D., Protein kinase C substrates from bovine brain: purification and characterization of neuromodulin, a neuron-specific calmodulin-binding protein, J. Biol. Chem., 264 (1989) 1824-1828.
[3] Baudier, J., Deloulme, J.C., Van Dorsselaer, A., Black, D. and Matthes, H.W.D., Purification and characterization of a brain-specific protein kinase C substrate, neurogranin (pl7), J. Biol. Chem., 266 (1991) 229-237.
[4] Chen, S.-J., Klann, E., Gower, M.C., Powell, C.M., Sessoms, J.S. and Sweatt, J.D., Studies with synthetic peptide substrates derived from the neuronal protein neurogranin reveal structural determinants of potency and selectivity for protein kinase C, Biochemistry, 32 (1993) 1032-1039.
[6] Chicurel, M.E., Terrian, D.M. and Potter, H., mRNA at the synapse: analysis of a synaptosomal preparation enriched in hippocampal dendritic spines, J. Neurosci., 13 (1993) 4054-4063.
[7] Coggins, P.J., McLean, K. and Zwiers, H., Neurogranin, a B- 50/GAP-43-immunoreactive C-kinase substrate (BICKS), is ADP- ribosylated, FEBS Lett., 335 (1993) 109-113.
[8] Cohen, R.W., Margulies, J.E., Coulter, P.M. and Watson, J.B., Functional consequences of expression of the neuron-specific, pro- tein kinase C substrate RC3 (neurogranin) in Xenopus oocytes, Brain Res., 627 (1993) 147-152.
[9] Fedorov, N.B., Pasinelli, P., Oestreicher, A.B., DeGraan, P.N.E. and Reymann, K.G., Antibodies to postsynaptic PKC substrate neuro- granin prevent long-term potentiation in hippocampal CA1 neurons, Eur J. Neurosci., 7 (1995) 819-822.
[10] Gerendasy, D.D., Herron, S.R., Watson, J.B. and Sutcliffe, J.G., Mutational and biophysical studies suggest RC3/neurogranin regu- lates calmodulin availability, J. Biol. Chem., 269 (1994) 22420- 22426.
[11] Houbre, D., Duportail, G., Deloulme, J.-C. and Baudier, J., The interactions of the brain-specific calmodulin-binding protein kinase C substrate, neuromodulin (GAP 43), with membrane phospholipids, J. Biol. Chem., 266 (1991) 7121-7131.
[12] Huang, K.P., Huang, F.L. and Chen, H.C., Characterization of a 7.5-kDa protein kinase C substrate (RC3 protein, neurogranin) from rat brain, Arch. Biochem. Biophys., 305 (1993) 570-580.
[13] Ito, A., Saito, N., Hirata, M., Kose, A., Tsujino, T., Yoshihara C., Ogita K., Kishimoto A., Nishizuka Y. and Tanaka C., Immunocyto- chemical localization of the oz subspecies of protein kinase C in rat brain, Proc Natl. Acad. Sci. USA, 87 (1990) 3195-3199.
[14] Klann, E., Chen, S.-J. and Sweatt, J.D., Increased phosphorylation of a 17-kDa protein kinase C substrate (P17) in long-term potentia- tion, J. Neurochem., 58 (1992) 1576-1579.
[15] Landry, C.F., Watson, J.B., Kashima, T. and Campagnoni, A.T.,
Fig. 2. Perikaryal and dendritic immunostaining of NG in hippocampal CA1 pyramidal cells. A: peroxidase immunostaining in both cell bodies of pyramidal cells (py) and their apical dendrites (arrow) at light-microscopical magnification. B: in a dendritic shaft, immunolabelling is associated with filamentous structures of the cytoskeleton (arrows). C: in a cell body, dispersed NG-IR is apparent in the cytoplasm close to the plasma membrane (asterisks) whereas the cell nucleus (nu) is unlabelled. Electron-dense NG-IR is concentrated near the membrane of mitochondria (mi; arrow heads) and at vesicles of the. trans-Golgi apparatus (ga; arrows). D: in a control section, immunostaining is absent from the perikaryal cytoplasm, endoplasmic reticulum (er), Golgi apparatus (ga), mitochondria (mi), and dendritic shafts (ds). E) Granular NG-IR is visible at the postsynaptic (arrow) and subsynaptic (arrow head) membranes of a soma-proximal dendritic spine head. Brain sections: 60 /~m (A), 800/~ (B-E). Scale bars: 100 /zm (A), 800 nm (C,D), 200 nm (B,E).
154 M. Neuner-Jehle et a l . / Brain Research 733 (1996) 149-154
Cellular influence on RNA sorting in neurons and glia: an in situ hybridization histochemical study, MoL Brain Res., 27 (1994) 1-11.
[16] Martzen, M.R. and Slemmon, J.R., The dendritic peptide neuro- granin can regulate a calmodulin-dependent target, J. Neurochem., 64 (1995) 92-100.
[17] Neuner-Jehle, M., Rhyner, T.A. and Borb61y, A.A., Sleep depriva- tion differentially alters the mRNA and protein levels of neurogranin in rat brain, Brain Res., 685 (1995) 143-153.
[18] Pandel, H.K., Zwiers, H. and Wang, J.H., Phosphorylase kinase phosphorylates the calmodulin-binding regulatory regions of neu- ronal tissue-specific proteins B-50 (GAP-43) and neurogranin, J. Biol. Chem., 268 (1993) 6207-6213.
[19] Ramakers, G.M., DeGraan, P.N., Urban, I.J., Kraay, D., Tang, T., Pasinelli, P., Oestreicher, A.B. and Gispen, W.H., Temporal differ- ences in the phosphorylation state of pre- and postsynaptic protein kinase C substrates B-50/GAP-43 and neurogranin during long-term potentiation, J. Biol. Chem., 270 (1995) 13892-13898.
[20] Represa, A., Deloulme, J.C., Sensenbrenner, M., Ben-Ari, Y. and Baudier, J., Neurogranin: Immunocytochemical localization of a brain-specific protein kinase C substrate, J. Neurosci., 10 (1990) 3782-3792.
[21] Reynolds, E.S., The use of lead citrate at high pH as an electron opaque stain in electron microscopy, J. Cell BioL, 17 (1963) 208-212.
[22] Rhyner, T.A., Borb61y, A.A. and Mallet, J., Molecular cloning of forebrain mRNAs which are modulated by sleep deprivation, Eur. J. Neurosci., 2 (1990) 1063-1073.
[23] Saito, N., Kose, A., Ito, A., Hosoda, K., Moil, M., Hirata, M., Ogita, K., Kikkawa, U., Ono, Y., Igarashi, K., Nishizuka, Y. and Tanaka,
C., Immunocytochemical localization of /3 n subspecies of protein kinase C in rat brain, Proc. Natl. Acad. Sci. USA, 86 (1989) 3409-3413.
[24] Scaife, R.M., Wilson, L. and Purich, D.L., Microtubule protein ADP-ilbosylation in vitro leads to assembly inhibition and rapid depolymerization, Biochemistry, 31 (1992)310-316.
[25] Somogyi, P. and Takagi, H., A note on the use of picric-acid-para- formaldehyde-glutaraldehyde fixative for correlated light and elec- tronmicroscopic immunocytochemistry, Neuroscience, 7 (1982) 1779-1783.
[26] Steward, O. and Banker, G.A., Getting the message from the gene to the synapse: sorting and intracellular transport of RNA in neurons, Trends Neurosci., 15 (1992) 180-186.
[27] Wakim, B.T., Alexander, K.A., Masure, H.R., Cimler, B.M., Storm, D.R. and Walsh, K.A., Amino acid sequence of P-57, a neurospe- cific calmodulin-binding protein, Biochemistry, 26 (1987) 7466- 7470.
[28] Watson, J.B., Battenberg, E.F., Wong, K.K., Bloom, F.E. and Sutcliffe, J.G., Subtractive cDNA cloning of RC3, a rodent cortex- enriched mRNA encoding a novel 78 residue protein, J. Neurosci. Res., 26 (1990) 397-408.
[29] Watson, J.B., Sutcliffe, J.G. and Fisher, R.S., Localization of the protein kinase C phosphorylation/calmodulin-binding substrate RC3 in dendritic spines of neostriatal neurons, Proc. Natl. Acad. Sci. USA, 89 (1992) 8581-8585.
[30] Watson, J.B., Szijan, 1. and Coulter II, P.M., Localization of RC3 (neurogranin) in rat brain subcellular fractions, Mol. Brain Res., 27 (1994) 323-328.
