Michael L. Brines

Glutamate receptor subunits GluR1 and GluR2/3 distribution shows reorganization in the human epileptogenic hippocampus

The AMPA‐type glutamate receptor subunits GluR1 and GluR2/3 were localized by immunohistochemistry with subunit‐specific antibodies in hippocampi removed surgically from patients with temporal lobe epilepsy for the control of seizures. The flip and flop splice variants of the subunits were localized by in situ hybridization histochemistry with specific oligoprobes. In patient hippocampi that were not the seizure focus, the GluR1 subunit proteins were diffusely expressed on the dendrites of neurons in all regions. In contrast, in these same hippocampi, the GluR2/3 subunit proteins were expressed strongly on the soma and proximal dendrites of principal neurons in all regions. The flip variant of these subunits was localized in the hilus and fields of Ammons Horn (CA), while the flop variants were prominent on the dentate granule cells. In the epileptogenic hippocampus, while immunoreactivity was decreased in all fields that showed neuronal loss, there was an increased expression of GluR1 on the dendritic excrescences on the proximal dendrites of hilar neurons and CA3 pyramidal neurons, as well as expression of GluR2/3 in hilar neuron excrescences. Electron microscopic examination confirmed that the GluR1 immunoreactivity was only in dendritic processes, particularly dense at the postsynaptic membranes. Such expression of GluR1 may provide for an enhanced glutamatergic response by these neurons. GluR2/3 was also significantly increased on the dendrites of dentate granule cells in the epileptogenic hippocampus and may provide some protection against excitotoxic injury by reducing calcium flux into neurons.

Cell-type specific expression of Na+, K(+)-ATPase catalytic subunits in cultured neurons and glia: evidence for polarized distribution in neurons.

Na+,K(+)-ATPase (the sodium pump) is a family of proteins consisting of catalytic (alpha) and glycoprotein (beta) subunit isoforms which are differentially expressed in excitable tissue. To gain insight into the cell-type distribution of sodium pump protein, we determined the expression pattern of fetal rat telencephalic cultures, of telencephalic cultures depleted of neurons, and of pure astrocyte cultures. Isoform-specific antibodies were used for immunoblotting and immunohistochemistry, with supplemental [3H]ouabain binding to assess levels of functional alpha 2/alpha 3 protein. The results show that neurons of mixed telencephalic cultures uniquely express alpha 3 and high levels of alpha 1. The marked similarity in the distribution of microtubule-associated protein-2 and alpha 1 immunocytochemical staining strongly suggests that alpha 1 subunits are enriched in dendrites. Further, highly correlative growth cone-associated protein-43 and alpha 3 staining is consistent with a preferential expression of alpha 3 subunits in axons, which are also characterized by low levels of alpha 1 and no alpha 2 immunoreactivity. Process-bearing glia are intimately associated with neuronal aggregates and express high levels of both alpha 1 and alpha 2 protein, as well as GFAP. Interestingly, polygonal, flat glia not within neuronal aggregates are weakly immunopositive only for alpha 1 and GFAP. Pure astrocytic cultures possess appreciable alpha 1 protein and GFAP, but lack both alpha 2 and alpha 3 immunoreactivity. As predicted by the immunohistochemical findings, [3H]ouabain binding was low in pure astrocytic cultures, and much higher in the neuron-enriched mixed cultures. These observations confirm that neurons express all three catalytic isoforms of the sodium pump. They also suggest that specific alpha-isoforms may be polarized to targeted membrane regions of neurons. Further, glia intimately associated with neurons express alpha 2, bind significant amounts of [3H]ouabain, and possess much higher levels of alpha 1 and GFAP compared to glia not near neurons. Thus, neurons may regulate glial sodium pump expression.

Inhibition of α2/α3 sodium pump isoforms potentiates glutamate neurotoxicity

Abstract Excessive stimulation of neurons by glutamic acid initiates a destructive cascade of ion fluxes, cellular swelling, and death. Homeostatic mechanisms which rectify these disturbances depend largely upon transmembrane ion gradients maintained by Na+, K+-ATPase (NaP). We proposed that the neurotoxicity of glutamate is enhanced when the NaP capacity is exceeded, and therefore, that the degree of neuronal death varies inversely with endogenous NaP activity. To test this concept, we directly reduced NaP activity in cultured rat telencephalic cells using either the specific inhibitor ouabain, or dcAMP, and assessed whether these treatments increased glutamate-induced neuronal death. Since rodent NaP catalytic subunits possess both low (α1) and high (α2/α3) affinity for ouabain, we were able to inhibit selectively the α2 (principally glial) and α3 (neuronal) catalytic subunits without affecting the α1 isoform. Brief exposures (5–60 min) to high ouabain concentrations (1–10 mM), which blocks the activity of all three catalytic subunits, killed differentiated neurons but spared glia. In contrast, differential inhibition of the α2/α3 isoforms (by 1 μM ouabain) was not of itself toxic, but produced a supersensitivity to glutamate. [3H]Ouabain binding studies confirmed that the glutamate neurotoxicity observed varied inversely with the degree of NaP inhibition. Further, this relationship was not absolutely dependent upon ouabain, since reduction in α2/α3 pump activity induced by dcAMP also amplified glutamate toxicity. We conclude glutamate excitotoxicity. Since the distribution of NaP is highly heterogenous in the nervous system, with similar cell types varying greatly in isoform expression, constitutive levels of this isoenzyme could constitute a major factor in the survival of stimulated neurons. Further, factors which directly affect pump activity, such as activation of protein kinase A, may modulate excitotoxicity.

Dynorphin and the kappa 1 ligand [3H]U69,593 binding in the human epileptogenic hippocampus.

The distribution of dynorphin (DYN), one of its binding sites (kappa 1 receptor) and their relationship to neuronal loss and granule cell hyperexcitability was examined in hippocampi from patients with temporal lobe epilepsy (TLE). In hippocampi that were not the seizure focus (mass associated temporal lobe epilepsy, MaTLE; and paradoxical temporal lobe epilepsy, PTLE) DYN-like immunoreactivity was localized in the dentate granule cells and their mossy fiber terminals within the hilus and area CA3. In hippocampi that were the seizure focus (MTLE), 89% showed an additional band of immunoreactivity confined to the inner molecular layer (IML) of the dentate gyrus, representing recurrent mossy fiber collaterals. In 11% of MTLE patients no staining was found in the IML (MTLE/DYN-). The MTLE/DYN- hippocampi were also characterized by a significantly lower degree of cell loss than in MTLE hippocampi in the dentate granule cell layer, the hilus and CA3. Both MTLE and MTLE/DYN- hippocampi showed evoked epileptiform bursting in granule cells while MTLE showed greater polysynaptic EPSPs and spontaneous excitatory activity. Thus granule cell recurrent collateral sprouting may account for only some aspects of hyperexcitability. In 30% of the MTLE group, hilar neurons of a variety of morphological types expressed DYN immunoreactivity in their somata and dendrites. The density of [3H]U69,593 binding sites in MaTLE and PTLE patients was highest in areas CA1 and the subiculum-regions having little or no DYN-staining. In the dentate molecular layer, hilus and CA3--regions with the most DYN immunoreactivity--there was a low density of ligand binding. The significance of this transmitter/receptor mismatch is yet unknown.

Quantitative Autoradiographic Analysis of lonotropic Glutamate Receptor Subtypes in Human Temporal Lobe Epilepsy: Up‐regulation in Reorganized Epileptogenic Hippocampus

Medically intractable temporal lobe epilepsy is a common disease typically associated with hippocampal damage (sclerosis) and synaptic remodelling. These changes could include increased glutamate receptor expression, enhancing excitability and the potential for neuronal injury. We directly assessed this hypothesis using quantitative in vitro receptor autoradiography to determine the densities of glutamate‐, NMDA‐, quisqualateh‐amino‐3‐hydroxy‐5‐methyl‐isoxazoleproprionic acid (AMPA)‐ and kainic acid‐preferring binding sites in surgically removed hippocampi from patients with mesial temporal lobe epilepsy (sclerosis; MTLE) and patients with mass‐associated temporal lobe epilepsy (no sclerosis: MaTLE), compared with autopsy material. Neuronal cell counts and in situ total protein densities were also obtained. In general, MaTLE and autopsy binding densities were indistinguishable. In contrast, some regions of MTLE hippocampi exhibited decreased receptor densities, with a corresponding loss of protein. In the hilus and CA1, however, ligand binding densities did not differ from the comparison groups in spite of markedly reduced protein content, consistent with increased glutamate receptor density. Kainate‐preferring sites were distributed differently from the other glutamate subtypes and were uniformly decreased throughout the MTLE hippocampus, except for a unique expression within the outer dentate molecular layer. Along with increased NMDA and AMPA receptor densities in the hilus and CA1, this distinctive population of kainate receptors establishes that increased glutamate receptor expression is a feature of the remodelled MTLE hippocampus. These observations suggest that enhanced sensitivity to glutamate may be an important element in the pathophysiology of temporal lobe epilepsy.

Regional Distributions of Hippocampal Na+, K+‐ATPase, Cytochrome Oxidase, and Total Protein in Temporal Lobe Epilepsy

Summary Na+, K+‐ATPase (the sodium pump) is a ubiquitous enzyme that consumes ATP to maintain an adequate neuronal transmembrane electrical potential necessary for brain function and to dissipate ionic transients. Reductions in sodium pump function augment the sensitivity of neurons to glutamate, increasing excitability and neuronal damage in vitro. Temporal lobe epilepsy (TLE) is one disease characterized by hyperexcitability and marked hippocampal neuronal losses that could depend in part, on impaired sodium pump capacity secondary to changes in sodium pump levels and/or insufficient ATP supply. To assess whether abnormalities in the sodium pump occur in this disease, we used [3H]ouabain to determine the density of Na+, K+‐ATPase for each anatomic region of hippocampus by in vitro autoradiography. Tissues were surgically obtained from epileptic patients with hippocampal sclerosis and compared with specimens from patients with seizures originating from temporal lobe tumors and autopsy controls. Changes in cellular population arising from neuronal losses or gliosis were assessed by protein densities derived from quantitative computerized densitometry of Coomassie‐stained tissue sections. We estimated regional differences in capacity for ATP generation by determining cytochrome c oxidase (CO) activity. Principal neurons of hippocampus exhibit high levels of sodium pump enzyme. Both epilepsy groups exhibited slight but significant increases in sodium pump densityhnit mass of protein in the dentate molecular layer, CA2, and subiculum as compared with autopsy controls. Greater hilar sodium pump density was also observed in sclerotic hippocampi. In contrast, CO activity was reduced in both epilepsy types throughout hippocampus. Results suggest that although sodium pump protein in surviving neurons appears to be upregulated in epilepsy, sodium pump capacity may be limited by the reduced levels of CO activity. Functional reduction in sodium pump capacity may be an important factor in hyperexcitability and neuronal death.

Parathyroid hormone-related protein protects against kainic acid excitotoxicity in rat cerebellar granule cells by regulating L-type channel calcium flux.

The parathyroid hormone-related peptide (PTHrP) and PTH/PTHrP receptor genes are widely expressed in the CNS and both are highly expressed in the cerebellar granule cell. We have shown previously that PTHrP gene expression in granule cells is depolarization-dependent in vitro and is regulated specifically by Ca2+ influx via L-type voltage-sensitive calcium channels (L-VSCCs). Kainic acid induces long-latency excitotoxicity in granule cells via L-VSCC-mediated Ca2+ influx. Here, we show that PTHrP is just as effective as the L-VSCC blocker, nitrendipine (NTR), in preventing kainate excitotoxicity. A competitive inhibitor of PTHrP binding abrogates its neuroprotective effect. Both NTR and PTHrP decrease 45Ca2+ influx to the same degree. These findings suggest that PTHrP functions in an autocrine/paracrine neuroprotective feedback loop that can combat L-VSCC-mediated excitotoxcity.

The cardiac glycoside ouabain potentiates excitotoxic injury of adult neurons in rat hippocampus

We demonstrate that the enzyme family responsible for the restoration of the transmembrane cation balance, namely the sodium pump (Na+, K(+)-ATPase), plays a critical role in whether glutamate injures adult neurons in vivo. Partial inhibition of the sodium pump by the cardiac glycoside ouabain in young adult rats is not itself damaging. This treatment, however, markedly potentiates ordinarily subtoxic dosages of the glutamate analog kainic acid to produce limbic seizures and widespread neurodegeneration within the hippocampus in a pattern closely resembling that observed for human temporal lobe epilepsy.

Parathyroid Hormone-related Peptide Is Produced by Cultured Cerebellar Granule Cells in Response to L-type Voltage-sensitive Ca2+ Channel Flux via a Ca2+/Calmodulin-dependent Kinase Pathway

Parathyroid hormone (PTH)-related peptide (PTHrP) is expressed in the adult mammalian brain, but its function is unknown. Here we show that PTHrP and the PTH/PTHrP receptor are products of cerebellar granule cells in primary culture. Granule cells maintained under depolarizing conditions (25 mM K+) make and release PTHrP. Further, PTHrP-(1-36) stimulates cAMP accumulation in granule neurons in a dose-dependent manner with half-maximal activation at ∼16 nM. Granule cell PTHrP mRNA is activity-dependent, and the pathway of regulation depends absolutely on the flux of Ca2+ ions through the L-type voltage-sensitive Ca2+ channel and the Ca2+/calmodulin kinase cascade. PTHrP is therefore a neuropeptide whose regulation depends upon L-type voltage-sensitive Ca2+ channel activity, and the gene is expressed under conditions that promote granule cell survival.

Cytoarchitectural relationships between [3H]ouabain binding and mRNA for isoforms of the sodium pump catalytic subunit in rat brain

We examined the cell type-specific expression of the alpha 1, alpha 2, and alpha 3 subunits of the sodium pump in rat brain using in situ hybridization and [3H]ouabain autoradiography. These techniques allowed us to colocalize mRNA and functional alpha 2/alpha 3 pumps on adjacent sections. The perikarya of many neurons possessed high levels of alpha 1 and/or alpha 3 transcripts, while alpha 2 mRNA appeared to be present in only a few neuronal types. [3H]Ouabain binding in general paralleled the distribution of alpha 3 mRNA-positive neurons. The regional variation of alpha 1 and alpha 3 transcripts was complex and varied. Large neurons of the olfactory bulb and piriform cortex expressed high levels of alpha 3 transcripts, but low levels of alpha 1 mRNA. In frontal cortex, neurons of layers II-III were enriched in alpha 1 mRNA, while those in layer V exhibited high levels of alpha 3 transcripts. In the hippocampus, principal neurons expressed all three alpha subunit mRNAs. CA subfield pyramidal neurons exhibited a high alpha 3/alpha 1 ratio, while dentate granule cells and hilar pyramidal neurons expressed approximately equal levels of alpha 1 and alpha 3. In the cerebellum, Purkinje and Golgi cells were rich in alpha 3 mRNA, while the granule cells appeared to express only alpha 1 transcripts. The distribution of functional sodium pump protein, as localized by [3H]ouabain binding, was highest in the neuropil of the hippocampus and cerebral cortex, and lowest over perikarya and white matter. [3H]ouabain did not bind to alpha 1 pump units, as confirmed by the complete absence of labeling over the choroid plexus, a tissue expressing only alpha 1 mRNA. In the cerebellum, regions of dense [3H]ouabain binding were localized to the granule cell layer, the inner third of the molecular layer in the basket region, and the deep cerebellar nuclei. Surprisingly, the dense neuropil in the outer 2/3 of the molecular layer lacked high [3H]ouabain binding. Thus, functional alpha 3 sodium pump units appear distributed to the axon terminals and not to apical dendrites of Purkinje, Golgi and basket cells. A similar pattern of increased [3H]ouabain binding in axonal but not dendritic fields of alpha 3-enriched neurons was present in the cerebral cortex and the hippocampus. Considering that many alpha 3-enriched neurons are of the Golgi I type with long axons, the alpha 3 isoform may be preferentially directed into axons to function in presynaptic membranes.