Abstract
Janus, the ancient Roman God of Gates and Doors had two faces: one looked into the past, and the other, into the future. Do neurons possess a Janus face when it comes to neurotransmitters, or a given neuron is to be forever solely γ-aminobutyric acid (GABA) ergic, glutamatergic, dopaminergic, peptidergic, or YOURPREFERREDTRANSMITTERergic? The answer is that the terminals of many neurons are homes to even more than two neurotransmitters. All this in spite of the “one neuron–one transmitter” usual misinterpretation of Sir Henry Hallett Dale's postulate, originally meant to indicate that a metabolic process taking place in the cell body can influence all processes of the same neuron. A large variety of neurons in the CNS, many of them GABAergic, produce and release chemicals that satisfy some of the criteria used to define neurotransmitters. The usual scenario for a dual-transmitter terminal is that the fast-acting transmitter such as GABA or glutamate is stored in regular synaptic vesicles, whereas a neuropeptide is stored in dense core vesicles (1). The vesicular zinc found in many glutamatergic terminals also may be considered to be a second neurotransmitter, based on its vesicular packaging with the aid of a specific vesicular transporter, and its postsynaptic actions through high-affinity binding sites and permeation through certain channels (2). Whenever a “fast” and a “slow” neurotransmitter are present in the same presynaptic terminal, it is customary to assume that their release can be differentially regulated (1). There is little convincing experimental support for this phenomenon in the mammalian CNS. The coexistence of two “fast” neurotransmitters in the same terminal is less frequent, but not unheard of. In neonatal sympathetic neurons cocultured with cardiac myocytes, norepinephrine and acetylcholine coexist and have opposite actions on the cardiac muscle cells (3). Very recently we learned that brain-derived neurotrophic factor acting at the low-affinity neurotrophin receptor p75NTR, perhaps as part of a programmed developmental switch, can convert the phenotype of the sympathetic neuron from noradrenergic to cholinergic (4). Other examples of two fast neurotransmitters released from the same neuron include GABA and glycine in interneurons of the spinal cord (5) and glutamate and dopamine in ventral midbrain dopamine neurons (6). Of all CNS neurons, the granule cells of the dentate gyrus appear to be the champions of neurotransmitter colocalization: glutamate, enkephalin, dynorphin, zinc, and finally GABA (2,7–9). With this many transmitters in a single neuron, there are probably different ways in which they can be released. Dynorphin and other opioid peptides can be released directly from the dendrites to inhibit excitatory transmission (8). A similar mechanism may take place for GABA, as described in cortical GABAergic neurons (10).
To produce fast GABAA receptor–mediated inhibitory events in target cells. These cells include interneurons, CA3 pyramidal cells, mossy cells, and, subsequent to the epilepsy-related mossy fiber sprouting, other granule cells.
To induce slow GABAB receptor–mediated inhibitory events in the target neurons mentioned earlier.
To cause presynaptic inhibition through GABAB receptors of glutamate or GABA release from other neighboring terminals, or from the same terminal that just released the GABA (20).
To elevate ambient levels of GABA producing tonic GABA conductances in hippocampal neurons equipped with nondesensitizing high-affinity GABAA receptors (21,22).
To buffer pH changes inside the granule cells, akin to the role of GABA in plants (23).
It would be foolish to attempt to speculate on the outcome of all of these possibilities and their interactions. Much of what we want to make of the role of GABA in dentate gyrus granule cells depends on what we believe to be the function of these neurons. They have been described as being the guardians of the hippocampus against an overpowering entorhinal input (24) or as merciless excitotoxic killers of various other cell types (25). Furthermore, the outcome of inhibiting interneurons by the GABA released from phenotypically altered granule cells must be considered. Preliminary findings, after kindling-like repetitive stimulations in slices, point to the existence of granule cell–mediated synaptic GABA responses not only in CA3 pyramidal cells but also in interneurons (11). This disinhibitory alternative must be seriously considered, particularly if we accept that the role of a dentate granule cell is to communicate with a handful of CA3 pyramidal cells while silencing most others. Neuroanatomists tell us that a single granule cell makes synaptic contacts with at least an order of magnitude more interneurons than do the 15 or so CA3 pyramidal cells it innervates (26).
Without any detailed experimental evidence, it will take a long time to decide whether the GABAergic phenotype of granule cells is pro- or antiepileptogenic. Nevertheless, by simultaneously releasing fast excitatory and inhibitory transmitters at times of neuronal conflagration, as in epilepsy, granule cells have revealed their ultimate Janus faces. In ancient Rome, of the many temples dedicated to Janus, one was called Ianus Geminus. It was a double-gated temple built in the Forum, serving a symbolic function. When its gates were closed, showing only one face, it signified peace within the Roman Empire. When the gates were open, both faces showing, it meant war. Alternatively, if the second face of Janus really looks into the past, the presence of GABA in granule cells during epilepsy may just be the reactivation of a developmental program reminding us of their GAD-given right to be GABAergic.