From ‘soup physiology’ to normal brain science

If we knew what it was we were doing, it wouldn't be called research, would it? (Albert Einstein) When we published our paper in the Journal of Physiology 42 years ago (Krnjević & Phillis, 1963), there was little information and even less consensus about signal transmission between cells in the brain. It all seems so obvious now: nerve cells talk to each other by sending out and recognizing chemical signals. Some act at a distance, some close by: albeit vastly differing in size – from simple elements to large proteins – the signals are virtually all chemical. How else could nerve cells communicate? By direct linkage in a continuous network proposed Golgi (Golgi et al. 2001), the inventor of the silver stain. But the neurone theory, which postulated real separation between nerve cells, won the day (Waldeyer, 1891; Cajal, 1909). How then are signals carried between cells? Influential brain scientists, from Cajal (1909) to Lorente de Nó (1939), believed that excitation was mediated by passive flow of the action current from the axon across the narrow synaptic gap. Current flow could not easily account for inhibition (though cf. Brooks & Eccles, 1947); but inhibition was of no importance in the brain (Fulton, 1949). Researchers working on peripheral junctions were generally more open to suggestions that signals might be chemical. Albeit initially based on somewhat shaky evidence, Loewi's (1921) proposed humoral mechanism of vagal inhibition was viewed quite seriously by the end of the decade. If acetycholine (ACh) was a mediator in the heart, what about ganglia? With improved bioassays and perfusions, good evidence was obtained of cholinergic transmission in ganglia (Feldberg & Gaddum, 1934; MacIntosh, 1938) and even in skeletal muscles (Dale, 1938). Before the end of the 1930s, slow muscarinic and fast nicotinic transmission at parasympathetic and other junctions had been distinguished, and adrenaline (and a close derivative) identified as the sympathetic transmitters. Humoral transmision (‘soup physiology’ according to the electrophysiologists) was the accepted mode of operation at peripheral junctions. Would the peripheral transmitters also play a major role in brain function (Dale, 1938; Feldberg, 1950; Rothballer, 1959)? There was little progress in relevant studies on the CNS, largely because of obvious technical difficulties. A turning point came early in the 1950s when Eccles (long a fierce opponent of chemical views) succeeded in recording synaptic potentials inside spinal motoneurones (Brock et al. 1951). What did he find? First, EPSPs (as he named them), very much like EPPs in muscle, though clearly not cholinergic. Second, and most important, IPSPs, manifested by hyperpolarization and large conductance increase, which could not be explained by electrical transmission. Henceforth, Eccles became an enthusiastic advocate of the chemical hypothesis (Eccles, 1957) – much to the dismay of his former brothers-in-arms and the amused delight of the ‘soup physiologists’. An essential element, however, was missing: the nature of the presumed central transmitter(s). Although recurrent motor axons made cholinergic (nicotinic) synapses on spinal Renshaw cells (Eccles et al. 1954), EPSPs and IPSPs in motoneurones were clearly not mediated by a peripheral humoral agent. So, in Eccles's department in Canberra, David Curtis, John Phillis (then a graduate student) and J. C. Watkins began to test (on spinal neurones) many possible excitatory or inhibitory substances, applied by iontophoresis from multibarrelled microelectrodes. Microiontophoresis had proved useful for very localized and/or transient drug applications at the muscle end-plate (Nastuk, 1953; del Castillo & Katz, 1957). For the CNS studies, micropipettes had five barrels: unit firing was recorded through the wider central channel; and various agents could be delivered by iontophoresis from outer barrels. Many such tests showed that spinal neurones are strongly inhibited by GABA and related ω-amino acids (Curtis et al. 1959) and just as strongly excited by glutamate and related amino acids (Curtis et al. 1960). For several reasons, however, the authors concluded that these agents could not be the synaptic transmitters. Thus, by the early 1960s, the identity of central transmitters seemed as elusive as ever (Burgen, 1964). That was the situation when we started experiments on cortical neurones (in cats). Much encouraged by the Babraham Institute's new director, John Gaddum, we tested all the peripheral transmitters, as well as the neuroactive amino acids. Much tedious work went into filling electrodes with the various solutions and, after allowing 1–2 days for diffusion into the tips, many hours were spent testing units in neocortex or cerebellum (the record was over 200 at one sitting). Before the end of John Phillis's 15 months at Babraham, we had data from over 4000 neurones (mostly in cats, but also some rabbits and monkeys, under various conditions of anaesthesia and even in non-anaesthetized ‘cerveaux isolés’). Many units were found at random, because they fired spontaneously or could be revealed by small applications of glutamate (a particularly useful procedure). We also tested motor cortex neurones identified by antidromic pyramidal stimulation; and chemically induced effects were compared with short-latency synaptic excitation or inhibition elicited via cortico-cortical (including transcallosal) or thalamocortical inputs. Fast transmitter-like effects were consistently seen only with amino acids: excitation with l-glutamate and inhibition with GABA. Though such close derivatives as cysteic and especially homocysteic were also powerful excitants, for several reasons l-glutamate seemed the likely physiological transmitter. As the most plentiful amino acid in the brain, it would never be in short supply. Its known very effective uptake (Stern et al. 1949) ensured rapid removal of synaptically released glutamate – without the need for degradation by a specific enzyme (the absence of such enzymes was a major argument against a transmitter role). Being coupled with Na+ influx and thus potentially electrogenic, the uptake might even cause excitation. The minimal effects of glutamine or d-glutamate suggested a selective receptor for l-glutamate. We were also impressed by glutamate's quick action – seen best with very brief iontophoretic pulses; the lack of obvious desensitization during long applications; as well l-glutamate's high potency: firing could be elicited by current pulses releasing only some 10 fmol (much less than the mean glutamate content of a cortical cell). All this made a strong case in favour of l-glutamate as the major transmitter of excitation. Or so it seemed to us. For others, however, its very abundance –in the cytosol or as a constituent of many proteins – went against what was expected of a transmitter. It was too well known as an important element in brain biochemistry – even as a food additive that enhances flavour (back to ‘soup physiology’). With no bioassay available, nor any ready-made pharmacology, the search for selective antagonists was long bedevilled by the existence of multiple receptors. Glutamate came to be accepted as the major transmitter only slowly, over the next few decades. For GABA, the climate of opinion was more favourable. It had been discovered in the brain only a decade earlier (Roberts & Frankel, 1950), had no obvious neurochemical role, and had already been identified as an inhibitory agent (Bazemore et al. 1957). The crayfish stretch receptor provided an excellent bioassay (Florey, 1954). In contrast to glutamate, for which antagonists had to be forged de novo, many known convulsants (including picrotoxin and bicuculline) were available for testing. For technical reasons, the increased Cl− conductance elicited by GABA and IPSPs could be recorded relatively easily. Parallel studies of GABAergic inhibition in crustacean muscle (Kuffler & Edwards, 1958; Kravitz et al. 1963) offered a useful model. Hence, GABA was accepted much sooner than glutamate as a major CNS transmitter. Our findings on the slow excitation by ACh opened a different perspective. A histochemical study of AChE-rich fibres (with Ann Silver) later revealed numerous putative cholinergic fibres travelling to cortex from the deepest region of the forebrain. Here was a potential cholinergic modulatory system, with a previously unsuspected origin at the base of the forebrain, which could raise neuronal excitability in the cortex and thus perhaps enhance cognitive function and wakefulness. We were fortunate in being new to the field and therefore not excessively inhibited by expectations. So we drew what turned out to be the right conclusions. Why were these significant? By proposing that glutamate and GABA could be major transmitters, they supported the shift in emphasis from electrical to chemical transmission; and they indicated specific targets for drugs acting, respectively, on inhibitory or excitatory pathways. Subsequent developments have led to the routine use of blockers of GABA or AMPA and NMDA receptors in a wonderful variety of studies. Such manipulations of synaptic transmission were quite inconceivable 40 years ago. The success of the chemical theory enormously enriched the possibilities of experimental research: it is difficult to envisage how electrical transmission could have lent itself to the current amazing expansion and flowering of brain science. The original classic paper reviewed in this article and published in The Journal of Physiology can be accessed online at: DOI: 10.1113/jphysiol.2005.096883http://jp.physoc.org/cgi/content/full/jphysiol.2005.096883/DC1 This material can also be found as part of the full-text HTML version available from http://www.blackwell-synergy.com Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.