(Early evolution of neurons - Current Biology Magazine, William B. Kristan, Jr. October 24, 2016). Early evolution of neurons How did a structure as complex as our own brain ever evolve? Although biologists have pondered this question since Charles Darwin, the explosion of molecular information in recent years has provided new insights into this question, particularly its first step: the evolution of neurons. Meshing information about genomes with insights from more classical anatomical, physiological, and developmental approaches has led to some remarkable insights and surprises. Because ‘phylogenomics’ is still a young fi eld, however, there are arguments about which genes to include in comparisons, how much to weigh genetic versus ‘classical’ features, and which algorithms to use in making such comparisons. One source of serious discussion is the explanation for a feature being present in one clade (a group of animals with a common ancestor) but absent in a second clade. Does the feature’s absence in clade 2 mean that the feature was never present in the ancestors of clade 2, or was it present in clade 2’s ancestors but subsequently lost? A second phylogenomic problem is posed by convergent evolution (or ‘homoplasy’ in genetic terminology): a feature or a molecule that is present in two clades might have evolved independently in each clade. Both of these problems, secondary loss and homoplasy, confound the interpretation of evolutionary relationships. For the moment, the only solution to these problems is to compare more genes in more animals to see whether the features that are missing from one species, for instance, can be found in other closely-related species. The purpose of this primer is not to consider the evolution of brains, however, but the more modest goal of determining the evolution of neurons, the information processing cells that compose brains. Even this more limited goal is, at this juncture, beyond our reach, but the journey to this goal has already uncovered some remarkable relationships and has made clearer what are the key questions and how they can be approached. What is a neuron? Leaving aside the sticky phylogenetic problems for the moment, let us proceed to seemingly more solid ground: what are the features that determine that a given cell is a neuron? Such a definition seems, at first pass, to be simple enough: neurons have long processes, generate action potentials, receive input from other neurons (or, if they are sensory neurons, from appropriate stimuli), and provide output to other cells (for example, neurons, muscles, glands) via synapses. Unfortunately, this ground is also not so solid. For one thing, not every bona fi de neuron has every one of these features. For instance, some perfectly good neurons have no processes, some vertebrate neurons do not generate action potentials, and some small (less than a millimeter in any dimension) invertebrate animals get along just fine without fast action potentials in any of their neurons. Furthermore, many neuron-like features are found in cells that are readily identifiable as muscle, gland, or even epithelial cells. For instance, the epithelial cells of some jellyfish generate action potentials that radiate out to other epithelial cells via gap junctions. In fact, even single-celled organisms generate action potentials, and non-animal eukaryotes (such as choanoflagellates and other protists) and even some prokaryotes (bacteria) communicate with other members of their species by releasing chemicals into their watery environment to elicit responses in conspecific cells. Even anatomical features are not sacrosanct; for instance, each muscle cell in the nematode worm Caenorhabditiselegans has a long slender process that connects to its motor neuron, rather than vice versa, and some synapses in vertebrates manage to function without such defining features as presynaptic vesicles. An alternative approach to defi ning a cell as a neuron is to determine whether it contains ‘neuron-specifi c molecules’: molecules that are found exclusively in cells that are clearly neurons. Among the molecules used for this purpose have been those associated with neuronal function, including voltage-sensitive channels, synapse-specific proteins - both presynaptic and postsynaptic - as well as neuronal morphogens responsible for the specifi cation of neurons during development. This approach also has its diffi culties, because all of these molecules are also found in cells that, by most definitions, are not neurons. Even single-celled organisms, like choanoflagellates, protozoa, and bacteria, have homologs of many of these seemingly neuronspecifi c molecules. The phylogenetic relationships of gap junction proteins (innexins, panexins, and connexins) have also been studied, but they are even less specifi cally expressed in neurons than the other molecules being used, so these proteins are less useful in tracking neuronal evolution. The usual working definition of a neuron is a cell that transmits information from one cell (or from a stimulus) to one or many other cells via synapses. Practical useful markers for neurons include their morphology - having long, thin processes - and their expression of voltage-gated channels, synaptic molecules, and neuron-specifi c developmental molecules. Granting that these are less than perfect descriptors, these criteria are widely used to paint a broad brush-stroke picture of the evolutionary origins of neurons. To establish a context for such a discussion, it is useful to consider the outlines of the origins of the major animal clades. Early evolution of the major animal phyla Phylogenetics is critical for determining the early stages of animal evolution because they occurred before the appearance of recognizable animal fossils (Figure 1). Starting long after the time that the eukaryotes (plants, animals, fungi and protists) separated from the prokaryotes (bacteria and archaea) more than a billion years ago (some say 4 billion), multicellular organisms formed and separated into the five major animal groups - the clades Ctenophora, Poriphera, Placozoa, Cnidaria, and the more than two dozen phyla that make up the Bilateria, which includes chordates such as humans - by the time the fi rst trace of fossil animals appeared, at about 600 million years ago. A variety of explanations have been proposed for the lack of animal fossils during this long period: the inability to fossilize (for example due to a lack of hard structures); low density of organisms due to inhospitable environments; or the possibility that the animals were not only soft-bodied, but also microscopically small. The environment was certainly made inhospitable by a major glaciation that blanketed much of the earth before 600 million years ago, which would have restricted both the habitable environment and limited the earth’s supply of oxygen. As the glaciers receded and the environment warmed, oxygen became more available and the animals became larger and more abundant. Starting in the Precambrian Period, the fossil record gives hints about the evolution of nervous systems and behaviors. The earliest fossilized animals (560-550 million years ago) were probably sessile fi lter feeders or grazers. Trails left by the early grazers were straight and simple, but they became more circuitous in later times (550-540 million years ago), and finally showed signs of digging into the substratum by the beginning of the ‘Cambrian explosion’ of fossils (~540 million years ago). These trails disappeared by 525 million years ago and were replaced by animals with hard coverings shaped into a wide variety of spikes, shells, and plates. The rich array of external armor and weapons in the fossil record strongly suggests that animals started to prey upon each other. The larger size of these animals put a premium on keeping different parts of the body coordinated, and their predatory behavior favored animals capable of making quick movements to obtain food, and to avoid becoming someone else’s food. Both demands favored the evolution of a fast-conducting system like neurons. The first clear indication of nervous tissue was the appearance of well-formed eyes and faint outlines of nervous systems in fossils from ~525 million years ago. At a molecular level, many of the ‘neuron-specific molecules’ (voltagegated channels, molecules that form synaptic structures) were already present in all major animal clades before the earliest fossils. Even some bacteria have genes homologous to those making these molecules, which means that these genes were present in the common prokaryotic/eukaryotic ancestor, which could have been as long as 4 billion years ago. The functions of all these genes in single-celled organisms is not known, but a reasonable guess is that voltage-gated channels functioned to regulate intracellularions and water of the ancestral prokaryotes, to keep them from bursting in the hypotonic water that was their environment, and were only later specialized for communication. In fact, modern bacteria use voltage-gated potassium channels to communicate the presence of metabolites to their companion bacteria that have formed a biofilm, a grouping of many bacterial cells. The K+ ions released by the activated bacteria depolarize nearby cells, which activates their voltage-gated K channels (Kv) causing them to release their own K+ ions, a process that propagates across the biofilm.

Evolution of voltage-gated channels Voltage-gated channels (Nav, Kv, Cav) are composed of homologous subunits, with Kv channels formed from a tetramer of a single subunit, whereas both Nav and Cav channels are monomers encoded by genes generated by a double duplication of the Kv gene, with modifi cations of the ion channels to make them selective either for Ca2+ or for Na+ ions. Kv channels were probably the only voltage-gated channels in the earliest animals and were used to regulate cell volume by changing the ionic content of the cell in response to cell membrane stretch. Cav channels probably appeared next, as a way to control the internal metabolic state of the cell, and in later organisms, to regulate the beating of cilia and the contraction of muscles. Cells with the proper combination of Cav and Kv channels could then generate action potentials, which expanded the cellular capabilities in many ways. With action potentials already possible, what was the selective advantage of adding Nav channels? One possibility is that cells could then use Cav channels for other purposes, like releasing transmitters or to avoid the build-up of intracellular Ca2+ to a toxic concentration. An alternative explanation is that, because Na-dependent action potentials are shorter in duration and conduct more rapidly than Ca-dependent ones, Nav channels were selected only later, when predation made rapid movements become increasingly benefi cial. Making action potentials shorter in duration may have been augmented by the evolution of Kv channels that had faster kinetics, which would quickly turn off the fast depolarization caused by Nav channels and make behaviors as fast as possible. Evolution of synaptic transmission Surprisingly, single-celled organisms not only have voltage-gated channels but also have many of the genes for presumed synapse-specifi c molecules, such as enzymes for producing and releasing transmitters and structural proteins that produce postsynaptic responses to the transmitter. Based upon both electrophysiological and phylogenetic studies of cnidarians and ctenophores, the earliest type of true chemical synapses were likely to be peptidergic. A possible precursor of synaptic transmission is found in choanofl agellates, a clade of more than 125 species that are unicellular, although the cells of some species aggregate to form colonies. In these colonies, the cells move water past the colony by beating their flagella. Each of these cells can release transmitters that act on receptors in nearby cells to produce movements of the whole colony. This proto-hormonal capability is even more organized in some sponges. A sponge takes water in through many openings (‘pore canals’, from whence Porifera received its name), pushes it through channels lined by cells (called ‘choanocytes’ because they are similar to the unicellular choanoflagellates) with beating flagella that force the water into a large central cavity, from which it exits through the osculum (‘little mouth’). A strong mechanical stimulation to the body causes cells lining the channels to release transmitters, including glutamate, GABA and nitric oxide (NO), which are carried by the water to cause coordinated contractions of the muscles in the body wall and osculum. In effect, sponges use these transmitters as hormones, with flowing water taking the part of blood in our own endocrine system, using many neuron-like molecules for this purpose. Placozoa, a separate clade with just a single species, Tricoplax, is a small, dome-shaped animal with just six different cell types, with which they glide over the substrate on cilia, detect food (single-celled algae), trap the food (their body forms a dome over it), digest the food (digestive enzymes are secreted onto the food), and absorb the digested food. Remarkably, Tricoplax has no cells that resemble either neurons or muscles, but it does have many genes used to make voltage-gated channels, presynaptic and postsynaptic structures, and even transmitters. Gland cells around its edges, for instance, contain the peptide FMRFamide, probably to coordinate its feeding movements. Single-celled organisms and small animals survived well using only Kv and Cav channels, but larger and faster animals evolved during the period when Nav channels were being established. In all cases, the transmitter receptors found in sponges, choanofl agellates, and bacteria are metabotropic: binding the signaling molecule to the receptor activates a cascade of intracellular pathways in the target cells. This mode of transmission produces slow, prolonged movements. The next evolutionary step was the progressive specialization of cell types, including neurons. The origin(s) of neurons It is a reasonable guess that most neurons in most animals are evolutionarily derived from epithelial cells. Evidence in support of this notion includes the fact that epithelial cells in jellyfish generate action potentials that conduct from cellto-cell across gap junctions, and that neurons in triploblastic animals (most Bilateria) are derived from the ectodermal layer, the same layer that produces epithelial cells. As usual, there are exceptions to this generality. Some neurons in cnidarians (the clade that includes the jellyfi sh), for instance, appear to be endodermal in origin, and there are specialized myoepithelial cells in some cnidarians that have both mechanosensory and contractile properties. From sampling their extant species, the Bilateria, Ctenophora, and Cnidaria all have neurons, whereas Placozoa and Porifera do not. The significance of this neuronal distribution depends upon the relationships among these five clades, indicated in this figure as a coarse phylogenetic tree. Both morphological and molecular features are consistent with the relationships among four of the groups (Bilateria, Cnidaria, Placozoa, and Porifera), though the placement of Ctenophora (comb jellies) is in dispute. Classical features (comparative anatomy, development), along with some molecular data, indicate that ctenophorans and cnidarians are sister groups, closer to the bilaterians than are the placozoans. More extensive molecular studies, however, suggest that Ctenophora is the most basal of the five clades, splitting off from the metazoan (animal) line even before Porifera did so. This uncertainty in placing Ctenophora is indicated in Figure 1 by the dashed line connecting ctenophores to the tree. If the ctenophores are truly basal to the sponges, then even the simplest origin of neurons would have two plausible scenarios that could explain this phylogenetic tree. The first possibility is that the common ancestor to all major animal clades had neurons but that Porifera and Placozoa lost them early in evolution. The second possibility is that neurons have two independent origins, one in the ctenophoran ancestor and a second in the shared ancestor to the cnidarian and bilaterian branches, so that the placozoan and poriferan lineages need never have had neurons. If the classical phylogenetic tree turns out to be accurate, however, then neurons could have had a single origin in an organism that gave rise to the cnidarians, ctenophores, and bilaterians. In this scheme, the placozoan and poriferan lineages would not have lost neurons but rather never had any. Working out the details of these early events is currently a hot research topic that will likely be resolved by genomic analysis of more species in each of the clades. Just having voltage-gated channels and synaptic molecules clearly does not automatically make a cell into a neuron. Minimally, these molecules need to be made in an appropriate number, then moved to the proper location and inserted into the cell membrane. To infl uence cells at a distance, long and thin processes need to be fashioned, and the terminals of these processes must be lined up with the correct locations on their synaptic partners. None of this anatomical detail can be extracted from the early fossil record, of course, and the molecular data are mixed. Molecules related to neuronal development, axonal outgrowth, and synapse formation have been investigated as markers for neurons. These molecules parallel the evolution of voltage-gated channels, expanding significantly in the Cambrian Period and beyond, which is a promising result. The difficulty in using them as neuronal markers, however, is that sponges and Trichoplax - two animal phyla without neurons - have homologs of these molecules. In fact, singlecelled organisms (protozoa and even bacteria) have homologs of neuronal developmental genes. The functions of these molecules in organisms without neurons is unknown, but their presence in these neuron-less animals undercuts their usefulness as neuronal identifiers.

Small animals, such as the placozoan Trichoplax, or the presumed precursors of the major phyla, could coordinate their behavior through connections among epithelial cells. As the animals got larger and faster, however, there would be advantage to having some of these epithelial cells be specialized for rapid conduction. In this scenario, the original nervous system would be composed of a net of electrically connected ‘proto-neurons’. These proto-neurons might all have served a sensory function, probably mechanosensory, responding to a stimulus at any site on the body, spreading out in all directions. The first true chemical synapses were probably neuromuscular, as a short-distance specialization of the neuroendocrinelike interactions used by extant choanofl agellates and sponges. Interestingly, muscle cells appeared to have evolved in parallel with neurons, with smooth muscles appearing before striated muscles. The contractile apparatus of smooth muscles uses many of the same molecules as used by striated muscles (actin and myosin, for example) but lacks the specialized mechanisms (for example, t-tubules and troponin) that allow striated muscles to contract quickly. Compared to striated muscles, smooth muscles generate more tension for a given amount of energy expended and can contract over a very large range of cell lengths, a necessity for soft-bodied animals. As animals evolved rigid structures (for predation and protection initially, and ultimately to use as skeletons) and became speedier, the muscles, too, became faster, albeit at the expense of increased energy expenditure. Once the cellular mechanisms had evolved to make chemical synapses, one can imagine that neurons began making synapses with one another, so that some of them could be specialized to accept input from other neurons rather than from outside stimuli; i.e., these neurons became interneurons. As the predator-prey competition ramped up, there would be advantages to being able to sense both food and predators in more ways, particularly at some distance. One can imagine that detecting chemical gradients could use the molecular tools available to the early multicellular animals, followed by sensation at a distance, such as vision (well-formed fossil eyes are found at 525 million years ago) and substrate vibrations. Having interneurons would allow both efficiency (for example, a single interneuron could sense different modalities of input from one location, rather than having different interneurons for each modality) and flexibility (for example, input from one location could be ignored if a stronger or more important input came in from another location). Although consistent with the scant data available, the interpretations in this section are highly speculative. Barring some dramatic new fossil finds from well before the Cambrian Period, these speculations can be tested only by molecular characterizations of many more species in all the major animal phyla. In addition, there needs to be anatomical, pharmacological, and physiological studies to localize the expression of molecules that can best distinguish neurons of different sorts. Simply having a particular set of molecules expressed in a given animal does not determine how these molecules function, as will be documented in the next section.

Examples of surprising neuronal functions Electrophysiological studies on poriferans, ctenophorans, and especially cnidarians have shown some remarkable features about how neuron-like cells control behaviors in these animals. All of them have rich complements of voltage-gated channels, and ctenophores and cnidarians have well-defi ned nervous systems, with a variety of sensory neuron types, interneurons, and motor neurons. Typical excitatory chemical synaptic potentials are recorded between neurons and from neurons to muscle cells. Interestingly, these phyla appear to coordinate their behaviors by modifying their voltage-gated channel properties, rather than by variations in their synaptic properties, which is the more common strategy used by bilaterian nervous systems. Two examples, both from experiments on cnidarian jellyfish, give a flavor for how such coordination strategies work. The first example is from studies of the bell jelly, Polyorchis, which, like all jellyfi sh, is radially symmetric and contracts all the muscles in its body (the ‘bell’) simultaneously both to swim and to escape from a stimulus directed at any site on the bell. For the contraction to be effective, all the muscles in the bell need to contract simultaneously. Because action potentials in these motor neuronal axons conduct slowly, however, contractions at the site of stimulation would occur earlier than at distant sites, unless there were some compensatory mechanism - which there is. Action potentials originating at the site of stimulation are slower to rise and are prolonged (they are carried mostly by Cav channels), whereas the action potentials become shorter in duration and conduct more rapidly as the action potentials move along the axon, because they are carried mostly by Nav channels. As a consequence of this variation in action potential shape and conduction velocity, the muscles all around the bell receive a bolus of transmitter at nearly the same time so that they contract nearly simultaneously. The second example of behaviorally significant variations in voltage-gated channels is from studies of the pink helmet jelly, Aglantha. These jellyfish have two modes of swimming: a slow, weak, rhythmic contraction of the whole bell that moves the body rhythmically in a bell-first direction, trailing its tentacles as it feeds, and a rapid, much stronger contraction that propels the animal in the same direction but much more quickly in response to a strong mechanical stimulus. Remarkably, both the weak and strong contractions result from single action potentials in the same giant motor axons. These axons have the remarkable ability to generate action potentials using either Nav or Cav channels. When the motor neurons are activated by rhythm-generating interneurons, the depolarizations are just large enough to activate low-threshold Cav channels, which sets up a Ca2+- dependent action potential that propagates in the motor neuronal axons; these action potentials cause weak contractions of muscles all around the body. Strong sensory input, on the other hand, provides much larger excitatory synaptic input that activates higher-threshold Nav channels, producing larger, faster action potentials in these same axons; these action potentials release more transmitter and produce much stronger muscle contractions. In this way, the same motor neurons mediate two different behaviors by activating two different voltage-gated channels. Closing thoughts Three kinds of gated channels probably evolved independently: voltage-gated channels, stretch-gated channels, and ligand-gated channels (probably used by animals initially for finding algae and bacteria by sensing their exuded chemicals). Because the oldest cells had only Kv channels, they could not generate action potentials, although they might have been able to produce a primitive form of propagation of activity, as in bacteria (see the section ‘Early evolution of the major animal phyla’ above). Although initially used for other purposes, voltage-gated channels are largely used to produce action potentials; stretch-sensitive channels almost certainly underlie mechanosensation; and ligand-gated channels were incorporated into endocrine cells and postsynaptic structures in neurons. Essentially every study of early evolution calls for making comparisons of more genes in more animals, to resolve critical issues like the timelines for the splitting of evolutionary branches and the definitive appearance of neurons. In addition, studies aimed at finding the functions of different gene products will give a better idea of the significance of gene duplications and modifications. In addition to the technical problems in using phylogenomics to study the evolution of neurons (see opening paragraphs above), there is a biological one: the fact that the same molecule can be used for different functions in different animals complicates the use of molecular homologies for defining cellular function. To understand the full importance of the evolution of molecules will require a complex interplay between molecular, anatomical, and functional studies, aided by computational modeling. The biophysical properties of the jellyfish neurons (see ‘Examples of surprising neuronal functions’), for instance, could not have been predicted by a molecular description of the voltage-gated channels that they express without sophisticated electrophysiological studies. Did neurons evolve more than once? Almost certainly. Even if ctenophores and cnidarians are sister groups, with a neuron-carrying ancestor, some cnidarian neurons derive from endodermal cells rather than from epidermal cells, as is the norm, and the same epitheliomuscular cells in Hydra, a cnidarian, can be transformed into neurons by perturbing neurogenesis. It may turn out that, once a cell type has acquired the basic molecular constituents to make action potentials and synapses, making these cells into neurons was relatively easy. In this case, neurons may have been gained and lost regularly in the evolution of the various animal clades. It is rather appealing to consider that our most lofty thoughts and aspirations are being produced by cells with their origins in many different kinds of tissues from a panoply of animals. The multiple origins of neurons may, if fact, be why defining ‘neuron’ is so difficult, and why defining the origin of neurons is so complex. On an even broader scale, it is interesting that some lineages (Cnidaria, for example) showed relatively little morphological change in the last 500 million years, whereas the Bilateria has shown enormous radiation into multiple phyla, varying both in size and body structure. Fossil Cambrian jellyfish are similar to the ones swimming in our oceans today, but there is little similarity between the bodies of earthworms and tigers, except for their bilaterally symmetric body form. Despite their huge differences in evolutionary history, bilaterians and cnidarians have been evolving for about the same length of time. There is a tendency to conclude from such comparisons that jellyfish have gotten stuck in an evolutionary dead end, in which the particular combination of cellular and molecular properties have stifled their progress toward more advantageous shapes and functions. In fact, the proximal goals of evolution are survival and reproduction, not the proliferation of species. Jellyfish (and other Cnidaria, plus Ctenophora and Placozoa) have taken a different path to these goals. What jellyfi sh may have been doing since the Cambrian is refining their molecular and cellular resources so that they were able to withstand such ravages as several ice ages, significant rearrangements of the continents, and a meteor collision that wiped out many animal species on the planet. In fact, jellyfish are greatly increasing in abundance as their predators (mainly fi sh) diminish and the ocean water warms and becomes more acidic, so their survival mechanisms seem to have made them capable of thriving during the current changes in their environment. It is interesting to speculate what the animal life on our planet will be like in another 100 million years.

რა არის ის თვისებები, რომლებიც განსაზღვრავს, რომ მოცემული უჯრედი არის ნეირონი? ასეთი განმარტება თავიდანვე საკმაოდ მარტივი ჩანს: ნეირონებს აქვთ ხანგრძლივი პროცესები, წარმოქმნიან მოქმედების პოტენციალს, იღებენ ინფორმაციას სხვა ნეირონებისგან (ან, თუ ისინი სენსორული ნეირონები არიან, შესაბამისი სტიმულიდან) და აწვდიან გამომავალს სხვა უჯრედებს (მაგ. ნეირონები, კუნთები, ჯირკვლები) სინაფსების მეშვეობით. სამწუხაროდ, ეს ახსნაც არც ისე მყარია, ვინაიდან ყველა ნეირონს არ აქვს ყველა ეს თვისება. მაგალითად, ზოგიერთ სრულყოფილ ნეირონსაც კი არ აქვს ეს პროცესები, ხერხემლიანი ცხოველების ზოგიერთი ნეირონიც არ წარმოქმნის მოქმედების პოტენციალს და ზოგიერთი პატარა (მილიმეტრზე ნაკლები ნებისმიერ განზომილებაში) უხერხემლო ცხოველი კარგად ახერხებს ცხოვრებას მიუხედავად იმისა, რომ სწრაფი მოქმედების პოტენციალი არ გააჩნიათ თავიანთ ნებისმიერ ნეირონში. გარდა ამისა, ნეირონის მსგავსი მრავალი თვისება გვხვდება უჯრედებში, რომლებიც ადვილად იდენტიფიცირებადია, როგორც კუნთების, ჯირკვლების (ენდოკრინული) ან თუნდაც ეპითელური უჯრედები. მაგალითად, ზოგიერთი მედუზას ეპითელური უჯრედები წარმოქმნის მოქმედების პოტენციალებს, რომელიც გავლენას ახდენს სხვა ეპითელურ უჯრედებზე. სინამდვილეში, ერთუჯრედიანი ორგანიზმებიც კი წარმოქმნიან მოქმედების პოტენციალებს, და არაცხოველური ევკარიოტები (როგორიცაა ქოანოფლაგელატები და სხვა პროტისტები) და ზოგიერთი პროკარიოტიც კი (ბაქტერიები) ურთიერთობენ თავიანთი სახეობის სხვა წევრებთან ქიმიკატების გამოთავისუფლებით თავიანთ წყლიან გარემოში, რათა გამოიწვიონ რეაგირება შესაბამის უჯრედებში. ანატომიური თვისებებიც კი არ არის საკრალური; მაგალითად, ნემატოდის ჭიის Caenorhabditiselegans-ის თითოეულ კუნთოვან უჯრედს აქვს გრძელი წვრილი წანაზარდი, რომელიც უკავშირდება მის მოტორულ ნეირონს და არა პირიქით, და ხერხემლიანებში ზოგიერთი სინაფსი ახერხებს ფუნქციონირებას ისეთი განმსაზღვრელი მახასიათებლების გარეშე, როგორიცაა პრესინაფსური ვეზიკულები. უჯრედის ნეირონად განსაზღვრის ალტერნატიული მიდგომა არის იმის დადგენა, შეიცავს თუ არა ის „ნეირონისთვის სპეციფიკურ c მოლეკულებს“: მოლეკულებს, რომლებიც გვხვდება მხოლოდ უჯრედებში, რომლებიც აშკარად ნეირონები არიან. ამ მიზნით გამოყენებულ მოლეკულებს შორის იყო ნეირონების ფუნქციასთან დაკავშირებული მოლეკულები, მათ შორის ძაბვისადმი მგრძნობიარე არხები, სინაფსის-სპეციფიკური ცილები - როგორც პრესინაფსური, ასევე პოსტსინაფსური - ასევე ნეირონების მორფოგენები, რომლებიც პასუხისმგებელნი არიან განვითარების დროს ნეირონების სპეციფიკაციაზე. ამ მიდგომას ასევე აქვს თავისი სირთულეები, რადგან ყველა ეს მოლეკულა ასევე გვხვდება უჯრედებში, რომლებიც, უმეტესი განმარტებით, არ არიან ნეირონები. ერთუჯრედიან ორგანიზმებსაც კი, როგორიცაა ქოანოფლაგელატები, პროტოზოები და ბაქტერიები, აქვთ მრავალი ამ ერთი შეხედვით ნეირონსპეციფიკური c მოლეკულის ჰომოლოგები. what are the features that determine that a given cell is a neuron? Such a definition seems, at first pass, to be simple enough: neurons have long processes, generate action potentials, receive input from other neurons (or, if they are sensory neurons, from appropriate stimuli), and provide output to other cells (for example, neurons, muscles, glands) via synapses. Unfortunately, this ground is also not so solid. For one thing, not every bona fi de neuron has every one of these features. For instance, some perfectly good neurons have no processes, some vertebrate neurons do not generate action potentials, and some small (less than a millimeter in any dimension) invertebrate animals get along just fine without fast action potentials in any of their neurons. Furthermore, many neuron-like features are found in cells that are readily identifiable as muscle, gland, or even epithelial cells. For instance, the epithelial cells of some jellyfish generate action potentials that radiate out to other epithelial cells via gap junctions. In fact, even single-celled organisms generate action potentials, and non-animal eukaryotes (such as choanoflagellates and other protists) and even some prokaryotes (bacteria) communicate with other members of their species by releasing chemicals into their watery environment to elicit responses in conspecific cells. Even anatomical features are not sacrosanct; for instance, each muscle cell in the nematode worm Caenorhabditiselegans has a long slender process that connects to its motor neuron, rather than vice versa, and some synapses in vertebrates manage to function without such defining features as presynaptic vesicles. An alternative approach to defi ning a cell as a neuron is to determine whether it contains ‘neuron-specifi c molecules’: molecules that are found exclusively in cells that are clearly neurons. Among the molecules used for this purpose have been those associated with neuronal function, including voltage-sensitive channels, synapse-specific proteins - both presynaptic and postsynaptic - as well as neuronal morphogens responsible for the specifi cation of neurons during development. This approach also has its diffi culties, because all of these molecules are also found in cells that, by most definitions, are not neurons. Even single-celled organisms, like choanoflagellates, protozoa, and bacteria, have homologs of many of these seemingly neuronspecifi c molecules.