Seeing Neurons: Golgi, Cajal and the Neuron Doctrine

Stain every cell in a slice of brain and you see almost nothing useful: a crowd of cell bodies, and between them a tangle of threads that no eye can follow. Stain only a few cells, but completely, and each one stands out with every branch.

That second kind of stain arrived in 1873, and it turned an old question into one that could be argued with evidence. Is the nervous system built of separate cells, like liver or skin, or is it one continuous network of fused fibres? The professor’s framing for it is “Neuron theory vs reticular theory.” It took about eighty years, a stain nobody fully understands, a lot of embryos and finally the electron microscope to decide.

The idea: to see one cell, stain almost none

The stains of the nineteenth century coloured what they reached. Take the stains developed by the German psychiatrist and neuropathologist Franz Nissl. A Nissl dye such as cresyl violet carries a positive charge, so it sticks to negatively charged RNA and DNA. Neurons pack their cell bodies with ribosome-rich membranes, and so a Nissl stain marks every cell body and nucleus in the section. It is excellent for counting cells and seeing layers. But axons and dendrites, the long thin processes where the question of fusion lives, stay almost invisible.

Camillo Golgi’s black reaction worked the other way. He hardened small blocks of nervous tissue in potassium dichromate for days, then moved them into silver nitrate. Where the two meet inside a cell, an insoluble salt crystallizes:

2 Ag⁺ + CrO₄²⁻ → Ag₂CrO₄ (silver chromate, which looks black in the cell)

The surprise is where it forms. Most cells stay pale. A few fill with the deposit from cell body to the finest twig, so in a thick section one neuron can be traced whole against a yellowish background. The professor calls this “Selective Labeling of a small group of neurons.”

“Selective” needs care. It does not mean Golgi could choose which cells were marked. The stained cells are a small, apparently random sample, and the fraction varies from piece to piece of tissue. Why one neuron takes the stain and its neighbour does not is still unknown. Think of a forest at night with lights strung on a handful of trees. If every tree were lit you would see a glow; with a few lit you see the shape of each lit tree, though you cannot pick which ones get lights.

Two readings of one slide

The reticular theory was older than Golgi’s stain. In the early 1870s the German anatomist Joseph von Gerlach, using other stains, proposed that the fine branches of nerve cells fuse into a continuous net. Golgi’s own reading differed in its details and changed over the years. In his preparations dendrites seemed to end freely, and he thought they mainly served nutrition. The thin branches of axons, by contrast, seemed to merge into a diffuse network through the grey matter. So his “reticulum” was a network of axon branches, and it would be wrong to say he thought every dendrite fused with the next.

The neuron theory said the opposite: each nerve cell is a separate cell, bounded by its own membrane, as the cell theory holds for every other tissue. A continuous membrane boundary is what makes a cell a separate unit (see that lesson), and separate compartments are what allow each one to be controlled on its own, the theme of that lesson. Under this theory, signals pass between cells at points of contact.

How could two careful observers look at similar preparations and disagree? The light microscope cannot separate two points closer than about 0.2 micrometres. Where two fine branches touch, or merely pass close at different depths in a section 100 micrometres or more thick, the image cannot tell contact from fusion. Each camp filled that blind spot with what its theory predicted. For Golgi, the lack of visible fusion meant only that the joins were too fine to see. For Cajal, the free endings he found over and over were the evidence, and an unseen connection was no evidence at all.

Santiago Ramón y Cajal came to the stain in 1887, when the Madrid histologist Luis Simarro showed him Golgi preparations. From 1888, working first on the cerebellum of birds, Cajal described axon branches that end close to other cells without joining them. He also changed where he looked.

Evidence from development: fibres grow out of single cells

An adult brain is a thicket. An embryo is a young wood: fewer cells, shorter branches, and little myelin, the fatty sheath that wraps many mature axons and can block the stain. Cajal turned to embryonic and young tissue because a whole cell fits in a section there and its branches can be counted.

He was not the first to argue from embryos. Wilhelm His cut embryos into complete, numbered series of sections and rebuilt their shapes in drawings and models. The neural tube, the hollow tube of tissue that becomes the brain and spinal cord, showed dividing cells next to its central cavity and young nerve cells farther out. In the late 1880s His argued that each nerve fibre grows out from a single cell, which he called a neuroblast. If a fibre is the outgrowth of one cell, it began as part of that cell and not as a strand of a net. Around the same time the Swiss psychiatrist Auguste Forel reached a similar conclusion from the way damage spread, or failed to spread, after injury.

In 1890, in the spinal cord of chick embryos, Cajal described swellings at the tips of young axons, some cone-shaped, some fringed with spikes. He named the structure the growth cone and read it as the advancing end of a fibre that lengthens behind it. Every one of his specimens was fixed and dead, so the motion was an inference from a series of still shapes; watching living growth cones belongs to a later lesson.

Cajal then added a rule about direction, which Arthur van Gehuchten in Belgium proposed independently. By the law of dynamic polarization, a signal enters a neuron through its dendrites and cell body and leaves through its axon. In the retina and the olfactory bulb, where it is obvious which end receives the outside world, the cells line up in exactly that order. A fused net has no preferred direction. A chain of separate, polarized cells does, which matches the professor’s point from the simple reflex circuit that the direction of information flow matters.

A prize shared by opponents, and what the electron microscope showed

In 1906 Golgi and Cajal shared the Nobel Prize in Physiology or Medicine “in recognition of their work on the structure of the nervous system.” By then most anatomists favoured separate cells, and the words had arrived: Wilhelm Waldeyer is usually credited with “neuron” in 1891, and Charles Sherrington with “synapse” in 1897. Golgi still titled his Nobel lecture “The neuron doctrine, theory and facts” and argued against it. He could hold out because no light microscope could show the gap.

The direct test came half a century later. In 1955, Eduardo De Robertis and H. Stanley Bennett used the electron microscope on synapses in frog sympathetic ganglia and earthworm nerve cord, and in 1956 Sanford Palay did the same in rat brain. Both reported that at a chemical synapse the membranes of two cells stay separate across a narrow cleft, 10 to 20 nanometres wide, with small vesicles crowded in the terminal. These match the professor’s “three signs of a synaptic specialization”: small vesicles in the presynaptic terminal, a narrow cleft, and a thickened postsynaptic membrane. Palay and George Palade had already reported vesicles in mammalian synapses at a 1954 meeting, so priority is hard to rank. The cleft is about ten times smaller than the light microscope’s limit, which is why it took a new instrument to show it.

There is one real exception to separateness. At an electrical synapse, channels made of proteins called connexins cross from one cell’s membrane to the next, and ions and small molecules pass straight through. These gap junctions join some neurons, especially in development. They still do not make a syncytium, a single cell with many nuclei. Each neuron keeps its own membrane, the gap junction’s two membranes are a few nanometres apart, the channels pass only molecules up to roughly one kilodalton, and they can close. Golgi’s network, in the sense he meant it, does not exist; coupling through regulated pores does.

Design problem

Suppose neurons were fused into a continuous network. What would you predict if you injected one neuron with a dye too large to pass through gap junctions, or if you cut one axon? Design one test that was available in 1900 and one available today, and say what each theory predicts.