Electron microscopy · Lobeworks/17

Electron microscopy is a family of imaging methods that form a picture with a beam of electrons instead of light, and because an accelerated electron's wavelength is about a hundred thousand times shorter than that of visible light, it resolves structures down to about a nanometre, far below the 200 nm limit of any lig

Electron microscopy. Electron microscopy is a family of imaging methods that form a picture with a beam of electrons instead of light, and because an accelerated electron's wavelength is about a hundred thousand times shorter than that of visible light, it resolves structures down to about a nanometre, far below the 200 nm limit of any light microscope.

A light microscope cannot separate two points closer than about half the wavelength of its light, which is why a synapse, with its 20 nm cleft and its 40 nm vesicles, is a blur under it. An electron accelerated through 120,000 volts behaves as a wave of about 3 picometres. No glass can bend it, so the lenses are coils whose magnetic fields curve the electrons' paths to a focus, and the whole column runs in vacuum. In a transmission microscope (TEM) the beam passes through a section a few tens of nanometres thick; in a scanning one (SEM) it sweeps the surface of a block and the electrons that bounce back are counted.

Contrast comes from heavy metals. Tissue is fixed, stained with osmium, which binds to membranes, and with uranyl salts, then embedded in resin: membranes show dark, the inside of cells pale, and vesicles, mitochondria and the dense bodies of synapses become visible.

Electrons become light at the end. Below the section a scintillator glows where electrons land, and cameras film the glow; the fly brain of 2018 was filmed by a two-by-two array of scientific CMOS cameras at about 50 megapixels a second.

It is how every connectome is made. Serial sections cut by an ultramicrotome, or faces exposed by FIB-SEM, are imaged one after another and stacked into a volume.

It sees shapes only. A micrograph has no colour and no label, shows no activity, and does not show reliably which transmitter a synapse releases or where the tiny gap junctions are.

Electron microscopy sees all of a brain's structure and none of its activity.

The sample has to be dead, fixed and in vacuum, so everything a neuron was doing is gone before the first image is taken.

Questions: How does a flood-filling network get past a gap in a membrane without leaking? It is never asked whether a voxel is membrane in general; it is asked whether the voxel belongs to the object it is already tracing. Its input is the image and its own current guess of the object, so at a gap where the stain failed it can see that the shape it has been following continues on one side and stays there. A method that first marks membranes and then floods the regions between them has no such memory, and pours through the hole into the neighbouring cell. What does neuropil look like under an electron microscope? Like the cut end of a bundle of spaghetti of many thicknesses. Each strand is a neurite cut across, outlined in dark membrane, some carrying vesicles and mitochondria, and the dark specks among them are synapses. There are almost no cell bodies: in an insect they sit in a rind outside the neuropil, which is why a plot of every synapse in a brain draws its neuropils. Why are gap junctions missing from the fly connectome? Because at the resolution and staining used for whole fly brains they cannot be identified reliably. A gap junction joins two membranes across a gap of a few nanometres, and in a micrograph at 4 to 8 nm per pixel it looks like two membranes pressed together. The worm connectome of 1986 did map them, from finer images of a much smaller animal; the fly connectomes record chemical synapses only, so any current that flows directly between fly neurons is absent from their models. Why can a light microscope not map synapses? Because light cannot separate points closer than about 200 nanometres, its diffraction limit, and a synaptic cleft is about 20 nm wide and a vesicle about 40 nm. An electron accelerated through 120,000 volts behaves as a wave of about 3 picometres, so an electron microscope resolves around a nanometre, enough to follow every membrane and count vesicles. Why does osmium make membranes visible to an electron microscope? Because it binds the lipids of membranes and leaves heavy atoms there. Electrons passing through a section are scattered by heavy nuclei and pass almost straight through light ones, so wherever osmium sits the image is dark. Without it the membranes, which are the outlines every segmentation follows, would be nearly invisible. Why must a section be so thin for an electron microscope? Because the electrons of a transmission microscope have to pass through it, and in thick material they scatter again and again until the image is a blur. Sections of a few tens of nanometres let most electrons through, scattered once by the heavy metal on membranes and straight elsewhere, which is the contrast the image is made of. The thinness also sets the depth resolution: the fly brain of 2018 was cut into 7,062 sections of about 40 nm, so each image averages 40 nm of tissue.