by Christoph
Figure 1. Shadow sculpture by Rashad Alakbarov. Frontispiece: detail of this image. Source
I first saw the image to your right a while ago on Twitter (now 𝕏) and soon realized that it's found all over social media, not really "gone viral" but with 70 hits upon googling. I will put down a few thoughts about it here because it illustrates so well how microscopists, especially cryo-electron tomographers, work. And what distinguishes their way of working from that of artists, and where both meet.
How do you "see" the image and what's on/in it? I don't want to be intrusive, so I'll say how and what I see, you may follow it for yourself. Receptors in the retina of my eyes register incident light by hue and intensity, and the optic nerves of both eyes then add positional information as they transmit these signals to the brain. This is, in a nutshell, how I see. The neurons in the visual cortex of my brain then assemble this stream of signals into a three-dimensional image by matching it with previously stored images. This is what I consciously see. I don't need to dwell on the term "consciousness," but in principle visual perception works similarly in all vertebrates, humans included. A hallmark is the unconscious ability to recognize patterns in new images based on stored images. Thus, I "see" what I have seen before. This does not work error-free, just think of optical illusions. And when the perception of familiar patterns fails, a feedback loop can be consciously set in motion: "focus on this part" or "look at it from a different angle." Until a previously unknown pattern becomes a familiar one, and practicing that pattern, for example, by dancing, singing, talking or drawing, helps me with fine-tuning the visual memorization (neuroscientists would agree).
In this image, in the shadow on the wall, I instantly saw a repeating, regular pattern I'm familiar with, the central 12-pointed star, an islamic ornament. This because during my studies of biochemistry and microbiology I dabbled in such patterns that I had first seen in floor and wall mosaics in the cathedral of Cefalù, Sicily. All I needed to recreate such patterns was a sheet of paper, pencil, ruler, compass, and getting into Pythagoras' theorem (see here an example from Anna Maksimova's tutorial).
(click to enlarge)
Figure 2. A Top view of a chemoreceptor array in a Salmonella enterica minicell. OM, outer membrane; IM, inner membrane. Scale bar: 100 nm. Source. B Side view of a membrane-bound chemoreceptor array (MA) and a cytoplasmic chemoreceptor array (CA) of Vibrio cholerae. The cytoplasmic array is composed of two parallel CheA/W baseplates approximately 35 nm apart. The chemoreceptors are sandwiched between the two baseplates and are hexagonally packed with a 12 nm center-to-center spacing. OM, outer membrane; IM, inner membrane. Scale bar: 50 nm. Source
Knowing this pattern so well, I immediately noticed that the shadow is slightly distorted. Do you see that too? It becomes even clearer when you see other images of this artwork by the Azerbaijani artist Rashad Alakbarov, for example here and here. This distortion made me realize that the shadow came from a light source that the artist had placed behind a rather chaotic looking pile of mikado sticks (square-section metal tubes of varying lengths for this image, wooden sticks for the other two linked images, apparently). Could I reconstruct the three-dimensional shape of this stack by just looking at it from one angle? No way.
I have no idea if the artist intended an allusion to Plato's Allegory of the Cave with this work (my take on Plato: you can't perceive reality but only see its shadow). From the many comments on this image on "social" media, however, I have learned that esoterically inclined people like to understand it − and are comfortable with it − in the sense of the psychatrist and psychoanalyst C.G. Jung: "In all chaos there is a cosmos, in all disorder a secret order" (cit.). Microscopists can do little with such a definitive sounding blurb.
The actual work of the electron microscopists starts when they get to "see" such striking patterns as in Figure 2, an array of fairly regular hexagonal tiles in a ~100 nm wide "patch" in Figure 2A, and ~15 nm thick and ~100 nm long "sandwiches" that either localize in the cytoplasm or abut the inner cell membrane in Figure 2B. They know that these images are "shadow images," signals cast onto a sensitive digital screen by a focused electron beam directed through a super-thin three-dimensional object (the signals recorded by the sensor are, of course, numerical values that can only be viewed by converting them into pixels).
Figure 3. Principle of electron tomography. a Schematic representation of data acquisition. A flexible rope knot represents the object, emphasizing that electron tomography can retrieve 3D information from structures with individual topologies. A set of projection images is recorded on a charge-coupled device camera while the object is tilted incrementally around an axis perpendicular to the electron beam. Owing to the limited accuracy (eucentricity) of the tilting device, the specimen has to be recentred and refocused at each tilt angle. Automated procedures have been developed to perform this task with negligible exposure of the object to the electron beam. b The back-projection method explains the principle of the 3D reconstruction in an intuitive manner. For each projection, a back-projection body is calculated, and the sum of all projection bodies yields the density distribution of the original object − the tomogram. To compensate for the fact that high-resolution features change more rapidly with tilt angle than do low-resolution features, an appropriate weighting function has to be applied to the data in the 2D images before calculating the reconstruction. Source
In Figure 2, the "patch" is from Salmonella and the "sandwiches" are from Vibrio, but I still dare to show these, at first glance, completely different structures from different organisms in one figure. In about a dozen bacteria of distant phylogenetic relationships exactly these two structures were found in astonishingly similar dimensions. The measured dimensions of the "patches" and "sandwiches" suggested that they were different views of the same cell structure but this had to be proven first.
The researchers used two common techniques to make sure that their own pattern recognition capability was not playing tricks on them here. First, they obtained power spectra − technically comparable to X-ray crystallography − of Vibrio "patch" images showing that they have periodicities like crystalline structures (see here, top insert). Second, they used subvolume averaging to verify that the "patches" showing a hexagonal pattern in the top view (Figure 2A) are indeed hexagonal (see here, bottom insert).
Note that microscopists have no control over how their snap-frozen cells are "caught" in a sample, whether they rest on their sides or stand upright or at an angle, they know they find identical three-dimensional cellular structures in a wide variety of orientations.
An aside. There is no Wikipedia entry on "chemosensory arrays" in bacteria and archaea yet, but if you want to see and know more, I recommend you to take a look at the respective chapter of the Cellstructureatlas, which we already called a source of excellent teaching and learning material here and here.
To make a three-dimensional "scan" of relevant cellular structures possible, cryo-electron microscopists use a trick called "back projection." First, a series of images (=data) is taken of an object at different angles, that is, the object is tilted incrementally around an axis perpendicular to the electron beam (Figure 3). Second, for each projection, a back-projection body is computationally calculated, and the sum of all projection bodies yields the density distribution of the original object, the tomogram. As Sali et al. (2003) write: "The quality of a tomogram depends critically on covering as wide a tilt range as possible (typically ±70°), with tilt increments as small as possible (1° to 3°). However, each additional exposure to the beam increases the amount of radiation damage and the cumulative dose must not exceed a tolerable limit." That is, a limit of approximately 100 electrons/Å2.
(click to enlarge)
Figure 4. Architecture of Chemosensory Arrays in E. coli. (A) The side view and the top-down view of a signaling core unit is depicted in cartoon. The interface 1 within the core unit is marked with a red line. (B) Repeats of the signaling core units assemble into arrays through interface 2, marked in black lines. Red and black circles in broken lines show CheA-filled and CheA-empty rings in baseplate, respectively. (C) Tomographic image of the chemosensory arrays in the side view near a flagellar motor in a lysed E. coli cell. (D) Cryo-electron tomography (cryo-ET) image of chemosensory arrays in the top view in a lysed E. coli cell. The insert panel shows subtomogram averaging of the hexagonally packed receptor trimers-of-dimers. Source
Using such tomograms obtained from many hundreds of computational "aligned" individual images, Yang & Briegel (2019) were able to chart the substructures of the chemosensory arrays of E. coli at near atomic resolution and map their spatial arrangement (Figure 4). However, since electron density maps are hardly "readable" except for very trained eyes, the authors have represented them diagrammatically (Figure 4A+B). They explain: "The chemoreceptor homodimers readily form trimers-of-dimers through the interaction at their cytoplasmic tips (Figure 4A). To form the signaling core units, receptor trimers bind both the kinase and coupling protein following a strict stoichiometry of 6 receptor trimers-of-dimers: 1 dimeric CheA:2 monomeric CheWs (Figure 4B). Among the five domains of the kinase CheA, the P5 domain directly binds to the receptor trimers. The P5 domain topologically resembles two tandem SH3 domains. Similarly, CheW is composed of two β-barrels sandwiching a hydrophobic core. CheW and P5 bind each other in an alternating order and form pseudo sixfold symmetric rings that link core units together. Within the individual signaling core unit, CheW and P5 establish interface 1; among the signaling core units, CheW and P5 form interface 2, which is crucial for forming the extended array lattice. Through the networks of CheAs and CheWs, the receptor trimers-of-dimers are arranged into a rigid hexagonal pattern with a spacing of ~12 nm. In the side view, arrays are distinguishable as a continuous layer parallel to the inner membrane (Figure 4C). The chemoreceptors can be seen as pillar-like densities that extend between the baseplate and the membrane."
(click to enlarge)
Figure 5. Molecular architecture of bacterial chemosensory arrays. Red: chemoreceptor trimers-of-dimers, green: CheW adaptor, blue: CheA kinase, turqois: bacterial membrane (IM). Source. See here for the complete Figure: Construction and refinement of an atomistic model of the bacterial chemosensory array. X‑ray structures of the kinase CheA, adaptor protein CheW, and chemoreceptor from the thermophilic bacterium T. maritima were used to model key array substructures, which were arranged according to cryo-ET densities to produce models of the extended array architecture, namely the CheA2-trimer and CheA2-hexamer assemblies. Subsequently, molecular dynamics flexible fitting (MDFF) simulations with symmetry restraints were carried out to refine the structures of the component models to their array-bound conformations. Source
With the structural data of the baseplate components, the hexagons, and the intriguing trimers-of-dimers receptor oligomers in hand, or better, in files: would it be possible to build a complete three-dimensional model, the "pile of mikado sticks," of a chemosensory array at near atomic resolution? Goh et al. (2016) attempted and managed to do just that (Figure 5). Their computational methodology, outlined in detail in their review, is too complex to recapitulate here in brief, but see the legend to Figure 5. I will limit myself to showing their model for a chemosensory array of Thermotoga.
Annoyingly, this is where the impossibility of approximating the three-dimensionality of models on paper or a computer screen, that is, two-dimensionally, comes into play. I wish I could at least see this array animated, or, better yet, in a version I could actively zoom into and wander around like in Google Earth. Is that too much to ask? The data is there and Keith Cassidy has selected only one possible view for the image of the Thermotoga chemosensory array. But the effort of programming an animation of the image is easily comparable to Rashad Alakbarov's work of piling up the "Mikado sticks" by trial-and-error tinkering such that when illuminated they produce this distinct shadow of a 12-pointed star in Figure 1.
One last point. If you are missing clues from the Thermotoga model of the chemosensory array as to how it might be embedded in the cellular context, you can be helped. Goodsell & Lasker (2023) have modeled a membrane-bound chemosensory array into a Caulobacter cell, of course also based on structural data (see here).
Do you want to comment on this post? We would be happy about it! Please comment on Mastodon, Bluesky, or on 𝕏 (formerly Twitter).
Comments