by Christoph
(click to enlarge for full image)
Figure 8.1. Photomicrographs of Hyphomicrobium sp. strain ZV580 swarmer cells at various times during rosette formation. Bar, 10 µm. Source. Frontispiece: fluorescent image of rosettes of Caulobacter crescentus cells. Chromosomal DNA is labeled in blue (DAPI), cell membranes are in red (FM4-64). The fluorescently labeled protein is DivJ, a histidine kinase that localized at the base of the stalk (yellow). By Jennifer Heinritz, Jacobs-Wagner lab, Yale MSI. Cropped from Source
Among the Alphaproteobacteria, a number of species from phylogenetically distantly related orders are known for forming rosettes during growth. Despite commonalities, they all have their own peculiarities when it comes to the first cell‑cell contact(s) during "rosetting". Phaeobacter inhibens (order Rhodobacterales) cells attach to each other by their "sticky ends" at one pole ("fibrils"), and preferably at the air/liquid interphase (part 1). Caulobacter crescentus (order Caulobacterales; correct scientific name: Caulobacter vibrioides) cells attach to each other by their mega-sticky holdfast at the tip of their stalks wherever they encounter each other, or otherwise to almost any imaginable surface except their own cell bodies (below).
Hyhomicrobium sp. rosettes
Hyphomicrobium sp. strain ZV580 (order Hyphomicrobiales) cells start "rosetting" at the air/liquid interphase like Phaeobacter, or upon attachment to cover slips or glass slides for microscopy, for example. Their prosthecae (stalks) point not inwards as in Caulobacter but outwards, as they serve as "birth canal" for budding daughter cells, one at a time. The long stalks of Hyphomicrobium with still-attached daughter cells are visible in Figure 8.1.
Rosette formation by Hyphomicrobium occurs among virtually all cells at the air/liquid interface after they have shed their flagella. Moore & Marshall (1981) found in a growing culture that: "At 2 h there was an average of three cells per rosette. Although the majority of single cells were still motile at this time, no motility was observed in the rosettes. By 10 h, the average number of cells per rosette had doubled. Rosettes containing as few as 3 and as many as 10 cells were present. At 24 h, there were approximately 25,000 rosettes per mm2, with an average of 12 cells each."
(click to enlarge)
Figure 8.2. (A) Orientation of cells of Hyphomicrobium vulgare ZV580 at an oil-water interface. Phase contrast (×3,500). (B) showing the more amorphous extracellular adhesive material produced by H. vulgare ZV580 (×72,700). Source
These authors observed under the microscope that there are apparently two modes of attachment of individual motile cells to the liquid/air interface, reversible and non-reversible sorption, and corroborated observations made by Marshall & Cruickshank (1973) at the water/oil interface earlier. Cells that had attached to the interface in parallel rotated like propellers and detached again after a while. Cells that had attached perpendicularly to the interface with their thicker end shed their subpolar flagella after a few hours but remained attached to the interface by secreting an "adhesin" (Figure 8.2, B). They then began to form stalks at the opposing pole (Figure 8.2, A), concommitant with forming rosettes by attachment of further cells to the interface-adherent pole (there are no images for this step). The lousy quality of Figure 8.2 is explained by the year of publication, 1973, and by the low-resolution digitization ~2 decades later.
It is not known which geometrical rules/routes the rosettes follow when they later assemble into a mature biofilm (difficult to trace microscopically), but, as with Phaeobacter (part 1), rosette formation is an oligo‑cellular transitional stage from (motile) single cells to a multicellular biofilm.
Caulobacter crescentus rosettes
(click to enlarge)
Figure 9. (A) Caulobacter crescentus rosette viewed with structural illumination super-resolution microscopy. Overlay of cells stained using HADA to label peptidoglycan (blue), holdfast using AF-594 WGA (red), and residual DNA in the holdfast labeled using YOYO-1 (green). Source. (B) Cells of Caulobacter sp. Z-0024 after pure culture growth in PYG medium at optimal pH and temperature (TEM, ×20,000). Cropped from original image in Source
Hughes et al. (2012) summarize the dimorphic life cycle of Caulobacter crescentus as follows: "The newborn swarmer cell is equipped with a flagellum and pili at a single pole. Incapable of DNA replication, the swarmer cell dedicates its energy towards motility and dispersal. With time, the flagellar pole of the swarmer cell undergoes differentiation. It secretes a polysaccharide adhesin known as the holdfast, which mediates permanent surface attachment of the cell. Then the flagellum and pili are lost from that pole and replaced by the growing stalk, which is a thin extension of the cell envelope. The stalked cell is reproductively mature and gives off daughter swarmer cells, marking the completion of the dimorphic life cycle." (also addressed in STC in Stalking Caulobacter).
If many – very many – Caulobacter colonize a surface, as in lab experiments when using a high‑titer inoculum, one can soon see a complete, jungle-like biofilm (like here). So how does rosette formation of Caulobacter come about then? Some of those very many Caulobacter that are bound to colonize a surface eventually fail to find a free spot and cling to their siblings instead. As you can see in Figure 9A, they cling to each other exclusively via their holdfast (red dot). (You also see in Figure 9A a green dot indicating bound residual DNA; Hernando-Pérez et al. (2018) know the reason but it has nothing directly to do with rosette formation). Among Caulobacter rosettes, younger ones comprise ~6 cells with almost no stalk elongation yet (Figure 9, (A) of C. crescentus, (B) of Caulobacter sp. Z‑0024), while older rosettes include ~21 cells, most of which have extended stalks. A mature C. crescentus rosette with fully extended stalks (1–2 cell body lengths) is shown in the Frontispiece. Stalks several times the length of the cell body, around 30 µm(!), can be found in cells starved for phosphate, but I have never seen those in rosettes (ref.).
Because Caulobacter rosettes are so easy to obtain in the lab, they are excellent objects for the precise localization of proteins in stalked and swarmer cells. To give an example: Jacobs et al. (2001) showed that DivJ, a protein involved in cell cycle regulation, is exclusively localized close to the stalked pole of the cells, and nowhere in the (daughter) swarmer cells (see this diagram, or Fig. 1 in Stalking Caulobacter). Fluorescence microscopy of a rosette with labeled DivJ thus already contained the necessary control (13 of 14 cells positive) in one single experiment (see Frontispiece).
(click to enlarge)
Figure 10. Dispersal of sessile–motile aggregates. (a) SEM of a typical C. crescentus rosette with numerous stalked cells and a flagellated predivisional cell (yellow). (b) Sample trajectories of 18 C. crescentus rosettes simultaneously recorded (in 10 min) via phase contrast microscopy. Those individual rosettes remaining in the view after 10 min are circled. A rosette is visible in the zoomed-in view (inset). (c) The mean-squared displacement (MSD) of nine rosettes (purple) demonstrates a crossover lag time (Δt ~1 s) demarcating the ballistic and diffusive regimes. The MSD of six individual solitary swarmer cells (green) shows a similar ballistic-diffusive transition, implying similarity in the temporal organization of power sources. Dashed lines, fits of the experimental data with both ballistic (MSD ∝ Δt2) and diffusive (MSD ∝ Δt) activities. Error bars correspond to standard deviations. The difference in sample size between b and c is due to the fact that not all trajectories in b are sufficiently long for MSD calculations. Source
These easy-to-obtain Caulobacter rosettes are also suitable for playing billiards, so to speak. Are they motile? Can they actively move, or only be shoved around on the billards table by diffusion in the liquid medium? These questions were seriously investigated in a fun project by Zeng & Liu (2020) who traced the trajectories of 18 C. crescentus rosettes simultaneously via phase contrast microscopy over 10 min (Figure 10). They compared the movement patterns of the rosettes with those of single swarmer cells, and calculated from the traces that both move by active swimming and by diffusion, both with similar efficiency. One or two already flagellated daughter swarmer cells would suffice to move the entire rosette before they separate from the stalked mother cell. As long as the cells anchored in the rosette are reproductively active, this process can in principle repeat itself and the rosette can keep rolling. Although this is an observation from the lab, it cannot be ruled out that Caulobacter in their natural habitat use this distribution method to colonize new suitable surfaces. An aside: microbiologists who prefer to inoculate plates evenly by using glass beads rather than a Drigalski spatula or an inoculation loop could call this the "Caulo method".
An aside for the teachers among our readers. When you search for "Caulobacter rosette" on Google Images, you will find among the first 20 hits the two images labeled A and B in the composite I made, with the central rosette scaled to the same size for a better comparison. The source of "A" is a presentation of teaching materials provided on Slideshare by an author "Alpana Dave". The source of "B" is a captionized Flickr/Instagram image by Jennifer Heinritz, Jacobs-Wagner lab, Yale MSI (see Frontispiece). "A" is unsuitable as teaching material because no source is given. What is even worse is that “A” has obviously been manipulated in several ways: 1. the colored image has been converted to a black&white version to make the “SEM” label added in the presentation credible. 2. a scale bar has been added without indicating the scale. 3. on the left side, 2 cells have been photoshopped into the rosette. This cannot be passed off as an “honorable mistake”, especially since it could have been avoided by citing the source and describing the easily justifiable “edits”.
Author: Yes, I am aware that this part of the "Rosette Trilogy" is already way too long for a blog post (and will far exceed the self-imposed limit of ~1,000+ words at the end). But I didn't want to inflate the whole story into four parts. So please bear with me, or read on at a later date.
Escherichia coli rosettes
Anyone who has ever plated E. coli cells that were transformed with a resistance plasmid onto solid culture media containing antibiotics in Petri dishes will be familiar with these rare big gooey colonies that grin at you the next day. These are usually spontaneous lon– mutants which, due to this mutation, secrete quantities of exopolysaccharide, the biofilm in which the cells inside are quite well protected from the antibiotic (the resistance plasmid is not found in these cells).
(click to enlarge)
Figure 11. Sister-cell folding produces rosettes and follows an angular random walk. (A) Micrographs from a representative movie (2-s resolution) of E. coli sister-cell folding and parallel alignment. Time and scale indicated on all micrographs. New (blue) and old (red) poles are indicated. Inset: Illustration of cell folding documented in movies. (B) Later micrographs from the same video. (C) Micrographs from a representative movie (30-s resolution) of rosette formation in cells constitutively expressing GFP. New (blue) and old (red) poles are indicated. Source
But of course, lon+ cells are also able to form a biofilm, preferably at liquid/solid and liquid/air interfaces. Pratt & Kolter (1998) found that the cells must be motile (have beating flagella) to successfully contact a surface, but chemotaxis is not required. For proper attachment to a surface after initial contact by their flagella (flagellar tips?), the cells must be able to form Type 1 pili. Attached cells then grow and coat themselves and their progeny in exopolysaccharide, eventually shedding their flagella. Still-flagellated daughter cells aid in expansion of the biofilm, or leave it for other business (see here their diagram).
More recently, Puri et al. (2023) have observed some intricacies of biofilm formation by E. coli at liquid/air interfaces. First, ~300 µm long biofilm threads are oriented perpendicular to the liquid/air interface and not as Pratt & Kolter (1998) had found parallel to a surface (see here). This may be due to a unique feature of their experimental setup. Second, individual "threads" are for the most part clonal over their entire length. Third, the width of the individual "threads" is highly uniform, suggesting they consist of of ~4 aligned cell chains. The model developed by the authors can be seen on the left side of this diagram.
(click to enlarge)
Figure 12. Postulated stages of E. coli multicellular morphogenesis in hydrostatic environments. 2-cell: Self-recognizing Ag43 enables adhesion of sister cells after the first division event. Cells induce de novo flagella biosynthesis, and the propulsion of the growing flagella tail causes sister cells to perform cell folding by an angular random walk to align side by side. 4-cell: Sister cells adhere after the second division event and again perform cell folding to align in parallel and produce rosettes with quatrefoil configuration. Rosettes transition into chains, whose square cross‑sectional arrangement is approximately maintained. ~8-cell stage: Radial-projecting extracellular polymer type-1 fimbriae (indicated in red) control chain stability. ~16-cell stage: Extracelularly polymerizing curli (indicated in blue) maintain cellular positioning. ~200-cell stage: Transcriptional master regulator σ38 (encoded by rpoS) is induced (indicated by cells turning blue). ~1,000-cell stage: Extracellular polysaccharide poly-α-1,6-N-acetyl-D-glucosamine (PGA, indicated in orange) is produced and attaches multicellular communities to surfaces. The representations of adhesins are not to scale and have yet to be confirmed by electron microscopy. Modified from Source
In their latest study, Puri & Allison (2024) have tried to find out how this remarkable 4-fold symmetry of the "threads" comes about. They succeeded in tracing this back to the level of individual, growing cells in liquid culture, using time-lapse imaging of wild-type cells and a battery of mutants, each lacking motility, or the formation of pili, or curli, or the ability to secrete the exopolysaccharide poly-β-1,6-N-acetyl-D-glucosamine (PGA). As shown in Figure 11, flagellated wild-type cells expressing the polarly-located Ag43 outer-membrane protein maintain sister-cell adhesion and are "pushed" by random flagellar motion to fold once, and after the next cell division a second time, and finally into a regular quatrefoil shape that the authors call "rosette" (see Figure 12).
It is probably pointless to start a dispute about the minimum number of cells required to call their aggregation a "rosette". Nobody has yet had the idea of calling 3‑cell aggregates a "trinacria", a symbol for Sicily already found on coins from ancient times, and researchers casually call such 3-cell aggregates "rosettes", see here one from Caulobacter and in Figure 5B of part 2 ('Pb' in the lower right) from Planctomyces befkelii. If Puri & Allison (2024) now refer to 4‑cell aggregates of E. coli as "rosettes", this is perfectly acceptable.
One just has to bear in mind that, according to the model, the 4 cells of a minimal E. coli rosette are strictly clonal, that is in a direct mother-to-granddaughter relationship, and the rosette formed by folding twice (cell-doublet > quatrefoil). This clonality is maintained when the rosettes "grow out" into longer threads, which then align with each other. In contrast, cells of Phaeobacter (part 1), Planctomyces (part 2), and Caulobacter (this part) aggregate by their "sticky" pole (via stalks or fibrils) regardless of whether they are in a mother-daughter relationship (they are in a wider sense clonal, though).
The study of biofilm formation at air/liquid interfaces has recently made enormous progress, so that it is now even possible to capture the dynamics of this process. Just take a look at this short video that Lars Dietrich posted on 𝕏 these days, in which you can look over the shoulder of Pseudomonas aeruginosa as it forms a pellicle (see several single cells zapping around like comets).
MMB consortia are… rosettes
(click to enlarge)
Figure 13. Morphology and structure of MMB. (A) Cartoon depicting the morphology and internal organization of an MMB consortium. At the center of each MMB consortium lies an acellular space that is surrounded by a single layer of cells. Each cell harbors magnetosome organelles (black polygons aligned along cytoskeleton-like filaments), compartments for carbon or energy storage (gray circles), as well as other, currently unidentified structures. Scale bar ca. 1 μm. (B) SEM image of 2 MMB magnetically enriched from LSSM, possibly undergoing division. Scale bar, 1 μm. (C) Backscatter electron microscopy image of magnetosome chains within MMB cells (arrow). Magnetosome minerals appear to have 4–8 visible facets and are approximately 30–60 nm in diameter. Scale bar, 300 nm. Contrast and brightness of image (C) was increased for better visualization. LSSM, Little Sippewissett Salt Marsh; MMB, multicellular magnetotactic bacteria; SEM, scanning electron microscopy. Source
The acronym "MMB" stands for "multicellular magnetotactic bacteria" and were first mentioned in STC by Elio in Could We Have Started Out as Magnetotactic Bacteria? back in 2007.
It is relatively easy to "fish" MMB out of brackish sea water, tide pools, or lagoons using a magnet and thus enrich them in samples. However, so far they survive for a short time in the lab only and cannot be continuously cultivated. This was sufficient to examine the anatomy of these consortia in more detail and to observe that they divide as a consortium (Figure 13, B). But it was never possible to cultivate viable and propagating individual cells, and they thus seem to be obligatory multicellular bacteria. How they coordinate replication and cell division in the consortium remains enigmatic.
The symmetrical single-species MMB consortia are composed of 15–86 cells of Desulfobacterota (formerly within the phylum Deltaproteobacteria) arranged in a single layer enveloping an acellular, central compartment (Figure 13, A). Consortia range in size from 3–12 µm in diameter (Figure 14). The researchers working on MMB have agreed to call them "consortia" to indicate a specific form of organization, somewhat more precise than terms as "assemblage" or "biofilm" would. I deem it appropriate to talk of "rosettes" for MMB, as their overall anatomy – not the size, though – is fully comparable to that of choanoflaggelate rosettes, for which the term has become commonplace.
(click to enlarge)
Figure 14. Correlative microscopy analysis of MMB. (e) Composite false color image presenting data. MMB that are not false-colored were not labeled by one of the three FISH-probes. (Click here to see panels a–e in one image). (a) SEM image of magnetically enriched MMB. (b) Fluorescence microscopy (FM) identifies three MMB subpopulations using FISH probes specific for Group 1 (yellow), Group 3 (teal), and Group 4 (red). Unlabeled MMB likely belong to two populations not targeted with FISH. (c) Raman spectra of MMB separated by the respective FISH label and control. The characteristic regions of C-H at 2,800−3,100 cm–1, C-D at 2,040−2,300 cm–1, and greigite at 350 cm–1 are highlighted in gray. Each spectrum shown is an average with the shaded regions showing the standard deviation for each data set. Raman spectra were collected from several ROIs besides the one shown in this figure. The yellow spectra show Group 1 (n=38), the blue spectra show Group 3 (n=2), and the magenta spectra show Group 4 (n=3) MMB. The dark gray spectra are those of MMB that were not labeled by FISH (n=10) and the light gray spectra show MMB from the negative control (i.e., no D2O addition; n=27). (d) NanoSIMS hue-saturation-intensity images showing the 2H/1H ratios present in the ROIs outline in white boxes in b. The scale is 0 to 0.55 atom %. All MMBs were labelled with deuterium. All scale bars equal 5 μm. Source
Unlike the other bacterial rosettes presented here and in the previous parts, MMB rosettes are an organizing form on their own and not a transitional form to a biofilm. Citing the relevant literature, Schaible et al. (2024) say: "Live/dead staining experiments revealed that when cells become separated from the consortium, for example because of osmotic or mechanical stress, the consortium dismantles. This is followed by an immediate loss of magnetic orientation and motility and eventual loss of membrane integrity, leading to the death of both the separated cells and the consortium. MMB consortia consistently exhibit a high degree of magnetic optimization, excluding the possibility that the consortium is a mere aggregation of cells without underlying self-organization. Each cell within the consortium has multiple flagella, resulting in the whole consortium being peritrichously flagellated. When environmental conditions change, such as alterations in light exposure or magnetic fields, a coordinated response in motility occurs within fractions of a second . This collective response implies inter-cellular communication among individual cells, which is hypothesized to occur through the central acellular volume that the cells surround."
Until now, MMB rosettes were thought to be strictly clonal and with a uniform metabolism across their cells, but to their surprise, Schaible et al. (2024) recently obtained a more differentiated picture when they batch‑sequenced individual consortia and investigated their physiology using a battery of advanced single-cell analysis methods. They obtained Single Consortium Metagenomes (SCMs) of 22 individual MMB from a tidal pool sample, representing 8 distinct species of of 5 genera of MMB. (Meta)genome sizes varied between 6.1 and 9.1 Mb, with the GC‑content ranging from 36.2% to 38.4%, which is similar to the GC‑content observed in previously published MMB draft genomes. This alone indicates that MMB is a highly diversified group of Desulfobacterota. As is common among "socially competent" bacteria like, for example, in the very distantly related Mycococcus xanthus, the MBB (meta)genomes encode a high number of genes thought to be required for cell‑cell‑comunication.
The authors found by comparative genomics that all 22 (meta)genomes from individual MMB have such high numbers of single‑nucleotide polymorphisms (SNP) for each of the 8 species that one should rather speak of "consortia of strains" for each of them. No strict clonality here! And probing individual MMB with advanced Raman spectroscopy and NanoSims revealed that not only indiviual MMB showed a considerable degree of metabolic variation but individual cells within the consortia too (Figure 14).
On a whim,...
...I thought I would borrow the line «Rose is a rose is a rose.» from Gertrude Stein's acclaimed poem Sacred Emily and adapt it as a title for this trilogy. But now I realize that by appending "-tte" to the 3' end of "rose", I have readily disqualified myself as a poet. Poor rhythm, poor rhyme! And, also, despite some shared fragrance, the bacterial rosettes are way to diverse. I hope you will agree with this and still enjoyed the bouquet.
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