RESEARCH ARTICLE | MICROBIOLOGY BIOPHYSICS AND COMPUTATIONAL BIOLOGY Collective magnetotaxis of microbial holobionts is optimized by the three-dimensional organization and magnetic properties of ectosymbionts Daniel M. Chevriera,1 , Amélie Juhinb , Nicolas Menguyb , Romain Bolzonia , Paul E. D. Soto-Rodrigueza , Mila Kojadinovic-Sirinellia , Greig A. Patersonc , Rachid Belkhoud , Wyn Williamse , Fériel Skouri-Panetb , Artemis Kostaf , Hugo Le Guennof , Eva Pereirog , Damien Faivrea , Karim Benzerarab , Caroline L. Monteila , and Christopher T. Lefevrea,1 Edited by Edward DeLong, University of Hawai'i at Manoa, Honolulu, HI; received October 4, 2022; accepted January 17, 2023 Over the last few decades, symbiosis and the concept of holobiont—a host entity with a population of symbionts—have gained a central role in our understanding of life Significance functioning and diversification. Regardless of the type of partner interactions, under- standing how the biophysical properties of each individual symbiont and their assembly Symbiosis between a motile may generate collective behaviors at the holobiont scale remains a fundamental chal- microeukaryotic host and dozens lenge. This is particularly intriguing in the case of the newly discovered magnetotactic of nonmotile, surface-attached holobionts (MHB) whose motility relies on a collective magnetotaxis (i.e., a magnetic magnetic bacterial symbionts field-assisted motility guided by a chemoaerotaxis system). This complex behavior raises was recently discovered, where many questions regarding how magnetic properties of symbionts determine holobiont the host acquires geomagnetic magnetism and motility. Here, a suite of light-, electron- and X-ray-based microscopy field-guided navigation thanks to techniques [including X-ray magnetic circular dichroism (XMCD)] reveals that sym- chains of ferrimagnetic bionts optimize the motility, the ultrastructure, and the magnetic properties of MHBs nanoparticles (within organelles from the microscale to the nanoscale. In the case of these magnetic symbionts, the magnetic moment transferred to the host cell is in excess (102 to 103 times stronger than called magnetosomes) produced by the bacteria. Our findings Downloaded from https://www.pnas.org by INIST-CNRS CS10310 on February 27, 2023 from IP address 193.54.110.55. free-living magnetotactic bacteria), well above the threshold for the host cell to gain a magnetotactic advantage. The surface organization of symbionts is explicitly presented reveal magnetic dipoles of each herein, depicting bacterial membrane structures that ensure longitudinal alignment magnetosome chain consistently of cells. Magnetic dipole and nanocrystalline orientations of magnetosomes were also align and efficiently confer a shown to be consistently oriented in the longitudinal direction, maximizing the magnetic large magnetic moment to the moment of each symbiont. With an excessive magnetic moment given to the host cell, host. Remarkably, the calculated the benefit provided by magnetosome biomineralization beyond magnetotaxis can be magnetic moment is greatly in questioned. excess of that required to gain a collective m​agn​eto​tax​is | symbiosis | magnetosomes | biomineralization | holobiont magnetotactic advantage. These results not only show an Species are adapted to their environment thanks to the heritable changes in their structure, optimization of collective physiology, locomotion, dispersal, reproduction, and other behaviors that have passed the magnetotaxis during the course filter of natural selection over generations. Several evolutionary forces may trigger these of evolution owing to the changes and shape biodiversity, among which symbiosis has gained a lot of recognition three-dimensional organization in recent decades (1). By creating intimate and long-term physical interactions, different and magnetic properties of organisms may rapidly acquire a variety of functions, ranging from detoxification, to cell bacteria but also raise the defense, motility and, in most cases, metabolic abilities. Together, symbiotic microorgan- question on the magnetosome’s isms form a single ecological unit, called a holobiont (2), whose functioning allows its function beyond magnetic field partners/bionts to extend their ecological niche and colonize otherwise previously inac- guidance since they abundantly cessible habitats. The symbiont integration level may vary from a simple attachment to cover the host. cell surfaces to their integration into the cell cytoplasm. In the most extreme cases of endosymbiont host integration, symbiosis can even lead to the creation of new organelles and genome transfer to the host nucleus, often illustrated by the eukaryogenesis (3, 4). Symbioses are established between organisms from all domains of life and have been mostly studied in macrobial systems involving multicellular eukaryote models (5, 6). Symbioses in the microbial world are also widespread but less known even though they The authors declare no competing interest. are of prime interest to gain insights into the mechanisms driving life complexification This article is a PNAS Direct Submission. and evolution. However, observation and characterization of microbial symbioses have Copyright © 2023 the Author(s). Published by PNAS. This article is distributed under Creative Commons been challenging for many years; most symbionts defy commonly applied enrichment Attribution-NonCommercial-NoDerivatives License 4.0 and cultivation techniques (1). Even if metagenomic and single-cell genomic approaches (CC BY-NC-ND). may now circumvent some of these issues by enabling the reconstruction of genomes from 1 To whom correspondence may be addressed. Email: daniel.chevrier@cea.fr or christopher.lefevre@cea.fr. symbionts in their natural habitats (7), in silico approaches do not entirely reveal the This article contains supporting information online at holobiont structure and behavior. Therefore, interrogating the physical properties of micro- https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.​ bial symbiotic systems may bring insights on mechanisms or interactions that have shaped 2216975120/-/DCSupplemental. biodiversity. Improvements in microscale and nanoscale analytical instrumentation and Published February 27, 2023. PNAS 2023 Vol. 120 No. 10 e2216975120 https://doi.org/10.1073/pnas.2216975120 1 of 11 sample preparation strategies over the past few years have boosted MTB). As long as magnetotaxis provides a selective advantage biophysical research, permitting microbial holobiont studies from (20), the function is supposed to have been optimized over the the single-cell level to organelle level (8). For instance, approaches time of holobiont diversification. If so, to what extent did holo- such as synchrotron-based X-ray spectromicroscopy imaging, bionts converge toward the same motility behavior in response to focused ion beam-scanning electron microscopy, and cryoelectron chemical gradients and magnetic fields as in prokaryotes? In addi- microscopy (and associated cryopreparation protocols) have made tion, since cell polarity, flagella organization, magnetosome it possible to rigorously examine the ultrastructure and organiza- arrangement, number and polarity impact magnetic behavior in tion of individual cells, while other techniques such as nanoscale free-living MTB (21–25), how are these features organized by secondary ion mass spectrometry can trace the metabolic exchanges symbionts to maintain a successful magnetic guidance at the scale to decipher intracellular/intercellular chemical interactions (9). of the consortium? Microbial holobionts may adopt behavior patterns almost com- Here, we employ light-, electron- and X-ray-based microscopy pletely determined by the symbionts. Cell-to-cell interactions techniques to characterize the motility, the ultrastructure and the between bionts generate an emergent collective behavior that may magnetic properties of MHBs. The calculated magnetic dipole change the mechanical, biological and physicochemical properties of MHBs from magnetic field-based motility assays was found to of each partner over generations. Although debated among evo- be more than two orders of magnitude greater than free-living lutionary biologists (10, 11), a microbial holobiont can neverthe- MTB, an amount clearly in excess of that required to gain a less be seen as a main unit of selection. As such, any cell or magnetotactic advantage. Microscopy imaging of MHB ultras- organelle function, structure and organization optimizing the tructure [using scanning electron microscopy (SEM), transmis- collective behavior may therefore be selected in a given environ- sion electron microscopy (TEM), scanning-TEM (STEM), cryo mental context. These evolutionary processes may create a com- soft X-ray tomography (cryo-SXT)] and spectroscopic investiga- plete dependency in some biological systems, which is well tion of magnetosome magnetic properties [using X-ray magnetic illustrated by mitochondria and chloroplast evolution (12). Not circular dichroism (XMCD)] confirm a cohesive organization of all symbioses are obligatory, but most of them involve a metabolic ectosymbiotic cells and consistent magnetic dipole direction of dependency from at least one of the partners. Such is the case with magnetosome chains relative to the host, which confers an ultra- the endosymbiotic methanogenic archaea or denitrifying bacteria sensitive magnetotaxis property to the holobiont. Last, unlike in hydrogenosome-bearing protists (13, 14). Other benefits can previously reported magnetite (Fe3O4) crystal morphology in rely on structural integrity or locomotion. Indeed, similar to other Deltaproteobacteria, we found that ectosymbiotic bacteria pro- biological systems (15), symbiont interactions can generate a col- duce rhomboidal dodecahedron-like particles. From the micros- lective motion. For example, some flagellates are mobile owing to cale to the nanoscale, the results support hypotheses that the Downloaded from https://www.pnas.org by INIST-CNRS CS10310 on February 27, 2023 from IP address 193.54.110.55. thousands of ectosymbiotic bacteria (e.g., spirochetes) anchored positioning, organization, and number of ectosymbiotic cells and to the host membrane, which ensure its movement with their magnetosome chains are overly optimized by symbionts to insure flagella (16). a collective magnetotaxis similar to the magnetotaxis behavior Recently, a new collective behavior was described in a group of observed in MTB. Furthermore, these results stimulate the debate microbial holobionts ubiquitous in anoxic marine sediments. The of magnetosomes function beyond magnetotaxis since the addi- functioning of this holobiont relies on a multiscale cooperation tive magnetic moment is orders of magnitude in excess of that between a flagellate, euglenozoan protist, and dozens of nonflag- required for effective magnetotaxis. ellated, sulfate-reducing bacteria of the Deltaproteobacteria class, which biomineralize magnetic nanocrystals (17). First, the host Results and symbiont interdependency relies on metabolic exchanges. Some of them were identified based on the host ultrastructure and Magnetotactic Holobionts Have a Magnetotactic Behavior ectosymbiont genome (17). Their syntrophy is most likely based Similar to Free-Living Magnetotactic Bacteria. Since the first on the transfer of molecular hydrogen produced by mitochon- description of magnetotactic holobionts (MHBs) (17), we dria-derived organelles, called hydrogenosomes, from the host to have continuously collected samples in Carry-le-Rouet, France the bacteria that use it to reduce sulfate. Second, these holobionts (Mediterranean Sea), to have a sufficient number of south-seeking harbor a peculiarity compared with other symbiotic systems MHBs to perform an extensive characterization of their ultrastruc- described in such habitats: They adopt a collective magnetotaxis, ture and magnetotaxis. SEM and light microscopy observations (i.e., a motility guided by a chemoaerotaxis system and geomag- show two thick flagella, each measuring about two-to-three times netic fields). Similar to magnetotactic bacteria (MTB) (18, 19), the length of the cell body, which is typical in Euglenids (Movie S1 magnetotactic holobiont (MHB) motility is guided by geomag- and Fig. 1A). Transverse section of the flagella shows a canonical netic fields, which is assumed to facilitate their positioning in “9+2” microtubule axoneme structure (SI Appendix, Fig. S1) (26). preferred ecological niches, just below the oxic–anoxic boundaries SEM and TEM indicate that both flagella emerge from a large of sediments. The passive alignment in a magnetic field is ensured depression (Fig. 1) from the anterior pole where the protist is only by ectosymbionts through the biomineralization of chained lipidic partially covered by few magnetic ectosymbiotic bacteria (MEB) in vesicle-encapsulated ferrimagnetic nanocrystals called magneto- contrast to the rest of its cell body (Fig. 1 D and E). SEM observa- somes. Experimental observations evidenced that the magnetic tions at low electron energy show magnetic ectosymbiotic bacteria guidance was influenced by oxygen gradients, while a nearly com- (MEB) densely cover the surface of the protist (Fig. 1B), while plete genome of biomineralizing bacteria showed a complete loss imaging at higher electron energy confirmed that they contained of sensing and motility machinery (17). Therefore, magnetotaxis structurally aligned nanoparticle chains (Fig. 1C). appears to emerge from the collective effects of magnetosome We assessed the extent to which collective magnetotaxis chain positioning from ectosymbiotic bacteria and the host motil- evolved toward the same motility behavior as the one observed ity and chemoaerotaxis. in single-celled MTB. Magnetic moments of MHBs were The discovery of MHBs raises many questions regarding the obtained by analyzing their U-turn motion (Fig. 2 A and B pres- magnetic properties of this collective behavior and how it echoes ent optical image of MHB in recorded video and schematic of the magnetic properties developed in free-living prokaryotes (i.e., U-turn experiment, SI Appendix, Fig. S2 presents additional 2 of 11 https://doi.org/10.1073/pnas.2216975120 pnas.org Fig. 1. Observation of the motility apparatus of MHBs isolated from Carry-le-Rouet, France. (A–C) SEM images of cells prepared using a critical-point drying approach showing (A) a whole MHB, (B) a higher magnification of the same holobiont observed at 2 kV showing the surface of the holobiont, and (C) the same Downloaded from https://www.pnas.org by INIST-CNRS CS10310 on February 27, 2023 from IP address 193.54.110.55. as in (B) but observed at 10 kV using the backscattered mode showing the magnetosome chains inside the ectosymbiotic bacteria. (D) Stitched TEM images of longitudinal ultrathin sections of a MHB showing the vestibulum on the front of the protistan cells and the two flagella and (E) TEM image showing one flagellum emerging from this cavity. Nu: nucleus; MEB: magnetic ectosymbiotic bacteria; F: flagella; V: vestibulum. extracted motility parameters), which relates the radius and turn- [ ( ) ] 16 () 3 1 (a + c) (a ∗ c) ing time of MHBs upon external magnetic field reversal using A= 𝜋𝜂c ⋅ ⋅ ln − 2 , [2] 3 2 (a − c) b the following equation (27): where η corresponds to the media viscosity (i.e., Mediterranean ( ) A 2⋅M ⋅B 𝜏 Uturn = ⋅ ln . [1] M ⋅B kB T Sea viscosity at 25 °C), a (10 µm) and b (5 µm) are the major and minor axes of the ellipsoid, respectively, and c 2 = a 2 − b 2 . Here, τUturn corresponds to the time the MHB takes to make a The MHB U-turn time (τUturn) was determined by tracking the U-turn upon switching the applied magnetic field (B); M is the U-turn trajectory and then plotting the first derivative, corre- magnetic moment of the MHB; kB corresponds to the Boltzmann sponding to the instantaneous velocity (Fig. 2C), where the constant and T the temperature at which the experiment was per- change in sign corresponds to the change in direction. Then, by formed (25 °C). A is a constant related to a viscous torque. By con- taking the time difference between maximum and minimum sidering the MHB shape to be an ellipsoid (17), it is defined as: peaks (shown in Fig. 2D), the value of τ is obtained and therefore Fig. 2. Observation of the magnetic response of MHBs isolated from Carry-le-Rouet, France. (A) Optical microscopy image of a swimming MHB showing the position of the flagella. (B) Schematic of U-turn trajectory measurements during magnetic field reversal. (C) The trajectory plot of a U-turn at 3.5 mT magnetic field switching. The instantaneous velocity (µm s−1) is represented in color indicating the MHB motility direction where the blue color indicates upward motion and red color the downward displacement. (D) Time difference subtracted from first derivate peaks (t1 and t2) of the Y trajectory to yield τUturn. PNAS 2023 Vol. 120 No. 10 e2216975120 https://doi.org/10.1073/pnas.2216975120 3 of 11 M can be calculated from Eq. 1. A set of MHB U-turn trajec- in Fig. 3B with magnetosome chains in red, MEB membrane in tories (n = 22) were tracked to calculate τUturn and M. We yellow (only partially reconstructed due to limited contrast in the obtained an average absolute value of the magnetic moment M tomographic volume), protistan host in cyan (i.e., outer membrane = 1.8 ± 0.8 × 10−13 A m2. Similar magnetic moments have been and intracellular compartments), and dense intracellular granules found for Candidatus Magnetoglobus multicellularis [between in white (identified as phosphorus- and calcium-rich granules 9 ± 2 × 10−15 and 20 ± 3 × 10−15 A m2 (28)] and magnetotactic using X-ray energy-dispersive spectrometry (XEDS); SI Appendix, protists [6.7 × 10−12 A m2 (29) and 2.5 ± 1.2 × 10−13 A m2 (30)]. Fig. S4). The bottom side of the MHB (in contact with the carbon As a comparison, the magnetic moment of Magnetospirillum film of TEM grid) was flattened likely due to sedimentation or the gryphiswaldense (MSR-1) was previously determined to be 2.5 ± vitrification process. This can be observed from the “front” and 0.5 × 10−16 A m2 (31) or around 7.7 × 10−16 A m2 by direct “side” views of the reconstructed model (Fig. 3 B, Right). When measurement of the magnetosome chain using magnetic tweezers considering the magnetosome chains in red; SI Appendix, Fig. S5 (32). Similar magnetic moments have been found for other spe- shows only this reconstructed volume and the protistan host. cies of MTB as well (33, 34). We note the magnetosome organization has a general similarity to that identified with multicellular magnetotactic prokaryotes Ectosymbiotic Cells Are Parallel to Each Other and Oriented (MMPs) (35, 36); magnetosomes are found near the exterior of the along the Long Axis of Their Protistan Host. To obtain a consortium and they align in the direction from pole to pole. For global understanding of MEB organization in their native-state MMPs, however, magnetosomes organize into chain-like clusters, configuration (i.e., without congealing artifacts from fixation not distinct chains as for the MHB. processes), MHB samples were vitrified for cryo soft X-ray Thin-sections of MHB confirmed that the protistan host is tomography (cryo-SXT) imaging. Cryo-SXT offers a relatively covered by one layer of longitudinally arranged, curved rod- deep penetration and enhanced natural contrast of biological shaped ectosymbiotic bacteria (Figs. 1D and 4 A and B). materials in the water window (i.e., X-ray absorption from C and Bacterial cells are localized within an invagination of the outer N K-edges) without chemical fixation or staining. Fig. 3A presents membrane of their host (Fig. 4 A–C). Such channel-like struc- tomogram reconstructed slices of the vitrified MHB captured with tures may help the bacteria to maintain position at the surface an incident energy of 520 eV (raw X-ray image of this MHB and of the eukaryotic host and increase their surface contact with tilt-series images are shown in SI Appendix, Fig. S3, see Movie S2 the host for increased transfer of material for chemical symbi- for tomogram video). Owing to a voxel size of (12 nm)3 and the osis-related exchanges. Transversal sections of ectosymbionts high density of magnetite nanocrystals, magnetosome chains were revealed a consistent and unique morphology (Fig. 4 A–E). As distinguishable in MEB from tomography virtual slices and were previously reported (17), transversal sections indicate the pres- Downloaded from https://www.pnas.org by INIST-CNRS CS10310 on February 27, 2023 from IP address 193.54.110.55. assigned during volume segmentation (Materials and Methods). ence of wing-like structures in all ectosymbionts. Here, we show The segmentation of the entire volume reconstruction is presented that these structures are protrusions from the external Fig. 3. 3-D reconstruction of vitrified MHB using cryo soft X-ray tomography (cryo-SXT). (A) Virtual slices of X-ray tomography reconstruction through the Z-direction from top to bottom of MHB (Movie S2). (B) Volume reconstruction of magnetosome chains (red), ectosymbiotic bacteria membrane (yellow), the protistan host (cyan), and dense intracellular granules (white) from different viewpoints. 4 of 11 https://doi.org/10.1073/pnas.2216975120 pnas.org Downloaded from https://www.pnas.org by INIST-CNRS CS10310 on February 27, 2023 from IP address 193.54.110.55. Fig. 4. Organization and attachment of ectosymbiotic bacteria at the surface of their host. Transmission electron microscope images in bright field mode of ultrathin sections of MHB showing (A and B) the presence of only one layer of bacteria at the surface of their host, (C) the presence of bacteria in channels formed at the surface of their host, (C and D) the proximity of ectosymbionts with hydrogenosomes, (C–E, white arrows) the wing-like structures that are outgrowths of bacterial external membrane, and (C–F, black arrows) the presence of vesicles between the bacteria. Nu: nucleus; H: hydrogenosomes. All are transverse sections except for F, which is longitudinal. membrane only (Fig. 4 C–E) with one of the wing structures Transversal sections indicate that magnetosomes are always often below and the other above neighboring bacterial cells positioned in the lower half of the MEB cell body, closer to the (Fig. 4D). These structures could also be observed from SEM host (Fig. 4 A–E) or on side of the MEB with shorter inner cur- images where they appear as tapered edges (SI Appendix, vature radius. Similar magnetosome positioning at midcell was Fig. S6). The linking of adjacent wing structures may help to previously reported in the free-living M. gryphiswaldense MSR-1. improve ectosymbiotic cells cohesion. Longitudinal and trans- It was shown that the cytoskeletal determinant CcfM links the versal sections also reveal the presence of vesicles (i.e., round magnetoskeleton (i.e., magnetosome-specific cytoskeleton pro- structures) between bacteria that could also be involved in the duced by specific proteins such as MamK) to cell morphology in fixation of the symbionts (Fig. 4 C–F). These vesicles could also regions of inner positive-cell curvature (22). In the case of MEB, be part of the communication between MEB to coordinate their such positioning of the magnetosome chain could represent an activities such as cell division, similar to the vesicles previously advantage for optimizing the magnetic moment of the MHB. observed in MMPs (21). The presence of the ectosymbiotic Although the genome of MEB did not contain any orthologue of bacteria in the vicinity of hydrogenosomes produced by the CcfM, it is possible that another molecular pathway is involved protist was also confirmed (Fig. 4 C and D). in magnetosome positioning in MHBs. PNAS 2023 Vol. 120 No. 10 e2216975120 https://doi.org/10.1073/pnas.2216975120 5 of 11 All magnetosome chains are aligned along the long axis of the analyses; (39–41)]. Furthermore, it was important to identify host cell. Toward the front or back of the host, magnetosome MHBs that had intact flagella, so that a correlation between the chains are tilted inward (i.e., following the curvature of the host magnetization of magnetosome chains and the swimming direc- cell), indicating a close contact between MEB and the surface of tion (i.e., south-seeking) could be assessed. the host cell. From cryo-SXT volume reconstruction (Fig. 3) and Fig. 5 A and B present STXM images of two different MHBs TEM thin sections (Fig. 4), parallel magnetosome chains are typ- positioned in opposite directions and observed at 710 eV (i.e., at ically spaced 0.5 to 1 µm apart. Considering the potential for the Fe L3-edge), where magnetosome chains are most visible against interchain interactions to influence the ensemble magnetic prop- the host cell. At this energy, it is also possible to observe the flagella erties, micromagnetic calculations of simulated magnetosome of the deposited MHBs (confirmed with optical microscopy). We chain structures were conducted using MERRILL (v1.6.4) then utilized three energy points in the Fe L3-edge absorption (Materials and Methods) (37). Based on the average particle size, region that show the maximum XMCD response (708.8, 709.8, spacing, and number of particles per chain as measured by TEM, and 710.7 eV, see SI Appendix, Fig. S9C for identification of these a chain of 25 rhomboidal dodecahedral nanocrystals (see section energies) to collect maps with both circular polarizations of inci- below for the characterization of the nanocrystal shape) with a dent X-rays [circular polarized left (CPL) and right (CPR)]. Higher mid-sphere diameter of 60-nm and 10-nm spacing between grains resolution analyses on regions of interest for the first MHB show was built to model the adjacent stray fields. Based on this calcu- differences in signal intensity between circular polarizations lation (SI Appendix, Fig. S7), stray fields at least 1 µm from the (Fig. 5C). Based on the tilt direction of the sample holder, the chain (in both parallel and perpendicular directions) are consist- orientation of MHB (i.e., swimming direction) and the circular ently more than three orders of magnitude weaker than the rem- polarization, the resulting color in the XMCD difference maps nant magnetization (Mrs) of the entire chain (8.4 × 10−16 A m2, indicates the direction of magnetization. Here, the red color indi- Materials and Methods), ruling out significant interchain interac- cates magnetization in the right direction (Fig. 5C), while blue tions. Moreover, the magnetic moment for one of the modeled indicates magnetization in the left direction (Fig. 5D). As seen magnetosome chain structures is high enough to overcome from the presented XMCD difference maps, the projected mag- Brownian motion at 20 °C in a magnetic field of 10 µT. To impart netization of magnetosome chains is mostly uniform, pointing a magnetotactic advantage for MHBs over other microorganisms toward the flagella end of the MHB. This was confirmed by meas- of similar size (i.e., a net average bias (>50 % of the population) uring another MHB with the flagella on the opposite side (Fig. 5 of n cells aligned with the magnetic field direction against rand- B and D). In total, three MHB were analyzed (SI Appendix, omized orientation caused by Brownian motion), we considered Fig. S10 shows the third MHB without chemical fixation), con- how the magnetic moment of magnetosome chains decrease sistently demonstrating magnetic dipoles of magnetosome chains Downloaded from https://www.pnas.org by INIST-CNRS CS10310 on February 27, 2023 from IP address 193.54.110.55. because of their bending to follow curvature of the host cell are in the same direction among ectosymbionts with their magnetic (SI Appendix, Fig. S8). Regardless of these few chains at the south pole pointing toward the flagella, thus optimizing the swim- extreme ends of the cell, which are highly curved, we estimate that ming direction of the protistan host toward the south. This suggests net cell magnetization from only a few chains positioned close to magnetic dipole direction is maintained for dividing ectosymbi- the middle section of the host is sufficient to confer magnetotaxis onts. A similar conservation of magnetic dipole direction relative on the MHB (SI Appendix, Fig. S8 and Materials and Methods). to consortium structure during division has also been identified The total magnetic moment of the holobiont being 1.8 × 10−13 A for multicellular magnetotactic prokaryotes (MMPs) (42). m2 and that of a single magnetosome chain 8.4 × 10−16 A m2, with An inversion of individual magnetic dipole direction against most of the magnetosome chains aligned parallel to each other, the overall magnetization vector was found on a few occasions, our model would indicate there are about 215 chains attached to which occurs in the middle or at the end of magnetosome chains. the host which is consistent with microscopy observation (100 to These inversions are evident by the alternating red–blue regions 200 MEB/holobiont) (17). in the CPL–CPR difference maps (Fig. 5C). These inversions also confirm the magnetic field strength used to extract MHBs Magnetic Moments of Magnetosome Chains Align to Enhance from collected sediment did not disturb native magnetic Magnetotaxis Ability of Their Host. Considering magnetosome moments of magnetosomes. The crystalline orientation of mag- chain organization, negligible interchain interaction, and the netosome particles in one inversion region was revisited and above-calculated ensemble magnetic moment of MHBs, the examined using scanning transmission electron microscopy magnetic moment of each chain should contribute independently high-angle annular dark field (STEM-HAADF) imaging. and additively to the magnetic moment the host cell experiences. SI Appendix, Fig. S11 shows the STXM region in Fig. 5A (top This assumption was tested by measuring the native-state magnetic left region, purple frame) and plots the <111> alignment of moment of individual chains via scanning transmission X-ray individual particles as found from Fast Fourier transform (FFT) microscopy with X-ray magnetic circular dichroism (STXM- and stereographic projections. This demonstrates the consistent XMCD) in the absence of applied magnetic fields. alignment of magnetite’s easy axis <111> with the chain direction For STXM-XMCD sample preparation, MHB samples were despite the inversion of magnetic dipoles. magnetically extracted from environmental samples using weak We then performed further high-resolution transmission elec- magnetic fields (Materials and Methods). Further, when collecting tron microscopy (HR-TEM) analyses of individual magnetosomes XMCD, no magnetic fields were applied on the sample. Instead, for three MEB from three different MHBs, confirming expected the sample was tilted ~30° to the normal plane to probe the intrin- {111}, {100}, and {110} faces of magnetite nanocrystals (Fig. 6) sic magnetization of each magnetite nanocrystal (SI Appendix, and identifying the nanocrystal orientation in chain structures. Fig. S9 A and B for experimental setup) (38). MHBs with mag- The <111> crystallographic direction of each particle was almost netosome chains positioned perpendicular to the axis of rotation systematically aligned along the chain length. This is consistent of the sample holder were analyzed to optimize the XMCD signal. with the orientation generally observed for individual MTB This is because the magnetization easy axis is typically aligned biomineralizing prismatic magnetite (41). Upon close inspection along the magnetosome chain direction, i.e., <111> crystallo- of several individual magnetosomes, the shape of most magnetite graphic direction of magnetite [see below for high-resolution TEM nanocrystals is best described as a rhomboidal dodecahedron 6 of 11 https://doi.org/10.1073/pnas.2216975120 pnas.org A C B D Downloaded from https://www.pnas.org by INIST-CNRS CS10310 on February 27, 2023 from IP address 193.54.110.55. Fig. 5. Magnetic moment orientation of magnetosomes relative to the protistan flagella using soft X-ray scanning transmission X-ray microscopy (STXM) at the Fe L3-edge and X-ray magnetic circular dichroism (XMCD) without applied magnetic fields. (A and B) STXM images at 710 eV of MHB deposited in opposite directions. Fla: flagella; Dmg: X-ray beam damage from previous scans. Colored frames indicate the regions further analyzed. (C and D) Optical density maps (Left-side) and difference maps of circular polarization left (CPL) and circular polarization right (CPR) at 708.8 eV (where maximum XMCD signal was found) (Right-side). (SI Appendix, Figs. S12 and S13), a morphology that exhibits only symbiosis research gain from such collective motion and physical {110} faces. SI Appendix, Figs. S14 and S15 present additional characterization studies? HR-TEM analyses of individual particles and demonstrate the The first finding is that a similar field-guided motility has consistent <111> alignment with chain direction and further this emerged independently in free-living single-celled MTB and in unexpected nanocrystal shape for magnetite. This crystallographic microeukaryotes through symbioses, although the magnetosome form of magnetite from a magnetosome-producing bacterium of organelle has a common prokaryotic origin. The magnetic moment the Deltaproteobacteria class is unanticipated, which suggests determined using U-turn analysis was 1.8 ± 0.8 × 10−13 A m2 previous studies that correlated the nanocrystal morphology (n = 22), a value more than two orders of magnitude higher than formed by a bacterium and its phylogenetic position should be the magnetic moment of a single MTB (45–49). This magnitude reconsidered (41, 43, 44). of magnetic interaction with the geomagnetic field is in excess for the MHB. Our micromagnetic calculations on the magnetic Discussion dipole strength of simulated magnetosome chains and the size and shape of the entire holobiont estimate (considering Brownian Many microorganisms, including bacteria, microeukaryotes, and motion at 20 °C) that only a few magnetosome chains—not doz- archaea, live together permanently or at least transitorily by forming ens—positioned longitudinally along the midsection of the host microbial holobionts. These intimate relationships must not only cell would be sufficient for the protist cell to gain a magnetotactic satisfy metabolic requirements, but their physical assembly must advantage (in fields > 10 µT) over other motile microorganisms maintain, and even improve, holobiont integrity and motion. This of similar size (SI Appendix, Fig. S8 and Materials and Methods). study utilized a suite of imaging and characterization tools to gen- It was then hypothesized from this calculation and our character- erate insights on one of the most fundamental cases of collective ization of the MHB that perhaps more magnetosome chains were magnetotaxis known, ranging from motility in magnetic fields, necessary during periods of weak geomagnetic field strength in three-dimensional (3-D) organization of magnetosome chains Earth’s history. However, even with a field strength of ~1 µT, only around the host cell, polarity of magnetic dipoles, ultrastructure of a few additional magnetosome chains contributing to the additive bacterial cells at the surface of the host, down to the morphology magnetic moment of a holobiont would be required (SI Appendix, of magnetite nanocrystals. Based on the different scales of our anal- Fig. S8). Therefore, the dozens of magnetosome chains that cover ysis, the MHB studied here is a model system apt to be investigated the MHB are greatly in excess of what is required for effective with the electron and X-ray based imaging, which provide the ade- magnetotaxis. Similar observations have been made in the mag- quate spatial resolutions to capture the intricate organization at both netotactic multicellular prokaryotes where their magnetic moment cellular and nanoparticle size regimes. Based on MHBs, what can was optimized to a large degree (28, 36, 42). PNAS 2023 Vol. 120 No. 10 e2216975120 https://doi.org/10.1073/pnas.2216975120 7 of 11 Downloaded from https://www.pnas.org by INIST-CNRS CS10310 on February 27, 2023 from IP address 193.54.110.55. Fig. 6. Crystallography of the cuboctahedral/prismatic magnetite particles produced by a magnetic ectosymbiotic bacteria. (A) Transmission electron microscope bright-field image of a magnetic ectosymbiotic bacterium detached from its host and its single magnetosome chain (B). (C) High-resolution transmission electron microscopy (HR-TEM) images of the crystals annotated 1 to 6 in (B). Prismatic models drawn in white were superimposed on the image of these crystals with an acceptable match. Yellow arrows indicate the <111> direction: plain arrows correspond to in plane direction and dashed arrows are related to out-of-plane direction (for the latter, smaller arrow length corresponds to a higher out-of-plane angle). (D) FFT pattern of these HR-TEM indexed with the magnetite structure (spacegroup Fmma = 8.04 Å). (E) Stereographic projection oriented with respect to the orientation inferred from (D) using SingleCrystal software. The orientations of the models and the <111> directions were deduced from the stereographic projections. At the magnetosome level, we observed a uniformity of mag- to optimize holobiont stability and hydrodynamics. Chains and netic dipole direction in MEB. However, as presented in Fig. 5C, MEB cells are parallel to the microeukaryote/holobiont motility inversion of magnetosome magnetic dipole was detected on a few axis, and extracellular vesicles and MEB wing-like protrusions seem occasions, although its origin could not be determined by the to stick and arrange bionts together in this orientation. This reveals present work. We note that magnetic dipole inversions have been that beyond the chemical symbiosis previously described between reported for MTB strains MV-1 and AMB-1 when a similar host and ectosymbiotic bacteria (17), the holobiont collective behav- XMCD-STXM measurement was performed (38, 50). In the case ior emerges from physical interactions between biological interfaces of Magnetospirillum magneticum (AMB-1) studies, inversion of an (positioning and structural integrity of ectosymbiotic bacteria on entire magnetosome chain segment was found for a mature cell, the host) and between magnetic dipoles and the geomagnetic field whereas out-of-plane magnetic dipoles were found for recently (retention of magnetic dipole direction with respect to direction of biomineralized magnetosomes. For MHB, the alternating inver- motility). These physical constraints may act on biology to drive sions of individual magnetosomes do not appear to originate from the adaptive evolution of the holobiont. Without complying to immature magnetosomes as these particles are found mid chain, these physical restrictions, the consortium’s persistence would be spaced closely to other magnetosomes, and are of average particle challenged. size. Regardless of this irregularity to be further understood, the Such a biological constraint exerted by magnetotaxis suggests consistent alignment of chain magnetic dipoles is an indicator of an important ecological role for the holobiont. As hypothesized highly controlled cell division (24) and appropriation of chains previously (17), the host benefits from the same advantages as on the host cell to maintain the maximum magnetic moment. MTB in chemical gradients, namely, finding more easily optimal This is in contrast to a recent finding by Leão et al. (30) where chemical redox conditions in fluctuating environments. Collective bundles or clusters of bullet-shaped magnetosomes were discov- magnetotaxis may thus optimize holobiont efficiency for nutrient ered in a flagellated protist that showed a seemingly random organ- acquisition or avoidance of toxic substances. Sharing the function ization of magnetosome magnetic dipoles. with the protistan host may avoid the metabolic burden of syn- TEM examinations reveal additional interesting features: Bionts thesizing flagellar components for MEB, the energetic expense of have evolved specific structures and three-dimensional organization fueling flagellar motors and the exposure of some molecules that 8 of 11 https://doi.org/10.1073/pnas.2216975120 pnas.org could be recognized by predators. More environmental and Materials and Methods genomic data will resolve the adaptive history, but can the biophys- Sample Collection and Light Microscopy Observations. Samples were col- ical characterization of MHBs give insights into the syntrophy itself lected by free-diving at a water depth of 0.5 to 2 m in the Mediterranean Sea, in or the fundamental role of magnetosomes in MTB? Attention has Carry-le-Rouet (43.334222°N, 5.175278°E). One-liter glass bottles were filled been paid to the geolocation function of magnetosomes to explain to about 0.3 to 0.5 of their volume with sediment, then filled to their capacity the emergence of such a biomineral and organelle (43, 51, 52). with water that overlaid the sediment. Air bubbles were excluded. Once in the However, if this primary function requires only few magnetosome laboratory, samples were stored under dim light at room temperature (~25 °C). chains in MHBs, then why is the magnitude of magnetic interac- South-seeking magnetotactic holobionts were concentrated by placing a mag- tion with the geomagnetic field excessive, and why do MEB con- netic stirring bar (~10 mT) next to the bottles, above the sediment-water interface tinue to spend energy to produce such an excess number of for 2 h. Examination of magnetically concentrated cells was carried out using the magnetosomes? Here, MHBs provide further evidence that mag- hanging drop technique (61) under a Zeiss Primo Star light microscope equipped netosomes might also be involved in metabolic aspects and possibly with phase-contrast and differential interference contrast optics. The local mag- syntrophy. Similar to what has been proposed for MTB, MEB netic field used to determine magnetotaxis was reversed by rotating the stirring could act as a battery to fuel the protist (53). The close vicinity of bar magnet 180° on the microscope stage. magnetosomes to the protist’s external membrane could evidence a possible traffic of energy from prokaryotic cells toward the eukar- Movement and Magnetotactic Response Analysis. A customized magnetic yotic host, similar to how hydrogenosomes function but in the microscope equipped with a triaxial Helmholtz coilset and controller (C-SpinCoil- opposite direction (i.e., fueling the MEB with H2) (17, 54). XYZ, Micro Magnetics Inc.) and an Andor Zyla 5.5 high-speed camera was used Another hypothesis explaining such a large amount of magneto- (62). The 3D-axis Helmholtz coils can generate DC magnetic fields with a precision somes in MHBs is an antioxidant defense. Indeed, it was shown of 5% of Earth’s magnetic field (±2.5 µT). Using the setup, we programmed the switching of the magnetic field between −3.5 and +3.5 mT for the U-turn. For in model MTB strains of the Magnetospirillum genus that magne- U-turn measurements, the magnetic field was fixed for 2 s before switching. The tosomes exhibit a peroxidase-like activity (55, 56). Magnetosomes switching was repeated to collect at least three U-turns in the field of view. The were proposed to decrease and eliminate reactive oxygen species trajectories and U-turn of 22 MHB were extracted and smoothed by a tracking (ROS) in the cell. Thus, in MHBs, ectosymbiotic cells could serve script written in python and based on the OpenCV object tracking algorithms as a sink for the ROS produced by their protistan host during with the Channel and Spatial Reliability (CSR) tracker. The data are smoothed by metabolic activity or exposure to oxygen (57). Although little is a convolution-based smoothing approach. For both trajectory and U-turn meas- known about the physiology of this MHB protist and its closest urements, a 20X objective (N.A. 0.45) was used. The mathematical relation to relatives, it is possible that they require a high concentration of calculate the MHB magnetic moment from the U-turn time are described in the iron for metabolic needs. Since MTB are known to be very effi- Downloaded from https://www.pnas.org by INIST-CNRS CS10310 on February 27, 2023 from IP address 193.54.110.55. main text by Eqs. 1 and 2. The fsolve function from Octave/Matlab was used to cient in the uptake of iron (58), the protist could benefit from the obtain M from Eq. 1. Data fitting was done by the intrinsic fitting functions of iron uptake by its symbionts with the transfer of iron from the OriginPro, Version 2016 (OriginLab Corporation). ectosymbionts toward the protist. Magnetotaxis in MTB is usually described as a passive orientation Scanning Electron Microscopy (SEM). Magnetically concentrated MHB were and active swimming along the Earth’s magnetic field lines thanks fixed in a solution of 1% paraformaldehyde and deposited on a glass coverslip to magnetosomes and to flagella, respectively (59). However, it was coated with poly-L-lysine and stored at 4 °C. Before the observation, the sample shown that a potential magnetic sensing, via the widely used chem- was dehydrated in successive ethanol baths (50%, 70%, 96%, 100%) then pro- otaxis mechanism, might be actively involved in magnetotaxis (60). cessed through critical point drying (CPD) (Leica EM CPD300) before coating with carbon (Leica EM SCD500). Images were collected in the backscattered In M. magneticum AMB-1, a methyl-accepting chemotaxis protein and secondary electron modes using a Zeiss Ultra 55 FEG-SEM operating at 1 to (MCP) appears to interact with the protein MamK involved in 10 kV, a working distance of 4 mm and an aperture of 10 to 60 μm. magnetosome chain formation. This finding led to the hypothesis that the magnetic torque applied on the magnetosome chain might Transmission Electron Microscopy (TEM). TEM was used on ultrathin sections be relayed by MamK and transferred to MCPs at the cell poles via to characterize the ultrastructure of the magnetotactic holobionts. Thin-sectioned specific interactions between them. The chemotaxis signal trans- samples were prepared from magnetically concentrated protists fixed in 2.5% (w/v) duction system may then take over the subsequent response of the glutaraldehyde in sodium cacodylate buffer (0.1 M, pH 7.4) and kept at 4 °C for at cell. In MHBs, it is likely that such signal transduction system least 24 h. Due to the low biomass of MHB, fixed cells were embedded in a small between the MEB magnetosome chains and the host flagellar motor agarose plug to facilitate their transfer in the different solutions before the inclusion is absent as it was shown that none of the genes encoding classical in resin. Cells were postfixed 1 h with 1% (w/v) of osmium tetroxide. Cells were chemotaxis pathways were detected in the genome of MEB (17). then dehydrated with successive ethanol baths (30%, 50%, 70%, 90%, and 100%) Thus, to overcome potential swimming deviation against the Earth’s with increasing concentrations and finally embedded in the resin (Epon 812). magnetic field, a last hypothesis to consider is that the higher the Sections (40 nm thick and 3 mm long) were cut with the UC7RT ultramicrotome number of magnetosome chains, the better MHBs can maintain a (Leica Microsystems GmbH), deposited onto TEM copper grids and stained with swimming direction parallel to the magnetic field. uranyless solution for 10 min and Reynolds lead citrate 3% for 3 min. Electron Future studies that inspect MHBs under laboratory-based con- micrographs were recorded with a Tecnai G2 BioTWIN (FEI Company) equipped ditions and their diversity in aquatic environments will hopefully with a charge-coupled-device (CCD) camera (Megaview III, Olympus Soft imaging Solutions GmbH) with an accelerating voltage of 100 kV. The sizes of magneto- reveal more on the function of magnetosomes for eukaryotes. The somes were measured from TEM images using the ImageJ software (v1.48). finding of collective magnetotaxis has opened up an interdiscipli- nary field of study in biology and biophysics to answer the ques- High-Resolution Transmission Electron Microscopes (HRTEM) and X-ray tions this symbiosis between eukaryotes and prokaryotes generates Energy-Dispersive Spectrometry (XEDS). HRTEM was performed on cells on their evolution (i.e., were ectosymbiotic bacteria originally directly deposited onto TEM copper grids coated with a carbon film. HRTEM and recruited with the ability to produce magnetosomes or was the z-contrast imaging in the high-angle annular dark field (STEM-HAADF) mode, and formation of magnetosomes acquired by the bacteria after they elemental mapping by XEDS were carried out using a JEOL 2100 F microscope started their symbiosis with their host?) and functioning (i.e., what operating at 200 kV. This machine was equipped with a Schottky emission gun major advantages does the protist obtain by carrying dozens of and an ultra-high-resolution pole piece. HRTEM images were obtained with a biomineralizing bacteria on its surface?). Gatan US 4000 charge-coupled-device (CCD) camera. PNAS 2023 Vol. 120 No. 10 e2216975120 https://doi.org/10.1073/pnas.2216975120 9 of 11 Cryo Soft X-ray Tomography (cryo-SXT). Imaging was conducted at ALBA syn- center of a rhomboidal dodecahedron magnetosome with mid-sphere diameter chrotron using cryo transmission X-ray microscopy at Mistral beamline (Barcelona, of 60 nm (volume of 7.0 × 10−23 m3), which, with a magnetite saturation Spain) (63) under the awarded proposals 2018022677 and 2019023346. Using magnetization of 4.8 × 105 A m−1, equates to a uniformly magnetized mag- a similar approach to that described above, MHB samples were magnetically con- netosome moment of 3.36 × 10−17 A m2 (total chain moment of 8.40 × 10−16 centrated on the wall of an environmental sample bottle and then extracted by A m2; 0.84 fA m2). The magnetosome neighbor-to-neighbor edge separation is micropipette. Then, 5 µL of the magnetically concentrated MHB extract along with taken to be 10 nm. The net stray field is calculated as the sum of 25 magneto- 1 µL 100-nm Au nanoparticles (BBI Solutions concentrated 5X) were added to a some dipole fields at given distances parallel and perpendicular to the chain poly-l-lysine coated transmission electron microscopy grids (Quantafoil R2/2 holey axis (SI Appendix, Fig. S7). carbon, gold). Gold nanoparticles of 100 nm deposited on the grid served as fiducial Micromagnetic models were used to determine the effect of curved chains around markers for projection alignment prior to tomographic reconstruction. The grid was the host cell. Magnetosome size, shape, and neighbor-to-neighbor edge separation incubated horizontally for 1 to 2 min to allow deposition of MHB on the grid. The are the same as described above. The magnetosome chains were arranged in an arc grid was then vertically loaded into a Leica EM GP plunge freezer at 95% humidity, defined by a circle of a specified radius, while maintaining the face centered neigh- blotted from the back of the grid with filter paper (3 s blotting time) and then quickly bor-to-neighbor edge separation of 10 nm. Mesh inputs were generated using Trelis dropped into a liquid ethane container (−180 °C) cooled by liquid nitrogen. Vitrified v17.1. The micromagnetic calculations were performed using MERRILL v1.6.4 (37). cells were kept under cryogenic conditions until being transferred to the MISTRAL An example of a micromagnetic solution is shown in SI Appendix, Fig. S8. beamline cryo chamber for measurement. Tomograms of two MHB were collected. For a population of cells, the average alignment, , is dependent on A tilt series of projections from −65° to +65° was collected every 1° with the balance of magnetic energy rotating the cell toward the magnetic field and an incident X-ray energy of 520 eV. Exposure time varied from 1 to 2 s for each thermal energy of randomizing Brownian rotation. This can be expressed as: projection (2 s at higher angles). The sample was imaged at 0° before and after � � collecting the tilt series to ensure there was no significant beam damage at the ⟨ cos𝜃 ⟩ = L mB , achievable resolution. A 40-nm Fresnel zone plate was used with an effective kB T pixel size of 12 nm. The projections were normalized with the incoming flux and deconvolved with the measured point spread function of the optical system (64). where m is the net moment of each cell, B is the magnetic field intensity, T is the Alignment of projections was done with Etomo using Au fiducials of 100 nm. 1 temperature, kB is the Boltzmann constant, and L(x) = coth x − , is the Langevin Tomographic reconstruction and simultaneous iterations reconstruction tech- x nique (SIRT) deconvolution were performed using IMOD. Volume segmentation function. A magnetotactic advantage will be conferred to a population of cells if and visualization of tomograms was conducted using Microscopy Image Browser there is a net average bias of alignment of cells toward the magnetic field direction (65) and Amira (FEI) (66). Although this approach conserves the organization of where ≥ 0.5. Over the past 10,000 y in the Mediterranean region, the field MHB cells in their native-state configuration, partial detachment of ectosymbionts strength is consistently above 30 µT, which, at 20 °C, equates to a minimum moment was observed. However, the majority of bacteria and their overall organization at per cell of 0.24 fA m2. For an extremely weak field of 3 µT, a minimum moment per Downloaded from https://www.pnas.org by INIST-CNRS CS10310 on February 27, 2023 from IP address 193.54.110.55. the surface of their host were maintained. cell of 2.4 fA m2 is required to overcome Brownian motion and can be achieved with as few as two chains of magnetosomes (SI Appendix, Fig. S8). Scanning Transmission X-ray Microscopy (STXM) and X-ray Magnetic Data, Materials, and Software Availability. All study data are included in the Circular Dichroism (XMCD). Magnetotactic holobionts were magnetically article and/or SI Appendix. concentrated using a low-intensity magnet (~10 mT) for a maximum time of 30 min to avoid remagnetization or any magnetic disturbance/interference with ACKNOWLEDGMENTS. This work was supported by a grant from the CNRS—­ magnetosomes. Cells were then transferred onto a light microscopy slide and mission pour les initiatives transverses et interdisciplinaires (MITI), adaptation du magnetically transferred in a clean drop of filtered environmental seawater. Cells vivant à son environnement, projet SymbioAdapt and a project from the French aggregated at the edge of the filtered drop were then micromanipulated with National Research Agency (ANR SymbioMagnet-21-CE02-0034-01). Romain an InjectMan® NI2 micromanipulator and a CellTram® vario, hydraulic, manual Bolzoni PhD contract was supported by the CNRS—MITI. D.M.C. acknowledges microinjector from Eppendorf mounted to a Leica DM IL LED microscope and fur- research funding through a European Union Marie-Skłodowska Curie Action ther transferred in a drop of fixative buffer containing 2.5% (w/v) glutaraldehyde International Fellowship (MSCA-IF Project 797431: BioNanoMagnets). D.M.C. and in sodium cacodylate buffer (0.1 M, pH 7.4) prepared in filtered environmental D.F. acknowledge awarded ALBA synchrotron beamtimes (Proposals 2018022677 seawater. Fixation occurred for only 10 s at room temperature before the transfer and 2019023346), Mistral beamline staff for assistance in cryo-SXT experiments of the cells, using the micromanipulator, onto a TEM copper grid coated with a and CALIPSOplus funding for Proposal 2019023346. We acknowledge Soleil carbon film. A quick fixation was sufficient to conserve the ultrastructure of the Synchrotron for beamtime awarded (Proposal 20191124) for experiments on holobiont and allow observation of the biont’s magnetosomes in the TEM. After the Hermes beamline (STXM-XMCD). W.W. would like to acknowledge support their preparation, TEM grids were observed with STXM without previous TEM from the Natural Environmental Research Council through grants NE/V001233/1 analysis to avoid magnetic disturbance. and NE/S011978/1. We thank Suri for his help in ultrathin-sections’ preparation. STXM-XMCD measurements were performed at HERMES, the soft X-ray spec- We thank Jean-Michel Guigner for managing the TEM facility at IMPMC. G.A.P. tromicroscopy beamline at SOLEIL synchrotron (67, 68). A 25-nm Fresnel zone is funded by a Natural Environment Research Council Independent Research plate was employed with measurements conducted under vacuum conditions Fellowship (NE/P017266/1). at room temperature. Most STXM maps of MHB were collected at 710 eV (at the Fe L3-edge). Circularly polarized right (CPR) and left (CPL) light was used without applied magnetic field to conserve the native-state magnetic moment of magne- Author affiliations: aAix-Marseille Université, Centre national de la recherche scientifique (CNRS), Commissariat à l'énergie atomique et aux énergies alternatives (CEA), UMR7265, tosomes. To retrieve an XMCD signal without applied fields, the sample was tilted Bioscience and biotechnology institute of Aix-Marseille (BIAM), Saint-Paul-lez-Durance 30° relative to the focal plane to measure the intrinsic magnetic dipole moments 13108, France; bSorbonne Université, UMR CNRS 7590, Muséum national d'Histoire naturelle (MNHN), Institut de recherche pour le développement (IRD), Institut de of magnetosomes (38). XMCD maps were then generated from the difference of Minéralogie, de Physique des Matériaux et de Cosmochimie (IMPMC), 75005 Paris, France; OD-converted CPR and CPL images at 708.8 eV, the energy at which the strongest c Department of Earth, Ocean and Ecological Sciences, University of Liverpool, L69 7ZE XMCD signal was found. Other energies corresponded to 709.9 and 711.0 eV. Liverpool, UK; dSynchrotron Soleil, L'Orme des Merisiers, 91192 Gif-sur-Yvette Cedex, France; eSchool of GeoSciences, Grant Institute, University of Edinburgh, Edinburgh EH9 Axis2000 and IgorPro software were used to perform image work-up, analyses, 3JW, UK; fPlateforme de Microscopie de l'Institut de Microbiologie de la Méditerranée, and create XMCD maps. XMCD maps were obtained for at least five MEB of each Institut de Microbiologie, FR3479, Campus CNRS, 13402 Marseille cedex 20, France; and g ALBA Synchrotron Light Source, Cerdanyola del Vallés, Barcelona 08290, Spain of the three MHB measured. Author contributions: D.M.C., A.J., N.M., D.F., K.B., C.L.M., and C.T.L. designed research; Micromagnetic Simulations and Calculations. To determine the decay of D.M.C., A.J., N.M., R. Bolzoni, P.E.D.S.-R., M.K.-S., G.A.P., W.W., K.B., and C.T.L. performed research; R. Belkhou, F.S.-P., A.K., H.L.G., E.P., and D.F. contributed new reagents/analytic the stray field from a straight chain of 25 magnetosomes, we use a noninter- tools; D.M.C., A.J., N.M., P.E.D.S.-R., K.B., and C.T.L. analyzed data; N.M., P.E.D.S.-R., G.A.P., acting point dipole approximation. Each point dipole is taken to represent the D.F., and K.B. contributed partial writing; and D.M.C., C.L.M., and C.T.L. wrote the paper. 10 of 11 https://doi.org/10.1073/pnas.2216975120 pnas.org 1. J.-B. Raina et al., Symbiosis in the microbial world: From ecology to genome evolution. Biol. Open 7, 36. P. Leão et al., Ultrastructure of ellipsoidal magnetotactic multicellular prokaryotes depicts their bio032524 (2018). complex assemblage and cellular polarity in the context of magnetotaxis. Environ. Microbiol. 19, 2. L. Margulis, R. Fester, Symbiosis as a Source of Evolutionary Innovation: Speciation and 2151–2163 (2017). Morphogenesis (MIT Press, 1991). 37. P. O. Conbhui et al., MERRILL: Micromagnetic Earth related robust interpreted language laboratory. 3. P. López-Garcia, D. Moreira, The Syntrophy hypothesis for the origin of eukaryotes revisited. Nat. Geochem. Geophys. Geosystems 19, 1080–1106 (2018). Microbiol. 5, 655–667 (2020). 38. L. Le Nagard et al., Magnetite magnetosome biomineralization in Magnetospirillum magneticum 4. P. López-García, L. Eme, D. Moreira, Symbiosis in eukaryotic evolution. J. Theor. Biol. 434, 20–33 strain AMB-1: A time course study. Chem. Geol. 530, 119348 (2019). (2017). 39. F. C. Meldrum, S. Mann, B. R. Heywood, R. B. Frankel, D. A. Bazylinski, Electron-microscopy study of 5. P. Engel, N. A. Moran, The gut microbiota of insects - diversity in structure and function. Fems magnetosomes in a cultured coccoid magnetotactic bacterium. Proc. R. Soc. Lond. Ser. B. Biol. Sci. Microbiol. Rev. 37, 699–735 (2013). 251, 231–236 (1993). 6. P. Vandenkoornhuyse, A. Quaiser, M. Duhamel, A. Le Van, A. Dufresne, The importance of the 40. F. C. Meldrum, S. Mann, B. R. Heywood, R. B. Frankel, D. A. Bazylinski, Electron-microscopy study microbiome of the plant holobiont. New Phytol. 206, 1196–1206 (2015). of magnetosomes in 2 cultured vibrioid magnetotactic bacteria. Proc. R. Soc. Lond. Ser. B. Biol. Sci. 7. T. Woyke et al., Symbiosis insights through metagenomic analysis of a microbial consortium. Nature 251, 237–242 (1993). 443, 950–955 (2006). 41. M. Pósfai, C. T. Lefèvre, D. Trubitsyn, D. A. Bazylinski, R. B. Frankel, Phylogenetic significance 8. C. E. Harper, C. J. Hernandez, Cell biomechanics and mechanobiology in bacteria: Challenges and of composition and crystal morphology of magnetosome minerals. Front. Microbiol. 4, 344 opportunities. APL Bioeng. 4, 021501 (2020). (2013). 9. J. Decelle et al., Subcellular architecture and metabolic connection in the planktonic 42. M. Winklhofer, L. G. Abraçado, A. F. Davila, C. N. Keim, H. G. P. Lins de Barros, Magnetic optimization photosymbiosis between Collodaria (radiolarians) and their microalgae. Environ. Microbiol. 23, in a multicellular magnetotactic organism. Biophys. J. 92, 661–670 (2007). 6569–6586 (2021). 43. C. T. Lefèvre et al., Monophyletic origin of magnetotaxis and the first magnetosomes. Environ. 10. J.-C. Simon, J. R. Marchesi, C. Mougel, M.-A. Selosse, Host-microbiota interactions: From holobiont Microbiol. 15, 2267–2274 (2013). theory to analysis. Microbiome 7, 5 (2019). 44. E. C. T. Descamps, J.-B. Abbé, D. Pignol, C. T. Lefèvre, "Controlled biomineralization of magnetite in 11. D. Faure, J.-C. Simon, T. Heulin, Holobiont: A conceptual framework to explore the eco-evolutionary bacteria" in Iron Oxides, D. Faivre, Eds. (Wiley-VCH Verlag GmbH & Co. KGaA, 2016), pp. 99–116. and functional implications of host-microbiota interactions in all ecosystems. New Phytol. 218, 45. R. B. Frankel, R. P. Blakemore, Navigational compass in magnetic bacteria. J. Magn. Magn. Mater. 1321–1324 (2018). 15–8, 1562–1564 (1980). 12. P. López-García, D. Moreira, Open questions on the origin of eukaryotes. Trends Ecol. Evol. 30, 46. D. M. S. Esquivel, H. G. P. Lins De Barros, Motion of magnetotactic microorganisms. J. Exp. Biol. 121, 697–708 (2015). 153–163 (1986). 13. M. Müller et al., Biochemistry and evolution of anaerobic energy metabolism in eukaryotes. 47. N. Petersen, D. G. Weiss, H. Vali, "Magnetic bacteria in lake sediments" in Geomagnetism and Microbiol. Mol. Biol. Rev. MMBR 76, 444–495 (2012). Palaeomagnetism, F. J. Lowes et al., Eds. (NATO ASI Series, Springer, Netherlands, 1989), pp. 14. E. Hamann et al., Environmental Breviatea harbor mutualistic Arcobacter epibionts. Nature 534, 231–241. 254–258 (2016). 48. Y. Pan et al., Reduced efficiency of magnetotaxis in magnetotactic coccoid bacteria in higher than 15. T. Vicsek, A. Zafeiris, Collective motion. Phys. Rep. 517, 71–140 (2012). geomagnetic fields. Biophys. J. 97, 986–991 (2009). 16. A. M. Wier et al., Spirochete attachment ultrastructure: Implications for the origin and evolution of 49. E. Wajnberg, L. Desouza, H. Debarros, D. Esquivel, A study of magnetic-properties of magnetotactic cilia. Biol. Bull. 218, 25–35 (2010). bacteria. Biophys. J. 50, 451–455 (1986). 17. C. L. Monteil et al., A symbiotic origin of magnetoreception in unicellular eukaryotes. Nat. Microbiol. 50. S. S. Kalirai, D. A. Bazylinski, A. P. Hitchcock, Anomalous magnetic orientations of magnetosome 4, 1088–1095 (2019). chains in a magnetotactic bacterium: Magnetovibrio blakemorei strain MV-1. PloS One 8, e53368 18. D. A. Bazylinski, R. B. Frankel, Magnetosome formation in prokaryotes. Nat. Rev. Microbiol. 2, (2013). 217–230 (2004). 51. E. F. DeLong, R. B. Frankel, D. A. Bazylinski, Multiple evolutionary origins of magnetotaxis in bacteria. 19. C. T. Lefèvre et al., Diversity of magneto-aerotactic behaviors and oxygen sensing mechanisms in Science 259, 803–806 (1993). cultured magnetotactic bacteria. Biophys. J. 107, 527–538 (2014). 52. W. Lin et al., Origin of microbial biomineralization and magnetotaxis during the Archean. Proc. Natl. Downloaded from https://www.pnas.org by INIST-CNRS CS10310 on February 27, 2023 from IP address 193.54.110.55. 20. X. Mao, R. Egli, X. Liu, L. Zhao, Magnetotactic advantage in stable sediment by long-term Acad. Sci. U.S.A. 114, 2171–2176 (2017). observations of magnetotactic bacteria in Earth’s field, zero field and alternating field. PloS One 17, 53. J. M. Byrne et al., Redox cycling of Fe(II) and Fe(III) in magnetite by Fe-metabolizing bacteria. e0263593 (2022). Science 347, 1473–1476 (2015). 21. X.-X. Qian et al., Juxtaposed membranes underpin cellular adhesion and display unilateral cell 54. V. P. Edgcomb et al., Identity of epibiotic bacteria on symbiontid euglenozoans in O2-depleted division of multicellular magnetotactic prokaryotes. Environ. Microbiol. 22, 1481–1494 (2020). marine sediments: Evidence for symbiont and host co-evolution. ISME J. 5, 231–243 (2011). 22. D. Pfeiffer et al., A bacterial cytolinker couples positioning of magnetic organelles to cell shape 55. F. F. Guo et al., Magnetosomes eliminate intracellular reactive oxygen species in Magnetospirillum control. Proc. Natl. Acad. Sci. U.S.A. 117, 32086–32097 (2020). gryphiswaldense MSR-1. Environ. Microbiol. 14, 1722–1729 (2012). 23. M. Toro-Nahuelpan et al., MamY is a membrane-bound protein that aligns magnetosomes and the 56. K. Li et al., Magnetosomes extracted from Magnetospirillum magneticum strain AMB-1 showed motility axis of helical magnetotactic bacteria. Nat. Microbiol. 4, 1978–1989 (2019). enhanced peroxidase-like activity under visible-light irradiation. Enzyme Microb. Technol. 72, 72–78 24. C. T. Lefèvre, M. Bennet, S. Klumpp, D. Faivre, Positioning the flagellum at the center of a dividing (2015). cell to combine bacterial division with magnetic polarity. mBio 6, e02286 (2015). 57. T. Fenchel, B. Finlay, Oxygen and the spatial structure of microbial communities. Biol. Rev. Camb. 25. M. Toro-Nahuelpan et al., Segregation of prokaryotic magnetosomes organelles is driven by Philos. Soc. 83, 553–569 (2008). treadmilling of a dynamic actin-like MamK filament. BMC Biol. 14, 88 (2016). 58. D. Schüler, E. Baeuerlein, Dynamics of iron uptake and Fe3O4 biomineralization during aerobic and 26. M. L. Ginger, N. Portman, P. G. McKean, Swimming with protists: Perception, motility and flagellum microaerobic growth of Magnetospirillum gryphiswaldense. J. Bacteriol. 180, 159–162 (1998). assembly. Nat. Rev. Microbiol. 6, 838–850 (2008). 59. R. B. Frankel, D. A. Bazylinski, M. S. Johnson, B. L. Taylor, Magneto-aerotaxis in marine coccoid 27. D. Esquivel, H. Debarros, M. Farina, P. Aragao, J. Danon, Magnetotactic microorganisms in the Rio- bacteria. Biophys. J. 73, 994–1000 (1997). De-Janeiro Region. Biol. Cell 47, 227–233 (1983). 60. N. Philippe, L.-F. Wu, An MCP-like protein interacts with the MamK cytoskeleton and is involved in 28. M. Perantoni et al., Magnetic properties of the microorganism Candidatus Magnetoglobus magnetotaxis in Magnetospirillum magneticum AMB-1. J. Mol. Biol. 400, 309–322 (2010). multicellularis. Naturwissenschaften 96, 685–690 (2009). 61. D. Schüler, The biomineralization of magnetosomes in Magnetospirillum gryphiswaldense. Int. 29. F. F. T. de Araujo, M. A. Pires, R. B. Frankel, C. E. M. Bicudo, Magnetite and magnetotaxis in Algae. Microbiol. Off. J. Span. Soc. Microbiol. 5, 209–214 (2002). Biophys. J. 50, 375–378 (1986). 62. M. Bennet et al., Influence of magnetic fields on magneto-aerotaxis. PLoS One 9, e101150 (2014). 30. P. Leão et al., Magnetosome magnetite biomineralization in a flagellated protist: Evidence for an early 63. A. Sorrentino et al., MISTRAL: A transmission soft X-ray microscopy beamline for cryo nano- evolutionary origin for magnetoreception in eukaryotes? Environ. Microbiol. 22, 1495–1506 (2019). tomography of biological samples and magnetic domains imaging. J. Synchrotron Radiat. 22, 31. M. P. Pichel, T. A. G. Hageman, I. S. M. Khalil, A. Manz, L. Abelmann, Magnetic response of 1112–1117 (2015). Magnetospirillum gryphiswaldense observed inside a microfluidic channel. J. Magn. Magn. Mater. 64. J. Otón et al., Characterization of transfer function, resolution and depth of field of a soft X-ray 460, 340–353 (2018). microscope applied to tomography enhancement by Wiener deconvolution. Biomed. Opt. Express 7, 32. C. Zahn et al., Measurement of the magnetic moment of single Magnetospirillum gryphiswaldense 5092–5103 (2016). cells by magnetic tweezers. Sci. Rep. 7, 1–14 (2017). 65. I. Belevich, M. Joensuu, D. Kumar, H. Vihinen, E. Jokitalo, Microscopy image browser: A platform for 33. R. Nadkarni, S. Barkley, C. Fradin, A comparison of methods to measure the magnetic moment of segmentation and analysis of multidimensional datasets. PLOS Biol. 14, e1002340 (2016). magnetotactic bacteria through analysis of their trajectories in external magnetic fields. PLoS One 8, 66. D. Stalling, M. Westerhoff, H.-C. Hege, "38 - amira: A highly interactive system for visual data e82064 (2013). analysis" in Visualization Handbook, C. D. Hansen, C. R. Johnson, Eds. (Butterworth-Heinemann, 34. R. E. Dunin-Borkowski et al., Magnetic microstructure of magnetotactic bacteria by electron 2005), pp. 749–767. holography. Science 282, 1868–1870 (1998). 67. R. Belkhou et al., HERMES: A soft X-ray beamline dedicated to X-ray microscopy. J. Synchrotron 35. F. Abreu et al., Cell adhesion, multicellular morphology, and magnetosome distribution in the Radiat. 22, 968–979 (2015). multicellular magnetotactic prokaryote Candidatus Magnetoglobus multicellularis. Microsc. 68. S. Swaraj et al., Performance of the HERMES beamline at the carbon K-edge. J. Phys. Conf. Ser. 849, Microanal. Off. J. Microsc. Soc. Am. Microbeam Anal. Soc. Microsc. Soc. Can. 3, 1–9 (2013). 012046 (2017). PNAS 2023 Vol. 120 No. 10 e2216975120 https://doi.org/10.1073/pnas.2216975120 11 of 11