November Title: Homologs of the plastidal preprotein translocase Tic20 mediate organelle assembly in bacteria Authors: Anja Paulus 0 Frederik Ahrens 0 Annika Schraut 0 Hannah Hofmann 0 Tim Schiller 0 Thomas Sura Dörte Becher René Uebe rene.uebe@uni-bayreuth.de 0 Affiliations: Department of Microbiology, University of Bayreuth; Bayreuth , Germany Microbial Proteomics, Institute of Microbiology, University of Greifswald; Greifswald , Germany 2023 16 2023 821 830

Organelle-specific protein translocation systems are essential for organelle biogenesis and maintenance in eukaryotes but thought to be absent from prokaryotic organelles. Here, we identified that MamF-like proteins involved in the formation of bacterial magnetosome organelles share an ancient origin with Tic20 protein translocases found in chloroplasts. Deletion of mamF-like genes in the alphaproteobacterium Magnetospirillum gryphiswaldense results in severe defects in organelle positioning, biomineralization, and magnetic navigation. Consistent with translocase-like functions, these defects are caused by the loss of magnetosome targeting of a subset of organellar proteins containing C-terminal glycine-rich integral membrane domains. Our findings suggest that organelle-specific protein translocation systems may indeed play a role in bacterial organelle formation.

 -

Main Text: Eukaryotes possess macromolecular machineries like preprotein translocases of chloroplasts or mitochondria (Tic/Toc or Tim/Tom) to ensure biogenesis and maintenance of these organelles. Both translocases have a mixed phylogenetic origin and evolved through the combination of novel eukaryotic proteins with bacterial subunits of endosymbiotic origin (Muñoz-Gómez et al., 2015; Rast et al., 2015). Although the endosymbionts (cyano- and alphaproteobacteria, respectively) likely also owned membranous organelles, no organellespecific protein import pathways equivalent to eukaryotic translocases have been identified among bacteria yet (Greening and Lithgow, 2020). Therefore, bacterial organelle biogenesis must proceed through distinct mechanisms. Magnetosomes, unique magnetic organelles, for example, emerge from the inner cell membrane by invagination (Komeili et al., 2006) . In Magnetospirillum gryphiswaldense MSR-1 (MSR) and related alphaproteobacteria, this process is supposedly induced by assembly of membrane-embedded complexes of magnetosome proteins (MAPs) that initiate magnetosome membrane (MM) formation through a molecular crowdinglike mechanism (Raschdorf et al., 2016). Notably, several of these MAPs also facilitate subsequent steps of magnetosome biogenesis like iron uptake, magnetite biomineralization, or crystal growth (Uebe and Schüler, 2016) . Concurrently, a dynamic cytoskeletal network composed of the MM adaptor protein MamJ (Scheffel et al., 2006), which mediates magnetosome binding to the actin-like protein MamK (Komeili et al., 2006) and the membranebound mechanical scaffold MamY (Toro-Nahuelpan et al., 2019), orchestrates the assembly of approximately 30 magnetosomes into a linear magnetosome chain (MC). This arrangement produces a magnetic dipole that enables geomagnetic navigation. Although magnetosomes represent one of the best characterized bacterial organelles, the molecular mechanisms of magnetosome assembly and protein targeting are still poorly understood as it remained unknown how and when proteins are targeted to these organelles. Here, by combining bioinformatic indepth analyses with systematic molecular, cell biological, quantitative proteomic as well as biochemical studies and evaluation of magnetism-related phenotypes, we show that magnetosomal MamF-like proteins (MFPs) and the distantly related plastidal preprotein translocase Tic20 are part of an ancient, widespread Tic20/HOTT (Homologs of Tic Twenty, formerly DUF4870) superfamily that mediates organelle-specific protein targeting.

Results: MFPs are members of a common Tic20/HOTT superfamily

Recently, we discovered a ~10 kb region within the genome of MSR that encodes several previously unrecognized MAPs (Uebe et al., 2018b) . One of these proteins, MmxF, is closely related to the well-known MAPs MamF and MmsF (hereafter collectively termed MamF-like proteins, MFPs) that are thought to facilitate magnetite crystal growth by direct interaction with the crystal surface or metal ions (fig. S1A) (Murat et al., 2012; Lohße et al., 2014 ; Rawlings et al., 2014). Contrary to previous studies that failed to detect any homologs outside of magnetotactic bacteria, we identified highly significant homologies between MFPs, the DUF4870 (hereafter termed HOTT) as well as the Tic20 protein families (fig. S1B). Interestingly, while the widely distributed HOTT family encompasses only uncharacterized proteins, the Tic20 family, among uncharacterized cyanobacterial representatives, also comprises Tic20 from chloroplasts (fig. S1, C and D). In these photosynthetic organelles, Tic20 functions as a protein-translocating channel within the Tic complex located at the inner envelope membrane (IEM) (Kouranov et al., 1998; van Dooren et al., 2008; Kikuchi et al., 2009; KovácsBogdán et al., 2011) . The identified homologies are further supported by cluster analysis of sequences (CLANS) (17). Here, the Tic20 family forms a widely separated but connected cluster to the HOTT family, in which MFPs seem to represent a discrete subfamily (Fig. 1A). Similar results were obtained by phylogenetic analyses where seven subfamilies including MFPs form distinct branches within the HOTT family, which itself is separated from the Tic20 family clade by a long branch (fig. S1E). Remarkably, despite only remote sequence homologies, structural superimposition, and multiple sequence alignments of MFPs, HOTT and Tic20 family proteins revealed similar architectures, including a reentrant helix, followed by a conserved charged loop and two additional transmembrane helices (Fig. 1B and C). The HOTT (including the MFPs) and Tic20 families thus form a common protein superfamily.

MFPs are essential for MC formation and magnetotaxis

The distant homology to Tic20 suggests that MFPs, besides regulating biomineralization, may also play a role during magnetosome assembly. To explore this hypothesis, we first analyzed transmission electron micrographs (TEM) of MSR mutants carrying deletions of mamF-like genes in all possible combinations, including the MFP-free mutant ΔF3 (ΔmamFΔmmsFΔmmxF) (Fig. 2 and fig. S2). These analyses revealed that, in agreement with known MFP deletion phenotypes (Murat et al., 2012; Lohße et al., 2014 ), the average magnetite crystal size gradually decreased from the WT, the single and double deletion mutants to the ΔF3 strain while crystal numbers per cell were only slightly reduced (Fig. 2, A and B). Beyond these expected biomineralization defects, however, the simultaneous deletion of all MFPs also resulted in the unexpected loss of MC formation as magnetosomes are randomly distributed within cells of strain ΔF3 (Fig. 2C). This observation is also reflected by an increased fraction of magnetosomes that have no neighboring particle within distance of 35nm and an overall lower number of closeneighboring magnetosomes (Fig. 2, D and E, fig. S2, B and C). Notably, expression of each mamF-like gene within ΔF3 restored magnetosome crystal sizes at least partially, whereas MC formation was reconstituted by expression of mmsF or mmxF but not mamF (Fig. 2, A and C, fig. S2A). Thus, only MmsF and MmxF seem to have an additional function during MC assembly. 101 102

Both, crystal size and MC formation are important parameters that influence the cellular magnetic dipole moment and, therefore, strongly affect the cells’ ability to align with the geomagnetic field (Pfeiffer and Schüler, 2020). Consequently, MFP mutants with reduced crystal sizes and defective MCs are severely impaired in magnetotaxis while strains exhibiting long MCs (WT and ΔF3::mmsF) showed strong magnetic alignment (Fig. 2F). Collectively, these results indicate that MFPs are essential for magnetic navigation and thus organellar function through regulation of MC formation and magnetosome crystal sizes.

 MFPs mediate MM protein assembly

The absence of MCs suggests that the function of the cytoskeletal proteins MamK, MamY, or MamJ might be affected in strain ΔF3. Nevertheless, these proteins are present at WT levels in ΔF3 cell extracts and even appear to be functional since fluorescently labelled proteins still localized in WT-like filaments in ΔF3 strains (Fig. 3, A and B). In purified MM extracts of ΔF3, however, the amounts of MamY and MamJ were significantly reduced and only MamK was detectable at WT levels (Fig. 3A). Since MM localization of MamY was previously shown to depend on MamJ (Toro-Nahuelpan et al., 2019), we surmised that loss of MamJ MM targeting is the primary cause for the absence of MCs in ΔF3. This suggestion is supported by decreased MamJ abundances in MM extracts of all MFP mutant strains that simultaneously lack mmsF and mmxF as well as the phenotypic similarity of the mamJ deletion mutant (Scheffel et al., 2006) with strains ΔF3::mamF and ΔmmsFΔmmxF, respectively (fig. S2A, Fig. 3C). Consistently, quantitative proteomic analyses of WT and ΔF3 MM extracts revealed eleven proteins that are depleted in ΔF3 MM fractions. These include MamJ with a 44-fold and MamY with a 6-fold reduced abundance, while MamK was again identified in equal quantities in WT and ΔF3 MM extracts (Fig. 3D). Additionally, the magnetite crystal size-regulating proteins MamD, Mms5, and MamR (84-fold, 12-fold and 9-fold decrease, respectively) (Arakaki et al., 2014; Lohße et al., 2014 ) were also strongly depleted in ΔF3 MMs, which was verified by immunoblotting of isolated MM extracts of WT and ΔF3 strains expressing GFP fusions (Fig. 3D, fig. S3A). Interestingly, while MamF fails to mediate MamJ MM targeting, in vivo fluorescence imaging of WT and ΔF3 strains producing GFP-tagged MamD, Mms5, or MamR showed that expression of mmsF, mmxF, or mamF in ΔF3 individually restored magnetosome localization of MamD-, Mms5- and MamR-GFP, respectively (fig. S3B). These findings indicate that all MFPs are central for correct MM assembly but may differ in their “substrates”.

MamJ is embedded within the MM

To gain a better understanding of the MFP-mediated MM assembly, we chose to study MamJ as model “substrate” because of its strong depletion in the ΔF3 mutant and the easily detectable mamJ deletion phenotype (Fig. 3D, fig. S2A). Since MamJ-GFP has a cytosolic localization in the absence of other MAPs (ΔM05, fig. S4A) (Dziuba et al., 2020), we initially assumed that MamJ might be membrane-associated by binding to MmsF or MmxF. To assess this assumption, solubilized MM protein complexes were fractionated by size exclusion chromatography (SEC). To reduce complexity, we used MMs of strain ΔF3Δmms5ΔmamKΔmamY::GFP-mmsF in which, due to the deletion of all genes encoding MamJ interaction partners like MamK, MamY, and MFPs, merely a complemented, functional GFP-MmsF fusion protein (Murat et al., 2012) can interact with MamJ. In contrast to our expectations, MamJ and GFP-MmsF elution profiles overlapped only partially (Fig. 4A). Thus, only a small fraction of MM-bound MamJ might be directly associated with GFP-MmsF. To evaluate this possibility, solubilized magnetosomes of GFP-mmsF and GFP-mamF complemented ΔF3 mutant strains as well as the WT and ΔmamJ were subjected to co-immunoprecipitation (Co-IP). In these analyses, MamJ was only detected in Co-IP extracts of strain ΔF3::GFP-mmsF (Fig. 4B). Nevertheless, even in these extracts only minute amounts of MamJ were detectable as most MamJ remained in the unbound fraction. Thus, indeed only few molecules of MM-bound MamJ are in direct contact with MmsF. Next, to assess if the majority of magnetosome-bound MamJ molecules might instead be associated with other MAPs, carbonate extractions with isolated MMs from a ΔmamKΔmamY deletion strain were performed. This treatment disrupts protein-protein interactions but leaves protein-lipid interactions intact (Fujiki, Y., Hubbard, A.L., Fowler, S., and Lazarow Paul B., 1982; Vögtle et al., 2017) . Therefore, peripherally associated MamJ should be extractable from the MM. Interestingly, however, contrary to the well-known magnetosome-associated protein MamA (Taoka et al., 2006), which could be efficiently extracted from the MM, MamJ, similar to the integral MAP MamM (Barber-Zucker et al., 2016), remained bound to magnetosomes (Fig. 4C). While all tested proteins could be completely extracted from the MM by detergent treatments, even repeated and extended incubations with alkaline, acidic, or high salt buffer solutions failed to solubilize MamJ or MamM from the MM (fig. S4B). Collectively, these results indicate that magnetosome-bound MamJ behaves as an integral membrane protein.

A hydrophobic C-terminus is required for MM targeting of MamJ

To investigate which domain of MamJ is required for its MM targeting, alleles encoding fulllength (aa 1-426) or truncated mamJ-GFP-fusions were expressed in ΔmamJ. As shown previously (Scheffel et al., 2006), full-length MamJ-GFP restored MC formation and localized in a linear pole-to-pole spanning signal. Contrary, GFP fusions to the MamJ N-terminal (aa 1-80) or the alanine-rich central acidic repetitive (CAR, aa 81-333) domains revealed only soluble localization signals (Fig. 4D). Consequently, ΔmamJ-like magnetosome clusters were still observable by TEM. When GFP was fused to the MamJ C-terminus (aa 334-426) large fluorescent foci reminiscent of magnetosome clusters could be observed. Since a MamJ334-426mCherry fusion colocalized with signals from a GFP fusion to the essential MAP MamB (Uebe et al., 2011) in large fluorescent foci within mamJ, the MamJ C-terminus seems to mediate MamJ MM targeting but lacks the ability to bind the cytoskeleton (fig. S4C). Finally, immunoblotting of isolated MMs from complemented ΔmamJ strains confirmed that only fulllength MamJ and the MamJ334-426-GFP fusion protein were targeted to the MM (Fig. 4E). Consistent with its ability to mediate MM targeting, the MamJ C-terminus interacts with MmsF and MmxF, but not MamF, in bacterial two-hybrid (BACTH) assays (fig. S4D) and encompasses a putative integral membrane helix (IMH, aa 359-379) (Scheffel and Schüler, 2007). Moreover, this glycine-rich (37.9% glycine content) region proved to be essential for MC formation as only truncations within or in close proximity of the IMH (Δ335-358, Δ378-426) prevented reconstitution of MC formation in ΔmamJ although all variants retained their ability to bind the cytoskeleton (Fig. 4F). These results indicate that MamJ is inserted into the MM via its hydrophobic, glycine-rich C-terminus. Notably, also MamD and Mms5 contain C-terminal IMHs of high glycine content, which are preceded by alanine-rich domains. Thus, the three most strongly affected proteins of the ΔF3 mutant share a similar domain architecture. Moreover, despite only low sequence similarity with MamJ, the C-terminal domains of MamD and Mms5 do also interact with each MFP in BACTH assays (fig. S4E). Thus, in agreement with our fluorescence microscopy results, all three MFPs facilitate MM targeting of MamD and Mms5.

 MFPs indirectly facilitate magnetite biomineralization

Previous studies suggested that MFPs directly promote magnetite crystal growth through interaction of iron ions with a cluster of conserved acidic amino acids within loop 1 (Fig. 1C) (Rawlings et al., 2014). Our findings, however, suggest that mistargeting of crystal growthpromoting proteins like MamD or Mms5 is the main contributor for decreased magnetite crystal sizes in MFP mutants. To discriminate between both possibilities, MmsF variants, lacking all acidic (D34N, D36N, D37N, E38N) or charged (D34N, R35Q, D36N, D37N, E38N) residues of loop 1 were tested for their ability to increase magnetosome crystal sizes in ΔF3 (Fig. 5A). Consistent with an indirect role during magnetite growth, both tested variants restored crystal sizes to the same level as wild-type MmsF (Fig. 5C). Next, we examined if MmsF maintains its ability to improve crystal growth when most of the proteins mistargeted in ΔF3 are absent. To this end, we deleted mamR within the ΔA13Δmms5ΔmmxF mutant (Zwiener et al., 2021) that already lacks all MFPs and most MFP-dependent proteins due to the combined deletion of several accessory magnetosome gene operons (mamGFDC, mms5, mms6, and mamXY). Since all essential genes of the mamAB operon are maintained, the resulting ΔA13Δmms5ΔmmxFΔmamR mutant still formed tiny magnetite particles that are, because of the absence of MFPs, dispersed throughout the cell (Fig. 5B, data table S4). Upon complementation with mmsF, the ΔA13Δmms5ΔmmxFΔmamR mutant regained the ability to assemble MCs but crystal sizes showed only a minor increase compared to the expression of mmsF in the ΔF3 mutant (Fig. 5, B and C, data table S4). Thus, even though MmsF is functional in the ΔA13Δmms5ΔmmxFΔmamR mutant, as evidenced by MC restoration, the absence of MFP-targeted proteins prevented considerable magnetite crystal growth. In summary, our analyses indicate that MFPs indirectly regulate organelle positioning and magnetosome crystal size by mediating correct organelle assembly.

Discussion:

In this study, we showed that MFPs of magnetotactic bacteria are members of the yet uncharacterized HOTT (DUF4870) protein family, which itself constitutes an ancient superfamily with Tic20 preprotein translocases (Fig. 1, fig. S1). Intriguingly, despite only distant sequence homology, plastidal Tic20 and bacterial MFPs share similar structures and play essential roles in proper organelle assembly and function. Deletion of all mamF-like genes in MSR, for example, resulted in the mislocalization of several MAPs, leading to magnetosome mispositioning, reduced magnetite biomineralization as well as loss of magnetotaxis (Fig. 2). Likewise, Tic20-I mutants of A. thaliana demonstrated impaired plastidal import of photosynthesis-related preproteins, resulting in abnormal chloroplast development and albino phenotypes (Kikuchi et al., 2009; Kasmati et al., 2011) . Thus, in both cases only a small subset of organellar proteins is affected.

Tic20 mediates translocation of proteins across the inner plastidal envelop membrane (Kikuchi et al., 2013) , while MFPs seem to insert proteins into the magnetosome membrane. The putative substrate MamJ, for instance, is soluble in the absence of MFPs but tightly binds to the MM in their presence, thus requiring detergents for extraction (Fig. 3C, fig. S4, A and B). Moreover, membrane binding seems to be independent of protein-protein interactions as magnetosomebound MamJ is resistant against carbonate, high salt, acidic, or alkaline treatments and only a minor MamJ fraction interacts with MmsF (Fig. 4C, fig. S4B, Fig3, A and B). Also consistent with a membrane insertion, MamJ functionality and magnetosome targeting depends on the presence and integrity of a putative C-terminal IMH (Fig. 4E and F) which represents one of the very few conserved regions within this hypervariable protein (Dziuba et al., 2023). Interestingly, two further potential MFP substrates, MamD and Mms5, share several of these characteristics with MamJ. Both proteins can also only be extracted from magnetosomes by use of detergents (Arakaki et al., 2003) and possess a soluble, alanine-rich N-terminal domain followed by a moderately hydrophobic (grand average hydropathy scores 1.3-1.6, (Peschke et al., 2019), glycine-rich C-terminal domain that interacts with MFPs. While the Ala-rich domains lack any sequence homology, the C-terminal domains share an IMH with a GxxxG glycine zipper motif that is commonly found within membrane proteins (Kim et al., 2005) . Notably, such glycine zipper motifs are specifically enriched among chloroplast inner envelope membrane proteins that are inserted via a stop-transfer mechanism (Viana et al., 2010; Froehlich and Keegstra, 2011) . However, the GxxxG motif is unlikely to serve as a universal recognition motif for MFPs as MamG and Mms6 harbor similar motifs but do not interact with MFPs and remain largely unaffected by their deletion (fig. S3, A and B, fig. S4, D and E). Thus, further studies are required to determine how MFPs recognize their substrates. In the future, this knowledge could be exploited to sustainably produce tailored, multifunctional magnetosomes for use in biotechnology and biomedicine (Rosenfeldt et al., 2021).

Additionally, our study reassigns the role of MFPs during magnetosome formation. While previous studies assumed an active crystal growth-promoting activity (Murat et al., 2012; Rawlings et al., 2014) our data show that MFPs are central for MM protein assembly and fail to significantly improve crystal growth in the absence of their substrates (Fig. 5). Their contribution to magnetite crystal maturation thus represents only an indirect effect that is based on the direct MM-targeting of proteins which themselves mediate crystal growth (e.g., MamD, Mms5). Initially, we were surprised to identify homologs of the plastidal biogenesis factor Tic20 within an alphaproteobacterium that is distantly related to the ancestor of mitochondria. However, in contrast to Tic20, whose occurrence is restricted to plastid-containing eukaryotic linages (11 phyla) and cyanobacteria, the homologous HOTT family is widely spread among all domains of life (present in at least 38 bacterial, 13 archaeal, and 5 eukaryotic phyla). This extensive phylogenetic distribution and the larger number of divergent subfamilies suggest that the HOTT family is evolutionarily more ancient. Thus, although the short protein lengths, long evolutionary distances, and high sequence divergences among superfamily representatives (van Dooren et al., 2008; Machettira et al., 2011) prevented more precise phylogenetic reconstructions (fig. S1C and D), our findings support the hypothesis that Tic20 originated from a HOTT ancestor that probably already mediated protein targeting. In contrast, previous studies suggested that Tic20, similar to the mitochondrial inner membrane preprotein translocases (Tim17/22/23), evolved from bacterial LivH-like amino acid transporters (Rassow et al., 1999; Reumann, 1999; Bodył et al., 2010). In support of our data, however, several subsequent studies failed to detect any significant homologies between LivH, Tic20, or Tim17/22/23, respectively (Gross and Bhattacharya, 2009; Kasmati et al., 2011; Žárský and Doležal, 2016) .

In summary, we not only present the first functional characterization of the yet uncharacterized HOTT protein family but also provide insights into its evolutionary relationship and functional conservation with the Tic20 protein family. Therefore, our findings also represent the first evidence for at least primitive organelle-specific translocases within bacteria and shed new light on the evolution of eukaryotic organelles.

 References

Almagro Armenteros, J.J., Tsirigos, K.D., Sønderby, C.K., Petersen, T.N., Winther, O., Brunak, S., Heijne, G. von, and Nielsen, H. (2019). SignalP 5.0 improves signal peptide predictions using deep neural networks. Nature biotechnology 37, 420-423. https://doi.org/10.1038/s41587-0190036-z.

Arakaki, A., Webb, J., and Matsunaga, T. (2003). A novel protein tightly bound to bacterial magnetic particles in Magnetospirillum magneticum strain AMB-1. The Journal of biological chemistry 278, 8745-8750. https://doi.org/10.1074/jbc.M211729200.

Arakaki, A., Yamagishi, A., Fukuyo, A., Tanaka, M., and Matsunaga, T. (2014). Co-ordinated functions of Mms proteins define the surface structure of cubo-octahedral magnetite crystals in magnetotactic bacteria. Molecular Microbiology 93, 554-567. https://doi.org/10.1111/mmi.12683.

Barber-Zucker, S., Uebe, R., Davidov, G., Navon, Y., Sherf, D., Chill, J.H., Kass, I., Bitton, R., Schüler, D., and Zarivach, R. (2016). Disease-Homologous Mutation in the Cation Diffusion Facilitator Protein MamM Causes Single-Domain Structural Loss and Signifies Its Importance. Sci Rep 6, 31933. https://doi.org/10.1038/srep31933.

Blum, H., Beier, H., and Gross, H.J. (1987). Improved silver staining of plant proteins, RNA and DNA in polyacrylamide gels. Electrophoresis 8, 93-99. https://doi.org/10.1002/elps.1150080203. Blum, M., Chang, H.-Y., Chuguransky, S., Grego, T., Kandasaamy, S., Mitchell, A., Nuka, G., Paysan-Lafosse, T., Qureshi, M., and Raj, S., et al. (2021). The InterPro protein families and domains database: 20 years on. Nucleic acids research 49, D344-D354. https://doi.org/10.1093/nar/gkaa977.

Bodył, A., Mackiewicz, P., and Stiller, J.W. (2010). Comparative genomic studies suggest that the cyanobacterial endosymbionts of the amoeba Paulinella chromatophora possess an import apparatus for nuclear-encoded proteins. Plant biology (Stuttgart, Germany) 12, 639-649. https://doi.org/10.1111/j.1438-8677.2009.00264.x.

Bundy, A. (1986). Catalogue of Artificial Intelligence Tools. In Catalogue of Artificial Intelligence Tools (Springer, Berlin, Heidelberg), pp. 7–161.

Chevrier, D.M., Cerdá-Doñate, E., Park, Y., Cacho-Nerin, F., Gomez‐Gonzalez, M., Uebe, R., and Faivre, D. (2022). Synchrotron‐Based Nano‐X‐Ray Absorption Near‐Edge Structure Revealing Intracellular Heterogeneity of Iron Species in Magnetotactic Bacteria. Small Science 2. https://doi.org/10.1002/smsc.202100089.

Clements, A., Bursac, D., Gatsos, X., Perry, A.J., Civciristov, S., Celik, N., Likic, V.A., Poggio, S., Jacobs-Wagner, C., and Strugnell, R.A., et al. (2009). The reducible complexity of a mitochondrial molecular machine. Proceedings of the National Academy of Sciences of the United States of America 106, 15791-15795. https://doi.org/10.1073/pnas.0908264106. Cox, J., and Mann, M. (2008). MaxQuant enables high peptide identification rates, individualized p.p.b.-range mass accuracies and proteome-wide protein quantification. Nature biotechnology 26, 1367-1372. https://doi.org/10.1038/nbt.1511.

Ducret, A., Quardokus, E.M., and Brun, Y.V. (2016). MicrobeJ, a tool for high throughput bacterial cell detection and quantitative analysis. Nat Microbiol 1, 16077. https://doi.org/10.1038/nmicrobiol.2016.77.

Dziuba, M.V., Paulus, A., Schramm, L., Awal, R.P., Pósfai, M., Monteil, C.L., Fouteau, S., Uebe, R., and Schüler, D. (2023). Silent gene clusters encode magnetic organelle biosynthesis in a non-magnetotactic phototrophic bacterium. The ISME journal 17, 326-339. https://doi.org/10.1038/s41396-022-01348-y.

Dziuba, M.V., Zwiener, T., Uebe, R., and Schüler, D. (2020). Single-step transfer of biosynthetic operons endows a non-magnetotactic Magnetospirillum strain from wetland with magnetosome biosynthesis. Environmental Microbiology 22, 1603-1618. https://doi.org/10.1111/14622920.14950.

Frickey, T., and Lupas, A. (2004). CLANS: a Java application for visualizing protein families based on pairwise similarity. Bioinformatics 20, 3702-3704. https://doi.org/10.1093/bioinformatics/bth444.

Froehlich, J.E., and Keegstra, K. (2011). The role of the transmembrane domain in determining the targeting of membrane proteins to either the inner envelope or thylakoid membrane. The Plant Journal 68, 844-856. https://doi.org/10.1111/j.1365-313X.2011.04735.x. Fujiki, Y., Hubbard, A.L., Fowler, S., and Lazarow Paul B. (1982). Isolation of intracellular membranes by means of sodium carbonate treatment: application to endoplasmic reticulum. J Cell Biol, 97-102.

Galperin, M.Y., Wolf, Y.I., Makarova, K.S., Vera Alvarez, R., Landsman, D., and Koonin, E.V. (2021). COG database update: focus on microbial diversity, model organisms, and widespread pathogens. Nucleic acids research 49, D274-D281. https://doi.org/10.1093/nar/gkaa1018. Greening, C., and Lithgow, T. (2020). Formation and function of bacterial organelles. Nat Rev Microbiol 18, 677-689. https://doi.org/10.1038/s41579-020-0413-0.

Gross, J., and Bhattacharya, D. (2009). Mitochondrial and plastid evolution in eukaryotes: an outsiders' perspective. Nature reviews. Genetics 10, 495-505. https://doi.org/10.1038/nrg2610. Haft, D.H., Loftus, B.J., Richardson, D.L., Yang, F., Eisen, J.A., Paulsen, I.T., and White, O. (2001). TIGRFAMs: a protein family resource for the functional identification of proteins. Nucleic acids research 29, 41-43. https://doi.org/10.1093/nar/29.1.41.

Heyen, U., and Schüler, D. (2003). Growth and magnetosome formation by microaerophilic Magnetospirillum strains in an oxygen-controlled fermentor. Applied microbiology and biotechnology 61, 536-544. https://doi.org/10.1007/s00253-002-1219-x.

J. Brocher (2015). The BioVoxxel image processing and analysis toolbox (Paris, France). Jumper, J., Evans, R., Pritzel, A., Green, T., Figurnov, M., Ronneberger, O., Tunyasuvunakool, K., Bates, R., Žídek, A., and Potapenko, A., et al. (2021). Highly accurate protein structure prediction with AlphaFold. Nature 596, 583-589. https://doi.org/10.1038/s41586-021-03819-2.

Karimova, G., Pidoux, J., Ullmann, A., and Ladant, D. (1998). A bacterial two-hybrid system based on a reconstituted signal transduction pathway. Proceedings of the National Academy of Sciences of the United States of America 95, 5752-5756. https://doi.org/10.1073/pnas.95.10.5752. 431 432 433 434 435 436 437

Lohße, A., Kolinko, I., Raschdorf, O., Uebe, R., Borg, S., Brachmann, A., Plitzko, J.M., Müller, R., Zhang, Y., and Schüler, D. (2016). Overproduction of Magnetosomes by Genomic Amplification of Biosynthesis-Related Gene Clusters in a Magnetotactic Bacterium. Applied and environmental microbiology 82, 3032-3041. https://doi.org/10.1128/AEM.03860-15. Lohße, A., Ullrich, S., Katzmann, E., Borg, S., Wanner, G., Richter, M., Voigt, B., Schweder, T., and Schüler, D. (2011). Functional analysis of the magnetosome island in Magnetospirillum gryphiswaldense: the mamAB operon is sufficient for magnetite biomineralization. PLoS ONE 6, e25561. https://doi.org/10.1371/journal.pone.0025561.

Lucy, L.B. (1974). An iterative technique for the rectification of observed distributions. The Astronomical Journal 79, 745. https://doi.org/10.1086/111605.

Machettira, A.B., Gross, L.E., Sommer, M.S., Weis, B.L., Englich, G., Tripp, J., and Schleiff, E. (2011). The localization of Tic20 proteins in Arabidopsis thaliana is not restricted to the inner envelope membrane of chloroplasts. Plant Mol Biol 77, 381-390. https://doi.org/10.1007/s11103011-9818-5.

Martínez-García, E., Calles, B., Arévalo-Rodríguez, M., and Lorenzo, V. de (2011). pBAM1: an all-synthetic genetic tool for analysis and construction of complex bacterial phenotypes. BMC Microbiol 11, 38. https://doi.org/10.1186/1471-2180-11-38.

Mistry, J., Chuguransky, S., Williams, L., Qureshi, M., Salazar, G.A., Sonnhammer, E.L.L., Tosatto, S.C.E., Paladin, L., Raj, S., and Richardson, L.J., et al. (2021). Pfam: The protein families database in 2021. Nucleic acids research 49, D412-D419. https://doi.org/10.1093/nar/gkaa913.

Möglich, A. (2018). An Open-Source, Cross-Platform Resource for Nonlinear Least-Squares Curve Fitting. J. Chem. Educ. 95, 2273-2278. https://doi.org/10.1021/acs.jchemed.8b00649. Muñoz-Gómez, S.A., Slamovits, C.H., Dacks, J.B., Baier, K.A., Spencer, K.D., and Wideman, J.G. (2015). Ancient Homology of the Mitochondrial Contact Site and Cristae Organizing System Points to an Endosymbiotic Origin of Mitochondrial Cristae. Current Biology 25, 14891495. https://doi.org/10.1016/j.cub.2015.04.006.

Murat, D., Falahati, V., Bertinetti, L., Csencsits, R., Körnig, A., Downing, K., Faivre, D., and Komeili, A. (2012). The magnetosome membrane protein, MmsF, is a major regulator of magnetite biomineralization in Magnetospirillum magneticum AMB-1. Molecular Microbiology 85, 684-699. https://doi.org/10.1111/j.1365-2958.2012.08132.x.

Perez-Riverol, Y., Bai, J., Bandla, C., García-Seisdedos, D., Hewapathirana, S., Kamatchinathan, S., Kundu, D.J., Prakash, A., Frericks-Zipper, A., and Eisenacher, M., et al. (2022). The PRIDE database resources in 2022: a hub for mass spectrometry-based proteomics evidences. Nucleic acids research 50, D543-D552. https://doi.org/10.1093/nar/gkab1038.

Peschke, M., Le Goff, M., Koningstein, G.M., Vischer, N.O., Abdel-Rehim, A., High, S., van Ulsen, P., and Luirink, J. (2019). Distinct Requirements for Tail-Anchored Membrane Protein Biogenesis in Escherichia coli. mBio 10. https://doi.org/10.1128/mBio.01580-19. Pfeiffer, D., and Schüler, D. (2020). Quantifying the Benefit of a Dedicated "Magnetoskeleton" in Bacterial Magnetotaxis by Live-Cell Motility Tracking and Soft Agar Swimming Assay. Applied and environmental microbiology 86. https://doi.org/10.1128/AEM.01976-19.

Pfeiffer, D., Toro-Nahuelpan, M., Awal, R.P., Müller, F.-D., Bramkamp, M., Plitzko, J.M., and Schüler, D. (2020). A bacterial cytolinker couples positioning of magnetic organelles to cell shape control. Proceedings of the National Academy of Sciences of the United States of America 117, 32086-32097. https://doi.org/10.1073/pnas.2014659117.

Raschdorf, O., Forstner, Y., Kolinko, I., Uebe, R., Plitzko, J.M., and Schüler, D. (2016). Genetic and Ultrastructural Analysis Reveals the Key Players and Initial Steps of Bacterial Magnetosome Membrane Biogenesis. PLoS genetics 12, e1006101. https://doi.org/10.1371/journal.pgen.1006101.

Raschdorf, O., Müller, F.D., Pósfai, M., Plitzko, J.M., and Schüler, D. (2013). The magnetosome proteins MamX, MamZ and MamH are involved in redox control of magnetite biomineralization in Magnetospirillum gryphiswaldense. Molecular Microbiology 89, 872-886. https://doi.org/10.1111/mmi.12317.

Raschdorf, O., Plitzko, J.M., Schüler, D., and Müller, F.D. (2014). A tailored galK counterselection system for efficient markerless gene deletion and chromosomal tagging in Magnetospirillum gryphiswaldense. Applied and environmental microbiology 80, 4323-4330. https://doi.org/10.1128/AEM.00588-14.

Raschdorf, O., Schüler, D., and Uebe, R. (2018). Preparation of Bacterial Magnetosomes for Proteome Analysis. Methods in molecular biology (Clifton, N.J.) 1841, 45-57. https://doi.org/10.1007/978-1-4939-8695-8_5.

Rassow, J., Dekker, P.J., van Wilpe, S., Meijer, M., and Soll, J. (1999). The preprotein translocase of the mitochondrial inner membrane: function and evolution. Journal of molecular biology 286, 105-120. https://doi.org/10.1006/jmbi.1998.2455.

Rawlings, A.E., Bramble, J.P., Walker, R., Bain, J., Galloway, J.M., and Staniland, S.S. (2014). Self-assembled MmsF proteinosomes control magnetite nanoparticle formation in vitro. Proceedings of the National Academy of Sciences of the United States of America 111, 1609416099. https://doi.org/10.1073/pnas.1409256111.

Reumann, S. (1999). The endosymbiotic origin of the protein import machinery of chloroplastic envelope membranes. Trends in plant science 4, 302-307. https://doi.org/10.1016/s13601385(99)01449-1.

Richardson, W.H. (1972). Bayesian-Based Iterative Method of Image Restoration*. J. Opt. Soc. Am. 62, 55. https://doi.org/10.1364/JOSA.62.000055.

Richter, M., Kube, M., Bazylinski, D.A., Lombardot, T., Glöckner, F.O., Reinhardt, R., and Schüler, D. (2007). Comparative genome analysis of four magnetotactic bacteria reveals a complex set of group-specific genes implicated in magnetosome biomineralization and function. Journal of bacteriology 189, 4899-4910. https://doi.org/10.1128/JB.00119-07. 482 483 484 485

Rosenfeldt, S., Mickoleit, F., Jörke, C., Clement, J.H., Markert, S., Jérôme, V., Schwarzinger, S., Freitag, R., Schüler, D., and Uebe, R., et al. (2021). Towards standardized purification of bacterial magnetic nanoparticles for future in vivo applications. Acta Biomaterialia 120, 293303. https://doi.org/10.1016/j.actbio.2020.07.042. Sambrook, J., and Russell, D.W. (2001). Molecular cloning. A laboratory manual (Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press).

Scheffel, A., Gärdes, A., Grünberg, K., Wanner, G., and Schüler, D. (2008). The major magnetosome proteins MamGFDC are not essential for magnetite biomineralization in Magnetospirillum gryphiswaldense but regulate the size of magnetosome crystals. Journal of bacteriology 190, 377-386. https://doi.org/10.1128/JB.01371-07.

Scheffel, A., Gruska, M., Faivre, D., Linaroudis, A., Plitzko, J.M., and Schüler, D. (2006). An acidic protein aligns magnetosomes along a filamentous structure in magnetotactic bacteria. Nature 440, 110-114. https://doi.org/10.1038/nature04382.

Scheffel, A., and Schüler, D. (2007). The acidic repetitive domain of the Magnetospirillum gryphiswaldense MamJ protein displays hypervariability but is not required for magnetosome chain assembly. Journal of bacteriology 189, 6437-6446. https://doi.org/10.1128/JB.00421-07. Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M., Pietzsch, T., Preibisch, S., Rueden, C., Saalfeld, S., and Schmid, B., et al. (2012). Fiji: an open-source platform for biological-image analysis. Nature methods 9, 676-682. https://doi.org/10.1038/nmeth.2019. Schultheiss, D., Kube, M., and Schüler, D. (2004). Inactivation of the flagellin gene flaA in Magnetospirillum gryphiswaldense results in nonmagnetotactic mutants lacking flagellar filaments. Applied and environmental microbiology 70, 3624-3631. https://doi.org/10.1128/AEM.70.6.3624-3631.2004.

Taoka, A., Asada, R., Sasaki, H., Anzawa, K., Wu, L.-F., and Fukumori, Y. (2006). Spatial localizations of Mam22 and Mam12 in the magnetosomes of Magnetospirillum magnetotacticum. Journal of bacteriology 188, 3805-3812. https://doi.org/10.1128/JB.00020-06. The PyMOL Molecular Graphics System, Version 2.0 Schrödinger, LLC.

Toro-Nahuelpan, M., Giacomelli, G., Raschdorf, O., Borg, S., Plitzko, J.M., Bramkamp, M., Schüler, D., and Müller, F.-D. (2019). MamY is a membrane-bound protein that aligns magnetosomes and the motility axis of helical magnetotactic bacteria. Nat Microbiol 4, 19781989. https://doi.org/10.1038/s41564-019-0512-8.

Trifinopoulos, J., Nguyen, L.-T., Haeseler, A. von, and Minh, B.Q. (2016). W-IQ-TREE: a fast online phylogenetic tool for maximum likelihood analysis. Nucleic acids research 44, W232-5. https://doi.org/10.1093/nar/gkw256.

Tyanova, S., Temu, T., Sinitcyn, P., Carlson, A., Hein, M.Y., Geiger, T., Mann, M., and Cox, J. (2016). The Perseus computational platform for comprehensive analysis of (prote)omics data. Nature methods 13, 731-740. https://doi.org/10.1038/nmeth.3901.

Uebe, R., Ahrens, F., Stang, J., Jäger, K., Böttger, L.H., Schmidt, C., Matzanke, B.F., and Schüler, D. (2019). Bacterioferritin of Magnetospirillum gryphiswaldense Is a

Heterotetraeicosameric Complex Composed of Functionally Distinct Subunits but Is Not Involved in Magnetite Biomineralization. mBio 10. https://doi.org/10.1128/mBio.02795-18. Uebe, R., Junge, K., Henn, V., Poxleitner, G., Katzmann, E., Plitzko, J.M., Zarivach, R., Kasama, T., Wanner, G., and Pósfai, M., et al. (2011). The cation diffusion facilitator proteins MamB and MamM of Magnetospirillum gryphiswaldense have distinct and complex functions, and are involved in magnetite biomineralization and magnetosome membrane assembly. Molecular Microbiology 82, 818-835. https://doi.org/10.1111/j.1365-2958.2011.07863.x. Uebe, R., Keren-Khadmy, N., Zeytuni, N., Katzmann, E., Navon, Y., Davidov, G., Bitton, R., Plitzko, J.M., Schüler, D., and Zarivach, R. (2018a). The dual role of MamB in magnetosome membrane assembly and magnetite biomineralization. Molecular Microbiology 107, 542-557. https://doi.org/10.1111/mmi.13899. 566 567 magnetosome biosynthesis by large-scale deletion in Magnetospirillum gryphiswaldense. BMC Microbiol 21, 65. https://doi.org/10.1186/s12866-021-02124-2.

Acknowledgments: We would like to thank Gabriele Zaus and Ulrike Brandauer for technical assistance as well as Stephanie Bauer and Michael Gass for help during immunoblot analysis, two-hybrid assays, strain, and plasmid construction.

Funding: The authors are grateful for financial support by the DFG, German Research Foundation grant UE200/1-1 (RU) DFG, German Research Foundation grant INST 91/374-1 LAGG (RU) Further funding was received from the BMBF, Federal Ministry of Education and Research grant MagBioFab (RU)

 Author contributions:

Conceptualization and methodology: RU Investigation: AP, FA, AS, TSc, HH and RU (conducted the biochemical and cell biological studies), TS and DB (performed MS proteomics and analyzed MS data) Formal analysis: RU Funding acquisition: RU Supervision: RU Writing - Original draft: AP and RU Writing - Review & editing: all authors.

Competing interests: Authors declare that they have no competing interests. Data and materials availability: All data supporting the findings of this study are included in the main text, the supplementary materials or the auxiliary data files. The mass spectrometry proteomics data have been deposited to the ProteomeXchange (http://proteomecentral.proteomexchange.org) consortium via the PRIDE partner repository (Perez-Riverol et al., 2022) with the dataset identifier PXD032959

 Supplementary Material

Materials and Methods Supplementary text Figs. S1 to S5 Tables S1 to S3 Data tables S1-5 (auxiliary data files) 630 631 632 633 634 635 636 (A) CLANS analysis of 229 TIC20/HOTT superfamily proteins. Protein sequences (colored dots) with greater pairwise similarity, are clustered closer together and connected by lines when pairwise BLAST P E-values are below 10−5. Accession numbers are listed in data table S2. (B) Superimposition (RMSD 3.419 Å, 61 to 61 atoms) of the Alpha fold prediction of the HOTT family member MamF (MSR-1, AF: Q6NE74, purple) (Jumper et al., 2021; Varadi et al., 2022) with the PDB-structure of Tic20 (C. reinhardtii, PDB: 7xZI, grey) (Liu et al., 2023) exhibits equal architectures including a reentrant-like helix, followed by a conserved charged loop (dashed orange line) and two additional transmembrane helices. All HOTT homologs have a short loop (dashed blue line), while Tic20 representatives contain an extended multi helical loop (dashed yellow line) in between the reentrant helix. RMSDs of superimpositions are listed in data table S2. (C) Multiple aa sequence alignment of highly conserved transmembrane regions of representatives of each subfamily highlights a shorter length of HOTT family members in comparison to the Tic20 family. Amino acid residues are shaded using the similarity color scheme and the Pam250 score matrix. Identical (100%), strongly (80-100%) or weakly (60-80%) similar residues are shaded in black, dark grey or light grey, respectively. Positions of predicted or definitive IHMs and alpha helices are indicated by purple (HOTT family) or grey (Tic20 family) cylinders. The reentrant-like helices of MamF (MSR-1) as well as Tic20 (C. reinhardtii) and the conserved acidic amino acids within loop 1 are marked with dashed lines, blue, yellow, and orange respectively. Accession numbers are listed in data table S2. (See also fig. S1 and data table S2) Fig. 2. MFPs are essential for MC formation and magnetotaxis. (A) Deletion of mamF-like genes affects magnetite biomineralization and results in smaller crystal sizes. Violin plot displaying the magnetite crystal size distribution of the MSR WT (black), single (green), double (blue), and triple (ΔF3, yellow) MFP mutants as well as single MFP complemented ΔF3 strains (red). Particle diameters given in nm were measured from TEM micrographs. The number of analyzed particles [N] is indicated below each plot. The min, max, and mean values are indicated by bars. Statistical significance was estimated using an unpaired two-tailed MannWhitney U-test. Asterisks indicate the points of significance, *, P value ≤ 0.05; **, P value ≤ 0.01; ***, P value < 0.001; ****, P value < 0.0001; ns, not significant (P ≥ 0.05). Raw data is provided in data table S3. (B) Deletion of mamF-like genes has only a slight effect on particle numbers per cell. Violin plot showing the distribution of magnetosome numbers per cell in the WT, ΔF3, and the MFP complemented strains. Magnetosomes per cell were measured from TEM micrographs. The number of analyzed cells [N] are indicated. Coloring and statistical analysis like Fig. 2A. Raw data is provided in data table S3. (C) Magnetosome chain formation is disrupted in the absence of mmsF and mmxF. TEM micrographs of representative WT and various MFP mutant cells. Insets visualize normalized 637 638 cellular distribution of magnetosomes (black circles) from at least 16 cells and 593 particles. (Scale bars: 500 nm) (D) Quantitative magnetosome neighbor analyses (qMNA; for details see fig. S2B and S2C) visualized in violin plots as the frequency of magnetosomes per cell that have no neighboring magnetosome within 35 nm around the outer edge of the magnetite crystal. Statistical significance was estimated using an unpaired two-tailed t-test with Welchs’ correction. The number of analyzed particles [N] measured from TEM micrographs are indicated. The min, max, and mean values are indicated by bars. Coloring and statistical significance similar to Fig. 2A. Raw data is provided in data table S3. (E) qMNA histograms of the WT, double deletion mutants, ΔF3 and single MFP complemented strains demonstrating the frequency of particle neighbor numbers within 35 nm around the outer edge of a magnetite crystal. Insets illustrate representative magnetosome particle distribution schemes. Sample size and coloring similar to Fig. 2A. Raw data is provided in data table S3. (F) MFPs are essential for magnetotaxis. i Swim halo aspect ratio of the WT and various MFP mutant strains within a homogenous 600 µT magnetic field 24 h after inoculation. Dots colored similar to Fig. 2A represent aspect ratios of individual biological replicates. Bars represents the mean. Statistical significance was estimated using an unpaired two-tailed t-test (labels similar to Fig. 2A). Raw data is provided in data table S3. ii Representative swim halos of the WT and various MFP mutant strains within a homogenous 600 µT magnetic field formed after 72 h. (Scale bar: 1 cm) (See also figure S2 and data table S3.) Fig. 3. MFPs mediate MM protein assembly. (A) Immunodetection of MamK, MamY, and MamJ in whole cell extracts and isolated MM fractions of the WT, ΔF3, and ΔF3::6His-mmsF strains. Molecular weight standards are indicated. Unspecific bands are marked with asterisks. (B) Magnetosome-specific cytoskeletal proteins are still functional in ΔF3. Representative 3DSIM micrographs showing WT-like filaments of fluorescently labelled MamK, MamY, and MamJ proteins in ΔF3 strains. BF, bright field image. (Scale bars: 2 µm) (C) Immunodetection of the MM adaptor protein MamJ in whole cell extracts and isolated MM fractions of the WT, single and double deletion mutants, the triple deletion mutant, and its single MFP complemented strains. MamJ, which is present in all cell extracts, is absent in MM fractions of all strains carrying simultaneous deletions of mmsF and mmxF. Electrophoretic mobility corresponds to a ~90 kDa protein (22). (D) The protein composition of the MM in ΔF3 is strongly altered. Volcano plot showing the log2fold change of proteins in purified magnetosome fractions of ΔF3 compared to the WT as 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 determined by liquid chromatography–tandem mass spectrometry (LC-MS/MS). Magnetosome island (MAI)-encoded proteins (squares) with significantly altered abundance (log2 ratio <-1.5 or >1.5 and P < 0.05, vertical and horizontal dotted lines, respectively) compared to the WT are labeled. Circles represent non-MAI encoded proteins. Significantly depleted or enriched proteins are labelled with red and blue symbols, respectively. Non-significant proteins are depicted by grey or black symbols, respectively. Values are the mean of three biological replicates. (See also data table S4.) (See also figure S2A and S3, data table S4, and supplementary text) Fig. 4. MamJ is inserted into the MM via its C-terminal domain. (A) Size exclusion chromatography (SEC) elution profiles of MamJ and GFP-MmsF overlap only partially. Preparative separation of solubilized MAP complexes of isolated magnetosomes from ΔF3DΔmamKΔmamY::GFP-mmsF using a Superose 6 increase 10/300 GL column. MAP complex elution fractions 16-31 were separated by SDS-PAGE and subsequently analyzed by Western blot immunodetection using MamJ, GFP, and MamM antibodies. Molecular weights of the SEC standard calibration are indicated. (B) Only minor amounts of magnetosome-bound MamJ directly interact with GFP-MmsF. Coimmunoprecipitation of solubilized magnetosomes from ΔF3::GFP-mmsF, ΔF3::GFP-mamF, WT and ΔmamJ mutant strains using anti-GFP nanobodies. Input, unbound, and immunoprecipitated (IP) fractions were analyzed by SDS-PAGE and subsequent Western immunoblots using GFP and MamJ antibodies. Molecular weight standards are indicated. (C) Magnetosome-bound MamJ behaves as an integral membrane protein. Western blot and SDSPAGE analysis of buffer E- (negative control), carbonate-, and SDS-treated (positive control), magnetosome pellet (P) and supernatant (S) fractions from a ΔmamKY mutant show that MamJ and the integral MAP MamM are only detectable in supernatant fractions upon treatment with SDS. In contrast, carbonate efficiently extracted the peripheral and highly abundant MamA protein as shown by silver-stained SDS PAGE (black arrow). Molecular weight standards are indicated. (D) Domain structure of MamJ containing a central acidic repetitive (CAR) (red), alanine- and glycine-rich domain (black and purple, respectively). TEM and 3D-SIM imaging of full-length (aa 1-426) or truncated versions of MamJ fused to GFP and expressed in ΔmamJ reveal that only fulllength MamJ-GFP is functional to restore the MCs. BF, bright field image. (Scale bars: SIM 2 µm, TEM 0.5 µm) (E) The C-terminal domain mediates magnetosome targeting of MamJ. Cell extracts and isolated magnetosomes of ΔmamJ strains producing different MamJ truncations fused to GFP were analyzed by immunoblotting using GFP antibodies. A Western blot against MamM is shown as loading control. Molecular weight standards are indicated, and unspecific bands are marked with asterisks. (F) The putative C-terminal IMH of MamJ is essential for MC reconstitution in ΔmamJ. TEM and 3D-SIM imaging of truncated MamJ-GFP variants produced in ΔmamJ. Domain structure is depicted analogue to Figure 4D. MamJ variants that maintained or lost the ability to reconstitute 714 715 716 717 718

MC formation are colored in blue and apricot, respectively. BF, bright field image. (Scale bars: SIM 2 µm, TEM 0.5 µm) (See also figure S4) Fig. 5. MFPs have an indirect effect on magnetite biomineralization. (A) Representative TEM micrographs of ΔF3 expressing mmsF mutant alleles lacking acidic (D34N, D36N, D37N, E38N) or charged (D34N, R35Q, D36N, D37N, E38N) residues within the loop between IMH1 and IMH2 (see Figure 1 B and C). (Scale bars: 500 nm) (B) Representative TEM micrographs of the ΔA13Δmms5/mmxF/mamR mutant and the complemented strain ΔA13Δmms5/mmxF/mamR::mmsF. The positions of the small magnetosomes are indicated by arrows. (Scale bars: 500 nm) (C) Violin plot showing the magnetite crystal size distribution of strain ΔF3 expressing WT and mutant mmsF variants as well as strains ΔA13Δmms5/mmxF/mamR and ΔA13Δmms5/mmxF/mamR::mmsF. The number of analyzed particles [N] is indicated. The min, max, and mean values are given by bars. Statistical significance was estimated using an unpaired two-tailed Mann-Whitney U test (*, P value ≤ 0.05; **, P value ≤ 0.01; ***, P value < 0.001; ****, P value < 0.0001; ns, not significant (P ≥ 0.05)). Raw data is provided in data table S5. (See also Figure 1 and data table S5) input were added to the equilibrated beads and after incubation for 1 h at 4 °C on a roll mixer, beads were sedimented by centrifugation (5 min, 2,500 x g, 4 °C; Eppendorf 5424 R, FA-45-2411 rotor). The supernatant (unbound) and 50 µL of the unused input were diluted 1:1 with 2x SDS-sample buffer (100 mM Tris, 20% (v/v) glycerol (v/v), 4% (w/v) SDS, 0.2% (w/v) bromophenol blue, 200 mM DTT, pH 6.8) and stored at -20°C until further analysis. The beads were washed three times with 500 μL buffer 2, sedimented by centrifugation (5 min, 2,500 x g, 4 °C) and the supernatant was discarded. Finally, the beads were transferred to new 1.5 mL reaction tubes, resuspended in 80 μL 2x SDS-sample buffer, boiled for 5 min at 95 °C to dissociate immunocomplexes (IP) from beads and sedimented by centrifugation (5 min, 2,500 x g, 10 °C). Samples were then evaluated via SDS-PAGE and subsequent Western immunoblots by loading 10 µL (input) and 20 µL (unbound, Co-IP) aliquots, respectively. Sample preparation for mass spectrometry

Magnetosome samples were prepared for mass spectrometry using filter aided sample preparation (FASP) as described elsewhere ( Wiśniewski, 2016 ). Briefly, aliquots containing 100 µg of total protein were reduced with tris(2-carboxyethyl)phosphine followed by mixing with 200 µL of buffer UA (8 M urea in 0.1 M Tris/HCl, pH 8.5) and loading on a Microcon YM 30 (Merck-Millipore, Darmstadt, Germany) filtration device by centrifugation. Subsequently, proteins were alkylated by adding iodoacetamide in buffer UA and digested for 18 hours at 37 °C with trypsin at a ratio of 1:100 (trypsin:protein) in 40 µL 50 mM Tris/HCl. Peptides were eluted by centrifugation with 100 µL 50 mM Tris/HCl, twice. Eluates were pooled and resulting peptides were purified by Pierce C18 Tips 100 µL (Thermo Fisher Scientific, Waltham, USA). Therefore, C18 tips were wetted with 200 µL of 70% acetonitrile and equilibrated with 200 µL 3% acetonitrile. Peptides were bound by aspirating and dispensing ten times. Bound peptides were washed and eluted with water and 60% acetonitrile, respectively. Eluted peptides were dried and stored at -80 °C until further use.

Mass spectrometry of magnetosomes

Purified peptides were reconstituted with 10 µL 0.1% acetic acid and analyzed by reversed phase liquid chromatography (LC) electrospray ionization (ESI) MS/MS using a QExactive Hybrid-Quadrupol-Orbitrap mass spectrometer (Thermo Fisher Scientific, Waltham, USA). In brief, nano reversed phase LC columns (20 cm length x 100 µm diameter) packed with 3.0 µm C18 particles (Dr. Maisch GmbH, Ammerbuch Entringen, Germany) were used to separate the purified peptides with an EASY nLC 1000 system (Thermo Fisher Scientific, Waltham, USA). The peptides were loaded with buffer A (0.1% acetic acid) and subsequently eluted by a nonlinear gradient of 166 min from 2% to 99% buffer B (0.1% acetic acid, 99.9% acetonitrile) at a flow rate of 300 nL min-1. A full scan was recorded in the Orbitrap with a resolution of 70,000. The twelve most abundant precursor ions were consecutively isolated by the quadrupole and fragmented via higher-energy collisional dissociation (HCD) with a normalized collision energy of 27.5%. MS2 scans were recorded with a resolution of 17,500. Unassigned charge states, singly charged ions, as well as ions of charge 7 and higher were rejected and the lock mass correction was enabled.

Database searching and quantification was done with MaxQuant version 1.6.3.4. (Cox and Mann, 2008) with the published genome sequence of MSR (GenBank: CP027526.1) (Uebe et al., 2018b) . The MaxQuant generic contaminants database was used. Database search was based on a strict tryptic digestion with two missed cleavages permitted. Carbamidomethylating on cysteine was considered as a fixed modification and oxidation of methionine as a variable modification.

MaxQuant computed LFQ intensities were loaded into Perseus 1.6.2.2 (Tyanova et al., 2016) and log2 transformed. Putative contaminants, reverse hits, and proteins identified by site only were removed and a list containing proteins identified in all samples was exported to Excel. Mean log2 differences and respective P values were obtained by a two-sided two sample t-test over three biological replicates.

Epifluorescence microscopy

For epifluorescence microscopy, MSR strains were grown in 3 mL FSM at 2% O2 at 28 °C without shaking for approximately 20 h. When required, gene expression was induced by addition of 100 ng mL-1 anhydrotetracycline or 2 mM isopropyl β-D-1-thiogalactopyranoside (IPTG). For imaging, cells were immobilized on 1% agarose pads supplemented with FSM media components (except for peptone and yeast extracts). Therefore, 3 µL of cell suspension were pipetted on agarose pads and covered with a coverslip. Samples were then imaged with an Olympus BX81 microscope equipped with a 100× UPLSAPO100XO objective (NA1.4), an OrcaER camera (Hamamatsu), and DIC contrast. Epifluorescence micrographs were recorded in Z-stacks with 750 ms exposure time per image and then deconvoluted employing 200 iterations of the Richardson-Lucy algorithm (Richardson, 1972; Lucy, 1974) using the DeconvolutionLab 2.0.0 plugin (Sage et al., 2017) the ImageJ Fiji package (Schindelin et al., 2012). Structured illumination microscopy

3D-SIM (striped illumination at 3 angles and 5 phases) was performed on an Eclipse Ti2-E N-SIM E fluorescence microscope (Nikon) equipped with a CFI SR Apo TIRF AC 100×H NA1.49 Oil objective lens, a hardware based ‘perfect focus system’ (Nikon), LU-N3-SIM laser unit (488/561/640 nm wavelength lasers) (Nikon), and an Orca Flash4.0 LT Plus 17 sCMOS camera (Hamamatsu). Calibration of the objective correction collar and SIM grating focus was performed with TetraSpeck fluorescent beads (T-7279 TetraSpeck microspheres). For 3D-SIM imaging, cells were prepared as described above except that 8-Well µ-Slides with 1.5H (170±5 µm) D 263 M Schott glass bottom (ibidi GmbH, Gräfelfing, Germany) were used. 3D SIM z-series were acquired with 120 nm z-step spacing and exposure times in the range of 20 to 100 ms at 25 to 75% laser power. EM515/30 and EM595/31 filters and fluorescence excitation with 488 nm and 561 nm lasers were used for imaging of GFP and mCherry/mScarlet-I, respectively. Image reconstruction was performed in NIS-Elements 5.01 (Nikon) using the ‘stack reconstruction’ algorithm with the following parameter settings: The ‘illumination modulation contrast’ was set to ‘auto’. The ‘high resolution noise suppression’ was set to 0.1. Transmission electron microscopy

For transmission electron microscopy (TEM) analysis, cells were grown at 28 °C under microaerobic conditions (2% O2) over-night. One milliliter cell suspension (OD565 ~ 0.2-0.3) was then concentrated by centrifugation at 1.000 x g for 3 min, followed by resuspension in ~50 µL of residual medium. Afterwards cells were adsorbed onto carbon coated copper mesh grids (CF200-CU, Electron Microscopy Sciences, Pennsylvania) and washed with ddH2O twice. Images were recorded using Zeiss EM 902A and Jeol JEM-1400 Plus electron microscopes at an accelerating voltage of 80 kV. For data processing, interpretation, and analysis, the software packages DigitalMicrograph (Gatan) and the ImageJ Fiji package (Schindelin et al., 2012) were used. For determinations of magnetite particle numbers per cell, at least 57 cells were analyzed and at least 300 particles were measured for analysis of magnetite particle diameters.

Magnetosome chain formation was analyzed from TEM images as depicted in Figure S2. Briefly, TEM images were segmented using the ImageJ plugin Trainable Weka Segmentation v3.2.28 (Arganda-Carreras et al., 2017) for extraction of cell boundaries and a difference of Gaussians (DoG) method (sigma1 = 1, sigma2 = 2) (Bundy, 1986) to enhance the edges of the images and enable extraction of magnetosomes by thresholding. After manual curation to prevent erroneous segmentation of polyphosphate granules, binary magnetosome images were subjected to particle number analyses using the inbuilt “Analyze Particles” function of Fiji. For neighbor analyses, the “Neighbor Analysis” function of the Fiji BioVoxxel toolbox plugin 2.5.0 (J. Brocher, 2015) was used in particle neighborhood mode with 35 nm neighborhood radius. Intracellular magnetosome distribution maps were generated with MicrobeJ (Ducret et al., 2016 ) from at least 593 particles and 16 cells.

Bioinformatic analyses

For the detection of remote MFP homologies, Hidden Markov Model-based HHPred analyses were performed (Zimmermann et al., 2018) . Using individual MFP sequences or MFP alignments as queries, the domain of unknown function DUF4870 (HOTT) and the Tic20 protein families were consistently the only hits with significant statistical support in various databases like PFAM (Mistry et al., 2021), TIGRFAMs (Haft et al., 2001), or COG (Galperin et al., 2021). Subsequently, 5200 protein sequences of the identified DUF4870 (HOTT) and Tic20 protein families were retrieved from the InterPro database (Blum et al., 2021). After an initial Cluster analysis of sequences (CLANS) (Frickey and Lupas, 2004), 229 DUF4870 (HOTT) and Tic20 family proteins of bacterial, archaeal and eukaryotic origin were selected to achieve a broad phylogenetic distribution. These sequences were again analyzed by CLANS using default settings for 150,000 iterations. Only if the P value for a pair of sequences is less than 10−5 in the all-against-all BLAST search, the corresponding edges between nodes are shown as gray or black lines. To generate maximum-likelihood trees of MamF-like proteins or DUF4870 (HOTT) and TIC20 family proteins 59 sequences were aligned using MAFFT 7.474 (Katoh et al., 2019) , respectively. Trimmed alignments (TrimAI 1.3 Capella-Gutiérrez et al., 2009, no gaps and gap threshold 0.7, respectively) were then used to infer maximum-likelihood trees with IQ-Tree 1.6.11 (Trifinopoulos et al., 2016) under the LG+G4 or mtInv+F+I+G4 models as suggested by ModelFinder (Kalyaanamoorthy et al., 2017). Bootstrap support was derived by ultrafast bootstrap approximation with 1000 iterations. The phylogenetic trees were visualized and annotated by iTOL online tool (Letunic and Bork, 2019) .

To infer the phylogenetic distribution of the DUF4870 (HOTT) and Tic20 families, a total of 16840 rRNA sequences (16S or 18S) from organisms encoding these protein families were initially retrieved from the NCBI nucleotide database. After filtering for sequences of at least 1200 or 1655 nt length and removal of sequences with similarities above 97% (CD-Hit, Huang et al., 2010), a total of 456 (HOTT) and 297 (Tic20) rRNA sequences were aligned using MAFFT 7.490 (Katoh et al., 2019) , respectively. Trimmed alignments (TrimAI 1.3, no gaps) were then used to infer maximum-likelihood trees with IQ-Tree 1.6.12 (Trifinopoulos et al., 2016) under the GTR+F+R10 or TIM3+F+R3 models as suggested by ModelFinder (Kalyaanamoorthy et al., 2017). Bootstrap support was derived by ultrafast bootstrap approximation with 1000 iterations. The phylogenetic trees were visualized and annotated by iTOL online tool (Letunic and Bork, 2019) . All sequences and alignments were edited and analyzed using Geneious 8.1.4 (Biomatters, Auckland, New Zealand).

All sequences and alignments were edited and analyzed using Geneious 8.1.4 (Biomatters, Auckland, New Zealand). 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185

For protein signal sequence predictions, SignalP 5.0 (Almagro Armenteros et al., 2019) was used. Hydrophobicity analyses were performed using the grand average of hydropathy (GRAVY) calculator (http://www.gravy-calculator.de) (Kyte and Doolittle, 1982; Peschke et al., 2019) .

Figures and RMSD of superimpositions of AlphaFold database (AF) predictions (Jumper et al., 2021; Varadi et al., 2022) of Tic20/HOTT superfamily members with the PDB (Protein data bank) structure of Tic20 (C. reinhardtii, PDB ID: 7xZI) were generated with PYMOL (The PyMOL Molecular Graphics System, Version 2.0 Schrödinger, LLC).

Quantification and statistical analysis

All statistical analyses were performed with GraphPad Prism 7 software (GraphPad Software, Inc., La Jolla, CA, USA). All data were analyzed using two-tailed Student’s t or Mann-Whitney U tests, respectively. A P value of less than 0.05 was considered statistically significant. Further information about statistical details and methods is indicated in the figure legends, text, or methods. If not stated otherwise, values are given as mean ± standard deviation of the indicated sample size. Violin plots, bar plots, and intracellular magnetosome distribution maps were generated by Fit-o-Mat 0.752 (Möglich, 2018), Prism 7 software (GraphPad) and MicrobeJ (Ducret et al., 2016), respectively. Unless indicated otherwise, all experiments were performed at least twice.

 Supplementary Text

Altered composition of the MM in MFP mutants

In addition to the eleven proteins that showed a significantly decreased abundance in MM fractions of strain F3, our quantitative proteomic analyses also revealed that 26 proteins are significantly enriched compared to the WT (data table S4). Among these proteins, the three MAPs MamC, MamW, and MamX showed a three-fold increase (Fig. 3D). While increased MamC levels could be confirmed by Western immunoblots (fig. S5A), the deletion of mamC or mamW had only very weak effects on magnetosome formation (Scheffel et al., 2008; Lohße et al., 2011) . Since MamX, on the other hand, plays a role in magnetosome redox control (Raschdorf et al., 2013), the enrichment of these MAPs does not seem to play a role for the ΔF3 phenotype. Beside these MAPs, 23 significantly enriched proteins have no known function in magnetosome biomineralization and are encoded outside the genomic magnetosome island that contains almost all MAP genes (Lohße et al., 2011) . Strikingly, 13 of these proteins contain predicted secretory N-terminal signal sequences (SignalP 5.0) (data table S4) (Almagro Armenteros et al., 2019) and thus likely reside within the periplasm or the outer membrane. To test for a putative role in magnetosome biogenesis, we fluorescently labeled three of the most strongly enriched proteins (fig. S5, B and C). When produced in the WT, all proteins showed a peripheral cell localization, whereas mCherry-labeled MmeA, a MAP with a known N-terminal signal peptide (Richter et al., 2007), showed a magnetosome chain-like localization pattern. Thus, at least the tested proteins (MSR1_01270, MSR1_09970, MSR1_21580) do not localize to the MM in vivo. We thus concluded that only proteins with a decreased abundance in F3 MM extracts are relevant for our analyses. 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231

Fig. S1. MFPs are members of a common Tic20/HOTT protein superfamily, which are distributed in all domains of life. (A) A phylogenetic analysis of various MFPs reveals that one of the newly identified MAPs (MSR1_02690) is an ortholog of MmxF (amb1026) from M. magneticum AMB-1 (Rawlings et al., 2014). The tree is based on 36 MFPs sequences from different magnetotactic bacteria inferred under the best-fitting LG+G4 substitution model. Grouped MmxF, MamF and MmsF homologs are shaded with blue, green, and red boxes, respectively. MSR MFPs are highlighted in bold and MmxF is marked with an arrow. Red dots above branches are bootstrap values in percent, denoted in the legend. Scale bar represents expected substitutions per site. Accession numbers are listed in data table S1. (B) HHPred analysis of the MSR MFPs reveals significant homologies to the DUF4870 (HOTT) and Tic20 protein families. E-values (average indicated by a black line) indicate highly significant homologies of MamF, MmsF, and MmxF (green, red, blue squares, respectively) to different HOTT and Tic20 protein family models. The averaged hit probability of the templates is represented by grey bars with grey lines indicating the standard deviation. Only hits with Evalues < than 10-3 are shown. (C) A phylogenetic analysis reveals a narrow distribution of the Tic20 family. The phylogenetic tree is based on 16S/18S rRNA sequences from 297 organisms containing at least one Tic20 family protein. The tree was inferred under the best-fitting TIM3+F+R3 substitution model. Scale bar represents expected substitutions per site. (D) A phylogenetic analysis reveals the wide distribution of the HOTT family in all domains of life. The phylogenetic tree is based on 16S/18S rRNA sequences from 456 organisms containing at least one HOTT family protein. The tree was inferred under the best-fitting GTR+F+R10 substitution model. Scale bar represents expected substitutions per site. (E) Unrooted phylogenetic tree of the Tic20/HOTT superfamily inferred under the best-fitting mtInv+F+I+G4 substitution model. The MFPs and six other HOTT groups form distinct subfamilies within the HOTT family that is widely separated from the TIC20 family. HOTT Subfamily 1 exclusively includes eukaryotic sequences whereas groups 2–6 contain only prokaryotic representatives. Cyanobacteria (in teal) are the only organisms containing HOTT and Tic20 family proteins. Bootstrap values (blue dots) are denoted in the legend. Scale bar represents expected substitutions per site. Accession numbers are listed in data table S1. (F) Comparison of protein length distributions (in aa) in the Tic20/HOTT superfamily using violin plots. Please note the short length of most HOTT family members in comparison to Tic20. The min, max, and mean values are indicated by bars. The number of analyzed proteins of each group [N] is given below. (See also Figure 1 and data table S1) 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282

Fig. S2. Characterization of MSR deletion phenotypes using quantitative magnetosome neighbor analysis. (A) Overview of different MSR deletion phenotypes. Representative TEM micrographs of MFP single deletions show WT-like phenotypes (left), while mutants lacking mmsF and mmxF but contain mamF phenocopy the ΔmamJ strain (right) (Scale bars: 500 nm). (B) MSR WT magnetosomes with an average size of 35 nm are arranged in almost perfect chainlike structures in which most particles have exactly two neighboring particles within a distance of 35 nm (see particle A). Exceptions can be found at the end of chains where a wider particle spacing can be observed (e.g. particle B that has only one neighboring particle within 35 nm) or particle C that has no neighbor. In the absence of MCs the frequency of neighbor-less particles increases strongly while strains with aggregated magnetosomes frequently have neighbor numbers ≥ 3. (C) For qMNA images of different mutant strains were collected by TEM. Micrographs were subsequently segmented into magnetosomes and cell bodies. Segmented binary images were then used to analyze the overall number of magnetosomes per cell, the number of neighboring magnetosomes of each individual particle and the distribution of intracellular magnetosome localizations using Fiji plugins. Visualization of the results was performed using Fit-o-mat 0.752 (Möglich, 2018), Prism 7 (GraphPad Prism Software Inc., San Diego, California), and MicrobeJ (Ducret et al., 2016). (See also Figure 2, data table S3)

Fig. S3. Verification of the ΔF3 proteome analysis. (A) Western blot analyses (GFP) of MM fractions from WT and ΔF3 strains expressing mamG-, mamD-, mms5-, and mamR-GFP confirm a strong reduction of MamD, Mms5, and MamR in ΔF3 magnetosomes compared to the WT. In contrast, MamG-GFP is equally abundant in both strains. (B) Epifluorescence imaging of WT, ΔF3, and complemented ΔF3 strains producing MamD , Mms5-, MamR-, or MamG-GFP fusion proteins reveals MM targeting of MamD-, Mms5-, and MamR-GFP by all three MFPs. In ΔF3, MamD-, MamR-, and partially Mms5-GFP show only a soluble localization whereas MamG-GFP exclusively localizes in a punctuate pattern. DIC, differential interference contrast image. (Scale bars: 5 µm) (See also Fig. 3) 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363

Fig. S4. Characterization of potential MFP substrate proteins (A) 3D-Structured illumination microscopy of MamJ-GFP reveals a soluble localization pattern in the MSR ΔM05 mutant that lacks all magnetosome genes. Similarly, MamJ334-426-GFP has a soluble localization in the ΔmmsF/mmxF mutant that forms ΔmamJ-like magnetosome clusters but is unable to target MamJ to the MM. BF, bright field image. (Scale bars: 2 µm) (B) MamJ is resistant to prolonged treatments (20 h) with basic (0.1 M N-cyclohexyl-3aminopropanesulfonic acid (CAPS), pH 11), acidic (0.1 M glycine, pH 2.5), or high salt (1 M NaCl, 10 mM Hepes, 1 mM EDTA, pH7.4) buffer solutions. Magnetosome pellet fractions were loaded on SDS-PAGE and subsequently analyzed by Western blot (αMamJ, αMamM, αMamC) or silver staining (MamA). In an independent experiment, magnetosomes were incubated for one hour in 1% SDS as a positive control. The sample was fractionated into supernatant and magnetosome pellets by centrifugation. TCA-precipitated supernatant and the pellet fractions were loaded onto SDS-PAGE. Subsequent Western blot immunodetections of MamJ, MamM, and MamC indicate an integral membrane protein behavior for all three proteins as they could be only extracted by SDS. Contrarily, the MM associated protein MamA could be readily extracted by SDS and CAPS-buffer treatments. Buffer E treatments served as negative controls. (C) 3D-Structured illumination colocalization microscopy of MamB-GFP and MamJ334-426mCherry in ΔmamJ indicate that residues 334-426 mediate MM targeting of MamJ (see also Fig. 3, Fig. 4, fig. S2A and Supplementary text). Micrographs are maximum intensity projections. MamB-GFP is colored in green; MamJ-mCherry is colored in magenta; White-colored regions in the overlay indicate colocalization. BF, bright field image. (Scale bars: 2 µm) (D) Interaction analysis of MamJ using a BACTH assay based on reconstitution of CyaA adenylate cyclase activity in the cya− strain E. coli BTH101. Fusion of proteins to the fragmented catalytic CyaA domains T25 and T18 from Bordetella pertussis only confers cAMPdependent expression of a lacZ reporter gene upon protein-protein interaction-mediated reconstitution of CyaA. lacZ expression is indicated by blue color formation and enhanced growth on X-Gal containing M63 maltose-mineral salts agar. The MamJ C-terminal domain interacts with MmsF and MmxF but not MamF in BACTH assays. (Ei) Domain structures of putative MFP substrates MamJ, MamD and Mms5 as well as the MamD-homologs MamG and Mms6 which are not targeted to the MM in an MFP-dependent manner (fold change (FC) of the proteomic analysis is indicated; nd, not detected). Alanine and glycine contents of the alanine- (black) and glycine-rich (purple) domains are given in %, respectively. Numbers above and below the proteins represent domain boundaries or protein lengths, respectively. Multiple aa sequence alignment of the glycine-rich domains. Amino acid residues are shaded using the Rasmol color scheme. A consensus motif is given below the alignment. The numbers at the beginning and end of each line indicate the position of the first and last amino acid of the respective protein within the alignment. Asterisks indicate position of conserved glycin residues. (Eii) Interaction analysis of MamD, Mms5, MamG, Mms6 C-terminal domains using a BACTH assay based on reconstitution of CyaA adenylate cyclase activity in the cya− strain E. coli BTH101. BACTH assay showing the interaction between the C-terminal domains of MamD and Mms5 with MamF, MmsF, and MmxF. Note that the highly homologous proteins MamG and Mms6 do not interact with the MFPs. (See also Fig. 4) 1382 1383 1384 1385 1386

Fig. S5. Verification of the proteomic data (A) Volcano plot showing the log2-fold depletion (red) or enrichment (blue) of proteins in purified magnetosome fractions of ΔF3 compared to the WT as determined by LC-MS/MS. Circles represent non-MAI encoded proteins, non-significant proteins are depicted by grey or black symbols, respectively. Proteins with predicted N-terminal signal peptide sequences and high enrichment in ΔF3 used for mCherry-labelling are highlighted (red circles). MmeA, whose abundance remained unaffected in ΔF3 MM fractions and used as a reference for mCherry labelling is also highlighted. Coloring otherwise identical to Fig. 3D. (See also Data table S3) (B) 3D-Structured illumination microscopy confirms that mCherry-labeled proteins (see Figure S3D) with potential secretory signal sequences are localized near the cell surface in the MSR WT. Except for the known MAP MmeA no protein showed MC localization. BF, bright field image. (Scale bars: 2 μM) (See also Figure 3)

Table S1. Strains used in this study.

Strain E. coli DH5α WM3064 BTH101 M. gryphiswaldense Wild type (WT)

WT::mamD-GFP WT::mamR-GFP

WT::mms5-GFP

 WT::mamG-GFP WT::Ptet-mmeA-mCherry WT::MSR1_01270-mCherry WT::MSR1_09970-mCherry WT::MSR1_21580-mCherry

WT::MSR1_03900-mCherry ΔM05 ΔM05::mamJ-GFP ΔmamJ ΔmamJ::mamJ-GFP ΔmamJ::mamJ1-80-GFP ΔmamJ::mamJ81-333-GFP ΔmamJ::mamJ334-426-GFP ΔmamJ::mamJΔ335-343-GFP ΔmamJ::mamJΔ335-348-GFP ΔmamJ::mamJΔ335-353-GFP ΔmamJ::mamJΔ335-358-GFP ΔmamJ::mamJΔ378-426-GFP Genotype/Description Host for cloning: F- Φ80lacZΔM15 Δ(lacZYA-argF) U169 recA1 endA1 hsdR17 (rk-, mk+) phoA supE44 thi-1 gyrA96 relA1 λDonor strain for conjugation: thrB1004 pro thi rpsL hsdS lacZ∆M15 RP4-1360 ∆(araBAD)567 ∆dapA1341::[erm pir (wt)] BACTH assay reporter strain:

 F- cya-99 araD139 galE15 galK16 rpsL1 (StrR)

hsdR2 mcrA1 mcrB1 MSR-1 R3/S1; RifR, SmR spontaneous mutant WT with mamD-GFP after Tn5 insertion, TetR WT with mamR-GFP after Tn5 insertion, TetR WT with mms5-GFP after Tn5 insertion, TetR WT with mamG-GFP after Tn5 insertion, TetR

 WT with mmeA-mCherry

after Tn5 insertion, KanR

 WT with MSR1_01270-mCherry

after Tn7 insertion, CamR

 WT with MSR1_09970-mCherry

after Tn7 insertion, CamR

 WT with MSR1_21580-mCherry

after Tn7 insertion, CamR

 WT with MSR1_03900-mCherry

after Tn7 insertion, CamR Non-magnetic mutant with large MAI operon deletions ΔM05 with mamJ-GFP after Tn5 insertion, TetR mamJ deletion strain ΔmamJ with mamJ-GFP after Tn5 insertion, TetR ΔmamJ with mamJ1-80-GFP after Tn5 insertion, TetR ΔmamJ with mamJ81-333-GFP after Tn5 insertion, TetR ΔmamJ with mamJ334-426-GFP after Tn5 insertion, TetR ΔmamJ with mamJΔ334-343-GFP after Tn5 insertion, TetR ΔmamJ with mamJΔ334-348-GFP after Tn5 insertion, TetR ΔmamJ with mamJΔ334-353-GFP after Tn5 insertion, TetR ΔmamJ with mamJΔ334-358-GFP after Tn5 insertion, TetR ΔmamJ with mamJΔ378-426-GFP after Tn5 insertion, TetR Reference/Source

Invitrogen William Metcalf at

UIUC

Euromedex (Schultheiss et al.,

2004) This study This study This study This study This study This study This study This study

This study (Dziuba et al., 2020)

This study (Scheffel et al., 2006)

This study This study This study This study This study This study This study This study This study Strain ΔmamJ::mamJΔ396-426-GFP ΔmamJ::mamJΔ407-426-GFP ΔmamJ::mamJΔ417-426-GFP ΔmamJ::mamB-GFP ΔmamJ::mamB-GFP:: Plac-mamJ334-426-mCherry ΔmamF ΔmmsF ΔmmxF ΔmamF/mmsF ΔmamF/mmxF ΔmmsF/mmxF ΔmmsF/mmxF::mamJ334-426-GFP ΔF3 ΔF3::mamF ΔF3::mmsF ΔF3::mmxF ΔF3::6His-mmsF ΔF3::Ptet-mCherry-mamK ΔF3::Ptet-mScarlet-I-mamY ΔF3ΔmamJ::mamJ-GFP ΔF3::GFP-mamF ΔF3::GFP-mmsF ΔF3::mamD-GFP ΔF3::mamR-GFP ΔF3::mms5-GFP ΔF3::mamG-GFP ΔF3::mamD-GFP::mmsF ΔF3::mamR-GFP::mmsF ΔF3::mms5-GFP::mmsF ΔF3::mamG-GFP::mmsF ΔF3::mamD-GFP::mamF ΔF3::mamR-GFP::mamF This study This study This study

This study ( Lohße et al., 2014 ) ( Lohße et al., 2014 )

This study ( Lohße et al., 2014 )

This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study Strain ΔF3::mms5-GFP::mamF ΔF3::mamG-GFP::mamF ΔF3::mamD-GFP::mmxF ΔF3::mamR-GFP::mmxF ΔF3::mms5-GFP::mmxF ΔF3::mamG-GFP::mmxF ΔF3::mmsF D34N, D36N, D37N, E38N ΔF3::mmsF D34N, R35Q, D36N, D37N, E38N ΔmamKY ΔF3Δmms5ΔmamKΔmamY::GFPmmsF ΔA13ΔmmxFΔmms5 ΔA13ΔmmxFΔmms5ΔmamR ΔA13ΔmmxFΔmms5ΔmamR::mmsF Genotype/Description ΔF3 with mms5-GFP and mamF after Tn5 insertion, KanR, TetR ΔF3 with mamG-GPF and mamF after Tn5 insertion, KanR, TetR ΔF3 with mamD-GFP and mmxF after Tn5 insertion, KanR, TetR ΔF3 with mamR-GFP and mmxF after Tn5 insertion, KanR, TetR ΔF3 with mms5-GFP and mmxF after Tn5 insertion, KanR, TetR ΔF3 with mamG-GPF and mmxF after Tn5 insertion, KanR, TetR ΔF3 with mmsF D34N, D36N, D37N, E38N after Tn5 insertion, TetR ΔF3 with mmsF D34N, R35Q, D36N, D37N, E38N after Tn5 insertion, TetR mamK, mamY double deletion strain mamF, mmsF, mmxF, mms5, mamK, mamY codeletion strain; ΔF3Δmms5ΔmamKΔmamY with GFP-mmsF after Tn5 insertion, TetR mamGFDC, mms5, mms6, mamXY codeletion strain mamGFDC, mms5, mms6, mamXY, MamR codeletion strain ΔA13ΔmmxFΔmms5ΔmamR with mmsF after Tn5 insertion, TetR This study This study This study This study This study This study

This study (Toro-Nahuelpan et al., 2019)

This study (Zwiener et al., 2021)

This study This study

Table S2. Plasmids used in this work.

Plasmid pUT18C pUT18C-zip; pKT25-zip pUT18C-mamJ pKT25C-mamK pUT18N-mmsF pKT25C-mmsF pUT18N-mmxF pKT25C-mmxF pUT18N-mamF pKT25C-mamF pUT18N-mms6CT pUT18N-mamGCT pUT18N-mamDCT pUT18N-mms5CT pBAM-Tet-mamD-GFP pBAM-Tet-mamR-GFP pBAM-Tet-mms5-GFP pBAM-Tet-mamG-GFP pBAM-Tet-mamJ-GFP pBAM-Tet-mamJ1-80-GFP pBAM-Tet-mamJ81-333-GFP Description BACTH vector designed to express a given polypeptide fused in frame at its N-terminal end with T18 fragment; ColE1 ori; AmpR BACTH vector designed to express a given polypeptide fused in frame at its C-terminal end with T18 fragment; ColE1 ori; AmpR BACTH vector designed to express a given polypeptide fused in frame at its N-terminal end with T25 fragment; p15A ori; KanR BACTH vector designed to express a given polypeptide fused in frame at its C-terminal end with T25 fragment; p15A ori; KanR Derivatives of pUT18C and pKT25 with a 114 bp DNA fragment encoding for a leucine zipper (positive control for BACTH) BACTH plasmid coding for T18-MamJ BACTH plasmid coding for T25-MamK BACTH plasmid coding for T25-MamY BACTH plasmid coding for MamJ334-426-T25 BACTH plasmid coding for MmsF-T18 BACTH plasmid coding for T25-MmsF BACTH plasmid coding for MmxF-T18 BACTH plasmid coding for T25-MmxF BACTH plasmid coding for MamF-T18 BACTH plasmid coding for T25-MamF BACTH plasmid coding for Mms6-T18 BACTH plasmid coding for MamG-T18 BACTH plasmid coding for MamD-T18 BACTH plasmid coding for Mms5-T18 R6K ori, Tn5, AmpR, TetR, PmamG-mamD-GFP R6K ori, Tn5, AmpR, TetR, PmamG-mamR-GFP R6K ori, Tn5, AmpR, TetR, PmamG-mms5-GFP R6K ori, Tn5, AmpR, TetR, PmamG-mamG-GFP R6K ori, Tn5, AmpR, TetR, PmamG-mamJ-GFP R6K ori, Tn5, AmpR, TetR,

 PmamG-mamJ1-80-GFP

R6K ori, Tn5, AmpR, TetR,

 PmamG-mamJ81-333-GFP

Reference/Source (Karimova et al.,

1998) (Karimova et al.,

1998) (Karimova et al.,

1998) (Karimova et al.,

1998) (Karimova et al.,

1998) (Toro-Nahuelpan

et al., 2019) (Toro-Nahuelpan

et al., 2019) (Toro-Nahuelpan et al., 2019) This study This study This study This study This study This study This study This study This study This study This study (Uebe et al.,

2018a) This study This study

This study (Kolinko et al.,

2014) This study This study Plasmid pBAM-Tet-mamJ335-426-GFP pBAM-Tet-mamJΔ335-343-GFP pBAM-Tet-mamJΔ335-353-GFP pBAM-Tet-mamJΔ334-359-GFP pBAM-Tet-mamJΔ378-426-GFP pBAM-Tet-mamJΔ396-426-GFP pBAM-Tet-mamJΔ407-426-GFP pBAM-Tet-mamJΔ417-426-GFP pBAM-Tet-GFP-mmsF pBAM-Tet-GFP-mamF pBAM-Tet-mmsF pBAM-Tet-mamF pBAM-Tet-mmxF pBAM-Tet-mmsF D34N, D36N, D37N, E38N pBAM-Tet-mmsF D34N, R35Q, D36N, D37N, E38N pBAM1-mmsF pBAM1-mamF pBAM1-mmxF pBAM2-His6-mmsF pBAM2-MSR1_09970-mCherry pBAM2-MSR1_01270-mCherry pBAM2-MSR1_21580-mCherry pBAM2-MSR1_03900-mCherry pBAM2-mamJ334-426-mCherry pBAM160-mamA-GFP pBAM160-mCherry-mamK pBAM160-mmeA-mCherry pBAM160-mScarlet-I-mamY pORFM-GalK-MCS pOR-ΔmamR pOR-∆mamJ Description R6K ori, Tn5, AmpR, TetR,

 PmamG-mamJ335-426-GFP

R6K ori, Tn5, AmpR, TetR,

 PmamG-mamJΔ335-343-GFP

R6K ori, Tn5, AmpR, TetR,

 PmamG-mamJΔ335-348-GFP

R6K ori, Tn5, AmpR, TetR,

 PmamG-mamJΔ335-353-GFP

R6K ori, Tn5, AmpR, TetR,

 PmamG-mamJΔ335-358-GFP

R6K ori, Tn5, AmpR, TetR,

 PmamG-mamJΔ378-426-GFP

R6K ori, Tn5, AmpR, TetR,

 PmamG-mamJΔ396-426-GFP

R6K ori, Tn5, AmpR, TetR,

 PmamG-mamJΔ407-426-GFP

R6K ori, Tn5, AmpR, TetR,

 PmamG-mamJΔ417-426-GFP

R6K ori, Tn5, AmpR, TetR, PmamG-GFP-mmsF R6K ori, Tn5, AmpR, TetR, PmamG-GFP-mamF R6K ori, Tn5, AmpR, TetR, PmamG-mmsF R6K ori, Tn5, AmpR, TetR, PmamG-mamF R6K ori, Tn5, AmpR, TetR, PmamG-mmxF R6K ori, Tn5, AmpR, TetR, PmamG-mmsF R6K ori, Tn5, AmpR, TetR, PmamG-mmsF R6K ori, Tn5, AmpR, KanR, PmamG-mmsF R6K ori, Tn5, AmpR, KanR, PmamG-mamF R6K ori, Tn5, AmpR, KanR, PmamG-mmxF p15A ori, Tn5, AmpR, KanR PmamG-6His-mmsF p15A ori, Tn7, AmpR, CamR,

 PG-MSR1_09970-mCherry

p15A ori, Tn7, AmpR, CamR,

 PG-MSR1_01270-mCherry

p15A ori, Tn7, AmpR, CamR,

 PG-MSR1_21580-mCherry

p15A ori, Tn7, AmpR, CamR,

 PG-MSR1_03900-mCherry

p15A ori, Tn7, AmpR, CamR,

 Plac-mamJ334-426-mCherry

R6K ori, Tn5, AmpR, KanR, Ptet-mamA-GFP R6K ori, Tn5, AmpR, KanR, Ptet-mCherry-mamK R6K ori, Tn5, AmpR, KanR,

 Ptet-mmeA-mCherry

R6K ori, Tn5, AmpR, KanR, Ptet-mScarlet-I-mamY universal in-frame deletion/in-frame fusion vector for GalK counterselection; npt galK TetR mobRK2 pORFM-GalK-MCS with ∆mamR deletion construct pORFM-GalK-MCS with Reference/Source

This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study

This study (Pfeiffer et al.,

2020) This study This study

This study (Raschdorf et al.,

2014) This study This study

Plasmid pOR-mamB-GFP Description ∆mamJ deletion construct pORFM-GalK-MCS with

Table S3. PCR Primers used in this work.

Oligonucleotides dmmxF_do_for dmmxF_do_rev dmmxF_up_rev dmmxF_up_for MamR_Kpn_for MamR-GFP_SacI_rev MmsF_Kpn_for MamF_Sac_rev MmeA_NdeI_for MmeAoStop_SacI_rev mms6_C-term2_for mms6_C-term2_rev mamG_C-term_for mamG_C-term_rev mamD_C-term2_for mamD_C-term2_rev mms5_C-term2_for mms5_C-term2_rev MmsF_for MmsF_rev MamF_for MamF_rev MmxF_for MmxF_rev MamY_oSTop_Kpn_rev MamY_Nde_for mamJ_1-80.FOR mamJ_335-392.FOR mamJ_335-392.REV mamJ_1-80.REV MamJ-E294_NdeI_for MamJ-E333_KpnI_rev MamJ-A334_NdeI_for MamJ-F364_KpnI_rev MamJ-G365_NdeI_for MamJ-N395_KpnI_rev MamJ-V396_NdeI_for MamJ-CAR_for MamJ-CAR_rev MamJ1-80_NdeI_rev MamJ344S_for MamJ349A_for MamJ354V_for MamJ359L_for Kpn-10G_for MamJ416S_rev MamJ406T_rev MamJ386V_rev

Sequence CCGAAGCCGTCTGATGCTGAACGGATCCGCCGCGCC CATCATCGGCAAGGAGGCGCAGA GGATCCGTTCAGCATCAGACGGCTTCGGCCGCAGC CCGCAGCTCCCGCAGCTGGC GGTACCATGATTTGGACAGCAGTGATC GAGCTCTCACTTATACAGCTCGTCCATGC GGTACCATGGTTGAAGCAATCCTTCG GAGCTCTCAGATCAGGGCGACTACAT CATATGGCCCTGAAGACGACCCATGC GAGCTCGCGAACGTAGACCTGCACCT AGGCTGTCGGCGGCACCATC GGGACAGCGCGTCGCGCAG AGGCTGTCGGCAGCACCTTG GAGCAGGCTCGGCGGAGGC AGGCCGCTGGCAGCGCCATG GTTCCTCGCCGACAGCCGCCAGAA AGACCAGCAGCAGCGCCATGCT GGACGGCTTCGGCCGCAGC AGGTTGAAGCAATCCTTCGG GGATCCGGTCGGCCACCCA AGGCCGAGACTATTTTGATC GGATCAGGGCGACTACATG AGATCGCACAGACTGTCGGG GGATGCGGTCGGCGATATA GATCGGTACCCGCATCGGAGATGGGGGTT CGTACATATGTTGATGAACTTTGTCAACAATG GTAACATATGGCAAAAAACCGGCGTGATCGC GCCTCATATGACCCGCCAGCCTAACAAGAT GCTGGGTACCTTTATTCTTATCTTCAGCATCACAT ACGAGGTACCGTCCTGGGAACGAATGGG GATCAGCATATGGAGAGCGTTGCATCAGCG GCTACGGGTACCTTCGACCGCCACAGCAAC GATCAGCATATGGCCACCCGCCAGCCTAAC GCTACGGGTACCGAAAATCCCCCCCAGAAGGT GATCAGCATATGGGCGTCGCCGGATCGGCG GCTACGGGTACCATTGCTTCCAACGAGGCGGCCTCC GATCAGCATATGGTGGTCGCCGGGACGCGC GATCAGCATATGCCTGTGCCGGTTGCCGATC GCTACGGGTACCTTCGACCGCCACAGCAAC GCTGACATATGGTCCTGGGAACGAATGGG TCAGTTAAGAAGCGCGCCC GCCCCGGTTCAGGAAGTTC GTTCCCGTGGAAGACCTTCTG CTTCTGGGGGGGATTTTCG GGTACCGGAGGCGGAGGCG CGATGAACAACTACCGCAACTTACCTC AGTTTGCGCCAACCGGCGC GACCACTCCATCGACGAATCCG

Source This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study This study 1412 1414 1415 1416 1417 1418 1419 1420 1421 1422 Oligonucleotides MamJ376T_rev 6His-MmsF-NdeI_for 6His-MamF-NdeI_for MmsF_SphI_rev MamF_SphI_rev MamJS410_SmaI_rev MamJA334_for MamJ_rev Mms5_SacI_rev Mms5_KpnI_for MmxF_SacI_rev MmxF_KpnI_for MamF_KpnI_for MamF_SacI_rev MmsF_SacI_rev MmsF_KpnI_for

Sequence AGTGAACACACCCCGCACCG AATTCATATGCATCACCACCATCATCACGTTGAAGCAATCCTTCGGAGC AATTCATATGCATCACCACCATCATCACGCCGAGACTATTTTGATC TAAGGCATGCTCAGATCCGGTCGGCCACCCA TAAGGCATGCTCAGATCAGGGCGACTACATG CCCGGGCACTTACCTCGATAGTTTGCGC AGGCCACCCGCCAGCCTAAC GTTTATTCTTATCTTCAGCATCACATTTCG GAGCTCTCAGACGGCTTCGGCCGC GGTACCATGGCTGGTGGGACCGCG GAGCTCTCAGATGCGGTCGGCGAT GGTACCATGATCGCACAGACTGTCG GGTACCATGGCCGAGACTATTTTG GAGCTCTCAGATCAGGGCGACTAC GAGCTCTCAGATCCGGTCGGCCAC GGTACCATGGTTGAAGCAATCCTTCG

 Data tables (separate excel files)

Data table S1. Accession numbers of proteins used for phylogenetic analyses of MFPs (Figure S1A) and the whole Tic20/HOTT superfamilies (Figure S1E) Data table S2. Accession numbers of proteins used for CLANS analysis (Figure 1A) and structural/multiple sequence alignments (Figure 1 B and C) Data table S3. Data for magnetosome crystal size (Figure 2A), magnetosome number per cell (Figure 2B), qMNA (Figure 2D and E), and magnetotaxis analyses (Figure 2F). Data table S4. List of proteins enriched or depleted in the ΔF3 MM fraction as determined by quantitative proteomic analysis. Related to Figures 3D and S3D.

Data table S5. Data for magnetosome crystal size analysis for Figure 5B.

 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 Kalyaanamoorthy, S. , Minh , B.Q. , Wong , T.K.F. , Haeseler , A. von, and Jermiin , L.S. ( 2017 ). ModelFinder: fast model selection for accurate phylogenetic estimates . Nat Methods 14 , 587 - 589 . https://doi.org/10.1038/nmeth.4285. Kasmati , A.R. , Töpel , M. , Patel , R. , Murtaza , G. , and Jarvis , P. ( 2011 ). Molecular and genetic analyses of Tic20 homologues in Arabidopsis thaliana chloroplasts . The Plant Journal 66 , 877 - 889 . https://doi.org/10.1111/j. 1365 - 313X . 2011 . 04551 .x. Katoh , K. , Rozewicki , J. , and Yamada , K.D. ( 2019 ). MAFFT online service: multiple sequence alignment, interactive sequence choice and visualization . Briefings in bioinformatics 20 , 1160 - 1166 . https://doi.org/10.1093/bib/bbx108. Kikuchi , S. , Bédard , J. , Hirano , M. , Hirabayashi , Y. , Oishi , M. , Imai , M. , Takase , M. , Ide , T. , and Nakai , M. ( 2013 ). Uncovering the protein translocon at the chloroplast inner envelope membrane . Science (New York, N.Y.) 339 , 571 - 574 . https://doi.org/10.1126/science.1229262. Kikuchi , S. , Oishi , M. , Hirabayashi , Y. , Lee , D.W. , Hwang , I. , and Nakai , M. ( 2009 ). A 1- megadalton translocation complex containing Tic20 and Tic21 mediates chloroplast protein import at the inner envelope membrane . The Plant cell 21 , 1781 - 1797 . https://doi.org/10.1105/tpc.108.063552. Kim , S. , Jeon , T.-J., Oberai , A. , Yang , D. , Schmidt , J.J. , and Bowie , J.U. ( 2005 ). Proceedings of the National Academy of Sciences of the United States of America 102 , 14278 - 14283 . https://doi.org/10.1073/pnas.0501234102. Kolinko , I. , Lohße , A. , Borg , S. , Raschdorf , O. , Jogler , C. , Tu , Q. , Pósfai , M. , Tompa , E. , Plitzko , J.M. , and Brachmann , A. , et al. ( 2014 ). Biosynthesis of magnetic nanostructures in a foreign organism by transfer of bacterial magnetosome gene clusters . Nature nanotechnology 9 , 193 - 197 . https://doi.org/10.1038/nnano. 2014 . 13 . Komeili , A. , Li , Z. , Newman , D.K. , and Jensen , G.J. ( 2006 ). Magnetosomes are cell membrane invaginations organized by the actin-like protein MamK . Science (New York, N.Y.) 311 , 242 - 245 . https://doi.org/10.1126/science.1123231. Kouranov , A. , Chen , X. , Fuks , B. , and Schnell , D.J. ( 1998 ). Tic20 and Tic22 are new components of the protein import apparatus at the chloroplast inner envelope membrane . The Journal of cell biology 143 , 991 - 1002 . https://doi.org/10.1083/jcb.143.4.991. Kovács-Bogdán , E. , Benz , J.P. , Soll , J. , and Bölter , B. ( 2011 ). Tic20 forms a channel independent of Tic110 in chloroplasts . BMC Plant Biol 11 , 133 . https://doi.org/10.1186/ 1471 - 2229-11-133. Kyte , J. , and Doolittle , R.F. ( 1982 ). A simple method for displaying the hydropathic character of a protein . Journal of molecular biology 157 , 105 - 132 . https://doi.org/10.1016/ 0022 - 2836 ( 82 ) 90515 - 0 . Letunic , I. , and Bork , P. ( 2019 ). Interactive Tree Of Life (iTOL) v4: recent updates and new developments . Nucleic acids research 47, W256 - W259 . https://doi.org/10.1093/nar/gkz239. Liu , H. , Li , A. , Rochaix , J. -D. , and Liu , Z. ( 2023 ). Architecture of chloroplast TOC-TIC translocon supercomplex . Nature 615 , 349 - 357 . https://doi.org/10.1038/s41586-023-05744-y. Lohße , A. , Borg , S. , Raschdorf , O. , Kolinko , I. , Tompa , E. , Pósfai , M. , Faivre , D. , Baumgartner , J. , and Schüler , D. ( 2014 ). Genetic dissection of the mamAB and mms6 operons reveals a gene set essential for magnetosome biogenesis in Magnetospirillum gryphiswaldense . Journal of bacteriology 196 , 2658 - 2669 . https://doi.org/10.1128/JB.01716- 14 . ( 2020 ). An automated oxystat fermentation regime for microoxic cultivation of Magnetospirillum gryphiswaldense . Microb Cell Fact 19 , 206 . https://doi.org/10.1186/s12934- 020-01469-z. Uebe , R. , and Schüler , D. ( 2016 ). Magnetosome biogenesis in magnetotactic bacteria . Nat Rev Microbiol 14 , 621 - 637 . https://doi.org/10.1038/nrmicro. 2016 . 99 . Uebe , R. , Schüler , D. , Jogler , C. , and Wiegand , S. ( 2018b ). Reevaluation of the Complete Genome Sequence of Magnetospirillum gryphiswaldense MSR-1 with Single-Molecule RealTime Sequencing Data . Genome announcements 6 . https://doi.org/10.1128/genomeA. 00309 - 18 . van Dooren , G.G. , Tomova , C. , Agrawal , S. , Humbel , B.M. , and Striepen , B. ( 2008 ). Toxoplasma gondii Tic20 is essential for apicoplast protein import . Proceedings of the National Academy of Sciences of the United States of America 105 , 13574 - 13579 . https://doi.org/10.1073/pnas.0803862105. Varadi , M. , Anyango , S. , Deshpande , M. , Nair , S. , Natassia , C. , Yordanova , G. , Yuan , D. , Stroe , O. , Wood , G. , and Laydon , A. , et al. ( 2022 ). AlphaFold Protein Structure Database: massively expanding the structural coverage of protein-sequence space with high-accuracy models . Nucleic acids research 50, D439 - D444 . https://doi.org/10.1093/nar/gkab1061. Viana , A.A.B. , Li , M. , and Schnell , D.J. ( 2010 ). Determinants for stop-transfer and post-import pathways for protein targeting to the chloroplast inner envelope membrane . The Journal of biological chemistry 285 , 12948 - 12960 . https://doi.org/10.1074/jbc.M110.109744. Vögtle , F.-N. , Burkhart , J.M. , Gonczarowska-Jorge , H. , Kücükköse , C. , Taskin , A.A. , Kopczynski , D. , Ahrends , R. , Mossmann , D. , Sickmann , A. , and Zahedi , R.P. , et al. ( 2017 ). Landscape of submitochondrial protein distribution . Nat Commun 8 , 290 . https://doi.org/10.1038/s41467-017-00359-0. Wiśniewski , J.R. ( 2016 ). Quantitative Evaluation of Filter Aided Sample Preparation (FASP) and Multienzyme Digestion FASP Protocols . Analytical chemistry 88 , 5438 - 5443 . https://doi.org/10.1021/acs.analchem.6b00859. Žárský , V. , and Doležal , P. ( 2016 ). Evolution of the Tim17 protein family . Biology direct 11 , 54 . https://doi.org/10.1186/s13062-016-0157-y. Zimmermann , L. , Stephens , A. , Nam , S.- Z. , Rau , D. , Kübler , J. , Lozajic , M. , Gabler , F. , Söding , J. , Lupas , A.N. , and Alva , V. ( 2018 ). A Completely Reimplemented MPI Bioinformatics Toolkit with a New HHpred Server at its Core . Journal of molecular biology 430 , 2237 - 2243 . https://doi.org/10.1016/j.jmb. 2017 . 12 .007. Zwiener , T. , Mickoleit , F. , Dziuba , M. , Rückert , C. , Busche , T. , Kalinowski , J. , Faivre , D. , Uebe , R. , and Schüler , D. ( 2021 ). Identification and elimination of genomic regions irrelevant for