A novel septal protein of multicellular heterocystous cyanobacteria is associated with the divisome

1140 | Molecular Microbiology. 2020;113:1140–1154.wileyonlinelibrary.com/journal/mmi 1 | INTRODUCTION Cyanobacteria are a monophyletic group that includes species having diverse morphologies, ranging from unicellular species to filamentous and cell-differentiating strains (Garcia-Pichel, Zehr, Bhattacharya, & Pakrasi, 2020; Rippka, Stanier, Deruelles, Herdman, & Waterbury, 1979). The major characteristics of prokaryotic multicellularity are the manifestation of cell-cell attachment sites, mechanisms of cell-cell communication (e.g., through junctional complexes) and a division of labor among cells in the colony (Claessen, Rozen, Kuipers, Søgaard-Andersen, & Wezel, 2014; Flores, Herrero, Forchhammer, & Maldener, 2016). The filamentous cyanobacterium Anabaena sp. strain PCC 7120 (hereafter Anabaena) reproduces through random trichome (i.e., filament) breakage and grows by intercalary cell division; the multicellular phenotype of Anabaena is maintained through incomplete segregation of cells following division (Claessen et al., 2014; Rippka et al., 1979). Individual cells within an Anabaena filament possess an individual cytoplasmic Received: 13 November 2019 | Revised: 3 February 2020 | Accepted: 4 February 2020 DOI: 10.1111/mmi.14483 RESEARCH ARTICLE A novel septal protein of multicellular heterocystous cyanobacteria is associated with the divisome Benjamin L. Springstein 1 | Sergio Arévalo2 | Andreas O. Helbig3 | Antonia Herrero 2 | Karina Stucken 4 | Enrique Flores 2 | Tal Dagan 1 This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made. © 2020 The Authors. Molecular Microbiology published by John Wiley & Sons Ltd 1Institute of General Microbiology, Christian-Albrechts-Universität zu Kiel, Kiel, Germany 2Instituto de Bioquímica Vegetal y Fotosíntesis, CSIC and Universidad de Sevilla, Seville, Spain 3AG Proteomics & Bioanalytics, Institute for Experimental Medicine, Christian-AlbrechtsUniversität zu Kiel, Kiel, Germany 4Department of Food Engineering, Universidad de La Serena, La Serena, Chile Correspondence Benjamin L. Springstein, Institute of General Microbiology, Christian-AlbrechtsUniversität zu Kiel, Kiel, Germany. Email: [email protected]. edu Present address Benjamin L. Springstein, Department of Microbiology, Blavatnick Institute, Harvard Medical School, Boston, MA, USA Funding information Deutsche Forschungsgemeinschaft, Grant/ Award Number: STU513/2-1; Spanish Government and European Regional Development Grant, Grant/Award Number: BFU2016-77097-P and BFU2017-88202-P Abstract Cyanobacteria are unique among the eubacteria as they possess a hybrid Gram phenotype, having an outer membrane but also a comparably thick peptidoglycan sheet. Furthermore, the cyanobacterial divisome includes proteins specific for both the Gram types as well as cyanobacteria-specific proteins. Cells in multicellular cyanobacteria share a continuous periplasm and their cytoplasms are connected by septal junctions that enable communication between cells in the filament. The localization of septal junction proteins depends on interaction with the divisome, however additional yet unknown proteins may be involved in this process. Here, we characterized Alr3364 (termed SepI), a novel septal protein that interacts with the divisome in the multicellular heterocystous cyanobacterium Anabaena sp. strain PCC 7120. SepI localized to the Z-ring and the intercellular septa but did not interact with FtsZ. Instead, SepI interacted with the divisome proteins ZipN, SepF and FtsI and with the septal protein SepJ. The inactivation of sepI led to a defect in cell filament integrity, colony and cell morphology, septum size, nanopore formation and peptidoglycan biogenesis, and inability to differentiate heterocysts. Our results show that SepI plays a role in intercellular communication and furthermore indicate that SepI functions in the coordination of septal junction localization during cell division. | 1141 SPRINGSTEIN ET al. membrane and peptidoglycan (PG) sheets, yet, all cells are engulfed in a continuous outer membrane and hence are connected by a shared periplasm (Wilk et al., 2011). In the absence of combined nitrogen, Anabaena produces heterocysts, which are specialized cells where atmospheric nitrogen is fixed into a bioavailable nitrogen form (ammonia). Heterocyst differentiation includes an inhibition of oxygenic photosynthesis, hence they supply a micro-aerobic environment that increases the efficiency of nitrogen fixation (Kumar, Mella-Herrera, & Golden, 2010). As a result of this cell specialization, a metabolic communication between heterocysts and vegetative cells in the Anabaena filament is essential to exchange photosynthates and combined nitrogen products between the two cell types (Herrero, Stavans, & Flores, 2016). Thus, the unique characteristics of cyanobacterial multicellularity hinge upon mechanisms for inter-cellular communication. The communication among cells in the Anabaena filament is supported by cell-cell joining structures that are termed “septal junctions” and are analogous to the eukaryotic gap junctions (Flores et al., 2016; Wilk et al., 2011). The Anabaena septal junctions are putatively composed of several proteins, among which are SepJ, FraC and FraD, whose deletion abrogated intercellular molecular exchange, hence they were considered putative septal junction components (Flores et al., 2007; Merino-Puerto, Mariscal, Mullineaux, Herrero, & Flores, 2010; Mullineaux et al., 2008; Nürnberg et al., 2015). The role of FraD in septal junctions recently gained additional support with the observation of septal junction structures that link the cytoplasms of neighboring cells and contain FraD (Weiss, Kieninger, Maldener, Forchhammer, & Pilhofer, 2019). Septal junctions mediate molecular diffusion between adjacent cells and are likely regulated by gating, thus predating the gated mechanism of eukaryotic gap junctions (Flores, Nieves-Morión, & Mullineaux, 2018; Nieves-Morión, Mullineaux, & Flores, 2017; Nürnberg et al., 2015; Weiss et al., 2019). The Anabaena septal junctions traverse the septal PG through holes, termed “nanopores”, whose formation is dependent on PG amidases (AmiC) and the PG binding protein SjcF1 (Bornikoel et al., 2017; Rudolf et al., 2015). The inactivation of sepJ, fraC and fraD is associated with filament fragmentation mainly under diazotrophic conditions, indicating a function of these proteins in filament integrity (Merino-Puerto et al., 2010; Nayar, Yamaura, Rajagopalan, Risser, & Callahan, 2007). Furthermore, heterocyst-formation is strictly dependent on the SepJ function but not on FraC and FraD (Flores et al., 2007; Merino-Puerto et al., 2010). SepJ is thought to interact with the PG sheet by means of its coiled-coil (CC) domain. Furthermore, SepJ forms multimers, hence it might form a structure that is reminiscent of the gap junctions buildup, which consist of two connexin hexamers (Ramos-León, Mariscal, Battchikova, Aro, & Flores, 2017). The localization of SepJ to the septa in Anabaena is dependent on its interaction with the divisome during cell division, nonetheless no direct interaction was found between SepJ and FtsZ (Ramos-León, Mariscal, Frías, Flores, & Herrero, 2015). Cell division in bacteria is governed by the divisome, a multi-protein complex that is formed at the future division site upon polymerization of FtsZ (the prokaryotic tubulin homolog), which assembles into the Z-ring below the cytoplasmic membrane. FtsZ is tethered to the cytoplasmic membrane by an interaction with proteins such as FtsA and ZipA that are linked to the cytoplasmic membrane through an amphipathic helix (FtsA) or a transmembrane helix (ZipA). The divisome governs chromosome segregation and PG remodeling as well as septum invagination and ultimately cell septation (den Blaauwen, Hamoen, & Levin, 2017). In Anabaena, FtsZ is an essential cellular component that includes an N-terminal peptide specific to filamentous heterocystous cyanobacteria (CorralesGuerrero et al., 2018; Zhang, Hugenin, Friry, Huguenin, & Friry, 1995). Additionally, several other components of the Anabaena divisome have been identified, including FtsK, FtsW, FtsQ and FtsI. Although generally characterized as Gram-negative bacteria, cyanobacteria possess conserved Gram-positive bacteria-specific cell division proteins such as DivIVA (also known as Cdv3) and SepF, suggesting that a clear Gram classification is not applicable for that phylum (MacCready, Schossau, Osteryoung, & Ducat, 2017; Marbouty, Saguez, Cassier-Chauvat, & Chauvat, 2009; Miyagishima, Wolk, & Osteryoung, 2005). Furthermore, cyanobacteria lack the FtsZ membrane tethering proteins FtsA and ZipA and possess cyanobacterial-specific cell division proteins such as ZipN (also known as Ftn2) and Ftn6 (Koksharova & Wolk, 2002; Marbouty, Mazouni, Saguez, Cassier-Chauvat, & Chauvat, 2009; Marbouty, Saguez, et al., 2009; Miyagishima et al., 2005). The Anabaena ZipN was found to recruit FtsZ to the cytoplasmic membrane and interact with a wide array of divisome proteins, including Ftn6, SepF, FtsW, FtsX, FtsQ and FtsI (Camargo et al., 2019). ZipN also interacts with SepJ and regulates its septal localization. As such, ZipN links cell division-related processes to septum formation, intercellular communication and filament integrity (Camargo et al., 2019). In this work, we aimed to identify additional proteins that play a role in the coordinated process of cell division and the formation of septal junctions. Our results identified a novel coiled-coil-rich protein (CCRP) in Anabaena that is localized in the septum, interacts with the divisome, and has a putative role in filamentation. 2 | RESULTS 2.1 | SepI is ubiquitous in filamentous heterocystforming cyanobacteria A survey for proteins containing a high coiled-coil content (using COILS) and hence a putative function in Anabaena multicellularity revealed the gene alr3364 (termed here sepI), which includes two N-terminal coiled-coil domains and a FtsK-like domain at the C-terminus (Figure 1a). Notably, SepI shows resemblance to parts of SepJ (Supplementary Figure S1). In addition to the two N-terminal coiled-coils, the C-terminal non-coiled-coil domain of SepI (from aa 241 on) has a similar amino acid composition to the linker domain of SepJ (Herrero et al., 2016) and is also rich in proline (15%), serine (11%) and threonine (8%), suggesting similar involvements in cellular 1142 | SPRINGSTEIN ET al. processes of both, the C-terminal SepI domain and the linker domain of SepJ. Unlike SepJ, which is a polytopic multipass transmembrane (TM) protein (Figure 1b), we found no evidence for TM domains in the SepI sequence (predicted with TMHMM). A computational prediction of SepI localization suggested that it is localized in the cytoplasm (predicted using PSORTb) and interacts with the cytoplasmic membrane through its C-terminus (predicted using Gneg-mPLoc and PSIPRED). The lack of a signal peptide and of an amphipathic helix in SepI (as predicted using SignalP-5.0 and AmphipaSeek respectively) are also consistent with a putative cytoplasmic localization (Figure 1b). A survey for homologs to sepI in cyanobacterial genomes showed that sepI is ubiquitous in heterocystous cyanobacteria and sepI homologs are rarely found in non-heterocystous species (Figure 1c, Supplementary File 1). Notably, none of the investigated cyanobacterial genomes of subsection I contained a sepI homolog (Supplementary File 1). Furthermore, the genomic neighborhood of sepI homologs, including the upstream genes alr3361, alr3362 and alr3363, is highly conserved in heterocystous cyanobacteria (Figure 1c; Supplementary File 1). Although these genes could form an operon together with sepI, a specific transcription start-site has been previously identified for sepI (Mitschke, Vioque, Haas, Hess, & Muro-Pastor, 2011). Downstream of sepI, ORFs alr3365 and alr3366 are located in the same orientation. Nonetheless, RT-PCR tests did not support the occurrence of common transcripts (Supplementary Figure S2a,b). 2.2 | SepI localizes to the mid-cell and the septa To investigate the intracellular localization of SepI, we expressed a SepI-GFP fusion (in which the GFP was added to the C-terminus of SepI) from the predicted native promoter (PsepI, encompassing FIGURE 1 Domain architecture, predicted localization and conservation of SepI. (a) Schematic depiction of the domain architecture of SepI. The CC domain, the FtsK-like domain and the domain similar to the linker domain of SepJ are indicated. (b) Schematic depiction of subcellular localization of selected divisome and septal junction proteins from Anabaena. Localization of FtsZ, ZipN and FtsI is inferred from (Camargo et al., 2019). N indicates the N-terminus and C indicates the C-terminus of the respective protein. For practical reasons, indication of the C-terminus for FtsZ and SepI was omitted. Close localization of SepI to the cytoplasmic membrane indicates a likely interaction with the cytoplasmic membrane. The precise localization of SepJ is still under debate with the N-terminal CC and linker domain localized in the periplasm or cytoplasm (Ramos-León et al., 2017, 2015). (c) Gene order in the neighborhood of genes encoding SepI homologs in cyanobacterial genomes. Note that alr3361, alr3362, and alr3363 have homologs in all cyanobacteria genomes tested with only few exceptions (i.e., they are universal). Homologs of alr3524 and alr3522 are abundant in all organisms (see full lists and accession numbers in Supplementary File 1). Organisms are marked according to their morphological phenotype: I, unicellular cyanobacteria that divide by binary fission; II, unicellular organisms that can divide in multiple planes; III, filamentous; IV, filamentous and heterocyst forming; V, heterocyst forming and true branching or multiseriate filaments D E F | 1143 SPRINGSTEIN ET al. 602 bp upstream of sepI; see Supplementary Figure S2a). Additionally, to observe the putative effects from protein overexpression, we also expressed SepI-GFP from the comparably strong copper-inducible PpetE promoter. Both constructs were carried on plasmid pRL25C. Regardless of the promoter, SepI-GFP localized to the mid-cell in ring forms reminiscent of the Z-ring and to the septa between neighboring cells (Figure 2a). Notably, the septal localization of SepI-GFP between the vegetative cells and heterocysts seemed to be restricted to the vegetative cell side as it was never observed at the heterocyst side (zoom-in in Figure 2a). Further examination of 150 samples of SepI-GFP localization in the neighborhood of heterocysts showed that none of the samples contained apparent septal SepI-GFP signals at the heterocysts side (this is unlike what we have observed for example for SepJ-YFP, which shows clear localization at the heterocyst-side; see also the localization of SepJ-GFP in Flores et al., 2007). We are aware, however, that the limited resolution of epifluorescence microscopy is likely not sufficient in order to fully discard the presence of SepI in the heterocysts. 2.3 | SepI is involved in filament integrity as well as cell and colony morphology To further elucidate the functional role of SepI in Anabaena, we generated a ΔsepI::C.S3 mutant strain (Supplementary Figure S2c). We obtained two independent fully segregated mutant strains, both showing the same phenotypic properties. We also verified by semi-quantitative RT-PCR that the genomically inserted C.S3 cassette (Elhai & Wolk, 1988) did not affect the expression of the FIGURE 2 SepI is associated with the Z-ring and the septa. Merged GFP fluorescence and chlorophyll autofluorescence (red) and bright field micrographs of (a) Anabaena WT or (b) ΔsepI::C.S3 mutant cells expressing SepI-GFP from PsepI or PpetE promoters. Cells were either grown on BG11 or BG110 plates. White triangles indicate Z-ring formation that simultaneously nucleated from more than one site. Blue triangles highlight the SepI-GFP localization between vegetative cells and heterocysts, which seems restricted to the vegetative cell. White arrows indicate heterocysts. Scale bars: 5 µm D E 1144 | SPRINGSTEIN ET al. downstream genes alr3365 and alr3366 (Supplementary Figure S2b). In the presence of nitrate, the ΔsepI mutant showed no detectable growth defect in liquid culture in comparison to the wild-type strain (Supplementary Figure S2d). However, the ΔsepI mutant was unable to grow diazotrophically (i.e., on BG110 plates or in BG110 liquid medium; Figure 3e, Supplementary Figure S2e). Strain ΔsepI readily fragmented when incubated in BG110 medium (Figure 3a,b), suggesting a function of SepI in filament integrity. Notably, we observed that the ΔsepI mutant never differentiated heterocysts. Additionally, the ΔsepI mutant was characterized by an abnormal cell shape and colony morphology under standard growth conditions (i.e., in BG11 liquid or on BG11 plates; Figure 3c), with an increase in cell size (Figure 3d). The colony morphology and cell morphology defects could be complemented by the expression of sepI or a sepI-gfp fusion from PsepI in the ΔsepI mutant (Figure 3c). Furthermore, SepI-GFP was localized in the septa and as rings at the mid-cell in the ΔsepI mutant background (Figure 2b). Additionally, the cell area of the WT expressing sepI-gfp from PsepI was decreased in comparison to the ΔsepI mutant or the WT (Figure 3d), which adds support to a role of SepI in cell size regulation. Diazotrophic growth of ΔsepI was not complemented by the wild type sepI gene expressed from PsepI. We note that since we were able to isolate two independent mutant clones with the same phenotype it is unlikely that the inability of diazotrophic growth was due to a secondary mutation. An alternative explanation for the lack of complementation may be a negative effect of sepI overexpression on heterocyst differentiation by its placement in a replicative plasmid with an unequal copy number per individual cell in the complemented strain, as has been shown previously for other genes (e.g., Nieves-Morión, Lechno-Yossef et al., 2017). FIGURE 3 SepI is involved in filament integrity and cell and colony morphology. (a) Merged chlorophyll autofluorescence and bright field micrographs of Anabaena wild type and ΔsepI mutant strain grown in BG11 liquid medium or 24 hr after transfer to BG110 liquid medium. Scale bars: 10 µm. (b) Filament lengths (in cells per filament) of Anabaena WT and ΔsepI mutant strain grown in BG11 or 24 hr after transfer to BG110. (c) Micrographs of Anabaena WT and ΔsepI mutant filaments as well as of ΔsepI mutant complemented with pRL25C-based plasmids carrying PsepI::sepI or PsepI::sepI-gfp grown in liquid BG11 medium (left). Right, micrographs showing colony morphology in solid medium of the respective strains. Scale bars: (left) 5 µm or (right) 500 µm. (d) Cell surface area in µm2 of Anabaena WT, ΔsepI mutant, ΔsepI mutant expressing SepI from PsepI and Anabaena WT expressing SepI-GFP from PsepI. Error bars indicate standard error of the mean (SEM) (WT: n = 758; ΔsepI: n = 1285; ΔsepI + PsepI::sepI: n = 1487; WT + PsepI::sepI-gfp: n = 749). Values indicated with asterisks are significantly different from the WT. ****: p < .0001 (Dunnett’s multiple comparison test and one-way ANOVA). (e) Spot assays of Anabaena WT and ΔsepI mutant strain. Both strains were grown on BG11 plates, resuspended in BG11 liquid medium and adjusted to an OD750 of 0.4 and spotted in triplicates of serial 10-fold dilutions on BG11 or BG110 plates. Note: re-streaking of green colonies that arose in higher dilutions of the ΔsepI mutant grown on BG110 still exhibited a diazotrophic growth defect and were not spontaneous revertant mutants D E G H F | 1145 SPRINGSTEIN ET al. 2.4 | SepI is involved in septal nanopore formation and intercellular molecular transfer A defect in filament integrity is commonly observed in mutants of genes encoding proteins that function in the septal junctions (e.g., FraC, FraD and SepJ; Herrero et al., 2016). Additionally, mutant strains of such genes are characterized by a decreased number of nanopores in the septal PG disks and a decreased efficiency of intercellular molecular transfer (Nürnberg et al., 2015). To test if the ΔsepI mutant is characterized by similar properties, we isolated the PG layer (i.e., the sacculi) from Anabaena wild type and the ΔsepI mutant. In addition, we measured the diffusion through septal junctions using the fluorescent markers calcein and 5-carboxyfluorescein (5-CF). In the ΔsepI mutant, two different perforation types in the septal PG disks were observed, some with a tiny diameter (in “sealed disks”) and others with a very large diameter (defining “unsealed disks”) (Figure 4a). Although we cannot exclude that unsealed disks correspond to not yet fully divided cells and would eventually become normal septal disks (see Figure S5 in (Bornikoel, Staiger, Madlung, Forchhammer, & Maldener, 2018)), it should be noted that we rarely observed such large perforations in FIGURE 4 Altered nanopore formation in the ΔsepI mutant. (a) Representative transmission electron microscopy micrographs of Anabaena WT and ΔsepI mutant septal PG disks showing nanopore formations. Scale bars 0.5 µm. Note the presence in the mutant of “sealed” (middle panels) and “unsealed” (right panels) septal disks. (b) Number of nanopores per septal PG disk in Anabaena WT and ΔsepI mutant grown in BG11 or BG110. Number of recorded septal discs (n): Anabaena WT BG11, n = 11; ΔsepI BG11, n = 13; Anabaena WT BG110, n = 9; ΔsepI BG110, n = 3. Only “sealed” septal disks were counted. (c) Nanopore diameter in µm for Anabaena WT and ΔsepI mutant grown in BG11 or BG110. Number of nanopores analyzed (n): Anabaena WT BG11, n = 100; ΔsepI BG11, n = 26; Anabaena WT BG110, n = 108; ΔsepI BG110, n = 7. Data correspond only to “sealed” PG disks. For (b) and (c): the significance of the differences in nanopore number and size between mutant and wild type were assessed by the Student’s t test: **p < .005, ***p < .0005, ****p < .0001. ns indicates no significant difference. Error bars indicate standard deviation D E F 1146 | SPRINGSTEIN ET al. the wild type. The septal disks with those large perforations were not included in the measurement of nanopore numbers and diameter. The number of nanopores was strongly reduced in the ΔsepI mutant and we never observed more than four nanopores in any of the isolated sacculi, regardless of the availability of combined nitrogen. This observation should be compared to about 40 nanopores per septal PG disk in the wild-type grown under the same conditions (Figure 4a,b). Nanopore diameter was unaffected in the ΔsepI mutant in BG11 medium but slightly (although significantly) reduced in the ΔsepI mutant under nitrogen deprivation (Figure 4c). Using FRAP analysis, we studied the intercellular transfer of calcein and 5-CF between vegetative cells of filaments incubated in the absence of combined nitrogen for 48 hr. The recovery rate constants (R) for calcein and 5-CF were decreased in the ΔsepI mutant compared to the wild type (Figure 5a). Notably, the decrease in R was accompanied by an increased number of noncommunicating cells (R < 0.01 s−1) for the two markers (Figure 5b). We suggest that the two populations of cells in the ΔsepI mutant may correspond to cells with sealed or unsealed septal PG disks (see Discussion). 2.5 | SepI localization is explained by interaction with divisome proteins and SepJ The absence of TM domains in SepI led us to hypothesize that its localization to the Z-ring and septal junctions is due to an interaction with other proteins. To examine that possibility, we tested for an interaction of SepI with several divisome or septal junctionrelated proteins using an array of bacterial adenylate cyclase twohybrid (BACTH) assays. Our results showed that SepI specifically interacts with ZipN (a cyanobacterial FtsZ membrane-tethering protein), SepF, SepJ and weakly with FtsI (Figure 6a). No interaction was detected with FtsZ, FtsQ, FtsK, FtsW, FtsK, FtsE, MreB, Ftn6, MinC, FraC, FraD or the filament forming and septal-localized CCRP All4981 (Supplementary Figure S3). Notably, SepI also weakly interacted with the CC domain of SepJ (Figure 6a), suggesting that the CC domains of SepI and SepJ are involved in the interaction of both proteins. We further confirmed the interaction of SepI with ZipN (50% coverage), SepF (37% coverage) and SepJ (6% coverage) using extracts from Anabaena cells expressing PsepI::sepI-gfp and anti-GFP co-immunoprecipitations followed by LC-MS/MS analytics (Figure 6b). This approach also identified several other CC domain-containing protein interactors of SepI (Figure 6b). Inspired by its interaction with SepJ, we also investigated the localization pattern of a SepJ-YFP fusion protein in the ΔsepI mutant but found no difference in SepJ-YFP localization compared to the wild type (Supplementary Figure S4). Additionally, we also did not find alterations of FtsZ localization in the ΔsepI mutant, regardless of the availability of combined nitrogen (Supplementary Figure S5). The Z-ring localization of SepI remains therefore best explained by its interaction with ZipN and SepF, while its localization at the septa could be explained by an interaction with SepJ. FIGURE 5 Decreased fluorescence recovery rate and noncommunicating cells in the ΔsepI mutant. (a) Recovery rate constants of fluorescence recovery after photobleaching (FRAP) experiments for calceinand 5-CF-labelled vegetative cells of Anabaena WT and the ΔsepI mutant 48 hr after transfer to BG110 liquid medium. Number of recorded bleached cells (n): calcein PCC 7120 (WT), 57; calcein alr3364 (sepI), 27; 5-CF PCC 7120 (WT), 51; 5-CF alr3364 (sepI), 40. Data are depicted as boxplots (Krzywinski & Altman, 2014a). Orange, quartile group 2 (from Q1 to median); blue, quartile group 3 (from median to Q3). Median values (M) are indicated for each strain and marker. The significance of the difference between mutant and wild type was assessed by the non-parametric Mann-Whitney U test (Krzywinski & Altman, 2014b) (P values are indicated on the top). (b) Distribution of cells of Anabaena WT (blue) and the ΔsepI mutant (orange) showing different R values. The percentage of cells within R value intervals of 0.01 s−1 (i.e., <0.01; from 0.01 to 0.02; from 0.02 to 0.03; etc.; >0.15) was plotted for the transfer of calcein (top) and 5-CF (bottom). The insets show the distribution of cells with R < 0.01 s-1 and R > 0.01 s-1. The significance of the difference of this distribution between the mutant and the wild type was assessed by the χ2 test (P indicated on top) D E | 1147 SPRINGSTEIN ET al. 2.6 | SepI is involved in PG processing Because of the interaction of SepI with several divisome-associated proteins, its localization to the Z-ring and the apparent alterations in cell and colony morphology in the ΔsepI mutant, we next examined whether PG biogenesis is altered in the ΔsepI mutant. For this, we stained Anabaena wild type and the ΔsepI mutant with HADA, a fluorescently labeled D-amino acid, and a fluorescently labeled vancomycin (Van-FL). HADA stains active sites of transpeptidase activity (Hsu et al., 2017) while Van-FL binds to inserted and non-crosslinked PG precursors (Daniel & Errington, 2003). We found that HADA staining was unaffected in the ΔsepI mutant even upon nitrogen stepdown-induced cell morphological changes and filament fragmentation (Figure 7a). Like HADA, Van-FL readily stained the septa and the mid-cell regions of the Anabaena filament. Notably, however, Van-FL more readily stained the lateral cell walls of the ΔsepI mutant than the WT, especially when FIGURE 6 SepI-interacting proteins. (a) BACTH assays of E. coli BTH101 cells co-expressing indicated T25 and T18 translational fusions of all possible pair-wise combinations of SepI, FtsI, ZipN, SepF, SepJ and the CC domain of SepJ (SepJ-CC). E. coli cells were subjected to beta-galactosidase assay in triplicates from three independent colonies grown for 2 d at 20°C. Quantitative values are given in Miller units, and the mean results from three independent colonies are presented. Negative: N-terminal T25 fusion construct of the respective protein co-transformed with empty pUT18C. Positive: Zip/Zip control. Error bars indicate standard deviations (n = 3). Values indicated with asterisks are significantly different from the negative control. **p < .01, ***: p < .001, ****: p < .0001 (Dunnett’s multiple comparison test and one-way ANOVA). (b) Excerpt of possible specific interactors of SepI-GFP identified by co-immunoprecipitation, their description/annotation and a description of specific features of the interactors. The full list of interactors is shown in Supplementary File 2 D E 1148 | SPRINGSTEIN ET al. incubated in BG110 medium. In the Anabaena WT, lateral cell wall staining could be readily observed in heterocysts rather than in vegetative cells similar to what has previously been reported for HADA (Zhang, Lin, Xing, & Zhang, 2018). In contrast, 48 hr after nitrogen-stepdown, Van-FL frequently stained the whole cell wall of a large proportion of ΔsepI mutant cells (Figure 7a), indicating alterations in nascent PG synthesis. As we consistently observed an increase in cell diameter in the ΔsepI mutant (Figure 3c,d and Figure 7a), we measured mid-cell diameter (determined as HADAstained rings) in Anabaena wild type and the ΔsepI mutant and found that mid-cell diameter was significantly increased in the ΔsepI mutant (Figure 7b). Thus, consistent with a role of SepI in the Anabaena divisome, these observations suggest that SepI influences PG biogenesis. 3 | DISCUSSION Based on the presence of an outer membrane, cyanobacteria are generally classified as Gram-negative bacteria. However, cyanobacteria also possess an unusually thick PG sheet between the cytoplasmic and outer membranes that bears similarities with that of Gram-positive bacteria (Hahn & Schleiff, 2014; Hoiczyk & Hansel, 2000). In accordance with their ambiguous cell envelope phenotype, the cyanobacterial divisome has previously been shown to include a combination of Gram-positive-like, Gram-negative-like and cyanobacterial-specific proteins (Koksharova & Wolk, 2002; MacCready et al., 2017; Marbouty, Saguez, et al., 2009; Miyagishima et al., 2005). Here, we identified a novel cyanobacterial divisome-associated protein, SepI (Figure 8), which is ubiquitous in filamentous, FIGURE 7 SepI is involved in PG biogenesis and septum diameter determination. (a) Merged chlorophyll autofluorescence and HADA or Van-FL fluorescence micrographs of Anabaena WT and ΔsepI mutant strain grown in BG11 or 48 hr after transfer to BG110. White arrows indicate heterocysts. Scale bars: 5 µm. (b) Measurement of mid-cell diameter determined from HADA-labeled Anabaena WT and ΔsepI mutant cells grown in BG11. Lower part shows a depiction of measured rings within an Anabaena filament. Green indicates a fully developed PG-ring and the purple arrow indicates the measured distance that is plotted on the upper part. Error bars indicate standard deviations (Anabaena WT: n = 600; ΔsepI mutant: n = 630) D E