Phylogenetic origin and sequence features of MreB from the wall-less swimming bacteria Spiroplasma

Spiroplasma are wall-less bacteria which belong to the phylum Tenericutes that evolved from Firmicutes including Bacillus subtilis. Spiroplasma swim by a mechanism unrelated to widespread bacterial motilities, such as flagellar motility, and caused by helicity switching with kinks traveling along the helical cell body. The swimming force is likely generated by five classes of bacterial actin homolog MreBs (SMreBs 1-5) involved in the helical bone structure. We analyzed sequences of SMreBs to clarify their phylogeny and sequence features. The maximum likelihood method based on around 5,000 MreB sequences showed that the phylogenetic tree was divided into several radiations. SMreBs formed a clade adjacent to the radiation of MreBH, an MreB isoform of Firmicutes. Sequence comparisons of SMreBs and Bacillus MreBs were also performed to clarify the features of SMreB. Catalytic glutamic acid and threonine were substituted to aspartic acid and lysine, respectively, in SMreB3. In SMreBs 2 and 4, amino acids involved in inter- and intra-protofilament interactions were significantly different from those in Bacillus MreBs. A membrane-binding region was not identified in most SMreBs 1 and 4 unlike many walled-bacterial MreBs. SMreB5 had a significantly longer C-terminal region than the other MreBs, which possibly forms a protein-protein interaction. These features may support the functions responsible for the unique mechanism of Spiroplasma swimming.


Introduction
The bacterial phylum Tenericutes that evolved from the phylum Firmicutes is comprised of the class Mollicutes, which includes the genera Spiroplasma and Mycoplasma [1]. Species of these genera are characterized by small genome size, being mostly parasitic or commensal, and absence of peptidoglycan. In addition, they cannot equip conventional machineries for bacterial motility, such as flagella or type IV pili, due to the lack of a peptidoglycan layer. Instead, they have developed three unique motility systems: mobile type gliding, pneumoniae type gliding, and Spiroplasma swimming [2].
Spiroplasma have a helical cell shape and infect plants and arthropods [3,4]. 3 They swim by rotating the cell body in viscous environments, including host tissues.
To rotate the cell body, Spiroplasma change the helicity handedness at the front end by forming a kink, and propagate the kink along the cell body to back (Fig. 1A&B) [5,6,7,8]. The helical cell shape originates from a flat ribbon structure (Fig. 1C) extending along the innermost line of the cell, and kink propagation is likely caused by a structural change of the ribbon [5,9,10,11]. Electron microscopy and proteome analyses have revealed that the flat ribbon is composed of fibril, a Spiroplasma specific protein that forms filaments, and MreB proteins, which are bacterial actin homologs ( Fig. 2A) [9,10,11]. Tenericutes lack respiration pathway to generate membrane potential, and produce ATP through glycolysis and arginine fermentation [12]. Then, the energy for cell activities is mostly supplied from ATP rather than membrane potential in many species. Actually, the energy for mobile and pneumoniae type gliding mechanisms is supplied from ATP [2]. Spiroplasma swimming may also be based on ATP energy. Regarding the component proteins of the ribbon-shaped swimming machinery, fibril is related to adenosyl homocysteine nucleosidase that cannot hydrolyze ATP [13]. On the other hand, MreB is well known as an ATPase like other actin superfamily proteins [14,15]. These facts 4 suggest that the MreBs are employed in the Spiroplasma swimming.
Generally, as seen in eukaryotic actin, the MreB molecule has a U-shaped structure that can be divided into four domains ( Fig. 2A) [15]. The groove of the U-shaped structure functions as a nucleotide-binding pocket, such as for ATP and GTP ( Fig. 2A, S3A) [15]. A previous study reported that MreB polymerizes into non-helical filaments with juxtaposed subunits in which two protofilaments interact antiparallelly in the presence of nucleotides (Fig. 2B) [15]. Molecular dynamics (MD) simulations suggest that the inter-and intra-protofilament interactions are comprised of four helices, two strands, and three loops (H3, H5, and H8 shown in Fig. 2A and Fig. S3C for inter-protofilament interaction; H9, S12, S13 and loops between S6-H1, S9-S10, and S10-S11 shown in Fig. 2A and Fig. S3D for intra-protofilament interaction) [16]. Unlike the eukaryotic actin filament, MreB filaments bind to the cell membrane directly [17]. A previous study reported that many MreBs have two consecutive hydrophobic amino acids for membrane binding in the hydrophobic loops in domain IA, and some MreBs of Gram-negative bacteria have an amphipathic helix at the N-terminus as an additional factor for membrane binding ( Fig. 2A) [17].
In various phyla of rod-shaped bacteria, MreB is well-conserved in the genomes and is responsible for cell shape maintenance [18]. The MreB filament binds to the membrane to form an elongasome complex with other proteins, which precisely locates the peptidoglycan-synthesizing enzymes to retain the rod shape of the cells [18]. Many bacterial species represented by Escherichia coli have a single copy of the mreB gene on their genome [18]  10.1 [20].

Structure prediction and sequence analyses
All monomeric MreB structures in this study were predicted using Rosetta comparative modeling by templating a structure of Caulobacter crescentus MreB (PDB ID: 4CZJ), using Robetta software [15,21]. Sequence comparisons among amino acid sequences were performed by Clustal Omega [22]. Duplicated SMreBs 1, 4, and 5 in the Apis group species were incorporated into the analyses. Seven of the 19 amino acid sequences contained in SMreB3 of the Apis group had no annotation. Since these proteins were coded in a position suggesting SMreB3 as an ortholog on the genome, they were included in the analyses. SMreB3 of the S. melliferum strain CH-1 (AHA83319.1) was excluded from the analyses because it was approximately 40% shorter than the other SMreB3 sequences. Two sequences, MreB4 of S. eriocheiris strain CCTCC_M_207170 (AHF58342.1) and a protein from S. culicicola, which is unannotated but can be classified to MreB3 (WP_025363622.1), were not used for the analyses because the initiation codons were miss-assigned. The predictions of secondary structures and amphipathic helices were performed using PSIPRED 4.0 and AmphipaSeeK software, respectively [23,24]. Sequence identity and similarity were calculated as the ratio of amino acids with identity and strong similarity defined by the Gonnet PAM 250 matrix over the total amino acid number excluding gap regions.

Evolutional relationships between Spiroplasma and other MreBs
We analyzed the phylogenetic relationships among SMreBs and other MreBs. We  (Table S1) (Data set 1) [3,4,25]. Based on these sequences, we constructed a maximum likelihood phylogenetic tree (Fig. 3, S7). We adopted this phylogenetic tree because the significant difference was not found among the topologies of a maximum likelihood phylogeny using 4,000 sequences, the Neighbor-Joining phylogeny using 5,002 sequences, and the phylogeny shown in Fig. 3 (and Fig. S7).
The largest radiation of the phylogenetic tree consisted of MreB family proteins of conventional bacteria, which were further divided into Gram-negative and positive 9 bacteria. Previous studies of MreB family proteins have primarily focused on proteins in these radiations [18]. In our phylogenetic tree, an MreB radiation of candidate phyla radiation (CPR) [1], which is composed of Parcubacteria and Microgenomates superphyla, was formed next to the radiation of conventional MreBs. The CPR MreBs were split into two clades. Three candidate species of Parcubacteria in candidate phyla Yanofskybacteria, Taylorbacteria, and Candidatus Izimaplasma, belonging to Tenericutes phylum (HMreB) was identified between clades of SMreB and BMreBH (Fig. 3, S1). Haloplasma contractile is known to perform characteristic motility via tentacle-like structures extending from a coccoidal body [26]. HMreB may be involved in this characteristic motility of Haloplasma [26,27], while the motility of Izimaplasma has not been reported yet.

Development of Spiroplasma swimming
Based on the phylogeny results, we assessed the evolutional background prior to Spiroplasma acquiring swimming motility using a phylogenetic tree of 16S rRNA, including newly obtained Spiroplasma species (Fig. 1D 1D, 3). A previous study suggested that Haloplasma and Spiroplasma acquired their MreBs by independent gene duplications [27]. However, HMreB was located closer 1 1 to SMreB and MreBH than to the other clades in our phylogenetic tree (Fig. 3, S1).
Accordingly, Spiroplasma likely acquired SMreBs through the duplication of MreBH (Fig. 1D). Fibril, another component of the ribbon structure of Spiroplasma cells, was likely acquired after Spiroplasma and Izimaplasma separated, because it is specific for Spiroplasma [9,10,13,27].
Amino acids that are specifically conserved among SMreBs may be essential for intrinsic functions of SMreBs. Sequences of Citri group SMreBs were more conserved than those of the Apis group, corresponding to the phylogenetical spread among Spiroplasma species represented in the 16S rRNA phylogeny (Table S3,
Moreover, the consecutive hydrophobic amino acids cannot be found in 46 out of 88 sequences of the Apis group SMreBs 1, 3, 4, and 5 (Fig. S6). These SMreBs may not directly interact with the membrane. We next focused on the N-terminal amphipathic helix found in some MreBs of Gram-negative bacteria ( Fig. 2A). All N-terminal regions of the Citri group SMreB3, except for S. melliferum strain AS576 (QCO24517.1), were predicted to take amphipathic helices ( Fig. 4 purple box, Fig.   S4, S6). This suggests that Citri group SMreB3 proteins also have affinity to the membrane through the helix (Fig. 2F).
In particular, SMreB5 had the longest C-terminal sequence of all SMreBs (Table S5).
Such terminal extension is a feature reminiscent of a bacterial tubulin homolog, FtsZ, featured with a long unstructured region. This is the interaction region for several proteins, including FtsA, ZipA, and ZapD involved in Z-ring formation, which causes cell constriction for bacterial cytokinesis [32]. The extension of the SMreB C-termini may play a role in the protein-protein interactions (Fig. 2F).