ARTICLE https://doi.org/10.1038/s41467-020-18777-y OPEN Immunodominant proteins P1 and P40/P90 from human pathogen Mycoplasma pneumoniae David Vizarraga 1, Akihiro Kawamoto 2,3, U. Matsumoto4, Ramiro Illanes 1, Rosa Pérez-Luque1, Jesús Martín1, Rocco Mazzolini5, Paula Bierge6, Oscar Q. Pich 6,7, Mateu Espasa8, Isabel Sanfeliu8, Juliana Esperalba9, Miguel Fernández-Huerta9, Margot P. Scheffer10, Jaume Pinyol7, Achilleas S. Frangakis 10, Maria Lluch-Senar5, Shigetarou Mori 11, Keigo Shibayama11, Tsuyoshi Kenri 11, Takayuki Kato 2,3, Keiichi Namba 2,12,13, Ignacio Fita1, Makoto Miyata 4,14 ✉ & David Aparicio 1 ✉ 1234567890():,; Mycoplasma pneumoniae is a bacterial human pathogen that causes primary atypical pneu- monia. M. pneumoniae motility and infectivity are mediated by the immunodominant proteins P1 and P40/P90, which form a transmembrane adhesion complex. Here we report the structure of P1, determined by X-ray crystallography and cryo-electron microscopy, and the X-ray structure of P40/P90. Contrary to what had been suggested, the binding site for sialic acid was found in P40/P90 and not in P1. Genetic and clinical variability concentrates on the N-terminal domain surfaces of P1 and P40/P90. Polyclonal antibodies generated against the mostly conserved C-terminal domain of P1 inhibited adhesion of M. pneumoniae, and serology assays with sera from infected patients were positive when tested against this C-terminal domain. P40/P90 also showed strong reactivity against human infected sera. The archi- tectural elements determined for P1 and P40/P90 open new possibilities in vaccine devel- opment against M. pneumoniae infections. 1 Instituto de Biología Molecular de Barcelona (IBMB-CSIC), Parc Científic de Barcelona, Baldiri Reixac 10, 08028 Barcelona, Spain. 2 Graduate School of Frontier Biosciences, Osaka University, 1-3 Yamadaoka, Suita, Osaka 565-0871, Japan. 3 Institute for Protein Research, Osaka University, 3-2 Yamadaoka, Suita, Osaka 565-0871, Japan. 4 Graduate School of Science, Osaka City University, Osaka 558-8585, Japan. 5 EMBL/CRG Systems Biology Research Unit, Centre for Genomic Regulation (CRG), The Barcelona Institute of Science and Technology, Dr Aiguader 88, 08003 Barcelona, Spain. 6 Laboratori de Recerca en Microbiologia i Malalties Infeccioses, Institut d’Investigació i Innovació Parc Taulí (I3PT), Hospital Universitari Parc Taulí, Universitat Autònoma de Barcelona, 08208 Sabadell, Spain. 7 Departament de Bioquímica i Biologia Molecular, Institut de Biotecnologia i Biomedicina, Universitat Autònoma de Barcelona, 08193 Bellaterra, Barcelona, Spain. 8 Departament de Microbiologia, Hospital Universitari Parc Taulí, Universitat Autònoma de Barcelona, 08208 Sabadell, Spain. 9 Departament de Microbiologia, Hospital Universitari Vall d´Hebron, Universitat Autònoma de Barcelona, 08035 Barcelona, Spain. 10 Buchmann Institute for Molecular Life Sciences, Max-von-Laue Str. 15, 60438 Frankfurt, Germany. 11 Department of Bacteriology II, National Institute of Infectious Diseases, Tokyo, Japan. 12 RIKEN Center for Biosystems Dynamics Research and SPring-8 Center, 1-3 Yamadaoka, Suita, Osaka 565-0871, Japan. 13 JEOL YOKOGUSHI Research Alliance Laboratories, Osaka University, 1-3 Yamadaoka, Suita, Osaka 565-0871, Japan. 14 The OCU Advanced Research Institute for Natural Science and Technology (OCARINA), Osaka City University, Osaka 558-8585, Japan. ✉email: miyata@sci.osaka-cu.ac.jp; daacri@ibmb.csic.es NATURE COMMUNICATIONS | (2020)11:5188 | https://doi.org/10.1038/s41467-020-18777-y | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18777-y M ycoplasma pneumoniae is a human pathogen respon- influenzae, a vaccine for M. pneumoniae is not yet available sible for upper and lower respiratory tract infections1. despite the considerable efforts4. It is estimated that this bacterium is responsible for up M. pneumoniae binds to host target cells by means of a polar to 40% of community-acquired pneumonias in persons of all structure known as the attachment organelle, and glides in the ages2. In addition to being a severe respiratory pathogen, M. direction of this differentiated structure at a speed of ~1 µm/s pneumoniae may induce clinically significant manifestations in (Fig. 1a)5. Gliding motility in M. pneumoniae and related species extrapulmonary sites and/or immunologic effects in as many as such as Mycoplasma genitalium is caused by a unique mechanism 25% of the infections3. Unlike for other important respiratory completely different from the motility of Mycoplasma mobile, pathogens, such as Streptococcus pneumoniae and Haemophilus which has been studied in further detail6–8. Adherence of M. a P1 – P40/P90 (Nap) Attachment organelle b 2000 4000 6000 8000 bp MPN140 MPN141 (p1) MPN142 (orf6) P40 P90 RepMP4 RepMP2/3 RepMP5 c Cytoadherence epitope 1527–1547 1 60 1383 1395 1627 aa P1 29 1521 Crystal P1 60 1518 Cryo-EM P1 445 RA 1120–1141 1 25 1003 1218 aa P40 P90 23 1003 Crystal P40 P90N Transmembrane Signal peptide N-domain C-domain Cytoplasmic domain 6 x His tag helix Fig. 1 The Nap complex in M. pneumoniae. a A phase-contrast immunofluorescence microscopy image of M. pneumoniae cells (left). Locations of the Nap (surface of the attachment organelle) were visualized by using anti-P1 monoclonal antibody and Alexa Fluor® 546 secondary antibody (red). Bar = 2 µm. Three-dimensionally printed model of M. pneumoniae cell (right). The positions of the attachment organelle and the Nap structure are indicated. The cell surface is partly modeled as a transparent canopy to show the internal core structure of the organelle. The arrow indicates gliding direction. b Scheme of the p1 operon that contains three genes: MPN140, MPN141(p1), and MPN142(orf6). The locations of repetitive elements (RepMPs) in MPN141 and MPN142 are shown by colors. The regions corresponding to P40 and P90 are also shown in MPN142. c Scheme of native P1 and P40/P90 proteins and the recombinant constructs used for crystallization and cryo-EM analysis. The major cytoadherence epitope of P1 reported previously is indicated (1383–1395). The boundary between P40 and P90N proteins identified in this study (445 R/A) is indicated by arrows (Supplementary Fig. 5). Positions of 6 × His-tag used for purification of recombinants are shown by red boxes. 2 NATURE COMMUNICATIONS | (2020)11:5188 | https://doi.org/10.1038/s41467-020-18777-y | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18777-y ARTICLE pneumoniae to cells of the respiratory tract is mediated by a resolution, has agreement R and Rfree factors of 18.7 and 22.9, network of adhesins and cytoadherence accessory proteins9,10. respectively (Fig. 2a, b and Supplementary Table 1). Within this network the 170 kDa protein P1 was identified as a The structure of the ectodomain from P1 was also studied by major determinant for cytoadherence and gliding motility with single-particle analysis using cryo-EM (Figs. 1c and 2b and antibodies against P1 preventing both adhesion and motility9–12. Supplementary Fig. 2). Although P1 particles were clearly For about 40 years, P1 has been assumed to be responsible for observed in the cryo-EM micrographs, the quality of the map binding to sialic acid oligosaccharide receptors from the host was too poor to refine the atomic model owing to severe cells13. Because all these relevant properties, P1 has been orientation bias. To alleviate this problem, we collected data by attracting attention since the late 1970s, although it was soon using grids covered with a thin film of graphene oxide (GO), recognized that accessory proteins were also required for its which drastically improved the data quality27 (Supplementary functioning14. P1, together with the P40/P90 polypeptides, forms Fig. 2a). A total of 68,014 selected particles yielded a 3D EM map a transmembrane complex called the “Nap”15,16. The structure of at an overall resolution of ∼2.9 Å, according to the gold-standard the Nap complex of M. genitalium, a human urogenital pathogen Fourier shell correlation (FSC) 0.143 criterion (see “Methods”) closely related to M. pneumoniae, was recently revealed by single- (Supplementary Table 2). particle cryo-electron microscopy (cryo-EM) at 9 Å and cryo- The structure of P1 consists of a large N-terminal domain, with electron tomography (cryo-ET) at ∼15 Å resolution17. The M. Asn60 as the first residue visible in the X-ray and the cryo-EM genitalium Nap structure consists of a dimer of a P140–P110 maps, and a smaller C-terminal domain (residues Gly1395– complex protruding outward from the mycoplasma membrane Asp1521) (Fig. 2). The N-terminal domain of P1 has a β-sheet and forming a large knob with a diameter of ∼15 nm. P140 and propeller topology of seven consecutive blades or β-sheets, each P110 are the M. genitalium orthologues of P1 and P40/P90, with four antiparallel strands, except for β-sheet VII that contains respectively. Therefore, it was most unexpected to find that in M. only two strands. Interestingly, these two strands correspond genitalium the binding site for sialic acid oligosaccharides is in exactly to a peptide with important immune properties, which P110 and not in P14017,18. P1 is considered to be one of the most had been defined as the cytoadherence epitope28 (Fig. 2c). The immunodominant proteins in M. pneumoniae cells, playing a lengths of connections between the β-sheets of the propeller are major role in the immune response of infected patients and very diverse, ranging from just seven residues between sheets VI accordingly in diagnosis and epidemiologic studies3,19,20. The M. and VII, to 315 residues between sheets IV and V. Links between pneumoniae genome contains repeated regions, denominated adjacent β-strands in the β-sheets of the propeller, also vary RepMPs. The majority (75%) of RepMPs has homology with widely in length, ranging from just 11 to 219 residues. The longest MPN141 (P1) and MPN142 (P40/P90) (Fig. 1b). Homologous intra and inter β-sheet connections cluster together creating a recombination between RepMPs and either MPN141 or MPN142 crown-like structure on one side of the β-propeller opposite to the generate variability within antigenic regions of P1 and P40/P90, position where the C-terminal domain is located. In the crown respectively, providing an essential strategy to evade the immune there are 13 disordered loops, spanning 117 residues, which could host system21–23. P40/P90 consists of two polypeptides from the not be built (Fig. 2c). These disordered regions, essentially the proteolytic cleavage of a 130 kDa translate encoded by MPN14224 same in the X-ray and cryo-EM structures, are spread over the N- (Fig. 1b). Historically little attention has been dedicated to P40/ terminal domain surface (Supplementary Figs. 1 and 3a). P90 and the contribution of P40/P90 to cytoadherence is much Therefore, the N-terminal domain of P1 can be described as a less studied than for P1. Even the presence of the P40 polypeptide structured core surrounded by a fuzzy surface. The C-terminal in the Nap has been questioned and the reason for and domain contains only one disordered loop (residues Thr1483– mechanism of the MPN142 cleavage remain unknown25. Asn1495) and has an elongated fold with the same unique In the present work, we report the structures of M. pneumoniae topology (according to DALI29) as the C-terminal domain of proteins P1 and P40/P90 and of complexes of P40/P90 bound to adhesin P110 from M. genitalium. sialylated oligosaccharides. The results elucidate the antigenic The cryo-EM and X-ray maps of P1 are well defined for most organization of these immunodominant proteins allowing the of the N-terminal domain with the residues clearly identifiable, in mapping of the genetic and clinical variability. Moreover, ser- general, in both maps (Fig. 2b). However, from residue Pro1388 ological tests on different fragments of P1 reveal the recognition to the C-end, a region that includes the hinge between domains preference of IgG antibodies for conserved over variables regions, and the whole C-terminal domain, the density in the cryo-EM suggesting an immune system strategy of selecting these regions for map becomes weak, indicating that there are hinge-like move- long-term immunity. The information is considered pivotal for ments between domains. In the crystal structure, the C-terminal clarifying the motility, infectivity, and epidemics of M. pneumoniae. domain is anchored by crystal contacts and the density is clear and crisp, while in the top part of the crown, with few crystal contacts, temperature factors are high (Supplementary Fig. 3b) Results and chain tracing was difficult. Together the cryo-EM and X-ray Structure determination of P1. Crystals were obtained of the M. maps provide accurate structural information about the whole P1 pneumoniae Nap protein P1 extracellular region (residues Thr29- ectodomain, proving the mobility of domains with respect to one Asp1521). This P1 construct, empirically chosen after many crys- another and confirming the characteristic presence of a large tallization attempts, excludes the predicted N-end of the signal number of disordered loops. secretion peptide (spanning residue Met1 to Leu59 according to the structural properties prediction program Psipred26). In this con- struct are also excluded the (predicted) transmembrane and intra- Crystal structure of P40/P90N. Diffracting crystals could not be cellular regions at the C-end of P1 (residues Tyr1522 to Ala1627) obtained with constructs of the complete extracellular region of (Fig. 1c). These P1 crystals were solved simultaneously with the P40/P90, which is predicted by Psipred26 to span from Ala25 to determination of the crystal structure from the M. genitalium Pro1113 (with Met1 to Leu24 corresponding to a signal N-pep- orthologous protein P140, by averaging between crystals from both tide) (Fig. 1c). After numerous attempts, high-resolution dif- proteins (with 41% sequence identity) despite neither molecular fracting crystals were grown from a construct, referred henceforth models nor experimental phases were available (see “Methods”) as P40/P90N, spanning residues Ser23-Val1003 that, by sequence (Supplementary Fig. 1). The final refined model of P1, at 1.9 Å alignment, corresponds to the N-terminal domain of the M. NATURE COMMUNICATIONS | (2020)11:5188 | https://doi.org/10.1038/s41467-020-18777-y | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18777-y a Crown N-terminal β-propeller 90° 90° C-terminal Cytoplasmic domain b Cryo-EM X-ray c P1 β-propeller Blades I II III IV V VI VII 77 142 183 393 494 540 573 791 1107 1140 1163 1376 1384 1394 N C AGT repeats Disordered regions Fig. 2 X-ray and cryo-EM structures of P1. a Ribbon representation, with three 90° apart views, of the complete extracellular region of P1. The large N- terminal domain is organized around a seven-blade β-propeller (light orange) from where the crown (green) emerges. The C-terminal domain (cyan) is immediately followed by a predicted transmembrane helix, which requires the proximity of this domain to the cell membrane. b P1 map densities from cryo- EM (left panels) and X-ray crystallography (right panels). Insets show representative regions from the N-terminal domain (top) and the C-terminal domain (bottom). The quality of the cryo-EM density provides the clear identification of residues in the N-terminal domain (at an estimated resolution of 2.9 Å). On the contrary, the C-terminal domain is poorly defined in the cryo-EM map, indicating high flexibility with respect to the N-terminal domain. The core of the N-terminal domain and, especially, the C-terminal domain are well defined in the X-ray map, while the very top of the N-terminal domain presents high- temperature factors and was difficult to trace. c Topology of the β-propeller in P1. Loops contributing to the crown are colored green. Disordered loops are indicated with black dots and the stretch of serines (the AGT repeats) with a red cross. Strands and loops are not in scale. genitalium orthologous protein P110 (Supplementary Fig. 4). The structure, at 2.65 Å resolution, has agreement R and Rfree factors P40/P90N crystals, with two subunits in the asymmetric unit, of 21.4 and 23.4, respectively (Supplementary Table 1). were solved by molecular replacement using as a searching model The structure of P40/P90N reveals a large globular domain that the structure of P110 (PDB entry 6R3T), with a sequence identity is structurally closely related to the N-terminal domain of P110, between the extracellular regions of 44%. The refined P40/P90N with a RMSD of 1.2 Å for 742 equivalent residues18 (Fig. 3a and 4 NATURE COMMUNICATIONS | (2020)11:5188 | https://doi.org/10.1038/s41467-020-18777-y | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18777-y ARTICLE a Crown N-terminal β-propeller 90° 90° C-terminal P110 Cytoplasmic domain b P40 P90 Blades I II III IV V VI VII 26 51 63 157 478 511 560 754 805 860 874 941 950 987 C N Cleavage site R445A Sialic binding site Disordered regions c 3SL α-B-D-Glc P235 B-D-Gal W222 Neu5AC T632 P237 Fig. 3 X-ray structure of P40/P90 and of its complex with oligosaccharide 3SL. a Ribbon representation, with three 90° apart views, of the crystal structure from the P40/P90 N-terminal domain that is organized, similarly to P1, around a seven-blade β-propeller (light orange) from where it emerges the crown (pink). The sialic binding site found in P40/P90 is explicitly indicated. The predicted C-terminal domain of P40/P90 is also shown (cyan), modeled from the C-terminal domain of the orthologous adhesin P110 from M. genitalium. b Topology of the β-propeller from P40/P90. Loops contributing to the crown are colored pink. The first blade β-sheet contains only three strands in P40/P90 placing the N-end at the propeller face opposite to the one in P1. Localizations of the disordered loops and sialic binding site are indicated with black dots (black) and a star (blue), respectively. The position of the cleavage site, which results in the corresponding P40 and P90 separate polypeptides, is depicted with a pentagram (pale blue). Strands and loops are not in scale. c Binding of 3SL to the N-terminal domain of P40/P90. Ribbon representation of the P40/P90N structure in complex with oligosaccharide 3SL. The inset shows a detail of the residues shaping the binding pocket. The electron density omit map (at 0.9 sigma) corresponding to 3SL is also shown. NATURE COMMUNICATIONS | (2020)11:5188 | https://doi.org/10.1038/s41467-020-18777-y | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18777-y Supplementary Fig. 4). The sequence of P40/P90N presents two P90 form an heterodimer showing a single peak in a gel filtration long insertions with respect to P110 of more than 60 residues profile (Supplementary Fig. 5c). each (Gly297–Thr368 and Leu395–Thr462 that are referred as S1 and S2, respectively) (Fig. 3b and Supplementary Fig. 4). Both Complexes of P40/P90N with sialylated oligosaccharides. The insertions, completely disordered in the crystal structures presence of a sialic binding site in P40/P90 was confirmed by the determined, are found between β-propeller blades II and III. crystal structures of complexes of P40/P90N with oligosacchar- P40/P90, like P1, has a seven-blade β-propeller topology with ides 3SL (neuraminic acid forming an α2–3 linkage to a lactose many of the connections between the β-strands clustering monosaccharide) and 6SL (with an α2–6 linkage) (Supplementary together and forming the crown (Fig. 3b). Table 1). This result is consistent with the facts that sialylated P40 and P90 had often been considered as two independent oligosaccharides act as cell receptors for M. pneumoniae and M. proteins resulting from the posttranslational processing of the genitalium34–36, and that adhesin P110 from M. genitalium, the MPN142 gene product25,30. In our hands, the purified P40/ orthologue of P40/P90 contains a binding site specific for neur- P90 samples present a characteristic SDS-PAGE gel pattern aminic acid18. The sialic binding site of P40/P90 is in a location indicating a slow progression of cleavage in a few specific locations structurally equivalent to the one in P110, in the upper part of the (Supplementary Fig. 5a). The primary scission occurs between crown (Fig. 3a, c and Supplementary Fig. 7). Only the neuraminic Arg445 and Ala446, located toward the C-end of insertion S2 acid moiety interacts directly with the P40/P90 β-hairpin that (Fig. 3b and Supplementary Figs. 4 and 5a). The two polypeptides contains the X-Tyr/Phe–Ser/Thr motif (Leu630–Phe631–Thr632) produced by this primary cleavage, the shorter spanning from the characteristic for the binding to sialic compounds18,34. P40 N-end to Arg445 and the longer from Ala446 to the C-end, can functions as the external rim of the sialic binding site pocket, correspond to the subunits P40 and P90 that are found in which would become exposed without P40 (Supplementary mycoplasma cells30. When the primary scission site was mutated to Fig. 6b). Surface plasmon resonance (SPR) indicates that sialy- avoid cleavage, the P40/P90 variant protein (Ser445–Ser446) lated oligosaccharides bind to P40/P90 with dissociation con- presented, unexpectedly, a similar proteolysis pattern with cleavage stants (Kd) in the micromolar range for 3SL (∼8.0) and 6SL taking place between Arg455 and Ala456, still in the S2 insertion (∼2.9), similar to the values reported for P11018 (Supplementary (Supplementary Fig. 5a). Studies with mycoplasma cells had Fig. 8) although, according to in vivo studies, 6SL binds at a much reported two cleavage sites occurring between Arg445–Ala44630 lower affinity than 3SL37,38. For P1, SPR shows no binding to and (probably) Leu454–Arg45525,31, in good agreement with the sialylated oligosaccharides, in agreement with what is observed two cleavage Arg–Ala sites (RA motifs) now found in vitro. for M. genitalium17. Regardless of the extend of proteolysis, all P40/P90N samples behave as a single and sharp peak by SEC–MALS (Supplementary Fig. 5b) indicating that cleavage takes place only when folding is Structural relationships of P1 and P40/P90. The structural completed and that a unique three-dimensional structure of P40/ results reveal that P1 (1627 residues) and P40/P90 (1218 residues) P90N is maintained. Structural stability of the cleaved protein present a similar overall organization and topology in the extra- probably relies on the large interface between the P40 and P90 cellular regions. The similarity is maintained between the trans- polypeptides, with 75 hydrogen bonds, 6 salt bridges, and a buried membrane helices and the cytoplasmic regions where the surface area of 10,480 Å2, which gives an estimated free energy of sequence identity of P1 and P40/P90 is high (43%). Despite the −55 kcal with an expected 100% probability of formation different size, the close relationship of both proteins suggests that according to the Complexation Significance Score algorithm in P1 and P40/P90 have a recent common phylogenetic ancestor. PISA32 (Supplementary Fig. 6a). Addition of EDTA to the P40/ Superposition of N-terminal domains from P1 and P40/P90, with P90 samples halts proteolysis completely, which suggests a key role a sequence identity of only 12%, gives a RMSD of 3.5 Å for 361 for divalent metals. No metals were found in the crystal structures equivalent residues, which correspond mainly to residues from of P40/P90, but EDTA was required for the crystallization the β-sheets of the propellers. However, β-sheet I has only three conditions. The presence of significant amounts of Zn2+ ions in strands in P40/P90, but four in P1 and this “extra” first strand in a purified P40/P90 sample was detected by inductive coupled the first β-sheet of P1 places the N-end in the propeller face that is plasma–mass spectrometry (ICP–MS), although the concentration the opposite to the one in P40/P90 (Fig. 4). The “extra” first of protein was higher than the one found for the ion (see strand of β-sheet I from P1 starts with a large β-bulge (residues “Methods”). The possible presence in the P40/P90 samples of an 69–77) occluding the center of the propeller, contrary to P40/P90 undetected protease, which could be coming from the over- where the center is empty. Linkage between the N- and C- expression in Escherichia coli, was analyzed by SEC–MALS. terminal domains of P1 changes with respect to P40/P90, due to Different constructs of P40/P90N, with and without the C- the absence of the last two strands in β-sheet VII. Therefore, β- terminal domain, always produced a sharp peak at the molecular sheets I and VII, structurally contiguous within the propeller, weight corresponding to the construct, without any indication of present the largest differences between P1 and P40/P90. The other species. Altogether these results suggest that cleavage of P40/ hypothesis of a recent common origin is further supported by the P90 is a self-proteolytic process. However, we have not yet been close structural relationships between the C-terminal domains able to find the residues that could be responsible for this from the non-orthologous proteins P1 and P110, with a sequence enzymatic activity. Besides insertions S1 and S2, the structure of identity of 19% and a RMSD of 2.0 Å for 73 equivalent residues P40/P90 presents only three short disordered loops (residues (Supplementary Fig. 9a), The hypothesis is also reinforced by the Ser118–Gly127, Ser168–Gly175, and Ala774–Arg777) and the N- presence of the characteristic “AGT repeats” motif in the same terminal domain from P40/P90 can be seen as a well-defined loop of the propellers, between the second and third β-strands of globular structure with two long floppy pendants (Supplementary the fourth blade, of P1 and P110 (Supplementary Fig. 9b). Fig. 3a). The complete extracellular region of P40/P90 was generated by combining the structure of the N-terminal domain Mapping of epitopes and genetic variability onto P1 and P40/ with the structure of the Pro1004–Pro1113 region of P40/P90 P90. Epitopes reported for six monoclonal antibodies (mAbs) modeled according to the P110 C-terminal domain because the against P1 with inhibiting cytoadherence activities for M. pneu- sequence identity is high (68%) and the expected RMSD low (∼0.6 moniae (aN, P1.62, P1.26, M51, M58, and 6E7)28,39 (Supple- Å)33 (Fig. 3a and Supplementary Fig. 4). In solution, P1 and P40/ mentary Fig. 10) are all exposed on the surface of P1, as expected 6 NATURE COMMUNICATIONS | (2020)11:5188 | https://doi.org/10.1038/s41467-020-18777-y | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18777-y ARTICLE consistent with the fact that the epitope of this mAb is located at a P1 P40/P90 variable site. The structures of P1 and P40/P90 served as a IV IV topographic map to determine positions of known variable III III regions. RepMPs from P1 and P40/P90 are exclusively located at V V the N-terminal domains of these proteins, with the C-terminal domains remaining genetically conserved (Fig. 5b, c). II VI 1 1 II 1 2 2 3 VI 2 3 4 Nter Polyclonal antibodies against constructs of the C-terminal 1 I domain. Adherence-inhibiting antibodies recognizing the con- 4 VII I VII 2 stant C-terminal domain of P1 could be more effective in eliciting 3 Nter immunoprotection than antibodies to variable regions within the N-terminal domain. Previous studies had shown that gliding and binding of M. pneumoniae were affected by addition of a 90° 90° monoclonal antibody generated against a large recombinant IV IV fragment of P1 (residues Ala1160–Gln1518) that contains the C- III terminal domain12. To consolidate these results and to better V III V delineate the binding site of antibodies during an immune response, polyclonal antibodies were generated against two con- II structs of P1 spanning residues Lys1376–Asp1521 and VI VI 1 II 1 Ala1400–Asp1521 (see “Methods”). The shortest construct cor- 4 2 1 2 3 responds accurately to the C-terminal domain, while the longest 1 2 3 VII 2 3 Nter VII I construct also includes the hinge between domains and the two 4 Nter I strands from blade VII (Fig. 6a). Quantitative PCR (qPCR) was used to measure the amount of genomic DNA (gDNA) from the M. pneumoniae cells in solution, the fraction not adhered to the glass (see “Methods”). The M. pneumoniae cells had been pre- viously incubated with different percentages of sera from rats challenged with the P1 constructs. The analysis determines how Cter Cter much the M. pneumoniae cells are affected by the presence of the polyclonal antibodies. Without exposure to sera, 81% of the M. Fig. 4 Comparison of P1 and P40/P90. Comparison of the β-propeller pneumoniae cells attached to the plastic surface upon incubation structures and of their connections with the corresponding C-terminal of these (scraped) cells for 2 h in 24-well plates. Then, to evaluate domains in P1 (left panels) and P40/P90 (right panels). The C-terminal the possible toxicity by other components of the sera, such as the domain is tangential to the propeller ring in P1 and protrudes radially in complement that enhances the ability of antibodies and phago- P40/P90, where the C-terminal domain was modeled from the orthologous cytic cells to clear microbes, different percentages (1, 3, 6, and adhesin P110 from M. genitalium. The N-ends are situated in opposite faces 10%) of serum obtained from pre-immune bleed (PPI) of chal- of the propeller in P1 and P40/P90. The center of the propeller is occluded lenged animals were tested. The total number of recovered in P1 but empty in P40/P90. mycoplasma after incubation only decreased with the highest concentrations (6 and 10%) of PPI when compared with samples to be accessible by the antibodies (Fig. 5a). Epitopes from mAbs not treated with serum (Supplementary Fig. 11a). Thus the low aN, P1.62, P1.26, and 6E7 are located in the N-terminal domain, concentrations (1 and 3%) were selected to test the inhibitory whereas epitopes from mAbs M51 and M58 are in the C-terminal effects on the adhesion capability of M. pneumoniae cells domain. The epitope from mAb 6E7 corresponds to the strands (Fig. 6b). Toxicity effects can be subtracted, at least in part, by from the last β-sheet of the β-propeller, close to the C-terminal normalization with the corresponding values of PPI serum and domain. The mAbs, P1.62 and P1.26 are reported to bind to dual the measures for concentrations of 10% are also reported (Sup- epitopes (Supplementary Fig. 10), separated on the structure of plementary Fig. 11b). As a final checking, the specificity of anti- P1 (Fig. 5a). Although dual epitopes of P1.26 are both exposed in bodies for the C-terminal constructs was also assessed. Serum the structure of P1, one of them positioned on Asn980–Lys987 from challenged rats presented no significant differences with might become hidden in the Nap complex. their PPI when incubated with the corresponding constructs Two other reported epitopes (Trp810–Tyr817 and Phe1124– before the quantitative adhesion assay. Arg1131), recognized by sera from many of the M. pneumoniae Incubation of serum from rats challenged with the longest C- infected patients20, are located in the N-terminal domain (Fig. 5a). terminal domain construct resulted in the number of the M. The first of these exist inside the P1 molecule, while the second is pneumoniae cells attached being decreased significantly with exposed in the P1 surface, although its accessibility might also be reduction ratios with respect to PPI samples of 0.31 and 0.56 for limited in the Nap complex. These two epitopes might be added serum concentrations of 1% and 3%, respectively (p values immunodominant for antibody production; however, antibodies of 9 × 10−6 and 1 × 10−6 by applying t-test one-tailed distribu- against these epitopes may not have cytoadherence-inhibitory tion; two samples equal variance). Reduction ratios for the short activities. construct were 0.22 and 0.33, for added serum concentrations of Sequences of P1 and P40/90 exhibit sequence variations 1% and 3%, respectively (p values of 0.016 and 3 × 10−4). depending on the M. pneumoniae strain. The two major variation Together these results reveal that M. pneumoniae adhesion types, known as 1 and 2, are harbored in strains M129 and FH, capabilities are significantly reduced by polyclonal antibodies respectively22,31,40. For P1 the main source of variability is generated against C-terminal domain constructs of P1, with RepMP4 and RepMP2/3, while for P40/P90 is RepMP541. It has inhibition being highest for the longest construct that includes the been reported that mAb P1.62 only binds to type 2 P1 protein42, cytoadherence peptide. NATURE COMMUNICATIONS | (2020)11:5188 | https://doi.org/10.1038/s41467-020-18777-y | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18777-y a P1.62 AGT repeat Patient’s P1.62 sera P1.26 P1.26 Patient’s sera 90° 90° 90° aN 6E7 M51 M58 b 90° 90° P1 N-Domain C-Domain Genetic Variability 1 515 683 1367 Clinical RepMP4 RepMP2/3 AGT repeat c 90° 90° P40/P90 N-Domain C-Domain Genetic Variability ! 900 Clinical RepMP5 Fig. 5 Mapping of P1 epitopes. Variability in P1 and P40/P90. a Mapping into the P1 structure of all known epitopes (Supplementary Fig. 10). Most epitopes are located on the protein surface, although accessibility by antibodies appears to be difficult for a few of them. b Three views of the P1 surface, 90° apart from each other, with the genetically variable regions (brown), RepMP4 (light pink), RepMP2/3 (light blue), and the C-terminal domain (cyan). c The corresponding three views of the P40/P90 surface, with the genetically variable regions (brown), RepMP5 (green), and the C-terminal domain (cyan). The C-terminal domain, modeled from the structure of P110, is depicted as ribbons. Sites of genetic and clinical variability are also represented for P1 and P40/P90 in the lower part of the corresponding panels. Clinical variability refers to the variation exhibit by clinical isolates characterized so far. In turn, “genetic variability” indicates regions that are potentially variable by DNA recombination between RepMP elements in M. pneumoniae genome. Presently, genetic variability regions are wider than clinical variability. 8 NATURE COMMUNICATIONS | (2020)11:5188 | https://doi.org/10.1038/s41467-020-18777-y | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18777-y ARTICLE a C-Terminal A1400-D1521 C-Terminal K1376-D1521 Cytoadherence epitope b pAb inhibition assay 1.1 1 0.9 Relative amount (gDNA) 0.8 Serum 1% 0.7 Serum 3% 0.6 0.5 0.4 0.3 0.2 0.1 0 PPI Anti-A1400-D1521 Anti-K1376-D1521 Sera Fig. 6 Inhibition adhesion assays of polyclonal antibodies generated against constructs of the P1 C-terminal domain. a Residues corresponding to the P1 constructs used as antigens to generate polyclonal antibodies are colored in cyan in the structure of P1. The short construct (left, residues Ala1400–Asp1521) includes only the C-terminal domain. The long construct (right, residues Lys1376–Asp1521) includes also the hinge, which connects the N-terminal domain with the C-terminal domain, and the two strands from the last propeller β-sheet, also known as the cytoadherence epitope. b Sera raised against any of the two constructs inhibit adhesion of M. pneumoniae cells even at low concentrations, with 3% showing a significantly higher inhibition than 1%. Error bars represent the mean ± standard deviation. *p < 0.05 and ***p < 0.001. Source data are provided as a Source Data file. Recognition of P1 and P40/P90 by sera from M. pneumoniae For P40/P90, an additional evaluation of the sera reactivity patients. Determination of the structural organization from the P1 with recombinant proteins lacking either insertion 1 (S1), or and P40/P90 ectodomains encouraged us to assess the immuno- insertions 1 and 2 (S1S2) was performed to assess the genic activity of different constructs from these proteins using total contribution of the insertions and the loss of the proteolytic site. sera from 19 patients with positive diagnostic for M. pneumoniae Construct S1 was detected by 14 out of 19 sera (74%), the same based on serological tests (Table 1). Clinically relevant information, ratio to that of the entire P40/P90. In contrast, remarkably, there including the IgG and IgM values, was available for all sera. In is an important decrease to just seven positive sera (37%) when addition, sera from five patients that tested negative for M. pneu- the P40/P90 S1S2 construct was evaluated. Furthermore, even for moniae were also included in the study for control purposes. First, most of these seven positive sera the response was substantially the immunodominant nature of the adhesins was confirmed, reduced. Therefore, results indicate that P40/P90 is a strong revealing that 17 out of the 19 sera (90%) tested positive for P1. immunogenic protein, with insertion 2 (S2) acting as an Similarly, 14 sera (74%) tested positive for P40/P90; all positive sera immunodominant region, although recognized by sera with low for P40/P90 were positive for P1. Of note, eight sera (42%) were IgG levels. also positive for P140. This result is not unexpected because P1 and P140 show extensive antibody cross-reactivity43. Alternatively, some of the P140 positive patients might have been infected with Discussion M. genitalium in addition to M. pneumoniae. Next, the sera Adherence of M. pneumoniae to sialylated receptors on the sur- response was examined with the same two C-terminal domain face of respiratory cells was elucidated years ago34,35,37,44. Since constructs of P1 (Lys1376–Asp1521 and Ala1400–Asp1521) the identification of the critical role of P1 in M. pneumoniae described before in the generation of polyclonal antibodies. The cytoadherence, it has been accepted that this protein was number of positive sera, 5 out of 19 (26%), decreased substantially responsible for binding to sialic acid oligosaccharides11–14,45. In with respect to the full length P1, although importantly all these addition to P1, other cytadherence-accessory proteins had been positive sera correspond to the ones with highest IgG levels. reported as required for binding to sialic acids14,46. The present NATURE COMMUNICATIONS | (2020)11:5188 | https://doi.org/10.1038/s41467-020-18777-y | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18777-y study demonstrates that the binding in vitro to sialic acid 7.40 72.4 ++ ++ ++ ++ ++ ++ ++ 19 receptors relies in fact on P40/P90 (Fig. 3c and Supplementary Fig. 7), however, affinities and specificities could be tuned through other proteins or due to the density and distribution of 44.5 1.80 the oligosaccharides in the surface37,38. This important result is 18 − − − − − − − consistent with the recent finding that binding to sialic acid oli- gosaccharides is mediated by P110 in M. genitalium18. Cleavage >200 6.20 ++ ++ ++ ++ ++ ++ ++ of the P40/P90 polypeptide chain occurs in vitro at two exposed 17 Arg–Ala motives (Arg445–Ala446 and Arg455–Ala456) located in the central part of the N-terminal domain, producing two 2.80 25.2 polypeptide chains that can correspond to the P40 and ++ 16 − − − − − − P90 subunits found in vivo (Supplementary Figs. 4 and 5a). Cleavage appears to be a divalent metal ion dependent auto- 49.50 proteolysis taking place only after the folding is completed. 12.00 ++ ++ ++ ++ ++ Nevertheless, P40/P90 continues to behave as a single globular 15 − − protein structurally similar to P110. This is in agreement with the extensive interface between the P40 and P90 polypeptides indi- 64.4 >27 ++ ++ ++ ++ cating that they form a very stable ensemble (according to 14 − + − PISA32) (Supplementary Fig. 6). However, P40 is located at the outer surface part of the Nap and due to the absence of covalent 33.6 >27 ++ ++ bonds might become separated from P90 and from the Nap 13 + − − − − complex under stringent or denaturing conditions. The loss of P40 only in certain conditions could explain the discrepancies in 15.9 >27 ++ the past about whether or not P40 was part of the adhesion 12 − − − − − − complex. The sialic acid oligosaccharides binding site is located in a loop of the P90 polypeptide with P40 contributing only to the >200 entrance of the pocket (Fig. 3c and Supplementary Fig. 6). ++ ++ ++ ++ ++ ++ ++ 2.4 Table 1 ELISA immunoassays for antigen detection by sera from patients infected with M. pneumoniae. 11 There are clear structural relationships between P1 and P40/ P90 suggesting a common protein ancestor, a hypothesis that is reinforced by the structural similarities found between P1 with aIgG and IgM levels against a P1 C-terminal fragment of M. pneumoniae determined by the commercial Liaison M. pneumoniae IgG, IgM kit (DiaSorin). 14.3 ++ 10 19 − − − − − − the non-orthologous protein P110 from M. genitalium (Fig. 4 and Supplementary Fig. 9). This common ancestor could have been encoded in a mobile element, thus facilitating transmission within 2.15 >27 ++ ++ ++ ++ − − − different species of the pneumoniae cluster. The presence of 9 multiple copies of the adhesin genes in the genomes of several mycoplasmas supports this hypothesis. In each species, the 0.25 ++ ++ ++ 12 adhesin genes may have evolved to mediate adherence to different − − − − 8 hosts and tissues. Molecular systems where one of the catalytic sites, at the interface of two paralogs, is lost during evolution can 34.1 >27 ++ ++ ++ ++ be related with the situation in F-type ATPase/synthase, widely − − − 7 −, negative (value < 0.080); +, weak positive (0.080 ≤ value ≤ 0.100); ++, strong positive (value > 0.100). distributed among most living organisms47, raising the possibility of a similar evolutionary process for the Nap. Interestingly, the 0.72 >27 ++ ++ ++ sequence identity between P40/P90 and P1 is low in the N- − + − − 6 terminal domains (12%), increases in the C-terminal domains (17%), and becomes high for the transmembrane and cytoplasmic 0.44 ++ ++ ++ regions (43%). The increased variation of the sequences from the 15 − − − − 5 most exposed regions of the Nap, the N-terminal domains, can reflect specific attachment and motility requirements but also a 4.65 ++ ++ mycoplasma survival strategy related with enhanced antigenic 14 − − + − + 4 variability. To date, 13 antigenic variants have been identified in clinical isolates for P1 and 5 for P40/P9048 (Fig. 5b, c). Regions of P1 and 62.6 ++ ++ ++ 17 + − − − 3 P40/P90 presenting genetic and clinical variability cluster at the surface of the upper part of the N-terminal domain, where antibodies could easily block access to the receptor binding site. 35.7 ++ ++ ++ ++ 2.8 − − − 2 In fact, mapping the known genetic variability of M. pneumoniae onto the Nap structure, modeled according to the information available from M. genitalium with the structures of P1 and P40/ 42.8 >27 ++ ++ ++ − − − − P90, reveals that the sialic binding site pocket is completely 1 surrounded by these variable regions (Fig. 7). P1 and P40/P90 present many sequence insertions, which amount to more than C-terminal 1400 IgG–IgM levelsa C-terminal 1376 200 extra residues in each protein, with respect to their ortho- P40/P90s1s2 IgM (UA/ml) IgG (UA/ml) P40/P90s1 logues P140 and P110 in M. genitalium (Supplementary Figs. 1 P40/P90 Antigen and 4). Interestingly, most of the insertions correspond to dis- P140 ordered regions in the structures of the M. pneumoniae proteins. P1 In P1 there are 14 disordered loops, while in P40/P90 there are 4, 10 NATURE COMMUNICATIONS | (2020)11:5188 | https://doi.org/10.1038/s41467-020-18777-y | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18777-y ARTICLE a Human alveolar membrane Sialic acid P40/P90 P1 M. pneumoniae membrane b 90° 90° RepMP4 RepMP2/3 RepMP5 Genetic variation AGT repeat Fig. 7 Structure of the M. pneumoniae Nap. a Representation of the whole M. pneumoniae Nap ectodomain as a dimer of a P1-P40/P90 heterodimer, modeled from the information recently available for M. genitalium. Binding to a receptor from the host cell, a human alveolar epithelial cell, is indicated. The sialic acid binding site is located in a pocket at the top of the Nap that is not far from the twofold symmetry axis of the Nap. The crowns from P1 (green) and P40/P90 (pink) are separated by the β-propellers (brown) from the C-terminal domains (cyan), which anchor the Nap to the mycoplasma membrane. b Three views, 90° apart from each other, of the Nap surface depicting the sialic acid compound 3SL (spheres) bound into the binding site of the two P40/ P90 subunits of a Nap. Regions with genetic variability of P1 and P40/P90 (brown) cover most of the surface from the upper part of the Nap. RepMP5 from P40P90 is depicted in green, while RepMP4 and RepMP2/3 from P1 are depicted in pink and blue, respectively. C-terminal domains from both P1 and P40/ P90 are depicted in cyan. The inset shows the sialic binding site pocket, with a 3SL molecule bound, surrounded by genetically variable regions. NATURE COMMUNICATIONS | (2020)11:5188 | https://doi.org/10.1038/s41467-020-18777-y | www.nature.com/naturecommunications 11 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18777-y but 2 of them more than 60 residues long (insertions S1 and S2). P1 has still to be seen as a major player in the Nap functioning The C-terminal domain from P1 contains only one disordered during gliding and also in the adhesion to host cells. Moreover, loop and no genetic variability, pointing to the interest of proteins others than the Nap components, including P30, P65, immunogenic studies with this domain. HMW1, HMW2, or PrpC/PrkC, are known to be essential for Polyclonal antibodies against two constructs from the C- binding and gliding57–60. These proteins might support func- terminal domain of P1 (Lys1376–Asp1521 and tioning and conformation of the Naps to play its roles. Ala1400–Asp1521) inhibit the cytoadherence of M. pneumo- Integration of previous knowledge with the structural data niae cells (Fig. 6b). In addition, most of the sera from M. obtained in this work provides important clues for a better pneumoniae infected patients presenting a high IgG content understanding of the functioning of Naps in M. pneumoniae, also recognize these C-terminal domain constructs (Table 1). although many questions remain open about the adhesion and On the contrary, these C-terminal domain constructs were gliding motility mechanisms. The structural framework obtained poorly recognized by sera from the early stages of infection in this work also allows the mapping of epitopes and of the where a high IgM content is present. The full length extra- genetic and clinical variability of P1 and P40/P90 explaining cellular region of P1 was strongly recognized by all sera, many of the immunogenic properties of these immunodominant independently of the IgG or IgM ratios, confirming the proteins in M. pneumoniae. These results can help in the devel- immunodominance of P119. These results indicate that the C- opment of vaccines and therapeutics for inhibiting the infectivity terminal domain might be a relatively weak or poorly accessible of this respiratory pathogen. antigenic region, in agreement with what the structural infor- mation suggests, but effective to generate more specific IgGs Methods antibodies capable of inhibiting mycoplasma cytoadherence. Cloning, expression, and purification of P1 and P40/P90 constructs. Regions Large P1 constructs, such as the whole extracellular region49 or corresponding to the MPN141 and MPN142 genes from M. pneumoniae were a long commercial peptide (residues 1160–1521, LIAISON M. amplified from synthetic clones (Supplementary Tables 3 and 4), using primers P1F and P1R and P40/P90F and P40/P90R, respectively (Supplementary Table 5). The pneumoniae IgM and IgG, Biotrin International Ltd), maybe PCR fragments were cloned into the expression vector pOPINE61 (gift from Ray better suited for diagnosis than small C-terminal domain con- Owens plasmid #26043, Addgene, Watertown, USA) to generate constructs, with a structs, which might, however, provide a significant protective C-terminal His-tag, comprising residues 29–1521 for P1 and 23–1003 and 23–1114 response. P40/P90, often overlooked in immunological studies for a short and a long construct of P40/P90, respectively. Recombinant proteins were obtained after expression at 22 °C in B834 (DE3) cells (Merck), upon in spite of pioneering studies showing P90 was an important induction with 0.8 mM IPTG at 0.6 OD600. Cells were harvested and lysed by immunogen50, is recognized by 74% of the sera from infected sonication in 1xPBS, 40 mM imidazole and centrifuged at 49,000 × g at 4 °C. patients. Interestingly, the serologic response decreased sharply Supernatant was loaded into a HisTrap 5 ml column (GE Healthcare) pre- for P40/P90 constructs when the second insertion (S2 residues equilibrated in 1xPBS with 40 mM imidazole as binding buffer and 1xPBS with Leu395–Thr462) was eliminated. Most of the sera that recog- 400 mM imidazole as elution buffer. Soluble aliquots were pooled and loaded onto a Superdex 200 GL 10/300 column (GE Healthcare) in buffer consisting of TRIS nize S2 have a high IgM content, dominant in the initial stages 20 mM pH 7.4 and 150 mM NaCl. Buffers prepared for P40/P90 purification were of the immune response. S2 presents the highest genomic (64%) supplemented with 1 mM EDTA to prevent proteolysis. Mutants of P40/P90 were and clinical (∼80%) variability and is highly exposed and produced following the same protocol as for the wild type, but using the corre- mobile on the surface of P40/P90. Exposure of S2 is likely sponding primers showed in Supplementary Table 5. Cleavage sites in P40/P90 were determined with N-terminal Edman sequencing of the bands obtained by increased by the cleavage at its C-end that generates the two SDS-PAGE of purified samples of the P40/P90 constructs (Supplementary Fig. 5a). polypeptides P40 and P90. Altogether these results suggest that Mixtures of P1 with either the short or the long P40/P90 constructs were used for S2, located within P40, might be acting as a decoy to divert the the preparation of complexes in a 1:1 ratio. The presence of stable P1-P40/P90 immune response (Table 1). The high immune response against heterodimers, verified by gel filtration, was the only complex detected in the mixtures (Supplementary Fig. 5c). P40/P90, with unique structural peculiarities, invites a recon- For cryo-EM data collection, P1 of M. pneumoniae M129 strain (MPN141) was sideration of its therapeutic possibilities. expressed and purified as described previously62. The p1 gene corresponding to the Gliding of mycoplasmas from the pneumoniae cluster, amino acid residues 60–1518 of P1 adhesin was codon-optimized, synthesized, and including M. pneumoniae and M. genitalium, is thought to be inserted between NdeI and XbaI sites in pCold I vector (TAKARA BIO, Shiga, JAPAN). The resulting plasmid, pP1-1, wherein p1 was fused to an N-terminal 6 × caused by a mechanism where the force, generated by ATP His-tag and a Factor Xa site. BL21 or BL21 (DE3) pLysS cells carrying pP1-1 were energy51, is transmitted to the Nap that appears to work as a leg/ cultured at 37 °C until the culture density reached an OD600 value of 0.3. The cells foot on the surface of the mycoplasma cell7,8,12,52–56. Gliding of were subjected to cold shock at 15 °C for 30 min, followed by the addition of IPTG M. pneumoniae cells is always in the direction of the attachment to a final concentration of 1.0 mM, and cultured overnight at 15 °C. The cells were organelle, which indicates that the gliding machinery, including harvested by centrifugation (4000 × g, 10 min, 4 °C) and washed twice with a TN buffer comprising 20 mM Tris-HCl and 500 mM NaCl. The suspension was frozen the Naps at the cell surface, should provide a well-defined at −80 °C. The cell pellet was thawed, suspended in a binding buffer (20 mM Tris- directionality. To achieve a directional gliding, the release of the HCl pH 8.0, 500 mM NaCl, 20 mM imidazole) containing 1 mM PMSF. Cells were sialylated receptors is as essential as the binding because myco- then disrupted with a Sonication (NIHONSEIKI, Tokyo, Japan). Unbroken cells plasma cells cannot move forward if they remain anchored to the were removed by centrifugation (60,000–140,000 × g, 60 min, 4 °C). The supernatant was loaded onto a HisTrap HP column (GE Healthcare, Milwaukee, surface receptors where they have first attached. In the Nap, the WI) equilibrated with the binding buffer. The bound proteins were eluted with a conformational changes required by the foot/leg activity during linear gradient of 10–200 mM imidazole in TN buffer. Fractions containing P1 gliding have to be synchronized with the binding/release to the adhesin were dialyzed against a TNG buffer (20 mM Tris-HCl (pH 8.0), 150 mM cell receptors, which are thought to be randomly distributed on NaCl) and loaded onto a HiLoad 16/60 Superdex 200 prep grade column (GE Healthcare) equilibrated with the TNG buffer at a flow speed of 1 mL/min. The the host surface. The binding site in P40/P90 could be regulated collected P1 adhesin fractions were combined, dialyzed against 20 mM Tris-HCl by the interaction with P1 as found in M. genitalium17. This may (pH 8.0), and applied to an anion exchanger, Q Sepharose Fast Flow equilibrated be the reason why P1 was expected to contain the binding site. In with the buffer, and eluted with a linear gradient of NaCl from 0 to 0.4 M. The the cryo-EM map of P1, the weak density of the C-terminal purified protein was concentrated using an Amicon Ultra 30k spin filter (Millipore, Darmstadt, Germany) if necessary. domain indicates flexibility with respect to the N-terminal domain proving that P1 can experience significant conforma- tional changes consistent with the observations that antibodies Crystallization and X-ray data collection of P1 and P40/P90N. Ninety-six-well plates crystallization screenings were performed on both Cartesian (Cartesian (TM) that bind to the C-terminal domain can interfere with the Dispensing Systems) and Phenix (Art Robbins Instruments) robots mixing 150 nL attachment and gliding of mycoplasma cells. Therefore, although of reservoir condition with 150 nL of protein solution at concentrations of 6 mg/ P40/P90 contains the binding site to the sialic acid cell receptors, mL for each of the P1 and P40/P90 proteins. The optimized crystals of P1 were 12 NATURE COMMUNICATIONS | (2020)11:5188 | https://doi.org/10.1038/s41467-020-18777-y | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18777-y ARTICLE obtained after mixing 1 µL of P1 at 6 mg/mL with 1 µL 20% PEGMME 2000 and quality of the final map (Fig. 2b), the P1 structure presented a significant amount of PCB (sodium propionate, sodium cacodylate, and BIS-TRIS propane pH 4) as a disordered residues located at the N-end (missing residues 29–59) and in 14 loops reservoir condition in a ratio of 2:1, respectively. P40/P90N optimized crystals were (102–105, 228–230, 259–268, 278–282, 298–300, 337–348, 831–847, 870–888, obtained mixing 1 µL of P40/P90N at 6 mg/mL with 1 µL 20% PEG 3350, 0.3% 923–928, 941–944, 1226–1232, 1308–1324, 1341–1349, and 1482–1495). The final MetOH in a ratio 1:1, respectively. To obtain crystals of P40/P90N in complex with refined structure has been deposited in the PDB with code 6RC9. 3SL or 6SL, protein samples were previously incubated with 10 mM of each oli- gosaccharide 1 h at 20 °C. All attempts for P40/P90 crystallization were not suc- Determination of the P40/P90 structures. The P40/P90 crystals belong to space cessful. Crystals used for data collection were flash frozen in liquid nitrogen with group P212121 and contain two subunits in the asymmetric unit. The structure was 15% glycerol as cryo-protectant. solved by molecular replacement with program Phaser68 using as searching model X-ray data collection was carried out at Xaloc beamline (ALBA Synchrotron, the structure from the orthologous protein P110 from M. genitalium (PDB code Spain). Data were processed with Xia263 using XDS64, Aimless and Pointless65 6R3T). The structure of P40/P90 was refined at 2.65 Å resolution, giving agreement from the CCP4 suit of programs66. factors R and Rfree of 21.4% and 23.4%, respectively (Supplementary Table 1). Crystals of complexes from P40/P90 with 3SL and 6SL belong to space C2, although with pseudo-orthorhombic unit cell dimensions, and contain one subunit Determination of the P1 crystal structure. The crystals from P1 belong to space in the asymmetric unit. The 3SL crystals showed a well-defined oligosaccharide group C2 and could contain, by packing considerations, only one subunit in the with high occupancy, at 3.2 Å. The 6SL crystals, at 2.8 Å resolution, presented some asymmetric unit. We tried to solve these P1 crystals by heavy atoms and SeMet disorder and an overall weak density for the oligosaccharide that was interpreted as derivatives, but finally the resolution was achieved by density modification tech- partial occupancy. niques applying solvent flattening and averaging with crystals from the M. geni- Tracing and refinement of all the structures was performed alternating talium orthologous protein P140, whose structure was also unknown at that time17. interactive and automatic cycles with programs Coot69 and Buster70, respectively. With this procedure the structures of P1 and of P140 (with a sequence identity of The final refined structures have been deposited in the PDB with codes: 6RJ1 for 41% between their ectodomains) were solved simultaneously. The P140 containing P40/P90, 6TLZ for the 3SL complex, and 6TM0 for the 6SL complex. crystals used corresponded to crystals from the complex of P140 with the N- domain of P110 (P140–P110N) and to crystals from P140 alone. Steps followed for the structure determination of P1 and P140 were the following: Single-particle cryo-EM. The purified P1 protein construct was applied onto a Quantifoil holey carbon grid (R1.2/1.3, Cu, 200 mesh) covered with a thin film of (1) A partial molecular replacement solution was obtained for the P140–P110N GO flakes (SIGMA-ALDRICH), blotted for 4.5 s at 4 °C in 100% humidity, and crystals using as searching model the N-domain from the P110 structure17. plunge frozen in liquid ethane by using a Vitrobot Mark IV (Themo Fisher Sci- This partial solution of the P140–P110N crystals contained four P110N entific). Cryo-EM imaging was obtained using a Titan Krios (Themo Fisher Sci- subunits in the asymmetric unit. The four P110N subunits were organized entific) operating at 300 kV acceleration voltage and equipped with a Cs corrector as two pairs and within each pair the two subunits were related by accurate (CEOS, GmbH) and a Falcon III direct electron detector (Themo Fisher Scientific). local (non-crystallographic) twofold axis. In this partial solution, no density A total of 5279 movies were obtained in the electron counting mode with a physical was visible that could correspond to P140. pixel size of 0.69 Å/pixel and total dose of 60 e/Å2 with 36.33 s exposure time. The (2) From this partial solution of the P140–P110N crystals, a difference (Fo–Fc) data were automatically collected using EPU software with a defocus range of map was generated at about 3 Å resolution. Positive densities in this −0.75 to −2.75 µm and were fractionated into 78 movie frames. difference map, expected to correspond mainly to P140, were tentatively reinforced by averaging (without applying any mask) using one of the local twofold symmetries. This unmasked averaging also weakens density from Single-particle cryo-EM image processing. The movie frames were subsequently other subunits in the crystal not related by the local twofold symmetry. In aligned to correct for beam-induced movement and drift using MotionCor271, and this averaged density it was possible to differentiate continuous regions of contrast transfer function was estimated using Gctf72. A total of 4,574,424 particles strong and of weak density, but it was not yet possible to identify any images were automatically picked using Gautomatch (http://www.mrc-lmb.cam.ac. consisting molecular feature from P140. uk/kzhang/) and several rounds of 2D classification and 3D classification were (3) The cryo-electron tomography (CET) map of a whole M. genitalium Nap17 performed using RELION-2.173. A total of 295,195 particle images from good 2D was then used to define an initial possible mask for P140. The CET map, at classes were selected for making the initial model of P1 using cryoSPARC74. First ∼17 Å resolution, corresponded to a dimer of P140–P110 complexes, where and second 3D classification were calculated with angular sampling interval at 7.5 the fitting of the P110 structures was unambiguous. The CET map was and 3.7 degrees, respectively. Finally, 68,014 particles from the best 3D class were placed on the (Fo–Fc) averaged map by superposing the corresponding fitted 3D refined, producing a reconstruction with a resolution of 2.88 Å and a B-factor of P110 structures. Then, a putative “P140 initial map” was generated as the 89 Å2 with the gold-standard FSC criteria (FSC ≥ 0.143). The local resolution was (Fo–Fc) averaged density inside the CET envelop corresponding to P140. estimated using RELION-2.1. The processing strategy is described in Supplemen- After some polishing (avoiding overlapping with neighbor subunits or tary Fig. 2a–d and the model refinement statistics in Supplementary Table 2. retaining regions with continuous densities) the assumed “P140 initial map” was used to perform a molecular replacement search (using program SEC–MALS analysis. The molecular weight of the P40/P90 short construct was PHASER) obtaining a reasonable possible solution with four P110N measured using a Superose 6 10/300 GL (GE HEalthcare) column in a Prominence structures and four “P140 initial maps.” Attempts to use the density from liquid chromatography system (Shimadzu) connected to a DAWN HELEOS II the CET maps directly for phasing, and not just as a mask, were unsuccessful multi-angle light scattering (MALS) detector and an Optilab T-REX refractive in our hands. index (dRI) detector (Wyatt Technology). ASTRA 7 software (Wyatt Technology) (4) Iterative cycles of density averaging (and solvent flattening), alternating with was used for data processing and result analysis. A dn/dc value of 0.185 ml/g manual readjustments of the P140 mask, were then performed with the (typical of proteins) was assumed for calculations. P140–P110N crystals. The procedure converged giving clearer and more continuous density, although no secondary structures that could correspond to P140 were visible. Likely, the almost parallel orientation of the two non- Antibody production and validation. Polyclonal antibodies against adhesin P1 crystallographic twofold axes weakened the phasing power of this averaging. were produced by Eurogentec. Briefly, rats were injected with peptides In spite of the limitations, the new density (within the updated mask) Lys1376–Asp1521 or Ala1400–Asp1521 and bleeds were collected at day 28. Pre- allowed obtaining a molecular replacement solution for the P1 crystals. immunization bleeds were also collected as negative controls. Specificity of the Surprisingly, the new density did not provide a solution for the P140 alone polyclonal antibodies was tested by western blot against the recombinant peptides crystals. and protein extract of Mycoplasma cells. The dilution used for each antibody (5) Density modification with averaging within the P140–P110N crystal and was 1:500. now also with the P1 crystals (using program DMMULTI67) improved The production of polyclonal antibodies has complied with ethical regulations quickly the maps. The iterative cycles of DMMULTI were alternated with for animal testing and research. It complies with the following association’s cycles of phase extension for the P1 crystals (using the “autobuild” protocol requirements: Federation of European Laboratory animal Science Associations and in Phenix at 1.94 Å resolution) and with manual readjustments of the UK Home Office Animals Scientific Procedures Act. averaging masks (that were updated about forty times). Model building was The regulations followed by Eurogentec are: 2010/63/EU and 01/2005/EU. carried forward for P1 and P140 in parallel. With about 50% of the structures traced, it was possible to obtain a molecular replacement solution Quantitative adhesion assay. M. pneumoniae (ATCC 29343) was grown in a T75 for the P140 alone crystals that, with six subunits in the crystal asymmetric flask in Hayflick medium during 24 h, harvested by scraping, and resuspended in unit, facilitated the completion of the P1 and P140 structures. Hayflick medium to a final concentration of 107 cells/ml. Incubation, during 1 h at P1 was then refined, giving agreement factors R and Rfree of 18.7% and 22.9%, room temperature on a rotating wheel, of 1 ml of a cell suspension with sera (at 1, respectively (Supplementary Table 1). The difference between these two factors was 3, 6, and 10% concentrations) containing polyclonal antibodies raised against used traditionally to assess how well the available molecular model explains the peptides Lys1376–Asp1521 or Ala1400–Asp1521, in presence or absence of 2 µg corresponding experimental diffraction data. The difference between both factors is from the corresponding peptide. A preimmunization serum was used as a negative related with the bias introduced in the model during refinement. Despite the control. After incubation, cell suspensions were seeded in 24-well plates and NATURE COMMUNICATIONS | (2020)11:5188 | https://doi.org/10.1038/s41467-020-18777-y | www.nature.com/naturecommunications 13 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18777-y incubated at 37 °C to allow cells adhere to the plastic. After 2 h, supernatants 2. Ferwerda, A., Moll, H. A. & de Groot, R. Respiratory tract infections by (containing the fraction of the cells that did not adhere to the plastic) and attached Mycoplasma pneumoniae in children: a review of diagnostic and therapeutic cells were collected and processed for gDNA extraction using MasterPure DNA measures. Eur. J. Pediatr. 160, 483–491 (2001). Purification Kit (EpiCentre #MCD85201). Samples were obtained in triplicate, and 3. Atkinson, T. P., Balish, M. F. & Waites, K. B. Epidemiology, clinical the relative amount of cells was quantified by qPCR with SYBR Green PCR Master manifestations, pathogenesis and laboratory detection of Mycoplasma Reagent (Thermo #4367659), using a LightCycler 480 Real-Time PCR machine pneumoniae infections. FEMS Microbiol. Rev. 32, 956–973 (2008). (Roche) with the following conditions: denaturation at 95 °C for 10 s; 40 cycles of 4. Parrott, G. L., Kinjo, T. & Fujita, J. A compendium for Mycoplasma amplification of 95 °C for 15 s and 60 °C for 1 min; melting curve at 95° for 15 s, pneumoniae. Front. Microbiol. 7, 513 (2016). 60 °C for 15 s, 95° continuous. The following oligonucleotides were used: qPCR 5. Radestock, U. & Bredt, W. Motility of Mycoplasma pneumoniae. J. Bacteriol. Myco FW 5′-ACGATGATTACAGGCGGTTC-3′ and qPCR Myco RV 5′- GTTG 129, 1495–1501 (1977). GTGGCCTCTTGTTGAT-3′. 6. Miyata, M. & Hamaguchi, T. Prospects for the gliding mechanism of Mycoplasma mobile. Curr. Opin. Microbiol. 29, 15–21 (2016). Immunoassays. Diagnostic of M. pneumoniae was conducted using the Liaison M. 7. Miyata, M. & Hamaguchi, T. Integrated information and prospects for gliding pneumoniae IgG, IgM kit (DiaSorin) using a 1/100 dilution of the patient sera. mechanism of the pathogenic bacterium Mycoplasma pneumoniae. Front. Indirect ELISA assays were performed on 96-well plates Immulon 4 HBX 96-well Microbiol. 7, 960 (2016). plates (ThermoFisher) incubating 1 µg of each antigen at 4 °C overnight. 1/100 8. Nishikawa, M. S. et al. Refined mechanism of Mycoplasma mobile gliding dilutions of each patient serum were added to the plate and detected using an anti- based on structure, ATPase activity, and sialic acid binding of machinery. human IgG antibody conjugated with HRP (Thermofisher Scientific). Upon mBio 10, e02846-19 (2019). incubation for 30 min with 100 µl of substrate (Thermofisher Scientific), 100 µl of 9. Krause, D. C. Mycoplasma pneumoniae cytadherence: unravelling the tie that sulfuric acid 25% was added to stop the reaction and absorbance was read at 450 binds. Mol. Microbiol. 20, 247–253 (1996). nm on a Triturus ELISA instrument (Grifols) device. Reference filter was set at 620 10. Krause, D. C. & Baseman, J. B. Mycoplasma pneumoniae proteins that nm. selectively bind to host cells. Infect. Immun. 37, 382–386 (1982). 11. Feldner, J., Gobel, U. & Bredt, W. Mycoplasma pneumoniae adhesin localized Ethical approval. Ethical approval for the study was obtained from the Ethics to tip structure by monoclonal antibody. Nature 298, 765–767 (1982). Committees for Research with Drugs from the Parc Taulí (Ref 2019/664) and Vall 12. Seto, S., Kenri, T., Tomiyama, T. & Miyata, M. Involvement of P1 adhesin in d’Hebron University Hospitals (PR(AG)24/2020). gliding motility of Mycoplasma pneumoniae as revealed by the inhibitory effects of antibody under optimized gliding conditions. J. Bacteriol. 187, 1875–1877 (2005). Inductive coupled plasma–mass spectrometry (ICP–MS). Samples were diges- 13. Hu, P. C. et al. Mycoplasma pneumoniae infection: role of a surface protein in ted with 1 ml of HNO3 plus 0.5 ml of H2O2 in the closed teflon reactor during 24 h the attachment organelle. Science 216, 313–315 (1982). at 90 °C. The Zn measure was realized with ICP–MS in an instrument Agilent 14. Krause, D. C., Leith, D. K., Wilson, R. M. & Baseman, J. B. Identification of 7500ce with respect to a calibration line prepared by dilution from a ZN standard solution (Inorganic Ventures). Mycoplasma pneumoniae proteins associated with hemadsorption and virulence. Infect. Immun. 35, 809–817 (1982). 15. Burgos, R. et al. Mycoplasma genitalium P140 and P110 cytadhesins are Surface plasmon resonance. Binding kinetics were determined by SPR using a reciprocally stabilized and required for cell adhesion and terminal-organelle Biacore 3000 biosensor platform (GE Biosystems) equipped with a research-grade development. J. Bacteriol. 188, 8627–8637 (2006). streptavidin-coated biosensor chip SA. The chip was preconditioned with three 1- 16. Scheffer, M. P. et al. Structural characterization of the NAP; the major min injections of 1 M NaCl and 50 mM NaOH. Subsequently, the second and third adhesion complex of the human pathogen Mycoplasma genitalium. Mol. flow cells of the chip were loaded, respectively, with oligosaccharides 6SL-PAA- Microbiol. 105, 869–879 (2017). biotin and 3SL-PAA-biotin (Carbosynth) at 10 µg/ml diluted in 20 mM TRIS pH 17. Aparicio, D. et al. Structure and mechanism of the Nap adhesion complex 7.4, 150 mM NaCl, and 0.05 % tween20. It is noteworthy to mention that the from the human pathogen Mycoplasma genitalium. Nat. Commun. 11, 2877 ligands used contains a long polyacrylamide chain connecting the biotinylated (2020). chemically linked tag with the oligosaccharides, which should avoid most steric 18. Aparicio, D. et al. Mycoplasma genitalium adhesin P110 binds sialic-acid clashes between biotin and the P1 and P40/P90 samples used. The immobilization human receptors. Nat. Commun. 9, 4471 (2018). levels acquired were at ∼200 response units and the first cell was left blank to serve 19. Chourasia, B. K., Chaudhry, R. & Malhotra, P. Delineation of as a reference. A series of diluted concentrations were injected using a flow rate of 30 µl/min at 25 °C (0.9, 1.9, 3.7, and 7.5 µM of P40/P90 for 6SL titration and 1.0, immunodominant and cytadherence segment(s) of Mycoplasma pneumoniae 2.0, 4.0, 16.0, and 32.0 µM of both P40/P90 and P1 for 3SL titration). The proteins P1 gene. BMC Microbiol. 14, 108 (2014). were allowed to associate and dissociate for 60 and 120 s, respectively, for P1 and 20. Jacobs, E., Pilatschek, A., Gerstenecker, B., Oberle, K. & Bredt, W. 120 and 240 s, respectively, for P40/P90. Dissociations were followed by a regen- Immunodominant epitopes of the adhesin of Mycoplasma pneumoniae. J. eration step of 30 s with 0.05% SDS at 30 µl/min. The data were fitted assuming a Clin. Microbiol. 28, 1194–1197 (1990). Langmuir 1:1 binding model using BiaEvaluation 3.1 software to determine the 21. Dandekar, T. et al. Re-annotating the Mycoplasma pneumoniae genome equilibrium dissociation constant KD. sequence: adding value, function and reading frames. Nucleic Acids Res. 28, 3278–3288 (2000). 22. Su, C. J., Chavoya, A. & Baseman, J. B. Regions of Mycoplasma pneumoniae Reporting summary. Further information on research design is available in the Nature cytadhesin P1 structural gene exist as multiple copies. Infect. Immun. 56, Research Reporting Summary linked to this article. 3157–3161 (1988). 23. Wenzel, R. & Herrmann, R. Repetitive DNA sequences in Mycoplasma Data availability pneumoniae. Nucleic Acids Res. 16, 8337–8350 (1988). Atomic coordinates and structure factors for the reported crystal structures of P1, P40/ 24. Sperker, B., Hu, P. & Herrmann, R. Identification of gene products of the P1 P90, and the P40/P90-3SL and P40/P90-6SL complexes have been deposited into the operon of Mycoplasma pneumoniae. Mol. Microbiol. 5, 299–306 (1991). Protein Data Bank (PDB) under accession codes 6RC9, 6RJ1, 6TLZ, and 6TM0, 25. Layh-Schmitt, G. & Herrmann, R. Localization and biochemical respectively. Cryo-EM density has been deposited in the EM Data Base under the characterization of the ORF6 gene product of the Mycoplasma pneumoniae P1 accession code EMD-30233. Atomic coordinates of the Cryo-EM structure have been operon. Infect. Immun. 60, 2906–2913 (1992). deposited in the PDB under the accession code 7BWM. Source data are provided with 26. McGuffin, L. J., Bryson, K. & Jones, D. T. The PSIPRED protein structure this paper. Other data are available from the corresponding authors upon reasonable prediction server. Bioinformatics 16, 404–405 (2000). request. 27. Pantelic, R. S., Meyer, J. C., Kaiser, U., Baumeister, W. & Plitzko, J. M. Graphene oxide: a substrate for optimizing preparations of frozen-hydrated samples. J. Struct. Biol. 170, 152–156 (2010). Received: 24 April 2020; Accepted: 8 September 2020; 28. Dallo, S. F., Su, C. J., Horton, J. R. & Baseman, J. B. Identification of P1 gene domain containing epitope(s) mediating Mycoplasma pneumoniae cytoadherence. J. Exp. Med. 167, 718–723 (1988). 29. Holm, L. & Rosenstrom, P. Dali server: conservation mapping in 3D. Nucleic Acids Res. 38, W545–W549 (2010). 30. Widjaja, M., Berry, I. J., Pont, E. J., Padula, M. P. & Djordjevic, S. P. P40 and References P90 from Mpn142 are targets of multiple processing events on the surface of 1. Narita, M. Pathogenesis of extrapulmonary manifestations of Mycoplasma Mycoplasma pneumoniae. Proteomes 3, 512–537 (2015). pneumoniae infection with special reference to pneumonia. J. Infect. 31. Ruland, K., Himmelreich, R. & Herrmann, R. Sequence divergence in the Chemother. 16, 162–169 (2010). ORF6 gene of Mycoplasma pneumonia. J. Bacteriol. 176, 5202–5209 (1994). 14 NATURE COMMUNICATIONS | (2020)11:5188 | https://doi.org/10.1038/s41467-020-18777-y | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18777-y ARTICLE 32. Krissinel, E. & Henrick, K. Inference of macromolecular assemblies from 60. Romero-Arroyo, C. E. et al. Mycoplasma pneumoniae protein P30 is required crystalline state. J. Mol. Biol. 372, 774–797 (2007). for cytadherence and associated with proper cell development. J. Bacteriol. 33. Chothia, C. & Lesk, A. M. The relation between the divergence of sequence 181, 1079–1087 (1999). and structure in proteins. EMBO J. 5, 823–826 (1986). 61. Berrow, N. S. et al. A versatile ligation-independent cloning method suitable 34. Bensing, B. A. et al. Structural basis for sialoglycan binding by the for high-throughput expression screening applications. Nucleic Acids Res. 35, Streptococcus sanguinis SrpA adhesin. J. Biol. Chem. 291, 7230–7240 (2016). e45 (2007). 35. Manchee, R. J. & Taylor-Robinson, D. Utilization of neuraminic acid receptors 62. Kenri, T. et al. Production and characterization of recombinant P1 adhesin by mycoplasmas. J. Bacteriol. 98, 914–919 (1969). essential for adhesion, gliding, and antigenic variation in the human 36. Sobeslavsky, O., Prescott, B. & Chanock, R. M. Adsorption of Mycoplasma pathogenic bacterium, Mycoplasma pneumoniae. Biochem. Biophys. Res. pneumoniae to neuraminic acid receptors of various cells and possible role in Commun. 508, 1050–1055 (2019). virulence. J. Bacteriol. 96, 695–705 (1968). 63. Winter, G. xia2: an expert system for macromolecular crystallography data 37. Kasai, T. et al. Role of binding in Mycoplasma mobile and Mycoplasma reduction. J. Appl. Crystallogr. 43, 186–190 (2010). pneumoniae gliding analyzed through inhibition by synthesized sialylated 64. Kabsch, W. XDS. Acta Crystallogr. D Biol. Crystallogr. 66, 125–132 (2010). compounds. J. Bacteriol. 195, 429–435 (2013). 65. Evans, P. Scaling and assessment of data quality. Acta Crystallogr. D Biol. 38. Williams, C. R. et al. Sialylated receptor setting influences Mycoplasma Crystallogr. 62, 72–82 (2006). pneumoniae attachment and gliding motility. Mol. Microbiol. 109, 735–744 66. Collaborative Computational Project, Number 4. The CCP4 suite: programs for (2018). protein crystallography. Acta Crystallogr. D Biol. Crystallogr. 50, 760–763 (1994). 39. Gerstenecker, B. & Jacobs, E. Topological mapping of the P1-adhesin of 67. Cowtan, K. D., Zhang, K. Y. J. and Main, P. International Tables for Mycoplasma pneumoniae with adherence-inhibiting monoclonal antibodies. J. Crystallography. Vol. F, ch. 15.3, pp. 407–412 (2012). Gen. Microbiol. 136, 471–476 (1990). 68. McCoy, A. J. Solving structures of protein complexes by molecular replacement 40. Su, C. J., Chavoya, A., Dallo, S. F. & Baseman, J. B. Sequence divergency of the with Phaser. Acta Crystallogr. D Biol. Crystallogr. 63, 32–41 (2007). cytadhesin gene of Mycoplasma pneumoniae. Infect. Immun. 58, 2669–2674 69. Emsley, P. & Cowtan, K. Coot: model-building tools for molecular graphics. (1990). Acta Crystallogr D. Biol. Crystallogr. 60, 2126–2132 (2004). 41. Spuesens, E. B. M. et al. Sequence variations in RepMP2/3 and RepMP4 70. Bricogne, G. et al. BUSTER 2.11.12. Cambridge, United Kingdom: Global elements reveal intragenomic homologous DNA recombination events in Phasing Ltd. (2017). Mycoplasma pneumoniae. Microbiology 155, 2182–2196 (2009). 71. Zheng, S. Q. et al. MotionCor2: anisotropic correction of beam-induced motion 42. Jacobs, E., Vonski, M., Oberle, K., Opitz, O. & Pietsch, K. Are outbreaks for improved cryo-electron microscopy. Nat. Methods 14, 331–332 (2017). and sporadic respiratory infections by Mycoplasma pneumoniae due 72. Zhang, K. Gctf: Real-time CTF determination and correction. J. Struct. Biol. to two distinct subtypes? Eur. J. Clin. Microbiol. Infect. Dis. 15, 38–44 193, 1–12 (2016). (1996). 73. Kimanius, D., Forsberg, B. O., Scheres, S. H. & Lindahl, E. Accelerated cryo- 43. Morrison-Plummer, J., Lazzell, A. & Baseman, J. B. Shared epitopes between EM structure determination with parallelisation using GPUs in RELION-2. Mycoplasma pneumoniae major adhesin protein P1 and a 140-kilodalton Elife 5, e18722 (2016). protein of Mycoplasma genitalium. Infect. Immun. 55, 49–56 (1987). 74. Punjani, A., Rubinstein, J. L., Fleet, D. J. & Brubaker, M. A. cryoSPARC: 44. Baseman, J. B., Cole, R. M., Krause, D. C. & Leith, D. K. Molecular basis for algorithms for rapid unsupervised cryo-EM structure determination. Nat. cytadsorption of Mycoplasma pneumoniae. J. Bacteriol. 151, 1514–1522 Methods 14, 290–296 (2017). (1982). 45. Razin, S. & Jacobs, E. Mycoplasma adhesion. J. Gen. Microbiol. 138, 407–422 (1992). Acknowledgements 46. Waldo, R. H. 3rd, Jordan, J. L. & Krause, D. C. Identification and This work was supported by grants BFU2018-101265-B-100 (MINECO) to I.F., complementation of a mutation associated with loss of Mycoplasma BIO2017-84166-R (MINECO) to J.P., and BES-2015-076104 (MINECO) to M.L.S., JSPS pneumoniae virulence-specific proteins B and C. J. Bacteriol. 187, 747–751 KAKENHI Grant Number JP25000013 to K.N., and a FEDER project from Instituto de (2005). Salud Carlos III (ISCIII, Acción Estratégica en Salud 2016). This work has also been 47. Guo, H., Suzuki, T. & Rubinstein, J. L. Structure of a bacterial ATP synthase. funded by the Platform Project for Supporting Drug Discovery and Life Science Research Elife 8, e43128 (2019). (BINDS) from AMED under Grant Number JP19am0101117 to K.N. (support number 48. Kenri, T. et al. Periodic genotype shifts in clinically prevalent Mycoplasma 1282), by the Cyclic Innovation for Clinical Empowerment (CiCLE) from AMED under pneumoniae strains in Japan. Front. Cell. Infect. Microbiol. 10, 385 (2020). Grant Number JP17pc0101020 to K.N., and by JEOL YOKOGUSHI Research Alliance 49. Svenstrup, H. F., Nielsen, P. K., Drasbek, M., Birkelund, S. & Christiansen, G. Laboratories of Osaka University to K.N. This work was supported by a Grant-in-Aid for Adhesion and inhibition assay of Mycoplasma genitalium and M. pneumoniae Scientific Research on the Innovative Area “Harmonized Supramolecular Motility by immunofluorescence microscopy. J. Med. Microbiol. 51, 361–373 (2002). Machinery and Its Diversity” (MEXT KAKENH, JP24117002 to M.M., JP25117530 and 50. Leith, D. K., Trevino, L. B., Tully, J. G., Senterfit, L. B. & Baseman, J. B. Host JP15H01337 to T.K.), Grants-in-Aid for Scientific Research (B) and (A) (MEXT discrimination of Mycoplasma pneumoniae proteinaceous immunogens. J. KAKENHI, JP24390107, JP17H01544), JST CREST (JPMJCR19S5), Osaka City Uni- Exp. Med. 157, 502–514 (1983). versity (OCU) Strategic Research Grant 2018 for top priority researches to M.M. D.A. 51. Mizutani, M. & Miyata, M. Behaviors and energy source of Mycoplasma acknowledges a María de Maeztu Unit of Excellence grant MDM-2014-0435. Thanks are gallisepticum gliding. J. Bacteriol. 201, e00397-19 (2019). given to O. Conchillo and Prof. X. Daura for their enlightening suggestions and to Prof. 52. García-Morales, L., González-González, L. Querol, E. and Piñol, J. A P. Loewen for a careful reading of the manuscript. Many thanks are given to the XALOC minimized motile machinery for Mycoplasma genitalium. Mol. Microbiol. 100 beamline team at ALBA for their support during data collection and to the Crystal- 125–138 (2016). lography Platform at the Barcelona Science Park (PCB), especially to Dr. Roman Bonet 53. Miyata, M. Centipede and inchworm models to explain Mycoplasma gliding. for his expert advice during SEC–MALS analysis. We also acknowledge to Dr. Sílvia Trends Microbiol. 16, 6–12 (2008). Barceló from the Molecular Interactions Unit, IDIBELL and to Dr. Marta Taules from 54. Seybert, A. et al. Cryo-electron tomography analyses of terminal organelle the Scientific and Technologic Department, UB for their technical support during SPR mutants suggest the motility mechanism of Mycoplasma genitalium. Mol. experiments. Finally, we are very grateful to Dr. Juan Delgado from the Immunology Microbiol. 108, 319–329 (2018). Department of the Hospital Universitari Parc Taulí for his helpful advices on performing 55. Nakane, D., Kenri, T., Matsuo, L. & Miyata, M. Systematic structural analyses ELISA analyses. of attachment organelle in Mycoplasma pneumoniae. PLoS Pathog. 11, e1005299 (2015). 56. Kawamoto, A. et al. Periodicity in attachment organelle revealed by electron cryotomography suggests conformational changes in gliding mechanism of Author contributions Conceived and designed the experiments: D.V., I.F., T.K., K.N., M.M., and D.A. Per- Mycoplasma pneumoniae. mBio 7, e00243-16 (2016). formed the experiments: D.V., A.K., U.M., R.I., R.P.L., J.M., R.M., P.B., O.Q.P., M.P.S., 57. Chang, H. Y., Jordan, J. L. & Krause, D. C. Domain analysis of protein P30 in T.K., T.A.K., and D.A. Analyzed the data: D.V., A.K., U.M., R.I., P.B., O.Q.P., M.E., I.S., Mycoplasma pneumoniae cytadherence and gliding motility. J. Bacteriol. 193, J.E., M.F.H., M.P.S., J.P., A.S.F., M.L.S., S.M., K.S., T.K., T.A.K., I.F., and D.A. Con- 1726–1733 (2011). tributed reagents/materials/analysis tools: A.S.F., M.E., K.N., M.M., and I.F. Wrote the 58. Hasselbring, B. M., Sheppard, E. S. & Krause, D. C. P65 truncation impacts paper: D.V., A.S.F., O.Q.P., M.L.S., T.K., M.M., I.F., and D.A. P30 dynamics during Mycoplasma pneumoniae gliding. J. Bacteriol. 194, 3000–3007 (2012). 59. Page, C. A. & Krause, D. C. Protein kinase/phosphatase function correlates with gliding motility in Mycoplasma pneumoniae. J. Bacteriol. 195, 1750–1757 Competing interests The authors declare no competing interests. (2013). NATURE COMMUNICATIONS | (2020)11:5188 | https://doi.org/10.1038/s41467-020-18777-y | www.nature.com/naturecommunications 15 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18777-y Additional information Open Access This article is licensed under a Creative Commons Supplementary information is available for this paper at https://doi.org/10.1038/s41467- Attribution 4.0 International License, which permits use, sharing, 020-18777-y. adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Correspondence and requests for materials should be addressed to M.M. or D.A. Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless Peer review information Nature Communications thanks Jeremy Derrick and the other, indicated otherwise in a credit line to the material. If material is not included in the anonymous, reviewer(s) for their contribution to the peer review of this work. Peer article’s Creative Commons license and your intended use is not permitted by statutory reviewer reports are available. regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/ Reprints and permission information is available at http://www.nature.com/reprints licenses/by/4.0/. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. © The Author(s) 2020 16 NATURE COMMUNICATIONS | (2020)11:5188 | https://doi.org/10.1038/s41467-020-18777-y | www.nature.com/naturecommunications