ARTICLE https://doi.org/10.1038/s41467-020-16511-2 OPEN Structure and mechanism of the Nap adhesion complex from the human pathogen Mycoplasma genitalium David Aparicio1,4, Margot P. Scheffer2,4, Marina Marcos-Silva3,4, David Vizarraga1, Lasse Sprankel2, Mercè Ratera 1, Miriam S. Weber 2, Anja Seybert2, Sergi Torres-Puig3, Luis Gonzalez-Gonzalez 3, Julian Reitz2, Enrique Querol3, Jaume Piñol3, Oscar Q. Pich 3 ✉, Ignacio Fita1 ✉ & Achilleas S. Frangakis 2✉ 1234567890():,; Mycoplasma genitalium is a human pathogen adhering to host target epithelial cells and causing urethritis, cervicitis and pelvic inflammatory disease. Essential for infectivity is a transmembrane adhesion complex called Nap comprising proteins P110 and P140. Here we report the crystal structure of P140 both alone and in complex with the N-terminal domain of P110. By cryo-electron microscopy (cryo-EM) and tomography (cryo-ET) we find closed and open Nap conformations, determined at 9.8 and 15 Å, respectively. Both crystal structures and the cryo-EM structure are found in a closed conformation, where the sialic acid binding site in P110 is occluded. By contrast, the cryo-ET structure shows an open conformation, where the binding site is accessible. Structural information, in combination with functional studies, suggests a mechanism for attachment and release of M. genitalium to and from the host cell receptor, in which Nap conformations alternate to sustain motility and guarantee infectivity. 1 Instituto de Biología Molecular de Barcelona (IBMB-CSIC) and Maria de Maeztu Unit of Excellence, Parc Científic de Barcelona, Baldiri Reixac 10, 08028 Barcelona, Spain. 2 Buchmann Institute for Molecular Life Sciences, Max-von-Laue Str. 15, 60438 Frankfurt, Germany. 3 Institut de Biotecnologia i Biomedicina and Departament de Bioquímica i Biologia Molecular, Universitat Autònoma de Barcelona, Bellaterra, 08193 Barcelona, Spain. 4These authors contributed equally: David Aparicio, Margot P. Scheffer, Marina Marcos-Silva. ✉email: oquijada@tauli.cat; ifrcri@ibmb.csic.es; achilleas.frangakis@biophysik.org NATURE COMMUNICATIONS | (2020)11:2877 | https://doi.org/10.1038/s41467-020-16511-2 | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16511-2 T he human pathogen Mycoplasma genitalium, a member of Structural similarities of P140 and P110. The structure of the the pneumoniae cluster of mycoplasmas, binds to eukar- extracellular region of P140, with a bulky N-terminal domain yotic cells by means of its adhesion complex, the Nap. This (residues 23–1243) and a small C-terminal domain (residues complex is formed by two heterodimers, each consisting of 1244–1351), has an overall shape resembling the capital letter P proteins P110 and P1401–6. In addition to their roles in cytad- (Fig. 1a). The N-terminal domain consists of a seven-bladed (β- herence and motility, P110 and P140 are immunodominant sheet) propeller and a “crown” formed by the clustering of the proteins and constitute the main target of host antibodies during long polypeptide segments that emerge from the propeller infection7–9. Antibiotic resistance to human pathogens from the (Supplementary Fig. 3). β-Sheets I to VI each have four strands, pneumoniae cluster10–13 is increasing at an alarming rate, mak- while β-sheet VII, the last β-sheet in the propeller that connects ing it necessary to explore novel therapeutic strategies. Anti- directly with the C-terminal domain, has only two strands adherence molecules, aimed at preventing the establishment of (Fig. 1b). P140 and P110 share many features of domain orga- infection, are attractive potential antimicrobial drugs14,15. A deep nization and the topology of the secondary structural elements, understanding of the Nap structure and adhesion mechanism suggesting a common ancestor, although the degree of con- will facilitate the development of anti-adherence therapies. servation differs for the N-terminal and C-terminal domains Recently, we determined the crystal structure of the extracellular (Supplementary Fig. 4). The N-terminal domains, with an RMSD region of P110 and demonstrated its binding to sialic acid of 3.5 Å between the Cα atoms of 359 structurally equivalent receptors6. Here, we address the structure and mechanism of the residues (∼28%), are markedly different in the crown. In contrast, Nap adhesion complex and reveal an intricate interplay between the C-terminal domains are closely related, with 74 equivalent P110 and P140. residues (∼71%) and an RMSD of 2.2 Å. Results Binding site analysis of the Nap. The binding site for the sia- Crystal structure of P140 and in complex with P110N. Crystals lylated oligosaccharides, identified in the structure of P1106, is were obtained from the extracellular region of P140 (residues located at the interface between P140 and P110 in the crystal 23–1351) (Fig. 1, Supplementary Figs. 1 and 2), both alone and structure of the P140–P110N complex (Fig. 2a–d). Interaction of in complex with the N-terminal domain of P110 (P110N: the two subunits changes the position of the sialic binding β- residues 23–827) (Fig. 2a, Supplementary Fig. 2). The structure hairpin while the interfering loop (P140 residues 807–827) inserts of P140, for which there are no molecular models or experi- into the binding pocket of P110N, sterically interfering with the mental phases available, was determined by density modifica- binding of oligosaccharides to the complex (Fig. 2c–d). In tion techniques, starting with a mask derived from the agreement with this, surface plasmon resonance analysis shows sub-tomogram-averaged map of the whole Nap obtained by that, in solution, sialylated compounds 3SL and 6SL (neuraminic cryo-electron tomography (cryo-ET) (see Methods). With four acid forming an α2–3 or an α2–6 linkage to a lactose mono- heterodimers in the asymmetric unit, the P140–P110N crystals saccharide, respectively), which bind to P110 alone6, do not bind were refined at 2.65 Å resolution to a final model with agree- either to P140 alone or to P140–P110 complexes (Supplementary ment R and Rfree factors of 18.7 and 22.4, respectively (Sup- Fig. 5). The structure of the P140–P110N complex suggests that plementary Table 1). P110 residues Gln460–Asp461, from the binding β-hairpin, and a b IV Interfering loop V III P140 Crown Vl 1 1 2 II 3 2 4 80 Å N-domain Nter Vll I β-Propeller 70° N–C domain interface IV V III C-domain 40 Å Nter (K53) Vl 1 1 2 II Vll 2 Cter (N1347) 3 Nter 4 I N–C domain Plasma interface membrane Cytoplasm Cter Fig. 1 Structure of P140. a P140 contains a large N-terminal domain and a smaller C-terminal domain, which is followed in sequence by a predicted transmembrane helix. The N-terminal domain has two distinct regions: the β-propeller (green) and the crown (blue). b Two views of the β-propeller, 70° apart. The first and last β-sheets of the propeller (I and VII, respectively) interact with the C-terminal domain and are structurally contiguous in the propeller ring. A large β-bulge (residues 69–77, in cyan) occludes the center of the propeller ring. The presence of only two strands in the last β-sheet (VII) constrains the interface between the N-terminal and C-terminal domains. 2 NATURE COMMUNICATIONS | (2020)11:2877 | https://doi.org/10.1038/s41467-020-16511-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16511-2 ARTICLE b c d Interfering 3’Sialyllactose Loop 468–480 Interfering loop loop T816 N815 N812 N895 β-Hairpin S458 Q460 β-Hairpin T898 N634 D461 R600 Loop 586–612 a Sialic acid e Cryo-EM P140-P110 binding site (closed conformation) P110 N–C domain P140 P110N interface P140 X-ray P140-P110N (closed conformation) g f Fraction of free mycoplasma cells 1.0 P110-R600A dP110 0.8 R834 D836 P110-RQD P110-W838F W838 0.6 P110-Y830A P110-R834G K+ 0.4 P110-D836L H 2O H2O 0.2 P110-WT Y830 WT 0.0 N–C domain interface 0 1 × 1004 2 × 1004 3 × 1004 mutated residues –1 hRBCs·μl Arg600, close to the receptor-binding site, play an important role Single-particle cryo-EM of the P140–P110 extracellular region. in the interaction with P140 (Fig. 2b, d). P110 variants carrying Using a sample of P140–P110 complexes, with the complete an R600A substitution or the triple substitution RQD (R600A, extracellular region included for both subunits (P140 residues Q460A, and D461A) were introduced by transposon delivery into 23–1351 and P110 residues 23–938), we performed single-particle an M. genitalium P110 null mutant1. Strains expressing the P110- cryo-electron microscopy (cryo-EM). We obtained a map with an RQD variant protein, which was barely detectable, showed a null overall resolution of 4.1 Å, although non-isotropic (Fig. 2e, Sup- binding capacity phenotype (Fig. 2g, Supplementary Fig. 6b). The plementary Table 2, Supplementary Figs. 7 and 8). The variant protein P110-R600A was well expressed, but the strain P140–P110N X-ray structure could be fitted as a rigid-body presented no capacity for adherence and characterization of cell without modifications into the P140–P110 cryo-EM map with motility was not feasible. UCSF Chimera16 (Supplementary Table 3, Supplementary Fig. 7, NATURE COMMUNICATIONS | (2020)11:2877 | https://doi.org/10.1038/s41467-020-16511-2 | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16511-2 Fig. 2 Closed conformation of the P140–P110 heterodimers by X-ray and cryo-EM. a X-ray crystal structure of the P140–P110N heterodimer. b Network of hydrogen bonds in the vicinity of the oligosaccharide-binding site between P140 (blue) and P110N (yellow). Rearrangements experienced by P110N (c) and P140 (d) with respect to the structures of P110 alone (orange) and P140 alone (light blue), respectively. Rearrangements of the ß-hairpin and the interfering loop are indicated with red arrows. Oligosaccharide 3SL (c) is depicted as it was found in the structure of P110 alone6, showing that it would have steric clashes in the closed conformation of the P140–P110N complex. e Cryo-EM structure of the P140–P110 heterodimer. The C-terminal domain of P140 is missing in the structure because it was not defined in the cryo-EM map (Supplementary Fig. 7). The C-terminal domain of P110 was visible in the cryo-EM map, although with a local resolution lower than that for the N-terminal domains. f Details of the interface (depicted in green) between the N- terminal and C-terminal domains, close to where a potassium ion was found in the structures of P110 alone. g Hemadsorption assays by fluorescence- activated cell sorting (FACS) analysis. Graphic represents inverse Langmuir plots (see also Supplementary Information) containing a fixed amount of mycoplasma cells and increasing amounts of human red blood cells (hRBCs). Plots represent the best-fitting curves to a series of hemadsorption measurements obtained from at least two biological repeats for each strain. We performed a double-gating strategy, using a preliminary FL3-H/FL2-H gate following an SSC-H/FL1-H gate. Mutations are either in the vicinity of the oligosaccharide-binding site at the interface between P140 and P110 (residues explicitly shown in panel b) or close to the interface between the N-terminal and C-terminal domains of P110 (residues shown in panel f). R2 for each fitting curve: WT (0.8576), dP110 (0.9054), P110-WT (0.9921), P110-Y830A (0.9663), P110-W838F (0.8339), P110-R834G (0.8757), P110-D836L (0.9993), P110-R600A (0.8663), and P110-RQD (0.9999). Supplementary Movie 1). Therefore, the structure of the complete Nap particles where the extracellular region was well P140–P110 complex found by cryo-EM corresponds to the con- defined. The best map, with an overall resolution of 9.8 Å, allowed formation of the X-ray P140–P110N structure, where access to accurate rigid-body fitting with Chimera16 of the P140 structure the sialylated oligosaccharides binding site is occluded (Supple- alone and the P110 structure alone, confirming the presence of a mentary Movie 1). In the P140–P110 cryo-EM map, there is no dimer of P140–P110 heterodimers in each Nap (Fig. 3a–c, Sup- density for the C-terminal domain of P140, whereas the C- plementary Fig. 9, Supplementary Table 2). The arrangement of terminal domain of P110, which is absent in the P140–P110N P140–P110 heterodimers is essentially identical to the X-ray and complex, is visible, albeit with a weak density (Fig. 2e, Supple- cryo-EM P140–P110 structures, with the interactions between mentary Fig. 7). This indicates a significant flexibility of the C- both subunits preventing access to the sialylated oligosaccharide- terminal domains with respect to the bulkier N-terminal binding site. This is the “closed” conformation of the Nap domains. In the P140–P110 complex found by cryo-EM, the (Supplementary Movie 2). C-terminal domains act like stalks interface between subunits spans 2758 Å2 with 30 hydrogen connecting the large N-terminal domain of each subunit with the bonds and an estimated Gibbs free energy of –20 kcal/mol, outer surface of the cell membrane. The two P110 subunits are resulting in a 100% probability of the formation of the complex almost parallel to the dimer axis and face away from each other, (PISA server17). while the two P140 subunits of the Nap adopt a “V-shaped” arrangement with the C-terminal domains very close to each other (Fig. 3c). The distance between heterodimers is large Motility analysis of mutants. Five mutations were introduced (Fig. 3b, c), suggesting that their interaction is weak. This is in close to the interface between the N-terminal and C-terminal agreement with MALS measurements where only heterodimers domains, to check if either adhesion or motility was affected. The are detected when mixing equimolecular amounts of constructs five mutated residues (Y830A, R834G, D836L, W838F, and from the extracellular regions of P110 and P140 (Supplementary G839F) were chosen in the vicinity of the potassium-binding site Fig. 10). Therefore, there are two interfaces in the Nap between found in P1106 (Fig. 2e, f, Supplementary Table 4). The P110- the extracellular regions of P140 and P110, which we name the G839F strain showed no detectable levels of adhesins, while P110- “tight” and “loose” interfaces (Fig. 3b, e). The cryo-EM map of D836L exhibited hemadsorption values similar to those of wild- the Nap shows density corresponding to the Nap transmembrane type cells (Fig. 2g). The P110-Y830A and P110-R834G variants region and the positions of the transmembrane helices (their N- showed intermediate hemadsorption phenotypes, characterized terminal ends) can be seen (Fig. 3a, c). The close proximity of the by unusually high Bmax values (see Methods). Hemadsorption of P140 C-terminal domains also brings the N-terminal ends of the the P110-W838F mutant was severely impaired, with high Kd and corresponding transmembrane helices closer together. For P140 Bmax values. Motility was examined by time-lapse micro- and P110, sequence analysis indicates the presence of one cinematography to monitor the movements of individual cells for transmembrane helix containing one and two Engelman motifs 120 s (Supplementary Fig. 6, Supplementary Table 5). In the (GXXXG, with X any residue, in general hydrophobic), respec- P110-W838F mutant, cells were completely non-motile and, in tively, which are characteristic of high-affinity interactions in agreement with this extreme phenotype, phase contrast images membrane helices18,19. revealed the presence of large aggregates resulting from the inability of cells to spread. The Y830A, R834G, and D836L var- iants also exhibited altered gliding velocities, indicating that Cryo-electron tomography of the in situ Nap. The structure of structural integrity at the interface between the N-terminal and the in situ Nap complex was also determined by cryo-ET from C-terminal domains, away from the cell receptor-binding site, is mildly lysed M. genitalium cells3. Classification of Nap volumes critical for motility in M. genitalium. and subsequent sub-tomogram averaging provided an improved map for the most abundant class (~85%) at 15 Å resolution, in which the four subunits, two from P110 and two from P140, are Single-particle analysis of the Nap. Next, we performed single- clearly distinguished and present a nearly perfect twofold sym- particle cryo-EM using a purified sample of Nap complexes, metry (Fig. 3d–f, Supplementary Fig. 11). The rigid-body fitting obtained as previously described3 (see Methods). The purified of the structures of P110 alone and P140 alone into the cryo-ET Nap complexes contain full-length P140 and P110 proteins, density with Chimera16 reveals major differences from the cryo- including the transmembrane helices and cytoplasmic regions, EM structure of the Nap (Fig. 3, Fig. 4a, Supplementary Table 3). which are required for formation of tetramers. Classification In the cryo-ET structure, the longest axis of the four subunits indicated that only ∼15–20% of the images corresponded to runs parallel to the axis of twofold symmetry of the Nap. This 4 NATURE COMMUNICATIONS | (2020)11:2877 | https://doi.org/10.1038/s41467-020-16511-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16511-2 ARTICLE a Cryo-EM Nap d Cryo-ET Nap 117Å 124Å Crowns P110 Propellers P140 C-terminal domains Membranes Tight interface view 90° 90° 90° 90° b e Closed Open tight 117Å 124Å tight interface interface Loose interface Top view Loose interface 90° 90° c f Heterodimers interaction Loose interface view Fig. 3 The structure of Nap by single-particle cryo-EM and in situ cryo-ET. a–c Three different views of the single-particle cryo-EM from the Nap (beige surface) with the structures of P140 (blue) and P110 (yellow) fitted into the density. In a the crown, propeller, and C-terminal domain are indicated. b Top view of the Nap, depicting the loose and tight interfaces. An approximate twofold axis perpendicular to the cell membrane relates the pair of P110–P140 heterodimers of a Nap. Fitting of the P140 and P110 subunits is unambiguous. The interface between the P110 and P140 subunits, defined as tight in the cryo-EM structure of the Nap, is the same found in the cryo-EM structure of the P140–P110 heterodimer. The interfering loop and the binding site are indicated with a black star. c The lateral view along the loose interface shows the V-like shape adopted by the two P140 subunits. d–f Three views of the cryo-ET map of the Nap, with the structures of P140 and P110 accurately fitted. The membrane bilayer of the mycoplasma cell is clearly defined in the lower part of the lateral views (d, f). e The top view shows that the tight interface is wider in the cryo-ET “open” conformation than in the cryo-EM “closed” conformation of the Nap. f The lateral view along the loose interface shows an interaction between heterodimers. NATURE COMMUNICATIONS | (2020)11:2877 | https://doi.org/10.1038/s41467-020-16511-2 | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16511-2 a Closed Open conformation conformation Human epithelial cells membrane Receptor release Receptor binding Lateral view Mycoplasma membrane b 13° P110 D3 D3’ D2 D2’ D1 Bottom D1’ view D4 D4’ P140 13° c P140 D1 D2 D1’ 46Å D2’ 36Å C-terminal domains P110 D3 D4 D3’ D4’ 51Å 47Å Fig. 4 Structural rearrangements between “open” and “closed” conformations. a “Closed” (left panel) and “open” (right panel) conformations of the Nap complex. The “closed” conformation associated with the release state of the sialylated oligosaccharides cell receptors. The “open” conformation associates with the bound and ready-to-bind state. b Bottom view along the Nap twofold axis, depicting points of contact (black circles) between subunits and the mycoplasma cell outer membrane. These points also mark the start (the N-terminal ends) of the transmembrane helices. The N-terminal end positions are changing from a squared to rhomboid shape, when transitioning from the closed to the open conformation. c Changes to the relative positions of the C- terminal domains from P140 (upper panels) and P110 (lower panels). implies that between the cryo-ET and cryo-EM structures there is P140 and P110 subunits, which is very wide in the cryo-EM a rotation of P110 relative to P140 in the heterodimer and an structure of the Nap, is narrower in the cryo-ET structure, with a increase of 10 Å in the closest distance between the C-terminal few interactions between P140 and P110 (in particular, P140 domains of P140 subunits (Fig. 4b, c, Supplementary Fig. 12, residues 1175–1179 interacting with the P110 loop 750–755) that Supplementary Movies 3 and 4). The loose interface between are absent in the cryo-EM Nap (Fig. 3c, f). The tight interface 6 NATURE COMMUNICATIONS | (2020)11:2877 | https://doi.org/10.1038/s41467-020-16511-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16511-2 ARTICLE between P140 and P110, which is in the “closed” conformation in was centrifuged at 25,000 × g at 4 °C and the supernatant applied to a 5 ml Histrap the cryo-EM structure of the Nap, undergoes major rearrange- column (GE Healthcare, Pittsburgh, USA) equilibrated with 1× PBS as binding buffer and 1× PBS with 500 mM imidazole as elution buffer. Soluble aliquots of ments in the cryo-ET structure, becoming on average ∼7 Å wider His6-tagged P110N were pooled and loaded onto a HiLoad Superdex 200 16/60 (Fig. 3a, b, d, e). This widening “opens” the sialylated column (GE Healthcare) in buffer consisting of 50 mM Tris pH 7.4 and 150 mM oligosaccharide-binding site, making it accessible for binding in NaCl. The region corresponding to the MG_192 gene from M. genitalium (strain the cryo-ET structure of the Nap. Therefore, the cryo-ET and G37, residues 23–938, P110) was amplified from a synthetic clone (Supplementary cryo-EM structures of the Nap correspond to conformations that, Table 6) using primers P110-F and P110-R as Forward and Reverse, respectively (Supplementary Table 4). The region corresponding to the MG_191 gene from M. with respect to the sialylated host cell receptors, are respectively genitalium (strain G37, residues 23–1351) was amplified from a synthetic clone “open”, with the binding site accessible and ready for binding, (Supplementary Table 6) using primers P140F and P140R as Forward and Reverse, and “closed”, with the binding site occluded and not accessible for respectively (Supplementary Table 4). Protein production and purification for P140 binding. In the cryo-ET map, the cell membrane surrounding the and P110 followed the same protocol as for P110N6. Nap is visible as a double layer that is perpendicular to the Nap twofold axis. The positions of the C-terminal domains are largely Preparation of P140–P110 and P140–P110N complexes and SEC-MALS ana- lysis. Central fractions of the Superdex 200 16/60 column (GE Healthcare) from different between the cryo-ET and cryo-EM Nap structures, P140 and either P110N or P110 were mixed in a 1:1 ratio to a final concentration of changing from a squared to rhomboid shape, respectively ∼11 mg/ml (measured with an absorption coefficient of 1.32). Each mixture was (Fig. 4b). These differences also affect the positions of the incubated for 30 min at room temperature to obtain the heterodimeric complex. transmembrane helices that follow in sequence immediately after For protein oligomerization experiments, prior to sample injections, the multi- the C-terminal domains, which provide a mechanism for com- angle light scattering (MALS, WYATT Technologies & Corporation) detector was normalized with 25 µl of BSA at 5 mg/ml. Then 150 µl of each purified P140, P110, munication between the Nap extracellular and intracellular and P140–P110 sample at 0.6, 2, and 0.5 mg/ml, respectively, were injected at regions. On the intracellular side, the cryo-ET map shows a 0.5 ml/min using a Superdex 200 increase 10/300 GL column (GE Healthcare) compact density, similar to the one reported previously3, with a equilibrated with a running buffer consisting of 50 mM Tris pH 7.4 and 150 mM volume suited to about 400 residues and an approximate twofold NaCl. All samples were purified by filtering through 0.22 µm filters. symmetry, where the four subunits appear to interact closely as they cannot be individualized at the resolution available. Preparation of purified Nap complexes. A P110His strain was generated for purification of the Nap complexes from M. genitalium G37 cells (ATCC 33530). Produced by genetic engineering3, the strain carries a 6xHistag insertion in the Discussion MG192 gene. Four liters of the P110His strain grown in SP4 medium in suspension at 37 °C at 150 r.p.m. was harvested by centrifugation (16,000 × g, 4 °C, 30 min). Integration of the information obtained from the structural The pellet was washed twice with PBS with calcium and magnesium, followed by approaches shows that the Nap is composed of two P140–P110 cell disruption by sonication in PBS in the presence of 1 mM EDTA, 5 mM β- heterodimers with a loose interface between them in the extra- mercaptoethanol, 0.1 mM PMSF, and EDTA-free cocktail of protease inhibitors cellular region, suggesting that subunits in the Nap are held (Roche Diagnostics, Mannheim, Germany). The pellet generated after centrifuga- tion (70,000 × g, 4 °C) was resuspended in 75 mM Tris pH 7.4, 400 mM NaCl, 5% together mainly by interactions in the intracellular and trans- glycerol, and 2% n-octyl-β-D-glucopyranoside detergent by homogenization in a membrane regions. Each heterodimer of the Nap appears to glass homogenizer. Solubilization of membranes was done overnight at 4 °C in an undergo large structural rearrangements between the “open” and orbital shaker. Solubilized membranes were centrifuged at 50,000 × g, 30 min (4 °C) “closed” conformations, which are associated with attachment and the supernatant was purified by Ni2+-affinity chromatography in 75 mM Tris and release to cell receptors, respectively (Fig. 4, Supplementary pH 7.4, 400 mM NaCl, 5% glycerol, and 0.5% octylglucoside detergent. The purified Nap complex was obtained by Superose 6 size-exclusion chromatography equili- Fig. 12, Supplementary Movie 4). The “open” conformation brated with the same buffer. corresponds to a state in which the P140 and P110 extracellular regions from the four subunits of the Nap interact only weakly Crystallization of P140 alone and the P140–P110N complex. Screening for with each other, allowing P110 to remain in a conformation that initial crystallization conditions was performed with 150 nl droplets in 96-well is ready to bind or is bound to sialylated cell receptors. The plates on a Cartesian robot (Cartesian (TM) Dispensing Systems) for both P140 “closed” conformation corresponds to a state in which the tight alone and the P140–P110N complex. Optimized crystals from P140 were prepared by mixing 1 µl P140 at 6.5 mg/ml and 1 µl reservoir solution containing 20% interaction of the extracellular regions of P140 and P110 occlude PEG3350, 0.1 M BisTris Propane pH 7.5, and 0.2 M sodium sulfate hydrate at 20 °C the cell receptor-binding site. The “closed” conformation could in hanging drop in 24-well plates. Crystals from the P140–P110N complex were occur rapidly to release the Nap from the sialylated cell receptor. prepared by mixing 1 µl of P140–P110N complex at 10.95 mg/ml and 1 µl reservoir To avoid been trapped in the overall most stable state, the cycling solution containing 23% PEG 500MME, 4% PGA, and 0.1 M sodium cacodylate between “open” and “closed” conformations would require a net pH 6.5 at 20 °C in hanging drop in 24-well plates. All crystals were flash-cooled in liquid nitrogen with 20% glycerol as a cryo-protectant. input of energy in each Nap complex. The structural rearrange- ments in the Nap extracellular region can be transmitted to the X-ray data collection and structure determination. X-ray diffraction experi- cell interior through repositioning of the transmembrane helices. ments were performed at the Xaloc Beamline (ALBA, Spain). Data were processed In vivo variants of Nap complexes indicate that structural rear- with Xia221 using XDS22, Aimless and Pointless23 from the CCP4i suite of pro- rangements can play a critical role in both cell adhesion and grams24. P140 crystals belong to the space group C2 with six subunits in the motility. Results from this work, together with previous data, asymmetric unit, while P140–P110N crystals belong to the space group P21 and contain four heterodimers in the crystal asymmetric unit. For the P140–P110N could form the basis for developing therapeutic strategies tar- crystals, a partial molecular replacement solution was obtained with Phaser soft- geting the adhesion machinery of mycoplasmas from the pneu- ware25 using the N-terminal domain of the available structures of P110 (PDB code moniae cluster. 6RT3) as a search model. In these P140–P110N crystals, an initial mask was tentatively defined for P140 using the sub-tomogram averaged map of a whole Nap complex obtained from cryo-ET images of M. genitalium ghost cells. Density Methods modification in the crystal of the P140–P110N complex, by iterative non- Cloning, expression, and purification of P110N, P140, and P110. The region crystallographic symmetry averaging and solvent flattening followed by phase corresponding to the MG_192 gene from M. genitalium (strain G37, residues extension with DM26, allowed us to obtain a rough density for P140, in which a few 23–827, P110N) was amplified from a synthetic clone (Supplementary Table 6) secondary structural elements were recognizable. This density was sufficient to using primers P110N-F and P110N-R as Forward and Reverse, respectively provide an initial molecular replacement solution for the P140 crystals (with six (Supplementary Table 7). The PCR fragment was cloned into the expression vector subunits in the asymmetric unit) and for crystals of the orthologous protein P1 pOPINE20 (gift from Ray Owens plasmid #26043, Addgene, Watertown, USA) to from M. pneumoniae (now deposited in the PDB with code 6RC9). Averaging generate a C-end His6-tagged protein. The recombinant protein was obtained after within and between crystals, while updating and refining the masks, provided expression in B834 (DE3) cells (Merck, Darmstadt, Germany) at 20 °C o/n upon electron density maps from which it was possible to progressively build in parallel induction with 1 mM IPTG at 0.6 OD600. Cells were harvested and lysed in 1× the structures of P140 and P1. The weak selenium methionine anomalous data, phosphate-buffered saline (PBS) buffer by sonication. Subsequently, the cell extract available from both the P140 and the P140–P110N crystals, were used to confirm NATURE COMMUNICATIONS | (2020)11:2877 | https://doi.org/10.1038/s41467-020-16511-2 | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16511-2 the sequence assignment to the P140 structure. Final models were traced with particle defocus estimation using a search range from 300 nm around the estimated Coot27 and refined with Refmac528 (Supplementary Table 1). values from the whole micrographs. To correct for local motion and for radiation damage, we used the Bayesian polishing function of Relion 3.0, in which the resolution-dependent decay caused by radiation damage is taken into account36. Cryo-ET and sub-tomogram averaging. Ghost cells were prepared in a similar The polished particles were subjected to another round of 2D classification to manner as reported previously3. Briefly, cells of the adherently growing M. geni- remove remaining junk particles. The remaining 404,064 particles for the hetero- talium G37 strain (ATCC 33530) were cultivated on 300-mesh lacey carbon EM dimer and 28,000 particles for the Nap complex were used for 3D classification grid (PLANO GmbH, Wet zlar, No. S166-3) in 3 ml SP4 medium, rinsed for 1 min using the initially generated density map filtered to 60 Å as reference. For the with PBS before incubation with 20 mM TEA pH 7.5, 0.5 M KCl, and 1% (v/v) heterodimer, six classes were generated using 3D classification, of which one class Triton X-100 for 1–5 min in the Vitrobot Mark IV (Thermo Fisher Scientific, (37,009 particles) showed electron density in the stalk region for P110 and was Waltham, USA) humidity chamber at 95% humidity and 15 °C. Before plunge refined using 3D auto-refine. For the Nap complex, six classes were generated using freezing, 5 nm gold fiducial markers (Protein A-conjugated to gold beads, Cell 3D classification, of which two classes (11,737 particles) were tetrameric. The one Biology Department, UMC Utrecht) were added. Tilt-series were recorded using tetrameric class containing 5612 particles was refined using 3D auto-refine. Gold- SerialEM 3.5.829 on a 300 kV Titan Krios (Thermo Fisher Scientific) in energy- standard Fourier shell correlations (FSCs) were calculated during the 3D refine- filtered transmission electron microscopy (EFTEM) mode using a GIF Quantum® ment between two independently refined halves of the data. The resulting map was SE post-column energy filter (Gatan, Inc., California, USA) set to 20 eV slit width. A post-processed to exclude solvent regions from the FSC calculation and to perform K2 Summit detector (Gatan Inc.) was operated in counting mode with a dose rate of sharpening with a temperature factor of –78 Å2 for the heterodimer and –218 Å2 ~1.0e–/pixel/s, using a frame rate of 0.25 s and a total dose in the range of 40–50e–/ for the Nap complex resulting in a global resolution of 4.1 Å for the heterodimer Å 2. A nominal magnification of ×64,000 was used with the K2 in super-resolution and 9.8 Å for the Nap complex, using the 0.143 criterion. Local resolution was mode, resulting in a pixel size of 1.1 Å/pixel. The tilt-series generally covered an estimated using Relion 3.0, showing high resolution in the core regions of the angular range from –60° to +60° with an angular increment of 3° and defocus set dimer and low resolution in the surface and stalk regions (Supplementary Figs. 8 between –2 and –4 µm using either dose symmetric or bidirectional acquisition. and 9). Tomograms were reconstructed by super-sampling SART30 with 3D contrast transfer function (CTF) correction. For volume visualization and isosurface ren- dering, the EMpackage in Amira was used (Thermo Fisher Scientific)31. Sub- Surface plasmon resonance. For binding assay experiments, a Biacore 3000 tomogram averaging was performed in a similar manner as reported previously3. In biosensor platform (GE Biosystems) equipped with a research-grade streptavidin- short, Nap complexes were manually selected and pre-aligned according to their coated biosensor chip SA was used. The chip was docked into the instrument and orientation on the membrane, which provided a strong constraint lowering the preconditioned with three 1-min injections of 1 M NaCl in 50 mM NaOH. Both degrees of freedom for sub-tomogram averaging. This first pre-aligned average was 3SL-PAA-biotin and 6SL-PAA-biotin (Carbosynth) oligosaccharides were injected used as the starting reference template for the iterative sub-tomogram averaging over the second and third flow cell, respectively, at 10 µg/ml diluted in HBS-P procedure, thus completely avoiding any initial model bias. The “open” structure for (10 mM Hepes, pH 7.4, 0.15 M NaCl, and 0.005% P20). The first cell was left blank the cryo-ET data is maintained when using the “closed” structure as a starting to serve as a reference. The running buffer consisted of HBS-P at a flow rate of reference (Supplementary Fig. 13). We then averaged 11,000 Nap sub-tomograms. 30 µl/min and the immobilization levels acquired were ~160 and ~180 response Post-processing involved refinement by projection matching and dose weighting of units for 3SL-PAA-biotin and 6SL-PAA-biotin, respectively. A series of diluted the wedge components. After classification, 8800 Nap sub-tomograms were inclu- purified extracellular P140 and P140–P110 samples in HBS-P (1.25, 2.5, 5, 10, and ded in the final average. All processing was performed using sub-tomograms with a 20 µM) were injected over the flow cell surface at 30 µl/min. Interaction analysis voxel size of 0.44 nm. The 0.5 criterion was used for resolution estimation. For were performed at 25 °C and the protein was allowed to associate and dissociate for classification of sub-tomograms, the averaged particles were normalized, and single 60 and 90 s, respectively, followed by a 30 s regeneration injection step of 0.05% projection slices in X, Y, and Z directions were extracted using custom MATLAB SDS at 30 µl/min. scripts, which are available on GitHub (https://github.com/uermel/Artiatomi). Subsequently, the two-dimensional (2D) projection slices were imported into Strains, culture conditions, and primers. M. genitalium was grown in SP4 Relion32 and 2D classification without image alignment was performed. Constituent medium (Hardy Diagnostics) at 37 °C in tissue culture flasks. Mutants were iso- sub-tomograms from the respective 2D classes were summed together to create 3D lated on SP4 agar plates supplemented with puromycin (3 µg/ml). All M. genita- classes. Molecular graphics rendering and analyses were performed with either lium strains used in this work are listed in Supplementary Table 4. Escherichia coli UCSF Chimera16 or UCSF ChimeraX33 (Resource for Biocomputing, Visualization, XL1-Blue strain (Agilent, Santa Clara, USA) was used for cloning and plasmid and Informatics at the University of California, San Francisco, USA). In all cases, amplification purposes. It was grown in LB broth or on LB-agar plates containing rigid-body transformations were used to fit X-ray structures into the cryo-EM 100 µg/ml ampicillin. Primers used are listed in Supplementary Table 7. density maps using the fit-in-map function in UCSF Chimera. DNA manipulation and mutant construction. Plasmid DNA was purified using Single-particle cryo-EM. For single-particle cryo-EM, a 3.5 μl aliquot of 20 μg/ml GeneJET Plasmid Miniprep Kit (Thermo Fisher Scientific). PCR products and purified P140–P110 heterodimer, or 25 μg/ml purified Nap complex, was applied to DNA fragments were recovered from agarose gels using NucleoSpin Gel and PCR a (30 s) glow-discharged R1.2/1.3 Quantifoil grid (Quantifoil, Großlöbichau, Ger- Clean-up Kit (Macherey-Nagel, Düren, Germany), and digested using the corre- many), and plunge-frozen in liquid ethane (Vitrobot Mark IV, Thermo Fisher sponding restriction enzymes (Thermo Fisher Scientific) when necessary. For Scientific) at 100% humidity, 4 °C, blot force –3, wait time 60 s, with a blotting time transformation of M. genitalium, plasmids were purified using the GenElute HP of 15 s for the heterodimer and 20 s for the Nap complex. Dose-fractionated movies Midiprep Kit (Sigma-Aldrich, St. Louis, USA) following the manufacturer’s were collected at a nominal magnification of ×130,000 (1.05 Å per pixel) in instructions. nanoprobe EFTEM mode at 300 kV with a Titan Krios (Thermo Fisher Scientific) P110 variants were generated in a two-step PCR procedure using DNA from electron microscope using a GIF Quantum S.E. post-column energy filter in zero plasmid pTnPacP110-WT as a template6. For each mutant, the first PCR round loss peak mode and a K2 Summit detector (Gatan Inc.). For the heterodimer, 32 was performed using primer COMmg192-F and the specific mg192 reverse primer, frames of 0.2 s each were collected and for the Nap complex 34 frames per or primer COMmg192-R and the specific mg192 forward primer (Supplementary micrograph were collected. The camera was operated in dose-fractionation Table 4). To reconstitute the mutated full-length MG_192 alleles, we conducted counting mode with a dose rate of ~7.7 electrons per Å2 s−1 for the heterodimer splicing by overhang extension (SOE) PCR using the specific amplicon pair and ~7.4 electrons per Å2 s−1 for the Nap complex, resulting in both cases in a total obtained for each mutant as a template (R600A, Y830A, R834G, D836L, W838F, dose of ~50 electrons per Å2. Defocus values ranged from –1 to –4 μm with and G839F) and primers COMmg192-F and COMmg192-R. Then, the mutated marginal (<0.1 μm) astigmatism. P110 alleles were digested with ApaI and XhoI and ligated into a similarly digested pMTnPac plasmid backbone6 to generate the corresponding pMTnPacP110 plasmid series. The P110 variant carrying the triple substitution RQD was obtained Single-particle cryo-EM image processing. Relion 3.0 was used for the whole- using primers Q460AD461Amg192-F and -R to introduce the double substitution image processing workflow32 unless stated otherwise. Beam-induced motion cor- Q460A and D461A into the pMTnPacP110-R600A plasmid, which was used as the rection was performed using UCSF MotionCor234. The CTF of each micrograph DNA template for the PCR reaction. Sequencing analysis of the different was estimated using GCTF35. As reference for autopicking the heterodimer, the TnPacP110 minitransposons, using primers Tnp3, RTPCR192-F, RTPCR192-R, crystal structure filtered to 20 Å was used. As a reference for the Nap complex, the and PacUp, ruled out the presence of additional mutations in the crystal structure of the dimer was fitted into the cryo-ET data and subsequently MG_192 sequence. These plasmids were transformed into a M. genitalium lowpass filtered to 20 Å. Cross-checking experiments with Laplacian-of-Gaussian MG_192 null mutant to create the different P110 variant strains. Identification of autopicking were performed for the Nap to verify that the structure was not the minitransposon insertion site in the individual clones was done by sequencing reference-biased. For the heterodimer, 1,024,402 particles were extracted from 2751 using the PacDown primer and chromosomal DNA as a template. micrographs, and for the Nap complex, 160,203 particles were extracted from 2225 micrographs, using a 270 pixel box in both cases. All images were normalized to make the average intensity of the background equal to zero during pre-processing. Transformation and screening. M. genitalium MG_192 null mutants were False-positive particles were removed manually or by unsupervised 2D classifica- transformed by electroporation using 5 µg of plasmid DNA of the different tion. The CTF-refinement function from Relion 3.0 was used to perform per- minitransposons, as previously described37. Puromycin-resistant colonies were 8 NATURE COMMUNICATIONS | (2020)11:2877 | https://doi.org/10.1038/s41467-020-16511-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16511-2 ARTICLE picked, propagated, and stored at –80 °C. For screening purposes, strains were References further propagated in 25 cm2 tissue culture flasks with puromycin and lysed using 1. Burgos, R. et al. Mycoplasma genitalium P140 and P110 cytadhesins are 0.1 M Tris-HCl pH 8.5, 0.05% Tween-20, and 250 µg/ml Proteinase K for 1 h at reciprocally stabilized and required for cell adhesion and terminal-organelle 37 °C. Then, Proteinase K was inactivated at 95 °C for 10 min. M. genitalium lysates development. J. Bacteriol. 188, 8627–37 (2006). were screened by sequencing using the PacDown primer. By contrast, the MG_192 2. Mernaugh, G. R., Dallo, S. F., Holt, S. C. & Baseman, J. B. Properties of alleles were fully re-sequenced to rule out the presence of undesired mutations. adhering and nonadhering populations of Mycoplasma genitalium. Clin. Infect. Dis. 17(Suppl 1), S69–78 (1993). Sequencing reactions. Sequencing reactions were performed with the BigDye® 3. Scheffer, M. P. et al. Structural characterization of the NAP; the major v3.1 Cycle Sequencing kit using 2.5 μl of genomic DNA, following the manu- adhesion complex of the human pathogen Mycoplasma genitalium. Mol. facturer’s instructions (Thermo Fisher Scientific). All sequencing reactions were Microbiol. 105, 869–879 (2017). analyzed using an ABI PRISM 3130xl Genetic Analyser at the Servei de Genòmica i 4. Nakane, D., Adan-Kubo, J., Kenri, T. & Miyata, M. Isolation and Bioinfòrmatica (UAB). characterization of P1 adhesin, a leg protein of the gliding bacterium Mycoplasma pneumoniae. J. Bacteriol. 193, 715–22 (2011). 5. Garcia-Morales, L., Gonzalez-Gonzalez, L., Querol, E. & Pinol, J. A minimized SDS-PAGE. Whole-cell lysates were obtained from mid-log phase cultures grown motile machinery for Mycoplasma genitalium. Mol. Microbiol 100, 125–38 in 75 cm2 flasks. Protein concentration was determined with the PierceTM BCA (2016). Protein Assay Kit (Thermo Fisher Scientific), and similar amounts of total protein were separated by SDS-PAGE following the standard procedures. 6. Aparicio, D. et al. Mycoplasma genitalium adhesin P110 binds sialic-acid human receptors. Nat. Commun. 9, 4471 (2018). 7. Ma, L., Jensen, J. S., Mancuso, M., Myers, L. & Martin, D. H. Kinetics of Quantitative hemadsorption assay. We used 109 mycoplasma cells during the genetic variation of the Mycoplasma genitalium MG192 Gene in hemadsorption assay. Fluorescence-activated cell sorting (FACS) data were experimentally infected chimpanzees. Infect. Immun. 84, 747–53 (2015). acquired using a FACSCalibur (Becton Dickinson, Franklin Lakes, USA) equipped 8. Wood, G. E. et al. Persistence, immune response, and antigenic variation of with an air-cooled 488 nm argon laser and a 633 nm red diode laser and analyzed Mycoplasma genitalium in an experimentally infected pig-tailed macaque with the CellQuest-Pro and FACSDiva software (Becton Dickinson). Hemad- (Macaca nemestrina). Infect. Immun. 81, 2938–51 (2013). sorption was quantified using flow cytometry as previously described38 with few 9. Wood, G. E., Patton, D. L., Cummings, P. K., Iverson-Cabral, S. L. & Totten, P. modifications. Binding of mycoplasma cells to red blood cells can be modeled in an A. Experimental infection of pig-tailed macaques (Macaca nemestrina) with inverse Langmuir isothermal kinetic function: Mycoplasma genitalium. Infect. Immun. 85, e00738-16.(2017). Bmax ½RBC 10. Cao, B. et al. High prevalence of macrolide resistance in Mycoplasma Mf ¼ 1  : ð1Þ K d þ ½RBC pneumoniae isolates from adult and adolescent patients with respiratory tract The plots represent the best-fitting curves to a series of hemadsorption infection in China. Clin. Infect. Dis. 51, 189–194 (2010). measurements obtained from at least two biological repeats for each strain. We 11. Parrott, G. L., Kinjo, T. & Fujita, J. A compendium for Mycoplasma performed a double-gating strategy, using a preliminary FL3-H/FL2-H gate pneumoniae. Front. Microbiol. 7, 513 (2016). following an SSC-H/FL1-H gate, as described previsouly38. 12. Bradshaw, C. S. et al. Azithromycin failure in Mycoplasma genitalium urethritis. Emerg. Infect. Dis. 12, 1149–52 (2006). 13. Couldwell, D. L. & Lewis, D. A. Mycoplasma genitalium infection: current Time-lapse microcinematography. Gliding properties of the wild type and treatment options, therapeutic failure, and resistance-associated mutations. mutant strains were analyzed by time-lapse microcinematography as previously Infect. Drug Resist. 8, 147–61 (2015). described5. Samples from mid-log phase cell cultures were grown overnight on 14. Ofek, I., Hasty, D. L. & Sharon, N. Anti-adhesion therapy of bacterial diseases: eight-well ibiTreat µ-slides (Ibidi, Gräfelfing, Germany) filled with 200 µl of SP4 prospects and problems. FEMS Immunol. Med. Microbiol. 38, 181–91 (2003). medium. Prior to the observation, medium was replaced with fresh SP4 pre- 15. Sharon, N. Carbohydrates as future anti-adhesion drugs for infectious warmed at 37 °C and cell movement was examined at 37 °C and 5% CO2 using a diseases. Biochim. Biophys. Acta 1760, 527–537 (2006). Nikon Eclipse TE 2000-E inverted microscope equipped with a CO2 Microscope 16. Pettersen, E. F. et al. UCSF chimera—a visualization system for exploratory Cage Incubation System (Okolab, San Bruno, USA). Images were captured at 2 s research and analysis. J. Comput. Chem. 25, 1605–12 (2004). intervals for a total of 2 min for all different strains. The frequency of motile cells was determined by examining approximately 250 isolated cells of each strain. 17. Krissinel, E. & Henrick, K. Inference of macromolecular assemblies from Given that cells of the P110-W838F mutant were mainly aggregated, only 64 crystalline state. J. Mol. Biol. 372, 774–97 (2007). individual cells could be analyzed. Mean velocities were obtained from the analysis 18. Russ, W. P. & Engelman, D. M. The GxxxG motif: a framework for of at least 25 motile, isolated cells, by measuring the traveled distance and dividing transmembrane helix-helix association. J. Mol. Biol. 296, 911–9 (2000). this value by the time spent in microcinematography. Analysis of the gliding 19. Seto, S., Kenri, T., Tomiyama, T. & Miyata, M. Involvement of P1 adhesin in properties was performed using the ImageJ software with the MTrack2, MTrackJ, gliding motility of Mycoplasma pneumoniae as revealed by the inhibitory wrMTrack, and Stack Deflicker plugins (https://imagej.nih.gov/ij/). effects of antibody under optimized gliding conditions. J. Bacteriol. 185, 1875–7 (2005). 20. Berrow, N. S. et al. A versatile ligation-independent cloning method suitable Reporting summary. Further information on research design is available in for high-throughput expression screening applications. Nucleic Acids Res. 35, the Nature Research Reporting Summary linked to this article. e45 (2007). 21. Winter, G. xia2: an expert system for macromolecular crystallography data Data availability reduction. J. Appl. Crystallogr. 43, 186–190 (2010). Atomic coordinates and structure factors for the reported crystal structures of P140 and 22. Kabsch, W. Xds. Acta Crystallogr D Biol. Crystallogr. 66, 125–32 (2010). the P140–P110N complex have been deposited in the Protein Data Bank under accession 23. Evans, P. Scaling and assessment of data quality. Acta Crystallogr. Sect. D 62, codes 6RUT and 6S3U, respectively. The cryo-electron microscopy density for the 4.1 Å 72–82 (2006). density map of the heterodimer was deposited in the Electron Microscopy Data Bank 24. CCP4. The CCP4 suite: programs for protein crystallography. Acta Crystallogr (EMDB) under the accession code EMD-10890 and the fitted P140–P110N complex D. Biol. Crystallogr 50, 760–763 (1994). under the accession code 6YRK. The 9.8 Å density map of the single-particle cryo-EM 25. McCoy, A. J. Solving structures of protein complexes by molecular density map of the Nap was deposited in the EMDB under the accession codes EMD- replacement with Phaser. Acta Crystallogr. Sect. D Biol. Crystallogr. 63, 32–41 10260. The 15 Å in situ density map of the in situ cryo-ET Nap was deposited in the (2007). EMDB under the accession code EMD-10259. The source data underlying Fig. 2g are 26. Cowtan, K. D. & Main, P. Phase combination and cross validation in iterated provided as a Source Data file. Other data are available from the corresponding authors density-modification calculations. Acta Crystallogr. Sect. D 52, 43–48 (1996). upon reasonable request. 27. Emsley, P. & Cowtan, K. Coot: model-building tools for molecular graphics. Acta Crystallogr D Biol. Crystallogr 60, 2126–32 (2004). 28. Murshudov, G. N. et al. REFMAC5 for the refinement of macromolecular crystal Code availability structures. Acta Crystallogr. Sect. D Biol. Crystallogr. 67, 355–367 (2011). The reconstruction, sub-tomogram averaging, and classification code referenced in the 29. Mastronarde, D. N. Automated electron microscope tomography using robust methods is available: prediction of specimen movements. J. Struct. Biol. 152, 36–51 (2005). (i) Artiatomi: http://github.com/uermel/Artiatomi 30. Kunz, M. & Frangakis, A. S. Three-dimensional CTF correction improves the (ii) Classification: https://github.com/lsprankel/Aparicio_2020. resolution of electron tomograms. J. Struct. Biol. 197, 114–122 (2017). 31. Pruggnaller, S., Mayr, M. & Frangakis, A. S. A visualization and segmentation Received: 22 November 2019; Accepted: 6 May 2020; toolbox for electron microscopy. J. Struct. Biol. 164, 161–5 (2008). 32. Zivanov, J. et al. New tools for automated high-resolution cryo-EM structure determination in RELION-3. eLife 7, e42166 (2018). NATURE COMMUNICATIONS | (2020)11:2877 | https://doi.org/10.1038/s41467-020-16511-2 | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16511-2 33. Goddard, T. D. et al. UCSF ChimeraX: meeting modern challenges in Competing interests visualization and analysis. Protein Sci. 27, 14–25 (2018). The authors declare no competing interests. 34. Zheng, S. Q. et al. MotionCor2: anisotropic correction of beam-induced motion for improved cryo-electron microscopy. Nat. Methods 14, 331–332 (2017). 35. Zhang, K. Gctf: real-time CTF determination and correction. J. Struct. Biol. Additional information Supplementary information is available for this paper at https://doi.org/10.1038/s41467- 193, 1–12 (2016). 020-16511-2. 36. Scheres, S. H. A Bayesian view on cryo-EM structure determination. J. Mol. Biol. 415, 406–418 (2012). Correspondence and requests for materials should be addressed to O.Q.P., I.F. or A.S.F. 37. Pich, O. Q., Burgos, R., Planell, R., Querol, E. & Pinol, J. Comparative analysis of antibiotic resistance gene markers in Mycoplasma genitalium: application to Peer review information Nature Communications thanks Jeremy Derrick, Makoto studies of the minimal gene complement. Microbiology 152, 519–527 (2006). Miyata, and the other, anonymous, reviewer(s) for their contribution to the peer review 38. Garcia-Morales, L., Gonzalez-Gonzalez, L., Costa, M., Querol, E. & Pinol, J. of this work. Peer review reports are available. Quantitative assessment of mycoplasma hemadsorption activity by flow cytometry. PLoS ONE 9, e87500 (2014). Reprints and permission information is available at http://www.nature.com/reprints Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in Acknowledgements published maps and institutional affiliations. We thank Utz Ermel for providing custom Matlab scripts. This work was supported by grants BFU2018-101265-B-100 and BIO2017-84166-R from MINECO (Spain). A.S.F. acknowledges the Deutsche Forschungsgemeinschaft Grant FR1653/6-1 and SFB902, as well as the Loewe Dynamem funding for the project. D.A. acknowledges María de Open Access This article is licensed under a Creative Commons Maeztu Unit of Excellence grant MDM-2014-0435. Many thanks are given to the Attribution 4.0 International License, which permits use, sharing, XALOC beamline team at ALBA for all the support during data collection and, in adaptation, distribution and reproduction in any medium or format, as long as you give particular, to Dr. Barbara M. Calisto for her many contributions to the mycoplasma appropriate credit to the original author(s) and the source, provide a link to the Creative structural studies. Many thanks are also given to the Crystallography Platform at the Commons license, and indicate if changes were made. The images or other third party Barcelona Science Park (PCB). material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory Author contributions regulation or exceeds the permitted use, you will need to obtain permission directly from Conceived and designed the experiments: D.A., M.P.S., J.P., O.Q.P., I.F., and A.S.F. the copyright holder. To view a copy of this license, visit http://creativecommons.org/ Performed the experiments: D.A., M.P.S., M.M.-S., D.V., M.R., M.S.W., A.S., J.R., L.S., licenses/by/4.0/. S.T.-P., L.G.-G., and O.Q.P. Analyzed the data: D.A., M.P.S., M.M.-S., D.V., M.S.W., J.R., L.S., J.P., O.Q.P., I.F., and A.S.F. Contributed reagents/materials/analysis tools: E.Q., J.P., I.F., and A.S.F. Wrote the paper: D.A., M.S., O.Q.P., I.F., and A.S.F. © The Author(s) 2020 10 NATURE COMMUNICATIONS | (2020)11:2877 | https://doi.org/10.1038/s41467-020-16511-2 | www.nature.com/naturecommunications