Université Paris Descartes École doctorale Frontières du vivant (ED 474) Unité de pathogénèse des Infections Vasculaires – Institut Pasteur

NEW INSIGHTS INTO MENINGOCOCCAL PATHOGENESIS:

EXPLORING THE ROLE OF THE MAJOR PILIN PilE IN THE FUNCTIONS OF TYPE IV PILI

Par Paul Kennouche

Thèse de doctorat de Biologie Dirigée par le Dr. Guillaume Duménil

Présentée et soutenue publiquement le 14 juin 2018

Devant un jury composé de : Pr. Jeremy Derrick Rapporteur Pr. Han Remaut Rapporteur Dr. Olivera Francetic Examinatrice Dr. Alexandra Walczak Examinatrice Dr. Guillaume Duménil Directeur de thèse

À mes formidables grand-mères,

À toi Nenès, qui ne m’auras pas vu « gagner le dernier bac ». À toi Manou, c’en est fini de « l’École Nationale Scientifique ».

Outline

INTRODUCTION 1

1 A HISTORICAL OVERVIEW OF THE DIVERSITY OF PROKARYOTIC APPENDAGES 3

1.1 THE FIRST OBSERVED APPENDAGES ARE ASSEMBLED BY TYPE THREE SECRETION SYSTEMS ...... 3

1.1.1 Flagella: rotating bacterial filaments 4

1.1.1.1 Diversity of flagellar systems 4

1.1.1.2 Functions: motility and more 5

1.1.1.3 Structure and assembly 7

1.1.2 The injectisome: needles assembled by the type three secretion system 8

1.1.2.1 Relationships to the 8

1.1.2.2 Diversity of injectisomes 8

1.1.2.3 A translocation machine 9

1.1.2.4 Structure and assembly 9

1.2 FIMBRIAE: A CATCH-ALL TERM FOR THIN PROKARYOTIC APPENDAGES...... 12

1.2.1 Fimbriae of diderm need to cross two membranes 12

1.2.1.1 Curli: unique fibers 12

¬ Discovery of functional amyloid pili 12

¬ Pili for adhesion and formation 13

¬ Structure and assembly 13

1.2.1.2 -usher pili: a diverse class of pili assembled by two conserved 15

¬ CU pili are widespread among diderm bacteria 15

¬ CU pili are powerful adhesins 15

¬ Structure and assembly 17

¬ Structural features of adhesion 17

1.2.1.3 Pseudopili assembled by the type II secretion system 18

¬ Discovery and distribution 18

¬ A secretion machine with high substrate specificity 18

¬ Structure and assembly 19

1.2.1.4 The newly discovered type V pili 20

¬ Distribution 20

¬ Multifunctional pili 20

¬ Structure and assembly 20

1.2.2 A monoderm-specific : the -dependent pilus 22

1.2.2.1 The first pili ever discovered in monoderm bacteria 22

1.2.2.2 Pili for adhesion and aggregation 22

1.2.2.3 Assembly of a -anchored pilus 23

1.2.3 Archaeal pili: an unexplored diversity 25

1.2.3.1 Hami: archaeal grappling hooks 25

1.2.3.2 Archaeal cannulae and bacterial spinae: intercellular communication fibers? 26

1.2.3.3 Mth60 fimbriae: species-specific multifunctional fimbriae 28

1.2.3.4 Archaella: the archaeal motility structure 28

¬ Discovery of an archaeal flagellum unrelated to the bacterial flagellum 28

¬ Motility and more… 29

¬ Structure and assembly 29

1.2.4 Pili found in all three types of : 2 different strategies to reach the surface 31

1.2.4.1 Pili assembled by type 4 secretion systems 31

1.2.4.2 Type IV pili: the all-in-one prokaryotic appendages 33

¬ Distribution and discovery 34

¬ TFP: multi-tasking champions 34

¬ Structure and assembly 34

2 TYPE FOUR FILAMENTS: MULTIFUNCTIONAL HOMOLOGOUS SYSTEMS 37

2.1 A CONSERVED BIOSYNTHESIS MACHINERY FORMED BY 3 COMPLEXES ...... 38

2.1.1 The inner membrane complex 38

2.1.1.1 The prepilin peptidase cleaves the leader peptide of the class III signal 38

2.1.1.2 The assembly platform initiates pilus assembly 39

2.1.1.3 The ATPases: powering pilus assembly and retraction 41

2.1.2 The outer membrane complex: crossing the outer membrane 44

2.1.3 The filament 45

2.1.3.1 Major pilins: major components of the pilus 46

¬ Structure 46

¬ Post-translational modifications 48

2.1.3.2 Minor (pseudo)pilins: a start/stop button? 48

2.2 ONE MACHINERY, MANY FUNCTIONS ...... 51

2.2.1 TFF mediate attachment through surface adhesion 51

2.2.2 TFF allow prokaryotes to move in various ways 53

2.2.3 TFF allow the formation of multicellular communities through aggregation 55

2.2.4 TFF allow selective secretion 56

2.2.5 TFF generate genetic diversity by providing transformation competence 58

2.2.6 TFF can be hijacked by phages 59

2.2.7 TFF can act as nanowires to allow extracellular respiration 60

2.2.8 TFF can enable surface sensing by mechanotransduction 61

3 TYPE IV PILI OF : A CASE STUDY 63

3.1 A HUMAN OBLIGATE ...... 63

3.1.1 The Neisseriaceae family: a diversity of commensal bacteria 63

3.1.2 Meningococcus has a high carriage rate 64

3.1.3 Meningococcal disease: a rare but deadly disease 64

3.2 OF NEISSERIA MENINGITIDIS...... 67

3.2.1 Hyperinvasive lineages: a few clonal complexes cause most disease cases 67

3.2.2 Multiple surface structures involved in infection 68

3.2.2.1 The protective capsule 68

3.2.2.2 The pro-inflammatory lipooligosaccharide 70

3.2.2.3 Metabolic adaptations 70

3.2.2.4 Several adhesins contribute to colonization of the human host 71

¬ Minor adhesins 71

¬ Opacity proteins 72

¬ Type IV pili: long distance adhesins 72 3.2.3 Infection models as a tool for in vivo identification of virulence factors 74

3.3 NEISSERIAL TYPE IV PILI: LINKING STRUCTURE AND FUNCTION ...... 75

3.3.1 Specificities of the meningococcal machinery 76

3.3.1.1 PilC-like proteins 76 3.3.1.2 Minor pilins 76

3.3.2 Pilus biogenesis 77 3.3.2.1 Role of the components of the piliation machinery 77

3.3.2.2 Pilus structure 78 3.3.3 TFP-dependent functions in Neisseria meningitidis 79

3.3.3.1 Pilus retraction enables 79 3.3.3.2 Pilus retraction enables natural competence 80

3.3.3.3 Pilus retraction enables phage infection 81

3.3.3.4 Pilus retraction enables the formation of fluid aggregates 82

3.3.3.5 Adhesion to human cells 83

¬ Adhesion to epithelial cells: two putative receptors 84

¬ Adhesion to endothelial cells: a CD 147-dependent adhesion? 85

¬ TFP: one adhesin or multiple adhesins? 85

3.3.3.6 Cellular response and signaling 86

OBJECTIVES: UNDERSTANDING HOW TFP MEDIATE SO MANY FUNCTIONS 89

RESULTS 91

1 SUBMITTED ARTICLE: MECHANISMS OF MENINGOCOCCAL TYPE IV PILI MULTIPLE FUNCTIONS REVEALED BY DEEP MUTATIONAL SCANNING. 93

2 ADDITIONAL RESULTS 129

2.1 CHARACTERIZING THE IMPORTANCE OF PILE IN COMPETENCE FOR TRANSFORMATION ...... 129

2.2 EXPLAINING THE PHENOTYPE OF THE “SHORT PILI” MUTANTS...... 130

2.2.1 Mutants with short pili have retractile pili 130

2.2.2 A role for minor pilins in pilus assembly 131

2.3 EXPLORING ADHESION TO HUMAN CELLS ...... 134

2.3.1 Deep mutational scanning shows a specific role of tyrosine residues in adhesion 134

2.3.2 Cholesterol-binding by TFP 136

2.3.3 Meningococcal TFP are electrically conductive 140

DISCUSSION 143

1 ADHERING UNDER FLOW, LEARNING FROM OTHER BACTERIA 144

2 REGULATION OF MENINGOCOCCAL PILIATION, A MATTER OF BISTABILITY? 147

3 CONSERVATION AMONG TFP-BEARING PROKARYOTES 150

3.1 PILIATION: HOMOLOGOUS STRUCTURES WITH DIFFERENT PROPERTIES...... 151

3.1.1 Folding PilE 151 3.1.2 Bistability and pilus length 151

3.2 COMPETENCE: ELECTROPOSITIVE GROOVES TO BIND DNA?...... 152

3.3 AGGREGATION THROUGH ELECTROSTATIC COMPLEMENTARITY ...... 153

3.4 ADHESION: A CONSERVED MECHANISM FOR TYPE IVA PILI? ...... 155

3.5 USING TFF TO UNDERSTAND HOW TFP MEDIATE THEIR FUNCTIONS ...... 157

CONCLUSION 159

SUPPLEMENTARY MATERIALS AND METHODS 161

REFERENCES 165

ACKNOWLEDGEMENTS 195

List of figures:

Figure 1: Summary of the introduction. 1 Figure 2: Early observation of 2 types of appendages: flagella and pili. 2 Venn diagram 4 Figure 3: Flagellum functions, structure and assembly. 6 Figure 4: Injectisome functions, structure and assembly. 10 Figure 5: Curli pili assembly and appearance. 14 Figure 6: Chaperone-usher pilus structure, functions and assembly. 16 Figure 7: Type 2 secretion system assembly. 19 Figure 8: Type V pilus structure and assembly. 21 Figure 9: Sortase-dependent pilus appearance and assembly. 24 Figure 10: Archaeal hami structure and functions. 26 Figure 11: The archaeal cannulae and the bacterial spinae share similar features. 27 Figure 12: Mth60 fimbriae functions and appearance. 28 Figure 13: Archaellum structure and functions. 30 Figure 14: Type IV secretion systems structure, function and assembly. 32 Figure 15: Type IV pili structure, functions and assembly. 35 Figure 16: Type four filaments share a conserved machinery. 37 Figure 17: Conservation of the class III signal peptide. 39 Figure 18: The inner membrane complex. 41 Figure 19: Structure of the ATPases PilF and PilT. 43 Figure 20: Structure/function relationship of secretins. 45 Figure 21: Conservation of type IV pilins structure. 47 Figure 22: Structure and functions of minor pseudopilins. 48 Figure 23: Type IV filaments are involved in a wide array of functions. 51 Figure 24: TFF-dependent adhesion to biotic and abiotic surfaces. 52 Figure 25: Diverse motility phenotypes can be achieved by TFF-bearing prokaryotes. 54 Figure 26: TFF-mediated aggregation. 56 Figure 27: Proposed mechanisms for protein secretion. 57 Figure 28: Proposed models for DNA uptake. 58 Figure 29: Phage binding to Type IV pili. 59 Figure 30: Type IV pili as nanowires. 61 Figure 31: Type IV pili as mechanosensors. 62 Figure 32: Diversity of the Neisseria genus. 63 Figure 33: Epidemiology of meningococcal disease. 65 Figure 34: Development of meningococcal disease. 67 Figure 35: envelope of Neisseria meningitidis. 69 Figure 36: Main adhesins used by N. meningitidis and corresponding cell receptors. 71 Figure 37: Mechanisms of protein variation in N. meningitidis. 74 Figure 38: Genetic organization of the meningococcal piliation machinery. 77 Figure 39: Structure of the meningococcal type IV pilus. 79 Figure 40: Twitching motility in N. meningitidis. 80 Figure 41: Natural competence for transformation. 81 Figure 42: Characterization of the meningococcal MDA phage 82 Figure 43: Aggregation promotes adhesion. 83 Figure 44: Putative adhesins and associated receptors of meningococcal type IV pili. 86 Figure 45: Cellular response to meningococcal adhesion. 87 Figure 46: Results summary. 91

Figure 47: Identification of two putative DNA-binding grooves at the surface of the pilus. 130 Figure 48: Characterization of the retractile properties of the α1N mutants. 132 Figure 49: Interactions between PilE and minor pilins are required to increase the proportion of piliated cells. 133 Figure 50: Introducing tyrosine residues in PilE enhances early adhesion to human cells 135 Figure 51: Identification of putative cholesterol binding sites in TFP. 137 Figure 52: TFP can interact with liposomes. 138 Figure 53: Hyper-adhering TFP have a higher affinity for cholesterol. 139 Figure 54: Conductive pili as adhesion facilitators. 140 Figure 55: Adhesion model for N. meningitidis 146 Figure 56: Speculative model for piliation bistability in N. meningitidis 150 Figure 57: Multiple sequence alignment of type IV pilins 151 Figure 58: Different proportions of cell display TFP depending on the expressed major pilin. 152 Figure 59: Putative DNA binding sites can be identified at the surface of type IVa pili. 153 Figure 60: Widespread role of electrostatic interactions in aggregation 155 Figure 61: Adhesion site of the PAK pilus from P. aeruginosa 157

List of abbreviations:

A. tumefasciens - Agrobacterium tumefasciens P. gingivalis -

C. diphteriae – Corynebacterium diphteriae S. Typhimurium - enterica serotype Typhimurium CMF – Cytoplasmic membrane-anchored filaments S. pyogenes - cAMP – cyclic AMP S. pneumoniae – c-di-GMP – cyclic dimeric GMP SCA – Statistical Coupling Analysis

CU- Chaperone Usher SEM – Scanning Electron Microscopy

DSC - Donor Strand Complementation SPI – Salmonella Pathogenicity Island

DUS - DNA Uptake Sequence T2SS - Type Two Secretion Systems

E. coli – T3SS - Type Three Secretion Systems

EM – Electron Microscopy T4SS – Type Four Secretion Systems

EPEC – EnteroPathogenic E. coli TEM – Transmission Electron Microscopy

HGT – Horizontal Transfer TLR - Toll-Like Receptor

IL – Interleukin TNF - Tumor Necrosis Factor

N. meningitidis – Neisseria meningitidis TFF - Type Four Filaments

N. gonorrhoeae – TFP - Type Four Pili

PAFr – Platelet Activating Factor Receptor UPEC – UroPathogenic E. coli

PAMP – Pathogen Associate Molecular Pattern V. cholerae – Vibrio cholerae

P. aeruginosa –

?

Figure 1: Summary of the introduction. We first describe the variety of structure function relationships found in prokaryotic appendages. We then focus on a specific subset of appendages: the type IV filaments and highlight the commonalities between the archaellum, the T2SS and the TFP. Finally, we present the role of TFP in the pathogenic bacterium Neisseria meningitidis and highlight the outstanding questions in the field. The question mark symbolizes the questions we addressed during this doctoral work. Cy: , IM: inner membrane, OM: outer membrane, Peri: , Pepti: peptidoglycan. Adapted from Berry and Pelicic 2015, Chang et al. 2016, Wang et al. 2017, Daum et al. 2017, Lopez-Castilla et al. 2017 and from figures by Arthur Charles-Orszag.

Introduction 1 Most prokaryotes are unicellular organisms. Because of the constraints associated with this lifestyle, they have evolved many ways to interact and cope with their environment. One strategy they have developed to expand their field of interaction is the display of external appendages. These appendages are used for several functions: motility, adhesion, formation of pluricellular communities. Historically, they have been separated in 2 classes: the flagella and the fimbriae (or pili).

Figure 2: Early observation of 2 types of appendages: flagella and pili. (A-F) Bacteria observed by light microscopy after Leifston . x1000. Adapted from Leifson (1951). (A) Alcaligenes sp. with lophotrichous flagellation. (B) Alcaligenes bronchisepticus. (C) P. aeruginosa with monotrichous flagellation. (D) Pseudomonas sp. with polar multitrichous flagellation. (E) Salmonella wichita. (F) Alcaligenes sp. showing predominantly bipolar lophotrichous flagellation. (G) Gold-palladium sputter- coated E. coli observed by SEM. Bacterium shows a long and thick flagellum (white arrow) and several shorter pili (e.g. black arrows). x20000. Adapted from Duguid (1954)

This classification is based on the ability to visualize flagella by combining simple light microscopy and specific staining methods as early as 18771–3 (Figure 2a-f). It was only 70 years later in 1949, that the first non-flagellar appendages were reported using electron microscopy (Figure 2g)4,5. Since their discovery, these non-flagellar appendages have been coined with several names, but, pili and fimbriae are the ones that were retained over time. We will use these two terms indistinctively through the manuscript.

The focus of this thesis is on the molecular mechanisms underlying the multiple functions of type IV pili in Neisseria. meningitidis. Before presenting the results we obtained during this doctoral work, we will introduce our biological question following the outline presented on

2 Figure 1. In order to illustrate the various mechanisms prokaryotes are using to interact with their environment, we will first present an overview of the assembly mechanisms, structures, functions and distributions of the appendages found in prokaryotes. The reader should know that this overview although quite detailed does not aim to be comprehensive, given the striking diversity of surface structures formed by prokaryotes. We will then focus on an evolutionarily- related class of appendages virtually present in all prokaryotes and grouped under the name Type Four Filaments (TFF)6 or Cytoplasmic Membrane-anchored Fibers (CMF)7 to provide a more extensive description of the system and its structure-function relationships. Finally, we will present the specificities of Type Four Pili (TFP) in the model organism Neisseria meningitidis that we used throughout our study.

1 A historical overview of the diversity of prokaryotic appendages

385 occurrences of the term flagellum can be found in the transcripts of the so-called Dover Intelligent Panda trial8. This trial opposed eleven parents of students to the Dover school district. This design: A pseudoscientific school district had required the teaching to their children of intelligent design as an alternative theory of the neo- creationism theory to evolution. One of the cornerstones of this pseudoscientific theory is that a structure movement aiming as complex as the bacterial flagellum could not have evolved through natural selection. This to prove the existence of God. It argument can be broken down to: “What good is it for bacteria to only have half a flagellum?”. states that some An excellent evolutionary perspective is provided by Pallen & Matzke on that topic in their 2006 biological features are too complex to 8 review . This is also a very tangible illustration of the complexity of the nanomachines that be explained by prokaryotes have evolved to assemble functional appendages. In the following sections, we try natural selection and result from a to illustrate this diversity and to keep an evolutionary logic in the presentation of these different designer’s work. appendages as presented in the Venn Diagram.

For historical reasons, we will first describe the Type Three Secretion Systems (T3SS) that comprise the bacterial flagellum and the related injectisome, we will then describe the thinner appendages grouped under the name pili. 1.1 The first observed appendages are assembled by type three secretion systems

Through the manuscript we will use the term flagellum specifically for the bacterial flagellum. We will refer to the archaeal rotary appendage (initially named archaeal flagellum) as the archaellum. Because of their unrelated evolutionary origins, the archaellum will be discussed later on in the section dedicated to archaeal pili (Section 1.2.3.4).

3

Venn diagram: Distribution of prokaryotic appendages. The appendages presented in the introduction are located in the diagram depending on the prokaryotes by which they are expressed. Appendages belonging to the type four filament family are in white and the filament using the type three secretion systems are in gray. Filaments in black do not belong to common families.

T5P: Type five pili, T4P: Type four pili, CU: Chaperone Usher.

1.1.1 Flagella: rotating bacterial filaments

Flagella are rotating micrometer-long helical filaments with diameters ranging from 20 to 60nm actuated by a rotary motor which is driven by ion-gradients across the bacterial cytoplasmic membrane9,10.

1.1.1.1 Diversity of flagellar systems

Flagella were first observed using light microscopy and specific staining techniques in the late XIXth century. At that time, the number, position and curvature of flagella was observed to be very variable depending on the bacterial species (Figure 2a-f). For example, Vibrio cholerae and Pseudomonas aeruginosa express polar flagella while Escherichia coli and Salmonella enterica express lateral (or peritrichous) flagella. This diversity of flagella is particularly well- illustrated in marine Vibrio (e.g. Vibrio parahaemolyticus) which can express both polar and peritrichous flagella upon induction11. Furthermore, the motors of these flagella are also powered by different energy sources: a proton-motive force for the polar flagella and a sodium- motive force for the peritrichous ones12. The subcellular location of these appendages can also be variable. While most flagella are located in the extracellular milieu, some are covered in a membrane sheath13. The most extreme example of this diversity is illustrated by the bacterial class of Spirochetes which possess endoflagella: polar flagella residing in the periplasm and connecting the two poles of the bacteria. These periplasmic flagella give the Spirochetes their typical spiral shape and were one of the first observations of a cytoskeletal structure in prokaryotes14,15 (Figure3b).

4 1.1.1.2 Functions: motility and more

Very early on, flagella have been associated with . Bacteria use their flagella for at least two distinct types of motility: swimming in liquids thanks to polar flagella and, swarming over solid substrates thanks to their peritrichous-induced flagella12. Bacteria are propelled at speeds ranging from 25 to 35µm/s for E. coli16 and can attain speeds of 160µm/s for Bdellovibrio bacteriovorus17. Possessing motility-mediating appendages provides quite obvious advantages (moving away from threats and moving towards nutrients) and the evolutionary benefit they provide is further supported by their widespread distribution in bacterial species. Regulation of motility is coupled to environmental sensing to provide directional motility. This is achieved through sensing of chemicals (chemotaxis) but also through sensing of other stimuli: light for , temperature for thermotaxis and touch for thigmotaxis. Chemotaxis has been particularly well-characterized in E. coli and it has been shown that an alternation between directional swimming in the presence of attractants and a random movement named tumbling in the presence of repellents ensured efficient net displacements10.

5

Figure 3: Flagellum functions, structure and assembly. (A) Illustration of the diverse roles of flagella in niche colonization. can use them to adhere and reversibly form , to approach cells in order to secrete effectors to promote host cell killing or invasion. Invasion can also be facilitated by mere movement. (B) Electron micrograph of the spirochete Treponema zuelzerae showing typical arrangement of intracellular flagella (top). Sketch of the observed structures (bottom). (C) Simplified representation of the structure of the bacterial flagellum showing the three subassemblies that make up the flagellum. OM: outer membrane; PG: peptidoglycan layer; IN: inner membrane. (D) Cross-sections of the Salmonella typhimurium and Campylobacter jejuni flagella models showing respectively a 11 and 7 protofilaments assembly. Red coloring highlights the region surrounding the lumen of the flagellum. (E) Screenshots of a video illustrating the successive steps of flagellum assembly. The T3SS is first assembled (steps A to C) and help secrete flagellar components distally under the control of the pentameric distal cap protein (D-O). Adapted from Galkin et al. 2008, Erhardt et al. 2010, Chaban et al. 2015, Holt 1978.

6 In , flagella have been shown to have other key functional roles in host Biofilm: Aggregate of colonization than mere motility (Figure 3a). They have notably been implicated in host surface microorganisms mechanosensing. This sensing is important to switch developmental state and its importance embedded in a slimy extracellular has been described for biofilm formation in several species (P. aeruginosa, V. cholerae, V. polymeric parahaemolyticus…)18, and more recently for synthesis of a surface adhesin in Caulobacter substance made of secreted DNA, 19 crescentus to mediate surface anchoring . proteins and Interestingly, surface adhesion can be directly mediated by the flagellar filament. The flagellin monomers that make up the filament in P. aeruginosa bind to mucins, which promotes Mucins: epithelial colonization20. Using the flagella-dedicated T3SS, pathogens can also secrete glycosylated proteins making bacterial effectors in their environment thus bypassing the need for an injectisome10 (cf next up part of the protective viscous section). Using flagellar motility, bacteria can even penetrate tissues by invading cell layers mucus that covers through disruption of intercellular junctions. This holds particularly true for the endoflagellated animal epithelial cells. Spirochetes21,22. Finally, flagella can help facilitate bacterial dispersion by promoting bacterial . S. typhimurium actively translocates flagellin into the of its host cells, which induces pyroptosis of the host cell, thus leading to a local inflammation and the Pyroptosis: Programmed cell recruitment of new macrophages. These macrophages are infected by S. typhimurium and will death associated disseminate the infection23,24. The many functions exerted by flagella highlight both the with the release of highly importance and the complexity of these nanomachines. inflammatory cytokines upon cell 1.1.1.3 Structure and assembly burst

Recent structural insights provide support to the understanding of the various functions mediated by the flagellum. The structure of the flagellum can be broken down into three main components: the which makes up the engine of the flagella, the rotating filament and the hook that joints the basal body and the filament. These three components make up about 30 distinct proteins which will not be detailed here.

Because the assembly of all these proteins has a very high metabolic cost, it is a tightly regulated and ordered process. Details on the genetic regulation of this assembly can be found in this review25. The architecture and assembly steps of the flagellum are illustrated in Figure3c and e. To begin with, basal body membrane components and the T3SS membrane components are exported by the Sec general secretory pathway26 . Namely, the inner membrane MS-ring starts assembling at the cytoplasmic membrane and the C-ring in the cytoplasm. This supports and co-occurs with the assembly of the T3SS27,28 which in turn enables the assembly of the peptidoglycan-anchored P-ring and of the L-ring at the outer membrane. Flagella of mono- derm bacteria lack Assembly of these axial components allows proton-motive dependent unfolding and secretion both the L-ring and of the components of the periplasm-spanning rod and the hook by the T3SS29. Once these the P-ring. elements are assembled, the flagellin subunits are exported through a recently-described T3SS-dependent injection-diffusion mechanism30. Flagellins diffuse through the 2nm-diameter hollow structure spanning the bacterial membranes and are folded and incorporated at the tip

7 through the action of a capping-chaperone31. This gives rise to a micrometer-long filament containing tens of thousands of flagellin subunits. There is a diversity in the quaternary structures of these filaments as observed by cryo-EM of purified filaments. For example, the flagellum of S. typhimurium is composed of eleven protofilaments while the one from Campylobacter jejuni only has seven protofilaments32 (Figure 3d). Rotation of the filament is enabled through the action of transmembrane stators , that transduce the ion-motive force into a rotation force (torque) to the cytoplasmic C-ring. This torque is eventually transduced to the connected rod, hook and filament thus generating motion33. To generate thrust, the filament has to adopt a supercoiled state. Specific point mutations in the flagellin were found to result in straight but still rotating filaments unable to propel the bacteria34. The molecular mechanism of flagellum supercoiling and transition between supercoiled states is still under investigation but recent technical progress in cryo-EM are bringing us closer to a comprehensive understanding of the flagellum structure-function relationship35,36.

1.1.2 The injectisome: needles assembled by the type three secretion system

The injectisome (or non-flagellar T3SS) is found in many symbiotic diderm bacterial species (pathogenic but also commensal ones). This complex is composed of more than tenty relatively well-conserved proteins. Upon contact with target cells, the injectisome mediates the translocation of so-called effector bacterial proteins in the cytoplasm of these target cells37. This contact-dependent phenomenon was first described in Yersinia pestis in 199438 and the injectisome was first visualized four years later in 199839.

1.1.2.1 Relationships to the flagellum

In spite of the profound differences in both structure and function of the injectisome and the flagellum, the evolutionary-relatedness of their core T3SS export proteins has long been recognized40. Yet, there has been considerable debate on the evolutionary scenario that led to the evolution of these two nanomachines. While an absolute consensus has not yet been reached, the prevalent scenario states that the injectisomes evolved from flagellum ancestors. This is supported by several pieces of evidence including: the relatively high conservation of encoding for the different injectisomes, the fact that unlike flagellar genes they are often found on mobile genetic elements and that their distribution does not match the phylogeny of the bacterial species bearing them26,41.

1.1.2.2 Diversity of injectisomes

Eight injectisome families have been identified so far. As mentioned earlier, the relatedness of these families does not relate to the phylogeny of the bacterial species expressing them. Yet, the family of injectisomes expressed by a bacterium does correlate with the cell-type it will interact with. This diversity is well-illustrated at the structural level. Indeed, the distal structures

8 allowing the bacteria to reach the plasma membrane of the target cells can be of 3 different types. The shortest is simply made of a stiff hollow needle with a 7nm outer-diameter, a 2.5nm inner diameter and a 60 nm-length42 (Figure 4c). This short needle can also be extended by a filament of approximately 600nm with a 12nm diameter which is thought to enable animal pathogens to reach the by going through the mucus layer43 (Figure 4d). Finally, in the T3SS of plant pathogens, the needle with a 7nm diameter can be extended by as many as a few micrometers forming the Hrp-pilus. This so-called pilus is thought to be needed to go SPI : across the plant cell wall37 (Figure 4d). A single bacterium can harbor more than one Salmonella Pathogenicity injectisome family (Burkholderia pseudomallei currently holds the record with three different Island – Region of functional T3SS44. Having two different injectisomes is thought to provide the bacterium an the genomes of Salmonella spp. advantage by allowing it to interact with and invade more niches. This has notably been shown acquired by HGT for Salmonella Typhimurium which uses the SPI-1 T3SS to invade the intestinal epithelium and and encoding for virulence traits. SPI-2 T3SS at later stages for deep tissue invasion45.

1.1.2.3 A translocation machine

Another source of variability for the injectisome is the nature of the effectors translocated through the needle. Over 100 effectors have been described and many remain to be identified. Identification of such effectors is hampered by the limited understanding we have of T3SS export signals and the key role of dedicated chaperones to enable the export of specific effectors26.

While the function of the injectisome has so far been limited to the translocation of unfolded effector protein into a target cell, the functions of these effectors are manyfold (Figure 4a).

Effectors have been involved in host colonization through: manipulation of the transcription profiles of target cells to facilitate invasion through differentiation of preferred cell types or on the contrary to maintain their niche, immune suppression to enable bacterial proliferation both in commensals and pathogens and nutrient acquisition by stimulating nutrient export by host cells46. In order to secrete these effectors and to hijack host cells, the injectisome first needs to be assembled.

1.1.2.4 Structure and assembly

The injectisome shares many structural features with the previously-described flagellum. It is composed of: a cytoplasmic sorting platform47 underlying the inner-membrane associated export apparatus and of the needle complex (NC). The NC can be easily purified and imaged. It comprises a pair of rings spanning the inner membrane connected through a neck to a pair of rings spanning the outer membrane. Finally, this envelope-spanning basal body is connected to the needle through an inner-rod (Figure 4b,c)39,48. The NC is hollow with a 20nm diameter that is used for effector secretion49. This whole structure has been elucidated with exquisite details in a recent high-throughput in situ cryo-electron tomography of the Salmonella T3SS42.

9 An interesting structural feature of the injectisome when compared to the flagellum is the acquisition of a secretin (the structure comprising the neck and the two rings spanning the outer membrane). The secretin family is a specialized family of integral outer membrane protein also found in the Type IV pilus and Type II secretion systems to enable (pseudo)pilus extrusion. Phylogenetic evidence shows that the ancestral injectisome acquired secretins from both systems on at least three different occasions during evolution which have led to functional injectisomes41. Similarly to what was described for the flagella, there is a sequential assembly of the components of the injectisome50 (Figure 4e). The Sec system initiates export and membrane insertion of the basal body and the inner membrane export system. The export apparatus and secretins are first assembled followed by the inner rings. The outer and inner rings are then linked which triggers the recruitment of the cytoplasmic sorting platform. This will trigger the secretion process, starting by the inner rod and needle components. Just like for the hook of the flagellum, the length of the needle is accurately controlled through a mechanism that remains unclear. As for the flagellum, polymerization also occurs at the tip of the needle. Finally, upon cell-contact the translocation pore is formed through the interaction of three translocator proteins at the tip of the needle: a scaffolding hydrophilic tip protein51 and two hydrophobic translocases with the pore-forming activity52. By a switch in the selectivity of the cytoplasmic platform, these proteins get assembled at the tip of the needle and cap it. The tip protein can also assemble as a longer filament polymerizing at the distal end of the needle in animal pathogens (e.g. enteropathogenic E. coli)43. In as little as 10 seconds, translocation of the effector proteins was initiated and could last for a dozen of minutes53.

Figure 4: Injectisome functions, structure and assembly. (A) Different functions of T3SS in the interaction of bacteria with eukaryotic cells. (B) 3D structure of the needle complex of S. typhimurium based on a cryo-EM map showing 5 distinct rings. OR1: outer ring 1; OR2: outer ring 2; IR1: inner ring 1; IR2: inner ring 2. (C) Average of the in situ cryo-ET structure of a full injectisome. OM: Outer Membrane; PG: Proteoglycan; IM: Inner membrane (D) Representative examples of the filaments assembled by T3SS. Upper panels: scanning electron micrographs showing the long filaments extending the needle complex at the surface of EPEC during infection of HEL cells. Lower left panel: electron micrograph of the plant pathogen P. syringae displaying a long Hrp-pilus stained by immunogold. Scale bar: 100nm. Lower right panel: shows negatively stained S. typhimurium with short needle complexes embedded in its membrane. (E) Model illustrating the sequential assembly of the needle complex. Adapted from Galàn et al. 2018, Jin et al. 2001, Knutton et al. 1998 and Puhar et al. 2014.

10 Legend on previous page

11 The recent possibility to visualize prokaryotic nanomachines in situ at extremely high resolution offers considerable insights in their structure/function relationship. Extending these high- throughput studies to a multiplicity of organism will undoubtedly benefit the overall understanding of these complex nanomachines54. Here, we have shown the functional divergence that can arise from relatively conserved structures. Indeed, type three secretion systems offer a very good example of how phylogenetically-related bacterial appendages despite conserved core components have evolved radically different functions (motility and protein translocation) through the acquisition of new modular components. In the following sections, we will continue to explore this surprising diversity and will describe the many ways prokaryotes have evolved to assemble multi-functional appendages. 1.2 Fimbriae: a catch-all term for thin prokaryotic appendages

Fimbriae are such thin structures that the first time they were visualized at the surface of cells by electron microscopy, an argument immediately emerged to determine whether they were real surface structures or preparation artifacts. Indeed, one of the two seminal publications reporting the observation of “thin filaments” stated that these structures were preparation artifacts due to drying of the capsule or of “slime substance”4 whereas the second publication indicated that they were true filamentous appendages present at the surface of bacteria5. Within a few years, the existence of pili was confirmed: they were shown to be associated with a hemagglutinin activity, a change in bacterial electrophoretic mobility and could be easily purified from the bacterial surface through “vigorous agitation”55,56. Since then, many different types of pili have been discovered and the list is still growing57. While pili were first identified in diderm bacteria, they have since then been observed both in monoderm bacteria and . Their various biogenesis modes, structures and functions will be discussed in the following sections based on their distribution in these three prokaryotic classes.

1.2.1 Fimbriae of diderm bacteria need to cross two membranes

Appendages expressed in diderm bacteria need to traverse two membranes before reaching the extracellular milieu. Traversal of the inner membrane is most often mediated by the general secretion Sec-pathway, but bacteria have often evolved dedicated outer-membrane pores to allow passage or assembly of appendages through this second membrane.

1.2.1.1 Curli: unique amyloid fibers ¬ Discovery of functional amyloid pili Curli fibers were first described by Olsen et al. in 1989 in E. coli for their role in fibronectin : binding58. These pili are produced by Enterobacteriaceae and have been particularly well- protein aggregates that characterized in E. coli and Salmonella spp. They are micrometric, curved fibers with a 4 to 6nm polymerize in diameter forming a dense meshwork surrounding the bacteria (Figure 5a). An outstanding a typical fibrillar 59 cross-ß sheet feature of these pili is that they are functional amyloids . There is growing evidence for the structure involvement of amyloids in (human) pathologic conditions. Prions, a family of amyloids with

12 self-perpetuating infectious capabilities have been implicated in several neurodegenerative diseases and their aggregation properties are still poorly understood60. The role of amyloids in human pathologies has made curli a particularly interesting experimental model to understand the fundamental processes involved in the formation of amyloid fibrils and maybe discover inhibitors of amyloid aggregation61.

¬ Pili for adhesion and biofilm formation It was only a decade ago that functional amyloids were found to be quite widespread. Since then, they have even been found in humans where they can play role as diverse as promoting the biosynthesis of melanin or initiating the activation of the coagulation factor FXII 62,63.

Curli were initially shown to play an important role in through fibronectin binding58 and have since then also been demonstrated to bind many other extracellular host proteins64. They are also important for bacterial internalization65 and one of their central role is to participate in biofilm formation66,67. In E. coli, curli are the major protein component of the biofilms and make up 85% of the extracellular material68. The dense extracellular webs formed by curli protect bacterial colonies and promote niche persistence (Figure 5a). Finally, curli have been found to be a pathogen-associated molecular pattern (PAMP) recognized by TLR1 and TLR2, thereby inducing a strong immune response of the host69–71.

¬ Structure and assembly How the structure mediating all these functions is assembled has been relatively well described using the E. coli model. The curli assembly system is relatively simple and is made of only seven curli specific genes (csgA-G). CsgD is a regulatory component in charge of controlling the expression of the csg genes and more generally genes involved in biofilm formation72. All the other Csg components are first secreted through the inner membrane via the Sec-system. The assembly mechanism is summarized in Figure 5b.

The curli filament is a heteropolymeric assembly of CsgA:CsgB at a 1:20 ratio with a typical amyloid cross-ß spine: a vertical stacking of ß sheets perpendicular to the filament axis, providing the structure a very high stability73. The field is still lacking an accurate native structure of the fiber. Yet, it has been shown that CsgA fibers can spontaneously assemble in vitro. Fiber formation is accelerated by the addition of CsgB (which acts as a nucleator) or that of pre- formed curli “seeds”74,75. In the bacterium, the nucleating action of CsgB is dependent on the presence of CsgF which interacts with CsgB and positions it near the secretion pore CsgG through an uncharacterized interaction76 . The structure of the nonameric secretion pore CsgG has been elucidated and suggests a diffusion-mediated secretion of its CsgA, CsgB and CsgF substrates. This diffusion-based secretion is facilitated by the formation of a CsgG:CsgE complex where CsgE plays the role of a plug by closing the periplasmic access of the CsgG pore77. This local diffusion of CsgA in the extracellular milieu, results in surface-assembly of the

13 curli. This growth has recently been shown to be polar, which suggests a proximal addition of CsgA to the fiber78.

Figure 5: Curli pili assembly and appearance. (A) Curli fibers observed by EM. The two panels on the left show by freeze-fracture EM the dense meshwork formed by curli in E. coli biofilms. Third picture shows an E. coli cell covered with curli by transmission electron microscopy. Rightmost picture shows curli polymerized in vitro using purified CsgA. (B) Model summarizing the formation of curli fibers by translocation of the CsgA curlin through the CsgG outer membrane channel and by incorporation at the tip or the base of the growing pilus. Accumulation of toxic fibre aggregates is prevented by CsgC and proteolysis. IM: inner membrane; OM: outer membrane; PG: peptidoglycan.

Adapted from Van Gerven et al. (2015) and Hospenthal et al. (2017).

Finally, the system has evolved a chaperone, CsgC, that inhibits CsgA amyloid formation in the peripslam79. CsgC binds to fibrils but not to monomeric CsgA and inhibits fibril elongation, presumably though a capping mechanism78. Interestingly, this CsgC chaperone can also inhibit aggregation of the a-synuclein amyloid involved in Parkinson’s disease79.

14 Curli fibers are a good example of bacteria taking advantage of self-assembling proteins to colonize new environment with a relatively simple assembly machinery. New structural insights should greatly benefit the understanding of the mechanism of curli assembly as well as the structural bases behind the functions of curli.

1.2.1.2 Chaperone-usher pili: a diverse class of pili assembled by two conserved proteins

Another example of fimbriae found in diderm bacteria is that of the chaperone-usher (CU) pilus family. This family likely constitutes the most extensively studied pilus family with a very well- characterized assembly mechanism. This understanding also resides in an even more reduced assembly machinery. Indeed, the polymerization of pilin subunits is mediated through the action of only two proteins: a chaperone that helps to fold and polymerize pilins80 and the pore- Pilin: 81,82 Protein subunits forming usher which supports the growth of the pilus through the outer membrane . that form the pilus Major pilin form ¬ CU pili are widespread among diderm bacteria most of the pilus and are expressed Chaperone usher pili are widely distributed in diderm bacteria and are found at the surface of at much higher many bacterial pathogens (Yersinia, Salmonella, Shigella spp. …). They have very diverse levels than so- called minor morphologies and can take the appearance of both a 2 micron-long rod with a 7nm diameter pilins. and a thin flexible fibrillum with a 2nm diameter or just one of these two components83 (Figure 6a-c). Because of this intrinsic diversity, CU pili have been given a variety of names according to: their appearance (UPEC’s type I pilus and ’s type III fimbriae), the species in which they are found (Hib pilus of Type B), or even the pathology they are associated to (P(yelonephritis-associated) pilus). To facilitate, the naming of the CU pili, a classification according to 6 phylogenetic clades has recently been proposed84. Most of the literature on CU pilus originates from 2 types of pili found at the surface of UPEC: the type I and P pili. The following sections will focus on the structure/function relationships of these two pili.

¬ CU pili are powerful adhesins Type I and P pilus are crucial for the virulence of uropathogenic E. coli (UPEC). While type I pilus are required for invasion of the bladder85,86, type P pilus are important for kidney invasion87. The importance of these pili is due to their ability to mediate adhesion to the surface of the epithelium. This adhesion is mediated by a dedicated adhesin found at the distal tip of the CU pilus88–90. CU pili are also important for cell invasion and the formation of long-lasting biofilms91,92(Figure 6d-e). The importance of CU pili has also been demonstrated in other species such as P. aeruginosa and Klebsiella pneumoniae93,94. Whether the role of CU in biofilm formation is simply due to substrate adhesion is unclear at the moment.

15 Figure 6: Chaperone-usher pilus structure, functions and assembly. (A) Electron micrographs of Klebsiella type 54 covered with type III fimbriae. x 40000. (B-C) Electron micrographs of E. coli. (B) Piliated virulent strain CN1016 displays type I pili. (C)Avirulent strain HB10 lacks visible pili. (D-E) Scanning electron micrographs of E. coli MS2027 showing the role of chaperone usher pili in biofilm formation on polystyrene. Left: piliated bacteria. Right: unpiliated bacteria. (F) Freeze-etch microscopy of purified P pili evidences the tip fibrillum. (G) Electron micrographs of rotary shadowed sheared P pili. The arrow points to the partially uncoiled pilus rod. Tip fibrillum is also evident at the tip of these pili. (H) Top panel illustrates the structure of the type I and the P pilus. Both pili possess very similar characteristics and polymerized through the coordinated action of a chaperone and the usher pore. The mechanism of polymerization is illustrated in the bottom panel and described in the main text and relies on the coordinated action of the periplasmic chaperone (in yellow) and of the N and C-terminal domains of the outer-membrane usher. Adapted from Bullit et al. 1995, Kuehn et al. 1992, Duguid 1959, Ong et al. 2008, Connell et al. 1996 and Hospenthal et al. 2017.

16 ¬ Structure and assembly Type I and P pili have a “typical“ CU pilus structure that includes a long basal rod and finishes with a much shorter and thinner flexible fibrillum95 (Figure 6f-g). The adhesins (FimG or PapG) are located at the tip of theisfibrillum96. The fibrillum itself is composed of 3-4 different minor pilins (FimG/F/H or PapG/F/E/K) assembled in an elongated state while the rod connected to the fibrillum is composed of a single protein: the major pilin (FimA or PapA) which adopts a compact quaternary structure97(Figure 6h) .

The assembly process is summarized on Figure 6h. In brief: each pilin in the growing pilus stabilizes the next one by donating its N-terminal ß-strand and thereby completing the next pilin’s fold. All the CU pilins share a very similar fold: an incomplete immunoglobulin fold, IgG domain: similar to the canonical immunoglobulin fold at the exception that it lacks the 7th ß-sheet Protein fold arran- ged in two layers required for proper folding and possesses an extra N-terminal ß-sheet98. This results in the of 7 to 9 ß-strands exposure of an unstable hydrophobic groove with 5 hydrophobic pockets at its surface (P1 to with a Greek key topology. P5)99,100. Pilins are exported in the periplasm through the Sec pathway where they will be stabilized by a dedicated chaperone101. This stabilization is mediated through a process named Donor Strand Complementation (DSC) whereby the chaperone fills the hydrophobic groove (pockets P1 to P4) of the pilin by yielding one of its own ß-strands. This chaperone-pilin complex is captured by the outer membrane-inserted usher98,102. The usher has a N-terminal periplasmic domain linked to a domain that plugs the pore and two C-terminal periplasmic domains. Initially, the chaperone-pilin complex is captured by the N-terminus of the usher103,104. Of all pilin-chaperone complexes, the usher has the highest affinity for the adhesin-chaperone complex which explains why the adhesin is the first pilin to be polymerized at the usher105,106. This complex is spontaneously transferred to the C-terminal periplasmic domain of the usher. This causes a lowering of the plug domain and frees the pore. The usher binds another pilin- chaperone complex through its N-ter. This positions the complex so that polymerization gets initiated through Donor Strand Exchange (DSE)103,104. DSE is the process whereby the N- terminal ß-strand of an incoming pilin replaces the ß-strand of the chaperone through a zip-in zip-out process107. DSE is initiated at the free P5 pocket and then proceeds to P4, P3,P2 until P1108. Upon completion of this process, the chaperone is released in the periplasm and the transfer of the second chaperone-pilin complex from the N-terminus of the usher to the C- terminus is thought to promote the translocation of the elongating pilus through the usher108,109. DSE exchange then proceeds sequentially to form the pilus. As there is no ATP in the periplasm, the elongation of the CU pilus is thought to be solely driven by a combination of Brownian motion, usher translocation and steric hindrance of the wide quaternary rod structure to prevent sliding back in the periplasm110.

¬ Structural features of adhesion Unlike the other pilins, the adhesins located at the tip of the pilus do not have a N-terminal Lectin : Carbohydrate ß-strand. Instead, they possess a lectin N-terminal domain that provides them with their binding protein

17 adhesion function and prevents their insertion in the growing pilus through DSE. Adhesin change conformation upon assembly and switch from a linear conformation to an angled one, which is also thought to participate in preventing the pilus to slip back into the usher106. Another atypical conformational change has been observed for the type I pilus adhesin (FimH). Indeed, the affinity of the adhesin to its receptor is increased under flow through a catch bond mechanism111,112. This increased adhesion under flow is thought to allow bacteria to maintain attachment under urine flow and to facilitate dissemination in the absence of shear. Interestingly bacteria can bind cells when they only express the fibrillum of the CU pilus and not the rod96. Yet, this does not hold true under flow and is explained by the capacity of the rod to uncoil under flow113–116 (Figure 6g). This uncoiling is thought to absorb most of the tensile force that would otherwise be applied at the level of the adhesin and enables bacteria to maintain adhesion under flow117–120. Hence, there is a structural cooperativity between the rod and the adhesin to maintain adhesion under flow. The structural basis behind the flexibility of the rod has recently been elucidated97,121–123. The rod has hollow lumen and forms extensive inter-monomer interactions. There are two families of interactions: the strong hydrophobic interactions involved in DSE and the weak hydrophilic non-essential interactions. The last ones can be broken and would thus enable rod elongation. Some of these hypotheses were recently tested, and suggest through point mutations of the amino acids involved in uncoiling that, a too flexible rod leads to a decrease in virulence123.

The CU pili are prototypical examples of how different proteins with homologous quaternary structures have cooperatively evolved to mediate a similar function: adhesion under flow.

1.2.1.3 Pseudopili assembled by the type II secretion system ¬ Discovery and distribution Another diderm-specific type of appendage was discovered through its function: protein secretion. The observation that the starch debranching enzyme pullulanase of could not be addressed to the outer face of the outer membrane when expressed in E. coli K-12 led to the discovery of the type II secretion system (T2SS) in 1987124. This secretion system has since then been observed in many proteobacteria (e.g. P. aeruginosa, V. cholerae, E. coli)125–127. Soon after the discovery of the T2SS, the similarity between the proteins of this secretion system and those of the type IV pilus was established128,129. But, in experimental conditions, as opposed to type IV pili, the T2SS is not involved in the formation of a pilus. Instead, the T2SS is believed to form short periplasmic fibers called pseudopili to drive protein secretion130. Yet, when overexpressing the genes encoding the T2SS, pili can be found at the cell surface131–133 (Figure 7a). Studies of this secretion system thus shed light on pilus assembly mechanisms and justifies their presence in this section dedicated to pili.

¬ A secretion machine with high substrate specificity The T2SS has the ability to recognize specific periplasmic folded proteins134,135 and to translocate them through the outer membrane. A T2SS can secrete a single substrate (e.g.

18 pullulanase in K. pneumoniae) or more than twenty different proteins (e.g. in V. cholerae) depending on the system136. Over a hundred substrates have been described so far for bacterial T2SS. Just like for the T3SS, some bacterial strains (e.g. P. aeruginosa) express up to three distinct T2SS137–139. The functions of the effector secreted by various T2SS have recently been reviewed in great detail139. In brief, these effectors are important for pathogenesis and can induce host cell death140, biofilm formation141,142, surface adherence143, immune dampening144,145 but also for environmental survival through nutrient assimilation for example146,147.

¬ Structure and assembly A unifying nomenclature was introduced for the machinery of the T2SS. The machinery of the General Secretory Pathway (Gsp) is usually composed of 11 to 16 proteins (GspA to GspP and GspS)148,149 (Figure 7b).

Figure 7: Type 2 secretion system assembly. (A) Electron micrograph of an E. coli K-12 strain engineered to overexpress the pullulanase secreton from Klebsiella oxytoca. Pili primarily composed of the pseudopilin PulG are found at the surface of the bacterium. (B) Model of the assembly of the T2SS. The minor pseudopilins complex recruits the polytopic inner membrane platform and the alignment complex allows to link this assembly platform to the secretin. This leads to the recruitment of the polymerization ATPase and the major pseudopilins is assembled as a short pseudopilus. This results in protein secretion through a mechanism that remains to be elucidated. Adapted from Sauvonnet et al. (2000) and Thomassin et al. (2017).

Before interacting with the T2SS, effector proteins are translocated through the inner membrane via the Tat or the Sec pathway150,151 . The T2SS can be broken down in three parts: an assembly platform, an outer membrane translocation protein (the secretin) and the pseudopilus. The assembly platform is associated to the inner membrane and formed of an initial GspC,F,L,M complex. These components recruit the cytoplasmic ATPase (GspE) which will hydrolyze ATP to allow pseudopilus assembly152,153 and link the whole system to the secretin (GspD). Prior to pilus assembly, the inner membrane-inserted major pseudopilin (GspG) and minor pseudopilins (GspH-K) first need to be cleaved by a dedicated prepilin peptidase (GspO) that recognizes their N-terminal class III signal sequence154,155. Upon cleavage, pilus assembly can be initiated. The four minor pilins are necessary for efficient pseudopilus formation and are

19 thought to form a tip complex priming pilus assembly156. Pilus is then elongated through interactions with the assembly platform and forms a helical pseudopilus hold together by hydrophobic interaction between the N-terminus of the major pilins157–159. Finally, pseudopilus growth allows the secretion of periplasmic protein through the secretin (GspD) that forms a pore in the outer membrane160,161 and is connected to the assembly platform through an interaction with GspC162–165 (Figure7). While this system has been shown to have very specific substrate secretion, the way it recognizes its protein substrates is still poorly understood and could involve other factors than the T2SS per se, such as inner membrane-anchored chaperones and lipids of the inner membrane itself149.

1.2.1.4 The newly discovered type V pili ¬ Distribution Type V pili is the latest class of pili that has been described in diderm bacteria. Although they were first observed in Porphyromonas gingivalis in 1984166, their structure and assembly mechanism have only started to be elucidated in a study published in 201657. This cornerstone study revealed pilin folds and assembly mechanisms different from all the ones that had been described until then (these pilins are lipidated, have a unique structure and are particularly long). While these pili were initially observed in P. gingivalis, more than 1800 members of this pilin family have been detected and are mostly distributed in the Bacteroidetes phylum and ubiquitously in the human gut microbiome57. These pili exist as two morphologically different types at the surface of P. gingivalis: long numerous ones of approximately 1 micron and short ones of ~100nm (Figure 8b,c). The different major pilins composing these two types of pili share similar Type V pilus assembly characteristics57,167.

¬ Multifunctional pili Type V pili of P. gingivalis are important virulence factors74,168. They are involved in surface adhesion to various host proteins169,170, cell internalization171, avoidance of immunity172 and also auto-aggregation57,173.

¬ Structure and assembly Type V pilins are exported as precursors with an unusually long leader peptide174,175. They are initially secreted through the Sec pathway before getting lipidated at a conserved cysteine residue. The lipidation signal sequence that precedes the lipidation site is then cleaved in the periplasm175 (Figure8a,d). Pilins then cross the outer membrane through an unknown mechanism. Their N-terminal end (including the lipidated cysteine) will be cleaved again, this will initiate pilus polymerization and growth57,176. There are at least three different pilins involved in pilus assembly: the tip minor pilin, the major pilin that forms most of the pilus173,177 and the anchoring minor pilin which is also involved in pilus length regulation178,179.

20

Figure 8: Type V pilus structure and assembly. (A) Overview of the current knowledge of the assembly system of the type V pilus. (B-C) Electron micrographs of negatively stained Porphyromonas gingivalis displaying surface type V pili. (B) Strain MPG1 with thin and short fimbriae. (C) Strain ATCC 33277 with numerous long fimbriae. (D) Illustration of the maturation steps of the type V pilin sequence depending on its location in the bacterium. (E) Representation of the structural changes at the pilin level allowing pilus polymerization. Adapted from Hamada et al. (1996) and Hospenthal et al. (2017).

Crystal structures of over 20 members of this pilin family have revealed that they all share a Greek-key β sandwich fold prior to cleavage. Outer membrane cleavage of the tip pilin and major pilins leads to the removal of their N-terminal β-strand revealing a destabilizing hydrophobic groove. Analogous to what happens in the chaperone usher pilus system, this groove is thought to be stabilized through a donor strand exchange mediated by the C- terminal β-strand of the next pilin subunit57 (Figure 8e). The base minor pilin lacks a N-terminal cleavage sequence and therefore remains anchored to the outer membrane thanks to its lipidated cysteine. The tip minor pilin on the contrary does not have an available C-terminal β- strand and cannot participate in DSE, ensuring its distal positioning57,180. An alternative mechanism of DSE involving the most N-terminal β-strands was also proposed recently 180.

21 Further structural work will be needed to dissociate between these 2 two hypotheses. Just like the adhesins of the CU pilus, some tip pilins of the type V pilus display a C-terminal lectin domain which is likely responsible for the adhesive properties of these pili. Other tip pilins do not have such domains and their adhesive properties cannot be explained at this point57. The auto-aggregative properties of the type V pili have not been investigated structurally, but, mutants lacking the anchor pilin, had longer pili and were shown to be hyperaggregative thus suggesting a role for pilus length in aggregation179. Further work will be required to fully apprehend the assembly of these recently described pili and understand how they mediate their virulence-associated functions.

1.2.2 A monoderm-specific pilus: the sortase-dependent pilus 1.2.2.1 The first pili ever discovered in monoderm bacteria

For almost 20 years, fimbriae were thought to be exclusively found in diderm bacteria. It was only in 1967 that the first fimbriae were visualized at the surface of a monoderm bacterium: Corynebacterium renale181 (Figure9a,b). Since then, pili have been observed at the surface of many monoderm pathogens, including: Group A Streptococcus182, Streptococcus pneumoniae183 but also more recently on commensals such as Lactobacillus rhamnosus184 (Figure9c) and Bifidobacterium bifidum185. They are relatively long fibers (2 to 5 microns) but are thinner than diderm fimbriae with a diameter of only 3nm. These pili share a common mode of assembly which is dependent on the action of a sortase. While sortase-dependent are also polymers of major and minor pilins, as opposed to the pilins of diderm bacteria, the pilins of monoderm bacteria are covalently linked to form the sortase-dependent pilus.

1.2.2.2 Pili for adhesion and aggregation

Sortase-dependent pili are important virulence factors186. In Corynebacterium diphteriae, pili have been shown to play the role of adhesins. The three different types of pili expressed by C. diphteriae determine cell tropism187. Yet, expression of minor pilins has been shown to be both necessary and sufficient to maintain adhesion to most cell types in C. diphteriae and Group B Streptococcus187,188. These minor pilins can form heterodimeric assemblies at the surface of the cell in a sortase-dependent fashion in the absence of pili189. The unified view in this field is that minor pilins are the ones who play the pivotal role of adhesins.

Interestingly, sortase-dependent pili are also involved in the formation of biofilms. Mixed cultures of Streptococcus oralis and Actinomyces naeslundii T14V were shown to form tight interactions which were suggested to be dependent on the presence of these fimbriae190. This is further supported by a study in Streptococcus pyogenes showing the importance of these pili in both adhesion and interbacterial aggregation191.

22 1.2.2.3 Assembly of a peptidoglycan-anchored pilus

The general principle directing pilus assembly in monoderm bacteria was first elucidated in the model organism C. diphteriae192. A hallmark feature of sortase-dependent pili is that their pilins are linked through covalent bonds. Most sortase-dependent pili are composed of a major pilin that forms the shaft of the pilus and of two minor pilins: one anchoring the pilus to the and a tip-located adhesin (Figure 9d). While the location of the adhesin has been subject to intense debate, cryo-EM structures of S. pneumoniae pili support the idea that the adhesin is only found at the tip of the filament193.

Pilins are exported to the membrane through the Sec pathway. They have a specific sorting motif (LPXTG) at their C-terminus and also possess a pilin motif with a conserved lysine (YPKN). In the membrane, the sorting motif is recognized by a pilus-dedicated sortase which cleaves the pilin between the threonine and the glycine residue of the LPXTG motif. The sortase then covalently bonds the C-terminal threonine of the cleaved pilin and the conserved lysine of the pilin motif of the next pilin in the assembling pilus192,194–197. This process is repeated and leads to the elongation of a linear pilus until the final incorporation of the anchor minor pilin198. To terminate pilus assembly, the basal pilin is linked to an amino group of the lipid II of the cell wall peptidoglycan through the action of a housekeeping sortase199–202 (Figure 9d).

An interesting feature of these pilins is their very specific structure. While there is great size and sequence variation in sortase-dependent pilins, their structure is very conserved and is composed of a modular assembly of immunoglobulin-like domains195,203–205. These structures were found to bear two unusual internal isopeptide bonds that stabilize the pilin tertiary structure. These bonds have recently been shown to allow energy dissipation through reversible unfolding when high forces are applied to pili206. Just like diderm bacteria and the uncoiling of their chaperone-usher pilus, monoderm bacteria have evolved a way to maintain pilus-mediated adhesion under mechanical constraint. This mechanism involves the ternary structure of the pilin and not the quaternary pilus structure as is the case for the chaperone-usher pilus.

23

Figure 9: Sortase-dependent pilus appearance and assembly. (A-B) Electron micrographs of Corynebacterium renale. (A) unpiliated strain 5 (B) piliated strain 35. (C) Immunogold electron microscopy of Lactobacillus rhamnosus GG showing the presence of sortase-dependent pili at its surface. Scale bar: 500nm. (D) Top panel: Typical genetic organization of a sortase-dependent pilus . Bottom panel: simplified assembly model of the pilus by addition of pilins from the tip to the base through the action of membrane-anchored sortase. Pilus growth is terminated by anchoring the fiber to a cell wall lipid through the action of another sortase. Adapted from Kankainen et al. 2009, von Ossowski 2017 and Yanagawa et al. 1968. 24 Minor pilins at the tip have some additional specificities: the available structures of adhesin display surprising features. In S. pneumoniae, the minor pilin adhesion domain exhibits an unprecedented structure in prokaryotes that resembles the eukaryotic collagen-binding integrin domain207. In S. pyogenes, the adhesin domain bears an internal thioester bond. This bond is required for adhesion and suspected to mediate a novel mode of adhesion through the formation of a covalent bond with the adhesin receptor208,209. The reactivity of this thioesther bond has been shown to be regulated by mechanical forces to promote stable adhesion under application of a tension, a mechanism reminiscent of the catch-bond of the E. coli adhesin FimH210. This novel bacterial adhesion mechanism is suspected to be widespread in monoderm bacteria211 and underlines the variety of mechanisms evolved by monoderm and diderm bacteria to mediate adhesion under mechanical stress.

1.2.3 Archaeal pili: an unexplored diversity

Just like monoderm and diderm bacteria, archaea have evolved specific appendages with very unusual structures. Because of the difficulties associated with laboratory maintenance and genetic manipulation of archaea expressing these appendages, their assembly, structure and function remain poorly understood. Recently research has intensified to decipher how these structures function. We provide an overview of recent archaea-specific findings in the following section.

1.2.3.1 Hami: archaeal grappling hooks

Hami have only been observed at the surface of the non-cultivated euryarchaeon Candidatus Altiarchaeum hamiconexum until now (Figure 10a-c). These micrometer-long filaments with a 7nm diameter span both the inner and outer membrane and were first observed in 2005212,213. They have an unusual structure, with three prickles emanating from the fiber in a repeated pattern every 46nm and display a tripartite barbed grappling hook at their distal end212,214(Figure 10f). The filament is mainly composed of a single protein with an N-terminus homologous to proteins that form the S-layer212,213. S-layer: Surface layer found in From microscopic observations, it has been concluded that hami are important to promote most archaea and 214 many bacteria. It is surface adhesion as illustrated by their firm attachment to surfaces and filamentous bacteria . a 2D semi- These filaments are also important for biofilm formation, participate in the formation of mixed crystalline assembly of poorly structure between filamentous bacteria and Candidatus Altiarchaeum hamiconexum and are conserved also found to promote auto-aggregation214–216 (Figure 10b,d,e). While the lattice-like (glyco)proteins. organization of Candidatus Altiarchaeum hamiconexum isolates (Figure 10c) and the hook- structure found at the surface of the hami point to a role of this terminal structure in auto- aggregation, structural studies and the ability to cultivate this organism would undoubtedly provide new insights in the function and assembly mechanisms of these unusual filamentous structures.

25 Figure 10: Archaeal hami structure and functions. (A) Thin section of high pressure frozen biofilm sample of the SM1 archaeum image by electron microscopy shows an extremely dense crown of hami surrounding the cell. (B-C) Scanning electron micrographs of biofilms of the SM1 archaeum. (B) Cells form a dense network connected by hami. Scale bar 2 µm (C) Typical hexagonal organization of the cellular network. Scale bar 1µm. (D-E) Scanning electron micrographs showing the ability of SM1 hami to interact with bacterial filaments. (F) Negatively stained hami in a SM1 biofilm. Dotted line on the bottom right image represent the location of the membrane, the first prickle appears to lie at the surface of the outer membrane. Typical structure is illustrated on the right and shows regularly spaced prickles. The filament ends with a triple fish hook structure. Adapted from Perras et al. (2014) Probst et al. (2014) Moissl et al. (2005).

Pyrodictium: 1.2.3.2 Archaeal cannulae and bacterial spinae: intercellular communication fibers? Strict anaerobe and hyperthermo- Cannulae are another example of atypical extracellular filaments. They were first observed in phile archaea 1983 in the Pyrodictium genus where they form dense networks of bundling hollow tubules found in deep-sea 217 hydrothermal with a 25nm outer diameter (Figure11b,c). These fast-growing filaments connect mother and vents. daughter cells and can measure between 30 and 150 microns218. Cannulae solely attached to

26 one cell with a free end can also be found (Figure11d). These filaments appear to be anchored in the periplasm of cells and not in the cytoplasm and are made up of at least 3 glycoproteins219. Because cells without cannulae have never been observed, these filaments have been hypothesized to be necessary for survival. A function in nutrient exchange as well as a function in adhesion have been speculated but never demonstrated experimentally. These hollow tubes form a network that connects cells and eventually lead to the formation of multicellular aggregates (Figure11a). These structures are reminiscent of the bacterial spinae (Figure 11d) which were first observed in 1970220 and have since then been identified in many diderm bacterial species221. These structures also connect distant cells and have been hypothesized to be involved in cell-cell exchanges222. Given the omnipresence of such structures in populations of Pyrodictium and diderm bacteria, understanding to what end they form such structures could reveal a novel inter-prokaryotic communication mechanism.

Figure 11: The archaeal cannulae and the bacterial spinae share similar features. (A) Typical flakes appearing in cultures of Pyrodictium. The multicellular assemblies are hold together by the cannulae. Scale bar: 20mm. (B) Scanning electron micrograph showing a dense network of Pyrodictium cells connected by tubular cannulae. Scale bar:20µm. (C) Time-lapse microscopy of the growth of Pyrodictium abyssi TAG 11 showing cannulae connecting mother and daughter cells upon division. Some cannulae with a free end can also be observed (black arrows). (D) Negatively-stained Roseobacter sp. strain YSCB observed by transmission electron microscopy. (Left) Cells are connected to each other by their spinae. (Right) Spinae are randomly distributed at the cell surface. Adapted from Rieger and Racel et al. (1995) and Horn et al. (1999), Bernadac et al. (2012).

27 1.2.3.3 Mth60 fimbriae: species-specific multifunctional fimbriae

Yet another species-specific type of pilus was identified in Methanothermobacter thermautotrophicus and named Mth60 fimbriae223. These fibers have a particularly narrow diameter of 4-5nm and can extend up to 10 microns from the cell surface and are mostly composed of the Mth60 protein that is not related to any other known class of pilin. These filaments are necessary to adhere to a variety of surfaces (nylon, chitin…) and to form intercellular aggregates. This adhesion depends on Mth60 as shown by antibody-mediated inhibition of surface adhesion223. Structural and functional data are still lacking and could reveal yet another fiber assembly mechanism and pilus-dependent adhesion mechanism.

Figure 12: Mth60 fimbriae functions and appearance. (A) Staining of Methanothermobacter Thermoautotrophicus with Alexafluor-488 allows visualization of Mth60 fimbriae with fluorescence microscopy. (B) Scanning electron micrographs of Methanothermobacter thermoautotrophicus adhering to chitin. (C) Scanning electron micrographs of Methanothermobacter thermoautotrophicus adhered to Ni grids show the formation of multicellular aggregates covered with Mth60 fimbriae.

Adapted from Thoma et al. (2008).

1.2.3.4 Archaella: the archaeal motility structure ¬ Discovery of an archaeal flagellum unrelated to the bacterial flagellum Archaeal flagella were observed as early as 1956 in the archaeum Halobacterium salinarum224. Some of their specific traits were observed in the same organism in 1984225, even before the definitive introduction of the three-domain classification of life226,227. While these filaments appeared to rotate and propel the “archaebacteria” just like it did for the “eubacteria”, the archaebacterial flagella were found to be organized as right-handed helices by default as opposed to bacterial flagella. In addition, bundled flagella did not separate as individual fibers upon inversion of the direction of rotation225. It is only in 1996, that a consensus started emerging on the originality of the archaeal flagellum in comparison to the bacterial one228. This

28 realization came from the fact that the archaeal flagellins that form the flagellum share homologies with type IV pilins, and not with the bacterial flagellins229 (Figure13a). This homology, including a specific class III export signal, later proved to extend to the whole Defined in section 2.1.1.1 assembly machinery and not only the flagellins230. Archaeal flagella are also thinner than bacterial flagella with a diameter of 10 to 14nm somewhere between the diameter of pili and bacteria flagella231 (Figure13a). While the field is still debating the relevance of a change in nomenclature, we will use the term archaellum instead of archaeal flagellum in the rest of the manuscript231–233.

¬ Motility and more… Just like the flagellum, the archaellum is able to propel archaea by generating thrust upon rotation of the archaellum. This archaellum-mediated motility has now been demonstrated in many archaeal species225,230,234. The range of swimming speeds observed is quite similar to those generated by the bacterial flagellum and is situated between 3 microns per second for the slowest archaea and 380 microns per second (for the fast-swimming Methanocaldococcus jannaschii)235. Just like bacteria, archaea can modify their swimming direction by changing the rotation direction of their archaellum. In H. salinarum and many other archaea this directional motility is regulated by the chemotaxis system236,237which is extremely conserved between bacteria and archaea238. Interestingly the archaellum is a multi-function appendage. In addition to swimming, it can be used by archaea to adhere to various abiotic and biotic surfaces239–242 (Figure13b,c,e). The same archaellum also plays a role in biofilm formation by forming dense networks of filaments connecting archaea239–241 (Figure13d,e). Such observations have even been made in mixed populations, where P. furiosus could only colonize a substrate indirectly using its archaellum by adhering to Methanopyrus kandleriiwhich which had itself adhered to the substrate with its archaellum243. All these observations suggest a prominent role of the archaellum in niche colonization and symbiosis formation.

¬ Structure and assembly The machinery necessary for the assembly of the archaellum is much more reduced than that of the flagellum and only counts a dozen of proteins231 (Figure 13f). As for most appendages described so far, the shaft of the archaellum is composed of a single protein: the major archaellin. In some archaea, minor archaellins can also be found but their location in the archaellum is unclear thus far. Just like pseudopilins, the archaellins are inserted in the membrane and need to get cleaved by a dedicated peptidase prior to getting assembled as a polymer244,245. The structures of two archaella obtained by cryo-EM studies have recently been elucidated246,247, they show that the archaellins have a conserved N-terminus architecture similar to that of the pseudopilins and type IV pilins. The N-terminus makes up the hydrophobic core of the archaellum while the exposed C-terminal domain has an Ig-like fold. The archaellum

29 is not hollow, thereby excluding a polymerization mechanism of archaellins at the tip of the archaellum as the one described for the bacterial flagella.

Figure 13: Archaellum structure and functions. (A) Structural comparison of the bacterial flagellin, type IV pilin and the archaeal flagellin highlights homology between the archaeal flagellin and the bacterial type IV pilin. This structural homology is also visible at the level of the filament. (B,D) Scanning electron micrographs of the archaeum Methanocaldococcus villosus sp. nov. showing (B) heavy flagellation and (D) inter-archaeal interaction mediated by these flagella. Scale bar 1µm. (C) Scanning electron micrograph of Methanococcus maripaludis cells attached to molybdenum by their archaellum. (E) Scanning electron micrograph of Pyrococcus furiosus cells adhered to sand grains and each other via their archaellum. Scale bar: 2µm. (F) Partial model of the archaellum machinery of P. furiosus inferred from subtomogram averaging. Dotted lines between the polar cap and FlaH represent possible interactions.

Adapted from Poweleit et al. (2016), Daum et al. (2017), Jarrell et al. (2011), Näther et al. (2006).

30 Another difference between the archaellum and the flagellum is the energy source they use. Indeed, motor rotation is powered by ATP-hydrolysis and not ion gradientss248,249. Archaella are inserted at the level of polar caps composed of hexameric arrays of uncharacterized proteins247. The assembly machinery and motor fueling archaellum rotation is partially known. In contrast with the T2SS and Type IV pili machinery, a large part of the archaellum machinery protrudes in the cytoplasm. FlaI is the ATPase assembled as a hexameric module in the cytoplasm249 and is linked to FlaJ, the polytopic membrane protein that is believed to play the role of the assembly platform250,251. FlaX or FlaCDE (depending on the archaeal phylum considered) is inserted in the membrane and form rings around FlaH and FlaI thus linking the assembly platform FlaJ and the ATP-binding proteins FlaH-I251. This complex likely forms the motor of the archaellum as recently visualized by cryo-EM247,252. Two additional proteins, FlaF and FlaG are proposed to act as stators and to be anchored in the S-layer where they could stabilize the growing archaellum and allow rotation transmission247,252,253. The details of pilus assembly, how rotation is achieved, how adhesion and auto-aggregation are achieved are among the many question that remain to be answered in this system.

1.2.4 Pili found in all three types of prokaryotes: two different strategies to reach the surface

While monoderm, diderm bacteria and archaea have very different cell envelopes, two families of appendages are found in all three of these families. These include the Type 4 secretion systems and the Type 4 pili, of which the machineries have been adapted to the constraints of all prokaryotes. Unlike what is indicated by their misleading nomenclature, these two systems are completely unrelated. Yet, both systems mediate multiple functions, including competence for transformation. This could explain their prevalence in prokaryotes.

1.2.4.1 Pili assembled by type 4 secretion systems

The Type 4 Secretion Systems (T4SS) are versatile nanomachines found in monoderm and diderm bacteria and in some archaea. T4SSs can be encoded on plasmids and are involved in horizontal gene transfer which might explain how widespread they are254. As indicated by their name, they are involved in secretion of both proteins and DNA. Just like the T3SS, T4SS can form different types of appendages using a homologous secretion apparatus: an effector translocator needle (Figure14c) and a conjugation pilus (Figure14d). Alternatively, T4SS can Conjugation: 255 DNA transfer from form a DNA uptake and release machinery which does not involve the buildup of a filament a bacterial donor (Figure14a). In the following section we will present how the filamentous systems mediate their cell to a recipient cell by direct cell- functions. to-cell contact Bacterial sex was an intensive area of research in the 1950s. Researchers were trying to understand the molecular basis for genetic recombination between E. coli bacteria256. These genetic exchanges were shown to occur through contact-dependent transmission of the F episome257,258. It was only using electron-microscopy that the conjugation pilus was first observed in 1964259. Since then, conjugation pili have been discovered in many bacterial

31 species and are of growing interest because of their involvement in the spread of antibiotic- resistance260. This class of appendage has an extensive diversity and has been separated in two subfamilies: the type IVA secretion system (e.g. the F-pilus of E. coli and the Agrobacterium tumefasciens VirB/VirD4 system) and the type IVB secretion system (e.g. Legionella pnemophila Dot/Icm system). Only a few proteins are conserved between these two families261.

Figure 14: Type IV secretion systems structure, function and assembly. (A) Diversity of functions mediated by the T4SS. (B) Top: Maturation steps of the F pilus TraA pro-pilin. Bottom: Machinery of the T4SS and F pilus can be separated in three components: the T4SS inner membrane complex (IMC), the outer membrane complex (OMC) and the F pilus, which extends in the extracellular milieu (VirB2 is a TraA homologue). OM: Outer Membrane, PG: Peptidoglycan, IM: Inner Membrane. (C) Scanning electron microscopy of Helicobacter pylori 26695 adhering to AGS cells. T4SS-dependent filaments required for cell adhesion and invasion are visible at the cell surface (white arrow). Scale bar 0.5µm. (D) Electron micrographs of E. coli expressing the F pilus. (E) Top: Atomic structure of the F-pilus determined by cryo-EM shows a five-start helix. Bottom: central section showing the electrostatic potential of the central lumen modified with phospholipids. Blue represents a positive charge and red a negative charge. Adapted from Hospenthal et al. (2017), Galan et al. (2018), Ishibachi (1967), Rohde et al. (2003). The conjugation pilus is made of 4 protein assemblies: the pilus, the type 4 secretion apparatus, a type 4 coupling protein (which couples DNA binding and the secretion apparatus262) and the relaxosome which nicks and separates DNA strands to enable transfer of ssDNA261,263. The type IVA family is usually made up of twelve proteins: VirB1-11 and VirD4 (in

32 the A. tumefasciens nomenclature) presented in Figure 14b264. The T4SS spans the whole cell envelope in diderm bacteria. It is composed of an inner membrane complex with 4 stacked structures arranged in dodecamers (composed of VirB3-6 and VirB8) linked via a stalk (VirB10) to an outer membrane complex made of two layers and 14 copies of VirB7, VirB9 and VirB10. This allows the polymerization of a pilus (made of the major pilin VirB2 and minor pilin VirB5)265– 267. Interestingly, the structure of a type IVB secretion systems shows a very similar structural organization despite the poor sequence conservation between the proteins of these 2 systems268. Three ATPase are also important to power pilus formation and DNA transfer. VirB4 and VirB11 are necessary for pilus formation while the VirD4 ATPase acts as a coupling protein and is necessary for DNA recruitment. To assemble pili, the propilin VirB2 composed of 3 α- helices is addressed to the inner membrane in a Sec-independent manner and processed by TraQ and LepB269 before being acetylated at the N-ter by TraX270 (Figure 14b). As revealed by the recent cryo-EM structure of two conjugation pili, the pilin is then lipidated by addition of a negatively-charged phosphatidylglycerol molecule271.

The pilus itself has an outer diameter of 9nm and a hollow lumen with a 3nm diameter. Because of the presence of the phospholipid modifications, the lumen is slightly electronegative; this has been proposed to facilitate DNA transport through the lumen271 (Figure14e). The events leading to DNA transfer to a target cell remain very elusive, but, it has been shown that conjugation can happen at distances of 12 microns which supports the idea that conjugation pili not only bring cell in contact but also mediate DNA transport272. Interestingly, it has also been shown that the assembly of an extracellular pilus is not absolutely required to mediate DNA transfer but that the expression of VirB2 and VirB5 is required273,274. This suggests that the major pilin VirB2 and the minor pilin VirB5 could form a pseudo-complex (very short pilus) to either push DNA through the outer membrane or simply allow DNA passage through the T4SS275.

The T4SS has also been shown to assemble needles necessary for effector translocation276. The homolog of VirB2 in Helicobacter pylori was found to form the surface and tip of such needles and has been shown to bind to integrins of the target cells (Fig.12c). Mutations in the VirB2 integrin-binding site significantly decrease translocation efficiency277.

How conjugation pili and needles are assembled by the T4SS is still poorly characterized. Similarly, how molecule translocation (DNA or effectors) is mediated by the different appendages assembled by T4SS is not well-understood. The recent elucidation of the structure of conjugative pili should greatly help the understanding of T4SS.

1.2.4.2 Type IV pili: the all-in-one prokaryotic appendages

Another example of an even more widespread appendage mediating horizontal gene transfer in prokaryotes is the Type IV pilus (T4P). T4P have been identified in more than 1800 species in monoderm, diderm bacteria and archaea6,278,279.

33 ¬ Distribution and discovery It was in 1975 that TFP were first proposed to form a unique class of fimbriae. This was based on the observation that these pili were involved in a specific form of bacterial motility named twitching motility280–282. It is only in the late 1980s that the phylogenetic relevance of grouping these appendages under the name of type IV pili became clear283,284. Presence of TFP had only been documented in diderm bacteria until 2002. TFP were formally identified for the first time in a monoderm bacterium in Ruminococcus albus in 2002285 where they were associated with cellulose adhesion (Figure 15a). TFP were then identified in Clostridium species in 2006 where they were associated with twitching motility 286,287. This was quickly followed by their identification in multiple species of archaea in 2007245,288–290. Archaeal TFP are different from the archaellum in both their appearance (TFP are thinner), function (TFP do not promote swimming through filament rotation) and machinery (Figure 15a).

All TFP observed so far have a diameter of 6 to 10nm and can extend from 200nm to 10 microns from the cell surface. Most of these pili have a unique property in that they are retractile291–293. Retraction of TFP has never been observed in archaea so far. These retractile properties enable the bacteria to generate considerable forces: in the order of a 100pN per pilus294 and is undeniably a critical factor in their multifunctionality.

¬ TFP: multi-tasking champions TFP are by far the appendages mediating the highest number of documented functions. They are important for twitching motility291, protein secretion295, surface adhesion, auto-aggregation (Figure 15b), biofilm formation296, competence for transformation297, host cell manipulation298– 300, surface sensing301,302, phage susceptibility303,304 and the most recently described function: electron transfer305. These functions will be described in greater detail in Section 2.2. and Section 3.3. How all of these functions are mediated by a single appendage is the focus of this manuscript.

¬ Structure and assembly Type IV pili of diderm bacteria are classically separated in two classes: type IVa (e.g. N. meningitidis, N. gonorrhoeae and P. aeruginosa) and type IVb pili (e.g. Toxin-coregulated pilus (Tcp) of Vibrio cholerae and Bundle forming pilus (Bfp) of EPEC)283. Prepilins of the type IVa class have short leader peptides (less than 10 amino acids) while type IVb pilin tend to have longer leader peptides (15 to 30 amino acids). The piliation machineries of type IVa and IVb pili also present several differences. Machineries of the type IVa are usually more complex (with up to 20 different proteins) and are scattered through the genome. Machineries of the type IVb pilus are limited to 10-12 different genes which are clustered in the genome296,306. TFP of monoderm bacteria possess both characters of type IVa and type IVb pili and are thus not easily

34 classified in one of these two classes279. For archaeal TFP, it is still unclear how they relate to this classification.

The structure of TFP and of the piliation machinery has been studied with extensive care and has greatly benefited from the recent technical advances in cryo-EM. To this date, four pili structures have been resolved at very high resolution by cryo-EM (for N. meningitidis, N.

Figure 15: Type IV pili structure, functions and assembly. (A) Electron micrographs of prokaryotes expressing TFP. Diderm bacterium: Neisseria gonorrhoeae, monoderm bacterium: Ruminococcus albus, archaeum: Sulfolobus solfataricus. (B) Type IV pili functions. Top panels: Scanning electron micrographs of EPEC shows TFP- dependent multicellular aggregate formation. Bottom panel: Localized adherence of EPEC to HEp-2 cells by microcolony formation (arrows). (C). Simplified illustration of a type IVa pilus machinery. Subcellular organization of these proteins was determined by subtomogram averaging. OM: Outer Membrane, PG: Peptidoglycan, IM: Inner Membrane. (D) Illustration of the minimal machinery required to assemble TFP. The inner membrane complex PilMNOP forms stable interactions with PilG. This leads to the recruitment of the assembly ATPase PilF. This complex can then polymerize mature PilE upon cleavage of its signal sequence By the peptidase PilD.

Adapted from Hospenthal et al. 2017, Goosens et al. 2017, Ajon et al. 2011, Rakotoarivonina et al. 2002 and Craig et al. 2006. gonorrhoeae, P. aeruginosa and Myxococcus xanthus) ,all of which are type IVa pili307–310. Yet, earlier studies with lower resolution structures of type IVb pili have shown an overall conservation of the pilus structure between type IVa and type IVb pili311–313. Structures of integral piliation machineries have also been described309,310,314 in V. cholerae, Myxococcus

35 xanthus and Thermus thermophilus in situ using cryo-electron tomography (Figure15c). While the protein making up the machineries of type IVa and type IVb pili are poorly conserved, these studies have also revealed very similar structures and assemblies between these two systems. This suggests that both systems roughly follow the same assembly mechanisms.

Combining these in situ data and the recent in vitro data on the minimal machinery needed to assemble TFP315 (Figure15d) , it can be summarized as follows for diderm bacteria (we use the nomenclature from N. meningitidis). Just like for the T2SS, the TFP machinery can be broken down into three parts: the extracellular pilus, an outer membrane translocation pore (the secretin) and the inner membrane and cytosolic components of the assembly platform309,310,314. Most of the non-cytosolic components of the TFP machinery are first exported through the Sec system. A complex of four inner membrane proteins (PilMNOP) recruit the inner membrane polytopic protein (PilG) and the assembly ATPase (PilF)310,315,316. This is sufficient to initiate pilus assembly in the periplasm315,317,318. Yet, to cross the outer-membrane, the TFP assembly system has to use a secretin (PilQ) which is connected to the rest of the machinery by one of the protein (PilP) from the four-protein inner membrane complex described above319. Secretins are assembled as rings in the periplasm and inserted in the outer membrane where they form a pore that will enable pilus extrusion upon assembly320,321. Their stability, assembly and proper localization are facilitated by named pilotins that surround the secretin pore318,322. Just like all the systems described so far, in most TFP the shaft of the filament is composed of a single protein: the major pilin (PilE). The location of minor pilins is still controversial323–325 and unlike most of the pili presented so far, they have never been observed at the tip of the pilus. Prepilins, are addressed to the inner membrane thanks to their class III signal peptide245,284. Their cytoplasmic signal sequence is then cleaved by a dedicated prepilin peptidase (PilD) which will allow their incorporation in the growing helical pilus326. As explained earlier, pili are retractile. Retraction is often mediated by a second cytoplasmic ATPase: the retraction ATPase (PilT) that replaces the assembly ATPase and disassembles the pilus327. Retraction can also occur through the action of minor pilins in the absence of a retraction ATPase, as recently evidenced in V. cholerae328. Minor pilins have also been involved in the initiation of pilus assembly but are not essential for pilus assembly in the absence of a retraction ATPase156,318,329. The extent to which these mechanisms are conserved in monoderm bacteria and archaea is unclear at this point. Yet, the virtual presence of these appendages in all classes of prokaryotes suggests a very ancient origin of TFP6. Their spread through the species and the ages as well as the multiple functions they accomplish underscores their central role in the life of prokaryotes. The rest of this thesis will be focused on these fascinating appendages.

36 2 Type Four Filaments: multifunctional homologous systems

As evidenced in the previous chapter, TFP not only form a superclass of appendages but also have striking homologies with the T2SS and the archaellum. All of these filaments are polymers of pilins which share a class III signal peptide and have quite conserved assembly mechanisms and machineries (Figure16a). This class of appendages has been coined Type Four Filaments (TFF)6. Combining the knowledge acquired on these three related systems greatly helps understand how TFF emerge at the prokaryotes’ surface and mediate their various functions. To provide a detailed understanding of TFF, we will therefore discuss the evidence obtained from all these systems in the following sections.

Figure 16: Type four filaments share a conserved machinery.

(A) Illustration of the overall conservation between the different TFF machineries. Functional homologies between the components of the different systems are color-coded.

(B) Distribution of the different proteins involved in TFF assembly in archaea and bacteria. Black squares indicate that the signature motif of the protein was present in at least one species in the analyzed phylum. White squares indicate that the protein could not be detected.

Adapted from Berry and Pelicic 2015.

37 2.1 A conserved biosynthesis machinery formed by 3 complexes

As mentioned earlier, all these systems share common mechanisms of biosynthesis (Figure 16a). This is illustrated by the fact that four proteins of the machinery are extremely conserved in these 3 systems and in all prokaryotes: the major pilin, the prepilin peptidase, the polytopic inner membrane assembly protein and the elongation ATPase6 (Figure 16b). We provide an overview of the proteins involved in assembly of TFF in the following section by separating them in 3 subcomplexes: the inner membrane complex, the outer membrane complex (for diderm bacteria) and the filament. In this section, we will use the nomenclature of N. meningitidis and refer to the major constituent of all filaments under the unifying term pilin.

2.1.1 The inner membrane complex 2.1.1.1 The prepilin peptidase cleaves the leader peptide of the class III signal

The prepilin peptidase PilD is a bifunctional enzyme present in the inner membrane and virtually found in all TFF-expressing prokaryotes. This enzyme has a key role in the processing of the class III signal peptide of the (major and minor) pilins of the TFF family6. As shown on Figure 17, the class III signal peptide is constituted of a leader peptide composed of hydrophilic amino acids almost always terminated by a glycine residue and followed by a stretch of hydrophobic amino acids245. Upon translation, the prepilin is addressed to the inner membrane via the Sec pathway330,331. The positively charged leader peptide remains in the cytoplasm while the hydrophobic stretch is inserted in the inner membrane332,333.

PilD has two functional domains that are thought to be associated with the two different activities of this enzyme. It first cleaves the cytoplasmic leader peptide of the prepilins after the hyperconserved glycine residue and subsequently N-methylates the new amino-terminal in the cleaved pilin334. N-methylation is not observed in archaea335 and the archaeal PilD may be responsible for other N-terminal post-translational modifications that have been observed in archaeal pilins336. This difference is also supported at the structural level, with the protease domain being the only one conserved between the bacterial and archaeal PilD337.

Cleavage of the leader peptide is required for subsequent polymerization of pilins as filaments, but methylation is not necessary for assembly338,339. The amino acids required for leader peptide cleavage are variable between species, but the Glycine-1of PilE is almost always required for this activity. Interestingly, it has been shown experimentally that PilD from a given species can cleave pilins from different species or even from a different filament class, thereby illustrating the overall conservation of these mechanisms in the TFF family326,340,341. The role of the N-terminal methylation is still unclear but has recently been proposed to facilitate inner membrane extraction of the pilin by the assembly platform component PilO342.

38 Figure 17: Conservation of the class III signal peptide. Sequences of major and minor (pseudo)pilins of the different TFF system are shown show an overall conservation. The class II signal sequence is composed of a leader peptide of hydrophilic amino acids almost always terminated with a glycine followed by a stretch of hydrophobic amino acids that fold as an α-helix. Prepilin cleavage site lies between the glycine and the first hydrophobic amino acid of the leader peptide. Blue coloring indicates hydrophobic residues and orange coloring indicates charged ones.

Adapted from Berry and Pelicic (2015)

2.1.1.2 The assembly platform initiates pilus assembly

The assembly platform is anchored to the inner membrane in diderm prokaryotes and allows the initiation of filament assembly. In most of the TFP and T2SS systems that have been studied, this platform is composed of 4 to 5 proteins: PilM,N,O,P and PilG (Figure 18a). In T2SS, PilM and PilN are merged in a single protein (GspL).

Of these proteins, the only one consistently found in the TFF family is PilG6. This polytopic inner membrane protein was discovered early in the 1990343 and is absolutely required to assemble surface pili315,344. PilG interacts with the major pilin157,345, the assembly ATPase PilF152,346–348 and PilMNOP152,349 (Figure 18b).

PilMNOP forms a supramolecular complex with a 1:1:1:1 stoichiometry315,345,349–355. The interactions between these proteins have been elucidated in great details and can be summarized as follows: The periplasmic portion of PilP interacts with the secretin PilQ and serves as a conduit between PilMNO and PilQ. PilP also interacts with PilN and PilO which form heterodimers in the membrane. PilM which is mainly cytoplasmic is then recruited by the cytoplasmic portion of PilN to the PilNOP complex and forms a ring below the inner membrane. The presence of this PilMNOP supramolecular complex allows the recruitment of the PilG

39 dome152,349, eventually leading to the recruitment of the assembly ATPase PilF (Figure 15d). Such a scenario is supported by direct visualization of subtomograms of the piliation machinery309,310 (Figure 18a). Just like PilG, PilMNOP also interacts with the major pilin157,342,349,355,356and the assembly and retraction ATPases (PilF and PilT) 152,346–348,357. Because of the complexity of these interactions, it is still unclear whether the motion that has been proposed to be generated by the assembly ATPase358,359 is transmitted to PilG, PilMNOP or both to allow pilus assembly. Because the archaellum is a rotating structure230 and the T2SS pseudopilus has also been suggested to rotate360, a rotational assembly mechanism of TFF pilus is currently favored. Because PilMNOP are directly connected to PilQ which is anchored to the peptidoglycan, it seems unlikely that these components might be the ones rotating, unless if the complex they formed is only transient. This suggests that PilG could be the rotating structure involved in rotational pilus assembly310.

40

Figure 18: The inner membrane complex. (A) Molecular structure of the assembled inner membrane complex as visualized by cryo- electron tomography. Top panel: Top view of the structure, middle panel: side view, bottom panel: bottom view. Figure was generated using Pymol and the PDB-deposited structure: 3jc9. (B) Protein interaction network constructed from bacterial Two-hybrid detection system using meningococcal proteins of the TFP machinery.

Adapted from Georgiadou et al. 2008

2.1.1.3 The ATPases: powering pilus assembly and retraction

Major pilins of the TFF have to be extracted from the membrane in order to polymerize as filaments. This process requires an energy source that is always provided by the cytoplasmic traffic ATPase PilF 6,361,362. For TFF disassembly, the situation is more complex. In archaea, filament retraction has not yet been documented. For the T2SS, what powers the disassembly process is still unclear and is thought to be driven by proteolysis or to happen spontaneously149.

41 For the bacterial TFP, the situation is even more complex. TFP retraction has been observed in diderm and monoderm bacteria and documented for all subtypes of TFP but with different modalities 292,327,328,363. The disassembly ATPase PilT is only present in bacteria expressing type IVa pili. Thanks to the action of PilT, TFP can be retracted at extremely rapid speeds of up to 1 μm.s-1 (representing the disassembly of approximately a thousand units of PilE per second) and generate considerable forces in the range of 100pN per pilus 292,294,364. Retraction in type IVb pili is mediated by minor pilins and generates much lower forces of about 4pN328. For the Tad pilus (sometimes referred to as a type IVc pilus), the mediator of retraction has not been identified yet363. The structures of both ATPases (PilF and PilT) have now been deciphered in great detail. Both proteins are member of the AAA+ (ATPases Associated with diverse cellular Activities) family365. As monomers, this family of protein displays a characteristic structure composed of two distinct motifs: the Walker A box which is necessary for ATP-binding and the Walker B box which is important for Mg2+ binding to catalyze ATP hydrolysis366. All of the TFF ATPases form bi-lobed structures with a N-terminal domain linked by a flexible segment to a bulkier C-terminal domain which contains the Walker boxes. These proteins assemble as hexameric rings250,367–370 (Figure 19a). Most recent structures, suggest that each monomer goes through three different conformations depending on its ATP-binding state (ATP bound, ATP hydrolysis and ADP-ATP exchange). Conformational states are symmetrical in the hexamer. Cycling between these open and closed conformations would trigger clockwise rotation for PilF and anticlockwise for PilT (Figure 19b). This rotation is accompanied by a vertical deformation of the structure from a saddle shape to a planar shape that could transduce the movement necessary to scoop PilE outside of the membrane (or of the pilus) to PilG or PilM, as previously discussed358,359 (Figure 19a).

Two additional paralogues of the PilT ATPase can be found in the type IVa machinery: PilT2 and PilU. The role of these AAA+ has mostly been studied in pathogenic Neisseria and P. aeruginosa but remains unclear. pilU mutants have very diverse phenotypes depending on the model in which they are studied. While pilU mutations have been shown to induce loss of motility in P. aeruginosa, they barely affect motility in Neisseria and do not abolish retraction in either case371,372. This mutation also results in an increased adhesion in pathogenic Neisseria371,373 but in a decreased ability for bacteria to form aggregates371,374,375. In P. aeruginosa, PilU has a polar localization at the piliated pole unlike PilF and PilT which are located at both poles376. Interactions between PilU, PilF and PilT have also been documented349 and suggest that by interacting with PilF and/or PilT, PilU could favor elongation over retraction. A mutation in pilU could result in the display of shorter pili which would account for all the phenotypes described above. Yet, pilU mutants appear hyperpiliated in P. aeruginosa thus disproving this hypothesis372.

42

Figure 19: Structure of the ATPases PilF and PilT. (A) Crystal structure of the hexameric assembly ATPase PilF of Geobacter metallireducens. PilF monomers show a typical bilobed structure. The hexamers were observed when bound to ADP or to AMP-PNP. A structural transition from a closed to an open state is observed and accompanied by vertical deformation upon ATP hydrolysis. (B) Model explaining pilus (de)polymerization by the action of PilF or PilT. Monomers in both hexamers alternate between three states which causes rotation of the motor. Clockwise for PilF and counterclockwise for PilT. This rotation results in vertical deformation of the hexameric assembly. This motion is transmitted to the assembly platform (PilG?) and results in (de)polymerization of the pilus.

Adapted from McCallum et al. 2017.

More work is needed to elaborate a working hypothesis to explain the function of this ATPase. The other PilT paralogue, PilT2 is found in pathogenic Neisseria but not P. aeruginosa. pilT2 mutants are also affected in their retraction properties: they are more piliated than the WT373 and retract their pili and move at a reduced speed377. This is reminiscent of the phenotypes

43 observed in the pilU mutants of P. aeruginosa. Interestingly, PilU of P. aeruginosa presents stronger sequence homologies to neisserial PilT2 than to neisserial PilU. It is therefore possible that upon this second duplication of PilT, neisserial PilU and PilT2 diverged to mediate different functions. This would explain the discrepancies of phenotypes observed between different species. Just like PilU, PilT2 was also shown to interact with PilF, PilT and PilU349. These interactions suggest two possibilities: either that heterocomplexes could be formed between the different ATPases or that hexamers of different ATPases could form stacking interactions to fine-tune the dynamics of TFP Yet, there is no evidence to support the formation of such complexes thus far.

2.1.2 The outer membrane complex: crossing the outer membrane

Once assembled in diderm bacteria, TFF have to cross the outer membrane to reach the cell surface. This is mainly achieved through the action of the secretin PilQ which is often assisted by a variety of pilotins. Assembly and extrusion of TFF are two distinct events. Indeed, pilQ/pilT double mutants, grow TFP that remain stuck in the periplasm317.

Secretins form a conserved class of pore-forming membrane proteins which are also found in the T3SS. To be addressed to the outer membrane and form stable oligomers, these proteins are assisted by outer membrane proteins named pilotins (reviewed in detail elsewewhere322). Depending on the system considered, pilotins are either absolutely required to address the secretins to the outer membrane378, to anchor them to the peptidoglycan379, for their proper oligomerization380 or simply to increase their stability318. There is no unifying model to explain how pilotins mediate the assembly and transport of the secretins to the outer membrane.

Over the years the structures of many secretins have been obtained by electron microscopy 381– 386 and more recently at near-atomic resolution thanks to the advances made in the field of cryo- electron microscopy48,387–389. These recent studies demonstrate considerable variation in the structure of secretins. While for all the structures observed so far, the C-terminal domains form a barrel in barrel structure where both the inner barrel and the outer barrel are a transmembrane assembly of β-strands, there is considerable variation in the number and organization of the N-terminal domains that form the periplasmic channel (Figure 20a). This variability is further illustrated by the observation that depending on the model studied, the number of monomers forming the pore can vary between 12 to 19387. All of these models agree to show that in the absence of a growing TFF, the secretin is gated both at the level of the β- barrel and of the periplasmic domains. As PilQ can be very abundant at the cell surface390, such a gating prevents periplasm leakage and entry of toxic compounds. Upon TFF assembly, the secretin undergoes significant structural change leading to its opening and the extrusion of the filament314,321. A model is presented on Figure 20b.

44 Figure 20: Structure/function relationship of secretins. (A) Top: Cryo-EM structure of the PilQ secretin from P. aeruginosa. Cross- sectional view highlights the conserved structural features found in all secretins. OM: outer membrane. Bottom: Structural comparison of secretins from different bacteria shows the great variability that exist among these structures. (B) Left: Structure of the GspD secretin from V. cholerae. Native conformation of the secretin does not allow aspecific protein leakage . Right: Transition from a closed to an open state. Model recapitulating the structural modifications that secretins have to undergo upon pilus assembly to allow pilus extrusion or protein secretion through the secretin pore.

Adapted from Yan et al. 2017 and Koo et al. 2016.

2.1.3 The filament

The high-resolution structures of a dozen of TFF are available to this date: the structure of five TFP307,308,391,392, three archaella246,247,393 and one pseudopilus159. Most of these type four filaments are essentially composed of a single protein: the major pilin. Exceptions have been observed in purified archaella, where up to 5 major archaellins can be found at similar levels in the archaellum231,394. Other members of the pilin family are expressed at much lower levels and

45 make up the minor pilins. Their precise location and role is still under debate and will be discussed in this section.

2.1.3.1 Major pilins: major components of the pilus

Major pilins are expressed at much higher levels than their counterpart minor pilins. This represents a major metabolic cost for the cells which can tightly regulate pilin expression at the transcriptional level296,391,395.

¬ Structure Dozens of pilin crystal structures are available throughout the literature311,325,396–407. This extensive structural characterization has provided unprecedented insights in the overall conservation of the pilin fold and in the varied stabilization methods they utilize. Overall, pilins resemble lollipops, with a long stick (α-helix) covered at its end by a globular domain (Figure 21a-c). The N-terminal α-helix is the most conserved structural feature of pilins. It also has the highest sequence conservation. This α-helix can be separated in two parts: the amino-terminal one (α1N) of approximately 21 residues and the carboxy-terminal one of about 30 amino acids (α1C). This separation is often materialized by a kink induced by a proline or glycine residue. The α1N domain protrudes from the globular domain and forms the transmembrane region408 while the α1C is packed in contact with the globular head. The α-helix is connected at its end to a ß-sheet made of antiparallel ß strands by the αß-loop. Stapled to the last ß strand lies an additional structured region (the Disulfide-region in type IVa pili). Most of the structural diversity lies in the αß-loop and the D(isulfide)-region that are surface exposed in the quaternary structure of the pilus296. The name of the D-region comes from the fact that it is stabilized by a disulfide bond formed between two cysteine residues. Yet, this only appears to be the case for type IVa and a minority of type IVb pilins. Depending on the studied TFF, different stabilization strategies have been adopted. They include : coordination of a Ca2+ cation by pseudopilins159,409 (Figure 21c), a dense network of hydrogen bonds or coordination of a zinc atom in type IVb pilins and pilins of gram positive bacteria410,411(Figure 21b) and a Ig- like ß sandwich fold in archaea 393.

46 Figure 21: Conservation of type IV pilins structure. (A) Type IVa pilins have a conserved fold. A very hydrophobic N-terminal α-helix with a kink (in cyan) connected to a ß-sheet (in gray) by a variable αßloop (in magenta). Most pilins are stabilized by a disulfide bond that connects the D-region (in blue). (B) Type Ivb pilins share a very similar fold but display a longer D-region (C) Pseudopilins share the same organization but lack the disulfide bond. Instead, structural stabilization is allowed by coordination of a calcium ion (orange sphere). Adapted from Giltner et al. 2012.

47 ¬ Post-translational modifications Another conserved feature of these pilins is that they are all post-translationally modified. As mentioned in the previous section, all these pilins are cleaved in their type III signal sequence and can subsequently undergo N-terminal methylation. In addition to this modification, extensive glycosylation of pilins has been reported398 with up to 7 glycosylation sites on a single pilin246. Additional modifications have been reported, such as the addition of phosphate- containing molecules412. All these different types of modifications can be combined on a single pilin413. With the advances of mass spectroscopy, this list of post-translational modifications identified in pilins is likely to grow.

The functions associated to these modifications are variable. Glycosylation is not critical for pilus assembly in bacteria in most cases, but it is necessary for filament assembly in archaea414,415. Glycosylation and other post-translational modifications have also been associated with filament functions: adhesion416,417, aggregation413,418 and more recently, phage binding419. Because extensive glycosylation dramatically modifies the surface of TFF, these modifications have also been suggested to facilitate immune escape413. To our knowledge, other than cleavage and N-methylation, post translational modifications has never been reported for pseudopilins of the T2SS. Because the pseudopilus does not reach the exterior of the cell, their absence could support a role of these post translational modifications in immune escape or in functions mediated in the environment.

2.1.3.2 Minor (pseudo)pilins: a start/stop button?

In addition to the major pilin, most systems also comprise other pilins expressed at much lower levels: the minor pilins. In this section, we will only discuss minor (pseudo)pilins of the bacterial TFP and T2SS as the situation is not so clear for archaea.

Figure 22: Structure and functions of minor pseudopilins. (A) Left -Crystal structure of the pseudohelical quaternary complex formed in vitro by PilIJK. Right - Proposed helical assembly model of the PilIJK structure with PilE or PilH. The red ellipse indicates a conserved PilK interface that was proposed to block addition of pilins at the top of this structure. This suggests a terminal position of this complex in the (pseudo)pilus and a role in initiation or termination of pilus assembly. (B) Model for the role of the multiple minor (pseudo) pilins in the initiation of Type IVa pilus and pseudopilus assembly. The PilHIJK complex binds the major (pseudo)pilin and initiates elongation upon recruitment by the assembly platform. (C) Model for the observed role of the minor pilin TcpB in Type IVb pilus (dis)assembly. Adapted from Korotkov and Hol, 2008, Ng et al. 2016, Nivaskumar et al. 2016.

48 Legend on previous page

49 Four core minor pseudo(pilins) are very conserved between the T2SS and Type IVa pili: PilH-K. These minor pseudopilins are not absolutely required for piliation but their absence results in dramatically reduced piliation levels and abolishes TFF functions318,420–422. These proteins have been found to interact with each other and several lines of evidence point to a model where PilI, PilJ and PilK form a trimeric assembly at the inner membrane, this complex would then be bound by PilH which could serve as a linker to bind PilE and prime pilus assembly324,422–426 (Figure 22a,b). The role and localization of these minor pilins is still controversial. They have been reported in the pilus324 and have also been proposed to make up the tip of the pilus423,425 but such a structure has never been visualized. Recent studies of the type IV pilus machinery by cryo-EM subtomogram averaging also support the presence of a minor (pseudo)pilin complex in the inner membrane at the basis of the machinery prior to pilus assembly310. While PilH-K likely facilitate pilus assembly, their presence at the tip remains to be demonstrated. Presence of PilH-K inside of the fiber may indicate that this complex also plays a role during pilus elongation and could participate in interrupting pilus growth. Such a dual role is observed for TcpB, the unique minor pilin found in the type IVb pilus of V. cholerae328 (Figure 22c). Furthermore, cryo-EM subtomogram averaging of the Tcp pilus suggests a similar position to that of the PilH-K complex and an additional role in recruiting the PilG polytopic membrane protein309. Altogether, this supports a conserved role of minor (pseudo)pilins in the initiation of the assembly and retraction of TFF. Whether these minor (pseudo)pilins are found at the tip of the TFF will need further investigation.

Additional minor pilins are found in some type IVa pilus systems. Because of their great species- dependent variability, we will only discuss the meningococcal minor pilins PilV, PilX and ComP and their functions in section 3.3.1.2.

50 2.2 One machinery, many functions

TFF have been involved in an incredibly diverse array of functions that are represented in Figure 23. Most of these functions are shared between the different systems and we try to highlight functional homologies and divergences in the following section.

Figure 23: Type IV filaments are involved in a wide array of functions. Representation of the diversity of functions that have been described for TFF so far. TFF share various functions. Interestingly, TFP appear to be extremely multifunctional as they are able to mediate all of these functions.

2.2.1 TFF mediate attachment through surface adhesion

For prokaryotes, appendages are a privileged way to explore surrounding surfaces and to colonize new environments. In most cases, this process is initiated by adhesion to a surface. The importance of TFF in surface adhesion is clearly established, but the structures responsible for adhesion remain unclear6.

Both TFP and the archaellum have been directly involved in binding biotic and abiotic surfaces6,427 (Figure 24) and these two types of appendages can have complementary roles in archaea240. An anecdotal role for T2SS in adhesion to murine intestinal epithelial cells by EHEC has also been reported143.

Conflicting evidence has been provided in various model organisms regarding the mechanisms behind adhesion. Two related points are currently unclear: does adhesion happen at the tip or along the length of the TFF and what are the proteins responsible for the adhesive properties of TFF?

51 To date, the only report of live observations of TFP-mediated adhesion has shown tip-mediated adhesion of P. aeruginosa to abiotic surfaces428. These observations are also supported by the identification of a binding site on the major pilin located at the tip of TFP and involved in adhesion to biotic and abiotic surfaces 429,430. Using electron microscopy to monitor adhesion of TFF to surfaces shows both interactions over extended length and at the tip 240,241,431 (Figure 24b-e). Interactions over extended fiber lengths are also supported by AFM adhesion force measurements made by pulling TFP-attached live cells from a substrate432 but the interpretation of these experiments can be subject to debate. Both types of observations have been made repeatedly in various organisms and would support the coexistence of these two types of adhesion.

Figure 24: TFF-dependent adhesion to biotic and abiotic surfaces. (A) Scanning electron micrograph of a biofilm of Pyrococcus furiosus adhering to sintered quartz over extended length of their archaellum. (B-C) Scanning electron micrographs of Methanococcus maripaludis adhering to glass (B) or silicon (C) with their archaella in bundles (thick arrows) or as single filaments (thin arrows). Scale bar: 100 nm. (D-E) Scanning electron micrographs of N. meningitidis adhering to human epithelial cells. Close contact between TFP (thin fibers) and cell membrane protrusions (large fibers) can be observed (white arrows). It is unclear whether this contact occurs at the tip or along the length of the pilus. Scale bar: 1 µm.

Adapted from Mikaty et al. (2009), Näther et al. (2006) and Jarrell et al. (2011).

The molecular mediators supporting adhesion are also under intense debate. No adhesin complex has ever been convincingly observed at the tip of TFF and the major pilin remains the most likely mediator to explain tip-mediated adhesion. This is also supported by findings made in P. aeruginosa and E. coli which show the direct role of the major pilin as an adhesin429,430,433. Furthermore, the intrinsic adhesive properties of the major pilin would also explain the

52 adhesion of TFP over extended length. The situation is even more complex when it comes to archaea, where some TFP have been found to be composed of up to 6 different major pilins434. In Haloferax volcanii, expression of only one pilin is sufficient to restore piliation but 2 of the 6 pilins do not restore their adhesive properties to these pili434. This further suggests that major pilins are sufficient to provide TFP adhesiveness. The mechanism behind archaellum- dependent adhesion remains to be investigated. The argument over the role of other proteins in TFP-dependent adhesion is not discussed here and will only be presented in the context of N. meningitidis in section 3.3.3.5.

TFP-mediated adhesion is not only useful to colonize a site, but also to explore it using TFP- dependent motility.

2.2.2 TFF allow prokaryotes to move in various ways

Two distinct types of motility can be mediated by TFF: twitching motility that is mediated by TFP in bacteria and swimming that is mediated by the archaellum.

Twitching motility is virtually present in all bacteria expressing retractile TFP291. This motility is characterized by a jerky cell motion. This peculiar motion results from successive steps of pilus elongation and attachment to a substrate followed pilus retraction and detachment435,436. Because of the differences that exist in the bacterial cell shape and in the localization of TFF, motion coordination is regulated by very different mechanisms in these organisms (Figure 25)437. In some species, coordination of TFP elongation and retraction with external stimuli allows phototaxis438.

So far, in archaea, no evidence has been found that TFP have the ability to retract. Yet, a swimming motion dependent on the rotation of the archaellum is widespread in archaeal species231,439. Just like for the bacterial flagellum, the rotation direction of the archaellum can be switched to determine swimming direction and speed (Figure 25f). This was first observed in Halobacterium salinarum where the direction of rotation of the archaellum is inverted following blue light emission to permit avoidance of this blue light by archaea237. The archaellum can also be coupled to a chemosensory transduction pathway, allowing archaea to swim towards attractants and away from repellents440,441. Archaellum are very good propellers and archaea can reach speeds of up to 0.5mm.s-1 235. Just like TFP, archaella are not only involved in motion but also in surface adhesion and biofilm formation427.

53

Figure 25: Diverse motility phenotypes can be achieved by TFF-bearing prokaryotes. Summary of the different forms of motility that can be adopted using TFF. For each species, typical tracks are shown on the left and the direction of the movement is indicated by an arrow. The forces generating this motion are depicted on the middle right column and where necessary, the mechanism is detailed in the text on the right. Adapted from Maier and Wong 2015 and Kinosita et al. 2016.

54 2.2.3 TFF allow the formation of multicellular communities through aggregation

The presence of TFF at the cell surface has also been shown to promote intercellular contacts. Aggregate formation has been documented both in bacteria and archaea. While the possibility for TFP to interact with each other has been discovered long ago in bacteria442, this capacity was only recently demonstrated for TFP and archaella in archaea241,289. How TFF mediate these interbacterial interactions is still unclear. TFP and archaella can form bundles of filaments through parallel interactions241,442. In several species, it has been observed that loss of the bundling capacity was always accompanied by a loss of the aggregative capacity, but the opposite is not true443,444. A two-step mechanism of aggregation has been proposed to explain the previous observations (Figure 26c-e). It proposes that TFF first form bundles at the surface of separate cells and then, bundles from different cells coil around one another to initiate interbacterial interaction445. Yet, this model remains highly speculative and would require further support. Aggregation was also shown to be regulated by the number of filaments present at the cell surface323. This is consistent with the proposition that the aggregation function is directly mediated by the major pilin.

Moreover, many variants of the major pilin leading to a loss of aggregation have been isolated. Most of these variants result in modifications of the pilus surface charge, which could explain the observed loss of interaction418,443,444,446. The observation that different species can form co- aggregates using their TFP or their archaella (Figure 26a,b) is also in favor of a partial conservation of the mechanism behind TFF-mediated aggregation447,448.

Observation of inter-species co-aggregates also questions the role of aggregate formation. Indeed, aggregation has traditionally been viewed as a means for cells to efficiently and durably colonize the environment. The importance of aggregation has been demonstrated in bacterial pathogenicity 443,449. This does not explain the advantage there could be for prokaryotes to form inter-species aggregates. A possibility is that these interactions, by bringing cells in close contact, favor DNA transfer as was observed between Neisseria family members (Figure 26b)447. Such conclusions are further supported by the observation of facilitated intra-species DNA transfer in the presence of TFP: through cell mating in E. coli 450and through UV-induced TFP-dependent aggregation in hyperthermophilic archaea289,451. Cell aggregation could therefore facilitate DNA exchange and DNA repair between individuals, thereby increasing the fitness of the bacterial population.

Aggregation can also eventually lead to biofilm formation, thus also protecting the bacterial community from environmental hazards 452. Interestingly, other TFF: the T2SS are involved in the buildup of this protective through their protein secretion activity 141,142.

55 Figure 26: TFF-mediated aggregation. (A) Scanning electron micrograph of a coculture of the archaea Methanocaldococcus villosus (black arrows) and Pyrococcus furiosus (white arrows) grown on glassy carbon. Archaea form mixed-species biofilm through archaellum-dependent interactions. Scale bar: 5µm. (B) Scanning electron micrograph showing the interaction of an aggregate of N. elongata cells (on the left) and an aggregate of N. gonorrhoeae cells (on the right). Note the presence of a few N. elongata cells in close association with the gonococcal aggregate. Cells were grown for 3hours on epithelial cells. Scale bar: 1µm. (C-D) Scanning electron micrograph of V. cholerae cells forming aggregates through TFP-dependent interactions. Bundles and superbundles between TFP of different cells can be observed (white arrows). (E) Proposed three-step mechanism for TFP-dependent aggregation.

Adapted from Higashi et al. (2011), Jude and Taylor (2011) and Weiner et al. (2012).

2.2.4 TFF allow selective protein secretion

The main function of T2SS is the secretion of folded proteins. At least two models (Figure 27) have been proposed to explain protein secretion: the piston model, where polymerization of the pseudopilus would push the substrate through the secretin upon successive cycles of elongation and disassembly, and, the Archimede’s screw model which predicts a continuous secretion mediated through pseudopilus rotational assembly combined to a continuous incorporation of the substrate. The mechanism linking pseudopilus polymerization and protein

56 secretion is still unclear360,453. Mechanistic details explaining substrate specificity of this machinery are also lacking.

TFP have also been involved in protein secretion. Preliminary evidence of their role was observed in P. aeruginosa where the deletion of the major type IV pilin gene negatively impacted protein secretion by the T2SS 421. This observation provides further support to the common origin of T2SS and TFP and their shared mechanism of assembly. Since then, other examples of protein secretion by TFP were described. When the major pilin of the sheep pathogen Dichelobacter nodosus was deleted, it lost both its pili and its ability to secrete a protease454. This was also observed in the human pathogen Francisella novicida455. In V. cholerae the secretion of a encoded in the TFP operon was also lost upon deletion of the major pilin and secretion of this factor was shown to be independent of the action of the T2SS 456. These observations suggest that both types of TFF can mediate protein secretion separately. Yet, the interactions documented between components of the T2SS and TFP machineries have functional effects on secretion and suggest overlapping mechanisms in protein secretion between these two systems421,425.

Figure 27: Proposed mechanisms for protein secretion. (A) Assembly of the pseudopilus is initiated by the interaction of the minor pilin complex PilHIJK, the assembly platform PilGMNOP and recruitment of the ATPase PilF. (B) The piston model of secretion predicts that the pseudopilus pushes the secreted protein through the secretin by successive cycles of elongation and retraction (through an uncharacterized mechanism of disassembly). (C). The Archimedes’ screw model predicts a rotational elongation of the pseudopilus with continuous incorporation of the secreted proteins in or around the pseudopilus. Pseudopilus is expected to disassemble when it reaches the extracellular milieu. This could be due to the small concentrations of Ca2+ found in the extracellular environment.

Adapted from Campos et al. 2013

57 2.2.5 TFF generate genetic diversity by providing transformation competence

Not only can TFF mediate protein export but they can also import DNA. Transformation competence is the ability for a cell to take up exogenous DNA and incorporate it in its genome. This ability is advantageous for several reasons: DNA can be used as a nutrient, to repair DNA and to acquire new traits. This natural competence can either be regulated by environmental factors or can be unregulated457. In the latter case, mechanisms involving specific DNA sequence recognition are used to ensure preferential take up of DNA from the same species 458. TFP and competence pseudopili are involved in the initial phases of DNA capture. Some TFP bind DNA and directly promote its uptake upon retraction inside the cells459while, for others, no DNA binding activity has been detected but DNA uptake is still dependent on pilus retraction. Natural competence is present in many prokaryotes that do not express TFP and has been proposed to be mediated by so-called competence pseudopili460. Just like T2SS and based on genetic studies, these structures are thought to only span the periplasm in diderm bacteria and the cell wall in monoderm bacteria. Evidence for the existence of such is very recent and was provided by observation of TFP-like structures binding DNA at the surface of S. pneumoniae461. There are currently two competing models to explain DNA uptake462: a retraction-dependent model where long TFP are required in order to bind and bring back DNA in the cell and a plug model where the pilus would only be required to open the cell wall or the secretin for DNA to penetrate the cell (Figure 28). This new member of the TFF family further blurs the frontiers between pili and pseudopili and provides additional support to the unity of the TFF family.

Figure 28: Proposed models for DNA uptake. These models are illustrated in a monoderm bacterium (A) Retraction-dependent model. In this model, long TFP are required to initiate DNA binding. This weak interaction, facilitates higher affinity interaction the DNA with components of the competence machinery upon pilus retraction, eventually leading to the entry of single strand DNA in the bacterial cytoplasm. (B) An alternative hole in the wall model proposes that the pilus only passively helps bringing DNA near competence machinery by forming a channel through the thick peptidoglycan layer in monoderm bacteria or by keeping the secretin in an open conformation in diderm bacteria. Adapted from Muschiol et al. 2015

58 2.2.6 TFF can be hijacked by phages

This ability to import DNA has been hijacked by phages which have evolved the capacity to bind TFP of many bacterial species and subsequently infect them304,463–466. Phages were observed to bind at the tip or sides of the pilus, suggesting that they bind the major pilin (Figure 29a,b). Upon binding, TFP must be retracted for phage infection to occur466 . It was also observed that phage-resistant bacteria were non-motile. This resulted in the understanding that twitching motility is dependent on TFP retraction467. Following TFP retraction, viral DNA is injected inside the bacterial host through an unknown mechanism (Figure 29c). Pilin glycosylation was recently described as an efficient defense mechanism evolved by some P. aeruginosa strains to prevent phage binding to the major pilin, nicely illustrating the perpetual arms race happening between bacteria and phages419. Only one example of binding to an archaeal filament has been described so far468. While these filaments were not formally identified in this study, Sulfolobus islandicus, the infected species is known to express UV- inducible TFP and likely represents the first example of virus binding to archaeal TFP469. To this date, virus binding has not been described for the other types of TFF.

Figure 29: Phage binding to Type IV pili. (A) Electron micrographs showing CBK phage binding by their tail to the sides of the Caulobacter crescentus pilus (white arrow). Scale bar: 120nm. Top: After 15s of infection. Bottom: After 15min of infection, pili retraction has brought CDK in close contact with the cell. (B) Similar observations were made for P. aeruginosa and the PB1 phage which binds TFP. (C) A simple model showing that upon phage binding, retraction of TFP allows injection of the page genome through the cell envelope by mechanisms that are still poorly characterized.

Adapted from Skerker and Shapiro (2000) and Bradley and Pitt (1974).

59 2.2.7 TFF can act as nanowires to allow extracellular respiration

Another unexpected TFP-specific function was discovered more recently.

In 1988, lake Oneida, one of the largest freshwater lake in the state of New-York became a source of interest for a group of scientists who noticed that it contained abnormally high levels of reduced Mn2+. Because manganese is usually found in its insoluble oxidized form in the environment (Mn4+), researcher hypothesized that a novel metabolic process could be explaining why they observed such high levels of reduced manganese. This led to the discovery that Shewanella onedensis was able to mediate extracellular respiration through the reduction of various metal oxides470. A similar observation was made in the bacterium Geobacter sulfurreducens which uses Fe(III) oxides as an electron acceptor in anaerobic conditions471.

Reduction of these extracellular oxides was then shown to be dependent on the presence of filaments protruding from the cell surface and contacting the insoluble oxides (Figure 30a). These appendages were named microbial nanowires305,472. Surprisingly, at least three types of microbial nanowires can be differentiated473. While for Shewanella onedensis microbial nanowires are composed of cytochrome-enriched membrane protrusions474, in G. sulfurreducens it is TFP that make up the microbial nanowires responsible for extracellular respiration305. TFP were also found to be responsible for the extracellular respiration in the bacterium Synechocystis sp. PCC 6803475. These findings have been confirmed by conductivity measurements of purified TFP and by direct visualization of electron transfer through these TFP476,477 (Figure 30c). These experiments reveal metallic-like conductivity of TFP.

The molecular mechanism allowing electron transport has also been partially deciphered and is believed to rely on electron transfer between successive aromatic amino acids through pi-pi stacking interactions477,478 (Figure 30b). A remarkable feature of the major pilin from G. sulfurreducens is that it is extremely short (only 61 amino acids) and enriched in aromatic amino acids in comparison to pilin from other model bacteria. Its structure was elucidated and shows a classical pilin fold408. Structure of the pilus has been modelled and revealed a dense packing of aromatic amino acids inside the fiber. Two models are competing concerning the precise mechanism of electron transport inside the fiber but both were obtained from sparse data and agree on the role of aromatic amino acids in electron transfer 479,480. Interestingly, the pilin of Synechocystis sp. PCC 6803 which also has conductive TFP is as long as pilins from non- conductive organisms (148 amino acid-long) but shows significant enrichment in aromatic amino acids in the C-terminal part of the protein475. This indicates, that having a short pilin is not a prerequisite to assemble conductive pili. More structural and mutagenesis efforts are required to determine how electron transport occurs inside TFP of these bacteria and how widespread these conductive properties are in TFF.

Despite the novelty of this function, yet another function of TFP was elucidated even more recently.

60 Figure 30: Type IV pili as nanowires. (A) Transmission electron micrographs showing the tight interaction between TFP of Geobacter sulfurreducens (red arrows) and crystalline Fe(III) oxides. Scale bar: 500 nm. (B) Atomic model of the Geobacter sulfurreducens pilus obtained by energy minimization. This model predicts very short distances (<5 angströms) between the different aromatic amino acids of the pilus. This could account for electron-transport by overlapping pi-pi orbitals which is consistent with the metallic-like conductivity observed in these TFP. (C) Visualization of charge propagation along TFP of Geobacter sulfurreducens using electrostatic force microscopy. Charge injection in the pilus (red arrow) is visualized by an increase in phase shift (in white on the figure). Progressive charge dissipation is visible at 80minutes. Scale bar: 100nm. Adapted from Reguera et al. (2005), Malvankar et al. (2014) and Xiao et al. (2016).

2.2.8 TFF can enable surface sensing by mechanotransduction

Because TFF, are involved in surface exploration and play a major role in the lifestyle of prokaryotes (through surface adhesion, aggregation and even biofilm formation), it was postulated as early as 1935 that prokaryotes must have ways to identify that they are approaching a surface481,482. Recently, TFP were shown to mediate such surface sensing in P. aeruginosa via mechanotransduction. A series of publication demonstrated that TFP-mediated contact triggers signaling pathways that eventually lead to increased cAMP and c-diGMP levels. This increase in secondary messengers results in upregulation of the expression of virulence factors (including TFP) which allows surface colonization483–485(Figure 31b).

61 This surface sensing is dependent on retraction of surface-attached TFP and was also shown to be dependent on several components of the TFP machinery, including PilY1 (PilC1 in N. meningitidis)483–485(Figure 31a). The mechanical signal that is detected is still unknown but, when under tension, TFP have been shown to deform486,487 and this morphological alteration of TFP has been proposed to be the signal that was detetced484. Surface sensing through pilus retraction was also recently described in the Tad-pilus expressing Caulobacter crescentus302 suggesting that such mechanisms may be widespread in TFF-expressing organisms.

Figure 31: Type IV pili as mechanosensors. (A) Analogy between the von Willebrand factor(vWF) secreetd by endothelial cells and the P. aeruginosa PilY1 protein. When a blood vessel is injured, shear of the blood flow is deforms the vWF and results in coagulation through platelet recruitment. Similarly, PilY1 is thought to detect mechanical stimuli upon cell adhesion and triggers pathogenic program in P. aeruginosa. (B) Model recapitulating the actors involved in the production of secondary messengers for surface detection (a mechanical stimulus? ). These signals first lead to an increase in cAMP levels which results in an increased transcription of the PilH-PilV operon (FimU-PilE in P. aeruginosa). Through the action of PilC1, this leads to an increase in c-di-GMP levels which eventually leads to biofilm formation.

Adapted from Ellison and Brun (2015) and O’Toole and Wong (2016).

Because of the functional variability that exists between TFF, we decided to focus our study on TFP, for which a role has been found in all of the previously described functions.

62 3 Type IV pili of Neisseria meningitidis: a case study

In order to understand the structure-function relationships of TFP, we decided to use the pathogen Neisseria meningitidis as a model system. This model is particularly interesting for at least five reasons. First, meningococcal TFP have been shown to be necessary for virulence of the bacterium in vivo488,489 and understanding how to block pili biogenesis or functions could help cure infections. Second, TFP are the only appendages present at the surface of the bacterium which makes it a relatively simple model. Third, the TFP of N. meningitidis are extremely versatile which makes them an ideal tool to study the different functions on a single structure. Fourth, these pili have been very well studied and their structure has recently been elucidated, which greatly contributes to our understanding of their functions 307. Fifth, N. meningitidis is a fast-growing organism that can easily be genetically modified. 3.1 A human obligate pathogen 3.1.1 The Neisseriaceae family: a diversity of commensal bacteria

The bacterium N. meningitidis or meningococcus is a member of the Neisseria genus. This genus is composed of diderm ß-proteobacteria with mainly coccoid and less frequently rod- shaped morphologies490 (Figure 32a). The Neisseria genus is part of the Neisseriaceae family which also includes the Kingella and Eikenella genera. The Neisseria genus counts two major These genera are composed of short pathogens: Neisseria meningitidis, the causative agent of meningococcal disease and Neisseria bacilli that are gonorrhoeae, the causative agent of gonorrhea. Other members of the genus are less virulent usually found as commensals of the and only rarely found as opportunistic pathogens in humans. While the two species of nasopharynx an can pathogenic Neisseria are specific to their human host, commensal Neisseria are found in sometimes act as opportunistic surprisingly diverse environments: mammalian and non-mammalian hosts and even on rare pathogens in occasions in the environment. immunocompro- mised individuals.

Figure 32: Diversity of the Neisseria genus. (A) Scanning electron micrographs of diverse members of the Neisseria genus reveals a variety of cellular shapes (from coccoid to rod-shaped). Scale bars: 2µm (four leftmost panels) and 1µm (four rightmost panels). (B) Phylogenetic tree of human-specific Neisseria species based on the comparison of a subset of 896 genes. Adapted from Liu et al. 2015 and Marri et al. 2010

63 These bacteria have been found in various organs and tissues in their different hosts. In the human host, commensal Neisseria are only found in the mouth and nasopharynx490. This is in contrast to the two pathogenic diplococci N. gonorrhoeae and N. meningitidis. N. gonorrhoeae or gonococcus is also found in the human urogenital tract and rectum and is usually transmitted during sexual activity. According to the World Health Organization, around 78 million new infections occur every year. There is growing concern about gonorrhea because of the rapid rise of antibiotic resistance in gonococcus491.

3.1.2 Meningococcus has a high carriage rate N. meningitidis is an opportunistic pathogen mainly found as a commensal in the nasopharynx where it adheres to the epithelium492. In North America and Europe, it is estimated that 10 to 35% of the population carries the bacterium493. Carriage rates are extremely variable depending on the context and age. Age was found to be a central variable in meningococcal carriage. Rates increase from infancy (around 5%) to peak around the age of 19 (around 24%) and decreases through adulthood to reach about 8% in 50 years old494 (Figure 33a). The higher meningococcal prevalence in teenagers is thought to be associated with their social behavior as it was reported that they tend to have more social contacts than other age classes495. Transmission of the bacterium happens through close contact via exchange of respiratory and throat secretions (typically coughing or kissing).

3.1.3 Meningococcal disease: a rare but deadly disease

While carriage rates are quite high, meningococcal disease is relatively rare, with rates between 0.2 to 1000 cases per 100,000 individuals and an estimated 1.2 million cases of meningococcal diseases each year496. No direct link has been found between carriage rates and onset of the disease. These rates greatly vary with the geographical area and the age group studied. In Europe and North America, these rates are always below 1 per 100,000 individuals. In these regions, children under 5 are the most susceptible to meningococcal disease while they only carry the bacterium at a low rate (Figure 33b). Rate of meningococcal disease can also dramatically increase during epidemics. In the past decades, such epidemics have mostly been observed in sub-Saharan Africa (in the so-called “meningitis belt”) with rates between 10 to 1000 cases per 100,000 individuals and the highest incidence in children under the age of 12493,497 (Figure 33b,c).

64 Figure 33: Epidemiology of meningococcal disease. (A) Meningococcal carriage in the Europe as a function of age. Each circle is a data point and circle radius is proportional to the sample size. (B) Epidemiology of meningococcus by region. In Europe, infants are the most-at-risk population while in Africa (data shown for Burkina Faso here), children remain at risk until they adolescence. (C) World map showing the geographical distribution of the different meningococcal serotypes. The “meningitidis belt” is indicated by the dotted line. Purple areas show most recent meningococcal outbreaks and associated serotypes. Adapted from Christensen et al. (2010), Pelton (2016) and Stephens et al. (2007).

65 Different bacterial serogroups are found in the different regions of the world. Thirteen N. meningitidis serogroups have been defined on the basis of the composition of their capsule498. Five serogroups are responsible for 90% of the cases of disease: serogroups A, B, C, W135 and Y499. Serogroups A, W135 and X are responsible for most epidemics in sub-Saharan Africa while the serogroup B is the biggest concern in Europe and America500,501. Serogroup Y recently emerged in North America and serogroup C is often responsible for disease in young adults all over the planet502,503 (Figure 33c). have now been developed for all these serogroups. They are all based on capsule polysaccharides except for serogroup B which has polysaccharides with low immunogenicity due to their resemblance to the polysialic acids of human neurons504. The most recently developed vaccines for serogroup B were licensed in 2013 and May 2017 and both show promising results505,506. Yet, the mortality rate of meningococcal disease remains elevated with 10-15% fatal cases. Morbidity is also elevated with 10% of the nonfatal cases requiring amputation or leading to deafness or severe cognitive decline507.

In individuals with meningococcal disease, the bacterium somehow crosses the nasopharyngeal epithelium to reach the bloodstream. Why and how this happens remains unclear. Possible risk factors include damages to the respiratory mucosa possibly following infection by the influenza virus508, during winter because of dry air507 and because of the Harmattan dry and dusty wind in the meningitis belt509. This could help the bacterium cross the epithelial barrier and reach the bloodstream where it will trigger invasive meningococcal disease. Another risk factor associated to the blood phase of the bacterium is a deficiency in Complement: circulating proteins of the complement system which results in reduced clearance of bacteria in the proteins that blood510,511. Upon extracellular bacterial division and dissemination, the disease can evolve participate in innate immunity towards two different pathologies (that can also be combined): meningitis and by boosting meningococcemia (Figure 34). Meningitis is the most common form of disease following phagocytosis, inflammation and infection. Meningococcal meningitis accounts for 50% of bacterial meningitis in children in bacterial killing France512. In meningitis, the bacterium manages to cross the blood brain barrier and upon activation proliferates in the cerebrospinal fluid which will result in a severe inflammation of the meninges Meninges: membranes upon invasion. This invasion is thought to be mediated by interaction with the bacterium surrounding the resulting in the opening of the tight junctions between endothelial cells513 (Figure 45b). brain and forming a barrier between In meningococcemia, upon attachment to blood vessels514 (probably capillaries), bacteria the blood circulation and divide in the blood stream eventually filling up the blood vessels as observed in post-mortem the cerebrospinal samples515. This uncontrolled bacterial growth can result in systemic inflammation which leads fluid. to important vascular damages (coagulation, blood leakage…). In patients, hemorrhagic skin lesions called purpura fulminans are often observed as a consequence of vascular damage. In the worst cases, vascular dysfunction can result in disseminated intravascular coagulation, drop of blood pressure accompanied with organ hypo-perfusion and eventually organ failure or ischemic extremities resulting in limb amputation. The course of progression of the disease can be as rapid as 12 hours between the apparition of the first rash and death516. Diagnosis and

66 treatment of meningococcemia must be very rapid to increase chances of survival. How meningococcus triggers this deadly disease at the molecular level is the focus of the next section.

Figure 34: Development of meningococcal disease. N. meningitidis is an airborne pathogen. In the nasopharynx it can invade the epithelium through several mechanisms. Once it has traversed this barrier, N. meningitidis can invade the blood vessels where it triggers massive inflammation. This can result in sever vascular damages and meningococcal disease manifesting as meningitis and/or septicemia. ECM: extracellular matrix. Adapted from Virji (2009).

3.2 Virulence of Neisseria meningitidis

N. meningitidis and N. gonorrhoeae are the only two pathogenic members of the Neisseria genus and are closely related to the commensal N. lactamica. (Figure 32b) Comparisons between whole genomes of virulent species and commensal species was expected to reveal the virulence factors involved in meningococcal pathogenicity. Yet, no known virulence factors (defined as specific factors favoring the pathogenicity of a bacterium) were identified using these approaches517,518. An additional intraspecific genetic variability exists in N. meningitidis. Indeed, non-invasive (carriage) strains can be separated from invasive strains by genome comparison.

3.2.1 Hyperinvasive lineages: a few clonal complexes cause most disease cases

The low ratio between the low rates of disease and the very high rates of carriage indicates that some bacteria may carry specific factors that facilitate invasion of the host. In 1998, a method using MultiLocus Sequence Typing (MLST) i.e. determining the nucleotide sequences of the loci of seven housekeeping genes was used to characterize isolates of N. meningitidis. This method allowed to classify meningococcal isolates by sequence types (ST) and revealed that all the hyperinvasive/hypervirulent lineages clustered in closely related ST519. These

67 hyperinvasive lineages are involved in a much higher disease rate than their carriage rate would predict. With the advent of Next Generation Sequencing (NGS), about 5,000 meningococcal whole genomes are now available and many more isolates have been sequenced520. As of the 14th of February 2018, 44,855 isolates of Neisseria spp. have been sequenced and form 13,481 distinct ST. These ST are grouped in 53 clonal complexes (CC) based on the fact that they share at least 4 similar alleles among the 7 loci that are sequenced in MLST. Again, only a few clonal complexes form hyperinvasive lineages that are associated with invasion of the host502,521. The phylogenetic proximity of these hyperinvasive bacteria led to the hypothesis that they could possibly share specific genetic determinants of virulence. Through comparison of the whole genomes of hyperinvasive lineages and non-invasive lineages, a single 8kb-long region was identified as significantly associated with hypervirulent lineages522. This region encodes a functional filamentous bacteriophage that is secreted through the TFP secretin and can form micrometer-long filament at the surface of bacteria. This phage can infect N. meningitidis strains in a TFP-dependent manner and uses the major pilin as a receptor523. It does not seem to provide an advantage in the bloodstream, but it is involved in promoting interbacterial aggregation and thus increases colonization of the epithelium. The authors of this study propose that this increased epithelial colonization increases the chances of bacterial invasion and thus of disease464. Whether this is true remains to be demonstrated.

3.2.2 Multiple surface structures involved in infection

Neisseria meningitidis also possesses many classical surface structures which are not restricted to hyperinvasive lineages but have been shown to be important for infection.

3.2.2.1 The protective capsule

Unlike gonococcus, meningococcus can be encapsulated. Almost all of the hyperinvasive isolates are found to be encapsulated while a majority of carriage isolates are non- capsulated524. Neisseria meningitidis probably acquired its capsule through horizontal gene transfer from Haemophilus influenza or Pasteurella multilocida525,526. The capsule forms the outermost layer of the bacterium and is formed of lipid-anchored polysaccharides pointing away from the bacterial surface. For all the serogroup except serogroup A, it is composed of polysaccharides that derive from sialic acid (Figure 35a). N-acetylneuraminic acid is incorporated in the capsule of N. meningitidis and helps dampen the immune reaction upon invasion as it is also commonly expressed at the surface of human cells527,528. The capsule forms a hydrated gel around the bacterium with a thickness estimated around 100nm (Figure 35b) that was found to protect the bacterium against complement-mediated killing following antibody recognition in the blood529. This is further supported by the increased infection susceptibility of individuals with deficiencies in their complement system and supports its implication in bacterial virulence510,511. Genes required for capsule synthesis and surface translocation are clustered in a single region of the chromosome named cps which is divided

68 in 6 regions: A which encodes genes for capsule synthesis and polymerization, C for capsule transport, B for capsule translocation, D and D’ for lipooligosaccharide synthesis and finally E for the tex gene which is a putative transcription factor498,530. Thanks to their natural competence, meningococci are able to exchange parts of their cps locus with meningococci from a different serogroup in a phenomenon called capsule switching531. This represents another mean for the bacterium to escape immunity and has been observed during outbreaks in 2000-2003 where a serogroup C strain from a hypervirulent clonal complex switched capsule with a bacterium from serogroup W-135 leading to epidemics caused by these new serogroup W-135 bacteria532–534. The protective capsule is necessary for blood survival of meningococci but not sufficient for its pathogenic phase. The outer membrane of the bacterium where the capsule is anchored also presents additional virulence factors.

Figure 35: Cell envelope of Neisseria meningitidis. (A) Representation of the outermost layer of the outer membrane. N. meningitidis has a thick capsule made of a polymeric assembly of N-acetylneuraminic acid (NeuAc) anchored to membrane phospholipids. Lipooligosaccharide (LOS) is also found at the surface of the bacterium and is anchored to lipid A. Its structure is illustrated on the figure. Hep: Heptose, Kdo: Ketodeoxyoctonic acid, Glc: Glucose, GlcNac: N-acetylglucosamine, Gal: Galactose. Illustration from Arthur Charles-Orszag. (B) Meningococcal polysaccharide capsule observed by transmission electron microscopy. Images were taken on bacteria of the serogroup W, strain ATCC 35559. Adapted from Ganesh et al. 2017.

69 3.2.2.2 The pro-inflammatory lipooligosaccharide

In diderm bacteria, a major virulence factor of the outer membrane is the endotoxin or (LPS). These glycolipids are the major lipid components of the outer leaflet of the outer membrane. In Neisseria meningitidis, just like in Haemophilus influenzae the endotoxin is short because it lacks the long polysaccharidic O-antigen that usually decorates the LPS and is thus called the lipooligosaccharide (LOS). This LOS is composed of three units: lipid A that anchors the LOS to the membrane, lipid A is itself connected to a variable core oligosaccharide containing heptoses that serve as a linkage for highly variable short oligosaccharides chains535 (Figure 35a) . Variations in structure and composition of the LOS can be used for classification and twelve different immunotypes have been differentiated (L1- L12)536,537. Non-sialylated shorter LOS of the L1, L8 and L10 immunotypes are more common in non-invasive strains while sialylated L3,7,9 immunotypes are almost always associated to invasive strains538,539. This suggests a role for the LOS in meningococcal virulence. Activation of the immune system by endotoxins has been observed and described for some time540. The LOS of N. meningitidis binds several molecules present at the surface of human cells. In invasive strains rather than carriage strains, this results in the activation of the TLR-4 pathway541,542 and is followed by secretion of various pro-inflammatory cytokines (including TNF-α, IL-1 and IL- 6)543,544. Circulating levels of LOS are good predictors of the severity of the disease. Very high LOS levels result in an uncontrolled inflammatory response which triggers vascular damages and can eventually lead to septic shock545,546. The LOS, as the most potent pro-inflammatory molecule expressed by N. meningitidis has a pivotal role in the pathobiology of meningococcal infections. But, in order to promote inflammation during infection, bacteria first have to invade and multiply in their vascular niche547. This requires specific metabolic adaptations.

3.2.2.3 Metabolic adaptations

Neisseria meningitidis is a fastidious bacterium. It is a strict aerobe which requires 5-7% of CO2, a carbon source (glucose, pyruvate or lactate) and reduced sulfur (cysteine, cystine or thiosulfate) to grow548–550.

Another crucial element required for growth is iron551–553. Yet, in the human blood, iron is sequestered by various proteins and unavailable for potential pathogens554,555 and has been shown to be a limiting factor for meningococcal invasion553,556,557. N. meningitidis has evolved several transporters and iron-binding proteins to acquire iron from its human host during the blood phase558. The bacterium can scavenge iron from extracellular hemoglobin (usually located inside red blood cells) and from haptoglobin-hemoglobin complexes559 but also from transferrin, an iron-sequestering protein found circulating in the serum560 and lactoferrin, a transferrin protein found in phagocytic cells and secretions such as mucus558,561. Neisseria meningitidis uses protein that specifically bind human iron-containing proteins. The primary iron source is obtained through scavenging of the iron contained in transferrins via a complex formed by two outer membrane transferrin-binding proteins TbpA and TbpB that form the

70 transferrin receptor562 as well as the lactoferrin receptor which is formed by two outer membrane lactoferrin-binding proteins LbpA and LbpB563. With these proteins, iron is imported in the periplasm and is finally translocated to the cytoplasm through the Ferric- binding protein (Fbp) transporter564. In addition to a requirement for iron, it has also been suggested that meningococcus has a requirement for zinc that it can scavenge from the human calprotectin565,566.

In the vascular niche, meningococcus grows as dense microcolonies that are likely under nutrient limitation and oxidative stress. The different environments encountered by the bacterium presumably result in bacterial adaptations. While it has been suggested that carbon metabolism and the oxidative stress response could be affected during infection, evidence in vivo is still lacking567. Such studies are starting to emerge547 through the use of the recently- developed humanized mouse model of infection488,489 and should help get a more accurate view of the metabolic pathways needed during infection.

3.2.2.4 Several adhesins contribute to colonization of the human host

In order to colonize their host, bacteria need to colonize their niche and maintain it. Bacteria have evolved many systems to adhere to host cells568. For carriage, meningococcus first needs to adhere to the nasopharyngeal epithelium. During the invasion and later on during the blood phase or cerebrospinal fluid invasion, bacteria need to adhere to endothelial cells. Meningococcus has evolved several mechanisms to adhere to human cells exclusively. Most of its adhesins are presented in the following section (Figure 36).

Figure 36: Main adhesins used by N. meningitidis and corresponding cell receptors. Long-distance interactions are initiated by TFP. Upon retraction, the other meningococcal adhesins (partially covered by the capsule) can interact with their cellular receptors. The ß2-Adrenergic receptor (ß2AR) does not directly bind pili but is required for stable TFP- dependent adhesion. From Coureuil et al. (2017).

¬ Minor adhesins Several outer membrane proteins have been shown to promote bacterial adhesion to human cells in the past decade. Their importance in pathogenesis is ill-defined. In vitro, these adhesins are expressed at relatively low levels compared to the major adhesins (see below). Minor

71 adhesins include: the Adhesion and penetration protein (App)569,570 and the Neisserial adhesin A (NadA)571,572 that bind epithelial but not endothelial cells, the Neisseria hia homologue A (NhhA) which binds epithelial cells and both laminin and heparan sulfate of the extracellular matrix573,574 and the meningococcal serine protease A (MspA) which binds both endothelial and epithelial cells575. These adhesins appear to only have a minor role in adhesion as compared to the following major adhesins.

¬ Opacity proteins Opacity proteins were first discovered in Neisseria gonorrhoeae. Two types of colonies could be observed on plates: light-colored transparent ones and opaque ones. This difference is due to the expression of opacity proteins by the opaque colonies576. Two distinct opacity proteins can be expressed by Neisseria meningitidis: Opa and Opc. Both proteins are major adhesins for adhesion to endothelial and epithelial cells577,578.

Opa proteins bind members of the CEACAM (carcinoembryonic antigen-related cell adhesion molecule) family on human cells579. To mediate adhesion, significant levels of expression of CEACAM are required. Such levels can be reached upon vascular inflammation580 and probably facilitate adhesion of meningococcus581,582. Under non-inflammatory conditions, Opa can still promote cellular invasion through its interaction with CEACAM-1582,583.

The partners of Opc are less clearly defined and recombinant expression of Opc at the surface of E. coli is not sufficient to induce adhesion to endothelial cells584. It is currently thought that Opc indirectly binds integrins at the surface of the cell. This indirect binding would result from a primary binding of Opc to serum-circulating vitronectin or fibronectin. Bound vitronectin/integrin would then bind to their integrin receptors at the cell surface584–586. Opc could also bind to vitronectin/integrin directly in the extracellular matrix. Both Opa and Opc can also bind heparan sulfate proteoglycans of the extracellular matrix thus providing further support for the implication of the extracellular matrix in bacterial binding587,588.

Both opacity proteins are ß-barrel integral outer membrane proteins588,589 subject to phase variation (Figure 37a). In addition, Opa is subject to (this process is described in the next section on TFP and in Figure 37b)590,591. Because of their size and outer membrane location, opacity proteins are masked by the capsule in capsulated bacteria. In capsulated bacteria and bacteria with a sialylated LOS, opacity proteins are no longer sufficient to mediate adhesion578 . This is probably due to the repulsion generated between the negative charge of the bacterial capsule and the negative charge of the human cell membrane. To overcome this repulsion, long distance interactions must be mediated by the other major adhesin of meningococcus: their type IV pili578,592.

¬ Type IV pili: long distance adhesins The adhesion of N. meningitidis just like the adhesion of N. gonorrhoeae is favored by the presence of TFP. All invasive meningococcus isolates collected from patients appear to be

72 piliated. Piliation was found to favor adhesion to organ explants and both cultures of human epithelial and endothelial cells593,594. The role of pili in the adhesion process was initially confirmed by the observation that adhesion was affected by changes in the amino acid sequence of the major pilin444,595,596. Subsequent work showed that pili-dependent adhesion also relies on the expression of other proteins of the piliation machinery and several pilus receptors have since then been proposed592,597–600. An overview of the current state of knowledge regarding pilus-dependent adhesion is provided in Section 3.3.3.5. Because TFP are necessary for virulence and exposed at the bacterial surface, meningococcus has also evolved mechanisms to avoid recognition of TFP by the immune system. In both N. meningitidis and N. gonorrhoeae, pilin and other virulence factors have very variable sequences. This high sequence variability is generated by a mechanism of antigenic variation601 that has been extensively studied602,603. In brief, in addition to the expressed major pilin locus (pilE), additional silent cassettes (pilS) with homology to the 3’ of pilE are found in the genome. Recombination can occur spontaneously between one or several of the pilS cassettes and the pilE locus (Figure 37b). These recombination events result in the expression of a pilin with a new C-terminal sequence thus providing a mean to escape immune recognition while maintaining piliation.

A specificity of N. meningitidis is that it expresses two classes of major pilins. Class I pilins are homologous to those of N. gonorrhoeae and they are all recognized by the monoclonal antibody SM1604,605 while class II pilins are not. Both classes of pilin have different genomic locations606 and can both mediate high adhesion to endothelial cells and moderate adhesion to epithelial cells but to a lesser extent for class II pili594,607. While class I pili are subject to antigenic variation, class II pili are not606,608,609. Mass spectrometry analysis of class II pilin revealed that their pilin can carry up to 5 glycosylations exposed at the surface of the pilus unlike class I pilin which only bears a single glycan413,610–613. This suggests that by covering the surface of class II pili, glycans may prevent antibody recognition of TFP. Because of the fundamental differences between these two classes of pili, it is possible that they mediate cell- adhesion via different mechanisms and this should be investigated. Evaluating the relative importance of these virulence factors during the course of infection and discovering new ones should be made easier using appropriate (animal) models.

73 Figure 37: Mechanisms of protein variation in N. meningitidis. (A) Phase variation events happen during DNA replication or repair. As exemplified, in the case of nadA, tandem repeats are found in the gene promoter and can lead to the formation of a DNA loop (3) and DNA mispairing. Upon replication this leads to a shortening of the promoter region that results in a reduction in nadA transcription. Such mispairing event can also happen in the ORF of the gene and induce translational frameshift thereby resulting in the apparition of premature stop codons and precluding protein expression. (B) Antigenic variation in pilE results from the donation of partial gene sequences from the pilS silent pseudogenes. This recombination event happens in the variable part of the gene and results in virtually infinite and extensive remodeling of the gene sequence.

Adapted from Davidsen and Tønjum 2005.

3.2.3 Infection models as a tool for in vivo identification of virulence factors

Elucidating the relative importance of meningococcal virulence factor in infection has been hampered by the strict human-specificity of meningococcus. The field is still lacking an animal model recapitulating all the steps of the disease, but, significant progress has been achieved recently. First studies were only performed on human cell cultures. Subsequently, infection

74 models have been developed in the mouse. An intraperitoneal model was developed in 1933 by co- injecting bacteria and hog’s mucin614. It was later found that supplementing bacteria with iron was sufficient to kill the mice552,615 thereby helping to understand the metabolic needs of the bacterium. This model has been used to evaluate the virulence of different meningococcal strains by measuring minimal lethal doses of bacteria556. Yet, this model does not recapitulate the invasiveness or the blood stage of the bacterium but rather indicates the ability of bacteria to multiply and trigger inflammation. To evaluate invasiveness of the bacteria, a neonatal mouse model was used with intranasal instillation of bacteria in 5-day old mice616. This model was used to confirm the importance of the LOS and the capsule in host invasion539. Yet, because adhesion of meningococci is specific to human cells, these models lacked the specific interaction of bacteria with their epithelial and endothelial niches. Addressing this limitation was later attempted in a transgenic mouse model expressing human CD46617. CD46 had been identified as a putative pilus receptor at the time598. In this model, intranasal instillation of bacteria could lead to the death of adult mice and bacteria could be recovered in the blood, thus showing that this model can recapitulate invasion. Yet, invasion could only be observed on rare occasions following antibiotic treatment and the pathological features of the infection could not be observed (e.g. meningitis or loss of vascular integrity) unless if infection was performed intraperitoneally617. Another transgenic mouse model was developed more recently with mice expressing CEACAM. Expression of CEACAM was necessary for persistence of meningococcus in the nasopharynx in the absence of antibiotic treatment and this was confirmed to be dependent on the expression of the Opa protein618. Yet, no invasion could be observed suggesting that an additional human factor is necessary to cross the epithelium and transgenic mice expressing both CD46 and CEACAM could be interesting to recapitulate the disease. The only model recapitulating symptoms of the disease is the recent human skin xenograft mouse model. In this model, human blood vessels of the transplanted skin merge with mouse vessels and are eventually perfused with mouse blood. Upon intra-venous infection, this model shows typical clinical sign of vascular dysfunction in the human blood vessels. This model has shown that pili-dependent adhesion is necessary to initiate vascular colonization. Even more interestingly, the model demonstrates that inflammation and vascular damages are strictly dependent on the presence of TFP thereby showing that circulating bacteria are not sufficient to promote inflammation488,489. This validates that TFP are one of the key virulence factors in meningococcal disease and substantiate the need to understand how they mediate their various functions.

3.3 Neisserial type IV pili: linking structure and function

The focus of my doctoral work is on TFP. While we have discussed functions of TFF in Section 2.2, here, we present functional findings specific to the neisserial TFP.

75 The reference strain we used in this study is called 2C4.3 (clone 12). It is a derivative of strain 8013 isolated in the blood of a patient at Institut Pasteur. 8013 is a serogroup C clinical isolate of the ST-18 clonal complex619. This strain does not express Opa or Opc adhesins but expresses hyperadherent class I type IV pili and is capsulated595 thus making it an ideal tool to understand the role of pili-dependent adhesion and aggregation.

3.3.1 Specificities of the meningococcal machinery While the piliation machinery of type IVa pili is relatively-well conserved between bacterial species, to the best of our knowledge some components are only found in the Neisseriaceae family. Interestingly the five components described below are required both for pilus functions and assembly.

3.3.1.1 PilC-like proteins

Pathogenic Neisseria and Kingella kingae express two copies of PilC-like proteins592,620,621. This is rather uncommon as only one PilC protein is found in most systems622,623 (e.g. PilY1 in P. aeruginosa). While these two proteins were first described to localize at the tip of TFP624, they appear to mainly reside in the outer membrane625, similarly to what has been observed in other species expressing a single copy of PilC 626,627. One of the specificity of these two proteins is that they have partially redundant roles in pilus assembly. Indeed, mutants lacking only PilC1 or PilC2 still express pili at their surface while mutants lacking both proteins no longer do592,620,628. Yet, they also bear additional functions including regulation of pilus retraction 629 and regulation of adhesion to various human cellular types592,625,630,631. Unlike gonococcus, meningococcus only requires PilC1 for adhesion to most cell types. These differences in ability to mediate adhesion could stem from the fact that these proteins are expressed from separate loci and have different regulation systems. Notably, PilC1 is induced upon cell-contact while PilC2 is not. This induction was shown to be necessary to achieve proper adhesion630,632. Another possibility to explain their different functions comes from the fact that their N-terminus is quite divergent. While both proteins have around 80% sequence identity, it was shown that the N-terminus of PilC1 is important for the function of adhesion633. These differences in function added to the fact that PilC proteins undergo phase variation in vivo result in the ability for meningococcus to form heterogeneous populations of bacteria expressing either non- adherent pili or adherent pili620,634. This could be of interest to first enable colonization of the host through adhesion and then allow spreading to new body compartments (or hosts) upon loss of pili-dependent adhesion.

3.3.1.2 Minor pilins

Another notable difference of pathogenic Neisseria resides in the number of minor pilins they express. While most bacteria with TFP and T2SS only express a conserved set of four minor (pseudo)pilins that are required for pilus assembly (PilH-K in pathogenic Neisseria)318,420. N.

76 meningitidis also expresses three additional minor pilins (PilV, PilX and ComP) which are not strictly required for piliation. These minor pilins are associated with different functions: ComP is required for competence for transformation373,635,636, PilV is required for adhesion and cellular response597,637 and PilX is required for aggregation, adhesion and cellular response638,639. Because of their structural resemblance to the major pilin PilE, it was hypothesized that these minor pilins could insert in the pilus and mediate their functions from within the pilus325,599,636. Yet, a recent study in our lab established that blocking minor pilins inside of the periplasm did not affect their functions. Furthermore, this study demonstrated that pilV and pilX mutants display fewer pili at their surface. This decrease in the number of pili is sufficient to explain the functional defects observed in these mutants323. This illustrates the complexity of pilus assembly and piliation regulation in N. meningitidis. Because of the tight relationship between piliation and TFP functions, it is crucial to understand how pili assemble in order to understand how they mediate their functions.

3.3.2 Pilus biogenesis

The rest of the proteins of the piliation machinery are rather conserved but genes of the machinery are scattered through the genome, which makes it even more complex to understand their role and regulation (Figure 38).

Figure 38: Genetic organization of the meningococcal piliation machinery. All genes with a known role in TFP assembly are depicted on this chromosomal view. Color codes for functional homologies that are identified on the figure. PilS1-5 are pseudogenes that do not encode for proteins. Gene orientation is indicated by the arrow and immediately neighbor genes are nested in each other on the illustration. Distances represented on the chromosome are not to scale. Adapted from Imhaus (2013).

3.3.2.1 Role of the components of the piliation machinery While the role of many proteins of the piliation machinery is still unclear, the minimum set of proteins necessary to assemble pili in N. meningitidis is now known. The expression of eight proteins in E. coli is sufficient to assemble periplasmic TFP in the absence of the retraction ATPase PilT315. As shown in Figure 15d, these proteins form an inner membrane complex (PilMNOP) which recruits the assembly ATPase PilF and the polytopic inner membrane platform

77 PilG. This macromolecular complex is able to assemble the major pilin PilE in a filament after it has been processed by the prepilin peptidase PilD.

For the filament to exit the periplasm and become exposed at the cell surface, the secretin PilQ is required318. Multimerization of PilQ is facilitated by the PilW. The secretin is notably connected to the assembling pilus through PilP-dependent interactions319. TFP are dynamic fimbriae and their dynamics is regulated by an alternation of elongation and retraction cycles. How this dynamic is orchestrated is poorly understood; yet in the presence of the PilT retraction ATPase several other proteins are required to maintain bacterial piliation. The exact role of these proteins is still unclear but they all participate to some extent in counteracting retraction and favoring elongation. These proteins include the PilW lipoprotein, the minor (pseudo)pilins PilH-K, the PilC proteins and the minor pilins PilV and PilX and ComP318. Because of their overlapping roles, both PilC1 and PilC2 need to be mutated to induce loss of piliation592. Similarly, pilV/pilX double mutants and pilV/comP double mutants no longer have pili while simple mutants maintain some degree of piliation640. In the absence of PilT, piliation is (partially) restored in the previous mutants but they all lose some of the functions associated to TFP318,640. This suggests that these proteins might play a dual role in pilus assembly and pilus functions. Another possibility is that PilT is required for these functions to take place.

3.3.2.2 Pilus structure

The recent elucidation of the structure of the meningococcal TFP by cryo-EM at a resolution of 6Å has benefited to the understanding of the mechanisms behind TFP functions307. This structure shows a very similar organization to previously determined TFP structures: a hydrophobic core composed of packed α1 helices of the pilins stabilizes the interactions of the jointed globular heads that protrude at the pilus surface (Figure 39a). Yet, an unexpected finding is that the α1N of the α -helix of PilE is partially melted between amino acids 15 and 23. This melted helix is only found in the assembled pilus and not in the crystal structure of PilE. This non-helical conformation of the major pilin in the pilus has since then also been observed in the Type IVa pili of P. aeruginosa and N. gonorrhoeae308 as well as in the T2SS pseudopilus159 but not in the archaellum246,393.

The authors of this study have proposed that melting of the α-helix could account for the observed deformability of TFP under tension486,639 (Figure 39b,c). This is an interesting example of convergent evolution with the chaperone usher pilus which relies on a set of weak interactions for flexibility. While very different from the chaperone usher mechanism, the model proposed here for TFP would result in similar mechanical properties of the pilus and could account for the ability of meningococcus to adhere under flow515.

78

Figure 39: Structure of the meningococcal type IV pilus. (A) Cryo-EM map of the pilus shows three different possible helical modes based on the surface connectivity of the globular domains of PilE. A cross-sectional view shows the fitting of the crystal structure of PilE. Red arrows highlight the connectivity of the rod-like density at the center of the fiber. (B) Structural model of pilus stretching. When a force is applied to the pilus (grey arrow), partial melting of the α-helix occurs and results in fiber extension. This elongation induces changes in the regions of the protein that are surface-exposed. (C) Immunostaining of TFP reveals the exposure of new epitopes (recognized by the SM1 antibody) upon stretching of the fiber. Scale bar:1 µm. Adapted from Kolappan et al. 2016 and Biais et al. 2010

3.3.3 TFP-dependent functions in Neisseria meningitidis 3.3.3.1 Pilus retraction enables twitching motility In N. meningitidis, very few studies have examined the importance of twitching motility. As for other bacteria, N. meningitidis needs retractile TFP to move on surfaces and uses PilT to retract its TFP641 . The influence of other components of the piliation machinery on cell motility has only been poorly characterized. Twitching motility has only been quantified for pilV and pilX mutants and has yielded contradictory results. In a first study, speed of pilV mutants was found to be similar to that of WT (on average 1.5 μm.s-1) and pilX mutants were found to be slower than the WT323. Two other studies showed that motility was increased in pilV aggregates373 and conserved in pilX638. Given the differences in pili length and number expressed by these mutants it is difficult to make sense of such results323. Interpretation of meningococcal motility

79 is further complicated by the non-polar distribution of TFP in the bacterium (Figure 25a and 40a). For this reason, future studies evaluating twitching motility should also examine other parameters than the average bacterial speed (e.g. directional persistence) (Figure 40b). Twitching motility remains an interesting proxy to evaluate retraction properties of TFP.

Figure 40: Twitching motility in N. meningitidis. (A) Time lapse video of N. meningitidis stained with fluorescent NHS-ester. The bacterial track is indicated in red and 3 to 5 pili can be visualized at the bacterial surface. Motion of bacteria depends on pilus retraction in a tug of war mechanism. Time is indicated in seconds. Scale bar: 5 µm. (B) Typical track of N. meningitidis showing various turns and sudden speed increase. Track is shown over 7s. Adapted from Eriksson et al. 2015.

3.3.3.2 Pilus retraction enables natural competence

Another function of TFP that requires pilus retraction is transformation competence. Only the first steps of transformation: binding and uptake of DNA (import from the extracellular milieu to the periplasm) are dependent on the piliation machinery642,643.

In Neisseria meningitidis, efficiency of transformation by exogenous DNA is improved 30-fold644 in the presence of a 12 nucleotide-long motif (5’-ATGCCGTCTGAA-3’) called a DUS (DNA Over 2000 occurrences of 645,646 Uptake Sequence) on the exogenous DNA sequence . This increased efficiency results such motifs are from an increase in DNA uptake but not in DNA binding645. By NMR, DUS recognition has been found in the meningococcal shown to be dependent on direct electrostatic interactions between DNA and the minor pilin genome. Specific ComP400,644. In the absence of ComP, bacteria are a 1000-fold less competent373,644. This recognition and capture of suggests a central role for ComP in DNA uptake. meningococcal DNA is thought to How DNA is bound is still under debate. Indeed, purified TFP only bind to DNA with a very low prevent hazardous affinity. This affinity is further decreased in pili purified from a comP mutant. This has led to the recombination with foreign suggestion that ComP could be inserted in the pilus and responsible for initial DNA binding. DNA645. Yet, whole bacteria have been found to bind DNA just as well in the absence of ComP and, the

80 presence of ComP in the pilus has never been convincingly documented373. An alternative hypothesis has been proposed where DNA binding would be mediated by the PilQ secretin. Indeed, PilQ monomers and multimerized secretin bind DNA with a high affinity, independently of the presence of DUS647–649. It has also been proposed that the inner membrane platform protein PilG could bind DNA648. This unspecific DNA-binding allowed by pilus retraction could thus promote interaction of ComP with DUS to facilitate the initial uptake in the periplasm. Once DNA enters the periplasm, the competence effector ComE biases its diffusion towards the periplasm via a ratchet mechanism650. Subsequent steps will allow translocation of single stranded DNA in the bacterial cytoplasm independently of the TFP machinery (Figure 41).

Figure 41: Natural competence for transformation. Representation of the competence machinery in N. meningitidis. Double-stranded DNA (dsDNA) is proposed to be bound by the minor pilin ComP, PilQ (Q° and/or PilE. Whether ComP is found in the pilus is still unclear. Upon pilus retraction by PilT (T), DNA can enter the periplasm, where it is bound by the protein ComE. dsDNA diffusion is then biased towards the periplasm via a ratchet-like mechanism. The competence machinery then allows translocation of single-stranded DNA in the bacterial cytoplasm. Adapted from Hepp et al. 2016

3.3.3.3 Pilus retraction enables phage infection

An outcome of this natural competence is that this system can also be used by and can result in phage infection. For N. meningitidis, a single filamentous phage that uses TFP as a receptor has been characterized, the MDA (Meningococcal Disease Associated) phage that is found in hyperinvasive lineages (see Section 3.2.1). This phage was shown to bind extracellular portions of the pilus (presumably PilE) through a specific adsorption protein651. Furthermore, this phage uses the TFP secretin PilQ for secretion. In the absence of this protein, the phage is no longer assembled and exported from the bacterium522,523. This MDA phage can form filaments at the surface of the bacterium that promote biofilm formation (Figure 42a,b). MDA phage thereby enables meningococcus to form aggregates even in the absence of assembled pili652.

81

Figure 42: Characterization of the meningococcal MDA phage (A) Immunostaining of a bacterial biofilm adhering to an epithelial cell monolayer. A bottom layer of bacteria adheres to cells with their pili while the upper layers of bacteria are aggregated by the mean of their MDA phage filament assemblies. (B)Transmission electron micrographs reveals that the MDA phage forms pili-like filaments at the surface of N. meningitidis.

Adapted from Bille et al. 2017

3.3.3.4 Pilus retraction enables the formation of fluid aggregates

Unlike the previous TFP-associated functions, formation of bacterial aggregates is not dependent on TFP retraction. Yet, in N. meningitidis, aggregates have very different properties in the presence or in the absence of pilT. While WT bacteria form spherical aggregates with fluid properties, pilT mutants form highly viscous aggregates with irregular shapes. These differences are due to the loss in the intermittence of interactions between bacteria in the pilT mutant and highlights the importance of pili dynamics in aggregation653.

Because aggregation results from pilus-pilus interactions, two pilins have been proposed as the mediator of aggregation in meningococcus, the major pilin PilE and the minor pilin PilX. As mentioned earlier, the pilX mutant is no longer able to form aggregates and, based on its structure, PilX was proposed to mediate aggregation by inserting in the pilus and acting as a hook that would counteract pilus retraction325,638. This hypothesis has since then been refuted by showing that pilX mutants have fewer pili than the WT and that the number of pili determines the ability for bacteria to form aggregates323. These findings point back to PilE as the sole mediator of aggregation at the molecular level. Several studies have identified mutations in PilE that can affect aggregation in meningococcus. Antigenic variation was observed to affect both pilus bundling and aggregation444. Post-translational modifications of PilE were also shown to be important for pilus bundling and aggregation413,418. While it is tempting to speculate from these results that PilE-PilE interactions are responsible for the aggregative properties of the pilus, we should be cautious when interpreting these results. Indeed, the limitation of these studies resides in the lack of accurate quantification of piliation which precludes proper results interpretation given the importance of the number of pili per bacterium for aggregation 323. Another observation made in the previous studies is that mutants with a defect in aggregation also have a defect in adhesion to human cells at 4hours. By restoring their ability to form aggregates, bacteria regain their ability to efficiently adhere to cells. Differences in the adhesion patterns are also observed between non-aggregative and aggregative strains. While,

82 aggregative strains have localized adherence patterns with bacteria forming microcolonies at the cell surface, non-aggregative strains show diffused adherence with bacteria adhering as individual cells444 (Figure 43). The link between aggregation and adhesion was initially observed at late time points (4 hours of adhesion) suggesting that interbacterial interactions are important to amplify colonization following initial adhesion. Later on, a study suggested that aggregation was even necessary to initiate adhesion. This suggestion was based on the observation that cells infected with a bacterial inoculum filtered to remove aggregates adhered as inefficiently as unpiliated bacteria638 . Other studies have since then suggested that initial adhesion under flow is similar in aggregative and non-aggregative strains515,653. Whether adhesion and aggregation can be dissociated is still under debate. In the next section, we try to only discuss factors that are specifically involved in adhesion.

Figure 43: Aggregation promotes adhesion. (A-B) Scanning electron micrographs of Hec-1B cells infected by N. meningitidis. (A) Clone 12 expressing aggregative pili shows localized adherence and forms microcolonies at the cell surface. (B) Fewer bacteria are found adhering as individuals to epithelial cells by clones expressing non-aggregative pili.

Adapted from Marceau et al. 1995.

3.3.3.5 Adhesion to human cells

Adhesion is still poorly understood in N. meningitidis. Bacteria can only adhere to human cells and adhesion has mainly been observed to epithelial and endothelial cells. Early studies suggested that different pathways could be used for adhesion to both kinds of cells. Indeed, meningococcus adheres more efficiently to endothelial cells than epithelial cells and this adhesion is less susceptible to inter- and intra-strain variation and pili classes607. This is further supported by the fact that upon adhesion, bacteria trigger two distinct pathways depending on whether they adhere to epithelial or endothelial cells654. Because adhesion to both kinds of cells is pili-dependent, these findings support the fact that different receptors could be targeted by TFP on endothelial and epithelial cells.

Receptors and adhesins described for TFP are summarized in Figure 44 and their relevance is discussed in the following paragraphs.

83 ¬ Adhesion to epithelial cells: two putative receptors The first TFP receptor to be identified was CD46, a transmembrane glycoprotein involved in complement regulation. A lot of indirect evidence pointed to CD46 as an epithelial cell receptor: incubation of epithelial cells with α-CD46 antibodies inhibited bacterial binding and so did pre-incubation of bacteria with CD46. In addition, transfection of CHO cells (a cell line derived from hamster ovaries) with CD46 allowed bacterial adhesion to these otherwise non- colonized cells598,655. A transgenic mouse model expressing human CD46 was later developed. While mortality of transgenic mice was increased upon peritoneal infection, this increase in virulence was not dependent on the presence of TFP and could be due to an unrelated altered interferon-response in these transgenic mice656. Mortality was slightly higher upon intranasal infection and was dependent on the presence of pili. Yet, bacteria were not able to invade the mouse and no bacteria could be recovered from the blood617. Since the publication of these results, no other study has been able to reproduce these experiments. On the contrary, other groups found no correlation between levels of CD46 and adhesion, no antibody-mediated inhibition of adhesion, no bacterial adhesion upon transfection of CD46 in CHO cells and no decrease in adhesion upon silencing of CD46657–659. Furthermore, the bacterial adhesin at the surface of TFP that would bind CD46 has never been identified. Altogether, these results argue against the fact that CD46 is a TFP receptor, but this protein could be involved in interaction with another adhesin or involved in a subsequent infection step.

One of the studies showing the absence of adhesion inhibition by α-CD46 antibodies reported a new putative receptor: the Platelet Activating Factor-receptor (PAFr)658. These data are supported by co-localization experiments on epithelial tissues and epithelial cell lines, inhibition of adhesion with α-PAFr antibodies and transfection of Chem-1 cells (a cell line derived from the rat) with PAFr allowed bacterial adhesion to these otherwise non-colonized cells. Interaction between pili and PAFr was confirmed by ELISA. Furthermore, this study identified post-translational modifications of PilE as the molecular mediators of interaction with PAFr. Both glycosylation and phosphocholine modifications seemed equally important for PAFr binding. Unlike the vast majority of meningococcal strains, the 8013 reference strain we use in the lab bears a phosphoglycerol post-translational modification instead of a phosphocholine. Engineering the 8013 strain to express a phosphocholine greatly increased adhesion to epithelial cells and interaction with PAFr thus confirming the role of this post- translational modification in epithelial cell binding. Given the importance of the phosphoglycerol post-translational modification in aggregation in the 8013 strain418, it would be of great interest to evaluate the aggregation properties of the modified phosphocholine- expressing 8013 strain and to test how this influences adhesion at later time points. To our knowledge, the role of PAFr has not been addressed in endothelial cells.

84 ¬ Adhesion to endothelial cells: a CD 147-dependent adhesion? The most recently proposed receptor of meningococcal TFP at the surface of endothelial cells is CD147600, a membrane protein used as a marker of brain capillaries660. Interestingly, another pathogen, measles virus has also been found to use CD147 as a receptor661. Previous reports had established CD46 as a receptor for measles virus but this interaction was later shown to be restricted to a specific strain662. This is reminiscent of what has been shown for meningococcus and could explain some of the discrepancies observed in the literature.

The role of CD147 in meningococcal adhesion was shown on brain-derived endothelial cells and on human brain sections. As for the previous examples, these conclusions are supported by co-localization of adherent bacteria with CD147, antibody-mediated inhibition of adhesion, hyper-adhesion on cells overexpressing CD147 and decreased adhesion for CD147-silenced cells. Interactions were documented between CD147 and both PilE and PilV pilins. Incubation of cells with both pilins also inhibited adhesion. The limit of this study resides in the weakness of the interactions documented between pilins and CD147. Authors propose that when assembled in a quaternary structure, these proteins could mediate higher affinity interactions or simply that the number of binding sites present at the surface of TFP would suffice to initiate adhesion600. The relevance of this receptor at the surface of epithelial cells has not yet been addressed.

¬ TFP: one adhesin or multiple adhesins? A lot of the research effort has been dedicated to identifying a TFP receptor, but early observations on the nature of the adhesin also provide insights in the adhesion mechanism. PilV, PilE and PilC1 have been proposed to act as adhesins mediating TFP-dependent adhesion.

While PilC1 has been shown to be required for adhesion to both endothelial and epithelial cells592,625, no receptor for this putative adhesin has ever been described thus questioning its direct role in adhesion.

Both PilE and PilV have been shown to bind CD147600 and PilE has also been involved in PAFr- dependent adhesion658. While the direct role of PilV is questioned by the recent finding that it is probably located in the periplasm and influences the number of pili at the cell surface323, a direct role of PilE has been suggested by early studies which observed direct binding of PilE to cells444 and an influence of changes in the primary sequence of PilE on adhesion663,664. All these observations make PilE an appealing target to investigate the molecular mechanisms of adhesion.

85

Figure 44: Putative adhesins and associated receptors of meningococcal type IV pili. The three different receptors described for TFP-dependent adhesion to endothelial cells are represented. The pilus component that act as the adhesin is indicated. PTM: Post Translational Modifications. Adapted from Coureuil et al. (2017).

3.3.3.6 Cellular response and signaling

Upon adhesion of meningococcus to cells, an almost instantaneous deformation of the plasma membrane is observed665,666. This membrane deformation is accompanied by the recruitment of several cellular components including ezrin and F-actin that form a cortical Ezrin: 300,667 Protein member plaque (Figure 45b,c). The chain of events leading to the formation of this cortical plaque of the Ezrin- has been described. Membrane deformation is the first event to occur and is quickly followed Radixin-Moesin 665 (ERM) family that by the recruitment of ezrin, itself followed by the recruitment of F-actin . Interestingly, acts as a membrane deformation has been shown to be ATP-independent and independent of tubulin crosslinker 637,666 between actin or actin recruitment . On the host side, only two factors required for membrane filaments and the deformation have been identified: membrane cholesterol637(Figure 45c) and α-Actn4668. The plasma membrane. first could alter membrane fluidity and thereby prevent the formation of membrane protrusions637. The second one was shown to determine the strength of adhesion of TFP to the cell, which could explain this defect. On the bacterial side, this process is absolutely dependent on the presence of TFP. The minor pilin PilV is required to trigger cellular response in endothelial cells, but not in epithelial cells637,654. As mentioned earlier, pilV mutants have less pili and this was proposed to be the reason why they fail to trigger cellular response. The retraction ATPase PilT speeds up membrane deformation but is not required for membrane deformation665. Finally, some PilE sequence variants have been shown to lose cellular response upon adhesion to endothelial cells but not upon adhesion to epithelial cells and vice-versa 669. The fact that these pilE mutants maintained cellular response on one of the two cell types indicates that their defect in membrane response is unlikely to be due to a decrease in their piliation. This rather suggests that PilE is directly responsible for the deformation of plasma membrane through two different pathways on epithelial and endothelial cell lines654,669. This is reminiscent of what we described earlier for adhesion. In line with this, a study to be published, reveals that cellular membrane deformation is a passive process that only relies on the adhesive properties of TFP (Figure 45a). Indeed, physical modelling predicts that the interactions

86 between the lipids of a cell membrane and an adhesive fiber with a very small diameter such as the TFP (6nm) is sufficient to deform the membrane and initiate protrusions along TFP (Charles-Orszag et al. 2018, in revision). Again, these findings point to the major pilin PilE as the key to understand TFP-mediated adhesion and more generally, TFP functions.

Figure 45: Cellular response to meningococcal adhesion. (A) Cellular response is dependent on the presence of the CD-147 receptor and the ß2 Adrenergic Receptor (ß2AR). These two proteins form a complex together with α-Actn4 to ensure proper adhesion of N. meningitidis. (B) Upon adhesion, the membrane deforms below the microcolonies. This leads to the recruitment of several cellular components, including junctional proteins. This participates in tight junction opening and facilitates invasion of the underlying tissue by N. meningitidis. (C) Scanning electron micrographs of N. meningitidis adhering to epithelial cells show typical cellular response. These membrane protrusions are lost when cholesterol is depleted from the cell membrane (MßCD). Inset: Transmission electron micrograph showing the intensity of the cellular remodeling around the microcolony. Adapted from Coureuil et al. 2017 and Mikaty et al. 2009.

87 In this introduction, we have seen that prokaryotes have evolved a tremendous variety of surface structures that are central in the ’s lifestyle. In spite of their structural diversity, these appendages mediate common functions, including: motility, adhesion, auto- aggregation and competence for transformation. While the mechanisms underlying a few of these functions are very well understood (e.g. adhesion of the chaperone-user pilus), this is not the case for the vast majority of pili-dependent functions. The variability that we observed in the assembly mechanisms and the structure of prokaryotic appendages is likely to be reflected by a great variability in the mechanisms underlying the functions of these appendages. It is therefore necessary to focus on phylogenetically related appendages to understand their structure-function relationships.

The TFF family is an extremely versatile family with many conserved structural features. After decades of intensive research, we now have a good understanding of the organization of their piliation machinery and the assembly mechanisms of these filaments. The homologous assembly mechanisms of TFF further support common structure-function relationships. Yet, the mechanisms by which they mediate their diverse functions are still ill-defined. This is particularly true in the case of TFP, “the prokaryotic Swiss Army knives”6. We have shown that the pathogen Neisseria meningitidis is a particularly well-suited model organism to study multifunctional TFP and decipher how a single surface structure can mediate so many functions. This is the objective of the present doctoral work and we have taken advantage of the deep mutational scanning technique to achieve this goal.

88 In N. meningitidis, TFP are polymers of the major pilin PilE, a relatively short protein of approximately 160 residues. It appears quite surprising that a single relatively simple protein would be able to mediate so many functions. In combination with other experimental evidence, this led to the hypothesis that the major pilin PilE was only required to build a protein scaffold that would allow the display of other proteins at the surface of the pilus: the minor pilins. These minor pilins were hypothesized to be the actual mediators of functions: comP for competence, PilV for adhesion and cellular response and PilX for adhesion and aggregation. Yet, a study conducted in the lab convincingly showed that two of these minor pilins (PilV and PilX) could mediate their functions from within the bacterial cell. Mutants lacking PilV or PilX have a reduced number of pili, indicating a role of these proteins in pilus biogenesis. This reduction in the number of pili is sufficient to explain the loss of function observed in these mutants. Therefore, the minor pilins PilV and PilX do not appear to be the direct mediator of TFP functions. This observation brought us back to questioning the direct role of PilE in TFP functions. In order to investigate if PilE is the direct mediator of these functions and how it exerts them, we undertook a global approach. Because levels of piliation were shown to determine levels of function, this approach needed to simultaneously evaluate the importance of individual amino acids for each of the studied functions (aggregation, adhesion and competence) and for piliation. We did so by taking advantage of a novel technique called deep mutational scanning679 and combining it with four different selection methods. This technique allows to investigate the functional consequences of almost 2000 different single mutations in the PilE sequence. We present our strategy and the outcome of this study in the Results section. We then put these results in the existing literature framework and highlight the new questions they raise in the final Discussion section.

Objectives 89

90

Figure 46: Results summary.

In vitro models of infection show the importance of TFP is at least twofold. Interaction with host cells is initiated by pilus-dependent adhesion. Later on in time, this TFP-dependent adhesion is amplified by the capacity of bacteria to form microcolonies through pilus-pilus interactions. These two functions of TFP can be separated. TFP-mediated adhesion is dependent on specific interactions mediated by the tip of TFP. These interactions can be disturbed using antibodies specific for the tip of TFP or by mutating residues at the tip of then pilus. This adhesion is also dependent on the presence of minor pilins which regulate the number of pili present at the bacterial surface. Interbacterial aggregation is mediated by pilus-pilus interactions and cannot happen below a given piliation threshold. To reach this piliation threshold, the minor pilin PilX is required as it regulates the number of pili. The amino-terminus of PilE is also critically involved in aggregation because it regulates pilus length. Below a critical length, pilus-pilus interactions are too weak to initiate the formation of aggregates. Pilus-pilus interactions are certainly mediated by a surface-exposed electropositive patch of amino acids (including lysine 140). This model highlights how the structure of the major pilin evolved to carry different functional domains while maintaining its ability to assemble as pili.

Results 91

92

1 Submitted article: Mechanisms of meningococcal type IV pili multiple functions revealed by deep mutational scanning.

93 Mechanisms of meningococcal type IV pili multiple functions revealed by deep mutational scanning.

Authors:

Paul Kennouche1,2, Arthur Charles-Orszag1,2, Anne-Flore Imhaus1, Guillaume Duménil1,2*

Affiliations:

1: Pathogenesis of vascular infections unit, INSERM, Institut Pasteur, 75015 Paris, France 2: Université Paris Descartes, 75006 Paris, France * Corresponding author

Abstract:

Type IV pili (TFP) are multifunctional micrometer-long filaments expressed at the surface of many prokaryotes. In Neisseria meningitidis, TFP are homopolymers of the major pilin PilE. They are crucial for virulence as they mediate interbacterial aggregation and adhesion to host cells although the mechanisms behind these functions remain unclear. Here, we simultaneously determined the regions of PilE involved in pili display, auto-aggregation and adhesion to human cells by using deep mutational scanning. Mining of this extensive functional map of the pilin sequence provides new mechanistic insights: first, the hyperconserved a1-domain of PilE was found to be involved in the balance between pili length and number; moreover, we identified an electropositive cluster of residues centered around Lys140 necessary for aggregation; finally, we show the importance of the tip of TFP in adhesion. Overall, these results support a direct role of PilE in aggregation and adhesion to host cells and identify these specific functional domains of PilE.

Introduction:

In order to interact with their environment, prokaryotes display a great variety of surface organelles. Type

IV filaments (TFF) constitute a class of closely-related organelles that are characterized by the class III signal peptide, a conserved amino terminal signature sequence of their major constituent 128,234. They

94 comprise type IV pili (TFP), the bacterial type II secretion system (T2SS) and the archaellum (archaeal flagellum). These very widespread filaments have a central role in the lifestyle of the organisms expressing them. Indeed, they play key roles in metabolism (nutrient acquisition and respiration 7,305, horizontal gene transfer, surface colonization (motility and attachment) and interaction with host cells in the case of pathogens 6. TFF are helical homopolymers of major pseudopilins for T2SS, type IV pilins for TFP and archaeal flagellins for the archaellum. These building blocks of the filaments share a highly conserved α- helical hydrophobic N-terminus which is buried inside the fiber while their highly variable C-terminus is rather exposed at the surface of the filament 362. Monomers are assembled by complex machineries which share functional and structural homology. Among the 10-18 proteins involved in pilus assembly, three are virtually found in every system expressing TFF: a specific prepilin peptidase involved in the cleavage of the prepilin leader sequence, an ATPase necessary for filament polymerization and a platform-like membrane protein which interacts with the ATPase to drive filament assembly 6.

Specifically, type IV pili are micrometer-long retractile filamentous appendages found at the surface of several pathogenic bacteria such as Neisseria meningitidis, Neisseria gonorrhoeae, Pseudomonas aeruginosa and Vibrio cholerae 6,283. They mediate several functions including competence for transformation 670,671, interbacterial aggregation 672,673, twitching motility 327,467 and adhesion to host cells

607,674,675. The multiple functions carried by TFP place them as central components in the pathogenicity of these bacteria.

N. meningitidis (or meningococcus), used in this study as a model of TFP-expressing bacterium, is a diderm bacterium found as a commensal in the human nasopharynx. Occasionally, the bacterium can breach the oropharyngeal epithelial barrier and colonize human blood vessels, eventually leading to sepsis and/or meningitis 514. In a humanized mouse model, this vascular colonization has been shown to be dependent on the ability of bacteria to adhere to human endothelial cells 488, to resist shear stress through reorganization of the host cell plasma membrane 431,515 and to form fluid aggregates653. All these functions have been shown to be dependent on the presence of TFP, thus highlighting the importance of this virulence factor in the pathogenesis of Neisseria meningitidis.

Despite the importance of TFP and their numerous functions, their mode of action remains unclear. In addition to the major pilin PilE that constitutes TFP, meningococcus also expresses three minor pilins (PilV,

95 PilX and ComP) that have been involved in TFP functions. They are designated as minor pilins as they all share a high degree of homology with PilE but are expressed at much lower levels. While mutants in the pilus assembly machinery fail to express any pili on their surface and lose all pili-dependent functions, mutants in pilX, pilV or comP have more complex phenotypes 635,638,676. pilV mutants have a moderately reduced piliation (70% of wild type) leading to a severe defect in remodeling of the host cell plasma membrane upon adhesion associated with a defect in adhesion, especially under flow conditions 323,431,676.

Interestingly, the pilV mutant maintains some aggregative properties. The pilX mutant has a more pronounced defect in piliation (30% of wild type) and can neither form aggregates nor adhere to human host cells 323,638. The comP mutant has a slightly increased piliation, is deficient for competence but retains all other T4P-dependent functions 373. It has been proposed that these minor pilins could be inserted periodically in the PilE scaffold and would specifically mediate these functions 325,599. Yet, a recent study in our lab has shown that the reduced piliation of pilV and pilX mutants are sufficient to explain their phenotypes

323. Indeed, when these piliation defects were reproduced by tuning the expression of the pilus assembly

ATPase PilF, a similar loss-of-function phenotype was observed. Combined with sub-cellular localization experiments, these results suggest that PilV and PilX are important for pilus biogenesis rather than for directly carrying pilus functions. These findings and the homopolymeric nature of TFP point back to PilE as the main mediator of TFP-associated functions. This hypothesis implies that PilE is under many functional constraints. First, to assemble retractile TFP, PilE must be folded so that it can interact reversibly with several members of the piliation machinery and itself 349. Second, PilE needs to enable interaction between

TFP in a parallel and antiparallel fashion to support pili bundling and aggregation. Third, to mediate adhesion, the major pilin needs a region that can interact with a cellular receptor.

Placing the major pilin at the center of pilus assembly and functions thus raises several questions. Is PilE directly involved in all these functions? How can such a short protein (around 160-amino acid in N. meningitidis) mediate so many different functions? Are different domains involved in different functions or is a single common domain sufficient? To shed light on these questions, we aimed to identify mutations in PilE which would disrupt specific properties of TFP. In particular, mutations that would maintain piliation but affect individual functions. Saturating mutagenesis approaches have previously been used successfully in N. gonorrhoeae on PorA and the C-terminus of PilE 677,678. We decided to extend this approach by taking

96 advantage of the deep mutational scanning method 679 and applied it to the full length PilE. This provided us with functional maps of PilE for piliation, aggregation and adhesion. By combining analysis of these maps and a thorough analysis of piliation, adhesion and aggregation of de novo generated PilE mutants, we could evaluate the importance of PilE in piliation, adhesion to human cells and interbacterial aggregation and decipher the molecular determinants behind these functions.

97 Results:

Deep mutational scanning of PilE

In order to generate a library of N. meningitidis strains carrying mutations in the pilE gene but otherwise identical, we first produced a plasmid library of single-point mutations in pilE using error-prone PCR. In N. meningitidis, PilE is naturally subject to a phenomenon called antigenic variation, whereby recombination events between the expressed pilE locus and silent pilE loci mediate sequence variation of PilE 601. This mechanism is dependent on the presence of a guanine quartet upstream of the pilE locus 680. To prevent subsequent antigenic variation in PilE, this plasmid library of single-point mutations in pilE was thus transformed in a strain of N. meningitidis with a chloramphenicol-resistance cassette inserted in the guanine

681 595 quartet upstream of PilE (G4) . In this study, we used the PilESB sequence as a reference sequence .

For this reason, in the rest of this paper, the reference strain will be indicated as PilESB or SB. Upon next- generation sequencing of the genomic pilE locus of transformed bacteria, we found that this library was at saturation levels approximating 90% for single point mutants (Fig. S1A). Single point mutations made up

35% of the mutant library and only single mutants were considered for the remainder of the analysis (Fig.

S1B-D). On average, each amino acid of the PilE protein sequence was mutated in 6 alternative amino acids.

This initial library was then submitted to three different selection schemes to assess surface piliation, auto- aggregation and adhesion to host cells (Fig. 1A). For the piliation-based selection, live bacteria immunolabeled with an anti-pilus monoclonal antibody (20D9)682 were sorted by FACS (Fluorescence-

Activated Cell Sorting). Using flow cytometry, we could distinguish the piliation level of reference mutants.

As shown in Fig. S2A, a mutant lacking the prepilin peptidase PilD was negative for piliation using this cytometry-based approach. We could also observe the increased piliation of a retraction-deficient strain

(pilT), but also the more subtle defect in piliation of a pilV strain (Fig. S2B). For the aggregation selection, bacterial suspensions from the initial mutant library were allowed to aggregate for 2 hours prior to filtration through transwells with 5 µm pores. The method was validated by mixing PilESB and pilD mutants (1:1) in an initial suspension. We could observe a 3-log enrichment in aggregating bacteria in the collected fraction

(Fig. S2C). Because of the variability associated with this method, this filtration was repeated three times to

98 increase the efficacy of the method (Fig. S2D). Finally, for the adhesion selection, human umbilical vein endothelial cells (HUVEC) were infected with the initial mutant library for 4 hours and washed extensively to only retain the adherent bacteria. Again, this method was validated by using mixed inoculates of a pilV mutant and PilESB or a mutant lacking the elongation ATPase (pilF) and PilESB (1:1) for which we could observe a 2-log enrichment in the adherent population (Fig. S2E). After selection, we obtained four libraries

(initial, piliation, aggregation and adhesion), in which the pilE gene was then sequenced using next- generation sequencing. The frequency of each single point mutation in the selected libraries was compared to the frequency of the same mutation in the initial library. This ratio is indicative of whether mutations are beneficial or detrimental to the function considered. We will refer to the log2 of this ratio as “the mutation score” (Fig. 1A). Overall, 1147 different point mutations were associated with three functional quantitative measures, generating 3441 data points.

Stop and synonymous mutations were used to control the validity of the resulting functional mapping. Stop mutations are expected to induce a complete loss of piliation and function while synonymous mutations should be indistinguishable from the reference strain. Indeed, stop mutations were selected against in all three libraries as shown by their distribution at the upper left of the volcano plots (Fig. 1B). Conversely, synonymous mutations were found to be neutral or positively selected. We also observed that the mutation scores for each function were correlated to the mutation scores for piliation (Fig. 1C). This is consistent with the observation that piliation is a good predictor of function323.

In summary, using deep mutational scanning, we generated a valid dataset associating an average of 6 mutations per PilE amino acid with a systematic quantification of piliation, aggregation and adhesion to host cells. On this basis, we began to exploit the wealth of results generated by this functional screen in terms of piliation, aggregation and adhesion. For these three functions, a similar strategy was followed, data extracted from the different libraries are usually presented in the first panel of each figure followed by de novo generation of individual mutants to validate the results.

99

Figure 1: Deep mutational scanning of PilE.

(a) Representation of the overall approach used in the study. On the top, the three methods used for selection of pilE mutants from the initial library to generate the adhesion, aggregation and piliation libraries are represented. Colored stars represent point mutations. Differences in mutation frequencies after selection are depicted. On the bottom, sequencing of all libraries using Next Generation Sequencing (N.G.S.) is shown. Reads were analyzed, and a mutation score was calculated for each amino acid mutation in the piliation, adhesion and aggregation libraries. Resulting data can be represented as an average per amino acid for each library as seen on the graph. The zoomed area depicts the extent of data obtained for each amino acid with 4 to 7 different mutations tested in three different assays. Upon analysis, we selected mutants of interest, de novo generated these individual mutations and tested their phenotypes. (b) Volcano plots of piliation, adhesion and aggregation mutation scores for all pilE mutations. Synonymous mutations are highlighted in orange and stop mutations in blue. (c) Graphs showing the linear correlation between the mutation scores of piliation, adhesion and aggregation for all pilE mutations.

100 Identification of a core region necessary for pilin folding

A synthetic view of the deep mutational scanning for piliation was obtained by averaging piliation mutation scores for each amino acid. These average values are plotted according to pilin sequence numbering in

Fig. 2A. The degree of conservation of PilE residues (based on the alignment of PilE sequences extracted from the Neisseria Multi Locus Sequence Typing website 683 is color-coded. In this representation, amino acids V33, V47, V66 and V82 had among the lowest average piliation mutation scores (Fig. 2A). In addition, even the most conservative mutations of these amino acids appeared deleterious to piliation (PilEV33L,

PilEV47L, PilEV82A) with high statistical significance (Fig. 2B). This suggests that these amino acids play a key role in piliation and we chose to focus on this subset.

We thus generated these conservative mutations de novo and tested the phenotypes of the corresponding mutants. As a reference, we also generated the PilEW108S mutant that was previously reported in N. gonorrhoeae to lose surface piliation 684 and was also predicted to have impaired piliation in our screen (Fig.

2A-B). Analysis of these point mutants using flow cytometry showed an important decrease in the proportion of piliated bacteria in the mutant population (Fig. 2C). This was also supported by their markedly decreased ability to form aggregates in solution (Fig. 2D). Interestingly, using the recently published pilus structure 307, we could observe that V47, V66 and V82 form a hydrophobic cluster in PilE (Fig 2E). The involvement of

V82 in this cluster likely favors the proper packing of the first (β1) against the alpha-helix (α1), suggesting an impaired folding of PilE in the mutants. In line with this hypothesis, we detected a decrease in PilE levels in whole cell lysates of mutant bacteria compared to PilESB, especially in PilEV82A (Fig. 2F, Fig.

S3). Overall, these results confirm the predictive ability of this method and identify a patch of key hydrophobic amino acids under high structural constraints to maintain PilE proper folding. These results constitute an example of what can be extracted from this dataset. Full analysis of the available information concerning pilus assembly will require further studies. Nevertheless, knowledge of the piliation level of each mutant was essential to interpret the following results related to aggregation and adhesion.

101

Figure 2: Identification of core regions necessary for piliation.

(a) Piliation mutation scores averaged for each amino acid in the PilE sequence. Mean values ± SEM of the different mutations are indicated. Individual mutations that are discussed in this section are highlighted in orange and blue. The black line indicates the region of the N-terminus that shows little effect on piliation upon mutation. Background colors indicate the conservation of each amino acid in the PilE sequence among the diversity of Neisseria isolates (dark blue, highly conserved; light blue, variable). (b) Volcano plot of piliation mutation scores for all PilE mutations. Outliers from panel A are highlighted. (c) Proportion of piliated bacteria relative to that of pilESB as measured by flow cytometry on a selection of de novo generated mutants. Mean values ± SEM are indicated for each strain. N⩾3 independent experiments were performed. (d) Quantification of aggregation after 30 minutes relative to pilESB- GFP. Mean values ± SEM are indicated for each strain. N⩾3 independent experiments. (e) On the left, side view of the structure of PilE. V33, V47, V66 and V82 are highlighted in orange on the structure. In the center, close up on these 3 amino acids reveals the close hydrophobic packing of V47, V66 and V82. V82 is engaged in the packing of the β-sheet β1 over the α-helix α1 probably explaining its key structural role. On the right, distances between amino acids are indicated in ångströms. Figure was made with Pymol using the structure of the pilus deposited in PDB under the reference 5KUA 307 (f) Western blot analysis from whole cell lysates to quantify pilin levels. Blot was probed for PilE and Rmp4.

102 Mutations in the hyperconserved N-terminus of PilE affect the balance between pilus length and abundance

A surprising outcome of the analysis of piliation levels in the mutant library is that mutations in most of the highly-conserved regions were detrimental for piliation except for those in the hyperconserved N-terminus of PilE (α1) that appeared tolerant to mutations (Fig. 2A, underlined sequence). In certain cases, mutants in this region even appeared as more piliated than the wild type according to the FACS-based selection.

For instance, this was the case for PilEL3H, PilEI4N and PilEV9M (Fig. S4A). Because this observation was quite unexpected, we undertook the analysis of the piliation of a subset of mutants in the α1 region in more detail. Using the cytometry-based approach on individual mutants rather than the whole library, we can obtain three parameters describing piliation (Fig. S4B): (i) the population piliation, i.e. the average piliation level in the whole population calculated as the mean fluorescence intensity of all the bacteria; (ii) the percentage of piliated bacteria and (iii) the average piliation level per piliated bacterium, i.e. the mean fluorescence intensity of individual piliated bacteria. Analysis of the piliation level of the pilESB reference strain revealed that in the conditions used here (piliation is measured on bacteria directly harvested from the plates), not all bacteria have pili (Fig. S2A) and only around 7% of the PilESB population is piliated (Fig.

S4C-D). We next examined the piliation of PilE mutants generated de novo through the study and compared it to the piliation of PilESB. Logically, we could see that the population piliation and percentage of piliated bacteria are very well correlated in PilE mutants (Fig. S4E). More interestingly, the percentage of piliated bacteria and the piliation per bacterium were also strongly correlated (Fig. 3A). In other words, the more piliated individual bacteria are, the higher the proportion of piliated bacteria in the population will be. In contrast, mutants in the α1N of PilE were outliers in this distribution. All of them had a higher percentage of piliated bacteria than the one expected for mutants with such low levels of piliation per bacterium, except for PilES34A which had the opposite phenotype (Fig. 3A, indicated in orange and light blue respectively).

Analysis of the piliation of mutants in the α1N region of PilE by flow cytometry thus reveals that they form a particular class of mutants with a larger population of piliated bacteria but a lower level of piliation per bacterium. Interestingly, the percentage of piliated bacteria was greatly diminished in the pilX mutant and increased in the pilT mutant.

103 We then wondered how this decreased piliation translated into the number and length of TFP expressed by these bacteria. To this end, we examined these outliers using immunofluorescence and scanning electron microscopy on single bacteria. We found that PilEL3H, PilEI4N and PilEV9M expressed numerous very short pili at their surface. On the contrary, PilES34A mutants displayed few, very long pili (Fig. 3C, Fig. S4F).

Differences in the piliation of these mutants were not due to changes in the expression of the pilin protein

(Fig.S4G), thus pointing to a change in the balance between initiation of pilus assembly and pilus extension.

Availability of this new mutant class allowed us to investigate how the balance between pilus length and number affects TFP-dependent functions. Interestingly, mutants with short pili had almost completely lost the ability to form aggregates (Fig. 3B). Among these, only the mutant with the highest piliation, PilEI4N had maintained a residual ability to form aggregates. On the other hand, the mutant PilES34A with long pili had no significant advantage in aggregation compared to the wild-type (Fig. 3B). These results suggest that pili have to reach a minimal length to promote aggregation but that above this threshold there is no further gain of function (at least not at the cell density used here). For the aggregation process to occur, the interaction between TFP from two closely located bacteria has to be maintained during pilus retraction, thus bringing the two bacteria together. A mutant in the retraction ATPase pilT was used to lengthen pili and test their interactions. Introduction of this mutation was sufficient to restore aggregation in mutants with short pili (Fig.

3D). This supports the notion that mutations in these strains do not affect the intrinsic ability of the pili to form pili-pili interactions but that in these strains, pili are too short to form stable interactions.

In order to confirm that the pilus-mediated interbacterial interaction force was modified by pilus length, we used a co-aggregation assay between PilESB and PilEI4N or PilESB and PilES34A. Indeed, it was shown that the position of bacteria in aggregates is dependent on the interaction forces that they are able to exert, with bacteria exerting lower forces segregating at the periphery of the aggregate, and reciprocally for bacteria

685 exerting higher forces . In both cases, we could observe co-aggregates of the mutant and the PilESB strains. Interestingly, PilEI4N segregated at the periphery of the co-aggregates (Fig.3E,G), while PilES34A was concentrated at the center of the co-aggregates (Fig.3F,G). This suggests that because of their reduced interaction surface, the short TFP of PilEI4N generate lower interbacterial interaction forces, while the long

TFP of PilES34A generate higher interbacterial interaction forces than those of PilESB, due to their increased

104 interaction surface. These lines of evidence support the idea that pilus length regulates interbacterial interaction forces and therefore that pilus length is a critical determinant of aggregate formation.

The observation that a subset of mutants with short pili could no longer mediate aggregation prompted us to examine whether these findings could be generalized to the N-terminus of PilE using our deep mutational scanning data. When compared to average piliation levels, defects in aggregation were most severe for mutations in the N-terminal region of PilE (Fig. S4H), thereby generalizing the importance of the N-terminus in aggregation, probably through regulation of the equilibrium between percentage of piliated bacteria and piliation per bacterium.

We then looked at the adhesion properties of these α1N mutants. Surprisingly, they did not show any significant difference in 4h-adhesion to HUVEC cells when compared to PilESB, except for PilEL3H which adhered slightly better than PilESB (Fig. 3H). This is unexpected as aggregation has been shown to favor adhesion 444,638,686. To better understand the adhesion process, we then examined adhesion after 30 minutes, prior to microcolony formation, and could observe that PilEL3H adheres significantly better than

PilESB (Fig.3I, Fig. S5). In the case of adhesion, pilus length does not seem to be a critical parameter. On the contrary, PilES34A which has longer pili has a slight defect in adhesion (Fig.3I). Analysis of the α1N mutants thus highlights the importance of pilus length in aggregation but suggests that length does not matter for adhesion, rather pilus number is important.

105

Figure 3: The hyperconserved N-terminus of PilE regulates the balance between pilus length and abundance. (a) Piliation per bacterium as a function of the proportion of piliated bacteria normalized relatively to pilESB values as measured by flow cytometry with the 20D9 monoclonal antibody. Black dots indicate all the individual de novo generated PilE mutants that were characterized. Blue dots indicate mutants of the piliation machinery. Dotted lines indicate 95% confidence interval. Orange and light blue dots indicate outliers from the distribution. (b) Quantification of aggregation normalized relatively to pilESB for a subset of outliers. Mean values ± SEM are indicated for each strain. N⩾3 independent experiments. (c) Scanning electron microscopy of single bacteria on a cellulose filter. Pili are preserved by incubating filters with 20D9 prior to post-fixation and processing. Scale bars: 500nm. (d) Quantification of aggregation normalized relatively to pilESB for mutants lacking the retraction ATPase pilT. Mean values ± SEM are indicated for each strain. N⩾3 independent experiments. (e-f) Relative bacterial occupancy in the co-aggregates, as a function of the distance from the aggregate center. For each aggregate, radial fluorescence intensities of pilESB-mCherry and GFP-expressing mutant strain were divided by the maximum intensity in the corresponding channel. Intensity distributions were averaged over several co-aggregates. Mean values ± SEM are indicated for each strain. N=3 independent experiments. (g) Representative images of co-aggregates of pilESB-mcherry and GFP- expressing indicated mutant after 30 minutes of incubation at a 1:1 ratio and an OD of 0.3. N=3 independent experiments. Scale bar: 15 µm. (h) Quantification of adhesion after infection of HUVEC for 4 hours (MOI 100), expressed as Colony Forming Units on GCB plates at appropriate dilution. Mean values ± SEM are indicated for each strain. N⩾3 independent experiments. (i) Quantification of adhesion after infection of HUVEC for 30 minutes (MOI 500) by microscopy, expressed relatively to pilE106SB- GFP. Mean values ± SEM are indicated for each strain. N⩾3 independent experiments. Auto-aggregation is dependent on a patch of charged amino acids centered around lysine 140

The functional mapping of the major pilin performed in this study also provides new insights on the regions of PilE necessary for auto-aggregation. Although aggregation is tightly linked to the number of TFP 323 and to their length, as shown above, the structural basis for aggregation remains unclear. Previous work indicates that the PilE-dependent aggregative properties of TFP in N. meningitidis rely on electrostatic interactions between pili of different bacteria 418. Since in our library screening approach we simultaneously measured aggregation and piliation we chose to mine the results using the ratio between these two parameters. We plotted the ratio of the mutation score for aggregation over the mutation score for piliation for each amino acid. It can be noted that, as expected, the α1N mutants with short pili identified above also show an aggregation defect with this analysis (Fig. 4A, indicated in green).

To investigate the potential role of pilus-pilus electrostatic interactions in bacterial aggregation, we focused on the negatively and positively charged amino acids in the pilin sequence, indicated in red and blue respectively (Fig. 4A). Among them, we found three that show a decrease in aggregation upon mutation

(K103, K144 and H149) and one leading to a strong increase in aggregation upon mutation (E99). Losing the negative charge increased aggregation and, reciprocally, losing a positive charge decreased aggregation. Strikingly, these four amino acids delineate a cavity in the pilus structure which is overhung by the bulky lysine 140 (Fig. 4B). We therefore focused on this area and generated several mutations around

K140. Since we suspected that this region might be the epitope of the 20D9 monoclonal antibody we used for piliation quantification, we also used a nanobody (F10-I) directed against TFP to measure the piliation of these mutants (Charles-Orszag et al., in revision). While there was an excellent agreement between the

2 measurements made with 20D9 and F10-I (R =0.94), we found two outliers in this distribution: PilEQ122E and

PilEK140Q (Fig. 4C), showing that these two amino acids are important constituents of the 20D9 epitope. This observation was confirmed by immunofluorescence (Fig. 4D) and the level of piliation measured using the

F10 nanobody for these two mutants was used for the rest of the study. We then characterized the ability of mutants in the K140 area to form aggregates. The level of aggregation could be represented as a function of piliation per bacterium for the series of mutants around K140 as well as mutants generated earlier (Fig.

4E). As previously observed, below a piliation threshold, aggregation is lost 323. Above this threshold, aggregation increased progressively and was correlated with the piliation level. We could observe for

107 instance that the pilX mutant is below the threshold of piliation required to form aggregates, in agreement with a previous study 323 (Fig. 4E, indicated in dark blue). In contrast, mutations located in the K140 region appeared as two groups of outliers in the distribution. A group of two mutations with normal or increased piliation and a lower positive charge of the pilus (PilEQ122E and PilEK140Q) had a decreased ability to form aggregates (Fig. 4E-F, indicated in orange). On the contrary, mutations increasing the overall positive charge led to higher aggregation levels. Notably, PilED143Y had decreased piliation levels but normal aggregation levels (Fig. 4E-F, indicated in light blue). This supports the notion that local charge and organization in this region are crucial for aggregation and further indicates a central role of the positive charge of lysine 140 in aggregation (Fig. 4B).

We took advantage of the residual aggregation of PilEQ122E to characterize the ability of these mutated pili to mediate pili-pili interactions. To this end, we used co-aggregation experiments with PilESB. We could observe that the mutant with a decreased positive charge (PilEQ122E) is enriched at the edge of the aggregate suggesting that these less positively charged pili mediate weaker interactions (Fig. 4G,H). On the contrary, the mutant less piliated than PilESB but with an increased overall positive charge (PilED143Y) is enriched at the center of the aggregate suggesting that these more positively charged pili mediate stronger interactions

(Fig. 4G,I). This confirms that local charge modifications of the pilus are sufficient to modify interbacterial interactions 685.

The adhesion properties of the non-aggregative mutants were then analyzed. We could observe that these mutants adhered more than PilESB after 30 minutes (Fig. 4J) and had similar adhesion levels at 4 hours

(Fig. 4K). This indicates that aggregation and adhesion can be dissociated in these single amino acid mutants.

Therefore, we identified an electropositive region centered around K140 in the major pilin of TFP which is crucial for aggregation. We also show that modifying the local charge is sufficient to affect interbacterial interactions. This demonstrates the importance of this region of PilE in the aggregation process.

108

Figure 4: Auto-aggregation is dependent on a charged patch of amino acids centered around lysine 140. (a) Ratio of individual aggregation mutation scores over piliation mutation scores for each amino acid in the PilE sequence derived from library screening. Mean values ± SEM are indicated for each amino acid position. Positively charged amino acids (R, K and H) are indicated in blue. Negatively charged amino acids (E, D) are indicated in red. Mutants with short pili identified previously are highlighted in green. (b) Pilus (PDB: 5KUA) view highlighting a novel region involved in aggregation shows a dense cluster of charged amino acids surrounding the protruding K140. Light blue and orange-colored amino acids refer to the outliers defined in panel 4E. Dark blue region refers to the charged cavity composed of E99, K103, K144 and H149, surrounding the bulky K140. (c) Piliation per bacterium as measured by flow cytometry using 2 different reagents to label pili, the 20D9 monoclonal antibody and the F10-I nanobody. Values obtained are in very good agreement, with the exception of two mutants, indicated in orange. (d) Immunofluorescence of GFP-expressing bacteria. Pili were stained using 20D9 or F10-I as indicated. Images are representative of the appearance of piliated bacteria. Scale bar: 2µm. (e) Relative aggregation expressed as a function of relative piliation per bacterium (measured with 20D9 except for PilEQ122E and PiEK140Q for which the F10-I value is reported). Each dot represents a PilE point mutant. Values were normalized to those of the PilESB strain. The gray area corresponds to mutants below the aggregation threshold. Hypo-aggregative mutants are highlighted in orange while hyper-aggregative mutants are in light blue. Dotted lines indicate 95% confidence interval. (f) Quantification of aggregation for the two hypo-aggregative mutants and one of the hyper-aggregating mutants. Aggregation levels are expressed relatively to those of PilESB-GFP. Mean values ± SEM are indicated for each strain. N⩾3 independent experiments. (g) Representative images of co-aggregates of PilESB-mcherry and GFP- expressing indicated mutant after 30 minutes of incubation at a 1:1 ratio and an OD of 0.3. N⩾3 independent experiments. Scale bar: 15µm. (h-i) Relative bacterial occupancy in the co-aggregates, as a function of the distance from the aggregate center. For each aggregate, radial fluorescence intensities of PilESB-mCherry and GFP-expressing mutant strain were divided by the maximum intensity in the corresponding channel. Intensity distributions were averaged over several co-aggregates. Mean values ± SEM are indicated for each strain. N=3 independent experiments. (j) Quantification of adhesion after infection of HUVEC for 30 ⩾ minutes (MOI 500) by microscopy, expressed relatively to PilESB-GFP. Mean values ± SEM are indicated for each strain. N 3 independent experiments. (k) Quantification of adhesion after infection of HUVEC for 4 hours (MOI 100), expressed as Colony109 Forming Units. Mean values ± SEM are indicated for each strain. N⩾3 independent experiments.

Identification of two regions of PilE required for initial adhesion to human cells

The high-throughput evaluation of adhesion generated by deep mutational scanning was then analyzed to decipher how pili mediate adhesion at the structural level. Various studies have found that the ability of bacteria to adhere to human cells is tightly linked to their ability to form aggregates, presumably by allowing the accumulation of bacteria in three dimensions at the site of adhesion 444,638,686. The objective of this section was to explore whether regions of the pilin specifically involved in adhesion could be identified. We took advantage of the simultaneous measure of adhesion and aggregation to tease out the pili-host cell interaction region, by analyzing the ratio of these two properties. The ratio of the mutation score for adhesion over the mutation score for aggregation was averaged over a window of 5 consecutive amino acids to facilitate the identification of regions involved in adhesion. The distance of the amino acids from the N- terminus of the folded PilE monomer is color-coded (Fig. 5A).

Consistently with what we have previously shown, we could see that for mutations in the a1N adhesion was superior to aggregation (Fig. 3B,H and Fig.5A). On the contrary, we found two regions with a severe defect in adhesion: amino acids around Y50 and amino acids around T130 (highlighted in blue and green respectively). Interestingly, in the folded protein, these regions are found at the tip of the pilus as indicated by the red coloring in Fig. 5A. In particular, the conserved region surrounding Y50 has been shown to only be exposed at the tip of the pilus in previous studies 486,639,687.

By characterizing the early TFP-mediated adhesion of mutants generated de novo throughout the study (30 min), we found that the most significant phenotypes (more than a 2.5-fold loss in adhesion) correspond to mutations at the tip of the pilus (Fig. S6A). By de novo mutating residues in these two regions, we were able to find mutations that showed very low adhesion while maintaining the ability to aggregate. (Fig.5B, indicated in blue and green). These mutants have similar phenotypes as the pilC1 mutant typically described as a non-adhesive yet aggregative mutant. This suggests that aggregation and initial adhesion are two distinct functions mediated by TFP that become linked at later time points when bacteria proliferate and start to form aggregates on the cellular surface.

A role for the pilus tip in adhesion could explain why, as described above, pilus number rather than length could be important for adhesion. Under this hypothesis, a prediction would be that the piliation level per

110 bacterium which is mostly related to pilus length, would not correlate with early adhesion. Accordingly, we could not find any statistical correlation between the levels of piliation per piliated bacterium and early adhesion in the mutants characterized in this study (Fig. 5C). In contrast, and as would be expected from the above results characterizing aggregation, after 4 hours of infection, the adhesion of a subset of these mutants correlated tightly with their piliation levels (Fig. S6B-C). This confirms that early adhesion is not dependent on the amount of pili as detected by cytometry, which is in support of the hypothesis of a tip- mediated adhesion as this mechanism would not be influenced by pilus length.

Overall, this mutational analysis is thus in favor of a tip-mediated adhesion mechanism. At the structural level, amino acids that affect initial adhesion upon mutation form a structure with a three-fold symmetry at the pilus tip (Fig 5D) each including a region close to Y50 and T130.

111 Figure 5: PilE is involved in early adhesion through residues displayed at the tip of TFP.

(a) Ratio between the adhesion and aggregation mutation scores for each amino acid in the PilE sequence. The average ratio of these scores was computed and then averaged over a moving window of 5 amino acids. Background colors indicate the axial position of the amino acid in PilE as color-coded on the structure on the right of the graph (distal tip in red, proximal tip in yellow). Blue and green bars highlight the two regions identified to be important for adhesion. (b) Early adhesion of PilE mutants as a function of aggregation. Values are normalized to that of PilESB-GFP. Each dot represents a PilE mutant except for pilX and pilC1 mutants (highlighted in purple). Mutants with conserved aggregation and diminished adhesion are indicated in green and blue. A blown-up view of the zone of interest (lower quarter) is presented on the right panel. (c) Early adhesion of PilE mutants as a function of piliation per bacterium (measured with 20D9). Values are normalized to that of PilESB-GFP. Each dot represents a PilE mutant. (d) Amino acids presented in panel b are highlighted in blue and green on the structure of PilE and the pilus. On the left, top view of the pilus. On the right, close-up showing the putative binding site displayed on a single PilE subunit.

112 TFP-mediated adhesion to human cells is initiated at the tip of the pilus

The previous findings prompted us to examine microscopically how pili adhere to cells to evaluate whether tip-mediated adhesion could be visualized. Using crude pili preparations and spinning disk confocal microscopy, we could detect pili adhering to the surface of HUVEC cells by immunofluorescence. To evaluate whether they were adhering through their tip or on their side, we applied flow on these cell-bound immunolabeled pili (Fig. 6A-B). We could observe that mechanical constraints generated by the flow led to a change in pilus position, bending it down onto the cell surface in the direction of the flow as shown in Fig.

6B and Movie S1. This result is in favor of a scenario where pili bind to cells by an extremity. In principle, the purified pili used in this experiment could interact with cells through their distal tip or through their hydrophobic proximal base. To distinguish between these two situations, we used the SM1 monoclonal antibody which specifically labels the distal tip of the pilus 605,687. After adhesion of crude pili preparations to cells, we labeled pili with both the F10-I nanobody and SM1 monoclonal antibody and looked for fields of view where we could visualize both pili bound to host cells and pili bound to the substrate. All pili were labeled by the F10-I monoclonal nanobody. Yet, while pili bound to the substrate were labeled by SM1 at the tip, pili bound to cells were not, suggesting that the epitope is covered by the cell membrane (Fig. 6C).

We could also observe on the xz-views that substrate-bound pili lied flat on the substrate while host cell- bound pili were anchored to the cell surface by the tip. Altogether, this strongly argues that TFP bind human endothelial cells through their distal tip. Finally, we reasoned that binding of SM1 to the tip of the pilus should prevent adhesion to host cells. Upon incubation of bacteria with the SM1 antibody for 15 minutes prior to adhesion, we observed a dose-dependent inhibition of early bacterial adhesion and obtained a significant inhibition with 20 ng.mL-1. Incubation of bacteria with similar amounts of the 20D9 monoclonal antibody had no effect on early adhesion (Fig. 6D). The binding of SM1 to the tip of TFP strongly supports the hypothesis that PilE is indeed the protein present at the tip of TFP. Inhibition of early adhesion by incubation of bacteria with this antibody argues that adhesion to human cells is initiated at the tip of the pilus by PilE.

113 Figure 6: TFP-mediated adhesion to human cells is initiated at the tip of the pilus.

(a) Representation of the experimental setup. Pili crude preparations are added on top of HUVEC cells and flow is applied. If pili are anchored to the cell surface by the tip, flow should bend down the pilus towards the cell. This change of orientation is visualized by spinning-disk confocal microscopy. Figure was adapted from the Servier medical art library. (b) Time-lapse of a cell-adhering pilus fluorescently labeled with F10 at three different focal planes, with z3 being the higher focal plane and z1 the lower focal plane, closest to the cell surface. Flow is initiated at t0. Arrow indicates flow direction. Scale bar, 1µm. Video presented in Movie S1. (c) Spinning-disk confocal immunofluorescence of pili stained with both SM1 (magenta) and F10-I (green). In each field of view, a substrate- and a cell-anchored pilus are presented. Upper panel: z-projection from the stacks, bottom panel: side view. Dotted line outlines the orientation of pili, evidencing specific tip-anchoring of TFP to cells. Scale bar, 2µm. (d) Quantification of adhesion after infection of HUVEC cells for 30 minutes, expressed as number of GFP-expressing adherent bacteria over number of cell nuclei. Prior to infection, bacteria were incubated at 37°C with or without antibody for 15 minutes. Mean values ± SEM are indicated for each strain. N=5 independent experiments.

114 Discussion:

How TFP mediate such a vast array of functions remains an unanswered question. Using deep mutational scanning of the major pilin PilE, we have uncovered new insights on pilus assembly and pilus-mediated aggregation and adhesion (Figure 47).

In this study, we first identified regions of the major pilin involved in the display of TFP on the bacterial surface. The introduction of a cytometry-based technique to measure piliation was critical for the investigation by providing a quantitative analysis of the piliation levels per individual bacterium in contrast with previous studies. As an illustration of the type of information offered by this approach, we describe a hydrophobic patch of three highly conserved amino acids necessary for piliation because of their role in maintaining the proper tertiary structure of PilE. This approach also led to an unexpected observation related to the amino terminal portion of the protein: the a1N region of PilE is quite tolerant to mutations and appears to regulate piliation dynamics. Mutations in this area of the protein lead to the formation of short but numerous pili. Interestingly, in these mutant strains, bacteria failed to aggregate because of the reduced length of their pili. A role for the a1N region in the balance between pilus length and number is in contrast to the current view supported by structural data that suggest that the hyperconservation of this region is critical for pilus assembly 307,308. Experimental results in line with our observations can however be found in the literature. Data obtained in P. aeruginosa show that the amino terminal region can be mutated without losing pilus expression 688. In addition, mutants isolated in N. gonorrhoeae and V. cholerae based on their inability to form microcolonies while retaining pilus expression contained very similar point mutations in the amino terminal region of the pilin 449,686. Using Coomassie staining of pili preparations, these studies concluded that this phenotype was not linked to defects in piliation. Taking into consideration the results of our study we would predict that these mutants are less piliated at the individual bacterium level but that a larger proportion of cells in the population is piliated, which would explain their inability to form aggregates.

Several mechanisms could account for this new role of the amino-terminus of PilE we observed. This region has been proposed to be involved in regulation of self-expression of the major pilin PilA in P. aeruginosa

689, yet, we have shown that this is not the mechanism at play here. We would favor a hypothesis where mutations in the amino-terminal part of the protein affect interaction with the piliation machinery. This portion

115 of PilE is buried in the inner membrane prior to pilus assembly. It is therefore possible that the interactions of the a1N of PilE with membrane lipids is affected upon mutation, thereby modifying the accessibility of

PilE to the machinery. In N. meningitidis, PilE has been shown to interact with several components of the machinery (PilG, PilN and PilO) at least in part through its N-terminal domain 349. Additional interactions have been identified in P. aeruginosa between the major pilin and both the minor pseudopilin FimU (PilH in

N. meningitidis) and the minor pilin PilE (PilV and PilX in N. meningitidis) 329,425. All these proteins have been proposed to participate in the initiation of pilus biogenesis. To the best of our knowledge, such interactions have not been investigated in meningococcus but could explain the peculiar phenotypes displayed by a1N mutants. The molecular determinants behind the regulation of piliation we describe here will need further investigation.

We next identified a region of PilE required for aggregation. The importance of lysine 140 for pilus bundling and bacterial adhesion had been suggested based on sequence comparison of hyper- and hypo-adherent variants 444 and more recently by structural data 307 where it appears as a hook-like structure protruding from the pilus. Yet, no mutagenesis study had been conducted to convincingly show this. Here, we provide compelling evidence that lysine 140 and the electropositive cavity in its vicinity are crucial for aggregation.

The role of this local positive charge is supported by the detrimental effect observed upon introduction of a negative charge in this region (PilEQ122E has an aggregation defect) and the beneficial effect observed upon withdrawal of a negative charge (PilED143Y is hyper-aggregative). Co-aggregation experiments further support the importance of this region in determining pili-pili interaction forces. Yet, the molecular mechanism of pili-pili interaction through lysine 140 remains elusive. Two mechanisms are possible: whether the lysine

140 from one pilus interacts with the lysine 140 of another pilus (e.g. through coordination of metal ions) or lysine 140 from one pilus interacts with a different region of another pilus (e.g. through salt bridges). More structural and biochemical work is required to further decipher the molecular mechanism of aggregation.

While there is only limited conservation of these residues across species, the results we obtained provide new evidence for a widespread role of electrostatic interactions between bacterial TFP in the function of aggregation.

A major finding of this study is that adhesion is initiated at the tip of TFP by PilE. This is supported by several lines of evidence. Data from deep mutational scanning indicates that amino acids located at the tip of TFP

116 are important for adhesion. This was validated by quantifying early adhesion of pili tip mutants to human cells. Initial adhesion of bacterial cells could be inhibited by the SM1 monoclonal antibody which is specific for the tip of the pilus and bacteria with short pili adhered just as well as bacteria with long pili. Finally, direct microscopic visualization of the adhesion of purified pili to human cells also showed that they bound cells via their distal tip. Furthermore, such a mechanism of binding has previously been reported in P. aeruginosa by imaging fluorescently labeled pili of live bacteria adhering to quartz 428. Here, we provide evidence for the general relevance of this mode of adhesion in the context of adhesion to human cells.

Remarkably, we observed that the initial adhesion site contains the SM1 epitope 605 and that early adhesion can be inhibited using this antibody. If an adhesin other than PilE was located at the tip of TFP and responsible for the tip-dependent binding to cells, we should not observe inhibition of adhesion by SM1 as it specifically binds PilE. This is in favor of a direct role of PilE in adhesion at the tip of TFP and implies that the tip of the pilus is free of other proteins that would cover it, contrarily to what was suggested for PilI-K in the type II secretion system 423. Whether this results in the presence of three binding sites at the tip of the pilus or one oligomeric site is presently unclear. Interestingly, the SM1 site is very well conserved and was used to define class I pili 605,690. This epitope has also been shown to be exposed upon stretching of TFP in

N. gonorrhoeae 486 and upon binding to endothelial cells for N. meningitidis 639. It could indicate that after initial cell binding via the tip of the pilus, the fiber is stretched upon pilus retraction and/or because of the shear stress exerted by the blood flow. Pilus stretching would therefore uncover new binding sites and enable firmer cell binding along the length of the pilus fiber. This two-step mechanism would reconcile our data and the biophysical evidence indicating that pili can be stretched upon force application and adhere over extended lengths with hydrophobic surfaces 691,692. Unlike what has been described in the literature so far for other types of pili 112,693, such a process would consist in a catch bond mechanism at the polymer level and not at the single adhesin monomer level. This means that initial cell-attachment of PilE at the tip of TFP enables deformation of the fiber upon tensile force application which results in an increased adhesion of TFP to cells due to the progressive unmasking and attachment of new PilE binding sites along the fiber via a zip-in mechanism 694.

A direct role of PilE in TFP functions remains quite controversial. In particular, PilC1 and minor pilins PilV and PilX have been shown to be required for adhesion 431,592,638,676. While the minor pilins PilV and PilX have

117 been proposed to be inserted in the fiber of TFP, a recent report rather suggests that they exert their functions from the periplasm 323. Quite similarly, an initial report indicated that PilC1 is located at the tip of the pilus 624, but it has been shown and is now widely accepted that PilC1 is primarily exposed at the outer membrane of the bacterium 625. In this article, we demonstrate that initial adhesion is mediated by the tip of

TFP. Yet, we also show that early adhesion is dependent on PilC1, PilV and PilX. The previous lines of evidence tend to indicate that these three proteins do not exert their adhesion function at the tip of the pilus.

How to explain their adhesion defects, then? pilV and pilX mutants have been shown to express less pili at their surface 323. Because a decrease in the number of pili results in a decrease in the number of tips available for adhesion, this could explain the adhesion defect we observe in pilV and pilX mutants. We observed a decrease in piliation of the pilC1 mutant by flow cytometry (data not shown) and we hypothesize that piliation defects similar to those observed for pilV could explain the adhesion defect of pilC1 mutants.

This is further supported by the phenotypic similarity we observe between PilV and PilC1 (Fig. 5C) and the fact that pilC1/2 mutants are non-piliated 592. All these observations strongly support that PilE is the direct mediator of TFP tip-dependent adhesion. While subsequent interactions may be at play between TFP and human cells, adhesion via the tip is likely the only interaction necessary to initiate adhesion. Consequently, the hyperconserved region of PilE we identified to be involved in adhesion makes an appealing therapeutic target to prevent adhesion of meningococcus (and presumably other pathogens expressing class I pili) to human cells and subsequent vascular damages.

118 Material and methods:

Antibodies and chemicals:

The following antibodies were used for Western blots and immunofluorescence: (i) Polyclonal serum anti-

PilE 695, anti-PilV 431 and anti-RMP4; (ii) Mouse monoclonal antibody, anti-PilE mouse monoclonal antibody clone 20D9 682, and clone SM1 696 (iii) Camelidae nanobody, anti-PilE clone F10-I (Charles-Orszag et al, in revision). The following goat secondary antibodies were used for immunofluorescence, Western blot and flow cytometry: anti-mouse or anti-rabbit IgG (H+L) coupled to horseradish peroxidase (Jackson Immuno-

Research Laboratories) and anti-rabbit or anti-mouse IgG (H+L) coupled to Alexa Fluor 488, 568 or 647

(Life Technologies) and mouse anti-his tag (Biolegend). 40,6-diamidino-2-phenylindole (DAPI) was purchased from Life Technologies and Hoechst 33342 from Invitrogen. Trypsin-EDTA (0.05%) was purchased from Gibco.

Bacterial growth conditions and mutagenesis

Neisseria meningitidis 8013/clone 12 (2C43) strain expressing the SB pilin variant was used in this study

595. In particular, a strain containing a deletion of the guanine quartet upstream of the pilE gene 681 was used to minimize antigenic variation, see below for the construction of this strain. Bacteria were grown on

Gonococcus Medium Base (GCB, Difco) agar plates supplemented with Kellogg’s supplements 697 and, when required, 100 µg/ml kanamycin, 2 µg/ml erythromycin, 50µ/ml spectinomycin, or 5 µg/ml chloramphenicol at 37°C in moist atmosphere containing 5% CO2. Before adhesion and aggregation assays, bacteria were grown shaking at 37°C -5% CO2. Bacteria were inoculated at an OD of 0.05 in Human

Endothelial-SFM (Gibco) supplemented with 10% heat-inactivated FBS and grown for 2h-2h30. Escherichia coli was grown on liquid or solid Luria-Bertani medium (Difco) containing 20 µg/ml chloramphenicol, 50

µg/ml spectinomycin, 30 µg/ml kanamycin, or 200 µg/ml erythromycin when necessary at 37°C. Guanine quartet mutant was generated as described by Tan el al. 681 using fusion PCR with primers PK-30,31,32 and 33 and a chloramphenicol resistance cassette (see able 1). The insertion of the chloramphenicol resistance cassette was verified by PCR. PilE point mutations. In order to increase transformation efficiency in N. meningitidis, we modified the previous vector (TOPO-SB-Kan 323) used for PilE mutagenesis as follows: we amplified genomic DNA of the strain transformed with TOPO-SB-Kan by PCR using the primer pair PK-

119 10/21. This fragment was then cloned in Topo-pCR2.1 (Invitrogen). This plasmid was named TOPO-SB2-

Kan and used for pilE mutagenesis. To generate point mutations in pilE, we used a single-primer one-step mutagenesis process698. Briefly, a mutagenic primer (listed in Table 1) was used to amplify plasmid Topo-

SB2-Kan by PCR using Phusion polymerase. The PCR product was digested with FastDigest DpnI at 37°C for 8 hours and then inactivated at 80°C for 5 minutes. This product was transformed in XL1 Blue cells which were then selected on Kanamycin-containing LB plates. Plasmids from single colonies were sequenced and transformed in N. meningitidis as described above.

Mutant library generation and analysis

Random mutagenesis was performed using the GeneMorph II Random Mutagenesis Kit. Following manufacturer’s instructions, the mutagenic PCR was run with Mutazyme II polymerase using the modified

Topo-SB2-Kan described earlier as template and primers PK-13 and PK-14. Mutagenic PCR products were purified and used as megaprimers to PCR amplify TOPO-SB-Kan with Phusion polymerase. The PCR product was digested with FastDigest DpnI at 37°C for 8 hours and then inactivated at 80°C for 5 minutes.

This product was transformed in XL1 Blue cells then selected on Kanamycin-containing LB plates. Mixed plasmids from all the transformants were purified and transformed in N. meningitidis. Transformants were selected on plates containing kanamycin, collected the next day and kept frozen until submitted to selection.

Piliation selection. Bacteria from the library were treated as described in the flow cytometry section.

Bacterial suspension was then sorted with MoFlo® Astrios for 1h and collected in warm Human Endothelial-

SFM +10%FBS. The sorted population was then plated on Kanamycin-containing GCB plates. Adhesion selection. Bacteria from the library were used to run a 2h adhesion assay following the protocol described below at the exception that all of the final cell suspension was plated on Kanamycin GCB plates. Aggregation selection. Bacteria from the library were left to aggregate at an OD of 0.6 for 2hours in Millicell Cell Culture

Inserts for 24-well plates with a 3 µm pore diameter. Medium was flown-through several times. The membrane was then vortexed in medium to collect the bacteria which were then plated. This process was repeated two times to significantly enrich the library in aggregation+ bacteria. Genomic DNA was then extracted from the different libraries. DNA from each library was amplified by two separate PCR (20 cycles) using two reverse pairs of barcoded primers (Primers PK-34 to PK-47) to obtain a more homogeneous coverage upon sequencing. The PCR products were purified, and concentration was normalized using

120 Fragment analyzer (AATI) prior to sequencing. Paired-end sequencing (2 x 300bp) was performed using an

Illumina Miseq sequencer at the Pasteur genomics platform. Fastq files were then assembled, filtered and analyzed for single nucleotide variations using CLC Genomics workbench. The read counts for single amino acid variation were further analyzed using the EdgeR-dedicated script in the SARTools package 699,700.

Resulting p-values and log values of the ratio were used for downstream analysis.

Each selection was done at least in triplicates on three separate days. The whole selection process was done twice, and the two biological replicates were sequenced twice. This means that we ran a total of 4 sequencing-runs (two biological duplicates and two sequencing duplicates).

Flow cytometry

Bacteria were resuspended directly from the GCB plates in Human-Endothelial medium+10%FBS to an OD of 0.1 and incubated for 30minutes at 37°C. 1µL of an antibody mix of 20D9 and anti-mouse IgG (H+L) coupled to Alexa Fluor 488 or 647 each at 100µg.mL-1 or 1µL of an antibody mix of F10-I at 30µg.mL-1 and mouse anti-his tag at 100µg.mL-1 and anti-mouse IgG (H+L) coupled to Alexa Fluor 488 or 647 at 100µg.mL-

1 was added to 10µL of the bacterial suspension and incubated at 37°C for 15minutes. 500µL of warm medium was added to the tubes and samples were analyzed using a Gallios cytometer (Beckman Coulter).

Signal was gated in order to exclude cell doublets or bigger aggregates.

Aggregation assays

GFP-bacteria precultured for 2 hours up to an OD of approximately 0.2 were pelleted (14200g for 1min), washed and concentrated at an O.D. of 0.6 in Human Endothelial-SFM + 10%FBS. 500 µL of this suspension was dispensed in a well of a 96 well µ-plate with square wells (Ibidi). Plates were incubated for

30min at 37°C in moist atmosphere containing 5% CO2 and 3 fluorescence images per well were captured using a 4x objective. The surface covered by bacterial aggregates was quantified in each field of view using a homemade macro in Fiji 701. Co-aggregation assays were performed similarly to aggregation assays with the exception that bacteria were concentrated to an O.D. of 0.3 and that 250 µL of one bacterial suspension was mixed with 250 µL of another bacterial suspension. Bacterial localization in co-aggregates was determined using the Radial Profile Plot ImageJ plugin on aggregates of similar size.

Adhesion assays

121 Human umbilical vein endothelial cells (HUVEC - PromoCell) were used between passages 1 and 8 and grown in Human Endothelial-SFM (Gibco) supplemented with 10% heat-inactivated FBS (PAA Laboratories) and 40 µg/ml of endothelial cell growth supplement (Sigma-Aldrich) and passed every 2-3 days. bEnd.3 cells were grown in DMEM (Gibco) supplemented with 10% heat-inactivated FBS. For 2-4h adhesion, 105 cells were seeded in Costar® 24-Well TC-Treated plates the day before infection. On the next day, cells were rinsed once and infected with midlog-phase bacteria (MOI 100) at 37°C-5%CO2. Inoculum was plated and quantified by CFU count. After 30minutes of infection, cells were rinsed 3 times and incubated for 90 min. Again, cells were rinsed 3 times. Infection was stopped at this stage for 2h adhesion but pursued for 2 more hours for the 4h adhesion. Cells were rinsed 3 times and incubated in trypsin-EDTA (0.05%) until detached. Trypsin was inactivated by adding culture medium. The infected cells were vortexed and plated at appropriate dilutions on GCB plates. Ratio of CFU after infection over CFU from the inoculum was compared between strains.

For early adhesion assays, 3.5 104 cells were seeded in Cellstar® 96 well plates (Greiner) the day before infection. On the next day, cells were rinsed once. Hoechst was added to the medium at a final concentration of 1 μg/ml and cells were infected with midlog-phase GFP-bacteria (MOI 500) for 30minutes at 37°C-

5%CO2. Wells were rinsed 3 times and then fixed with 4% formaldehyde for 30min at room temperature.

Wells were rinsed 3 times with PBS and then imaged with a 40x objective. Each infection was run in duplicate and 16 images were taken per well. A homemade macro in Fiji was used to count the number of bacteria and nuclei per field of view. These values were summed for each well and a ratio of the total number of bacteria over the total number of nuclei was used to evaluate the number of bacteria per cell.

Scanning electron microscopy of type IV pili

Hydrophilic mixed cellulose esters filters, 13 mm in diameter containing 0.025 µm pores (MF-Millipore,

VSWp01300) were placed in 24-well plate, layered with 500 µL of a bacterial suspension at an OD600nm of

1 in Human Endothelial-SFM + 10% FBS and incubated for 30 minutes at 37°C and 5% CO2 in a humidified incubator. 500 µL of pre-warmed 8% PFA in 0,1M HEPES pH 7.4 was then added to the well and incubated at room temperature for 45 min. After three washing steps in HEPES, type IV pili were blocked for 20 minutes in HEPES-0.2% gelatin (HEPES-G) and incubated with 2 µg.mL-1 20D9 antibody followed by 10 µg/mL Anti

Mouse-AlexaFluor 568 in HEPES-G, each for 1h at room temperature. Samples were then post-fixed

122 overnight at 4°C with 2.5% EM grade glutaraldehyde in HEPES, washed in HEPES, post-fixed with 1%

OsO4 in HEPES for 1h, washed in distilled water, dehydrated in graded series of ethanol (25, 50, 75, 90 and

100%), critical point dried in liquid CO2 in a Leica EM CPD300, mounted on aluminum stubs and sputter- coated with 15 nm gold/palladium in a Gatan PEC 682 gun ionic evaporator. Imaging was performed in an

Auriga scanning electron microscope (Carl Zeiss) operated at 7 kV with an in-lens secondary electrons detector.

Pili purification:

Crude pili preparations were obtained by suspending bacteria to an OD of 10 in cold PBS. Bacteria were vortexed for 1min at max speed and then incubated on ice for 2min. These steps were repeated two more times. To separate pili from debris and bacteria, the suspension was then centrifuged for 5 minutes at 9000g.

This was repeated twice.

Immunofluorescence:

For immunofluorescence, samples were fixed with PBS containing 4% formaldehyde for 30minutes at room temperature. After rinsing, samples were blocked with PBS containing 0.2% skin fish gelatin (PBSG) for

30minutes. Samples were then incubated with appropriate combinations of the following: 20D9 diluted in

PBSG at 2.5µg/mL and/or SM1 diluted at 2µg/mL or F10-I diluted at 4µg/mL at for 1hour and subsequently incubated with the secondary antibody anti-mouse-AlexaFluor 491 or 568 diluted in PBSG at 10 µg/mL and/or Dylight650 (Invitrogen) conjugated mouse anti-his tag for 1hour at room temperature or overnight at

4°C.

SDS–PAGE and Western blot:

Protein samples preparation, SDS–PAGE separation, transfer and immunoblotting were performed using standard techniques 702. Proteins were separated by SDS–PAGE in Tris-glycine gels containing 12–15% acrylamide, transferred onto PVDF membranes (Thermo Scientific) using semi-dry electrotransfer (Biorad). Membranes were blocked with PBS + 0.1% Tween-20 + 5% milk for 30 minutes and incubated for 1 hour with polyclonal antibodies (anti-PilE at 1/5,000, anti-PilV at 1/1,500) diluted in blocking solution. Membranes were then incubated with horseradish peroxidase (HRP)-coupled anti-rabbit antibody (1/10,000) in the blocking solution. Peroxidase activity was detected by enhanced

123 chemiluminescence (ECL-plus, Pierce) and recorded with a G:BOX Chemi multi-purpose imaging system from Syngene. Protein quantities were analyzed using the Fiji software.

Primers and strains used in this study:

See supplementary tables 1 and 2.

Statistical analysis

Graphpad was used to plot graphs and run appropriate statistical tests. Statistical significance was defined by p<0.05 (*), p<0.01 (**) and p<0.001 (***).

Competing interests:

No competing interests to declare.

Acknowledgements:

This work was supported by the Integrative Biology of Emerging Infectious Diseases (IBEID) laboratory of excellence; The VIP European Research Council starting grant (GD); Paris Descartes University PhD fellowship, the FRM PhD fellowship (FDT20170437205) and Programme Bettencourt (Ecole Doctorale

Frontières du Vivant, FdV) funding. NGS was performed at the Genomics Platform, member of “France

Génomique” consortium (ANR10-INBS-09-08). We acknowledge the ultrapole platform and cytometry platform of the Center for Innovation and Technological Research (CRT) at Institut Pasteur for support in conducting this study. We would like to thank Mumtaz Virji and Darryl J Hill for providing the SM1 antibody.

We are also extremely grateful to Daria Bonazzi and Olivera Francetic for critical reading of the manuscript and to all members of the Dumenil lab for helpful discussions.

124 Supplementary figures:

Figure S1: Obtaining near-saturating libraries with single point mutations

(a) Number of single amino acid protein variants observed in each library. Numbers were calculated using data from the sequencing runs. Dotted line indicates the theoretical maximum number of variants that could have been obtained by single- nucleotide mutations of pilE.

(b-d) Distribution of mutation counts per pilE read in each library. Frequency in the initial library follows a Poisson distribution with maximal for 1 mutation per read. Only results from single mutations have been used in the study.

125

Figure S2: Validation of selection methods

(a-b) Flow cytometry analysis of pili expression using the 20D9 monoclonal antibody. The Pili + gate was used to separate piliated and non piliated bacteria. pilD mutants appear as non-piliated while pilV mutant is hypo-piliated and the pilT mutant is hyper-piliated. Insert in panel B is a blown-up view of the piliated population. (c) Ratio between the colony forming units of pilESB and pilD-GFP recovered from the filter and filtrate fractions after 2 hours of static co-incubation in transwell inserts. Mean ratio ± SEM is indicated. N≥3 independent experiments. (d) Brightfield images showing the aggregation of the initial mutant library before selection, after 1 round of selection and after 3 rounds of selection. Scale bar: 200µm. (e) Ratio between the colony forming units of pilESB over pilF-GFP or pilV-GFP recovered from HUVEC cells after 4hours of infection (MOI100). Appropriate dilutions of adherent bacteria were plated on antibiotic-supplemented GCB plates. Mean ratio ±SEM is indicated. N≥3 independent experiments.

Figure S3: Pilin quantification in whole cells

Quantification of α-PilE and α-Rmp4 blots of whole cell lysates. PilE levels were normalized using Rmp4 and values are normalized to pilESB values. Mean ratio ±SEM is indicated. N=3 independent experiments.

126

Figure S4: Pilus length is critical for aggregation and affected by mutations in the N-terminus (a) Volcano plot of piliation mutation scores for all PilE mutations. Some outliers from Fig. 3A are highlighted. (b) Illustration explaining the 3 different parameters that can be extracted from analysis of piliation by flow cytometry, population piliation, piliation per bacterium and percentage of piliated bacteria. (c) Flow cytometry scatter plot (side scatter intensity as a function of pili fluorescence intensity) of pilESB and pilD stained with 20D9. Piliated bacteria are located in the rectangular gate. (d) Percentage of piliated bacteria in the total population detected by flow cytometry in pilESB and pilD strains. Mean ratio ±SEM is indicated. N=7 independent experiments. (e) Population piliation as a function of the proportion of piliated bacteria normalized relatively to PilESB values as measured by flow cytometry with the 20D9 monoclonal antibody. Black dots indicate all the PilE mutants that were characterized. Blue dots indicate mutants of the piliation machinery. Dotted lines indicate 95% confidence interval of the linear fit. (f) Representative immunofluorescence images of GFP-expressing PilE mutants using the 20D9 monoclonal antibody. Scale bar: 1µm. (g) Quantification of α-PilE and α-Rmp4 blots of whole cell lysates. PilE levels were normalized using Rmp4 and values are normalized to pilESB values. Mean ratio ±SEM is indicated. N=3 independent experiments. (h) Ratio between the aggregation and piliation mutation scores for each amino acid in the PilE sequence127. The average ratio of these scores was computed and then averaged over a moving window of 5 amino acids. Zones significantly deviating from zero are highlighted and the N-ter of PilE is indicated with a black line.

Figure S5: Characterization of early adhesion

(a) Z-projections of images captured after 30 minutes of infection. Adherent GFP-expressing bacteria are in green and nuclei (stained with Hoechst) in blue. Scale bar: 50µm. (b) Number of adherent bacteria per cells as a function of MOI. Dotted line: linear fit of the data. (c-f) Quantification of adhesion of the indicated mutants after infection of HUVEC cells for 30 minutes, expressed as number of GFP-expressing adherent bacteria over number of cell nuclei. Mean value ±SEM is indicated. N≥3 independent experiments.

Figure S6:

(a) Axial position of the point mutations as a function of normalized early adhesion of the corresponding PilE mutants. The red area delineates mutants at the tip of the pilus (highest z values) which display the strongest adhesion phenotypes (points at the exterior of the central zone of pilESB-like adhesion, marked by black dotted lines). (b-c) Similar mutants are presented on both graphs. Each dot represents a PilE mutant unless otherwise indicated. The dotted line corresponds to a linear fit of the data. (b) 4-hour adhesion to HUVEC cells as a function of piliation per bacterium (measured with 20D9). Values are normalized to that of pilESB-GFP. (c) 30-minute adhesion to HUVEC cells as a function of piliation per bacterium (measured with 20D9). Values are normalized to that of pilESB-GFP. N≥3 independent experiments.

128 2 Additional results

2.1 Characterizing the importance of PilE in competence for transformation

Just like for piliation, adhesion and aggregation, we investigated the role of PilE in competence using deep mutational scanning. In order to do so, we incubated the initial mutant library with DNA containing an antibiotic resistance cassette. Recombinant bacteria were subsequently selected on antibiotic-containing plates. These bacteria formed the competence library. We describe these results in the following section.

We used a similar analysis method to the one we had used earlier for adhesion and tried to identify domains in PilE that played a significant role in competence. This led to the identification of four patches of amino acids (Figure 47a). These patches form a single cluster of amino acids (in red on Figure 47c). This cluster lies at the interface between the globular heads of two PilE subunits and resides at the intersection between two grooves that run through the pilus surface (arrows on Figure 47c). Interestingly, at the level of this cluster of amino acids, these two grooves narrow to a diameter of approximately 2nm (yellow arrow on Figure 47c). This diameter is compatible with the dimensions of the DNA double helix and supports the hypothesis that these furrows could play their role in competence as binding sites for DNA.

We then examined some of the mutants we had previously generated and characterized their competence in order to detect additional sites important for competence (Figure 47b). This led to the identification of two new patches (in blue on Figure 47c): additional residues around the 2nm bottleneck described earlier and residues 74-75 that lie at the center of one of the two grooves (represented by a blue arrow in Figure 47c). This could indicate that this groove is a DNA binding site. Yet, more extensive studies of the DNA-binding ability of the mutant pili are required to reach such conclusions.

Another observation we made was that the most N-terminal amino acids appeared to be important for competence (Figure 47a). As these residues are unlikely to be exposed in the extracellular milieu, the explanation for this phenotype has to be indirect. We have shown earlier that mutations in these N-terminal amino acids resulted in the presence of short pili at the bacterial surface. This could explain why these mutations result in lower competence for transformation. We therefore decided to further characterize the short pili mutants from the previous study to better understand their phenotype.

129

Figure 47: Identification of two putative DNA-binding grooves at the surface of the pilus. (A) Ratio between the competence and piliation mutation scores for each amino acid in the PilE sequence. The average ratio of these scores was computed and then averaged over a moving window of 5 amino acids. Important regions are highlighted in red. (B) Relative competence as a function of piliation per bacterium of PilE mutants. Outliers are indicated in blue. Mean

values are indicated for each strain. N⩾3 independent experiments (C) Structure of the pilus showing the amino acids we identified to be important for competence (in red: amino acids from panel A, in blue: from panel B). The inset helps to visualize the two grooves indicated by dotted arrows) passing through the 2nm narrowing formed by the red cluster of amino acids.

2.2 Explaining the phenotype of the “short pili” mutants 2.2.1 Mutants with short pili have retractile pili

We could observe that pilEL3H was 10-fold less competent than the reference strain and the competence of the other short pili mutants were not strikingly affected. In contrast the mutant

with long pili (pilES34A) displayed an increased competence (Figure 48c). These elements do not rule out that pilus length can influence competence for transformation, but they do not confirm its necessity. Because competence is known to be dependent on TFP retraction, we

next investigated TFP retraction in these mutants by examining their motility. Again, PilEL3H displayed a severe defect in twitching motility but the other mutants twitched at similar or higher speed than the reference strain (Figure 48a). The fact that mutants with short pili have a

130 similar motility to that of the reference strain is unexpected given that motility is driven by retraction of pili attaching at a significant distance. We decided to examine the displacements of these bacteria in more detail by evaluating their directionality. This revealed that movements of the reference strain had a higher directionality compared to that of the mutant strains and confirms that the reference strain has a more efficient motility (Figure 48b). We could also

confirm that the pilEL3H mutant is in a subdiffusive regime and cannot readily displace. Just like for competence, this defect in twitching motility could be due to a retraction defect or to the fact that pili are too short to mediate significant motility. In order to differentiate between these

two options, we deleted the pilT gene in the pilEL3H mutant. Indeed, if the PilEL3H TFP are non-

retractile, we should not observe a difference between the piliation of the pilEL3H mutant and

the piliation of the pilT pilEL3H double mutant. Yet, we could see that deleting pilT in the pilEL3H

mutant had very similar effects on piliation to deleting pilT in the PilESB strain (Figure 48d). This

demonstrates that the phenotypes we observed in pilEL3H are not due to a retraction defect but rather to the size of the pili it expresses. This also shows that the increase in the proportion of piliated cells in these mutants is not due to a retraction defect, as a pilT mutation results in the doubling of this proportion (Figure 48d). We then investigated the origin of this phenotype in more detail.

2.2.2 A role for minor pilins in pilus assembly

We know from the literature that the N-terminus of PilE potentially interacts with proteins from the alignment subcomplex PilG, PilN and PilO349, the minor pseudopilin PilH703,704 and probably the minor pilins PilV, PilX and ComP. To understand the interactions that could mediate an increase in the percentage of piliated cells we examined the structure of the pilus. The observation of this structure reveals that the amino acids disturbing the percentage of piliated bacteria are engaged in close interactions with amino acids of a neighboring PilE monomer in the pilus (Figure 49a). Because the N-terminus of minor pilins is very conserved with the N- terminus of the major pilin, we can imagine that similar interactions could be engaged between PilE and minor pilins. For example, L3 is in close contact with Y27 of the next monomer (Figure 49a). The L3H mutation could induce a stronger interaction because of aromatic stacking.

131

Figure 48: Characterization of the retractile properties of the α1N mutants. (A) Measurement of average bacterial speed on bacterial tracks captured for by live microscopy for 2minutes.

Mean values ± SEM are indicated for each strain. N=3 independent experiments. (B) α-value extracted from the MSD that was computed on all the tracks obtained for each strain. Interpretation of this value is indicated on the figure. (C) Transformation competence measured as the ratio between the number of bacteria in the inoculate and

the number of transformant bacteria recovered. Mean values ± SEM are indicated for each strain. N⩾3 independent experiments. (D) Piliation per bacterium as a function of the percentage of piliated bacteria measured by flow cytometry.

Dotted arrows indicate the effects of the pilT mutation on piliation. Mean values are indicated for each strain. N⩾3 independent experiments.

Evidence for interaction between the major pilin and the minor pilins (PilV and PilX) were obtained by cysteine cross-linking by Anne-Flore Imhaus during her PhD (unpublished data Figure 49b). These data indicated that L3 from PilE and Y27 from PilV and PilX are in close contact in the cell. This supports the possibility that the changes we observe in the proportion of piliated bacteria are due to a modified interaction between PilE and minor (pseudo)pilins. To test this hypothesis, we deleted the minor pilin pilV gene in the short pili mutants. We then measured their piliation by flow cytometry. Strikingly, deletion of pilV resulted in a 4-fold decrease in the proportion of piliated cells in PilEL3H (Figure 49c). Overall, pilV deletion brought all of the studied mutants back into the piliation distribution that we have established earlier in the article.

132 pilV deletion had no significant effect on the piliation per bacterium but strongly reduced the percentage of piliated bacteria. Altogether, this suggests that the change in the proportion of piliated bacteria is dependent on the presence of minor pilins and that this phenotype is probably unrelated to decrease in pilus length. These overlapping phenotypes are certainly due to the many interactions mediated by the N-terminal part of PilE.

Figure 49: Interactions between PilE and minor pilins are required to increase the proportion of piliated cells. (A) Interactions found in the core of the pilus structure between two pilE monomers are highlighted for the mutated amino acids we found to be affected by mutation. (B) Western blots showing disulfide crosslinking between PilE and minor pilins. Electrophoresis was run under non- reducing conditions. Top blot was incubated with the α-PilV antibody and the bottom one with an α-PilX antibody. Location of the heterodimer is indicated by the black arrow. Figure from Anne-Flore Imhaus (Imhaus, 2013) (C) Piliation per bacterium as a function of the percentage of piliated bacteria measured by flow cytometry. Dotted arrows indicate the effects of the pilV mutation on piliation. The curve is the reference piliation curve that derives from results presented in the article. Mean values are indicated for each strain. N⩾3 independent experiments

133 2.3 Exploring adhesion to human cells

In the course of our adhesion study, not only did we find mutants that had lost their ability to adhere, but we also found hyper-adhering mutants. These mutations were of interest as they specifically increased adhesion to human cells and not to mouse cells (Figure 50a). This increased adhesion was also dependent on the presence of PilV and PilC1 (Figure 50a), thereby suggesting that the phenotype we observed was not due to a novel kind of adhesion but to a reinforcement of the WT adhesion.

2.3.1 Deep mutational scanning shows a specific role of several tyrosine residues in adhesion We repeatedly observed that mutations introducing tyrosine residues in the PilE sequence had significant effects on adhesion. To validate these observations, we used our deep mutational scanning data to analyze if there was an effect of the mutation type we introduced in PilE. We could first confirm the validity of our analysis as we found that introducing stop mutations was as deleterious to piliation as it was to adhesion. In contrast, mutating residues to tyrosine significantly increased adhesion over piliation (Figure 50b). We initially focused on two

mutations located in the N-terminus of PilE and predicted to be buried in the fiber: PilED26Y and

PilES34Y. These mutants were less piliated than PilESB and were found to be non-aggregative (Figure 50c) but hyper-adhering at early time points (Figure 50d). This defect in aggregation resulted in the loss of their adhesion advantage over time (Figure 50d) and could be visualized by microscopy through their diffuse adherence pattern at 2 hours at the cell surface (Figure 50e,f). Because, early adhesion was increased in these mutants, we next examined the patterns of plasma membrane remodeling (cellular response) we could observe upon adhesion of individual bacteria. Indeed, while membrane deformation upon meningococcal adhesion has been described for the adhesion of microcolonies, it has only been recently described at the level of individual bacteria and can only be detected using TIRF (Total Internal Reflection Fluorescence) microscopy for the reference strain because of the very weak intensity of the response (Charles-Orszag et al., in review 2018). In contrast, we could observe a very intense response around individual mutant bacteria but not in the reference strain using conventional spinning disk confocal microscopy (Figure 50f).

134 Figure 50: Introducing tyrosine residues in PilE enhances early adhesion to human cells (A) Ratio of the piliation mutation scores and the adhesion mutation scores were depending on the amino acid they introduced in the PilE sequence. These values were then averaged and are presented on this graph. Significant deviation from zero indicates that introduction of a given amino acid affects one function more significantly than the other. Tyrosine mutation score is highlighted in white. (B) Quantification of bacterial adhesion at 30minutes to HUVEC and bEnd.3 cells. HUVEC: primary human umbilical vein endothelial cells. bEnd.3: immortalized cell line from brain endothelial mouse cells. Mean values ± SEM are indicated for each strain. N⩾3 independent experiments (C) Fluorescence microscopy illustrating the different adhesion patterns between the three strains after 30 minutes and 2hours of adhesion. PilESB eventually forms microcolonies while PilED26Y and PilES34Y have a diffuse adherence pattern.

(D) Quantification of bacterial adhesion to HUVEC over time. Mean values ± SEM are indicated for each strain. N⩾3 independent experiments. (E) Immunofluorescence of bacteria adhering to HUVEC after 30minutes or 2hours. Cells were infected with GFP- expressing bacteria at MOI 200. Nuclei were stained with Hoechst prior to infection. (F) Immunofluorescence of bacteria adhering to HUVEC after 2 hours of infection. Cells were infected with GFP- expressing bacteria at MOI 200. Bacterial pili were immunostained with the monoclonal antibody 20D9 and ezrin with a rabbit polyclonal antibody.

135 This confirms an increase in the affinity of TFP to the cell membrane. The specific role of the hydrophobic aromatic tyrosine in this adhesion prompted us to examine whether TFP could be binding to a lipid substrate.

2.3.2 Cholesterol-binding by TFP This led to the observation that PilE contained several putative cholesterol binding sites in its sequence, including a site at the tip of the pilus for which we have previously described a role in the initiation of adhesion. The specificity of these motifs named CRAC (Cholesterol Recognition Amino acid Consensus) or CARC (inverted CRAC) domains is that their central

invariable amino acid is a tyrosine (Figure 51a,c). In this context, PilED26Y is predicted to result

in a more favorable putative CRAC site and PilES34Y results in the formation of a new CARC site. This could explain the hyper-adhesion of these mutants. Reciprocally, at the exception of the

mutant PilEL52M, we observe decreased adhesion when mutating the key amino acids in the CARC binding site at the tip of the pilus (Figure 51b). These results are in partial support of our cholesterol-binding hypothesis. One of the problem for this model is that all these amino acids are buried inside the fiber. Yet, we can observe that these sites are located at the junction of the globular heads of four monomers. Because of the previously discussed elastic properties of TFP, these regions could become exposed upon stretching of the pilus fiber. This could provide a mechanism to explain how adhesion propagates along the fiber.

To assess whether cholesterol binding by TFP was a reasonable hypothesis, we depleted the cholesterol of the cellular membrane of HUVEC by using methyl-ß-cyclodextrin. We could observe a dose-dependent decrease in adhesion upon cholesterol depletion (Figure 51d). Upon repletion of membrane cholesterol using methyl-ß-cyclodextrin-cholesterol complexes, adhesion could be restored to basal levels or even increased by longer incubations.

We then undertook to analyze interaction between TFP and cholesterol. In order to do that, we co-incubated multilamellar vesicles (MLV) composed for 70% of POPC (a phosphatidylcholine lipid) and 30% of cholesterol with crude pili preparations. Upon imaging of the fluorescently labeled mixtures, we could observe some interactions between vesicles and TFP. These images show interactions along the length of the fiber and potentially at the tip of the pilus (Figure 52). Because quantification of the interactions between pili and lipids using microscopy was quite tedious, we used flow cytometry to conduct in-depth analysis of this interaction.

136

Figure 51: Identification of putative cholesterol binding sites in TFP. (A) Consensus sequences of cholesterol binding sites is presented at the top. The two cholesterol binding sites Identified in the α1 of PilE are highlighted below on the PilE sequence. Some of the mutations we characterized in these binding sites are indicated by arrows. Color indicates the observed phenotype (green for maintained or increased adhesion, red for deficient adhesion).

(B) Quantification of adhesion at 30 minutes of mutants from panel A. Mean values ± SEM are indicated for each strain. N⩾3 independent experiments (C) Structure of the pilus highlighting the putative cholesterol binding sites (CARC in blue, CRAC in orange). Top inset shows the location of these sites at the junction of 4 pilins. Bottom inset shows the continuity of these cholesterol binding sites in the helical pilus. (D) Quantification of bacterial adhesion at 30minutes in the presence of methyl-ß-cyclodextrin (MßCD). Prior to adhesion, cells were incubated for 2 hours at the indicated concentration of MßCD. For repletion experiments, incubation was followed by an incubation in the presence of 4mM MßCD-cholesterol for 30minutes or 1hour. Mean values ± SEM are indicated. N⩾3 independent experiments.

137

Figure 52: TFP can interact with liposomes. Immunofluorescence of MLV incubated with and wihtout TFP. MLV were stained with the FM4-64 lipid dye, pili were stained with the F10-I nanobody and FITC dextran was added to the medium to distinguish lipid vesicles. Medium was solidified by addition of low- melt agar before imaging. Scale bar: 1µm.

With this technique, we could detect binding of TFP to MLV. Binding was observed both to MLV composed of pure POPC and to MLV composed of POPC:cholesterol (7:3) (Figure 53a). This lipid-binding is dose-dependent and is enhanced in the presence of cholesterol (Figure 53c).

We then utilized pili from the hyper-adhering mutant strain PilES34Y and the hypo-adhering

PilEY50C and analyzed their lipid binding properties. We could observe a significant increase in lipid binding by the PilES34Y mutant pili. We could also detect a small decrease in lipid binding in the hypo-adhering mutant. This supports the hypothesis that adding tyrosine residues could enhance lipid binding, in part through cholesterol binding. In order to determine the role of cholesterol more specifically, we tried to prevent unspecific lipid binding by TFP. In order to do that, we introduced negatively charged phospholipids (POPG, a phosphoglycerol lipid) in the composition of the MLV we used. Addition of these negatively charged lipids completely prevented TFP binding to POPC:POPG (7:3) MLV. Residual TFP binding could be observed to POPC:POPG:cholesterol vesicles but these events were too rare to yield interpretable results (Figure 53b). Appropriate experimental conditions still need to be found to evaluate the influence of cholesterol on TFP-binding. A final line of evidence supporting a role for cholesterol in TFP-dependent adhesion lies in the vesicle deformation we observed by flow cytometry. Indeed, vesicle shape was dramatically affected by the addition of TFP in a cholesterol-dependent manner and a dose-dependent manner (Figure 53d). Again, these alterations were enhanced for the PilES34Y mutant pili. These observations are also consistent with the increased cell membrane deformation induced by bacterial adhesion we previously described (Figure 50f). This also suggests that efficient adhesion to lipid vesicles is only observed in the presence of cholesterol and enhanced by the presence of an extra tyrosine residue. Yet, because the amino acids we identified to be able to mediate an increase in adhesion are buried inside the fiber, we also examined another hypothesis that could explain this phenotype.

138

Figure 53: Hyper-adhering TFP have a higher affinity for cholesterol. (A-B) Flow cytometry scatter plot (side scatter intensity as a function of pili fluorescence intensity) of MLV alone or co-incubated with TFP and stained with 20D9. MLV composition is indicated and represented on the left. (C) Quantification of the fluorescence increase detected by flow cytometry as a function of the concentration of pili added to POPC or POPC: cholesterol MLV. Experiments were run independently with pili of the PilESB, PilES34Y or PilEY50C strains. Mean values are indicated. N⩾3 independent experiments (D) Quantification of the shaped change detected by flow cytometry (using the forward scatter parameter) as a function of the concentration of pili added to POPC or POPC: cholesterol MLV. Experiments were run independently with pili of the PilESB, PilES34Y or PilEY50C strains. Mean values are indicated. N⩾3 independent experiments.

139 2.3.3 Meningococcal TFP are electrically conductive

Aromatic amino acids and tyrosine in particular have been shown to be involved in a specific function of TFP in G. sulfurreducens: electrical conductivity (see Section 2.2.7 of the introduction). Such a function has never been investigated for the TFP of N. meningitidis.

Figure 54: Conductive pili as adhesion facilitators. (A) Structure of the TFP from G. sulfurreducens (ARC-1 model) and N. meningitidis (PDB:5KUA). Spacefilling representations of the aromatic residues F1, Y24 and Y27 are shown in yellow. S34 is shown in blue in the rightmost panel. (B) Model explaining the role conductive pili could have in overcoming negative repulsive interactions due to negative surface charges. Figure adapted from the Malvankar lab (Yale university)

(C-D) Experiments performed by the Malvankar lab. Mean values ± SD are indicated for each experiment. N⩾3 independent experiments. (C) Conductivity measurements made on TFP from N. gonorrhoeae and G. sulfurreducens. In the aromatic mutant, all 5 of the aromatic amino acids are substituted by alanine residues. (D)

Left panel: conductivity measurement of TFP from PilESB (WT) and PilES34Y. Right panel: Adhesion measurement of

TFP from PilESB (WT) and PilES34Y adhering to a mica substrate by Atomic Force Microscopy.

140 Indeed, while the aromatic amino acids (F1, Y24, Y27) are conserved between the TFP of N. meningitidis and those of Geobacter sulfurreducens, they do not show the same organization in the fiber and electrons would have to be transmitted over considerable distances in order for the meningococcal pilus to be conductive (Figure 54a). Inserting, tyrosine residues inside the fiber (at position 26 or 34) could therefore promote electron transport in the pilus. Because both the bacterial surface and the human cell surface are negatively charged, bacteria need a way to overcome this repulsive interaction in order to attach to the cellular surface. TFP are known to be responsible for this initial interaction in N. meningitidis. If the pili were conductive, this would result in electron transfer between the less negatively charged bacterial surface and the human cell and provide a way to generate an attractive interaction (Figure 54b).

Our collaborators from the Malvankar lab undertook to measure the conductivity of TFP from N. meningitidis and N. gonorrhoeae and were able to show that these pili are indeed conductive. While the conductivity is 10 times weaker than that of G. sulfurreducens, it is more than 100 times superior to that of a non-conductive mutant with no aromatic residues (Figure 54c). This shows that meningococcal pili are able to conduct electrons.

Furthermore, we could observe that the TFP of the PilES34Y mutant were three times more conductive than those of PilESB. This increase in conductivity also induced a 3-fold increase in adhesion to a negatively charged mica surface as measured by AFM (Figure 54d). This increase in adhesion is similar to the one we observed for bacterial adhesion (Figure 51b). These preliminary results provide substantial support for a role of additional tyrosine residues as electron transfer facilitators and potentially support the importance of the conductivity of TFP in the adhesion of bacterial pathogens.

141

142

Discussion143 In this section, we put in perspective the results presented earlier. We will not extensively discuss the results from the article and will instead present more speculative perspectives. We first discuss the adhesion results presented in the additional results section and propose a comprehensive model for this process. We then discuss regulation of piliation at the population level, indicate the actors we have identified that could take part in this regulation and propose experiments to get a better understanding of this phenomenon. Finally, we recapitulate our results for all the meningococcal TFP-dependent functions and extrapolate on what this tells us about the functions of TFP in particular and TFF in general.

1 Adhering under flow, learning from other bacteria

Deep mutational scanning provided several unexpected results concerning adhesion. Overall, we could identify mutations of amino acids that are not exposed at the surface of the pilus but that promote adhesion and we could only identify mutations that inhibit adhesion at the tip of the pilus. While a clear view of the adhesion process is still missing, we have gathered sufficient preliminary evidence on this process to build a 4-step adhesion model (Figure 55).

Because both the bacterial surface and the cell surface are negatively charged, interaction between these two cells is not favored. In N. meningitidis, TFP are known to be the initial mediator of this interaction and we could observe this electrostatic repulsion at the TFP level in our MLV-TFP binding experiments. Addition of negatively charged lipids significantly prohibited interaction between TFP and the MLVs (Figure 53b).

It would therefore be advantageous for N. meningitidis to have evolved a way to overcome this electrostatic repulsion. The observation that meningococcal TFP are electrically conductive could explain how bacteria overcome this repulsion (Figure 55 step1).

By introducing aromatic residues (tyrosine) at non-exposed sites in the fiber, we could observe an increase in cell adhesion. This increased adhesion to cells was accompanied by an increased electrical conductivity of the mutated TFP and an increased adhesion force to the negatively- charged abiotic mica surface. We could also detect a local depolarization of the human cells in the seconds preceding bacterial adhesion using fluorescent reporters of membrane potential (data not shown) These data support the view that pilus conductivity could facilitate initial strong yet non-specific interactions through electron transfer. Using AFM, the detachment forces between TFP and a mica abiotic surface were estimated to be around 2nN. Such forces could be sufficient to explain retraction-resistant anchoring at the cell surface. Indeed, the retraction force of a single pilus is estimated to be in the 100pN range294.

Yet, because of the human specificity of TFP-dependent adhesion, we do not believe that this electrostatic favored adhesion is sufficient to explain the whole adhesion process. We expect TFP to engage interaction with a cellular receptor following the overcoming of the electrostatic barrier (Figure 55 step2). We have shown here that this high-affinity adhesion happens at the tip of TFP. This is also consistent with the orientation of TFP that would be generated by the

144 initial electrostatic interaction we just mentioned. The nature of the human receptor is still under debate. In order to clarify the nature of the interactions happening at the tip of TFP, we should examine the effect of the mutations we generated at the tip of the pilus on the affinity to the receptors described in the literature (CD147 notably600).

Upon tip-anchoring of the TFP, two types of forces are expected to act on adhesion. An environmental force : shear-stress is exerted on bacteria in blood vessels or upon cell washing in vitro and, in all conditions: TFP retract upon adhesion705. Shear stress is exerted indirectly on TFP while retraction forces are directly applied on TFP (under the hypothesis that more than one pilus is attached to the surface). Because TFP have been suggested to deform upon adhesion and the structural basis behind this observation has recently been elucidated 307,486,639, we would predict that the forces exerted on TFP during adhesion will result in their deformation (Figure 55 step 3). This deformation probably participates in dissipating most of these forces exerted on the bacteria as this has been previously suggested for other pilus types. Furthermore, this deformability was recently suggested to be important for adhesion and pathogenicity for the type I pilus of E. coli123. The importance of TFP deformation for adhesion should be investigated. Because the portion of PilE behind this flexibility is thought to be in the α-helix, it would be interesting to evaluate the consequences of the mutations we generated in the α1N on TFP flexibility and to try and correlate them to adhesion forces under controlled flow. These observations would further expand the importance of the N-terminus of PilE and justify the extreme selection pressure exerted on this portion of the protein. In addition to the mechanical resistance it provides, we expect this flexibility to induce conformational changes that reveal new binding sites along the length of the pilus. Indeed, the site we observed to be responsible for tip-dependent adhesion has been shown to be exposed upon pilus deformation486. Because we and others have observed that TFP can also adhere to the cell membrane over extended lengths, we propose that this deformation could initiate the propagation of adhesion along the pilus to the same cellular receptor or a secondary receptor present in the cell membrane (Figure 55 step4). This strategy would also present the advantage for the bacterium to hide the majority of its antigenic adhesion sites from the immune system and allow their conservation without compromising immune escape and survival of the bacterium. In order to demonstrate the validity of such an adhesion model, we could use an experimental setup quite similar to the one we used in the article. Namely, we could monitor adhesion using live microscopy by first labelling tip-anchored TFP without using formaldehyde fixation and then applying an increasing flow to monitor if adhesion progresses along the pilus length upon shear increase.

The importance of the retraction force on adhesion is also supported by the observation that a pilT mutant does not adhere more efficiently than the WT at early time points (data not shown). This is quite surprising given the fact that in a pilT mutant, the proportion of bacteria is 5 times higher than in the WT and that these bacteria are thought to be more piliated. Keeping up with this reasoning, this would mean that the pili of pilT mutants are 5 times less efficient in adhesion

145 than the WT ones, which is equivalent to the adhesion efficiency of a pilC1 mutant. The importance of pilus retraction has been clearly documented in P. aeruginosa706. As explained in the next section, we believe that in P. aeruginosa all bacteria in the population display similar piliation status. If the same proportion of bacteria is piliated in the WT and the pilT mutant in P. aeruginosa, this could explain why differences in adhesion are more easily detected in P. aeruginosa than N. meningitidis. Again, this reveals the need to understand the mechanisms behind piliation regulation in N. meningitidis in order to understand how TFP mediate their diverse functions.

Figure 55: Adhesion model for N. meningitidis To adhere to human cells, bacteria first have to overcome electrostatic repulsion using long-distance interaction with their electrically conductive TFP (1). This positions the pilus ideally to initiate adhesion via a binding site located at the tip of TFP (2). Upon adhesion, TFP are retracted and the bacteria has to face blood flow in the vessels. Adhesion is maintained by partial dissipation of these forces through pilus elongation (3). This elongation results in the exposure of binding sites along the pilus and could participate in bacterial anchoring and the typical cellular response observed upon bacterial adhesion. Upper panel makes use of illustrations from the Servier medical art library.

146 2 Regulation of meningococcal piliation, a matter of bistability?

In this study, we have provided evidence that only a minor proportion of the bacteria is piliated in the meningococcal population. To our knowledge, this phenomenon has never been documented in the literature for pathogenic Neisseria. While these observations might only be relevant under laboratory conditions, we do not believe that they are due to technical limitations. Indeed, we never observed more than 20% of piliated bacteria in our reference background but could detect as much as 90% of piliated bacteria in a retraction-deficient mutant (pilT). Furthermore, the proportion of piliated cells we observe is in good agreement with the proportion of bacteria that have been estimated to participate in the formation of aggregates in liquid culture in previous studies323,418 and to what can be observed microscopically in immunofluorescence studies of meningococcal TFP (Figure 58).

Additionally, such phenotypic differences in genetically identical bacteria have previously been described for other bacterial appendages. In Salmonella enterica707 and B. subtilis708 , flagella expression is only seen in a minority of cells (~20%). Similarly, in S. pneumoniae, type I pili are only expressed in ~30% of the cells709,710.

This ability for a genetically identical population to display two different phenotypes is named bistability. This phenotypic heterogeneity is due to bimodal gene expression and is under transcriptional control. Epigenetic mechanisms behind this regulation can range from simple feedback loops to DNA methylation711. In the examples we mentioned previously, feedback loops have been shown to be responsible for bistability in appendage assembly. In the case of the type I pilus a single positive feedback loop is responsible for the observed phenotype712. Briefly, the transcriptional activator RrlA positively regulates its own expression and the expression of the pilus operon. This single positive feedback loop is sufficient to explain bistability in the population. Yet, quite interestingly, RrgA, the major component of the pilus interacts with RrlA and this interaction results in a negative regulation of pilus expression. This negative regulation loop is thought to enhance the robustness of this bistability.

In the case of N. meningitidis, we have identified piliation bistability and propose that the major pilin PilE is part of this transcriptional regulation loop. Two lines of evidence in our work are in favor of this conclusion. First, single point mutations in the α1N of PilE are sufficient to dramatically increase or decrease the proportion of piliated cells. The mutations that we found to increase the proportion of piliated cells introduce polar amino acids that are more prone to interact with other proteins (Leucine to Histidine, Isoleucine to Asparagine). Reciprocally, the mutation decreasing the proportion of piliated cells involved the loss of a polar amino acid (Serine to Alanine). This would support the view that interaction of PilE with an unknown factor would bias the system equilibrium towards a piliated state. Second, we observed a gradual increase in the proportion of piliated cells when piliation per cell is increased. This could mean that an increase in [PilE] not only leads to an increase in the piliation per bacterium but also in the proportion of piliated

147 cells. Just like the first observation, this would place PilE as a positive regulator of piliation. This hypothesis could be tested by measuring the evolution of the percentage of piliated cells in bacterial population expressing an inducible version of PilE. Other lines of evidence from the literature are in favor of a regulatory role of PilE in piliation. In P. aeruginosa and M. xanthus, primary sequence of the major pilin was found to be involved in the regulation of its own transcription689,713,714. In P. aeruginosa, this regulation was recently shown to be mediated by intramembrane interactions between the α1N of PilE and the sensor kinase of a two-component system (PilS). This feedback loop is negative: high levels of PilE in the inner membrane induce a repression in the transcription of PilE. Yet, the situation is expected to be quite different in N. meningitidis. There is no PilS-homolog in the meningococcal genome and the transcription of PilE is thought to be under the control of the σ70 factor (housekeeping sigma-factor) unlike in P. aeruginosa where it is under the control of a σ54 factor (nitrogen-limiting factor)715,716. These observations along with our inability to detect differences in the PilE protein levels in our mutants suggest that the levels of PilE are not tightly regulated. Because PilE appears to play a regulatory role, it would suggest on the contrary that PilE may play its role through the regulation of the expression of other proteins of the piliation machinery.

Two questions arise from these observations: By what means does PilE regulate piliation? And, what are the components of the piliation machinery that are regulated by this mechanism?

Because of its membrane location, PilE needs to interact with an intermediate in order to influence transcription. A reasonable hypothesis would be that PilE levels are detected by a sensor through protein-protein interactions. A two-component signaling system could be involved in such a system. There are only four two-component signaling systems in Neisseria meningitidis717,718 and preliminary data obtained with a colleague from the lab indicate that the deletion of one these two-component signaling systems results in a considerable increase in the percentage of piliated cells without perturbing the cellular levels of PilE. The percentage of piliated bacteria in mutants of each of the two-component signaling systems should be examined in order to determine if they play a role in this regulation. Again, this supports the view that regulation of the proportion of piliated bacteria is not controlled through regulation of pilE transcription.

What component could then be under this PilE-dependent regulation? The easiest way to answer this question would be to use transcriptomics to determine the differences in gene expression between the reference strain, a pilEL3H mutant and a pilE deletion mutant. Yet, we already dispose of some information concerning the elements that could play a role in this regulation. We observed that the percentage of piliated cells sharply increases in a pilT mutant. This effect is independent of the one observed to be mediated by PilE as the double mutant pilEL3H pilT displayed an additive increase in piliation (Figure 48d). To confirm this effect, it would be interesting to investigate the proportion of piliated cells expressed by a strain with

148 inducible expression of pilT. Furthermore, we also observed (data not shown) that the percentage of piliated cells could be artificially increased upon induction of the assembly ATPase PilF. Taken together, these elements suggest that transcriptional regulation could act on the ratio between machineries in an assembly mode (in the presence of PilF) and in a retraction mode (in the presence of PilT). We have also shown that PilV (and presumably other minor pilins) is required to observe the influence of PilE on the percentage of piliated cells (Figure 49c). This could indicate two things: either PilV is required for PilE to mediate its transcriptional effects or pilV is one of the targets of the PilE-dependent transcriptional regulation. The first hypothesis is reasonable given the putative interaction we described earlier between PilE and PilV. To test the second hypothesis, we could simply compare the levels of the PilV protein in the reference strain, a pilEL3H mutant and a pilE deletion mutant. Looking at the western blot on Figure 49b and other replicates, we can observe that PilV levels seem to be increased in the PilEL3C mutant and we would therefore favor the hypothesis that PilV levels are indirectly regulated by PilE. This would indicate a new role for minor pilins that will need to be investigated in N. meningitidis. Because of the dispersion of the genes of the piliation machinery throughout the meningococcal genome, it is difficult to make other predictions on the nature of the other proteins that could be under the transcriptional control of PilE. Data on the transcriptional regulation of proteins of the piliation machinery is still lacking in meningococcus and could facilitate the understanding of the mechanisms behind piliation bistability. We summarize our current understanding of this system and draft a preliminary model of the actors involved in this bistability in Figure 56.

The ability for Neisseria meningitidis and other prokaryotes to maintain two subpopulations of piliated and non-piliated cells at all time is an example of bet-hedging. While TFP appear to be advantageous for cell motility and adhesion, the absence of pili is likely to favor bacterial dispersion, cell invasion and immune escape in the human host. Indeed, N. meningitidis has evolved several other mechanisms to generate TFP diversity in its population, in particular through antigenic variation of pilE, post-translational modifications of PilE and phase variation of pilC. Loss of piliation was recently shown to provide a selective advantage over time in cultures of N. gonorrhoeae719.Under the conditions used in their study, the authors could only attribute part of the absence of piliation to antigenic variation. The observation that a significant proportion of non-piliated cells had intact pilE sequences supports the existence of an additional mechanism to generate diversity in the piliation phenotype. We propose that this mechanism is bistability. This could mean that bistability in the expression of TFP is a shared feature in pathogenic Neisseria and potentially other pathogenic bacteria.

149

Figure 56: Speculative model for piliation bistability in N. meningitidis This model proposes that a complex between the major pilin PilE and minor pilin PilV could act as a regulating signal for gene transcription. This signal could be sensed in the inner membrane by the sensors of a two-component signaling system (TCSS) and lead to simple positive feedback loop of PilV and potentially other regulatory loop for other proteins of the piliation machinery that would promote pilus extension. The bottom graph represents how retraction or extension are affected in the different background that are listed on the x-axis. The balance between these two components can explain the nature of the changes we observed in the percentage of piliated cells in the population. IM: inner membrane, OM: outer membrane, P: Periplasm, C: Cytoplasm

3 Conservation among TFP-bearing prokaryotes

While our study has shed some light on the mechanisms behind the functions of meningococcal TFP, it is still unclear if these mechanisms can be extrapolated to other prokaryotes with TFP. By examining pilin sequence conservation and the available structures of TFP, we try to infer how broadly applicable these findings might be for the different functions we have described.

150

Figure 57: Multiple sequence alignment of type IV pilins The alignment of the first 53 amino acids of major and minor subunits from a variety of TFF is shown. The category to which they belong is indicated on the right and the protein name is indicated on the left. Conserved amino acids are highlighted for different functions. In blue: aromatic amino acids contributing to electron transfer, in purple: adhesion site identified at the tip of the pilus, in green: helix breaking residues and in orange: amino acids important for pilus length regulation and/or bistability. The arrow points the non-conserved V47. Adapted from Wang et al. 2017

3.1 Piliation: homologous structures with different properties 3.1.1 Folding PilE

During our study, we identified three valine residues that form a hydrophobic core required for proper folding of the major pilin and pilus assembly. The alignment in Figure 59 shows that these amino acids are not conserved in most TFP. This indicates that this finding is not a general feature required for PilE folding or pilus assembly and, there is certainly an important variation in the combination of amino acids that maintain the pilin fold between different species.

3.1.2 Bistability and pilus length

It seems likely that pilus bistability is not present in all prokaryotes. Indeed, by simply observing TFP expression by immunostaining, one can easily see that the majority of N. gonorrhoeae cells (and similarly for N. meningitidis) are not piliated, while the majority of P. aeruginosa cells are piliated (Figure 58). Even more interestingly, by replacing the native major pilin of N. gonorrhoeae by that of P. aeruginosa, we observe a similar pattern to that seen in populations of P. aeruginosa, with most cells presenting pili. This emphasizes the role of the major pilin in this bistability process and the species-dependence of this phenomenon. Yet the residues we identified to be important for bistability and pilus length are very conserved (Figure 57). This

151 indicates that there are probably other residues that remain to be identified that could be involved in this regulation.

Furthermore, this high conservation (even observed for the T2SS) could also indicate that these amino acids play an important role in another aspect of pilus assembly. Such a role could be mediated through the interactions we discussed previously between the major pilin and minor pilins (Figure 49). A role of the minor pilins in regulation of pilus length has been demonstrated in other systems such as the Pap pilus720 and the type V pilus178,179 and could also explain why the mutants we obtained display shorter pili.

Overall, these results suggest that pilins have evolved a great diversity of strategies to conserve similar folds and assemble functional pili. This is also important in a therapeutic perspective for the development of broadly applicable strategies to preclude TFP assembly. Such strategies appear unlikely to be successful if focused on the major pilin. The development of strategies to inhibit other broadly distributed members of the piliation machinery with better conserved properties such as the elongation ATPase PilF might be more relevant. Yet, the intracellular location of such targets represents a serious challenge to overcome.

TFP !-PilE !-PilA (PilE) (PilE) (PilE) (PilA)

N. gonorrhoeae P. aeruginosa

Figure 58: Different proportions of cell display TFP depending on the expressed major pilin. Immunofluorescence staining of pili from N. gonorrhoeae and P. aeruginosa. The upper line indicates the antibody used to stain pili (in red). From left to right: WT N. gonorrhoeae; N. gonorrhoeae co-expressing PilA from P. aeruginosa and PilE; same strain but stained with an anti-PilA antibody; WT P. aeruginosa. Adapted from Winther- Larsen et al. 2007 and Heiniger et al. 2010.

3.2 Competence: electropositive grooves to bind DNA?

For competence for transformation, we have only provided preliminary evidence that DNA binding could be occurring in one of the furrows found at the surface of the pilus. No obvious charged patch could be identified to account for high-affinity binding of DNA. This is in agreement with the relatively low-affinity DNA binding that was identified for meningococcal TFP 636. We examined how such findings could be extrapolated to the two other bacterial species for which TFP-dependent DNA binding has been described and for which high resolution structures are available: N. gonorrhoeae643 and P. aeruginosa721.

152 The surface of the TFP of P. aeruginosa presents a continuous electropositive stripe wrapping around the helical filament. This stripe makes an appealing target for DNA binding but no mutagenesis studies supporting this prediction can be found in the literature. In contrast, the surface of the gonococcal pilus is mostly electronegative, which is predicted to be unfavorable for DNA-dependent interactions. Yet, a right-handed helical stripe of electropositive amino acids could support DNA binding function (Figure 59). Again, there is no evidence in the literature to support such a mechanism of DNA binding. Finally, in support with the results we presented earlier for the competence of N. meningitidis (Results section 2.1), the groove we identified (blue path in Figure 47d) follows a helical stripe of electropositive patches resembling the stripe identified in P. aeruginosa ((Figure 59).

Figure 59: Putative DNA binding sites can be identified at the surface of type IVa pili. Structure of the pili from P. aeruginosa (PAK), N. gonorrhoeae (Ng) and N. meningitidis (Nm) are color- coded based on their surface electrostatic potential. Putative DNA binding sites are indicated by green arrows. For Nm, we used the results from deep mutational scanning to represent the DNA binding site.

Adapted from Wang et al. 2017

These observations support the view that the major pilin could be directly responsible for this DNA binding activity through the formation of positively charged quaternary surfaces. So far, we were only able to observe such surface arrangements for type IVa pili with relatively low DNA-binding properties, but these could probably be extended to other types of TFP. High affinity DNA binding by a TFP-like competence pilus has been reported in S. pneumoniae461 and it would be of great interest to obtain its structure in order to see whether similar patterns can be found at the pilus surface. The higher affinity of this pilus for DNA could be a valuable tool to decipher the molecular interactions that are at play between the amino acids of the major pilin in the pilus and the nucleotides of the DNA double helix. 3.3 Aggregation through electrostatic complementarity

For aggregation, we could also confirm the importance of electrostatic interactions between long TFP. A peculiarity of the TFP we studied here, is that they display a hook-like structure at

153 their surface. There is considerable variation in this structure between different bacterial species (Figure 59) and even among different strains of N. meningitidis. These natural variations can even induce loss of the aggregative properties of N. meningitidis595. In the literature, TFP-dependent auto-aggregation has been well-described for relatively few bacterial species; essentially the pathogenic Neisseria, V. cholerae and E. coli. A clear mechanism is still lacking to explain aggregate formation, partly due to the lack of high- resolution structures for these pili. The aggregation of V. cholerae is the one that was most extensively studied until now. Two studies have examined the influence of mutating amino acids of the major pilin of the Tcp pilus443,722. Very similarly to what we have found, authors identified a central role of charged residues. Point mutations in charged amino acids were sufficient to completely inhibit aggregation or to promote it. By combining a mutational and a structural approach, the authors identified a positively-charged protruding region in the D- region and a negatively charged depression at the surface of the pilus. Depending on the biotype examined, the opposite pattern (a positively charged depression and a negatively charged protuberance) could also be observed (Figure 60). From these observations and co- aggregation experiments, authors have speculated that the protruding region of the pilus from one bacterium could intercalate in the depression of the pilus from another bacterium and lead to aggregate formation. While most point mutations engineered by the authors do not appear to directly support this hypothesis, by combining multiple substitutions the authors provide further support for their model. This also shows the predictive limitations of single point mutations. In our case, this would suggest that the electropositive hook could interact with a negative surface on the pilus. We did not identify such a surface in our work and inspection of the structure, only shows a positively charged depression at the surface of the pilus (Figure 60). Yet, the upper face of the hook-structure appears positively charged while the lower face is negatively charged. This arrangement could imply that the hook is “bipolar” and acts like a magnet (Figure 60). This would mean that the only region required for these interactions is the hook and would also explain why we did not identify another region involved in aggregation. Because these two sites are in close vicinity, it is complicated to demonstrate such a mechanism. We could try to swap charges (K140D D142K) at these two locations and observe if aggregation can be restored. The only obvious limitation of this mechanism is that it would imply that TFP of different bacteria can only interact in a parallel fashion and not in an anti-parallel fashion (Figure 60).

While much of the aggregation mechanism remains to be clearly established, both examples of aggregation from a type IVa (TFP of N. meningitidis) and a type IVb pilus (Tcp of V. cholerae) seem to rely on surface electrostatic interactions involving the highly variable D-region established between the pili of two or more bacteria and indicate an overall conservation of this function in TFP.

154

Figure 60: Widespread role of electrostatic interactions in aggregation (A) Structure of the V. cholerae TcpA major pilin. The two variants presented (from the classical biotype on the left and the El Tor biotype on the right) have relatively well-conserved structure at the exception of two areas. The depression highlighted by an arrow and the protruding region circled by an ellipse. Note the inversion in electrostatic potential. (B) Pilus structure color-coded by electrostatic potential. No obvious electronegative groove can be found to explain a direct electrostatic-dependent interaction at the exception of the lower face of the hook (Left). A model for bundling and aggregation based on the opposed charges found at the hook surface is proposed. The pilus is represented in gray and the blue and red rectangles respectively account for the positive and negative hook surfaces (Right). Adapted from Lim et al. 2010 and Wang et al. 2016 3.4 Adhesion: a conserved mechanism for type IVa pili?

In the case of adhesion, there is reasonable evidence that TFP from different bacterial species bind different receptors as substantiated by the differences in cell tropism mediated by TFP. It is therefore not possible to envisage a common adhesion site that would be conserved among TFP. Indeed, we can see that the adhesion site we propose for N. meningitidis is not significantly conserved between species (Figure 57 in purple). Yet, it does appear to be conserved between

155 the two pathogenic Neisseria. This is reasonable as both bacterial species are human obligate pathogens. Because the TFP receptor for N. gonorrhoeae is still unknown, it would be interesting to investigate if the recently described CD147 receptor for N. meningitidis can also serve as a receptor for N. gonorrhoeae. We should also examine the consequences of the mutations we identified in N. meningitidis on gonococcal adhesion. While various bacteria interact with different cell types and even cell species, there are a few common points between all of these TFP-bearing prokaryotes. The first one is that they have to overcome the electrostatic repulsion between their negatively- charged surface and that of the surfaces they bind to. We can see on the previous alignment that there is a good conservation of the aromatic residues that could be involved in electrical conductivity for type IVa pilins (Figure 57 in blue). These residues are not conserved for other TFF. This provides further support for a specific function of these aromatic residues and indicates that it would have only emerged in a fraction of TFF-bearing prokaryotes. It also raises the question of how this electrostatic repulsion can be overcome by type IVb pili-expressing bacteria. Because most bacteria also express several other virulence factors that are required for cell adhesion (toxins, pili, flagella), these additional virulence factors could be the ones supporting this role.

Another common point between TFP-bearing prokaryotes is the conditions they face upon attachment. Indeed, it is common for them to be subject to high shear stress upon attachment. As mentioned earlier, a common mechanism used by bacteria to maintain attachment is to disperse some of the energy resulting from the shear through pilus deformation and such deformations have been observed in type IV pili. We have also suggested that this mechanism might be at play in the adhesion of N. meningitidis and could even be responsible for adhesion amplification via a zip-in mechanism (Figure 55). Recent structures of Type IVa pili307,308 and T2SS159 but not the archaellum246 have revealed that part of the α-helix is melted in the assembled pilus and could account for this flexibility. Where observed, this melted helix is always flanked by helix-breaking glycine or proline residues (Figure 57 in green). These residues are very well-conserved in type IVa pili and the T2SS. While the potential reasons behind the conservation of these attributes in T2SS will require further investigation, their conservation in type IVa pili further support a functional role of this flexibility. This role is also in good agreement with their role in adhesion.

Interestingly, the residues involved in the adhesion of another type IVa pilus (the PAK pilus) to its N-glycan receptors723,724 on cells have been identified 725–727 and are found in the C-terminal D-region (Figure 61). This adhesion was shown to happen at the tip of purified PAK pili429. Indeed, the binding region is buried along the pilus as it is masked by the neighboring pilin subunit and is only available at the tip of the pilus (Figure 61). Again, the position of this binding site strongly suggests that a rearrangement of the fiber upon extension would unmask these binding sites along the fiber and could promote a stronger adhesion. This reveals that the

156 model of adhesion we proposed based on our findings (Figure 55) may be relevant for type IVa pili in general and not simply for the pili of pathogenic Neisseria.

These observations also reveal that a single model is unlikely to explain the adhesion of all TFF and that a more advanced phylogenetic analysis could greatly participate in clarifying this point.

Figure 61: Adhesion site of the PAK pilus from P. aeruginosa The D-region binding site of the pilus is highlighted in red and magenta. Magenta indicates the cluster of amino acids available at the tip of the pilus as shown on the top left inset. The bottom left inset shows the neighboring pilin monomer partially blocking the binding site. Structure from PDB :5vxy

3.5 Using TFF to understand how TFP mediate their functions

As exemplified by our study, TFP seem to have evolved several independent functions through distinct specialized regions of the major pilin PilE. Because the vast TFF family shares common evolutionary origins as well as common functions, a global evolutionary approach could be useful in understanding how the major pilin has evolved its different functions and what are the domains required for these different functions and for piliation.

This could theoretically be achieved using the Statistical Coupling Analysis (SCA) method728. This approach allows to detect and differentiate functional protein sectors. These so-called sectors are non-contiguous networks of amino acids in the protein that have coevolved (to mediate a common function). In brief, this technique relies on the alignment of a large repertoire of evolutionarily-related sequences. Based on this multiple sequence alignment, the co-evolution of all couples of amino acids is evaluated. Groups of amino acids that consistently evolved together form sectors.

Because TFF form a diverse repertoire with many known major pilin sequences, we could use SCA to identify sectors in the major pilin. Coupling this analysis with the functional mapping we have established in our study for PilE should help expand and generalize our findings to TFP and TFF. What it potentially means, is that analyzing data that are already available could suffice

157 to obtain a global view of the mechanisms underlying TFF functions. While this remains an ambitious goal, such an analysis should significantly improve our current understanding of the structure/function relationships in TFF and highlights the crucial importance of studying the diversity of the too often overlooked non-pathogenic prokaryotes.

158

In this doctoral work, using deep mutational scanning, we have provided evidence that the major pilin PilE has evolved several specialized domains to mediate multiple functions. These functions can be dissociated and loss or gain of functions can be observed by mutating single amino acids of PilE. We identified four distinct functional regions at the surface of the pilus: v The distal tip (in magenta) which is involved in initial adhesion to human cells, v The electropositive hook centered around lysine 140 (in green) which is involved in auto- aggregation, v Two neighboring regions in the groove of the quaternary structure (in blue) which may be involved in DNA binding, v The α1N region (in light gray) which is involved in piliation regulation but not in pilus assembly per se. These findings strongly support the view that PilE is the main mediator of TFP functions. Yet, they open many new questions concerning the actual role of minor pilins and PilC1. They also open new research avenues in the field, including: deciphering the succession of events allowing TFP-dependent adhesion but also characterizing regulation of piliation at the population level and at the individual level. Combined with our results, these perspectives will undoubtedly benefit the understanding of meningococcal pathogenesis and could help develop new treatments for the too often fatal meningococcal disease.

Conclusion159

160

PilE

Materials &

Methods 161 In this section, we only present the methods used to obtain the additional results presented in the Additional results section. Methods used for the submitted article are described in the main body of the article.

¬ Cholesterol depletion and repletion: Stock solutions of depletion medium (Endo-SFM+10%FBS+4 or 2 or 1mM Methyl-beta- cyclodextrin (MβCD purchased from Sigma)) were made fresh on the day of infection. Cholesterol-depletion was performed 2 hours prior to infection by replacing the cell medium, rinsing with PBS 3 times and replacing it with the depletion medium. After 2 hours of depletion, infection was carried in the depletion medium.

For repletion experiments, after 2 hours of depletion, cells were rinsed 3 times with PBS and incubated for 30minutes or 1hour in repletion medium (Endo-SFM+10%FBS+1mM MβCD- cholesterol (purchased from Sigma)). Cells were then rinsed 3 times and infection was done Endo-SFM+10%FBS.

¬ Preparation of Multilamellar vesicles: The following reagents were purchased from Avanti Polar Lipids: 1-palmitoyl-2-oleoyl-sn- glycero-3-phosphocholine (POPC, reference 850457C), 1-palmitoyl2-oleoyl-sn-glycero-3- [phospho-rac-(1-glycerol)] (POPG, reference 840457C) and cholesterol (reference 700000P).

Multilamellar vesicles were prepared at a 10mM concentration at one of the following ratio: 7:3 POPC:Cholesterol, 7:3 POPC:POPG, 4:3:3 POPC:POPG:Cholesterol, or pure POPC. Lipids were prepared by reverse phase evaporation: they were initially dissolved in chloroform and the chloroform was evaporated in a rotary evaporator under nitrogen flow. Lipids were then resuspended in a 3:1 mixture of ether: 25 mM HEPES, 150 mM NaCl, pH 7.4 and acetone was evaporated in a rotary evaporator under nitrogen atmosphere. Buffer volume was adjusted, and vesicles were extruded by successive filtration through 0.4 and 0.2 μm polycarbonate filters. Vesicles were then sonicated for 15 to 20minutes and their charge, diameters and dispersity was characterized by dynamic light scattering. All the following experiments with MLVs were performed using the resuspension buffer (25 mM HEPES, 150 mM NaCl, pH 7.4).

¬ Imaging liposome-bound pili: POPC:Cholesterol (7:3) vesicles at a concentration of 100µM were incubated with crude pili preparations at a concentration of approximately 0.5 µM for 1hour at 37°C in 96-well plates suited to microscopy. This solution was then incubated with FM4-64 at a final concentration of 0.5µg/mL, fluorescently labeled (AlexaFluor 647) F10-I at a final concentration of 10ng/mL and FITC-dextran 15 kDa at a final concentration of 20 µg/mL for 30minutes. A similar volume of 2% low melting point agarose solution made in the resuspension buffer was added to the well to gel the solution and facilitate imaging.

162 ¬ Flow cytometry of pili prep: For flow cytometry experiments, crude pili preparations were quantified by western blot. Liposomes at a concentration of 100µM were incubated with crude pili preparations at concentrations ranging from 0 to 0.6 µM for 1hour at 37°C. The F10-I nanobody and a mouse α-His antibody were added at a final concentration of 2ng/mL and 6ng/mL respectively and incubated for 30minutes at 37°C. An AlexaFluor 647 goat anti mouse antibody was finally added and incubated at 37°C for 15 minutes. Samples were then diluted by addition of 400µL of suspension buffer and analyzed using a Gallios cytometer (Beckman Coulter). Signal analysis was done as described in Temmerman et al. 2009 729.

¬ Pili purification: Pili purification was performed as previously described. Briefly, bacteria were suspended and vortexed in 150mM ethanolamine pH 10.5. Bacteria and pili were separated by centrifugation and pili were precipitated for 1 hour by adding 9% volume of a saturated ammonium sulfate solution in 150mM ethanolamine and then centrifuged. Pili suspensions were rinsed with PBS twice and finally resuspended in PBS.

¬ Twitching motility assay: Midlog-phase GFP-bacteria were pelleted (14200g for 1min), washed and normalized to an O.D. of 0.05 in Human Endothelial-SFM + 10% FBS. 100µL of this suspension was dispensed in a channel of µ-Slide VI 0.4 (Ibidi) and incubated at 37°C - 5% CO2 for 30min. Each channel was extensively washed by pipetting up and down. Single bacteria were tracked for 2 minutes. An image was captured every 500ms using a 40x objective. Experiments were performed in triplicates. For analysis, the Fiji-integrated plugin Trackmate was used701,730. Only tracks lasting more than 1minute were analyzed. For MSD analysis, tracks obtained in Trackmate were processed using DiPer731.

¬ Mutant library selection for competence: Bacteria were transformed with genomic DNA containing a resistance cassette to erythromycin and then plated on erythromycin-containing plates. This formed the competent library.

¬ Competence assay Bacterial mutants were resuspended in GCB-transformation medium to an OD of 0.5. 10 µL of a 10-6 dilution was plated on GCB-kanamycin plates. 200µL of the appropriate bacterial suspensions were dispensed in 24-well plates and 50ng of the pMGC2 plasmid containing a spectinomycin resistance cassette was added to each well. These suspensions were incubated for 30minutes at 37°C shaking with 5%CO2. 800µL of warm GCB-transformation medium was added and cells were incubated for another 2h30. Appropriate dilutions of these bacteria were then plated on GCB-spectinomycin containing plates and CFU were counted on the next day.

163 The ratio between the transformed bacteria and the initial number of bacteria in the inoculate was used as an indication of transformation frequency.

¬ Immunofluorescence of infected cells For immunofluorescence, 35,000 cells were plated per well one day before infection in 96-well plates. On the next day, cells were infected at a MOI of 200 and incubated with bacteria at 37°C,

5% CO2 for 30 minutes. Cells were washed extensively, and incubation was prolonged for 1h30 and washing repeated or stopped at this point. Cells were then fixed with 4%formaldehyde in PBS for 30 minutes at room temperature. After rinsing, samples were permeabilized for 5 minutes in PBS containing 0.1% Triton X-100, rinsed and then blocked with PBS + 0.2% skin fish gelatin (PBSG) for 30 minutes. They were then incubated with appropriate combinations of the following: 20D9 diluted in PBSG at 2.5µg/mL and anti-ezrin polyclonal rabbit antibody (gift from Paul Mangeat) diluted in at 1:1000 in PBSG for 1hour, rinsed and subsequently incubated with the secondary antibodies Anti Mouse-AlexaFluor 568, Anti Rabbit-AlexaFluor 647 (Life Technologies) diluted in PBSG at 10 µg/mL and DAPI at 0.1 µg/mL for 1 hour at room temperature or overnight at 4°C. Samples were then washed extensively mounted in Mowiol and imaged.

¬ bEnd.3 cell culture bEnd.3 cells were grown in DMEM (Gibco) supplemented with 10% heat-inactivated FBS and passed when they reached 80% confluence.

164

References165 1. Shunk, I. V. A modification of Loeffler’s flagella stain. J. Bacteriol. 5, 181–187 (1920).

2. Loeffler, F. Eine neue Methode zum Färben der Mikroorganismen, im besonderen ihrer Wimperhaare und Geisseln. Cent. f. Bakt 6, 209–224 (1889).

3. Zettnow, E. Romanowski’s Farbung bei Bakterien. Z. Hyg. Infekt. (1899).

4. Anderson, T. F. The Nature of the Bacterial Surface. eds. A. A. Miles N. W. Pirie, 76, (1949).

5. Houwink, A. L. & van Iterson, W. Electron microscopical observations on bacterial cytology; a study on flagellation. Biochim. Biophys. Acta 5, 10–44 (1950).

6. Berry, J.-L. & Pelicic, V. Exceptionally widespread nanomachines composed of type IV pilins: the prokaryotic Swiss Army knives. FEMS Microbiol. Rev. 39, 134–154 (2015).

7. Nivaskumar, M. & Francetic, O. Type II secretion system: A magic beanstalk or a protein escalator. Biochimica et Biophysica Acta - Molecular Cell Research 1843, 1568–1577 (2014).

8. Pallen, M. J. & Matzke, N. J. From The Origin of Species to the origin of bacterial flagella. Nat. Rev. Microbiol. 4, 784–790 (2006).

9. Jarrell, K. F. & McBride, M. J. The surprisingly diverse ways that prokaryotes move. Nat. Rev. Microbiol. 6, 466–476 (2008).

10. Chaban, B., Hughes, H. V. & Beeby, M. The flagellum in bacterial pathogens: For motility and a whole lot more. Semin. Cell Dev. Biol. 46, 91–103 (2015).

11. Allen, R. D. & Baumann, P. Structure and arrangement of flagella in species of the genus Beneckea and Photobacterium fischeri. J. Bacteriol. 107, 295–302 (1971).

12. Atsumi, T., McCartert, L. & Imae, Y. Polar and lateral flagellar motors of marine Vibrio are driven by different ion-motive forces. Nature 355, 182–184 (1992).

13. Sjoblad, R. D., Emala, C. W. & Doetsch, R. N. Invited review: bacterial flagellar sheaths: structures in search of a function. Cell Motil. 3, 93–103 (1983).

14. Wolgemuth, C. W., Charon, N. W., Goldstein, S. F. & Goldstein, R. E. The Flagellar of the Spirochetes. J Mol Microbiol Biotechnol 11, (2006).

15. Holt, S. C. Anatomy and Chemistry of Spirochetes. 42, 114–160 (1978).

16. Lowe, G., Meister, M. & Berg, H. C. Rapid rotation of flagellar bundles in swimming bacteria. Nature 325, 637–640 (1987).

17. Lambert, C. et al. Characterizing the flagellar filament and the role of motility in bacterial prey-penetration by Bdellovibrio bacteriovorus. Mol. Microbiol. 60, 274–286 (2006).

18. Belas, R. Biofilms, flagella, and mechanosensing of surfaces by bacteria. Trends Microbiol. 22, 517–527 (2014).

19. Hug, I., Deshpande, S., Sprecher, K. S., Pfohl, T. & Jenal, U. Second messenger-mediated tactile response by a bacterial rotary motor. Science 358, 531–534 (2017).

20. Lillehoj, E. P., Kim, B. T. & Kim, K. C. Identification of Pseudomonas aeruginosa flagellin as an adhesin for Muc1 mucin. Am. J. Physiol. Cell. Mol. Physiol. 282, L751–L756 (2002).

21. Comstock, L. E., Denee Thomas, D. & Thomas, D. D. Characterization of Borrelia burgdorferi invasion of cultured endothelial cells. Microb. Pathog. 10, 137–148 (1991).

22. Thomas, D. D. & Comstock, L. E. Interaction of Lyme disease spirochetes with cultured eucaryotic cells. Infect. Immun. 57, 1324–6 (1989).

23. Fink, S. L. & Cookson, B. T. Pyroptosis and host cell death responses during Salmonella infection. Cell. Microbiol. 9, 2562–2570 (2007).

24. Sun, Y.-H., Rolán, H. G. & Tsolis, R. M. Injection of flagellin into the host cell cytosol by Salmonella enterica serotype Typhimurium. J. Biol. Chem. 282, 33897–901 (2007).

25. Chevance, F. F. V. & Hughes, K. T. Coordinating assembly of a bacterial macromolecular machine. Nat. Rev. Microbiol. 6, 455–465 (2008).

26. Diepold, A. & Armitage, J. P. Type III secretion systems: the bacterial flagellum and the injectisome. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 370, 20150020 (2015).

27. Li, H. & Sourjik, V. Assembly and stability of flagellar motor in Escherichia coli. Mol. Microbiol. 80, 886–899 (2011).

28. Morimoto, Y. V. et al. Assembly and stoichiometry of FliF and FlhA in Salmonella flagellar basal body. Mol. Microbiol. 91, 1214– 1226 (2014).

166 29. Minamino, T. & Namba, K. Distinct roles of the FliI ATPase and proton motive force in bacterial flagellar protein export. Nature 451, 485–488 (2008).

30. Renault, T. T. et al. Bacterial flagella grow through an injection-diffusion mechanism. Elife 6, (2017).

31. Maki-Yonekura, S., Yonekura, K. & Namba, K. Domain movements of HAP2 in the cap-filament complex formation and growth process of the bacterial flagellum. Proc. Natl. Acad. Sci. 100, 15528–15533 (2003).

32. Galkin, V. E. et al. Divergence of quaternary structures among bacterial flagellar filaments. Science 320, 382–5 (2008).

33. Blair, D. F. & Berg, H. C. The MotA protein of E. coli is a proton-conducting component of the flagellar motor. Cell 60, 439–449 (1990).

34. Hyman, H. C. & Trachtenberg, S. Point mutations that lock Salmonella typhimurium flagellar filaments in the straight right-handed and left-handed forms and their relation to filament superhelicity. J. Mol. Biol. 220, 79–88 (1991).

35. Calladine, C. R., Luisi, B. F. & Pratap, J. V. A ‘mechanistic’ explanation of the multiple helical forms adopted by bacterial flagellar filaments. J. Mol. Biol. 425, 914–928 (2013).

36. Killian, J. A. & von Heijne, G. How proteins adapt to a membrane–water interface. Trends Biochem. Sci. 25, 429–434 (2000).

37. Cornelis, G. R. The type III secretion injectisome. Nat. Rev. Microbiol. 4, 811–825 (2006).

38. Rosqvist, R., Magnusson, K. E. & Wolf-Watz, H. Target cell contact triggers expression and polarized transfer of Yersinia YopE cytotoxin into mammalian cells. EMBO J. 13, 964–72 (1994).

39. Kubori, T. et al. Supramolecular structure of the salmonella typhimurium type III protein secretion system. Science (80-. ). 280, 602–605 (1998).

40. Hueck, C. J. Type III protein secretion systems in bacterial pathogens of animals and plants. Microbiol. Mol. Biol. Rev. 62, 379–433 (1998).

41. Abby, S. S. & Rocha, E. P. C. The Non-Flagellar Type III Secretion System Evolved from the Bacterial Flagellum and Diversified into Host-Cell Adapted Systems. PLoS Genet. 8, e1002983 (2012).

42. Hu, B., Lara-tejero, M. & Kong, Q. In Situ Molecular Architecture of the Salmonella Type III Secretion Machine Article In Situ Molecular Architecture of the Salmonella Type III Secretion Machine. Cell 168, 1065–1074 (2017).

43. Sekiya, K. et al. Supermolecular structure of the enteropathogenic Escherichia coli type III secretion system and its direct interaction with the EspA-sheath-like structure. Proc. Natl. Acad. Sci. 98, 11638–11643 (2001).

44. Sun, G. W. & Gan, Y.-H. Unraveling type III secretion systems in the highly versatile Burkholderia pseudomallei. Trends Microbiol. 18, 561–568 (2010).

45. Hansen-Wester, I. & Hensel, M. Salmonella pathogenicity islands encoding type III secretion systems. Microbes Infect. 3, 549–559 (2001).

46. Puhar, A. & Sansonetti, P. J. Type III secretion system. Curr. Biol. 24, R784–R791 (2014).

47. Lara-Tejero, M., Kato, J., Wagner, S., Liu, X. & Galán, J. E. A sorting platform determines the order of protein secretion in bacterial type III systems. Science 331, 1188–91 (2011).

48. Worrall, L. J. et al. Near-atomic-resolution cryo-EM analysis of the Salmonella T3S injectisome basal body. Nature 540, 597–601 (2016).

49. Radics, J., Königsmaier, L. & Marlovits, T. C. Structure of a pathogenic type 3 secretion system in action. Nat. Struct. Mol. Biol. 21, 82–87 (2014).

50. Diepold, A. & Wagner, S. Assembly of the bacterial type III secretion machinery. FEMS Reviews 38, 802–822 (2014).

51. Mueller, C. A. et al. The V-Antigen of Yersinia Forms a Distinct Structure at the Tip of Injectisome Needles. Science (80-. ). 310, 674–676 (2005).

52. Cornelis, G. R. & Wolf-Watz, H. The Yersinia Yop virulon: a bacterial system for subverting eukaryotic cells. Mol. Microbiol. 23, 861–867 (1997).

53. Schlumberger, M. C. et al. Real-time imaging of type III secretion: Salmonella SipA injection into host cells.

54. Egelman, E. H. Cryo-EM of bacterial pili and archaeal flagellar filaments. Curr. Opin. Struct. Biol. 46, 31–37 (2017).

55. Duguid, J. P., Smith, I. W., Dempster, G, Edmunds, P. N. Non-flagellar filamentous appendages (fimbriae) and haemagglutinating activity in Bacterium coli. J Pathol Bacteriol 70, 335–348 (1955).

56. Brinton, C. C. Non-Flagellar appendages of bacteria. Nature 183, 782–786 (1959).

167 57. Xu, Q. et al. A Distinct Type of Pilus from the . Cell 165, 690–703 (2016).

58. Olsén, A., Jonsson, A. & Normark, S. Fibronectin binding mediated by a novel class of surface organelles on Escherichia coll. Nature 338, 652–655 (1989).

59. Chapman, M. R. et al. Role of Escherichia coli curli in directing amyloid fiber formation. Science 295, 851–5 (2002).

60. Chiti, F. & Dobson, C. M. Protein Misfolding, Functional Amyloid, and Human Disease. Annu. Rev. Biochem. 75, 333–366 (2006).

61. Cegelski, L. et al. Small-molecule inhibitors target Escherichia coli amyloid biogenesis and biofilm formation. Nat. Chem. Biol. 5, 913–919 (2009).

62. Fowler, D. M. et al. Functional Amyloid Formation within Mammalian Tissue. PLoS Biol. 4, e6 (2005).

63. Fowler, D. M., Koulov, A. V., Balch, W. E. & Kelly, J. W. Functional amyloid – from bacteria to humans. Trends Biochem. Sci. 32, 217–224 (2007).

64. Barnhart, M. M. & Chapman, M. R. Curli Biogenesis and Function. Annu. Rev. Microbiol. 60, 131–147 (2006).

65. Gophna, U. et al. Curli fibers mediate internalization of Escherichia coli by eukaryotic cells. Infect. Immun. 69, 2659–65 (2001).

66. Vidal, O. et al. Isolation of an Escherichia coli K-12 mutant strain able to form biofilms on inert surfaces: involvement of a new ompR allele that increases curli expression. J. Bacteriol. 180, 2442–9 (1998).

67. Römling, U. et al. Occurrence and regulation of the multicellular morphotype in Salmonella serovars important in human disease. Int. J. Med. Microbiol. 293, 273–285 (2003).

68. Hung, C. et al. Escherichia coli biofilms have an organized and complex extracellular matrix structure. MBio 4, e00645-13 (2013).

69. Tükel, Ç. et al. Toll-like receptors 1 and 2 cooperatively mediate immune responses to curli, a common amyloid from enterobacterial biofilms. Cell. Microbiol. 12, 1495–1505 (2010).

70. Tükel, Ç. et al. Responses to Amyloids of Microbial and Host Origin Are Mediated through Toll-like Receptor 2. Cell Host Microbe 6, 45–53 (2009).

71. Tükel, Ç. et al. CsgA is a pathogen-associated molecular pattern of Salmonella enterica serotype Typhimurium that is recognized by Toll-like receptor 2. Mol. Microbiol. 58, 289–304 (2005).

72. Brombacher, E., Dorel, C., Zehnder, A. J. B. & Landini, P. The curli biosynthesis regulator CsgD co-ordinates the expression of both positive and negative determinants for biofilm formation in Escherichia coli. Microbiology 149, 2847–2857 (2003).

73. Van Gerven, N., Klein, R. D., Hultgren, S. J. & Remaut, H. Bacterial Amyloid Formation: Structural Insights into Curli Biogensis. Trends Microbiol. 23, 693–706 (2015).

74. Wang, M. et al. Fimbrial proteins of porphyromonas gingivalis mediate in vivo virulence and exploit TLR2 and complement receptor 3 to persist in macrophages. J. Immunol. 179, 2349–58 (2007).

75. Hammer, N. D., Schmidt, J. C. & Chapman, M. R. The curli nucleator protein, CsgB, contains an amyloidogenic domain that directs CsgA polymerization. Proc. Natl. Acad. Sci. 104, 12494–12499 (2007).

76. Nenninger, A. A., Robinson, L. S. & Hultgren, S. J. Localized and efficient curli nucleation requires the chaperone-like amyloid assembly protein CsgF. Proc. Natl. Acad. Sci. U. S. A. 106, 900–5 (2009).

77. Goyal, P. et al. Structural and mechanistic insights into the bacterial amyloid secretion channel CsgG. Nature 516, 250–253 (2014).

78. Sleutel, M. et al. Nucleation and growth of a bacterial functional amyloid at single-fiber resolution. Nat. Chem. Biol. 13, 902–908 (2017).

79. Evans, M. L. et al. The bacterial curli system possesses a potent and selective inhibitor of amyloid formation. Mol. Cell 57, 445–55 (2015).

80. Barnhart, M. M. et al. PapD-like chaperones provide the missing information for folding of pilin proteins. Proc. Natl. Acad. Sci. 97, 7709–7714 (2000).

81. Dodson, K. W., Jacob-Dubuisson, F., Striker, R. T. & Hultgren, S. J. Outer-membrane PapC molecular usher discriminately recognizes periplasmic chaperone-pilus subunit complexes. Proc. Natl. Acad. Sci. U. S. A. 90, 3670–4 (1993).

82. Thanassi, D. G. et al. The PapC usher forms an oligomeric channel: Implications for pilus biogenesis across the outer membrane. Proc. Natl. Acad. Sci. 95, 3146–3151 (1998).

83. Waksman, G. & Hultgren, S. J. Structural biology of the chaperone–usher pathway of pilus biogenesis. Nat. Rev. Microbiol. 7, 765– 774 (2009).

84. Nuccio, S.-P. & Baumler, A. J. Evolution of the Chaperone/Usher Assembly Pathway: Fimbrial Classification Goes Greek. Microbiol. Mol. Biol. Rev. 71, 551–575 (2007).

168 85. Connell, I. et al. Type 1 fimbrial expression enhances Escherichia coli virulence for the urinary tract. Proc. Natl. Acad. Sci. 93, (1996).

86. Mulvey, M. A. et al. Induction and evasion of host defenses by type 1-piliated uropathogenic Escherichia coli. Science 282, 1494– 7 (1998).

87. Roberts, J. A. et al. The Gal(al-4)Gal-specific tip adhesin of Escherichia coli P-fimbriae is needed for pyelonephritis to occur in the normal urinary tract. 91, 11889–11893 (1994).

88. Abraham, S. N., Goguen, J. D. & Beachey, E. H. Hyperadhesive mutant of type 1-fimbriated Escherichia coli associated with formation of FimH organelles (fimbriosomes). Infect. Immun. 56, 1023–9 (1988).

89. Lund, B., Lindberg, F., Marklund, B. I. & Normark, S. The PapG protein is the alpha-D-galactopyranosyl-(1----4)-beta-D- galactopyranose-binding adhesin of uropathogenic Escherichia coli. Proc. Natl. Acad. Sci. 84, (1987).

90. Dodson, K. W. et al. Structural basis of the interaction of the pyelonephritic E. coli adhesin to its human kidney receptor. Cell 105, 733–43 (2001).

91. Martinez, J. J., Mulvey, M. A., Schilling, J. D., Pinkner, J. S. & Hultgren, S. J. Type 1 pilus-mediated bacterial invasion of bladder epithelial cells. EMBO J. 19, 2803–12 (2000).

92. Ong, C. L. Y. et al. Identification of type 3 fimbriae in uropathogenic Escherichia coli reveals a role in biofilm formation. J. Bacteriol. 190, 1054–1063 (2008).

93. Langstraat, J., Bohse, M. & Clegg, S. Type 3 fimbrial shaft (MrkA) of Klebsiella pneumoniae, but not the fimbrial adhesin (MrkD), facilitates biofilm formation. Infect. Immun. 69, 5805–5812 (2001).

94. Vallet, I., Olson, J. W., Lory, S., Lazdunski, A. & Filloux, A. The chaperone/usher pathways of Pseudomonas aeruginosa: Identification of fimbrial gene clusters (cup) and their involvement in biofilm formation. Proc. Natl. Acad. Sci. 98, 6911–6916 (2001).

95. Kuehn, M. J., Heuser, J., Normark, S. & Hultgren, S. J. P pili in uropathogenic E. coli are composite fibres with distinct fibrillar adhesive tips. Nature 356, 252–255 (1992).

96. Lindberg, F., Lund, B., Johansson, L. & Normark, S. Localization of the receptor-binding protein adhesin at the tip of the bacterial pilus. Nature 328, 84–7 (1987).

97. Hospenthal, M. K. et al. The Cryoelectron Microscopy Structure of the Type 1 Chaperone-Usher Pilus Rod. Structure 0, (2017).

98. Choudhury, D. et al. X-ray structure of the FimC-FimH chaperone-adhesin complex from uropathogenic Escherichia coli. Science 285, 1061–6 (1999).

99. Verger, D., Bullitt, E., Hultgren, S. J. & Waksman, G. Crystal structure of the P pilus rod subunit PapA. PLoS Pathog. 3, e73 (2007).

100. Puorger, C., Vetsch, M., Wider, G. & Glockshuber, R. Structure, folding and stability of FimA, the main structural subunit of type 1 pili from uropathogenic Escherichia coli strains. J. Mol. Biol. 412, 520–35 (2011).

101. Vetsch, M. et al. Pilus chaperones represent a new type of protein-folding catalyst. Nature 431, 329–333 (2004).

102. Sauer, F. G. et al. Structural Basis of Chaperone Function and Pilus Biogenesis. Science (80-. ). 285, 1058–1061 (1999).

103. Remaut, H. et al. Fiber Formation across the by the Chaperone/Usher Pathway. Cell 133, 640–652 (2008).

104. Phan, G. et al. Crystal structure of the FimD usher bound to its cognate FimC–FimH substrate. Nature 474, 49–53 (2011).

105. Saulino, E. T., Thanassi, D. G., Pinkner, J. S. & Hultgren, S. J. Ramifications of kinetic partitioning on usher-mediated pilus biogenesis. EMBO J. 17, 2177–2185 (1998).

106. Nishiyama, M., Vetsch, M., Puorger, C., Jelesarov, I. & Glockshuber, R. Identification and characterization of the chaperone-subunit complex-binding domain from the type 1 pilus assembly platform FimD. J. Mol. Biol. 330, 513–25 (2003).

107. Sauer, F. G., Pinkner, J. S., Waksman, G. & Hultgren, S. J. Chaperone Priming of Pilus Subunits Facilitates a Topological Transition that Drives Fiber Formation. Cell 111, 543–551 (2002).

108. Remaut, H. et al. Donor-Strand Exchange in Chaperone-Assisted Pilus Assembly Proceeds through a Concerted β Strand Displacement Mechanism. Mol. Cell 22, 831–842 (2006).

109. Ng, T. W., Akman, L., Osisami, M. & Thanassi, D. G. The usher N terminus is the initial targeting site for chaperone-subunit complexes and participates in subsequent pilus biogenesis events. J. Bacteriol. 186, 5321–31 (2004).

110. Waksman, G. Structural and Molecular Biology of a Protein-Polymerizing Nanomachine for Pilus Biogenesis. J. Mol. Biol. 429, 2654–2666 (2017).

111. Sauer, M. M. et al. Catch-bond mechanism of the bacterial adhesin FimH. Nat. Commun. 7, 10738 (2016).

112. Thomas, W. E., Trintchina, E., Forero, M., Vogel, V. & Sokurenko, E. V. Bacterial adhesion to target cells enhanced by shear force.

169 Cell 109, 913–923 (2002).

113. Björnham, O., Nilsson, H., Andersson, M. & Schedin, S. Physical properties of the specific PapG–galabiose binding in E. coli P pili- mediated adhesion. Eur. Biophys. J. 38, 245–254 (2009).

114. Jass, J. et al. Physical Properties of Escherichia coli P Pili Measured by Optical Tweezers. Biophys. J. 87, 4271–4283 (2004).

115. Fällman, E., Schedin, S., Jass, J., Uhlin, B.-E. & Axner, O. The unfolding of the P pili quaternary structure by stretching is reversible, not plastic. EMBO Rep. 6, 52–6 (2005).

116. Bullitt, E. & Makowski, L. Structural polymorphism of bacterial adhesion pili. Nature 373, 164–167 (1995).

117. Forero, M., Yakovenko, O., Sokurenko, E. V, Thomas, W. E. & Vogel, V. Uncoiling Mechanics of Escherichia coli Type I Fimbriae Are Optimized for Catch Bonds. PLoS Biol. 4, e298 (2006).

118. Zakrisson, J., Wiklund, K., Axner, O. & Andersson, M. The Shaft of the Type 1 Fimbriae Regulates an External Force to Match the FimH Catch Bond. Biophys. J. 104, 2137–2148 (2013).

119. Zakrisson, J., Wiklund, K., Axner, O. & Andersson, M. Helix-like biopolymers can act as dampers of force for bacteria in flows. Eur. Biophys. J. 41, 551–560 (2012).

120. Miller, E., Garcia, T., Hultgren, S. & Oberhauser, A. F. The Mechanical Properties of E. coli Type 1 Pili Measured by Atomic Force Microscopy Techniques. Biophys. J. 91, 3848–3856 (2006).

121. Habenstein, B. et al. Hybrid Structure of the Type 1 Pilus of Uropathogenic Escherichia coli. Angew. Chemie Int. Ed. 54, 11691– 11695 (2015).

122. Hospenthal, M. K. et al. Structure of a Chaperone-Usher Pilus Reveals the Molecular Basis of Rod Uncoiling. Cell 164, 269–278 (2016).

123. Spaulding, C. N. et al. Functional role of the type 1 pilus rod structure in mediating host-pathogen interactions. Elife 7, e31662 (2018).

124. D ’enfert, C., Ryterl, A., Pugsley, A. P. & Schwartz, M. Cloning and expression in Escherichia coli of the Klebsiella pneumoniae genes for production, surface localization and secretion of the lipoprotein pullulanase. EMBO J. 6, 3531–3538 (1987).

125. Francetic, O., Belin, D., Badaut, C. & Pugsley, A. P. Expression of the endogenous type II secretion pathway in Escherichia coli leads to chitinase secretion. EMBO J. 19, 6697–703 (2000).

126. Sandkvist, M. et al. General Secretion Pathway (eps) Genes Required for Toxin Secretion and Outer Membrane Biogenesis in Vibrio cholerae. J. Bacteriol. 179, 6994–7003 (1997).

127. Filloux, A. et al. Protein secretion in gram-negative bacteria: transport across the outer membrane involves common mechanisms in different bacteria. EMBO J. 9, 4323–9 (1990).

128. Hobbs, M. & Mattick, J. S. Common components in the assembly of type 4 fimbriae, DNA transfer systems, filamentous phage and protein-secretion apparatus: a general system for the formation of surface-associated protein complexes. Mol. Microbiol. 10, 233– 243 (1993).

129. Nunn, D. Bacterial Type II protein export and pilus biogenesis: more than just homologies? Trends Cell Biol. 9, 402–408 (1999).

130. Campos, M., Cisneros, D. A., Nivaskumar, M. & Francetic, O. The type II secretion system – a dynamic fiber assembly nanomachine. Res. Microbiol. 164, 545–555 (2013).

131. Sauvonnet, N., Vignon, G., Pugsley, A. P. & Gounon, P. Pilus formation and protein secretion by the same machinery in Escherichia coli. EMBO J. 19, 2221–2228 (2000).

132. Vignon, G. et al. Type IV-like pili formed by the type II secreton: specificity, composition, bundling, polar localization, and surface presentation of peptides. J. Bacteriol. 185, 3416–28 (2003).

133. Durand, E. et al. Type II protein secretion in Pseudomonas aeruginosa: the pseudopilus is a multifibrillar and adhesive structure. J. Bacteriol. 185, 2749–58 (2003).

134. Hirst, T. R. & Holmgren, J. Conformation of protein secreted across bacterial outer membranes: a study of enterotoxin translocation from Vibrio cholerae. Proc. Natl. Acad. Sci. 84, (1987).

135. Pugsley, A. P. Translocation of a folded protein across the outer membrane in Escherichia coli. PNAS 89, 12058–12062 (1992).

136. Sikora, A. E., Zielke, R. A., Lawrence, D. A., Andrews, P. C. & Sandkvist, M. Proteomic analysis of the Vibrio cholerae type II secretome reveals new proteins, including three related serine proteases. J. Biol. Chem. 286, 16555–66 (2011).

137. Ball, G., Durand, E., Lazdunski, A. & Filloux, A. A novel type II secretion system in Pseudomonas aeruginosa. Mol. Microbiol. 43, 475–85 (2002).

170 138. Cadoret, F., Ball, G., Douzi, B. & Voulhoux, R. Txc, a new type II secretion system of Pseudomonas aeruginosa strain PA7, is regulated by the TtsS/TtsR two-component system and directs specific secretion of the CbpE chitin-binding protein. J. Bacteriol. 196, 2376–86 (2014).

139. Cianciotto, N. P. & White, R. C. Expanding Role of Type II Secretion in Bacterial Pathogenesis and Beyond. Infect. Immun. 85, e00014-17 (2017).

140. Tauschek, M., Gorrell, R. J., Strugnell, R. A. & Robins-Browne, R. M. Identification of a protein secretory pathway for the secretion of heat-labile enterotoxin by an enterotoxigenic strain of Escherichia coli. Proc. Natl. Acad. Sci. 99, 7066–7071 (2002).

141. Baldi, D. L. et al. The type II secretion system and its ubiquitous lipoprotein substrate, SslE, are required for biofilm formation and virulence of enteropathogenic Escherichia coli. Infect. Immun. 80, 2042–52 (2012).

142. Johnson, T. L. et al. The Type II Secretion System Delivers Matrix Proteins for Biofilm Formation by Vibrio cholerae. J. Bacteriol. 196, 4245–4252 (2014).

143. Ho, T. D., Davis, B. M., Ritchie, J. M. & Waldor, M. K. Type 2 secretion promotes enterohemorrhagic Escherichia coli adherence and intestinal colonization. Infect. Immun. 76, 1858–65 (2008).

144. McCoy-Simandle, K. et al. Legionella pneumophila type II secretion dampens the cytokine response of infected macrophages and epithelia. Infect. Immun. 79, 1984–1997 (2011).

145. Mallama, C. A., McCoy-Simandle, K. & Cianciotto, N. P. The Type II Secretion System of Legionella pneumophila Dampens the MyD88 and Toll-Like Receptor 2 Signaling Pathway in Infected Human Macrophages. Infect. Immun. 85, e00897-16 (2017).

146. von Tils, D., Blädel, I., Schmidt, M. A. & Heusipp, G. Type II secretion in Yersinia—a secretion system for pathogenicity and environmental fitness. Front. Cell. Infect. Microbiol. 2, 160 (2012).

147. Sikora, A. E. Proteins Secreted via the Type II Secretion System: Smart Strategies of Vibrio cholerae to Maintain Fitness in Different Ecological Niches. PLoS Pathog. 9, e1003126 (2013).

148. Pugsley, A. P. The complete general secretory pathway in gram-negative bacteria. Microbiol. Rev. 57, 50–108 (1993).

149. Thomassin, J.-L., Santos Moreno, J., Guilvout, I., Tran Van Nhieu, G. & Francetic, O. The trans-envelope architecture and function of the type 2 secretion system: new insights raising new questions. Mol. Microbiol. 105, 211–226 (2017).

150. Pugsley, A. P., Kornacker, M. G. & Poquet, I. The general protein-export pathway is directly required for extracellular pullulanase secretion in Escherichia coli K12. Mol. Microbiol. 5, 343–52 (1991).

151. Voulhoux, R. et al. Involvement of the twin-arginine translocation system in protein secretion via the type II pathway. EMBO J. 20, 6735–41 (2001).

152. Py, B., Loiseau, L. & Barras, F. An inner membrane platform in the type II secretion machinery of Gram-negative bacteria. EMBO Rep. 2, 244–8 (2001).

153. Possot, O. M., Letellier, L. & Pugsley, A. P. Energy requirement for pullulanase secretion by the main terminal branch of the general secretory pathway. Mol. Microbiol. 24, 457–464 (1997).

154. Pugsley, A. P. & Dupuy, B. An enzyme with type IV prepilin peptidase activity is required to process components of the general extracellular protein secretion pathway of Klebsiella oxytoca. Mol. Microbiol. 6, 751–760 (1992).

155. Dupuy, B., Taha, M.-K., Possot, O., Marchal, C. & Pugsley, A. P. PuIO, a component of the pullulanase secretion pathway of Klebsiella oxytoca, correctly and efficiently processes gonococcal type IV prepilin in Escherichia coli. Mol. Microbiol. 6, 1887–1894 (1992).

156. Cisneros, D. A., Pehau-Arnaudet, G. & Francetic, O. Heterologous assembly of type IV pili by a type II secretion system reveals the role of minor pilins in assembly initiation. Mol. Microbiol. 86, 805–818 (2012).

157. Nivaskumar, M. et al. Pseudopilin residue E5 is essential for recruitment by the type 2 secretion system assembly platform. Mol. Microbiol. 101, 924–941 (2016).

158. Polar N-terminal Residues Conserved in Type 2 Secretion Pseudopilins Determine Subunit Targeting and Membrane Extraction Steps during Fibre Assembly. J. Mol. Biol. 429, 1746–1765 (2017).

159. López-Castilla, A. et al. Structure of the calcium-dependent type 2 secretion pseudopilus. Nat. Microbiol. 1 (2017). doi:10.1038/s41564-017-0041-2

160. Koster, M. et al. The outer membrane component, YscC, of the Yop secretion machinery of Yersinia enterocolitica forms a ring- shaped multimeric complex. Mol. Microbiol. 26, 789–97 (1997).

161. Bitter, W., Koster, M., Latijnhouwers, M., de Cock, H. & Tommassen, J. Formation of oligomeric rings by XcpQ and PilQ, which are involved in protein transport across the outer membrane of Pseudomonas aeruginosa. Mol. Microbiol. 27, 209–19 (1998).

171 162. de Groot, A. et al. Exchange of Xcp (Gsp) secretion machineries between Pseudomonas aeruginosa and Pseudomonas alcaligenes: species specificity unrelated to substrate recognition. J. Bacteriol. 183, 959–67 (2001).

163. Michel, G. P. F., Durand, E. & Filloux, A. XphA/XqhA, a novel GspCD subunit for type II secretion in Pseudomonas aeruginosa. J. Bacteriol. 189, 3776–83 (2007).

164. Korotkov, K. V. et al. Structural and Functional Studies on the Interaction of GspC and GspD in the Type II Secretion System. PLoS Pathog. 7, e1002228 (2011).

165. Douzi, B., Ball, G., Cambillau, C., Tegoni, M. & Voulhoux, R. Deciphering the Xcp Pseudomonas aeruginosa type II secretion machinery through multiple interactions with substrates. J. Biol. Chem. 286, 40792–801 (2011).

166. Yoshimura, F., Takahashi, K., Nodasaka, Y. & Suzuki1, T. Purification and Characterization of a Novel Type of Fimbriae from the Oral Anaerobe Bacteroides gingivalis. J. Bacteriol. 160, 949–957 (1984).

167. Hamada, N., Sojar, H. T., Cho, M.-I. & Genco, R. J. Isolation and Characterization of a Minor from Porphyromonas gingivalis. Infect. Immun. 64, 4788–4794 (1996).

168. Malek, R. et al. Inactivation of the Porphyromonas gingivalis fimA gene blocks periodontal damage in gnotobiotic rats. J. Bacteriol. 176, 1052–9 (1994).

169. Nakamura, T., Amano, A., Nakagawa, I. & Hamada, S. Specific interactions between Porphyromonas gingivalis fimbriae and human extracellular matrix proteins. FEMS Microbiol. Lett. 175, 267–72 (1999).

170. Amano, A. et al. Molecular interactions of Porphyromonas gingivalis fimbriae with host proteins: kinetic analyses based on surface plasmon resonance. Infect. Immun. 67, 2399–405 (1999).

171. Hajishengallis, G., Wang, M., Harokopakis, E., Triantafilou, M. & Triantafilou, K. Porphyromonas gingivalis fimbriae proactively modulate beta2 integrin adhesive activity and promote binding to and internalization by macrophages. Infect. Immun. 74, 5658– 66 (2006).

172. Hajishengallis, G., Wang, M., Liang, S., Triantafilou, M. & Triantafilou, K. Pathogen induction of CXCR4/TLR2 cross-talk impairs host defense function. Proc. Natl. Acad. Sci. U. S. A. 105, 13532–7 (2008).

173. Park, Y. et al. Short fimbriae of Porphyromonas gingivalis and their role in coadhesion with Streptococcus gordonii. Infect. Immun. 73, 3983–9 (2005).

174. Shoji, M. et al. Recombinant Porphyromonas gingivalis FimA preproprotein expressed in Escherichia coli is lipidated and the mature or processed recombinant FimA protein forms a short filament in vitro. Can. J. Microbiol. 56, 959–967 (2010).

175. Shoji, M. et al. The major structural components of two cell surface filaments of Porphyromonas gingivalis are matured through lipoprotein precursors. Mol. Microbiol. 52, 1513–1525 (2004).

176. Nakayama, K., Yoshimura, F., Kadowaki, T. & Yamamoto, K. Involvement of arginine-specific cysteine proteinase (Arg-gingipain) in fimbriation of Porphyromonas gingivalis. J. Bacteriol. 178, 2818–24 (1996).

177. Lee, J. Y., Sojar, H. T., Bedi, G. S. & Genco, R. J. Porphyromonas (Bacteroides) gingivalis fimbrillin: size, amino-terminal sequence, and antigenic heterogeneity. Infect. Immun. 59, 383–9 (1991).

178. Nagano, K., Hasegawa, Y., Murakami, Y., Nishiyama, S. & Yoshimura, F. FimB regulates FimA fimbriation in porphyromonas gingivalis. J. Dent. Res. 89, 903–908 (2010).

179. Hasegawa, Y. et al. Anchoring and length regulation of Porphyromonas gingivalis Mfa1 fimbriae by the downstream gene product Mfa2. Microbiology 155, 3333–3347 (2009).

180. Hall, M., Hasegawa, Y., Yoshimura, F. & Persson, K. Structural and functional characterization of shaft, anchor, and tip proteins of the Mfa1 fimbria from the periodontal pathogen Porphyromonas gingivalis. Sci. Rep. 8, 1793 (2018).

181. Yanagawa, R., Otsuki, K. & Tokui, T. Electron microscopy of fine structure of corynebacterium renale with special reference to pili. Jpn. J. Vet. Res. (1967).

182. Mora, M. et al. Group A Streptococcus produce pilus-like structures containing protective antigens and Lancefield T antigens. Proc. Natl. Acad. Sci. 102, 15641–15646 (2005).

183. Barocchi, M. A. et al. A pneumococcal pilus influences virulence and host inflammatory responses. Proc. Natl. Acad. Sci. U. S. A. 103, 2857–62 (2006).

184. Kankainen, M. et al. Comparative genomic analysis of Lactobacillus rhamnosus GG reveals pili containing a human- mucus binding protein. Proc. Natl. Acad. Sci. U. S. A. 106, 17193–17198 (2009).

185. Turroni, F. et al. Role of sortase-dependent pili of Bifidobacterium bifidum PRL2010 in modulating bacterium-host interactions. Proc. Natl. Acad. Sci. U. S. A. 110, 11151–6 (2013).

172 186. Mandlik, A., Swierczynski, A., Das, A. & Ton-That, H. Pili in Gram-positive bacteria: assembly, involvement in colonization and biofilm development. Trends Microbiol. 16, 33–40 (2008).

187. Mandlik, A., Swierczynski, A., Das, A. & Ton-That, H. Corynebacterium diphtheriae employs specific minor pilins to target human pharyngeal epithelial cells. Mol. Microbiol. 64, 111–124 (2007).

188. Dramsi, S. et al. Assembly and role of pili in group B streptococci. Mol. Microbiol. 60, 1401–1413 (2006).

189. Chang, C., Mandlik, A., Das, A. & Ton-That, H. Cell surface display of minor pilin adhesins in the form of a simple heterodimeric assembly in Corynebacterium diphtheriae. Mol. Microbiol. 79, 1236–1247 (2011).

190. Palmer, R. J., Kazmerzak, K., Hansen, M. C. & Kolenbrander, P. E. Mutualism versus independence: strategies of mixed-species oral biofilms in vitro using saliva as the sole nutrient source. Infect. Immun. 69, 5794–804 (2001).

191. Manetti, A. G. O. et al. Streptococcus pyogenes pili promote pharyngeal cell adhesion and biofilm formation. Mol. Microbiol. 64, 968–983 (2007).

192. Ton-That, H. & Schneewind, O. Assembly of pili on the surface of Corynebacterium diphtheriae. Mol. Microbiol. 50, 1429–1438 (2003).

193. Hilleringmann, M. et al. Molecular architecture of Streptococcus pneumoniae TIGR4 pili. EMBO J. 28, 3921–3930 (2009).

194. Ton-That, H., Marraffini, L. A. & Schneewind, O. and pilin elements involved in pilus assembly of Corynebacterium diphtheriae. Mol. Microbiol. 53, 251–261 (2004).

195. Kang, H. J., Coulibaly, F., Clow, F., Proft, T. & Baker, E. N. Stabilizing isopeptide bonds revealed in gram-positive bacterial pilus structure. Science 318, 1625–8 (2007).

196. Manzano, C. et al. Sortase-Mediated Pilus Fiber Biogenesis in Streptococcus pneumoniae. Structure 16, 1838–1848 (2008).

197. Neiers, F. et al. Two Crystal Structures of Pneumococcal Pilus Sortase C Provide Novel Insights into Catalysis and Substrate Specificity. J. Mol. Biol. 393, 704–716 (2009).

198. Linke, C. et al. Crystal structure of the minor pilin FctB reveals determinants of Group A streptococcal pilus anchoring. J. Biol. Chem. 285, 20381–9 (2010).

199. Nobbs, A. H. et al. Sortase A utilizes an ancillary protein anchor for efficient cell wall anchoring of pili in . Infect. Immun. 76, 3550–60 (2008).

200. Swierczynski, A. & Ton-That, H. Type III pilus of corynebacteria: Pilus length is determined by the level of its major pilin subunit. J. Bacteriol. 188, 6318–25 (2006).

201. Swaminathan, A. et al. Housekeeping sortase facilitates the cell wall anchoring of pilus polymers in Corynebacterium diphtheriae. Mol. Microbiol. 66, 961–974 (2007).

202. Budzik, J. M., Oh, S.-Y. & Schneewind, O. Cell wall anchor structure of BcpA pili in anthracis. J. Biol. Chem. 283, 36676– 86 (2008).

203. Krishnan, V. et al. An IgG-like Domain in the Minor Pilin GBS52 of Streptococcus agalactiae Mediates Lung Epithelial Cell Adhesion. Structure 15, 893–903 (2007).

204. Kang, H. J., Paterson, N. G., Gaspar, A. H., Ton-That, H. & Baker, E. N. The Corynebacterium diphtheriae shaft pilin SpaA is built of tandem Ig-like modules with stabilizing isopeptide and disulfide bonds. Proc. Natl. Acad. Sci. U. S. A. 106, 16967–71 (2009).

205. Kang, H. J., Coulibaly, F., Proft, T. & Baker, E. N. Crystal Structure of Spy0129, a Streptococcus pyogenes Class B Sortase Involved in Pilus Assembly. PLoS One 6, e15969 (2011).

206. Echelman, D. J. et al. CnaA domains in bacterial pili are efficient dissipaters of large mechanical shocks. Proc. Natl. Acad. Sci. 113, 2490–2495 (2016).

207. Izoré, T. et al. Structural Basis of Host Cell Recognition by the Pilus Adhesin from Streptococcus pneumoniae. Structure 18, 106– 115 (2010).

208. Linke-Winnebeck, C. et al. Structural model for covalent adhesion of the Streptococcus pyogenes pilus through a thioester bond. J. Biol. Chem. 289, 177–89 (2014).

209. Pointon, J. A. et al. A highly unusual thioester bond in a pilus adhesin is required for efficient host cell interaction. J. Biol. Chem. 285, 33858–66 (2010).

210. Echelman, D. J., Lee, A. Q. & Fernández, J. M. Mechanical forces regulate the reactivity of a thioester bond in a bacterial adhesin. J. Biol. Chem. 292, 8988–8997 (2017).

211. Walden, M. et al. An internal thioester in a pathogen surface protein mediates covalent host binding. Elife 4, (2015).

173 212. Moissl, C., Rachel, R., Briegel, A., Engelhardt, H. & Huber, R. The unique structure of archaeal ‘hami’, highly complex cell appendages with nano-grappling hooks. Mol. Microbiol. 56, 361–370 (2005).

213. Perras, A. K. et al. S-layers at second glance? Altiarchaeal grappling hooks (hami) resemble archaeal S-layer proteins in structure and sequence. Front. Microbiol. 6, 543 (2015).

214. Perras, A. K. et al. Grappling archaea: Ultrastructural analyses of an uncultivated, cold-loving archaeon, and its biofilm. Front. Microbiol. 5, 397 (2014).

215. Probst, A. J. et al. Biology of a widespread uncultivated archaeon that contributes to carbon fixation in the subsurface. Nat. Commun. 5, 5497 (2014).

216. Henneberger, R., Moissl, C., Amann, T., Rudolph, C. & Huber, R. New insights into the lifestyle of the cold-loving SM1 euryarchaeon: natural growth as a monospecies biofilm in the subsurface. Appl. Environ. Microbiol. 72, 192–9 (2006).

217. Stetter, K. O., König, H. & Stackebrandt, E. Pyrodictium gen. nov., a New Genus of Submarine Disc-Shaped Sulphur Reducing Archaebacteria Growing Optimally at 105°C. Syst. Appl. Microbiol. 4, 535–551 (1983).

218. Horn, C., Paulmann, B., Kerlen, G., Junker, N. & Huber, H. In vivo observation of cell division of anaerobic hyperthermophiles by using a high-intensity dark-field microscope. J. Bacteriol. 181, 5114–8 (1999).

219. Nickell, S., Hegerl, R., Baumeister, W. & Rachel, R. Pyrodictium cannulae enter the periplasmic space but do not enter the cytoplasm, as revealed by cryo-electron tomography. J. Struct. Biol. 141, 34–42 (2003).

220. Moll, G. & Ahrens, R. Ein neuer Fimbrientyp. Arch. Mikrobiol. 70, 361–368 (1970).

221. Bernadac, A. et al. Structural properties of the tubular appendage spinae from marine bacterium Roseobacter sp. strain YSCB. Sci. Rep. 2, 950 (2012).

222. Bayer, M. E. & Easterbrook, K. Tubular spinae are long-distance connectors between bacteria. J. Gen. Microbiol. 137, 1081–1086 (1991).

223. Thoma, C. et al. The Mth60 fimbriae of Methanothermobacter thermoautotrophicus are functional adhesins. Environ. Microbiol. 10, 2785–2795 (2008).

224. Houwink, A. L. Flagella, Gas and Cell-wall Structure in Halobacterium halobium; an Electron Microscope Study. J. Gen. Microbiol. 15, 146–150 (1956).

225. Alam, M. & Oesterhelt, D. Morphology, function and isolation of halobacterial flagella. J. Mol. Biol. 176, 459–475 (1984).

226. Woese, C. R. & Fox, G. E. Phylogenetic structure of the prokaryotic domain: the primary kingdoms. Proc. Natl. Acad. Sci. U. S. A. 74, 5088–90 (1977).

227. Woese, C. R., Kandler, O. & Wheelis, M. L. Towards a natural system of organisms: proposal for the domains Archaea, Bacteria, and Eucarya. Proc. Natl. Acad. Sci. U. S. A. 87, 4576–9 (1990).

228. Jarrell, K. F., Bayley, D. P. & Kostyukova, A. S. The archaeal flagellum: a unique motility structure. J. Bacteriol. 178, 5057–64 (1996).

229. Faguy, D. M., Jarrell, K. F., Kuzio, J. & Kalmokoff, M. L. Molecular analysis of archaeal flagellins: similarity to the type IV pilin – transport superfamily widespread in bacteria. Can. J. Microbiol. 40, 67–71 (1994).

230. Albers, S.-V. & Jarrell, K. F. The archaellum: how archaea swim. Front. Microbiol. 6, 23 (2015).

231. Jarrell, K. F. & Albers, S.-V. The archaellum: an old motility structure with a new name. Trends Microbiol. 20, 307–312 (2012).

232. Eichler, J. Response to Jarrell and Albers: the name says it all. Trends Microbiol. 20, 512–513 (2012).

233. Wirth, R. Response to Jarrell and Albers: seven letters less does not say more. Trends Microbiol. 20, 511–512 (2012).

234. Szabó, Z. et al. Flagellar motility and structure in the hyperthermoacidophilic archaeon Sulfolobus solfataricus. J. Bacteriol. 189, 4305–9 (2007).

235. Herzog, B. & Wirth, R. Swimming behavior of selected species of Archaea. Appl. Environ. Microbiol. 78, 1670–4 (2012).

236. Marwan, W., Schäfer, W. & Oesterhelt, D. Signal transduction in Halobacterium depends on fumarate. EMBO J. 9, 355–362 (1990).

237. Marwan, W., Alam, M. & Oesterhelt, D. Rotation and switching of the flagellar motor assembly in Halobacterium halobium. J. Bacteriol. 173, 1971–1977 (1991).

238. Briegel, A. et al. Structural conservation of chemotaxis machinery across Archaea and Bacteria. Environ. Microbiol. Rep. 7, 414– 419 (2015).

239. Bellack, A., Huber, H., Rachel, R., Wanner, G. & Wirth, R. Methanocaldococcus villosus sp. nov., a heavily flagellated archaeon that adheres to surfaces and forms cell-cell contacts. Int. J. Syst. Evol. Microbiol. 61, 1239–1245 (2011).

174 240. Jarrell, K. F., Stark, M., Nair, D. B. & Chong, J. P. J. Flagella and pili are both necessary for efficient attachment of Methanococcus maripaludis to surfaces. FEMS Microbiol. Lett. 319, 44–50 (2011).

241. Näther, D. J., Rachel, R., Wanner, G. & Wirth, R. Flagella of Pyrococcus furiosus: Multifunctional organelles, made for swimming, adhesion to various surfaces, and cell-cell contacts. J. Bacteriol. 188, 6915–6923 (2006).

242. Zolghadr, B. et al. Appendage-mediated surface adherence of Sulfolobus solfataricus. J. Bacteriol. 192, 104–10 (2010).

243. Schopf, S., Wanner, G., Rachel, R. & Wirth, R. An archaeal bi-species biofilm formed by Pyrococcus furiosus and Methanopyrus kandleri. Arch. Microbiol. 190, 371–377 (2008).

244. Albers, S.-V., Szabó, Z. & Driessen, A. J. M. Archaeal homolog of bacterial type IV prepilin signal peptidases with broad substrate specificity. J. Bacteriol. 185, 3918–25 (2003).

245. Szabó, Z. et al. Identification of diverse archaeal proteins with class III signal peptides cleaved by distinct archaeal prepilin peptidases. J. Bacteriol. 189, 772–8 (2007).

246. Poweleit, N. et al. CryoEM structure of the Methanospirillum hungatei archaellum reveals structural features distinct from the bacterial flagellum and type IV pili. Nat. Microbiol. 2, 16222 (2016).

247. Daum, B. et al. Structure and in situ organisation of the Pyrococcus furiosus archaellum machinery. Elife 6, e27470 (2017).

248. Streif, S., Staudinger, W. F., Marwan, W. & Oesterhelt, D. Flagellar Rotation in the Archaeon Halobacterium salinarum Depends on ATP. J. Mol. Biol. 384, 1–8 (2008).

249. Ghosh, A., Hartung, S., van der Does, C., Tainer, J. A. & Albers, S.-V. Archaeal flagellar ATPase motor shows ATP-dependent hexameric assembly and activity stimulation by specific lipid binding. Biochem. J. 437, 43–52 (2011).

250. Reindl, S. et al. Insights into FlaI Functions in Archaeal Motor Assembly and Motility from Structures, Conformations, and Genetics. Mol. Cell 49, 1069–1082 (2013).

251. Banerjee, A., Neiner, T., Tripp, P. & Albers, S. V. Insights into subunit interactions in the Sulfolobus acidocaldarius archaellum cytoplasmic complex. FEBS J. 280, 6141–6149 (2013).

252. Briegel, A. et al. Morphology of the archaellar motor and associated cytoplasmic cone in Thermococcus kodakaraensis. EMBO Rep. 18, 1660–1670 (2017).

253. Banerjee, A. et al. FlaF Is a β-Sandwich Protein that Anchors the Archaellum in the Archaeal Cell Envelope by Binding the S-Layer Protein. Structure 23, 863–872 (2015).

254. Chandran Darbari, V. & Waksman, G. Structural Biology of Bacterial Type IV Secretion Systems. Annu. Rev. Biochem. 84, 603–629 (2015).

255. Cascales, E. & Christie, P. J. The versatile bacterial type IV secretion systems. Nat. Rev. Microbiol. 1, 137–149 (2003).

256. Lederberg, J. & Tatum, E. L. Gene recombination in Escherichia coli. Nature 158, 558 (1946).

257. Lederberg, J., Cavalli, L. L. & Lederberg, R. M. Sex compatibility in Eschericha coli. Genetics 720–730 (1952).

258. Cavalli, L. L., Lederberg, J. & Lederberg, E. M. An infective factor controlling sex compatibility in Bacterium coli. J. Gen. Microbiol. 8, 89–103 (1953).

259. Brinton, C. C., Gemski, P., Carnahan, J., Jr. & Carnahan, J. A new type of bacterial pilus genetically controlled by the fertility factor of E. coli K12 and its role in chromosome transfer. Proc. Natl. Acad. Sci. U. S. A. 52, 776–83 (1964).

260. Thomas, C. M. & Nielsen, K. M. Mechanisms of and Barriers to, Horizontal Gene Transfer between Bacteria. Nat. Rev. Microbiol. 3, 711–721 (2005).

261. Grohmann, E., Christie, P. J., Waksman, G. & Backert, S. Type IV secretion in Gram-negative and Gram-positive bacteria. Molecular Microbiology 107, 455–471 (2018).

262. Redzej, A. et al. Structure of a VirD4 coupling protein bound to a VirB type IV secretion machinery. EMBO J. 36, 3080–3095 (2017).

263. Ilangovan, A. et al. Cryo-EM Structure of a Relaxase Reveals the Molecular Basis of DNA Unwinding during Bacterial Conjugation. Cell 169, 708–721.e12 (2017).

264. Hospenthal, M. K., Costa, T. R. D. D. & Waksman, G. A comprehensive guide to pilus biogenesis in Gram-negative bacteria. Nat. Rev. Microbiol. 15, 365–379 (2017).

265. Fronzes, R. et al. Structure of a Type IV Secretion System Core Complex. Science (80-. ). 323, 266–268 (2009).

266. Low, H. H. et al. Structure of a type IV secretion system. Nature 508, 550–553 (2014).

267. Chandran, V. et al. Structure of the outer membrane complex of a type IV secretion system. Nature 462, 1011–1015 (2009).

175 268. Ghosal, D., Chang, Y.-W., Jeong, K. C., Vogel, J. P. & Jensen, G. J. In situ structure of the Legionella Dot/Icm type IV secretion system by electron cryotomography. EMBO Rep. 18, 726–732 (2017).

269. Majdalani, N., Moore, D., Maneewannakul, S. & Ippen-Ihler, K. Role of the propilin leader peptide in the maturation of F pilin. J. Bacteriol. 178, 3748–54 (1996).

270. Maneewannakul, K., Maneewannakul, S. & Ippen-Ihler, K. Characterization of traX, the F plasmid locus required for acetylation of F-pilin subunits. J. Bacteriol. 177, 2957–64 (1995).

271. Costa, T. R. D. et al. Structure of the Bacterial Sex F Pilus Reveals an Assembly of a Stoichiometric Protein-Phospholipid Complex. Cell 166, (2016).

272. Babic, A., Lindner, A. B., Vulic, M., Stewart, E. J. & Radman, M. Direct Visualization of Horizontal Gene Transfer. Science (80-. ). 319, 1533–1536 (2008).

273. Banta, L. M. et al. An Agrobacterium VirB10 mutation conferring a type IV secretion system gating defect. J. Bacteriol. 193, 2566– 74 (2011).

274. Jakubowski, S. J. et al. Agrobacterium VirB10 domain requirements for type IV secretion and T pilus biogenesis. Mol. Microbiol. 71, 779–794 (2009).

275. Backert, S., Fronzes, R. & Waksman, G. VirB2 and VirB5 proteins: specialized adhesins in bacterial type-IV secretion systems? Trends Microbiol. 16, 409–413 (2008).

276. Rohde, M., Püls, J., Buhrdorf, R., Fischer, W. & Haas, R. A novel sheathed surface of the Helicobacter pylori cag type IV secretion system. Mol. Microbiol. 49, 219–234 (2003).

277. Kwok, T. et al. Helicobacter exploits integrin for type IV secretion and kinase activation. Nature 449, 862–866 (2007).

278. Pohlschroder, M. & Esquivel, R. N. Archaeal type IV pili and their involvement in biofilm formation. Front. Microbiol. 6, 190 (2015).

279. Melville, S. & Craig, L. Type IV Pili in Gram-Positive Bacteria. Microbiol. Mol. Biol. Rev. 77, 323–341 (2013).

280. Ottow, J. C. G. Ecology, Physiology, and Genetics of Fimbriae and Pili. Annu. Rev. Microbiol. 29, 79–108 (1975).

281. Henriksen, S. D. & Henrichsen, J. Twitching motility and possession of polar fimbriae in spreading streptococcus sanguis isolates from the human throat. Acta Pathol. Microbiol. Scand. Sect. B Microbiol. 83 B, 133–140 (1975).

282. Kaiser, D. Social gliding is correlated with the presence of pili in Myxococcus xanthus. Proc. Natl. Acad. Sci. 76, (1979).

283. Strom, M. S. & Lory, S. Structure-function and biogenesis of the type IV pili. Annu. Rev. Microbiol. 47, 565–596 (1993).

284. Dalrymple, B. & Mattick, J. S. An Analysis of the Organization and Evolution of Type 4 Fimbrial (MePhe) Subunit Proteins. J Mol Evol 25, 261–269 (1987).

285. Rakotoarivonina, H. et al. Adhesion to cellulose of the Gram-positive bacterium Ruminococcus albus involves type IV pili. Microbiology 148, 1871–1880 (2002).

286. Myers, G. S. A. et al. Skewed genomic variability in strains of the toxigenic bacterial pathogen, Clostridium perfringens. Genome Res. 16, 1031–40 (2006).

287. Varga, J. J. et al. Type IV pili-dependent gliding motility in the Gram-positive pathogen Clostridium perfringens and other Clostridia. Mol. Microbiol. 62, 680–694 (2006).

288. Zolghadr, B., Weber, S., Szabó, Z., Driessen, A. J. M. & Albers, S.-V. Identification of a system required for the functional surface localization of sugar binding proteins with class III signal peptides in Sulfolobus solfataricus. Mol. Microbiol. 64, 795–806 (2007).

289. Fröls, S. et al. UV-inducible cellular aggregation of the hyperthermophilic archaeon Sulfolobus solfataricus is mediated by pili formation. Mol. Microbiol. 70, 938–952 (2008).

290. Fröls, S. et al. Response of the hyperthermophilic archaeon Sulfolobus solfataricus to UV damage. J. Bacteriol. 189, 8708–18 (2007).

291. Mattick, J. S. Type IV Pili and Twitching Motility. Annu. Rev. Microbiol. 56, 289–314 (2002).

292. Gurung, I. et al. Functional analysis of an unusual type IV pilus in the Gram-positive Streptococcus sanguinis. Mol. Microbiol. 99, 380–392 (2016).

293. Burrows, L. L. Weapons of mass retraction. Molecular Microbiology 57, 878–888 (2005).

294. Maier, B. et al. Single pilus motor forces exceed 100 pN. Proc. Natl. Acad. Sci. U. S. A. 99, 16012–7 (2002).

295. Kirn, T. J., Bose, N. & Taylor, R. K. Secretion of a soluble colonization factor by the TCP type 4 pilus biogenesis pathway in Vibrio cholerae. Mol. Microbiol. 49, 81–92 (2003).

176 296. Giltner, C. L., Nguyen, Y. & Burrows, L. L. Type IV pilin proteins: Versatile molecular modules. Microbiol. Mol. Biol. Rev. 76, 740– 772 (2012).

297. Chen, I. & Dubnau, D. DNA uptake during bacterial transformation. Nat. Rev. Microbiol. 2, 241–249 (2004).

298. Coureuil, M. et al. Meningococcus hijacks a ??2-adrenoceptor/??-arrestin pathway to cross brain microvasculature endothelium. Cell 143, 1149–1160 (2010).

299. Coureuil, M. et al. Meningococcal type IV pili recruit the polarity complex to cross the brain endothelium. Science 325, 83–7 (2009).

300. Merz, A. J., Enns, C. A. & So, M. Type IV pili of pathogenic Neisseriae elicit cortical plaque formation in epithelial cells. Mol. Microbiol. 32, 1316–1332 (1999).

301. Persat, A. et al. Type IV pili mechanochemically regulate virulence factors in Pseudomonas aeruginosa. doi:10.1073/pnas.1502025112

302. Ellison, C. K. et al. Obstruction of pilus retraction stimulates bacterial surface sensing. Science (80-. ). 358, 535–538 (2017).

303. BRADLEY, D. E. Shortening of Pseudomonas aeruginosa Pili after RNA-Phage Adsorption. J. Gen. Microbiol. 72, 303–319 (1972).

304. Waldor, M. K. & Mekalanos, J. J. Lysogenic conversion by a filamentous phage encoding cholera toxin. Science 272, 1910–4 (1996).

305. Reguera, G. et al. Extracellular electron transfer via microbial nanowires. Nature 435, 1098–1101 (2005).

306. Pelicic, V. Type IV pili: e pluribus unum? Mol. Microbiol. 68, 827–837 (2008).

307. Kolappan, S. et al. Structure of the Neisseria meningitidis Type IV Pilus. Nat. Commun. 7, 1–12 (2016).

308. Wang, F. et al. Cryoelectron microscopy reconstruction of the Pseudomonas aeruginosa and Neisseria gonorrhoeae Type IV pili at sub-nanometer resolution. Structure 25, 1423–1435.e4 (2017).

309. Chang, Y. W. et al. Architecture of the Vibrio cholerae toxin-coregulated pilus machine revealed by electron cryotomography. Nat. Microbiol. 2, 16269 (2017).

310. Chang, Y.-W. W. et al. Architecture of the type IVa pilus machine. Science 351, aad2001 (2016).

311. Ramboarina, S. et al. Structure of the Bundle-forming Pilus from Enteropathogenic Escherichia coli. J. Biol. Chem. 280, 40252– 40260 (2005).

312. Li, J. et al. Vibrio cholerae Toxin-Coregulated Pilus Structure Analyzed by Hydrogen/Deuterium Exchange Mass Spectrometry. Structure 16, 137–148 (2008).

313. Kolappan, S., Roos, J., Yuen, A. S. W., Pierce, O. M. & Craig, L. Structural Characterization of CFA/III and Longus Type IVb Pili from Enterotoxigenic Escherichia coli. J. Bacteriol. 194, 2725–2735 (2012).

314. Gold, V. A., Salzer, R., Averhoff, B. & Kühlbrandt, W. Structure of a type IV pilus machinery in the open and closed state. Elife 4, (2015).

315. Goosens, V. J. et al. Reconstitution of a minimal machinery capable of assembling periplasmic type IV pili. Proc. Natl. Acad. Sci. 114, E4978–E4986 (2017).

316. Friedrich, C., Bulyha, I. & Søgaard-Andersen, L. Outside-in assembly pathway of the type IV pilus system in Myxococcus xanthus. J. Bacteriol. 196, 378–90 (2014).

317. Wolfgang, M., van Putten, J. P., Hayes, S. F., Dorward, D. & Koomey, M. Components and dynamics of fiber formation define a ubiquitous biogenesis pathway for bacterial pili. EMBO J. 19, 6408–6418 (2000).

318. Carbonnelle, E., Helaine, S., Nassif, X. & Pelicic, V. A systematic genetic analysis in Neisseria meningitidis defines the Pil proteins required for assembly, functionality, stabilization and export of type IV pili. Mol. Microbiol. 61, 1510–1522 (2006).

319. Balasingham, S. V. et al. Interactions between the Lipoprotein PilP and the Secretin PilQ in Neisseria meningitidis. J. Bacteriol. 189, 5716–5727 (2007).

320. Collins, R. F. et al. Structure of the Neisseria meningitidis Outer Membrane PilQ Secretin Complex at 12 Å Resolution. J. Biol. Chem. 279, 39750–39756 (2004).

321. Collins, R. F. et al. Interaction with Type IV Pili Induces Structural Changes in the Bacterial Outer Membrane Secretin PilQ. J. Biol. Chem. 280, 18923–18930 (2005).

322. Koo, J., Burrows, L. L. & Lynne Howell, P. Decoding the roles of pilotins and accessory proteins in secretin escort services. FEMS Microbiol. Lett. 328, 1–12 (2012).

323. Imhaus, A.-F. & Duménil, G. The number of Neisseria meningitidis type IV pili determines host cell interaction. EMBO J. 33, 1767–

177 83 (2014).

324. Giltner, C. L., Habash, M. & Burrows, L. L. Pseudomonas aeruginosa minor pilins are incorporated into type IV Pili. J. Mol. Biol. 398, 444–461 (2010).

325. Helaine, S., Dyer, D. H., Nassif, X., Pelicic, V. & Forest, K. T. 3D structure/function analysis of PilX reveals how minor pilins can modulate the virulence properties of type IV pili. Proc. Natl. Acad. Sci. 104, 15888–15893 (2007).

326. Nunn, D. N. & Lory, S. Product of the Pseudomonas aeruginosa gene pilD is a prepilin leader peptidase. Proc. Natl. Acad. Sci. U. S. A. 88, 3281–5 (1991).

327. Merz, A. J., So, M. & Sheetz, M. P. Pilus retraction powers bacterial twitching motility. Nature 407, 98–102 (2000).

328. Ng, D. et al. The Vibrio cholerae Minor Pilin TcpB Initiates Assembly and Retraction of the Toxin-Coregulated Pilus. PLoS Pathog. 12, e1006109 (2016).

329. Nguyen, Y. et al. Structural and Functional Studies of the Pseudomonas aeruginosa Minor Pilin, PilE. (2015). doi:10.1074/jbc.M115.683334

330. Arts, J., van Boxtel, R., Filloux, A., Tommassen, J. & Koster, M. Export of the Pseudopilin XcpT of the Pseudomonas aeruginosa Type II Secretion System via the Signal Recognition Particle-Sec Pathway. J. Bacteriol. 189, 2069–2076 (2007).

331. Francetic, O., Buddelmeijer, N., Lewenza, S., Kumamoto, C. A. & Pugsley, A. P. Signal Recognition Particle-Dependent Inner Membrane Targeting of the PulG Pseudopilin Component of a Type II Secretion System. J. Bacteriol. 189, 1783–1793 (2007).

332. Dupuy, B., Taha, M. K., Pugsley, A. P. & Marchal, C. Neisseria gonorrhoeae prepilin export studied in Escherichia coli. J. Bacteriol. 173, 7589–98 (1991).

333. Strom, M. S. & Lory, S. Mapping of export signals of Pseudomonas aeruginosa pilin with alkaline phosphatase fusions. J. Bacteriol. 169, 3181–8 (1987).

334. Balaban, M. et al. Secretion of a pneumococcal type II secretion system pilus correlates with DNA uptake during transformation. Proc. Natl. Acad. Sci. 111, E758–E765 (2014).

335. Jarrell, K., Ding, Y., Nair, D. & Siu, S. Surface Appendages of Archaea: Structure, Function, Genetics and Assembly. Life 3, 86–117 (2013).

336. Ng, S. Y. M. et al. Genetic and mass spectrometry analyses of the unusual type IV-like pili of the archaeon Methanococcus maripaludis. J. Bacteriol. 193, 804–814 (2011).

337. Hu, J., Xue, Y., Lee, S. & Ha, Y. The crystal structure of GXGD membrane protease FlaK. Nature 475, 528–531 (2011).

338. Pepe, J. C. & Lory, S. Amino acid substitutions in PilD, a bifunctional enzyme of Pseudomonas aeruginosa. Effect on leader peptidase and N-methyltransferase activities in vitro and in vivo. J. Biol. Chem. 273, 19120–9 (1998).

339. Strom, M. S. & Lory, S. Amino acid substitutions in pilin of Pseudomonas aeruginosa. Effect on leader peptide cleavage, amino- terminal methylation, and pilus assembly. J. Biol. Chem. 266, 1656–1664 (1991).

340. Nunn, D. N. & Lory, S. Cleavage, methylation, and localization of the Pseudomonas aeruginosa export proteins XcpT, -U, -V, and - W. J. Bacteriol. 175, 4375–82 (1993).

341. Ng, S. Y. M. et al. Different minimal signal peptide lengths recognized by the archaeal prepilin-like peptidases FlaK and PibD. J. Bacteriol. 191, 6732–40 (2009).

342. Santos-Moreno, J. et al. Polar N-terminal Residues Conserved in Type 2 Secretion Pseudopilins Determine Subunit Targeting and Membrane Extraction Steps during Fibre Assembly. J. Mol. Biol. 429, 1746–1765 (2017).

343. Nunn, D., Bergman, S. & Lory, S. Products of three accessory genes, pilB, pilC, and pilD, are required for biogenesis of Pseudomonas aeruginosa pili. J. Bacteriol. 172, 2911–9 (1990).

344. Takhar, H. K., Kemp, K., Kim, M., Howell, P. L. & Burrows, L. L. The Platform Protein Is Essential for Type IV Pilus Biogenesis. J. Biol. Chem. 288, 9721–9728 (2013).

345. Karuppiah, V., Collins, R. F., Thistlethwaite, A., Gao, Y. & Derrick, J. P. Structure and assembly of an inner membrane platform for initiation of type IV pilus biogenesis. Proc. Natl. Acad. Sci. 110, E4638–E4647 (2013).

346. Sandkvist, M., Bagdasarian, M., Howard, S. P. & DiRita, V. J. Interaction between the autokinase EpsE and EpsL in the cytoplasmic membrane is required for extracellular secretion in Vibrio cholerae. EMBO J. 14, 1664–73 (1995).

347. Arts, J. et al. Interaction domains in the Pseudomonas aeruginosa type II secretory apparatus component XcpS (GspF). Microbiology 153, 1582–1592 (2007).

348. Bischof, L. F., Friedrich, C., Harms, A., Søgaard-Andersen, L. & van der Does, C. The Type IV Pilus Assembly ATPase PilB of Myxococcus xanthus Interacts with the Inner Membrane Platform Protein PilC and the Nucleotide-binding Protein PilM. J. Biol.

178 Chem. 291, 6946–57 (2016).

349. Georgiadou, M., Castagnini, M., Karimova, G., Ladant, D. & Pelicic, V. Large-scale study of the interactions between proteins involved in type IV pilus biology in Neisseria meningitidis: characterization of a subcomplex involved in pilus assembly. Mol. Microbiol. 84, 857–873 (2012).

350. Johnson, T. L., Scott, M. E. & Sandkvist, M. Mapping Critical Interactive Sites within the Periplasmic Domain of the Vibrio cholerae Type II Secretion Protein EpsM. J. Bacteriol. 189, 9082–9089 (2007).

351. Possot, O. M., Vignon, G., Bomchil, N., Ebel, F. & Pugsley, A. P. Multiple interactions between pullulanase secreton components involved in stabilization and cytoplasmic membrane association of PulE. J. Bacteriol. 182, 2142–52 (2000).

352. Michel, G., Bleves, S., Ball, G., Lazdunski, A. & Filloux, A. Mutual stabilization of the XcpZ and XcpY components of the secretory apparatus in Pseudomonas aeruginosa. Microbiology 144, 3379–3386 (1998).

353. Ayers, M. et al. PilM/N/O/P Proteins Form an Inner Membrane Complex That Affects the Stability of the Pseudomonas aeruginosa Type IV Pilus Secretin. J. Mol. Biol. 394, 128–142 (2009).

354. Sampaleanu, L. M. et al. Periplasmic Domains of Pseudomonas aeruginosa PilN and PilO Form a Stable Heterodimeric Complex. J. Mol. Biol. 394, 143–159 (2009).

355. Tammam, S. et al. PilMNOPQ from the Pseudomonas aeruginosa type IV pilus system form a transenvelope protein interaction network that interacts with PilA. J. Bacteriol. 195, 2126–35 (2013).

356. Gray, M. D., Bagdasarian, M., Hol, W. G. J. & Sandkvist, M. In vivo cross-linking of EpsG to EpsL suggests a role for EpsL as an ATPase-pseudopilin coupling protein in the Type II secretion system of Vibrio cholerae. Mol. Microbiol. 79, 786–798 (2011).

357. McCallum, M. et al. PilN Binding Modulates the Structure and Binding Partners of the Pseudomonas aeruginosa Type IVa Pilus Protein PilM. J. Biol. Chem. 291, 11003–15 (2016).

358. Mancl, J. M., Black, W. P., Robinson, H., Yang, Z. & Schubot, F. D. Crystal Structure of a Type IV Pilus Assembly ATPase: Insights into the Molecular Mechanism of PilB from Thermus thermophilus. Structure 24, 1886–1897 (2016).

359. McCallum, M., Tammam, S., Khan, A., Burrows, L. L. & Howell, P. L. The molecular mechanism of the type IVa pilus motors. Nat. Commun. 8, 15091 (2017).

360. Nivaskumar, M. et al. Distinct docking and stabilization steps of the pseudopilus conformational transition path suggest rotational assembly of type IV pilus-like fibers. Structure 22, 685–696 (2014).

361. Planet, P. J., Kachlany, S. C., DeSalle, R. & Figurski, D. H. Phylogeny of genes for secretion NTPases: Identification of the widespread tadA subfamily and development of a diagnostic key for gene classification. Proc. Natl. Acad. Sci. 98, 2503–2508 (2001).

362. Peabody, C. R. Type II protein secretion and its relationship to bacterial type IV pili and archaeal flagella. Microbiology 149, 3051– 3072 (2003).

363. Ellison, C. K. et al. Obstruction of pilus retraction stimulates bacterial surface sensing. Science (80-. ). 358, 535–538 (2017).

364. Biais, N., Ladoux, B., Higashi, D., So, M. & Sheetz, M. Cooperative retraction of bundled type IV pili enables nanonewton force generation. PLoS Biol. 6, 907–913 (2008).

365. Iyer, L. M., Leipe, D. D., Koonin, E. V & Aravind, L. Evolutionary history and higher order classification of AAA+ ATPases. J. Struct. Biol. 146, 11–31 (2004).

366. Walker, J. E., Saraste, M., Runswick, M. J. & Gay, N. J. Distantly related sequences in the alpha- and beta-subunits of ATP synthase, myosin, kinases and other ATP-requiring enzymes and a common nucleotide binding fold. EMBO J. 1, 945–51 (1982).

367. Yamagata, A. & Tainer, J. A. Hexameric structures of the archaeal secretion ATPase GspE and implications for a universal secretion mechanism. EMBO J. 26, 878–90 (2007).

368. Misic, A. M., Satyshur, K. A. & Forest, K. T. P. aeruginosa PilT Structures with and without Nucleotide Reveal a Dynamic Type IV Pilus Retraction Motor. J. Mol. Biol. 400, 1011–1021 (2010).

369. Lu, C. et al. Hexamers of the Type II Secretion ATPase GspE from Vibrio cholerae with Increased ATPase Activity. Structure 21, 1707–1717 (2013).

370. Satyshur, K. A. et al. Crystal Structures of the Pilus Retraction Motor PilT Suggest Large Domain Movements and Subunit Cooperation Drive Motility. Structure 15, 363–376 (2007).

371. Park, H.-S. M., Wolfgang, M. & Koomey, M. Modification of type IV pilus-associated epithelial cell adherence and multicellular behavior by the PilU protein of Neisseria gonorrhoeae. Infect. Immun. 70, 3891–903 (2002).

372. Whitchurch, C. B. & Mattick, J. S. Characterization of a gene, pilU, required for twitching motility but not phage sensitivity in Pseudomonas aeruginosa. Mol. Microbiol. 13, 1079–1091 (1994).

179 373. Brown, D. R., Helaine, S., Carbonnelle, E. & Pelicic, V. Systematic functional analysis reveals that a set of seven genes is involved in fine-tuning of the multiple functions mediated by type IV pili in Neisseria meningitidis. Infect. Immun. 78, 3053–3063 (2010).

374. Chiang, P. & Burrows, L. L. Biofilm formation by hyperpiliated mutants of Pseudomonas aeruginosa. J. Bacteriol. 185, 2374–2378 (2003).

375. Eriksson, J., Eriksson, O. S. & Jonsson, A.-B. Loss of meningococcal PilU delays microcolony formation and attenuates virulence in vivo. Infect. Immun. 80, 2538–47 (2012).

376. Chiang, P., Habash, M. & Burrows, L. L. Disparate subcellular localization patterns of Pseudomonas aeruginosa Type IV pilus ATPases involved in twitching motility. J. Bacteriol. 187, 829–39 (2005).

377. Kurre, R., Höne, A., Clausen, M., Meel, C. & Maier, B. PilT2 enhances the speed of gonococcal type IV pilus retraction and of twitching motility. Mol. Microbiol. 86, 857–865 (2012).

378. Guilvout, I., Chami, M., Engel, A., Pugsley, A. P. & Bayan, N. Bacterial outer membrane secretin PulD assembles and inserts into the inner membrane in the absence of its pilotin. EMBO J. 25, 5241–5249 (2006).

379. Siewering, K. et al. Peptidoglycan-binding protein TsaP functions in surface assembly of type IV pili. Proc. Natl. Acad. Sci. U. S. A. 111, E953-61 (2014).

380. Bose, N. & Taylor, R. K. Identification of a TcpC-TcpQ Outer Membrane Complex Involved in the Biogenesis of the Toxin- Coregulated Pilus of Vibrio cholerae. J. Bacteriol. 187, 2225–2232 (2005).

381. Nouwen, N. et al. Secretin PulD: association with pilot PulS, structure, and ion-conducting channel formation. Proc. Natl. Acad. Sci. U. S. A. 96, 8173–7 (1999).

382. Nouwen, N., Stahlberg, H., Pugsley, A. P. & Engel, A. Domain structure of secretin PulD revealed by limited proteolysis and electron microscopy. EMBO J. 19, 2229–36 (2000).

383. Collins, R. F. et al. Three-dimensional structure of the Neisseria meningitidis secretin PilQ determined from negative-stain transmission electron microscopy. J. Bacteriol. 185, 2611–7 (2003).

384. Schmidt, S. A. et al. Structure-function analysis of BfpB, a secretin-like protein encoded by the bundle-forming-pilus operon of enteropathogenic Escherichia coli. J. Bacteriol. 183, 4848–4859 (2001).

385. Berry, J.-L. et al. Structure and Assembly of a Trans-Periplasmic Channel for Type IV Pili in Neisseria meningitidis. PLoS Pathog. 8, e1002923 (2012).

386. Collins, R. F., Davidsen, L., Derrick, J. P., Ford, R. C. & Tønjum, T. Analysis of the PilQ secretin from Neisseria meningitidis by transmission electron microscopy reveals a dodecameric quaternary structure. J. Bacteriol. 183, 3825–32 (2001).

387. D’Imprima, E. et al. Cryo-EM structure of the bifunctional secretin complex of Thermus thermophilus. Elife 6, e30483 (2017).

388. Koo, J., Lamers, R. P., Rubinstein, J. L., Burrows, L. L. & Howell, P. L. Structure of the Pseudomonas aeruginosa Type IVa Pilus Secretin at 7.4 Å. Structure 24, 1778–1787 (2016).

389. Yan, Z., Yin, M., Xu, D., Zhu, Y. & Li, X. Structural insights into the secretin translocation channel in the type II secretion system. Nat. Struct. Mol. Biol. 24, 177–183 (2017).

390. Newhall, W. J., Wilde, C. E., Sawyer, W. D. & Haak, R. A. High-molecular-weight antigenic protein complex in the outer membrane of Neisseria gonorrhoeae. Infect. Immun. 27, 475–82 (1980).

391. Piepenbrink, K. H. et al. Structural and Evolutionary Analyses Show Unique Stabilization Strategies in the Type IV Pili of Clostridium difficile. Structure 23, 385–396 (2015).

392. Li, J., Egelman, E. H. & Craig, L. Structure of the Vibrio cholerae Type IVb Pilus and Stability Comparison with the Neisseria gonorrhoeae Type IVa Pilus. J. Mol. Biol. 418, 47–64 (2012).

393. Braun, T. et al. Archaeal flagellin combines a bacterial type IV pilin domain with an Ig-like domain. Proc. Natl. Acad. Sci. U. S. A. 113, 10352–7 (2016).

394. Gerl, L., Deutzmann, R. & Sumper, M. Halobacterial flagellins are encoded by a multigene family Identification of all five gene products. FEBS Lett. 244, 137–140 (1989).

395. Long, C. D. et al. Modulation of gonococcal piliation by regulatable transcription of pilE. J. Bacteriol. 183, 1600–9 (2001).

396. Craig, L. et al. Type IV pilin structure and assembly: X-ray and EM analyses of Vibrio cholerae toxin-coregulated pilus and Pseudomonas aeruginosa PAK pilin. Mol. Cell 11, 1139–1150 (2003).

397. Craig, L. et al. Type IV Pilus Structure by Cryo-Electron Microscopy and Crystallography: Implications for Pilus Assembly and Functions. Mol. Cell 23, 651–662 (2006).

398. Parge, H. E. et al. Structure of the fibre-forming protein pilin at 2.6 Å resolution. Nature 378, 32–38 (1995).

180 399. Nguyen, Y., Jackson, S. G., Aidoo, F., Junop, M. & Burrows, L. L. Structural Characterization of Novel Pseudomonas aeruginosa Type IV Pilins. J. Mol. Biol. 395, 491–503 (2010).

400. Berry, J.-L. et al. A Comparative Structure/Function Analysis of Two Type IV Pilin DNA Receptors Defines a Novel Mode of DNA Binding. Structure 24, 926–934 (2016).

401. Gorgel, M. et al. High-resolution structure of a type IV pilin from the metal-reducing bacterium Shewanella oneidensis. BMC Struct. Biol. 15, 4 (2015).

402. Piepenbrink, K. H. et al. Structure of Clostridium difficile PilJ exhibits unprecedented divergence from known type IV pilins. J. Biol. Chem. 289, 4334–45 (2014).

403. Hartung, S. et al. Ultrahigh resolution and full-length pilin structures with insights for filament assembly, pathogenic functions, and potential. J. Biol. Chem. 286, 44254–65 (2011).

404. Gerald F. Audette, Randall T. Irvin, and & Hazes*, B. Crystallographic Analysis of the Pseudomonas aeruginosa Strain K122-4 Monomeric Pilin Reveals a Conserved Receptor-Binding Architecture†,‡. (2004). doi:10.1021/BI048957S

405. Keizer, D. W. et al. Structure of a pilin monomer from Pseudomonas aeruginosa: implications for the assembly of pili. J. Biol. Chem. 276, 24186–93 (2001).

406. Hazes, B., Sastry, P. A., Hayakawa, K., Read, R. J. & Irvin, R. T. Crystal structure of Pseudomonas aeruginosa PAK pilin suggests a main-chain-dominated mode of receptor binding. J. Mol. Biol. 299, 1005–1017 (2000).

407. Xu, X.-F. et al. NMR structure of a type IVb pilin from Salmonella typhi and its assembly into pilus. J. Biol. Chem. 279, 31599–605 (2004).

408. Reardon, P. N. & Mueller, K. T. Structure of the type IVa major pilin from the electrically conductive bacterial nanowires of Geobacter sulfurreducens. J. Biol. Chem. 288, 29260–6 (2013).

409. Korotkov, K. V. et al. Calcium is essential for the major pseudopilin in the type 2 secretion system. J. Biol. Chem. 284, 25466– 25470 (2009).

410. Piepenbrink, K. et al. Structural and Evolutionary Analyses Show Unique Stabilization Strategies in the Type IV Pili of Clostridium difficile. Structure 3, 385–396 (2014).

411. Piepenbrink, K. H. et al. Structural and Evolutionary Analyses Show Unique Stabilization Strategies in the Type IV Pili of Clostridium difficile. Structure 23, 385–396 (2015).

412. Stimson, E. et al. Discovery of a novel protein modification: alpha-glycerophosphate is a substituent of meningococcal pilin. Biochem. J. 316 ( Pt 1), 29–33 (1996).

413. Gault, J. et al. Neisseria meningitidis Type IV Pili Composed of Sequence Invariable Pilins Are Masked by Multisite Glycosylation. PLoS Pathog. 11, 1–24 (2015).

414. Chaban, B., Voisin, S., Kelly, J., Logan, S. M. & Jarrell, K. F. Identification of genes involved in the biosynthesis and attachment of Methanococcus voltae N -linked glycans: insight into N -linked glycosylation pathways in Archaea. Mol. Microbiol. 61, 259–268 (2006).

415. VanDyke, D. J. et al. Identification of a putative acetyltransferase gene, MMP0350, which affects proper assembly of both flagella and pili in the archaeon Methanococcus maripaludis. J. Bacteriol. 190, 5300–7 (2008).

416. Virji, M. et al. Pilus-facilitated adherence of Neisseria meningitidis to human epithelial and endothelial cells: modulation of adherence phenotype occurs concurrently with changes in primary amino acid sequence and the glycosylation status of pilin. Mol. Microbiol. 10, 1013–1028 (1993).

417. Jen, F. E. C. et al. Dual Pili Post-translational Modifications Synergize to Mediate Meningococcal Adherence to Platelet Activating Factor Receptor on Human Airway Cells. PLoS Pathog. 9, e1003377 (2013).

418. Chamot-Rooke, J. et al. Posttranslational modification of pili upon cell contact triggers N. meningitidis dissemination. Science 331, 778–82 (2011).

419. Harvey, H. et al. Pseudomonas aeruginosa defends against phages through type IV pilus glycosylation. Nat. Microbiol. 3, 47–52 (2018).

420. Winther-Larsen, H. C. et al. A conserved set of pilin-like molecules controls type IV pilus dynamics and organelle-associated functions in Neisseria gonorrhoeae. Mol. Microbiol. 56, 903–917 (2005).

421. Lu, H.-M., Motley, S. T. & Lory, S. Interactions of the components of the general secretion pathway: role of Pseudomonas aeruginosa type IV pilin subunits in complex formation and extracellular protein secretion. Mol. Microbiol. 25, 247–259 (1997).

422. Cisneros, D. A., Bond, P. J., Pugsley, A. P., Campos, M. & Francetic, O. Minor pseudopilin self-assembly primes type II secretion pseudopilus elongation. EMBO J. 31, 1041–1053 (2012).

181 423. Korotkov, K. V & Hol, W. G. J. Structure of the GspK-GspI-GspJ complex from the enterotoxigenic Escherichia coli type 2 secretion system. Nat. Struct. Mol. Biol. 15, 462–468 (2008).

424. Yanez, M. E., Korotkov, K. V., Abendroth, J. & Hol, W. G. J. The Crystal Structure of a Binary Complex of two Pseudopilins: EpsI and EpsJ from the Type 2 Secretion System of Vibrio vulnificus. J. Mol. Biol. 375, 471–486 (2008).

425. Nguyen, Y. et al. Pseudomonas aeruginosa minor pilins prime type IVa pilus assembly and promote surface display of the PilY1 adhesin. J. Biol. Chem. 290, 601–611 (2015).

426. Douzi, B. et al. The XcpV/GspI pseudopilin has a central role in the assembly of a quaternary complex within the T2SS pseudopilus. J. Biol. Chem. 284, 34580–9 (2009).

427. Wirth, R. Colonization of Black Smokers by Hyperthermophilic Microorganisms. Trends in Microbiology 25, 92–99 (2017).

428. Skerker, J. M. & Berg, H. C. Direct observation of extension and retraction of type IV pili. Proc. Natl. Acad. Sci. U. S. A. 98, 6901–4 (2001).

429. Lee, K. K. et al. The binding of Pseudomonas aeruginosa pili to glycosphingolipids is a tip-associated event involving the C- terminal region of the structural pilin subunit. Mol. Microbiol. 11, 705–713 (1994).

430. Giltner, C. L. et al. The Pseudomonas aeruginosa type IV pilin receptor binding domain functions as an adhesin for both biotic and abiotic surfaces. Mol. Microbiol. 59, 1083–1096 (2006).

431. Mikaty, G. et al. Extracellular Bacterial Pathogen Induces Host Cell Surface Reorganization to Resist Shear Stress. PLoS Pathog. 5, e1000314 (2009).

432. Lu, S. et al. Nanoscale Pulling of Type IV Pili Reveals Their Flexibility and Adhesion to Surfaces over Extended Lengths of the Pili. Biophysj 108, 2865–2875 (2015).

433. Hyland, R. M. et al. The bundlin pilin protein of enteropathogenic Escherichia coli is an N-acetyllactosamine-specific lectin. Cell. Microbiol. 10, 177–187 (2008).

434. Esquivel, R. N., Xu, R. & Pohlschroder, M. Novel archaeal adhesion pilins with a conserved N terminus. J. Bacteriol. 195, 3808–18 (2013).

435. Jin, F., Conrad, J. C., Gibiansky, M. L. & Wong, G. C. L. Bacteria use type-IV pili to slingshot on surfaces. Proc. Natl. Acad. Sci. 108, 12617–12622 (2011).

436. Marathe, R. et al. Bacterial twitching motility is coordinated by a two-dimensional tug-of-war with directional memory. Nat. Commun. 5, 3759 (2014).

437. Maier, B. & Wong, G. C. L. How Bacteria Use Type IV Pili Machinery on Surfaces. Trends Microbiol. 23, 775–788 (2015).

438. Bhaya, D., Takahashi, A. & Grossman, A. R. Light regulation of type IV pilus-dependent motility by chemosensor-like elements in Synechocystis PCC6803. Proc. Natl. Acad. Sci. U. S. A. 98, 7540–5 (2001).

439. Shahapure, R., Driessen, R. P. C., Haurat, M. F., Albers, S. V. & Dame, R. T. The archaellum: A rotating type IV pilus. Mol. Microbiol. 91, 716–723 (2014).

440. Schlesner, M. et al. Identification of Archaea-specific chemotaxis proteins which interact with the flagellar apparatus. BMC Microbiol. 9, 56 (2009).

441. Quax, T. E. F. et al. Structure and function of the archaeal response regulator CheY. Proc. Natl. Acad. Sci. 115, 201716661 (2018).

442. Taylor, R. K., Millert, V. L., Furlong, D. B. & Mekalanost, J. J. Use of phoA gene fusions to identify a pilus colonization factor coordinately regulated with cholera toxin. Genetics 84, 2833–2837 (1987).

443. Kirn, T. J., Lafferty, M. J., Sandoe, C. M. P. & Taylor, R. K. Delineation of pilin domains required for bacterial association into microcolonies and intestinal colonization by Vibrio cholerae. Mol. Microbiol. 35, 896–910 (2000).

444. Marceau, M., Beretti, J.-L. L. & Nassif, X. High adhesiveness of encapsulated Neisseria meningitidis to epithelial cells is associated with the formation of bundles of pili. Mol. Microbiol. 17, 855–863 (1995).

445. Jude, B. A. & Taylor, R. K. The physical basis of type 4 pilus-mediated microcolony formation by Vibrio cholerae O1. J. Struct. Biol. 175, 1–9 (2011).

446. Park, H.-S. M. et al. Structural alterations in a type IV pilus subunit protein result in concurrent defects in multicellular behaviour and adherence to host tissue. Mol. Microbiol. 42, 293–307 (2001).

447. Higashi, D. L. et al. N. elongata produces type IV pili that mediate interspecies gene transfer with N. gonorrhoeae. PLoS One 6, e21373 (2011).

448. Weiner, A., Schopf, S., Wanner, G., Probst, A. & Wirth, R. Positive, Neutral and Negative Interactions in Cocultures between Pyrococcus furiosus and Different Methanogenic Archaea. Microbiol. Insights 5, MBI.S8516 (2012).

182 449. Chiang, S. L., Taylor, R. K., Koomey, M. & Mekalanos, J. J. Single amino acid substitutions in the N-terminus of Vibrio cholerae TcpA affect colonization, autoagglutination, and serum resistance. Mol. Microbiol. 17, 1133–1142 (1995).

450. Yoshida, T. et al. Purification and characterization of thin pili of IncI1 plasmids ColIb-P9 and R64: formation of PilV-specific cell aggregates by type IV pili. J. Bacteriol. 180, 2842–8 (1998).

451. Ajon, M. et al. UV-inducible DNA exchange in hyperthermophilic archaea mediated by type IV pili. Mol. Microbiol. 82, 807–817 (2011).

452. O’Toole, G. A. & Kolter, R. Flagellar and twitching motility are necessary for Pseudomonas aeruginosa biofilm development. Mol. Microbiol. 30, 295–304 (1998).

453. Shevchik, V. E., Robert-Baudouy, J. & Condemine, G. Specific interaction between OutD, an Erwinia chrysanthemi outer membrane protein of the general secretory pathway, and secreted proteins. EMBO J. 16, 3007–3016 (1997).

454. Kennan, R. M., Dhungyel, O. P., Whittington, R. J., Egerton, J. R. & Rood, J. I. The Type IV Fimbrial Subunit Gene (fimA) of Dichelobacter nodosus Is Essential for Virulence, Protease Secretion, and Natural Competence. J. Bacteriol. 183, 4451–4458 (2001).

455. Hager, A. J. et al. Type IV pili-mediated secretion modulates Francisella virulence. Mol. Microbiol. 62, 227–237 (2006).

456. Kirn, T. J., Bose, N. & Taylor, R. K. Secretion of a soluble colonization factor by the TCP type 4 pilus biogenesis pathway in Vibrio cholerae. Mol. Microbiol. 49, 81–92 (2003).

457. Lorenz, M. G. & Wackernagel, W. Bacterial gene transfer by natural genetic transformation in the environment. Microbiol. Rev. 58, 563–602 (1994).

458. Hamilton, H. L. & Dillard, J. P. Natural transformation of Neisseria gonorrhoeae : from DNA donation to . Mol. Microbiol. 59, 376–385 (2006).

459. Schaik, E. J. Van et al. DNA Binding : a Novel Function of Pseudomonas aeruginosa Type IV Pili DNA Binding : a Novel Function of Pseudomonas aeruginosa Type IV Pili. J. Bacteriol. 187, 1455–1464 (2005).

460. Chen, I. & Dubnau, D. DNA uptake during bacterial transformation. Nat. Rev. Microbiol. 2, 241–249 (2004).

461. Laurenceau, R. et al. A Type IV Pilus Mediates DNA Binding during Natural Transformation in Streptococcus pneumoniae. PLoS Pathog. 9, e1003473 (2013).

462. Muschiol, S., Balaban, M., Normark, S. & Henriques-Normark, B. Uptake of extracellular DNA: competence induced pili in natural transformation of Streptococcus pneumoniae. BioEssays 37, (2015).

463. Skerker, J. M. & Shapiro, L. Identification and cell cycle control of a novel pilus system in Caulobacter crescentus. EMBO J. 19, 3223–3234 (2000).

464. Bille, E. et al. A virulence-associated filamentous bacteriophage of Neisseria meningitidis increases host-cell colonisation. PLoS Pathog. 13, e1006495 (2017).

465. Karaolis, D. K. R., Somara, S., Maneval, D. R., Johnson, J. A. & Kaper, J. B. A bacteriophage encoding a pathogenicity island, a type-IV pilus and a phage receptor in cholera bacteria. Nature 399, 375–379 (1999).

466. Bradley, D. E. & Pitt, T. L. Pilus-dependence of Four Pseudomonas aeruginosa Bacteriophages with Non-contractile Tails. J. Gen. Virol. 24, 1–15 (1974).

467. Bradley, D. E. A function of Pseudomonas aeruginosa PAO polar pili: twitching motility. Can. J. Microbiol. 26, 146–54 (1980).

468. Quemin, E. R. J. et al. First insights into the entry process of hyperthermophilic archaeal viruses. J. Virol. 87, 13379–85 (2013).

469. Jaubert, C. et al. Genomics and genetics of Sulfolobus islandicus LAL14/1, a model hyperthermophilic archaeon. Open Biol. 3, 130010 (2013).

470. MYERS, C. R. & NEALSON, K. H. Bacterial Manganese Reduction and Growth with Manganese Oxide as the Sole Electron Acceptor. Science (80-. ). 240, 1319–1321 (1988).

471. Lovley, D. R., Stolz, J. F., Nord, G. L. & Phillips, E. J. P. Anaerobic production of magnetite by a dissimilatory iron-reducing microorganism. Nature 330, 252–254 (1987).

472. Gorby, Y. A. et al. Electrically conductive bacterial nanowires produced by Shewanella oneidensis strain MR-1 and other microorganisms. Proc. Natl. Acad. Sci. U. S. A. 103, 11358–63 (2006).

473. Sure, S., Ackland, M. L., Torriero, A. A. J., Adholeya, A. & Kochar, M. Microbial nanowires: an electrifying tale. Microbiology 162, 2017–2028 (2016).

474. Pirbadian, S. et al. Shewanella oneidensis MR-1 nanowires are outer membrane and periplasmic extensions of the extracellular electron transport components. Proc. Natl. Acad. Sci. U. S. A. 111, 12883–8 (2014).

183 475. Sure, S. et al. Inquisition of Microcystis aeruginosa and Synechocystis nanowires: characterization and modelling. 108, 1213–1225 (2015).

476. Malvankar, N. S. et al. Tunable metallic-like conductivity in microbial nanowire networks. Nat. Nanotechnol. 6, 573–579 (2011).

477. Malvankar, N. S., Yalcin, S. E., Tuominen, M. T. & Lovley, D. R. Visualization of charge propagation along individual pili proteins using ambient electrostatic force microscopy. Nat. Nanotechnol. 9, 1012–7 (2014).

478. Vargas, M. et al. Aromatic amino acids required for pili conductivity and long-range extracellular electron transport in Geobacter sulfurreducens. MBio 4, e00105-13 (2013).

479. Xiao, K. et al. Low Energy Atomic Models Suggesting a Pilus Structure that could Account for Electrical Conductivity of Geobacter sulfurreducens Pili. Sci. Rep. 6, 23385 (2016).

480. Feliciano, G. T., Steidl, R. J. & Reguera, G. Structural and functional insights into the conductive pili of Geobacter sulfurreducens revealed in molecular dynamics simulations. Phys. Chem. Chem. Phys. J. Name 0, 1–3 (2015).

481. Zobell, C. E. & Allen, E. C. The Significance of Marine Bacteria in the Fouling of Submerged Surfaces. J. Bacteriol. 29, 239–51 (1935).

482. O’Toole, G. A. & Wong, G. C. Sensational biofilms: Surface sensing in bacteria. Curr. Opin. Microbiol. 30, 139–146 (2016).

483. Siryaporn, A., Kuchma, S. L., O’Toole, G. A. & Gitai, Z. Surface attachment induces Pseudomonas aeruginosa virulence. Proc. Natl. Acad. Sci. U. S. A. 111, 16860–5 (2014).

484. Persat, A., Inclan, Y. F., Engel, J. N., Stone, H. A. & Gitai, Z. Type IV pili mechanochemically regulate virulence factors in Pseudomonas aeruginosa. Proc. Natl. Acad. Sci. U. S. A. 112, 7563–8 (2015).

485. Luo, Y. et al. A hierarchical cascade of second messengers regulates Pseudomonas aeruginosa surface behaviors. MBio 6, e02456- 14 (2015).

486. Biais, N., Higashi, D. L., Brujic, J., So, M. & Sheetz, M. P. Force-dependent polymorphism in type IV pili reveals hidden epitopes. Proc. Natl. Acad. Sci. U. S. A. 107, 11358–63 (2010).

487. Beaussart, A. et al. Nanoscale adhesion forces of Pseudomonas aeruginosa type IV pili. ACS Nano 8, 10723–10733 (2014).

488. Melican, K. et al. Adhesion of Neisseria meningitidis to Dermal Vessels Leads to Local Vascular Damage and Purpura in a Humanized Mouse Model. PLoS Pathog. 9, e1003139 (2013).

489. Join-Lambert, O. et al. Meningococcal Interaction to Microvasculature Triggers the Tissular Lesions of Purpura Fulminans. J. Infect. Dis. 208, 1590–1597 (2013).

490. Liu, G., Tang, C. M., Exley Correspondence, R. M. & William, S. Non-pathogenic Neisseria: members of an abundant, multi-habitat, diverse genus. Microbiology 161, 1297–1312 (2015).

491. Wi, T. et al. in Neisseria gonorrhoeae: Global surveillance and a call for international collaborative action. PLOS Med. 14, e1002344 (2017).

492. Stephens, D. S., Hoffman, L. H. & McGee, Z. A. Interaction of Neisseria meningitidis with Human Nasopharyngeal Mucosa: Attachment and Entry into Columnar Epithelial Cells. J. Infect. Dis. 148, 369–376 (1983).

493. Caugant, D. A. & Maiden, M. C. J. Meningococcal carriage and disease—Population biology and evolution. Vaccine 27, B64–B70 (2009).

494. Christensen, H., May, M., Bowen, L., Hickman, M. & Trotter, C. L. Meningococcal carriage by age: A systematic review and meta- analysis. Lancet Infect. Dis. 10, 853–861 (2010).

495. Mossong, J. et al. Social Contacts and Mixing Patterns Relevant to the Spread of Infectious Diseases. PLOS Med. 5, e74 (2008).

496. Jafri, R. Z. et al. Global epidemiology of invasive meningococcal disease. Population Health Metrics 11, 17 (2013).

497. Pelton, S. I. The Global Evolution of Meningococcal Epidemiology Following the Introduction of Meningococcal Vaccines. Journal of Adolescent Health 59, S3–S11 (2016).

498. Harrison, O. B. et al. Description and nomenclature of Neisseria meningitidis capsule locus. Emerg. Infect. Dis. 19, 566–73 (2013).

499. Peltola, H. Meningococcal Disease: Still with Us. Clin. Infect. Dis. 5, 71–91 (1983).

500. Greenwood, B. Meningococcal meningitis in Africa. Trans. R. Soc. Trop. Med. Hyg. 93, 341–353 (1999).

501. Gagneux, S. P. et al. Prospective Study of a Serogroup X Neisseria meningitidis Outbreak in Northern Ghana. J. Infect. Dis. 185, 618–626 (2002).

502. Caugant, D. A. Genetics and evolution of Neisseria meningitidis: Importance for the epidemiology of meningococcal disease. Infect. Genet. Evol. 8, 558–565 (2008).

184 503. Stephens, D. S., Greenwood, B. & Brandtzaeg, P. Epidemic meningitis, meningococcaemia, and Neisseria meningitidis. Lancet 369, 2196–2210 (2007).

504. Finne, J., Leinonen, M. & Mäkelä, P. H. Antigenic similarities between brain components and bacteria causing meningitis. Implications for vaccine development and pathogenesis. Lancet (London, England) 2, 355–7 (1983).

505. Watson, P. S. & Turner, D. P. J. Clinical experience with the meningococcal B vaccine, Bexsero ® : Prospects for reducing the burden of meningococcal serogroup B disease. Vaccine 34, 875–880 (2016).

506. Flacco, M. E. et al. Immunogenicity and safety of the multicomponent meningococcal B vaccine (4CMenB) in children and adolescents: a systematic review and meta-analysis. Lancet Infect. Dis. (2018). doi:10.1016/S1473-3099(18)30048-3

507. Sharip, A. et al. Population-Based Analysis of Meningococcal Disease Mortality in the United States. Pediatr. Infect. Dis. J. 25, 191– 194 (2006).

508. Jacobs, J. H. et al. The Association of Meningococcal Disease with Influenza in the United States, 1989–2009. PLoS One 9, e107486 (2014).

509. Palmgren, H. Meningococcal disease and climate. Glob. Health Action 2, (2009).

510. Densen, P. Interaction of Complement with Neisseria meningitidis and Neisseria gonorrhoeae. Clin. Microbiol. Rev. 2, 11–17 (1989).

511. Figueroa, J., Andreoni, J. & Densen, P. Complement deficiency states and meningococcal disease. Immunol. Res. 12, 295–311 (1993).

512. Bingen, E. et al. Bacterial Meningitis in Children: A French Prospective Study. Clin. Infect. Dis. 41, 1059–1063 (2005).

513. Coureuil, M., Lécuyer, H., Bourdoulous, S. & Nassif, X. A journey into the brain: Insight into how bacterial pathogens cross blood- brain barriers. Nature Reviews Microbiology 15, 149–159 (2017).

514. van Deuren, M., Brandtzaeg, P. & van der Meer, J. W. Update on meningococcal disease with emphasis on pathogenesis and clinical management. Clin. Microbiol. Rev. 13, 144–66, table of contents (2000).

515. Mairey, E. et al. Cerebral microcirculation shear stress levels determine Neisseria meningitidis attachment sites along the blood- brain barrier. J. Exp. Med. 203, 1939–50 (2006).

516. Kirsch, E. A., Barton, R. P., Kitchen, L. & Giroir, B. P. Pathophysiology, treatment and outcome of meningococcemia: a review and recent experience. Pediatr. Infect. Dis. J. 15, 967–78 (1996).

517. Snyder, L. A. & Saunders, N. J. The majority of genes in the pathogenic Neisseria species are present in non-pathogenic Neisseria lactamica , including those designated as ‘virulence genes’. BMC Genomics 7, 128 (2006).

518. Hotopp, J. C. D. et al. Comparative genomics of Neisseria meningitidis: core genome, islands of horizontal transfer and pathogen- specific genes. Microbiology 152, 3733–3749 (2006).

519. Maiden, M. C. J. et al. Multilocus sequence typing: A portable approach to the identification of clones within populations of pathogenic microorganisms. Proc. Natl. Acad. Sci. 95, 3140–3145 (1998).

520. Harrison, O. B. et al. Neisseria genomics: current status and future perspectives. Pathog. Dis. 75, (2017).

521. Rouphael, N. G. & Stephens, D. S. Neisseria meningitidis: Biology, microbiology, and epidemiology. Methods in Molecular Biology 799, 1–20 (2012).

522. Bille, E. et al. A chromosomally integrated bacteriophage in invasive meningococci. J. Exp. Med. 201, 1905–1913 (2005).

523. Meyer, J. et al. Characterization of MDAФ, A temperate filamentous bacteriophage of Neisseria meningitidis. Microbiol. (United Kingdom) 162, 268–282 (2016).

524. Claus, H., Maag, R., Maiden, M. C. J., Frosch, M. & Vogel, U. Many carried meningococci lack the genes required for capsule synthesis and transport. Microbiology 148, 1813–1819 (2002).

525. Lâm, T.-T., Claus, H., Frosch, M. & Vogel, U. Sequence analysis of serotype-specific synthesis regions II of Haemophilus influenzae serotypes c and d: evidence for common ancestry of capsule synthesis in Pasteurellaceae and Neisseria meningitidis. Res. Microbiol. 162, 483–487 (2011).

526. Schoen, C. et al. Whole-genome comparison of disease and carriage strains provides insights into virulence evolution in Neisseria meningitidis. Proc. Natl. Acad. Sci. U. S. A. 105, 3473–8 (2008).

527. Kahler, C. M. et al. The (alpha2-->8)-linked polysialic acid capsule and lipooligosaccharide structure both contribute to the ability of serogroup B Neisseria meningitidis to resist the bactericidal activity of normal human serum. Infect. Immun. 66, 5939–47 (1998).

528. Estabrook, M. M., Griffiss, J. M. & Jarvis, G. A. Sialylation of Neisseria meningitidis lipooligosaccharide inhibits serum bactericidal

185 activity by masking lacto-N-neotetraose. Infect. Immun. 65, 4436–44 (1997).

529. Uria, M. J. et al. A generic mechanism in Neisseria meningitidis for enhanced resistance against bactericidal antibodies. J. Exp. Med. 205, 1423–1434 (2008).

530. Frosch, M., Weisgerber, C. & Meyer, T. F. Molecular characterization and expression in Escherichia coli of the gene complex encoding the polysaccharide capsule of Neisseria meningitidis group B. Proc. Natl. Acad. Sci. U. S. A. 86, 1669–73 (1989).

531. Swartley, J. S. et al. Capsule switching of Neisseria meningitidis. Proc. Natl. Acad. Sci. U. S. A. 94, 271–6 (1997).

532. Aguilera, J.-F., Perrocheau, A., Meffre, C., Hahné, S. & W135 Working Group. Outbreak of serogroup W135 meningococcal disease after the Hajj pilgrimage, Europe, 2000. Emerg. Infect. Dis. 8, 761–7 (2002).

533. Raghunathan, P. L. et al. Predictors of Immunity after a Major Serogroup W-135 Meningococcal Disease Epidemic, Burkina Faso, 2002. J. Infect. Dis. 193, 607–616 (2006).

534. Beddek, A. J., Li, M.-S., Kroll, J. S., Jordan, T. W. & Martin, D. R. Evidence for capsule switching between carried and disease- causing Neisseria meningitidis strains. Infect. Immun. 77, 2989–94 (2009).

535. Zughaier, S., Steeghs, L., van der Ley, P. & Stephens, D. S. TLR4-dependent adjuvant activity of Neisseria meningitidis lipid A. Vaccine 25, 4401–4409 (2007).

536. Mandrell, R. E. & Zollinger, W. D. Lipopolysaccharide serotyping of Neisseria meningitidis by hemagglutination inhibition. Infect. Immun. 16, 471–5 (1977).

537. Scholten, R. J. P. M. et al. Lipo-oligosaccharide immunotyping of Neisseria meningitidis by a whole-cell ELISA with monoclonal antibodies. J. Med. Microbiol. 41, 236–243 (1994).

538. Jones, D. M. et al. The lipooligosaccharide immunotype as a virulence determinant in Neisseria meningitidis. Microb. Pathog. 13, 219–224 (1992).

539. Mackinnon, F. G. et al. Demonstration of lipooligosaccharide immunotype and capsule as virulence factors for Neisseria meningitidis using an infant mouse intranasal infection model. Microb. Pathog. 15, 359–366 (1993).

540. Ulevitch, R. J. & Tobias, P. S. Recognition of Gram-negative bacteria and endotoxin by the innate immune system. Curr. Opin. Immunol. 11, 19–22 (1999).

541. Zughaier, S. M. et al. Physicochemical characterization and biological activity of lipooligosaccharides and lipid A from Neisseria meningitidis. J. Endotoxin Res. 13, 343–357 (2007).

542. Zughaier, S. M. et al. Neisseria meningitidis lipooligosaccharide structure-dependent activation of the macrophage CD14/Toll- like receptor 4 pathway. Infect. Immun. 72, 371–80 (2004).

543. Braun, J. M. et al. Proinflammatory Responses to Lipo-oligosaccharide of Neisseria meningitidis Immunotype Strains in Relation to Virulence and Disease. J. Infect. Dis. 185, 1431–1438 (2002).

544. Deghmane, A.-E. et al. Differential Modulation of TNF-α–Induced Apoptosis by Neisseria meningitidis. PLoS Pathog. 5, e1000405 (2009).

545. Brandtzaeg, P. et al. Plasma endotoxin as a predictor of multiple organ failure and death in systemic meningococcal disease. J. Infect. Dis. 159, 195–204 (1989).

546. Brandtzaeg, P. et al. Meningococcal endotoxin in lethal septic shock plasma studied by gas chromatography, mass-spectrometry, ultracentrifugation, and electron microscopy. J. Clin. Invest. 89, 816–823 (1992).

547. Capel, E. et al. Peripheral blood vessels are a niche for blood-borne meningococci. Virulence 8, 1808–1819 (2017).

548. Chapin, C. W. Carbon Dioxid in the Primary Cultivation of the Gonococcus. J. Infect. Dis. 23, 342–343 (1918).

549. Port, J. L., DeVoe, I. W. & Archibald, F. S. Sulphur acquisition by Neisseria meningitidis. Can. J. Microbiol. 30, 1453–7 (1984).

550. Frantz, I. D. & Jr. Growth Requirements of the Meningococcus. J. Bacteriol. 43, 757–61 (1942).

551. Archibald, F. S. & DeVoe, I. W. Iron in Neisseria meningitidis: minimum requirements, effects of limitation, and characteristics of uptake. J. Bacteriol. 136, 35–48 (1978).

552. Calver, G. A., Kenny, C. P. & Lavergne, G. Iron as a replacement for mucin in the establishment of meningococcal infection in mice. Can. J. Microbiol. 22, 832–8 (1976).

553. Holbein, B. E., Jericho, K. W. & Likes, G. C. Neisseria meningitidis infection in mice: influence of iron, variations in virulence among strains, and pathology. Infect. Immun. 24, 545–51 (1979).

554. Evans, R. W., Crawley, J. B., Joannou, C. L. & Sharma, N. D. Iron proteins. Wiley Chichester 27–86 (1999).

555. Griffith, E. & Williams, P. The iron-uptake systems of pathogenic bacteria, fungi and protozoa. in Iron and infection: molecular,

186 physiological and clinical aspects 87–212 (John Wiley & Sons, 1999).

556. Holbein, B. E. Enhancement of Neisseria meningitidis infection in mice by addition of iron bound to transferrin. Infect. Immun. 34, 120–5 (1981).

557. Holbein, B. E. Iron-controlled infection with Neisseria meningitidis in mice. Infect. Immun. 29, 886–91 (1980).

558. Perkins-Balding, D., Ratliff-Griffin, M. & Stojiljkovic, I. Iron Transport Systems in Neisseria meningitidis. Microbiol. Mol. Biol. Rev. 68, 154–171 (2004).

559. Dyer, D. W., West, E. P. & Sparling, P. F. Effects of serum carrier proteins on the growth of pathogenic neisseriae with heme-bound iron. Infect. Immun. 55, 2171–5 (1987).

560. Mickelsen, P. A. & Sparling, P. F. Ability of Neisseria gonorrhoeae, Neisseria meningitidis, and commensal Neisseria species to obtain iron from transferrin and iron compounds. Infect. Immun. 33, 555–64 (1981).

561. Mickelsen, P. A., Blackman, E. & Sparling, P. F. Ability of Neisseria gonorrhoeae, Neisseria meningitidis, and commensal Neisseria species to obtain iron from lactoferrin. Infect. Immun. 35, 915–20 (1982).

562. Griffiths, E., Stevenson, P. & Ray, A. Antigenic and molecular heterogeneity of the transferrin-binding protein of Neisseria meningitidis. FEMS Microbiol. Lett. 69, 31–36 (1990).

563. Schryvers, A. B. & Morris, L. J. Identification and characterization of the human lactoferrin-binding protein from Neisseria meningitidis. Infect. Immun. 56, 1144–9 (1988).

564. Khun, H. H., Kirby, S. D. & Lee, B. C. A Neisseria meningitidis fbpABC mutant is incapable of using nonheme iron for growth. Infect. Immun. 66, 2330–6 (1998).

565. Stork, M. et al. Zinc Piracy as a Mechanism of Neisseria meningitidis for Evasion of Nutritional Immunity. PLoS Pathog. 9, e1003733 (2013).

566. Stork, M. et al. An Outer Membrane Receptor of Neisseria meningitidis Involved in Zinc Acquisition with Vaccine Potential. PLoS Pathog. 6, e1000969 (2010).

567. Echenique-Rivera, H. et al. Transcriptome Analysis of Neisseria meningitidis in Human Whole Blood and Mutagenesis Studies Identify Virulence Factors Involved in Blood Survival. PLoS Pathog. 7, e1002027 (2011).

568. Pizarro-Cerdá, J. & Cossart, P. Bacterial Adhesion and Entry into Host Cells. Cell 124, 715–727 (2006).

569. Hadi, H. A., Wooldridge, K. G., Robinson, K. & Ala’Aldeen, D. A. A. Identification and characterization of App: An immunogenic autotransporter protein of Neisseria meningitidis. Mol. Microbiol. 41, 611–623 (2001).

570. Serruto, D. et al. Neisseria meningitidis App, a new adhesin with autocatalytic serine protease activity. Mol. Microbiol. 48, 323–34 (2003).

571. Capecchi, B. et al. Neisseria meningitidis NadA is a new invasin which promotes bacterial adhesion to and penetration into human epithelial cells. Mol. Microbiol. 55, 687–698 (2004).

572. Comanducci, M. et al. NadA, a novel vaccine candidate of Neisseria meningitidis. J. Exp. Med. 195, 1445–54 (2002).

573. Peak, I. R., Srikhanta, Y., Dieckelmann, M., Moxon, E. R. & Jennings, M. P. Identification and characterisation of a novel conserved outer membrane protein from Neisseria meningitidis. FEMS Immunol. Med. Microbiol. 28, 329–34 (2000).

574. Scarselli, M. et al. Neisseria meningitidis NhhA is a multifunctional trimeric autotransporter adhesin. Mol. Microbiol. 61, 631–644 (2006).

575. Turner, D. P. J. et al. Characterization of MspA, an immunogenic autotransporter protein that mediates adhesion to epithelial and endothelial cells in Neisseria meningitidis. Infect. Immun. 74, 2957–64 (2006).

576. Swanson, J. Studies on gonococcus infection. XIV. Cell wall protein differences among color/opacity colony variants of Neisseria gonorrhoeae. Infect. Immun. 21, 292–302 (1978).

577. Virji, M. et al. Expression of the Opc protein correlates with invasion of epithelial and endothelial cells by Neisseria meningitidis. Mol. Microbiol. 6, 2785–2795 (1992).

578. Virji, M., Makepeace, K., Ferguson, D. J. P. P., Achtman, M. & Moxon, E. R. Meningococcal Opa and Opc proteins: their role in colonization and invasion of human epithelial and endothelial cells. Mol. Microbiol. 10, 499–510 (1993).

579. Virji, M., Watt, S. M., Barker, S., Makepeace, K. & Doyonnas, R. The N-domain of the human CD66a adhesion molecule is a target for Opa proteins of Neisseria meningitidis and Neisseria gonorrhoeae. Mol. Microbiol. 22, 929–39 (1996).

580. Hammarström, S. The carcinoembryonic antigen (CEA) family: structures, suggested functions and expression in normal and malignant tissues. Semin. Cancer Biol. 9, 67–81 (1999).

187 581. Griffiths, N. J., Bradley, C. J., Heyderman, R. S. & Virji, M. IFN-γ amplifies NFκB-dependent Neisseria meningitidis invasion of epithelial cells via specific upregulation of CEA-related cell adhesion molecule 1. Cell. Microbiol. 9, 2968–2983 (2007).

582. Muenzner, P. et al. Carcinoembryonic antigen family receptor specificity of Neisseria meningitidis Opa variants influences adherence to and invasion of proinflammatory cytokine-activated endothelial cells. Infect. Immun. 68, 3601–7 (2000).

583. Virji, M. et al. Critical determinants of host receptor targeting by Neisseria meningitidis and Neisseria gonorrhoeae : identification of Opa adhesiotopes on the N-domain of CD66 molecules. Mol. Microbiol. 34, 538–551 (1999).

584. Virji, M., Makepeace, K. & Moxon, E. R. Distinct mechanisms of interactions of Opc-expressing meningococci at apical and basolateral surfaces of human endothelial cells; the role of integrins in apical interactions. Mol. Microbiol. 14, 173–184 (1994).

585. Unkmeir, A. et al. Fibronectin mediates Opc-dependent internalization of Neisseria meningitidis in human brain microvascular endothelial cells. Mol. Microbiol. 46, 933–946 (2002).

586. Virji, M. et al. Opc- and pilus-dependent interactions of meningococci with human endothelial cells: molecular mechanisms and modulation by surface polysaccharides. Mol. Microbiol. 18, 741–54 (1995).

587. De Vries, F. P., Cole, R., Dankert, J., Frosch, M. & Van Putten, J. P. M. Neisseria meningitidis producing the Opc adhesin binds epithelial cell proteoglycan receptors. Mol. Microbiol. 27, 1203–1212 (1998).

588. Prince, S. M., Achtman, M. & Derrick, J. P. Crystal structure of the OpcA integral membrane adhesin from Neisseria meningitidis. Proc. Natl. Acad. Sci. U. S. A. 99, 3417–21 (2002).

589. Fox, D. A. et al. Structure of the Neisserial outer membrane protein Opa₆₀: loop flexibility essential to receptor recognition and bacterial engulfment. J. Am. Chem. Soc. 136, 9938–46 (2014).

590. Aho, E. L., Dempsey, J. A., Hobbs, M. M., Klapper, D. G. & Cannon, J. G. Characterization of the opa (class 5) gene family of Neisseria meningitidis. Mol. Microbiol. 5, 1429–1437 (1991).

591. Hobbs, M. M. et al. Recombinational reassortment among opa genes from ET-37 complex Neisseria meningitidis isolates of diverse geographical origins. Microbiology 144, 157–166 (1998).

592. Nassif, X. et al. Roles of pilin and PilC in adhesion of Neisseria meningitidis to human epithelial and endothelial cells. Proc. Natl. Acad. Sci. U. S. A. 91, 3769–3773 (1994).

593. Heckels, J. E. Structure and function of pili of pathogenic Neisseria species. Clin. Microbiol. Rev. 2 Suppl, S66-73 (1989).

594. Virji, M. et al. The role of pili in the interactions of pathogenic Neisseria with cultured human endothelial cells. Mol. Microbiol. 5, 1831–1841 (1991).

595. Nassif, X. et al. Antigenic variation of pilin regulates adhesion of Neisseria meningitidis to human epithelial cells. Mol. Microbiol. 8, 719–725 (1993).

596. Virji, M. et al. Pilus-facilitated adherence of Neisseria meningitidis to human epithelial and endothelial cells: modulation of adherence phenotype occurs concurrently with changes in primary amino acid sequence and the glycosylation status of pilin. Mol. Microbiol. 10, 1013–1028 (1993).

597. Winther-Larsen, H. C. et al. Neisseria gonorrhoeae PilV, a type IV pilus-associated protein essential to human epithelial cell adherence. Proc. Natl. Acad. Sci. U. S. A. 98, 15276–81 (2001).

598. Källström, H., Liszewski, M. K., Atkinson, J. P. & Jonsson, A.-B. Membrane cofactor protein (MCP or CD46) is a cellular pilus receptor for pathogenic Neisseria. Mol. Microbiol. 25, 639–647 (1997).

599. Coureuil, M. et al. Meningococcus Hijacks a β2-Adrenoceptor/β-Arrestin Pathway to Cross Brain Microvasculature Endothelium. Cell 143, 1149–1160 (2010).

600. Bernard, S. C. et al. Pathogenic Neisseria meningitidis utilizes CD147 for vascular colonization. Nat. Med. 20, 725–31 (2014).

601. Olafson, R. W. et al. Structural and antigenic analysis of meningococcal piliation. Infect. Immun. 48, 336–342 (1985).

602. Borst, P. Molecular genetics of antigenic variation. Parasitol. Today 7, 29–33 (1991).

603. Davidsen, T. & Tønjum, T. Meningococcal genome dynamics. Nat. Rev. Microbiol. 4, 11–22 (2006).

604. Diaz, J.-L., Virji, M. & Heckels, J. E. Structural and antigenic differences between two types of meningococcal pili. FEMS Microbiol. Lett. 21, 181–184 (1984).

605. Virji, M., Heckels, J. E., Potts, W. J., Hart, C. A. & Saunders, J. R. Identification of Epitopes Recognized by Monoclonal Antibodies SM1 and SM2 Which React with All Pili of Neisseria gonorrhoeae but Which Differentiate between Two Structural Classes of Pili Expressed by Neisseria meningitidis and the Distribution of Their Encoding Sequences in the Genomes of Neisseria spp. Microbiology 135, 3239–3251 (1989).

606. Wörmann, M. E. et al. Sequence, distribution and chromosomal context of class I and class II pilin genes of Neisseria meningitidis

188 identified in whole genome sequences. BMC Genomics 15, 253 (2014).

607. Virji, M., Alexandrescu, C., Ferguson, D. J. P. P., Saunders, J. R. & Moxon, E. R. Variations in the expression of pili: the effect on adherence of Neisseria meningitidis to human epithelial and endothelial cells. Mol. Microbiol. 6, 1271–1279 (1992).

608. Helm, R. A. & Seifert, H. S. Frequency and rate of pilin antigenic variation of Neisseria meningitidis. J. Bacteriol. 192, 3822–3823 (2010).

609. Davies, J. K. et al. The use of high-throughput DNA sequencing in the investigation of antigenic variation: Application to Neisseria species. PLoS One 9, (2014).

610. Gault, J., Malosse, C., Duménil, G. & Chamot-Rooke, J. A combined mass spectrometry strategy for complete posttranslational modification mapping of Neisseria meningitidis major pilin. J. Mass Spectrom. 48, 1199–1206 (2013).

611. Stimson, E. et al. Meningococcal pilin: a glycoprotein substituted with digalactosyl 2,4-diacetamido-2,4,6-trideoxyhexose. Mol. Microbiol. 17, 1201–1214 (1995).

612. Terje Hegge, F. et al. Unique modifications with phosphocholine and phosphoethanolamine define alternate antigenic forms of Neisseria gonorrhoeae type IV pili.

613. Chamot-Rooke, J. et al. Alternative Neisseria spp. type IV pilin glycosylation with a glyceramido acetamido trideoxyhexose residue. Proc. Natl. Acad. Sci. U. S. A. 104, 14783–8 (2007).

614. Miller, C. P. EXPERIMENTAL MENINGOCOCCAL INFECTION IN MICE. Science 78, 340–1 (1933).

615. Holbein, B. E., Jericho, K. W. & Likes, G. C. Neisseria meningitidis infection in mice: influence of iron, variations in virulence among strains, and pathology. Infect. Immun. 24, 545–51 (1979).

616. Salit, I. E. & Tomalty, L. Experimental meningococcal infection in mice: a model for mucosal invasion. Infect. Immun. 51, 648–52 (1986).

617. Johansson, L. et al. CD46 in meningococcal disease. Science 301, 373–375 (2003).

618. Johswich, K. O. et al. In vivo adaptation and persistence of Neisseria meningitidis within the nasopharyngeal mucosa. PLoS Pathog. 9, e1003509 (2013).

619. Nassif, X., Puaoi, D. & So, M. Transposition of Tn1545-delta 3 in the pathogenic Neisseriae: a genetic tool for mutagenesis. J. Bacteriol. 173, 2147–54 (1991).

620. Jonsson, A. B., Nyberg, G. & Normark, S. Phase variation of gonococcal pili by frameshift mutation in pilC, a novel gene for pilus assembly. EMBO J. 10, 477–488 (1991).

621. Kehl-Fie, T. E., Miller, S. E. & St Geme, J. W. Kingella kingae expresses type IV pili that mediate adherence to respiratory epithelial and synovial cells. J. Bacteriol. 190, 7157–63 (2008).

622. Aim, R. A., Hallinan, J. P., Watson, A. A. & Mattick, J. S. Fimbrial biogenesis genes of Pseudomonas aeruginosa: pilW and pilX increase the similarity of type 4 fimbriae to the GSP protein-secretion systems and pilY1 encodes a gonococcal PilC homologue. Mol. Microbiol. 22, 161–173 (1996).

623. Orans, J. et al. Crystal structure analysis reveals Pseudomonas PilY1 as an essential calcium-dependent regulator of bacterial surface motility. Proc. Natl. Acad. Sci. U. S. A. 107, 1065–70 (2010).

624. Rudel, T., Boxberger, H. -J & Meyer, T. F. Pilus biogenesis and epithelial cell adherence of Neisseria gonorrhoeae pilC double knock-out mutants. Mol. Microbiol. 17, 1057–1071 (1995).

625. Rahman, M., Källström, H., Normark, S. & Jonsson, a B. PilC of pathogenic Neisseria is associated with the bacterial cell surface. Mol. Microbiol. 25, 11–25 (1997).

626. Hoppe, J. et al. PilY1 Promotes Legionella pneumophila Infection of Human Lung Tissue Explants and Contributes to Bacterial Adhesion, Host Cell Invasion, and Twitching Motility. Front. Cell. Infect. Microbiol. 7, 63 (2017).

627. Bohn, Y.-S. T. et al. Multiple roles of Pseudomonas aeruginosa TBCF10839 PilY1 in motility, transport and infection. Mol. Microbiol. 71, 730–747 (2009).

628. Rudel, T., van Putten, J. P., Gibbs, C. P., Haas, R. & Meyer, T. F. Interaction of two variable proteins (PilE and PilC) required for pilus- mediated adherence of Neisseria gonorrhoeae to human epithelial cells. Mol. Microbiol. 6, 3439–50 (1992).

629. Morand, P. C. et al. Type IV pilus retraction in pathogenic Neisseria is regulated by the PilC proteins. EMBO J. 23, 2009–2017 (2004).

630. Taha, M.-K. et al. Pilus-mediated adhesion of Neisseria meningitidis: the essential role of cell contact-dependent transcriptional upregulation of the PilC1 protein. Mol. Microbiol. 28, 1153–1163 (1998).

631. Scheuerpflug, I., Rudel, T., Ryll, R., Pandit, J. & Meyer, T. F. Roles of PilC and PilE proteins in pilus-mediated adherence of Neisseria

189 gonorrhoeae and Neisseria meningitidis to human erythrocytes and endothelial and epithelial cells. Infect. Immun. 67, 834–43 (1999).

632. Taha, M.-K., Giorgini, D. & Nassif, X. The pilA regulatory gene modulates the pilus-mediated adhesion of Neisseria meningitidis by controlling the transcription of pilC1. Mol. Microbiol. 19, 1073–1084 (1996).

633. Morand, P. C., Tattevin, P., Eugene, E., Beretti, J.-L. & Nassif, X. The adhesive property of the type IV pilus-associated component PilC1 of pathogenic Neisseria is supported by the conformational structure of the N-terminal part of the molecule. Mol. Microbiol. 40, 846–856 (2001).

634. Rytkönen, A. et al. Neisseria meningitidis undergoes PilC phase variation and PilE sequence variation during invasive disease. J. Infect. Dis. 189, 402–9 (2004).

635. Wolfgang, M., van Putten, J. P., Hayes, S. F. & Koomey, M. The comP locus of Neisseria gonorrhoeae encodes a type IV prepilin that is dispensable for pilus biogenesis but essential for natural transformation. Mol. Microbiol. 31, 1345–1357 (1999).

636. Cehovin, A. et al. Specific DNA recognition mediated by a type IV pilin. Proc. Natl. Acad. Sci. U. S. A. 110, 3065–70 (2013).

637. Mikaty, G. et al. Extracellular bacterial pathogen induces host cell surface reorganization to resist shear stress. PLoS Pathog. 5, (2009).

638. Helaine, S. et al. PilX, a pilus-associated protein essential for bacterial aggregation, is a key to pilus-facilitated attachment of Neisseria meningitidis to human cells. Mol. Microbiol. 55, 65–77 (2005).

639. Brissac, T., Mikaty, G., Duménil, G., Coureuil, M. & Nassif, X. The meningococcal minor pilin PilX is responsible for type IV pilus conformational changes associated with signaling to endothelial cells. Infect. Immun. 80, 3297–3306 (2012).

640. Imhaus, A.-F. Rôle et mode d’action des pilines mineures des pili de type IV de Neisseria meningitidis. http://www.theses.fr (2013).

641. Eriksson, J. et al. Characterization of motility and piliation in pathogenic Neisseria. BMC Microbiol. 15, 92 (2015).

642. Wolfgang, M. et al. PilT mutations lead to simultaneous defects in competence for natural transformation and twitching motility in piliated Neisseria gonorrhoeae. Mol. Microbiol. 29, 321–330 (1998).

643. Aas, F. E. et al. Competence for natural transformation in Neisseria gonorrhoeae: components of DNA binding and uptake linked to type IV pilus expression. Mol. Microbiol. 46, 749–760 (2002).

644. Berry, J.-L., Cehovin, A., McDowell, M. A., Lea, S. M. & Pelicic, V. Functional Analysis of the Interdependence between DNA Uptake Sequence and Its Cognate ComP Receptor during Natural Transformation in Neisseria Species. PLoS Genet. 9, e1004014 (2013).

645. Frye, S. A., Nilsen, M., Tønjum, T. & Ambur, O. H. Dialects of the DNA Uptake Sequence in Neisseriaceae. PLoS Genet. 9, e1003458 (2013).

646. Ambur, O. H., Frye, S. A. & Tønjum, T. New functional identity for the DNA uptake sequence in transformation and its presence in transcriptional terminators. J. Bacteriol. 189, 2077–85 (2007).

647. Assalkhou, R. et al. The outer membrane secretin PilQ from Neisseria meningitidis binds DNA. Microbiology 153, 1593–1603 (2007).

648. Lang, E. et al. Identification of neisserial DNA binding components. Microbiology 155, 852–862 (2009).

649. Sinha, S., Ambur, O. H., Langford, P. R., Tonjum, T. & Kroll, J. S. Reduced DNA binding and uptake in the absence of DsbA1 and DsbA2 of Neisseria meningitidis due to inefficient folding of the outer-membrane secretin PilQ. Microbiology 154, 217–225 (2008).

650. Hepp, C. & Maier, B. Kinetics of DNA uptake during transformation provide evidence for a translocation ratchet mechanism. Proc. Natl. Acad. Sci. 113, 12467–12472 (2016).

651. Meyer, J. et al. Characterization of MDAФ, A temperate filamentous bacteriophage of Neisseria meningitidis. Microbiol. (United Kingdom) 162, 268–282 (2016).

652. Bille, E. et al. A virulence-associated filamentous bacteriophage of Neisseria meningitidis increases host-cell colonisation. PLOS Pathog. 13, e1006495 (2017).

653. Bonazzi, D. et al. Intermittent pili-mediated forces fluidize Neisseria meningitidis aggregates promoting vascular colonization. Cell in press (2018).

654. Lécuyer, H., Nassif, X. & Coureuil, M. Two strikingly different signaling pathways are induced by meningococcal type IV pili on endothelial and epithelial cells. Infect. Immun. 80, 175–86 (2012).

655. Källström, H. et al. Attachment of Neisseria gonorrhoeae to the cellular pilus receptor CD46: Identification of domains important for bacterial adherence. Cell. Microbiol. 3, 133–143 (2001).

656. Katayama, Y., Hirano, A. & Wong, T. C. Human receptor for measles virus (CD46) enhances nitric oxide production and restricts

190 virus replication in mouse macrophages by modulating production of alpha/beta interferon. J. Virol. 74, 1252–7 (2000).

657. Kirchner, M., Heuer, D. & Meyer, T. F. CD46-independent binding of neisserial type IV pili and the major pilus adhesin, PilC, to human epithelial cells. Infect. Immun. 73, 3072–3082 (2005).

658. Jen, F. E. C. et al. Dual Pili Post-translational Modifications Synergize to Mediate Meningococcal Adherence to Platelet Activating Factor Receptor on Human Airway Cells. PLoS Pathog. 9, e1003377 (2013).

659. Tobiason, D. M. & Seifert, H. S. Inverse relationship between pilus-mediated gonococcal adherence and surface expression of the pilus receptor, CD46. Microbiology 147, 2333–2340 (2001).

660. Sameshima, T. et al. Expression of emmprin (CD147), a cell surface inducer of matrix metalloproteinases, in normal human brain and gliomas. Int. J. Cancer 88, 21–27 (2000).

661. Watanabe, A. et al. CD147/EMMPRIN acts as a functional entry receptor for measles virus on epithelial cells. J. Virol. 84, 4183–93 (2010).

662. Dörig, R. E., Marcil, A., Chopra, A. & Richardson, C. D. The human CD46 molecule is a receptor for measles virus (Edmonston strain). Cell 75, 295–305 (1993).

663. Virji, M. et al. Pilus-facilitated adherence of Neisseria meningitidis to human epithelial and endothelial cells: modulation of adherence phenotype occurs concurrently with changes in primary amino acid sequence and the glycosylation status of pilin. Mol. Microbiol. 10, 1013–1028 (1993).

664. Marceau, M., Forest, K., Béretti, J. L., Tainer, J. & Nassif, X. Consequences of the loss of O-linked glycosylation of meningococcal type IV pilin on piliation and pilus-mediated adhesion. Mol. Microbiol. 27, 705–715 (1998).

665. Soyer, M. et al. Early sequence of events triggered by the interaction of Neisseria meningitidis with endothelial cells. Cell. Microbiol. 16, 878–895 (2014).

666. Eugène, E. et al. Microvilli-like structures are associated with the internalization of virulent capsulated Neisseria meningitidis into vascular endothelial cells. J. Cell Sci. 115, 1231–1241 (2002).

667. Merz, A. J. & So, M. Attachment of piliated, Opa- and Opc- gonococci and meningococci to epithelial cells elicits cortical actin rearrangements and clustering of tyrosine-phosphorylated proteins. Infect. Immun. 65, 4341–9 (1997).

668. Maïssa, N. et al. Strength of Neisseria meningitidis binding to endothelial cells requires highly-ordered CD147/β 2-Adrenoceptor clusters assembled by alpha-Actinin-4. Nat. Commun. 8, (2017).

669. Miller, F. et al. The hypervariable region of meningococcal major pilin PilE controls the host cell response via antigenic variation. MBio 5, e01024-13 (2014).

670. Seifert, H. S., Ajioka, R. S., Paruchuri, D., Heffron, F. & So, M. Shuttle mutagenesis of Neisseria gonorrhoeae: Pilin null mutations lower DNA transformation competence. J. Bacteriol. 172, 40–46 (1990).

671. Froholm, L. O., Jyssum, K. & Bøvre, K. Electron microscopical and cultural features of neisseria meningitidis competence variants. Acta Pathol. Microbiol. Scand. Sect. B Microbiol. Immunol. 81 B, 525–537 (1973).

672. Swanson, J., Kraus, S. J. & Gotschlich, E. C. Studies on gonococcus infection. I. Pili and zones of adhesion: their relation to gonococcal growth patterns. J. Exp. Med. 134, 886–906 (1971).

673. Blake, M. S., MacDonald, C. M. & Klugman, K. P. Colony morphology of piliated Neisseria meningitidis. J. Exp. Med. 170, 1727– 36 (1989).

674. Woods, D. E. et al. Role of pili in adherence of Pseudomonas aeruginosa to mammalian buccal epithelial cells. Infect. Immun. 29, 1146–51 (1980).

675. Virji, M. & Heckels, J. E. The Role of Common and Type-specific Pilus Antigenic Domains in Adhesion and Virulence of Gonococci for Human Epithelial Cells. Microbiology 130, 1089–1095 (1984).

676. Winther-Larsen, H. C. et al. Neisseria gonorrhoeae PilV, a type IV pilus-associated protein essential to human epithelial cell adherence. Proc. Natl. Acad. Sci. 98, 15276–15281 (2001).

677. Chen, A. & Seifert, H. S. Saturating mutagenesis of an essential gene: A majority of the Neisseria gonorrhoeae major outer membrane porin (PorB) is mutable. J. Bacteriol. 196, 540–547 (2014).

678. Obergfell, K. P. & Seifert, H. S. The Pilin N-terminal Domain Maintains Neisseria gonorrhoeae Transformation Competence during Pilus Phase Variation. PLOS Genet. 12, e1006069 (2016).

679. Fowler, D. M. & Fields, S. Deep mutational scanning: A new style of protein science. Nature Methods 11, 801–807 (2014).

680. Cahoon, L. A. & Seifert, H. S. An alternative DNA structure is necessary for pilin antigenic variation in Neisseria gonorrhoeae. Science 325, 764–7 (2009).

191 681. Tan, F. Y. Y., Wörmann, M. E., Loh, E., Tang, C. M. & Exley, R. M. Characterization of a novel antisense RNA in the major pilin locus of Neisseria meningitidis influencing antigenic variation. J. Bacteriol. 197, 1757–1768 (2015).

682. Pujol, C., Eugène, E., Marceau, M. & Nassif, X. The meningococcal PilT protein is required for induction of intimate attachment to epithelial cells following pilus-mediated adhesion. Proc. Natl. Acad. Sci. U. S. A. 96, 4017–4022 (1999).

683. Jolley, K. A. & Maiden, M. C. BIGSdb: Scalable analysis of bacterial genome variation at the population level. BMC Bioinformatics 11, 595 (2010).

684. Zhang, Q. Y., DeRyckere, D., Lauer, P. & Koomey, M. Gene conversion in Neisseria gonorrhoeae: evidence for its role in pilus antigenic variation. Proc. Natl. Acad. Sci. U. S. A. 89, 5366–70 (1992).

685. Oldewurtel, E. R., Kouzel, N., Dewenter, L., Henseler, K. & Maier, B. Differential interaction forces govern bacterial sorting in early biofilms. Elife 4, (2015).

686. Park, H. S. M. et al. Structural alterations in a type IV pilus subunit protein result in concurrent defects in multicellular behaviour and adherence to host tissue. Mol. Microbiol. 42, 293–307 (2001).

687. Forest, K. T. et al. Assembly and antigenicity of the Neisseria gonorrhoeae pilus mapped with antibodies. Infect. Immun. 64, 644– 652 (1996).

688. Strom, M. S. & Lory, S. Amino acid substitutions in pilin of Pseudomonas aeruginosa. J. Biol. Chem. 266, 1656–1664 (1991).

689. Kilmury, S. L. N. & Burrows, L. L. Type IV pilins regulate their own expression via direct intramembrane interactions with the sensor kinase PilS. Proc. Natl. Acad. Sci. U. S. A. 201512947 (2016). doi:10.1073/pnas.1512947113

690. Virji, M. & Heckels, J. E. Antigenic Cross-reactivity of Neisseria Pili: Investigations with Type- and Species-specific Monoclonal Antibodies. Microbiology 129, 2761–2768 (1983).

691. Lu, S. et al. Nanoscale Pulling of Type IV Pili Reveals Their Flexibility and Adhesion to Surfaces over Extended Lengths of the Pili. Biophysj 108, 2865–2875 (2015).

692. Beaussart, A. et al. Quantifying the forces guiding microbial cell adhesion using single-cell force spectroscopy. Nat. Protoc. 9, 1049–1055 (2014).

693. Nilsson, L. M., Thomas, W. E., Trintchina, E., Vogel, V. & Sokurenko, E. V. Catch bond-mediated adhesion without a shear threshold: trimannose versus monomannose interactions with the FimH adhesin of Escherichia coli. J. Biol. Chem. 281, 16656–63 (2006).

694. Zeng, L., Zhang, L., Wang, P. & Meng, G. Structural basis of host recognition and biofilm formation by Salmonella Saf pili. Elife 6, 1–24 (2017).

695. Morand, P. C. et al. Type IV pilus retraction in pathogenic Neisseria is regulated by the PilC proteins. EMBO J. 23, 2009–17 (2004).

696. Virji, M., Heckels, J. E. & Watt, P. J. Monoclonal Antibodies to Gonococcal Pili: Studies on Antigenic Determinants on Pili from Variants of Strain P9. Microbiology 129, 1965–1973 (1983).

697. Kellogg, D. S., Cohen, I. R., Norins, L. C., Schroeter, A. L. & Reising, G. Neisseria gonorrhoeae. II. Colonial variation and pathogenicity during 35 months in vitro. J. Bacteriol. 96, 596–605 (1968).

698. Huang, Y. & Zhang, L. An In Vitro Single-Primer Site-Directed Mutagenesis Method for Use in Biotechnology. in 375–383 (Humana Press, New York, NY, 2017). doi:10.1007/978-1-4939-6472-7_26

699. Varet, H., Brillet-Guéguen, L., Coppée, J.-Y. & Dillies, M.-A. SARTools: A DESeq2- and EdgeR-Based R Pipeline for Comprehensive Differential Analysis of RNA-Seq Data. PLoS One 11, e0157022 (2016).

700. Robinson, M. D., McCarthy, D. J. & Smyth, G. K. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26, 139–140 (2010).

701. Schindelin, J. et al. Fiji: an open-source platform for biological-image analysis. Nat. Methods 9, 676–682 (2012).

702. Harlow, E. & Lane, D. Antibodies, a Laboratory Manual. Cold Spring Harb. Lab. Press 649–654 (1988). doi:10.1016/0968- 0004(89)90307-1

703. Cisneros, D. A., Bond, P. J., Pugsley, A. P., Campos, M. & Francetic, O. Minor pseudopilin self-assembly primes type II secretion pseudopilus elongation. EMBO J. 31, 1041–1053 (2012).

704. Giltner, C. L., Habash, M. & Burrows, L. L. Pseudomonas aeruginosa minor pilins are incorporated into type IV Pili. J. Mol. Biol. 398, 444–461 (2010).

705. Pujol, C., Eugène, E., Marceau, M. & Nassif, X. The meningococcal PilT protein is required for induction of intimate attachment to epithelial cells following pilus-mediated adhesion. Proc. Natl. Acad. Sci. U. S. A. 96, 4017–22 (1999).

706. Basso, P. et al. Pseudomonas aeruginosa pore-forming exolysin and type IV pili cooperate to induce host cell lysis. MBio 8, e02250- 16 (2017).

192 707. Koirala, S. et al. A nutrient-tunable bistable switch controls motility in Salmonella enterica serovar Typhimurium. MBio 5, e01611- 14 (2014).

708. Kearns, D. B. & Losick, R. Cell population heterogeneity during growth of Bacillus subtilis. Genes Dev. 19, 3083–3094 (2005).

709. Basset, A. et al. Expression of the type 1 pneumococcal pilus is bistable and negatively regulated by the structural component RrgA. Infect. Immun. 79, 2974–2983 (2011).

710. De Angelis, G. et al. The Streptococcus pneumoniae Pilus-1 Displays a Biphasic Expression Pattern. PLoS One 6, e21269 (2011).

711. Casadesús, J. & Low, D. A. Programmed Heterogeneity: Epigenetic Mechanisms in Bacteria. (2013). doi:10.1074/jbc.R113.472274

712. Basset, A. et al. An epigenetic switch mediates bistable expression of the type 1 pilus genes in Streptococcus pneumoniae. J. Bacteriol. 194, 1088–91 (2012).

713. Wu, S. S. & Kaiser, D. Regulation of expression of the pilA gene in Myxococcus xanthus. J. Bacteriol. 179, 7748–58 (1997).

714. Bertrand, J. J., West, J. T. & Engel, J. N. Genetic analysis of the regulation of type IV pilus function by the Chp chemosensory system of Pseudomonas aeruginosa. J. Bacteriol. 192, 994–1010 (2010).

715. Laskos, L., Dillard, J. P., Seifert, H. S., Fyfe, J. A. . & Davies, J. K. The pathogenic neisseriae contain an inactive rpoN gene and do not utilize the pilE σ54 promoter. Gene 208, 95–102 (1998).

716. Carrick, C. S., Fyfe, J. A. . & Davies, J. K. The normally silent σ54 promoters upstream of the pilE genes of both Neisseria gonorrhoeae and Neisseria meningitidis are functional when transferred to Pseudomonas aeruginosa. Gene 198, 89–97 (1997).

717. Parkhill, J. et al. Complete DNA sequence of a serogroup A strain of Neisseria meningitidis Z2491. Nature 404, 502–506 (2000).

718. Tettelin, H. et al. Complete genome sequence of Neisseria meningitidis serogroup B strain MC58. Science 287, 1809–15 (2000).

719. Zöllner, R., Oldewurtel, E. R., Kouzel, N. & Maier, B. Phase and antigenic variation govern competition dynamics through positioning in bacterial colonies. Sci. Rep. 7, 12151 (2017).

720. Båga, M., Norgren, M. & Normark, S. Biogenesis of E. coli Pap pili: PapH, a minor pilin subunit involved in cell anchoring and length modulation. Cell 49, 241–251 (1987).

721. van Schaik, E. J. et al. DNA binding: a novel function of Pseudomonas aeruginosa type IV pili. J. Bacteriol. 187, 1455–64 (2005).

722. Lim, M. S. et al. Vibrio cholerae El Tor TcpA crystal structure and mechanism for pilus-mediated microcolony formation. Mol. Microbiol. 77, 755–770 (2010).

723. Saiman, L. & Prince, A. Pseudomonas aeruginosa pili bind to asialoGM1 which is increased on the surface of cystic fibrosis epithelial cells. J. Clin. Invest. 92, 1875–1880 (1993).

724. Bucior, I., Pielage, J. F. & Engel, J. N. Pseudomonas aeruginosa Pili and Flagella mediate distinct binding and signaling events at the apical and basolateral surface of airway epithelium. PLoS Pathog. 8, e1002616 (2012).

725. Lee, K. K., Doig, P., Irvin, R. T., Paranchych, W. & Hodges, R. S. Mapping the surface regions of Pseudomonas aeruginosa PAK pilin: the importance of the C-terminal region for adherence to human buccal epithelial cells. Mol. Microbiol. 3, 1493–1499 (1989).

726. Irvin, R. T. et al. Characterization of the Pseudomonas aeruginosa pilus adhesin: confirmation that the pilin structural protein subunit contains a human epithelial cell-binding domain. Infect. Immun. 57, 3720–6 (1989).

727. Wong, W. Y. et al. Structure-Function Analysis of the Adherence-Binding Domain on the Pilin of Pseudomonas aeruginosa Strains PAK and KB7. Biochemistry 34, 12963–12972 (1995).

728. Halabi, N., Rivoire, O., Leibler, S. & Ranganathan, R. Protein Sectors: Evolutionary Units of Three-Dimensional Structure. Cell 138, 774–786 (2009).

729. Temmerman, K. & Nickel, W. A novel flow cytometric assay to quantify interactions between proteins and membrane lipids. J. Lipid Res. 50, 1245–54 (2009).

730. Tinevez, J.-Y. et al. TrackMate: An open and extensible platform for single-particle tracking. Methods 115, 80–90 (2017).

731. Gorelik, R. & Gautreau, A. Quantitative and unbiased analysis of directional persistence in cell migration. Nat. Protoc. 9, 1931– 1943 (2014).

193

194

Acknowledgements195

To all the jury members: A Guillaume. Jeremy Derrick, Olivera Je voudrais te remercier Francetic, Han Remaut and de m’avoir donné Aleksandra Walczak l’opportunité de réaliser for accepting to be part of my examining committee ma thèse dans ton and taking on their precious time to evaluate my laboratoire et par la même occasion de m’avoir work. In particular, I would like to thank the permis de faire d’enrichissantes rencontres. Merci rapporteurs for their thorough reading of the de ton investissement qui nous a assuré des manuscript and Olivera Francetic for her insightful conditions de travail très confortables. Merci aussi comments on the article and helpful discussions. de m’avoir ouvert de nouveaux horizons. Enfin, merci pour le champagne, les bières et les pisco sour !

To the members of my thesis advisory committee: Aux fantastiques collègues et ami.e.s : Marc Lecuit, Sven van Teeffelen and my tutor Aux occupant.e.s du Bureau Du Bonheur, à Catherine Werts commencer par sa co-fondatrice Dai dai dai for following the progression of my doctoral work Darinetta, merci pour les caffè, les conseils, les and their precious advice. blaguina, le niveau sonore, les heures de visionnage de clip et surtout, l’arrosage des plantes ! Aux autres locataires à temps partiel : Tomas pour ses diagnostics et sa contagieuse bonne humeur, To all the collaborators from outside of the lab: Daiki pour sa patience, Marie-Paule pour ses Pierre Henri Commere for his great availability chaussettes à paillettes et Isabelle pour toutes ses and technical help in bacterial sorting and flow délicates attentions. cytometry, Aux visiteurs du BDB (par ordre de squat

Nikhil Malvankar and members of his lab for décroissant) : la grosse Pierre pour ses performing and driving the experiments on recommandations éclectoélectroniques, Arthur electrically conductive pili, pour son accueil chaleureux et son aide précieuse, Alexis Vogel and Alexandre Chenal for their Dorian pour ses beaux yeux, Valeria pour les indispensable work for lipid-binding studies, Anna bananas et les cumpleaños, Jean-Philippe pour Sartori-Rupp and Gérard Pehau-Arnaudet for ses blagues de bon goût, Sylvie pour son assistance their help in cryo-EM and Matthieu Piel and indispensable et ses petites piques, Youxin et à Rafaele Attia for their initial assistance in Camille la plus jeune recrue, bon courage pour la microfluidics experiments. suite!

196

Aux ancien.ne.s: Hebert, Flore, Corinne et Ximing A la famille bis, et aux voisin.e.s : Aline, Sylvain, Andrea, Sophie, coloc de l’angoisse, Fabiane et Barbara. babzous du turfu et L. roots à leurs heures perdues, les indispensables de la vie.

Louise, la belle plante, coloc de l’éternel et pour

l’éternité (?). Merci pour la danse, les danses, les meilleurs cadeaux de la Terre, la joie, les quarts de pizza, les débats, les pérégrinations pédestres, les A tou.te.s les ami.e.s chambres à air, les acronymes, les 7 minutes. sans qui la thèse aurait été beaucoup नम#ते ma sœur.

beaucoup beaucoup Adddddelise pour les bons plats (FAF 2019, on y trop pénible, vous êtes croit), les belles plantes, l’arbitrage des discussions au top. de la table carrée, blue lagoon, le bingewatching pop

culturel, les conseils mode et capillaires, les cailloux De l’ENS : Valentin, Irini, Clothilde, Arthur, et l’amouuur. Mateusz, Clémence P, Katy, Elizabeth, Yanniv, T’es belle ma belle.

Hemini, Imogen, Aurore, Adèle, Félix et Camille. Charlotte pour les traquenards, les parties de baby, les criiiiiis, les ambiances lumineuses, les conseils De la prépa : Alma, Emma, Mathilde B, Mathilde puces, les HISTOIRES, le GrafIK DESiGn (non T, Sophie, Clémence D, Claire, Marie, Anaïs et rémunéré), la TDR, les FOS, les ouh yeaaaah, les Margaux. conques.

Poissonne un jour, poissonne toujours. De FdV (par ordre ALPHABÉTIQUE) : Bérengère,

Carlos, Clara, Dany, Frances, Lise, Marion et Lucas, coloc adoptif, ce que la Suisse a fait de Quentin. meilleur. Merci pour ta drôlerie, ton sans-gêne, tes fondues, ton velours et tes ateliers créatifs. Aux inclassables : Marie-Alphie, Ewelina, Please come back from Yosemitiiii. Alexandre, Christina et Alexandra. Juliette, ex-coloc encore un peu coloc. Merci pour

l’éducation cinématographique, les rires, les Merci Arthur de m’avoir accompagné pendant la vacances, le yoga et le chocolat. majeure partie de cette aventure. C’était drôlement chouette. Et aussi aux colocs figurantes : Dragana et Daisy, on se voit au Canada!

197

Aux nouvelles branches du Papa d’avoir su grandir avec buisson familial - Laurent, A la famille qui a toujours été là. nous, merci pour ta drôlerie, Ginette, Valérie, Eva, Jean- ta vivacité, ton exigence, ta Michel, Anne-Sophie, Nina curiosité et ton amour. et Chantal.

Papou pour les souvenirs Maman de tout nous avoir d’enfance, l’escalade, les donné, j’espère que ça en valait randonnées, le jardinage, merci la peine. Merci de tes de m’avoir aidé à devenir une valeureux conseils et de ton tronche d’ail épanouie et (un éternelle et indispensable peu) moins susceptible. présence. Pour tout, merci.

Manou d’avoir été et d’être Enfin à la compagne d’enfance, encore la meilleure des Emilie, merci pour ton aide confidentes, merci d’être dans les premiers moments restée si jeune et si drôle, merci difficiles, merci pour les pour ton amour et ton soutien. premières bêtises et pour celles qui restent à venir.

198

SUMMARY RÉSUMÉ

Type IV pili (TFP) are micrometric filaments Les pili de type IV (PT4) sont des filaments involved in multiple functions at the surface of micrométriques qui exercent de multiples many prokaryotes. fonctions à la surface de nombreux procaryotes. In Neisseria meningitidis, TFP are Chez Neisseria meningitidis, les PT4 sont des homopolymers of the major pilin PilE. Their role homopolymères de la piline majeure PilE. Leur in interbacterial aggregation and adhesion to implication dans l’agrégation interbactérienne et human cells provides them a central role in l’adhésion aux cellules humaines leur donne un meningococcal virulence. However, the rôle central dans la virulence du méningocoque. mechanisms behind the multiple functions of Cependant, les mécanismes permettant aux PT4 TFP are still too elusive. d’exercer ces diverses fonctions restent encore During this Ph.D., we simultaneously trop élusifs. determined the regions of PilE involved in pili Durant ce doctorat, nous avons simultanément assembly, auto-aggregation, human cell déterminé les régions de PilE impliquées dans adhesion, and transformation competence l’assemblage des pili, l’auto-agrégation, using the deep mutational scanning technique. l’adhésion aux cellules humaines et la This functional map of the pilin sequence compétence à la transformation en utilisant la coupled with its detailed analysis offers new technique de deep mutational scanning. perspectives on the mechanisms behind TFP L’obtention de cette carte fonctionnelle de la multiple functions. séquence de la piline couplée à son analyse First, the hyperconserved α1N domain of PilE is détaillée nous offre de nouvelles perspectives sur involved in regulating the balance between pili les mécanismes de fonctionnement des PT4. length and number; furthermore, we have Tout d’abord, le domaine hyperconservé α1N de identified a group of electropositive amino PilE est impliqué dans la régulation de la balance acids around lysine 140 required for entre la longueur et le nombre des pili ; par aggregation; Finally, we show the importance ailleurs, nous avons identifié un groupe d’acides of the distal end of TFP in adhesion. aminés électropositifs autour de la lysine 140 In summary, these results support a direct role requis pour l’agrégation ; enfin, nous montrons for PilE in bacterial aggregation and adhesion l’importance de l’extrémité distale des PT4 dans and identify areas specifically involved in these l’adhésion. functions. En résumé, ces résultats sont en faveur d’un rôle This work also opens up new perspectives on direct de PilE dans l’agrégation et l’adhésion the pathogenicity of Neisseria meningitidis and bactérienne et identifient les domaines could contribute to the development of new spécifiquement impliqués dans ces fonctions. therapies to treat pathologies caused by Ces travaux ouvrent aussi de nouvelles meningococcus. perspectives sur la pathogénicité de Neisseria meningitidis et pourraient participer au développement de nouvelles thérapies pour combattre les pathologies provoquées par le méningocoque.