Water, water, every where—the advent of hydrated pili structure

Abstract

Protein structure is defined by its surrounding water. Naturally, this is also true for the gigantic protein filaments that pathogenic bacteria use to attach to and colonize their host cells. Structures of these filaments for which water can be identified with high confidence have stayed elusive for the longest time. In this issue of The FEBS Journal, Petrova et al. unveil the first fully hydrated Type I pilus structure from uropathogenic E. coli. In a powerful showcase of the current capabilities of cryo‐EM, their structure beautifully visualizes internal water channels that traverse the pilus, thus marking the advent of hydrated pilus structural biology.

Protein structure is defined by its surrounding water. Naturally, this is also true for the gigantic protein filaments that pathogenic bacteria use to attach to and colonize their host cells. Structures of these filaments for which water can be identified with high confidence have stayed elusive for the longest time. In this issue of The FEBS Journal, Petrova et al. unveil the first fully hydrated Type I pilus structure from uropathogenic E. coli. In a powerful showcase of the current capabilities of cryo‐EM, their structure beautifully visualizes internal water channels that traverse the pilus, thus marking the advent of hydrated pilus structural biology.

Abbreviations

CDS

correlated double sampling

cryo‐EM

cryogenic electron microscopy

EMDB

Electron Microscopy Data Bank

EMPIAR

Electron Microscopy Public Image Archive

ETEC

enterotoxigenic Escherichia coli

NeCEN

The Netherlands Centre for Electron Nanoscopy

PDB

Protein Data Bank

UPEC

uropathogenic Escherichia coli

Introduction

The title, a quotation from Coleridge's poem The Rime of the Ancient Mariner ‘Water, water, every where/Nor any drop to drink’ describes the situation of being surrounded by water without being able to make good use of it. A situation akin to solving protein structures in vitreous ice without being able to incorporate water into the structural analysis. Water is the mother liquor of life that allows biomolecules to take the shape they evolved to fulfill their biological roles [1]. The amazing speed at which proteins assume their correct fold is magnitudes faster than a random process, such as the movement of long amino acid chains thrust around by Brownian motion, and is thought to be guided by the shell of water molecules in direct contact with residues at the protein surface. The tetrahedral hydrogen bonding network formed by water restricts the side‐chain conformations of amino acid residues at the protein surface. These restraints, in tandem with the hydrophobic effect, perpetuate to the protein core and determine overall protein architecture [2, 3]. Internal water molecules are crucial to allow for functionally relevant polar residues in the hydrophobic core of the protein, while also providing room for conformational flexibility through water's ultrafast movement and its flexible hydrogen bond geometry.

The precise identification of hydration sites on protein surfaces and interiors has traditionally relied on high‐resolution X‐ray crystallography [3]. However, interpretation of bound waters is often constrained by the crystal packing environment and nonphysiological crystallization conditions. Cryo‐EM circumvents these limitations, allowing proteins to be visualized under near‐native conditions. However, resolving ordered waters requires high resolution.

Owing to recent advances in hardware and image‐processing methods [4, 5], cryo‐EM has become capable of resolving structures at atomic or near‐atomic resolutions sufficient to enable the visualization of water. This holds true even in challenging cases involving protein polymers beyond the reach of crystallography and has now been demonstrated for bacterial pili [6, 7]. These cryo‐EM structures, resolved with bound water molecules, reveal new aspects of the biology of these remarkable protein polymers.

High‐resolution cryo‐EM of frozen hydrated bacterial pili

Bacterial pili are hair‐like protein polymers composed of pilin subunits and are produced by a wide variety of Gram‐positive and Gram‐negative bacteria. Multiple types of pili have been reported, reflecting variations in pilin composition and assembly mechanisms [8]. Pili perform several functions, including adhesion to and colonization of host tissues, motility, and protein secretion, all of which are critical for ‘successful’ infection by pathogenic bacteria. Therefore, detailed structural knowledge of pili provides an essential basis for the development of antimicrobial agents and vaccines. However, their large size, insolubility, and inherent flexibility have historically posed challenges for high‐resolution structural analysis [9].

In this issue of The FEBS Journal, Petrova et al. [6] report a near‐atomic‐resolution cryo‐EM structure of the type I pilus from uropathogenic Escherichia coli (UPEC), a pathogen responsible for chronic urinary tract infections that are increasingly resistant to standard antibiotic treatment. The authors utilized modern cryo‐EM instrumentation at the NeCEN public facility (Leiden, The Netherlands) to obtain a structure of the type I pilus composed of 15 subunits of the major pilin protein FimA in its native assembly. Their helical reconstruction of the type I pilus at an exceptionally high resolution of 1.85 Å provided an excellent map quality, which was sufficient to assign densities corresponding to water molecules with high confidence, thus allowing them to construct a detailed water network for this biomedically relevant pilus. The interpretation of cryo‐EM density as water molecules, especially when distinguishing them from noise, is often subject to debate, and several approaches have been proposed to address this issue [10, 11]. By cross‐referencing the slightly earlier released 2.2 Å hydrated structure of the homologous (142 out of 159 residues are identical) type I pilus from enterotoxigenic Escherichia coli (ETEC) [7], Petrova et al. were able to further validate their water assignments and probe sequence‐specific hydration sites of the type I pilus. Their findings provide critical insights into the molecular architecture underlying pilus stability and flexibility and the mechanisms underlying bacterial adhesion.

A water channel in the pilus filament

As mentioned above, the UPEC type I pilus is composed of 15 subunits of the major pilin protein FimA. Previous studies have described that subunit interactions within the pilus rod can contribute to its biomechanical properties [12]. However, the incorporation of water molecules provides new insights into these interactions. Petrova and coauthors describe that in addition to the well‐established hydrogen bonding and hydrophobic interactions, two new types of water–protein interactions: ‘direct bridging’ and ‘extended hydrogen bond clusters’, stabilize the intersubunit surfaces, thus contributing to both stability and flexibility of the pilus rod.

Interestingly, the hydration sites at inter‐subunit contact regions described by Petrova et al. differ from those previously identified by high‐resolution X‐ray crystallography of the isolated FimA subunit [13], highlighting the importance of high‐resolution cryo‐EM for understanding the physiological hydration of pili. Their observations afford the visualization of hitherto undiscovered water molecules, revealing that the regions between subunits are filled with a network of water forming helically arranged water channels traversing the length of the pilus filament (Fig. 1, left panel). Remarkably, despite having distinct pilin structures and different assembly mechanisms, the recent 1.8 Å‐resolution structure of type IV pili from ETEC [7] also revealed similar water channels (Fig. 1, right panel). As suggested by high‐resolution cryo‐EM structures of F‐actin filaments [14], the existence of water channels does not only stabilize subunit‐subunit interactions but also illuminates the dynamic pathways that underlie the remarkable flexibility and function of pilus filaments.

Fig. 1.

Solvent channels in type I and type IV pilus filaments. Cryo‐EM structures of the type I pilus from UPEC (left: EMDB ID: 66617; PDB ID: 9X67) and the type IV pilus from ETEC (right: EMPIAR ID: 13411; EMDB ID: 60902; PDB ID: 9IUF) highlight the solvent channels formed between pilus subunits. FimA (type I) and CofA (type IV) are shown as white strands, and inter‐subunit water molecules lining the channels are depicted as red spheres. Surface waters are omitted.

Water visualization by cryo‐EM

Both type I and type IV pilus structures have now been resolved at exceptional resolutions of better than 2 Å, revealing for the first time the full hydration network that had long been missed, despite its obvious presence around the protein [6, 7]. In fact, at the time of writing, out of a total of 2474 entries for helical reconstructions, the EMDB contains only 20 helical reconstructions at resolutions of better than 2 Å, of which the majority are amyloid fibers, that is, hydration structures of bacterial pili are still hard to obtain.

So, this naturally brings us to the question of how water visualization using cryo‐EM worked out so beautifully in this study by Petrova et al. [6], and our previous work on ETEC Type IV pilus structure [7]. Structure determination by cryo‐EM is a complex, multistep process involving experimenter‐dependent steps such as purification, cryo‐grid freezing, or image processing, which makes it somewhat difficult to pinpoint the reasons for the success of a particular project. Nevertheless, some insight might be gleaned from the conditions of data acquisition in these studies. In both studies, images were acquired using zero loss energy filtering and a Gatan K3 direct electron detector in correlated double sampling (CDS) mode [15]. These are hardware configurations that might be favorable for projects aiming for the visualization of hydration networks. Additionally, the high level of maintenance and microscope tuning afforded at the public use cryo‐EM facilities used in both studies could be another important but easily overlooked factor in achieving the necessary resolution required to visualize water networks.

Excitingly, the problem of accurate assignment of hydration sites and the potential visualization of loosely bound water using high‐resolution cryo‐EM maps has attracted considerable attention as a new avenue of research by cryo‐EM [10, 11]. We hope that increased deposition of high‐resolution pili raw image data sets to the raw electron microscope image data base EMPIAR [16], as was done here for the type IV pilus data set (EMPIAR ID: 13411), will aid in the improvement of image processing software for helical assemblies and their hydration shells. Together with further improved cryo‐EM hardware [17], this could jump‐start the routine high‐resolution analysis of hydrated pili.

Conclusions

The exciting advent of routine mapping of hydrated pili structures at better than 2 Å is anticipated to allow structure‐based and artificial intelligence‐guided drug design, similar to what is already feasible for the development of novel inhibitors of the gastric proton pump [18]. This could then, for instance, lead to the development of small‐molecule disruptors of the water glue that conveys stability and flexibility to enterotoxic pili, potentially providing relief to patients of chronic urinary tract infections resistant to existing pharmacological treatments.

Conflict of interest

The authors declare no conflict of interest.

Author contributions

All authors discussed the results by Tatiana Petrova et al., drafted the commentary, and edited its contents.