Abstract

The Role of Two HtrA/DegP Family Proteases, MamE and MamO, in Protein Sorting and Biomineralization in Magnetospirillum magneticum str AMB-1

by

Anna Wiedmann

Doctor of Philosophy in Molecular and Cell Biology

University of California, Berkeley

Professor Arash Komeili, Chair

Magnetotactic bacteria (MTB) are a diverse group of bacteria capable of synthesizing magnetosomes, membrane-bound prokaryotic organelles that biomineralize magnetic minerals. Magnetosomes allow cells to align passively with geomagnetic field lines. This is thought to facilitate a bacterium's search for favorable environments, a process referred to as magnetotaxis. To execute biomineralization, a specific set of proteins needs to be localized to the magnetosomes. I have taken both a candidate and a global approach to understanding how these magnetosome proteins (MPs) are sorted to the magnetosome.

In a global approach I attempted to identify magnetosome sorting signals by assaying magnetosome localization of N-terminal truncations of MPs. To carry out this approach I unsuccessfully attempted to optimize magnetosome purification protocols to separate inner membrane proteins from magnetosome membrane proteins.

In the more successful candidate approach, I show that the HtrA/DegP family protease MamE is essential for localization of several soluble, as well as membrane bound MPs. Using site-directed mutagenesis, I show that MamE's role in protein sorting is independent of its protease activity, but that protease activity is essential for a novel 20nm crystal size transition step. Site-directed mutagenesis also revealed that the protease activity of a second HtrA/DegP family protease, MamO, is essential for biomineralization. Furthermore, these studies suggest that MamE and MamO's additional functional domains, not usually found in HtrA/DegP family proteases, are required for magnetosome formation. Lastly, this work highlights the step-wise assembly of magnetosomes, illustrates that several levels of control underlie magnetosome formation, and proposes that by modulating MamE's protease activity, cells could potentially exert control over their degree of magnetism and thus magnetotaxis.

Abstract

Regulation of Biomineralization by an HtrA Protease in Magnetotactic Bacteria

by

Patrick Jeffrey Browne

Doctor of Philosophy in Microbiology

University of California, Berkeley

Professor Arash Komeili, Chair

Magnetotactic bacteria are a unique group of bacteria characterized by their ability to manufacture magnetic crystals for the purpose of aligning with geo-magnetic fields. They offer a unique opportunity for merging the study of bacterial cellular biology, classic molecular genetics, biochemistry, and evolution, both on the small scale of individual proteins, and a larger system of proteins with a defined complex function. This dissertation will first look at the genetics and evolution of magnetotactic bacteria as a whole before focusing on MamE, an HtrA protease conserved throughout magnetotactic bacteria. HtrA proteases are ubiquitous throughout all domains of life and serves several different functions. This dissertation will show that MamE is a multifunctional protein and will also highlight other magnetosome related genes that either regulate MamE’s activity or are targets of proteolysis.

The first chapter of this dissertation is a to-be-submitted review article and introduces the genetics of magnetotactic bacteria (MTB), a paraphyletic group of gram-negative bacteria that use around a hundred genes, located in a defined gene island, to create magnetosomes, membrane-bound compartments containing a magnetic iron crystal. By organizing magnetosomes into a chain, cells are able to align with geo-magnetic fields. It will discuss the most well-conserved genes whose general functions are known. It also describes the interesting evolutionary history of the magnetosome gene island (MAI) which despite being an island seemingly capable of horizontal gene transfer, does not contain other hallmarks of recent HGT and closely matches the phylogenetic tree.

The second chapter, a published primary research article (Hershey et al., PLOS Biology 2016, and) is focused on MamO, a protein that also appears to also be an HtrA protease similar to MamE. However, closer examination shows that is contains no protease activity and is instead a metal-binding protein most likely involved in crystal nucleation. It is also shown to not only be a target of MamE proteolysis but also regulate MamE’s activity. Finally, it contains an analysis of MamOs throughout MTB and shows that they are not actually a single protein family. Instead, four separate lineages of MTB have all acquired inactive proteases in a convergent manner.

The third chapter, a published primary research article (Hershey et al., Journal of Biological Chemistry 2016) is the first to focus primarily on MamE. Unlike MamO, this chapter shows that MamE is a bonafide protease and identifies several targets of MamE, including MamE itself, MamO, and MamP. It also interrogates the kinetics of MamE’s protease activity and shows that MamE processes its targets in a regulated manner.

The fourth chapter, to be submitted as a primary research article, also investigates MamE and its targets, this time beginning with a proteomic approach. Cells carrying an inactive form of MamE have deformed smaller crystals. Using a novel method of cell lysis and magnetosome separation, wild-type and mutant magnetosome are compared. This method uncovered a few potential additional targets of MamE proteolysis, one of which was confirmed in vitro. This protein, MamD was also further investigated and determined to possibly function as a biomineralization inhibitor. This chapter also discusses the proteome of the magnetosome as a whole, defining proteins enriched in the magnetosome

Finally, the fifth chapter contains unpublished material related to MamE and ends with concluding thoughts and perspectives. It includes a preliminary investigation of how magnetosome proteins reach the magnetosome, a poorly understood topic, but one that MamE also plays a role in. It will also identify one further potential target of MamE proteolysis, MamT, and investigates how the magnetochrome domain, shared by MamE, MamP, and MamT is crucial to the function of all three proteins but not in identical ways. Finally, it will include thoughts on ways to move forward in the field of magnetosome genetics and molecular biology.

Magnetotactic bacteria (MTB) are a phylogenetically diverse group of bacteria remarkable for their ability to biomineralize magnetite (Fe3O4) or greigite (Fe3S4) in organelles called magnetosomes. MTB can account for up to 30% of the microbial biomass in a an aquatic habitat [1]. MTB also take up large amounts of iron from the environment in order to form magnetosomes—they are estimated to take up anywhere from ~1-50% of dissolved iron inputs into the ocean [2]. However, relatively little is known about the ecology of MTB, other than species diversity in sampled habitats [3]. On the contrary, much is known about the genetics and cell biology of MTB. The majority of genes required for magnetosome formation are encoded by a magnetosome gene island (MAI) and many of their functions are known. However, no previous studies have involved unbiased genetic analysis of biomineralization genes. This work expands the genetic toolset for studying MTB in the lab and identifies novel genes—or functions of genes—that have an impact on biomineralization. Importantly, this dissertation also lays the groundwork to bridge the gap between genetic and environmental studies of MTB.

The first chapter of this dissertation, a published review article, covers the work that has been done on the genetics of MTB and where the field is headed. We focus on the genetics behind the formation of magnetosomes and biomineralization. Then, we cover the history of genetic discoveries in MTB and key insights that have been found in recent years and provide a perspective on the future of genetic studies in MTB.

The second chapter of this dissertation, a published primary research article, takes an unbiased approach to study genes, both inside and outside the MAI, needed for magnetosome formation under different growth conditions. First, we developed the use of random barcoded transposon mutagenesis (RB-TnSeq) in Magnetospirillum magneticum AMB-1 and generated a library of tens of thousands of unique mutant strains. The RB-TnSeq library allowed us to determine the essential gene set of AMB-1 under standard growth conditions. We also performed magnetic selection screen in varied growth conditions to uncover novel genes required for biomineralization in high iron and anaerobic growth conditions. We discovered more nuanced functions for the MAI gene mamT and several ex-MAI genes that may be important for biomineralization.

The third chapter of this dissertation, a published review article written in collaboration with Dr. Andy Tay, covers the potential uses for microfluidics devices for studying MTB from separating cells based on magnetic properties to directed evolution to single cell analysis. Being able to study magnetotactic bacteria in highly controlled, quantifiable environments will be a boon to the field. Additionally, using microfluidics devices to mimic natural environments will be valuable for linking the cell biology and genetics of MTB to their ecology. I end the dissertation with a concluding chapter on the future prospects for employing genetics to understand the molecular mechanisms of biomineralization in a variety of environmental contexts.

Cellular organization is critical for life to flourish. Historically, this area of research has focused on eukaryotic systems, but recent advances have resulted in bacteria as model organisms to study cell biology. This research is aimed to elucidate how bacteria organize their internal structures. To address this question, the bacterial model organism, Magnetospirillum magneticum AMB-1 (AMB-1) was used and the role of two bacterial actin proteins, MamK and MamK-like, were studied.

The first chapter of this dissertation, a published review article (Cornejo-Warner et al., Current Opinion in Cell Biology 2014) written in collaboration with fellow Komeili lab member Elias Cornejo-Warner, introduces the topic of bacterial cell compartmentalization, highlighting examples from Bacillus subtilis and Caulobacter crescentus. Additionally, the organization of bacterial organelles is discussed with examples of two model systems, the magnetosomes of magnetotactic bacteria and the carboxysomes of cyanobacteria.

The second chapter of this dissertation, a combination of a published primary research article (Abreu et al., Journal of Bacteriology 2014) and unpublished supporting data, focuses on a bacterial actin protein, MamK-like. MamK-like is encoded in the magnetosome islet, a region of DNA distinct from the more heavily studied magnetosome island in AMB-1. The work described in this chapter, with assistance in molecular biology from Ertan Ozyamak and in vitro work from Soumaya Mannoubi, resulted in determining that this protein does indeed play a role in AMB-1 organelle alignment.

The third chapter of this dissertation (unpublished work) focuses on the efforts that were taken to further understand the mechanism of MamK turnover by Flourescence Recovery After Photobleaching (FRAP) studies, a continuation of previous work published by Olga Draper (Draper et al. Molecular Microbiology 2011).

The fourth chapter of this dissertation (unpublished work) focuses on the experiments developed (ex vivo: Bacterial two hybrid and in vitro: MamK affinity tagged pulldowns and Mass Spectrometry) to identify MamK and MamJ interacting proteins.

The fifth chapter of this dissertation is a conclusion of the work as it stands and proposals for future directions that could lead to a more comprehensive understanding of the organelle organization phenotype and bacterial actin behavior.

Despite their diminutive size and seemingly simple construction, bacteria lead remarkably complex lives. In order to fulfill their biological roles of growth and reproduction, they must integrate a wealth of information about their environment and, depending upon the suitability of the available conditions for survival, they can act to relocate themselves to more preferred locales. Doing so requires that bacteria be able to sense environmental stimuli and relay signals induced by those stimuli to various locomotive apparatuses. Once a cell has fulfilled its nutrient quota to support replication, cell division can occur. Cell division is also intricately timed and regulated in bacterial cells, which are now known to possess intracellular organization, cytoskeletal features, and, in some species, compartmentalization. Therefore, division of a bacterial cell must coordinate disassembly, reassortment, and segregation of these cell biological features. In this work, I investigate the connection between the cell cycle and bacterial organelle formation in the magnetotactic bacterium Magnetospirillum magneticum AMB-1. Magnetotactic bacteria, including AMB-1, are defined by their ability to synthesize chains of intracellular membrane-bounded magnetic minerals, which facilitate bacterial alignments with and responses to geomagnetic fields. To determine the role of the cell cycle in governing the production of these bacterial organelles, called magnetosomes, I disrupted homologs of regulatory factors known to control the progression of the cell cycle as well as polar organelle development in related Alphaproteobacteria. Surprisingly, mutants in the CtrA regulatory pathway were viable, indicating alternative mechanisms of cell cycle progression in AMB-1; in addition, magnetosome formation was also unaffected. Notably, motility was the only feature of AMB-1 disrupted by the CtrA pathway mutations. While subsequent studies to probe upstream regulators of motility in AMB-1 failed to yield additional insight, my results suggest a terminal role for CtrA in the transcription of flagellar biosynthesis genes. This role appears to be ancestral in the Alphaproteobacteria. Further, I have developed protocols which should enable future investigations of the cell cycle in AMB-1 and the temporal changes in gene expression which allow its progression. Preliminary studies indicate that genes involved in signal transduction and possibly magnetosome membrane formation vary their expression throughout the AMB-1 cell cycle. Continued investigation of the connections between the CtrA pathway, magnetosome gene expression, and the cell cycle may elucidate regulation of motility in magnetotactic bacteria and illustrate novel mechanisms of cell cycle progression in these unique organisms.

Bacteria make a wide variety of organelles to assist in specific cellular functions. How these organelles are assembled is poorly understood. One model for prokaryotic organelle biogenesis is the magnetosome, a lipid-bound compartment that contains a magnetic crystal allowing magnetotactic bacteria (MTB) to navigate along magnetic field lines in their environments. To create magnetosomes, MTB localizes many proteins specifically to the developing magnetosome membrane. How these proteins are sorted is an active area of research. This work explores magnetosome formation and investigates the sorting of a class of magnetosome proteins that are only localized to magnetosomes under magnetite-forming conditions. The first chapter of this dissertation, an unpublished review article, introduces the topic of magnetosome assembly in magnetotactic bacteria. It explores the selection of Magnetospirillum magneticum AMB-1 and Magnetospirillum gryphiswaldense MSR-1 as model species for studying magnetosome formation, including early work that created genetic tools enabling molecular characterization. In addition, studies are described that used these tools to identify proteins critical to magnetosome formation. Chapter 1 also describes a working model of magnetosome biogenesis and identifies areas where more research is needed to fill in our understanding of this complex process. The second chapter of this dissertation, an unpublished primary research article, investigates the localization of magnetite shaping protein Mms6 in AMB-1. Mms6 had previously been shown to localize to magnetosomes after iron is added to iron-starved cells. This change in localization suggested that either new Mms6 was produced and targeted to magnetosomes upon addition of iron, or pre-existing Mms6 was able to relocalize after iron addition. Using pulse-chase analysis combined with microscopy, we determine that pre-synthesized Mms6 in the cytoplasm is relocalized to magnetosomes in response to environmental cues. Our findings identify several magnetosome proteins and Mms6 protein domains critical to this dynamic localization behavior. The third chapter of this dissertation, an unpublished primary research article, identifies another protein, MamD, that is also sorted based on environmental conditions. MamD is a magnetosome protein that has been shown to bind magnetite crystal and inhibit magnetosome membrane growth. We use fluorescence microscopy to show that MamD, like Mms6, only localizes to magnetosomes when biomineralization of magnetite crystal is possible. In addition, we show that MamD magnetosome localization requires several of the same magnetosome proteins required for Mms6 localization. Our results suggest that there may be a step in magnetosome assembly where a specific cohort of proteins is conditionally recruited to assist in the nucleation and development of magnetite crystal.

The ability of living organisms to transform inorganic elements into insoluble crystalline structures is an underexplored theme in biology. A group of aquatic bacteria, called magnetotactic bacteria, produces chains of nanometer-sized magnetic crystals within their cells, allowing them to align in the earth’s geomagnetic fields. Understanding the mechanism for this process has become increasingly important as the demand for customized nanoparticles in medical and industrial applications has grown. In particular, the protein factors required to transform soluble iron into magnetite (Fe3O4) are likely to contain novel mechanisms for manipulating insoluble inorganic compounds.

This thesis describes the biochemical and genetic features of two such factors, MamO and MamE. A historical account is provided describing the discovery of magnetotactic bacteria, the development of Magnetospirillum magneticum AMB-1 as a model system and the identification of specific genes required to produce magnetite. This previous body of work led to the identification of MamO and MamE as predicted trypsin-like proteases that are required for biomineralization in AMB-1. Genetic, biochemical and structural studies reported here showed that the MamO protease domain is catalytically inactive and incapable of serine protease activity. Instead it has a novel di-histidine motif that participates in direct binding to transition metals. This motif is required for biomineralization in vivo, confirming that the MamO protein is a repurposed trypsin-like scaffold that promotes magnetite nucleation by binding directly to iron precursor atoms. Genomic and phylogenetic analysis of related serine proteases in other magnetotactic bacteria showed that the repurposing of trypsin-like proteases has occurred numerous times independently during the evolution of magnetosome formation. Also described is the observation that three biomineralization factors, MamE, MamO and MamP are proteolytically processed in AMB-1. MamE and MamO are both required for these proteolytic events, as are the predicted catalytic residues from MamE. However, consistent with its newly assigned pseudo-protease classification, the predicted MamO active site is dispensable. This suggested that MamE directly processes these targets in a manner that requires MamO. The proteolytic activity of MamE was reconstituted in vitro with a recombinant form of the protein. MamE cleaved a custom peptide substrate based on an in vivo cleavage site in MamO with positive cooperativity, and its auto-proteolytic activity could be stimulated by both substrates and peptides that bind to its regulatory domains. These enzymatic properties suggested that a switch-like regulatory mechanism modulated MamE-dependent proteolysis during biomineralization. This regulatory paradigm was confirmed by showing that both catalytically inactive and constitutively active alleles of mamE caused severe biomineralization defects in vivo.

Although the genes required for biomineralization were known previously, the molecular mechanisms by which each protein promotes magnetite synthesis had not been explored. The results of these studies define biochemical functions for two of the four factors required for magnetite nucleation in AMB-1. Furthermore, describing the evolutionary repurposing of a trypsin scaffold along with the phylogenetic description of its evolutionary history add broad biological interest to magnetosome research.

Though small in size, there is a growing appreciation for the complex ultrastructure of bacteria and archaea. This complexity and beauty is exemplified by the diverse protein- and lipid-bounded organelles that have been discovered. The first chapter of this dissertation, a published review article, introduces different lipid-bounded organelles that have been found in bacteria and archaea. The best-studied lipid-bounded organelles in bacteria are the magnetosomes of magnetotactic bacteria. This chapter discusses, in depth, the mechanism of magnetosome formation in two Magnetospirillum spp. that make cubooctahedral-shaped magnetite crystals within the magnetosome lumen. This chapter also discusses what is known about other, more mysterious, organelles, including bullet- shaped magnetosomes, anammoxosomes, and nucleus-like organelles in archaea.

Tools for genome editing are a major limiting factor when attempting to elucidate the structure and function of organelles. As such, there are few model systems for studying organelle formation. The second chapter of this dissertation, a published primary research article, describes a method for genome editing in Desulfovibrio magneticus RS-1. This work is the first example of gene editing for an anaerobic bacterium that makes bullet- shaped magnetosomes and marks a major step in magnetosome research.

In addition to making magnetosomes, D. magneticus makes ferrosomes, which are membrane-bounded organelles that contain iron, oxygen, and phosphorus. Ferrosomes were discovered serendipitously when former Komeili lab postdoctoral scholar, Dr. Meghan Byrne, observed that D. magneticus cells transitioning out of iron starvation are full of electron-dense granules, now named ferrosomes. The third chapter of this dissertation uncovers the genetic basis of ferrosomes. Using the genetic method we developed for D. magneticus, we show that ferrosomes require a set of genes that encode proteins associated with isolated ferrosomes. In addition to D. magneticus, diverse bacteria, and perhaps archaea, require a similar set of genes to make ferrosomes. Finally, we show that ferrosomes likely have an important role in iron homeostasis during anaerobic metabolism. Future research on bullet-shaped magnetosomes and ferrosome formation, function, and regulation are introduced in the final chapter of this dissertation.