investigating the mechanism of escherichia coli min protein dynamics

Summary 37 Functionality of tagged proteins in vivo 46 Preparation of phospholipid vesicles 49 Functionality of proteins used in this study 52 MinD associates with phospholipid membranes

INVESTIGATING THE MECHANISM OF

ESCHERICHIA COLI MIN PROTEIN DYNAMICS

by

Laura L Lackner

Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy

Advisor: Piet de Boer, PhD

Department of Molecular Biology and Microbiology CASE WESTERN RESERVE UNIVERSITY

January 2006

UMI Number: 3193524

3193524 2006

UMI Microform Copyright

All rights reserved This microform edition is protected against unauthorized copying under Title 17, United States Code.

ProQuest Information and Learning Company

300 North Zeeb Road P.O Box 1346 Ann Arbor, MI 48106-1346

by ProQuest Information and Learning Company

CASE WESTERN RESERVE UNIVERSITY

SCHOOL OF GRADUATE STUDIES

We hereby approve the dissertation of

*We also certify that written approval has been obtained for any

proprietary material contained therein

Min system dynamics spatially regulate the activity of MinC 27

Chapter 2: ATP-dependent interactions between Escherichia coli

Summary 37

Functionality of tagged proteins in vivo 46

Preparation of phospholipid vesicles 49

Functionality of proteins used in this study 52

MinD associates with phospholipid membranes in an ATP-

Cooperative binding of MinD to phospholipid vesicles 57 MinE stimulates dissociation of MinD from phospholipid

Role of nucleotide hydrolysis in MinD/MinE-dissociation

MinD-dependent recruitment of the division inhibitor MinC

Chapter 3: Roles of MinE domains in the regulation of Min

Purification of MinE deletion mutants 115

DMinE is necessary and sufficient to stimulate the ATPase

DMinE is necessary and sufficient to stimulate dissociation

of MinD and MinC from the phospholipid membrane 116 Affinity of DMinE for the phospholipid membrane 119

DMinE removes Gfp-MinD from the membrane in vivo, but TSMinE is required to efficiently establish MinD

Discussion 129

Chapter 4: In vitro and in vivo characterization of the MinD

Lipid preparation, sedimentation and ATPase assays 153

Yeast strains, plasmids, and two-hybrid analyses 155

MinDK16R forms spiral-like structures in the presence

Discussion 164

The role of ATP in regulating the interactions of MinD with its

How does MinE stimulate the ATPase activity of MinD? 186

What role does MinE play in the formation of Min spirals? 192

Is the association of MinE with the membrane self-enhancing? 194 Are additional proteins required for MinDE dynamics? 195

A conserved mechanism for the oscillation of proteins in bacteria 197

Appendix A: Supplementary Data – FRET studies suggest that the

MinD-MinD and MinD-MinE interactions are membrane dependent 201

Labeling proteins with AlexaFluor dyes 205

Fluorescent labeling of MinD and MinE 207

FRET between A488-MinD and A594-MinD is

FRET between A594-MinD and A488-MinE is

List of Tables 2-1 In vivo functionality of His-tagged Min proteins 78

2-2 ATP- and Mg++-dependent association of H-MinD with

3-1 Location of Gfp-MinD when co-expressed with the MinE fusions 136

3-2 Cellular distribution and oscillation cycle time of MinE-Gfp deletion

mutants and Gfp-MinD in the presence of native MinCDE 137

4-1 In vivo and in vitro properties of MinD and MinDK16R 171

List of Figures

1-2 Model of MinCDE action in Z-ring placement 34

2-1 Pole-to-pole oscillation of Gfp-MinC-H in the presence of H-MinD

2-3 ATP-dependent association of MinD with phospholipid vesicles 84

2-4 Self-enhanced association of MinD with phospholipid vesicles 85

2-5 MinE-stimulated dissociation of MinD from vesicles 86

2-6 Stabilization of a MinE-MinD-membrane complex in the presence

2-7 Recruitment of MinE to MinD-decorated vesicles 88

2-8 Recruitment of MinC to MinD-decorated vesicles 89

2-9 Microscopy of phospholipid vesicles showing MinD-mediated

recruitment and MinE-stimulated release of MinC 91

2-10 MinE-stimulated release of both MinC and MinD from phospholipid

2-11 MinE stimulates the specific release of Gfp-MinC-H from

3-1 MinE derivatives and their in vitro properties 138

3-2 Gel filtration analysis of purified MinE-M proteins 139

3-3 MinE1-33-M stimulates the ATPase activity of H-MinD in the

3-4 Min1-33-M stimulates dissociation of H-MinD and Gfp-MinC-H from

3-5 Nucleotide hydrolysis requirements for MinE1-33-M stimulated

release of H-MinD and GFP-MinC-H from phospholipid vesicles 142

3-6 The N-terminus of MinE has affinity for phospholipid vesicles 143

3-7 Localization of GFP-MinD in the presence of the MinE deletion

3-8 Cellular location of Gfp-MinD in the presence of increasing levels

3-9 Localization of MinE deletion mutants in the presence of MinD

and co-localization of MinE deletion mutants and MinD 146

4-1 The association H-MinDK16R with phospholipid vesicles is

4-2 Gel filtration analysis of MinD and MinDK16R 174

4-3 MinE stimulates the release of MinC but not MinDK16R from

4-4 MinE does not stimulate the ATPase activity of MinDK16R 176

4-5 MinE alters the cellular distribution of Gfp-MinDK16R 177

4-6 Localization of Gfp-MinDK16R in the presence of MinE deletion

4-7 Localization of MinE-Gfp and MinE1-33-Gfp in the presence

4-8 Localization of MreB and Gfp-MinDK16R 180

4-9 Localization of Gfp-MinDK16R in the absence of MreB 181

B-1 Microscopy of phospholipid vesicles showing the

MinD.ATP-mediated recruitment and MinE-stimulated release

I would like to thank my committee members Dr Lloyd Culp, Dr Jonatha Gott, Dr Pieter de Haseth, Dr Sandra Lemmon, Dr Phil Rather, and Dr Patrick Viollier for their advice and support I would also like to thank other members of the Department of Molecular Biology and Microbiology for help and suggestions during the course of my research

Thanks to the Cell and Molecular Biology Training Grant and Jo Ann Wise for financial support

For their love and encouragement, I would like to thank my parents, Herman and Sue Lackner, and my sister, Mandy I am also thankful for the wonderful friends I have made here at Case Finally, I wish to thank Chris for his love and support and for his patience during the more stressful and crazy times

Investigating the Mechanism of Escherichia coli Min Protein Dynamics

Abstract

by Laura L Lackner

In Escherichia coli the pole-to-pole oscillation of the MinC division inhibitor is

required for proper placement of the division machinery The membrane

association/dissociation cycle of MinC is driven by the MinD ATPase and the MinE topological specificity factor, which themselves undergo a coupled

oscillatory localization cycle During this cycle, MinD and a portion of MinE assemble on the membrane along one cell half in a pattern resembling a test tube (the MinD/E tube), while another portion of MinE assembles at the rim of this tube forming a ring-like structure (E-ring)

To understand the biochemical mechanisms behind Min protein dynamics,

we investigated the interactions of purified Min proteins with ATP and

phospholipid vesicles We found that the ATP-bound form of MinD binds

phospholipid vesicles in a cooperative fashion and recruits both MinC and MinE

to the vesicles In addition, we found that MinE stimulates the dissociation of MinC, MinD, and itself from the vesicles Thus, ATP and MinE play critical roles

in regulating the association and dissociation, respectively, of the Min proteins and the membrane

MinE has two known functional domains: the N-terminal anti-MinCD domain (DMinE) and the C-terminal topological specificity and dimerization domain (TSMinE) We investigated the contributions of each domain of MinE in the regulation of Min protein-membrane dissociation We found that DMinE is

necessary and sufficient to stimulate dissociation of the Min proteins from the membrane However, TSMinE is required to establish Min protein oscillation in a majority of cells In addition, TSMinE is also required for E-ring formation

These results suggest that TSMinE and perhaps E-ring formation play a critical role in Min protein oscillation by helping to initiate and/or sustain Min protein dynamics

Characterization of the MinD mutant MinDK16R revealed an additional role for MinE While MinE does not induce MinDK16R oscillation, MinE does promote the accumulation of MinDK16R into static membrane-associated spiral-like

structures We propose that these structures may represent a stage in the dynamic Min protein oscillation cycle, therefore, suggesting that MinE has at least two roles in Min protein dynamics, regulating Min protein-membrane

dissociation and regulating Min protein spiral formation

Chapter 1 Introduction

Introduction

Cell division, the partitioning of a mother cell into two daughter cells, is an essential biological process However, the molecular details of this process are

poorly understood, even in the unicellular organism Escherichia coli

In E coli, cell division occurs by the coordinated constriction of the three cell

envelope layers that make up the cell wall This event takes place quite

accurately at mid-cell following the segregation of the newly replicated sister chromosomes to opposite cell halves, which ensures that each daughter cell receives a complete copy of the genome Many players involved in the

processes of constriction and division site selection have been identified and their roles in these processes will be discussed below

Division proteins and the septal ring

A number of proteins involved in cell division were initially identified by Hirota and co-workers who isolated thermosensitive mutants that failed to divide

at elevated temperatures (Hirota et al., 1968) As a result of continued growth

in the absence of division, these mutants formed long filaments and were, therefore, given the designation filamentous temperature sensitive, fts

Additional cell division proteins have since been identified, and to date, 10 proteins are known to be required for division: FtsZ, A, K, Q, L, B, W, I, N and ZipA (Addinall and Holland, 2002; Buddelmeijer and Beckwith, 2002; Errington

et al., 2003; Margolin, 2000; Rothfield et al., 1999)

All 10 proteins assemble at the site of division into a multi-protein complex called the septal ring (Fig 1-1) Formation of this complex is thought to initiate with the assembly of FtsZ into a ring-like structure, the Z-ring, which requires the activity of at least one of the two essential FtsZ binding proteins, ZipA or FtsA (Hale and de Boer, 1999; Pichoff and Lutkenhaus, 2002) Following addition of both ZipA and FtsA to the Z-ring, the remaining division proteins are recruited to the maturing septal ring in a defined, sequential order (see Fig 1-1) (Addinall and Holland, 2002; Buddelmeijer and Beckwith, 2002; Chen and Beckwith, 2001; Errington et al., 2003; Hale and de Boer, 2002; Margolin, 2000; Pichoff and Lutkenhaus, 2002; Rothfield et al., 1999) After assembly is

complete and, most likely, following an as of yet unknown cue to begin

constriction, the proteins of the septal ring function in concert to mediate the coordinated invagination of the three cell envelope layers: the inner membrane, the cell wall (the peptidoglycan layer), and the outer membrane Unfortunately, very little is known about the roles of many of the division proteins in this

process

FtsZ and the Z-ring

FtsZ is the most widely conserved of the division proteins It has been identified in almost all eubacteria, in many archaea, and in some organelles of prokaryotic origin, plastids and the mitochondria of primitive eukaryotes

(Errington et al., 2003; Gilson and Beech, 2001; Margolin, 2000; Osteryoung

and McAndrew, 2001; Rothfield et al., 1999) Similar to its function in E coli,

FtsZ most likely plays a critical role in the division of these organisms and organelles

In E coli, FtsZ, the most abundant of the division proteins (~15,000

molecules per cell), exhibits a dynamic cell cycle-dependent cellular distribution (Bi and Lutkenhaus, 1991; Den Blaauwen et al., 1999; Lu et al., 1998) Prior to the onset of division, cytoplasmic FtsZ assembles into a membrane-associated structure, the Z-ring The Z-ring remains associated with the leading edge of the invaginating septum during division and dissipates upon completion of division, at which point FtsZ is again cytoplasmic Interestingly, during its

existence, the Z-ring itself is highly dynamic Only 30% of FtsZ is incorporated into the ring at any given time (Stricker et al., 2002), and the FtsZ subunits within the ring are continually exchanged with the cytoplasmic pool of FtsZ, such that the Z-ring completely turns over approximately every 18s (Anderson

et al., 2004)

FtsZ is a homologue of the eukaryotic tubulins While the primary sequence

of FtsZ shows weak homology to tubulin, the crystal structure of FtsZ from

Methanococcus jannaschii and that of tubulin from bovine brain are remarkably

similar (Löwe, 1998; Nogales et al., 1998) In addition, FtsZ, like tubulin, is a GTPase that undergoes GTP-dependent polymerization forming polymers that closely resemble tubulin protofilaments (Bramhill and Thompson, 1994; de Boer

et al., 1992a; Erickson et al., 1996; Mukherjee et al., 1993; Mukherjee and Lutkenhaus, 1994) Polymerization of FtsZ is required for nucleotide hydrolysis

as the GTPase active site is comprised of amino acids from two adjacent

monomers (Wang and Lutkenhaus, 1993) In turn, GTP hydrolysis triggers changes in polymer conformation and/or polymer disassembly (Lu et al., 2000; Mukherjee and Lutkenhaus, 1998)

The Z-ring formed in vivo is most likely comprised of FtsZ protofilaments Based on the rate of Z-ring dynamics in vivo in combination with the kinetics of FtsZ assembly in vitro, it is estimated that protofilaments only ~1/15 the length

of the circumference of the cell can be formed during the time it takes the Z-ring

to completely turn over (Anderson et al., 2004; Chen and Erickson, 2005) Thus, the Z-ring is proposed to be an assemblage of many short protofilaments (Anderson et al., 2004), the arrangement of which is unknown

ZipA and FtsA

ZipA and FtsA are the earliest known recruits to the nascent Z-ring The proteins join the Z-ring independently of one another via direct interactions with

a conserved sequence at the extreme C-terminus of FtsZ (Hale and de Boer, 1997; Hale and de Boer, 1999; Hale et al., 2000; Haney et al., 2001; Ma and Margolin, 1999; Mosyak et al., 2000; Pichoff and Lutkenhaus, 2002; Wang et al., 1997) As mentioned above, the presence of either ZipA or FtsA is required for Z-ring assembly (Hale and de Boer, 1999; Pichoff and Lutkenhaus, 2002) Therefore, ZipA and FtsA appear to have a partially redundant role in Z-ring formation, as both proteins most likely act to stabilize the assembling structure

In contrast, both proteins are required for the recruitment of the remaining division components, implying that the proteins have distinct roles in septal ring

assembly (Hale and de Boer, 2002; Pichoff and Lutkenhaus, 2002)

Interestingly, a gain of function mutation in FtsA bypasses the requirement for ZipA in division suggesting that FtsA may play a more dominant role in the recruitment of downstream division components (Geissler et al., 2003)

ZipA is a poorly conserved division protein that was identified as a

FtsZ-interacting protein in E coli using Far Western analysis (Hale and de Boer,

1997; Margolin, 2000) ZipA is anchored to the inner membrane via an

N-terminal transmembrane domain that is followed by a highly charged region, a flexible Pro/Gln rich linker, and a globular FtsZ-binding domain (Hale and de Boer, 1997; Hale et al., 2000) In vitro, ZipA promotes the bundling of FtsZ protofilaments, and this bundling activity could act to stabilize the growing Z-ring

in vivo (Hale et al., 2000) Additionally, ZipA serves as a membrane anchor for FtsZ, which may provide support for the assembling Z-ring

FtsA is a highly conserved septal ring component (Margolin, 2000) The protein belongs to the actin/Hsp70/sugar kinase family of ATPases (Bork et al.,

1992), and the crystal structure of FtsA from Thermotoga maritima revealed that

FtsA was most similar to the actin subfamily (van Den Ent and Löwe, 2000) FtsA, like actin, has been shown to bind ATP; however, the ability of FtsA to form polymers following ATP binding or to hydrolyze the bound nucleotide has yet to be demonstrated (Sánchez et al., 1994) FtsA is peripherally tethered to the inner membrane via a C-terminal amphipathic helix, which is required for recruitment to the septal ring (Pichoff and Lutkenhaus, 2005) Thus similar to

ZipA, FtsA may contribute to Z-ring stability by serving as a membrane anchor providing support for the nascent Z-ring

FtsK

FtsK is a large polytopic membrane protein with two distinct functional domains (Begg et al., 1995; Liu et al., 1998) The N-terminal membrane

spanning domain (NFtsK) plays an essential but unknown role in division

(Draper et al., 1998) The large cytoplasmic C-terminal domain (CFtsK), which

is dispensable for division, functions in the final stages of chromosome

segregation, which include decatenation of topologically linked chromosomes and resolution of chromosome dimers The enzymes responsible for

decatenation and dimer resolution, TopoIV and XerC/D, respectively, are

spatially and temporally regulated by CFtsK, which ensures that these

processes and, consequently, the final stages of segregation are coordinated with cell division (Aussel et al., 2002; Espeli et al., 2003; Steiner et al., 1999) Additionally, CFtsK is an ATP-dependent DNA translocase and may, therefore, play an additional role in segregation by clearing any remaining DNA from the septal pore prior to completion of division (Aussel et al., 2002) FtsK, therefore, directly links chromosome segregation and cell division

Additional essential septal ring components

With exception of FtsI, very little is known about the functions of the

remaining essential division proteins FtsI, also known as penicillin binding

protein 3 (PBP3), is required for the synthesis of peptidoglycan at the site of constriction (Adam et al., 1997; Ishino and Matsuhashi, 1981) Two lines of evidence suggest that FtsW functions with FtsI in the synthesis of septal

peptidoglycan First, FtsW is a homolog of RodA, which is proposed to act with PBP2, a murein synthase in the same family as FtsI, in peptidogylcan synthesis during elongation (Ikeda et al., 1989) Second, the presence of FtsW within bacterial genomes is strongly correlated with the presence FtsI (Errington et al., 2003; Margolin, 2000)

The roles FtsQ, L, B, and N play in division, other than serving to recruit downstream septal ring components, are unknown (Buddelmeijer and Beckwith, 2002; Errington et al., 2003; Margolin, 2000)

Non-essential septal ring components

In addition to the essential division proteins that compose the septal ring, non-essential septal ring components have been identified: ZapA, FtsE and X, EnvC, and AmiC (Bernhardt and de Boer, 2004; Bernhardt and de Boer, 2003; Gueiros-Filho and Losick, 2002; Schmidt et al., 2004) (Fig 1-1) While these proteins are not required for division, they most likely aid in the efficacy of the division process

Like FtsA and ZipA, ZapA interacts directly with FtsZ (Gueiros-Filho and Losick, 2002) While a deletion of ZapA has no noticeable effect on division (Johnson et al., 2004), ZapA has been shown to bundle FtsZ protofilaments in

vitro suggesting ZapA may play a non-essential role in Z-ring stability and/or architecture (Gueiros-Filho and Losick, 2002; Low et al., 2004)

The roles of FtsE and X in division are unclear as the proteins are only required for division during growth in the absence of salt (Gill et al., 1986;

Schmidt et al., 2004) The proteins have homology to ABC transporters; FtsE resembles the ATP-binding cassette (ABC) component and FtsX resembles the membrane component (Schmidt et al., 2004) ABC transporters use energy from ATP to transport a variety of components into and out of the cell The salt requirement suggests FtsEX may import an ion (Gill et al., 1986); however, it has yet to be demonstrated what cargo the complex transports or even if the complex actually functions as a transporter

EnvC and AmiC are the only two septal ring components that reside entirely

in the periplasm (Bernhardt and de Boer, 2004; Bernhardt and de Boer, 2003) Both proteins are involved in some aspect of peptidoglycan breakdown, and both EnvC and AmiC mutants have defects in daughter cell separation

(Bernhardt and de Boer, 2004; Bernhardt and de Boer, 2003; Heidrich et al., 2001; Höltje and Heidrich, 2001) Thus, both proteins are proposed to function

in the cleavage of septal peptidoglycan to facilitate daughter cell separation

Spatial regulation of Z-ring assembly

Z-ring assembly initiates the formation of a mature, division competent septal ring, and therefore, the site of Z-ring assembly dictates where the cell will

divide In E coli, Z-ring assembly and, consequently, division occur quite

accurately at mid-cell (Trueba, 1982; Yu and Margolin, 1999) This ensures that each daughter receives a copy of the newly replicated and segregated chromosome The spatial regulation of Z-ring assembly is mediated by two partially redundant negative regulatory systems, nucleoid occlusion and the Min system These systems act in concert to restrict Z-ring assembly to mid-cell by inhibiting assembly at all other sites

Nucleoid occlusion

Nucleiod occlusion is a phenomenon in which Z-ring assembly is inhibited

on portions of the membrane in close vicinity to the nucleoid Thus, assembly is spatially restricted to the nucleoid free regions of the cell, the cell poles and mid-cell following nucleoid segregation (Mulder and Woldringh, 1989; Sun and Margolin, 1998; Sun and Margolin, 2001; Sun and Margolin, 2004; Yu and Margolin, 1999) These regions will be referred to as potential division sites (PDSs) The action of the Min system blocks Z-ring assembly at the polar PDSs (discussed below) leaving only the mid-cell site available The availability

of this site is regulated with the cell cycle as this region only becomes DNA-free following chromosome replication and segregation (Draper and Gober, 2002) Therefore, nucleoid occlusion may also temporally regulate Z-ring assembly by preventing assembly at the mid-cell PDS prior to chromosome segregation The molecular mechanism underlying nucleoid occlusion is poorly

understood The nucleoid and its associated factors have been proposed to block Z-ring formation indirectly as a result of steric hindrance (Woldringh et al.,

1991) Recent studies, however, suggest that nucleoid-associated proteins may directly interfere with Z-ring formation, although the two mechanisms are

not mutually exclusive The DNA-binding proteins Noc in Bacillus subtilus and SlmA in E coli were recently identified as nucleoid occlusion factors, and

accordingly, these proteins are required to block Z-ring assembly in the area surrounding the nucleoid (Bernhardt and de Boer, 2005; Wu and Errington, 1994) Both proteins are proposed to interact with FtsZ directly to inhibit Z-ring formation, although the mechanism of inhibition is unknown When either Noc

or SlmA mutants are combined with Min system mutants, the spatial regulation

of Z-ring assembly is lost Consequently, multiple, randomly distributed Z-rings are formed, none of which can recruit enough FtsZ and/or downstream division components to form a productive septal ring, and, as a result, division is

blocked (Bernhardt and de Boer, 2005; Wu and Errington, 1994)

The Min system

Spatial regulation of Z-ring assembly by the Min system is independent of nucleoid occlusion, which is evident in annucleate cells produced by

chromosome segregation mutants In these cells, Z-rings assemble at or near mid-cell suggesting that the Min system alone is sufficient to regulate the proper placement of the Z-ring (Sun et al., 1998; Yu and Margolin, 1999) In contrast,

in cells that lack a functional Min system, nucleoid occlusion itself is not

sufficient to direct Z-ring assembly to mid-cell In these cells, Z-ring assembly and, consequently, division occur at any one of the three PDSs dictated by

nucleoid occlusion If division occurs at a polar PDS, a chromosome-less

minicell and a multinucleate filament are produced (Adler et al., 1967; de Boer

et al., 1989)

The first minicell mutant in E coli was isolated in 1967 by Alder and

colleagues (Adler et al., 1967) The mutations responsible for the minicell

phenotype were mapped to the minB locus (Davie et al., 1984), which was later cloned and characterized (de Boer et al., 1988; de Boer et al., 1989) The minB

locus encodes three proteins, MinC, MinD, and MinE, all of which are required for proper Min system function

MinC, the division inhibitor

MinC is a 231 amino acid protein comprised of two domains of roughly equal size (Cordell et al., 2001; de Boer et al., 1989; Hu and Lutkenhaus, 2000) The N-terminal domain interacts with FtsZ and is required and sufficient to inhibit FtsZ polymerization in vitro and Z-ring formation in vivo (Hu and Lutkenhaus, 1999; Hu and Lutkenhaus, 2000; Johnson et al., 2002; Pichoff and Lutkenhaus, 2001) The mechanistic basis of this inhibition is unknown The C-terminal domain is required for homodimerization of the protein and for interaction with activators of MinC function, MinD and DicB (Cordell et al., 2001; Hu and

Lutkenhaus, 2000; Johnson et al., 2002; Szeto et al., 2001b) The DicB protein

is encoded on the cryptic prophage Kim, and under normal conditions,

expression of DicB is actively repressed (Béjar et al., 1988; Cam et al., 1988)

In the absence of these activators, MinC blocks cell division only when present

at a level at least 25-fold greater than normal In the presence of either

activator, physiological levels of MinC are sufficient to inhibit division (de Boer

et al., 1990; de Boer et al., 1992b)

inhibition First, MinD recruits MinC to the membrane, and second, the

MinC/MinD complex targets nascent septal ring assemblies on the membrane (Johnson et al., 2002; Johnson et al., 2004) As a consequence, MinC is

concentrated in the vicinity of the ring where the protein can interfere with ring assembly and/or integrity MinD also interacts with MinE, and this

Z-interaction imparts topological specificity to MinCD, as MinE relieves the

inhibitory activity of the complex specifically at mid-cell (de Boer et al., 1989) MinD belongs to a family of ATPases with deviant Walker A motifs This family also includes the ParA-type proteins involved in chromosome and

plasmid partitioning and NifH, a component of the nitrogenase complex (de Boer et al., 1991; Draper and Gober, 2002; Gerdes et al., 2000; Lutkenhaus and Sundaramoorthy, 2003; Motallebi-Veshareh et al., 1990) The members of this family show significant conservation in both the sequence and arrangement

of the three motifs involved in nucleotide binding and hydrolysis: the Walker A

(P-loop), A’, and B motifs (Motallebi-Veshareh et al., 1990) The presence of two lysines in the Walker A motif is the signature feature of this family (Koonin, 1993) The first lysine, known as the signature lysine, is unique to this family while the second is common to all Walker A motifs In addition to sequence similarities, members of this family also share a high degree of structural

similarity as revealed by the crystal structures of MinD-homologues from

Archaeoglobus fulgidus, Pyrococcus furiosus, and Pyrococcus horikoshii, the

ParA-like protein Soj from Thermus thermophilus, and NifH from Azobacter

vinelandii (Cordell and Lowe, 2001; Georgiadis et al., 1992; Hayashi et al.,

2001; Leonard et al., 2005; Sakai et al., 2001) The significance of these

similarities will be discussed in Chapter 5

MinE

MinE is an 88 amino acid protein with two known functional domains The N-terminal domain (DMinE, amino acids 1-33) is required and sufficient to

interact with MinD and to counteract MinCD-mediated division inhibition (Huang

et al., 1996; Ma et al., 2003; Pichoff et al., 1995; Raskin and de Boer, 1997; Zhao et al., 1995) The C-terminal topological specificity domain (TSMinE, amino acids 34-88) is required to suppress the inhibitory activities of MinCD specifically at mid-cell Additionally, this domain mediates homodimerization of the protein (King et al., 1999; King et al., 2000; Pichoff et al., 1995; Raskin and

de Boer, 1997; Shih et al., 2002; Zhang et al., 1998; Zhao et al., 1995) The structure of the C-terminal domain as a dimer has been solved (King et al.,

2000) and reveals that the protein dimerizes in an anti-parallel fashion The terminal domain is not shown in the structure but is predicted to be a nascent helix (King et al., 1999)

N-Min system dynamics spatially regulate the activity of N-MinC

The site of Z-ring assembly and, therefore, division is determined by the cellular distribution of MinC, which is regulated by both MinD and MinE In addition, MinD and MinE each influence the cellular distribution of the other (Fig 1-2) In MinD- cells, MinC and MinE are cytoplasmic, and MinC is unable

to efficiently inhibit Z-ring assembly (Hu and Lutkenhaus, 1999; Raskin and de Boer, 1997; Raskin and de Boer, 1999a) As a result, cells show a minicelling division phenotype (Min-) where cells frequently divide close to either cell pole (de Boer et al., 1989) In MinE- cells, MinC and MinD assemble along the entire membrane This static, uniform distribution of MinC blocks Z-ring assembly at all sites resulting in the formation of long, non-septate filamentous cells (Sep-) (de Boer et al., 1989; Hu and Lutkenhaus, 1999; Raskin and de Boer, 1999a; Raskin and de Boer, 1999b; Rowland et al., 2000) In cells expressing all three Min proteins, a WT scenario, MinC undergoes a rapid and dynamic localization cycle in which the protein oscillates from one cell pole to the other every 20-30

s (Hu and Lutkenhaus, 1999; Raskin and de Boer, 1999a) As a result of this pole-to-pole oscillation, the time-averaged concentration of the division inhibitor

is greatest at the cell poles and lowest at mid-cell, and this concentration

differential is proposed to direct Z-ring assembly to mid-cell (Hale et al., 2001;

Howard et al., 2001; Huang et al., 2003; Kruse, 2002; Meinhardt and de Boer, 2001) This dynamic distribution of MinC is driven by MinD and MinE, which themselves undergo a coupled oscillatory localization cycle (Fu et al., 2001; Hale et al., 2001; Hu and Lutkenhaus, 1999; Raskin and de Boer, 1999a; Shih

et al., 2002) (Fig 1-3) At the start of this cycle, MinD and a portion of MinE assemble on the membrane along one half of the cell in a pattern resembling a test tube (MinD/E tube), while another portion of MinE assembles at the rim of this tube forming a ring (E-ring) The MinD/E tube undergoes a continuous cycle

of disassembly from the membrane at one cell end and concurrent reassembly

on the membrane at the opposite cell end The E-ring closely follows the

location of the MinD/E tube, always appearing to be associated with its rim (for

a more detailed explanation of MinDE dynamics see Fig 1-3 and the

accompanying legend) MinC plays no role in the mechanism of oscillation, and the cellular location of MinC follows that of MinD (Hale et al., 2001; Hu and Lutkenhaus, 1999; Raskin and de Boer, 1999a; Raskin and de Boer, 1999b; Rowland et al., 2000)

While the MinD/E tube and E-ring do not appear to have an underlying structure in live cells analyzed by conventional 2D fluorescence microscopy, spiral-like accumulations of the proteins in what would represent the MinD/E tube and E-ring have been observed in fixed cells analyzed by 3D

deconvolution microscopy This observation suggests that the bulk of Min protein dynamics occurs along spiral tracks within the cell (Shih et al., 2003)

Conservation of the Min proteins

MinD is well conserved among many bacteria and archaea and is also found

in a number of eukaryotic chloroplasts (Colletti et al., 2000; Gerard et al., 1998; Margolin, 2000) Interestingly, this high degree of conservation is not shared with MinC and/or MinE Many organisms that posses clear MinD homologues lack MinC and/or MinE suggesting that the mechanism by which MinD regulates division site placement is not conserved in all organisms (Margolin, 2000;

Rothfield et al., 1999) For example, B subtilis has both MinC and MinD but lacks MinE In B subtilis, DivIVA imparts topological specificity to MinCD-

mediated division inhibition in a manner very different from that of MinE (Cha and Stewart, 1997) DivIVA, which stably associates with the cell poles, recruits MinCD to the poles and, thereby, spatially restricts the inhibitory activity of the complex to the polar PDSs (Marston and Errington, 1999b; Marston et al., 1998) Therefore, Min protein oscillation is not required to spatially regulate Z-ring assembly in all organisms

Summary of this work

It is clear that Min protein dynamics are essential for proper placement of

the Z-ring in E coli (Fig 1-2) Prior to starting the work presented here, very

little was known about the molecular mechanism underlying these dynamics Localization studies suggested that the dynamics of MinD involved a reversible association of the protein with the membrane (Raskin and de Boer, 1999b; Rowland et al., 2000) Furthermore, these studies suggested a role for MinE in

modulating this reversible interaction, as indicated by the findings that the oscillation of MinD is dependent on MinE, the frequency of oscillation is

inversely related to the cellular MinD to MinE ratio, and the movements of MinE appear to be physically linked to those of MinD (Fu et al., 2001; Hale et al., 2001; Raskin and de Boer, 1999a; Shih et al., 2002) These ideas were

supported by the findings that MinE stimulated the ATPase activity of MinD in a phospholipid-dependent fashion and that this stimulation appeared to be

required for oscillation These findings suggested that MinD.ATP might interact directly with the membrane and that MinE-stimulated nucleotide hydrolysis might interfere with this interaction (Hu and Lutkenhaus, 2001) Mathematical models of MinDE dynamics provided additional support In these models, the assumption that MinE forces the release MinD from the membrane is critical for the establishment of oscillation (Howard et al., 2001; Kruse, 2002; Meinhardt and de Boer, 2001)

With the aim of gaining a better understanding of the roles ATP and MinE play in the reversible association of the Min proteins with the membrane, we explored the interactions between the Min proteins and phospholipid vesicles in vitro (Chapter 2) We found that, in contrast to the ADP-bound form of MinD, the ATP-bound form binds phospholipid vesicles in a cooperative fashion and recruits both MinC and MinE to these vesicles We further found that MinE stimulates the dissociation of both MinC and MinD from the vesicles, and that the MinE-stimulated dissociation of MinD requires nucleotide hydrolysis These results demonstrate that ATP and MinE play critical roles in regulating the

association and dissociation, respectively, of the Min proteins with the

membrane

To further characterize the role of MinE in Min protein dynamics, we

investigated the contributions of each domain of MinE in the regulation of Min protein-membrane dissociation in vitro and in vivo (Chapter 3) We found that the monomeric DMinE domain is both required and sufficient to stimulate the ATPase activity of MinD and to cause dissociation of Min protein complexes from phospholipid vesicles in vitro DMinE also removed MinD from the

membrane in vivo, and was capable of inducing MinDE oscillation in a minor fraction of cells in a population and did so without forming an E-ring The results suggest that neither TSMinE-mediated dimerization of MinE, nor formation of an E-ring, is needed to promote Min protein oscillation per se, but that TSMinE contributes critically to the robustness of oscillation by helping to initiate and/or sustain protein dynamics in the majority of cells in a given population

In a previous study, we identified a MinD Walker A motif mutant, MinDK16R, that interacts with the membrane and MinE, but interestingly, does not oscillate (Raskin, 2001) To try to gain an understanding of why MinE does not promote the oscillation of MinDK16R, we further characterized the K16R mutation in vitro and in vivo (Chapter 4) We found that while MinDK16R can interact with the membrane, the interaction is no longer regulated by nucleotide or MinE

Therefore, the K16R mutation abrogates the reversible association of MinD with the membrane that is required for oscillation of the protein Interestingly, while MinE does not induce MinDK16R oscillation, MinE does promote the

accumulation of MinDK16R into static membrane-associated spiral-like structures We propose that these structures may represent a stage in the highly dynamic Min protein oscillation cycle

Fig 1-1 Organization of the septal ring

A, Model showing the order with which the division proteins are recruited to the

maturing septal ring The essential division proteins are shown in black and the essential division proteins in red

non-Adapted from (Hale and de Boer, 2002)

B, The membrane topologies of the septal ring components These components

include: FtsZ (Z), FtsA (A), ZipA (ZA), ZapA (P), FtsE (E), FtsX (X), FtsK (K), FtsQ (Q), FtsL (L), FtsB (B), FtsW (W), FtsI (I), FtsN (N), AmiC (C) and EnvC (V)

Adapted from Errington et al., 2003 Courtesy of J.E Johnson

Fig 1-2 Model of MinCDE action in Z-ring placement

(+) denotes a potential division site (PDS) that is available for Z-ring formation, and (-)

denotes a PDS where Z-ring formation is inhibited by MinCD See text on page 27-28

for more detail

Fig 1-3 Model of MinDE dynamics

At the start of the cycle, MinD and a portion of MinE assemble on the membrane along one half of the cell in a pattern resembling a test tube (MinD/E tube), while another portion of MinE assembles at the rim of this tube forming a ring (E-ring) The rim of the D/E-tube along with the E-ring retracts back towards the pole and a portion of MinD becomes cytoplasmic When the rim of the D/E-tube reaches the cell pole, the E-ring dissipates and the bulk of MinE also appears cytoplasmic Meanwhile, assembly of a new tube begins on the membrane of the opposite cell pole Again, a portion of MinE accumulates with MinD in the new tube while another concentrates at its rim, forming a new E-ring The new MinD and MinE assemblies retract towards the pole, dissipate upon reaching it, form new membrane-associated assemblies on the initial pole, and a new cycle starts A full cycle takes on average ~40-50 seconds

Chapter 2

ATP-dependent interactions between Escherichia coli Min proteins

and the phospholipid membrane in vitro

SUMMARY

Proper placement of the division apparatus in E.coli requires pole-to-pole

oscillation of the MinC division inhibitor MinC dynamics involves a membrane association/dissociation cycle that is driven by the activities of the MinD ATPase and the MinE topological specificity factor, which themselves undergo coupled oscillatory localization cycles To understand the biochemical mechanisms underlying Min protein dynamics we studied the interactions of purified Min proteins with phospholipid vesicles, and the role of ATP in these interactions

We show that: i) The ATP-bound form of MinD (MinD.ATP) readily associates with phospholipid vesicles in the presence of Mg++, whereas the ADP-bound form (MinD.ADP) does not; ii) MinD.ATP binds membrane in a self-enhancing fashion; iii) Both MinC and MinE can be recruited to MinD.ATP-decorated

vesicles; iv) MinE stimulates dissociation of MinD.ATP from the membrane in a process requiring hydrolysis of the nucleotide; v) MinE stimulates dissociation of MinC from MinD.ATP-membrane complexes, even when ATP hydrolysis is blocked

The results support and extend recent work by Hu et al (Hu, Z., E P Gogol, and J Lutkenhaus 2002 Proc Natl Acad Sci USA 99: 6761-6766), and support models of protein oscillation wherein MinE induces Min protein dynamics by stimulating the conversion of the membrane-bound form of MinD (MinD.ATP) to the cytoplasmic form (MinD.ADP) The results also indicate that MinE-stimulated dissociation of MinC from the MinC-MinD.ATP-membrane complex can, and may, occur prior to hydrolysis of the nucleotide