University of Alberta

Investigating the mechanism of FinO-mediated inhibition of bacterial

conjugation

by

David Cameron Arthur

A thesis submitted to the Faculty of Graduate Studies and Research in partial fulfillment of the requirements for the degree of

Doctor of Philosophy

Department of Biochemistry

Edmonton, Alberta Spring 2009 Library and Archives Bibliotheque et 1*1 Canada Archives Canada Published Heritage Direction du Branch Patrimoine de I'edition 395 Wellington Street 395, rue Wellington OttawaONK1A0N4 Ottawa ON K1A 0N4 Canada Canada

Your file Vote reference ISBN: 978-0-494-55313-8 Our file Notre reference ISBN: 978-0-494-55313-8

NOTICE: AVIS:

The author has granted a non­ L'auteur a accorde une licence non exclusive exclusive license allowing Library and permettant a la Bibliotheque et Archives Archives Canada to reproduce, Canada de reproduire, publier, archiver, publish, archive, preserve, conserve, sauvegarder, conserver, transmettre au public communicate to the public by par telecommunication ou par Nnternet, prefer, telecommunication or on the Internet, distribuer et vendre des theses partout dans le loan, distribute and sell theses monde, a des fins commerciales ou autres, sur worldwide, for commercial or non­ support microforme, papier, electronique et/ou commercial purposes, in microform, autres formats. paper, electronic and/or any other formats.

The author retains copyright L'auteur conserve la propriete du droit d'auteur ownership and moral rights in this et des droits moraux qui protege cette these. Ni thesis. Neither the thesis nor la these ni des extraits substantiels de celle-ci substantial extracts from it may be ne doivent etre imprimes ou autrement printed or otherwise reproduced reproduits sans son autorisation. without the author's permission.

In compliance with the Canadian Conformement a la loi canadienne sur la Privacy Act some supporting forms protection de la vie privee, quelques may have been removed from this formulaires secondaires ont ete enleves de thesis. cette these.

While these forms may be included Bien que ces formulaires aient inclus dans in the document page count, their la pagination, il n'y aura aucun contenu removal does not represent any loss manquant. of content from the thesis.

•+• Canada To:

Tom and Fiona Abstract

Bacterial conjugation is repressed by a two-component system comprising the antisense RNA FinP and its protein co-factor FinO. FinO protects FinP from cellular degradation and mediates base-pairing to its target traJ mRNA leading to translational inhibition of the transcriptional activator TraJ and subsequent downregulation of conjugation genes. In this work, we explore the mechanism of

FinO-mediated intermolecular duplexing of FinP and traJ mRNA. Biochemical assays show that FinO performs strand exchange on an SLII-derived duplex RNA and mediates duplexing of SLII to its complement SLIIc from traJ mRNA in an ATP- independent fashion reminiscent of RNA chaperones. This strongly suggests that

FinO destabilizes base pairs within SLII and SLIIc which present a barrier to duplex formation. Also, the strand exchange and duplexing processes are shown to be important for repression of bacterial conjugation in vivo. Scanning alanine mutagenesis shows that the strand exchange area of FinO resides in a lysine-rich area on the N-terminal a-helix containing a solvent exposed tryptophan residue which is hypothesized to play a role in destabilizing portions of the stem regions of

SLII and SLIIc. N-terminal truncation mutants abolishing strand exchange and duplexing bind significantly tighter than wild-type FinO indicating that FinO uses a portion of its binding energy to destabilize RNA duplexes. Secondly, we examine the details of how FinO binds to its target RNAs. We attempt to crystallize a FinO-SLII complex using a variety of RNA and protein constructs. However, despite our efforts, the best crystals only diffract to 4.5 A. Site specific protein-RNA cross-linking experiments highlight positively charged regions of the N-terminal alpha-helix and globular region of FinO which contact SLII. RNase footprinting studies locate the areas of SLII contacting FinO. FinO protects the lower 3' side of the SLII stem and 3' tail from RNase digestion. The loop region of SLII is not contacted by FinO as

shown previously. N-terminal truncation mutants exhibit similar RNase footprinting

patterns as FinO, indicating the globular region of FinO plays the primary role in

binding to SLII. A model is proposed highlighting the relationship between the

binding and catalytic components of FinO-mediated inhibition of conjugation. Acknowledgements

First, I would like to thank my supervisor, Dr. Mark Glover, for his guidance and patience throughout my Ph.D. program. Thank you for taking a chance on a student who did not have the best academic record but was eager to jump into the research environment. Thanks for marching me up Heart mountain, Roche Miette, Mt. Glasgow (almost...), and Mt. Rundle. Finally, thanks for the chair lift conversations at the Colorado "Ski-stone" meetings. I would need another thesis to fully thank my FinO colleague and good friend Dr. Ross Edwards. Ross patiently taught me almost everything I know today and more importantly shared many experiences with me, good and bad. I think of Ross when I hear the words: All Blacks, Wolfenstein, Black Dog, baked glassware, Linux, Friday morning hockey, and underscore. Dr. Steven Chaulk is the Newfoundlander who taught me everything I know about RNA. Since this makes up a sizeable chunk of this thesis, I owe him a great deal of thanks. I would also like to thank Steve for being such a good friend. Sitting on his front porch sharing a cold one, "jamming", torturing ourselves watching the Leafs, and winter BBQs were just a few things I'll remember. I would like to thank Dr. Alex Ghetu for helping me get started with the FinO project. Alex taught me a lot of techniques and helped out with many experiments in this thesis. I would like to thank the other members of the Glover Lab, past and present, for their helpful discussions and friendship: Scott, Jason, Nina, Megan, Gina, Jun, Stephen, Charles, Danny, Ruth, Joyce, Jody, and Diana. Thank you to my collaborators and other departmental members for helpful discussions during my graduate studies: Dr. Laura Frost, Dr. Michael Gubbins, Dr. Leo Sypracopoulos, Dr. Andrew MacMillan, Dr. Paul Scott, Dr. Tracy Raivio, Dr. Daelynn Buelow, Craig Garen and Sheraz Khan. Finally, thank you to my fellow graduate students from the Department of Biochemistry for so many "good times". Table of Contents

Page Chapter 1: Introduction

Bacterial conjugation 1

Repression of conjugation - from past to present 4

Themes in antisense RNA regulation of gene expression in prokaryotes 15

RNA chaperones 21

Organization of the thesis 28

References 39

Chapter 2: FinO acts as an RNA chaperone to facilitate F'inP-traJ mRNA interactions

Overview 49

Introduction 50

Results 51

Discussion 59

Materials and Methods 63

References 80

Chapter 3: Strategies for crystallization of the FinO-SLII RNA complex

Overview 85

Introduction 85

Results 89

Discussion and future directions 97

Materials and Methods 100

References 120 Chapter 4: Examining the molecular details of the interaction of FinO with its target RNAs

Overview 125

Introduction 125

Results 128

Discussion and future directions 136

Materials and Methods 141

References 157

Chapter 5: General Discussion

Overall summary 161

Orientation of FinO on FinP SLII 163

Implications for FinP and traJ mRNA duplexing 165

The N-terminus of FinO and disorder as a possible mechanism for its chaperoning function 170

Concluding remarks: The bigger picture 171

References 177

Appendix A: Assaying FinO structural homologs for RNA chaperone activity

Introduction 181

Results and discussion

Assaying ProQ for RNA chaperone activity 185

Assaying NMB1681 for RNA chaperone activity 188

Materials and Methods 189

References 197 Appendix B: Biochemical characterization of the envelope stress accessory protein, CpxP

Overview 199

Introduction 200

Results 202

Discussion 207

Materials and Methods 209

References 220 List of Tables

Page

Chapter 2

Table 2-1 - Effect of FinO mutations on rates of strand exchange, duplexing, and conjugative inhibition 76

Table 2-2 - FinO double alanine mutant primer oligonucleotides 77

Table 2-3 - FinO single alanine mutant primer oligonucleotides 78

Table 2-4 - DNA templates for in vitro transcription of RNAs for Chapter 2 79

Chapter 3

Table 3-1 - Data collection statistics for FinO 26-186 W36A/SLII guga-5 crystal collected at ALS 8.3.1 117

Table 3-2 - FinO truncation mutants Gateway attB PCR oligonucleotides 118

Table 3-3 - DNA templates for in vitro transcription of RNAs in Chapter 3 119 List of Figures

Page

Chapter 1

Figure 1-1 - The bacterial conjugation cycle 29

Figure 1-2 - Organization of the genes of the F-plasmid transfer operon and their functions 30

Figure 1 -3 - Summary of early fertility inhibition experiments 31

Figure 1-4 - Summary of the components of fertility inhibition 32

Figure 1-5 - Model of FinOP repression of bacterial conjugation 33

Figure 1-6 - Loop-loop interactions in antisense RNA regulatory systems 34

Figure 1-7 - Regulation of ColE1 DNA replication by RNA I and Rom 35

Figure 1-8 - The action of RNA chaperones 36

Figure 1 -9 - RNA chaperone activity of the HIV-1 nucleocapsid protein 37

Figure 1-10 - RNA chaperone activities of Hfq 38

Chapter 2

Figure 2-1 - RNA and protein constructs used in this study 68

Figure 2-2 - FinO can perform strand exchange on SLII-derived duplex RNAs 69

Figure 2-3 - Residues in the N-terminal a-helix of FinO are critical for strand exchange 71

Figure 2-4 - The N-terminus of FinO is important for facilitating sense- antisense interactions 72

Figure 2-5 - RNA strand exchange deficient FinO mutants bind RNA tighter than wild-type FinO 74

Figure 2-6 - In vivo stabilization of FinP by FinO and FinO derivatives 75 Chapter 3

Figure 3-1 - X-ray crystal structure statistics from the Protein Data Bank (PDB) as of July 22, 2008 108

Figure 3-2 - Employing SLII stem truncations and tetraloops for FinO-SLII crystallization 109

Figure 3-3 - Optimization of FinO-SLII gnra-5 complexes 111

Figure 3-4 - Optimization of FinO 33-186 W36A/SLII gaaa-5 complex with polyamines 112 Figure 3-5 - The effect of crystallizing the FinO 33-186 W36A/SLII gaaa-5 complex in the presence of cryoprotectants 113

Figure 3-6 - Screening complexes of FinO 33-186 W36A with SLII-derived duplexes of varying length containing blunt or overhanging ends 114

Figure 3-7 - Employing the U1A crystallization module to help crystallize the FinO/SLII complex 115

Chapter 4

Figure 4-1 - Overview of constructs used in the study 147

Figure 4-2 - Site specific cross-linking of FinO and SLII 148

Figure 4-3 - RNase V1 cleavage of 5' and 3' end labeled SLII in the absence and presence of various FinO constructs 150 Figure 4-4 - RNase V1 cleavage of 5' and 3' end labeled SLIIc in the absence and presence of various FinO constructs 151

Figure 4-5 - Limited RNase I digest of 5' and 3' end labeled SLII and SLIIc in the absence and presence of various FinO constructs 152

Figure 4-6 - RNase I overdigestion of 5' and 3' end labeled SLII and SLIIc in the absence and presence of various FinO constructs 153

Figure 4-7 - Summary of RNase V1 and I cleavage reactions of SLII and SLIIc 155

Figure 4-8 - FinO binding to SLII requires a terminal 3' OH on the 3' tail of SLII 156

Chapter 5

Figure 5-1 - Model of the FinO 45-186-SLII interaction 173

Figure 5-2 - The many roles of FinO in FinP-fraJ mRNA intermolecular duplexing 174

Figure 5-3 - Model of the FinO-mediated SLII-SLIIc recognition complex 176 Appendix A

Figure A-1 - Comparison of the ProQ (1 -121) homology model to FinO (33-184) 193

Figure A-2 - Assaying ProQ-His6 for RNA binding and chaperone activity 194

Figure A-3 - Assaying Neisseria meningitidis strain MC58 protein of unknown of unknown function NMB1681 for RNA chaperone activity 196

Appendix B

Figure B-1 - Wild-type CpxP is mainly a-helical and may make a slight structural rearrangement at alkaline pH 214

Figure B-2 - CpxP is a dimer at pH 5.8 and 8.0 215

Figure B-3 - Preliminary crystallizaton of wild-type CpxP 216

Figure B-4 - Limited trypsin digest of wild-type CpxP removes flexible areas of the

protein 217

Figure B-5 - SAXS and in vivo chemical cross-linking confirms that CpxP is a dimer 218

Figure B-6 - Model of alkaline pH induced activation of the Cpx signal transduction pathway 219 List of Abbreviations

Amp - ampicillin

APA-Br - 4-azidophenacyl bromide

ATP - adenosine triphosphate

CD - circular dichroism cDNA - complementary DNA

DEPC - diethypyrocarbonate

Dmax - maximum particle dimension

DM - double mutant (U1A)

DMS - dimethylsulfate

DNA - deoxyribonucleic acid

DTNB - dithio-bis(2-nitrobenzoic acid) (Ellman's Reagent)

DTT - dithiolthreitol

EDTA - ethylenediaminetetraacetic acid

EMSA - electrophoretic mobility shift assay

E signal - encapsidation signal element

F+ cell - E. coli cells contain F plasmid

F" cell - E. coli cells lacking F plasmid

F/7/s - F' plasmid with his operon fin - fertility inhibition

Fin+ - phenotype with fertility inhibition

Fin" - phenotype without fertility inhibition fis - fertility inhibition site of action

F/ac - F' plasmid with lac operon

F plasmid - fertility plasmid

FRET - fluorescence resonance energy transfer

GNRA - tetraloop consisting of guanine, any base, purine, adenine GST - glutathione-s-transferase

HDV - Hepatitis delta virus

HEPES - (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid )

Hfq - E. coli Host factor I

HFT - high frequency transfer

HIC - hydrophobic interaction chromatography

HIV-1 - human immunodeficiency virus type 1

HK - histidine kinase

IAA - isoamyl alcohol

IHF - integration host factor

IncF - incompatibility group F ki - apparent first order strand exchange rate constant k2 (kapp) - apparent second order duplexing rate constant

Ka - association constant

Kan - kanamycin kDa - kilo-dalton

KH - hnRNP K homology domain

LB - Luria-Bertani broth

LOF - loss of function

MAD - multi-wavelength anomalous dispersion

MALDI-TOF - matrix assisted laser desorption/ionization time-of-flight

MALLS - multiangle laser light scattering

MBP - maltose binding protein

MCSG - Midwest center for structural genomics miRNA - microRNA

MME - monomethyl ether mRNA - messenger RNA

MW - molecular weight NC - nucleocapsid ncRNA - non-coding RNA

NdAg - N-terminal domain of Hepatitis delta antigen

NRMSD - normalized root mean squared deviation

OMP - outer membrane protein

ORF - open reading frame oriT - origin of transfer oriV - origin of replication

PAGE - polyacrylamide gel electrophoresis

PBS - primer binding site

PCR - polymerase chain reaction

PDB - Protein Data Bank

PEG - polyethylene glycol

PML - pseudo mother liquor

PMSF - phenylmethanesulphonylfluoride

PONDR - Predictor of natural disordered regions

QELS - quasi-elastic light scattering

RBD - RNA binding domain

RBS - ribosomal binding site

R-factor - resistance factor

Rg - radius of gyration

RH - radius of hydration

RNA - ribonucleic acid

RNAi - RNA interference

RNAP - RNA polymerase

RNase - ribonuclease

RR - response regulator

RRM - RNA recognition motif SAXS - small angle x-ray scattering

SDS - sodium dodecyl sulfate

Sll - SLII derivative missing seven nucleotide loop

SUA - Sll without 5' and 3' single stranded tails

SII+4 - Sll with four added base pairs siRNA - small interfering RNA

SL - stem-loop snRNP - small nuclear ribonucleoprotein sRNA - small RNA

Str - streptomycin ti/2 - half-life

T4SS - Type IV secretion system

TAR - transactivation response element

TCEP - tris(2-carboxyethyl) phosphine

TEMED - N,N,N',N'-Tetramethylethylenediamine

Tm - melting temperature tra - transfer tris - tris(hydroxymethyl)aminomethane tRNA - transfer RNA ts - temperature sensitive

UTR - untranslated region

U-turn - uridine turn

WT - wild-type

YUNR - loop motif consisting of pyrimidine, uridine, any base, purine Chapter 1

Introduction

Bacterial Conjugation

Conjugation is the process by which bacteria horizontally spread genetic elements by cell-cell contact throughout a population (2). The process was discovered by Lederberg and Tatum (60), and was fundamental for early work in bacterial molecular genetics. This work demonstrated that genes from the Escherichia coli chromosome could be mobilized and provided the basis for the concept of genetic recombination (102). The 100 kb extra-chromosomal plasmid which was found to be transferred in the study and has since shown to be the paradigm for bacterial conjugation is the fertility or F-plasmid from E. coli strain K-12 (reviewed in (33)). A representation of the conjugation cycle is shown in Figure 1-1. Donor bacteria harboring the F plasmid (F+) possess tubular structures called sex pili which can extend into the environment and contact acceptor (F) cells which lack the plasmid in a process called mating (16) (i). Once an F" cell is attached to the F+ bacterium, the donor cell retracts its pilus, through depolymerization of its F-pilin subunits (32, 33), bringing the acceptor cell into close contact forming a stable conjugation junction (ii and iii) (20). Often, complex mating aggregates can form with heterogeneous numbers of donor and recipient cells in close contact (1). The area of cell-cell contact forms a channel through which the DNA is transferred by a type IV secretion system (T4SS) which is composed of many transfer proteins (59). A signal in response to mating pair formation initiates strand transfer at the origin of transfer (oriT) which is nicked by a multi-protein relaxosomal complex (105) The DNA single strand to be transferred is then fed 5' to 3' through the channel (47), possibly accompanied by the relaxase which is covalently attached to the 5' end, into the recipient cell (iv) (64). During the transfer step, the F+ cell replicates a new DNA strand from the non-mobilized F-

1 plasmid strand (52) while the transferred strand is replicated in the recipient cell (45). The replication process, which may progress by rolling circle replication similar to (|>X174 phage (38) or through another mechanism (105), produces two double stranded F-plasmids: one of which is re-established in the donor cell, and a new plasmid in the acceptor cell. Now that the F" cell has received the F-plasmid, it becomes F+ and proceeds to "infect" other F" cells in the population (v) (16). The machinery responsible for moving the F-plasmid from F+ to F" cells is encoded from a 33.3 kb transfer region of the F-plasmid (33). Shown in Figure 1-2, the transfer (tra) genes are arranged in a

polycistronic fashion under the control of the PY promoter. The genes are classed into groups for pilus synthesis and assembly, mating pair stabilization, DNA transfer, and surface and entry exclusion. As mentioned above, the pilus structure, which is needed for mating pair formation, is composed of F-pilin subunits encoded by the traA gene (32). When traA is expressed, it produces a larger protein called propilin which contains a signal sequence directing it to the inner membrane (66). After cleavage of the signal sequence by a peptidase, the traQ gene product positions the mature pilin correctly in the inner membrane (66). This location is where the pool of pilin subunits is housed for use during pilus assembly (71). Acetylation of the N-terminus of F-pilin has been shown to occur post-translationally and is carried out by traX gene product (66). Assembly of the pilin subunits into a tubular filament initiates at the base of the pilus, taking subunits from the inner membrane pool. Likewise, after mating pair contact, the pilus is retracted to bring the cells into close contact. This is carried out by depolymerization of the filament which returns the pilin subunits to the inner membrane (59). A large number of proteins are required for assembly of the pilus however the exact functions of most proteins are currently unknown. Genetic studies have characterized their location in the T4SS and their importance in various stages of pilus formation. Pilus tip formation on the cell surface is performed by the traL, traE, traK (a putative secretin), and traC genes as well as the N-terminus of traG (59)

2 Pilus extension is carried out by traB, traF (a potential thioredoxin/chaperone), traH, traV (a lipoprotein), and the trbC-traW fusion (59). Once the mating pair or aggregate has been established, stabilization occurs through traG (particularly the C-terminus), trail, and traN (59). TraN is an outer membrane protein of donor cells which, in the F-plasmid, interacts with OmpA and a component of the lipopolysaccharide in the recipient cell to stabilize cell-cell interactions. Surface and entry exclusion mechanisms exist to prevent conjugation between two cells containing closely related plasmids belonging to the same exlusion group (33). The traT gene codes for an outer membrane protein which blocks mating pair stabilization (surface exclusion) and traS encodes an inner membrane protein which blocks a signal for DNA transfer (entry exclusion) possibly through an interaction with rraG (3). Once cell-cell contact has been established, a signal is generated to initiate transport of the DNA strand to the recipient cell. The cytoplasmic relaxosome complex is responsible for readying the DNA for transport and is coded for by the traM, tral, and traYgenes (33). TraM recognizes the mating signal after mating pair stabilization and is thought to couple the relaxosome to the DNA transfer machinery (transferosome). It has been shown to interact with DNA and the transfer coupling protein encoded by traD (65). The traM gene is located upstream of the tra cluster and has its own promoter. The tral gene product is involved in oriT nicking and is covalently attached to the 5' end of the nicked strand (33). It also functions as a ATP-dependent helicase unwinding the plasmid DNA so that one strand can be transferred (33). Two accessory proteins are required for the nicking reaction: TraY and the host- encoded integration host factor (IHF). These are thought to bend the DNA to present a suitable substrate for Tral (106). TraD is an inner membrane protein which couples the relaxosome to the transferosome and is thought to be involved in feeding the single stranded DNA through the conjugative pore (65).

The genes of the tra operon are under positive regulation by the TraJ transcriptional activator which is encoded from the traJ gene upstream of the tra gene cluster (Figure 1-2) (101). It recognizes PY,

3 which is the promoter of traY, the first gene in the transfer operon (33) and has been shown to require host-encoded regulators such as ArcA (84) and the leucine-responsive regulator protein (9). A number of mechanisms have been shown to promote negative regulation of the tra operon, mainly at the level of traJ expression, including DNA adenine methylation (9), Hfq destablization of traJ mRNA (see below) (100), and GroEL-assisted proteolytic degradation of TraJ (109). However, the primary focus of this thesis is on the plasmid encoded fertility inhibition system which is the primary mechanism for repression of bacterial conjugation.

Repression of conjugation - from the past to the present

R-factors and fertility inhibition One of the earliest, large-scale civilian uses of antibiotics was for the treatment of bacterial dysentery in Japan after WWII. While the antibiotics were initially effective, multi-drug resistant strains rapidly arose (97). Some pathogenic, drug-resistant Shigella strains were found to pass their resistance to E. coli cells, thereby suggesting that bacterial conjugation was the mechanism for transfer (97). The genetic elements responsible for the resistance were termed resistance factors or R- factors. They belong to a group of F-like plasmids which, as the name implies, are similar to the F-plasmid except that in addition to carrying antibiotic resistance or virulence genes, they have the ability to negatively regulate conjugative transfer. The classic F-plasmid from the E. coli K12 strain, on the other hand, was found to be constitutively active for conjugation. The majority of F-like plasmids only transfer plasmids at a high frequency for a short period of time (96) because they have a repression system originally referred to as t (22) or ft (98) but later renamed fertility inhibition or fin+ (26). This was shown in established strains (i.e. grown to stationary phase) containing the derepressed (Fin) F plasmid and a repressed (Fin+) F-like plasmid (e.g. R100 R-factor), as transfer of F to F'R" cells occurs at a very low frequency (0.1-1%) compared to cells harboring only F (97).

4 Due to the metabolic burden imposed by the conjugation process, particularly with F-pilus synthesis and assembly, fertility inhibition is an important process to ensure that the cell is not wasting precious energy under conditions where conjugation is not needed. Also, in the repressed state, the cell is more resistant to male-specific phages such as the f2 and Q(3 RNA containing phage or the f1 DNA containing phage, which adsorb to the sides and tip of the F-pilus respectively and pose a threat to the cell (16, 68).

Discovery of fertility inhibition components It was initially thought that a cytoplasmic repressor product derived from R-factors solely inhibited F/ac conjugation by inhibiting transcription of one or more transfer genes (68). The site of action of the repressor was called traO (later renamed fisO). It was thought to be either a true operator sequence or associated with the traJ gene since the presence of the repressor led to inhibition of DNA transfer, F-pilus formation and surface exclusion which were known to be controlled by the traJ gene (28). Isolated R-factors which had a mutation in fisO were derepressed for transfer because they were insensitive to the repressor. Interestingly, the repressor is not plasmid specific since a repressor from one plasmid could lead to transfer inhibition of a number of plasmids which do not have a functional repressor gene (26). As it turns out, the repressor, later renamed FinO (34), was not the only component involved in the inhibition of conjugative transfer. In 1971, Finnegan and Willetts characterized a plasmid-specific product called traP (28) found in Flac1 mutants insensitive to inhibition by R100. Mutations in traP, later named finP (34) were similar to those of fisO mutants as they were insensitive to R100 FinO and were derepressed, however, they were recessive in nature. This was demonstrated in experiments utilizing a transient heterozygote (finP/finP*) intermediate strain which consisted of three plasmids: the F/ac mutant (finP), R100, and Fhis (finP*)2. A schematic of the experiments is shown in Figure 1-3.

1 F' plasmid carrying the chromosomal lac operon (2) 2 F' plasmid containing the chromosomal his operon (26) 5 Here, an exponential donor strain consisting of wild-type F/ac or its mutants was mated with an established intermediate recipient strain containing Fhis, R100, or both plasmids. Mating was stopped by T6 phage killing of the remaining donor F/ac cells. Transient heterozygotes were formed when F/ac (Lac+) donor cells were mated with cells harboring Fhis (Lac). This culture was then immediately mixed with a streptomycin-resistant (Strr) F"R" recipient strain. After a period of mating, the culture was plated to select for Lac7Strr transconjugate cells. The table in Figure 1-3 shows the results of the experiments. As expected, when wild-type F/ac cells, containing a functional finP gene, were mated with an intermediate strain containing no additional plasmid (i) or Fhis (ii), the cells could re-transfer F/ac efficiently due to the absence of the FinO repressor. Unexpectedly, when the exponential F/ac strain was mated with an established strain containing R100, re-transfer of F/ac was also efficient (iii). This was attributed to either a slow production of FinP or a slowly forming FinO-FinP interaction (28). This period where cells containing F-like factors can transfer unrepressed for a few generations is referred to as high frequency transfer (HFT) (68). When the F/ac strain was mated with an established R100/F/}/s strain re- transfer of F/ac was inhibited demonstrating that both FinO (R100) and FinP (Fhis) are needed to form a functional repression complex (iv). Interestingly, strains containing the Flac finP mutants were also inhibited from re-transfer to the F"R" strain after forming a transient heterozygote strain containing the mutant plasmids R100 and Fhis (vii). In contrast, re- transfer occurred readily from transient heterozygote strains containing F/ac fisO mutants, R100, and Fhis (x). This demonstrated that while the FinO repressor was insensitive to both types of mutations, the finP mutations were recessive and could be rescued by wild-type finP from Fhis. Flac fisO mutations were c/s-dominant rendering the plasmid derepressed for transfer (28). The Willetts group went on to show that many R-factors and F-like Col plasmids, which produce colicins that destroy other cells in the environment which do not have the Col plasmid (14, 75), can repress transfer of F/ac and other derepressed R and Col plasmids, demonstrating that the FinOP system is a general mechanism

6 for transfer inhibition (26, 34). It was found by the same group that traJ was indeed the site of action of FinO (27).

Genetic location of finP and finO The finP gene locations for F (104) and the R6-5 F-like plasmids (87) were determined by restriction endonuclease mapping. Results for both plasmids were similar, mapping the gene near traJ and it was later narrowed down to a location between traM and traJ (51, 86) (Figure 1-2). The finO gene location from plasmid R6-5 (87) and R100 (19, 67) were mapped to the very end of the tra operon, within 1 kb from the origin of replication (oriV) (Figure 1-2). It was later found that the naturally de- repressed F plasmid contains an IS3 insertion sequence in its finO gene rendering its gene product non-functional (Figure 1-2) (11, 108). In contrast, repressed F-like plasmids have functional uninterrupted finO genes whose products are able to aid in inhibition of their own and F plasmid transfer to recipient cells.

The nature of finP and finO After the discovery of the gene locations, the nature of the finP and finO products were still under debate. If the finP gene was to encode a protein, it would be very small (24-52 amino acids) (86). Mullineaux and Willetts suggested that an 81 base pair restriction fragment housed the finP gene (73). The fragment also contained two previously characterized finP mutations and was suggested to be expressed from the DNA strand complementary to the leader sequence of the traJ gene. They hypothesized that the finP gene product was an RNA due to the presence of a strong transcriptional terminator (consisting of a 12 base pair perfectly inverted repeat with many GC pairs and a sequence of four thymines 3' to the repeat) which would disrupt an open reading frame (ORF), producing a smaller protein without a termination codon (73). Also, a large number of finP alleles exist within the plasmid incompatibility group F (IncF) of F-like conjugative plasmids (103). The RNA hypothesis was confirmed by Finlay ef a/, upon the sequencing of five of the six f/nP alleles from different IncF plasmids (25).

7 The sequences were aligned showing conserved sequences in the promoter region and other areas of the gene with the exception of the six- seven nucleotide loop sequence between the 12 bp inverted repeat. Each allele showed variability in this region showing that this area is likely responsible for the plasmid specificity of finP (28). This area also contained stop codons and frame shift mutations in different alleles re- enforcing the idea that FinP was an RNA (25). Two groups later showed that finP does indeed produce a small antisense RNA: Frost et al. isolated an 80 nucleotide FinP from total cellular RNA (31), and Koraimann et al. showed that finP mutations, which were made to interrupt potential ORFs, did not affect the ability of FinP to inhibit conjugation (57). Secondary structure predictions of the five FinP RNA sequences were performed showing two stem-loop structures separated by a single stranded region (25). A similar result was obtained for a 72 nucleotide FinP RNA from plasmid R1 (57). van Biesen et al. enzymatically probed the structure of the 79 nucleotide in vitro transcribed FinP confirming the prediction results (Figure 1-4)(93). The cleavage data showed that FinP consists of two stem-loops, SLI and SLII, separated by a four nucleotide single stranded spacer. At the 5' end of the RNA is a four nucleotide single stranded leader and at the 3' end is a six nucleotide single stranded tail. SLI contains an eight nucleotide loop and harbors an AA mismatch within its stem region. SLII (previously the 12 bp inverted repeat) contains a seven nucleotide loop (area of allelic variation discussed above) and has 14 contiguous base pairs. The authors also analyzed the structure of the 5' untranslated region (UTR) of a 211 nucleotide in vitro transcribed traJ mRNA (Figure 1-4). The RNA is fully complementary to FinP RNA and contains 3 stem loops (from 5' to 3'): SLIM, SLIIc, and SLIc (93). SLIc contains the ribosomal binding site within the 3' portion of its loop and upper part of the stem and the AUG start codon for the traJ ORF at the lower 3' side of the stem. It has an eight nucleotide loop and contains UU and AC mismatches in its stem region. SLIIc is a 12 bp contiguous stem loop with a seven nucleotide loop. 5' to SLIIc is a long A/U rich

8 single stranded stretch of nucleotides which is preceded by the small SLIM stem loop (93). The nature of the finO gene product was also controversial. The ORF of the R100 finO gene encoded a protein with a molecular weight of 21,268 Da (67, 108), but the isolation of multiple active FinO constructs, some of which arose due to internal deletion mutations, made it hard to determine which was the correct sequence (19, 67). The finO gene products cloned from the R6-5 plasmid also gave multiple sizes (87). The issue was solved by Yoshioka et al., showing that the R100 finO gene did indeed encode a 186 amino acid (21.2 kDa) protein (108). Another confusing aspect regarding finO was that there appeared to be two alleles of the gene with varying abilities to repress F plasmid conjugative transfer: Type I repressed 100-1000 fold, and type II repressed 20-50 fold (103). van Biesen and Frost discovered that the difference between the two types of alleles was due to an orf286 gene upstream of the type I finO sequences which was not present in type II plasmids (91). This showed that the levels of repression in different plasmids are not due to finO sequence differences but rather the amount of FinO present in the cell and that a c/s-acting orf286 product enhances the expression of FinO at the translational level leading to increased inhibition of transfer (91). The crystal structure of a FinO truncation mutant (26-186) was solved by Ghetu et al. in 2000 (Figure 1-4)(36). It was previously determined by limited proteolysis experiments that the first 25 amino acids of FinO is disordered in the presence of its RNA substrate (37) and removal of this region was necessary for crystallization. The overall fold of FinO is novel. It forms an elongated, primarily a-helical structure consisting of a long N-teminal helix (a1) connected to a mixed a/p core and ending with a long C-terminal helix (a6) which packs against the bottom of the N-terminal helix (Figure 1-4A). Residues 26-32 and 185 and 186 could not be modeled due to flexibility. Interestingly, the FinO structure also consists of a hole 5 A in diameter formed between the isl­ and C-terminal helices and the core region. An electrostatic potential surface representation shows large areas of positively charged residues throughout the N-terminus and on one face of the core region. 9 FinO binding protects FinP from RNase degradation Following the characterization of the FinOP gene products, the focus shifted towards understanding the biochemical mechanism of conjugative repression. After determining that finP produces an antisense RNA, Mullineaux and Willetts proposed that the repression of conjugation occurs through a mechanism by which FinO and FinP RNA interact with traJ or its mRNA to downregulate its expression (73). Upon the discovery of finP, one hypothesis was that FinO and FinP interact to form a functional repressor complex (28). FinO would interact with finP post-transcriptionally since the relative levels of gene expression from the finP promoter remained the same upon addition of finO to the system (73). Indirect evidence for the interaction surfaced when it was discovered that R100 FinP RNA levels increased four to ten-fold in the presence R100 FinO in vivo (18). Frost et al. confirmed this result, reporting a ten-fold increase in an 80 nucleotide F finP transcript upon incubation with a recombinant plasmid harboring finO from R6-5 (31). FinO appeared to stabilize the FinP structure after it was produced either by aiding in its folding or by blocking its degradation (31). Previously characterized mutations in finP affecting its action or the action of FinOP (fisO mutants) (28, 73) generally affected transcript stability. The group found that the fisO305 mutation (C30U in SLI, Figure 1-4) had the most deleterious effect, as both conjugative repression and transcript stability were compromised in the presence of FinO (31). Interestingly, in the absence of FinO, higher concentrations of cellular FinP and inhibition of conjugation resulted if the finP gene was supplied from a multi-copy plasmid, thus mimicking the effect of FinO (31, 57). Lee et al. showed that FinO is able to extend the half-life of FinP RNA from seven minutes in its absence to >40 minutes in its presence (61). In this study, FinP was synthesized in the absence of traJ mRNA which suggests that FinO blocks the degradation of the antisense RNA (61). FinP contains a potential RNase E cleavage site at the bottom 3' side of SLI, an area which also harbors the C30U mutation of the fisO305 finP mutant (61). To test whether RNases were involved in the

10 deleterious effect of the mutation, the group transcribed FinP in the E. coli cells containing temperature sensitive (ts) mutations in two genes encoding endoribonucleases which degrade cellular E. coli RNAs: rne (RNase E) or rnc (RNase III) (10, 74). For me'5, the half-life of FinP was extended from the 5 minutes at the permissive temperature of 30°C to 20 minutes at 43°C for a larger 150 nucleotide FinP transcript and from 5 to 10 minutes for the 80 nucleotide FinP (61). For mcte cells, there was no difference on the half-life suggesting that RNase E is involved in cellular degradation of FinP whereas RNase III is not. In addition, two other E. coli ribonucleases, RNase II and PNPase (74), were later determined not to be involved in FinP degradation (49). Jerome et a/., followed up on this study by demonstrating that the primary FinP RNase E cleavage site in vitro was located at the single stranded spacer (A36; Figure 1-4) between stem-loops I and II (49). Mutating the FinP spacer from 5'-GACA-3' to GCCC increased the half- life in a wild-type rne+ strain from 14 to 48 minutes highlighting the importance of the site for cleavage in vivo. Furthermore, when FinO-GST was incubated with FinP and RNase E in vitro, FinO was able to protect the spacer from RNase E digestion (49). In agreement with previous data (31), the finP305 (formally fisO305) was not protected from RNase E at the U30 mutation site in the absence and presence of FinO-GST (49). This indicates that the deleterious nature of the mutation results from a new RNase E cleavage site which cannot be protected by FinO. Therefore, FinO-FinP interactions play an important role in stabilizing the cellular levels of FinP enabling repression of conjugation (Figure 1-5 i and iv).

FinO enhances FinP-traJ mRNA duplexing Insight into the mechanism of FinP-traJ mRNA interaction was first provided by Koraimann et al. who showed that individual mutations in the loop of SLII of FinP could disrupt repression of conjugation (57). In their assay, plasmids which contained wild-type and single mutant finP genes were transformed into cells containing R1-19 (FinO) and mating efficiencies were measured. The group showed that mating inhibition

11 was completely disrupted when an A to C transversion occurred at the middle nucleotide of the SLII loop, while a slightly smaller disruption of inhibition occurred for the same mutation introduced on the 3' side of the same loop. In contrast, mating inhibition was comparable to wild-type FinP for an A to C mutation in the 3' side of the SLI loop or with individual mutations which disrupt base pairing within the stem of SLII (57). This indicated that the SLII loop was important for repression of conjugation, presumably through a mechanism involving loop-loop "kissing" interactions between FinP SLII and SLIIc from traJ mRNA. This interaction is prevalent in antisense RNA inhibition systems (see below). Loop-loop interactions between the two RNAs could lead to further base pairing which could block access to the RBS on traJ mRNA, thus preventing translation of TraJ and activation of conjugation genes (Figure 1-5). van Biesen et al. used an in vitro binding experiment to measure the kinetics of FinP-traJ mRNA base pairing (duplexing) in the absence of

FinO (93). They measured a second order duplex rate constant (kapp) of 5 x 105 M"V1 which is typical of natural antisense systems (93). The information from the two studies inferred a great deal about the nature of the traJ repression but it did not explain the role of the key player in the fertility inhibition system, FinO. The first direct evidence for a FinO-FinP complex was shown by van Biesen and Frost who used electrophoretic mobility shift assays to measure the equilibrium binding constants for the interaction (92). GST-FinO makes a specific complex 5 1 with an approximate Ka of 4 x 10 M" for both FinP and a 211 nucleotide traJ mRNA. Binding experiments with truncation derivatives of FinP showed that the specific interaction with FinP was located at the SLII domain, while a complex could not be formed with SLI alone. It was also shown that GST-FinO was able to increase the apparent second-order rate constant of in vitro F'mP-traJ mRNA duplexing five-fold relative to GST alone (92), suggesting that FinO appears to have a dual role in conjugative repression (Figure 1-5). First, FinO is able to stabilize FinP through specific interactions with SLII preventing its degradation in the cell so that levels are high enough for an interaction with traJ mRNA. Secondly, FinO actively promotes F'mP-traJ mRNA duplexing,

12 presumably to inhibit the translation of TraJ. It was further shown that when FinO was present, it was able to compensate for the mutations in the SLI and SLII of FinP, particularly in the loop of SLII, which were shown previously to disrupt mating inhibition in R1-19 (FinO) containing cells (57, 58). In addition, FinO was able to inhibit TraJ expression from a traJ-lacZ reporter construct in the presence of mutant FinP RNAs which had previously been shown to disrupt inhibition in a FinO" environment (58). The traJ inhibition was shown to be the result of duplex formation between traJ mRNA and FinP (49, 61). After formation, the intermolecular duplex appears to be rapidly degraded by the double- stranded specific RNase III (49). The results from the two studies suggest that FinO is crucial for efficient inhibition of traJ expression and conjugative DNA transfer and appears to initiate FinP-fraJ mRNA interactions in a different manner than the RNA alone situation, which may rely on loop-loop ("kissing") interactions for recognition. This was confirmed recently by Gubbins et al. which studied interactions between wild-type and mutated SLI RNAs from FinP and SLIc from traJ mRNA in vitro (40). Four nucleotide transversion mutations were introduced at different areas of the SLI loop, including the anti-RBS, and the kapp of duplexing with wild-type SLIc was measured in the absence and presence of FinO. In addition, the SLI(16-18) mutation which disrupts the anti-RBS was introduced into FinP using a moderate copy number plasmid. The mutated finP plasmid was then transformed into a finP strain, in the absence or presence of a finO plasmid, and conjugative DNA transfer and the levels of TraJ protein were then measured (40). In all experiments, FinO was able to compensate for the reduced inhibition of DNA transfer and traJ expression due to the mutations. The authors also demonstrated that SLI/SLIc duplexing appears to initiate from the loop region and that the single stranded tails and internal base mismatches in the stem region are required for efficient duplex progression (40).

13 FinO interaction with its target RNAs While it was known that FinO forms specific complexes with FinP and traJ mRNA and that SLII seemed to be the primary binding determinant in FinP, nothing was known about the mode of binding. Jerome and Frost showed that FinO binds to its target RNAs in vitro with high affinity and in a non-sequence specific manner (48). Equilibrium binding constants were measured for GST-FinO interactions with a wide variety of truncation and sequence mutants of FinP, and to a lesser extent traJ mRNA. The authors confirmed that SLII was the area of FinP which primarily contacts FinO and that the single stranded spacer region between SLI and SLII, and the 3' tail of SLII were very important for high affinity interactions (Figure 1-4). SLI, which bound poorly to FinO by itself, was chosen to have a 5' single stranded leader tail but no spacer at the 3' end. Binding could be improved significantly if the SLII 3' GAUUUU tail was added to the SLI RNA, highlighting the importance of the tails. It was also shown that the 3' tail needs > six nucleotides for strong interactions with GST-FinO (48). For traJ mRNA interactions, SLIIc and the six nucleotide single stranded spacer 3' to the stem-loop were required for high affinity interactions with GST-FinO (48). The results show that FinO recognizes the overall shape of the hairpin RNAs and that single stranded regions 5' and 3' to the stem region are important for high affinity interactions. Deletion studies were used to determine the areas of FinO which made interactions with FinP and traJ mRNA (37, 81). It was determined by limited proteolysis in the presence of SLII RNA and through electrophoretic mobility shift assays (EMSAs) that FinO has two RNA binding domains: 26-61 comprising the N-terminal a-helix, and 62- 186 which consisted of the core globular region and the C-terminal a-helix (37). FinO 1-61 and 62-186 fragments by themselves bind weakly to SLII indicating that both domains are needed for high affinity interactions with SLII. FinO was also shown to bind as a monomer to SLII (37).

14 Themes in antisense RNA regulation of gene expression in prokaryotes Gene regulatory circuits encompass an important part of functional processes in bacteria. In order for bacterial cells to survive, they must be able to regulate their genes in order to adapt to a particular environment. Negative regulation is important for ensuring metabolic energy is not wasted on unneeded processes. This applies to the bacterial conjugation process, particularly for the formation of the F-pilus which would require energy for F-pilin synthesis, pilus assembly, extension and retraction. In addition, sex pili are highly susceptible to phage attack which can infect and kill donor cells (16). For this reason, F- like plasmids have developed the FinOP fertility inhibition system. The c/s-acting FinP antisense RNA is required for duplexing to traJ mRNA and blocking synthesis of TraJ, probably through occlusion of ribosomal binding (Figure 1-5). This down-regulates conjugation genes including the traA gene and other pilin-specific genes. Non-coding RNAs (ncRNAs) are commonly used in both prokaryotes and eukaryotes to regulate a number of cellular processes (21). In eukaryotic cells, an increasing number of 21-25 nucleotide microRNAs (miRNAs) are being discovered which function in regulating gene expression of cellular processes related to developmental timing, apoptosis, metabolism, cardiogenesis, and myogenesis (53, 107). Also, changes in the expression of miRNAs have been linked to diseases such as cancer (53). The miRNAs are derived from larger stem-loop precursors which are processed by the enzymes Drosha and Dicer (41). One strand from the resulting mature miRNA-containing duplex is used to pair with partially complementary sequences in the target mRNA to repress translation in a process called RNA interference (RNAi) (41). The majority of ncRNAs have been discovered in prokaryotes (95). Antisense RNAs are employed primarily for the regulation of plasmid systems, but they are also emerging in chromosomal gene regulation (6). In addition to conjugation repression, they play crucial roles in the regulation of plasmid replication and copy number, transposition, post-segregational killing, control of stress responses,

15 osmoregulation, mRNA degradation and virulence (95). Most non-coding antisense RNAs are quite small (~50 to 300 nucleotides), structured, and are often fully complementary to their target RNAs (6). Binding of the antisense RNA to the target sequence generally results in translational repression (5). Antisense RNAs can be c/s-acting where their target mRNA is located in the same DNA segment or trans-acting where the antisense RNA is transcribed from another locus (6). The frans-acting RNAs are usually less complementary to their target RNAs. The following is a short review of the themes in antisense regulation of bacterial genes and how they relate to the FinOP fertility inhibition system using examples from classic and recently discovered systems.

Antisense-target RNA recognition by loop-loop ("kissing") interactions Due to their diffusible nature, antisense RNAs must recognize their target RNAs efficiently within the cell to quickly perform their regulatory duties. Second order pairing rate constants in vitro for most antisense systems are on the order of 106 M"1-s"1 which appears to correlate with the efficiency of in vivo control, thus showing that the RNA interactions are not merely diffusion limited (8). Commonly antisense- target mRNA pairings initiate with a transient, reversible "kissing" complex where a loop or single stranded region from the antisense RNA forms a small number of base pairs with a loop from the target mRNA. (6, 8) This rate-limiting kissing intermediate is followed by duplex progression and possible formation of a full duplex RNA (46). To account for this rapid association, many antisense RNAs contain a 5'-YUNR-3' (Y=pyrimidine, N=any base, and R=purine) sequence in one or more of their loops which pre-forms the loop into /\-form geometry aiding in the formation of the initial base pairs with the target RNA (29). A comparative sequence analysis of antisense and target recognition loops from many plasmids showed the ubiquitous nature of the RNA structure element in antisense systems (29). Initially discovered in the anti-codon loop of tRNAs (77), YUNR or U-turn motifs consist of a sharp bend in the phosphate backbone after the conserved uridine, followed by stacking of bases 3' to the turn, allowing this part of the loop to adopt a C3' endo /\-form stacking

16 conformation (Figure 1-6A)(29). A classic example of an antisense system utilizing loop-loop interactions is the CopA/CopT control of the R1 plasmid replication (Figure 1-6B). Here CopA, the antisense RNA, binds to the leader region of CopT (repA mRNA) preventing expression of the Tap leader peptide which controls expression of the plasmid replication initiator protein RepA (4). An interaction between stem-loops II of CopA and II' of CopT, which contains a U-turn motif, has been shown to occur through an initial kissing intermediate which can be visualized by native gel electrophoresis (76). The hoklsok post-segregational killing system of the R1 plasmid also uses a YUNR sequence for antisense-target mRNA recognition (29). In E. coli cells which harbor the R1 plasmid, the 5' single stranded tail of the Sok (Suppressor of killing) antisense RNA (anti-toxin) associates with the YUNR loop sequence in the 5' side of the mok (modulator of killing) domain of hok (host killing) mRNA (29). The mok stem-loop contains the Shine-Dalgarno and start codon for translation of the Hok protein (toxin) which is required for killing cells lacking the R1 plasmid after segregation (95). After Sok RNA binds to the mok domain, intermolecular duplexing resumes leading to translation repression of Hok (29). It was initially hypothesized that FinP RNA may make initial loop- loop interactions with its traJ mRNA target since mutations in the loop of SLII resulted in a decrease of conjugative inhibition (18, 31, 57). Both loops of FinP contain YUNR sequences making the hypothesis more attractive. However, evidence for kissing interactions is scarce, as attempts have failed to visualize kissing complexes on native gels with either SLI/SLIc (40) or SLII/SLIIc (Ghetu, A.F., unpublished data) using stem-loop pairs which are complementary only in the loop areas. It is not known whether the lack of evidence for a kissing intermediate is due to its transient nature or whether it exists at all. In contrast to other antisense systems, the fertility inhibition system relies on a protein co-factor, FinO, to function properly. Not only does FinO increase the levels of FinP so that it can pair with its target RNA, but it also significantly increases the rate of pairing between FinP and traJ mRNA in vitro (Figure 1-5) (92). This would account for the fact that FinO can rescue mutations in both

17 loops of FinP which disrupt RNA pairing in its absence (40, 58). Therefore, FinP may initially pair to traJ mRNA using its structured loops, but the recognition step is likely not as simple as other antisense systems due to the involvement of FinO.

Duplex Progression After initial contact of the antisense RNA with its target RNA, whether it be through loop-loop interactions (e.g. R1 CopA/CopT system; Figure 1-6B) or loop-single strand interactions (e.g. R1 hoklsok system; see above), the RNA-RNA complex must continue to form a stable intermolecular duplex in order to fulfill its negative regulation. This may be through formation of a complex which blocks the RBS and prevents translation of the target mRNA or through formation of a new substrate for ribonucleases such as the double strand specific RNase III which degrade the mRNA (6). For some antisense systems, conversion to a fully complementary intermolecular duplex does not occur and the process of forming a stable complex can have many intermediates. The highly structured nature of both antisense and target RNAs must be destabilized in order to promote intermolecular duplexing (8). This is accomplished primarily through the presence of mismatched or bulged out nucleotides within the stem-loops which disrupt RNA helices (56). Removal of bulged nucleotides or conversion of mismatched nucleotides to canonical Watson-Crick base pairs can have dire consequences on duplex progression (56) Intermolecular duplexes which initiate by loop- loop interactions accumulate torsional stress arising when both RNAs twist around one another (95). In addition to mismatched or bulged out nucleotides which help to unwind the stem-loops, duplex progression sometimes proceeds by multiple steps where distal areas of RNAs, such as single stranded tails, base pair with complementary sequences to relieve the stress (6, 95). The CopA/CopT antisense system proceeds by a multi-step mechanism forming an unusual four-helical complex upon intermolecular duplexing (Figure 1-6B) (55) As mentioned above, a kissing interaction occurs between loops II and II' of the two RNAs. Next, the intermolecular duplex progresses from the loop-loop interaction with

18 the aid from bulged residues within the stems of the RNAs (56) The single stranded section between stem-loop II and stem-loop I of CopA forms another intermolecular duplex with the 3' single stranded end of CopT, which contains the tap Shine-Dalgarno sequence and prevents translation of the peptide (54). The topological stress during the formation of the four-helical structure probably stops further duplex progression (8). The mechanism of duplex progression for the FinOP system will likely differ from other antisense systems due to the presence of FinO. Whereas SLII and SLIIc have contiguous Watson-Crick base pairing throughout their stems, SLI and SLIc have mismatches which would probably aid in duplex progression. Removal of the upper mismatches to Watson-Crick base pairs leads to decreases in SLI/SLIc pairing rates in vitro (40). Thus, in the absence of FinO, SLI and SLIc may be the ideal positions to start intermolecular duplexing. It was also determined that single stranded tail regions in SLI and SLIc are important for intermolecular duplexing, indicating that these areas may be additional starting points for duplex progression (40). FinO has been shown to greatly enhance FinP-fraJ mRNA duplexing rates (36, 92) and this can be attributed to its ability to destabilize and anneal the highly stable SLII and SLIIc stem-loops which otherwise would be a huge thermodynamic barrier to overcome. This aspect of duplex progression will be discussed in Chapter 2 of the thesis.

The role of protein co-factors in plasmid antisense systems The majority of plasmid antisense systems consist only of RNA components. They have all the tools needed to form duplexes including YUNR recognition motifs, bulges and mismatches in stems, and other single stranded regions. The first antisense regulation system identified was the RNA l/RNAII system in the ColE1 plasmid to control replication (23). As shown in Figure 1-7, an RNA primer called RNA II is synthesized by RNA polymerase from an initiation site 555 bp upstream of the origin of replication (1-7Ai). RNA II adopts a specific structure (Aii) which keeps it hybridized to the template strand of the DNA at the origin of replication

19 making an RNA-DNA hybrid (Aiii) which is then cleaved by RNase H to produce primers for replication initiation (Aiv and Av). RNA I is an antisense RNA (1-7Bi) which negatively regulates replication by binding to RNA II through three kissing complexes (Bii). The complex is then converted to a stable duplex that prevents persistent hybridization of RNA II to the DNA strand thus preventing the priming step (Biii). RNA-one modulator (Rom) is a 63 amino acid protein which binds as a dimer to an unstable loop-loop intermediate, decreasing the rate of its dissociation by 100-fold (88) (Figure 1-7 Bii, C) (23). It does not associate with individual loops or double stranded RNA. Instead, it binds to the kissing complex in a non-sequence specific manner, recognizing its overall shape (24). However, the fact that Rom exerts only a small effect on the overall rate of RNAI/RNAII pairing and ColE1 copy number shows that it appears to be dispensible (95). In contrast to Rom, FinO is absolutely required for inhibition of conjugation. FinO is needed for protecting FinP from degradation by RNase E (49) so that the antisense RNA can bind to traJ mRNA, and is responsible for enhancing the pairing rate of the two RNAs (Figure 1-5) (92). As will be demonstrated in Chapter 2, FinO is responsible for active RNA-RNA duplex formation through destabilization and annealing events. Thus FinO is a very important protein co-factor as it is needed for multiple steps in fertility inhibtion. A third protein, the E. coli Host Factor I (Hfq) (Figure 1-10), is involved in countless chromosomally encoded antisense activities regulating processes such as iron metabolism (35), response to oxidative stress (111), and galactose metabolism (70). Hfq binds to both antisense (also known as small RNAs; sRNAs) and target mRNAs, often stabilizing the sRNA, and promoting pairing of the two RNAs leading to either increased translation, translational inhibition (Figure 1-10C) or degradation of the message (Figure 1-10B) (39). Interestingly, Hfq has been shown to bind traJ mRNA and affect its stability leading to down- regulation of TraJ and TraM synthesis (100). Electrophoretic mobility shift assays demonstrated that Hfq bound to the single stranded A/U rich sequence 5' to SLIIc (Figure 1-4), however binding to FinP was not

20 observed, suggesting the mechanism of repression of TraJ synthesis and F plasmid transfer is independent from the FinOP system (100). Recently, it has been shown that like FinO, Hfq has RNA chaperone activities (35). The mechanisms by which it performs these functions are discussed in the next section.

RNA chaperones

RNAs are very complex molecules which are functionally defined by the structure they adopt. Even though they have only four chemical building blocks, they are able to use local secondary structure motifs to form interactions on a global scale and fold into compact, functional structures. Motifs include helices, hairpins, internal loops, and junctions which can form tertiary interactions with nucleotides very distant in the primary sequence (42). Tertiary interactions can also globally position helices into specific conformations using interaction motifs such as the tetraloop-receptor interaction, ribose zipper and co-axial stacking interactions (42). They often involve metal ion interactions which allow phosphate-lined helices to come into close contact through charge neutralization (e.g., Mg2+ bridges) (42). A major consequence of the myriad of possible interactions results from rapid misfolding events at both the secondary and tertiary structure level (43, 99). At the secondary structure level, an RNA can become kinetically trapped into a functionally incorrect conformation (Figure 1-8i). The RNA must then perform the impossible and overcome a huge energy barrier to assume the functional conformation. On the other hand, at the tertiary level, interactions between various motifs can be unstable, due to their weak nature, so that a heterogeneous population of structures exists making the functional conformation difficult to achieve. In a world where RNA catalyzes a number of key reactions in biology, nature has designed proteins which can help RNAs achieve the impossible.

21 RNA chaperones and specific RNA binding proteins RNA chaperones are a class of RNA binding proteins which have the ability to remodel structured RNAs. Chaperone functions, which have been characterized primarily in vitro, include the resolution of kinetically trapped or misfolded ribozyme structures (e.g. T4 phage thymidylate synthase (td) group I intron), enhancement of hammerhead ribozyme activity, RNA annealing, helix destabilization, and strand exchange activities (78). Generally, chaperones bind target RNAs with weak affinity and low specificity, destabilizing RNA-RNA interactions (Figure 1-8 ii and iii) and allowing it to refold into its correct structure (78). Alternatively, they may have annealing activity where two complementary RNA strands are brought together to base-pair (Figure 1-8iv) (78). This is thought to be accomplished through a "molecular crowding" mechanism where the local concentration of complementary RNAs is increased to promote intermolecular base-pairing (78). In contrast to RNA-dependent ATPases (RNA helicases), the activities accomplished by RNA chaperones are performed in the absence of ATP. However, it has been shown that some RNA-dependent ATPases possess chaperone activity (12). RNA chaperones also perform within a certain "window of activity" where their duties can only be accomplished at a certain protein concentration range (15). Many RNA chaperones are small proteins that often contain RNA binding motifs such as the RNA recognition motif (RRM), hnRNP K homology (KH) domain, double-stranded RNA binding domain, and zinc- finger motif(15). A number of chaperones also have the propensity to form homotypic protein-protein interactions which may be needed to accelerate processes such as intermolecular annealing of complementary RNAs (15). The exact mechanism of how RNA chaperones perform their RNA re-modeling is currently unknown but a recent hypothesis is explained below.

In contrast to RNA chaperones, specific RNA binding proteins work to stabilize functional RNA tertiary structures and block or prevent mis-folding into kinetically trapped conformations (15). However, some of these proteins have RNA chaperone activity which allows them to help RNA fold into the correct structure by destabilizing local kinetic traps and

22 stabilizing the correct fold (43). An "induced fit" mechanism usually accompanies binding as both the protein and RNA make conformation changes to optimize binding interactions (99). Unlike RNA chaperones, the specific RNA binding proteins usually remain bound to their RNA to prevent mis-folding (78).

Examples of RNA chaperones This section summarizes the current understanding of the molecular mechanism underlying two of the best-studied RNA chaperones: the human immunodeficiency virus Type I (HIV-1) nucleocapsid protein and the Sm-like Hfq which mediates diverse RNA- RNA interactions in E. coli. The HIV-1 nucleocapsid protein (NC; also called NCp7 due to its 7 KDa size) is a protein which is involved in a number of key events in the life cycle of the retrovirus including packaging of RNA genome, reverse transcription, and integration (62). The HIV-1 retroviral Gag polyprotein is cleaved by a viral protease during maturation into three protein components: the matrix, capsid and the small 55 amino acid NC (80). The NC from HIV-1 is highly basic and contains two CCHC-type zinc finger domains preceded by an N-terminal 3m helix (Figure 1-9A) (80). It binds with high affinity to single stranded RNA and DNA using its zinc finger domains to contact exposed bases (62). The NC contacts double stranded RNA and tRNA through electrostatic interactions with its N- terminal 3m helix (62). The NC binding site on RNA was found to be ~ 8 nucleotides indicating that many protein molecules can bind to a single nucleic acid (62). The first requirement for the NC comes when it is still part of the Gag polyprotein. The encapsidation of genomic RNA into the immature virus particles is preceeded by recognition of the RNA encapsidation signal element (termed E) by the NC portion of Gag (50) The HIV-1 E signal resides downstream of the primer binding site at the 5' end of the RNA genome and consists of four stem-loops (50). The NC has been shown to interact specifically with SL3 of the E signal using its zinc finger domains and N-terminal 3i0 helix to bind the loop and major groove

23 portions of the stem-loop respectively (Figure 1-9A) (17). The viral RNA genome is packaged as a dimer containing two identical positive strand RNAs which are connected by a limited number of base pairs (80). Dimerization provides insurance from breaks and other damage to the viral genome and can lead to increased genetic variation as a result of template switching during the reverse transcription (79). The NC, after cleavage from the Gag polyprotein, has been shown to promote a conformational change in the dimer interaction during virus maturation leading to increased base pairing and a more stable dimer (80). The protein is also involved extensively in reverse transcription (Figure 1-9B and C). The first step in the process involves annealing of a cellular tRNA to a complementary region, called the primer binding site (PBS) near the 5' end of the positive RNA strand (Figure 1-9B) (80). The NC was shown to be important in this step as binding of the two RNAs requires a small section of the PBS to be destabilized and is carried out by the protein (62). In addition, it was determined that the NC is needed for generating the first few PBS-tRNA base pairs, which is the rate limiting step in the bimolecular interaction (62). After this step, cDNA synthesis by reverse transcriptase combined with degradation of the RNA template by its RNase H domain leads to generation of the negative strand DNA (Figure 1-9C i and ii) (62). The cDNA contains a stretch of nucleotides (r region) complementary to the 3' end of the partially digested RNA positive strand (R region). Annealing of the cDNA to this area (called negative strand transfer) leads to positive strand cDNA synthesis by reverse transcription (Figure 1-9C iv). However, in the absence of the NC, the strand transfer event has an efficiency of only 3% (80). This is because the cDNA is able to self-prime at its 3' end by folding back over on itself at the trans-activation response (TAR) element (80). This initiates positive strand cDNA synthesis resulting in a strand which is not complementary to the RNA positive strand (Figure 1-9C iii) (80). The NC increases the efficiency to 65% by destabilizing base pairs in the negative cDNA TAR region and structures in the complementary TAR region of the RNA positive strand before promoting annealing of the two (Figure 1-9C iv) (62, 80). In contrast to the binding kinetics of the PBS-tRNA reaction, the

24 rate limiting step in the strand transfer reaction is the destabilization of the TAR elements allowing annealing of the complementary strands (62). The mechanism of NC chaperone activity is still not known but clues have developed over recent years (62). High affinity binding to single stranded nucleic acids is a consequence of the zinc finger motifs, which were shown to be important in the duplex destabilizing activity of the protein (62). However, this destabilizing activity was found to be quite weak when compared to other single stranded binding proteins such as E. co// SSB (62). The NC can only melt a limited number of base pairs in its target RNAs. In contrast, the annealing activity of the protein was found to be very strong. This is due to its nucleic acid aggregating ability which is similar to other molecules such as polyamines, cobalt hexamine, and cationic detergents (62). However, the strength of the individual functions complement each other during processes such as tRNA annealing to the PBS. Here, the rate limited step is the formation of a small number of intermolecular base pairs so the NC needs only to destabilize small portions of the tRNA and PBS (62). It then uses its non­ specific nucleic acid aggregation activity to efficiently anneal the two unwound complementary sections (62). NC-mediated destabilization of the large TAR stem-loops is aided by bulged out nucleotides within the stem region (62). In contrast to other RNA chaperones, NC was found to function in a non-cooperative fashion, as it did not rely on protein-protein interactions to mediate destabilization or annealing events (62). In summary, the NC protein is a model nucleic acid chaperone as it exhibits both destabilizing and annealing activites and is vital for a great number of retroviral maturation processes.

As mentioned above, the Hfq protein is responsible for a great number of gene regulation events in E. coli and other bacteria. The protein was initially characterized as a host factor in E. coli which is needed for replication of the positive RNA strand of Qp bacteriophage (30). Later it was determined that Hfq plays a broader role in E. coli physiology, as mutation of its gene causes a number of deleterious phenotypes such as a decreased growth rate and decreased negative supercoiling at stationary phase, increased osmosensitivity, and

25 increased sensitivity to UV light (90) Hfq is an 11.2 kDa, basic protein which associates into a doughnut-like hexameric ring as determined by electron microscopy and X-ray crystallography studies (Figure 1-10A) (70, 83, 111). The protein is homologous to the Sm and Sm-like (Lsm) family of RNA binding proteins which exist in the spliceosome (111). The hexameric crystal structure from Staphylococcus aureus has a highly positively charged proximal face which was co-crystallized with a six nucleotide U-rich RNA oligomer (Figure 1-10A) (83). The E. coli structure is also hexameric but is more positively charged on the proximal and distal faces as well as the sides of the ring, which may enable binding of poly(A) tails of mRNAs (7). Hfq has been shown to bind to single stranded A/U rich sequences which are next to structured areas (7). It was suggested by Muffler et al. that Hfq binding to its mRNA targets may change their structures in a way that leads to translational activation/repression or transcript degradation in a manner similar to RNA chaperones (72). Recent in vitro and in vivo data have supported this hypothesis, demonstrating that Hfq displays the hallmark characteristics of this class of proteins. Firstly, Hfq is able to rescue splicing of the td gene of T4 bacteriophage in vivo (69) Here, in the absence of ribosomes, the presence of a stop codon in the upstream exon sequence leads to interactions with the 3' end of the intron preventing splicing from occurring. RNA chaperones have been shown to rescue splicing activity by disrupting the exon-intron trapped interaction (13). Additionally, it was demonstrated by RNase footprinting and toeprinting assays that Hfq can re-model one of its targets, the 5' UTR of E. coli ompA mRNA, preventing binding of the 30 S ribosomal subunit (Figure 1-1 OB) (69). The 30 S subunit protects the mRNA from RNase E degradation and therefore Hfq binding leads to increased degradation of the transcript (94). As shown in a previous section, Hfq is also involved in promoting sRNA-mRNA interactions to regulate gene expression for a number of host encoded processes. One target of Hfq is the small RNA OxyS which binds to rpoS mRNA which codes for the as subunit of RNA polymerase, involved in regulating a number of genes during the growth transition to stationary phase and in response to stress conditions (Figure 1-10C) (110). Zhang

26 et al. showed that Hfq led to destabilization of some parts of OxyS upon Hfq binding which enhanced pairing with rpoS mRNA (111). The structural rearrangement of the sRNA was not extensive but may be enough to account for the increased interaction between the two RNAs in vitro and in vivo. Similar enhancements occurred for other RNA pairings such as Spot42 sRNA with its target galK mRNA which is involved in sugar metabolism (70). This interaction seems to be tripartite as Hfq was shown to bind both Spot42 and galK mRNA (70). Thus Hfq might enhance interactions between the two RNAs by bringing complementary sequences into close proximity through RNA binding. In summary, Hfq has diverse mechanisms for promoting post-translational regulation of E. coli genes including destabilization and annealing activities leading to sRNA-mRNA pairing and RNA remodeling events which lead to increased message degradation.

The role of protein disorder in the mechanism of RNA chaperones Although many RNA chaperones have been identified, the mechanism of ATP-independent RNA remodeling still remains elusive. Tompa and Csermely recently suggested a link between areas of structural disorder in protein and RNA chaperones and their chaperone function (89). The authors looked at the amount of protein disorder, through Predictors of Natural Disorder Regions (PONDR) which are neural network predictors which output the amount of intrinsic disorder of a protein in a given region based on its inputted amino acid sequence (http://www.pondr.com) (63). They tested 27 RNA chaperones and found that the amount of disorder was as high as regulatory, cell-signaling, and cancer associated proteins which display high levels of natural disorder. In addition, RNA chaperones contain the longest stretches of disordered residues out of all the classes. For chaperones such as NCp9, the prion protein, and hnRNP A1, deletion of the disordered regions leads to loss of chaperone function (89). Disordered regions are well suited for RNA chaperone function as they have the potential to bind a wide variety of RNA targets in a fast, reversible manner (89). This would allow chaperones to bind to a misfolded RNA, destabilize its structure,

27 dissociate, and move on to another misfolded area. An entropy transfer mechanism may be at play as the ordering of local chaperone structure through RNA binding may subsequently provide energy needed for local unwinding (disordering) of the RNA substrate helping to rescue the RNA from folding traps (89). This mechanism may explain the RNA chaperone activity of the HIV-1 NC as this protein is largely disordered (48% disorder; (89)) and has been shown to destabilize portions of its RNA targets (62). In Chapter 2 we demonstrate that truncation mutants which remove disordered areas of the N-terminus also abolish the RNA chaperone activity of FinO. This would suggest that its chaperone activity may proceed via an entropy transfer mechanism.

Organization of the thesis

The objective of the thesis work is to determine the mechanism of FinOP mediated repression of bacterial conjugation. The details of how FinO promotes FinP-fraJ mRNA intermolecular duplexing will be discussed in Chapter 2. It was discovered that FinO has RNA chaperone activity in the N-terminal a-helix of its structure which is responsible for the antisense-sense pairing. An extensive mutational analysis is used to locate the RNA chaperone region and a model of action is proposed. In Chapter 3, details of the experimental design and results of FinO-FinP SLII complex crystallization attempts will be discussed. The details of interaction of FinO with its target RNAs will be explored in Chapter 4 using techniques such as RNase footprinting, electrophoretic mobility shift assays, and site-specific cross-linking. Chapter 5 will summarize the experimental data and propose an up to date model of FinOP inhibition of conjugation. Appendix A will test whether two recently discovered FinO structural homologs exhibit RNA chaperone activity. This work was in collaboration with Dr. Janet Wood at the University of Guelph and investigators from the Midwest Center for Structural Genomics. Finally, Appendix B will present a biophysical characterization of the envelope stress protein CpxP. This work was accomplished during the last two years of my graduate work in collaboration with Dr. Tracy Raivio from the Biological Sciences department at the University of Alberta.

28 Figure 1-1: The bacterial conjugation cycle

Donor (F+) and recipient (F) cells are shown without common pili and flagellae for simplicity, (i) A donor cell, which harbors all the genes necessary for pilus formation and DNA transfer, attaches to a recipient cell using its sex pilus. (ii) The pilus retracts bringing the donor cell into closer proximity to the recipient, (iii) The mating pair (or aggregate) is stabilized so that the two cell membranes are in close contact for DNA transfer, (iv) DNA transfer step by which one strand of the donor F-plasmid is transferred to the recipient cell through a conjugation pore employing a T4SS. (v) The recipient cell, now having a copy of the F-plasmid, dissociates from the donor cell and goes on to transfer its plasmid to other recipient cells. 29 Synthesis Pilus & Assembly Mating Pair Stabilizator

F-plasmid IS3 \7 J tfir«TTTvy • , M J Y I K B P1 V C W U N F| '•' •*- flnP tra operon

(iii) JNA Surface & Processing Entry & Transport Exclusion

Figure 1-2: Organization of the genes of the F-plasmid transfer operon and their functions The tra genes are shown by white boxes with their names above or below the box while intervening genes such as trbC are shown by black boxes for simplicity. Genes are colour coded to their specific function in the conjugation process. The traV promoter (Py) is noted as is the origin of transfer (oriT). The regulatory genes finP and finO are noted and the IS3 insertion mutation in the finO gene of the F-plasmid, which renders it non­ functional, is shown above the gene. The traG gene is split into N- and C-terminal segments which are involved in different functions.

30 1) Mating + Mating F 2) T6 phage /ae rhls F" F R100 his Lactose Streptomycin R100 O Donor strain Recipient Transient Recipient (T6sStrs) Intermediate heterozygote strain strain (Lac+/Lac_) (T6rStrr) (T6rStr8) Donor strain OD L^

Donor Plasmids in Growth on F;ac intermediate Lac/Str strains strain plates

Yes

ii his Yes WT iii R100 Yes

iv F... / R100 his Inhibited

V Yes traP (finP) vi F„- Yes mutants his

F„. / R100 Inhibited vii his

viii Yes traO ix (fisO) Fhis Yes mutants

X Ft. /R100 Yes his

Figure 1-3: Summary of early fertility inhibition experiments

Experiments were performed by Finnegan and Willetts (28). Top: Schematic showing overview of mating experiments. Donor strains containing a wild-type (i-iv), finP mutant F/ac (v-vii), or fisO mutant F/ac (viii-x) plasmid were mixed with an intermediate recipient strain in the absence or presence of additional plasmids (Fhis, R100, or both). For the case where Fhis and R100 were mixed with an F/ac plasmid, a transient heterozygote donor strain was produced which was tested for.its donor ability by mating with a recipient strain. Transconjugants contained the F/ac plasmid from donor cells and the Strr gene from recipient cells grew on lactose/streptomycin plates. T6 bacteriophage was used to kill initial donor strains containing the F/ac plasmid after formation of the transient heterozygote donor strains. Graph schematics show the growth state of individual cultures. OD=optical density and t=time. Exponential cultures are shown by rising arrow and stationary phase cultures are shown by a sigmoidal curve. Table: Summary of results from the experiments.

31 U G G U/G\ C A G AA\ A A 60 u )G) G-C 90 U /G/ A-U U Af RBS G-C c G 10 50G-C cJ G\100 A A G-C u \uj U U A-U A-U U G c C G-C SLID SLIIC G:i *G-C SUc G-C U-A A-U U-A A-U 70 U4A! G-C G-C BOAJU 20 30 40 . r, C"G C G C G 5-G UUAAAAUUUGAAAUUGAAAAUCGC ACUGUC '-'UAUC 3 1 110

5'UTRoffraJmRNA

f c C 20 u G C A u u U C G C60 C A A- u C A 50 G • c u -A G -c c -G G c SLI ,c -G A "u SLII A A C G A U G- C 10 G -U u- A G -C3C A- U A -U G- C70 U -A C- G A -U 40 U G C -G G •c „, 5-GAUA GACA GAUUUU-3 1 t RNase E

FinP RNA

Figure 1-4: Summary of the components of fertility inhibition

The genes which are involved in the regulation of conjugative transfer are shown as boxes. The traJ gene codes for the TraJ transcription factor which activates transcription of the tra operon at the promoter of traY (Py). The 5' UTR of traJ mRNA is shown above the traJ gene and is the target for the antisense RNA FinP (secondary structure shown below traJ gene). Pairing of the two RNAs presumably occludes ribosomal binding to SLIc of traJ mRNA (RBS) and prevents TraJ translation at the start codon (shown). The FinP gene is transcribed in the opposite direction of traJ within the non-coding region between the traM and traJ genes. The primary RNase E cleavage site is indicated by an arrow. The product of the finO gene, which is located at the end of the tra operon, is needed for antisense-target RNA pairing. A cartoon representation of the crystal structure of a 26-186 truncation derivative is shown (PDB ID=1DVO)(36). FinO is mainly a-helical with two short p-sheets and consists of two major domains: the long N-terminal a1 helix, and the globular region comprising the remainder of the protein. The fold is stabilized by interactions between a1 and the C-terminal region of

32 5'UTRfraJmRNA Conjugation (iii) V Inhibition

FmPltraJ mRNA duplex FinO \^/ Ny (vi) (IV)

5' UTR traJ mRNA

Figure 1-5: Model of FinOP repression of bacterial conjugation

The secondary structures of FinP antisense RNA (black) and traJ mRNA (gray) are shown. In the absence of the FinO protein co-factor, FinP is rapidly degraded by RNase E (i). The 30 S subunit of the ribosome is then free to bind to the 5' UTR of traJ mRNA (ii) and initiate translation of the TraJ transcription factor which activates expression of the tra genes allowing conjugation to proceed (iii). The presence of FinO stabilizes FinP, preventing RNase E degradation (iv). It also enhances intermolecular duplex formation between FinP and traJ mRNA (vi), possibly through a "kissing" loop intermediate (v). Formation of the duplex presumably prevents binding of the ribosome thus preventing translation of TraJ and expression of conjugation genes.

33 Anti-codon

CopT ""^ Stable Duplex Kissing complex Inhibition of Tap translation

Figure 1-6: Loop-loop interactions in antisense RNA regulatory systems

A) Crystal structure of the anti-codon loop of yeast tRNAPhe (PDB ID=4TNA) (44) in different orientations showing the U-turn structure. Cm=2' O-methylated cytidine, Gm=2' O-methylated guanosine. A cartoon representation of the phosphate backbone is shown in blue highlighting the abrupt conformational change. Hydrogen bonding between loop nucleotides is shown by dotted lines and base stacking at the anti-codon sequence is shown by black bars at the right. B) The CopA/CopT regulation of R1 plasmid replication frequency. The secondary structures of CopA antisense RNA (black) and the repA mRNA leader are shown. The CopT segment (gray) is the target of CopA. Intermolecular duplexing proceeds by a loop-loop "kissing' interaction between stem-loops II (CopA) and II' (CopT). The stable duplex contains an unusual four-helix junction. 34 A B RNAII UNA II

COIE1 DNA (j) (RNAP

RNA I

Persistent Hybridization (M) Ori of RNA II to DNA

RNA I / RNA II duplex Non-persistent hybridizaton; Ori No priming •

Figure 1-7: Regulation of ColE1 DNA replication by RNA I and Rom

A) (i) DNA replication requires RNA primers which are formed by RNA Polymerase (RNAP) 555 bp upstream of the origin of replication, (ii) RNAP produces RNA II which folds into a specific structure that allows persistent hybridization of the RNA to the DNA template strand (iii). RNAP continues to synthesize RNA II and RNase H cleaves the RNA/DNA hybrid to produce short RNA primers (iv) which initiate DNA replication (v). B) Regulation occurs via RNA I which is transcribed in the opposite direction to RNA II (i). RNA I binds to RNA II by three loop-loop interactions, the first being stabilized by the ROM protein dimer (ii). Formation of an extended duplex prevents persistent hybridization with the DNA template and primer formation (iii). C) Cartoon representation of the ROM dimer crystal structure (PDB ID=2IJK)(85).

35 IV

VII

©DA.

Figure 1-8: The action of RNA chaperones

Large RNAs such as ribozymes (above) have the tendency to fold into functionally incorrect structures (i). The RNA is kinetically trapped in this conformation because energy is required to break base-pairs in order to fold into the correct structure. RNA chaperones exist to help RNAs achieve their functional state (ii). The proteins bind non- specifically to RNAs and destabilize local secondary structure (iii). The RNA is then free to fold into the correct structure by itself or may be aided through strand annealing events by the chaperone (iv). Now that the RNA is properly folded (v), it may perform its specific function (vi and vii).

36 B

R U5 PBS U3 R

(ii)

Destabilization at (+) & (-) TAR sequences Fold-back Strand transfer Self-priming

PBS U3 R (iv) + RNA - DNA "X Figure 1-9: RNA chaperone activity of the HIV-1 nucleocapsid protein

A) Crystal structure of NCp7 (blue) bound to the SL3 hairpin of y-RNA (gray) (PDB ID=1A1T)(17). The N-terminal 3io helix interacts with the SL3 major groove, while the zinc-finger motifs bind to nucleotides within the loop. The zinc atoms are shown in white. B) NCp7 anneals a cellular tRNA molecule (black) to the primer binding site (PBS) (gray) to initiate reverse transcription from the RNA (+) strand. The complementary region is shown in red. C) NCp7 (blue circles) is involved in DNA (-) strand transfer for reverse transcription of the RNA (+) strand, (i and ii) After tRNA priming, reverse transcriptase elongates the primer producing a DNA (-) strand which is complementary to the 3' portion of the RNA (+) strand (iv). RNase H degrades the RNA after synthesis of the (-) strand (ii). However, in the absence of NCp7, the (-) strand can fold back at its trans-activation region (TAR), which is highly structured, and self-prime to make a DNA (+) strand which is not complementary to the RNA (+) strand, preventing further reverse transcripition (iii). NCp7 resolves this by destabilizing DNA (-) and RNA (+) strand TAR sequences allowing the complementary sequences to anneal at the (+) strand R and (-) strand r areas (iv). U5 and U3 are unique 5' and 3' genomic sequences in the (+) strand. Elements of this figure adapted from (62, 80)

37 A

(i) (ii) B Fast Growth

Protection from RNase E SD AUG ompA mRNA °t- Degradation by RNase E ompA structural Hfq binding Rearrangement

5'J JL_1 OxyS sRNA Hfq-mediated ~° Hfq binding OxyS destabilization rpo$ mRNA

Hfq-mediated Intermolecular Duplex Formation & Translational inhibition

Figure 1-10: RNA chaperone activities of Hfq

A) Crystal structure of the Hfq hexamerfrom E. coli (PDB ID=1HK9)(82)(left) and hexamer with an 5'-(AU5G)-3' RNA oligonucleotide from S. aureus (PDB ID=1KQ2)(83) (right). Left: Surface representation showing the molecular shape of the distal face of the Hfq hexamer with a cartoon representation of secondary structure. One subunit is shown in blue. Right: Surface representation of the proximal face of the Hfq hexamer with the RNA oligonucleotide bound in blue. B) Hfq increases the degradation of ompA mRNA. During fast growth conditions, the 30 S ribosomal subunit binds to ompA mRNA, protecting the message from RNase E degradation and leading to translation. Under slow growth stress conditions, the concentration of Hfq increases and binds to ompA mRNA rearranging the structure and preventing ribosomal binding. In addition, the mRNA is exposed to RNase E and is subsequently degraded upon Hfq binding. C) Hfq enhances antisense-target mRNA pairing. Hfq binds to the OxyS antisense RNA (black) destabilizing portions of two stem loops which enhances its interaction with one of its targets, rpoS mRNA (gray). Hfq mediates the pairing of the two RNAs through an unknown stoichiometry and mechanism.

38 References

1. Achtman, M. (1975) Mating aggregates in Escherichia coli conjugation. J Bacteriol, 123: 505-515.

2. Adelberg, E.A. and Pittard, J. (1965) Chromosome Transfer in Bacterial Conjugation. Bacteriol Rev, 29: 161-172.

3. Anthony, K.G., Klimke, W.A., Manchak, J., and Frost, L.S. (1999) Comparison of proteins involved in pilus synthesis and mating pair stabilization from the related plasmids F and R100-1: insights into the mechanism of conjugation. J Bacteriol, 181: 5149-5159.

4. Blomberg, P., Nordstrom, K., and Wagner, E.G. (1992) Replication control of plasmid R1: RepA synthesis is regulated by CopA RNA through inhibition of leader peptide translation. EMBO J, 11: 2675-2683.

5. Brantl, S. (2002) Antisense RNAs in plasmids: control of replication and maintenance. Plasmid, 48: 165-173.

6. Brantl, S. (2007) Regulatory mechanisms employed by cis- encoded antisense RNAs. CurrOpin Microbiol, 10: 102-109.

7. Brennan, R.G. and Link, T.M. (2007) Hfq structure, function and ligand binding. CurrOpin Microbiol, 10: 125-133.

8. Brunei, C, Marquet, R., Romby, P., and Ehresmann, C. (2002) RNA loop-loop interactions as dynamic functional motifs. Biochimie, 84: 925-944.

9. Camacho, E.M. and Casadesus, J. (2002) Conjugal transfer of the virulence plasmid of Salmonella enterica is regulated by the leucine-responsive regulatory protein and DNA adenine methylation. Mol Microbiol, 44: 1589-1598.

10. Carpousis, A.J. (2007) The RNA degradosome of Escherichia coli: an mRNA-degrading machine assembled on RNase E. Annu Rev Microbiol, 61: 71-87.

11. Cheah, K.C. and Skurray, R. (1986) The F plasmid carries an IS3 insertion within finO. J Gen Microbiol, 132: 3269-3275.

12. Chu, V.B. and Herschlag, D. (2008) Unwinding RNA's secrets: advances in the biology, physics, and modeling of complex RNAs. CurrOpin Struct Biol, 18: 305-314.

13. Clodi, E., Semrad, K., and Schroeder, R. (1999) Assaying RNA chaperone activity in vivo using a novel RNA folding trap. EMBO J, 18: 3776-3782. 39 14. Clowes, R.C. (1961) Colicine factors as fertility factors in bacteria: Escherichia coliK-12. Nature, 190: 988-989.

15. Cristofari, G. and Darlix, J.L. (2002) The ubiquitous nature of RNA chaperone proteins. Prog Nucleic Acid Res Mol Biol, 72: 223-268.

16. Curtiss, R., 3rd (1969) Bacterial conjugation. Annu Rev Microbiol, 23: 69-136.

17. De Guzman, R.N., Wu, Z.R., Stalling, C.C., Pappalardo, L, Borer, P.N., and Summers, M.F. (1998) Structure of the HIV-1 nucleocapsid protein bound to the SL3 psi-RNA recognition element. Science, 279: 384-388.

18. Dempsey, W.B. (1987) Transcript analysis of the plasmid R100 traJ and finP genes. Mol Gen Genet, 209: 533-544.

19. Dempsey, W.B. and Mclntire, S.A. (1983) The finO gene of antibiotic resistance plasmid R100. Mol Gen Genet, 190: 444- 451.

20. Durrenberger, M.B., Villiger, W., and Bachi, T. (1991) Conjugational junctions: morphology of specific contacts in conjugating Escherichia coli bacteria. J Struct Biol, 107: 146-156.

21. Eddy, S.R. (2001) Non-coding RNA genes and the modern RNA world. Nat Rev Genet, 2: 919-929.

22. Egawa, R., and Hirota, Y. (1962) Inhibition of fertility by multiple drug-resistance factor (R) in Escherichia coli K12. Japanese Journal of Genetics, 37: 66.

23. Eguchi, Y., Itoh, T., and Tomizawa, J. (1991) Antisense RNA. Annu Rev Biochem, 60: 631-652.

24. Eguchi, Y. and Tomizawa, J. (1991) Complexes formed by complementary RNA stem-loops. Their formations, structures and interaction with ColE1 Rom protein. J Mol Biol, 220: 831-842.

25. Finlay, B.B., Frost, L.S., Paranchych, W., and Willetts, N.S. (1986) Nucleotide sequences of five IncF plasmid f/nP alleles. J Bacteriol, 167: 754-757.

26. Finnegan, D. and Willetts, N. (1972) The nature of the transfer inhibitor of several F-like plasmids. Mol Gen Genet, 119: 57-66.

27. Finnegan, D. and Willetts, N. (1973) The site of action of the F transfer inhibitor. Mol Gen Genet, 127: 307-316.

40 28. Finnegan, D.J. and Willetts, N.S. (1971) Two classes of F/ac mutants insensitive to transfer inhibition by an F-like R factor. Mol Gen Genet, 111: 256-264.

29. Franch, T., Petersen, M., Wagner, E.G., Jacobsen, J.P., and Gerdes, K. (1999) Antisense RNA regulation in prokaryotes: rapid RNA/RNA interaction facilitated by a general U-turn loop structure. J Mol Biol, 294: 1115-1125.

30. Franze de Fernandez, M.T., Eoyang, L, and August, J.T. (1968) Factor fraction required for the synthesis of bacteriophage Qbeta- RNA. Nature, 219: 588-590.

31. Frost, L, Lee, S., Yanchar, N., and Paranchych, W. (1989) finP and fisO mutations in FinP anti-sense RNA suggest a model for FinOP action in the repression of bacterial conjugation by the F/ac plasmid JCFLO. Mol Gen Genet, 218: 152-160.

32. Frost, L.S., Finlay, B.B., Opgenorth, A., Paranchych, W., and Lee, J.S. (1985) Characterization and sequence analysis of pilin from F-like plasmids. J Bacteriol, 164: 1238-1247.

33. Frost, L.S., Ippen-lhler, K., and Skurray, R.A. (1994) Analysis of the sequence and gene products of the transfer region of the F sex factor. Microbiol Rev, 58: 162-210.

34. Gasson, M.J. and Willetts, N.S. (1975) Five control systems preventing transfer of Escherichia coli K-12 sex factor F. J Bacteriol, 122: 518-525.

35. Geissmann, T.A. and Touati, D. (2004) Hfq, a new chaperoning role: binding to messenger RNA determines access for small RNA regulator. EMBO J, 23: 396-405.

36. Ghetu, A.F., Gubbins, M.J., Frost, L.S., and Glover, J.N. (2000) Crystal structure of the bacterial conjugation repressor FinO. Nat Struct Biol, 7: 565-569.

37. Ghetu, A.F., Gubbins, M.J., Oikawa, K., Kay, CM., Frost, L.S., and Glover, J.N. (1999) The FinO repressor of bacterial conjugation contains two RNA binding regions. Biochemistry, 38: 14036-14044.

38. Gilbert, W. and Dressier, D. (1968) DNA replication: the rolling circle model. Cold Spring Harb Symp Quant Biol, 33: 473-484.

39. Gottesman, S. (2004) The small RNA regulators of Escherichia coli: roles and mechanisms*. Annu Rev Microbiol, 58: 303-328.

41 40. Gubbins, M.J., Arthur, D.C., Ghetu, A.F., Glover, J.N., and Frost, L.S. (2003) Characterizing the structural features of RNA/RNA interactions of the F-plasmid FinOP fertility inhibition system. J Biol Chem, 278: 27663-27671.

41. Hammond, S.M. (2005) Dicing and slicing: the core machinery of the RNA interference pathway. FEBS Lett, 579: 5822-5829.

42. Hendrix, D.K., Brenner, S.E., and Holbrook, S.R. (2005) RNA structural motifs: building blocks of a modular biomolecule. Q Rev Biophys, 38: 221-243.

43. Herschlag, D. (1995) RNA chaperones and the RNA folding problem. J Biol Chem, 270: 20871-20874.

44. Hingerty, B., Brown, R.S., and Jack, A. (1978) Further refinement of the structure of yeast tRNAPhe. J Mol Biol, 124: 523-534.

45. Hiraga, S. and Saitoh, T. (1975) F deoxyribonucleic acid transferred to recipient cells in the presence of rifampin. J Bacteriol, 121: 1000-1006.

46. Hjalt, T. and Wagner, E.G. (1992) The effect of loop size in antisense and target RNAs on the efficiency of antisense RNA control. Nucleic Acids Res, 20: 6723-6732.

47. Ihler, G. and Rupp, W.D. (1969) Strand-specific transfer of donor DNA during conjugation in E. coli. Proc Natl Acad Sci USA, 63: 138-143.

48. Jerome, L.J. and Frost, L.S. (1999) In vitro analysis of the interaction between the FinO protein and FinP antisense RNA of F-like conjugative plasmids. J Biol Chem, 274: 10356-10362.

49. Jerome, L.J., van Biesen, T., and Frost, L.S. (1999) Degradation of FinP antisense RNA from F-like plasmids: the RNA-binding protein, FinO, protects FinP from ribonuclease E. J Mol Biol, 285: 1457-1473.

50. Jewell, N.A. and Mansky, L.M. (2000) In the beginning: genome recognition, RNA encapsidation and the initiation of complex retrovirus assembly. J Gen Virol, 81: 1889-1899.

51. Johnson, D., Everett, R., and Willetts, N. (1981) Cloning of F DNA fragments carrying the origin of transfer oriT and the fertility inhibition gene finP. J Mol Biol, 153: 187-202.

52. Kingsman, A. and Willetts, N. (1978) The requirements for conjugal DNA synthesis in the donor strain during F/ac transfer. J Mol Biol, 122: 287-300.

42 53. Kloosterman, W.P. and Plasterk, R.H. (2006) The diverse functions of microRNAs in animal development and disease. Dev Cell,U: 441-450.

54. Kolb, F.A., Engdahl, H.M., Slagter-Jager, J.G., Ehresmann, B., Ehresmann, C, Westhof, E., Wagner, E.G., and Romby, P. (2000) Progression of a loop-loop complex to a four-way junction is crucial for the activity of a regulatory antisense RNA. EMBO J, 19: 5905-5915.

55. Kolb, F.A., Malmgren, C, Westhof, E., Ehresmann, C, Ehresmann, B., Wagner, E.G., and Romby, P. (2000) An unusual structure formed by antisense-target RNA binding involves an extended kissing complex with a four-way junction and a side-by- side helical alignment. RNA, 6: 311-324.

56. Kolb, F.A., Westhof, E., Ehresmann, C, Ehresmann, B., Wagner, E.G., and Romby, P. (2001) Bulged residues promote the progression of a loop-loop interaction to a stable and inhibitory antisense-target RNA complex. Nucleic Acids Res, 29: 3145- 3153.

57. Koraimann, G., Koraimann, C, Koronakis, V., Schlager, S., and Hogenauer, G. (1991) Repression and derepression of conjugation of plasmid R1 by wild-type and mutated FinP antisense RNA. Mol Microbiol, 5: 77-87.

58. Koraimann, G., Teferle, K., Markolin, G., Woger, W., and Hogenauer, G. (1996) The FinOP repressor system of plasmid R1: analysis of the antisense RNA control of traJ expression and conjugative DNA transfer. Mol Microbiol, 21: 811-821.

59. Lawley, T.D., Klimke, W.A., Gubbins, M.J., and Frost, L.S. (2003) F factor conjugation is a true type IV secretion system. FEMS Microbiol Lett, 224: 1-15.

60. Lederberg, J., and Tatum, E.L. (1946) Gene recombination in Escherichia coli. Nature, 158: 558.

61. Lee, S.H., Frost, L.S., and Paranchych, W. (1992) FinOP repression of the F plasmid involves extension of the half-life of FinP antisense RNA by FinO. Mol Gen Genet, 235: 131-139.

62. Levin, J.G., Guo, J., Rouzina, I., and Musier-Forsyth, K. (2005) Nucleic acid chaperone activity of HIV-1 nucleocapsid protein: critical role in reverse transcription and molecular mechanism. Prog Nucleic Acid Res Mol Biol, 80: 217-286.

43 63. Li, X., Romero, P., Rani, M., Dunker, A.K., and Obradovic, Z. (1999) Predicting Protein Disorder for N-, C-, and Internal Regions. Genome Inform Ser Workshop Genome Inform, 10: 30- 40.

64. Llosa, M., Gomis-Ruth, F.X., Coll, M., and de la Cruz Fd, F. (2002) Bacterial conjugation: a two-step mechanism for DNA transport. Mol Microbiol, 45: 1-8.

65. Lu, J. and Frost, L.S. (2005) Mutations in the C-terminal region of TraM provide evidence for in vivo TraM-TraD interactions during F-plasmid conjugation. J Bacteriol, 187: 4767-4773.

66. Maneewannakul, K., Maneewannakul, S., and Ippen-lhler, K. (1993) Synthesis of F pilin. J Bacteriol, 175: 1384-1391.

67. Mclntire, S.A. and Dempsey, W.B. (1987) Fertility inhibition gene of plasmid R100. Nucleic Acids Res, 15: 2029-2042.

68. Meynell, E., Meynell, G.G., and Datta, N. (1968) Phylogenetic relationships of drug-resistance factors and other transmissible bacterial plasmids. Bacteriol Rev, 32: 55-83.

69. Moll, I., Leitsch, D., Steinhauser, T., and Blasi, U. (2003) RNA chaperone activity of the Sm-like Hfq protein. EMBO Rep, 4: 284- 289.

70. Moller, T., Franch, T., Hojrup, P., Keene, D.R., Bachinger, H.P., Brennan, R.G., and Valentin-Hansen, P. (2002) Hfq: a bacterial Sm-like protein that mediates RNA-RNA interaction. Mol Cell, 9: 23-30.

71. Moore, D., Sowa, B.A., and Ippen-lhler, K. (1981) Location of an F-pilin pool in the inner membrane. J Bacteriol, 146: 251-259.

72. Muffler, A., Traulsen, D.D., Fischer, D., Lange, R., and Hengge- Aronis, R. (1997) The RNA-binding protein HF-I plays a global regulatory role which is largely, but not exclusively, due to its role in expression of the sigmaS subunit of RNA polymerase in Escherichia coli. J Bacteriol, 179: 297-300.

73. Mullineaux, P., and Willetts, N.S., Promoters in the transfer region of plasmid F, in Plasmids in bacteria, D.R. Helinski, Cohen, S.N., Clewell, D.B., Jackson, D.A., and Hollaender, A. 1985, Plenum Press: New York. p. 605-614.

74. Nicholson, A.W. (1999) Function, mechanism and regulation of bacterial ribonucleases. FEMS Microbiol Rev, 23: 371-390.

44 75. Ozeki, H. and Howarth, S. (1961) Colicine factors as fertility factors in bacteria: Salmonella typhimurium strain LT2. Nature, 190: 986-988.

76. Persson, C, Wagner, E.G., and Nordstrom, K. (1990) Control of replication of plasmid R1: structures and sequences of the antisense RNA, CopA, required for its binding to the target RNA, CopT. EMBO J, 9: 3767-3775.

77. Quigley, G.J. and Rich, A. (1976) Structural domains of transfer RNA molecules. Science, 194: 796-806.

78. Rajkowitsch, L, Chen, D., Stampfl, S., Semrad, K., Waldsich, C, Mayer, O., Jantsch, M.F., Konrat, R., Blasi, U., and Schroeder, R. (2007) RNA chaperones, RNA annealers and RNA helicases. RNA Biol, 4: 118-130.

79. Rein, A. (2004) Take two. Nat Struct Mol Biol, 11:1034-1035.

80. Rein, A., Henderson, L.E., and Levin, J.G. (1998) Nucleic-acid- chaperone activity of retroviral nucleocapsid proteins: significance for viral replication. Trends Biochem Sci, 23: 297-301.

81. Sandercock, J.R. and Frost, L.S. (1998) Analysis of the major domains of the F fertility inhibition protein, FinO. Mol Gen Genet, 259: 622-629.

82. Sauter, C, Basquin, J., and Suck, D. (2003) Sm-like proteins in Eubacteria: the crystal structure of the Hfq protein from Escherichia coli. Nucleic Acids Res, 31: 4091-4098.

83. Schumacher, M.A., Pearson, R.F., Moller, T., Valentin-Hansen, P., and Brennan, R.G. (2002) Structures of the pleiotropic translational regulator Hfq and an Hfq-RNA complex: a bacterial Sm-like protein. EMBO J, 21: 3546-3556.

84. Strohmaier, H., Noiges, R., Kotschan, S., Sawers, G., Hogenauer, G., Zechner, E.L., and Koraimann, G. (1998) Signal transduction and bacterial conjugation: characterization of the role of ArcA in regulating conjugative transfer of the resistance plasmid R1. J Mol Biol, 277: 309-316.

85. Struble, E.B., Ladner, J.E., Brabazon, D.M., and Marino, J.P. (2008) New crystal structures of ColE1 Rom and variants resulting from mutation of a surface exposed residue: Implications for RNA- recognition. Proteins, 72: 761-768.

86. Thompson, R. and Taylor, L. (1982) Promoter mapping and DNA sequencing of the F plasmid transfer genes traM and traJ. Mol Gen Genet, 188: 513-518.

45 87. Timmis, K.N., Andres, I., and Achtman, M. (1978) Fertility repression of F-like conjugative plasmids: physical mapping of the R6-5 finO and finP cistrons and identification of the FinO protein. Proc Natl Acad Sci USA, 75: 5836-5840.

88. Tomizawa, J. (1990) Control of ColE1 plasmid replication. Interaction of Rom protein with an unstable complex formed by RNA I and RNA II. J Mol Biol, 212: 695-708.

89. Tompa, P. and Csermely, P. (2004) The role of structural disorder in the function of RNA and protein chaperones. FASEB J, 18: 1169-1175.

90. Tsui, H.C., Leung, H.C., and Winkler, M.E. (1994) Characterization of broadly pleiotropic phenotypes caused by an hfq insertion mutation in Escherichia coli K-12. Mol Microbiol, 13: 35-49.

91. van Biesen, T. and Frost, L.S. (1992) Differential levels of fertility inhibition among F-like plasmids are related to the cellular concentration of finO mRNA. Mol Microbiol, 6: 771-780.

92. van Biesen, T. and Frost, L.S. (1994) The FinO protein of IncF plasmids binds FinP antisense RNA and its target, traJ mRNA, and promotes duplex formation. Mol Microbiol, 14: 427-436.

93. van Biesen, T., Soderbom, F., Wagner, E.G., and Frost, L.S. (1993) Structural and functional analyses of the FinP antisense RNA regulatory system of the F conjugative plasmid. Mol Microbiol, 10: 35-43.

94. Vytvytska, O., Moll, I., Kaberdin, V.R., von Gabain, A., and Blasi, U. (2000) Hfq (HF1) stimulates ompA mRNA decay by interfering with ribosome binding. Genes Dev, 14: 1109-1118.

95. Wagner, E.G., Altuvia, S., and Romby, P. (2002) Antisense RNAs in bacteria and their genetic elements. Adv Genet, 46: 361-398.

96. Watanabe, T. (1963) Episome-Mediated Transfer of Drug Resistance in Enterobacteriaceae. Vi. High-Frequency Resistance Transfer System in Escherichia Coli. J Bacteriol, 85: 788-794.

97. Watanabe, T. (1963) Infective heredity of multiple drug resistance in bacteria. Bacteriol Rev, 27: 87-115.

98. Watanabe, T., Nishida, H., Ogata, C, Arai, T., and Sato, S. (1964) Episome-Mediated Transfer of Drug Resistance in Enterobacteriaceae. Vii. Two Types of Naturally Occurring R Factors. J Bacteriol, 88: 716-726.

46 99. Weeks, K.M. (1997) Protein-facilitated RNA folding. Curr Opin Struct Biol, 7: 336-342.

100. Will, W.R. and Frost, L.S. (2006) Hfq is a regulator of F-plasmid TraJ and TraM synthesis in Escherichia coli. J Bacterid, 188: 124-131.

101. Willetts, N. (1977) The transcriptional control of fertility in F-like plasmids. J Mol Biol, 112: 141-148.

102. Willetts, N., Bacterial conjugation. A historical perspective, D.B. Clewell. 1993, New York: Plenum Press, 1-22.

103. Willetts, N., and Maule, J. (1986) Specificities of IncF plasmid conjugation genes. Genetical Research, 47: 1-11.

104. Willetts, N., Maule, J., and Mclntire, S. (1976) The genetic locations of traO, finP, and tra-4 on the E. coli K12 sex factor F. Genetical Research, 26: 255-263.

105. Willetts, N. and Wilkins, B. (1984) Processing of plasmid DNA during bacterial conjugation. Microbiol Rev, 48: 24-41.

106. Williams, S.L. and Schildbach, J.F. (2007) TraY and integration host factor oriT binding sites and F conjugal transfer: sequence variations, but not altered spacing, are tolerated. J Bacteriol, 189: 3813-3823.

107. Yang, Z. and Wu, J. (2006) Small RNAs and development. Med SciMonit, 12: RA125-129.

108. Yoshioka, Y., Ohtsubo, H., and Ohtsubo, E. (1987) Repressor gene finO in plasmids R100 and F: constitutive transfer of plasmid F is caused by insertion of IS3 into F finO. J Bacteriol, 169: 619- 623.

109. Zahrl, D., Wagner, A., Tscherner, M., and Koraimann, G. (2007) GroEL plays a central role in stress-induced negative regulation of bacterial conjugation by promoting proteolytic degradation of the activator protein TraJ. J Bacteriol, 189: 5885-5894.

110. Zhang, A., Altuvia, S., Tiwari, A., Argaman, L, Hengge-Aronis, R., and Storz, G. (1998) The OxyS regulatory RNA represses rpoS translation and binds the Hfq (HF-I) protein. EMBO J, 17: 6061- 6068.

111. Zhang, A., Wassarman, K.M., Ortega, J., Steven, A.C., and Storz, G. (2002) The Sm-like Hfq protein increases OxyS RNA interaction with target mRNAs. Mol Cell, 9: 11-22.

47 48 Chapter 2

FinO acts as an RNA chaperone to facilitate F'mP-traJ mRNA interactions1

Overview

Repression of bacterial conjugation occurs through a two-component system involving the antisense RNA FinP and its protein co-factor FinO. FinP forms an intermolecular duplex with the 5' UTR of traJ mRNA presumably leading to occlusion of the ribosome binding site, preventing translation of the TraJ transcriptional activator and expression of downstream conjugation genes. FinO is required to protect FinP from cellular degradation and to enhance the rate of duplexing between the antisense RNA and its target traJ mRNA. The mechanistic details of how FinO facilitates RNA-RNA duplexing are currently not known. Here, it is shown that FinO is able perform strand exchange and intermolecular annealing activities on FinP SLII derivatives in vitro. This strongly suggests that FinO is able to destabilize stable RNA substrates thus exhibiting RNA chaperone behavior. Using alanine scanning mutagenesis studies, the location of the activity is mapped to a 10 amino acid region in the N-terminal a-helix of FinO consisting primarily of lysine residues and a solvent exposed tryptophan residue. Removal of this region from FinO completely disrupts its chaperoning activity. Interestingly, the catalytically dead mutant, FinO 45-186, binds 20-fold tighter to FinP SLII than wild-type FinO. Furthermore, a mutant which contains the catalytic region but lacks a 25 residue disordered region of the N-terminus (FinO 26-186) binds 4-fold tighter and has approximately 10-fold less annealing activity than wild-type FinO. The mutations that disrupt in vitro strand exchange and annealing activities also affect bacterial conjugation in vivo highlighting the importance of the N-terminal region. Taken together, the data suggests an inverse relationship between the chaperone and RNA binding activities of FinO and

1 Part of this work was previously published: Arthur D.C., Ghetu, A.F., Gubbins, M.J., Edwards, R.A., Frost, L.S., Glover J.N. (2003) EMBO J. 22: 6346-6355.

49 hints toward an entropy transfer mechanism where FinO uses its binding energy to destabilize and re-anneal its RNA substrates.

Introduction

In order for antisense regulation of gene expression to proceed, pairing between the antisense and target mRNA must occur efficiently. In RNA-only mechanisms, loop-loop interactions enhance the recognition process allowing the formation of the first few intermolecular base-pairs (2) and duplex progression is expedited using bulges and mismatched base- pairings within the RNA structures (19). In the FinOP inhibition system of bacterial conjugation, the protein co- factor FinO substitutes for these structured RNA elements leading to a significant 50-fold enhancement of the rate of intermolecular duplexing between FinP antisense RNA and its target traJ mRNA in vitro (7, 29). FinP and traJ mRNA can however form an intermolecular duplex in vitro in the absence of FinO. FinP has a 5'-YUNR-3' (Y=pyrimidine, N=any base and R=purine) U-turn motif in both SLI and SLII domains (Figure 2-1) which would aid in recognition of the 5' UTR of traJ mRNA (5). In addition, both SLI from FinP and SLIc from traJ mRNA contain mismatches in the stems (A12/A27 in FinP; U86/U101 and U84/U103 in SLIc)(Figure 2-1) which would aid in intermolecular duplex progression (10). However, SLII and SLIIc contain 14 and 12 contiguous base pairs respectively (Figure 2-1), which would present a huge kinetic barrier to overcome for complete duplex formation. The two 1 1 stem-loops are very stable (AGSui = -28 kcalmol" and AGSLMC = -23 kcalmol" ) (RNAfold) (9) and would require base pairs to be destabilized in order for intermolecular strand annealing to occur.

It is therefore hypothesized that FinO is able to overcome the kinetic barrier by destabilizing base pairs within the SLII and SLIIc stem-loops of FinP and traJ mRNA respectively. This would account for the large increase in the bimolecular association rate constant (kapp; k2 in this paper) between the two RNAs in its presence. Interestingly, previous studies have shown that the N-terminal 25 amino acids are important in FinP-fraJ mRNA intermolecular duplexing activity. Truncation to FinO 26-186 (used for X-ray

50 crystallographic studies) reduced the k2 aproximately nine-fold compared to wild-type FinO (7). The goal of the present study is to expand on the previous knowledge to obtain mechanistic details of how FinO increases the rate of F'mP-traJ mRNA duplexing. Does FinO destabilize base pairs in SLII and SLIIc? If it does, is it actively involved in intermolecular annealing events or does it just keep strands apart so complementary sequences can come together? Site- directed mutagenesis was used to explore these questions producing FinO mutants which have point mutations or portions of the N-terminus deleted. The mutants were then used in in vitro strand exchange and annealing assays using SLII derivatives and in vivo conjugation assays to characterize their activity. It is shown here that FinO has RNA chaperone activity in a stretch of amino acids in its N-terminal a-helix containing a number of lysines and a solvent exposed tryptophan residue. Using electrophoretic mobility shift assays, it is shown that there is an inverse relationship between the binding and chaperone activity of FinO, suggesting that the protein may use its binding energy to remodel its RNA substrates during the intermolecular duplexing process.

Results

FinO facilitates RNA strand exchange To test whether FinO might be involved in destabilization events during F\nP-traJ mRNA duplexing, we used a strand exchange assay based on an unwinding assay for DEAH-box splicing factors (32, 33). Figure 2-2A gives an overview of the experimental setup. An SLII duplex derivative (SII) was designed which is missing the seven nucleotide loop region but retains the 14 base pair stem, 5' spacer and 3' tail (Figure 2-1). As mentioned in Chapter 1, FinO recognizes the overall shape of the RNA and requires the tails for high affinity binding (15). Using electrophoretic mobility shift assays (EMSAs), it was determined that FinO binds to SI I with an equal affinity as SLII (data not shown). In the assay, one of the strands in the duplex, SII(A), is radiolabeled with y32P-ATP and annealed to a non-radiolabeled (cold) SII(B) strand (Figure 2-2A). After its formation, the duplex is incubated at

51 37°C (well below the Tm of the duplex, >50°C; data not shown) for 2 hours with an excess of cold SII(A) strand in the absence or presence of wild-type FinO (herein referred to as FinO) or mutants. If strand exchange occurs, the radiolabeled SII(A) strand will be exchanged for the excess cold SII(A) strand forming a cold Sll duplex thereby isolating the radiolabeled SII(A) strand. EMSAs showed that FinO does not form a stable complex with SII(A) (data not shown). The reaction was stopped by adding sodium dodecylsulfate (SDS) to denature FinO and the products of the reaction were resolved on a 15% native polyacrylamide gel. The kinetics of strand exchange was monitored by removing aliquots of each reaction at various time points, stopping them with SDS and loading on a continuously running gel. Apparent first-order rate constants (ki) were calculated from the data using the equation shown in the Materials and Methods. The results of the experiments (shown in Figure 2-2B, 2-2C, and Table 2-1) demonstrate that FinO is able to promote strand exchange of the Sll duplex in a time dependent manner. In the absence of the protein, strand exchange does not occur within the two hour time course. Upon the addition of FinO the strand exchange rate constant increases 25-fold, suggesting FinO may destabilize portions of the stable duplex. FinO strand exchange proceeded through an ATP-independent mechanism (data not shown) which is characteristic of many RNA chaperones (4). Decreasing temperature and the addition of increasing amounts of Mg2+ to the reaction decreased the levels of strand exchange significantly probably due to stabilization of the Sll duplex (Figure 2-2D). Also, the addition of four base pairs to Sll (SII+4 duplex; Figure 2-1) reduced the FinO-mediated strand exchange rate likely due to the increased stability of the duplex compared to Sll (Table 2-1). This also shows that FinO probably acts locally to destabilize base pairs and is not processive along the helix. Since the single stranded 5' spacer and 3' tails were shown to be important in FinO binding to its RNA substrates (15), we wanted to test whether these areas were important for strand exchange. An Sll duplex was designed which lacked the single stranded tails (SNA) (Figure 2-1). The radiolabeled modified duplex was then incubated with an excess of SII(A) which lacked the 5' spacer, called SllA(A), and the strand exchange rate

52 constants were determined in the absence or presence of FinO. Surprisingly, FinO was able to perform strand exchange on the SUA as well or better than Sll (Figure 2-2B and 2-2C). It was shown by Jerome et al. that removal of the tails from SLII decreased FinO-GST binding by 25-fold (15). In the current study, a stable FinO-SllA complex was not detected by EMSAs at 4°C (data not shown) as higher concentrations of FinO (>0.7 jaM) produced smeary shifts and aggregates. Therefore, it appears that the recognition of the double-stranded RNA by FinO is not the rate-limiting step in the strand exchange reaction and that the single stranded 5' spacer and 3' tail regions of Sll are not important for the process. It also became apparent that strand exchange is highly dependent on the concentration of FinO. The level of strand exchange was examined at various concentrations of FinO (Figure 2-2E). Reactions employing Sll or SNA duplexes were incubated with various concentrations of FinO. After two hours at 37°C the reactions were stopped by addition of SDS and the products were resolved by native PAGE. It can be seen that strand exchange does not occur until a final FinO concentration of 0.25 ^M. Optimal activity occurs at 1 |j.M, however activity starts to decrease again at 4 JJ.M, indicating that strand exchange is concentration dependent. The decrease in strand exchange levels may be due to non-specific interactions between FinO and its RNA substrates. EMSAs showed that FinO can form non-specific aggregates on the RNA at similar protein concentrations (data not shown).

FinO strand exchange activity resides in the N-terminal cc-helix It was shown previously that the FmP-traJ mRNA intermolecular duplexing rate constant decreased approximately nine-fold when the first 25 amino acids of FinO are deleted (7). Knowing this and that FinO performs strand exchange on SLII-derived RNA duplexes, we wondered if various truncations of the N-terminus had an effect on strand exchange. FinO truncation mutants were produced which had the first 25 (FinO 26-186) and 44 (FinO 45-186) amino acids deleted and their effect on Sll duplex strand exchange was measured (Figure 2-2B and 2-2C, Table 2-1). Interestingly, the effect of removing the N-terminal 25 residues, which decreased the strand exchange rate by ~ 1.5-fold compared to FinO, was not as large as for

53 the F\r\P-traJ mRNA duplexing reaction. This indicated that this region of FinO is more important for the annealing component of the duplexing reaction, whereas the strand exchange activity was further towards the C- terminus. In stark contrast, the effect of removing the first 44 amino acids completely abrogated strand exchange activity (Figure 2-2B). Strand exchange activity was actually less than in the absence of any FinO protein, indicating that the construct seemed to stabilize the RNA duplex. This indicates that the area responsible for strand exchange is located in the N- terminal a-helix between residues 26-44. An additional FinO construct was produced comprising only the N- terminal a-helix 1 (FinO 1-61) (Figure 2-1). Knowing that strand exchange activity resided in a segment of the N-terminus, we wanted to test whether this fragment was sufficient for strand exchange or if the rest of the protein was required. Surprisingly, FinO 1-61 does exhibit strand exchange of the Sll duplex (Figure 2-2B, 2-2C, Table 2-1), however it appears that the rest of FinO is required for full activity. This may be due to the RNA binding activity of the construct as it was previously shown that FinO 1-61 does not bind with high affinity to SLII (8). Similarly, FinO 1-61 bound very poorly to the Sll substrate at 4°C (Table 2-1).

Alanine scanning mutagenesis identifies the strand exchange region of FinO It was shown above that strand exchange activity is located between residues 26-44 in the N-terminal helix of FinO. We attempted to identify which residues in this area were important for strand exchange using site- direct mutagenesis to create double amino acid to alanine substitutions within the region spanning residues 26-44. A summary of the results is shown in Figure 2-3A and 2-3B. Strand exchange reactions were performed as before using the Sll duplex RNA in the absence or presence of the double alanine mutants or FinO and FinO 45-186 as positive and negative controls respectively. Each reaction was incubated at 37°C for two hours before stopping with SDS and running on a 15% native gel. For each mutant, the levels of strand exchange (percent of SII(A) released) were quantified and the results are presented relative to FinO (Figure 2-3B). The results show that the P34A/K35A, W36A, K37AA/38A, K39A/K40A, and Q41A/K42A mutants

54 had the greatest effect on the levels of strand exchange. The T32A/P33A and L43A mutations also had a small effect on strand exchange activity. Therefore, the area encompassing residues 32-42, which maps to the very end of the N-terminal helix on the FinO 26-186 crystal structure (7), is the catalytic area of FinO. Single alanine point mutations were created to identify individual residues which are critical to strand exchange (Figure 2-3C and 2- 3D). As with the double mutation experiments, the W36A mutation had the biggest effect on strand exchange while mutations P34A, K35A, K40A and K42A also led to a significant decrease in activity. Interestingly, while in the double mutations K37A, V38A and K39A had large effects on strand exchange, whereas the single mutations did not decrease the activity significantly.

The N-terminal 44 amino acids of FinO are critical for SLII/SLIIc duplexing Next, we wanted to explore the relationship between the strand exchange and duplexing activities of FinO. Were the same residues that catalyzed strand exchange also important to intermolecular duplexing? We performed duplexing assays between SLII of FinP (SLIIX) and the SLIIc domain of the 5' UTR of traJ mRNA (SLIIcx), which are the primary high affinity binding targets of FinO (Figure 2-1) (15). In the assay, SLIIX is 5' end- 32 labeled with y P-ATP and is incubated at 37°C with an excess of its SLIIcx complement in the absence or presence of FinO or its mutant derivatives (Figure 2-4A). The reactions were stopped after one hour by adding SDS to denature FinO and the products were run on a 10% native gel (Figure 2-4B and Figure 4-4C). By themselves, the two complementary stem-loops did not form an intermolecular duplex over the time course, probably due to their extremely stable nature. However, upon the addition of FinO, SLIIX and SLIIcx form a duplex within 10 minutes (Figure 2-4B) with an apparent 5 1 1 second-order duplexing rate constant (k2) of 1.4 x 10 M" s" (Table 2-1). Therefore, FinO is able to duplex both FinP and traJ mRNA as well as minimal binding targets derived from the two RNAs. The SLIIx-SLIICx duplexing rate constant decreased nine-fold in the presence of FinO 26-186 relative to FinO (Figure 2-4B and Table 2-1) The result was similar to the mutant's decrease in the FmP-traJ mRNA duplexing

55 rate constant (~ nine-fold; Table 2-1) (7). This indicates that the first 25 amino acids are specific for some aspect of the duplexing reaction and FinO can anneal various complementary RNA targets. As with the strand exchange reactions, removal of the first 44 amino acids of FinO completely abolished duplexing activity for both the SLIIx-SLIIcx and F'mP-traJ mRNA reactions (Figure 2-4B and Table 2-1) indicating that similar N-terminal residues are important for both duplexing and strand exchange.

To test this hypothesis, we performed SLIIx-SLIIcx duplexing assays in the presence of the double alanine mutants used in the strand exchange assay and looked for decreases in the levels of intermolecular duplexing relative to FinO. Figures 2-4C and 2-4D show that all the double mutants spanning residues 32-42 had decreased duplexing activity to some degree, with the P34A/K35A, W36A, K37AA/38A, K39A/K40A, and Q41A/K42A mutants having the biggest effect. Therefore, the N-terminal region spanning residues 32-42 appear to be involved in both strand exchange and duplexing activities. However, the decrease in duplexing activity due to the alanine substitutions was not as severe as for the strand exchange assay. This indicates that there is an additional component of the N-terminus, mainly the first 25 amino acids, which is specific for the duplexing activity of FinO. Since the N-terminal FinO 1-61 fragment could perform strand exchange, we wondered if it could also duplex SLIIX and SLIIcx. As shown in Figure 2-4B and Table 2-1, the construct was able to form an intermolecular duplex, albeit at a much lower rate than FinO. This highlights the importance of the N-terminal helix for strand exchange and duplexing. However, the remainder of FinO is required for efficient activity.

Like strand exchange, the SLIIx-SLIIcx duplexing reaction is dependent on the concentration of FinO (Figure 2-4E). Maximal FinO duplexing activity occurs at final FinO concentrations of 0.5 to 2 \iM and duplexing decreases at concentrations greater than 4 |a.M. Reactions with the other truncation constructs showed a similar pattern although the level of duplexing was much less than FinO.

56 FinO may destabilize duplex structures using its free energy of binding Unlike RNA helicases which use ATP to destabilize base pairs (31), FinO performs its strand exchange and duplexing activities in an ATP- independent fashion. Since these processes would require an energy input by FinO to break base pairs in the stems of SLII and SLIIc, we asked whether the protein may use its free energy of binding to destabilize its RNA substrates. The binding affinities of several FinO mutants were compared using an EMSA (Figure 2-5). Briefly, SLII was 5' 32P end-labeled and incubated with increasing concentrations of FinO, FinO 1-186 W36A, FinO 26-186 or FinO 45-186. Binding reactions were incubated on ice for 30 minutes prior to loading onto an 8% native gel at 4°C. The results show an inverse relationship where the mutants deficient in strand exchange and duplexing bind tighter to SLII than wild-type FinO (four, five and 20-fold increase in affinity (Ka) for FinO 26-186, FinO 1-186 W36A, and FinO 45-186 respectively) (Table 2-1). This may be explained by FinO having two components to its free energy of binding. One part may be the favourable RNA contacts made by the protein and the other part may be derived from unfavourable base pair destabilization. If the catalytic area of FinO is removed, the unfavourable component of the binding process is removed thus leading to more favourable interactions and an increase in the overall binding affinity. However, it should be noted that the EMSA experiments and the strand exchange and duplexing assays were performed at different temperatures (4°C vs. 37°C). Control experiments show that FinO binding affinities are not affected by the presence of excess cold SII(A) RNA which is a component of the strand exchange assay (data not shown).

The catalytic portion of FinO is important for bacterial conjugation Knowing that FinO has strand exchange and duplexing activity in vitro, we wanted to test whether these processes were important for inhibition of bacterial conjugation in vivo. Mating assays were performed where we looked at the ability of the FinO mutants (supplied in trans as GST fusion proteins) to repress F-plasmid transfer from donor to recipient E. coli cells. Donor cells contained a pGEX-FinO plasmid and pOX38-Km which is derived from the F-plasmid and is FinO-deficient and kanamycin resistant. Recipient

57 cells, which are spectinomycin-resistant, were mixed with donor cells and transconjugant cells which were kanamycin and spectinomycin resistant were selected. The efficiency of mating was calculated as the ratio of tranconjugate to donor cells. Table 2-1 summarizes the data which shows that as expected, FinO- GST is able to repress conjugation significantly. Levels of conjugative transfer were only 5% of the no FinO assay. Both the N-terminal FinO-GST truncation mutants, however, severely disrupted inhibition of conjugation. Conjugative transfer in the presence of FinO 26-186-GST was 87% of the no FinO assay whereas the FinO 45-186-GST mutant was completely unable to repress conjugation. FinO W36A-GST also significantly disrupted mating inhibition as transfer in the presence of the mutant was 36% of the no FinO assay. The results indicate that the strand exchange process appears to play a role in conjugation in vivo as deletion of the N-terminal 32-42 catalytic region completely disrupts mating repression. Also, mutation of the key residue in strand exchange, tryptophan 36, resulted in a decrease of repression highlighting the importance of this amino acid in the process. Surprisingly, the presence of two double alanine mutants, FinO K37A/V38A- GST and K39A/K40A-GST which severely affected strand exchange, did not affect repression of conjugation. This seems to indicate that conjugation repression proceeds by a different mechanism than the simple in vitro strand exchange experiment, but W36 is still an important residue, whereas others such as K39 and K40 play a lesser role in vivo. FinO 26-186 mutant also severely affected repression of mating suggesting that the N-terminal 25 residues also play an important role in the process. These residues were very important for the in vitro intermolecular duplexing activity of FinO so it is possible they mediate annealing of the complementary strands of FinP and traJ mRNA in vivo. Previous studies has shown that FinO protects FinP from RNase E degradation, stabilizing RNA levels so it can interact with traJ mRNA (6, 16, 20). A control experiment was performed to make sure that the negative effect of the mutants was not due to a decreased protection of FinP RNA levels from RNase E. Plasmids expressing wild-type or mutant FinO-GST

58 were introduced into a FinP-expressing E. coli strain and the levels of FinP transcripts were measured by northern blot analysis at various time points after rifampicin blocking of transcription initiation (Figure 2-6). In the absence of FinO-GST, FinP is degraded (t1/2=50 min) whereas in its presence the half- life of the RNA is extended past 2 hours. Similarly, all of the mutants tested in the mating assay were able to stabilize FinP to the same level as wild-type FinO, indicating that the disruption of conjugative repression was not due to the inability of the mutant to protect FinP from cellular degradation.

Discussion

Here it is shown that FinO is able to perform strand exchange on a duplex RNA derived from the SLII domain of FinP antisense RNA in vitro. This strongly suggests that FinO is able to destabilize otherwise stable duplex regions of its targets FinP SLII and SLIIc of traJ mRNA. In addition, FinO is able to promote intermolecular duplexing of SLII and SLIIc in vitro. The complementary RNAs are not able to base pair in the absence of FinO due to their highly stable stem areas which pose a kinetic barrier to duplex formation. Thus, FinO acts as an RNA chaperone, remodeling its target RNAs through strand exchange and annealing events, thereby enhancing the rate of intermolecular duplexing between FinP and traJ mRNA. These actions would account for the 20-fold repression of bacterial conjugation by FinO in this study. The chaperone area was mapped to the N-terminal 44 amino acids of FinO. Deletion of this area to FinO 45-186 completely abolishes both strand exhange and intermolecular duplexing activities. The first 25 amino acids were specific for duplexing activity as removal of the residues led to a nine­ fold decrease in the duplexing rate for both FmP/traJ mRNA and SLII/SLIIc while the strand exchange rate only decreased 1.5-fold. The debilitating effect of the truncations extended to the in vivo mating assays as both FinO 26-186 and FinO 45-186 failed to repress bacterial conjugation. This shows that FinO-mediated strand exchange and annealing are important processes in vivo as they are needed for efficient FmP-traJ mRNA interactions which

59 would down-regulate expression of TraJ and subsequent genes needed for bacterial conjugation. It can be seen that the SI I strand exchange process is much slower than the SLII/SLIIc duplexing reaction. FinO-mediated strand exchange is completed by two hours whereas complete duplex formation occurs within 10 minutes. Whereas FinO makes high affinity interactions with the SI I duplex, EMSAs reveal that FinO does not form a stable complex with SII(A). Increasing the concentration of the SI I (A) strand moderately increases the strand exchange rate indicating that coordination of the single strand by FinO may be the rate limiting step in the process. For the duplexing reaction, SLII and SLIIc are both high affinity targets of FinO. Thus the recognition step in duplexing is likely enhanced by FinO-RNA interactions. The SLII domain contains a 5'-YUNR-3' (Y=pyrimidine, N=any base, R=purine) U-turn motif in its loop which probably aids in the recognition process. This motif has been shown to be important for making initial base pair interactions during RNA- RNA pairing in a number of RNA-only antisense systems such as CopA/CopT control of plasmid R1 replication (5). However, even though SLII and SLIIc could form loop-loop contacts, they cannot form a full duplex in the absence of FinO. The protein would be needed to destabilize portions of the stem regions of both RNAs which present a kinetic barrier to duplex formation. In addition, it is likely that FinO aids in the annealing process of SLII-SLIIc duplexing either by preventing the destabilized strands from "snapping back" or by increasing the local concentration of complementary strands, enhancing the intermolecular base pairing process. Alanine scanning mutagenesis showed that residues 32-42 of the N- terminal a-helix of FinO represent the catalytic region of the protein. This unique area consists mainly of lysine residues surrounding a solvent-exposed tryptophan. Mutation of P34, K35, W36, K40 and K42 to alanine had serious consequences for Sll strand exchange. We propose that the lysine residues in this region contact the phosphodiester backbone of the RNA allowing the tryptophan to intercalate the stem region thus disrupting base pairing in this region. A similar mechanism, whereby stacking interactions between aromatic amino acids and RNA help to unwind sections of double-stranded RNA has been shown for helicases (17, 22, 30). Mutation of residues in the

60 catalytic region of FinO was more deleterious for strand exchange than duplexing indicating that other areas of FinO are important for the duplexing reaction. Indeed, removal of the first 25 amino acids reduced the rate of both FinP/fraJ mRNA and SLII/SLIIc duplexing significantly. Interestingly, this region of the protein has been shown to be disordered in the presence of SLII (8) and removal of this area increases the binding affinity to SLII four-fold. It is possible that it is involved in transient RNA contacts which may aid in the annealing step of duplexing or it could be involved in protein-protein interactions. FinO catalyzes Sll strand exchange and SLII/SLIIc duplexing in the absence of ATP. To explain this interesting finding, we suggest that FinO uses a portion of its binding energy to destabilize its RNA targets. Indeed, we noticed an inverse relationship between FinO-SLII binding affinity and its catalytic activity. The 45-186 truncation mutation completely abolishes strand-exchange and duplexing in vitro and mating repression in vivo. However, this mutant bound to SLII 20-fold tighter than FinO. Likewise, the 26-186 and W36A mutations both disrupted catalytic activities while binding to SLII four and five-fold tighter than FinO. The catalytic area also resides in the most disordered region of FinO. The crystal structure of FinO 26-186 showed that the highest temperature (B) factors are located throughout the N-terminal a-helix (residues 26-61) (7). Interestingly, the FinO 1-61 construct demonstrated strand exchange and SLII/SLIIc duplexing activity (albeit at much lower rates than FinO) even though it was shown previously that this construct was primarily disordered by circular dichroism (8). This indicates that high affinity interactions are important for FinO recognition of its RNA substrates while low affinity interactions, in the N-terminal region, are important for the catalytic activities of FinO.

7 The rate of FinO-mediated FinP-fraJ mRNA duplexing (k2 = 2.5 x 10 M"1s"1) was much higher than the rate of FinO-mediated SLII-SLIIc duplexing 5 1 1 (k2 = 1.4 x 10 M" s" ). This may be due to the presence of the SLI and SLIc domains of FinP and traJ mRNA respectively. Compared to SLII and SLIIc, these stem-loops contain fewer base pairs and harbor base mismatches which make them less stable. As shown in the CopA/CopT system, the presence of bulges or mismatches in antisense and target mRNAs is

61 important for duplex progression into the stable inhibitory structure (19). Here, the antisense RNA, CopA, forms initial base pairs with its target RNA, CopT, through loop-loop interactions (18). Further intermolecular base pairing to the final four-helix structure is achieved through internal mismatches in both stem-loops which destabilize intramolecular base pairing. Likewise, for the FinOP system, removal of the A12/A27 mismatch in SLI and the U86/U101 mismatch in SLIc to Watson-Crick base pairing has shown to decrease pairing rates between the two stem-loops (10). Therefore, the presence of SLI and SLIc may explain the enhancement in the pairing rate of FinP and traJ mRNA. ATP-dependent DNA helicases often unwind large tracts of duplex DNA with significant energy input from ATP hydrolysis. In contrast, most RNA helicases only need to unwind short segments of duplex and therefore, the processivity afforded by continuous ATP hydrolysis may not be absolutely required for many RNA remodeling processes. This is the case for a class of proteins called RNA chaperones, which like FinO, have the ability to remodel RNA structure without energy input from ATP. One of the most critical functions inherent in these proteins is the ability to lower the energy barrier needed to resolve kinetically trapped, misfolded secondary or tertiary RNA structures in vivo (4, 11, 21, 34). Recently, a number of RNA chaperones have been found to have helix destabilizing activities and the ability to promote strand annealing. Examples include the N-terminal domain of the Hepatitis Delta Antigen (NdAg) (14) and the E. coli host factor I (Hfq) (23, 35). Perhaps the best characterized candidate for a functional homolog for FinO is the nucleocapsid protein (NCp7) of HIV-1. This nucleic acid chaperone mediates a conformational change in the viral RNA genome to the mature, more stable dimeric state. It also facilitates binding of the primer tRNA to its binding site to initiate reverse transcription, and is subsequently involved in two strand transfer reactions leading to the final DNA copy of the genome (25). Like FinO, NCp7 can bind kissing stem loop structures and facilitate their transition into an extended duplex, however, in contrast to FinO, it can stably bind single-stranded nucleic acids leading to annealing or strand exchange events (27, 28). Recently, it has been shown that NCp7 has helix destabilizing properties in vitro which are crucial for annealing of the minus-

62 strand DNA to the 3' terminus of the RNA genome, and preventing self- priming reactions (1, 13). These melting events are limited to a few base pairs, enough to initiate annealing to the complementary strand. It is probable that FinO might act in a similar fashion to NCp7, destabilizing a small portion of FinP and traJ mRNA allowing the two complementary stem loops to duplex. Like NCp7, the mechanism of FinO-mediated RNA-RNA interactions is not fully understood. Further structural studies using NMR and crystallographic strategies are needed to investigate possible helix destabilization activities and sense-antisense interactions.

Materials and Methods

Preparation of proteins FinO(45-186) was cloned into the pGEX-KG vector using the protocol described previously for cloning other FinO fragments, including FinO(26- 186) (7). FinO alanine substitutions were introduced by PCR overlapping amplification (12). The primers used for creating the double and single alanine point mutants are shown in Table 2-2 and 2-3 respectively. Expression and purification of FinO and all FinO derivatives used in this study also follow a protocol we described previously (8). To prevent protein degradation, purified proteins were divided into 50 pL aliquots at < 3 mg/mL, flash frozen in liquid nitrogen and stored at -70 °C until required.

Preparation of RNA substrates Various RNAs used in this study were prepared by T7 in vitro run-off transcription reactions using the appropriate DNA template and the complementary primer as described (8)!. The nucleotide sequence of SLII, FinP and the traJ RNA templates, as well as the primer are given in (8), while the DNA templates used for synthesis of SLIIX, SLIIcx, SII(A), SII(B), SII+4(A), and SII+4(B) are shown in Table 2-4. Sll sequences were designed to allow for maximal transcriptional yields of both strands while maintaining base-pair complementarity. SllA(A) and SIIA(B) RNA strands were purchased from Dharmacon Research.

63 The duplexes used in this study were prepared as follows: 5 nM of the 32P-labeled strands were combined with an excess of their complementary cold strand (50 nM) in 10 mM Tris-HCI (pH 8.1) and 100 mM NaCI and annealed by slow cooling to 23 °C from an initial temperature of either 85 °C for SII/SllA/SII+4, or 100 °C for SLIIx/SLIIcx.

Strand exchange assay Strand exchange assays were set-up on ice in 12 pL reaction volumes consisting of 6 pL 2X reaction buffer (50 mM Tris (pH 8.1), 100 |ig/mL BSA, 10% glycerol, 0.1 % B-mercaptoethanol and 20 mM NaCI), 1 uL of 20 units/uL RNAguard (Amersham Biosciences) (in 50 mM KCI, 20 mM HEPES-KOH (pH 7.6) and 5 mM DTT), 1.5 pl_ protein (in 50 mM MES (pH 6.5), 0.1% B-mercaptoethanol and 150 mM NaCI; 8 uM to give a final concentration of 1 uM, unless otherwise indicated), 1 pL of the labeled SII/SllA/SII+4 duplex (5 nM to give a final duplex concentration of -400 pM), 2 pL of the unlabeled SII(A)/SIIA(A)/SII+4(A) strand (1.25 pM to give a final concentration of ~200 nM) and 0.5 pL of distilled water. For Mg2+ titration studies, MgCI2 was added to the 2X reaction buffer at 0.02, 0.2, 2, 10, 20, 200 mM. Strand exchange assays were initiated by placing reaction tubes at 37 °C (or 4°C, 21°C, or 30°C for temperature variation experiments) and stopped after two hours by addition of an equal volume of stop solution (5% glycerol, 0.4% SDS and 20 mM EDTA). For time course experiments, aliquots were taken from a scaled up reaction mixture at various time points and were stopped. Samples were subjected to 15% non-denaturing PAGE (180 V) at room temperature to separate the free and duplexed 32P-labeled strands. For rate determination assays, samples were loaded onto a continuously running gel. To visualize bands corresponding to duplexed and single-stranded 32P-labeled RNA, a Molecular Dynamics storage phosphor screen was exposed to the gels for ~12 hours and scanned using a Storm 840 phosphorimager (Molecular Dynamics). Bands were quantified with ImageQuanNT software (Molecular Dynamics). See below for determination of apparent first order strand exchange rates.

64 Duplexing assays

Duplexing between SLIIX and SLIIcx was performed in 50 uL reaction mixtures containing 25 uL of 2x reaction buffer (50 mM Tris (pH 8.1), 100 ug/mL BSA, 10% glycerol, 0.1 % p-mercaptoethanol and 40 mM NaCI), 5 uL of protein at 10 uM, in 20 mM MES (pH 6.5), 0.1% p-mercaptoethanol, and 32 60 mM NaCI, 5 uL of SLIICx at 1 uM, 5 uL of P-labeled SLIIX at 50 nM and

10 uL of ddH20. To initiate duplexing, labeled SLIIx, preincubated at 37 °C,

was added to reaction mixtures, also preincubated at 37 °C and lacking SLIIX. 5 uL aliquots were taken at various time points and added to 5 uL of cold stop buffer (5% glycerol, 0.4% SDS and 20 mM EDTA). Samples were subjected to 10% non-denaturing PAGE (180 V) for 2 hours prior to visualization and

quantification of the labeled SLIIX using ImageQuantNT software, as described above. FinP-fraJ RNA duplexing assays with FinO(45-186) were performed as described previously (7). See below for determination of second order duplexing rates.

Derivation of rate constants Strand exchange rates were determined by fitting data to the equation,

kt [A*] = [A*B]0(1-e ) + c

where k is the apparent first order rate constant of strand exchange, [A*B]0 is the initial concentration of duplex RNA, and [A*] is the concentration of 32P-

labeled single stranded RNA relative to [A*B]0 at time t (calculated as the ratio of band intensities of A* divided by A*B + A* at each time, t, minus the ratio obtained at time zero).

The second order apparent rate constant (k2) for duplex formation was determined essentially as described (24). Briefly, the second order rate equation for the duplexing reaction is,

at at where [A*B], [A*], and [B] are the concentrations of duplex, 32P-labeled strand A and unlabeled strand B, respectively, at time t. When the unlabeled strand

is in large excess over the labeled strand ([B]0 » [A*]0), duplexing can be expressed as a pseudo-first order reaction,

65 and k! can be determined from a plot of ln[A*] vs t. k2 is then determined from

^ by the equation k2 = &, /[B]0.

Electrophoretic Mobility Shift Assays2 To determine the association constants of FinO and FinO-derived proteins for SLII or Sll duplex, EMSAs were performed with increasing concentrations of protein at 4°C as described previously (15), with the following modification. Binding reactions contained 5 |il_ of the reaction buffer used in the strand exchange assays (with an additional 10% glycerol and 1 mM EDTA), 4 ^L of protein in 50 mM MES (pH 6.5), 0.1% p-mercaptoethanol, 450 mM NaCI and 100 |ag/ml BSA, and 1 jaL of labeled RNA at a concentration of 500 pM.

Mating assays3 Mating assays were performed essentially as described (26) using Escherichia coli MC4100 cells bearing the F derivative plasmid pOX38-Km (3) and various pGEX-FinO plasmids. The presence of both plasmids was confirmed by agarose gel electrophoresis of plasmid DNA isolated from the E. coli strains, and GST-FinO protein expression levels were assayed by Western blot analysis using anti-GST antibodies (Sigma) and anti-FinO antiserum. All GST-FinO proteins used in these studies were expressed at similar levels, within approximately 20% of wild type. The ratio of transconjugants to donors was calculated, allowing mating efficiency to be compared with the control of conjugal transfer of pOX38-Km alone.

Northern Blot analysis to determine FinP half-life4 The half-life of FinP RNA isolated from the GST-FinO-expressing E. coli MC4100 strains used for the mating assays was assessed by Northern blot analysis as previously described (16, 26). Equivalent amounts (35 \xg) of

2 Some EMSAs were performed by Alexandru F. Ghetu 3 Experiments were performed by Michael J. Gubbins 4 Experiments were performed by Michael J. Gubbins

66 total RNA were loaded in each lane of the gel used for the Northern analysis. After measurement of FinP band intensities, the blots were stripped and re- probed for the control RNA (tRNAser) and the FinP intensities were normalized based on the amounts of tRNAser detected in same lanes.

67 A

u c C- - C 60 C A A-U A A 60 U C A 50G-C G-C SO U U-A G-C A-U C-G G-C SLI C_G A"U SLII E-DS-C A A C-G A A G-C A-U G-C U U A-U 10G-U y-A U C- G-C 30 A-U SLIM SLIIc G-C SLlc A-U G-C 70 U-A U-A C-G A-U 70 A-U 40 U*G 20 1C C-G G-C C-G 3° ' C-G GAUA GACA* GAUUUU 5'- G UUAAAAUUUGAAAUUGAAAAUCGC ACUGUC .-3' 1 -3' FinP RNA 5'UTRoffraJmRNA

B C-G C-G U-A • 1 C-G U-A C-G C-G G-C C-G A-U A-U A-U G-C G-C G-C G-C G-C G-C A-U G-C SII(A*>) c'-o. SII(B) SllA(A) £;£ SIIA(B) Sll +4(A) *:£ SII+4(B G-C U-A G-C U-A A-U U-A A-U G-C A-U G-C C-G G-C C-G U*G C-G U»G G-C U»G G-C ACA GAUUUU-3 5'' '3' 5'- GGACA GAUUUU Sll duplex SUA duplex SII+4 duplex

D r T P P K W K V K K Q K L A E

C-G G-C G-C A-u U-A C-G A-U G-C G-C U-A C-G A-U U«G G-C GGACA GAGUCC-3 5'-•GAAAAUCGC ACUGUC-3' SLII,, SLIICv

Figure 2-1: RNA and protein constructs used in this study

A) Sequence and secondary structures of FinP and the 5' UTR of traJ mRNA. The AUG start codon and ribosomal binding site (RBS) of traJ mRNA are boxed. B) Sequences of the RNA duplexes used in the strand exchange experiments. The A and B strands are shown base paired at their complementary regions. The extra 4 base pairs of the SII+4 duplex are highlighted in bold. C) Sequences of the RNA constructs used in intermolecular duplexing assays. D) Summary of FinO constructs used in study. A ribbons representation of the FinO 26-186 crystal structure (7) is shown. The amino acid sequence of residues 26-45 is shown above the structure. FinO residues 32-45, which are critical for strand exchange activity, are shown in black in the sequence and in the structure. The catalytic Trp 36 is annotated on the structure. The N-terminal 25 amino acids, not shown in the structure but important to the duplexing activity of FinO, is shown by a dashed line. Finally, position 61 is shown on the structure in reference to the FinO 1-61 C-terminal truncation construct.

68 A FinO + J MB A >. ABB + A 37 °C X.J B No Protein FinO FinO (26-186) FinO(45-186) FinO(1-61) FinO/SHA • • • • m m • • * mjm " Duplex • •

SII(A) •

0 120 0 120 0 120 0 120 0 120 0 120

g D 4 RT 30 37 ( C) Duplex • *Mft"

SII(A) • . ""«i#

0.44 +/- 0.2 0.04 +/- o.oi Duplex •

SII(A) •

^ FinO FinO (26-186) FinO(45-186) FinO(1-61) FinO/SllA 0 .1 .25 .5 1 2 4 8 .1 .25.5 1 2 4 8 .1 .25 .5 1 2 4 8 .1 .25 .5 1 2 4 8 .1 .25 .5 1 2 4 8 Duplex • • . . SII(A) • "ZLmm ••riiiiiiiiniiiliif'

(Figure 2-2 legend on p. 70)

69 Figure 2-2: FinO can perform strand exchange on SLII-derived duplex RNAs (p. 69)

A) Schematic of the strand exchange experiment. The Sll duplex is radiolabeled with 32P (star) at the 5' end of strand A and incubated with FinO and a molar excess of unlabeled (cold) A strand at 37°C. Over the two hour time course FinO exchanges the radiolabeled (hot) A strand for the cold A strand. The released hot A strand is then resolved from the non- exchanged hot duplex by gel electrophoresis. B) Comparison of strand exchange over time for various FinO truncation mutants. Radiolabeled Sll was incubated with FinO, FinO 26-186, FinO 45-186, FinO 1-61 (each at 1 uM final concentration) or No protein. Radiolabeled SUA duplex was incubated with FinO at a final concentration of 1 jxM. For each construct, an aliquot from the strand exchange reaction was taken and stopped at 0, 1, 5, 15, 30, 60, 90, and 120 minutes and loaded on a continuously running 15% native gel. Duplex and released SII(A) strand are marked by an arrowhead beside the gels. C) Quantification of strand exchange, shown as "Percentage SII(A) released", as a function of time in minutes for the strand exchange reaction shown in B). The apparent first order rate constants (ki; see Materials and Methods) for each construct as a fraction of FinO (ki= 1.1 x 10~2 s"1) and the standard deviations (derived from at least three independent rate determinations) are shown next to each plot. No Protein (•), FinO 1-61 (•), FinO 26-186 (A), FinO/SII (•), FinO/SMA(o). D) The dependence of temperature and MgCb on FinO-mediated Sll strand exchange. Strand exchange reactions were stopped after two hours and run on a 15% native gel. The reaction temperatures were 4°C, room temperature (21 °C), 30°C, and 37°C. For the Mg2+ titration, the final concentration of MgCb added in the reaction buffer is noted in each lane (0, 0.01, 0.1, 1, 5, 10, 100 mM). Duplex and released SII(A) strand bands are noted by arrowheads in both gels. E) Concentration dependence of FinO truncation constructs on strand exchange. Reactions with radiolabeled Sll (No Protein, FinO, FinO 26-186, FinO 45- 186, and 1-61) or radiolabeled SUA (FinO) were run for two hours before being stopped and loaded on a 15% native gel. The final concentration of each of the FinO constructs (in uM) is noted above the gel. Duplex and released SII(A) strand bands are noted by arrowheads beside the gels.

70 A B ^

Sll duplex* j

811(A)*

*.&&&

FinO Mutants

y />^W>W#>VW<^ 100 *• -^ Sit duplex • IS 80 TO ^~ to ,1

SII(A)* « w 40

<6,S 20 1H11

FinO Mutants

Figure 2-3: Residues in the N-terminal a-helix of FinO are critical for strand exchange

Strand exchange reactions containing radiolabeled Sll duplex and FinO double (A,B) or single (C,D) alanine point mutants (noted) were incubated for two hours before being stopped and loaded on a 15% native gel (shown in A and C). Sll duplex and the released SII(A) bands are noted next to the gels. The quantified levels of strand exchange for each mutant, expressed as "Percent SII(A) released relative to FinO" is shown in B and D along with standard deviations (results from at least three independent experiments). The FinO and FinO 45-186 constructs as well as No Protein were used as controls.

71 A O SLII. Duplex 37 °C SLilc, B SLII,

No Protein FinO FinO (26-186) FinO(4S-186) Fin0(1-61)

lz P-SLII,/SLIIcx Duplex •• «*«NPNMi (. » S^--**!-

... "P-SLIIX • fMgpMNP •#-*-

Time (min)

1 1 k2(M' s" ) 1 (tO.4) 0.11(t0.07) 0.01 (t 0.0008)

D

n 60 ]

3zP-SLiySLIIc, 55 - Duplex ^

HP-SUI,><|| , H:

FinO Mutants

.*P *° FinO FinO (26-186) FinO(4S-186) FinO(1-61)

0 o- »• ©• 1 > 4 8 « «'«• 1 2 4 8 O- «.' o- 12 4 8 «• O- O- 1 2 4 8 "P-BLiysuic,, Duplex '

^ HfeUL Ml

ure 2-4 legend on p. 73)

72 Figure 2-4: The N-terminus of FinO is important for facilitating sense-antisense interactions (p. 72)

A) Schematic of the SLII/SLIIc duplexing reaction. SLII is radiolabeled with 32P (star) and incubated with FinO and a molar excess of SLIIc for one hour at 37°C. Formation of a radiolabled intermolecular SLII-SLIIc duplex is monitored by native gel electrophoresis. B) Comparison of SLIIx-SLIIcx duplex over time for various FinO truncation mutants. Radiolabeled SLIIX was incubated with FinO, FinO 26-186, FinO 45-186, FinO 1-61 (each at 1 uM final concentration) or No protein. An aliquot from the duplexing reaction was taken out and stopped at 0, 10, 20, 30, and 60 minutes for No Protein and FinO 45-186; 0, 0.25, 0.5, 0.75, 1, 1.5, 2, 3, 4, and 10 for FinO; and 0, 1, 2, 4, 8, 15, 30, and 60 min for FinO 26-186 and 32 32 FinO 1-61. Samples were loaded onto a 10% native gel. P-SLIIx/SLIIcx duplex and P-SLIIX bands are noted next to the gels. The second order duplexing rate constants (k2) and standard deviations (results from at least three independent experiments) of each construct 5 1 1 relative to FinO (k2= 1.4 x 10 M" s~ ) are noted below each gel. C) SLIIx/SLIIcx duplexing experiments with the FinO N-terminal double alanine mutants. Reactions were incubated for 60 minutes after which they were stopped and loaded onto a 10% native gel. No Protein, 32 32 FinO, FinO 26-186 and FinO 45-186 were used as controls. P-SLIIx/SLIIcx duplex and P- SLIIx bands are noted next to the gels. D) Quantification of the levels of duplexing for each mutant, expressed as a percentage of the amount of intermolecular duplex formed relative to FinO, from the gel in C). Error bars represent standard deviations calculated from at least 3 different independent experiments. E) Concentration dependence of FinO truncation constructs on intermolecular duplexing of SLIIx/SLIIcx. Reactions were run for one hour before being stopped and loaded on a 10% native gel. The final concentration of each of the FinO 32 32 constructs (in uM) is noted above the gel. P-SLIIx/SLIIcx duplex and P-SLIIX bands are noted next to the gels.

73 FinO FinO{26-186) FinO(45-186) Bound SUl • ; Free SLII •| WS«<,

>

Figure 2-5: RNA strand exchange deficient FinO mutants bind RNA tighter than wild type FinO5

Representative gel electrophoretic mobility shift assays for FinO, FinO 26-186 or FinO 45-186 binding to SLII RNA. Samples containing 50 pM SLII were incubated with protein at the concentrations indicated. These and similar experiments were used to determine the relative FinO-RNA association constants shown in Table 2-1.

This work was completed by Alexandru F. Ghetu

74 ^ 4>' r*T / Protein «# ^v ^ ^4?- JS$? $? F-plasm id + + + + 4" + + Time {mirt} 30 60120 0 120 0 120 0 120 0 120 0 120 0 120 post-rifampicin **

Figure 2-6. In vivo stabilization of FinP by FinO and FinO derivatives

Stabilization of FinP in cells expressing FinP and the indicated FinO proteins was examined at the given times after the addition of rifampicin by Northern blot analysis. As controls, FinP stability was examined in the absence of FinO, and hybridization was performed on RNA extracted from cells not harboring the F-plasmid.

This work was completed by Michael J. Gubbins

75 Table 2-1. Effect of FinO mutations on rates of strand exchange, duplexing, and conjugative inhibition

Relative Relative rate of Relative rate Relative mating 1 3 1 3 2 4 affinities (Ka) strand exchange (k-,) of duplexing ' (k2) efficiency

Protein RNA substrate

SLII Sll SUA SII+4 SLII/SLIIc FinP/fraJ mRNA

None 0.04(10.01) — — <0.01 0.02 (±0.01) 1

FinO 1.0 (±0.1) 1.0(10.3) 1.3 (±0.3) 0.4 (±0.2) 1.0 (±0.4) 1.0 (±0.4) 0.05 (±0.03)

26-186 4.0( + 0.3) 0.7 (±0.2) — — 0.11 (±0.07) 0.12 (±0.08) 0.87 (±0.20)

45-186 20 (±2) < 0.04 — — < 0.01 0.02 1.0 (±0.2)

1-61 0.004 (±0.001) 0.4 (±0.2) — _ 0.01 —

W36A 5.0(10.9) — 0.5 (±0.2) 0.4 (±0.1)

K37A/V38A — 0.6 (±0.3) 0.06 (±0.01)

K39A/K40A 1.4(10,2) — 0.4 (±0.2) 0.02

All rates are as a percentage of FinO 2 Efficiencies are as a percentage of mating in the absence of protein

3 7 1 5 ForFinO;Ka = 5(±1) x 10 M" ; k,= 1.1 (±0.3) x 10'V ; k2 (SLII and SLIIc ) = 1.4 (±0.2) x 10 M"V

7 1 1 k2(FinP-fraJmRNA) =2.5 (±1.0) x 10 M" s" This work was completed by Michael J. Gubbins Table 2-2: FinO double alanine mutant primer oligonucleotides*

Forward Primers:

T26A/I27A: 5'-CGG AGC CGG AAA GCC GCC ATC AAT GTC ACC-3'

I28A/N29A: 5'-CGG AAA ACC ATC GCC GCT GTC ACC ACG CCA-3'

V30A/T31 A: 5'-ACC ATC ATC AAT GCC GCC ACG CCA CCA AAA-3'

T32A/P33A: 5'-ATC AAT GTC ACC GCG GCA CCA AAA TGG AAG-3'

P34A/K35A: 5'-GTC ACC ACG CCA GCA GCA TGG AAG GTG AAA-3'

K37A/V38A: 5'-CCA CCA AAA TGG GCG GCG AAA AAG CAG AAA-3'

K39A/K40A: 5'-AAA TGG AAG GTG GCA GCG CAG AAA CTG GCG-3'

Q41A/K42A: 5'-AAG GTG AAA AAG GCG GCA CTG GCG GAG AAG-3'

E45A/K46A: 5'-CAG AAA CTG GCG GCG GCG GCT GCC CGG GAA-3'

Reverse Primers:

T26A/I27A: 5'-GGT GAC ATT GAT GGC GGC TTT CCG GCT CCG-3'

I28A/N29A: 5'-TGG CGT GGT GAC AGC GGC GAT GGT TTT CCG-3'

V30A/T31 A: 5'-TTT TGG TGG CGT GGC GGC ATT GAT GAT GGT-3'

T32A/P33A: 5'-CTT CCA TTT TGG JGC CGC GGT GAC ATT GAT-3'

P34A/K35A: 5'-TTT CAC CTT CCA JGC JGC TGG CGT GGT GAC-3'

K37A/V38A: 5'-TTT CTG CTT TTT CGC CGC CCA TTT TGG TGG-3'

K39A/K40A: 5'-CGC CAG TTT CTG CGC JGC CAC CTT CCA TTT-3'

Q41A/K42A: 5'-CTT CTC CGC CAG J_GC CGC CTT TTT CAC CTT-3'

E45A/K46A: 5'-TTC CCG CGG AGC CGC CGC CGC CAG TTT CTG-3'

* The alanine codon sequences (forward) and complementary sequences (reverse) are underlined

77 Table 2-3: FinO single alanine mutant primer oligonucleotides*

Forward Primers:

T32A: 5'-ATC AAT GTC ACC GCG CCA CCA AAA TGG-3'

P33A: 5'-AAT GTC ACC ACG GCA CCA AAA TGG AAG-3'

P34A: 5'-GTC ACC ACG CCA GCA AAA TGG AAG GTG-3'

K35A: 5'-ACC ACG CCA CCA GCA TGG AAG GTG AAA-3

W36A: 5'-ACG CCA CCA AAA GCG AAG GTG AAA AAG-3'

K37A: 5'-CCA CCA AAA TGG GCG GTC AAA AAG CAG-3'

V38A: 5'-CCA AAA TGG AAG GCG AAA AAG CAG AAA-3'

K39A: 5'-AAA TGG AAG GTG GCA AAG CAG AAA CTG-3'

K40A: 5'-TGG AAG GTG AAA GCG CAG AAA CTG GCG-3'

Q41 A: 5'-AAG GTG AAA AAG GCG AAA CTG GCG GAG-3'

K42A: 5'-GTG AAA AAG CAG GCA CTG GCG GAG AAG-3'

L43A: 5'-AAA AAG CAG AAA GCG GCG GAG AAG GCT-3'

Reverse Primers:

T32A: 5'-CCA TTT TGG TGG CGC.GGT GAC ATT GAT-3'

P33A: 5'-CTT CCA TTT TGG TGC CGT GGT GAC ATT-3'

P34A: 5'-CAC CTT CCA TTT TGC TGG CGT GGT GAC-3'

K35A: 5'-TTT CAC CTT CCA TGC TGG TGG CGT GGT-3'

W36A: 5'-CTT TTT CAC CTT CGC TTT TGG TGG CGT-3'

K37A: 5'-CTG CTT TTT GAC CGC CCA TTT TGG TGG-3'

V38A: 5'-TTT CTG CTT TTT CGC CTT CCA TTT TGG-3'

K39A: 5'-CAG TTT CTG CTT TGC CAC CTT CCA TTT-3'

K40A: 5'-CGC CAG TTT CTG CGC TTT CAC CTT CCA-3'

Q41 A: 5'-CTC CGC CAG TTT CGC CTT TTT CAC CTT-3'

K42A: 5'-CTT CTC CGC CAG JGC CTG CTT TTT CAC-3'

L43A: 5'-AGC CTT CTC CGC CGC TTT CTG CTT TTT-3'

* The alanine codon sequences (forward) and complementary sequences (reverse) are underlined

78 Table 2-4: DNA templates for in vitro transcription of RNAs for Chapter 2*

SLIIX: 5'-GGA CTC GCC GAT GCA GGG AGA CGT GAA CTC CCT GCA TCG ACT GTC CTA TAG TGA GTC GTA TTA-3'

SLIIcx: 5'-GGA CAG TCG ATG CAG GGA GTT CAC GTC TCC CTG CAT CGG CGA GTC CTA TAG TGA GTC GTA TTA-3' SII(A): 5'-GGT CCT GCA TCG ACT GTC CTA TAG TGA GTC GTA TTA-3' SII(B): 5'-AAA ATC GCC GAT GCA GGA CCT ATA GTG AGT CGT ATT A-3' SII+4(A): 5'-GGA ACT CCC TGC ATC GAC TGT CCT ATA GTG AGT CGT ATT A-3' SII+4(B): 5'-AAA ATC GCC GAT GCA GGG AGT TCC TAT AGT GAG TCG TAT TA-3'

* The T7 RNAP primer binding site is underlined

79 References

1. Bernacchi, S., Stoylov, S., Piemont, E., Ficheux, D., Roques, B.P., Darlix, J.L., and Mely, Y. (2002) HIV-1 nucleocapsid protein activates transient melting of least stable parts of the secondary structure of TAR and its complementary sequence. J Mol Biol, 317: 385-399.

2. Brunei, C, Marquet, R., Romby, P., and Ehresmann, C. (2002) RNA loop-loop interactions as dynamic functional motifs. Biochimie, 84: 925-944.

3. Chandler, M. and Galas, D.J. (1983) ISf-mediated tandem duplication of plasmid pBR322. Dependence on recA and on DNA polymerase I. J Mol Biol, 165: 183-190.

4. Cristofari, G. and Darlix, J.L. (2002) The ubiquitous nature of RNA chaperone proteins. Prog Nucleic Acid Res Mol Biol, 72: 223-268.

5. Franch, T., Petersen, M., Wagner, E.G., Jacobsen, J.P., and Gerdes, K. (1999) Antisense RNA regulation in prokaryotes: rapid RNA/RNA interaction facilitated by a general U-turn loop structure. J Mol Biol, 294: 1115-1125.

6. Frost, L, Lee, S., Yanchar, N., and Paranchych, W. (1989) finP and fisO mutations in FinP anti-sense RNA suggest a model for FinOP action in the repression of bacterial conjugation by the Flac plasmid JCFL0. Mol Gen Genet, 218: 152-160.

7. Ghetu, A.F., Gubbins, M.J., Frost, L.S., and Glover, J.N. (2000) Crystal structure of the bacterial conjugation repressor FinO. Nat Struct Biol, 7. 565-569.

8. Ghetu, A.F., Gubbins, M.J., Oikawa, K., Kay, CM., Frost, L.S., and Glover, J.N. (1999) The FinO repressor of bacterial conjugation contains two RNA binding regions. Biochemistry, 38: 14036-14044.

9. Gruber, A.R., Lorenz, R., Bernhart, S.H., Neubock, R., and Hofacker, I.L. (2008) The Vienna RNA Websuite. Nucleic Acids Res, 36: W70- 74.

10. Gubbins, M.J., Arthur, D.C., Ghetu, A.F., Glover, J.N., and Frost, L.S. (2003) Characterizing the structural features of RNA/RNA interactions of the F-plasmid FinOP fertility inhibition system. J Biol Chem, 278: 27663-27671.

11. Herschlag, D. (1995) RNA chaperones and the RNA folding problem. J Biol Chem, 270: 20871-20874.

80 12. Ho, S.N., Hunt, H.D., Horton, R.M., Pullen, J.K., and Pease, L.R. (1989) Site-directed mutagenesis by overlap extension using the polymerase chain reaction. Gene, 77: 51-59.

13. Hong, M.K., Harbron, E.J., O'Connor, D.B., Guo, J., Barbara, P.F., Levin, J.G., and Musier-Forsyth, K. (2003) Nucleic acid conformational changes essential for HIV-1 nucleocapsid protein- mediated inhibition of self-priming in minus-strand transfer. J Mol Biol, 325: 1-10.

14. Huang, Z.S., Su, W.H., Wang, J.L., and Wu, H.N. (2003) Selective strand annealing and selective strand exchange promoted by the N- terminal domain of hepatitis delta antigen. J Biol Chem, 278: 5685- 5693.

15. Jerome, L.J. and Frost, L.S. (1999) In vitro analysis of the interaction between the FinO protein and FinP antisense RNA of F-like conjugative plasmids. J Biol Chem, 274: 10356-10362.

16. Jerome, L.J., van Biesen, T., and Frost, L.S. (1999) Degradation of FinP antisense RNA from F-like plasmids: the RNA-binding protein, FinO, protects FinP from ribonuclease E. J Mol Biol, 285: 1457-1473.

17. Kim, J.L., Morgenstern, K.A., Griffith, J.P., Dwyer, M.D., Thomson, J.A., Murcko, MA, Lin, C, and Caron, P.R. (1998) Hepatitis C virus NS3 RNA helicase domain with a bound oligonucleotide: the crystal structure provides insights into the mode of unwinding. Structure, 6: 89-100.

18. Kolb, F.A., Engdahl, H.M., Slagter-Jager, J.G., Ehresmann, B., Ehresmann, C, Westhof, E., Wagner, E.G., and Romby, P. (2000) Progression of a loop-loop complex to a four-way junction is crucial for the activity of a regulatory antisense RNA. EMBO J, 19: 5905-5915.

19. Kolb, F.A., Westhof, E., Ehresmann, C, Ehresmann, B., Wagner, E.G., and Romby, P. (2001) Bulged residues promote the progression of a loop-loop interaction to a stable and inhibitory antisense-target RNA complex. Nucleic Acids Res, 29: 3145-3153.

20. Lee, S.H., Frost, L.S., and Paranchych, W. (1992) FinOP repression of the F plasmid involves extension of the half-life of FinP antisense RNA by FinO. Mol Gen Genet, 235: 131-139.

21. Lorsch, J.R. (2002) RNA chaperones exist and DEAD box proteins get a life. Cell, 109: 797-800.

22. Marians, K.J. (2000) Crawling and wiggling on DNA: structural insights to the mechanism of DNA unwinding by helicases. Structure, 8: R227-235.

81 23. Moller, T., Franch, T., Hojrup, P., Keene, D.R., Bachinger, HP., Brennan, R.G., and Valentin-Hansen, P. (2002) Hfq: a bacterial Sm- like protein that mediates RNA-RNA interaction. Mol Cell, 9: 23-30.

24. Persson, C, Wagner, E.G., and Nordstrom, K. (1988) Control of replication of plasmid R1: kinetics of in vitro interaction between the antisense RNA, CopA, and its target, CopT. EMBO J, 7: 3279-3288.

25. Rein, A., Henderson, L.E., and Levin, J.G. (1998) Nucleic-acid- chaperone activity of retroviral nucleocapsid proteins: significance for viral replication. Trends Biochem Sci, 23: 297-301.

26. Sandercock, J.R. and Frost, L.S. (1998) Analysis of the major domains of the F fertility inhibition protein, FinO. Mol Gen Genet, 259: 622-629.

27. Takahashi, K., Baba, S., Koyanagi, Y., Yamamoto, N., Takaku, H., and Kawai, G. (2001) Two basic regions of NCp7 are sufficient for conformational conversion of HIV-1 dimerization initiation site from kissing-loop dimer to extended-duplex dimer. J Biol Chem, 276: 31274-31278.

28. Urbaneja, M.A., Wu, M., Casas-Finet, J.R., and Karpel, R.L. (2002) HIV-1 nucleocapsid protein as a nucleic acid chaperone: spectroscopic study of its helix-destabilizing properties, structural binding specificity, and annealing activity. J Mol Biol, 318: 749-764.

29. van Biesen, T. and Frost, L.S. (1994) The FinO protein of IncF plasmids binds FinP antisense RNA and its target, traJ mRNA, and promotes duplex formation. Mol Microbiol, 14: 427-436.

30. Velankar, S.S., Soultanas, P., Dillingham, M.S., Subramanya, H.S., and Wigley, D.B. (1999) Crystal structures of complexes of PcrA DNA helicase with a DNA substrate indicate an inchworm mechanism. Cell, 97: 75-84.

31. von Hippel, P.H. and Delagoutte, E. (2001) A general model for nucleic acid helicases and their "coupling" within macromolecular machines. Cell, 104: 177-190.

32. Wagner, J.D., Jankowsky, E., Company, M., Pyle, A.M., and Abelson, J.N. (1998) The DEAH-box protein PRP22 is an ATPase that mediates ATP-dependent mRNA release from the spliceosome and unwinds RNA duplexes. EMBO J, 17: 2926-2937.

33. Wang, Y., Wagner, J.D., and Guthrie, C. (1998) The DEAH-box splicing factor Prp16 unwinds RNA duplexes in vitro. Curr Biol, 8: 441-451.

34. Weeks, K.M. (1997) Protein-facilitated RNA folding. Curr Opin Struct Biol, 7. 336-342.

82 35. Zhang, A., Wassarman, K.M., Ortega, J., Steven, A.C., and Storz, G. (2002) The Sm-like Hfq protein increases OxyS RNA interaction with target mRNAs. Mol Cell, 9: 11-22.

83 84 Chapter 3

Strategies for crystallization of the FinO-SLII RNA complex1

Overview The FinO-FinP interaction is critical to the repression of bacterial conjugation. Through RNA binding, FinO protects FinP from RNase E degradation and enhances its interaction with traJ mRNA. This leads to intermolecular RNA duplexing which prevents translation of the TraJ transcriptional activator and downstream conjugation genes. Here we attempt to crystallize FinO in complex with the SLII domain of FinP to obtain high resolution information on the interaction. SLII derivatives which vary in length and loop sequence are tested with FinO N-terminal truncation mutants to search for optimal crystallization complexes. The effect of polyamines, divalent cations, and cryo-additives are explored. Sll duplexes with various lengths and either blunt or overhanging ends were tested with the FinO constructs to increase RNA crystal contacts. Also, an SLII derivative was synthesized to allow binding of both U1A and FinO in order to increase protein-protein and protein-RNA crystal contacts.

Introduction X-ray crystallography is an important tool for determining the three-dimensional structure of macromolecules. Since the first crystal structure of myoglobin in the late 1950s (20) there have been more than 44000 structures solved according to recent statistics from the RCSB Protein Data Bank (PDB) (3). The overwhelming majority of structures solved comprise proteins (93.3%), followed by nucleic acid-protein complexes (4.3%), and DNA (1.6%) (Figure 3-1). RNA and RNA-protein complexes represent only 0.7% and 1.2% of the total crystal structures on the PDB respectively (Figure 3-1). Even though the number of structures solved from these categories is growing, crystallization of RNA and RNA- protein complexes remains challenging. The following are a few reasons

1 Part of this work was performed by Ross A. Edwards 85 why high quality crystals are hard to achieve for RNA and RNA-protein samples and some methods which can be employed to help solve the problems.

The chemical and structural nature of RNA Unlike proteins which have a large diversity of chemical groups on their surface, the surface of a folded RNA is dominated by the negatively charged phosphate backbone. This limits the number and precision of crystal contacts which form between neighboring molecules leading to poorly ordered crystals which fail to diffract X-rays to a suitable resolution needed for structure solution (19). Most buffers used for RNA crystallization require the presence of a combination of different cations to circumvent this problem. Monovalent cations, such as potassium, have been shown to mediate crystal contacts between RNAs (5). Divalent cations, such as Mg2+ are often required for RNA folding and activity in additon to aiding in the formation of crystal contacts (14). Polyamines, such a spermine and spermidine are often used to improve the quality of crystals (14) In addition, weak tertiary interactions between subdomains of large RNAs and RNA misfolding into alternative structures can lead to poorly ordered crystals due to conformational heterogeneity (19). As mentioned above, monovalent cations and Mg2+are often required to mediate folding into the correct structure and RNA annealing protocols are essential to ensure that the RNA has folded into its functional conformation (14).

Varying the RNA sequence One of the most critical considerations for RNA crystallization is the design of the sequence (11, 14, 19). Sequence changes in loop regions, the length of a helix, or the addition of stable 5'-GNRA-3' (N=any base, R=purine) tetraloops to duplex regions can have profound effects on the ability of the RNA to crystallize (33). Crystal contacts can be introduced by engineering crystallization modules into the RNA. For example, crystals of the group II intron domains 5 and 6 and the hepatitis delta virus were improved by introducing a 5'-GAAA-3' tetraloop in one

86 domain of the RNA and a tetraloop receptor in the minor groove of another (10). The new intermolecular interactions provided crystal contacts between neighboring molecules. Stacking interactions and intermolecular base pairing between ends of neighbouring RNAs can be introduced by engineering blunt or overhanging ends to an RNA helix, thus facilitating the ordering of molecules to improve crystallization (17, 35). An increase in crystal contacts was also be achieved by modifying RNA sequences which reside on the surface to bind the RNA binding domain of the U1A spliceosomal protein (8). The high affinity U1A binding sequence is a 10 nucleotide loop with two CG closing base pairs (27) and binding of the protein promotes protein-protein crystal contacts which significantly improve RNA crystallization (8). U1A can also be labeled with selenomethionine thereby providing a route to structure solution using the multiwavelength anomalous dispersion (MAD) technique (9).

The purity of the RNA sample Another source of heterogeneity which can affect crystal growth is the purity of the RNA sample. There are two methods to produce RNAs: chemical synthesis and in vitro transcription from a linearized plasmid or chemically synthesized DNA template (14). Chemically synthesized RNA is usually advantageous over in vitro transcription because milligram quantities of the RNA can be produced with little heterogeneity (31). However, the method is limited to the production of relatively short RNAs (< 80 nucleotides) as the frequency of abortive RNAs (failed coupling reactions) increases with increasing RNA length (14). In vitro transcription from the strong T7 promoter can produce much larger RNAs, however the technique suffers from heterogeneity at the 3' ends leading to a mixed sample with varying RNA lengths (23). The extra N+1-3 nucleotides are incorporated randomly by the RNA polymerase and can be very hard to resolve from the N-length RNA using standard RNA purification protocols. This can be averted by incorporated self-cleaving ribozymes at the 5' and 3' end which cut the desired RNA to the proper length (7).

87 The integrity of the RNA-protein complex sample The considerations for RNA crystallization, highlighted above, need to be extended to the crystallization of RNA-protein complexes. Once again the RNA construct is arguably the most important component of the interaction as it will be the hardest to crystallize. Using RNA truncations at duplex and single stranded areas which are not involved in the interaction may improve crystallization of the complex (11). Biochemical footprinting assays should be used in advance to determine which areas of the RNA are not involved in protein binding (6). Likewise, removing flexible areas of the protein which are not involved in the interaction may enhance crystals as well. This can be assayed by limited proteolysis in the presence of RNA followed by mass spectrometry analysis to determine the identity of the truncated protein fragment (13). The interaction between the RNA and protein should be stable as low affinity interactions may produce a heterogeneous sample of unbound components and complex which may hinder crystallization. Binding experiments should be performed at crystallization concentrations and complexes can be analyzed using light scattering techniques (static or dynamic) and gel-filtration chromatography to ensure proper formation and to assay for aggregation (24, 29). Often, the crystallization temperature is important as the complex may be more stable at lower temperatures (11). Depending on the purity of the protein solution, RNase inhibitors may have to be added to control RNA degradation. Assaying the long term stability of the complex solution will ensure that the RNA-protein complex is stable for the duration of the crystallization experiment and that aggregation and RNA degradation is not affecting crystal growth.

Goals ofFinO-SLII complex crystallization The crystal structure of FinO 26-186, solved by Ghetu et al. (12), was an important step towards determination of the mechanism by which FinO facilitates inhibition of bacterial conjugation. Naturally, the next step in the process is to attempt to determine the crystal structure of FinO in complex with one of its RNA targets. SLII was chosen as the substrate

88 RNA because it is the FinP high affinity binding site of FinO and would likely be simpler to crystallize than the entire FinP RNA. The structure would show the molecular details of the FinO-RNA interaction including: the role of the 5' and 3' single stranded tails of SLII which are required for FinO binding, the orientation of FinO on SLII which could give mechanistic clues into how FinO performs its intermolecular RNA duplexing activity, and the role of the N-terminal a-helix of FinO in RNA binding. In this chapter, various attempts are made at crystallizing the FinO-SLII complex. In spite of extensive optimization techniques, the best crystals diffracted to only 4.5 A, well short of a resolution needed to solve the structure. Nevertheless, the various optimization techniques and results are discussed in hopes that future researchers who are struggling to crystallize an RNA-protein complex may use the information to help grow well-diffracting crystals.

Results

Determining the optimal FinO and SLII hairpin constructs The FinO 26-186 construct used to solve the FinO crystal structure (12) was trimmed of the N-terminal 25 amino acids because they were found to be accessible to trypsin in the presence of SLII and thereby flexible in nature (13). In our first attempts at crystallizing the FinO-SLII complex, we used four FinO constructs: FinO 1-186, 1-186 W36A, 26-186, and 26-186 W36A. The W36A point mutants exhibit much less RNA duplexing and strand-exchange activity than wild-type FinO allowing them to bind tighter to SLII (See Chapter 2). In addition to wild-type SLII, various mutant SLII constructs were designed based on the length of their stem region. The SLII derivatives also contained 5'-GNRA-3' tetraloops for added stem-loop stability with varying sequences in hope of producing crystal contacts (15, 19). The structure of a GAAA tetraloop is defined by an abrupt turn in the phosphodiester backbone after the invariant G nucleotide (marked by asterisk; Figure 3-2A). Stacking interactions and multiple hydrogen

89 bonding interactions stabilize the structure of the loop and closing duplex base pairs (4). As shown in Figure 3-2B, the general nomenclature of the RNAs is SLII gnra-x, where gnra represents the sequence of the tetraloop from 5' to 3' (e.g. gaaa and guga) and -x indicates the number of base pairs removed from the top of the wild-type 14 bp SLII stem. Therefore SLII guga-5 is a hairpin containing a tetraloop with sequence 5'-GUGA-3' and nine base pairs of the lower stem region of SLII. All RNAs contain the required 5' four nucleotide and 3' six nucleotide single stranded tails. It is our hope we can find the optimal stem length for crystallization by screening SLII tetraloops of different sizes. Complexes were made with different combinations of the four FinO constructs and nine SLII dervatives: wild-type SLII, SLII gaaa-0, SLII guga-1, SLII guga-2, SLII gaga-3, SLII gaaa-4, SLII guga-5, SLII gaaa-6, and SLII guga-7 (Figure 3-2B). Various commercial and homemade sparse matrix screens were used to screen the complexes producing a number of crystal forms ranging from small needles to blocks. The most promising crystal contained the FinO 26-186 W36A/SLII guga-5 complex producing small blocks which grew in 0.1 M propionic acid pH 4.5, 0.2 M ammonium sulfate, and 30% PEG 4000. A dataset for this crystal was collected at the synchrotron (Figure 3-2C). It diffracted to 4.5 A with a space group of 0222^ however, we could not solve the structure using molecular replacement with the FinO 26-186 model. This may have been due to the small amount of usable data or that the structure of FinO in the complex was quite different from the search model. The statistics for the data set are given in Table 3-1. In the next round of crystallization trials, we further truncated the FinO 26-186 W36A truncation mutant to FinO 33-186 W36A. The area between residues 26 and 32 does not have any density in the FinO 26- 186 crystal structure indicating that these residues are flexible (see Figure 2-1D) (12). Therefore, we wanted to test the new FinO construct with SLII guga-5 under the conditions that grew the FinO 26-186 W36A/SLII guga-5 crystals to see if the crystals could be improved. We could not reproduce the crystals in the original conditions possibly due to the use of new stock solutions or different FinO or SLII purification protocols.

90 Therefore we decided to re-screen the new complex with commercial screens. The only condition which crystallized the FinO 33-186/SLII guga-5 complex was 0.4 M ammonium dihydrogen phosphate, producing long thin needles in approximately seven days (Figure 3-2D). Crystals were also found in this condition for FinO 33-186 W36A in complex with two other SLII-5 tetraloops: GAGA and GAAA (Figure 3-2D). At the time we did not test whether the crystals contained RNA. However, later crystals which grew in similar conditions (see Figure 3-3D) did contain the SLII derivative, increasing the likelihood that all the crystals contained RNA. To ensure we were not missing crystallization opportunities between residues 26-32, three additional FinO W36A truncation mutants were constructed: FinO 27-186 W36A, FinO 29-186 W36A, and FinO 31- 186 W36A and their ability to crystallize was tested, along with FinO 33- 186 W36A, in conditions which varied the ammonium dihydrogen phosphate concentration of the well solution from 0.25-0.5 M. The concentrations of the complexes were varied from 2.5-10 mg/mL and as an additional parameter, we varied the sequence of the GNRA tetraloop: 5'-GUGA-3', 5'-GAGA-3' and 5'-GAAA-3'. The sequence of the tetraloop was varied to increase crystal contacts between neighboring complexes. The results, summarized in Figure 3-3A, show that while nearly all FinO constructs produced crystals with SLII guga-5, the largest crystals contained FinO 29-186 W36A and 33-186 W36A. In each case, needles or rods were obtained around nine days, often growing out of brown precipitate. It appeared that FinO 33-186 W36A in complex with SLII gaaa-5 produced the best looking crystals (Figure 3-3B). Two crystal forms grew in 0.4 M ammonium dihydrogen phosphate. The first appeared from the initial screen at one month producing small, oval shaped crystals (Figure 3-2D) and the second produced long rods at nine days which had thickness (largest width approximately 0.03 mm) and well defined edges (Figure 3-3B). The crystals were very sensitive to the concentration of the precipitant. Unfortunately, both these crystal forms did not diffract well at our home source with the only a few low resolution (30 A) spots visible. Two-dimensional plates also occurred at

91 approximately 17 days for the FinO 33-186 W36A/SLII gaga-5 complex (Figure 3-3C). Most of these were quite small (0.08 x 0.08 mm) in dimension but had well defined edges. Once again diffraction of the crystals was poor with few reflections in the 28-12.5 A resolution range at both home and the synchrotron. For both crystals above, three cryoprotectants were used to freeze the crystals: 35% v/v glycerol, 35% v/v ethylene glycol, and 35% v/v polyethylene glycol 400. These cryoprotectants were chosen based on the 0.4-0.45 M ammonium dihydrogen phosphate well solution (22). The percentage of cryoprotectant was determined systematically by dunking crystals in increasing percentages, freezing and monitoring their diffraction. Since the sequence of the tetraloop seemed to play a role in crystallization of the FinO 33-186 W36A/SLII gnra-5 complex, we tried two additional loop sequences with pyrimidines at the "N" position: 5'- GCAA-3' and 5'-GUAA-3'. However, crystals did not materialize from these complexes. To ensure that RNA was in the FinO 33-186 W36A/SLII gaaa-5 crystals, we harvested a rod crystal from day seven and washed it three times with pseudo-mother liquor (0.425 M ammonium dihydrogen phosphate) to remove the excess RNA which did not crystallize in the drop. The washes were kept to monitor the removal of the excess RNA from the crystal. Finally, the crystal was dissolved in distilled water. The wash and crystal solutions were radiolabeled with y-32P-ATP and run on a 20% urea denaturing polyacrylamide gel. The results indicate that SLII gaaa-5 was in the crystal and that it was not degraded significantly (Figure 3-3D).

Optimizing the crystallization conditions with additives and cryoprotectants Due to the poor diffraction results of the two crystal forms we decided to optimize the crystals by adding various polyamines, divalent cations (MgCI2 and MgS04), and cobalt (III) hexamine which are known to stabilize nucleic acids and often improve the quality of RNA containing crystals (14, 26). For the FinO 33-186 W36A/SLII gaaa-5 rods, we added

92 0.5 and 1 mM of either spermine or spermidine to the complex solution. The chemical structures of the two polyamines are shown in Figure 3-4A. As shown in Figure 3-4B, compared to complexes without polyamines, the crystals grew much thicker (0.05 mm) and longer (0.3 mm) with 0.5 mM sperimine. However, increasing the concentration of spermine to 1 mM had negative effects on the crystals, slightly deforming them. The effect of spermidine was not as advantageous as sperimine, as the rods did not increase in girth or length (Figure 3-4B). The crystal enhancing effects of polyamines did not translate to the FinO 33-186 W36A/SLII gaga-5 complex as crystals did not grow under these conditions. Similarly, the addition of Mg2+ and cobalt (III) hexamine had negative effects on crystallization of both complexes. Therefore it appears that the crystals are very sensitive to the effect of polyamines and other stabilizing agents. Unfortunately, the diffraction quality of the larger FinO 33-186 W36A/SLII gaaa-5 crystals did not improve in the presence of 0.5 mM spermine (20 A resolution at synchrotron). Therefore, we wondered if the poor diffraction quality could be due to the cryopreservation process. RNA crystals tend to be very fragile due to high solvent content and typically have less diffracting power than protein crystals (14, 19). Therefore finding the correct cryoprotectant is important to protect the already poorly diffracting crystals from radiation damage. We decided to try growing the FinO 33-186 W36A/SLII gaaa-5 (+ 0.5 mM spermine) crystals in the presence of various concentrations (up to 10% v/v) of the three cryoprotectants we previously used for freezing crystals. We hoped this would minimize any negative effects the cryoprotectant may have at the time of freezing because the crystal would have already grown in its presence. The crystal could then be transferred step-wise into higher concentrations of cryoprotectant. The results show that the crystals could be grown in the presence of all three cryoprotectants, however, the order of quality was: ethylene glycol > PEG 400 » glycerol (Figure 3-5). All crystals could only be grown in cryoprotectant concentrations less than 2.5% v/v. The presence of the cryoprotectant did not seem to help the quality of diffraction for the FinO 33-186 W36A/SLII gaaa-5 crystals.

93 We also tried seeding techniques to try and improve the quality of crystals. FinO 33-186 W36A/SLII gaaa-5 crystals were crushed and streaked through freshly prepared drops containing the same complex and crystallization condition. The quality of the crystals which ensued was much poorer than the original crystals taking on a long, thin needle form rather than the thicker rods (data not shown).

The use of modular SI I duplexes to improve crystal contacts In previous studies we determined that a nine base pair SLII construct produced the best diffracting crystals. Assuming that FinO binds to the bottom portion of the SLII hairpin near the 5' and 3' tails (later confirmed by biochemical studies; see Chapter 4), it would appear that crystal contacts were occurring at the top of the duplex since the length of the stem and the sequence of the tetraloop were factors in crystallization. To follow up on this finding and to improve the number of crystal contacts, we tested various SLII-derived duplexes lacking the loop region (denoted SI I) for their ability to crystallize in complex with FinO 33-186 W36A. The nine different duplexes spanned a stem length of 8-10 base pairs while retaining the critical 5' four nucleotide and 3' six nucleotide single stranded tails. In addition to varying the stem length, the top of the Sll duplexes also varied from blunt end to having one or two nucleotide overhangs. To form the duplexes, we synthesized six Sll strands: Sll- 4A, SII-5A, SII-6A, SII-4B, SII-5B, and SII-6B. As shown in Figure 3-6A, different combinations SII(A) (shown in bold) and SII(B) strands were annealed to produce the nine duplexes. This combinatorial crystallization technique was used successfully for crystallizing the catabolite gene activator protein with its target DNA (32) and the ribosomal L30-RNA complex (17). FinO 33-186 W36A bound to all the duplexes as well as wild-type SLII and its tetraloop derivatives (data not shown).

The nine FinO 33-186 W36A/SII duplex complexes were formed and screened using commercially available sparse matrix screens. Only two crystals resulted from the screens: a short needle crystal (0.02 x 0.2 mm) of FinO 33-186 W36A with the blunt ended SII-5A5B after eight days (in 0.1 M sodium citrate tribasic dihydrate pH 5.6, 1 M lithium sulfate

94 monohydrate, and 0.5 M ammonium sulfate), and a very long, thin crystal (0.02 x 1 mm) from FinO 33-186 W36A with the SII-5A4B overhang duplex at day 105 (in 4 M sodium formate) (Figure 3-6B). Both crystals did not diffract, indicating that the tetraloop appears to be needed for either crystal contacts or stability of the Sll duplex as its removal seems to disrupt the ability of the complex to form crystals which diffract.

Employing the U1A crystallization module The negatively charged phosphate backbone of RNA provides less opportunity for crystal contacts compared with proteins (19). For this reason, the RNA binding domain (RBD) from the U1A spliceosomal protein is often used to help increase the number of intermolecular interactions within the RNA crystal (8). The U1A RBD is a small, highly basic 102 residue protein which binds with high affinity (KD ~10 nM) to stem-loop II of U1 RNA in the U1 small nuclear ribonucleoprotein (snRNP) (18). The crystal structure of a U1A 1-98 double mutant (DM) (Y31H, Q36R) in complex with a 21 nucleotide RNA derived from stem- loop II of U1 snRNA shows that U1A binds to its RNA target at the 5' side of the 10 nucleotide loop (27). Ferre-D'Amare and Doudna used the U1A DM mutant to solve the structure of the hepatitis delta virus (HDV) ribozyme (8). They swapped an area of the HDV ribozyme which was not critical for its function for the stem-loop RNA recognition sequence for U1A DM. Crystals of the U1A/HDV ribozyme complex were of high quality, diffracting to 2.3 A. The primary reason for this enhancement was due to the increased formation of protein-protein crystal contacts between molecules in the asymmetric unit and neighbouring lattice molecules (Figure 3-7A). Inspired by this work, we set out to introduce a U1A DM binding site onto SLII RNA so that both FinO and U1A could bind. The hope was that we could crystallize a tripartite RNA-protein complex so we could visualize the molecular details of the FinO-SLII interface. We have shown that the loop of SLII is not required for the interaction with FinO so we decided to engineer the 10 nucleotide U1 stem-loop II snRNA loop with two closing CG base pairs (also required for U1A binding) onto the nine

95 base pair SI 1-5 stem which was shown to be important for crystallization with FinO 33-186 W36A. Three constructs were made with 0 (U1ASII0), 1 (U1ASII1), and 2 (U1ASII2) base pairs in between the U1A binding site and the FinO site (Figure 3-7B). The spacer base pairs may be needed to accommodate binding of both proteins. Once again, the 5' and 3' single stranded spacers, required for FinO binding, are included in the constructs. Before setting up trays, we performed preliminary EMSA experiments to characterize the binding of FinO 33-186 W36A and U1A DM onto the RNA constructs. Both proteins were able to bind readily to U1ASII1 RNA (Figure 3-7C). There appeared to be two bands in the U1ASII1 RNA only lane which is probably due to annealing problems with the U1A binding area as both RNA bands bind FinO, however, the faster migrating species does not bind to U1A DM as well. It did not matter which protein was added to U1ASII1 RNA first as the binding of one protein did not inhibit binding of the other. Also, EMSAs showed that both proteins could bind equally well to U1ASII0 and U1ASII2 (Figure 3-7C). Furthermore, longer FinO constructs (FinO 27-186 W36A, 29-186 W36A, and 31-186 W36A) could bind as well as FinO 33-186 W36A indicating that the unstructured 27-32 region is not important for binding to the SLII portion of UlASIIx RNA and does not interfere with U1A binding (data not shown). A 1:1:1 molar ratio complex appeared to be optimal for crystallization trials as the majority of U1ASII RNA was shifted to the tripartite complex, however, there was a small portion of binary FinO-RNA and U1A-RNA complexes remaining (Figure 3-7C). The FinO 33-186 W36A/U1ASII1/U1A DM 1:1:1 complex was formed and put into crystallization trials at two different concentrations (5 and 10 mg/mL) employing a number of different commercial crystallization screens. We kept an aliquot of the complex and assayed it on a native gel two days after the trials were set up. The gel showed that the complex appeared to have formed properly and was still intact after the two days (data not shown). However, we did not see crystals in any of the 192 conditions screened.

96 Discussion and future directions

Ke and Doudna wrote in their review of the crystallization of RNA and RNA-protein complexes that "crystallization is predictably the least predictable aspect of a structure determination project" (19). This statement epitomizes this chapter which aims to summarize the attempts at crystallizing the FinO-SLII complex to obtain high resolution information on the interaction. Disappointingly, we were unable to obtain crystals which diffracted to a resolution that would allow us to solve the structure. The highest quality crystals of FinO 26-186 W36A in complex with a nine base pair truncated SLII tetraloop (SLII guga-5) diffracted to 4.5 A at the synchrotron. We further truncated FinO 26-186 W36A to FinO 33-186 W36A to remove the flexible 27-32 residues which may impede proper crystallization. We managed to crystallize this construct with various SLII gnra-5 RNAs however, even though the crystals were attractive visibly, they diffracted very poorly both at home and at synchrotron sources. Like many studies, we found that the sequence and length of the RNA construct were important factors in crystallization (14, 19). Three SLII-5 tetraloop sequences out of the five screened produced crystals: GAAA, GAGA, and GUGA with FinO 33-186 W36A. Another stable tetraloop sequence which was not tried is 5'-UNCG-3'. Like the GNRA tetraloop, the UNCG family is commonly found in ribosomal RNA stem-loop structures (21) and would provide four additional loop sequences to screen. In many cases needles, clusters of needles, small rods, and plates were obtained from the crystallization screens. These crystal forms are very common in RNA crystallization (19) and tend to diffract poorly thereby necessitating further optimization. The primary strategy was to employ RNA stabilizing additives such as polyamines (spermine and spermidine), Mg2+, and cobalt (III) hexamine. Out of all the compounds tried, spermine had the greatest effect, leading to multi-dimensional growth and creation of defined crystal faces and edges from the cylindrical rod crystals. However, the crystals still diffracted poorly suggesting that more optimization was needed. Other variables such as

97 + + + monovalent ions (Na , K vs. NH4 ) were tried unsuccessfully. One optimization tool which is often successful but was not tried in this study is the Hampton Additive Screen (Hampton Research, CA) which analyzes the effect of small molecules on crystal growth. The screen consists of large variety of molecules such as amino acids, various salts, polyamines, volatile and non-volatile organic compounds, chaotropes, reducing agents, carbohydrates, etc. Some of these molecules have been shown to enhance the size and quality of crystals (34). On the FinO side of the complex, truncations appeared to have a large effect on the frequency of crystallization. FinO 33-186 W36A produced the most crystals from the various screens tested. Perhaps the quality of the crystals could be improved by truncating the N-terminal a- helix further. Our initial goal was to obtain as much information on the FinO-SLII interaction, including the role of the N-terminal helix. However, it appears this domain of may interact weakly with SLII and therefore may interfere with crystallization of the complex. We showed in Chapter 2 of this thesis that deletion of the first 44 amino acids which comprise the catalytic portion of FinO, increased the binding affinity of the protein to SLII 20-fold. Also, our footprinting studies (Chapter 4) showed that FinO 45-186 appears to bind similarly to SLII as wild-type FinO and FinO 33- 186 W36A. The latter information was not known at the time of the crystallization study but may help to design future FinO-SLII crystallization trials using truncation constructs such as FinO 45-186 and possibly FinO 62-186.

While the the U1A crystallization module study did not produce any crystals, its potential was not fully explored. The effect of the spacer base pairs between the U1A and FinO binding domains was not tested as we only used one of the three UlASIIx RNA constructs. The spacer proved to be an important variable in crystallization of the HDV ribozyme (8). The annealing protocol for the U1ASII RNAs needs to be optimized as there are two RNA species on the native gels. Alternatively folded RNA species may lead to heterogeneity in the RNA-protein sample making it difficult to produce crystals. From the native gel in Figure 3-7C it can be seen that in addition to the tripartite complex there is a small

98 amount of FinO 33-186 W36A/U1ASII and U1A DM/U1ASII binary complexes existing. These may be an additional source of sample heterogeneity and could possibly be removed by gel filtration chromatography prior to crystallization setup. Also, FinO 45-186 or FinO 62-186 might be better crystallization constructs in the tripartite FinO/RNA/U1A complex as the presence of the N-terminus in FinO 33- 186 W36A may have hindered crystal growth. More crystallization conditions and additives should be tested to ensure all possibilities for crystal growth are explored. A number of lessons can be learned from this crystallization study. First, if possible, biochemical footprinting and activity assays should be performed to get information on the RNA-protein complex before starting crystallization trials. This can help develop rational protein and RNA constructs so as to maximize crystal contacts and prevent flexibility in the system which would hinder crystal growth. Secondly, in RNA-containing systems, all attempts should be explored to maximize crystal contacts. This may be through changing the RNA or protein sequence or by using additives which mediate RNA-RNA or RNA-protein interactions. The U1A crystallization motif is especially promising for RNA crystallization as U1A has been shown to mediate a large number of crystal contacts thus improving the diffracting power of crystals considerably (8). Finally, the purity of the RNA and protein components should not be understated. RNA should be homogeneous in length and be folded properly using optimized annealing protocols. Protein samples should be as pure as possible and be free of contaminants such as RNases which could compromise the RNA sample. The RNA-protein complex itself should be fully studied and its integrity checked after setting up crystallization trays. When possible, crystals should be checked to verify the existence of RNA by washing and radiolabeling dissolved crystals.

99 Materials and Methods

Cloning ofFinO constructs

FinO constructs were cloned using Gateway Technology (Invitrogen) and the overlap extension PCR technique to introduce mutations (16). Primers were synthesized by Integrated DNA technologies. Forward afrB PCR primers were designed to have the necessary homologous recombination sites and a cleavage site for PreScission protease (GE Healthcare). The sequences for the forward and reverse primer of FinO 1-186 (FOWTFWD/F0186REV), FinO 1-186 W36A (FOWTFWD/F0186REV)*, FinO 26-186 (F026FWD/F0186REV), FinO 26-186 W36A (F026FWD/F0186REV)*, FinO 27-186 W36A (F027FWD/F0186REV)*, FinO 29-186 W36A (F029FWD/F0186REV)*, FinO 31-186 W36A (F031W36A/F0186REV), and FinO 33-186 W36A (F033W36AFWD/F0186REV) are shown in Table 3-2. The primers were used to PCR off a pGEX-KG plasmid containing either the FinO 1-186 or FinO 1-186 W36A (marked by asterisk above) gene (13) using Platinum Pfx DNA Polymerase with standard protocol (Invitrogen). The amplified inserts were then recombined into pDONR 201 before being transferred into the pDEST-15 expression plasmid, which contained the glutathione­ s-transferase gene using standard Gateway protocol. Expression clones were then transformed into subcloning efficency E. coli DH5a cells (Invitrogen) and colonies screened for the insert using the BsrG1 restriction enzyme (New England Biolabs). Plasmids which tested positive for the insert were then sent for DNA sequencing to verify the correct FinO sequence (Molecular Biology Services Unit, University of Alberta Department of Biological Sciences). Plasmids were transformed into E. coli BL21-DE3 cells (Invitrogen) for protein expression.

Expression and Purification of FinO and U1A DM constructs FinO expression and purification is described in the "Materials and Methods" section of Chapter 2. The U1A 1-98 (Y31H, Q36R), referred to as U1A DM, was cloned into the pT7 plasmid by Oubridge et al. (28). The plasmid was a 100 generous gift from A. Ferre D'Amare (Fred Hutchinson Cancer Research Center, Seattle WA). The pT7/U1A DM plasmid (Ampr) was transformed into BL21 Gold cells (Stratagene) conferring kanamycin resistance and plated onto LB-Agar with kanamycin and ampicillin. Freshly picked colonies were used to innoculate 5 mL LB-Kan (0.05 mg/ml_)/Amp (0.1 mg/mL) cultures which were then grown overnight at 37°C. The starter cultures were then used to innoculate 1 L cultures of LB-Kan (0.05 mg/ml_)/Amp (0.1 mg/mL) which were grown overnight at 37°C. The cultures were pelleted by centrifugation at 4000 rpm and the cells frozen at -80°C for future use. The pellets were resuspended at 4°C in lysis buffer (25 mM Tris-HCI pH 8, 100 mM NaCI) with one Complete EDTA-free protease inhibitor cocktail tablet (Roche), stirred for 20 minutes, and then sonicated by six 10 second bursts with a one minute wait between bursts to allow the suspension to cool. The lysate was then centrifuged at 13000 rpm for 45 minutes. 0.05 volumes of 4 M ammonium sulfate and 0.03 volumes of 10% v/v poly(ethyleneimine) pH 7.5 were added to the pooled supernatant. Another protease inhibitor tablet was added and the suspension was stirred at 4°C for 20 minutes followed by centrifugation at 13000 rpm for 30 minutes. 0.7 volumes of saturated ammonium sulfate was added to the supernatant which was then stirred for 20 minutes and centrifuged at 13000 rpm for 20 minutes. The supernatant was discarded and pellets resuspended gently in lysis buffer. The first U1A DM purification steps were assayed by 18% SDS- PAGE. The resuspended U1A DM was loaded onto an SP-Sepharose Fast Flow ion-exchange column (GE Healthcare), equilibrated with low salt buffer (25 mM Tris-HCI pH 8.0). The column was then washed with 10 column volumes of low salt buffer. U1A DM was eluted from a linear gradient formed from low salt to high salt buffer (25 mM Tris-HCI pH 8.0, 1 M NaCI). U1A DM fractions were assayed by 18% SDS-PAGE, pooled, and concentrated to approximately 2 mL using an Amicon Ultra-free 5 K MWCO spin filtration device (Millipore, Fisher Scientific). The concentrated sample was brought up in 40 mL of high salt hydrophobic interaction chromatography (HIC) buffer (25 mM Tris-HCI pH

101 8.0, 2 M ammonium sulfate) and applied to a butyl sepharose Fast Flow HIC column (GE Healthcare) equilibrated in high salt HIC buffer. The column was washed with 10 column volumes of high salt HIC buffer. U1A DM was eluted from a linear gradient formed from high salt to low salt HIC buffer (25 mM Tris-HCI pH 8.0). U1A DM fractions were assayed by 18% SDS-PAGE, pooled, and concentrated using an Amicon Ultra-free 5 K MWCO spin filtration device (Fisher Scientific). The sample was applied to a Superdex 75 26/60 gel filtration column (GE Healthcare), equilibrated with 25 mM potassium phosphate pH 7.4, 100 mM NaCI. U1A DM fractions were assayed by 18% SDS- PAGE, pooled, and concentrated. The molecular weight of U1A DM after purification was 11190 ± 5 Da by MALDI-TOF mass spectrometry (University of Alberta Chemistry) corresponding to the N-terminal methionine being cleaved after synthesis, resulting in U1A 2-98 DM (2). The concentration of U1A 2-98 DM was calculated using the experimentally determined extinction coefficient at 280 nm of 4774 M"1cm" 1 (Alberta Peptide Institute).

Synthesis and purification ofRNA constructs The wild-type SLII from FinP, SLII gaaa-0, SLII guga-1, SLII guga- 2, SLII gaga-3, SLII gaaa-4, SLII guga-5, SLII gaaa-6, and SLII guga-7 RNA constructs were synthesized by in vitro transcription from DNA templates (Institute of Biomolecular Design, IBD, University of Alberta) (listed in Table 3-3)2. The T7 DNA primer: 5'-TAATACGACTCACTATA- 3' was synthesized by IBD. Prior to transcription, the templates were purified by urea denaturing polyacrylamide gel electrophoresis (PAGE). Template bands were visualized by UV shadowing, cut out and extracted from the gel slice by electroelution. Eluted samples were ethanol precipitated and resuspended in distilled water. The T7 primer was annealed to each template DNA by heating the DNA oligonucleotides to 90°C followed by slow cooling to room temperature. Transcriptions were carried out in 10 mL reactions: 0.5 mL 20X transcription buffer (0.8 M

2 The T7 in vitro transcription and purification of the SLII gnra-x constructs was performed by Ross A. Edwards. 102 Tris-HCI pH 8.1, 20 mM spermidine, 0.2% Triton X-100, 100 mM dithiothreitol), 1.6 ml_ 50% v/v polyethylene glycol 8000, 3.2 mL of 0.1 M solution of nucleoside triphosphates (adenosine triphosphate, uridine triphosphate, cytosine triphophate, guanosine triphosphate), 0.56 mL 1 M magnesium chloride, 0.05 mL 100 U/mL inorganic pyrophosphatase (Sigma-Aldrich), 0.064 mL 62.5 |iM annealed T7 primer/DNA template, 1 mL 0.5 mg/mL T7 RNA polymerase. Transcriptions were incubated at 37°C for four hours and stopped by phenol/chloroform extraction. RNA samples were ethanol precipitated and purified by urea denaturing PAGE. The RNA bands were visualized by UV shadowing, cut out, and extracted from the gel slice by electroelution. Eluted samples were ethanol precipitated, resuspended in 10 mM Tris-HCI pH 7.5, 1 mM EDTA, and quantified by UV absorbance using extinction coefficients calculated using an online oligonucleotide calculator (Ambion). The SLII gaaa-5, SLII gaga-5, SLII guga-5, SLII guaa-5, SLII gcaa-5, U1ASII0, U1ASII1, U1ASII2, SII-4A, SII-4B, SII-5A, SII-5B, Sll- 6A, and SII-6B RNAs (sequences shown in Figure 3-2B, 3-6A, 3-7B) were chemically synthesized using an Applied Biosystems DNA synthesizer which was modified for RNA synthesis using Dharmacon 2' ACE chemistry (30). The 2' OH of the synthesized RNAs were deprotected in 100 mM acetic acid titrated to pH 3.8 with N,N,N',N'- Tetramethylethylenediamine (TEMED) for 30 minutes at 60°C followed by drying down by SpeedVac concentration (Savant). Pellets were resuspended in 10 mM Tris-HCI pH 7.5, 1 mM EDTA, ethanol precipitated and dried down. Precipitated RNAs were resuspended in distilled water, mixed with denaturing gel load buffer (95% v/v formamide, 18 mM EDTA, 0.025% w/v sodium dodecyl sulfate, 0.025% w/v bromophenol blue, and 0.025% w/v xylene cyanol) and loaded onto an 8% urea denaturing gel. RNA bands were cut out and extracted by electroelution. Eluted samples were ethanol precipitated, resuspended in 10 mM Tris-HCI pH 7.5, 10 mM NaCI, 1 mM EDTA, and quantified by UV absorbance using extinction coefficients calculated from the sequence using an online oligonucleotide calculator.

103 Stem-loop RNAs (all constructs except Sll RNAs) were annealed prior to use in crystallization experiments to ensure they adopted the correct structure. This was carried out by heating the RNAs to 95°C followed by slow cooling to room temperature. Various Sll duplexes (Figure 3-6A) were annealed by incubating complementary strands at the concentration needed for crystallization in 10 mM Tris-HCI pH 8.1, 100 mM NaCI, and 1 mM EDTA. The RNA duplex samples were heated to 95°C followed by slow cooling to room temperature.

Characterization and formation of crystallization complexes To determine the correct stoichiometry of the FinO-SLII crystallization complexes, 100 pmol of SLII, SLII-x gnra, or Sll-x duplex was incubated with increasing amounts of FinO in a 10 \xL binding reaction: 5 pi 2X binding buffer (50 mM Tris-HCI pH 8.1, 50 mM NaCI, 2 mM EDTA, 2 mM TCEP), 1 pL of 100 jiM SLII, SLII-x gnra (annealed in 10 mM Tris-HCI pH 8.1, 10 mM NaCI, 1 mM EDTA) or SLII-x duplex (annealed in 10 mM Tris-HCI pH 8.1, 100 mM NaCI, 1 mM EDTA), 1 pi of FinO construct at increasing concentrations of 50, 100, 120, 150, and

200 fiM, and 3 pL ddH20. Complexes were incubated on ice for 30 minutes before adding 10 pL of 20% v/v glycerol and loading onto an 8% native gel which was pre-equilibrated with 1X tris-glycine pH 8. Gels were run for 2 hours at 4°C and bands were visualized by ethidium bromide. Binding reactions in which the RNA species was almost completely shifted to a complex (usually 1.2:1 molar ratio of RNA to protein) were chosen for crystallization trials.

Similarly, formation of the tripartite FinO/U1ASIIx/U1A DM complex was monitored by electrophoretic mobility shift assays to determine the optimal stoichiometry of its components. The binding reactions were 10 pL: 5 (iL 2X binding buffer, 1 pL of UlASIIx (where x refers to 0, 1, or 2 spacer base pairs; annealed in 10 mM Tris-HCI pH 8.1, 10 mM NaCI, 1 mM EDTA) at increasing concentrations, 1 pL of 100 (iM

FinO 33-186 W36A, 1 pL of 100 pM U1A 2-98 DM, and 2 pL of ddH20. Complexes were incubated on ice for 30 minutes before adding 10 pL of

104 20% v/v glycerol and loading onto a 12% native gel which was pre- equilibrated with 1X tris-glycine pH 8. Gels were run for 3-4 hours at 4°C and bands were visualized by ethidium bromide. Binding conditions in which the RNA species was almost completely shifted to a tripartite complex (e.g. 1:1:1 molar ratio of U1ASII1:FinO 33-186 W36A:U1A DM) were chosen for crystallization trials. The effect of additives, such as polyamines and divalent cations, on the complexes was monitored by EMSAs by including the additive into the 2X binding buffer. Crystallization complexes were formed quickly on ice. The 10X binding buffer consisted of: 250 mM Tris-HCI, pH 8.1, 250 mM NaCI, 10 mM EDTA, 10 mM TCEP, any additives). Upon mixing the components, a precipitate usually formed. This was likely due to the high concentration of the molecules but it disappeared after further mixing (complex formation). The complexes were incubated for 30 minutes on ice before setting up trays at 4°C.

Crystallization Setup FinO-SLII/SLII-x gnra/SII-x duplex and FinO-U1ASIIx-U1A DM complexes were put through a number of commercial crystallization screens including: the Classics suite (Nextal/Qiagen), Natrix (Hampton Research), Wizard I, and Wizard II (Emerald Biosystems). Hanging drop vapor diffusion crystallization setups were performed with 0.5 mL of crystallization solution in the well. 1 u.L of complex was mixed with 1 jaL of well solution on a siliconized glass coverslip which was applied to the top of the well and sealed with vacuum grease to initiate the vapor diffusion process. All trays were setup at 4°C and incubated in a vibration-free 4°C crystallization incubator (Molecular Dimensions). The dimensions, time of growth, and form of the resulting crystals are explained in the text, figures, and figure legends.

105 Harvesting crystals and X-ray diffraction screening Crystals were harvested using a suitable sized cryo loop (Hampton Research) and transferred from the crystallization drop to a mother liquor solution which was similar to the well condition. From there, the crystal was transferred to various cryoprotectants of different concentrations depending on the number of crystals available and the condition in which the crystal grew (22), and then quickly frozen using liquid nitrogen. If only one crystal resulted, usually 25% v/v glycerol was tried first as a cryoprotectant. Other cryoprotectants used were ethylene glycol and PEG 400. Some crystals were transferred to the cryoprotectant solution in a stepwise manner starting with a very low concentration (5%) and then into increasing amounts of the same cryoprotectant. Crystals were screened at home using an R-AXIS IV X-ray diffractometer (Rigaku) with a 0.3 or 0.5 mm collimator and image plate detector. Ten to 60 minute exposures at 0 and 90° were taken with a 0.5° oscillation. The crystal to detector distance was 80 mm. Some crystals were also screened using Beamline 8.3.1 at the Advanced Light Source synchrotron (Lawrence Berkeley National Laboratory in Berkeley, California) with a 0.03-0.1 mm collimator and an ADSC Q315 3x3 CCD detector. One to five second exposures at 0 and 90° were taken with a 0.5-1° oscillation at a crystal to detector distance of 150 ± 50 mm. FinO 26-186 W36A/SLII guga-5 crystals were screened at 350 mm. A data set was collected for one 26-186 W36A/SLII guga-5 crystal and processed with Denzo and Scalepack from the HKL suite3 (25). The data set statistics are shown in Table 3-1.

Verifying RNA presence in complex crystals Spermine-optimized FinO 33-186 W36A/SLII gaaa-5 crystals were harvested and transferred into 2 (iL of pseudo-mother liquor (425 mM ammonium dihydrogen phosphate) in a sitting drop bridge for the first wash. After a few minutes the crystal was transferred to 2 \xL of fresh pseudo-mother liquor for the second wash. The first wash solution was

3 This work was completed by Ross A. Edwards 106 transferred to a microfuge tube. This step was repeated for third wash, transferring the second wash solution into a microfuge tube. Finally, the crystal was transferred into 2 |al_ of fresh pseudo-mother liquor. 8 uL of ddH20 was added to the crystal to dissolve it. The 10 \iL sample was then transferred to a microfuge tube. The third wash solution was transferred to a microfuge tube. As a control, 2 \iL of fresh pseudo-mother liquor was added to a microfuge tube. 8 uL of ddH20 was added to the three wash solutions and control. All five 10 ^L samples were radiolabeled in 30 uL reactions: 10 |^L of crystal, wash or control solution, 6 ^il_ of 5X forward reaction buffer, 2 uL y32P-ATP (6000 Ci/mmol, Perkin Elmer), 2 u.L T4 polynucleotide kinase (10 U/|iL)(lnvitrogen), 2 ]xL Superaseln (Ambion) and 8 ^.L ddH20. 100 pmol of SLII gaaa-5 was radiolabeled in another reaction to act as a marker. The labeling reactions were incubated for 10 minutes at 37°C and purified using the QIAquick nucleotide removal kit (Qiagen), eluting in 50 p.L of 10 mM Tris-HCI pH 8.5. Purified radiolabeled samples were run on a 20% urea denaturing gel for analysis.

107 Figure 3-1: X-ray crystal structure statistics from the Protein Data Bank (PDB) as of July 22, 2008

Top: Pie chart representation of the number of X-ray crystal structures for the protein, DNA, RNA, nucleic acid-protein complexes, and other. Below: Pie chart representation of the number of DNA-protein, RNA-protein, and other crystal structures from the nucleic acid-protein category above.

108 A B N R G A G-C -0 GAAA A-U -1 GUGA G-C -2 GUGA G-C -3 GAGA G-C -4 GAAA A-U -5 GUGA C-G -6 GAAA G-C -7 GUGA U-A A-U G-C C-G U»G G-C 5' -GGACA GAUUUU -3' SLII gnra-x

D NP 1:1 NP 1:1 NP 1:1

I FinO 33-186 W36A/ RNA

I RNA

SLII guga-5 SLII gaga-5 SLII gaaa-5

a is wmmm Day 7 Day 14 Day 36

(Figure 3-2 legend on p. 110)

109 Figure 3-2: Employing SLII stem truncations and tetraloops for FinO-SLII crystallization (p. 109)

A) Structure of the 5'-GAAA-3' tetraloop from 4.5 S RNA from the E. coli signal recognition particle (PDB ID = 1HQ1) (1). A cartoon representation of the phosphodiester backbone is shown in blue. The dramatic change in direction of the backbone after the invariant G is marked by an asterisk. The bases of the tetraloop are shown in blue while the closing GC base pair is shown in grey. B) Schematic showing design of SLII gnra-x RNAs. The 14 base pair SLII stem sequence is shown with the critical 5' and 3' single stranded tails and a generic GNRA tetraloop. The stem truncation positions and their respected GNRA sequences are boxed. For example, SLII guga-1 is 13 base pairs and contains a 5'-GUGA-3' tetraloop sequence. C) Diffraction pattern of the FinO 26-186 W36A/SLII guga-5 crystals. D) Ethidium stained 8% native polyacrylamide gel of SLII guga-5, gaga-5, and gaaa-5 RNAs alone and in a 1:1 molar ratio complex with FinO 33- 186 W36A. Shown at the right of the gel slices are the positions of the free RNA and complex bands. Photos of crystals of the three FinO 33-186 W36A-SLII gnra-5 complexes from the 0.4 M ammonium dihydrogen phosphate condition are shown below each gel slice.

110 FinO 27-186 W36A FinO 29-186 W36A FinO 31-186 W36A FinO 33-186 W36A SLII guga-5 SLII guga-5 SLII guga-5 SLII guga-5 Day 9 Day 9 Day 9 Day 9

0.45MNH4H2PO4 >dNH4H /INH4H 0.5MNH4H2PO4 10 mg/mL 10 mg/mL 10 mg/mL 10 mg/mL

FinO 33-186 W36A FinO 33-186 W36A FinO 33-186 W36A FinO 33-186 W36A FinO 33-186 W36A SLII gaaa-5 SLII gaaa-5 SLII gaaa-5 SLII gaaa-5 SLII gaaa-5 Day 9 Day 9 Day 9 Day 9 Day 9 0.35 M NH4H2PO4 0.4 M NH4H2P04 0.4 M NH4H2P04 0.45MNH4H2PO4 0.5MNH4H2PO4 10 mg/mL 5 mg/mL 10 mg/mL 10 mg/mL 10 mg/mL D

FinO 33-186 W36A FinO 33-186 W36A SLII gaga-5 SLII gaga-5 Day 17 Day 17 0.25MNH4H2PO4 0.3 M NH4H2P04 5 mg/mL 5 mg/mL J PML 1 2 SLII gaaa-5 I—

Figure 3-3: Optimization of FinO-SLII gnra-5 complexes

A) Crystallization drop photos from SLII guga-5 RNA in complex with 27-186 W36A, 29- 186 W36A, 31-186 W36A, and 33-186 W36A (10 mg/mL) from day nine. The concentration of the NH4H2PO4 precipitant is noted. B) Crystallization drop photos from FinO 33-186 W36A in complex with SLII gaaa-5 at day nine. The concentration of the complex and NH4H2PO4 precipitant are noted. C) Crystallization drop photos from FinO 33-186 W36A in complex with SLII gaga-5 at day 17. The concentration of the complex and NH4H2PO4 precipitant are noted. D) FinO 33-186 W36A/SLII gaaa-5 crystal contains RNA. The crystal was washed three times with pseudomother liquor (PML) and dissolved in water. Crystal (X) and three wash samples along with a PML sample were radiolabeled with y32P-ATP, purified, and run on a 20% urea denaturing polyacrylamide gel alongside a radiolabeled SLII gaaa-5 marker.

111 A

Spermine

H;M

Spermidine

No Polyamine 0.5 M spermine 1 M spermine 1 M spermidine

O^MNH.H^O, 0.4MNH4H2PO4 0.4 M NH4H2PO,, 0.4MNH4H2PO4

No Polyamine 0.5 spermine 1 M spermine 1 M spermidine

0.425 MNH4H2PO„ 0.425 M NH4H2P04 0.45 M NH4 H.P04 0.42b M NH4H2P04

Figure 3-4: Optimization of FinO 33-186 W36A/SLII gaaa-5 complex with polyamines

A) Chemical structures of spermine and spermidine, the two polyamines used in this study. B) Crystallization drop photos of 10 mg/mL FinO 33-186 W36A/SLII gaaa-5 complex at seven days with various concentrations (final) of spermine and spermidine. In the top four photos, the concentration of the NH4H2PO4 precipitant is 0.4 M while it is 0.425 or 0.45 M in the bottom four.

112 No cryoprotectant 1%v/vPEG400 2.5% v/v PEG 400 1 % v/v glycerol

1 % v/v ethylene glycol 2.5% v/v ethylene glycol 5% v/v ethylene glycol

Figure 3-5: The effect of crystallizing the FinO 33-186 W36A/SLII gaaa-5 complex in the presence of cryoprotectants

Crystallization drop photos of 10 mg/mL FinO 33-186 W36A/SLII gaaa-5 complex crystallized in the presence of 0.5 M spermine and cryoprotectants such as PEG 400, glycerol, and ethylene glycol at various concentrations. The concentration of the precipitant was 0.425 M NH4H2P04.

113 A 3' 5' 3' 5' 3' 5' e-c G G A-U A-U A C-G C-G C-G G-C G-C G-C U-A U-A U-A SII-4(A) A-U A-U A-U 0-C G-C G-C C-G C-G C-G U.G U"G U.G G-C G-C G-C 5' -GACA GAUUUU -3' 5' -GACA GAUUUU -3' 5" -GACA GAUUUU-3' SII-4A4B SII-4A5B SII-4A6B

3' 5' 3' 5' 3* 5'

c A-U A C-A-GU C-G C-G G-C G-C G-C U-A U-A U-A S1I-5(A) A-U A-U A-U G-C G-C G-C C-G C-G C-G IMG U»G U.G G-C G-C G-C 5" -GACA GAUUUU -3' 5" -GACA GAUUUU -3' 5' -GACA GAUUUU-3 SII-5A4B SII-5A5B SI1-5A6B

3' 5' 3" 5' 3' 5' c u u C-G C-G C-G G-C G-C G-C U-A U-A U-A SII-6(A) A-U A-U A- U G-C G-C G-C C-G C-G C-G U.G U« G U« G G-C G-C G-C 5' -GACA GAUUUU -3' 5' -GACA GAUUUU -3' 5" -GACA GAUUUU-3 SII-6A4B SII-6A5B SII-6A6B

B

FinO 33-186 W36A FinO 33-186 W36A SII-5A5B SII-5A4B

Figure 3-6: Screening complexes of FinO 33-186 W36A with SLII-derived duplexes of varying length containing blunt or overhanging ends

A) Sequences of Sll constructs used in this study. Six Sll single strands were made: Sll- 4A, SII-4B, SII-5A, SII-5B, SII-6A, and SII-6B. These were combined to make nine different duplexes which were each bound to FinO 33-186 W36A at 10 mg/mL and put through various commercial sparse matrix screens. B) Photos of two crystals which materialized from the screen. FinO 33-186 W36A/SII-5A5B crystallized in 0.1 M sodium citrate tribasic dihydrate pH 5.6, 1 M lithium sulfate, and 0.5 M ammonium sulfate at eight days. FinO 33-186 W36A/SII-5A4B crystallized in 4 M sodium formate at day 105.

114 A

i ii iii fb» JL ftv m

B c u G c c u u U1ADM u c binding A C C-G site G-C C-G Spacer base pair(s) G-C A-U A-U A-U C-G C-G C-G G-C FinO G-C G-C U-A U-A U-A A-U binding A-U A-U G-C site G-C G-C C-G C-G C-G U«G U«G U«G G-C G-C G-C b' -GACA GAUUUU-3' 5'-GACA GAUUUU- S'-GACA GAUUUU. U1ASII0 U1ASII1 U1ASII2

FinO U1A FinO then U1A UlAthenFinO FinO/U1A FinO/U1A

(V *) N V V K K V «j- K *V *> - - - • X <• N •*.• s- A -•-'.- • ' *.' K" N." Is." K* IV K' K.' ^ >v K- iv .Jr *>••*»•• •»•'

~*-U1A/RNA/FinO ',-* FinO/RNA— ~* U1A/RNA—•

RNA

U1ASII1 U1ASII0 U1ASII2

(Figure 3-7 legend on p. 116)

115 Figure 3-7: Employing the U1A crystallization module to help crystallize the FinO/SLII complex (p. 115)

A) Various U1A-mediated crystal contacts in the hepatitis delta virus (HDV) crystal structure (PDB ID = 1CX0) (8). (i) U1A crystal contacts around a crystallographic 3-fold axis (denoted by black triangle), (ii) Three HDV-U1A complexes in the asymmetric unit forming around a crystallographic 3-fold axis, (iii) U1A-U1A interaction across a 2-fold crystallographic axis from neighboring complexes of neighboring asymmetric units. B) Sequences of RNAs used in this study to bind both U1A DM and FinO 33-186 W36A. The U1A DM binding domain is shown in blue and comprises the 10 nucleotide loop and closing 2 C G base pairs. The FinO 33-186 W36A binding site is shown in black and comprises a nine base pair SLII stem sequence and the 5' and 3' single stranded tails. The three RNA constructs differ in the number of spacer G C base pairs separating the two protein binding domains. This varies from none (U1ASII0), one (U1ASII1), or two (U1ASII2). U1ASII1 was the only construct used in crystallization trials. C) EMSAs showing the formation of the tripartite U1A/RNA/FinO complex. Left. FinO 33-186 W36A and U1A DM binding to U1ASII1. NP - No protein. The ratio of the FinO 33-186 W36A/U1ASII1, U1A/U1ASII1, and FinO 33-186 W36A/U1ASII1/U1A complexes is shown above the gel. The U1ASII1 ratio component is shown in bold. In these experiments, the protein concentrations were held the same while the RNA concentration was varied. We tried two binding scenarios to investigate if binding of one protein prevented binding of the other. In the first case, FinO 33-186 W36A was added to U1ASII1 before U1A and in the second case the order was reversed. The identity of the bands are annotated beside the gels. Right FinO 33-186 W36A and U1A DM binding to U1ASII0 and U1ASII2 RNAs.

116 Table 3-1: Data collection statistics for FinO 26-186 W36A/SLII guga-5 crystal collected at ALS 8.3.1

Data set

Wavelength (A): 1.1 Resolution (A): 50-4.5 Observations: 19275 Number of unique reflections: 5587 Redundancy: 3.4 Completeness (overall/last shell1)(%): 98.6/98.7 R-factor (overall): 0.094 R-factor (last shell): 0.45 l/oil) (overall/last shell): 12.0/2.6

1. Last shell is 4.7-4.5 A

117 Table 3-2: FinO truncation mutants Gateway attB PCR oligonucleotides

Forward Primers (PCR off FinO 1-186 pGEX-KG)*:

FOWTFWD: S'-GGGGACAATTTGTACAAAAAAGCAGGCTTCCTGGAAGTTCTGTTCCAGGGGCCC/ArG/ACyAG/AGCAG/AAACG/ACCGGra-a'

F026FWD: S'-GGGGACAATTTGTACAAAAAAGCAGGCTTCCTGGAAGTTCTGTTCCAGGGGCCCACC^rCATC^rGrC^CC/ACGCCA-a1

F031W36AFWD: 5'-GGGGACAAGTTTGTACAAAAAAGCAGGCTTCCTGGAAGTTCTGTTCCAGGGGCCC^CC/>CGCC/>CC^A4/tGCG^GGTG/>M 3'

F033W36AFWD: 5'-GGGGACAAGTTTGTACAAAAAAGCAGGCTTCCTGGAAGTTCTGTTCCAGGGGCCCCC/^CCAAMGCG^GGrGy^/^yA-3^

F045FWD: S'-GGGGACAAGTTTGTACAAAAAAGCAGGCTTCCTGGAAGTTCTGTTCCAGGGGCCCG/AG/MGGCTGCCCGGGA/iGCyqG/AG-S1

Forward Primers (PCR off FinO 1-186 W36A pGEX-KG)*:

, , F026FWD: 5 -GGGGACMTTTGTACAAAAAAGCAGGC^^^CCTGGAAGTTCTG^rTCCAGGGGCCCACC/^TC;A^C/=^/^rG^C/ACC/^CGCCA-3

F027FWD: S'-GGGGACAATTTGTACAAAAAAGCAGGCTTCCTGGAAGTTCTGTTCCAGGGGCCCATCATC^TGTC/ICC^CG-S'

F029FWD: 5'-GGGGACAATTTGTACAAAAAAGCAGGCTTCCTGGAAGTTCTGTTCCAGGGGCCCMTGrC/iCC/ACGCC^CC/ii-3

Reverse Primers (PCR off FinO 1-186 or FinO 1-186 W36A pGEX-KG)**:

F0186REV: 5'-GGGGACCACTTTGTACAAGAAAGCTGGGTCCTAtcaTTGTTCATCAAGCACGGCCTGAAGTTC-S

F061 REV: 5'-GGGGACCACTTTGTACAAGAAAGCTGGGTCCTAtta rCTGGCCrGCGCTTr7TTTGC-3'

* Normal type: attB1 sequence; Underlined: PreScission Protease sequence; Italics: FinO coding sequence; Bold: Alanine codon ** Normal type: attB2 sequence; Italics: FinO coding sequence; small case: Stop codon sequence Table 3-3: DNA templates for in vitro transcription of RNAs in Chapter 3*

FinP SLII: 5'-AAAATCGCCGATGCAGGGAGACGTGAACTCCCTGCATCGACTGTCCTATAGTGAGTCGTATTA-3'

SLII gaaa-0: 5'-AAAATCGCCGATGCAGGGAGTTTCCTCCCTGCATCGACTGTCCTATAGTGAGTCGTATTA-3'

SLIIguga-1: 5'-AAAATCGCCGATGCAGGGATCACTCCCTGCATCGACTGTCCTATAGTGAGTCGTATTA-3'

SLII guga-2: 5'-AAAATCGCCGATGCAGGGTCACCCCTGCATCGACTGTCCTATAGTGAGTCGTATTA-3'

SLII gaga-3: 5'-AAAATCGCCGATGCAGGTCTCCCTGCATCGACTGTCCTATAGTGAGTCGTATTA-3'

SLII gaaa-4: 5'-AAAATCGCCGATGCAGTTTCCTGCATCGACTGTCCTATAGTGAGTCGTATTA-3'

SLII guga-5: 5'-AAAATCGCCGATGCATCACTGCATCGACTGTCCTATAGTGAGTCGTATTA-3'

SLII gaaa-6: 5'-AAAATCGCCGATGCTTTCGCATCGACTGTCCTATAGTGAGTCGTATTA-3'

SLII guga-7: 5'-AAAATCGCCGATGTCACCATCGACTGTCCTATAGTGAGTCGTATTA-3'

* The T7 RNAP primer binding site is underlined; The loop sequence is in bold References

1. Batey, R.T., Sagar, M.B., and Doudna, J.A. (2001) Structural and energetic analysis of RNA recognition by a universally conserved protein from the signal recognition particle. J Mol Biol, 307: 229- 246.

2. Ben-Bassat, A., Bauer, K., Chang, S.Y., Myambo, K., Boosman, A., and Chang, S. (1987) Processing of the initiation methionine from proteins: properties of the Escherichia coli methionine aminopeptidase and its gene structure. J Bacterid, 169: 751-757.

3. Berman, H.M., Westbrook, J., Feng, Z., Gilliland, G., Bhat, T.N., Weissig, H., Shindyalov, I.N., and Bourne, P.E. (2000) The Protein Data Bank. Nucleic Acids Res, 28: 235-242.

4. Correll, C.C. and Swinger, K. (2003) Common and distinctive features of GNRA tetraloops based on a GUAA tetraloop structure at 1.4 A resolution. RNA, 9: 355-363.

5. Correll, C.C, Wool, I.G., and Munishkin, A. (1999) The two faces of the Escherichia coli 23 S rRNA sarcin/ricin domain: the structure at 1.11 A resolution. J Mol Biol, 292: 275-287.

6. Ehresmann, C, Baudin, F., Mougel, M., Romby, P., Ebel, J.P., and Ehresmann, B. (1987) Probing the structure of RNAs in solution. Nucleic Acids Res, 15: 9109-9128.

7. Ferre-D'Amare, A.R. and Doudna, J.A. (1996) Use of cis- and trans-ribozymes to remove 5' and 3' heterogeneities from milligrams of in vitro transcribed RNA. Nucleic Acids Res, 24: 977-978.

8. Ferre-D'Amare, A.R. and Doudna, J.A. (2000) Crystallization and structure determination of a hepatitis delta virus ribozyme: use of the RNA-binding protein U1A as a crystallization module. J Mol Biol, 295: 541-556.

9. Ferre-D'Amare, A.R., Zhou, K., and Doudna, J.A. (1998) Crystal structure of a hepatitis delta virus ribozyme. Nature, 395: 567- 574.

10. Ferre-D'Amare, A.R., Zhou, K., and Doudna, J.A. (1998) A general module for RNA crystallization. J Mol Biol, 279: 621-631.

11. Garber, M., Gongadze, G., Meshcheryakov, V., Nikonov, O., Nikulin, A., Perederina, A., Piendl, W., Serganov, A., and Tishchenko, S. (2002) Crystallization of RNA/protein complexes. Acta Crystallogr D Biol Crystallogr, 58: 1664-1669.

120 12. Ghetu, A.F., Gubbins, M.J., Frost, L.S., and Glover, J.N. (2000) Crystal structure of the bacterial conjugation repressor FinO. Nat Struct Biol, 7: 565-569.

13. Ghetu, A.F., Gubbins, M.J., Oikawa, K., Kay, CM., Frost, L.S., and Glover, J.N. (1999) The FinO repressor of bacterial conjugation contains two RNA binding regions. Biochemistry, 38: 14036-14044.

14. Golden, B.L. and Kundrot, C.E. (2003) RNA crystallization. J Struct Biol, 142: 98-107.

15. Heus, H.A. and Pardi, A. (1991) Structural features that give rise to the unusual stability of RNA hairpins containing GNRA loops. Science, 253: 191-194.

16. Ho, S.N., Hunt, H.D., Horton, R.M., Pullen, J.K., and Pease, L.R. (1989) Site-directed mutagenesis by overlap extension using the polymerase chain reaction. Gene, 77: 51-59.

17. Hoggan, D.B., Chao, J.A., Prasad, G.S., Stout, CD., and Williamson, J.R. (2002) Combinatorial crystallization of an RNA- protein complex. Acta Crystallogr D Biol Crystallogr, 59: 466-473.

18. Jessen, T.H., Oubridge, C, Teo, C.H., Pritchard, C, and Nagai, K. (1991) Identification of molecular contacts between the U1 A small nuclear ribonucleoprotein and U1 RNA. EMBO J, 10: 3447-3456.

19. Ke, A. and Doudna, J.A. (2004) Crystallization of RNA and RNA- protein complexes. Methods, 34: 408-414.

20. Kendrew, J.C, Bodo, G., Dintzis, H.M., Parrish, R.G., Wyckoff, H., and Phillips, D.C. (1958) A three-dimensional model of the myoglobin molecule obtained by x-ray analysis. Nature, 181: 662- 666.

21. Leulliot, N., Baumruk, V., Abdelkafi, M., Turpin, P.Y., Namane, A., Gouyette, C, Huynh-Dinh, T., and Ghomi, M. (1999) Unusual nucleotide conformations in GNRA and UNCG type tetraloop hairpins: evidence from Raman markers assignments. Nucleic Acids Res, 27: 1398-1404.

22. McFerrin, M.B., and Snell, E.H. (2002) The development and application of a method to quantify the quality of cryoprotectant solutions using standard area-detector X-ray images. J Appl Crystallogr, 35: 538-545.

23. Milligan, J.F., Groebe, D.R., Witherell, G.W., and Uhlenbeck, O.C. (1987) Oligoribonucleotide synthesis using T7 RNA polymerase and synthetic DNA templates. Nucleic Acids Res, 15: 8783-8798.

121 24. Niesen, F.H., Koch, A., Lenski, U., Harttig, U., Roske, Y., Heinemann, U., and Hofmann, K.P. (2008) An approach to quality management in structural biology: biophysical selection of proteins for successful crystallization. J Struct Biol, 162: 451-459.

25. Otwinowski, Z., and Minor, W., Processing of X-ray diffraction data collected in oscillation mode, in Macromolecular crystallography, part A, C.W. Carter Jr., and Sweet, R.M. 1997, Academic Press: New York. p. 307-326.

26. Ouameur, A.A. and Tajmir-Riahi, H.A. (2004) Structural analysis of DNA interactions with biogenic polyamines and cobalt(lll)hexamine studied by Fourier transform infrared and capillary electrophoresis. J Biol Chem, 279: 42041-42054.

27. Oubridge, C, Ito, N., Evans, P.R., Teo, C.H., and Nagai, K. (1994) Crystal structure at 1.92 A resolution of the RNA-binding domain of the U1A spliceosomal protein complexed with an RNA hairpin. Nature, 372: 432-438.

28. Oubridge, C, Ito, N., Teo, C.H., Fearnley, I., and Nagai, K. (1995) Crystallisation of RNA-protein complexes. II. The application of protein engineering for crystallisation of the U1A protein-RNA complex. J Mol Biol, 249: 409-423.

29. Park, H.H. and Wu, H. (2007) Crystallization and preliminary X-ray crystallographic studies of the oligomeric death-domain complex between PIDD and RAIDD. Acta Crystallogr Sect F Struct Biol Cryst Commun, 63: 229-232.

30. Scaringe, S.A. (2001) RNA oligonucleotide synthesis via 5'-silyl-2'- orthoester chemistry. Methods, 23: 206-217.

31. Scaringe, S.A., Wincott, F.E., and Caruthers, M.H. (1998) Novel RNA synthesis method using 5'-0-silyl-2'-0-orthoester protecting groups. J Am Chem Soc, 120: 11820-11821.

32. Schultz, S.C., Shields, G.C., and Steitz, T.A. (1990) Crystallization of Escherichia coli catabolite gene activator protein with its DNA binding site. The use of modular DNA. J Mol Biol, 213: 159-166.

33. Scott, W.G., Finch, J.T., Grenfell, R., Fogg, J., Smith, T., Gait, M.J., and Klug, A. (1995) Rapid crystallization of chemically synthesized hammerhead RNAs using a double screening procedure. J Mol Biol, 250: 327-332.

34. Sousa, R. (1995) Use of glycerol, polyols and other protein structure stabilizing agents in protein crystallization. Acta Crystallogr D Biol Crystallogr, 51: 271-277.

122 35. Wedekind, J.E. and McKay, D.B. (1999) Crystal structure of a lead-dependent ribozyme revealing metal binding sites relevant to catalysis. Nat Struct Biol, 6: 261-268.

123 124 Chapter 4

Examining the molecular details of the interaction of FinO with its target RNAs1

Overview

Conjugation is repressed through a mechanism involving the 79 nucleotide FinP antisense RNA. The RNA is composed of two stem- loops: SLI and SLII linked by a single stranded spacer and flanked by two single-stranded tails. SLII was previously determined to be the high affinity FinO binding site. FinO mediates base-pairing of FinP to its target traJ mRNA which leads to translational inhibition of the transcriptional activator TraJ and subsequent downregulation of conjugation genes. Little is known about how FinO binding to its RNA targets leads to the facilitation of FinP and traJ mRNA pairing. Here, we investigate the mechanism through which FinO binds specifically to FinP SLII and its complement SLIIc from traJ mRNA using low-resolution solution methods. The residues of FinO which contact SLII were determined by cross-linking experiments employing single cysteine mutants conjugated to the UV- activated 4-azidophenacyl bromide. Enzymatic footprinting experiments, employing various FinO truncation mutants, highlight the areas of SLII and SLIIc that are contacted by FinO.

Introduction

Bacterial conjugation in F-like systems is repressed by the two- component FinOP fertility inhibition system (7). FinP is a 79 nucleotide non-coding RNA which is transcribed from the antisense strand of the F- plasmid between the traM and traJ genes (6, 25, 32). FinO is a 21.2 kDa protein co-factor expressed from the end of the tra operon (33). It protects FinP RNA from cellular degradation (9, 17, 20) and promotes

1 Part of this work was previously published: Ghetu, A.F., Arthur, D.C., Kerppola, T.K., and Glover J.N. (2002) RNA 8: 816-823. 125 intermolecular base-pairing of FinP to the 5' untranslated region (UTR) of traJ mRNA (34). This leads to translational inhibition of the transcriptional activator TraJ, which activates downstream conjugation genes of the tra operon. FinP consists of two stem-loops (SLI and SLII) separated by a four nucleotide single stranded spacer and beginning with a four nucleotide leader sequence 5' to SLI and ending with a six nucleotide tail 3' to SLII (35) (Figure 4-1 A). It has been shown that the spacer region is susceptible to attack by the RNase E endonuclease, a component of the E. coli degradosome (17). FinO binds with high specificity to the SLII domain in a non-sequence specific manner, protecting FinP from degradation, thereby increasing its half-life in the cell (16, 20). Although the duplex portion of SLII is important, the critical requirement for efficient FinO binding is the single stranded spacer and 3' tail areas of FinP (16). It was shown that the loop region of SLII is not needed for binding FinO (see Chapter 2) (10). FinO binds similarly to the 5' UTR of traJ mRNA (16) which consists of three stem-loops: SLIM, SLIIc and SLIc (16) (Figure 4-1B). SLIIc is the high affinity FinO binding site and is separated by a six nucleotide spacer from SLIc which contains the ribosomal binding site (RBS) and the AUG start codon (boxed in Figure 4-1B). Upstream of SLIIc is a large A/U rich region which binds the E. coli host factor Hfq (36) and a third stem-loop (SLIM) at the 5' end of traJ mRNA. The molecular details of how FinO binds to its target RNAs are not fully understood. Previous studies showed that FinO contains two RNA binding domains: residues 26-61, comprising the solvent exposed N- terminal a-helix; and residues 62-186, comprising the stably folded core (Figure 4-1D) (12). Both FinO 1-61 and FinO 62-186 form complexes with SLII, however at lower affinities than full length FinO suggesting the presence of both domains is necessary for optimal RNA binding (12). Interestingly, truncations of wild-type FinO to FinO 26-186 and FinO 45- 186 increase its binding affinity 4 and 20-fold respectively, demonstrating that removing flexible portions of the N-terminus improves binding to SLII (see Chapter 2). This finding suggests that there is a minimum amount of the N-terminal helix needed for RNA binding and that addition of more N-

126 terminal residues leads to greater flexibility which decreases RNA binding. In-gel fluorescence resonance energy transfer (FRET) experiments showed that the N-terminal a-helix may make a conformational change to interact with the single stranded tails of SLII RNA, contradicting the previous thought that the helix would interact with the upper portion of the SLII stem (10). Attempts to crystallize the FinO- SLII complex remain unsuccessful (see Chapter 3) probably due to the dynamic nature of the N-terminal a-helix. In the crystal structure of FinO 26-186, the a-helix demonstrated substantially higher B-factors than the globular region indicating significant flexibility (11). Complexes employing FinO truncation derivatives lacking the first 32 amino acids (no density was seen for this region in the FinO 26-186 crystal structure) and a variety of SLII-like RNA constructs with stem truncations, tetraloops and sequence variations still resulted in crystals which diffracted poorly (Chapter 3). In this chapter, we use solution techniques to obtain more information about the interaction between FinO and its target RNAs. Cross-linking experiments between FinO cysteine mutants derivatized with the heterobifunctional 4-azidophenacyl bromide and SLII RNA show that FinO contacts the RNA at positively charged areas of the N-terminal a-helix and globular domain. Enzymatic footprinting determined where wild-type FinO and its truncation mutants FinO 33-186 W36A and FinO 45-186 contact FinP SLII and traJ SLIIc. Most interactions occur at nucleotides on the lower 3' side of the stem and 3' tail. Weaker interactions were apparent at the lower 5' side of the RNA stems and we did not observe interactions between FinO and the 5' tail of the RNAs. The upper half of the stem and loop regions of both RNAs were exposed to the RNases in the presence of FinO indicating that the protein did not contact this area. The truncation mutants bound to the RNAs in the same manner as full length FinO suggesting that FinO makes primary RNA contacts with its globular region, not the N-terminal a-helix. We also show there may be structure in the 5' tail of SLII and 3' tail of SLIIc which could contribute to the decreased RNase attack.

127 Results

Cross-linking demonstrates that FinO contacts SLII at positively charged areas of the N- and C-terminal a-helices and core globular domain In order to facilitate F\r\P-traJ mRNA pairing, FinO must first bind to its target RNAs. Previous studies show that FinO binds specifically and with high affinity to the SLII and SLIIc domains of FinP and traJ mRNAs respectively (Figure 4-1 A, B) (12, 16). The protein interacts with its target RNAs by recognizing the overall shape, rather than through sequence-specific interactions (16). An electrostatic surface potential representation of FinO 26-186 (Figure 4-2B) shows that FinO contains dense areas of positively charged residues throughout the entire protein. To determine if residues within these areas of FinO are important for RNA binding, we performed cross-linking experiments between SLII RNA and single cysteine FinO mutants labeled with the heterobifunctional 4- azidophenacyl bromide (APA-Br) cross-linking agent via a thioether linkage (Figure 4-1C). Ten FinO single cysteine mutants were generated at various positively charge residues in the N- and C-terminal a-helices and in the globular region (Figure 4-1C). In addition, glutamate 147 at the negatively charged bottom of FinO was also modified to cysteine. Two natural cysteine residues near the surface of FinO (C135 and C142) were also tested. To make the single cysteine mutants, we first developed a cysteine-less mutant (FinO 3C->S) where all three natural cysteines were mutated to serine. As this mutant should not be capable of conjugating to APA-Br, we would not expect any cross-linking to SLII. Each of the 14 mutants were labeled with APA-Br and bound to SLII. After a 10 minute incubation period at 4°C, the complexes were irradiated with UV light at 302 nm to activate the azide moiety of APA (boxed Figure 4-1C). Attack of the resulting nitrene on RNA or protein within 10 A of the cysteine residue generated cross-linked products which were resolved on an SDS- PAGE gel (Figure 4-2A). FinO-SLII cross-links are identified by ethidium bromide and coomassie staining. To ensure that the FinO-SLII cross­ links were specific, we performed parallel control experiments where each of the modified FinO cysteine mutants was incubated with yeast tRNA,

128 which does not bind FinO with high affinity (12, 29, 34), and tested their ability to form cross-links. As can be seen in Figure 4-2A, FinO was specific for SLII and did not efficiently form cross-links with tRNA. The results of the cross-linking experiments are shown in Figure 4-2A and summarized on an electrostatic surface potential representation of FinO 26-186 (Figure 4-2B). As expected, FinO (3C->S) did not form cross-links with SLII whereas cross-links formed from mutants in the N- terminus (K37C, K40C, K42C, and K46C), globular region (K81C, R121C, K125C, K165C), and C-terminal helix (K176C). The strongest cross-links resulted from the positively charged region encompassing residues R121, K125, and R165 showing that this region of FinO makes strong contacts with SLII. RNA contacts also formed on both front and back sides of FinO (Figure 4-2B). However, cross-linking indicates that residues on the front face appear to make the strongest contacts. The weak cross-links may result from SLII wrapping its single stranded tails around the back of the protein. Mutants from areas near the negatively charged bottom face of FinO (S142C and E147C) and rear face of the globular region (S135C and R170C) do not form cross-links indicating that they are not important in SLII binding (Figure 4-2A and B).

FinO binds to the lower duplex region of SLII and SLIIc We used enzymatic footprinting experiments to determine which areas of the target RNAs contact by FinO. RNase V1 is a 15.9 kDa endoribonuclease which specifically cleaves after nucleotides involved in base pairs and stacking interactions (23). SLII and SLIIc RNAs were radiolabeled with y-32ATP at the 5' or 3' end and subjected to limited RNase V1 digestion alone or in complex with FinO 1-186 WT, 33-186 W36A, or 45-186 (constructs shown in Figure 4-1D) at a 1:1 molar ratio. Before adding the ribonuclease to the reaction, an aliquot from each binding reaction was run on an 8% native gel to ensure the RNA-protein complex had formed (Figure 4-1E). The cleavage reaction was incubated for 60 minutes at 4°C and the products of the reaction were resolved on a 15% urea-denaturing gel. The results of the RNase V1 cleavage experiments with 5' and 3' end labeled SLII are shown in Figure 4-3A.

129 Footprints on the gel are areas where the band intensity in the FinO-SLII complex lanes is significantly decreased relative to the "No Protein" lane. This indicates that FinO protects this area of SLII from cleavage by RNase V1. Visible footprints (indicated by square brackets) were observed at residues G32-C39 on the 5' 32P-SLII gel (Figure 4-3A; left)) and residues A34-C39 on the 3' 32P-SLII gel (Figure 4-3A; right) corresponding to the lower 3' portion of the SLII double-stranded stem region. Also, a minor footprint was observed on the 5' 32P-SLII gel at positions C7 and G8 mapping to the lower 5' portion of the SLII stem. Quantification of the footprint regions is shown in Figure 4-3B. For this study, we chose band intensity decreases, or protection values (See Materials and Methods) > 2-fold relative to the "No Protein" sample as being significant. This value is shown as a horizontal rule across each graph. To address the differences between the 5' and 3' end-labeled data, we chose to report a FinO footprint occurring at residues A34-C39 of SLII. We compared three different FinO truncation constructs (Figure 4-1D) in this study to see what role the N-terminus has in RNA binding. The results show that the shortest construct, FinO 45-186, binds in a similar manner to the longer FinO 33-186 W36A and wild-type FinO 1- 186. This indicates the globular region of FinO, residues 62-186, appears to play the primary role in SLII binding.

An area of intense RNase V1 cleavage in the presence of all FinO constructs occurred at residues G11 to G14 on the 5' portion of the SLII stem with minor cleavages surrounding it (U10 and G15-G16) (Figure 4- 3A). Minor cleavages also occurred between C26 and G32 at the upper 3' side of the stem. Taken together, the RNase V1 cleavage results show that FinO only contacts the lower portion of the SLII stem region, leaving the upper portion exposed for attack. The RNase V1 cleavage results for traJ SLIIc were similar to SLII (Figure 4-4). FinO protected the lower portion of the 3' stem of SLIIc from RNase V1 cleavage at positions C32-U37 on 5' 32P-SLIIc and A36-A40 on 3' 32P-SLIIc. The discrepancy between the 5' and 3' end-labeled SLIIc experiments is unclear. RNase footprinting assays were repeated at least twice. Based on the results, we conservatively report a footprint from A36

130 to G39 to encompass strong cleavages similar to both gels. Like SLII, there appeared to be a very weak footprint at the lower 5' stem of SLIIc at position C10 (marked by asterisk in Figure 4-4A, left gel). Both 5' and 3' end-labeled SLIIc gels have strong RNase V1 cleavages at the upper portion of the stem on the 5' side. In the presence of all FinO constructs, cleavages occur at positions C14 to G18 with weaker cleavages surrounding this area. The upper 3' portion of the SLIIc stem shows weaker cleavages from C28 to G34. The results indicate that like SLII, FinO binds at the lower part of the SLIIc stem region while leaving the upper half exposed to RNase V1. A summary of the RNase V1 results for SLII and SLIIc is shown in Figure 4-7. FinO-protected nucleotides in footprint regions are enclosed within solid boxes, while black arrowheads denote areas of the RNA where RNase V1 cleavages of various magnitudes occur in the presence of FinO.

FinO does not bind to the loop region of its target RNAs It is shown in Chapter 2 that when the seven nucleotide loop is removed from SLII, leaving a 14 base pair duplex with tails (denoted Sll), FinO binds with the same affinity as wild-type SLII indicating the loop is not necessary for specific binding. We wished to confirm this result using RNase I footprinting to probe for interactions between FinO and single- stranded areas of SLII or SLIIc, including the loop and 5' and 3' tails. RNase I is a 27.2 kDa single strand specific ribonuclease which, like RNase V1, does not have sequence specificity (24). We performed the experiments in exactly the same way as the RNase V1 footprinting. A 1:1 molar ratio FinO-RNA complex was formed and RNase I digestion was carried out in a 60 minute reaction at 4°C. The cleavage products were resolved on a 15% urea-denaturing polyacrylamide gel. Figure 4-5 shows the results for the limited RNase I digestion of SLII and SLIIc in the absence and presence of various FinO constructs. For both RNAs the most intense cleavage was observed, as expected, at the single-stranded loop regions. There was a discrepancy as to where the cleavages occurred between the 5' and 3' end-labeled gels for both SLII and SLIIc.

131 For 5' P-SLII, strong cleavages occurred between nucleotides G16 to C23, while for 3' 32P-SLII, cleavages resided between U20 and U25. This difference may indicate that cleavage is processive, as the RNase stays attached to the RNA and makes cleavages at adjacent regions. Interestingly, for both SLII datasets, cleavage intensities increase at the loop and nucleotides in the 5' upper portion of the stem upon binding all FinO constructs. Additional weaker cleavages result at positions U19 and U26 for the 3' 32P-SLII gel. This phenomenon also occurs for the SLIIc cleavage data. For 5' 32P-SLIIc, strong cleavages occur between A19 and G25 with weaker cleavages at G18 and A26. Intense cleavages result between loop residues A21 to A27 for the 3' 32P-SLIIc dataset with a weaker cleavage at C28. A summary of the data for SLII and SLIIc is shown in Figure 4-7. Large open arrowheads denote intense cleavages which were similar in both the 5' and 3' end-labeled SLII or SLIIc data and occur in the presence of all FinO constructs. Smaller open arrowheads highlight weaker cleavages. Confirming previous results, we show that the lack of RNase I protection at the loop region of the RNAs indicates that FinO does not bind this area (10). The increase in RNase I loop cleavage upon incubation with the FinO constructs may result from FinO binding to one or both of the single stranded tails, protecting them from cleavage and thereby increasing RNase I attack at the loop region. This is discussed in more detail below.

FinO protects the 3' tails of SLII and SLIIc from RNase I degradation and enhances cleavage of the 5' tail of SLIIc Electrophoretic mobility shift assays show that a 3' single-stranded tail of at least six nucleotides on SLII (and FinP) is required to bind FinO- GST (16). Removing the tail in its entirety led to a 5.5-fold decrease in affinity for FinO-GST. Removing the four nucleotide 5' tail of SLII (spacer between SLI and SLII in FinP) led to a more modest 1.3-fold decrease in affinity and removal of both tails decreased affinity for FinO-GST by nearly 14-fold (16).

132 We wanted to confirm the importance of the single-stranded tails of SLII and also SLIIc of traJ mRNA by performing RNase I cleavage assays to see if FinO binding protects these areas of the RNAs. In the 5' 32P-SLII cleavage gel in Figure 4-5 it appears that the FinO constructs protect residues of the 3' tail from degradation by RNase I. However, none of the other gels, with the possible exception of 5' 32P-SLIIc, clearly show protection of the tail residues. For all the gels, we see enhanced cleavage at the loop region upon binding of FinO. This suggests that FinO may be protecting the single stranded tails from cleavage, thereby leading to an increase in the loop region. Upon performing an initial experiment to determine the amount of RNase I to add for limited cleavage conditions of 5' 32P-SLII alone (see Materials and Methods), we noticed at a higher concentration of RNase I (0.1 U/jaL final), the full-length SLII band was cleaved to 39-42 nucleotides in length, corresponding to the removal of the 3' tail (Figure 4- 6Aii, lane 0). This result was similar to the 5' 32P-SLII cleavage gel in Figure 4-5, however more pronounced. We repeated the experiment, adding FinO 1-186 WT to 5' end-labeled SLII at increasing concentrations. The cleavage pattern from the 15% urea-denaturing gel is shown in Figure 4-6AN and the corresponding 8% native binding gel from the binding reactions, prior to the addition of RNase I, is shown in Figure 4-6AL As the concentration of FinO increases, the 3' tail of SLII becomes protected from RNase I digestion. In accordance with the results from the limited RNase I digestion, cleavage of the loop region increases as the FinO concentration increases (Figure 4-6Aiii), showing the increasing RNase I accessibility at this site upon protection of the 3' tail. Similar protection occurs for FinO 33-186 W36A and FinO 45-186 (not shown). When the same experiment was repeated with 3' 32P-SLII, we did not observe protection of the 5' tail of SLII (Figure 4-6B). The RNase T1 band corresponding to position G5 of SLII is noted by the arrowhead on the 15% urea-denaturing gel. It can be seen in the "No Protein" lane that the full-length is not cleaved to a band near G5. However, the full-length band does increase as the different 1:1 molar ratio FinO-SLII complexes

133 are formed. The loss in signal in the "No Protein" lane would be due to the 3' tail (and the 3' 32P label) being degraded in the absence of protein. The counts are restored when the FinO constructs are bound to SLII. To determine whether or not FinO binds to the one or both of the single-stranded tails of SLIIc, we also probed the RNase I accessibility of the tails in this molecule in a set of experiments similar to those performed on SLII (Fig. 4-6C, D). Figure 4-6C shows the 5' 32P-SLIIc data at a final RNase I concentration of 0.1 U/pL. One sees that the full- length band is partially cleaved to G39 and A40, however, the 3' tail digestion is not nearly as significant as SLII. Although not clearly shown, the limited RNase I digest of 5' 32P-SLIIc displays the same partial cleavage of the 3' tail (Figure 4-5). When the FinO constructs are bound to 5' 32P-SLIIc, it can be seen that the site is protected from RNase I digestion. However, contrary to the SLII data, the full-length counts decrease relative to the "No Protein" lane upon addition of FinO. The total counts in the "No Protein" lane are significantly (~3-fold) less than the input and T1 ladder from the same experiment (Figure 4-6C). Since a similar amount of RNA was loaded in each lane, this finding suggests that the 5' 32P label and possibly 5' tail nucleotides are being removed. This finding was confirmed in Figure 4-6Di showing the results from the RNase I digestion of 3' 32P-SLIIc. Here we see that approximately five nucleotides are digested from the 5' end of SLIIc. Interestingly, as with the 5' 32P-SLIIc experiment, we see that digestion of the 5' end occurs upon binding of the FinO constructs. This suggests that FinO is somehow freeing the 5' tail of SLIIc for attack by RNase I. Also, the total counts from the "No Protein" lane are greatly reduced compared to the FinO-SLIIc complex lanes. This indicates RNase I is digesting the 3' tail and in the process removing the 3' 32P label. When the FinO constructs bind to 3' 32P-SLIIc enhanced cleavage is seen at the loop areas (Figure 4-6DN). In these lanes, the total counts are very similar to the RNase T1 ladder and input lanes indicating protection at the 3' tail by FinO. A summary of the data is shown in Figure 4-7 where dashed boxes on the schematics of SLII and SLIIc indicate protection from RNase

134 I upon FinO binding. We show that FinO protects SLII from RNase I at the 3' tail region whereas the 5' tail of SLII did not appear to be accessible to RNase I. Enhanced cleavage at the loop region of SLII is seen when FinO protects the 3' tail from digestion. FinO also protects SLIIc from digestion at the 3' tail, but interestingly, the 5' tail cleavage is enhanced upon the binding of FinO. These cleavages are indicated by "V" on the SLIIc schematic. FinO may increase the single-stranded character of the 5' tail upon binding leading to new sites of RNase I attack.

A terminal 3' OH on the 3' tail of SLII is required for FinO binding During our footprinting experiments we discovered that 3' end labeling SLII RNA with 3',5'-cytidine [5' 32P] diphosphate (pCp) completely abolished binding to all FinO constructs (Figure 4-8B). It appears that the phosphate on the 3' terminus of the cytidine nucleotide (Figure 4-18A) interferes with the SLII binding mechanism of FinO, suggesting that a free 3' OH on the terminal nucleotide of the 3' tail of SLII is needed for the specific interaction. To test this hypothesis, we first removed the 3' phosphate with T4 polynucleotide kinase and then oxidized the resulting 2', 3' c/'s-diol to a 2', 3' dialdehyde using sodium periodate (18). We followed the FinO-SLII interaction using electrophoretic mobility shift assays during both steps. The results, shown in Figure 4-8C, indicate that all FinO constructs bind efficiently to 3' dephosphorylated SLII, however the interaction is disrupted when the diol is converted to the dialdehyde. Therefore, the terminal 3' OH of SLII appears to be critical for binding FinO. The results of the experiments were similar for wild-type FinO, FinO 33-186 W36A, and FinO 45-186, indicating that residues C- terminal to 45 are important for forming specific interactions with the 3' tail of SLII.

135 Discussion

FinO contacts SLII at positively charged residues throughout the protein We used site-specific cross-linking to map the SLII binding interface of FinO. A large set of single cysteine mutants were generated spanning areas of the N- and C-terminal helices, and globular region. The majority of residues mutated were positively charged residues such as arginines and lysines, however a negatively charged patch at the bottom of FinO was also tested. The UV-activated cross-linker APA-Br was conjugated to the mutants and their ability to cross-link to SLII while in complex was measured. We found that FinO cross-links to SLII at multiple sites. However, the strongest cross-links involve residues 121, 125, and 165, comprising a large positively charged patch at the front face of the globular domain. Weaker cross-links formed at N-terminal residues 37, 40, 42, and 46, residue 81 at the back face of globular domain of FinO, and at residue 176 of the C-terminal helix. The negatively charged area at the bottom of FinO did not form cross-links, indicating that the area does not contact SLII. Therefore, it appears that FinO interacts with SLII primarily through electrostatic interactions. This agrees with previous studies that show that FinO binds in a non- sequence specific manner to SLII (16).

FinO binds to SLII and SLIIc at the lower half of the stem and the 3' tail FinO binds its RNA targets with high specificity and affinity (12). In this study we used RNase footprinting experiments to determine the areas of FinP SLII and traJ mRNA SLIIc which contact FinO. FinO protects the RNAs from RNase V1 cleavage at the lower half of the stem region (Figures 4-3 and 4-4). The biggest footprint occurs at residues comprising the 3' side of the stem regions whereas residues towards the middle and upper portions of the stems are cleaved readily in the presence of all FinO constructs. During a previous attempt at crystallizing the FinO-SLII complex we designed SLII tetraloop hairpins varying the stem lengths while preserving the four nucleotide 5' tail and six nucleotide 3' tail (Chapter 3). We found that if we delete seven of the 14 base pairs

136 from the top portion of the stem of SLII, the RNA construct binds to FinO as well as the full-length 14 base pair tetraloop. Taken together, this data suggests that FinO binds to the lower half of the duplex. FinO also protects the 3' tails of SLII and SLIIc from digestion by RNase I. In addition, we showed that FinO appears to require a 3' OH on the 3' tail for specific interactions with SLII. A 3' tail of at least 6 nucleotides in length was previously shown to be critical for high affinity FinO interactions (16). The same study showed that the 5' tail has a modest role in binding FinO. We were not able to detect significant FinO protection of the 5' tail in this study (Figure 4-6B). This may, however, be a limit of the RNase I cleavage assays as the four nucleotide 5' tail of SLII may be too short for endonucleolytic cleavage (5). Also, the 5' tail may be partially structured which may inhibit its digestion by the nuclease (see below). Surprisingly, the 5' tail of SLIIc is cleaved by RNase I more readily in the presence of FinO (Figure 4-6D). It appears from Figure 4-6D that SLIIc is clipped of the first five nucleotides. This indicates that FinO is not binding to the 5' tail and appears to be structurally rearranging it so it has greater exposure to RNase I. We also show that the loop area of SLII and SLIIc do not interact with FinO (Figure 4-5). This result confirms previous data which shows that FinO binds SLII-like duplex RNAs without loops (Chapter 2) and also that FinO binds a number of different FinP sequences from different plasmids which differ in their SLII loop region (16).

In addition to enzymatic footprinting experiments, we attempted chemical footprinting experiments using dimethylsulfate (DMS), diethylpyrocarbonate (DEPC) and hydroxyl radicals to probe for evidence of RNA-protein interactions (data not shown). Unfortunately, we did not visualize any footprints, which we attributed to the FinO having non- sequence specific RNA interactions (in the case of DMS and DEPC) and possibly the size of the probes being too small. The interactions between FinO and its target RNAs being non-sequence specific in nature would be in accordance with previous studies where length, rather than the sequence of the tails was the determinant of high affinity interactions (16).

137 Therefore, the overall shape of the lower part of the stem and 3' tail is the requirement for high affinity FinO-SLII and FinO-SUIc interactions.

FinO residues C-terminal to alanine 44 play the primary role in RNA binding Footprinting experiments show that all three FinO constructs bound to SLII and SLIIc in the same way. We were hoping to see differences in the RNase cleavage patterns for FinO 1-186, FinO 33-186 W36A and FinO 45-186, thereby allowing us to assign a role for the N- terminus in RNA binding. The fact that we did not see any differences suggests that the N-terminal 44 amino acids do not play a key role in RNA binding. However, we cannot rule out interactions between N- terminal residues and other areas of the RNA. If they were to occur, they would be transient in nature and therefore not applicable to a footprinting technique, which requires stable interactions between the protein and RNA so the RNA can be protected from RNase attack. Indeed, we show that N-terminal residues 37-46 form specific cross-links with SLII RNA. Due to its flexibility, the N-terminus could adopt a large number of conformations thus extending its area of RNA contact. This property would also prevent visualization of N-terminal RNA footprints. In-gel fluorescence resonance energy transfer (FRET) experiments show that residues 37 and 42 of FinO are in close proximity to the 5' tail of an SLII- like RNA duplex, indicating significant flexibility in the N-terminus or a possible structural rearrangement in its a-helix upon RNA binding (10). The may be ruled out since circular dichroism studies show that FinO does not make any large structural changes in its structure upon binding SLII (unpublished data).

The single stranded tails may contribute to the orientation of the stem loops in FinP and traJ mRNA Single stranded tails (also called dangling ends) adjacent to RNA duplexes play an important role in stabilizing double-stranded regions in a number of different RNAs such as tRNAs (21), pseudoknots (3), and siRNAs (26). Purine bases immediately 3', or to a lesser extent 5', to the

138 duplex region are able to interact stably with the adjacent closing base pair through stacking interactions (22, 31). The strength of the interaction depends on the type of closing base pair and the number of residues in the tail (27, 31). Freier et al. measured the free energy for different stacking scenarios (8). They found the greatest stabilization occurs when a 3' single-stranded purine forms an inter-strand stacking interaction with the guanine from the preceding GC base pair (increase in stability of -1.7 kcalmol"1 for G and -1.8 kcalmol"1 for A). When the base pair was CG, the increase in stability from the intra-strand stacking interaction was -1.3 kcalmol"1 for G and - 1.1 kcalmol"1 for A. 3' tails stacked more significantly than 5' tails; the latter having a typical increase in stability less than -0.5 kcalmol"1 Ohmichi et al. show that longer tails can increase duplex stability (27). RNAs with tails consisting of four single stranded adenines 3' to a GC base pair increased stabilization by -2.5 kcalmol"1 compared to -1.5 kcalmol"1 for one adenine. The majority of the stabilization was found to occur through the first adenine but the addition of subsequent bases increased the interaction. Also, they show that longer 5' tails with four adenines significantly stabilize the duplexes compared to one adenine, indicating that 5' stabilization of the duplex can occur with longer tails. It can be seen from the schematics of SLII and SLIIc in Figures 4- 1A and B, and 4-7 that both RNAs have the correct elements to form stacking interactions at the 3' tail and possibly at the four nucleotide 5' tail for SLII. In this study, the 3' tail of SLII, which has G40 and A41 adjacent to the G5C39 closing base pair, was very susceptible to RNase I cleavage in the absence of FinO and was protected from digestion in the presence of FinO. The enzyme seemed to cleave the tail down to 39-41 nucleotides in length (Figure 4-6A). In contrast, SLIIc appeared to be less susceptible to RNase I attack. A portion of SLIIc was cleaved down to 39 and 40 nucleotides, however much less than SLII. The difference may lie with the poly(U) sequence at the end of SLII which would give the tail more single strand character and serve as a better substrate for RNase I (24). The 3' tail of SLIIc is also six nucleotides in length but has a greater variety of bases in its sequence than SLII.

139 O'Toole et al. show that duplex 3' overhangs with a 5' purine- pyrimidine 3' dinucleotide sequence following the closing base pair, as in SLIIc, stabilize a duplex more than a 5' purine-purine 3' sequence (as in SLII) (26). The NMR structure of the T2 bacteriophage gene 32 mRNA pseudoknot has a purine-pyrimidine dinucleotide 3' tail adjacent to a GC base pair (15). The structure of the 3' region shows the adenine makes an inter-strand stacking interaction with the guanine from the closing base pair and the cytosine appears to be stacking with the adenine. The authors report the cytosine in a near-C3' endo sugar conformation which is indicative of an >A-form helix (28). In this study, 3' 32P-SLIIc also appears to be cleaved strongly by RNase V1 at nucleotide A40, which may indicate a stacking interaction between this base and the C9G39 closing base pair (Figure 4-4A). The 5' tail of SLII appears to be protected from cleavage by RNase I and this may be due to a potential stacking interaction between A4 and the closing G5C39 base pair or because of its short length combined with its proximity to the stem region. One possible role for the stacking interactions at the tails of SLII and SLIIc may be to help fold FinP and the 5' UTR of traJ mRNA into their appropriate conformation. The 5' portion of SLII corresponds to the 4 nucleotide spacer between SLI and SLII and the 3' tail of SLIIc represents the spacer region between SLIIc and SLIc. Partial stacking between the G5C39 closing base pair in SLII and A4 and a potential stacking interaction with C3 could extend the helical character of SLII and constrain it to a specific angle with SLI. This could put the two helices into a near-coaxially stacked orientation, which is a common for helices in structured RNAs (4, 19). Likewise, A40 and possibly C41 from the 3' tail of SLIIc may stack with the closing C9G39 base pair to put an angle between SLIIc and SLIc in the 5' UTR of traJ mRNA. Further biochemical and structural studies with both FinP and traJ mRNA in their entirety are needed to confirm this hypothesis.

140 Materials and Methods

Production ofFinO cysteine point mutants2 FinO and variants containing single cysteines were expressed as GST-fusions from the pGEX-KG vector (11, 12). All substitutions in FinO were introduced using the PCR overlapping amplification protocol (14). Initially, we constructed a cysteine-free finO clone, in which the codons encoding the three cysteines in the wild-type protein were mutated to serines. Cysteine codons were then substituted for the codons for residues Lys 37, Lys 40, Lys 42, Lys 46, Lys 81, Arg 121, Lys 125, Glu 147, Arg 165, Arg 170 or Lys 176. Proteins containing native cysteine residues at either position 135 or 142 were also prepared. DNA sequencing was used to confirm the presence of the cysteine substitutions. Expression and purification of all the cysteine mutants was as described (12). Protein concentrations were determined using the BIORAD Bradford assay, which was calibrated for true molar concentration by amino acid analysis. We used a dithionitrobenzoate assay (DTNB, Sigma) (13) to determine the reactivity of the cysteine residues at pH 7.0. This assay revealed that at least 90% of the thiol groups were accessible to DTNB for each cysteine mutant.

Cloning of FinO truncation constructs The cloning procedures are described in Chapter 3 "Materials and Methods".

Expression and Purification of FinO truncation constructs FinO expression and purification is described in the "Materials and Methods" section of Chapter 2.

2 This work was performed by Alexandru F. Ghetu 141 Preparation and labeling of RNA constructs for RNase footprinting experiments SLII and SLIIc constructs (sequences shown in bold in Figure 4- 1A and B) were chemically synthesized using an Applied Biosystems DNA synthesizer which was modified for RNA synthesis using Dharmacon 2' ACE chemistry (30). The 2" OH of the synthesized RNAs were deprotected in 100 mM acetic acid titrated to pH 3.8 with N,N,N',N'- Tetramethylethylenediamine (TEMED) for 30 minutes at 60°C followed by drying down by SpeedVac concentration (Savant). Pellets were resuspended in 10 mM Tris-HCI pH 7.5, 1 mM EDTA, ethanol precipitated and dried down. Precipitated RNAs were resuspended in 10 mM Tris-HCI pH 7.5, 1 mM EDTA and quantified by UV absorbance using extinction coefficients calculated using an online oligonucleotide calculator (Ambion). For 5' end-labeling experiments, 200 pmol SLII or SLIIc was added to a 60 pL reaction consisting of 6 ^L 10X T4 polynucleotide kinase reaction buffer, 8 pL y32P-ATP (6000 Ci/mmol, Perkin Elmer), 8 uL 10 U/p.L T4 polynucleotide kinase (New England Biolabs), and 10 ^L Superaseln (Ambion). Reactions were incubated for 10 minutes at 37°C. For 3' end-labeling experiments, 100 pmol of SLII or SLIIc was added to a 60 uL reaction consisting of 6 pL 10X RNA Ligase I reaction buffer, 30 p.L 3',5'-cytidine [5' 32P] diphosphate (pCp) (Perkin Elmer), 2 |aL 20 U/|iL T4 RNA Ligase I (New England Biolabs), 6 JJ.L dimethyl sulfoxide, and 5 \xL Superaseln. Reactions were incubated overnight at 4°C. 5' and 3' labeling reactions were then added to 60 (J.L formamide gel load buffer (95% formamide, 18 mM EDTA, 0.025% sodium dodecyl sulfate, 0.025% bromophenol blue, and 0.025% xylene cyanol) and heated to 90°C for 2 minutes before being loaded onto an 8% urea-denaturing gel for purification. Gels were exposed for an optimized period of time and bands corresponding to the full length RNAs were cut out, crushed and soaked in 0.3 M sodium acetate pH 5.3 with a small volume of 25:24:1 phenol/chloroform/isoamyl alcohol (IAA) (50 ^L per 0.5 mL of NaOAC) and incubated on a shaker overnight at 37°C. The RNA samples were then filtered from the crushed acrylamide, chloroform/IAA extracted, and ethanol precipitated. The 5' end-labeled pellets were resuspended in 200 142 uL10 mM Tris-HCI pH 7.5, 10 mM NaCI and 1 mM EDTA. The 3' end-

labeled pellets were resuspended in 50 pl_ ddH20. An additional step was needed to remove the 3'-phosphate from the added cytidine of the 3' end-labeled RNAs as it prevented binding of the FinO constructs. 7 pL 10X T4 polynucleotide kinase buffer, 4 \iL polynucleotide kinase, 5 jaL of

Superaseln and 4 p.L ddH20 was added to the 50 JJL samples. Reactions were incubated for 10 minutes at 37°C and loaded onto an 8% urea-denaturing gel. Gel purification of the 3' dephosphorylated end- labeled RNAs was the same as above and ethanol precipitated pellets were resuspended in 200 ^L 10 mM Tris-HCI pH 7.5, 10 mM NaCI, and 1 mM EDTA. Prior to use in the footprinting experiments, all labeled RNA stocks were annealed by heating to 95°C for 1 minute followed by slow cooling to room temperature.

Protein-RNA cross-linking The cross-linker APA (Sigma) was initially dissolved in methanol to a final concentration of 208 mM. 1 fil_ APA stock was added to 100 |iL 80 nM protein solution containing the buffer 10 mM Tris (pH 7.0), 600 mM NaCI and 1 mM EDTA. The reaction mixture was then incubated in the dark for two hours at room temperature. Excess APA was subsequently removed using a BIORAD P-30 spin column pre-equilibrated with 10 mM Tris (pH 7.0), 600 mM NaCI and 1mM EDTA.The cross-linking reactions were performed with 42 |iM protein and 81 JJ.M SLII or yeast tRNA (Type X, Sigma) that had been pre-incubated for 10 min. at 4°C, in 10 mM Tris (pH 7.0), 600 mM NaCI and 1 mM EDTA. Reactions were performed in a 96-well plate that was placed on ice under a 302 nm UV-light source (115V, 60Hz and 160 mA) at a distance of approximately 4 cm. Samples were exposed to UV-light for 10 minutes, mixed with 3X load buffer (150 mM Tris pH 6.8, 30% (v/v) glycerol, 6% (w/v) sodium dodecyl sulfate, 0.3% (w/v) bromophenol blue) and electrophoresed for 70 minutes at 130 V on a 15% denaturing polyacrylamide gel. Gels were then stained with ethidium bromide to allow detection of the RNA, followed by Coomassie staining to visualize protein.

143 RNase footphnting Native electrophoretic mobility shift assays were used to find 1:1 molar ratio FinO-RNA complexes prior to footprinting experiments. The binding reactions for the RNase footprinting experiments were performed in 14 |al_ reactions: 1.5 ^L 10X structure buffer (500 mM Tris-HCI pH 7.5,

32 1 M NaCI, 100 mM MgCI2), 1.5 (aL purified, annealed 5' or 3' P-SLII or SLIIc (in 10 mM Tris-HCI pH 7.5, 10 mM NaCI, 1 mM EDTA), 1.5 ^L protein sample (at 10X final concentration in 50 mM HEPES pH 7, 200 mM NaCI, 1 mM EDTA, 5 mM TCEP), 1.5 ^L 2 ng/^iL tRNA (Ambion),

and 8 \xL ddH20. Reactions were incubated on ice for 30 minutes and aliquots of 5 (o.L of the reaction was removed and added to 5 JJ.L 20% glycerol and loaded onto an 8% native gel to assay binding. 1 jal_ of RNase V1 or I (at the appropriate concentration; see below) was added to the remaining 9 |al_ of the binding reaction. RNase cleavage experiments were incubated at 4°C for 1 hour and stopped immediately with 120 jxL 0.3 M NaOAc pH 5.3 and 130 pL of phenol/chloroform/IAA. Samples were then chloroform extracted and ethanol precipitated. Pellets were resuspended in 4 (iL formamide gel load buffer and loaded on a 15% urea-denaturing sequencing gel. Gels were exposed overnight and quantified using ImageQuant software (GE Healthcare).

To determine the optimal amount of RNase V1 or I to add for the footprinting assays, we performed cleavage assays with end labeled SLII or SLIIc (at the same concentration as the footprinting experiments) where the concentration of RNase V1 or I was increased until low-affinity cleavages resulted. The reactions were incubated at the same temperature and length of time as the footprinting experiments. A final RNase V1 concentration of 0.001 U/JJ.L and RNase I concentration of 0.01 U/|JL gave specific cleavages with an adequate signal to noise. The alkaline hydrolysis ladders were performed in 10 |il_ reactions: 1 j^L end- labeled SLII or SLIIc, 1 nL 2 ^g/|aL tRNA, 8 ^L 1X alkaline hydrolysis buffer (50 mM sodium carbonate pH 9.2, 1 mM EDTA). Samples were incubated at 90°C for seven minutes after which they were ethanol precipitated and loaded on the denaturing gel. RNase T1 ladders were

144 performed in 9 pi reactions: 1 pi end-labeled SLII or SLIIc, 1 [iL 2 pg/pL tRNA, and 7 pL 1X sequencing buffer (50 mM sodium citrate, 7 M urea, and 1 mM EDTA titrated to pH 5). Reactions were incubated at 90°C for two minutes and then put on ice. 1 \\L 5 U/^L RNase T1 (Ambion) was added to the chilled reaction which was then incubated at 60°C for 20 minutes followed by phenol/chloroform extraction and ethanol precipitation before loading onto the denaturing gel. Assigning the RNase V1 cleavages can be problematic since the enzyme does not have sequence specificity and its products run differently on a denaturing gel when compared to RNase T1 (which cleaves after guanines in single-stranded sections) and alkaline hydrolysis ladders (2). The latter cleavage products leave a phosphate or cyclic phosphate 3' to the nucleotide cleaved whereas the RNase V1 products leave a 3'OH, which migrates slower on a gel (5). We got around this problem using two different methods. First, for the 5' 32P SLII RNase V1 cleavage assays, we chemically synthesized short RNA markers which had the same sequence as SLII and had a 3'OH. SLII markers were 10 (5'-GACAGUCGAU-3') and 15 (5'- GACAGUCGAUGCAGG-3') nucleotides in length (shown in Figure 4-3A, left). The oligmers were synthesized, purified and labeled in the same manner as SLII and SLIIc (see above). The other method to assign RNase V1 cleavages was to use T4 polynucleotide kinase, in the absence of ATP, to remove the 3' phosphate from the RNase T1 and alkaline hydrolysis products (1). Precipitated alkaline hydrolysis and T1 cleavage products were resuspended in 6 pL ddH20. 1 (j.L 10X polynucleotide kinase buffer, 1 ^L 10 U/pL T4 polynucleotide kinase and 2 |j.L Superaseln were added to the RNA sample. The kinase reactions were incubated for 10 minutes at 37°C after which they were phenol/chloroform extracted, ethanol precipitated and loaded on the denaturing gel. To quantify the footprinting data, we first normalized the total counts in each FinO-RNA complex lane to the total counts of the "No Protein" lane to account for lane loading discrepancies. Then for each band of interest, representing a position in the RNA, the fraction (f) of the 145 total counts for that band was calculated (fP]i) where p is a FinO-RNA complex, either wild-type, 33-186 W36A or 45-186, and i is the nucleotide position of the RNA. We also determined the fraction of the total counts for the "No Protein" sample (fnp,i) at that position. Finally, we divide fnPii by fp,i to get the protection factor which is defined as the magnitude by which the RNA was protected from RNase cleavage by each FinO derivative. The values in Figure 4-3B and 4-4B are an average of two independent experiments. We decided on a protection value of two or greater to represent a significant footprint. This is shown in the figures as a horizontal rule across the graph.

Periodate oxidation of SLII 3' end labeled SLII was dephosphorylated with T4 polynucleotide kinase to remove the terminal 3' phosphate using methods explained above. The resulting 32P-SLII RNA has an extra cytidine nucleotide at the 3' tail (Figure 4-8A). For each periodate reaction, 1 nL of dephosphorylated, annealed, 32P-SLII was added to 99 pL of 10 mM

Nal04 (dissolved in ddH20; Sigrna-Aldrich). The reactions were incubated at 4°C for 40 minutes in the dark followed by ethanol precipitation. The pellets were washed and resuspended in 7 ^L ddH20. The periodate-treated 32P-SLII resuspensions were then added to 1 \xL 10X structure buffer (see above), 1 (iL 2 mg/mL tRNA, and 1 \iL 10 jaM FinO (wild-type, 33-186 W36A, or 45-186) or 1X structure buffer for the no protein reaction. Parallel binding reactions were performed for untreated, annealed, dephosphorylated 32P-SLII RNA using 1 JIL RNA, 1 jil 10X structure buffer, 1 pL 2 mg/mL tRNA, 1 pL 10 ^M FinO (wild-type, 33-186 W36A, or 45-186) or 1X structure buffer for the no protein reaction, and 7 pL ddH20. All 10 |iL binding reactions were incubated at 4°C for 30 minutes before adding 10 (iL of 20% glycerol and loading onto an 8% native polyacrylamide gel, equilibrated with 1X tris-glycine pH 8.0 at 4°C. The gel was run for 2 hours at 130 V and the gels were exposed overnight.

146 A C C A 20 U G B GUG U&N C A U U C A G \A\ U C G-C A A u JG) C A A-U 20-G-C u /G/ C A G-C A-U u- Al U-A G-C G-C 30 C' G RBS C-G G-C 30 G-C c G\ C-G A-U A A G-C u \ly j A A C-G U U A-U A-U A-U G-C G-U 10 U-A U G C-G A C G-C A-U G-C G-C SLI A-U G-C SLII G-C SLIII SLIIc U-A A-u SLIc U-A A-U U-IAI U-A C-G G-C 10 G-C .„ AHUI A-U 1 U»G .„ 40 C-G/ G-c/° « „ c, C-G 1 C-G , C-0 5' GAUA GACA GAUUUU-3 5-G UUAAAAUUUGAAAUUGAAAAUCGC ACUGUC UAUC-3 « FinP 5'UTRoffraJmRNA

N, K37i f*1P K4° |K42 K46B-* r ^ i , cj* ! Ntt-..' 180

/ K125 .J. R121 J, '*": ' *A\ V' *

>"*• "*%? ^5142 Vi :-J'W

E147

Figure 4-1: Overview of constructs used in the study

(A), (B) The secondary structures of FinP and traJ mRNA respectively. The ribosomal binding site and start codon of traJ mRNA are boxed. The derivative RNAs of focus, SLII and SLIIc, are shown in bold and numbered accordingly. (C) Boxed: The chemical structure of 4-azidophenacyl bromide (APA-Br), the cross-linking agent used in the FinO- SLII cross-linking experiments. Cartoon representation of the front (left) and rear (right) faces of FinO 26-186. Blue spheres indicate the residues which are labeled with APA-Br for the cross-linking experiment. The N- and C-termini are indicated. (D) The crystal structure of FinO 26-186 highlighting the protein constructs used in the experiments. The W36A mutation is shown as a sphere on the structure. The N-terminal 32 amino acids could not be seen in the structure and are drawn in (dashed section). Below is a scaled linear representation of the primary structure of FinO showing where each construct begins. The shades of grey correspond to the crystal structure colouring. (E) A representative 8% native PAGE demonstrating each of the FinO constructs in a 1:1 molar complex with 5' 32P-SLII.

147 A

FinO-FinO } Crosstinking «-FinO/SLtl *—Free FtnO

B

K;- OQ;C40 v •-.' Or> §€4-',£)^

f V S * A S Si - I, \

if

, ,#' /-'•' JCM170 • Sgap*™ • G '• O.;; 135 .••'-J'] •. r

0>- - • <• • '••. * .. i- •.: .t

(Figure 4-2 legend on p. 149)

148 Figure 4-2: Site specific cross-linking of FinO and SLII (p. 148)

(A) Cross-linked protein/RNA complexes were analyzed by SDS-PAGE in which RNA containing species were visualized by ethidium bromide staining (left), while protein- containing species were detected by Coomassie blue staining (right). Indicated on the bottom of each gel pair are the various FinO cysteine mutants used. Indicated to the left of the top ethidium bromide stained gel are the positions of free SLII, free tRNA, the FinO/SLII cross-linked product containing one protein bound to one SLII (resulting from specific interactions), and the FinO/tRNA cross-linked product (resulting from non-specific interactions). To the right of the top Coomassie stained gel are the positions of free FinO, the FinO/SLII cross-linked product, and non-specific cross-linking products between FinO molecules. Indicated by arrows on the bottom two gel pairs is the position of the FinO/SLII cross-linked product. 3C -> S indicates a protein in which all the native cysteine residues have been replaced with serine. (B) Electrostatic surface representations of the front (left) and rear (right) faces of FinO (26-186) Magenta circles indicate the sites of APA attachment that showed significant cross-linking to SLII RNA, while yellow circles indicate sites that do not cross-link to RNA.

149 5' P-SLII 3' P-SLIi Nucleotide Position

Figure 4-3: RNase V1 cleavage of 5' and 3' end labeled SLII in the absence and presence of various FinO constructs

(A) 15% urea-denaturing polyacrylamide gels showing the products of RNase V1 (0.001 U/nL final concentration) cleavage reactions. M10 and M15 are synthesized SLII RNA markers of 10 and 15 nucleotides in length (see Materials and Methods). The lanes OH and T1 represent the alkaline hydrolysis and RNase T1 cleavage of denatured SLII respectively. SLII nucleotide positions are indicated at the right of the gels. Large and small arrowheads indicate major or minor cleavages by RNase V1 in the presence of FinO. Vertical brackets represent significant footprints on SLII RNA resulting from FinO protection of SLII from RNase V1 attack. (B) Bar graphs showing the quantification of the footprint areas of the gels in (A). The left axis shows the degree of FinO protection from RNase V1 relative to the "No protein" reaction. In black is FinO 1-186 WT, white is FinO 33-186 W36A, and in grey is FinO 45-186. Data above the horizontal rule in each graph represent significant protection (£ 2-fold) by FinO. The black bars below the x-axis highlight the footprint.

150 B 5' P-SLIIc

£ J CL O TI 2 o Jt C30 C31 C32 U33 G34 C35 A36 U37 IC3S G39

L.20 z E p **#* «*•• I

G34 C35 A36 U37 C38 G39 A40

5' P-SLIIc 3' P-SLIIc Nucleotide Position

Figure 4-4: RNase V1 cleavage of 5' and 3' end labeled SLIIc in the absence and presence of various FinO constructs

(A) 15% urea-denaturing polyacylamide gels showing the products of RNase V1 (0.01 U) cleavage reactions. The lanes OH and T1 represent the alkaline hydrolysis and RNase T1 cleavage of denatured SLIIc respectively. SLIIc nucleotide positions are indicated at the right of the gels. Large and small arrowheads indicate major or minor cleavages by RNase V1 in the presence of FinO. Vertical brackets represent significant footprints on SLIIc RNA resulting from FinO protection of SLIIc from RNase V1 attack. (B) Bar graphs showing the quantification of the footprint areas of the gels in (A). The left axis shows the degree of FinO protection from RNase V1 relative to the "No protein" reaction. In black is FinO 1-186 WT, white is FinO 33-186 W36A, and in grey is FinO 45-186. Data above the horizontal rule in each graph represent significant protection (> 2-fold) by FinO. The black bars below the x-axis highlight the footprint.

151 5' P-SLII 3' P-SLII 5' P-SLIIc 3' P-SLIIc

Figure 4-5: Limited RNase I digestion of 5' and 3' end labeled SLII and SLIIc in the absence and presence of various FinO constructs

Products of the RNase I (0.01 U/|xL final concentration) cleavage reactions were resolved on 15% urea-denaturing polyacrylamide gels. The radiolabeled RNA construct is noted below each of the gels. The lanes OH and T1 represent the alkaline hydrolysis and RNase T1 cleavage of denatured SLII or SLIIc respectively. The RNA nucleotide positions are indicated at the left of the gels. Large and small arrowheads indicate major or minor cleavages by RNase I in the presence of FinO.

152 A B

(MM Final FinO 1-186 WT) 0 .25 .5 1 2.5 5 FinO-SLII 0) 3" 32P-SLII SLII

(ii)

|G40

5' P-SLIIc

. (iii) D

W& •» m *•» ««•• m •» m IG18 •m «m _ w* «* m «# m |G16 •**

5' 32P-SLII

3' P-SLIIc

(Figure 4-6 legend on p. 154)

153 Figure 4-6: RNase I overdigestion of 5' and 3' end labeled SLII and SLIIc in absence and presence of various FinO constructs (p. 153)

In all experiments, the RNAs were digested with RNase I at a final concentration of 0.1 U/nL. (Ai) 8% native EMSA showing binding reactions of 5' 32P-SLII with increasing amounts of FinO 1-186 WT before the addition of RNase I. The final concentration of FinO WT in |iM in each reaction is indicated on top of the gel. (Aii and Aiii) Slices from a 15% urea-denaturing gel showing the products of the RNase I digest of 5' 32P-SLII at the 3' end (Aii) and loop region (Aiii) in the presence of increasing amounts of FinO 1-186 WT. The lanes correspond to the binding reactions in (Ai). The lanes OH and T1 represent the alkaline hydrolysis and RNase T1 cleavage of denatured 5' 32P-SLII respectively. SLII nucleotide positions are indicated at the right of the gel. (B) Slice from a 15% urea- denaturing gel showing the products of the RNase I digest of 3' 32P-SLII at the 5' end in the absence and presence of various FinO constructs. 1:1 molar ratio FinO-3' 32P-SLII complexes were formed prior to exposure to RNase I. Nucleotide position G5 is indicated at the right of the gel. (C) Slice from a 15% urea-denaturing gel showing the products of the RNase I digest of 5' P-SLIIc at the 3' end in the absence and presence of various FinO constructs. 1:1 molar ratio FinO-5' 32P-SLIIc complexes were formed prior to exposure to RNase I. Nucleotide position G39 is indicated next to the gel. (D) Slices from a 15% urea-denaturing gel showing the products of the RNase I digest of 3' 32P-SLIIc at the 5' end (Di) and loop region (Dii) in the absence and presence of various FinO constructs. 1:1 molar ratio FinO-3' 2P-SLIIc complexes were formed prior to exposure to RNase I. SLIIc nucleotide positions are indicated next to the gel.

154 <

•- *• Strong and weak RNase V1 cleavage in presence of FinO [>- E>- Strong and weak RNase I cleavage in presence of FinO

GC Protection from RNase V1 cleavage upon FinO binding

; G CI Protection from RNase I cleavage upon FinO binding

V VV Enhancement of RNase I cleavage upon FinO binding

Figure 4-7: Summary of RNase V1 and I cleavage reactions of SLII and SLIIc

Secondary structures of SLII and SLIIc showing the results from the RNase cleavage reactions. Large and small black arrowheads denote stong and weak RNase V1 cleavages, respectively in the presence of the FinO constructs. Boxes indicate footprints where FinO protected the RNAs from RNase V1 cleavage. Large and small open arrowheads denote strong and weak RNase I cleavages, respectively in the presence of FinO. Dashed boxes indicate areas of protection from RNase I cleavage by FinO. Areas where a "V" resides indicates enhanced cleavage by RNase I in the presence of the FinO constructs.

155 A A C C U G U U G-C A-U G-C G-C G-C A-U C-G G-C U-A A-U G-C C-G U"G G-C 32 -GACA GAUUUUpCp- SLIIWT

B FinO 33-186 W36A 45-186

4 FinO-SLII

32 < iP-SLI I

# / f O <§> & & & *°

4 FinO-SLII

: ^ 32 lKUP»«»* P-SLII Untreated Treated (2', 3' c/s-diol) (2',3C dialdehyde)

Figure 4-8: FinO binding to SLII requires a terminal 3' OH on the 3' tail of SLII

A) A secondary structure schematic of SLII RNA after the addition of 3',5'-cytidine [5' 32Pr ] diphosphate (pCp) (in bold red). A terminal 3' phosphate results from the modification. B) 8% native gel showing results of binding reactions between FinO constructs (noted above gel) and P-SLII which has a terminal 3' phosphate on the 3' tail. NP - No protein. Triangles represent increase in final concentration of FinO: 0.25, 0.5, 1, 2.5, 5, and 10 u.M. The positions of free 32P-SLII and the FinO-32P-SLII are noted by arrows. C) 8% native gels showing results of binding reactions between FinO constructs (noted above gels) and 32P-SLII which has been dephosphorylated (untreated; left) or dephosphorylated and treated with 10 mM Nal04 (treated; right), The concentration of the FinO constructs are at 1 \iM final. The positions of free 32P-SLII and the FinO-32P-SLII are noted by arrows. 156 References

1. Brown, T.S. and Bevilacqua, P.C. (2005) Method for assigning double-stranded RNA structures. Biotechniques, 38: 368-372.

2. Donis-Keller, H., Maxam, A.M., and Gilbert, W. (1977) Mapping adenines, guanines, and pyrimidines in RNA. Nucleic Acids Res, 4: 2527-2538.

3. Du, Z., Giedroc, DP., and Hoffman, D.W. (1996) Structure of the autoregulatory pseudoknot within the gene 32 messenger RNA of bacteriophages T2 and T6: a model for a possible family of structurally related RNA pseudoknots. Biochemistry, 35: 4187- 4198.

4. Duckett, D.R., Murchie, A.I., Clegg, R.M., Bassi, G.S., Giraud- Panis, M.J., and Lilley, D.M. (1997) Nucleic acid structure and recognition. Biophys Chem, 68: 53-62.

5. Ehresmann, C, Baudin, F., Mougel, M., Romby, P., Ebel, J.P., and Ehresmann, B. (1987) Probing the structure of RNAs in solution. Nucleic Acids Res, 15: 9109-9128.

6. Finlay, B.B., Frost, L.S., Paranchych, W„ and Willetts, N.S. (1986) Nucleotide sequences of five IncF plasmid finP alleles. J Bacteriol, 167: 754-757.

7. Finnegan, D. and Willetts, N. (1972) The nature of the transfer inhibitor of several F-like plasmids. Mol Gen Genet, 119: 57-66.

8. Freier, S.M., Alkema, D., Sinclair, A., Neilson, T., and Turner, D.H. (1985) Contributions of dangling end stacking and terminal base- pair formation to the stabilities of XGGCCp, XCCGGp, XGGCCYp, and XCCGGYp helixes. Biochemistry, 24: 4533-4539.

9. Frost, L, Lee, S., Yanchar, N., and Paranchych, W. (1989) finP and fisO mutations in FinP anti-sense RNA suggest a model for FinOP action in the repression of bacterial conjugation by the Viae plasmid JCFLO. Mol Gen Genet, 218: 152-160.

10. Ghetu, A.F., Arthur, D.C., Kerppola, T.K., and Glover, J.N. (2002) Probing FinO-FinP RNA interactions by site-directed protein-RNA crosslinking and gelFRET. RNA, 8: 816-823.

11. Ghetu, A.F., Gubbins, M.J., Frost, L.S., and Glover, J.N. (2000) Crystal structure of the bacterial conjugation repressor FinO. Nat Struct Biol, 7: 565-569.

157 12. Ghetu, A.F., Gubbins, M.J., Oikawa, K., Kay, CM., Frost, L.S., and Glover, J.N. (1999) The FinO repressor of bacterial conjugation contains two RNA binding regions. Biochemistry, 38: 14036-14044.

13. Hall, K.B. and Fox, R.O. (1999) Directed cleavage of RNA with protein-tethered EDTA-Fe. Methods, 18: 78-84.

14. Ho, S.N., Hunt, H.D., Horton, R.M., Pullen, J.K., and Pease, L.R. (1989) Site-directed mutagenesis by overlap extension using the polymerase chain reaction. Gene, 77: 51-59.

15. Holland, J.A., Hansen, M.R., Du, Z., and Hoffman, D.W. (1999) An examination of coaxial stacking of helical stems in a pseudoknot motif: the gene 32 messenger RNA pseudoknot of bacteriophage T2. RNA, 5: 257-271.

16. Jerome, L.J. and Frost, L.S. (1999) In vitro analysis of the interaction between the FinO protein and FinP antisense RNA of F-like conjugative plasmids. J Biol Chem, 274: 10356-10362.

17. Jerome, L.J., van Biesen, T., and Frost, L.S. (1999) Degradation of FinP antisense RNA from F-like plasmids: the RNA-binding protein, FinO, protects FinP from ribonuclease E. J Mol Biol, 285: 1457-1473.

18. Kurata, S., Ohtsuki, T., Suzuki, T., and Watanabe, T. (2003) Quick two-step RNA ligation employing periodate oxidation. Nucleic Acids Res, 31: e145.

19. Laederach, A., Chan, J.M., Schwartzman, A., Willgohs, E., and Altman, R.B. (2007) Coplanar and coaxial orientations of RNA bases and helices. RNA, 13: 643-650.

20. Lee, S.H., Frost, L.S., and Paranchych, W. (1992) FinOP repression of the F plasmid involves extension of the half-life of FinP antisense RNA by FinO. Mol Gen Genet, 235: 131-139.

21. Limmer, S., Hofmann, H.P., Ott, G., and Sprinzl, M. (1993) The 3'- terminal end (NCCA) of tRNA determines the structure and stability of the aminoacyl acceptor stem. Proc Natl Acad Sci U S A, 90: 6199-6202.

22. Liu, J.D., Zhao, L., and Xia, T. (2008) The dynamic structural basis of differential enhancement of conformational stability by 5'- and 3'-dangling ends in RNA. Biochemistry, 47: 5962-5975.

23. Lowman, H.B. and Draper, D.E. (1986) On the recognition of helical RNA by cobra venom V1 nuclease. J Biol Chem, 261: 5396-5403.

158 24. Meador, J., 3rd, Cannon, B., Cannistraro, V.J., and Kennell, D. (1990) Purification and characterization of Escherichia coli RNase I. Comparisons with RNase M. Eur J Biochem, 187: 549-553.

25. Mullineaux, P., and Willetts, N.S., Promoters in the transfer region of plasmid F, in Plasmids in bacteria, D.R. Helinski, Cohen, S.N., Clewell, D.B., Jackson, D.A., and Hollaender, A. 1985, Plenum Press: New York. p. 605-614.

26. OToole, A.S., Miller, S., and Serra, M.J. (2005) Stability of 3' double nucleotide overhangs that model the 3' ends of siRNA. RNA, 11: 512-516.

27. Ohmichi, T., Nakano, S., Miyoshi, D., and Sugimoto, N. (2002) Long RNA dangling end has large energetic contribution to duplex stability. J Am Chem Soc, 124: 10367-10372.

28. Saenger, W., Principles of nucleic acid structure. Springer advanced texts in chemistry, C.R. Cantor. 1984, New York: Springer-Verlag.

29. Sandercock, J.R. and Frost, L.S. (1998) Analysis of the major domains of the F fertility inhibition protein, FinO. Mol Gen Genet, 259: 622-629.

30. Scaringe, S.A. (2001) RNA oligonucleotide synthesis via 5'-silyl-2'- orthoester chemistry. Methods, 23: 206-217.

31. Sugimoto, N., Kierzek, R., and Turner, D.H. (1987) Sequence dependence for the energetics of dangling ends and terminal base pairs in ribonucleic acid. Biochemistry, 26: 4554-4558.

32. Thompson, R. and Taylor, L (1982) Promoter mapping and DNA sequencing of the F plasmid transfer genes traM and traJ. Mol Gen Genet, 188: 513-518.

33. Timmis, K.N., Andres, I., and Achtman, M. (1978) Fertility repression of F-like conjugative plasmids: physical mapping of the R6-5 finO and finP cistrons and identification of the FinO protein. Proc Natl Acad Sci USA, 75: 5836-5840.

34. van Biesen, T. and Frost, L.S. (1994) The FinO protein of IncF plasmids binds FinP antisense RNA and its target, traJ mRNA, and promotes duplex formation. Mol Microbiol, 14: 427-436.

35. van Biesen, T., Soderbom, F., Wagner, E.G., and Frost, L.S. (1993) Structural and functional analyses of the FinP antisense RNA regulatory system of the F conjugative plasmid. Mol Microbiol, 10: 35-43.

159 36. Will, W.R. and Frost, L.S. (2006) Hfq is a regulator of F-plasmid TraJ and TraM synthesis in Escherichia coli. J Bacteriol, 188: 124-131.

160 Chapter 5

General discussion

Overall summary

This work has shown that FinO is the essential component of the FinOP inhibition system of F-like conjugation. FinO acts as an RNA chaperone, performing strand-exchange and intermolecular annealing activities on its target RNAs. These actions are necessary for destabilization of the extremely stable SLII and SLIIc hairpins of FinP and traJ mRNA respectively, which would otherwise act as a kinetic barrier to intermolecular duplex formation. The catalytic area of FinO is located between residues 32-42 on the N-terminal a-helix. This region consists primarily of positively charged lysine residues centered around a solvent exposed tryptophan. Mutation of W36 to alanine negatively affects FinO- mediated in vitro strand-exchange and intermolecular duplexing. Inhibition of conjugation is also disrupted by the mutant, strongly suggesting that strand-exchange and annealing activities occur in vivo. Other mutations in this region such as P34A, K35A, K40A and K42A disrupt strand-exchange significantly. The critical lysines in this area may be important for contacting the duplex RNA at the site of destabilization. A truncation mutant which removes the N-terminal 44 amino acids (FinO 45-186) completely disrupts strand-exchange, duplexing, and inhibition of conjugation. Interestingly, the construct has been shown to bind to SLII 20-fold tighter than wild-type FinO. Similarly, deletion to FinO 26-186 causes a 10-fold disruption in intermolecular duplexing and a significant 4-fold increase in binding affinity. In addition, mutation of W36A leads to a 5-fold increase in binding affinity. This inverse relationship between the ATP-independent chaperone activity and RNA binding suggests that FinO may use a portion of its binding energy to perform its strand-exchange and duplexing activities. It was previously determined that FinO binds to its RNA targets with high specificity in a non-sequence specific manner (6, 9, 17, 20). 161 Site-specific cross-linking experiments showed that FinO interacts with SLII RNA using positively charged residues located throughout the protein. The strongest contacts occur at a large positively charged patch on the front side of globular region of the protein encompassing residues R121, K125, and R165. Other areas generated weaker cross-links, including residues in the N-and C-terminus and along the back of the globular domain of FinO. We performed RNase footprinting assays to determine the areas of SLII and SLIIc which contact FinO. FinO primarily contacts the 3' portion of the lower stem and 3' tail of the RNAs. In contrast to previous studies (9), we found that the 5' tail was not important for FinO binding. In fact, the 5' tail of SLIIc was degraded by RNase I to a greater degree in the presence of FinO. Consistent with previous studies, we found that the loops of SLII and SLIIc are not important for FinO binding. All FinO constructs (wild-type, 33-186 W36A, and 45-186) bound to SLII and SLIIc similarly indicating that the N-terminus does not play a primary role in binding to the hairpins. However, it was shown that the C- terminal truncation fragment, FinO 1-61, exhibits a small amount of chaperone activity even though the rest of FinO is required to bind efficiently. This indicates that the N-terminal a-helix is likely using low- affinity interactions with SLII or SLIIc leading to strand-exchange and annealing activities, while high affinity interactions are reserved for residues closer to the globular domain. We employed various RNA and protein constructs for FinO-SLII complex crystallization trials. The best crystals resulted from a complex consisting of a nine base-pair SLII tetraloop (SLII guga-5) with FinO 26- 186 W36A. However, the crystals only diffracted to 4.5 A, well short of the resolution needed to solve the complex structure. Attempts were made to improve the crystals by removing N-terminal residues 26-32 of the FinO W36A mutant. This truncation removes a flexible stretch of residues which lack density in the FinO 26-186 crystal structure (5). On the RNA side, we screened different 5'-GNRA-3' SLII tetraloop sequences and employed Sll duplexes with blunt and overhanging ends. However, none of the complexes produced crystals diffracting close to 4.5 A. Attempts using the U1A RNA binding domain crystallization module to

162 increase the quality of the crystals also failed but remains a promising approach as the technique was not fully explored.

Orientation of FinO on FinP SLII

While we identified areas of SLII and SLIIc which are contacted by FinO, and highlighted which areas of FinO contact SLII, we are lacking a complete picture as to the orientation of FinO on its target RNAs. However, we can use the data from this work and previous studies to build a model of how FinO may bind SLII. A qualitative model of FinO 45-186 in complex with a model SLII RNA is shown in Figure 5-1. The stem-loop consists of a 14 base pair stem, seven nucleotide loop, and single stranded four nucleotide 5' and six nucleotide 3' tails. RNase footprinting experiments showed that FinO appears to bind SLII and SLIIc in a similar manner for wild-type FinO, FinO 33-186 W36A, and FinO 45-186. This suggests that the N-terminus of FinO does not contact the RNA with high affinity and may be oriented away into solution. Limited proteolysis experiments at 25°C have demonstrated that the N-terminal a1 helix of wild-type FinO is digested to residue 50 in the presence of SLII compared to residue 62 (start of globular domain) in the absence of RNA (6). This would suggest that residues 1-49 of the N-terminus are unstructured and do not play a primary role in binding SLII. In Figure 5-1, the N-terminal a-helix projects away from the RNA prior to residue Q62. The largest RNA binding interface would reside in a highly positively charged region of the globular domain consisting of helices a4 and a6. Cross-linking experiments have identified four residues (labeled in boxes in Figure 5-1) which interact strongly with SLII: K118 (a4; data not shown in Chapter 4), R121 (oc4), K125 (a4), and R165 (a6). Although not tested, four other residues lie in this area and may interact with SLII: R122 (cc4), R129 (a4), K168 (a6), and R171 (a6) (Figure 5-1). RNase footprinting experiments demonstrated that the lower 3' portion of the stem and 3' tail of SLII are contacted by FinO. Strong protection from RNase cleavage was seen for nucleotides U35-C39 at the 3' stem and 163 nucleotides G40-U45 of the 3' tail of SLII (Figure 5-1; large asterisks). Weaker FinO protection was seen at nucleotides A34 of the 3' stem and C7 and G8 of the 5' stem (Figure 5-1; small asterisks). Accordingly, the lower 3' portion of the RNA stem has been modeled into the positively charged region of the globular domain of FinO. Positioning the rest of the RNA at an angle to FinO would place nucleotides 7 and 8 near Q62 which projects towards the RNA and may form a hydrogen bond with the RNA. Q59 also projects towards the RNA and may interact alongside Q62 (Figure 5-1). The 3' tail of SLII has been shown in this work and previous studies to be required for specific FinO interactions (9). Accordingly, we modeled the phosphate backbone of the tail projecting across a positively charged patch on one side of FinO, consisting of residues K118, K119, and R121 of a4. The 3' tail heads toward K81 (a2) at the rear face of FinO which forms a cross-link with SLII (Figure 5-1). The path of the 3' tail is constrained to this side of FinO as residues on the other side of a6 (R170 of a6, C135 of a5, and C142 of (33) did not form cross-links with SLII (see Figure 4-2B). We showed that FinO did not protect the 5' tail of SLII from RNase I attack. Previous studies have shown that the 5' tail is not as important for FinO binding as the 3' tail since removal of the tail only reduces the Ka of the interaction 1.3-fold compared with a 5.5-fold reduction after removal of the 3' tail (9). The lack of protection may be due to the short length of the tail and its proximity to the stem, thus preventing digestion or because FinO does not interact with it. Thus we have modeled the tail so that it projects away from the protein (Figure 5-1). Two questions remain regarding the model. First, the cross- linking study showed that K176 at the C-terminal end of a6 interacts with SLII. The residue lies on the rear face of FinO away from the RNA, however, its side chain points towards a hole in the globular domain (Figure 5-1). Conformational flexibility of a6 may allow it to interact with SLII since crystallographic B-factors for this area are high relative to the globular region (5). Finally, the solvent exposed hydrophobic residue I66, part of a 310 helix C-terminal to a1 (Figure 5-1), may lie close to a minor 164 groove of the RNA, potentially forming a hydrophobic interaction with bases in this area thereby stabilizing the interaction with the RNA (2). This residue is peculiar in that it faces into the solvent away from a6. However, it may sandwich against the neighboring Y67 lying in proximity to a6 residue 1175, thereby constituting a three way structural stabilization of FinO. More experiments need to be performed to determine if it has any importance for RNA binding.

Implications for FinP and traJ mRNA duplexing

In this work, we have attempted to decipher the mode by which FinO interacts with its minimal binding targets and how FinO is able to remodel the complementary RNAs leading to formation of an intermolecular duplex. While this model in vitro system has been important for characterizing the function of FinO, the ultimate goal is to characterize the mechanism of intermolecular duplexing between the parent RNAs, FinP and the 5' UTR of traJ mRNA, which leads to the inhibition of conjugation in vivo. A detailed model outlining how FinO may mediate FinP-fraJ mRNA intermolecular duplexing in vivo is presented in Figure 5-2. During the high frequency transfer period of conjugation, the TraJ transcription factor is expressed from the traJ gene. The 30 S subunit of the ribosome is capable of binding the 5' UTR of traJ mRNA unabated, leading to translation of the protein (Figure 5-2i). FinP antisense RNA is also transcribed from the opposite strand of the traJ, however, due to degradation by RNase E, its levels are too low to inhibit conjugation. Eventually, FinP is stabilized by increased FinO levels as high affinity binding of the protein to the SLII domain protects the RNA from cellular degradation (Figure 5-2N). FinO likely binds to the 5' UTR of traJ mRNA at the SLIIc domain since the RNA contains the same structural elements as FinP SLII and has been shown to bind to FinO in a similar manner. As hypothesized above, FinO would bind to SLII and SLIIc at the 3' lower portion of the RNA stem and 3' tail. The N-terminus of FinO would be

165 oriented into the solution and the SLI and SLIc domains would be free for pairing (Figure 5-2iii). Many antisense RNAs use loop-loop interactions to recognize their target RNAs (1). Both loops of FinP contain the 5'-YUNR-3' U-turn motif (Y=pyrimidine; N=any base, R=purine) which is common to plasmid- encoded antisense RNAs and has been shown to be important for initial loop-loop interactions between the antisense and target RNAs (4). However, unlike other antisense systems, it appears that the U-turn sequences are not critical for FinP-fraJ mRNA duplexing due to the presence of FinO. We were not able to identify a stable loop-loop interaction using native gel-electrophoresis for either SLII/SLIIc (unpublished results) or SLI/SLIc (7) pairings when RNAs were made where the loops were complementary but the stem and 5' and 3' tails were not. This shows that loop-loop interactions, if they exist, are transient in nature and further intermolecular base-pairing is required for visualization by gel electrophoresis. Also, FinO can overcome FinP SLI and SLI I loop mutations which, in the absence of the protein, disrupt mating inhibition in vivo (13). Likewise, FinO can compensate for 3-4 nucleotide loop mutations in SLI of FinP which disrupt the ability of SLI to duplex with wild-type SLIc in vitro and disrupt conjugation repression in vivo (7). This highlights the importance of FinO in the FmP-traJ mRNA recognition step of duplexing. FinO may be needed to bring the RNAs into closer proximity for contact, perhaps through transient loop-loop interactions (7). We show that the N-terminal 25 amino acids are important for intermolecular duplexing as removal of the residues led to a 10-fold decrease in the apparent second-order duplexing rate for both SLII-SLIIc and FinP-fraJ mRNA pairings. In contrast, this region was not terribly important for Sll strand-exchange, as removal of the residues led to a moderate 1.5-fold decrease in the first-order strand exchange rate. Therefore, residues 1-25 (in red; Figure 5-2iii) may be involved in the antisense-target RNA recognition step. A large fraction of this unstructured region consists of positively charged residues which may be important for neutralizing the phosphate backbones of the two complementary RNAs allowing initial intermolecular base pairs to form.

166 Previous studies have shown that FinO binds to SLII as a monomer (6). However, the intermolecular duplexing activity of FinO is dependent on the protein concentration, indicating that FinO may use protein-protein interactions to bring a FinO-FinP complex into close contact with a FmO-traJ mRNA complex (double arrow, Figure 5-2iii). Although untested, the C-terminal a-helix (a6) of FinO may be a good candidate for protein-protein interactions. Cross-linking studies have shown that R165 of a.6, whose side chain faces the front face of FinO, is important for RNA binding. However, R170, whose side chain faces to the back side of FinO, and nearby C135 (a5) do not interact with SLII. Using this knowledge and the structural model for the FinO-SLII complex (Figure 5-1), a qualitative model was made which shows how dimerization of FinO may help position SLII and SLIIc for loop-loop interactions. Figure 5-3A shows multiple views of this recognition complex where a6 from one FinO (blue) interacts with a6 from another FinO molecule (gray). The second FinO is rotated 180° around its long axis relative to the first. The proposed interaction can be stabilized by a hydrophobic interaction between V160 of both FinO molecules and through two salt-bridges involving residues K168 and D167 from both FinOs (Figure 5-3B). The orientation of the two complexes would allow for loop-loop interactions between SLII and SLIIc as the N-terminal a1 helix of both FinO molecules point away from the RNAs. Therefore, the FinO a6 dimerization would enhance F\r\P-traJ mRNA interactions by bringing the SLII and SLIIc domains into close contact for initial base pairing (Figure 5-2iv). In addition, since SLI of FinP is 5' to SLII, it would project away from the protein while SLIc of traJ mRNA, which is 3' of SLIIc, would lie on the same side as SLI in the model (Figure 5-2iii and iv). However, the importance of protein-protein interactions needs to be assessed as the stoichiometry of the duplexing complex remains undetermined. After an initial F\nP-traJ mRNA interaction has occurred, the next step involves progression of the intermolecular duplex. The duplexes of SLII and SLIIc have continuous A-form helices which need to be destabilized for intermolecular duplexing. In the absence of protein, SLII is not able to form an intermolecular duplex with SLIIc within the reaction 167 time, however, in its presence a stable duplex is able to form in approximately an hour, indicating that FinO is able to destabilize SLII and SLIIc, overcoming the kinetic barrier to duplex formation. Indeed, it was shown that FinO is able to destabilize stable SLII-like duplexes leading to strand-exchange events. Figure 5-2iv shows that residues 32-42 (red) of the N-terminus of FinO are responsible for destabilizing the stems of SLII and SLIIc so the unwound strands can pair with their complement. The critical residue responsible for the catalytic activity of FinO is tryptophan 36 (Figure 5-2iv; enlargement). When it was mutated to alanine, the in vitro strand-exchange activity decreased five-fold while the intermolecular SLII-SLIIc duplexing activity decreased approximately 1.5-fold. Furthermore, conjugation repression was reduced seven-fold when FinO 1-186 W36A was expressed in vivo compared to wild-type FinO. It is hypothesized that the aromatic ring of W36 intercalates with the RNA duplex, disrupting base pairing within the stem in a similar fashion to RNA helicases (11, 14, 21, 22). Association of the residue with the RNA would be enhanced by the lysine residues which flank W36 (Figure 5-2iv). In contrast to SLII and SLIIc, SLI from FinP and SLIc from traJ mRNA are less stable hairpins containing fewer base pairs due to mismatches within the duplex regions (positions A12/A27 for SLI and A83/C102 and U85/U100 for SLIc)(Figure 4-1 A, B). Unpaired nucleotides within duplex regions of antisense RNAs such as CopA from plasmid R1 have been shown to be important for promoting intermolecular RNA-RNA pairings (8, 12). Gubbins ef al. showed that removing the upper stem mismatches in SLI (A27U) and SLIc (U85A) led to a significant decrease in pairing rates in the presence of FinO compared to wild-type SLI and SLIc suggesting that the mismatches are important for stable duplex formation (7). It was also shown in the study that duplex progression probably starts from a loop-loop interaction. This would agree with the model as the loops of SLI and SLIc would be close enough to interact (Figure 5-2iv). Therefore, due to the lower stability of the hairpins, FinO may play a lesser role in the duplexing of SLI and SLIc. However, due to flexibility in the N- terminal a-helix of FinO, it may be possible that the chaperone region aids in duplex progression of this area.

168 FinO is also required for intermolecular annealing of the complementary RNAs. The N-terminal 25 amino acids of FinO would likely play the primary role in RNA annealing (Figure 5-2v) as non-specific interactions with the separated strands need to be made to ensure they do not revert into hairpin structures. RNA binding would also be required to prevent repulsion of the phosphate backbones of the juxtaposed RNA strands, ensuring proper intermolecular base pairing ensues. It was demonstrated in this work that FinO is not processive during strand- exchange, as the addition of four base pairs to the top of the SI I duplex led to a reduction in the rate of strand-exchange. Therefore it is possible that globular domain of FinO remains associated with the lower 3' portion of SLII and SLIIc while the N-terminus repeatedly binds non-specifically and weakly to areas of the two domains (and potentially SLI and SLIc) which need to be destabilized and annealed to complementary sequences. At a certain point, duplex progression would encroach on the FinO-binding elements of SLII and SLIIc preventing high affinity binding of FinO. FinO would then dissociate either continuing to form non-specific contacts with the RNAs which lead to destabilization/annealing events (Figure 5-2vi) or allowing the intermolecular duplex to progress by itself until completion (Figure 5-2vii). It has been shown in this work that FinO does not need the 5' and 3' single stranded tail regions of SI I to mediate strand-exchange. However, it is likely that the tails would be needed for the F\nP-traJ mRNA recognition event during intermolecular duplexing. We did not test whether FinO-mediated duplexing is affected by removal of the single stranded tails of SLII and SLIIc. However, Gubbins ef al. demonstrated that removal of the tail regions from SLI and SLIc led to a five-fold decrease in the FinO-mediated rate of duplex formation suggesting that the tails are an important part of the process (7). Whether this is due to a reduction in FinO binding or because the single stranded tails are required for efficient duplex progression remains undetermined.

169 The N-terminus of FinO and disorder as a possible mechanism for its chaperoning function

Multiple lines of evidence have shown that the N-terminus of FinO is highly disordered. For instance, FinO 26-186 was crystallized at 4°C, however residues 26-32 could not be visualized in the final structure due to flexibility (5). Crystallographic B-factors for the structure were highest throughout the entire N-terminal a-helix indicating flexibility. FinO performs its RNA chaperone activites at 37°C so it is likely that the N- terminus is much more flexible than at 4°C. Indeed, limited proteolysis of wild-type FinO at 25°C showed that FinO is cleaved to 62-170 and 62-174 in the absence of SLII, while it is cleaved to 50-186 in its presence (6). FinO 1-61 can bind to SLII but the affinity for the interaction is decreased more than two orders of magnitude relative to FinO 1-186. Interestingly, the fragment can perform strand-annealing activities, albeit at a lower rate than full-length FinO. FinO 45-186, which is devoid of RNA chaperone activity, bound to SLII with the greatest affinity indicating that there is an inverse relationship between FinO binding to its RNA targets and its chaperone activity. Indeed, an RNA chaperone would not be expected to bind too strongly to its targets as it would need to dissociate to facilitate subsequent cycles. Tompa and Csermely showed that a great majority of RNA and protein chaperones have intrinsic structural disorder (19). In the study, the region of chaperone activity almost invariantly could be traced to a disordered region and deletion of this region led to the abolishment of chaperoning activity, as is the case for FinO 45-186. They went on to suggest that disordered RNA chaperones have versatility in recognition of their RNAs so that they can bind a wide variety of misfolded RNAs and help them to refold into their functional structure. For FinO, while recognition of its target RNAs appears to be specific for the globular core region, the unstructured N-terminus could transiently interact with areas on FinP and traJ mRNA where the chaperoning activity of FinO is needed for intermolecular strand-exchange and annealing processes (Figure 5-2 iv-vi). Indeed, it was shown that N-terminal residues 37, 40, 42, and 46

170 are able to cross-link to an unknown area of SLII. FinO would therefore sacrifice some its binding energy, due to the presence of the unstructured N-terminus, to provide energy for its chaperoning activities which have been shown to occur, like all RNA chaperones, in an ATP-independent manner (16). Many RNA chaperones keep the strands of their substrate in proximity to increase the rate of refolding and prevent reversion to a misfolded state (19). The N-terminus of FinO may also be important for low affinity interactions with destabilized single strands which need to be kept from reforming intramolecular base pairs in order to base pair with their complementary sequence. Therefore, this work highlights the importance of disordered regions in proteins with RNA chaperone activity as we have shown that the N-terminus of FinO may be required at multiple steps during intermolecular RNA duplexing.

Concluding remarks: The bigger picture

Structure-function relationships are important for all biological processes. Enzymes use three-dimensionally folded structures to specifically bind substrates and efficiently catalyze reactions. Likewise, the double stranded DNA helix stores vital genetic information and many RNA molecules fold into complex structures for protein recognition and catalysis. The work in this thesis has demonstrated that FinO is also built specifically for its respected functions. The protein consists of two domains, each responsible for its own specific task. Interestingly, the catalytic RNA chaperone domain of FinO is the least structured part of the protein. However, as mentioned above, this property is common to the majority of RNA and protein chaperones. The majority of plasmid-encoded antisense systems do not require a protein co-factor. Therefore, why is FinO needed for the inhibition of bacterial conjugation? The answer may lie in the pathogenic strategy of enterobacteria. In optimal environments, bacteria which harbor antibiotic resistance or virulence factors want to transfer the genes throughout the population. However, they also want to limit production of the conjugation machinery because it is energetically costly to the cell

171 and exposes them to attack from bacteriophages. After donor bacteria transfer F-like plasmids un-repressed for a short period of time, inhibition of conjugation starts as FinO levels increase and the protein begins to bind FinP, increasing its cellular concentration by protecting it from degradation. At a certain cellular concentration, FinO would switch into RNA chaperone mode and facilitate the rapid shut down of conjugation by annealing FinP to traJ mRNA thus preventing translation of TraJ and expression of the genes needed for bacterial conjugation. Therefore, the intricate timing of FinO expression and its important functions allow pathogenicity and ensure the survival of bacteria during the costly bacterial conjugation process.

172 Figure 5-1: Model of the FinO 45-186-SLII interaction

Qualitative molecular model of FinO 45-184 with a SLII-like stem-loop RNA based on the biochemical data from this work. The 14 base pair stem and seven nucleotide loop of the RNA stem-loop was made using Coot (3) from the PDB structures 405D (stem) (15) and 1SZY (loop) (18). The RNA was then modeled with FinO 45-184 using Coot, restraining the complex with data from the FinO-SLII cross-linking and RNase footprinting experiments. The 5' and 3' tails were derived from 5-GACA-3' and 5'-GAUUUU-3' sequences from PDBs 1AJF (10) and 1JGQ (23) respectively. Left. Cartoon representations of FinO 45-184 from the front (left, top) and side (left, bottom). The model stem-loop is shown in gold. The positively charged residues which may be important for RNA contact are shown as blue sticks and are labeled. The FinO residues which formed cross-links with SLII are shown as blue sticks and labeled in boxes. The nucleotides of the RNA which FinO protected from RNase cleavage are indicated by large (strong protection) and small (weak protection) asterisks. This corresponds to A34-U45 and C7-G8 on the 3' and 5' side of the RNA respectively. Right Electrostatic surface representation of FinO 45-186 with the model RNA (gold) from the front (top, right) and side (bottom, right) showing where the RNA is thought to contact the protein. The orientations of the protein and RNA are the same as the Left images. In each image, Q62 is labeled to indicate where the globular domain begins. 173 FinO

%J.

Inhibitory FinP-traJ mRNA Duplex

(Figure 5-2 legend on p. 175)

174 Figure 5-2: The many roles of FinO in FinP-fraJ mRNA intermolecular duplexing (p. 174)

Schematic cartoon highlighting the proposed steps during FinO-mediated FmP-traJ mRNA duplexing in vivo, i) traJ mRNA (gray) and FinP antisense RNA (black) are transcribed in opposite directions from traJ. The 30 S ribosomal subunit binds to structures within SLIc of the 5' UTR of traJ mRNA leading to translation of the TraJ transcription factor, ii) FinO binds to FinP and traJ mRNA. FinO protects FinP from endonucleolytic degradation thereby stabilizing its levels, iii) FinO binding enhances the antisense-target mRNA recognition possibly through protein-protein interactions (double arrow) or using the N- terminal 25 amino acids (shown in red), iv) FinP and traJ mRNA form initial base pairs, possibly through loop-loop interactions. FinO begins to destabilize the stem of SLII and SLIIc using residues 32-42 (shown in red) of the N-terminus. Enlargement: Close up of residues P34-K37 interacting with the upper stem of SLII. It is proposed that W36A intercalates with the stem, destabilizing base pairs within the helix. Surrounding lysine residues are thought to bring W36 close to the RNA helix, v) Duplex progression involves the N-terminus of FinO (red) for destabilization and annealing events, vi) After the duplex has progressed to a certain point, FinO can no longer bind to the 3' portion of SLII or SLIIc and dissociates. The protein may still form non-specific interactions with the progressing intermolecular duplex, continuing its remodeling activities, vii) The stable inhibitory duplex is formed, preventing the 30 S ribosomal subunit from binding. FinO goes on to bind other FinP and traJ mRNA molecules.

175 Figure 5-3: Model of FinO-mediated SLII-SLIIc recognition complex

A) Qualitative molecular model of the symmetrical interaction between two complexes of FinO 45-184 with an SLII-like RNA model. The individual complexes (from Figure 5-1) are shown in blue and gray with the N-and C-terminus of FinO and 5' and 3' ends of the RNA model labeled. The 5' and 3' single stranded tails are not shown for simplicity. The docking was performed using Coot (3). Top left: Front view of interaction. Topright: Side view. Bottom left: Top view. B) Close up of residues involved in potential FinO a6 dimerization interface. D167 and K168 side chains are shown as sticks while V160 side chain is shown as spheres.

176 References

1. Brunei, C, Marquet, R., Romby, P., and Ehresmann, C. (2002) RNA loop-loop interactions as dynamic functional motifs. Biochimie, 84: 925-944.

2. Draper, D.E. (1999) Themes in RNA-protein recognition. J Mol Biol, 293: 255-270.

3. Emsley, P. and Cowtan, K. (2004) Coot: model-building tools for molecular graphics. Acta Crystallogr D Biol Crystallogr, 60: 2126- 2132.

4. Franch, T., Petersen, M., Wagner, E.G., Jacobsen, J.P., and Gerdes, K. (1999) Antisense RNA regulation in prokaryotes: rapid RNA/RNA interaction facilitated by a general U-turn loop structure. J Mol Biol, 294: 1115-1125.

5. Ghetu, A.F., Gubbins, M.J., Frost, L.S., and Glover, J.N. (2000) Crystal structure of the bacterial conjugation repressor FinO. Nat Struct Biol, 7: 565-569.

6. Ghetu, A.F., Gubbins, M.J., Oikawa, K., Kay, CM., Frost, L.S., and Glover, J.N. (1999) The FinO repressor of bacterial conjugation contains two RNA binding regions. Biochemistry, 38: 14036-14044.

7. Gubbins, M.J., Arthur, D.C., Ghetu, A.F., Glover, J.N., and Frost, L.S. (2003) Characterizing the structural features of RNA/RNA interactions of the F-plasmid FinOP fertility inhibition system. J Biol Chem, 278: 27663-27671.

8. Hjalt, T.A. and Wagner, E.G. (1995) Bulged-out nucleotides in an antisense RNA are required for rapid target RNA binding in vitro and inhibition in vivo. Nucleic Acids Res, 23: 580-587.

9. Jerome, L.J. and Frost, L.S. (1999) In vitro analysis of the interaction between the FinO protein and FinP antisense RNA of F-like conjugative plasmids. J Biol Chem, 274: 10356-10362.

10. Kieft, J.S. and Tinoco, I., Jr. (1997) Solution structure of a metal- binding site in the major groove of RNA complexed with cobalt (III) hexammine. Structure, 5: 713-721.

11. Kim, J.L., Morgenstern, K.A., Griffith, J.P., Dwyer, M.D., Thomson, J.A., Murcko, M.A., Lin, C, and Caron, P.R. (1998) Hepatitis C virus NS3 RNA helicase domain with a bound oligonucleotide: the crystal structure provides insights into the mode of unwinding. Structure, 6: 89-100.

177 12. Kolb, F.A., Westhof, E., Ehresmann, C, Ehresmann, B., Wagner, E.G., and Romby, P. (2001) Bulged residues promote the progression of a loop-loop interaction to a stable and inhibitory antisense-target RNA complex. Nucleic Acids Res, 29: 3145- 3153.

13. Koraimann, G., Teferle, K., Markolin, G., Woger, W., and Hogenauer, G. (1996) The FinOP repressor system of plasmid R1: analysis of the antisense RNA control of traJ expression and conjugative DNA transfer. Mol Microbiol, 21: 811-821.

14. Marians, K.J. (2000) Crawling and wiggling on DNA: structural insights to the mechanism of DNA unwinding by helicases. Structure, 8: R227-235.

15. Pan, B., Mitra, S.N., and Sundaralingam, M. (1998) Structure of a 16-mer RNA duplex r(GCAGACUUAAAUCUGC)2 with wobble C.A+ mismatches. J Mol Biol, 283: 977-984.

16. Rajkowitsch, L, Chen, D., Stampfl, S., Semrad, K., Waldsich, C, Mayer, O., Jantsch, M.F., Konrat, R., Blasi, U., and Schroeder, R. (2007) RNA chaperones, RNA annealers and RNA helicases. RNA Biol, 4: 118-130.

17. Sandercock, J.R. and Frost, L.S. (1998) Analysis of the major domains of the F fertility inhibition protein, FinO. Mol Gen Genet, 259: 622-629.

18. Schweisguth, D.C. and Moore, P.B. (1997) On the conformation of the anticodon loops of initiator and elongator methionine tRNAs. J Mol Biol, 267: 505-519.

19. Tompa, P. and Csermely, P. (2004) The role of structural disorder in the function of RNA and protein chaperones. FASEB J, 18: 1169-1175.

20. van Biesen, T. and Frost, L.S. (1994) The FinO protein of IncF plasmids binds FinP antisense RNA and its target, traJ mRNA, and promotes duplex formation. Mol Microbiol, 14: 427-436.

21. Velankar, S.S., Soultanas, P., Dillingham, M.S., Subramanya, H.S., and Wigley, D.B. (1999) Crystal structures of complexes of PcrA DNA helicase with a DNA substrate indicate an inchworm mechanism. Cell, 97: 75-84.

22. von Hippel, P.H. and Delagoutte, E. (2001) A general model for nucleic acid helicases and their "coupling" within macromolecular machines. Cell, 104: 177-190.

178 23. Yusupova, G.Z., Yusupov, M.M., Cate, J.H., and Noller, H.F. (2001) The path of messenger RNA through the ribosome. Cell, 106: 233-241.

179 180 Appendix A

Assaying FinO structural homologs for RNA chaperone activity

Introduction After the crystal structure of FinO 26-186 was solved in 2000 (2), it was determined that FinO had a completely novel protein fold as searches failed to identify any structural homologs from the Protein Data Bank. The fold of FinO confirms the old adage that structure leads to function. The protein is divided into two domains, resides 1-61 and 62- 186, which bind to RNA with different affinities based on their functional needs (3). The N-terminal 1-61 domain has a weak affinity to RNA by itself, was shown to harbor the RNA chaperone activity of FinO (1, 3). RNA chaperones typically bind to their substrates weakly and non- specifically (10). Association with target RNAs is needed for the destabilization of misfolded secondary and tertiary structures, allowing the RNAs to fold into their native confirmation. However, chaperones are not needed for maintaining the structure of the properly folded RNA (10). Similarly, for FinO, weak RNA binding at the N-terminal RNA domain is needed to efficiently catalyze destabilization events which lead to FinP- traJ mRNA duplexing. In contrast, the 62-186 domain of FinO is important for the RNA recognition step, binding specifically to the 3' lower portion of FinP SLII and traJ mRNA SLIIc with high affinity [Chapter 4 and (6)]. Specific binding to both complementary RNAs would not only bring them into proximity for intermolecular duplexing but may also anchor the protein on the RNA, positioning the N-terminal chaperone domain close to areas of the RNA which need to be destabilized and re-annealed with complementary sequences (see Chapter 5). In addition, the high affinity interaction with FinP protects the RNA from RNase E degradation within the cell (7). At a certain point in the intermolecular duplexing process, FinO would not be needed anymore and would dissociate from the stable

181 duplex structure, which is degraded by RNase III (7), allowing the protein to repeat its chaperone function with another pair of RNAs. Recently, two proteins have surfaced as structural homologs of FinO: the E. coli osmotic regulatory protein ProQ and the protein of unknown function NMB1681 from Neisseria meningitidis serogroup B strain MC58.

The N-terminus of ProQ is homologous to the FinO globular domain Osmotic regulation is critical for the survival of bacterial cells. Osmotic upshifts, or increases in the external solute concentration, lead to water efflux from the cell, causing a decrease in cytoplasmic volume and macromolecular crowding (12). The cellular changes trigger the activation of osmoregulatory transporters involved in moving K+ or organic osmoprotectants into the cell to restore hydration (12). One such osmoprotectant/hT symporter is ProP, which transports proline and glycine betaine into the cell in response to an osmotic upshift (12). E. coli ProP is a large integral membrane protein with an extended C-terminal a- helical domain which forms a coiled-coiled dimer with an adjacent ProP (12). The dimerization was shown to be important for the response of ProP to osmotic upshifts in vivo (12). A positive regulator, ProQ, is needed for amplifying the response to an osmotic upshift (9). An insertion lesion in the proQ locus impairs osmoactivation of ProP but does not affect its transcription or translation suggesting that it acts on ProP post- translationally (8, 9). ProQ is a highly basic, 26 kDa cytoplasmic protein (12) which contains two domains, an N-terminal a-helical domain and a C-terminal p- sheet domain, which are separated by a flexible linker (13). The amino acid sequence of the N-terminal domain is very similar to the sequence of the FinO globular domain (Expectation value = 1.6 x 10~8 and 25% sequence identity) while a structure to sequence alignment was gapless for ProQ residues 7-121 and FinO residues 70-184 (12). Figure A-1 shows cartoon and electrostatic representations of the ProQ homology model which is based on the structure of the globular domain of FinO (12). The majority of polar and charged side chains were placed on the

182 surface of the model while hydrophobic residues were buried. In addition, the stereochemistry of the model was validated using PROCHECK, returning an overall positive score of 0.14 (12). Sequence to structure alignments predicted that the C-terminal domain of ProQ is similar to proteins which contain a five-strand (5-meander characteristic of SH3 domains and Sm motifs (12). The individual domains were tested for their ability to activate ProP in cells lacking the proQ gene (13). It was found that the N-terminal fragment could partially complement the deletion whereas full activition was restored with a construct which contained both N- and C-terminal domains lacking the linker (13). This showed that the N-terminus is responsible for amplifying the ProP-mediated response to an osmotic upshift, however a component of the C-terminus is needed for full activity possibly by interacting with another protein through its SH3/Sm-like fold (13).

Neisseria meningitidis serogroup B strain MC58 protein of unknown function NMB 1681 is structurally similar to FinO The bacterial pathogen Neisseria meningitidis is responsible for the devastating meningitis disease which has caused outbreaks throughout the world but remains an epidemic in sub-Saharan Africa (14). Its prevalence ranges from rare to 1000 cases in 100000 each year (14). The disease spreads from the transfer of the meningococci through very close contact with saliva or respiratory secretions, after which the bacteria penetrate the mucosal barrier and colonize the epithelial tissues of the upper respiratory tract (14). Part of the pathogenic success of the bacterium comes from its ability to alter its genetic material and surface structure. N. meningitidis is sub-divided into 13 serogroups, six of which are pathogenic (A, B, C, W-135, X, Y), based on the composition of their outer capsular polysaccharide (14). Serogroup A meningococci are responsible for the epidemic outbreak of meningitis in Africa, while serogroup B has produced prolonged outbreaks in many industrialized countries around the world leading to many deaths (14). The individuals most at risk from the disease are infants (14).

183 The complete genome of Neisseria meningitidis serotype B strain MC58 (isolated from an invasive infection) was sequenced in 2000 (15). A structural genomics project is underway at the Midwest Center for Structural Genomics (MCSG) (www.mcsg.anl.gov/) to determine the protein structures for the N. meningitidis strain MC58 using high- throughput X-ray crystallography. Presently, out of the 877 protein targets cloned, only 18 structures have been inputted into the Protein Data Bank (from MCSG website). As with many structural genomics projects, the major bottlenecks in the process are poor solubility and crystallizability of the targets (4). A recent Dali search (5) for FinO structural homologs returned a match for the PDB ID 2HXJ with Z-scores ranging from 7.8-9.0 for the six molecules in the asymmetric unit of the crystal. In comparison, statistically significant search hits from the PDB database result in Z- scores > 2 (5). The retrieved structures were for the 141 amino acid protein of unknown function NMB1681 from N. meningitidis strain MC58. Figure A-3A shows cartoon representations of the six proteins (in various shades of gray) surrounding the FinO 26-186 crystal structure (blue and yellow). The structure for NMB1681 looks similar to FinO as it is mainly a-helical with two short anti-parallel 3-strands. Each molecule in the asymmetric unit forms a compact globular structure with varying degrees of a-helicity in its N-terminal residues. The break in the N-terminal helix for four of the molecules is at the same three residue stretch of threonine- selenomethionine-serine. In all cases, the highest crystallographic B- factors map to the N-terminal helix suggesting high flexibility in this region.

Since the fold of FinO highly complements its function, we wondered if ProQ and NMB1681 might exhibit RNA chaperone activity. We used the intermolecular duplexing and strand-exchange assays developed in Chapter 2 to investigate this possibility. In addition to binding the 5' UTR of traJ mRNA, ProQ was able to mediate Sll strand exchange and SLII-SLIIc duplexing. This preliminary data suggests ProQ may mediate osmotic activation of ProP at the post-transcriptional level.

184 In contrast, we found that NMB1681 did not exhibit duplexing of SLII- SLIIc suggesting it may not act as a chaperone. However, more experiments need to be carried out to confirm this hypothesis.

Results and discussion

Assaying ProQ for RNA chaperone activity

ProQ appears to bind to the 5' UTR of traJ mRNA but not the SUIc domain Prior to subjecting ProQ to RNA chaperone assays, we wanted to determine if the protein exhibited RNA binding activity as it is a highly basic protein with homology to the FinO RNA binding domain. Experiments are currently underway to determine the RNA targets (if any) of ProQ using pull-down assays with a His6-tagged version of the protein (being conducted in the lab of Dr. Janet Wood, University of Guelph). Until then, we decided to test ProQ binding with two RNAs: SLIIc and the 5' UTR of traJ mRNA (1-112) (Figure A-2A). We set up electrophoretic mobility shift assays (EMSAs) with the two radiolabeled RNAs and increasing amounts of His6-tagged ProQ or wild-type FinO as a control. The reactions were incubated at 4°C before loading onto an 8% native gel. FinO bound to both RNAs as expected, producing clearly resolved RNA-protein complex bands. For traJ mRNA, two supershifts resulted over the protein concentration range, indicating that two FinO molecules could bind to one traJ mRNA (Figure A-2B). Jerome ef a/, also observed binding of multiple FinO-GST monomers onto a similar traJ mRNA construct (6). ProQ-His6 also bound to traJ mRNA, however, the majority of the RNA was shifted to a higher order complex which remained in the well (Figure A-2B). A very small amount of 1:1 complex was visualized on the gel indicating that ProQ may oligermize on the traJ mRNA. We found that ProQ-His6 did not bind to SLIIc at protein concentrations < 10 uM (data not shown). This suggests that SLIIc is not a specific binding target of ProQ, however, a wider protein titration and competition experiments are needed to confirm this preliminary result.

185 ProQ mediates intermolecular duplexing between SLII and SLIIc RNAs To test if ProQ acts as an RNA chaperone, we performed intermolecular duplexing assays with radiolabeled SLII and SLIIc (see Figure 2-4A for a description of the assay). The two RNAs do not form an intermolecular duplex in the absence of FinO due to their extremely stable stem regions which need to be destabilized prior to intermolecular annealing. In accordance with previous studies, FinO was able to facilitate formation of a full SLII-SLIIc duplex within the duration of the assay (Figure A-2C). Surprisingly, ProQ-His6 mediated the formation of the intermolecular duplex at concentrations 2.5-fold lower than wild-type FinO. Like FinO, the annealing behavior of ProQ was dependent on protein concentration (compare Figure 2-4E to Figure A-2C). However, unlike FinO, ProQ is able to efficiently mediate duplexing at very high protein concentrations suggesting its annealing activity may function by a different mechanism. A wider protein titration is still needed to confirm this result.

ProQ mediates strand-exchange ofSII Since ProQ mediates intermolecular duplexing between SLII and SLIIc, we decided to test if the protein could perform strand exchange on the SI I duplex used to characterize the chaperone activity of FinO (see Figure 2-2A for a description of the assay). In the absence of protein, the radiolabeled Sll duplex does not exchange its radiolabeled SII(A) strand for the cold SI 1(A) strand which is present in a molar excess (Figure A- 2D). In contrast, wild-type FinO is able to mediate strand exchange at final concentrations greater than 5 uM (Figure A-2D). The concentration at which FinO performed strand exchange was 5 times higher than normal experiments, possibly due to an older FinO stock solution.

Interestingly, ProQ-His6 was also able to perform strand-exchange, again at concentrations much lower than FinO (0.25 uM final). Like the intermolecular duplexing assay, an increased ProQ-His6 concentration did not affect ProQ strand-exchange (Figure A-2D).

186 Discussion The preliminary study determined that ProQ appears to act as an RNA chaperone to facilitate intermolecular duplexing and strand exchange. ProQ appeared to bind to traJ mRNA as a higher order complex. Previous data showed that ProQ-His6 purifies as monomers or dimers (12), suggesting that ProQ would need to form a higher order oligomer upon RNA binding. ProQ did not bind to the SLIIc domain suggesting that it may need the single stranded section region of traJ mRNA, located upstream of SLIIc, for specific contacts. The C-terminal domain of ProQ is predicted to adopt an SH3 or Sm-like fold. Sm-like proteins, such as Hfq have been shown to form ring-like homo-hexamers or heptamers (11). It is possible that ProQ may form such a structure upon binding RNAs such as traJ mRNA. Interestingly, traJ mRNA is a target of Hfq, which binds at the A/U-rich single stranded sequence upstream of the SLIIc domain (16). Multi-angle laser light scattering experiments are underway to determine the oligomeric state of the ProQ sample. To characterize the binding activity of ProQ further, a larger set of RNA constructs with different structures and sequences will be needed. Also, information about possible in vivo RNA targets would be useful. Like Hfq, ProQ can mediate intermolecular base pairing of complementary RNAs (17). Hfq has also been shown to destabilize areas of the antisense RNA, OxyS, possibly enhancing the interaction with its target fhIA mRNA (17). ProQ was able to mediate Sll strand exchange indicating that it may be able to destabilize double stranded RNA. The lack of dependence on protein concentration for the chaperone assays suggests that ProQ uses a different mechanism than FinO to remodel its RNA substrates. Testing different N- and C-terminal truncation mutants of ProQ (13) would be useful for assigning RNA binding and chaperone functions to the two domains of ProQ. Also, as a control, the ability of Hfq to mediate SLII-SLIIc intermolecular duplexing and Sll strand exchange should be tested to compare with the ProQ results.

Previous studies have shown that mutations in the proQ locus impair osmotic activation of the ProP transporter (9). However, the

187 transcription and translation of proP was not compromised indicating that ProQ acted on ProP post-translationally (8, 9). The results from this preliminary study suggest that ProQ may act to regulate the translation of mRNAs encoding proteins which interact with ProP. This may proceed through an unidentified antisense-mRNA interaction to inhibit translation or by altering the mRNA structure leading to an enhancement of translation or increased degradation of the transcript. However, cellular RNA targets of ProQ still need to be identified before any hypotheses can be made.

Assaying NMB1681 for RNA chaperone activity

NMB1681 does not exhibit SLII-SLIIc intermolecular duplexing activity Since NMB1681 has similar structural features as FinO, we decided to test whether the bacterial protein exhibited RNA chaperone activities. Purified protein was obtained from the MCSG group which solved the structure of the protein. We performed SLII-SLIIc intermolecular duplexing assays with increasing concentrations of NMB1681 and 1 uM final concentrations of wild-type FinO, FinO 33-186 W36A and FinO 45-186 as controls (Figure A-3B). We found that while wild-type FinO and FinO 33-186 W36A was able to form duplexes with either radiolabeled SLII (left, Figure A-3B) or SLIIc (right) NMB1681 could not mediate duplex formation. As with previous experiments, FinO 45- 186 could not form and SLII-SLIIc duplex.

Discussion From the experiments, it appears that NMB1681 does not exhibit intermolecular annealing activity despite its structural similarity to FinO. However, more experiments need to be performed before we can rule out the possibility that it behaves as an RNA chaperone. Due to time constraints, we did not try the strand-exchange and RNA binding assays. The protein has a predicted pi of 8.7, indicating that it might have RNA binding activity. As with ProQ, more research is needed to identify any specific RNA targets.

188 Materials and Methods

Expression and purification of ProQ-His6^

ProQ-His6 was expressed and purified as described in previous

studies (12). ProQ-His6 was purified in 50 mM HEPES pH 7.3, 600 mM NaCI, 1 mM EDTA, and 5% v/v glycerol.

Expression and purification ofNMB1681 Purified NMB1681 was a gift from Kemin Tan at the MCSG. The protein was concentrated to 90 mg/mL in 20 mM HEPES pH 8.0, 250 mM NaCI and 2 mM DTT.

In vitro SLII-SLIIc intermodular duplexing and Sll strand exchange assays SLII and SLIIc RNA preparation is described in Chapter 4 Materials and Methods. The synthesis of the Sll strands and formation of the Sll duplex is described in Chapter 2 Materials and Methods. Radiolabeling of the RNA substrates is described in Chapter 4 Materials and Methods. The in vitro assays were performed as described in

Chapter 2 Materials and Methods section. ProQ-His6 samples were diluted to specific working concentrations in 50 mM HEPES pH 7.3, 600 mM NaCI, 1 mM EDTA, and 5% glycerol. The protein could not be transferred to lower salt concentrations during dilution because it would crash out of solution (12). NMB1681 samples were diluted in stock buffer (see above).

Preparation of SLIIc and traJ mRNA (1-112) forEMSA experiments SLIIc was chemically synthesized, radiolabeled and purified according to Chapter 4 Materials and Methods. The traJ mRNA (1-112) sample was produced by in vitro run-off transcription due to its large size. traJ inserts were generated by PCR. Primers and templates were synthesized by Integrated DNA technologies.

1 This work was performed by Ross A. Edwards 189 The traJ mRNA insert was formed in two PCR reactions. The first reaction templates were: traJ 5'temp (5-TGA AAT TGA AAA TCG CCG ATG CAG GGA GAC GTG AAC TCC CTG CAT CGA CTG TCC ATA-3'), and traJ 3'temp (5-TAT GGA CAG TCG ATG CAG GGA GTT CAC GTC TCC CTG CAT CGG CGA TTT TCA ATT TCA-3'). The primers for the first PCRs were used to extend make a larger 114 base pair template: traJ fwdl (5'-GGG CGT GGT TAA TGC CAC GTT AAA ATT TGA AAT TGA AAA TCG CCG-3") and traJ revl (5'-GAT ACA TAG GAA CCT CCT CAC AAA GGA TTC TAT GGA CAG TCG ATG CAG-3'). The first PCR produced two products 84 (traJ fwd/traJ 3'temp) and 87 (traJ rev/traJ 5'temp) nucleotides in length which self-prime and extend to produce the 114 base pair template. PCR reactions were performed with Platinum Pfx DNA polymerase (Invitrogen) using standard protocol specified by manufacturer. The PCR annealing time was 30 seconds and the temperature was varied from 55°C for 5 cycles, 61°C for 10 cycles and then 55°C for 15 cycles. The extension time was 15 seconds at 68°C. The PCR products were then run on a 4% agarose gel and purified using the QIAquick gel extraction kit (Qiagen) to remove any 72 nucleotide product. The primers for the second PCR reaction were: traJ fwd2 (5'- GCG CGC TAC GTA ATA CGA CTC ACT ATA GGG CGT GGT TAA TGC CAC-3') containing the Snabl restriction sequence, and traJ rev2 (5'- GCG CGC GAA TTC GAT ACA TAG GAA CCT CCT-3') which contained the EcoRI sequence. The second PCR reaction was performed for 30 cycles with an annealing temperature held at 55°C for 30 seconds for this reaction and the extension time was 15 seconds at 68°C, and contained the purified traJ PCR product from above and the traJ fwd2 and traJ rev2 primers to produce the 153 base pair insert. The PCR product was purified using the QIAquick PCR purification kit and eluted in 50 uL of 10 mM Tris-HCI pH 8.5. The insert was then cloned into the LITMUS 28i in vitro transcription vector (New England Biolabs, NEB) which contains the T7 promoter. 10 ug of the vector and insert were digested with Snabl (NEB) and EcoRI (NEB) using NEBuffer 4 and standard conditions. Restriction

190 digests were performed for 2.5 hours at 37°C and then purified by the QiaQuick PCR purification kit (Qiagen). The digested vector was treated with Calf Intestinal Alkaline Phosphatase (Invitrogen) for 1 hour at 50°C and purified by the QiaQuick PCR purification kit. A 20 uL ligation reaction was performed with 90 fmol of traJ insert and 30 fmol of LITMUS 28i vector and 1 U of T4 DNA Ligase (NEB) for 18 hours at 14°C. The reaction was diluted to 100 uL and transformed into E. coli DH5a cells (Invitrogen). Colonies were screened for the insert by PCR and confirmed by restriction digest before sending the plasmid for sequencing to verify the traJ DNA sequence. Large scale amounts of the traJ plasmid were grown up and purified by Maxiprep (University of Alberta Biochemistry DNA Core Lab). 100 ug of fraJ-LITMUS 28i plasmid was linearized with 10 U of EcoRI and Stul per ug of plasmid in NEBuffer 2. Digestion products were purified using the QiaQuick PCR purification kit. Transcriptions were carried out in 1 mL reactions: 50 uL 20X transcription buffer (0.8 M Tris-HCI pH 8.1, 20 mM spermidine, 0.2% Triton X-100, 100 mM dithiothreitol), 160 uL 50% v/v polyethylene glycol 8000, 320 uL of 0.1 M solution of nucleoside triphosphates (adenosine triphosphate, uridine triphosphate, cytosine triphophate, guanosine triphosphate), 56 uL 1 M magnesium chloride, 10 uL 100 U/mL inorganic pyrophosphatase (Sigma-Aldrich), 50 ug of linearized plasmid, 100 uL 0.5 mg/mL T7 RNA polymerase. Transcriptions were incubated at 37°C for four hours and stopped by phenol/chloroform extraction. RNA samples were ethanol precipitated and purified by urea denaturing PAGE. The RNA bands were visualized by UV shadowing, cut out, and extracted from the gel slice by electroelution. Eluted samples were ethanol precipitated, resuspended in 10 mM Tris-HCI pH 7.5, 1 mM EDTA, and quantified by UV absorbance using extinction coefficients calculated using an online oligonucleotide calculator (Ambion). Radiolabeling and annealing of both SLIIc and traJ mRNA (1-112) are described in the Materials and Methods section of Chapter 4.

191 ProQ-His6 EMSA with traJ mRNA (1-112) and SLIIc Binding experimetns were formed on ice in 10 p.L reactions: 5 |jL 2 X binding buffer (50 mM Tris-HCI pH 7.5, 200 mM NaCI, 2 mM EDTA), 1 pL of wild-type FinO (in 50 mM HEPES pH 7.0, 200 mM NaCI, 1 mM

EDTA, 5 mM TCEP), ProQ-His6 (in 50 mM HEPES pH 7.3, 600 mM NaCI, 1 mW\ EDT/\, 5% gtycevoV) a\ \

2 uL ddH20. Reactions were mcubateci for 20 rr\\r\u\es behove a

192 FinO (33-184) HI!

»'£!.''.;

ProQ (1-121)

Figure A-1: Comparison of the ProQ (1-121) homology model to FinO (33-184)

Structures of the crystal structure of FinO 26-186 (top) [PDB ID=1DVO (2); no density for residues 26-32 and 185-186) and the homology model for the N-terminal 121 amino acids of ProQ (bottom). Left. Cartoon representations of the structures colored from blue at the N-terminus through to red at the C-terminus. Right Electrostatic potential surface representations of both protein structures. Blue represents postively charged residues while red represents negatively charged residues.

2 Adapted from Smith, M.N. et. al. (2004) Biochemistry 43: 12979-12989. 193 A

10 A A U U U G

6-C SUM U-A G-C C-G 20 30 5-GGG UUAAAAUUUGAAAUUGAAAAUCGC 1 5'UTRoflraJmRNA(1-112)

FinO 1-186 WT ProQ-His6 B 5~i 5~To [fiM Final] -*-(xPmQ)-traJ

(2 FmO)-traJ- FinO-traJ- -*- PmQ-traJ ~*-32P-traJ mRNA

FinO WT ProQ-His6 NP 1 .08 .2 .4 1 2 5 10 [^M Final]

• SLII/SLlIc duplex

P-SLII

D FinO 1-186 WT ProQ-His„ NP .5 1 2.5 5 10 .1 .25 .5 1 2.5 5 10 NP [(iM Final]

Sll duplex

32 -*- P-SII(A)

(Figure A-2 legend on p. 195)

194 Figure A-2: Assaying ProQ-His6 for RNA binding and chaperone activity (p. 194)

A) Sequence and secondary structure of traJ mRNA (1-112). The SLIIc domain is shown in bold letterining. Additional G nucleotides (italic and underlined) at the 5' and 3' end result from cloning of traJ DNA template for in vitro transcription. The ribosomal binding site (RBS) and AUG start codon in SLI are shown. B) EMSA studies of wild-type FinO and ProQ-His6 with the 5' UTR of traJ mRNA. Binding reactions were run on an 8% native gel for 2-3 h at 4°C. The final protein concentration (in \xM) for each reaction is shown above each gel. Left: Radiolabeled traJ mRNA is incubated with increasing amounts of wild-type FinO. Bands representing FinO-fraJ and (2FinO)-fraJ complexes are annotated at the left of the gel. Right. Binding reactions of traJ mRNA with increasing amounts of ProQ-His6. Bands representing ProQ-traJ and (x ProQ)-traJ complexes are annotated at the right of the gel. C) Concentration dependence of ProQ-His6 on intermolecular duplexing of SLII/SLIIc. Wild-type FinO at a final concentration of 1 uM is used as a control duplexing reaction. Reactions were run for one hour before being stopped and loaded on a 10% native gel. The final concentration of ProQ-HiS6 (in \iM) is noted above the gel. NP=No Protein. The 32P-SLII/SLIIc duplex and 32P-SLII bands are noted at the right of the gel. D) Concentration dependence of wild-type FinO (control) and ProQ-HiS6 on Sll strand exchange. Reactions were run for two hours before being stopped and loaded on a 15% native gel. The final protein concentration (in ^M) is noted above the gel. The 32P-SII duplex and 32P-SII(A) band is noted at the right of the gel.

195 A

AT-,

B

P-Duplex

P-SLIIx

32 32, P-SLII/SLIIc P-SLIIc/SLII

Figure A-3: Assaying Neisseria meningitidis strain MC58 protein of unknown function NMB1681 for RNA chaperone activity

A) Cartoon representations of the crystal structures of six molecules of NMB1681 (shown in various shades of gray) (PDB ID = 2HXJ). For comparison, the structure of FinO 26- 186 is shown in blue (a-helices) and yellow (S-strands) in the middle of the diagram. The N- and C-terminus are noted. B) Concentration dependence of NMB1681 on intermolecular duplexing of SLII/SLIIc. Reactions with wild-type FinO, FinO 33-186 W36A, and FinO 45-186 (at 1 uM final concentration) were used as controls. Reactions were run for one hour before being stopped and loaded on a 10% native gel. The increasing final protein concentration of NMB1681 is represented by a triangle above the gel. Final concentrations (in uM): 0.16, 0.4, 0.8, 2, 4, 10, 20, 50. The 32P-intemolecular duplex and 32P-SLII or SLIIc bands (collectively annotated as SLIIx) are noted at the right of the gel. NP=No protein.

196 References

1. Arthur, D.C., Ghetu, A.F., Gubbins, M.J., Edwards, R.A., Frost, L.S., and Glover, J.N. (2003) FinO is an RNA chaperone that facilitates sense-antisense RNA interactions. EMBO J, 22: 6346- 6355.

2. Ghetu, A.F., Gubbins, M.J., Frost, L.S., and Glover, J.N. (2000) Crystal structure of the bacterial conjugation repressor FinO. Nat Struct Biol, 7: 565-569.

3. Ghetu, A.F., Gubbins, M.J., Oikawa, K., Kay, CM., Frost, L.S., and Glover, J.N. (1999) The FinO repressor of bacterial conjugation contains two RNA binding regions. Biochemistry, 38: 14036-14044.

4. Goulding, C.W. and Perry, L.J. (2003) Protein production in Escherichia coli for structural studies by X-ray crystallography. J Struct Biol, 142: 133-143.

5. Holm, L. and Sander, C. (1995) Dali: a network tool for protein structure comparison. Trends Biochem Sci, 20: 478-480.

6. Jerome, L.J. and Frost, L.S. (1999) In vitro analysis of the interaction between the FinO protein and FinP antisense RNA of F-like conjugative plasmids. J Biol Chem, 274: 10356-10362.

7. Jerome, L.J., van Biesen, T., and Frost, L.S. (1999) Degradation of FinP antisense RNA from F-like plasmids: the RNA-binding protein, FinO, protects FinP from ribonuclease E. J Mol Biol, 285: 1457-1473.

8. Kunte, H.J., Crane, R.A., Culham, D.E., Richmond, D., and Wood, J.M. (1999) Protein ProQ influences osmotic activation of compatible solute transporter ProP in Escherichia coli K-12. J Bacteriol, 181: 1537-1543.

9. Milner, J.L. and Wood, J.M. (1989) Insertion proQ220::Tn5 alters regulation of proline porter II, a transporter of proline and glycine betaine in Escherichia coli. J Bacteriol, 171: 947-951.

10. Rajkowitsch, L, Chen, D., Stampfl, S., Semrad, K., Waldsich, C, Mayer, O., Jantsch, M.F., Konrat, R., Blasi, U., and Schroeder, R. (2007) RNA chaperones, RNA annealers and RNA helicases. RNA Biol, 4: 118-130.

11. Schumacher, M.A., Pearson, R.F., Moller, T., Valentin-Hansen, P., and Brennan, R.G. (2002) Structures of the pleiotropic translational regulator Hfq and an Hfq-RNA complex: a bacterial Sm-like protein. EMBO J, 21: 3546-3556.

197 12. Smith, M.N., Crane, R.A., Keates, R.A., and Wood, J.M. (2004) Overexpression, purification, and characterization of ProQ, a posttranslational regulator for osmoregulatory transporter ProP of Escherichia coli. Biochemistry, 43: 12979-12989.

13. Smith, M.N., Kwok, S.C., Hodges, R.S., and Wood, J.M. (2007) Structural and functional analysis of ProQ: an osmoregulatory protein of Escherichia coli. Biochemistry, 46: 3084-3095.

14. Stephens, D.S., Greenwood, B., and Brandtzaeg, P. (2007) Epidemic meningitis, meningococcaemia, and Neisseria meningitidis. Lancet, 369: 2196-2210.

15. Tettelin, H., Saunders, N.J., Heidelberg, J., Jeffries, A.C., Nelson, K.E., Eisen, J.A., Ketchum, K.A., Hood, D.W., Peden, J.F., Dodson, R.J., Nelson, W.C., Gwinn, M.L., DeBoy, R., Peterson, J.D., Hickey, E.K., Haft, D.H., Salzberg, S.L., White, O., Fleischmann, R.D., Dougherty, B.A., Mason, T., Ciecko, A., Parksey, D.S., Blair, E., Cittone, H., Clark, E.B., Cotton, M.D., Utterback, T.R., Khouri, H., Qin, H., Vamathevan, J., Gill, J., Scarlato, V., Masignani, V., Pizza, M., Grandi, G., Sun, L, Smith, H.O., Fraser, CM., Moxon, E.R., Rappuoli, R., and Venter, J.C. (2000) Complete genome sequence of Neisseria meningitidis serogroup B strain MC58. Science, 287: 1809-1815.

16. Will, W.R. and Frost, L.S. (2006) Hfq is a regulator of F-plasmid TraJ and TraM synthesis in Escherichia coli. J Bacteriol, 188: 124-131.

17. Zhang, A., Wassarman, K.M., Ortega, J., Steven, A.C., and Storz, G. (2002) The Sm-like Hfq protein increases OxyS RNA interaction with target mRNAs. Mol Cell, 9: 11-22.

198 Appendix B:

Biochemical characterization of the envelope stress accessory protein, CpxP1

Overview

The two component CpxAR signal transduction system is used to combat stress to the bacterial cell envelope. Responding to misfolded or aggregated proteins at the periplasmic side of the inner membrane, CpxA, an inner membrane spanning histidine kinase, phosphorylates the cytoplasmic CpxR response regulator which activates genes needed to restore the cell envelope to its normal state. Another gene which is activated codes for an accessory protein, CpxP, which is needed for down-regulation of the pathway. CpxP is thought to associate with CpxA, preventing it from being activated in absence of a proper stress signal. Once the pathway is activated, CpxP is thought to be removed from CpxA in a manner dependent on the periplasmic DegP protease. Previous mutagenesis studies have isolated two areas of CpxP which are important for its inhibition activity. These are thought to be located at predicted a-helical domains at the N- and C-terminus of CpxP. Here, we expand on this study by characterizing the structure of CpxP using biochemical and biophysical studies. Circular dichroism experiments show that CpxP adopts a mainly a-helical structure and that it may take on a slightly more compact form at pH 8.0, an extracellular condition known to activate the Cpx stress response. Static light scattering experiments show that CpxP forms a dimer at pH 5.8 and 8.0. Finally, we use limited proteolysis to find a suitable CpxP crystallization construct which will be used to optimize crystals obtained with the wild-type protein.

1 This work was in collaboration with Dr. Tracy Raivio from the department of Biological Sciences at University of Alberta. Part of this work has been submitted for publication (1). 199 Introduction

To survive sub-optimal growth conditions, bacteria have evolved extensive mechanisms to protect their cellular envelope. These include the GE stress response pathway and the BaeSR and CpxAR two component signal transduction systems (21). The focus of this chapter will be on the CpxAR system which responds to protein misfolding events caused by a number of stresses such as an increase in extracellular pH (3) and the overexpression of the outer membrane lipoprotein NIpE (24), P-pilin (in the absence of its periplasmic PapD chaperone) (11), and E. co//type IV bundle forming pilius subunit, BfpA (17). The system involves the inner membrane histidine autokinase (HK) CpxA and the cytoplasmic response regulator (RR) CpxR. After sensing the presence of misfolded protein or aggregates at the periplasmic side of the inner membrane, CpxA phosphorylates CpxR at a conserved aspartate residue (23). CpxA also acts as a phosphatase to remove the phosphate from CpxR when the Cpx pathway is turned off (23). Phosphorylation of the CpxR transcription factor enhances its DNA- binding to genes in the Cpx regulon which are involved in restoring the envelope to its normal state (21). These include genes for the periplasmic DegP protease which acts to degrade misfolded proteins; protein folding proteins such as the PpiA and PpiD peptidyl-prolyl- isomerases; the periplasmic disulfide oxidase DsbA; and CpxA and CpxR themselves leading to autoactivation and amplification of the signal (21, 22). The mechanism of induction of the CpxAR system remains unclear as it is not known whether the periplasmic sensing domain of CpxA binds unfolded proteins directly or is activated indirectly through sensing the levels of other periplasmic folding proteins (22). Under non- stressful conditions, an accessory protein, CpxP, is required to keep the Cpx pathway in an inhibited state. Overexpression of the protein has been shown to downregulate the Cpx stress response in a mechanism which involves the sensing domain of CpxA (22). It is thus thought, although not shown yet, that CpxP interacts with the sensing domain and

200 prevents CpxA from receiving stress signals (2). Upon activation, CpxP is thought to be degraded by DegP either directly or through is association with misfolded proteins, thus freeing CpxA to transduce the stress signal (2, 9). However, the exact cellular role of CpxP is still under debate since the Cpx pathway can be activated when the protein is overexpressed or when it is absent from the cell (22). Recent data by Buelow et al. has shown that both CpxP and DegP are needed for early activation of the Cpx response to alkaline pH stress and that DegP is upstream of CpxP in the Cpx pathway (1). However, depending on the inducing cue, the fate of CpxP is not always proteolysis by DegP. While an increase in alkaline extracellular environment and an overproduction of BfpA caused CpxP to be degraded, the overexpression of NIpE did not (1) suggesting that activation by NIpE is different from the response to misfolded proteins. The cpxP gene is located in the Cpx regulon and, like cpxA and cpxR, its expression is enhanced upon activation of the pathway (3). It is therefore thought that CpxP plays a role to fine-tune the response. Binding of the protein to CpxA would prevent unnecessary signaling noise from inducing a response and its presence would be important for shutting down the pathway after the stress response is alleviated (1, 2). Mutagenesis studies have identified two classes of CpxP mutants which disrupt the ability to inhibit the Cpx response (2). Five loss-of- function (LOF) mutations localized to an area between residues 55 and 61 of a predicted a-helix at the N-terminus of CpxP while one LOF mutant resided at residue 128 of a predicted C-terminal a-helix. One class of mutants disrupted inhibition but was present at normal levels, while the second class, disrupted inhibition and was present at decreased levels indicating structural instability. The N-terminal residues are highly conserved indicating that this stretch of residues in the predicted a-helix are very important for CpxP function, possibly through interactions with CpxA, and also for structural stability (2). The C-terminal mutant is also highly conserved and is thought to reside on the outside of the protein possibly being important for intermolecular or intramolecular protein- protein interactions (2).

201 In this chapter, we use biochemical and biophysical studies to characterize the structure of CpxP at two different pH values representing the off (pH 5.8) and on (pH 8.0) state of the Cpx pathway. Using far UV circular dichroism (CD) experiments, we show that CpxP adopts a mainly a-helical structure at both pH values. Near UV CD demonstrates that CpxP may make a small structural change upon conversion from pH 5.8 to 8.0. This was confirmed by measuring the change in a-helical ellipticity at the two pH values. The melting temperature of the protein increases upon incubation in the pH 8.0 buffer suggesting the protein adopts a slightly more compact structure. Static light scattering experiments showed that CpxP forms a dimer in solution. Preliminary crystallization experiments with wild-type CpxP resulted in small block crystals with well defined edges. Limited proteolysis was used to remove flexible portions of CpxP for crystallization studies. Both the N- and C-terminal portions were digested by trypsin indicating disorder in these regions. The resulting CpxP 40-151 trypsin fragment contains the two conserved functional regions as determined by mutagenesis studies (2) and will be used for future crystallization experiments.

Results

Circular dichroism shows that CpxP adopts a mainly a-helical fold and may make a small structural rearrangement under alkaline pH conditions The conserved residues which affect the ability of CpxP to inhibit the Cpx pathway are predicted to reside on two conserved helices at the N- and C-terminus of the protein (2). We wanted to further characterize CpxP biochemically to learn more about its structure and function. Circular dichroism (CD) is a useful technique to obtain information about the secondary structure of an unknown protein (13). We purified a periplasmic wild-type CpxP-maltose binding protein (MBP) fusion protein, cleaved the N-terminal MBP tag and performed far- and near-UV CD spectroscopy on the purified protein. We performed the in vitro experiments at two pHs since an increase from pH 5.8 to 8.0 has been shown to induce the Cpx signal transduction pathway (2). Since CpxP is

202 degraded at pH 8.0, but not at 5.8, in the presence of the DegP protease, we wondered if CpxP may make a structural change at pH 8.0 which renders it more suitable for DegP proteolysis. Far-UV experiments (255- 190 nm) showed that wild-type CpxP has a characteristic a-helical spectrum with two minima located at 208 and 222 nm (Figure B-1A). The spectrum did not change upon a shift to a higher pH indicating that CpxP is folded similarly at both pH values. Using the DICHROWEB webserver (26) to deconvolute the spectra it was estimated that the percentage of a- helix is between 50-56% for pH 5.8 and 54-59% for pH 8.0. Near-UV experiments (320-255 nm) were used to probe for structural changes in the tertiary structure of CpxP at pH 5.8 and 8.0. In this range, CD signals for the individual aromatic amino acids can be seen at higher concentrations (1 mg/mL for this study). The signals are highly dependent on the number of each residue and the chemical environment they reside in within the structure (13). The spectrum of wild-type CpxP at each pH shows that it is folded into a stable structure with overlapping peaks between 260-305 nm consistent with signals from phenylalanine (255-270 nm), tyrosine (275-283 nm), and tryptophan residues (285-305 nm) (Figure B-1B). The majority of peaks do not change with increasing pH with the exception of the 297 nm minimum at pH 8.0 which is abolished at pH 5.8, and possibly the signal at 285 nm which may be decreased at pH 8.0. Both these peaks are in the tryptophan range showing that small rearrangements may be occurring within the wild-type CpxP tertiary structure upon a pH shift from 5.8 to 8.0. To test this further, the 220 nm a-helix CD signal of wild-type CpxP was monitored as a function of temperature. The loss of CD signal at higher temperatures indicates a loss of helicity and conversion to a less folded state. The melting temperature (Tm) was measured for each pH and the results were summarized in Figure B-1C. At pH 8.0 (open circles), the Tm of wild-type CpxP increases 3.3 ± 0.2°C showing that the protein adopts a slightly more stable structure. This is consistent with the near-UV CD experiments which show that this conversion may be due to a slight structural rearrangement within CpxP.

203 CpxP is a dimer at both pH 5.8 and 8.0 It has been previously thought that CpxP may form a homodimer since histidine kinases have been shown to dimerize within the cell (25). To test the oligomeric state of wild-type CpxP, we performed size- exclusion chromatography combined with multi-angle laser light scattering (MALLS) to obtain the molecular weight of the protein in solution at pH 5.8 and 8.0. The elution volume of CpxP on the gel filtration column was independent of pH (Figure B-2). Light scattering data determined that at both pH values, wild-type CpxP formed a dimer (pH 5.8, MW = 33771 ± 1222 Da; pH 8.0, MW = 34360 ± 505 Da). In-line dynamic light scattering data allowed for an estimation of the apparent radius of hydration (RH) for the CpxP dimer particles. At pH 5.8, the RHwas 2.9 ± 0.4 nm, whereas at pH 8.0, the RH was 3.0 ± 0.5 nm, indicating little change in the CpxP particle size within error. Thus, the light scattering data suggests that a shift to a more alkaline pH does not change the dimerization state of CpxP or size of the dimer particle in vitro.

Preliminary crystallization of wild-type CpxP We attempted to crystallize CpxP to gain high resolution structural information on the protein. Purified wild-type CpxP was put through commercial sparse matrix screens at 2.5-5 mg/mL concentrations using the hanging drop vapor diffusion method. The 5 mg/mL CpxP sample crystallized in 0.1 M phosphate-citrate pH 4.2, 40% v/v ethanol, and 5% w/v PEG 1000. The crystals took the form of small blocks with well defined edges (Figure B-3). Unfortunately, due to the presence of ethanol, the crystals were extremely unstable during harvesting. A grid screen was developed to try and lower the concentration of ethanol while increasing the concentration of the PEG 1000 precipitant. However, this screen did not produce crystals at the lower ethanol concentrations. Future experiments will attempt to find a stable pseudo-mother liquor in which the crystals can be quickly transferred to once the well has been opened. Perhaps increasing the drop size may help alleviate the harvesting problems. The other two conditions which produced crystals were: 0.1 M potassium thiocyanate and 30% w/v PEG monomethyl ether

204 (MME) 2000 and 0.15 M potassium bromide and 30% w/v PEG MME 2000. The two similar conditions gave the same type of oval shaped crystals with 2.5-5 mg/mL CpxP (Figure B-3). Attempts at optimizing the crystals with various potassium salts (KCI, Kl, KBr, and KSCN) at different concentrations did not improve the crystals. Therefore, the crystallization of CpxP appears to be feasible as nice looking crystals have been grown. However, different protein constructs should be tried to improve the crystals (see next section). Also, additional conditions need to be screened to ensure that all crystallization space is explored. Finally, the concentration of the CpxP sample in the crystallization trials could be increased to 10-20 mg/mL as the 2.5-5 mg/mL concentrations used in this study may have been too low.

Limited proteolysis of wild-type CpxP removes the flexible portions of the N- and C-terminus In order for a macromolecular crystal to diffract X-rays well, its molecules must be well ordered. This is often judged externally by examining the faces of the crystal which should have sharp edges and be free of defects (4). Disruptions in the regular packing of macromolecules can lead to increased mosaicity, a measure of how well molecules are ordered within the crystal (8). While crystals will never be perfect, they can be improved by removing flexible areas of the protein which are non­ essential to its function and may increase the mosaicity (16). Since wild-type CpxP crystallized in a few conditions, we decided to use limited proteolysis to remove areas of the protein which may be unstructured and hindering formation of well-ordered crystals. This technique was successful for generating well-diffracting crystals of FinO as the N-terminal 25 amino acids were found to be unstructured in the presence of its FinP SLII RNA substrate (6, 7). Wild-type CpxP was incubated with increasing amounts of trypsin, which cleaves specifically after positively charged residues such as lysine and arginine, for 60 minutes at 37°C. CpxP was digested to two bands at a weight ratio of 1/100 w/w trypsin:CpxP (Figure B-4A). Mass spectrometry determined the molecular weight of the two CpxP fragments to be 15990 and 13341 ±

205 5 Da. Analysis using the online program MS-Digest (University of California, San Francisco) showed that the larger fragment corresponded to the removal of the last 15 amino acids of CpxP. This fragment is cleaved further, removing the first 20 amino acids to the smaller fragment which is stable to further increases in trypsin and exists past eight hours under the reaction conditions above (data not shown). Figure B-4B shows the primary sequence of CpxP minus its N-terminal periplasmic signal sequence. The sequence of the wild-type CpxP used in the experiments starts at H20 (numbering consistent with Buelow et al. (2)). Therefore the two typsin fragments on the SDS-PAGE in Figure B-4A correspond to residues 20-151 and 40-151. We used three different online servers to predict the secondary structure of wild-type CpxP based on its primary sequence (12, 18, 19). All three servers claimed to have prediction accuracy past 75% using machine learning techniques based on sets of previously solved structures. Consistent with the CD results, all predictions indicate that CpxP is primarily a-helical (Figure B-4B). It can also be seen that the N- and C-termini are predicted to be unstructured. Two of the servers (Porter and SSpro) predict that the N-terminal a-helix starts near residue 40 and the C-terminal helix ends near residues 151-152. This corresponds well to our proteolytic mapping studies which produced the stable fragment 40-151. Both CpxP trypsin fragments contain the regions which are important for inhibition of the Cpx pathway: residues 55-61 and residue 128 (2) (Figure B-4B). The two fragments were subsequently cloned as MBP fusions2 and will be used for future crystallization studies. Hopefully, these constructs will help to improve the quality of the crystals, which were grown in the preliminary experiments using the wild-type construct.

2 The CpxP fragments were cloned by Daelynn Buelow from the lab of Dr. Tracy Raivio 206 Discussion

In this study, we sought to obtain structural and biochemical information about the bacterial envelope stress inhibitory protein, CpxP. Using far-UV CD studies we determined that CpxP is primarily a-helical at pH 5.8 and 8.0 representing the inactive and active states of the Cpx signal transduction pathway respectively. Near-UV studies showed that CpxP is a well-folded protein and may make a slight structural rearrangement upon shifting to a more alkaline pH. Two peaks in the spectrum, corresponding to the signals of tryptophan, undergo small changes indicating that the environment of the residues may change upon transfer to the pH 8.0 buffer. We determined that CpxP is a dimer at both pH values in vitro using size exclusion chromatography combined with MALLS. Dynamic light scattering showed that the radius of hydration of CpxP does not change much upon the pH change. Buelow et at. discovered that CpxP is a dimer in vivo by over-expressing CpxP and CpxP fusion proteins in E. coli cells and exposing the cells to formaldehyde thereby crosslinking proteins which are in close contact (Figure B-5A) (1). The dimer interaction was also shown by small angle x-ray scattering analysis (SAXS) of wild-type CpxP (1). Figure B-5B shows the low resolution molecular envelope of CpxP (pH 5.8) which was shown not to change significantly upon a pH increase from 5.8 to 8.0. The radius of gyration (Rg) of CpxP, which was calculated using two different methods, reproducibly decreased approximately 1 A from pH 5.8 to 8.0 indicating the CpxP dimer may make a small compaction in its structure (1). This finding was confirmed in the maximum dimension of the particle (Dmax), which decreased upon an increase to pH 8.0 (66 to 60 A) (1). Therefore, CpxP forms a dimer in the cell at both pH 5.8 and 8.0 showing that its degradation by DegP upon an increase in alkaline pH is not a result of a change in its oligomeric state. Although CpxP appears to make a slight structural rearrangement at pH 8.0, it remains to be tested whether this small change is important for the function of CpxP at the alkaline pH inducing cue.

207 The current model of the Cpx signal transduction pathway in response to envelope stress from alkaline extracellular conditions is shown in Figure B-6. At pH 5.8, the Cpx pathway is inactive due to the presence of the CpxP inhibitory protein, which is hypothesized to interact with periplasmic sensing domain of the CpxA histidine kinase protein, which spans the inner membrane of the cell (step 1). Upon an increase in pH, envelope-residing proteins will become misfolded or aggregated (Step 2). CpxP is then degraded in a DegP-dependent manner, either directly by the protease or through an interaction with a misfolded protein (Step 3) (9). DegP was found to reside upstream of CpxP in the Cpx pathway (1) and both proteins are required for activation of the pathway in response to an alkaline pH inducing cue (1). The absence of CpxP leads to the activation of CpxA, possibly through direct binding of misfolded proteins to the CpxA sensing domain or by sensing the levels of resident chaperone proteins (Step 4). After activation, the cytoplasmic kinase domain of CpxA phosphorylates the CpxR response regulator at a conserved aspartate residue (Step 5). Phosphorylation increases the binding of CpxR to consensus sequences in the cpx regulon, thereby activating genes coding for proteins, such as DegP, peptidyl-prolyl- isomerases, and disulfide oxidases needed to restore the bacterial envelope to its normal state (Step 6). The cpxA and cpxR genes are also activated to amplify the response. Once the envelope stress has been alleviated, CpxA uses its cytoplasmic phosphatase domain to dephosphorylate CpxR (Step 7). Activation of the cpxP gene is also needed to rapidly shut-off the pathway (1).

A recent structural investigation has shown that DegP appears to use an oligomerization mechanism to regulate its proteolysis and chaperoning functions (14) In the hexameric form DegP6, the active site can only be accessed by unfolded proteins and proteolytic cleavage is slow. However, formation of larger DegP12 and DegP24 assemblies stimulates protease activity 15-fold. Also the higher order complexes create a large pore, which in theory could house fully folded proteins up to 300 kDa in size (14). An example of a DegP chaperoning substrate is the outer membrane protein (OMP) class (e.g. OmpA) which form 3-barrel

208 structures. DegP has been shown to facilitate transport of OMPs through the periplasm to the outer membrane (14). This finding in conjunction with a previous study by Isaac et al. (9) supports a model where CpxP would be degraded by DegP through its association with misfolded proteins. Envelope stress would lead to the accumulation of misfolded proteins or aggregates. CpxP would be titrated away from CpxA by these proteins to the higher order DegP assembly where they would both be degraded. However, since this study showed that CpxP is stably folded at pH 5.8 and 8.0, it still remains undetermined how CpxP would present itself as a DegP substrate. To investigate the structure of CpxP in greater detail, we decided to attempt to crystallize the wild-type protein. We obtained three hits from one sparse matrix screen with the best looking crystals being small blocks with well defined edges. However, more optimization needs to be performed to increase the diffraction quality of the crystals. We used limited proteolysis with trypsin to remove flexible regions of CpxP to hopefully improve crystallization in the future. Trypsin cleaved wild-type CpxP to two fragments over the time course. The first fragment was missing the C-terminal 15 residues of CpxP (20-151) and the second fragment was further removed of both the C-terminal residues and the N- terminal 20 amino acids (40-151). Secondary structure predictions showed that the proteolysis likely cleaved off the N- and C-terminal unstructured regions. We cloned both fragments and will use them in future crystallization studies to hopefully improve the quality of the CpxP crystals.

Materials and Methods

Expression and purification of CpxP WT for biochemical assays and crystallization CpxP WT was over-expressed as an MBP fusion in the DB300 degP null strain (1). Cells containing the pMCP plasmid (22) were grown at 30°C in LB media with 2 g/L glucose and 100 )ig/mL ampicillin until they reached an O.D.600 of 0.7 at which point they were induced with 0.2

209 mM IPTG and grown overnight at 22°C. Cells were then harvested, resuspended and osmotically shocked to release the periplasmic fusion protein as per the pMAL protein fusion and purification system manual (New England Biolabs). The shock fluid was then applied to amylose resin (New England Biolabs) which was pre-equilibrated with wash buffer (25 mM Tris-HCI pH 7.5, 150 mM NaCI, 1 mM EDTA, 1 mM DTT). After loading, the column was washed with 10 volumes of wash buffer and the CpxP-MBP fusion protein was eluted with elution buffer (wash buffer + 10 mM final concentration maltose). The fusion was then buffer exchanged into 50 mM Tris-HCI pH 7.5, 150 mM NaCI, and 1 mM CaCI2, and concentrated to approximately 2 ml_ using an Amicon Ultra-15 10 K MWCO spin concentrator (Millipore, Fisher Scientific). The concentrated sample was quantified using the theoretical extinction coefficient at 280 nm of 78840 M"1cm"1 calculated using ProtParam from the primary sequence of CpxP WT-MBP (5). Factor Xa protease (GE Healthcare) was then added to 2.5% w/w ratio Factor Xa:CpxP-MBP and incubated for 16 h at 4°C. The cleavage reaction was monitored by SDS-PAGE and was stopped with phenylmethylsulfonyl fluoride (PMSF) (Sigma-Aldrich). The cleavage reaction was diluted in wash buffer and applied to the amylose resin to separate CpxP from MBP and residual fusion protein. The CpxP-containing flow-through was concentrated using an Amicon Ultra-15 5 K MWCO spin concentrator (Millipore, Fisher Scientific) and applied to a Superdex 75 26/60 gel filtration column (GE Healthcare) which was equilibrated in buffer containing 250 mM NaCI, 1 mM EDTA and 50 mM sodium phosphate (pH 5.8 or 8.0). The purified wild-type CpxP sample was subjected to MALDI-TOF mass spectrometry (University of Alberta Chemistry Mass Spectrometry Facility), where it was found to have an average m/z of 17336 + 5 Da, corresponding to the protein having lost two C-terminal amino acids (S165 and Q166) (Figure B-4B). Amino acid analysis (Alberta Peptide Institute) was used to calculate an experimental extinction coefficient at 280 nm of 12214 M" 1cm"1 which was subsequently used for protein quantifications.

210 Size-exclusion chromatography with Multi-angle Laser Light Scattering (MALLS)3 50 |j.L of a 2 mg/mL wild-type CpxP sample was injected at 0.5 mL/minute onto a Superose 6 HR 10/300 gel filtration column (GE Healthcare) pre-equilibrated with 50 mM sodium phosphate (pH 5.8 or 8.0), 250 mM NaCI, and 1 mM EDTA. The effluent was directly passed over an in-line DAWN EOS MALLS, an Optilab rEX differential refractive index detector and an in-line Wyatt quasi-elastic light scattering (QELS) instrument (Wyatt Technologies, Santa Barbara, CA). Light scattering data was processed using the ASTRA version 4.90 software. Averages of molecular weights and radii of hydration (RH) were calculated from the elution peak (15) A minimum of two runs per pH were collected from which the average and standard deviation were determined.

Circular dichroism CD experiments were performed on a Jasco J720 spectropolarimeter. Far-UV (255-190 nm) experiments used 0.25 mg/mL wild-type CpxP samples in a thermostated fused silica cell with a path length of 0.05 cm at 24°C. For secondary structure analysis, spectra were collected with CpxP samples in 50 mM sodium phosphate pH (5.8 or 8.0) to reduce unwanted signal contributions from NaCI and EDTA at wavelengths below 200 nm. Spectra in this buffer were identical to spectra from the 250 mM NaCI and 1 mM EDTA buffer. Raw ellipticities were subtracted from the buffer and converted into mean residue ellipticities using an average amino acid weight of 117.1 Da, calculated from the wild-type CpxP primary sequence. Secondary structure predictions were performed using the DICHROWEB webserver (26). Several algorithms with reference datasets 4 and 7 were compared and those which gave a normalized root mean square deviation (NRMSD) < 0.1 were used to calculate the fractions of secondary structure. These

3 We would like to thank Dr. Paul Scott (University of Alberta, Department of Biochemistry) for his assistance with the MALLS experiments and helpful discussions 211 algorithms were CONTIN/LL (20) and CDSSTR (10). For melting temperature (Tm) experiments, ellipticity was recorded at 220 nm for 5 minutes at each temperature, measuring a reading every second. From the data, an average ellipticity at each temperature was calculated. Wild- type CpxP samples were in 250 mM NaCI, 1 mM EDTA, and 50 mM sodium phosphate (pH 5.8 or 8.0), and were at a concentration of 0.25 mg/mL in a 0.05 cm silica cell. The samples were equilibrated for 5 minutes at each temperature before data collection. Temperatures ranged from 24 to 62°C, employing a Lauda water bath (Brinkmann Instruments) to control the temperature of the cell. The loss of ellipticity at 220 nm indicates unfolding of CpxP and Tm is defined as being the temperature at which the ellipticity at 220 nm decreases by half. Melting temperatures were calculated using the SigmaPlot 2001 software (SPSS

Inc.). Tm values represent an average of two experiments. Near-UV (320-255 nm) experiments used 1 mg/mL samples of CpxP WT in a 1 cm path length thermostated fused silica cell at 24°C. Wild-type CpxP samples were incubated in 50 mM sodium phospate pH (5.8 or 8.0), 250 mM NaCI, 1 mM EDTA. Ellipticity data was converted to mean residue ellipticity as described above. Both far-UV and near-UV spectra were an average of 12 scans.

Crystallization of wild-type CpxP Wild-type CpxP protein solutions, in 50 mM sodium phosphate pH 7.0, 250 mM NaCI, and 1 mM EDTA and at 2.5 and 5.0 mg/mL concentrations, were put through a number of commercial crystallization screens including: the JCSG+ Suite (Qiagen) and PACT suite (Qiagen). Hanging drop crystallization setups were performed with 0.5 mL of crystallization solution in the well. 1 )j.L of protein was mixed with 1 \iL of well solution on a siliconized glass coverslip which was applied to the top of the well and sealed with vacuum grease to initiate the vapour diffusion process. All trays were setup at room temperature.

212 Limited proteolysis of wild-type CpxP 5 |ag of purified wild-type CpxP in 50 mM sodium phosphate pH 7.0, 250 mM NaCI, and 1 mM EDTA was digested with 5 x10"4, 0.001, 0.005, 0.01, 0.05, and 0.1 |ag of trypsin (Sigma-Aldrich) for 60 minutes at 37°C. Reactions were stopped with PMSF and the reaction products were separated on an 18% SDS-PAGE gel. MALDI-TOF mass spectrometry (University of Alberta Chemistry Mass Spectrometry Facility) was used to identify the molecular weights of the two digestion products from a parallel trypsin digest of CpxP at a 1/100 w/w ratio of trypsin:CpxP. The two fragments had average m/z of 15990 and 13336 ± 5 Da corresponding to residues 20-151 and 40-151, respectively (Figure B-4B).

213 210 220 230 240 Wavelength (nm)

50' pHS.8 40' _ pH8.0 m u w 30

20-

10-

Phe 1 10- r-Tyr-, •^^ i— — Trp - 70- 280 290 300 Wavelength (nm)

tuuu i i i i i i i i i i i i i i i i 1 1 1 1 't'"!'"!""! r'T" ^E_ c . o -6000 - CM Ci -8000 - i.9 ip t " 48.3 ± 0.2 / / -10000 - El l a> 3 1 -12000 - / m„&± ©^ QS: • -14000- S %r j^ 1 JS^O^^^ -16000 - • • • • • .••'•_••'•.''•• • •»• i .'•••_• * 30 35 40 45 50 55 60 Temperature (°C)

Figure B-1: Wild-type CpxP is mainly ct-helical and may make a slight structural rearrangement at alkaline pH

A) Far-UV spectrum of CpxP in 50 mM sodium phosphate pH 5.8 (solid) and 8.0 (dashed). B) Near-UV spectrum of CpxP in 250 mM NaCI, 1 mM EDTA, and 50 mM sodium phosphate pH 5.8 (black) or 8.0 (grey). The wavelength ranges for the signals of Phe, Tyr and Trp are indicated at the bottom. The far and near UV CD experiments were repeated in duplicate and representative spectra are shown. C) CpxP melting experiment at pH 5.8 (•) and 8.0 (o). Samples were in 250 mM NaCI, 1 mM EDTA and 50 mM sodium phosphate pH 5.8 or 8.0. Protein unfolding was tracked by measuring the mean residue ellipticity (MRE) at 220 nm (a-helix) at temperatures between 30 and 62°C. The resulting melting temperature (Tm) of CpxP at pH 5.8 and 8.0 is shown by black and open lettering respectively beside the curve. Each MRE220nm point on the graph is an average of 300 CD measurements (see Materials and Methods). In addition, each pH experiment was repeated twice. 214 0.06

17.5 18 Volume (mL)

Figure B-2: CpxP is a dirtier at pH 5.8 and 8.0.

Size-exclusion chromatography and multi-angle laser light scattering determination of molecular weight of wild-type CpxP at pH 5.8 and 8.0. Light scattering over the elution profile of CpxP shown in solid (pH 5.8) and dashed (pH 8.0) lines. Superimposed across the peaks in filled (pH 5.8) and empty (pH 8.0) circles indicate molecular weights as a function of elution volume.

215 0.1 M phosphate-citrate pH 4.2 0.1 M KSCN 40% v/v ethanol 30% w/v PEG MME 2000 5%w/vPEG 1000

Figure B-3: Preliminary crystallization of wild-type CpxP

Crystallization drop photos of 5 mg/mL wild-type CpxP samples. The conditions used to crystallize the protein are noted below the images.

216 A / Trypsin M ^ —^^^ H 20 • 17336(20-164) 15 -.«.rfi ii|ijw: *£$* %'•' «s; . •15990(20-151) •13336(40-151) 10 ••

KDa B

Porter 1 I I I YASSPF I I SSpro EEE 1 1 I I 20 30 40 50 GO ISEFH AAEVGJ;GDN W HPGEELTQRS TQSHI\ dFDGIS LTEHQRQQMR P TQ

l l 1 1 I I 1 1 1 1 i i i i 1 11 i 70 80 90 100 DLMQQARHEQ PPVNVSELET MHRLVTAENF DENAVRAQAE Ui >

1 1 l I i l l I i 1 l l 110 120 130 140 KMANEQIARQ VEMAKVRNQM YRLLTPEQQA VLNEKHQQRM H l I 1 1 I I i Zl DEE 150 160 EQLRDVTQWQ KSSSLKLLSS SNSRSQ

Figure B-4: Limited trypsin digest of wild-type CpxP removes flexible areas of protein

A) Limited proteolysis of wild-type CpxP with trypsin. Reactions contained 5 ng of purified CpxP with 5 xlO"4, 0.001, 0.005, 0.01, 0.05, and 0.1 ng of trypsin. The molecular weights and corresponding residues of full-length CpxP (17336 Da; residues 15-164) and its digested fragments (15590 Da; residues 20-151 and 13336 Da; residues 40-151) are shown on the right as determined by MALDI-TOF mass spectrometry. B) Schematic of the wild-type CpxP primary sequence (residues 15-166) with amino acid numbering from (2). Shown in bold lettering is the sequence which remained from the cloning of the wild- type construct. Shown in gray at the C-terminus are two residues which get cleaved during protein purification (as determined by mass spectrometry). The mutations important for the inhibition function are shown in bold (Q55P, M59T, R60Q, D61E, D61V, and Q128H) (2). The results from three secondary structure prediction programs are shown above the sequence. Rectangular blocks represent cc-helices and E represents extended (P) strand. The sequence which is underlined in black highlights the smaller truncation fragment (residues 40-151) and the black underline plus the dotted underline represents the larger truncation fragment (residues 20-151). 217 214 A 214 — - CpxP-MBP dimer 118 — 118 • MBP dimer 92 — 92 — • CpxP-MBP monomer • MBP monomer 52.2 —— *Mfc 52.2 —

35.7 — 35.7 — -*- CpxP dimer

28.9 28.9

20.8 — 20.8 — ^ -*- CpxP monomer **»**»». •***» Lane: 1 Lane: 1

B

Figure B-5: SAXS and in vivo chemical cross-linking confirms that CpxP is a dimer

A) In vivo cross-linking shows that over-expressed CpxP and a CpxP-MBP fusion form a dimer within the cell (1). Formaldehyde was added to E. coli cell cultures to cross-link the proteins. B) SAXS molecular envelope of wild-type CpxP at pH 5.8 (1). The envelope is rotated 90° about the axis shown to highlight its shape.

4 SAXS experiments and analysis were performed by Ross A. Edwards and formaldehyde cross-linking was performed by Daelynn Buelow. 218 Cpx regulon

Figure B-6: Model of alkaline pH induced activation of the Cpx signal transduction pathway

1) In the absence of envelope stress, the CpxA histidine autokinase is inhibited by the CpxP dimer. 2) Upon a shift to pH 8.0, envelope stress is seen likely due to protein misfolding events and protein aggregation. 3) CpxP is removed from CpxA in a DegP- dependent manner. Either CpxP and misfolded proteins are degraded directly by the protease or CpxP is degraded through its association with misfolded proteins. 4) The absence of CpxP allows the sensor domain of CpxA to respond to the alkaline pH cue possibly through binding of a misfolded protein directly or through a resident chaperone protein (square). CpxA is now activated (star) and autophosphorylates itself. 5) Activated CpxA uses its kinase domain to transfer a phosphate to the CpxR response regulator. 6) This activates the transcription factor increasing its DNA binding to consensus sequences in the Cpx regulon. A number of genes are activated which help to restore the bacterial envelope to its normal state. Activation of the gene coding for CpxR ensures amplification of the response through a positive feedback loop. 7) Once the stress is relieved, CpxA dephosphorylates CpxR and the CpxP which was expressed upon CpxR activation helps to shut down the pathway by binding to the periplasmic CpxA sensor domain.

219 References

1. Buelow, D.R., Arthur, D.C., Edwards, R.A., Glover, J.N., and Raivio, T.L. (2007) CpxP is an a-helical dimer that fine-tunes turn- on and shut-off of the Cpx envelope stress response. Submitted,

2. Buelow, D.R. and Raivio, T.L. (2005) Cpx signal transduction is influenced by a conserved N-terminal domain in the novel inhibitor CpxP and the periplasmic protease DegP. J Bacteriol, 187: 6622- 6630.

3. Danese, P.N. and Silhavy, T.J. (1998) CpxP, a stress-combative member of the Cpx regulon. J Bacteriol, 180: 831-839.

4. Drenth, J., Principles of protein x-ray crystallography. 2nd ed. Springer advanced texts in chemistry, C.R. Cantor. 1999, New York: Springer-Verlag. p 11.

5. Gasteiger, E., Hoogland, C, Gattiker, A., Duvaud, S., Wilkins, M.R., Appel, R.D., and Bairoch, A., Protein Identification and Analysis Tools on the ExPASy Server in The Proteomics Protocols Handbook, J.M. Walker. 2005, Humana Press: Totowa, N.J. p. 571-607.

6. Ghetu, A.F., Gubbins, M.J., Frost, L.S., and Glover, J.N. (2000) Crystal structure of the bacterial conjugation repressor FinO. Nat Struct Biol, 7: 565-569.

7. Ghetu, A.F., Gubbins, M.J., Oikawa, K., Kay, CM., Frost, L.S., and Glover, J.N. (1999) The FinO repressor of bacterial conjugation contains two RNA binding regions. Biochemistry, 38: 14036-14044.

8. Gilliland, G.L. and Ladner, J.E. (1996) Crystallization of biological macromolecules for X-ray diffraction studies. Curr Opin Struct Biol, 6: 595-603.

9. Isaac, D.D., Pinkner, J.S., Hultgren, S.J., and Silhavy, T.J. (2005) The extracytoplasmic adaptor protein CpxP is degraded with substrate by DegP. Proc Natl Acad SciUSA,W2: 17775-17779.

10. Johnson, W.C. (1999) Analyzing protein circular dichroism spectra for accurate secondary structures. Proteins, 35: 307-312.

11. Jones, C.H., Danese, P.N., Pinkner, J.S., Silhavy, T.J., and Hultgren, S.J. (1997) The chaperone-assisted membrane release and folding pathway is sensed by two signal transduction systems. EMBOJ, 16: 6394-6406.

220 12. Karypis, G. (2006) YASSPP: better kernels and coding schemes lead to improvements in protein secondary structure prediction. Proteins, 64: 575-586.

13. Kelly, S.M., Jess, T.J., and Price, N.C. (2005) How to study proteins by circular dichroism. Biochim Biophys Acta, 1751: 119- 139.

14. Krojer, T., Sawa, J., Schafer, E., Saibil, H.R., Ehrmann, M., and Clausen, T. (2008) Structural basis for the regulated protease and chaperone function of DegP. Nature, 453: 885-890.

15. Lu, J., Edwards, R.A., Wong, J.J., Manchak, J., Scott, P.G., Frost, L.S., and Glover, J.N. (2006) Protonation-mediated structural flexibility in the F conjugation regulatory protein, TraM. EMBO J, 25: 2930-2939.

16. McPherson, A., Malkin, A.J., and Kuznetsov, Y.G. (1995) The science of macromolecular crystallization. Structure, 3: 759-768.

17. Nevesinjac, A.Z. and Raivio, T.L. (2005) The Cpx envelope stress response affects expression of the type IV bundle-forming pili of enteropathogenic Escherichia coli. J Bacteriol, 187: 672-686.

18. Pollastri, G. and McLysaght, A. (2005) Porter: a new, accurate server for protein secondary structure prediction. Bioinformatics, 21: 1719-1720.

19. Pollastri, G., Przybylski, D., Rost, B., and Baldi, P. (2002) Improving the prediction of protein secondary structure in three and eight classes using recurrent neural networks and profiles. Proteins, 47: 228-235.

20. Provencher, S.W. and Glockner, J. (1981) Estimation of globular protein secondary structure from circular dichroism. Biochemistry, 20: 33-37.

21. Raivio, T.L. (2005) Envelope stress responses and Gram-negative bacterial pathogenesis. Mol Microbiol, 56: 1119-1128.

22. Raivio, T.L., Popkin, D.L., and Silhavy, T.J. (1999) The Cpx envelope stress response is controlled by amplification and feedback inhibition. J Bacteriol, 181: 5263-5272.

23. Raivio, T.L. and Silhavy, T.J. (1999) The sigmaE and Cpx regulatory pathways: overlapping but distinct envelope stress responses. CurrOpin Microbiol, 2: 159-165.

221 24. Snyder, W.B., Davis, L.J., Danese, P.N., Cosma, C.L., and Silhavy, T.J. (1995) Overproduction of NIpE, a new outer membrane lipoprotein, suppresses the toxicity of periplasmic LacZ by activation of the Cpx signal transduction pathway. J Bacteriol, 177: 4216-4223.

25. Surette, M.G., Levit, M., Liu, Y., Lukat, G., Ninfa, E.G., Ninfa, A., and Stock, J.B. (1996) Dimerization is required for the activity of the protein histidine kinase CheA that mediates signal transduction in bacterial chemotaxis. J Biol Chem, 271: 939-945.

26. Whitmore, L. and Wallace, B.A. (2004) DICHROWEB, an online server for protein secondary structure analyses from circular dichroism spectroscopic data. Nucleic Acids Res, 32: W668-673.

222