Structure, Function, and Regulation of Multidrug Export Proteins Among the RND Superfamily in Gram-Negative Bacteria Mathew David Routh Iowa State University

Home , Dissociation constant, Rnd (GTPase)

Iowa State University Capstones, Theses and Graduate Theses and Dissertations Dissertations

2010 Structure, function, and regulation of multidrug export proteins among the RND superfamily in Gram-negative bacteria Mathew David Routh Iowa State University

Follow this and additional works at: https://lib.dr.iastate.edu/etd Part of the Physics Commons

Recommended Citation Routh, Mathew David, "Structure, function, and regulation of multidrug export proteins among the RND superfamily in Gram- negative bacteria" (2010). Graduate Theses and Dissertations. 11401. https://lib.dr.iastate.edu/etd/11401

This Dissertation is brought to you for free and open access by the Iowa State University Capstones, Theses and Dissertations at Iowa State University Digital Repository. It has been accepted for inclusion in Graduate Theses and Dissertations by an authorized administrator of Iowa State University Digital Repository. For more information, please contact [email protected].

Structure, function, and regulation of multidrug export proteins among the RND superfamily in Gram-negative bacteria

Mathew David Routh

A dissertation submitted to the graduate faculty

in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

Major: Molecular, Cellular and Developmental Biology

Program of Study Committee: Edward Yu, Major Professor Drena Dobbs Kai-Ming Ho Michael Shogren-Knaak Jeff Beethem

Iowa State University

Ames, Iowa

2010

TABLE OF CONTENTS

ABSTRACT vi

CHAPTER 1. GENERAL INTRODUCTION 1 DISSERTATION ORGANIZATION 1 RESEARCH AIMS AND SIGNIFICANCE 3 Aim 1: Understanding the structure and molecular mechanism of the multidrug transport protein AcrD. 5 Aim 2: Providing insights to how MDR pumps are locally regulated by the TetR family of proteins. 5 FIGURES AND TABLES 7 REFERENCES 8

CHAPTER 2. LITERATURE REVIEW: RND TRANSPORTERS 11 INTRODUCTION 11 BACKGROUND 13 RND FUNCTION 16 SUBSTRATE CAPTURE AND TRANSPORT 20 Drug entry: 21 Periplasmic drug binding: 22 STRUCTURAL STUDIES OF AcrB 24 TM domain : 25 Periplasmic region: 26 TolC docking domain: 26 Porter domain: 26 ASYMMETRIC AcrB 29 MexB STRUCTURE 30 EVIDENCE TO SUPPORT THE ASYMMETRIC MODEL 32 TolC STRUCTURE 33 iii

AcrA STRUCTURE 34 AcrA/AcrB/TolC COMPLEX 36 OTHER RND PROTEINS 37 INHIBITION 41 OTHER FUNCTIONS 43 CONCLUDING REMARKS 44 FIGURES AND TABLES 46 REFERENCES 59

CHAPTER 3. STRUCTURAL AND FUNCTIONAL STUDIES OF AcrD 73 ABSTRACT 73 INTRODUCTION 74 PROTEIN PRODUCTION 75 Expression 75 Purification 75 Detergent exchange 76 CRYSTALLIZATION 77 Detergents 78 DATA COLLECTION 79 AcrD-LIGAND CRYSTALLIZATION 79 FLUORESCENT POLARIZATION 80 ISOTHERMAL TITRATION CALORIMETRY (ITC) 81 STD-NMR 83 STRUCTURAL MODELING 84 DISCUSSION 86 FIGURES AND TABLES 89 REFERENCES 110

CHAPTER 4. TetR REVIEW 114 ABSTRACT 114 ARTICLE OUTLINE 115 iv

INTRODUCTION 115 THE AcrR REGULATOR 119

Crystal structure of AcrR in space group of P222 1 121

Crystal structure of AcrR in space group of P31 123 Crystallization of AcrR-IR 125 THE CmeR REGULATOR 126 Crystal structure of CmeR 128 CONCLUSIONS AND PERSPECTIVES 131 ACKNOWLEDGEMENTS 135 FIGURES AND TABLES 136 REFERENCES 147

CHAPTER 5. CRYSTAL STRUCTURES OF CmeR-BILE ACID COMPLEXES FROM CAMPYLOBACTER JEJUNI 154 ABSTRACT 154 INTRODUCTION 155 MATERIALS AND METHODS 157 Preparation and crystallization of the CmeR-ligand Complexes 157 X-ray data collection, processing, and structural Refinement 158 Polarization 158 RESULTS 159 Structure of the CmeR-Taurocholate complex 160 Structure of the CmeR-Cholate complex 161 DISCUSSION 162 ACCESSION NUMBERS 167 ACKNOWLEDGEMENTS 167 FIGURES AND TABLES 168 REFERENCES 179

CHAPTER 6. CONCLUDING REMARKS 183 RND TRANSPORTERS AND AcrD 183 TetR REGULATORS, AcrR AND CmeR 186 REFERENCES 189

ABSTRACT

Multidrug binding proteins are able to recognize structurally unrelated compounds. These proteins play a crucial role as drug resistant export proteins in both eukaryotic and prokaryotic cells. Furthermore, multidrug binding proteins exist that regulate transcription of multidrug transporters and play an important role in responding to incoming toxins. To study the function of multidrug binding proteins and exporters, AcrD of Escherichia coli (E. coli), CmeR of Campylobacter jejuni , and AcrR of E. coli were chosen as models. AcrD, a multidrug efflux pump, functions to rid the bacterial cell of incoming hydrophilic substrates by harboring a diverse ligand binding cavity capable of recognizing the deleterious compounds. By using fluorescence polarization and isothermal titration calorimetry, we show that AcrD specifically interacts with aminoglycosides and anthracyclines with dissociation constants in the low micromolar range. By modeling the AcrD structure, hydrophilic patches were identified that could allow substrates to be captured from the cytoplasm or periplams and exported completely out of the bacterial cell. The transcriptional regulators CmeR and AcrR interact with inducing ligands by utilizing a similar mechanism, whereby a diverse binding pocket exists allowing recognition of various substrates. This work shows how CmeR recognizes large, negatively charged bile salts by harboring a large hydrophobic surface with appropriately spaced polar residues to stabilize bile acid binding. Furthermore, by comparing the structures of AcrR and CmeR, a model can be developed to describe transcriptional regulation. Upon ligand interaction in the C-terminal domain, adjacent N-terminal domains separate, breaking bonds with the DNA operator, thus releasing repression. Interestingly, kinetics studies reveal that AcrR, CmeR, and AcrD interact with ligands in the low micromolar range, which may be a critical feature of mutlidrug binding proteins.

CHAPTER 1. GENERAL INTRODUCTION DISSERTATION ORGANIZATION

This dissertation contains six chapters. Chapter 1 is a general introduction to the dissertation organization and the research aims and significance of the studies. The two research aims stated in Chapter 1 are specifically addressed in Chapters 3-5. Chapter 2 is a literature review of the resistance, nodulation, and cell division (RND) superfamily of transporters. First, this review takes a look at the history and highlights specific experiments that have provided novel insights into the function of RND proteins. In particular, studies focused on the well-studied RND transporters AcrB of E. coli and MexB of Pseudomonas aeruginosa (P. aeruginosa) are discussed. Recently obtained X-ray crystal structures of these two proteins are then closely examined. Furthermore, the important roles other members of the RND family play in the life cycle of Gram-negative bacteria are closely examined. Finally, the review highlights advances made in the development of efflux pump inhibitors (EPIs) and the implications of these molecules. Chapter 3 addresses the first research aim of understanding the structure and function of RND transporters. In the first research aim, a crystallization protocol is developed to delve into the structure of the RND transporter AcrD of E. coli . Furthermore, biochemical and biophysical tools are used to understand the structure-function relationship of AcrD. As the crystal structure yet eludes us, structural and binding site predictions are used to validate the available biochemical data and enhance the understanding of multiligand interaction. Multidrug binding proteins similar to AcrD have been shown to harbor large binding cavities capable of interacting with structurally diverse ligands. As AcrD is able to recognize a wide array of antibacterial compounds, including aminoglycosides, sodium dodecyl sulfate, tetracyclines, and bile salts, the docking predictions aid in providing valuable insights into the diverse ligand binding sites. To this point, AcrD harbors a large multifaceted uniquely polar binding cleft. It is suggested that the wide range of polar 2

residues are employed to interact with structurally dissimilar hydrophilic ligands. Furthermore, we provide initial evidence that AcrD is able to recognize the anthracycline class of antibiotics through fluorescence polarization experiments using with the substrate daunorubicin HCl (DR). The binding assays suggest that DR and gentamicin bind with similar affinity to AcrD, whereby dissociation constants of 6 µM and 3 µM are observed, respectively. The 4 th Chapter is a review article published in the journal BBA Proteins and Proteomics (1). This review is a compilation of research performed in our lab on the TetR family of transcriptional repressors. Specifically, the repressors AcrR and CmeR of E. coli and C. jejuni , respectively, are discussed. Additional work is included in the text describing recent studies on the transcriptional regulators AcrR and CmeR. The work focuses on structural and functional characterization of these proteins. Contributions to this work are described as follows: Mathew Routh collected binding data, peformed crystallization experiments on the AcrR-IR complex, and wrote the manuscript. Chih-Chia Su performed modeling studies, collected data, and performed crystallization studies. Qijing Zhang prepared protein samples for crystallization and kinetic studies, and Edward Yu designed experiments and assisted in writing the manuscript. The research presented in Chapter 5 is in preparation for submission to the Journal of Biochemistry . This work describes the crystallization of CmeR in complex with the bile acids cholate and taurocholate, and identifies a novel, distinct binding site within CmeR. Contributions made to this work include the following: Mathew Routh resolved the structure, identified binding site, analyzed data, and wrote manuscript. Zhangqi Shen prepared CmeR protein for crystallization. Jack Su performed much of the crystallization experiments. Qijing Zhang aided in experimental design and edited the manuscript, and Edward W. Yu assisted in the experimental design, data analysis, and manuscript writing. Chapter 6 is the general conclusion of the studies addressed in the thesis. This chapter summarized the results from the structural and functional studies on AcrD that facilitated a greater understanding of RND transporters and how they are 3

able to recognize a diverse array of ligands. Furthermore, the chapter concludes the work discussed in Chapters 4 and 5 on the TetR family of repressors. Through the work in these chapters, a model can be suggested to show how these proteins regulate transcritption and allow rapid induction upon interaction with structurally dissimilar compounds. Other topics in this chapter include future directions and potential applications deriving from the insights of this work.

RESEARCH AIMS AND SIGNIFICANCE

Multidrug resistance has been called one of the world’s most pressing public health problems. For epidemiological purposes, multidrug resistant (MDR) bacteria are defined as bacteria able to resist one or more classes of antimicrobials (2). For instance, Methicillin-resistant Staphylococcus aureus (MRSA) and vancomycin- resistant enterococci (VRE) are resistant to most available antibacterial drugs and receive special attention in healthcare facilities (3). Usually, MDR infections have similar clinical manifestations as those caused by susceptible pathogens. However, treating the infections becomes increasingly difficult in resistant microorganisms. For example, vancomycin is one of the few effective treatments for MRSA and is considered the last line of defense against many resistant pathogens (4, 5). Although new antimicrobials are becoming available, resistance to each new agent has already emerged in clinical isolates (6-10). Increased mortality has recently been associated with Salmonella typhimurium infections by linking data from the Danish Surveillance Registry for Enteric Pathogens with the Civil Registration System and the Danish National Discharge Registry (11). This study suggests that patients infected with S. typhimurium infections resistant to ampicillin, chloramphenicol, streptomycin, sulfonamide, and tetracycline had a 2-year death rate 4.8 times higher, whereas additional quinolone resistance increased the mortality rate to 10.3 times the general population. Comparatively, patients infected with drug susceptible strains were only 2.3 times more likely to die 2 years after infection than people in the general Danish 4

population. This study highlights the necessity to increase understanding of the fundamental mechanisms associated with drug resistance. Drug resistant bacterial infections were first reported shortly after the initial mass production of antibiotics in the 1940’s. In fact, a report published in 1947 found that, of 100 staphylococcus infections tested, 38 were classified as highly resistant to penicillin ( 12). This initial resistance was primarily associated with individual enzymes inactivating specific antibiotics, such as β-lactamases on penicillin. As novel antibiotics were implemented to combat resistant pathogens, selective pressure led to fundamentally new methods of drug resistance. Currently, there are roughly three major mechanisms utilized by bacteria to evade the toxic effects of biocidal agents, including enzymatic modification (Fig 1A), target alteration (Fig 1B), and reduced uptake due to the presence of drug efflux pumps (Fig 1D) or a decrease in porin expression (Fig 1C). Enzymatic modification involves two classes of enzymes, including those that degrade specific antibiotics ( 13) and enzymes that chemically modify the antibacterial compound ( 14), thus decreasing drug toxicity. The second mechanism employed by bacteria is alteration of the drug target. Nearly all relevant fluoroquinolone resistance has been attributed to target alteration, whereby specific mutations inhibit the drugs interaction to DNA gyrase and topoisomerase IV ( 15). Although highly effective, these mechanisms are limited by inhibiting only specific classes of antibiotics. A more critical issue is the broad spectrum of antibiotic resistance associated with multidrug efflux pumps. Ubiquitous in most living cells, multidrug efflux transporters have gained recognition as the major contributor to drug resistance observed in many pathogenic microorganisms (16, 17). These transporters are capable of capturing and exporting toxic compounds before they reach their cellular target, thereby decreasing the effectiveness of the drug ( 18, 19). The first drug transporter identified, TetA, in 1980 (20 ), conferred tetracycline resistance to its host. Following this breakthrough, ensuing discoveries identified proteins capable of eliminating various structurally unrelated compounds. 5

Recently, tremendous strides have been made to understand the mechanisms that govern RND transporter function. These works have focused on both the transporters as well as the regulatory networks that control their expression. To provide novel insights into multidrug resistance and the regulation of the associated export proteins, two specific aims have been pursued in this work. Aim 1: Understanding the structure and molecular mechanism of the multidrug transport protein AcrD of E. coli . Elucidating the three-dimensional structure of AcrD will provide fundamental insights to the mechanism of drug efflux. Furthermore, the protein structure is a pivotal tool in the design of effective efflux pump inhibitors. Based on the homologous protein AcrB and previously obtained biochemical evidence on AcrD, we hypothesize that AcrD utilizes the proton-motive-force to transition through three functional states (access, binding, and extrusion) to export substrates. It has been suggested that RND transport proteins harbor a multifaceted binding cleft capable of interacting with a diverse array of potential substrates (21-25). The studies in chapter 3 help to delineate how AcrD is able to recognize structurally dissimilar ligands. Additionally, initial clues to the AcrD transport mechanism are illustrated through the homology modeling experiments in the same chapter. Aim 2: Providing insights to how MDR pumps are locally regulated by the TetR family of proteins. The soluble regulators that govern multidrug efflux pump exression provided the initial clues to how multidrug binding proteins are able to recognize such dissimilar agents. Crystallization experiments on QacR of Staphylococcus auerius first showed the multifaceted nature of the binding cavity, thus illustrating the importance of this feature (26). Functionally, these proteins regulate transcription by interacting with the promoter sequence upstream of the open reading frame, thereby inhibiting polymerase initiation (27). It seems the N-terminal domain of each regulator interacts with consecutive major grooves of the DNA. We hypothesize that ligand interaction to the C-terminal domain of CmeR of C. jujuni and AcrR of E. coli initiates structural changes in the N-terminal domain forcing protein/DNA separation. 6

The work performed in Chapters 4 and 5 support this hypothesis by showing that, upon ligand interaction, the distance between N-terminal domains separate to ~54 Å in CmeR, although in unliganded AcrR, this distance is 37 Å. A docking study showed that the hydrophobic tunnel of CmeR should be able to accommodate large, negatively charged bile acid molecules, such as taurocholate and cholate ( 28). We hypothesized that bile acids should anchor to a mostly hydrophobic surface with appropriately positioned polar and positively charged residues. Various hydrophobic residues are located throughout the binding cavity of CmeR, including H72, F99, F103, F137, S138, Y139, V163, C166, T167, K170 and H174. The anionic bile acids suggested length would span nearly the entire ligand- binding tunnel of the regulator, respectively. The large tunnel, possibly consisting of multiple mini-pockets that may be employed to interact with different ligands, is rich in aromatic residues and contains three positively charged amino acids (two histidines and one lysine). We previously predicted that these positively charged residues would be crucial for CmeR to recognize negatively charged ligands. The work in chapter 5 is based on this hyposthesis and provides clues to how CmeR is able to recognize both anionic and neutral ligand by employing various charged and polar residues.

FIGURES AND TABLES Figure 1

A B

C D

Figure 1 is a schematic diagram representing the major mechanisms bacteria utilize to resist drug toxicity. (A) Enzymatic breakdown of drug: the scissors represent inactivating enzymes while the red barbell shapes are active drugs and half barbells are inactive. (B) Target site mutation: single point mutations in the drug’s target site can reduce affinity making the drug ineffective. The target is represented in green and the drug is the red barbell. (C) Decrease of porin expression: Many drugs poorly spontaneously diffuse across the membrane and therefore must utilize porin molecules to enter the cell. Mutations altering porin expression levels can reduce accumulation of drugs within the cell. The porin molecules are colored green and drugs are the red barbell. (D) Active efflux: As drugs enter the cell, drug exporters reduce the intracellular concentration of drug. The blue, green, and orange molecules are a schematic representation of a inner membrane transporter, membrane fusion protein, and outer membrane channel of an RND efflux complex, respectively. The drugs are again represented as red barbells.

REFERENCES

1. Routh, M. D., Su, C.-C., Zhang, Q., and Yu, E. W. (2009) Structures of CmeR and AcrR: Insight into the mechanisms of transcriptional repression and multidrug recognition in the TetR family of regulators. Biochim. Biophys. Acta 1794 , 844-851.

2. IOM (1998), eds. Harrison, P. F. & Lederberg, J. (National Academy Press, Washington, DC), pp. 8-74.

3. Shlaes, D. M., Gerding, D. N., John, J. F., Jr., Craig, W. A., Bornstein, D. L., Duncan, R. A., Eckman, M. R., Farrer, W. E., Greene, and W. H., Lorian, V. , et al. (1997) Infect Control. Hosp. Epidemiol. 18 , 275-291.

4. Gonzales, R. D., Schreckenberger, P. C., Graham, M. B., Kelkar, S., DenBesten, K., and Quinn, J. P. (2001) Infections due to vancomycin- resistant Enterococcus faecium resistant to linezolid, Lancet 357 , 1179.

5. Soltani, M., Beighton, D., Philpott-Howard, J., and Woodford, N. (2001) Mechanisms of Resistance to Quinupristin-Dalfopristin among Isolates of Enterococcus faecium from Animals, Raw Meat, and Hospital Patients in Western Europe, Antimicrob. Agents. Chemother. 45 , 645-646.

6. Pai, M. P., Rodvold, K. A., Schreckenberger, P. C., Gonzales, R. D., Petrolatti, J. M., and Quinn, J. P. (2002) Risk factors associated with the development of infection with linezolid- and vancomycin-resistant Enterococcus faecium , Clin. Infect. Dis. 35 , 1269-1272.

7. Pillai, S. K., Sakoulas, G., Wennersten, C., Eliopoulos, G. M., Moellering, R. C., Jr., Ferraro, M. J., and Gold, H. S. (2002) Linezolid resistance in Staphylococcus aureus : characterization and stability of resistant phenotype, J. Infect. Dis. 186 , 1603-1607.

8. Hershberger, E., Donabedian, S., Konstantinou, K., and Zervos, M. J. (2004) Quinupristin-dalfopristin resistance in gram-positive bacteria: mechanism of resistance and epidemiology, Clin. Infect. Dis. 38 , 92-98.

9. Mangili, A., Bica, I., Snydman, D. R., and Hamer, D. H. (2005) Daptomycin- resistant, methicillin-resistant Staphylococcus aureus bacteremia, Clin. Infect. Dis. 40 , 1058-1060.

10. Sabol, K., Patterson, J. E., Lewis, J. S., 2nd, Owens, A., Cadena, J., and Jorgensen, J. H. (2005) Emergence of daptomycin resistance in Enterococcus faecium during daptomycin therapy, Antimicrob. Agents Chemother. 49 , 1664-1665. 9

11. Helms, M., Vastrup, P., Gerner-Smidt, P., and Molbak, K. (2002) Excess mortality associated with antimicrobial drug-resistant Salmonella typhimurium , Emerg. Infect. Dis. 8 , 490-495.

12. Barber, M. (1947) Staphylococcal infection due to penicillin-resistant strains, British Medical Journal 29, 863-865.

13. Thompson, K. S., and Smith, M. E. (2000) Version 2000: the new β- lactamases of Gram-negative bacteria at the dawn of the new millennium, Microbes Infect. 2 , 1225-1235.

14. Wright, G. D. (1999) Aminoglycoside-modifying enzymes, Curr. Opin. Microbiol. 2 , 499-503.

15. Hooper, D. C. (2000) Mechanisms of action and resistance of older and newer fluoroquinolones, Clin. Infect. Dis. 31 , S24-28

16. Piddock, L. J. V. (2006) Clinically relevant bacterial chromosomally encoded multi-drug resistance efflux pumps, Clinl. Microbiol. Rev. 19 , 382-402.

17. Levy, S. (1992) Active efflux mechanisms for antibiotic resistance, Antimicrob. Agents Chemother. 36 , 695–703.

18. Bolhuis, H., van Veen H. W., Poolman B., Driessen, A. J., and Konings, W.N. (1997) Mechanisms of multidrug transporters, FEMS Microbiol. Rev. 21 , 55– 84.

19. Nikaido, H. (1994) Prevention of drug access to bacterial targets: permeability barriers and active efflux, Science 264 , 382–388.

20. McMurry, L., Petrucci, R. E., Jr., and Levy, S. B. (1980) Active efflux of tetracycline encoded by four genetically different tetracycline resistance determinants in Escherichia coli , Proc. Natl. Acad. Sci. USA 77 , 3974-3977.

21. Murakami, S., Nakashima, R., Yamashita, E., and Yamaguchi, A. (2002) Crystal structure of bacterial multidrug efflux transporter AcrB, Nature 419 , 587-593.

22. Yu, E. W., McDermott, G., Zgurskaya, H. I., Nikaido, H., and Koshland, D. E., Jr. (2003) Structural basis of multiple drug-binding capacity of the AcrB multidrug efflux pump, Science 300 , 976-980.

23. Murakami, S., Nakashima, R., Yamashita, E., Matsumoto, T., and Yamaguchi, A. (2006) Crystal structures of a multidrug transporter reveal a functionally rotating mechanism, Nature 443 , 173-179. 10

24. Seeger, M. A., Schiefner, A., Eicher, T., Verrey, F., Diederichs, K., and Pos, K. M. (2006) Structural asymmetry of AcrB trimer suggests a peristaltic pump mechanism, Science 313 , 1295-1298.

25. Sennhauser, G., Amstutz, P., Briand, C., Storchenegger, O., and Grutter, M. G. (2007) Drug export pathway of multidrug exporter AcrB revealed by DARPin inhibitors, PLoS Biol. 5, e7.

26. Schumacher MA, Miller MC, Grkovic S, Brown MH, Skurray RA, Brennan RG. Structural mechanisms of QacR induction and multidrug recognition. Science. 2001;294:2158–2163.

27. Ramos JL, nez-Buenzo MM, Molina-Henares AJ, Tera'n W, Watanabe K, Zhang X, Gallegos MT, Brennan R, Tobes R. The TetR family of transcriptional repressors. Microbiol. Mol. Biol. Rev. 2005;69:326–356.

28. Gu R, Su C-C, Shi F, Li M, McDermott G, Zhang Q, Yu EW. Crystal structure of the transcriptional regulator CmeR from Campylobacter jejuni . J. Mol. Biol. 2007;372:583–593.

CHAPTER 2. LITERATURE REVIEW: RND TRANSPORTERS

INTRODUCTION Multidrug resistance (MDR) is a devastating problem associated with the treatment of many bacterial infections and chemotherapeutic resistant cancers. One mechanism of MDR is the expression of integral membrane efflux pumps that actively extrude an array of structurally dissimilar drugs as they cross the cell membrane (1, 2). When over expressed in cancerous cells, P-glycoprotein, an efflux pump in human cells, which naturally functions to prevent toxic molecules from crossing the blood-brain barrier, causes cancer cells to become resistant to various chemotherapeutic agents (3-5). Additionally, bacteria harbor a diverse range of efflux pumps capable of extruding a variety of antibacterial compounds. The efflux pumps effectively reduce the intracellular concentrations of the antibiotics and make eliminating the infection increasingly difficult (4, 6, 7). Treatment of these drug resistant bacterial strains is exacerbated by the ease at which resistant markers are genetically transferred between species. In fact, plasmid mediated genetic transfer has been implicated in quinolone and fluoroquinolone resistant infections in many human isolates. In 1998, a plasmid, qnrA1, was identified that harbored genes able to protect the bacterial pathogen by utilizing various mechanisms, including drug efflux by the innermembrane transporter QepA (8). As efflux pumps are capable of exporting a library of structurally diverse ligands, the transfer of these resistance genes is a major concern due to the risk of an epidemic of multidrug resistant bacterial diseases (9). Multidrug transporters are classified into five families based on sequence and functional similarities, these families include: primary transporters of the ATP- binding cassette (ABC) family (Fig. 1B) (10), and secondary transporters in the resistance-nodulation-division (RND) (Fig. 1A) (11), multidrug and toxic compound extrusion (MATE) (Fig. 1E) (12, 13), major facilitator (MF) (Fig. 1C) (14-16), and small multidrug resistance (SMR) families (Fig. 1D) (17). Of these classes of efflux pumps, the RND family confers resistance to the broadest range of antibacterial 12

agents (18, 19) and is the major cause of multidrug resistance in many pathogenic Gram-negative bacteria, including Burkholderia pseudomallei, Escherichia coli O157:H7, Haemophilus influenzae, Neisseria gonorrhoeae, Neisseria meningitides, Pseudomonas aeruginosa, Salmonella typhimurium, Stenotrophomonas maltophilia, and Vibrio cholerae (20-23). This illustrates why RND transporters are considered a primary contributor to multidrug resistance associated with Gram-negative bacteria. Although members of the RND family have been identified that promote a drug resistant phenotype in primarily gram-negative bacteria, they are ubiquitous in all domains of life, including Gram-positive bacteria, archaea, and eukaryotes (11). The prevalence of these transporters highlights the critical need to fully understand the functional mechanism of these pumps. Transporters in the RND family are energized through the proton-motive force (PFM) with proton translocation occurring in the transmembrane (TM) domain (24). RND pumps in Gram-negative bacteria generally function as tripartite efflux complexes in conjunction with a membrane fusion protein (25, 26) and an outer membrane channel (27, 28) (Fig. 1A) to export substrates completely out of the bacterial cell. The transporters share common structural features, including 12 transmembrane (TM) helices that harbor large periplasmic loops between TM1 and 2, and TM 7 and 8 which extend ~70 Å into the periplasm (11, 29-33). The outer membrane channel completely spans the outer membrane and extends an additional 100 Å into the periplasm. The combined 170 Å allows the outer membrane channel and inner membrane transporter to make weak interactions. These contacts are further stabilized by membrane fusion proteins, which interact with both the outer and inner membrane proteins (34-36), thus providing a critical component of the tripartite complex. The transport complexes AcrA-AcrB-TolC of E. coli and MexA-MexB-OprM of Pseudomonas aeruginosa are two well studied examples of RND efflux complexes, whereby AcrB and MexB function as inner membrane transporters. Crystal structures are available for both AcrB (29-33) and MexB (34). These structures indicate an asymmetric, functional rotating mechanism for ligand export (31-34). The 13

significant role that these pumps play in multidrug resistance associated with virulent strains of bacteria highlights the critical need to further understand the mechanism these transporters employ to recognize and expel structurally dissimilar antibiotics. This review will highlight the genetic, biochemical, and structural studies undertaken to gain valuable insights into RND transporters, with a particular focus on information garnered from the research on AcrB and MexB.

BACKGROUND The RND family of proteins was first described by Saier and colleagues, et al. 1994 (35), and was based on homology between functionally related membrane proteins (Fig. 2). Initially, five proteins were classified into the RND family (35) with two of those being previously identified as transporters conferring resistance to heavy-metal ion toxicity. It was suggested that resistance was due to extrusion of heavy-metal ions via active transport. The cobalt, cadium, zinc (Czc) (36-38) and the cobalt nickel resistant (Cnr) (39, 40) efflux systems were identified as close homologues to three other proteins involved in nodulation of legumes by rhizobia (NolGHI) (41, 42) and a protein which plays a role in cell division (EnvD) (43). Based on the variety of functions these proteins were implicated in, the resistance/nodulation/cell division family adopted its unique title. With the genes for these membrane proteins being preceded upstream by genes transcribing a class of membrane fusion proteins, it was suggested that members of the RND family may similarly function to actively transport substrates across the inner and outer membranes of Gram-negative bacteria (35). Although initially found in only Gram- negative bacteria, further homology studies and hydophobicity analysis identified proteins belonging to the RND family in all domains of life (11). Newly identified proteins from this phylogenetic analysis identified seven sub-families of RND transporters, including a protein implicated in Niemann-Pick type C (NP-C) disease, a lipid storage disorder in humans (44-46), as well as three sub-families associated with multidrug export (11). Among the proteins classified as drug transporters, the acriflavine-resistance protein B (AcrB) (47) was identified as a member of the 14

primarily Gram-negative bacterial hydrophobe/amphiphile efflux-1 (HAE1) family (11). A sequence alignment highlighting important residues and regions of AcrB with other members of the RND family is illustrated in figure 2. Although AcrB was identified as a member of the RND family within the last ten years, the multidrug resistant phenotype associated with the acr gene locus was identified more then 40 years ago (48). Intriguingly, during a study undertaken on E. coli strains resistant to the basic dye acriflavine, a cross-resistance to other basic dyes was also observed. Nakamura, et al. 1965, identified a single locus responsible for the phenotype, and termed this resistance marker acrA (later called acrB ). It was observed that, after selecting for resistance to any one of the basic dyes, including acriflavine, methylene blue, toluidine blue, crystal violet, methyl green, and pyronine B, that cross-resistance to all the other dyes was acquired. Furthermore, he noticed that the resistance to the acidic dyes eosin Y and erythrosin were not affected by the acrA locus. It was therefore suggested that the acrA gene product produced a membrane-associated factor that effected permeability to cationic dyes. The proposed model, at that time, stated that acrA mutations affected the biosynthesis of a 58-kDa inner membrane protein and, in the mutant acrA strain; acriflavine, or other basic dyes, would bind to the protein resulting in the increased resistance (48). Further studies indicated that resistance to structurally dissimilar compounds, including phenethyl alcohol, sodium dodecyl sulfate, as well as the antibiotics novobiocin and nalidixic acid were also mediated by the acrA locus. This could not easily be explained by the previously hypothesized model; therefore further inquisition was needed to delineate the resistant mechanism (49). Following the molecular cloning of the acrA DNA fragment, using a complementation test, it was discovered that the open-reading-frame (ORF) of acrA contained two genes, acrA and acrB (at the time termed acrE as the name acrB had already been used for another acriflavine-resistant gene) (50). The acrA ORF encodes a 397-residue lipoprotein containing a 24-residue periplasmic signal peptide at the N-terminus, while the acrB gene product transcribes a 1049 amino acid transmembrane protein. These gene products were identified as close 15

homologues to the previously identified E. coli proteins EnvC and EnvD (Fig. 2), respectively (50). Sequence analysis identified AcrB as a membrane protein with 12 membrane spanning sequences. Moreover, it was surmised that AcrB may function as a transporter, which was hypothesized based on homology between AcrB and the heavy-metal transporter CzcA (50) (Fig. 2). Characterization of AcrA suggested that this was a periplasmic protein that may aid in transport, but at the time of its discovery, little was known about the proteins function. At the time that the acrA/acrB genes were cloned, another multidrug resistance determinant was identified in Pseudomonas aeruginosa (51). It was observed that a siderophore-deficient P. aeruginosa mutant, when grown in an iron- deficient media, became less susceptible to drug-induced toxicity compared to the wild-type strain. Further investigation identified an ORF composed of three genes, multiple efflux genes A, B, and outer-membrane protein M ( mexA/mexB/oprM , previously identified as orfA/orfB/orfC), associated with the drug-resistant phenotype. Mutants defective in mexA, mexB, or oprM exhibited increased susceptibility to tetracycline, chloramphenicol, ciprofloxacin, streptonigrin, and dipyridyl, consistent with a role for these gene products in multiple antibiotic resistance. It was further identified that MexA, MexB, and OprM exhibited sequence homology to AcrB, AcrA, and a previously described family of outer-membrane channels, respectively. Based on this homology and the previous implication of the active transport of pyoverdine (a siderophore in P. aeruginosa ) by MexA and MexB, it was suggested that these proteins most likely function as multidrug transporters (51). The clinical implications of MexA/MexB/OprM were further explored in a study of 11 patients treated with with β-lactam antibiotics to combat a P. aeruginosa infection (52). It was found, in addition to resistance to β-lactams, these bacterial strains acquired cross-resistance to various unrelated antibiotics, including quinolones, tetracyclines, and trimethoprim. Using immunoblotting techniques, with polyclonal antibody raised against OprM, the cause of the multidrug resistance phenotype was identified as over-expression of MexA/MexB/OprM. Additionally, 16

AcrB was shown to effectively promote a drug resistant phenotype. In this case, inactivation of the constitutively expressed RND pump AcrB causes E. coli to become more susceptible to lipophilic β-lactams, whereby the minimum inhibitory concentrations (MIC) of lipophilic penicillin and cloxacillin is reduced from 512 µg/ml in the wild type strain to approximately 2 µg/ml in the mutant (53).

RND FUNCTION With these RND transporters being implicated in a majority of acquired and intrinsic multidrug resistance cases among Gram-negative organisms, it was important to understand the molecular mechanisms and recurring features that promote the phenotype. Overall, the hallmarks of the RND family of transporters are a transmembrane domain consisting of 12 α-helical segments (TMS) along with two large-periplasmic domains located between TMS 1 and 2 and TMS 7 and 8 (29) (Fig. 3). This overall architecture was first confirmed using the phoA gene fusion method to study the MexB protein. Alkaline phosphatase is enzymatically activated only after it is translocated from the cytoplasm to the periplasm, thus allowing for determination of periplasmic spanning segments based on relative phosphatase activity (54). Figure 3 is a schematic representation of the results of the phoA fusion experiments on MexB. Similar experiments were performed to determine the architecture of the RND transporters MexD, also of P. aeruginosa (55), and CzcA, a heavy metal transporter from Ralstonia sp (56). Based on the results, it was concluded that an overall similar topology exists throughout the RND family. Moreover, as the N- and C-terminal halves of these proteins share sequence similarities, it was suggested that the 1-plus-5 repeated architecture may have evolved via a tandem duplication event. Interestingly, the three nodultation- dependent RND proteins, NolGHI, complex to form the full-length RND protein, whereby NolG, H, I are homologous to the N-terminal, middle, and C-terminal portions of AcrB, respectively (35). Proteins among the RND family act as proton/ligand antiporters, in which the inner-membrane protein functions as the proton-motive-force (PMF) dependent 17

transporter (57). Initially, indirect methods were used to provide evidence for this energy-driven efflux mechanism. The PMF was examined using drug susceptibility studies of both AcrA/AcrB/TolC and MexA/MexB/OprM. In whole-cell experiments, AcrB mutants of the putative proton-relay network increased the accumulation and thus, susceptibility to cephalothin and cephaloridine (58). Similar results were obtained for MexA/MexB/OprM (59), whereby MexB-null mutants accumulated drugs more readily than wild-type strains. Additionally, deenergization of the cytoplasmic membrane using the proton conductor carbonyl cyanide m-chlorophenylhydrazone (CCCP) always produced a strong increase in the accumulation level of substrates, which indicated a potential requirement of a proton gradient for drug efflux (59). Finally, a mutant overproducing MexAB-OprM accumulated less tetracycline and chloramphenicol than the parent strain and was more resistant to a wide range of antimicrobial compounds, including b-lactams (60). In a novel series of experiments using Purified AcrB reconstituted into proteoliposomes, direct evidence was provided for a PMF-driven substrate transport mechanism (61). In this extensive work, AcrB was shown to extrude a fluorescent derivative of phosphatidylethanolamine (PE), 7-nitrobenz-2-oxa-1,3-diazol-4-yl (NBD)-PE, in a proton-dependent manner. As most of the natural AcrB substrates are lipophilic and readily diffuse across the membrane, attempts made to directly measure accumulation of these compounds into inverted vesicles were unsuccessful. It may be that the naturally slow kinetics (turnover rate of less than one per minute for most substrates (62)) of the pump and the removal of the membrane fusion protein, AcrA, during purification procedures may have further hindered the reconstitution experiments. P revious to the NBD-PE studies, only inconsistent results could be obtained. For example, using an everted vesicle system, it was suggested that AcrB may transport taurocholate, a bile acid with a pKa of ~1.5 that does not readily cross the lipid membrane, utilizing the PMF. As those vesicles without AcrB also showed accumulation of taurocholate, these results were inconclusive (63). 18

In this regard, an intuitive method was used to measure the transport of the NBD-PE, whereby AcrB was reconstituted into donor vesicles harboring fluorescently labeled PE and a rhodamine-PE derivative. The rhodamine conjugate quenches the fluorescently labeled PE’s signal within the donor vessicle (61). As NBD-PE molecules are transported, the quenching is diminished due to separation of the fluorescent signal and rhodamine quenching. Presumably, the NBD-PE would become reinserted into the original membrane as it is expelled by the AcrB pump. To counteract this event, an excess of AcrB-free, non-labeled, (acceptor) vesicles were used to trap the extruded phospholipids. In this fashion, the fluorescence signal is quenched in presence of rhodamine in the donor vesicles. As the fluorescent-PE is exported by AcrB and trapped into the acceptor vesicles, AcrB activity can be measured through the increase of fluorescence signal. Figure 4 illustrates the overall experimental design. These studies allowed for the examination of the effect of pH and the PMF on substrate transport by AcrB. The results indicated that donor vesicles suspended in an acidic external environment were able to transport labeled PE. Furthermore, using known substrates of AcrB, including bile acids, erythromycin, and cloxacillin, PE transport was eliminated. This indicated that AcrB has a strong preference for these lipophilic molecules. Additionally, by incorporating various drugs into the AcrB-containing proteoliposome system, proton efflux could be measured. As the internal proton concentration increased beyond that of the external buffer during transport, proton translocation provided compelling evidence of the involvement of a proton- dependent transport system. To more directly observe proton/drug antiport, the proteoliposome system was energized via the valinomycin-induced flux of K +, which was converted into a proton gradient in the presence of KCl. Using the fluorescent pH indicator pyranine, proton efflux, associated with drug influx, could be confirmed by directly measuring the the change in intravesicular pH. Interestingly, the addition of the membrane fusion protein AcrA accelerated extrusion of substrates (61), which was interpreted as AcrA connecting the donor and acceptor vesicles. An Alternative explanation later surfaced, whereby AcrA may be required to activate the pumping 19

mechanism. This was suggested based on results stemming from a series of studies using another RND transporter purified into proteoliposomes. Although the exact mechanism of membrane fusion protein activation remains unresolved, the critical importance to RND transporters was illustrated in experiments using the AcrB homologue AcrD of E. coli . In contrast to AcrB, AcrD transports many hydrophilic compounds including aminoglycosides that do not readily traverse the cellular membrane. After purification and reconstitution of AcrD into inverted membrane vesicles, it was shown that aminoglycoside transport only occurred when AcrA was supplemented into the proteoliposome (equivalent to periplasm) (64). It was previously recognized that AcrA was required for AcrD activity in viable cells (65) , but this was not unexpected as AcrA is required to make appropriate stabililizing contacts between the outer membrane channel TolC and AcrD. Interestingly, in AcrD- proteoliposomes, the tripartite complex is not required, thus these results strongly suggest MF proteins may perform a critical function in activating RND transporters. In Similar experiments using the heavy-metal ion exporter CzcA, it was observed that this RND protein was able to transport toxic divalent cations such as Zn 2+ , Co 2+ , or Cd 2+ . The Requirent of a proton gradient for substrate transport was again observed using inverted membrane vesicles. The proton gradient was created by dilluting the external buffer and thus forming a higher internal (periplasmic) proton concentration (56). This provided initial indications for a cation/proton antiporter for the CzcA system as well. Supporting results were obtained when the vesicle interior was acidified using the diffusion of NH3 from proteoliposomes loaded with 0.5 M 2+ NH 4Cl. Data from these experiments using Zn indicated sigmoidal kinetics with a Hill coefficient of 2. Furthermore, these studies showed that the transport was saturable and that this channel may function by a two-channel mechanism, whereby one channel may be used to transport metal ions and the other used to import protons. Additional results from the CzcA inverted-vesicle system suggested the importance of well-conserved, charged residues, including Asp402, Asp408, and 20

Glu415, located in TM-helix 4 of CzcA (56). The significance of the charged residues was observed following a single-point mutation of each amino acid. Following the mutation, divalent-cation transport was completely abolished. This report was later corroborated by site-directed mutagenesis studies of MexB (67) . In MexB, the Asp402 and Asp408 of CzcA are replaced by a tandem pair of aspartate residues, Asp407 and Asp408 on transmembrane segment (TMS) 4. As was the case in CzcA, both of the Asp residues were critical for the function of MexB. Additionally, Lys939, located in TMS 10, was also found to be essential to drug export (67). Finally, Thr978 of AcrB of TMS-11 was identified as another vital component (68). Together, it is suggested that these residues constitute a proton- relay system that allows for the active transport of RND substrates. The alignment in table 1 further identifies the residues critical to the proton-relay network.

SUBSTRATE CAPTURE AND TRANSPORT RND transporters, including AcrB and MexB, seemingly provide multidrug resistance to the widest range of deleterious compounds among multidrug export proteins. Understanding the nature of multiligand recognition and substrate transport is a primary focus of the work on these proteins. To this point, AcrB has been shown to provide significant levels of resistance to various structurally unrelated compounds, including detergents, simple organic solvents, bile acids, various cationic dyes, and antibiotics such as penicillins, cephalosporins, fluoroquinolones, macrolides, chloramphenicol, tetracyclines, novobiocin, fusidic acid, oxazolidinones, and rifampicin (69, 70). Recently, direct evidence from the crystal structures of AcrB-ligand complexes has provided details into ligand recognition. This will be discussed below (30). Prior to the resolution of the AcrB crystal structures, an abundance of data was available to provide insights into the drug uptake, recognition, and extrusion mechanisms of RND transporters.

Drug entry: In the early days of studies on AcrB, it was presumed that this protein, along with other members of the RND family, acted simply as cytosolic vacuum cleaners. This hypothesis seemed likely since AcrB transverses the inner 21

membrane and utilizes a membrane fusion protein and outer-membrane channel to completely rid the cell of substrates. Early indications that an additional periplasmic or cytoplasmic membrane entry point also existed arose initially from insights into P- glycoprotein, an ABC-transporter that could potentially capture ligands directly from the plasma membrane (71). Additionally, it was observed that ceftriaxone, a dianionic β-lactam not able to passively diffuse across the cytoplasmic membrane, could be extruded from drug resistant strains of P. aeruginosa overexpressing MexB (72) . From this, it was postulated that MexB could capture substrates from either the periplasm or from the outer leaflet of the cytoplasmic membrane. Analogous to this work, it was observed that a strong correlation exists between the lipophilicity of β- lactam side-chains and the efficiency to which they are exported by AcrB. By comparing MIC values between an AcrB-knockout mutant and a wild-type Salmonella typhimurium strain overproducing AcrB, It seemed that highly lipophilic compounds, including nafcillin and cloxacillin, were better substrates and more readily transported by the multidrug exporter (73). In contrast, compounds presenting hydrophilic side chains were not exported as efficiently. Importantly, since it was previously suggested that resistance to these antibiotics primarily occurred due to β-lactamase hydrolytic activity, Salmonella typhimurium lacking the hydrolyzing enzyme were utilized. The previously established resistance model had to be reevaluated. The work on AcrB initiated a paradigm shift, whereby lactam- antibiotic resistance could be enhanced through the presence of multidrug exporters that capture and export the antibiotics from the periplasm to the cytoplasm, respectively. Intriguingly, multidrug efflux pumps can illicit a more-than-additive resistance phenotype by working synergistically with various simple pumps. For instance, MexAB/OprM alone produces a modest 4 µg/ml MIC to tetracycline, but functioning in conjunction with the inner-membrane spanning tetracycline efflux pump TetA, the MIC increases to ~512 µg/ml. These results suggest that TetA specifically exports tetracycline from the cytoplasm to the periplasmic space, which alone produces minimal resistance. As the antibiotic accumulates, the periplasmic vacuum 22

MexAB/OprM ultimately eliminates tetracycline completely from the cell. This indirectly supports the presence of a periplasmic entry points for ligands. Furthermore, this work illustrates how bacteria can adapt to toxic environments and evade antibiotic toxicity by utilizing cooperating efflux pumps, an avenue that requires further exploration (74). The most convincing evidence for the presence of a periplasmic, substrate entry point was established using, the previously mentioned, AcrD-inverted proteoliposome experiments. In these experiments, aminoglycoside ligands induced proton transport when the substrates were present in either the exterior (cytoplasm) or interior (periplasm) of the lipid vesicle (64). This provided evidence for a substrate entry point from both the cytoplasm, as well as the periplasmic space to the RND transporters, since aminoglycosides used in the experiments do not passively transverse the membrane.

Periplasmic drug binding: Structural studies (discussed below) from AcrB have provided the most intriguing evidence regarding ligand recognition and capture. Prior to the availability of these three-dimensional structures, three independent examples were published suggesting that the two large periplasmic loops (between TMS 1-2 and TMS 7-8) were critical in ligand recognition. Two of those, published in 2002, utilized protein chimeras from the fusion of two RND proteins that differ in substrate specificities to analyze the critical binding regions. One chimera was constructed using the N-terminal residues of AcrB and C-terminal residues of MexB (75). AcrB and MexB have strikingly similar substrate specificities; however they are highlighted by varying degrees of resistance levels to those substrates. These differences were utilized for experimental analysis of the chimeric proteins. First, MexAB-OprM provides increased resistance to cinoxacin, while AcrAB-TolC expression shows higher resistant levels for positively charged ethidium bromide and oleandomycin. Taking this into consideration, when expressing chimeras in E. coli harboring the appropriate MFP and OM components, the experiments revealed the critical regions required for substrate interaction. These regions are located 23

primarily within the periplasmic loops of the RND transporters. In fact, these studies showed that the 200 C-terminal residues of MexB were sufficient to convert the specificity of AcrB to the ligands ethidium bromide, oleandomycin, and cinoxacin to a more typical MexB phenotype. Additionally, the N-terminal residues 1-60 and 612- 849 seem to affect the specificities of these hybrid transporters (75). Elkins and Nikaido, et al. 2002, corroborated this analysis using AcrB/AcrD hybrids (76). Having high sequence similarity, yet strikingly different substrate profiles, these chimeras provided strong evidence to support the notion that the periplasmic loops are responsible for ligand recognition. Chimeras were constructed by replacing the two ~300 amino acid residue loop regions between TMS 1-2 and 7- 8 of either AcrB or AcrD with the corresponding region of the other RND-transporter. Results illustrated that if the periplasmic loops of AcrD were replaced with the corresponding region of AcrB, an AcrB-specific substrate resistance profile was acquired. Similarly, by introducing the loops of AcrD into AcrB, this AcrB/AcrD chimera was able to recognize aminoglycosides, which are generally specific to AcrD. Importantly, since AcrA and TolC are utilized by both efflux pumps, these results can not be explained due to a variation in the MFP or OM channel (76). Finally, in a study using a different approach, spontaneous point mutations were identified within another RND transporter that altered specificity. Single-point mutations were identified in MexD of P aeruginosa that increased the range of substrates it could recognize (77). Intriguingly, all of these single-point mutations were located within the two periplasmic loops of the protein. Specifically, Q34K, E89K, A292V and P328L were found in the first loop, located between transmembrane domains (TMD) 1 and 2, whereas F608S and N673K were contained in the second loop, located between TMD7 and TMD8. Together, these examples highlight the proposed mechanism, whereby substrates enter the transporter through the periplasm, or potentially the outer-leaflet of the inner membrane, and are captured in specific binding sites within the periplasmic loop regions for transport. 24

In the years after this series of insightful experiments, more evidence accumulated to suggest the periplasmic domain was indeed responsible for ligand recognition. In 2004, in vitro random mutagenesis was carried out on the mexB gene to test for substrate specificity (78). Many of the mutations that garnered decreased sensitivity were isolated to the putative proton-relay network or Gly220, a residue stabilizing the trimer. The most interesting mutations appeared in the periplasmic domain, a region later identified as the binding cleft. Mutations that increased substrate induced toxicity included Ala618 and Arg716 which are located on the opposing walls of the proposed binding cleft, a region which will be discussed in more detail later in the text. Additionally, studies on the AcrB homologue YhiV also suggested the importance of the periplasmic domain in substrate recognition (79). An E. coli strain overproducing YhiV (lacking AcrB and AcrF) showed increased resistance to levofloxacin and other aromatic substrates after repeated exposure to levofloxacin. The higher resistance levels were attributed to a periplasmic loop mutation of Val610 into an aromatic phenylalanine residue. Interestingly, this mutation decreased resistance levels to non-aromatic compounds, which further highlights the importance of the local environment formed within the substrate binding site.

STRUCTURAL STUDIES OF AcrB Structural information provides an important means to understand how proteins function. As a major step in uncovering the transport mechanism of RND tranporters, the high-resolution (3.5Å) crystal structure was solved for AcrB in the spacegroup R32 in 2002 (29). The structure revealed a single AcrB monomer within an asymmetric unit. The AcrB homo-trimer was produced by utilizing the inherent three-fold crystallographic symmetry of the R32 spacegroup. Overall, AcrB takes on a compact topology, initially described as a jellyfish appearance, with a transmembrane (TM) domain comprised of 12 transmembrane helices from each monomer. The periplasmic portion of AcrB, which extends ~70Å into the periplasm, is divided into the TolC-docking domain and the porter (pore) domain. Figure 5 displays a single monomer with specific domains and subdomains indicated. 25

TM domain: The transmembrane domain of AcrB, a 50Å-thick membrane spanning region, is composed of 12 membrane spanning α-helices from each monomer, which confirmed earlier topological predictions (Figure 5). The TM domain has been identified as the location for proton import and functions to sequester the energy from the proton-motive-force to drive substrate translocation. Within the TM domain, a central core composed of helices α4 and α10 is surrounded by the rest of the TM domain helices. The central core of each monomer harbors three titratable residues implicated in proton transport. These residues, D407 (TM4), D408 (TM4), and K940 (TM10) have been proposed to drive the conformational changes required for ligand export by sequentially transferring protons down their concentration gradient through various transition states of the transport cycle (29) (Figure 6). Site-directed mutagenesis studies, discussed previously for CzcA, AcrB and MexB (56, 61, 67), have provided supporting evidence to this model, whereby substitution of the amino acids eliminates effective drug resistance. Similar results were observed for AcrB. Using matrix-assisted laser desorption ionization time-of- flight (MALDI-TOF) mass spectrometry, it was shown that D408 specifically reacts with dicyclohexylcarbodiimide (DCCD), which, In a pH-dependent manner, can be used to react with and measure pKa values for membrane-buried carboxylates (80). The apparent pKa of the amino acid D408 (7.4) would allow binding and release of protons under physiological conditions. Not surprisingly, D408 was not protected from carbodiimide labeling while substrates were incorporated into the system, thus providing evidence to the working model that states drug and proton transport are spatially separated. Furthermore, this work suggests the importance of D408 as a critical element in the proton translocation pathway. In fact, crystallization experiments were performed using mutants of the putative proton-relay network (81). In this work, site-directed mutagenesis was used to mutate residues D407, D408, K940, and T978 (helix 11) to alanine. Previously, using alanine scanning experiments it was observed that Thr978, a polar residue in close proximity to Asp407, also plays an essential role in the proton translocation pathway. The 26

resulting structures of the mutant proteins were essentially identical which suggested the relevance of the observed conformations. Interestingly, the structures were strikingly different from the previously identified AcrB crystal forms. At that time, it was hypothesized that these mutant proteins revealed a transient state that existed during the transport cycle (81). Figure 6 identifies the local conformational shifts resulting from the proton-relay mutants and highlights key residues in the proton- relay network.

Periplasmic region: The periplasmic region extends from the TM domain to approximately 70Å into the periplasmic space and can be divided into two stacked domains. Adjacent to the transmembrane domain, the porter domain (formerly described as the pore domain) extends 40Å into the periplasm. The upper part, distal to the inner membrane, extends the additional 30Å and is referred to as the TolC docking domain (29) (Figures 5, 7, 8).

TolC docking domain: The TolC docking domain is composed of two homologous subdomains termed DN and DC (Figure 5) (29). These subdomains are each comprised of a four-stranded mixed β-sheet, whereby two antiparallel β- strands are located parallel to an additional β-sheet hairpin structure incorporated from the other half of the protomer or one of the other monomers. Additionally, a vertical hairpin motif is located at the apex of each of the subdomains. As the three TolC docking domains, one from each monomer, come together, a funnel-like structure is formed with a larger opening at the top. Intriguingly, this opening is similar in diameter to the bottom of the outer membrane channel TolC. The six vertical haipins of AcrB coincide with the six α-helix-turn-α-helix motifs at the bottom of TolC (the TolC structure will be discussed below) (82). The funnel narrows as it extends down the TolC docking domain and leads to the central pore of the porter domain.

Porter domain: The porter domain can be divided into four additional subdomains which include PN1, PN2, PC1, and PC2 (figures 5, 7, 8b) (29). The 27

PN1 and PN2 subdomains comprise the segment between the transmembrane helices TM1 and TM2, while the PC1 and PC2 subdomains form a similar segment between TM7 and TM8. Structurally, a commonality exists between all of the subdomains, whereby each subdomain is comprised of a repeat of two β-strand-α- helix-β-strand motifs sandwiched together. In this repeated motif, the helices are located externally to the four-stranded antiparallel β-sheets. Additionally, the PN1 and PC1 domains consist of an extra antiparallel β-strand, thus forming a five- stranded antiparallel β-sheet. In the trimeric AcrB structure, three porter domains, one from each monomer, come together to form a central pore that extends perpendicular to the membrane plane to the TolC docking domain funnel. The most interesting features of AcrB may be the vestibules that are open to the periplasm which are located at monomer interfaces as well as the deep cleft located between subdomains PC1 and PC2. Figures 7 and 8 depict ribbon diagrams and surface electrostatics of the trimeric AcrB protein and highlight the previously mentioned vestibules and clefts. The vestibules lead directly to a large central cavity (Fig 8A), which was initially proposed to be the location for ligand recognition. It was suggested that ligands may enter the central cavity through the vestibules, located at the monomer interfaces, positioned just above the inner-membrane plane (29, 30). This would allow entry from either the periplasmic space or the outer-leaflet of the inner membrane. Indeed, the first crystal structure solved for AcrB/ligand complexes, including co-crystallization with ethidium bromide (Et), dequalinium (Dq), rhodamine 6G (R6G), and ciprofloxacin (Cip), indicated that the central cavity, may be the ligand binding site (30). The crystallographic structures of trimeric AcrB co- crystallized with these four structurally dissimilar ligands showed that they bound to the upper portion of the 5000 cubic angstrom central cavity using distinct but overlapping binding sites, whereby a majority of the important contacts were formed from residues within the TM domain. It was proposed that upon binding, the ligands may exit through the central pore (30). Intriguingly, the central pore, the predicted exit pathway for ligands, extending from the central cavity to the TolC docking 28

domain was blocked by Asn109 of the porter domain, therefore contradicting the mechanistic claim. Furthermore, these results conflicted with previous experiments, whereby studies using domain swapping (previously mentioned in the text) indicated the periplasmic domain plays the most critical role in determining the ligand recognition profile (75, 76). It was suspected that ligands could be interacting with non-specific hydrophobic patches without being on the correct path for ligand export (76). To answer these contradictions and attempt to delineate the correct mechanism and pathway for ligand extrusion, Asn109 of AcrB was mutated to alanine and crystallized with various ligands, including Cip, R6G, Et, nafcillin (Naf), and Phe-Arg-β-napthylamide (83). In wild-type AcrB, Asn109 blocks the most direct pathway from the central cavity to the central pore by protruding into the pore interior. Thus, mutating the residue presumably would open the exit route and allow for further analysis of the extrusion pathway. Although AcrB function was not significantly altered, based on observed MIC values, a second substrate binding site was identified. The ligands were located in a region that corresponds to the binding cleft, located 15Å above the membrane plane (Fig. 8). The cleft, which lies perpendicular to the membrane plane, is approximately 40Å in lenth, 20Å in width, and penetrates AcrB ~15Å. The cleft was originally speculated to be the site of AcrA interaction. Figure 8b highlights the cleft by removing the TolC-docking domain and the TM domain. In attempts to determine if this binding was physiologically relevant, site-directed mutagenesis was performed on those amino acids that specifically interacted with bound ligands (83). Mutating these residues, which included Phe664, Phe666, or Glu673 severely inhibited AcrB function, determined by testing the MIC in correlation with substrate accumulation experiments. Importantly, the results suggested the relevance of this periplasmic binding site. Additional questions remained regarding the periplasmic substrate binding site. As inhibition could have occurred as a result of a mechanism unrelated to ligand recognition, further analysis of the binding cleft is needed. Moreover, these results still lack a fundamental mechanism to explain how drug export occurs and more specifically, where the 29

actual exit pathway for ligand export resides. An alternative hypothesis exists, whereby ligands may enter the second binding site through the clefts and completely bypass the central cavity. Indeed, based on the recent crystal structures of AcrB in an asymmetric conformation (discussed below) the binding site in the cleft region seems to be the most likely candidate for ligand recognition. In either instance, the central pore is the likely exit pathway for ligand extrusion.

ASYMMETRIC AcrB Recently, three groups have independently solved the crystal structure of AcrB in an asymmetric form (31-33). These structures have provided valuable insights and provided a working model for the detailed mechanism of RND transporter function. The structures suggest a functional rotating model to describe substrate efflux. The new asymmetric structures indicate that individual protomers adopt distinct conformations that correspond to one of three functional states during drug export. Based on the role each protomer plays, these states have been denoted as access, binding and extrusion. Initially, drugs are able to enter into a hydrophobic binding pocket in the access protomoer, but are unable to make appropriate binding contacts due to the diminished size of the tunnel interior. A conformational shift results in a transition from the access state to the binding state. In the binding state the size of the tunnel increases and allows AcrB to accommodate substrates. Indeed, an asymmetric structure presented by Murakami and Yamaguchi unequivocally identified 9-bromo-minocyclin (brominated minocycin derivative) and doxorubicin in only the binding protomer, indicating the physiological relevance of the binding state (31). Ligand recognition occurs within the binding cavity, whereby the substrates primarily interact with aromatic and hydrophobic patches in the cavity. Intriguingly, during the binding state, the access tunnel remains open. This may allow those compounds that are not AcrB substrates to exit the tunnel before being exported. In another asymmetric structure, two tunnels were identified that led to the binding protomer of AcrB. It was proposed that tunnel 1 may provide an entrance point from the outer leaflet of the inner membrane while 30

tunnel 2, positioned ~15Å above the membrane would allow entrance or exit of molecules from or into the perisplasmic space, respectively (32). Transition from the binding to the extrusion state closes the entrance tunnels and opens the exit tunnel leading to the central pore and the TolC-docking domain funnel. The ligand is forced out of the binding cavity owing to the decrease in volume associated with a switch to the expulsion state. This transition allows AcrB to transfer the bound ligand to TolC completing ligand export and initiating a transition back to the access state. Figure 9 illustrates the differences among the three protomers and highlights the proposed exit pathway in the asymmetric AcrB model. Based on structural evidence, the transition from the binding to the expulsion state is the rate-limiting and energy sequestering step in drug transport. The tirtratable residues Asp407, Asp408, and Lys940 undergo small conformational changes in the expulsion state presumably due to protonation of Asp407 and Asp408 which leads to cycling between the various states.

MexB STRUCTURE Following the identification of the asymmetric conformations of AcrB, the 3.0Å resolution crystal structure of MexB from P. aeruginosa was solved (34). Intriguingly, MexB shares a similar topology as AcrB and also adopts an aymmetric conformation. As is the case for AcrB, each subunit of MexB resides in a distinct conformational state. The MexB crystal structure provided support to the proposed transport model, whereby the subunits transition through three states in a functional rotating fashion to drive substrate export. Not surprisingly, with a 69.8% identity and an 83.2% similarity in sequence, a high degree of structural relatedness exists between AcrB and MexB (34). Superimposition of the tripartite structures revealed an RMSD (root mean squared deviation) between models of 1.4Å. Furthermore, with a common functional role in the TM domain and the sequences being nearly identical, a higher degree of structural similarity exits in this domain (rmsd < 1). This indicates proton 31

translocation may function identically among RND proteins and drive similar conformational changes throughout the protein. The most intriguing structural differences between MexB and AcrB locate to the porter domain and specifically the access conformation. Compared to the binding and extrusion states, which show an RMSD of 1.0Å, when superimposed over the same AcrB protomers, the Access subunit exhibits striking differences (RMSD of 1.6Å). Interestingly, in the access state of MexB, the entrance channel leading to the binding pocket is closed. Upon transition to the binding conformation, the entrance channel opens. This seems to be the result of the PC2 subdomain shifting towards the TM domain and PC1. It was suggested that this conformational difference is due to the specificity for its cognate MFP MexA, whereby the pore domain is suggested to make stabilizing contacts to the MFP. Indeed, a chimeric analysis of MexB and AcrB indicated that residues 589-612 of the PC1 subdomain are notably required for specificity to their respective MFP (75). Additionally, T578 also of PC1 of MexB seems to play a pivotal role in MFP binding (84). The pivotal role of the porter domain and the MFP in forming the entire tripartite complex is further highlighted by the docking domain. The outer-membrane channel is thought to interact transiently with low affinity to the RND protein without the presence of the MFP. Thus, it was not surprising that the docking domains of MexB and AcrB were highly similar. Fortuitously, an n-dodecyl-D-maltoside (DDM) molecule was identified in the binding cleft of the Binding protomer (34). The detergent DDM was used to solubilize the membrane protein. This substate-bound crystal structure further highlights the critical features of multidrug binding identified previously in the AcrB/substrate co-crystals. Briefly, a large cavity rich in exposed aromatic amino acids harboring pockets of polar residues highlights the multifaceted binding pocket. Results indicate the binding pockets harbor distinct but overlapping sites for ligand recognition. The polar residues seem to perform a role in ligand specificity, whereby the DDM forms hydrogen bonds with an asparagine and a serine residue. 32

The altered conformation of MexB compared to AcrB indicates some degree of structural flexibility. Furthermore the flexibility previously observed in various crystal structures of both AcrB (29-33) as well as the associated MFP proteins MexA (85, 86) and AcrA (87) suggests a critical need for mobility. The flexibility presumably plays a role for both tripartite association and pump function. Importantly, the structure of MexB-DDM along with the genetic evidence discussed below further supports the functional rotating model of substrate transport.

EVIDENCE TO SUPPORT THE ASYMMETRIC MODEL Genetic evidence supporting the functional rotating mechanism to describe substrate export has been established using two ingenious approaches. First, Takatsku, et al. 2007 (88) and Seeger, et al. 2008 (89), Introduced cysteine residues at appropriate locations in AcrB monomers by using site-directed mutagenesis. The idea was as follows: when two cysteine residues come within 6.4Å, the sulfur atoms from their side chains would form disulfide bonds. If the asymmetric structure is correct, it should be possible to detect the transition between the access, binding, and extrusion states in these mutants. As subdomains change in conformation, cross-links would form and inhibit any further changes in transition states. Indeed, the results suggested that the disulfide bonds did lock AcrB from transitioning between states, thus inhibiting drug export. One specific example supports the presence of the extrusion conformation, whereby disulfide formation between the PC2 subdomain and TM7 inhibited export activity (88). The access and binding conformations reveal an approximately 10Å distance between these cysteine residues, while a mere 3.3 Å separates the corresponding cysteines in the extrusion state of the S562C_T837C mutant. This result implies that AcrB, at least partially, exists in a conformation resembling the extrusion state, whereby PC2 and TM7 transition closer together. Intriguingly, this inhibition is alleviated with the addition of the reducing agent di-thiothreitol (DTT). These results, illustrating the existence of the extrusion conformation, support the crystallographic structure of AcrB in complex with designed ankyrin repeat protein inhibitors (DARPins), whereby the DARPins 33

bound to the access and binding states, but did not interact with the extrusion protomer (33). Furthermore, in another series of experiments, a covalently-linked AcrB trimer (with a short linker sequence connecting each monomer) was developed to test the functional rotating mechanism (90). In this way, a single protomer within the trimer could be manipulated. MIC values of the linked trimer indicated that this protein complex was completely functional. Intriguingly, after point mutations were introduced into the proton-relay network of just one protomer, the linked AcrB trimer became nonfunctional. Additionally, inhibiting a single AcrB protomer by introducing disulfide bonds in the external cleft of the periplasmic domain inhibited drug transport for the entire complex (90). Together, these studies provide supporting evidence to the functionally rotating mechanism and indicate the physiological importance of the transition through the three distinct monomeric states. Although it has been illustrated that the AcrB binding cleft is critical to ligand recognition and export, it is difficult to determine if the central cavity harbors ligand binding sites relevant to extrusion, or if crystal structures identifying ligands within the central cavity are simply an artifact of high ligand concentrations used during crystallization. The question remains as to how the substrates may enter the central pore during export. Furthermore, it is difficult to ascertain how AcrB manages to export such a diverse range of compounds without eliciting the multifaceted central cavity to recognize these molecules. It seems, with these newly resolved asymmetric structures more time is needed to answer the questions that remain on the function of RND transporters. Of those questions, the most important work seems to be on understanding the roles played by the membrane fusion protein AcrA and outer membrane channel TolC

TolC STRUCTURE At 2.1Å resolution, the TolC homotrimer is seen as a tapered cylinder approximately 140Å in length (82). This comprises a 40Å long Outer membrane β- barrel domain, which anchors a contiguous α-helical barrel domain that projects 100 34

Å across the periplasmic space. A third subdomain, of mixed α/β structure, forms a strap around the mid-section of the α-helical barrel. The average accessible interior diameter of the single central TolC pore is 19.8 Å. Three TolC monomers each contribute four β-strands to form the twelve-stranded β-barrel, which is constitutively open to the cell exterior. The periplasmic barrel comprises twelve antiparallel α- helices (two continuous long helices and two pairs of shorter helices from each monomer) that pack laterally side-by-side and form two separate interfaces. The helices follow a left-handed superhelical twist that tends to be less tightly wound in the upper half compared to helices in a conventional two-stranded coiled coil, enabling the helices to lie on the surface of a cylinder. In the lower half of the a barrel, neighboring helices form six pairs of regular two-stranded coiled coils, but one from each monomer folds inward at the periplasmic end. This constricts the entrance to TolC and establishes a resting state with a closed opening, whereby an effective diameter of approximately 3.9Å is established. The closed state is reflected in the small conductance of TolC in lipid bilayers (91). After the TolC and AcrB crystal structures were resolved, it was proposed that TolC and AcrB may directly interact with each other by virtue of their extended periplasmic domains. With TolC harboring the 100 Å periplasmic α-helical domain and AcrB protruding 70 Å into the periplasm, this notion seems very likely. The first experiments indicating this resulted from a study showing disulfide bond formation between engineered cysteine residues located on the periplasmic hairpin turns of AcrB and the coiled-coil region of TolC (92). Furthermore, it has been illustrated that the outermembrane channels TolC and OprM weakly interact with AcrB and MexB, respectively (93, 94). This interaction can be further stabilized by the membrane fusion proteins AcrA and MexB, thus highlighting the function of these membrane fusion proteins (Figure 10).

AcrA STRUCTURE Membrane fusion proteins act as a bridge to fasten the inner membrane transporter to its cognate outer membrane channel. The crystal structure of approximately two-thirds of AcrA (res 45-312), the membrane fusion protein of the 35

AcrB/AcrA/TolC tripartite complex, was elucidated in 2006 (87, 95). This structure revealed a protein with an overall topology similar to that of the previously resolved MexA structure (85, 86). Briefly, the three-dimensional structure of AcrA revealed a linear-arranged three-domain protein. The domains identified include an alpha- helical hairpin domain composed of two α-helices connected by a hairpin-turn that forms a 105 Å long structure, a lipoyl domain with two sets of four β-strand lipoyl half motifs connected by an α-helical linker, and a β-barrel domain composed of six antiparallel β-strands along with a single α-helix. An additional membrane proximal (MP) domain, identified through re-examination of the homologous MexA crystallographic data, and a yet unresolved C-terminal domain are intrinsically disordered in the AcrA crystals and, thus lacking in the AcrA structure (85-87, 95). Recently, successful cross-linking experiments of AcrA to both AcrB and TolC (86) along with extensive biochemical and genetic data have helped to illustrate the intermolecular contacts in the tripartite complex. The compiled data indicates that the α-helical hairpin domain, which is the least conserved of the domains, contacts TolC while the three other domains interact exclusively with AcrB (Figure 10). A more detailed description of the formation of the tripartite efflux complex is discussed below. Membrane fusion proteins, once thought to only function as adaptor molecules, have recently gained recognition as critical components in drug efflux. How the proteins aid in transport is an unanswered question. To this point, it has been shown that AcrA is essential for aminoglycoside transport by the RND transporter AcrD (64). In inverted membrane vesicles, AcrD was unable to transport aminoglycosides without the addition of AcrA to the internal portion of the vesicle, which corresponded to the periplasmic space in the experiment. Moreover, the membrane fusion protein MacA is critical for optimal catalytic efficiency of the ABC- type drug transporter MacB, increasing the rate of ATP-hydrolysis 45-fold (96). Together, this data indicates that membrane fusion proteins may stimulate substrate export through a direct interaction with the transporter possibly allowing efficient transition through transient states. Other evidence suggests that membrane fusion 36

proteins may interact directly with substrates to facilitate drug efflux. Indeed, conformational changes are induced in EmrA (97) and CusB (98) when they bind their respective ligands, indicating a possible transport role for these proteins. Dissociation constants in the nM and µM range have been identified for CusB and EmrA to their substrates, respectively (97, 98). Furthermore, the crystal structure of CusB bound to its cognate heavy metal ions was recently solved, illustrating the binding capacity of this membrane fusion protein (99). A three-dimensional structure of the entire tripartite complex may be needed to understand the role(s) of membrane fusion proteins. Using the three-dimensional structures of AcrA, AcrB, and TolC it became possible to develop a model of the elusive complex.

AcrA/AcrB/TolC COMPLEX By utilizing extensive cross-linking experiments along with previously published crystal structures, Symmons and Koronakis et al. 2009 were able to propose a structure for the complete AcrA/AcrB/TolC complex and expand on the previously suggested efflux mechanism (86). Overall, the cross-liking experiments revealed a 610 kD 270Å long complex with a stoichiometry of AcrB3-AcrA3-TolC3 per transporter. Docking of AcrA onto AcrB occurs along the exposed interface between adjacent subdomains PN1-PN2 and PC1-PC2 of the AcrB porter domain. Binding to AcrB orients AcrA slightly askew from vertical creating an appropriate distance for TolC to interact with both AcrA and AcrB. This interaction forces TolC to shift from a closed to an open state, which may allow antibiotics to be transported without any additional conformational shifts in TolC, whereby ligands traverse TolC through natural diffusion. It has been proposed that information regarding the bound state may be transduced from the asymmetrical structures of AcrB to AcrA and ultimately TolC, suggesting the membrane fusion and outer membrane proteins could perform a functional role in drug export. The proposed mechanism describes these two proteins to be functioning as a sort of peristaltic pump by pushing ligands out as they enter the TolC channel in a manner similar to food passage through the digestive tract. The modeled tripartite complex can be viewed in Figure 10. To gain 37

further insights and decipher a final mechanism for multidrug efflux, the crystal structure of the AcrA/AcrB/TolC awaits.

OTHER RND PROTEINS AcrB and MexB are the most characterized members of the RND transporter family and play important roles in drug resistance. Importantly, multidrug resistance is a feature displayed by organisms other than E. coli and P. aeruginos . Notably, this section will highlight the roles played by RND transporters in the biocidal resistance associated with other Gram-negative bacterial species, including Campylobacter jejuni and Neisseria gonorrhoeae. C. jejuni is the major causative agent of human enterocolitis and is responsible for more than 400 million cases of diarrhea each year worldwide (100). The Gram-negative microbe generally resides in the intestinal tracts of animals. Furthermore, it can be transmitted to humans via contaminated food, water, or raw milk. For antibiotic treatment of human campylobacteriosis, fluoroquinolones and macrolides are frequently prescribed (101). Unfortunately, Campylobacter has developed resistance to both classes of antimicrobials, especially to fluoroquinolones (102-104). In part, this resistance is due to the expression of drug efflux transporters. C. jejuni harbors multiple drug efflux transporters of different families (105). Among them, CmeABC, an RND-type efflux pump, is the primary antibiotic efflux system and is the best functionally characterized transporter in Campylobacter (106, 107). CmeABC consists of three components including an outer membrane protein (CmeC), an inner membrane drug transporter (CmeB), and a periplasmic fusion protein (CmeA). CmeABC contributes significantly to the intrinsic and acquired resistance of Campylobacter to structurally diverse antimicrobials including fluoroquinolones and macrolides (106-111). It has been found that CmeABC functions synergistically with target mutations in conferring and maintaining high- level resistance to fluoroquinolones and macrolides (108, 109, 111-113). This efflux pump also plays an important role in the emergence of fluoroquinolone-resistant 38

Campylobacter under selection pressure (114). Inactivation of CmeABC reduced the frequency of emergence of fluoroquinolone-resistant mutants, while overexpression of CmeABC increased the frequency of emergence of the mutants (114). In addition to conferring antibiotic resistance, CmeABC also has important physiological functions. It has been shown that CmeABC is functionally interactive with CmeDEF, another RND-type efflux pump, in maintaining optimal cell viability in Campylobacter , possibly by extruding endogenous toxic metabolites (115 ). Double mutations in CmeABC and CmeDEF appeared to be lethal to C. jejuni strain 11168 and significantly reduced the growth of strain 81-176 in conventional media. Another important function of CmeABC is bile resistance. As an enteric pathogen, C. jejuni must possess means to adapt in the animal intestinal tract, where bile acids are commonly present. Mutations of CmeB in C. jejuni resulted in a drastic increase in the susceptibility to various bile acids and a severe growth defect in bile-containing media or in chicken intestinal extracts (116 ). When inoculated into chickens, the CmeB mutant failed to colonize the inoculated birds. These findings provide compelling evidence that CmeABC, by mediating the resistance to bile acids, is essential for Campylobacter adaptation to the intestinal environment. These findings also strongly suggest that bile resistance is a natural function of this RND-type efflux pump. Within the strict human pathogen N. gonorrhoeae , the mtr (multiple transferable resistance) system was first identified by Maness and Sparling ( 117 ) when they isolated a spontaneous mutant that exhibited increased resistance to multiple, structurally diverse antimicrobial hydrophobic compounds. Similar to the story of AcrB and MexB, It was originally thought that mtr modified outer membrane permeability ( 118 ). However, subsequent cloning/sequencing experiments ( 119-122 ) showed that the mutation was located within a gene encoding a transcriptional repressor (MtrR) of a downstream, but transcriptionally divergent, operon ( mtrCDE ) encoding the tripartite MtrCDE efflux pump. Similar to other RND-type pumps of Gram-negative bacteria, the three proteins are a cytoplasmic membrane transporter (MtrD), a membrane fusion protein (MtrC) and an outer membrane channel protein 39

(MtrE). Directly or indirectly, other proteins also participate in efflux mediated by the pump. Veal and Shafer ( 123 ) identified an accessory protein (MtrF) which, for reasons that are not yet clear, is required for efflux activity when the host strain is expressing high levels of the pump during stressful conditions. Energy supplied by the TonB-ExbB-ExbD system is also needed for inducible antimicrobial resistance mediated by MtrC-MtrD-MtrE ( 124 ). There is evidence that gonococcal efflux pumps can contribute to levels of bacterial resistance to classical antibiotics since inactivation of efflux pump-encoding genes can enhance susceptibility to pump substrates ( 121,122, 125 ). Moreover, mutations that increase efflux pump gene expression can also increase antimicrobial resistance of N. gonorrhoeae . From a clinical perspective, the important question is whether efflux pumps can influence the efficacy of antibiotic treatment. In this respect, work on the MtrCDE efflux pump in clinical isolates indicates that this is indeed the case. As an example, over-expression of the mtrCDE operon due to mutations in the mtrR -coding sequence, which encodes a repressor of mtrCDE expression, or its promoter can provide gonococci with a two-fold increase in resistance to penicillin ( 125 ). However, when strains have co-resident mutations in other chromosomal genes that influence affinity of penicillin for penicillin-binding proteins (PBPs) or drug influx, resistance can become clinically significant ( ≥ 2.0 g/ml). The outbreak of penicillin-resistant gonorrhea that occurred in Durham, NC in the 1980s ( 126 ) due to a strain (termed FA6140) that had mtrR mutations as well as other mutations that both decreased binding of penicillin to PBP-1 and PBP-2 and influx of penicillin ( 127 ) is an example of the impact efflux can have on gonococcal resistance to antibiotics. Thus, while introduction of the mtrR mutations from penicillin-resistant strain FA6140 by transformation into highly penicillin-sensitive strain FA19 resulted in only a two-fold increase in resistance, inactivation of the mtrD gene, which encodes the inner membrane transporter protein, in resistant strain FA6140 decreased resistance from 4 to 0.25 g/ml. This decrease in resistance, due to loss of efflux activity, was intriguing, as it represented a transition from clinical resistance to sensitivity and provides support for the notion that inhibitors of efflux 40

pumps could reverse antibiotic resistance exhibited in pathogenic organisms. In addition to penicillin, gonococcal clinical isolates bearing mtrR mutations can express decreased susceptibility to macrolides and tetracycline ( 128 ). In fact, a cluster of azithromycin-resistant gonococci identified in patients was found to be overexpressing the mtrCDE genes ( 129 ). It has been suggested ( 130 ) that efflux pumps endow bacteria with the ability to resist natural or man-made antimicrobial agents in their local environment and that such resistance is important for their survival in ecosystems. For strict human pathogens, such as gonococci, that do not naturally exist for long periods of time outside the human body, these antimicrobial agents would be compounds (e.g., antimicrobial peptides, long chain fatty acids, bile salts, certain hormones) that are at the front-line of the innate host defense system. In this respect, the MtrCDE efflux pump appears to recognize certain antimicrobial peptides ( 131 ), progesterone ( 122 ) and bile salts ( 122 ). In support of the hypothesis that efflux pumps can promote bacterial survival during infection, Jerse et al . ( 132 ) found that the MtrCDE efflux pump is required for survival of gonococci in the lower genital tract of experimentally- infected female mice ( 122 ). This is a unique example of how a mechanism of antibiotic resistance can actually increase in vivo fitness and is likely due to the ability of the MtrC-MtrD-MtrE pump to recognize both classical antibiotics (e.g., penicillin) and host-derived antimicrobials.

INHIBITION Information may be extracted from the structural data of multi-drug binding proteins that could potentially aid in the rational design of drugs able to act as efflux pump inhibitors (EPIs). These EPIs would inhibit the function of MDR pumps and combat multidrug resistant bacteria. Multidrug export may be blocked in various fashions, including altering regulation, inhibiting assembly, closing or plugging the outer membrane channel, collapsing the energy coupling mechanism, or creating a binding competitor. 41

To inhibit protein expression, incorporating antisense oligonucleotides that interfere with transcription of the RND genes could reasonably be employed. Additionally, small interfering oligonucleotides would also function in inhibiting translation of the export proteins. Intriguingly, using the antisense strategy has previously shown promise, whereby the function of AcrAB was severely inhibited (133). Furthermore, studies indicated that chlorpromazine, an anti-psychosis compound, reduced expression of acrB. Unfortunately, the required concentrations of chlorpromazine to inhibit expression are extremely high and would produce deleterious side-effects if administered at those levels (134) A potential lead compound to block the assembly of the RND complex has yet to be identified, although a natural peptidase inhibitor has shown promise as a successful blocking agent. The use of this type of adjuvant may prove to be a useful mechanism to inhibit drug resistance. It is known that the assembly of the efflux complex is required for function. Indeed, globomycin, an inhibitor of the signal peptidase II (135, 136), which removes lipoprotein signal sequences from exported membrane proteins, may block the assembly of the complex and specifically inhibit AcrA. Initial results using globomycin showed that, when used at subinhibitory concentrations, the inhibitor could restore chloramphenicol transport and reduce resistance levels (137). One of the most attractive ideas is to design an EPI that would block the exit pathway of the outer membrane channel. In E. coli , TolC is utilized by various inner membrane transporters to export substrates, thus making it an ideal inhibitory target. In vitro assays showed that hexammine cobalt can severely inhibit the conductivity of TolC after reconstitution (138). To examine these results, TolC was co- crystallized with hexammine cobalt. The Hexammine cobalt was shown to form hydrogen bonds with Asp374 of each of the TolC monomers with strikingly high affinity (20 nM) (138). Although hexammine cobalt is not a potential EPI due to its harmful effects on cell viability, this illustrates how a blocker of an outer membrane channel could function and may suggest a design strategy for such an inhibitor. 42

Energy uncouplers such as carbonyl cyanide m-chlorophenylhydrazone (CCCP) and potassium cyanide are used in the laboratory settings to abolish substrate efflux by RND proteins (139). They presumably reduce the viability of the bacterium and cause cell death via the dissipation of the proton-motive force of the membrane. The downfall of these molecules is they are highly noxious and cytotoxic. This is exacerbated by the fact that these molecules are also substrates of the RND transporters themselves. The most promising EPIs have been designed and synthesized as antibiotic antagonists. Interestingly, these compounds exert no antibacterial effect on their own, but when co-administered with antibacterial compounds they are able to effectively reduce the MIC of co-administered antibiotics. Studies suggest that these EPIs, such as phenylalanyl arginyl b-naphthylamide (PAbN), are able to restore antibiotic activity in efflux mediated MDR bacteria by competing favorably for substrate recognition (139). The competition results in an increase in the intracellular concentration of the antibiotic to levels required for antimicrobial activity. In a study in 1999, PAbN and other homologous EPI compounds were characterized. It was observed that a group of peptidomimetic molecules were active at preventing fluoroquinolone efflux in P. aeruginosa strains overexpressing the MexAB efflux system (140). Several derivatives were made from this lead compound. Of these, PAbN along with two other derivatives that effectively inhibit efflux of quinolones, including levofloxacin were identified. Intriguingly, PAbN showed a broad range of activity against many efflux pumps of E. coli and other Gram-negative organisms (140). Importantly, it seems that PAbN may be effective against specific compounds that interact with overlapping regions, while not inhibiting efflux of other structurally distinct molecules. Indeed, the inhibitor PAbN only minimally affected the MIC of carbenicillin, although the MICs of erythromycin and levofloxacin were reduced to a degree similar to removal of the efflux pump. Presumably, this observation may be due to the unaffected molecules binding to a distinct site within the large binding cleft. This distinct binding site may remain unaltered by the EPI. Using this model for the rational design of an inhibitor, it may 43

be possible to design an EPI with the ability to inhibit efflux of a broad range of structurally dissimilar substrates. As shown, this EPI may be globally effective against efflux pumps from various infectious microorganisms.

OTHER FUNCTIONS Although this review has focused primarily on RND transporters and their function on multidrug resistance, this does not seem to be the primary function of these transporters. In fact, RND efflux systems are critical to the virulence of various microorganisms (142). The findings suggest that their physiological role in virulence is primarily the evasion of host derived antibacterial compounds, including bile salts, fatty acids, and steroid hormones. The efflux, thus allows the microorganism to survive in a deleterious environment. Indeed, studies indicate that AcrB has a much higher affinity to host derived bile acids, which reside within the intestinal environment, than for many antibiotics (143). This indicates that AcrB may have initially evolved as a bile acid efflux protein with its antibiotic efflux capacity only developing later. Furthermore, expression levels of AcrAB of Salmonella typhimurium (144) and AcrD, AcrE, EmrK, MdtA, and MdtE of E. coli (145) are increased in the presence of indole, a substrate within the intestinal tract, which promotes survival of the invasive bacteria in this deleterious niche. In a similar case, during macrophage infection by Salmonella , the expression levels of the MFS efflux gene emrAB and the RND transporter genes mdsABC and mdtABC are increased, suggesting a role during virulence (146) An intriguing study using strains of P. aeruginosa lacking MexAB-OprM further illustrated the importance of efflux pumps during infection (147). The strains lacking MexAB-OprM showed reduced ability to invade Madin-Darby Canine Kidney (MDCK) cells. P. aeruginosa invasiveness was restored using two methods. Not surprisingly, infectivity could be restored through the complementation of the disrupted genes. The second method involved the addition of the culture supernatant from MDCK cells previously infected with wild-type P. aeruginosa (harboring MexAB-OprM). Following addition of the supernatant, the invasiveness was completely restored. Intriguingly, this suggests that this RND efflux system 44

exports a virulence factor during macrophage infection that can be used to complement the deficient P. aeruginosa strain.

CONCLUDING REMARKS The availability of the three-dimensional structures of these efflux transporters has, for the first time, allowed us to propose a fundamental model of substrate export. However, there is still quite a mountain to climb in understanding the function of these proteins. To date, no RND transporter structures have been published outside of the HAE1 subfamily, although our group has recently solved the structure for the first RND protein outside of this subfamily (148). The crystal structure is of an RND protein among the heavy-metal efflux (HME-RND) subfamily of transporters. Interestingly, structures for the HME-RND protein from CusA of E. coli , solved in both apo-form and a liganded conformation, may illustrate a distinct mechanism to describe substrate transport. CusA, in the silver and copper bound form does not crystallize in an asymmetric conformation, thus proposing a possible new model, whereby a swinging arm motion in the periplasmic domain driven by the PMF may provide the conformational shifts that facilitate substrate export. Furthermore, the crystal structures of AcrB, MexB, and CusA are all from Gram- negative bacteria. As RND transporters exist in all other domains of organisms, it will be interesting to see structural and functional differences that exist for those RND transporters that do not utilize membrane fusion proteins or outer membrane channels to operate. Questions still remain as to how these RND proteins operate and whether the functional rotating model can be used to describe all RND transporters, or is this mechanism specific to transporters among the HAE1 subfamily? Among the HAE1 subfamily, the entire role for membrane fusion proteins in transporter activation also remains to be determined. Crystal structures of other RND proteins and a structure of the entire tripartite complex may be needed to answer these intriguing questions. Furthermore, as new crystal structures are published, it may be possible to design efflux pump inhibitors capable of knocking out an entire subfamily of multidrug exporters. An inhibitor capable of this would 45

make once futile antibiotics functional once again. Finally, as many of these transporters are globally as well as locally regulated, understanding the regulatory network that controls each of these RND pumps (discussed in chapters three and four) will provide futher insights into the complexity of multidrug resistance.

FIGURES AND TABLES

Figure 1

Figure 1 displays a schematic diagram of the five families of multidrug exporters. (A) The resistance/nodulation/cell division family (teal) illustrated as a tripartite complex with a membrane fusion protein (red) and outer membrane channel (blue). (B) The ATP binding cassette family utilizes the energy derived from ATP hydrolysis to drive drug export. The ATP binding domains are shown in yellow and the Transmembrane domains are dark blue. (C) The major facilitator family shown as a red and blue dimer. (D) The small multidrug resistance family is shown as a teal dimer. (E) The recently discovered multidrug and toxic compound extrusion family utilizes either sodium ion or proton influx to drive drug extrusion. A structure for this protein remains unsolved, but the protein is represented as an orange diamond.

Figure 2 TM1 EnvD ----MANFFIRRPIFAWVLAIILMMAGALAILQLPVAQYPTIAPPAVSVSANYPGADAQT 56 AcrB ----MPNFFIDRPIFAWVIAIIIMLAGGLAILKLPVAQYPTIAPPAVTISASYPGADAKT 56 MexB ----MSKFFIDRPIFAWVIALVIMLAGGLSILSLPVNQYPAIAPPAIAVQVSYPGASAET 56 Cnr MIESILSGSVRYRWLVLFLTAVVAVIGAWQLNLLPIDVTPDITNKQVQINSVVPTMSPVE 60 Czc MIKDFIETALRNRITTIITAIVAVLFGIWAWIDIRKEAYSDIADTQVRLIAKFPGKAAVE 60 : . : . . : : : * : . *: : : * .

EnvD VQDTVTQVIEQNMNGIDNLMYMSSTSDSAGSVTITLTFQSGTDPDIAQVQVQNKLQLATP 116 AcrB VQDTVTQVIEQNMNGIDNLMYMSSNSDSTGTVQITLTFESGTDADIAQVQVQNKLQLAMP 116 mexb VQDTVVQVIEQQMNGIDNLRYISSESNSDGSMTITVTFEQGTDPDIAQVQVQNKLQLATP 116 Cnr VEKRVTYPIETAIAGLNGVESTRSMSRN-GFSQVTVIFKESANLYFMRHEVSERLAQARP 119 Czc VEERVTLPIERVLNAIPKVAVRRSRTIN-GLVVFQFVFEDGTDDYFARMRLMERVADAD- 118 *:. *. ** : .: : * : . * . . *:..:: : : .: ::: *

EnvD LLPQEVQQQGISVEK------SSSSYLMVAGFVSDN 146 AcrB LLPQEVQQQGVSVEK------SSSSFLMVVGVINTD 146 mexb LLPQEVQRQGIRVTK------AVKNFLMVVGVVSTD 146 Cnr NLPENVEPQMGPVSTGLGEVFHYSVEYQYPDGTGASIKDGEPGWQSDGSFLTERGERLDD 179 Czc -IPEDVHPALGPMSS------PVGEIYRYVLESS 145 :*::*. : . .

EnvD PGTTQDDISDYVASNVKDTLSRLNGVGDVQLFG-AQYAMRIWLDADLLNKYKLTPVDVIN 205 AcrB GTMTQEDISDYVAANMKDAISRTSGVGDVQLFG-SQYAMRIWMNPNELNKFQLTPVDVIT 205 mexb GSMTKEDLSNYIVSNIQDPLSRTKGVGDFQVFG-SQYSMRIWLDPAKLNSYQLTPGDVSA 205 Cnr RVSRLAYLRTVQDWIIRPQLRTTPGVADVDSLGGYVKQFVVEPDTGKMAAYGVSYADLAR 239 Czc ENHTPMELRTIQDWIVMPKMLQIPGIADVVTFGGLPKQYHVVTSPDKLIRYKLTIGDVIR 205 : : : *:.*. :* : .. : : :: *:

EnvD QLKVQNDQIAAGQLGGTPALPGQQLNASIIAQTRFKNPEEFGKVTLRVNSDGSVVRLKDV 265 AcrB AIKAQNAQVAAGQLGGTPPVKGQQLNASIIAQTRLTSTEEFGKILLKVNQDGSRVLLRDV 265 mexb AIQAQNVQISSGQLGGLPAVKGQQLNATIIGKTRLQTAEQFEKILLKVNPDGSQVRLKDV 265 Cnr ALEDTNLSVGAN------FIRRSGESYLVRADARIKSADEISRAVIAHG--KMSHHVGQV 291 Czc AIQENNLNTGGN------LLLQGEQGFPIRSLGAIRDPKHIENIVVKTVN-GVPVFIRDL 258 :: * . ... : . : . : ...: . : : ::

EnvD ARVELGGENYNVIARINGK------PAAGLGIKLATGANALDTAKAIKAKLAELQP 315 AcrB AKIELGGENYDIIAEFNGQ------PASGLGIKLATGANALDTAAAIRAELAKMEP 315 mexb ADVGLGGQDYSINAQFNGS------PASGIAIKLATGANALDTAKAIRQTIANLEP 315 Cnr ARVKIGGELRSGAASRNGN------ETVVGSALMLVGANSRTVAQAVGDKLEQISK 341 Czc GSVEISHPIPSGVLGYTVQNDEEGLIDVDSSVQGLVAMRRWGDPNEMGERIREKVKEINE 318 . : :. . . . : : .:. . : : ::. TM2 TM3 EnvD -FFPQGMKVLYPYDTTPFVQLSIHEVVKTLFEAIMLVFLVMYLFLQNMRATLIPTIAVPV 374 AcrB -FFPSGLKIVYPYDTTPFVKISIHEVVKTLVEAIILVFLVMYLFLQNFRATLIPTIAVPV 374 mexb -FMPQGMKVVYPYDTTPVVSASIHEVVKTLGEAILLVFLVMYLFLQNFRATLIPTIAVPV 374 Cnr -TLPPGVVIVPTLNRSQLVIATIETVAKNLIEGALLVVAILFALLGNWRAATIAALVIPL 400 Czc NYLPKGVQLRNTYDRTDLVNYTLRTIGKTLVEGVVVVSLVLIFFIGSVRASLVVVATIPF 378 :* *: : . : : .* ::. : *.* *. ::* :: :: . **: : . .:*. TM4 EnvD VLLGTFAILAAFGYSINTLTMFGMVL AIGLLV DD AIVVV ENVERVMMEDKLPPK------428 AcrB VLLGTFAVLAAFGFSINTLTMFGMVL AIGLLV DD AIVVV ENVERVMAEEGLPPK------428 mexb VLLGTFGVLAAFGFSINTLTMFGMVL AIGLLV DD AIVVV ENVERVMAEEGLSPR------428 Cnr SLLVSAIGMNQFHISGNLMSLG--AL DFGLII DGAVIIV ENSLRRLAERQHREGRLLTLD 458 Czc AMLFAFLLMNMTGIPASLLSLG--AI DFGIIV DGAVIMV ENIMRRYRDATPEEKSHG--- 433 :* : : . . ::: .: :*::: *.*:::* ** * :

TM5 TM6 EnvD ---EATEKSMSQIQGALVGIAMVLSAVFIPMAFFGGSTGAIYRQFSITIVSAMALSVLVA 485 AcrB ---EATRKSMGQIQGALVGIAMVLSAVFVPMAFFGGSTGAIYRQFSITIVSAMALSVLVA 485 mexb ---EAARKSMGQIQGALVGIAMVLSAVFLPMAFFGGSTGVIYRQFSITIVSAMALSVIVA 485 Cnr DRLQEVVQSSREMVRPTVYGQLVIFMVFLPSLTFQGVEGKMFSPMVITLMLALASAFVLS 518 Czc -ILAFTRDAASEVGTEILFSILIIILAYLPIFSFERIEGRLFKPMAFTISFAILGALVFA 492 . .: :: : ::: .::* * * :: : :*: *: :.:.: EnvD LILTPALCATLLKPVS-AEHHENKGGFFGWFNTTFDHSVNHYTNSVGKILGSTGRYLLIY 544 AcrB LILTPALCATMLKPIAKGDHGEGKKGFFGWFNRMFEKSTHHYTDSVGGILRSTGRYLVLY 545 mexb LILTPALCATMLKPIEKGDHGEHKGGFFGWFNRMFLSTTHGYERGVASILKHRAPYLLIY 545 Cnr LTFVPAMVAVMLRKKVAETEVR------VIVATKESYRPWLEHAVARPMPFIGAG 567 Czc MAVIPVIMSIIYKHYFESKNPGP----IEWHNPFYDWIEARYKRLIEFIVDRSKKAVRYT 548 : . *.: : : : . * : : : TM7 EnvD ALIVAGMVVLFLR-LPSSFLPEEDQG-VFLTMIQLPAGATQERTQKVLDQVTDYYLKNEK 602 AcrB LIIVVGMAYLFVR-LPSSFLPDEDQG-VFMTMVQLPAGATQERTQKVLNEVTHYYLTKEK 603 mexb VVIVAGMIWMFTR-IPTAFLPDEDQG-VLFAQVQTPPGSTAERTQVVVDSMREYLLEKES 603 Cnr IATVAVATVAFTF-VGREFMPTLDELNLNLSSVRIPSTSIDQSVAIDLPLERAVLSLPEV 626 Czc FSVVTIFLAIGMFSLGTEFLPEMDEGGFNIRIFFPVGISLPEARKFMPKIRQTVYKNEQV 608 *. : *:* *: . : . : : :

EnvD ANVESVFTVNGFSFSGQAQNAGMAFVSLKPWEERNGDENSAEAVIHRAKMELGKIRDGFV 662 AcrB NNVESVFAVNGFGFAGRGQNTGIAFVSLKDWADRPGEENKVEAITMRATRAFSQIKDAMV 663 mexb SSVSSVFTVTGFNFAGRGQSSGMAFIMLKPWEERPGGENSVFELAKRAQMHFFSFKDAMV 663 Cnr QTVYSKAGTASLAADPMPPNASDNYIILKPKSEWPEGVTTKEQVIERIREKTAPMVGNNY 686 Czc SVVISQLGRNDDGTDPLPPNRLEVLIGLKDYSKWKEKITKQELLLRMKNDLEATLPGARI 668 * * . . : ** . .. : : .

EnvD IPFNMPAIVELGTATGFDFELIDQAGLGHDALTQARNQLLGMAAQHPASLVSVRPNGLED 722 AcrB FAFNLPAIVELGTATGFDFELIDQAGLGHEKLTQARNQLLAEAAKHPDMLTSVRPNGLED 723 mexb FAFAPPSVLELGNATGFDLFLQDQAGVGHEVLLQARNKFLMLASQNP-ALQRVRPNGMSD 722 Cnr DVTQPIEMRFNELIGGVRSDVAVKVYGENLDELAATAQRIAAVLKKTPGATDVRVPLTSG 746 Czc SFSQPIMDNLSEAIMGTIADLAVFVSGNDLKIMRGIGNEVLKEIKEMKGASEYGIEQEAE 728 * : . . . : : :.

EnvD TAQFKLEVDQEKAQALGVSLSDINQTISTALGGTYVNDFIDRGR------VKKLYVQA 774 AcrB TPQFKIDIDQEKAQALGVSINDINTTLGAAWGGSYVNDFIDRGR------VKKVYVMS 775 mexb EPQYKLEIDDEKASALGVSLADINSTISIAWGSSYVNDFIDRGR------VKRVYLQG 774 Cnr FPTFDIVFDRAAIARYGLTVKEVADTISTAMAGRPAGQIFDGDR------RFDIVIRL 798 Czc SPQLTISINREAAARFGINVIDIQQMIEAAIGMQRISTLYEGPSDIPPKTPARFGIVVRF 788 . : .: *:.: :: : * . . : : : :

EnvD DAKFRMLPEDVDKLYVRSANGE-----MVPFSAFTTSHWVYGSPRLERYNGLPSMEIQGE 829 AcrB EAKYRMLPDDIGDWYVRAADGQ-----MVPFSAFSSSRWEYGSPRLERYNGLPSMEILGQ 830 mexb RPDARMNPDDLSKWYVRNDKGE-----MVPFNAFATGKWEYGSPKLERYNGVPAMEILGE 829 Cnr PGEQRENLDVLGALPVMLPLSEGQARASVPLRQLVQFRFTQGLNEVSRDNGKRRVYVEAN 858 Czc SKDYRASKQAIENMPIISPKGE-----RIPLSQLADIEVIDGPTMIFRQEGRRVVTVRTN 843 . * : : : .: :*: : . * : * :* : : : TM8 EnvD AAPG--TSSGDAMALMENLASKLPAGIGYDWTGMSYQERLSGNQAPALVAISFVVVFLCL 887 AcrB AAPG--KSTGEAMELMEQLASKLPTGVGYDWTGMSYQERLSGNQAPSLYAISLIVVFLCL 888 mexb PAPG--LSSGDAMAAVEEIVKQLPKGVGYSWTGLSYEERLSGSQAPALYALSLLVVFLCL 887 Cnr VGGRDLGSFVDDAAARIAKEVKLPPGMYIEWGGQFQNLQAATKRLAIIVPLCFILIAATL 918 Czc IRGRDQGGFVSELQKRVKKKIKLPDGYEIRFGGQYENLSRVGKKLGIVIPITVLIIFGVL 903 . . :** * : * : .: : .: .::: * TM9 YM10 EnvD AALYESWSIPVSVMLVVPLGIVGVLLAATLFNQKNDVYFMVGLLTTIGLSA KNAILIVEF 947 AcrB AALYESWSIPFSVMLVVPLGVIGALLAATFRGLTNDVYFQVGLLTTIGLSA KNAILIVEF 948 mexb AALYESWSIPFSVMLVVPLGVIGALLATSMRGLSNDVFFQVGLLTTIGLSA KNAILIVEF 947 Cnr YMAIGSAALTATVLTASPLALAGGVFALLLRGIPFSISAAVGFIAVSGVAV LNGLVLISA 978 Czc YLLYRNLKYVYVALACIPLSLLGGIYALLLRGYYFNVSGGVGFISLFGIAT MAGVLFVSR 963 . .: **.: * : * : . .: **::: *::. .::::.

TM11 EnvD AKDLMEK-EGKGVVEATLMAVRMRLRPILM TSLAFILGVLPLAISNGAGSGAQNAVGIGV 1006 AcrB AKDLMDK-EGKGLIEATLDAVRMRLRPILM TSLAFILGVMPLVISTGAGSGAQNAVGTGV 1007 mexb AKELHE--QGKGIVEAAIEACRMRLRPIVM TSLAFILGVVPLAISTGAGSGSQHAIGTGV 1005 Cnr IRKRLD--DGMAPDAAVIEGAMERVRPVLM TALVASLGFVPMAIATGTGAEVQKPLATVV 1036 Czc TNHLLVEEPDISTKAAVKKAAVIQLRPMLM TMLLALLGLIPATLGTGVGSDVQRPLATVI 1023 .. . . *. . ::**::* * * **.:* .:..*.*: *..:. : TM12 EnvD MGGMVSATLLAIFFVPVFFVVIRRCFKG------1034 AcrB MGGMVTATVLAIFFVPVFFVVVRRRFSRKNEDIEHSHTVDHH------1049 mexb IGGMVTATVLAIFWVPLFYVAVSTLFKDEASKQQAEAEKGQ------1046 Cnr IGGLVTATVLTLFVLPALCGIVLKRRTAGRPEAQAALEA------1075 Czc VGGLFSAMCLVLTILPSLYLVVVGERKPSAEELEEMSHKKHIPFLDFVNELSEEPLEEED 1083 :**:.:* *.: :* : : .

EnvD ------AcrB ------mexb ------Cnr ------Czc EDDEPVSKKKKKPAKKRKKT 1103

Figure 2 : A sequence alignment of members of the RND transporter family using ClustalW. TM segments are highlighted as follows: α1 and α7 are yellow, α2 and α8 are green, α3 and α9 are teal, α4 and α10 are magenta, α5 and α11 are gray, and α6 and α12 are highlighted red. Residues critical to the proton relay pathway are colored blue. Alighnment was performed using clustalw (149). Transmembrane regions were predicted using TMPRED (150)

Figure 3

Figure 3: Schematic representation of the topologyof the MexB protein and members of the family. Symbols: filled circle , positively charged amino acid; shaded circle , negatively charged amino acid; open circle, uncharged amino acid. Amino acid residues in transmembrane segments were expressed by a one-letter code. Arrow : one-letter code number in parentheses represents the fusion site-amino acid residue-AP activity (units). Shaded rectangles around circles , putative TMS. a through e are weak hydrophobic segments. The overall topology is suggested to be common throughout the RND family. This picture was taken from Guan et al. 1999 (54).

Figure 4

Figure 4 . A schematic representation of the phospholipids extrusion assay of reconstituted AcrB (blue rectangles). AcrB was mixed with acceptor liposomes, wictout any protein or labeled lipids. The NBD fluorescence from the NBD-labeled phosphatidylethanolamine (orange) in the donor vesicles is initially quenched by fluorescent emission energy transfer to the rhodamine-labeled phosphatidylethanolamine (red). Following export of phospholipids to the acceptor vesicles, the fluorescent emission increases from NBD as the surface density of labeled phospholipids decreases indicating transport. The diagram was taken from a 2009 review by Nikaido and Takatsuka (151).

Figure 5

Figure 5 is a ribbon model of a single AcrB monomer taken from pdb accession code 1IWG. Subdomains are individually colored, whereby TM helices 1-6 are brown, TM helices 7-12 are purple, PC1 is Green, PC2 is yellow, PN1 is blue, PN2 is red, DC is orange, and DN is colored cyan. Subdomains PC1, PC2, PN1, and PN2 comprise the porter domain of AcrB, while the DN and DC subdomains come together to form the TolC-docking domain, respectively. The image was created using pymol (152). 53

Figure 6 A K940

K940

D408

D407

Figure 6 . A) A close-up depiction of the proton relay network of wild-type AcrB. Residues D408, K940, D407, and T978 are all within hydrogen bond distance of less than 3.2Å. The image was rendered using pymol and PDB accession code 1IWG. B) A similar view of Mutated T978A, and non functional AcrB. In this ribbon diagram the proton relay network is disturbed, therefore no efflux antibiotic efflux occurs. The PDB accession code utilized for the mutant AcrB was 2HQG. 54

Figure 7

Figure 7 . A ribbon model of AcrB viewed from the top, perpendicular to the membrane plane. Features illustrated include the central pore (purple circle) which extends from the central cavity to the TolC docking domain funnel (orange circle). The vestibules, which are located at the monomer interfaces and lead to the central cavity, are depicted with purple arcs. Additionally, the suggested binding clefts are displayed as teal arcs. Individual AcrB monomers are displayed as red, green, and blue respectively. This image was created using Pymol and PDB accession code 1IWG.

Figure 8A

Figure 8B

Figure 8 . A) Surface electrostatic potential of AcrB with bound R6G (yellow). The model was built using PDB 1T9V, which has a mutated Asp109 to alanine. This mutation allowed R6G to enter a second binding site, with the first being the central binding cavity (entrance from the vestibules) and the second locating to the binding cleft. B) After removal of the TolC-docking domain and transmembrane domain, the binding cleft can be easily displayed along with bound R6G (yellow). Subdomains, vestibules and binding clefts are labeled black, purple, and teal, respectively. Both images were made using Pymol (152).

Figure 9

Figure 9. Tunnels in the porter domain of trimeric AcrB peristaltic drug efflux pump. The AcrB monomers are presented in blue (access), yellow (binding) and red (extrusion). The tunnels are highlighted as green surfaces in a ribbon model of the AcrB trimer and might function as transport paths of drugs. Tunnel 1 might serve as entrance for drugs from the outer leaflet of the inner membrane towards the hydrophobic substrate binding pocket. Tunnel 2 might serve as an alternative entrance for substrates entering via the periplasm or as an exit duct for non- substrates. Tunnel 3 in the open monomer is the exit pathway for substrates towards TolC and the outside medium. Inset: in the T monomer (yellow), a hydrophobic pocket is defined by phenylalanine, valine, isoleucine and tyrosine side chains at the PN2/PC1 interface. Bound minocycline is depicted with the observed electron density in a 2Fo-Fc electron density map contoured 1 s (T. Eicher, M. Seeger, K.M. Pos and colleagues, unpublished data). Panels (A) and (B) represent in each case a one-third conversion of a cycle. This figure was originally published by Eicher, Brandstatter, and Pos (153).

Figure 10

Figure 10 . Schematic drawing based on the X-ray structures of the tripartite multidrug efflux system AcrAB–TolC of Gram-negative E. coli . AcrB (RND component, in blue color) resides in the inner membrane and is responsible for substrate recognition/selection and energy transduction. Drugs are captured at the outer leaflet of the inner membrane and extruded in a coupled exchange with protons. TolC (OMF component, yellow) forms a pore in the outer membrane which is extended by a long periplasmic conduit. AcrA (MFP component, red) mediates contact between AcrB and TolC. The presence of all three components is essential for the MDR phenotype. From Seeger (32).

References

1. Piddock, L. J. V. (2006) Clinically relevant bacterial chromosomally encoded multi-drug resistance efflux pumps, Clinl. Microbiol. Rev. 19 , 382-402.

2. Levy, S. (1992) Active efflux mechanisms for antibiotic resistance, Antimicrob. Agents Chemother. 36 , 695–703.

3. Juliano, R. L., Ling, V. (1976) A surface glycoprotein modulating drug permeability in Chinese hamster ovary cell mutants, Biochim Biophys Acta. 455 , 152-62.

4. Debenham, P. G., Kartner, N., Siminovitch, L., Riordan, J. R., and Ling, V. (1982) DNA-mediated transfer of multiple drug resistance and plasma membrane glycoprotein expression, Mol Cell Biol. 8, 881–889.

5. Schinkel, A. H. (1999) P-Glycoprotein, a gatekeeper in the blood–brain barrier, Adv. Drug Del. Rev. 36 , 179-194

6. Levy, S. (1992) Active efflux mechanisms for antibiotic resistance, Antimicrob. Agents Chemother. 36 , 695–703.

7. Bolhuis, H., van Veen H. W., Poolman B., Driessen, A. J., and Konings, W.N. (1997) Mechanisms of multidrug transporters, FEMS Microbiol. Rev. 21 , 55– 84.

8. Martinez-Martinez, L., Pascual, A., Jacoby, G. A. (1998) Quinolone resistance from a transferable plasmid, Lancet. 351 , 797-799.

9. Courvalin, P. (1994) Transfer of antibiotic resistance genes between gram- positive and gram-negative bacteria, Antimicrob. Agents Chemother. 38 , 1447-1451.

10. Higgins, C. F. (1992) ABC transporters: from microorganisms to man, Annu. Rev. Cell Biol. 8 , 67-113.

11. Tseng, T. T., Gratwick, K. S., Kollman, J., Park, D., Nies, D. H., Goffeau, A. and Saier, M. H., Jr. (1999) The RND permease superfamily: an ancient, ubiquitous and diverse family that includes human disease and development proteins, J. Mol. Microbiol. Biotechnol. 1 , 107–125.

12. Brown, M. H., Paulsen, L. T., and Skurray, R. A. (1999) The multidrug efflux protein NorM is a prototype of a new family of transporters, Mol. Microbiol. 31 , 394–395. 60

13. Morita, Y., Kodama, K., Shiota, S., Mine, T., Kataoka, A., Mizushima, T., and Tsuchiya, T. (1998) NorM, a putative multidrug efflux protein, of Vibrio parahaemolyticus and its homolog in Escherichia coli , Antimicrob. Agents Chemother. 42, 1778-1782.

14. Griffith, J. K., Baker, M. E., Rouch, D. A., Page, M. G., Skurray, R. A., Paulsen, I. T., Chater, K. F., Baldwin, S. A., and Henderson, P. J. (1992) Membrane transport proteins: Implications of sequence comparisons, Curr. Opin. Cell Biol. 4 , 684-695.

15. Marger, M., and Saier, M. H., Jr. (1993) A major superfamily of transmembrane facilitators that can catalyze uniport, symport and antiport, Trends Biochem. Sci. 18 , 13–20.

16. Pao, S. S., Paulsen, I. T., and Saier, M. H. Jr. (1998) Major facilitator superfamily, Microbiol. Mol. Biol. Rev. 62 , 1-34.

17. Paulsen, I. T., Skurry, R. A., Tam, R., Saier, M. H., Jr., Turner, R. J., Weiner, J. H., Goldberg, E. B., and Grinius, L. L. (1996) The SMR family: A novel family of multidrug efflux proteins involved with the efflux of lipophilic drugs, Mol. Microbiol. 19 , 1167–1175.

18. Schweizer, H. P. (2003) Efflux as a mechanism of resistance to antimicrobials in Pseudomonas aeruginosa and related bacteria: unanswered questions, Genet. Mol. Res. 31 , 48–62.

19. Nikaido, H. (1998) Antibiotic resistance caused by Gram-negative multidrug efflux pumps, Clin. Infect. Dis. 27 , S32-S41.

20. Moore, R. A., DeShazer, D., Reckseidler, S., Weissman, A., and Woods, D. E. (1999) Efflux-mediated aminoglycoside and macrolide resistance in Burkholderia pseudomallei, Antimicrob. Agents. Chemother. 43 , 465-470

21. Livermore, D.M. (2003) Bacterial resistance: origins, epidemiology, and impact, Clin Infect Dis. 36 , S11-S23.

22. Poole, K. (2003) Overcoming multidrug resistance in Gramnegative bacteria, Curr Opin Invest Drugs 4 , 128-139.

23. Helms, M., Vastrup, P., Gerner-Smidt, P., and Molbak, K. (2002) Excess mortality associated with antimicrobial drug-resistant Salmonella typhimurium. Emerg. Infect. Dis. 8 , 490-497.

24. Paulsen, I. T., Brown, M. H., and Skurray, R. A. (1996) Proton-dependent multidrug efflus systems, Microbiol Rev. 60 , 575-608 61

25. Zgurskaya, H. I., and Nikaido, H. (2000) Cross-Linked Complex between Oligomeric Periplasmic Lipoprotein AcrA and the Inner-Membrane-Associated Multidrug Efflux Pump AcrB from Escherichia coli , J. Bacteriol. 182 , 4264- 4267.

26. Mikolosko, J., Bobyk, K., Zgurskaya, H. I. and Ghosh, P. (2006) Conformational flexibility in the multidrug efflux system protein AcrA, Structure 14 , 577-587.

27. Fralick, J. A. (1996) Evidence that TolC is required for functioning of the Mar/AcrAB efflux pump of Escherichia coli , J. Bacteriol. 178 , 5803-5805.

28. Koronakis, V., Sharff, A., Koronakis, E., Luisi, B., and Hughes, C. (2000) Crystal structure of the bacterial membrane protein TolC central to multidrug efflux and protein export, Nature 405 , 914-919.

29. Murakami, S., Nakashima, R., Yamashita, E., and Yamaguchi, A. (2002) Crystal structure of bacterial multidrug efflux transporter AcrB, Nature 419 , 587-593.

30. Yu, E. W., McDermott, G., Zgurskaya, H. I., Nikaido, H., and Koshland, D. E., Jr. (2003) Structural basis of multiple drug-binding capacity of the AcrB multidrug efflux pump, Science 300 , 976-980.

31. Murakami, S., Nakashima, R., Yamashita, E., Matsumoto, T., and Yamaguchi, A. (2006) Crystal structures of a multidrug transporter reveal a functionally rotating mechanism, Nature 443 , 173-179.

32. Seeger, M. A., Schiefner, A., Eicher, T., Verrey, F., Diederichs, K., and Pos, K. M. (2006) Structural asymmetry of AcrB trimer suggests a peristaltic pump mechanism, Science 313 , 1295-1298.

33. Sennhauser, G., Amstutz, P., Briand, C., Storchenegger, O., and Grutter, M. G. (2007) Drug export pathway of multidrug exporter AcrB revealed by DARPin inhibitors, PLoS Biol. 5, e7.

34. Sennhauser, G., Bukowska, M. A., Briand, C., and Grütter, M. G. ( 2009) Crystal structure of the multidrug exporter MexB from Pseudomonas aeruginosa , J. Mol. Biol. 389, 134-45.

35. Saier, M. H. Jr., Tam, R., Reizer, A., and Reizer, J. (1994) Two novel families of bacterial membrane proteins concerned with nodulation, cell division and transport, Molecular Microbiol. 11 , 841-847.

36. Nies, D. H. (1992) CzcR and CzcD, gene products affecting regulation of resistance to cobalt, zinc, and cadmium ( czc system) in Alcaligenes eutrophus, J. Bacteriol. 174 , 8102-8110. 62

37. Nies, D. H. (1992) Resistance to cadmium, cobalt, zinc, and nickel in microbes, Plasmid 27 , 17-28.

38. Nies, D. H., Nies, A., Chu, L., and Silver, S. (1989) Expression and nucleotide sequence of a plasmid-determined divalent cation efflux system from Alcaligenes eutrophus. Proc. Natl. Acad. Sci. USA 86 , 7351-7355.

39. Collard, J.-M., Provoost, A., Taghavi, S., and Mergeay, M. (1993) A new type of Alcaligenes eutrophus CH34 zinc resistance generated by mutations affecting regulation of the cnr cobalt-nickel resistance system. J. Bacteriol. 175 , 779-784.

40. Liesegang, H., Lemke, K., Siddiqui, R. A., and Schlegel, H.-G. (1993) Characterization of the inducible nickel and cobalt resistance determinant cnr from pMOL28 of Alcaligenes eutrophus CH34, J. Bacteriol . 175, 767-778

41. Baev, N., Endre, G., Petrovics, G., Banfalvi, Z., and Kondorosi, A. (1991) Six nodulation genes of nod box locus 4 in Rhizobium meliloti are involved in nodulation signal production: nodM codes for D-glucosamine synthetase, Mol. Gen. Genet. 228 , 113-124.

42. Gottfert, M. (1993) Regulation and function of rhizobial nodulation genes, FEMS Microbiol. Revs. 104 , 39-64.

43. Klein, J. R., Henrich, B., and Plapp, R. (1991) Molecular analysis and nucleotide sequence of the envCD operon of Escherichia coli, Mol. Gen. Genet. 230 , 230-240.

44. Erickson, R.P., Aviles, R.A., Zhang, J., Kozloski, M.A., Garver, W.S., and Heidenreich, R.A. (1997) High-resolution mapping of the spm (Niemann-Pick Type C) locus on mouse chromosome 18, Mamm. Genome. 8 , 355–356.

45. Vanier, M.T., Duthel, S., Rodriguez-Lafrasse, C., Pentchev, P., and Carstea, E.D. (1996) Genetic heterogeneity in Niemann-Pick C disease: A study using somatic cell hybridization and linkage analysis, Am. J. Hum. Genet. 58 , 118– 125.

46. Watari, H., Blanchette-Mackie, E.J., Dwyer, N.K., Glick, J.M., Patel, S., Neufeld, E.B., Brady, R.O., Pentchev, P.G., and Strauss, J.F., III. (1999) Niemann-Pick C1 protein: Obligatory roles for N-terminal domains and lysosomal targeting in cholesterol mobilization, Proc. Natl. Acad. Sci. USA. 96 , 805–810.

47. Ma, D., Cook, D.N., Alberti, M., Pon, N.G., Nikaido, H., and Hearst, J.E. (1993) Molecular cloning and characterization of acrA and acrE genes of Escherichia coli, J. Bacteriol. 175, 6299–6313. 63

48. Nakamura, H. (1965) Gene-controlled resistance to acriflavine and other basic dyes in Escherichia coli, J. Bacteriol. 90 , 8-14.

49. Nakamura, H. (1968) Genetic determination of resistance to acriflavine, phenethyl alcohol, and sodium dodecyl sulfate in Escherichia coli, J. Bacteriol. 96 , 987-996.

50. Ma, D., Cook, D. N., Alberti, M., Pon, N. G., Nikaido, H., and Hearst, J. E. (1993) Molecular cloning and characterization of acrA and acrE genes of Escherichia coli, J. Bacteriol. 175 , 6299-6313.

51. Poole, K., Krebes, K., McNally, C., and Neshat, S. (1993) Multiple antibiotic resistance in Pseudomonas aeruginosa : evidence for involvement of an efflux operon, J. Bacteriol. 175 , 7363-7372.

52. Ziha-Zarifi, I., Llanes, C., Kohler, T. Pechere, J.-C., and Plesiat, P. (1999) In vivo emergence of multidrug-resistant mutants of Pseudomonas aeruginosa overexpressing the active efflux system MexA-MexB-OprM, Antimicrob. Agents Chemother. 43 , 287-291.

53. Nikaido, H., Basina, M., Nguyen, V., and Rosenberg, E.Y. (1998) Multidrug efflux pump AcrAB of Salmonella typhimurium excretes only those beta- lactam antibiotics containing lipophilic side chains, J. Bacteriol. 180, 4686– 4692.

54. Guan, L., Ehrmann, M., Yoneyama, H., and Nakae, T. (1999) Membrane topology of the xenobiotic-exporting subunit, MexB, of the MesA,B-OprM extrusion pump in Pseudomonas aeruginosa, J. Bacteriol. 274 , 10517-10522.

55. Gotoh, N., Kusumi, T., Tsujimoto, H., Wada, T., and Nishino, T. (1999) Topological analysis of an END family transporter, MexD of Pseudomonas aeruginosa, FEBS Letters 458 , 32-36.

56. Goldberg, M., Pribyl, T., Juhnke, S., and Nies, D. (1999) Energetics and topology of CzcA, a cation/proton antiporter of the Resistance-Nodulation-Cell Division protein family, J. Biol. Chem. 37 , 26065-26070.

57. Paulsen, I. T., Brown, M. H., and Skurray, R. A. (1996) Proton-dependent multidrug efflux systems, Microbiol. Rev. 60 , 575-608

58. Ma, D., Cook, D. N., Alberti, M., Pon, N. G., Nikaido, H., and Hearst, J.E. (1995) Genes acrA and acrB encode a stress-induced efflux system of Escherichia coli, Mol. Microbiol.16, 45–55.

59. Li X.-Z., Nikaido H., and Poole K. (1995) Role of MexA-MexB-OprM in antibiotic efflux in Pseudomonas aeruginosa, Antimicrob Agents Chemother , 39, 1948-1953. 64

60. Poole, K., Tetro, K., Zhao, Q., Neshat, S., Heinrichs, D. E., and Bianco, N. (1996) Expression of the multidrug resistance operon mexA-mexB-oprM in Pseudomonas aeruginosa : mexR encodes a regulator of operon expression, Antimicrob. Agents Chemother. 40 , 2021-2028.

61. Zgurskaya, H. I., and Nikaido, H. (1999) Bypassing the periplasm: reconstitution of the AcrAB multidrug efflux pump of Escherichia coli , Proc. Natl. Acad. Sci. USA 96 , 7190-7196.

62. Nagano, K., and Nikaido, H. (2009) Kinetic behavior of the major multidrug efflux pump AcrB of Escherichia coli , Proc. Natl. Acad. Sci. USA. 106 , 5854- 5858.

63. Thanassi, D. G., Cheng, L. W., and Nikaido, H. (1997) Active efflux of bile salts by Escherichia coli, J. Bacteriol . 179, 2512-2518.

64. Aires, J. R., and Nikaido, H. (2005) Aminoglycosides are captured from both periplasm and cytoplasm by the AcrD multidrug efflux transporter of Escherichia coli , J. Bacteriol. 187 , 1923-1929.

65. Elkins, C. A., and Nikaido, H. (2002) Substrate specificity of the RND-type multidrug efflux pumps AcrB and AcrD of Escherichia coli is determined predominantly by two large periplasmic loops, J. Bacteriol. 184 , 6490-6498.

66. Goldberg, M., Pribyl, T. S. Juhnke, D.H. Nies, Energetics and topology of CzcA, a cation/proton antiporter of the resistance-nodulation-cell division protein family, J. Biol. Chem. 274 (1999) 26065–26070.

67. Guan, L., and Nakae, T. (2001) Identification of essential charged residues in transmembrane segments of the multidrug transporter MexB of Pseudomonas aeruginosa, J. Bacteriol. 183, 1734–1739.

68. Takatsuka, Y., and Nikaido, H. (2006) Threonine-978 in the transmembrane segment of the multidrug efflux pump AcrB of Escherichia coli is crucial for drug transport as a probable component of the proton relay network, J. Bacteriol. 188 , 7284–7289.

69. Sulavik, M. C. (2001) Antibiotic susceptibility profiles of Escherichia coli strains lacking multidrug efflux pump genes, Antimicrob. Agents Chemother. 45 , 1126–1136.

70. Nikaido, H. (1996) Multidrug efflux pumps of gram-negative bacteria, J. Bacteriol. 178, 5853–5859.

71. Gottesman, M. M., and Pastan, I. (1993) Biochemistry of multidrug resistance mediated by the multidrug transporter, Annu. Rev. Biochem. 62 , 385-427. 65

72. Li, X. Z., Ma, D., Livermore, D. M., and Nikaido, H. (1994) Role of efflux pump(s) in intrinsic resistance of Pseudomonas aeruginosa : active efflux as a contributing factor to beta-lactam resistance, Antimicrob. Agents Chemother . 38, 1742–1752.

73. Nikaido, H., Basina, M., Nguyen, V., and Rosenberg, E. Y. (1998) Multidrug efflux pump AcrAB of Salmonella typhimurium excretes only those beta- lactam antibiotics containing lipophilic side chains, J. Bacteriol. 180, 4686– 4692.

74. A. Lee, W. Mao, M.S. Warren, A. Mistry, K. Hoshino, R. Okumura, H. Ishida, O. Lomovskaya, Interplay between efflux pumps may provide either additive or multiplicative effects on drug resistance, J. Bacteriol. 182 (2000) 3142– 3150.

75. E.B. Tikhonova, Q. Wang, H. Zgruskaya, Chimeric analysis of the multicomponent multidrug efflux transporters from Gram-negative bacteria, J. Bacteriol. 184 (2002) 6499–6507.

76. C.A. Elkins, H. Nikaido, Substrate specificity of the RND-type multidrug efflux pumps AcrB and AcrD of Escherichia coli is determined predominantly by two large periplasmic loops, J. Bacteriol. 184 (2002) 6490–6498.

77. W. Mao, M.S. Warren, D.S. Black, T. Satou, T. Murata, T. Nishino, N. Gotoh, O. Lomovskaya, On the mechanism of substrate specificity by resistance nodulation division (RND)-type multidrug resistance pumps: the large periplasmic loops of MexD from Pseudomonas aeruginosa are involved in substrate recognition, Mol. Microbiol. 46 (2002) 889–901.

78. J.K. Middlemiss, K. Poole, Differential impact of MexB mutations on substrate selectivity of the MexAB-OprM multidrug efflux pump of Pseudomonas aeruginosa, J. Bacteriol. 186 (2004) 1258–1269.

79. J.A. Bohnert, S. Schuster, E. Fähnrich, R. Trittler, W.V. Kern, Altered spectrum of multidrug resistance associated with a single point mutation in the Escherichia coli RND-type MDR efflux pump YhiV (MdtF), J. Antimicrob. Chemother. 59 (2007) 1216–1222.

80. Seeger, M. A., Ballmoos, C. V., Verry, F., and Pos, K. M. (2009) Crucial role of Asp408 in the proton translocation pathway of multidrug transporter AcrB: evidence from site-directed mutagenesis and carbodiimide labeling, Biochem. 48 , 5801-5812.

81. Su, C.-C., Li, M., Gu, R., Takatsuka, Y., McDermott, G., Nikaido, H., and Yu, E. W. (2006) Conformation of the AcrB multidrug efflux pump in mutants of the putative proton relay pathway, J. Bacteriol. 188 , 7290-7296. 66

82. Koronakis, V., Sharff, A., Koronakis, E., Luisi, B., and Hughes, C. (2000) Crystal structure of the bacterial membrane protein TolC central to multidrug efflux and protein export, Nature 405 , 914-919

83. Yu, E. W., Aires, J. R., McDermott, G., and Nikaido, H. (2005) A periplasmic drug- binding site of the AcrB multidrug efflux pump: a crystallographic and site- directed mutagenesis study, J. Bacteriol. 187 , 6804-6815.

84. Middlemiss, J. K. and Poole, K. (2004) Differential impact of MexB mutations on substrate selectivity of the MexAB–OprM multidrug efflux pump of Pseudomonas aeruginosa. J. Bacteriol . 186 , 1258–1269.

85. Akama, H., Matsuura, T., Kashiwagi, S., Yoneyama, H., Narita, S., Tsukihara, T., Nakagawa, A. & Nakae, T. (2004) Crystal structure of the membrane fusion protein, MexA, of the multidrug transporter in Pseudomonas aeruginosa , J. Biol. Chem. 279 , 25939-25942

86. Symmons, M. F., Bokma, E., Koronakis, E., Hughes, C., and Koronakis, V. (2009) The assembled structure of a complete tripartite bacterial multidrug efflux pump, Proc. Natl. Acad. Sci. USA 106 , 7173-7178.

87. Mikolosko, J., Bobyk, K., Zgurskaya, H. I. and Ghosh, P. (2006) Conformational flexibility in the multidrug efflux system protein AcrA, Structure 14 , 577-587.

88. Y. Takatsuka, H. Nikaido, Site-directed disulfide cross-linking shows that cleft flexibility in the periplasmic domain is needed for the multidrug efflux pump AcrB of Escherichia coli, J. Bacteriol. 189 (2007) 8677–8684.

89. Seeger, M. A., von Ballmoos, C., Eicher, T., Brandstatter, L., Verrey, F., Diederichs, L., and Pos, K. M. (2008) Engineered disulfide bonds support the functional rotation mechanism of multidrug efflux pump AcrB, Nat. Struct. Mol. Biol. 15 , 199-205.

90. Takatsuka, Y., and Nikaido, H. (2009) Covalently linked trimer of the AcrB multidrug efflux pump provides support for the functional rotating mechanism, J. Bacteriol. 191 , 1729-1737.

91. Andersen, C., Hughes, C., Koronakis, V. (2002) Electrophysiological behavior of the TolC channel-tunnel in planar lipid bilayers, J. Membrane Biol. 185 , 83- 92.

92. Tamura, N., Murakami, S., Oyama, Y., Ishiguro, M., and Yamaguchi, A. Direct interaction of multidrug efflux transporter AcrB and outer membrane channel TolC detected via site-directed disulfide cross-linking, Biochemistry 44 , 111115-21. 67

93. Touze, T., Eswaran, J., Bokma, E., and Koronakis, E. (2004) Interactions underlying assembly of the Escherichia coli AcrAB–TolC multidrug efflux system , Molec. Microbiol. 53 , 697-706.

94. Mokhonov, V. V., Akama, A., and Nakae, T. (2004) Role of the membrane fusion protein in the assembly of resistance-nodulation-cell division multidrug efflux pump in Pseudomonas aeruginosa , Biochem. Biophys. Res. Commun. 322 , 483–489.

95. Higgins, M. K., Bokma, E., Koronakis, E., Hughes, C., and Koronakis, V. (2004) Structure of the periplasmic component of a bacterial drug efflux pump, Proc. Natl. Acad. Sci. USA 101 , 9994-9999.

96. Tikhonova, E. B., Devroy, V. K., Lau, S. Y., and Zgurskaya, H. I. (2007) Reconstitution of the Escherichia coli macrolide transporter: the periplasmic membrane fusion protine MacA stimulates the ATPase activity of MacB, Mol. Microbiol. 63 , 895-910.

97. M.I. Borges-Walmsley, J. Beauchamp, S.M. Kelly, K. Jumel, D. Candlish, S.E. Harding, N.C. Price, A.R. Walmsley, Identification of oligomerization and drug-binding domains of the membrane fusion protein EmrA, J. Biol. Chem. 278 (2003) 12903–12912.

98. I. Bagai, W. Liu, C. Rensing, N.J. Blackburn, M.M. McEvoy, Substrate-linked conformational change in the periplasmic component of a Cu(I)/Ag(I) efflux system, J. Biol. Chem. 282 (2007) 35695–35702.

99. Su, C.-C., Yang, F., Long, F., Reyon, D., Routh, M. D., Kuo, D. W., Mokhtari, A. K., Van, Ornam, J. D., Rabe, K. L., Hoy, J. A., Lee, Y. J., Rajashankar, K. R., and Yu, E. W. (2009) Crystal structure of the membrane fusion protein CusB from Escherichia coli. J. Mol. Biol. 393 , 342-355.

100. Ruiz-Palacios, G. M. (2007) The health burden of Campylobacter infection and the impact of antimicrobial resistance: playing chicken, Clin.Infect.Dis. 44 , 701-703.

101. Blaser, M. J., and Engberg, J. (2008) Clinical aspects of Campylobacter jejuni and Campylobacter coli infections, in Campylobacter (Nachamkin, I., Szymanski, C. M., Blaser, and M. J. Eds.) 3 rd ed., pp 99-121.

102. Engberg, J., Aarestrup, F. M., Taylor, D. E., Gerner-Smidt, P., and Nachamkin, I. (2001) Quinolone and macrolide resistance in Campylobacter jejuni and C. coli : resistance mechanisms and trends in human isolates, Emerg.Infect.Dis. 7 , 24-34. 68

103. Gibreel, A., and Taylor, D. E. (2006) Macrolide resistance in Campylobacter jejuni and Campylobacter coli, J.Antimicrob.Chemother. 58 , 243-255.

104. Luangtongkum, T., Jeon, B., Han, J., Plummer, P., Logue, C. M., and Zhang, Q. (2009) Antibiotic resistance in Campylobacter: emergence, transmission and persistence, Future.Microbiol. 4 , 189-200.

105. Zhang Q., Plummer P. (2008) Mechanisms of antibiotic resistance in Campylobacter , in Campylobacter (Nachamkin, I., Szymanski, C. M., and Blaser, M. J. Eds.) 3rd ed., pp 263-276.

106. Lin, J., Michel, L. O., and Zhang, Q. (2002) CmeABC Functions as a Multidrug Efflux System in Campylobacter jejuni , Antimicrob.Agents Chemother. 46 , 2124-2131.

107. Pumbwe, L., and Piddock, L. J. (2002) Identification and molecular characterisation of CmeB, a Campylobacter jejuni multidrug efflux pump, FEMS Microbiol.Lett. 206 , 185-189.

108. Luo, N., Sahin, O., Lin, J., Michel, L. O., and Zhang, Q. (2003) In vivo selection of Campylobacter isolates with high levels of fluoroquinolone resistance associated with gyrA mutations and the function of the CmeABC efflux pump, Antimicrob. Agents Chemother. 47 , 390-394.

109. Cagliero, C., Mouline, C., Payot, S., and Cloeckaert, A. (2005) Involvement of the CmeABC efflux pump in the macrolide resistance of Campylobacter coli , Journal of Antimicrobial Chemotherapy 56 , 948-950.

110. Mamelli, L., Prouzet-Mauleon, V., Pages, J. M., Megraud, F., and Bolla, J. M. (2005) Molecular basis of macrolide resistance in Campylobacter: role of efflux pumps and target mutations, J. Antimicrob.Chemother. 56 , 491-497.

111. Ge, B., McDermott, P. F., White, D. G., and Meng, J. (2005) Role of efflux pumps and topoisomerase mutations in fluoroquinolone resistance in Campylobacter jejuni and Campylobacter coli , Antimicrob. Agents Chemother. 49 , 3347-3354.

112. Lin, J., Yan, M., Sahin, O., Pereira, S., Chang, Y. J., and Zhang, Q. (2007) Effect of macrolide usage on emergence of erythromycin-resistant Campylobacter isolates in chickens, Antimicrob. Agents Chemother. 51 , 1678-1686.

113. Cagliero, C., Mouline, C., Cloeckaert, A., and Payot, S. (2006) Synergy between efflux pump CmeABC and modifications in ribosomal proteins L4 and L22 in conferring macrolide resistance in Campylobacter jejuni and Campylobacter coli, Antimicrob. Agents Chemother. 50 , 3893-3896. 69

114. Yan, M., Sahin, O., Lin, J., and Zhang, Q. (2006) Role of the CmeABC efflux pump in the emergence of fluoroquinolone-resistant Campylobacter under selection pressure, J. Antimicrob. Chemother. 58 , 1154-1159.

115. Akiba, M., Lin, J., Barton, Y. W., and Zhang, Q. J. (2006) Interaction of CmeABC and CmeDEF in conferring antimicrobial resistance and maintaining cell viability in Campylobacter jejuni , Journal of Antimicrobial Chemotherapy 57 , 52-60.

116. Lin, J., Sahin, O., Michel, L. O., and Zhang, Q. (2003) Critical role of multidrug efflux pump CmeABC in bile resistance and in vivo colonization of Campylobacter jejuni , Infect. Immun. 71 , 4250-4259.

117. Maness, M. J., and Sparling, P. F. (1973) Multiple antibiotic resistance due to a single mutation in Neisseria gonorrhoeae , J. Infect. Dis. 128 , 321-330.

118. Guymon, L.F., Walstad, D. L., and Sparling, P. F. (1978) Cell envelope alterations in antibiotic-sensitive and –resistant strains of Neisseria gonorrhoeae , J. Bacteriol. 136 , 391-401.

119. Pan, W., and Spratt, B. G. (1994) Regulation of the permeability of the gonococcal cell envelope by the mtr system, Mol. Microbiol. 11 , 769-775.

120. Delahay, R. M., Robertson, B. D., Balthazar, J. T. Shafer, W. M., and Ison, C. (1997) Involvement of the gonococcal MtrE in the resistance of Neisseria gonorrhoeae to toxic hydrophobic compounds, Microbiology 143 , 2127-2133.

121. Hagman, K. E., Pan, W., Spratt, B. G., Balthazar, J. T., Judd, R. C., and Shafer, W.M. (1995) Resistance of Neisseria gonorrhoeae to antimicrobial hydrophobic agents is modulated by the mtrCDE efflux system, Microbiololgy 141 , 611-622.

122. Hagman, K. E., Lucas, C. E., Balthazar, J. T., Snyder, L., Nilles, M., Judd, R. C., and Shafer, W. M. (1997) The MtrD protein of Neisseria gonorrhoeae is a member of the resistance/nodulation/division protein family constituting part of an efflux system, Microbiology 143 , 2117-2125.

123. Veal, W. L., and Shafer, W. M. (2003) Identification of a cell envelope protein (MtrF) involved in hydrophobic antimicrobial resistance in Neisseria gonorrhoeae , J. Antimicrob. Chemother. 51 , 27-37.

124. Rouquette-Loughlin, C., Stojiljkovic, I., Hrobowski, T., Balthazar, J. T. and Shafer, W. M. (2002) Inducible, but not constitutive resistance of gonococci to hydrophobic agents due to the MtrC-MtrD-MtrE efflux pump requires the TonB-ExbB-ExbD proteins, Antimicrob. Agents Chemother. 46 , 561-565. 70

125. Veal, W. L., Nicholas, R. A., and Shafer, W. M. (2002) Overexpression of the MtrC-MtrD-MtrE efflux pump due to an mtrR mutation is required for chromosomally mediated penicillin resistance in Neisseria gonorrhoeae , J . Bacteriol. 184 , 5619-5624.

126. Faruki, H., Kohmescher, R. N., McKinney, W. P., and Sparling, P. F. (1985). A community-based outbreak of infection with penicillin-resistant Neisseria gonorrhoeae not producing penicillinase (chromosomally-mediated resistance), N. Engl. J. Med. 313 , 607-611.

127. Olesky, M., Hobbs, M., and Nicholas, R. A. (2002) Identification and analysis of amino acid mutations in porin IB that mediate intermediate-level resistance to penicillin and tetracycline in Neisseria gonorrhoeae , Antimicrob. Agents Chemother. 46 , 2811-2820.

128. Zarantonelli, L., Borthagary, G., Lee, E. H., Veal, W., and Shafer, W. M. (2001) Decreased susceptibility to azithromycin and erythromycin mediated by a novel mtrR promoter mutation in Neisseria gonorrhoeae , J. Antimicrob. Chemother. 47 , 651-654.

129. Johnson, S. R., Sandul, A. L., Parekh, M., Wang, S. A., Knapp, J. S., and Trees, D.L. (2003) Mutations causing in vitro resistance to azithromycin in Neisseria gonorrhoeae , Internatl. J. Antimicrob. Agents 21 , 414-419.

130. Rouquette-Loughlin, C., Veal, W. L., Lee, E.-H., Zarantonelli, L., Balthazar, J. T., and W. M. Shafer (2002) Antimicrobial efflux systems possessed by Neisseria gonorrhoeae and Neisseria meningitidis viewed as virulence factors, in Microbial Drug Efflux (Paulsen, I., and Lewis, K. ed.) pp 187-200.

131. Shafer, W. M., Qu, X.-D., Waring, A. J., and Lehrer, R. I. (1998) Modulation of Neisseria gonorrhoeae susceptibility to vertebrate antibacterial peptides due to a member of the resistance/nodulation/division efflux pump family, Proc. Natl. Acad. Sci. U.S.A. 95 , 1829-1833.

132. Jerse, A. E., Sharma, N. D., Bodner, A. N. B., Snyder, L. A., and Shafer, W. M. (2003) A gonococcal efflux pump system enhances bacterial survival in a female mouse model of genital tract infection, Infect. Immun. 71 , 5576-5582.

133. Van Bambeke, F., Pagès, J.-M., and Lee, V.J. (2006) Inhibitors of bacterial efflux pumps as adjuvants in antibiotic treatments and diagnostic tools for detection of resistance by efflux, Recent Patents Anti-Infect Drug Disc. 1, 157–175.

134. Bailey, A. M., Paulsen, I. T., and Piddock, L. J. (2008) RamA confers multidrug resistance in Salmonella enterica via increased expression of acrB, 71

which is inhibited by chlorpromazine, Antimicrob. Agents Chemother. 52, 3604–3611.

135. Kiho, T., Nakayama, M., Yasuda, K., Miyakoshi, S., Inukai, M., and Kogen, H. (2004) Structure–activity relationships of globomycin analogues as antibiotics, Bioorg. Med. Chem. 12 , 337–361.

136. Tokuda, H., and Matsuyama, S. (2004) Sorting of lipoproteins to the outer membrane in E. coli , Biochim. Biophys. Acta 1693, 5–13.

137. Malléa, M., Chevalier, J., Eyraud, A., and Pagès, J.-M. (2002) Inhibitors of antibiotic efflux pump in resistant Enterobacter aerogenes strains, Biochem. Biophys. Res. Comm. 293, 1370–1373.

138. Higgins, M.K., Eswaran, J., Edwards, P., Schertler, G.F., Hughes, C., and Koronakis, V. J. (2004) Structure of the ligand-blocked periplasmic entrance of the bacterial multidrug efflux protein TolC, Mol Biol. 34 , 2697-702.

139. Pagès, J.-M., Masi, M., and Barbe, J. (2005) Inhibitors of efflux pumps in Gram-negative bacteria, Trends Mol. Med. 11, 382–389.

140. Lomovskaya, O., Warren, M. S., Lee, A., Galazzo, J., Fronko, R., Lee, M., Blais, J., Cho, D., Chamberland, S., Renau, T. E., Leger, R., Hecker, S.J., Watkins, W., Hoshino, K., Ishida, H., and Lee, V. J. (2001) Identification and characterization of inhibitors of multidrug resistance efflux pumps in Pseudomonas aeruginosa : novel agents for combination therapy, Antimicrob. Agents Chemother. 45, 105–116.

141. Renau, T. E., Leger, R., Flamme, E. M., Sangalang, J., She, M. W., Yen, R., Gannon, C. L., Griffith, D., Chamberland, S., Lomovskaya, O., Hecker, S. J., Lee, V. J., Ohta, T., and Nakayama (1999) Inhibitors of efflux pumps in Pseudomonas aeruginosa potentiate the activity of the fluoroquinolone antibacterial levofloxacin, J. Med. Chem. 42, 4928–4931.

142. Piddock, L. J. (2006) Multidrug-resistance efflux pumps - not just for resistance, Nat. Rev. Microbiol. 4 , 629-36.

143. Zgurskaya, H. I., and Nikaido, H. (1999) Bypassing the periplasm: reconstitution of the AcrAB multidrug efflux pump of Escherichia coli , Proc. Natl. Acad. Sci. USA 96 , 190-7195.

144. Nikaido, E., Yamaguchi, A., and Nishino, K. (2008) AcrAB multidrug efflux pump regulation in Salmonella enterica serovar Typhimurium by RamA in response to environmental signals, J. Biol. Chem. 283 , 24245-24253. 72

145. Hirakawa, H., Inazumi, Y., Masaki, T., Hirata, T., and Yamaguchi, A. (2005) Indole induces the expression of multidrug exporter genes in Escherichia coli, Mol. Microbiol. 55, 1113-1126.

146. Nishino, K., Nikaido, E., and Yamaguchi, A. (2009) Regulation and physiological function of multidrug efflux pumps in Escherichia coli and Salmonella , Biochim Biophys Acta (BBA) Proteins Proteomics 1794 , 834-843.

147. Hirakata, Y., Srikumar, R., Poole, K., Gotoh, N., Suematsu, T., Kohno, S., Kamihira, S., Hancock, R. E. W., and Speert, D. P. (2002) Multidrug efflux systems play an important role in the invasiveness of Pseudomonas aeruginosa, J. Exp. Med. 196, 109-118.

148. Long, F., Su, C.-C., Routh, M. D., Wang, W., Rajashankar, K. R., and Yu, E. W. (2010) Crystal structures of the CusA heavy-metal efflux pump suggests a methionine mediated mechanism, Nature (submitted)

149. K. Hofmann & W. Stoffel (1993) TMbase - A database of membrane spanning proteins segments Biol. Chem. Hoppe-Seyler 374,166

150. Thompson, J.D., Higgins, D.G. and Gibson, T.J. (1994) CLUSTAL W: improving the sensitivity of progressive multiple sequence alignments through sequence weighting, position specific gap penalties and weight matrix choice. Nucl. Acids Res. 22:4673-4680.

151. Biochimica et Biophysica Acta 1794 (2009) 769–781 Mechanisms of RND multidrug efflux pumps Hiroshi Nikaido, Yumiko Takatsuka

152. The PyMOL Molecular Graphics System, Version 1.2r3pre, Schrödinger, LLC.

153. Eicher, T., Brandstatter, L., and Pos, K. M. (2009) Structural and functional aspects of the multidrug efflux pump AcrB, Biological Chem. 390 , 693-699.

CHAPTER 3. STRUCTURAL AND FUNCTIONAL STUDIES OF AcrD

ABSTRACT Here, we report the crystallization and preliminary X-ray data analysis of the multidrug efflux pump AcrD of Escherichia coli (E. coli) . AcrD is a cytoplasmic membrane protein consisting of 1037 amino-acid residues. The transport protein belongs to the previously characterized resistance, nodulation, and cell division (RND) superfamily of transporters. Substrates of AcrD include polar, hydrophilic aminoglycosides such as amikacin, gentamicin, and kanamycin. Furthermore, AcrD presumably transports other structurally unrelated toxic compounds, including sodium dodecyl sulfate (SDS), novobiocin, tetracycline, nalidixic acid, norfloxacin, fosfomycin, and bile acids. The recombinant AcrD protein was expressed in E. coli and purified to greater than 95% using a combination of hydroxyapatite, metal affinity, and size exclusion chromatography. The protein was crystallized using the hanging-drop vapor diffusion method. X-ray diffraction data were collected from cryo-cooled crystals at a synchrotron light source. The best crystal diffracted to 4.2 Å and diffraction data were complete to 5.8 Å resolution. The spacegroup was determined to be P2, with unit-cell parameters: a = 114.0, b = 117.2, c = 178.4 Å. Additionally, binding studies utilizing fluorescence polarization and STD-NMR techniques were used to analyze protein-ligand interactions. Intriguingly, we illustrate that the anthracyline family of chemotherapeutic agents is recognized by AcrD. Furthermore, we identify the binding epitope of gentamic using STD-NMR. Finally, as the crystal structure yet eludes us, we use computer-aided modeling to suggest the AcrD structure and identify the ligand binding region within a cleft in the porter domain. Interestingly, a possible second binding site is located in the TolC- docking domain, which suggests a probable substrate pathway from the cell interior to TolC by passing through both binding sites.

INTRODUCTION The multidrug restant (MDR) phenotype in bacteria has been primarily attributed to transporters within the cytoplasmic membrane (1). These pumps are capable of exporting a wide range of antibacterial compounds and their overproduction causes increased resistance to antimicrobial agents. The multidrug resistance associated with these proteins has become a major concern with increasing occurrences of resistant pathogens in the clinical setting (2). Multidrug resistance efflux pumps are grouped into five families of transporters. These families include primary transporters belonging to the ATP-binding cassette (ABC) family (3) and secondary transporters among the major facilitator superfamilies (MFS) (4,5), as well as the multidrug and toxic compound extrusion (MATE) (6-8), the resistance nodulation division (RND) (9), and the small multidrug resistance (SMR) families (10). Among the five known families of multidrug transporters, the RND family tends to play major roles in the intrinsic resistance of gram-negative bacteria (11, 12). In fact, E. coli harbor 37 putative multidrug efflux transporters (13, 14). Thus far, ~20 of these transporters have been identified as contributors to multidrug resistance (14) with seven belonging to the RND family of exporters. The E. coli efflux pump AcrD is a prototypical member of the RND family of multidrug transporters. As a member of the RND family, it is suggested that AcrD is a trimeric protein comprised of topological features including 12 transmembrane (TM) α-helices with two large (~300 residue each) periplasmic extensions between TMs 1-2 and TMS 7-8 from each monomer (9, 15). The transporter AcrD is homologus to the well-characterized MDR efflux pump AcrB. Intriguingly, it has been suggested that AcrD is able to sequester ligands from both the periplasmic space and the cytoplasmic interior, which has not been shown for AcrB (16). This disparity may be due to the nature of the inducing ligands. AcrB recognizes mainly negatively charged and hydrophobic drugs that readily pass the inner membrane, which may make a cytoplasmic entry point unnecessary (1, 17, 18). The aminoglycosides that are extruded by AcrD are positively charged and extremely hydrophilic molecules. Presumably, these drugs are not able to cross the 75

membrane efficiently so a cytoplasmic entrance to AcrD is critical to protect the bacterial cell (19). Aminoglycosides are an important class of antibacterial compounds that are commonly used as a broad-spectrum antibiotic and are reemerging as key treatments of drug resistant infections. First discovered in 1944, aminoglycosides disrupt the integrity of the cell membrane and inhibit protein synthesis by binding the 16S rRNA of bacteria (20, 21). Understanding the binding mode of these aminoglycosides as well as other AcrD substrates will provide valuable insights into the novel mechanism of ligand capture and extrusion. Additionally, the determination of subtrate recognition profiles will facilitate the discovery of novel aminoglycosides that may not be recognized by AcrD or the development of efflux pump inhibitors that may block AcrD-substrate interaction. Importantly, the array of ligands AcrD recognizes is distinct of those for AcrB and the detailed ligand binding mechanism will provide information as to how a MDR efflux pump is able to recognize and extrude cationic, hydrophilic compounds.

PROTEIN PRODUCTION Expression: The C-6XHis AcrD protein was overproduced in E. coli BL21 Gold/pSport1_acrD cells. After optimizing expression conditions, a protocol developed in our lab was used to purify AcrD. In brief, the AcrD-pSport1 vector was transformed into the BL21 Gold and grown overnight at 37ºC in a 10 ml liquid Luria– Bertani (LB) broth culture. Following overnight growth, cells were transferred to four 1-liter LB cultures supplemented with 100 µg/ml of ampicillin. The culture was grown with shaking (210 rev min -1) at 310 K. Expression of AcrD was induced by 1 mM isopropyl β-d-1-thiogalactopyranoside (IPTG) after the OD 600 reached .4. Cells were then harvested by centrifugation at 6000 rev min -1 for 10 minutes at 277 K and frozen and stored at 193 K.

Purification: The membrane protein was extracted and purified as follows. Briefly, the cells were suspended in low-salt buffer containing 100mM sodium 76

phosphate (pH7.2), 10% glycerol, 1mM EDTA, and 1mM phenylmethylsulfonylfluoride (PMSF), and then disrupted by passing through a French press three times. The membrane fraction of the pellet was collected and washed twice with a high-salt buffer containing 20mM sodium phosphate (pH7.2), 2M KCl, 10% glycerol, 1mM EDTA, and 1mM PMSF and once with 20 mM HEPES- NaOH buffer (pH7.5). The membrane protein was then solubilized in 2% (w/v) n- dodecy-β-D-maltoside ( β-DDM). Insoluble material was removed by ultracentrifugation at 100,000×g. The extracted membrane protein was subsequently loaded into a Hydroxy Apatite column and eluted with 8 mM Na-phosphate buffer (pH 7.0) supplemented with 1% β-DDM. This elution was then added to a Nickel- affinity column, washed with a buffer containing 20mM HEPES-NaOH (pH7.5), 50mM imidazole, and 0.05% β-DDM, and eluted in a buffer consisting of 20mM HEPES-NaOH (pH7.5), 400mM imidazole, and 0.05% β-DDM. The eluted protein fractions were collected and dialyzed against 20mM HEPES-NaOH (pH7.5) buffer and concentrated to 20 mg/ml for crystallization. The C-6XHis AcrD protein was concentrated to 20 mg/ml in a solution containing 20 mM Na-HEPES (pH 7.5) and .05% β-DDM (or other detergent as indicated). Exchange of primary detergents was performed using gel filtration chromatography. The purity of C-6XHis AcrD protein was judged using a10% SDS–PAGE stained with Coomassie Brilliant Blue. Western blots were used to detect the 6X his-tag and determine the presence of purified C- 6XHis AcrD. Typical starting and ending volumes were 15 ml and 1 ml, respectively. A YM-50 Centriprep concentrator (Millipore, 50 kDa molecular-weight cutoff) was used for concentrating protein. Generally, 10-20 mg of purified C-6XHis AcrD was obtained.

Detergent exchange: Detergent exchange to any of the six detergents used for crystallization was performed using a G-200 sizing colulmn. β-DDM solubilized AcrD was loaded onto the sizing column, which was pre-equilibrated with a buffer containing 20 mM Na-HEPES (pH 7.5) and the indicated detergent. The volume and length of the G-200 sizing column was 120 ml and 60 cm, respectively. A flow rate 77

of 0.5 ml min -1 was used to collect 2 ml fractions. The fractions were analyzed for the presence of AcrD using a 10% SDS–PAGE stained with Coomassie Brilliant Blue. Fractions harboring AcrD were pooled and concentrated using the YM-50 Centriprep concentrator to 20 mg/mL and and were subsequently used for crystallization. Following extensive crystallization trials with the detergents exchanged through gel-filtration chromatography, it was determined which of the six primary detergents were most suitable for crystallization. After choosing the optimal detergents, steps were taken to solubilize AcrD from the membrane fraction. During exchange through the gel-filtration column, residual β-DDM may remain in the detergent-AcrD micelle, therefore hindering crystallization. To avoid this, the identical purification protocol was used, but the primary solubilizing detergent was replaced with either Fos-choline 12 (FC-12) or Cymal-6.

CRYSTALLIZATION The full-length C-6XHis AcrD protein was crystallized in 24-well plates using the hanging drop vapor-diffusion method. As an initial step to understanding how the RND family transporter AcrD functions, crystallization trials were performed in the absence of ligands. Typically, AcrD crystals were grown using the hanging-drop vapor diffusion method with a 20mg/ml protein solution in 20mM tris (8.5) and .05% β-DDM. An ambient temperature of 25ºC was maintained throughout crystal growth. Initial crystallization trials were performed using Hampton reagents, including Crystal Screen 1, 2, Lite, and Membfrac. Initially, Reagent 15 of Crystal Screen Light produced the most promising results. This condition was optimized by manipulating buffers, precipitants, salts, detergents and additive solutions. In the beginning, all crystallization trials were carried out using β-DDM as the primary detergent. Following rigorous screening, the best quality AcrD- β-DDM crystals grew in a mother solution containing PEG 1500 15.5%, .25M NaCl, .1M tris 8.5, and 3% Glycerol.

Detergents : With initial X-ray diffraction only achieving 15Å, steps were taken to improve cystal packing. It has previously been shown that the choice of primary and addition of secondary detergents is a crucial factor in the success of crystal formation for membrane proteins. Indeed, crystallization of the Na +/H + Antiporter NhaA was attempted in the presence of ten various detergents, although crystal growth was achieved in only Fos-choline 12, α-DDM, and β-DDM. Of these detergents, high-quality diffraction was obtained with only α-DDM and the addition of a secondary detergent n-Octyl-β-D-Glucopyranoside (β-OG), which was identified following an extensive screen of secondary detergents (22). In this regard, we initially set up secondary detergent screens using Detergent Kits 1, 2, and 3 from Hampton Research to improve results. Using our previously established conditions, the addition of n-Octyl-b-D-thiomaltoside (OTM), FC-12, CHAPS, Cymal-6, dodecyl dimethyl glycine, and tetra DM improved the AcrD- β-DDM crystal quality. The most dramatic improvement was observed using CHAPS as the secondary detergent. The addition of 1% CHAPS improved resolution limits to ~5.0 Å with a complete data set complete to 5.8 Å. Figure 1 displays a representative diffraction image with diffraction statistics highlighted in table 1. Detergent selection is a critical factor in membrane protein crystallization. The tendency of a detergent to aid in crystallization or possibly denature the protein is a fine line. Qualitatively, predictions can be made about the deleterious effects of a detergent by considering the size and charge of the polar headgroup along with the length of the alkyl tail. These factors affect the CMC and size of the micelle (23). This was taken into consideration when selecting primary detergents, thus a range of detergents varying in these parameters were chosen. Using gel-filtration chromatography, β-DDM was exchanged with various detergents, including FC-12, Cymal-6, Cymal-5, CHAPS, β-OG, OTM, and Lauryldimethylamine-oxide (LDAO). The most promising results were established with FC-12 and Cymal-6. Therefore, the ensuing steps involved purification of AcrD and solubilization with these two detergents. After purification, crystallization with AcrD-Cymal-6 only produced needle shaped (poor quality) crystals, while AcrD-FOS-choline-12 crystals had 79

improved birefringence, which indicated higher qualtity crystals (Figure 2). Both crystal forms were produced in nearly identical conditions, which highlights the necessity for the correct choice of primary detergent. Additionally, secondary detergents were once again screened, although this did not improve diffraction quality. Table (2) illustrates the screening experiments and corresponding results. The needle shaped cymal-6-AcrD crystals did not diffract while the FC-12 solubilized AcrD protein diffracted to ~10Å. Although, the diffraction quality did not improve beyond the AcrD-β-DDM crystal, these experiments outline the importantce of correct detergent selection.

DATA COLLECTION For data collection, a single native crystal of C-6XHis AcrD was flash-cooled in a cryoprotectant solution containing the mother solution (PEG 1500 15.5%, .25M NaCl, .1M tris 8.5, and 3% Glycerol) plus 25% glycerol at 100 K. The glycerol concentration was gradually increased to 25% by 5% increments. The best crystal diffracted anisotropically to a resolution of 5 Å. Fig. 1 depicts one of the diffraction images of the native AcrD crystal. Diffraction data sets were obtained from the native C-6XHis AcrD crystals at the Advanced Photon Source (APS, beamline 24IDC) at cryogenic temperature (100 K) using an ADSC Quantum 315 CCD-based detector. The beam size was 50 X 20 mm.

AcrD-LIGAND CRYSTALLIZATION The integral membrane protein AcrD plays an important role in protecting the bacterial cell from various toxic chemicals, including aminoglycosides and other antibiotic compounds (14, 19). The addition of substrates and other small molecules has been shown to stabilize transport proteins and aid in crystallization. In lieu of this, attempts were made to improve diffraction limits by adding the antibiotic substrates. Initially, gentamicin and daunorubicin HCl (DR) were soaked into preformed crystals. Trials to soak in antibiotics and co-crystallize the drug/protein were not able to improve resolution. Crystals of the AcrD-gentamicin complex had 80

similar diffraction as apo-AcrD and crystallized in the identical space group. Therefore, co-crystallization of AcrD with the same two ligands was initiated. Co- crystallization successfully produced crystals, although these were poor quality crystals with data sets complete to less than 10Å.

FLUORESCENT POLARIZATION Fluorescence polarization assays were used to determine the drug binding affinities of AcrD. The experiments were done using a ligand binding solution containing 20 mM HEPES (pH 7.5), 0.05% DDM, and 1 M DR. The AcrD protein solution and 1 M DR was titrated into the ligand binding solution until the polarization (P) became saturated. In this assay, the protein–drug interaction would reach equilibrium within 1 min. As this is a steady-state approach, fluorescence polarization measurements were taken after incubation for 5 min at each corresponding concentration of the protein and drug to ensure that the binding had reached equilibrium. It should be noted that the detergent concentration was kept constant at all times to eliminate the change in polarization generated by drug–DDM micelle interaction. All measurements were performed at 25 °C using a PerkinElmer LS55 spectrofluorometer equipped with a Hamamatsu R928 photomultiplier. The excitation wavelength was 485 nm. Fluorescence polarization signals (in P) were measured at an emission wavelength of 595 nm, respectively. Each titration point recorded was an average of 15 measurements. Data were analyzed using the equation, P={( Pbound -Pfree )[protein]/( KD+[protein])}+ Pfree , where P is the polarization measured at a given total protein concentration, Pfree is the initial polarization of free ligand, Pbound is the maximum polarization of specifically bound ligand, and [protein] is the protein concentration. The titration experiments were repeated three times to obtain the average KD value. Curve fitting was accomplished using the program ORIGIN (24).

In order to determine if detergent solubilized AcrD is present in an active state and able to bind gentamicin and DR, binding assays were performed to test 81

substrate-AcrD interaction. Importantly, the fluorescence studies identified an antibacterial compound, DR, from the anthracycline family that is recognized by AcrD. Binding kinetics were assayed using fluorescence polarization, a technique that measures the tumbling rates of fluorescent molecules. Using a constant concentration of 1 µM DR, the concentration of AcrD was increased from 2 to 150 µM to create the curve indicated in Figure 3. The dissociation constant of 6 µM, suggests that AcrD is fully functional and binds DR in a range previously observed between AcrB and its hydrophobic substrates (25). As gentamicin does not harbor fluorescent characteristics, isothermal titration calorimetry was performed to develop a better understanding of AcrD function.

ISOTHERMAL TITRATION CALORIMETRY (ITC): Previously purified AcrD protein was dialyzed against sodium phosphate buffer (7.2) with additional .02% β-DDM (buffer A) using a dialysis cassette with a 20,000 MWCO. Dialyzed AcrD was then used in all ensuing ITC experiments. All calorimetric experiments were performed with a VP−ITC MicroCalorimeter (MicroCal, Inc., Northampton, MA) in buffer A, at 25 °C. All solutions were thoroughly degassed under vacuum for 5 min before being used. Ligand (gentamicin) was prepared in the dialysate of the protein buffer to minimize artifacts due to different compositions of solutions. The reaction cell contained 1.4 mL of protein in buffer A, and the reference cell contained distilled water only. The injection syringe was filled with ligand solution and rotated at 310 rpm during equilibration and experimentation. Titration experiments consisted of 61 injections. The volume of the first injection was 2.5 L, and the subsequent injection volumes were 5 L. Injection speed was 0.5 L/s with 2-min intervals between injections. For blank experiments, separate titrations of the ligand solution in the buffer were performed to determine the heat of ligand dilution. The blank run was subtracted from the experimental ITC to counteract the effects of the heat from ligand dilution. The heat of dilution was subtracted prior to analysis of the data.

A Leverber−Marquardt algorithm performed by MicroCal Origin scientific plotting software was used to fit the incremental heat of the nth titration [( Q(i)] of the total heat, Qt

where V0 is the volume of the sample solution. For the model of the single set of identical independent sites, the following equation was used:

where n is the stoichiometry of the binding reaction, [P] t is the total AcrD

concentration in the sample vessel, Hb is the binding enthalpy, and Kb is the binding

constant. The binding entropies, Sb, were calculated using the following equation of the thermodynamics:

Standard deviations for Hb, Sb, and Kb were calculated from multiple titration runs. The calorimetry experiments revealed a stoichiometry of .398:1 gentamicin:AcrD or 1.2 gentamicin molecules per AcrD trimer, respectively. Additionally, a dissociation constant of 3.8 ± .95 µM was estimated. The system is exothermic with a heat change of -906.1 ± 39.9 kcal/mole and is driven entropically, whereby the change in entropy is 21.8 e.u (Figure 4). Not surprisingly, the binding constant of gentamicin is distinctly similar to that of daunorubicin, although they belong to different families of drugs and different techniques were used for analysis.

STD-NMR To assess the interaction between AcrD and gentamicin, novel saturation transfer difference-nuclear magnetic resonance (STD-NMR) studies were undertaken. STD-NMR is a powerful technique used for studying protein-ligand interactions. The STD NMR experiment is based on transfer of saturation from the protein to a bound ligand. This method is capable of identifying the binding epitope of a ligand when bound to protein. Ligand-protons that are in close contact with the receptor receive a higher degree of saturation, and as a result stronger STD NMR signals can be observed. All NMR experiments were performed on a Bruker Avance DRX 900-MHz spectrometer equipped with a 5-mm inverse triple-resonance probe head at 300 K. NMR samples were prepared as described above then dialyzed 3X against 500 L

of 99.9% D 2O Sodium Phosphate buffer (pH 7.0). Protein concentrations in the NMR samples were in a range between 1 and 50 M. Gentamicin concentration ranged from 1 to 4 mM. The pulse scheme of the 1D STD NMR spectra is previously described by Mayer (26). Subtraction of the 1D STD spectra were performed internally via phase cycling after every scan to minimize artifacts arising from temperature and magnet instability. For samples with higher concentrations of H2O, the water suppression by gradient tailored excitation (WATERGATE) scheme for suppression of the residual HDO signal was employed (27). The on-resonance irradiation of the protein was performed at a chemical shift of .25 ppm. Off-resonance irradiation was applied at 30 ppm, where no protein signals were present. 1D STD NMR spectra were multiplied by an exponential line-broadening function of 1−3 Hz prior to Fourier transformation. The irradiation power of the selective pulses in all STD NMR experiments was set to

(γ/2 π)B1 = 86 Hz. Selective presaturation of the protein was achieved by a train of Gauss-shaped pulses of 50-ms each, separated by a 1-ms delay. The number of selective pulses n determines the presaturation period, and the standard value was 40 pulses, leading to a total length of the saturation train of 2.04 s. The additional delay (d1) was set to 100 ms in all STD experiments. The total number of scans was 84

256 or 512. Also, 16 dummy scans were applied, typically using 12 ppm spectral widths for the 1D STD NMR spectra. All 1D STD spectra of samples containing

ligands were recorded with a 30-ms spin-lock pulse, or so-called T1ρ filter, after the

π/2 pulse with a strength of ( γ/2 π)B1 = 4960 Hz, which eliminates the background protein resonances to facilitate analysis. For the group epitope mapping analysis, the STD integrals of the individual protons of gentamicin are referenced to the strongest STD signal in each spectrum, which is assigned to a value of 100%. Proton peaks were previously identified in NMR experiments for gentamicin by Deubner et al. 2003 (28). The differential STD effects within gentamicin then yield information on the proximity of the individual protons to the protein surface. Indeed, the STD-NMR results indicated that gentamicin does interact with AcrD. The studies suggest that protons at positions C22, C24, C26, C7, C13, C11 form the closest interactions with AcrD, whereby their percent saturation, relative to the maximum (max std is group C22), is above 60% (Fig 5 and Table 3). A representative STD-NMR experiment is presented in Figure 6, with Figure 5 and Table 3 highlighting the experimental results.

STRUCTURAL MODELING To build a structural model for AcrD, the ESyPred3D Web Server 1.0 was used (29). In the ESyPred3D system, alignments are obtained by combining, weighting and screening the results of several multiple alignment programs. The final three dimensional structures are built using the modeling package MODELLER (30). To build the model, the crystal structure of MexB, accession code 2V50, was utilized. The homologous MexB protein shares 60.7% identity with AcrD. The output result, from the computer-aided modeling, was a single AcrD monomer, with overall structural features similar to that of AcrB and MexB. The transmembrane domain (TM) was comprised of 12 α-helical segments. In the periplasmic region two individual domains are formed from various smaller subdomains. The subdomains PC1, PC2, PN1, and PN2 (utilized from AcrB nomenclature) combine to make up the 85

porter domain, while the TolC-docking domain is comprised of the subdomains DC and DN, respectively (Fig. 7). To model the trimeric state of AcrD, monomeric AcrD was superimposed over each of the individual protomers of trimeric AcrB (accession code 1IWG) using the program COOT (31) (Fig. 8). The AcrD trimer displayed intriguing features, including a large central cavity, vestibules at monomer interfaces, and clefts located between subdomains PC1 and PC2 that are also present in the AcrB trimeric structure(Fig. 9). Previously, AcrB had been suggested to bind ligands within the central cavity, which is located in the transmembrane region. Ligands presumably could enter this central cavity through adjacently located vestibules. Alternatively, recent evidence suggests that ligands may interact within the clefts, which are located approximately 15Å above the membrane plane (Fig. 9). To further evaluate AcrD-ligand interaction, AutoDock vina was used to predict the most favored binding modes of ten previously identified ligands (32), including sodium dodecyl sulfate (SDS), fosfomycin (Fos), Gen, amikacin (Amk), norfloxacin (Nor), novobiocin (Nov), deoxycholic acid (Dca), tetracycline (Tet), nalidixic acid (Nal), and DR. Table 4 lists the ligands along with the associated minimum inhibitory concentrations (MIC) of plasmids with and without AcrD. Daunorubicin is not listed in table 4, as the work reported here describes, for the first time, AcrD-DR interaction. Intriguingly, AutoDock Vina identified two distinct binding sites within AcrD, with neither locating within the central binding cavity. Binding site 1 is located directly in the hydrophilic binding cleft (Fig. 10). This is nearly identical to the location of bound minocycline in the recently solved asymmetric crystal structure of AcrB (33). However, in the AcrD binding groove many of the hydrophobic phenylalanines present in AcrB are replaced with hydrophilic residues, including F615 to Ser, F617 to Pro, and F178 to Tyr. Additionally, binding site 1 seems to harbor multiple binding patches capable allowing AcrD to recognize diverse substrates. For instance, a more hydrophobic patch exists within the binding cavity allowing AcrD to accommodate the hydrophobic tail of SDS (Fig. 11F). The aminoglycosides Gen and Amk appear to bind in an overlapping location that may be the primary drug recognition site 86

(Figures 11A and 11B), as seven of the ten docked molecules have at least partial interaction in this region. Figure 10A displays binding site 1 with all docked drugs, while 10B provides a closer view of the binding site while removing overlapping ligands. The most interesting results from the docking experiment were the identification of a possible second binding site. The second site was identified in a location stacked between the TolC-docking domain and the porter domain. Autodock identified a binding site in this region for each of the tested substrates (Fig. 12). Table 5 lists each ligand along with the binding energy for each binding site. Similar to binding site 1, this second binding site is multifaceted and capable of recognizing diverse ligands. Figures 13A through 13J illustrate the individual binding sites of each drug. Presumably, drugs could interact with binding site 1 initiating conformational shifts that transfer the ligand to the second site. To identify a possible exit pathway, Caver was used to identify tunnels that pass through AcrD (34). The starting point was set as the Gen binding site, which allowed for the identification of tunnels leading into and out of the suggested cavity. As displayed in figure 14, a passage was identified leading from the the central cavity as well as the periplasmic space to binding site 1. Furthermore, the tunnel extended from binding site 1 to binding site 2 and ultimately exited toward the TolC-docking domain funnel.

DICUSSION The studies reported here describe the function and suggest the structure of the RND transporter AcrD. As indicated, the results of these studies indicate that AcrD does indeed specifically interact with the aminoglycoside Gen by primarily associating with the R1 and R2 groups (figure 5). The binding assay with Gen, using ITC revealed a KD of 3.8 ± .95 µM. Not surprisingly, it was observed that the chemotherapeutic drug DR, of the anthracycline family, also binds to AcrD with a

similar affinity ( KD = 6 µM). We hypothesize, this dissociation constants is in an ideal range, whereby AcrD is able to export substrates while they are still at sublethal 87

concentrations. Interestingly, in similar experiments using the homologus protein AcrB, binding affinities to AcrB substrates were nearly identical. Using fluorescence polarization, dissociation constants were observed for ethidium bromide, proflavin, and rhodamine 6G of 8.7, 14.5, and 5.5 M, respectively (25). Along with the binding studies we attempted crystallization of AcrD in both apo and ligand-bound forms utilizing diverse primary and secondary detergents. To this point, diffraction limits have been obtained to lower than 5Å. Although diffraction limits have been drastically improved, there is still a mountain to climb in determing the three-dimensional structure of AcrD. Therefore, this challenge was circumvented using structural modeling of AcrD. A model was created using MexB as a template, whereby the overall architecture generally observed in RND transporters was maintained. After docking several ligands into the AcrD model, two distinct binding sites were located. It seems possible that ligands may enter binding site 1 through the central cavity (cytoplasm) or simply be captured from the periplasmic space from the binding cleft. We suggest that substrate interaction will close binding site 1, forcing the ligand into the second binding site and ultimately out of the transporter. Alternatively, the second binding site may not be functional in substrate recognition or expulsion. As the Autodock program searches for probable binding sites, it may have located this area due to its large hydrophobic area. Currently, no other RND protein has been identified to bind ligands in this region. Using CAVER to highlight solvent channels a specific pathway was suggested leading through both binding sites to the TolC-docking domain. At this point, the substrate could enter the interior of TolC and be ushered completely out of the bacterial cell. Although this suggests a possible pathway in provides clues to the mechanism, detailed structural changes are yet undefined. A model to describe the function of RND transporters AcrB and MexB has recently been described (33, 35- 37), whereby a functionally rotating mechanism guides drug export. A similar situation may occur, but our modeling studies fail to detect such subtle 88

conformational differences. In order to improve the AcrD model, the crystal structure of both apo-AcrD and substrate-bound AcrD awaits.

FIGURES AND TABLES

Figure 1

Figure 1. A representative diffraction image of AcrD-β-DDM crystals along with a close-up view (red box) displaying resolution to approximately 6.8Å. A synchrotron light source at the APS facility at Argonne National Laboratories was used to collect data.

Table 1

Table 1. Diffraction statistics of the highest diffracting crystal (AcrD- β-DDM with additional CHAPS). The diffraction limit for this data set was approximately 6.2 Å.

Figure 2

Figure 2 : Crystal shapes obtained through screening of various primary and secondary detergents.

Table 2 AcrD crystallization Secondary Primary Detergent Detergent Crystals diffraction β-DDM Fos-choline 12 rectangular 8-9 Å (figure 2b) CHAPS rectangular 6-7 Å (fi gure 2a) None cubic 12-15 Å (figure 2e) Cymal-6 Bar 10-12 Å (figure 2f) OTM Rectangular 8-9 Å Fos-chol 12 ( figure 2d ) Cymal-4 Cubic salt (figure 2c) Cymal-6 Diamond 9-11 Å Cymal-6 OTM needle none Cymal-5 None no crystals CHAPS None no crystals β-OG None no crystals OTM None no crystals LDAO None no crystals

Table 2. Table displaying the various primary and secondary detergents used during crystallization trials and the resulting crystals and resolution. AcrD-β-DDM crystals along with a secondary detergent of CHAPS provided the highest diffracting crystals.

Figure 3A

kD=6µM

Figure 3B

Daunorubicin-HCl Figure 3. A) Representative fluorescence polarization of AcrD with Daunorubicin (DR). Statistical analysis revealed a KD of 6 ± 0.3 M. B) schematic representation of DR molecule used in binding studies.

Figure 4A

Figure 4B

Figure 4. A) Binding isotherm of AcrD-gentamicin interaction. Isothermal titration calorimetry suggested a KD value of 3.8 M. The stoichiometry suggests that approximately .398 gentamicin molecules bind an AcrD monomer (1.2 molecules/trimer). In the asymmetric model, only the binding protomer interacts with extruding ligands. The ITC data may provide supporting evidence to the asymmetric model, whereby a gentamicin molecule may only be recognized by the binding protomer. B) Schematic structure of gentamicin used in the ITC assay.

Figure 5

Figure 5. The components of gentamicin are displayed. Five varieties of gentamicin can be identified after purification with differences indicated in R1, R2, or R3 groups. Circled in red are those protons with the strongest STD-NMR signals. Labeled withing the gentamicin schematic diagram are the corresponding peaks within the NMR spectrum. This is taken from Deubner, et al. 2003.

Table 3

Table 1. Three STD-NMR experiments were run in replicate to identifify the binding epitope of gentamicin. Results suggest that protons at positions 26, 27, 7, 22, 24, and 5 (Figure 3 and highlighted in yellow) play a significant role in AcrD interaction.

Figure 6

Figure 6. Representative STD-NMR spectrum of gentamicin with AcrD. 20 peaks are identified and labeled based on the proton number from figure 3. Top is reference spectrum, while bottom is spectrum after saturation transfer from AcrD to gentamicin.

Figure 7

Figure 7. A ribbon diagram representing the modeled structure of AcrD. Coloring scheme is as follows: TM1-TM6, raspberry; TM7-TM12, split pea; PC1, marine; PC2, yellow; PN1, orange; PN2, hot pink; DN, Deep teal; DC, ruby. The model was generating using ESyPred3D with MexB (PDB accession 2V50) as the homologus structure. The image was created using Pymol.

Figure 8

Figure 8. Ribbon diagram representation of Trimeric AcrD with protomers colored red, blue, and green, respectively. Monomer was generated by ESyPred3D, while the trimer was created by superimposing three monomeric models of AcrD over trimeric AcrB (PDB accession code 1IWG) using the program coot. Image was made using Pymol.

100

Figure 9

Figure 9. A Top view (perpendicular to the membrane) of the predicted AcrD trimeric structure. Displayed in this view are the suggested ligand entry points including the vestibules (violet), binding clefts (teal), central pore (purple), The TolC docking domain funnel (orange). Individual monomers are colored blue, green, and red, respectively. The image was created using Pymol.

101

Table 4

Antibiotics recognized by AcrD and MIC

WT MIC AcrD expressing MIC Compound Drug Family (µg/ml) (µg/ml) Reference Tetracyline Tetracyclines 0.39 0.78 14 Nalidixic Acid Quinolone* 0.78 1.56 14 Norfloxacin Fluoroquinolone 0.0025 0.05 14 Fosfomycin Epoxide** 1.56 3.13 14 Novobiocin aminocoumarin 1.56 6.25 14 anionic SDS detergent 50 >400 14 Deoxycholic Acid Bile Acid 1.25 >40000 14 Gentamicin aminoglycoside 1.5 6 19 Amikacin aminoglycoside 0.8 3 19

*first synthetic quinolone **only member of the epoxide family of antibiotics

Table 4 lists the substrates used for docking studies along with the associated drug family. Additionally, MIC values for the various ligands with and without AcrD expression are listed in µg/ml. The MIC was not tested for Daunorubicin HCL, therefore it is unlisted.

102

Table 5

Substrate binding energy and binding site

Amikacin Deoxycholate Daunorubicin Fosfomycin Gentamicin kCal/mol site kCal/mol site kCal/mol site kCal/mol site kCal/mol site 1 -9.0 2* -8.7 2* -10.0 2* -4.7 2* -10.2 2* 2 -8.6 2 -8.2 1* -9.5 2 -4.6 2 -9.8 1* 3 -8.1 1* -8.2 2 -9.3 O -4.5 2 -9.6 1 4 -8.0 1 -8.1 2 -8.2 1* -4.5 2 -9.6 1 5 -7.9 1 -8.1 O -9.1 1 -4.3 1* -9.6 1 6 -7.9 2 -8.0 1 -9.1 2 -4.2 1 -9.5 2 7 -7.6 2 -8.0 2 -9.1 O -4.2 2 -9.3 2 8 -7.6 2 -8.0 1 -9.1 1 -4.2 2 -9.2 1 9 -7.5 1 -8.0 2 -8.9 1 -4.1 2 -9.2 1

Naldixic acid Norfloxacin Novobiocin SDS Tetracycline kCal/mol site kCal/mol site kCal/mol site kCal/mol site kCal/mol site 1 -7.3 2* -8.7 2* -10.1 2* -5.9 1* -10.4 2* 2 -7.2 2 -8.3 2 -10.3 2 -5.5 1 -9.9 1* 3 -7.2 2 -8.2 O -9.9 2 -5.3 O -9.3 O 4 -7.2 1* -8.2 1* -9.8 2 -5.3 2* -9.3 1 5 -7.1 2 -7.9 2 -9.5 2 -5.3 2 -9.2 2 6 -7.1 O -7.9 2 -9.4 2 -5.3 2 -9.1 2 7 -6.9 O -7.9 2 -9.3 1* -5.2 2 -9.0 2 8 -6.9 2 -7.8 2 -9.2 1 -5.2 2 -9.0 2 9 -6.9 1 -7.7 1 -9.0 1 -5.2 1 -8.9 1

Table 5 lists the substrates used for docking studies along with the associated binding energies and binding site. Binding sites are indicated as 1, 2, and O for the binding cleft site, TolC-docking domain site, and other, respectively. Those indicated with a (*) were used for binding analysis and imaging.

103

Figure 10

A B

Figure 10. As a result of docking studies using the program Autodock Vina, two suggested and distinct binding regions were located. Illustrated in figure 8A is an overlay of the substrates in binding site 1, which is located in the binding cleft between subdomain PC1 and PC2. B) A zoomed in view of binding site 1 highlights the multifaceted binding site. Withing binding site 1, amikacin (yellow), tetracycline (pink), SDS (teal), and nalidixic acid (blue-grey) binding to distinct but overlapping locations. The image was created using Pymol.

104

Figure 11

A B

C D

105

E F

G H

I J

Figure 11. Binding site 1 of AcrD identified using Autodock Vina with displayed surface electrostatic potential around docked ligands. The prevailing features of the individual binding sites include charged and hydrophilic surfaces. A) Amikacin (yellow) B) Gentamicin (purple) C) Daunorubicin (gray) D) Norfloxacin (white) E) Fosfomycin (orange) F) SDS (tan) G) Deoxycholic acid (dark blue) H) Nalidixic acid (dark green) I) Novobiocin (Brown) J) Tetracyline (light pink). Oxygen and nitrogen atoms are diplayed as red and blue, respectively. Positive (blue) and negative (red) charges on the AcrD surface displayed using Pymol. 106

Figure 12

Figure 12. The second binding site suggested from Autodock Vina is highlighted in figure 10. Illustrated in figure 8A is an overlay of the substrates in binding site 2, which is located in the region between the porter domain and TolC-docking domain B) A zoomed in view of binding site 2 highlights the multifaceted binding site. Withing binding site 2, Duanorubicin (orange), tetracycline (pink), nalidixic acid (gray), and gentamicin (purple) bind to distinct but overlapping locations. The image was created using Pymol.

107

Figure 13

B A

C D 108

E F

G H

I J

Figure 13. Binding site 2 of AcrD identified using Autodock Vina with displayed surface electrostatic potential around docked ligands. The prevailing features of the individual binding sites include charged and hydrophilic surfaces. A) Amikacin (yellow) B) Deoxycholic acid (dark blue) C) Daunorubicin (gray) D) Gentamicin (purple) E) Fosfomycin (orange) F) Nalidixic acid (dark green) G) Norfloxacin (white) H) Novobiocin (Brown) I) SDS (tan) J) Tetracyline (light pink). Oxygen and nitrogen atoms are diplayed as red and blue, respectively. Positive (blue) and negative (red) charges on the AcrD surface displayed using Pymol.

109

FIGURE 14

Figure 14. To identify if there were tunnels connecting the binding sites to the exterior solvent, Caver was used to generate probably ligand entrance and exit pathways. A) Using the amikacin binding site 1 as a starting point, it was intriguing to identify an entrance from the cleft as well as the central cavity. Furthermore, a tunnel can be observed leading to the second binding site and ultimately out of the AcrD TolC-docking domain funnel. B) top view of the tunnels generated by caver showing the exit pathway of suggested ligands. Pictures were generated using Pymol, whereby AcrD is displayed in trimeric form in green ribbon, while the tunnel surfaces are shown in red.

110

REFERENCES

1. Nikaido, H. (1998) Multiple antibiotic resistance and efflux, Curr. Opin. Microbiol. 1 , 516-523.

2. Nikaido, H. (1998) Antibiotic resistance caused by Gram-negative multidrug efflux pumps. Clin. Infect. Dis. 27 , (Suppl. 1):S32–S41.

3. Higgins, C. F. (1992) ABC transporters: from microorganisms to man, Annu. Rev. Cell Biol. 8 , 67-113.

4. Marger, M., and Saier, M. H., Jr. (1993) A major superfamily of transmembrane facilitators that can catalyze uniport, symport and antiport, Trends Biochem. Sci. 18 , 13–20.

5. Pao, S. S., Paulsen, I. T., and Saier, M. H. Jr. (1998) Major facilitator superfamily, Microbiol. Mol. Biol. Rev. 62 , 1-34.

6. Brown, M. H., Paulsen, L. T., and Skurray, R. A. (1999) The multidrug efflux protein NorM is a prototype of a new family of transporters, Mol. Microbiol. 31 , 394–395.

7. Morita, Y., Kodama, K., Shiota, S., Mine, T., Kataoka, A., Mizushima, T., and Tsuchiya, T. (1998) NorM, a putative multidrug efflux protein, of Vibrio parahaemolyticus and its homolog in Escherichia coli , Antimicrob. Agents Chemother. 42 , 1778-1782.

8. Griffith, J. K., Baker, M. E., Rouch, D. A., Page, M. G., Skurray, R. A., Paulsen, I. T., Chater, K. F., Baldwin, S. A., and Henderson, P. J. (1992) Membrane transport proteins: Implications of sequence comparisons, Curr. Opin. Cell Biol. 4 , 684-695.

9. Tseng, T. T., Gratwick, K. S., Kollman, J., Park, D., Nies, D. H., Goffeau, A. and Saier, M. H., Jr. (1999) The RND permease superfamily: an ancient, ubiquitous and diverse family that includes human disease and development proteins, J. Mol. Microbiol. Biotechnol. 1 , 107–125.

10. Paulsen, I. T., Skurry, R. A., Tam, R., Saier, M. H., Jr., Turner, R. J., Weiner, J. H., Goldberg, E. B., and Grinius, L. L. (1996) The SMR family: A novel family of multidrug efflux proteins involved with the efflux of lipophilic drugs, Mol. Microbiol. 19 , 1167–1175. 111

11. Schweizer, H. P. (2003) Efflux as a mechanism of resistance to antimicrobials in Pseudomonas aeruginosa and related bacteria: unanswered questions, Genet. Mol. Res. 31 , 48–62.

12. Nikaido, H. (1998) Antibiotic resistance caused by Gram-negative multidrug efflux pumps, Clin. Infect. Dis. 27 , S32-S41.

13. Paulsen, I. T., Sliwinski, M. K., and Saier, M. H. Jr. (1998) Microbial genome analyses: global comparisons of transport capabilities based on phylogenies, bioenergetics and substrate specificities, J. Mol. Biol. 277 , 573-592.

14. Nishino, K., and Yamaguchi, A. (2001) Analysis of a complete library of putative drug transporter genes in Escherichia coli , J. Bacteriol. 183 , 5803- 5812.

15. Goldberg, M., Pribyl, T., Juhnke, S., and Nies, D. H. (1999) Energetics and topology of CzcA, a cation/proton antiporter of the resistance-nodulation-cell division protein family, J Biol Chem. 274 , 26065-70.

16. Aires, J. R., and Nikaido, H. (2005) Aminoglycosides are captured from both periplasm and cytoplasm by the AcrD multidrug efflux transporter of Escherichia coli , J. Bacteriol. 187 , 1923-1929.

17. Tsukagoshi, N., and Aono, R. (2000) Entry into and release of solvents by Escherichia coli in an organic-aqueous tow-liquid-phase system and substrate specificity of the AcrAB-TolC solvent-extruding pump, J. Bacteriol. 182 , 4803– 4810.

18. White, D. G., Goldman, J. D., Demple, B., and Levy, S. B. (1997) Role of the acrAB locus in organic solvent tolerance mediated by expression of marA , soxS , or robA in Escherichia coli. J. Bacteriol. 179, 6122–6126.

19. Rosenberg, E. Y., Ma, D., and Nikaido, H. (2000) AcrD of Escherichia coli is an aminoglycoside efflux pump, J. Bacteriol. 182 , 1754-1756.

20. Von Ahsen, U., Davies, J., and Schroeder, R. (1991) Antibiotic inhibition of group I ribozyme function, Nature 353 , 368-370.

21. Fourmy, D., Recht, M. I., Blanchard, S. C., and Puglisi, J.D. (1996) Structure of the A site of Escherichia coli 16S ribosomal RNA complexed with an aminoglycoside antibiotic, Science 274 , 1367-1371

22. Screpanti, E., Padan, E., Rimon, A., and Hunte, M. H. (2006) Crucial steps in the structure determination of the Na+/H+ antiporter NhaA in its native conformation. J Mol Biol 362 , 192–202. 112

23. Prive, G. G. (2007) Detergents for the stabilization and crystallization of membrane proteins, Methods 41 , 388–397.

24. ORIGIN Ver. 7.5. OriginLab Corporation, Northampton, MA, USA.

25. Su, C.-C., Nikaido, H., and Yu, E. W. (2007) Ligand-transporter interaction in the AcrB multidrug pump determined by fluorescence polarization assay, FEBS Letters 581 , 4972–4976.

26. Mayer, M., and Meyer, B. (2001) Group Epitope Mapping by Saturation Transfer Difference NMR To Identify Segments of a Ligand in Direct Contact with a Protein Receptor, J. Am. Chem. Soc. 123 , 6108–6117.

27. Piotto, M., Saudek, V., and Sklenar, V. (1992) Gradient-tailored excitation for single-quantum NMR spectroscopy of aqueous solutions, J. Biomol NMR 2 , 661-665.

28. Deubner, R., Schollmayer, C., Wienen, F., and Holzgrabe, U. (2003) Assignment of the major and minor components of gentamicin for evaluation of batches, Magn. Reson. Chem. 41 , 589-598.

29. Lambert, C., Leonard, N., De Bolle, X., and Depiereaux, E. (2002) ESyPred3D: Prediction of proteins 3D structures Bioinformatics , 18 , 1250- 1256.

30. Sali, A., and Blundell, T. L. (1993) Comparative protein modelling by satisfaction of spatial restraints, J. Mol. Biol. 234 , 779-815.

31. Emsley, P., and Cowtan, K. (2004) Coot: Model-building tools for molecular graphics, Acta Crystallog. Sect. D , 60 , 2126-2132.

32. Trott, O., Olson, A. J. (2010) AutoDock Vina: improving the speed and accuracy of docking with a new scoring function, efficient optimization and multithreading, Journal of Computational Chemistry 3, 455-461.

33. Murakami, S., Nakashima, R., Yamashita, E., Matsumoto, T., and Yamaguchi, A. (2006) Crystal structures of a multidrug transporter reveal a functionally rotating mechanism, Nature 443 , 173-179.

34. Pet řek, M., Otyepka, M., Banáš, P., Košinová, P., Ko ča, J. and Damborský, J. (2006) CAVER: A New Tool to Explore Routes from Protein Clefts, Pockets and Cavities, BMC Bioinformatics 7 , 316.

35. Seeger, M. A., Schiefner, A., Eicher, T., Verrey, F., Diederichs, K., and Pos, K. M. (2006) Structural asymmetry of AcrB trimer suggests a peristaltic pump mechanism, Science 313 , 1295-1298. 113

36. Sennhauser, G., Amstutz, P., Briand, C., Storchenegger, O., and Grutter, M. G. (2007) Drug export pathway of multidrug exporter AcrB revealed by DARPin inhibitors, PLoS Biol. 5, e7.

37. Sennhauser, G., Bukowska, M. A., Briand, C., and Grütter, M. G. (2009) Crystal structure of the multidrug exporter MexB from Pseudomonas aeruginosa , J. Mol. Biol. 389, 134-45.

114

CHAPTER 4. TetR REVIEW

STRUCTURES OF AcrR AND CmeR: INSIGHT INTO THE MECHANISM OF TRANSCRIPTIONAL REPRESSION AND MULTI-DRUG RECOGNITION IN THE TetR FAMILY OF REGULATORS

Mathew D. Routh a, Chih-Chia Su b, Qijing Zhang c and Edward W. Yu a, b, d,

aMolecular, Cellular and Developmental Biology Interdepartmental Graduate Program, Iowa State University, IA 50011, USA Department of Biochemistry, Biophysics and Molecular Biology, Iowa State University, Ames, IA 50011, USA cDepartment of Veterinary Microbiology and Preventive Medicine, College of Veterinary Medicine, Iowa State University, Ames, IA 50011, USA dDepartment of Physics and Astronomy, Iowa State University, Ames, IA 50011, USA

Abstract

The transcriptional regulators of the TetR family act as chemical sensors to monitor the cellular environment in many bacterial species. To perform this function, members of the TetR family harbor a diverse ligand-binding domain capable of recognizing the same series of compounds as the transporters they regulate. Many of the regulators can be induced by a wide array of structurally unrelated compounds. Binding of these structurally unrelated ligands, to the regulator, results in a conformational change that is transmitted to the DNA-binding region causing the repressor to lose its DNA-binding capacity and allowing for the initiation of transcription. The multi-drug binding proteins AcrR of Escherichia coli and CmeR from Campylobacter jejuni are members of the TetR family of transcriptional repressors that regulate the expression of the multidrug resistant efflux pumps AcrAB and CmeABC, respectively. To gain insights into the mechanisms of transcriptional regulation and how multiple ligands induce the same physiological response, we determined the crystal structures of the AcrR and CmeR regulatory proteins. In this review, we will summarize the new findings with AcrR and CmeR, and discuss the novel features of these two proteins in comparison with other regulators in the TetR family. 115

Keywords: TetR family; Escherichia coli AcrR; Campylobacter jejuni CmeR; Transcriptional regulator; Multidrug resistance

ARTICLE OUTLINE

1. Introduction 2. The AcrR regulator

2.1. Crystal structure of AcrR in space group of P222 1

2.2. Crystal structure of AcrR in space group of P31 3. The CmeR regulator 3.1. Crystal structure of CmeR 4. Conclusions and perspectives Acknowledgements References

INTRODUCTION

Bacterial infections are commonly treated with various classes of antibiotics. The clinical treatment is necessary for curing infectious diseases, but an unintended consequence of the treatment is the selection of bacterial pathogens with elevated levels of resistance to antibiotics. Constant emergence and spread of antibiotic resistance has become a major threat to the health of humans and animals (1). Bacterial organisms utilize multiple mechanisms to combat antibiotics and antimicrobial agents. One important mechanism that gives rise to multidrug resistance (MDR) is the expression of multidrug efflux transporters that are capable of reducing the intracellular concentration of toxic compounds (2-7). The expression of these transporters is tightly controlled at the transcriptional level by global and local regulators (2). Transcriptional regulators serve as the cellular intermediate between chemical stress and response. In most cases, the regulators recognize the same series of compounds as the transporters they regulate (7). Transcriptional activators, belonging to one of three families, including MerR, AraC, and LysR, are 116

able to drive transcription of MDR pumps through their interaction with their cognate promoters and RNA polymerase (8, 9). The global activators SoxS, Rob, and MarA (belonging to the AraC family of regulators) in Escherichia coli are able to activate a group of 40 promoters (known as the mara/soxs/rob regulon ) through their interaction with the marbox (10). Due to variations in affinity between the activators and the marbox , each of the 40 promoters in the mara/soxs/rob regulon will respond differently to varying levels of the activators and allow for stress specific responses. Local transcriptional regulation of MDR pumps is achieved by the action of DNA binding proteins capable modulating expression levels of specific proteins. In bacteria, local transcriptional regulation can involve either one-component or two- component regulatory systems. Two-component regulatory systems control protein expression through the function of a membrane-bound sensor kinase and a cytoplasmic response regulator, which is a DNA-binding protein (10-18). The membrane-bound kinase is responsible for receiving external signals and transmitting the information into the cell by phosphorylating the DNA-binding protein. The phosphorylated DNA-binding protein then modulates gene transcription by interacting with its cognate DNA. A key feature of two-component regulatory systems is the phosphorylation between sensor kinase and response regulator. On the other hand, One-component bacterial transcriptional regulators modulate gene expression levels using a single two-domain protein where one domain receives signals and the other domain binds specific DNA sequences to regulate transcription (8, 9, 19). Information flow between the two domains is through conformational changes, contrasting the phosphorylation events required in two-component systems. Structural analyses revealed that almost 95% of all known prokaryotic transcriptional factors employ the helix–turn–helix (HTH) motif to bind their target DNAs (9). Prokaryotic transcriptional repressors are classified in families based on their functional and sequence similarities and include proteins that belong to the TetR, MarR, or LacI family or of repressors (8, 9, 19). Generally, repressors bind their cognate operator region as a dimer to locally inhibit transcription and upon ligand induction, the repressor will release from the operator to allow transcription of 117

the adjacent MDR transporter. Members of the TetR family, such as AcrR of E. coli , are two-domain proteins and bind a specific inverse repeat (IR) operator site through the interaction of their N-terminal HTH motif. Inducing ligands interact with the protein inside a ligand binding pocket located in the C-terminal region of the protein which induces a conformational change in the DNA binding region, resulting in the repressor releasing from the operator site. The transcriptional regulation that is provided by the activators and repressors creates a tightly controlled cellular environment that is able to respond rapidly to various cellular stresses. Presumably, these transcriptional factors act as cytosolic chemical sensors by responding to threatening levels of toxic compounds (8, 9, 19).

Naming of the TetR family comes from the most completely characterized member of the family, the TetR protein (19). TetR regulates transcription of the tetA gene. TetA confers resistance to tetracycline by pumping the antibiotic out of the bacterial cell (19-23). Members of the TetR family are defined based on structural, sequence, and functional similarities. Structurally, the TetR repressors are all-helical proteins that harbor two distinct domains, a larger C-terminal domain and a smaller N-terminal domain (19). Sequence similarities exist, most notably, in the N-terminal domain of the protein, which in all characterized TetR family members is the DNA- binding domain. In most cases, the N-terminal domain contains three α helices with two of the helices forming the signature HTH motif characterized by this family. Most likely due to the varying array of ligands that TetR family members recognize, sequence similarities do not exist in the C-terminal ligand-binding domain although the crystal structures reveal an overall similar topology among members of the TetR family. It is predicted that the structural similarities in the C-terminal domain can be attributed to a common mechanism of ligand induction used by these regulatory proteins (19).

Insights into the function of the TetR family have come from the crystal structures of the QacR in various conformations (24, 25). QacR is a transcriptional regulator that represses the transcription of the qacA gene. QacA is a membrane associated MDR 118

pump of the MFS family of transporters. Its expression is associated with increased resistance levels of Staphylococcus aureus to quaternary ammonium compounds. Interestingly, QacR binds to its 28 bp cognate operator site (upstream of QacA) in a dimer to DNA ration of 2:1 (25). The interaction seems to be cooperative, in which the binding of one dimer initiates a change in the local DNA structure, altering the distance between consecutive major grooves in B-form DNA from 34Å to 37Å, which drives the binding of the other dimer. This DNA-binding mode is in direct contrast to that seen for TetR, where a single TetR dimer binds to a 15bp operator sequence. In spite of the number of dimers involved in DNA-binding, it seems that the interactions of QacR, TetR, and possibly other members of the TetR family bind the palindromic operator with a similar mechanism. In both QacR and TetR, The N- terminal DNA binding domain consists of 3 α-helices, of which α3, known as the recognition helix, makes the most base-specific DNA interactions. Helices α1 and α2 are associated with mainly non-sequence specific interactions to the phosphate backbone of the DNA strands.

QacR can be induced by a wide array of structurally dissimilar cytotoxic compounds. Many of these ligands, including ethidium bromide, dequalinium, rhodamine 6G, malachite green, berberine, crystal violet, proflavin, DB75, DB359, pentamidine, and hexamidine have been co-crystallized with QacR to provide information of ligand induction and multi-drug recognition (25-28). Briefly, an inducing ligand binds inside the ligand-binding pocket of the C-terminal domain with a ratio of one ligand per dimer. Structurally distinct drugs are recognized inside the binding pocket due to the presence of at least two distinct, multi-faceted binding sites. Binding initiates a helix-to-coil transition of α5 and elongates the C-terminal of the helix by one turn in one of the two QacR monomers. This transition relocates α6 and the tethered DNA-binding domain, increasing the distance between the two DNA-binding domains by 11Å (from 37Å to 48Å) compared to the DNA-bound structure. The increased distance between the DNA-binding domains disrupts the interactions between QacR and DNA resulting in QacR releasing from its cognate operator (29). Induction of TetR occurs through a mechanism distinct from that 119

observed in QacR. Tetracycline-Mg 2+ binding in both monomers causes helix α6 to shift creating a β-turn which pushes the loop between helices α6 and α7 closer to the inducer. Contacts are made between tetracycline-Mg 2+ and this loop region resulting in the movement of helix α6. Due to van der Waals contacts, the movement of α6 drives the movement of helix α4. Helix α4 acts as a pendulum that separates helix α3 and α3´ by 3Å in the ligand-bound conformation which allows TetR to release from the operator (19). The ligand-bound structure of CmeR will be discussed later in this review to provide additional information on the C-terminal domain and ligand recognition. It is important to note that, as revealed by the QacR and TetR crytstal structures, proteins that are homologous in structure and function may utilizize slightly different mechanisms to perform the same function, i.e., repression. Thus, it is critical to understand the regulatory mechanism of other members of the TetR family.

Understanding the molecular mechanisms of transcriptional regulation is vital due to the potential that these regulatory proteins can offer for new drug targets. Recently, the crystal structures of AcrR (30) and (31), a transcriptional regulator of the AcrAB efflux pump in Escherichia coli , and CmeR (32), a regulator that represses the expression of CmeABC in Campylobacter jejuni , have been determined. Induction of AcrR is initiated through an interaction of cationic and neutral ligands. In contrast, CmeR more favorably recognizes anionic and uncharged compounds. In this review we will describe the structural features of these two regulatory proteins and discuss the valuable insights that they provide for delineating the mechanisms of gene regulation and multidrug recognition by these TetR family regulators. The details described in this manuscript will add to the model provided by QacR and TetR.

THE AcrR REGULATOR

E. coli AcrB is a prototypical multidrug transporter that belongs to the resistance- nodulation-division (RND) superfamily of MDR pumps (33) and (34). Of all currently 120

characterized multidrug transporters, AcrB possesses the widest range of ligand recognition. It is capable of recognizing many structurally dissimilar compounds, including most of the currently administered antibiotics, chemotherapeutic agents, bile salts, dyes, and detergents (35) and (36). This inner membrane efflux pump functions in conjunction with the periplasmic membrane fusion protein, AcrA (37), and the outer membrane channel protein, TolC (38), to export a diverse range of compounds completely out of the bacterial cell.

The expression of AcrAB is modulated by the transcriptional regulator AcrR, whose open reading frame is located 141 bp upstream of the acrAB operon and is transcribed divergently (39). Transcription of the acrR gene gives rise to a 215 amino acid protein, which shares N-terminal sequence and structural similarities to members of the TetR family (19). The signatures of the TetR family of regulators include a homologous N-terminal three-helix DNA-binding domain and a diverse C- terminal ligand-binding domain (19). Experimental evidence suggests that the 24 base pair palindromic inverted repeat (IR) sequence (5 ′TACATACATT TGTG AATGTATGTA 3′), located between the acrR and acrAB genes and overlapping with the acrAB promoter, is the target DNA for AcrR (39) and (40). It has been demonstrated through fluorescence polarization and gel filtration

that AcrR binds to this IR as a dimer of dimers, with a dissociation constant ( KD) of 20.2 nM (30) and (40). This suggests that the binding of AcrR to its IR resembles that of QacR, which binds a 28 bp IR1 sequence as a dimer of dimers, but is distinct from many other members of the TetR family where the interaction consists of a dimer bound to an 15 bp IR (19), (41), (42) and (43). The diverse C-terminal region of the TetR family of regulators possesses unique sequences, which allows different regulators in the family to accommodate specific sets of inducing ligands. Upon ligand binding, the AcrR regulator is presumed to dissociate from its target DNA to allow the expression of the AcrAB efflux complex which, in turn, protects the bacterial cell from toxic substances. 121

Recent studies indicate that AcrR interacts with the same set of antimicrobial agents as AcrB with strikingly similar affinities (40) and (44). Su et al. (40) demonstrated that AcrR binds ethidium bromide (Et), proflavin (Pf), and rhodamine 6G (R6G) with

dissociation constants of 4.2, 10.1, and 10.7 M (40); while the KD values for AcrB with these ligands are 8.7, 14.5, and 5.5 M (44), respectively. These affinities also coincide with those observed for QacR (45), BmrR (46), and TtgV (47). The KDs in this range may be optimal to initiate the expression of MDR pumps while the antibiotic concentration is at the sub-inhibitory level. Each AcrR monomer binds an inducing ligand, and thus a dimer can accommodate two identical molecules. Fluorescence polarization assays also suggest that AcrR binds many structurally unrelated ligands in distinct but possibly overlapping binding sites(40). For instance, Et-saturated AcrR can accommodate Pf with equal affinity as apo-AcrR while Et and R6G seem to be competing for the same binding site (40). The 1:1 ligand-to- monomeric AcrR stoichiometry is similar to that of TetR (41), but distinct from the 1:2 ligand-to-monomer ratio of QacR (42). AcrR is unique in that its ligand binding mode is similar to TetR while its mode of DNA-binding is related to that of QacR. Exploration of the AcrR induction may provide us with new insight into the mechanisms that the TetR family utilizes to regulate genes.

Crystal structure of AcrR in space group of P222 1: The crystal structure of

AcrR in space group P222 1 (30) is illustrated in Fig. 1a. It reveals a dimeric protein composed almost entirely of α-helices with an overall architecture similar to members of the TetR family, including TetR (42) and (48), QacR (41) and (49), EthR (50) and (51), CprB (52), CmeR (32), ActR (53), HapR (54), and IcaR (55). Each subunit in the dimer comprises nine helices ( α1 to α9 and α1′ to α9′). Of this two- domain protein, helices α1 to α3 make up the N-terminal DNA binding domain while the larger C-terminal ligand-binding domain consists of helices α4 through α9. A high degree of conservation exists in the DNA-binding region among members of the TetR family of transcriptional regulators (30). This can be attributed to the critical role of specific amino acids in contacting the phosphate backbone of the DNA strand and to the overall function of the TetR family proteins in transcriptional repression. When 122

the N-terminal domain of AcrR (amino acids 12–62) is aligned against QacR and CmeR, 43% amino acid identity is observed in both cases (30). Superimposition of this domain to QacR reveals a very similar overall structure, which is reflected by an overall rmsd of 1.2 Å calculated over the C α atoms.

As the DNA-binding mode of AcrR is expected to be similar to that of QacR, a speculative model of DNA-bound AcrR (Fig. 1b) was generated by aligning its domains to those of DNA-bound QacR (30). This model suggests that R45, G46, Y49, W50, H51 and K55 are important for IR binding. Among these residues, R45 interacts directly with four different bases of the DNA, confirming its critical role for IR recognition. Therefore, it is not surprising that a recent sampling indicated six of 36 isolated fluoroquinolone-resistant E. coli strains had a mutation at codon 45 (Arg → Cys), and all six R45C mutants showed evaluated resistance to multiple antibiotics (56).

The C-terminal regulatory domain of AcrR comprises six helices, including helices α4 through α9. Helices α8 and α9 form the majority of the dimerization surface with small contributions from helices α6 and α7, creating a 2002 Å2/monomer buried, mostly hydrophobic contact region (30). Owing to differences in inducing ligands and varying specificity, the C-terminal domain of AcrR possesses little sequence homology to other members of the TetR family of transcription regulators. It is intriguing that superposition of AcrR (30), QacR (49), and CmeR (32) reveals significant topological similarities in the C-terminal domains. This, in part, may be attributed to a similar functional role that the C-terminal domains play in recognizing inducing ligands and transmitting the signal to the N- terminal DNA-binding regions. Like other members of the TetR family, a large internal cavity, with a total volume of 350 Å3 is formed in the C-terminal region of AcrR (Fig. 1c). This cavity, surrounded by helices α4 through α8 of each monomer, is predicted to form the multidrug-binding pocket in AcrR (30). It should be noted that the C-terminal domain of apo-QacR does not have a ligand-binding cavity. A unique characteristic of AcrR is the presence of three openings to the drug-binding pocket. 123

Two of the openings are located at the front and side surfaces of each monomer and appear to be orthogonal to each other. The loop between helices α4a and α4b contributes to form part of the openings. The other opening is located at the dimer interface and is partially blocked by the loop between helices α8′ and α9′ from the second subunit. It is likely that drug molecules may enter AcrR through the loop region between helices α4a and α4b.

Docking of ligands into the AcrR structure suggested that the large cavity created by helices α4 to α8 could accommodate different drugs, including Et, Pf and ciprofloxacin (Cip) (30). In each case, the bound drug was completely buried in the AcrR molecule, and strong interaction was observed between bound drug and the regulator. Predictions also indicated that the binding sites for Et, Pf and Cip are distinct, but partially overlapped in each monomer of AcrR. Fig. 1d depicts the multidrug binding site formed by the C-terminal domain of the regulator. The extensive binding pocket, which is created by helices α4 through α8, is mostly hydrophobic in nature, with W63, I70 and F114 predicted to make important hydrophobic contacts with the inducing ligands. In addition to these hydrophobic interactions, residues E67 and Q130 are predicted to make electrostatic interactions with bound drug to secure the binding. Among these amino acids, E67 seems to be of particular importance for drug recognition. It was found that a mutation of this residue by an alanine, E67A, abolished the binding of Pf, Et and R6G to the regulator (30).

Crystal structure of AcrR in space group of P31: Recently, a new crystal structure of AcrR with space group P31 (31), which is distinct from the P222 1 space group structure, was determined. A comparison of these two structures reveals considerable conformational changes at both the N-terminal and C-terminal regions, suggesting that these two structures represent different conformational states of AcrR. These crystal structures have provided valuable insight into the mechanisms of ligand binding and AcrR regulation. 124

The overall structure of AcrR with the space group P31 is quite similar to the

P222 1 space group structure. A detailed comparison, however, reveals a significant change in conformation at the N-terminal DNA-binding domain. This change results in an overall rmsd of 2.8 Å calculated over the C α atoms at the N-terminal domains (residues 7–65), in contrast to the < 0.7 Å rmsd of the C-terminal domains (residues 73–210) (13). Fig. 2 illustrates a superposition of these two AcrR structures using the program ESCET 57).

Conformational changes between the P31 and P222 1 structures seem to be predominantly rigid-body translation and rotation at the N-terminal domain. These movements lead to a downward shift of the entire N-terminal DNA-binding domain of the P31 structure (with respect to the orientation shown in Fig. 1a) by 2.6 Å, and a rotation of 10° towards the subunit interface of the dimer when compared with that of the structure of P222 1 (Fig. 2). As a consequence of these movements, the two N- terminal domains of the AcrR dimer, in the P31 structure, move closer to each other by approximately 2 Å. The center-to-center distance between recognition helices α3 and α3′ (as measured by the distance between C α atoms of Y49 and Y49 ′) decreases from 42 Å in the P222 1 structure to 39 Å in the P31 structure (31). To bind two consecutive major grooves of B-DNA, the center-to-center distance has to be 34 Å. This distance is thought to increase upon drug binding, which in turn inhibits the binding of the regulator to its operator DNA. With the observed center-to-center distances in the crystal structures, it is likely that the conformation of the DNA-bound form of AcrR is more similar to the P31 structure, while its drug-bound form is more closely related to the P222 1 structure. In addition to these differences, R45, an N- terminal amino acid previously identified to be critical for DNA binding and AcrR regulation (56), undergoes a significant conformational change. The C α–Cα distance

between R45 and R45 ′ decreases from 40 Å in the P222 1 structure to 35 Å in the

P31 structure (31).

When examining the C-terminal domain, the most striking conformational change involves the amino acid E67. This residue may act as a molecular switch 125

that drives the change of conformations throughout the AcrR molecule. Superimposition of the two AcrR structures reveals that the C α atom of E67 shifts by 4.2 Å. This shift initiates considerable changes in the C-terminal domain of AcrR, including helix α4a shifting towards the N-terminal domain by 2.3 Å and a local unwinding of the N-terminal end of helix α6 which shortens the helix by one turn. The local unwinding and overall change result in the disruption of the hydrogen bonded

network connecting the N- and C-terminal domains. In the P222 1 structure, it was found that R106 is H-bonded with E67 in the drug binding site, and the C-terminal domain residue R105 is H-bonded with the N-terminal domain residues Q14 and

D18, respectively (Fig. 1a) (30) and (31). These H-bonds are missing in the P31 structure.

Based on the P31 and P222 1 structures of AcrR, we suspect that the changes in conformation of the N-terminal DNA-binding and C-terminal drug-binding domains of AcrR are cooperative, due to the formation of H-bonds at the interface between these two domains (Fig. 1a). In the DNA-bound form of AcrR, the structure of the

regulator may be closer to the P31 structure. Thus, the side chain of E67 may point outside the drug-binding pocket, exposed to the solvent. Drug-binding within the C- terminal domain may induce structural changes resulting in a conformation more

closely related to the P222 1 structure, in which the side chain of E67 flips into the interior of the hydrophobic core. This change may also be accompanied by the formation of new H-bonds between E67 and R106, R105 and Q14, and R105 and D18. The crystal structures of both DNA-bound and drug-bound AcrR would be necessary to confirm the change in conformation upon DNA and drug binding.

Crystallization of AcrR-IR: In efforts to further understand the conformational changes that take place throughout AcrR, crystallization attempts were made with C-6XHis AcrR and Double stranded DNA helices of various lengths (IR1-IR4, table X) (Integrated DNA technologies). Initially, C-6XHis AcrR was purified as outlined previously (12). The purified C-6XHis AcrR protein was mixed with the indicated IR fragment in a 4:1 molar ratio and subsequently passed through 126

a G-200 gel fitration column pre equilibrated with buffer. Fractions were collected and analyzed with 15% SDS–PAGE stained with Coomassie Brilliant Blue to detect AcrR. Furthermore, a 1% agarose gel was used on fractions containing AcrR to further detect the presence of DNA. Those fractions with AcrR-IR were pooled and concentrated to 20 mg/ml using a YM-10 Centriprep concentrator. A gel-mobility- shift assay was performed to further confirm AcrR-IR interaction.

The full-length C-6XHis AcrR protein was crystallized in 24-well plates using the hanging drop vapor-diffusion method crystals. An ambient temperature of 25ºC was maintained throughout crystal growth. Initial crystallization trials were performed using Hampton reagents, including Crystal Screen 1, 2, and Lite. With little success, subsequent screens were performed by mixing various polyethylene glycols (PEGs ranging from PEG 200 to PEG 20000) with different buffers (between pH 3.5 and pH 9.5), salts, and additives. Following extensive screening, a mother

solution containing PEG 6000 14%, .3M CaCl 2, .02M MgCl 2, .1M NaMES (6.5), Glycerol 5%, and 20 mM dithiothreitol produced the highest diffracting crystal (Figure 3). With this condition, the four distinct IR strands were screened with the IR2- C- 6XHis AcrD complex crystals diffracting to 3.2 Å (Figure 4). For data collection, a single native crystal of IR2- C-6XHis AcrD was flash- cooled in a cryoprotectant solution containing the mother solution plus 25% glycerol at 100 K. The glycerol concentration was gradually increased to 25% by 5% increments. The diffracted anisotropically to a resolution of 3.2 Å. Fig. X depicts one of the diffraction images of the native crystal. Diffraction data sets were obtained from the native IR2- C-6XHis AcrD crystals at the Advanced Photon Source (APS, beamline 24IDC) at cryogenic temperature (100 K) using an ADSC Quantum 315 CCD-based detector. The beam size was 50 X 20 mm.

THE CmeR REGULATOR

Campylobacter jejuni is the leading cause of food-borne enteritis to humans in the USA as well as other developed countries (58). C. jejuni is able to infect animal 127

hosts, and colonize the intestinal tracts of these animals. To withstand the various deleterious conditions both in vitro and in vivo , this Gram-negative microorganism harbors 13 putative MDR transporter genes (according to the genomic sequence of NCTC 11168) that may be used to extrude antimicrobial compounds that Campylobacter may encounter in the intestinal tract (59) and (60). Some of these multidrug transporters have been linked to the intrinsic and acquired resistance of Campylobacter to various antibiotics (61). Currently, CmeABC and CmeDEF (belonging to the RND-family of MDR proteins) are the only two, of the 13 predicted, antibiotic-resistance transporters that have been functionally characterized in this Gram-negative microorganism (62-65).

A primary contributor to antibiotic resistance in C. jejuni is the CmeABC efflux system. CmeABC is a tripartate RND efflux transporter (63). It consists of an inner membrane efflux pump, CmeB, a periplasmic membrane fusion protein, CmeA, and an outer membrane channel, CmeC. Together, these three proteins effectively mediate the extrusion of commonly used antibiotics, metal ions, and lipophilic compounds out of the bacterial cell (63-67). Importantly, CmeABC is a key player in C. jejuni for resistance to bile salts, which are ubiquitously present in the intestinal tract. CmeABC deletion mutants are unable to colonize in the intestinal tract of chickens (68), indicating the essential role of CmeABC in adapting to the in vivo environment. Expression of the CmeABC efflux complex is inducible by bile salts (67) and potentially by other unidentified ligands. Understanding the regulation of cmeABC is an important step in the elucidation of the mechanism of multi-drug extrusion in Campylobacter .

Transcription of cmeABC is controlled by the transcriptional regulator CmeR (69). The gene for cmeR is located immediately upstream of cmeABC and encodes a 210 amino acid protein that shares N-terminal sequence and structural similarities to members of the TetR family of transcriptional repressors (19) and (70). CmeR is a two-domain protein with an N-terminal DNA-binding region and a predicted C- terminal ligand-binding domain. The 16 bp IR sequence, 128

5′TGTAATA AA TATTACA 3′, located between cmeR and cmeABC is shown to be the operator site of CmeR binding and transcriptional repression (69). Transcriptional repressors of the TetR family bind their IR operator sites which generally overlap the promoter sequences of the regulated genes. This interaction inhibits RNA-polymerase from binding or blocks the transcriptional initiation event to repress gene expression. Alterations that affect CmeR–operator binding, including deletions of CmeR and single nucleotide mutations in the operator site, releases the repression of CmeR and results in overexpression of CmeABC (69) and (71). In addition, CmeR–DNA interactions are inhibited when inducing ligands, such as bile salts, interact with CmeR and cause a conformational change in the protein that renders it unable to bind to its operator DNA. Guo et al. (72) used DNA microarrays and real-time quantitative reverse transcription-PCR to analyze the regulatory network of cmeR. The results showed that in addition to repressing the transcription of cmeABC , CmeR functions as a pleiotropic regulator and modulates the expression of at least 28 other genes in C. jejuni (72). The array of genes regulated by CmeR outlines the important role this regulator plays in the adaptive response to the intestinal environment.

Crystal structure of CmeR: Recently, the crystal structure of CmeR has been determined (Fig. 5a) (32). This work revealed novel structural features of a TetR family regulator, and has brought new insights into the mechanisms of transcriptional regulation and ligand recognition. Like AcrR, CmeR is a dimeric two- domain molecule with an entirely helical architecture with similar topology to other members of the TetR family. Distinct from the other members of the TetR family, CmeR exhibits a unique crystal structure that lacks the α3 helix (replaced by a random coil) which is involved in DNA recognition. Along with this unique characteristic, a large center-to-center distance (54 Å as measured by the separation between C α atoms of Y51 and Y51 ′ from the other subunit) was observed between the two N-termini of the dimer. In addition, a large flexible ligand-binding pocket is found to form in the C-terminal domain of CmeR. Each monomer forms a 20 Å long tunnel-like cavity in the ligand-binding domain of CmeR and occupies a 129

volume of about 1000 Å3 (Fig. 5b), which is approximately three times of that of AcrR. As CmeR recognizes anionic and neutral ligands, the structure offers the first glance on how anionic and uncharged ligands are bound by a regulator from the TetR family.

The crystal structure of the dimeric CmeR regulator is shown in Fig. 5a. This structure revealed that each subunit of CmeR is composed of nine α helices, in which the short recognition α3 helix, presumably formed by residues 47–53, is missing (32). To facilitate comparisons with other TetR members, α3 was excluded from the numbering of the α-helical segments. Thus, helices α1, α2, and the random loop (residues 47–53) form the N-terminal DNA-binding domain, and helices α4 through α10 form the C-terminal ligand-binding domain.

The N-terminal DNA-binding domain of CmeR exhibits several distinct features compared with the rest of the TetR family members. First, helix α1, consisting of 23 residues, is relatively long among all structurally known TetR regulators. For example, the corresponding helices α1 in QacR (49), TetR (48), and EthR (50) are composed of only 16, 13, and 17 residues, respectively. As mentioned above, the structure of CmeR does not consist of the third N-terminal helix. This is, perhaps, the most striking feature that makes CmeR distinct from the other TetR family members. To date, the CmeR regulator is the only observed case of a random coil replacing helix α3 in a TetR family member. Presumably, the TetR regulators possess a HTH DNA-binding motif formed by helices α2 and α3. Owing to its important role in recognizing target DNA, helix α3 is named the “recognition helix” (19). Thus, we reasoned that the flexible coil might need to transform into a helix when the regulator binds target DNA (Fig. 5c). Since CmeR is a pleiotropic regulator of a large set of genes and is predicted to bind multiple operator sites, with many of those not being of the consensus IR sequence located in the promoter region of cmeABC (72), it could be postulated that the flexibility of the DNA-binding domain permits CmeR to recognize multiple cognate DNA sites. 130

One other unique feature of the CmeR structure is its large center-to-center distance between the two N-termini of the dimer. This center-to-center distance (according to the separation between C α atoms of Y51 and Y51 ′) was measured to be 54 Å (32). The corresponding distances are 39 Å and 35 Å in the apo forms of QacR (49) and TetR (73). These center-to-center distances increase upon ligand binding. For the ligand-bound dimers of QacR (49), TetR (48), EthR (51), and YfiR (74) these distances become 41 Å, 38 Å, 52 Å and 54 Å, respectively. Thus, the relatively large center-to-center distance observed for CmeR reflected the fact that CmeR was liganded (32). Indeed, the crystal structure indicated that a glycerol molecule was bound in each subunit of the CmeR dimer (Fig. 5a) (32).

The C-terminal domain of CmeR consists of helices α4 through α10, with helices α4, α5, α7, α8 and α10 forming an anti-parallel five-helix bundle. In view of the crystal structure, helices α6, α8, α9 and α10 are involved in the formation of the dimer. Dimerization occurs mainly by couplings between pairs of helices ( α6 and α9′, α8 and α10 ′, and their identical counter pairs). A surface area of 1950 Å2 per monomer is buried in the contact region of the dimer (32). The interaction surface is mostly hydrophobic in character. The C-terminal domain of CmeR is distinct in that helix, α9, which is between the two anti-parallel helices α8 and α10, deviates from the direction of α8 by 40°. Thus, helix α9 bends toward the next subunit of the dimer, interacting with α6′ and α7′a to secure interaction between the dimer.

The C-terminal domain forms a large tunnel-like cavity in each subunit of CmeR. This tunnel, surrounded by mostly hydrophobic residues of helices α4–α9, opens horizontally from the front to the back of each protomer. The length of this tunnel is approximately 20 Å. Helices α7 and α8 from one subunit, and α9′ from the other subunit of the regulator make the entrance of the tunnel. Helices α4–α6, however, contribute to form the end of this hydrophobic tunnel. Each hydrophobic tunnel, occupying a volume of about 1000 Å3, spans horizontally across the C- terminal domain and can be seen through from the front to the back of the dimer without obstruction. This unique feature, not found in other structures of the TetR 131

family of regulators, highlights the flexibility of the CmeR regulator (32). As indicated above, the crystal structure of CmeR revealed the presence of a glycerol molecule inside this large ligand-binding tunnel. Glycerol binds identically in each subunit, as indicated by the crystallographic two-fold symmetry of the CmeR dimer (32). This ligand-binding mode is different from that of QacR in which one dimer of QacR binds one drug (49), but similar to that of TetR, which interacts with tetracycline in a manner of 1:1 monomer-to-drug molar ratio (48). The volume of the ligand-binding tunnel of CmeR is large enough to accommodate a few of the ligand molecules. Additional water molecules fill the portion of the large tunnel that is unoccupied by ligand. The structure suggests that CmeR might be able to bind more than one drug molecule at a time, or possibly accommodate a significantly larger ligand that spans across the entire binding tunnel. Indeed, a docking study showed that the hydrophobic tunnel of CmeR should be able to accommodate large, negatively charged bile acid molecules, such as taurocholate and cholate (32). Fig. 5d demonstrates the extensive predicted ligand-binding site, and important residues that are critical for ligand recognition in the tunnel. The bound bile acids are predicted to anchor to several hydrophobic, polar and positively charged residues, including H72, F99, F103, F137, S138, Y139, V163, C166, T167, K170 and H174. These anionic ligands were predicted to span almost the entire length of the ligand- binding tunnel of the regulator, respectively. The flexibility of this large ligand-binding tunnel suggests that CmeR is a multiple ligand binding protein (32).

CONCLUSIONS AND PERSPECTIVES

In the past decade, several crystal structures of the TetR family of regulators have been determined. These structures have allowed us to glance through the phenomenon of multidrug recognition. In particular, the structures of six QacR–drug complexes (48) and QacR simultaneously bound by two drugs (75) revealed the presence of multiple binding sites within an extensive, sizeable drug-binding pocket of the regulator. It is quite conclusive that QacR recognizes a combination of drugs, both positively charged and neutral, by using multiple, proximal and distinct drug 132

binding sites. Based on the structures of AcrR, it is very likely this regulator employs a similar mechanism that QacR uses for drug binding. It is expected that AcrR binds cationic and neutral charged drugs by utilizing aromatic and acidic residues. The predicted drug-binding pocket at the C-terminal domain of AcrR indeed consists of multiple hydrophobic and aromatic residues with a buried glutamate critical for drug binding. A similar drug-binding pocket has been found in BmrR in which multiple aromatic side chains are stacked against the positively charged tetraphenylphosphonium ligand (76). The charge of the bound ligand was further neutralized by negatively charged acidic residue(s) in the ligand-binding site (49, 75- 77).

In the case of CmeR, CmeR tends to bind anionic and uncharged ligands, including the large bile salts such as cholate and deoxycholate. Based on the crystal structure, the mechanisms by which CmeR employs to recognize anionic compounds seem to be an “analog” to those of QacR and AcrR. The C-terminal domain of CmeR forms a large, sizable drug-binding tunnel that occupies a volume of 1000 Å3 (32). This tunnel, possibly consisting of multiple mini-pockets for different ligands, is rich in aromatic residues and contains three positively charged amino acids (two histidines and one lysine). It is very likely that CmeR uses these positively charged residues, in an analogous manner, to recognize negatively charged ligands.

The structure of CmeR indicated that anionic bile acids, such as cholate and deoxycholate, are probably bound in the large hydrophobic tunnel via hydrophobic and aromatic stacking interactions. These bound ligands presumably are further neutralized by one or more of the positively charge residues, including lysine and histidine, in the binding tunnel to secure the binding. This binding mode is quite different from that shown in the MarR-salicylate structure in which the anionic salicylates are not bound in hydrophobic pockets, but in openings that are exposed to solvent (78). Although the binding sites for salicylates possess positively charged arginines to neutralize the formal negative charge of salicylates, they do not contain any aromatic residue. Thus, the crystal structures of the CmeR–ligand complex and 133

other regulatory proteins bound by anionic ligands are needed to provide a clearer understanding on the mechanisms of anionic ligand recognition utilized by these regulators.

CmeR possesses a very unique structural feature at the N-terminal domain, in which it does not have the recognition α3 helix. It is not yet known if this unique feature is related to its function. CmeR acts as a pleiotropic regulator and modulates the expression of many other genes in C. jejuni . As helix α3 is critical for contacting with target DNA, the lack of helix α3 in the structure of CmeR could explain the fact that CmeR binds its IR in a much weaker fashion when compared with the binding of TetR, QacR and AcrR to their target DNAs (19). Detailed interaction between CmeR and target DNA awaits the atomic resolution structure of the CmeR–IR complex.

Knowledge of how these proteins are able to recognize multiple, diverse ligands may prove beneficial to understanding how the MDR transporters, themselves, bind and extrude these deleterious compounds. Structural comparisons from the ligand bound conformations of AcrB (79, 80) with that of QacR (24, 26-29) and the predicted drug bound AcrR (30, 31) conformations reveal similar mechanisms of ligand recognition. Much like AcrR and QacR, AcrB interacts with its various ligands inside a large central cavity utilizing an overlapping but distinct subset of residues for each of the specific ligands to create a multi-faceted binding domain. The major contributors to protein-ligand interactions in AcrB, AcrR, and QacR seem to be aromatic residues providing hydrophobic interactions that secure the drug molecule in the binding cavity. QacR and AcrR require electrostatic interactions from negatively charged amino acids to neutralize the positive charges on the inducing ligands, while neutralization of the positive charges is not necessary for the interaction of these ligands with AcrB. These binding variations may be due to the functional differences of these proteins, in which tight electrostatic interactions may hinder the ability of AcrB to expel the compounds at an adequate rate, but are beneficial to the binding of the transcriptional repressors. Although similar features exist for multi-drug recognition between the regulator and protein pump, multidrug 134

binding is not identical and must be further explored. At this time, it is critical to determine the structure of other multi-drug binding proteins to reveal specific features that are required for ligand recognition and to determine if other TetR family regulators have similar drug binding patterns as the proteins they regulate.

With the structures of AcrR, CmeR, and other multidrug binding proteins, Information may be extracted from the structural data that will aid in the rational design of drugs which are able to inhibit the function of MDR pumps and combat multidrug resistant bacteria. Multidrug export may be blocked in two distinct fashions. First, an anti- MDR drug can be manufactured that directly blocks the pumping mechanism. This direct method will inhibit drug export by either blocking the opening to the central cavity and not allowing the drug to enter, allosterically inhibiting ligand binding by interfering with the binding site, or by holding the protein in single conformation that disrupts drug expulsion. If multidrug binding sites are similar between the regulator and MDR pump then information garnered from the repressor can be utilized in the design of some of these drugs. A second, indirect method of drug design would inhibit multidrug transport by blocking the expression of the MDR pumps. This can be achieved by not allowing derepression of the TetR regulator by either blocking the opening to the binding cavity or by blocking the ligand binding event. The latter would require a drug designed to bind within the cavity that does not induce the conformational change required for operator release. A plausible design to inhibit AcrR-operator is a drug that binds and overlaps both predicted binding sites but features a negatively charged group that blocks the amino acid E67 from transitioning to the binding cavity. Blocking the transition of E67, which is the switch that drives operator release, would allow AcrR to maintain in its state of transcriptional repression and block polymerase binding to inhibit expression of AcrB (30, 31). With the advent of these new drugs, other antibacterial agents that have previously been deemed ineffective may once again become a useful treatment to fight bacterial infections. 135

Acknowledgments

This work was supported by National Institutes of Health Grants GM074027 (to E.W.Y.) and DK063008 (to Q.Z.). M.D.R. is a recipient of a Roy J. Carver Trust predoctoral training fellowship.

136

FIGURES AND TABLES

Figure 1a

137

Figure 1b

138

Figure 1c

139

Figure 1d

Figure d

Fig.1. Structure of AcrR. (a) Crystal structure of AcrR in space group of P2221. Ribbon diagram of the AcrR homodimer generated by crystallographic symmetry. The hydrogen bonds at the interface between the C-terminal and N-terminal domains of each subunit AcrR, shown as dotted lines in the left subunit, are between Glu67 and Arg106; between Gln14 and Arg105; and between Asp18 and Arg105. These H-bonds are absent in the P31 structure. (b) Speculative model of AcrR in its DNA-bound form. The N- and C-terminal domains of the ligand-bound dimeric AcrR were individually aligned with those of the DNA-bound QacR (1JT0) to generate the DNA-bound form of AcrR. The model of the 24 bp IR model is shown as a space- filling model. It is expected that two dimers of AcrR (in orange and deep olive ribbons) bind one double-stranded IR. (c) Electrostatic surface potential of one subunit of AcrR. This view shows the large cavity spanning from the side surface (front) to the subunit interface (back) of the C-terminal domain of one subunit of AcrR. Blue (+15 kBT) and red (−15 kBT) indicate the positively and negatively charged areas, respectively, of the protein. (d) Binding site prediction for AcrR. Residues, W63, E67, I70, F114 and Q130, which are predicted to be important for drug binding are shown as yellow sticks. The figure was prepared using PyMOL. 140

Figure 2

Fig. 2. Structural comparison of the P31 and P2221 structures of AcrR. Superimposition of the dimeric AcrR structures was performed using the program ESCET (green, P31 structure; orange, P2221 structure). Residue E67 in each subunit is shown as a stick model.

141

Figure 3

Figure 3. AcrR-IR2 crystal in a mother solution containing PEG6000, as the precipitant. Resolution extended to approximatel 3.2Å (Fig. 5) although crystal twinning did not allow for proper indexing.

142

Figure 4

3.2Å

Figure 5. Representative diffraction pattern of AcrR-IR2 crystals with resolution reaching 3.2Å. A synchrotron light source was utilized at the Argonne National Laboratory’s Advanced Photon Source Facility for data collection.

143

Figure 5a

144

Figure 5b

145

Figure 5c

146

Figure 5d

Fig. 5. Structure of CmeR. (a) Crystal structure of CmeR. Ribbon diagram of the ligand bound CmeR homodimer generated by crystallographic symmetry. The bound glycerols were shown as space-filling models. (b) Electrostatic surface potential of one subunit of CmeR. This view shows the long tunnel spanning through the C- terminal domain of CmeR. Blue (+15 kBT) and red (−15 kBT) indicate the positively and negatively charged areas, respectively, of the protein. (c) Speculative model of CmeR in its DNA-bound form. The N- and Cterminal domains of the ligand-bound dimeric CmeR were individually aligned with those of the DNA-bound QacR (1JT0) to generate the DNA-bound form of CmeR. The two DNA recognition α3 helices (red) in the dimer of CmeR are included in the model. Each helix of the bound DNA is shown in orange thread. (d) Binding site prediction for CmeR. Residues, H72, F99, F103, F137, S138, Y139, L163, C166, T167, K170 and H174, which are predicted to be important for drug binding are shown in stick models.

147

REFERENCES

1. Davies JE. Origins, acquisition and dissemination of antibiotic resistance determinants. In: Chadwick DJ, Goode J, editors. Antibiotic Resistance: Origins, Evolution, Selection and Spread. Chichester, United Kingdom: John Wiley and Sons Ltd; 1997. pp. 15–35.

2. Grkovic S, Brown MH, Skurray RA. Transcriptional regulation of multidrug efflux pumps in bacteria. Semin. Cell Dev. Biol. 2001;12:225–237.

3. Levy SB. Antibiotic resistance: consequences of inaction. Clin. Infect. Dis. 2001;33:S124–S129.

4. Levy SB. Active efflux mechanisms for antimicrobial resistance. Antimicrob. Agents Chemother. 1992;36:695–703.

5. Saier MH, Paulsen IT, Jr, Sliwinski MK, Pao SS, Skurray RA, Nikaido H. Evolutionary origins of multidrug and drug-specific efflux pumps in bacteria. FASEB J. 1998;12:265–274.

6. McKeegan KS, Borges-Walmsley MI, Walmsley AR. The structure and function of drug pumps: an update. Trends Microbiol. 2003;11:21–29.

7. Ahmed M, Borsch CM, Taylor SS, Vazques-Laslop N, Neyfakh AA. A protein that activates expression of a multidrug efflux transporter upon binding the transporter substrates. J. Biol. Chem. 1994;269:28506–28513.

8. Grkovic, S., Brown, M.H., and Skurray, R.A. (2001) Transcriptional regulation of multidrug efflux pumps in bacteria, Semin. Cell. Dev. Biol. 12 , 225–237.

9. Schumacher, M. A., and Brennan, R. G. (2002) Structural mechanisms of multidrug recognition and regulation by bacterial multidrug transcription factors, Molecul. Microbiol. 45 , 885-893.

10. Martin, R. G., and Rosner, J. L. (2001) The AraC transcriptional activators, Curr. Opin. Microbiol. 4 , 132-137.

11. Parkinson, J.S. and Kofoid, E.C. (1992) Communication modules in bacterial signaling proteins, Annu. Rev. Genet. 26 , 71–112.

12. Parkinson, J.S. (1993) Signal transduction schemes of bacteria, Cell 73 , 857– 871.

13. Hoch, J.A. and Silhavy, T.J. eds (1995) Two-component signal transduction, ASM Press. 148

14. Stock, A.M. (2000) Two-component signal transduction, Annu. Rev. Biochem. 69 , 183–215.

15. Hoch, J.A. (2000) Two-component and phosphorelay signal transduction, Curr. Opin. Microbiol. 3 , 165–170.

16. Inouye, M. and Dutta, R. eds (2003) Histidine kinases in signal transduction, Academic Press.

17. Ventre, L., Filloux, A., Lazdunski, A. (2004) Two-component signal transduction systems: a key to the adaptative potential of Pseudomonal aeruginosa . In: Ramos JL, editor. Pseudomonas. Vol. 2. London, England: Kluwer; pp. 257–288.

18. Beier, D., Gross, R. (2006) Regulation of bacterial virulence by two- component systems, Curr. Opin. Microbiol. 9,143–152.

19. Ramos, J. L., nez-Buenzo, M. M., Molina-Henares, A. J., Tera'n, W., Watanabe, K., Zhang, X., Gallegos, M. T., Brennan, R., and Tobes, R. (2005) The TetR family of transcriptional repressors, Microbiol. Mol. Biol. Rev. 69 , 326–356.

20. Hinrichs, W., Kisker, C., Du¨vel, M., Mu¨ller, A., Tovar, K., Hillen, W., and Saenger. W. (1994) Structure of the Tet repressor-tetracycline complex and regulation of antibiotic resistance, Science 264 , 418–420.

21. Kisker, C., Hinrichs, W., Tovar, K., Hillen, W., and Saenger. W. (1995) The complex formed between Tet repressor and tetracycline-Mg2_ reveals mechanism of antibiotic resistance, J. Mol. Biol. 247 , 260–280.

22. Levy, S. B. (1984) Resistance to tetracyclines p. 191–204. In L. E. Bryan (ed.), Antimicrobial drug resistance, Academic Press, New York, N.Y.

23. Saenger, W., Orth, P., Kisker, C., Hillen, W., and Hinrichs. W. (2000) The tetracycline repressor-A paradigm for a biological switch, Angew. Chem. Int. ed. Eng l. 39, 2042–2052.

24. Schumacher, M.A., Miller, M.C., Grkovic, S., Brown, M.H., Skurray, R.A., and Brennan R.G. (2001) Structural mechanisms of QacR induction and multidrug recognition, Science 294 , 2158–2163.

25. Schumacher, M.A., Miller, M.C., Grkovic, S., Brown, M.H., Skurray, R.A., and Brennan, R.G. (2002) Structural basis for cooperative DNA binding by two dimers of the multidrug binding protein QacR, EMBO J. 21 (2002) 1210–1218 149

26. Schumacher, M. A., Miller, M. C., and Brennan, R. G. (2004) Structural mechanism of the simultaneous binding of two drugs to a multidrug-binding protein, EMBO J. 23 , 2923-30.

27. Murray, D. S., Schumacher, M. A., and Brennan, R. G. (2004) Crystal structures of QacR-diamidine complexes reveal additional multidrug-binding modes and a novel mechanism of drug charge neutralization, J. Biol. Chem. 279 , 14365-71.

28. Grkovic, S., Hardie, K. M., Brown, M. H., and Skurray, R. A. (2003) Interactions of the QacR multidrug-binding protein with structurally diverse ligands: implications for the evolution of the binding pocket, Biochemistry. 42 , 15226-36.

29. Schumacher, M. A., and Brennan, R. G. (2003) Deciphering the molecular basis of multidrug recognition: Crystal structures of the Staphylococcus aureus multidrug binding transcription regulator QacR, Research in Microbiol. 154 , 69-77.

30. Li M, Gu R, Su C-C, Routh MD, Harris KC, Jewell ES, McDermott G, Yu EW. Crystal structure of the transcriptional regulator AcrR from Escherichia coli. J. Mol. Biol. 2007;374:591–603.

31. Gu R, Li M, Su C-C, Long F, Routh MD, Yang F, McDermott G, Yu EW. Conformational change of the AcrR regulator reveals a possible mechanism of induction. Acta Crystallogr. 2008;F64:584–588.

32. Gu R, Su C-C, Shi F, Li M, McDermott G, Zhang Q, Yu EW. Crystal structure of the transcriptional regulator CmeR from Campylobacter jejuni . J. Mol. Biol. 2007;372:583–593.

33. Tseng TT, Gratwick KS, Kollman J, Park D, Nies DH, Goffeau A, Saier MH., Jr The RND permease superfamily: an ancient, ubiquitous and diverse family that includes human disease and development proteins. J. Mol. Microbiol. Biotechnol. 1999;1:107–125.

34. Ma D, Cook DN, Alberti M, Pon NG, Nikaido H, Hearst JE. Genes acrA and acrB encode a stress-induced efflux system of Escherichia coli. Mol. Microbiol. 1995;16:45–55.

35. Nikaido H. Multidrug efflux pumps of gram-negative bacteria. J. Bacteriol. 1996;178:5853–5859. 150

36. Okusu H, Ma D, Nikaido H. AcrAB efflux pump plays a major role in the antibiotic resistance phenotype of Escherichia coli multiple-antibiotic- resistance (Mar) mutants. J. Bacteriol. 1996;178:306–308.

37. Zgurskaya HI, Nikaido H. Cross-linked complex between oligomeric periplasmic lipoprotein AcrA and the inner-membrane-associated multidrug efflux pump AcrB from Escherichia coli . J. Bacteriol. 2000;182:4264–4267.

38. Koronakis V, Sharff A, Koronakis E, Luisi B, Hughes C. Crystal structure of the bacterial membrane protein TolC central to multidrug efflux and protein export. Nature. 2000;405:914–919.

39. Ma D, Alberti M, Lynch C, Nikaido H, Hearst JE. The local repressor AcrR plays a moderating role in the regulation of acrAB genes of Escherichia coli by global stress signals. Mol. Microbiol. 1996;19:101–112.

40. Su C-C, Rutherford DJ, Yu EW. Characterization of the multidrug efflux regulator AcrR from Escherichia coli. Biochem. Biophys. Res. Commun. 2007;361:85–90.

41. Schumacher MA, Miller MC, Grkovic S, Brown MH, Skurray RA, Brennan RG. Structural basis for cooperative DNA binding by two dimers of the multidrug- binding protein QacR. EMBO J. 2002;21:1210–1218.

42. Orth P, Schnappinger D, Hillen W, Saenger W, Hinrichs W. Structural basis of gene regulation by the tetracycline inducible Tet repressor-operator system. Nat. Struct. Biol. 2000;7:215–219.

43. Grkovic S, Brown MH, Schumacher MS, Brennan RG, Skurray RA. The staphylococcal QacR multidrug regulator binds a correctly spaced operator as a pair of dimers. J. Bacteriol. 2001;183:7102–7109.

44. Su C-C, Yu EW. Ligand-transporter interaction in the AcrB multidrug pump determined by fluorescence polarization assay. FEBS Lett. 2007;581:4972– 4976.

45. Grkovic S, Hardie KM, Brown MH, Skurray RA. Interactions of the QacR multidrug-binding protein with structurally diverse ligands: implications for the evolution of the binding pocket. Biochemistry. 2003;42:15226–15236.

46. Vazquez-Laslop N, Markham PN, Neyfakh AA. Mechanism of ligand recognition by BmrR, the multidrug-responding transcriptional regulator: mutational analysis of the ligand-binding site. Biochemistry. 1999;38:16925– 16931.

47. Guazzaroni M-E, Krell T, Felipe A, Ruiz R, Meng C, Zhang X, Gallegos M-T, Ramos JL. The multidrug efflux regulator TtgV recognizes a wide range of 151

structurally different effectors in solution and complexed with target DNA. J. Biol. Chem. 2005;280:20887–20893.

48. Hinrichs W, Kisker C, Duvel M, Muller A, Tovar K, Hillen W, Saenger W. Structure of the Tet repressor-tetracycline complex and regulation of antibiotic resistance. Science. 1994;264:418–420.

49. Schumacher MA, Miller MC, Grkovic S, Brown MH, Skurray RA, Brennan RG. Structural mechanisms of QacR induction and multidrug recognition. Science. 2001;294:2158–2163.

50. Dover LG, Corsino PE, Daniels IR, Cocklin SL, Tatituri V, Besra GS, Futterer K. Crystal structure of the TetR/CamR family repressor Mycobacterium tuberculosis EthR implicated in ethionamide resistance. J. Mol. Biol. 2004;340:1095–1105.

51. Frenois F, Engohang-Ndong J, Locht C, Baulard AR, Villeret V. Structure of EthR in a ligand bound conformation reveals therapeutic perspectives against tuberculosis. Mol. Cell. 2004;16:301–307.

52. Natsume R, Ohnishi Y, Senda T, Horinouchi S. Crystal structure of a γ- butyrolactone autoregulator receptor protein in Streptomyces coelicolor A3(2). J. Mol. Biol. 2003;336:409–419.

53. Willems AR, Tahlan K, Taguchi T, Zhang K, Lee ZZ, Ichinose K, Junop MS, Nodwell JR. Crystal structure of the Streptomyces coelicolor TetR-like protein ActR alone and in complex with actinorhodin or the actinorhodin biosynthetic precusor (S)-DNPA. J. Mol. Biol. 2008;376:1377–1387.

54. De Silva RS, Kovacikova G, Lin W, Taylor RK, Skorupski K, Kull FJ. Crystal structure of the Vibrio cholerae quorum-sensing regulatory protein HapR. J. Bacteriol. 2007;189:5683–5691.

55. Jeng WY, Ko TP, Liu CI, Guo RT, Liu CL, Shr HL, Wang AH. Crystal structure of IcaR, a repressor of the TetR family implicated in biofilm formation in Staphylococcus epidermidis. Nucleic Acids Res. 2008:1567–1577.

56. Webber MA, Piddock LJV. Absence of mutations in marRAB or soxRS in acrB-overexpressing fluoroquinolone-resistant clinical and veterinary isolates of Escherichia coli . Antimicrob. Agents Chemother. 2001;45:1550–1552.

57. Schneider TR. A genetic algorithm for the identification of conformationally invariant regions in protein molecules. Acta Crystallogr. 2002;D58:195–208.

58. Friedman CR, Neimann J, Wegener HC, Tauxe RV. Epidemiology of Campylobacter jejuni infections in the United States and other industrialized 152

nations. In: Nachamkin I, Blaser MJ, editors. Campylobacter. Washington, D.C.: ASM Press; 2000. pp. 121–138.

59. Parkhill J, Wren BW, Mungall K, Ketley JM, Churcher C, Basham D. The genome sequence of the food-borne pathogen Campylobacter jejuni reveals hypervariable sequences. Nature. 2000;403:665–668.

60. Lin J, Akiba M, Zhang Q. Multidrug efflux systems in Campylobacter . In: Ketley JM, Konkel ME, editors. Campylobacter. Norfolk, UK.: Molecular and Cell Biology, Horizon Bioscience; 2005. pp. 205–218.

61. Zhang Q, Plummer PJ. Mechanisms of antibiotic resistance in Campylobacter jejuni . In: Nachamkin I, Szymanski CM, Blaser MJ, editors. Campylobacter. Washington, D.C: ASM Press; 2008. pp. 263–276.

62. Akiba M, Lin J, Barton YW, Zhang QJ. Interaction of CmeABC and CmeDEF in conferring antimicrobial resistance and maintaining cell viability in Campylobacter jejuni . J. Antimicrob. Chemother. 2006;57:52–60.

63. Lin J, Michel LO, Zhang Q. CmeABC functions as a multidrug efflux system in Campylobacter jejuni . Antimicrob. Agents Chemother. 2002;46:2124–2131.

64. Pumbwe L, Piddock LJ. Identification and molecular characterization of CmeB, a Campylobacter jejuni multidrug efflux pump. FEMS Microbiol. Lett. 2002;206:185–189.

65. Cagliero C, Mouline C, Payot S, Cloeckaert A. Involvement of the CmeABC efflux pump in the macrolide resistance of Campylobacter coli . J. Antimicrob. Chemother. 2005;56:948–950.

66. Luo N, Sahin O, Lin J, Michel LO, Zhang Q. In vivo selection of Campylobacter isolates with high levels of fluoroquinolone resistance associated with gyrA mutations and the function of the CmeABC efflux pump. Antimicrob. Agents Chemother. 2003;47:390–394.

67. Lin J, Cagliero C, Guo B, Barton YW, Maurel MC, Payot S, Zhang Q. Bile salts modulate expression of the CmeABC multidrug efflux pump in Campylobacter jejuni . J. Bacteriol. 2005;187:7417–7424.

68. Lin J, Sahin O, Michel LO, Zhang Q. Critical role of multidrug efflux pump CmeABC in bile resistance and in vivo colonization of Campylobacter jejuni . Infect. Immun. 2003;71:4250–4259.

69. Lin J, Akiba M, Sahin O, Zhang Q. CmeR functions as a transcriptional repressor for the multidrug efflux pump CmeABC in Campylobacter jejuni . Antimicrob. Agents Chemother. 2005;49:1067–1075. 153

70. Grkovic S, Brown MH, Skurray RA. Regulation of bacterial drug export systems. Microbiol. Mol. Biol. Rev. 2002;66:671–701.

71. Cagliero C, Maurel M-C, Cloeckaert A, Payot S. Regulation of the expression of the CmeABC efflux pump in Campylobacter jejuni : identification of a point mutation abolishing the binding of the CmeR repressor in an in vitro selected multidrug resistant mutant. FEMS Microbiol. Lett. 2007;267:89–94.

72. Guo B, Wang Y, Shi F, Barton Y-W, Plummer P, Reynolds DL, Nettleton D, Grinnage-Pulley T, Lin J, Zhang Q. CmeR functions as a pleiotropic regulator and is required for optimal colonization of Campylobacter jejuni in vivo. J. Bacteriol. :1879–1890. 190.

73. Orth P, Cordes F, Schnappinger D, Hillen W, Saenger W, Hinrichs W. Conformational changes of the Tet repressor induced by tetracycline trapping. J. Mol. Biol. 1998;279:439–447.

74. Rajan SS, Yang X, Shuvalova L, Collart F, Anderson WF. Crystal structure of YfiR, an unusual TetR/CamR-type putative transcriptional regulator from Bacillus subtilis . Proteins: Struct. Funct. Genet. 2006;65:255–257.

75. Schumacher MA, Miller MC, Brennan RG. Structural mechanism of the simultaneous binding of two drugs to a multidrug-binding protein. EMBO J. 2004;23:2923–2930.

76. Heldwein EE, Brennan RG. Crystal structure of the transcription activator BmrR bound to DNA and a drug. Nature. 2001;409:378–382.

77. Murray DA, Schumacher MA, Brennan RG. Crystal structures of QacR- diamidine complexes reveal additional multidrug-binding modes and a novel mechanism, of drug charge neutralization. J. Biol. Chem. 2004;279:14365– 14371.

78. Alekshun MN, Levy SB, Mealy TR, Seaton BA, Head JS. The crystal structure of MarR, a regulator of multiple antibiotic resistance, at 2.3 Å resolution. Nat. Struct. Biol. 2001;8:710–714.

79. Yu, E. W., McDermott, G., Zgurskaya, H. I., Nikaido, H., and Koshland, D. E., Jr. (2003) Structural basis of multiple drug-binding capacity of the AcrB multidrug efflux pump, Science 300 , 976-980.

80. Murakami, S., Nakashima, R., Yamashita, E., Matsumoto, T., and Yamaguchi, A. (2006) Crystal structures of a multidrug transporter reveal a functionally rotating mechanism, Nature 443 , 173-179.

154

CHAPTER 5. CRYSTAL STRUCTURES OF CmeR-BILE ACID COMPLEXES FROM CAMPYLOBACTER JEJUNI

Mathew D. Routh 1, Zhangqi Shen 2, Chih-Chia Su 3, Qijing Zhang 1,2 , and Edward W. Yu 1,3,4*

1Molecular, Cellular and Developmental Biology Interdepartmental Graduate Program, Iowa State University, IA 50011, USA 2Department of Veterinary Microbiology, College of Veterinary Medicine, Iowa State University, Ames, IA 50011, USA 3Department of Chemistry, Iowa State University, Ames, IA 50011, USA 4Department of Physics and Astronomy, Iowa State University, Ames, IA 50011, USA

ABSTRACT

The TetR family of transcriptional regulators are diverse proteins capable of sensing and responding to various structurally dissimilar antimicrobial agents. Upon detecting these agents, the regulators allow transcription of an appropriate array of resistance markers to counteract the deleterious compounds. Campylobacter jejuni CmeR is a pleiotropic regulator of multiple proteins, including the membrane-bound multidrug efflux transporter CmeABC. CmeR represses the expression of CmeABC and is induced by bile acids, which are also substrates of the CmeABC tripartite pump. The multiligand-binding pocket of CmeR has been shown to be very extensive and consists of several positively charged and multiple aromatic amino acids. Here we describe the crystal structures of CmeR in complexes with the bile acids, taurocholate and cholate. These structures reveal how negatively charged compounds are bound by a TetR family member. Taurocholate and cholate are structurally similar, differing by only the anionic charged group. However, these two bile acids bind distinctly in the binding tunnel. Cholate spans the novel bile acid binding-site adjacent to and without overlapping with the previously determined glycerol-binding site. The negatively charged pentanoate group of cholate is neutralized by a charge-charge interaction. Unlike chloate, taurocholate binds in an anti-parallel orientation and its anionic aminoethanesulfonate moiety interacts with polar side chains to neutralize the formal negative charge. These structures underscore the promiscuity of the multifaceted binding pocket of CmeR. 155

INTRODUCTION The ability of bacteria to adapt and respond to diverse classes of toxic compounds allows these organisms to survive in a variety of harsh environments. Campylobacter jejuni , the leading bacterial cause of food-borne enteritis in humans, is able to flourish in the intestinal mucosa due to its rapid response to bile acid intrusion ( 1, 2 ). This Gram-negative enteric pathogen has become increasingly resistant to common antibacterial agents encountered during the course of an infection. The intrinsic and acquired resistance to these diverse classes of toxic chemicals is facilitated through the expression of multidrug resistant (MDR) efflux transporters. The MDR pumps are capable of effectively lowering the intracellular concentration, thus compromising the effectiveness of the antibacterial compounds. Based on the genomic sequence of C. jejuni NCTC 11168, this organism harbors 13 putative MDR transporters that belong to five different classes, including the ATP- binding cassette (ABC) superfamily, the resistancenodulation- division (RND) family, the multidrug and toxic compound extrusion (MATE) family, the major facilitator (MF) superfamily, and the small multidrug resistance (SMR) family ( 3, 4 ). Currently, only two RND efflux transporters, CmeABC and CmeDEF, in C. jejuni have been functionally characterized ( 5-8). Among these five different families of transporters, members of the RND superfamily exhibit the broadest range of substrate specificity and are usually the primary contributor to the intrinsic multidrug resistance associated with Gram- negative organisms ( 9, 10 ). CmeB, a prototypical RND family transporter, is the major efflux transporter in C. jejuni . This inner membrane efflux pump functions as a tripartite protein complex along with a periplasmic membrane fusion protein, CmeA, and an outer membrane channel, CmeC, to extrude deleterious compounds from the bacterial cell ( 6). The CmeABC complex recognizes and protects C. jejuni from a diverse set of antibacterial compounds, including commonly used antibiotics, metal ions, and lipophilic compounds ( 2, 6-8, 11 ). In addition, CmeABC plays a major role in conferring resistance to bile acids, which are ubiquitously present in the intestinal tract ( 2, 12 ). It has been reported that mutant strains of C. jejuni lacking a functional 156

CmeABC are unable to colonize in the intestinal tract of chickens ( 2). The essential role of CmeABC for the growth of C. jejuni in the intestinal mucosa highlights the importance of this efflux complex to the pathogenicity of the bacterium. The expression of CmeABC is controlled by the transcriptional regulator CmeR, whose open reading frame is located immediately upstream of the cmeABC operon and is transcribed divergently ( 13 ). Transcription of the cmeR gene gives rise to a 210 amino acid protein, which shares N-terminal sequence and structural similarities to members of the TetR family ( 14, 15 ). CmeR is a two-domain protein with an N-terminal DNAbinding motif and a C-terminal multiligand-binding domain. Experimental evidence suggests that the 16 base pair palindromic inverted repeat (IR) sequence, 5 ′TGTAATA AA TATTACA 3′, located between cmeR and cmeABC is the operator site for CmeR binding and transcriptional repression ( 13 ). Bile acids induce the expression of cmeABC by inhibiting the binding of CmeR to this operator (12 ). The crystal structure of CmeR revealed that CmeR exhibits a unique secondary structural feature among all known structures of the TetR family of regulators ( 16 ). To date, CmeR is the only regulator in the TetR family that lacks the N-terminal helix-turnhelix (HTH) DNA-binding motif, in which the recognition helix α3 is replaced by a random coil ( 16, 17 ). Along with this unique characteristic, a large center-to-center distance (54 Å as measured by the separation between C α atoms of Y51 and Y51 ′ from the other subunit) was observed between the two N-termini of the dimer, making it incompatible with the distance between two consecutive major grooves of B-DNA. Each monomer of CmeR consists of nine helices, and numbered with helix α3 being skipped to facilitate comparisons to other member of the TetR family. As a result, the N-terminal domain of CmeR comprises helices α1 and α2 along with this random coil (Fig. 1). The larger C-terminal domain is composed of helices α4-α10, forming a very large hydrophobic tunnel for substrate binding. This tunnel is about 20 Å long with a volume of approximately 1000 Å 3, which is distinctly larger than the binding pockets of many other members of the TetR family. Surprisingly, a fortuitous glycerol molecule was found to bind in the binding tunnel of 157

each monomer ( 16 ). Residues F99, F103, F137, S138, Y139, V163, C166, T167 and K170 are responsible for forming this glycerol binding site. The structure also suggests that CmeR binds glycerol in a manner of 1:1 monomer-to-ligand molar ratio. Although glycerol may not be a natural inducer, the binding may mimic that of natural ligands, such as the C4-dicarboxylates. At this point, the C4-dicarboxylates have not been identified as transcriptional inducers, but Recent evidence suggests CmeR acts as a pleiotropic regulator by modulating the expression of up to 28 genes, including many of those for C4-dicarboxylate transport and utilization (18 ). The volume of the ligand-binding tunnel of CmeR is large enough to accommodate a few of the ligand molecules. Additional water molecules fill the portion of the large tunnel that is unoccupied by ligand. Thus, CmeR might be able to bind more than one drug molecule at a time, or possibly accommodate a significantly larger ligand that spans across the entire binding tunnel. This tunnel, possibly consisting of multiple binding sites for different ligands, is rich in aromatic residues and contains four positively charged amino acids (three histidines and one lysine). Based on the structural information and the fact that bile acids induce transcription of cemABC , we hypothesize that CmeR may utilize these positively charged residues to recognize negatively charged ligands, like bile acids. To elucidate how CmeR recognizes these large anionic ligands, we here report the crystal structures of CmeR in complexes with taurocholate and cholate, respectively.

MATERIALS AND METHODS Preparation and crystallization of the CmeR-ligand complexes: Recombinant CmeR containing a 6 x His tag at the N-terminus was overexpressed in Escherichia coli strain JM109 using the pQE30 vector. The cloning, expression and purification procedures have been described previously ( 16, 19). The purified protein was extensively dialyzed against buffer containing 10 mM Na-phosphate pH 7.2 and 100 mM NaCl and concentrated to 10-15 mg/ml. Prior to crystallization trials, taurocholate or cholate was added to the protein solution at a final concentration of 2 mM and then incubated overnight at 4º C. 158

The CmeR-taurocholate and CmeR-cholate complexes were crystallized at room temperature using hanging-drop vapor diffusion. Briefly, a 4-l drop containing equal volume of protein-ligand solution and reservoir buffer (24% PEG 3350, 0.1 M Tris-HCl pH 8.5, 5% JM 600 pH 7.0, 20 mM sodium acetate) was equilibrated against 500 l of reservoir buffer. Crystals of the CmeR-bile acid complexes appeared within two weeks with typical dimensions of 0.2 x 0.2 x 0.2 mm.

X-ray data collection, processing, and structural refinement: X-ray intensity data were collected at 100 K using beamline-24IDC at the Advanced Photon Source. The crystals were cryoprotected with a solution containing 32% PEG 3350, 0.1 M Tris-HCl pH 8.5, 5% JM 600 pH 7.0, 20 mM sodium acetate. Diffraction data sets were processed with DENZO and scaled with SCALEPACK ( 20 ). Both the

CmeR-taurocholate and CmeR-cholate crystals took the space group of P21212 with unit cell dimensions that were isomorphous to the previously determined CmeRglycerol complex (Table 1). The structures of the CmeR-taurocholate and CmeR-cholate complexes were determined using the PHENIX suite of programs for crystallographic computing ( 21 ). The initial phases were calculated by molecular replacement as implemented in Phaser ( 22) using the previously determined CmeR-glycerol structure (2QCO) with the bound glycerol and water molecules removed as the starting model. Model building was performed using the program Coot ( 23). Refinement of both structures was carried out using CNS ( 24) and PHENIX ( 23). The final model was verified by inspection of the simulated annealing composite omit maps. The simulated annealing electron density omit maps of the bound taurocholate and cholate are shown in Fig. 2. In the final models of both taurocholate and cholate bound structures, 100% of the residues are within either the most favored or additional allowed regions of the Ramachandran plot analysis, as defined by the program PROCHECK ( 25). Polarization: Fluorescence polarization assays were used to determine the drug binding afiinity of CmeR to the conjugated bile acid cholyllysyl fluroscein (CLF) 159

(BD Biosciences). The experiments were done using a ligand binding solution containing 20 mM Phosphate buffered saline (pH 7.2) and 2 M CLF. The CmeR protein solution and 2 M CLF was titrated into the ligand binding solution until the polarization (P) became saturated. In this assay, the protein–drug interaction would reach equilibrium within 1 min. As this is a steady-state approach, fluorescence polarization measurement was taken after incubation for 5 min for each corresponding concentration of the protein and drug to ensure that the binding had reached equilibrium. All measurements were performed at 25 °C using a PerkinElmer LS55 spectrofluorometer equipped with a Hamamatsu R928 photomultiplier. The excitation wavelength was 480 nm. Fluorescence polarization signals (in P) were measured at an emission wavelength of 517 nm, respectively. Each titration point recorded was an average of 15 measurements. Data were

analyzed using the equation, P={( Pbound -Pfree )

[protein]/( KD+[protein])}+ Pfree , where P is the polarization measured at a given total protein concentration, Pfree is the initial polarization of free ligand, Pbound is the maximum polarization of specifically bound ligand, and [protein] is the protein concentration. The titration experiments were repeated for three times to obtained

the average KD value. Curve fitting was accomplished using the program ORIGIN (26 ).

RESULTS To understand how CmeR recognizes anionic ligands, we solved the crystal structure of CmeR in complex with the conjugated and non-conjugated bile acids, taurocholate (Tch) and cholate (Chd), respectively. Crystals of the bile acid bound

complexes belonged to the space group P21212, with the asymmetric unit being occupied by one CmeR molecule. The symmetry operators were used to identify the dimeric state of CmeR. These bile acids were found to bind within the ligand-binding tunnel and interact with the regulator using a surprisingly novel binding site which does not overlap with the previously determined glycerol-binding site. However, the 160

overall conformation of the bile acid bound CmeR structures are very similar to that of the CmeR-glycerol structure.

Structure of the CmeR-Taurocholate complex: The crystal structure of the

CmeR-Tch complex was refined to 2.49 Å resolution with a final R work of 23.5% and

Rfree of 28.9%, revealing that Tch binds within the ligandbinding tunnel in a position adjacent to the previously identified glycerol-binding site (Figure 1). The Tch binding site utilizes a distinct set of amino acids to accommodate the elongated structure of the bile acid, while leaving the glycerol-binding site unoccupied. The four-ring system of the bound Tch is completely buried in the CmeR binding tunnel, leaving its negatively charged 2-aminoethanesulfonate group in the 5β position oriented at the entry point and exposed to the solvent. This four-ring skeleton, mimicking the steroid backbone, consists of three hydroxyl groups located at the 3 α, 7 α and 12 α positions. The CmeR-Tch structure demonstrates that the 3 α- hydroxyl group in the A ring makes a hydrogen bond with the positively charged residue H72 (Fig. 3). The C ring and the 12 α-hydroxyl group of Tch, however, face directly toward helix α8 in the binding tunnel. This orientation facilitates the interaction between the 12-hydroxyl oxygen and K170 of α8, allowing them to form a second hydrogen bond to anchor the bound Tch. Interestingly, the 7 α-hydroxyl group of the B ring does not have a hydrogenbonded partner. With the closest polar residue not within 5 Å, this hydroxyl group seems to be left behind in the binding tunnel without making dipole-dipole interaction with any residues that form the wall of this binding site. Perhaps the most striking feature for Tch binding is found at the other end of the molecule which harbors the anionic, conjugated ethanesulfonate tail. Tch is bound in such a way that the long 2-aminoethanesulfonate moiety at the 5 β position is extended slightly out of the binding tunnel exposed to the solvent, while still in close enough proximity to interact with residues forming the entrance of the tunnel. Within 5 Å of this negatively charged sulfonate group, there are no positively charged histidines, lysines or arginines available to neutralize the formal negative 161

charge of this sulfonate moiety. Instead, this conjugated acidic tail is engaged to interact with the side chain hydroxyl oxygen atom of residue Y137 and the side chain carbamoyl nitrogen of residue Q134, thus forming two additional hydrogen bonds with these residues. The repressor protein further anchors this bound bile acid molecule through a water-mediated hydrogen bond between T167 and carbonyl oxygen of the 2-aminoethanesulfonate group of Tch tosecure the binding. Surprisingly, the large molecule of Tch does not occupy the entire volume of the tunnel. The four-ring backbone and the 2-aminoethanesulfonate tail of the bound Tch are not linear in shape, but rather curve upward and result in a concave conformation. Thus, the end-to-end length of the molecule is significantly shorter than it was expected and only reaches 16.1 Å. In doing so, CmeR is able to accommodate and create a novel bile acid-binding pocket for Tch. This new binding pocket is distinct from the previously determined glycerol-binding site. Thus, the bound Tch only spans this bile acid-binding site and leaves the glycerol-binding site unoccupied. As the binding tunnel of CmeR is mostly hydrophobic in nature, the bound Tch is also found to make extensive hydrophobic contacts with residues forming the wall of this tunnel. It is observed that at least nine hydrophobic amino acids, including four aromatic residues (F103, W129, F137 and Y139), that line the inside wall of the tunnel are involved in Tch binding (Table 2).

Structure of the CmeR-Cholate complex: The crystal structure of the

CmeR-Chd complex was refined to a resolution of 2.57 Å, resulting in R work and R free of 21.5% and 27.3 %, respectively. This structure revealed that the binding mode for Chd, which differs from Tch by its 5 β-cholanoate group, is very distinct (Figure 2b). Surprisingly, the bound Chd was found to orient in an opposite direction when compared with Tch. Thus, the bound Chd and Tch are anti-parallel to each other. For Chd binding, Chd is completely buried within the hydrophobic tunnel in a way that its non-conjugated 5 β-cholanoate tail is inserted into the end of the tunnel, leaving its four-ring steroid backbone anchored closer to the entrance. In this manner, the anionic pentanoate moiety of Chd directly interacts with the positively 162

charged H174 of helix α9, forming a hydrogen bond (Fig. 4). Unlike Tch, in which its anionic ethanesulfoate group is stabilized by charge-dipole interaction, the structure of CmeRChd suggests that the negatively charged end of Chd is neutralized by this positively charged histidine residue. A second interaction between CmeR and the pentanoate group of the bound Chd is established through the backbone carbonyl oxygen of A108 from helix α6, forming another hydrogen bond to anchor this bile acid. Important interactions were also found at the 3 α and 12 α-hydroxyl groups of the four-ring system. These two hydroxyl groups contribute two hydrogen bonds with C166 and K170 of helix α8, respectively, to stabilize the steroid backbone. Like Tch, the 7 α-hydroxyl group of Chd does not form a hydrogen bond with any amino acid. The bound Chd molecule is significantly curved upward and exhibits a boat-like conformation. As a result, the endto- end length of the molecule is only 12.5 Å. The curved structure of Chd also makes interactions with nine different hydrophobic amino acids, including four aromatic residues (F103, W129, F137 and Y139) that create the wall of the tunnel (Table 2). Overall, Chd and Tch share the same ligand-binding pocket. These bile acids do not span the entire tunnel, but rather bend into a concave structure. In this conformation, these bile acids occupy a novel distinct binding site that is not overlapped with the previously determined glycerol-binding site (Fig. 5). Based on the structures of CmeR-bile acid complexes, it is observed that the glycerol-binding site in the tunnel remains unoccupied upon bile acid binding. Instead, several solvent molecules are found in this glycerol site. Thus, it is very likely that the large ligand-binding tunnel of CmeR could accommodate a bile acid and a glycerol molecule simultaneously.

DISCUSSION With the rising incidences of multidrug resistant strains of bacteria, it has become increasingly important to understand how individual proteins are able to recognize such diverse substrates. The crystal structures of the QacR multidrug binding protein in complex with its respective ligands have provided many insights 163

into the mechanism of multidrug binding ( 27, 28), but these reports have primarily involved positively charged compounds. The CmeR-bile acid complexes reveal how a TetR family protein specifically interacts with anionic ligands, whereby a primarily hydrophobic tunnel with appropriately spaced polar and charged residues is used to stabilize bile acid binding. Prior to this structural data, the crystal structure of MarR-salicylate has provided evidence on how regulatory proteins recognize anionic compounds ( 29). The negatively charged salicylate binds to MarR within a solvent exposed crevice, rather than a large pocket, and interacts with arginines to neutralize its formal charge. The binding crevice lacks the familiar aromatic residues that are critically important in other multidrug binding proteins (15, 17, 27). It is intriguing that the multidrug binding protein TtgR seems to utilize a different mechanism to recognize negatively charged antibiotics and plant antimicrobials ( 30 ). The hydrophobic environment is provided in the ligand binding pocket at the C-terminal regulatory domain. In addition, a positively charged histidine and a polar asparagine are also found to involve in the binding. For CmeR, this regulator seems to share a similar mechanism with TtgR to recognize negatively charged bile acids. Within the multifaceted binding tunnel there are at least seven aromatic residues, five phenylalanines, one tyrosine and one tryptophan, lining the hydrophobic surface to accommodate staking interactions with the ligands. In addition, charge neutralization is observed in bile acid binding, whereby the positively charged histidine interacts with the negatively charged cholanoate group of Chd. In fact, within the binding tunnel of CmeR, we have found four positively charged residues, including H72, K170, H174 and H175′ (Table 2). These residues, which underscore the diversity of the CmeR binding tunnel, probably function to neutralize charges and accommodate the binding of anionic and neutral ligands. This phenomenon is clearly demonstrated in the structures of CmeR-Tch and CmeR-Chd, in which the negatively charged ligands are secured in the binding tunnel by residue K170. Surprisingly, the two elongated bile acids did not bind in the same orientation inside the tunnel of CmeR, but were actually bound anti- 164

parallel to each other. Chd was bound in an orientation where its A ring was located close to the tunnel opening. However, the bound Tch molecule displayed a contrasting orientation, whereby the corresponding A ring was buried deeply inside the far end of the tunnel. Because of the difference in orientation, the conserved four-ring systems of Tch and Chd were found to bind in different environments. Intriguingly, only one hydrogen bond is conserved in both structures. Residue K170 forms an important hydrogen bond with the 12 α-hydroxyl group in the C ring of both Tch and Chd to anchor the steroid backbone and create an overlapping site with very dissimilar binding modes for these molecules. In the case of Chd, CmeR further anchors this ligand by using the positively charged H72 residue. For Tch binding, the regulator chooses H174 at the other side of the tunnel to interact with the ligand. Tch and Chd are similar in chemical structure and have identical charge. The different binding modes of these two bile acids indeed highlight the promiscuity of the multiligand-binding tunnel of CmeR. To demonstrate that purified C-6XHis CmeR indeed interacts and tightly binds to bile acids, we performed fluorescence polarization assays with the fluroscein conjugated bile acid CLF. Using this method, we were able to determine the binding affinity of the ligand. The observed dissociation constant of 50.2 ± 5.8 µM appears similar to those kD values obtained for AcrR to its inducing ligands proflavin, rhodamine 6G, and ethidium bromide (31). The approximate five-fold reduction in affinity of Cmer-CLF compared to AcrR-ligand is most likely do to the nature of the ligand, whereby the large fluroscein moiety may interfere with the molecule entering the cavity. It is worth noting that the high number of hydrogen bonds formed between CmeR and bile acids (compared to AcrR) may be required to provide the desolvation energy to initiate ligand binding. Previously, the crystal structure of CmeR was fortuitously resolved in complex with a glycerol molecule ( 16 ). This structure suggested that at least two distinct binding sites existed within the tunnel. Indeed, one of these predicted binding sites was occupied by the bound glycerol. Interestingly, the CmeR-bile acid structures indicated that the large molecules of Chd and Tch did not span both predicted 165

binding sites, but instead took the distinct second site and left the glycerol-binding site unoccupied. On the basis of these crystal structures, it is possible that CmeR can accommodate a bile acid and a glycerol molecule at the same time (Fig. 5). Indeed, positive electron density was observed in the glycerol binding site. Since glycerol was purposely not used in crystallization trials, we could not unambiguously identify the density. If this is indeed a natural ligand, it is fascinating to note the ability of CmeR to recognize such a large bile acid molecule and simultaneously bind a second natural inducer. Such a phenomenon has been previously observed with the crystal structure of QacR simultaneously bound to two ligands, proflavin and ethidium ( 28), and has been predicted through biochemical analysis to occur in many other proteins, including AcrR ( 31, 32), TtgV ( 32) and MdfA ( 34). The plasticity and promiscuity of the multiligand-binding tunnel of CmeR were further underscored by these CmeR-ligand complex structures. As mentioned previously, glycerol and bile acids have distinct binding sites within the tunnel. In the glycerol-bound structure, the bile acid binding site was unoccupied and filled with water molecules. This empty site was surrounded with several aromatic residues, including F61, F103, F111, W129, F137 and Y139. When Tch occupied this bile acid site, which was observed from the CmeR-Tch structure, all of these aromatic residues shifted outward and seemingly participated to expand the internal size of this binding site, probably accommodating the large size of the bile acid. It is worth noting that the formal negative charge of Tch was not neutralized by positively charged residues. Instead, electrostatic neutralization was achieved by interactions between the anionic Tch and the positive dipoles of two side chains, Y139 and Q134. Thus, charge-charge electrostatic interaction is not essential for binding negatively charged ligands. Similar drug-regulator interaction has been found in QacR, in which the QacR regulator neutralized one end of the positively charged pentamidine by using carbonyl and side chain oxygen atoms ( 35). Interestingly, the bound Chd rather employed another mechanism to neutralize its formal negative charge, whereas the anionic pentanoate group was compensated by the formal positive charge of H72. 166

CmeR is able to create a binding pocket that can discern between bile acids and other ringed compounds by employing helices α4, α5, α6, α7, α8, and α9 to create a curved hydrophobic binding pocket with appropriately placed charged/polar residues that are able to associate with the unique structure and amphipathic properties of the bile acids. Similar binding architectures have previously been observed for bile acids binding to the nuclear receptor FXR (36) and the binding inhibition observed in the interaction to pancreatic phospholipase A2 (37). Our structures indicate at least two possible orientations of bile acids whereby the sulfonic group of TCH is facing the exterior and the 3-α-hydroxyl of the A ring is facing this direction in CHD. The disparate binding of modes of the molecules is most likely associated with the increased lenth of TCH as the steroid backbone is identical in both molecules with hydroxyl groups at positions 3 α, 7 α, and 12 α for each. Interestingly, only the 3 α and 12 α hydroxyl groups make hydrogen bond contacts with CmeR while the hydroxyl group at 7 α does not make any contacts. Presumably the 7-α-dehydroxylated derivatives of CHD and TCH, deoxycholate and taurodeoxycholate, would ligand to CmeR in an identical fashion. Whether an analogous binding mechanism is observed for the 12-α-dehydroxylated derivatives, chenodeoxycholate and taurochenodeoxycholate, or if they utilize an entirely different orientation to form hydrogen bond contacts is an intriguing question. In summary, the ability of CmeR to bind two very similar bile acids in quite distinct manners highlighted the plasticity and promiscuity of the ligand-binding tunnel of this regulator. This plasticity is very likely applicable to other multiligand binding proteins, including the AcrR multidrug regulator. Further, neutralization of the negatively charged bile acids can take place using the proximal positively charge residues or the nearby polar groups. The proximal and distinct bile acid and glycerol- binding sites of CmeR highlights the capacity of this regulator, whereby the sizeable hydrophobic tunnel indeed consists of multiple mini-pockets to accommodate diverse ligands.

167

ACCESSION NUMBERS Coordinates and structural factors have been deposited in the Protein Data Bank with accession numbers 3HGY (CmeR-taurocholate) and 3HGG (CmeR-cholate).

ACKNOWLEDGEMENTS This work was supported by NIH Grants DK063008 (to Q.Z.) and GM074027 (to E.W.Y.). M.D.R. was a recipent of a Roy J. Carver Trust predoctoral training fellowship. This work is based upon research conducted at the Northeastern Collaborative Access Team beamlines of the Advanced Photon Source, supported by award RR- 15301 from the National Center for Research Resources at the National Institutes of Health. Use of the Advanced Photon Source is supported by the U.S. Department of Energy, Office of Basic Energy Sciences, under Contract No. DE-AC02-06CH11357.

168

FIGURES AND TABLES

Table 1

169

Table 2

170

Figure 1

Figure 1. Structure of a CmeR-ligand complex. Ribbon diagram of the taurocholate- bound CmeR homodimer generated by crystallographic symmetry. TCH interacts within an ~20 Å long tunnel that resides in the CmeR interior. The interaction is stabilized through polar and hydrophobic interactions. The taurocholate is shown as a stick model (cyan, carbon; blue, nitrogen; red, oxygen).

171

Figure 2

172

Figure 2. Electron density maps of the bile acid binding pocket. A, Stereoview of the simulated annealing electron density omit map of the bound taurocholate contoured at 1.5 σ. The electron density map is shown as grey mesh. The structure of taurocholate is shown as a stick model (cyan, carbon; blue, nitrogen; red, oxygen). The surrounding secondary structural elements are shown as slate ribbons. B, Stereoview of the simulated annealing electron density omit map of the bound cholate contoured at 1.5σ. Theelectron density map is shown as grey mesh. The structure of cholate is shown as a stick model (green, carbon; blue, nitrogen; red, oxygen). The surrounding CmeR secondary structural elements are shown as slate ribbons. The electron density omit maps were calculated with a starting temperature of 2,000 K and by excluded the bound ligands from the models.

173

Figure 3

174

Figure 3. The taurocholate binding site. A, Amino acid residues within 4.2 Å from the bound taurocholate (cyan, carbon; blue, nitrogen; red, oxygen; orange, sulfur). The side chains of selected residues are shown as slate sticks (slate, carbon; blue, nitrogen; red, oxygen). Residues from the next subunit of CmeR are shown as magenta sticks (magenta, carbon; blue, nitrogen; red, oxygen). A water molecule (OW) hydrogenbonded with the bound taurocholate is shown as red sphere. B, Schematic representation of the CmeR and taurocholate interactions shown in panel A. Dotted lines indicate hydrogen bonds.

175

Figure 4

176

Figure 4. The cholate binding site. A, Amino acid residues within 4.2 Å from the bound cholate (green, carbon; blue, nitrogen; red, oxygen). The side chains of selected residues are shown as slate sticks (slate, carbon; blue, nitrogen; red, oxygen). Residues from the next subunit of CmeR are shown as magenta sticks (magenta, carbon; blue, nitrogen; red, oxygen). B, Schematic representation of the CmeR and cholate interactions shown in panel A. Dotted lines indicate hydrogen bonds.

177

Figure 5

Figure 5. The bile acid and glycerol-binding sites of CmeR. This is a composite figure showing the locations of the bound ligands in the ligand binding tunnel. The ligands shown in stick models are taurocholate (cyan), cholate (green) and glycerol (pink).

178

Figure 6

Figure 6A

0.25

0.20

0.15 Data: Data1_B Model: Hill Equation: y=Vmax*x^n/(k^n+x^n) Weighting: 0.10 y No weighting

Y Axis Title Y Chi^2/DoF = 0.00004 FP R^2 = 0.99444 0.05 Vmax 0.32879 ±0.01607 k 50.25936 ±5.77434 n 1 ±0 0.00

-20 0 20 40 60 80 100 120 140 CmeRX Axis [µM] Title

Figure 6B

Figure 6. A) Representative fluorescence polarization of CmeR with CLF. Statistical analysis revealed a KD of 50.3 ± 5.8 M. B) schematic representation of the CLF molecule used in binding studies. Cholate moiety is to the left while the fluorescein moiety os on the right. 179

REFERENCES

1. Friedman, C. R., Neimann, J., Wegener, H. C., and Tauxe, R. V. (2000) Epidemiology of Campylobacter jejuni infections in the United States and other industrialized nations. Campylobacter , edited by I. Nachamkin & M. J. Blaser, pp. 121-138, Washington, D. C., ASM Press.

2. Lin, J., Sahin, O., Michel, L. O., and Zhang, Q. (2003) Critical role of multidrug efflux pump CmeABC in bile resistance and in vivo colonization of Campylobacter jejuni . Infect. Immun. 71 , 4250–4259.

3. Parkhill, J., Wren, B. W., Mungall, K., Ketley, J. M., Churcher, C., Basham, D., Chillingworth, T., Davies, R. M., Feltwell, T., Holrovd, S., Jagels, K., Karlvshev, A. V., Moule, S., Pallen, M. J., Penn, C. W., Quail, M. A., Raiandream, M. A., Rutherford, K. M., van Vliet, A. H., Whitehead, S., and Barrell, B. G. (2000) The genome sequence of the food-borne pathogen Campylobacter jejuni reveals hypervariable sequences. Nature 403 , 665–668.

4. Lin, J., Akiba, M., and Zhang, Q. (2005) Multidrug efflux systems in Campylobacter . Campylobacter: Molecular and Cell Biology, edited by J. M. Ketley & M. E. Konkel, pp. 205-218. Norfolk, U.K.: Horizon Bioscience.

5. Akiba, M., Lin, J., Barton, Y. W., and Zhang, Q. J. (2006) Interaction of CmeABC and CmeDEF in conferring antimicrobial resistance and maintaining cell viability in Campylobacter jejuni . J. Antimicrob. Chemother. 57 , 52-60.

6. Lin, J., Michel, L. O., and Zhang, Q. (2002) CmeABC functions as a multidrug efflux system in Campylobacter jejuni . Antimicrob. Agents Chemother. 46 , 2124-2131.

7. Pumbwe, L., and Piddock, L. J. (2002) Identification and molecular characterization of CmeB, a Campylobacter jejuni multidrug efflux pump. FEMS Microbiol. Lett. 206 , 185-189.

8. Cagliero, C., Mouline, C., Payot, S., and Cloeckaert, A. (2005) Involvement of the CmeABC efflux pump in the macrolide resistance of Campylobacter coli . J. Antimicrob. Chemother. 56 , 948-950.

9. Tseng, T. T., Gratwick, K. S., Kollman, J., Park, D., Nies, D. H., Goffeau, A., and Saier, M. H., Jr. (1999) The RND permease superfamily: an ancient, ubiquitous and diverse family that includes human disease and development proteins. J. Mol. Microbiol. Biotechnol. 1 ,107–125.

10. Schweizer, H. P. (2003) Efflux as a mechanism of resistance to antimicrobials in Pseudomonas aeruginosa and related bacteria: unanswered questions. Genet. Mol. Res. 2 , 48–62. 180

11. Luo, N., Sahin. O., Lin, J., Michel, L. O., and Zhang, Q. (2003) In vivo selection of Campylobacter isolates with high levels of fluoroquinolone resistance associated with gyrA mutations and the function of the CmeABC efflux pump. Antimicrob. Agents Chemother. 47 , 390-394.

12. Lin, J., Cagliero, C., Guo, B., Barton, Y. W., Maurel, M. C., Payot, S., and Zhang, Q. (2005) Bile salts modulate expression of the CmeABC multidrug efflux pump in Campylobacter jejuni . J. Bacteriol. 187 , 7417-7424.

13. Lin, J., Akiba, M., Sahin, O., and Zhang, Q. (2005) CmeR functions as a transcriptional repressor for the multidrug efflux pump CmeABC in Campylobacter jejuni . Antimicrob. Agents Chemother. 49 , 1067-1075.

14. Grkovic, S., Brown, M. H., and Skurray, R. A. (2002) Regulation of bacterial drug export systems. Microbiol. Mol. Biol. Rev. 66 , 671-701.

15. Ramos, J. L., Martinez-Bueno, M., Molina-Henares, A. J., Teran, W., Watanabe, K., Zhang, X. D., Gallegos, M. T., Brennan, R., and Tobes, R. (2005) The TetR family of transcriptional repressors. Microbiol. Mol. Biol. Rev. 69 , 326-356.

16. Gu, R., Su, C.-C., Shi, F., Li, M., McDermott, G., Zhang, Q., and Yu, E. W. (2007) Crystal Structure of the transcriptional regulator CmeR from Campylobacter jejuni . J. Mol. Biol. 372 , 583–593.

17. Routh, M. D., Su, C.-C., Zhang, Q., and Yu, E. W. (2009) Structures of CmeR and AcrR: Insight into the mechanisms of transcriptional repression and multidrug recognition in the TetR family of regulators. Biochim. Biophys. Acta 1794 , 844-851.

18. Guo B, Wang Y, Shi F, Barton Y-W, Plummer P, Reynolds DL, Nettleton D, Grinnage-Pulley T, Lin J, Zhang Q. CmeR functions as a pleiotropic regulator and is required for optimal colonization of Campylobacter jejuni in vivo. J. Bacteriol. :1879–1890. 190.

19. Su, C.-C., Shi, F., Gu, R., Li, M., McDermott, G., Yu, E. W., and Zhang, Q. (2007) Preliminary structural studies of the transcriptional regulator CmeR from Camplobacter jejuni . Acta Crystallog. Sect. F , 63 , 34-36.

20. Otwinowski, Z., and Minor, M. (1997) Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 276 , 307-326.

21. Afonine, P.V., Grosse-Kunstleve, R.W., and Adams, P.D. (2005) The Phenix refinement framework. CCP4 Newsletter 42 , contribution 8. 181

22. McCoy, A. J., Grosse-Kunstleve, R. W., Adams, P. D., Winn, M. D., Storoni, L. C., and Read, R. J. (2007) Phaser crystallographic software. J. Appl. Crystallog. 40 , 658-674.

23. Emsley, P., and Cowtan, K. (2004) Coot: Model-building tools for molecular graphics. Acta Crystallog. Sect. D , 60 , 2126-2132.

24. Brünger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse- Kunstleve, R. W., Jiang, J. S., Kuszewski, J., Nilges, M., Pannu, N. S., Read, R. J., Rice, L. M., Simonson, T., and Warren, G. L. (1998) Crystallography & NMR system: A new software suite for macromolecular structure determination. Acta Crystallog. Sect. D , 54 , 905-921.

25. Laskowski, R. A., MacArthur, M. W., Moss, D. S., and Thornton, J. M. (1993) PROCHECK: a program to check the stereochemical quality of protein structures. J. Appl. Crystallog. 26 , 283-291.

26. ORIGIN Ver. 7.5. OriginLab Corporation, Northampton, MA, USA.

27. Schumacher, M. A., Miller, M. C., Grkovic, S., Brown, M. H., R. A. Skurray, R. A., and Brennan, R. G. (2001) Structural mechanisms of QacR induction and multidrug recognition. Science 294 , 2158-2163.

28. Schumacher, M. A., Miller, M. C., and Brennan, R. G . (2004) Structural basis for cooperative DNA binding by two dimers of the multidrug-binding protein QacR . EMBO J. 23 , 2923-2930.

29. Alekshun, M. N., Levy, S. B., Mealy, T. R., Seaton, B. A., and Head, J. S. (2001) The crystal structure of MarR, a regulator of multiple antibiotic resistance, at 2.3 Å resolution. Nature Struct. Biol. 8 , 710-714.

30. Alguel, Y., Meng, C., Terán, W., Krell, T., Ramos, J. L., Gallegos, M.-T., and Zhang, X. (2007) Crystal structures of multidrug binding protein TtgR in complex with antibiotics and plant antimicrobials. J. Mol. Biol. 369 , 829-840.

31. Su, C.-C. Rutherford, D. J., and Yu, E. W. (2007) Characterization of the multidrug efflux regulator AcrR from Escherichia coli . Biochem. Biophys. Res. Commun. 361 , 85–90.

32. Li, M., Gu, R., Su, C.-C., Routh, M. D., Harris, K. C., Jewell, E. S., McDermott, G., and Yu, E. W. (2007) Crystal structure of the transcriptional regulator AcrR from Escherichia coli . J. Mol. Biol. 374 , 591–603.

33. Guazzaroni, M.-E., Krell, T., Felipe, A., Ruiz, R., Meng, C., Zhang, X., Gallegos, M.-T., and Ramos, J. L. (2005) The multidrug efflux regulator TtgV recognizes a wide range of structurally different effectors in solution and complexed with target DNA. J. Biol. Chem. 280 , 20887-20893. 182

34. Lewinson, O., and Bibi, E. (2001) Evidence for simultaneous binding of dissimilar substrates by the Escherichia coli mulitdrug transporter MdfA. Biochemistry 40 , 12612-12618.

35. Murray, D. S., Schumacher, M. A., and Brennan, R. G. (2004) Crystal structures of QacR-diamidine complexes reveal additional multidrug-binding modes and a novel mechanism, of drug charge neutralization. J. Biol. Chem. 279 , 14365-14371.

36. Mi, L. Z., Devarakonda, S., Harp, J. M., Han, Q., Pellicciari, R., Willson, T. M., Khorasanizadeh, S., and Rastinejad, F. (2003) Structural basis for bile acid binding and activation of the nuclear receptor FXR , Mol. Cell 11 , 1093-1100.

37. Pan, Y. H., and Bahnson, B. J. (2007) Structural basis for bile salt inhibition of pancreatic phospholipase A2, J. Mol. Biol. 369 , 439-450.

183

CHAPTER 6: CONCLUDING REMARKS

RND TRANSPORTERS AND AcrD

Although much progress has been made to understand the mechanism of facilitated drug export by RND transporters, only minimal information exists for AcrD and the transport of aminoglycosides and anthracylines. Numerous structural models are available for the RND transporters AcrB and MexB, but these proteins do not readily recognize hydrophilic ligands such as aminoglycosides. Therefore, it is important to develop a model for transport of hydrophilic ligands, compounds which do not readily cross the cellular membrane. With an aim to develop novel antibiotics against these pumps, it is important to understand the nature of ligand recognition in the transporters along with the proteins that regulate their expression. The structures we have modeled for AcrD and previous structures for AcrB indicate that drug binding is localized primarily within the central binding cavity or a periplasmic binding cleft located 15 Å above the membrane plane (tunnel 2) (1-7). It is hypothesized that the binding pocket in the binding cleft is more relevant based on previous experiments using domain swapping of AcrD and AcrB (8). It was shown in homology studies that, as expected, the binding cleft region of AcrD is distinctly more polar, allowing interation with the hydrophilic character of aminoglycosides. Alternatively, AcrB exhibits a hydrophobic surface in the same area (1-5, 7). Importantly, either instance indicates that substrate binding occurs through overlapping but distinct sites to discriminate between dissimilar compounds. Furthermore, the recognition sites of multidrug binding proteins are composed of hydrophobic and aromatic residues that recognize the specific shape of inducing ligands. In some instances, polar residues are recruited to stabilize ligand-binding through hydrogen bonds, which seems to be the case in aminoglycoside binding by AcrD. Although the hydrophilic aminoglycosides are quite distinct from the hydrophobic AcrB substrates, it is intrigueing that these ligands bind with strikingly similar affinities. Through fluorescent polarization (FP) experiments it was observed that AcrB binds Ethidium Bromide (Et), Proflavin (Pf), and Rhodamine-6-G (R6G) 184

with dissociation constants of 8.7, 14.5, and 5.5 µM (9), respectively, while AcrD, as displayed through ITC and FP experiments, binds gentamicin and daunorubicin-HCl with dissociation constants of 3 and 6 µM (Chapter 3). Presumably, transport in this binding range inhibits the antibacterial compound from reaching toxic levels. Furthermore, if binding affinities were increased into the nM scale, it could be surmised that export of important metabolites would be a deleterious side effect. As continued exploration of RND proteins has shed new light and insights into the structure and function of these proteins, the functional rotating mechanism has emerged as the leading model (3-6). Recent structures of AcrB and MexB have pointed to an asymmetric model that seemingly contradicted previous symmetric structures (1, 2, 7). Some evidence indicated that the symmetrical AcrB structures represent a resting state of the trimer. It is also likely that the observed symmetry is just an artifact of the crystallization process, whereby, as suggested by Seeger et al. 2006 (4), merohedral twinning was present in the R32 spacegroup. These crystals were in fact twinned R3 crystals with noncrystallographic symmetry. As a result, seemingly suitable models were observed with higher than expected R-factors. A high twinning fraction results in a blurred electron density, which cannot be computationally deconvoluted, and explains the difficulties encountered with the trigonal crystal form. Of note is the recent crystallization of CusA by our group, which crystallized in a symmetric conformation (10). The heavy-metal transporter CusA crystallized in the R32 spacegroup in both the liganded and unliganed state. Although CusA represents a different subfamily of RND transporters, a distinct functional mechanism could be suggested which highlights the need to obtain additional structural information of RND proteins in hopes of fully understanding the transport mechanism. Another interesting case exists, whereby MdtBC, an RND transporter in the same subfamily as AcrB and MexB (HAE1), functions as a heterotrimer. For MdtBC, the only functional state exists as a B2C trimer (11). This poses a question as to whether this trimer can function through the same bi-site activation method, in which ligand interaction to the neighboring protomer is used to initiate 185

conformational changes. The AcrB model assumes that each monomer plays an equal role in the three-stage cycle. Additionally, why does this protein only function in this B2C trimeric state? It is intriguing that inactivation of MdtC, by mutating the proton-relay-network, did not alter substrate transport, which indicates that it may play a unique role from that of MdtB or even AcrD monomers. As it stands, the asymmetric model provides the only reasonable transport mechanism. The model suggested a functionally rotating mechanism to describe transport, whereby each protomer of the trimer cycles through successive states. The mechanism is said to resemble Boyer’s binding change mechanism of the F1Fo ATPase, where conformational cycling of α and β subunits through the various states, loose, tight and open, leads to the synthesis of ATP (12). Intriguingly, the F1Fo ATPase utilizes a rotational mechanism in which proton translocation in the Fo portion drives an internal rotation of the single-copy γ-subunit of F1, which causes the asymmetry and drives the sequential conformational changes. The counterpart RND systems do not have a comparable γ-subunit to drive the asymmetry; therefore it is unknown how this asymmetry occurs. Indeed, the γ-rod of the ATPase is tilted towards one of the three α/β subunits. Likewise, in one monomer of the AcrB trimer the pore α-helix is tilted towards the neighbouring monomer's PN2 subdomain (3-5). Presumably, ligand interaction can trigger the asymmetric transport mechanism. Recently, an AcrB structure revealed a novel binding partner with an unknown function (13). A transmembrane protein YajC co-crystallized with AcrB and this complex induced a functionally significant rotation of the porter domain of AcrB relative to the transmembrane domain which could drive the asymmetry of AcrB. Deletion of YajC only slightly reduced ampicillin resistance, which suggests this protein is not critical for drug transport. The most intriguing issues of the efflux mechanism arrise while discussing topics encompasing the entire tripartite complex. The flexible multidomain membrane fusion protein AcrA seems to be critical to drug efflux (14). How this protein initiates AcrD facilitated aminoglycoside efflux is an unexplored question. Is AcrA required to interact with ligands or is it critical for initiating conformational 186

changes to AcrD? Furthermore, molecular dynamics experiments have suggested that membrane fusion proteins may initiate a twisting motion in the outer membrane channel in addition to the role in activating the RND transporter (15). Presumably, once AcrA is bound to AcrD, TolC is recruited and the twisting motion initiated By AcrA opens the TolC channel. The twisting motion may also lead to a peristaltic action aiding in the movement of substrates along TolC. As yet, these are untested ideas. To unravel these enticing questions, new and ingenious experiments using cross-linking, NMR, and mutational studies must be explored on AcrD, AcrB and other RND transporters.

TetR REGULATORS, AcrR AND CmeR Through the work on AcrR and CmeR, a model can be suggested to describe transcriptional regulation and multidrug binding by proteins of the TetR family. The studies indicate that AcrR utilizes a random coil-to-helix transition, which is initiated by a glutamic acid residue 67 (16-18). The coil-to-helix transition shortens helix α4 resulting in separation of the N-terminal domains. Separation of the N-terminal domains is a critical factor in repressor release. This mechanism contrasts what is seen in the repressor TetR, whereby contacts are made between tetracycline-Mg 2+ and a loop region formed between helices α6 and α7 (19, 20). The movements of α6, due to interaction with tetracycline-Mg 2+ , and subsequently α4 act as a pendulum to separate helix α3 and α3´ by 3Å in the ligand-bound conformation. These conformational changes force the release of TetR from the operator. It is important to note that, as revealed by the QacR, TetR, and AcrR crytstal structures, proteins that are homologous in structure and function may utilizize slightly different mechanisms to perform the same task, i.e., transcriptional repression (21). Multidrug recognition is a critical factor in MDR proteins. The transporters and regulators must be able to discern between deleterious compounds and beneficial metaboloites. Based on our crystallization studies on CmeR, We provide evidence that the overall shape and size of the binding pocket along with appropriately spaced polar or charged residues is a vital feature to discriminate 187

between molecules (22, 23). The most intriguing feature of the CmeR crystal structures is the overall size and shape of the CmeR binding pocket, which seems to perfectly accommodate the ~20 Å-long bile acids. With the charged residues located on the more C-terminal half of the pocket and aromatic residues on the other, the pocket seems to be engineered precisely to bind the aliphatic bile acid compounds. Prior to the CmeR-bile acid structure, AcrB-deoxycholate crystals have provided the only evidence to how multidrug binding proteins recognize these ligands (24). The structure suggests deoxycholate is stabilized within AcrB through 4 van der Waals contacts with Phe665 and a hydrogen bond to a Ser715 with no charge neutralizing residues in the vicinity of the carboxyl group. This contrasts CmeR, which does utilize charge neutralization to associate with negatively charged bile acids. Furthermore, CmeR also employs aromatic residues to form additional hydrophobic interactions. This suggests that CmeR may be more adapted to specifically interact with these anionic ligands. Indeed, within the multifaceted binding pocket there are at least four aromatic residues (three phenylalanines and one tryptophan), three positively charge His and one positively charged Lys residue that function in neutralizing formal charges. A question emerges as to whether other multidrug binding proteins recognize anions through aromatic stacking and charge neutralization, or if CmeR has optimized its binding for these bile acids? As a multidrug binding protein, CmeR binds the same series of bile acids as CmeABC, the transporter it regulates. CmeABC is a multidrug binding protein capable of recognizing and extruding many structurally dissimilar antibiotics, in addition to bile acids (25-30). Interestingly, antibiotics have not been shown to initiate CmeABC expression or bind CmeR. Initially, this seems counterintuitive, but during the time when C. jejuni is being inundated with antibiotics, expression of CmeABC would have already initiated due to the presence of bile acids in the colonized intestinal tract. This places selective pressure on CmeABC to recognize these molecules but places minimal emphasis on CmeR to do the same. Therefore, CmeABC has evolved to recognize antibiotics while there is little advantage if CmeR were to recognize the same drugs. 188

How structurally different molecules, including glycerol, TCH, and CHD initiate the same conformational changes in CmeR remains an unresolved question. Currently, only the ligand-bound crystal structures have been solved for this TetR family regulator. Homology models indicate that ligand binding within the binding pocket initiates a conformational shift throughout the molecule resulting in separation of the N-terminal domains to the 54Å center-to-center distance identified in the crystal structures. Whether a coil-to-helix transition initiates these changes as it does with QacR and AcrR by elongating helices 5 and 4, respectively, or if a completely distinct mechanism exists is unknown. The crystal structures of apo- CmeR and the CmeR-DNA complex are needed to further resolve these important questions.

189

REFERENCES

1. Murakami, S., Nakashima, R., Yamashita, E., and Yamaguchi, A. (2002) Crystal structure of bacterial multidrug efflux transporter AcrB, Nature 419 , 587-593.

2. Yu, E. W., McDermott, G., Zgurskaya, H. I., Nikaido, H., and Koshland, D. E., Jr. (2003) Structural basis of multiple drug-binding capacity of the AcrB multidrug efflux pump, Science 300 , 976-980.

3. Murakami, S., Nakashima, R., Yamashita, E., Matsumoto, T., and Yamaguchi, A. (2006) Crystal structures of a multidrug transporter reveal a functionally rotating mechanism, Nature 443 , 173-179.

4. Seeger, M. A., Schiefner, A., Eicher, T., Verrey, F., Diederichs, K., and Pos, K. M. (2006) Structural asymmetry of AcrB trimer suggests a peristaltic pump mechanism, Science 313 , 1295-1298.

5. Sennhauser, G., Amstutz, P., Briand, C., Storchenegger, O., and Grutter, M. G. (2007) Drug export pathway of multidrug exporter AcrB revealed by DARPin inhibitors, PLoS Biol. 5, e7.

6. Sennhauser, G., Bukowska, M. A., Briand, C., and Grütter, M. G. (2009) Crystal structure of the multidrug exporter MexB from Pseudomonas aeruginosa , J. Mol. Biol. 389, 134-45.

7. Yu, E. W., Aires, J. R., McDermott, G., and Nikaido, H. (2005) A periplasmic drug-binding site of the AcrB multidrug efflux pump: a crystallographic and site-directed mutagenesis study, J. Bacteriol. 187 , 6804-6815.

8. Elkins, C. A., and Nikaido, H. (2002) Substrate specificity of the RND-type multidrug efflux pumps AcrB and AcrD of Escherichia coli is determined predominantly by two large periplasmic loops, J. Bacteriol. 184 , 6490-6498.

9. Su, C.-C., Yu, E. W. (2007) Ligand-transporter interaction in the AcrB multidrug pump determined by fluorescence polarization assay, FEBS Lett. 581 , 4972–4976.

10. Long, F., Su, C.-C., Routh, M. D., Wang, W., Rajashankar, K. R., and Yu, E. W. (2010) Crystal structures of the CusA heavy-metal efflux pump suggests a methionine mediated mechanism, Nature (submitted to Nature)

11. Kim, H. S., Nagore, D., and Nikaido, H. (2010) Multidrug efflux pump MdtBC of Escherichia coli is active only as a B2C heterotrimer, J. Bacteriol. 192 , 1377-86. 190

12. Boyer, P. D. (1997) The ATP synthase--a splendid molecular machine, Annu. Rev. Biochem. 66 , 717-49.

13. Törnroth-Horsefield, S., Gourdon, P., Horsefield, R., Brive, L., Yamamoto, N., Mori, H., Snijder, A., and Neutze, R. (2007) Crystal structure of AcrB in complex with a single transmembrane subunit reveals another twist, Structure 15 , 1663-73.

14. Aires, J. R., and Nikaido, H. (2005) Aminoglycosides are captured from both periplasm and cytoplasm by the AcrD multidrug efflux transporter of Escherichia coli , J. Bacteriol. 187 , 1923-1929.

15. Vaccaro, L., Koronakis, V., and Sansom, M. S. (2006) Flexibility in a drug transport accessory protein: molecular dynamics simulations of MexA, Biophys. J. 91 , 558-64.

16. Routh, M. D., Su, C.-C., Zhang, Q., and Yu, E. W. (2009) Structures of CmeR and AcrR: Insight into the mechanisms of transcriptional repression and multidrug recognition in the TetR family of regulators. Biochim. Biophys. Acta 1794 , 844-851.

17. Li, M., Gu, R., Su, C.-C., Routh, M. D., Harris, K. C., Jewell, E. S., McDermott, G., and Yu, E. W. (2007) Crystal structure of the transcriptional regulator AcrR from Escherichia coli . J. Mol. Biol. 374 , 591–603

18. Gu R, Li M, Su C-C, Long F, Routh MD, Yang F, McDermott G, Yu EW. Conformational change of the AcrR regulator reveals a possible mechanism of induction. Acta Crystallogr. 2008;F64:584–588.

19. Orth, P., Saenger, W., and Hinrichs, W. (1999) Tetracycline-chelated Mg2+ ion initiates helix unwinding in Tet repressor induction, Biochemistry 38 , 191- 8.

20. Orth, P., Schnappinger, D., Hillen, W., Saenger, W., and Hinrichs, W. (2000) Structural basis of gene regulation by the tetracycline inducible Tet repressor- operator system, Nat. Struct. Biol. 7 , 215-9.

21. Ramos, J. L., nez-Buenzo, M. M., Molina-Henares, A. J., Tera'n, W., Watanabe, K., Zhang, X., Gallegos, M. T., Brennan, R., and Tobes, R. (2005) The TetR family of transcriptional repressors, Microbiol. Mol. Biol. Rev. 69 , 326–356.

22. Gu, R., Su, C.-C., Shi, F., Li, M., McDermott, G., Zhang, Q., and Yu, E. W. (2007) Crystal Structure of the transcriptional regulator CmeR from Campylobacter jejuni . J. Mol. Biol. 372 , 583–593. 191

23. Crystal structures of CmeR-bile acid complexes from Campylobacter jejuni Mathew D. Routh1, Zhangqi Shen2, Chih-Chia Su3, Qijing Zhang1,2, and Edward W. Yu1,3,4*

24. Drew, D., Klepsch, M. M., Newstead, S., Flaig, R., De Gier, J. W., Iwata, S., and Beis, K. (2008) The structure of the efflux pump AcrB in complex with bile acid, Mol. Membr. Biol. 25 , 677-82.

25. Lin, J., Sahin, O., Michel, L. O., and Zhang, Q. (2003) Critical role of multidrug efflux pump CmeABC in bile resistance and in vivo colonization of Campylobacter jejuni . Infect. Immun. 71 , 4250–4259.

26. Lin, J., Akiba, M., and Zhang, Q. (2005) Multidrug efflux systems in Campylobacter . Campylobacter: Molecular and Cell Biology, edited by J. M. Ketley & M. E. Konkel, pp. 205-218. Norfolk, U.K.: Horizon Bioscience.

27. Akiba, M., Lin, J., Barton, Y. W., and Zhang, Q. J. (2006) Interaction of CmeABC and CmeDEF in conferring antimicrobial resistance and maintaining cell viability in Campylobacter jejuni . J. Antimicrob. Chemother. 57 , 52-60.

28. Lin, J., Michel, L. O., and Zhang, Q. (2002) CmeABC functions as a multidrug efflux system in Campylobacter jejuni . Antimicrob. Agents Chemother. 46 , 2124-2131.

29. Pumbwe, L., and Piddock, L. J. (2002) Identification and molecular characterization of CmeB, a Campylobacter jejuni multidrug efflux pump. FEMS Microbiol. Lett. 206 , 185-189.

30. Cagliero, C., Mouline, C., Payot, S., and Cloeckaert, A. (2005) Involvement of the CmeABC efflux pump in the macrolide resistance of Campylobacter coli . J. Antimicrob. Chemother. 56 , 948-950.