UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Molecular studies of the structure and function of pseudomonas aeruginosa OprD: an imipenem specific… Huang, Hongjin 1994

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1995-059804.pdf [ 3.6MB ]
JSON: 831-1.0088172.json
JSON-LD: 831-1.0088172-ld.json
RDF/XML (Pretty): 831-1.0088172-rdf.xml
RDF/JSON: 831-1.0088172-rdf.json
Turtle: 831-1.0088172-turtle.txt
N-Triples: 831-1.0088172-rdf-ntriples.txt
Original Record: 831-1.0088172-source.json
Full Text

Full Text

MOLECULAR STUDIES OF THE STRUCTURE AND FUNCTION OFPseudomonas aeruginosa OprD: AN IMIPENEM SPECIFIC PORINbyHONGeTIN HUANGB.Sc., Sichuan University, Sichuan, China, 1986M.Sc., Academy of Miiitary Methcai. Sciences, Beijing, China, 1989A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIES(Department of Microbiology and Immunology)We accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIAMay, 1995© Hongjin Huang, 1995In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)Department of_________________The University of British ColumbiaVancouver, CanadaDateDE-6 (2188)ABSTRACTPseudomonas aeruginosa OprD is a specific porin which facilitates theuptake of basic amino acids and imipenem, a carbapenem antibiotic with highpotency against P. aeruginosa. To permit further studies of OprD, the oprDstructural gene was cloned and expressed in Escherichia coli on a 2.1-kbBamHIIKpnI fragment. DNA sequencing predicted a 420 amino acid mature OprDprotein with a 23 amino acid signal peptide. In addition, a putative oprD regulatorygene opdE was sequenced, which predicted a hydrophobic protein of 402 aminoacids.A set ofF. aeruginosa isogenic strains with genetically defined levels of OprDwere constructed and utilized to characterize the in vivo function of OprD. Theresults clearly demonstrated that OprD could be utilized by imipenem andmeropenem but, even when substantially overexpressed, could not be signfficantlyutilized by other p-lactams, quinolones or aminoglycosides. Regarding its functionin uptake of nutrients, OprD selectively facilitated the diffusion of basic aminoacids and gluconate under growth-rate limiting conditions. Competitionexperiments confirmed that imipenem shared common binding sites with basicamino acids in the OprD channel, but not with gluconate or glucose. In vitrofunctional studies using purified OprD provided direct evidence for the presence ofa specific binding site(s) for imipenem in the OprD channel, with an I value of 1.4.tM.An OprD topology model was proposed based on sequence alignment with E.Ucoli porin OmpF and structure predictions. Sixteen n-strands were predicted,connected by short turns at the periplasmic side, whereas the eight external loopswere of variable length but tended to be much longer. In addition, multiplesequence a]ignments between OprD and seven representatives from the porinsuperfamily indicated that OprD was the first specific porin that could be alignedwith members of the so-called porin superfamily. PCR-based site directedmutagenesis was performed to separately delete short stretches (4-8) of amino acidresidues from each of the predicted external loops. Six out of eight mutantsexpressed in both E. coli and P. aeruginosa, maintained substantial resistance totrypsin treatment in the context of outer membranes, and formed functionalchannels, which supported the general accuracy of the model. The loop 2 deletionmutant only partially reconstituted supersusceptibility to imipenem in an OprDdefective background, and showed much lower affinity to imipenem in themacroscopic conductance inhibition experiment, indicating its involvement ininiipenem binding. Deletions in loops 5, 7 or 8 resulted in a channel with enhancedpermeability to antibiotics, but which retained the imipenem binding site(s).A model of the channel architecture of OprD was constructed based on thesedata, and the mechanism by which imipenem and basic amino acids pass throughthe OprD channel was discussed.mTABLE OF CONTENTSABSTRACTTABLE OF CONTENTSLIST OF FIGURESLIST OF TABLESLIST OF ABBREVIATIONSACKNOWLEDGMENTSDEDICATIONINTRODUCTIONA. Pseudoinonas aeruginosaB. Pseudomonas aeruginosa Outer Membrane1. Outer Membrane Structure2. Porins: General Porins and Specific Porins .3. Antibiotic Uptake Across the Outer Membrane4. Imipenem5. OprD: A Specific Porin for Basic Amino Acids andC. Structure Analysis of Porin Proteins1. X-ray Crystallography and Others2. Crystal Structure of General Porin Channels.3. Crystal Structure of Specific Porin Channels4. Prediction of Porin Structures: Porin SuperfamilyD. Model Membrane Studies of PorinsE. Aims of This StudyMATERIALS AND METHODSA. Strains, Plasmids and Growth ConditionsB. Genetic Manipulations1. General DNA Techniques2. DNA Fragment Isolationivvmxxixmxiv11226810Imipenem . 15191920232526283030303035iv3. DNA Sequencing. 354. Transfer of DNA into P. aeruginosa 365. Oligonucleotide Synthesis and Purification 37C. Cloning Strategy for the oprD Gene 38D. Allele Replacement Mutagenesis 391. Construction of pXH1 392. Selection of the oprD::Km’1. Mutant 39E. Overexpression of the oprD Gene in P. aeruginosa 41F. Electrophoresis 41G. Growth Experiments 41H. Purification of OprD 43I. Immunological Techniques 451. Production and Purification of OprD Antibodies 452. Western-Immunoblotting 46J. Prediction of the Topology Model of OprD 461. Sequence Alignment 462. Structural Characteristics 47K. PCR-Based Site Directed Deletion Mutagenesis 481. Two PCR Strategies 482. Construction of pMBK19R and pMBE19R 503. Polymerase Chain Reaction 52L. Whole Cell Lysate 52M. Trypsinization Studies 56N. Black Lipid Bilayer Analysis 561. Single Channel Conductance 562. Macroscopic Conductance Experiment 573. Zero Current Membrane Potential 580. Assays 581. Protein Assay 582. Nitrocefin Assay 59V3. iVilnimum Inhibitory Concentration (MIC) Determination . - 59RESULTS 61CHAPTER ONE Analysis of Two Gene Regions Involved in theExpression of OprDA. Molecular Cloning of the oprD GeneB. Overexpression of the oprD Gene in E. coli CE1248C. Nucleotide Sequencing of the oprD GeneD. Nucleotide Sequencing of the opdE GeneE. Summary• . . 61• . . . 6161• . . . 64• . . . 6871CHAPTER TWO Functional Characterization of OprD: In vivo andIn vitro 73A. IntroductionB. Construction of a Defined OprD-Defective Mutant H729C. Overexpression of the oprD Gene in P. aeruginosaD. Function of OprD in Antibiotic UptakeE. Function of OprD in Sugar TransportF. Competition ExperimentsG. Purification of OprDH. Black Lipid Bilayer AnalysisI. SummaryCHAPTER THREE Structural Characterization of OprD:Topology ModelA. IntroductionB. Prediction of an OprD Topology ModelC. PCR-Based Site Directed Deletion MutagenesisD. Characterization of the Deletion Mutants737475788084889194Membrane969696101104viE. Trypsin Susceptibility of the Deletion VariantsF. Revised OprD ModelG. SummaryCHAPTER FOUR Structure/Function Relationships: FunctionalAlterations of Deletion MutantsA. IntroductionB. Effects of Deletion on Imipenem/Meropenem Susceptibilities.C. Effects of Deletion on Other Antibiotic SusceptibilitiesD. Effects of Deletion on Sugar TransportE. Purification of the Mutant OprDsF. Effects of Deletion on the Physical Properties of the ChannelG. SummaryDISCUSSIONA. Function of OprD m Antibiotic UptakeB. Function of OprD in Nutrient Uptake: Is Lysine the BestSubstrate9C. Prediction of an OprD Membrane Topology Model .D. Molecular Architecture of the OprD ChannelE. The Journey of Imipenem Through the OprD Channel.F. General Porins and Specific Porins131131134137141146148REFERENCES 151APPENDIX 165A. Revised OprD Model by Multiple Alignments and AmphipathicityCalculations 165109• 112113115• 115• • 115118121• . 121• . 124130vuLIST OF FIGURESFigure 1. Schematic representation of the outer membrane andpeptidoglycan ofF. aeruginosa 3Figure 2. Structures of thienamycin and imipenem 11Figure 3. Structures of some antibiotics and amino acids that penetratethrough the OprD channel 18Figure 4. Diagram of plasmid pXH1 utilized for allele replacementmutagenesis 40Figure 5. Diagram of plasmid pXH2 utilized for the overexpression ofthe oprD gene 42Figure 6. Schematic diagram showing two PCR strategies 49Figure 7. Diagram of plasmid pMBK19R 51Figure 8. Diagram of plasmid pMBE19R 53Figure 9. Restriction endonuclease maps of F. aeruginosa PAO 1 strainH 103 chromosomal DNA derived by Southern hybridization. 62Figure 10. Diagram of plasmid pBK19R 63Figure 11. Overexpression of the oprD gene in E. coli 65Figure 12. Nucleotide and the deduced amino acid sequence of theoprD gene 66Figure 13. Nucleotide and the deduced amino acid sequence of theopclE gene 69Figure 14. Construction of an OprD defective mutant in F. aeruginosa . 76Figure 15. SDS-PAGE demonstrating overexpression and mutagenesisof the oprD gene 77Figure 16. Function of OprD in sugar transport 83Figure 17. Competition between L-lysine and imipenem for theOprD channel 87Figure 18. SDS-PAGE of samples from solubilization stages 89Figure 19. Purification of OprD 90viiiFigure 20. Chart recording of step wise increase of the membranecurrent formed by 1% oxidized cholesterol in n-decanein the presence of purified proteins 92Figure 21. Sequence alignment between OprD and OmpF 98Figure 22. Membrane topology model of OprD 99Figure 23. Expression of OprD derivatives in the outer membraneof E. coli 106Figure 24. Expression of OprD derivatives in the outer membrane ofP. aeruginosa OprD-defective strain 11729 108Figure 25. Western-immunoblot demonstrating expression of OprDderivatives in P. aeruginosa whole cell lysates 110Figure 26. Trypsinization studies of OprD derivatives 111Figure 27. Competition between L-lysine and imipenem for OprDAL2 . . 119Figure 28. Competition between L-lysine and imipenem for OprDAL5/7/8 122Figure 29. Comparison of heat-modiflabilities between the purifiednative OprD and mutant OprDs 123Figure 30. Comparison of single channel conductance between nativeand mutant OprDs 125Figure 31. Macroscopic conductance inhibition experiments 129Figure 32. Schematic representation of the predicted interiorarchitecture of the OprD channel 142Figure 33. Multiple sequence alignments between OprD andrepresentatives of porin superfamily 168Figure 34. Revised membrane topology model of OprD 171ixLIST OF TABLESTable I. Strains 31Table II. Plasmids 32Table III. PCR-based site specific deletion mutagenesis 54Table IV. Influence of OprD expression levels on antibioticsusceptibility ofF. aeruginosa strains 79Table V. Influence of OprD expression levels on antibioticsusceptibility of E. coli strains 81Table VI. Effects of basic amino acids and gluconate on imipenemsusceptibilities of P. aeruginosa strains expressingdifferent levels of OprD 86Table VII. Deletion mutagenesis of the predicted loops 105Table VIII. Effects of deletions on imipenem/meropenemsusceptibilities ofF. aeruginosa strains 117Table IX. Effects of deletions on other antibiotic susceptibilities ofP. aeruginosa strains 120Table X. Average single-channel conductance of the native andmutant OprD pores in different salt solutions 127Table XI. Zero-current membrane potentials 128Table XII. Homologies and identities among the porin superfamily. . . . 167xLIST OF ABBREVIATIONSAp ampicillinbp base pairCb carbenicillinCFP cefpiromeCli chioramphenicolCIP ciprofloxacinCTX cefotaximeDEAE diethylaminoethylEDTA ethylenecliamine tetraacetateFLER fleroxacinFPLC fast protein liquid chromatographyGM gentamicinIMIP imipenemIPTG isopropylthiogalactosideLBNS Luria broth, normal saltLDAO N, N, thmethyldodecylamine-N-oxideLPS lipop olysaccharidekb kilobase pairkD kilodaltonKDO 2-keto-3-deoxyoctulosonic acidKm kanamycinxMERO meropenemMOPS 3-N-morpholino propane sulfonic acidoctyl-POE octyl-polyoxyethyleneOD optical densityPAGE polyacrylamide gel electrophoresisPBPs penicillin binding proteinsPCR polymerase chain reactionPMSF phenyl methyl sulfonyl fluoridePst phosphate specific transportrpm revolutions per minuteSDS sodium lauryl sulfateStr streptomycinTPCK N-tosyl-L-phenylalanine chloromethyl ketonexliACKNOWLEDGMENTSI am most grateful to my supervisor, Bob Hancock, for his excellentsupervising, constant support and kindness.I would like to thank Richard Siehnel for his technical assistance andmembers of my supervisory committee for their guidance and helpful advices.I am also deeply grateful to my friends, especially Kathy, Manhong and Huafor their encouragement, friendship and patience during the last several years.The financial support of Canadian Cystic Fibrosis Foundation is gratefullyacknowledged.xiuDEDICATIONTo my beloved Mom and Dad.xivINTRODUCTIONA. Pseudomonas aeruginosa.Pseudomonas aeruginosa is an opportunistic human pathogen that causesa variety of infections, usually in immunocompromised hosts such as burn victimsand cancer patients, or children with cystic fibrosis (Schimpif et at, 1970). P.aeruginosa causes 10-15% of all nosocomial infections, making it second only to E.coli as the most frequently hospital-acquired pathogen (Young, 1984). This Gramnegative rod also makes several different toxins, some of which may cause shock,while others kill tissue cells or hydrolyse structural tissue proteins such as elastin(Liu, 1974). Given its multifactorial virulence, it is not surprising that P.aeruginosa is able to cause a wide variety of diseases such as bacteremia, urinarytract infections, endocarditis and gastrointestinal infections (Pollack, 1990). P.aeruginosa is becoming a major clinical problem since it has a high, naturalresistance to many commonly used antibiotics, including first and secondgeneration penicilhins and cephalosporins, tetracycline, chioramphenicol andvancomycin (Bryan, 1979). It has been shown that the permeability of the P.aeruginosa outer membrane to p-lactam antibiotics and also some other simpleorganic compounds is from twelve (Nicas and Hancock, 1983a) to one hundred fold(Yoshimura and Nikaido, 1982) lower than that of the permeability of E. coli outermembrane to the same or similar compounds and clearly this lower permeabilityof the outer membrane layer plays a major role in the intrinsic antibiotic resistance1of this organism (Nikaido and Hancock, 1986).B. Pseudomonas aeruginosa Outer Membrane1. Outer Membrane Structure.The cell envelope of P. aeruginosa consists of two membranes separated bya layer of peptidoglycan and a cellular compartment called the periplasm. Theinnermost, cytoplasmic (inner) membrane is a typical phospholipid bilayermembrane which is studded with a wide variety of polypeptides. The majorfunctions of cytoplasmic membrane proteins are in cellular energization, transportof nutrients and export of toxic byproducts (Cronan et al., 1987). Peptidoglycan islocated underneath the outer membrane and is the major determinant of cell shapeand osmotic stability (Oliver, 1987). Thus, the periplasm is primarily locatedbetween the peptidoglycan and the cytoplasmic membrane. It functions in theprocessing and traffic of molecules entering or leaving the cell (Oliver, 1987).The outer membrane is biologically unusual in that, unlike the cytoplasmicmembrane, it is an asymmetric bilayer (Fig. 1), in which the inner monolayer iscomposed of phospholipid, whereas the outer monolayer contains the unique lipidspecies lipopolysaccharide (LPS) (Lugtenberg and van Alphen, 1983; Nikaido andNakae, 1979). The basic LPS consists of three regions: (a) the hydrophobic,biologically-active endotoxin lipid A, (b) the rough core and (c) the 0-antigen region,which is immunodominant (Rietschel et al., 1984). The lipid A region is typical in2c ++Figure 1: Schematic representation of the outer membrane and peptidoglycanof P. aeruginosa.3that a single backbone structure corresponding to glucosaminyl- p-(16)glucosamine is substituted with six or seven saturated or hydroxyl fatty acidresidues (Karunaratne et at, 1992). This region is antigenically and chemicallyconserved. The rough core is covalently bound to lipid A. It contains 11heterogeneous sugar residues, including an unique octose (2-keto-3-deoxyoctulosonic acid [KDO]) as well as glucose, rhamnose and galactosamineresidues. In addition, this fraction contains phosphate and alanine. Analysissuggested that there are 1l’—16 phosphate residues per chain ofF. aeruginosa LPScore oligosaccharide, which is much higher than that of Enterobacteriaceae, suchas Salmonella minnesota, where only 1 or 2 phosphate residues were present perchain (Drewry et al., 1975; Kropinski et at, 1979). The rough core may be cappedby repeating tn- to pentasaccharide units termed the 0-antigen. It has been shownthat the 0-antigen portion of P. aeruginosa often contains such sugars as glucose,rhamnose, glucosamine, fucosamine and quinavosamine (Kropinski et at, 1985).This latter repeating saccharide portion is one of the most immunogenic antigensof smooth Gram-negative bacteria and determines the 0-serotype of such bacteria(for reviews, see Nikaido and Hancock, 1986; Hancock et al., 1994).The asymmetric distribution and chemical characteristics of LPS give theouter membrane many of its unique barrier properties. As mentioned above, thepresence of a large amount of phosphate in the core region of P. aeruginosa LPSresults in the strong surface negative charge (Sherbert and Lakshmi, 1973). LPSis anchored in the outer membrane in part, by the fatty acyl chains of its Lipid A4portion (Morrison, 1985). In addition, the non-covalent cross-bridging of adjacentLPS molecules with clivalent cations (IVIg or Ca2) (Rottem and Leive, 1977), andthe hydrophobic interactions between the outer membrane proteins and Lipid A(Nikaido and Vaara, 1985), also contribute to stabilize LPS in the outer membrane.The combination of surface negative charge and divalent cation cross-bridging ofLPS makes P. aeruginosa and other Gram-negative bacteria resistant tohydrophobic antibiotics, bile salts, detergents, proteases, lipases and lysozyme(Nikaido and Vaara, 1985).The P. aeruginosa outer membrane also contains a few species of “major”proteins. These include the murein lipoproteins, the multifunctional protein OprFand porins (Fig. 1). Two lipoproteins have been identified in P. aeruginosa, OprIand OprL, both of them are inserted in the inner phospholipid monolayer and arenon-covalently associated with peptidoglycan (Mizuno, 1979; Hancock et al.,198 la). Therefore they are structural proteins that stabilize the architecture of theouter membrane-peptidoglycan complex by seating the outer membrane onto thesurface of the peptidoglycan. Multifunctional protein OprF is also strongly but noncovalently associated with peptidoglycan and plays an important role in outermembrane stabilization and cell shape determination (Gotoh et al., 1989; Woodruffand Hancock, 1989). Porins are a group of proteins forming trans-outer-membrane,water-fi]led channels. In general, porins have monomer molecular weights in therange of 28 kD to 48 kD, are present in membrane as oligomers (usually trimers),are often strongly but non-covalently associated with the underlying peptidoglycan5and with LPS, and have a high content of a-sheet structure. In P. aeruginosa,OprB, OprC, OprD, OprE, OprF, OprP and OprO have been identified as porins (forreview, see Hancock et al., 1990).2. Porins: General Porins and Specific Porins.Porins are generally divided into two classes: non-specific (general) porinsand specific porins (Nikaido and Vaara, 1985). General porins form water-filledchannels that permit the passive diffusion of hydrophilic molecules below a certainsize, and thus are responsible for the non-specific exclusion limit of the outermembrane. Specific porins also produce water-filled channels, which containstereospecific substrate-binding sites (hancock, 1987). The diffusion of the specificsubstrate is accelerated when the solute concentration is low, but it is slowed downwhen the concentration is high, producing saturation-type kinetics.General porins take up molecules based on size, electrical charge andhydrophilicity (Nikaido and Vaara, 1985). Though it used to be a controversialissue, OprF is a major non-specific porin in P. aeruginosa. Its pore-forming propertywas conlirmed both in model membrane systems (Benz and Hancock, 1981) and inintact cells (Beffido et al., 1992). The channel diameter was estimated to be 20 A,about twice the width of E. coli porin channels, and can allow the passage ofsaccharides with molecular weights of approximately 3,000 (Nikaido and Hancock,1986). However, only 400 out of 200,000 OprF molecules per cell are proposed to6form such large channels; the rest appear to form small channels that are predictedto be antibiotic impermeable (Woodruff et al., 1986). Besides OprF, minor outermembrane proteins OprC and OprE are also general porins with small channel size(Yoshihara and Nakae, 1989). The above cited data explains the low outermembrane permeability of P. aeruginosa compared to E. coli (Angus et al., 1982;Yoshimura and Nikaido, 1982; Nicas and Hancock, 1983a). This, in turn, wasproposed to be the major basis for the high intrinsic resistance of P. aeruginosa tohydrophilic antibiotics (Nikaido and Hancock, 1986).To overcome the low permeability and to permit the effective uptake ofessential nutrients available at low concentrations in the medium, several specificporins are present in the P. aeruginosa outer membrane. OprB, which is inducedby growth in the presence of glucose (Hancock and Carey, 1980), forms a channelthat prefers D-glucose and D-xylose (Trias et al., 1988). OprD was discovered dueto its role in the facilitated uptake of imipenem (Trias and Nikaido, 1990a), acarbapenem which shows excellent activity against P. aeruginosa. However, thenatural substrate for OprD is not imipenem, but its structural analogues,presumably basic amino acids and small peptides containing those amino acids(Trias and Nikaido, l990b). OprP is induced by growth under phosphate starvation(0.15 mM or less) conditions (Hancock et al., 1982). Mutational studiesdemonstrated that OprP is an important component of the high-affinity, phosphate-starvation-inducible, phosphate specific transport (Pst) system of P. aeruginosa(Poole and Hancock, 1986). OprP shows 100-fold preference for phosphate over7other anions by virtue of a phosphate binding site with a Kd of 0.3 mM (Hancockand Benz, 1986). In addition, another porin OprO, which is highly homologous toOprP (Siehnel et at, 1992), forms pyrophosphate-specific channels (Hancock et al.,1992).3. Antibiotic Uptake Across the Outer Membrane.(a) THE HYDROPHILIC PATHWAY. Hydrophiic antibiotics, including avariety of p-lactam antibiotics, tetracycline and chloramphenicol (Foulds, 1976),can pass across the Gram-negative bacterial outer membrane through the water-filled channels formed by porins. The strongest supporting data has been obtainedby comparing porin-deficient mutants with their isogenic wild-type strains. Suchmutants have significant increases in IVIIC for some but not all p-lactams (Hancockand Bell, 1988), as well as 10- to 100-fold-lower rates of -1actam permeation thantheir porin-sufficient parent strains (Hancock, 1987). P. aeruginosa wild-type cellshave a 12-fold lowered permeability to -1actam antibiotics compared to E. coliYoshimura and Nikaido, 1982; Nicas and Hancock, 1983a) and a consequenthigher resistance to hydrophilic antibiotics (Brown, 1975). Therefore, wild-type P.aeruginosa cells behave like porin-deficient mutants despite the high copy numberof their major porin OprF (Angus et al., 1982). This is primarily due to the lowactivity of this porin (only 0.4% of the OprF in the outer membrane forms largepores). Loss of OprF by mutation decreased the outer membrane permeability to the8-lactam nitrocefln by a further six-fold (Nicas and Hancock, 1983a) but had oniya small effect (up to three fold) on sensitivity to many antibiotics (Nicas, 1983;Woodruff and Hancock, 1988).Certain antibiotics can utilize the channels of specific porins to enhance theiruptake since they resemble the specific substrate of the given channel. Diffusionthrough such specific channels can make a major contribution in P. aeruginosa,which has very low outer membrane permeability. One excellent example isimipenem and the related zwitterionic carbapenems. It was shown that OprDproduced a diffusion channel with a specific binding site for basic amino acids andtheir structural analogue imipenem (Trias and Nikaido, 1990). The diffusion ofimipenem through this channel followed saturation kinetics (Trias et al., 1989).(b) THE HYDROPHOBIC PATHWAY. Due to the unusual asymmetricstructure and presence of LPS cross-bridged by divalent cations in the outermembrane (see above), most wild-type Gram-negative bacteria (including P.aeruginosa) exclude moderately hydrophobic antibiotics that are quite effectiveagainst Gram-positive bacteria (Nikaido and Vaara, 1985). These antibioticsinclude macrolides, novobiocin, the more hydrophobic -lactams, rifamycin SV andactinomycin D (Nikaido et al., 1983). Permeabilization to hydrophobic antibioticscan be achieved when the structure of the outer membrane bilayer is modified bymutational alterations of LPS components Nikaido and Vaara, 1985), or byaddition of compounds, which remove (e.g. EDTA) or competitively displace (e.g.polycations) divalent cations from their LPS binding sites (Hancock, 1984; Nikaido9and Hancock, 1986).(c) THE SELF-PROMOTED PATHWAY. The self-promoted pathway hasbeen postulated for the uptake of polycationic antibiotics, like polymyxin andaminoglycosides, across the outer membrane of P. aeruginosa (Hancock, 1981;Hancock et at, 1981b; Nicas and Hancock, 1983b). It involves the displacement ofdivalent cations from LPS by these polycations, thus destroying the LPS cross-bridging and destabilizing the outer membrane (Hancock et at, 198 lb; Nicas andHancock, 1983b). Because this can result in enhancement of uptake of lysozyme,-lactams and hydrophobic fluorescent dyes across the outer membrane, it wasproposed that such interactions promote the uptake of the interacting polycationicantibiotics itself. As further evidence in favour of self-promoted uptake, EDTA, athvalent cation chelator that removes Mg2 from outer membrane sites, causessimilar enhancement of uptake of lysozyme and -lactams (Nicas and Hancock,1983b) as well as enhanced killing by the polycationic antibiotics (Sykes andMorris, 1975).4. Imipenem.Imipenem, or N-formimidoyl thienamycin, is derived from thienamycin, anatural product of the soil organism Streptomyces cattleya (Kahan et at, 1979).Thienamycin has unique structural feature (Fig. 2) that distinguishes it from allnatural and synthetic -lactam antibiotics previously described (Albers-Schonberg10OHHThienamycin —CH2—CH2—NH2Imipenem —OH2—CHNH—CH=NHFigure 2: Structures of thienamycin and imipenem.11et at, 1978). It is the first representative of a new class of antibiotics, thecarbapenems, carbapen- to denote the substitution of carbon for the sulfur moleculeof the five-membered ring, and -em to signify the double bond in the ring (Fig. 2).The unusual trans configuration of the C-6 alkyl side chain and its directattachment to the -lactam ring of thienamycin further differentiates thisantibiotic from the penicillins and the cephalosporins (Fig. 2), both of which havecis-acylamine side chains in the C-6 position. Since thienamycin breaks downspontaneously at high concentrations, synthesis of the amidine derivative Nformimidoyl thienamycin (Leanza et at, 1979) provided a stable, crystallinecompound, imipenem (Fig. 2). Imipenem was the first carbapenem antibiotic to bedeveloped for use in humans.Imipenem is of umque interest because it has an unusually broad spectrum,high potency and no cross-resistance to other -lactam antibiotics. Significantly,activity against the pathogen P. aeruginosa was substantially improved over other-lactains. Comparative studies (Rolinson et al., 1986) showed that the majoradvantage of iniipenem was its broad spectrum and high potency against isolatesexhibiting -lactamase-mediated resistance to one or more of the penicillins andcephalosporins. Furthermore, imipenem had greater bactericidal activity in vitroand greater protective efficacy in experimental infections against diversepathogenic species (Kropp et al., 1980). Therefore, imipenem is particularly usefulin the treatment of pathogens with high intrinsic resistance to many drugs, forexample, P. aeruginosa, and infections caused by mixtures of bacteria for which a12combination of antibiotics would be normally used. In other cases, however,imipenem did not show superior activity (cf. other -lactams) (Rolinson et al.,1986), which may be due to the overestimation of the non-specific permeability ofimipenem across outer membrane Weffido and Hancock, personal communication),and the derepression of the chromosomally-encoded -lactamases (see below). Sinceimipenem is hydrolysed and thereby inactivated by the renal dipeptidase,dehydropeptidase (Kropp et al., 1982), it is administered in combination with equalamount of cilastatin, a dehydropeptidase inhibitor.The high potency and unusually broad spectrum of antimicrobial activity ofimipenem is due to three aspects. Firstly, it is able to penetrate the outermembrane of many Gram-negative bacteria. Imipenem has a compact structurewith a molecular weight of 299 and is zwitterionic, both of which features facilitateits diffusion through the outer membranes of gram-negative bacteria by distinctporin channels (Yoshimura and Nikaido, 1985; Lipman and Neu, 1988). In P.aeruginosa, imipenem can overcome the poor outer membrane permeability bypenetrating through the specific porin OprD (Trias and Nikaido, 1990a). Secondly,it has high affinity for the critical penicillin binding proteins (PBPs) from a broadrange of bacteria. In E. coli and P. aeruginosa, imipenem showed the highestaffinity to PBP-2 and appreciable affinity to most other PBPs (Hashizume et al.,1984). The binding to PBP-1 and PBP-2 is probably the main reason for itsbactericidal action, namely rounding of cells at subinhibitory concentrations andlysis at higher concentrations. Thirdly, imipenem is a poor substrate for a broad13range of p-lactamases from Gram-positive and Gram-negative bacteria (Kahan etal., 1983). This stability is due, in part, to the trans conformation of the side chainon the 6-position of imipenem (Fig. 2). Finally, imipenem is however, a potentinducer of chromosomal cephalosporinases, a class of -lactamases that areproduced in some aerobic Gram-negative bacteria in the presence of selected-lactam antibiotics and are capable of hydrolysing many p-lacts. Howeverimipenem is only weakly hydrolysed by these p-lactamases (Livermore and Yang,1987; Tausk et al., 1985). These properties may account for the general lack ofcross-resistance of imip enem with other -lactam antibiotics.Correspondingly, resistance to imipenem can be mediated by three ways.Firstly, the constituents of the outer membrane maybe modified to prevent thepassage of imipenem. P. aeruginosa mutants that are resistant only to imipenembut remain susceptible to most other -lactams have been isolated both fromclinical and laboratory sources. Their resistance is usually due to the decreased orlack of expression of OprD (Büscher, et al., 1987; Lynch et al., 1987; Quinn et al.,1986). Genetic analysis shows that the elimination of OprD results from generearrangements in the oprD coding region or the upstream promoter region(Yoneyama and Nakae, 1993). Secondly, the structure of PBPs may be altered toreduce the effect of imipenem on the cell wall in, for example, Streptococcus faecium(HeDinger and Brewer, 1991). Thirdly, -lactamases may be expressed which arecapable of hydrolysing and thereby inactivating the -lactam ring of imipenem.Pseudomonas maltophilia is known to be uniformly resistant to imipenem. Saino14et al. (1982) demonstrated, in P. maltophilia, the presence of an induciblepenicillinase, L-1, which is an unusual zinc metalloenzyme that can hydrolyseirnipenem. Fortunately, enzymes of this type appear to be extremely rare in otherspecies, although they also have been reported sporadically in Bacteroids fragilis(Yotsuji et al., 1983), Aerornonas hydrophila (Massidda et al., 1991) and Serratiamarcescens (Osano et al., 1994). These imipenem-hydrolysing -lactamasesconstitute a unique class of Zn2-containing p-lactamases analogous to themetalloproteases whereas most other -1actamase are related to serine proteases.In addition, in P. aeruginosa, it appears the full expression of resistance toimipenem requires both reduced permeability due to loss of OprD and slowhydrolysis mediated by derepressed chromosomal p-lactamase (Livermore, 1992).In Enterobacter, the over-production of group I cephalosporinase and/or thedecreased outer membrane permeability due to the deletion of certain porin(s) alsoconferred imipenem resistance (Lee et al., 1991; Thomson et al., 1993)5. OprD: A Specific Porin for Imipenem and Basic Amino Acids.As stated above, imipenem is highly potent against P. aeruginosa. However,during clinical therapy of P. aeruginosa, imipenem-resistant isolates arise at asignificant rate (Quinn et al., 1986), and usually the resistant strains are not cross-resistant to other antibiotics, cannot hydrolyse or modifr imipenem and do not showany alterations in the affinity or copy number of penicillin-binding proteins. On the15other hand, they lack an outer membrane protein with apparent molecular weight46 kD, which has been identified as protein D2 (now called OprD) (Trias et al.,1979; Büscher, et al., 1987; Lynch et al., 1987; Quinn et al., 1986).OprD is one of the porins in the P. aeruginosa outer membrane (Yoshiharaand Nakae, 1989). The heat-modifiability property of OprD is like that of E. coliOmpA, it runs at a lower molecular weight when solubilized in SDS at lowtemperature, and at the monomer molecular weight when solubilized in SDS athigh temperature. Natively, it is present in the outer membrane as trimers(Yoshihara et at, 1991). OprD forms small diffusion pores, which has beendemonstrated in both liposome sweffing assays (Yoshihara and Nakae, 1989) andin the black lipid bilayer system (Ishii and Nakae, 1993).Purified OprD has been found capable of allowing the size-dependent uptakeof small hydrophilic molecules, and it was suggested that OprD could permit ageneral diffusion of monosaccharides, disaccharides and amino acids at asignificant rate (Yoshthara and Nakae, 1989; Trias and Nikaido, 1990a; Yoshiharaet at, 1991). From this perspective, the channel has been proposed to behave as ageneral porm. More importantly, the following evidence indicates that OprD is alsoa specific channel for imipenem, basic amino acids and their structural analogues:(a) In vitro liposome sweffing assays demonstrated that the OprD channelallowed the diffusion of imipenem at a rate much higher than expected given itsmolecular weight, which was the behaviour expected for a specific ligand whosediffusion was facilitated by the given channel (Trias and Nikaido, 1990a).16(b) In vivo experiments performed with intact cells carrying a plasmidexpressing the gene for L-1 -lactamase from P. maltophilia, showed that the OprDchannel was selective for imipenem over other p-lactam antibiotics (Trias andNikaido, 1990a).(c) Regarding other antibiotics, Trias and Nikaido (1990a) suggestedcarbapenem derivatives, such as Sm-7338, Sch 33755, Sch 33440, meropenem andpanipenem (Fukuoka et al., 1993), containing oniy one basic group at position 2 ofthe molecule also utilized the OprD channel for the facilitated diffusion. However,it was recently demonstrated that the addition of a second basic group at position1 or 6 of a carbapenem which already contained a basic group at position 2,abrogated the role of OprD in its activity (Fung-Tomc et al., 1995)(d) Regarding nutrients, basic amino acids and some small peptidescontaining these amino acids (Fig. 3) were shown to be competitive inhibitors of thediffusion of imipenem (Trias and Nikaido, 1990b).(e) P. aeruginosa showed higher susceptibility to imipenem in minimalmedium than it did in rich medium such as Mueller-Hinton medium. Thesusceptibility was decreased by the addition of basic amino acids to the minimalmedium, whereas the susceptibility to other antibiotics was not influenced. It wassuggested that the decrease in susceptibility to imipenem was related tocompetition with basic amino acids for permeation through the OprD channel(Fukuoka et al, 1991).17OHImipenem —CH2NH—CH=NHCOOHOH CH3 CH3C—N<CH3MeropenemCOOHLysineNH2 1CH2—CH2NCOOHArginine NH2%1.CH—CH2NH—C —NH2NHCOOHCH_NH%NHistidine NH2 CH2 cH —NHCOOHFigure 3: Structures of some antibiotics and amino acids that penetratethrough the OprD channel.18C. Structure Analysis of Membrane Proteins.1. X-ray Crystallography and Others.Two major approaches have been taken to obtain the structural informationon porin proteins. X-ray crystallography solves the structure at the highest possibleresolution, i.e. the atomic level. So far, four general porins and one specific porinhave been crystallized and analyzed in atomic detail. This has created a milestonein our understanding of porin functions. However, this method has the limitationsof being highly technical, requiring specific training and expertise, time consuming,having no guarantee of success, and requiring large amount of highly purifiedprotein. For most porins, in the absence of crystallographic data, structural modelshave been built using known sequences as the starting point (see details below),followed by using genetic, immunological and biochemical approaches to test andmodify the predicted structure. Gene fusion techniques using -galactosidase,alkaline phosphatase and p-lactamase as reporter enzymes have been successfulin studies of the folding of inner-membrane proteins, which contain transmembranesegments composed of hydrophobic residues forming alpha helices (Manoil andBeckwith, 1986). However, since the structure of outer membrane proteins is morerigid by virtue of extensive p-structure, and consequently more dependent ontertiary interactions, the fusion of the reporter enzyme could cause severeperturbations of the native configuration, and fusion junctions may not correspondto ioop regions. Therefore, such techniques are not suitable for the study of outer19membrane protein topology. Another technique, linker insertion mutagenesis,involving introduction of a short stretch of amino acids, modifies the proteins in amore subtle way. It has been successfully used to study the topology of several E.coli porins including LamB (Boulain et al., 1986), PhoE (Bosch and Tommassen,1987) and P. aeruginosa OprF (Wong et al., 1993).2. Crystal Structures of General Porin Channels.The crystal structures of four general porins, including two F. coli porinsOmpF and PhoE (Cowan et al., 1992), Rhodobacter capsulatus porin (Weiss andSchulz, 1992) and Rhodopseudomonas blastica porin (Kreusch et al., 1994) havebeen published. Although they do not share substantial sequence homology, theirstructures revealed striking similarity.(a) BARREL TOPOLOGY. All of the general porins form trimers of identicalmonomers, each monomer consisting of a 16-stranded anti-parallel p-barrelenclosing a pore. The p-strands are amphipathic, in that they are composed ofalternating polar and non-polar amino acid residues which are exposed to theaqueous channel and hydrophobic membrane interior respectively. The p-strandsare connected by seven short p-hairpin turns at the periplasmic side (smooth end),and by eight long loops exposed at the cell surface (rough end). Six external loopspack together and partially cover the entrance to the barrel. The longest loop L3,contains a short piece of s-helix, folds inside the barrel and constricts the width of20the channel. These loop structures cause the channel to lie off center at an angleof about 16° to the barrel axis. The remaining loops are involved in monomerinteractions.(b) PORE ARCHITECTURE. The shape of the pore varies as it traverses themembrane and it can be divided into three parts: the mouth, constriction zone(eyelet), and exit zone. The pore entrance (the mouth) is narrowed by long loops atthe rough end of the barrel to a diameter of 11-49 A. About halfway through themembrane, the cross section decreases to 7 x 11 A where the internal loop (L3) andsome side chains from barrel walls constrict the size of pore (eyelet). The crosssection of the channel increases abruptly to 15 x 22 A right after the eyelet sincethe pore size in this region is simply defined by the barrel walls (exit zone). ForOmpF and PhoE, the three pores are separated over a distance of 30 A which spansthe entire passage through the core of the membrane (Cowan et al., 1992). Incontrast, the three pores in R. capsulatus porin, each with an eyelet determiningthe solute exclusion limit, run separately over a distance of only 20 A, and then thethree pores merge into one channel at the periplasmic side (Weiss et al., 1991).(c) EYELET. The eyelet region of OmpF and PhoE is lined on one side bynegatively charged residues Asp”31106 (OmpF/PhoE numbering), Glu’17”°, whereason the other side it is lined by positively charged residues Lys’6116,Arg42137,Arg82175and Arg’32”26,giving rise to a strong transverse electrical field. These residues arestrictly conserved among eight different porins from enteric bacteria (Jeanteur etal., 1991). Essentially the same arrangement is observed for R. capsulatus and R.21blastica porins. Several experiments indicate that the eyelet region is important forthe selectivity as well as the determination of solute exclusion size for the generalporins (Benson et al., 1988; lVlisra and Benson, 1988; Bauer et al., 1989). The mostdirect evidence comes from the crystal structure of a mutant OmpF protein(Jeanteur et al., 1994b). With a single mutation Gly- ll9-Asp in the eyelet region,X-ray structure analysis reveals a locally altered peptide backbone, with the sidechain of residue Asp- 119 protruding into the channel, causing the original eyeletregion to be subdivided into two intercommunicating compartments of 34 A indiameter. The functional consequences of this structural modification included areduction of the channel conductivity by about one-third, altered ion selectivity andvoltage gating, and decrease of permeation rates of various sugars by 2—’12 fold.(d) TR11VIER STABILITY. The trimer is stabilized by both hydrophobic andhydrophilic interactions between the monomers. The hydrophobic contacts aremade by residues from the barrel walls. Along the trimer axis, large hydrophobicresidues pack together to fifi up the space completely, leaving no room for water.Away from the trimer axis, the p-sheets of the barrel walls pack in a highlycomplementary manner, resulting in extensive contact between the monomers. Thehydrophilic interactions primarily involve a loop L2, which reaches into the poreof a neighbouring monomer where it participates in extensive hydrogen bondingand a few salt bridges. This loop also fills the gap in the wall of the adjacentmonomer left by L3 which folds inside the barrel.(e) AROMATIC RING. The membrane-facing surface of trimeric porins can22be subdivided into three main areas. Starting from the bottom (periplasmic side),the first zone is characterized by the presence of many aromatic residues, mainlyTyr and Phe. Moving upwards is the expected non-polar region composed mainlyof Leu, Val, and Ala. On top is the second aromatic ring at the nonpolar/polarborder of the interface between protein and membrane, which functions to anchorthe protein in the membrane. The flat aromatic surface is ideal for packing withfatty acyl chains, therefore protecting the porin conformation against adversemembrane fluctuations. For Tyr and Trp, the combined properties of hydrophobicityand the ability to form hydrogen bonds is favourable at the polar/nonpolar interfaceseparating regions with dramatically different dielectric constants.3. Crystal Structure of Specffic Porin Channel.Very recently, the first crystal structure of a specific porin of E. coli, LamB,was solved at 3.1 A resolution (Schirmer et al., 1995). LamB, originally discoveredas the receptor of bacteriophage ?, is also a specific porin for maltose andmaltodextrins (Benz et al., 1987; Freundlieb et al., 1988; Gehring et al., 1991), thusit has another name, maltoporin. The X-ray structural analysis of LamB reflects ageneral similarity to the structure of nonspecific porins. On the other hand, it ismore sophisticated to allow for the specific binding and efficient transport ofmaltose and maltodextrins.Active maltoporin is a trimer, with each monomer consisting of an 18-23stranded antiparallel a-barrel, instead of the 16-stranded structure of generalporins. Each monomer contains an independent channel, and all three monomersof the trimer are required for phage adsorption (Iviarchal and Hofnung, 1983).Similar to the general porins, the n-strands are connected by short turns at theperiplasmic side, whereas the cell surface connections are made by long loops. Thethird surface ioop, L3, is entirely folded into the channel, while Li and L6 from thesame monomer and L2 from a neighbouring monomer fold inside to differentextents, forming the constriction zone toward the middle of the channel. The otherloops form a sort of umbrella covering the entrance of most of the channel.As with the general porins, the eyelet region is defined by L3 and a few sidechains from the barrel walls, but the lumen at the channel entrance is furtherconstricted by residues from Li and L6, presumably to increase the selectivity. Thepore has a diameter of 5 to 6 A, considerably smaller than that of OmpF. Chargedresidues are distributed pairwise in the eyelet region and form an electrostaticfield.The most interesting feature is a series of aromatic residues arranged alonga left-handed helical pathway from the inlet to the outlet of the channel. This path(the “greasy slide”) guides the diffusion of sugars through stacking interactions.The hydrophobic faces of glycosyl moieties are known to stack with aromaticresidues in sugar binding proteins (Spurlino, 1991). Other charged residues in thevicinity of the “greasy slide” interact with the hydroxyl group of the sugars and mayaccount for the stereospecificity of the channel. The positions of all selected24mutations with altered affinities toward maltodextrin (Ferenci and Lee, 1982)cluster at the pore eyelet region.4. Prediction of Porin Structure: Porin Superfamily.In the last few years, porin genes from many pathogenic Gram-negativebacteria have been cloned and sequenced. These porins are the focus of manystudies because of their potential use as vaccines, or for bacterial typing, and theirrole in antibiotic resistance. Since their three dimensional structures are usuallyunknown, prediction of their folding patterns is important for further investigation.As porins are comprised of antiparallel p-strands tranversing the membrane, thefirst approach to structure prediction was to identify segments causing thepolypeptide strands to reverse their direction, i.e., turn prediction. According toPaul and Rosenbush (1985), amino acids can be divided into three groups: turnpromoters (N, D, E, G, P, S), turn blockers (A, Q, I, L, M, F, W, Y) and otherresidues. Turns are then predicted as a segment of three or more residuescontaining at least one turn promoter and no turn-blockers. Another approach isbased on the fact that the transmembrane p-strands are amphipathic (Vogel andJahnig, 1986), with one face created by every second amino acid being in contactwith the hydrophobic core of the membrane, and the other facing the hydrophilicpore lumen. Therefore the p-strands should have high amphipathic values.The concept of a “porin superfamily” was proposed by Jeanteur et al (1991).25This superfamily consists of over 30 general porins from five distantly relatedspecies. Though they do not share significant homology based on overall sequence,their transmembrane segments can be aligned with good homology. Therefore,aligning the porin sequence with those of known structures, provides anotherapproach in the structure prediction. Jeanteur et al. (1994a) first combined turnprediction and amphipathicity calculations with multiple alignments, greatlyimproving the quality of these predictions. The relevance of this predictive methodwas confirmed by the crystal structures of OmpF, PhoE and Rhodobactercapsulatus porins (Cowan et al., 1992; Weiss and Schulz, 1992). The specific porinsLamB and Tsx, however, are not able to align with the porin superfamily. Thereforethey form a distant family or families of their own (Jeanteur et al., 1994a)Multiple alignment and topology prediction are complementary tasks. On theone hand, multiple alignment gives better accuracy to a prediction because, if thesequences are properly aligned, predictions of topological elements will bereinforced by being predicted in all aligned sequences. On the other hand,prediction of topology will help to align sequences because predicted topologicalelements can be lined up together. In addition, multiple sequence alignmentshighlight important conserved features of the sequences, thus structuralinformation or biochemical information on one species of porins can be related tomore distant porin species.D. Model Membrane Studies of Porins.26A variety of model systems have been used to investigate the physicalproperties of porins in vitro. The two most-utilized systems are the liposomesweffing assay (Nikaido and Rosenberg, 1981; Hancock, 1986) and black lipidbilayers (Hancock, 1986). These permit one to probe the function of porins inallowing the passage of medium-sized sugars, p-lactams, amino acids and small-to medium- sized ions, respectively. Our laboratory uses the second system in whichporins in detergent solution are added to the aqueous salt solution bathing a planarlipid bilayer. Individual porin molecules then spontaneously insert in a time-dependent fashion into the membrane, an event that can be measured as stepincreases in the conductance between two electrodes placed on either side of themembrane. This method has the rather unique property of having single moleculesensitivity, since amplifying the current through a single channel forming unit, by1O91Ob0 fold, results in events that can be read out on a chart record. In addition,it is capable of providing an estimate of the channel diameter, a precisemeasurement of the ion selectivity of a variety of anions and cations, ameasurement of the heterogeneity of individual channels, and direct evidence forthe presence of substrate binding site(s) in the channel.This system revealed that general porins have the following properties:porins form water-filled channels, with the size of the channel largely determiningthe exclusion limit of the outer membrane for hydrophilic compounds; smallchemicals pass through the middle of the channel in a manner similar to theirdiffusion through bulk water; porin channels are usually either cation or anion27selective, ranging from 2 to 30 fold; porin channels are not voltage gated or, in mostcases, voltage regulated. For specific porins such as OprP and LamB, it was foundthey contained substrate binding sites which, when occupied by substrates, blockthe passage of ions through the channel (Hancock, 1986).E. Aims of This Study.OprD, a specific porin for imipenem and basic amino acids, provides anexcellent model for studying the mechanisms of antibiotic and nutrient uptakethrough the specific porins. Even though previous work and other work performedduring the investigations described here, suggested that OprD had a specificbinding site for imipenem, there still remained many questions. What is thefunction of OprD in transport of other antibiotics and nutrients? What is the foldingpattern of OprD in the outer membrane? Where are the specific binding site(s)located and which residues are responsible for the specific binding? What is themechanism of facilitated diffusion of imipenem and basic amino acids through theOprD channel? Attempts to solve these questions made the goals of this thesis: (1)to further investigate the substrate selectivity of OprD by using isogeneic mutantsexpressing different levels of OprD; (2) to predict an OprD membrane topologymodel and verify it by site-directed mutagenesis; (3) to locate the specific bindingsite(s) for imipenem by studying the functional alterations of the mutants; and (4)to elucidate the molecular architecture of OprD channel in order to understand the28mechanism of imipenem uptake through the channel, which was the ultimate goalof this study.29MATERIALS AND METHODSA. Strains, Plasmids and Growth Conditions.All strains used in this study are listed in Table I and all plasmid used arelisted in Table II. Strains were routinely grown on Luria Broth (LB) medium (1.0%Tryptone, 0.5% yeast extract, 0.5% NaC1) or LB agar containing, in addition 2%agar. For experiments involving growth on specific carbon sources, P. aeruginosastrains were grown on BM2 minimal media (Hancock and Carey, 1979). P.aeruginosa strains were also grown on Mueller-Hinton broth. VB1VI]\’l media is ‘/13(Voger and Bonner, 1956) medium containing 0.3% trisodium citrate as a carbonsource and was selective for P. aeruginosa since E. coli cannot utilize citrate. Theformulation for VBMIV[ (per liter) was as follows: Na3citrate, 3.Og; citric acid, 2.Og;K2HPO410.Og; NaNH4POx4H2,3.5g; adjust (or check) PH 7.0, after autoclavingand cooling add: 0.8 ml of 1M MgSO4x7H2Oand 0.08 ml of 1M CaC12.All mediacomponents were obtained from Difco Laboratories, Detroit, Michigan. Antibioticswere used in selective media at the following concentrations, for E. coli: ampicillin75 jig/mi; chioramphenicol 25 jig/ml; kanamycin 35 jig/ml; for P. aeruginosa:carbenicillin 500 jig/mi; kanamycin 300 jig/ml; streptomycin 500 p,g/ml.B. Genetic Manipulations.1. General Techniques.30Table I:StrainE. coilDH5 aCE 1248S17-1P. aeruginosaHi 03H63 6H673H729Strains.DescriptionsupE44 hsdRl7 recAl endAl gyrA96 1/il-i relAlPorin deficient strain: OmpF, OmpC, PhoFMobilizing donor strain in biparental matingPAO 1 prototroph: wild type reference strainH103 oprF::H103 opdE:: Tn501, imipenem resistant strainH103 oprD::Reference/SourceHanahan, 1983Van der Ley, et al., 1985Simon, et al., 1983Hancock & Carey, 1979Woodruff& Hancock, 1988Huang et a!., 1992This study31TableII:Plasmids.PlasmidDescriptionReference/SourcepTZ18R/19Rgeneral cloningvector,APRPharmaciapBK18R/19RpTZ18R/19Rwitha2.1-kbBainHJ/KpnIfragment encodingtheThisstudyoprDgenepE37/65pTZ19Rwitha4.0-kbEcoRI fragment encodingtheN-terminalThisstudyportionoftheoprDgenepRK767low-copy-number shuttlevectorDittaetal.,1985pD2-45pRK767witha9.1-kbBell fragment encodingtheoprDHuangetal.,1992regulatorygenepUCP18/19pUC18/19-derivedbroadhostrangeplasmidwhichcanbeSchweizer,1991maintainedinbothE.coliandP.aeruginosapNOT19pUC19withtheuniqueNdeIsitechangedtoaNotIsiteSchweizer,1992pMOB3acassettecontainingaportableoriT, thesacBgenefromSchweizer,1992Bacillussubtilisandachioramphenicol-resistancegeneallowingpositiveselectionfor bothoriTandsacB.pUC4KAPAavectorcontainingakanamycinresistantcKmR)interposonPharmaciaflankedbysymmetricalrestrictionsites_______con’t...TableII:Plasmids(con’t)PlasmidDescriptionReference/SourcepXHlaplasmidconstructedfor theallelereplacement mutagenesisofThisstudytheoprDgenepXH2pUCP19witha2.1-kbBarnHJ/KpnI fragment encodingtheoprDThisstudygenepMTZ19RmodifiedpTZ19RwhicheliminatedtherestrictionsitesSalT, PstI,ThisstudySphIandHindIIIinthemultiplecloningsitepMBK19RpMTZ19Rwitha2.1-kbBainHlIKpnI fragment encodingtheoprDThisstudygenepMBE19RpMTZ19Rwitha1.2-kbBarnIfl/EcoRI fragment encodingN-terminalThisstudypartoftheoprDgenepHE1pMBK19Rwithadeletionof24bpfromtheregionencodingtheThisstudypredictedloopLiofOprDpHE2pMBK19Rwithadeletionof24bpfromtheregionencodingtheThisstudypredictedloopL2ofOprDpHE3pMBK19Rwithadeletionof24bpfromtheregionencodingtheThisstudypredictedloopL3of OprD_____con’t...TableII:Plasmids(con’t)PlasmidDescriptionReference/SourcepHE4pMBK19Rwithadeletionof 12bpfromtheregionencodingtheThisstudypredictedloopL4of OprDpHE5pMBK19Rwithadeletionof24bpfromtheregionencodingtheThisstudypredictedloopL5of OprDpHE6pMBK19Rwithadeletionof12bpfromtheregionencodingtheThisstudypredictedloopL6of OprDpHE7pMBK19Rwithadeletionof 24bpfromtheregionencodingtheThisstudypredictedloopL7of OprDpHE8pMBK19Rwithadeletionof 24bpfromtheregionencodingtheThisstudypredictedloopL8of OprDpHP1’-’pHP8pUCP19containingthecorrespondingmutantoprDgenewithThisstudydeletionsintheregionencodingthepredictedloopsL1-’L8All general DNA techniques such as DNA isolation, agarose gelelectrophoresis, radioactive labelling of oligonucleotides, colony blotting, Southernblotting and transformation were performed as described in Sambrook et at (1989).Other methods included slot lysis gel electrophoresis (Sekar, 1987). DNA restrictionand modifying enzymes (Bethesda Research Laboratories (BRL), Burlington,Canada; Boehringer Mannheim, Mannheim, Germany; Pharmacia, Uppsala,Sweden) and Vent DNA polymerase (New England Biolabs, Beverly, MA)were used according to the manufacturer’s method.2. DNA Fragment Isolation.DNA fragments were isolated by the band interception technique (Winbergand Hammarskjörd, 1980) using DEAE paper and the manufacturer’s method(Schleicher and Schuell Inc., Keene, N.H.). In addition, PCR fragments werepurified using the QIAEX Gel Extraction Kit (Qiagen Inc., Chatsworth, California)following the manufacturer’s protocol.3. DNA Sequencing.Plasmid DNA for sequencing was isolated using Qiagen columns (QiagenInc., Chatsworth, California) following the manufacturer’s protocol. Sequencingreactions were set up according to the manufacturer’s method, containing 1 jig of35template DNA, 3.2 pmol of primer and components from an Applied Biosystems Inc.(ABI, Foster City, California) Taq DyeDeoxy Terminator Cycle Sequencing Kit.Sequencing reactions were carried out using an Ericomp thermocycler (96°C for 30sec, 50°C for 15 sec, 60°C for 4 mm; 25 cycles), run on an ABI 370A automated DNAsequencer, and analyzed using ABI 373A Data Collection and Analysis programsfor the Macintosh computer.Two strategies were utilized to obtain the complete sequence from a longfragment. Timed exonuclease III digestions (Erase-a-base, Promega, Madison, WI)were employed to create ordered deletions for both strands of the oprD gene. Forthe strand from Ba,nRZ[ to KpnI (Fig. 10), BamI-]J generated a 5’-protruding endand the adjacent primer binding site was protected from digestion by the 3’-overhang of the SphI restriction site. For the other strand, ClaI and Kvnl were usedas the 5’-protruding and 3’-overhang restriction site respectively. The reaction wascarried out at 30°C, and digestion proceeded at about 210 bp per minute. For eachstrand, samples were taken at 12 time points at 1 mm time intervals. The secondstrategy was a combination of subcloning and building of oligonucleotide primerswhich was used to sequence the oprD regulatory gene and PCR products.4. Transfer of DNA into P. aeruginosa.Plasn-iids were transferred into P. aeruginosa by transformation (Olsen et al.,1982). Briefly, cells to be transformed were grown to an 0D550 of 0.20.6 in LBNS.36Cells from 20 ml culture were pelleted and resuspended in 10 ml cold, 0.15 MMgC12,placed on ice for 5 mm. This step was repeated except that the cells werekept on ice for 20 mill. The cells were then pelleted and resuspended in 1 ml ice-cold 0.15 ml MgC12.For each transformation, 100 ng of DNA was added to 0.2 mlcells and the mixture was placed on ice for 60 mm followed by a 3 mm heat pulseat 37°C. LBNS (0.5 ml) was added and the mixture was incubated at 37°C for 2.5h to allow the expression of the plasmid’s antibiotic resistance gene. Aliquots (0.1or 0.25 ml) of cells were plated on selective methum and grown for 24 to 48 h.Alternatively, biparental mating was also used. Overnight cultures of theplasmid-containing E. coli S17-1 strain were grown in LBNS at 30°C with shaking.The recipient strain was grown overnight at 42°C with shaking. Samples (0.1 ml)of both the donor and recipient strains were mixed in 2 ml fresh LBNS andincubated for 10 to 30 mm at room temperature. The cell mixture was filtered ontoa 0.45 gm membrane. The filter was then plated on a non-selective agar plate andincubated at 30°C overnight. The cells were subsequently washed off the ifiter withsterile saline, serially diluted, spreaded onto selective plates and incubated at 37°Cfor up to 48 hours.5. Oligonucleotide Synthesis and Purification.Oligonucleotides were synthesized on an ABI (Foster City, California) model392 DNAIRNA synthesizer accorthng to the manufacturer’s protocol. The37synthesized oligonucleotides were incubated at 55°C overnight followed by dryingin the Speed Vac Concentrator (Fisher Scientific, Ottawa, ON). The oligonucleotideswere resuspended in 1.5 ml of 0.5 M ammonium acetate and further purified on aC18 SEP-PAK cartridge (Millipore, Milford, Massachusetts) as described byAtkinson and Smith (1984). The 0.5 M ammonium acetate solution containing theoligonucleotides was loaded onto a prepared C18 SEP-PAK column, washed withwater and eluted with 20% acetonitrile (if the oligonucleotides were less than 40bases) or 40% acetonitrile (if the oligonucleotides were more than 40 bases). Theoligonucleotides were either lyophilized or ethanol precipitated beforequantification by A absorbance.C. Cloning Strategy for the oprD Gene.OprD was partially purified by Susan Farmer from P. aeruginosa PAO1strain H103 grown in BM2 minimal medium containing succinate as a carbonsource and the N-terminal amino acid sequence was determined by Sandy Kielland,(University of Victoria, Canada) to be D A F V S D Q A E A K G F I E D S. Takinginto account codon bias in P. aeruginosa, a corresponding 48 mer nucleotide oligopooi was deduced. This was then radiolabelled withy-32P-ATP and used as a probein Southern hybridization analysis with P. aeruginosa chromosomal DNA that hadbeen singly or pairwise digested with several restriction enzymes: &oRI, BamFH,Konl, ClaI. The chromosomal restriction map was made and the N-terminus of the38oprD gene was located, while the expression of OprD could be in either of the twodirections. Fragments of corresponding sizes were isolated from chromosomaldigests and ligated into plasmid pTZ18/19R to construct mini-libraries. Theresulting colonies were screened with the same probe described above.P. Allele Replacement Mutagenesis.1. Construction of pXH1.Firstly, a kanamycin-resistance (Km’) conferring -fragment frompUC4KAPA was isolated as the 1.3-kb SalT fragment and cloned into the XhoI siteon the oprD gene, yielding pBK19R::, which left 0.6-kb and 0.7-kb, respectivelyof chromosomal DNA sequence on either side of the Km’ cassette. Then the 3.6-kbBamHlIKpnI fragment containing the oprD::L2 insert was cloned into the similarlycleaved pNOT19. Subsequently, the MOB3 cassette was isolated as a 5.8-kb NotIfragment from pMOB3 and inserted into the unique NotI site on pNOT19 withoprD::KmR to generate plasmid pXH1 (‘-‘12-kb) (Fig. 4).2. Selection of the oprD::Km mutant.pXHl was transferred into P. aeruginosa by biparental mating. Plasmidcointegrates in which the entire plasmid was inserted into the P. aeruginosachromosome due to homologous recombination were isolated by plating on VBMM39CmNoti oriTA SacBpXH112KbE. coilonNotiBamHIoprD::Km :*ø KpnFigure 4: Diagram of plasmid pXIIl utilized for allele replacement mutagenesis.The lighter shaded arrow between the BamI-H and KpnI restriction sitesrepresents the P. aeruginosa oprD gene coding region, whereas the black thick barin the middle represents the 1.3-kb kanamycin-resistance 2-interposon that wasused to interrupt the oprD gene. The fragment between the two NotI sites is the5.8-kb MOB3 cassette. The orientations of the oprD gene and ampicillin-resistancemarker are indicated. Abbreviations: Ap: ampiciliin-resistance gene; oriT: origin oftransfer; Cm: chloramphenicol resistance gene; SacB: sucrose expression results insusceptibility to sucrose; Km: kanamycin-resistance L-interposon; E. coli ori: E. colispecific origin of replication.40Km Cb plates. Several colonies that were Km’ and CbR were then grown onMueller-Hinton kanamycin plates containing 5% sucrose to select for arecombination event deleting the sacB gene and other vector sequences. Themutants were characterized by Southern analysis and the isolation of outermembrane proteins.E. Overexpression of the oprD Gene in P. aeruginosa.To overexpress OprD, the 2.1-kb BainIHIKpnI fragment from pBK19Rcontaining the oprD gene was cloned into pUCP19 to form the plasmid pXH2, sothat the direction of expression of the oprD gene was in the same orientation as thelac promoter (Fig. 5). It was then transformed into E. coli Si 7-1 and mobilized backinto P. aeruginosa H103, H636 and H729.F. Electrophoresis.Proteins were separated by electrophoresis through 7%, 11% or 15% SDSpolyacrylamide gels (SDS-PAGE) as previously described (Hancock and Carey,1979).G. Growth experiment.41ApSFpXH26 6 KbPlac LacZBamHI —KpnIOprDFigure 5: Diagram of plasmid pXFI2 utilized for the overexpression of the oprDgene.The dark shaded arrow between the BamHI and KpnI restriction sitesrepresents the P. aeruginosa oprD gene coding region. The orientations of theampicillin resistance gene, the oprD gene, origin of replication and the lac promoterare indicated. The 1.8-kb stabilizing fragment is for the maintenance of the plasmidin P. aeruginosa. Abbreviations: Ap: ampicfflin resistance gene; SF: stabilizingfragment; Plac: lac promoter; On: origin of replication for E. coli.42For growth experiments, each strain was grown to mid-exponential phase onminimal medium with the specific carbon source, glucose, gluconate or pyruvate.They were then subcultured 1 in 50 into prewarmed fresh media containing theindicated levels of saccharides and grown with shaking at 37°C. Samples (1 mleach) were taken at regular intervals for measurements of optical density at 600nm. Growth rate was calculated by the equation: t=ln 2/g, where jt was growth rate,expressed in hours and, g was doubling (generation) time which was determinedfrom a semi-logarithmic plot of the growth curve.H. Purification of OprD.For the purification of OprD from E. coli, strain CE1248(pBK19R) was usedand cultures were grown at 37°C with 50 ig/ml ampicillin, 0.4% glucose and 1 mMisopropyithiogalactoside (IPTG) to an 0D600 of 0.8 to 1.0. For the purification ofOprD from P. aeruginosa, strain H636pXII2) lacking OprF and overexpressingOprD was used, and cultures were grown at 37°C with 500 g/m1 carbeniciffin and0.4% glucose to an 0D600 of 0.8 to 1.0. Mutant proteins OprDAL2 and OprDAL5,were purified from CE 1248(pHE2) and CE 1248(,pHE5) respectively. The wholepurification procedure could be divided into the following three steps:(i) Isolation of outer membranes:Cultures were harvested and the cell pellet was resuspended in cold 20%sucrose containing 50 jig/nil DNaseI. The cell suspension was passed twice through43a French pressure cell (American Instrument Co., Inc. Silver Spring, MD) at 15,000psi. After unbroken cells were removed by low-speed centrifugation (3000 rpm, 10mm), the cell lysate was applied to a 2-step sucrose gradient (50% and 70%) andcentrifuged in L8-70 Ultracentrigue (Beckman Instrument, mc, Fullerton, CA) at21,000 rpm overnight. The outer membrane band was collected from the interfaceof the above two sucrose steps and diluted with distilled water, and centrifuged at45,000 rpm for 1 h to get rid of sucrose.(ii) Detergent solubilization:The outer membrane fraction was subjected to a 3-step differential detergentsolubilization to concentrate OprD and remove other membrane components.Firstly, the pellet was extracted with 10 mM Tris-HC1 (pH8.0), 0.5% octylpolyethylene (octyl-POE) (Bachem Bioscience Inc., Philadelphia, PA), followed bycentrifugation at 45,000 rpm for 1 h. The supernatant was reserved and the pelletwas extracted with 10 mM Tris-HC1, 3% Octyl-POE, 0.2 M NaC1 followed bycentrifugation as above. Finally OprD was largely extracted from the pellet with10 mM Tris-HC1, 3%Octyl-POE, 0.1 M NaCl and 5 mM EDTA followed bycentrifugation, and this supernatant was thalysed against 10 mM Tris-HC1, 5 m1VIEDTA and 0.08% N, N, dimethyldodecylamine-N-N-oxide (LDAO) (Fluka Chemika,Ronkonkoma, NY).ciii) Fast Protein Liquid Chromotography (FPLC):The solubilized protein was loaded onto an FPLC anion exchange columnMono Q, bed volume1.0 ml, flow rate0.5 to 1.0 ml/min) that had been44equilibrated with 10 mM Tris-HC1, 5 mM EDTA and 0.08% LDAO. The protein waseluted by applying a linear gradient of buffer which contained the aboveingredients plus 0 to 1.0 M NaC1. After the first run, the fractions which containedthe least contaminants were pooled and subjected to a second run with a muchflatter salt gradient and a lower elution speed. OprD was eluted in a purified formduring this step. The purified OprD was aliquoted and frozen at -70°C.I. Immunological Techniques.1. Production and Purification of OprD antibodies.With the help from Bill Masin and Mike McClymont, anti-OprD polyclonalantibodies were raised in New Zealand White rabbits as described by Poole andHancock (1986). FPLC-purified (100 jig) OprD was injected subcutaneously on days0, 14, 28, 42, and 56. A booster shot of 200 jig OprD was given on day 68. For thefirst injection, OprD was mixed with equal volume of Freund’s incomplete adjuvant(Difco, Detroit, MI, USA. Subsequently it was injected in FreuncVs completeajuvant. Two weeks after the last injection, the rabbit was bled and the serum wasisolated.The antiserum was purified by absorbing against whole cells of P. aeruginosaOprD-defective strain H729 as follows. Cells from 5 ml overnight culture in LBNSwere harvested by centrifugation at 7,000 rpm for 10 mm and washed twice byresuspension and centrifugation with sterile saline. The cell pellet was resuspended45directly into 1 ml of antiserum and incubated with shaking for 45 mm at roomtemperature. The cells were then pelleted and the antiserum-containingsupernatant was absorbed a second time against a fresh batch of washed cells.Whole cell absorbtion effectively removed most antibodies directed against otherouter membrane components and gave reasonably clean backgrounds when usedat a 1:2000 dilution in Western-immunoblotting.2. Western-immunoblotting.Western-immunoblotting was done as previously described cMutharia andHancock, 1983).J. Prediction of The OprD Topology Model1. Sequence Alignment.The first criteria utilized for the modeffing was based on sequencealignments of P. aeruginosa OprD with E. coli OmpF, PhoE and Rhodobacterporins. PCGENE program was used to perform the pairwise and multiplealignments, the gap penalty was adjusted to optimize the alignment and minimizegaps in the known transmembrane segments. It is clear that surface loop regionsof the porins undergo maximal variation, so the nonhomologous regions andlorsmall deletions (or insertions) would be preferentially located at the surface loops.Conversely, the transmembrane regions are more conserved. From these46alignments, OprD showed highest homology with E. coli OmpF compared to PhoEand Rhodobacter porins in the n-strand regions, so the transmembrane segmentswere primarily predicted according to the consensus between OprD and OmpFsequences.2. Structural Characteristics.The primary model was then adjusted to the structural characteristics ofporins as confirmed by the known structures.(a) TURN. The p-strands were connected by short turns at the periplasmicside and turns were mainly identified according to the definition of Paul andRosenbush (1985). The newly published porin structures have shown thatperiplasmic loops were very short, involving only a few residues (Cowan et al.,1992; Weiss and Schulz, 1992). As a consequence, the prediction of the periplasmicp-turns also took into account the frequency of residue occurrence within theseturns in E. coli and Rhodobacter porins.(b) TRANSMEMBRANE STRANDS. Porins have high content (more than60%) of p-sheet conformation (Kieffel et al., 1985). And the p-strands were typicallyamphipathic in that they were composed of alternating polar and non-polar aminoacids which were exposed to the aqueous porin channel and hydrophobic membraneinterior respectively. This requirement was not strict, however, because internalresidues could be hydrophobic if they were buried by internal structures.47(c) EXTERNAL LOOPS. The external loops were long and of variable length,they contained many polar residues and were hydrophilic.(d) CONSERVED RESIDUES. Some residues were conserved among manyporin, for example, the existence of a aromatic ring at the water/lipid interface andlarge excess of negatively charged residues at the level of the LPS headgroup on theouterface of the membrane.This model was further modified by Dr. Denis Jeanteur, using a computerprogram to perform multiple alignments with porin superfamily and calculate theamphipathicity of the predicted a-strands (see Appendix).K. PCR-based Site Specific Deletion Mutagenesis.1. Two PCR Strategies.Strategy I, direct extension (Vallette et al, 1989), was applied to those loopencoding regions with convenient restriction sites adjacent to the nucleotidesequence to be deleted. This procedure required 2 synthetic oligonucleotides as theprimers to amplify the nucleotide sequence of interest. Primer ‘a’ contained therestriction site and the desired deletion. Primer ‘b’ annealed at another end of thetargeted sequence and was oriented in the opposite direction (Fig. GA).For those loop-encoding regions without convenient restriction sites locatednear the sites of mutagenesis, strategy II, overlap extension (Ho et al, 1989), wasemployed. This method required 2 pairs of primers, an external pair ‘a’ and ‘d’48I. Direct extension.ni R2I IPrimer a Primer bII. Overlap extension.Primer a Primer c>F”________\Primer b Primer dPrimer a____________—CDI Primerdoverlap extensionRi R2Figure 6: Schematic diagram showing two PCR strategies.The primer labels presented correspond to those listed in Table III. Ri andR2 were two unique restriction sites flanking the mutated site and they were usedfor the later cloning procedure. I. Direct extension. The deleted sequence is shownby the broken circle, primer ‘a’ was the mutagenic oligo which contained restrictionsite Ri and the deletion. II. Overlap extension. The deleted sequence is shown asthe thick bar in the middle, the solid and broken lines are the template sequenceson either side of the deletion site. Primers ‘b’ and ‘c’ were designed such that their5’ ends were complementary to the template sequence on one side of the deletionand their 3’ ends were complementary to the template sequence on the other sideof the deletion. The first step PCR products AB and CD thus overlapped at thedeletion site.49which hybridized at each end of the target sequence, and an internalcomplementary pair ‘b’ and ‘c’ that hybridized to either side of the desired site of themutation and contained the desired deletion. The first stage PCR involved separateamplification of the oprD gene with primers a/b or c/d, yielded two fragments ARand CD, which overlapped with one another. These PCR products were mixed andthe second PCR utilizing primer aid, involved extension of the overlap and, resultedin the mutant product (Fig. 6B).To design primers, the PCGENE program was used to minimize the chanceof non-specific binding and primer-dimer interactions. For the mutagenic primers,at least 20 nucleotides from each side of the deletion site were included.2. Construction of pMBK19R and pMBE19R.Plasmids pMBK19R and pMBE19R were derivatives of pBK19R andconstructed as follows. Plasmid pTZ19R was digested by SalT and HindIII and, theresultant large fragment was blunt ended by Klenow fragment and then self-ligatedto form plasmid pMTZ 19R. This procedure eliminated the restriction sites Sail,PstI, SphI and HindITI in the multiple cloning site. The 2.1-kb BamIlI/KpnIfragment from pBK19R containing the oprD gene was cloned into pMTZ19R to formpMBK19R (Fig. 7), which was used as the template for the mutagenesis of thepredicted loops L3 to L8 of OprD. The 1.2-kb BamHl/EcoRI fragment from pBK19Rcontaining N-terminal of the oprD gene was cloned into pMTZ 19R to construct50EcoRISad!*KpnISail*ClaI* XmaliT AiJ1c.* NcoI\4{jJP 44%%4.AV \5000EcoRI I*pstI‘oprDj 4000 pMBK19R 1000 mp5009 bpsSail3000 2000/ \BamHIIXbaIFigure 7: Diagram of plasmid pMBK19R.The hatched bar represents the 2.1-kb BamH[IKpnJ fragment containing theoprD gene coding region. The position and orientation of oprD and the ampicillinresistance marker are indicated by the stippled arrows. The restriction siteslabelled with asterisks were the unique sites utilized for the PCR-cloningmutagenesis of the predicted loops L3 to L8 (Table III).51p1VIBE19R (Fig. 8), which was used as the template for the deletion mutagenesis ofpredicted loops Li and L2.3. Polymerase Chain ReactionThe reaction mixture (total volume 50 jtl) contained: 5 tl of 10 x Ventreaction buffer, 400 tM each dNTPs, 10 ng DNA template, 1 tM of each primer,and 2 units of Vent polymerase (New England Biolab, Beverly, MA). The reactionswere carried out for 20 cycles using a DNA thermal cycler (Ericomp Inc.). Eachcycle included a heat denaturation step at 94 °C (1 mm), followed by annealing ofthe primer at 50-55 °C (2 mm) and primer extension at 72 °C (1-1.5 mm).The PCR strategy, oligonucleotide primers, DNA template, and restrictionsites used for the mutagenesis of each of the predicted ioops are listed in Table III.L. Whole Cell Lysate.Cells from a 1.5 ml overnight culture were harvested and resuspended in 100jtl TE (10 mM Tris-Ci, 1 mM EDTA) containing 1 mM phenyl methyl sulfonylfluoride (PMSF). The cell suspensions were frozen on dry ice followed by thawingat 37°C for three times. The cell lysate was thoroughly sonicated for 5 mm and thencentrifuged for 5 mm, 5 to 10 Ltl of supernant was heated at 90 °C for 10 mm insolubilization-reduction mix (Hancock and Carey, 1979) and run on SDS-PAGE.52*ECORJPstIj*S all3500500/pMBE19R— 3000 10003925 bps25001500/\jiamllhi 2000XbaIFigure 8: Diagram of plasmid pMBE19R.The hatched bar represents the 1.2-kb BamFH/EcoRI fragment containingthe N-termina]. coding region of the oprD gene. The position and orientation of theincomplete oprD gene and the ampicillin-resistance marker are indicated by thestippled arrows. The restriction sites labelled with asterisks were the unique sitesutilized for the PCR mutagenesis of the predicted loops Li and L2 (Table III).53TableIII:PCR-basedsitespecificdeletionmutagenesis.PredictedMutagenesisPCRCloningloopsstrategyaPrimersforPCRbtemplatesites1IpMBE19RSail/BamHI1-a:CACACAGGAAACAGCTATGACCATG(reverseprimer)1-b:93OCCAGTCGACCGCGGTCCCCGCTGCCGCT879GflGCGGAGCAGCAGGTCGAGg592-a:853AocAGccTcoAccTGcTGcTccoc76pMBE19R2-b:1098GCTGTAGTCATCGCGCGGCTr0781053GCCGGTGCCGGTCTrGTCGGA332-c:1033TCCGACAAGACCGGCACCGGC0531078AAGCCGCGCGATGACTACAGC982-d:CGTFGTAAAACGACGGCCAGT(universalprimer)3-a:1219GGcrrccAGcTGcAGAGcAGc2391264GcAGGCcAcTrCAccGAXIGc84pMBK19R3-b:1280TGcGcCAGAGCCGTFGCGGCcGAA574-a:1189GGCAGCCGCCTGTrCCCGCAGACC212pMI3K19R4-b:1422GAGTFCGGCGCCGTACAGGGA4021389ATCGGTGATrGCGTAGCGGCC3694-c:1369GGCCGCTACGCAATCACCGAT3g,TCCCTGTACGGCGCCGAACTC4224-d:3-b5-a.1372CGCTACGCAATCACCGATAACCTC955-b:1576CCAAGTGGTGTfGCTGATGTC4.21GTflGTGCGGTAGATG’1TGAA05-c:1501rccATcTAccGcAcAAAc5211546GAcATcAGccAccAcrrx765-d:3-b6-a:1645GATrATATCGGCTTCGGCCGC6651678GCAGGTGGCGACTCQATrrrC98pMBK19R6-b:16g4GTAOrrcTrATAGccGAcGYFGTF661II I II2 3 4 5 6EcoRI/SaIIPstJJNcoIPstI/NcoIIpMBK19RPstI/NcoIXnzaIII/ClaIcon’t..a.StrategyI wasdirectextension,strategyIIwasoverlapextension.b.Thelabels,a,b,c,d,correspondtothelabelsinFigure6.Alloligosarelistedfrom5’to3’end.Position1wasthestartsiteofBamBIonpBK19R,thenumber wasthenucleotidepositiononthetemplate.Allof theprimerswithnumbersinthemiddlewerethemutagenicprimers,wherethenumbersindicatethecorrespondingnucleotidesfromthetemplatethatwereomittedintheoligonucleotide.c.Underlinednucleotidesweretherestrictionsitesbuiltintotheoligonucleotide.TableIII:PredictedloopsPCR-basedsitespecificdeletionmutagenesis(con’t).MutagenesisstrategyaPrimersforPCRb7-a:1828AAGGACATCGATGGCACCAAG8481873AAGAACTACGGCTACGGCGAG937-b:universal primerPCRtemplatepMBK19R7I-8II8-a:1800GAC1TrCATGOTCCGCTATATCAA23pIVIBK19RClaI/KpnI8-b:2040GATCAGGCGGAACTCGYrCTG1995.2GGCACGGTGCCAGGCCTGGCG78-c:1975CGCCAOGCCTGGCACCGTGCC9952020CAGAACGAGTFCCGCCTGATC20408—d:universal primerCloningsitesClaIIKpnIM. Typsinization StudiesOprD variants in P. aeruginosa outer membrane samples were digestedusing trypsin (TPCK treated, Sigma) at a concentration of 1 mg/mi, in 10 mM TrisCl (PH 8.0) at 37 °C for 1 hour. Untreated samples were incubated in the sameconditions except that trypsin was omitted. Proteolysis was stopped by heating at90 °C for 10 mill in solubilization-reduction mix (Hancock and Carey, 1979). Thetrypsimzed samples were run on SDS-PAGE and analyzed by Western-immunoblotwith anti-OprD polyclonal antibody.N. Black Lipid Bilayer analysis.The techniques and instrument for this procedure were detailed by Benz andHancock (1981) and Benz et al. (1985). The apparatus included a Teflon chamberdivided into two compartments by a Teflon wall that contained a small hole (0.1mm for single channel conductance experiments; 2 mm for macroscopic conductanceexperiment). Electrodes dipped into the aqueous solution on both sides of the hole.A membrane was formed across the hole by painting a solution of 1.5% (w/v)oxidized cholesterol in n-decane. Bilayer formation was indicated by the membraneturning optically black to incident light.1. Single Channel Conductance.56For single channel conductance measurements, one electrode was connectedto a millivolt voltage source. The other was connected to a Keithley 427 currentamplifier to boost the output 109-fold, a Tektronik 511 1A storage oscifioscope tomonitor the amplified output, and a chart recorder (Huston Instrument). PurifiedOprD was diluted at least 1000 times in 0.1% Triton-X 100 and 5 j.tl was added toone side of the chamber. For each experiment, more than 100 events were recorded.2. Macroscopic Conductance Experiment.For macroscopic conductance experiments and zero-current potentialmeasurements, one electrode was again connected to a voltage source while theother was connected to a Keithley 6 1OC electrometer. The experiment was initiatedby adding 5 to 10 jil of 1:10 diluted purified OprD to the bathing solution (1.0 MKC1) on either side of the lipid bilayer membrane. The increase in conductance(measured as current increase) was followed until the rate of increase had sloweddown considerably. At this time membrane conductance had increased by 2 to 3orders of magnitude. The bathing solutions in both compartments of the chamberwere stirred gently (approx. 60 rev./min) with a magnetic stir bar and aliquots (60jil) of imipenem solution (20 jiM) were added carefully to both compartments.Sufficient time (usually 2 mm) was allowed for the new current level to beestablished before addition of the next aliquots.573. Zero-Current Membrane Potential.The experiment was initiated exactly as described above for macroscopicconductance experiment using a bathing solution of 0.1 M KC1. After the membraneconductance had increased two orders of magnitude, the applied voltage was turnedoff and the Keithley 610 electrometer switched to measure voltages. Aliquots (60il) of 3 M KC1 solution were added to the compartment on one side of themembrane (the concentrated side) and equal aliquots of 0.1 M KC1 solution wereadded at the same time to the other side. The solutions in the compartments werestirred (approx. 120 rev./min) to allow relatively rapid equilibration. Theconcentration gradient of KC1 across the membrane provided a chemical potentialwhich was the driving force for ion movement. Ions then diffused across the porinchannels according to the ion selectivity characteristics of the channel until thevoltage caused by preferential one of the ions balanced the chemical potential. Atthis stage the zero-current potential was measured and fitted to the GoldmanHodgkin-Katz equation (Benz et al., 1978) to determine the relative permeabilitiesof anions (Pa) and cations (Pc).0. Assays.1. Protein Assay.Protein concentrations were estimated with a modified Lowry assay58(Sandermann and Stromiger, 1972). A 1 mg/mi solution of bovine serum albuminwas used as a standard.2. Nitrocefin Assay.Whole cell -lactamase levels were measured by a nitrocefin assay aspreviously described by Angus et al. (1982). Briefly, cells of 20 ml cultures weregrown to0D600=0.5—’O.8, harvested and resuspended to the same final OD (1.0) in10 miVi Na-Hepes (P117.0). 3 ml cell suspension was passed twice through a smallFrench pressure cell at 500 psi. 0.1 ml cell lysate was added to 0.65 ml nitrocefinsolution (0.1 mg/ml in Hepes buffer) in a semi-microcuvette, and the kinetics ofnitrocefin hydrolysis was monitored at 0D495, using a Perkin-Elmer (Lambda 3)dual-beam spectrophotometer coupled to a Perkin-Elmer 561 chart record.3. Minimum Inhibitory Concentration MIC) Determination.To determine the MIC, each strain was grown overnight in Mueller-Hintonmedium. Mueller-Hinton or other media agar plates containing serial two-folddilutions of appropriate antimicrobial agents were inoculated with iO cells in a 10tl volume. MICs were determined at least three times and were assessed after 18h of incubation at 37°C. The MICs were taken as the lowest antibiotic concentrationat which cell growth was inhibited. The influence of basic amino acids and59gluconate on inilpenem susceptibilities of F. aeruginosa strains was determined byusing BM2 miiilmal metha containing 20 mM carbon source (glucose or succinate)supplemented with an basic amino acid or glucose or gluconate.60RESULTSCHAPTER ONE Analysis of Two Gene Regions Involved in the Expressionof OprD.A. Molecular Cloning of the oprD Gene.According to chromosomal restriction mapping, the N-terminus of the oprDgene was located between BamFTI and EcoRI sites, and the expression of OprDcould be in either of the 2 directions (Fig. 9B). Since the molecular weight of OprDwas 46 klJ, the oprD gene should be around 1.3-kb. Two fragments were cloned: a4.0-kb EcoRI fragment and a 2.0-kb BainHJ/KpnI fragment which should havecovered the whole oprD gene no matter in which direction the gene was expressed.The 4.0-kb EcoRI fragment was cloned into pTZ18R in two orientations, generatingplasmids pE37 (correct orientation) and pE65 (inverse orientation), meanwhile, the2.1-kb BamRIIKpnI fragment was cloned into pTZ 18R and PTZ 19R, giving rise topBK18R (inverse orientation) and pBK19R (correct orientation) (Fig. 10)respectively.B. Overexpression of the oprD Gene in E. coli CE 1248.Expression studies with the various subclones were performed in E. coliCE 1248, a mutant which lacks the major E. coli porins OmpF, OmpC and PhoE.Plasniids pBK18R, pBK19R, pE37 and pE65 were transformed into E. coli CE 124861A ClallSstISadEcoRV EcoRIEcoRI BgIII Sad EcoRIBoll Psti XhoI BamHI BglIIPstI I Ecu BglI i PsLI PstI I Bollii. . . . . )i Ji I.0 Tn501 4.0 6.0 .0 PopdE‘ORF2’B.SairPstrEcoRICIa!SaltClal EcoRI KpnI BamEl I tpnl BamHII ih6.014.0 KbpoprDFigure 9: Restriction endonuclease maps ofF. aeruginosa PAO1 strain 11103chromosomal DNA derived by Southern hybridization.(A) Map of the region surrounding the insertion site of Tn50 1 (marked bysolid triangle) in strain 11673, thick bar indicates DNA cloned in pD2-45. (B) Mapof the region surrounding the oprD gene, thick bar indicates DNA cloned inpBK19R. Asterisks indicate restriction sites discovered by DNA sequencing only.62EcoRIS acT!KpnTh:SalTClaTXmaIIINcoI\3000XhoIEcoRI\PstI oprD1000 mpSalT3000 2000/PlacBainHISalIIPstTISphTjHindillFigure 10: Diagram of pBK19R.The hatched bar represents the 2.1-kb Bam.HZ[/KpnI fragment containing theoprD gene coding region. The position and orientation of the oprD gene, lacpromoter and ampicillin-resistance marker are indicated.63and the transformants were grown on LB medium with 1 mM isopropyl- p-Dthiogalactopyranoside (IPTG) to induce the high expression of lac promoteradjacent to the insert.E. coli CE1248(pBK19R) carrying the 2.1-kb BamIH/KpnI fragment clonedin the same orientation as the lac promoter revealed expression in E. coli of a outermembrane protein migrating with the same mobility as OprD, and the expressionlevel was almost equivalent to that of the E. coli major outer membrane proteinOmpA (Fig. 11, lanes 5 & 6). When cloned in the reverse orientation to the lacpromoter in pBK18R, only weak expression was observed (Fig. 11, lane 8),suggesting that the cloned BamH[/KpnI fragment contained an oprD genepromoter that could be recognized by E. coli, and that OprD was weakly expressedfrom this promoter. E. coli CE 1248(pE37) and E. coli CE 1248(pE65) containing the4.1-kb EcoRI insert did not result in production of a band of equivalent molecularweight to OprD (Fig. 11, lane 9 & 10), proving it did not contain the entire gene.C. Nucleotide Sequencing of oprD Gene.Both strands of the 2.1-kb BamHIIKpnI fragment containing the oprD genewere sequenced. Within this fragment, an open reading frame containing 1,329nucleotides was obtained (Fig. 12). It encoded a 443 amino acid preprotein. Aminoacids 24’4O were identical in sequence to the N-terminal sequence obtained fromthe purified protein, whereas the ftrst 23 amino acids had the features typical of a641 2 3 4 5 6 7 8 9 1096— —=—OprDEOmpAO•••••j•— — I — — — — — —: iiaFigure 11: Overexpression of the oprD gene in E. coli.SD S-PAGE demonstrating the cloning of DNA fragment influencing theexpression of OprD in E. coil. Lanes: 1, molecular weight marker; 2, fl636; 3,CE1248; 4, CE1248(pTZ19R); 5 & 6, CE1248(pBK19R); 7, CE1248(pTZ18R); 8,CE1248(pBK18R); 9, CE1248(pE37); 10, CE1248(pE65). IPTG was added to all E.coli cultures to induce expression from the lac promoter.65GGATCCAAAGCGAACATACTGACCTCTCCTGTTCGACCGTCGTTCATGGACAcTTAQCCCCTcccTccGGGAAGGGccccGccGTAAcTGccGcGCAG 99GATACTTCGCCGCCCGGCCAAAGCAAGCCCACACATCCGCCCGCCCCAGcTTGGcGcGccTcTccAGccGMcGccccATMTGccGGccAMTGM 198TACAGCGCGACGCCGAACATAAGACATGCCGTGGATACAAACGCATTCGCCACAGACMCTCGATGGCAACCAACCCTTGAAGCAGACGGATTACMTC 297AGGTTTCWGCATAATTCGTTTGCTTTcAAAcAcMTAGccTcGcTcTcGAAaAGAccAAcTGGAATAcATAGGcGAAGccATTTTccpTTTTcA 396CGGAGTTTGCTTATACCTCTTTCATCACAGTAAGAGGGGCCGTACGGAACATGACATTTTTATTAcAAGGCCCCGcCMTCGGGAWGCGACTTGAGA 495AGCGACCTCAACAAGAGTGACCAACCCCGCGACATACGTCATTTTTTCAACTGcGCACCTACGCAGATGCGACATGCGTCATGCAATTTTGCGACAGCA 594693TGTGATGGCAGAGATAATTTCAAAACCAAAGGAGCAATCACAATGAAAGTGATGAAGTGGAGCGCCATTGCACTGGCGGTTTCCGCAGGTAQCACTCAG 792- --- N K V N K N S A I A L A V S A G S T Q 19TTCGCCGTGGCCGACGCATTCGTCAGCGATCAGGCCGAAGCGAAGGGGTTCATcGAAGACAGCAGCCTCGACCTGCTGCTCcGCMCTACTATTTcC 891F A V A D A F V S V Q A E A K G F I E V S S L 0 1 L I R N Y Y F N 52CGTGACGGCAAGAGCGGCAGCGGGGACCGCGTCGACTGGACCCAAGGCTTCCTCACCACCTATGAATCCGGCTTCACCCAP.GGcACTGTGGGCTTCGGC 990R D G K S G S G D R V V W T Q 0 F L T T Y E S 0 F T 0 0 T V 6 F 0 85GTCGATGCCTTCGGCTACCTGGGCCTGAAGCTCGACGGCACCTCCGACAAGACCGGCACCGGCAACCTGCCGGTGATGAACGACGGCAAGccGcGCGAT 1089V V A F G Y I 0 1 K L V 0 T S D K T 6 T 0 N L P V N N D G K P R 0 118GACTACAGCCGCGCCGGCGGCGCCGTGAAGGTGCGCATCTCCAAGACCATGCTGAAGT600GCGAGATGCAACCGACCGCCCCGGTcTTcGccGCTGGc 1188V Y S R A G 0 A V K V R I S K T N L K W G E M 0 P T A P V F A A G 151GGCAGCCGCCTGTTCCCGCAGACCGCGACCGGCTTCCAGCTGCAGAGCAGCGAATTCGAAGGGCTCGACCTCGAGGCAGGCCACTTCACCGAGGGCAAG 1287G S R I F P Q T A T G F 0 L Q S S E F E G L V L E A G H F T E 0 K 184GAGCCGACCACCGTCAAATCGCGTGGCGAACTCTATGCCACCTACGCAGGCGAGAcCGCCAAGAGCGCCGATTTCATTGGGGGCCGcTAcGcAATCACC 1386E P T T V K S R C E L Y A T Y A G E T A K S A V F 1 6 G R Y A I T 217GATAACCTCAGCGCCTCCCTGTACGGCGCCGAACTCGAAGACATCTATCGCCAGTATTACCTCAACAGCAACTACACCATCCCACTCGCATCCGACCAA 1485D N I S A S I Y G A E L E 0 I Y R 0 Y Y L N S N Y T I P L A S V 0 250TCGCTGGGCTTCGATTTCAACATCTACCGCACAAACGATGAAGGCAAGGCCAAGGCCGGCGACATCAGCAACACCACTTGGTCCCTGGCGGCAGCCTAC 1584S I C F V F N I Y R T N V E G K A K A G D I S N T T U S L A A A Y 283ACTCTGGATGCGCACACTTTCACCTTGGCCTACCAGAACCTCCATGGCGATCAGCCGTTTGATTATATCGGCTTCCCCCGCAACGGCTCTGGCGCAGGT 1683T L D A H T F T I A Y Q K V H G V Q P F 0 Y I 6 F 6 R N C S 6 A 6 316GGCGACTCGATTTTCCTCGCCAACTCTGTCCAGTACTCCGACTTCAACGGCCCTGGCGAGAAATCCTGGCAGGCTCGCTACGACCTGAACCTAGCCTCC 1782C V S I F L A N S V 0 Y S D F N C P G E K S U 0 A R Y D L N I A S 349TATGGCGTTCCCGGCCTGACTTTCATGGTCCGCTATATCAATGGCAAGGACATCGATGGCACCAAGATGTCTGACAACAACGTCGGCTATAAGAACTAC 1881Y C V P G L T F N V R Y I N G K V I D C T K N S 0 N N V 6 Y K N Y 382GGCTACGGCGAGGATGGCAAGCACCACGAAACCAACCTCGAAGCCAAGTACGTGGTCCAGTCCGGTCCGGCCAAGGACCTGTCGTTCCGCATCCGCCAG 1980C Y C E V C K H H E T N L E A K Y V V Q S C P A K 0 L S F R I R Q 415GCCTGGCACCGTGCCAACGCCGACCAGGGCGAAGGCGACCAGAACGAGTTCCGCCTGATCGTCGACTATCCGCTGTCGATCCTGTAATCGACCGACAGG 2079A W H R A N A V 0 6 E C V 0 N E F R L I V D Y P L S I L * 443CAACGAAAAAACCCCCCATCGCCCGGTTTTTTCTTCTTCCCCGCAACGCCCCTATAAAGGAAGCGCGTAGGTACCGACCTCGAAT 2164Figure 12: Nucleotide and the deduced amino acid sequence of the oprD gene.The sequence is oriented in the same orientation as the map in Figure 9Band goes from the Bam[H site to the rightmost KvnI site. A typical Shine-Dalgarnosequence appears between nucleotides 723-726 while a prethcted terminator stem-loop appears between nucleotides 2084-2112 (underlined with a dashed brokenline). The accession number in the EMBL data library for this sequence is Z 14065.66bacterial signal sequence:MKVMKWSAIALAVSAGSTQFAVAPolar - -Hydrophobic core 4--iThe 420 amino acid sequence of the mature protein predicted certain typicalfeatures observed for other outer membrane proteins including an overall negativecharge and a typical amphipathicity plot with alternating hydrophobic andhydrophilic stretches (Siehnel et al., 1990)The GC content of the whole gene was 61%, and at the third codon position,it reached 81.8%, which are features typical of a P. aeruginosa gene. The 6nucleotides AAGGAG which were 8 bp upstream of the starting codon ATG werea typical bacterial Shine-Dalgarno sequence, whereas the 29 nucleotides 16 bpdownstream from the stop codon TAA could form a hairpin structure and functionas a transcriptional terminator (Fig. 12).An attempt was made to match the putative OprD sequence to other outermembrane protein sequences obtained from P. aeruginosa and to the OmpF andTo1C porins from E. coli, using the method of Needleman and Wunsch (1970) witha bias parameter of 0 and a gap penalty of 4 with 10 random runs. The alignmentscores obtained were 1.4, -0.5, 1.4,2.6 and 1.1 for P. aeruginosa OprF, OprH, OprPand E. coli OmpF and TolC, respectively. Although OprD showed the highesthomology to OmpF, none of these scores were considered significant above 3standard deviations.67D. Nucleotide Sequencing of the opdE Gene.In an attempt to mutagenize the oprD gene, strain H673 opdE::Tn501 wasisolated by Howard Meadows in our laboratory. The 1VIIC of imipenem for H673 was12 tg/ml, whereas the MIC for the parent strain H103 was 1.5 jig/mi, and SDSPAGE of outer membrane preparations revealed that HM2 was OprD-deflcient(data not shown). The region of the chromosome equivalent to that surrounding thetransposon insertion site was cloned by Eileen Rawling and Richard Siehnel fromthe parent strain H103 into vector pRK767, to create plasmid pD2-45 containinga 6-kb EcoRI/PstI fragment. When mobilized back into strain H673 by triparentalmating by Francis Beffido, pD2-45 was able to complement this strain to imipenemsusceptibility (MIC=1.5 jig/mi) and OprD deficiency (data not shown).It was first assumed that the OprD gene had been cloned. However thesequencing of 3931 base pairs of DNA surrounding the site of transposon insertionfailed to reveal a sequence corresponding to the N-terminal amino acid sequenceof OprD, and no open reading frame equivalent to an outer membrane protein (i.e.containing a signal sequence) was predicted. In addition, no new protein bandswere observed in E.coli containing plasmid pD2-45. Furthermore an N-terminalspecific oligonucieotide failed to hybridize to plasmid pD2-45 and indeed hybridizedwith sequences in the P. aeruginosa chromosome with an entirely differentrestriction pattern (Fig. 9).The DNA sequenced predicted 4 large open reading frames (Fig. 13), 3 of68GAATTCCTCTTCGGGGAGCCAGGCGATTCCCAGCCCCGCCAGCGCGGCATCCACGATATTCGGCGAGGTGTTGAAAATGAGCTGTCCATCGACACGGAC 99F E E E P L W A I G I 0 A I A A D V I N P S T N F I. L a G 0 V R V 33GTTCACGTGTCGATCTTTCCGCTGAAAATCCCAGGCATACAGGCCGCCGCCGGACTGCATGCGCATGTTGATGCAGTTGTGGTCGACCAGATCGCGAGG 198N V H R D K R Q F 0 W A Y L G G 0 S Q N R N N I C N H 0 V I D R P 66ACTCCTGGGCTTCGGATGTGCCGCAAAGTAGGCCGGGGCCGCGACGACCGCCATGCGCACTGGCGGCCCAATCGGCACGGCGATCATGTCCTTGTCTAT 297S R P K P H A A F Y A P A A V V A N R V P P G I P V A I N 0 K 0 I 99GGTGTCGCCCAGGCGTACGCCGGCATCGAACCGGTCGGCCACGATGTCCCGAAAGCCATAGTTGATGTCGAACTCCACCTTGATGTCTGGATATTCCAG 396T 0 G L R V C A 0 F R D A V I 0 R F C Y N I D F E V K I D P Y E L 132CAGCAACG000TGAGCCTGGGTAGCAACAGGGTTCGCTGGATGTGATCGCCACAGGTAATGCGAACCGTGCCACTTGGTTTGTCGCCCAGCGCCGACAG 495I L P T L R P L L L T R 0 I H 0 0 C T I R V T C S P K D R L A S L 165CTCGTCCAGTTCCGCCTCGATCTCGTCGAACCGATTGCCGATGGCATTCAACAGGCGCTCCCCTGCCGCCGTGGGCGAAACGCTGCGGGTGGTGC000T 594E 0 I E A E I E 0 F R N C I A N I L R E G A A T P S V S R T T R T 198GAGTAAGCGGATCTGCAGGCGCGCCTCCAGGCCGCTTATCGACTGGCTCAATGCCGACTGCOTCACGCCCAGTTGGGCGCCGGCACCGGTGAAGGTTCC 693I L R I 0 L R A E L G S I S 0 S L A S 0 T V C L 0 A A A R T F T 0 231CTCGCGGGCGACCGCAACGAAGGACACGAGGTCGTTGAGCTTACGTTTGATCATGGCCAATTCTTCCACGGACCATTAATTAGTAGAGCTTATATACCC 792E R A V A V F S L L 0 N L N R K I N - - - -ORF3 251ATTAAGAATTGTTTAGCTAGTAACAGGCCGCCGCATACCCGAATATTCATTTGCAACAATCATCTCCGCCTACCGCACCCTGACTTCCCTTCCCTGCCG 891ACTCAGCCCAGGCGTTTGTCACCAGCGGAGCTGATCTCTTCTTTTCACTCTTTCGATAAGCCGGTTTTTTCATGACAACCCGCGCACTCGATACCGCCA 990opdE- - - - N T T R A L D T A 263ACGAAAACCCTGAACAATCGGGCTCCTGGAGTGGCGTCCTGGCCATTGCGGTTTGCGCCTTCGCACTGGTCGCGTCGGAGTTCCTGCCGGTCAGCCTGC 1089N E N P E 0 S G S W S C V I A I A V C A F A I V A S E F L P V S L 296TGACTCCCATCGCCAACGACCTGGGAACTACCGA000CATGGCAGGCCAGGGCATCGCCATCTCCGGCGCCTTCGCCGTTTTAACCAGCCTGTTCATTT 1188L T P I A N 0 L C T T E C N A G 0 G I A I S G A F A V L T S I F I 3291287S S V A C S L N R K T L L I G L T A A N G N S C A I V A L A P N Y 362TCGTCTACATGCT000CCGGGCCCTGATCGGCATAGTGATCGGCGGCTTCTGGTCGATGTCGGCACCCACCCCCATGCGCCTGGTGCCTGCCAACGACG 1386F V Y N L C R A L 1 C I V I 0 0 F W S N S A A T A N R L V P A N 0 395TGCCGCGAGCCCTGGCCCTCGTCAATGGCGCCAACGCTCTGGCGACAGTGGTCGCCGCGCCGCTGGGCGCCTGGCTAGGCACCCTCATCGGCTGGCGAG 1485V P R A L A L V N C C N A I A T V V A A P L C A W L C T I I G U R 428GGGCTTTTCTCTGCCTTGTGCCGGTAGCCCTGGTGGCACTGGCCTGGCAATGGACCACCCTGCCCTCCATGC000CCGGCGCGCGTGCTCCcGGCCCGG 1584O A F L C L V P V A L V A L A U Q U T T L P S M R A C A R A P G P 461GCAATGTCTTCACGGTATTCCCTCTGCTCAAGCGTCCCGGTGTGATGCTCGGCATGCTCGCCAGCAGCCTCTTCTTCAT000CCAGrTTTCCCTGrTCA 1683G N V F T V F A L I K R P 0 V M I C N L A S S L F F N G 0 F S L F 494CCTATGTGCGACCATTCCTGCACACGGTCACCGGCGTACATGGCGCGCATGTTTCGCTCGTACTGCTGGTCATCCGTGCAGC000CTTTATCGGCACCC 1782T Y V R P F I E T V T G V H G A H V S I V I I V I C A A G F I G T 527TGCTGATCGACCGGGTTCTGCAACGGCGCTTCTTCCAGACACTCGTCGCCATCCCGTTGCTGATCGCCCTGATCGCCCTGGTACTGACGGTCCTTGGCG 1881I I I 0 R V I 0 R R F F 0 T L V A I P 1 I N A I I A L V I T V I G 560GCTGGCCCGCCATCGTTGTCGTCCTGCTCGGATTGTCGGGACTGACCGGTACCTCGGCCCCCGTCGGTTGGT000CCTGGATCGCCAGGGTGTTCCCAG 1980O W P A I V V V I I C L U G I T C T S A P V C U U A U I A R V F P 593ACCACGCCCAAGCCGGTGGCGCCCTGTTCGTCGCCGTGCTCCAACTCTCCATTGCCCTGGGCTCCACATT000TGGTCTGCTGTTCGATCGCACTGGCT 2079E 0 A E A G G C L F V A V V 0 1 5 1 A I C S T I C C I L F 0 R T C 626ATCAGCCGACCTTCTTCGCCAGCOCCGCGATGCTGCTGATCGCAGCCTTCCTGACCATCCTCACCGCACGCTCCAAAOCCCCCGCCGGCTAGACCCCGG 2178Y Q A T F F A S A A N I I I A A F L T I L T A R S K A P A C * 656GAACGCCCCGACGCGACTTCCCTCGCGCCCCAGGCCAGCTCGTCGAGCCGAATCCCACCACGTCGATCTGATCOATGGAGAACGCCATGGAAACCAAGC 2277ORF2- - -- N E N A N E T K 666ACAGCAATCGAGCTCGCTCTCCCAAGGGTGCCCTGA000GCCCAGTCCTTGCCGGTGCGCTGATCGCTCTCGTCGCCTGCCAGACCAGTCCGGCGGCAA 2376H S N R A R S P K C A I R C A V I A C A I N A I V C C 0 T S P A A 699CGACTTCGTCAAACACCGGAGGAACCAACATGCAGCTGCAATTGACCCAGCAGT000ACAAGACCTTTCCCCTCAGCGCAAAGGTCGAACATCCCAAGG 2475T T S S N T G C T N N 0 I Q I T Q E U 0 K T F P L S A K V E H P K 732TCACCTTCGCCAATCGCTACGGCATCACCCTGGCAGCTGACCTGTACCTGCCGAAGAACCGTCGCCGCGATCGGCTGCCGGCAATCGTGATCGGCGGTC 2574VT F A N R Y CI TI A AOL Y L P K N R G GD RIP Al VI GO 765CGTTCGGCGCGGTCAACGAGCAGTCCTCCGCACTCTATGCGCAAACCATGGCCGAACGCGGATTCGTCACGCTGGCGTTCGACCCATCGTATACCGGTO 2673P F C A V K E 0 S S C I Y A 0 T N A E R G F V T I A F 0 P S Y T G 79869AGAGCGGCGGTCAGCCACGCAACGTCGCTTCGCCGGATATCAATACCGA.AGACTTCAGCGCGGCAGTGGATTTCATCAGTTTGTTGCCGGAAGTGAATC 2772E S G G 0 P R N V A S P D I N T E D F S A A V D F I S L L P E V N 831GCGAGCGCATCGGCGTCATCGGCATCTGCGGCTGGGGTGGCATGGCGCTGAACGCGGTGGCCGTGGACAAGCGCGTCAAGGCGGTGGTGACCAGCACCA 2871R E R I G V I G C G W G G N A I N A V A V D K R V K A V V T S T 8642970N Y D N T R V N S K G Y N 0 5 V T L E 0 R T R T L E 0 L G 0 Q R W 897AGGACGCGGAAAGCGGTACCCCCGCCTATCAGCCGCCCTACAACGAACTGAAGGGTGGTGAGGCACAGTTCCTCGTCGACTACCACGACTACTACATGA 3069K D A ES G T PAY OP P Y NE L KG GE GO F LV DY H DY Y N 930CACCCCGTGGCTACCACCCGCGGGCAGTCAACTCCGGTAACGCCTGGACGATGACCACGCCGCTGTCGTTCATGAACATGCCGATCCTCACCTACATCA 3168T P R G Y H P R A V N S G N A W T N T T P L S F N N M P 1 L T Y I 963AGGAGATCTCGCCACGCCCGATCCTGTTAATCCACGGCGAAAGGGCCCATTCACGCTACTTCAGCGAGACCGCCTACGCCGCTGCCGCAGAGCCAAAGG 3267K E I S P R P 1 1 L I H G E R A H S R Y F S E T A Y A A A A E P K 996AGCTGCTGATCGTTCCGGGAGCCAGTCATGTCGACCTGTACGACCGGCTGGACAGGATTCCTTTCGATCGGATTGCCGGATTCTTCGACGAGCATCTGT 3366E L I I V P G A S H V 0 L Y 0 R L D R I P F D R I A G F F D E H L 1029AGCGTCGTGCACGCCAGGGCAACAGCGCCGGGAGATTGATTCGGNCCGCTCCCCCGCGTCCTGTCGCGCACCTCTCCGGCTTTTTCCGCGCCAGCGAGG 3465*TCCCGCTCCGCTCGAGACCTCGCCCCTCCCTGGCACCCCTTTCAAGCAACCGCCGCCCGCGTCACGATCCCGTCCACCAACCGCGCAATCCCCAATGGG 3564TTGCCATCCTTCAGCGCTTCCGGCAGCAACGCGTCCGGGTAGTTCTGGTAGCACACCGGGCGCAGGAAACGGTCGATGACCAGGGTGCCCACCGAGGTA 3663CCGCGGGCGTCCGAAGTGACCGGGTACGGNCCACCGTGGACCATCGCGTCGCAGACTTCCACACCGGTCGGGTAGCCGTTGAGCAGCAGGCGTCCTGCC 3762TTCTGTTCCAGGAGCGGTACCAGGTCGNCGAAGGACGCCAGGTCTTCCGCTTCGGCGATCAGGGTCGCGGTGANCTGCCCGTGCAGCCCATGCAGCGCG 3861CGCTTCAGTTCGGCGTGGTCGGCGACCTCGACGACCACGCTGGCCGGGCCGTTGACTTCTTCCTGCAG 3929Figure 13: Nucleotide and the deduced amino acid sequence of the opdE gene.The transposon insertion site from H673 is indicated by a solid triangle afternucleotide 1243. The sequence is oriented in the same direction as the map givenin Figure 9A, and goes from the leftmost EcoRI site to the third PstI site. ORF3 wason the complimentary strand and began at base 747, read to the left and proceededbeyond the beginning of this sequence (no stop codon was encountered but wassubsequently identified by sequencing beyond the EcoRI site by R. Siehnel). Theaccession number in the EMBL data library for this sequence is Z14065.70which had a codon usage typical ofF. aeruginosa genes (>80% G+C in position 3 ofcodons). One of these open reading frames overlapped the region of transposoninsertion in F. aeruginosa strain H673 and was thus named opdE (for putativeregulator of QrD expression). This open reading frame was 402 amino acids longwith a predicted Mr of 41,592 (Fig. 13). The sequence was quite hydrophobic, with6 1.3% non polar amino acids, 29.4% uncharged amino acids and only 34 chargedresidues. All secondary structure prediction methods used suggested that thisprotein was an integral membrane protein containing as many as 12 membranespanning -helices. Only 85 nucleotides after the end of the opdE gene, anotherlarge open reading frame (1110 bp, predicted to encode a 370 amino acid protein)was predicted, whereas a third open reading frame of greater than 747 bp(predicted by single stranded sequencing past the EcoRI site to be 978 bp in length)was predicted to be encoded by the other strand (Fig. 13). These sequences, calledorf2 and orf3, might also be involved in OprD expression since no obviousterminator appears between opdE and orf2, suggesting a potential operonstructure. A screen of the EMBL Swiss pro database revealed that the proteinencoded by the opdE gene was homologous to chloramphenicol-resistance proteinsfrom Streptomyces liviclans and Rhodococcus fascian, and the multidrug-resistanceprotein EmrB from E. coli.E. Summary.71The oprD structural gene was cloned as a sequence homologous to an N-terminal specific oligonucleotide probe. The 2.1-kb BamHI/KpnI fragment, clonedin plasmid pBK19R in the same orientation as the lac promoter, revealed highexpression of OprD in E. coli outer membrane. DNA sequencing predicted an openreading frame containing 1,329 bp nucleotides, which encoded a 420 amino acidmature OprD protein with a 23 amino acid signal sequence. The sequence hadcertain typical features observed for other outer membrane proteins. In addition,a putative oprD regulatory gene opdE was sequenced, which predicted ahydrophobic protein of Mr 41,592.72CHAPTER TWO Functional Characterization of OprD: In vivo and In vitroA. Introduction.Previous studies suggested that P. aeruginosa OprD is a specific porin forbasic amino acids and imipenem (Trias and Nikaido, 1990). Regarding its functionin the uptake of carbon sources, it was suggested OprD had signfficant non-specificpermeability to monosaccharides and disaccharides (Yoshihara and Nakae, 1989;Yoshihara et al., 1991). From the literature, there was confficting data about thefunction of OprD in the uptake of fluoroquinolone antibiotics. It was shown thatsome, but not all fluoroquinolone-resistant mutants were also cross resistant toimipenem and lacked OprD. For example, lVlichea-Hamzehpour et al. (1991)demonstrated that decreased fluoroquinolone susceptibility was associated with adecrease or loss of OprD and proposed that OprD can catalyze the facilitateddiffusion of fluoroquinolone as it does for imipenem.Our previous data indicated two gene regions were involved in the expressionof the oprD gene. One turned out to be the oprD structural gene, and the otherregion (the opdE gene) might encode a protein influencing the expression of OprD.In keeping with this hypothesis, the cloned oprD gene in E. coli was expressedpoorly from its owii promoter Fig. 11, lane 8) and the level of OprD observed in theouter membrane of P. aeruginosa was influenced by both the growth medium andcarbon source (Hancock and Carey, 1980). However, all the strains, includingOprD-defective strains, that had been used in prior studies of the in vivo function73of OprD, were genetically undefined, and many were from clinical sources. I believethis was the major cause of unclear and controversial results. This chapterdescribes the construction of a set of isogenic mutants expressing geneticallydefined levels of OprD, and utilizing them to further investigate the substrateselectivity of OprD in vivo.Regarding the in vitro functions of OprD, very limited work has been doneto study the physical properties of OprD in the black lipid bilayer system. Ishui andNakae (1993) measured the single channel conductance of OprD, which was 20 to30 Ps. In addition, they observed larger channels (400 pS). Furthermore, the ionselectivity of the OprD channel was unknown. In addition, no direct evidence hadbeen obtained from black lipid bilayer studies to prove the presence of a specificbinding site(s) for imipenem within the OprD channel. This chapter describes thepurification of OprD and a thorough analysis of its in vitro functions in the blacklipid bilayer system.B. Construction of a Defined OprD Defective Mutant H729.The improved method of Schweizer (1992) for allele replacement wasutilized to replace the wild-type oprD gene in strain H103 with an oprD::Qinterposon-mutated gene (see Material and Methods). Southern hybridization ofchromosomal DNA with a32P-labelled oprD gene probe confirmed that in strain11729, the oprD chromosomal gene was interrupted by a 1.3-kb Sail fragment74containing the Km’42 interposon (Fig. 14). In the Southern blot, a 4.5-kb band waspresent in both the mutant and parent strains (Fig. 14), indicating there wasanother gene with high homology to the oprD gene in P. aeruginosa, which mightbe the oprE gene (Yoshinori et al., 1993). Sodium dodecyl sulfate-polyacrylamidegel electrophoresis (SDS-PAGE) of the outer membrane proteins of H729 confirmedthe lack of OprD (Fig. 15, lanes 8, 9).C. Overexpression of the oprD Gene in P. aeruginosaTo overexpress the oprD gene in P. aeruginosa, the oprD gene was subclonedinto plasmid pUCP19 to form plasmid pXTI2, so that the direction of expression ofthe oprD gene was in the same orientation as the lac promoter (Fig. 5). PlasmicipXH2 was then mobilized back into the P. aeruginosa wild type strain 11103 andthe oprF:: mutant strain 11636. Since P. aeruginosa does not have the lacrepressor gene, IPTG was not added to the medium. SDS-PAGE of the outermembrane proteins demonstrated that in strain H103(pXH2), OprD was expressedto a level almost equivalent of that ofF. aeruginosa major outer membrane proteinOprF (Fig. 15, lane 4, and in H636(pXH2), OprD appeared to be the predominantouter membrane protein (Fig. 15, lane 7).Plasmid pXH2 was also transferred into the OprD defective strain H729. Theresults indicated that the loss of OprD could be complemented and that the oprDgene was also overexpressed in H729(pXH2) (Fig. 15, lane 10).75EcoRI XhoI1< 1.1kb >1probeEcoRI olsail XhoI SailB2.4kbC 1245—.Figure 14: Construction of an OprD defective mutant in P. aeruginosa.Genomic Southern hybridization fflustratmg the interruption of the oprDgene (shaded bar) with a kanamycin resistance L-interposon (black bar). Thephysical maps of the wild type (A) and the mutant (B) oprD gene region are shown.Genomic DNAs were digested to completion (C) with SalT and hybridized to the 32P-labelled 0.3-kb SalI-EcoRI fragment shown in panel A. Lanes: 1. H729; 2. 11103.The molecular sizes of the fragments (in kb) are indicated on the left.761 2 3 4 5 6 7 8 9 1096#67!43 ;;30,20.24Figure 15: SDS-PAGE demonstrating overexpression and mutagenesis of theoprD gene.The banding position of OprD is indicated by the arrow on the right. Lanes:1, molecular marker; 2, H103; 3, H1O3(,pUCP19); 4, H103(pXH2); 5, 11636; 6,H636(pUCP19); 7, H636(pXH2); 8, H729; 9, H729(pUCP19); 10, H729(pXH2). Foreach lane, 20 tg outer membrane proteins were added.— m.’.. —— •—•——— E—-OprDI iOprF<— OprL77D. Function of OprD in Antibiotic SusceptibilityTo reexamine the role of OprD in quinolone uptake and also to confirm, ingenetically defined strains, that it was a specific pore for imipenem, MICdeterminations were performed by the agar dilution procedure in Mueller-Hintonmedium (Table IV). Nine strains which represented different expression levels ofthe oprD and oprF gene were utilized. The rationale behind this experiment wasthat, if OprD could act as channel for certain antibiotics, the overexpression of thisporin would increase susceptibility whereas the lack of the porin would decreasesusceptibility to these antibiotics; otherwise, the amount of OprD would notinfluence antibiotic susceptibility.Two carbapenems were used, imipenem and meropenem, MICs for wild typestrain H103 were 4.0 ji.g/ml and 0.5 jig/ml respectively. The OprD-overexpressingstrain H103(pXH2) showed the lowest MICs, 0.5-1.0 jig/ml for imipenem and 0.06-0.12 tg/ml for meropenem, which were 4 to 8-fold lower than that of 11103. Incontrast, the OprD-defective strain 11729 showed the highest MIC, 16 jig/ml forimipenem and 2.0 for meropenem, which were 4-fold higher than that of H103CPable IV). The overexpression of OprD from plasmid pXH2 in the OprD defectivestrain 11729 restored susceptibility to levels equivalent to those observed inH103(pXH2). In contrast, the loss by mutation of OprF did not influence the MICsto carbapenems, indicating that OprF did not function as a major uptake route forcarbapenems (Table IV).78TableIV:InfluenceofOprDexpressionlevelsonantibioticsusceptibilityofP.aeruginosastrains.StrainsRelevantpropertiesaMICs(tg/ml)”OprDOprFIMIPMEROCTXCFPCIPFLRGMH103++4.00.504.,pUCP19)+-4.00.508.,pXH2)+++- EachMICwasdeterminedthreetimesindependently. Abbreviations: IMIP, imipenem; MERO,meropenem;CTX,cefotaxime;CFP, cefpirome;CIP, ciprofloxacin;FLER, fleroxacin;GM, gentamicin.Other p-lactams (cefotaxime and cefpirome), quinolones cciprofloxacin andfleroxacin) arid an aminoglycoside (gentamicin), did not show significant differencesin MICs, regardless of the expression levels of OprD (Table IV). (N.B. the 2-4 foldincrease in susceptibility to cefotaxime and cefpirome of strains containing theplasmids pUCP19 or pXII2, was due to the plasmid encoded -lactamase). Theadditional absence of OprF had little effect on antibiotic resistance. These resultsindicated that OprD did not significantly facilitate the passage of these antibioticsacross the outer membrane.We also examined the influence of OprD expression on antibioticsusceptibility of E. coli. The oprD gene was overexpressed to high levels in E. coliCE l248(pBK19R), a strain with mutations preventing the production of porinsOmpF, OmpC and PhoE. However, even in this background, MICs for thecarbapenems were much lower than those for P. aeruginosa strains andoverexpression of OprD had no effect (Table V), presumably due to the higherintrinsic outer membrane permeability of E. coli.E. Function of OprD in Sugar TransportSpecific porins, with the exception of the sucrose porin of E. coli (Schülein etal., 1991), are generally poorly permeable to non-specffic substrates. To investigatethe role of OprD in sugar transport, a set of isogenic strains: P. aeruginosa 11729,H 103, and H103(pXH2), expressing different levels of OprD were used. Control80Table V: Influence of OprD expression levels on antibiotic susceptibility of E.coli strains.MICs(JIg/ml)aStrains OprD level IMIP MERO CTX CFPCE1248- 0.06 0.016 0.16 0.16CE1248(pTZ19R)- 0.03 0.016 0.16 0.64CE1248(pBK19R) ++ 0.03 0.016 0.16 0.64a. MICs were determined by the agar dilution method on Mueller-Hintonplates. Each MIC was determined three limes independently. Abbreviations:IMIP, imipenem; MERO, meropenem; CTX, cefotaxime; CFP, cefpirome.81experiments demonstrated very similar rates of growth for all three strains onLuria Broth (growth rate 1.26 doublings per hour of all three strains) and MuellerHinton broth (growth rate 1.10-1.23 doublings per hour). Therefore, strains weregrown in the BM2 minimal medium with either glucose, gluconate or pyruvate ascarbon sources, at concentrations varying from 0.5 mM to 10 mM. Growth curveswere constructed for each sugar at each concentration and utilized to calculategrowth rates. We reasoned that, if OprD could facilitate the transport of a certainsugar, at the growth rate limiting concentrations, different growth rates would beobserved depending on the oprD gene expression level.For gluconate, at 0.5 mM concentration, the growth rates of wild type strainH103 and OprD-defective strain H729 were only 60% and 20% respectively, of thatof OprD-overexpressing strain H103(pXET1), and these differences were statisticallysignificant (p <0.05 by Student’s t test) (Fig. 16A). These data were thus consistentwith outer membrane permeation being the rate limiting step for growth ongluconate, and further suggested OprD was the major porin involved in gluconatepassage across the outer membrane at growth-limiting conditions.As the initial concentration of gluconate in the medium was increased, thegrowth rates for H 103 and H729 increased, whereas the growth rate of the OprDoverexpressing strain H103(pXH2) remained stable. The growth rates of the threestrains converged with increasing saccharide concentrations and eventually becamenot significantly different at 10 mM gluconate. This result indicated that outermembrane permeation through OprD ceased to become rate limiting at high82ABC1.20—.—s::. 0.60 I::./__________IIIIIIIII0.0024681012024681012024681012Concentration(mM)Figure16:FunctionofOprDinsugartransport.InfluenceofsubstrateconcentrationsonthegrowthrateofH103(circle),itsOprD-overexpressingstrainH103(pXH2)(square)andOprD-defectivestrain11729(diamond).(A) gluconate;(B) glucose, (C)pyruvate. Datapointsaretheaverageresultsfrom3experimentswithlessthanorequalto10%variationbetweenexperiments.saccharide concentrations, at which other porins could substitute for OprD (Fig.16A).To further exclude the possibility that the slow growth rate of the OprDdefective strain H729 had resulted from metabolic disturbances due to themutation, another carbon source, glucose, was used as a control. At 0.5 mM glucoseconcentration, the growth rates for H103 and H729 were 110% and 90% that ofH103(pXH2), which were not significantly different (p>0.5). As the glucoseconcentration was increased, the growth rates for all three stains increased butremained very close for the three strains (Fig. 1GB). The results were reasonablesince, as previously demonstrated, in the presence of glucose, P. aeruginosa strainsinduce a specific porin OprB (Hancock and Carey, 1980). The OprB levels in allthree strains were found to be the same, and approximately 5 fold lower than thelevel of OprD in strain H103(pXH2). These results indicated that loss oroverexpression of OprD did not affect the normal growth of the cell.Since growth in pyruvate leads to higher expression levels of OprD in P.aeruginosa (Hancock and Carey, 1980), it was questioned whether OprD could alsoallow the specffic passage of pyruvate. The results showed no significant differencein the growth rates of P. aeruginosa Hl03(pXII2), Hl03 and H729 at pyruvateconcentrations from 1 ml\’l to 10 mM (Fig. 16C), indicating that OprD was not ableto function as the predominant channel for the transport of pyruvate.F. Competition Experiments.84Since OprD can facilitate the permeation of basic amino acids (Trias andNikaido, 1990b) and the above suggested a role in uptake of gluconate andzwitterionic carbapenems, competition experiments were performed as describedby Fukuoka et al (1991) to determine if common binding sites were shared by thesesubstrates. The susceptibilities of the isogenic variants to imipenem weredetermined using BM2 medium supplemented with basic amino acids or gluconatearid were compared with the results obtained in unsupplemented BM2 glucose orsuccinate respectively (Table VI). The MICs for H103 and H103(pXH2) wereincreased 8-16 fold and 4-8 fold respectively by the addition of 50 mM basic aminoacids. However only a 2 fold effect was observed for the OprD-deficient mutant11729. Such a change in 1VIIC is usually considered to be not significant. In addition,the effect of L-lysine concentrations in BM2 glucose medium on the susceptibilitiesof the isogenic variants to imipenem was determined. Figure 17 showed that theMICs for H103 and H103pXH2) increased as the concentration of L-lysineincreased, until they reached the same level as the OprD-deflcient mutant H729.In contrast, the susceptibility of the OprD defective strain H729 was notsignificantly influenced by the addition of basic amino acids. The results suggestedthat imipenem and basic amino acids shared a common binding site(s) in OprDchannel.The same competition experiments were also performed with gluconate orglucose at the concentrations of 20 mM, 50 mM, 100 mM and 150 mM (Table VI).In contrast to the resuIts for the basic amino acids, the susceptibilities of 11103 and85TableVI:Effectsofbasicaminoacidsandgluconateonimipenemsusceptibilitiesof P.aeruginosastrainsexpressingdifferentlevelsofOprD.MICs(jig/mi)indifferentmediaaStrainsOprDBM2’+BM2”+BM2b+BM2C+BM2C+levelsBM2”50mM50mM50mMBM2C20-150mM20-150mML-lysineL-arginineL-histiclineGluconateGlucoseH729- EachMICwasdeterminedthreetimesindependently.b.GlucosewasusedasthecarbonsourceinBM2media.c.SuccinatewasusedasthecarbonsourceinBM2media.20i10— 1-00:i /I0.1 I I0 50 100 150 200 250Lysine (mM)Figure 17: Competition between L-lysine and imipenem for the OprD channel.Effect of L-lysine concentration in BM2 glucose medium on the susceptibilityof 11103 (open circle), its OprD-overexpressing strain H103(pXII2) (triangle) andOprD-defective strain 11729 (filled circle) to imipenem.87Hl03(pXH2) were not affected by the presence of gluconate or glucose (the 2 foldincrease in the presence of gluconate was also observed for the negative control11729), indicating that no common OprD binding sites were involved for gluconateand imipenem.G. Purification of OprD.OprD was purified from both E. coli and P. aeruginosa following theprocedures described in Materials and Methods. For the purification of OprD fromE. coli, the porin deficient strain CE 1248 containing plasmid pBK19R was grownin the presence of glucose and IPTG to repress the expression of LamB and inducethe high levels of expression of OprD. Similarly, to purify OprD from P. aeruginosa,11636(XH2) overexpressing OprD and lacking major porin OprF was used. Cellswere grown in glucose to avoid the contamination of OprB (Dl). Differentcombinations of salt and detergent concentrations were tested to optimize thesolubilization conditions. OprD was maximally extracted with 10 mM Tris-HC1, 3%Octyl-POE and 0.1 M NaCl in the presence of 0.5 mM EDTA (Fig. 18, lane 5). Whensolubilized at room temperature prior to electrophoresis (Fig. 18, lane 6), OprD ranat a lower apparent molecular weight, demonstrating OmpA-like heat modifiability.When applied to an anion exchange column (Mono Q, Phamacia), thedetergent/EDTA soluble protein eluted in two major peaks (Fig. 19A). The firstpeak was the unbound proteins washed out before applying the salt gradient (Fig.8896-..67433020___ ___Figure 18: SDS-PAGE of samples from solubilization stages.Lanes: 1, molecular weight markers; 2, outer membrane; 3, 0.5% octyl-POEsoluble fraction; 4, 3% octyl-POE/NaCl soluble fraction; 5 & 6, 3% octylPOE/NaCIIEDTA soluble fraction; 7, 3% octyl-POEINaC1/EDTA insoluble fraction.The sample of lane 6 was unheated. Solid triangle indicates the banding positionof the heat-modifiable form of OprD.1234567—— —E—OprDa-i—89A 0.5 100I A A // \ I0.1 I 20II0 010 20 30 40 50Fraction numberB1234966743 EOprD—43O2OFigure 19: Purification of OprD.(A) Mono-Q FPLC elution proflIe. Buffer A: 10 mM Tris-HCJ (P118.0), 5 mMEDTA, and 0.08% LDAO. Buffer B: buffer A plus 1.0 M NaC1. Broken line indicatesthe salt gradient and numbers refer to peaks 1 and 2. (B) SDS-PAGE of samplesehited form Mono-Q column. Lanes: 1, peak 1 sample; 2, peak 2 sample; 3, purifiedOprD (heated); 4, purified OprD (unheated). The solid triangle indicates thebanding position of the heat-modifiable form of OprD.90l9B, lane 1). OprD was eluted in the second peak along with traces of otherproteins of higher molecular weight (Fig. 19B, lane 2). Those fractions were pooledand subjected to a second run with a much flatter salt gradient and lower elutionspeed, OprD was eluted as a single purified protein (Fig. 19B, lane 3). The purifiedOprD stifi retained its heat-modifiability characteristic (Fig. 19B, lane 4) and couldform functional channels in the black lipid bilayer system (see below).H. Black Lipid Bilayer Analysis.The purified OprD protein was added at nanomolar concentrations to theaqueous solution bathing a black lipid bilayer membrane. Membrane conductanceincreased in a stepwise fashion (Fig. 20A), presumably due to the incorporation ofindividual porin units into the membrane as suggested for other porins (Benz andHancock, 1981). For 170 measured single-channel events, the average singlechannel conductance in 1 M KC1 was 20 pS, which was at least 10 times smallerthan those of most other porins studied to date. The only exception was the E. colinucleotide-specffic porin Tsx, with the average single channel conductance of 10 Ps.The purified OprD from E. coli and P. aeruginosa did not show any differences insingle channel conductance, indicating that OprD expressed from the cloned genewas properly folded in the E. coli outer membrane. Ishii and Nakae (1993)demonstrated occasional “open” channels of OprD with a much higher conductivity(400 p5), especially in the presence of LPS. However, I did not observe any of such91Figure 20:I.Chart recording of stepwise increase of the membrane current formedby 1% oxidized cholesterol in n-decane in the presence of purifiedproteins.(A) Native OprD. (B) OprDAL5. The aqueous phase contained 1 M KC1,PH7.0, the temperature was 20°C and the applied voltage was 20 mV. Note that theresolution of the instrument was higher in (A) than in (B).ABHi__I4iii1 Iii_,, 1.4111100 Psf500 Ps1i if i I1U ii 4!1E1 :r.fi +tI1[J1, IiI .1jj jUIj :j[i ; 1 liiiII i111 IM :FtI: ii ‘iii Ii?_— 1Iii lit II I ijtli j1HiI 1HS —— •.- I__ —itt if ii, :.I I liiii J j’ ijiliii92events and one possibility for such large channels might be the contamination ofother porins in their preparation.To examine the ion-selectivity of the OprD channel, single channelconductance was measured in salts of varying cation or anion sizes (Table )Q.Increasing the size of cations from K to Cs to Li (the last being highly hydrated)while keeping Cl- as the anion resulted in a steady decrease in the average singlechannel conductance (Table X). In contrast, the single channel conductance waslittle affected by changing the size of anion from Cl- to MOPS- (Table X). The resultssuggested that OprD channel was cation selective. Consistent with this, zerocurrent membrane potential measurements confirmed that OprD channelexhibited 2.6-fold preference for cations over anions (Table XI). The weak cationselectivity of OprD channel was in good agreement with its preference for basicamino acids. Increasing the salt concentration (3.0 M KC1) did not result in a linearincrease in conductance (Table X). A similar result for E. coli hemolysin (Benz etal., 1989) has been interpreted as due to the surface charges at the pore mouthwhich caused a substantial sur[ace potential. For OprD, these charges would beassumed to be due to anionic (acidic) amino acids that would tend to attract cationsand repel anions.To demonstrate that the OprD channel possessed specific binding site(s) forimipenem, a macroscopic conductance inhibition experiment was performed. Largebilayer membranes (2 mm2)were formed in 1 M KC1. A small amount of purifiedOprD was added to one side of the membrane, and the conductance started to rise93rapidly for 10-40 mm, and thereafter continued to rise at a decreasing rate. At thistime, membrane conductance had increased 2-3 orders of magnitude and morethan 1,000 channels were present in the membrane. Aliquots (60 tl) of imipenemsolution (20 jil’1) were added to the aqueous solution at both sides of the membrane,and the conductance decreased to a new level over a period of about 2 mm (seeChapter 4, Fig. 31). The ability of imipenem to block KC1 movement provided directevidence that OprD contained imipenem binding site(s). In addition, by plotting thedata as % inhibition of conductance as a function of imipenem concentration, it waspossible to derive an I5 value (i.e., a concentration of imipenem resulting in 50%inhibition of the original conductance) of 1.4 jiM.I. Summary.A set ofF. aeruginosa isogenic strains with genetically defined levels of OprDwere constructed and utilized to characterize the in vivo substrate selectivity of thisporin. To determine the role of OprD in antibiotic uptake, nine strains representingdifferent levels of OprD and OprF were used to determine the MICs of differentantibiotics. The results clearly demonstrated that OprD could be utilized byimipenem and meropenem but, even when substantially overexpressed, could notbe significantly utilized by other p-lactams, quinolones or aminoglycosides. To testthe function of OprD in the transport of carbon sources, strains were grown inminimal medium with limiting concentrations of the carbon sources, glucose,94gluconate or pyruvate. The results indicated that OprD selectively facilitated thediffusion of gluconate under growth-rate limiting conditions. In contrast, it did notfunction as the predominant channel for the transport of glucose or pyruvate.Competition experiments confirmed that imipenem shared common binding siteswith basic amino acids in the OprD channel, but not with gluconate or glucose. Inaddition, OprD was purified and was able to reconstitute channels in black lipidbilayer model membranes. OprD formed very small pores with an average singlechannel conductance in 1.0 M KC1 of 20 Ps, and the channel was weakly cationselective. When large numbers of OprD channels were incorporated into lipidbilayer membranes, addition of imipenem resulted in progressive decrease inmembrane conductance, indicating the presence of specific binding site(s) foriniipenem in the OprD channel. This allowed the calculation of an I value of 1.4M.95CHAPTER THREE Structural Characterization of OprD: MembraneTopology ModelA. Introduction.The crystal structures of 5 porins was a milestone in our understanding ofporin functions. To fully understand the molecular mechanism involved in thefacilitated uptake of basic amino acids and imipenem, a detafled knowledge of themolecular structure of OprD is required. The amino acid sequence of OprD (Fig. 12;Yoneyama et al., 1992) was typical of porins: charged residues were distributedalmost uniformly along the primary sequence and as a consequence there were noclear hydrophobic stretches which would be predicted to span the membrane as analpha helix. Therefore it is very likely that the structured transmembrane segmentsare essentially composed of p-strands. In this chapter, the first OprD topologymodel was constructed by multiple alignments together with secondary structurepredictions. PCR-mediated site-directed mutagenesis was then employed toseparately delete the predicted external loops and to verify the accuracy of themodel.B. Prediction of an OprD Topology Model.In a previous paper (Jeanteur et al., 1994a), the alignments of 30 non-specfficporins from 5 distant families were reported. Alignment of OprD was not96considered in detail. However attempts to match the OprD sequence with other P.aeruginosa porins and the E. coli porins OmpF and To1C showed that OprD had thehighest homology to OmpF with an alignment score using the Needileman andWunsch method (1970) of 2.6, which was close to 3.0, the minimal score requiredfor an alignment to be considered significant. Based on the sequence alignmentbetween OmpF and OprD, sixteen p-strands were predicted (Fig. 21). Alignmentwas very clear for the N and C terminal p-strands, but the homology was weakerin the middle part of the sequence, similar to that reported for other porins(Jeanteur et al., 1991). The 16 transmembrane segments had the typicalamphipathic features of porin p-strands in that they were composed of alternatingpolar and non-polar residues exposed to the aqueous channel and hydrophobicmembrane interior respectively (Fig. 22). The sizes of the predicted p-strands(10’’21 residues) were in agreement with the lengths of p-strands observed for thesolved porin structures, and the ends of the these p-strands were often composedof aromatic residues, which may function as one of the stabilizing forces for thebarrel structure (Cowan et al., 1992).The n-strands were connected by short turns at the periplasmic side and bylong loops at the cell surface. Consistent with the larger number of amino acids inOprD than in the other members of the porin superfamily, the 8 external loops wereoften slightly longer than the ones observed for the known porin structures. Thepredicted loop L3 (Q13 to T165) was a long loop (Fig. 22). It was the longest externalloop observed in the E. coli porin OmpF and PhoE, R. capsulatus and97OprD DAFVSDQAEAKGFIEDSSLDLLLRNYYFNRDGKSGSGDRVDWTQG 45OmpF AEIYNKDGNKVDLYGKAVGLHYFSKGNGENSYGGN GDMTYARLG 44OprD FLTTYESGFTQGTVGFGVDAFGYLGLKLDGTSDKTGTGNLPVMNDGKP 93I IIOmpF FICGETQINS DLTGYGQWEYNFQGNNSEGADAQTGNKTR 82OprD RDDYSRAGGAVKVRISKTMLKWGEMQPTAPVFAAGGSRLFPQTATGFQ 141II::: I: I :OmpF LAFAGLKYADV GSFDYGRNYGVVYDALGYT 112OprD LQSSEFEGLDLEAGHFTEGKEPTTVKSRGELYATYAGETAKSADFIGG 189II I I :: I:OmpF DMLPEFGGDTAYSDDFFVGRVGGVATYRNSNFFGLVDGLNFAVQYLGK 160OprD RYAITDNL SASLYGAELEDIYRQYYLNSNYTIPLASDQSLGFDFNIY 236I I: :: :QmpF NERDTARRSNGDGVGGS I SYEYEGFGIVGAYGAADRTNLQEAQPLGNG 208OprD RTNDEGKAKAGDISNTTWSLAAAYTLDAHTFTLAYQKVHGDQPFDYIG 284I: : I: :: :: :1OmpF KK AEQWATGLKYDANNIYLAANYGETRNATPITNKF 244QprD FGRNGSGAGGDS I FLANSVQYSDFNGPGEKSWQARYDLNLASYGVPGL 332I :::::I:: I: IIOmpF TNTSGFANKTQDVLLVAQYQFDF GL 269OprD TFMVRYINGKDIDGTKMSDNNVGYKNYGYGEDGKHHETNLEAKYVVQS 380I: I IOmpF RPSIAYTKSKAKDVEGIGDVDL VNYFEVGATYYFNK 305OprD GPAKDLSFRIRQAWHRANADQGEGDQNEFRLIVDYPLS IL 4201:1 : : I::OmpF NMSTYVDYIINQIDSDNKLGVGSDD TVAVGIVYQF 340Figure 21: Sequence alignment between OprD and OmpF.Dashed lines indicate identical amino acids and ‘ indicates conservativesubstitutions. The known a-strands of OmpF are marked by and thepredicted n-strands of OprD are marked by____98Figure 22: Membrane topology model of OprD.The sixteen predicted transmembrane n-strands are boxed, and the 8external loops are labelled as Li to L8. The deleted amino acid residues arepresented as unfilled letters.99L2L3L5L7IEMvIilNIFKNLITNNSAGVLIGKNDSGTPTDL6QHIFID)GRRDKLFGT80DTYS250KNL8llKDiNNAGYF140ENSDY____GGNTRGN340GID)RGSS160FTGG__TK__L4DwFDyGNLSTRAELGGA100TFsGS___EAQLDGTNAG23LI1VpRNP20RLGTDLAyFMGHELvKAAPLFKwF40LvFVTLQAF280A30THAR41SPswG70KLKIy‘7pNLHL26PTLvRSAGSTQSGEREQYRS130RYAALflVP33ITJVIGGiGG170R1,ELPGQVI N1RP37390FFFsGEGLAHVGFGLAI(ALGEIHVSVSKTPTAVIDT220TKPSIAILLA10•TVMFAFVFQFAKPSEGLVTPv 210NTVNLlylK50PyAA___FKLAG310NVIAL42EG60wAASRN270GLV]pVG12oGKE__PEANKPQGySVSAFTE___T18032380CF”TQGMQVQARRhoclopseudomonas blastica porin crystal structures, in which it completely foldedinside the pore forming the constriction zone for general porins and part of theconstriction zone for the specific porin LamB (Cowan et al., 1992; Weiss and Schulz,1992; Kreusch et al., 1994; Schirmer et al., 1995).C. PCR-Based Site Directed Deletion Mutagenesis.To test the validity of the predicted external loops, site-specific deletionmutagenesis was performed to separately delete short stretches of amino acids (4-8)from each of the predicted loop to see if these deletions were tolerated. Thedeletions were made around the middle of the predicted loops, as shown by theunfilled letters in Fig. 22. For the shorter loops, L4 and L6, only 4 amino acids weredeleted, and for the remaining six longer loops, 8 amino acids were deleted. Therationale was that the external loops can undergo substantial variation withoutaffecting the configuration of the protein. In contrast, p-strands buried in themembrane are more conserved and more sensitive to deletions or insertions(Jeanteur et al., 1991). If the predicted loops were correct, the consequent deletionmutant protein would retain the native conformation and assemble correctly in theouter membrane. Otherwise, if the deletion happened in the transmembraneregions, the protein would lack one of the major stabilizing forces of the structureand the mutant protein could not assemble properly in the outer membrane.Early methods for site-directed mutagenesis using single stranded DNA gave101low efficiencies of obtaining the desired mutation. The development of PCR (Saikiet al., 1985), however, provided a new approach (Vallette et aL, 1989). Thesemethods use primers bearing the mutations, which, after PCR, were thenincorporated into the PCR products. By using an appropriate strategy, the mutationfrequency could reach 100%. In this work, 2 PCR strategies, direct extension andoverlap extension were used as described in Material and Methods (Fig. 6, TableIII).Since all of the restriction enzyme sites used in the PCR/cloning procedureshad to be unique, plasmid pBK19R was modified to eliminate the unnecessaryrestriction sites. Plasmid pMBK19R was constructed to eliminate the restrictionsite PstI in the multiple cloning site (Fig. 7), which would interfere with themutagenesis of the predicted loops L3, L4 and L5. Plasmid pMBE 19R, containingonly the N-terminus of the oprD coding region, was constructed to eliminate oneSail site in the multiple cloning site, and another Sail site and one EcoRI site inthe C-terminus of the oprD gene (Fig. 8). After PCR-mutagenesis of the predictedloops Li and L2 using pMBE19R as the template, the 1.2-kb EcoRI fragmentencoding the C-terminus of OprD was then cloned back in the correct orientationto complete the oprD coding region. E. coli CE 1248 containing pMBK19Rexpressed OprD at the same level as E. coli CE1248(pBK19R), whereas E. coliCE1248(pMBE19R), containing only part of the oprD gene, did not express OprD(data not shown).The major limitations of PCR were the unspecific products and the102unexpected mutations (“errors”) generated by the DNA polymerase. To obtain agood yield of the desired PCR products, and to minimize the error and unspecificproducts, I optimized the conditions with respect to DNA polymerase, templateamount and thermal cycling conditions. First, instead of Taq DNA polymerase, thehigh-fidelity thermophilic Vent DNA polymerase was utilized. The fidelity of theVent DNA polymerase was 5 to 15 fold higher than that of Taq DNA polymerase,due in part to an integral 3’-5’ proofreading exonuclease activity. Second, differentamounts of DNA template ranging from 1 pg to 100 ng were tested. Ten ng turnedout to be the optimal amount, since insufficient yield was obtained when less than10 ng of template DNA was used and unspecific product was present when toomuch template was used. Finally, I varied the annealing temperature (50-60°C),used a short extension time (1-4.5 mm instead of 3 mm) and fewer cycles of PCR(20 cycles instead of 30) to minimize errors.After inserting the mutagenic PCR products back into the parental plasmid,a simple and rapid primary screening was utilized to identify the desired deletionmutants. Plasmids were isolated from the transformants and digested with thesame pair of restriction enzymes used in the PCR/cloning procedure to generate thefragments of interest. The digestion mix was analyzed on 1.5—’2.0 % agarose gels,small but readily observable differences were noted between the correspondingfragments containing the desired deletion (12’-’24 bp) and the original gene (datanot shown). From the mutagenesis of each predicted loop, 5 mutant plasmidscontaining the deletions were selected and the whole PCR-amplified regions were103sequenced. The exact amino acid positions of deletion and the identities of thedeleted amino acids from 8 deletion mutants are summarized in Table VII.D. Characterization of the Deletion MutantsThe expression of deletion mutant OprD derivatives was examined in theporin deficient strain E. coli CE 1248. The outer membranes containing the deletionmutations were isolated and examined by SDS-PAGE (Fig. 23A, and on Westernimmunoblot using an anti-OprD polyclonal antiserum (Fig. 23B). The mutantpolypeptides from the deletion mutagenesis of six predicted loops, Li, L2, L5, L6,L7 and L8, co-fractionated with the outer membranes, were typically heatmodifiable, and were expressed at similar levels compared to cells expressing wildtype OprD (Fig. 23A, lanes 4, 5, 8-11). They also showed a slightly increasedelectrophoretic mobility as compared to wild type OprD, consistent with thedeletion of a few amino acid residues. These results indicated that deletions of shortstretches of amino acids in these six predicted loops did not substantially changethe native conformation of OprD, such that the consequent mutant proteins wereassembled into the outer membrane, a result suggesting that these loops wereaccurately predicted. The deletion of the predicted loop L3 caused diminished butobservable expression (Fig. 23A, lane 6), as confirmed by Western-immunoblotresults (Fig. 23B, lane 6), indicating that this deletion was tolerated. However,since the deleted stretch had 4 negatively charged residues (Fig. 22), which could104TableVII:Deletionmutagenesisofthepredictedloops.PlasmidsDeletionAminoacidsMutagenesisE.coliP.aeruginosasitesadeletedefficiencypHE1pHP1Tyr-26TyrTyrPheAsnArgAspGLyLys60%pHE2pHP2Asn-84AsnLeuProValMetAsnAspGly80%pHE3pHP3Glu-146GluPheGluGlyLeuAspLeuGlu100%pHE4pHE4Asn-196AsnLeuSerAla90%pHE5pHE5Asp-240AspGluGlyLysAlaLysAlaGly100%pHE6pHP6Asn-288AsnGlySerGly100%pHE7pHP7Met-349MetSerAspAsnAsnValGlyTyr100%pHE8pHP8Asn-398AsnAlaAspGlnGlyGluGlyAsp100%a.Position1isthestartsiteofthematureOprD, seeFigure22foractualaminoaciddeleted.I. CA1 2 3 4 5 6 7 8 9 10 119643 — — — —— - — — — — —— — — — —B9667123456743 — — —48 9 10 11— — — — — OprDFigure 23: Expression of OprD derivatives in the outer membrane of E. coli.(A) SDS-PAGE and (B) Western-immunoblot. The banding position of OprDis indicated by an arrow on the right. Lane 1 contained molecular weight markers.Lanes 2’-ll contained outer membranes from CE 1248 cells containing the followingplasmids: lane 2, pMTZ19R; 3, pMBK19R; 4, pHE1; 5, pHE2; 6, pHE3; 7, pHE4; 8,pHE5; 9, pHE6; 10, pHE7; 11, pHE8. For each lane, 20 jig of outer membraneprotein was loaded.— — —— u— OprDN106be important for protein folding, the deletion may have perturbed the OprDstructure sufficiently to lead to reduced protein production or unstable products.The deletion of the originally predicted loop L4 did not permit stable expression ofan OprD protein (Fig. 23, lane 7). The deletion may have involved atransmembrane segment or a much less flexible turn. This and the potential foralternative positioning of p-strands 7 and 8 led to a modification of the model(Appendix, Fig. 34).To confirm the above conclusions based on the deletion mutagenesis, theexpression of these OprD derivatives was examined in the native host P.aeruginosa. AU of the mutant oprD genes were subcloned in the same orientationas the lac promoter into the shuttle plasmid pUCP19 (Schweizer, 1991). Therecombinant plasrnids were then transformed into the P. aeruginosa OprD-defectivestrain H729. Examination of plasmid-encoded -lactamase levels indicated nosignificant difference (p > 0.5) in -lactamase levels for any of the transformants(Table VIII), suggesting that the plasmids were present in similar copy numbers.SDS-PAGE (Fig. 24A) as confirmed by Western-immunoblot (Fig. 24B) analysisshowed the same proffle as observed in E. coli, with the exception of the mutantdeleting 8 residues of predicted loop L3 for which no expression was observed (Fig.24, lane 8), further confirming that six of the predicted loops were accurate. Theloop L3 mutant grew much slower than the remaining mutants.To exclude the possibility that lack of expression of the OprD derivativeswith deletions in the predicted loops L3 and L4 was due to the blocked transport107A1 2 3 4 5 6 7 8 9 10 11 121396 .-6743;3O —2O — —B1 2 3 4 5 6 7 8 9 11 12 139667— OprIJ302014Figure 24: Expression of the OprD derivatives in the outer membrane of P.aeruginosa OprD-defective strain 11729.(A) SDS-PAGE and (B) Western-immunoblot. The banding position of OprDis indicated by an arrow on the right. Lanes: 1, molecular weight markers; 2, 11103;3, 11729. Lanes 4l1 contained outer membranes from 11729 containing thefollowing plasmids: lane 4, pUCP19, 5, pXH2; 6, pHP1; 7, pHP2; 8, pHP3; 9, pHP4;10, pHP5; 11, pflP6; 12, pHP7; 13, pHP8. For each lane, 20 jig of outer membraneprotein was loaded.108of the mutant peptides to the outer membrane, whole cell lysates were made andassessed by Western-immunoblotting using anti-OprD antibody. There were nodetectable bands corresponding to OprDL3 or OprDL4 in the whole cell lysate(Fig. 25, lane 6 & 7), confirming the complete degradation of those two mutantproteins.E. Trypsin Susceptibility of the Deletion VariantsThe above results indicated that the OprD derivatives were properly locatedand assembled in the outer membrane. To probe the configuration of these OprDderivatives in P. aeruginosa, trypsin susceptibility assays were performed.Outer membrane proteins tend to be protease resistant because of theirpossession of extensive n-structures with the linking surface loops tightly packedand folded in towards the porin channel (Cowan et al., 1992). In our studies,trypsin treatment of the outer membrane from H729 containing wild type OprDprotein resulted in substantial retention of full sized OprD with a small amount ofdegradation to two protected fragments of apparent molecular masses of 32 kD and16 kD (Fig. 26, lane 3). Similar results were obtained for P. aeruginosa strain H729expressing OprD derivatives with deletions in loops L7 and L8 (Fig. 26, lanes 8, 9).Derivatives with deletions in loops Li, L2, L5 and L6 also were substantiallytrypsin resistant, although one or two additional fragments of mass 25 kD and 40kD, were generated by trypsin treatment (Fig. 26, lanes 4 to 7). This is in1091234567.891011--—.aFigure 25: Western-immunoblot demonstrating expression of OprD derivativesin P. aeruginosa whole cell lysates.Lanes: 1, 11729; lanes 2’-41 contained whole cell lysates from H729contpining the following plasmids: lane 2, pUCP19; 3, pXH2; 4, pHP1; 5, pHP2; 6,pHP3; 7, pHP4; 8, pHP5; 9, pHP6; 10, pHP7; 11, pHP8.110123458966743302014789Figure 26: Tr3rpsinization studies of OprD derivatives.Western-immunoblot of trypsinized outer membrane samples of P.aeruginosa strain H729 containing OprD derivatives. Lane 1 is molecular weightmarker and lane 2 is the untreated wild type OprD control. Lanes 3-9 containedtrypsin-treated outer membranes from H729 with the following plasmids: lane 3,pXH2; 4, pHP1; 5, pHP2; 6, pHP5; 7, pHP6; 8, pHP7; 9, pHP8.—p —. — OprD111agreement with the proposal that deletions of these predicted loops could havecaused local modifications of OprD configuration, leading to the exposure of certaintrypsin susceptible sites. Nevertheless these data were generally consistent withthe correct folding of the OprD derivatives. Increasing the amounts of trypsinandlor incubation time resulted in the generation of more fragments for both wildtype and mutant OprD.F. Revised OprD Model.Based on the above data, the OprD model was refined in collaboration withDr. Denis Jeanteur. By combining multiple alignments with amphipathicitycalculations, it was demonstrated that, although OprD was a specific porin for basicamino acids and imipenem, in contrast to other members of the non-specific porinsuperfamily, it could be aligned almost as well to OmpF as was OmpF to thestructurally-related porin from Rhodobacter capsulatus (Appendix, Table XII).Detafled examination indicated that the alignment was stronger in the predictedmembrane spanning regions and on this basis, OprD was the first specific porinthat could be included in the porin superfamily alignment (Appendix, Fig. 33). Incontrast, neither OprD nor other members of the porin superfamily could besuccessfully aligned with other specific porins such as E. coli porin LamB or Tsx(Jeanteur et al., 1994a).From the multiple alignments and amphipathicity calculations, sixteen 3-112strands were predicted and could be aligned to those of other members of the porinsuperfamily. Four other segments, according to our membrane criteria, could alsobe predicted as transmembrane segments, but were rejected in the alignmentprocedure. In addition, the placement of p-strands 7 and 8 relied on the fact thatdeletion of the predicted L4 in the original model, did not result in expression. Thepredicted loop 3 (S130 to R169)was as long or longer than any other loops (Appendix,Fig. 34). Four periplasmic turns, Ti, T4, T5 and T6 were clearly predicted by turnpropensity analysis (Appendix, Fig. 33). Most of these turns were short (2 to 9residues) and of about the same length as those determined from the knownstructures. A revised OprD membrane topology model was proposed (Appendix, Fig.34), which was in general agreement with the previous model.G. Summary.The first OprD topology model was proposed based on the sequencealignment with E. coli porin OmpF and structural predictions. Sixteen p-strandswere predicted, connected by short turns at the periplasmic side, whereas the eightexternal loops were of variable length but tended to be much longer. PCR-based sitedirected mutagenesis was performed to separately delete short stretches (4-8) ofamino acid residues from each of the predicted external loops. These mutants werecharacterized by DNA sequencing, expression of the mutant OprD derivatives, and.assessments of trypsin susceptibility. The deletion mutants from the predicted113external loops Li, L2, L5, L6, L7, and L8 were tolerated in both E. coil and P.aeruginosa, whereas the L3 mutant was only expressed in F. coil and the L4mutant was not expressed in either bacterium. In addition, expressed mutantproteins maintained substantial resistance to trypsin treatment in the context ofouter membranes. Based on this model, Denis Jeanteur performed multiplesequence alignments between OprD and seven representatives from the porinsuperfamily. OprD was the first specific porin that could be aligned with membersof the so-called porin superfamily. Utilizing this alignment in conjugation withamphipathicity calculations, a revised OprD model was proposed.114CHAPTER FOUR Structure/Function Relationships: FunctionalAlterations of Deletion Mutants.A. Introduction.The OprD topology model predicted 8 external loops, including 3 long loops,L2, L3 and L7. At least one of them might fold inside the channel to form the‘eyelet’ region, and the specific binding site for imipenem would be anticipated tobe located in that region. To elucidate the organization of the channel and to locatethe specific binding site for imipenem, the deletion mutants were examinedregarding their function in antibiotic and sugar transport. In addition, twointeresting OprD derivatives, OprDL2 and OprDAL5 were purffied and analyzedin the black lipid bilayer system. As shown below, the results are used to proposea molecular architecture for the OprD channel and to explore the transportmechanism of imipenem through the specific porin.B. Effects of Deletion on Imipenem/Meropenem Susceptibilities.To determine if any of the deletions influenced the function of OprD as achannel for imipenem and the related carbapenem antibiotic meropenem, MICswere assessed for those two antibiotics in the OprD-defective strain 11729background. As described earlier (Chapter 2), strain H729 expressing excess OprDfrom plasmid pXH2 had an MIC that was 16 to 32 fold lower than those observed115for strain 11729 and 11729 carrying the vector pUCP19 (Table VIII). Similarly therewas a 16 to 32 fold reduction in MIC for strain 11729 expressing the mutant OprDswith deletions in ioops Li, L5, L6, L7 or L8. In contrast, the ioop L2 deletionexpressed from plasmid pHP2 resulted in oniy a 2 to 4 fold reduction in MIC toimipenem and meropenem (Table VIII).To demonstrate the differences in MICs were only due to the deletionsperformed on the predicted loops, the following control experiments wereperformed. Three control antibiotics were used, including two polycationicantibiotics (gentamicin and polymyxin) and trimethoprim which diffuse cross theouter membrane through the self-promoted pathway and the hydrophobic pathwayrespectively. There were no significant differences in MIC for any of the strainsstudied (n.b. a 2 fold difference in MIC is considered by convention to be withinexperimental variability), indicating that the mutant proteins did not grosslydisrupt the outer membrane since it had retained its barrier property (Table VIII).To exclude the possibility that the differences in MICs resulted from the differentcopy numbers of the plasmids encoding OprD derivatives, we measured theplasmid-encoded -lactamase levels in all the strains. The results indicated thatthe plasmid copy numbers were similar (Table VIII). In addition, there were nosiginificant differences in growth rates for all the strains, indicating that thedeletions did not cause metabolic disturbances.It has been previously demonstrated that lysine will compete with imipenemfor uptake through OprD Fukuoka et al., 1993), resulting in an increasing MIC as116TableVIII:Effectsofdeletionsonimipenem/meropenemsusceptibilitiesofP. aeruginosastrains.-lactamaseactivityMICs(,ig/ml)aStrains(nmolnitrocefin/mgofcellpermm)ImipenemMeropenemGentamicinPolymyxinTrimethoprimH7295.4±0.616.,pUCP19)166±1516.±530.50.1252.04.0100H729(jHP1)129±441.00.1251.04.0100H729(pHP2)’128±264.Oc1.002.02.0100H729(pHP5)122±211.00.1251.04.0100H729(jHP6)188±561.,pHP7)135±±350. of imipenem/meropenem.c.TheboldnumbersindicatetheIVIICs of imipenem/meropenemwhichweresignificantlydifferentfromthecontrols.I.-zIa function of lysine concentration. In contrast, lysine had no significant effect onimipenem MICs measured using the L2 deletion mutant OprD strain H729(pHP2)(Fig. 27). These data suggested that this deletion substantially influenced thepassage of both imipenem and lysine through OprD.C. Effects of Deletions on Other Antibiotic Susceptibilities.MICs of other antibiotics for strain H729 expressing the OprD deletionderivatives were also assessed. It was interesting to note that the deletion in L5 ledto the enhancement of susceptibilities to different kinds of antibiotics, including 3-lactams (cefpirome, cefotaxime and aztreonam), quinolones (ciprofloxacin andfleroxacin), chioramphenicol and tetracycline (Table IX). For chioramphenicol, thesusceptibility increased 32-fold. Similar results were obtained for the deletions inL7 and L8, except that there were no differences in susceptibility to tetracycline(Table IX). The results indicated that the deletions may have resulted in a moreopen channel leading to a significant increase in the permeability of the channelto antibiotics that were normally excluded. This could be explained by the deletionseither changing OprD from a specific porin to a general porin, or converting OprDto a specific channel with high general permeability as observed for the E. colisucrose porin ScrY (Schülein et al., 1991).To determine if the deletions in loops L5, L7 and L8 also affected thepassage of imipenem and lysine through OprD (i.e. from specific to non-specific118200 50 100 150 200 250Lysine (mM)Figure 27: Competition between L-lysine and imipenem for OprDL2.Effect of L-lysine concentration in BM2 glucose medium on the susceptibilityof OprD-defective strain H729 (filled circle), 11729 expressing OprD with deletionin the predicted loop 2, H729(pHP2) (open circle) and 11729 expressing nativeOprD, 11729 (pXII2) (triangle) to imipenem.119TableIX:EffectsofdeletionsonotherantibioticsusceptibilitiesofP. aeruginosastrains.MICs(tgIm1Y1MICsweredeterminedbytheagardilutionmethodonMueller-Hintonplates.EachMICwasdeterminedthreetimesindependently.Theboldlettersindicatethestrains whichshoweddifferencesinantibioticsusceptibilities.TheboldnumbersindicatetheIVllCs ofantibiotics whichweredifferentfromthecontrols.StrainsCefpiromeCefotaximeAztreonamCiprofloxacinFleroxacinTetracyclineChioramphenicolH7291.°2.0°0.5°0.0156°0.06°0.78°0.78°H729(jHP6)’05°2OC05C00312°0125°625312°H729(pHP8)b0.5c2.0°0.5c0.0312°0.125°6.253.12°a. b. c.I.uptake), competition experiments were performed. Figure 28 demonstrated thatthose deletions did not influence the binding of imipenem or lysine in the OprDchannel.D. Effects of Deletion on Sugar Transport.To determine if the deletions in L2, L5, L7 and L8 affected the function ofOprD as a channel for the transport of gluconate, strain 11729 expressing thoseOprD derivatives were grown in BM2 minimal medium with gluconate as thecarbon source at concentrations ranging from 0.5 to 10 mM. Except H729(pHP2),the growth rates of all the strains did not show significant differences from that ofstrain H729 expressing native OprD, suggesting that those deletions did not affectthe passage of gluconate through OprD channel. For some unknown reason,H729(pHP2) did not grow in BM2 with gluconate as the carbon source.E. Purification of the Mutant OprDs.To investigate the in vitro functions altered by the deletion mutagenesis,OprDAL2 and OprDAL5 were purified (Fig. 29, lane 4 & 6) following similarprocedures described for wild type OprD. The purified OprDL2 retained OmpAlike heat modifiability, as observed for native OprD (Fig. 29, lane 5). However,OprDL5 did not change banding position when solubilized at low temperature1212010:.../--A -I,—I—I —/ ——— / —0.1 I P I I0 50 100 150 200 250Lysine (mM)Figure 28: Competition between L-lysine and imipenem for OprDAL5/7/8.Effect of L-lysme concentration in BM2 glucose medium on the susceptibilityof OprD-defective strain 11729 (filled circle), 11729 expressing OprD with deletionin the predicted loop 5, H729(pHP5) (open circle, broken line) and H729 expressingnative OprD, H792(pXH2) (triangle) to imipenem. The same results were obtainedfor strain H729(11P7) and H729(11P8) as that observed for 11729(HP5).12212345679667”—43 —__— E— OprD3O.-i.2OFigure 29: Comparison of heat-modiflabilities of the purified native OprD andmutant OprDs.Lanes: 1, molecular weight markers; 2 & 3, native OprD; 4 & 5, OprDL2;6 & 7, OprDAL5. Lanes 2, 4, and 6 are heated samples, while lanes 3, 5 and 7 arethe corresponding unheated samples.123prior to electrophoresis (Fig. 29, lane 7). The results suggested that the deletionmay have affected the structural stability in such way that SDS could completelydenature the protein at room temperature. Nevertheless, the purified OprDAL5was still active, which was confirmed by being able to reconstitute channels in theblack lipid bilayer model membrane (see below).F. Effects of Deletion on the Physical Properties of the Channel.The pore characteristics of OprDAL2 and OprDL5 were further studied invitro in black lipid bilayer experiments. At nanomolar concentrations, both porinswere able to increase the specific conductance of the lipid bilayer by several ordersof magnitude. The time-course of the increase was similar to that of OprD. After arapid increase during 1O4O mm, the membrane conductance increased at a muchslower rate. The addition of the porins at much lower concentrations to the aqueousphases bathing lipid bilayer membranes allowed the resolution of stepwise increasein conductance. For OprDL2, the average single channel conductance (26 pS in1.0 M KC1) was slightly bigger than that of OprD (Fig. 30). However, mutant OprDwith deletion in the predicted Loop L5 showed more than 30-fold increase in theaverage single channel conductance (675 pS in 1.0 M KC1) compared with that ofthe wild type OprD (Fig. 20B; Fig. 30). Measuring the average single channelconductance of the loop L5 deletion variant in KC1 solutions ranging inconcentration from 0.3 to 3.0 M showed a linear relationship between the salt124100 80 60 40 200Conductance(pS)Figure30:ComparisonofsinglechannelconductancebetweennativeandmutantOprDs.Histogramoftheconductancesobservedwithmembranesformedby1%oxidizedcholesterol inn-decaneuponadditiontotheaqueous phaseofpurifiedOprD,OprDL2andOprDAL5. Gistheaveragesinglechannelconductanceofmorethan100eventsrecorded.1020304050601020304050605006007008009001000concentration and single channel conductance (Table X). The much bigger singlechannel conductance confirmed the previous hypothesis that the deletion increasedthe general permeability of the channel, which in turn, increased thesusceptibilities of H729(HP5) to different antibiotics (Table IX).Earlier results showed that OprD was a weakly cation-selective channel. Forboth OprDL2 and OprDAL5, increasing the size of cation caused a steadydecrease in the average single channel conductance, while it was little affected bychanging the size of anion (Table X), indicating they were stifi cation selectivechannels. Zero current membrane potential measurements confirmed that they hadsimilar cation preferences to that of the wild type OprD (Table XI).As described earlier (Chapter 2), macroscopic conductance due to nativeOprD, in 1.0 M KC1, could be inhibited by imipenem with an I value of 1.4 jiM(Fig. 31). In contrast, for OprDAL2, no decrease in conductance was observed upto the tenth addition of imipenem (Fig. 31, 0.2 jiM each addition). These resultsindicated that OprDAL2 had a much lower affinity for imipenem, which furthersuggested the deleted stretch was involved in the specific binding of imipenem. Thisin vitro property was also in agreement with the in vivo functional data forH729(HP2). In case of OprDL5, the progressive decrease in conductance upon theaddition of imipenem solution was stifi observed (Fig. 31), suggesting that thedeletion did not affect the specific binding for imipenem. Therefore OprDàL5 wasstill a specific porin but with much higher general permeability, like the E. colisucrose porin ScrY (Schulein et al., 1991).126Table X: Average single-channel conductance of the native and mutantOprD pores in different salt solutions.Aqueous salt Conductance (pS)solutions WT OprD OprDAL2 OprDAL50.3MKC1 a a 229LOMKC1 20 26 6753.OMKC1 26 31 17111.OMCsC1 15 <10 557l.OMLiC1 <10 2721.0 M KMOPS 19 25b 640a. The single channel conductance was too small for the resolution of theequipment to detect.b. 1.0 M KMOPS was replaced by 1.0 M KNO3,since the membrane was toonoisy in 1.0 M KMOPS in the presence of OprDL2.127Table XI: Zero-current membrane potentials.Porins Pc/PaWTOprD 2.60± 0.39OprD AL2 2.90 ± 1.30OprD AL5 2.61 ± 0.27128120+_+__+o- —e——+—+—+80-oc.)— 6040-0200 I I I0 0.4 0.8 1.2 1.6 2.0Imipenem Concentration(1iM)Figure 31: Macroscopic conductance inhibition experiments.Macroscopic conductance inhibition experiments using native OprD (filledcircle), OprDAL2 (plus) and OprDAL5 (open circle). Purified protein was added tothe salt solution (1.0 M KC1) and the increase in membrane conductance due toinsertion of porin pores was followed until the rate of increase had slowed (1’-’2hours). At this time, aliquots of imipenem solution were added to the bathing saltsolutions (volume 6 ml) in both compartments of the lipid bilayer chamber (i.e., toeach side of the membrane) to increase the concentration in steps of 0.2 jiM andstirred until the conductance stabilized (about 2 mm). After a stable conductancelevel was achieved additional aliquots were added to each side of the membrane.129G. Summary.In Chapter 3, an OprD topology model was proposed and 8 deletion mutantswere made, one from each of the predicted external loops. Six of these deletionmutants could be expressed in the P. aeruginosa outer membrane. The effects ofdeletions on the in vivo and in vitro functions of OprD were examined in thischapter. OprD derivatives with deletions in loops, Li, L5, L6, L7 and L8reconstituted similar imipenem supersusceptibility in the P. aeruginosa OprDdefective background. In contrast, L2 deletion mutant only partially reconstitutedthe supersusceptibility. Consistent with this, competition experiments showed thatlysine had no significantly antagonistic effect on imipenem MICs for H729(HP2).Furthermore, purified OprDL2 showed much lower affinity to imipenem inmacroscopic conductance inhibition experiments. These data indicated that L2 wasinvolved in imipenem binding. Another interesting mutant, L5 deletion mutant,resulted in supersusceptibility to many antibiotics, Further analysis confirmed thatthis deletion had changed OprD to a ScrY-like porin, a specific porin with highgeneral permeability.130DISCUSSIONOprD is an interesting porin from the P. aeruginosa outer membrane sinceit facilitates the diffusion of basic amino acids and imipenem, a potent antibioticthat has been used for the therapy of P. aeruginosa infections. In this study, theoprD gene was cloned, nucleotide-sequenced and overexpressed in both E. coli andP. aeruginosa. These and other genetic manipulations provided effective approachesto investigating structure-function relationships in OprD, in an attempt to addressthe mechanism by which imipenem and nutrients diffuse through this specificporin.A. Function of OprD in Antibiotic Uptake.In this study, the genetic approach was employed to define the in vivo roleof OprD in antibiotic uptake. Previous studies using clinical isolates had indicateda role for OprD in uptake of zwitteronic carbapenems, including imipenem.However, these clinical isolates had different genetic backgrounds and there wasno convincing evidence presented that they were isogenic, differing only in theirability to produce OprD. Also, it was known that there were at least one regulatorylocus (opdE) and one poorly understood multiple-antibiotic resistance locus (nfxC)(Fukuda et al., 1990) that could influence OprD expression and imipenemsusceptibffity. Furthermore, the potential for obtaining double mutants when usingclinical strains or when selecting directly with antibiotics was reported (Zhou et al.,1311993). All these indicated the importance of utilizing truly isogenic strains obtainedwithout direct antibiotic selection.The strains constructed had identical genetic backgrounds except for theirOprD levels (Pab1e IV), as confirmed by their outer membrane protein proffles (Fig.15). In addition, there were no significant differences in the growth rates of any ofthe strains in either rich medium or minimal medium. Therefore, the differencesin MICs were only due to differing OprD levels. OprD was only moderatelyexpressed in wild type strains like H103 in most growth media (Fig. 15).Nevertheless this level of expression was sufficient to enhance the uptake ofimipenem and meropenem, since the OprD deficient mutant 11729 was four foldmore resistant to these antibiotics than the parent strain H103 (Table IV). Howeverthe level of OprD in strain H103 was insufficient to permit maximal uptake, andouter membrane permeability to imipenem and meropenem was stifi rate-limiting,since overexpression of OprD led to enhanced susceptibility to both antibiotics.These data confirmed OprD could facilitate the diffusion of imipenem/meropenemacross P. aeruginosa outer membrane.The MIC of imipenem for the OprD-defective strain H729 was 16 jig/mi, aconcentration at which an isolate is considered resistant from a clinical perspective(Barza, 1985). It indicated a very slow diffusion of imipenem through non-specificporins and possibly non-porin pathways. The rate of such nonspecific diffusionprocesses is essentially proportional to the external concentration of imipenem(Trias et aL, 1989). At an external concentration of 16 gg/ml, the diffusion rate of132imipenem through OprD would be predicted to be at least ten times higher thanthat through non-specific porins, and more than 95% of the imipenem moleculeswould diffuse through the saturable, specific channel formed by OprD (Trias andNikaido, 1990b).From a structural perspective, the facilitated diffusion of imipenem was dueto the presence of a specific binding site(s) in the OprD channel, as confirmed bymodel membrane studies. The first crystal structure of a specific porin LamBrevealed that a series of residues, located at the most constricted portion of andalong the channel, interacted with maltose in a highly stereospecific fashion.Therefore, it might be anticipated that the high affinity of imipenem for the OprDchannel resulted from similar binding process. Evidence for this included that someother carbapenems, for example, meropenem (Livermore and Yang, 1989), bipanem(Catchpole et al., 1992) and panipenem (Fukuoka et al., 1993), with similarstructures to imipenem, could also utilize the OprD channel for the facilitateddiffusion. All of these carbapenems have a single basic group at position 2 of acarbapenem. However, carbapenems lacking a basic residue (Trias and Nikaido,1990a) or with one more basic group at position 1 or 6 (Fung-Tomc et al., 1995)could not use the OprD channel, reflecting the critical requirement for certainsubstrate structures by the channel. On the other hand, deletion of the bindingsites located in L2 caused loss of the ability of imipenem to inhibit macroscopicconductance (Fig. 31).Some authors have also suggested a role for OprD in uptake of quinolones133Vlichea-Hamzehpour et al, 1991) or other antibiotics (Satake et al., 1990) in spiteof their different chemical structures from imipenem. These conclusions were basedin part on poorly defined clinical or experimental animal isolates, or on in vitromodel liposome sweffing experiments that have been criticized on other grounds(Bellido et al., 1992; Trias et al., 1989). The in vivo experiments described here,utilizing genetically defined isogenic variants, were more definitive. If OprD wereto be involved in uptake of quinolones and other antibiotics, one would expect thatthe substantial overexpression of OprD in H 103(jXH2) and H729pXH2) wouldincrease the normal low outer membrane permeability of P. aeruginosa and thusdecrease MICs of these antibiotics. In contrast no significant alteration insusceptibility was observed. Therefore, OprD could only facilitate the diffusion ofimipenem and those carbapenems with only one positive group at position 2.B. Function of OprD in Nutrient Uptake: Is Lysine the Best Substrate?There is no doubt that the original purpose of having OprD in the P.aeruginosa outer membrane was not for transport of imipenem, which would tendto result in cell death, but for the facilitated uptake of essential nutrients. Previouswork demonstrated that basic amino acids and small peptides containing theseamino acids were the natural substrates of the OprD channel (Trias and Nikaido,1990b; Fukuoka et al., 1991). In this study, competition experiments using isogenicmutants confirmed that basic amino acids shared common binding sites with134imipenem in the OprD channel.The function of OprD in transport of carbon sources was not examined indetail in the literature although other authors suggested a role for uptake of smallsugars, based on results from liposome sweffing experiments (Yoshihara andNakae, 1989; Yoshihara et al., 1991). In this thesis, three isogenic strains includingOprD-defective strain 11729, wild type strain H103 and OprD-overexpressing strainH103(jX112) were utilized to investigate the role of OprD in transport of threecommonly used carbon sources for P. aeruginosa: gluconate, glucose and pyruvate.Previous work demonstrated that OprF was the major porin for uptake of di-, tn-and tetrasaccharides (Beffido et al., 1992). However OprF levels did not influencegrowth on gluconate. Consistent with this, it was demonstrated here that OprD wasthe major porin for gluconate in that its absence (in strain H729 oprD::) led to anearly three fold decrease in growth rate on 0.51.0 mM gluconate (compared withthe parent strain 11103), whereas its overexpression resulted in a 70% increase ingrowth rate (Fig. 16A). As expected for an outer membrane diffusion-limitedprocess, these differences disappeared at high concentrations, at whichconcentrations other porins including OprF might be expected to functionadequately. In addition, no significant differences in the rates of growth on glucoseor pyruvate or on rich media were observed regardless of OprD expression levels(Fig. 16), indicating that the results for gluconate were not due to metabolicdisturbances caused by the loss or overexpression of OprD. The result obtained forglucose did not indicate that OprD is unable to permit the passage of glucose since135these studies were performed in strains capable of being induced for the glucosespecific porm OprB. However, given the approximately 5 fold higher level of OprDthan OprB in strain H103(pXH2) grown on glucose, it seems likely that OprD hasat best a minor role in glucose uptake, as confirmed in part by the data in Table VI.To see if gluconate might be an analog of basic amino acids and imipenem,we compared their three-dimensional structures by using the computer programHyperChem. Except for the common possession of a carboxyl group, gluconate wasnot structurally related to the basic amino acids or imipenem. This was alsoconfirmed by competition experiments (Table VI), which suggested that commonbinding sites were not involved in imipenem and gluconate passage through OprD.Three possibilities were proposed to explain how the OprD channel might facilitatethe transport of gluconate. First, since both basic amino acids and imipenemcontain carboxyl groups and since the only difference between gluconate andglucose is that the former has a carboxyl group, the carboxyl group might functionin directing the molecules to the channel and in binding to sites within the channel.Second, there might be two functional domains in the OprD channel, one for thebinding of basic amino acids and the other one for gluconate. Third, given the lowouter membrane permeability of P. aeruginosa, gluconate may pass through thechannel in a nonspecific fashion. This latter suggestion would make the OprDchannel analogous to the sucrose porin ScrY which has the properties of bothsubstrate-specific and general porins (Schülein et al., 1991).Despite the fact that basic amino acids could diffuse efficiently through the136OprD channel, there were a few pieces of evidence which suggested that basicamino acids may not be the best natural substrate of this channel. First, theaffinity of the channel toward basic amino acids was much lower than its affinitytoward the presumably “unnatural” analog, imipenem. A competition experimentshowed that the concentration of lysine (100 to 200 mM) required to completelyblock the transport of imipenem was 2,000 to 4,000 times higher than thecorresponding imipenem concentration (0.05 mM) (Fig. 17). Consistent with this,I did not observe the inhibition of the macroscopic conductance when titrated withlysine solution in model membrane system, possibly because the small volume ofthe testing chamber (6 ml) could not build up high enough concentration of lysineto block the transport of KC1. Second, the channel clearly preferred the D-isomersof lysine and arginine over their L-isomers (Trias and Nikaido, 1990b). Third, theOprD channel could also facilitate the diffusion of a wide variety of peptides,including di-, tn- and tetrapeptides with basic amino acids at the COOH-terminalposition, or dipeptides with basic amino acids at theNH2-terminal position followedby a small amino acid residue at the COOH-terminal position (Tnias and Nikaido,1990b). These suggested that imipenem is really a dipeptide analogue, a suggestionthat matches with its unique susceptibility to hydrolysis by renal dipeptidase. Thusthe fit of basic amino acids into the binding site may be rather poor.C. Prediction of an OprD Membrane Topology Model.137Functional studies have indicated that OprD could facilitate the uptake ofbasic amino acids and imipenem by virtue of possessing specific binding sites, butno previous work had been done to identify the structural characteristics of thisporin or the specffic binding sites which are involved in antibiotic and nutrientuptake through this specific channel. In the absence of crystallographic data, thisthesis presented a prediction of the topology model of OprD and an approach toassess the accuracy of this model.The published crystal structures of the general porins and specific porinsreveal consensus structures. They all form trimers of identical subunits, eachmonomer subunit consisted of 16 or 18 anti-parallel p-strands forming a barrelsurrounding a pore. These strands are connected by very short loops on theperiplasmic face of the porin whereas the ioops on the outside of the bacteria areof variable length but in general are longer (Cowan et al., 1992; Weiss and Schulz,1992; Kreusch et aL, 1994; Schirmer et al., 1995). Analysis of the structures of afamily of bacterial general porins by sequence alignment and structure predictionssuggested similar structures for all of the general porins in the porin superfamily(Jeanteur et aL, 1991). P. aeruginosa OprD, like other specific porins, wasconsidered unlikely to align with the porin superfamily. However, thetransmembrane segments of OprD showed good homology with the known-strands of OmpF (Fig. 21). A more detailed study by Dr. Denis Jeanteursubsequently indicated that OprD was the first specific porin which could bealigned with the porin superfamily (Appendix, Fig. 33)138To verify the accuracy of the model, an efficient site-directed deletionmutagenesis technique was developed in this study. The utilization of PCR resultedin a high ratio of mutant to wild type fragments. The two strategies used here,direct extension and overlap extension, are of general applicability. The rapidscreening step allowed us to sequence only those clones with the desired deletions.By using Vent DNA polymerase and optimizing the PCR conditions, the averageerror frequency was lowered to 1 in 2,000 base pairs. Given the substantial savingin time, general feasibility for obtaining the deletion, and low rate of undesiredmutations, these techniques can be applied to any membrane protein.The OprD model provided a prediction of the flexible segments (loops) ofOprD. Generally speaking, insertions or deletions in porins should be non-disruptive of overall structures only if they occur in the surface loops. Comparisonbetween the crystal structures of OmpF and PhoE demonstrated that all theevolutionary insertions or deletions were restricted to the ioop regions and a singleshort turn (Cowan et al., 1992). Consistent with this, sequence comparisons ofporins from distant families showed that the ioop regions often varied substantiallyin length, in contrast to the highly conserved p-strands (Jeanteur et al., 1994a).One reason is that selective pressure from the environment, eg. antibiotics orphages may play a role in allowing certain regions to evolve at rates higher thanothers. Another possible explanation is that the external loops simply have morefreedom to change without altering porin secretion, folding or the ability to form atransmembrane channel. For example, deletions of certain PhoE cell surface139exposed regions did not interfere with the translocation across the inner membraneor the incorporation into the outer membrane (Agterberg et al., 1989). Moreover,spontaneous deletions located in the OmpF or OmpC external loops could producemutant OmpF or OmpC proteins which were not only active but also allowed thepassage of large maltodextrins (Benson et al., 1988; lVlisra and Benson, 1988). Incontrast, the membrane spanning segments contribute especially to theconformation required for stability, folding or outer membrane localization, sincestudies involving deletion mutagenesis of PhoE, removing either the first (Boschet al., 1988) or the last (Bosch et al., 1989) transmembrane segment, drasticallyaffected or completely inhibited incorporation into the outer membrane.Three criteria were used to evaluate whether the mutant proteins folded intonear-native configurations. First, the polypeptides encoded by mutant OprD allelesin E. coli and the native host P. aeruginosa were identified (Fig. 23 & 24). The highlevels of expression and correct location in the outer membrane of OprD derivativescontaining deletions of the presumed loops Li, L2 and L5 to L8, were consistentwith our model. The deletion of the predicted L3 was tolerated in E. coli, butresulted in reduced expression. The same phenomenon was observed for certaindeletion mutants of PhoE (Agterberg et al., 1989). We anticipated that, as for othermembers of the porin superfamily, L3 may be involved in forming the ‘eyelet’region. In E. coli porins, the individual residues involved in the eyelet are highlyconserved, and mutations in this region could affect the size, conductivity andspecificity of the channel (Jeanteur et al., 1994b). In addition, such mutations also140destabilized the trimer in many cases (Lakey et al., 1991). These may account forthe diminished expression. The L3 deletion mutant did not direct the productionof any OprD in the outer membrane of P. aeruginosa. This may reflect an innateand more efficient ability of P. aeruginosa to proteolyse unstable products,compared to E. coli. A second criterion used to assess configuration was trypsinsusceptibility (Fig. 26). All of the OprD derivatives were resistant to digestion tosome extent, indicating that the deletions did not cause extensive alterations inconfiguration. The third criterion was functional activity (Table VIII). All of thetolerated deletion mutant OprDs could form functional channels and did not grosslydisrupt the outer membrane since it retained its barrier properties againstpolymyxin, gentamicin and trimethoprim (Table VIII). Five of the deletion mutantscould reconstitute fuil imipenem and meropenem susceptibility. Only one, the ioopL2 deletion mutant, lost the ability to reconstitute full susceptibility, suggesting apossible role for loop L2 in imipenem binding.D. Molecular Architecture of the OprD Channel.Structural and functional studies allowed construction of a moleculararchitecture for the interior of the OprD channel (Fig. 32). In the crystal structureof specific porin LamB, more than one loop folds inside the channel, including L3,which was entirely folded into the barrel to form the eyelet, whereas Li and L6from the same monomer and L2 from the adjacent monomer were folded inward to1411,4,6 1,4,6-_5,7,8 5,7,82,3? 2,3?Figure 32: Schematic representation of the predicted interior architecture of theOprD channel.142different extents. Similarly, I predict that five loops may fold inside the OprDchannel. L2 and possibly L3 are proposed to completely fold into the barrel, andtogether with some residues from the barrel walls, form the constriction zonetoward the middle of the channel. In addition, L5, L7 and L8 would be partially orcompletely folded into the channel to further restrict the lumen at the channelentrance (especially L5). The other ioops, Li, L6 and possibly L4, together with thesurface-exposed portion of the partially inward folded loops would be arrayed onthe surface of OprD. One of these loops might reach into the neighbouring monomerand be involved in stabilizing the trimer. This model is consistent with all thefunctional data presented in this thesis.Several pieces of evidence supported the involvement of L2 in the eyeletregion. First, the L2 mutant only partially reconstituted the supersusceptibility toimipenem in the OprD-defective background 11729. For meropenem, the differencein MICs between 11729 and H729(pHP2) was only 2-fold, which by convention,could be considered insignificant (Table Viii). The results indicated that the uptakeof imipenem and meropenem was seriously affected by this deletion. In addition,competition experiments demonstrated that the deletion substantially influencedthe uptake of lysine, since lysine lost its antagonistic effects on imipenem MIC forH729(pHP2) (Fig. 27). Consistent with this, imipenem was unable to inhibit themacroscopic conductance of KC1 through channels formed by OprDAL2 (Fig. 31).All of these data indicated that the deleted stretch was critical for the efficientbinding of imipenem and basic amino acids. From the crystal structure of LamB,143all the residues identified to be responsible for the binding of maltodextrin werelocated in the eyelet region (Hofnung, 1995). Therefore, L2 was very likely locatedin the eyelet region of the OprD channel. However, our data did not preclude theco-involvement of loop L3 which, in all of the structurally defined porins, folds intothe center of the channel to form the ‘eyelet’ region determining channel diameterand selectivity.The placements of L5, L7 and L8 were based on the fact that the deletionsperformed on these predicted loops led to the enhancement in susceptibilities tothose antibiotics which cross outer membrane through the hydrophilic pathway(Table IX. The antibiotics included -1actams, quinolones, chloramphenicol andtetracycline (the latter for the L5 deletion mutant only). Control experimentsindicated that the deletions did not disrupt the integrity of the outer membrane, orsubstantially affect the growth rates of the mutants. Therefore, thesupersusceptibilities were only due to the deletions resulting in more open channelswith higher permeability. In good agreement with this, the single channelconductance of OprDzL5 (675 pS) was more than 30 times higher than that ofnative OprD (20 pS). Interestingly, these deletions did not affect the specificbinding site(s), as confirmed by the antagonistic effects of lysine concentration onimipenem MICs for the mutants (Fig. 28), and the retention of the ability ofimipenem to inhibit macroscopic conductance for the OprDL5 channels (Fig. 31).Therefore, these 3 loops were not involved in the eyelet region. Instead, they weresimilar to Li, L2 and L6 of LamB, which folded inward and restricted the size of144the entrance to this channel. The MICs of chioramphenicol and tetracycline forH729(pHP7) and H729(jHP8) were 4-fold and 8-fold higher than the MICs forH729(pHP5), respectively (Table IX). The chemical structures of these twoantibiotics are quite bulky, and the four-ring-structured tetracycline is even bulkierthan chloramphenicol. In order to allow the maximum passage of these twoantibiotics, I propose that the channel has to be more open. Therefore thedifferences in MICs could be explained if the deletion of the predicted L7 and L8 didnot open the channel as widely as did the deletion of L5. Based on this, I assumedthat L5 contributed the most to restricting the channel size, whereas L7 and L8only partially restricted the channel size.Regarding other loops such as Li and L6, the deletions did not significantlyaffect susceptibilities to imipenem or the other antibiotics tested (Table WIT & IX),suggesting that these two loops were not as important in determining the channelsize or selectivity. Therefore, Ll, L6 arid possibly L4 are proposed to be completelyexposed at the cell surface, and with surface-exposed parts of L7 and L8, they covermost of the outer surface of OprD.The modelling of channel architecture supported the assumption that outermembrane permeability is important for antibiotic susceptibility. The singlechannel conductances of the channels could be ordered as OprD (20 pS) <OprDL2(26 pS) < OprDAL5 (675 pS). The small size of the native OprD channelpresumably served to maintain the low intrinsic permeability of the P. aeruginosaouter membrane (Nicas and Hancock, l983a). Deletion of the predicted loop L2145slightly opened the channel, so it did not enhance the susceptibilities to theantibiotics tested, except imipenem and meropenem. The L7 and L8 deletionderivatives, presumably opened the channel significantly, resulting in increasedsusceptibilities to p-lactams, quinolones, and chioramphenicol. However these twomutant channels were stifi not as large as the L5 deletion mutant channel sincethey were not able to increase the susceptibility to tetracycline (Table IX). The onlydifference between H729(pXH2) and H729(J)HP5) was that the latter one containeda deletion of 8 amino acids from OprD, resulting in the presence of a large channelin P. aeruginosa outer membrane. The supersusceptibility of H729(jHP5) thussupported the important role of the outer membrane of P. aeruginosa as a barrierto antibiotics (Nikaido and Hancock, 1986).E. The Journey of Imipenem Through the OprD Channel.The molecular architecture of OrpD is helpful in understanding the processof imipenem uptake. The journey of imipenem and basic amino acids through theOprD channel may be depicted as follows. The initial prescreening regarding sizeand charge might be done by the loops exposed at the cell surface. These couldfunction as a primary filter, concentrating the substrates such as imipenem andbasic amino acids, especially when they were in low concentrations in the medium.The substrate molecule would then enter the mouth of the channel which would beconstricted by L5, L7 and L8. Similar to the maltose transport system, small146peptides and imipenem are long molecules, which exceeded, in their long axis, theexclusion limit of the OprD channel. Therefore, the residues located at the mouthwould orient imipenem and peptides so that they would be aligned to the pore axis.It is unknown whether the same “greasy slide” could be applied to the OprDchannel, but for the efficient transport of such extended molecules, there must beresidues located along the channel to guide their diffusion. The substrate moleculeswould then encounter the constriction zone about half way through the channel, atwhich position side chains, largely comprising charged residues from L2, possiblyL3, and the barrel wall would bind to the substrates in a highly stereospecificmanner influenced by the size, geometry and charge of the substrate molecule. Thiswould presumably account for the specificity of the channel. After the passagethrough the narrow constriction zone, the imipenem molecule would be effectivelyreleased into the bulk solvent. In spite of the structural similarity in side chainsbetween imipenem and basic amino acid containing dipeptides, it is obvious thatimipenem has a carbapenem nucleus which is quite different from the peptidebackbone (Fig. 3), and this might be the reason for the different affinities betweenimipenem and basic amino acids as described before. The higher affinity ofimipenem suggested that imipenem could fit in the specific binding site(s) betterthan basic amino acids. On the other hand, it is possible that the specific bindingsite(s) might be located in such a way that they could have good but not very highaffinity for various nutrients of similar structures, such as basic amino acids andsmall peptides containing these amino acids. The carbapenems with one additional147positive group at position 1 or 6 might be excluded from the mouth due to theirbulkiness or they might not be able to fit into the specific binding site(s). Regardingthe uptake of gluconate, certain positive charged residues located at the filter ormouth might attract its carboxyl group. After gluconate entered the channel, itssmall size might permit the diffusion at a reasonable speed.F. General Porins and Specific Porins.Porins have been subdivided into 2 classes, specific porins and general porins(Hancock, 1987). R. capsulatus porin has been classified as a general porin.However, in the crystal structure, a solute binding site was observed within thepore at the external side of the eyelet, with an unknown solute co-crystallized in it(Weiss and Schulz, 1992). Furthermore, this porin has been reported to bindefficiently to tetrapyrrols (Bollivar and Bauer, 1992). Therefore, the R. capsulatuspormn that had been previously suggested to belong to a class of general porins canbehave as both specific and non-specific porins, depending on the solute. Based onthis observation, Schulz (1993) first proposed that all porins might have specificsubstrates, but that these specific substrates have been detected in oniy few porins.The first supporting data were from crystal structures of general porins and specificporin. Although there is no detectable sequence homology and a different numberof n-strands, the resemblance of the maltoporin folding to that of the general porinsis obvious. In addition, general porins also generate a local electrostatic field near148the channel constriction zone, sufficient to orient small hydrophilic molecules andrepel hydrophobic ones (Schulz, 1993), suggesting that there is only a quantitativedifference between the ifitering and binding. The alignment of the OprD sequencewith the porin superfamily provided further strong evidence to support Schulz’sproposal about the dual nature of porins.From the crystal structure of LamB, the entrance of the pore was highlycovered by surface-exposed loops. In addition, several loops folded inside to furtherrestrict the entrance. Therefore it was not surprising that LamB has a low singlechannel conductance (0.15 nS) (Benz et al., 1986). Consistently, it has been foundthat many specific porins have pore sizes which are about one order of magnitudesmaller than those of the general diffusion pores, such as OmpF of E. coli (1.8 nS).For example, P. aeruginosa OprD (0.02 nS), OprP (0.25 nS) (Hancock and Benz,1986), and E. coli Tsx (0.01 nS) (Benz et al., 1988), all demonstrated very low singlechannel conductances in model membrane studies. However, there are also specificporins with high conductivity, almost equivalent to that of OmpF, for example, E.coli ScrY (1.4 nS) and P. aeruginosa OprDL5 (0.67 nS). From a structuralperspective, all porins have an analogous n-barrel structure, with at least oneinternal loop defining the selectivity. “Specific” porins have more loops folded intothe channel to preffiter molecules by further restricting the entrance of moleculesbased on their size and charge. Therefore they would have a decreased exclusionlimit for molecules other than the given substrates, which in turn, would make theuptake of those specific substrates as their predominant functions. “General”149porins, in contrast, would have less loops (only one) folded inside, and thus offermore open channels which would allow the non-specific diffusion of manyother molecules at a sigiifflcant rate. Despite these differences, it seems possiblethat many porins are both specific and general in nature.150REFERENCESAlbers-Schonberg G., B.H. Arison, O.D. Hensens, J. Hirschfleld, K. Hoogsteen, E.A.Kaczka, R.E. Rhodes, J.S. Kahan, F.M. Kahan, R.W. Ratcliffe, E. Walton, L.J.Ruswinkle, R.B. Morin, and B.G. Christensen. 1978. Structure and absoluteconfiguration of thienamycin. J. American Chemical Society. 100: 6491-6499.Angus, B.L., A.M. Carey, D.A. Caron, A.M.B. Kropinski, and R.E.W. Hancock. 1982.Outer membrane permeability in Pseudomonas aeruginosa: comparison of a wild-type with an antibiotic-supersusceptible mutant. Antimicrob. Agents Chemother.12: 299-309.Agterberg, M., H. Adriaanse, E. Tijhaar, A. Resink, and J. Tommassen. 1989. Roleof the cell surface-exposed regions of outer membrane protein PhoE of Escherichiacoli K12 in the biogenesis of the protein. Eur. J. Biochem. 185: 365-370.Atkinson, T. and M. Smith. 1984. Purification of oligonucleotides obtained by smallscale (0.2 jimol) automated synthesis by gel electrophoresis. In “OligonucleotideSynthesis: A practical Approach”, M.J. Gait (ed.). IRL Press, Oxford, 35-8 1.Barza, M. 1985. Imipenem: first of a new class of beta-lactam antibiotics. Annalsof Internal Medicine. 103: 552-560.Bauer, K., M. Struyvé, D. Bosch, R. Benz, and J. Tommassen. 1989. One singleresidue is responsible for the special interaction between polyphosphate and theouter membrane porin PhoE of Escherichia coli. J. Biol. Chem. 264: 16393-16398.Bellido, F., N.L. Martin, R.J. Siehnel, and R.E.W. Hancock. 1992. Reevaluation,using intact cells, of the exclusion limit and role of porin OprF in Pseudomonasaeruginosa outer membrane permeability. J. Bacteriol. 174: 5196-5203.Benson, S.A., J.L.L. Occi, and B.A. Sampson. 1989. Mutations that alter the porefunction of the OmpF porin of Eseherichia coli K12. J. Mol. Biol. 203: 96 1-970.Benz, R., K. Janko, and P. Lauger. 1978. Formation of large ion-permeablemembrane channels by the matrix protein (porin) of Escherichia coli. Biochim.Biophys. Acta 511: 238-247.Benz, R., and R.E.W. Hancock. 1981. Properties of the large ion-permeable poresformed from protein F of Pseudomonas aeruginosa in lipid bilayer membranes.Biochim. Biophys. Acta 646: 298-308.Benz, R., A. Schmid, and R.E.W. Hancock. 1985. Ion selectivity of gram-negativebacterial porins. J. Bacteriol. 162: 722-72 7.151Benz, R.A., A. Schmid, T. Nakae, and G.H. Vos-Scheperkeuter. 1986. Poreformation by LamB of Escherichia coli in lipid bilayer membranes. J. Bacteriol.165: 978-986.Benz, R., A. Schmid, and G.H. Vos-Scheperkeuter. 1987. Mechanism of sugartransport through the sugar-specific LamB channel of Escherichia coli outermembrane. J. Membr. Biol. 100: 2 1-29.Benz, R.A., A. Schmid, T. Nakae, C, Maier, and E. Bremer. 1988. Characterizationof the nucleotide-binding site inside the Tsx channel of Escherichia coli outermembrane reconstituted experiments with lipid bilayer membranes. Eur. J.Biochem. 176: 699-705.Benz, F., A. Schmid, W. Wagner, and W. Doebel. 1989. Pore formation by theEscherichia coli hemolysin: evidence for an association-dissociation equilibrium ofpore-forming aggregates. Infec. Immun. 57: 887-895.Boffivar, D.W., and C.E. Bauer. 1992. Association of tetrapyrrole intermediates inthe bateriochlorophyll a biosynthetic pathway with the major outer-membraneporin protein of Rhodobacter capsulatus. Biochem. J. 282: 471-476.Bosch, D., and J. Tommassen. 1987. Effects of linker insertion on the biogenesisand functioning of the Escherichia coli outer membrane pore protein PhoE. Mol.Gen. Genet. 208: 485-489.Bosch, D., W. Voorhout, and J. Tommassen. 1988. Export and localization of N-terminally truncated derivatives of Escherichia coli K- 12 outer membrane proteinPhoE. J. Biol. Chem. 263: 9952-9957.Bosch, D., M. Scholten, C. Verhagen, and J. Tommassen. 1989. The role of carboxyterminal membrane-spanning fragment in the biogenesis of Escherichia coli K-12outer membrane protein PhoE. Mol. Gen. Genet. 216: 144-148.Boulain, J.C., A. Charbit, and M. Hofnung. 1986. Mutagenesis by random linkerinsertion into the lamB gene of Escherichia coli K-12. Mol. Gen. Genet. 205: 339-348.Brown, M.R.W. The role of the cell envelope in resistance. In “Resistance ofPseudomonas aeruginosa”, M.R.W. Brown (Ed.), printed in Great Britain by J. W.Arrowsmith Ltd, Bristol, 71-107.Bryan, L.E. 1979. Resistance to antimicrobial agents: The general nature of theproblem and the basis of resistance. In “Pseudomnonas aeruginosa: ClinicalManifestations of Infection and Current Therapy”, R.G. Doggett (Ed.), Academic152Press, New York, 2 19-270.Büscher, K.H., W. Cullmann, W. Dick, and W. Opferkuch. 1987. Imipenemresistance in Pseudomonas aeruginosa resulting from diminished expression of anouter membrane protein. Antimicrob. Agents Chemother. 31: 703-708.Catchpole, C.R., R. Wise, D. Thornber, and J.M. Andrews. 1992. In vitro activity ofL-627, a new carbapenem. Antimicrob. Agents Chemother. 36: 1928- 1934.Comette, J.L., K.B. Cease, H. Margalit, J.L. Spouge, J.A. Berzofsky, and C. Delisi.1987. Hydrophobicity scales and computational techniques for detectingamphipathic structures in proteins. J. Mol. Biol. 195: 659-685.Cowan, S.W., T. Schirmer, G. Rummel, M. Steiert, R. Ghosh, R.A. Pauptit, J.Jansonius, and J.P. Rosenbush. 1992. Crystal structures explain functionalproperties of two E. coli porins. Nature 358: 727-733.Cronan, J.E., R.B. Gennes, arid S.R. Maloy. 1987. Cytoplasmic membrane. In“Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology”,F.C. Neidhardt (ed.) ASM Publications, Washington, D.C., 3 1-55.Devereux, J., P. Flaeberli, and 0. Simthies. 1984. A comprehensive set of sequenceanalysis programs for the VAX. Nucl. Acids Res. 12: 387-395.Ditta, G., T. Schmidhauser, E. Yakobson, P. Lu, X.W. Liang, D.R. Finlay, D.Guiney, and D.R. Helinski. 1985. Plasmids related to the broad host range vector,pRK29O, useful for gene cloning and for monitoring gene expression. Plasmid 13:149-153.Drewry, D.T., K.C. Symes, G.W. Gray, and S.G. Wilkinson. 1975. Studies ofpolysaccharide fractions from the lipopolysaccharide of Pseudomonas aeruginosaNCTC 1999. Biochem. J. 149: 93- 106.Eisenberg, D., R.M. Weiss, and T.C. Terwilliger. 1982. The helical hydrophobicmoment: a measure of the amphiphilicity of a helix. Nature 299: 37 1-374.Ferenci, T., and K.S. Lee. 1982. Direct evolution of the lambda receptor ofEscherichia coli through affinity chromatographic selection. J. Mol. Biol. 160: 431-444.Foulds, J. 1976. toiF Locus in Escherichia coli: chromosomal location andrelationship to loci cmlB and tolD. J. Bacteriol. 128: 604-608.Freundlieb, S., U. Ehmann, and W. Boos. 1988. Facilitated diffusion of p153nitrophenyl-ft-D-maltohexaoside through the outer membrane of Escherichia coli.J. Biol. Chem. 263: 3 14-320.Fukuda, H., M. Hosaka, K. Hirai, and S. lyobe. 1990. New norfioxacin resistancegene in Pseudomonas aeruginosa PAO. Antimicrob. Agents Chemother. 34: 175 7-1761.Fukuoka, T., N. Masuda, T. Takenouchi, N. Sekine, M. Tijima, and S. Ohya. 1991.Increase in susceptibility of Pseudoinonas aeruginosa to carbapenem antibiotics inlow-amino-acid media. Antimicrob. Agents Chemother. 35: 529-532.Fukuoka, T., S. Ohya, T. Narita, M. Katsuta, M. lijima, N. Masuda, H. Yasuda, J.Trias, and H. Nikaido. 1993. Activity of carbapenem panipenem and role of theOprD (D2) protein in its diffusion through the Pseudoinonas aeruginosa outermembrane. Antimicrob. Agents Chemother. 37: 322-327.Fung-Tomc, J. C., E. Huczko, J. Banville, M. Menard, B. Kolek, E. Gradeiski, R. E.Kessler and D. P. Bonner. 1995. Structure-activity relationships of carbapenemsthat determine their dependence on porin protein D2 for activity againstPseudomonas aeruginosa. Antimicrob. Agents Chemother. 39: 386-393.Gehering, K., C.H. Cheng, H. Nikaido, and B.K. Jap. 1991. Stoichiometry ofmaltodextrin-binding sites in LamB, an outer membrane protein from Escherichiacoli. 173: 1873-1878.Gotoh, N., H. Wakebe, E. Yoshihara, T. Nakae, T. Nishino, 1989. Role of protein Fin maintaining structural integrity of the Pseudornonas aeruginosa outermembrane. J. Bacteriol. 171: 983-990.Hanahan, D. 1983. Studies on the transformation of Escherichia coli with plasmids.J. Mol. Biol. 166: 557-580.Hancock, R.E.W. and A.M. Carey. 1979. Outer membrane of Pseudomonasaeruginosa. Heat- and 2-mercaptoethanol-modifiable proteins. J. Bacteriol. 140:902-910.Hancock, R.E.W. and A.M. Carey. 1980. Protein D1-a glucose-inducible, pore-forming protein from the outer membrane of Pseudoinonas aeruginosa. FEMSMicrobiol. Lett. 8: 105-109.Hancock, R.E.W. 1981. Aminoglycoside uptake and mode of action-with specialreference to streptomycin and gentamicin. J. Antimicrob. Chemother. 1981: 249-276.154Hancock, R.E.W., R.T. Irvin, J.W. Costerton, and A.M.Carey. 1981a. Pseuclomonasaeruginosa outer membrane: peptidoglycan-associated proteins. J. Bacteriol. 145:628-631.Hancock, R.E.W., V.J. Raffle, and T.I. Nicas. 198 lb. Involvement of the outermembrane in gentamicin and streptomycin uptake and killing in Pseudomonasaeruginosa. Antimicrob. Agents Chemother. 19: 777-785.Hancock, R.E.W., K. Poole, and R. Benz. 1982. Outer membrane protein P ofPseudoinonas aeruginosa: regulation by phosphate deficiency and formation ofsmall anion-specific channels in lipid bilayer membranes. J. Bacteriol. 150: 730-738.Hancock, R.E.W. 1984. Alteration in outer membrane permeability. Aim. Rev.Microbiol. 38: 237-264.Hancock, R.E.W., and R. Benz. 1986. Demonstration and chemical modification ofa specffic phosphate binding site in the phosphate-starvation-inducible outermembrane porin protein P of Pseudoinonas aeruginosa. Biochim. Biophys. Acta 860:669-707.Hancock, R.E.W. 1987. Role of porins in outer membrane permeability. J. Bacteriol.169: 929-933.Hancock, R.E.W., and A. Bell. 1988. Antibiotic uptake into gram-negative bacteria.Eur. J. Clin. IVlicrobioL 7: 7 13-720.Hancock, R.E.W., R. Siehnel, and N. Martin. 1990. Outer membrane proteins ofPseudomonas. Mol. Microbiol. 4: 1069-1075.Hancock, R.E.W., C. Egli, R. Benz, and R.J. Siehnel. 1992. Overexpression inEscherichia coli and functional analysis of a novel PPi-selective porin, OprO, fromPseudomonas aeruginosa. J. Bacteriol. 174: 47 1-476.Hancock, R.E.W., D.N. Karunaratne, and C.B. Egli. 1994. Molecular organizationand structural role of outer membrane macromolecules. In “Bacterial Cell Wall”,J.M. Ghuysen and R. Hakenbeck (Eds.) Elsevier Science B.V., 263-2 79.Hashizume, T., F. Ishino, J.I. Nakagawa, S. Tamaki, and M. Matsuhashi. 1984.Studies on the mechanism of action of imipenem (N-formimidoyl thienamycin) invitro: binding to the penicillin-binding proteins (PBPs) in Escherichia coli andPseudomonas aeruginosa, and inhibition of enzyme activities due to the PBPs in E.coli. J. Antibiotics. 37: 394-400.155Hellinger, W.C., and N.C. Brewer. 1991. Imipenem. Mayo Clin. Proc. 66: 1074-108 1.Ho. S.N., H.D. Hunt, R.M. Horton, J.K. Pullen, and L.R. Pease. 1989. Site-directedmutagenesis by overlap extension using polymerase chain reaction. Gene 77: 51-59.Hofnung, M. 1995. An intelligent channel (and more). Science 267: 473-474.Huang, H. and R.E.W. Hancock. 1993. Genetic definition of the substrate selectivityof outer membrane porin protein OprD of Pseudomonas aeruginosa. J. Bacteriol.175: 7793-7800Huang, H., R. Siehnel, F. Beffido, E. Rawling, and R.E.W. Hancock. 1992. Analysisof two gene regions involved in the expression of the imipenem-specific, outermembrane porin protein OprD of Pseudomonas aeruginosa. FEMS Microbiol. Lett.97: 267-274.Ishii, J., and T. Nakae. 1993. Lipopolysaccharide promoted opening of the porinchannel. FEBS Lett. 3: 25 1-255.Jeanteur, D., J.H. Lakey, and F. Pattus. 1991. The bacterial porin superfamily:sequence alignment and structure prediction. Mol. Microbiol. 5: 2153-2164.Jeanteur, D., J.H. Lakey, and F. Pattus. 1994a. The porin superfamily: diversityand common features. In “Bacterial Cell Wall”. Ghuysen, J.M., and Hakenbeck, R.(Eds), Amsterdam: Elsevier Science, 363-380.Jeanteur, D., T. Schirmer, D. Fourel, V. Simonet, G. Rummel, C. Widmer, J.P.Rosenbush, F. Pattus, and J.M. Pages. 1994b. Structural and functional alterationsof a colicin-resistant mutant of OmpF porin from Escherichia coli. Proc. Nati. Acad.Sci. USA. 91: 10675-10679.Kahan, J. S., F.M. Kahan, R. Goegelman, S.A. Currie, M. Jackson, E.0. Stapley,T.W. Miller, A.K. lVliuler, D. Hendlin, S. Mochales, S. Hernandez, H.B. Woodruff,and J. Birnbaum. 1979. Thienamycin, a new -lactam antibiotic. I. Discovery,taxonomy, isolation and physical properties. J. Antibiot. 32: 1-12.Kahan, F.M., H. Kropp, J.G. Sundelof, and J. Birnbaum. 1983. Thienamycin:development of imipenem-cilastatin. J. Antimicrob. Chemother. 12(suppl): S1-S35.Karunaratne, D.N., J.C. Richards, and R.E.W. Hancock. 1992. Characterization oflipid A from Pseudo,nonas aeruginosa 0-antigen B band lipopolysaccharide by 1Dand 2D NMR and mass spectral analysis. Arch. Biochem. Biophy. 299: 368-376.Kieffel, B., R.M. Garavito, W. Baumeister, and J.P. Rosenbush. 1985. Secondary156structure of a channel-forming protein from E. coli outer membrane. EMBO. J. 4:1589-1592.Kreusch, A., A. Neubüser, E. Schiltz, J. Weckesser, and G.E. Schulz. 1994.Structure of the membrane channel porin from Rhodopseudornonas blastica at 2.0A resolution. Protein Sci. 3: 58-63.Kropinski, A.M., L.C. Chan, and F.H. Milazzo. 1979. The extraction and analysisof lipopolysaccharide from Pseudomonas aeruginosa strain PAO, and three roughmutants. Can. J. lVlicrobiol. 25: 390-398.Kropinski, A.M., B. Jewell, J. Kuzio, F. Milazzo, and D. Berry. 1985. Structure andfunctions of Pseudomonas aeruginosa lipopolysaccharide. In “Pseudoinonasaeruginosa: Nre Therapeutic Approaches from Basic Research”, D.P. Speert andR.E.W. Hancock (Eds.) Karger, Basel, Switzerland, 58-73.Kropp, H., J.G. Sundelof, J.S. Kahan, F.M. Kahan, and J. Birnbaum. 1980.MK0787 (N-formimidoyl thienamycin): evaluation of in vivo and in vitro activities.Antimicrob. Agents Chemother. 1980. 17: 993-1000.Kropp, H., J.G. Sundelof, R. Hajdu, and F.M. Kahan. 1982. Metabolism ofthienamycin and related carbapenem antibiotics by the renal dipeptidase,dehydropeptidase-I. Antimicrob. Agents Chemother. 22: 62-70.Kropp H., L. Gerckens, J.G. Sundelof, and F.M. Kahan. 1985. Antibacterial activityof imipenem: the first thienamycin antibiotic. Rev. Infec. Dis. 7 (Suppl 3): S389-S410.Kyte, J. and R.F. Doolittle. 1982. A simple method for displaying the hydropathiccharacter of a protein. J. Mol. Biol. 157: 105-132.Lakey, J.H., E.J.A. Lea, and F. Pattus. 1991. OmpC mutants which allow growthon maltodextrins show increased channel size and greater voltage sensitivity. FEBSLett.. 278: 3 1-34.Leanza W.J., K.J. Wildonger, T.W. Miller, and B.G. Christensen. 1979. Nacetimidoyl- and N-formimidoylthienamycin derivatives: antipseudomonal 3 -lactam antibiotics (letter). J. Med. Chem. 22: 1435-1436.Lee, E.H., M.H. Nicolas, M.D. Kitzis, G. Pialoux, E. Collatz, and L. Gutmann. 1991.Association of two resistance mechanism in a clinical isolate of Enterobacter cloacaewith high-level resistance to imipenem. Antimicrob. Agents Chemother. 35: 1093-1098.157Lipman B., and H.C. Neu. 1988. Imipenem: a new carbapenem antibiotic. Med.Clin. North. Am. 72: 567-579.Liu, P.V. 1974. Extracellular toxins of Pseudomonas aeruginosa. J. Infect. Dis. 130(suppi): S94-S99.Livermore, D.M., and Y.J. Yang. 1987. p-lactamase lability and inducer power ofnewer [3-lactam antibiotics in relation to their activity against [3-lactamase-inducibility mutants of Pseudoinonas aeruginosa. J. Infect. Dis. 155: 775-782.Livermore, D.M., and Y. Yang. 1989. Comparative activity of meropenem againstPseudomonas aeruginosa strains with well-characterized resistance mechanism. J.Antimicrob. Chemother. 24: S149-S159.Livermore, D.M. 1992. Interplay of impermeability and chromosomal betalactamase activity in imip enem-resistant Pseudomonas aeruginosa. Antimicrob.Agents. Chemother. 36: 2046-2048.Lugtenberg, B. and L. van Aiphen. 1983. Molecular architecture and functioningof the outer membrane of Escherichia coli and other gram-negative bacteria.Biochim. Biophys. Acta 737: 5 1-115.Lynch, M.J., G.L. Drusano, and fl.L.T. Mobley. 1987. Emergence of resistance toimipenem in Pseudomonas aeruginosa. Antimicrob. Agents. Chemother. 31: 1892-1896.Manoil, C., and J. Beckwith. 1986. A genetic approach in analyzing membraneprotein topology. Science 233: 1403-1408.Marchal, C., and M. Hofnung. 1983. Negative dominance in gene lamB: randomassembly of secreted subunits issued from different polysomes. EMBO J. 2: 81-86.Massidda, 0., G.M. Rosso]irii, and G. Satta. 1991. The Aeromonas hydrophila cphAgene: molecular heterogeneity among class B mettalo- [3-lactamases. J. Bacteriol.173: 4611-46 17.Michea-Hamzhehpour, M., Y.X. Furet, and J.C. Pechere. 199 la. Role of protein D2and lipopolysaccharide in diffusion of quinolones through the outer membrane ofPseudomonas aeruginosa. Antimicrob. Agents Chemother. 35: 209 1-2097.Michea-Hamzhehpour, M., C. Lucain, and J.C. Pechere. 199 lb Resistance topefloxacin in Pseudomonas aeruginosa. Antimicrob. Agents Chemother. 35: 512-518.158Misra, R., and S.A. Benson. 1988. Genetic identification of the pore domain of theOmpC porin of Escherichia coli K12. J. Bacteriol. 170: 36 11-3617.Mizuno, T. 1979. A novel peptidoglycan-associated lipoprotein found in the cellenvelope of Pseudoinonas aeruginosa and Escherichia coli. J. Biochem. 86: 991-1000.Morrison, D.C. 1985. Nonspecific interaction of bacterial lipopolysaccharides withmembranes and membrane components. In “Handbook of Endotoxin”, L.J. Berry(Ed.) Elsevier, Amsterdam, Vol 3: 25-55.Mutharia, L.M., and R.E.W. Hancock. 1983. Surface localization of Pseudomonasaeruginosa outer membrane porin protein F using monoclonal antibodies. Infect.Immun. 42: 1027-1033.Needleman, S.B., and C.D. Wunsch. 1970. A general method applicable to theresearch for similarities in the amino acid sequence of two proteins. J. Mol. Biol. 48:443-453.Nicas, T.I. 1983. Ph.D thesis “Role of the outer membrane of Pseudoinonasaeruginosa in antibiotic resistance”. Univ. British Columbia, Canada.Nicas, T.I., and R.E.W. Hancock. 1983a. Pseudoinonas aeruginosa outer membranepermeability: isolation of a porin protein F-deficient mutant. J. Bacteriol. 153: 281-285.Nicas, Ti., and R.E.W. Hancock. 1983b. Alterations of susceptibility to EDTA,Polymyxin B and Gentamicin in Pseudoinonas aeruginosa by divalent cationregulation of outer membrane protein Hl. J. Gen. lVlicrobiol. 129: 509-5 17.Nikaido, H. and T. Nakae. 1979. The outer membrane of Gram-negative bacteria.Adv. Microbiol. Physiol. 20: 163-250.Nikaido, H., and E.Y. Rosenberg. 1981. Effect of solute on diffusion rates throughthe transmembrane pore of the outer membrane of B. coil. J. Gen. Physiol. 77: 121-?Nikaido, H., E.Y. Rosenberg, and J. Foulds. 1983. Porin channels in Escherichiacoli: studies with -lactams in intact cell. J. Bacteriol. 153: 232-240.Nikaido, H. and R.E.W. Hancock. 1986. Outer membrane permeability ofPseudomonas aeruginosa. In “The Bacterial”, J.R. Sokatch (ed.). Academic Press,New York, VolX: 145-193.Nikaido, H., and M. Vaara. 1985. Molecular basis of bacterial outer membrane159permeability. Microbiol. Rev. 49: 1-32.Oliver, D.B. 1987. Periplasm and protein secretion. In “Eseherichia coli andSalmonella typhiinurium: Cellular and Molecular Biology”, F.C. Neidhardt (Ed.)ASM Publications, Washington, D.C., 56-69.Olsen, R.H., W. DeBusscher, and W.R. McCombie. 1982. Development of broad hostrange vectors and gene banks: self-cloning of the Pseudomonas aeruginosa PAOchromosome. J. Bacteriol. 150: 60-69.Osano, E., Y. Arakawa, R.W. Yankun, M. Ohta, T. Horn, H. Ito, F. Yoshimura, andN. Kato. 1994. Molecular characterization of the enterobacterial metallo 3-lactamase found in a clinical isolate of Serratia marcescens that shows imipenemresistance. Antimicrob. Agents Chemother. 38: 7 1-78.Paul, C. and J. Rosenbush. 1985. Folding patterns of porin and bacteriorhodopsin.EMBO. J. 4: 1593-1597.Pollack, M. 1990. Pseudomonas aeruginosa. In “Principles and Practice of InfectiousDiseases”, G.L. Mandell, R.G. Douglas and J.E. Bennett (Eds). ChurchillLivingstone, New York, 1673-1691.Poole. K. and R.E.W. Hancock. 1986. Isolation of a Tn50 1 insertion mutant lackingporin protein P of Pseudomonas aeruginosa. Mol. Gen. Genet. 202: 403-409.Quinn, J.P., E.J. Dudek, C.A. DiVincenzo, D.A. Lucks and S.A. Lerner. 1986.Emergence of resistance to inzipenem during therapy for Pseudomonas aeruginosa.J. Infect. Dis. 154: 289-294.Rietschel, E.T., H. Mayer, H.W. Wollenweber, U. Zahringer, 0. Luderitz, 0.Westphal, and H. Brade. 1984. Bacterial lipopolysacchanides and their lipid Acomponent. In “Bacterial Endotoxin: Chemical, Biological and Clinical Aspects”,J.Y. Homma, S. Kanegasaki, 0. Ludenitz, T. Shiba, and 0. Westphal (Eds.) VerlagChemie, Weinheim, pp. 11-22.Rolinson, G.N. 1986. -lactam antibiotics. J. Antimicrob. Chemother. 17: 5-36.Rottem, S., and L. Leive. 1977. Effect of variations in lipopolysaccharides on thefluidity of the outer membrane of Escherichia coli. J. Biol. Chem. 252: 2077-208 1.Saiki, R.K., S. Scharf, F. Faloona, K.B. Muffis, G.T. Horn, H.A. Erlich, and N.Arnheim. 1985. Enzymatic amplification of -globin genomic sequences andrestriction site analysis for diagnosis of sickle cell anemia. Science. 230: 1350-1354.160Saino, Y., F. Kobayashi, M. Inoue, S. lVlitsuhashi. 1982. Purification and propertiesof inducible penicillin -lactamase isolated from Pseudomonas maltophilia.Antimicrob. Agents Chemother. 22: 564-570.Sambrook, J., E.F. Fritisch, and T. Maniatis. 1989. Molecular Cloning. ALaboratory Manual. Cold Spring Harbour Laboratory Press, Cold Spring Harbour,New York.Sandermann and Strominger. 1972. Purification and properties ofC55-isoprenoidalcohol phosphokinase from Staphylococcus aereus. J. Biol. Chem. 247: 5123-5131.Satake, S., E. Yoshihara, and T. Nakae. 1990. Diffusion of -lactam antibioticsthrough liposome membranes reconstituted from purified porins of the outermembrane of Pseuclornonas aeruginosa. Antimicrob. Agents Chemother. 34: 685-690.Schimpif, S.C., M. Moody, V.M. Young. 1970. Relationship of colonization withPseud,omonas aeruginosa to develop of Pseuclomonas bacteremia in cancer patients.Antimicrob. Agents Chemother. 10: 240-244.Schirmer, T., T.A. Keller, Y.F. Wang, and J.P. Rosenbush. 1995. Structural basisfor sugar translocation through maltoporin channels at 3.1 A resolution. Science267: 512-514.Schülein, K., R. Ebner, K. Jaheris, J.W. Lengeler, and F. Titgemeyer. 1991. Asugar-specific porin, ScrY, is involved in sucrose uptake in enteric bacteria. Mol.lVlicrobiol. 5: 941-950.Schulz, G.E. 1993. Bacterial porins: structure and function. Current Opinion in CellBiology 5: 70 1-707.Schweizer, H.P. 1991. Escherichia-Pseudo,nonas shuttle vectors derived frompUC18/19. Gene 97: 109-112.Schweizer, H.P. 1992. Allelic exchange in Pseudomonas aeruginosa using novelColE 1-type vectors and a family of cassettes containing a portable oriT and thecounter-selectable Bacillus subtilis sacB marker. Mol. Microbiol. 6: 1195-1204.Sekar, V. 1987. A rapid screening procedure for the identification of recombinantbacterial clones. 5: 11-13Sherbert, G.V., M.S. Lakshmi. 1973. Characterization of Escherichia coli cellsurface by isoelectric equilibrium analysis. Biochim. Biophys. Acta 298: 50-58.161Siehnel, R.J., N.L. Matin, and R.E.W. Hancock. 1990. Sequence and relatedness inother bacteria of the Pseudoinonas aeruginosa oprP gene coding for the phosphate-specific porin P. Mol. Microbiol. 4: 83 1-838.Siehnel, R.J., C. Egli, and R.E.W. Hancock. 1992. Mol. lVlicrobiol. Polyphosphateselective porin OprO of Pseudoinonas aeruginosa: expression, purification andsequence. Mol. Microbiol. 6: 23 19-2326.Simon, R., U. Preifer, and A. Pühler. 1983. A broad host range mobilization systemfor in vivo genetic engineering: transposon mutagenesis in Gram negative bacteria.Bio/Technology 1: 784-790.Sykes, R.B., and A. Morris. 1975. Resistance of Pseudoinonas aeruginosa toantimicrobial drugs. Prog. Med. Chem. 12: 333-93.Spurlino, J.C., G-Y. Lu, and F.A. Quiocho. 1991. The 2.3-A resolution structure ofthe maltose or maltodextrin-biiiding protein, a primary receptor of bacterial activetransport and chemotaxis. J. Biol. Chem. 266: 5202-5219.Tausk F., M.E. Evans, L.S. Patterson, C.F. Federspiel, and C.W. Stratton. 1985.Imipenem-induced resistance to antipseudomonal -lactams in Pseudornonasaeruginosa. Antimicrob. Agents Chemother. 28: 41-45.Thomson, K.S., C.C. Sanders, and H. Chmel. 1993. Imipenem resistance inEnterobacter. Eur. J. Clin. IVlicrob. Infect. 12: 610-613.Towbin, M., T. Staehlin, and J. Gordon. 1979. Electrophoretic transfer of proteinsfrom polyacrylamide gels to nitrocellulose sheets: procedure and some applications.Proc. Natl. Acad. Sci. U.S.A. 76: 4350-4354.Trias, J., E.Y. Rosenberg, and H. Nikaido. 1988. Specificity of the glucose channelformed by protein Dl of Pseudomonas aeruginosa. Biochim. Biophys. Acta 938: 493-496.Trias, J., J. Dufresne, R.C. Levesque, and H. Nikaido. 1989. Decreased outermembrane permeability in imipenem-resistant mutants of Pseudornonasaeruginosa. Antimicrob. Agents Chemother. 33: 120 1-1206.Trias, J., and H. Nikaido. 1990a. Outer membrane protein D2 catalyzes facilitateddiffusion of carbapenems and penems through the outer membrane of Pseudomonasaeruginosa. Antimicrob. Agents Chemother. 34: 52-57.Trias, J., and H. Nikaido. 1990b. Protein D2 channel of the Pseudornonasaeruginosa outer membrane has a binding site for basic amino acids and peptides.162J. Biol. Chem. 265: 15680-15684.Vallette, F., E. Mege, A. Reiss, and M. Adesnik. 1989. Construction of mutant andchimeric genes using the polymerase chain reaction. Nuci. Acids Res. 17: 723-733.Van der Ley, P., H. Amesz, J. Tommassen, and B. Lugtenberg. 1985. Monoclonalantibodies directed against the cell surface-exposed part of PhoE pore protein of theEscherichia coli K-12 outer membrane. Eur. J. Biochem. 147: 40 1-407.Vogel, H. and D.M. Bonner. 1956. Acetylornithase of E. coli: partial purificationand properties. J. Biol. Chem. 218: 97-106.Vogel, H., and F. Jahnig. 1986. Models for the structure of outer membraneproteins of Escherichia coli derived form Raman spectroscopy and predictionmethods. J. Mol. Biol. 190: 19 1-199.Weiss, M.S., U. Abele, J. Weckesser, W. Welte, E. Schiltz, and G.E. Shulz. 1991.Molecular architecture and electrostatic properties of a bacterial porin. Science 254:1627-1630.Weiss, M.S. and G.E. Schulz. 1992. Structure of porin refined at 1.8 A resolution.J. Mol. Biol. 207: 493-509.Winberg, G. and M.L. Hammarskjörd. 1980. Isolation of DNA from agarose gelsusing DEAE paper. Application to restriction site mapping of adenovirus type 16DNA. Nucl. Acids Res. 8: 253-264.Wong, S.Y., H. Jost, and R.E.W. Hancock. 1993. Linker-insertion mutagenesis ofPseudomonas aeruginosa outer membrane protein OprF. Mol. Microbiol. 10: 283-292.Woodruff, W.A., T.R. Parr, R.E.W. Hancock, L.F. Hanne, T.I. Nicas, and B.H.Iglewski. 1986. Expression in Escherichia coli and function of Pseudomonasaeruginosa outer membrane porin protein F. J. Bacteriol. 167: 473-479.Woodruff, W.A. and R.E.W. Hancock. 1988. Construction and characterization ofPseudomonas aeruginosa protein F-deficient mutants after in vitro and in vivoinsertion mutagenesis of the cloned gene. J. Bacteriol. 170: 2592-2598.Woodruff, W.A. and R.E.W. Hancock. 1989. Pseudornonas aeruginosa outermembrane protein F: structural role and relationship to the Escherichia coli OmpAprotein. J. Bacteriol. 171: 3304-3309.Yoneyama, H., E. Yoshihara, and T. Nakae. 1992. Nucleotide sequence of the163protein D2 gene of Pseudoinonas aeruginosa. Antimicrob. Agents Chemother. 36:1791-1793.Yoneyama, H., and T. Nakae. 1993. Mechanism of efficient elimination of proteinD2 in outer membrane of imipenem-resistant Pseudoinonas aeruginosa. Antiniicrob.Agents Chemother. 37: 2385-2390.Yoshishara, E., and T. Nakae. 1989. Identification of porins in outer membrane ofPseudornonas aeruginosa that form small diffusion pores. J. Biol. Chem. 264: 6297-6301.Yoshishara, E., H. Yoneyama, and T. Nakae. 1991. In vitro assembly of thefunctional porin trimer from dissociated monomers in Pseudoinonas aeruginosa. J.Biol. Chem. 266: 952-957.Yoshimura, F., and H. Nikaido. 1982. Permeability of Pseudomonas aeruginosaouter membrane to hydrophilic solutes. J. Bacteriol. 152: 636-642Yoshimura, F., and H. Nikaido. 1985. Diffusion of -1actam antibiotics through theporin channels of Escherichia coli K-12. Antimicrob. Agents Chemother. 27: 84-92.Yoshinori, Y., T. Nishikawa, and Y. Komatsu. 1993. Cloning and nucleotidesequence of anaerobically induced porin protein El (OprE) of Pseudoinonasaeruginosa PAO1. Mol. lVlicrobiol. 8: 993-1004.Yotsuji, A., S. Minami, M. Inoue, and S. Mitsuhashi. 1983. Properties of novel 3-lactamase produced by Bacteroides fragilis. Antimicrob. Agents Chemother. 24:925-929.Young, L.S. 1984. The clinical challenge of infections due to Pseudoinonasaeruginosa. Rev. Infec. Dis. 6: 5603-607.Zhou, X.Y., M. Kitzis, and L. Gutmann. 1993. Role of cephalosporinase incarbapenem resistance of clinical isolates of Pseudomonas aeruginosa. Antimicrob.Agents Chemother. 37: 1387-1389.164APPENDIXA. Revised OprD Model by Multiple Alignments and Amphipathicity Calculations.(a) MULTIPLE ALIGNMENTS. Alignments of closely related sequenceswere performed using classical alignment tools such as those available in the GCGpackage (Devereux et al., 1984). For distant sequences such tools are not veryaccurate, mainly because they tend to introduce gaps which incur the same penaltyall along the sequence. For porins, it is clear that the loop regions are much morevariable and even very long gaps may be easily introduced without problem.Conversely, insertion of gaps in the transmembrane regions should be heavilypenalized. Therefore that was taken into account manually in the final alignmentpredictions(b) MEMBRANE CRITERIA. The hydrophobicity ‘H’ of a segment wasdefined as usual (Kyte and Doolittle, 1982). The hydrophobicity of each residue wastaken from the PRIFT table (Cornette et at, 1987). The hydrophobic moment <it>was introduced by Eisenberg et al. (1982) to describe the amphipathicity of theprotein segments. The membrane criteria value, using a linear combination ofhydrophobicity and hydrophobic moment, H + <it>, determined the transmembranesegments quite precisely (Jeanteur et al., 1991).(c) TURN PREDICTIONS. According to Paul and Rosenbush (1985), a turncould be defined as a segment consisting of 3 or more residues containing at leastone turn promoter and no turn blockers. We refined this criteria by computing a165frequency matrix of residue occurrence within short periplasmic turns, externalloops and transmembrane strands from those porins with known structures (Cowanet at, 1992; Weiss and Schulz, 1992) and from a set of very closely related porins.Using this linear matrix of turn frequencies, we predicted turns by identifying themas segments of 3 residues with a ‘turn promoter’ propensity that was 3 times higherthan the ‘turn blocker’ propensity.In general, a n-strand was defined as a segment with high value of themembrane criteria, no gaps, no turn predictions and sequence conservation. Incontrast, a loop was defined by its low value of membrane criteria, the presence ofgaps, turn predictions and sequence variability.166TableXII:Homologiesandidentitiesamongtheporinsuperfamily.IdentitiesHIOMPPAECOOMPFOPRDOPRENMPORANSPORRHODORBALSTJCAH.influenzaE.coilP.aeruginosaP.aeruginosaNmeninglidisNsccaR.capsulatusR.blasticaP2OmpFOprDOprEPorAlPorHomologiesM93268M74489Z14065P32722X52995X65461P312443HIOMPPA21%9%8%17%15%14%15%ECOOMPF30%15%12%16%21%18%16%OPRD18%23%39%13%11%11%11%OPRE15%20%51%11%11%11%10%NIVIPORA27%26%22%22%47%16%13%NSPOR26%32%21%21%58%15%13%RHODO26%28%19%18%27%26%28%RBLASTICA25%26%18%17%24%25%40%AlignedsequenceswereanalyzedwithDISTANCE(Devereuxetal.,1984)usingathresholdof1.0foridentitiesand0.6forhomologies.Identitiesbetweensequencesareshownonthetoppartof thetablewhilehomologiesareinthebottom.ThecolumnheadingsgivethenameusedinFig.33,thespecies,genenameandaccessionnumber.AccessionnumbersbeginningwiththeletterParefromtheSwissportdatabase,othersarefromtheGenebank/EMBLdatabase.Wheretheaccessionnumberisprovided,thesequenceandgenenamecorrespondtothatentry.-ciFigure 33: Multiple sequence alignments between OprD and representatives ofporin superfamily.Membrane spanning strands are boxed in solved structures. The lines andnumbers under the alignment represent the predicted p-strands. The residueswhich were predicted to face the hydrophobic core of the membrane are shaded,some of them presenting certain poiar properties are in yellow. Aromatic residuesare shown in bold, charged residues involved in the eyelet are bold and in biggersize. Major turn predictions are indicated by *The hydrophobicity plus hydrophobic moment (H + <m>) was calculated from32 sequences (Jeanteur et ciL, 1994) and is shown in black. The hydrophobicmoment was calculated using a periodicity of 1/2 and 1/2.5 in order to take intoaccount untwisted and twisted p-strands. Each column represents a (H + <m>)calculated with a window of 9 residues centred at the current position. In yellowthe calculation is displayed for OprD sequence alone.1681.0-HIOMPPAECOOMPFOPNDOPRENMPORANSPORPRODORBLASTICA——NIOMPFADOFADIVNKYVIGNNT...FKKGRAKIIADEITTAEDKEYSLLNSKKYIPTNGNT6‘7MUlOGV6NLLAQERDLRTLDSRTNLS..1<ECOOMPFDAQTGNKTRIIF61<AID....VSDGRjNV.GVVYDAIGYTDMLPEFGGDTAYSDDFFVGRVGV.1RNISNFFGLVDIGNAQLGKNIEROPRSRAGGAVKVRISKIMKGEMQPTAVA•GGSR.LFPQTAT6FQLSSEFEGLDLC...AGHFTEGKEPTTVKSRGEYTAETAKSADFNPATDNLSASOFRESLGLTAKAKVSNTE6GTLQPKLVTNDGR.LLPVTFEGGQVTSTDLKDFTLV....AGQLEMSKGRNSTDNRSLSIRNSSASSRDSNK..FYY’SOKNKOLT10NAPORA...TQWGNRESFGAEFTNGRVANQFDOASOAIDPWDSNNSVASQLGIFKRHDDMPSRDSPEFSGFSGSQVIQNSKSAVTPAVYTKOT.NNNLILCDNSFOR...KGWGTRESFSE6FKNGKLNTALKDNSDSVDFWESSDANASVLOFGKLKNVSERKSRDSPFAGFSAS00RDNANPEDKHVPRODSGEDTFSAFKEG;DAL.SASEALFGDLYEvGY’rD...LDDRGGNDIPYLTGDERLTAEDN.VITSGASAS5105KVRBLASTICA....GTAGNAQWSNGTSGNVD.TAFDSVALTYDSEMGYEASSFGDAQSSFFAYNSKYDAS5ALDNVNGATSSF0NVSVDIPDQTV**________________45N71.5I0.5-I-AVVVNNRGTKK66SIEOSSSTEDNQEQQHGALRNQSSRHK’Tl<NFGD.6Y5VETRVSKASKEKAAEIVNKDCNIH0Y‘KVIKY5KGN6ENSV66N[pTYR6K€TQN5.101VQENQNNSSGASAFVSDQAEAKREDSS0LRNYVFNROGKSGSG0RDTQGITTESGFTQISSOFYGLKLDGTSOKTGTGNLPVMNSGKPRDDYS6IEOSK’SIRNFYINTDNRNGTASPSKQEEGQGINQSGFTQIREDLIGVRLDGSGRAGI(SSLSROPGTVFFLESNGEPVNOFA0SV0KG0RNYQLQLTEAQAANSSGASGQVKVTKVTKAKSRI611<1SD5SF61<SEDLGD.6K’VQEQDSAG6GA01VSK’SSSRTKETVNNVSTKNKTATEIADGSRSKHENLSN.NN‘I5EQNTSASTDEKSDRSMNGDDIWNFSSRSRLISTTDS...GE6SKNESGAKIE‘KNVR6QVKDR6V6LSIDTIISSRRNVITIETOO.SV6KkMQjS000KS***************123-,-HIOHPPASS5VTVSEVSNDQSKSN.NIIAASHrNYK05NHSYTQKIPKANAADASTOTTYPHHSKKQEVNGL5SSS..D6LSSSSYAKTKNKAKHEKSYFSSQECOOHPF..OTARKSINSOSSSSE...KSS‘IASADIRTLNQKAQPL5KGKKIAKQATSKD.N.NYANSK1KNATFITNKFTNTSGFANIK1Q5LVOPROLYSAKLGSIYRQ5NSNTPLASDQSSDNTRYNOESKAKASDISNTTSAATD.AHTTAQKVHSDQPFDYIRFSKNGSGASSDSFANSOFREYYYGNLDSFYKQHFSINNQI.SFVKDRFDSSSOGKNSSKSSRADSY.VSSDYYRS5VTK5EVONKASLTIS.RHS5.5QLNGDSDFPFLNRSDGEGSTIITONHPORAVPAVVSKPSSDVYRNKN...55ANAKAKNANVSRNAFELFI.ISS5SDQAK5TOPLKNHQVHR15EEG.SNAAQDLSENSDKTKNSTTEATNSFOR....NDVKTKHS05NEN...SRFSATAKRKVLSODHLELFNSNTTLSSGVYKSYQAYYRV50•N.NLSAQEGFKADVAOAKKNERTKATRH050..SETSEDSAQEAAA1..SNTSSEKTG1SPSYALMADMEQKA1KS.AITNKYAO5ELDRDFARAVFDLTFVAAAATAIVSHKASSSRBLASTICA....0SSLV1GKSASSN...DIMSAAITOAGSIV5KGIFSAKNSATSNYSNDLSTARIOQIIN*****NN151115-EMEDTN5NKKRN5VSQ5EKEREQALSSUKL..HKLTYEAYSRTRTTSVSSKQVA5KVKTEKEKSSKKYQSF5RSATKi]KDVESlSOVSLVNYFESTYNK..NSTYSINQISSONKLSVGSSSIASVQSSSFNSPSEKSQR0NLASYGVPSLTFMVRYINSKSIOGTKMSSNNVSYKNYGTSEOGKHHETNLEAKYVQSFA.KDSHRQAHRANASQGESSQNOFRIop5LSGKFQRASERTQKSSFATVGVPSLTFNTIYLSSSKIKTARGOQSEWERSIDL10TF.KSSIKNASFRSSLPAAGSSNNQRDQSENRISIPL5RFG...NAPRISYHFDFIERGKKSENTSYOQ50OFSKRTSISALKKNTGISNYTQINASSRHKSKMS...NVFRLSYHFKAKENSVKQSNSQYNQI50DFSKRTSLSSLKEGKSSSKTQSTVLSRHKSISATTSIQLollOTIDSYTI5.55SLSSGSVSADDLPNSDMIVOSKKAS.....SATTRISDIDIR.......ASASTIAYS5E.ASKSSSPANETVSGK—a1_i—H)SHQFASKOOMPFOPROOPRE—1A NSPOKRBLAS11CA—IFigure 34: Revised membrane topology model of OprD.The sixteen predicted transmembrane n-strands are boxed, and the 8external loops are labelled as Li to L8. The deleted amino acid residues arepresented as unfilled letters.17i£2£3TK%(©GL0GKA0TpG£6K£1GRHGYINIT80DLFNK340YKDTIN0FE£4£5RAyY360£8140NSSGG160GGRGRGTRTKAFGV©LSGA100AEGEGDMG©4@©LGDGTpLIIFELLGTLED©YITDAQ020RKApT200____S0NDFLGRNLV40L____FVIF280LGKHESDG70KLK____V210N300330WF______ApHHWLVR5LRyTNHAR41oTVR130RQITSGELQGIG17VNGTR370IGFSETyFHNIV390JFFAK11AlLL0LVALRDKLEFVLDTIAIANFANAsPKTVMFASG23ALKLL1A10TGLTR19NLAEVsoFKGyA270FYDS26LN320VIAQEG6WAlAGTQT220OsVGGI0LTG310RK______________L42vGTEEFFpAQAAFFGMi2oDLDTGsAAAHEQAK5KS


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items