OUTER MEMBRANE PROTEIN OprFof P. aeruginosabyEileen Grace RawlingB.Sc. (Microbiology), University of British Columbia, 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) required standardThe University of British ColumbiaAugust 1995© Eileen Grace Rawling, 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 I i i’iH Joc0(V ‘IThe University of British ColumbiaVancouver, CanadaDate______________DE-6 (2188)11ABSTRACTAn OprF-deficient mutant of P. aeruginosa strain M-2 wasconstructed by 2 mutagenesis. This strain was unable to grow in lowosmolarity media and was 70% the length of the parental strain. Theseresults confirmed that these phenotypes were not strain specific.Consistent with the appearance of OprF-deficient strains in clinicalsituations, an OprF-deficient strain was shown to not have a majorgrowth disadvantage in an in vivo chamber model.Plasmids encoding truncated and cysteine-to-serine mutants ofOprF were constructed by linker-insertion mutagenesis, PCR andsubcloning of a previously characterized mutant. Analysis of theresulting proteins indicated that between 102 and 163 N-terminal aminoacids of OprF was required for protein expression as determined byWestern immunoblotting. All of the truncated mutants expressed wereassociated with the outer membrane, indicating that the N-terminal 163amino acids of OprF were sufficient for this association. None of thetruncated-OprF mutants tested were associated with the peptidoglycan,indicating that more than 215 N-terminal amino acids of OprF arerequired for this association. Strains that encoded at least two cysteineswere 2-mercaptoethanol modifiable. The full-length cloned OprF and thetruncated mutant expressing the N-terminal 290 amino acids were111normally heat modifiable. When denatured prior to electrophoresis, theremaining truncated mutants had apparent molecular weights lowerthan those of untreated samples indicating that between 215 and 290amino acids were required for the protein to be normally heat modified.Although the truncated-OprF mutants were able to grow in lowosmolarity media and were significantly longer than the OprF-deficientstrain, the C-terminal half of OprF appeared to be required for the wild-type growth and length.The cysteine-to-serine OprF mutants were associated with theouter membrane and were heat modifiable. Analysis of these mutantswith and without 2-mercaptoethanol was consistent with the hypothesisthat wild-type OprF has two disulphide bondsMonoclonal antibody reactivity with overlapping octapeptidesequivalent to the primary sequence of OprF localized three epitopes.Indirect immunofluorescence and opsonic phagocytosis studies identifiedthe surface location of epitopes binding nine OprF-specific monoclonalantibodies. This information was incorporated into an updatedsecondary structure model of OprF.ivTABLE OF CONTENTSPageABSTRACT iiTABLE OF CONTENTS ivList of Tables ixList of Figures xACKNOWLEDGEMENTS xiiiINTRODUCTIONA. P. aeruginosa 1B. Gram-negative outer membranes1. Introduction 22. Phospholipids 43. LPS 44. Proteinsa). Lipoproteins 5b). Porins 6C. OmpA 9D. OprF 14E. Aims of this study 22VMATERIALS AND METHODSA. Bacterial strains and plasmids.23B. Media and growth conditions1. In vitro growth 232. In vivo growth 29C. DNA procedures1. Plasmid isolation 302. Plasmid introductiona). Transformation 30b). Electroporation 30c). Biparental mating 313. 2 mutagenesis of oprF 314. Southern blottinga). chromosomal DNA isolation 31b). Labeling of the DNA probe 32c). Southern blotting 325. Oligonucleotide synthesis 326. Polymerase chain reaction 337. DNA sequencing 33D. Protein procedures1. SDS-PAGE 342. Outer membrane preparation 34vi3. Preparation of peptidoglycan-associated proteins.... 354. Determination of protein concentration 35E. Immunological techniques1. Antibodies 352. Colony immunoblotting 363. Western immunoblotting 374. Immunofluorescent labeling 375. Opsonic phagocytosis 386. Overlapping-octapeptide analysis 39F. Cell-length measurement1. Image analysis 392. Microscopy 40G. Growth studies 40RESULTSCHAPTER 1. CONSTRUCTION AND EXPRESSION OF OprFMUTANTS IN P. aeruginosa.A. Introduction 42B. Construction of an OprF-deficient mutant ofP. aeruginosa strain, M-2 43C. Construction of truncated versions of OprF 48D. Site-directed mutagenesis of the cysteines of OprF 54viiE. Expression of OprF mutants in P. aenosa. 56F. Outer membrane and peptidoglycan association of OprFmutants 67G. Summary 72CHAPTER 2. FUNCTIONAL ANALYSIS OF OprF MUTANTS IN P.aeruginosa.A. Introduction 74B. Growth of OprF mutants 74C. Cell-length of OprF mutants 84D. Summary 89CHAPTER 3. MONOCLONAL ANTIBODY STUDIES.A. Introduction 91B. Linear-epitope mapping using overlapping octapeptides 91C. Monoclonal antibody binding to OprF mutants 96D. Surface accessibility of epitopes binding OprF-specificmonoclonal antibodies 100E. Conservation of epitopes binding OprF-specificmonoclonal antibodies 103F. Summary: A secondary structure model of OprF 104DISCUSSIONA. Aims of this study 111viiiB. Genetic complementation of (2-mutagenized OprF inP. aeruginosa 111C. Protein analysis of mutated versions of OprF 113D. Functional analysis of truncated-OprF mutants 119E. Analysis of the disulphide region of OprF 120F. Analysis of epitopes binding OprF-specific monoclonalantibodies 123G. Conclusion and perspectives 126REFERENCES 128ixLIST OF TABLESTable Title PaneI. Bacterial Strains 24II. Plasmids 26III. Amino acid residues introduced by linker-insertiontruncation of OprF 52IV. Characteristics of truncated-OprF mutants inP. aeruginosa 62V. Apparent molecular weight of cystreine-to-serine mutantsin P. aeruginosa 65VI. Relative length of truncated OprF in P. aeruginosa 87VII. Summary of OprF-specific monoclonal antibodyreactivity with truncated OprF in P. aeruginosa 97VIII. Summary of OprF-specific monoclonal antibodyreactivity with cysteine-to-serine mutants of OprF inP. aeniginosa 99IX. Indirect-immunofluorescence labeling of intactP. aeruginosa with OprF-specific monoclonalantibodies 101xLIST OF FIGURESFigure Title Page1. Schematic representation of the outer membrane of aGram-negative bacterial cell 32. Map of plasmid pWW2500 and restriction map of oprF 453. Southern blot of a biotinylated oprF probe hybridized tochromosomal DNA digests of P. aeruginosa strain M-2 474. Map of oprF and OprF showing the restrictionendonuclease sites and the length of the mature protein 505. Western immunoblot of whole cell lysates of H636containing plasmids encoding truncated OprF 586. Western immunoblots of OprF and truncated OprFmutants with and without TCA pretreatment 617. Western immunoblot of whole cell lysates of cysteine-toserine mutants 648. SDS-PAGE of outer membrane preparations of H 103/pER,H636/pER and H636/pER326t 689. Western immunoblot of outer membrane preparations oftruncated versions of OprF 69xiLIST OF FIGURES (continued).10. Western immunoblot of outer membrane preparationsof truncated OprF solubilized in Triton-EDTA or Tritonlysozyme 7111. InvivogrowthofHlo3andH636 7612. Growth of OprF-deficient strains in LB no salt orLB 200 mM salt 7713. Growth of H636/pER17O-26t and H636/pER with orwithout the addition of MgC12 7914. Growth of truncated versions of OprF in low osmolarityand high osmolarity media 8115. Western immunoblot of H636/pER29Ot, H636/pER andH103/pER 8316. Growth of cysteine-to-serine mutants in LB no salt orLB + 200 mM salt 8517. Length of strains of P. aeruginosa containing plasmidsencoding full-length or truncated versions of OprF 8618. ELISA readings of monoclonal antibodies MA7-1, MA7-2,and MA5-8 reacting with individual pins derivatized withoverlapping octapeptides from OprF 93xiiLIST OF FIGURES (continued).19. ELISA readings of mouse arid rabbit polyclonal serumreacting with individual pins derivatized withoverlapping octapeptides from OprF 9420. Opsonic phagocytosis of M-2 by OprF-specificmonoclonal antibodies 10221. Secondary structure model of OprF 10522. Proposed secondary structure model of OprF 107xiiiACKNOWLEDGEMENTSI would like to acknowledge the support and guidance of mysupervisor, R.E.W. Hancock, and the members of my supervisorycommittee. The financial assistance of the Canadian Cystic FibrosisFoundation and R.E.W. Hancock is gratefully acknowledged.I would also like to acknowledge the friendship and assistance ofthe members of the Hancock laboratory and the Department ofMicrobiology and Immunology.1INTRODUCTIONA. Pseudomonas aeruginosa.Pseudomonas aeruginosa is a rod-shaped, Gram-negativebacterium with a polar flagellum and polar p111. Its species name isdeduced from one of the extracellular pigments it produces, pyocyanin,which is the colour of oxidized copper. Other extracellular compoundsinclude several enzymes, toxins, and extracellular slime, all of whichcontribute to its growth versatility and pathogenicity (Young, 1980). P.aeruginosa is a common inhabitant of soil and water that presents littleproblem in healthy individuals, but is a serious problem in patients thathave lowered resistance to infection due to severe burns,immunosuppresive therapy, and for conditions that include cancer,transplants and cystic fibrosis (Blackwood and Pennington, 1981). It is anosocomial, opportunistic organism that can produce local infections inburn sites, the urinary tract, the respiratory tract, ears and eyes, as wellas a generalized septicemia. P. aeruginosa infections are difficult to treatdue to their intrinsic resistance to antibiotics (Bryan, 1979; Hancock andBell, 1988). P. aeruginosa is more resistant to antibiotics than otherGram-negative bacteria, particularly to the f3-lactam antibiotics (Hancockand Bell, 1988). The in vitro minimal inhibitory concentration (MIC) ofthese antibiotics is eight to one hundred times higher than that of2Escherichia coli (Rolinson, 1986). P. aeruginosa is also resistant to manyother compounds including detergents, disinfectants, bile salts andlysozyme (Hancock, 1984).B. Gram-negative outer membranes.1. IntroductionThe outer membranes of Gram-negative bacteria contain a numberof different proteins, including porins, in an asymmetric lipid bilayercomposed of an inner layer of phospholipid (Smit et al., 1975; Kamio andNikaido, 1976) and an outer layer of lipopolysaccharide (LPS) (Figure 1).The outer membrane acts as a molecular sieve, limiting the access ofmany compounds to the cell. The water-filled porins provide the majorpathway for the uptake of small hydrophilic molecules with thespecificity and the size of the pore determining the access. Hydrophobicmolecules, which can pass directly through the outer membrane of somebacteria such as N. gonorrhea (Martinez de Tejada and MoriyOn, 1993),are largely excluded from E. coli and P. aeruginosa (Hancock et al., 1994).This resistance is attributed to the structure of the LPS, as describedbelow. A third route into the cell, the self-promoted uptake pathway, isalso attributed to the structure of the LPS and allows the uptake ofpolycationic antibiotics and polycationic peptides (Hancock and Bell,1988; Vaara, 1992).3LPSPhospholipidFigure 1. Schematic representation of the outer membrane of a Gramnegative bacterial cell. The locations of the LPS, phospholipid, lipoproteins(L), porins (P) and peptidoglycan are indicated.42. PhospholipiclsThe phospholipids found in the outer membrane are similar tothose found in the cytoplasmic membrane, with phosphatidylethanolamine being the predominant species (Cronan Jr., 1979; Conradand Gilleland, 1981). Acidic phospholipids, including phosphatidylglycerol, cardiolipin and an unidentified lipid are also present in P.aeruginosa (Conrad and Gilleland, 1981).3. LPSLPS is a unique molecule found only in Gram-negative bacteria.Lipid A or endotoxin contains the hydrophobic component of themolecule and is highly conserved between bacterial species (Reeves1994). It is composed of from 5 to 7 fatty acids, which anchor it in themembrane, attached to a diglucosamine disaccharide with phosphates inthe 1 and 4’ positions. Attached to Lipid A is the hydrophilic coreoligosaccharide which varies from species to species. The core region iscomposed of 2-keto-3-deoxyoctonate (KDO), heptose, hexose, andphosphate. In smooth strains, between 10-25 % of molecules have theimmunodominant, sereotype-specific 0-antigen attached. The 0-antigenis composed of varying numbers of repeating oligosaccharide units whichform a capsule like structure over the bacterium. P. aeruginosa contains2 different types of LPS; B-band which is composed of LPS with 0-antigen and A-band LPS which has an antigenically and chemically-5distinct, shorter-chain polyrhamnan polysaccharide (Rivera et al., 1988;Rivera and McGroarty, 1989). The negative charges present on the LPScore oligosaccharide and KDO are partially neutralized by the binding ofdivalent cations such as Mg2 and Ca2. This binding stabilizes the LPSLPS and LPS-protein interactions and limits the uptake of hydrophobiccompounds. The addition of divalent-cation chelators such as EDTAdestabilizes the outer membrane which allows the uptake of hydrophobiccompounds (Hancock and Bell, 1988; Vaara, 1992). This divalent-cationbridging is also involved in self-promoted uptake. It has been proposedthat compounds such as polycationic antibiotics and cationic peptidescan displace the divalent cations, allowing not only the uptake of thecompounds themselves, but also the uptake of hydrophobic compounds(Hancock and Bell, 1988).4. Proteinsa) LipoproteinsOne type of protein which contributes to the stability of the outermembrane is the lipoproteins. The predominant lipoprotein of E. coil isthe Braun or murein lipoprotein (110). It is composed of 58 amino acidresidues in a primarily alpha-helical conformation, with the hydrophilicand the hydrophobic side chains located at opposite sides of the helix.The fatty acyl chains, linked to the N-terminus of the protein, anchor itin the outer membrane (Braun and Wu, 1994). About 1/3 of these6molecules are covalently attached to the peptidoglycan via the C-terminus of the protein and the remaining 2/3 are non-covalentlyassociated (Braun, 1975). Mutants lacking this protein (lpp) (Yem andWu, 1978) or that were deficient in the bound form of the protein (Funget al., 1978) were more sensitive to EDTA, had cell surface blebbing,impaired septum formation (Yem and Wu, 1978) and had leakyperiplasms indicating a structural role for this protein. The analogousprotein in P. aeruginosa is OprI which has 23-30% amino acid sequenceidentity with the Braun lipoprotein (Hancock et al., 1990). Althoughsome strains of P. aeruginosa were reported to have both covalent andnon-covalent forms, H 103, the wild-type laboratory strain used in thisstudy, does not have the covalently-bound form of this protein (Hancocket al., 1981). E. coli also has a 21 kDa non-covalently associatedlipoprotein, called the peptidoglycan associated protein, or PAL(Lazzaroni and Portalier, 1992). The P. aeruginosa equivalent is OprLwhich, unlike PAL, is a major constituent of outer membrane. Whengrown in rich media, OprI, OprL and OprF are the most prevalent outermembrane proteins in P. aeruginosa (Nikaido and Hancock, 1986).b) PorinsProteins of the outer membrane of E. coli include porins as well asproteases, phospholipase A, pili, flagella and inducible receptors for theuptake of iron-siderophore complexes and vitamin B 12. (Martin and7Hancock, 1990). The TonB-dependent receptor, FepA, is a high-affinityprotein specific for the uptake of ferric enterobactin in E. coli. It appearsto be a gated-porin and it has been suggested that other TonB-dependentligand-specific outer membrane receptors function in the same way (Rutzetal., 1992).Porins are classified as general or specific porins. Specific porinsallow the diffusion of small hydrophilic molecules, but have a specificbinding site which is an advantage when the nutrient is at a lowconcentration. At high nutrient concentrations these sites are saturated.Unlike the general porins, the expression of these porins is co-regulatedwith a complex transport system (Poole and Hancock, 1984; Benz, 1988).Two examples of this type are LamB from E. coli and OprP from P.aeruginosa which are specific for the uptake of maltose andmaltodextrins, and phosphate respectively, and which both function astrimeric proteins (Benz, 1988). LamB has recently been crystallized,showing that each subunit is composed of 18 anti-parallel f3 strands,with 3 of the longer external ioops folding into the barrel (Schirmer et al.,1995). Each subunit has a series of aromatic residues leading from theentrance to the constriction of the pore which have been called a “greasyslide”. These are thought to be a series of sugar-binding sites that alignthe sugar allowing its guided diffusion through the pore (Schirmer et al.,1995). Although OprP from P. aeniginosa and PhoE from E. coli are both8expressed in low phosphate media, they have minimal primary sequencehomology (Siehnel et aL, 1989) and differ in their anion-bindingpreference. PhoE has a weak affinity for anions while OprP has a 100-fold preference for the binding of phosphate over other anions (Hancocket at., 1986). Other specific porins include the minor outer membraneproteins Tsx of E. coli, for the uptake of nucleosides, (Benz, 1988) andTolC of E. coli, which appears to be selective for peptides (Benz, 1994),and OprB of P. aeruginosa, for the uptake of glucose, (Hancock andCarey, 1980) and OprD of P. aeruginosa, for the uptake of basic aminoacids (Benz, 1994) and imipenem in P. aeruginosa (Hancock et at., 1990).General trimeric porins are the most abundant proteins in E. coli(Nikaido, 1993). They allow the passage of small hydrophilic molecules,with an exclusion limit of 6O0 Da. and exclude hydrophobic and/orlarge molecules. Three trimeric, general porins have been identified in E.coli: OmpF, OmpC and PhoE. OmpF and OmpC are weakly cationselective and PhoE has a slight preference for anions (Benz et al., 1985).OmpF is expressed at higher levels than OmpC in low osmolarity mediaand the reverse is true in high osmolarity media. PhoE is expressedwhen the level of phosphate is low. The structures of OmpF and PhoE,which have 63% amino acid sequence identity, have been determined byX-ray crystallography (Cowan et at., 1992). Like the first crystallizedporin from Rhodobacter capsulatus, which has low sequence homology9with OmpF and PhoE, these porins are comprised of a f3-barrel composedof 16 anti-parallel 13-strands, with short turns at the periplasmic face andlonger cell-surface exposed loops. One of the eight external loops, loop 3,folds inside the barrel narrowing its diameter to create the “eyelet”region. Mutants in this region have been shown to have an increasedchannel size (Benson et al., 1988). Six of these loops partially shield theentrance of the pore. Loop 2 is involved in the subunit-subunitinteractions. Another conserved feature is the distribution of chargedamino acids in the eyelet region of the channel, with the basic sidechains on one side of the channel and the acidic side chains on theother. Lysine125of PhoE, which is located on loop 3, has been shown tobe primarily responsible for the difference in ion selectivity between PhoEand OmpF (Hancock et al., 1986). P. aeruginosa does not appear to haveany general trimeric porins belonging to the porin superfamily (Jeanteuret al., 1994). It does, however, have a general porin OprF which isrelated to OmpA of E. coli.C. OmpA.OmpA, one of the major outer membrane proteins in E. coli, has acopy number of about 1O per cell. The 325-residue mature protein issynthesized with a 21-residue signal sequence (Chen et al., 1980) andhas a high 13-sheet content (Jeanteur et al., 1994). Unlike trimeric porins10which require heating in SDS to disassociate into monomers (asdetermined by the decrease in apparent molecular weight on SDS-PAGE),OmpA has only been observed as a monomer on SDS-PAGE and isthought to be a monomer in its native state (Saint et at., 1993; Nikaido,1993). OmpA is associated both with the Braun lipoprotein and thepeptidoglycan (Braun, 1975).OmpA contributes to the stability of the cell membrane and to theshape of the cell and appears to function as a porin. OmpA-deficientmutants were shown to have a lag in growth when grown in minimalmedia, but had growth rates only slightly less than wild-type strains(Manning et at., 1977). These strains did not grow as well at 42° C,especially in minimal media, and appeared to have reduced uptake ofamino acids (Manning et at., 1977). Unlike either single mutant, doublemutants, deficient in both OmpA and in the Braun lipoprotein, describedin section 4a, were almost spherical, required cations for growth, but didnot require osmotic protection (Sonntag et at., 1978). Electronmicroscopy showed that the outer membranes of these cells wereblebbed. This was also seen in the lipoprotein mutants when they werestarved for magnesium. The micrographs also indicated that thepeptidoglycan of the double mutants was no longer associated with theouter membrane. OmpA has been shown to be non-covalently associatedwith both the peptidoglycan and the peptidoglycan-bound Braun11lipoprotein (Endermann et at., 1978; Lugtenberg and van Alphen, 1983).The double mutants were also shown to be more susceptible tohydrophobic antibiotics and detergents (Sonntag et at., 1978).Recently, OmpA has been shown to function as a porin by twodifferent groups. Sugawara and Nikaido (1992) showed that OmpAfunctioned as a nonspecific-porin for small molecules in the osmoticswelling of proteoliposomes. They estimated the size of the channel wassimilar to that of OmpF, about 1 nm in diameter, but that the rate ofdiffusion was about 100 times less. Saint et at. (1993), used a black lipidbilayer to show that OmpA has a diameter of about 0.6-0.7 nm and wasweakly anion selective.OmpA has also been shown to function as a receptor for severalphages, for the action of colicins K and L (Morona et aL, 1984), and to beinvolved in F’-mediated conjugation (Morona et at., 1984; Ried andHenning, 1987). This has proved to be useful for the construction of asecondary structure model. Morona et al. (1984) determined thatmutations in regions containing residues 25, 70, 110 and 154 resulted inphage resistance and proposed a model in which the N-terminal domainof the protein was composed of eight anti-parallel 13-strands with theregions noted above exposed on the surface of the cell. The susceptibilityto cleavage in intact cells of trypsin-cleavage sites inserted into theseregions is consistent with this model (Freudl et at., 1986; Freudl, 1989).12The same approach has been used to confirm the location of theperiplasmic turns (Ried et al., 1994). The N-terminal domain of OmpAterminates with a region that is required for insertion of OmpA into theouter membrane (Morona et at., 1984) and has sequence homology withthe last strand of the porin super family (Jeanteur et al., 1991). Thissequence identity followed by a proline-rich region which resembles thehinge region of immunoglobulins (Chen et at., 1980) and is the proposedsite of the characteristic trypsin cleavage site.The C-terminal domain of OmpA has been shown to be inessentialfor the association of the N-terminus with the outer membrane or withits ability to function as a phage and colicin receptor (indicating that theN-terminus was inserted into the membrane in the same conformation asthe wild-type protein) (Bremer et at., 1982). The C-terminal domain hasbeen shown to be the immunodominant portion of the protein inexperiments using purified OmpA, or intact E. coli as the antigen(Puohiniemi et at., 1990). In the secondary structure model of OmpA, theC-terminal domain is located entirely in the periplasm. This assignmentwas made primarily because, unlike the N-terminal domain, which isprotected by the outer membrane, the C-terminal domain is completelycleaved by proteases (Chen et at., 1980). In these experiments only theN-terminal region remained associated with the outer membrane (Chenetal., 1980).13OmpA-like proteins are present in E. coli as well as other entericand nonenteric bacteria including P. aeruginosa . Studies have used theproperty of heat-modifiability, which is the increase in the apparentmolecular weight of the protein due to the unfolding of the protein whenboiled in SDS or denatured in trichioroacetic acid (TeA), phenol or urea,to identify these related proteins (Beher et al., 1980; Spinola et al., 1993).Beher et at. (1980) identified heat modifiable proteins of similar molecularweight in twenty three strains of enteric and non-enteric Gram-negativebacteria. Of the strains tested, cleavage of these proteins with trypsinresulted in a characteristic membrane-bound fragment. Also, all of theproteins from the strains of Enterobacteriaceae were antigenically related.Primary structure analysis has shown significant C-terminal homologywith a number of proteins including OmpA proteins in strains of E. coli(Bremer et at., 1982), Salmonella typhimurium (Freudl and Cole, 1983),and Shigella dysenteriae (Braun and Cole, 1984), PAL of E. coli (Hardhamand Stamm, 1994), P6 of Haemophilus influenzae (Nelson et at., 1988),Pill of Neisseria gonorrhoeae (Gotschlich et al., 1987), a 31 kDa protein ofHaemophilus somnus (Won and Griffith, 1993), a 21 kDa protein ofBordetella avium (Gentry-Weeks et at., 1992), TpN5O of Treponemapallidum (Hardham and Stamm, 1994) and OprF of P. aeruginosa(Woodruff, 1988). The homology of the C-terminal portion of OmpA fromE. coli with OprF includes 56 identical amino acid residues and 3614conservative substitutions in the 180 C-terminal amino acids of OprFand in the corresponding 168 amino acids of OmpA (Woodruff, 1988).Only two regions in the C-terminal portion of these proteins had noapparent homology: the region located between amino acids 170-200containing four cysteines in OprF and the region between the twocysteines of OmpA, amino acids 290 and 300.D. OprF.As well as being related to the OmpA family, OprF of P. aeruginosais also part of the OprF family of the fluorescent Pseudomonadaceae. Allnine members tested from the rRNA homology group I contain an OprFlike protein as determined by Southern blotting with a P. aeruginosa oprFprobe (Ullstrom et al., 1991). This group of strains have also beenshown, by Western immunoblotting, to bind at least two of the ten OprFspecific monoclonal antibodies tested (Martin et al., 1993). Restrictionmapping and Southern blotting of 17 serotypes and 42 clinical isolates ofP. aeruginosa indicated that one copy of the gene was present in all ofthese strains and that the restriction pattern was generally conserved(Ullstrom et al., 1991). The exceptions to this were serotype 12, whichhad been sequenced by Duchene et al. (1988) and one clinical strain,both of which contained an additional KpnI site in the C-terminal half ofthe gene (Ullstrom et al., 1991). Comparison of the DNA sequence of15H 103, our laboratory wild-type strain which is serotype 5, with that ofthe serotype 12 strain showed only 16 nucleotide changes, all of whichare silent (Martin et al., 1993).The OprF from two other species of rRNA homology group I havealso been sequenced. These are the OprF from the plant pathogen, P.syringae, and from the plant growth-promoting rhizobacteria, P.fluorescens. Comparison of the DNA sequences of oprF from P.aeruginosa and P. syringae showed 72% identity while the regionsflanking the genes had only 34% identity (Ulistrom et al., 1991). The C-terminal regions had even higher homology with 85% identity and 10%conservative substitutions. Included in this homology is the proline-richhinge region also present in OmpA, and the region containing the fourcysteines. OprF appears to be conserved within strains of P. syringae.Nine strains representing seven different pathovars of P. syringae had anOprF-like protein of the same molecular weight as that of P. aeruginosaand all were heat and 2-mercaptoethanol modifiable and antigenicallycross reactive. Like OprF from P. aeruginosa, OprF from P. syringaereconstituted both large and small channels as determined by blacklipid bilayer (Ullstrom et al., 1991).The oprF sequence of P. fluorescens codes for a mature protein of302 amino acid residues, smaller than the 326 and 320 of P. aeniginosaand P. syringae, respectively (De Mot et al., 1992). An amino acid16sequence comparison of these three OprF molecules showed 71 %identity in the C-terminal half and only 35 % in the N-terminal half.Notable differences are the longer proline-rich region in P. fluorescensOprF and the lack of the cysteine-containing region of P. aeruginosa andP. syringae OprFs. It has been suggested that this region may be relatedto the root-adhesive properties of this species (De Mot et al., 1994).Analysis of protease cleaved peptides of OprF from whole cells or outermembrane samples of P. fluorescens located the proline-rich region nearthe surface of the cell, and retention of the resulting fragments in theouter membrane indicated that the C-terminal domain is not locatedsolely in the periplasm but loops through the membrane (De Mot et al.,1994).Another member of the OprF family appears to be the CD proteinfrom Branhamella catarrhalis, a human pathogen. This 453 amino acidprotein is heat modifiable and has surface-exposed epitopes (Murphy etal., 1993). It has significant amino acid homology with the OprF familyhaving -35% identity and 55-61% similarity (varying with the straincompared) (Murphy et al., 1993). It has two cysteines which correspondto cysteines 3 and 4 (numbered from the N-terminus) in P. aeruginosaand P. syringae OprF.OprF, with 200,000 copies per cell, is one of the major outermembrane proteins of P. aeruginosa. OprF is synthesized with a signal17sequence of 24 amino acids residues; the mature protein consists of 326amino acid residues (Duchene et aL, 1988). The oligomeric structure ofOprF is controversial. Unlike the trimeric porins which are readilyisolated as trimers and which require the native oligomeric conformationto function as porins (Hancock, 1987), the majority of OprF is isolated asmonomers which have porin function (Woodruff and Hancock, 1988).Oligomeric OprF have been observed on Western immunoblots, albeit insmall amounts (Mutharia and Hancock, 1985), and chemical cross-linking showed that purified OprF could form dimers and trimers (Angusand Hancock, 1983). It can be expressed in E. coli under control of itsown promoter, on a low copy number vector (Woodruff, 1988) due to thesimilarity of the RNA polymerase binding site sequence to that of the E.coli RNA polymerase binding site consensus sequence (Duchene et al.,1988). In addition to their size, high copy number, similar promotersequence and high degree of C-terminal homology, as described above,both OprF and OmpA are heat modifiable (Woodruff, 1988),immunologically cross reactive (Woodruff and Hancock, 1989; Martin,1992), and peptidoglycan associated (although OprF is released from thepeptidoglycan at lower temperatures than OmpA) (Lugtenberg and vanAlphen, 1983). Both also have a characteristic trypsin cleavage site,although that of OprF appears to be less accessible in outer membranes(Mutharia and Hancock, 1985). Trypsin digestion of outer membrane18samples showed that OmpA is easily cleaved (Chen et al., 1980) but, evenwith ten times more enzyme, OprF is only partially cleaved (Mutharia andHancock, 1985).Both OmpA and OprF are bifunctional; they are general-diffusionporins and have a role in the structural integrity of the cell. The functionof OprF has been studied, in part, by the construction of OprF-deficientstrains by chemical mutagenesis (Gotoh et al., 1989a; Nicas andHancock, 1983) or by c’-cartridge insertional mutagenesis (Woodruff andHancock, 1989). The effect of the loss of OprF on the structural integrityof these strains was shown in several ways. These strains grew poorly inlow osmolarity media (Gotoh et al., 1989a; Woodruff and Hancock, 1989)and were more sensitive to osmotic shock as measured by a decrease inviability (Gotoh et al., 1989a) or by periplasmic leakage as measured byan increase in extracellular f3-lactamase activity (Gotoh et al., 1989a;Woodruff and Hancock, 1989). Electron micrographs of osmoticallyshocked cells have shown blebbing and plasmolysis of OprF-deficientcells (Gotoh et al., 1989a; Gotoh et al., 1989b). Even in media with anosmolarity optimal for growth, the leakage of 13-lactamase from theperiplasm was higher in the OprF-deficient strains than in the wild-typestrains (Gotoh et al., 1989a; Woodruff and Hancock, 1989), as was theuptake of the hydrophobic probe, N-phenylnapthylamine (NPN) (Woodruffand Hancock, 1989).19The cell shape was also shown to be affected by OprF. OprFdeficient cells have been shown to be 67% of the length of the parentalstrain by image analysis (Woodruff and Hancock, 1989), or to be shorterand wider than the wild-type strain by electron microscopy (Gotoh et al.,1989a). The expression of cloned OprF in the almost spherical E. coli lppompA strain, described in the previous section, resulted in an increase incell length of 43% (Woodruff and Hancock, 1989). Hardham and Stamm(1994) confirmed these results and also showed that cloned TpN5O, anOmpA-like protein from T. pallidum, expressed in this strain could alsosubstitute for OmpA, resulting in wild-type length cells and growth at 42°C (Hardham and Stamm, 1994).As well as having a role in the structural integrity of the cell, OprFalso functions as a general diffusion porin, allowing the uptake ofhydrophilic compounds (Hancock et at., 1990). The exclusion limits ofboth P. aeruginosa and OprF have been controversial. Evidence for a lowexclusion limit of P. aeruginosa includes that it has a greater resistanceto some antibiotics than E. coli (Gotoh et at., 1989a) and the results ofplasmolysis (Caulcott et at., 1984; Yoneyama and Nakae, 1986) andliposome swelling studies using whole membranes (Yoshihara et at.,1988). Evidence for a high exclusion limit include its ability to grow onpentamethionine (Miller and Becker, 1978) and susceptibility to somelarge antibiotics (Siehnel et at., 1989). Nikaido et at. (1991) showed that20tetrasaccharides were able to diffuse across isolated outer membrane bythe osmotic swelling of proteoliposomes that had incorporated pieces ofouter membrane. Bellido et at. (1992) showed that a wild-type strainprovided with the ability to transport and metabolize raffinose andrelated sugars was able to grow on both a disaccharide and atrisaccharide. This study also demonstrated plasmolysis with atetrasaccharide of 660 daltons. These experiments indicated that theexclusion limit of P. aeruginosa was greater than that of E. coli whichexcludes tetrasaccharides (Bellido et at., 1992). The exclusion limit ofpurified OprF is also controversial. Liposome swelling assays withpurified OprF indicated that the exclusion limit was low, excluding evendisaccharides (Yoshihara and Nakae, 1989). This work was criticized byNikaido et al. (1993) for imprecise experimental procedures. Nikaido etal. (1991) showed that tetrasaccharides were able to diffuse throughOprF by the osmotic swelling of proteoliposomes. Bellido et al.( 1992)showed that the growth rate of an OprF deficient-strain was three to fivefold lower than the wild-type on a disaccharide and a trisaccharideindicating that OprF has a role in the uptake of these sugars. Earlierradioefflux experiments (Hancock et at., 1979) and liposome swellingassays (Yoshimura et at., 1983) also supported a large exclusion limit.Purified OprF’ was shown to have both large (4-7 nS) and small (0.36-0.38 nS) pores in black lipid bilayer experiments (Benz and Hancock,211981; Woodruff and Hancock, 1988), with the majority of the channelsbeing small (Woodruff and Hancock, 1988). Two hypothesis have beenproposed to explain the contradictory evidence of the large channel sizeof OprF and the general low permeability of the cell, both of whichsuggest that OprF functions poorly in the uptake of some substrates likeantibiotics. Woodruff et al. (1989) proposed that less than 1% of the ofOprF formed large channels, therefore limiting the access of largermolecules. In this paper, it was also proposed that the difference in thesize of the pore could be due to alternate bonding of the four disuiphidespresent in OprF. Nikaido et al. (1991) proposed that despite the largeexclusion limit of OprF, it was the structure of the channel of OprF thatlimited the uptake of some solutes.Although the C-terminal domain of OprF has a high degree ofamino acid homology with OmpA, calculation of the antigenic index,which is based on measures of secondary structure and on predictivemethods, of the N-terminal domain of these two proteins showed a bettercorrelation for these regions than for the C-terminal domains (Martin andHancock, 1990). A secondary structure model of OprF (Wong et al.,1993) shows the N-terminal domain of OprF composed of eight f3 strandsin the same basic configuration as the model of OmpA. The location ofC-terminal region of OprF however, differs significantly from the modelproposed for OmpA. The model of OmpA located the entire C-terminal22domain of the protein in the periplasm, while that of OprF was composedof transmembrane f3 strands and included surface-exposed loops. Thisconformation was primarily based on the location of permissive sites forlinker insertion mutagenesis (Wong et al., 1993), for protection of outermembrane samples from cleavage by trypsin and by surface accessibilityof the monoclonal antibody, MA5-8 (Mutharia and Hancock, 1985).E. Aims of this study.The aims of this study were to study the structure and function ofOprF. The study of the function of proteins often includes mutation ofthe protein and analysis of the resulting phenotype. This is ideallyconfirmed by genetic complementation. In this study cloned OprF wasreturned to an OprF-deficient strain of P. aeruginosa and the resultingphenotype assessed. The conserved sequence of the C-terminal domainof OprF indicates that it has an important function in the cell. To studythe role of the C-terminal portion of OprF in the integrity of the outermembrane, truncated mutants of OprF were constructed and expressedin P. aeruginosa. A number of approaches were used to gain a greaterunderstanding of the secondary structure of OprF. These included themutation of cysteines and the location of the epitopes binding OprFspecific monoclonal antibodies both in the primary and secondarystructure of OprF.23MATERIALS AND METHODSA. Bacterial strains and plasmids.The strains of E. coli and P. aeruginosa used in this study arelisted in Table I and the plasmids used are listed in Table II.B. Media and growth conditions.1. In vitro growth.E. coli strains were grown on Luria Broth (LB) (1% tryptone,0.5% yeast extract, 1% NaC1, pH 7.0), and P. aeruginosa strains weregrown on Mueller-Hinton broth or on Luria Broth High Salt (LBHS) (1%tryptone, 0.5% yeast extract, 200mM NaC1, pH 7.0). The low osmolaritymedia used in growth experiments was either PP2 (1% proteose peptoneno. 2) or Luria Broth No Salt (LBNoS) (1% tryptone, 0.5% yeast extract,pH7.0). Media were solidified with 2% Bacto-agar. All mediacomponents were from Difco Laboratories, Detroit, Michigan. Selectivemedia used antimicrobial agents at the following concentrations for E.coli: tetracycline at 25 pg/ml, ampicillin at 75 ig/ml, kanamycin at 25j.ig/ml; and for P. aeruginosa: tetracycline at 200 pg/mi, kanamycin at250jig/mi, streptomycin at 500 jig/ml, and carbenicillin at 300 jig/ml.Plates were spread with 50 jil of dimethylformamide containing 50 jig of24Table I: Bacterial Strains.Strain Characteristics ReferenceE. coliK-12DHSci. endAl hsdRl 7 supE44 thi-1 recAl gyrA96 BRL, Burlington, Ont.relAl A(argF-1acZYA)U169 4)80 dlacZzM15C441 Kans , has helper function on Simon et al., 1983chromosomeC441a C44 1/ pWW2500 This studyP. aeruqinosaH103 PAO1 Cmr prototroph; wild type Hancock and Carey, 1979reference strainH636 H103 oprF:: Woodruff and Hancock, 1988M-2 strain isolated from GI tract of normal Stieritz and Holder, 1975miceM-2F- M2oprF::) ThisstudyPA1O3 pathogenic strain Mutharia and Hancock, 1983ATCC-33348- 17 serotype strains of International Benz, R. 199464 Antigen Typing Scheme25Table I. Bacterial Strains (continued).CF832 Clinical isolate from a Cystic Fibrosis Patient Rivera and McGroarty, 1989CF1452CF2314CF3660-1CF6094CFLaughlinCF4349CF1278CF22 1CF3790CF9490H496-H510 Environmental isolates Hancock and Chan, 198826Table II: Plasmids.Name Characteristics! Origin ReferencepHP45C2-Tc pHP45 + 2.0 kb fragment coding for TcrQ Fellay et al., 1987cartridgepUCP19 pUC19 + 1.8 kb P. aeruginosa stabilizing Schweizer, 1991fragmentpWW2500 pRZ 102 + 2.5 kb Sail fragment containing Woodruff and Hancock, 1988the N-terminal half of oprF with the SmrQ cartridge inserted into the Smal sitepWW5 pUC8 + 2.0 kb fragment from pWW4 Woodruff and Hancock, 1989coding for truncated OprF (a.a. 17 1-300 ofOprF were deleted by homologousrecombination)pRW5 pUCP19 + 1.47 kb HinDIII-EcoRI fragment R. Wong, unpublished resultsfrom pRW3 coding for OprFpRW3O7 linker -insertion at a.a. 188 of pRW3 Wong et al., 1993(pTZ19R + oprF with a weakened promoter)pRW3O9 as above, with the linker inserted Wong et al., 1993ata.a. 210pRW3 10 as above, with the linker inserted Wong et al., 1993at a.a. 215pRW31 1 as above, with the linker inserted Wong et al., 1993ata.a. 231pRW3 12 as above, with the linker inserted Wong et al., 1993at a.a. 290pER1O2 pRW5 deleted in the 1.1 Kprtl fragment This studycoding for the C-terminal 2/3 of OprFpERlO2t pER1O2 + 2 kb fragment coding for Tcr This studypER163 pUCP19 + 0.6 kb HinDIII-EcoRI fragment This studywith a translational stop signal followinga.a. 16327Table II: Plasmids (continued).pERl63t pER163 + 2 kb fragment coding for Tcr This studypER17O-26 pER1O2 + 0.7kb KpnI fragment from pWW5 This studypER17O-26t pER17O-26 + 2 kb fragment coding for Tcr This studypER188 pUCP19 + 1.47 kb HinDIII-EcoRI fragment This studycoding for the N-terminal 188 a.a. of QprFconstructed by linker insertionmutagenesis of pRW3O7pER 188t pER 188 + 2 kb fragment coding for Tcr This studypER2 13 pUCP19 + 1.47 kb HinDIII-EcoRI fragment This studycoding for the N-terminal 210 a.a. of OprFconstructed by linker insertionmutagenesis of pRW3O9pER2l3t pER21O + 2kb fragment coding forTcr This studypER215 pUCP19 + 1.47 kb HinDIII-EcoRI fragment This studycoding for the N-terminal 215 a.a. of OprFconstructed by linker insertionmutagenesis of pRW3 10pER2l5t pER215 + 2kb fragment coding forTcr This studypER231 pUCP19 + 1.47 kb HinDill-EcoRl fragment This studycoding for the N-terminal 231 a.a. of OprFconstructed by linker insertionmutagenesis of pRW3 11pER29O pUCP19 + 1.47 kb HinDIII-EcoRI fragment This studycoding for the N-terminal 290 a.a. of OprFconstructed by linker insertionmutagenesis of pRW3 12pER29Ot pER29O + 2 kb fragment coding for Tcr This studypER pUCP19 + 2 kb fragment coding for Tcr This studypER326t pRW5 + 2 kb fragment coding for Tcr This studyTable II: Plasmids (continued).pERC 185s pRW5 - 0.7 kb HindIII-SalJ fragment + This studycorresponding fragment with a..a. 185mutated from a cysteine to a serinepERC191s pRW5 - 0.45 kb SalI-EcoRI fragment + This studycorresponding fragment with a.a. 191mutated from a cysteine to a serinepERC185s+ pERl85s - SalI-EcoRI fragment + This studyC 19 is corresponding fragment from pERC 19 is28295-bromo-4-chloro-indolyl-J3-D-galactoside when manipulating pUCP 19-based plasmids.2. In vivo growth.The in vivo chamber model, developed by Day et al. was usedfor the growth of P. aeruginosa in mice. Briefly, a Millipore filter (0.2 iimpore size) was glued on one end of a chamber made from a 1-cm sectionof a 1-cc plastic syringe barrel. Bacteria, resuspended in 0.9% saline at10 cells per ml, as assessed by total count in a Petroff-Hausserbacteria counter, were added to the chambers and a second filter wasattached to the open end. Four chambers were surgically implanted inthe peritoneal cavity of B6D2 Fl mice. The chambers were removed fromthe mice after 4, 8, 16, 20, 24 or 48 hours after implantation and platedfor a viable count.C. DNA procedures.1. Plasmid isolation.Plasmid DNA was routinely isolated by an alkaline lysismethod (Sambrook et al., 1989). Plasmid DNA was isolated forsequencing either using Quiawell-8 plasmid purification system (QuiagenInc., Chatsworth, CA.) or by the modified alkaline-lysis/PEG precipitationprocedure described in the manual for the Taq DyeDeoxy terminatorcycle sequencing kit from Applied Biosystems Inc. (Mississauga, Ont.).30Routine screening of transformants for inserts or for the presence ofplasmid DNA was by the slot-lysis method of Sekar (1987).2. Plasmid introduction.a.) TransformationCompetent E. coli were prepared by the CaC12 methodand transformed as described in Sambrook et al. (1989). P. aeruginosawas transformed by the method of Olsen et al. (1982). This method issimilar to the CaC12 method used for E. coil, the main difference beingthe use 0.15 M MgCi2 for the preparation of the cells instead of CaC12.b.) ElectroporationThe procedure used for the electroporation is describedin Farinha et al. (1990). The buffer used for preparation of the cells was15% (w/v) glycerol in 1mM Hepes- 1mM MgCi2. Cells were frozen inaliquots at -70° C until required. The plasmid DNA was prepared by analkaline lysis method (Sambrook et al., 1989) repeating the rinsing of theDNA in 70% ethanol to reduce the salt content. Either 2.5 or 5 jil of DNAwas added to 50 jii of the prepared cells. The samples were gently mixed,and 25 il was added to each chilled cuvette. After 10 mm incubation onice, the samples were electroporated. The settings used were 1.8 kV,2002, and 25 pP in cuvettes having a 0.1 cm electrode gap. Theequipment used was a Gene Pulser (Blo-Rad Laboratories, Richmond,CA). Immediately after the electroporation, high salt media containing 5031mM MgCl2was added to the cuvette. The cells were then gentlysuspended, and allowed to grow at 37°C for 1.5-1.75 hr before plating onselective media.c.) Biparental matingThe method used for biparental mating was based onthe method described in Goldberg and Ohman (1984). Overnight cultureswere grown in broth with shaking at 37°C for the donor strain and 42°Cfor the recipient strain. Aliquots of 0. lml of both the donor and therecipient strains were added to 2 ml of fresh broth. The mixture wasthen filtered through a 0.45 jim membrane. The membrane was placedcell side up on a non-selective agar plate and incubated overnight at30°C. The cells were then washed off the membrane with fresh media,diluted, spread onto selective plates and incubated at 30° C for 1 to 2days.3.)-cartridge mutagenesis of oprF.The method used for the -cartridge mutagenesis of P.aeruginosa M-2 was basically the same as described in Woodruff andHancock (1988) and utilized the plasmid pWW2 500 described in thatpaper. In this study, biparental mating was done using the donor helperstrain C441.4. Southern blotting.a.) Chromosomal DNA isolation32Chromosomal DNA was isolated by the methoddescribed in Current Protocols In Microbiology (1987) and quantified bymeasurement of the absorption at 260nm.b.) Labeling of DNA probeThe DNA probe was made by isolating a 1.47 kbHinclIII/EcoRI fragment from pRW5 coding for oprF. This fragment waslabeled with biotin using a Bionick Labeling System from BRL(Burlington, Ont.).c.) Southern blottingChromosomal DNA was digested with either PstI orSmal and the fragments separated by electrophoresis in an agarose gel.The DNA was transferred to Biotrans nylon membrane (ICN BiomedicalsCanada Ltd., St. Laurent, Quebec) by capillary transfer as described inCurrent Protocols in Molecular Biology (1987). The PhotoGene NucleicAcid Detection System (BRL, Burlington, Ont.) was used as described inthe manufacturer’s instructions for the remainder of the procedure.5. Oligonucleotide synthesis.Oligonucleotides were synthesized on a Applied Biosystems392 DNA/RNA Synthesizer (Mississauga, Ont.) exactly as described inthe manufacturers instructions.336. Polymerase Chain Reaction.For the construction of the truncated OprF mutants byPolymerase Chain Reaction (PCR), the buffer used was 20 mM Tris, pH8.3, 1.5mM MgC12,25mM KC1, 0.5 %Tween and 1 mg/mi gelatin. Tothis 5% formamide, 1.25 units Taq polymerase (BRL, Burlington, Ont.),250 jiM dNTP’s (BRL) , and ijiM each of the 5’and the 3 primers(described in Chapter 1) were added. DNA was amplified in an MJResearch thermal cycler (MJ Research, Watertown, Mass) at 95° C for 15seconds, 55° C for 30 seconds and 72° C for 90 seconds and this cyclewas repeated 30 times. The product was gel purified and digested withthe appropriate restriction enzymes for ligation into pUCP19. Vent DNApolymerase (Fisher Scientific, Vancouver, B.C.) was used for themutagenesis of two of the cysteines of OprF. The primers used aredescribed in Chapter 1. The procedure followed was exactly as aboveexcept for the use of the supplied buffer.7. DNA seQuencing.DNA was sequenced with an Applied BiosystemsIncorporated (ABI, Mississauga, Ont.) automated fluorescent sequencingsystem, model 373A, and analyzed with an ABI 675 sequence-editorprogram. The polymerase chain reaction and dye-terminator chemistrywas used as described in ABI’s protocols.34D. Protein procedures.1. SDS-PAGE.Proteins were prepared for electrophoresis by solubilizationin 2% (w/v) SDS with or without the addition of 2% (v/v) 2-mercaptoethanol. Samples prepared from whole cell lysates of P.aeruginosa were then sonicated 3 times for 15 seconds in a sonicatingwater bath. Samples were then heated to 1000 C for 10 mm. Heattreated samples were boiled for 30 mm in SDS or treated with 5% (w/v)TCA for 30 mm on ice (Hancock and Carey, 1979). Proteins wereseparated by electrophoresis with 11%, 12% or 14% polyacrylamide gelsas previously described (Hancock and Carey, 1979). The gels werestained with Coomassie Brilliant Blue R250 (Bio-Rad) for visualization ofthe proteins or used for Western immunoblotting.2. Outer Membrane Preparation.Outer membranes were prepared by the method of Hancockand Carey (1979). Briefly, after centrifugation, mid-log phase cells werebroken with a French pressure cell. The resulting lysate, in 20 % (w/v)sucrose, was loaded onto a 50/60 or 50/70% sucrose gradient andcentrifuged in a swinging bucket rotor. The lower band containing theouter membrane was collected, suspended in distilled water and pelletedby ultracentrifugation.353. Preparation of peptidoglycan-associated proteins.Peptidoglycan-associated proteins were prepared by themethod of Hancock et al. (1981). Briefly, outer membrane preparationswere suspended in 2% Triton X- 100 in buffer and sonicated. Theinsoluble membrane was pelleted by ultracentrifugation. The pellet wassuspended as above with the addition of 10 mM EDTA. This step wasrepeated resulting in the Triton-EDTA soluble fraction. The remainingpellet was digested with lysozyme and diluted with the Triton-X- 100 inbuffer with the addition of 10 mM MgSO4,sonicated and pelleted asabove. The supernatant was called the Triton-lysozyme soluble fraction.4. Determination of protein concentration.A modified Lowry assay was used to determine proteinconcentrations (Peterson, 1977). This method uses SDS to solubilizemembrane proteins and also denatures the proteins making the resultsmore reproducible.E. Immunological techniques.1. AntibodiesOprF-specific monoclonal antibodies MA7-1, MA7-2, MA7-3(designated IG1 in Pennington et al., 1986) MA7-4, MA7-5, MA7-6, MA7-7, and MA7-8 were prepared at Oncogen (Seattle, WA) and are describedin Pennington et al. (1986). These antibodies were prepared either by36injecting mice 4 times with purified OprF in combination with J5 LPS orby extensive immunization over a 6 month period with a variety ofantigen preparations which included heat killed P. aeruginosa ATCC27318, LPS prepared from this strain, live P. aeruginosa ATCC 27317,and then finally injected with outer membranes in combination withpurified OprF. Monoclonal antibodies MA4-4 and MA5-8 were made byL. Mutharia et al. (1983) using purified OprF or outer membranesprepared from H 103. FPLC-purified OprF was used to make thepolyclonal rabbit and polyclonal mouse sera as described in Rawling etal. (1995).2. Colony immunoblotting.Bacterial colonies, grown overnight on agar plates,were transferred to nitrocellulose by contact. The filters were hung in achloroform-vapor-saturated chamber for 30 mm to allow cell lysis. Thefilters were then washed in a buffer containing 3% (w/v) bovine serumalbumin (BSA) in phosphate buffered saline (PBS) with 150 mM NaCl, 5mM MgC12, 1 jig/mL DNAase I, and 40 j.ig/mL of lysozyme. After rinsing3 times with PBS the filters were incubated with primary antibody(diluted 1:2000 in 1% BSA in PBS) for 2 h at 37° C with shaking. Afterwashing 3 times with PBS, the secondary antibody, goat anti-mouseimmunoglobulin G alkaline phosphatase conjugate (BioRad), used at adilution of 1:2000, was added and incubated for 2 h at 37°C with37shaking. After washing 3 times in PBS, the filters were washed in 0.1 MTris-HC1 pH 9.6 and then developed with the substrate solution (50pg/ mL 5-bromo-4-chloro-3 -indolyl-phosphate, 10 ig/mL nitrobluetetrazolium, and 40 jig/mL MgC12 in 0.1 M Tris-HC1 pH 9.6).3. Western immunoblotting.After separation by SDS-PAGE, proteins wereelectrophoretically transferred to nitrocellulose from the polyacrylamidegel using the BioRad Trans-Blot electrophoretic transfer cell, with acooling pack, at a constant voltage of bOy for 1 h. The buffer used fortransfer contained 25 mM Tris-HC1, 192 mM glycine, and 20% (v/v)methanol, pH 8.3. After transfer, proteins were immunologically detectedas described above for colony immunoblots.4. Immunofluorescent labeling.Cell surface indirect immunofluorescent labeling ofcells was performed as described by Martin et al. (1993). Briefly, mid-logarithmic cells were collected by centrifugation, washed with PBS,allowed to air dry on a glass slide, and then fixed in 100% ethanol. Thefixed cells were incubated with an OprF-specific monoclonal antibodydiluted 1:100 in PBS with 1% fetal calf serum for 30 mm at roomtemperature. After washing with PBS, the secondary antibody(fluorescein isothiocyanate-conjugated goat anti-mouse immunoglobulin(Sigma Chemical Co.) was added at a dilution of 1:20. After washing and38the addition of a drop of Sigma Mounting Medium, a cover slip wassealed on the slide with the use of clear fingernail polish. The slides wereexamined with a Zeiss microscope fitted with a halogen lamp and filtersset for emission at 525 nm.5. Opsonic phagocytosis.Opsonic phagocytosis was performed using a modification ofthe procedures of Battershill et al. (1987). Mouse peritonealmacrophages from 2-4 month old B6D2(fl) mice were incubatedovernight in 8-well chamber slides (NUNC, Naperville, IL) to allowadherence. Antibody concentrations were standardized by an enzyme-linked immunosorbent assay (ELISA) with purified OprF. Themonoclonal antibody, at a dilution of 1:10, and the bacteria at a ratio of20:1 (bacteria to macrophage, suspended in 20 jiL bacterial buffer-S mMHepes and 1 mM MgC12), were added to the macrophage monolayer. Theplates were then incubated for 90 mm at 37°C in a CO2 atmosphere.After rinsing, cells were stained with Diff-Quick (Canlab, Vancouver,B.C.) and viewed with a Zeiss microscope fitted with a lOOx oilimmersion objective. The number of bacteria in 100 macrophages wererecorded. Experiments were carried out on 3 different days, and theresults from each day were tabulated and subjected , separately, to theMann-Whitney test to determine the significant of differences fromcontrols.396. Overlapping-octapeptide analysis.Support-coupled overlapping octapeptides starting at everysecond amino acid position of OprF were purchased from ChironMimotopes (Clayton, Australia). The peptides, attached to polyamidepins in arrays of 96 pins, were used as antigens in ELISA studies inwhich the pins were inserted into ELISA plates as described previously(Geysen et al., 1987). The ELISAs were performed, in two independenttrials, as described in Geysen et al. (1987) except that the absorbancewas assessed periodically to ensure that readings were in the linearrange of the machine. Positive and negative controls supplied with thekit were successfully performed (data not shown). Antibodies were usedat dilutions of 1:10,000 to 1:2500 for the monoclonal antibodies and1:5,000, 1:500, or 1:350 for the polyclonal antibodies. Antibodies wereremoved from the pins by sonication of the blocks for 10 mm in asolution of 0.1 M phosphate buffer, 1% (w/v) SDS and 0.1 % (v/v) 2-mercaptoethanol preheated to 60° C. The blocks were rinsed at in 60° CdH2O and then in methanol at 60°C.F. Cell-length measurement.1. Image analysis.Bacteria were grown to mid-logarithmic phase (opticaldensity of -0.5 at 600 nm.) in high osmolarity media. A sample was air40dried on a microscope slide, heat fixed and stained with crystal violet.Cells were viewed by oil immersion with a Zeiss Universal microscopeusing x 100 magnification. Images were digitized and analyzed with aSEM-IPS image analysis system (Kontron, Munich, Germany).2. Microscopy.Bacteria were grown in high osmolarity media with 200jig/ml tetracycline to mid-logarithmic phase (75 klett units) or to earlylogarithmic phase (40 klett units). Samples were prepared as describedfor image analysis. To prevent biased readings, measurement of thesamples was done in a single-blinded fashion by S. Farmer. Thesamples, identified by a code number, were viewed at x 1000magnification by oil immersion using a Zeiss IIIRS microscope fitted withphase rings and connected to a television monitor. Measurement of thecells was done directly from the monitor. At least 100 cells of eachstrain, from 3 different experiments, were measured and the resultsanalyzed by Student’s t-test using the Bonferonni correction.G. Growth studies.Strains of P. aeruginosa, with wild-type or mutated OprF wereassessed for their growth in low osmolarity media. Overnight cultures,grown in high salt media, were diluted 1:100 into high salt and low saltmedia, with or without the addition of antibiotics. When tetracycline wasadded, the media was warmed for 15 mm to allow the evaporation of the41ethanol into which the tetracycline was dissolved. Cultures were grownin a shaking water bath at 37° C, and 160-180 rpm. Cell density wasdetermined using a Klett-Summerson photometer with a green filter.42RESULTSCHAPTER 1. CONSTRUCTION AND EXPRESSION OF OprFMUTANTS IN P. aeruginosaA. Introduction.The function of outer membrane proteins can be studied throughinactivation of the oprF by chemical mutagenesis, the insertion oftransposons or interposons or by gene replacement, and an examinationof the resulting phenotypes of the cell. Ideally, the mutation should beconfirmed by genetic complementation. Although OprF-deficient strainsof P. aeruginosa have been constructed by chemical mutagenesis (Nicasand Hancock, 1983; Gotoh et al., 1989a) and by transposon andinterposon mutagenesis (Woodruff and Hancock, 1988), and the clonedgene expressed in E. coli (Woodruff and Hancock, 1989), attempts tocomplement this mutation in P. aeruginosa had been unsuccessful(Woodruff, 1988). In this study, a cloned variant of oprF, with aweakened promoter, was used to complement an OprF-deficient strain.Various mutants, including deletion mutants, have been used tostudy regions of OmpA (Maclntyre et al., 1988; Klose et al., 1988a; Freudland Henning, 1988; Klose et al., 1989; Ried et al., 1994; Tanji et al.,1991) and PhoE (Bosch et al., 1988) required for translocation across the43cytoplasmic membrane and for incorporation into the outer membrane.In this study, deletion mutants of oprF were constructed to determine therole of the C-terminal domain in the incorporation of the protein into theouter membrane and also to study its role in outer membrane stability,as assessed by growth in low osmolarity media and cell length. Cysteineto-serine mutants of OprF were constructed by site-directed mutagenesisand expressed in P. aeruginosa to quantitate the disulphide bond(s)present in the native OprF and to determine the identity of the cysteinesthat are bonded. The stability of the outer membranes of these strainswas also assessed.B. Construction of an OprF-deficient mutant of P. aeruginosa,strain M-2.OprF-deficient strains constructed by chemical mutagenesis wereshown to have a high frequency of reversion (Nicas and Hancock, 1983)and also exhibited phenotypic variation (Gotoh et al., 1989a). Thegeneration times of three chemically-mutagenized OprF-deficient strainsin a low osmolarity medium with 17 mM NaC1 were 452, 1,204 and 1,806mm and with 42.5 mM NaC1 was 125, 144, and 301 mm, respectively(Gotoh et al., 1 989a). Comparison of insertionally mutagenized OprFdeficient strains, constructed by Tn 1 mutagenesis or by -cartridgemutagenesis, also showed variation with generation times of 80 and 142mm in a low osmolarity medium with 50 mM NaC1 and generation times44of 128 mm and no growth in a medium without added salt, respectively(Woodruff, 1988). These data were consistent with the suggestion thatthe more strongly affected a-cartridge insertion mutant had the correctOprF-deficient phenotype and that the other mutants had suffered fromphenotypic modulation due to secondary mutations. Woodruff andHancock (1988) had constructed and characterized only one OprFdeficient -cartridge mutagenized strain, H636 (Woodruff, 1988;Woodruff and Hancock, 1988). To ensure that the observed phenotypeassociated with this method of mutagenesis was not limited to the straintested, a strain used for mouse pathogenicity studies (Stieritz andHolder, 1975), phagocytosis studies (Battershill et aL, 1987; Speert andThorson, 1991), and killing assays (Speert and Thorson, 1991; Speert etal., 1994), strain M-2, was selected for mutagenesis. The plasmid,pWW2500, described in Woodruff and Hancock (1988) was used for themutagenesis procedure (Figure 2A). Plasmid pWW2 500 contains an 2-cartridge, which codes for resistance to streptomycin and has stopcodons in all reading frames, inserted into the SmaI site of a Sailfragment coding for the N-terminal 60% of OprF. The Sail fragment isflanked by sequence coding for the 1S50 elements of the transposon Tn5and confers kanamycin resistance. This plasmid is unable to replicate inP. aeruginosa and can undergo homologous recombination (Jorgenson etal., 1979) between the plasmid oprF:2 and the chromosomal oprF. For45AKn SailSmaIQpWW250016.2kbSmaISailB4SmaI SailoprFFigure 2. A. Map of the plasmid pWW2500 showing the location of theSail and SmaI sites and the Tn5 E] oprF , and 12-cartridge E]DNA. B. Restriction map of oprF showing the 12 insertion site.46the mutagenesis of M-2, pWW2500 was first transformed into the E. colihelper strain C44 1. The resultant strain, selected on the basis ofkanamycin resistance, was used for biparental mating with M-2.Transconjugants were first screened for streptomycin resistance,indicating the rescue of the a-cartridge by incorporation into thechromosome, and then for kanamycin sensitivity, indicating the loss ofvector sequences, a result consistent with a double cross-over event. Thesite of the 2-cartridge insertion into oprF is indicated in Figure 2B. Theloss of OprF was then determined by colony immunoblot using the OprFspecific monoclonal antibody MA7- 1. To ensure that the )-cartridge wasonly inserted into the oprF gene, chromosomal DNA, isolated from theparent strain and the selected mutants, was digested with PstI and SmaIand analyzed by Southern blot using a biotin-labeled oprF probe. Figure3 shows the hybridization pattern of three oprF :f mutants and theparent strain M-2. In the PstI digest, the probe hybridized to a 2.4 kbfragment from the strain M-2 (lane 1) and a 4.4 kb fragment from threeM-2F- strains (lanes 3, 5, and 7), an increase of 2.0 kb which correlatedwith the insertion of the 2.0 kb Smal fragment. In the Smal digest,which released the -cartridge, the probe bound to identically-sizedfragments in all strains (lanes 2, 4, 6, and 8). These results confirmedthat the OprF-deficient strains of M-2 constructed by mutagenesis472.4kb41 2 3 4 5 6 7 8Figure 3. Southern blot of a biotinylated oprF probe hybridized tochromosomal DNA digests of P. aeruginosa strain M-2 (lanes 1 and 2)and three M-2F- strains (lanes 3-8). The DNA was digested with PstI (oddnumbered lanes), or with SmaI (even numbered lanes).4.4kb48contained one copy of the )-cartridge inserted into the chromosomaloprF gene.Restriction endonuclease mapping of a 5 kb region surroundingand including oprF in the 17 serotypes of P. aeruginosa has shown that,with the exception of serotype 12, there is strong conservation of therestriction endonuclease sites studied in and upstream of oprF and thatthere is some heterogeneity of the KpnI sites several kilobasesdownstream of oprF (Woodruff, 1988). This is consistent with the datapresented here. The PstI fragment and the smaller of the two SmaIfragments from M-2 identified in the Southern blot were the same size asthe corresponding fragments from H 103 (Woodruff and Hancock, 1988).However, the larger Smal fragment was 1.5 kb smaller than that fromH 103 indicating that the downstream Smal sites, as well as the KpnIsites may be heterogeneous. No reversion of M-2F- or H636 wasobserved after growth in high-salt media for twenty four hours or in lowsalt media for eight hours.C. Construction of truncated versions of OprF.Attempts to clone oprF, and certain other outer membrane proteingenes, into high copy number vectors have proven unsuccessful(Duchene et al., 1988; Woodruff, 1988). These constructs are thought tobe lethal due to high levels of protein produced as a result of their49efficient promoters in combination with the high copy number of theplasmids (Woodruff, 1988). The weakening of the oprF promoter by site-directed mutagenesis (Wong et at., 1993) permitted the cloning of oprFinto a high copy number, 4.5 kb Escherichia-Pseudomonas shuttle vector,pUCP19 (Schweizer, 1991) resulting in the 6.0 kb plasmid, pRW5 (R.Wong, unpublished results). All plasmids constructed in this study werebased on pUCP19 and maintained the weakened oprF promoter.Plasmids encoding truncated versions of OprF were constructedusing three basic approaches. The first utilized a plasmid previouslycharacterized in this laboratory, pWW5 (Woodruff et at., 1986).Sequencing of this plasmid had shown that amino acids 17 1-300 hadbeen deleted (Finnen et at., 1992). A KpnI fragment, coding for aminoacids 103-170 and 30 1-326, from pWW5 was used to replace the wild-type KpnI fragment from pRW5. Screening was done first by slot lysis, toselect for insertion of the fragment, and then by colony immunoblot.Colonies selected were reactive with OprF-specific monoclonal antibodiesMA7- 1 (N-terminus specific) and MA5-8 (C-terminus specific) (seeChapter 3, B) to ensure the correct orientation of the fragment. Theresulting plasmid was designated pER17O-26, referring to the N-terminal170 and the C-terminal 26 amino acid residues that it encodes(Figure 4).The second approach utilized plasmid constructs made by R. Wongin this laboratory by linker insertion mutagenesis (Wong et at., 1993).HindlilKpnISailKpnI/EcoRII..,1::::::.:.:::::;:::•;÷:x:::::::,:::.‘.:.....:.:.50100150200250300STRAINH103(wild-type)H636oprF:PLASMIDSpER326pER1O2pER163pER188pER215pER231pER29OpER170-26___Figure4.MapofoprFandOprFshowingtherestrictionendonucleasesitesandthelengthofthematureproteinTheencodedlengthof thetruncatedOprFareindicatedbelow.51Strains constructed by this method were based on plasmid, pRW3, whichencoded oprF variants with the mutated oprF promoter described aboveand had a 12 bp insertion, including a unique PstI site, inserted intodifferent regions of the oprF sequence. An adapter oligonucleotide, withPstI overlaps and stop codons in all three reading frames, was designedso that upon insertion the PstI site would be lost and an XbaI siteintroduced. Initial screening was done by colony immunoblotting.Colonies that were reactive with the N-terminus specific monoclonalantibody MA7- 1 and non-reactive with the C-terminus specificmonoclonal antibody MA5-8 were selected for restriction endonucleaseanalysis. HindIII-EcoRI fragments were isolated from plasmids thatlacked the PstI site and contained an unique XbaI site and then insertedinto pUCP19. Plasmids constructed this way were designated pER188,pER213, pER215, pER231, and pER29O (Figure 4). The plasmid numberindicates the number of amino acids that were encoded in the mutantprotein. For example, pER 188 encodes the N-terminal 188 amino acidsof OprF. This method of truncation resulted in the insertion of a varyingnumber of nucleotides before the intended stop codon. The amino acidswhich they encoded varied according to the reading frame of the originalinsertion and are listed in Table III. Insertions at amino acids 188 and215 added three amino acid residues: aspartate, leucine and histidine.Insertions at amino acids 213, 231, and 290 added six additional amino52Table III. Amino acid residues introduced by linker-insertiontruncation of OprF.insertion sites amino acids amino acidsplasmids (amino acid) inserteda replacedpER188 188 DLH DNVpER2 13 213 TCTSLD VQLDVKpER215 215 DLH LDVpER231 231 TCTSLD YADIKNpER29O 290 TCTSLD VNAVGYa The amino acids listed are followed by the stop codon, TAG53acids residues: threonine, cysteine, threonine, serine, leucine, andaspartate.As none of the linker mutants described above had insertions nearthe prospective hinge (proline-rich) region of the protein, polymerasechain reaction was used to produce proteins truncated in this region.The primer for the coding strand included the HindIII site in order tofacilitate cloning as well as for maintenance of the mutated promoter.The primers for the non-coding strand were designed to introduce a stopcodon after a selected amino acid, followed by an EcoRI site to permitcloning. The resulting HindIII-EcoRI fragments were cut with theappropriate enzymes and cloned into pUCP19. Final selection was madeby colony immunoblotting, selecting for colonies that were reactive withMA7-1. Two mutants were constructed in this way, pER158 and pER163(Figure 4), but after sequencing it was shown that pER 158 containedseveral unintended mutations. Repetition of the PCR did not produce anaccurate fragment and therefore this construction was not pursuedfurther. The insert in pER163 was fully sequenced to ensure that it wasfree from errors.The addition of carbenicillin to growth media was required tomaintain pUCP19-based plasmids in H636. However, the resultingfilamentation compromised the study of cell length. To solve thisproblem, a 2.0 kb fragment coding for tetracycline resistance from54pHP45c2-Tc (Fellay et al., 1987) was inserted into the EcoRI site ofselected plasmids. The control plasmid, pUCP19, with the addedtetracycline resistance cartridge was named pER. The plasmid encodingthe entire 326 amino acids of OprF was named pER326t. Otherplasmids had a “t” added to their name to differentiate them from theinitial constructs.D. Site-directed mutagenesis of the cysteines of OprF.The effect of heat and suboptimal concentrations of reducingagents on OprF indicated that the four cysteines of this protein form twodisuiphide bonds (Hancock and Carey, 1979; Martin, 1992). SDS-PAGEshowed OprF migrating at four different apparent molecular weights.The band with the lowest molecular weight corresponded with OprF thathad not been modified by heat or by 2-mercaptoethanol. The proteinwith the highest apparent molecular weight was the heat-modified, fullyreduced OprF. The remaining intermediate molecular weight bandsappeared to be OprF that was not heat modified but that had either oneor both of the disuiphide bonds reduced.Previous attempts to quantify the disuiphide bonds of OprF weredone by biochemical methods. Martin (1992) found that the method ofThannhauser et al. (1984), in which the protein is denatured withguanidine thiocyanate and the broken disuiphide bonds measured in a55colourimetric assay, identified only one disulphide bond. The method ofIyer and Klee (1973), which measures the rate of reduction, showed thatthe disulphide bonds were more available in heat denatured samples andthe method of Needleman et al. (1970) indicated that no free thiols werepresent. These results indicated that two disuiphide bonds may bepresent but not readily available for assay.In this study, a more direct approach was used to address thisproblem, the cysteines of OprF were replaced with serines. Site-directedmutagenesis, by PCR, changed the DNA sequence from TGC, encoding acysteine residue, to TCC, encoding a serine residue. Two of the fourcysteines, C 185 and Cl 91, (referring to their relative position from the N-terminus of the protein) were mutated using this method. For theconstruction of mutant C185S, the coding-strand primer was the sameas that used in the construction of pER 163, and included the mutatedpromoter and the HindIII site (Figure 4). The primer for the noncodingstrand contained a one-basepair mutation, changing the cysteine to aserine, and included the existing Sail site (Figure 4). The coding-strandprimer for the construction of C 19 iS contained the Sail site and also hada one-base mutation changing the cysteine to a serine. The noncodingstrand primer included the wild-type translational stop codon and addedan EcoRI site to permit cloning. The resulting fragments were digestedwith the appropriate enzymes and cloned into a pRW5 derivative in56which the corresponding fragment had been removed. These plasmidswere named pERC185S and pERC191S, respectively. A plasmidcontaining both of the mutated cysteines, pERC185S+C191S, wasconstructed by the replacement of the HindIII-Sall fragment frompERC19 iS with that from pERC185S. Following antibiotic selection,colonies were selected by colony immunoblotting that were reactive withmonoclonal antibody MA7- 1, indicating that OprF was produced, andnon-reactive with monoclonal antibody MA7-8. It has been shown thatdisulphide bond formation is required for the binding of MA7-8 (Finnenet al., 1992; Martin, 1992). Selected plasmids were sequenced to ensurethat the desired mutation was the only mutation present.E. Expression of OprF mutants in P. aeruginosa.Previous attempts to introduce cloned OprF into P. aeruginosa hadbeen unsuccessful (Woodruff, 1988). This failure may have been due toplasmid copy number, as discussed in Chapter i C, or due to the use ofconjugation as the method of plasmid introduction. Like OmpA (Havekeset al., 1976; Achman et al., 1978; Manoil and Rosenbusch, 1982) OprFmay have a role in the stabilization of mating pair formation (Nicas,1983). In this study, two alternative methods of plasmid introductionwere tested. The first utilized a transformation method modified for usewith P. aeruginosa (Olson et al., 1982). Despite the high number of57transformants obtained with the control strain H 103 and plasmidpUCP19, no transformants were recovered using the OprF-deficientstrain H636 and the plasmids pRW5 or pUCP19. There are a number ofreasons why this method may have been unsuccessful including therepeated incubations on ice and/or the solution used to make the cellscompetent which contained only 0.15 M MgC12. A preliminary studyindicated that H636 had a greater decrease in viability than H 103 whenheld on ice. Also, studies done at room temperature, for 1 hour,indicated that a buffer, pH 7.2, containing salt for osmotic protection aswell as MgCl2was required to maintain the viability of H636, but notH103.The second method tested was electroporation. Farina et aL (1990)found that water and dilute ionic solutions resulted in extensive cell lysisduring the preparation of P. aeruginosa for electroporation, and that theaddition of glycerol solved this problem. The solution used forelectroporation was composed of 15% (w/v) glycerol in 1mM Hepes- 1mMMgC12 and the incubations on ice were kept to a minimum. The high saltmedium used for the recovery period after electroporation included 50mM MgC12which had been shown to enhance P. aeruginosa recovery (M.Bains, personal communication). Initial selection was made by platingcells on a high osmolarity medium containing the appropriate antibiotic.The resulting colonies were shown to contain plasmids of the correct sizeby slot lysis. Figure 5 is a Western immunoblot of whole-cell lysates of58Ia.41 2 3 4 5 6 7 8 9 10 11Figure 5. Western immunoblot of whole cell lysates of H636 containingplasmids coding for truncated OprF. The monoclonal antibody used wasMA7-1. Lane 1, molecular weight markers (from top to bottom: 142.9,97.2, 50.0, 35.1, 29.7, 21.9 kDa), lane 2, H 103/pER; lane 3,H636/pER; lane 4, H636/pERlO2t; lane 5, H636/pERl63t; lane 6,H636/pER17O-26t; lane 7, H636/pERl88t; lane 8, H636/pER2l3t;lane 9, H636/pER2l5t; lane 10, H636/pER29Ot; lane 11,H636/pER326t.59recombinant OprF and truncated versions of OprF expressed in H636.The antibody used was MA7- 1 which bound to an epitope localized toamino acids 55-62 (Chapter 3B). Strain H636 containing pERl63t (lane5), pER17O-26t (lane 6) pERl88t (lane 7), pER2l3t (lane 8), pER2l5t(lane 9), pER29Ot (lane 10) or pER326t (lane 11) and the wild-typecontrol H 103/pER (lane 2) produced protein(s) binding MA7-1 and, withspecific exceptions, corresponding in relative mobility to the approximatenumber of amino acids encoded. These results show that the weakenedpromoter allowed the expression of the cloned full-length OprF and thetruncated-OprF mutants in an OprF-deficient strain of P. aeruginosa.Unexpectedly, two bands reactive with MA7- 1 were seen with the strainH636/pER2 13t (lane 8). One was truncated as expected, but the otherappeared to have the same molecular weight as the native protein. Insome experiments H636/pER29Ot appeared to have the same molecularweight as the native protein (lane 10). This suggested that theseplasmids were able to recombine into the chromosome, although itappeared that the plasmid was also retained in the case ofH636/pER2 13t. This apparent recombination was not always observedwith these constructs and was rarely observed with the other plasmids.The plasmids pER (lane 3), pERlO2t (lane 4) and pER23lt did notproduce a protein detectable by MA7- 1 in the E. coli strain DH5c or inthe P. aeruginosa strain H636 indicating either that these proteins had60not been synthesized or that they had been degraded. While this wasexpected for the negative control H636/pER, (H636 has an 2-cartridgeinserted at amino acid 102) and H636/pERlO2t, it was surprising that aproduct was not observed in H636/pER23 it. Restriction mapping of thisplasmid indicated that oprF was present and did not appear to berearranged, but these results were not confirmed by sequencing. Theseresults indicated that between 102 and 163 N-terminal amino acids ofOprF are required for the production of a stable protein.The property of “heat modifiability” refers to an increased apparentmolecular weight of a protein on SDS-PAGE when pretreated by boilingin SDS or treated with denaturants such as TCA or urea compared to thesame protein solubilized at lower temperatures. When pretreated withTCA or boiled in SDS, the recombinant OprF, encoded on pER326t(Figure 6A, lane 3), pER29Ot and the OprF from the control strain, H 103,were all heat modifiable (Table IV). The truncated versions of OprFencoded on pERl63t, pER17O-26t, pERl88t and pER2l5t, however, hada decreased apparent molecular weight when pretreated with TCA orboiled in SDS (Table IV, examples in Figure 6B). A portion of eachsample tested retained the apparent molecular weight of the untreatedsample (Figure 6). This incomplete modification has been observedpreviously (Martin, 1992). These results indicated that between 215 and290 amino acids are required for the protein to be heat modifiable.61B.Figure 6. Western immunoblots of OprF and truncated OprF mutantswith and without TCA pretreatment. Odd numbered lanes pretreatedwith TCA. A. Lane 1, molecular weight markers; lanes 2 and 3,H636/pER326t (lower—weight bands in lane 2 presumed to be frequentlyobserved degradation products of OprF). B. Lane 1, molecular weightmarkers; lanes 2 and 3, H636/pERl63t; lanes 4 and 5, H636/pER17O-26; lanes 6 and 7, H636/pERl88t.A.123 1 23456762Table IV. Characteristics of truncated-OprF mutants inP. aeruginosa.strain/plasmid apparent mwa apparent mwa 2-ME modifiableb(kDa) after heating(kDa)H103/pER 37 41.5 +H636/pER———H636/pERl63t 23.5 20—H636/pERl88t 27 23.5 +H636/pER17O-26t 28 25.5—H636/pER2l5t 33 28.5 +H636/pER29Ot 35 39 +H636/pER326 37 41.5 +a as assessed by Western immunoblottingb + indicates an increase in apparent molecular weight after treatmentwith 2 —mercaptoethanol63These strains were also analyzed for modifiability with 2-mercaptoethanol. Full-length OprF from the wild type strain H 103/pERand from the strain H636/pER326t, and truncated OprF encoded onplasmicis pER 188t, pER2 15t, and pER29Ot, all ran at a higher apparentmolecular weight with the addition of 2-mercaptoethanol indicating thatthe cysteines encoded on these plasmids were present and had formeddisuiphide bonds (Table IV). The remaining truncated oprF did notcontain cysteines and their apparent molecular weights were not affectedby the addition of 2-mercaptoethanol (Table IV).The cysteine-to-serine OprF mutants were also introduced into theOprF-deficient strain H636 by electroporation. Colonies, selected byantibiotic resistance, were shown to contain plasmicls of the correct sizeby slot lysis. Whole cell lysates of these strains were prepared with andwithout 2-mercaptoethanol pretreatment. Western immunoblottingidentified proteins in all of the samples that reacted with the monoclonalantibody MA7-l (Figure 7, Table V). Bands, with apparent molecularweights greater than 50 kDa, observed in the samples untreated with 2-mercaptoethanol (lanes 6-8), were not detected in the samples pretreatedwith 2-mercaptoethanol (examples in lanes 3 and 5) or in the native OprF(lane 3 and 9). These bands may be multimers of OprF stabilized byintra or intermolecular disulphide bonding.641 2 3 4 5 6 7 8 9 10Figure 7. Western immunoblot of whole cell lysates heated to 1000 C for10 minutes. The monoclonal antibody used was MA7-1. Lanes 1 and10, molecular weight markers (from top to bottom: 142.9, 97.2, 50.0,35.1, 29.7 kDa). The samples in lanes 2-5 were treated with 2-mercaptoethanol. Lane 2, H636/pER; lanes 3 and 6, H636/pERC185S;lanes 5 and 7, H636/pERC191S; lane 8, H636/pERC185S+C191S; lane4 and 9, H103/pUCP19. F* indicates the heat-modified position of OprFfrom H103/pER.65Table V. Apparent molecular weight of cysteine-to-serine mutantsin P. aeruginosa.apparent molecular weightaunheated heated unheated heatedstrain/plasmid (kDa) (kDa) +2-ME +2-ME(kDa) (kDa)H103/pER 37 41 40.5 43.5H636/pERC185S+C191S 38 41.5 40.5 43.5H636/pERC191S 38 42 40.5 43.539H636/pERC185S 8.5 42.5 40.5 43.540a as assessed by Western immunoblotting.66All of the samples appeared to be heat modifiable with and withoutpretreatment with 2-mercaptoethanol. Bands with apparent molecularweights of 41 (H103/pER, lane 9), 41.5 (H636/pERC185S+C191S, lane8), 42 (H636/pERC191S, lane 7), 42.5 (H636/pERC185S, Table V) and43 kDa (all samples pretreated with 2-mercaptoethanol, lanes 3-5, TableV) were not detected in samples heated to 45° C.The apparent molecular weight of the native OprF (lane 9)increased with the addition of 2-mercaptoethanol (lane 4) as did that ofthe cysteine-to-serine mutants (examples in lanes 3 and 5; Table V)indicating the presence of at least one disulphide bond in all of thesamples. With the addition of 2-mercaptoethanol, the apparentmolecular weight of all of the samples was similar, indicating that thecysteines located at amino acids 185 and 191 were not essential for thesynthesis of a stable, full length protein.Without the addition of 2-mercaptoethanol, a band(s) with anapparent molecular weight higher that of the native OprF (37 kDa) andlower than the native OprF pretreated with 2-mercaptoethanol (40.5) wasobserved in all of the mutants. Strain H636/pERC185S+C191S had oneband with an apparent molecular weight of 38 kDa indicating that thecysteines at amino acid 176 and 205 had formed a disulphide bond.H636/pERC191S had a major band of 39 kDa and a minor band of 38kDa, and H636/pERC185S had two bands of 38.5 and 40 kDa (Table V).67This suggested that more than one combination of the non-mutagenizedcysteines may be able to form a disuiphide bond and furthermore, thatthe residual cysteines in the single mutants pair in such a way to confera different apparent molecular weight, possibly due to a differentconfiguration of the disuiphide region.F. Outer membrane and peptidoglycan association of OprF mutants.An SDS-PAGE of outer membrane preparations of H 103/pER andH636/ER and H636/pER326t is shown in Figure 8. These resultsindicated that the cloned OprF was associated with the outer membranewhen expressed in the OprF-deficient strain, H636. By visualcomparison with other proteins in these samples, it appeared that thecloned OprF was produced at roughly the same levels as the native OprF.Accordingly, outer membranes of the truncated OprF strains wereprepared. Figure 9 shows a Western immunoblot of these samples. Thetruncated and full-length OprF expressed in H636 containing pERl63t(lane 6), pER17O-26t (lane 5), pERl88t (lane 4), pER2l5t (lane 3),pER29Ot (lane 2) and the wild-type control (lane 1) were all associatedwith the outer membrane. The cysteine-to-serine OprF mutants werealso associated with the outer membrane (M. Bains, personal6841 2 4Figure 8. SDS-PAGE of outer membrane preparations of H 103/pER(lane 1); H636/pER (lane2); H636/pER326t (lane 3); molecular weightmarkers (from top to bottom: 97.4, 66.2, 45.0, 31.0, 21.5 kDa) (lane 4).The position of OprF is indicated.3691 2Figure 9. Western immunoblot of outer membrane preparations oftruncated versions of OprF. Monoclonal antibody used was MA7- 1.Lane 1, H 103/pER; lane 2, H636/pER29Ot; lane 3, H636/pER2l5t; lane4, H636/pERl88t; lane5, H636/pER17O-26t; lane 6, H636/pERl63t;lane 7, molecular weight markers (from top to bottom: 142.9, 97.2, 50.0,35.1, 35.1, 29.7 and 21.9 kDa).3 4 5 6 770communication). These results indicated that the N-terminal 163 aminoacids of OprF were sufficient for outer membrane association and thatthe wild-type conformation of the disuiphide region was not required.OprF has been shown to be non-covalently associated with thepeptidoglycan in strain H103 (Hancock and Carey, 1979). Thisassociation has been proposed to contribute to cell shape and stabilitysince the peptidoglycan is involved in both functions. To showpeptidoglycan association, Triton-EDTA and Triton-lysozyme solublefractions were prepared as described in the Materials and Methodssection. The Triton-EDTA soluble fraction contained proteins that werestabilized in the outer membrane by LPS association with divalentcations like MgC12. Incubation of the Triton-EDTA insoluble fraction withlysozyme released outer membrane proteins that are associated with thepeptidoglycan. Figure 10 shows a Western immunoblot of Triton-EDTAfractions (lanes 2, 4, 6, 8, 11, and 13) and Triton-lysozyme fractions(lanes 3, 5, 7, 9, 12, and 14) from strains expressing full-length ortruncated OprF. OprF from H 103/pER and H636/pER326t was foundprimarily in the Triton-lysozyme soluble fraction (lanes 3 and 5). A smallamount was observed in the Triton-EDTA fraction (lanes 2 and 4) and thepellet remaining after Triton-lysozyme treatment. This may have beendue to incomplete digestion of the peptidoglycan. All of the truncatedproteins tested were located in the Triton-EDTA soluble fraction (lanes 6,7112345 678•19 10 11 12 13 14Figure 10. Western immunoblot of outer membranes preparations oftruncated OprF’ solubilized in Triton-EDTA (lanes 2, 4, 6, 8, 11, and 13)or Triton-lysozyme (lanes 3, 5, 7, 9, 12, and 14). Monoclonal antibodyused was MA7-1, Molecular weight markers (lanes 1 and 10), from top tobottom, 112, 84, 53.2, 34.9, 28.7, and 20.5 kDa. H103/pER (lanes 2and 3); H636/pER326t (lanes 4 and 5); H636/pERl63t (lanes 6 and 7);H636/pER17O-26t (lanes 8 and 9); H636/pER2l5t (lanes 11 and 12);H636/pERl88t (lanes 13 and 14). An arrow indicates the position ofH636/pER17O-26t,728, 11, and 13) indicating that they were stabilized by association withLPS and divalent cations in the outer membrane, but were not associatedwith the peptidoglycan. A small amount was seen in the initial Triton-X100 solubilization. These data indicated that more than 215 amino acidsof OprF were required for association with the peptidoglycan and that,like the wild type OprF, the recombinant OprF was peptidoglycanassociated.G. Summary.Plasmids encoding truncated mutants of OprF were constructed byPCR, linker-mutagenesis, and use of an existing mutant. The resultingplasmids, and a plasmid encoding full length OprF, were successfullytransferred to the OprF-deficient strain H636 by electroporation.Plasmids that encoded at least 163 N-terminal amino acids of OprFexpressed proteins detectable by Western immunoblot. These proteinswere all associated with the outer membrane, but only the full-lengthOprF was associated with the peptidoglycan. Plasmids that encodedtruncated-OprF mutants with at least two cysteines produced proteinsthat were 2-mercaptoethanol modifiable. Like the native OprF, the OprFencoded on plasmid pER326t and pER29Ot were heat modifiable,showing an increase in the apparent molecular weight when boiled inSDS or pretreated with TCA. In contrast, the remaining truncated73mutants of OprF had a decrease in their apparent molecular weight withthese treatments.Two cysteine-to-serine OprF mutants were constructed by PCR anda third mutant containing both of the cysteine-to-serine mutations wasconstructed by genetic manipulation. These plasmids wereelectroporated into the OprF-deficient strain, H636. The encodedproteins were associated with the outer membrane of this strain andwere heat and 2-mercaptoethanol modifiable. They had an apparentmolecular weight higher than than that of the native OprF and lowerthan that of the native OprF pretreated with 2-mercaptoethanolindicating that mutation of cysteines located at amino acids 185 and 191had affected disuiphide bond formation. The two single cysteine-toserine OprF mutants produced two bands detected by Westernimmunoblotting which indicated that the remaining three cysteines maybe able to bond in more than one conformation.74CHAPTER 2. FUNCTIONAL ANALYSIS OF OprF MUTANTS INP. aeruginosa.A. Introduction.The contribution of OprF to the stability of the outer membrane ofP. aeruginosa has been studied with OprF-deficient strains (Woodruff andHancock, 1988; Gotoh et al., 1989a). These studies have shown thatOprF has a role in the shape of the cell and in growth in low osmolaritymedia. These properties were assessed in the mutants described inChapter 1.B. Growth of OprF mutants.Despite the instability of the outer membrane of OprF-deficientstrains, as described in the Introduction, strains with this defect havebeen observed in clinical situations. This indicated that OprF is not anessential component of the cell and that its deficiency may in fact confersome advantage in vivo (Piddock et al., 1992; Chamberland et al., 1990).To determine if H636 had a growth disadvantage in vivo, H636 and itsparent strain H 103 were grown in chambers implanted in the peritoneumof mice. This model is useful because it permits the growth of more thanone strain of bacteria per animal, in separate chambers, thus reducingthe animal-to-animal variation. Also, because the ends of the chambers75are made from 0.2 jim pore size filters, the bacteria are retained in thechamber (permitting their recovery), the host cells are excluded(simplifying the results), and in vivo nutrients are taken up (permittinggrowth of the bacteria). Results from one experiment are shown inFigure 11. After a lag of 4 hours for H 103 and between 4 and 8 hours forH636, both strains grew, with doubling times of 40 and 53 mmrespectively, reaching stationary phase after about 20 hours. Thisindicated that OprF-deficient strains did not have a major growthdisadvantage in vivo.Previous studies have shown that OprF is required for the growthof P. aeruginosa in low osmolarity media (Woodruff and Hancock, 1988;Gotoh et al., 1989a; Nicas and Hancock, 1983). The OprF deficient strainM-2F-, described in Chapter 1, was tested for its ability to grow in highand low osmolarity media. The strains M2F- and H636 as well as theirwild-type parent strains were able to grow at a similar rate in the highosmolarity medium (Figure 12B), but only the parent strains were able togrow in the low osmolarity medium (Figure 12A). This indicated that theinability to grow in a low osmolarity medium is due to the 2-cartridgemutagenesis of OprF and is not limited to the strain H636.To determine if the entire OprF was required for growth underthese conditions, the truncated OprF mutants, described in Chapter 1,were grown in high and low osmolarity media. Strain H636 had beenViableCount(logcfu/ml)ITJc?i (DCi).P’01-100O00 b-I 0C o1%)m0oCi)I.,0C •0 CR 0ON77A. B.400 4001100 110010 107I I I I I I 7 I I I I012345678 012345678Time (hours) Time (hours)Figure 12. Growth of OprF-deficient strains in A) LB no NaC1 or B) LB200 mM NaC1. The strains are indicated as follows: H 103(0); H636 (•);M2 (a); M2F- (a).78shown to be unable to grow in low osmolarity media (Woodruff andHancock, 1988). With the addition of 200 mM NaC1, the generation timeof H636 had been shown to be similar to its parent strain, H103(Woodruff, 1988). In the initial growth studies, the addition oftetracycline to ensure plasmid maintenance appeared to inhibit thegrowth of the controls. H 103/pER had a lag period of 3-4 hours in bothhigh salt and low salt media and H636/pER was unable to grow, in the 6hours tested, in either medium. When plated with tetracycline at 100pg/mi, the H 103/pER colonies were small and had rough edges.Normally colonies of H 103 are larger and smooth. As the outermembrane of P. aeruginosa is stabilized by divalent cations, and astetracycline can act as a divalent-cation chelator (Nikaido and Thanassi,1993), MgC12was added to the growth media. The addition of 5 mMMgCl2 permitted the expected growth of H 103/pER in both high and lowsalt media. The effect of the addition of 5 mM or 15 mM MgC12 on thestrains H636/pER17O-26 and H636/pER is shown in Figure 13. Theaddition of 15 mM MgC12 resulted in growth of both strains in the highosmolarity medium but did not permit the growth of the strain,H636/pER in the low osmolarity medium. The addition of 15 mM MgC12to the low osmolarity medium permitted growth of the strainH636/pER17O/26 at a rate greater than with the addition of 5 mMMgC12. Like H 103/pER, H636/pER17O-26 was able to grow in the high79A. B.1 1—I Io 0o 0o 00.1 0.10.05 I I 0.050123456Time (hours) Time (hours)Figure 13. Growth of H636/pER17O-26t (A) and H636/pER (B) in mediawith and without the addition of MgC12. Mueller Hinton broth (0), MuellerHinton broth + 15 mM MgCl2(), PP2 (•), PP2 + 5mM MgC12 (a), and PP2 +15mM MgC12(+).0123456I I I I I80osmolarity medium without the addition of MgC12 after about threehours. With the addition of 5 mM MgC12 to the growth media, all of thestrains had smooth colonies. These results suggest that the addition ofMgC12 protected the cells from the divalent-cation chelating (Nikaido andThanassi, 1993) effects of tetracycline.Figure 14 shows an example of the growth of the truncated OprFstrains in low and high osmolarity media with the addition of 5 mMMgC12 and 200 tg/ml tetracycline. All of the strains tested were able togrow at about the same rate in the high osmolarity medium. In the lowosmolarity medium H636/326t grew at the same rate as the wild typestrain and H636/pER showed little or no growth during the 7 hourstested. In the low NaC1 medium, the strains with truncated OprF grew ata rate similar to that of the wild-type strain for between 0 and 3 hourswith the length of time of this initial growth varying in independentexperiments (N.B. in the growth experiment of strain H636/pER17O-26shown in Figure 13 there was no period of wild-type growth rate with theaddition of 5 mM MgC12). This initial ability to grow at a rate near that ofthe wild-type strain may have been due to carry over from the highosmolarity medium used for the growth of the overnight culture or to thestage of growth of the overnight cultures used in these experiments.After this initial phase, the remaining strains appeared to fall into 2groups in all experiments. H636/pER29Ot and H636/pER2l5t hadgrowth rates less than that of the full length OprF but81A. B.400 400100 / / 1oo//__4•I7/ _—_/ •1./_/ — _/_(10 0 0 0 0 0 0 108I I I I 8 I I I I012345678 012345678Time (hours) Time (hours)Figure 14. Growth of truncated forms of OprF in A) LB no salt + 200 pg/m1tetracycline/5 mM MgC12 or B) LB + 200mM salt + 2OOig/ml tetracycline/5mM MgC12. The strains are indicated as follows: H636/pER (0);H636/pERl63t(•); H636/pER17O/26t (s); H636/ 188t (•);H636/pER2l5t (•); H636/pER29Ot (+); and H636/pER326 (.). H103 wasidentical to H636/pER326t and was omitted for clarity.82greater than the other strains, with the rate of H636/pER29Ot beinggreater than that of strain H636/pER2 15t. A Western immunoblot of thewhole cell lysates of the strain H636/pER29Ot and the control strainH 103/pER taken after 7 hours of growth in the low osmolarity medium isshown in Figure 15. The apparent molecular weight of the truncatedOprF expressed in strain H636/pER29Ot was lower than that of thestrain H 103/pER indicating that in this experiment the plasmid had notrecombined with the chromosome as had been observed previously(Figure 5). The strains H636/pERl63t, H636/pER17O-26t orH636/pERl88t always grew at a rate higher than the OprF deficientstrain, H636/pER, but lower than that of the other strains. Although therate of strain H636/pERl63t was higher than that of H636/pERl88t andH636/pER170-26t in the experiment presented in Figure 14, this wasnot the case in all experiments. The results from these experimentsindicated that the OprF-deficient strain containing the plasmidpER/326t was able to grow at the same rate as the wild-type strain inboth high and low osmolarity media. The strains containing thetruncated versions of OprF were able to grow in the low osmolaritymedium, but not at the same growth rate as the wild-type strain. Thesestrains also appeared to be more susceptible to the chelating effects oftetracycline than the wild-type strain or the strain containing the fulllength OprF.831 2 3 4Figure 15. Western immunoblot of H636/pER29Ot (lane 2), H636/pER(lane 3), and H 103/pER (lane 4) The molecular weight markers (lane 1)are, from top to bottom, 112, 84, 53.2, 34.9, and 28.7 kDa.84The three cysteine-to-serine OprF mutant strains were also testedfor their ability to grow in a low salt medium. As shown in Figure 16,these strains grew at the same rate as the wild-type strain in both lowsalt and high salt media, indicating that the apparent disruption of thedisuiphide-bond region (Chapter 1 E) did not effect the ability of thesestrains to grow in a low osmolarity medium.C. Length of OprF mutants.Previous studies have shown that OprF-deficient strains wereshorter than their wild-type parent strains as judged by image analysis(Woodruff and Hancock, 1988) and by electron microscopy (Gotoh et al.,1989a). In this study, cells were grown with 200 pg/ml tetracycline inLBHS to mid log phase (50 Klett units) fixed, stained and measureddirectly from a video monitor connected to a phase contrast microscope.The study was conducted blind; the slides were coded and measurementswere made by Susan Farmer. At least 100 cells of each strain from 3different experiments were measured. The exception to this was thestrain H636/290t which was assessed twice. All of the cells had similarwidths but varied in their length. An example of one of theseexperiments is shown in Figure 17 and Table VI. The negative control,H636/pER, was 61% the length of its parent strain, H 103/pER. StrainH636 with pERl63t, pERl88t, pER17O—26t, pER2l5t, pER29Ot, or85A. B.400 4001oo /// iooz :II,II, —1o 1o7 I I I I I I 7 I I I I012345678 012345678Time (hours) Time (hours)Figure 16. Growth of cysteine-to-serine muants in A) LB no salt orB) LB + 200mM salt. The strains are indicated as follows:H103/pUCP19 (0); H636/pUCP19 (•); H636/pERC185s (a);H636/pERC3 (•); H636/pERC185s+C 19 is (•).862.502.00E1.50 -S- 1.00 -0.50 -0.001 2 3 4 5 6 7 8 9 10Figure 17. Length of strains of P. aeruginosa containing plasmidsencoding full-length or truncated versions of OprF. The strains are asfollows: H 103/pER (positive control) (lane 1), H636/pER (negativecontrol), (lane 2), H636/pERl63t (lane 3), H636/pERl88t (lane 4),H636/pER17O-26t (lane 5), H636/pER2l5t (lane 6), H636/pER290t(lane 7), H636/pER326t (lane 8), M-2 (lane 9), and M-2F- (lane 10). Atleast 100 cells were measured and the mean normalized to the length ofH 103/pER as determined by image analysis.87Table VI. Relative length of P. aeruginosa with truncated OprF.Strain/plasmid OprFt/OprFa P value b(%) cfH636 cfHlO3 cfH636 cfH636/pER /pER /pER326t /pER29OtH103/pER 100± 14 +-- +H636/pER 61±20 - + + +H636/pERl63t 77 ± 19 + + + +H636/p13R170-26t 86 ± 14 + + +-H636/pERl88t 85 ± 14 + + +-H636/pER2l5t 88 ± 12 + + +-H636/pER29Ot 91 ± 13 + +--H636/pER326t 97 ± 13 +---a OprFt (truncated versions of OprF) / wild-type OprF (%).b + indicates strains that are significantly different as assessed by theStudent’s t-test with P0.05.88H636/326t were 77, 85, 86, 88, 91 and 97% the length of H 103/pERrespectively. All of the strains were statistically larger than H636/pERand smaller than H 103/pER with the exception of H636/pER326t (TableVI). There was no statistical difference between strain H103/pER andstrain H636/pER326t. There was also no statistical difference betweenH636/pER29Ot and H636/pER326t and between the truncated-OprFstrains, including H636/pER29Ot, with the exception of H636/pERl63t.H636/pERl63t appeared to be of intermediate size between H636/pERand H636 containing pERl88t, pER17O-26t, pER2l5t and pER29Ot. Thestrain H636/pERl63t was statistically different from H636/pER and allof the truncated OprF strains. These results indicated that the clonedfull-length OprF was able to complement the size defect in the OprFdeficient strain, H636. Although the truncated versions of OprF alsoappeared to affect the size of the OprF deficient strain, with the exceptionof H636/pER29Ot, the increase in size appeared to be less than that withthe full-length protein.Also measured were the OprF-deficient strain, M-2F-, described inChapter 1 and its parent strain, M-2 (Figure 17). The strain M-2F- wasabout 70% of the length of the parent strain M-2, indicating that theeffect of the a-cartridge mutagenesis of oprF on the size of the cell wasnot limited to the strain H636. Although the length of the cysteine-toserine mutants were not analyzed in a blinded study, the length of 2589cells from one experiment, grown without the addition of carbenicillin,were measured. Filamented cells were not observed and the mutantsappeared to be about the same length as the wild-type strain,H 103/pUCP19.E. Summary.The c-cartridge mutagenesis of OprF in the strain M-2 resulted inthe same phenotype as the previously c-cartridge mutagenized strainH636. M-2F- was unable to grow in a low salt medium and was 70% ofthe length of its parent strain. In vivo growth of strain H636, in a mousechamber-model, was similar to that of its parent strain H 103, indicatingthat OprF was not required for growth of P. aeruginosa in vivo. Thestrain H636 containing cloned full-length OprF, truncated OprF andcysteine-to-serine mutagenized OprF were able to grow at the same rateas the wild-type strain, H 103/pER, in the high salt medium. In the lowsalt medium, however, only the strain containing the cloned full-lengthOprF and the cysteine-to-serine OprF mutants were able to grow like thewild-type strain. The strains of H636 containing the plasmids pER 163t,pER 170-26t, pER 188t, pER2 15t, or pER29Ot were able to grow in the lowosmolarity medium, but at a lower rate than pER326 or the wild-typestrain, H 103/pER. A similar pattern was observed in the length of thetruncated-OprF strains. Although these mutants were longer than the90OprF-deficient strain H636/pER, the entire OprF was required for a wildtype length cell. The cysteine-to-serine mutants appeared be similar inlength to the wild-type strain, H 103/pUCP19, indicating that wild-typeconformation of this region is not required for cell length.91CHAPTER 3. MONOCLONAL ANTIBODY STUDIES.A. Introduction.Although X-ray crystallography of outer membrane proteins is theideal method for the study of their structure, these proteins are verydifficult to crystallize. Sequence analysis (Martin, 1992; Woodruff andHancock, 1989) and the interaction of OprF-specific monoclonalantibodies with mutant OprF derivatives (Finnen et al., 1992; Wong et al.,1993) and with OprF-derived peptides (Rawling et al., 1995) havecontributed to the knowledge of the secondary structure of OprF. In thisstudy, linear epitopes and surface-exposed epitopes were identified.These data were included in an updated secondary-structure model ofOprF. The conservation of epitopes in strains of P. aeruginosa was alsodetermined.B. Linear-epitope mapping using overlapping octapeptides.Previous studies in this laboratory using truncated forms of OprF,including TnphoA derivatives, partially localized the epitopes recognizedby 10 OprF-specific monoclonal antibodies. Unfortunately, the epitopesfor most of the antibodies could only be crudely localized by this meansand it was not known to what extent the structure of recombinant C-terminal deletion derivatives reflected the structure of the native protein.92Overlapping octapeptides, starting at every second amino acid andcovering the entire 326 amino acids of OprF were synthesized on 160pins. ELISAs were performed with 10 monoclonal and 2 polyclonal antiOprF antibodies. Reactivity was observed with only 3 of the 10monoclonal antibodies as shown in Figure 18. MA7- 1 bound to 3 pinsclerivatized with the peptides VRGTYETG (a.a. 53 to 60, pin 27),GTYETGNK (a.a. 55 to 62, pin 28) and YETGNKKV (a.a. 57 to 64, pin 29).The binding of MA7- 1 to pin 28 was approximately three times that of 29and 10 times that of 27. MA7-2 bound primarily to the peptideNLADFMKQ (a.a. 237 to 244, pin 119) but did bind with low affinity tothe peptide IKNLADFM (a.a. 235 to 242, pin 118). MA5-8 bound to thepeptide with the sequence TAEGRAIN (a.a. 307 to 314, pin 154) 3 timesgreater than to the peptide NATAEGRA (a.a. 305 to 312, pin 153). MA7-3, MA7-4, MA7-5, MA7-6, MA7-7, MA7-8, and MA4-4 showed noreactivity, indicating that they did not recognize linear epitopes.To examine the extent of distribution of linear epitopes on the OprFsequence, the pins were reacted with polyclonal antibodies from pooledmouse and rabbit serum. The mouse serum was tested at a workingdilution of 1:5000 and repeated at a dilution of 1:350. The results weresimilar and those from the 1:350 dilution are presented in Figure 19.The antibodies bound with high affinity to pins 153 and 154, which isthe same region as the epitope for MA5-8 and bound more weakly to93U)QCuCu.00Cu.0E=U)Cu0=Cu.00.0EU)Cu0=Cu.00C.,.0Figure 18. ELISA readings of monoclonal antibodies (A) MA7- 1,(B) MA7-2 and (C) MA5-8 reacting with individual pins derivatized withoverlapping octapeptides from OprF. Background values were subtracted. Thepositions of the middle amino acids of the OprF-derived peptides, starting fromthe N-terminus, are indicated on the X-axis. These data are representative oftwo independent trials with essentially identical results.2A. MA7-11 50 100 150 200 250 300B. MA7-21 50 100 150 200 250 300C. MA5-8I. I I —020201 50 100 150 200 250 300Approximate OprF amino acid iocation94Eto=a’C.).00C,,.0EU)0.00C,,.0Figure 19. ELISA readings of (A) mouse, and (B) rabbit polyclonal serumreacting with individual pins derivatized with overlapping octapeptidesfrom OprF. Background values were subtracted. The positions of themiddle amino acids of the OprF-derived peptides, starting from theN-terminus, are indicated on the X-axis. These data are representative oftwo independent trials with essentially identical results.32103210A.. Polyclonal mouse anti-OprF AbL . Iii L . 1U J.1 50 100 150 200 250 300B. Polyclonal rabbit anti-OprF AbL I.LIJLhL1J ..50 100 150 200 250Approximate OprF amino acid location30095peptides composed of a.a. 15 to 22 (pin 8), a.a. 281 to 288 (pin 141), anda.a. 295 to 306 (pins 148-150). This serum bound purified OprF in astandard ELISA assay when used at a dilution of 1:100,000 (R. Wong,personal communication), and specifically recognized OprF on Westernimmunoblots of whole-cell lysates of wild-type P. aeruginosa, H 103, atdilution of 1:1,000.No high-affinity antibodies were detected in the polyclonal rabbitserum when tested at a dilution of 1:500 (Figure. 19). By increasing thedevelopment time, multiple weaker binding sites were observed. Of thesesites, those that bound with an absorbance of at least 0.5 includedpeptides of a.a. 13 to 20 (pin 7), 83 to 90 (pin 42), 117 to 124 (pin 54),127 to 134 (pin 59), 213 to 220 (pin 107), 231 to 238 (pin 116), 241 to248 (pin 121), 259 to 266 (pin 130), 277 to 284 (pin 139), 281 to 288 (pin141), and 289 to 296 (pin 145). Peptides of a.a. 245 to 252 (pin 123) and295 to 306 (pin 149 and 150) had approximately 2- to 2.5-times greaterELISA reading than the other pins. This serum bound specifically toOprF on Western immunoblots of whole-cell lysates of wild-type P.aeruginosa at a dilution of 1:500 (A. Sukhan, personal communication).The two polyclonal sera appeared to share weak binding only to pins 141and 149-150. These results suggested that, as for the monoclonalantibodies, the majority of the polyclonal antibodies produced weredirected against conformational epitopes.96C. Binding of OprF-specific monoclonal antibodies to OprF mutants.The location of the OprF-specific monoclonal antibody epitopes inthe primary-structure of OprF was useful for the analysis of thetruncated-OprF mutants and the disuiphide-bond mutants. The resultsfrom colony immunoblotting, and was confirmed in selected casesWestern immunoblotting, of the truncated-OprF mutants is shown inTable VII. The strains H636/pER326t and the positive control H 103/pERbound all of the monoclonal antibodies tested confirming that the clonedOprF was expressed in P. aeruginosa and had assumed a conformationsimilar to that of the native OprF. The negative control strain,H636/pER, truncated by 2 mutagenesis at amino acid 102, did not bindany of the antibodies, including the N-terminal specific MA7- 1 (sectionB), indicating that no stable protein was produced. All of the remainingtruncated-OprF mutants were expressed and stable in P. aeruginosa asshown by their ability to bind MA7-1. The strains H636/pERl63t andH636/pERl88t did not bind any other antibodies, including MA7-8 andMA4-4. The epitopes for these antibodies have been previously locatedbetween amino acids 152 and 210 (Finnen et al., 1992). This indicatesthat part or all of these epitopes lie downstream of amino acid 188.H636/pER2l5t bound both MA7-8 and MA4-4 indicating that thistruncated protein was not degraded and, as both of these antibodiesTableVII.SummaryofOprF-specificmonoclonalantibodyreactivitywithtruncatedOprFinP.aeruginosa.MonoclonalantibodyreactivityStrain/plasmidMA7-1MA7-2MA7-3MA7-4MA7-5MA7-6MA7-7MA7-8MA44MA5-8H103/pER++++++++++H636/pER----------H636/pERl63t+---------H636/pER17O-26t+--------+H636/pERl88t+---------H636/pER2l5t+------++-H636/pER29Ot++++++++++1-H636/pER326t++++++++++98require disulphide bonds for binding (Finnen et al., 1992), it appearedthat the disuiphide region of this protein had assumed the wild-typeconformation. As expected, the strain H636/pER29Ot bound 9 of the 10monoclonal antibodies tested but also reacted with the monoclonalantibody MA5-8 in some experiments. As the epitope binding thisantibody has been located between amino acid residues 305 to 312(section B), it appears that this plasmid had recombined with thechromosome resulting in a full-length protein. These results areconsistent with those presented in Figure 5.The reactivity of the cysteine-to-serine mutants, encoded onplasmids pERC185S, pERC191S and pERC185S+C191S, with the OprFspecific monoclonal antibodies was also studied. As shown in Table VIII,these mutants bound all of the antibodies with the exception of MA7-8and MA4-4. The epitopes for MA7-8 and MA4-4 have been previouslylocalized to the cysteine-containing region of OprF by the analysis oftruncated mutants (Finnen et al., 1992) and papain- and cyanogenbromide-cleaved peptides (Rawling et al., 1995). It has also been shownthat OprF pretreated with 2-mercaptoethanol no longer binds theseantibodies (Rawling et al., 1995). These results indicated that thesemutations had affected the conformation of the disuiphide-bond region.TableVIII.SummaryofOprF-specificmonoclonalantibodyreactivitywithcysteine-to-serinemutantsinP.aeruginosa.Strain!plasmidMonoclonalantibodyreactivityMA7-1MA7-2MA7-3MA7-4MA7-5MA7-6MA7-7MA7-8MA44MA5-8+++++++++++++++++++++--++--++--+H103/pERH636/pERH636/pERC185SH636/pERC191sH636/pERC185S+C1915++++++++100D. Surface accessibility of the epitopes.The surface accessibility of epitopes binding monoclonal antibodiescan provide information useful in making secondary-structure modelsand for studying antigenic conservation. In this study, the OprF-specificmonoclonal antibodies were used in indirect immunofluorescence studiesand as opsonins in phagocytosis to identify surface-localized epitopes.The results of the indirect-immunofluorescent labeling of P. aeruginosastrain M-2, are presented in Table IX. High levels of fluorescence wereseen with MA7-1, MA7-3, MA7-4, MA7-7 MA7-8, MA4-4 and MA5-8.Uniform but weaker fluorescence was observed with MA7-5 and MA7-6.No fluorescence was observed with MA7-2 and with the negative controlantibody MA 1-3, specific for the lipoprotein, OprI. Another approach tostudy surface exposed epitopes was to use the monoclonal antibodies asopsonins for phagocytosis of the strain M-2 by mouse peritonealmacrophages. The antibody concentrations used were standardized byELISA with purified OprF and used at a dilution of 10-2. The results fromthis study, shown in Figure 20, confirmed previous data which hadshown that MA5-8 and MA4-4 were opsonic and that MA5-8 bindingresulted in greater uptake than MA4-4 (Battershill et al., 1987). MA7-3,MA7-5, MA7-6, MA7-7 and MA7-8 consistently opsonized strain M-2.MA7- 1 produced significant uptake in 1 out of 3 experiments, althoughwhen less diluted antibody was used, significant uptake was consistently101Table IX. Indirect immunofluorescence labeling of intact P.aeruginosa strain, M-2, with OprF-specific monoclonal antibodiesand the control antibody, MA1-3.Monoclonal antibody ImmunofluorescenceaMA7-1 ++MA7-2MA7-3 ++MA7-4 ++MA7-5 +MA7-6 +MA7-7 ++MA7-8 ++MA4-4 ++MA5-8 ++MA13ba- no fluorescence or only slight fluorescence observed, + cells uniformlyfluorescent, ++ cells highly fluorescent..b negative-control monoclonal antibody specific for a non-surfacelocalized epitope of outer membrane protein OprI.102Figure 20. Opsonic phagocytosis of M-2 by OprF-specific monoclonalantibodies. Results represent the average number of bacteria per mouseperitoneal macrophage. Each different shaded bar represents data froman independent experiment with the control bacteria per macrophage(obtained by using the negative control antibody MA1-3) subtracted. Thestars indicate samples that were significantly different from the control(by the Mann-Whitney test P -0.05).03 **C)hhh i7-1 7-2 7-3 7-4 7-5 7-6 7-7Antibodyiii4-4 5-8***7-8103observed. An increase in the concentration of MA7-2 and MA7-4 had noeffect on uptake; MA7-2 was no better than the control antibody, andMA7-4 produced significant results in only 1 of the 3 experiments. Themean uptake of cells with the negative-control monoclonal antibodyMA 1-3 was 1.5 bacteria per macrophage and which was subtracted fromthe values shown in Figure 20. The combined results from these surfacelocalization studies suggested that the epitopes for nine of themonoclonal antibodies were surface exposed and that the epitoperecognized by MA7-2 was not exposed in strain M-2.E. Conservation of epitopes binding OprF-specific monoclonalantibodies.Forty six P. aeruginosa strains including the 17 IATS serotypeisolates, 15 clinical strains from different countries, 11 environmentalisolates, and three laboratory isolates, including M-2, were studied forconservation of epitopes by colony immunoblotting with the 10monoclonal antibodies. All of the isolates, except for the negativecontrols H636 and M-2F- reacted with all of the antibodies and inselected cases this was confirmed by Western immunoblotting. Thesedata indicate that these epitopes of OprF were strongly conserved in P.aeruginosa.104F. Summary: A secondary structure model of OprF.Previous secondary structure models of OprF have been derivedfrom experimental evidence and predictive methods based on amino acidsequence analysis. The experimental results include the estimation of 13-strand content by CD analysis (Siehnel et al., 1989), the surfacelocalization of sites permissive for linker-insertion mutagenesis (Wong etaL, 1993) and the surface localization of the disulphide-bond dependentepitope of the monoclonal antibody, MA4-4 (Mutharia and Hancock,1985). CD analysis suggested a 13-sheet content of 62% in the nativeprotein, with 50% in the C-terminal 149 amino acid residues, which ismuch higher than the 25% predicted by amino acid sequence analysis(Siehnel et al., 1989; Jeanteur et al., 1994). Raman spectroscopy ofOmpA predicted 55% 13-sheet, with 60% in the N-terminal 177 aminoacids (Vogel and Jahnig, 1986). The original model (Siehnel et al., 1989)was modified by Wong et al. (1993) to include the surface localization ofthe permissive linker-insertion sites (Figure 21) (Wong et al., 1993). Ithas been shown that the loop regions of porins are more likely to toleratethe insertion of foreign DNA than the transmembrane strands (Cowan etal., 1992; Ried et al., 1994; Wong et at., 1993). The length of the loopsconnecting the transmembrane 13-strands were adjusted to conform tothe known structures of other porins. The periplasmic loops of these105‘981NP V TT VDG DP AT N YG C:CGNE K ioQ LE DN KPY K S LJ K S D i& A V AM T V N D R ?J D G E D H 290N K G H I S L 1J S V V A T A GR2 G NQ131G C:CDV*:S31D C Y EN31° AvD2tHEKEE DFGKVaFigure 21. Secondary structure model of OprF. The N-terminus is onthe bottom left-hand side of the model. Sites of linker-insertionmutagenesis indicated by shaded boxes. Predicted regions oftransmembrane f3- strands are boxed.106porins are all shorter than the surface-exposed loops which are ofvariable length.The updated model presented in this study has been modified fromthat presented in Rawling et al. (1995) to include the location of 13-strands predicted by the method of Jeanteur et al. (1994). These modelshave incorporated the results from the mimotoping and surface-localization studies, presented in section B and D respectively. Unlikeother prediction methods, the method of Jeanteur et al. (1994) has beentailored for the analysis of porins. Porins are a unique class of proteinspossessing transmembrane strands with f3-strand conformation(Jeanteur et al., 1994). It had been previously believed thattransmembrane strands of proteins could only have an ce-helicalconformation.The N-terminal domain of the current model is similar to that ofthe previous models (Rawling et al., 1995; Wong et al., 1993) in that it iscomposed of eight transmembrane 13-strands connected by shortperiplasmic loops and longer surface-exposed loops (Figure 22). Thelocation of the 13-strands in the primary structure of OprF was based onthose predicted by Jeanteur et al. (1994). This method of predictionresults in a conservative estimate in the length of the 13-strands (Jeanteuret al., 1994) and the length of the strands were accordingly extended toinclude contiguous regions of alternating hydrophobic and hydrophilic< MA7-4,5,7 ><- MA7-6-E—MA7-3 -. MA7-8, MA4-4 -MA71 V TN 200A DpAy TD NGKRDC:CNMirK E140 NK PR KK D G L G DN D DAV MA7-2190 AV20 N R S G H81A H N R QD GV EIioo Q Q G S V DN“210 y STD DG70r1 G E C:C R STflV ITMlli DGETNTG 1YIILI H AF ISDANALPIoIjallEio V iLl GIA iAlIFiIFI PIN HG iL Al L‘ii KiGi P K I T1 yE L EH SoH_jINt A L!_JF vT D DU IK12 N L° E G290N 40 T pY1 G N239 TAER80Q P I []T G KKR00 YGGN OK LSE270107MA5-80 a 0210 RE R290 Ij-GI ri hr IN1 IA I AEl hAl NNI I viI [J A IE20 LJhAE R___j 300 L!_VANFigure 22. Modified secondary structure model of OprF. The locationsof epitopes are indicated by outlined letters and by arrows. The N-terminus is on the bottom left-hand side of the model. Predicted regionsof transmembrane 13-strands are boxed.108amino acid residues. This resulted in a number of minor changes to theprevious models primarily affecting strands 6 and 7. The surface-exposed epitope for monoclonal antibody MA7- 1, identified bymimotoping, was located on the surface exposed loop 2. Theconcentration of tyrosine and phenylalanine residues at the hydrophobicprotein-lipid interface, the excess of negatively-charged residues at thelevel of the LPS core and the length of the f3-strands were consistent withthat of known porin structures (Jeanteur et al., 1994). The f3-sheetcontent of this region was similar to that predicted by CD analysis.The model of the C-terminal domain of OprF, like that of the N-terminal domain of OprF, is composed of eight transmembrane strands(Figure 22). This is consistent with the number of strands present in theknown structures of porins. Interestingly, Jeanteur et al. (1994)predicted only two C-terminal domain 13-strands, the first and seventh J3-strands presented in this model. Jeanteur et al.(1994) predicted that theregion located between these two 13-strands was composed ofamphipathic x-helices. In the model presented in this study, thetransmembrane strands in the C-terminal domain have been included as13-strands to reflect the data from CD analysis, although the total 13-sheetcontent is about 10 % lower than the predicted value. The epitopesbinding nine OprF-specific monoclonal antibodies have been shown to belocated in the C-terminal half of OprF (Finnen et al., 1992; Rawling et al.,1091995; Wong et al., 1993). In this study, eight of these epitopes wereshown to be surface exposed (Table IX, Figure 20). The region bindingone of these monoclonal antibodies, MA5-8, localized to eight aminoacids by mimotoping (Figure 19), was accordingly located in the loopconnecting f3-strands 15 and 16. The epitope for MA7-2 also localized bymimotoping did not appear to be surface located in wild-type cells (TableIX, Figure 20; Martin et al., 1994) and was located primarily in theeleventh transmembrane strand.The conformation and location of the cysteine-containing regionfrom the previous model (Wong et al., 1993) have been retained. Thesurface exposure of the epitopes binding the 2-mercaptoethanol sensitivemonoclonal antibodies support this location (Table VIII and Figure 20)and the results from the analysis of the cysteine-to-serine mutants(Chapter 1E) support the proposal that the cysteines form two disulphidebonds. Also in agreement with the location of this region is the surfacelocation of the proline-rich region of OprF from P. fluorescens whichappears to replace the cysteine-containing region of OprF from P.aeruginosa (De Mot et al., 1994).The region between amino acids 260 and 275 has been located inthe periplasm in this model. This region has few nonpolar amino acidresidues and is therefore unlikely to be a transmembrane strand. It hasbeen highly conserved in OmpA-related outer membrane proteins with 7110of the 14 amino acid residues being perfectly conserved in the 10proteins analyzed (De Mot et al., 1992). The C-terminal region of OprF(amino acid residues 163-326) has homology with the B. subtilis protein,MotB (not an outer membrane protein) and with the non-covalentlypeptidoglycan associated lipoproteins (PALS) and has been proposed tobe a site for peptidoglycan interaction (De Mot and Vanderleyden, 1994).111DISCUSSIONA. Aims of this study.The aims of this study were to confirm the function OprF bygenetic complementation and to assess the role of the C-terminal domainof the protein in cell length and growth on low osmolarity media.Another aim was to assess the disulphide bond region of OprF includingthe determination of the number of disulphide bonds present in thenative protein and the identification of the cysteines bonded. It washoped that these studies and those with the OprF-specific monoclonalantibodies would increase the understanding of the structure andfunction of this protein.B. Genetic complementation of Q-cartridge mutagenized OprF inP. aeruginosa.The function of OprF has been studied by the construction ofOprF-deficient strains by chemical mutagenesis (Gotoh et al., 1989a;Nicas and Hancock, 1983) and by transposon and interposon insertionalmutagenesis (Woodruff and Hancock, 1988). Although the length ofthese strains and their ability to grow in a low osmolarity medium wasgenerally the same, there was some phenotypic variation (described inChapter 1-B). To ensure that the more extreme results observed with 2-112cartridge mutagenesis of OprF were not strain specific, this method wasused to mutagenize the oprF gene in strain M-2. The resulting strain, M2F-, was unable to grow in a low osmolarity medium (Figure 12) and wasabout 30% shorter than its parent strain (Figure 17). This confirmedthat the phenotype previously observed was not strain-specific and thatOprF has a significant role in cell length and growth on low osmolaritymedium.The expression of certain outer membrane proteins, which canhave copy numbers of 10 per cell, from high copy number plasmids hasproven lethal (Duchene et al., 1988; Woodruff, 1988). In order tofacilitate the genetic complementation of an OprF-deficient strain of P.aeruginosa, the promoter of oprF was weakened by site directedmutagenesis (Wong et aL, 1993). The less efficient promoter permittedthe cloning of oprF into a E. coli-P. aeruginosa shuttle vector, pUCP19,which has a copy number of about 100 in E. coli and 10-25 in P.aeruginosa (Schweizer, 1991). Although the control plasmid, pUCP19,was introduced into the wild-type strain by transformation and byelectroporation, only electroporation resulted in the successful uptake ofplasmid DNA by the OprF-deficient strain, albeit at a low frequency. Thisunderscores the contribution of OprF to the cell. OprF appears toprovide protection from the potentially damaging procedures of113centrifugation, low temperatures and incubation of the cells in lowosmolarity buffers, all of which are required in transformation.Visual comparison indicated that the level of expression of OprF,encoded on plasmid pER326t, in the OprF-deficient strain H626 wassimilar to that of the parent strain, H 103/pER (Figure 8). Like the nativeOprF, the recombinant OprF had the same apparent molecular weight,was heat and 2-mercaptoethanol modifiable (Table IV), was associatedwith the outer membrane (Figure 8) and the peptidoglycan (Figure 10),and bound all of the OprF-specific monoclonal antibodies tested (TableVII). This strain was able to grow at the same rate as the wild-type strainin the low osmolarity medium (Figure 14) and regained the wild-type celllength (Figure 17, Table VI). The genetic complementation of the OprFdeficient strain, H636, confirmed the role of OprF in the growth of P.aeruginosa in low osmolarity media and in the determination of celllength.C. Protein analysis of mutated versions of OprF.The construction of plasmid-encoded truncated and cysteine-toserine mutants of OprF has enabled the assessment of the role of the C-terminal domain and the disulphide-containing region of OprF in theexpression and conformation of the encoded protein. In this study, itwas shown that between 102 and 163 N-terminal amino acids were114required for the stable expression of OprF as detected by Westernimmunoblot (Figure 5). This correlated with the minimum number ofamino acids required for the expression of OmpA in E. coli. Bremer et at(1982) constructed truncated-OmpA mutants by exonuclease digestion.A construct encoding the N-terminal 133 amino acids produced atruncated protein detected 45 s after radioactive labeling. No protein wasdetected 1 hr after labeling which indicated that the protein was unstableand had been degraded. A protein truncated at amino acid 193 wasstable (Bremer et at., 1982) as was a protein that contained a frame shiftafter amino acid 188 (188fs+1) (Klose et at., 1988b). It should be notedthat this latter mutation resulted in the addition of 18 non-OmpA aminoacids which could have affected the conformation of this protein or itsinsertion into the outer membrane. Recently, Ried et at (1994)constructed an OmpA mutant truncated at amino acid residue 171 bysite directed mutagenesis which also produced a stable protein.All of the truncated versions of OprF expressed in P. aeruginosawere associated with the outer membrane (Figure 9). These resultsindicated that the N-terminal 163 amino acids were sufficient forassociation with the outer membrane. Secondary structure models ofOprF (Figures 21 and 22) and OmpA (Morona et at., 1984; Klose et at.,1989; Ried et at., 1994) predict an N-terminal domain composed of 8transmembrane (3-strands, terminating at amino acids 160 and 170115respectively. A series of plasmids containing overlapping deletions ofOmpA were examined by immunoelectron microscopy for theirassociation with the outer membrane (Kiose et al., l988b). All of theconstructs containing amino acids 154-180, including the frameshiftmutant 188fs+ 1 and the mutant truncated at amino acid 193 describedabove, were associated with the outer membrane and all of those missingthis region were located in the periplasm. Although association with theouter membrane was not determined, the mutant truncated at aminoacid residue 171 was associated with the cell envelope (Ried et al., 1994).The importance of this region was confirmed by Kiose et al (1989). Inthat study, the eighth transmembrane strand, predicted to be composedof amino acids 158-170, was shown to be required for association withthe outer membrane by the mutation of leu’64 to pro and val’66 to asp.The resulting full-length, double mutant was located in the periplasm.These results indicate that the proposed eighth transmembrane strand isessential for association with the outer membrane. Unfortunately, inthis study, a mutant designed to terminate within the proposed eighthtransmembrane strand of OprF was found to be rearranged. Ihypothesize that the observed rearrangement in the PCR-amplified regionof this mutant was required to permit non-lethal production of afragment expressing the epitope binding the monoclonal antobody, MA7-1161. Mutants, truncated within the proposed eighth 13-strand woulddetermine if the results observed in OmpA are applicable to OprF.The outer membranes of many Gram-negative bacteria containOmpA- or OprF-like proteins which have been identified in part by thedistinctive property of heat modifiability (see Introduction). Thischaracteristic has been proposed to be due to the resistance of the f3-barrel structure to denaturation (Cowan et at., 1992), requiring boiling inSDS for unfolding. The extended structure of the unfolded protein has ahigher apparent molecular weight in SDS-PAGE than the folded formsince the latter is less compact. This property has also been used toassess the conformation of truncated mutants of OmpA. It has beensuggested that non-heat modifiable mutants are not folded correctly(Klose et at., 1988b; Ried et at., 1994). An OmpA mutant truncated atamino acid 193 and the frameshift mutant l8Ofs+ 1 were heat modifiableas well as being able to bind phages and colicins indicating that thesemutants were correctly assembled into the outer membrane (Kiose et at.,1 988b). The increase in the apparent molecular weight of the frameshiftmutant upon heating was slight compared to that of the wild-type strainand was attributed to the loss of the C-terminal portion of the proteinrather than to an effect of the additional 18 non-OmpA amino acids(Klose et at., 1988b). Interestingly, not all OmpA mutants have both theproperty of phage sensitivity and of heat modifiability indicating that117these mutated proteins may be inserted into the outer membrane in thewild-type conformation without the property of heat modifiability.Double mutants of OmpA, with amino acid changes in the proposed firstor eighth f3-strand, conferred phage sensitivity and produced a trypsinfragment of the expected-size but were not heat modifiable (Klose et al.,1988a). Without heating, these mutants had the apparent molecularweight of the heat-modified wild-type protein. OmpA mutants truncatedat amino acids 228 or 274 had a similar degree of heat modifiability asthe wild-type protein (Ried et al., 1994). Trypsin-cleaved fragmentsderived from mutants truncated at amino acid 228 and 274 and from thewild-type strain had the same apparent molecular weight, indicatingcorrect membrane insertion. It has been proposed that insertion into theouter membrane protects the N-terminal domain of OmpA from digestionby proteases (Chen et al., 1980). When heat treated, the trypticfragments derived from the native protein and from the mutantstruncated at amino acid 228 and 274 had a decrease, instead of theexpected increase, in their apparent molecular weight when compared tountreated samples (Ried et at., 1994). Another OmpA mutant, truncatedat amino acid 171 also exhibited this aberrant mobility (Ried et at.,1994). These studies indicated that although mutations in the Nterminal domain of the protein can affect heat modifiability, the entireprotein is not required for this property and that the portion of OmpA118that is required for the property of heat modifiability terminates betweenamino acids 171 and 193.In the study presented here, OprF truncated at amino acid 290and the full-length plasmid-encoded protein were heat modifiable,showing the expected increase in their apparent molecular weight (TableIV). The truncated-OprF mutants encoded on pERl63t, pER17O-26t,pER 188t or pER2 15t had a decreased apparent molecular weight whendenatured. These mutants did not appear to be degraded and themutants truncated at amino acid 188 and 215 remained 2-mercaptoethanol modifiable. These results indicated that between 215and 290 amino acids of OprF were required for the wild-type heatmodifiability. The amino acid alignment of OmpA and OprF indicatedthat regions located between amino acids 197 and 204 and betweenamino acids 215 and 222 respectively have significant homology with theeighth f3-strand of these proteins and has been predicted to form a f3-strand in both proteins (Jeanteur et at., 1994). This strand, like theproposed eighth 13-strand of both proteins, has strong homology to thesixteenth (terminal) 13-strand of the crystallized nonspecific porins, and inthese proteins is required for the correct membrane insertion of theseporins (Jeanteur et at., 1994). The contribution of this region to theformation of an SDS-stable (heat modifiable) structure in OmpF iscompatible with the experimental data, but further studies are required119to test this hypothesis. This region does not appear to be required forthis property in OmpA. Ried et al. (1994) suggested that the decreasedmobility of the non-heat modifiable OmpA mutants could be due to thedisassociation of a protein dimer. The relative change in the apparentmolecular weight observed with the OprF mutants is not consistent withdimer disassociation. It is also inconsistent with a heat/TCA inducedcleavage of the protein since the change in apparent molecular weightwas similar for all mutant proteins and no common fragment wasobserved. Another explanation is that these truncated versions of OprFmay be partly denatured by SDS before the heat treatment and thefurther treatment with heat or TCA may actually induce secondarystructure, resulting in the increased mobility on SDS-PAGE.D. Functional analysis of truncated OprF mutants.The conservation of the sequence of the C-terminal domain ofOmpA-like proteins, including OprF, is consistent with the hypothesisthat it has an important function in the cell. C-terminal deletionmutants were analyzed for their ability to grow in a low salt medium andfor their effect on cell length. Interestingly, the plasmid pERl63t, whichencoded only the N-terminal region of OprF, produced a significantincrease in cell length of the OprF-deficient strain (Figure 18) andpermitted its growth in a low salt medium, although at a rate lower than120the wild-type OprF (Figure 14). This suggested that this protein was notonly associated with the outer membrane but was inserted into themembrane and may have folded into a wild-type conformation. Thereappeared to be a further increase in cell length with the expression ofportions of the C-terminal region. A similar increase in length wasobserved with truncated versions of OprF encoded on pER17O-26t,pER 188t, and pER2 15t which suggested that the proline-alanine regionmay have a role in cell length. Without further testing, it was notpossible to draw any conclusions about the length of strainH636/pER29Ot. These results indicated that the entire OprF wasrequired for wild-type cell length.The growth of the truncated mutants in a low osmolarity mediumalso appeared to be affected by the number of amino acids encoded.When grown with tetracycline to maintain their plasmids, the mutantstruncated at amino acids 215 or 290 grew at a higher rate than theothers. The remaining mutants all grew at a rate higher than thenegative control strain. But, as with cell length, the entire proteinappeared to be required for wild-type growth in a low osmolarity medium.These experiments indicate that the C-terminal region of OprF had asignificant, though not complete, role in the contribution of OprF to thegrowth of P. aeruginosa in low osmolarity media.121E. Analysis of the disuiphide region of OprF’.The cysteine-containing region of OprF is highly conserved in P.aeruginosa and P. syringae. Thirty-two of the 34 amino acids thatcompose this region are perfectly conserved and 1 of the remaining 2amino acids is a conservative substitution. Amino acid alignment ofOmpA and the OprFs from P. aeruginosa and P. syringae locate thissequence within the proline-rich so called hinge region of OmpA.Similarly, alignments with OprF from P. fluorescens, which lacks thiscysteine region, also suggest that its insertion occurred in the prolinerich region. The titration of OprF with 2-mercaptoethanol, resulting in athree-step increase in apparent molecular weight, led to the proposalthat the four cysteines of OprF form 2 disulphide bonds (Hancock andCarey, 1979; Martin, 1992). Previous attempts to measure the numberof disulphide bonds by chemical methods identified one bond but alsoindicated that there were no free thiols present (Martin, 1992).In this study, another method was used in an attempt tounambiguously quantify the number of bonds and also to identify thecysteines involved in each bond. The cysteine-to-serine mutagenesis ofamino acid residues 185 and 191 appeared to result in the disruption ofa single disulphide bond as assessed by the change in apparentmolecular weight of these mutated proteins under non-denaturingconditions and by their failure to react with the 2-mercaptoethanol122sensitive monoclonal antibodies, MA7-8 and MA4-4. In the absence of 2-mercaptoethanol treatment, the apparent molecular weights of theseproteins were higher than that of the native OprF but were stillmodifiable with 2-mercaptoethanol to a position identical to that of OprF.This was consistent with the presence of one disulphide bond in each ofthe mutants and two in the native protein. An alternative explanation isthat these mutations altered the conformation of the protein resulting inthe bonding of free thiols. However, this explanation is not favored bythe observation that no free thiols have been detected (Martin, 1992).The apparent ability of the mutant proteins to bond in more than oneconfiguration (Figure 7) prevented the definitive identification of thecysteine residues involved in each bond. The mutation of the remainingtwo cysteine residues has been hampered by technical problems (ManjeetBains, personal communication), but may provide further insight into theanalysis of this region. I suggest that the best method of resolving thisdilemma would be to insert a methionine residue between amino acidresidues 185 and 191. Cleavage with cyanogen bromide and the analysisof the resulting peptides, with and without the addition of 2-mercaptoethanol, would unambiguously identiir one of the three possiblecombinations of bonding between the four cysteine residues (i.e. thatequivalent to the model in Figure 22). If required, the insertion of123methionines before and after these amino acids would distinguishbetween the two remaining combinations.No function has been assigned to the disuiphide region of OprF.Woodruff et al. (1986) proposed that alternate disuiphide bonding couldaccount for the two porin sizes identified by black lipid bilayer methods.The results with the cysteine-to-serine mutants and the double mutantsare consistent with this possibility. This region has a high degree ofhomology with the calcium binding repeats present in the eukaryoticextracellular matrix protein, thrombospondin and with the CD protein ofBranhamella catarrhalis, a respiratory tract pathogen (De Mot andVanderleyden, 1994). However, the significance of this observation isunknown.This study has shown that the native conformation of thedisuiphide region is not required for association with the outermembrane, heat modifiability (Figure 6), wild-type growth in lowosmolarity media (Figure 16) or for wild-type cell length, but has notidentified a clear role for this region.F. Analysis of epitopes binding OprF-specific monoclonalantibodies.Previous analysis of truncated mutants of OprF (Finnen et al.,1992), and protease- and cyanogen bromide-cleaved peptides (Rawling et124al., 1995) localized the epitopes binding 10 OprF-specific antibodies toregions of OprF that were 42-90 amino acids long. These regions were asfollows: MA7-1, amino acids 24-112; MA7-2, amino acids 250-273;MA7-3, amino acids 188-230; MA7-4, MA7-5 and MA7-7, amino acids188-278; MA7-6, amino acids 198-240; MA7-8 and MA4-4, amino acids152-2 10; MA5-8, amino acids 30 1-326. In this study, three of theseepitopes, binding MA7- 1, MA7-2 and MA5-8, were localized to regions ofeight amino acids by overlapping peptide methodology (Figure 22). Nineof these epitopes appeared to be surface exposed in strain M-2 asassessed by opsonic phagocytosis studies (Figure 20) and by indirectimmunofluorescence labeling (Table IX). The remaining epitope bindingmonoclonal antibody MA7-2 did not appear to be surface exposed. Theresults of indirect-immunofluorescence labeling of the strain H692(Martin et al., 1993), which has a rough LPS, with the OprF-specificmonoclonal antibodies were in agreement with the results from strain M2, with the exception of the binding of MA7-2. Unlike strain M-2, strainH692 showed a moderate level of fluorescence when reacted withmonoclonal antibody MA7-2 (Martin et al., 1993). This difference mightbe due to the masking of this epitope in strain M-2 by the smooth LPS(Martin et al., 1993). The surface localization of the epitope binding MA5-8, (amino acids 307-314) is in agreement with recent studies (Hughes etal., 1992; Von Specht et al., 1995). In both of these studies, polyclonal125sera raised against peptides composed of amino acids 308-326 (VonSpecht et al., 1995) and 305-318 (Hughes et al., 1992) bound to surfaceexposed epitopes as determined by immunofluorescence of intactbacteria or by opsonic-phagocytosis assays, respectively. The surfacelocalization of a region composed of amino acids 188-216 (Von Specht etal., 1995), and the lack of reactivity to serum against a peptide composedof amino acids 189-203 (Hughes et al., 1992) is also consistent with thedata in this study. In this study, the truncated OprF encoded onpER2 15t (Table VII) bound the surface localized 2-mercaptoethanol-sensitive monoclonal antibodies MA7-8 and MA4-4 (Figure 20, Table IX),while the truncated OprF encoded on pER188 did not (Table VII). Onemay speculate that a disuiphide bond involving the cysteines located atamino acids 191 and/or 205 may be required for the production ofantibodies reacting with this area of OprF. Polyclonal sera raised againstpeptides composed of amino acid residues 166-189, 2 15-226, 260-292Von Specht et al. (1995), 7-19, 177-190 or 242-56 (Hughes et al., 1992)did not bind surface-exposed epitopes. These results are in agreementwith the assignment of these regions in the proposed model of OprF(Figure 22). However, polyclonal sera raised against peptides composedof amino acid residues 54-67, 98-111, or 136-149 did not appear to bindsurface exposed epitopes (Hughes et al., 1992). In the present study, theepitope binding monoclonal antibody MA7- 1 was located between amino126acid residues 55 and 62 (Figure 18) and appeared to be surface exposedas assessed by indirect immunofluorescence labeling and by opsonicphagocytosis studies (Figure 20, Table VIII). Cells were highly fluorescentwhen labeled with MA7- 1. Unlike the other monoclonal antibodiesbinding surface-exposed epitopes, a higher concentration of MA7- 1 wasrequired for the consistent uptake of opsonized bacteria by macrophages.The reason for the discrepancy between the two studies is not clear but itmay be a reflection of the different methodologies used. Other conflictingdata are the surface (Hughes et al., 1992) and non-surface (Von Specht etal., 1995) localization of the region between amino acid residues 26 1-274and 260-292, respectively. Further studies are required tounambiguously determine the location of surface exposed loops in OprF.G. Conclusions and perspectives.This study has shown that the C-terminal domain of OprF has arole in the outer membrane stability of P. aeruginosa as assessed by celllength and growth in low osmolarity media and is required forpeptidoglycan association and is partially required for the property ofheat modifiability. It would be of interest to determine if this region isalso required for the porin function of the protein. The truncatedmutants constructed in this study could be assessed for pore formingability using black lipid bilayer methodology. The analysis of the pore127function and pore size of the cysteine-to-serine mutants of OprF wouldindicate whether wild-type conformation of the disuiphide region isrequired for porin function and may also be useful in testing thealternate-disuiphide bonding hypothesis proposed by Woodruff et al.,(1986) to account for the two sizes of pore measured. As well, theanalysis of cyanogen bromide-cleaved peptides, as described in thediscussion, may unambiguously determine the number andconfiguration of the disuiphide bond(s).This study has also shown that several regions of OprF, bindingOprF-specific monoclonal antibodies, are surface exposed and highlyconserved in strains of P. aeruginosa. These properties make it anpossible candidate for vaccine and diagnostic use.128REFERENCESAchtman, M., S. Schunchow, R. Helmuth, G. Morelli, and P.A.Manning 1978. Cell-cell interactions in conjugating Escherichia coli Con-mutants and stabilization of mating aggregates. Mol. Gen. Genet. 164:171-183.Angus, B.L. and R.E.W. Hancock 1983. Outer membrane porin proteinsF, P, and Dl of Pseudomonas aeruginosa and PhoE of Escherichia coli:chemical cross-linking to reveal native oligomers. J. Bacteriol. 155: 1042-1051.Battershill, J., D.P. Speert, and R.E.W. Hancock 1987. Use ofmonoclonal antibodies against protein F of Pseudomonas aeruginosa asopsonins for phagocytosis by macrophages. Infect. Immun. 55: 2531-2533.Beher, M.G., C. Schnaitman, and A.P. Pugsley 1980. Major heat-modifiable outer membrane protein in gram-negative bacteria:comparison with the OmpA protein of Escherichia coli. J. Bacteriol. 143:906-913.Beilido, F., N.L. Martin, R.J. Siehnel, and R.E.W. Hancock 1992.Reevaluation in intact cells of the exclusion limit and role of porin OprFin Pseudomonas aeruginosa outer membrane permeability. J. Bacteriol.174:Benson, S.A., J.L.L. Occi, and B.A. Sampson 1988. Mutations thatalter the pore function of the OmpF porin of Escherichia coli K12. J. Mol.Biol. 203: 96 1-970.Benz, R. 1988. Structure and function of porins from gram-negativebacteria. Ann. Rev. Microbiol. 42: 359-393.Benz, R. and R.E.W. Hancock 1981. Properties of large ion-permeablepores formed from protein F of Pseudomonas aeruginosa in lipid bilayermembranes. Biochim. Biophys. Acta. 646: 298-308.Benz, R., A. Schmid, and R.E.W. Hancock 1985. Ion selectivity ofgram-negative bacterial porins. J. Bacteriol. 162: 722-727.Benz, R. 1994. Uptake of solutes through bacterial outer membranes. In:Bacterial cell wall. J.M. Ghuysen and R. Hakenbeck, eds. Elsevier,Amsterdam. p. 397-424.129Blackwood, L.L. and J.E. Pennington 1981. Influence of mucoidcoating on clearance of Pseudomonas aeruginosa. Infect. Immun. 32:443-448.Bosch, D., W. Voorhout, and J. Tommassen 1988. Export andlocalization of N-terminally truncated derivatives of Escherichia coil K- 12outer membrane protein PhoE. J. Biol. Chem. 263: 9952-9957.Braun, G. and S.T. Cole 1984. DNA sequence analysis of the Serratiamarcescens ompA gene: implications for the organization of anenterobacerial outer membrane protein. Mol. Gen. Genet. 195: 321-328.Braun, V.1975. Covalent lipoprotein from the outer membrane ofEscherichia coil. Biochim. Biophys. Acta. 415: 335-377.Braun, V. and H.C. Wu 1994. Lipoproteins, structure, function,biosynthesis and model for protein export. In: Bacterial cell wall. J.-M.Ghuysen and R. Hakenbeck, eds. Elsevier, Amsterdam. p. 3 19-342.Bremer, E., S.T. Cole, I. Hindennach, U. Henning, E. Beck, C. Kurz,and H. Schaller 1982. Export of a protein into the outer membrane ofEscherichia coii K12 Stable incorporation of the OmpA protein requiresless than 193 amino-terminal amino-acids residues. Eur. J. Biochem.122: 223-23 1.Bryan, L.E. 1979. Resistance to antimicrobial agents: The generalnature of the problem and the basis of resistance. In: Pseudomonasaeruginosa: Clinical manifestations of infection and current therapy. R.Doggett, eds. Academic Press, New York. p. 219-270.Caulcott, C.A., M.R.W. Brown, and I. Gonda 1984. Evidence for smallpores in the outer membrane of Pseudomonas aeruginosa. FEMSMicrobiol. Lett. 21: 119-123.Chamberland, S., F. Malouin, H.R. Rabin, T. Schollaardt, T.R. ParrJr, and L.E. Bryan 1990. Persistence of Pseudomonas aeruginosa duringciprofloxacin therapy of a cystic fibrosis patient: transient resistance toquinolones and protein F-deficiency. J. Antimicrob. Chemother. 25: 995-1010.Chen, R., W. Schmidmayr, C. Kramer, U. Chen-Schmeisser, and U.Henning 1980. Primary structure of major outer membrane protein 11*(ompA protein) of Escherichia coil K12. Proc. Natl. Acad. Sci. U. S. A. 77:4592-4596.130Conrad, R.S. and H.E. Gilleland 1981. Lipid alterations in cellenvelopes of polymyxin-resistant Pseudomonas aeruginosa isolates. J.Bacteriol. 148: 487-497.Cowan, S.W., T Schirmer, G. Rummel, M. Steiert, R. Ghosh, R.A.Pauptit, J.N. Jansonius, and J.P. Rosenbusch 1992. Crystal structuresexplain functional properties of two E. coli porins. Nature. 358: 727-733.Cronan Jr., J.E. 1979. Phospholipid synthesis and assembly. In:Bacterial outer membranes. John Wiley and Sons, Toronto. p. 35-66.Day, S.E.J., K.K. Vasli, R.J. Russel, and J.P. Arbuthnott 1980. Asimple method for the study in vivo of bacteria growth and accompanyinghost response. J. Infect. Dis. 2: 39-5 1.De Mot, R., P. Proost, J. Van Damme, and J. Vanderleyden 1992.Homology of the root adhesin of Pseudomonasfluorescens OE 28.3 withporin F of Pseudomonas aeruginosa and Pseudomonas syringae. Mol.Gen. Genet. 231: 489-493.De Mot, R. and J. Vanderleyden 1994a. MicroCorrespondence. Mol.Microbiol. 13: 379-380.De Mot, R., G. Schoofs, A. Roelandt, P. Declerck, P. Proost, J. VanDamme, and J. Vanderleyden 1994b. Molecular characterization of themajor outer-membrane protein OprF from plant root-colonizingPseudomonasfluorescens. Microbiology. 140: 1377-1387.Duchene, M., A. Scheiezer, F. Lottspeich, G. Krauss, M. Marget, K.Vogel, B. von Sprecht, and H. Domdey 1988. Sequence andtranscriptional start site of the Pseudomonas aeruginosa outer membraneporin protein F gene. J. Bacteriol. 170: 155-162.Endermann, R., C. Kramer, and U. Henning 1978. Major outermembrane proteins of Escherichia coli K- 12: evidence for protein 11*being a transmembrane protein. FEBS Lett. 86: 21-24.Farinha, M.A. and A.M. Kropinski 1990. High efficiency electroporationof Pseudomonas aeruginosa using frozen cell suspensions. F’EMSMicrobiol. Lett. 70: 22 1-226.Fellay, R., J. Frey, and H. Krisch 1987. Interposon mutagenesis of soiland water bacteria: a family of DNA fragments designed for in vitroinsertional mutagenesis of Gram-negative bacteria. Gene. 52: 147-154.131Finnen, R.L., N.L. Martin, R.J. Siehnel, W.A. Woodruff, M. Rosok,and R.E.W. Hancock 1992. Analysis of the major outer membraneprotein OprF from Pseudomonas aeruginosa using truncated derivativesand monoclonal antibodies. J. Bacteriol. 174: 4977-4985.Freudi, R., S. Maclntyre, M. Degan, and U. Henning 1986. Cell surfaceexposure of the outer membrane protein OmpA of Escherichia coli K- 12.J. Mol. Biol. 188: 49 1-494.Freudi, R. and S.T. Cole 1983. Cloning and molecular characterizationof the ompA gene from Salmonella typhimurium. Eur. J. Biochem. 134:497-502.Freudl, R. and U. Henning 1988. On the role of the mature part of anEscherichia coli outer membrane protein (OmpA) in translocation acrossthe plasma membrane. J. Mol. Biol. 203: 5 17-519.Freudi, R. 1989. Insertion of peptides into cell-surface-exposed areas ofthe Escherichia coli OmpA protein does not interfere with export andmembrane assembly. Gene. 82: 229-236.Fung, J., T.J. MacAlister, and L.I. Rothfield 1978. Role of mureinlipoprotein in morphogenesis of the bacterial division septum:phenotypic similarity of lkyD and lpo mutants. J. Bacteriol. 133: 1467-1471.Gentry-Weeks, C.R., A. Hultsch, S.M. Kelly, J.M. Keith, and R.Curtiss III 1992. Cloning and sequencing of a gene encoding encoding a21 -kilodalton outer membrane protein from Bordetella avium andexpression of the gene in Salmonella typhimurium. J. Bacteriol. 174:7729-7742.Geysen, H.M., S.J. Rodda, T.J. Mason, G. Tribbick, and P.G. Schoofs1987. Strategies for epitope analysis using peptide synthesis. J. Immun.Meth. 102: 259-274.Goldberg, J.B. and D.E. Ohman 1984. Cloning and expression inPseudomonas aeruginosa of a gene involved in the production of alginate.J. Bacteriol. 158: 1115-1121.Gotoh, N, H. Wakebe, and E. Yoshihara 1989a. Role of protein F inmaintaining structural integrity of the Pseudomonas aeniginosa outermembrane. J. Bacteriol. 171: 983-990.132Gotoh, N., H. Wakebe, and T. Nishino 1989b. Ultrastructural aspects offragility of Pseudomonczs aeruginosa outer membrane devoid of protein F.FEMS Microbiol. Lett. 59: 5 1-54.Gotschlich, E.C., M.S. Blake, E.J. Lytton, and M. Seiff 1987.Gonococcal protein III. Purification and chemical characterization of theprotein, and the DNA sequence of the structural gene. Antonie vanLeeuwnhoek. 53: 455-459.Hancock, R.E.W., G.M. Decad, and H. Nikaido 1979. Identification ofthe protein producing transmembrane diffusion pores in the outermembrane of Pseudomonas aeruginosa PAO 1. Biochim. Biophys. Acta.554: 323-329.Hancock, R.E.W., R.T. Irwin, J.W. Costerton, and A.M. Carey 1981.The outer membrane of Pseudomonas aeruginosa: peptidoglycanassociated proteins. J. Bacteriol. 145: 628-631.Hancock, R.E.W. 1984. Alterations in outer membrane permeability.Ann. Rev. Microbiol. 38: 237-264.Hancock, R.E.W., A. Schmidt, K. Bauer, and R. Benz 1986. Role oflysines in ion selectivity of bacterial outer membrane porins. Biochim.Biophys. Acta. 860: 263-267.Hancock, R.E.W. 1987. Model membrane studies of porin function. In:Bacterial outer membranes as model systems. M. Inouye, eds. JohnWiley & Sons, Inc. New York.Hancock, R.E.W., R.J. Siehnel, and N.L. Martin 1990. Outermembrane proteins of Pseudomonas aeruginosa . Mol. Microbiol. 4: 1069-1075.Hancock, R.E.W., D.N. Karunaratne, and C. Bernegger-Egli 1994.Molecular organization and structural role of outer membranemacromolecules. In: Bacterial cell wall. J.-M. Ghuysen and R.Hakenbeck, eds. Elsevier, Amsterdam. p. 263-280.Hancock, R.E.W. and A. Bell 1988. Antibiotic uptake into Gram-negative bacteria. Eur. J. Clinical Microbiol. and Infect. Dis. 7: 7 13-720.Hancock, R.E.W. and A.M. Carey 1979. Outer membrane ofPseudomonas aeruginosa: heat- and 2-mercaptoethanol-modifiableproteins. J. Bacteriol. 140: 902-910.133Hancock, R.E.W. and A.M. Carey 1980. Protein D 1-a glucose inducible,pore-forming protein from the outer membrane of P. aeruginosa. FEMSMicrobiol. Lett. 8: 105-109.Hancock, R.E.W. and L. Chan 1988. Outer membranes ofenvironmental isolates of Pseudomonas aeruginosa. J. Clin. Microbiol.26: 2423-2424.Hardham, J.M. and L.V. Stamm 1994. Identification andcharacterization of the Treponema pallidum tpn5O gene, ompA homolog.Infect. Immun. 62: 10 15-1025.Havekes, L.M., B.J.J. Lugtengerg, and W.P.M. Hoekstra 1976.Conjugation deficient E. coli K12 F- mutants with heptose-lesslipopolysaccharide. Mol. Gen. Genet. 146: 43-50.Hughes, E.E., L.B. Gilleland, and H.E. Gilleland 1992. Syntheticpeptides representing epitopes of outer membrane protein F ofPseudomonas aeruginosa that elicit antibodies reactive with whole cells ofheterologous immunotype strains of P. earuginosa. Infect. Immun. 60:3497-3503.Iyer, K.S. and W.A. Klee 1973. Direct spectrophotometric measurementof the rate of reduction of disulfide bonds. J. Biol. Chem. 248: 707-7 10Jeanteur, D., J.H. Lakey, and F. Pattus 1991. The bacterial porinsuperfamily: sequence alignment and structure prediction. Mol.Microbiol. 5: 2153-2164.Jeanteur, D.., J.H. Lakey, and F. Pattus 1994. The porin superfamily:diversity and common features. In: Bacterial cell wall. J.-M. Ghuysenand R. Hakenbeck, eds. Elsevier, Amsterdam. p. 363-380.Jorgenson, R.A., S.J. Rothstein, and W.S. Reznikoff 1979. Arestriction enzyme cleavage map of Tn5 and location of a region encodingneomycin resistance. Mol. Gen. Genet. 177: 65-72.Kamlo, Y. and H. Nikaido 1976. Outer membrane of Salmonellatyphimurium: accessibility of phospholipid head groups to phospholipaseC and cyanogen bromide activated dextran in the external medium.Biochemistry. 15: 256 1-2570.134Kiose, M., S. Maclntyre, H. Schwarz, and U. Henning 1988a. Theinfluence of amino substitutions within the mature part of an Escherichiacoli outer membrane protein (OmpA) on assembly of the polypeptide intoits membrane. J. Biol. Chem. 263: 13297-13302.Kiose, M., H. Schwarz, S. Maclntyre, R. Freudi, M.-L. Eschbach, andU. Henning 1988b. Internal deletions in the gene for an Escherichia coliouter membrane protein define an area possibly important forrecognition of the outer membrane by this polypeptide. J. Biol. Chem.263: 13291-13296.Kiose, M., F. Jahnig, I. Hindennach, and U. Henning 1989.Restoration of membrane incorporation of an Escherichia coli outermembrane protein (OmpA) defective in membrane insertion. J. Biol.Chem. 264: 21842-2 1847.Lazzaroni, J. and R. Portalier 1992. The excC gene of Escherichia coli K-12 required for cell envelope integrity encodes the peptidoglycanassociated lipoprotein (PAL). Mol. Microbiol. 6: 735-742.Lugtenberg, B. and L. van Aiphen 1983. Molecular architecture andfunctioning of the outer membrane of Escherichia coli and other gramnegative bacteria. Biochim. Biophys. Acta. 737: 51-115.Maclntyre, S., R. Freucil, M.-L. Eschbach, and U. Henning 1988. Anartificial hydrophobic sequence functions as either an anchor or a signalsequence at only one of two positions within the Escherichia coli outermembrane protein OmpA. J. Biol. Chem. 263: 19053-19059.Manning, P., A.P. Pugsley, and P. Reeves 1977. Defective growthfunctions in mutants of Escherichia coli K12 lacking a major outermembrane protein. J. Mol. Biol. 116: 285-300.Manoil, C. and J.P. Rosenbusch 1982. Conjugation-deficient mutantsof Escherichia coli distinguish classes of functions of the outer membraneOmpA protein. Mol. Gen. Genet. 187: 148-156.Martin, N.L., E.G. Rawling, R.S.Y. Wong, M. Rosok, and R.E.W.Hancock 1993. Conservation of surface epitopes in Pseudomonasaeruginosa outer membrane porin protein OprF. FEMS Microbiol. Lett.113: 261-266.135Martin, N.L. and R.E.W. Hancock 1990. Function and structure of themajor components of the outer membrane of gram-negative bacteria. In:Advances in Brucellosis Research. G.L. Adams, eds. Texas A & MUniversity Press, College Station, TX. p. 55-75.Martin, N.L. 1992. Ph.D. thesis. University of British Columbia.Vancouver.Martinez de Tejada, G. and I. Moriyón 1993. The outer membranes ofBrucella spp. are not barriers to hydrophobic permeants. J. Bacteriol.175: 5273-5275.Miller, R.V. and J.M. Becker 1978. Peptide utilization in P. aeruginosa:evidence for membrane-associated peptidase. J. Bacteriol. 133: 165-171.Morona, R., M. Klose, and U. Henning 1984. Escherichia coli K- 12outermembrane protein (OmpA) as a bacteriophage receptor: analysis ofmutant genes expressing altered proteins. J. Bacteriol. 159: 570-578.Murphy, T.F., C. Kirkham, and A.J. Lesse 1993. The major heat-modifiable outer membrane protein CD is highly conserved amongstrains of Branhctmella catarrhctlis. Mol. Microbiol. 10: 87-97.Mutharia, L.M. and R.E.W. Hancock 1983. Surface localization ofPseudomonas aeruginosa outer membrane porin protein usingmonoclonal antibodies. Infect. Immun. 42: 1027-1033.Mutharia, L.M. and R.E.W. Hancock 1985. Characterization of twosurface-localized antigenic sites on porin protein F of Pseudomonasaeruginosa. Can. J. Microbiol. 31: 381-390.Needleman, S.B. and C.D. Wunsch 1970. A general method applicableto the search for similarities in the amino acid sequence of two proteins.J. Mol. Biol. 48: 443-453.Nelson, M.B., M.A. Apicella, T.F. Murphy, H. Vankeulen, L.D. Spotila,and D. Rekosh 1988. Cloning and sequencing of Haemophilus influerizaeouter membrane protein P6. Infect. Immun. 56: 128-134.Nicas, T.I. and R.E.W. Hancock 1983. Pseudomonas aeruginosa outermembrane permeability: isolation of a porin F-deficient mutant. J.Bacteriol. 153: 281-285.136Nicas, T. 1983. Ph.D. thesis. University of British Columbia. Vancouver.Nikaido, H. 1993. Transport across the bacterial outer membrane. J.Bioenerg. Biomemb. 25: 581-589.Nikaido, H. and R.E.W. Hancock 1986. Outer membrane permeabilityof Pseudomonas aeruginosa. In: The bacteria: a treatise on structure andfunction. J.R. Kotatch, eds. Academic Press, Inc. New York. p. 145-193.Nikaido, H. Nikaido, K., and S. Harayama. 1991. Identification andcharacterization of porins in P. aeruginosa. J. Biol. Chem. 266:770-779.Nikaido, H. and D.G. Thanassi 1993. Penetration of lipophilic agentswith multiple protonation sites into bacterial cells: tetracyclines andfluoroquinolones as examples. Antimicrob. Agents Chemother. 37: 1393-1399.Olson, R.H., G. DeBusscher, and W.R. McCombie 1982. Developmentof broad-host-range vectors and gene banks: self-cloning of thePseudomonas aeruginosa PAO chromosome. J. Bacteriol. 150: 60-69.Pennington, J.E., G.E. Small, M.E. Lostrom, and G.B. Pier 1986.Polyclonal and monoclonal antibody therapy for experimentalPseudomonas aeruginosa pneumonia. Infect. Immun. 54: 239-244.Peterson, G.L. 1977. A simplification of the protein assay method ofLowry et al. which is more generally acceptable. Anal. Biochem. 83: 346-356.Piddock, L.J.V., M.C. Hall, F. Bellido, M. Bains, and R.E.W. Hancock1992. A pleiotropic, posttherapy, enoxacin-resistant mutant ofPseudomonas aeruginosa. Antimicrob. Agents Chemother. 36: 1057-1061.Poole, K. and R.E.W. Hancock 1984. Phosphate transport inPseudomonas aeniginosa. Involvement of a periplasmic phosphate-binding protein. Eur. J. Biochem. 144: 607-6 12.Puohiniemi, R., M. Karvonen, J. Vuopio-Varkila, A. Muotiala, I.M.Helander, and M. Sarvas 1990. A strong antibody response to theperiplasmic C-terminal domain of the OmpA protein of Escherichia coli isproduced by immunization with purified OmpA or with whole Escherichiacoli or Salmonella typhimurium bacteria. Infect. Immun. 58: 1691-1696.137Rawling, E.G., N.L. Martin, and R.E.W. Hancock 1995. Epitopemapping of the Pseudomonas aeruginosa major outer membrane porinprotein OprF. Infect. Immun. 63: 38-42.Reeves, P. 1994. Biosynthesis and assembly of lipopolysaccharide. In:Bacterial cell wall. J.-M. Ghuysen and R. Hakenbeck, eds. Elsevier,Amsterdam. p. 281-318.Ried, G., R. Koebnik, I. Hindennach, B. Mutschler, and U. Henning1994. Membrane topology and assembly of the outer membrane proteinOmpA of Escherichia coli K12. Mol. Gen. Genet. 243: 127-135.Ried, G. and U. Henning 1987. A unique amino acid substitution in theouter membrane protein OmpA causes conjugation deficiency inEscherichia coli K-12. FEBS Lett. 223: 387-390.Rivera, M., L.E. Bryan, R.E.W. Hancock, and E.J. McGroarty 1988.Heterogeneity of lipopolysaccharides from Pseudomonas aeruginosa:analysis of lipopolysaccaride chain length. J. Bacteriol. 170: 512-521.Rivera, M. and E.J. McGroarty 1989. Analysis of a common-antigenlipopolysaccharide from Pseudomonas aeruginosa. J. Bacteriol. 171:2244-2248.Rolinson, G.N. 1986. Beta-lactam antibiotics. J. Antimicrob. Chemother.17: 5-36.Rutz, J.M., J. Liu, J.A. Lyons, J. Goranson, S.K. Armstrong, M.A.McIntosh, J.B. Feix, and P.E. Kiebba 1992. Formation of a gatedchannel by a ligand-specific transport protein in the bacterial outermembrane. Science. 258: 47 1-475.Saint, N., E. De, S. Julien, N. Orange, and G. Moile 1993. lonophoreproperties of OmpA of Escherichia coli. Biochim. Biophys. Acta. 1145:119-123.Sambrook, J., E.F. Fritsch, and T. Maniatis 1989. Molecular cloning:a laboratory manual. Cold Spring Harbour Laboratory, Cold SpringHarbour, NY.Schirmer, T., T.A. Keller, Y. Wang, and J.P. Rosenbusch 1995.Structural basis for sugar translocation through maltoporin channels at3.1 A resolution. Science. 267: 512-514.138Schweizer, H.P. 1991. Escherichia-Pseudomonas shuttle vectors derivedfrom pUC18/19. Gene. 97: 109-112.Siehnel, R.J., N.L. Martin, and R.E.W. Hancock 1989. Functional andstructure of the porin proteins OprF and OprP of Pseudomonasaeruginosa. In: Pseudomonas: biotransformations, pathogenesis, andevolving biotechnology. S. Silver, eds. A.S.M. Washington, D.C. p. 328-342.Simon, R., U. Priefer, and A. Pflhler 1983. A broad host rangemobilization system for in vivo genetic-engineering: transposonmutagenesis in Gram-negative bacteria. Bio/Technol. 1: 784-791.Smit, J., Y. Kamio, and H. Nikaido 1975. Outer membrane ofSalmonella typhimurium: chemical analysis and freeze-fracture studieswith lipopolysaccharide mutants. J. Bacteriol. 124: 942-958.Sonntag, I., H. Schwarz, Y. Hirota, and U. Henning 1978. Cellenvelope and shape of Esherichia coli: multiple mutants missing theouter membrane lipoprotein and other major outer membrane proteins.J. Bacteriol. 136: 280-285.Speert, D.P., M. Bond, R.C. Woodman, and J.T. Curnutte 1994.Infection with Pseudomonas cepacia in chronic granulomatous disease:role of nonoxidative killing by neutrophils in host defense. J. Infect. Dis.170: 1524-1531.Speert, D.P. and L. Thorson 1991. Suppression by human recombinantgamma interferon of in vitro macrophage nonopsonic and opsonicphagocytosis and killing. Infect. Immun. 59: 1893-1898.Spinola, S.M., G.E. Griffiths, K.L. Shanks, and M.S. Blake 1993. Themajor outer membrane protein of Haemophilus ducreyi is a membrane ofthe OmpA family of proteins. Infect. Immun. 61: 1346-135 1.Stieritz, D.D. and l.A. Holder 1975. Experimental studies of thepathogenesis of infections due to Pseudomonas aeruginosa: description ofa burned mouse model. Infect. Immun. 131: 688-69 1.Sugawara, B. and H. Nikaido 1992. Pore-forming activity of OmpAprotein of Escherichia coli. J. Biol. Chem. 267: 2507-2511.Tanji, Y., J. Gennity, S. Pollitt, and M. Inouye 1991. Effect of OmpAsignal peptide mutations on OmpA secretion, synthesis, and assembly. J.Bacteriol. 173: 1997-2005.139Thannhauser, T.W., Y. Konishi, and H.A. Scheraga 1984. Sensitivequantitative analysis of disulfide bonds in polypeptides and proteins.Anal. Biochem. 138: 181-188.Ulistrom, C.A., R.J. Siehnel, W.A. Woodruff, S. Steinbach, and R.E.W.Hancock 1991. Conservation of the gene for outer membrane proteinOprF in the Pseudomonaceae: Sequence of the Pseudomonas syringaeOprF gene. J. Bacteriol. 173: 768-775.Vaara, M. 1992. Agents that increase the permeability of the outermembrane. Microbiol. Rev. 56: 395-411.Vogel, H., F. Jahnig 1986. Models for the structure of outer-membraneproteins of Escherichia coli derived from Raman spectroscopy andprediction methods. J. Mol. Biol. 190: 19 1-199.Von Specht, B., B. Knapp, G. Muth, M. Broker, K. Hungerer, K.Diehi, M. Massarrat, A. Seeman, and H. Domdey 1995. Protection ofimmunocompromised mice against lethal infection with Pseudomonasaeruginosa by active or passive immunization with recombinant P.aeruginosa outer membrane protein F and outer membrane protein Ifusion proteins. Infect, and Immun. 63: 1855-1862.Won, J. and R.W. Griffith 1993. Cloning and sequencing of the geneencoding a 3 1-kilodalton antigen of Haemophilus somnus. Infect. Immun.61: 2813-2821.Wong, R.S.Y., H. Jost, and R.E.W. Hancock 1993. Linker-insertionmutagenesis of Pseudomonas aeruginosa outer membrane protein OprF.Mol. Microbiol. 10: 283-292.Woodruff, W.A., Parr, Jr., R.E.W. Hancock, L.F. Hanne, T.I. Nicas,and B.H. Iglewski 1986. Expression in Escherichia coli and function ofPseudomonas aeruginosa outer membrane porin protein F. J. Bacteriol.167: 473-479.Woodruff, W.A. 1988. Ph.D. thesis. University of British Columbia.Vancouver.Woodruff, W.A. and R.E.W. Hancock 1988. Construction andcharacterization of Pseudomonas aeruginosa porin protein F-deficientmutants after in vivo and in vitro mutagenesis of the cloned protein Fgene in Escherichia coli. J. Bacteriol. 170: 2592-2589.140Woodruff, W.A. and R.E.W. Hancock 1989. Pseudomonas aeruginosaouter membrane protein F: structural role and relationship to theEscherichia coli OmpA protein. J. Bacteriol. 171: 3304-3309.Yem, D.W. and H.C. Wu 1978. Physiological characterization of anEscherichia coli mutant altered in the structure of murein lipoprotein. J.Bacteriol. 133: 14 19-1426.Yoneyama, H. and T. Nakae 1986. A small diffusion pore in the outermembrane of Pseudomonas aeruginosa. Eur. J. Biochem. 157: 33-38.Yoshihara, E., N. Gotoh, and T. Nakae 1988. In vitro demonstration bythe rate assay of the presence of small pores in the outer membrane ofPseudomonas aeruginosa. Biochem. Biophys. Res. Commun. 156: 470-476.Yoshihara, E. and T. Nakae 1989. Identification of porins in the outermembrane of Pseudomonas aeruginosa that form small diffusion pores. J.Biol. Chem. 264: 6297-6301.Yoshimura, F., L.S. Zalman, and H. Nikaido 1983. Purification andproperties of Pseudomonas aeruginosa. J. Biol. Chem. 258: 2308-2314.Young, L.S. 1980. The role of exotoxins in the pathogenesis ofPseudomonas aeivginosa infections. J. Infect. Dis. 142: 626-630.