INTRODUCTION

Top Previous Next References

Lipopolysaccharides (LPS) of gram-negative bacteria are major components of the cell wall. The hydrophobic lipid A component of LPS secures these molecules in the outer membrane, while the core oligosaccharide links the lipid A region to the O antigen or O polysaccharide. The location of these molecules in the outer leaflet of the outer membrane permits interaction of LPS with the external milieu. As a result, early research focused on the role of LPS as a virulence determinant and on its use as a vaccine candidate. Since that time, studies have expanded to include analysis of the chemistry and biosynthesis of the O-antigenic region due to its immunogenicity, serotype specificity, and serum resistance properties.

With the medical and environmental importance of Pseudomonas aeruginosa, major efforts have been directed toward understanding the factors relevant to initial bacterial attachment, evasion of host defenses, and establishment of infection. LPS is one of these factors, and over the last decade remarkable advancements have been made in the fields of P. aeruginosa LPS chemistry and biosynthesis, in particular the progress in the area of O-antigen synthesis and assembly. While the O-antigen synthesis pathways of P. aeruginosa have many properties in common with other characterized LPS systems, unique features of synthesis have been identified. In this review, O-antigen synthesis pathways for the two LPS molecules produced by P. aeruginosa, A band and B band, are discussed in detail, along with future research areas. The role of LPS in the biology and pathogenesis of P. aeruginosa is also examined.

P. AERUGINOSA, THE PATHOGEN

Top Previous Next References

Occurrence of P. aeruginosa Infections

P. aeruginosa typically causes disease only in individuals with impaired host defenses and is thus referred to as an opportunistic pathogen. Such compromised individuals include patients undergoing immunosuppressive therapies (e.g., cancer treatment), those receiving treatment for traumatic skin damage (burn wounds), those with human immunodeficiency virus infections, and those with cystic fibrosis (CF) (78, 107). Recent clinical data indicates P. aeruginosa to be the fourth leading cause of nosocomial infection and the foremost cause of hospital-acquired pneumonia (95). Acquisition of this pathogen within the hospital setting is attributed to contaminated environmental reservoirs (e.g., sinks and respirators), as well as patient-to-patient spread (182).

CF patients in particular are highly susceptible to chronic pulmonary infections with P. aeruginosa. The lung environment of these patients appears to provide a unique niche that promotes chronic microbial colonization. Mutations in the CF transmembrane regulator protein (CFTR) interfere with chloride ion transport in CF patients. This electrolyte imbalance causes dehydration within the lungs and production of a viscous mucous, which significantly impairs mucociliary clearance mechanisms, allowing persistent bacterial colonization (115). Mutations in CFTR are also known to cause undersialylation of epithelial cell surface receptors, which increases P. aeruginosa adherence to host tissue (59, 121). These patients typically experience a progression in pulmonary pathogens, with Haemophilus influenzae and Staphylococcus aureus infections occurring in infants and children and P. aeruginosa and Burkholderia cepacia infections occurring during adolescence and adulthood (186). H. influenzae and S. aureus pulmonary infections are usually controllable with antimicrobial therapies, but effective clearance of these organisms allows subsequent colonization by and chronic establishment of P. aeruginosa within the lungs of these patients (238). More than 80% of CF patients over the age of 26 years are colonized with P. aeruginosa (68). It is these chronic P. aeruginosa respiratory infections which account for most of the pulmonary deterioration and mortality in CF patients, since this organism is usually the only pathogen recovered postmortem from the sputum and lung tissue (67).

Effective antibiotic therapy of P. aeruginosa infections has been problematic, largely due to the high intrinsic resistance of this organism to antimicrobial agents. This resistance is a result of the low permeability of the outer membrane (81), combined with the presence of both -lactamases (84, 178) and multidrug efflux pumps (119, 183, 184). The outer membrane is thought to reduce the passage of hydrophobic antibiotics due to the highly charged bacterial surface that is stabilized by divalent cations (180). The uptake of small hydrophilic antimicrobial agents, such as -lactams, occurs via porin proteins, but P. aeruginosa has only 1 to 5% the permeability of Escherichia coli for -lactams (83). -Lactam resistance is heightened through the presence of periplasmic -lactamases, which can be plasmid or chromosomally encoded (80). Recently, various efflux systems (mexAB-oprM [138, 184], mexCD-oprJ [183], and mexEF-oprN [184]) which are able to export structurally unrelated antibiotics, providing P. aeruginosa with multidrug resistance, have been described.

Efforts have been made to gain an in-depth understanding of the above-mentioned resistance mechanisms in an attempt to design more effective therapies for P. aeruginosa infections. One approach has been the coadministration of -lactams with -lactamase inhibitors. However, resistance to this treatment combination has also developed (79). There is also the approach of downregulating or inactivating multidrug efflux pumps for enhanced susceptibility to various antibiotics. Interestingly, -lactamase inhibitors have recently been shown to serve as substrates for the MexAB-OprM pump, which suggests that pump inactivation may enhance the efficacies of -lactam--lactamase inhibitor therapeutic combinations (139). The issue of outer membrane permeability has also been addressed through the coadministration of antibiotic with permeabilizing compounds such as cationic peptides (80, 181). These peptides interact with divalent cation binding sites present on LPS molecules and permeabilize the outer membrane, allowing enhanced antibiotic uptake during this process. In future, one or a combination of these approaches may facilitate improved management of P. aeruginosa infections.

LPS molecules are located in the cell wall and thus play an important structural role while mediating interaction with the neighboring environment. The tripartite nature of LPS divides the molecule into a hydrophobic lipid A region, which replaces phospholipids in the outer membrane, a central core oligosaccharide region, and a repeating polysaccharide portion referred to as O antigen or O polysaccharide. The terms "smooth" and "rough" are often used to describe the LPS phenotype. Attachment of the O antigen to core-lipid A results in a smooth LPS phenotype, while core-lipid A lacking O antigen is referred to as rough LPS. The contribution of smooth LPS to virulence has been demonstrated repeatedly by using various animal model systems. A study by Cryz et al. (40) showed that a wild-type strain of P. aeruginosa with smooth LPS was more virulent than were isogenic mutants. In a burned-mouse infection model, the mutant, which has rough LPS, has a 50% lethal dose more than 1,000-fold higher than that of the wild-type strain. This increase in the 50% lethal dose of a strain producing rough LPS demonstrates that O antigen is critical for virulence and that studies directed toward an understanding of LPS biosynthesis are essential. Recent studies by Preston et al. (185) with a mouse cornea infection model and Tang et al. (220) with a neonatal-mouse challenge model confirm that intact smooth LPS is required for P. aeruginosa virulence. In vitro experiments have shown that rough mutants of P. aeruginosa deficient in O-antigen synthesis are sensitive to the killing effects of human serum while wild-type strains with smooth LPS are serum resistant (44, 82). Since the focus of this review centers on LPS biosynthesis in P. aeruginosa, we address the role of lipid A, core oligosaccharide, and O antigen in the pathogenesis of this organism (see below).

When LPS is shed by bacteria into host tissues, it is usually bound by LPS binding protein (136, 222), which is transferred to the CD14 receptor (2395) on macrophages, thereby inducing secretion of cytokines including tumor necrosis factor alpha, interleukin-1 (IL-1), IL-6, IL-8, and IL-10 (150). These cytokines are known markers of inflammatory responses. The lipid A region of LPS, composed of a phosphorylated diglucosamine moiety substituted with fatty acids, is thought to be responsible for most of the biological activities of LPS (also referred to as endotoxin) through the induction of these immunomodulating molecules. Release of these inflammatory mediators enhances host defenses against bacterial infections. Excessive LPS stimulation of the immune system can occur, whereby elevated levels of activated and recruited immune cells results in septic shock and even death (150). P. aeruginosa is one of the top three pathogens responsible for sepsis due to gram-negative bacteria, and LPS from this organism is capable of overstimulating the immune system (62). Interestingly, P. aeruginosa lipid A is less toxic than that of enteric organisms, probably due to the presence of mostly pentaacyl rather than hexaacyl chains (125).

Attached to the lipid A is the core oligosaccharide region of LPS, which can be subdivided into an inner and an outer core. The inner core contains L-glycero-D-manno-heptose and 3-deoxy-D-manno-octulosonic acid (KDO), while the outer core is composed of hexose sugars such as D-glucose (D-Glc). The chemical structure of the P. aeruginosa serotype O5 core oligosaccharide is shown in Fig. 1C. The genetics involved in synthesis of this core region are currently being investigated by our laboratory but are not included in this review (52). The outer core region of P. aeruginosa LPS has recently been proposed to function as a bacterial ligand for association and entry of the organism into corneal cells during the infection stage of keratitis (243). In vitro studies have revealed the terminal D-Glc moiety of the outer core to be the critical residue in binding of the LPS molecule to mouse corneal epithelial cells (243). Recently, Pier et al. (179, 180) assessed the role of CFTR as a receptor for the binding of P. aeruginosa to host epithelial surfaces. In these studies, CFTR-minus epithelial cell lines showed poor ingestion and uptake of P. aeruginosa compared to cell lines transfected with CFTR. These authors have proposed that the presence of CFTR in normal hosts could be a defense mechanism whereby efficient epithelial-cell ingestion of P. aeruginosa, followed by cellular desquamation and swallowing, could facilitate bacterial killing by the digestive system of the host. In this model, the outer core region is believed to be the ligand mediating this association, since strains producing semirough LPS (core plus one O-antigen repeat) and rough LPS (lacking O antigen) with a complete core region were ingested more readily than were those expressing wild-type smooth LPS (180). These data provide evidence demonstrating that LPS variations can affect the uptake of P. aeruginosa by host epithelial cells. These new findings may help elucidate specific host-bacterium interactions and favor the development of therapeutics for chronic P. aeruginosa pulmonary infections in CF patients.

View larger version (40K): [in this window] [in a new window]   FIG. 1.   Structures of O-antigen and core oligosaccharide P. aeruginosa LPS. (A) A-band O-antigen structure of P. aeruginosa (7). (B) B-band O-antigen structures of P. aeruginosa serotypes O2, O5, O6, O10, O11, O15, O16, O17, O18, and O20 (109-111). The O2, O5, O16, O18, and O20 serotypes compose a cross-reactive serogroup due to similarities in their O-repeat units (130). (C) Core oligosaccharide structure of P. aeruginosa serotype O5 (199). LD-Hep, L-glycero-D-manno-heptose; Kdo, 2-keto-3-deoxy-D-manno-octulosonic acid; CONH2O-carbamoyl.

As mentioned above, attachment of the O antigen or O polysaccharide to core-lipid A results in the smooth form of LPS. This O-antigenic region is highly immunogenic and elicits a strong antibody response from the host. Resistance to serum is conferred by the presence of O antigen on the cell surface, and the extent of this serum resistance is influenced by O-antigen structure, chain length, and the amount of O antigen substituted on core-lipid A (reviewed in reference 97). P. aeruginosa produces two forms of O antigen, known as A band (homopolymer) and B band (heteropolymer). The A-band O-polysaccharide region is composed of D-rhamnose (D-Rha) residues arranged as trisaccharide repeating units linked 12, 13, 13 (Fig. 1A) (7). The A-band D-rhamnan polysaccharide is composed of approximately 70 D-Rha residues, which is equivalent to 23 repeating units (242). This is shorter than B-band O antigen (composed of 50 repeating units [129]), which is thought to mask underlying A-band polysaccharide, since A⁺B⁺ strains are not agglutinable with A-band-specific monoclonal antibodies (MAbs) (135). B-band O antigen is composed of di- to pentasaccharide repeating units of various monosaccharides. The composition of the B-band O-antigen trisaccharide repeat of serotype O5 (the serotype of the common laboratory strain PAO1) is di-N-acetylmanosaminuronic acid and N-acetyl-6-deoxygalactose (Fuc2NAc) (Fig. 1B) (109). Several studies have demonstrated that P. aeruginosa LPS confers serum resistance and elicits a protective immune response (40, 63, 82). Recently, a panel of P. aeruginosa LPS-deficient mutants was used to determine that B-band LPS confers serum resistance to the organism while A-band LPS is not protective against serum-mediated lysis (44). Studies have also indicated that while antibodies directed toward B-band LPS are highly protective of neutropenic mice upon challenge with smooth LPS strains, antibodies against A-band LPS are not protective (86).

P. aeruginosa is known to undergo differential expression of some of its virulence factors during infection. These phenotypic changes have for the most part been observed during pulmonary infections in CF patients (reviewed in reference 78). The most dramatic change is the conversion of the organism from a nonmucoid to a mucoid phenotype due to the production of copious amounts of the exopolysaccharide alginate (reviewed in reference 78). This cell surface alteration occurs once P. aeruginosa is well established within the lungs of CF patients and correlates with poor lung function (177). Accompanying the onset of alginate production is initiation of the microcolony mode of growth within the lungs of CF patients, representing a bacterial biofilm composed of cells embedded within an alginate matrix (128). With the emergence of mucoid P. aeruginosa within the lungs, there are cell surface changes with respect to LPS phenotype. Chronic P. aeruginosa isolates from CF patients either lack B-band O antigen entirely or express smaller amounts (70, 82, 131, 177, 182), while the level of A-band O polysaccharide is maintained (135). A study by Lam et al. (135) examined 250 P. aeruginosa isolates from the lungs of CF patients and determined that 68% expressed A-band LPS but none expressed B-band O antigen. Many of these clinical isolates are therefore nontypeable or polytypeable due to the lack of O-antigen polymer or deficiency in high-molecular-weight O-antigen polymer production (131, 177, 182). Such changes in surface properties are problematic for serotyping and epidemiological studies of P. aeruginosa isolates from CF patients (221). Evans et al. (64) demonstrated that 8 of 13 CF patient P. aeruginosa isolates expressing rough LPS could coproduce endogenous smooth LPS along with serotype O11 LPS when transformed with O-antigen genes (wbp, previously rfb) from P. aeruginosa serotype O11. Therefore, in some cases, these rough P. aeruginosa isolates from CF patients have acquired mutations in their wbp region. At present, the mechanisms responsible for these mutations and changes in LPS phenotype are not known.

It is interesting that A band is the LPS molecule selectively maintained on the P. aeruginosa cell surface during chronic CF lung infections. The presence of anti-A-band antibodies within CF patients correlates with both increased duration of P. aeruginosa infections and lower pulmonary function (135). Once P. aeruginosa has colonized the lungs, these LPS modifications are probably beneficial for evasion of host defenses (A band is less immunogenic) and for alteration of susceptibility to antibiotics, since loss of B-band O antigen confers resistance to aminoglycosides (17, 98). A recent study by Asboe et al. (8) examined P. aeruginosa isolates from human immunodeficiency virus-infected patients with respiratory infections. Multiple isolates from single patients infected with the same bacterial strain were shown to become serum sensitive and polyagglutinable/nontypeable. These characteristics, as mentioned above, are associated with a loss of or decrease in B-band O antigen. A-band LPS was not examined in that study. It would be interesting to determine if these respiratory isolates express A-band LPS and if the selection toward B-band deficiency follows a similar mechanism to that of P. aeruginosa pulmonary isolates from CF patients.

LPS heterogeneity can be achieved through variations in the sugar moieties within the O-antigen repeating unit, the type of stereochemistry ( or ) of the glycosyl linkages within the O-antigen repeat, the addition of noncarbohydrate moieties to the O antigen (i.e., O acetylation), and the proportion of smooth versus rough LPS molecules. Most strains of P. aeruginosa have a capping frequency (core-lipid A molecules substituted with long-chain O antigen) of between 0.2 and 14% (82, 191, 232). Although the regulatory mechanisms governing LPS synthesis and changes in LPS production in P. aeruginosa have not been determined, several studies have been performed to examine the influence of environmental parameters on LPS synthesis. Growth at elevated temperatures decreases B-band O-antigen chain length as the temperature increases (123, 165). Along with this decrease in the level of long-chain O antigen is an increase in the amount of the semirough (SR) B-band LPS (i.e., core-lipid A capped with one O-repeat unit). A study by Kropinski et al. (123) demonstrated that the proportion of SR LPS in P. aeruginosa PAO1 increases from 19.3 to 37.6% when the growth temperature shifts from 15 to 45°C. Interestingly, a recent study by Makin and Beveridge (153) revealed complete loss of B-band O antigen when PAO1 cells were grown at 45°C. The reason for the observed differences between these two studies is not known; however, the data does indicate a correlation between B-band O-antigen synthesis and temperature. The influence of various osmotic conditions on P. aeruginosa LPS synthesis has also been examined. Under conditions of high NaCl, MgCl₂, glycerol, and sucrose and low pH and phosphate, there is a decrease in the amount of long-chain B-band O antigen. Both elevated temperature and variation in nutrient levels resulted in modest increases in the chain length of A-band O polysaccharide (153, 165). Future work in this area should be directed at determining how these environmental conditions influence LPS synthesis at the molecular level for a fuller understanding of the regulatory networks controlling the expression of these cell surface molecules.

Biofilm formation has been intensely studied over the years due to its environmental and medical relevance. Bacterial biofilms can develop on solid substrata and generally consist of cells entwined in a protective matrix of extracellular polysaccharides. Establishment of a biofilm requires initial bacterial attachment to a solid surface to allow the growth and development of a mature biofilm. Attachment of P. aeruginosa to surfaces is known to increase transcription of alginate biosynthetic genes, thereby increasing alginate production, which enhances the growth of the biofilm. Transcriptional fusions with a lacZ reporter gene have demonstrated that the transcriptional activities of two alginate structural genes, algD and algC, increase in attached cells compared to planktonic cells (46, 47, 90). To date, regulatory components that are involved in sensing environmental conditions and controlling the transcription of alginate genes from the algC and algD promoters are reasonably well defined (reviewed in references 78 and 162).

The contribution of other polysaccharide molecules (i.e., LPS) to biofilm formation have been less well studied. However, the presence or absence of long-chain LPS polymers and the differences in the chemical nature of LPS molecules influence the physiochemical characteristics of the cell surface. In wild-type P. aeruginosa strains that are nonmucoid and devoid of alginate, the predominant surface polysaccharide is B-band LPS (129). The B-band O polymers are highly anionic and extend beyond the layers of A-band O polysaccharide and outer membrane proteins (98, 129, 154). Makin and Beveridge (154) conducted a study to examine the relationship between cell surface hydrophobicities and relative adhesive properties among P. aeruginosa strains with various LPS phenotypes. They found that A⁺B⁺ and AB⁺ cells possess the lowest surface hydrophobicity and the lowest surface charge (154). P. aeruginosa cells expressing these predominantly anionic LPS phenotypes (A⁺B⁺ and AB⁺) adhere to glass more efficiently, implying that electrostatic interactions among the B-band O-antigen polymers play a role in binding (154). In contrast, the ability of the bacteria to adhere to polystyrene was shown to correlate with the relative hydrophobicity of the bacterial cells in the descending order A⁺B > AB > A⁺B⁺ > AB⁺ (154). Therefore, changes in the production of either A-band or B-band LPS in P. aeruginosa influence the surface characteristics and probably modify the binding capabilities of these bacteria.

Beveridge et al. (12) recently examined the production of LPS by P. aeruginosa during biofilm formation in vitro and observed that as biofilms mature and thicken, the bacteria undergo changes from an A⁺B⁺ LPS phenotype to an A⁺B phenotype. This phenotypic change was reversible when cells were removed from the biofilm and allowed to grow in the planktonic mode. Interestingly, a longitudinal study examining P. aeruginosa serial isolates from a number of CF patients with chronic pulmonary infections over a period of several years also showed that the ability of the bacteria to produce B-band LPS was progressively lost over time (135). Thus, there is a need to investigate which form of LPS favors the establishment of biofilms by P. aeruginosa. Such a study was conducted by Flemming et al. (69), who examined a series of P. aeruginosa strains including wild-type serotype O6 (A⁺B⁺) and its rough mutant derivatives A28 (A⁺B, with a complete core oligosaccharide), R5 (AB) (although the A band was detected from whole-cell lysate of this strains by Western immunoblotting, no A-band LPS was produced on the surface of the bacteria due to a deficiency in the LPS core), and Gt700 (AB). The LPS structures and other properties of these strains have been characterized previously (44, 159, 160). On the basis of hydrophobic interaction chromatography and a salt aggregation test, the hydrophobic character of these strains was ranked as R5  A28 > Gt700 > O6. In addition, the anionic characteristics of cell surfaces were determined by electrostatic interaction chromatography and by zeta-potential measurements, ranking the strains as R5 > A28  Gt700 > O6. Both R5 and A28, which possess more strongly hydrophobic surfaces, demonstrated a significantly greater capacity to form biofilms on stainless steel and glass surfaces. Interestingly, both these adherent mutant strains also have more anionic surfaces based on the zeta-potential measurements (69).

Confocal microscopy was also used to examine the biofilms formed by each of these bacterial strains (134). Both A28 and R5 mutants form luxurious biofilms that spread out evenly on the substratum (glass surface), while the wild-type O6 strain develops a biofilm composed of "microcolonial" clusters of cells that appear to form clumps (Fig. 2). The formation of these clumped biofilms is probably due to specific electrostatic interactions among O-antigen polymers of the LPS molecules. These data imply that the properties of LPS on the surface of P. aeruginosa cells could affect the way in which a biofilm is formed on solid surfaces. Motility due to the presence of flagella generally appears to promote the initial attachment and recolonization in flow systems (50, 120). A recent study by O'Toole and Kolter (174) also demonstrated that flagellum-associated motility is essential for biofilm formation. However, Flemming et al. (69) found that the strongly adherent mutant, R5, lacked flagella and was nonmotile. The presence of flagella did not appear to influence either the hydrophobicity or cell surface anionic character, as seen in the similar rankings of R5 (with no flagella) and A28 (possessing flagella). From these studies, the observations that rough mutants, devoid of A-band and B-band O-antigen polymers, form a more stable biofilm are consistent with the situation for P. aeruginosa, which under certain environmental conditions (i.e., interactions within the CF host) is regulated to produce less high-molecular-weight LPS. At present, the mechanism of LPS regulation is not clearly defined.

View larger version (81K): [in this window] [in a new window]   FIG. 2.   The architecture of the biofilm formed by P. aeruginosa is influenced by changes in the LPS phenotype. Confocal micrographs of biofilms of fluorescein-stained P. aeruginosa strains grown in glass laminar-flow cells as described by Palmer and Caldwell (175) are shown. After the 9-h growth, the biofilms of the various bacterial strains were approximately 13 to 15 µm thick. (A through C) Optical sections (x-y plane) at 1-µm intervals through biofilms of wild-type strain O6 (A+B+), strain A28 (A+B, and contains complete core oligosaccharide), and strain R5 (AB, contains truncated core oligosaccharide, and A-band LPS is not produced on the surface but accumulated in the bacterial cells), respectively. The characteristics of LPS in the O6 strain and its isogenic mutants have been described by Dasgupta (45). Note the foci of microcolonial growth of wild-type O6 bacterial cells; this mode of growth on a substratum is maintained throughout the biofilm development. The biofilms formed by the mutants, A28 and R5, are evenly spread out, with the most confluent biofilm being formed by the most hydrophobic R5 strain. (D) Optical sections (x-z plane, cross sections) through biofilms of strains O6, A28, and R5 in ascending order of cell surface hydrophobicity (69). The differences in their surface hydrophobicity or hydrophilicity are also revealed by differential fluorescein staining characteristics: cells from both strain A28 and R5 exclude the stain (appeared dark [B and C]) whereas cells of O6 are positively stained (appeared bright [A]). The numbers on the micrographs indicate the distance of the optical sections from the substratum, and the scales of magnification are indicated by the bars. These confocal micrographs were kindly provided by R. Palmer, Center for Environmental Biotechnology, University of Tennessee, Knoxville, Tenn.

In a recent study (134), biofilm formation in the wild-type strain PAO1 (serotype O5; A⁺B⁺) and its derived wzm mutant (AB⁺; A-band polysaccharide transport mutation [196]) was observed by confocal microscopy. The wild-type PAO1 strain was found to form clumping biofilms similar to those of the O6 strain (data not shown). The AB⁺ wzm mutant formed biofilms similar to those of the rough mutants A28 and R5; i.e., the biofilms are evenly spread out on the substratum (reference 134 and data not shown). This nonmicrocolonial biofilm formation by the wzm mutant is unexpected, since the predominant B⁺ LPS phenotype should provide more hydrophilic charge to this mutant. Instead, the biofilm growth pattern was more widespread and resembled that of B mutants. Therefore, the physicochemistry with the presence of both A-band and B-band O antigens in wild-type strains appears to promote interactions among polymers of both LPS types. These interactions probably contribute to the cell clumping that leads to microcolony formation in the initial phase of biofilm generation.

GENETICS OF A-BAND O-POLYSACCHARIDE BIOSYNTHESIS

Top Previous Next References

Identification and Organization of A-Band O-Polysaccharide Genes

Several reports described the identification of a neutral polysaccharide composed of D-Rha produced by P. aeruginosa (116, 202, 242). However, it was the study performed by Rivera et al. (191) that distinguished A-band LPS in strain PAO1 as another form of LPS that is separable from the predominant B-band LPS by gel filtration chromatography. They examined fractionated LPS from PAO1 by Western immunoblotting techniques with A-band- and B-band-specific MAbs and by chemical analysis, revealing that this organism coproduces two antigenically and chemically distinct LPS molecules (191, 192). Subsequent immunochemistry studies have demonstrated that 14 of the 20 P. aeruginosa IATS (International Antigenic Typing Scheme) O serotypes produce A-band LPS (42, 135). Although the existence of A band in other gram-negative bacteria has not been studied, reports of LPS bearing the same O-polysaccharide structure as A band have emerged. Organisms expressing these D-rhamnan LPSs include P. syringae pv. morsprunorum C28 (211), P. syringae pv. cerasi 435 (225), B. cepacia (32), and Stenotrophomonas (Xanthomonas) maltophilia (235). Interestingly, the last two bacterial species are among pathogens associated with lung infections in CF patients.

Genes encoding proteins involved in synthesis of a D-rhamnan-containing LPS molecule have been described only for P. aeruginosa. The A-band O-polysaccharide gene cluster was isolated from a P. aeruginosa PAO1 (IATS O5) cosmid library on the basis of its ability to restore A-band LPS synthesis to the A-band deficient mutant rd7513 (142). This cosmid clone, designated pFV3 (containing a 27-kb insert [Fig. 3A]), complemented A-band LPS synthesis in five (O7, O13, O14, O15, and O16) of the six A-band-deficient IATS strains (O7, O12, O13, O14, O15, and O16), indicating that it encoded genes responsible for expression of this cell surface molecule (142). The DNA region responsible for complementation of A-band LPS in rd7513 was localized to a 1.6-kb KpnI fragment, pFV39, through the generation of subclones from pFV3 and the use of transposon mutagenesis (142). Identification of this region led us to focus our sequencing efforts on the 5' half of the cosmid clone pFV3. Sequence analysis of this 15-kb region revealed eight genes thought to be involved in A-band LPS synthesis (42, 195-197). These LPS genes have been named according to the nomenclature of Reeves et al. (189), whereby wb- represents genes involved in O-antigen biosynthesis and wz- denotes genes involved in O-antigen assembly. Following this system, the P. aeruginosa A-band O-polysaccharide-specific genes have been given the designation wbp_{A band}. The wbp_{A band} genes and the predicted functions of the encoded proteins are listed in Table 1. Lightfoot and Lam (143) used pulsed-field gel electrophoresis and Southern hybridization experiments to map the wbp_{A band} cluster to between 10.5 and 13.3 min on the 75-min map of PAO1 (Fig. 4).

View larger version (22K): [in this window] [in a new window]   FIG. 3.   Location of A-band and B-band O-antigen genes within their respective cosmid clones. (A) The wbpA band gene cluster isolated from strain PAO1 (serotype O5) is contained on cosmid pFV3. This cluster is composed of eight genes which are probably transcribed from a common 70-like promoter located upstream of rmd (197). Genes encoding proteins involved in A-band O-polysaccharide synthesis are shown in gray, while non-LPS genes are shown in white. (B) The wbpO5B band gene cluster is found on the cosmid pFV100. (C) wbpO6B band gene cluster. For the B-band clusters, genes encoding proteins involved in B-band O-polysaccharide synthesis for both O5 and O6 are shown in black while non-LPS genes are shown in white.

View larger version (19K): [in this window] [in a new window]   FIG. 4.   The genetic map (75 min) of strain PAO1 illustrates the approximate chromosomal locations of genes involved in synthesis of polysaccharide molecules in P. aeruginosa. Map positions are indicated for genes encoding proteins involved in the synthesis and regulation of alginate (reviewed in reference 78) and in the synthesis of the inner core region (waaFCGP); (52) and A-band and B-band O antigens of the LPS molecule (143). A newly identified locus (orf477 orf488 orf303), which may be involved in synthesis of a unique polysaccharide molecule in P. aeruginosa, is also shown (197). The map is not to scale.

These eight genes are arranged contiguously on pFV3 and are probably transcribed from a promoter found upstream of the first gene, rmd, in this cluster. Three potential promoter sequences have been located upstream of rmd, which correspond to the E. coli ⁷⁰ consensus sequence (197). However, two of these putative promoters are nonfunctional on the basis of protein expression experiments with an E. coli background (197). Chromosomal rmd mutants were generated through insertion of a gentamicin resistance (Gm^r) cassette within rmd. These mutants no longer express A-band LPS and could be complemented for A-band synthesis by using two subclones that represented the same insert DNA in either orientation (197). This suggested the presence of a functional promoter upstream of rmd; however, the precise transcriptional start site has not yet been determined.

The layout of the wbp_{A band} genes is such that the biosynthetic genes involved in synthesis of the A-band sugar D-Rha (rmd, gmd, and wbpW) are positioned at the beginning of the cluster, followed by those coding for an ATP binding cassette (ABC) transport system (wzm and wzt) and those involved in assembly of the D-rhamnan repeating unit (wbpX, wbpY, and wbpZ) (Fig. 3A). This genetic arrangement is similar to that of the O-antigen gene cluster of E. coli O9a, whereby manC and manB, encoding enzymes required for synthesis of the O9a sugar D-mannose (D-Man), reside at the 5' end of the cluster and are followed by wzm and wzt, coding for the ABC transport system (106); the 3' end of the cluster contains genes (wbdA, wbdB, and wbdC) coding for the D-mannosyltransferases that act to assemble the O9a O-repeating unit (105, 106). The wbd_EcO9a gene cluster contains one additional gene, wbdD, located between wzt and wbdA, that does not yet have an assigned function for its gene product. The organization of the Yersinia enterocolitica O:3 cluster is somewhat different from that of the wbp_{A band} and wbd_EcO9a clusters in that wzm and wzt are positioned among genes encoding the O:3 polysaccharide sugar 6-deoxy-L-altrose (244). The wbb gene arrangement of Klebsiella pneumoniae O1 and O8 and Serratia marcescens O16 is different from the above-mentioned clusters in that wzm and wzt are located at the extreme 5' ends of their respective gene clusters (35, 102, 218).

The conservation of the A-band O-polysaccharide genes has been determined within the 20 IATS reference strains of P. aeruginosa (42, 195, 196, 197). Southern hybridization analysis has established that all eight genes are present in the 20 IATS strains of P. aeruginosa. However, there are differences in the hybridization profiles of some strains that appear to be serotype specific. The majority of this variation is observed for IATS strains O12, O13, O15, and O16, which do not express A-band LPS (42).

The DNA region on pFV3 flanking the wbp_{A band} cluster has also been examined (195). The 1.5 kb of sequence 5' to the first A-band gene, rmd, showed no significant similarity to any sequences encoding proteins in the GenBank databases. However, the sequence downstream of the last gene in the cluster, wbpZ, revealed one complete open reading frame (ORF) which coded for a protein of 54 kDa. This ORF is located on the complementary DNA strand and, as shown in Fig. 3A, is transcribed in the opposite direction to that of the A-band gene cluster. A GenBank comparative search of the predicted amino acid sequence encoded by this ORF revealed high homology to coenzyme A transferase proteins from the gram-positive organisms Thermoanaerobacterium thermosaccharolyticum (52.9% identity; GenBank accession no. Z69031) and Clostridium kluyveri (51.3% identity; GenBank accession no. P38946). On the basis of this homology, we have assigned this ORF the designation psecoA. Downstream of psecoA, we identified the 3' end of a gene whose product displayed homology to the C terminus of a DNA helicase II enzyme (UvrD) from E. coli (66.3% identity; GenBank accession no. P03018). Like psecoA, this P. aeruginosa uvrD homologue is transcribed in the opposite direction to that of the wbp_{A band} cluster. Interestingly, uvrD is also located downstream of the enterobacterial common antigen (ECA) genes wecA to wecG (rfe to rff) at 85 min on the chromosome of E. coli K-12 (166). It is intriguing that another uvr gene homologue was identified during characterization of the B-band LPS gene cluster, wbp_{B band}, in strain PAO1. Burrows et al. (21) identified uvrB to be flanking wbp_{B band}, also at the 3' end. UvrB is known to function as part of a UvrABC complex that removes thymidine dimers (241). In E. coli, this complex is released from DNA after excision in the presence of UvrD, which is known to unwind DNA duplexes during nucleotide excision and mismatch repair (31, 124, 168). We are curious about whether uvrB and uvrD function as chromosomal markers in other P. aeruginosa serotypes for the B-band and A-band O-antigen biosynthetic gene clusters, respectively.

gmd and rmd. Sugar nucleotide precursors synthesized within the cell cytoplasm are used as donor molecules in the synthesis of cell surface polysaccharides (205). These sugar monomers can be transferred to a lipid carrier molecule, identified as undecaprenol phosphate (Und-P), at the cytoplasmic face of the inner membrane (237). The pathway we have proposed for synthesis of the sugar nucleotide precursor GDP-D-Rha for A-band O-polysaccharide synthesis is illustrated in Fig. 5. This pathway includes GDP-D-Man as an intermediate in GDP-D-Rha synthesis for two reasons. First, early work by Markovitz (156, 157) showed that a soil isolate, Pseudomonas strain GS (ATCC 19241), produced a capsular polysaccharide composed equally of the C-4 epimers D-Rha and D-talomethylose (D-Tal). Crude cell extracts from this bacterium were found to convert GDP-D-Man to GDP-D-Rha and GDP-D-Tal, via a GDP-4-keto-6-deoxy-Man intermediate (157). From this work, Markovitz identified two enzyme fractions from partial column purifications which possessed GDP-D-Man dehydratase and GDP-D-4-keto-6-deoxy-D-Man reductase activity, respectively. Second, Lightfoot and Lam (143) demonstrated that the protein coded for by the A-band gmd gene (formerly called gca) was involved in synthesis of GDP-D-Rha from GDP-D-Man. In that study, supernatants of whole-cell lysates from the A-band mutant rd7513 carrying the gmd gene on a plasmid (pFV39) were incubated with [¹⁴C]GDP-D-Man. Data from paper chromatography experiments revealed that rd7513(pFV39) was able to convert GDP-D-Man to a product which migrated at the same rate as the GDP-D-Rha standard. Based on this chromatography data and on the amino acid homology of Gmd to other dehydratase enzymes (42), we propose that Gmd functions as a GDP-D-Man dehydratase converting GDP-D-Man to GDP-4-keto-6-deoxy-D-Man (Fig. 5).

View larger version (15K): [in this window] [in a new window]   FIG. 5.   Proposed biosynthetic pathways for sugar nucleotide precursors used for synthesis of polysaccharide molecules within P. aeruginosa. GDP-D-Man is a common precursor of both D-mannuronic acid residues of alginate and the D-Rha residues of A-band O polysaccharide. Synthesis of GDP-D-Rha requires WbpW (GMP-PMI), Gmd (GDP-D-Man dehydratase), and Rmd (GDP-4-keto-6-deoxy-D-Man reductase) (42, 197). The D-mannuronic acid residues of alginate are formed through the activities of AlgA (GMP-PMI) (206), AlgC (PMM) (245), and AlgD (GDP-D-Man dehydrogenase) (54). The PMM activity of AlgC is required for GDP-D-Rha synthesis (197, 240), while the phophoglucomutase activity of AlgC is also needed for UDP-D-Glc and TDP-L-Rha for core biosynthesis (39). PGI, phosphoglucose isomerase; GalU, glucose-1-phosphate uridyltransferase; RmlA, glucose-1-phosphate thymidyltransferase; RmlB, dTDP-glucose-4,6-dehydratase; RmlC, dTDP-4-keto-L-rhamnose-3,5-epimerase; RmlD, dTDP-L-rhamnose synthetase.

algC and wbpW/algA/orf477. Synthesis of the D-mannuronic and L-guluronic acid residues of alginate also proceeds via a GDP-D-Man intermediate (54, 206, 245) (Fig. 5). For this reason, it has previously been suggested that common enzymes may be involved in the formation of the GDP-D-Man precursor for synthesis of both A band and alginate (140). In fact, an alginate enzyme, AlgC, is required for the synthesis of a number of cell surface molecules within P. aeruginosa. AlgC possesses both phosphomannose mutase (PMM) and phosphoglucose mutase (PGM) activities, which are essential for alginate and core oligosaccharide synthesis (39, 245). PMM is required for the conversion of mannose-6-phosphate to mannose-1-phosphate in the synthesis of GDP-D-Man (245) (Fig. 5), while PGM activity is required for conversion of glucose-6-phosphate to glucose-1-phosphate (39). This PGM activity was identified through complementation of the core-deficient mutant AK1012 (39). The cellular pool of glucose-1-phosphate is required for formation of the precursors UDP-D-glucose (D-Glc) and dTDP-L-rhamnose (L-Rha) (Fig. 5), which are presumably the donor sugar nucleotides for the D-Glc and L-Rha residues of the outer core region in P. aeruginosa (Fig. 1C). Synthesis of the O antigen itself may also be affected by an algC mutation in strains that contained D-Glc and/or L-Rha moieties in the O-repeat unit, such as serotype O6 (Fig. 1B).

wbpL. The mechanism of polysaccharide assembly for LPS and capsule synthesis begins through the activities of a glycosyltransferase and Und-P. These enzymes, termed initiating enzymes, contain numerous membrane-associated domains to allow for interactions with the hydrophobic Und-P acceptor. The best-characterized initiating enzymes are WecA_Ec (Rfe) and WbaP_{Se (Salmonella enterica)}, which transfer N-acetylglucosamine-1-phosphate (GlcNAc-1-P) (1, 167) or N-acetylgalactosamine-1-P (GalNAc-1-P) (4) and galactose-1-phosphate (Gal-1-P) (226), respectively, onto Und-P. In our laboratory, an enzyme designated WbpL, which exhibits high homology to glycosyltransferases responsible for initiating polysaccharide synthesis, has been identified within P. aeruginosa (21). The wbpL gene is located within the wbp_{B band} cluster; however, wbpL::Gm^r mutations affect both A-band and B-band LPS synthesis (195). No such initiating glycosyltransferase homologue was found with the wbp_{A band} cluster. Figure 6 shows that wbpL mutations abrogate both A-band and B-band synthesis in the cross-reactive serogroups O2, O5, and O16. Based on results from spectroscopic analysis, Fuc2NAc is the first sugar of the serotype O5 repeating unit attached to the core region (198). Since all three serotypes contain a Fuc2NAc moiety in their O-repeat unit (Fig. 1B), WbpL is believed to initiate B-band O-antigen synthesis through the addition of Fuc2NAc-1-P to Und-P (21). Heteropolymeric initiation and synthesis differ from the homopolymeric counterparts, since the former involve the initiating sugar within each O-repeating unit. In the case of B-band O-antigen synthesis, for example, WbpL would be responsible for the addition of Fuc2NAc in the formation of each trisaccharide repeat. In the homopolymers, the initiating "priming sugar" does not form part of the O repeat, since subsequent enzymes act to transfer additional sugar residues processively onto this initial sugar moiety. These differences reflect the cellular location of assembly components for heteropolymeric and homopolymeric O antigens. This is briefly discussed below but is reviewed in detail by Burrows et al. (28).

View larger version (46K): [in this window] [in a new window]   FIG. 6.   LPS analysis of WbpL mutants in the related serotypes O2, O5, and O16 (195). (A) Silver-stained SDS-PAGE gel; (B and C) Western immunoblots reacted with the A-band-specific MAb N1F10 and the B-band cross-reactive MAb 18-19, respectively. All of the wbpL mutants are deficient in both A-band and B-band LPS, with the parental phenotype being restored through complementation with a subclone containing wbpLO5 (pFV110). Serotype O2 reacts only with MAb 18-19 (C) as a core-plus-one-O-repeat unit, while serotypes O5 and O16 are able to react with the full B-band ladder. The substitution of a guluronic acid residue in the O2 O antigen (Fig. 1B) may be responsible for the formation of an epitope that is not recognized by 18-19 above core-plus-one. *, serotype O16 does not express A-band LPS. Reproduced from reference 195 with permission.

wbpX, wbpY, and wbpZ. Following the initiation event catalyzed by WbpL, homopolymeric biosynthesis proceeds by way of particular glycosyltransferases recognizing acceptor and donor molecules while catalyzing specific glycosidic linkages. The homopolymeric assembly process occurs on Und-P by using the sugar primer GlcNAc provided by the initiating glycosyltransferase at the cytoplasmic face of the inner membrane (230). The process of elongation occurs at the nonreducing terminus of the nascent polysaccharide (228). After assembly or concomitant with this process, the O polysaccharide is transported to the periplasm for attachment to core-lipid A. For the majority of homopolymers characterized thus far, this translocation process occurs via an ABC transport system (16, 106, 155, 218, 244).

View larger version (14K): [in this window] [in a new window]   FIG. 7.   Proposed glycosyltransferase mechanism for assembly of A-band (A) and serotype O5 B-band (B) O antigens (195). Note that WbpL participates in initiating both synthesis pathways with, probably, a GlcNAc residue for A band and a Fuc2NAc residue for B band. The specificities of these enzymes remain to proven biochemically. Modified from reference 195.

wzm and wzt. Currently, there are two known mechanisms of homopolymeric O-antigen assembly and export across the inner membrane. The more common of the two systems are the ABC transport systems reported for organisms such as E. coli O9a (106), K. pneumoniae O1 and O8 (35, 102), and S. marcescens O16 (218). These systems are composed of two proteins, a hydrophilic ATP-binding component and an integral membrane component, which associate to form paired homodimers (66). The second mechanism has been reported only for S. enterica serovar Borreze O:54 O-polysaccharide synthesis (100). This unique system contains a glycosyltransferase (WbbF) that is believed to couple polymerization and export functions by forming a pore-like structure at its C terminus (100). In either case, these export systems facilitate the translocation of O polysaccharide to the periplasm, allowing its ligation to separately synthesized core-lipid A molecules. Both of these systems proceed independently of an O-antigen polymerase, known as Wzy (discussed below), and are therefore commonly referred to as Wzy-independent pathways (230).