INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Escherichia coli is a major cause of urinary tract infections, septicemia, and neonatal meningitis. E. coli possesses several virulence factors, such as outer membrane proteins, capsules, and lipopolysaccharide (LPS), which allow the bacteria to evade the host defense system and promote infection. LPS and capsules are thought to provide resistance to the bactericidal effects of serum (10, 28-30, 46, 59, 61).

The complex carbohydrate LPS is a major constituent of the outer membrane of gram-negative bacteria. LPS serves as a barrier and shields the bacteria from detergents, free fatty acids, hydrophobic antibiotics, complement, and phagocytosis (40, 56, 63). LPS consists of three parts: lipid A (the toxic portion of the molecule), core oligosaccharide, and O antigen or O polysaccharide. The O antigen is made of sugar residue subunits linked to form a chain. The structure of the O-antigen subunit determines the serogroup of E. coli. The O75 subunit is composed of four sugars: N-acetylglucosamine, rhamnose, galactose, and the side sugar mannose (15).

Specific enzymes construct the oligosaccharide units, polymerize the units together, and ligate the growing O-antigen chain to the core oligosaccharide. Many of the enzymes necessary for the synthesis of the O antigen are encoded by genes in the rfb cluster. The rfb locus maps at 44 min on the E. coli chromosome and encodes enzymes necessary for the synthesis of sugar residues as well as proteins involved in the biosynthesis of the O polysaccharide (for a review, see reference 50). The rfbBDAC genes and the rfbKM genes encode the synthesis of TDP-rhamnose (26) and GDP-mannose (26, 63), respectively. Both of these sugar precursors are necessary to complete the O75 O-antigen subunit. The length of the O-antigen chain is regulated by a protein encoded by the rol (regulator of O-antigen length) gene (3). Normally, a wild-type strain has a bimodal distribution of O-antigen chains, where there exists a higher percentage of long chains (15 to 20 O-antigen lengths) than of short chains (1 to 5 O-antigen lengths). Mutation of the rol gene destroys the ability to regulate O-antigen length, producing a random length of O-antigen chains (3). Other genes, such as rfc and rfaL, are necessary to complete the assembly of the O antigen by polymerizing the O-antigen subunits and ligating the O antigen to the core, respectively (50).

In addition to LPS, many pathogenic bacteria also possess a capsule. The capsular antigens (K antigens) are acidic polysaccharides made of linear polymers of repeating carbohydrate subunits. The K5 antigen consists of repeating units of -glucuronic acid and N-acetylglucosamine and is identical to an intermediate in the synthesis of heparin (39). The K antigens can be divided into two groups (groups I and II) on the basis of biochemical and physical characteristics (25). The K5 antigen is a group II K antigen. The genes encoding the group II capsules have a common organization. Three gene clusters encode group II capsular antigen synthesis. Regions 1 and 3 are common to all group II capsular antigens, while region 2 is unique for each capsular antigen (6). K5 region 2, consisting of the kfiABCD genes, has been analyzed in the most detail (44).

Of the more than 160 different O antigens, only a small number are associated with disease (42). For example, serotypes O1, O2, O4, O6, O7, O18, and O75 are primarily isolated from urinary tract infections (41). The ability of bacteria to cause disease correlates with the sensitivity of the bacteria to serum. Bacteria able to resist the effects of complement are known as serum resistant; those that are not are considered sensitive to serum. E. coli strains lacking O antigens (rough) are more sensitive to serum (42, 58, 61), while strains which cause severe infections, such as bacteremia and urinary tract infections, have a high percentage of serum resistance (reviewed in reference 60). However, studies by Russo and colleagues using an O4 strain indicated that the loss of the O antigen has only minor effects on serum sensitivity (48).

Conflicting evidence has been presented as to whether capsular polysaccharides actually provide serum resistance. Of the more than 80 types of K antigens, only a few appear to be responsible for serum resistance (2, 9, 16, 34, 47, 48, 64). Evidence suggests that the capsule does not impart resistance by impeding complement components or antibody from binding to the bacterial surface because macromolecules can easily permeate the polysaccharide capsule (54). However, capsular antigens K1 and K54 seem to be protective against the bactericidal effects of serum (10, 34, 47, 64). K5, on the other hand, does not seem to impart serum resistance (8, 55). Thus, only certain O and K antigens may provide resistance to serum.

Many previous studies failed to utilize proven isogenic mutants of wild-type pathogenic strains. Genetically defined mutants are superior for the evaluation of putative virulence factors. We report five new chromosomally defined mutants of an otherwise wild-type O75 pathogenic strain to address the importance of the O75 and K5 antigens in protection from complement-mediated lysis. The genes chosen for mutation were the rfbD, rfbKM, rol, kfiC, and kfiC-rfbD genes, producing distinctly different phenotypes: rough (O75), core plus one partial O-antigen subunit, random distribution of O-antigen chain lengths, acapsular (K5), and O75 K5, respectively. The five mutants were analyzed by LPS analysis, sensitivity to K5-specific phage, outer membrane profiles, and serum assays and compared to the wild-type strain. All mutants were complemented in trans, showing restoration of the wild-type phenotype and serum resistance.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains, plasmids, and media. E. coli GR-12 was originally isolated from a patient with pyelonephritis (57). This strain is O75:K5 and is serum resistant. E. coli XL1-Blue, used for cloning with pBluescript II KS vectors, was obtained from Stratagene (La Jolla, Calif.). E. coli DH1 was used as a host with the pBR322 vectors. For cloning with the suicide vectors pGP704 and pGP704Km, SY327  pir (38) was used for initial cloning and SM10  pir (53) was used for conjugation. All strains are described in Table 1. Details of the construction of all plasmids used in this study are given in Table 2.

Mating procedure. GR-12 (or SMB20) and SM10 with specific plasmids were grown to the mid-log phase and combined in a 1:4 ratio of donors to recipients. The mixture was filtered, and the filter was placed on a fresh Luria-Bertani (LB) plate. The plate was incubated at 37°C for approximately 2 h. The bacteria on the filter were resuspended in 1 ml of BSG, and 50- to 200-µl samples were plated on minimal medium containing cysteine and the appropriate antibiotics. To optimize mating, a nonshaking culture of SM10 (with a plasmid) was mixed with a well-vortexed culture of GR-12 grown to the late stationary phase. After mating, the bacteria were gently resuspended in 1 ml of SOC (36) and incubated at 37°C for 45 min (11). Cells were washed twice with BSG, and samples were plated as before.

Construction of an rfbD mutation in GR-12. The chloramphenicol acetyltransferase gene (cat) obtained by gel purification from pUC18CML (51) digested with HindII and SmaI was inserted in the rfbD gene of plasmid pSMB102, a subclone of plasmid pRAB2401 (4). Isolates which had the cat gene in the same orientation as the rfbD gene were chosen (22) to avoid disruption of the transcription of upstream genes. The rfbD::cat construct from the new plasmid, pSMB103, was ligated to pGP704Km to make pSMB104. SM10 electroporated with pSMB104 was conjugated with GR-12. The vector pGP704 and its derivative pGP704Km cannot replicate in GR-12 and thus act as suicide vectors. Transconjugants were selected on minimal medium supplemented with cysteine and chloramphenicol to eliminate SM10 (unable to grow on minimal medium). Transconjugants were further screened for kanamycin sensitivity to distinguish single recombinants (in which the entire plasmid was integrated into the chromosome) from double recombinants.

Construction of an rfbKM mutation in GR-12. A subclone of pRAB2401, pSMB105, was recloned to eliminate the NruI sites, creating pSMB106. A portion of the sequence between the rfbK and rfbM genes was deleted, and a cat gene was inserted. As before, the orientation of the cat gene which corresponded to the predicted direction of transcription of the rfbKM genes was chosen (22). The new plasmid, pSMB107, was subcloned into pGP704. After ligation, the mixture was digested with KspI (found in the pBluescript II KS vector only) before electroporation into SY327 to eliminate self-ligation of pSMB107. The new plasmid, pSMB108, was transferred into SM10 and conjugated with GR-12. Transconjugants were selected as described above.

Construction of a rol deletion in GR-12. In order to delete the rol gene, upstream and downstream sequences were amplified with primers with engineered restriction sites so as to clone the products side by side, deleting the rol gene (0.82 kb of sequence deleted). Primers ugd F and ugd R and primers his F and his R (Table 3) were used to amplify by PCR the upstream ugd gene and the downstream his gene, respectively. The ugd PCR product was first cloned into pBluescript II KS (pSMB110), and then the his PCR product was cloned (pSMB111). The cat gene with BamHI ends was inserted between the two cloned genes, creating pSMB112. The ugd'-'his::cat fragment of DNA was cloned into pGP704. The new plasmid, pSMB114, was transferred into SM10 and conjugated with GR-12. Transconjugants were selected as described above.

Construction of a kfiC mutation in GR-12 and the rfbD mutant. The kfiCD gene from region 2 of the K5 kps capsular cluster was cloned by PCR amplification with the kfiC F and kfiD R primers (Table 3). The kfiCD PCR product was digested with SalI and ligated to pBluescript II KS, creating pSMB250. The cat gene was inserted into the kfiC gene in the same orientation, resulting in plasmid pSMB251-Cm, for construction of the K5 capsular mutant. The aphA gene from pBluescript II KS-Km was inserted into the kfiC gene in the same orientation, resulting in plasmid pSMB251-Km, for construction of the double mutant, since the rfbD mutant was already Cm^r. The new construct was removed from each plasmid (pSMB251-Cm and pSMB251-Km) and ligated to pGP704, making pSMB252-Cm and pSMB252-Km, respectively. The new plasmids were transferred into SM10. SM10 pSMB252-Cm was mated with GR-12 to create a K5 mutant. SM10 pSMB252-Km was mated with SMB20 (rfbD mutant) to create an O75 K5 double mutant.

PCR amplification. A small amount of a fresh colony was resuspended in 50 µl of PCR mixture (1× PCR buffer, 2 mM MgCl₂, 0.5 U of Amplitaq [Perkin-Elmer Cetus, Norwalk, Conn.], 200 µM each deoxynucleoside triphosphate [Boehringer Mannheim Biochemicals, Indianapolis, Ind.]). A DNA concentration of 1 ng was used for plasmid amplification. The mixture was sealed with approximately 60 µl of paraffin oil. The PCR consisted of 27 cycles: 1 min of denaturation at 94°C, 2 min of annealing at 50°C, and 3 min of extension at 72°C. The PCR concluded with 7 min at 72°C and 10 min at 4°C. A modified PCR was used for the kfiC F and kfiC R primers by increasing the annealing temperature to 60°C. In order to amplify products larger than 4.5 kb, it was necessary to use Taq Extender PCR Buffer and Taq Extender PCR Additive from Stratagene. These reactions were done with a 5-min extension time.

DNA manipulations. In order to purify a small amount of plasmid to screen isolates, the cleared-lysate method of preparation was used as described by Maniatis et al. (36). Larger-scale preparations of plasmids for cloning and storage were made with Qiagen tips (Qiagen, Inc., Chatsworth, Calif.). Chromosomal DNA preparations were made with 0.5-ml overnight cultures in accordance with the protocol of Puregene, Research Triangle Park, N.C. Digestion with restriction endonucleases, ligations, transformations, electrophoresis on agarose gels, and DNA blotting by capillary action were done as described by Maniatis et al. (36). T4 DNA ligase was used for ligations (New England BioLabs, Beverly, Mass.). Restriction endonucleases used for cloning procedures were from New England BioLabs and Boehringer. DNA fragments were cut from high-strength analytical-grade ultra-pure DNA-grade agarose (Bio-Rad, Hercules, Calif.) and purified with Jetsorb according to the manufacturer's instructions (Genomed, Research Triangle Park, N.C.).

LPS analysis. LPS was isolated from a 1-cm square patch of bacteria on an agar plate by the procedure of Hitchcock and Brown (21). The preparations were separated by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) on a 6% stacking-15% separating gel as described by Laemmli (32) with Tris-glycine buffer for GR-12 and mutant strains. For further analysis of the rfbKM mutant, Tricine-SDS buffer was used (49). The samples were electrophoresed at a constant 30 A on a minigel apparatus (Mighty Small II SE 250; Hoefer Scientific, San Francisco, Calif.). The gel was fixed overnight in a solution of 5% acetic acid and 40% ethanol. The gel was silver stained by the procedure of Tsai and Frasch (62).

Sensitivity to K5-specific bacteriophage. GR-12, mutants, and complemented mutants were tested for sensitivity to K5-specific bacteriophage (G. Schmidt, Borstel, Germany) by cross-streaking of bacteria against a phage suspension on Luria agar.

Outer membrane protein analysis. Outer membrane proteins were isolated by detergent solubilization as described previously, except that cells were broken by passage through a French press (1). After pelleting of insoluble outer membrane proteins, proteins were washed in saline, centrifuged in Sorvall (SS34; 20,000 rpm, 90 min, 4°C), and resuspended in 25 µl of 2× Laemmli buffer without dye. Proteins were separated by SDS-PAGE on a 4% stacking-12% separating gel at 100 V through the stacking gel and 200 to 250 V through the separating gel on a minigel apparatus (Mighty Small II SE 250). A low-molecular-weight protein ladder was used as a standard (Pharmacia Fine Chemicals, Piscataway, N.J.). Proteins were visualized by staining with 0.25% Coomassie brilliant blue R-250 for 20 min and then were destained as described previously (36).

Serum assays. Blood was collected from five healthy individuals who gave informed consent and then was allowed to clot. Serum was collected, pooled, and stored at 70°C in aliquots. Serum assays were performed with normal human serum, serum plus EGTA, and heat-inactivated serum. BSG with 100 mM EGTA and MgCl₂ was added to serum to chelate calcium, inactivating the classical pathway of complement. Heated serum was serum heated to 56°C for 30 min to inactivate complement components. Assays were performed by a previously described procedure with slight modifications (60). The concentration of serum in the assays was 80%. Bacteria were inoculated into a culture of LB medium from a fresh LB plate and grown to the mid-log phase (optical density at 600 nm, 0.5). Antibiotics were added for the growth of complemented mutants. Cells were added directly or were diluted twofold in saline and then added to normal human serum, serum plus EGTA, or heat-inactivated serum to obtain approximately 3 × 10⁷ or 3 × 10⁵ CFU per ml, respectively, in a serum concentration of 80%. BSG or BSG plus EGTA was used as a diluent for the serum reactions and viable counts. The reaction mixture was incubated at 37°C for 3 h. Viable counts were made by taking samples at 0, 1, 2, and 3 h. At least two samples were taken at each time point. Dilutions from samples were plated onto LB plates and incubated overnight at 37°C, and counts were determined. Percentages for each time point were obtained from the surviving viable count versus the initial viable count. Each strain was assayed at least two different times in 80% serum and 80% heat-inactivated serum.

Complementation of mutants. Single mutants were complemented with the wild-type gene on a plasmid. The plasmids were electroporated into the mutants, the complemented mutants were selected by resistance to tetracycline, and the presence of the plasmids was confirmed. SMB20 was complemented with the rfbD gene on plasmid pGH1610 (Table 2). This plasmid contains the 3' end of the galF gene and the entire rfbB and rfbD genes, as well as the 5' end of the rfbA gene, inserted into the bla gene of pBR322. SMB122 was complemented with the rol gene on plasmid pRAB10 (3). For complementation of SMB43, the rfbKM genes were cloned into plasmid pBR322, creating pSMB109 (Table 2), and electroporated into SMB43. The kfiCD genes were removed from pSMB250 and cloned into pBR322, creating pSMB255 (Table 2). pSMB255 was electroporated into SMB213. The vector plasmid in each case was also electroporated into the mutants to confirm that the vector was not responsible for serum resistance.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

In order to study the importance of the O75 O antigen and the K5 capsular antigen in conferring resistance to serum, chromosomal mutations in LPS synthesis and K5 synthesis genes were made in serum-resistant, pathogenic strain GR-12. Several genes determined to be important in LPS synthesis on the basis of preliminary O75 sequence analysis were mutated: the rol gene, the putative rfbD gene, and the putative rfbKM genes (5, 22). The kfiC gene, required for the synthesis of the K5 capsule, was chosen for mutation to make a K5 mutant.

Rationale for mutations in GR-12. We sought to evaluate and compare the effects of LPS and capsular mutations on serum resistance. A rough mutant, having no O polysaccharide, and a capsule-deficient mutant should establish the role of the O75 and K5 antigens in resistance to complement-mediated lysis. In order to create a rough mutant, the rfbD gene was targeted for mutation. The putative rfbD gene encodes the synthesis of TDP-rhamnose, which is necessary to make the O75 subunit. A diagram of the O75 subunit is shown in Fig. 1. The rfbD gene was inactivated because the sugar rhamnose is unique to the O75 subunit and is not required for the synthesis of other polymers, such as the core oligosaccharide, lipid A, and peptidoglycan (50). Without TDP-rhamnose, the mutant strain is predicted to lack O-polysaccharide units attached to the core. In order to create an acapsular mutant, the kfiC gene, encoding a putative glycosyltransferase necessary for the synthesis of the K5 capsular antigen, was targeted (45, 52). A double mutant was also made with mutations in the kfiC and rfbD genes to determine the effect on serum resistance of the loss of both the O75 and the K5 antigens.

View larger version (5K): [in this window] [in a new window]   FIG. 1.   Structure of the O75 subunit (15). The O75 subunit is made of four sugars: GlcNAc, N-acetylglucosamine; Rha, rhamnose; Gal, galactose; and Man, mannose (side-chain sugar).

Construction of mutations in GR-12. A brief description of the construction of the mutants follows; details are given in Materials and Methods and Tables 1, 2, and 3. The kfiC, rfbD, and rfbKM genes, to be mutated by insertion of an antibiotic resistance gene, were initially subcloned onto a plasmid. For the kfiC and rfbKM genes, a portion of the gene was deleted. A cat gene was inserted in the same orientation as the wild-type gene to disrupt the gene(s) and allow for the selection of mutants. The aphA gene was used for creation of the kfiC mutation in the double mutant (kfiC-rfbD mutant). The rol gene was completely deleted by insertion of a cat gene between the upstream and downstream sequences of the rol gene. The mutated gene in each case was further subcloned into the suicide vector (pGP704) or a derivative of the suicide vector. The suicide vector was placed into a donor strain, SM10, and conjugated with GR-12 or SMB20 (for the double mutant) (Fig. 2). Transconjugants were selected on minimal medium with appropriate antibiotics and confirmed by PCR (except for the rfbKM mutant) and Southern analysis. The mutants were designated as follows: SMB20 (rfbD mutant), SMB43 (rfbKM mutant), SMB122 (rol deletion mutant), SMB213 (kfiC mutant), and SMB316 (kfiC-rfbD mutant).

View larger version (18K): [in this window] [in a new window]   FIG. 2.   Diagram of construction of chromosomal mutants. Genes deleted or mutated are indicated. Only relevant restriction sites are included: A, SalI; C, ClaI; D, NdeI; E, EcoRV; H, HpaI; K, KpnI; M, MunI; N, NruI; P, PstI; and S, SnaBI. Plasmids are described in Table 2. Crossed lines denote recombination.

Phenotypic analyses of GR-12 and mutants. The LPS phenotype of a wild-type strain, such as GR-12, is characterized by a bimodal distribution of O-antigen subunits visualized on a polyacrylamide gel (23, 43). Strains possessing a smooth LPS phenotype have long O-antigen chains; a rough strain has only core oligosaccharide. A semirough strain has also been described; the LPS is characterized by the substitution of only one O-antigen subunit on the core oligosaccharide. An example is an rfc mutant (35) in which the loss of O polymerase function does not allow polymerization of the O-antigen subunits.

View larger version (58K): [in this window] [in a new window]   FIG. 3.   Visualization of LPSs of GR-12, mutants, and complemented mutants. (A) SDS-PAGE. Lanes: 1, GR-12 (wild type); 2, SMB213 (kfiC mutant); 3, SMB122 (rol mutant); 4, SMB122(pRAB10); 5, SMB43 (rfbKM mutant); 6, SMB43(pSMB109); 7, SMB20 (rfbD mutant); 8, SMB20(pGH1610); 9, SMB316 (kfiC-rfbD double mutant). (B) Tricine-SDS buffer system. Lanes: 10 and 13, GR-12; 11, SMB20 (rfbD mutant); 12, SMB43 (rfbKM mutant).

View larger version (128K): [in this window] [in a new window]   FIG. 4.   Analysis of outer membrane protein profiles of GR-12 and mutants. A 5-µl sample of each strain was loaded in each lane. Lanes: 1, protein standard (from top: molecular weight of 94,000 [94K], 67K, 43K, 30K, and 20.1K); 2, GR-12 (wild type); 3, SMB122 (rol mutant); 4, SMB43 (rfbKM mutant); 5, SMB20 (rfbD mutant); 6, SMB213 (kfiC mutant); 7, SMB316 (kfiC-rfbD mutant).

Serum assays of GR-12 and mutants. GR-12 is serum resistant. The sensitivity of the mutants in serum can be compared to that of GR-12. Since isogenic mutants were constructed, any loss of resistance could be attributed to mutation of a particular gene.

View larger version (14K): [in this window] [in a new window]   FIG. 5.   Serum assays of wild-type strain GR-12 and mutants in 80% serum (A) or 80% heat-inactivated serum (B) with an inoculum of 3 × 105 CFU/ml. The assays were performed for 3 h; samples were taken in duplicate at the start and at each hour. The graphs are a result of at least two different assays for each strain. Results are shown as percent viability. Error bars indicate standard deviations.

View larger version (23K): [in this window] [in a new window]   FIG. 6.   Serum assays (80% serum) of GR-12 and O75, K5, and O75 K5 mutants at an inoculum of 3 × 107 CFU/ml. The assays were performed for 3 h. Samples were taken in duplicate at the start and at each hour. The graph is a result of at least two different assays for each strain. Results are shown as percent viability. Error bars indicate standard deviations.

View larger version (22K): [in this window] [in a new window]   FIG. 7.   Serum assays (80% serum) with 10 mM EGTA plus MgCl2 of GR-12 and O75, K5, and O75 K5 mutants at an inoculum of 3 × 105 CFU/ml. The assays were performed for 3 h. Samples were taken in duplicate at time zero and at each hour. A control assay was performed with 80% heat-inactivated serum plus EGTA and the double mutant (SMB316). The graph is a result of at least two different assays for each strain. Results are shown as percent viability. Error bars indicate standard deviations.

Complementation of mutants. In order to confirm that the loss of function of the mutated gene caused the failure to produce a wild-type phenotype and the loss of serum resistance, the wild-type gene was transferred to the mutants in trans on a plasmid. The wild-type gene on a plasmid was introduced into the mutants via electroporation, and complemented mutants were selected on the basis of resistance to tetracycline. Complementation of mutants is described in Materials and Methods. All single mutants were also electroporated with the cloning vector alone; i.e., SMB122 was electroporated with pACYC184 and SMB20, SMB43, and SMB213 were electroporated with pBR322 to confirm that the vector plasmid itself was not complementing the capacity of the mutants to resist the bactericidal effect of serum.

View larger version (24K): [in this window] [in a new window]   FIG. 8.   Serum assays (80% serum) of GR-12 and each complemented mutant at an inoculum of 3 × 105 CFU/ml. The assays were performed for 3 h. Samples were taken in duplicate at time zero and at each hour. The graph is a result of at least two different assays for each strain. Results are shown as percent viability. Error bars indicate standard deviations.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Many previous studies, in which a role in serum resistance was assigned to LPS or capsules, used mutants in which the exact genetic lesion was unknown. Furthermore, the potential for secondary mutations in these strains was not always investigated. Therefore, although correlations were made as to the role of LPS and capsules in serum resistance, the evidence was not conclusive. Through creation of a defined mutant strain from an otherwise wild-type pathogenic strain, it is possible to definitively point to the mutation as the cause for the loss or gain of a specific attribute.

GR-12 and the five mutants were analyzed phenotypically for the K5 capsule and LPS. K5-specific phage sensitivity testing confirmed that GR-12 and LPS single mutants were sensitive to K5-specific phage, while SMB213 (kfiC mutant) and SMB316 (double mutant) were resistant. GR-12 had the typical bimodal distribution of LPS; SMB122 (rol mutant) had a random distribution of LPS, as expected; and SMB20 (rfbD mutant) was rough, with only the core being present. We predicted that the rfbKM mutant would have one of several phenotypes: rough, core plus one partial O-antigen subunit, or O-antigen chains with no side-chain sugars present. Previous studies with E. coli reported the identification of an O4 strain which had O-antigen chains without the side-chain sugars normally present (24). The rfbKM mutant (SMB43) appeared rough on SDS-PAGE but, after analysis with a Tricine-SDS buffer system, SMB43 was discovered to have the core plus one partial O-antigen subunit. The partial O-antigen subunit, consisting of N-acetylglucosamine, rhamnose, and galactose, was ligated to the core without the side-chain sugar mannose, but polymerization of the partial O-antigen subunit was not apparent. Thus, the side-chain sugar mannose is not necessary for ligation of the O-antigen subunit to the core but is required for polymerization of the O-antigen subunits.

Analyzing the mutants with serum assays proved that the O75 antigen is more important than the K5 antigen in serum resistance. Rough mutant SMB20 was completely killed in serum by the second hour of incubation. In comparison, K5 mutant SMB213 increased in numbers in the first hour and then slowly decreased in viability by the third hour of incubation in serum. A complete loss of the O75 antigen made the bacterium sensitive to serum. This work is in contrast to work done by Russo et al (48). They made an isogenic mutant lacking O antigen in an O4 pathogenic strain. Contrary to the expected results, the O4 antigen played only a minor role in serum resistance in comparison to the K54 antigen. However, our work did agree with that of Russo et al. in that the loss of both O and K antigens was correlated with a synergistic effect on serum sensitivity (48). The O75 K5 mutant was undetectable by the end of the first hour of incubation at both inoculum sizes tested, while the O75 mutant was not as sensitive to serum when the inoculum size was increased. The K5 mutant, in contrast, was resistant to serum at the larger inoculum size.

Complement may be activated by either the classical or the alternative pathway. The classical pathway can be blocked by the addition of a calcium chelator, such as EGTA. GR-12 and all mutants, except for the double mutant, were resistant to serum when the classical pathway was blocked. The alternative pathway was not sufficient to kill the rough mutant (O75 K5⁺ mutant) but was able to kill the double mutant over a 3-h incubation. Thus, the O75 antigen is important in resistance to the classical pathway, and the K5 antigen may provide resistance to the alternative pathway, as indicated by the difference in sensitivity between the O75 K5⁺ mutant and the O75 K5 mutant in serum plus EGTA. These findings agree with studies done by Devine and Roberts (14) in which O75 strains were affected by the classical complement pathway and K5 strains showed a delayed sensitivity to serum and were resistant to the alternative complement pathway.

The rfbKM mutant was also very sensitive to serum. Although the serum sensitivity was similar to that of the rough mutant in the first hour, the rfbKM mutant was not completely killed by the third hour of incubation. Substitution of the partial O-antigen subunit on the core oligosaccharide must have given the rfbKM mutant some protection against serum complement components in the second and third hours of incubation in serum.

Several studies have indicated the importance of long-chain LPS (18, 27). Smooth strains of Salmonella and E. coli have been shown to interact with terminal components of the complement cascade, yet insertion of the membrane attack complex never occurs (30). The membrane attack complex, formed by the activation of long O-antigen side chains at points distal to the vulnerable lipid layer, is sterically hindered from insertion into the outer membranes of smooth strains (27, 28). Although smooth strains are less susceptible to serum, the degree of susceptibility also depends on the coverage of LPS (59). Grossman et al. (17) showed that an increase in LPS coverage correlated with an increase in survival in serum. The present study demonstrated that, in the case of the O75 antigen, a bimodal distribution of O75 O antigens is crucial in protecting the bacteria from complement-mediated lysis. The rol deletion mutant could still produce long-chain LPS but in fewer full-length chains than the wild-type strain. Loss of the majority of long O-antigen chains made the rol deletion mutant sensitive to serum but to a lesser degree than the rough mutant. The rol deletion mutant exhibited almost 2 log units of death in the first hour and continued to decrease in numbers in the second and third hours of incubation, while GR-12 increased in numbers. These studies showed the importance of regulation of the length of the O75 O antigens for survival in serum.

Through restoration of the wild-type phenotype with the wild-type gene in trans, the mutations causing the defect in the synthesis of LPS were demonstrated to be responsible for the loss of serum resistance. All complemented mutants regained a wild-type bimodal distribution of LPS or K5-specific phage sensitivity and serum resistance.

When compared to the studies of Russo et al. (48), our results established the fact that the roles of the O and K antigens may be different in different strains. Our finding was that the O75 antigen plays a more important role in serum resistance than does the K5 antigen. However, Russo et al. found these roles to be reversed in an O4:K54 strain (48). Although one antigen may not be critical in serum resistance, it may play a more important role in another aspect of virulence, such as resistance to phagocytosis or to phagocytic killing. Thus, the antigens may have a complementary relationship in virulenceone antigen providing resistance to one aspect of the host defense system and the other antigen providing resistance to another part of the host defense system. Testing of other isogenic mutants of different O:K pathogenic strains is necessary to prove this point.

This is the first study in which defined LPS and capsular mutants were created for a pathogenic strain, tested in serum resistance assays, and complemented, showing restoration of the wild-type phenotype along with serum resistance. The mutants from this study will allow us to further understand the capacity of the O75 and K5 antigens in protection against the host defense system and to address important aspects of LPS synthesis.