Loss of Resistance to Ingestion and Phagocytic Killing by O⁻ and K⁻ Mutants of a Uropathogenic Escherichia coli O75:K5 Strain

Abstract

To determine the importance of the O75 O antigen and the K5 capsular antigen in resistance to phagocytosis and phagocytic killing, we used previously described O75⁻ and K5⁻ mutants from an O75⁺ K5⁺ wild-type uropathogenic Escherichia coli strain in phagocytosis assays with polymorphonuclear leukocytes (PMNs) and monocytes. At a 10-to-1 ratio of bacteria to phagocytes and in the presence of 10% serum, the parental strain GR-12 was resistant to both PMNs and monocytes over a 2-h incubation period. The O75⁻ and K5⁻ mutants were similar in sensitivity to killing by both PMNs and monocytes, decreasing in viability by 80% in the first hour. Yet, a significant difference in killing between the O75⁻ and K5⁻ mutants was observed in the first 15 min of incubation. The K5⁻ mutant decreased in numbers by almost 60%, while the O75⁻ mutant increased in numbers similarly to GR-12 in the first 15 min. The difference in killing was found not to be due to the rate of opsonization. To further determine the mechanism of resistance, a fluorescence assay was used to differentiate attached and internalized bacteria. The K5 capsule hindered the association of both the wild-type strain and the O75⁻ mutant in the initial incubation time with PMNs. In conclusion, both the K5 capsule and O75 O antigen play crucial roles in resistance to phagocytosis over time.

Escherichia coli strains causing urinary tract infections, septicemia, and neonatal meningitis typically belong to a restricted set of O and K antigen serogroups. The O antigen is part of the complex carbohydrate lipopolysaccharide (LPS). LPS also consists of lipid A (toxic portion of the molecule) and core oligosaccharide. In addition to LPS, many pathogenic bacteria have an outermost acidic polysaccharide or capsular antigen (K antigen). The capsule is made of linear polymers of repeating carbohydrate subunits. Of the more than 160 different O antigens and 80 different K antigens, only a few are associated with disease (1, 18, 20). These major outer membrane antigens of pathogenic strains are thought to provide resistance to the host defense system, promoting colonization and infection.

Complement-mediated lysis and phagocytosis are the first line of defense against invading microorganisms. The bacteriolytic activity of serum usually kills most gram-negative bacteria, but those that are resistant may then be susceptible to ingestion and killing by phagocytic cells (15). Polymorphonuclear leukocytes (PMNs) and monocytes are the major phagocytic cells that phagocytize and lyse bacteria. K antigens and in some cases O antigens are thought to provide resistance to phagocytosis (9, 17, 25, 33). The capsule may provide protection to the organism by masking the underlying opsonized surface and serving as a physical barrier to the phagocytic cell (5, 10, 17, 29).

Previously, we investigated the role of the O75 O antigen and the K5 antigen in resistance to complement-mediated lysis by using proven isogenic mutants in serum resistance assays. The use of genetically defined mutants is critical for the evaluation of the precise role of putative virulence factors. We constructed an O75⁻ mutant and a K5⁻ mutant of the uropathogenic strain GR-12. The O75 O antigen was demonstrated to be more important than the K5 antigen in serum resistance. The K5 capsule antigen, in contrast, played only a minor role in resistance to serum (4). Although the K5 capsule was not crucial in serum resistance, it may play a role in protection from phagocytic activity.

The K1 antigen has been speculated to be crucial in resistance to phagocytosis, a possibility supported by results of several studies (2, 6, 11, 19, 34). However, comparison of K5⁺ and K1⁺ strains in phagocytosis assays suggested that the K5 antigen did not have antiphagocytic properties (6). Certain O antigens are also speculated to provide resistance to killing (6, 16, 24). To determine whether the O75 O antigen and the K5 capsular antigen have a significant role in resistance to phagocytosis and to phagocytic killing, we analyzed the O75⁻ and K5⁻ mutants along with GR-12 by several different phagocytosis assays not only to determine the importance of the antigens in providing resistance to phagocytosis but also to differentiate the step at which resistance was provided by the O and K antigens.

MATERIALS AND METHODS

Bacterial strains, media, and reagents.

The O75:K5 E. coli strain GR-12 was originally isolated from a patient with pyelonephritis (28). SMB20 (O⁻ mutant) and SMB213 (K⁻ mutant) are defined chromosomal mutants derived from GR-12 and were described previously (4).

Luria broth and Luria agar (1.6% [wt/vol] agar; Difco Laboratories, Detroit, Mich.) were used for standard growth of strains. Dulbecco’s phosphate-buffered saline (DPBS) was used in 1× and 10× concentrations from Gibco BRL (Grand Island, N.Y.). Sterile deionized, distilled water (ddH₂O; Gibco BRL) was used for lysis of erythrocytes and as a diluent. Water and DPBS had endotoxin levels of less than 0.25 endotoxin units/ml. Fluorescein isothiocyanate isomer 1 (FITC) was made at a concentration of 0.15 mg/ml in 1× DPBS. FITC and ethidium bromide were obtained from Sigma (St. Louis, Mo.).

Isolation of PMNs and monocytes.

Human serum was collected from five healthy donors with no recent history of antibiotic therapy as previously described by Burns and Hull (4). Blood for separation of PMNs and monocytes was donated from an individual into heparin (143 USP units of sodium heparin) Vacutainer tubes (Becton Dickinson, Franklin Lakes, N.J.) in 20- to 40-ml volumes. A discontinuous density gradient was made by placing 1 ml of a 5:3 ratio of Mono-Poly resolving medium (M-PRM; ICN Biomedicals, Aurora, Ohio) and ddH₂O onto 3 ml of M-PRM in a 15-ml polypropylene conical tube (Corning Inc., Corning, N.Y.). A 4-ml volume of heparin-treated blood was layered on top of the density gradient and separated by spinning in a tabletop centrifuge (Centra GP8R; International Equipment Company, Needham Heights, Mass.) at 750 × g for 50 min; the centrifuge was allowed to stop without the brake. For isolation of PMNs, 16 ml of blood was separated. When monocytes were isolated, 40 ml of blood was separated.

After separation, the plasma layer was discarded and the appropriate fraction either mononuclear or PMNs as described in the M-PRM manufacturer’s directions, was removed and washed with DPBS. The washed cells were pelleted at 250 × g for 10 min. The buffer was discarded, and the cells were resuspended in the remaining buffer. The erythrocytes were lysed by adding cold ddH₂O; after 20 s, the isotonicity was restored by adding 2× DPBS. The cells were pelleted at 130 × g for 8 min. The lysis step was repeated if necessary. Cells were resuspended in 1/10 the volume of blood originally collected. The viability and concentration of the cells were determined by trypan blue exclusion. In each experiment, at least 98% of the PMNs and 95% of the monocytes were viable. The purity of the preparation was determined by using 0.1 N HCl. The PMN fraction was an average of 96% PMNs. The mononuclear fraction contained on average of 26% monocytes, with only 1% contaminating PMNs.

Phagocytosis killing assays.

The procedure for the phagocytosis killing assay was adapted from that of Rest and Speert (22). Bacteria were grown to the exponential phase and counted in a hemocytometer. For certain experiments, bacteria were preopsonized by incubation in 10% human serum for 30 min at 37°C. Reactions consisted of a 10:1 ratio of bacteria to phagocytes, 10% serum or 10% heat-inactivated serum as a complement source, 1 mM MgCl₂ and CaCl₂, and DPBS as a diluent. A concentration of 4 × 10⁶ PMNs/ml or 1.5 × 10⁶ to 2.0 × 10⁶ monocytes/ml was used in the assays as appropriate. The reactions were incubated in Falcon polypropylene snap cap tubes (Becton Dickinson) at 37°C, with rotation, for 2 h. Samples in duplicate were removed at 0, 15, 30, 60, and 120 min and diluted in ddH₂O to lyse the phagocytes releasing internalized bacteria. To separate PMNs from unattached bacteria, additional samples were taken, placed on ice, and spun in a low-speed HF-120 capsule centrifuge (Tomy Seiko Co., Tokyo, Japan) for 10 s to pellet the PMNs. The supernatant was subsequently removed and diluted. Samples from dilutions were plated on LB plates and incubated at 37°C overnight, and viable counts were determined. Percentages for each time point were obtained by dividing the surviving viable count by the initial viable count. Statistical analysis was done by the two-sample t test (35).

Attached versus internalized bacteria.

Bacteria grown to the exponential phase were pelleted and washed twice with saline. Approximately 3 × 10⁹ CFU of bacteria were resuspended in 1 ml of FITC (0.15 mg/ml) in DPBS. The suspension was protected from light and rotated at 37°C for 30 min to label the bacteria. Viability was determined before and after labeling. Bacteria were viable after labeling with 0.15 mg of FITC per ml in DPBS. The reactions were set up as described above; 50-μl samples were taken at 0, 5, 15, and 30 min and immediately placed on ice. The PMNs were separated from unattached bacteria as described above, resuspended in an equal volume of DPBS, and placed on ice. To quench the extracellular fluorescence of attached bacteria, 2.5 μl of a 0.1% ethidium bromide solution was added to the samples prior to visualization as described by Arduino et al. (3). The mixture was placed on a glass slide and overlaid with a coverslip, and the number of attached and/or internalized bacteria was determined by examination with a Leitz fluorescence microscope (Wetzlar, Germany) with an FITC filter under oil immersion (magnification of ×1,000). Bacteria attached to the PMNs appeared yellow or yellow-green with yellow-orange centers, whereas the internalized bacteria were an intense green. Twenty-five individual PMNs were examined per sample, and attached and ingested bacteria per PMN were counted. The results were averaged for four individual experiments.

RESULTS

To investigate whether the O75 O antigen and the K5 capsular antigen provide resistance to phagocytosis or phagocytic killing, phagocytosis assays were performed with an O75⁻ mutant (SMB20), a K5⁻ mutant (SMB213), and the parental strain (GR-12). Previously, an O75⁻ mutant and a K5⁻ mutant were constructed in the uropathogenic E. coli strain GR-12 by mutating the rfbD gene and the kfiC gene, respectively (4).

PMN phagocytosis killing assays.

The wild-type strain and mutants at the exponential phase of growth were mixed with PMNs at a 10:1 ratio of bacteria to PMNs with 10% serum or 10% heat-inactivated serum added. The reaction mixtures were incubated with rotation at 37°C for 2 h, and viable counts were determined in duplicate at 0, 15, 30, 60, and 120 min. To control for sensitivity of the bacterial strains to 10% serum, the same reactions were performed without PMNs added. The cumulative results of seven different assays using blood from seven different donors are shown in Fig. 1 as percent viability. A 15-min time point was performed for only five of the seven assays. The wild-type strain (GR-12) maintained its number for the first hour and then increased in number in the second hour of incubation. Both SMB20 (O75⁻ mutant) and SMB213 (K5⁻ mutant) decreased approximately 80% in viability in the first and second hours of incubation. These results were significantly different from the sensitivity exhibited by GR-12 (Table 1). The K5⁻ mutant was more sensitive to killing by PMNs in the first 15 min of incubation compared to GR-12 and the O75⁻ mutant. An average of only 42.5% ± 11.8% of the K5⁻ mutant was still viable in the first 15 min, compared to 148.8% ± 61.3% and 130.0% ± 34.1% for the O75⁻ mutant and GR-12, respectively (n = 5; P < 0.02 for SMB213 compared to SMB20 and P < 0.001 for SMB213 compared to GR-12). Yet, at 30 min there was no statistical difference in the survival of the O75⁻ mutant compared to the K5⁻ mutant. GR-12 and the mutants were not sensitive to 10% serum alone (Fig. 1). When 10% heat-inactivated serum was used as the complement source, GR-12 and the mutant strains were not sensitive to killing by PMNs (data not shown).

FIG. 1.

PMN phagocytosis killing assays. Wild-type strain GR-12 (squares) and O75⁻ (triangles) and K5⁻ (circles) mutants were mixed with PMNs at a 10-to-1 ratio of bacteria to PMNs. Assays were performed in 10% serum with (closed symbols) and without (open symbols) PMNs (as a control) for 2 h; samples were taken in duplicate at the start and at 15, 30, 60, and 120 min. The graph represents results of seven different assays for each strain, although the 15-min time was taken for only five of the seven assays. Error bars indicate standard deviations.

TABLE 1.

Viability following incubation with PMNs or monocytes

Cell type	Time (min)	Mean % viability ± SD			Pa (n)
		GR-12	SMB20 (O−)	SMB213 (K−)	GR-12 vs SMB20	GR-12 vs SMB213	SMB20 vs SMB213
PMN	15	130 ± 34	149 ± 61	45 ± 12	NS (5)	<.001 (5)	<.02 (5)
	30	107 ± 34	41 ± 26	27 ± 13	<0.01 (7)	<0.001 (7)	NS (7)
	60	103 ± 30	19 ± 12	21 ± 11	<0.001 (7)	<0.001 (7)	NS (7)
	120	199 ± 75	33 ± 27	27 ± 19	<0.002 (7)	<0.002 (7)	NS (7)
Monocyte	15	162 ± 32	105 ± 45	46 ± 20	<0.02 (6)	<0.001 (6)	<0.05 (6)
	30	119 ± 37	26 ± 23	27 ± 30	<0.001 (6)	<0.001 (6)	NS (6)
	60	81 ± 61	16 ± 13	15 ± 16	<0.1 (6)	<0.1 (6)	NS (6)
	120	99 ± 48	28 ± 29	8 ± 9	<0.02 (6)	<0.01 (6)	NS (6)

^aDetermined by the two-sample t test for statistical analysis (35). NS, not significant. 

Preopsonization phagocytosis killing assays.

To determine whether the difference in killing of SMB20 (O75⁻ mutant) and SMB213 (K5⁻ mutant) was due to a difference in the kinetics of opsonization of the bacteria, GR-12 and the mutants were preopsonized in 10% serum for 30 min to equalize the amount of time of opsonization prior to the addition to PMNs. The K5⁻ mutant was still killed more rapidly than the O75⁻ mutant despite preopsonization (Fig. 2). We noted an approximately 60% decrease in viability of SMB213 and an increase in viability of SMB20 at the 15-minute time point.

FIG. 2.

Preopsonization PMN phagocytosis assays of wild-type strain GR-12 (squares) and O75⁻ (triangles) and K5⁻ (circles) mutants at a 10-to-1 ratio of bacteria to PMNs. GR-12 and mutants incubated with PMNs (closed symbols) were preopsonized in 10% serum for 30 min prior to addition to PMNs. Controls were done with 10% serum and no PMNs added (open symbols). Assays were performed for 1 h; samples were taken at the start and at 15, 30, and 60 min. The graph represents results of three different assays of each strain. Error bars indicate standard deviations.

Separation of nonassociated bacteria from PMNs.

The assay described above (without preopsonization) was repeated except that PMNs were separated from nonassociated bacteria and the viable number of bacteria in the supernatant was determined. The bar graph in Fig. 3 compares GR-12 and the mutants by the number of viable bacteria in the supernatant compared to the initial number of viable bacteria in the reaction. At time zero, all bacteria are nonassociated, as approximately 100% are in the supernatant. The numbers of viable SMB213 (K5⁻ mutant) decreased by 80% in the supernatant at 15 min, while the viable numbers of GR-12 and SMB20 (O75⁻ mutant) exhibited no decrease in the supernatant fraction. By 30 min, the number of viable bacteria in the supernatant for both mutants had declined to 20%, while 65% of the parent remained in the supernatant. These results indicate that the K5⁻ mutant associated with PMNs faster than GR-12 and the O75⁻ mutant in the first 15 min, but by 30 min equal numbers of the K5⁻ and O75⁻ mutants were associated with the PMNs.

FIG. 3.

Bacterial viability in supernatant fraction following separation from PMNs. PMN phagocytosis assays were performed with wild-type strain GR-12 and mutants; bacteria associated with PMNs were separated from bacteria in supernatant at 0, 15, and 30 min; bacterial counts were compared to the initial total number of bacteria. The bar graph represents results of three different assays of each strain.

Uptake versus attachment assays.

Since preopsonization of the mutants did not predispose SMB20 (O75⁻ mutant) to faster killing by PMNs, similar to findings for SMB213 (K5⁻ mutant), we investigated whether the difference in killing was due to the kinetics of association. We performed a phagocytosis assay developed by Arduino et al. in which FITC-labeled bacteria are used along with ethidium bromide to quench the extracellular fluorescence of the attached bacteria, allowing distinction between attached and ingested bacteria (3). Bacteria were labeled with slight modification, using a lower concentration of FITC (0.15 mg/ml) in DPBS to retain bacterial viability as described in Materials and Methods. Average viabilities after 30 min of labeling for GR-12, SMB213, and SMB20 were 96.5, 94.5, and 87.0%, respectively. Labeled bacteria were added as before; samples were taken at 0, 5, 15, and 30 min and immediately placed on ice. After separation of PMNs from nonassociated bacteria, ethidium bromide was added to quench extracellular fluorescence of bacteria immediately before viewing under the microscope.

Results of the fluorescence assay are presented in Fig. 4. After 5 min of incubation with PMNs, there were averages of 3.32 ± 0.96 bacteria ingested and 1.28 ± 0.22 bacteria attached per PMN for SMB213 (K5⁻ mutant). For GR-12 and SMB20 (O75⁻ mutant), however, less than one bacterium was attached and/or ingested per PMN. At 15 min, the K5⁻ mutant increased its association with PMNs to 5.05 ± 0.77 ingested and 1.51 ± 0.13 attached, compared with 0.66 ± 0.37 ingested and 0.70 ± 0.18 attached for the O75⁻ mutant. GR-12 was similar to the O75⁻ mutant at 15 min, with approximately one bacterium attached and one ingested per PMN. Although at 30 min GR-12 and the O75⁻ mutant were similar in the number of associated bacteria per PMN, the O75⁻ mutant was ingested by PMNs in greater numbers than was GR-12. For GR-12, 2.79 ± 1.00 bacteria were attached per PMN, compared with only 0.51 ± 0.22 attached for the O75⁻ mutant. This difference was reversed, however, when the numbers of ingested bacteria were compared: for the O75⁻ mutant, 2.78 ± 0.80 bacteria were ingested per PMN, while for GR-12 the number was only 1.44 ± 0.80. For the K5⁻ mutant, the numbers of bacteria associated with PMNs were similar for the 15- and 30-min time points. Shown in Fig. 4B are the results of using heat-inactivated serum as a complement source. After 60 min, the K5⁻ mutant was both associated with and internalized by PMNs, while the O75⁻ mutant and GR-12 were relatively resistant to both attachment and ingestion.

FIG. 4.

Fluorescence assay to determine the number of attached and internalized bacteria per PMN for the wild-type strain GR-12 and mutants with 10% serum (A) or 10% heat-inactivated serum (B). FITC-labeled bacteria were added to PMNs as described in the text, samples were taken at 0, 5, 15, and 30 min (A) or 60 min (B), and nonassociated bacteria were separated. Ethidium bromide was added to quench the fluorescence of extracellular bacteria. Samples were viewed under a microscope, and attached and internalized bacteria were counted. Solid bar denote mean bacteria internalized, and hatched bars denote attached bacteria. The graph depicts results of four different assays of each strain.

We note that viability of the bacteria should be considered when the bacteria are labeled. Originally a higher concentration of FITC (1.0 mg/ml) in carbonate buffer was used to label the bacteria. The carbonate buffer along with the high concentration of FITC reduced the viability of the bacteria. When bacteria labeled with this higher concentration of FITC in carbonate buffer were used, the PMNs quickly internalized the mutants and GR-12 at the same rate (data not shown). Thus, results when nonviable bacteria are used should be interpreted with caution. In addition, the results of the fluorescence assay do not completely coincide with the supernatant viability at 30 min, although the same trends are seen. GR-12 seems to be associated and ingested in greater numbers when FITC labeled than when not labeled (compare Fig. 3 and 4). Thus, the labeling reaction may also increase the association of the bacteria with phagocytic cells.

Monocyte phagocytosis killing assays.

Although PMNs are a major defense mechanism against invading bacteria, monocytes are also thought to play a role in clearance of bacteria. The phagocytosis killing assay was repeated with monocytes at the same ratio, 10:1. The results of these experiments are presented in Fig. 5 and Table 1. There was donor-to-donor variability in killing, as depicted by the large error bars in the graph. Nonetheless, the difference in the means between GR-12 and the mutants were significant at all time points, as shown in Table 1. The difference in the killing of the O75⁻ and K5⁻ mutants at the 15-min time point was also significant. The viability of the K5⁻ mutant at 15 min was 46 ± 20, compared to 105 ± 45 and 162 ± 32 for the O75⁻ mutant and GR-12, respectively. There was no significant difference in killing of the mutants at subsequent time points.

FIG. 5.

Monocyte phagocytosis killing assays of wild-type strain GR-12 (squares) and O75⁻ (triangles) and K5⁻ (circles) mutants at a 10-to-1 ratio of bacteria to monocytes. Assays were performed in 10% serum with (closed symbols) and without (open symbols) monocytes (as a control) for 2 h; samples were taken in duplicate at the start and at 15, 30, 60, and 120 min. The graph represents results of six different assays for each strain. Error bars indicate standard deviations.

The bactericidal activities of PMNs and monocytes are similar for the O75⁻ and K5⁻ mutants. Monocytes are able to kill the mutants just as efficiently as PMNs. There was a significant difference in killing by PMNs and the monocytes at the 2-h time point, at which monocytes were more effectively able to kill GR-12 than were PMNs (P < 0.05).

DISCUSSION

The use of genetically defined mutants of a pathogenic strain allows for more conclusive determination of the importance of a specific virulence factor. We previously constructed proven isogenic mutants in a wild-type pathogenic strain, creating an O75⁻ mutant and a K5⁻ mutant. The mutants were analyzed phenotypically and tested in serum resistance assays. The O75 O antigen was shown to be more important than the K5 antigen in serum resistance (4). Although the K5 antigen was not critical in serum resistance, we sought to determine if it played a role in resistance to phagocytosis.

This is the first study in which isogenic mutants were used in phagocytosis assays. Previous studies to determine the role of O and K antigens in resistance to phagocytosis are for the most part inferential. Genetically defined mutants are superior for determining the importance of a virulence factor because it is possible to attribute the cause for loss or gain of a specific attribute to the mutation. In our study, we can directly compare an O75⁺ K5⁺ strain with an O75⁻ strain and a K5⁻ strain to reliably determine the role of the O75 and K5 antigens in resistance to phagocytic killing. The wild-type strain and the O75⁻ and K5⁻ mutants were tested in phagocytosis killing assays with PMNs. The wild-type strain was resistant to killing, whereas the mutants were sensitive. The O75⁻ and K5⁻ mutants were equally sensitive to killing at 30-min and subsequent time points. These results support the evidence that certain O and K antigens provide resistance to killing by phagocytes. In this study, the K5 capsule imparts resistance to phagocytic killing, as shown by comparison of the viability of the wild-type strain (bearing the K5 capsule) to that of the K5⁻ mutant in the phagocytosis killing assay. The O75 O antigen also imparts resistance to phagocytic killing similarly to the K5 antigen.

Despite similar killing of the mutants at 30-min and later time points, there was a significant difference in killing of the mutants at the 15-min time point, with the K5⁻ mutant exhibiting a more rapid decrease in viability than the O75⁻ mutant. Thus, the K5 capsule was more important in the survival of the organism during the initial incubation with PMNs. We demonstrated that the difference was not due to a difference in opsonization because despite preopsonization, there was still a significant difference in killing of the two mutants.

The kinetics of phagocytic killing may be divided into several stages: rate of opsonization, rate of attachment, rate of internalization, and rate of killing (7, 10). Since the difference in the killing of the mutants is not due to the rate of opsonization, we further analyzed the association of the mutants with PMNs and demonstrated that the K5⁻ mutant associated more rapidly with PMNs. To confirm the difference in rate of association and to differentiate the type of association (attached versus internalized bacteria), we used a fluorescence assay which distinguishes between attached and internalized bacteria (3). Our results indicate that the K5 capsule prevented initial association with the PMNs, and bacteria lacking the K5 capsule quickly attached and were internalized by PMNs. After the O75⁻ mutant eventually associated with PMNs, it was internalized faster than the wild-type strain. This indicated that in the presence of the K5 capsule, the O75 O antigen hinders internalization. Thus, we conclude that the K5 antigen prevents attachment to phagocytic cells and that both the O75 and K5 antigens confer protection from internalization.

When heated serum was used in the fluorescence assay, greater numbers of the K5⁻ mutant than of the O75⁻ mutant or parent strain GR-12 associated with PMNs. This finding indicates that complement is not necessary for the attachment and internalization of the unencapsulated mutant, but complement is necessary for attachment of the encapsulated strains. It is also interesting that although the K5⁻ mutant was internalized by PMNs in the presence of heat-inactivated serum, it was not killed. Thus, despite ingestion of the K5⁻ mutant, heat-labile factors were not present to completely stimulate the mechanisms necessary for killing the mutant.

In addition to PMNs, monocytes are an important defense to invading microorganisms. The rate and pattern of killing by monocytes paralleled those of the PMNs. These results support the findings of other groups who have found similar killing activity by monocytes (21, 23) but are in conflict with studies by Verbrugh (32), who found that monocytes had a lower bactericidal activity than PMNs. We also report that monocytes were significantly better at killing GR-12 at the 2-h time point. Grunwald et al. proposed that a CD14-defined phagocytosis pathway by monocytes may be able to destroy gram-negative pathogens which are resistant to both complement-mediated lysis and phagocytosis by PMNs (8). Our finding that GR-12 is more sensitive to killing by monocytes than PMNs supports this hypothesis.

Capsules are thought to impair phagocytosis by several mechanisms: by their hydrophilic nature, by serving as camouflage, and by masking the underlying opsonized surface (5, 10, 17, 29). Most investigations have focused on the K1 antigen, and more studies on the role of other K antigens and O antigens in phagocytosis are needed. The K5 capsule may prevent initial association with PMNs because the phagocytic cell does not recognize the opsonized bacteria. O antigens activate complement components that are recognized by complement receptors present on phagocytic cells. Antibodies to LPS which opsonize the bacterium are also formed. If the opsonized surface is covered by a capsule, the phagocytic cell will not recognize the opsonized bacteria. Kim et al. showed that the K5 capsule prevented anti-LPS monoclonal antibody from interacting with its target because it conferred protection to an unencapsulated but not an encapsulated, bacterial strain in a neonatal rat model (14).

Furthermore, the K5 capsule is poorly immunogenic due to its identity with an intermediate in the synthesis of heparin (30). The K5 capsule avoids immune recognition, and thus no antibodies against the K5 capsule are produced (13). The K5 capsule is also thought to not activate the alternative pathway (26). Thus, the K5 capsular surface is not opsonized by antibodies or complement.

The capsule may prevent association with the PMN simply due to its hydrophilicity and negative charge (17, 24, 31), characteristics that result in repulsion from the phagocytic cell (12). The K5⁻ mutant associates with PMNs in the absence or presence of complement, although faster in the presence of complement. Little is known about opsonization in the urinary tract other than the work of Suzuki et al. showing that there is a small amount of opsonic activity in urine (27). In the bladder, bacteria lacking a capsule may be phagocytized more efficiently than encapsulated strains and thus cleared from the bladder either by phagocyte killing or by voiding of the urine. Thus, if there is not enough complement present to effectively opsonize the bacteria, the K5 capsule plays a crucial role in preventing attachment and internalization.

This is the first study in which defined O75⁻ and K5⁻ mutants were used in phagocytosis killing assays to determine the role of the O75 and K5 antigens in resistance to phagocytic killing. We have shown that both the K5 capsular antigen and the O75 O antigen are important in resistance to phagocytic killing. However, the K5 antigen plays a more critical role in the initial encounter between bacteria and phagocytic cells.