Neutrophils upon Interaction with Escherichia coli - Infection and ...

3 downloads 0 Views 944KB Size Report
recurrent episodes of acute pyelonephritis: one with renal scars (12 strains) and one without renal scarring (11 strains). The majority of strains (from both groups) ...
Vol. 59, No. 11

INFECTION AND IMMUNITY, Nov. 1991, p. 4168-4172

0019-9567/91/114168-05$02.00/0

Extracellular Release of Reactive Oxygen Species from Human Neutrophils upon Interaction with Escherichia coli Strains Causing Renal Scarring HASSAN MUNDI,1 BENGT BJORKSTEIN,1 CATHARINA SVANBORG,2 LENA OHMAN,3 AND CLAES DAHLGREN3* Department of Pediatrics1 and Department of Medical Microbiology, Linkoping University, Linkoping, and Department of Clinical Immunology, University of Lund, Lund,2 Sweden Received 6 May 1991/Accepted 30 August 1991

The production of reactive oxygen metabolites by neutrophils plays a key role both in the host defense against invading microorganisms and in tissue damage resulting from infection. In the present study we measured the ability of different uropathogenic Escherichia coli strains to induce generation of oxygen metabolites upon interaction with human neutrophils. The strains were selected to represent two groups of patients with recurrent episodes of acute pyelonephritis: one with renal scars (12 strains) and one without renal scarring (11 strains). The majority of strains (from both groups) induced a pronounced neutrophil respiratory burst activity. When the intracellular and extracellular oxidative responses were measured separately, it was found that the response induced by nonscarring strains was primnarily of intracellular (intraphagosomal) origin, whereas a proportionally larger fraction of the response induced by the scarring strains was extracellular. Since reactive oxygen products are toxic to the renal tissue, this release can be of importance in the development of renal scars.

Escherichia coli strains that cause acute pyelonephritis special combinations of virulence factors. Many of these factors are missing in strains isolated from pyelonephritis patients who develop renal scars. We have therefore suggested that other bacterial characteristics are of importance for the development of the scars and/or that the host defense plays a more significant role than the bacteria in this process (30). Inflammatory cells, especially the professional phagocytes (i.e., neutrophil granulocytes and macrophages), are essential for the elimination of bacteria from the urinary tract (39). To kill invading microbes, the neutrophils are equipped with an arsenal of bactericidal molecules (17), involving both oxygen-independent and oxygen-dependent mechanisms. The oxygen-dependent pathway is characterized by an increased uptake of molecular oxygen, followed by superoxide anion (02-) and hydrogen peroxide (H202) formation (2, 8, 22). Together with myeloperoxidase and a halide, H202 constitutes a powerful antimicrobial system. A fraction of the O2-forming NADPH oxidase is localized to the plasma membrane of the neutrophil (4, 8, 36). Activation of this oxidase can result in an extracellular release of oxygen metabolites, which could be of importance for the killing of nonphagocytized bacteria. In order to contribute effectively to the killing process without significant destruction of tissue components, the cellularly produced oxidative metabolites should be formed within the phagocytic vacuoles (24). When released extracellularly, the bactericidal molecules are highly toxic to the renal tissue (35). In the present study we investigated the interaction between neutrophils and a number of E. coli strains isolated from patients suffering from pyelonephritis with and without renal scarring. The strains isolated from the scarred group gave rise to a more pronounced extracellular release of metabolites than did the strains isolated from the nonscarred possess

*

Corresponding author. 4168

group. The results suggest that the capacity of E. coli strains to cause extracellular neutrophil release of oxygen metabolites can be a key factor in the scarring process.

MATERIALS AND METHODS Patients and bacterial strains. The 23 E. coli strains were selected from a larger group of previously described strains (30) that were isolated from girls with a history of recurrent acute pyelonephritis (more than two episodes). The subjects were all included in a urinary tract infection follow-up program with regular intervals of clinical and radiological investigations at the Division of Nephrology at the Children's Hospital, Gothenburg, Sweden. The level of infection was determined by clinical, radiologic, and laboratory criteria (25). Acute pyelonephritis was diagnosed by the presence of significant bacteriuria in combination with a temperature >38°C, at least two abnormal laboratory findings of C-reactive protein (>20 mg/ml), and a microsedimentation rate -25 mrii/h or a transiently decreased renal concentrating capacity. Vesicouretric reflux (VUR) was diagnosed by means of voiding cystourethrography and graded 1 to 5 (23). Renal scarring was defined as calyceal deformity combined with parenchymal reduction on urography (21). The patients were divided into two groups, those with and those without renal scarring.

The bacterial strains isolated from the patients were first characterized according to the method of Edwards and Ewing (15) and then further analyzed for serotype, hemolysin production, sensitivity to the bactericidal effect of human serum, and binding properties (the relevant surface characteristics of the bacteria are summarized in Table 1). The strains were stored in deep agar slants, at 4°C, until used. For studies of interaction with neutrophils, the bacterial strains were grown on nutrient agar plates for 16 to 18 h and then transferred to liquid growth medium (tryptic soy broth) and grown overnight on a rotary shaker at 37°C. Thereafter

VOL. 59,

1991

E. COLI ACTIVATION OF NEUTROPHILS

TABLE 1. Surface characteristics of the E. coli strains isolated from patients with acute pyelonephritis with or without renal scarring Strain no.

Scarring strains 1 2 3 4 5 6 7 8

9 10 11 12

Nonscarring strains 13 14 15 16 17 18 19 20 21 22 23 MS, mannose sensitive; a

Agglutinationa

Antigen 0

K

H

MS

MR

2 2 75 75 1 7 1 8 91 3 25 91

1 1 + 5

4 6

-

+

+

-

-

+

-

+

1

-

+

-

1

55

+

-

-

+

-

-

+ -

-

19 28 4 1 7

-

-

-

6 17 16 1 1 16 2 7 7 4 4

2 53 1 1 1 1 1 7 1 7 13

1 18

+

-

-

-

7 7 6 4 6

+ -

+ + + + + + + + +

1 7

-

32 -

+

-

+ + +

-

-

-

+

MR, mannose resistant.

the bacteria were harvested and washed three times in phosphate-buffered saline (pH 7.3) and resuspended (109/ml) in Krebs-Ringer phosphate buffer supplemented with glucose (10 mM), Ca2+ (1 mM), and Mg2+ (1.5 mM) (KRG; pH 7.2). Fluorescein isothiocyanate labelling of the bacteria was performed as described by Hed (20). The fluorescein isothiocyanate-labelling procedure does not affect cell viability or the physicochemical surface properties of the bacteria as measured with hydrophobic interaction chromatography and two-phase partitioning (unpublished observations). Preparation of human polymorphonuclear neutrophils. To obtain normal neutrophils, heparinized blood from apparently healthy volunteers was separated by the method of B0yum (6). After dextran sedimentation, hypotonic lysis, and Hypaque-Ficoll centrifugation, the neutrophils were washed twice and resuspended to 2 x 107 cells per ml in KRG and transferred to a melting ice bath. Samples were withdrawn from the cell suspension and used in the chemiluminescence (CL) or phagocytosis assay. Phagocytosis system. The fluorescence technique described by Hed (20) was used to determine phagocytebacterium interaction. In short, the neutrophils (5 x 104) were placed on multispot microscope slides and incubated for 20 min at 37°C in a humid environment. After the slides were washed, a bacterial suspension was added and the slides were incubated for another 30 min. The slides were washed and kept on ice until examined under the microscope. With this technique the total number of cell-associated bacteria (surface-adherent as well as internalized bacteria) is determined (20). CL measurement. CL was measured in a six-channel

4169

Biolumat LB 9505 (Berthold, Wildbad, Germany) with disposable 4-ml polypropylene tubes. The tubes, each containing a 1.0-ml reaction mixture of KRG, neutrophils (2 x 106), and luminol (5 x 10' M), were placed in the Biolumat and allowed to equilibrate for 5 to 10 min at 37°C. Two different reaction mixtures were used for quantification of intracellularly and extracellularly generated luminescence: the tubes were supplied with azide (1 mM) and horseradish peroxidase (HRP; 4 U) for extracellularly generated CL, whereas superoxide dismutase (200 U) and catalase (2,000 U) were added to the tubes for the measurement of intracellularly generated CL. Bacterial suspension (100 ,ul) was added to activate the system, and the light emission was then recorded continuously. Chemicals. Luminol was obtained from Sigma (St. Louis, Mo.). The scavengers catalase and superoxide dismutase as well as HRP (grade 1) were obtained from Boeringer Mannheim. NaOH (0.1 M) was used to dissolve luminol to 0.02 M. All substances were further diluted in KRG (pH 7.2). Statistics. The Mann-Whitney-Wilcoxon sign rank test was used for statistical analysis. RESULTS Bacterial binding to the neutrophils. Binding of the E. coli strains to the neutrophils was measured by microscopic examination. The total number of cell-associated bacteria did not correlate with the known adhesin expression of the strains. Consistent with previous studies, most of the isolates expressing mannose-resistent adhesins showed a low degree of cell association. Even though the significance is low (P < 0.05), our present results indicate that the bacterial strains in the scarring group have an increased capacity for

binding to neutrophils (Fig. 1). CL response of neutrophils. Luminol-amplified CL was used to investigate the metabolic response induced in neutrophils upon interaction with the E. coli bacteria. Most of the bacterial strains induced such a response. However, the responses varied in both quantity and time course. The response induced by the scarring strains was not significantly higher than that elicited by the nonscarring strains, nor did the time to peak differ significantly between the two groups (data not shown). Extracellular and intracellular CL activity. If the oxygen metabolites generated during neutrophil phagocytosis are to be restricted to the intracellular compartment, they should be produced after closing of the phagocytic vacuole. We have previously shown that luminol-amplified CL is dependent on the presence of the granule enzyme myeloperoxidase (12); this knowledge can be exploited to determine the relationship between intracellularly and extracellularly produced CL activity (7, 29). The detection system for the extracellular response contained azide (a low-molecularweight myeloperoxidase inhibitor that gains access to intracellular compartments) combined with an azide-insensitive peroxidase (HRP). Azide alone totally inhibited the CL response, but in combination with HRP the response was

regenerated. Since HRP is a large protein that does not have

sites, it can be used to detect extracellularly released oxygen metabolites. The system for quantification of the intracellular response contained superoxide dismutase and catalase, large molecular scavengers of 02 and H202, respectively. In the presence of these scavengers the extracellularly produced 2- and H202 will be removed, thus providing a measurement of the intracellular response. In our experiments, the extracellular peak value was

access to intracellular

4170

INFECT. IMMUN.

MUNDI ET AL. co

200

-

022

0

0 cn 0

0 11

x

E 3.0 a 0

@ 10 cv

* 9

0

0 0

0 0 o 0 (D

CO0

._

4._

100

l-

* 1 5122

* 4 6 0 8

@18

-o0L.0

@ 20

@21 * *

@ 17 @ 14

13,15

@7 @3

*

5

E z

1.5

.E

0

0)

0 0

!i non-scarring

C-

0

20

40

60

80

Time (min) FIG. 2. Representative time traces of intracellular (solid line)

and extracellular (dotted line) CL produced by neutrophils following an interaction with E. coli bacteria (strain 22). The system used to measure intracellular activity contained superoxide dismutase (200 U) and catalase (2,000 U), whereas the system used to measure extracellular activity contained HRP (4 U) and azide (1 mM).

scarring

FIG. 1. Adherence of different E. coli strains when interacting with human neutrophils. The number of associated bacteria per 100 neutrophils was quantified, and the symbols represent the mean value from two experiments. The strain number (Table 1) is given to the right of each symbol. Bacterial adhesion was higher among the scarring strains than among the nonscarring strains (P < 0.05).

always reached at an earlier time point than the intracellular peak. This suggested that adhesion of the bacteria to the phagocyte cell surface was associated with a greater leakage of oxygen metabolites than was the internalization. Representative curves of the response are shown in Fig. 2. There was no difference in the intracellular activity induced by the two groups of bacteria (Fig. 3a). Most strains belonging to the nonscarring group did not induce any extracellular activity, whereas most of the strains belonging to the scarring group did. We thus conclude that proportionally more extracellular activity was induced by the bacteria associated with scarring (Fig. 3b; P < 0.01). DISCUSSION We interpret the presented data to mean that there are qualitative differences in the ability to induce an extracellular release of reactive oxygen species from human neutrophils between uropathogenic E. coli bacteria isolated from patients with and without renal scarring. Under circumstances ideal for the host, the interaction between neutrophils and invading bacteria should result in bacterial killing with minimal damage to host tissues. However, during neutrophil phagocytosis, the cellularly produced oxygen metabolites that are released inside the phagosome may leak out to the surrounding tissue, where they can inflict considerable damage (17). The techniques commonly used to measure the production of reactive oxygen metabolites cannot distinguish between intracellular and extracellular production (33). With the luminol-amplified CL technique, however, extracellular and intracellular events

connected with a cellular response can be measured (3, 7, 11, 12, 19, 27, 29). The CL reaction is peroxidase dependent and totally inhibited by the myeloperoxidase inhibitor azide (12, 13, 16, 37). In addition, the H202 scavenger catalase, the 02 scavenger superoxide dismutase, and the azide-insensitive HRP are large proteins that have no access to intracellular sites; this means that the extracellular and intracellular CL responses can be detected separately (9-11, 27, 29). The majority of bacterial strains (both scarring and nonscarring) activated the neutrophils to generate oxygen radicals. Most of the activity induced by the nonscarring strains was limited to an intracellular compartment, possibly the phagosome, whereas the scarring strains induced a substantial extracellular release of the oxygen metabolites produced; this could be an effect of a lack of phagosome formation or vacuole closure. The ability of neutrophils to release oxygen metabolites could be important under conditions in which intracellular sequestering of bacteria is incomplete (28, 41), but the goal of killing must then be balanced against the threat made by the released metabolites on the host tissue. The role of bacterial properties in the induction of renal scarring is a matter of debate (18, 21, 30, 40). The formation of scars has been assumed to result from the inflammatory tissue reactions during acute pyelonephritis (42). Bacterial properties which enhance the acute inflammatory response, e.g., binding to Gala1-4Galp-containing receptors, would then be expected to increase the renal scarring frequency. However, strains from patients with renal scarring were rarely found to belong to the most uropathogenic E. coli clones (14, 30). The ability of E. coli to induce renal scars was thus suggested to be due to other unidentified virulence properties (40). This hypothesis was tentatively refuted by the finding that the scarring strains, as a group, lacked selectivity in terms of clonal identification markers. The findings of the present study provide the first suggestion for a mechanism whereby the scarring strains might have an increased propensity for attacking the renal tissue. Neutrophil-E. coli interactions are initiated by binding of bacteria to the phagocyte membrane, a process that is

E. COLI ACTIVATION OF NEUTROPHILS

VOL. 59, 1991 331

4171

* 11

10I

10

-

0 14

0 0 0 0

0 c 0) 0

0

5

co

0 9

iE

0

5

-

.E D 0

@ 22

0

0

0

0 11

@4

a

@ 20 * 13 . 18 @ 21 * * 15,23

non-scarring

@9

*12

@6

:* 86

scarring

b

02 aee~0'22

non-scarring

01.7 scarring

FIG. 3. Intracellular (a) and extracellular (b) CL induced in human neutrophils by the different E. coli strains. The response was quantified as the peak CL values (108 cpm) obtained, and the symbols represent the mean value obtained from three experiments. The strain number (Table 1) is given to the right of each symbol. The intracellular response did not differ between the two groups, whereas a higher extracellular release was induced by scarring bacterial strains than by the nonscarring strains (P < 0.01).

influenced by the surface characteristics of both the phagocyte and the prey (31). Binding is thus enhanced by bacterial lectins recognizing mannose residues (type 1 fimbriae) and fimbriae recognizing sialic acid residues (S fimbriae) (1, 32). Receptors for P fimbriae (i.e., the globoseries of glycolipids) are not abundant on human phagocytes (5, 38). Consistent with this, we show in the present study that most of the strains expressing mannose-resistant adhesins bound poorly to the phagocytes. There was no pattern in adhesin expression that could explain the bacterial binding to the neutrophils. However, although both the bacterial lipopolysaccharide and the capsular polysaccharide influence phagocyte-bacterium interactions, the differences between the scarring and nonscarring strains could not be explained by the serotype. Therefore, the bacterial determinant of extracellular release of oxidative metabolites remains to be identified. The data concerning the relationship between inflammatory response and renal scarring are somewhat of a paradox. In experimental models, scarring can be caused by the acute inflammatory response (18, 40), a process that can be interrupted by treatment with antibiotics or anti-inflammatory agents (18, 26). In humans, however, bacteria which cause the highest acute inflammatory response (measured as fever and elevated C-reactive protein (CRP), erythrocyte sedimentation rate, and number of leukocytes in the urine) rarely cause renal scars (14, 30). This suggests that there may be qualitative differences between the inflammatory responses of patients with and without renal scarring. The results presented in this study indicate that the extracellular release of oxidative metabolites is a key to the scarring process. The fact that scavengers of oxidative metabolites can protect the

renal tissue against the development of scars (34, 35) further strengthens this conclusion, but it should once again be pointed out not only that bacterial characteristics are of importance for the outcome of a bacterial phagocyte interaction but also that host defense could play a significant role. Thus, the role of phagocyte host factors for the oxygen metabolites released from phagocytes and for the damage to the human kidney inflicted by infection needs further investigation. ACKNOWLEDGMENTS This work was supported by grants from the Swedish Medical Research Council, the King Gustav V 80-Year Foundation, the Swedish Society against Rheumatism, the County Council of Ostergotland, and the medical faculty, University of Lund. The skillful technical assistance of Kerstin Hagersten is gratefully acknowledged, as is advice on statistics from Erik Leander. REFERENCES 1. Bar-Shavit, Z., I. Ofek, R. Goldman, D. Mirelman, and N. Sharon. 1977. Mannose residues on phagocytes as receptors for attachment of Escherichia coli and Salmonella typhi. Biochem. Biophys. Res. Commun. 78:455-460. 2. Bellavite, P. 1988. The superoxide forming enzymatic system of phagocytes. Free Radical Biol. Med. 4:225-261. 3. Bender, J. G., and D. E. Van Epps. 1983. Analysis of the bimodal chemiluminescence pattern stimulated by chemotactic factors. Infect. Immun. 41:1062-1070. 4. Bjerrum, 0. W., and N. Borregaard. 1989. Dual granule localization of the dormant NADPH oxidase and cytochrome b559 in human neutrophils. Eur. J. Haematol. 43:67-77. 5. Bluhmenstock, E., and K. Jann. 1982. Adhesion of piliated Escherichia coli strains to phagocytes: differences between

4172

6.

7.

8. 9.

10.

11. 12. 13.

14.

15.

16. 17. 18. 19. 20.

21. 22.

23. 24.

25.

MUNDI ET AL.

bacteria with mannose-sensitive pili and those with mannoseresistent pili. Infect. Immun. 35:264-269. B0yum, A. 1968. Isolation of mononuclear cells and granulocytes from human blood. Isolation of mononuclear cells by one centrifugation and of granulocytes by combining centrifugation and sedimentation at 1xg. Scand. J. Clin. Lab. Invest. Suppl. 97:77-89. Briheim, G., 0. Stendahl, and C. Dahlgren. 1984. Intra- and extracellular events in luminol-dependent chemiluminescence of polymorphonuclear leukocytes. Infect. Immun. 45:1-5. Clark, R. A. 1990. The human neutrophil respiratory burst oxidase. J. Infect. Dis. 161:1140-1147. Dahlgren, C. 1987. Difference in extracellular radical release after chemotactic factor and calcium ionophore activation of the oxygen radical generating system in human neutrophils. Biochim. Biophys. Acta 930:33-38. Dahlgren, C. 1988. Effect on extra- and intracellularly localized, chemoattractant induced, oxygen radical production in neutrophils following modulation for ligand-receptor interaction. Inflammation 12:335-349. Dahlgren, C. 1988. Analysis of luminol-dependent CL from granule depleted neutrophil cytoplast reveals two different light emitting mechanisms. J. Biolumin. Chemilumin. 2:25-33. Dahlgren, C., and 0. Stendahl. 1983. Role of myeloperoxidase in luminol-dependent chemiluminescence of polymorphonuclear leukocytes. Infect. Immun. 39:736-741. DeChatelet, L. D., G. D. Long, P. S. Shirley, D. A. Bass, M. J. Thomas, F. W. Henderson, and M. Cohen. 1982. Mechanism of the luminol-dependent chemiluminescence from human neutrophils. J. Immunol. 129:1589-1593. DeMan, P., I. Claeson, I. M. Johansson, U. Jodal, and C. Svanborg-Eden. 1989. Bacterial attachment as a predictor of renal scarring in boys with urinary tract infection. J. Pediatr. 115:915-922. Edwards, P. R., and W. H. Ewing. 1986. Identification of Enterobacteriacea. Elsevier, New York. Edward, S. W. 1987. Luminol and lucigenin-dependent chemiluminescence of neutrophils: role of degranulation. J. Clin. Lab. Immunol. 22:35-39. Gallin, J. I., 1. M. Goldstein, and R. Snyderman. 1988. Inflammation. Basic principles and clinical correction. Raven Press, New York. Glausser, M., J. M. Lyons, and A. I. Braude. 1978. Prevention of chronic pyelonephritis by suppression of acute suppuration. J. Clin. Invest. 61:403-407. Hallet, M. B., and A. K. Campbell. 1983. Two distinct mechanisms for stimulation of oxygen-radical production by polymorphonuclear leukocytes. Biochem. J. 216:459-465. Hed, J. 1977. The extinction of fluorescence by crystal violet and its use to differentiate between attached and ingested microorganisms in phagocytosis. FEMS Microbiol. Lett. 1:357361. Hodson, C. J., and S. Wilson. 1965. Natural history of chronic pyelonephritic scarring. Br. Med. J. 2:191-194. Hurst, N. P. 1987. Molecular basis of activation and regulation of the phagocyte respiratory burst. Ann. Rheum. Dis. 42:265272. International Reflux Study Committee. 1981. Medical versus surgical treatment of primary vesicoureteral reflux: a prospective international reflux study in children. J. Urol. 125:277-283. Jesaitis, A. J., E. S. Buescher, D. Harrison, M. T. Quinn, C. A. Parkos, S. Livesey, and J. Linner. 1990. Ultrastructural localization of cytochrome b in the membranes of resting and phagocytosing human granulocytes. J. Clin. Invest. 85:821-835. Jodal, U., U. Lindberg, and K. Lincoln. 1975. Level diagnosis of symptomatic urinary tract infections in childhood. Acta Pediatr.

INFECT. IMMUN.

Scand. 64:201-208. 26. Linder, H., I. Engberg, C. van Kooten, P. de Man, and C. Svanborg-Eden. 1990. Effects of anti-inflammatory agents on the mucosal inflammation induced by infection with gram-negative bacteria. Infect. Immun. 58:2056-2060. 27. Lock, R., and C. Dahlgren. 1988. Characteristics of the granulocyte CL reaction following an interaction between human neutrophils and Salmonella typhimurium bacteria. APMIS 96: 299-305. 28. Lock, R., C. Dahlgren, M. Linden, 0. Stendahl, A. Svensbergh, and L. Ohman. 1990. Neutrophil killing of two type 1 fimbriabearing Escherichia coli strains: dependence on respiratory burst activation. Infect. Immun. 58:37-42. 29. Lock, R., A. Johansson, K. Orselius, and C. Dahlgren. 1988. Analysis of horseradish peroxidase-amplified chemiluminescence produced by human neutrophils reveals a role for the superoxide anion in the light emitting reaction. Anal. Biochem. 173:450-455. 30. Lomberg, H., M. Hellstrom, U. Jodal, I. 0rskov, and C. Svanborg Eden. 1989. Properties of E. coli in patients with renal scarring. J. Infect. Dis. 159:579-582. 31. Magnusson, K. E., C. Dahlgren, G. Maluszynska, E. Kihlstrom, T. Skogh, 0. Stendahl, G. Soderlund, L. Ohman, and A. Walan. 1985. Non-specific and specific recognition mechanisms of bacterial and mammalian membranes. J. Dispersion Sci. Technol. 6:69-89. 32. Marre, R., J. Hacker, W. Henkel, and W. Goebel. 1986. Contribution of cloned virulence factors from uropathogenic Escherichia coli strains to nephropathogenicity in an experimental rat pyelonephritis model. Infect. Immun. 54:761-767. 33. Metcalf, J. A., J. I. Gallin, W. M. Nausseef, and R. K. Root. 1986. Laboratory manual of neutrophil function. Raven Press, New York. 34. Ransley, P. G., and R. A. Risdon. 1981. Reflux nephropathy: effects of antimicrobial therapy on the evolution of the early pyelonephritic scar. Kidney Int. 20:733-742. 35. Roberts, J. A., J. K. Roth, G. Dominigue, R. W. Lewis, S. Kaack, and G. Baskin. 1982. Immunology of pyelonephritis in the primate model. V. Effect of superoxide dismutase. J. Urol. 128:1394-1400. 36. Segal, A. W. 1988. The molecular and pathology of chronic granulomatous disease. Eur. J. Clin. Invest. 18:433-443. 37. Stevens, P., D. J. Winston, and L. Van Dyke. 1978. In vitro evaluation of opsonic and cellular granulocyte function by luminol-dependent chemiluminescence: utility in patients with severe neutropenia and cellular deficiency states. Infect. Immun. 22:41-51. 38. Svanborg-Eden, C., L. M. Bjursten, R. Hull, S. Hull, K. E. Magnusson, Z. Moldovano, and H. Leffler. 1984. Influence of adhesins on the interaction of Escherichia coli with human phagocytes. Infect. Immun. 44:672-680. 39. Svanborg-Eden, C., L. Hagberg, R. Hull, K. E. Magnusson, and L. Ohman. 1987. Bacterial virulence versus host resistance in the urinary tracts of mice. Infect. Immun. 55:1224-1232. 40. Topley, N., R. Steadman, J. M. Knowlden, and T. D. Williams. 1989. Type 1 fimbriate strains of Escherichia coli initiate renal parenchymal scarring. Kidney Int. 36:609-616. 41. Weiss, J., L. Kao, M. Viktor, and P. Elsbach. 1985. Oxygendependent intracellular and oxygen-dependent extracellular killing of Escherichia coli S 15 by polymorphonuclear leukocytes. J. Clin. Invest. 76:206-212. 42. Winberg, J., I. Bollgren, G. Kallenius, R. Mollby, and S. B. Svensson. 1982. Clinical pyelonephritis and focal renal scarring. A selected review of pathogenesis, prevention and prognosis. Pediatr. Clin. North Am. 29:801-814.