1033

Human Neutrophil Responses to Pathogenic Escherichia coli Are Receptor-Specific and Selectively Augmented by Recombinant Human Tumor Necrosis Factor-a R. Steadman, N. Matthews, M. Lichodziejewska, and J. D. Williams

Institute of Nephrology, Kidney Research Unit Foundation, University of Wales College of Medicine, Cardiff Royal Infirmary, and Department of Medical Microbiology, University Hospital of Wales, Cardiff, United Kingdom

Both the in vitro activation of human neutrophils (PMNL) and the in vivo initiation of renal parenchymal scarring by Escherichia coli are directly dependent on the expression of proteinacious, hairlike fimbriae bearing specific mannosedependent adhesins [1, 2]. PMNL respiratory burst activation, comprehensive degranulation, and phagocytosis are stimulated in response to organisms bearing these mannosesensitive (type 1) fimbriae [1, 3, 4]. PMNL activation is essential for the initiation of tissue damage, which precedes the formation of renal scars characteristic of E. coli-induced chronic pyelonephritis [2, 5]. However, in the absence of scarring, there is an extensive inflammatory infiltrate in response to nonfimbriated organisms [2]. Thus neutrophil recruitment but not necessarily activation may be stimulated by organisms that are relatively nonvirulent in terms of scarring potential. Previous studies have investigated the binding of E. coli to different structures on the PMNL surface, and at least three glycoproteins have been isolated (but not characterized) from PMNL membranes that will bind to type 1 fimbriae in a mannose-inhibitable manner [6]. In addition, it is clear from other studies that Fe receptor expression is also necessary for type 1 fimbriae-dependent binding to PMNL but not for

the binding of non-type l-fimbriated strains [7, 8]. The binding of these non-type l-fimbriated organisms, in contrast, is dependent on the expression of complement receptors (CRI and CR3) on the PMNL cell membrane [7, 9]. While Fc receptor-dependent activation of PMNL by IgG-coated particles [10] and CR3-dependent activation by zymosan [11, 12] have been studied and shown to involve independently regulated effector responses, the PMNL responses to pathogenic bacteria have not previously been linked to receptor occupancy. The interaction of PMNL with the inflammatory cytokine tumor necrosis factor (TNF) results in the augmentation of PMNL responses to a variety of ligands in vitro including unopsonized zymosan [13, 14], the chemotactic peptide formylmethionylleucylphenylalanine and phorbol ester [15-17]. TNF also augments an increased expression of cell surface receptors including the CDII/CDI8 family of adhesion proteins as well as the C3b receptor (CRl) [18]. This augmentation of PMNL responses is highly selective [19], and the intracellular mechanism of TNF stimulation is not yet elucidated. We examined the receptor specificity of particular bacterial-PMNL interactions and their potential up-regulation by TNF.

Methods Received 13 July 1990; revised 18 December 1990. Financial support: British Council Scholarship (to M.L.). Reprints or correspondence: Dr. J. D. Williams, Institute of Nephrology, Kidney Research Unit Foundation, University of Wales College of Medicine, Royal Infirmary, Cardiff, UK cn ISZ. The Journal of Infectious Diseases 1991;163:1033-1039 © 1991 by The University of Chicago. All rights reserved. 0022-1899/91/6305-0015$01.00

Bacterial strains. Four uropathogenic strains of E. coli were stored on Dorset egg slopes (Oxoid, Basingstoke, UK) and, before use, were subcultured at least three times overnight either in Oxoid nutrient broth no. 2 (NB2) or modified Vogel-Banner's medium (MVBM) to allow maximal expression or suppression of fimbriae, respectively. Bacteria were then harvested by centrifugation (200

Downloaded from http://jid.oxfordjournals.org/ at New York University on July 22, 2015

The effect of recombinant human tumor necrosis factor-a (rhTNFq) on neutrophil (PMNL) response to uropathogenic Escherichia coli was assessed. A strain expressing mannose-sensitive adhesins (type 1 fimbriae) stimulated significant primary granule, secondary granule, and leukotriene B4 (LTB4) release. The same strain grown to suppress fimbrial expression and three non-type 1-fimbriated strains stimulated only background low-level PMNL activation. The binding of the type 1-fimbriated strain to PMNL was inhibitable by n-mannose and concanavalin A, while that of non-type 1-fimbriated strains was not but was inhibited by antibodies to the PMNL complement receptors 1 and 3. TNFa (10-9 M) synergistically augmented the non-type 1-fimbriated E. coli-stimulated LTB4 release and additively increased secondary granule release without affecting primary granule release. In contrast, none of the responses to the type 1-fimbriated strain were augmented by TNFa. In the absence of mannose-sensitive adhesins, the activation ofPMNL by E. colimay involveboth complement receptors 1 and 3 and be augmented by TNFa.

1034

Steadman et aI.

g for 15 min) and washedtwice in Oxoid PBS, pH 7.3. Organisms

Table 1. Escherichia coli strains used in this study. Fimbriation Strain 504 SC DF AB

NB2 Type 1 P

MVBM

0

6 1 11

NT

K U 1 U U

NOTE. Strain 504 only was hemolytic. NT = not a types 1, 2, 3, 4, 5, 6, 7, 8,9, 11, 15, 17, 18,25, or 75. U = not Kl or K5. NB2 = nutrient broth no. 2; MVBM = modified Vogel-Bonner's medium.

BSA(!FA buffer). The preparations werefound to haveretained85% ± 13% of the PMNL present in the original suspension (counting three randomlyselectedmedium-power fields of each of 10cytospin preparations). FITC-Iabeled concanavalin A (ConA; 10lLg/ml) placed on the slides at room temperature for 20 min was used to visualize mannose-containing cell surface glycoproteins. This binding was totally inhibited by 1% (wt/vol) n-mannose in IFA buffer or by a 25-foldexcessof unlabeledConA. Cell surfacereceptor antigenexpression was confirmed after blocking nonspecific sites with normal humanABserum (NHS; 1:5in IFAbuffer) at room temperature for 20 min, usingmonoclonal anti-CRtand CR3antibodies (Serotec, Oxford, UK) in IFA buffer. Goat anti-mouse FITC-Iabeled antiserum (lCN Biomedicals) wasusedto visualize specific antibody binding.Nonimmune mouseserumandlabeled goatanti-mouse antiserum in the absenceof specific antibodyformedthe nonspecific controls. In a separate series of experiments, PMNL were pretreated for 20 min withdosesof rhTNFa in PBS,pH 7.3, at 37°Cbeforecytocentrifugation and fixation in 4% (wt/vol) paraformaldehyde. Bacterial adherence. Bacteria were grown overnight in either NB2 or MVBM in the presence of 1 ILCi/ml tritiated amino acids, harvested by centrifugation, washedin PBS, pH 7.3, as above, and suspended in IFA buffer at a concentration of 1 x 109 cfu/ml. Labelingefficiency wasbetween 11780 and 1/1349 (cpm/cfu) for 10 preparations of strain 504 grown in MVBM and between 1/9104 and 1113,090 (cpm/cfu) for 10 preparations of strain 504 grown in NB2. PMNL cytocentrifuge preparations were overlaidwith 10 ILl of radiolabeled E. coli suspension in IFA buffer containing 0.1% (wt/vol)sodiumazide for 60 min at room temperature. Slides were then washedextensively in PBS,pH 7.3,and adherentbacteria were removed for scintillationcounting with 1% (vol/vol) Triton X-lOO in PBS, pH 7.3. The number of adherent organismswas calculated after subtractionof backgroundvaluesfrom incubationscarried out on slideswithoutneutrophils. In parallel experiments, doses of both NHS and ConA were used to block bacterial adherenceby preincubationwiththe PMNL preparations for 30 min at roomtemperature. Treatment ofPMNLwith rhTNFa. PMNL were incubated with concentrations of rhTNFa from 10-8 to 10-14 Min KRPG for up to 60 min at 37°C. For cell sortinganalysisby IFA,the PMNL were immediately centrifuged at 2000 g for 20 s, resuspended, placed on ice, and treated with 50 JlI of monoclonalanti-CR3(anti-CDllb; Serotec) at 1/2000dilution as previouslydescribed [14]. The fixed cells were stored in the dark at 4°C until analyzedby fluorescenceactivated cell sorting (FACS) on a FACS 440 (Becton Dickinson, Oxford, UK). Percentageincreases aboveinitial fluorescence were recorded. PMNL stimulation. After preincubation at 1 x 107/ml with

Downloaded from http://jid.oxfordjournals.org/ at New York University on July 22, 2015

were resuspended to an optical density of 1.0at 560 om (5 X 108 cfu/ml). Bacterial characteristics. The strains, characterized by 0 and K serotyping, were further characterized by the mannose-sensitive hemagglutination of guinea pig or human erythrocytesand by their hemolyticpotential (table 1). The hemolyticcapacity of individual strains was measured by plating onto diagnostic sensitivitytesting agar plates (DST; Oxoid) containing 10% (vol/vol) sheep erythrocytesor by the lysis of sheep erythrocytes(10% vol/vol) (measured as an increasein absorbanceat 415 run) in Krebs-Ringerphosphate buffer, containing 11 mM n-glucose, 0.54 mM Ca2+, and 1.2 mM Mg2+ (KRPG). Fimbrial expression. Strains were tested for their mannosesensitiveagglutination of either 3% (vol/vol) guinea pig or human erythrocytes prepared from citrated, peripheral blood and washed twice in PBS, pH 7.3, in the presence or absence of 2.5 % (wt/vol) n-mannose.Strains werealso tested for P fimbrialexpressionusing a specific latex agglutination test kit (PF test; BradsureBiologicals, Loughborough, UK). Indirect immunofluorescence assay (IFA). Dried bacterial samples were fixed with cold 3.5% (wt/vol) paraformaldehyde (BDH, Poole, UK) in PBS,pH 7.1, for 15min and washedthree times with PBS,pH 7.3,for 5 min. The plates were incubatedat roomtemperature for 15 min with a 1:50 dilution of the appropriate specific antifimbrial antiserum [20,21], washed, and incubatedfor 15 min at roomtemperature witha 1:50dilutionoffluorescein isothiocyanate (FITC)-conjugated goat anti-rabbit IgG (ICN Biomedicals, High Wycombe, UK) in PBS, pH 7.3, as previously described [21]. The number of bacteria specifically stained with antifimbrial antiserum was counted directly. Fimbriationwas confirmedby electron microscopyof the pellets of eachorganismresuspended in 2%(wt/vol)phosphotungstic acid, containing0.2% (wt/vol) bovineserum albumin (BSA) , pH 7.0, and dried onto carbon-coatedcelloidin films on copper grids (Emscope Laboratories, Ashford, UK). For specificimmunolocalization, the anti-P or anti-type 1 fimbrial antibodies were diluted 1:50 in 1% (wt/vol) BSA(ICN Biomedicals) in PBS, pH 7.3.The washedgrids were then floated on a drop of goat anti-rabbit antibody linked to colloidal gold (AuroprobeEM GAR 10 om; Janssen Pharmaceuticals, Wantage, UK), as previously described [21]. Recombinant human TNFa (rhlNFa)preparation. rhTNFa (gift of G. R. Adolf, Ernst-Boehringer Institute, Vienna) was diluted in PBS, pH 7.3,and stored in aliquots of 100 ng/lOO ILl at -70°C. Dilutions were made in KRPG immediately before each experiment and kept at 4°C. Preparation ofhumanPMNL. Normal human leukocytes were isolatedfrom citratedperipheralblood by dextransedimentation and renderedplasma-freeand platelet-poorby washing three times with PBS, pH 7.3. PMNL were purified by density gradient centrifugation at 400 g for 35 min at 23°C on cushions of Ficoll-Hypaque (Pharmacia, MiltonKeynes, UK).Microscopic examination of Neatstained(GuestMedical,Sevenoaks, UK), cytocentrifuged (Cytospin II; ShandonSouthernProducts, Runcorn, UK)preparations revealed that the cells were >98% PMNL [1]. PMNLsurfaceglycoprotein expression. Cytocentrifugepreparationsof PMNL suspensions werepreparedand fixed in4%(wt/vol) paraformaldehyde in PBS,pH 7.1. The slideswerewashed three times in PBS, pH 7.3, and once in PBS, pH 7.3, containing0.8% (wt/vol)

JID 1991;163 (May)

TNF Augments CR3 Response of PMNL to E. coli

110 1991;163 (May)

1035

Table 2. Inhibition of binding of labeled Escherichia coli to cytocentrifuge preparations of neutrophils. Percentage inhibition of binding of strain Treatment

504 (NB2)

None D-mannose (1 %) Concanavalin A (250 ug/rnl) Normal human serum (20% vol/vol) Anti-complement receptor 1 (I: 100) Anti-complement receptor 3 (1: 100) NOTE.

Data are mean

± SD

504 (MVBM)

0 82.5 68.2 10.6 8.3 5.6

± 9.9 ± 10.2 ± 7.1 ± 8.6 ± 12.3

SC (NB2)

0 1.6 3.3 11.9 68.3 68.2

SC (MVBM)

0

± 6.4 ± 4.8 ± 12.6

± 13.2 ± 8.3

0.1 7.1 13.7 67.6 38.3

± 19.9 ± 10.0 ± 2.2 ± 9.4 ± 6.4

AB (NB2)

0

± 13.2 o ± 32.2 20.7 ± 9.0 2.5

53.4 36.7

± 18.0

± 11.8

0 3.6 11.8 17.3 50.6 69.9

± 14.7 ± 9.8

± 12.8

± 18.0 ± 1.4

from seven experiments. NB2 = nutrient broth no. 2; MVBM = modified Vogel-Bonner's medium.

Characterization of immunoreactive LTB4 releasedfrom PMNL.

Selected samples ofsupernatants (500 ILl) containing immunoreactive LTB4 identified in the RIA were injected ontoa Nucleosil C18 5-ILm reverse-phase column (25.4 em x 4.6mm) (Hichrom, Reading, UK) and eluted for 30 min in a methanol-water-acetic acid (65:35:0.1) mobile phase brought to pH5.6 with ammonium hydroxide. Theflow ratewas maintained at 1ml/min throughout each run, and l-ml fractions were collected. Each fraction was stored under argon at -20°C until analyzed. Duplicate samples of100 J.tl ofeluent were dried down under vacuum and resuspended inRIA buffer. The elution time of PMNL-generated immunoreactive LTB4 and integrated UV absorption (270 om) were compared with those ofauthentic LTB4 and [3H]LTB4 eluted in the same system. LTB4 eluted at 25.3 ± 1.7 min (n = 3), and its stereoisomer, 5S, 12S LTB4, eluted at 26.8 ± 1.5 min (n = 3). Statistics. Data are expressed as mean ± SD, and differences between treatments were analyzed using thetwo-tailed Mann-Whitney U test unless stated otherwise.

Results Bacterial surface expression. Strain 504 grown in NB2 expressed type 1 fimbriae (table 1). Electron microscopic (EM) observation of carbon grids demonstrated that >90 % of the bacterial cells were fimbriated. Similarly, strain SC grown in NB2 expressed P fimbriae (>85 % of the cells observed under EM). Strains AB and DF were nonfimbriated.

None of the strains grown in MVBM expressed fimbriae detectable by hemagglutination, IFA, or EM, but all possessed condensed capsular material. Bacterial adherence to PMNL. Radiolabeled strain 504 grown in NB2 to express type 1 fimbriae adhered, in a ratio equivalentto 50:1-100:1, to each neutrophilon cytocentrifuge slides prepared at a density of 5 X 104 PMNL/slide. This adherence was dependent both on the dose of bacteria and on the numbers of PMNL in the original suspension spun onto the slide. Nonfimbriated radiolabeled strain 504 (MVBMgrown)also adheredto PMNL but in significantly fewernumbers ("-'5:1 to 10:1 on PMNL preparations; P < .01). The adherenceof strain 504 expressingtype 1 fimbriaewas inhibited in the presence of 1% (wt/vol) D-mannose and by increasing dosesof ConA. The bindingof nonfimbriated strain 504, however, was unaffectedby either n-mannose or ConA (table 2). NHS (20% [vol/vol] in IFAbuffer) did not affect the binding of type l-fimbriated or nonfimbriated organisms to PMNL. In contrast, both anti-CRI and -CR3 antibodies inhibitedthe bindingof nonfimbriated strain504, whilethebinding of type l-fimbriated strain 504 was unaffected at the same antibody concentrations. Similarly, the bindingof the P fimbriated strain SC and the nonfimbriated strains AB and SC to the PMNL preparations wasalso independent of mannosecontaining glycopeptides but wasinhibitableby both the anti-CRt and anti-CR3 antibodies (table 2). Pretreatment of PMNL with doses of rhTNFa significantly increased the binding of the strains that did not express type I fimbriae (SC and AB) and of strain 504 (cultured to suppress type l-fimbrial expression) to the PMNL preparations in a dose-dependent manner, reaching a maximum of 50%100% at 10-9 M TNF. In the same seven experiments, however, the binding of strain 504 expressing type I fimbriae was not significantly increased by TNF pretreatment (figure 1). No differences in binding of either FITC-ConAor FITC-goat anti-mouse antiserum were observed after rhTNFa treatment of the cells. Neutrophil responses to E. coli strains. E. coli strain 504 grown in NB2 generated significantLTB4, B12 BP, and MPO release (P < .01, P < .02, P < .001, respectively; figures 2A, 3A, 4A) after incubation in KRPG at 37°C for 60 min com-

Downloaded from http://jid.oxfordjournals.org/ at New York University on July 22, 2015

10-9 M rhTNFa in KRPG or with KRPG alone for 20 min, 1 x 1()6 PMNL were incubated inKRPG at 37°C for upto 60 min alone or with 100 ILl oftheE. coli strains ata bacteria-to-cell ratio of100/1. Incubations were ended bycentrifugation at 11,000 g for1min, and the supernatants were taken to assay for myeloperoxidase (MPO), vitamin B12 binding protein (B12 BP), leukotriene B4 (LTB4), or lactate dehydrogenase, as described previously [1]. LTB4 radioimmunoassay. The radioimmunoassay (RIA) for LTB4 was a modified version of that described previously [19]. Briefly, 100ILl ofsupernatant from thecellstimulation was assayed induplicate using 100 ILl ofspecific antiserum [22] against standard dilutions of LTB4 from 10to 0.04 ng/ml (l ng to 4 pg/tube). Free LTB4 was separated from antibody bound byadsorption to dextrancoated charcoal (l % [wt/vol] dextran T70[Pharmacia], 1% [wt/vol] GSX charcoal [BDH, Poole, UK]) at4°Cand centrifugation at3500 g and 4°Cfor10min. Thesupernatant was decanted into4 mlofscintillation fluid (Optiphase MP; urn, Bromma, Sweden) andcounted in a scintillation counter (Rackbeta; LKB).

1036

Steadman et al.

•

250

**

*

200

110 1991;163 (May)

NoTNF

o

-11 TNF (log M)

* II

-10 TNF (log M)

~

-9

TNF (log M)

Figure 1. Adherence of radiolabeled Escherichia coli to cytocentrifuge preparations of neutrophils (PMNL) preincubated withtumor necrosis factor (TNF). E. coli wereaddedto slidesat 100:1. Results are percentageof adherentorganisms compared withthoseadherent in the absence of TNF (100 %) and are mean ± SD of seven experiments. NB = nutrientbrothno. 2; MVBM = modified VogelBonner's medium. *, P < .05; **, P < .01.

OJ

c

'6 c :0

150

Q)

OJ

co C

100

Q)

o (j) o,

50

o

A

504 (MVSM) No fimbriae

SC

AS No fimbriae

P fimbriae

NS2

B I2J II

0.6

-TNF +TNF

..J

**

*

Z

:E

MVSM

**

1.4

l2] -TNF

II

1.2

+TNF

0-

lD

o

1.0

~

.....

OJ c

Q)

(f)

co Q)

~

'it

co

'it (J)

...J

I-

I-

...J

PMNL 504 alone

sc

OF

AS

PMNL alone

pared with unstimulated controls. Growth of strain 504 in MVBM to suppress fimbrial expression resulted in a significantly reduced generation of LTB4 (P < .01), B 12 BP (P < .05), and MPO (P < .02; figures 2B, 3B, 4B). None of the other strains caused significant MPO release. Significant (P < .05) B 12 BP release from PMNL occurred with all four strains grown in nutrient broth and with strain 504 grown in MVBM. None of the other strains grown in MVBM, however, released significant amounts of B 12 BP above that released by unstimulated controls. Neutrophil responsesto rhTNFa. rhTNFa caused a doseand time-dependent increase in CR3 expression in parallel to the release of B I2 BP from PMNL, which reached a maximum twofold augmentation at 20 min with a dose of 10-9 M. There was no LTB4 , lactate dehydrogenase, or MPO release in response to any of the doses of rhTNFa used. E. coli stimulation of rhTNFa-pretreated PMNL. The LTB4 release from PMNL exposed to strains not expressing type 1 fimbriae (including strain 504 grown in MVBM) was synergistically augmented following rhTNFa preincubation.

504

sc

Figure 2. Leukotriene B4 (LTB4) release from neutrophils (PMNL) incubated with Escherichia coli grownin nutrient broth no.2 (NB2) (A) or modified Vogel-Bonner's medium (MVBM) (B). PMNLwere pretreated with or without 10-9 M recombinant humantumornecrosis factor-a (TNF). Results are mean ± SD for three experiments, each with PMNL from different donor. *, P < .05; **, P < .01.

AS

The response to type l-fimbriated strain 504 grown in NB2, however, was unaffected (figures 2A and 2B). The release of B 12 BP was also increased but in an additive manner compared with that released by PMNL incubated with either rhTNFa or bacteria alone (figures 3A and 3B). There was a small, nonsignificant augmentation by rhTNFa of the MPO release stimulated by all of the strains tested (figures 4A and 4B). In contrast, none of the PMNL responses stimulated by type l-fimbriated strain 504 was augmented by doses of rhTNFa (figures 2A, 3A, and 4A). There was no significant lactate dehydrogenase release above background in response to any strain used, either in the presence or absence of rhTNFa.

Discussion rhTNFa, at a concentration of 10-9 M, selectively augmented the neutrophil responses to uropathogenic strains of E. coli that were not expressing type 1 (mannose-specific) adhesins. The generation of LTB4 was increased in a syner-

Downloaded from http://jid.oxfordjournals.org/ at New York University on July 22, 2015

504 (NS) Type 1 fimbriae

TNF Augments CR3 Response of PMNL to E. coli

JID 1991;163 (May)

A

B

NS2

100

Figure 3. Vitamin B12 binding protein (B12 BP) release from neutrophils (PMNL) incubated with Escherichia coli grown in nutrient broth no. 2 (NB2) (A) or modified KJ Vogel-Bonner's medium (MVBM) ~ (B). PMNL were pretreated with or c, without 10-9 M recombinant hu- ~ man tumor necrosis factor-a (TNF). cD Results are mean ± SD for three rf. experiments, each with PMNL from different donor. *, P < .05; **,

80

(?2

- TNF

fI

+ TNF

*

(1)

~

20

504

sc

o 504

B

release from neutrophils (PMNL) incubated with Escherichia coli grown in nutrient broth no. 2 (NB) (A) ormodified Vogel-Bonner's medium (MVBM) (B). PMNL were pretreated with or without 10-9 M recombinant human tumor necrosis factor-a (TNF). Results are mean ± SD for three experiments, each with PMNL from different donor.

24

sc

II

AS

MV~

12

E]

I::F

l2] -TNF +TNF

- TNF + TNF

II

20

8

16

o

12

ll.. ~

8

4

4

o

PMNL alone

504

sc

gistic fashion, in manycases from undetectablelevels «0.04 ng/l x !Q6 PMNL). There was also a small, though nonsignificant, increase in MPO release after rhTNFa pretreatment. In addition, B12 BP release was additively increased above that of controls. These results suggest that a specific cell surface-bacterium interactionwasaugmented in PMNL after rhTNFa pretreatment and that the response was selective. This augmentation was not inhibited by increasing the capsuleexpression ofthese strainsbygrowth in MVBM. Capsular polysaccharide is believed to inhibit leukocyte activation through a physicochemical charge interaction with the cell surface [23, 24]. After pretreatmentwith rhTNFa, however, the LTB4 response of PMNL to encapsulated strains was augmented, emphasizing the potential of this cytokine for selective amplification of the inflammatory response. Besidescontributing to the overall surfacechargeofthe cell membrane, many of the glycoproteins on the PMNL surface function as specific receptors. These mediate adherence to different substrates and to other cells [25, 26] or function as receptors for solubleligandssuchas formylated peptides, immunoglobulins, and complement components [11, 12,27,28].

OF

AB

o

PMNL alone

504

sc

AB

Both CRI and CR3 have been shown to be involved in the binding of E. coli strains in addition to acting as receptors for the complementcomponents C3b and iC3b, respectively [9]. CR3 is probablyalso involved in the bindingof zymosan to PMNL [12, 29, 30]. PMNL activation by the type l-fimbriated strain 504, however, was not augmented by pretreatment with rhTNFa at doses that augmented the responses to strains not expressing type 1 fimbriae despite the PMNL-to-E. coli ratio being suboptimal [1]. Thesedoses of rhTNFa have also been shownto be optimal for the augmentation of PMNL responses to zymosan [14, 19]. Since FITC-conjugated antibody binding andFITC-conjugated ConA bindingto the PMNL cytocentrifuge preparationswasnot affected byrhTNFa pretreatment, it is unlikely thatTNF affected the number of type l-fimbriated binding sites or IgO receptors [31]. Thus, our results suggestthat the observed selective modulation of PMNL responsesmaybe CR3-dependent and that type l-fimbriatedorganisms activate the cell through an alternative mannose-dependent interaction. Olycoproteins of 150, 100, and 70-80 kDa on the PMNL membrane have beenimplicated in the mannose-specific bind-

Downloaded from http://jid.oxfordjournals.org/ at New York University on July 22, 2015

PMNL alone

AS

I::F

A

~ ~ rf.

40

cD

PMNL alone

28

*

('II

20

Figure 4. Myeloperoxidase (MPO)

-TNF +TNF

*

en

40

ELI

rIA 60

(1) (f)

c,

o

~

80

(1)

60

P< .01.

m

MVBM

ell

(1)

~

1037

1038

Steadman et al.

Acknowledgment

Wethank A. Roberts (Departmentof MedicalMicrobiology, Charing Cross Hospital, London) for serotyping the strains used in this study, the Departmentof Hematology (University Hospitalof Wales,

Cardiff)for FACS analyses,and Cheryl Patterson for help in manuscript preparation.

References 1. Steadman R, Topley N, Jenner DE, Davies M, Williams JD. Type 1 fimbriate Escherichia coli stimulatesa unique pattern of degranulation by human polymorphonuclearleukocytes. Infect Immun 1988; 56:815-22. 2. Topley N, SteadmanR, MackenzieRK, Knowlden J, WilliamsJD. Type 1 fimbriatestrainsof Escherichia coli initiaterenalparenchymalscarring. Kidney Int 1989;36:609-16. 3. Blumenstock E, Jann K. Adhesionof piliated Escherichia coli strains to phagocytes: differences between bacteria with mannose-sensitive pili and those with mannose-resistant pili. Infect Immun 1982;35: 264-9. 4. SvanborgEden C, Bjorksten LM, Hull R, et al. Influence of adhesins on the interactionof Escherichia coli with human phagocytes.Infect Immun 1984;44:672-80. 5. Harber MJ, Topley N, Jenner DE, et al. Virulence factors of urinary pathogens in relation to kidney scarring. In: Asscher AW, Brurnfitt W, eds. Microbial diseases in nephrology. Chichester: 1. Wiley & Sons, 1986:69-82. 6. Rodrigues-Ortega M, Ofek I, Sharon N. Membrane glycoproteins of human polymorphonuclear leukocytes that act as receptors for mannose-specific Escherichia coli. Infect Immun 1987;55:968-73. 7. Wright SD, Jong MTC. Adhesion-promoting receptors on human macrophages recognize Escherichia coli by binding to lipopolysaccharide. J Exp Med 1986;164:1876-88. 8. Salmon JE, Kapur S, Kimberley RP. Opsonin-independentligation of Fc g receptors. The 3G8 receptors on neutrophilsmediatethe phagocytosis of concanavalin A-treated erythrocytes and non-opsonized Escherichia coli. J Exp Med 1987;166:1798-813. 9. Fine DP, Harper BL, Carpenter ED, Davis CP, CavalloT, Kukian rc Complement-independent adherenceof Escherichia coli to complement receptors in vitro. J Clin Invest 1980;66:465-72. 10. Fitzharris P, Cromwell 0, Moqhel R, et al. Leukotriene B4 generation by human neutrophils following IgG-dependentstimulation. Immunology 1987;61:449-55. 11. Williams JD, Lee TH, Lewis RA, Austen KF. Intracellular retention of the 5-lipoxygenase pathway product leukotrieneB4 by human neutrophils activatedwith unopsonizedzymosan. J ImmunoI1985;134: 2624-30. 12. WilliamsJD, Topley N, AlobaidiHM, Harber MJ. Activationof human polymorphonuclear leukocytesby particulate zymosan is related to both its major carbohydrate components: glucan and mannan. Immunology 1986;58:117-24. 13. KlebanoffSJ, Vadas MA, Harlan JM, et al. Stimulationof neutrophils by tumor necrosis factor. J Immunol 1986;136:4220-5. 14. Petersen M, SteadmanR, MatthewsN, WilliamsJD. Zymosaninduced leukotriene B4 generation by human neutrophils is augmented by rhTNFa but not chemotactic peptide. Immunology 1990;70:75-81. 15. BerkowRL, DodsonMR. Biochemical mechanisminvolved in the priming of neutrophils by tumor necrosis factor. J Leukoc BioI 1988; 44:1345-52. 16. Ferrante A, Nandoskar M, Walz A, Goh DHB, Kowanko rc Effects of tumor necrosis factor alpha and interleukin 1 alpha and beta on human neutrophilmigration, respiratory burst and degranulation. Int Arch Allergy Appl Immunol 1988;86:82-91. 17. BerkowRL, WangD, Larrick JW, Dodson RW, HowardTH. Enhancementof neutrophil superoxide production by preincubation withrecombinant human tumor necrosis factor. J ImmunoI1987;139:3783-91. 18. Berger M, WatzlerEM, WallisRS. Thmor necrosis factor is the major

Downloaded from http://jid.oxfordjournals.org/ at New York University on July 22, 2015

ing of type 1 fimbriae [6], and type l-fimbriated E. coli (in contrast to non-type l-fimbriated organisms) bind in a mannose-inhibitable manner to the Fc receptors of phagocyticcells [8]. Previous studies havedemonstrated that type l-fimbriated E. coli bind in a mannose-inhibitable manner to the Fc-y III receptor on PMNL and that Fc receptor cappingdoesnotaffect the binding of E. coli thatare not expressingtype 1fimbriae [7, 8]. In addition,CRI blocking hasbeen shown to inhibitthe binding of non-type l-fimbriatedorganisms, andlipopolysaccharide in thebacterialoutermembrane may mediatethis CRI binding as well as the binding to the CDIlICDI8 glycoproteins [7]. Our presentfindings confirmand extendthese data by examining in detail the activation of PMNL after E. coli binding to thesesurface receptorsand the selective augmentation of neutrophil responses by rhTNFa The results demonstrate that E. coli thatdo not express type 1adhesins bindto PMNL complement receptors in a mannerinhibitable by specific antibodies to the receptors and that the binding stimulates LTB4 synthesis as well as the release of secondary lysosomal contents. The expression of CR3 was increased after rhTNFa priming of the cells, and the LTB4 response to non-type l-fimbriated organisms was synergistically augmented. A strain of E. coli expressing type 1 fimbriae, however, bound to mannose-containing glycoproteins on thePMNLandstimulated the releaseof the primary granule contents in addition to that of LTB4 • The PMNL responses to this strain, unlike thoseofthe non-type l-fimbriatedstrains, however, werenot increased after TNF pretreatment. Theseresults furtherdemonstrate distinct intracellular pathways of neutrophil activation as previously describedfor other particulate stimuli[12, 19,32]. The mechanism of primingof PMNL byTNF is still not clear, although it nowseemslikely that intracellular arachidonic acid metabolites, calciuminflux,andcyclic nucleotidesare not involved [14]. Proteinkinaseactivation has been suggested, although to date the identity of the proteins specifically phosphorylated in response to TNF is notknown [15]. In conclusion, these results support the viewthat the neutrophildoes not respondin an identicalmanner to all phagocyticstimuli. Theprimary(azurophil) granuleenzyme release in responseto E. coli seemsto involve Fc receptor binding, is activated by few stimuli, and is not increasedeither in response to direct activation or after priming by rhTNFa. In contrast, the PMNL responses to non-type l-fimbriated E. coli binding to complement receptors are significantly augmentedbyprimingwith rhTNFa. Thus, the different inflammatory pathways withinPMNLare independently controlled and independently primed by rhTNFa.

JID 1991;163 (May)

JID 1991;163 (May)

19.

20.

21.

22.

23.

25.

monocyte productthat increasescomplement receptorexpressionon mature human neutrophils. Blood 1988;71:151-8. SteadmanR, PetersenMM, Topley N, et al. Differential augmentation by recombinant humantumornecrosis factor-aof neutrophil responses to particulate zymosan and glucan. J Immunol 1990;144:2712-8. Pere A, NowickiB, Saxen H, Siitonen A, Korhonen TK. Expression of P, type-l, and type-1C fimbriae of Escherichia coli in the urine of patients with acute urinary tract infection. J Infect Dis 1987;156: 567-74. Lichodziejewska M, Topley N, SteadmanR, et al. Variable expression of P-fimbriae in Escherichia coli urinary tract infection. Lancet 1989;1:1414-8. Rokach J, HayesEC, Girard Y,et al. The development of sensitiveand specificradioimmunoassays for leukotrienes. Prostaglandins Leukot Essent Fatty Acids 1984;13:21-5. (/)hman L, Hed J, Stendahl 0. Interaction between human polymorphonuclear leukocytes and two different strains of type 1fimbriae-bearing Escherichia coli. J Infect Dis 1982;146:751-7. Steadman R, Knowlden JM, Lichodziejewska M, Williams JD. The influence of net surface charge on the interactionof uropathogenic Escherichia coli with human neutrophils. Biochim Biophys Acta 1990;1053:37-42. Miller U, BaintonDF, BorregaardN, SpringerTA. Stimulated mobilization of monocyte MAC-l and p150-95 adhesin proteins from an intracellularvesicularcompartmentto the cell surface.J Clin Invest 1987;80:535-44.

1039

26. Zimmerman GA, McIntyre TM. Neutrophil adherence to human endotheliumin vitro occursby CDW18 (Mol, MAC-1/LFA-1/gp150,96) glycoprotein-dependent and independent mechanisms. J Clin Invest 1988;81:531-7. 27. Ohta H, OkajimaF, Machio U. Inhibitionby islet-activating proteinof chemotactic peptide-induced early breakdown of inositol phospholipids and Ca2+ mobilization in guineapig neutrophils. J BioIChern 1985;260:15771-80. 28. Mantovani B. Different rolesforIgGandcomplement receptors inphagocytosis by polymorphonuclear leukocytes. J lmmunoI1975;115:15-21. 29. RossDG, CainJA, Lachmann PI Membrane complement receptortype 3 (CR3) has lectin-like properties analogous to bovine conglutinin and functions as a receptor for zymosan and rabbit erythrocytes as well as receptor for iC3b. J Immunol 1985;134:3307-15. 30. RossGD,CainJA, MyonesBL, NewmanSL, LachmannPI Specificity of membranecomplement receptor type three (CR3) for l3-glucans. ComplementInflamm 1987;4:61-74. 31. PerussiaB, Kobayashi M, RossiME, AnegonI, TrinchieriG. Immune interpheronenhancesfunctional properties of human granulocytes: roleof Fcreceptorsandeffectof lymphotoxin, tumor necrosisfactor, and granulocyte-macrophage colony-stimulating factor. J Immunol 1987;134:765-74. 32. Topley N, Steadman R, PetersenMM, Spur B, Williams JD. Phagocytosis, the respiratory burst and the 5-lipoxygenase pathway are independently regulated in humanpolymorphonuclear leukocytes. Adv Biosci 1987;66:123-9.

Downloaded from http://jid.oxfordjournals.org/ at New York University on July 22, 2015

24.

TNF Augments CR3 Response of PMNL to E. coli