Kidney International, Vol. 59 (2001), pp. 893–904

CELL BIOLOGY – IMMUNOLOGY – PATHOLOGY

Escherichia coli-induced inducible nitric oxide synthase andcyclooxygenase expression in the mouse bladder and kidney

MIRJANA POLJAKOVIC, MAJ-LIS SVENSSON, CATHARINA SVANBORG, KJELL JOHANSSON,BENGT LARSSON, and KATARINA PERSSON

Department of Clinical Pharmacology; Department of Microbiology, Immunology and Glycobiology;and Department of Ophthalmology, Lund University Hospital, Lund, Sweden

Escherichia coli-induced inducible nitric oxide synthase and The inducible forms of nitric oxide synthase (iNOS)cyclooxygenase expression in the mouse bladder and kidney. and cyclooxygenase (COX-2) are two enzymes involved

Background. The host response to urinary tract infection in inflammatory responses [1]. iNOS can be induced inincludes the production of different inflammatory mediators.

a wide variety of human cells, including monocytes/mac-We investigated the cellular localization and time course ofrophages [2], hepatocytes [3], and vascular smooth mus-inducible nitric oxide synthase (iNOS) and cyclooxygenase

(COX-2) expression in the mouse bladder and kidney after cle cells [4], and may cause the production of sustained,bacterial infection. high levels of nitric oxide (NO). Accumulating evidence

Methods. Experimental urinary tract infection in mice was indicates that excessive production of NO plays a patho-established by intravesical inoculation of a clinical uropathogen

genic role in both acute and chronic models of inflamma-Escherichia coli (E. coli) AD 110. Urine was collected at 6-, 12-,tion [5]. This has led many investigators to examine the24-, and 72-hours postinstillation, and the nitrite concentrationrole of iNOS and NO in pathophysiological conditions.was determined. The induction of iNOS and COX-2 was stud-

ied by immunohistochemistry and reverse transcription-poly- Prostaglandins, produced by COX, are biological me-merase chain reaction (RT-PCR). diators in all stages of inflammation [6]. The COX en-

Results. Nitrite levels in the urine had increased threefold zyme exists in two isoforms [7], of which COX-1 is pres-at 6 and 12 hours postbacterial instillation. Bladders from mice

ent constitutively in various types of cells, includinginstilled with AD 110, but not with phosphate-buffered saline,vascular endothelial cells, platelets, and cells in the kid-showed a large number of iNOS– and COX-2–expressing in-ney and gastrointestinal tract [8]. COX-2 expression isflammatory cells. The inflammatory cell activation peaked at

6 and 12 hours postinstillation and had vanished by 72 hours. rapidly induced by numerous proinflammatory stimuli,iNOS expression was detected in some urothelial cells after 24 including endotoxin, growth factors, interleukin-1 (IL-1),and 72 hours, but COX-2 expression was not detected. In the and tumor necrosis factor-a (TNF-a) and is thought tokidney, infection activated an iNOS and COX-2 response, as

be involved in inflammation, cellular differentiation, andshown by immunoreactivity in inflammatory cells at all timemitogenesis [1, 8]. The induction of COX-2 transcriptionpoints. A strong epithelial iNOS response was observed in the

renal pelvis at 12, 24, and 72 hours postinstillation, but COX-2 is rapid, transient, and like iNOS, inhibited by glucocorti-was not detected. Enhanced tissue expression of iNOS and coids like dexamethasone [1].COX-2 after bacterial instillation was also demonstrated by Nitric oxide and prostaglandins may work coopera-RT-PCR. tively and synergistically in both physiological and patho-Conclusions. E. coli AD 110 induced expression of iNOS

logical conditions [1]. During the inflammatory responseand COX-2 in the urinary tract. Inflammatory cells expressedto infection, endotoxins and cytokines induce rapid up-both iNOS-and COX-2, but epithelial cells expressed only

iNOS and with a later onset than in the inflammatory cells. regulation of COX-2 [6] and iNOS [5]. Coinduction ofThis suggests that the epithelial iNOS response is not caused iNOS and COX-2 has been shown to occur in severalby direct bacterial activation, but more likely is by mediators cell types, including murine macrophages [9] and smoothinvolved in the inflammatory response.

muscle cells [10]. Lately, it has become evident that theinteraction between the COX and NOS pathways is im-portant in the regulation of the inflammatory process.Key words: prostaglandins, urinary tract infection, inflammatory re-

sponse, bacterial infection. Hence, endogenously formed NO increased the prosta-glandin biosynthesis in in vivo and ex vivo models ofReceived for publication October 25, 1999inflammation and endotoxic shock [9, 11], and inhibitionand in revised form August 24, 2000

Accepted for publication September 14, 2000 of NO synthesis by NOS inhibitors caused a reductionof prostaglandin production in vivo and in vitro [9, 12, 2001 by the International Society of Nephrology

893

Poljakovic et al: iNOS and COX-2 in E. coli-infected mice894

13]. In contrast, high concentrations of NO may inhibit cultured on TSA overnight at 378C, harvested in sterilephosphate-buffered saline (PBS; pH 7.4), centrifuged atthe induction of COX-2, as reported in cultured murine

macrophages when stimulated with lipopolysaccharide 3200 r.p.m. for 20 minutes, and resuspended in ster-ile PBS to a final concentration of approximately 109(LPS) [14].

Urinary tract infections (UTIs) are among the most CFU/mL.common bacterial infections in humans, and most UTIs

Infection procedureare caused by Escherichia coli (E. coli). The urothelialcells are the first cells to make contact with invading The mouse bladder was emptied by gentle compres-

sion on the lower abdomen, and urine was saved forbacteria, and the urothelium is triggered to produce vari-ous cytokines that attract cells of the innate immune preinfection measurements of nitrite (discussed later

in this article). Infection was induced by intravesicalsystem to the site of infection [15]. The bacteria arecleared from the urinary tract through the action of injection of E. coli as previously described [19]. After

anesthesia, 0.1 mL of bacterial suspension or PBS wasinflammatory cells, especially by polymorphonuclear(PMN) cells [16]. Previously, it has been demonstrated slowly instilled into the bladder, transurethrally, through

a soft polyethylene catheter (outer diameter 0.61 mm;that patients with UTI have an elevated nitrite concen-tration and iNOS activity in the urine compared with KEBOLab, Malmo, Sweden). The catheter was immedi-

ately withdrawn after inoculation, and no further manip-healthy controls [17]. Elevated gaseous NO concentra-tions in urinary bladder in patients with lower UTI have ulations were carried out. The animals were placed in

the cages after instillation and allowed food and wateralso been reported [18]. However, the cell types responsi-ble for the induction of iNOS have not been systemati- ad libitum.

To exclude artifacts caused by the instillation proce-cally investigated, and furthermore, the time course foriNOS expression has not been established. Likewise, in- dure, some mice were sacrificed without any treatment

and served as an additional control. Urine samples ob-formation on COX-2–expressing cells in the host defenseagainst UTIs is lacking, and its role in UTI remains tained before sacrifice were stored on ice or at 48C until

further analysis. The mice were sacrificed at 6, 12, 24,unclear.The present study investigated the cellular localization and 72 hours postinstillation by carbon dioxide (CO2)

asphyxia. The bladders and kidneys were removed andand time course of iNOS and COX-2–immunoreactivecells in the bladder and kidney in mice that were trans- processed for immunohistochemistry as described later

in this article.urethrally challenged with a uropathogenic strain ofE. coli.

Nitrite assay

Nitric oxide production was measured as the urineMETHODS

nitrite levels using the Griess reaction assay [20]. Briefly,Mice 10 mL of centrifuged urine were transferred to 96-well

plates and mixed with 20 mL of water and 100 mL ofC3H/HeN mice were purchased from Møllegaard &Bomholtgard (Ry, Denmark). Female mice were used Griess reagent [one part 0.1% N-(1-naphtyl) ethylenedi-

amine dihydrochloride in water and one part 1% sulfanil-at 9 to 10 weeks of age. The mice were kept on a nitrate/nitrite-free pellet diet (Altromin 1324N; Chr. Petersen amide in 5% concentrated H3PO4 (both purchased from

Sigma Chemical Co., St. Louis, MO, USA)]. The mixtureA/S, Ringsted, Denmark) and sterile water, starting oneweek before the experiment. A few days prior to infec- was incubated for five minutes in room temperature and

read at 540 nm by spectrophotometry (Labsystemstion, groups of five mice were separated into individualcages, and urine samples were collected and examined Multiscan PLUS; Labsystems AB, Lund, Sweden). Con-

centrations were determined in comparison to a standardfor bacterial growth, and the PMN cell content was deter-mined in a hemocytometer chamber. Mice with more curve (dilutions of 100 mmol/L NaNO2). The lower de-

tection limit of the assay was 1 mmol/L nitrite.than 50 3 104 leukocytes/mL in preinoculation urinesamples or with a positive bacterial culture were ex-

Immunohistochemistrycluded from the experiment. The experimental protocolwas approved by the Animal Ethics Committee (Lund Bladders and kidneys were immersion-fixed for four

hours in ice-cold 4% formaldehyde in PBS (pH 7.4)University, Lund, Sweden).and then rinsed for three days in PBS containing 15%

Bacteria sucrose. Both fixation and rinsing were performed at 48C,after which the specimens were frozen in isopenthane atEscherichia coli AD 110 is a clinical isolate that ex-

presses type 1 and P-fimbrial adhesions. The bacteria 2408C and stored at 2708C until sectioning. Sectionswere cut (10 mm) in a cryostat (Leica CM3050; Leicawere maintained on tryptic soy agar (TSA; Difco Labora-

tories, Detroit, MI, USA). For infection, bacteria were Microsystems AB, Askik, Sweden) and preincubated

Poljakovic et al: iNOS and COX-2 in E. coli-infected mice 895

with PBS containing 0.2% Triton X-100 and 0.1% bovine merase chain reaction (RT-PCR) was performed ac-serum albumin (BSA) for two hours at room tempera- cording to the Perkin Elmer PCR-kit (GeneAmptRNAture. Sections were incubated with primary antisera (di- PCR kit; Perkin Elmer, Foster City, CA, USA) usingluted with PBS containing 0.2% Triton X-100 and 0.1% 2 mg total RNA, oligo-dT as the first-strand primer andBSA): rabbit polyclonal antibody raised to murine iNOS MuLV reverse transcriptase according to manufacturer’s(1:1000; Santa Cruz Biotechnology, Inc., Santa Cruz, CA, instructions. The primers for mouse iNOS were obtainedUSA) or rabbit polyclonal antiserum raised to murine from DNA technology (DNA Technology Aps, AarhusCOX-2 (1:400; Cayman Chemicals, Ann Arbor, MI, C, Denmark) as follows: iNOS sense, 59-CTT CCG AAGUSA) overnight in a moisture chamber at room tempera- TTT CTG GCA GCA GCG-39; iNOS antisense,ture. The sections were rinsed in PBS and incubated 59-GAG CCT CGT GGC TTT GGG CTC CTC-39, am-for 90 minutes with fluorescein isothiacyanate (FITC)- plifying a 487 bp product. The primer set for mouseconjugated donkey anti-rabbit IgG (1:80) or Texas Red COX-2 was ordered from Oxford Biomedical Research,(TR)-conjugated F(ab9)2 fragment donkey anti-rabbit Inc. (Oxford, MI, USA) and amplified a 724 bp product.IgG (1:160) diluted in PBS (Jackson Immunoresearch PCR was performed in an automated thermal cyclerLaboratories, Inc., West Grove, PA, USA). The sections (Gene Amp 2000; Perkin-Elmer), with one cycle at 958Cwere rinsed in PBS and mounted in glycerol with for 2 minutes, followed by 35 cycles at 958C for 60 sec-p-phenylenediamine to prevent fading. onds, at 558C (iNOS) or 538C (COX-2) for 60 seconds,

To identify the inflammatory cells, a double-label im- at 728C for 60 seconds, and a final extension at 728C formunofluorescence method was used. Sections were first 7 minutes. RNA isolated from LPS-stimulated mouseincubated with RB6-8C5, a rat monoclonal antibody spe- macrophages (RAW 264.7) were run in parallel as posi-cific for murine neutrophils (a kind gift from Dr. A. tive controls. PCR products were separated by 2% aga-Sjostedt, Umea University, Sweden) overnight. After rose gel electrophoresis, and bands were visualized byrinsing in PBS, the iNOS or COX-2 antibody was added, ethidium bromide staining.and the sections incubated again overnight. The RB6-8C5 staining was visualized by incubating the sections for Analysis of data90 minutes with TR-conjugated F(ab9)2 fragment donkey Data are expressed as mean 6 SEM and were com-anti-rat IgG (1:160; Jackson Immunoresearch Labora- pared by an unpaired Student t test. Differences weretories, Inc.). After rinsing, the sections were incubated considered significant at P values , 0.05.for 90 minutes with (FITC)-conjugated donkey anti-rab-bit IgG (1:80) to visualize iNOS or COX-2. The sectionswere mounted as previously described. RESULTS

In control experiments, no immunoreactivity was de- Nitrite levels in urinetected in sections incubated with the secondary antibody.

The nitrite concentrations in preinoculation urine sam-The specificity of the iNOS antibody was confirmed byples did not differ (P 5 0.217) between the groups (PBS;incubating the antibody with an excess of antigen. No14 6 2 mmol/L, N 5 15; AD 110; 18 6 2 mmol/L, N 5iNOS immunoreactivity was detected in sections incu-23). The time-dependent nitrite accumulation in urinebated with absorbed antibody. All micrographs of theof mice infected with E. coli AD 110 is summarizedimmunolabeled sections were obtained using a digitalin Figure 1A. Elevated nitrite levels were observed atcamera system (Nikon E400 microscope and Optronix6- and 12-hours postinstillation. At six hours, a threefoldDEI-750 camera), and the pictures were captured using(53 6 19 mmol/L) increase in nitrite accumulation wasappropriate filter settings for FITC and TR. Adobetfound when compared with mice instilled with PBS onlyPhotoshope was used for image handling, and the three-(18 6 4 mmol/L, P 5 0.17). After the initial six hourcolor channels were handled separately. Only the back-peak, the nitrite concentration in the urine of infectedground level, contrast and brightness of the entire imagemice remained elevated for up to 72 hours, but the differ-were changed in the final picture.ence was not significant compared with time-matched PBS

Hematoxylin and eosin staining controls. The mean nitrite levels in mice instilled withPBS were approximately 15 mmol/L, and no changes inThe morphology of the inflammatory cells was exam-nitrite levels over time were noted in this group. (Fig. 1B).ined after staining of the tissue sections with hematoxylin

and eosin (Apoteksbolaget, Malmo, Sweden).Bladder immunohistochemistry

Reverse transcription-polymerase chain reaction Bladders obtained from infected mice were examinedfor iNOS and COX-2 expression by immunohistochemis-The bladders and kidneys were homogenized in TRI-try at 6, 12, 24, and 72 hours postinstillation and com-zol reagent (Life Technologies AB, Taby, Sweden), and

total RNA was extracted. Reverse transcription-poly- pared with the PBS controls. The results are presented

Poljakovic et al: iNOS and COX-2 in E. coli-infected mice896

Table 1. Kinetics of the inducible nitric oxide synthase (iNOS) andcyclooxygenase-2 (COX-2) response to infection in the bladder

iNOS COX-2

Inflammatory Urothelial Inflammatory UrothelialStimulus cells cells cells cells

PBS 2 2 2 2E. coli 1 6, 12 (24) 1 24, 72 1 6, 12 (24) 2

Symbols indicate the presence (1) and absence (2) of the enzymes. Thetimes without a parenthesis indicate that many cells (massive infiltration) wereobserved and the times within the parenthesis indicate that a few inflammatorycells were observed. The times given indicate the time after instillation.

flammatory cells were observed in the smooth musclelayer, although they were mainly found in the submu-cosa, around vessels, and close to the urothelium. Someinflammatory cells were seen crossing the urotheliuminto the bladder lumen (Figs. 2D and 3D). At 24 hourspostinstillation, the number of iNOS- and COX-2–expressing inflammatory cells had decreased, and thecells were usually located in the submucosa. At 72 hourspostinstillation, very few of the iNOS- and COX-2–expressing inflammatory cells remained. A similar distri-bution of iNOS- and COX-2–expressing inflammatorycells was seen when investigating consecutive sections(Figs. 2B and 3B). Hematoxylin and eosin staining ofthe bladder tissue sections showed that the majority ofthe inflammatory cells in the bladder were PMN cells. Bydouble-label immunofluorescence, using the neutrophilmarker RB6-8C5, practically all (.90%) of the inflam-matory cells in the bladder that were positive for iNOSor COX-2 were identified as neutrophils (Fig. 4 A–C).iNOS-positive urothelial cells were found in 50% of thebladders at 24 and 72 hours (Fig. 2 E, F). The iNOS-positive urothelial cells were mainly found in the apicaland medial parts of the urothelium and in a nonuniformpattern. Some urothelial cell shedding was also seen(Fig. 2F). The bladder urothelial cells did not expressCOX-2 at any time.

The bladders were analyzed by RT-PCR 6 and 24hours after bacterial instillation. The expression of iNOSand COX-2 mRNA was enhanced in bladders fromFig. 1. Nitrite concentration in individual urine samples obtained afterE. coli AD 110-infected mice when compared with PBS-instillation of (A) E. coli AD 110 and (B) phosphate-buffered saline

(PBS). The preinoculation nitrite values are given at time 0 (h) and instilled mice (Fig. 5). iNOS and COX-2 PCR productsare expressed as mean 6 SEM (N 5 15 to 23).

in the bladder were of the expected size and of the samesize as iNOS and COX-2 produced by LPS-stimulatedmouse macrophages (positive controls).

in Table 1 as positive/negative immunolabeling in uro-Kidney immunohistochemistrythelial cells and inflammatory cells.

Kidney sections obtained from infected mice were ex-Mice instilled with PBS or mice sacrificed without anyamined for iNOS and COX-2 expression by immunohis-instillation treatment did not show any iNOS (Fig. 2A)tochemistry 6, 12, 24, and 72 hours postinstillation andor COX-2 (Fig. 3A) immunoreactivity in the bladder.compared with the PBS controls. The results are pre-Numerous iNOS (Fig. 2 B, C) and COX-2 (Fig. 3 B,sented in Table 2 as positive/negative immunolabelingC)–expressing inflammatory cells were found in bladders

of infected mice at 6 and 12 hours postinstillation. In- in various parts of the kidney and in inflammatory cells.

Poljakovic et al: iNOS and COX-2 in E. coli-infected mice 897

Fig. 2. Immunohistochemistry of inducible nitric oxide synthase (iNOS) in mouse bladder sections. (A) PBS-instilled control bladders did notshow any iNOS immunoreactivity, U (urothelium). (B and C) iNOS-positive inflammatory cells (arrows) are seen at 6 and 12 hours postinstillationwith E. coli. (D) iNOS-positive cells (arrows) crossing through the urothelium. (E and F ) iNOS-positive urothelial cells (arrowheads) at 72 hourspostinfection. (F) Some urothelial cell shedding was observed. Scale bars: A–C 5 30 mm; D 5 10 mm; and E and F 5 15 mm.

No iNOS-positive (Fig. 6A) or COX-2–positive (Fig. immunoreactivity was also observed in the tubular epi-thelial cells in the outer parts of the cortex.7A) inflammatory cells were found in the kidneys of

PBS-instilled mice. The PBS controls and the non- Inducible NOS- and COX-2–positive inflammatorycells were found in kidneys of E. coli-infected mice atinstilled control mice showed iNOS immunolabeling in

tubular epithelial cells of the outer medulla and inner all times. The inflammatory cells were mainly found closeto the epithelium lining the renal pelvis (Figs. 6C andcortex at all times (Fig. 6B). In some PBS controls, iNOS

Poljakovic et al: iNOS and COX-2 in E. coli-infected mice898

Fig. 3. Immunohistochemistry of cyclooxygenase-2 (COX-2) in mouse bladder sections. (A) No COX-2 immunoreactivity was detected in thecontrol bladders. The COX-2 antibody displayed some background labeling in the urothelium (U). (B and C) COX-2–positive inflammatory cells(arrows) are seen at 6 and 12 hours postinstillation with E. coli. (D) COX-2–positive cells (arrows) crossing through the urothelium. Scale bars:A–C 5 30 mm and D 5 10 mm.

7B). The majority of the iNOS and COX-2–positive in- The kidneys were analyzed by RT-PCR 6 and 24 hoursflammatory cells was identified as neutrophils by the after bacterial instillation. The expression of iNOSRB6-8C5 antibody (Fig. 4 D–F). Distinct staining for mRNA was enhanced in kidneys from E. coli AD 110-iNOS was found in the transitional and columnar epithe- infected mice when compared with PBS-instilled micelial cells lining the renal pelvis at 12, 24, and 72 hours (Fig. 5). Bacterial instillation caused detectable increasespostinstillation (Fig. 6 C, D). However, renal epithelial in COX-2 mRNA expression in a few, but not all, kidneycells in infected mice did not express COX-2 (Fig. 7B). samples (Fig. 5).At 72 hours postinstillation, a positive iNOS labelingwas observed in the tubular epithelial cells in the outer-

DISCUSSIONmost parts of cortex in the majority of the infected miceBacterial colonization of the urinary tract is counter-(Fig. 6E). Glomeruli were devoid of iNOS immunoreac-

acted by the urine flow and secreted receptor analogues,tivity. Similar to PBS-instilled mice, E. coli-infected micebut also by shedding of urothelial cells with attachedshowed iNOS immunolabeling in the tubular epithelial

cells of the outer medulla/inner cortex at all times. bacteria [15]. If the bacteria succeed to colonize, they

Poljakovic et al: iNOS and COX-2 in E. coli-infected mice 899

Fig. 4. Double immunolabeling of COX-2 (A, green fluorescence) and the neutrophil marker RB6-8C5 (B, red fluorescence) in the same bladdersection six hours after E. coli infection. (C) Note that practically all COX-2–positive inflammatory cells are identified as neutrophils (double-positive cells, yellow fluorescence). Double immunolabeling of COX-2 (D, green fluorescence) and the neutrophil marker RB6–8C5 (E, redfluorescence) in the same kidney section 24 hours after E. coli infection. (F ) Arrows show COX-2–positive inflammatory cells that were identifiedas neutrophils (double-positive cells, yellow fluorescence). Arrowheads show neutrophils that do not stain for COX-2. Scale bar 5 10 mm.

activate epithelial cells to produce inflammatory media- tious agents and contribute to tissue damage and in-creased microbial virulence [21]. In the present study,tors and cause an inflammatory response involving neu-

trophils and other cell types. The magnitude of the in- we examined the expression of iNOS and COX-2 in themouse urinary tract following intravesical instillation offlammatory response is determined by the virulence of

the invading microbe and the ability of the host to re- a uropathogenic E. coli strain.The production of NO into the urine in response tospond. The contribution of NO to the infectious process

is complex. NO may both protect the host against infec- infection was quantitated as nitrite, a stable end product

Poljakovic et al: iNOS and COX-2 in E. coli-infected mice900

Fig. 5. Reverse transcription-polymerase chain reaction (RT-PCR) analysis of iNOS and COX-2 mRNA expression in the mouse bladder (upperpanel; lanes 1 through 3) and kidney (lower panel; lanes 4 through 9). Lane 1 and lanes 4 and 5 show control treatment (PBS instillation for 24hours). Lane 2 and lanes 6 and 7 show expression of iNOS and COX-2 six hours after E. coli infection. Lane 3 and lanes 8 and 9 show expressionof iNOS and COX-2 24 hours after E. coli infection.

Table 2. Kinetics of the iNOS and COX-2 response to infection in the kidney

Outer medulla/ Epithelium lining Inflammatory cellsStimulus Cortex inner cortex the renal pelvis in the renal pelvis

iNOSPBS 1 24 1 6, 12, 24, 72 2 2E. coli 1 72 1 6, 12, 24, 72 1 12, 24, 72 1 6, 12, 24, 72

COX-2PBS 2 2 2 2E. coli 2 2 2 1 6, 12, 24, 72

Symbols indicate the presence (1) and absence (2) of the enzymes. The times given indicate the time after instillation.

of NO [20]. Most uropathogens are members of the fam- the bladder lumen. Thus, in agreement with analysis ofleukocyte-enriched urinary pellet from patients withily Enterobacteriaceae, known to reduce nitrate to nitrite.

Urine contains nitrates from dietary source, but the mice UTI [22], neutrophils are a likely primary source of ni-trite in the mouse urine.in our study were fed a nitrate/nitrite-free diet to ensure

that we measured nitrite formed from NO. The nitrite Prostanoids have been suggested to be inflammatorymediators during bladder inflammation [23], and the uri-levels in the urine from the E. coli-infected mice were

higher than the levels detected in PBS treated mice, nary content of prostaglandins is increased in acute bac-terial cystitis [24]. COX-2 is normally undetectable ineven three days after infection. There was a considerable

variation in this response. However, the results support the adult bladder but is expressed in the bladder duringfetal development [25]. We found expression of COX-2the notion that increased urinary nitrite excretion co-

incides with bacterial infection. in inflammatory cells that migrated into the bladder afterE. coli infection, supporting an involvement of COX-2The cellular origin of NO/nitrite was examined by

immunohistochemistry. A large number of iNOS- and in the host response during UTI. Whether COX-2 isessential for the urinary tract host response is not yetCOX-2–positive inflammatory cells were observed in

E. coli-infected bladders at 6 and 12 hours postinstilla- known, but mice lacking the COX-2 gene were reportedto have a normal inflammatory response to bacterialtion. Most of the inflammatory cells were identified as

PMN cells by hematoxylin and eosin staining and as invasion of the peritoneum [26]. The kinetics of iNOSand COX-2 expression in inflammatory cells were identi-neutrophils by immunohistochemistry using the RB6-

8C5 antibody. Time course studies showed that the cal, indicating that iNOS and COX-2 are triggered bysimilar mechanism/mediators. Double immunolabelingiNOS- and COX-2–positive cells migrated from the sub-

mucosa out to the bladder lumen and that the number studies are needed to clarify whether iNOS and COX-2really are coexpressed in the same inflammatory cells.of inflammatory cells in the bladder decreased by 24

hours. The elevated urinary nitrite levels coincided in Mucosal surfaces are important sites of microbiallyinduced diseases [27]. In vitro studies using human uro-time with the migration of iNOS-positive cells toward

Poljakovic et al: iNOS and COX-2 in E. coli-infected mice 901

Fig. 6. Immunohistochemistry of iNOS in mouse kidney sections. (A) PBS-instilled control mice did not show any iNOS-positive inflammatorycells (rp) renal papilla. (B) iNOS-positive tubular epithelial cells (arrowheads) in the outer medulla/inner cortex of PBS-instilled mice (gl) glomeruli.(C and D) iNOS-positive inflammatory cells (arrows) and epithelial cells (arrowheads) in kidneys infected with E. coli. (E) iNOS-positive tubularepithelial cells in the outermost parts of the cortex at 72 hours after infection. Scale bars: A–D 5 30 mm and E 5 10 mm.

Poljakovic et al: iNOS and COX-2 in E. coli-infected mice902

Fig. 7. Immunohistochemistry of COX-2 in mouse kidney sections. (A) No COX-2 immunoreactivity was detected in the control kidneys (rp)renal papilla. (B) COX-2–positive inflammatory cells (arrows) in kidneys infected with E. coli are shown. Note that no COX-2 immunoreactivitywas found in the epithelial cells lining the renal pelvis. Scale bars 5 30 mm.

epithelial cell lines have shown that uropathogenic mucosa by interfering with urothelial cell differentia-tion [35].E. coli stimulate the production of proinflammatory cyto-

kines like IL-6 and IL-8 through a direct effect of bacteria Notably, a constitutive iNOS immunoreactivity wasobserved in tubular epithelial cells in the outer medullaon urothelial cells [28]. In vivo infection activates a rapid

epithelial chemokine response [29]. In urothelial cells of and inner cortex of PBS control mice. This immunoreac-tivity to iNOS in the outer medulla was observed alsothe bladder, iNOS expression at first was detected 24 to

72 hours after bacterial instillation. In the mouse UTI in noninstilled control mice, which excludes that the ex-pression was caused by the instillation procedure. A con-model, most of the bacterial inoculum is cleared within

24 to 48 hours, with the greatest reduction in bacterial stitutively expression of iNOS has previously been re-ported in the outer medulla of the rat kidney [36, 37],number occurring between 6 and 24 hours [30]. Thus,

the majority of the bacteria should be cleared from the suggesting a more diverse function of this isoform apartfrom modulation of inflammation. Constitutive expres-bladder when the urothelial cells express iNOS. This

late onset of iNOS in urothelial cells compared with sion of iNOS may produce NO necessary for regulationof medullary blood flow [38].inflammatory cells suggests that the urothelial induction

of iNOS is not caused by bacteria per se, but rather by Following E. coli infection, neither the epithelium lin-ing the renal pelvis nor the bladder urothelial cells weremediators secreted by inflammatory cells. Experiments

on isolated human urothelial cells have confirmed that induced to express COX-2. This is in contrast to thesituation in other epithelial cells, like airway epitheliumbacteria are not a sufficient stimulus to induce iNOS

expression (unpublished observations). In rats, systemic [39] and intestinal epithelial cells [40], in which COX-2expression was induced in response to cytokines or bac-administration of E. coli LPS causes iNOS expression in

inflammatory cells and urothelial cells within three to terial infection. Thus, epithelial cells in the mouse uri-nary tract do not seem to express the COX-2 enzyme,six hours [31, 32], while intravesical administration of

LPS failed to increase bladder iNOS mRNA [33]. This or alternatively, the stimulus for COX-2 induction inurothelial cells may be different than for iNOS induction.difference may be explained by the poor LPS respon-

siveness of epithelial cells in the urinary tract [34]. The Increases in COX-2 mRNA expression were detected inonly a few kidney samples from infected mice. This islater onset of iNOS expression in urothelial cells also

suggests that NO is not a component of the urothelial likely to reflect that only the inflammatory cells andnot epithelial cells, as shown by immunohistochemistry,cell proinflammatory program. Induction of iNOS in uro-

thelial cells during the course of inflammation may rather contribute to the overall increase in kidney COX-2 ex-pression.serve other functions in the host response. NO may be

involved in urothelial cell shedding by promoting dele- Several in vitro and in vivo studies have proposed arole for renal NO production under pathophysiologicaltion of infected and damaged cells and may damage the

Poljakovic et al: iNOS and COX-2 in E. coli-infected mice 903

4. Junquero DC, Scott-Burden T, Schini VB, et al: Inhibition ofconditions. Isolated mouse and rat kidney epithelial cellscytokine-induced nitric oxide production by transforming growth

express iNOS following exposure to cytokines [41, 42]. factor-b1 in human smooth muscle cells. J Physiol 454:451–465,In patients with renal transplant rejection or infection, 1992

5. Moncada S, Higgs EA: Molecular mechanisms and therapeuticthere is evidence of increased NOS activity and iNOSstrategies related to nitric oxide. FASEB J 9:1319–1330, 1995expression in cells isolated from the urine [43]. In our 6. Vane JR, Mitchell JA, Appleton I, et al: Inducible isoforms of

UTI model, E. coli AD 110 induced iNOS expression cyclooxygenase and nitric-oxide synthase in inflammation. ProcNatl Acad Sci USA 91:2046–2050, 1994in epithelial cells lining the renal pelvis. Similar to the

7. Xie W, Chipman JG, Robertson DL, et al: Expression of a mitogen-situation in the bladder, the induction of epithelial iNOS responsive gene encoding prostaglandin synthase is regulated byin the kidney occurred later than the induction of iNOS mRNA splicing. Proc Natl Acad Sci USA 88:2692–2696, 1991

8. Dewitt DL, Meade EA, Smith WL: PGH synthase isoenzymein inflammatory cells. However, compared with the blad-selectivity: The potential for safer nonsteroidal antiinflammatoryder, the up-regulation of iNOS in kidney epithelial cells drugs. Am J Med 95:40–44, 1993

occurred earlier and was more pronounced, suggesting 9. Salvemini D, Misko TP, Masferrer JL, et al: Nitric oxide activatescyclooxygenase enzymes. Proc Natl Acad Sci USA 90:7240–7244,that epithelial NO may be an important component of1993the kidney host response to bacterial infection. The rela-

10. Inoue T, Fukuo K, Morimoto S, et al: Nitric oxide mediates in-tionship between severity of experimental pyelonephritis terleukin-1-induced prostaglandin E2 production by vascular

smooth muscle cells. Biochem Biophys Res Commun 194:420–424,and NO production was recently investigated in mice1993[44]. Inhibition of NOS was shown to be associated with

11. Rettori V, Gimenko M, Lyson K, et al: Nitric oxide mediatesincreased sensitivity to experimental E. coli pyelonephri- norepinephrine-induced prostaglandin E2 release from the hypo-

thalamus. Proc Natl Acad Sci USA 89:11543–11546, 1992tis of LPS nonresponder mice (C3H/HeJ), but not of12. Sautebin L, Dirosa M: Nitric oxide modulates prostacyclin biosyn-LPS responder mice (C3H/HeN). Furthermore, NO inhi-

thesis in the lung of endotoxin-treated rats. Eur J Pharmacolbition increased the infection rate in C3H/HeJ mice in- 262:193–196, 1994fected with E. coli bearing Dr fimbriae, but not P fimbriae 13. Salvemini D, Manning PT, Zweifel BS, et al: Dual inhibition

of nitric oxide and prostaglandin production contributes to the[44]. Thus, both tissue receptors and bacterial virulensantiinflammatory properties of nitric oxide synthase inhibitors. Jfactors seem to influence the contribution of NO to the Clin Invest 96:301–308, 1995

antibacterial defense mechanism during experimental 14. Swierkosz TA, Mitchell JA, Warner TD, et al: Co-induction ofnitric oxide synthase and cyclo-oxygenase: Interactions betweenpyelonephritis.nitric oxide and prostanoids. Br J Pharmacol 114:1335–1342, 1995In summary, the results show that E. coli AD 110

15. Svanborg C, Agace W, Connell H, et al: Urinary tract infectionsinduced iNOS and COX-2 expression in the urinary tract. and the mucosal immune system, in Mucosal Immunology (2nd

ed), edited by Ogra PL, Mestecky J, Lamm ME, et al, London,Two cell types contributed to this response: inflamma-Academic Press, 1999, pp 1381–1393tory cells that expressed both iNOS and COX-2 immuno-

16. Hedlund M, Svensson M-L, Hang L, et al: Bacterial adherence,reactivity and epithelial cells that only expressed iNOS transmembrane signaling and epithelial cytokine responses, in Re-

cent Advances in the Pathogenesis of Gastrointestinal Bacterial In-and with a later onset than the inflammatory cells. Thisfections, edited by Rampal P, Boquet P, Paris, John Libbey Euro-may suggest that the epithelial induction of iNOS is nottext, 1998, pp 65–79caused by direct bacterial activation but more likely by 17. Smith SD, Wheeler MA, Weiss RM: Nitric oxide synthase: An

mediators involved in the inflammatory response. endogenous source of elevated nitrite in infected urine. KidneyInt 45:586–591, 1994

18. Lundberg JON, Ehren I, Jansson O, et al: Elevated nitric oxide inACKNOWLEDGMENTS the urinary bladder in infectious and noninfectious cystitis. Urology48:700–702, 1996This project was supported by the Swedish Medical Research Coun-

19. Hagberg L, Engberg I, Freter R, et al: Ascending, unobstructedcil (12601), the Royal Physiographic Society, the National Board ofurinary tract infection in mice caused by pyelonephritogenic Esche-Health and Welfare, the Swedish Society of Medicine, and the Founda-richia coli of human origin. Infect Immunol 40:273–283, 1983tions of Crafoord, Magnus Bergwall, Tage Blucher, Ake Wiberg, and

20. Green LC, Wagner DA, Glogowski J, et al: Analysis of nitrate,Memorial Lars Hierta. A preliminary report has previously been pub-nitrite and [15N] nitrate in biological fluids. Anal Biochem 126:131–lished in abstract form (Immunol Lett 69:122, 1999). Mr. Peter Hoglund138, 1982is acknowledged for statistical consultation and Mr. Bjorn Frendeus

for initial help with the UTI model. 21. MacMicking J, Xie QW, Nathan C: Nitric oxide and macrophagefunction. Annu Rev Immunol 15:323–350, 1997

Reprint requests to Katarina Persson, Ph.D., Department of Clinical 22. Wheeler MA, Smith SD, Garcıa-Cardena G, et al: BacterialPharmacology, Lund University Hospital, SE-221 85 Lund, Sweden. infection induces nitric oxide synthase in human neutrophils. JE-mail: Katarina.Persson@klinfarm.lu.se Clin Invest 99:110–116, 1997

23. Mikhailidis DP, Jeremy JY, Dandona P: Urinary bladder prostan-oids-their synthesis, function and possible role in the pathogenesisREFERENCES and treatment of disease. J Urol 137:577–582, 1987

24. Farkas A, Alajem D, Dekel S, et al: Urinary prostaglandin E2 in1. Salvemini D, Seibert K, Marino MH: New concepts in inflamma-acute bacterial cystitis. J Urol 124:455–457, 1980tion and therapy. DN&P 9:204–219, 1996

25. Park JM, Yang T, Arend LJ, et al: Cyclooxygenase-2 is expressed2. Reiling N, Ulmer AJ, Duchrow M, et al: Nitric oxide synthase:in bladder during fetal development and stimulated by outlet ob-mRNA expression of different isoforms in human monocytes/mac-struction. Am J Physiol 273(4 Pt 2):F538–F544, 1997rophages. Eur J Immunol 24:1941–1944, 1994

26. Morham SG, Langenbach R, Loftin CD, et al: Prostaglandin3. Geller DA, Lowenstein CJ, Shapiro RA, et al: Molecular cloningsynthase 2 gene disruption causes severe renal pathology in theand expression of inducible nitric oxide synthase from human

hepatocytes. Proc Natl Acad Sci USA 90:3491–3495, 1993 mouse. Cell 83:473–482, 1995

Poljakovic et al: iNOS and COX-2 in E. coli-infected mice904

27. Kagnoff MF, Eckmann L: Epithelial cells as sensors for microbial localization of mRNA encoding inducible nitric oxide synthase inrat kidney. Am J Physiol 267(5 Pt 2):F748–F757, 1994infection. J Clin Invest 100:6–10, 1997

28. Hedges S, Bjarnadottir M, Agace W, et al: Immunoregulatory 37. Morrissey J, McCracken R, Kaneto H, et al: Location of aninducible nitric oxide synthase mRNA in the normal kidney. Kid-cytokines modify Escherichia coli induced uroepithelial cell IL-6

and IL-8 responses. Cytokine 8:686–697, 1996 ney Int 45:998–1005, 199438. Brezis M, Heyman SN, Dinour D, et al: Role of nitric oxide in29. Agace W, Hedges S, Andersson U, et al: Selective cytokine pro-

duction by epithelial cells following exposure to Escherichia coli. renal medullary oxygenation: Studies in isolated and intact ratkidneys. J Clin Invest 88:390–395, 1991Infect Immunol 61:602–609, 1993

30. Haraoka M, Hang L, Frendeus B, et al: Neutrophil recruitment 39. Watkins DN, Garlepp MJ, Thompson PJ: Regulation of the induc-ible cyclo-oxygenase pathway in human cultured airway epithelialand resistance to urinary tract infection. J Infect Dis 180:1220–1229,

1999 cells (A549) by nitric oxide. Br J Pharmacol 121:1482–1488, 199740. Eckmann L, Stenson WF, Savidge TC, et al: Role of intestinal31. Cook HT, Bune AJ, Jansen AS, et al: Cellular localization of

inducible nitric oxide synthase in experimental endotoxic shock in epithelial cells in the host secretory response to infection by inva-sive bacteria: bacterial entry induces epithelial prostaglandin Hthe rat. Clin Sci (Colch) 87:179–186, 1994

32. Persson K, Poljakovic M, Johansson K, et al: Morphological synthase-2 expression and prostaglandin E2 and F2a production. JClin Invest 100:296–309, 1997and biochemical investigation of nitric oxide synthase and related

enzymes in the rat and pig urothelium. J Histochem Cytochem 41. Markewitz BA, Michael JR, Kohan DE: Cytokine-induced ex-pression of a nitric oxide synthase in rat renal tubule cells. J Clin47:739–749, 1999

33. Olsson LE, Wheeler MA, Sessa WC, et al: Bladder instillation Invest 91:2138–2143, 199342. Mohaupt MG, Schwobel J, Elzie JL, et al: Cytokines activateand intraperitoneal injection of Escherichia coli lipopolysaccharide

up-regulate cytokines and iNOS in rat urinary bladder. J Pharmacol inducible nitric oxide synthase gene transcription in inner medul-lary collecting duct cells. Am J Physiol 268(4 Pt 2):F770–F777,Exp Ther 284:1203–1208, 1998

34. Hedlund M, Wachtler C, Johansson E, et al: P fimbriae-depen- 199543. Smith SD, Wheeler MA, Zhang R, et al: Nitric oxide synthasedent, lipopolysaccharide-independent activation of epithelial cyto-

kine responses. Mol Microbiol 33:693–703, 1999 induction with renal transplant rejection or infection. Kidney Int50:2088–2093, 199635. Elgavish A, Robert B, Lloyd K, et al: Nitric oxide mediates the

action of lipoteichoic acid on the function of human urothelial 44. Nowicki B, Singhal J, Fang L, et al: Inverse relationship betweenseverity of experimental pyelonephritis and nitric oxide productioncells. J Cell Physiol 169:66–77, 1996

36. Ahn KY, Mohaupt MG, Madsen KM, et al: In situ hybridization in C3H/HeJ mice. Infect Immunol 67:2421–2427, 1999