Toll-like receptors TLR2 and TLR4 initiate the innate immune response of the renal tubular epithelium to bacterial products

Abstract

Summary

Renal tubular epithelial cells (TECs) respond diffusely to local infection, with the release of multiple cytokines, chemokines and other factors that are thought to orchestrate the cellular constituents of the innate immune response. We have investigated whether the Toll-like receptors TLR4 and TLR2, which are present on tubular epithelium and potentially detect a range of bacterial components, co-ordinate this inflammatory response acting through nuclear factor-kappa B (NF-κB). Primary cultures of TECs were grown from C57BL/6, C3H/HeN, C3H/HeJ, TLR2 and TLR4 knock-out mice. Cell monolayers were stimulated with lipopolysaccharide (LPS) and synthetic TLR2 and 4 agonists. The innate immune response was quantified by measurement of the cytokines tumour necrosis factor (TNF)-α and KC (IL-8 homologue) in cell supernatants by enzyme-linked immunosorbent assay. Cultured TECs grown from healthy mice produced the cytokines TNF-α and KC in response to stimulation by LPS and synthetic TLR2 and TLR4 agonists. Cells lacking the respective TLRs had a reduced response to stimulation. The TLR2- and TLR4-mediated response to stimulation was dependent on NF-κB signalling, as shown by curcumin pretreatment of TECs. Finally, apical stimulation of these TLRs elicited basal surface secretion of TNF-α and KC (as well as the reverse), consistent with the biological response in vivo. Our data highlight the potential importance of TLR-dependent mechanisms co-ordinating the innate immune response to upper urinary tract infection.

Introduction

Urinary tract infections are among the most common bacterial infections affecting humans, and a major cause of morbidity and mortality. Chronic infection in children can also lead to renal scarring and subsequent end-stage renal failure. During pyelonephritis, renal tubular epithelial cells (TECs) are one of the first host cells to come into contact with invading bacteria and influence the innate immune response via the production of cytokines and anti-microbial peptides [1]. This cytokine production is important in the recruitment and activation of inflammatory cells such as neutrophils, macrophages and lymphocytes. Cellular recruitment is a vital part of the innate immune response, which plays a central role in maintaining sterility of the urinary tract. In particular, neutrophils have been shown to be of key importance in the process of bacterial clearance [2]. Neutrophils are attracted along a chemotactic gradient, and the chemokine interleukin (IL)-8 has been shown to play a critical role in neutrophil recruitment. TECs have been shown to secrete a number of cytokines, including IL-8, during infection [3]. However, an excessive inflammatory response may lead to irreversible renal injury. Therefore the local response by tubular epithelial cells may have both beneficial and detrimental effects, the overall balance of which will determine the clinical outcome. Despite these important findings, our understanding of the mechanisms which initiate these local influences remains unclear. Toll-like receptors (TLRs) are a recently discovered group of pattern recognition receptors that recognize conserved molecular motifs found on a variety of organisms including bacteria, viruses and fungi [4]. So far, 11 mammalian TLRs have been described. TLR4 is the receptor which recognizes lipopolysaccharide (LPS) [5,6], a constituent of the cell wall of Gram-negative bacteria such as Escherichia coli, which accounts for up to 80% of UTIs [7]. Meanwhile, TLR2 reacts with a wider spectrum of bacterial products found in both Gram-positive and -negative bacteria including lipoproteins, peptidoglycans and lipoteichoic acid [8,9]. Constitutive expression of mRNA for TLR2 and TLR4 as well as associated molecules has been demonstrated on mouse tubular epithelial cells. Furthermore, C-C chemokine secretion was found to be strictly dependent upon TLR2 and TLR4 [10]. Mice with defective TLR4 have been shown previously to have reduced neutrophil recruitment and impaired bacterial clearance during experimental pyelonephritis [11]. More recently, using bone marrow chimeric mice, it has been shown that TLR4 is required on intrinsic renal cells as well as bone marrow-derived cells for the effective control of ascending UTI [12]. The most recently described mammalian TLR, TLR11, has been noted to be present on uroepithelial cells and protects against infection from uropathogenic E. coli in mice [13]. The interaction between TLRs and their specific ligands leads to activation of defined intracellular signalling pathways. The intracellular adaptor protein, myeloid differentiation factor 88 (MyD88) appears to be involved in signalling by all TLRs except TLR3. Activation of the MyD88 signalling pathway results in the translocation of nuclear factor nuclear factor-kappa B (NF-κB) and activation of mitogen activated protein kinases (MAPKs), leading to cytokine gene up-regulation [14].

The aim of our study was to determine the role of TLR2 and TLR4 in the generation of tumour necrosis factor (TNF)-α and KC (IL-8 homologue) by renal tubular cells during pyelonephritis. TNF-α has a variety of proinflammatory effects, but overproduction can lead to septic shock following Gram-negative infection [15]. KC is a murine homologue of IL-8, and an important neutrophil chemoattractant [16]. Furthermore, epithelial production of IL-8 plays a vital role in the transmigration of neutrophils across the epithelial cell layer and into the urinary tract lumen [3]. We further investigated whether the TLR-dependent production of these cytokines by TECs required activation of NF-κB. Although the vast majority of cases of pyelonephritis occur as a result of ascending infection, some may occur following haematogenous spread of bacteria via the bloodstream. We therefore examined whether cytokine production by TECs could occur as a result of either apical or basolateral exposure to bacterial products. An improved understanding of these molecular events may allow separation of the host response into those which are of benefit and those which have potentially detrimental effects.

Materials and methods

Antibodies and reagents

DMEM-F12 culture medium was purchased from gibco (Paisley, UK). All other cell culture reagents were obtained from Sigma (Poole, UK) except where stated otherwise. LPS from E. coli (serotype 0111:B4) was obtained from Sigma, purified LPS from E. coli (serotype R515 Re) was obtained from Alexis Biochemicals (Lausen, Switzerland), the synthetic lipoprotein Palm₃Cys-SKKK × 3HCl was obtained from EMC Microcollections (Tuebingen, Germany) and the synthetic lipid A analogue (ONO 4007) was a kind gift from ONO Pharmaceuticals (Osaka, Japan). Cyclohexamide and curcumin were purchased from Sigma. Rat anti-mouse CD68 and rat IgG2a isotype control were purchased from Serotec (Oxford, UK). Mouse anti-rat fluorescein isothiocyanate (FITC)-conjugated secondary was purchased from Jackson ImmunoResearch Europe Ltd (Soham, UK). FITC-conjugated mouse anti-pan cytokeratin monoclonal antibody was purchased from Sigma. Reagents for reverse transcription–polymerase chain reaction (RT–PCR) were purchased from Promega (Southampton, UK) and primers were constructed by MWG-Biotech (Ebersberg, Germany).

Cell culture and stimulation assays

C57/BL6, C3H/HeN and C3H/HeJ mice were purchased from Harlan (Oxon, UK). TLR2 and TLR4 knock-out mice were a kind gift from Professor Akira (Osaka, Japan). Kidneys from 6–8-week-old mice were used to grow primary cultures of proximal tubular cells (TECs) as described previously [17]. Kidneys were bisected and outer cortical tissue was separated. Tissue was then minced and digested in collagenase II (Worthington, Lakewood, CO, USA) for 20 min at 37°C. The digest was then passed through graduated sieves (250, 106, 75 and 40 nm). Cells trapped in the 40 nm sieve were collected by washing, centrifuged to obtain a pellet and resuspended in culture medium [Dulbecco’s modified essential medium (DMEM)]-F12 supplemented with 2% fetal calf serum (FCS), insulin transferrin selenite (ITS), T3 (10⁻¹² M), hydrocortisone (40 ng/ml) and penicillin/streptomycin (100 µ/ml penicillin, 100 µg/ml streptomycin). The cell suspension was added to 1% gelatin-coated culture plates and incubated at 37°C, 5% CO₂.

TECs were grown until confluent under direct light microscopy. Culture medium was changed to serum-free conditions 24 h prior to stimulation with bacterial products. Experiments were performed in triplicate using three consecutive wells of six-well plates.

Polarity

For polarity experiments, TECs were grown on 30 mm polycarbonate cell culture plate inserts (Millipore, Billerica, MA, USA) with a pore size of 0·4 µm. Inserts were coated with fibrinogen prior to seeding with cells. Cell growth was monitored by measuring transepithelial cell resistance (TER) using Millicell-ERS microvoltmeter (Millipore). TER was found to plateau at values between 550 and 600 Ω/cm², which was taken as indicating formation of a confluent monolayer. This corresponded to values obtained using the MDCK cell line, with which this technique has been validated in the literature [18]. TECs grew with the apical surface uppermost, therefore isolated stimulation of the apical surface was achieved by addition of LPS to the upper chamber. Similarly, stimulation of the basolateral cell surface was performed by addition of LPS to the lower chamber. Monitoring of TER following stimulation demonstrated that integrity of the monolayer was maintained.

Cytokine gene expression and protein biosynthesis

Total RNA was extracted from cells using a commercially available kit (Qiagen, Crawley, UK) according to the manufacturer’s instructions. RNA concentration was determined by spectrometer absorbance at 260 nm 5 µg of RNA was reverse transcribed in a reaction together with 160 ng oligo(dT)₁₅, 500 µM of each dNTP and 200 U Moloney murine leukaemia virus reverse transcriptase in 20 µl solution for 90 min at 37°C.

PCR was carried out using the primer sequences shown in Table 1 according to published conditions in a 25-µl reaction with 12·5 pmol each of forward and reverse primers, 200 µM of each dNTP and 3 U Taq polymerase. Products were stained on a 1·2% agarose gel and stained with ethidium bromide. Unconverted RNA for each sample was also run to demonstrate the absence of any genomic contamination.

Table 1.

Primer sequences used in reverse transcription–polymerase chain reaction.

		Reference
β actin	5′-GAG CAA GAG AGG TAT CCT GAC C-3′
	5′-GGA TGC CAC AGG ATT CCA TAC C-3′
TNF-α	5′-GGT GCC TAT GTC TCA GCC TCT-3′	[19]
	5′-CAT CGG CTG GCA CCA CTA GTT-3′
KC	5′-CCT TGA CCC TGA AGC TCC CTT GGT TC-3′	[20]
	5′-CGT GCG TGT TGA CCA TAC AAT ATG-3′
TLR2	5′-TCT GGG CAG TCT TGA ACA TTT − 3′	[21]
	5′-AGA GTC AGG TGA TGG ATG TCG − 3′
TLR4	5′-CAA GGG ATA AGA ACG CTG AGA-3′	[21]
	5′-GCA ATG TCT CTG GCA GGT GTA-3′
CD14	5′-CTA GTC GGA TTC TAT TCG GAG C-3′	[22]
	5′-AGA CAG GTC TAA GGT GGA GAG G-3′
MD-2	5′-CTG AAT CTG AGA AGC AAC AGT GG-3′	[23]
	5′-CAG TCT CTC CTT TCA GAG CTC TGC-3′
MyD88	5′-ATC CGA GAG CTG GAA ACG-3′	[22]
5′-GCA AGG GTT GGT TA ATC-3′

MyD88: myeloid differentiation factor 88; TLR: Toll-like receptor; TNF: tumour necrosis factor.

Concentrations of TNF-α and KC were determined in culture supernatants using commercially available enzyme-linked immunosorbent assay (ELISA) kits (R&D Systems, Oxon, UK) according to the manufacturer’s instructions. All samples were assayed in duplicate and data presented as the mean.

Statistics

Experiments were performed in triplicate with data shown as the mean with error bars depicting the standard deviation. P-values were calculated using the unpaired t-test or Mann–Whitney where there was a significant difference in variance. P-values of less than 0·05 were regarded as significant.

Results

Cell phenotype was confirmed by positive staining for cytokeratin (Fig. 1a) and brush border alkaline phosphatase together with demonstration of apical surface microvilli under electron microscopy. Absence of contamination with macrophages was confirmed by negative staining for CD68. Thioglycolate induced peritoneal macrophages were used as a positive control (Fig. 1b).

Fig. 1.

(a) Positive staining for cytokeratin in tubular cell cultures. (b) Positive staining of peritoneal macrophages with anti-CD68 antibody. (c) Simultaneous staining of tubular cell cultures for CD68 was negative.

Gene expression of TLR2, TLR4, CD14, MD-2 and MyD88 in TECs

The presence of mRNA for TLR2 and TLR4 was verified in primary cultures of TECs from C57/BL6 mice by RT–PCR (Fig. 2). The presence of message for CD14 and MD-2, which are both required in TLR4 activation by LPS, were also confirmed, along with that for MyD88, the adaptor molecule involved in the intracellular signalling pathway common to all TLRs and interleukin 1 and 18. RT–PCR was performed on RNA extracted from thioglycolate-induced peritoneal macrophages as a positive control. Lack of corresponding bands in unconverted mRNA was demonstrated to show absence of contamination with genomic DNA.

Fig. 2.

Expression of Toll-like receptors TLR2, TLR4 and related molecules in tubular epithelial cells (TECs) from C57/BL6 mice. Thioglycolate induced peritoneal macrophages (macro) from the same strain were used as a positive control. Cells were cultured under basal conditions. mRNA was extracted and reverse transcription–polymerase chain reaction (RT–PCR) was performed as described.

TECs produce the cytokines TNF-α and KC in a dose- and time-dependent manner following stimulation with LPS

Both TNF-α and KC production was seen clearly with the highest dose (10 µg/ml) at 2 h post-stimulation (Fig. 3). At 24 h, there was no significant difference between the responses to 100 ng/ml or 10 µg doses, suggesting that this represented a maximal response.

Fig. 3.

Dose–response and time-course for tumour necrosis factor (TNF)-α and KC secretion by tubular epithelial cells (TECs) from C57/BL6 mice in response to stimulation with lipopolysaccharide (LPS). TECs were stimulated with 1, 100 and 10 000 ng/ml of LPS and cell supernatant sampled over a period of 24 hours. TNF-α (a) and KC (b) protein concentration in samples as determined by enzyme-linked immunosorbent assay (ELISA). Data shown are representative of three separate experiments.

LPS and synthetic lipid A stimulate cytokine production in TECs via a TLR4-dependent mechanism

In order to assess the role of TLR4 signalling in TNF-α and KC production by TECs, we stimulated primary cultures of TECs from C3H/HeJ strain mice with LPS (Fig. 4c,d). Although these mice have TLR4, they have a point mutation in the tlr4 gene which results in substitution of histidine for proline, resulting in defective signalling [5]. C3H/HeN mice without this defect were used as a positive control (Fig. 4a,b). Interestingly, the response to LPS was not absent in TECs from C3H/HeJ mice, although the magnitude of the response was reduced in comparison with control cells from C3H/HeN mice. A problem that has been highlighted previously in using LPS preparations derived from bacteria is that of purity, and it has been shown that contaminants can lead to stimulation via other TLRs. To overcome this, we also stimulated cells concurrently with a highly purified source of LPS, and with a synthetic lipid A analogue. Both purified LPS and synthetic lipid A did not induce production of either TNF-α or KC in TECs from C3H/HeJ mice, indicating that the response seen with standard LPS formulation was a result of contaminants. In assessing the TNF-α response in C3H/HeN cells, it was found that the standard preparation of LPS was able to induce a greater response in comparison with lipid A. Higher doses of lipid A did not produce a greater response than that shown (data not shown).

Fig. 4.

Tumour necrosis factor (TNF)-α and KC production by tubular epithelial cells (TECs) in response to lipopolysaccharide (LPS) is dependent on the presence of a functional Toll-like receptor TLR4. TECs were cultured from C3H/HeJ, C3H/HeN, TLR2 knock-out, TLR4 knock-out and C57/BL6 mice. Cells were stimulated with LPS, a highly purified LPS preparation or synthetic lipid A (ONO 4007) diluted to a concentration of 1 µg/ml in sterile water. Culture supernatants were collected at 24 h post-stimulation, and TNF-α (a, c, e, g, i) and KC concentrations (b, d, f, h, j) determined by enzyme-linked immunosorbent assay (ELISA). Data shown are representative of two separate experiments. *P < 0·05, **P < 0·005 and ***P < 0·0005 highlight responses which are statistically significant when compared that seen from cells stimulated with an equivalent volume of sterile water alone.

We further assessed the roles of both TLR2 and TLR4 in cytokine production following stimulation with LPS using TECs grown from TLR2 and TLR4 knock-out mice which had been back-crossed onto a C57/BL6 background (Fig. 4g–j). Wild-type C57/BL6 mice were used to grow control cells (Fig. 4e,f). TECs from TLR4 knock-out mice failed to respond to either purified LPS or synthetic lipid A, confirming that the response to LPS is dependent upon TLR4. A response to crude LPS was seen in TLR4 knock-out TECs, due presumably to contamination. Interestingly, the response to both LPS preparations and lipid A in cells grown from TLR2 knock-out mice, although present, was significantly reduced in magnitude compared with that seen from wild-type cells (for stimulation with purified LPS, TNF-α production C57BL/6 v TLR2 knock-out P < 0·0001, KC production P < 0·005). This suggests that TLR2 might help to augment the response to LPS by TLR4, independent of stimulation by contaminants.

Synthetic lipoprotein stimulates cytokine production in TECs via TLR2

To assess the role of TLR2, we grew primary cultures of TECS from wild-type C57/BL6, TLR2 knock-out and TLR4 knock-out strain mice. These were then stimulated with a synthetic lipopeptide, Palm₃Cys-SKKKK (Fig. 5). In wild-type cells a response to stimulation was seen, both in terms of TNF-α and KC secretion into the culture medium. This response was absent in TLR2 knock-out cells, indicating that TLR2 is required for the response to lipopeptides. Although TLR4 knock-out cells demonstrated a response, the magnitude was reduced in comparison with that seen in wild-type cells (for stimulation with 10 µg lipoprotein, TNF-α secretion wild-type v TLR4 knock-out P < 0·005, KC P < 0·05). This further supports our theory of co-operation between TLR2 and TLR4.

Fig. 5.

Tumour necrosis factor (TNF)-α and KC production by tubular epithelial cells (TECs) in response to lipoprotein is dependent on Toll-like receptor TLR2. TECs from C57/BL6, TLR2 and TLR4 knock-out mice were grown as described and stimulated with 1 µg/ml and 10 µg/ml of Palm₃Cys-SKKKK, a synthetic lipoprotein dissolved in dimethylsulphoxide (DMSO). Cytokine concentration in the culture supernatant was determined at 24 h by enzyme-linked immunosorbent assay (ELISA). Data shown are representative of two separate experiments. *P < 0·05, **P < 0·005 and ***P < 0·0005 highlight responses which are statistically significant when compared with that seen from cells stimulated with an equivalent volume of DMSO alone.

TNF-α and KC production in response to LPS follows gene up-regulation which does not require new synthesis of protein intermediates but is NF-κB-dependent

In macrophages, TLR4 activation has been shown to lead to translocation of the nuclear factor NF-κB, which leads to cytokine production. This pathway requires recruitment of the intracellular adaptor protein MyD88, gene expression of which was demonstrated in Fig. 1. To examine whether NF-κB activation was required for the TNF-α and KC production seen, we pretreated cells with curcumin prior to stimulation. It has been reported previously that pretreatment with 5 µM curcumin inhibits JNK activation, while concentrations of 50 µM inhibit NF-κB activation [10]. Pretreatment of TECs with 50 µM curcumin was found to completely abolish any increase in TNF-α and KC in culture supernatants following stimulation with LPS (Fig. 6a). Analysis of RNA extracted from cells revealed that this was a consequence of failure of gene up-regulation (Fig. 6b). Therefore, activation of NF-κB was critical for TNF-α and KC production by TECs in response to LPS. Five µM curcumin did not abolish production of either cytokine; however, the magnitude of response was reduced in both, suggesting that JNK activation may be required for a maximal response. In order to assess if this was a direct effect or required the new synthesis of intermediate cytokines, cells were also pretreated with cyclohexamide prior to stimulation. Cyclohexamide inhibits any new protein synthesis, and failed to inhibit gene up-regulation (Fig. 6b).

Fig. 6.

Tumour necrosis factor (TNF)-α and KC gene upregulation in response to lipopolysaccharide (LPS) does not require new protein synthesis and is dependent on NF-κB activation. Tubular epithelial cells (TECs) were pretreated with either 100 µM cyclohexamide (CHX) or stated concentrations of curcumin for 1 h prior to stimulation with 1 µg/ml of LPS. Samples of culture supernatant were collected at 4 h post-stimulation and RNA extracted from cells. TNF-α (a) and KC (b) concentrations were determined by enzyme-linked immunosorbent assay (ELISA) on supernatants from curcumin and control groups. Reverse transcription–polymerase chain reaction (RT–PCR) was performed on extracted RNA as described. β actin was used as a housekeeping gene to ensure that pretreatment did not affect cell viability adversely (c).

TECs express immunological polarity

TECs grown in culture grow with their apical surfaces uppermost. During ascending UTI, bacteria will initially interact with the apical surface of epithelial cells. Renal epithelial cells are polarized cells with many surface proteins expressed either on the apical or basolateral cell surface. Similarly, any epithelial response to exogenous stimuli may be polarized. Addition of LPS to either the upper or lower chambers did not affect TER, suggesting that the integrity of the monolayer was unaffected (data not shown). Stimulation of the apical surface led to TNF-α secretion, as would be expected from previous experiments. However, independent sampling of the upper and lower chambers revealed that this secretion was polarized, occurring predominantly apically (Fig. 7a). Stimulation via the basolateral surface also led to a predominantly apical TNF-α response. In contrast, the KC response was non-polarized, with both apical and basolateral secretion occurring regardless of whether LPS was applied to the apical or basolateral cell surface (Fig. 7b).

Fig. 7.

Stimulation via the apical cell or basolateral cell surface results in polarised responses. Tubular epithelial cells (TECs) from C57/BL6 mice were grown on inserts within six-well plates as described. Trans-epithelial resistance (TER) was used to confirm confluency of the cell monolayer as described. Cells were then stimulated via either their apical surface, by addition of 1 µg/ml of purified lipopolysaccharide (LPS) to the upper chamber, or via the basolateral surface by addition of LPS to the lower chamber. Monitoring of TER demonstrated that the monolayer remained intact following stimulation. Supernatant samples were obtained from upper and lower chambers and tumour necrosis factor (TNF)-α (a) and KC (b) concentrations determined via enzyme-linked immunosorbent assay (ELISA) as described previously. Data shown are representative of three separate experiments. **P < 0.005; n.s.: non-significant difference.

Discussion

The findings presented in this paper support the hypothesis that TECs play a vital role in initiating the innate immune response to ascending urinary tract infection. Innate immunity is critical for the clearance of bacteria, and epithelial cells, the first host cells to interact with invading bacteria, are in a key position to initiate this response. They are capable of synthesizing chemotactic and inflammatory cytokines in response to bacterial products through engagement of TLRs expressed on their surface.

Previous studies have shown that mouse TECs are able to synthesize C-C chemokines via engagement of TLRs and therefore promote T cell infiltration [13]. However, in the context of acute UTI and pyelonephritis, the function of T cells is uncertain. The predominant infiltrating cell in the early stages of infection is the neutrophil, and neutrophil depletion renders normal mice susceptible to UTI [2]. Chemokines are the main neutrophil chemoattractants and are vital for migration into the inflamed kidney. Previous studies have shown that human TECs produce IL-8 in response to infection with E. coli [24]; however, their ability to do so in response to isolated bacterial products is unclear. It has been shown that human TECs, despite possessing TLR4, respond poorly to LPS [25]. This may be due to lack of CD14 expression [26]. It is clear from the work presented here that murine cells express both TLR4 and the required accessory molecules, and can produce KC in response to purified LPS. In addition, murine tubular epithelial cells produce TNF-α. This cytokine has a wide variety of proinflammatory effects, including stimulating local cells to up-regulate cytokine and adhesion molecule expression, which will further assist in the recruitment and activation of neutrophils into the kidney [27].

Although it is clear that TLR4 is an absolute requirement for the response to LPS, our studies using standard LPS preparations demonstrate that the response to Gram-negative organisms is not wholly dependent on TLR4. A response to crude LPS is seen in TLR4 defective or knock-out cells. This response was lost when using either highly a highly purified LPS or a synthetic lipid A analogue. This confirms that contaminants in the standard LPS preparation were able to stimulate cytokine production, possibly acting via other TLRs. A previous study has demonstrated that contaminants in commercial LPS preparations can signal through TLR2 [28]. Furthermore, dual stimulation of TLR2 and TLR4 on macrophages has been shown to have synergistic effects [29]. This would explain why our standard preparation of LPS was able to elicit a greater maximal response than the pure agonists on their own.

Although purified LPS and lipid A were able to elicit a response in TLR2 knock-out cells, the magnitude of the both TNF-α and KC secretion was significantly reduced in comparison with wild-type cells, suggesting that TLR2 may co-operate with TLR4 in order to achieve a maximal response to LPS. The response to TLR2 ligands in TLR4 knock-out cells was also reduced. Furthermore, the response to the impurities in standard LPS was greater in C3H/HeJ cells than in cells grown from TLR4 knock-out mice. Although C3H/HeJ mice have defective TLR4 signalling, TLR4 is still present and can therefore augment signalling via other TLRs such as TLR2. Previous studies have highlighted co-operation between TLRs. For example, TLR2 has been shown to co-operate with TLR1 and TLR6 [30]; however, as yet TLR4 has not been shown to co-operate with any other TLR. TLR4, however, requires a number of accessory molecules in order to respond to LPS, including CD14 and MD-2 [31,32]. Our findings suggest that in order to achieve a maximal response to pure agonists, co-operation between TLR2 and TLR4 is required. This adds a further facet to the synergy described previously when both TLRs are stimulated simultaneously.

By growing cells on inserts, we were able to stimulate the apical and basolateral cell surfaces separately. Both cell surfaces of TECs responded to LPS, suggesting that TLR4 is present on both the apical and basolateral sides. Another possibility is that this represents the ability of LPS to internalize from either cell surface, with TLR4 located within the cell, as has been demonstrated in intestinal epithelial cells [33]. It has been postulated that the reason for this is that the gut is normally a non-sterile environment and that if TLR4 were located on the cell surface, this would lead to an unwanted host response to commensal bacteria. By locating TLR4 within the cell, the response can be limited to those bacteria that are invading the epithelium. However, unlike the gut, the urinary tract is normally a sterile environment, and any bacteria within the tubular lumen would constitute infection. In this situation it would be beneficial for TLR4 to be located on the apical cell surface where it can sample luminal contents.

Isolated stimulation of apical and basolateral surfaces revealed the fascinating discovery that the cytokine responses can be polarized. It is already established that TECs are polarized with respect to their ability to transport ions and proteins across their membranes. It appears that this polarization extends to their immunological response. Stimulation of either the apical or basolateral cell membrane with purified LPS resulted in predominantly apical TNF-α secretion. KC secretion, however, did not demonstrate any polarity. Given the functions and effects of these cytokines, this might be expected. TNF-α is a potent inflammatory cytokine, and its systemic effects include the vascular collapse seen in patients with septic shock [34]. Such a response would not be of benefit to the host in the case of a localized infection. This may explain why there is little basolateral secretion of TNF-α which would then enter the systemic circulation. Conversely, TNF-α is needed for neutrophil activation, hence apical secretion into the lumen would be beneficial during infection of the upper urinary tract. KC, on the other hand, is needed for neutrophil chemotaxsis from the vascular compartment into the tubular lumen, hence the need for basolateral secretion. Studies have also demonstrated a key role for IL-8 and its receptor in transepithelial migration of neutrophils into the tubular during UTI [35]. Failure of epithelial transmigration leads to defective bacterial clearance and accumulation of neutrophils in the subepithelial space, resulting in greater tissue injury. Therefore, both apical and basolateral secretion may be needed, initially to recruit cells from the circulation, and subsequently to allow their passage across the epithelial membrane and into the tubular lumen where the bacteria are resident.

In terms of the pathways involved in cytokine production, our data confirm a critical role for NF-κB activation leading to cytokine gene up-regulation. Cytokine production was also found not to require the synthesis of any protein intermediates; however, the involvement of preformed intermediates cannot be ruled out. Whether TECs utilize exactly the same signalling pathways as macrophages remains to be determined, although the presence of genetic message for MyD88 raises this as a possibility.

Our study confirms previous findings that TLR2 and TLR4 are present on renal epithelial cells and demonstrates a role in the innate immune response to bacteria during ascending urinary tract infection. We also highlight that although TLRs have specific ligands, co-operation may be required between these TLRs and maybe other molecules in order to achieve a maximal response. This supports previous findings that rather than acting alone, TLRs might form receptor complexes within lipid rafts on the cell surface [36]. This might explain how using a relatively small number of TLRs, the innate system can provide tailored responses to a wide variety of molecules. Further complexity is added when we consider that in the clinical setting, cells are exposed to whole organisms rather than their individual components, therefore it is likely that a number of TLRs are stimulated simultaneously on each cell, and that the overall response will be the sum of a number of individual interactions. Endogenous ligands have been identified for both TLR2 and TLR4, suggesting that they potentially have a role in a variety of inflammatory renal conditions [37].