Identification of TLR4 as the Receptor That Recognizes Shiga Toxins in Human Neutrophils

This article is featured in In This Issue, p.4469

Shiga toxin (Stx)–producing Escherichia coli (STEC) infections are a worldwide health problem because of the severity of the deriving illnesses, which occur in sporadic form or as community-wide outbreaks (1–3). In humans, the main clinical manifestations of STEC infections are bloody diarrhea and its life-threatening sequela, hemolytic uremic syndrome (HUS). The latter represents the main cause of acute renal failure in early childhood, and it is characterized by thrombocytopenia and microangiopathic hemolytic anemia (1–3). Such a triad was also observed in adult patients during the dramatic STEC outbreak, which occurred in summer 2011, causing ∼4000 cases of bloody diarrhea and more than 800 HUS cases in Europe (4–7). Thus, the risks of STEC infections are 2-fold: the sudden appearance of food-borne outbreaks and the burden of a constant incidence rate of childhood HUS.

Infectious diseases always develop as a consequence of a balance between the virulence factors expressed by a pathogen and the host response to those challenges. In this view, Stx as STEC virulence factors, along with neutrophils as effectors of innate immunity, play an important role in the pathogenesis of HUS. STEC produce two main AB5 toxin variants—Shiga toxin 1 (Stx1) and Shiga toxin 2 (Stx2)—consisting of five B chains that are linked noncovalently to a single A chain (8, 9). The B pentamer directs the toxins toward cells harboring glycosphingolipids of the globoseries called globotriaosylceramide (Gb3Cer) and globotetraosylceramide (Gb4Cer) (10). Few human cells possess these receptors (11–14) whose expression characterizes the endothelial lining of the intestine, brain, and kidney, which are indeed the body sites mainly targeted by Stx. In particular, glomerular endothelial injury induced by Stx is considered central to the pathogenesis of HUS (9). The A chain is endowed with the enzymatic activity (15–17) and is responsible for an array of damage within intoxicated cells, as well as multiple cellular and host responses, culminating in the typical prothrombotic state and renal involvement seen in HUS. Among the toxin actions on endothelial cells involved in HUS pathogenesis, damage to DNA, ribotoxic stress, endoplasmic reticulum stress, activation of the apoptotic program, stimulation of the production of proinflammatory cytokines, and upregulation of adhesion molecules have been described (17–23). Moreover, activation of complement, formation of platelet-leukocyte complexes, and production of tissue factor–bearing microparticles in blood by Stx have been demonstrated in vitro and in the blood from patients with HUS (24–26). In this context, several lines of evidence indicate that Stx bind in vitro to platelets through Gb3Cer lipoforms similar to those on target endothelial cells (27–29). A role of platelets as Stx carriers in blood has been proposed by different authors; however, platelets internalize Stx within 2 h, becoming activated and aggregated (28). Because platelet aggregates are cleared by the reticuloendothelial system, direct binding of the toxins to these cells seems to be involved in thrombocytopenia, rather than in the passive transfer of Stx to the kidney and brain.

Neutrophils too are involved in HUS because, in patients, these leukocytes are found to be activated, degranulated, and hyporesponsive to other stimuli (30, 31). Moreover, neutrophilia is a typical finding in patients with HUS, and elevated neutrophil counts are considered a poor prognostic factor (32–34). In vitro, Stx bind to neutrophils from healthy donors through an unknown receptor (31, 35, 36), whereas neutrophils carrying Stx on their membrane have been found in HUS patients (31, 37, 38). Because free toxins have never been detected in the sera of patients with overt HUS, several authors have proposed that neutrophils can deliver Stx from the gut, which is the site of toxin production by the confined noninvasive STEC, to the kidney or the brain (31, 35). In this view, the Stx–neutrophil interaction could be considered a harmful phenomenon. On the other hand, neutrophils are enrolled in the innate immunity responses whose role in defending the host from bacterial infections has been clearly established. In fact, the relevance of the Stx–neutrophil interaction is ambivalent because two different studies with HUS patients showed that STEC-infected children with high amounts of Stx on neutrophil membranes had mild renal involvement and were at risk for neurologic complications, whereas patients with lower amounts of Stx on their granulocytes showed severe renal failure and lower risk of neurologic involvement (31, 37, 38). Thus, the identification of the Stx receptor on granulocytes is important to shed some light on the role of neutrophils either in the protection by or in their involvement in toxin actions during the pathogenesis of HUS.

The Stx–neutrophil interaction is not followed by the internalization of the toxins (35), which conversely occurs in the case of Gb3Cer- or Gb4Cer-bearing cells such as human endothelial cells (10). Indeed, these glycosphingolipids (Gb3Cer or Gb4Cer) are not expressed by human neutrophils (39). The neutrophil receptor interacts with the A chain of Stx (40, 41) and recognizes the A subunit of ricin, the well-known toxin from Ricinus communis, without inducing its internalization (41). This is in keeping with the notion that Stx1 and ricin share the same molecular mechanism of action on intracellular targets and homologous A chain folding and primary sequence (8). Partial unfolding of Stx1 causes loss of neutrophil-binding activity, preserving toxicity for human endothelial cells (42). Because neutrophils recognize molecular signatures common to different pathogen-derived molecules by means of specific receptors belonging to the large family of pattern-recognition receptors (43, 44), their involvement in the recognition of these toxic molecules has been suggested. However, different candidates belonging to this family, such as TLR1, TLR2, TLR3, TLR5, TLR6, TLR7, TLR8, TLR9, and the mannose receptor, have been excluded as Stx receptors (41).

Because the role of TLR4 was not previously examined, the aim of this study was: 1) to investigate whether TLR4 is the receptor mediating the binding of Stx to human neutrophils; 2) to show the consequence of such an interaction in term of cytokine, growth factor, and chemokine production by neutrophils; 3) to compare the latter responses to those induced by LPS, the well-known and prototypical TLR4 agonist.

The Stx1 producer E. coli C600 (H19J) and the Stx2 producer C600 (933W) were supplied by Dr. Alison O’Brien (Department of Microbiology and Immunology, Uniformed Services University of the Health Sciences, Bethesda, MD). Stx1 and Stx2 were purified by receptor analog affinity chromatography, the former on globotriose-Fractogel (IsoSep, Lund, Sweden) (45) and the latter on (Galα1-4Galβ-O-spacer)-BSA-Sepharose 4B (Glycorex, Lund, Sweden) (19), followed in both cases by a passage through ActiClean Etox columns (Sterogene Bioseparations, Carlsbad, CA) to remove trace endotoxin contaminant. Stx1 and Stx2 preparations contained low amounts of LPS (< 6 or 7.5 ng/mg, respectively), as assayed by using the Limulus amebocyte lysate Pyrogent plus (Cambrex, Walkersville, MD). LPS from E. coli serotype O111:B4 (TLRgrade) was obtained by Alexis.

Highly purified neutrophils were isolated under endotoxin-free conditions from buffy coats of different healthy donors after centrifugation over Ficoll-Paque, followed by dextran sedimentation and hypotonic lysis of contaminating erythrocytes, as described previously (36, 46). Neutrophils were further enriched by positively removing any eventual contaminating cells, to reach 99.7% purity using the EasySep Human Neutrophil Enrichment Kit (Stemcell Technologies, Vancouver, BC, Canada). For binding experiments with Stx, Eppendorf tubes were precoated with PBS containing 1% endotoxin-low (≤ 1 endotoxin units/mg) BSA (Sigma-Aldrich, St. Louis, MO) to avoid nonspecific loss of the toxins (47). Neutrophils (5 × 10⁵ cells/ml) were incubated for 90 min at 37°C with Stx1 or Stx2 (60 nM) in 250 μl of the same buffer or with the indicated amount of LPS in 250 μl of PBS. Competitors present in Stx-binding assays were LPS (10 μg simultaneously present) or anti-TLR4 Ab 15C1 (provided by Dr. G. Elson, NovImmune, Geneva, Switzerland) and its isotype IgG1k control (2.5 μg, added 30 min before Stx1 stimulation) (48). After incubation, the cells were spun down at 200 × g for 5 min and washed three times with 100 μl of the incubation buffer at 37°C. The extent of Stx binding to neutrophils was assessed by flow cytometry as described below. Alternatively, neutrophils carrying Stx or LPS on their membrane were resuspended in RPMI 1640 containing 10% low endotoxin FBS (≤0.5 endotoxin units/ml; BioWhittaker, Walkersville, MD) at 37°C and incubated up to 20 h for cytokine determinations on culture supernatants or for RNA isolation (see below).

Stx1 or Stx2 bound on neutrophils were detected by flow cytometry as described previously (49). Neutrophils carrying Stx1 or Stx2 (see above) were incubated with a mouse mAb against Stx1 or Stx2 in the presence of human serum to saturate FcRs on polymorphonuclear leukocytes (PMN). After incubation with FITC-goat anti-mouse IgG, flow cytometric analysis was used to reveal the neutrophil-bound fluorescence. Cells were visualized by a cytogram that combined forward-scatter versus 90-degree side-scatter, and fluorescence was analyzed by a cytogram that combined 90-degree side-scatter and fluorescence and by a single-fluorescence histogram. Neutrophils were checked by staining with mAb to Ags associated with granulocytes (FITC-CD16 and FITC-CD65, Beckman Coulter, Miami FL). The mean channel value of fluorescence (MCV) of the single histogram was chosen to measure the extent of binding of Stx to PMN (49). The single values were calculated by subtracting the control MCV (range, 0.4–0.6)—that is, the MCV of neutrophils from the same donor incubated with primary and secondary Abs in the absence of the toxin. The same values (MCV = 0.4–0.6) were obtained if anti-Stx mouse mAbs were omitted in the assay in the presence of toxins and secondary Abs. The assay has been validated previously by comparing control subjects and patients with HUS in a double-blind fashion (49) and by challenging Stx-positive PMN with a negative control Ab (50).

CXCL8 present in culture supernatants from control and Stx1-, Stx2- or LPS-treated neutrophils was quantified by a specific ELISA kit (Quantikine Human CXCL8 Immunoassay; R&D Systems, Minneapolis, MN). For other cytokines, customized detection panels (IL-6, CXCL8, IL-10, G-CSF, CCL2, CCL4, TNF-α) were purchased from Bio-Rad (Hercules, CA). The assays were performed in 96-well plates by Multiplexed Luminex-based immunoassay following the manufacturer’s instructions, as described previously (51). Microsphere magnetic beads coated with mAbs against the different target analytes were added to the wells. After 30 min of incubation, the wells were washed and biotinylated secondary Abs were added. After incubation for 30 min, beads were washed and then incubated for 10 min with streptavidin-PE conjugated to the fluorescent protein, PE (streptavidin-PE). After washing, a minimum of 100 beads per analyte were analyzed using the BioPlex 200 instrument (Bio-Rad). The concentrations of the samples were estimated from the standard curve using a fifth-order polynomial equation and expressed as picograms per milliliter after adjusting for the dilution factor (Bio-Plex Manager software 5.0). Samples below the detection limit of the assay were recorded as zero. The intra-assay MCV was 15%.

RNA isolation and reverse transcription were accomplished as described previously (52). Quantitative RT-PCR (QRT-PCR) was performed using SYBR Premix Ex Taq (Takara, Mountain View, CA) and gene-specific primers (Life Technology, Carlsbad, CA) available in the public database RTPrimerDB (http://medgen.ugent.be/rtprimerdb/index.php) under the following entry codes: TNF-α (3551), CXCL8 (3553), CCL4 (3535), G-CSF (8615), IL-6 (3545) CCL2 (3540), IL-10 (8230), and GAPDH (3539). Data were calculated with Q-Gene software (http://www.gene-quantification.de/download.html) and are expressed as mean normalized expression (MNE) units after GAPDH normalization.

Continuous variables were described through means and SD or SE. Data analysis was performed with GraphPad Prism 5. We assessed differences between groups using the Student t test for continuous variables after controlling normality and homoscedasticity assumptions.

Human neutrophils were isolated and prepared under endotoxin-free conditions from buffy coats of healthy donors (46), and further enriched to reach >99.7% purity by positively removing any eventual contaminating cells (53). The Stx/neutrophil interaction was then measured by indirect flow cytometric analysis by using a mouse monoclonal IgG to Stx1 or Stx2 and a fluorescent secondary goat anti-mouse IgG (49). The results depicted in the representative experiments in Fig. 1A and 2A show that LPS, the prototypical agonist of TLR4, strongly inhibited the recognition of Stx1 and Stx2 by neutrophils. Consistently, also the anti-TLR4 Ab 15C1, known to block TLR4-mediated responses by monocytes and granulocytes efficiently (48), impaired the binding of both toxin variants to neutrophils, whereas an isotype control Ab did not elicit any significant effect. Such a significant inhibition was obtained with three different donors (means ± SD in Fig. 1B and 2B), under conditions allowing full saturation of receptors by Stx. It should be noted that previous competition experiments using large excess of agonists for several membrane TLRs (TLR1, TLR2, TLR5, TLR6) or for mannose receptor showed no effect on the binding of Stx1 to neutrophils by the same technique (41).

In addition to their canonical role as professional phagocytes, neutrophils challenged with different agonists are capable of producing an array of cytokines and chemokines, which play a pivotal role in orchestrating inflammatory and immune responses (54). Thus, we initially sought the presence of CXCL8 (by ELISA) in the culture media harvested from neutrophils challenged with Stx and, by comparison, LPS, the natural/specific TLR4 agonist. Accordingly, after treatment with the two exotoxins we observed a significant increase in CXCL8 release by neutrophils, as shown in the representative experiment depicted in Fig. 3A. It should be noted that these results have been obtained by exposing neutrophils to a short pulse with toxins, followed by incubation in toxin-free medium for up to 20 h. Consistent with the involvement of TLR4 in neutrophil stimulation by Stx, 15C1 mAb significantly inhibited the upregulating effects induced by Stx1 and Stx2 on neutrophil-derived CXCL8, whereas its isotype control Ab did not (Fig. 3A). Because the mechanism of action of 15C1 in neutralizing LPS signaling is supposed to involve the direct recognition of TLR4 through its Ag-combining sites and the contemporary engagement of CD32A through its Fc portion (48), we tested the effect of IV.3 mAb, which specifically recognizes CD32A (55). However, as shown in Fig. 3A, IV.3 showed no effect on CXCL8 induction by Stx1 or Stx2, thus ruling out the involvement of CD32A in toxin binding. On the other hand, data support and confirm that the dampening effect of the anti-TLR4 Ab on the Stx-dependent induction of CXCL8 should be ascribed only to its interaction with TLR4. The same results, expressed as percentages in Fig. 3B, were consistently obtained with five neutrophil preparations, which differed in the absolute basal level of released CXCL8.

Stx are bacterial products; therefore, an important point to address in interpreting the results from the latter experiments is the possible role of an eventual contamination by LPS of our Stx batches. The preparations of Stx1 and Stx2 used for these studies, in fact, showed trace amounts of endotoxin (<6 or 7.5 pg/μg of protein, respectively, as measured with Limulus amebocyte lysate), thereby accounting for the presence of ∼25–30 pg/ml LPS per assay when 60 nM Stx was added to neutrophils. However, no significant CXCL8 release by neutrophils stimulated with 100 pg/ml LPS was observed (Fig. 3A). Moreover, heat treatment (30 min at 95°C) of Stx totally blocked the Stx1-induced upregulating effect on CXCL8 release at 6 h (1% ± 0.2%, mean percent activity ± SD; n = 3), substantially blocked the stimulation induced by Stx2 (known to be more heat-resistant than Stx1; 36% ± 4%; n = 3) (56), whereas it was ineffective toward LPS (known to be heat-stable; 159% ± 48%; n = 3) (57). Furthermore, partially unfolded Stx1 molecules, which display negligible neutrophil-binding activity (9% ± 3%, mean ± SD; n = 3) with respect to the standard Stx1 preparations, but comparable amounts of contaminating LPS, did not properly stimulate any production of CXCL8 by neutrophils (8% ± 2%, mean ± SD; n = 3). The latter conformational changes of the toxin responsible for the phenomenon was serendipitously obtained after a purification procedure (Stx1 batch 9) or purposely provoked by multiple freezing and thawing cycles (42). It is worth noting that the loss in neutrophil-binding activity is related to changes in conformation of specific Stx1 moieties, because the partially unfolded toxin was found to express both enzymatic and Gb3Cer-binding activities, thus maintaining toxicity for Gb3Cer-bearing eukaryotic cells (42). Taken together, the data unequivocally exclude any involvement of contaminating LPS in the response of neutrophils to Stx.

To gain more information on the panel of cytokines and chemokines upregulated and released by neutrophils after stimulation by Stx and LPS, we subsequently used multiplexed Luminex-based immunoassays to test the same culture supernatants in which we measured CXCL8 for the presence of IL-6, IL-10, G-CSF, CCL2, CCL4, and TNF-α. In Fig. 4, the absolute cytokine concentrations measured in extracellular supernatants of neutrophils isolated from a representative donor are reported, whereas the mean fold increases of the stimulated cytokines with respect to untreated controls from five different neutrophil preparations are shown in Supplemental Fig. 1. At first glance, it is evident that the pattern of induced cytokines by Stx1, Stx2, and LPS is highly similar at qualitative level; in other words, clearly the same cytokines/chemokines were either upregulated or not induced by the toxins in stimulated neutrophils (Fig. 4 and Supplemental Fig. 1). CXCL8, CCL4, and TNF-α were significantly induced showing similar fold increases for all the tested toxins, whereas a time-dependent, non-significant positive trend was observed with G-CSF (Supplemental Fig. 1). The relatively high standard deviations are related to the different extents of stimulation induced by all the toxins in neutrophils isolated from different donors (Supplemental Fig. 1). CXCL8 and CCL4 stood out as the most relevant mediators, because they were produced at nanogram-per-milliliter concentrations after treatment with both Stx1 and Stx2, whereas the proinflammatory cytokine TNF-α was produced at lower concentrations (picograms per milliliter), as well as G-CSF (Fig. 4). Conversely, the other tested cytokines were not induced by challenged leukocytes in the presence of Stx or LPS (Fig. 4); therefore, the mean fold increases of IL-6, CCL2, and IL-10 are not shown. Treatment of neutrophils with lower concentrations of LPS (100 pg/ml), as control for endotoxin contamination of Stx, showed no stimulation at 90 min and 20 h for all the tested cytokines/chemokines, thus excluding artifacts in the determination of the stimulation patterns triggered by Stx (Fig. 4).

In conclusion, the results on the panel of neutrophil-upregulated cytokines constitute a further convergence for Stx and LPS acting on the same neutrophil receptor, namely TLR4.

We subsequently performed cytokine gene expression studies by QRT-PCR in neutrophils stimulated for 90 min and 4 h with either Stx1 or LPS. As shown in Fig. 5, the more stimulated and quantitatively expressed mRNAs by both toxins at 90 min and 4 h were those encoding for CXCL8 and CCL4, followed by TNF-α and to a lesser extent by G-CSF, in line with the results obtained at protein level (Fig. 4). On the other hand, no induction of IL-6, CCL2, or IL-10 mRNAs (Fig. 5) was detectable in activated neutrophils, as expected (Fig. 4). The same production pattern was observed with Stx2 (Supplemental Fig. 2). More importantly, the addition of anti-TLR4 mAb 15C1 once again abrogated the upregulatory effects of LPS, Stx1, and Stx2 on cytokine mRNA induction (Fig. 5 and Supplemental Fig. 2), as well as commercially available (HM2247; Hycult Biotech, Uden, The Netherlands) anti-TLR4 mAb 3C3 (data not shown), in keeping with the notion that TLR4 is engaged by LPS and by both exotoxins.

The data provided by the current study clearly demonstrate that the interaction between human neutrophils and Stx is mediated by TLR4. The first line of evidence is based on competition binding experiments in the presence of exogenous LPS or anti-TLR4 neutralizing Abs performed by indirect flow cytometric analysis. Because some commercial LPS preparations have been found to be contaminated by lipopeptides, which are sensed by TLR2 (58), we used ultrapure TLR-grade LPS. However, minimal contaminations with lipopeptides should have not interfered in our assays because it is known that engagement of TLR1, TLR2, TLR5, and TLR6 with proper ligands (dipalmitoylcysteinylseryl-[lysyl]₄; tripalmitoylcysteinylseryl-[lysyl]₄, peptidoglycan, lipoteichoic acid and flagellin) show no interference with Stx1 binding to neutrophils (41). Because Stx are not internalized by neutrophils (35), the involvement of the intracellular TLR7, TLR8, and TLR9 (59) as toxin receptors has been excluded. Finally, TLR3 is not expressed by human neutrophils (60).

The second line of evidence supporting our conclusions is based on the comparison of neutrophil response to Stx1, Stx2, and LPS, such as the release of mediators orchestrating innate immunity. For example, LPS, through the stimulation of TLR4 on neutrophils, represents an effective inducer of chemokines (CXCL8, CCL4), proinflammatory cytokines (TNF-α), and growth factors (G-CSF) (54, 61). To our knowledge, we show for the first time that an identical pattern of neutrophil-derived mediators are expressed at mRNA level (by QRT-PCR) and translated into proteins (by ELISA and multiplexed Luminex-based immunoassays) by Stx-treated neutrophils. Both Stx variants induced the release of CXCL8 and CCL4 at nanogram per milliliter concentrations and TNF-α and G-CSF at picogram per milliliter concentrations. Because HUS is associated with the inflammatory response (62), our in vitro experiments are in line with a proinflammatory scenario, in which neutrophils challenged with Stx might participate in their own recruitment via CXCL8 (63), as well as to the transmigration of monocytes and NK cells through CCL4 (64, 65). Furthermore, the behavior of Stx-treated neutrophils parallel to that of LPS-treated cells is confirmed also by the lack of Stx-mediated induction of IL-6, IL-10, and CCL2, which is consistent with our previous findings using LPS (66, 67). However, human monocytes are also targeted by Stx and, in turn, release large amounts (nanogram-per-milliliter concentrations) of proinflammatory mediators, such as IL-1β, TNF-α, IL-6, and CXCL8 (47). In contrast to human neutrophils (39), monocytes harbor, apart from TLR4, also the specific Stx-receptor Gb3Cer (47); therefore, toxin effects on these cells have simply been explained by its engagement. However, in the light of the present demonstration of the interaction of Stx with TLR4, it would be interesting to study the relative contribution of the two receptors in the upregulating effects induced by Stx on human monocytes. A cooperation between TLR4 and Gb3Cer in mediating the interaction of Stx with target cells has recently been demonstrated to occur in primary human endothelial cells and colon carcinoma cells (68). In this latter study, the depletion of TLR4 reduced the binding of Stx to cells, thus defining a clear role for TLR4 in facilitating the task of Gb3Cer. This finding further confirms our results for Stx binding to TLR4 and explains the lack of Stx internalization in cells that do not express Gb3Cer, such as neutrophils (35).

TLR4 plays a central role in the transduction of extrinsic proinflammatory signals in macrophages and, in particular, is central to the detection and response to LPS (59). Besides TLR4, the sensing machinery is composed of a soluble LPS-binding protein that facilitates the interaction of LPS with membrane CD14 that, in turn, transfers LPS to TLR4 and its related coreceptor MD-2. This sophisticated machinery allows the recognition of LPS at picogram-per-milliliter concentrations, although CD14 is not necessary when higher concentrations of LPS are present, as observed in CD14-deficient macrophages (59, 69, 70). Because elevated LPS concentrations have been used in our competition experiments, the direct involvement of CD14 in Stx recognition is highly unlikely. However, it should be noted that TLR4 discriminates between ligands by recruiting different coreceptors (59). Hence, the demonstration of TLR4 involvement in the binding of Stx1 and Stx2 might constitute just the first step in the identification of other molecules constituting a multifaceted receptor system, similar to LPS recognition. Apart from TLR4, the recruited coreceptor molecules might be different for the two Stx variants. Once activated by LPS, the receptor system can signal through MyD88-dependent and MyD88-independent pathways, culminating in the rapid activation of proper transcription factors (59). However, LPS responsiveness is regulated in a cell-specific manner, as clearly shown in LPS-treated human neutrophils in which the MyD88-independent pathway is not mobilized (71). Accordingly, with Stx as agonist, we observed the induction of proinflammatory cytokines and chemokines (TNF-α, CXCL8, and CCL4) synthesized and released upon the activation of the MyD88-dependent pathway.

Finally, to complete the identikit, it is mandatory to review and compare the features of TLR4 with those attributed to the Stx receptor in different studies. TLR4 is a highly glycosylated protein containing terminal α-2,3-sialyl residues attached to β-linked galactose (72). Similarly, the Stx receptor is recognized by a lectin that specifically binds to galactose residues (41). The human myeloid leukemia cell HL-60 differentiated along the granulocyte line by treatment with all-trans-retinoic acid or DMSO showed TLR4 expression to be slightly induced (71, 73); accordingly, the former treatment slightly induced the binding of Stx1 to otherwise refractory HL60 (36). Apoptosis is delayed in Stx-treated neutrophils that showed a prolonged lifespan (36) and, likewise, a large body of evidence showed that LPS by interacting with TLR4 can delay the neutrophil spontaneous apoptotic program (74). The copy number of TLR4 on human neutrophils is modest (75–77) and less than the number of neutrophil receptors for Stx (35, 36) and for the homologous ricin A chain (41). Because TLR4 expression increases after exposure to LPS or proinflammatory cytokines (60), one possible explanation for this discrepancy is that the known degranulating effect induced by Stx on human neutrophils (78) can rapidly enhance the surface exposure of TLR4 (79). In line with our hypothesis, TLR4 cell-surface expression in circulating neutrophils greatly increased in patients with HUS within 24 h of disease diagnosis, whereas TLR4 membrane expression did not change in circulating monocytes (80).

In conclusion, our study provides additional elements to foster the understanding of the pathophysiology of human STEC-induced diseases. Within 3–5 d after STEC ingestion, patients suffer from abdominal cramps and watery diarrhea. These gastrointestinal prodromal symptoms are related to the intimate adhesion of the bacteria to the gut mucosa rather than to toxin actions (3, 9). Only one third of the infected patients experience bloody diarrhea caused by Stx acting on the endothelial lining of the intestine (3, 9). Approximately 1 wk after the onset of bloody diarrhea, 10–25% of patients develop HUS, when Stx move from the gut to the kidney (3, 9). The question of why only a low percentages of infected patients manifest bloody diarrhea and later HUS remains unexplained. What role do the neutrophils play in these key points of STEC-infection pathogenesis? Studies performed in patients have shown that Stx–neutrophil interactions seem to be a double-edged sword, because the presence of neutrophils fully loaded with Stx confers protection against acute renal failure, while contributing to the risk of neurologic complications (31,37, 38). Understanding the real significance of Stx/PMN interaction (toxin delivery from the gut to the kidney or protective sponge effect) could have an important effect on the therapeutic strategies for preventing the onset of HUS. For this purpose, the identification of the Stx receptor on neutrophils appears relevant, because TLR4 polymorphisms have been described in humans, which could influence the frequency and course of infectious diseases (81, 82). TLR4 polymorphisms could also contribute to explain the individual susceptibility to hemorrhagic colitis and HUS, because STEC-infections in humans give rise to a clinical spectrum ranging from mild to severe manifestations (3, 9). Moreover, among susceptible individuals, the different abilities of neutrophils to interact with or respond to Stx through TLR4 could contribute to the non-mandatory transition from hemorrhagic colitis to HUS. A validation of these suggestive hypotheses awaits the search for specific TLR4 polymorphisms in patients with STEC-derived hemorrhagic colitis and HUS.

We thank Prof. Giuseppe Remuzzi (Istituto di Ricovero e Cura a Carattere Scientifico–Istituto di Ricerche Farmacologiche Mario Negri, Centro Anna Maria Astori, Bergamo, Italy) and Dr. Christine Betts (Dipartimento di Medicina Specialistica, Diagnostica e Sperimentale, Università di Bologna) for critical readings of the manuscript.

The authors have no financial conflicts of interest.