Molecular Damage and Induction of Proinflammatory Cytokines in Human Endothelial Cells Exposed to Shiga Toxin 1, Shiga Toxin 2, and α-Sarcin

The prototype Stx1 and Stx2 producers E. coli C600 (H19J) and C600 (933W) were kindly supplied by Alison O'Brien (Department of Microbiology and Immunology, Uniformed Services University of the Health Sciences, Bethesda, MD). Stx1 was purified by receptor analogue affinity chromatography (25) on globotriose-Fractogel (IsoSep AB, Lund, Sweden). Stx2 was obtained according to the method described by Downes et al. (5). l-[4,5-³H]leucine (64 mCi/mmol, 1 μCi/ml) was obtained from Amersham Pharmacia Biotech (Bucks, United Kingdom). Disposable plastics for laboratory use were obtained from Costar (Broadway, Cambridge, MA). α-Sarcin and analytical-grade reagents were purchased from Sigma Chemical Co. (St. Louis, MO).

HUVEC cultures (15) were kindly provided by Janette A. M. Maier (University of Milan, Milan, Italy). HUVEC were cultured as previously described (3). The cells used in the present study were all of early passages and were checked for the expression of von Willebrand factor by immunocytochemistry using a rabbit anti-human von Willebrand factor (Dako, Milan, Italy) as the primary antibody.

Protein synthesis was measured as the rate of incorporation of labeled leucine during a 30-min incubation of the cell monolayers in the complete medium described above containing 0.4 mM leucine and trace amounts of [³H]leucine. This procedure has been described in detail elsewhere (24).

The fast halo assay was carried out as described previously (28). Briefly, after the treatments, the cells were resuspended at 4.0 × 10⁴/100 μl in ice-cold phosphate-buffered saline (8 g of NaCl, 1.15 g of Na₂HPO₄, 0.2 g of KH₂PO₄, and 0.2 g of KCl per liter) containing 5 mM EDTA: this cell suspension was diluted with an equal volume of 2.0% low-melting-point agarose in phosphate-buffered saline and immediately sandwiched between an agarose-coated slide and a coverslip. After complete gelling on ice, the coverslips were removed, and the slides were immersed in 0.3 M NaOH for 15 min at room temperature. Ethidium bromide (10 μg/ml) was directly added to NaOH during the last 5 min of incubation. The slides were then washed and destained for 5 min in distilled water. The ethidium bromide-labeled DNA was visualized by using a Leica DMLB/DFC300F fluorescence microscope (Leica Microsystems, Wetzlar, Germany), and the resulting images were digitally recorded on a PC and processed with an image analysis software (Scion Image). The extent of strand scission was quantified by calculating the nuclear diffusion factor (NDF), which represents the ratio between the total area of the halo plus nucleus and that of the nucleus. The data are expressed as relative NDF values (rNDF), calculated by subtracting the NDF value of control cells from that of treated cells.

Adenine release from DNA was measured by using as substrate the 2,251-bp [³H]DNA labeled in the purine ring of adenine obtained by PCR amplification of the 731-2981 region of the pBR322 plasmid (2). Enzymatic reactions were performed in 150 μl of 50 mM sodium acetate buffer (pH 4) containing 0.3 μg of substrate corresponding to 205.5 pmol of adenine having a specific radioactivity of 3,000 dpm/pmol and 1 pmol of Stx1 or Stx2. After 40 min at the indicated temperatures, DNA molecules were removed by passing the samples through Bond Elut NH2 columns as previously described (2), and the combined flowthrough and washing was measured by liquid scintillation counting.

The assay we used for the detection of caspase 3 (Caspase Colorimetric assay kit; MBL International Corp., Watertown, MA) is based on the spectrophotometric detection of the chromophore p-nitroanilide (pNA) after cleavage from the labeled substrate DEVD-pNA. The ratio between the absorbance of pNA from the apoptotic sample and that from an untreated control gives the fold increase in caspase 3 activity (relative activity). The assay was performed as described earlier (3).

HUVEC cells from 6-cm-diameter tissue culture dishes were harvested by lysis using the p38 MAP kinase assay kit (Cell Signaling Technology, Inc., Danvers, MA) according to the directions of the manufacturer. The assay was performed on 200 μg of protein extracts, to which immobilized monoclonal antibody specific for phosphorylated p38 MAP kinase was added. After an overnight incubation at 4°C, immunoprecipitates were repeatedly washed with the lysis buffer and with the kinase buffer supplied in the kit and then resuspended in 50 μl of the latter solution supplemented with 200 μM ATP and 1 μg of the kinase substrate ATF-2 fusion protein. Kinase reaction mixtures were incubated for 30 min at 30°C, and then the reaction was terminated with sodium dodecyl sulfate-polyacrylamide gel electrophoresis sample buffer. ATF-2 phosphorylation was detected by Western blotting with the phospho-ATF-2 specific (Thr71) antibody.

IL- 8 and GM-CSF proteins present in culture supernatants from control and toxin-treated cells were quantified by enzyme immunoassay using standards ranging from 31.25 to 2,000 pg/ml and from 7.8 to 500 pg/ml, respectively, according to the manufacturer's instructions (Quantikine human IL-8 and human GM-CSF immunoassay; R&D Systems, Inc., Minneapolis, MN).

In Fig. 1 the rate of protein synthesis in HUVEC treated with the two bacterial toxins (0.01 nM) was measured and compared to that obtained after incubation of the same cells with the fungal ribotoxin α-sarcin (10 μM) acting on the same rRNA sequence. A relatively high concentration of α-sarcin was used in these experiments, because it lacks a specific receptor and slowly enters the cells by endocytosis through acidic endosomes (1). With both Stxs the inhibition of protein synthesis started early (90 min) and reached a plateau value approaching 80% within 6 h. It is of note that the maximal level and the time course of the antiribosomal activity of Stx1 and Stx2 essentially coincided. In the case of α-sarcin the inhibition of translation progressively increased over 16 h of incubation, reflecting a slow, non-receptor-mediated internalization, and finally reached a plateau equal to 70%.

Since Stx1, in addition to depurinating 28S rRNA within ribosomes, has been shown to remove adenines from nuclear DNA in HUVEC (3), the closely related Stx2 was checked for the capacity to damage DNA. The effect of the toxins on nuclear DNA was studied by means of the fast alkaline halo assay (27, 28). With this technique, DNA fragments resulting from the cleavage of DNA itself diffuse out of the nuclear cage as an inverse function of their size, thus producing a concentric halo whose radius reflects the extent of DNA damage (Fig. 2): i.e., on microscopic observation (i) the smaller the fragments (highly damaged DNA) the greater the halo and (ii) the longer the fragment (intact DNA) the smaller the halo. It should be noted that the single-stranded DNA fragments may result from a direct breakage of DNA or from the presence of apurinic sites converted to DNA single-strand breaks at alkaline pH. Halo formation can be conveniently monitored at the single cell level with fluorescence microscopy, as seen in the representative micrographs shown in Fig. 2. As expected (3), Stx1-treated cells, unlike control cells (Fig. 2A), showed well-defined halos (Fig. 2B) indicative of DNA breakage. More interestingly, the same phenomenon was observed in Stx2-intoxicated cells (Fig. 2C). Consistently, both bacterial toxins in vitro depurinated at similar rates a [³H]DNA labeled in the purine ring of adenine (Table 1). To our knowledge, this is the first evidence that Stx2 damages nuclear DNA in human vascular endothelial cells: hence, both Stx1 and Stx2 could be considered N-glycosylases with broad substrate specificity, since the two ribotoxins remove the same single specific adenine from rRNA (6) and multiple adenines from nuclear DNA. The induction of nuclear damage as a function of intoxication time has also been studied (Fig. 1): Stx2 demonstrated a steady but steep increase over the 6 h, whereas Stx1 is slow to start in the first 3 h and then had a steep increase in the second 3 h. After 6 h, however, the difference between the amount of DNA lesions accumulated in Stx1-intoxicated HUVEC {rNDF [expressed as the mean ± the standard error of the mean (SEM)], 4.2 ± 0.77} and in Stx2-treated cells (rNDF, 3.1 ± 0.59) was not statistically significant (Fig. 1). α-Sarcin had no effect on nuclear DNA (Fig. 2D and Fig. 1) even after prolonged incubation (40 h). This finding leads to the conclusion that damage to nuclear DNA observed with Stxs is independent from their ribotoxic activity since α-sarcin induced similar ribosomal lesions in the absence of primary DNA damage. Accordingly, Olmo et al. did not show any direct DNA damage in rhabdomyosarcoma cells exposed to the fungal toxin, whereas internucleosomal DNA fragmentation secondary to apoptosis has been shown to occur in these cells (22). However, the latter finding was not observed in α-sarcin-treated HUVEC (not shown) which, consistently, barely activated caspase 3 (Fig. 1) and showed a very low apoptotic response. It is not surprising that very different cell types (tumor cells and endothelial cells) might respond differently to the same stimulus. Thus, α-sarcin should be considered a pure ribotoxin targeting uniquely ribosomes sparing DNA. It was therefore used as a tool to assess the effects induced on gene expression by ribotoxic stress (see below and Fig. 4 and 5) in the absence of any direct effect on the nucleus.

DNA lesions induced in HUVEC by Stxs are the result of a direct action of the toxins and are not secondary to the onset of apoptosis: indeed, Fig. 1 also shows that the DNA breaks appeared before the onset of the execution phase of apoptosis (as assayed with caspase 3 activation). As previously described (3), with Stx1 the caspase activity increased ∼3-fold after 16 h of treatment, i.e., more than 10 h after the maximum nuclear damage observed. With Stx2, the same threefold increase in caspase 3 activity occurred about 3 h after the onset of detectable DNA injuries. Notably, the kinetics of Stx-induced DNA damage are similar to the kinetics of caspase activation offset by 3 h (Fig. 1). This is not unexpected since Stx2 induces DNA lesions (which are known to be an apoptogenic stimulus) with more rapid kinetics than does Stx1. Moreover, Stx2, in contrast to Stx1, has been found to interact with mitochondrial Bcl-2, resulting in faster caspase 3 activation (30). As for α-sarcin, which is not DNA damaging (Fig. 1 and 2), it is noteworthy that the strong inhibition of translation induced by the fungal toxin acting on the same crucial sequence of rRNA targeted by Stxs was not sufficient per se to trigger the sustained activation of caspase 3 in the absence of DNA damage.

The response of HUVEC to toxin-induced stress was evaluated by monitoring the time course of the phosphorylation of ATF-2 fusion protein by p38 MAP kinase (Fig. 3A). It is well known that stress activated protein kinases are activated by specific ribosomal lesions (12, 29); accordingly, the pure ribotoxin α-sarcin triggered p38 MAP kinase activity in HUVEC in the absence of DNA damage. Moreover, the activation of the protein kinase clearly paralleled the inhibition of protein synthesis (Fig. 1) starting after 3 h of treatment, i.e., when protein synthesis in HUVEC was markedly inhibited by the two bacterial toxins (>50%, Fig. 1). In the case of α-sarcin the delayed inhibition of protein synthesis was accompanied by a delayed (6 h) and transient activation of p38 MAP kinase, declining at 24 h. Quantitative analysis of the immunoblot results by densitometric scanning showed differences between the two bacterial toxins in activating p38 MAP kinase, since phospho-ATF-2 levels are significantly higher in Stx1-treated cells at 3 and 6 h (Fig. 3B). Since protein synthesis was equally inhibited by the two bacterial toxins (Fig. 1), these differences probably reflect diverse mechanisms of binding to the substrate rather than different levels of 28S rRNA damage. It is well known that some antibiotics are capable of activating stress kinase response by binding to 28S rRNA (12).

All of the toxins tested on HUVEC under the conditions described above caused the induction of IL-8, a key chemokine in HUS pathogenesis, at times subsequent to p38 MAP kinase activation. As shown in Fig. 4, the ELISA determinations performed on the culture supernatants were negative at 6 h, whereas the chemokine release was detectable after 12 h and peaked at 24 h. There was a significant difference between the two bacterial toxins (Stx2 > Stx1) in the induction of IL-8 at 24 h.

When the same culture supernatants were checked for the presence of GM-CSF (Fig. 5), the release of the cytokine was found only with Stx2, as previously demonstrated by Matussek et al. (16), whereas Stx1 and α-sarcin were apparently ineffective. Again, the time course of GM-CSF induction by Stx2 clearly paralleled that observed with IL-8. Since GM-CSF was undetectable in control cells, we hypothesized the existence of a masked stimulation with Stx1 and α-sarcin at just detectable values of GM-CSF. At lower concentrations of Stxs (2 pM), the extent of ribosomal lesions in HUVEC was lower, protein synthesis was only 50% inhibited, and IL-8 and GM-CSF inductions were more pronounced (Table 2). Under these conditions, Stx1-treated HUVEC secreted detectable amounts of GM-CSF. Again, Stx2 clearly had stronger effects on gene expression than did Stx1, since the total amounts of cytokines induced by this toxin are twofold (IL-8) or fourfold (GM-CSF) higher than those secreted in the presence of Stx1. The results show that the same level of ribotoxic stress imposed by Stx1 or Stx2 (Fig. 1 and Table 2) led to quite different levels of gene expression (Fig. 4, Fig. 5, and Table 2). These findings cannot be accounted for solely in terms of pure inhibition of protein synthesis or MAP kinase activation. A further regulation independent from the ribotoxic stress was assumed to be operative in intoxicated cells. It should be noted that, although DNA lesions were in both cases delayed compared to RNA lesions (Fig. 1), they emerged before the upregulating effects triggered by Stxs (Fig. 4 and 5).