Staphylococcus aureus alpha toxin activates Notch in vascular cells

Abstract

Staphylococcus aureus infection is one of the leading causes of morbidity in hospitalized patients in the United States, an effect compounded by increasing antibiotic resistance. The secreted agent hemolysin alpha toxin (Hla) requires the receptor A Disintegrin And Metalloproteinase domain-containing protein 10 (ADAM10) to mediate its toxic effects. We hypothesized that these effects are in part regulated by Notch signaling, for which ADAM10 activation is essential. Notch proteins function in developmental and pathological angiogenesis via the modulation of key pathways in endothelial and perivascular cells. Thus, we hypothesized that Hla would activate Notch in vascular cells. Human umbilical vein endothelial cells were treated with recombinant Hla (rHla), Hla-H35L (genetically inactivated Hla), or Hank's solution (HBSS), and probed by different methods. Luciferase assays showed that Hla (0.01 µg/mL) increased Notch activation by 1.75 ± 0.5-fold as compared to HBSS controls (p < 0.05), whereas Hla-H35L had no effect. Immunocytochemistry and Western blotting confirmed these findings and revealed that ADAM10 and γ-secretase are required for Notch activation after inhibitor and siRNA assays. Retinal EC in mice engineered to express yellow fluorescent protein (YFP) upon Notch activation demonstrated significantly greater YFP intensity after Hla injection than controls. Aortic rings from Notch reporter mice embedded in matrix and incubated with rHla or Hla-H35L demonstrate increased Notch activation occurs at tip cells during sprouting. These mice also had higher skin YFP intensity and area of expression after subcutaneous inoculation of S. aureus expressing Hla than a strain lacking Hla in both EC and pericytes assessed by microscopy. Human liver displayed strikingly higher Notch expression in EC and pericytes during S. aureus infection by immunohistochemistry than tissues from uninfected patients. In sum, our results demonstrate that the S. aureus toxin Hla can potently activate Notch in vascular cells, an effect which may contribute to the pathobiology of infection with this microorganism.

Keywords: Alpha-toxin; HUVEC; Notch; Staphylococcus aureus.

Fig.1. HUVEC are an appropriate model to study Hla-induced Notch activation in EC in vitro

A) Over 99% of HUVEC express ADAM10 (left panel, red line) relative to unstained controls (blue) assessed by Flow cytometry. HUVEC also express Notch1 and Notch4 by Flow cytometry (middle and right panels), red line relative to isotype controls (blue). B) PCR of HUVEC cDNA expressing Notch1, 2 and 4 (top panel), while no amplification was observed in no-template controls (NTC) (bottom panel). C) Lactose dehydrogenase release indicates toxicity in vitro. LDH assays revealed significant toxicity starting at 1 µg/mL Hla or higher when compared to an untreated control (p=0.006), therefore Hla is not lethal on HUVEC at the doses used in follow-up experiments (0.1 µg/mL) or lower. D) ADAM10 is necessary for Hla toxicity in HUVEC: addition of 10μg/mL Hla induces toxicity (lane 6), but the same concentration in the presence of GI254023X (lane 5) brings LDH levels to baseline (lane1) and are not different to the genetically inactivated H35L levels (lane7, p=ns). E) While 1 μg/mL Hla increases LDH release relative to HBSS controls in non-targeting siRNA transfected HUVEC, this concentration does not induce LDH release in ADAM10 knockdown HUVEC (lane 4 vs 9, p<0.001). Representative results of two experiments, with technical triplicates, p value represents t test. Graphs represent mean and standard deviation.

Fig.2. rHla activates Notch1 in HUVEC, peaking between 30 and 45 minutes after Hla stimulation, at 0.01 µg/mL.

A) HUVEC were incubated with HBSS, increasing concentrations of rHla, or with 5mM EDTA for 10 minutes, followed by media incubation for 45 minutes. Quantification of cN1 after SDS-PAGE and immunoblotting indicates cN1 is not altered by the highest rHla concentration tested (0.1 µg/mL), but increased when incubated at lower concentrations (0.01 or 0.001 µg/mL), approximately 2 fold relative to HBSS or Untreated controls. In contrast, the positive control EDTA increased cN1 by 15 fold relative to control levels. B) HUVEC were treated with HBSS, 0.01 ug/mL Hla or 5mM EDTA for 10 minutes followed by different time periods on media. cN1 levels started to increase 15 minutes after Hla was removed, peaks between 30 and 45 minutes, and then starts to decrease 60 minutes after Hla incubation, returning to comparable to untreated or HBSS levels after that. Hla elicited approximately 2 fold increase in cN1 levels compared to HBSS controls. The positive control EDTA measured 30 minutes after stimulation shows an approximately 6 fold increase in cN1 levels compared to untreated or HBSS stimulated controls. Representative results of 3 independent experiments. Relative values of cN1 to β−actin are shown under each lane.

Fig.3. Hla induction of cN1 in HUVEC is specific.

A) HUVEC stimulated with HBSS (control), 0.01 µg/mL rHla, 0.01 µg/mL genetically inactivated Hla-H35L, or EDTA for 10 minutes and then incubated with media for 45 minutes. EDTA and rHla induced nuclear cN1 (upper right panel, white, yellow arrows), while the Hla-H35L mutant form fails to elicit cN1 in HUVEC (bottom right panel). B) cN1 Immunostain in DAPI positive nuclei shown in A) was quantified using ImageJ. Results indicate EDTA increased nuclear cN1 by four fold relative to HBSS controls, while Hla induced a two fold increase. C) Transfected HUVEC were stimulated for 15 minutes in the same conditions as in A, followed by 8 hours of media incubation and luciferase assay. The Luciferase Notch reporter indicates that the positive control EDTA increases Notch activation by approximately 5.5 fold, whereas rHla stimulates Notch by almost two-fold compared to HBSS controls; in contrast, Hla-H35L does not change Notch activation relative to controls. D) HUVEC stimulated in equal conditions for 10 minutes followed by 45 minutes in media prior to SDS-PAGE. rHla led to a 2-fold cN1 increase compared to HBSS controls, whereas EDTA revealed an approximately 6-fold increase. In comparison, the Hla-H35L construct had similar cN1 levels to HBSS. Representative results from three independent experiments are shown, *represents significance of 0.05, ** is less than 0.01; p values represent t test. Graphs represent meanand standard deviation.

Fig.4. ADAM10 and γ−secretase are necessary for Hla activation of Notch in HUVEC.

A) SDS-PAGE and immunoblot of HUVEC preincubated with DMSO, γ-secretase inhibitor (DAPT) or ADAM10 inhibitor (GI254023X), then stimulated with HBSS control, 0.01 µg/mL rHla, 0.01 µg/mL inactivated Hla-H35L, EDTA, or Hla with inhibitors for 10 minutes, followed by media for 45 minutes. Hla-induced cN1 levels (Lane2) were abrogated in the presence of DAPT and GI254023X (Lanes4 and 6). The EDTA positive control increased cN1 levels relative to HBSS controls (Lane7), B) HUVEC were preincubated with DMSO, DAPT or GI254023X, followed by DMSO, 0.01 µg/mLHla or 0.01 µg/mLHla with DAPT in HBSS for 15 minutes, fixed after 45 minutes. Hla increased cN1 (middle upper panel, white inset) compared to vehicle DMSO (Upper left panel inset), and the EDTA positive control increased cN1 as expected (upper right panel, inset). Both DAPT and GI254023X abrogated cN1 upregulation in response to Hla (Bottom left and middle panels; yellow arrows). Representative images of two independent experiments C) Quantification of B) revealed a two fold increase in cN1 levels after Hla stimulation relative to HBSS controls which was reversed in the presence of DAPT and GI254023X. EDTA increased cN1 levels by 5.5 fold relative to DMSO. D) HUVEC were transiently transfected with siRNA against ADAM10 or non-targeting, and 96 hours later stimulated as described in 2A prior to fix and staining. Despite moderate toxicity, cN1 was upregulated by Hla and EDTA compared to HBSS in control transfected cells (top panels) and ADAM10 siRNA markedly reduced cN1 presence in both Hla (lower left panel, yellow arrow) and EDTA treated cells (lower right panel). Cleaved Notch1 was pseudocolored white following Maximum Projection and No-neighbors deconvolution in Slidebook. E) SDS-PAGE of HUVEC transfected and stimulated as in 4D), and proteins were harvested after 48 hours. ADAM10 knockdown prevented Hla induction of cN1 as observed in non-targeting siRNA HUVEC (Lane 5 vs Lane 2). No difference was observed in EDTA stimulated HUVEC at 48 hours. Representative blots of three independent experiments, p values represent t test. Graphs represent mean and standard deviation.

Fig.5. Notch is upregulated in retinal EC of Notch reporter p6 pups treated with rHla, and in ex vivo aortic rings.

A) Littermate Notch reporter mice were injected daily with rHla subcutaneously from p2 to p5, and retinas harvested at p6 were stained with Isolectin B4 (white) and mounted with DAPI (Blue). Confocal images of whole mounts were taken at a 40X magnification, and mean YFP intensity (green) in areas overlapping Isolectin-B4 positive and DAPI positive areas (red arrows) were quantified using Image J. Representative images of 5–10 μM stacks are shown. Top panel shows overlay of lectin (White), Notch (Green), and nuclei (Blue). Bottom panels show Notch in that field (green), red arrowheads indicate activated Notch in nuclei of blood vesssels is higher in Hla treated pups (right panels). B) Quantification of the Mean Grey Intensity in the GFP channel (Notch) in DAPI positive nuclei revealed a 69% increase in Notch activity in Hla treated mice compared to H35L controls. For quantification purposes, each layer of each picture was quantified separately and the average of those layers is shown. Representative results of two independent litters are shown. C) Sprouts from aortic rings from Notch reporter mice incubated with 0.01 µg/mL H35L, Hla, or Hla with GI254023X for 5 days and stained for the endothelial marker Lectin (white) resulted in tip cells with activated Notch signal (middle panel, green, yellow arrow), but not in aortic rings incubated with H35L (left panel). Addition of the ADAM10 inhibitor GI254023X eliminated all Notch signal from the sprouts (right panel). Eight aortic rings per group with 6 to 10 sprouts per bead were counted and the percentage of positive tips per aortic ring is shown. Man Whitney test indicates a pvalue of <0.0001. Representative results of two independent experiments. Maximum projections from 30µM thick stacks stacks are shown. Graphs represent mean and standard deviation.

Fig.6. S. aureus derived Hla toxin activates skin endothelial Notch activation.

Notch reporter mice received subcutaneous S. aureus secreting Hla (WT Hla) or lacking Hla (ΔHla). A) The epidermal and dermal layers adjacent to the infection site were analyzed 36 hours after infection by histology, confocal images show increased Notch reporter (YFP) signal in skin infected with WT S. aureus compared to ΔHla. Representative images are shown. B) Quantification of the YFP signal in the skin adjacent to the infection area shows a significantly higher YFP intensity/DAPI area (p<0.01) and a higher YFP area/DAPI area (p<0.005). D) IHC of skin 16 days after infection shows sustained Notch activation (left panel, arrows, Green) in EC (red) of WT Hla skin compared to ΔHla strains with much lower Notch activation in EC (right panel, arrowheads, red). Graphs represent mean and standard deviation.

Fig.7. Notch is increased in EC of human liver after S. aureus infection.

Liver sections from a patient with S. aureus infection (right panels) had increased Notch expression (red) in PECAM-positive EC (green, white arrows), compared to a non-infected control (left panels). The red blood cells (RBC) indicate the vessel lumen.