Abstract
Background Although the effects on human organs by shock waves (SWs) induced by medical treatments or high-energy trauma are well recognized, little is known about the effects on the cellular level. Since blood vessel injury is a common finding after SW exposure, we assessed the in vitro effects of SWs on human umbilical vein endothelial cells (HUVECs).
Methods An in vitro trauma model was used to expose HUVEC monolayers to focused SWs or to shock waves plus cavitation (SWC), a subsequent phenomenon that is often considered the main cause of SW vascular injury.
Results SWs alone did not cause any changes in the studied variables. In contrast, HUVEC monolayers exposed to SWC exhibited discrete central lesions with extensive cell death. Cells peripheral to the main lesion area displayed disassembly of dense peripheral bands and formation of actin stress fibers, indicating increased intercellular gaps. Expression of P-selectin was enhanced 11-fold compared with controls, whereas expression of E-selectin and intercellular adhesion molecule 1 was enhanced 8-fold (p < .05) and 1.5-fold (p < .01), respectively. The latter responses were preceded by nuclear translocation of nuclear factor κB subunit p65 by 16% (p < .01). When compared with mechanically produced lesions used as controls, SWC lesions exhibited an impaired regeneration rate of the endothelial cell layer (p < .001). Redistribution of centrosomes toward the lesion borders was less effective in the SWC samples compared with the mechanically produced lesions (p < .01).
Conclusions SWC lesions were associated with a switch to an endothelial proinflammatory phenotype, with an impaired regeneration rate and changes in cytoskeletal functions.
Various medical treatments, for example, extracorporeal renal shock wave lithotripsy (ESWL), and high-energy trauma, such as explosions, expose the human body to shock waves (SWs). A major consequence of SWs is tissue injury, especially to gas-containing organs (ie, the middle ear, lungs, and large intestine), but also large blood vessels and the microcirculation in parenchymatous organs.1-3The ensuing injuries, ranging from rupture to hemorrhages, thrombus formation, and tissue edema, may proceed to multiple organ failure. In particular, blood vessel injuries have been attributed to perturbation and detachment of endothelial cells consistent with a phenotype switch from the quiescent noninflammatory and anticoagulative state to the opposite, a proinflammatory and procoagulative state.2,3
The vascular endothelium forms the interface between blood and surrounding tissue. When SWs traverse such boundaries (ie, tissues with different acoustic impedance), tensile forces are generated. If these tensile forces exceed the tensile strength of blood, cavitation (ie, rapidly imploding microscopic bubbles) is induced.4Cavitation bubbles generate liquid jets and high local temperatures and are generally considered to cause SW-associated endothelial injury.5-7
The pivotal role of the endothelial layer in the regulation of inflammatory processes, coagulation, and vascular permeability is well documented. Here we sought to assess the effects of SWs on endothelial cell functions related to regulation of inflammatory processes and regeneration of the endothelial monolayer. To this end, we used an established experimental method, the flyer-plate model,8in which we exposed endothelial monolayers to SWs alone or shock waves plus cavitation (SWC).
We assessed the effects of SWs or SWC on the expression of adhesion molecules for the attachment of leukocytes and platelets, namely, P-selectin, E- selectin, and intercellular adhesion molecule 1 (ICAM-1). Given that de novo synthesis of the inducible adhesion molecules E-selectin and ICAM-1 depends on the activation of nuclear transcription factor κB (NF-κB),9we assessed translocation of NF-κB subunit p65 from the cytosol to the nucleus. Furthermore, we evaluated the association between NF-κB and adhesion molecule induction by studying the effects of pyrrolidine dithiocarbamate (PDTC), a NF-κB inhibitor, on E-selectin expression.
Because the endothelial cytoskeleton, that is, actin (microfilaments), tubulin (microtubules), and vimentin (intermediate filaments), is essential for organization of surface receptors, monolayer permeability to fluids5,10and leukocytes,11cell shape changes, and cell motion during endothelial repair,12we studied the structure and redistribution of the cytoskeletal components after exposure to SWs or SWC. The morphologic and cytoskeletal changes in human umbilical vein endothelial cells (HUVECs), believed to be of significance for reendothelialization (ie, lamellipodia extrusion and centrosome redistribution toward the injured area), were also studied and compared with changes in mechanically induced lesions (MLs).
MATERIALS AND METHODS
Materials
Murine monoclonal antibodies to P-selectin (immunoglobulin [Ig]G1, 31791A), E-selectin (IgG1, 555648), and von Willebrand's factor (vWF) (IgG1, 555849) were from BD Pharmingen (San Diego, CA), ICAM-1 (IgG1, MCA532) was from Serotec (Oxford, England), and vimentin (IgG1, 074M) or tubulin (IgG1, 089M) was from BioGenex Laboratories (San Ramon, CA). Rabbit polyclonal antibodies to human NF-κB (PC137) was from Calbiochem (San Diego, CA). Fluorescein isothiocyanate (FITC)-conjugated AffinoPure F(ab′)2 fragment goat antimouse IgG (711-096-152) was from Jackson Immuno Research Laboratories (West Grove, PA).
The microscopes used were Nikon Diaphot 300 (for densitometric measurement in polystyrene vials and morphology using phase contrast) and Nikon Microphot FX (for fluorescence microscopy) (Nikon Corp., Tokyo, Japan). Microphotographs were taken with a Nikon CoolPix 990 (Nikon Corp.). Image analysis was performed with NIH Image software, version 1.61 (Bethesda, MD), on an Apple Macintosh computer, PowerBook G4 (Apple Inc., USA).
Preparation of HUVECs
HUVECs were harvested and cultured on glass cover slips (mean 13 mm, 2 mm thick; Labora Chemicon, Stockholm, Sweden), as described elsewhere.13,14Confluent monolayers from passages 2 and 3 were used for experiments. HUVECs were characterized by their cobblestone appearance and positive immunocytochemical staining for vWF.
Generation of SWs
SWs were generated with the flyer-plate model, described in detail elsewhere.8,14The experimental setup is shown in Figure 1. SW peak amplitude was 23 ± 5.9 MPa (mean ± standard error of measurement [SEM]). Vials used for SW exposure were polystyrene 24-well cell culture dishes (Nunc, Roskilde, Denmark). In each well, HUVECs were situated on a glass coverslip resting on a circular, 1.2 mm-thick, stainless steel ring (No 08281, Erlandsons Brygga, Stockholm, Sweden) with the cells facing upward. During SW exposure, the wells were filled with 0.4 or 2.4 mL Hanks' Balanced Salt Solution (HBSS)-1% albumin (Life Technologies, Paisley, United Kingdom). From previous experiments, the former volume was known to be associated with the generation of surface cavitation in the well, whereas the latter was not.14Thus, HUVECs were exposed to identical SWs with the absence or presence of cavitation. When present, surface cavitation was detected on the video monitor attached to the stereomicroscope and registered as droplets on the transparent lid placed on top of the vial. Wells with 0.4 mL, in which no cavitation was registered, were excluded from experiments.
HUVECs exposed to SWs or SWC and sham-exposed controls were kept in a cell incubator at 37°C until fixation. Monolayers examined for cell morphology and cytoskeletal changes were fixed and examined 30 minutes or 4 hours after their exposure. Given the known temporal appearance of P-selectin, E- selectin, and ICAM-1 on the endothelial cell surface,5,15cultures assayed for these adhesion molecules were fixed 7.5 minutes, 4 hours, or 6 hours postexposure, respectively. Similarly, cultures analyzed for NF-κB translocation were fixed at 30 minutes, considering our own pretrial experiment and the results of other studies.9,16
Mechanical Wounding
Mechanical wounding in vitro is an established method for studying processes involved in endothelial repair.12,17,18Circular endothelial lesions, of approximately the same size as obtained with the flyer plate model,14were accomplished by mechanical scraping with a wooden peg. Subsequently, cultures were kept in a cell incubator at 37°C prior to examination. The endothelial lesions were photographed, and their surface areas were measured using NIH Image at 30 minutes and 4 hours (n ≥ 9 in each group). After photography, the cultures were fixed in 95% ethanol for 5 minutes and stained for cytoskeletal elements.
Morphologic Examination
HUVEC monolayers, stained for P-selectin, E-selectin, and ICAM-1, were examined with phase contrast microscopy. Some cultures were counterstained with hematoxylin for further morphologic examination. Additionally, loss of membrane integrity was examined with trypan blue (0.2%). Morphometry was performed on micrographs using NIH Image (n ≥ 8 in each group).
Immunocytochemistry
Cytoskeletal assessment was performed in samples exposed to SWs or SWC and in mechanically accomplished lesions. The staining method of tubulin and vimentin has been described in detail elsewhere.19Briefly, cells were fixed with methanol for 5 minutes at −20°C and permeabilized with acetone at −20°C for 10 seconds. Cells were incubated with primary antibodies to tubulin or vimentin (dilution 1:20 or 1:250, respectively) and subsequently with FITC-conjugated secondary antibodies (dilution 1:20 or 1:50, respectively). The specimens (n ≥ 8 or more in each group) were examined by fluorescence microscopy, and areas of interest were photographed.
In tubulin-stained cultures, cells at the lesion border were examined for the position of the centrosome. The cell nucleus and the lesion border were used as reference points, as previously described.20Briefly, the centrosome was classified as “toward,” “middle,” and “away” with respect to the nucleus and the side closest to the lesion area. At least 50 cells in each well were analyzed to determine the percentage of cells with centrosomes in each of the three locations (n ≥ 5 in each group).
The immunofluorescence staining of F-actin was performed as follows: HUVECs were fixed in 4% paraformaldehyde for 15 minutes and washed three times in phosphate buffered saline (PBS). Cells were permeabilized with 0.1% Triton X-100 for 3.5 minutes, washed three times in PBS, and stained with tetramethylrhodamineisothiocyanate (TRITC)-phalloidin for 30 minutes at room temperature. Samples were examined with a fluorescence microscope, and areas of interest were photographed (n ≥ 8 in each group).
Immunostaining and Measurement of P-Selectin, E-Selectin, ICAM-1, and NF-κB
HUVECs exposed to SWs or SWC and sham-exposed controls were fixed in 4% paraformaldehyde for 15 minutes. To examine the involvement of NF-κB for adhesion molecule induction, 100 μM of PDTC (Sigma, Stockholm, Sweden), a NF-κB antagonist known to inhibit nuclear translocation of NF-κB,21was added to some wells 30 minutes prior to SW exposure. HUVECs incubated with 4-phorbol 12-myristate 13-acetate (1 × 10−5 M) (Sigma) in HBSS-1% albumin were used as positive controls. After rinsing with PBS, fixed samples were kept over night at 4°C in 5% nonfat dry milk in PBS. Without washing, cells were labeled with specific monoclonal antibodies (1 μg/mL) to the antigen examined (1.5 hours, room temperature). Following washings in PBS, primary antibodies were visualized with a BioGenex Fast Red Detection kit (AA000-5M; BioSite, Sweden) and examined under a microscope. Using fixed camera and microscope settings, the wells were photographed. Densitometric quantification of the antigen expression was then performed with NIH Image. Data given, expressed as pixel density, were calculated by subtracting pixel density in blanks from the value measured (n ≥ 8 in each group).
The staining procedure for NF-κB has been presented in detail elsewhere.16As described above for tubulin and vimentin, cells were fixed in methanol and permeabilized in acetone. After incubation with 1.5% goat serum for 20 minutes, cells were incubated with antibodies to NF-κB subunit p65. Immune complexes were detected with a Vectastain ABC kit and diaminobenzidine tetrahydrochloride (DAB) substrate for peroxidase (Vector). Subsequently, HUVEC cultures (n ≥ 6 in each group) were examined with a microscope and photographed. Densitometric assessment of the staining ratio, between the cytoplasm and the nucleus, was performed with NIH Image. More than 25 cells were analyzed on each micrograph.
Statistical Analysis
All measurements are expressed as mean ± SEM. Statistical analyses were performed with one-way analysis of variance Kruskal-Wallis statistics (Dunn's multiple comparison test) using the software PRISM (Graphpad Software, San Diego, CA). When only two groups were compared, the Mann-Whitney U-test was applied. A p value < .05 was considered statistically significant.
RESULTS
Effects of SWs
The cell morphology of monolayers exposed to SWs alone could not be distinguished from sham-treated controls at 30 minutes or 4 hours postexposure (not shown). Examination of actin, tubulin, and vimentin filaments at these time points revealed no effect of SWs (not shown), and no significant effects on P-selectin, E-selectin, or ICAM-1 expression could be demonstrated.
Effects of SWs and Additional Cavitation
Effects on Morphology and Cytoskeletal Redistribution
The effects of SWC on cell morphology were studied at 30 minutes and 4 hours after impact. Since mechanical wounding is an established method for studying endothelial repair in vitro,12,17,22MLs, studied at the same time points, were used as controls. As previously described,8exposure to SWC caused an immediate endothelial lesion in the HUVEC monolayers. This lesion could be divided into three discrete zones: a central area with total cell loss, a surrounding area (second zone) with signs of significant cell injury and variable cell loss, and a peripheral zone (third zone) with morphologically undamaged cells (Figures 2 and 3). The border between the second and third zones in SWC lesions (see the arrows in Figure 3) or the border between the cells and the denuded area in MLs is denoted as the lesion border.
In agreement with our previous report,14loss of cell membrane integrity in the second zone was confirmed with trypan blue staining. Using immunofluorescence, it was shown that SWC caused an immediate disassembly of actin, tubulin, and vimentin filaments in this area, changes that persisted at 4 hours (not shown). The cells in this zone detached over time, leaving this area more or less denuded at 4 hours. Morphometric assessment of the lesion area, that is, the central and second zones, revealed no significant changes between 30 minutes and 4 hours (4.31 ± 0.38 mm2 vs 4.35 ± 0.18 mm2) in the SWC lesions. The ML areas, however, were significantly larger when measured at 30 minutes compared with 4 hours, comprising 5.96 ± 0.37 mm2 and 5.15 ± 0.43 mm2, respectively (p < .001). This equals a regeneration rate of 22.5 ± 0.37 μm/h and is in agreement with regeneration rates for HUVEC cultures presented by others.23,24Thus, endothelial reendothelialization during the first 4 hours after injury was less efficient in the lesions caused by SWC compared with MLs, used as controls.
Cells in the third zone displayed a reduction in actin-dense peripheral bands (DPBs) and formation of stress fibers at 30 minutes (Figure 4C). These changes, known to be associated with increased intercellular permeability, for instance during inflammation,11were obvious in cells located several cell layers away from the lesion border and were not seen in MLs (Figure 4) or sham-treated controls (Figure 4). At 4 hours, stress fiber formation was less pronounced and the DPBs were again evident (Figure 5A). No effects on tubulin or vimentin filament organization were noted in the third zone.
As a response to the endothelial lesion, cells located at the lesion border began to elongate and extend lamellipodia toward the partly denuded area (second zone). As described by others, these changes were clearly seen in MLs as early as 30 minutes postexposure.23Although lamellipodia formation was likewise observed in SWC lesions at this time point, cells at the lesion border remained remarkably unchanged. These differences between the SWC lesions and MLs were even more distinct when cells were stained for cytoskeletal filaments at 4 hours, as illustrated in Figure 5.
Tubulin redistribution during endothelial repair requires relocation of microtubule organizing centers (ie, centrosomes) toward the lesion border.25Centrosome redistribution toward the lesion area was seen in cells at the lesion border, in SWC, and in ML samples. This redistribution was, however, delayed in SWC samples, at 30 minutes and 4 hours, as shown in Figure 6. Consequently, only 27.8 ± 1.1% of the centrosomes were located between the nucleus and the lesion edge in SWC lesions, whereas 60.0 ± 2.0% were found at this location in MLs at 30 minutes (p < .01). At 4 hours, the corresponding numbers were 47.8 ± 1.4% versus 69.6 ± 1.6% (p < .01). Thus, redistribution of cytoskeletal filaments, essential for cell spreading and important for efficient cell migration during endothelial repair,20,23was hampered in lesions caused by SWC.
Adhesion Molecules and NF-κB
Adhesion molecules for the attachment of leukocytes and platelets, namely, P-selectin, E-selectin, and ICAM-1, were assessed subsequent to SWC exposure. At 7.5 minutes, cytoplasmic staining of P-selectin (Figure 7) was evident in the second zone. This expression was 11-fold higher compared with that of sham-exposed controls (p < .01) and 6-fold higher compared with SW-exposed samples (not significant). As expected, expression of E-selectin and ICAM-1 did not differ between cells exposed to SWC and controls at this time point (not shown). At 4 hours, the remaining cells in the second zone were continuously positive for P-selectin (data not shown). Cells in the third zone displayed 8-fold expression of E-selectin compared with controls (p < .05) at this time point (Figure 8, B, D, and F). Expression of ICAM-1, in the same area, examined 6 hours postexposure, was 1.5-fold compared with controls (p < .01) (Figure 8, A, C, and E).
Given that NF-κB mediates the transcriptional activation of genes that encode for E-selectin and ICAM-1, the translocation of NF-κB subunit p65 was studied. Cells located in the third zone exhibited pronounced staining for p65 in the nucleus and less intense staining of the cytoplasm 30 minutes after exposure to SWC (Figure 9). This gave a nucleus to cytoplasm ratio of 1.49 in SWC samples compared with 1.27 in controls (p < .05), presumably indicating translocation of NF-κB and subsequent de novo synthesis of E-selectin and ICAM-1. The involvement of NF-κB in signal transduction was further emphasized by the fact that the presence of PDTC, a NF-κB antagonist, abolished the SWC-induced E-selectin expression (see Figure 8F).
DISCUSSION
Human organs exposed to SWs commonly exhibit injury to the vascular endothelium. To further investigate the mechanisms involved in this injury, we developed the flyer-plate model: an experimental model for SW studies in vitro.8,14In the current setting, the flyer-plate model generates SWs, which are almost identical to SWs used during ESWL26and are similar to SWs registered under body armor at high-energy missile impact.27In contrast to existing in vivo models, the flyer-plate model enables us to differentiate between the effects caused by the shock waves per se and the effects seen when cavitation bubbles are induced. In our previous studies, we could thus, for the first time, definitively connect the cell injury associated with SWs to the presence of cavitation.14The endothelial lesion caused by SWC was discrete and immediate. It did not progress over time and was not an effect of secondary reactive oxygen species induced by the imploding cavitation bubbles.14
The aim of this study was to elucidate mechanisms involved in the inflammatory response seen after SW exposure and to further characterize the endothelial lesion caused by SWC. The primary findings are that endothelial cells peripheral to the endothelial lesion exhibit formation of actin stress fibers and up-regulation of adhesion molecules to leukocytes. This study also indicates that cytoskeletal redistribution, necessary for cell spreading and migration during endothelial repair, is hampered after SWC exposure. The presence of cavitation was compulsory for all of these effects, supporting the notion that cavitation is a prerequisite for the majority of SW effects.7,28
Studies elucidating the inflammatory response after SW exposure are rare. In this study, we chose to evaluate the SW effect on the expression of P-selectin, E- selectin, and ICAM-1, leukocyte adhesion molecules pivotal for inflammatory processes. We demonstrate that SWC exposure causes cytoplasmic expression of P-selectin. This expression was, however, solely observed in the second zone (ie, in cells permeable to trypan blue), and the positive staining for P-selectin reflected permeabilization and exposure of P-selectin in the Weibel-Palade bodies rather than a true up-regulation of the antigen. Induction of E-selectin and ICAM-1, was, on the contrary, demonstrated in morphologically undamaged cells (third zone). This up-regulation, most probably, represented cell activation caused by SWC because it was preceded by nuclear translocation of NF-κB subunit p65 and abolished by the presence of PDTC. Consequently, our results indicate that the extravasation of leukocytes (eg, observed in kidneys after ESWL treatment)29can be partly explained by functional effects of SWC on the vascular endothelium. Although the precise mechanism initiating the signal cascade was not discerned, NF-κB activation was definitively connected to the presence of cavitation. When cavitation bubbles implode, they cause liquid jets, which, directly or via liquid turbulence, may affect the cell membrane.4,30NF-κB is known to be activated by changes in shear stress,31and it is plausible that SWC activate NF-κB through this mechanism.
When studying the effects of SWC on the endothelial cytoskeleton, we saw that cells located in the third zone also exhibited disassembly of DPBs and formation of stress fibers. Stress fiber induction by SWs was demonstrated by Seidl and colleagues when they studied the effects of multiple SWs on the endothelial layers in umbilical cords.32In their study, they speculated that the stress fibers were a consequence of collapsing cavitation bubbles, speculations that are strongly supported by our results. Because stress fibers are known to be associated with enhanced vascular permeability to fluids and leukocytes,11,33this SWC effect on the actin cytoskeleton might contribute to the tissue edema seen after SW exposure3but also facilitate the transmigration of leukocytes into the tissues. It may also be part of the explanation why cytotoxic drugs have enhanced effects when combined with SW treatment.34
Endothelial detachment is probably the most common feature of SW exposure.1,3,35Injury to the endothelial lining leads to increased vascular permeability and tissue edema. Concomitantly, the exposure of subendothelial structures (eg, collagen) makes platelets adhere and activate the coagulation cascade, causing thrombus formation and occlusion of the vessel. Thus, rapid and efficient reestablishment of the endothelial monolayer is of the utmost importance to avoid organ perturbation. Reendothelialization of vascular endothelial lesions after SW exposure has not been previously studied. Here we show that the regeneration rate and morphologic changes pivotal for endothelial repair during the initial hours after injury were impaired in SWC lesions compared with MLs used as controls in this study, as well as when compared with the results presented by others.23,24The pivotal role of the endothelial cytoskeleton for reendothelialization is well documented.36Therefore, we studied the effects of SWC on the cytoskeletal components during the first 4 hours of regeneration and could demonstrate that actin and tubulin redistribution, as well as centrosome relocation after injury, was absent in SWC lesions or delayed compared with MLs. Endothelial repair of an injury of this size, however, requires not only cell spreading and migration but also proliferation. Inhibition of proliferation caused by SWC could, hence, contribute to the delay in endothelial regeneration. Since our study focused on the SW effects on the cytoskeleton, this issue was not covered; hence, it cannot be concluded to what extent the cytoskeletal changes contributed to the delayed regeneration rate in SWC samples. Cell spreading and cell migration are, however, the dominating processes during early endothelial regeneration, and proliferation is known to be prominent by 24 hours after injury.37It is, therefore, believed that the impaired cytoskeleton had importance to the decreased regeneration efficiency seen in SWC lesions.
To summarize, we have shown that endothelial lesions caused by SWC in vitro are accompanied by a proinflammatory reaction in cells located outside the actual lesion area. Moreover, we have demonstrated that SWC lesions are associated with impaired endothelial repair that can be conferred to changes in cytoskeletal functions. These findings may have clinical implications on various medical treatments, such as ESWL, as well as blast- and high-energy missile trauma.
ACKNOWLEDGMENTS
We would like to acknowledge Mrs. E. Malm for excellent technical assistance, M.Sc. N. Roman for valuable discussions concerning SWs, and Associate Professor M. Risling, all three of the Swedish Defense Research Agency.