Am J Physiol Cell Physiol AJP: Gastrointestinal and Liver Physiology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Am J Physiol Cell Physiol 292: C1645-C1659, 2007. First published December 27, 2006; doi:10.1152/ajpcell.00533.2006
0363-6143/07 $8.00
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Supplemental Figures
Right arrow All Versions of this Article:
292/5/C1645    most recent
00533.2006v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via ISI Web of Science (1)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Malek, A. M.
Right arrow Articles by Alper, S. L.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Malek, A. M.
Right arrow Articles by Alper, S. L.

VASCULAR BIOLOGY

Hypertonicity triggers RhoA-dependent assembly of myosin-containing striated polygonal actin networks in endothelial cells

Adel M. Malek ,1,* Chang Xu,2,4,* Edward S. Kim,2,4 and Seth L. Alper2,3,4

1Cerebrovascular and Endovascular Division, Department of Neurosurgery, Tufts-New England Medical Center, Tufts University School of Medicine, 2Molecular and Vascular Medicine Unit and 3Renal Division, Beth Israel Deaconess Medical Center, and 4Department of Medicine, Harvard Medical School, Boston, Massachusetts

Submitted 17 October 2006 ; accepted in final form 20 December 2006


   ABSTRACT
TOP
 ABSTRACT
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 GRANTS
 REFERENCES
 
Endothelial cells respond to mechanical stresses of the circulation with cytoskeletal rearrangements such as F-actin stress fiber alignment along the axis of fluid flow. Endothelial cells are exposed to hypertonic stress in the renal medulla or during mannitol treatment of cerebral edema. We report here that arterial endothelial cells exposed to hypertonic stress rearranged F-actin into novel actin-myosin II fibers with regular 0.5-µm striations, in which {alpha}-actinin colocalizes with actin. These striated fibers assembled over hours into three-dimensional, irregular, polygonal actin networks most prominent at the cell base, and occasionally surrounding the nucleus in a geodesic-like structure. Hypertonicity-induced assembly of striated polygonal actin networks was inhibited by cytochalasin D, blebbistatin, cell ATP depletion, and intracellular Ca2+ chelation but did not require intact microtubules, regulatory volume increase, or de novo RNA or protein synthesis. Striated polygonal actin network assembly was insensitive to inhibitors of MAP kinases, tyrosine kinases, or phosphatidylinositol 3-kinase, but was prevented by C3 exotoxin, by the RhoA kinase inhibitor Y-27632, and by overexpressed dominant-negative RhoA. In contrast, overexpression of dominant-negative Rac or of dominant-negative cdc42 cDNAs did not prevent striated polygonal actin network assembly. The actin networks described here are novel in structure, as striated actin-myosin structures in nonmuscle cells, as a cellular response to hypertonicity, and as a cytoskeletal regulatory function of RhoA. Endothelial cells may use RhoA-dependent striated polygonal actin networks, possibly in concert with cytoskeletal load-bearing elements, as a contractile, tension-generating component of their defense against isotropic compressive forces.

mannitol; Rho kinase; blebbistatin; bovine aortic endothelial cells

EUKARYOTIC CELLS FROM YEAST TO HUMAN have developed survival mechanisms to allow adaptation to anisosmotic stress (74). Some of these ancient regulatory mechanisms have been preserved in cells and tissues of metazoans despite their protection from extreme osmotic fluctuations by organismal homeostasis. Hyperosmolar stress has been shown to trigger signaling cascades via MKK3/6 (76) and/or other kinases, leading to activation of the stress kinases p38, JNK, and/or ERK1/2 in endothelial cells (18, 60). The upstream events leading to activation of MKK3/6 remain unknown, but hypertonic activation of the small Rho family GTPases (15, 47) has been implicated in this process (26). Rho-GTPases also regulate the actin cytoskeleton, among their many other functions (34). In endothelial cells, RhoA GTPase activity and membrane translocation are required for the enhanced formation and parallel realignment of stress fibers induced by shear stress (48, 89) or by thrombin (57). RhoA activity has also been linked in subconfluent endothelial cells to the assembly of actin into podosomes (59). The role of RhoA in assembly of central stress fibers in endothelial cells subjected to elevated hydrostatic pressure is not known (69).

Mammalian endothelial cells are regularly exposed to hypertonic stress only in the renal medulla. However, endothelial cells throughout the body can be exposed to hypertonic stress during intravenous mannitol infusion for up to 48–72 h for acute and subacute treatment of brain injury or cerebral edema (84). Infusion of hypertonic solutes has a barrier repair potential (65) and increases endothelial membrane E-cadherin expression and the density of filamentous cortical actin (62), but extreme hypertonicity can disrupt endothelial cell barrier function (37). We (51) and others (2) have shown that hypertonic mannitol promotes apoptosis in serum-deprived endothelial cells, but apoptosis is attenuated by growth in the presence of compatible osmolytes or the delayed biosynthesis of uptake systems for these osmolytes (3).

Hyperosmolarity-induced actin remodeling has been observed in numerous cell types (16), including Dictyostelium (1), neutrophils (47, 64), Chinese hamster ovary cells (14), and LLC-PK1 cells (15). Transient (15 min) hypotonic exposure also induced actin reorganization in bovine aortic endothelial cells (BAEC) (42) and in human umbilical vein endothelial cells (HUVEC) (30), but the response to hypertonicity of the endothelial F-actin cytoskeleton has not been reported.

We describe here a novel rearrangement of the endothelial actin cytoskeleton, unassociated with apoptosis and distinct in morphology from the cortical actin web, the junction-associated actin filament system, and stress fibers (17). Acute exposure to hypertonic medium led to the de novo assembly of striated actin-myosin II fibers, which assembled over several hours into a dense but irregular polygonal network most prominent at the base of the cell. This striated polygonal actin network (SPAN) extended upward around the nucleus to form in some cells a three-dimensional geodesic structure. Hypertonic treatment of venous endothelial cells, fibroblasts, or epithelial cells did not lead to assembly of SPANs. Assembly of SPANs was independent of the microtubule cytoskeleton and insensitive to a range of kinase inhibitors. Although insensitive to antagonists of Rac and Cdc-42, SPAN assembly was blocked by several mechanistically distinct modulators of the RhoA pathway.

Our observations of hypertonicity-induced striated actin/myosin fibers of regular periodicity, and of their assembly into RhoA-dependent SPAN structures, are novel in several respects: as a newly described cellular response to hypertonicity, as striated actin-myosin superstructures of nonmuscle cells, as novel polygonal geodesic-like actin networks, and as a new cytoskeletal regulatory function of RhoA.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 GRANTS
 REFERENCES
 
Cell culture. Primary cultures of BAEC were prepared as described previously (51) or purchased from Clonetics or Cambrex (Walkersville, MD). Bovine brain microvascular endothelial cells (BMEC) were purchased from Cell Systems (Kirkland, WA). All endothelial cell preparations exhibited uptake of acetylated LDL. Confluent BAEC also responded to laminar shear stress with upregulation of endothelial nitric oxide synthase, downregulation of endothelin-1, and elongation and alignment of the long cellular axis parallel to the shear vector (52). Cells were grown in DMEM supplemented with 10% fetal calf serum (FCS) in a humidified incubator containing 5% CO2 at 37°C. For experimental purposes, BAEC were seeded on 25-mm coverslips and used after 24 h at ~50% confluence. Small-molecule inhibitors were from Sigma or Calbiochem.

Cell stimulation by hypertonicity. Fifty percent confluent BAEC were washed twice with DMEM and then incubated at 37°C in 5% CO2 for the indicated times in DMEM plus 10% FCS or calf serum supplemented with 300 mM mannitol or, as indicated, with equiosmolar NaCl (150 mM) or urea (300 mM). In some experiments treatment times and concentrations were varied as indicated. The stimulation period was terminated by rapid rinsing in ice-cold phosphate-buffered saline (PBS) followed by cell fixation for 30 min with 3% paraformaldehyde in PBS (140 mM sodium phosphate, 10 mM HEPES, pH 7.4), quenching with PBS containing 50 mM lysine HCl, and storage of coverslips at 4°C in PBS containing 0.02% sodium azide until later use. Fixation with methanol at –20°C gave similar results.

Cell volume measurement. Cells were loaded for 30 min at room temperature with 1.8 µM calcein-AM (for excitation at 495 nm) or with 4 µM BCECF-AM (for excitation at 440 nm) in Hanks' balanced salt solution (HBSS) containing 10 mM HEPES, pH 7.4, rinsed, and mounted in a 1-ml perfusion chamber on the stage of an Olympus IMT-2 microscope equipped with an Image-1 imaging system (Universal Imaging, West Chester, PA). Cells were perfused with HBSS and then with HBSS supplemented with 150 mM NaCl in either the absence or the presence of 100 µM bumetanide. Fluorescence emission images were acquired at 530 nm by IMAGE-1 FL software (Universal Imaging). Normalized volume at time t (Vt/Vo) was calculated as [(Fo/Ft) – Fb]/(1 – Fb), where Fo and Ft are fluorescence intensities at time = 0 and time t and Fb is background fluorescence intensity. Cellular fluorescence emission intensity values were corrected for combined efflux and bleaching of cell-associated dye of 0.1%/min. Values are reported as means ± SD.

cDNA constructs and BAEC transfection. cDNAs encoding 3x hemagglutinin (HA)-tagged human wild-type RhoA, 3x HA-tagged RhoA G14V, 3x HA-tagged RhoA T19N, 3x HA-tagged human Cdc42 T17N mutant, and 3x HA-tagged human Rac1 T17N mutant were from the University of Missouri-Rolla cDNA Resource Center (www.cdna.org). cDNAs encoding {alpha}-actinin-1-green fluorescent protein (GFP) and {alpha}-actinin-4-GFP were from Dr. Carol Otey (University of North Carolina at Chapel Hill).

BAEC were trypsinized for 2 min in an incubator at 37°C, and an aliquot of trypsinized cell suspension was counted by hemocytometer. The cells were pelleted at 100 g for 10 min. Then 5–7 x 105 cells were resuspended at room temperature in 100 µl of Basic Endothelial Nucleofector solution (Amaxa Biosystems, Köln, Germany), to which was added 5–7 µg of plasmid DNA. This mixed suspension was transferred to a 2-mm electroporation cuvette and inserted in the Nucleofector device (Amaxa). Electroporation with program T-05 was immediately followed by addition of 500 µl of prewarmed DMEM plus 10% FCS. The cell suspension was plated onto a 25-mm coverslip in a six-well plate and returned to the 37°C incubator for 36 h before use in experiments.

Fluorescence microscopy. Cells grown on glass coverslips were fixed with 3% paraformaldehyde in PBS. Fixed cells were blocked and permeabilized for 15 min with PBS containing 1% bovine serum albumin and 0.05% saponin. Cells were stained with FITC-phalloidin or Oregon Green 488 phalloidin (1:40, 30 min; Molecular Probes, Eugene, OR) and one of the following primary antibodies: mouse monoclonal anti-{alpha}-actinin (1:500, 2 h; Sigma), rabbit polyclonal anti-{alpha}-actinin (from Dr. John Hartwig, Harvard Medical School), and rabbit polyclonal anti-myosin IIB and anti-myosin IIA (1:500, 2 h; Covance, Princeton, NJ). Secondary antibodies were from Jackson Immunochemicals (Huntington, PA). Cells transfected with HA-tagged small GTPase constructs were stained with monoclonal anti-HA antibody (Molecular Probes) for 90 min, followed by Cy3 goat anti-mouse Ig for 1 h. Coimmunostaining with two rabbit polyclonal antibodies (see GoGoGoFig. 4) was by the microwave denaturation method (78), with appropriate single-antibody pre- and postmicrowave controls.


Figure 1
View larger version (80K):
[in this window]
[in a new window]

  		Fig. 1. Hypertonic stress with impermeant osmoles induces assembly of a polygonal network of punctate F-actin filaments. Bovine aortic endothelial cells (BAEC) grown to 50% confluence were exposed for 3 h to isotonic culture medium (a, control) or to isotonic medium to which was added 300 mM mannitol (b), 300 mM urea (c), or 150 mM NaCl (d). Alternatively, cells were also exposed to hypotonic culture medium for 3 h (e). Fixed cells were stained with FITC-phalloidin and imaged by confocal laser fluorescence microscopy. Scale bar, 5 µm.

 

Figure 2
View larger version (30K):
[in this window]
[in a new window]

  		Fig. 2. Colocalization of {alpha}-actinin with F-actin in punctate filaments of the polygonal network induced by hypertonic stress. Subconfluent BAEC treated for 3 h in mannitol-supplemented hypertonic medium were fixed and costained with rhodamine-phalloidin (a) and with anti-actinin antibody detected with FITC-conjugated anti-Ig (b) and imaged by confocal laser fluorescence microscopy and Scanalytics deconvolution software. In the merged image (c), yellow indicates colocalization. Scale bar, 5 µm.

 

Figure 3
View larger version (73K):
[in this window]
[in a new window]

  		Fig. 3. The hypertonicity-induced polygonal network is built from striated fibers of actin and myosin II. Subconfluent BAEC treated 3 h with culture medium in the absence (Iso; a–c) and presence (Hyper; d–f) of added mannitol were fixed and stained with Oregon Green 488-phalloidin (green) and with antibody to myosin IIB (red). White boxes in c and f are enlarged in inset (g and h). Scale bar, 5 µm.

 

Figure 4
View larger version (68K):
[in this window]
[in a new window]

  		Fig. 4. Myosin IIB and myosin IIA partially colocalize in stress fibers of isotonically treated BAEC and in striated polygonal actin networks (SPANs) of hypertonically treated BAEC. Subconfluent BAEC on coverslips were incubated 3 h in culture medium in the absence (Iso; a–c) or presence (Hyper; d–f) of 300 mM added mannitol and then fixed. Cells were immunostained with antibody to myosin IIB and with Oregon Green-conjugated anti-Ig and then postfixed and subjected to microwave treatment that was confirmed in parallel control experiments to denature the bound antibody (not shown). Fixed, microwaved coverslips were then immunostained with antibody to myosin IIA and Cy3-conjugated anti-Ig, postfixed, and imaged by laser confocal fluorescence microscopy. White boxes in c and f are enlarged in the insets. Scale bar, 5 µm.

 
Cells were imaged by conventional epifluorescence microscopy with an Olympus BH2 photomicroscope and then by confocal immunofluorescence microscopy using the NORAN, the Zeiss LSM510 META, or the Bio-Rad MRC-1024 laser-scanning confocal microscope. FITC, Oregon Green 488, and GFP were excited at 488 nm with a 505- to 530-nm band pass emission filter. Cy3 was excited at 543 nm with a 570- to 655-nm band pass emission filter. Figure 2 and Supplemental Fig. 2 are from deconvolution of MRC1024 x-y stacks with software from Scanalytics (Billerica, MA).1 Images were recorded from at least three separate experiments for each experimental condition, with assessment of between 26 and 62 individual cells at high magnification. Morphometric measurements are presented as means ± SE.

Terminal deoxynucleotidyltransferase-mediated dUTP nick end labeling analysis. BAEC fixed in 3% paraformaldehyde were permeabilized with 0.1% Triton X-100–0.1% sodium citrate for 5 min on ice. After permeabilization, the coverslips were washed with PBS and then incubated in terminal deoxynucleotidyltransferase-mediated dUTP nick end labeling (TUNEL) reaction mixture prepared from the TMR red in situ cell death detection kit (Roche, Germany) for 60 min at 37°C. Positive controls were performed by incubating fixed and permeabilized cells with DNase I (grade I, 200 U/ml) in 50 mM Tris·HCl, pH 7.5, plus 1 mg/ml BSA for 20 min at 37°C before labeling procedures.


   RESULTS
TOP
 ABSTRACT
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 GRANTS
 REFERENCES
 
Hypertonic stress induces formation of polygonal actin networks in BAEC. In the course of investigating anisosmotic regulation of gene expression and cell volume in BAEC, we noted a novel rearrangement of the F-actin cytoskeleton in cells exposed for 3–6 h to hypertonic mannitol. As shown in Fig. 1, addition to standard tissue culture medium of 300 mM mannitol led to formation of irregular polygonal networks of punctuated, linear arrays of F-actin as detected by phalloidin staining. This unusual response of the actin cytoskeleton was observed in ~60% of subconfluent BAEC, but infrequently in confluent cells (not shown). The actin polygons varied in their number of sides from three to approximately seven, and the punctuated, linear F-actin arrays comprising the sides of the polygons differed from the thinner, continuous actin fibers and cables of endothelium in resting and fluid shear stress-exposed states (52, 87). These novel structures also differed from the highly regular, usually trigonal, geodesic actin structures previously detected in endothelial cells (32). The punctuated F-actin filaments in hypertonically stressed cells contrasted with the continuous F-actin distribution in stress fibers of control cells (Fig. 1). The central cellular location of the hypertonicity-induced actin structure also differed from the peripheral preponderance of the increased number of F-actin stress fibers in BAEC exposed to hypotonic stress (Fig. 1).

Primary cultures of BMEC responded similarly to hypertonic mannitol with assembly of polygonal networks of punctuated linear F-actin arrays. Cell lines in which hypertonic mannitol exposure did not produce this polygonal rearrangement of punctate actin fibers included human HeLa cells, MDCK II cells, and mouse 3T3-L1 fibroblasts (not shown). Thus, among the cells tested, hypertonicity-induced assembly of polygonal networks of punctuated actin filaments was restricted to primary cultures of arterial or microvascular endothelial cells.

Hypertonicity-induced polygonal actin rearrangement does not require regulatory volume increase and is unaccompanied by apoptosis. The polygonal rearrangement of F-actin was also elicited by hypertonicity when, instead of 300 mM mannitol, the incremental 300 mosmol was contributed by the poorly permeant salt NaCl at 150 mM, but not with the permeant solute urea at 300 mM (Fig. 1). This suggested an important role of osmotic cell shrinkage for the actin rearrangement. We therefore assessed cell volume regulation in BAEC loaded with calcein-AM or with BCECF-AM by monitoring fluorescence intensity. BAEC exposed to hyperonic saline (300 mM NaCl) shrank within 1–2 min to 79.2 ± 2.3% (SD) of original volume and recovered within 15 min to 92.2 ± 2.2% of original volume (n = 32 cells imaged in 4 experiments, Supplemental Fig. 1; results were identical when volume was monitored by BCECF fluorescence excited at its pH-insensitive wavelength, not shown). The regulatory volume increase (RVI) was blocked completely by 100 µM bumetanide (Supplemental Fig. 1), or by selective omission of either Na+ (N-methyl-D-glucamine substitution) or Cl– (gluconate substitution, not shown), confirming the importance of Na+-K+-2Cl– cotransport in the RVI of BAEC (56). However, the presence of 100 µM bumetanide did not prevent assembly of the polygonal actin network (Supplemental Fig. 1), suggesting that RVI following cell shrinkage was not required for the polygonal rearrangement of actin.

In low-serum or serum-free conditions, endothelial cells exposed to prolonged hyperosmotic stress can undergo apoptosis (38, 51, 77). Therefore, we sought evidence of apoptosis in BAEC exposed to hypertonic mannitol medium for 3 h in the presence of 10% serum. Whereas DNase I treatment of BAEC elicited TUNEL positivity in nearly all cells, BAEC treated for 3 h in either isotonic or hypertonic medium in the presence of serum exhibited no TUNEL staining (data not shown). All hypertonicity data presented here were in the presence of serum, but hypertonicity-induced assembly of polygonal networks of punctuated actin filaments was also observed in the presence of 2% NuSerum or in the absence of serum (not shown).

{alpha}-Actinin colocalizes with F-actin in punctate fibers of the hypertonicity-induced polygonal actin network. We used immunofluorescence microscopy to examine the possible presence of actin-associated proteins along the punctate, linear F-actin arrays of the hypertonicity-induced polygonal actin network. Figure 2 shows a representative mannitol-treated cell costained with phalloidin and with antibody to {alpha}-actinin. F-actin and {alpha}-actinin colocalize in the punctate structures arrayed along the linear struts of the irregular polygonal network. This distribution of {alpha}-actinin resembles the periodic distribution along F-actin stress fibers previously described in fibroblasts in isotonic conditions (10, 19, 25, 44). Polygonal co-rearrangement of F-actin and {alpha}-actinin in response to hypertonic mannitol showed parallel dependence on osmolar concentration and time (not shown). Cells treated for 3 h with increasing mannitol concentrations showed increasingly prominent coalescence of F-actin and {alpha}-actinin into punctate arrays, and increasingly prominent polygonal rearrangement of those arrays, which were maximal at 300 mM mannitol (not shown). In cells exposed to 300 mM mannitol for varying periods of time, rearrangement of F-actin and of {alpha}-actinin was detectable after 15 min and progressively more prominent over several hours, peaking between 3 and 6 h (not shown). All subsequent hypertonic stress experiments were performed with 3-h exposure of BAEC to medium containing 300 mM added mannitol. In these conditions, ~60% of examined cells assemble the characteristic polygonal network of punctuated actin arrays. In these cells, restoration of isotonic medium led to disassembly of the polygonal structure over several hours, and its reappearance on second exposure to hypertonic medium. Continuation of hypertonic stress for 18–21 h beyond assembly of the irregular polygonal network of punctuated actin arrays led to progressive loss of punctuation in most actin fibers, accompanied by disappearance of the polygonal filament network from the cell. Abrupt exposure to 600 mM mannitol of these cells previously acclimated to 300 mM mannitol re-induced the assembly of the polygonal actin network (data not shown).

{alpha}-Actinin-1 is the most abundant nonmuscle form of {alpha}-actinin (54) and plays an important role in regulating cell shape and cell motility (27, 79). {alpha}-Actinin-4 (31) is a less abundant form of nonmuscle {alpha}-actinin. We expressed GFP::{alpha}-actinin-1 and GFP::{alpha}-actinin-4 fusion proteins in BAEC cells to assess a possible preferential incorporation into the hyperosmotic stress-induced polygonal actin network. {alpha}-Actinin-1-GFP was distributed at periodic intervals along F-actin stress fibers under isosmolar conditions (not shown), in a pattern similar to that of endogenous {alpha}-actinin (Fig. 2). The {alpha}-actinin-1-GFP puncti were 0.28 ± 0.01 (SE) µm in length along the filament axis (n = 26 in 3 cells) and were separated by a center-to-center distance of 0.80 ± 0.03 µm (n = 20 in 3 cells). After 3 h of hypertonic stress, {alpha}-actinin-1-GFP puncti increased in intensity and assembled into polygonal networks (not shown). Although the length of the {alpha}-actinin-1-GFP puncti in hypertonically stressed cells (0.29 ± 0.01 µm, n = 33 in 3 cells) did not differ from the isotonic value (P > 0.05), the center-to-center distance between {alpha}-actinin-1-GFP puncti in the polygonal networks decreased to 0.50 ± 0.01 µm (P < 0.0001, n = 43 in 4 cells). Similar apparent hypertonicity-induced contractions of punctuated actin filaments were evident, with interpuncta distances of 0.5 µm in phalloidin-stained cells exposed to hypertonicity (Fig. 2). {alpha}-Actinin-4-GFP expressed in BAEC exhibited similar localization under isotonic conditions and redistribution into punctuated polygonal networks under hypertonic conditions (not shown).

Hypertonicity-induced polygonal actin networks are composed of striated filaments of F-actin and myosin II. In addition to actin filaments and {alpha}-actinin, myosin II has been found in actin stress fibers of various types of nonmuscle cells (25, 86), including BAEC (39–41). Myosin II in stress fibers exhibits cell type-specific periodicities of 0.5–2 µM (10, 61, 66). Figure 3, a–c, shows a BAEC under isotonic conditions in which the continuous distribution of F-actin along stress fibers contrasted with a periodic distribution of myosin IIB. In mannitol-untreated cells, myosin IIB bands were 0.28 ± 0.01 µm in length along the stress fiber axis (n = 56 in 4 cells), with a center-to-center distance of 0.66 ± 0.01 µm (n = 61 in 4 cells). After BAEC exposure to 300 mM mannitol for 3 h, the punctate staining pattern and polygonal distribution of F-actin (Fig. 4d) and of myosin IIB (Fig. 4e) were particularly striking. The mean length of myosin IIB bands increased in mannitol-treated BAEC to 0.31 ± 0.01 µm (P < 0.001, n = 72 in 5 cells), and the center-to-center distance decreased to 0.50 ± 0.01 µm (P < 0.0001, n = 109 in 6 cells). Figure 3f shows that the hypertonicity-induced polygonal actin network consists of striated struts of alternating bands of actin and of myosin IIB. The length of phalloidin-stained bands along the filament axis of polygonal actin structures in mannitol-treated cells was 0.35 ± 0.01 µm (n = 108 in 6 cells), with center-to-center distance of 0.49 ± 0.01 µm (n = 110 in 6 cells). The branch angles of the striated actin-myosin filaments in hypertonically stressed BAEC varied considerably. However, in an arbitrary but representative selection of four- to seven-sided polygons in cells costained with phalloidin and with anti-myosin IIB, the mean branch angle was 109 ± 2° (n = 102 angles in 20 polygons from 8 cells).

BAEC express both myosin IIA and myosin IIB (39, 40). In isotonic conditions both isoforms were arrayed along stress fibers, where they partially overlap (Fig. 4, a–c). In mannitol-treated cells, both myosin isoforms redistributed into more prominent periodic patterns within polygonal structures (Fig. 4, d–f). The polarized distribution of myosin isoforms was somewhat concentric. Myosin IIB was present more at the cell center and at lower abundance at the cell periphery. Myosin IIA was less abundant at the cell center and more abundant in the area surrounding the nucleus, with staining intensity falling off more peripherally. The two myosin II isoforms have also been shown to localize differentially in leading and trailing regions of migrating BAEC (39, 40).

We have named the actin-myosin structures of hypertonicity-stressed BAEC "striated polygonal actin networks," or SPANs. A particularly prominent SPAN in a BAEC double-stained with phalloidin and anti-myosin IIB is presented in Fig. 5. The x-y confocal series illustrates how the geodesic-like SPAN can encompass the cell nucleus.


Figure 5
View larger version (76K):
[in this window]
[in a new window]

  		Fig. 5. Confocal series (A–I) of x-y plane images of a cell costained with Oregon Green 488-phalloidin (green) and antibody to myosin IIB (red); x-y plane of each panel is separated by 1.0 µm. Note that the SPAN climbs up around the nucleus, and the peripheral structure converges at the top of the cell. The BAEC shown is taller than most examined in this study. Scale bar, 10 µm.

 
Hypertonicity-induced assembly of SPANs is blocked by agents that disrupt actin-myosin cytoskeleton. Figure 6, c and d, reveals that inclusion of 2 µg/ml cytochalasin D to depolymerize F-actin during the 3-h incubation of BAEC under isotonic or hypertonic conditions completely disrupted both actin-based stress fibers and SPANs visualized with FITC-phalloidin. Figure 6 shows that inclusion of 10 µM blebbistatin to inhibit myosin ATPase during the 3-h BAEC incubation failed to disrupt F-actin stress fiber structures in isotonic conditions (Fig. 6e), whereas assembly of SPANs in hypertonically treated BAEC was abrogated (Fig. 6f). Inhibition of myosin ATPase activity by the less specific 2,3-butanedione monoxime (BDM, 10 mM) during the 3-h exposure completely disrupted both F-actin stress fibers in isotonically treated cells and SPANs in hypertonically treated cells (data not shown). A similar effect of BDM has been noted previously under isotonic conditions (11, 24).


Figure 6
View larger version (67K):
[in this window]
[in a new window]

  		Fig. 6. Disruption of the actin-myosin cytoskeleton prevents SPAN assembly in hypertonically treated BAEC. Subconfluent BAEC were treated for 3 h under isotonic or hypertonic conditions in the absence (control; a, b) or presence of 2 µg/ml cytochalasin D (Cyto D; c, d) or 10 µM blebbistatin (e, f). Cells were then fixed and stained with Oregon Green 488-phalloidin. Scale bar, 5 µm.

 
Thus pharmacological experiments confirm the requirement of integrity of both F-actin and myosin for SPAN assembly in BAEC and demonstrate greater blebbistatin sensitivity of hypertonic cytoskeletal rearrangement than of isotonic cytoskeletal maintenance. Consistent with these roles of actin and nonmuscle myosin, hypertonicity-induced SPAN assembly in BAEC was abolished by metabolic depletion of ATP (1-h preincubation in glucose-free medium containing 12 mM deoxyglucose and 25 nM antimycin A) and was attenuated by inclusion of either of the intracellular Ca2+ chelators Quin2-AM (10 µM) and BAPTA-AM (30 µM) (not shown).

Hypertonicity-induced SPAN assembly does not require integrity of protein synthesis, RNA transcription, or microtubular cytoskeleton. As evident from the representative phalloidin-stained BAEC exposed for 3 h either to 10 µM cycloheximide (Fig. 7, a and b) or to 5 µg/ml actinomycin D (Fig. 7, c and d) in the absence (Fig. 7, a and c) or presence (Fig. 7, b and d) of 300 mM mannitol, nominal inhibition of protein synthesis or of RNA transcription did not prevent assembly of SPANs. In addition, neither cycloheximide (Fig. 7a) nor actinomycin D (Fig. 7c) produced any apparent change in BAEC stress fiber structure under isotonic conditions. Thus hypertonicity-induced SPAN assembly does not require rapid synthesis of new protein or RNA and represents rearrangements of cytoskeletal polypeptides already present in cells maintained under isotonic conditions. We next examined the role of the microtubule cytoskeleton in hyperosmotic stress-induced SPAN assembly. Immunostaining of tubulin did not reveal microtubular structures colocalized in SPANs or actin stress fibers (not shown). As evident in Fig. 7, e–h, 3-h exposure to 10 µM nocodazole to depolymerize microtubules or to 10 µM taxol to interfere with dynamic microtubule remodeling led to no apparent alteration either in hypertonicity-induced assembly of SPANs (Fig. 7, f and h) or in the structure of stress fibers under isotonic conditions (Fig. 7, e and g). Immunostaining of BAEC with anti-tubulin antibody (not shown) confirmed disruption of the microtubular cytoskeleton under these conditions. These data together show that the microtubule cytoskeleton is not required in BAEC either for hypertonicity-induced assembly of SPANs or for maintenance of actin stress fiber structure in isotonic conditions.


Figure 7
View larger version (68K):
[in this window]
[in a new window]

  		Fig. 7. Hypertonicity-induced assembly of SPANs in BAEC does not require integrity of protein or RNA synthesis or of the microtubule cytoskeleton. Subconfluent BAEC on coverslips were preincubated in the presence of cycloheximide (CHX, 5 µg/ml, 30 min; a, b), actinomycin D (ActD, 5 µg/ml; c, d), nocodazole (10 µM, 30 min; e, f), or taxol (10 µM, 30 min; g, h) before 3 h of additional incubation in isotonic (a, c, e, g) or hypertonic medium supplemented with 300 mM mannitol (b, d, f, h) in the continued presence of drugs. Cells were then fixed and stained with Oregon Green 488 phalloidin. Scale bar, 5 µm.

 
Hypertonicity-induced assembly of SPANs in BAEC is accompanied by radial rearrangement of peripheral focal adhesions. Supplemental Fig. 2A shows that F-actin filaments of variable length extended from focal adhesions visualized by antibody to paxillin in BAEC under isotonic conditions. The focal adhesions tended to align in linear patterns along the axis of actin stress fibers. After 3-h exposure of BAEC to hypertonic mannitol, the paxillin-stained focal adhesions adopted a circumferential pattern around the assembled SPAN but did not extend prominently underneath the center of the SPAN, where smaller patches of paxillin were inconsistently associated with F-actin filaments (Supplemental Fig. 2B). The F-actin filaments extending radially out from the SPAN terminated in focal adhesions near or at the cellular periphery (Supplemental Fig. 2B). Immunostaining of isotonically and hypertonically treated cells with antibodies to phosphotyrosine and to vinculin colocalized with paxillin in focal adhesions (not shown).

Inhibitors of Rho kinase block hypertonicity-induced assembly of SPANs. Rho-GTPases play a central role in controlling organization of the actin cytoskeleton (34). Activated RhoA stimulates the formation of stress fibers (63), and acute enhancement of actin stress fibers in BAEC by hypotonic stress requires activation of Rho kinase (42). We used two pharmacological inhibitors of the Rho pathway for initial tests of its possible role in hypertonicity-induced assembly of SPANs in BAEC. Twenty-four-hour preincubation of BAEC with the RhoA inactivator C3 exoenzyme of Clostridium botulinum (10 µg/ml) (5), followed by coincubation during the 3-h exposure to mannitol, abolished SPAN assembly (Fig. 8d; n = 54) compared with cells unexposed to C3 exotoxin (Fig. 8b; n = 38). Since RhoA acts in part through the RhoA kinases (ROCKs), we tested the effect on SPAN assembly of exposure to the ROCK inhibitor Y-27632 (20 µM) during the 3-h hypertonic incubation. This treatment similarly resulted in complete inhibition of SPAN assembly in hypertonically treated BAEC (Fig. 8f). In isotonic conditions, C3 exoenyzme greatly reduced the length and number of actin stress fibers (Fig. 8c), and Y-27632-treated BAEC in isotonic conditions lost all actin stress fibers (Fig. 8e; n = 41 compared with 34 untreated cells). Similar results have been noted in other cell types (73).


Figure 8
View larger version (63K):
[in this window]
[in a new window]

  		Fig. 8. Inhibitors of Rho-GTPase and Rho A kinase (ROCK) prevent hypertonicity-induced assembly of SPANs in BAEC. Subconfluent BAEC were preincubated in the absence (control; a, b) or presence of the Rho GTPase inhibitor C3 exotoxin (10 µg/ml, 24 h; c, d) or the ROCK inhibitor Y-27632 (10 µM, 30 min; e, f), followed by 3 h of additional incubation in isotonic (a, c, e) or hypertonic medium supplemented with 300 mosM mannitol (b, d, f) in the continued presence of drugs. Cells were then fixed and stained with Oregon Green 488-phalloidin. Both inhibitors disrupt stress fibers in isotonically treated cells and SPANs in hypertonically treated cells. Scale bar, 5 µm.

 
To extend the results obtained with C3 exotoxin and Y-27632 suggesting a role of RhoA, we tested the effects of overexpression in BAEC cells of HA-tagged versions of wild-type RhoA cDNA, its dominant-negative T19N mutant, and its constitutively active G14V mutant. Thirty-six hours after transfection, cells were incubated for 3 h under isotonic or hypertonic conditions. Under isotonic conditions, the parallel array of stress fibers seen in untransfected cells (Fig. 9, a, e, i) was attenuated in cells overexpressing RhoA T19N and replaced by enhancement of peripheral actin filaments, with a faint central polygonal pattern (Fig. 9, b, f, j). However, cells overexpressing wild-type RhoA (Fig. 9, c, g, k) or constitutively activated G14V RhoA (Fig. 9, d, h, l) exhibited normal or increased stress fiber density. Isotonically treated cells overexpressing G14V RhoA also expressed small circles of actin (Fig. 9l) suggestive of possible early-stage podosome assembly, also regulated by RhoA in BAEC under select conditions (81).


Figure 9
View larger version (97K):
[in this window]
[in a new window]

  		Fig. 9. Role of RhoA on F-actin structure in BAEC in isotonic conditions. Subconfluent cultures of BAEC (non-Tx; a, e) were transfected with plasmids encoding hemagglutinin (HA)-tagged dominant-negative T19N RhoA (b, f), wild-type RhoA (c, g), or constitutively active G14V RhoA (d, h). Thirty-six hours after transfection, cells incubated in isotonic medium for 3 h were fixed and costained with Oregon Green 488-phalloidin and with anti-HA antibody followed by Cy3-conjugated secondary anti-Ig. Insets in white boxes are enlarged in i–l. Note the minimal effects of RhoA overexpression on patterns of F-actin stress fibers and the interference with parallel stress fiber arrangement when RhoA activity is inhibited. Scale bar, 5 µm.

 
Figure 10 reveals that the SPANs present in hypertonically stressed, nontransfected BAEC (n = 59; Fig. 10, a, e, i) were nearly absent from cells overexpressing dominant-negative T19N RhoA (n = 62; Fig. 10, b, f, j), although on occasion attenuated polygonal actin forms could be seen in subregions of the cell. In cells overexpressing wild-type RhoA (n = 28; Fig. 10, c, g, k) or constitutively active RhoA G14V (n = 26; Fig. 10, d, h, l), the greatly decreased frequency and complexity of mannitol-induced SPANs was accompanied by increased stress fiber density compared with the nontransfected cells on the same coverslips (n = 30, n = 27, respectively). These findings together suggest a requirement for optimal levels of RhoA activity in hypertonicity-induced SPAN assembly, acting through its downstream kinase ROCK.


Figure 10
View larger version (103K):
[in this window]
[in a new window]

  		Fig. 10. Role of RhoA in hypertonicity-induced assembly of SPANs in BAEC. Subconfluent BAEC (a, e) were transfected with plasmids encoding HA-tagged dominant-negative T19N RhoA (b, f), wild-type RhoA (c, g), or constitutively active G14V RhoA (d, h). Thirty-six hours after transfection, cells were incubated for 3 h in hypertonic medium supplemented with 300 mM mannitol. Cells were then fixed and costained with Oregon Green 488-phalloidin and with anti-HA, followed by Cy3-conjugated secondary anti-Ig. Insets in white boxes are enlarged in i–l. Note that RhoA inhibition (b, f, j) or RhoA overexpression (c, g, k, d, h, l) can prevent hypertonic assembly of SPANs. Scale bar, 5 µm.

 
Activities of Rac1, Cdc42, and MAP kinases are not crucial for hypertonicity-induced assembly of SPANs. Studies of neutrophils (47) and hepatocytes (7) have demonstrated that hypertonic stress can activate Rho family GTPases Rac1 and Cdc42. Dominant-negative Rac1 was recently shown to inhibit the hyperosmolarity-induced increase in actin filament density in lung microvascular endothelial cells (62). Therefore, we assessed possible contributions of Rac1 and Cdc42 to hypertonicity-induced SPAN assembly by overexpression in BAEC of HA-tagged dominant-negative T17N mutants of Rac1 and Cdc42. As shown in Fig. 11, a–d, overexpression of Rac1T17N substantially altered neither stress fibers in isotonic conditions nor SPAN assembly in hypertonic conditions (n = 38 transfected vs. 28 nontransfected cells). Consistent with this result, the Rac1 inhibitor NSC-23766 (200 µM) had no effect on hypertonicity-induced SPAN assembly in BAEC (n = 74 vs. 75 untreated cells, not shown). Similarly, overexpression in BAEC of the dominant-negative Cdc42 T17N mutant did not significantly alter stress fiber abundance under isotonic conditions (Fig. 11, e and f) or hypertonicity-induced assembly of SPANs (Fig. 11, g and h; n = 33 transfected vs. 29 nontransfected cells).


Figure 11
View larger version (51K):
[in this window]
[in a new window]

  		Fig. 11. Rac1 and CDC42 are not required for hypertonicity-induced assembly of SPANs in BAEC. Subconfluent BAEC were transfected with plasmids encoding HA-tagged dominant-negative Rac1 T17N (a–d) or dominant-negative Cdc42 T17N (e–h). Thirty-six hours after transfection, cells were incubated 3 h in isotonic or hypertonic medium supplemented with 300 mM mannitol. Cells were then fixed and costained with phalloidin and with anti-HA, followed by Cy3-conjugated secondary anti-Ig. Scale bar, 5 µm. Insets show higher magnification.

 
Hyperosmolarity has been shown to activate the stress kinases p38, JNK, and ERK1/2 in endothelial cells (18, 60). We therefore tested the effects on hypertonicity-induced SPAN assembly in BAEC of the c-JNK inhibitor SP-600125, the p38 inhibitor SB-239063, and the ERK1/2 inhibitor U-0126. However, neither stress fiber distribution under isotonic conditions nor SPAN assembly under hypertonic stress was altered by any of these MAP kinase inhibitors (not shown).


   DISCUSSION
TOP
 ABSTRACT
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
GRANTS
 REFERENCES
 
We have shown that subconfluent BAEC subjected for several hours to hypertonic stress reorganize their F-actin stress fibers into SPANs. The striations are characterized by colocalized actin and {alpha}-actinin alternating with myosin II and exhibit a characteristic striation repeat dimension of 0.5 µm. The striated filaments assemble into irregular polygons at the base of the cell and, in some cells, ascend to encompass the nucleus with an irregular geodesic dome. Hypertonic assembly of SPANs is prevented by inhibition of actin polymerization and by inhibition of myosin ATPase activity, but not by disruption of the microtubular cytoskeleton, and does not require synthesis of new mRNA or protein. Although hypertonic stress activates RVI and MAP kinase cascades, hypertonicity-induced assembly of SPANs is not prevented by inhibitors of RVI or of MAP kinases. RhoA activity is essential to hypertonicity-induced assembly of SPANs, as judged by the inhibitory effects of C3 exotoxin, ROCK inhibitor, and transfected dominant-negative RhoA. However, RhoA overactivity leading to enhanced stress fiber formation under isotonic conditions also prevents SPAN assembly in BAEC under hypertonic conditions. The effects of RhoA inhibitors were specific insofar as pharmacological or dominant-negative inhibitors of the Rho family GTPases Rac1 and Cdc42 did not prevent hypertonicity-induced SPAN assembly.

SPANs are novel, actin-myosin-based structures previously unreported in nonmuscle cells, and they differ in striation dimensions and polygon structure from the polygonal actin networks of fibroblasts and myocytes. Assembly of SPANs also represents a novel cellular response to hypertonic stress and a new type of actin cytoskeleton regulation by RhoA. The decrease in striation lengths of myosin II and {alpha}-actinin induced by hypertonicity suggests that SPANs are under contractile tension. Together with the hypertonicity-induced radial rearrangement of paxillin-positive, focal adhesion-like structures, the data suggest that SPANs provide mechanical stability to endothelial cells subjected to hypertonic stress.

Polygonal networks of F-actin in adherent, nonmuscle cells. Actin filaments can form triangular geodesic networks in a variety of nonmuscle cell types during reattachment to and spreading on the substratum in the hours after plating from suspension. These cells include embryonic fibroblasts from rat (43, 44) and chick (33, 49), epidermal cells (58), and human endothelial cells (32). Formation of polygonal networks in rat hepatocytes required several days (55), whereas the networks were present in resting bovine lens epithelial cells (53). These polygonal networks were usually regular trigonal structures in which {alpha}-actinin was localized to the vertices. Myosin, when studied, was absent from the vertices but present in the struts. The networks were usually observed in only a fraction of cells examined in each visual field.

Polygonal actin networks have also been reported to be induced by agonists. Thus CV1 kidney cells overexpressing plastins reorganized their actin filaments into regular geodesic structures with plastins excluded from the vertices (6). SLC4.1/F7 neonatal rat Schwann cells exhibited "starlike geodesic arrangements of actin" induced in 85% of cells by 1 µM sphingosine-1-phosphate and in 50% of cells by 1 µM lysophosphatidic acid (8). Irregular polygonal actin structures containing {alpha}-actinin in their vertices ("cross-linked actin networks," or CLANs) were induced in 50% of endothelial-like human trabecular meshwork cells in primary culture by 10- to 14-day exposure to 1 µM dexamethasone (88) and were detected as well in intact trabecular meshwork tissue from steroid-treated patients (12). CLAN formation in human trabecular meshwork cells also occurred within 3 h after plating onto fibronectin and was enhanced by antibodies to beta1- and beta3-integrins (22).

In each of these cases, the pattern of F-actin along polygonal struts appeared continuous, as has generally been true also for stress fibers of F-actin in nonmuscle cell types (10, 29, 45, 46). Myosin (10, 25, 39, 40, 61), tropomyosin (44), and {alpha}-actinin (10, 45, 61, 68) have often exhibited periodic staining patterns along actin stress fibers. SPANs of BAEC appear to be the first nonmuscle cell actin structures to exhibit actin periodicity coincident with the distribution of {alpha}-actinin. SPANs are also unusual in the presence of actin, {alpha}-actinin, and myosin II in both struts and vertices of the polygonal network.

RhoA activity is important for hypertonicity-induced assembly of SPANs in BAEC. RhoA, Rac1, and Cdc42 each produce specific effects on the actin cytoskeleton of cultured fibroblasts. Whereas RhoA activation stimulates the assembly of contractile actin-myosin filaments and stress fibers, Rac activation induces lamellipodia, and Cdc42 activation leads to assembly of filopodia (21, 34). In HUVEC, RhoA activation is required for the initial contraction and depolarization phases of cell elongation in response to laminar shear stress (89), whereas decreased RhoA activity is required for later-stage shear stress-induced alignment of BAEC (80). The importance of BAEC RhoA activation in hypertonicity-induced assembly of SPANs was demonstrated by inhibition of SPAN assembly by C3 exotoxin, by ROCK inhibitor Y-27632 (Fig. 8, d and f), and by overexpression of dominant-negative T19N-RhoA (Fig. 10b). BAEC myosin II is 70–90% myosin IIA and 10–30% myosin IIB (40), and both isoforms were present in SPANs (Fig. 5). However, only myosin IIB was regulated by RhoA in a BAEC wound healing model in which myosin IIB preferentially accumulated at the cell's trailing edge (39).

RhoA and Rho kinase signaling have been implicated in hypotonic stress-induced actin reorganization in HUVEC and in BAEC (30, 42). RhoA overexpression in isotonic medium enhanced formation of dense stress fibers in BAEC (23), while overexpression of either wild-type RhoA or constitutively active G14V RhoA suppressed hypertonicity-induced formation of SPANs (Fig. 10, c and d). Thus interference with hypertonic SPAN assembly by RhoA hyperfunction may reflect preferential recruitment of F-actin into stress fibers, a mechanism that may differ from that by which RhoA inhibitors block SPAN assembly. In contrast to the requirement of Rac1 and/or Cdc42 for hypertonicity-induced polymerization of subcortical actin in neutrophils and for filopodia generation in Swiss 3T3 cells (47, 64), neither Rac1 nor Cdc2 appeared to contribute to hypertonicity-induced SPAN assembly in BAEC (Fig. 11). RhoA also regulates c-JNK (13) and p38 MAPK (35), and hypertonic stress activates p38, JNK, and ERK1/2 in endothelial and Swiss 3T3 cells (18, 60). However, inhibitors of p38 MAPK, c-JNK, and ERK1/2 individually had no effect on hyperosmotic stress-stimulated SPAN assembly in BAEC, suggesting that no single MAP kinase is required for hypertonicity-induced RhoA/ROCK-dependent assembly of SPANs. The phosphatidylinositol 3-kinase inhibitor wortmannin was similarly without effect on SPAN assembly (not shown).

Actin filament branching and striation in hypertonicity-induced SPANs. In eukaryotic cells, actin branching is believed to be mediated mainly by the actin polymerization factors Arp2/3 and the formins. Cdc42 and Rac respectively activate WASP and Wave proteins to stimulate the Arp2/3 complex to form peripheral lamellipodia and filopodia. Arp2/3-mediated actin branching is characterized by 70° branch angles. Some polygon vertices do anchor acute angles. However, the average branch angle of 103–109° in SPANs does not suggest a central role for Arp2/3 in determination of branch angle for the larger polygons. As RhoA also activates formins such as mDia1 to stimulate linear elongation of branched filaments at barbed ends (34), formins may contribute to hypertonicity-induced SPAN assembly.

The uniform 0.5-µm striation periodicity of SPANs is identical to the myosin II periodicity noted along native stress fibers in endothelial cells of intact mouse aorta (87) and similar to that noted in intact stress fibers isolated from bovine carotid arterial endothelial cells (36). The 0.5-µm striation dimension also recalls the ~0.5-µm length of short, thick filaments assembled in vitro from antiparallel tetramers of Dictyostelium myosin II (50). The BAEC transition from stress fibers in isotonic medium to SPANs in hypertonic medium was accompanied by apparent contraction of actin-myosin filaments. Because myosin II bands of nonmuscle cells generally consist of laterally aligned bipolar filaments of 10–30 myosin molecules (75, 82, 83), the slightly increased myosin II band length in SPANs may reflect staggered lateral incorporation of additional myosin II subunits, possibly stimulated by RhoA/ROCK-regulated (70) myosin light chain phosphorylation.

The in vitro transformation from premyofibrils to nascent myofibrils (67) in embryonic chick cardiomyocytes bears some resemblance to the hypertonicity-induced transition of BAEC F-actin organization from continuous actin stress fibers to punctuated, striated filaments within polygonal actin networks. Each premyofibril contains a single uniform "stress fiber-like structure" of F-actin on which bands of {alpha}-actinin and myosin II are arrayed with a periodicity of 0.6–1.3 µm (4). The transformation to nascent myofibrils is accompanied by confinement of F-actin to I bands, while actinin-enriched Z bodies containing actin filament barbed ends from adjacent fibrils align to form beaded Z bands (28), and myosin II is gradually replaced with myosin I. The morphological similarity between the BAEC transition from stress fiber to SPAN and actin remodeling in embryonic cardiomyofibrillogenesis suggests possible mechanistic similarity of filament reorganization. Indeed, the RhoA pathway participates in cardiac myofibrillogenesis (5, 85).

Physiological significance of SPAN formation in the hypertonically stressed endothelial cell. Although hypertonicity promotes SPAN formation most effectively in subconfluent BAEC, SPAN assembly has been noted in occasional BAEC within confluent monolayers subjected to hypertonicity. Just as the actin stress fibers prominent in cultured BAEC reflect stress fibers present in native arterial and high-flow venous endothelia (see Ref. 17 for review), so SPANs may develop in native endothelium exposed to hypertonic stress. Such exposure may occur in the course of normal cycles of urinary concentration and dilution in renal vasa recta endothelial cells or pathologically during hyperosmolar metabolic states. Exposure to hypertonicity may also occur therapeutically during mannitol or hypertonic-hyperoncotic saline infusion after brain trauma or major ischemic stroke and as part of the treatment for shock (72). Although the brain parenchymal osmotic effects of such hypertonic infusion therapy are established, data on cerebrovascular endothelial barrier function generally indicate either no effect or deterioration. In contrast, infusion of hypertonic fluid has been shown to enhance the barrier function of rat pulmonary venular capillary endothelial cells in vivo (65), paralleling hypertonicity-induced increased transendothelial resistance of rat lung microvascular endothelial cell monolayers in vitro (62), accompanied by increased peripheral actin polymerization and E-cadherin expression. Evidence from transient treatment with hypertonic-hyperoncotic fluid treatment immediately following surgical repair of pediatric atrioventricular septal defects is also consistent with reduced pulmonary endothelial leak (71). Hypertonic-hyperoncotic fluid infusion has been associated with increased capillary flow and decreased ischemia-reperfusion injury in multiple tissues.

As hypertonicity-induced BAEC shrinkage is only partially compensated by acute RVI (Supplemental Fig. 1), SPAN assembly may be a cellular compensation for a reduction in the turgor component of cellular prestress pending intracellular accumulation of compatible osmoles. Centripetal tension possibly developed by the striated actin-myosin II struts of the perinuclear SPAN may stabilize an initially increased state of cellular prestress. The polygonal arrangement of the striated struts may at the same time provide incremental compressive load-bearing capability. In this way, SPANs might represent a tensegrity structure in which the thick, striated struts dispense with a requirement for load-bearing microtubules, perhaps accompanied by increased contribution from adhesive matrix receptors or microfilaments. Preliminary data with millimolar concentrations of the vimentin microfilament cytoskeletal inhibitor acrylamide indeed suggest the possibility that microfilaments may be required for SPAN assembly (data not shown). However, more specific vimentin antagonists will be needed to assess the role of the microfilament cytoskeleton in hypertonic SPAN assembly.

Conclusion. We have characterized a geodesic-like striated polygonal actin network (SPAN) as a novel RhoA-associated endothelial cell response to hypertonic stress. It is tempting to hypothesize that the SPAN represents a structural defense of the cell and its genetic material in the nucleus, achieved in part by increasing the maximal compressive force tolerated by the adjacent microtubular (9) and microfilament (20) cytoskeletons. Future study of the mechanical properties of BAEC harboring SPANs will define the contribution of this novel actin cytoskeletal structure to cellular prestress and stiffness as part of the cellular response to clinically relevant hypertonic stress and to other mechanical forces. More detailed study of the kinetics and regulation of SPAN assembly and disassembly should offer new paths to understanding regulation of the endothelial actin cytoskeleton.


   GRANTS
TOP
 ABSTRACT
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 GRANTS
REFERENCES
 
A. M. Malek was supported by the Boston Neurosurgical Foundation and by the Cam Neely Foundation for Cancer Care and the Boston Institute of Neurosurgery. C. Xu was supported by National Institutes of Health (NIH) Fellowship DK-69049. S. L. Alper was supported by American Heart Association Grant-in-Aid 9650781N and by NIH Grants CA-86207, DK-61051, HL-15157 (Boston Comprehensive Sickle Cell Center) and DK-34854 (Harvard Digestive Diseases Center).


   ACKNOWLEDGMENTS
 
We thank John Hartwig for antibody to {alpha}-actinin, Carol Otey for GFP-{alpha}-actinin cDNAs, Scott Pomeroy and John Fiala for assistance with confocal microscopy, Lianwei Jiang for the regulatory volume increase experiment, and Steen Hansen and John Hartwig for helpful discussion.


   FOOTNOTES
 

Address for reprint requests and other correspondence: S. L. Alper, Molecular and Vascular Medicine Unit, Beth Israel Deaconess Medical Center, 330 Brookline Ave, Boston, MA 02215 (e-mail: salper{at}bidmc.harvard.edu); A. M. Malek, Cerebrovascular and Endovascular Div., Dept. of Neurosurgery, Tufts-New England Medical Center, 750 Washington St., Proger 7, Boston, MA 02111 (e-mail: amalek{at}tufts-nemc.org)

The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

* A. M. Malek and C. Xu contributed equally to this work. Back

1 The online version of this article contains supplemental data. Back


   REFERENCES
TOP
 ABSTRACT
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 GRANTS
 REFERENCES
 
1. Aizawa H, Katadae M, Maruya M, Sameshima M, Murakami-Murofushi K, Yahara I. Hyperosmotic stress-induced reorganization of actin bundles in Dictyostelium cells over-expressing cofilin. Genes Cells 4: 311–324, 1999.[Abstract]

2. Alfieri RR, Cavazzoni A, Petronini PG, Bonelli MA, Caccamo AE, Borghetti AF, Wheeler KP. Compatible osmolytes modulate the response of porcine endothelial cells to hypertonicity and protect them from apoptosis. J Physiol 540: 499–508, 2002.[Abstract/Free Full Text]

3. Alfieri RR, Petronini PG, Bonelli MA, Desenzani S, Cavazzoni A, Borghetti AF, Wheeler KP. Roles of compatible osmolytes and heat shock protein 70 in the induction of tolerance to stresses in porcine endothelial cells. J Physiol 555: 757–767, 2004.[Abstract/Free Full Text]

4. Antin PB, Tokunaka S, Nachmias VT, Holtzer H. Role of stress fiber-like structures in assembling nascent myofibrils in myosheets recovering from exposure to ethyl methanesulfonate. J Cell Biol 102: 1464–1479, 1986.[Abstract/Free Full Text]

5. Aoki H, Izumo S, Sadoshima J. Angiotensin II activates RhoA in cardiac myocytes: a critical role of RhoA in angiotensin II-induced premyofibril formation. Circ Res 82: 666–676, 1998.[Abstract/Free Full Text]

6. Arpin M, Friederich E, Algrain M, Vernel F, Louvard D. Functional differences between L- and T-plastin isoforms. J Cell Biol 127: 1995–2008, 1994.[Abstract/Free Full Text]

7. Auer KL, Contessa J, Brenz-Verca S, Pirola L, Rusconi S, Cooper G, Abo A, Wymann MP, Davis RJ, Birrer M, Dent P. The Ras/Rac1/Cdc42/SEK/JNK/c-Jun cascade is a key pathway by which agonists stimulate DNA synthesis in primary cultures of rat hepatocytes. Mol Biol Cell 9: 561–573, 1998.[Abstract/Free Full Text]

8. Barber SC, Mellor H, Gampel A, Scolding NJ. S1P and LPA trigger Schwann cell actin changes and migration. Eur J Neurosci 19: 3142–3150, 2004.[CrossRef][ISI][Medline]

9. Brangwynne CP, MacKintosh FC, Kumar S, Geisse NA, Talbot J, Mahadevan L, Parker KK, Ingber DE, Weitz DA. Microtubules can bear enhanced compressive loads in living cells because of lateral reinforcement. J Cell Biol 173: 733–741, 2006.[Abstract/Free Full Text]

10. Byers HR, Fujiwara K. Stress fibers in cells in situ: immunofluorescence visualization with antiactin, antimyosin, and anti-alpha-actinin. J Cell Biol 93: 804–811, 1982.[Abstract/Free Full Text]

11. Chrzanowska-Wodnicka M, Burridge K. Rho-stimulated contractility drives the formation of stress fibers and focal adhesions. J Cell Biol 133: 1403–1415, 1996.[Abstract/Free Full Text]

12. Clark AF, Brotchie D, Read AT, Hellberg P, English-Wright S, Pang IH, Ethier CR, Grierson I. Dexamethasone alters F-actin architecture and promotes cross-linked actin network formation in human trabecular meshwork tissue. Cell Motil Cytoskeleton 60: 83–95, 2005.[CrossRef][ISI][Medline]

13. Coleman ML, Densham RM, Croft DR, Olson MF. Stability of p21Waf1/Cip1 CDK inhibitor protein is responsive to RhoA-mediated regulation of the actin cytoskeleton. Oncogene 25: 2708–2716, 2006.[CrossRef][ISI][Medline]

14. Di Ciano C, Nie Z, Szaszi K, Lewis A, Uruno T, Zhan X, Rotstein OD, Mak A, Kapus A. Osmotic stress-induced remodeling of the cortical cytoskeleton. Am J Physiol Cell Physiol 283: C850–C865, 2002.[Abstract/Free Full Text]

15. Di Ciano-Oliveira C, Sirokmany G., Szaszi K, Arthur WT, Masszi A, Peterson M, Rotstein OD, Kapus A. Hyperosmotic stress activates Rho: differential involvement in Rho kinase-dependent MLC phosphorylation and NKCC activation. Am J Physiol Cell Physiol 285: C555–C566, 2003.[Abstract/Free Full Text]

16. Di Ciano-Oliveira C, Thirone AC, Szaszi K, Kapus A Osmotic stress and the cytoskeleton: the R(h)ole of Rho GTPases. Acta Physiol (Oxf) 187: 257–272, 2006.[CrossRef][Medline]

17. Drenckhahn D, Ness W. The endothelial contractile cytoskeleton. In: Vascular Endothelium: Physiology, Pathology, and Therapeutic Opportunities. New Horizon Series, edited by Born GV, Schwartz CJ. Stuttgart, Germany: Schattauer, 1997, vol. 3, p. 1–25.

18. Duzgun SA, Rasque H, Kito H, Azuma N, Li W, Basson MD, Gahtan V, Dudrick SJ, Sumpio BE. Mitogen-activated protein phosphorylation in endothelial cells exposed to hyperosmolar conditions. J Cell Biochem 76: 567–571, 2000.[CrossRef][ISI][Medline]

19. Edlund M, Lotano MA, Otey CA. Dynamics of alpha-actinin in focal adhesions and stress fibers visualized with alpha-actinin-green fluorescent protein. Cell Motil Cytoskeleton 48: 190–200, 2001.[CrossRef][ISI][Medline]

20. Esue O, Carson AA, Tseng Y, Wirtz D. A direct interaction between actin and vimentin filaments mediated by the tail domain of vimentin. J Biol Chem 281: 30393–30399, 2006.[Abstract/Free Full Text]

21. Etienne-Manneville S, Hall A. Rho GTPases in cell biology. Nature 420: 629–635, 2002.[CrossRef][Medline]

22. Filla MS, Woods A, Kaufman PL, Peters DM. beta1 and beta3 integrins cooperate to induce syndecan-4-containing cross-linked actin networks in human trabecular meshwork cells. Invest Ophthalmol Vis Sci 47: 1956–1967, 2006.[Abstract/Free Full Text]

23. Flinn HM, Ridley AJ. Rho stimulates tyrosine phosphorylation of focal adhesion kinase, p130 and paxillin. J Cell Sci 109: 1133–1141, 1996.[Abstract]

24. Folsom TD, Sakaguchi DS. Disruption of actin-myosin interactions results in the inhibition of focal adhesion assembly in Xenopus XR1 glial cells. Glia 26: 245–259, 1999.[CrossRef][ISI][Medline]

25. Fujiwara K, Pollard TD. Fluorescent antibody localization of myosin in the cytoplasm, cleavage furrow, and mitotic spindle of human cells. J Cell Biol 71: 848–875, 1976.[Abstract/Free Full Text]

26. Garrington TP, Johnson GL. Organization and regulation of mitogen-activated protein kinase signaling pathways. Curr Opin Cell Biol 11: 211–218, 1999.[CrossRef][ISI][Medline]

27. Gluck U, Ben-Ze'ev A. Modulation of alpha-actinin levels affects cell motility and confers tumorigenicity on 3T3 cells. J Cell Sci 107: 1773–1782, 1994.[Abstract]

28. Handel SE, Wang SM, Greaser ML, Schultz E, Bulinski JC, Lessard JL. Skeletal muscle myofibrillogenesis as revealed with a monoclonal antibody to titin in combination with detection of the alpha- and gamma-isoforms of actin. Dev Biol 132: 35–44, 1989.[CrossRef][ISI][Medline]

29. Herman IM, Pollard TD. Comparison of purified anti-actin and fluorescent-heavy meromyosin staining patterns in dividing cells. J Cell Biol 80: 509–520, 1979.[Abstract/Free Full Text]

30. Hirakawa M, Oike M, Karashima Y, Ito Y. Sequential activation of RhoA and FAK/paxillin leads to ATP release and actin reorganization in human endothelium. J Physiol 558: 479–488, 2004.[Abstract/Free Full Text]

31. Honda K, Yamada T, Endo R, Ino Y, Gotoh M, Tsuda H, Yamada Y, Chiba H, Hirohashi S. Actinin-4, a novel actin-bundling protein associated with cell motility and cancer invasion. J Cell Biol 140: 1383–1393, 1998.[Abstract/Free Full Text]

32. Ingber DE. The architecture of life. Sci Am 278: 48–57, 1998.[ISI][Medline]

33. Ireland GW, Voon FC. Polygonal networks in living chick embryonic cells. J Cell Sci 52: 55–69, 1981.[Abstract]

34. Jaffe AB, Hall A. Rho GTPases: biochemistry and biology. Annu Rev Cell Dev Biol 21: 247–269, 2005.[CrossRef][ISI][Medline]

35. Jia Z, Vadnais J, Lu ML, Noel J, Nabi IR. Rho/ROCK-dependent pseudopodial protrusion and cellular blebbing are regulated by p38 MAPK in tumour cells exhibiting autocrine c-Met activation. Biol Cell 98: 337–351, 2006.[CrossRef][ISI][Medline]