Mesangial Cell Activation by Bacterial Endotoxin - Europe PMC

American Journal of Pathology, Vol. 139, No. 2, August 1991 Copyright X) American Association of Pathologists

Mesangial Cell Activation by Bacterial Endotoxin Induction of Rapid Cytoskeletal Reorganization and Gene Expression

Stuart L. Bursten,* Frazier Stevenson,t Frank Torrano,t and David H. Lovettt

sponses to peptide mitogens. (Am J Pathol 1991,

139:371-382)

From the Department of Medicine,' Seattle Veterans

Administration Medical Center, University of Washington, Seattle, Washington; and the Department of Medicine,t San Francisco Veterans Administration Medical Center, University of California at San Francisco, San Francisco, California

Cultured glomerular mesangial cells (MC) respond to low concentrations of bacterial endotoxin (ET) by secreting prostaglandins and interleukin-1. To evaluate further the nature of ET-induced mesangial cell activation; the authors evaluated the effects of this agent on MC morphology and cytoskeletal organization. Bacterial ET, in concentrations as low as 1 ng/ ml, induced reversible membrane ruffling cellular rounding and extension of many filopodia and lamellopodia Augmented fluid-phase pinocytosis occurred in parallel as determined by transmission electron microscopy and tritiated sucrose uptake. These cellular morphologic and functional changes were associated with an extensive, but reversible, depolymerization of actin microfilaments. Actin gene expression was also modified by ET At 4 to 6 hours after ET exposure, Northern blot analysis showed a twofold to fourfold increase in actin mRNA levels. In situ hybridizations ofET-stimulated cells at the light and electron microscopic levels demonstrated a markedly asymmetric distribution of actin mRNA, which was localized in the cellular periphery at filopodial and lamellopodial extensions, presumably sites of new actin protein synthesis. It is concluded that ET effects on MC are distinct from the nonspecific lytic or 'toxic' actions described for other cell types. Endotoxin induces a global activation of this cell type associated with major changes in membrane structure, cytoskeletal organization, and gene expression, which resemble in many respects the re-

Gram-negative bacterial infection is associated with a number of functional and structural alterations in renal glomerular and tubular function. These changes in renal function may be mediated by two separate pathways. First, bacterial cell wall components, primarily endotoxin, induce the synthesis of vasoactive agents and cytokines by circulating inflammatory cells, such as monocytes, which can subsequently modulate renal hemodynamic and cellular function. The most important of these secreted factors may include the eicosanoids, plateletactivating factor, tumor necrosis factor (TNF), and interleukin-1 (IL-1). Second, it has been suggested that bacterial endotoxin exerts direct effects on the intrinsic renal cell populations, leading either to direct cytotoxicity or to alterations in cellular functional phenotypes.1-3 Given their strategic location within the renal vascular bed, glomerular endothelial and mesangial cells may be particularly relevant targets of this direct form endotoxin action on the kidney. Several studies have demonstrated that intrinsic mesangial cells are sensitive to endotoxin, resulting in the secretion of tissue factor, prostaglandins, IL-1, and TNF.1' 34 Recently we have found that the biologically active component of endotoxin, lipid A, activates cultured mesangial cells by specifically inducing phosphatidic acid synthesis.5 Phosphatidic acid has several intrinsic biologic activities that closely resemble those of peptide growth factors. These include the rapid induction of calcium flux and the hydrolysis of polyphosphoinositol to form IP3.6 In addition, phosphatidic acid induces cytoplasmic alkalinization and expression of c-fos.7 These observations have lead us to postulate that the patterns of Supported by Research Funds of the Veterans Administration. Accepted for publication April 12, 1991. Address reprint requests to David H. Lovett, 11 1J Medical Service, Division of Nephrology, San Francisco VAMC, 4150 Clement St., San Francisco, CA 94121.

371

372

Bursten et al

AJP August 1991, Vol. 139, No. 2

endotoxic cellular activation are analogous to those resulting from peptide mitogens, per se. Alterations in cellular shape, cytoskeletal organization, and gene expression represent major patterns of cellular response to peptide mitogens."12 In addition, changes in cellular shape can actually modify the subsequent expression of multiple genes, as well as the response to further mitogenic or matrix stimuli.13-15 Evidence is presented here that bacterial endotoxin, in noncytolytic concentrations, modulates cellular shape, cytoskeletal organization, and gene expression in patterns that closely resemble those induced by peptide growth factors. The direct activation, without cytolysis, of the intrinsic glomerular mesangial cell population during bacteremic states may lead to the expression of functional cellular phenotypes that modulate the glomerular proliferative and hemodynamic responses characteristic of these disorders.

Materials and Methods

Animals, Cells Male Sprague-Dawley rats, 125 to 150 g, were used as a source of mesangial cells, which were prepared and characterized as reported in detail.'

Media and Reagents Proliferating mesangial cells (MC; 4th through 6th passage) were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 20% heatinactivated fetal calf serum (FCS, Irvine Scientific, Irvine, CA), 100 U/ml penicillin, 100 ,g/ml streptomycin and 300 ,ug/ml glutamine (growth medium). Medium to induce viable, noncycling cells (rest medium) consisted of Medium 199 (Gibco, Grand Island, NY), 0.5% bovine serum albumin (BSA), 50 U/ml penicillin, 50 ,ug/ml streptomycin, 300 ,ug/ml glutamine, and 1 x 1 -6 mol/I (molar) insulin.1 All media components were screened for the presence of exogenous endotoxin using a Limulus amoebocyte lysate assay sensitive to less than 0.1 ng/ml. Highly purified lipopolysaccharide (LPS) from Escherichia co#l 011 1:B8 was the gift of Dr. J. L. Ryan, Yale University and was quantified as previously reported.1 Highly purified Lipid A from the heptose-deficient mutant S. minnesota R595 was obtained from Ribi Immunochem, Butte, Montana. Stock solutions of these reagents were prepared by sonication into phosphate-buffered saline (PBS) containing 0.1% defatted BSA (Sigma Chemical Co., St. Louis, MO) and stored at - 700C until use.

Scanning Electron Microscopy (SEM) Mesangial cells cultured on etched glass cover slips were grown to near confluency. To prepare resting cells (serum deprived), cultures were washed twice in warm PBS and cultured in rest medium. The medium was changed every 48 hours, and the cells were rendered quiescent within 96 to 120 hours.1 The discs of either growing or resting cells were washed twice with warm DMEM and incubated for 18 hours with DMEM supplemented with 0.5% FCS. This medium was removed and replaced with fresh DMEM containing 0.5% FCS and with various concentrations of LPS or Lipid A. Controls were given fresh medium alone. Discs were processed for SEM at serial times by washing twice in PBS (25°C), followed by fixation in 3.7% buffered glutaraldehyde for 45 minutes at room temperature. The discs were then washed, postfixed in 2% osmium tetroxide for 1 hour, and dehydrated in a graded series of ethanol. The discs were processed through critical point drying and sputtercoated with gold for 4 minutes. Scanning electron microscopy was performed with a Fuji scanning electron microscope.

Transmission Electron Microscopy Groups of resting or proliferating MC were prepared, stimulated with 100 ng/ml Lipid A for 2 hours, and processed for transmission electron microscopy (TEM) in the tissue culture flasks as reported.16

Cytoskeletal Studies The organization of filamentous cytoskeletal actin was determined on MC grown on cover slips as above. At various times the cells were washed in PBS (250C) and fixed for 10 minutes in 3.7% buffered glutaraldehyde. Fixed cells were permeabilized with acetone (- 200C) for 3 minutes, washed, and reacted in the dark with 20 U/ml NDBphallicidin (Molecular Probes, Junction City, OR) and examined by epifluorescence.

Pinocytosis Assay Fluid-phase pinocytosis was analyzed using 3H-sucrose as the marker. Mesangial cells were grown in 2-cm2 wells to a final density of approximately 5 x 1 0 cells per well. The cultures were washed twice, and placed in DMEM supplemented with 0.5% FCS for 18 hours. Thereafter the cells were washed and given fresh DMEM supplemented with 0.5% FCS and 2 RCi/well 3H-sucrose (specific activ-

Mesangial Activation by Endotoxin

373


ity 10.1 Ci/mmol/l [millimolar], New England Nuclear, Boston, MA). Lipid A, at 100 ng/ml, was added to the experimental groups; controls were maintained in medium alone. At serial times, the medium was removed and the cell layers washed six times with cold PBS. The layers were then solubilized in 10% sodium dodecyl sulfate solution and the amount of incorporated radioactivity determined by P-scintillation counting. The cellular uptake of tritiated sucrose was determined in six wells for each time.

Northern Blot Analyses Total RNA from control and endotoxin-treated cultures was extracted by the guanidine isothiocyanate/cesium chloride method.17 For Northern blot analysis, 10 ,ug total RNA per lane were electrophoresed in 1% agarose gels containing formaldehyde and transferred to nitrocellulose filters. The positions of ribosomal 28 and 18s RNAs were determined by visualization of acridine orange-stained gels. Blots were first hybridized with a 32P nick-labeled, 1.8-kb full-length insert for human beta-actin (cut with Bam Hi from plasmid HF-beta-Al, the gift of Dr. Alan Pollock, UCSF), which recognizes the highly conserved beta-actin sequence of the rat. After hybridization for 16 hours at 420C, the blots were washed with 0.1 x SSC at 65°C and autoradiographed at - 70°C. The blots were then stripped, and rehybridized with a 32p-labeled antisense RNA probe to the RNA encoding for the 28S ribosomal protein. This RNA probe was transcribed from the plasmid pSPEE6.7 (the gift of Dr. Allan Pollock) with SP6 polymerase, using standard labeling conditions. Hybridization was performed at 550C, after which the blots were washed twice with 2x SSC (0.15 M NaCI, 0.015 M sodium citrate, pH 7.0) treated with 2 ,ug/ml RNAse at 370C for 20 minutes, and washed in 0.1 x SSC at 650C for 30 minutes. Autoradiographs of the actin and 28S RNA signals were quantified by densitometry.

Actin In Situ Hybridization Various study groups of MC were prepared for in situ localization of actin mRNA after growth to nearconfluence on glass cover slips. The cells were washed, fixed in 3.7% buffered paraformaldehyde for 20 minutes at room temperature and permeabilized in 70% ethanol for 5 minutes. After washing in PBS, the cells were hybridized. The full-length P-actin insert from plasmid HFbkAl was cut with Aval, yielding a 300-bp Aval-Aval fragment that was gel purified and subcloned into pGEM 3zf (Promega, Madison WI), yielding the plasmid pMID3. 35S-labeled (35S-CTP, NEN Research Products, Wilmington, DE) antisense actin RNA probes were generated with SP6 RNA polymerase after plasmid linearization with

EcoRl. Nonhybridizing control (sense) probes were generated with T7 RNA polymerase after linearization with Hinc II. An additional control for specificity of actin hybridization was performed by mixing labeled actin probe with increasing amounts of unlabeled probe to compete for binding. 35S-labeled antisense probes for murine interleukin-1 ,B (IL-1 P) were generated using SP6 polymerase, after linearization with Nco of a pGEM 7zf plasmid containing the full-length coding sequence for murine IL-1 P. All RNA transcripts were suspended in hybridization buffer consisting of 50% deionized formamide, 10% dextran sulfate, 10 mmol/l TRIS/HCI, pH 7.4, 1 mmol/l ethylenediaminetetra-acetic acid, 0.02% each of Ficoll, polyvinylpyrrlidine (PVP), and BSA, 600 mmol/l NaCI, 1 mg/ml yeast tRNA, and 100 ,ug/ml sonicated salmon sperm DNA. Probes were denatured at 650C for 10 minutes before use. The hybridizing cover slips were sealed with rubber cement and incubated at 370C for 18 hours. Thereafter, the sealant was removed and the slips washed in 2 x SSC for 15 minutes at 250C, 2 x SSC for 30 minutes at 37°C, 0.1 x SSC for 10 minutes at 500C, and 0.1 x SSC for 10 minutes at 250C. The slips were dehydated in ethanol, dipped in Kodak NTB-2 emulsion, and exposed for 5 days. After development, the slips were stained with Mayer's hematoxylin and mounted. Additional controls included cells treated with 100 ,ug/ml RNAse before hybridization. For electron microscopic actin mRNA localization, plasmid pMID3 was linearized with Eco Ri or Hinc 11 as above. Biotinylated RNA probes were transcribed with either SP6 RNA polymerase or T7 polymerase in a reaction mixture containing 40 mmol/l TRIS/HCI, pH 8.3, 6 mmol/l Mg acetate, 400 ,umol/l (micromolar) each of adenosine triphosphate, cytidine triphosphate, and guanosine triphosphate, 1.5 mmol/I spermidine, and 1 mmol/l biotin-1 1 -UTP (Enzo Biochemical, New York, NY). Biotinylated RNA probes were purified by serial precipitation. Cultured rat MC were washed twice in PBS and fixed with 3.7% buffered paraformaldehyde as above. After washing and ethanol dehydration, the cells were embedded in LR White resin (Ted Pella Co, Redding, CA). Ultrathin sections were cut and picked up on nickel grids. The grids were incubated overnight at 370C with the biotinylated RNA probes in the hybridization buffer detailed above. Thereafter the grids were washed five times in PBS, blocked with 1% BSA (RIA grade, Sigma) in PBS, pH 7.4, for 20 minutes, blotted on filter paper, and incubated with affinity-purified goat anti-biotin gamma G immunoglobulin (Vector Laboratories, Burlingame, CA) at 10 ,ug/ml in 1% BSA overnight. The grids were washed, reblocked in 1% BSA in PBS, pH 8.2, and incubated with protein A-colloidal gold complex, 15 nm, (E-Y Laboratories, San Mateo, CA) diluted 1:10 in PBS, pH 8.2, for 2 hours. The grids were washed four times with PBS, pH

374

Bursten et al


.1

_71; s~ft,.XSAJS8,,w


375


LI

4c 4-

ILI o

Control

*

+LPS

0.1

.5

1

10

100

500 1000

Concentration LPS/Lipid A (ng/mI) 0

Time (Hrs)

Figure 2. Time course of mesangial cell morphologic changes after exposure to 100 ng/ml LPS (0); controls were maintained in medium alone (0). Parameters of 'activation" as defined in Materials and Methods; data given as means + SEM

8.2, once with deionized water, and stained with uranyl acetate. The total number of gold particles per cell in the experimental and control groups were determined by counting over sections containing a complete longitudinal section of a given cell (n = 20). The cellular distribution of the gold particles was determined by counting gold particles associated with either lamellopodial or filopodial processes and comparing these with the total gold particle counts/cell.

Statistics For the morphologic (SEM) and cytoskeletal studies, the cells were evaluated relative to controls maintained in rest medium. Groups of 150 cells on three cover slips for each time or concentration point were considered. Cells were counted as having significantly altered morphology if any of the following changes were present: elongation, rounding, development of more than 10 membrane processes (evaginations, filopodia), or ruffling of more than one lamellopodial process. Cells were counted as having significantly altered actin cytoskeletal organization based on the presence of extensive depolymerization of the prominent stress fibers present in these cells. Data were analyzed by standard analysis of variance and multiple

Figure 3. Concentration dependence ofLPS and LipidA-mediated changes in mesangial cell morpbology. Cells were exposed to serial concentrations of LPS (0) of Lipid A (V) and the percentage of activated cells determined after 4 hours of stimulation. Mesangial cells, in which a resting state had been induced by prolonged culture in rest medium (see Matenals and Methods) were virtualsy nonresponsive to LPS (0) or Lipid A (not shown). Nonstimulated controls (v) varied by less than 5%.

comparison techniques, with P values of less than 0.05 considered significant.

Results Endotoxin Alters Mesangial Cell Morphology and Cytoskeletal Organization The first series of experiments analyzed the effects of highly purified LPS on the morphologic characteristics of cultured mesangial cells. The cells were exposed to a concentration of LPS (100 ng/ml), which was previously found to yield optimal stimulation of prostaglandin and IL-1 synthesis.' Nonstimulated MC appeared as interwoven, shinglelike structures on SEM, with a smooth and uniform surface appearance (Figure 1A). After 60 minutes of exposure to LPS, the MC demonstrated rounding or narrowing, with the development of numerous filopodia and membrane evaginations (Figure 1 B, C). At this concentration of LPS there was no obvious membrane blebbing, which is a frequent manifestation of cellular toxicity. By 2 hours of exposure, many of the cells had rounded up extensively and had developed a prominent pattern of lamellopodia. In many cases, the lamellopodia were concentrated at the leading edges of elongated, apparently

Figure 1. A: Scanning electron microscopic (SEM) appearance of nonstimulated cultured glomerular mesangial cells. Cells are arranged as smooth-surfaced, interwoven, shinglelike structures. B,C: LPS-stimulated cells after 60 minutes exposure to 100 ng/ml. Cells are rounding up, with extension of many filopodial processes. D: At 2 hours ofLPS stimulation, there is extensive development of lamellopodial processes, particularly at the leading edges, as shown here. (A, x 1400; B, x 1750; C,D, x 6500).

376

Bursten et al


Figure 4. Modulation of cellular cytoskeletal actin organization by Lipid A. A: Control, nonstimulated mesangial cells demonstrate

prominent, axially oriented actin filaments contained within stressfibers. B: Exposure to 100 ng/ml Lipid A for 15 minutes results in extensive actin filament dissolution, which by 1 hour of exposure (C) is virtually complete. (A,B, x 1400; C, x800).


377


c 100 ._

0

N 80

10o gc 60

" 40 C-

Figure 5. Time course of Lipid A-mediated actin filament depolymerization. Controls (0) were maintained in medium alone; experimental groups were exposed to 100 ng/ml Lipid A (0) or to the positive control, 5 ng/ml PMA (7). Data are given as the mean percentage of cells with depolvmerized actin + SEM.

u 20

0.S

0

motile cells (Figure 1 D). Resting cells (serum-deprived) did not respond to LPS with any discernible morphologic alterations. The time course of these LPS-dependent morphologic alterations of the mesangial cells is depicted in Figure 2. After 1 hour of LPS stimulation, approximately 20% of the cells demonstrated some degree of morphologic change, and by 4 hours nearly 80% of the cells had done so. These alterations in mesangial cell morphology were dependent on the concentration of LPS used (Figure 3). Concentrations of LPS as low as 1 ng/ml induced significant degrees of shape change when assessed after 4 hours of exposure. Concentrations of 10 ng LPS/ml activated nearly 50% of the cells. As expected, Lipid A demonstrated a greater degree of biologic activity, with concentrations as low as 500 picograms/mI inducing morphologic alterations in 50% of the cells (Figure 3). In contrast, resting cells were virtually unresponsive to LPS across a very broad concentration range. The absence of a shape change response by resting cells closely parallels our previous finding that resting mesangial cells fail to secrete prostaglandins or IL-1 in response to LPS stimulation.1 Modulation of cellular morphology is highly dependent on the organization of the cytoskeleton. In cultured mesangial cells, filamentous actin contained within stress fibers and the subcortical area is prominently revealed by staining with NBD-phallacidin (Figure 4A). Lipid A induced rapid changes in the state of filamentous actin polymerization of these cells. By 15 minutes there was extensive actin microfilament dissolution, which occurred before obvious changes in cellular morphology (Figure 4B). After exposure to Lipid A for 1 hour (Figure 4C), virtually all microfilamentous actin was depolymerized. Most actin staining in these cells was present in a punctate perinuclear pattern, although some staining also could be seen in the subcortical area. By 24 hours the

15

30

45

60

75

90

105

Time (Min.)

cells had resumed their previous flat appearance with well-organized microfilaments, even in the continued presence of Lipid A (not shown). The time course of the Lipid A-mediated changes in actin microfilament organization is shown in Figure 5 and is compared with the time course of the positive actin depolymerization control, phorbol myristate acetate.18

Ultrastructural and Pinocytosis Studies The changes in cellular morphology and actin polymerization induced by endotoxin or Lipid A were associated with prominent alterations in cellular ultrastructure. Transmission electron microscopy of control, nonstimulated cells demonstrated typical elongated mesangial cell morphology, with moderately abundant polyribosomes and prominent microfilaments (Figure 6A). Mesangial cells incubated with 100 ng/ml Lipid A for 2 hours showed major changes in morphologic organization, including cellular rounding, extension of multiple filopodial projections, and formation of numerous endocytic vacuoles (Figure 6B). There was no evidence at this level for any cytotoxic changes, such as membrane blebbing or loss of organellar integrity. Measurement of cellular pinocytosis rates using 3Hsucrose confirmed the impressions gained by TEM. Whereas damaged cells are unable to offset leakage of pinocytosed material through deteriorating membranes, activated cells are capable of increasing rates of pinocytosis by severalfold.19 Exposure of mesangial cells to 100 ng/ml of Lipid A resulted in an approximately eightfold increase in cellular rates of pinocytosis. The time course of this process closely parallels that of Lipid A-induced actin microfilament depolymerization (Figure 7). The cellular pinocytosis rate of nonstimulated cells was 45 + 15

378

Bursten et al


r :P;l

*

~A a

)t

Figure 6. A: Transmission electron micrographic (TEM) appearance of nonstimulated cultured mesangial cells showing typical elongated morphology, abundant polyribosomes, and prominent microfilaments (MF, arrous). B: MCs stimulated with 100 nglml Lipid Afor 2 hours demonstrate a rounded morphology, multiplefilopodial extensions, and many endoc'tic vacuoles (arrows). (A, X6000, inset, x 10,000; B, x8000).


379


utes), Lipid A-mediated stimulation of mesangial cell pinocytosis persisted over several hours of culture. Taken together, the ultrastructural and pinocytosis studies indicate that Lipid A induces a state of cellular activation that is unassociated with signs of cellular toxicity. Additional studies measuring release of radiolabeled chromium also excluded any direct toxic effects of Lipid A or endotoxin (data not shown).

10.0-

a.

5.0-

0

E

1.0

-

Endotoxin Effects on Cytoskeletal Actin Gene Expression

0)

0

,o

0.5 a.

o Um

*

Control Lipid A

C)

0.1 -

0y .

Ir

I.

I

0

1

2

s

s

3

4

I

Time (Hrs) Figure 7. Measurement ofmesangial cellpinocytosis rates in controls (0) and cultures exposed to 100 ng/ml Lipid A (0). Data given as means ± SEM

nanoliters/1 06 cells/hour. Lipid A-stimulated cells had pinocytosis rate of 355 + 70 nanoliters/1 06 cells/hour. These rates of basal and stimulated pinocytosis by mesangial cells closely resemble those exhibited by epidermal growth factor-stimulated A-431 cells (basal: 37.5 nanoliter/1 06 cells/hour versus epidermal growth factor (EGF)-stimulated: 378.1 nanoliter/1 06 cells/hour20). In contrast to the rapid and limited effect of EGF (