Complement C5b-9 Membrane Attack Complex Increases Expression of Endoplasmic Reticulum Stress Proteins in Glomerular Epithelial Cells*

In the passive Heymann nephritis (PHN) model of membranous nephropathy, complement C5b-9 induces glomerular epithelial cell (GEC) injury, proteinuria, and activation of cytosolic phospholipase A2 (cPLA2). This study addresses the role of endoplasmic reticulum (ER) stress proteins (bip, grp94) in GEC injury. GEC that overexpress cPLA2 (produced by transfection) and “neo” GEC (which expresses cPLA2at a lower level) were incubated with complement (40 min), and leakage of constitutively expressed bip and grp94 from ER into cytosol was measured to monitor ER injury. Greater leakage of bip and grp94 occurred in complement-treated GEC that overexpress cPLA2, as compared with neo, implying that cPLA2 activation perturbed ER membrane integrity. After chronic incubation (4–24 h), C5b-9 increased bip and grp94 mRNAs and proteins, and the increases were dependent on cPLA2. Expression of bip-antisense mRNA reduced stimulated bip protein expression and enhanced complement-dependent GEC injury. Glomerular bip and grp94 proteins were up-regulated in proteinuric rats with PHN, as compared with normal control. Pretreatment of rats with tunicamycin or adriamycin, which increase ER stress protein expression, reduced proteinuria in PHN. Thus, C5b-9 injures the ER and enhances ER stress protein expression, in part, via activation of cPLA2. ER stress protein induction is a novel mechanism of protection from complement attack.

Activation of the complement cascade near a cell surface leads to assembly of terminal components, exposure of hydrophobic domains, and insertion of the C5b-9 membrane attack complex into the lipid bilayer of the plasma membrane (1,2). Assembly of C5b-9 results in formation of transmembrane channels or rearrangement of membrane lipids with loss of membrane integrity. Nucleated cells require multiple C5b-9 lesions for lysis, but at lower doses, C5b-9 induces sublethal (sublytic) injury and various metabolic effects (1)(2)(3)(4)(5)(6)(7)(8). An example of sublytic C5b-9-mediated cell injury in vivo is passive Heymann nephritis (PHN) 1 in the rat, a widely accepted model of human membranous nephropathy (9). In PHN, antibody binds to visceral glomerular epithelial cell (GEC) antigens and leads to the in situ formation of subepithelial immune complexes (9,10). C5b-9 assembles in GEC plasma membranes, "activates" GEC, and leads to proteinuria and sublytic GEC injury (9 -14). Based on studies in GEC culture and in vivo, C5b-9 assembly induces transactivation of receptor tyrosine kinases (15), an increase in cytosolic free Ca 2ϩ concentration ([Ca 2ϩ ] i ), and activation of protein kinase C, as well as cytosolic phospholipase A 2 -␣ (cPLA 2 ) (16 -19). cPLA 2 is an important mediator of C5b-9-dependent GEC injury. First, arachidonic acid (AA) released by cPLA 2 is metabolized in GEC via cyclooxygenases-1 and -2 to prostaglandin E 2 and thromboxane A 2 (20), and inhibition of prostanoid production reduces proteinuria in PHN (21)(22)(23)(24)(25) and in human membranous nephropathy (26). Second, cPLA 2 may mediate GEC injury more directly (16). cPLA 2 (group IV PLA 2 ) is regulated by [Ca 2ϩ ] i and phosphorylation (27,28). Stimuli that raise [Ca 2ϩ ] i in the submicromolar range may induce translocation of cPLA 2 from cytosol to an intracellular membrane, where cPLA 2 would bind via its N-terminal Ca 2ϩ -dependent lipid binding or C2 domain, gaining access to phospholipid substrate. We have demonstrated that cPLA 2 is the major PLA 2 in GEC and that complement enhances cPLA 2 phosphorylation and catalytic activity (16). Moreover, C5b-9 increases free AA in GEC, and release of AA is amplified by overexpression of cPLA 2 (16,17). Recently, we demonstrated that cPLA 2 localizes and hydrolyzes phospholipids at the plasma membrane in GEC, the membrane of the endoplasmic reticulum (ER), and the nuclear envelope but not at mitochondria or Golgi (29). Thus, the activation of cPLA 2 and release of AA are compartmentalized to specific organelles, but it is presently unknown if hydrolysis of membrane phospholipids by cPLA 2 leads to injury of these organelles in GEC.
Based on studies in cell culture, complement attack may injure cells but may also activate pathways that restrict injury or facilitate recovery. One mechanism of protection from complement attack is "ectocytosis" (shedding) of C5b-9 complexes from cell membranes (1,2). There are undoubtedly other recovery mechanisms that require delineation. For example, ex-posure of cells to environmental stress increases expression of stress proteins in cellular compartments such as the ER. There are several types of ER stress responses, including the "unfolded protein response" (30 -32). ER stress proteins are believed to be induced by accumulation of abnormal proteins or depletion of ER Ca 2ϩ stores and include the glucose-related proteins (grp), grp94, bip (grp78), and erp72. Tunicamycin, a nucleoside antibiotic that blocks N-linked glycosylation and is believed to cause an accumulation of unfolded proteins in the ER, and the Ca 2ϩ ionophore, ionomycin, which can deplete Ca 2ϩ from intracellular stores, can induce ER stress proteins (30 -32). Under normal conditions, ER stress proteins might serve as protein chaperones for exocytosis from ER, and they might complex with defective proteins and target them for degradation. During stress, the induction of ER stress proteins may limit accumulation of abnormal proteins in cells (31,32). Generally, regulation of ER stress proteins appears to be controlled at the transcriptional level, and requires several hours and de novo protein synthesis. Moreover, exposure of cells to mild stress, which induces ER stress proteins, may be protective to additional insults (33,34), although prolonged or more substantial ER stress may lead to cell death by apoptosis (35).
The aim of the present study was to determine if C5b-9mediated activation of cPLA 2 and GEC injury are associated with induction of the ER unfolded protein response. We demonstrate that C5b-9 may damage the ER and enhance expression of ER stress proteins via activation of cPLA 2 . Induction of ER stress proteins limits complement-dependent GEC injury in culture and in PHN.

Methods
Cell Culture and Transfection-Rat GEC culture and characterization has been detailed previously (36). GEC were cultured in K1 medium on plastic substratum. The method for stable transfection of GEC (e.g. to produce GEC that stably overexpress cPLA 2 ) was described previously (16,17), and the same method was used to produce GEC that stably expressed bip antisense cDNA. GEC that express only the neomycin-resistance gene (neo) were used as control. Briefly, GEC were co-transfected with the expression vector of interest, and pRc/RSV (molar ratio 12.5:1), using a CaPO 4 technique. Colonies of GEC resistant to G418 were isolated and expanded. Transfectants were selected for a high level of cPLA 2 activity and protein expression or, in the case of bip antisense cDNA, inhibition of stimulated bip protein expression (determined by immunoblotting). Compared with parental or neo GEC, PLA 2 activity was increased ϳ5-fold in GEC that overexpress cPLA 2 , but was nevertheless within a physiological range, and comparable to the level in Madin-Darby canine kidney cells (16,17).
Incubation of GEC with Antibody and Complement-To activate complement, GEC in monolayer culture were incubated with rabbit anti-GEC antiserum (5% v/v) in modified Krebs-Henseleit buffer containing 145 mM NaCl, 5 mM KCl, 0.5 mM MgSO 4 , 1 mM Na 2 HPO 4 , 0.5 mM CaCl 2 , 5 mM glucose, and 20 mM Hepes, pH 7.4, for 40 min at 22°C (16,17). GEC were then incubated with normal human serum (diluted in Krebs-Henseleit buffer in acute incubations, or K1 medium in chronic incubations) or with heat-inactivated (decomplemented) human serum (56°C) in controls, at 37°C. In some experiments, antibodysensitized GEC were incubated with C8-deficient human serum or C8-deficient serum supplemented with purified C8 (80 g/ml undiluted serum). We have generally used heterologous complement to facilitate studies with complement-deficient sera, and to minimize possible signaling via complement-regulatory proteins, however, in earlier studies results of several experiments were confirmed with homologous (rat) complement. There was some variability in sublytic and lytic concentrations of complement among batches of sera. In GEC, complement is not activated in the absence of antibody, and antibody does not independently affect free [ 3 H]AA (16 -18).
Immunofluorescence microscopy for sheep IgG, rat IgG, and rat C3 was performed as described previously (11). Briefly, 4-m cryostat kidney sections were stained with fluorescein-conjugated IgG fractions of monospecific antisera. The immunofluorescence signals from whole glomeruli were evaluated using a Leitz immunofluorescence microscope with visual output connected to a Nikon UFX-II photomultiplier and camera, similar to a method described earlier (38). Densitometry readings were done under immersion oil, and the biopsy material was magnified 400 times, such that the glomerular cross-section filled the majority of the densitometry field. The time required to collect an image from a glomerulus is inversely proportional to immunofluorescence intensity. Times required to collect images from three representative glomeruli in each tissue section were recorded and averaged. Serum creatinine measurements were performed in the clinical biochemistry laboratory of the Royal Victoria Hospital (Montreal, Quebec).
Measurement of Free [ 3 H]AA-GEC phospholipids were labeled to isotopic equilibrium with [ 3 H]AA for 48 -72 h, as detailed previously (16 -20). Lipids were extracted from ϳ1 ϫ 10 6 cells and cell supernatants. Methods for extraction and separation of radiolabeled lipids by thin layer chromatography were published previously (16 -20).
Northern Hybridization and Immunoblotting-Northern hybridization was performed as described previously (20). Briefly, total RNA was extracted from GEC using the TRIzol reagent. RNA was separated by gel electrophoresis (1% agarose containing 1.9% formaldehyde) and was transferred to a nylon membrane. Coding regions of rat bip, grp94, and erp72 cDNAs were radiolabeled with [␣-32 P]dCTP using the Random Primer DNA Labeling System. Membranes were hybridized in buffer containing 1% bovine serum albumin, 7% SDS, 0.5 M sodium phosphate, pH 6.8, 1 mM EDTA, and 1-2 ϫ 10 6 cpm/ml of radiolabeled probe for 16 h at 42°C. Membranes were washed in buffer containing 0.5% bovine serum albumin, 5% SDS, 40 mM sodium phosphate, pH 6.8, 1 mM EDTA twice for 20 min at 65°C, and then buffer containing 1% SDS, 40 mM sodium phosphate, pH 6.8, 1 mM EDTA four times for 20 min at 65°C. Membranes were exposed to x-ray film with an intensifying screen at Ϫ70°C for 48 -72 h.
For immunoblotting, GEC or glomerular lysates were mixed with Laemmli sample buffer and were subjected to SDS-PAGE. Proteins were transferred to nitrocellulose paper, blocked with 5% fat-free dry milk in 20 mM Tris, 50 mM NaCl, pH 7.5, with 0.05% Tween 20 (15)(16)(17). Blots were incubated with primary antibodies. After washing with Tris-buffered saline/Tween 20 solution, blots were incubated with secondary antibody and developed using the enhanced chemiluminescence technique.
Quantification of Northern blots and immunoblots was performed by densitometry. Blots were scanned, specific bands of interest were selected, and the density of the bands was measured using National Institutes of Health Image software. Results are expressed in arbitrary units. Preliminary studies demonstrated that there was a linear relationship between densitometric measurements and the amounts of protein loaded onto gels.
Measurement of Complement-dependent Cytotoxicity-Complementmediated cell lysis was determined by measuring release of lactate dehydrogenase (LDH), as described previously (15,36). Specific release of LDH was calculated as [NS Ϫ HIS]/[100 Ϫ HIS], where NS represents the percentage of total LDH released into cell supernatants in incubations with normal serum, and HIS is the percentage of total LDH released into cell supernatants in incubations with heat-inactivated serum (36).
Statistics-Data are presented as mean Ϯ S.E. The t statistic was used to determine significant differences between two groups. For more than two groups, one-way analysis of variance was used to determine significant differences among groups, and where significant differences were found, individual comparisons were made between groups using the t statistic, and adjusting the critical value according to the Bonferroni method. Two-way analysis of variance was used to determine significant differences in multiple measurements among groups.

RESULTS
Complement Activates cPLA 2 and Induces GEC Injury-Incubation of GEC with antibody and complement increases free AA (16 -19). The increase is evident within 40 min, and persists for at least 24 h (217 Ϯ 46% control at 24 h). Stable overexpression of cPLA 2 in GEC amplifies the complement-induced increase in free AA (465 Ϯ 183% control at 24 h) (19). Table I shows levels of free [ 3 H]AA in complement-treated GEC that stably overexpress cPLA 2 . Incubation of antibody-sensitized GEC with normal serum as the complement source increased [ 3 H]AA more than 3-fold, as compared with heat-inactivated (decomplemented) serum. C8 is the key component of the C5b-9 membrane attack complex. Incubation of GEC with C8-deficient serum reconstituted with purified C8 also increased free [ 3 H]AA more than 3-fold, as compared with unreconstituted C8-deficient serum or with heat-inactivated serum. Therefore, PLA 2 -mediated release of AA is due to assembly of C5b-9, whereas C5b-7 has no significant effect.
Complement-induced activation of cPLA 2 leads to phospholipid hydrolysis at the membrane of the ER (29). We assessed whether hydrolysis of ER membrane phospholipids may induce injury to the ER compartment. Resting GEC express the ER stress proteins, bip and grp94, and these proteins are localized mainly in the ER (Fig. 1). The integrity of the ER membrane was monitored by the leakage of constitutively expressed bip and grp94, from the ER compartment into the cytosol. Brief incubation of GEC with complement (2.5% normal serum, 40 min) induced significant increases in the amounts of bip or grp94 in the cytosol of the GEC that overexpress cPLA 2 (a representative immunoblot is shown in Fig. 1A, and densitometric quantification is given in Table IIA). The amount of bip or grp94 in the cytosol represented only a minor proportion of the total (Ͻ15%), and brief incubation with complement did not induce significant increases in total ER stress protein expression ( Fig. 1). At the concentration of complement that induced increases in bip and grp94 in the cytosol of GEC that overexpress cPLA 2 , there were no significant increases in bip or grp94 in the cytosol of neo GEC (Table IIA). However, increases in bip and grp94 could be detected in the cytosol of neo GEC when the concentration of complement was increased (4.0% normal serum, Fig. 1B and Table IIB). These results suggest that activation of cPLA 2 by complement perturbed the ER membrane sufficiently to allow a small portion of bip or grp94 to leak into the cytosol. Sublytic complement did not cause bip to leak out of cells, i.e. through the plasma membrane (not shown).
Overexpression of cPLA 2 in GEC enhanced complement-mediated cytotoxicity (16). To assess the role of endogenous cPLA 2 in complement-mediated injury, antibody-sensitized neo GEC were incubated with serially increasing doses of complement in the presence or absence of the cPLA 2 inhibitor, MAFP (39). GEC injury was quantified by monitoring release of LDH, a sensitive measure of cell viability. Cytolysis was lower in the presence of MAFP (Table III), suggesting that inhibition of cPLA 2 activity reduces complement-mediated injury. FIG. 1. Complement-induced activation of cPLA 2 affects the integrity of the membrane of the ER. A, neo GEC (which express a low endogenous level of cPLA 2 ) or GEC that overexpress cPLA 2 were incubated in duplicate with anti-GEC antibody and 2.5% normal serum (NS; to form sublytic C5b-9), or heat-inactivated serum (HIS) in controls, for 40 min. After collection of supernatants, GEC plasma membranes were briefly permeabilized with digitonin (37 g/ml) to recover the cytosol. Then, membranes and contents of the ER were solubilized by the addition of 1% Triton X-100. Digitonin and Triton fractions were immunoblotted with antibodies to bip or grp94 (in duplicate). An increase in bip and grp94 is present in the digitonin fraction of complement-treated GEC that overexpress cPLA 2 (seventh and eighth lanes from the left). B, antibody-sensitized neo GEC were incubated with 4.0% normal serum or heat-inactivated serum in controls, for 40 min. bip and grp94 are increased in the digitonin fraction of complementtreated GEC.

TABLE II
Translocation of ER stress proteins from ER to cytosol A: Antibody-sensitized neo GEC and GEC that overexpress cPLA 2 were incubated with 2.5% normal serum, or heat-inactivated serum in controls, for 40 min (a representative experiment is shown in Fig. 1A). B: Antibody-sensitized neo GEC were incubated with 4.0% normal serum or heat-inactivated serum in controls (a representative experiment is shown in Fig. 1B). The preparation of cytosolic fractions is described in the legend to Fig. 1. Densitometry of bip and grp94 in cytosol is presented as normal serum/heat-inactivated serum (-fold control). GEC

Complement Enhances Expression of ER Stress Proteins in
GEC-Assembly of C5b-9 may result in cell injury and may in parallel activate mechanisms that restrict or limit the amount of injury. Because C5b-9-induced activation of cPLA 2 appeared to disrupt the integrity of the ER membrane, as a consequence, the function of the ER may have been altered. Thus, we proceeded to assess whether exposure of GEC to complement attack would lead to increased expression of ER stress proteins. Incubation of antibody-sensitized GEC that overexpress cPLA 2 with sublytic complement for 4 h resulted in increased mRNA levels of bip, grp94, and erp72 ( Fig. 2A). The signals on Northern blots were quantified by densitometric analysis. Bip, grp94, and erp72 mRNAs increased 1.7 Ϯ 0.5-fold (n ϭ 4, p Ͻ 0.02), 1.4 Ϯ 0.2-fold (n ϭ 6, p ϭ 0.01), and 1.3 Ϯ 0.2-fold (n ϭ 4, p Ͻ 0.04), respectively, as compared with control. Complement had no effect on the mRNA level of the cytosolic stress protein, Hsp70 (not shown). Incubation of GEC with C5b-9 induces an increase in [Ca 2ϩ ] i and activates protein kinases (17,18). To determine the effect of increased [Ca 2ϩ ] i on bip, grp94, and erp72 mRNAs, GEC were incubated with the Ca 2ϩ ionophore, ionomycin, for 24 h. Ionomycin increased bip, grp94, and erp72 mRNAs (Fig. 2B), 1.7-, 2.3-, and 1.4-fold, respectively, as compared with control (two experiments). The effect of tunicamycin, one of the most potent inducers of ER stress proteins, is shown for comparison (Fig. 2B). Tunicamycin increased bip, grp94, and erp72 mRNAs 5.5-, 5.4-, and 3.1-fold, respectively, as compared with control (two experiments).
The next series of experiments addressed changes in protein levels of bip, grp94, and erp72 in GEC, and the role of cPLA 2 . Incubation with ionomycin for 24 h increased ER stress protein expression (Fig. 3). The effect of ionomycin was more prominent in the GEC that overexpress cPLA 2 (1.6-to 2.7-fold increases), as compared with neo (1.3-to 1.7-fold increases). Tunicamycin (shown for comparison as a positive control) was generally a more potent inducer of ER stress proteins, and its effect was similar in neo GEC and GEC that overexpress cPLA 2 (1.7-to 2.9-fold increases). These results indicate that the effect of ionomycin is, at least in part, mediated via activation of cPLA 2 , whereas changes secondary to tunicamycin are cPLA 2independent. Chronic incubation of antibody-sensitized GEC with complement (4 -24 h) induced increases in bip and grp94 protein expression (Fig. 4, A-C). In these experiments, the increases in protein levels (30 -50% above control at 24 h) were evident mainly in the GEC that overexpress cPLA 2 . An upward trend was present in neo GEC, but the change did not reach statistical significance. It should be noted that complementinduced increases in [Ca 2ϩ ] i are similar in magnitude in neo GEC and in GEC that overexpress cPLA 2 (17). Thus, the effect of complement on bip and grp94 expression is, at least in part, mediated via activation of cPLA 2 . To address the involvement of endogenous cPLA 2 , antibody-sensitized neo GEC were incubated with a higher concentration of complement, with or with-out the cPLA 2 inhibitor, MAFP. At the higher dose, significant complement-induced increases in bip and grp94 protein levels were evident in neo GEC, and the increases were inhibited by MAFP (Fig. 4D). The complement-induced increases in bip and grp94 were functionally important, as demonstrated in the studies of complement-induced injury (discussed below). We did not study the effect of complement on erp72, because preliminary studies demonstrated that complement-induced changes in erp72 protein levels were not readily detectable.
Increases in the expression of ER stress proteins could be due to cPLA 2 -mediated phospholipid hydrolysis and membrane injury or, secondarily, to signals triggered by cPLA 2 -generated lipid products or their metabolites (prostanoids). GEC that overexpress cPLA 2 were incubated with ionomycin or ionomycin plus the cyclooxygenase inhibitor, indomethacin (10 M), at a concentration known to inhibit cyclooxygenase activity (20). Ionomycin increased bip and grp94 expression (as in Fig. 3), and this effect was not inhibited by indomethacin, suggesting that prostanoid production was not involved in ER stress protein up-regulation (Fig. 5). Products of PLA 2 -induced phospholipid hydrolysis include AA and lysophospholipid. Addition of AA (10 M) did not enhance bip or grp94 expression (Fig. 5), although a similar concentration of AA can induce prostanoid generation in GEC (20). Lysophosphatidylcholine was also ineffective in enhancing ER stress protein expression (Fig. 5). Thus, increases in bip and grp94 are most likely triggered by cPLA 2 -mediated phospholipid hydrolysis and membrane perturbation.
Bip Antisense mRNA Expression and Effects on GEC Injury-To determine if complement-mediated induction of ER stress proteins is functionally important, GEC were stably transfected with bip antisense cDNA. Three clones that demonstrated a reduction in tunicamycin-stimulated bip protein expression, as compared with neo GEC were selected for further studies (Fig. 6A). In these bip antisense clones, tunicamycin generally stimulated increases in grp94 and erp72 protein expression similar to neo GEC. The increases tended to be blunted in one of the clones (Fig. 6A); however, previous studies have reported that bip antisense mRNA expression may also decrease induction of other ER stress proteins (33,40). When the GEC clones that express bip antisense mRNA were incubated with 1 M ionomycin for 24 h, there was greater cytolysis  2. Expression of ER stress protein mRNAs in GEC. GEC that overexpress cPLA 2 were incubated with anti-GEC antibody and 2.5% normal serum (NS; to form sublytic C5b-9), or heat-inactivated serum (HIS) in controls, for 4 h (A). Total RNA was extracted and subjected to Northern blotting. The control incubations demonstrate that there is some basal expression of bip, grp94, and erp72 mRNAs in GEC. Complement increased expression of bip, grp94, and erp72 mRNAs. B, GEC that overexpress cPLA 2 were incubated with ionomycin (Iono, 1 M) or buffer (Ctrl) for 24 h. Ionomycin increased bip, grp94, and erp72 mRNAs. Tunicamycin (Tunic, 10 g/ml, 24 h) was employed as a positive control in these experiments.
(release of LDH) in these clones, as compared with neo GEC (Table IV). Thus, inhibition of the ionomycin-induced increase in bip expression by bip antisense mRNA results in enhanced GEC injury. There was also a tendency toward greater cytolysis with 24-h tunicamycin incubation, but the difference was not statistically significant (Table IV). By light microscopy, bip antisense clones had normal morphology and proliferated at rates similar to neo GEC under standard culture conditions. The bip antisense clones and neo GEC were sensitized with antibody and incubated with serially increasing doses of complement that induced minimal to moderate cell lysis at 18 h. This protocol allows for C5b-9 to increase ER stress protein expression (Fig. 4), but with increasing incubation time and complement dose, a portion of the cells will undergo lysis. After 18 h of incubation, lysis (LDH release) was consistently greater in the GEC clones that express bip antisense mRNA (Fig. 6B). Thus, attenuation of the C5b-9-induced increase in bip expression by bip antisense mRNA results in enhanced cytolysis, indicating that induction of bip plays a functionally important role in limiting the amount of complement-dependent injury. In other experiments, neo GEC and the GEC that express bip antisense mRNA were incubated with antibody and serially increasing doses of complement for only 40 min. During brief incubation, there is insufficient time for C5b-9 to increase ER stress protein expression. After 40 min, complement lysis was not enhanced in the bip antisense clones and actually tended to be lower in these clones, as compared with neo (Table V). Thus, basal expression of ER stress proteins did not modulate complement cytotoxicity. Moreover, these experiments indicate that expression of bip antisense mRNA in GEC does not exacerbate complement-induced lysis nonspecifically.
In the next series of experiments, we assessed whether other FIG. 3. Effect of ionomycin on expression of ER stress proteins in GEC. GEC that overexpress cPLA 2 or neo GEC were incubated with ionomycin (Iono, 1 M) or buffer alone (control; Ctrl) for 24 h. Tunicamycin (Tunic, 10 g/ml, 24 h) was employed as a positive control. Cell lysates were immunoblotted with antibodies to bip, grp94, or erp72. A, representative immunoblot. B, densitometric quantification of immunoblots. The control incubations demonstrate that there is some basal expression of bip, grp94, and erp72 proteins in GEC. Ionomycin stimulated increases in grp94, bip, and erp72, and the effect of ionomycin was more prominent in the GEC that overexpress cPLA 2 . ϩ, p Ͻ 0.04 Iono versus Ctrl; *, p Ͻ 0.001 Iono versus Ctrl and p Ͻ 0.004 cPLA 2 versus neo (Iono-treated GEC); X, p ϭ 0.06 Iono versus Ctrl; ϩϩ, p Ͻ 0.001 Iono versus Ctrl and p Ͻ 0.03 cPLA 2 versus neo (Iono-treated GEC); **, p Ͻ 0.01 Iono versus Ctrl (5 experiments). Tunicamycin (positive control) increased ER stress proteins in both neo GEC and GEC that overexpress cPLA 2 .

FIG. 4. Effect of complement and cPLA 2 on expression of ER stress proteins in GEC.
Antibody-sensitized GEC that overexpress cPLA 2 or neo GEC were incubated with 2.5% normal serum (NS; to form sublytic C5b-9) or heat-inactivated serum (HIS) in controls for 4, 6, or 24 h. Cell lysates were immunoblotted with antibodies to bip or grp94. A, representative immunoblot. (On this gel, the grp94 band appeared to run as a doublet.) B, densitometric quantification of immunoblots. Complement increased expression of bip and grp94 significantly in GEC that overexpress cPLA 2 (nine experiments), whereas upward trends in bip and grp94 expression occurred in neo GEC (six experiments). bip, p Ͻ 0.002 NS versus HIS and p ϭ 0.05 cPLA 2 versus neo (at all time points); grp94, p Ͻ 0.001 NS versus HIS and p Ͻ 0.035 cPLA 2 versus neo (at all time points). The effect of tunicamycin (Tunic) is shown for comparison (positive control). C, assembly of C5b-9 is required for increased expression of ER stress proteins. Antibody-sensitized GEC that overexpress cPLA 2 were incubated with 2.5% C8-deficient serum (C8DS) or 2.5% C8-deficient serum reconstituted with purified C8 (C8DSϩC8) for 24 h. Cell lysates were immunoblotted with antibodies to bip or grp94. A representative immunoblot and densitometric quantification of immunoblots are presented. C8DSϩC8 increased expression of bip and grp94 significantly, as compared with C8DS. bip, p Ͻ 0.0001 C8DSϩC8 versus C8DS; grp94, p Ͻ 0.005 C8DSϩC8 (six incubations). D, antibody-sensitized neo GECs were incubated with 4.0% normal serum (NS; to form sublytic C5b-9), 4.0% normal serum plus MAFP (25 M), or heat-inactivated serum (HIS) in controls, for 24 h. A representative immunoblot and densitometric quantification of immunoblots are presented. In control GEC (C), complement increased bip and grp94 expression significantly (p Ͻ 0.01), whereas MAFP (M) inhibited the complementinduced increases (five experiments).
stimuli that induce ER stress protein expression would affect complement-mediated GEC injury. We observed that some stimuli provided protective effects, whereas others enhanced complement lysis. GEC in culture are particularly sensitive to the cytotoxic effect of puromycin aminonucleoside (36), and injection of this compound into rats may induce GEC injury and proteinuria in vivo (41). Incubation of cultured GEC with puromycin aminonucleoside increased expression of bip at 2 and 4 h, however, expression returned to basal levels at 24 h (Fig. 7A). In GEC that were preincubated with puromycin aminonucleoside for 3 h, complement-dependent cytolysis was reduced significantly (Fig. 7B). We then examined the effects of pretreatment with tunicamycin and ionomycin (Fig. 3). Unlike puromycin aminonucleoside, 24-h preincubation with tunicamycin (a potent inducer of ER stress proteins) enhanced complement-mediated lysis (Table VI). Preincubation with tunicamycin for only 6 h did not affect complement-mediated GEC injury significantly, but this shorter preincubation time did not increase expression of ER stress proteins consistently (data not shown). Ionomycin pretreatment tended to enhance complement-mediated injury, although the change was not statistically significant (Table VI). We also evaluated if exposure of GEC to an in vitro model of ischemia-reperfusion injury (chemical anoxia) would increase expression of ER stress proteins and protect from subsequent complement attack. Incubation of GEC with 2-deoxyglucose ϩ antimycin A for 90 min (to inhibit glycolytic and/or oxidative metabolism), followed by re-exposure to glucose-replete culture medium for 24 h resulted in increased expression of bip and grp94 (Fig. 8). This protocol did not protect, but rather enhanced, complement-mediated injury significantly (Table VI).

Expression of ER Stress Proteins Is Enhanced in GEC Injury in Vivo-
The above experiments demonstrated that C5b-9 increases bip and grp94 protein expression in cultured GEC, but it is important to determine if analogous changes occur in C5b-9-mediated GEC injury in vivo. To address this question, we assessed levels of bip and grp94 in PHN, where GEC injury is due to C5b-9 assembly, and is associated with cPLA 2 activation and production of prostanoids. In our model of PHN, proteinuria begins to appear at approximately day 7 and is wellestablished at days 13 and 14 (see below). On day 14, expression of glomerular bip and grp94 was increased in rats with PHN, as compared with control (Fig. 9, A and B). Increases in levels of bip and grp94 proteins were not detected consistently on days 3 and 7 (data not shown). By analogy to puromycin aminonucleoside, GEC in vivo are sensitive to the cytotoxic effects of adriamycin, and injection of rats with adriamycin may lead to GEC injury, in association with protein-  -1-3). B, complement-mediated cytotoxicity in GEC that express bip antisense mRNA. Neo GEC and three clones of GEC that express bip antisense mRNA were incubated with anti-GEC antibody (40 min), and normal serum (heat-inactivated serum in controls) for 18 h. Complement-mediated cytotoxicity was determined by measuring release of LDH into cell supernatants. Complement induced greater cytotoxicity in the bipAS-1 and bipAS-3 GEC (p Ͻ 0.015 bipAS-1 versus neo, p Ͻ 0.003 bipAS-3 versus neo), whereas an upward trend in cytotoxicity was evident in bipAS-2 (p ϭ 0.078 bipAS-2 versus neo). Values represent five experiments. There were no significant differences in background LDH release, i.e. in incubations with heat-inactivated serum (not shown).

TABLE IV Effect of ionomycin and tunicamycin on LDH release in GECs that
stably express bip antisense mRNA neo and bip antisense (bipAS) GEC were incubated with ionomycin, 1 M (four experiments), or tunicamycin, 10 g/ml (six experiments). LDH release (reflecting cell lysis) was determined after 24 h. uria (41). A second group of rats was injected with a subnephritogenic dose of adriamycin, i.e. a dose that did not induce proteinuria for up to 14 days. Glomeruli of these rats showed a significant increase in grp94 expression, whereas bip expression showed an upward trend (Fig. 9B). Thus, levels of ER stress proteins also increased when GEC were injured in vivo by a toxin-mediated mechanism. Moreover, the results with adriamycin suggest that development of proteinuria is not essential for enhanced ER stress protein expression in GEC. Finally, injection of rats with tunicamycin enhanced glomerular expression of bip and grp94 (Fig. 9C), without inducing proteinuria. The experiments presented in Figs. 6 and 7 show that ER stress proteins restrict complement-mediated cytolysis in cultured GEC. In the next series of experiments, we assessed whether stimuli shown to induce ER stress protein expression would limit C5b-9-mediated GEC injury (i.e. proteinuria) in vivo. Rats were treated with a subnephritogenic dose of adriamycin, or tunicamycin, prior to the induction of PHN. Proteinuria developed in untreated rats with PHN (on days 7, 9, and 13), whereas the amount of proteinuria was significantly lower in the rats with PHN that had been pretreated with adriamycin or tunicamycin (Fig. 10). Thus, increased ER stress protein expression can reduce C5b-9-mediated GEC injury in vivo. By immunofluorescence microscopy and/or immunoblotting, there were no apparent differences in the amounts of glomerular sheep antibody IgG, rat IgG, and rat C3 among the three groups of rats ( Fig. 11 and Table VII), indicating that adriamycin and tunicamycin did not decrease proteinuria by reducing the amount of glomerular antibody deposition. Serum creatinine was not significantly different among the three groups of rats, indicating that most likely there were no significant differences in renal function (Table VII). Urine volumes ranged from 5.6 to 22.5 ml/24 h, and there was no consistent pattern among groups (data not shown). In the PHN rats pretreated with adriamycin or tunicamycin, we did not detect further increases in glomerular levels of bip and grp94 (day 14), as compared with the PHN-untreated group (data not shown).

DISCUSSION
In this study, we show that cPLA 2 activation in GEC was associated with leakage of luminal ER stress proteins (bip and grp94) into the cytosol (Fig. 1), suggesting that cPLA 2 -induced phospholipid hydrolysis resulted in impairment of ER membrane integrity. These results are in keeping with our previous studies, which demonstrated that overexpression of cPLA 2 in GEC exacerbates complement-mediated cytotoxicity (16) and that complement-induced activation of cPLA 2 leads to phospho-TABLE V Complement-mediated cytolysis (acute incubation) in GECs that express bip antisense mRNA neo GEC and three clones of GECs that express bip antisense mRNA were incubated with anti-GEC antibody (40 min) and normal serum (or heat-inactivated serum in controls) for 40 min. Complement-mediated cytolysis was determined by measuring release of LDH into cell supernatants. There were no significant increases in cytolysis in the bip antisense clones, as compared with neo. Values represent 6 -10 experiments.
Normal serum Specific LDH release

FIG. 7. Effects of puromycin aminonucleoside on bip expression (A) and complement-mediated GEC injury (B).
A, GEC that overexpress cPLA 2 were incubated with puromycin aminonucleoside (PA; 50 g/ml) for 2, 4, or 24 h. An increase in bip expression is evident at 2 and 4 h (representative immunoblot). The effect of tunicamycin is shown for comparison. B, GEC were preincubated without (Ctrl) and with puromycin aminonucleoside for 3 h (to induce bip expression) and were then incubated with antibody and normal serum (heat-inactivated serum in controls). LDH release (reflecting cell lysis) was determined after 40 min. Pretreatment with puromycin aminonucleoside reduced complement-dependent lysis significantly (p Ͻ 0.002 PA versus Ctrl, five experiments).

TABLE VI
Effects of tunicamycin, ionomycin, and chemical anoxia on complement-mediated GEC injury GECs that overexpress cPLA 2 were preincubated with tunicamycin (10 g/ml, 24 h), ionomycin (1 M, 24 h), or deoxyglucose (DG) ϩ antimycin A (as in Fig. 8) to induce expression of ER stress proteins. Then, GECs were incubated with antibody, followed by 5, 10, or 15% normal serum (or heat-inactivated serum in controls). LDH release (reflecting cell lysis) was determined after 40 min. Pretreatment with tunicamycin or prior exposure to chemical anoxia enhanced complementdependent lysis significantly (p Ͻ 0.01 tunicamycin versus untreated, four experiments; p Ͻ 0.04 DG ϩ antimycin A versus untreated, three experiments). A small upward trend in complement-dependent lysis was evident with ionomycin pretreatment (five experiments).  8. Effect of chemical anoxia on ER stress protein expression (representative immunoblot). GEC were incubated with glucose-free measurement buffer for 40 min at 37°C. Then, GEC were incubated with 10 mM 2-deoxyglucose (DG), 10 mM 2-deoxyglucose ϩ 10 M antimycin A, or buffer alone (Ctrl). Supernatants were removed after 90 min at 37°C, and GEC were placed into culture medium for 24 h. Lysates were immunoblotted with antibodies to bip or grp94. Deoxyglucose with or without antimycin A induced an increase in bip. There was a smaller increase in grp94 with deoxyglucose ϩ antimycin A. The effect of tunicamycin is shown for comparison. lipid hydrolysis at the membrane of the ER (29). Furthermore, we show that inhibition of endogenous cPLA 2 can attenuate complement-mediated cytotoxicity in GEC (Table III). The mechanism of bip and grp94 leakage from the ER may have involved dysregulation of ER protein pores (42), but it will require further delineation. Another consideration is that complement-induced cPLA 2 activation was associated with increased retrograde translocation of abnormal proteins from the ER (the abnormal proteins being coupled with bip or grp94) (42), although it is less likely that retrograde translocation could have occurred in an acute time frame. Impairment of ER membrane integrity could potentially be injurious by causing leakage of Ca 2ϩ and other ER luminal components as well as impairment in ER Ca 2ϩ uptake. Although association of cPLA 2 with membranes of cell organelles has been established in other cell types (27), there is little information on potential damage to organelle membranes due to phospholipid hydrolysis. In addition to the present study, overexpression of cPLA 2 in LLCPK 1 kidney epithelial cells was associated with disruption of the Golgi (43).
Chronic incubation of GEC with complement (4 -24 h) enhanced expression of bip and grp94 mRNAs and proteins (Figs. 2 and 4), although there was no detectable increase in the cytosolic stress protein, Hsp70. The increases in ER stress proteins were dependent on C5b-9 assembly (Fig. 4C). At lower doses, complement-induced increases in bip and grp94 expression were seen mainly in the GEC that overexpress cPLA 2 , but at a higher dose, expression was also induced in neo GEC, and the increases were blocked with MAFP (Fig. 4). Thus, increases in bip and grp94 were, at least in part, dependent on the activation of cPLA 2 . Further support for the role of cPLA 2 in ER stress protein induction was provided by experiments in which GECs were incubated chronically with the Ca 2ϩ ionophore, ionomycin. bip and grp94 protein expression was enhanced by ionomycin, and the changes were significantly greater in the GEC that overexpress cPLA 2 (Fig. 3).
Increases in ER stress proteins were not limited to cultured GEC but also occurred in vivo. Using the PHN model of C5b-9-induced GEC injury, it was demonstrated that glomerular bip and grp94 proteins were up-regulated in proteinuric rats with PHN (day 14), as compared with normal controls (Fig. 9). Rats with PHN show increased glomerular activity of cPLA 2 and AA metabolites (19,20). At present, there are no specific inhibitors of cPLA 2 that can be used in experimental animals (44). Thus, although bip and grp94 up-regulation in PHN is associated with the activation of cPLA 2 , proof for the functional importance of cPLA 2 in vivo will require development of suitable cPLA 2 inhibitors. Moreover, we have not been successful in establishing a proteinuric model of PHN in mice, and, consequently, cPLA 2 -null mice (45) could not be used to address the functional role of cPLA 2 .
The ER serves as a site for folding, assembly, and degradation of proteins (30 -32). Membrane and secreted proteins are translocated into the lumen of the ER shortly after initiation of synthesis, and resident ER luminal proteins, including bip and grp94, mediate protein folding. Moreover, bip and grp94 are believed to bind to misfolded or abnormal proteins and prevent their aggregation, either by rescuing such proteins from irreversible damage, or by increasing their susceptibility to proteolytic attack. Other ER proteins, including erp72, may participate in disulfide isomerization. Perturbation of the ER by stimuli, including accumulation of mutant proteins or Ca 2ϩ depletion, may increase expression of ER stress proteins. In yeast, misfolded proteins in the ER lead to activation of Ire1p endonuclease (31,32). Ire1p splices HAC1 (homologous to activating transcription factor (ATF) and CREB) mRNA in the FIG. 9. Expression of ER stress proteins in vivo. Glomeruli were isolated from normal rats (control) and from rats with PHN on day 14. Glomerular lysates were immunoblotted with antibodies to bip or grp94 (A, representative immunoblots; B, densitometric quantification). bip and grp94 expression was increased in glomeruli isolated from rats with PHN, as compared with control (*, p Ͻ 0.002; ϩ, p Ͻ 0.005 PHN versus control; 7-9 rats per group). B also shows the densitometric quantification of grp94 and bip expression in rats injected with subnephritogenic adriamycin, 6 mg/kg intravenously (ADR, day 14, **, p Ͻ 0.001 adriamycin versus control; 9 -12 rats per group), as well as in rats injected with tunicamycin, 1 mg/kg intraperitoneally (Tun, 24 h, the dots represent the values of two individual rats in each group).
FIG. 11. Glomerular deposition of sheep (Sh) IgG, rat IgG, and rat C3 in rats with PHN that were pretreated with adriamycin or tunicamycin, or were untreated. Kidney sections were stained with fluorescein-conjugated antibodies and visualized using immunofluorescence microscopy (ϫ250 magnification). nucleus, which is then religated, exits the nucleus, and is translated into Hac1p transcription factor. Hac1p translocates to the nucleus, binds to the unfolded response element, and induces expression of ER stress proteins. The mammalian unfolded protein response, although analogous to yeast, is more complex and appears to involve additional pathways and effectors (46).
C5b-9-induced sublethal cell injury may lead to a decline in cellular ATP, mitochondrial lipid perturbation, or loss of mitochondrial membrane potential (47)(48)(49), whereas at high doses, C5b-9 can induce mitochondrial damage and cell necrosis (50). Based on biochemical and morphologic observations (36,49), it is likely that, during complement-dependent GEC injury, integral membrane and secretory proteins are altered. Such proteins may include integrins, transporters, and/or cell junctional proteins, and these alterations may contribute to the permselectivity defect of the glomerular capillary wall in PHN. Besides inducing cell injury, C5b-9 assembly leads to activation of mechanisms that restrict injury or facilitate recovery of cells from complement attack. One such mechanism of protection is "ectocytosis" (shedding) of C5b-9 complexes from cell membranes (1,2). The present study identifies another mechanism that limits complement attack (Figs. 6, 8, and 10). Expression of bip antisense mRNA in GEC reduced stimulated bip protein expression, and when these GEC were incubated with complement for 18 h, cytolysis was enhanced, as compared with neo GEC (Fig. 6). Thus, induction of ER stress proteins is an important novel mechanism of protection from sustained complement attack. The capability of the GEC to recover or limit the severity of complement attack may depend on its capacity to re-synthesize or reassemble integral membrane proteins, which may require the presence of bip or grp94. Induction of these ER stress proteins during complement attack may also limit accumulation of abnormal proteins and help sustain physiological functions and viability (31,32).
Exposure of cells to mild stress, sufficient to induce upregulation of ER stress proteins, may be protective to additional insults (33,34), although progression to cell death may occur if the stress is more intense or prolonged (35). The present study demonstrates that preincubation of cultured GEC with a sublethal dose of puromycin aminonucleoside increased bip expression and protected GEC from complement attack (Fig. 7). On the other hand, prior exposure of cultured GEC to ischemia-reperfusion injury (chemical anoxia), despite increasing ER stress proteins (Fig. 8), exacerbated complementmediated cytotoxicity (Table VI). However, cellular effects of ischemia and reperfusion are complex and are manifested by a variety of changes that include a decline in cellular ATP levels, alterations of cellular redox state and perturbations in intracellular Ca 2ϩ homeostasis (34). In earlier studies, pretreatment with tunicamycin protected renal tubular epithelial cells from chemical anoxia or toxin-induced injury (33,34). In con-trast, ionomycin and tunicamycin, despite increasing ER stress proteins, either had no effect or exacerbated complementmediated cytotoxicity in cultured GEC (Table VI). These compounds may have induced multiple alterations in GEC leading to substantial injury, and with subsequent exposure to complement, the cells could not withstand the additional stress. However, it is not clear why the effect of tunicamycin was cytotoxic in culture, but protective in PHN. This question will require further study.
The functional role of ER stress proteins in cultured GEC was applicable to GEC injury in vivo. Induction of ER stress proteins by pretreatment with a subnephritogenic dose of adriamycin or tunicamycin significantly reduced proteinuria in PHN (Fig. 10), independently of changes in glomerular immunoglobulin or complement deposition ( Fig. 11 and Table VII). It should be noted that, although tunicamycin was reported to induce a renal lesion resembling acute tubular necrosis in mice (51), the PHN rats that had been pretreated with tunicamycin had no significant alterations in renal function (Table VII). Our observations provide a rationale for developing non-toxic methods to induce expression of ER stress protein in vivo, which may eventually have applications for therapy of glomerular disease. The importance of these results extends beyond complement-induced GEC injury. For example, preinduction of ER stress proteins prior to xenotransplantation may potentially be a means of reducing hyperacute xenograft rejection, which is complement-dependent (52).

TABLE VII
Glomerular antibody and C3 deposition and serum creatinine measurements in PHN Rats with PHN were untreated (4 rats) or were pretreated with adriamycin (ADR; 3 rats) or tunicamycin (Tun; 3 rats). All measurements were performed on day 14 (see Fig. 10). Glomeruli were isolated, and the amounts of sheep IgG and rat IgG were assessed by immunoblotting and quantified by densitometry. Values represent densitometry in arbitrary units. Kidney sections were stained with fluorescein-conjugated specific antibodies, and the amounts of glomerular sheep IgG, rat IgG, and rat C3 were assessed by immunofluorescence microscopy (IF) and quantified (see "Methods"). Values represent times required to collect images by the photomultiplier and camera (seconds) and are inversely proportional to fluorescence intensity. For all parameters, there are no significant differences among groups.