C1q Protects Against the

Cutting Edge: C1q Protects Against the Development of Glomerulonephritis Independently of C3 Activation This information is current as of June 13, 2013.

Daniel A. Mitchell, Philip R. Taylor, H. Terence Cook, Jill Moss, Anne E. Bygrave, Mark J. Walport and Marina Botto J Immunol 1999; 162:5676-5679; ; http://www.jimmunol.org/content/162/10/5676

References Subscriptions

Email Alerts

Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscriptions Submit copyright permission requests at: http://www.aai.org/ji/copyright.html Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/cgi/alerts/etoc

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 9650 Rockville Pike, Bethesda, MD 20814-3994. Copyright © 1999 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606.

Downloaded from http://www.jimmunol.org/ by guest on June 13, 2013

Permissions

This article cites 23 articles, 12 of which you can access for free at: http://www.jimmunol.org/content/162/10/5676.full#ref-list-1

●

Cutting Edge: C1q Protects Against the Development of Glomerulonephritis Independently of C3 Activation1 Daniel A. Mitchell,2* Philip R. Taylor,2* H. Terence Cook,† Jill Moss,‡ Anne E. Bygrave,* Mark J. Walport,* and Marina Botto3*

T

he role of complement in the development and expression of autoimmunity in humans appears to be paradoxical. Complement components are present at the sites of tissue injury in inflammatory lesions, including glomerulonephritis (GN);4 however, inherited homozygous deficiencies of early classical complement pathway proteins, especially C1q and C4, are strongly associated with the development of systemic lupus erythematosus (SLE) (1). The recent findings that C1q can bind specifically to the surface blebs of apoptotic keratinocytes (2) and that *Rheumatology Section and †Department of Histopathology, Hammersmith Campus, Imperial College School of Medicine, London, United Kingdom; and ‡Department of Histopathology, Charing Cross Campus, Imperial College School of Medicine, London, United Kingdom Received for publication February 5, 1999. Accepted for publication March 18, 1999. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 This work was supported by the Arthritis Research Campaign (ARC) and the Wellcome Trust. D.A.M. is the recipient of an ARC Ph.D. studentship. 2

These authors contributed equally to this paper.

3

Address correspondence and reprint requests to Dr. Marina Botto, Rheumatology Section, Division of Medicine, Imperial College School of Medicine, Hammersmith Campus, Du Cane Road, London W120NN, U.K. E-mail address: [email protected] 4 Abbreviations used in this paper: GN, glomerulonephritis; SLE, systemic lupus erythematosus; FcR, Fc receptor; C1qa2/2, C1q-deficient; H2-Bf/C22/2, factor B- and C2deficient; ANA, antinuclear Ab(s); AEU, arbitrary ELISA unit(s); wt, wild-type.

Copyright © 1999 by The American Association of Immunologists

●

the common autoantigens targeted in SLE can be found in high concentrations on the surface of cells which have undergone apoptosis (3) have led to the new hypothesis that C1q deficiency may cause SLE as a result of an impaired clearance of apoptotic cells (2, 4). In addition, immunization with apoptotic cells has been shown to stimulate autoantibody production (5), suggesting that defects in the pathways by which apoptotic cells are processed could play an important role in driving an autoimmune response. Recently, there has been considerable debate surrounding the role of complement in the induction and maintenance of inflammation, with growing evidence stressing the important role of Fc receptors (FcRs) in the mediation of the inflammatory responses triggered by immune complexes. Experiments involving the reverse passive Arthus reaction in the skin suggest a dominant role for FcRs, as opposed to complement activation, in initiating inflammation (6 –9), although, in a model of reverse passive Arthus reaction in the lungs, there was evidence that complement played a key role in the initiation of the inflammatory injury (10). FcR g-chain deficiency has been shown to confer marked protection against renal damage both in the spontaneous (NZB 3 NZW)F1 model of autoimmunity (11) and in a nephrotoxic serum GN model in C57BL/6 mice (12). In these experiments, the mice were protected from severe GN, whereas the glomerular deposition of IgG and C3 remained unaffected, providing strong evidence for a predominant role for FcRs in driving inflammation in autoimmune nephritis. Gene-targeted homozygous C1q-deficient (C1qa2/2) mice have been shown to develop a spontaneous autoimmune disorder with high titers of antinuclear Abs (ANA) and GN that was associated with renal IgG and C3 deposition (4). A striking feature of the GN in C1qa2/2 animals was the presence of increased numbers of apoptotic bodies in the glomeruli, a phenomenon also observed in the kidneys of C1qa2/2 animals without histological evidence of GN. These observations supported the hypothesis that C1q may protect against autoimmunity by serving as an opsonin in the efficient recognition and physiological clearance of apoptotic cells; however, these findings did not fully resolve the question surrounding the importance of complement activation in the development of the spontaneous GN. To address this question, we crossed the C1qa2/2 strain with gene-targeted factor B/C2-deficient (H2-Bf/C22/2) mice (13), generating mouse strains lacking both the classical and alternative pathways of complement activation in the presence or absence of C1q. These cohorts of mice were sacrificed after 8 mo and analyzed for the presence of autoantibodies and GN. Here, we demonstrate that C1qa2/2 mice that also 0022-1767/99/$02.00


C1q-deficient (C1qa2/2) mice develop antinuclear Abs and glomerulonephritis (GN) characterized by multiple apoptotic bodies. To explore the contribution of C3 activation to the induction of spontaneous GN, C1qa2/2 mice were crossed with factor B- and C2-deficient (H2-Bf/C22/2) mice. GN was present in 64% of the 45 C1qa/H2-Bf/C22/2 mice compared with 8% of the 65 H2-Bf/ C22/2 mice and none of the 24 wild-type controls. IgG was detected in the glomeruli of diseased C1qa/H2-Bf/C22/2 kidneys. However, glomerular staining for C3 was absent. Increased numbers of glomerular apoptotic bodies were detected in undiseased C1qa/H2-Bf/C22/2 kidneys. These findings support the hypothesis that C1q may play a role in the clearance of apoptotic cells without the necessity for C3 activation and demonstrate that the activation of C3 is not essential for the development of GN in this spontaneous model of lupus-like disease. The Journal of Immunology, 1999, 162: 5676 –5679.

The Journal of Immunology

5677

lack C2 and factor B develop GN without glomerular C3 deposition. Mice lacking C2 and factor B did not develop either GN or autoantibodies, showing a role for C1q alone or in conjunction with C4 in the protection against the development of autoimmunity.

Materials and Methods Mice C1qa2/2 and H2-Bf/C22/2 mice were generated as described previously (4, 13). All mice were bred in a mixed genetic background (129/Sv 3 C57BL/6) and kept in specific pathogen-free conditions but not in a germfree environment. The C1qa2/2 and H2-Bf/C22/2 mice were crossed to generate a C1qa/H2-Bf/C22/2 strain that was deficient in all three complement components. Animal care and procedures were conducted according to institutional guidelines.

Autoantibody assays and serum biochemistry

Histology Kidney portions were fixed in Bouin’s solution for 4 h, transferred into 70% ethanol, and processed into paraffin. The sections were stained with hematoxylin and eosin and scored for GN as described previously (17). Glomerular hypercellularity was graded on a scale of 0-IV; grade 0 represents no involvement, and grade IV represents severe proliferative GN in .90% of glomeruli. For electron microscopy, kidneys were fixed in 4% glutaraldehyde, postfixed in 1% osmium tetroxide, and embedded in Spurr’s resin. Fluorescence microscopy was conducted on snap-frozen sections incubated with FITC-conjugated polyclonal Abs to mouse IgG (Sigma, Poole, U.K.) and mouse C3 (Cappel/ICN Biomedicals, Aurora, OH). For C4 staining, a monoclonal rat anti-mouse C4 (Cedarlane, Ontario, Canada) and a monoclonal FITC-labeled mouse anti-rat IgG secondary Ab (Sigma) were used. Apoptotic bodies were quantified by light microscopy on coded sections. A cell was considered apoptotic when it showed loss of cell volume, chromatin condensation along the nuclear membrane with intensely basophilic staining, and/or nuclear fragmentation into spherical structures containing condensed chromatin.

Statistics Statistics were calculated using GraphPad Prism version 2.0 (GraphPad Software, San Diego, CA). Nonparametric statistical tests were applied throughout.

FIGURE 1. Autoantibody profiles. ANA levels in 45 C1qa/H2-Bf/ C22/2, 65 H2-Bf/C22/2, and 24 wt animals at 8 mo of age. Females and males are shown separately. Serum samples were titrated to endpoint. Small circles represent one mouse; large circles represent a variable number of animals as indicated below. E, animals with evidence of GN upon histological examination; F, absence of renal involvement.

Renal histology Histological examination showed GN in 29 of 45 (64%) of the C1qa/H2-Bf/C22/2 mice at 8 mo compared with only 5 of 65 (8%) of the H2-Bf/C22/2 mice and none of the 24 wt mice (x2 5 55.17, p , 0.0001) (Table I). In the C1qa/H2-Bf/C22/2 group, GN was observed predominantly in females (87% compared with 41% of the males). The severity of GN and levels of ANA did show a significant correlation, although the correlation was weak (Spearman correlation: p 5 0.0443, r 5 0.3013). Morphologically, the GN consisted of glomerular hypercellularity with increased numbers of cells in mesangial areas and capillary lumens (Fig. 2A). Renal functional analysis showed no differences in serum creatinine (nephritic kidneys: 37.30 6 3.20 mmol/l (mean 6 SEM); non-nephritic kidneys: 40.00 6 2.89) and in serum albumin (nephritic kidneys: 24.67 6 2.60 g/l; non-nephritic kidneys: 23.67 6 1.30). Immunostaining revealed the presence of prominent deposits of IgG in the glomeruli of C1qa/H2-Bf/C22/2 mice assessed as positive for GN. The staining was mostly mesangial, with some on the capillary wall. The wt mice with no histological evidence of GN had only weak focal staining in some glomeruli, a pattern also seen in the H2-Bf/C22/2 mice. Kidneys from C1qa2/2 mice showed staining that was similar to that observed in the C1qa/H2-Bf/C22/2 mice (Fig. 2B).

Results Autoantibody analysis Three cohorts of mice, consisting of 45 C1qa/H2-Bf/C22/2, 65 H2-Bf/C22/2, and 24 wild-type (wt) animals, were analyzed for the presence of autoantibodies at 3, 5, and 8 mo of age. IgG ANA were detected in 20% of the C1qa/H2-Bf/C22/2 mice at 5 mo, increasing to 40% at 8 mo (range 1:80 –1:1280). In comparison, low levels of ANA were detected in only 4% of the wt mice (titer 1:80) and in 1% of the H2-Bf/C22/2 mice (titer 1:80) at 8 mo of age (Kruskal-Wallis test, p , 0.0001) (Fig. 1). At 8 mo of age, Abs to ssDNA were detected in 17% of the C1qa/H2-Bf/C22/2 mice (range 14.6 –150 AEU) compared with only one of the wt mice (12.9 AEU) and none of the H2-Bf/C22/2 mice (Kruskal-Wallis test, p 5 0.0026). Abs to dsDNA were detected in only two C1qa/ H2-Bf/C22/2 animals.

Table I. Histological assessment of kidney sectionsa GN Strain

wt (1/1) 2/2

H2-Bf/C2

C1qa/H2-Bf/C22/2

Sex (n)

0

I

II

III

IV

? (11) / (13) ? (29) / (36) ? (22) / (23)

11 13 28 32 13 3

0 0 1 3 4 8

0 0 0 1 3 6

0 0 0 0 2 5

0 0 0 0 0 1

a All of the experimental animals were sacrificed at 8 mo of age to obtain agematched autopsy specimens. Bouin’s fixed kidney sections were scored for GN. Glomerular hypercellularity was graded on a 0 –IV scale, where 0 5 no involvement and IV represents severe proliferative GN in .90% of glomeruli.


Mice were bled at 3, 5, and 8 mo of age; at 8 mo, all of the animals were sacrificed. The serum was stored at 270°C before analysis. Levels of IgG ANA were sought by indirect immunofluorescence using Hep-2 cells (14). Anti-dsDNA Abs were detected by indirect immunofluorescence on Crithidia luciliae (15). Serum samples were screened at a 1/80 (ANA) or 1/20 (anti-dsDNA) dilution, and the positive samples were titrated to endpoint. Abs to ssDNA (calf thymus) were measured by ELISA as described previously (16). Samples were screened at a 1/50 dilution, and the results were expressed in arbitrary ELISA units (AEU) relative to a standard positive sample (derived from an MRL/Mp.lpr/lpr mouse) that was assigned a value of 100; samples were scored as positive at $7.0 U (3 SD above the lower limit of detection). Serum creatinine and serum albumin were measured by an autoanalyzer using standard methods.

5678

CUTTING EDGE

In the kidneys of wt mice, there was strong peritubular staining for C3, with very weak staining in mesangial areas. Similar weak mesangial C3 staining was seen in the undiseased kidneys of H2Bf/C22/2 mice and in the nephritic kidneys of C1qa/H2-Bf/C22/2 animals. No peritubular staining was seen in either of these groups. Kidneys from nephritic C1qa2/2 mice used as positive controls showed extensive mesangial C3 deposition (Fig. 2C). Immunostaining showed C4 in the mesangium of all of the experimental groups of mice in similar quantity and distribution (data not shown). There was no enhancement of the mesangial staining of C4 in the C1qa/H2-Bf/C22/2 animals with nephritis compared with any of the other groups of animals, including the healthy controls. Electron microscopy in selected cases showed expansion of mesangial areas with multiple electron-dense deposits. Some capillary loops showed subendothelial deposits with formation of a new layer of basement membrane on the luminal side of the deposits and mesangial cell interposition (Fig. 3). Cells with the morphology of macrophages were present in capillary lumens. As we have reported previously, the most striking histological feature of the GN in the C1qa2/2 mice was the presence of multiple apoptotic bodies in glomeruli (4). Assessment of the kidneys of C1qa/H2-Bf/C22/2, H2-Bf/C22/2, and wt mice with no histological evidence of GN revealed increased numbers of apoptotic bodies in the glomeruli of the C1qa/H2-Bf/C22/2 animals compared with the other two groups (wt (n 5 23): 0.174 1 0.388 (mean of apoptotic bodies in 50 glomeruli 6 SEM); H2-Bf/C22/2 (n 5 56): 0.089 6 0.288; C1qa/H2-Bf/C22/2 (n 5 16): 0.875 6 0.272; Kruskal-Wallis test, p , 0.0013).

Discussion We have reported recently that C1qa2/2 mice develop autoimmunity characterized by the production of ANA and immune com-

plex-mediated GN associated with the presence of increased numbers of apoptotic bodies (4). IgG and C3 were present in the glomeruli of the diseased kidneys, which suggested that complement was being activated, most likely by the alternative pathway. This possibility led to the question of whether the development of glomerular injury in this model was dependent upon the activation of C3 in glomeruli by the alternative pathway. To test this hypothesis, we crossed mice deficient in C1q with mice deficient in factor B and C2. Mice deficient in complement activation by disruption of the C2 and factor B genes did not develop spontaneous autoimmunity. When deficiency of C1q was added, renal damage and autoantibody production developed, suggesting a discrete role for the first component of the classical pathway, and possibly C4, in protection from autoimmunity. A striking feature of the glomeruli of aged, undiseased C1qa2/2 mice was the presence of elevated numbers of apoptotic bodies (4). This observation coupled with the knowledge that 1) C1q can directly bind to apoptotic cells (2), 2) the surface blebs of apoptotic cells express the common autoantigens of SLE (3), and 3) immunization with apoptotic cells can stimulate autoantibody production (5), has led to the hypothesis that C1q may be involved in the clearance of apoptotic cells, and that this activity may protect from the development of autoimmunity. An elevated number of apoptotic bodies were also present in the undiseased kidneys of the C1qa/H2-Bf/C22/2 mice but not in the H2Bf/C22/2 animals, suggesting that the proposed defect in the clearance of apoptotic cells in C1qa2/2 animals did not require C3 activation. Recent studies using both spontaneous and induced models of GN have suggested a dominant role for FcRs in the generation of immune complex-mediated renal damage. Deficiency of the FcR complex g-chain backcrossed onto the autoimmune-prone (NZB 3 NZW)F1


FIGURE 2. Renal histology. Light microscopy (row A) and immunofluorescence for mouse IgG (row B) and mouse C3 (row C) from a H2-Bf/C22/2 mouse (column 1), a C1qa/H2-Bf/C22/2 mouse (column 2), and a C1qa2/2 mouse (column 3). By light microscopy, the H2-Bf/C22/2 mouse shows normal glomerular morphology; there is faint glomerular localization of IgG and C3. The C1qa/ H2-Bf/C22/2 mouse shows enlarged glomeruli with marked hypercellularity. There is strong staining for IgG that is mainly mesangial. C3 staining is similar to that seen in the non-nephritic H2-Bf/C22/2 mouse. The C1qa2/2 mouse shows hypercellular GN with prominent mesangial staining for both IgG and C3 (all at 3340 magnification).

The Journal of Immunology

5679 In conclusion, the data described in this study of complementdeficient mice, when considered in association with recent studies on the role of FcRs in GN, support the hypothesis that the early components of the classical pathway protect from the development of nephritis. If this mechanism is impaired, as in C1q deficiency, renal inflammation may proceed, with FcRs as the dominant transducers of IgG-mediated injury.

Acknowledgments We thank the staff of our animal facility for the care of the animals used in this work.

References

background resulted in protection from the development of GN (11) without affecting the production of autoantibodies. An alternative approach to the same question involved the use of the nephrotoxic serum GN model in FcR g-chain deficient mice and also showed a dramatic attenuation of the inflammation (12). However, complement depletion in induced models of GN has been shown to attenuate the disease process (18, 19), and blockade of C5 in the spontaneous (NZB 3 NZW)F1 model of autoimmunity was also shown to ameliorate renal disease (20). Our initial studies in C1qa2/2 mice confirmed a protective role for C1q in the development of GN (4). Detection of glomerular C3 deposition in the diseased kidneys of C1qa2/2 animals indicated that complement was activated, but its relevance to the pathogenesis of the GN was unknown. The findings reported here, which showed GN with no C3 deposition in the C1qa/H2-Bf/C22/2 animals compared with the other two groups, would indicate that the development of glomerular damage associated with C1q deficiency occurs predominantly via a complement-independent, and perhaps FcR-mediated, mechanism. In addition, the presence of peritubular C3 staining in the kidneys of the wt mice, but not in kidneys of the H2-Bf/C22/2 mice or in the diseased kidneys of the C1qa/H2-Bf/ C22/2 mice, suggests that this staining may reflect deposited C3. In contrast, the low level of mesangial staining for C3 and C4 in all groups of mice is most likely the product of local synthesis, supporting a role for glomerular cells in the local production of complement components (21–23).


FIGURE 3. Electron microscopy. Electron micrographs of a glomerulus from a C1qa/H2-Bf/C22/2 mouse with GN are shown. The upper panel shows prominent mesangial electron dense deposits (37600 magnification). The lower panel is a higher power view of a capillary loop showing focal subendothelial electron dense deposits (arrows) and mesangial cell interposition (311,000 magnification).

1. Morgan, B. P., and M. J. Walport. 1991. Complement deficiency and disease. Immunol. Today 12:301. 2. Korb, L. C., and J. M. Ahearn. 1997. C1q binds directly and specifically to the surface blebs of apoptotic human keratinocytes: complement deficiency and systemic lupus erythematosus revisited. J. Immunol. 158:4525. 3. Casciola-Rosen, L. A., G. Anhalt, and A. Rosen. 1994. Autoantigens targeted in systemic lupus erythematosus are clustered in two populations of surface structures on apoptotic keratinocytes. J. Exp. Med. 179:1317. 4. Botto, M., C. Dell’Agnolla, A. E. Bygrave, E. M. Thompson, H. T. Cook, F. Petry, M. Loos, P. P. Pandolfi, and M. J. Walport. 1998. Homozygous C1q deficiency causes glomerulonephritis associated with multiple apoptotic bodies. Nat. Genet. 19:56. 5. Mevorach, D., J. L. Zhou, X. Song, and K. B. Elkon. 1998. Systemic exposure to irradiated apoptotic cells induces autoantibody production. J. Exp. Med. 188:387. 6. Sylvestre, D. L., and J. V. Ravetch. 1994. Fc receptors initiate the Arthus reaction: redefining the inflammatory cascade. Science 265:1095. 7. Hazenbos, W. L., J. E. Gessner, F. M. Hofhuis, H. Kuipers, D. Meyer, I. A. Heijnen, R. E. Schmidt, M. Sandor, P. J. Capel, M. Daeron, J. G. van de Winkel, and J. S. Verbeek. 1996. Impaired IgG-dependent anaphylaxis and Arthus reaction in FcgRIII (CD16) deficient mice. Immunity 5:181. 8. Sylvestre, D. L., and J. V. Ravetch. 1996. A dominant role for mast cell Fc receptors in the Arthus reaction. Immunity 5:387. 9. Sylvestre, D. L., R. Clynes, M. Ma, H. Warren, M. C. Carroll, and J. V. Ravetch. 1996 Immunoglobulin G-mediated inflammatory responses develop normally in complement-deficient mice. J. Exp. Med. 184:2385. 10. Hopken, U. E., B. Lu, N. P. Gerard, and C. Gerard. 1997. Impaired inflammatory responses in the reverse Arthus reaction through genetic deletion of the C5a receptor. J. Exp. Med. 186:749. 11. Clynes, R., C. Dumitru, and J. V. Ravetch. 1998. Uncoupling of immune complex formation and kidney damage in autoimmune glomerulonephritis. Science 279:1052. 12. Park, S. Y., S. Ueda, O. Hiroshi, Y. Hamano, M. Tanaka, T. Shiratori, T. Yamazaki, H. Arase, N. Arase, A. Karasawa, et al. 1998. Resistance of Fc receptor-deficient mice to fatal glomerulonephritis. J. Clin. Invest. 102:1229. 13. Taylor, P. R., J. T. Nash, E. Theodoridis, A. E. Bygrave, M. J. Walport, and M. Botto. 1998. A targeted disruption of the murine complement factor B gene resulting in loss of expression of three genes in close proximity, factor B, C2, and D17H6S45. J. Biol. Chem. 273:1699. 14. Molden, D. P., R. M. Nakamura, and E. M. Tan. 1984. Standardization of the immunofluorescence test for autoantibody to nuclear antigens (ANA): use of reference sera of defined antibody specificity. Am. J. Clin. Pathol. 82:57. 15. Isenberg, D. A., C. Dudeney, W. Williams, I. Addison, S. Charles, J. Clarke, and A. Todd-Pokropek. 1987. Measurement of anti-DNA antibodies: a reappraisal using five different methods. Ann. Rheum. Dis. 46:448. 16. Burlingame, R. W., and R. L. Rubin. 1990. Subnucleosome structures as substrates in enzyme-linked immunosorbent assays. J. Immunol. Methods 134:187. 17. Jevnikar, A. M., G. G. Singer, D. C. Brennan, H. W. Wu, V. R. Kelley. 1992. Dexamethasone prevents autoimmune nephritis and reduces renal expression of Ia but not costimulatory signals. Am. J. Pathol. 141:743. 18. Unanue, E., and F. J. Dixon. 1964. Experimental glomerulonephritis: participation of complement in nephrotoxic nephritis. J. Exp. Med. 119:965. 19. Couser, W. G. 1990. Mediation of immune glomerular injury. J. Am. Soc. Nephrol. 1:13. 20. Wang, Y., Q. Hu, J. A. Madri, S. A. Rollins, A. Chodera, and L. A. Matis. 1996. Amelioration of lupus-like autoimmune disease in NZB/W F1 mice after treatment with a blocking monoclonal antibody specific for complement component C5. Proc. Natl. Acad. Sci. USA 93:8563. 21. Sacks, S. H., W. Zhou, A. Pani, R. D. Campbell, and J. Martin. 1993. Complement C3 gene expression and regulation in human glomerular epithelial cells. Immunology 79:348. 22. Sacks, S. H., W. Zhou, R. D. Campbell, and J. Martin. 1993. C3 and C4 gene expression and interferon-g-mediated regulation in human glomerular mesangial cells. Clin. Exp. Immunol. 93:411. 23. Sheerin, N. S., W. Zhou, S. Adler, and S. H. Sacks. 1997. TNF-a regulation of C3 gene expression and protein biosynthesis in rat glomerular endothelial cells. Kidney Int. 51:703.