Renal Physiology

Trif is not required for immune complex glomerulonephritis: dying cells activate mesangial cells via Tlr2/Myd88 rather than Tlr3/Trif

Julia Lichtnekert, Volker Vielhauer, Daniel Zecher, Onkar P. Kulkarni, Sebastian Clauss, Stephan Segerer, Veit Hornung, Tanya N. Mayadas, Bruce Beutler, Shizuo Akira, Hans-Joachim Anders


Viral RNA or bacterial products can activate glomerular mesangial cells via a subset of Toll-like receptors (Tlr). Because Tlr2-deficient mice were recently found to have attenuated nephrotoxic serum nephritis (NSN), we hypothesized that endogenous Tlr agonists can activate glomerular mesangial cells. Primary mesangial cells from C57BL/6 mice expressed Tlr1-6 and Tlr11 mRNA at considerable levels and produced Il-6 when being exposed to the respective Tlr ligands. Exposure to necrotic cells activated cultured primary mesangial cells to produce Il-6 in a Tlr2/Myd88-dependent manner. Apoptotic cells activated cultured mesangial cells only when being enriched to high numbers. Apoptotic cell-induced Il-6 release was Myd88 dependent, and only purified apoptotic cell RNA induced Trif signaling in mesangial cells. Does Trif signaling contribute to disease activity in glomerulonephritis? To answer this question, we induced autologous NSN by injection of NS raised in rabbits in Trif-mutant and wild-type mice. Lack of Trif did not alter the functional and histomorphological abnormalities of NSN, including the evolution of anti-rabbit IgG and anti-rabbit-specific nephritogenic T cells. We therefore conclude that apoptotic cell RNA is a poor activator of Trif signaling in mesangial cells and that necrotic cells' releases rather activate mesangial cells via the Tlr2/Myd88 signaling pathway.

  • innate immunity
  • glomerulonephritis
  • kidney
  • necrosis
  • apoptosis
  • Toll-like receptor

danger recognition is an indispensable step for triggering innate immunity in multicellular organisms. Among the innate danger recognition receptors, the Toll-like receptors (Tlr) recognize common molecular patterns of pathogens like bacterial LPS and lipopeptides (Tlr1, -2, -4, -6) or bacterial proteins (Tlr5, -11). These Tlr are single-transmembrane molecules which are localized in the outer cell membrane (2). By contrast, the nucleic acid-specific Tlrs Tlr3, -7, -8, -9 are localized in intracellular endosomes (2). Signaling of the Tlr (and of the Il-1R family) is mediated by recruiting cytosolic adaptor molecules that bind to the intracellular Toll-interleukin 1 receptor (TIR) domain of the Tlr/Il-1Rs. The first of these intracellular adaptors identified was myeloid differentiation primary-response protein (Myd)-88 (24). However, experiments with Myd88-deficient cells revealed that at least signaling of LPS or viral dsRNA can be Myd88 independent and that other intracellular adaptors must exist (43). The TIR domain-containing adaptor protein-inducing IFN-β (Trif) was identified to exclusively mediate Tlr3 signaling and to contribute to Tlr4 signaling (9). Meanwhile, the role of Myd88- and Trif-mediated innate immunity in antimicrobial defense has been well established (2), but their role in noninfectious types of danger remains under debate. Is there evidence to support the so-called “danger signaling” hypothesis in which endogenous molecules, released during cell death or tissue remodeling, trigger tissue damage via innate pattern recognition receptors (23)?

Several studies have reported that Myd88-deficient mice develop less tissue injury in various noninfectious disease models, e.g., skin transplantation (8), encephalomyelitis (27), myocardial ischemia (10), and kidney transplantation (41). In renal ischemia-reperfusion injury, the contribution of Myd88 could clearly be devoted to Tlr2 signaling as Tlr2-deficient mice were protected from acute renal failure (20, 35). The role of Tlr2 for danger signaling in ischemic tissue damage was also confirmed for the brain (47), heart (6, 29), and liver (46). Thus certain endogenous danger signals trigger innate immune responses via the Tlr2/Myd88 signaling pathway, which contributes to tissue damage in ischemia-reperfusion injury. Furthermore, Tlr2/Myd88 signaling was shown to promote inflammation in pancreatic islets during experimental type 1 diabetes (15). Does this mechanism also apply to glomerulonephritis? Two recent studies from Brown et al. (4, 5) have demonstrated that Tlr2 signaling in nonimmune cells and immune cells of the kidney contributes to immune complex glomerulonephritis. In these studies, a Tlr2 ligand was coinjected with the nephrotoxic serum; hence, the role of Tlr2 in recognizing endogenous immunostimulatory molecules remains unclear. A number of self-molecules trigger innate immunity via the tlrs, e.g., biglycan via Tlr2/Tlr4 (34), mRNA via Tlr3 (12), U1snRNP RNA via Tlr7 (31), or hypomethylated CpG DNA via Tlr9 (17). Such endogenous Tlr agonists could be released from glomerular apoptotic and necrotic cells which are present in murine serum nephritis (22, 45, 46). In fact, blocking glomerular cell apoptosis by caspase inhibition was shown to reduce glomerular inflammation in nephrotic serum nephritis (NSN) (44). All of the aforementioned Tlr agonists except that for Tlr3 would trigger the Myd88-dependent pathway, which is consistent with the respective in vivo data mentioned before. However, it is not known whether Tlr3/Trif signaling also contributes to glomerulonephritis. A number of different shapes of endogenous RNAs can activate dendritic cells via Tlr3/Trif in vitro (3, 12, 25), but the role of endogenous RNA-mediated Trif signaling in glomerular cells and in glomerulonephritis is unknown to date. Glomerular mesangial cells (MC) express Trif and trigger glomerulonephritis upon recognition of viral RNA (26, 42). Therefore, we hypothesized that self-RNA released from dying cells would also activate MC to produce inflammatory mediators. To address this question, we performed experiments with primary MC (pMC) prepared from glomeruli of mice that lack functional Trif, Myd88, or Tlr2 and used Trif-mutant mice to characterize their phenotype upon induction of crescentic glomerulonephritis.


Animals and preparation of pMC.

Six-week-old C57BL/6 mice were obtained from Charles River (Sulzfeld, Germany). Trif-mutant-, Myd88-, or Tlr2-deficient mice in the C57BL/6 background (F6) were generated as previously described (1, 9, 38). All mice were housed in filter top cages with a 12:12-h light-dark cycle and unlimited access to food and water. Cages, bedding, nestlets, food, and water were sterilized by autoclaving before use. Kidneys were obtained under sterile conditions for the preparation of pMC. The capsule and medulla of the kidney were removed, and the renal cortices were diced in cold PBS and passed through a series of stainless steel sieves (150, 103, 63, 50, 45 μm). Glomeruli were then collected on the 50- and 45-μm sieve, washed with PBS, and treated with a 1 mg/ml solution of type IV collagenase (Worthington, Lakewood, NY) for 15 min at 37°C. Finally, the digested glomeruli were seeded into six-well plates with RPMI 1640-medium containing 20% FCS, 1% ITS (insulin, transferrine, selenium; Roche, Mannheim, Germany), penicillin (100 U/ml), and streptomycin (100 U/ml) and incubated at 37°C in a humidified atmosphere of 5% CO2. Mesangial cell outgrowth started at days 4–8. Other cell populations were subsequently scraped off and removed by gentle washing with fresh medium. After five passages, pMC were characterized by positive immunostaining for smooth muscle actin (Chemicon International, Hampshire, UK) and by negative staining for cytokeratin 18 (Chemicon International). Phalloidin (Invitrogen, Karlsruhe, Germany) and 4′-6-diamidino-2-phenylindole (Vector Laboratories, Burlingame, CA) were used to visualize the cytoskeleton and nuclei of pMC, respectively. MC preparations from different mice were used to avoid sample bias.

In vitro studies with pMC.

pMC from passages 7–11 were stimulated in 24-well plates (being 80% confluent) with Pam3CysSK4 (Invivogen, San Diego, CA), poly I:poly C RNA (Invivogen, Toulouse, France), ultrapure LPS (Invivogen), and imiquimod (Sequoia Research Products, Oxford, UK), CpG-ODN 1668 (TIB Molbiol, Berlin, Germany) with indicated concentrations for 24 h in RPMI 1640 containing 4% FCS in the presence or absence of 200 U/ml IFN-γ (PeproTech, Rocky Hill, NJ)+500 U/ml TNF-α (ImmunTools, Firesoythe, Germany). All non-LPS Tlr ligands were preincubated with polymyxin B (Invivogen) to block residual LPS contamination. Apoptotic cells were prepared by passing a whole mouse thymus through a 70-μm cell strainer to yield a single-cell suspension. Apoptosis was induced by incubating the cells in RPMI 1640 without serum in the presence of 10 mM dexamethasone for 3–4 h at 37°C. Apoptotic thymocytes were enriched using an Annexin V MicroBead Kit (Milteny Biotec, Bergisch Gladbach, Germany) following the manufacturer's protocol. For cell necrosis experiments, a murine MC line (30) and murine NIH3T3 fibroblasts (36) were also used. Cell death was induced by repeated freezing and thawing as described (33), and a necrotic cell supernatant was used in a ratio of culture medium to supernatant as indicated. The supernatant of dead cells did not show any Il-6 release. Cytokine levels were measured in cell supernatants using a commercial ELISA kit for IL-6 (OptEiA, BD Pharmingen) following the protocols provided by the manufacturer. The minimum detection level of this ELISA is 2 pg/ml. Apoptotic cell RNA was isolated by standard methods using a Qiagen RNA isolation kit (Qiagen, Hilden, Germany).

Induction and evaluation of autologous NSN.

NSN was induced in groups of C57BL/6 wild-type and Trif-mutant mice (n = 10/group) as described previously (39). In brief, nephrotoxic serum was prepared by immunizing rabbits with mouse glomerular basement membrane preparations. Nephrotoxic nephritis was induced in male mice following subcutaneous immunization in both flanks with 0.2 mg rabbit IgG (Jackson ImmunoResearch Laboratories, West Grove, PA) in Freund complete adjuvant (Sigma-Aldrich, Deisenhofen, Germany). Three days later, mice were injected intravenously with 50 μl heat-inactivated, filter-sterilized nephrotoxic rabbit serum. Spot urine samples were collected at weekly intervals. At day 21 after injection of nephrotoxic serum, mice were killed by cervical dislocation. All experimental procedures were carried out following German animal welfare legislation and had been approved by the local government authorities. All protocols followed the APS's Guiding Principles in the Care and Use of Animals. Blood and urine samples were collected from each animal at the end of the study period as described (39). Urinary albumin concentrations were measured by ELISA (Bethyl Lab, Montgomery, TX). From all mice, kidneys were fixed in 10% buffered formalin, processed, and embedded in paraffin. Two-micrometer sections for periodic acid-Schiff stains were prepared following routine protocols. Glomerulosclerosis was assessed as described elsewhere (39). Immunostaining was either performed on paraffin-embedded or frozen sections as described previously (39) using the following primary antibodies: anti-mouse CD3 (1:100, Serotec, Oxford, UK), anti-mouse IgG (1:100, M32015, Caltag Laboratories, Burlingame, CA), and a terminal transferase-dUTP-nick-end labeling (TUNEL) staining kit (Roche). Frozen sections were used to stain with anti-mouse C3 (1:200, GAM/C3c/FITC, Nordic Immunological Laboratories, Tilburg, The Netherlands). Negative controls included incubation with a respective isotype antibody. For quantitative analysis, glomerular cells were counted in 50 cortical glomeruli/section. Scoring of complement C3 deposits was done semiquantitatively from 0 to 3 (26).

Humoral and cellular rabbit IgG-specific response.

Circulating mouse anti-rabbit IgG was measured by ELISA in serum collected at the end of the study as described previously (39). The T cell response against rabbit IgG was determined by incubating splenocytes from nephritic mice of all groups with 20 μg/ml rabbit IgG (Jackson ImmunoResearch, Suffolk, UK). Splenocyte supernatants were harvested after 72 h and Ifn-γ levels were measured by ELISA as a marker of rabbit IgG-specific T cell activation (OptEiA, BD Pharmingen). Control cells were incubated with medium or 20 μg/ml sheep IgG (Jackson ImmunoResearch).

Flow cytometry.

Apoptotic cells were detected by flow cytometry using an Annexin V Microbead Kit (Milteny Biotec). Propidium iodide staining (Milteny Biotec) was used to exclude necrotic cells. Infiltrating renal macrophages were quantitated in single-cell suspensions from individual kidneys prepared as described elsewhere (40). Cells (0.5 × 106 to 1 × 106) were incubated with 1 μg of FITC-conjugated rat anti-mouse F4/80 (clone CL:A3–1; Serotec). The amounts of positively stained macrophages were expressed as the percentage of total renal cells.

Statistical analysis.

Data are expressed as means ± SD. Comparison between Trif-mutant and wild-type mice was performed using Student's t-test. Comparison between more than two groups was performed using univariate ANOVA. Post hoc Bonferroni's correction was used for multiple comparisons. P < 0.05 was considered to indicate statistical significance.


Trif and Myd88 signaling in pMC.

Glomerular injury usually involves the activation of MC, but whether Trif or Myd88 signaling contributes to this process is unknown. To answer this question, we first isolated pMC from wild-type mice. The pMC revealed the typical spindle-shaped appearance, stained positive for smooth-muscle actin and negative for the epithelial cell marker cytokeratin 18 (not shown). pMC expressed Tlr1-6 and Tlr11. Tlr7 and -9 mRNA levels were low or absent (Fig. 1A). Tlr8 was not studied here because mice generally do not express this receptor (1). Proinflammatory cytokines can specifically regulate Tlr3 expression in human mesangial cells (41). Upon exposure to a combination of the proinflammatory cytokines Tnf/Ifn-γ (200 and 500 U/ml, respectively), pMC significantly upregulated Tlr3 mRNA but neither of the other Tlrs (Fig. 1A) similar to what we recently observed in human MC or in a murine MC line (25, 40). Next, we exposed pMC to pI:C RNA (ligand to Tlr3) in the presence or absence of either Tnf/Ifn-γ or the cationic lipid lipofectamine and determined Il-6 production as a readout of MC activation by ELISA. In wild-type MC, pI:C RNA induced Il-6 only after cytokine prestimulation, a response that was absent in pMC from Trif-mutant mice (Fig. 1B) while pMC from Myd88-deficient mice showed a response similar to that in wild-type mice (Fig. 1B). By contrast, pI:C RNA complexed with cationic lipids was independent of either Myd88 or Trif (Fig. 1B), consistent with the potential of such RNA complexes to activate Tlr-independent cytosolic Tlr RNA recognition pathways (13). Stimulating pMC with the Tlr1/2 or Tlr4 agonists lipoprotein Pam3Cys or LPS also induced Il-6, consistent with the respective Tlr mRNA expression pattern (Fig. 1, A and C), and Trif-mutant pMC showed full induction of Il-6 with Pam3Cys while the LPS-induced response was lower than in wild-type MC (Fig. 1C). Pam3Cys recognition required Myd88 and Tlr2 while LPS recognition, though depending on Myd88, was Tlr2 independent (Fig. 1C). From these data we conclude that RNA can activate murine MC via a Trif-dependent or a Trif-/Myd88-independent pathway as MC lack Tlr7 and -8 expression. Tlr4 agonists activate the Trif- and the Myd88-dependent signaling pathway, and Tlr2/1 agonists exclusively depend on Myd88.

Fig. 1.

Toll-like receptor (Tlr) expression and Tlr signaling in primary mesangial cells (pMC). A: RNA was prepared from pMC cultured in the presence or absence of 200 U/ml Ifn-γ and 500 U/ml Tnf for 6 h. Tlr mRNA expression was determined by real-time RT-PCR. Values are means ± SD of 3 comparable experiments. *P < 0.05 vs. medium. B and C: pMC from wild-type, Trif-mutant, Myd88-, or Tlr2-deficient mice were stimulated with double-stranded poly I:C (pIC) RNA, Pam3Cys, or LPS as indicated for 24 h. Il-6 was determined in cell supernatants by ELISA. Values are means ± SD from 3 experiments all analyzed in triplicate. c.l., Cationic lipid; n.d., not detectable. **P < 0.01 vs. wild-type.

Role of Trif and Myd88 in mediating MC activation by necrotic cells.

Crescentic glomerulonephritis is frequently associated with focal glomerular necrosis, which may expose MC to the cellular contents of dying cells, i.e., necrotic or apoptotic cells. Necrotic cells can activate innate immunity by exposing intracellular molecules to pattern recognition receptors on antigen-presenting cells (3, 16). To test whether this mechanism also operates in glomerular MC, we exposed pMC to various types of necrotic cells in vitro. Among the three different cell types tested, necrotic fibroblasts were most potent to induce Il-6 production in wild-type pMC (Fig. 2A). Exposure to necrotic cell components is more likely to occur in states of glomerular injury which should happen in the presence of proinflammatory cytokines. In fact, necrotic cells elicited synergistic effects with Tnf/Ifn-γ on Il-6 release (Fig. 2A). Are these effects also mediated via Tlr2/Myd88 in MC as it has been shown for antigen-presenting cells (20)? To answer this question, we prepared pMC from wild-type, Trif-, Myd88-, or Tlr2-deficient mice and exposed them to increasing amounts of necrotic fibroblast supernatants. Wild-type MC released Il-6 in a dose-dependent manner in response to the necrotic fibroblast supernatant, a response that was completely lacking in Myd88- or Tlr2-deficient MC (Fig. 2B). pMC were the origin of Il-6 in the supernatants because Il-6 could not be detected in the supernatants of necrotic cells (not shown). By contrast, lack of Trif only partially reduced that response when MC were exposed to necrotic cells in a 2:1 or 1:1 ratio. Similar results were obtained when pMC supernatants were analyzed for the chemokine Cxcl9/Mig (Fig. 2C). These data suggest that necrotic cell components activate glomerular MC to secrete Il-6 or Cxcl9/Mig predominately via the Tlr2/Myd88- rather than the Trif-dependent pathway.

Fig. 2.

Necrotic cell supernatants induce Il-6 or Cxcl9/Mig production in pMC. A: pMC were cultured in the presence or absence of Tnf/Ifn-γ as indicated. pMC were stimulated with necrotic cell supernatants of a murine MC cell line (MMC), pMC, NIH3T3 renal fibroblasts, Pam3Cys, or LPS for 24 h as indicated. B: pMC from wild-type, Trif-mutant, Myd88-, or Tlr2-deficient mice were stimulated with increasing numbers of necrotic fibroblasts (NIH3T3) or pI:C RNA/cationic lipid (c.l.) complexes for 24 h as indicated. C: pMC from wild-type, Trif-mutant, Myd88-deficient mice were cultured in the presence or absence of Tnf/Ifn-γ (300 U/ml). pMC were stimulated with necrotic NIH3T3 cell supernatants. Il-6 (B) and Cxcl9/Mig (C) were determined in cell culture supernatants by ELISA. Values are means ± SD from 3 experiments all analyzed in triplicate. n.s., Not significant.

Apoptotic cells activate MC via Trif and Myd88.

Apoptosis, the other route of cell death, is considered to elicit fewer immunostimulatory effects on macrophages compared with cell necrosis (16, 31). To test whether this also applies to MC, we generated apoptotic cell preparations by incubating thymocytes with dexamethasone followed by magnetic bead enrichment for annexin V-positive cells as described in materials and methods (Fig. 3A). Exposing wild-type pMC to increasing numbers of apoptotic cells dose dependently induced Il-6 release up to fivefold (Fig. 3B), but the Il-6 levels reached were more than five times lower compared with the levels induced by necrotic fibroblast supernatants (Fig. 2A). pMC were the origin of Il-6 in the supernatants because Il-6 could not be detected in the supernatants of apoptotic cell cultures (not shown). This production of Il-6 required Myd88 and Tlr2, as Myd88- or Tlr2-deficient pMC hardly produced any Il-6 compared with wild-type pMC (Fig. 3B). By contrast, Trif was not required to elicit the Il-6 response (Fig. 3B). Obviously, apoptotic cells contain or release endogenous agonists that predominately activate the Myd88 signaling pathway in MC, which fosters the production of proinflammatory cytokines like Il-6. Apoptotic cell RNA might trigger a Trif-dependent response; hence, we repeated the experiment using RNA extracts from apoptotic thymocytes or pMC as a stimulus. The apoptotic cell RNA-induced Il-6 release of MC was almost entirely Trif dependent (Fig. 3C). Myd88-deficient pMC also showed a lower response upon stimulation with apoptotic cell RNA (Fig. 3C), indicating that the RNA preparation still contained agonists to Tlrs that signal via Myd88. Together, these data show that exposure to apoptotic cells has a moderate stimulatory effect on MC mediated via Myd88 while the Trif signaling is only activated by purified apoptotic cell RNA.

Fig. 3.

Apoptotic cells induce Il-6 production in pMC. A: apoptotic thymocytes were prepared and enriched by magnetic bead sorting for annexin V as described in materials and methods. The data shown illustrate the percentage of propidium iodine- and annexin V-positive thymocytes before (left) and after (right) magnetic bead enrichment. Note that after enrichment 99% of thymocytes were annexin V positive, and only 9% were positive for propidium iodine (late apoptotic cells). B: pMC from wild-type, Trif-mutant, Myd88-deficient, or Tlr2-deficient mice were stimulated with increasing amounts of apoptotic cells or pI:C RNA/cationic lipid complexes for 24 h as indicated. C: pMC from wild-type, Trif-mutant or Myd88-deficient mice were stimulated with 1 μg/ml of either MC RNA or apoptotic thymocyte RNA or Pam3Cys for 24 h as indicated. Il-6 was determined in cell supernatants by ELISA. Values are means ± SD from 3 experiments all analyzed in triplicate. *P < 0.05 vs. wild-type. **P < 0.01 vs. wild-type.

Crescentic glomerulonephritis and nephrotic syndrome develop independently of Trif.

What is the significance of Tlr2/Myd88- and Tlr3/Trif-dependent danger signaling in immune complex glomerulonephritis? To address this question, we induced NSN in Trif-mutant mice and wild-type mice (10 mice/group). Immunization with rabbit IgG induced comparable serum levels of anti-rabbit IgG in Trif-mutant or wild-type mice (Fig. 4A), and splenocytes from wild-type and Trif-mutant mice produced high levels of Ifn-γ upon exposure to rabbit IgG but not with sheep IgG or medium only (Fig. 4B). These data confirm that Trif is not required to generate the adaptive humoral and cellular immune response against the immunogen rabbit IgG, which is a basic requirement for the induction of the NSN model. In wild-type mice, autologous NSN was characterized by early albuminuria at week 1, increasing to massive albuminuria at week 3 (Table 1). Histopathologically, proteinuria was associated with crescentic glomerulonephritis with TUNEL+ cells within the glomerular tuft, indicating the presence of dying cells with glomerular lesions (Fig. 5, Table 1). Other markers of NSN included glomerular deposits of complement factor C3c (not shown) and IgG (Fig. 5) along the glomerular basement membrane and glomerular, periglomerular, and diffuse interstitial macrophage infiltrates (not shown). T cell infiltrates were mainly restricted to the periglomerular areas and the interstitial compartment (Fig. 5). Tubular interstitial damage was characterized by marked tubular atrophy, intratubular protein casts, and diffuse interstitial fibrosis (Fig. 5). Lack of Trif had no significant effect on the increase in albuminuria, the markers of nephrotic syndrome, serum creatinine levels, or the histopathological parameters of NSN (Table 1, Fig. 5). These data show that NSN develops independently of Trif.

Fig. 4.

Humoral and cellular immune responses in autologous nephrotoxic serum nephritis (NSN). A: serum levels of autologous mouse anti-rabbit IgG levels in wild-type and Trif-mutant mice were determined by ELISA at day 21 (n = 10/group). Titers in serial dilutions of serum are expressed in arbitrary units. Values are means ± SD. B: at the end of the study, splenocytes were prepared from mice of 2 serial experiments and incubated with rabbit IgG, sheep IgG, or medium only. After 72 h, Ifn-γ was measured in the supernatant of the cell suspensions by ELISA as a marker of rabbit IgG-specific T cell activation. Data represent respective Ifn-γ from individual mice (n = 10).

Fig. 5.

NSN in wild-type and Trif-mutant mice. A: renal sections from mice of all groups were stained with periodic acid-Schiff (PAS). Immunostaining for CD3-positive T cells, glomerular IgG, and terminal transferase-dUTP-nick-end labeling (TUNEL) stain was performed as described in materials and methods. White arrows indicate TUNEL- and DAPI double-positive nuclei in the glomerular compartment (encircled). Images are representative for 8–10 mice/group (original magnification PAS and CD3: IgG, ×200 and TUNEL, ×400). B: F4/80+ macrophages were identified by flow cytometry in kidney cell suspensions from mice of both groups (n = 8) at the end of the study. Values are mean percentages of F4/80+ cells per total renal cells ± SD.

View this table:
Table 1.

Parameters of nephrotoxic serum nephritis


We hypothesized that dying cells release endogenous ligands that activate pMC via Trif to produce inflammatory mediators. Our studies show that necrotic cells are more potent than apoptotic cells to activate pMC, a mechanism mediated via Tlr2/Myd88. By contrast, Trif partially contributes to the recognition of necrotic cell releases and of self-RNA released by apoptotic cells, but this effect was hardly detectable in vitro.

Our data provide the first evidence that exposure to necrotic cells is a robust stimulus for MC to produce Il-6 and that this effect is exclusively mediated via the Tlr2/Myd88 recognition pathway. The contribution of Myd88 is not surprising because Myd88-deficient mice develop less tissue injury in various noninfectious disease models, e.g., skin transplantation (8), encephalomyelitis (27), myocardial ischemia (10), kidney transplantation (41), and ischemia-reperfusion injury (35). Tlr2 and Tlr4 have also been shown to sense β cell death and to contribute to the initiation of type 1 diabetes (16). Myd88 is required for intracellular signaling of most Tlr (2), with MC expressing only Tlr1, Tlr2, and Tlr4–6. Tlr2 is a crucial element in this context because Tlr1 and Tlr6 require Tlr2 as a partner to form heterodimers for ligand binding (2). In fact, Tlr2 and Tlr4 have been identified to mediate immune activation upon exposure to a number of different endogenous molecules, including biglycan (34), fibrinogen (2), hyaloran (2), Tamm-Horsfall protein (28), and heat shock proteins (18). The latter are intracellular molecules which may not be able to activate TLR signaling before being released, e.g., during cell death (16). In fact, necrotic cells were shown to activate NF-κB via Myd88 and induce expression of genes involved in inflammatory and tissue-repair responses, including neutrophil-specific chemokine genes KC and macrophage-inflammatory protein-2, in viable fibroblasts and macrophages (21). In the latter study, Tlr2 was identified to mediate this process. In addition, in renal ischemia-reperfusion injury the contribution of Myd88 also depends on Tlr2 because Tlr2-deficient mice were protected from acute renal failure (20, 35). The role of Tlr2/Myd88 signaling in exacerbating glomerulonephritis was recently addressed by Brown et al. (5). Coinjection of a synthetic Tlr2 agonist aggravated the autologous NSN model but not in Tlr2-deficient mice, which, however, could also be explained by the impairment of adaptive immunity in autologous NSN (5). Another study reported from the same group analyzed heterologous NSN 2 h after NSN and TLR2 agonist injection in bone marrow of Tlr2 chimeric mice (4). This study allowed the group to conclude that the Tlr2 agonist-induced aggravation of NSN involved Tlr2 on intrinsic renal cells (5). Our data suggest that Tlr2 may also trigger MC activation upon exposure to necrotic cell releases. This finding supports the concept that pathogen recognition receptors like Tlr2 can also function as danger recognition receptors and serves as the rationale for a new hypothesis that intraglomerular necrotic lesions promote local inflammation via this mechanism.

Tlr2/Myd88 signaling entirely mediated necrotic cell-induced MC secretion of proinflammatory cytokines like Il-6 or chemokines like Cxcl9/Mig and partially mediated a similar response induced by apoptotic cells. The response induced by apoptotic cell RNA was largely Trif dependent. However, the overall in vitro responses were low compared with the effects induced by necrotic cells. We required pure apoptotic cell RNA extracts to induce Il-6 in mesangial cells. This experimental design is artificial and rather unlikely to mimic an in vivo scenario. Interestingly, NSN developed completely independently of Trif. Our analysis cannot completely rule out that the anti-rabbit IgG response developed more slowly in Trif-mutant mice, but Trif is usually not required to mount humoral immune responses to antigens delivered with Freund adjuvant (7). Obviously, self-RNA is less potent than viral RNA to trigger a robust immune activation. In mammalian RNA (and DNA), nucleoside modifications prevent the immunostimulatory effects known from viral RNA (11, 13, 37). By contrast, the stimulatory effects of bacterial RNA, RNA transcribed in vitro, or chemically synthesized RNA on dendritic cells or stably transformed 293 cells expressing Tlr3, Tlr7, or Tlr8 can be eliminated by distinct nucleoside modifications, such as m5C, m6A, m5U, pseudouridine, or 2′-O-methyl-U (13). The amount of intracellular RNA exceeds by far the amount of DNA, but demethylation also unmasks the immunostimulatory effects of self-DNA and methylation blocks the immunostimulatory effects of bacterial DNA (16). While this effect contributes to the lack of RNA/DNA-induced immune activation in tissue necrosis, apoptotic cell death provides additional mechanisms to avoid inflammation (16). Apoptosis, also named “the silent route of cell death,” prevents the release of intracellular content into the extracellular space (32). Rapid phagocytosis of apoptotic cells usually avoids local accumulation of apoptotic cell bodies. Endosomal processing of apoptotic cell bodies should expose self-RNA/DNA to Tlr3, -7, and -9 in glomerular macrophages and to Tlr3 in MC. However, the RNA/DNA modifications outlined above and additional yet unknown mechanisms prevent TLR activation. Hence, the weak Trif-dependent activation of MC upon exposure to apoptotic cell RNA noticed in vitro is not a crucial pathomechanism of NSN.


This work was supported by grants from the Deutsche Forschungsgemeinschaft (AN372/8-1, GRK 1202), the Fritz Thyssen Foundation, and the EU Integrated Project “INNOCHEM” (FP6-518167) to H.-J. Anders. J. Lichtnekert was funded by Grant GRK 1202. V. Vielhauer was supported by the Deutsche Forschungsgemeinschaft (VV231/2-1).


The authors thank Dr. H. Weighardt and Dr. C. J. Kirschning, Technical University, Munich, for generously providing Trif-, Myd88-, and Tlr2-deficient mice. The expert technical assistance of Dan Draganovic, Iana Mandelbaum, and Stephanie Pfeiffer is gratefully acknowledged. Parts of this project were prepared as a doctoral thesis (J. Lichtnekert) at the Medical Faculty of the University of Munich.


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


View Abstract