Neutrophil Extracellular Trap-Related Extracellular Histones Cause Vascular Necrosis in Severe GN

Abstract

Severe GN involves local neutrophil extracellular trap (NET) formation. We hypothesized a local cytotoxic effect of NET-related histone release in necrotizing GN. In vitro, histones from calf thymus or histones released by neutrophils undergoing NETosis killed glomerular endothelial cells, podocytes, and parietal epithelial cells in a dose-dependent manner. Histone-neutralizing agents such as antihistone IgG, activated protein C, or heparin prevented this effect. Histone toxicity on glomeruli ex vivo was Toll-like receptor 2/4 dependent, and lack of TLR2/4 attenuated histone-induced renal thrombotic microangiopathy and glomerular necrosis in mice. Anti–glomerular basement membrane GN involved NET formation and vascular necrosis, whereas blocking NET formation by peptidylarginine inhibition or preemptive anti-histone IgG injection significantly reduced all aspects of GN (i.e., vascular necrosis, podocyte loss, albuminuria, cytokine induction, recruitment or activation of glomerular leukocytes, and glomerular crescent formation). To evaluate histones as a therapeutic target, mice with established GN were treated with three different histone-neutralizing agents. Anti-histone IgG, recombinant activated protein C, and heparin were equally effective in abrogating severe GN, whereas combination therapy had no additive effects. Together, these results indicate that NET-related histone release during GN elicits cytotoxic and immunostimulatory effects. Furthermore, neutralizing extracellular histones is still therapeutic when initiated in established GN.

Rapidly progressive glomerulonephritis (RPGN) encompasses a heterogeneous group of disorders resulting in severe glomerular inflammation and injury. Clinically, RPGN is characterized by a rapid loss of GFR, hematuria, and proteinuria caused by characteristic glomerular lesions such as capillary necrosis and hyperplasia of the parietal epithelial cells (PECs) along Bowman’s capsule forming crescents. The pathogenesis of RPGN involves autoantibodies, immune complex–mediated activation of complement, the local production of cytokines and chemokines, and glomerular leukocyte recruitment.¹ RPGN is more common in ANCA-associated GN or anti–glomerular basement membrane (GBM) disease than in other forms of GN,^2,3 but why? The hallmark of severe GN is glomerular capillary necrosis leading to hematuria and plasma leakage.⁴ PEC exposure to plasma is sufficient to trigger crescent formation⁵ but inflammation and PEC injury serve as additional stimuli.⁶ What causes vascular necrosis inside the glomerulus?

Two separate lines of research hold potential answers to this question: (1) the discovery of neutrophil extracellular trap (NET) formation and (2) the toxicity of extracellular histones. NET formation results from an explosion-like directed expulsion of the neutrophil’s nuclear chromatin, which decondensates to generate a meshwork that can immobilize and kill bacteria during infections.⁷ Cytokine-induced NET formation also drives sterile injury including necrotizing GN.^8–11 Many cytosolic or chromatin-related components could account for the toxic and proinflammatory effect of NETs, such as proteolytic enzymes or intracellular molecules with immunostimulatory effects, referred to as danger-associated molecular patterns (DAMPs).¹² In this context, histones are particularly interesting candidate mediators.

Histones are nuclear proteins that wind up the double-stranded DNA to form chromatin. Dynamic modifications of histone residues regulate gene transcription by determining the accessibility of transcription factors to their DNA binding sites.¹³ However, when cell necrosis releases histones into the extracellular space, they display significant cytotoxic effects.^14–17 Histones contribute to fatal outcomes in murine endotoxinemia caused by microvascular injury and activation of coagulation.^15,18–20 We previously showed that dying renal cells release extracellular histones that promote septic and postischemic AKI.²¹ We further demonstrated that histones act as DAMPs by activating Toll-like receptor (TLR)-2 and -4 as well as NOD-like receptor family, pyrin domain containing 3 (NLRP3),^21,22 which was confirmed by other groups.^20,23,24 Because TLR2/TLR4-mediated pathology is an essential mechanism of crescentic GN,^25,26 we hypothesized that extracellular histones originating from glomerular cells and/or NETs may elicit toxic effects on glomerular endothelial cells and promote capillary necrosis and/or podocyte loss leading to proteinuria and crescent formation. Therefore, we hypothesized that either inhibiting NETosis or neutralizing extracellular histones might be therapeutic during severe GN.

Results

Glomerular TLR2 and TLR4 Expression in Severe Human GN

Based on our previous discovery that extracellular histones have agonistic effects on TLR2 and TLR4,²¹ we first asked whether these receptors are expressed in the healthy and diseased glomerulus. TLR2/TLR4 immunostaining of normal human kidney showed a weak granular positivity in all glomerular cells; TLR4 positivity was clearly noted in glomerular endothelial cells (Figure 1A). In addition, TLR2 was strongly positive in the cytoplasm of epithelial cells of the proximal and distal tubule, whereas this was less prominent for TLR4 (Figure 1A). Immunostaining of kidney biopsies of patients with ANCA-associated necrotizing and crescentic GN revealed prominent positivity also in PECs along the inner aspect of Bowman’s capsule (Figure 1B). Because glomerular crescents are largely formed by PECs,^27,28 glomerular crescents displayed TLR2 and TLR4 positivity (Figure 1C). Of note, isotype IgG staining was negative in all tissues (Supplemental Figure 1). Thus, the cells of the normal glomerulus express TLR2/TLR4 and PECs induce these TLRs in crescentic GN.

Figure 1.

TLR2 and TLR4 expression in human crescentic glomerulonephritis. (A–C) TLR2 and TLR4 immunostaining is performed on healthy kidney tissue (A) or on kidney biopsies of patients with newly diagnosed ANCA vasculitis and clinical signs of GN (B and C). B shows representative glomeruli unaffected by loop necrosis or crescent formation, while glomeruli affected by such lesions are shown in C. Original magnification, ×400.

Antihistone IgG Prevents Histone Toxicity on Glomerular Cells

Histones were previously shown to be toxic on pulmonary endothelial cells in vitro and in vivo.^15,18 We tested this effect on cultured glomerular endothelial cells and found that a total histone preparation was cytotoxic in a dose-dependent manner. Antihistone IgG derived from the BWA3 hybridoma is known to neutralize the toxic and immunostimulatory effect of extracellular histones.^15,24,29 Antihistone IgG almost entirely prevented histone toxicity on glomerular endothelial cells up to a histone concentration of 30 μg/ml (Figure 2A). Antihistone IgG also prevented histone-induced glomerular endothelial cells microtubule destruction in angiogenesis assays (Supplemental Figure 2, A and B). Histone-induced toxicity was also evident in cultured podocytes and PECs albeit at much higher histone concentrations compared with the toxic dose required killing endothelial cells (Figure 2A). Antihistone IgG also significantly reduced histone-induced detachment of cultured podocytes (Supplemental Figure 3). Thus, extracellular histones are toxic to glomerular cells, which can be blocked by antihistone IgG.

Figure 2.

NETosis-related extracellular histones kill glomerular cells. (A) Murine glomerular endothelial cells, podocytes, and PECs are incubated with increasing doses of histones together with either control IgG or antihistone IgG. Cell viability is determined after 24 hours by MTT assay. (B) Immunostaining of naïve (left) and TNF-α–activated neutrophils (right) in culture. Staining for elastase (red) and histones (green) illustrates how TNF-α triggers NETosis leading to NET formation (i.e., expelling cytoplasmic and nuclear contents in the extracellular space). (C–E) Similar experiments are performed on monolayers of glomerular endothelial cells, which appear flat and evenly laid out on scanning electron microscopy (C, left). However, neutrophil NETosis leads to severe injury and death of endothelial cells appearing as bulging white balls with corrugated surfaces adjacent to activated NETs (C, middle). This effect is almost entirely prevented by antihistone IgG demonstrated by significant reversal of the structural integrity of the endothelial cell monolayer (C, right). (D) Immunostaining for elastase, histone, and DAPI illustrate the same effect. (E) MTT assay analysis of endothelial cell viability allows quantifying this effect, which is identical for TNF-α and PMA, two known inducers of NETosis. Data in A and E represent the mean OD±SEM of three experiments measured at a wavelength of 570 nm. *P<0.05; **P<0.01; ***P<0.001 versus control IgG. GEnC, glomerular endothelial cell; Cont, control; DAPI, 4′,6-diamidin-2-phenylindol; Hist, histone.

NETs Kill Glomerular Endothelial Cells through Histone Release

In severe GN, neutrophils undergo NETosis, which deposits nuclear chromatin within the glomerular capillaries.⁸ Immunohistochemical staining displayed nuclear chromatin release from netting neutrophils including the spreading of histones outside the dying cells (Figure 2B). Neutrophils undergoing TNF-α– or phorbol 12-myristate 13-acetate (PMA)–induced NETosis on monolayers of glomerular endothelial cells destroyed this monolayer by inducing endothelial cell death, whereas TNF or PMA alone did not (Figure 2, C–E). This NETosis-related endothelial cell toxicity was entirely prevented by antihistone IgG (Figure 2, C–E). Thus, we conclude that netting neutrophils damage glomerular endothelial cells via the release of histones.

Histones Need TLR2/TLR4 to Trigger Glomerular Necrosis and Microangiopathy

It is not clear whether glomerular toxicity of extracellular histones is TLR2/TLR4 dependent. To answer this question, we exposed glomeruli isolated from wild-type and Tlr2/Tlr4-deficient mice to histones ex vivo. Histone exposure was cytotoxic to glomeruli, a process that was entirely prevented using glomeruli from Tlr2/Tlr4-deficient mice (Figure 3A). Lack of TLR2/TLR4 also prevented IL-6 and TNF expression in histone-exposed glomeruli (Supplemental Figure 4). We also studied the effects of extracellular histones on glomeruli in vivo. Because intravenous histone injection kills mice immediately by pulmonary microvascular injury,¹⁵ we injected histones directly into the left renal artery in anesthetized mice. Unilateral histone injection caused glomerular lesions within 24 hours ranging from minor endothelial fibrinogen positivity to thrombotic microangiopathy and global glomerular necrosis (Figure 3, B–F). The contralateral kidney remained unaffected (not shown). Histone injection into the renal artery of Tlr2/Tlr4-deficient mice showed significantly reduced glomerular lesions and fibrinogen positivity (Figure 3, B and F). These results demonstrate that extracellular histones induce glomerular injury in a TLR2/TLR4-dependent manner.

Figure 3.

Extracellular histones injure glomeruli in a TLR2/TLR4-dependent manner. (A) Glomeruli are isolated from wild-type and Tlr2/Tlr4-deficient mice and incubated with histones (30 µg/ml). After 12 hours, LDH release into the supernatant is measured as a marker of glomerular cell injury. Data represent the mean OD±SEM of three experiments measured at a wavelength of 492 nm. **P<0.01; ***P<0.001 versus control. (B) For intra-arterial histone injection, the abdominal aorta is prepared and a microcannula is placed into the left renal artery to inject histones directly into the kidney. Images show hematoxylin and eosin staining of representative glomeruli of the different groups as indicated. (C–E) Fibrinogen immunostaining displays three different staining patterns: diffuse positivity of glomerular endothelial cells, entire luminal positivity indicating microthrombus formation, and global positivity of glomerular loop indicating loop necrosis. (F) A quantitative analysis of these lesions reveals that histone injection massively increases luminal and global fibrinogen positivity, which is partially prevented in Tlr2/Tlr4-deficient mice. **P<0.01; ***P<0.001 versus saline; ^#P<0.05; ^##P<0.01 versus histone group. DKO, double knockout; LDH, lactate dehydrogenase. Original magnification, ×400 in B–E.

Extracellular Histones Contribute to Severe GN

On the basis of these results, we speculated that NET-related histone release may also contribute to severe GN in vivo. To address this question, we first used the peptidylarginine deiminase inhibitor Cl-amidine to block NETosis in a mouse model of heterologous anti-GBM nephritis.³⁰ Cl-amidine treatment prevented glomerular NET formation at day 7 as evidenced by reduced extracellular myeloperoxidase immunostaining (Figure 4A). Blocking glomerular NET formation preserved glomerular CD31⁺ endothelial cells, glomerular lesions, and BUN levels (Figure 4, A–E). Next, we applied the same neutralizing antihistone IgG as before in vitro. Mice were injected intraperitoneally with 20 mg/kg antihistone IgG or with 20 mg/kg control IgG 24 hours before the intravenous injection of a GBM antiserum raised in sheep. At the end of the study at day 7, only sheep IgG but no mouse IgG deposits were found in glomeruli, excluding any autologous anti-sheep IgG response contributing to GN (Supplemental Figure 5A). Antihistone IgG significantly reduced BUN levels after GBM antiserum injections (Figure 4F). This was associated with a significant reduction in crescent formation and global glomerular pathology with less severe lesions 7 days after antiserum injection (Figure 4, G and I). NET-related myeloperoxidase staining inside glomeruli was associated with focal loss of endothelial CD31 positivity as a marker of glomerular vascular injury (Figure 4, J and K). Antihistone IgG did not affect extracellular positivity but maintained CD31⁺ vasculature (Figure 4J), indicating a protective effect on NET-related vascular injury. Because histones were toxic to glomerular endothelial cells and podocytes in vitro, we assessed the glomerular capillary ultrastructure by transmission electron microscopy. In control mice with crescentic glomeruli, there was severe glomerular damage with fibrin deposits replacing large glomerular segments (fibrinoid necrosis). The capillary loops showed extensive GBM splitting and thinning, prominent endothelial cell nuclei, massive subendothelial edema with closure of the endothelial fenestrae, and obliteration of the capillary lumina. Subendothelial transudates (leaked serum proteins) and luminal platelets and neutrophils were also noted. Severe podocyte injury with diffuse foot process effacement, reactive cytoplasmic changes, and detachment from the GBM were apparent (Figure 5A). In contrast, glomeruli of mice injected with antihistone IgG showed restored endothelial fenestrations, flat appearing endothelial cells and preserved podocytes with intact foot processes (Figure 5A). Wilms' tumor 1 (WT-1)/nephrin coimmunostaining revealed that antihistone IgG largely prevented podocyte loss in antiserum-induced GN (Figure 5, B and C). This was consistent with significant reduction of albuminuria on day 7 after antiserum injection compared with control IgG-treated mice (Figure 5D). These results demonstrate that extracellular histones induce severe GN by causing glomerular vascular injury and podocyte loss, accompanied by proteinuria. These results were similar, after blocking NETosis using the peptidylarginine demininase inhibitor (Figure 5, E and F, Supplemental Figure 6).

Figure 4.

Neutralizing histones protects from severe GN. (A) CD31 and MPO immunostaining representing NETs formation in the glomeruli of vehicle group. PAD inhibition shows no NETs in terms of MPO positivity. (B) Quantification of mean fluorescence area for MPO and CD31 positivity in glomeruli. (C) The model of antiserum-induced GN using PAD inhibitor 24 hours after antiserum injection shows reduced proteinuria and BUN at day 7. (D) Glomerular lesions. (E) Crescents. (F) BUN levels are determined 1 and 7 days after intravenous injection of GBM antiserum. Mice are either treated with control IgG or antihistone IgG starting from the day before antiserum injection. (G) Representative hematoxylin and eosin stainings of glomeruli are shown. (H and I): Morphometric analysis of segmental and global glomerular lesions (left) and of glomeruli with crescents (right) as described in the Concise Methods. (J) CD31 and MPO immunostaining representing NETs formation in the glomeruli of control IgG group shows focal loss of endothelial cell positivity compared with the antihistone IgG group. (K) Quantification of mean fluorescence area for MPO and CD31 positivity in glomeruli. Data are means±SEM from five to six mice in each group. *P<0.05; **P<0.01; ***P<0.001 versus control IgG. Cont, control; Hist, histone; inh, inhibition; MPO, myeloperoxidase; PAD, peptidylarginine demininase; ve, vehicle. Original magnification, ×400.

Figure 5.

Neutralizing histones protects the glomerular filtration barrier in GN. (A) Transmission electron microscopy of antiserum-induced GN reveals extensive glomerular injury with fibrinoid necrosis (upper left), endothelial cell swelling, luminal thrombosis, and intraluminal granulocytes (upper middle and right). Podocytes show foot process effacement (all upper images). Preemptive treatment with antihistone IgG decreases most of these abnormalities, particularly endothelial cell and podocyte ultrastructure (lower images). (B and C) Immunostaining for WT-1 (red) and nephrin (green) is used to quantify podocytes. Antihistone IgG doubles the number of nephrin/WT-1+podocytes at day 7 of antiserum-induced GN. (D) The urinary albumin/creatinine ratio is determined at day 1 and day 7 after antiserum injection. (E and F) Urinary albumin/creatinine ratio and number of podocytes after blocking NETs with PAD inhibitor. Data represent the mean±SEM from five to six mice of each group. *P<0.05; ***P<0.001 versus control IgG. Cont, control; Hist, histone; PAD, peptidylarginine demininase.

Extracellular Histones Drive Glomerular Leukocyte Recruitment and Activation

Infiltrating leukocytes are not only a documented source of extracellular histones in severe GN,⁸ but they are also important effector cells.³¹ For example, in GBM antiserum-exposed glomerular endothelial cells, histone exposure triggered chemokine (c-x-c motif) ligand 2 expression (Supplemental Figure 5B). In vivo, antihistone IgG significantly reduced the numbers of glomerular neutrophils and macrophages as quantified by immunostaining (Figure 6A). Flow cytometry of renal cell suspensions allowed us to better distinguish renal mononuclear phagocyte populations. Antihistone IgG significantly reduced the numbers of activated (MHC II⁺) F4/80⁺ cells as well as of activated (CD86⁺) CD11b/CD103⁺ cells, and of CD4⁻ dendritic cells (Figure 6B). In fact, histones dose dependently induced activation markers like MHCII, CD40, CD80, and CD86 in cultured bone marrow-derived macrophages (BMDCs) as well, which was entirely prevented with antihistone IgG (Figure 6C). Taken together, extracellular histones trigger glomerular leukocyte recruitment and activation, which can be blocked with antihistone IgG in vitro and in vivo.

Figure 6.

Leukocyte recruitment and activation in GN. (A) Glomerular neutrophil (Ly-6B.2) and macrophage (Mac-2) infiltrates are quantified by immunostaining. Representative images are shown. (B) Leukocyte activation is quantified by flow cytometry of renal cell suspensions harvested 7 days after antiserum injection. Data represent the mean±SEM from five to six mice of each group. (C) Cultured bone marrow–derived dendritic cells are exposed to increasing doses of histones as indicated. After 24 hours, flow cytometry is used to determine the percentage of cells that express the activation markers MHC II, CD40, CD80, and CD86. Data are means±SEM from three independent experiments. *P<0.05; **P<0.01; ***P<0.001 versus control IgG. Cont, control; Hist, histone. Original magnification, ×400.

Extracellular Histones Trigger Intraglomerular TNF-α Release and Thrombosis

Activated mononuclear phagocytes are also an important source of proinflammatory cytokines in glomerular disease. Among these, TNF-α particularly contributes to podocyte loss, proteinuria, and glomerulosclerosis.³² Because antihistone IgG entirely prevented histone-induced TNF-α secretion in cultured macrophages and dendritic cells (Figure 7A), we next assessed glomerular TNF-α expression. Immunostaining displayed robust TNF-α positivity within the glomerular tuft, which localized not only in infiltrating cells but also in the inner and outer aspect of the glomerular capillaries (Figure 7B). Antihistone IgG strongly reduced glomerular TNF-α positivity, which was consistent with the corresponding renal mRNA expression levels (Figure 7C). TNF-α is not only an inducer of NETosis but also triggers a prothrombotic activation of (glomerular) endothelial cells and intravascular fibrin formation.^33–35 Our GN model displayed global fibrinogen positivity within glomerular capillaries, which was almost entirely prevented with antihistone IgG (Figure 7D). In addition, fibrinogen mRNA levels were reduced in the antihistone IgG group (Figure 7E). These results show that extracellular histones trigger intraglomerular TNF-α production and microthrombi formation within glomerular capillaries.

Figure 7.

Histones activate TNF-α production. (A) Cultured J774 macrophages and BMDCs respond to histone exposure by inducing the secretion of TNF-α, which is blocked by antihistone IgG. Data are means±SEM from three independent experiments. ***P<0.001 versus control IgG. (B and D) TNF-α and fibrinogen immunostaining on renal sections from both treatment groups taken at day 7 after antiserum injection. Representative images are shown. (C and E) Real-time RT-PCR for TNF-α and fibrinogen mRNA on renal tissue at day 7 after antiserum injection. Data are means±SEM from at least five to six mice in each group. *P<0.05 versus control IgG. (F) Immunostaining of renal sections of all groups for claudin-1 (red, marker for PECs and some tubular cells), WT-1 (green, marker for podocytes and activated PECs), and DAPI (blue, DNA marker) illustrates that in severe GN, crescents consist of WT-1+PECs, which is reversed with antihistone IgG. (G) Mouse PEC viability (MTT assay) when cultured in the presence of different serum concentrations together with a low concentration (20 μg/ml) of histones, that without serum histones reduces PEC viability. Together with serum histones rather promote PEC growth. (H): Similar experiment showing that blocking anti-TLR2 and anti-TLR4 antibodies and antihistone IgG neutralize the histone effect on PEC growth. Data are the mean OD±SEM of three experiments measured at a wavelength of 570 nm. **P<0.01; ***P<0.001 versus control IgG. (I) RT-PCR analysis of PECs stimulated with histones and various neutralizing compounds (antihistone IgG, heparin 50 μg/ml, activated protein C 500 nM, anti-TLR2 or anti-TLR4 1 ng/ml). Note that all of these interventions block histone-induced CD44 and WT-1 mRNA expression, which serve as markers of PEC activation. Data are means±SEM of three experiments. ^#P<0.05; **P<0.01; ***P<0.001 versus histone group. Cont, control; Hist, histone; DAPI, 4′,6-diamidin-2-phenylindol. Original magnification, ×400 in B and D; 200 in F.

Extracellular Histones Activate PECs via TLR2/TLR4

Mitogenic plasma proteins leaking from injured glomerular capillaries cause PEC hyperplasia and glomerular crescent formation.^5,27,28 In fact, in antiserum-induced GN glomerular crescents were positive for claudin-1/WT-1–positive cells (Figure 7F), where claudin-1 identifies PECs and WT-1 marks PEC activation.^36,37 In fact, PECs cultured in 10% serum started proliferating upon histone exposure (Figure 7G), which was entirely prevented by antihistone IgG or TLR2/TLR4 inhibition (Figure 7H). TLR2/TLR4 inhibition also blocked histone-induced expression of CD44 and WT-1 in PECs (Figure 7I). Previous reports documented that heparin and recombinant activated protein C (aPC) also block histone toxicity.^15,38 As such, the protective effect on PEC activation was shared by antihistone IgG, heparin, or activated protein C (aPC) (Figure 7I), the latter two suppressing histone cytotoxicity on glomerular endothelial cells just like antihistone IgG (Supplemental Figure 7). Thus, extracellular histones activate PECs in a TLR2/TLR4-dependent manner, a process that may act synergistically with other triggers of PEC hyperplasia during crescent formation and that can be blocked by antihistone IgG, aPC, or heparin.

Delayed Onset of Histone Neutralization Still Improves Severe GN

The results of preemptive histone neutralization prove their pathogenic contribution to severe GN but can it be also therapeutic in established disease? We used the aforementioned three therapeutic interventions (i.e., antihistone IgG, heparin, and aPC); all completely blocked histone toxicity on glomeruli ex vivo (Figure 8A). In another series of identical experiments, we initiated antihistone IgG, heparin, and aPC treatments 24 hours after GBM antiserum injection, a time point at which massive proteinuria and elevated BUN were already established (Figures 4A and 5D). All these treatments consistently and significantly reduced plasma creatinine levels, proteinuria, and podocyte loss at day 7 (Figures 8, B–D). Histone blockade also significantly reduced the percentage of glomeruli with global lesions or halted damage (Figure 8E). Glomerular crescents were reduced by 80% (Figure 8F), as were features of secondary tubular injury (Figure 8G). This was associated with less glomerular neutrophil and macrophage infiltrates as well as a significant reduction of intrarenal leukocyte subpopulations in number as well as their activation as identified by flow cytometry (Supplemental Figure 8). Combination therapy of all of the three treatments did not show any additive effect (Supplemental Figure 9). Thus, delayed onset of histone blockade with antihistone IgG, heparin, or aPC protects from renal dysfunction and structural injury during severe GN.

Figure 8.

Delayed histone blockade still improves GN. (A) Glomeruli are isolated from wild-type mice and incubated with histones in the presence or absence of antihistone IgG, heparin, or aPC as before. LDH release is measured in supernatants as a marker of glomerular cell injury. Data are the mean OD±SEM of three experiments. *P<0.05; **P<0.01; ***P<0.001 versus control IgG or vehicle group histone group, respectively. (B) Further experiments again use the model of antiserum-induced GN using antihistone IgG, heparin, or recombinant aPC initiated only after disease onset (i.e., 24 hours after antiserum injection, when the urinary albumin/creatinine ratio was around 80 μg/mg). (C) Data show plasma creatinine levels at day 7 and albuminuria also at day 2. (D) Podocytes are quantified as nephrin/WT-1+cells on renal sections at day 7. (E–G) Glomerular lesions (E), crescents (F), and tubular injury (G) are quantified by morphometry from sections stained with periodic acid–Schiff taken at day 7. Glomerular podocyte numbers are assessed by WT-1/nephrin costaining on glomerular cross-sections. Data represent the mean±SEM from five to six mice of each group. *P<0.05; **P<0.01; ***P<0.001 versus control IgG or vehicle, respectively. Cont, control; Hist, histone; LDH, lactate dehydrogenase.

Discussion

We hypothesized that NET formation inside the glomerulus releases extracellular histones that elicit toxic and immunostimulatory effects on glomerular cells during necrotizing and crescentic GN. Our data confirm this concept and also demonstrate that histone neutralization continues to be protective when it commences after disease onset, which implies a potential therapeutic use of histone-neutralizing agents in severe GN.

Necrotizing and crescentic GN, such as seen in ANCA-associated renal vasculitis or anti-GBM disease, is associated with neutrophil-induced glomerular injury. As first discovered in 2004, NETosis is a regulated form of neutrophil death that supports killing of extracellular bacteria.⁷ NETosis is not limited to antibacterial host defense and also occurs in sterile forms of inflammation, because it can be triggered by proinflammatory cytokines such as TNF-α. Our in vitro studies show that TNF-α is a sufficient stimulus to trigger NETosis-driven injury of glomerular endothelial cells. We also show that blocking NET formation prevents glomerular pathology in GN, which is consistent with previous findings in lupus nephritis.³⁰ NETosis releases many aggressive proteases, oxygen radicals, and potential DAMPs into the extracellular space that drive vascular injury in the glomerulus. For example, DAMPs activate TLRs and other pattern recognition receptors of the innate immune system. Our data demonstrate an essential role of histones in this context. The endothelial toxicity of extracellular histones was first described in a seminal article on sepsis, in which early lethality was due to microvascular endothelial cell injury in the lung.¹⁵ Subsequent reports further explored the thrombogenic potential of extracellular histones via direct activation of endothelial cells as well as of platelets.^{18–20,39–41} In infection and sepsis models, NETosis is the most likely source of extracellular histones. However, in mechanical trauma, toxic liver injury, cerebral stroke, and postischemic renal tubular necrosis histones are also released from dying tissue cells.^17,21

Our in vitro and in vivo data clearly demonstrate that extracellular histones are toxic to glomerular cells and promote glomerular injury in healthy mice upon intra-arterial injection or during severe antiserum-induced GN. The mechanisms of histone toxicity are not entirely clear but are thought to be due to their strong basic charge, a TLR-independent form of cytotoxicity.⁴² Although histones’ basic charge is needed inside the nucleus to neutralize acidic residues of the DNA, it has the capacity to damage cell membranes outside the cell.⁴² The polyanion heparin blocks this charge effect of histones, which may explain its antagonistic effect on histone toxicity in vitro and in vivo. However, we and others discovered that histones also elicit DAMP-like immunostimulatory activity by activating TLR2, TLR4, and NLRP3 in dendritic cells and possibly also in other cell types,^20–24 which is another pathway of how extracellular histones trigger sterile inflammation. Because TLR2 and TLR4 (but not NLRP3) are known to induce glomerular injury in the heterologous anti-GBM GN model and are expressed inside the glomerulus in human ANCA vasculitis, we further explored the histone-TLR2/TLR4 axis.^{25,26,43–45} Tlr2/Tlr4-deficient glomeruli were protected from histone-induced injury ex vivo and in vivo, implying that the histone-related glomerular injury refers to the TLR2/TLR4-dependent DAMP effect. In particular, the presence of serum turned the cytotoxic effect of histones on PECs into PEC proliferation, which was entirely TLR2/TLR4 dependent. Although PEC necrosis can be followed by excessive PEC recovery leading to PEC hyperplasia and crescent formation,⁶ concomitant plasma leakage and histone release provide additional mitogenic stimuli during severe GN.⁵

Our proof-of-concept experiments were based on preemptive histone neutralization with antihistone IgG. To explore a potential efficacy of histone blockade in severe GN, we also applied three different modes of histone inactivation after GN induction. Delayed onset of antihistone IgG was equally protective as preemptive therapy in terms of glomerular injury, proteinuria, and serum creatinine levels. The same applies to heparin treatment, which confirms previously published results in GN models.⁴⁶ Our data clearly document that heparin inhibits the direct toxic effects of histones on glomerular endothelial cells, which is consistent with the results of other investigators in other cell types.^{14,20,38,39,41} As previously reported, aPC degrades extracellular histones.¹⁵ In our work, it was as equally effective as antihistone IgG and heparin in abrogating extracellular histone toxicity in vitro and severe GN in vivo. The fact that combination therapy of all three of these agents did not show any additive effect supports the concept that their protective effect on GN is via histone neutralization and not via unrelated mechanisms.

Together, NETosis releases histones into the extracellular space, where they have toxic effects on glomerular endothelial cells and podocytes. Extracellular histone-induced glomerular injury is partially due to TLR2/TLR4. Preemptive as well as delayed onset of histone neutralization either by antihistone IgG, recombinant aPC or heparin abrogates all aspects of GBM antiserum-induced severe GN. We conclude that extracellular histones represent a novel therapeutic target in severe GN.

Concise Methods

Mice and Anti-GBM Nephritis Model

C57BL/6 mice were procured from Charles River Laboratories (Sulzfeld, Germany). Mice were housed in groups of five mice in standard housing condition with a 12-hour light/dark cycle. Cages, nest lets, food, and water were sterilized by autoclaving before use and mice were allowed to unlimited access to food and water. Mice aged 6–8 weeks mice received an intravenous injection of 100 µl of anti-GBM serum (PTX-001 sheep anti-rat GBM serum; Probetex, Inc.). Urine samples were collected on different time points after antiserum injection to evaluate the functional parameters of the kidney damage. On day 7, the mice were euthanized by cervical dislocation to collect plasma and kidney tissue. Kidneys were kept at −80°C for protein isolation and in RNAlater solution at −20°C for RNA isolation. A part of the kidney was also kept in formalin to be embedded in paraffin for histologic analysis.⁴⁷ We treated groups of mice either with Cl-amide 10 mg/kg intraperitoneally (EMD Millipore, Darmstadt, Germany), heparin 50 IU (Ratiopharm, Ulm, Germany) intraperitoneally, or aPC 5 mg/kg intraperitoneally (Lilly, UK), or 20 mg/kg control IgG intraperitoneally or antihistone IgG (clone BWA3; Immunomedics, Morris Plains, NJ) or combination of antihistone IgG, heparin, and aPC to neutralize the effects of extracellular histones. All animal studies were conducted according to the European equivalent of the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the local governmental authorities.

Assessment of Renal Pathology

Renal sections of 2 μm were stained with periodic acid–Schiff reagent. Glomerular abnormalities were scored in 50 glomeruli per section by a blinded observer. The following criteria were assessed in each of the 50 glomeruli and scored as segmental or global lesions if <50% or >50% of the glomerular tuft was affected by focal necrosis and capsule adhesions. Cellular crescents were assessed separately when more than a single layer of PECs were present around the inner circumference of Bowman’s capsule. Immunostaining was performed as described using the following primary antibodies: for WT-1/nephrin, neutrophils (Serotec, Oxford, UK), Mac-2 (Cedarlane, ON, Canada), TNF-α (Abcam, Inc., Cambridge, UK), and fibrinogen (Abcam, Inc.). Stained glomerular cells were quantified in 50 glomeruli per section.

Electron Microscopy

Kidney tissues and endothelial cell monolayers were fixed in 2.0% paraformaldehyde/2.0% glutaraldehyde, in 0.1 M sodium phosphate buffer, pH 7.4 for 24 hours, followed by three washes×15 minutes in 0.1 M sodium phosphate buffer, pH 7.4, and distilled water. For transmission electron microscopy, kidneys were postfixed in phosphate cacodylate-buffered 2% OsO4 for 1 hour, dehydrated in graded ethanols with a final dehydration in propylene oxide, and embedded in Embed-812 (Electron Microscopy Sciences, Hatfield, PA). Ultrathin sections (approximately 90-nm thick) were stained with uranyl acetate and Venable’s lead citrate. For scanning electron microscopy, after rinsing in distilled H₂O, cells on coverslips were treated with 1% thiocarbohydrazide, postfixed with 0.1% osmium tetroxide, dehydrated in ethanol, mounted on stubs with silver paste, and critical-point dried before being sputter coated with gold/palladium. Specimens were viewed with a JEOL model 1200EX electron microscope (JEOL, Tokyo, Japan).

Immunohistochemistry in Human Tissues

Formalin-fixed paraffin-embedded sections of renal biopsies from five participants with ANCA-positive RPGN, newly diagnosed in 2013, were drawn from the files of the Institute of Pathology at the Ludwig-Maximilians-University of Munich. Informed consent was obtained from all participants before renal biopsy. The renal biopsies were fixed in 4% PBS-buffered formalin solution and embedded in paraffin. Biopsies contained normal glomeruli and glomeruli exhibiting cellular, fibrocellular, or fibrous crescents. Controls consisted of normal kidney tissue from tumor nephrectomies. TLR2 and TLR4 expression was assessed by using specific antibodies (TLR2; LS Bio, Seattle, WA; and TLR4; Novus, Littleton, CO).

In Vitro Models

Cytotoxicity Assay

Mouse glomerular endothelial cells,⁴⁸ podocytes⁴⁹, and PECs⁵⁰ were cultured in 96-well plates with RPMI media without FCS and penicillin-streptomycin and allowed to adhere overnight. The cells were stimulated with the different concentrations of total calf thymus histones (10, 20, 30, 40, 50, and 100 µg/ml) with or without histone antibody for another 18–20 hours. The possibility of contaminants like bacterial endotoxin or other immunostimulatory contaminants of this histone preparation was ruled out as previously reported.²¹ Cytotoxicity assay was performed using Promega CellTiter 96 nonradioactive cell proliferation assay (MTT Assay Kit; Promega, Mannheim, Germany). Glomerular cells were also incubated with histones with or without antihistone IgG, heparin, and/or aPC. Lactate dehydrogenase assay using cytotoxicity detection kit (Roche Diagnostics, Mannheim, Germany) was used to assess cell death.

Podocyte Detachment Assay

Podocytes were grown at 33°C using modified RPMI media in the presence of IFN-γ in collagen-coated 10-cm dishes and 8×10⁴ cells were seeded and allowed to differentiate as podocytes at 37°C for 2 weeks in collagen plates without IFN-γ. Once the monolayers of podocytes were differentiated, the cells were treated with either histones or GBM antiserum with or without histone antibody and allowed for 18 hours. Detached cells that were present in supernatant were manually counted using a hemocytometer. Adhered cells were trypsinized and counted manually to calculate the percentage of cells detached.

In Vitro Tube Formation Assay

Matrigel was thawed overnight at 4°C to make it liquid. The addition of 10 µl in each well of µ-slide angiogenesis (ibidi, Munich, Germany) was allowed to solidify at 37°C. Glomerular endothelial cells were seeded at 1×10⁴ cells/well and stimulated with vascular endothelial growth factor and b-FGF as positive control or with histones with or without histone antibody. Tube formation as a marker of angiogenesis was assessed by light microscopy by taking series of pictures at 0, 4, 8, and 24 hours.⁵¹

NETosis Assay

Neutrophils were isolated from healthy mice by dextran sedimentation and hypotonic lysis of red blood cells. NETs were induced in vitro by adding TNF-α (Immunotools, Friesoythe, Germany) or PMA (Sigma-Aldrich, MO, USA) for 12 hours in with or without histone antibody. Later, the endothelial cell death was assessed by performing an MTT assay and immunofluorescence staining for histones (BWA3 clone), neutrophil elastase (Abcam, Inc.), and 4′,6-diamidin-2-phenylindol (Vector Laboratories, Burlingame, CA) after fixing with acetone.

BMDCs and J774 Macrophages

Bone marrow cells were isolated from healthy mouse and plated at 1×10⁶ cells per well and differentiated into BMDCs in the presence of GM-CSF (Immunotools). J774 macrophage cells were grown in RPMI media, plated at 1×10⁶ cells per well, and stimulated with different doses of histones with or without histone antibody for 18 hours. Supernatants were collected for TNF-α (BioLegend, San Diego, CA) and IL-6 ELISA (BD Biosciences, San Diego, CA). Flow cytometry for the activation markers MHC II, CD40, CD103, and CD86 (BD Biosciences) was also performed.

Flow Cytometry

Flow cytometric analysis of cultured and renal immune cells was performed on a FACS Calibur flow cytometer (BD Biosciences) as described.⁵² Kidney cell isolate was incubated with binding buffer containing either anti-mouse CD11c, CD11b, CD103, F4/80, and CD45 antibodies (BD Biosciences) for 45 minutes at 4°C in dark to detect renal mononuclear phagocyte populations. Anti-CD86 (BD Biosciences) was used as activation marker. Anti-CD3 and CD4 (BD Biosciences) were used to identify the respective T cell population.

RNA Preparation and Real-Time RT-PCR

RT and real-time RT-PCR from total renal RNA was prepared as described.⁵³ The SYBR Green Dye detection system was used for quantitative real-time PCR on a Light Cycler 480 (Roche). The following respectively reverse and forward gene-specific primers (300 nM; Metabion, Martinsried, Germany): 18s, AGGGCCTCACTAAACCATCC and GCAATTATTCCCCA TGAACG; TNF-α, CCACCACGCTCTTCTGTCTAC and AGGGTCTGGGCCATAGAACT; Fibrinogen (FGL-2), AGGGGTAACTCTGTAGGCCC and GAACACATGCAGTCACAGCC; WT-1, CATCCCTCGTCTCCCATTTA and TATCCGAGTTGGGGAAATCA; and CD44, AGCGGCAGGTTACATTCAAA and CAAGTTTTGGTGGCACACAG. Controls consisting of ddH₂O were negative for target and housekeeper genes.

Statistical Analyses

Data were expressed as the mean±SEM. Comparison between groups was performed by the two-tailed t test or ANOVA. A value of P<0.05 was considered statistically significant. All statistical analyses were calculated using GraphPad Prism software (GraphPad).

Disclosures

None.