Decay-accelerating factor induction by tumour necrosis factor-α, through a phosphatidylinositol-3 kinase and protein kinase C-dependent pathway, protects murine vascular endothelial cells against complement deposition

Abstract

We have shown that human endothelial cells (EC) are protected against complement-mediated injury by the inducible expression of decay-accelerating factor (DAF). To understand further the importance of DAF regulation, we characterized EC DAF expression on murine EC in vitro and in vivo using a model of glomerulonephritis. Flow cytometry using the monoclonal antibody (mAb) Riko-3 [binds transmembrane- and glycosylphosphatidylinositol (GPI)-anchored DAF], mAb Riko-4 (binds GPI-anchored DAF) and reverse transcription–polymerase chain reaction (RT–PCR), demonstrated that murine EC DAF is GPI-anchored. Tumour necrosis factor-α (TNF-α) increased EC DAF expression, detectable at 6 hr and maximal at 24–48 hr poststimulation. DAF upregulation required increased steady-state DAF mRNA and protein synthesis. In contrast, no increased expression of the murine complement receptor-related protein-Y (Crry) was seen with TNF-α. DAF upregulation was mediated via a protein kinase C (PKC)α, phosphoinositide-3 kinase (PI-3 kinase), p38 mitogen-activated protein kinase (MAPK) and nuclear factor-κB (NF-κB)-dependent pathway. The increased DAF was functionally relevant, resulting in a marked reduction in C3 deposition following complement activation. In a nephrotoxic nephritis model, DAF expression on glomerular capillaries was significantly increased 2 hr after the induction of disease. The demonstration of DAF upregulation above constitutive levels suggests that this may be important in the maintenance of vascular integrity during inflammation, when the risk of complement-mediated injury is increased. The mouse represents a suitable model for the study of novel therapeutic approaches by which vascular endothelium may be conditioned against complement-mediated injury.

Introduction

The complement cascade plays a central role in defence against infection and in the modulation of inflammatory responses.¹ In order to prevent bystander injury to host tissues following complement activation, a variety of soluble and membrane-bound complement regulatory proteins have evolved. These include the cell-surface proteins decay-accelerating factor (DAF, CD55), membrane cofactor protein (MCP, CD46), protectin (CD59) and complement receptor 1 (CR-1, CD35). DAF acts to prevent the formation and accelerate the decay of C3 and C5 convertases, the central amplification enzymes at the proximal end of the complement cascade.² MCP acts as a cofactor to Factor I in the cleavage and degradation of C3b, whilst CD59 acts distally to prevent the assembly of the C5b-9 membrane attack complex (MAC).^3,4 In addition, murine cells express complement receptor-related protein-Y (Crry), which combines the functions of DAF and MCP.^5,6 The importance of these regulatory proteins is well illustrated by the clonal disorder paroxysmal nocturnal haematuria, in which an acquired absence of DAF and CD59 on a subpopulation of erythrocytes renders them prone to complement-mediated lysis.⁷

In humans, there is a single DAF gene located on the long arm of chromosome 1.² In contrast, the mouse has two DAF genes (Daf1 and Daf2) located in tandem on chromosome 1 and these encode a glycosylphosphatidylinositol (GPI)-anchored form of DAF (GPI-DAF) and a transmembrane protein (TM-DAF), respectively.^8–10 However, recent work has demonstrated that, via alternative splicing, each gene has the potential to make both forms of murine DAF.^11,12 Further detailed analyses have established that the two forms of murine DAF are differentially expressed, with TM-DAF confined to the spleen and the testis.^8,11–13 In contrast, GPI-DAF is more widely distributed, in a pattern similar to human DAF, with expression found predominantly on haematopoietic, epithelial and endothelial cells in a variety of organs, including heart, lung, brain, liver, kidney and intestine.¹³

The generation of DAF-deficient mice has begun to provide insight into the function of DAF in vivo. Although DAF-deficient erythrocytes are more susceptible to induced complement-mediated lysis,¹⁴ Crry appears to be more important than either DAF or CD59 in the regulation of spontaneous complement activation on murine erythrocytes.¹⁵ Notwithstanding this, DAF has recently been shown to play a critical role in cytoprotection in two models of glomerulonephritis, with DAF-deficient mice significantly more susceptible to disease than control animals.^16,17 These data are consistent with our in vitro observations that DAF expression on the surface of human endothelial cells (EC) is induced by tumour necrosis factor-α (TNF-α), interferon-γ (IFN-γ), vascular endothelial growth factor (VEGF), basic fibroblast growth factor (bFGF) and thrombin, thus potentially providing enhanced cytoprotection in a variety of inflammatory and thrombotic situations against complement-mediated lysis.^18–21

In this study, we provide evidence that DAF expression is inducible on the surface of murine EC and demonstrate a functional role for this response in the protection of EC against complement activation. Using an in vivo model of immune complex-mediated nephritis we also demonstrate, for the first time, an increase in glomerular DAF expression in the face of ongoing inflammation.

Materials and methods

Monoclonal antibodies (mAbs) and other reagents

The following anti-DAF mAbs were used: hamster anti-mouse DAF immunoglobulins Riko-1, Riko-2, Riko-3 (DAF-GPI and DAF-TM specific), Riko-4 (DAF-GPI specific)²² and rat anti-mouse DAF MD1.¹³ mAb MJ7/18, rat anti-mouse endoglin, was obtained from the Developmental Studies Hybridoma Bank, University of Iowa (Iowa City, IA) and anti-Crry/p65 mAb 1F2 was from BD PharMingen (San Diego, CA). Protein kinase C (PKC) antagonists Gö6976 and GF109203X were from Calbiochem (Nottingham, UK). PKCβ specific inhibitor LY379196²³ was a gift from Dr K. Ways, Eli Lilly (Indianapolis, IN). Myristoylated PKC peptide inhibitor (myr-ψPKC) (myr-Arg-Phe-Ala-Arg-Lys-Gly-Ala-Leu-Arg-Gln-Lys-Asn-Val) was from Promega (Madison, WI). The p38 mitogen-activated protein kinase (MAPK) inhibitor (SB202190), nuclear factor-κB (NF-κB) inhibitor [proteasome inhibitor-1 (PSI)] and MEK-1 inhibitors (PD98059 and UO126) were from Calbiochem. Phosphoinositide-3 kinase (PI-3 kinase) inhibitors LY290042 and wortmannin were from Biomol (Plymouth Meeting, PA). Anti-PKC isozyme antibodies were from Transduction Laboratories (Lexington, KY). Rabbit anti-phospho PKCα was from Upstate Biotech (Lake Placid, NY). Recombinant human and murine TNF-α, IFN-γ, and interleukin (IL)-1α and -β, were from Pepro Tech (London, UK). Cycloheximide, actinomycin D and phosphatidylinositol-specific phospholipase C (PIPLC) were purchased from Sigma-Aldrich (Poole, UK). Normal mouse serum (NMS) was purchased from DAKO (Glostrup, Denmark), aliquoted and frozen at −70° prior to use. NMS serum (10–50%) was prepared fresh in Dulbecco's modified Eagle's minimal essential medium (DMEM) (Gibco BRL Life Technologies, Paisley, UK), without heparin, for each experiment (DAKO). In addition, sera from wild-type C57BL/6 mice and mice deficient in C1q (on a C57BL/6 background) were a kind gift from Dr M. Botto (Imperial College London, London, UK).

Animals

C57BL/6 mice were purchased from Harlan Olac (Bicester, Oxon, UK). Mice deficient in PKCβ²⁴ and PKCε²⁵ (on a C57BL/6 background), and H-2K^b-tsA58 transgenic mice (CBA/Ca × C57BL/10 background),²⁶ were bred in house. All mice were housed under controlled climatic conditions in filter-topped microisolator cages with autoclaved bedding. Irradiated food and drinking water were readily available. All animals were housed and studied according to United Kingdom Home Office guidelines. Sentinel mice were housed alongside test animals and screened regularly for a standard panel of murine pathogens.

Endothelial cell isolation and culture

The conditionally immortalized murine cardiac EC (MCEC) were isolated and cultured as previously described.²⁷ Similar EC lines were also isolated from PKCβ^−/− and PKCε^−/− mice. Knockout mice were first crossed with H-2K^b-tsA58 transgenics and the offspring crossed to generate H-2K^b-tsA58/PKCβ^−/− and H-2K^b-tsA58/PKCε^−/− mice. Cardiac EC were then obtained from these mice and from PKCβ^+/+ and PKCε^+/+ littermate controls, using a method previously described in detail.²⁷ MCEC were cultured in DMEM (Gibco), supplemented with 10% fetal bovine serum (FBS) (Helena Biosciences, Sunderland, UK), 100 IU/ml penicillin, 0·1 mg/ml streptomycin, 2 mm l-glutamine (Gibco), 10 U/ml heparin (Leo Laboratories, Prince Risborough, UK) and 30 µg/ml EC growth factor (Sigma) and cultured in 1% gelatin-coated tissue-culture flasks.

Flow cytometry

EC were analysed by flow cytometry using an Epics XL-MCL flow-cytometer (Coulter, Hialeah, FL), as previously described.¹⁸ In experiments involving pharmacological antagonists, each inhibitor was added 1 hr prior to TNF-α. In some experiments the results are expressed as the relative fluorescent intensity (RFI), which represents the mean fluorescent intensity (MFI) with test mAb divided by the MFI obtained using an isotype-matched irrelevant mAb. Cell viability was assessed by examination of EC monolayers prior to staining using phase-contrast microscopy, cell counting and estimation of Trypan blue or propidium iodide (PI) exclusion. In some experiments, EC, stained as described above, were examined by confocal microscopy using a Carl Zeiss LSM5 PASCAL laser-scanning microscope, and the fluorescence was recorded using pascal, version 3·2, software (Zeiss, Welwgn Garden City, UK).

Northern blotting

RNA was extracted from EC by passage of guanidium isothiocyanate lysates over RNEasy columns (Qiagen Ltd, Crawley, UK) and Northern blot analysis was performed, as previously described.¹⁸ The probe for GPI-DAF was a gift from Dr A. Spicer (University of California, Davis, CA).⁸ Blots were quantified by an Appligene Image Analysis System and densitometry was performed with image programme 1.52 software (National Institutes of Health, Bethesda, MD). Values were corrected with respect to ethidium bromide-stained rRNA-loading patterns and an arbitary value of 1 was assigned to unstimulated EC.

Reverse transcription–polymerase chain reaction (RT–PCR)

RT–PCR for GPI- and TM-DAF was performed as described previously.¹¹ Purified RNA was reverse transcribed according to standard protocols, using an RT–PCR kit (Sigma). The same sense primer was used for both GPI and TM isoforms (5′-CAATGGAATAATGCGAGGGG-3′), whilst separate antisense sequences were employed for GPI-DAF (5′-TGTATCCATTCTTCTTGGACA-3′) and TM-DAF (5′-GTTGAAAAGGTGGAGACTGG-3′). PCR reactions were carried out, in a total volume of 25 µl, using 1 µl of cDNA, 2·5 U of Taq DNA polymerase, 500 µm dNTPs and 10 pmol of each primer. Amplification was performed in a Biometra thermocycler for 30 cycles (94° for 1 min, 57° for 45 seconds, 72° for 2 min) with a final extension at 72° for 15 min.

Western blotting

Unstimulated or TNF-α-treated EC were lysed in 4 mm EDTA, 50 mm Tris–HCl, pH 7·4, in 150 mm NaCl with 25 mm sodium deoxycholic acid, 200 µm sodium orthovanadate, 10 mm sodium pyrophosphate, 100 mm sodium fluoride, 1% Triton-X-100, 1 mm phenylmethylsulphonyl fluoride and 5% protease inhibitor cocktail (Sigma). Following sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS–PAGE), the separated proteins were transferred to Immobilon-™P membranes (Millipore Corporation, Bedford, MA) which were blocked for 1 hr, at room temperature, in Tris-buffered saline (TBS)/0·1% Tween-20 (vol/vol) with 5% milk powder (wt/vol). The membrane was incubated with relevant primary Abs overnight at 4°, washed three times in TBS/0·1% Tween-20 and incubated for 1 hr at room temperature with an appropriate peroxidase-labelled secondary Ab, developed with an enhanced chemiluminescence substrate (Amersham Pharmacia Biotech, Little Chalfont, UK) and exposed to autoradiography film. Equal loading of lanes was confirmed by estimation of lysate protein content using the Bio-Rad D_c protein assay (Bio-Rad, Hercules, CA) and by Ponceau S staining of membranes prior to analysis.

C3-binding and cell-lysis assays

The methods used for detection of cell-surface C3 have been previously described.^18,19 EC were cultured in the presence and absence of TNF-α (10 ng/ml) for 24 hr, opsonized with the anti-endoglin mAb (MJ7/18) and polyclonal rabbit anti-rat immunoglobulin for 30 min and incubated with 50 µl of 10–30% NMS in DMEM for 3 hr at 37°, prior to analysis of C3 deposition by flow cytometry using fluorescein isothiocyanate (FITC)-labelled anti-mouse C3 (ICN Biomedicals, Irvine, CA). During the inhibition studies, the anti-DAF inhibitory mAb, Riko-1, or control mAbs were added to the assay to achieve a final concentration of 10 µg/ml. To estimate complement-mediated cell lysis, EC pretreated with TNF-α or medium alone for 48 hr were harvested with trypsin/EDTA, washed, opsonized as described above and incubated with 5–20% NMS in veronal-buffered saline for 60 min at 37°. Following further washing, EC were resuspended in phosphate-buffered saline (PBS), PI (Sigma) was added to a final concentration of 5 µg/ml and EC were analysed by flow cytometry using the FL2 channel. Lysis was calculated, in triplicate samples, as the number of PI-positive cells expressed as a percentage of the total number of cells.

Induction of glomerulonephritis and quantitative immunofluorescence

A heterologous nephrotoxic nephritis model was used in wild-type C57BL/6 mice. Nephrotoxic globulin was prepared as previously described.²⁸ Mice were injected intravenously with 4 mg of rabbit nephrotoxic globulin and then anaesthetized and exsanguinated at selected time-points prior to harvesting the kidneys. For light microscopy, kidneys were fixed for 4 hr in Bouin's solution, transferred to 70% ethanol and paraffin embedded. For immunofluorescence, samples were snap-frozen in isopentane and stored at −70°. Cryostat sections (5 µm) were cut, mounted on glass slides and fixed for 10 min in acetone, washed twice for 5 min in PBS and the primary antibody Riko-4 (GPI-DAF specific) was added at 10 µg/ml for 1 hr at room temperature. After further washing, FITC-labelled F(ab′)₂ goat anti-hamster IgG (Stratech, Luton, UK) was added for 1 hr at room temperature. This antibody is solid-phase adsorbed to prevent cross-reactivity with immunoglobulins from other species, including rabbit and mouse. Sections were then washed and mounted in Permafluor (Shandon, Pittsburgh, PA). Image analysis was performed as described previously²⁸ with the MFI of 25 glomeruli per section being expressed in arbitary fluorescence units (AFU).

Statistics

Differences between the results of experimental treatments were evaluated by analysis of variance (anova), with Newman-Keuls multiple comparison test, using graphpad 3.0 software (GraphPad Software, San Diego, CA). Differences were considered significant at P-values of <0·05.

Results

Expression of DAF on murine EC

Temperature-sensitive conditionally immortalized MCEC were isolated from H-2K^b-tsA58 transgenic mice.²⁷ Cells were expanded at the permissive temperature (33°), before switching to non-permissive conditions (38°) for experimentation. When cultured at 38°, MCEC do not express the SV40 large T antigen and are growth factor dependent.²⁷

As seen in Fig. 1(a), DAF was expressed constitutively on >90% of cardiac EC cultured at 38°, in a unimodal pattern. Experiments comparing mAb Riko-3 (binds GPI-DAF and TM-DAF) and mAb Riko-4 (binds GPI-DAF only)²² demonstrated that both mAbs bound resting EC equally well, suggesting that murine EC DAF is predominantly GPI-anchored, as indicated by immunocytochemistry in mice deficient in GPI-DAF.¹³ RFI ± SD values with Riko-3 were 4·8 ± 1·2 and with Riko-4 were 5·2 ± 1·8 (n = 3). Similar experiments performed with both primary MCEC and conditionally immortalized lung EC showed the same pattern of expression (data not shown). Furthermore, treatment of EC with PIPLC for 30 min at 37° significantly reduced DAF expression, also suggesting that the molecule is GPI-anchored (Fig. 1b). However, some residual binding of both mAbs was observed, raising the possibility of some PIPLC resistance. The specificity of PIPLC for GPI-anchored proteins alone was confirmed by its failure to alter surface expression of the TM-anchored protein, endoglin (data not shown).

Figure 1.

Constitutive expression of decay-accelerating factor (DAF) on mouse cardiac endothelial cells (MCEC). (a) DAF expression was analysed by flow cytometry using monoclonal antibody (mAb) Riko-3. Background fluorescence is shown by the black histogram and DAF expression by the white histogram. (b) DAF expression in the absence (hatched bars) or presence (white bars) of phosphatidylinositol-specific phospholipase C (PIPLC) was assessed by flow cytometry using mAbs Riko-3 [binds glycosylphosphatidylinositol (GPI)-DAF and a transmembrane protein (TM)-DAF)] or Riko-4 (binds GPI-DAF). Bars represent relative fluorescence intensity (RFI) ± standard deviation (SD) (n = 3), derived by dividing the mean fluorescence intensity (MFI) value obtained with test mAb by the MFI value obtained using isotype-matched irrelevant mAb. The figures are representative of three replicate experiments. *P < 0·05.

DAF, but not Crry, expression is induced by TNF-α on murine EC

To investigate the regulation of DAF, MCEC were treated with pro-inflammatory cytokines TNF-α, IFN-γ, IL-1α or IL-1β. The concentration ranges of cytokines used were based upon those previously demonstrated to induce the expression of vascular cell adhesion molecule-1 (VCAM-1) and intracellular adhesion molecule-1 (ICAM-1).²⁷ No significant change in DAF expression was observed following treatment with IFN-γ, IL-1α or IL-1β (data not shown). However, a dose-dependent increase in DAF expression was seen following treatment with TNF-α (Fig. 2a). A significant increase was detected at 6 hr poststimulation, with peak expression at 12 hr, which was maintained up to 48 hr (Fig. 2b). In 10 separate experiments, TNF-α upregulated DAF levels by up to fourfold higher than the basal level of unstimulated EC (the RFI ± SD of unstimulated EC was 4·19 ± 1·5 and of TNF-α-stimulated EC was 12·7 ± 5·2; P < 0·01). Experiments were also performed with primary cardiac EC, and these demonstrated a comparable increase in DAF following treatment with TNF-α, suggesting that the conditionally immortalized MCEC respond normally to cytokines, as we have previously reported.²⁷ Study of the immunofluorescence staining pattern of DAF showed significant upregulation by TNF-α and even distribution on the cell surface (Fig. 2d).

Figure 2.

Upregulation of decay-accelerating factor (DAF) expression following treatment with tumour necrosis factor (TNF-α). (a) Endothelial cells (EC) were incubated for 48 hr with TNF-α at the concentrations shown and analysed by flow cytometry using the monoclonal antibody (mAb) Riko-3. Data are expressed as relative fluorescence intensity (RFI) ± standard deviation (SD) (n = 3). (b) EC were treated with TNF-α (10 ng/ml) for up to 48 hr and DAF expression was analysed by flow cytometry and RFI ± SD (n = 3). (c) Lysates, prepared from TNF-α-treated (48 hr) or unstimulated (US) EC, were separated by sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS–PAGE) and transblotted to nitrocellulose. Blots were probed with mAb Riko-3 or Riko-4. (d) EC cultured in the absence (US) or presence of TNF-α for 48 hr (TNF) were analysed by confocal microscopy for DAF expression using the mAb, MD1. The figures are representative of three replicate experiments. *P < 0·05, **P < 0·01.

Costimulation of EC with combinations of TNF-α and IL-1β, or TNF-α and IFN-γ, showed no significant increase above stimulation with TNF-α alone (data not shown). Further analysis was performed using Western blotting. Following treatment with TNF-α for 48 hr, cell lysates were prepared and Western blots probed with mAbs Riko-3 or Riko-4. DAF, detectable as single band of 60 000 molecular weight (MW) by both mAbs, was weakly expressed by unstimulated EC and upregulated following treatment with TNF-α (Fig. 2c).

To investigate Crry regulation on MCEC, we compared unstimulated and TNF-α-stimulated EC and analysed Crry expression by flow cytometry using mAb 1F2. Crry was constitutively expressed with unimodal distribution on >95% of resting murine EC (RFI ± SD = 17·2 ± 0·8, n = 3). However, in contrast to DAF, no change in the expression of Crry was detected following stimulation with TNF-α for up to 48 hr (the RFI ± SD of unstimulated EC was 17·9 ± 1 and of TNF-α-stimulated EC was 18·5 ± 0·5; n = 3). In addition, no change in Crry expression following treatment of EC for up to 48 hr with either IL-1β or IFN-γ was seen. These data suggest that the upregulation of DAF may have a particular role in protecting EC against C3 deposition under inflammatory conditions.

TNF-α-induced DAF on EC requires gene transcription and de novo protein synthesis

To determine whether the observed increase in DAF was dependent upon gene transcription, MCEC were pretreated with actinomycin D prior to the addition of TNF-α. The presence of actinomycin D completely inhibited the TNF-α-induced increase in DAF expression (RFI ± SD of TNF-α-stimulated EC was 9·9 ± 2·8 and of actinomycin D-treated EC was 4·2 ± 1·4, n = 3). Cytotoxic effects of actinomycin D were excluded by parallel staining with PI, which showed that >90% of the EC were viable. To investigate changes in DAF steady-state mRNA following treatment of MCEC with TNF-α, RT–PCR and Northern blotting were employed. As seen in Fig. 3(a), a transcript of ≈600 bp, corresponding to GPI-DAF, was detected in MCEC treated for 8 hr with TNF-α. No PCR product was obtained from stimulated or unstimulated EC using primers specific for TM-DAF (data not shown). Northern blot analysis, using a GPI-DAF-specific probe, confirmed the presence of GPI-DAF in MCEC treated with TNF-α (Fig. 3b). A GPI-DAF mRNA transcript (≈3·1 kb) was detected at a low level in unstimulated EC. Treatment with TNF-α led to an increase in the 3·1-kb DAF mRNA transcript that was first detectable at 2 hr and was maximal at 4 hr poststimulation. In addition, a second minor GPI-DAF mRNA transcript (≈2·8 kb) was detectable following longer exposure (data not shown). Quantification of mRNA levels in resting and TNF-α-treated EC demonstrated a threefold increase above baseline 4 hr poststimulation (data not shown).

Figure 3.

Decay-accelerating factor (DAF) upregulation requires increased steady-state mRNA. (a) Total RNA was extracted from unstimulated and tumour necrosis factor-α (TNF-α)-treated endothelial cells (EC) and from murine testis (positive control). cDNA was prepared by reverse transcription using oligonucleotides specific for glycosylphosphatidylinositol (GPI)-DAF. Lane 1, reverse transcription–polymerase chain reaction (RT–PCR) kit control; lane 2, unstimulated EC; lane 3, EC stimulated with TNF-α for 8 hr; lane 4, murine testis. (b) Total RNA was isolated from unstimulated and TNF-α-treated EC, and Northern blots were prepared and probed with a GPI-DAF-specific probe. Lane 1, unstimulated; lane 2, TNF-α, 2 hr; lane 3, TNF-α, 4 hr; lane 4, TNF-α, 6 hr.

To investigate the role of protein synthesis, MCEC were treated with cycloheximide prior to the addition of TNF-α. Cycloheximide led to a complete abrogation of TNF-α-induced DAF (the RFI ± SD of TNF-α-stimulated EC was 9·1 ± 2·6 and of cycloheximide-D treated EC was 4·0 ± 1·5; n = 3). Therefore, these results suggest that the upregulation of DAF expression, following stimulation of MCEC with TNF-α, is associated with gene transcription, increased steady-state DAF mRNA and de novo protein synthesis.

TNF-α-induced DAF expression on MCEC is PKC-dependent

We have previously identified distinct PKC-dependent and -independent pathways for the regulation of DAF in human EC.^18–20 Western blotting analysis revealed that PKCα, δ, ε, θ and ζ were expressed in MCEC (Fig. 4a) and RT–PCR additionally demonstrated the presence of PKCβ (data not shown). To investigate the role of PKC in TNF-α-induced DAF expression, we initially used GF109203X, a pharmacological inhibitor of the classical and novel PKC isoenzymes.²⁹ Pretreatment with GF109203X significantly inhibited TNF-α-induced DAF upregulation (Fig. 4b). Subsequent experiments used Gö6976, a specific antagonist for the classical PKC isozymes PKCα and PKCβ.³⁰ In addition, we also tested the effect of LY379196, a selective PKCβ inhibitor.²³ Whilst Gö6976 inhibited TNF-α-induced DAF expression, LY379196 had no significant effect (Fig. 4b), suggesting that PKCα was the principal isoenzyme involved in TNF-α-induced DAF expression. This interpretation was supported by the use of a specific cell-permeable peptide antagonist of PKCα/β, myr-ψPKC,³¹ which also inhibited TNF-α-induced DAF upregulation (Fig. 4b).

Figure 4.

Analysis of protein kinase C (PKC) in endothelial cell (EC) decay-accelerating factor (DAF) upregulation. (a) Lysates were prepared from EC, separated by sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS–PAGE), transblotted to nitrocellulose and probed with PKC isoenzyme-specific antibodies. (b) EC were left unstimulated (US) or treated for 48 hr with tumour necrosis factor-α (TNF-α) (10 ng/ml) in the presence or absence of PKC inhibitors GF109203X (GF) (3 µm), Gö6976 (Go) (5 µm), PKCβ inhibitor LY397196 (LY) (60 nm) or the peptide inhibitor myr-ψPKC (PEP) (100 µm). EC were analysed by flow cytometry for DAF expression using monoclonal antibody (mAb) MD-1. The data are expressed as a percentage of the DAF expression on TNF-α-treated cells [mean ± standard error of the mean (SEM) from three separate experiments]. *P < 0·05, **P < 0·01. (c) DAF expression on PKCβ^−/−, PKCε^−/− and wild-type mouse cardiac endothelial cells (MCEC), unstimulated (US) or treated with TNF-α 10 ng/ml for 48 hr (TNF) and analysed by flow cytometry. Data are expressed as relative fluorescence intensity (RFI) ± standard deviation (SD) (n = 3). (d) Lysates were prepared from EC treated with TNF-α for up to 60 min. Blots were probed with anti-PKCα Ab C-20 and an anti-phospho-PKCα Ab. Lane 1, unstimulated; lane 2, TNF-α, 10 min; lane 3, TNF-α, 20 min; lane 4, TNF-α, 30 min; lane 5, TNF-α, 60 min. The figures are representative of three replicate experiments.

Further studies, using EC lines generated from mice deficient in PKCβ or the novel isoenzyme PKCε, demonstrated that treatment of the knockout EC with TNF-α resulted in upregulation of DAF equivalent to that in matched wild-type EC (Fig. 4c). It was also of note that whilst GF109203X significantly reduced the constitutive expression of DAF, no such decrease was seen with Gö6976, LY379196 or myr-ψPKC. This suggests that constitutive expression is regulated by another PKC isozyme expressed by EC. Finally, to demonstrate the ability of TNF-α to activate PKCα, use was made of an antibody specific for the phosphorylated form of PKCα. Treatment of MCEC with TNF-α resulted in PKCα phosphorylation (Fig. 4d), reaching a maximum of threefold above baseline 60 min poststimulation, as quantified by densitometry (data not shown).

Role of PI-3K, p38 and p42/44 MAPK, and NF-κB in TNF-α-induced DAF expression

TNF-α is known to exert its effects on vascular EC through a variety of downstream signalling pathways, including PI-3K, p38 MAPK, ERK1/2 and NF-κB. To define the signalling pathways involved in the regulation of DAF expression, we used pharmacological inhibitors of PI-3K (LY290042 and wortmannin), p38 MAPK (SB202190), MEK-1 (UO126 and PD98059) and NF-κB (PSI). Inhibition of PI-3K significantly inhibited DAF upregulation (P < 0·05), implicating activation of PI-3K in the response. Furthermore, inhibition of p38 MAPK also abrogated TNF-α-induced DAF expression (P < 0·05), whilst in contrast, inhibition of the ERK1/2 pathway with either UO126 or PD98059 had no significant effect. Finally, a role for NF-κB was suggested by the inhibition of DAF induction by PSI (P < 0·05).

TNF-α-induced DAF reduces complement deposition on MCEC

The binding of complement factor C3 to MCEC was used to establish whether DAF induction resulted in increased cytoprotection against complement activation. MCEC, opsonized with the monoclonal anti-endoglin MJ7/18 and polyclonal rabbit anti-rat immunoglobulin, were incubated with 30% NMS for 2 hr, and C3 binding was quantified using FITC-labelled anti-mouse C3 immunoglobulin. Treatment of MCEC for 48 hr with TNF-α reduced the deposition of C3 on the EC surface by up to 70% (P < 0·01) (Fig. 5a). Parallel flow-cytometric analysis demonstrated that MCEC expression of endoglin was not significantly altered by TNF-α (the RFI ± SD of unstimulated MCEC was 11·9 ± 2·1 and of TNF-α-treated EC was 10·5 ± 1·2). To confirm the role of DAF in the reduction of C3 binding, the inhibitory mAb hamster anti-mouse DAF, Riko-1, which does not significantly fix complement, was included in the assay. This led to a modest increase in the binding of C3 to unstimulated, opsonized EC exposed to 30% NMS (Fig. 5b). In addition, the reduction in C3 binding to EC pretreated with TNF-α was significantly reversed in the presence of Riko-1 (Fig. 5b), whereas isotype-matched control mAbs had no effect.

Figure 5.

Tumour necrosis factor-α (TNF-α) protects mouse cardiac endothelial cells (MCEC) against C3 deposition. (a) EC treated with TNF-α (10 ng/ml) (hatched bars), or medium alone (white bars), for 48 hr were incubated with anti-endoglin monoclonal antibody (mAb) MJ7/18 and polyclonal rabbit anti-rat immunoglobulin for 30 min at 4°. EC were then washed in Hanks' balanced salt solution (HBSS)/1% bovine serum albumin (BSA) prior to the addition of normal mouse serum (NMS). Binding of C3 was detected by flow cytometry using fluorescein isothiocyanate (FITC)-conjugated goat anti-mouse C3. The data are shown as C3 binding [mean relative fluorescence intensity (RFI) ± standard deviation (SD), n = 3]. The negative control represents C3 binding in the presence of heat-inactivated mouse serum. (b) C3 binding (mean RFI ± SD, n = 3), following addition of 10% NMS, to unstimulated EC (white bars) and TNF-α-treated EC (hatched bars), in the presence or absence of the inhibitory decay-accelerating factor (DAF) mAb, Riko-1. (c) C3 binding (mean RFI ± SD, n = 3), following addition of 30% heat-inactivated mouse serum (HIMS), and 30% serum from wild-type C57BL/6 mice (WT) or C1q-deficient mice (C1q^−/−). (d) Unstimulated (○) and TNF-α-treated EC (•) were opsonized with mAb MJ7/18 and polyclonal rabbit anti-rat immunoglobulin prior to exposure to different concentrations of mouse serum for 60 min. Propidium iodide (PI) (5 µg/ml) was added to the cell suspension and analysis was by flow cytometry. Percentage EC lysis was calculated as the number of PI-positive cells expressed as a percentage of the total number of cells (mean ± SD, n = 3). The figures are representative of three similar experiments. *P < 0·05, **P < 0·01.

To address the question of whether the deposition of C3 on EC following complement activation is entirely dependent upon the activity of the classical pathway, we compared serum from wild-type mice and mice deficient in C1q. As seen in Fig. 5(c), an 80% reduction in the deposition of C3 on the EC surface was seen when serum from C1q-deficient mice was used in place of wild-type serum. However, the residual C3 deposition, above that seen with heat-inactivated serum, suggests that although activation of the classical pathway is the predominant source of C3 on the cell surface, activation of the alternative pathway may also contribute (Fig. 5c).

To assess the physiological relevance of the reduction in C3 binding seen in EC treated with TNF-α, MCEC were opsonized and exposed to NMS; endothelial lysis was subsequently measured by the uptake of PI. As seen in Fig. 5(d), pretreatment of EC with TNF-α for 48 hr was cytoprotective, significantly reducing cell lysis following exposure to 10–20% NMS. These observations suggest that the increased levels of cell-surface DAF, seen in response to the treatment of EC with TNF-α, provides additional cellular protection against complement-mediated injury.

Expression of DAF in an in vivo model of glomerulonephritis

To investigate DAF upregulation in vivo, we used a heterologous nephrotoxic nephritis model in which foreign antibody to the glomerular basement-membrane causes a neutrophil influx that peaks at 2 hr and resolves by 24 hr.³² In addition, C3 is deposited on the glomerular basement-membrane. The severity of nephritis is dependent upon local production of cytokines, including TNF-α,³³ and this model therefore represents a means by which the ability of TNF-α to induce DAF expression during inflammation in vivo can be studied. Analysis of GPI-DAF demonstrated that DAF was weakly detectable in the glomeruli of saline-treated control mice and markedly upregulated by 2 hr post-treatment, with expression returning to baseline at 24 hr (Fig. 6a). To reduce the risk of non-specific binding of the secondary antibody to immunoglobulin deposited in the glomeruli, a solid-phase adsorbed, FITC-labelled F(ab′)₂ goat anti-hamster IgG was used. No significant binding of this antibody alone to glomerular sections from diseased mice was seen at 0, 2 or 24 hr post-treatment, and the 2-hr negative control is shown in Fig. 6(a). Quantification by measurement of AFU showed a significant increase at 2 hr (mean ± SD AFU = 70·7 ± 6·1) versus controls (mean AFU = 45·6 ± 4·2), with return to baseline at 24 hr (mean AFU = 43·4 ± 4·2) (Fig. 6b). These data demonstrate significant upregulation of DAF during an inflammatory response in vivo, suggesting that the expression of DAF above constitutive levels may provide enhanced protection against C3 deposition during inflammation.

Figure 6.

Analysis of glomerular decay-accelerating factor (DAF) expression during nephritis. (a) Wild-type C57BL/6 mice were treated by intravenous injection of rabbit anti-mouse glomerular basement membrane to induce heterologous nephrotoxic nephritis. Kidneys harvested from untreated mice and at 2 and 24 hr post-treatment (n = 5 per group) were stained with monoclonal antibody (mAb) Riko-4 followed by fluorescein isothiocyanate (FITC)-labelled goat anti-hamster IgG or with FITC-labelled goat anti-hamster IgG alone (negative control). (b) Quantification of fluorescence intensity of 25 glomeruli, expressed as arbitary fluorescence units (AFI) ± SD. **P < 0·01.

Discussion

We have previously shown that DAF expression on human vascular EC may be regulated via distinct agonist-specific pathways.^18–21 This led us to propose that the induction of DAF may be important in the maintenance of vascular integrity during inflammation, thrombosis and angiogenesis. However, although the role of DAF is well established in vitro, its function in vivo remains to be fully established.

The study of murine DAF is complicated in two ways. First, there are two homologous genes for DAF, in contrast to only one in humans. Murine GPI-DAF is widely distributed in a pattern similar to human DAF, whereas TM-DAF expression is largely restricted to the testis and splenic dendritic cells.^11–13 Second, rodents express Crry, which is an additional complement regulatory protein that combines the functions of human DAF and MCP.^5,6 Moreover, in addition to functional overlap, DAF and Crry are co-expressed in many tissues,³⁴ leading to questions about their relative importance. The initial analysis of gene-targeted mice revealed that DAF-deficient mice were fertile, whilst Crry knockout mice were non-viable owing to complement-mediated injury to the embryo.³⁵ This suggested that Crry played the dominant role in the control of murine complement activation. However, recent studies have demonstrated a distinct role for murine DAF in the protection of glomeruli against autologous complement-mediated damage in vivo, above and beyond that afforded by Crry.^16,17

In the present study, we found that DAF was constitutively expressed on murine cardiac EC under basal culture conditions. The binding patterns of the anti-DAF mAbs Riko-3 and Riko-4,²² and the effect of PIPLC, suggested that GPI-DAF predominates on EC. This was confirmed by RT–PCR, and Northern and Western blotting, and is consistent with the absence of DAF on the vascular endothelium of mice deficient in GPI-DAF.¹³

Vascular endothelium is continually exposed in vivo to low-level alternative and classical complement pathway activation through the ‘tick-over’ pathways,^36,37 and this may be increased following activation of the classical or alternative pathways during inflammation.³⁸ Therefore, induction of DAF expression by TNF-α may enhance endothelial cytoprotection under inflammatory conditions. DAF upregulation was dependent upon gene transcription, increased steady-state mRNA and de novo protein synthesis. Furthermore, the relatively delayed time course is directly comparable with that observed for cytokine induction of DAF on human EC.¹⁸ In contrast, although Crry was constitutively expressed on murine EC at significantly higher levels than DAF, no change in expression was observed following treatment of EC with TNF-α, IL-1β or IFN-γ for up to 48 hr. This suggests that Crry is more important in providing vascular cytoprotection under non-inflammatory conditions.

Investigation of the signalling pathways involved in DAF regulation confirmed a role for PKC. Inhibition of the classical and novel isozymes of PKC with GF109203X abrogated TNF-α-induced DAF upregulation. Furthermore, experiments with PKCβ- and PKCε-deficient EC, and the specific inhibitors Gö6976 and LY379196, suggested that PKCα was the predominant isozyme involved. This was supported by the demonstration that myr-ψPKC, a highly specific peptide inhibitor of PKCα/β, inhibited upregulation of DAF by TNF-α and by the fact that TNF-α induced phosphorylation of PKCα in MCEC. Evidence, to date, suggests that the regulation of DAF expression on murine and human EC differs somewhat between species and/or vascular beds. Optimum DAF upregulation on human EC was seen following exposure to a combination of TNF-α and IFN-γ, whereas IFN-γ had no effect on TNF-α-induced expression on murine EC. Furthermore, whilst the response in murine EC was PKC-dependent, TNF-α/IFN-γ-induced DAF in human EC was PKC-independent.¹⁸ However, the PKC inhibitor used at that time, RO31-8220, does not inhibit PKCξ, which has recently been implicated in TNF-α-induced ICAM-1 expression in HUVEC.³⁹ Hence, PKC cannot be completely excluded in cytokine-induced DAF upregulation in human EC.

Inhibition of PI-3K and p38 MAPK abrogated TNF-α-induced DAF upregulation, whereas inhibition of ERK 1/2 had no effect. Furthermore, PSI, which prevents the activation of NF-κB through the inhibition of IκBα degradation, also prevented DAF upregulation. The relationship between signalling pathways including PI-3K, PKC, p38 MAPK, ERK 1/2 and NF-κB remain poorly understood and may differ between cell types and in response to different stimuli. Moreover, these pathways may act sequentially or in parallel and may also demonstrate crosstalk.⁴⁰ Evidence, to date, suggests that PI-3K may act upstream of PKCα which is, in turn, upstream of p38 MAPK.^41–44 Activation of PI-3K can lead to the subsequent activation of PKCα during VEGF-induced migration of myeloma cells⁴⁴ and in the control of erythropoietin receptor signalling.⁴² Likewise, an association between the activation of PKCα and p38 MAPK has been reported during TNF-α-mediated signalling and in cytoprotective pathways induced by fibroblast growth factor-2.⁴³

DAF exerts its actions at the proximal end of the complement cascade by accelerating the decay of C3 and C5 convertases.² This is especially relevant in view of the relative inefficiency of C5 activation, which is critically dependent upon an absolute excess of activated C3. Hence, the observation that TNF-α significantly reduced deposition of C3 on opsonized EC exposed to mouse serum which, in turn, was reversed by Riko-1 (the inhibitory DAF mAb) is particularly important. Moreover, the relevance of this reduction in C3 deposition was confirmed by the significant fall in complement-mediated cell lysis observed in TNF-α-treated EC. We have previously shown that Riko-1, at the concentrations used, blocks the function of DAF.²² Therefore, the failure of Riko-1 to reverse completely the protective effect of TNF-α, suggests that additional complement regulatory mechanisms are induced. We have excluded increased Crry expression, but changes in other EC complement regulators, such as CR-1,⁴⁵ or structural changes in the cell membrane itself, may be involved.

Heterologous nephrotoxic nephritis was used to study the regulation of DAF in vivo. Previous work with this model has shown that DAF-deficient mice are more susceptible to nephritis than wild-type controls.¹⁶ Moreover, the role of TNF-α in the pathogenesis of the associated nephritis³³ suggested that this model represented a means by which the ability of TNF-α to induce EC DAF expression in vitro could be tested in vivo. The rapid upregulation of DAF seen in the in vivo experiments reflects the complex nature of the inflammatory response in nephrotoxic nephritis, with local and systemic generation of TNF-α, as well as other factors that might contribute to DAF upregulation, including complement deposition, additional cytokines such as IL-1 and neutrophil immigration.^18,32,33 These data suggest that the vascular endothelium can adapt to ongoing complement activation by upregulating the expression of DAF above constitutive levels.

Previous studies have demonstrated spontaneous C3 deposition on DAF-deficient erythrocytes¹³ and increased susceptibility of DAF knockout mice to glomerular injury.^16,17 These observations and data reported herein, point to a physiologically important role for DAF in murine complement regulation in addition to that reported for Crry. Furthermore, identification of CD97, a member of the EGF-TM7 group of seven-span transmembrane receptors, as a DAF ligand⁴⁵ suggests that DAF has functions above and beyond decay acceleration. Thus, DAF expression on the apical surface of EC and the presence of CD97 on activated leucocytes^46,47 places DAF in a position to modulate leucocyte/EC interactions. The structure and size of DAF, and the signalling properties of both molecules, suggests that their interaction during leucocyte adhesion to endothelium is more likely to provide a costimulatory signal than a primary adhesive event.

In summary, we have demonstrated that DAF is expressed on murine EC and can be upregulated in vivo during inflammation and in vitro by TNF-α through a PI-3K, PKCα, p38 MAPK and NF-κB-dependent pathway. This pathway may play an important role in maintaining vascular integrity during subacute and chronic inflammatory responses.

Abbreviations

bFGF

basic fibroblast growth factor

Crry

complement receptor-related protein-Y

DAF

decay-accelerating factor

DMEM

Dulbecco's modified Eagle's minimal essential medium

EC

endothelial cells

FBS

fetal bovine serum

FITC

fluorescein isothiocyanate

GPI

glycosylphosphatidylinositol

IFN-γ

interferon-γ

mAb

monoclonal antibody

MAPK

mitogen-activated protein kinase

MCEC

mouse cardiac endothelial cells

MCP

membrane cofactor protein

MFI

mean fluorescence intensity

NF-κB

nuclear factor-κB

NMS

normal mouse serum

PI

propidium iodide

PI-3 kinase

phosphoinositide-3 kinase

PKC

protein kinase C

PSI

proteasome inhibitor-1

RFI

relative fluorescence intensity

RT–PCR

reverse transcription–polymerase chain reaction

TM

transmembrane

TNF-α

tumour necrosis factor-α

VEGF

vascular endothelial growth factor