Selection of a C5a Receptor Antagonist from Phage Libraries Attenuating the Inflammatory Response in Immune Complex Disease and Ischemia/Reperfusion Injury

Construction of C5a C-terminal libraries

All of the constructed libraries are based on pJuFo plus [Ala²⁷]C5a vector including Cys²⁷Ala replacement in the C5a molecule (16). Altogether, five different libraries with modifications at the C5a C-terminus were constructed (Table I⇓). For all libraries, primer MH40 (5′ end of C5a) was combined with five different mutagenesis primers (3′ end of C5a). Sequences of all primers are given below. All mutagenesis primers introduced a His⁶⁷Phe replacement (Table I⇓). C5a mutants from libraries I–III had 73 aa. C5a mutants from libraries IVa and IVb had 75 aa. For library I, mutagenesis primer MH44 was used, resulting in Asp⁶⁹Phe, Met⁷⁰Lys, and Gln⁷¹Pro replacements and random mutagenesis of positions 72 and 73. For library II, primer MH45 was used resulting in His⁶⁷Phe and Asp⁶⁹Phe replacements and random mutagenesis of positions 70–73. In library III, primer MH46 created random mutagenesis at positions 69–73. In library IVa, Asp⁶⁹Phe, Met⁷⁰Lys, Gln⁷¹Pro replacements were introduced by primer MH47. In addition, positions 72, 73, and 75 were randomly mutated. At position 74, the VDN codon was used encoding all of the 20 naturally occurring amino acids except Arg, Gly, or Cys. Primer MH48, which was used for the construction of library IVb, introduced the same mutations as MH47, with the exception that at position 74 the RGM codon was used, which encodes for Gly and Cys. Thus, by combining libraries IVa and IVb, 19 of the 20 naturally occurring amino acids are encoded at position 74 except the Arg residue, which is conserved in all C5a sequences described (Table I⇓).

The sequences of the different primers were the following from 5′ to 3′: MH40, T CCT TCT AGA ACG CTG CAA AAG AAG ATA; MH44, G ACC GGT ACC TTA MNN MNN TGG CTT GAA TTT AAA AGA GAT ATT AGC ACG GA G; MH45, G ACC GGT ACC TTA TTA MNN MNN MNN MNN GAA TTT AAA AGA GAT ATT AGC ACG GAG; MH46, G ACC GGT ACC TTA TTA MNN MNN MNN MNN MNN TTT AAA AGA GAT ATT AGC ACG GAG; MH47, G ACC GGT ACC TTA TTA MNN VDN MNN MNN TGG CTT GAA TTT AAA AGA GAT ATT AGC ACG GAG; MH48, G ACC GGT ACC TTA TTA MNN RGM MNN MNN TGG CTT GAA TTT AAA AGA. Restriction sites XbaI and KpnI are underlined; D = G, A, or T; M = A or C; N = G, A, T, or C; R = A or G; and V = G, C, or A. All oligonucleotides were purchased from Pharmacia (Freiburg, Germany). Primers longer than 40 bases were purified by denaturing PAGE (17). The PCR reactions, DNA fragment digestion, ligation, transformation, growth of the libraries, helper phage infection, phage preparation, and DNA sequencing of selected clones was performed as described (18). The theoretical sizes of the libraries were from 1 × 10³ (library I) to 3.4 × 10⁷ (library III). We obtained 6 × 10⁶ (library IVb) to 5.4 × 10⁷ (libraries II and III) independent clones. Thus, the whole repertoire of possible clones was represented in each library.

Affinity enrichment of C5a mutants

Selection of C5a receptor binding clones was performed on U937 cells, which had been differentiated with 1 mM N⁶,2′-O-dibutyryladenosine 3′,5′-cyclic monophosphate (Bt₂cAMP; Boehringer Mannheim, Mannheim, Germany) (dU937 cells) for 3 days as described (16). For the first round of panning, 2.5×10⁷ cells grown in RPMI 1640 medium (Life Technologies, Rockville, MD) supplemented with 10% heat-inactivated FCS and penicillin (50 IU/ml)/streptomycin (50 μg/ml) at 37°C in a humidified atmosphere with 5% CO₂ were harvested and resuspended in 500 μl medium containing 100 μl phage (≈ 10¹³ pfu). This mixture was incubated at room temperature for 30 min with gentle agitation. Cells were pelleted (3 min, 800 × g) and washed five times with 1 ml medium. Specifically bound phage were eluted with 200 μl 0.1 M HCl/glycine, pH 2.2, supplemented with 0.1% BSA at room temperature for 10 min. Cells were pelleted by centrifugation and the supernatant was immediately neutralized with 37.5 μl 1 M Tris/HCl, pH 9.1. Then, 100 μl of neutralized eluate were used to infect Escherichia coli TG1 cells. Numbers of phage before and after panning were determined by plating infected E. coli TG1 cells onto TYE+amp+gluc (18).

Altogether, three panning cycles were performed. The washing steps were increased from five times to 10 times in rounds two and three. In panning round three, 2.5 × 10⁶ cells were used. All other steps were performed as described above.

Generation of the ΔpIII-A8 mutant

To obtain the ΔpIII-A8 mutant, gene III, encoding the viral coat protein pIII, was removed by digestion of pJuFo plus [Ala²⁷]C5a-[A8] with SpeI and NheI. SpeI and NheI produce compatible cohesive ends. The digested vector was religated yielding ΔpIII-A8, in which the pIII moiety of the Jun-pIII fusion protein was deleted (see Fig. 1⇓). The PCR reactions, DNA fragment digestion, ligation, transformation, growth of the libraries, helper phage infection, phage preparation, and DNA sequencing of the mutants was performed as described (18).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Cartoon representation of a filamentous phage expressing the selected C5aRA A8. The leucine zipper of transcription factor Jun is fused to the minor coat protein (pIII) of the filamentous M13 phage. The C5a mutant A8 is fused to the leucine zipper of Fos. Jun and Fos form a heterodimer that is stabilized by two disulfide bonds at the N and C terminus of Jun and Fos as described (53 ).

Purification of C5a mutants by affinity chromatography

For all in vitro and in vivo experiments, purified soluble C5a mutants were used. For protein expression TG1 bacteria were grown in 2× TY+amp medium containing 100 μg/ml ampicillin and 0.1% glucose. At an OD₆₀₀ of 0.9, isopropyl b-d-thiogalactoside was added to give a final concentration of 0.5 mM. Bacteria were grown overnight with shaking at room temperature. The next day, the periplasmic fraction was prepared by freezing and thawing. Periplasmic fractions from the selected C5a mutants selected were applied to Sepharose columns coupled with C5a-specific mAb 561 as described (16). Highly purified C5a mutants were obtained as determined by SDS-PAGE and subsequent silver staining.

Competitive ¹²⁵I-labeled recombinant human C5a (¹²⁵I-rhC5a) binding studies

¹²⁵I-rhC5a binding assays were conducted on dU937 cells and naive murine peritoneal cells obtained by peritoneal lavage essentially as described (19). Briefly, increasing concentrations of unlabeled C5a or a C5a mutant were tested against a constant amount of ¹²⁵I-rhC5a as tracer (≈17,000 cpm). After separation of the free ¹²⁵I-rhC5a by filtration, the cell bound ¹²⁵I-rhC5a was determined in a γ-counter (Packard, Canberra, Australia). The plot of cell-bound ¹²⁵I-rhC5a vs the concentration of unlabeled competitor yielded the half-maximal inhibitory concentrations (ID₅₀).

Degranulation assay

As an example for the C5a-induced degranulation of phagocytic cells, the N-acetyl-β-d-glucosaminidase release from dU937 cells was determined. The assay was performed exactly as described (20). The ED₅₀ of C5a or C5a mutants was determined as the concentration leading to half-maximal enzyme release.

Chemotaxis assay

Chemotactic activity of dU937 cells was assessed as described by Schwenk et al. (21). Modified Boyden chambers (Nucleopore, Tübingen, Germany) were filled with increasing concentrations of rhC5a (Sigma, Diesenhofen, Germany) or, in case of inhibition experiments, with a mixture of increasing concentrations of rhC5a and ΔpIII-A8 at a constant concentration of 10⁻⁶M. Subsequently, chambers were covered with polycarbonate filters (pore size, 3 μm; Nucleopore). Then, 100 μl of dU937 cells at a concentration of 5 × 10⁵/ml were applied to each chamber. Cells were allowed to migrate for 3 h at 37°C. Migrated cells were lysed by adding 0.1% Triton X-100 (Boehringer Mannheim), and β-glucuronidase activity in the lysates was determined photometrically using p-nitrophenol-β-d-glucuronide as substrate (Sigma). Chemotactic activity was expressed as chemotactic index, which is defined as the ratio of the number of migrating cells in presence of stimulus vs the number of migrating cells in presence of medium.

GTPase assay

To determine the GTP-ase activity of C5a or C5a mutants, the method of Gierschik (22) was used. Different concentrations of C5a or the C5a mutants were suspended in 50 μl of 50 mM triethanolamine-HCl, 5 mM MES pH 7.3 containing 5 mM MgCl₂, 143 mM NaCl, 1 mM EDTA, 0.16% BSA, 0.8 mM 5′-adenylylimidodiphosphate (App(NH)p), 0.1 mM ATP, 5 mM creatine phosphate, 0.4 mg/ml creatine phosphokinase, 0.1 μM GTP, and 0.2 μl [γ-³²P]dGTP (10.8 μCi/μl; 3000 Ci/mmol). The reaction was started by adding 50 μl of dU937 membranes at a concentration of 3 μg dissolved in the same buffer. Cell membranes were prepared as described by Laugwitz et al. (23). After 15 min incubation at 30°C, 700 μl of an ice-cold 5% (w/v) charcoal suspension in 10 mM H₃PO₄, pH 2.0, were added and the samples were centrifuged at 12,000 × g for 30 min at 4°C. Then, 500 μl of the supernatant were counted in a Beckmann LS 6000 SE counter (Beckman Coulter, Fullerton, CA).

Metabolic labeling and immunoprecipitation of C5aR

A stable cell line of C5aR-transfected RINm5F cells was used for the phosphorylation experiments (24). Cells were metabolically labeled with [³²P]orthophosporic acid (0.3–0.5 mCi/ml) for 3 h at 37°C, as described (25). Briefly, phosphorylation of C5aR was initiated by adding a saturating concentration of C5a (5 × 10⁻⁸ M) or ΔpIII-A8 (8 × 10⁻⁷ M). After 10 min incubation, the monolayers were lysed and C5aR was immunoprecipitated as described (26). Immunoprecipitates were analyzed by SDS-PAGE under reducing conditions (27) and autoradiography.

Detection of C5aR by imunoblotting

Immunoblotting of C5aR-transfected RINm5F cells was performed as described (24). In brief, cells were incubated with either rhC5a (5 × 10⁻⁸ M) or ΔpIII-A8 (8 × 10⁻⁷ M) and subsequently lysed in Laemmli buffer (27) supplemented with DTT, separated by SDS-PAGE, and transferred to a 0.22-μM nitrocellulose filter (Schleicher & Schuell, Keene, NH). Filters were blocked with PBS, 0.1% Tween 10, 3% BSA for 1 h at room temperature. Subsequently, filters were incubated with affinity-purified anti-C5aR rabbit IgGs diluted 1:400 in blocking buffer. The filter was washed with PBS, 0.1% Tween 20 and incubated with ¹²⁵I-labeled protein A (1 μCi/ml of blocking buffer) for 1 h at room temperature. Bound radioactivity was detected by autoradiography using Fuji RX film (Tokyo, Japan) at −80°C.

Measurement of intracellular Ca²⁺ concentration ([Ca²⁺]_i) of murine peritoneal cells

The peritoneal cavity of BALB/c mice was lavaged with 2 ml of ice-cold PBS. Cell staining using Diff-Quick (Baxter Merz & Dade, Dudingen, Switzerland) revealed 95–98% macrophages and 2–5% mast cells. The cells were centrifuged, resuspended in HBSS buffer without Ca²⁺, and subsequently loaded with fura-2/AM (Calbiochem, Bad Soden, Germany) as described (20). After reconstitution of the HBSS buffer with Ca²⁺, the C5a-induced rise in [Ca²⁺]_i was determined using the Luminescence Spectrometer LS 50 B (Perkin-Elmer, Beaconsfield, U.K.).

Animals

Female BALB/c mice (8- to 10-wk old) were purchased from Charles River Laboratory (Wilmington, MA) and were maintained under pathogen-free conditions.

Reverse passive Arthus reaction in the peritoneum

To induce the reverse passive Arthus reaction in the peritoneum, chicken egg albumin (20 mg/kg body weight; Sigma) was injected i.v., subsequently followed by an i.p. injection of IgG rich in Ab to chicken egg albumin (800 μg/mouse; ICN, Eschwege, Germany) exactly as described (7). Mice were killed 6 h after injury, and the peritoneal cavity was lavaged with 2 ml PBS, 0.1% ice-cold BSA. Peritoneal cells were stained using Diff-Quick (Baxter Merz & Dade) and subsequently assessed for differential cell count. Where indicated, mice were injected i.v. with 8 μg cobra venom factor (CVF) the day before IC challenge and again with 8 μg CVF 4 h before IC application. CVF was purified from Naja naja venom (Miami Serpentarium Laboratories, Miami, FL) according to a previously described procedure (28). In inhibition experiments, 200 μl (10⁻⁵ M) of the C5aRA ΔpIII-A8 were applied i.p. either at the time of IC challenge (T₀) or 1 h and 2 h later. In a further set of experiments, 100 μl of ΔpIII-A8 at a concentration of 10⁻⁵ M were given i.v. at T₀ or 1 h and 2 h after IC application. In a third set of experiments, 200 μl ΔpIII-A8 were injected i.p. and 100 μl i.v., both at T₀ and 1 h and 2 h later. Serum complement levels were determined as described (29).

Reverse passive Arthus reaction in skin and lung

Mice were anesthetized with ketamine and xylazine and shaved at their basolateral sides. The trachea was cannulated and 150 μg rabbit anti chicken egg albumin Ab (Sigma) was applied. In addition, 30 μg Ab were injected intradermally. Ag (chicken egg albumin, 20 mg/kg) was subsequently given i.v. Where indicated, 200 μl ΔpIII-A8 at a concentration of 7.3 × 10⁻⁶ M were given i.v. before the Arthus reaction, or, additionally, 100 μl ΔpIII-A8 (7.3 × 10⁻⁶ M) were applied 60 and 120 min after IC challenge. After 4 h, mice were killed and PMN accumulation in the skin was assessed by quantitation of myeloperoxidase (MPO). In the lung, the number of PMN was determined by bronchoalveolar lavage.

Bronchoalveolar lavage

The lungs were lavaged with 5 × 1 ml PBS following cannulation of the trachea. Lavaged cells were stained using May-Grünwald and Giemsa stain. The differential cell count of the bronchoalveolar lavage fluid was assessed with a hemocytometer (Neubauer Zählkammer Jürgens, Germany).

MPO assay

Skin punches (1 cm²) of the injection sites were assayed for MPO activity. MPO was extracted and determined as described (30). Briefly, homogenized tissue was suspended in 50 mM potassium phosphate buffer, pH 6.0, 0.5% hexadecyltrimethylammoniumbromide. Cells were broken by three cycles of freezing and thawing and subsequent sonification. The supernatant was mixed with 0.167 mg/ml O-dianisidine dihydrochloride (Sigma) and 0.0005% hydrogen peroxide. The MPO release was calculated by assessing the absorbance at 450 nm. A serial dilution of MPO from human PMN (Calbiochem-Novabiochem, Bad Soden, Germany) served as standard. Samples were run in duplicate.

Model of intestinal I/R injury

Mice were anesthetized with ether and were kept under a heating lamp. A laparatomy was performed, and I/R was induced by occluding the superior mesenteric artery with a microbulldog clamp. The abdomen was covered with warm, moist gauze during this period. Weight-matched control mice underwent sham operation in which the superior mesenteric artery was exposed but not occluded. After 60 min ischemia, the clamp was removed, the incision was closed with a silk suture, and the mice were returned to their cages. The bowel was reperfused for 60 min or until the animals died (50–55 min after starting the reperfusion). Bowel injury (hemorrhage, bowel wall edema, and purple black discoloration) was assessed macroscopically at the end of the reperfusion period or when animals died. Where indicated, mice were treated with CVF before I/R as described for the reverse passive Arthus reaction. The C5aRA was given twice: 200 μl (7.3 × 10⁻⁶ M) were applied i.v. before starting ischemia; another 100 μl were given i.v. directly before removing the clamp.

Quantitation of I/R-induced microvascular dysfunction

Intestinal reperfusion-induced dysfunction was quantitated by measuring the extravasation of plasma proteins into the lung as described (31). Evans blue dye binds avidly to albumin and was used as a marker of protein extravasation. This technology compares favorably with the methodology involving radiolabeled albumin (31). After 60 min of reperfusion, animals were killed and lung tissue was harvested. Pulmonary vessels were emptied by flushing saline into the right ventricle. Subsequently, the lungs were excised, weighed, and the right lung lobe was placed in 1 ml of formamide and homogenized. After incubation in formamide at 37°C for 16 h, the concentration of dye within the eluate was measured spectrophotometrically at 620 nm and expressed as ng dye/mg wet lung weight.

Statistical analysis

Statistical analysis was performed using the SigmaStat version 2.0 statistical package (Jandel, Erkrath, Germany). All data are given as mean ± SEM. First, we tested for a normal distributed population using the Kolomogoroff-Smirnov test. To analyze differences between two normally distributed groups, an unpaired t test was used. Comparison of the means of more than two groups were done by one-way ANOVA. When the differences in the mean values of the groups showed a significant difference, pairwise comparison was performed using the Tukey test. A p value <0.05 was considered to be significant, and a p value <0.001 was considered to be highly significant.

Panning of the C5a phage libraries on differentiated U937 cells

We constructed 5 C5a-C-terminal libraries, i.e., libraries I, II, III, IVa, and IVb. In libraries I, II, and III, the functionally important Arg⁷⁴ residue was deleted. In libraries IVa and IVb, C5a mutants with a length of 75 aa were constructed, in which no Arg⁷⁴ residue was allowed. In library III, positions 69–73 were randomly mutated (Table I⇑).

For the first panning cycle, 3.0 × 10¹² phage of a mixture of all five libraries were used, which is 6 × 10¹¹ phage from each library. Panning was performed in solution using C5aR-expressing dU937 cells. Next, 2.1 × 10⁸ phage were eluted from dU937 cells giving an output/input ratio of 7.2 × 10⁻³%. During the panning cycles, the number of eluted phage increased by a factor of ∼200, giving a final output/input ratio of 1.1% after three panning rounds. After the third pan, 50 clones were randomly picked and screened for consensus motifs. All clones selected belonged to library III, in which positions 69–73 were randomly mutated. The Phe⁶⁹Lys⁷⁰Pro⁷¹ motif present in library I, II, IVa, and IVb was not found. The motif most often selected included a Leu residue at positions 70, 71, or 72 and a Tyr or an Arg residue at position 73, while no particular amino acid was enriched at position 69, as we have described recently (19).

Inhibition of C5a binding to dU937 cells by the selected C5a mutants

Selected C5a mutants with C-terminal consensus motifs Ser⁷⁰Leu⁷¹Leu⁷²Arg⁷³, i.e., A10, B4, and B8, were expressed as soluble proteins, purified by affinity chromatography, and tested in competitive binding studies using dU937 cells (Table II⇓). All clones showed ID₅₀ values in the range of rhC5a, while the ID₅₀ of one mutant, A8, the structure of which is Jun-pIII/Fos-C5a-[1–66]-Phe⁶⁷Lys⁶⁸Arg⁶⁹Ser⁷⁰Leu⁷¹Leu⁷²Arg⁷³, was about 15-fold higher as compared with rhC5a (Table II⇓). Mutants A10, B4, B8, and A8 share the same C-terminal sequence except position 69, which is Arg in A8 and Ala, Glu, or His in A10, B4, and B8, respectively. In clone A8, a cartoon representation of which is depicted in Fig. 1⇑, the minor coat protein pIII was deleted, resulting in ΔpIII-A8. This deletion did not change binding affinity (Table II⇓). Mutant ΔpIII-A8 was used throughout additional experiments.

The ΔpIII-A8 mutant does not induce degranulation of dU937 cells

Mutants A10, B4, and B8 induced degranulation of dU937 cells and had ED₅₀ values quite similar to C5a (Table II⇑), whereas ΔpIII-A8 did not induce degranulation up to the highest concentration tested, which was 3 × 10⁻⁶ M. At that concentration, rhC5a was completely displaced from the receptor. ΔpIII-A8 was also devoid of any chemotactic activity (data not shown). To check whether signal transduction was blocked at the level of G-protein coupling, hydrolysis of [γ-³²P]GTP was determined. A dose-dependent increase in GTPase activity was observed for C5a and B4 but not for ΔpIII-A8, even at 2 × 10⁻⁶ M (Fig. 2⇓). Next, we tested the ability of ΔpIII-A8 to induce the phosphorylation of the C5aR. As shown in Fig. 3⇓, C5a caused a strong phosphorylation of the C5aR, whereas no ligand specific phosphorylation occurred for ΔpIII-A8.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

ΔpIII-A8 does not induce GTP hydrolysis in membranes of dU937 cells. Recombinant hC5a or mutant B4 increased GTPase activity from 11 (rhC5a) or 13 pmol/min/mg protein to 35 (rhC5a) and 28 pmol/min/mg protein, respectively. Values of one typical experiment of three are given.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

ΔpIII-A8 does not cause any phosphorylation of RINm5F cells of the human C5aR. A, The C5aR was immunoprecipitated from ³²P metabolically labeled cells before (−) and after (+) stimulation with C5a (5 × 10⁻⁸M) or ΔpIII-A8 (8 × 10⁻⁷M). B, Immunoblot of nonlabeled cell extracts before (−) and after (+) C5a (5 × 10⁻⁸ M) or ΔpIII-A8 (8 × 10⁻⁷ M) addition revealed with an anti-C5aR serum. Phosphorylation of the C5aR induces a mobility shift from 40 to 43 kDa.

Inhibition of C5a induced enzyme release and chemotaxis

To assess the potency of ΔpIII-A8 to inhibit C5a-induced degranulation of dU937 cells, increasing concentrations of ΔpIII-A8 were incubated with rhC5a at a concentration of 8.6 × 10⁻¹⁰ M (ED₈₀). The C5a mutant inhibited cell degranulation with an ID₅₀ of 7.85 ± 1.57 × 10⁻⁸ M (Fig. 4⇓A).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

ΔpIII-A8 inhibits C5a-induced cell degranulation and chemotaxis. A, Differentiated U937 cells were challenged with a constant concentration of rhC5a, i.e., 8.6 × 10⁻¹⁰ (ED₈₀) after a 10-min preincubation with increasing concentrations of ΔpIII-A8. ΔpIII-A8 inhibits cell degranulation with an ID₅₀ of 7.85 ± 1.57 × 10⁻⁸ M. B, Increasing concentrations of C5a were applied to dU937 cells in the absence or the presence of ΔpIII-A8 (10⁻⁶ M). Maximal cell migration induced by C5a at a concentration of 5 × 10⁻⁹ M was inhibited by 77%. Lower concentrations of C5a, i.e., 10⁻⁹ M (ED₉₀) or 5 × 10⁻¹⁰ M (ED₈₀) were inhibited by 92% and 100%, respectively. Data from one typical experiment of five are given. ∗∗, p < 0.001 (t test).

When ΔpIII-A8 was applied at a concentration of 1 × 10⁻⁶ M 10 min before the C5a challenge, the C5a-induced enzyme release was completely inhibited up to concentration of C5a of 3 × 10⁻⁹ M, which was the highest concentration tested. The simultaneous application of C5a (3 × 10⁻⁹ M) and ΔpIII-A8 (10⁻⁶ M) resulted in a 70% inhibition (data not shown). Simultaneous application of ΔpIII-A8 at a concentration of 10⁻⁶ M and increasing concentrations of rhC5a inhibited the C5a-induced chemotaxis between 77% and 100% (Fig. 4⇑B).

ΔpIII-A8 is a C5aRA on murine peritoneal macrophages

The A8 mutant was selected using the human histiocytic cell line U937. To test whether this mutant also binds to the murine C5aR and, moreover, whether it is able to displace C5a from the murine C5aR, we performed competitive binding studies with naive murine peritoneal macrophages. Indeed, ¹²⁵I-rhC5a was displaced from the murine C5aR with an ID₅₀ of 1.68 ± 0.81 × 10⁻⁸ M and thus did not differ from the data obtained with dU937 cells (Table II⇑). To analyze, whether ΔpIII-A8 also inhibits C5a-mediated effects in mice, the antagonist was tested for its potency to block rhC5a-induced rise in [Ca²⁺]_i of naive peritoneal macrophage. As shown in Fig. 5⇓, ΔpIII-A8 completely inhibited Ca-mobilization mediated by 10⁻⁹ M (ED₅₀) and 2 × 10⁻⁹ M (ED₇₅) C5a, when applied at a constant concentration of 10⁻⁶ M. Higher concentrations of C5a, i.e., 10⁻⁸ M or 2 × 10⁻⁸ M, were inhibited by 50–75%.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

ΔpIII-A8 inhibits C5a-induced [Ca²⁺]_i increase in murine peritoneal macrophage. Naive peritoneal macrophage were challenged with increasing concentrations of rhC5a in the absence or the presence of ΔpIII-A8 at a concentration of 10⁻⁶ M. ΔpIII-A8 was applied 10 min before the C5a stimulus. The C5a-induced rise in [Ca²⁺]_i was completely inhibited when C5a was applied at a concentration of 10⁻⁹ M or 2 × 10⁻⁹ M. At higher concentrations, the inhibition was between 50 and 95%. Data from one typical experiment of three are depicted. ∗∗, p < 0.001; ∗, p < 0.05 (t test).

Inhibition of neutrophil recruitment in IC peritonitis

To analyze the in vivo potency of ΔpIII-A8, it was tested in a murine model of IC peritonitis, i.e., the reverse passive Arthus reaction. BALB/c mice were challenged i.v. with chicken egg albumin and i.p. with anti-chicken egg albumin Ab. This treatment resulted in the influx of 3.8 ± 0.8 neutrophils × 10⁶/mouse, corresponding to 46.5 ± 7.3% of cells in the peritoneum 6 h after IC challenge (Fig. 6⇓). Administration of 200 μl (10⁻⁵ M) of ΔpIII-A8 i.p. at the time of IC challenge or 1 h and 2 h after IC challenge did not alter the PMN influx 6 h after initiation of IC peritonitis. A significant reduction in PMN recruitment was found when 100 μl (10⁻⁵M) of ΔpIII-A8 were applied i.v., independent of whether the molecule was given at the time of IC challenge or 1 h and 2 h after the initiation of IC peritonitis (PMN, 20.1 ± 5.9% or 20.1 ± 7.4%; p < 0.05). When 200 μl ΔpIII-A8 were injected i.p. and 100 μl i.v. (both 10⁻⁵ M) at the time of Ag/Ab application, and again 1 h and 2 h after IC application, PMN influx was abolished (9.5 ± 3.1%) (p < 0.001; ANOVA). An identical reduction in neutrophil recruitment was observed after complement depletion with CVF (10.1 ± 4.8%). Neutrophil accumulation in Ag and Ab controls was 2.1 ± 0.8 or 13.6 ± 6.4% of peritoneal cells, respectively.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6.

ΔpIII-A8 impairs PMN recruitment in IC peritonitis. Intraperitoneal challenge of BALB/c mice with IC resulted in an increase in the portion of PMN in the peritoneal cavity from 1 to 2% in untreated animals to 46.5 ± 7.3% 6 h after IC application. A combination of i.p. and i.v. application of ΔpIII-A8 1 h and 2 h after IC challenge abrogated PMN migration (46.5% ± 7.3% (n = 26) vs 9.5% + 3.1% (n = 6); p < 0.001, ANOVA and Tukey test). An identical reduction was found for CVF-treated animals (10.1% + 4.8% (n = 8); p < 0.001, ANOVA and Tukey test). PMN recruitment was also significantly blocked when ΔpIII-A8 was injected solely i.v. 1 h and 2 h after IC challenge (PMN, 20.1% + 5.9% (n = 12); p < 0.05, ANOVA and Tukey test) or at the time of IC application (T₀) (PMN, 20.1 + 7.4% (n = 8); p < 0.05, ANOVA and Tukey test). Intraperitoneal application of ΔpIII-A8 did not influence PMN migration. ∗∗, p < 0.001 .

Attenuation of the PMN influx in the reverse passive Arthus reaction in skin and lung

The properties of ΔpIII-A8 to serve as a C5aRA were further evaluated in two other models of IC diseases, i.e., the reverse passive Arthus reaction in the skin and in the lung. After i.v. application of OVA and a combination of intratracheal and subcutaneous application of anti-OVA Abs, a tremendous accumulation of neutrophils was observed, both in the skin and in the lung (Fig. 7⇓). The MPO activity in the skin, 4 h after challenging mice with IC, was 241.6 ± 51.6 μg/cm² as compared with 44.1 ± 7.1 μg/cm² or 57.3 ± 8.7 μg/cm² in mice receiving only the Ag or the Ab. When mice were treated with a single dose of 200 μl i.v. ΔpIII-A8 at a concentration of 7.3 × 10⁻⁶ M or, additionally, with 100 μl C5aRA 1 h and 2 h after initiation of the Arthus reaction, the MPO activity decreased to 101.2 ± 33.4 or 80.5 ± 23.7 μg/cm² (Fig. 7⇓A).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 7.

The PMN migration into the skin (A) and the lung (B) is attenuated by the C5aRA ΔpIII-A8. A, The i.v. application of a single dose ΔpIII-A8 (n = 5) just before initiation of the Arthus reaction (T₀) reduced the skin content of MPO to 26.8%; however, this was statistically not significant (p = 0.052, ANOVA and Tukey test). The additional i.v. application of ΔpIII-A8 (n = 7) 60 min and 120 min after IC challenge reduced the skin content of MPO significantly to 15.8% as compared with untreated animals (n = 6) (p = 0.023, ANOVA and Tukey test). B, The number of neutrophils determined in the BAL was significantly reduced from 3.2 ± 0.8 neutrophils × 10⁵/mouse to 0.7 ± 0.2 neutrophils × 10⁵/mouse in the group that received three doses of the C5aRA i.v. (p = 0.003, ANOVA and Tukey test).

In the lung of mice receiving only OVA or anti-OVA Ab, a maximum of 8 × 10³ PMN was found in the pulmonal lavage fluid taken 4 h after the onset of the Arthus reaction (Fig. 7⇑B). In mice challenged with ICs, the neutrophil number increased to 324 ± 86 × 10³. In mice receiving a single dose of 200 μl i. v. of the C5aRA (7.3 × 10⁻⁶ M), the neutrophil number decreased to 264 ± 91 × 10³. This reduction was statistically not significant. However, the additional application of 100 μl ΔpIII-A8 i. v. 60 and 120 min after initiation of the Arthus reaction resulted in a significant reduction of the neutrophil number to 73 ± 22 × 10³ (p = 0.003).

Effects of the C5aRA on I/R-induced local and remote organ injury

Occluding of the superior mesenteric artery for 60 min and subsequent reperfusion for 1 h resulted in severe local and remote tissue injury, i.e., in the bowel and the lung. In the group of untreated animals, 5 of 10 animals died 50–55 min after reperfusion. All animals suffered from heavy bowel injury (hemorrhage, bowel wall edema, and purple black discoloration located at different sites) (Table III⇓). In contrast, all animals receiving CVF survived and showed no signs of bowel injury at all. When mice were treated with 200 μl ΔpIII-A8 i.v. at a concentration of 7.3 × 10⁻⁶ M before arterial occlusion and another 100 μl of the C5aRA just before reperfusion, four of six animals survived. None of the surviving animals showed intestinal injury, while the two nonsurviving animals exhibited intestinal hemorrhage and bowel edema. Sham-operated controls had completely normal bowel appearance and survived.

To assess remote tissue injury, change in lung vascular permeability was assessed. I/R in untreated animals was associated with a significantly greater extravasation of plasma proteins into the pulmonary interstitium as compared with sham-operated controls or CVF- or C5aRA-treated animals (Table III⇑; p < 0.001; ANOVA). In fact, the concentration of Evans blue dye within the lungs of CVF- and C5aRA-treated animals was indistinguishable from sham-operated controls, suggesting that C5a is a key mediator of the intestinal I/R-induced increase in lung vascular permeability.