Citrullination of CXCL8 by peptidylarginine deiminase alters receptor usage, prevents proteolysis, and dampens tissue inflammation

Abstract

Biological functions of proteins are influenced by posttranslational modifications such as on/off switching by phosphorylation and modulation by glycosylation. Proteolytic processing regulates cytokine and chemokine activities. In this study, we report that natural posttranslational citrullination or deimination alters the biological activities of the neutrophil chemoattractant and angiogenic cytokine CXCL8/interleukin-8 (IL-8). Citrullination of arginine in position 5 was discovered on 14% of natural leukocyte-derived CXCL8(1-77), generating CXCL8(1-77)Cit(5). Peptidylarginine deiminase (PAD) is known to citrullinate structural proteins, and it may initiate autoimmune diseases. PAD efficiently and site-specifically citrullinated CXCL5, CXCL8, CCL17, CCL26, but not IL-1beta. In comparison with CXCL8(1-77), CXCL8(1-77)Cit(5) had reduced affinity for glycosaminoglycans and induced less CXCR2-dependent calcium signaling and extracellular signal-regulated kinase 1/2 phosphorylation. In contrast to CXCL8(1-77), CXCL8(1-77)Cit(5) was resistant to thrombin- or plasmin-dependent potentiation into CXCL8(6-77). Upon intraperitoneal injection, CXCL8(6-77) was a more potent inducer of neutrophil extravasation compared with CXCL8(1-77). Despite its retained chemotactic activity in vitro, CXCL8(1-77)Cit(5) was unable to attract neutrophils to the peritoneum. Finally, in the rabbit cornea angiogenesis assay, the equally potent CXCL8(1-77) and CXCL8(1-77)Cit(5) were less efficient angiogenic molecules than CXCL8(6-77). This study shows that PAD citrullinates the chemokine CXCL8, and thus may dampen neutrophil extravasation during acute or chronic inflammation.

Figure 1.

Modification of CXCL8 by PAD and production of pure citrullinated CXCL8(1–77)Cit₅. (A) Arginine residues that are incorporated in protein sequences may be converted into citrulline by PAD. (B) Recombinant CXCL8 (5 μM) was incubated with rabbit PAD, human PAD2, or human PAD4 at an enzyme/substrate (E/S) molar ratio of 1:20 or 1:200 for different time periods. The presence of Arg (♦) or Cit (▪) at position 5 was determined by Edman degradation. The percentage of Arg₅ or Cit₅ in the sequence was calculated from the amount of PTH-Arg and PTH-Cit that was detected by RP-HPLC after Edman degradation for five cycles. (C) Alternatively, recombinant CXCL8 was incubated for 90 min with PAD at an E/S molar ratio of 1:20, purified by C8 RP-HPLC, and eluted in an acetonitrile gradient, and UV absorption was detected at 214 nm. Part of the column effluent (0.67%) was connected online to an ion trap mass spectrometer, and the averaged spectra for the chromatographic peaks were deconvoluted to obtain the M_r of the proteins (top graph).

Figure 2.

Sensitivity of citrullinated CXCL8 to thrombin and plasmin cleavage. Recombinant CXCL8(1–77) and CXCL8(1–77)Cit₅ were incubated with thrombin (A) or plasmin (B) for different time periods at an enzyme/substrate molar ratio of 1:100. SDS-PAGE was performed under reducing conditions on Tris/tricine gels and proteins were visualized by silver staining (25). The bovine trypsin inhibitor (M_r 6,200) is visible as the M_r marker (arrows). Edman degradation and ion trap mass spectrometry analysis were used to calculate the percentage of conversion of CXCL8(1–77) or CXCL8(1–77)Cit₅ (black) into CXCL8(6–77) (white) or CXCL8(9–77) (dashed).

Figure 3.

Effect of citrullination on the receptor and GAG binding properties of CXCL8. Increasing concentrations of unlabeled chemokine were added together with ¹²⁵I-CXCL8 to HEK293 cells transfected with CXCR1 (A) or CXCR2 (B). Results represent the mean percentage (± the SEM) of remaining specific ¹²⁵I-CXCL8 binding (n = 4). Alternatively, GAG binding was evaluated by immobilizing low molecular weight heparin (C) or heparan sulfate (D) on EpranEx plates. CXCL8 isoforms bound to GAG were detected with biotinylated anti-CXCL8 antibodies and peroxidase-conjugated streptavidin. The optical densities obtained without addition of CXCL8 and with addition of 300 nM CXCL8(1–77) were set to 0 and 100%, respectively. Results represent the mean percentage of specific chemokine binding (six or more independent experiments) ± the SEM to heparin or heparan sulfate for human CXCL8(1–77) (♦), CXCL8(6–77) (▴), or CXCL8(1–77)Cit₅ (▪). Statistical differences between CXCL8(1–77) and CXCL8(1–77)Cit₅ or CXCL8(6–77) were detected using the Mann-Whitney U test (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

Figure 4.

Calcium signaling capacity and receptor desensitization of citrullinated CXCL8 in neutrophils and HEK293 cells transfected with CXCR1 or CXCR2. The increase in intracellular calcium concentration ([Ca²⁺]_i) in neutrophils and HEK293 cells transfected with CXCR1 or CXCR2 was measured using the ratiometric dye Fura-2. The cells were stimulated with CXCL8(1–77) (♦), CXCL8(6–77) (▴), or CXCL8(1–77)Cit₅ (▪). Values represent the mean (± the SEM) increase of [Ca²⁺]_i (nM) with a detection limit at 20 nM (dotted line; neutrophils, n = 5 [A]; HEK-CXCR1, n = 3 [C]; HEK-CXCR2, n = 5 [E]). Desensitization experiments were performed by rechallenging the cells with 5 nM of CXCL8(1–77) 100 s after the first stimulus. Results (mean ± the SEM) represent the percentage of inhibition of the second agonist by the first stimulus in comparison with buffer as the first stimulus (neutrophils, n = 5 [B]; HEK-CXCR1, n = 3 [D]; HEK-CXCR2, n = 5 [F]). Significant differences were calculated using the Mann-Whitney U test on paired values ( *, P < 0.05; **, P < 0.01 for comparison between CXCL8[1–77] and CXCL8[1–77]Cit₅).

Figure 5.

Phosphorylation of ERK1/2 by CXCL8 forms. The amount of phosphorylated ERK1/2 (pg phospho ERK1/2 per mg total protein) was measured by a specific ELISA after stimulation of serum-starved CXCR2-transfected HEK293 cells for 5, 10, or 20 min with medium (co), 10 nM of CXCL8(1–77), CXCL8(6–77), or CXCL8(1–77)Cit₅. Results represent the percentage ERK1/2 phosphorylation (± the SEM) compared with medium-treated control cells (three to four independent experiments). Statistical analysis was performed using the Mann-Whitney U test on paired values (*, P < 0.05 for comparison with CXCL8[1–77]).

Figure 6.

In vitro neutrophil chemotactic activity of citrullinated CXCL8. The chemotactic activity of CXCL8(1–77), CXCL8(1–77)Cit₅, and CXCL8(6–77) for neutrophils was measured using a Boyden microchamber (four to six independent experiments). The chemotactic index (± the SEM) was calculated by dividing the number of migrated cells toward test samples by the number of spontaneously migrated cells toward buffer. Statistical analysis was performed using the Mann-Whitney U test on paired values (*, P < 0.05; **, P < 0.01; ***, P < 0.001 for comparison with the corresponding concentration of CXCL8[1–77]).

Figure 7.

Effect of citrullination on CXCL8-induced angiogenesis and neutrophil extravasation in vivo. (A)Neovascularization induced by CXCL8(1–77), CXCL8(1–77)Cit₅, and CXCL8(6–77) was tested in a rabbit corneal micropocket. Angiogenesis was scored daily (score 0–4) from days 4 to 8. The angiogenic index was calculated by dividing the maximal neovascularization scores (occurring between days 5 and 7) by the spontaneous angiogenesis score obtained with PBS (co). (B and C) Induction of neutrophil infiltration was measured in NMRI mice by i.p. injection of 200 μl vehicle (0.9% NaCl = co), CXCL8(1–77), CXCL8(6–77), or CXCL8(1–77)Cit₅. After 2 h, mice were killed and the peritoneal cavity was washed. The total amount of leukocytes per microliter of peritoneal lavage solution was determined, and cytospins were stained with Hemacolor solutions to determine the percentage of neutrophils. The figures denote the median (squares), the 25–75% range (boxes), the nonoutlier range (whiskers), and the extreme values (⋄) acquired from 5–9 replicates (eyes; A) or 12–21 mice (B and C) per CXCL8 dose. Statistical analysis was performed using the Mann-Whitney U test (§, P < 0.05; §§, P < 0.01; §§§, P < 0.001 for comparison with vehicle alone; *, P < 0.05; **, P < 0.01 for comparison with CXCL8[6–77]).