COOH-terminal SAA1 peptides fail to induce chemokines but synergize with CXCL8 and CCL3 to recruit leukocytes via FPR2

The acute phase response is part of the inflammatory cascade. It is caused by infection, trauma, or cancer and comprises processes such as fever and leukocytosis to restore disturbances of homeostasis in the organism. One of the systemic effects during the inflammatory response is the induction of acute phase proteins in the liver.¹  Serum amyloid A (SAA) is a highly upregulated major acute phase protein in humans. Human SAA consists of the SAA1 variants SAA1α, -β, and -γ; the SAA2 forms SAA2α and -β; SAA3; and SAA4.²  Because the increase in concentration of SAA during the acute phase response is most pronounced for SAA1 and SAA2 variants (100- to 1000-fold increase), these are denominated “acute phase SAA” or “A-SAA.” In contrast, the concentration of SAA4, named “constitutive SAA” or “C-SAA,” remains practically stable. Human SAA3 protein, initially regarded as a pseudogene product,³  has only been detected at low levels in mammary gland epithelial cells.⁴ 

Being highly conserved, SAA takes part in the inflammatory response of most vertebrates. Nonetheless, the exact role of SAA and the differences in functions between distinct SAA variants are still not elucidated. Several functions have been attributed to A-SAA, such as antibacterial and antiviral activities,^5-9  stimulation of angiogenesis,^10-12  and a role in cholesterol transport.^13,14  In a recent review,²  the most important functions reported for SAA were listed according to the minimal effective SAA concentration required to exert the respective biological activity. Based on this parameter, SAA’s critical functions are its chemotactic activity and its chemokine-, cytokine-, and matrix-degrading enzyme-inducing capacities.¹⁵  Chemokines constitute a family of small (7-12 kDa) chemotactic cytokines that exert their chemotactic activity through specific G protein–coupled receptors (GPCRs).^16,17 

SAA1α is chemotactic for neutrophils, monocytes, immature monocyte-derived dendritic cells and T lymphocytes.^15,18-22  SAA1α exerts its chemotactic activity in a GPCR-dependent way via formyl peptide receptor 2 (FPR2),²³  and indirectly, by induction of chemokines via toll-like receptor 2 (TLR2). Subsequently, the induced chemokines cooperate with each other and with SAA1 to enhance leukocyte migration.^20,24  Several receptors, including TLR2, TLR4, FPR2, and the scavenger receptor CD36, are involved in the chemokine- and cytokine-inducing properties of SAA.^25-30  In our hands, CXCL8 induction in SAA1α-stimulated monocytes is blocked by TLR2 neutralization, whereas its direct neutrophil chemotactic effect is inhibited by an FPR2 antagonist.²⁴  The reports about the use of more than 1 receptor by SAA underline that SAA exerts its different activities by binding to distinct receptors. Lately, new insights in the functions of FPRs have revealed their importance in innate host defense, as critical receptors propagating chemotactic signals during microbial infections, inflammation, and cancer.^31-34 

Here, we provide evidence for the presence of a biologically active COOH-terminal fragment of SAA1α in the blood circulation. This peptide was first purified and identified from bovine serum via a standardized 4-step chromatographic procedure. It was subsequently chemically synthesized together with human equivalents. We demonstrate that these SAA1 peptides fail to induce chemokines and to chemoattract neutrophils and CD14⁺ monocytes on their own. However, these COOH-terminal fragments potently synergize with human CXCL8 and murine CXCL6 in in vitro and in vivo neutrophil chemotaxis, respectively, as well as with CCL3 in monocyte migration. Moreover, we show that the synergy between SAA1α fragments and CXCL8 in neutrophil chemotaxis and CCL3 in monocyte migration is mediated by FPR2. Because SAA1 fragments desensitize the chemotactic activity of intact SAA1α, natural proteolytic processing of SAA1 fine-tunes the inflammatory response. Indeed, these SAA1 peptides can sustain an elevated number of circulating neutrophils during host insults because of synergy with chemokines.

Monocyte chemotactic factors were concentrated and purified from tissue culture–grade newborn calf serum (Life Technologies, Paisley, United Kingdom) following a standardized 4-step chromatographic procedure, as described previously.³⁵  Each step was followed by assessing the monocyte chemotactic activity of the fractions via MultiScreen assays (Millipore Corporation, Billerica, MA) with the THP-1 monocytic cell line.³⁵  For more details, see the supplemental Methods (available on the Blood Web site).

Bovine (bo) SAA1(46-112), human (hu) SAA1(47-104), and hu SAA1(52-104) were synthesized by solid-phase peptide synthesis on an Activo-P11 automated synthesizer (Activotec, Cambridge, United Kingdom) using fluorenylmethoxycarbonyl (Fmoc) chemistry.³⁶  The processing of the SAA synthetic fragments is explained in supplemental Methods. Recombinant hu apo-SAA1 (rSAA1α; endotoxin level <30 pg/µg SAA1α) and CXCL8(6-77), identified as a predominant form produced by monocytes,³⁷  were purchased from PeproTech. Recombinant hu CCL3 was from R&D Systems. Lipopolysaccharide (LPS) from Escherichia coli (0111:B4) was obtained from Sigma-Aldrich. Recombinant murine (mu) CXCL6(9-78) was expressed in E coli and purified to homogeneity.³⁸  The selective FPR2 antagonist WRW₄^39,40  and the selective CXCR2 antagonist SB225002 were purchased from Calbiochem.

CD14⁺ monocytes were isolated from 1-day-old buffy coats from healthy donors (Blood Transfusion Center, Mechelen, Belgium) via density gradient centrifugation on Ficoll-sodium diatrizoate (Lymphoprep; Axis-Shield PoC AS, Oslo, Norway) and via magnetic cell separation (Miltenyi Biotec, Bergisch Gladbach, Germany).³⁵  Neutrophils were isolated from fresh blood as described.²⁴ 

The chemotactic potency of chemoattractants for CD14⁺ monocytes and neutrophils was determined in the Boyden microchamber assay (Neuro Probe, Gaithersburg, MD).^24,41  Changes in morphological shape after neutrophil stimulation with chemotactic agents were evaluated by microscopic counting.²⁴  SAA-mediated receptor signaling was determined by measuring phosphorylation of extracellular signal-regulated kinase 1/2 (ERK1/2) in adherent monocytes.⁴¹  The chemokine-inducing capacity of SAA was tested on CD14⁺ monocytes, fresh neutrophils, and ex vivo on murine peritoneal cells. Levels of CCL3 (sensitivity of the enzyme-linked immunosorbent assay [ELISA]: 0.7 ng/mL) and/or CXCL8 (sensitivity of the ELISA: 0.04 ng/mL) in supernatants were determined with sandwich ELISAs developed in our laboratory.^42,43  Biological assays are described in detail in supplemental Methods.

In vivo experiments in mice, approved by the local Ethics Committee for Animal Experiments (University of Leuven, Leuven, Belgium), were conducted in conformity with European legislation on animal usage for scientific purposes. Female NMRI mice (7-8 weeks old; 3 or 4 mice per group per experiment), kept in a specific pathogen-free environment (Elevage Janvier, Le Genest Saint Isle, France), were injected intraperitoneally with 200 µL phosphate-buffered saline (PBS), mu CXCL6 (100 ng), SAA1(46-112) (3 or 10 µg), hu rSAA1α (100 or 1000 ng), or a combination of mu CXCL6 (100 ng) and SAA1(46-112) (10 µg) or hu rSAA1α (100 ng). For antagonization experiments, mice were injected intraperitoneally with 200 µL PBS, SAA1(46-112) (10 µg) or a combination of 10 µg SAA1(46-112) and 25 µg SB225002. Mice were euthanized after 2 hours and peritoneal lavages were performed as previously described.²⁴  The percentage neutrophils (CD11b⁺Gr1⁺ cells) migrated into the peritoneal cavity toward the chemotactic agent(s) was evaluated by flow cytometric analysis (FACSCalibur; BD Biosciences, Heidelberg, Germany).

Proteolytic cleavage of intact SAA1α and statistical analysis were performed as depicted in supplemental Methods.

Previously, we isolated new chemokines (regakine-1 and CCL3 isoform 2) from bovine calf serum, based on their chemotactic activity for monocytes.^35,44  Using a similar isolation procedure and the MultiScreen assay with THP-1 cells to measure chemotactic activity, we purified another monocyte chemotactic factor (MCF-2) from bovine calf serum. After heparin-Sepharose and cation exchange chromatography steps, MCF-2 eluted upon reversed phase high performance liquid chromatography (RP-HPLC) as a single peak of THP-1 cell chemotactic activity in fractions 43 and 44 at 29% acetonitrile (Figure 1A) and corresponded to a 7 to 8 kDa protein on sodium dodecyl sulfate–polyacrylamide gel electrophoresis (Figure 1B). Mass spectrometry of the protein revealed a relative molecular mass (M_r) of 7287 (Figure 1C) with a by Edman degradation determined NH₂-terminal amino acid sequence of Arg-Gly-Pro-Gly-Gly. Both this NH₂-terminal amino acid sequence and the M_r of 7287 matched with the 67 amino acid (aa) long carboxy-terminal part of bovine SAA1, corresponding to amino acids 46 to 112 of the mature protein, here designated SAA1(46-112) (UniProt database accession number: P35541).

Recently, we have studied the mechanisms involved in the chemotactic activity of intact hu SAA1α on monocytes and neutrophils.^20,24  Therefore, we first evaluated whether the identified fragment SAA1(46-112) has similar chemotactic activity as ascribed to intact SAA1α. The 67 aa long bovine SAA1 fragment was therefore chemically synthesized and purified to homogeneity via RP-HPLC. Mass spectrometric analysis confirmed the correct synthesis and deprotection of SAA1(46-112) (Figure 1D).

After initial screening on THP-1 cells, we analyzed the chemotactic response of physiologically more relevant CD14⁺ monocytes toward SAA1(46-112) in the Boyden chemotaxis assay, in which activity for SAA1 has previously been reported.¹⁹  Compared with the potent monocyte chemoattractant activity of CCL3 (number of migrated cells = 1509 ± 213 at 1 ng/mL, n = 11, P = .0001; data not shown), SAA1(46-112) (maximal number of migrated cells = 161 ± 11 at 10 µg/mL, n = 3) did not attract CD14⁺ monocytes, isolated from human buffy coats. As a control, intact hu rSAA1α significantly attracted CD14⁺ monocytes (maximal number of migrated cells = 229 ± 42 at 3 µg/mL, n = 13, P = .015; Figure 2A). In contrast to intact SAA1α, SAA1(46-112) did not induce significant ERK1/2 phosphorylation in monocytes [162.3 ± 56.6% at 3 µg/mL, n = 5, P = .43 for SAA1(46-112) vs 428 ± 182.2% at 3 µg/mL, n = 4, P = .005 for intact SAA1α] (Figure 2B). Further, the strong neutrophil chemoattractant CXCL8 (1 ng/mL) attracted 1113 ± 151 neutrophils per 10 high power fields (HPFs) (n = 9, P = .0004) in the Boyden chamber assay (data not shown), whereas SAA1(46-112) did not significantly stimulate the migration of neutrophils (maximal number of migrated neutrophils = 170 ± 75/10 HPFs at 10 µg/mL, n = 6). On the other hand, intact SAA1α significantly stimulated the migration of neutrophils (213 ± 38/10 HPFs at 3 µg/mL, n = 10, P = .0022; Figure 2C).

In view of the apparently contradicting data between natural and synthetic SAA1(46-112), we postulated that the natural preparation still contained a contaminating chemoattractant with which it synergized. Indeed, we have previously shown that intact SAA1α synergizes with CXCL8 via binding to FPR2 and CXCR2, respectively, to chemoattract neutrophils.²⁴  Therefore, we investigated whether SAA1(46-112) still synergized with CXCL8 in neutrophil shape change and chemotaxis assays. Figure 3A shows that SAA1(46-112) at 300 to 1000 ng/mL dose-dependently synergized with CXCL8 (3-10 ng/mL) to chemoattract neutrophils. A maximal chemotactic index (CI; CI = 77.3 ± 9.7, n = 4, P = .03) was reached when SAA1(46-112) at 300 ng/mL (CI = 1.1 ± 0.2, n = 6) was combined with CXCL8 at 10 ng/mL (CI = 42.0 ± 4.6, n = 4) in the lower compartment of the Boyden chamber. In shape change assays, 3 µg/mL of SAA1(46-112) was needed to synergize with CXCL8 (5-25 ng/mL) to activate neutrophils (Figure 3B). Neutrophils were stimulated during 5 minutes with 3 µg/mL of SAA1(46-112) (net percentage of blebbed neutrophils being 1 ± 1%, n = 6), CXCL8 at 5 ng/mL (9 ± 5% blebbed neutrophils, n = 6) or at 25 ng/mL (21 ± 8% blebbed neutrophils, n = 6), or with a combination of both chemoattractants. For the combination of SAA1(46-112) (3000 ng/mL) with CXCL8 at 5 ng/mL or with CXCL8 at 25 ng/mL, the net percentage of blebbed neutrophils was significantly higher [56 ± 4%, P = .0048 and 65 ± 3%, P = .0049, respectively, n = 6] than the sum of these percentages when both chemoattractants were added separately (Figure 3B).

Next, we tested if SAA1(46-112) can interfere with intact SAA1α and CXCL8 in neutrophil chemotaxis. Therefore, we performed desensitization experiments, whereby neutrophils were preincubated (10 minutes) with different concentrations (1000-3000 ng/mL) of SAA1(46-112) before adding the cells to the upper compartment of the Boyden microchamber. Figure 3C shows that intact SAA1α (300 ng/mL, CI = 1.2 ± 0.1, n = 3) synergized significantly (CI = 42.7 ± 1.5, n = 12, P = .010) with CXCL8 (3 ng/mL, CI = 26.9 ± 0.6, n = 12) to attract neutrophils, as demonstrated previously.²⁴  This synergy was completely blocked (CI = 25.1 ± 0.9, n = 12, P = .0086) when neutrophils were pretreated with a threefold higher concentration (1000 ng/mL) of SAA1(46-112) than that of intact SAA1α (300 ng/mL) synergizing with CXCL8. Preincubation of neutrophils with SAA1(46-112) at 3000 ng/mL also fully inhibited (CI = 13.3 ± 0.7, n = 12, P = .0001) the synergy between intact SAA1α (300 ng/mL) and CXCL8 (3 ng/mL) (Figure 3C).

Recently, we showed that SAA1α did not synergize with CCL3 in the chemotaxis assay to attract monocytes, because this acute phase protein rapidly induced endogenously synergizing CCL3 in the target monocytes during the chemotaxis experiment.²⁰  In contrast, intact SAA1α did not rapidly induce chemokines in neutrophils and hence was able to synergize with exogenous CXCL8 in the neutrophil chemotaxis assay.²⁴  Before testing whether SAA1(46-112) could synergize with CCL3 to attract monocytes in vitro, we first verified chemokine induction by SAA1(46-112) in CD14⁺ monocytes (Figure 4A). SAA1(46-112) (3 µg/mL) did not stimulate production of CCL3 (1.94 ± 0.39 ng/mL, n = 3) (Figure 4Ai) or CXCL8 (0.11 ± 0.03 ng/mL, n = 5) (Figure 4Aii) in CD14⁺ monocytes within the duration of the chemotaxis assay (2 hours), neither within 24 hours (data not shown). In contrast, these cells produced 7.99 ± 5.36 ng/mL of CCL3 (n = 3) and 3.82 ± 2.80 ng/mL of CXCL8 (n = 5, P = .0117) upon 2 hours stimulation with intact SAA1α (1000 ng/mL) (Figure 4A).

Notwithstanding its lack of chemokine induction in monocytes, SAA1(46-112) was able to synergize with CCL3 to attract CD14⁺ monocytes. Indeed, a combination of SAA1(46-112) at 1 µg/mL (CI = 1.6 ± 0.3, n = 7) or 3 µg/mL (CI = 2.0 ± 0.3, n = 7) and CCL3 at 1 ng/mL (CI = 14.8 ± 0.4, n = 7) in the lower compartment of the Boyden chamber provoked a stronger chemotactic response of these cells (CI = 17.4 ± 0.7 n = 7, P = .025 or CI = 18.7 ± 0.9 n = 7, P = .015, respectively) than the sum of the CI when both chemoattractants were added separately to the lower wells (Figure 4B). In contrast, as already depicted previously,²⁰  intact SAA1α (3000 ng/mL) did not synergize with CCL3 (0.1-1 ng/mL) (Figure 4B), because SAA1α already cooperates with the rapidly induced (endogenous) chemokines (Figure 4A) to enhance monocyte chemotaxis.

The difference between intact SAA1α and SAA1(46-112) at inducing chemokines in vitro was confirmed ex vivo. Stimulation of peritoneal cells from untreated mice with 100 to 1000 ng/mL of intact hu SAA1α during 24 hours yielded a statistically significant increase in the production of the major murine neutrophil attractant CXCL6. In contrast, SAA1(46-112) failed to induce CXCL6 ex vivo at a dose as high as 10 µg/mL (Figure 5A), thereby confirming the in vitro induction data.

We previously demonstrated that neutrophil influxes into the peritoneal cavity of mice induced by intact SAA1α could be inhibited by a selective CXCR2 antagonist, indicating that the neutrophil recruitment was at least partially mediated via induction of CXCR2 ligands such as CXCL6.²⁴  Coinjection of CXCL6 (100 ng) together with intact SAA1α (100 ng) did not yield any increase, but rather a decrease in neutrophil recruitment (Figure 5Bi). Possibly, this is because of interference between the endogenous CXCL6 induced by SAA1α and the injected CXCL6. In contrast, the neutrophil recruitment by mu CXCL6 (100 ng) was further enhanced by coinjection of 10 µg of SAA1(46-112) (Figure 5Bii), which could only weakly (6.22 ± 1.06 × 10⁴ neutrophils/mL, n = 17) stimulate neutrophil accumulation on its own (Figure 5B). Because SAA1(46-112) failed to chemoattract neutrophils in vitro (Figure 2C) and to induce chemokines (Figures 4A and 5A), it was speculated that in vivo SAA1(46-112) synergized with trace amounts of constitutive chemokine. Indeed, the selective CXCR2 antagonist SB225002 significantly reduced the neutrophil influx upon intraperitoneal injection of SAA1 peptide. SAA1(46-112) alone at 10 µg provoked again a statistically significant neutrophil influx into the peritoneal cavity of 6.29 ± 0.92 × 10⁴ neutrophils per mL (n = 19), compared with 3.95 ± 0.72 × 10⁴ neutrophils per mL in mice injected with PBS (n = 19, P = .020). Combining SAA1(46-112) (10 µg) and SB225002 (25 µg) yielded a decreased intraperitoneal neutrophil influx (4.99 ± 1.41 × 10⁴ neutrophils/mL, n = 20, P = .021).

Next, we chemically synthesized the human equivalent of bo SAA1(46-112), corresponding to the amino acids 47 to 104 of mature hu SAA1α, designated SAA1(47-104) (M_r of 6309; Figure 6A). Compared with hu SAA1, bo SAA1 contains an insertion of 9 aa between residues 69 and 70 of mature hu SAA1.⁴⁵  Therefore, SAA1(47-104) contains 58 instead of 67 aa (Figure 6B). Similar to SAA1(46-112), SAA1(47-104) did not significantly attract CD14⁺ monocytes nor neutrophils in vitro (data not shown). In contrast, SAA1(47-104) at 300 ng/mL (CI = 1.8 ± 0.5, n = 6) or 1000 ng/mL (CI = 1.5 ± 0.4, n = 6) synergized with CXCL8 (1 ng/mL, CI = 20.5 ± 1.9, n = 8), reaching a CI of 40.6 ± 5.3 (n = 6, P = .0051) or 37.9 ± 4.4 (n = 6, P = .0202), respectively (Figure 7A). Moreover, like SAA1(46-112), SAA1(47-104) (300-3000 ng/mL) desensitized the synergy between intact SAA1α (300 ng/mL) and CXCL8 (3 ng/mL) in neutrophil chemotaxis (Figure 7B). In addition, SAA1(47-104) did not induce CCL3 nor CXCL8 in CD14⁺ monocytes (Figure 4A). Equivalently to bo SAA1(46-112), hu SAA1(47-104) (2000 ng/mL; CI = 1.6 ± 0.3) and CCL3 (1 ng/mL; CI = 28.4 ± 0.8) synergized in monocyte migration (CI = 32.9 ± 1.6, n = 6, P = .038).

The use of FPR2 by SAA1(47-104) for exerting chemotactic activity was confirmed via antagonization experiments, whereby this receptor was blocked by the selective FPR2 antagonist WRW₄ (Figure 7C). As shown in Figure 7A, SAA1(47-104) at 300 ng/mL (CI = 2.8 ± 1.2, n = 5) again (Figure 7C) synergized (CI = 62.0 ± 5.6, n = 5, P = .012) with CXCL8 at 1 ng/mL (CI = 33.7 ± 4.8, n = 5) on neutrophils. Treatment of neutrophils with WRW₄ completely blocked this synergistic interaction (CI = 31.4 ± 5.9, n = 5, P = .012). As a control, the chemotactic response of neutrophils toward the FPR2 agonist WKYMVm (CI = 43.5 ± 6.4, n = 5) was inhibited for 60.8 ± 11.4% upon treatment of the cells with WRW₄ (CI = 16.2 ± 3.9, n = 5, P = .0216) (Figure 7C).

Similarly, the synergistic interaction between 1 ng/mL of CCL3 and 3000 ng/mL of SAA1(46-112) on CD14⁺ monocytes (CI = 22.8 ± 1.0, P = .01) was reduced to a CI of 18.0 ± 1.2 in the presence of WRW₄ (n = 5, P = .016). Hence, this synergy was completely blocked, because the CI of CCL3 alone was 18.9 ± 0.4 (n = 5).

In order to identify the protease that generates the SAA1 peptide isolated from bovine serum, we incubated intact hu rSAA1α with matrix metalloproteinase-2 (MMP-2), cathepsin G, plasmin, and thrombin. Incubation (30 minutes, 37°C, enzyme/substrate ratio of 1:50) with cathepsin G resulted in complete degradation of intact SAA1α. No cleavage was observed after treatment (up to 16 hours at 37°C) with plasmin or thrombin. Only MMP-2 was able to cleave intact SAA1α. Incubation with MMP-2 (70% conversion after 1-hour incubation with an enzyme/substrate ratio of 1:50) resulted in formation of rSAA1(52-104) (data not shown), as previously described.^46,47  This physiologically relevant SAA1(52-104) has not yet been biologically characterized. We have therefore also chemically synthesized SAA1(52-104) to be tested for synergy. SAA1(52-104) at 1000 ng/mL (CI = 1.7 ± 0.5) synergized with CXCL8 at 3 ng/mL (CI = 32.6 ± 3.3) in neutrophil chemotaxis (CI = 52 ± 3.2, n = 7, P = .0049) (Figure 7D).

Multiple functions have been ascribed to the acute phase protein SAA, and, currently, the emergence of transgenic mouse models reveal novel insights on the importance of this highly conserved protein in host defense.^48,49  For many of those biological activities, including the stimulation of angiogenesis, its role in cholesterol transport, and its antibacterial activities, high concentrations of SAA are needed. In contrast, already at 30 to 300 ng/mL, SAA chemoattracts monocytes, immature monocyte-derived dendritic cells, and neutrophils and induces chemokines in several cell types.²  Previously, the 1000-fold induction of some acute phase proteins has deviated research toward effects at very high concentrations. We contemplate that the interesting biology of acute phase molecules is at their highest specific activities (ie, at their lowest concentrations, when these molecules get diluted in the vast volume of body fluids).

In previous studies, we have partly unraveled the mode of action of SAA1α to exert its monocyte and neutrophil chemotactic activity, which occurs either directly (via GPCR) or indirectly (via TLR) by autocrine or paracrine chemokine induction.^20,24  Using monocyte chemotaxis assays, we have purified a novel COOH-terminal SAA1 fragment from bovine newborn calf serum. To our knowledge, such fragments of SAA have not been studied in this context before. In contrast, rather than COOH-terminal fragments, NH₂-terminal SAA fragments are frequently reported in the context of amyloidosis.^46,47,50  The proteases responsible for SAA processing are members of the family of the matrix metalloproteinases (ie, MMP-1, -2, and -3), but also other proteinases can cleave SAA, resulting in fragments with different lengths.^47,51,52 

In contrast to the natural peptide, synthetic SAA1(46-112) lacked in vitro chemotactic activity. We are convinced that the synthetic SAA1 peptide fully matches the natural one and that no misfolding of synthetic peptide is involved. Indeed, no complex structural features such as disulfide bridges are present. We postulate that the chemotactic activity of natural SAA1(46-112) is because of synergy with a potent contaminating chemotactic factor, most probably a chemokine (see also supplemental Methods). Potential synergy also explains why SAA1(46-112) has in vivo chemotactic activity. Indeed, we demonstrated that its activity in vivo was reduced by a CXCR2 antagonist.

Although the COOH-terminal SAA1 peptides did not chemoattract human neutrophils in vitro on their own, these fragments potently synergized with CXCL8 to activate neutrophils. This synergy was even more pronounced than that observed between intact SAA1α and CXCL8 (Figure 3C). In neutrophil shape change assays, the synergy between SAA1(46-112) and CXCL8 (Figure 3B) was similar to that between intact SAA1α and CXCL8.²⁴  Intact SAA1α was able to significantly activate neutrophils on its own,²⁴  whereas SAA1(46-112) alone did not induce morphological changes. Hence, the COOH-terminal peptides are less potent than intact SAA1α to activate human neutrophils (chemotaxis, shape change), but are at least as potent as intact SAA1α to synergize with CXCL8. Such phenomenon of synergy in neutrophil chemotaxis was described for the first time by our research group between the chemokines regakine-1 and CXCL8.⁴⁴ 

Synergy between SAA1α and CXCL8 in neutrophil chemotaxis implies binding of these agonists to their cognate GPCR (ie, FPR2 and CXCR1/2, respectively).²⁴  Here, the use of FPR2 by COOH-terminal SAA1 peptides to synergize with CXCL8 or CCL3 in neutrophil or monocyte chemotaxis, respectively, was evidenced by inhibition of this synergy with the selective FPR2 antagonist WRW₄ and by desensitization of the synergy between intact SAA1α and CXCL8 on neutrophils by SAA1(47-104).

In contrast to intact SAA1α²⁰ , SAA1(46-112) and SAA1(47-104) synergized with CCL3 to chemoattract CD14⁺ monocytes (Figure 4B). This is because of the lack of chemokine induction in these cells upon stimulation with SAA1 fragment compared with the strong induction capacity of intact SAA1 (Figure 4A). The lack of chemokine induction by SAA1(46-112) also affects the in vivo mobilization of leukocytes, when compared with intact SAA1α. Coinjection of SAA1(46-112) and murine CXCL6, the functional equivalent of human CXCL8 in mice, into the peritoneal cavity caused a cooperative effect in neutrophil recruitment, which was not observed with intact SAA1α. We hypothesize that this latter outcome is because of cross-desensitization between injected CXCL6 and CXCR2 ligands induced by intact SAA1α. Indeed, in a previous study we pointed out that the in vivo neutrophil recruitment by intact SAA1α was predominantly mediated by the induction of CC and CXC chemokines (eg, in monocytes), which subsequently synergize which each other and with SAA1 to recruit neutrophils to the peritoneal cavity.²⁴  Synergy between CC and CXC chemokines in the attraction of neutrophils has been documented before.^53,54  In vivo enhancement of neutrophil recruitment into the peritoneal cavity has also previously been shown when chemokines were coinjected into the peritoneal cavity of mice.⁵⁵ 

Taken together, endogenous (eg, IL-1 and IL-6) or exogenous (eg, LPS) inflammatory mediators lead to induction of acute phase proteins in the liver. One of the major acute phase proteins in humans is SAA. Because SAA has a short half-life of ∼1 day,^1,56  degeneration of the acute phase protein occurs rapidly after production. SAA breakdown results in generation of circulating fragments of SAA, among which the COOH-terminal peptides described here. Compared with intact SAA1, these fragments have lost their chemokine inducing activity, but have maintained their capacity to synergize with chemokines in leukocyte migration. As such, these COOH-terminal SAA1 fragments can generate a positive and negative feedback loop on leukocyte chemotaxis. In the presence of chemokines, induced by intact SAA1 and other inflammatory mediators, COOH-terminal fragments synergize with these chemokines via binding to FPR2, maintaining the inflammatory response. On the other hand, when both intact SAA1 and COOH-terminal fragment are present, these fragments can desensitize the cooperation between intact SAA1 and chemokines to moderate the inflammatory response. Our data demonstrate that SAA1 and SAA1-derived peptides constitute distinct alarm transmitters, fine-tuning defense mechanisms, and hence SAA1 functions as an alarmin.⁵⁷ 

The authors thank Isabelle Ronsse, Noëmie Pörtner, Lotte Vanbrabant, Sofie Knoops, Maaike Cockx, Rik Janssens, Vincent Vanheule, Jennifer Vandooren, Sarah Abu El-Asrar Salama, and Maitreyi Sadanand Joshi for their technical assistance.

This work was supported by the Research Foundation of Flanders (FWO-Vlaanderen; projects G.0764.14 and G.0D25.17), the Interuniversity Attraction Poles Program initiated by the Belgian Science Policy Office (I.A.P. project P7/40), the Concerted Research Actions (G.O.A., GOA/12/017) and C1 funding (C16/17/010) of the University of Leuven. The Hercules Foundation of the Flemish government provided funding to purchase mass spectrometry equipment (contract AKUL/11/31). M.D.B. is a postdoctoral research fellow of the FWO-Vlaanderen, and M.G. is a “research expert” funded by the Rega Foundation.

Contribution: M.D.B. performed experiments, analyzed data, wrote part of the manuscript, and submitted the manuscript; M.G. gave technical advice, planned and performed experiments, analyzed data, and corrected the manuscript; N.B. performed experiments and analyzed data; G.O. was involved in proteolytic cleavage of SAA and advised on the writing; P.P. executed biochemical quality control of reagents, performed protein synthesis, and corrected the manuscript; S.S. isolated SAA fragments, wrote part of the manuscript, and corrected the manuscript; and J.V.D. designed the study, planned the experiments, analyzed the data, and corrected the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Jo Van Damme, Laboratory of Molecular Immunology, Rega Institute for Medical Research, Herestraat 49, Box 1042, 3000 Leuven, Belgium; e-mail: jo.vandamme@kuleuven.be.