Oxidative modifications of extracellular matrix promote the second wave of inflammation via β₂ integrins

Understanding the mechanism of leukocyte migration is essential for the treatment of chronic inflammation, which is a major factor contributing to many devastating diseases, including arthritis, diabetes, obesity, and atherosclerosis.^1-3  Neutrophil recruitment is the first wave of immune response directed to fight inflammation,⁴  primarily by secreting peroxidases. Generation of excess reactive oxygen and nitrogen species facilitate a speedy inactivation of pathogens while releasing chemotactic signals to promote a second wave of immune response, monocyte/macrophage migration.^5,6  Arriving macrophages play a central role in the resolution of acute inflammation, essentially by removing the debris and promoting tissue healing. However, excessive or uncontrolled macrophage accumulation contributes to chronic inflammation and a number of pathologies.⁷ 

To date, the mechanisms controlling the transition between the first and second wave of inflammation are not fully understood. Although chemokine gradients seem to be critical for macrophage migration, the importance of adhesive receptors and their respective ligands remains questionable. Macrophage receptors, integrins, are the major players in adhesion-mediated migration. Prominent among the leukocyte adhesion receptors are the 4 members of the integrin β₂ subfamily: α_Lβ₂ (CD11a/CD18, LFA-1), α_Mβ₂ (CD11b/CD18, Mac-1), α_Xβ₂ (CD11c/CD18, p150, 95), and α_Dβ₂ (CD11d/CD18).⁸  Although α_X is an important marker of pro-inflammatory macrophage activation, its expression on monocytes/macrophages is low, which reduces the contribution of α_X to macrophage migration.⁹  In contrast, α_Mβ₂ is a major β₂ integrin on macrophages and α_Dβ₂ is upregulated during pro-inflammatory macrophage activation.¹⁰  In our previous studies, we found that α_Mβ₂ and α_Dβ₂ are highly homologous multiligand receptors that share common ligands^11,12  and participate in macrophage migration and retention at the site of inflammation.¹⁰  A typical extracellular matrix (ECM) has a limited availability of ligands for β₂ integrins. The types of adhesive ligands available to mediate inflammation-specific and directed macrophage migration remain to be determined.

One of the possible mechanisms of directed macrophage migration is the modification of existing ECM during inflammation. Oxidation of polyunsaturated fatty acids (PUFA) by reactive oxygen species produced in inflamed tissues might generate protein modifications, which, in turn, might provide macrophages with inflammation-specific ligands. As oxidation substrates, PUFAs are readily available as a part of cellular membranes as well as from dietary sources, and their products were shown to exhibit a wide spectrum of biological activities.^13,14  Despite widely advertised opinion regarding the beneficial role of 3-PUFA (particularly docosahexaenoate [DHA]) for overall health, current results of clinical trials are questioning its protective role for the cardiovascular system.^15-17  Apparently, DHA-derived products generated in vivo may have effects distinct from DHA itself.

2-(ω-Carboxyethyl)pyrrole (CEP) is formed through adduction of the end products of DHA oxidation with the ε-amino groups of protein lysyl residues^18,19  (Figure 1; supplemental Figure 1, available on the Blood Web site). To develop tools for testing CEP function, ω-carboxyethylpyrrole–modified proteins were synthetized using Paal-Knorr reactions of γ-dicarbonyl compounds with the ε-amino group of lysyl residues of proteins.²⁰  γ-Dicarbonyl compounds was used to prepare CEP-modified keyhole limpet hemocyanin (CEP-KLH), bovine serum albumin (CEP-BSA), and human serum albumin). Using these proteins, highly specific monoclonal and polyclonal antibodies against CEP were generated and tested for specificity.¹⁸  Notably, a structurally similar protein modification, ethylpyrrole (EP), is generated through the alternative oxidative cleavage of DHA (Figure 1). Compared with CEP, this modification lacks a carboxyl group, which makes it an excellent control for CEP functional studies.²¹ 

CEP generation was reported to contribute to a number of inflammation-associated diseases, including macular degeneration, hyperlipidemia, atherosclerosis, thrombosis, and tumor progression.^21,22  However, the pro-inflammatory mechanism of CEP function is not clear. Accumulation of CEP in damaged tissue and induction of pro-inflammatory cytokines from macrophages in response to CEP represents the data that may explain the contribution of CEP to the augmentation of inflammatory responses.^23-25  Our investigation was outlined to seek a possible link between CEP and macrophage migration/accumulation at the site of inflammation.

Additional information is available in the supplemental Materials and methods.

Wild-type (WT; C57BL/6J) mice, β₂-deficient mice (B6.129S7-Itgb2^tm2Bay/J), and myeloperoxidase (MPO)-deficient mice (B6.129X1-Mpo^tm1Lus/J) were purchased from Jackson Laboratory (Bar Harbor, ME). All procedures were performed according to animal protocols approved by the Cleveland Clinic and East Tennessee State University Institutional Animal Care and Use Committee.

Adhesion assay was performed as described previously.¹¹  Two-dimensional (2D) cell migration assays were performed under sterile conditions using uncoated 6.5-mm diameter Transwell inserts with a pore size of 8 µm (Costar, Corning, NY), as described in our previous paper.¹² 

Three-dimensional (3D) migration assay was performed as described in our previous paper.¹⁰  Briefly, neutrophils or macrophages were labeled with PKH26 red or PKH67 green fluorescent dyes. Labeled leukocytes migrated through the custom-made fibrin matrix prepared in Transwell inserts for 24 (neutrophils) or 48 (macrophages) hours at 37°C in 5% CO₂. 100 nM N-formylmethionyl-leucyl-phenylalanine (FMLP), or 30 nM monocyte chemoattractant protein-1 (MCP-1) were added on the top of the gel to initiate the migration. Migrating cells were detected by Leica Confocal microscope (Leica-TCS SP8) and the results were analyzed by IMARIS 8.0 software.

Constructs for α_D I-domains, α_M I-domains, and the α_L I-domain in active and nonactive conformations were generated and recombinant proteins were isolated as described in our previous papers.^12,26  Real-time protein–protein interactions were analyzed using a Biacore3000 instrument (Biacore, Uppsala, Sweden) as described previously.^12,21 

Numerous data demonstrate the involvement of CEP in the inflammatory process.^21,23-25  We tested CEP and EP accumulation during acute inflammation in the peritoneal tissue after thioglycollate-induced peritoneal inflammation. Normal tissue is characterized by strong deposition of EP, but was devoid of CEP completely. However, induction of inflammation led to the marked accumulation of CEP in the peritoneal wall (supplemental Figure 2A) that significantly overlaps with macrophage staining after 72 hours (Figure 2 A-B), which indicates a link between CEP and macrophage accumulation during peritoneal inflammation. Anti-CEP monoclonal antibody was injected into the peritoneal cavity of WT mice 30 minutes before and 24 hours after injection of thioglycollate. We found that thioglycollate-induced accumulation of macrophages in the peritoneal cavity after 72 hours was dramatically reduced in the presence of anti-CEP antibody, whereas neutrophil accumulation after either 6 or 18 hours was not affected (Figure 2C; supplemental Figure 2B). Notably, anti-CEP antibody did not affect cell survival according to cell toxicity assay (Dojindo Molecular Technologies, Rockville, MD) (supplemental Figure 2C).

To detect CEP-modified ligands for macrophages within the peritoneal cavity, we collected mouse peritoneal exudates at 72 hours after injection of thioglycollate and performed immunoprecipitation with anti-CEP antibody. The sodium dodecyl sulfate (SDS)-electrophoresis demonstrated 2 bands with the molecular weights ∼70 and ∼49 to 51 kDa (reduced samples) (Figure 2D). Using mass spectrometry we identified 2 major proteins in these bands: fibrinogen and albumin. Serum albumin (69 kDa; 35 peptides, covering 51.3% sequences) and the α-chain of fibrinogen (67 kDa; 8 peptides, covering 15.1% sequences) were detected in the upper band.²⁷  ɤ-Chains of fibrinogen (47 kDa; 2 peptides, 4.8% of sequences) were identified in the lower band (Figure 2D). To confirm the presence of CEP-modified fibrinogen, we performed immunoprecipitation of mouse peritoneal exudates with anti-fibrinogen antibody. A subsequent western blot with anti-CEP antibody revealed a band that corresponds to the molecular weight of the fibrinogen α-chain (Figure 2E). The bands, which correspond to the β and ɣ chains, are masked by the heavy chains of immunoglobulin G. Therefore, peritoneal inflammation led to the formation of CEP-modified fibrinogen and albumin, which are major plasma proteins that leak into the tissue during inflammation.

Neutrophil activation during inflammation promotes the release of MPO and other enzymes leading to generation of reactive oxygen and nitrogen species, causing oxidation of polyunsaturated lipids.²⁸  Polyunsaturated phospholipids are abundant in the membranes of cells, particularly neutrophils.²⁹  We explored the possibility that neutrophil accumulation and activation may lead to generation of CEP-modified proteins within ECM. To test this hypothesis in vitro, a 3-dimensional fibrin matrix was generated in a Boyden chamber. Freshly isolated neutrophils were stained with green fluorescent dye and stimulated by an FMLP gradient to migrate through the fibrin matrix. After overnight migration, the fibrin matrix was stained with anti-CEP antibody (red fluorescence). We detected a dramatic increase in CEP deposition following neutrophil migration (Figure 3A-B). In parallel samples, fibrin gels were digested with plasmin. The generated protein fragments were tested by western blot using an anti-CEP antibody. A strong CEP signal was detected in western blot samples after neutrophil migration. The band at ∼180 kDa (nonreduced samples) corresponds to the molecular weight of the DD-fragment of fibrin, the major product of fibrin degradation by plasmin. The sequences from the β- and ɣ-chains of fibrin were detected in this band by mass spectrometry (β-chain: 10 peptides, 31% sequence coverage; ɣ-chain: 7 peptides, 21.74% sequence coverage). The band at ∼69 kDa corresponds to the molecular weight of bovine albumin, which was also verified by mass spectrometry (32 peptides, 59.31% sequence coverage). Bovine albumin (1% fetal bovine serum) was incorporated in the fibrin matrix for neutrophil support. Surprisingly, the second major product of fibrin degradation, the E-fragment (50 kDa) was not detected by anti-CEP antibodies, which indicates selectivity for protein modification by CEP. Notably, the control fibrin gel had only background staining, which is most likely because of the low level of preexisting modifications present in isolated fibrinogen (Figure 3 A,C). The contribution of neutrophils and macrophages to CEP formation was evaluated using neutrophil depleted (Ly6G antibody) and macrophage depleted (Clodronate-liposome) models. Neutrophil depletion dramatically reduced the formation of CEP in the peritoneal cavity of mice after thioglycollate induced peritoneal inflammation, whereas macrophage depletion had no significant effect. These data confirmed our in vitro results and emphasize the role of neutrophils in CEP generation (supplemental Figure 3).

Because MPO is a major peroxidase enzyme secreted by neutrophils especially upon their activation, we tested its role in CEP generation. Activation of human neutrophils with interleukin-1β (IL-1β; 200 ng/mL) or lipopolysaccharide (100 ng/mL) increased CEP levels in the media by more than twofold, thereby confirming that neutrophil activation promotes CEP generation (Figure 3D). The treatment of neutrophils with the MPO inhibitor, 4-ABH, dramatically reduced the CEP level. 4-ABH was used at a concentration known to have no effect on catalase or glutathione peroxidase activities or on the production of superoxide by neutrophils.³⁰  At the same time, resorcinol at a concentration of 10 µM had no effect on CEP production. It has been shown that at this concentration resorcinol inhibits eosinophil peroxidase but has low effect on MPO.³¹  In contrast to CEP, neither 4-ABH nor resorcinol inhibited the production of EP by neutrophils (supplemental Figure 4). Consistent with the main role for MPO in CEP generation, CEP levels in wounds of MPO-deficient mice were reduced by 50% as compared with WT controls (Figure 3E). To assess whether MPO is able to directly contribute to CEP–protein adduct formation, human fibrinogen was incubated with active recombinant MPO and DHA.³²  Resulting CEP formation was quantified by enzyme-linked immunosorbent assay (supplemental Figure 5A). As anticipated,³³  the formation of CEP–protein adducts required the presence of all 3 main components: DHA (as a lipid substrate), MPO (as a source of oxidation), and a protein (eg, fibrinogen, as a source of lysines). Incubation of fibrinogen with MPO or DHA alone was not sufficient for CEP generation. The presence of CEP adducts on fibrinogen was confirmed by western blot with monoclonal anti-CEP antibody (supplemental Figure 5B). These results clearly demonstrate that MPO-mediated DHA oxidation is one of the main mechanisms for CEP generation. Based on the results with MPO-deficient mice, MPO seems to be a clear source of CEP generation in vivo; however, it appears that other oxidative enzymes might participate in this process.

During macrophage migration, CEP may serve as a chemoattractant, an activator of adhesive receptors, or a direct ligand for adhesive receptors. First, using a 2D chemotactic assay, we found that neither CEP nor EP was able to stimulate macrophage chemotaxis. MCP-1 was used as a positive control, whereas BSA was used as a negative control (supplemental Figure 6A). Next, we found that preincubation with CEP did not increase macrophage adhesion to immobilized fibronectin³⁴  (ligand to different β₁, β₂, and β₃ integrins), which eliminates a potential effect of CEP on integrin activation (supplemental Figure 6B).

Based on these results, we hypothesized that CEP may serve as a direct ligand for macrophage adhesion and migration. Adhesion assay revealed strong binding of peritoneal macrophages to CEP-modified BSA and CEP-modified KLH, whereas adhesion to EP-modified BSA or BSA itself was not detected (Figure 4A). Because leukocyte integrins are the major migratory receptors on the macrophage surface, we evaluated a contribution of 2 macrophage integrin subfamilies β₁ and β₂ on CEP-mediated adhesion. The adhesion to CEP was significantly inhibited by anti-β₂, but not by anti-β₁ antibodies (Figure 4B). These data suggest that CEP may serve as a ligand for β₂-integrins.

The subfamily of β₂ integrins consists of 4 members, α_Mβ₂, α_Dβ₂, α_Lβ₂, and α_Xβ₂. Although α_Mβ₂, α_Dβ₂, and α_Lβ₂ demonstrate strong expression on macrophages, the level of integrin α_Xβ₂ is low, which reduces its potential role in integrin-mediated adhesion and migration. To further confirm the role of β₂ integrins, we used α_Mβ₂-, α_Dβ₂-, and α_Lβ₂-transfected HEK293 cells previously generated in our laboratory (Figure 4C).^11,12  We tested their ability to bind CEP-BSA and found that α_Dβ₂- and α_Mβ₂-transfected, but not α_Lβ₂-transfected or control mock-transfected cells, strongly adhered to CEP (Figure 4D). In contrast, adhesion to BSA or EP was not detected for α_Dβ₂- and α_Mβ₂-transfected cells (Figure 4E-F). HEK 293 cells express endogenous β₁ integrins including α₁β₁, α₂β₁, α₄β₁, and α₅β₁; therefore, the lack of adhesion of mock-transfected HEK293 cells to CEP confirmed integrin β₂ specificity for CEP.

The adhesion of α_Dβ₂- and α_Mβ₂-transfected cells was significantly inhibited by anti-CEP, anti-β₂, and anti-α_D (or anti-α_M) antibodies, but not by anti-β₁ antibody (Figure 4E-F). In addition, we demonstrated that preincubation of α_Mβ₂ and α_Dβ₂ cells with CEP in solution decreased the binding of blocking anti-α_M and anti-α_D antibodies more than twofold (supplemental Figure 7). These data prove the hypothesis that integrin α_Mβ₂ and α_Dβ₂ are receptors for CEP-modified proteins.

It has been shown that the ligand binding sites for most ligands of α_Mβ₂ and α_Dβ₂ integrins reside within the I-domain, the independently folded domain that sits on the top of the extracellular part of the integrin.³⁵  To verify our adhesion results, we expressed recombinant I-domains, α_D in active and inactive conformation, α_M in active and inactive conformation, and α_L in active conformation (Figure 5A) and tested their direct binding to the CEP-BSA using surface plasmon resonance. The binding to EP-BSA was evaluated as a control. The different concentrations of α_D and α_M I-domains flowed over the immobilized CEP or EP. We found that α_D and α_M I-domains in active conformation demonstrated concentration-dependent binding to CEP (Figure 5D-E). In contrast, binding to EP was not detected (Figure 5B-C). The distribution coefficient for CEP–I-domain interactions were calculated using BIAevaluation software (GE Healthcare) and correspond to 2.1 × 10⁻⁶ for α_M and 1.81 × 10⁻⁷ for α_D. This binding was activation dependent because α_M and α_D I-domains in nonactive conformation did not bind to CEP (Figure 5F-G). The specificity of binding was confirmed with blocking anti-α_M antibodies (supplemental Figure 8). Active α_L I-domain did not interact with immobilized CEP (Figure 5G). These data verified our results obtained in adhesion assay with α_Mβ₂, α_Dβ₂, and α_Lβ₂ transfected cells.

To characterize CEP as a migratory substrate for α_Mβ₂ and α_Dβ₂ integrins, we first demonstrated that β₂ deficiency significantly reduced macrophage adhesion to CEP (Figure 6A). Then, we compared the migration of WT and β₂-deficient macrophages within 3D CEP-enriched fibrin matrix. WT cells and β₂-deficient cells were labeled with green fluorescent dye PKH67 and red fluorescent dye PKH26, respectively. Macrophages were mixed at a ratio of 1:1 (Figure 6B), placed on the bottom of a fibrin gel in a Boyden chamber, and then migration against gravity was initiated by MCP-1 added to the top surface of the gel. The migration was evaluated after 48 hours (Figure 6C-D) and revealed that β₂ deficiency dramatically reduced the 3D migration of macrophages into the CEP-enriched matrix (Figure 6E). To exclude the contribution of a particular dye, the experiment was repeated with the opposite labeling conditions and revealed a similar result (supplemental Figure 9). To demonstrate that the difference between WT control and β₂-deficient macrophages was due to mesenchymal rather than amoeboid migration, this experiment was performed in the presence of ROCK inhibitor (Y-27632), known to block the amoeboid (adhesion-independent) component of cell migration. The presence of ROCK inhibitor did not alter the pattern of migration and the difference between WT control and β₂-deficient macrophages remained. This confirms that macrophage migration in 3D CEP-enriched matrix was integrin-dependent, or mesenchymal type (supplemental Figure 10). The quality of ROCK inhibitor was verified using M2-activated macrophages that strongly depend on amoeboid motility (data not shown).

To rule out the potential interplay between WT and β₂^−/− cells in 3D gel, we tested the migration of individual subsets of macrophages in CEP- and EP-enriched fibrin matrices (Figure 7A-B). Similar to the results of the adhesion experiments (Figure 4A), the migration of WT macrophages involving CEP was substantially stronger in comparison with EP. In contrast, the migration of β₂^−/−-deficient macrophages was similar between CEP and EP (Figure 7B), which clearly indicates the critical role of β₂ integrins, primarily α_Mβ₂ and α_Dβ₂ (Figures 4 and 5), in CEP-mediated migration.

Because neutrophils also express α_Mβ₂ and α_Dβ₂ integrins, neutrophil migration might also depend on the presence of CEP within the matrix. We evaluated the migration of human neutrophils within CEP-enriched fibrin gel; however, no additional effects of CEP on neutrophil migration were detected (Figure 7C). These results correspond to the published data showing that, in contrast to macrophages, neutrophil 3D migration is exclusively mediated by the amoeboid mode.³⁶ 

In the present study, we demonstrated that recruited neutrophils secrete reactive oxygen species that modify existing components of ECM by CEP adducts, which in turn serve as adhesive and migratory ligands for macrophage integrins. The following conclusions are drawn from this study. (1) Although CEP is not present in healthy tissues, levels are dramatically increased at the sites of inflammation. (2) Neutrophil activation and migration through ECM results in the generation of CEP-modified proteins. (3) CEP-modified proteins promote macrophage adhesion and migration. (4) Inhibition of CEP prevents macrophage recruitment. (5) Integrins α_Mβ₂ and α_Dβ₂ are specific macrophage receptors for CEP-mediated adhesion and migration. Thus, CEP is a natural inflammatory product generated during the first phase of inflammation by recruited neutrophils, facilitating the second wave of inflammation, namely, the recruitment of macrophages. By means of oxidation, neutrophils seem to “pave the road” for future macrophage invasion by modifying ECM with CEP.

CEP was initially described in the retina of patients with age-related macular degeneration.¹⁸  Later, however, the link between inflammation and CEP was demonstrated in several pathologies associated with chronic inflammation including atherosclerosis, tumor progression, aging, and others.^22,26,37-39  Several groups reported the secretion of pro-inflammatory mediators from macrophages after engagement with CEP-modified proteins.^21,23-25  In agreement with this, CEP induced M1 macrophage polarization.⁴⁰  Therefore, CEP function is closely related to the development of inflammation.

Previously, several receptors were detected for CEP on the surface of macrophages, including TLR2, CD36, TLR1, and TLR6.^21,38  Although these receptors contribute to macrophage activation and reprogramming by CEP, they do not directly mediate macrophage migration. Our data demonstrated that CEP is not able to serve as a chemoattractant for macrophages or as an integrin activation agonist. These results suggest that macrophages use another receptor for CEP that can mediate migration.

A subfamily of β₂ integrins are the major adhesive receptors on the surface of macrophages. Since α_Lβ₂ interacts only with cell counterreceptors and the expression of α_Xβ₂ on macrophages is low,⁹  2 other members of the subfamily, α_Mβ₂ and α_Dβ₂, are the best candidates for macrophage migration and retention in the extracellular matrix. α_Dβ₂ and α_Mβ₂ recognize a wide range of ligands, including fibronectin, thrombospondin, and vitronectin.^11,12  CEP modifications seem to generate the new inflammation-specific substrate for α_Mβ₂- and α_Dβ₂-mediated macrophage migration or retention. Importantly, even known β₂ integrin ligands such as fibrinogen can be “improved” by CEP modification. Inflammation promotes the leakage of fibrinogen from blood to the inflamed tissue. Importantly, although the affinity of soluble fibrinogen for β₂ integrins is low, it increases substantially upon immobilization or partial digestion because of the exposure of carboxyl groups of glutamic and aspartic acid.^41,42  Acidic side chains of ligands are required for the binding to Mg²⁺ in the MIDAS motif within the integrin I-domain, which is critical for integrin ligand recognition.⁴³  Likewise, the modification of fibrin with CEP (via the side-chain of lysine) reduces the positive charge on the fibrin surface and at the same time increases the number of negatively charged carboxyl groups, which are a major active element within the CEP structure. Thus, CEP adducts modify fibrinogen into the high-affinity ligand for α_Mβ₂ and α_Dβ₂ integrins. Indeed, we show that CEP but not EP, which modifies proteins in a similar manner but lacks a carboxyl group, is able to create an adhesive β₂ ligand out of many proteins, possibly imitating bacterial ligands by exposed carboxyl groups as proposed previously. In general, the high distribution coefficient for the α_M-CEP and α_D-CEP interactions demonstrate a strong affinity that exceeds the affinity for the binding of α_M or α_D I-domains to previously identified proteins.¹²  This makes CEP-modified proteins a preferential ligand for the macrophages. We found that the affinity for α_D-CEP interaction surpasses the affinity for α_M-CEP approximately 10-fold. We suggest that this effect is mediated by a difference in the electrostatic surfaces between the α_D I-domain and α_M I-domain.⁴⁴  Therefore, integrin α_Dβ₂ may play a more significant physiological role in the CEP-mediated macrophage adhesion and migration. Further studies are required to clarify this question.

It is well accepted that strong integrin-mediated adhesion often antagonizes or limits migration.^45,46  Because CEP-modified proteins serve as strong ligands for β₂ on macrophages, CEP might serve as a cell migration signal for cells with moderate integrin expression and as a retention/arrest signal for cells with the high integrin expression. We recently demonstrated that expression of α_Dβ₂ is upregulated on M1 macrophages in vitro and within atherosclerotic lesions. We showed that high expression of α_Dβ₂ mediates the retention of macrophages at the site of inflammation.¹⁰  Thus, CEP-modified proteins are the likely ligands responsible for these phenomena.

Recent studies indicate that rapid migration of neutrophils is mediated in an integrin-independent manner by an amoeboid mode of motility.³⁶  Our data correspond to these results (Figure 7C) and confirm that neutrophils do not need to use CEP as substrate for their migration. Importantly, it has been demonstrated that macrophages can use both mechanisms, amoeboid and mesenchymal migration, depending on environment and macrophage subset.⁴⁷  This, again, implies the selectivity of CEP ligand specificity toward macrophages.

Macrophage accumulation serves as defensive mechanism during acute inflammation⁴⁸  or wound healing,⁴⁹  but has a strong pathological outcome during the development of chronic inflammation.⁵⁰  Therefore, CEP generation may have pro-inflammatory as well as protective functions, depending on type of inflammation. The information obtained in our studies not only establishes the foundation for a new model of inflammation but also provides a new strategy for treatment of chronic inflammatory diseases. The advantage of CEP as a new therapeutic target resides in its unique formation in inflamed tissue. Therefore, the blocking of CEP in peripheral tissues during different chronic inflammatory diseases can prevent macrophage accumulation and further development of chronic inflammation.

These studies were supported by National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases grant DK102020 (V.P.Y.), National Institutes of Health, National Heart, Lung, and Blood Institute grants HL077213 (E.A.P.), HL126738 (E.A.P.), and HL071625 (T.V.B.), National Institutes of Health, National Eye Institute grant EY016813 (R.G.S.), National Institutes of Health grant C06RR0306551 for East Tennessee State University, and by the Canova endowment fund (T.V.B.).

Contribution: V.P.Y. designed the research, performed the experiments, analyzed the data, and wrote the manuscript; K.C., C.L.A., D.G., K.E.B., S.S., and X.Z.W. performed the experiments and analyzed the data; R.G.S. provided 2-(ω-carboxyethyl)pyrrole– and ethylpyrrole-modified proteins, anti-CEP polyclonal antibody, and edited the manuscript; E.A.P. designed the research and analyzed the data; and T.V.B. designed the research, analyzed the data, and edited the manuscript.

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

Correspondence: Valentin P. Yakubenko, Department of Biomedical Sciences, Quillen College of Medicine, East Tennessee State University, PO Box 70582, Johnson City, TN 37614; e-mail: yakubenko@etsu.edu.