The inflammatory macrophagea story of Jekyll and Hyde

Clinical Science (2003) 104, 27–38 (Printed in Great Britain) 27 GLAXOSMITHKLINE/MRS PAPER The inflammatory macrophage: a story of Jekyll and Hyde* Jeremy S. DUFFIELD Medical Research Council Centre for Inflammation Research, Medical School, University of Edinburgh, Teviot Place, Edinburgh EH8 9AG, Scotland, U.K. A B S T R A C T Recent investigations have highlighted new roles for the macrophage (Mφ) in the biology of inflammation. Selective depletion of Mφs from inflamed sites has confirmed their predominant role in immune-mediated damage. The components of this injury have been dissected. Mφs mediate death of stromal, parenchymal and other immune cells by engaging the death programme, resulting in apoptosis. In addition, Mφs induce destruction of matrix and extracellular structures both directly and indirectly by inducing stromal cells to release matrix metalloproteinases. However, there is another side to the inflammatory Mφ. Evidence is provided that Mφs at the same sites possess the ability to aid cell proliferation, secrete and stabilize new matrix components and induce resident cells to secrete matrix components themselves. Mφ phagocytosis of apoptotic cells brings about a change from the cell-killing matrix-degrading cell to the matrix-generating cell-proliferating tissue-healing cell. Just as both Mφ types are necessary at the inflamed site, the right balance of these two populations is required for healing and resolution. Evidence of excessive inflammation as a manifestation of impaired phagocytosis of apoptotic cells emphasizes that defects in the transition from one Mφ type to another may account for the uncontrolled excessive inflammation seen in disease. Recent insights into the mechanisms by which apoptotic cells signal the change of function to the Mφ offer the prospect of novel targets for manipulation of Mφs in the inflamed tissue. INTRODUCTION Inflammation is characterized by the presence of the inflammatory macrophage (Mφ), be it simple wounded skin or complex autoimmune inflammation, such as Type I diabetes mellitus. It is also characterized by the influx of other inflammatory cells, such as neutrophils and lymphocytes. The former die mainly by apoptosis at the inflamed site, whereas the latter both proliferate and die at the inflamed site. Parenchymal and stromal cells start to proliferate and die by apoptosis or sometimes necrosis. Stromal or mesenchymal cells become migratory and deposit excessive pathological matrix. Normal tissue structure and architecture is often destroyed, leading to organ dysfunction. Yet, amongst all this chaotic activity, there is evidence of co-ordination, since apparently severe * This paper was awarded the Sue McCarthy Prize at the GlaxoSmithKline\MRS Young Investigator sessions at the MRS Meeting, Royal College of Physicians, London, on 4 February 2002. Key words : apoptosis, inflammation, macrophage, phagocytosis, repair, scarring. Abbreviations : C1q, complement factor 1q ; Fc-γR, Fc-γ receptor ; IFN, interferon ; IL, interleukin ; LPS, lipopolysaccharide ; Mφ, macrophage ; MCP-1, Mφ chemoattractant protein-1 ; MMP, matrix metalloproteinase ; OPN, osteopontin ; PS, phosphatidylserine ; PSR, PS receptor ; SH2, Src homology 2 ; TIMP, tissue inhibitors of metalloproteinase ; TGF, transforming growth factor ; TNF, tumour necrosis factor ; VEGF, vascular endothelial growth factor. Correspondence : Dr Jeremy S. Duffield (e-mail j.duffield!ed.ac.uk). # 2003 The Biochemical Society and the Medical Research Society 28 J. S. Duffield inflammation can resolve, leaving histologically and functionally normal structures. So why do some types of inflammation resolve, some ‘ burn-out ’ with persistent scarring, whereas others are characterized by persistent inflammatory cell infiltrate ? Recently, it has become apparent that the Mφ is a major player in the inflammatory response. Furthermore, evidence has accumulated that there is more than one type of inflammatory Mφ ; indeed, there may be different populations of Mφ sub-serving different functions simultaneously. Newly described roles for the inflammatory Mφ are often diverse, sometimes contradictory, and in conflict with other roles. Initially, the inflammatory Mφ was shown, like the neutrophil, to produce cytotoxic mediators injurious to eukaryotic cells as well as prokaryotic cells. In addition to the capacity of Mφs to phagocytose debris, it was shown that Mφs phagocytose cells dying by apoptosis [1]. Mφs have also been shown to produce a myriad of cytokines, small, often soluble, peptide molecules that are the effectors of Mφ function. In the present paper, evidence is provided that a key role for the inflammatory Mφ is killing by apoptosis and clearance of both infiltrating leucocytes and proliferating resident stromal and parenchymal cells. Furthermore, ingestion by Mφs of the cells they have killed leads to dampening down of the pro-inflammatory activities of the Mφ, altering the Mφ to a cell that has the capacity to initiate tissue healing and repair. Table 1 Mφs Stimuli for classically and alternatively activated Terms in parentheses represent the receptors through which the stimuli act. TLR, toll-like receptors ; FcR, Fc-receptors ; CR, complement receptors. Mφ activation Classical Stimulus IFN-γ plus pro-inflammatory cytokines Bacterial lipoproteins (TLRs) Bacterial DNA (TLRs) Parasitic proteins/carbohydrates (TLRs) Opsonized particles (FcR, CR) Hypoxia Abnormal matrix IL-4 IL-10 IL-13 TGF-β Glucocorticoids Alternative Mφ ACTIVATION STATES Monocytes are the circulating precursors of tissue Mφs. They lack phagocytic capacity and their cell surface receptor\ligand complement renders them relatively inert whilst in the blood compartment. In vivo, they rapidly cross the endothelial barrier using selectins, integrins and other receptors of the immunoglobulin superfamily to initially role on the endothelium, then become tethered, and finally diapedese [2,3]. At the inflamed site, activated endothelial, epithelial and mesenchymal cells of the tissue release chemokines and cytokines, which signal to the naı$ ve monocyte. Binding of immune complexes or phagocytosis of opsonized particles and debris triggers activation through receptor signalling. Activating cues are also received from lymphocytes and disturbed matrix [4,5]. Products of invading organisms, as might be found in wounded tissue, also potently stimulate activation through specific receptor– ligand interactions. Once the monocyte leaves the circulation, it differentiates into the Mφ (defined by its capacity to phagocytose). Naı$ ve Mφs entering the inflamed site develop one of at least several phenotypic states depending on the array of information sensed at the cell surface by such receptor binding. Collectively these phenotypic states are known as classical activation (Table # 2003 The Biochemical Society and the Medical Research Society 1). Many factors can push the Mφ toward the classical activation phenotype. Ligation of chemokine receptors [such as those for Mφ chemoattractant protein-1 (MCP1)], exposure to pathological collagen deposition and hypoxia can all activate the Mφ [4–7]. There are, however, important differences in the phenotype of Mφs activated by different stimuli. For example, binding of the Th1 lymphokine interferon (IFN)-γ has been shown [8] to augment the activation of Mφs in response to bacterial products, such as lipopolysaccharide (LPS), pro-inflammatory cytokines, including tumour necrosis factor (TNF)-α, interleukin (IL)-1 and IL-12, or activated T-lymphocytes through CD40 ligand. The absolute amount of cytokines and toxic nitrogen radicals released by Mφs that have been exposed to IFN-γ and one of those activating stimuli is many fold higher than Mφs that have been exposed to activating stimuli without IFN-γ . Mφs can be activated by immune complexes signalling through the Fc-γ receptors (Fc-γRs), but, instead of NO production, toxic oxygen radicals predominate. Instead of high IL-12 release (known to be necessary in the development of cell-mediated immune responses) like when Mφs are treated with LPS, Fc-γR1 activation results in the release of low levels of this cytokine [9,10]. Stimulation of Mφs with IFN-γ also upregulates expression of those receptors necessary for effective presentation of antigens to the adaptive immune system. Thus differences in the pattern of cytokine release by Mφs may translate into differences in function, rendering the term ‘ classical activation ’ of Mφs an oversimplification. Not all cytokines, however, cause Mφs to become classically activated. Exposure of naı$ ve Mφs to Th2 lymphokines, such as IL-4, or the cytokine transforming growth factor (TGF)-β prevents the Mφ inflammatory " phenotype, even when such Mφs are challenged with pro- The inflammatory macrophage 29 inflammatory cytokines. These cytokines cause deactivation. In fact, several other cytokines (IL-10 and IL-13) and glucocorticoids have the same effect. The term deactivation is inadequate, however, since Mφs exposed to these ‘ non-inflammatory ’ cytokines do not remain naı$ ve or inert ; instead, they show enhanced capacity for antigen presentation and enhanced phagocytosis of debris and particles, though not pathogens. The ‘ alternatively activated ’ Mφ generates anti-inflammatory cytokines, suppresses synthesis of pro-inflammatory cytokines and is resistant to re-activation [8,11,12]. MR HYDE : INDUCTION OF TISSUE INJURY Tissue injury is a rather vague term without precise definition. It leads to organ dysfunction and damage of cellular structures, but might also imply the induction of cellular death. It indicates damage to the tissue architecture and therefore not only cells, but the extracellular matrix. The ability of Mφs to induce tissue injury is well established. In models of inflammation, such as severe glomerulonephritis in the kidney (induced by deposition of antibody on the basement membrane of highly specialized glomerular capillaries), depletion of Mφs in the early stages of disease brought about amelioration [13]. Kupffer cell depletion in three models of liver injury using liposomal clodronate attenuated disease in all cases [14]. In the T-cell-dependent autoimmune model of multiple sclerosis, experimental allergic encephalomyelitis (‘ EAE ’), depletion of Mφs prevented relapse completely. T-cells were still present, indicating that infiltrat- ing Mφs were crucial effectors of tissue injury [15]. The Mφ as the effector of tissue injury is supported further from depletion studies in allograft and tumour rejection [16,17]. Interestingly, however, Mφ depletion during an animal model of autoimmune diabetes not only blocked disease, but prevented the maturation of effector cytotoxic T-cells, due to an absence of the Mφ-derived cytokine IL-12 [18] (Figure 1). More than 30 years ago it was reported that cells can die by a controlled programme of death called apoptosis [19], unlike the chaotic destruction of necrosis. The principal differences between apoptosis and necrosis are both functional and morphological. Functionally, apoptotic cells maintain integrity of the plasma membrane and specifically disable many intracellular signalling and homoeostatic systems. Furthermore, DNA is specifically cleaved to render it unusable. These processes might be seen as non-inflammatory in that toxic intracellular vesicles are not released and DNA is safely disposed of [20]. Dramatic increases in apoptosis of cells at the inflamed site have been described [1,19,21,22]. Some of these apoptotic cells are neutrophils, inflammatory cells that are pre-programmed to die as soon as they exit the bone marrow, where they are produced in large numbers. Many are cells derived from the inflamed tissue itself, including resident stromal and parenchymal cells [23–29]. In the models of severe inflammation described above, there is evidence of dramatic increases in apoptosis [14,26,30–33]. What is the significance of these apoptotic cells ? If the loss of resident cells outstrips the generation of new cells by proliferation, the outcome is irrevocable loss of tissue, hypocellularity and failure of organ Figure 1 Diverse and disparate functions of the Mφ Monocytes can differentiate into classically or alternatively activated Mφs. Their respective functions are shown. Overlapping functions are also indicated. # 2003 The Biochemical Society and the Medical Research Society 30 J. S. Duffield Figure 2 Occurrence of Mφs and apoptotic mesangial cells during resolving Thy1.1 nephritis in the rat Thy 1.1 nephritis was induced in rats by using a standard protocol [23]. Kidneys were fixed for analysis at various time points until 14 days and 50 glomeruli per kidney were assessed. Apoptotic mesangial cells (i10) were labelled by terminal transferase deoxytidyl uridine end labelling (‘ TUNEL ’) staining and macrophages were labelled by anti-Mφ antibody (ED-1) immunostaining. Total cell number was assessed by haematoxylin and eosin staining. Values are means, with n l 6 for each time point. gcs, glomerular cross section. function [23,26,32]. The corollary of this is also true. Thus perhaps one of the reasons that some inflammation progresses to organ failure is that there may be too much cell loss by apoptosis. The frequency of apoptosis of resident cells correlates closely with the presence of Mφs. Studies of inflammation in resolving and also progressive nephritis show that apoptosis and Mφ influx are strongly correlated [23,33, 34], as has been found in other inflammatory models, such as hepatic fibrosis ( J. S. Duffield and J. P. Iredale, unpublished work) (Figure 2). In addition, apoptosis was seen mainly in mesenchymal cells, which produce an abundance of extracellular matrix and also have both contractile and migratory properties, sharing many features with the myofibroblast, which is known to be responsible for laying down scar tissue at the inflamed site. Importantly, the frequency of apoptotic myofibroTable 2 blast cells in progressive non-resolving kidney disease was substantially higher than that seen in milder resolving disease [23,33,34]. The implication is that in progressive disease there is indeed too much death of resident cells (see later). The correlation between the presence of Mφs and the occurrence of apoptosis in kidney and other inflammatory diseases [24,35] suggests that Mφs might induce death of resident parenchymal and stromal cells. This has been borne out in recent studies. Mφs extracted from inflamed glomeruli were able to bring about apoptosis of mesangial cells, a myofibroblast-like cell in the kidney. Similar observations have been made when classically activated Mφs were cultured with proximal tubular cells from the kidney, pancreatic β cells, vascular smooth muscle cells from atherosclerotic plaques, myofibroblastlike stellate cells in liver fibrosis and elsewhere (Table 2). Importantly, classically activated Mφs also appear to be able to regulate populations of other inflammatory leucocytes by inducing apoptotic cell death of neutrophils and lymphocytes (Table 2). What mechanisms do the inflammatory Mφs use ? Mφs have an arsenal of cytokines and free radicals that have been described as being pro-apoptotic to different cells. These range from the bactericidal NO, oxygen radicals and H O to specific cytokine ligand–receptor interac# # tions. Some of the Mφ-derived cytokines are pro-survival and pro-proliferative in some contexts, whereas in others they promote cell death. The development of specific blocking antibodies and genetically targeted knockout mice has helped immensely in deciphering which of these factors is (are) crucial for mediating cell death. In the kidney, death of both mesangial and tubular cells was induced by Mφ membrane-bound TNF-α. In rodents, the cytotoxic mediator NO was also able to potently induce kidney cell death [36–38]. In contrast, when Mφs were activated with opsonized particles, H O was # # responsible for bringing about apoptotic death [39]. The Mφ appears to employ different death-inducing factors in different tissues. For example, vascular smooth muscle cells are killed by ligation of the death receptor Fas by membrane Fas ligand, whereas neutrophils are hastened Induction of resident cell apoptosis by Mφs in inflammation EAE, experimental allergic encephalomyelitis ; GN, glomerulo-nephritis ; UUO unilateral ureteric obstruction. Organ Kidney Liver Vasculature Lung Skin Pancreatic Islets Central nervous system Immune cell interactions Disease GN, UUO Cirrhosis, hepatic failure Atherosclerosis Wounding Diabetes EAE Cell-type Mesangial, tubular Stellate, hepatocyte Smooth muscle Fibroblasts Fibroblasts, neutrophils β Cells T-cells Lymphocytes Authors Duffield et al. [36], Tesch et al. [34], and Lange-Sperandio et al. [50] Schumann et al. [14], and J. S. Duffield and J. P. Iredale (unpublished work) Boyle et al. [40] Polunovsky et al. [83] Desmouliere et al. [24], and Meszaros et al. [35,84] Jun et al. [18], Suk et al. [85], and Thomas and Kay [86] Pender and Rist [87] Saio et al. [88], and Brown and Savill [41] # 2003 The Biochemical Society and the Medical Research Society The inflammatory macrophage 31 to apoptotic death by the Mφ release of the soluble form of Fas ligand [40,41]. Tissue injury, however, is not only cell death. The matrix supporting the cellular architecture is progressively destroyed, resulting in collapse and distortion of normal structures. Many matrix components are crucial to normal tissue function and homoeostasis. One example of a specialized extracellular protein matrix is the glomerular basement membrane in the kidney that acts as a molecular sieve. During disease, this sieve malfunctions, resulting in leakage of plasma proteins and erythrocytes into the urine. The classically activated inflammatory Mφ plays a crucial role in bringing about matrix destruction through the production of matrix metalloproteinases (MMPs) both directly and indirectly. For example, classically activated Mφs up-regulate the activity of MMPs with gelatinase, caseinase and elastinase activity ; in particular MMP-9, MMP-2 (gelatinases), MMP-12 (metalloelastase) [42] and MMP-7 (matrilysin) [43]. Mφs are also known to release collagenases (MMP-1 and MMP-13) in particular circumstances [44]. In co-culture assays, Mφs can induce myofibroblasts to release collagenases. Thus particulateactivated Mφs induce lung fibroblasts in the rat to generate collagenase-3 (MMP-13) activity, whereas fibroblasts alone did not produce this collagen-degrading enzyme [45]. In studies of kidney disease, inflammatory Mφs of the glomeruli induced myofibroblast mesangial cells to produce stromelysin (MMP-3) [46]. Thus the classically activated Mφ may bring about degradation of normal and abnormal matrix. There is evidence that matrix destruction is not a linear unidirectional process, rather a dynamic turnover. During the early phases of normal wound healing, both Mφs and myofibroblasts express tissue inhibitors of metalloproteinases (TIMPs). These natural inhibitors of the MMPs bind specifically and block MMP activity. TIMP1 and -3 are up-regulated early in the injury response [47]. In other inflammatory models, such as ureteric obstruction, Mφs early in disease express MMP-1 and -2, but simultaneously express TIMP-1 and -2 [48]. How exactly Mφ-directed matrix degradation and its inhibition balance out has not been elucidated. However, conditioned medium from quiescent rat alveolar Mφs does not have overall proteolytic activity in enzymography assays, whereas medium from classically activated Mφs does. Furthermore, in whole tissue assays (i.e. those including the resident and inflammatory cells), expression of TIMP-1 predominates over MMPs in liver fibrosis during the early stages of disease, implying that overall proteolytic activity of tissue Mφs is inhibited, though the matrix may be turned over. Later, however, during the resolution phase of liver fibrosis, expression of the collagenase MMP1\13 predominates [25], suggesting that during this later phase Mφ function switches and overall proteolytic activity predominates. Thus the classically activated Mφ, analogous to Mr Hyde, kills inflammatory, stromal and parenchymal cells by apoptosis. In addition, this cell-type has overall matrix proteolytic activity, releasing many enzymes that break down extracellular structures. Why should such a Mφ exist ? It is likely that, since the inflammatory process functions to keep invading organisms from entering the host as well as bring about repair, a cell-killing Mφ that is able to migrate easily through matrix structures is essential for the first phase of wound healing ; that is containment and clearance of debris. It is also clear that if proliferation of resident cells were not kept in check the consequences might be even greater damage to the tissue. DR JEKYLL : EXTRACELLULAR MATRIX REMODELLING AND RESOLUTION OF INFLAMMATION Just as Mφ-depletion studies have shown a clear role for Mφs in killing cells and damaging extracellular structures, in other, often milder, resolving models of inflammation, Mφ-depletion has had different results. In the healing wound model, Leibovich and Ross [49] depleted Mφs using anti-Mφ serum and demonstrated decreased matrix production and fibrosis, indicating that Mφs were responsible for laying down matrix. In recent studies by Lange-Sperandio et al. [50], Mφs were unable to transmigrate and enter the injured renal interstitium following ureteric obstruction in mice lacking selectin, cell-surface molecules important in diapedesis. These selectin-deficient animals showed a much diminished laying down of scar tissue as well as a reduction in tubular cell apoptosis [50]. These observations have been corroborated in our own department. De Heer et al. [51], investigating the self-resolving model of rat Thy1.1 glomerulonephritis, found that in some rodent strains there were many influxing Mφs, whereas in others far fewer. The presence of Mφs correlated with fewer mesangial (myofibroblast) cells, but more matrix deposition. However, those strains with more Mφs showed resolution of disease, in contrast with those with fewer Mφs, which had persistent hypercellularity. The latter did not recover and went on to have permanent renal scarring. This interpretation of Mφ function was supported by Mφ-depletion studies using clodronate in the same disease model [52]. Thus such studies indicate that Mφs, or at least a population of Mφs at the inflamed area, are responsible for generating matrix components. This is in direct contrast with studies by Tipping et al. [13] who depleted Mφs in another model of kidney inflammation, progressive nephrotoxic nephritis. In these studies, depletion of Mφs attenuated inflammation and prevented proteinuria [13,53]. # 2003 The Biochemical Society and the Medical Research Society 32 J. S. Duffield Table 3 Induction of resident cell proliferation by Mφs in inflammation Organ Kidney Liver Vasculature Lung Skin Disease Nephrosis Partial hepatectomy Atherosclerosis Wounding Cell-type Mesangial, interstitial fibroblast Hepatocyte Smooth muscle Fibroblasts Fibroblasts Authors Diamond et al. [89], and Sharma et al. [90] Rai et al. [91] Ross et al. [62] Cao et al. [61], and Kodelja et al. [92] Hunt et al. [93] Direct evidence for Mφ-mediated matrix deposition has been derived from in vitro studies : when Mφs were alternatively activated by IL-4 or the cytokine TGF-β, followed by co-culture with myofibroblasts, the latter increased fibronectin production, proliferated more and produced more collagen I [43,54]. IL-4-activated Mφs themselves produced more fibronectin and other matrix proteins, such as TGF-β-inducible gene H3 (‘ βIG-H3 ’) [55]. Furthermore, Mφs in healing wounds highly expressed tissue transglutaminase [56], which cross-links matrix proteins, such as fibronectin, collagen, fibrinogen, laminin and osteopontin (OPN), rendering the polymeric structures resistant to breakdown by proteases. In rodent injury models of the myocardium, skin and other tissues (by freezing), a population of influxing Mφs abundantly produced the matrix protein OPN [57]. Its production by Mφs is limited to the early phase of the injury response. OPN is not only involved in normal fibrillogenesis of collagen, but also Mφ recruitment, maintaining Mφ activation and down-regulating enzymes which breakdown matrix, namely MMPs. In support of a role for TGF-β in signalling to myofibroblasts to generate new matrix, two recent studies of wound healing in knockout animals lacking either TGF-β or smad-3 (one " of the components involved in TGF-β receptor signalling) indicated Mφ-derived TGF-β as necessary for matrix deposition in the normal healing process [58,59]. The in vitro studies of Song et al. [43] and Pierce et al. [54] indicated that the alternatively activated Mφ can induce myofibroblast-like stromal cells to proliferate. Corroborative evidence for this comes from several studies in different organs (Table 3). Analysis of lowgrade inflammation in the kidney emphasizes the point : the presence of Mφs in the tissue correlated with mesangial (myofibroblast) proliferation [60]. Song et al. [43] and others [54,61,62] were able to confirm that Mφderived platelet-derived growth factor (‘ PDGF ’) and insulin-like growth factor (‘ IGF ’) were important prosurvival and pro-proliferative factors for myofibroblasts. There is also data to suggest that Mφs are important in angiogenesis, since they generate a multitude of angiogenic soluble cytokines, including vascular endothelial growth factor (VEGF) [63], basic fibroblast growth factor (‘ bFGF ’), insulin-like growth factor-1, TGF-α and TNF-α [64,65]. The importance of angiogenesis in normal healing of the injured tissue has recently been # 2003 The Biochemical Society and the Medical Research Society emphasized in the rat Thy1.1 model of nephritis that normally resolves. Blockade of the potent angiogenic cytokine VEGF was associated with the failure of healing and progression to scarring ensued instead [66]. Although not exclusively produced by Mφs, they represent a potent source of this cytokine. Collectively, these studies point to a different population of Mφs in those models of inflammation where healing is occurring, and indicate a role for Mφs in healing and repair. Thus the alternatively activated Mφ is analogous to the good Dr Jekyll. MR HYDE CAN CHANGE INTO DR JEKYLL Inflammatory and resident Mφs are phagocytic and are able to phagocytose debris, invading organisms and opsonized particles. They have another role in the inflamed tissue, namely the clearance of apoptotic cells. As described earlier, inflammatory and resident cells die by apoptosis. Mφs limit inflammation by rapidly clearing apoptotic cells, such as neutrophils, which, if left in the tissue, would undergo secondary necrosis, spilling proinflammatory proteases into the inflammatory milieu [1,20,67]. When Mφs ingest opsonized particles or foreign agents, this triggers a pro-inflammatory response. Ingestion of apoptotic cells does not trigger activation. Indeed, recent evidence indicates that Mφ ingestion of apoptotic cells is more than non-activating Mφs, it is deactivating. Several in vitro studies have now demonstrated that when Mφs ingest apoptotic cells, be they neutrophils, lymphocytes or stromal cells, such as mesangial cells, they become refractory to the production and release of pro-inflammatory mediators, including IL-1β, IL-6, IL-12, TNF-α, MCP-1 and Mφ-inhibitory protein (‘ MIP ’)-1α [37,68–70]. The same Mφs were not inert, but produced anti-inflammatory mediators, such as TGF-β, IL-10 (in some studies) and prostaglandin E (‘ PGE ’). # # In our own studies of models of inflammation in the kidney [36,37], resident myofibroblast mesangial cells were killed (by apoptosis) by inflammatory Mφs. Those Mφs phagocytosed the apoptotic mesangial cells and were inhibited from further killing (Figure 3). This process could be seen as analogous to a carnivore preying on an unsuspecting animal, killing it and devouring it, which leads to satiety. The carnivore no longer desires or The inflammatory macrophage 33 Figure 3 Phagocytosis of apoptotic cells inhibits inflammatory Mφs from inducing death of myofibroblast mesangial cells Mφs were differentiated from bone marrow using Mφ-colony-stimulating factor (‘ M-CSF ’). The purified Mφs were then cultured with mesangial cells (MC) in a 1 : 1.5 ratio. The co-culture was overlaid with pre-prepared apoptotic cells or similar sized latex beads. After incubation for 4 h, loose apoptotic cells were removed and the co-culture was activated with pro-inflammatory cytokines (IFN-γ and LPS). MC apoptosis was scored at 24 h and compared with MC apoptosis in MC cultures exposed to apoptotic cells and cytokines in the absence of Mφs. ***P 0.001 compared with ingestion of beads. is able to kill. Thus it seems that the process of inducing apoptosis in resident cells with subsequent binding and phagocytosis of the dead cells ‘ switches off ’ the Mφ. The switching off of Mφ-mediated killing in this model of kidney inflammation coincided with down-regulation of TNF-α release and up-regulation of TGFβ release [37]. " Furthermore, the effect was long lasting ; 48 h after ingestion of apoptotic cells, Mφs were incapable of responding to inflammatory stimuli ( J. S. Duffield, unpublished work). The immunosuppressive effect of apoptotic cells on Mφs has now been reported in animal models of multiple sclerosis [71], rheumatoid arthritis, [72], lung inflammation [73] and Trypanosome infection of the peritoneum [69], emphasizing its widespread implication in inflammation. It is important to emphasise that this switch from classically to alternatively activated cell-type is not complete. The alternatively activated Mφ does not produce NO, but releases IL-10 in abundance. However, in our studies [37], ingestion of apoptotic cells neither downregulated NO nor promoted IL-10 generation. Thus the Mφ that has ingested apoptotic cells exhibits many features of the alternatively activated Mφ, but it is not functionally identical. In the investigation by Voll et al. [70] of human monocytes binding apoptotic lymphocytes, monocytes behaved similarly to Mφs in that binding of apoptotic cells inhibited the activation normally induced by LPS. The monocytes also released anti-inflammatory cytokines. Since monocytes have little capacity to phagocytose particles or cells, it may be that binding alone of apoptotic cells is sufficient to signal to the monocytes\ Mφs to bring about the change in phenotype. Figure 4 Some of the receptor–ligand interactions occurring when a Mφ binds an apoptotic cell When an apoptotic cells engages with a Mφ several receptor–ligand interactions occur. New epitopes on the apoptotic cell surface, such as PS exposure, result in binding to the PSR. Intercellular adhesion molecule (‘ ICAM ’)-3 binds to CD14 and unknown ligands bind to the scavenger receptor CD36 and via a molecular bridge the integrin αvβ3 binds. CD31 expressed on live cells is repellent to CD31 on Mφs, but becomes adherent on the apoptotic cell. Homophilic engagement of CD31 on the Mφ can activate CD31 to bind, via SH2 domains, SHP-1 (SH2 domain-containing phosphatase), a phosphatase which counteracts the activation of cell signalling pathways. # 2003 The Biochemical Society and the Medical Research Society 34 J. S. Duffield Blockade of Mφ receptors known to be involved in recognition and binding of newly exposed epitopes on the surface of apoptotic cells, namely the scavenger receptor CD36, the integrin αvβ and the phosphatidyl$ serine (PS) receptor (PSR), has been reported to inhibit the immunosuppressive effects of apoptotic cells on Mφs [69,70,73]. The ligands for CD36 and αvβ remain $ unknown, but PSR recognizes the membrane phospholipid PS, which is exposed on the outer leaflet of the plasma membrane of apoptotic cells. Binding of agonistic antibodies to αvβ or CD36 on the Mφ cell surface $ concurrently with the activating stimulus LPS has been reported to block the activation normally induced by LPS [69,70], thus mimicking the effect of apoptotic cells binding to the Mφ. Furthermore, intracellular signalling from CD47, also known as integrin-associated protein (‘ IAP ’), can block certain pro-inflammatory signalling pathways. This cofactor is closely associated with αvβ $ and is necessary for its activation [74,75] (Figure 4). These observations open up the possibility that, by manipulating receptors on the Mφ that mediate apoptotic cell engulfment, it will be possible to manipulate the ongoing function or activity of a population of inflammatory Mφs. The population could be accelerated into the transition from killing matrix-degrading cells into reparative healing cells. Brown et al. [76] recently found that the cell-surface receptor CD31, located on both the Mφ and the apoptotic cell, bound tightly through homophilic interaction, thus mediating recognition. The same interaction between the same receptors on live cells was repulsive, that is promoted disengagement. Such a finding offers novel strategies for manipulating the function of the Mφ, as CD31 has properties of an inhibitory receptor : when ligated it becomes tyrosine phosphorylated at specific sites in the intracellular tail and this enables the binding of phosphatases, such as Src homology 2 (SH2) domain-containing phosphatase (‘ SHP-1 ’). This phosphatase can block many activating signalling pathways, thereby inhibiting activation of the Mφ. Ongoing research points to CD31 and other inhibitory receptor family members as being responsible for mediating the immunosuppressive effects of apoptotic cells on the Mφ (Figure 4). THE RIGHT AMOUNT OF MR HYDE AND DR JEKYLL It can be seen that there is evidence of both types of Mφ in inflammation from the descriptions of Mr Hyde, the classically activated Mφ, and Dr Jekyll, the alternatively activated Mφ, in different inflammatory diseases. For example, in the resolving Thy1.1 nephritis, characterized by initial mesangiolysis, there are activated Mφs [77] and the induction of apoptosis [23]. There is also evidence of Mφ-mediated matrix accumulation following the initial # 2003 The Biochemical Society and the Medical Research Society mesangiolytic injury [52]. This pattern of cell death, matrix destruction followed by healing, is re-iterated elsewhere [24,25,50]. It suggests that the population of Mφs changes with time from one predominantly classically activated to one predominated by alternatively activated Mφs. Our studies [36,37] indicate that this transition from one Mφ population to another is brought about by the actions of the classically activated Mφ. Thus the very action of killing cells leads to phagocytosis, which leads to a transition from classically to alternatively activated. In progressive inflammation, there is much more apoptosis of stromal and parenchymal cells than seen in resolving inflammation ([32,29], and G. L. Thomas and J. S. Duffield, unpublished work). It implies that there may be too many Mφs or too many activated Mφs at the inflamed site. On the other hand, too many of the alternatively activated Mφs also appears to be detrimental. For example, when TGF-β, a cytokine released by alternatively activated Mφs, is overexpressed in the lung, it promotes progressive uncontrolled fibrosis with a paucity of inflammatory cells. Glucocorticoids, often used as a treatment for severe inflammation, promote alternatively activated Mφs. When effective, these agents leave scarred organs. Moreover, alternatively activated Mφs are associated with protracted chronic inflammation that is seen in conditions such as psoriasis and rheumatoid arthritis [78,79]. Together, these observations beg the question : are there defects in the co-ordinated progression from a Mφ population predominated by classically activated Mφs to one predominated by alternatively activated Mφs that result in failed resolution of inflammation ? Evidence already exists that this is the case. Complement factor 1q (C1q), a soluble protein from the complement cascade and more recently recognized as a member of the collectin family of proteins, is able to bind apoptotic cells and promote phagocytic clearance. Inflammatory Mφs recognize C1q on apoptotic cells, which assists in mediating phagocytosis. The C1q knockout mouse exhibits failure to clear apoptotic cells in tissues in a C1q-dependent fashion. The animal develops excessive inflammation, glomerulonephritis and features of the autoimmune disease systemic lupus erythematosis (‘ SLE ’). Excessive apoptotic cells are seen in the organs. Furthermore, when nephritis is deliberately induced, the disease is much more dramatic in these animals [80–82], implying overactivation of inflammatory Mφs. CONCLUSION Inflammatory Mφs are a heterogeneous group of cells with diverse functions. Broadly, and simplistically, they fall into two groups : the ‘ angry ’ cell-killing matrixdestroying Mφ, analogous to Mr Hyde, and the matrix- The inflammatory macrophage 35 Figure 5 Proposed negative feedback for inflammatory Mφ populations entering the inflamed site Monocytes entering the inflamed tissue become classically activated and induce apoptotic death of other cells in the tissue. These newly apoptotic cells are phagocytosed by the killer Mφs and this process brings about a switch in Mφ function to a cell sharing many features with the alternatively activated Mφ. generating cell-nurturing tissue-regenerating Mφ, analogous to Dr Jekyll. As Mr Hyde turns into Dr Jekyll, there is now evidence that the classically activated Mφ can lose its cell-killing pro-inflammatory capacity and take on features of the alternatively activated antiinflammatory Mφ (Figure 5). It is clear that (i) both cell types are necessary in the inflamed tissue for normal damage limitation and healing, (ii) both cell-types are probably active concurrently, (iii) an excess of either is detrimental to resolution of inflammation, and (iv) co-ordinated changes in one overriding population to another are necessary for remodelling and resolution of the inflamed tissue. Central to this conversion is the killing and ingestion of apoptotic cells by classically activated Mφ. This process signals to the Mφ population to change from a killing to a reparative population. Logically, defects in the process of cell killing, apoptotic cell clearance or intracellular pathways signalling to the Mφ upon engulfment of apoptotic cells will perturb this normal process, leading to excessive or inadequate activation of Mφs with the consequence of excessive or persistent chronic inflammation. Evidence already exists for such a defect in studies of the C1q knockout mouse. With the discovery of a role for members of the inhibitory receptor family in clearance of apoptotic cells by Mφ, current research is directed at defining the precise mechanisms by which apoptotic cell engulfment leads to the reprogramming of Mφ function. Understanding and manipulating defects in clearance and inhibitory signalling, such as C1q deficiency, may well lead to new therapeutic interventions in the future. ACKNOWLEDGMENTS This work was supported by a Clinical Training fellowship from the Medical Research Council. Current studies are supported by a grant from the National Kidney Research Fund (NKRF). Professor John Savill and Dr Simon Brown are thanked for their useful comments. REFERENCES 1 Savill, J. S., Wyllie, A. H., Henson, J. E., Walport, M. J., Henson, P. M. and Haslett, C. (1989) Macrophage phagocytosis of aging neutrophils in inflammation. Programmed cell death in the neutrophil leads to its recognition by macrophages. J. Clin. Invest. 83, 865–867 Albelda, S. M., Smith, C. W. and Ward, P. A. (1994) Adhesion molecules and inflammatory injury. FASEB J. 8, 504–512 Tedder, T. F., Steeber, D. A., Chen, A. and Engel, P. (1995) The selectins : vascular adhesion molecules. FASEB J. 9, 866–873 Aliberti, J. C., Machado, F. S., Souto, J. T. et al. (1999) βChemokines enhance parasite uptake and promote nitric oxide-dependent microbiostatic activity in murine inflammatory macrophages infected with Trypanosoma cruzi. Infect. Immun. 67, 4819–4826 Jun, C. D., Yoon, H. J., Kim, H. M. and Chung, H. T. (1995) Fibronectin activates murine peritoneal macrophages for tumor cell destruction in the presence of IFN-γ. Biochem. Biophys. Res. Commun. 206, 969–974 Albina, J. E., Henry, Jr, W. L., Mastrofrancesco, B., Martin, B. A. and Reichner, J. S. (1995) Macrophage activation by culture in an anoxic environment. J. Immunol. 155, 4391–4396 Cook, D. N., Beck, M. A., Coffman, T. M. et al. (1995) Requirement of MIP-1 α for an inflammatory response to viral infection. Science (Washington, D.C.) 269, 1583–1585 2 3 4 5 6 7 # 2003 The Biochemical Society and the Medical Research Society 36 J. S. Duffield 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 Erwig, L. P., Kluth, D. C., Walsh, G. M. and Rees, A. J. (1998) Initial cytokine exposure determines function of macrophages and renders them unresponsive to other cytokines. J. Immunol. 161, 1983–1988 Gerber, J. S. and Mosser, D. M. (2001) Reversing lipopolysaccharide toxicity by ligating the macrophage Fc γ receptors. J. Immunol. 166, 6861–6868 Lake, F. R., Noble, P. W., Henson, P. M. and Riches, D. W. (1994) Functional switching of macrophage responses to tumor necrosis factor α (TNF α) by interferons. Implications for the pleiotropic activities of TNF α. J. Clin. Invest. 93, 1661–1669 Goerdt, S., Politz, O., Schledzewski, K. et al. (1999) Alternative versus classical activation of macrophages. Pathobiology 67, 222–226 Stein, M., Keshav, S., Harris, N. and Gordon, S. (1992) Interleukin 4 potently enhances murine macrophage mannose receptor activity, a marker of alternative immunologic macrophage activation. J. Exp. Med. 176, 287–292 Tipping, P. G., Leong, T. W. and Holdsworth, S. R. (1991) Tumor necrosis factor production by glomerular macrophages in anti-glomerular basement membrane glomerulonephritis in rabbits. Lab. Invest. 65, 272–279 Schumann, J., Wolf, D., Pahl, A. et al. (2000) Importance of Kupffer cells for T-cell-dependent liver injury in mice. Am. J. Pathol. 157, 1671–1683 Huitinga, I., Ruuls, S. R., Jung, S., Van Rooijen, N., Hartung, H. P. and Dijkstra, C. D. (1995) Macrophages in T cell line-mediated, demyelinating, and chronic relapsing experimental autoimmune encephalomyelitis in Lewis rats. Clin. Exp. Immunol. 100, 344–351 Bottino, R., Fernandez, L. A., Ricordi, C. et al. (1998) Transplantation of allogeneic islets of Langerhans in the rat liver : effects of macrophage depletion on graft survival and microenvironment activation. Diabetes 47, 316–323 Ushio, Y., Yamamoto, N., Sanchez-Bueno, A. and Yoshida, R. (1996) Failure to reject an allografted tumor after elimination of macrophages in mice. Microbiol. Immunol. 40, 489–498 Jun, H. S., Yoon, C. S., Zbytnuik, L., van Rooijen, N. and Yoon, J. W. (1999) The role of macrophages in T cell-mediated autoimmune diabetes in nonobese diabetic mice. J. Exp. Med. 189, 347–358 Kerr, J. F. R., Wyllie, A. H. and Currie, A. R. (1972) Apoptosis : a basic biological phenomenon with wideranging implications in tissue kinetics. Br. J. Cancer 26, 239–257 Savill, J. (1997) Apoptosis in resolution of inflammation. J. Leukocyte Biol. 61, 375–380 Ballard, K. J. and Holt, S. J. (1968) Cytological and cytochemical studies on cell death and digestion in the fetal rat foot : the role of macrophages and hydrolytic enzymes. J. Cell. Sci. 3, 245–257 Harrison, D. J. (1988) Cell death in the diseased glomerulus. Histopathology 12, 679–683 Baker, A. J., Mooney, A., Hughes, J., Lombardi, D., Johnson, R. J. and Savill, J. (1994) Mesangial cell apoptosis : the major mechanism for resolution of glomerular hypercellularity in experimental mesangial proliferative nephritis. J. Clin. Invest. 94, 2105–2116 Desmouliere, A., Redard, M., Darby, I. and Gabbiani, G. (1995) Apoptosis mediates the decrease in cellularity during the transition between granulation tissue and scar. Am. J. Pathol. 146, 56–66 Iredale, J. P., Benyon, R. C., Pickering, J. et al. (1998) Mechanisms of spontaneous resolution of rat liver fibrosis. Hepatic stellate cell apoptosis and reduced hepatic expression of metalloproteinase inhibitors. J. Clin. Invest. 102, 538–549 Shimizu, A., Kitamura, H., Masuda, Y., Ishizaki, M., Sugisaki, Y. and Yamanaka, N. (1995) Apoptosis in the repair process of experimental proliferative glomerulonephritis. Kidney Int. 47, 114–121 Shimizu, A., Masuda, Y., Kitamura, H. et al. (1996) Apoptosis in progressive crescentic glomerulonephritis. Lab. Invest. 74, 941–951 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 Sugiyama, H., Kashihara, N., Makino, H., Yamasaki, Y. and Ota, A. (1996) Apoptosis in glomerular sclerosis. Kidney Int. 49, 103–111 Sugiyama, H., Kashihara, N., Onbe, T. et al. (1997) Bcl-2 expression and apoptosis in nephrotoxic nephritis. Exp. Nephrol. 5, 481–489 Kim, Y. H., Kim, S., Kim, K. A. et al. (1999) Apoptosis of pancreatic β-cells detected in accelerated diabetes of NOD mice : no role of Fas-Fas ligand interaction in autoimmune diabetes. Eur. J. Immunol. 29, 455–465 Liu, D., Pavlovic, D., Chen, M. C., Flodstrom, M., Sandler, S. and Eizirik, D. L. (2000) Cytokines induce apoptosis in β-cells isolated from mice lacking the inducible isoform of nitric oxide synthase (iNOSk\k). Diabetes 49, 1116–1122 Shimizu, A., Nakao, N., Muroga, K. et al. (1994) Apoptosis in progression process of experimental crescentic glomerulonephritis. J. Am. Soc. Nephrol. 5, 769 Yang, B., Johnson, T. S., Thomas, G. L., Watson, P. F., Wagner, B. and Nahas, A. M. (2001) Apoptosis and caspase-3 in experimental anti-glomerular basement membrane nephritis. J. Am. Soc. Nephrol. 12, 485–495 Tesch, G. H., Schwarting, A., Kinoshita, K., Lan, H. Y., Rollins, B. J. and Kelley, V. R. (1999) Monocyte chemoattractant protein-1 promotes macrophagemediated tubular injury, but not glomerular injury, in nephrotoxic serum nephritis. J. Clin. Invest. 103, 73–80 Meszaros, A. J., Reichner, J. S. and Albina, J. E. (2000) Macrophage-induced neutrophil apoptosis. J. Immunol. 165, 435–441 Duffield, J. S., Erwig, L. P., Wei, X., Liew, F. Y., Rees, A. J. and Savill, J. S. (2000) Activated macrophages direct apoptosis and suppress mitosis of mesangial cells. J. Immunol. 164, 2110–2119 Duffield, J. S., Ware, C. F., Ryffel, B. and Savill, J. (2001) Suppression by apoptotic cells defines tumor necrosis factor-mediated induction of glomerular mesangial cell apoptosis by activated macrophages. Am. J. Pathol. 159, 1397–1404 Kipari, T. and Hughes, J. (2002) Macrophage-mediated renal cell death. Kidney Int. 61, 760–761 Duffield, J. S. and Savill, J. (2001) Macrophages ingesting opsonised zymosan induce glomerular mesangial cell apoptosis by hydrogen peroxide and TNFα generation. J. Am. Soc. Nephrol. 12, 589A Boyle, J. J., Bowyer, D. E., Weissberg, P. L. and Bennett, M. R. (2001) Human blood-derived macrophages induce apoptosis in human plaque-derived vascular smooth muscle cells by Fas-ligand\Fas interactions. Arterioscler. Thromb. Vasc. Biol. 21, 1402–1407 Brown, S. B. and Savill, J. (1999) Phagocytosis triggers macrophage release of Fas ligand and induces apoptosis of bystander leukocytes. J. Immunol. 162, 480–485 Gibbs, D. F., Warner, R. L., Weiss, S. J., Johnson, K. J. and Varani, J. (1999) Characterization of matrix metalloproteinases produced by rat alveolar macrophages. Am. J. Respir. Cell. Mol. Biol. 20, 1136–1144 Song, E., Ouyang, N., Horbelt, M., Antus, B., Wang, M. and Exton, M. S. (2000) Influence of alternatively and classically activated macrophages on fibrogenic activities of human fibroblasts. Cell. Immunol. 204, 19–28 Quinn, C. O., Scott, D. K., Brinckerhoff, C. E., Matrisian, L. M., Jeffrey, J. J. and Partridge, N. C. (1990) Rat collagenase. Cloning, amino acid sequence comparison, and parathyroid hormone regulation in osteoblastic cells. J. Biol. Chem. 265, 22342–22347 Mariani, T. J., Sandefur, S., Roby, J. D. and Pierce, R. A. (1998) Collagenase-3 induction in rat lung fibroblasts requires the combined effects of tumor necrosis factor-α and 12-lipoxygenase metabolites : a model of macrophage-induced, fibroblast-driven extracellular matrix remodeling during inflammatory lung injury. Mol. Biol. Cell 9, 1411–1424 Kitamura, M. (1998) TGF-β1 as an endogenous defender against macrophage-triggered stromelysin gene expression in the glomerulus. J. Immunol. 160, 5163–5168 # 2003 The Biochemical Society and the Medical Research Society The inflammatory macrophage 37 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Airola, K., Karonen, T., Vaalamo, M. et al. (1999) Expression of collagenases-1 and -3 and their inhibitors TIMP-1 and -3 correlates with the level of invasion in malignant melanomas. Br. J. Cancer 80, 733–743 Sharma, A. K., Mauer, S. M., Kim, Y. and Michael, A. F. (1995) Altered expression of matrix metalloproteinase-2, TIMP, and TIMP-2 in obstructive nephropathy. J. Lab. Clin. Med. 125, 754–761 Leibovich, S. J. and Ross, R. (1975) The role of the macrophage in wound repair. A study with hydrocortisone and antimacrophage serum. Am. J. Pathol. 78, 71–100 Lange-Sperandio, B., Cachat, F., Thornhill, B. A. and Chevalier, R. L. (2002) Selectins mediate macrophage infiltration in obstructive nephropathy in newborn mice. Kidney Int. 61, 516–524 De Heer, E., Prodjosudjadi, W., Davidoff, A., van der Wal, A., Bruijn, J. A. and Paul, L. C. (1998) Control of monocyte influx in glomerulonephritis in transplanted kidneys in the rat. Lab. Invest. 78, 1327–1337 Westerhuis, R., van Straaten, S. C., van Dixhoorn, M. G. et al. (2000) Distinctive roles of neutrophils and monocytes in anti-thy-1 nephritis. Am. J. Pathol. 156, 303–310 D’Souza, M. J., Oettinger, C. W., Shah, A., Tipping, P. G., Huang, X. R. and Milton, G. V. (1999) Macrophage depletion by albumin microencapsulated clodronate : attenuation of cytokine release in macrophage-dependent glomerulonephritis. Drug Dev. Ind. Pharm. 25, 591–596 Pierce, G. F., Mustoe, T. A., Lingelbach, J. et al. (1989) Platelet-derived growth factor and transforming growth factor-β enhance tissue repair activities by unique mechanisms. J. Cell Biol. 109, 429–440 Gratchev, A., Guillot, P., Hakiy, N. et al. (2001) Alternatively activated macrophages differentially express fibronectin and its splice variants and the extracellular matrix protein βIG-H3. Scand. J. Immunol. 53, 386–392 Haroon, Z. A., Lai, T. S., Hettasch, J. M., Lindberg, R. A., Dewhirst, M. W. and Greenberg, C. S. (1999) Tissue transglutaminase is expressed as a host response to tumor invasion and inhibits tumor growth. Lab. Invest. 79, 1679–1686 Murry, C. E., Giachelli, C. M., Schwartz, S. M. and Vracko, R. (1994) Macrophages express osteopontin during repair of myocardial necrosis. Am. J. Pathol. 145, 1450–1462 Ashcroft, G. S., Yang, X., Glick, A. B. et al. (1999) Mice lacking Smad3 show accelerated wound healing and an impaired local inflammatory response. Nat. Cell Biol. 1, 260–266 Koch, R. M., Roche, N. S., Parks, W. T., Ashcroft, G. S., Letterio, J. J. and Roberts, A. B. (2000) Incisional wound healing in transforming growth factor-β1 null mice. Wound Repair Regen. 8, 179–191 Diamond, J. R., Ding, G., Frye, J. and Diamond, I. P. (1992) Glomerular macrophages and the mesangial proliferative response in the experimental nephrotic syndrome. Am. J. Pathol. 141, 887–894 Cao, B., Guo, Z., Zhu, Y. and Xu, W. (2000) The potential role of PDGF, IGF-1, TGF-β expression in idiopathic pulmonary fibrosis. Chin. Med. J. (Beijing, Engl. Ed.) 113, 776–782 Ross, R., Masuda, J., Raines, E. W. et al. (1990) Localization of PDGF-B protein in macrophages in all phases of atherogenesis. Science (Washington, D.C.) 248, 1009–1012 Xiong, M., Elson, G., Legarda, D. and Leibovich, S. J. (1998) Production of vascular endothelial growth factor by murine macrophages : regulation by hypoxia, lactate, and the inducible nitric oxide synthase pathway. Am. J. Pathol. 153, 587–598 Leibovich, S. J., Polverini, P. J., Shepard, H. M., Wiseman, D. M., Shively, V. and Nuseir, N. (1987) Macrophage-induced angiogenesis is mediated by tumour necrosis factor-α. Nature (London) 329, 630–632 Sunderkotter, C., Goebeler, M., Schulze-Osthoff, K., Bhardwaj, R. and Sorg, C. (1991) Macrophage-derived angiogenesis factors. Pharmacol. Ther. 51, 195–216 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 Ostendorf, T., Kunter, U., Eitner, F. et al. (1999) VEGF mediates glomerular endothelial repair. J. Clin. Invest."'& 913–923 104, Meagher, L. C., Savill, J. S., Baker, A. and Haslett, C. (1992) Phagocytosis of apoptotic neutrophils does not induce macrophage release of thromboxane B2. J. Leukocyte Biol. 52, 269–273 Fadok, V. A., Bratton, D. L., Konowal, A., Freed, P. W., Westcott, J. Y. and Henson, P. M. (1998) Macrophages that have ingested apoptotic cells in vitro inhibit proinflammatory cytokine production through autocrine\paracrine mechanisms involving TGF-β, PGE2, and PAF. J. Clin. Invest. 101, 890–898 Freire-de-Lima, C. G., Nascimento, D. O., Soares, M. B. et al. (2000) Uptake of apoptotic cells drives the growth of a pathogenic trypanosome in macrophages. Nature (London) 403, 199–203 Voll, R. E., Herrmann, M., Roth, E. A., Stach, C., Kalden, J. R. and Girkontaite, I. (1997) Immunosuppressive effects of apoptotic cells. Nature (London) 390, 350–351 Magnus, T., Chan, A., Grauer, O., Toyka, K. V. and Gold, R. (2001) Microglial phagocytosis of apoptotic inflammatory T cells leads to down-regulation of microglial immune activation. J. Immunol. 167, 5004–5010 Van Lent, P. L., Licht, R., Dijkman, H., Holthuysen, A. E., Berden, J. H. and van den Berg, W. B. (2001) Uptake of apoptotic leukocytes by synovial lining macrophages inhibits immune complex-mediated arthritis. J. Leukocyte Biol. 70, 708–714 Huynh, M. L., Fadok, V. A. and Henson, P. M. (2002) Phosphatidylserine-dependent ingestion of apoptotic cells promotes TGF-β1 secretion and the resolution of inflammation. J. Clin. Invest. 109, 41–50 Brown, E., Hooper, L., Ho, T. and Gresham, H. (1990) Integrin-associated protein : a 50-kD plasma membrane antigen physically and functionally associated with integrins. J. Cell Biol. 111, 2785–2794 Lindberg, F. P., Bullard, D. C., Caver, T. E., Gresham, H. D., Beaudet, A. L. and Brown, E. J. (1996) Decreased resistance to bacterial infection and granulocyte defects in IAP-deficient mice. Science (Washington, D.C.) 274, 795–798 Brown, S., Heinisch, I., Ross, E., Shaw, K., Buckley, C. D. and Savill, J. (2002) Apoptosis disables CD31mediated cell detachment from phagocytes promoting binding and engulfment. Nature (London) 418, 200–203 Johnson, R. J., Lombardi, D., Eng, E. et al. (1995) Modulation of experimental mesangial proliferative nephritis by interferon-γ. Kidney Int. 47, 62–69 Djemadji-Oudjiel, N., Goerdt, S., Kodelja, V., Schmuth, M. and Orfanos, C. E. (1996) Immunohistochemical identification of type II alternatively activated dendritic macrophages (RM 3\1+++, MS-1p, 25F9−) in psoriatic dermis. Arch. Dermatol. Res. 288, 757–764 Szekanecz, Z., Haines, G. K., Lin, T. R. et al. (1994) Differential distribution of intercellular adhesion molecules (ICAM-1, ICAM-2, and ICAM-3) and the MS-1 antigen in normal and diseased human synovia. Their possible pathogenetic and clinical significance in rheumatoid arthritis. Arthritis Rheum. 37, 221–231 Botto, M., Dell’Agnola, C., Bygrave, A. E. et al. (1998) Homozygous C1q deficiency causes glomerulonephritis associated with multiple apoptotic bodies. Nat. Genet. 19, 56–59 Robson, M. G., Cook, H. T., Botto, M. et al. (2001) Accelerated nephrotoxic nephritis is exacerbated in C1q-deficient mice. J. Immunol. 166, 6820–6828 Taylor, P. R., Carugati, A., Fadok, V. A. et al. (2000) A hierarchical role for classical pathway complement proteins in the clearance of apoptotic cells in vivo. J. Exp. Med. 192, 359–366 Polunovsky, V. A., Chen, B., Henke, C. et al. (1993) Role of mesenchymal cell death in lung remodeling after injury. J. Clin. Invest. 92, 388–397 Meszaros, A. J., Reichner, J. S. and Albina, J. E. (1999) Macrophage phagocytosis of wound neutrophils. J. Leukocyte Biol. 65, 35–42 # 2003 The Biochemical Society and the Medical Research Society 38 J. S. Duffield 85 Suk, K., Kim, S., Kim, Y. H. et al. (2001) IFN-γ\TNF-α synergism as the final effector in autoimmune diabetes, a key role for STAT1\IFN regulatory factor-1 pathway in pancreatic β cell death. J. Immunol. 166, 4481–4489 86 Thomas, H. E. and Kay, T. W. (2000) Beta cell destruction in the development of autoimmune diabetes in the non-obese diabetic (NOD) mouse. Diabetes Metab. Res. Rev. 16, 251–261 87 Pender, M. P. and Rist, M. J. (2001) Apoptosis of inflammatory cells in immune control of the nervous system, role of glia. Glia 36, 137–144 88 Saio, M., Radoja, S., Marino, M. and Frey, A. B. (2001) Tumor-infiltrating macrophages induce apoptosis in activated CD8+ T cells by a mechanism requiring cell contact and mediated by both the cell-associated form of TNF and nitric oxide. J. Immunol. 167, 5583–5593 89 Diamond, J. R., Ding, G., Frye, J. and Diamond, I. P. (1992) Glomerular macrophages and the mesangial proliferative response in the experimental nephrotic syndrome. Am. J. Pathol. 141, 887–894 Sharma, S., Sankaran, R. T., Shan, Z., Gibbons, N. and Singhal, P. C. (1996) Escherichia coli–macrophage interactions modulate mesangial cell proliferation and matrix synthesis. Nephron. 73, 587–596 91 Rai, R. M., Lee, F. Y., Rosen, A. et al. (1998) Impaired liver regeneration in inducible nitric oxide synthase deficient mice. Proc. Natl. Acad. Sci. U.S.A. 95, 13829–13834 92 Kodelja, V., Muller, C., Politz, O., Hakij, N., Orfanos, C. E. and Goerdt, S. (1998) Alternative macrophage activation-associated CC-chemokine-1, a novel structural homologue of macrophage inflammatory protein-1α with a Th2-associated expression pattern. J. Immunol. 160, 1411–1418 93 Hunt, T. K., Knighton, D. R., Thakral, K. K., Goodson, III, W. H. and Andrews, W. S. (1984) Studies on inflammation and wound healing, angiogenesis and collagen synthesis stimulated in vivo by resident and activated wound macrophages. Surgery 96, 48–54 90 # 2003 The Biochemical Society and the Medical Research Society