Cells of the synovium in rheumatoid arthritis. Macrophages

The multitude and abundance of macrophage-derived mediators in rheumatoid arthritis and their paracrine/autocrine effects identify macrophages as local and systemic amplifiers of disease. Although uncovering the etiology of rheumatoid arthritis remains the ultimate means to silence the pathogenetic process, efforts in understanding how activated macrophages influence disease have led to optimization strategies to selectively target macrophages by agents tailored to specific features of macrophage activation. This approach has two advantages: (a) striking the cell population that mediates/amplifies most of the irreversible tissue destruction and (b) sparing other cells that have no (or only marginal) effects on joint damage.


Introduction
Macrophages (Mφ) are of central importance in rheumatoid arthritis (RA) due to their prominent numbers in the inflamed synovial membrane and at the cartilage-pannus junction, their clear activation status [1,2] (see Table 1 for overview), and their response to successful anti-rheumatic treatment [3]. Although Mφ probably do not occupy a causal pathogenetic position in RA (except for their potential antigen-presenting capacity), they possess broad pro-inflammatory, destructive, and remodelling potential and contribute considerably to inflammation and joint destruction in acute and chronic RA. Also, activation of this lineage extends to circulating monocytes and other cells of the mononuclear phagocyte system (MPS), including bone marrow precursors of the myelomonocytic lineage and osteoclasts [2,4,5].
Thus, before a causal factor for RA is known, monocytes/Mφ remain an attractive research focus for the following reasons: (a) the radiological progression of joint destruction correlates with the degree of synovial Mφ infiltration [1], (b) the therapeutic efficacy of conventional anti-rheumatic therapy coincides with downregulation of MPS functions [6], (c) therapies directed at cytokines made predominantly by Mφ are effective in RA [7], (d) conventional or experimental drugs can be selectively targeted to Mφ or their different subcellular compartments (for example, [2,8]), (e) differential activation of intracellular signal transduction pathways underlies different Mφ effector functions [9], and (f) more specific inhibitors of key metabolic enzymes or particular signal transduction pathways may become available as selective targets of antirheumatic therapy [9,10]. In addition, the amplifying role of Mφ in RA has emerged so clearly that the effects of antirheumatic therapy (whether specific or conventional) on monocytes/Mφ may become an objective readout of the effectiveness of treatment [11][12][13] (Stuhlmuller B, Hernandez MM, Haeupl T, Kuban RJ, Gruetzkau A, Voss JW, Salfeld J, Kinne RW, Burmester GR, unpublished data).
Cells of the MPS show clear signs of activation, not only in synovial and juxta-articular compartments such as the synovial membrane or the cartilage-pannus and bone-pannus junctions (including the subchondral bone), but also in extraarticular compartments (for example, peripheral blood and subendothelial space, the latter of which is the site of foam cell formation and development of atherosclerotic plaques in RA) ( Table 2). This activation underlines the systemic inflammatory character of RA and may contribute to the occurrence of cardiovascular events and its increased mortality (reviewed in [2,14,15]).

Biological functions of monocytes/ macrophages and their role in rheumatoid arthritis
The monocyte/Mφ system represents an integral part of the natural immune system and participates in the first-line response against infectious agents. Another crucial contribution to the body's homeostasis is the scavenging function of any debris generated by physiological or pathological processes. Thus, monocytes/Mφ possess multiple and powerful biological functions that may greatly affect onset and development of chronic inflammatory diseases like RA (see overview in Table 3) (reviewed in [16]).

Stimulation/regulation of monocyte/ macrophage activation in rheumatoid arthritis
The role of monocytes/Mφ in RA is conceivably the integrated result of stimulatory, effector, dually active, and autoregulatory mediators/mechanisms. At the tissue level, the scenario is characterized by the influx of pre-activated monocytes, their maturation into resident Mφ, their full activation, and their interaction with other synovial cells. The complexity of the interaction is the result of paracrine activation mechanisms generated via sheer cell-cell contact as well as of numerous autocrine mechanisms -nearly any soluble mediator shows abnormalities. A simplified scheme of this integrated system and the currently known mediators is provided in Figure 2. For ease of presentation, the parts are organized as incoming stimuli (both paracrine and soluble) (column a) and effector molecules (column b), although autocrine loops are also relevant (as discussed below).

Cell-cell interaction
A significant part of Mφ effector responses is mediated by cell contact-dependent signalling with different inflammatory or mesenchymal cells (as exemplified in the lower left quadrant of Figure 2).

Fibroblast-macrophage interaction
Because of the prominent numbers of Mφ and fibroblasts and their activated status in RA synovial tissue, the interaction of these cells is critical for the resulting inflammation and tissue damage. Indeed, the mere contact of these cells elicits the production of interleukin (IL)-6, granulocyte macrophage colony-stimulating factor (GM-CSF), and IL-8. The cytokine output can be enhanced or down-modulated not only by addition of pro-inflammatory or regulatory cytokines (for example, IL-4, IL-10, IL-13, or IL-1 receptor antagonist [IL-1RA]), Available online http://arthritis-research.com/content/9/6/224

Figure 1
Physiological/pathological differentiation of the mononuclear phagocyte system in rheumatoid arthritis (RA). (a) Physiological differentiation of the mononuclear phagocyte system (MPS) (steady-state cytokine and growth factor milieu). In the human MPS, monocytes (M) differentiate from a CD34 + stem cell via an intermediate step of monoblasts. Monocytes leave the bone marrow and remain in circulation for approximately 3 days. Upon entering various tissues, they differentiate into different types of resident macrophages (Mφ), including synovial macrophages. It is believed that these mature cells do not recirculate, surviving for several months in their respective tissues until they senesce and die. Some circulating monocytes retain the potential for differentiating into dendritic cells and osteoclasts (asterisk in the insert). The steady-state myeloid differentiation involves many factors, including granulocyte macrophage colony-stimulating factor (GM-CSF), interleukin (IL)-1, IL-6, and tumor necrosis factoralpha (TNF-α), which are produced by resident bone marrow macrophages (reviewed in [2]). (b) Increased plasticity of myeloid differentiation and its possible role in RA (augmented cytokine and growth factor milieu). Human bone marrow intermediate cells can differentiate into macrophages or dendritic cells in the presence of c-kit ligand, GM-CSF, and TNF-α. TNF-α, in turn, inhibits the differentiation of monocytes into macrophages in vitro and, together with GM-CSF, directs the differentiation of precursor cells into dendritic cells, another important arm of the accessory cell system. Also, either IL-11 or vitamin D 3 and dexamethasone induce the differentiation of bone marrow cells or mature macrophages into osteoclasts, cells involved in the destruction of subchondral bone in RA. Osteoclasts and dendritic cells can also be derived from circulating monocytes upon stimulation with macrophage colony-stimulating factor (M-CSF) or IL-4 plus GM-CSF. This plasticity, and its dependence on growth factors or cytokines that are clearly elevated in peripheral blood and bone marrow of patients with RA, may explain some differentiation anomalies in the disease and also the efficacy of some anti-rheumatic drugs. Non-specific enhancement of monocyte maturation and tissue egression, in turn, are consistent with the known alterations in inflammation (reviewed in [2]). The differentiation paths potentially relevant to RA are but also by neutralization of the CD14 molecule [17]. Also, in vitro, significant cartilage degradation occurs in co-cultures of mouse fibroblasts and Mφ, a response markedly exceeding that observed with each culture alone (reviewed in [2]). Furthermore, purified human synovial fibroblasts co-cultured with myelomonocytic cells induce cartilage degradation in vitro, but with a strong contribution of soluble IL-1 and tumor necrosis factor (TNF)-α [18].
T cell-macrophage interaction Accessory, inflammatory, effector, and inhibitory Mφ functions can be stimulated by fixed T cells or their plasma membranes if T cells are pre-activated and express activation surface molecules. In response to such interaction, monocytes produce metalloprotease (MMP), IL-1α, and IL-1β [19,20]. Also, T cells pre-stimulated in an antigen-mimicking fashion stimulate TNF-α and IL-10 production once in contact with monocytes [20]. Conversely, fixed T cells stimulated in an antigen-independent fashion (that is, with IL-15, IL-2, or a combination of IL-6 and TNF-α, the so-called Tck cells) induce monocyte production of TNF-α but not the antiinflammatory IL-10 [20,21]. These findings suggest that early RA may reflect antigen-specific T cell-Mφ interactions [22]. Conversely, chronic RA may be associated with antigenindependent interactions dominated by an exuberant cytokine milieu and Tck cells. This may also explain the relative paucity of IL-10 in the synovial membrane in chronic RA, as discussed below.
Several ligand pairs on T cells and monocytes/Mφ have been implicated in this interaction [20], although the importance of individual ligand pairs, as well as the influence of soluble mediators, remains unclear. Interestingly, T cells isolated from RA synovial tissue show phenotypical and functional features similar to Tck cells and the above-mentioned signal transduction pathways differentially contribute to the induction of TNF-α and IL-10 production in monocytes/Mφ by co-culture with Tck cells. If applicable in vivo in RA, this would allow selective therapeutic targeting of pro-inflammatory TNF-α and sparing of anti-inflammatory IL-10.

Interaction of macrophages with endothelial cells and natural killer cells
The interaction between monocytes and endothelial cells in RA ( Figure 2), critical for the sustained influx of activated monocytes in the synovial membrane, relies on the altered expression of integrin/selectin pairs on the surface of the two cell types (reviewed in [2]). Because the synovial cytokine milieu (including the Mφ-derived TNF-α) upregulates the expression of these ligand pairs, a self-perpetuating cycle ensues by which sustained Mφ-derived mechanisms lead to further influx and activation of circulating monocytes. Upon cell contact, monokine-activated CD56 bright natural killer cells induce monocytes to the production of TNF-α, thus representing another possible reciprocal loop of activation in RA [23].

Soluble stimuli
Cytokine stimuli with pro-inflammatory effects on macrophages Numerous cytokines with known or potential stimulatory activity on monocytes/Mφ have been identified, as schematically shown in the upper left quadrant of Figure 2. A systematic list of these stimuli and their known or potential functions is provided in Table 4. Some of these mediators are produced by monocytes/Mφ themselves and therefore activate Mφ in an autocrine fashion, as also exemplified in  Table 4. T-cell cytokines acting on Mφ (for example, IL-17) have been comprehensively reviewed elsewhere [24,25].

Bacterial/viral components and Toll-like receptors
The ability of bacterial toxins or superantigens to initiate the secretion of Mφ-derived cytokines is relevant in view of a possible microorganism etiology of RA and in view of side effects of anti-TNF-α therapy, particularly mycobacterial infections [26,27]. Lipopolysaccharide (LPS), for example, binds to Mφ through the CD14/LPS-binding protein receptor complex and, in vitro, stimulates the production of IL-1β, TNF-α, and macrophage inflammatory protein-1α. Staphylococcal enterotoxin B (SEB), a potent Mφ activator, enhances arthritis in MRL-lpr/lpr mice. Anti-TNF-α therapy, in this case, reverses both the severe wasting effects of SEB and the incidence of arthritis, indicating that TNF-α is central in this system. Finally, the staphylococcal enterotoxin A increases the expression of the Toll-like receptor (TLR)-4 in human Table 3 Monocyte

Function Mechanisms (Potential) role in rheumatoid arthritis
Clearance of Binding of immunoglobulins to Fc receptors Potential clearance of rheumatoid factor but further activation of immune complexes (Fc-γ-R I, IIA, IIB, and IIIA) monocytes/Mφ Opsonization of complexes by complement, leading to binding to Mφ complement receptors and further cell activation [101,102] (reviewed in [2,103]) Notably, inhibition of monocyte activation by Fc-γ-R IIB [102] Complement Binding of complement factors to complement Recognition of activated complement (soluble phase or on activation receptors 1 (CD35), 3 (CD11b), and 5a (CD88) immunoglobulin G-immune complexes) Promotion of phagocytosis and activation of monocytes/Mφ [103] Phagocytosis of Conventional (Fc-mediated) → lysosomal Scavenging of debris but potential import of arthritogenic molecules particulate antigens degradation and MHC-II antigen processing [103] Antigen presentation and activation of CD4 + and CD8 + T cells, possibly relevant to disease initiation or perpetuation (spreading of autoimmunity) (reviewed in [2]) Coiling phagocytosis → lysosomal degradation Involved in phagocytosis of Borrelia burgdorferi, active agent of Lyme and MHC-I antigen processing arthritis (reviewed in [2]) Clearance of Removal of pathogens and recognition of Induction of Mφ-derived cytokines by bacterial toxins or superantigens intracellular apoptotic cells via exposed intracellular [26,28,103] pathogens and membrane components Modulation of Mφ responses by mycobacterial lipoarabinomannan apoptotic cells [104,105] or Toll-like receptors [29,106] Persistence of obligate/facultative intracellular pathogens with arthritogenic potential [107,108] Antigen processing Enzymatic degradation of antigens and binding Important cognate functions upon antigen recognition via presentation and presentation of antigenic peptides to MHC molecules and of antigen on MHC-II molecules [109] and expression of membrane transport to the cell surface second signal molecules adjacent to T cells (reviewed in [2])

Chemotaxis and Attraction of other inflammatory cells and
Positive feedback between Mφ-derived cytokines and chemotactic angiogenesis induction of neo-vascularization factors (for example, IL-8 and monocyte chemoattractant protein-1) Promotion of angiogenesis by IL-8 and soluble forms of adhesion molecules (for example, vascular cell adhesion molecule-1 and endothelial-leukocyte adhesion molecule-1) [69] Wound healing Remodelling of tissue via interaction with Sustained monocyte recruitment at wound injury sites via monocyte fibroblasts chemoattractant macrophage inflammatory protein-1α Phagocytosis of matrix debris and endogenous production of IL-1, TNF-α, and so on as well as post-injury tissue remodelling (reviewed in [2]) Lipid metabolism Mφ synthesis of prostaglandins (PGs) E 2 and I 2 Pro-inflammatory activity of PGE 2 and PGI 2 and leukotrienes in Expression of scavenger receptor A (uptake of rheumatoid arthritis, but also autocrine negative feedback through oxidized low-density lipoprotein) peroxisome proliferator-activated receptors α and γ (reviewed in [2]) Fish-based diets are associated with clinical improvement of human and experimental arthritis (reviewed in [2]) Modulation of T cell-contact-induced production of IL-1β and TNF-α in monocytes by ligation of major histocompatibility complex-II, with subsequent enhancement of pro-inflammatory cytokines by known TLR-4 ligands (for example, LPS [28]).
TLRs are part of the recently discovered cellular patternrecognition receptors (PPRs) involved in first-line defense of the innate immune system against microbial infections. In addition to bacterial or viral components, some PPRs recognize host-derived molecules, such as the glycoprotein gp96, nucleic acids, hyaluronic acid oligosaccharides, heparan sulfate, fibronectin fragments, and surfactant protein A (reviewed in [29]). In RA, notably, functional TLR-2 and TLR-4 are expressed on CD16 + synovial Mφ, peripheral blood mononuclear cells, and synovial fibroblasts [30]. Also, their expression can be upregulated by cytokines present in the inflamed RA joint (for example, IL-1β, TNF-α, macrophage colony-stimulating factor, and IL-10); this suggests that activation of synovial cells via TLRs may contribute to disease processes [29], as supported by findings in experimental arthritis [31]. On the other hand, the chronic polyarthritis observed in mice with deletion of the DNase II gene, whose Mφ are incapable of degrading mammalian DNA, appears to occur independently of the nucleic acid-specific TLR-9 [32].   Table 4 Overview

Hormones
Females are affected by RA at a ratio of approximately 3:1 compared with males and experience clinical fluctuations during the menstrual cycle and pregnancy, indicating a major modulating role for sex hormones. Due to their expression of sex-hormone receptors and their cytokine response upon exposure to estrogens, monocytes/Mφ are strongly involved in hormone modulation of RA [33]. Indeed, physiological levels of estrogens stimulate RA Mφ to the production of the pro-inflammatory cytokine IL-1, whereas higher levels inhibit IL-1 production, conceivably mimicking the clinical improvement during pregnancy. Interestingly, selective estrogen receptor ligands inhibiting nuclear factor (NF)-κB transcriptional activity (but lacking estrogenic activity) can markedly inhibit joint swelling and destruction in experimental arthritis [34].

Cytokine stimuli with regulatory effects on macrophages
In addition to pro-inflammatory cytokines, several cytokines that regulate monocyte/Mφ function in RA have been described (summarized in the upper left quadrant of Figure 2). A systematic list of these cytokines is provided in Table 5. Interestingly, some of these molecules are produced by Mφ themselves (most notably, IL-10), so that autocrine regulation may also play a prominent role during the different clinical phases of RA. Other regulatory cytokines derive from other cell types present in the inflamed synovial membrane: T cells (for example, IL-4 and IL-13) or stromal cells (for example, IL-11). For these molecules, the reader is referred to recent publications or comprehensive reviews [25,35,36].

Monocyte/macrophage effector molecules with proinflammatory effects in rheumatoid arthritis
Mφ produce a number of pro-inflammatory cytokines, as schematically shown in the upper right quadrant of Figure 2. A systematic list of the pro-inflammatory ILs is provided in Table 4.
Tumor necrosis factor-alpha TNF-α is a pleiotropic cytokine that increases the expression of cytokines, adhesion molecules, prostaglandin E 2 , collagenase, and collagen by synovial cells. TNF-α exists in membrane-bound and soluble forms, both acting as proinflammatory mediators. Transmembrane TNF-α is involved in local, cell contact-mediated processes and appears to be the prime stimulator of the R75 receptor [37]. Interestingly, the transgenic expression of this form is alone sufficient to induce chronic arthritis [38]; likewise, a mutant membrane TNF-α, which uses both R55 and R75 receptors, can cause arthritis. Conversely, the soluble form of TNF-α, shed via MMP cleavage from the membrane-bound form, primarily stimulates the R55 receptor, acting transiently and at a distance [37].
In RA, TNF-α is mostly produced by Mφ in the synovial membrane and at the cartilage-pannus junction and possibly

Interleukin-1
In the RA synovial membrane, IL-1 is found predominantly in CD14 + Mφ [51]; also, IL-1 levels in the synovial fluid significantly correlate with joint inflammation [52]. The two existing forms of IL-1 (IL-1α and IL-1β) show some differ-ences (for example, low protein homology, stronger proinflammatory regulation of the IL-1β promoter, and secretion of inactive pro-IL-1β versus expression of membrane-bound IL-1α activity) but also strong similarities (that is, threedimensional structures of the essential domains, molecular masses of pro-peptides, and mature-form processing proteases), resulting in almost identical binding capacity to the IL-1 receptors and comparable function. In arthritis, IL-1 appears to mediate a large part of the articular damage, as it profoundly influences proteoglycan synthesis and degradation [43,53]. At the same time, IL-1 induces the production of MMP-1 and MMP-3 and enhances bone resorption; this is compatible with recent evidence from arthritis models and human RA suggesting that the tissue-destruction capacities of IL-1β may outweigh its genuine role in joint inflammation [53].

Interleukin-1 receptors
The IL-1 type I receptor (IL-1R1), which mediates cell activation via IL-1R accessory protein and IL-1 receptor-associated kinase (IRAK), is found on numerous cells in the synovial tissue of patients with RA [54]. In contrast, the type II receptor (IL-1R2) (also found in soluble form in serum), which lacks cell-activating properties and acts exclusively as a decoy receptor, is low in synovial tissue [55]. Similarly, IL-1RA, a soluble protein that blocks the action of IL-1 by binding to the type I receptor without receptor activation, has been detected only sporadically in RA synovial samples. In RA, the balance between IL-1 and its physiological inhibitor IL-1RA is therefore shifted in favor of IL-1, indicating a dysregulation crucial in promoting chronicity [53]. However, therapeutic application of IL-1RA (anakinra) appears to be only modestly effective in RA (reviewed in [56]). Therefore, it remains to be clarified whether the IL-1 pathway is a less suitable therapeutic target than TNFα (for example, due to functional redundancy in the IL-1 receptor superfamily) or whether the biological molecule IL-1RA is suboptimal for therapy.

Interleukin-15
IL-15, a cytokine of the IL-2 family with chemoattractant properties for memory T cells, is produced by lining layer cells Table 5 Overview  [135,136], IL-16 also has proinflammatory properties (that is, correlates with metalloprotease-3 levels, progression of joint destruction, and levels of other pro-inflammatory cytokines) [137,138].

of anti-inflammatory cytokines relevant to macrophage (dys)function in rheumatoid arthritis
IFN-β X --Clear anti-inflammatory and anti-destructive effects in experimental arthritides Therapy attempts in human RA thus far have been unsuccessful [149].

IFN-β, interferon-beta
Notably, peripheral or synovial T cells stimulated with IL-15 induce Mφ to produce IL-1β, TNF-α, IL-8, and monocyte chemotactic protein-1 [21,57] but not the regulatory IL-10. Because IL-15 is also produced by Mφ themselves, this cytokine may (re)stimulate T cells, possibly self-perpetuating a pro-inflammatory loop [57]. The expression of IL-15 in the RA synovial membrane, its biological function, and its successful targeting in experimental arthritis have generated large expectations on the use of a fully humanized anti-IL-15 antibody in clinical trials [56][57][58].

Interleukin-18
In the RA synovial membrane, this cytokine of the IL-1 family is expressed in CD68 + Mφ contained in lymphoid aggregates. CD14 + Mφ of the RA synovial fluid also express the IL-18 receptor [59]. The pro-inflammatory role of IL-18 in arthritis (and its potential suitability as a therapeutic target in RA) is indicated by the following findings: (a) IL-18 treatment markedly aggravates experimental arthritis [59], (b) intraarticular overexpression of IL-18 induces experimental arthritis, (c) IL-18 is involved in the development of experimental streptococcal arthritis (a strongly Mφ-dependent model), (d) IL-18 is selectively overexpressed in the bone marrow of patients with juvenile idiopathic arthritis and Mφ activation syndrome [5], (e) IL-18 can stimulate osteoclast formation through upregulation of RANKL (receptor activator of NF-κB ligand) production by T cells in RA synovitis, and (f) IL-18 mediates its action via classic induction of TNF-α, GM-CSF, and interferon (IFN)-γ [59] or functional Toll-like receptors TLR-2 and TLR-4 in synovial cells [30] or else through the induction of synovial acute-phase serum amyloid proteins.
The clinical relevance of synovial IL-18 is emphasised by its correlation with the systemic levels of C-reactive protein (CRP); also, IL-18 and CRP decrease in parallel in synovial tissue and serum following effective treatment with diseasemodifying anti-rheumatic drugs [60]. In addition, peripheral blood mononuclear cells of RA patients show low levels of the IL-18 binding protein (a natural inhibitor of IL-18) and reduced sensitivity to stimulation with IL-12/IL-18, indicating profound dysregulation of the IL-18 system [61].

Interleukin-23
The genuine role of IL-23, a cytokine of the IL-12 family predominantly produced by Mφ or dendritic cells, is unclear due to the sharing of the p40 subunit with IL-12 [62]. IL-23 has prominent pro-inflammatory functions, since transgenic expression in mice leads to multi-organ inflammation and premature death. IL-23 promotes various T-cell responses potentially relevant to RA [62]. Recent studies in experimental arthritis have demonstrated that mice lacking only IL-12 (p35 -/-) show exacerbated arthritis, whereas mice lacking only IL-23 (p19 -/-) are completely protected from arthritis [63]. In addition, activation of Mφ derived from arthritissusceptible rats is paradoxically associated with reduced levels of pro-inflammatory mediators but high expression of IL-23 (p19), whereas non-susceptible rats show the inverse phenotype. If these findings were transferable to human RA, IL-23 would have a pro-inflammatory role and IL-12 a protective one. At the present time, it is unclear whether these findings fit into the recently introduced M1/M2 paradigm of differential Mφ activation [64,65] and especially whether this paradigm can be exploited for a better understanding of the role of Mφ in RA.

Interleukin-27
IL-27, another cytokine of the IL-12 family, is expressed by monocytes/Mφ following common inflammatory stimuli and displays a variety of pro-and anti-inflammatory properties [66]. In support of a pro-inflammatory role in arthritis, neutralizing antibodies against IL-27p28 suppress experimental arthritis [67].

Chemokines and chemokine receptors
Chemokines (subdivided into the CXC, CC, C, and CX3C families) are small proteins specialized in differential recruitment of leukocyte populations via a number of transmembrane receptors. Chemokines not only favor monocyte influx into inflamed tissue, but also play a key role in activation, functional polarization, and homing of patrolling monocytes/Mφ [65]. Notably, monocytes/Mφ express only select types of the numerous chemokine receptors (for example, CCR1, 2, 5, 7, and 8 as well as CX3CR1), representing a partially specific basis for prominent trafficking of monocyte/Mφ in arthritis. In RA, synovial Mφ produce several chemokines (for example, CCL3 [or Mφ inflammatory protein 1α], CCL5 [or RANTES], and CX3CL1 [or fractalkine]) and at the same time carry chemokine receptors, indicating the presence of autocrine loops in disease (reviewed in [68]). At the same time, chemokines are upregulated by the Mφderived TNF-α and IL-1. Significantly, some chemokines expressed in synovial Mφ (for example, IL-8 and fractalkine) are powerful promoters of angiogenesis, thus providing a link between Mφ activation and the prominent neo-vascularization of the RA synovium [69]. In RA, angiogenesis may be further promoted via activation of Mφ by advanced glycation end products, whereas thrombospondin-2 seems to downregulate angiogenesis. Because the enlargement of the vascular bed potentiates the influx of activated monocytes, down-modulation of the chemokine system represents a multi-potential target of anti-rheumatic therapy, as indicated by the promising results of treatment with a CCR1 antagonist in RA [68].

Macrophage migration inhibitory factor
One of the first ILs ever discovered, migration inhibitory factor (MIF), is an early-response cytokine abundantly released by Mφ. MIF stimulates a number of Mφ functions in an autocrine fashion (for example, secretion of TNF-α, phagocytosis, and generation of reactive oxygen species [ROS]). In addition, MIF confers resistance to apoptosis in Mφ and synovial fibroblasts, thus prolonging the survival of activated, disease-relevant cells. In RA, MIF is overexpressed in serum and synovial tissue in correlation with disease activity. Also, polymorphisms in the promoter or coding region of the human MIF gene are associated with features of juvenile idiopathic arthritis or adult RA [70].

Monocyte/macrophage effector molecules with antiinflammatory/regulatory effects in rheumatoid arthritis
Mφ also produce anti-inflammatory cytokines, most notably IL-RA and IL-10, both cytokines engaged in autocrine regulatory loops (shown in the lower right quadrant of Figure 2) ( Table 5).

Interleukin-1 receptor antagonist
Differentiated Mφ constitutively express IL-1RA, which is upregulated by pro-inflammatory mediators, including IL-1 itself or GM-CSF, and induces strong anti-inflammatory effects. By means of this feedback mechanism, Mφ therefore contribute to the termination of inflammatory reactions (reviewed in [71,72]) (see above).
Interleukin-10 IL-10, a Th 2 -and Mφ-derived cytokine with clear autocrine functions, reduces HLA-DR expression and antigen presentation in monocytes and inhibits the production of proinflammatory cytokines, GM-CSF, and Fc-γ receptors by synovial Mφ. Consistently with cytokine and chemokine downregulation, IL-10 clearly suppresses experimental arthritis. In spite of IL-10 elevation in serum and synovial compartments of patients with RA [73], some studies suggest a relative deficiency of IL-10 [74]. A combined IL-4/IL-10 deficiency probably tilts the cytokine balance to a proinflammatory predominance. In addition, the ex vivo production of IL-10 by RA peripheral blood mononuclear cells is negatively correlated with radiographic joint damage and progression of joint damage, suggesting that high IL-10 production is protective in RA. Similarly to IL-4, however, treatment with recombinant IL-10 does not improve RA. This may be partially explained by upregulation of the proinflammatory Fc-γ receptors I and IIA on monocytes/Mφ (reviewed in [2]).

Monocyte/macrophage effector molecules with dual effects in rheumatoid arthritis
Cytokines with a dual role are indicated in Tables 4 and 5.
Interleukin-6 IL-6 is the most strikingly elevated cytokine in RA, especially in the synovial fluid during acute disease [75]. The acute rise is consistent with the role of IL-6 in acute-phase responses (Table 1). However, while IL-6 levels in the synovial fluid correlate with the degree of radiological joint damage, and IL-6 and soluble IL-6 receptors promote the generation of osteoclasts, this cytokine has phase-dependent effects; for example, it protects cartilage in acute disease but promotes excessive bone formation in chronic disease. While IL-6 is mostly produced by synovial fibroblasts and only partially by Mφ, two findings suggest that the striking IL-6 rise is a prominent outcome of Mφ activation: (a) the morphological vicinity of IL-6-expressing fibroblasts with CD14 + Mφ in the RA synovial tissue (reviewed in [2]) and (b) co-culture studies showing that IL-1 stimulates IL-6 production [17]. The role of IL-6 in experimental arthritis and the anti-arthritic effects of anti-IL-6 receptor antibodies suggest a role for anti-IL-6 therapy in RA [76] (reviewed in [77]).
Transforming growth factor-beta In RA, Mφ express different transforming growth factor-beta (TGF-β) molecules and TGF-β receptors in the lining and sublining layers, at the cartilage-pannus junction, and in the synovial fluid [78][79][80]. The pro-inflammatory effects of TGF-β are substantiated by induction of Mφ expression of Fc-γ receptor III (which elicits the release of tissue-damaging ROS) and promotion of monocyte adhesion and infiltration during chronic disease (reviewed in [80]). At the same time, TGF-β has anti-inflammatory properties; for example, it counteracts some IL-1 effects, including phagocytosis of collagen and possibly MMP production. A protective role of TGF-β in RA is also suggested by the association between TGF-β polymorphism and disease severity; that is, alleles associated with low TGF-β expression are correlated with stronger inflammation and poorer outcome [81]. Likewise, experimental arthritis is significantly ameliorated by activation of TGF-β via adenoviral expression of thrombospondin-1 [82]. The effects of TGF-β on tissue inhibitor of MMP (TIMP) are also unclear, as the regulation of MMP and TIMP may depend on different tissue domains (superficial versus deep cartilage layers) and may vary for intra-or extracellular digestion of collagen (reviewed in [2]).

Treatment of human rheumatoid arthritis with conventional anti-macrophage approaches
The role of Mφ-derived cytokines in the perpetuation of RA, the pathophysiological dichotomy between joint inflammation and cartilage destruction, and the crucial significance of activated synovial Mφ in relation to permanent joint damage [1] have led to a radical re-evaluation of the conventional antiinflammatory and disease-modifying treatments in relation to Mφ parameters in order to potentiate therapeutic effects (for example, via combination approaches [83]) and reduce side effects. For anti-Mφ effects of conventional anti-rheumatic therapy in RA (including methotrexate, leflunomide, antimalarials, gold compounds, corticosteroids, and non-steroidal anti-inflammatory drugs), the reader is referred to a recent comprehensive review [11]. Recent findings show that conventional and specific anti-rheumatic treatments predominantly target sublining rather than lining Mφ; also, different therapeutic approaches seem to result in similar histological changes in the inflamed synovial membrane, including significant reduction of sublining Mφ. This, in turn, is significantly correlated with the degree of clinical improvement [11,12]. Thus, different pathogenetic mechanisms may funnel into similar disease pathway(s), leading to massive activation of Mφ and providing the rationale for targeted anti-Mφ therapy.

Non-conventional and experimental antimacrophage therapy
Counteraction of monocyte/macrophage activation at a cellular level Apoptosis-inducing agents Physical elimination of disease-relevant cells (for example, activated Mφ or osteoclasts) by apoptosis is advantageous because it circumvents secondary tissue damage by restraining cellular organelles in apoptotic vesicles. Phagocytic incorporation of liposome-encapsulated non-amino-bisphosphonates by activated monocytes, for example, induces apoptosis in these cells [84] (Figure 3). Systemic application of encapsulated bisphosphonates in experimental arthritis not only counteracts joint swelling, but also prevents local joint destruction and subchondral bone damage [85]; in addition, it shows protective effects on remote bone damage. Studies in RA show that a single intra-articular administration of clodronate liposomes leads to Mφ depletion and decreased expression of adhesion molecules in the lining layer of RA synovial tissue [86]. Selective targeting of activated Mφ has also been demonstrated using either apoptosis-inducing immunotoxins coupled to anti-Fc-γ receptor I (CD64) antibodies or folate receptor-mediated targeting (reviewed in [2]). In general, liposome encapsulation can also be exploited for selective delivery of Mφ-modulating drugs [87] (reviewed in [2]) or gene therapy constructs (reviewed in [88]).

Control of gene transcription
The transcription of most cytokine genes in monocytes/Mφ depends on the activation of NF-κB and NF-κM transcription factors or that of the activator protein-1 (AP-1) complex. In RA synovial Mφ, the expression of NF-κB is more pronounced than that of AP-1, a selectivity that may bear important therapeutic implications [89]. Accordingly, the anti-arthritic effects of IL-4 may be based on the selective suppression of NF-κB in Mφ. IL-10 also downregulates the production of proinflammatory monokines, inhibiting the nuclear factors NF-κB,

Gene therapy in experimental arthritis
Gene therapy has been applied in experimental arthritis models to counteract Mφ-derived IL-1 and TNF-α or to deliver/overexpress protective IL-1RA, soluble IL-1 type I receptor-IgG fusion protein, and type I soluble TNF-α receptor-IgG fusion protein. This has been extended to (Mφderived) anti-inflammatory cytokines (that is, IL-4, IL-10, IL-13, IFN-β, or TGF-β) and to 'molecular synovectomy' (either by expression of herpes simplex virus-thymidine kinase with subsequent administration of ganciclovir or by overexpression of Fas-ligand/inhibitors of nuclear translocation of NF-κB, resulting in synovial cell apoptosis [88,91,92]). Therefore, gene therapy aimed at neutralizing pro-inflammatory Mφ products, overexpressing Mφ-regulating mediators, or simply eliminating overly activated Mφ remains promising for the treatment of arthritis.

Conclusion
The multitude and abundance of Mφ-derived mediators in RA and their paracrine and autocrine effects (including those directed to other cells of the myeloid lineage) indicate that Mφ are local and systemic amplifiers of disease severity and perpetuation. The main local mechanisms include (a) selfperpetuating chemokine-mediated recruitment of inflammatory cells, (b) cytokine-mediated activation of newly immigrated inflammatory cells, (c) cell contact-mediated activation of neighboring inflammatory cells, (d) cytokine-and cell contactmediated secretion of matrix-degrading enzymes, (e) activation of mature dendritic cells and cytokine-mediated differentiation of Mφ (and possibly B cells, T cells, and mesenchymal cells) into antigen-presenting cells, with possible effects on spreading of autoimmunity to cryptic epitopes, (f) neovascularization, with potentiation of cellular and exudatory mechanisms, and (g) (trans)differentiation of Mφ into osteoclasts involved in subchondral bone damage. At a systemic level, amplification of disease can proceed at least through the following mechanisms: (a) acute-phase response network, (b) systemic production of TNF-α, (c) anomalies in bone marrow differentiation, and (d) chronic activation of circulating monocytes.
Although uncovering the etiology of disease remains the ultimate goal of research, the efforts in understanding how activated Mφ influence disease have led to optimization strategies to selectively target activated Mφ in RA (Figure 3). This approach has at least two advantages: (a) striking the very cell population that mediates/amplifies most of the irreversible cartilage destruction and (b) minimizing adverse effects on other cells that may have no (or marginal) effects on joint damage.