CX₃CR1 Reduces Kidney Fibrosis by Inhibiting Local Proliferation of Profibrotic Macrophages

The renal mononuclear phagocyte system consists of an extensive network of mononuclear phagocytes (i.e., dendritic cells [DCs] and macrophages) (1, 2). Although there is no clear demarcation between these cell types, DCs are typically specialized at regulating other immune effector cells, especially T cells, whereas macrophages preferentially act as immune effectors in innate immune responses or contribute to tissue repair (3–6). In the healthy kidney, most resident immune cells display phenotypic and functional characteristics of DCs and can be distinguished from macrophages by expression of the CD11c molecule (7). Furthermore, CD11c⁺ kidney cells possess a transcriptome that resembles that of DCs but not of macrophages (8, 9). Although kidney DCs have been shown to perform sentinel and regulatory functions in inflammation and infection (10), macrophages contributed to host protection from infection and maintained tissue homeostasis (1, 11, 12). In contrast, inappropriate macrophage functions promoted fibrosis in various organs (13, 14). In the kidney, macrophages but not DCs mediated fibrosis in unilateral ureter ligation (UUO) and after ischemia/reperfusion (15–18). Several mechanisms contribute to renal fibrosis, including the production of the main profibrotic molecule TGF-β (14, 19). In the kidney, TGF-β promotes fibrosis by activating fibroblasts and myofibroblasts, modulating the expression of tissue inhibitors of matrix metalloproteases and by directly promoting synthesis of collagens (20). TGF-β secretion has been linked to monocytes and macrophages (14), suggesting that macrophages might mediate kidney fibrosis through this mediator.

Tissue macrophages partially originate from embryonic progenitors of the yolk sac or the fetal liver (21, 22), and from infiltrating bone marrow–dependent progenitors such as monocytes, whereas DCs can develop from monocytes or pre-DCs (23–25). The exact contribution of the different precursors to the renal mononuclear cell system is unclear. The chemokine receptor CX₃CR1, also known as fractalkine receptor, is ubiquitously expressed on most tissue macrophages and DCs but does not play a major role for their ontogeny, homeostatic migration, or colonization of tissues with resident phagocytes (26, 27), except kidney DCs (3) and intestinal macrophages (28–30). There is evidence that the survival of monocytes and tissue macrophages depends on CX₃CR1 (31-33), and this required interaction with cell surface–bound CX₃CL1 (34). CX₃CL1⁺ cells have been shown to be present within glomeruli and in the tubular compartment (35), but their exact identity is unclear.

Under inflammatory conditions, CCR2-dependent monocytes expressing the markers Ly6C and Gr1 enter inflamed tissues (36–38) and rapidly give rise to DCs and macrophages (31, 39–41). Inflammatory Gr1⁺ macrophages possess profibrotic properties (42), and their depletion or genetic CCR2 deficiency attenuated tissue damage and fibrosis in UUO (16), lupus nephritis (43), and ischemia/reperfusion (44). Also, CX₃CR1 has been implicated in monocyte recruitment into inflamed tissues, like arteriosclerotic vessels (45), the listeria-infected spleen (46), the eye (47), in colitis models (28), or the skin (48). Furthermore, CX₃CR1 may exert profibrotic functions (49–55).

DCs and macrophages of the kidney abundantly express CX₃CR1 (56). CX₃CR1 deficiency reduced entry of their progenitors into the kidney and attenuated symptoms and fibrosis in ischemia/reperfusion injury (44, 49), hypertensive kidney damage (54), and crescentic glomerulonephritis (3, 57). In contrast, the loss of renal phagocytes aggravated renal candidiasis (58) but not pyelonephritis (3). Also, studies on human biopsy material from fibrotic kidneys described upregulation of CX₃CR1, supporting a pathogenic role (59). However, in toxin- and in obstruction-induced liver fibrosis, CX₃CR1 was protective (50). The role of CX₃CR1 in the obstruction-induced standard kidney fibrosis model, UUO, has not been clarified yet. In this study, we have addressed this open question, assuming that the marked loss of renal DCs in CX₃CR1 deficiency would attenuate inflammation and fibrosis. However, we found that CX₃CR1 was protective in UUO because of a novel inhibitory effect on the other CX₃CR1-expressing mononuclear phagocyte type of the kidney, the macrophages.

Eight- to 14-wk-old Cd45.1, Cd11c^EYFP (60), Cx₃cr1^GFP/+, and Cx₃cr1^GFP/GFP mice (61), all on a C57BL/6 background, were bred and kept under specific-pathogen-free condition at the animal facilities of the University of Bonn and Monash Medical Centre. Institutional and Government Review Boards have approved all animal studies.

Adult male mice were anesthetized by inhalation of isofluorane. UUO was performed through a midline incision. The left ureter was tied with silk suture at two points and permanently ligated. Sham-operated mice underwent an identical midline incision but without ureteric ligation. At 3, 5, and 7 d postobstruction, mice were sacrificed, and kidneys were collected.

Intravital multiphoton microscopy of the intact kidney was performed on mice 3 d post-UUO, anesthetized by i.p. injection of 150 mg/kg ketamine hydrochloride and 10 mg/kg xylazine, as described previously (18, 62). The jugular vein was cannulated, and the left kidney was exteriorized through a lateral incision, immersed in bicarbonate buffered saline at 37°C, and covered with a coverslip. Mice were kept at 37°C. Time lapse recordings of GFP⁺ cells in the kidney of Cx₃cr1^GFP/+ and Cx₃cr1^GFP/GFP mice were captured using a Leica SP5 mulitphoton microscope system (Leica Microsystems, Mannheim, Germany) with a ×20 1.0 NA water immersion objective (18, 63). Fluorophores were excited by a SpectraPhysics MaiTai pulsed infrared laser tuned to 860 nm, with emissions captured by non-descanned detectors with 525/50 nm (GFP), 585/40 nm (tetramethylrhodamine isothiocyanate), and 650/50 nm (Alexa 633) filters. Celltracker Orange CMRA (C34551; Molecular Probes) was used as a counterstain to provide visual context. Images were collected with ∼120 μm Z-depth in 6-μm steps, with one Z-stack every 30s for 20–30 min and analyzed using Imaris (version 7.0; Bitplane AG). To determine GFP⁺ cell motility, individual cells were tracked in vivo and a motile cell defined as one that had moved at least the diameter of the cell body. To determine cell activity, a morphology index was used as previously described (18), relating to the sphericity of cells described by the rate of change in morphology index (ΔMI). Cells with relatively low ΔMI values change morphology less than cells with higher ΔMI values.

Methyl Carnoyl’s fixed paraffin-embedded murine kidneys were sectioned at 1 μm and stained with periodic acid–Schiff’s (PAS) for evaluation of hydronephrosis and overall injury. For the evaluation of fibrosis, Sirius Red stained sections as well as α-smooth muscle actin (αSMA) immunohistochemistry were evaluated using computer-based morphometric analyses by a pathologist in blinded fashion as described previously (64). In short, for both stainings, the percentage of positively stained area was calculated from 20 cortical fields per section, representing nearly the whole cortex. Indirect immunoperoxidase procedure was performed as described previously (64). HRP coupled to human αSMA (clone 1A4; DakoCytomation, Glostrup, Denmark), which cross-react with murine αSMA was used. Fluorescence microscopy was performed on 5-μm paraffin sections from 4% PFA-fixed kidneys. GFP-expressing cells in Cx₃cr1^GFP/+ mice were identified by nuclei staining and quantified using the IX71 microscope and the CellR software (Olympus).

Kidneys were digested with collagenase and DNAse-I (both from Sigma-Aldrich) as described previously (7). Briefly, kidneys were digested for 45 min, and tubular fragments from digested kidneys were removed by filtration through a 100-μm nylon mesh.

After treatment with Fc-blocking Abs (2.4G2) cells were stained with fluorochrome-labeled mAbs against F4/80 (CI:A3-1), CD45 (30-F11), CD11c (N418), KI67 (SolA15; isotype control: eBR2a), Gr1 (RB6-8C5), and Annexin V (number 550474; BD Biosciences). For intracellular cytokine staining, single-cell suspensions were incubated in RPMI 1640 medium (10% FCS, penicillin/streptomycin, and glutamine) with 1 μl/ml GolgiPlug and 1 μl/ml GolgiStop (BD Biosciences) for 4 h at 37°C. After cell surface staining, cells were fixed in Cytofix (BD Biosciences) and permeabilized with PermWash (BD Biosciences). Intracellular staining was performed for 30 min at room temperature using fluorochrome-labeled mAbs against TGFβ (TW7-16B4; isotype control: MOPC-21). For analysis of proliferation, BrdU was administered 3 and 4 d after ligation, and incorporation of BrdU was analyzed on day 5 after ligation by flow cytometry (number 557892; BD Biosciences). Cells were analyzed with a BD Canto II and a BD Fortessa (BD Biosciences), and data were analyzed using Flowjo (Tree Star). For concatenated data presentation, data from all mice within the same group was combined using Flowjo.

Single-cell suspensions from the bone marrows were prepared by flushing the bones with sterile and cold PBS. Bone marrow monocytes were enriched by excluding CD19⁺CD3⁺CD11c⁺ cells. Bone marrow cells (1–5 × 10⁶) were transferred into mice.

The protein levels of cytokines in the supernatant of mechanically homogenized kidneys were determined by Luminex assay. The growth factors M-CSF and GM-CSF were determined by using ELISA (R&D Systems). All other levels were determined by a flow cytometry–based bead assay (eBioscience).

Whole-tissue RNA from kidneys was extracted using NucleoSpin RNA II kit (Macherey-Nagel), according to the manufacturer’s instructions. RNA was reverse transcribed into cDNA using the High-Capacity Reverse Transcription Kit (Applied Biosystems). Quantitative PCR was performed for 40 cycles using SYBR Green (Applied Biosystems). QuantiTect Primer assays (Qiagen) were used to detect mouse gene expression of Stat1 (QT00162183), Stat6 (QT01042503), Arg1 (QT00134288), Hmox (QT00159915), Irf5 (QT00252623), Retnla (QT00254359), and Mrc1 (QT00103012). All samples were run in duplicates and normalized to Hprt primers (Invitrogen), forward primer (5′-GTCCCAGCGTCGTGATTAGCGAT-3′) and reverse primer (5′-GGGCCACAATGTGATGGCCTCC-3′).

Results are expressed as mean ± SEM. Comparisons were drawn using parametric T-Test and nonparametric Mann–Whitney test (Prism 5; Graphpad Software, San Diego, CA). Significant p values were expressed as follows: *p < 0.05; **p < 0.01; ***p < 0.001.

The chemokine receptor CX₃CR1 has been shown to promote renal inflammation in several models (3, 49, 54, 57–59, 65). Given that chronic kidney inflammation often leads to fibrosis, we hypothesized that CX₃CR1 deficiency should attenuate renal fibrosis. To test this hypothesis, we induced one of the standard kidney fibrosis models, UUO in mice expressing or lacking CX₃CR1. We used Cx₃cr1^GFP/GFP mice that homozygously express a GFP reporter under the control of the CX₃CR1 promoter and are consequently CX₃CR1 deficient. As controls, we used Cx₃cr1^GFP/+ mice, which express the GFP reporter heterozygously and CX₃CR1-competent wild-type mice (Cx₃cr1^+/+ mice). Surprisingly, kidney fibrosis in CX₃CR1-deficient mice was notably aggravated compared with control mice 7 d after UUO, as evidenced by higher Sirius Red collagen staining in the ligated kidneys of Cx₃cr1^GFP/GFP mice (Fig. 1A, 1B). Also, αSMA staining for profibrotic myofibroblasts was stronger in ligated kidneys of CX₃CR1-deficient mice (Fig. 1C, 1D). These findings revealed an unexpected antifibrotic role of CX₃CR1 in renal fibrosis.

To define the mechanisms underpinning the protective role of CX₃CR1, we focused on the cells that express CX₃CR1 in the kidney, namely mononuclear phagocytes, including DCs and macrophages. As reported before (3), CX₃CR1-deficient (Cx₃cr1^GFP/GFP) mice normally possess far less GFP⁺ renal phagocytes than CX₃CR1-competent (Cx₃cr1^GFP/+) mice (Fig. 2A, 2B). Also, on day 7, after ureter ligation, fewer phagocytes were seen, albeit only 10% less (Fig. 2A, 2B), suggesting the appearance of new phagocytes that partially compensated for their deficiency under homeostatic conditions. Of note, the message for the Th1 mediator Stat1 was below detection limit, and the message for the inflammatory cytokines IL-1α, IL-6, IFN-γ, and TNF and of the proresolving mediators, IL-10 and Hmox, did not differ between ligated kidneys of CX₃CR1-competent and -deficient mice (Fig. 2C). Thus, the aggravated fibrosis in CX₃CR1-deficient mice cannot be explained merely by stronger renal inflammation.

These findings led us to examine in more detail the renal GFP⁺ phagocytes in UUO using multiphoton microscopy, which allows analyzing their morphology and motility over time. In both healthy CX₃CR1-competent and CX₃CR1-deficient mice, GFP⁺ cells with dendritic protrusions were present within the interstitium (see Supplemental Videos 1, 2).

After UUO, GFP⁺ cells tended to form clusters around damaged tubules (see Supplemental Videos 3, 4). In addition to GFP⁺ cells with dendritic morphology, the kidneys of CX₃CR1-deficient mice with UUO contained larger cells (Fig. 3A) devoid of dendrites (Fig. 3B), which were more motile (Fig. 3C). Furthermore, the phagocytes in CX₃CR1-deficient mice contained a subset that underwent little shape change over the time of analysis (Fig. 3D). Although these data did not reach statistical significance, they prompted us to hypothesize that GFP⁺ phagocytes with monocyte/macrophage properties might be involved in the increased fibrosis in CX₃CR1-deficient mice with UUO.

We hypothesized that higher numbers of renal macrophages might explain the profibrotic effect of CX₃CR1 deficiency in UUO (Fig. 1) because it has been previously reported that renal macrophages but not DCs mediate fibrosis in this model (16, 18, 66). However, Cx₃cr1^GFP/GFP mice possessed similar or lower numbers of GFP⁺ renal phagocytes under homeostatic conditions, in experimental glomerulonephritis and in kidney infection compared with CX₃CR1-competent controls (3, 58), consistent with our present findings in UUO (Fig. 2A, 2B). We therefore discriminated between GFP⁺ kidney macrophages and DCs by costaining with F4/80 and CD11c. We found that CX₃CR1 deficiency reduced renal CD11c⁺ DC numbers in the ligated and contralateral kidney (Fig. 4A, upper gate in the dot plots) consistent with the morphological changes shown in Fig. 3. Next, we focused on CD11c⁻GFP⁺ cells (Fig. 4A, lower gate in the dot plots). We found that >90% of these cells expressed F4/80 (Fig. 4B) identifying these GFP/CX₃CR1-expressing cells as macrophages. Furthermore, these data also exclude considerable contaminations of other cells than monocytes/macrophages and DCs in the microscopy data (Figs. 2, 3). Because the microscopy findings above (Fig. 3) hinted at the appearance of monocytes/macrophages, we focused on the analysis of macrophages expressing the Gr1 marker, which designates mononuclear phagocytes recently recruited from the circulation (67, 68) and which is related to helper macrophages expressing Ly6C (4), a component of Gr1. Gr1⁺ macrophages were considerably more numerous already on day 3 after UUO (Fig. 4B, 4C). In the absence of CX₃CR1, these cells were more abundant than in CX₃CR1-competent mice, whereas the abundance of kidney-resident Gr1⁻ macrophages was unchanged (Fig. 4C). On day 7, fewer recruited macrophages were seen, whereas numbers of Gr1⁻ macrophages had further increased (Fig. 4C). Thus, the lack of CX₃CR1 reduced the number of CD11c⁺ DCs, but increased the numbers of CD11c^– Gr1⁺ macrophages in UUO. This shift from DCs to macrophages explains why the total number of GFP⁺ phagocytes was only somewhat reduced in CX₃CR1-deficient mice (Fig. 2), and the relative increase of macrophages might explain the increased kidney fibrosis in the mice.

Next, we examined whether CX₃CR1 also modulates the functionality of macrophages. To this end, we analyzed the production of TGF-β, the key profibrotic molecule in the kidney (65, 69). Most of the TGF-β–producing leukocytes indeed expressed F4/80 (Supplemental Fig. 1). Importantly, the frequencies (Supplemental Fig. 1) and numbers (Fig. 4D) of TGF-β–producing Gr1^– and Gr1⁺ macrophages were increased in ligated kidneys of CX₃CR1-deficient mice. M1/M2 differentiation markers such as Arg1, Mrc1, Irf5, Stat6, and Retnla1 were little altered in CX₃CR1 deficiency (Fig. 5), indicating that the functional changes of macrophages could not be explained by effects on M1/M2 differentiation (70, 71). These data demonstrate that CX₃CR1 not only negatively regulates the numbers of profibrotic macrophages but also their fibrogenic potential.

The inhibitory effects of CX₃CR1 on renal macrophages could theoretically be due to reduced recruitment of macrophage precursors such as monocytes, reduced intrarenal differentiation into macrophages, reduced proliferation, increased survival or be due to a combination of these factors. Therefore, we examined all of these potential mechanisms.

To establish a system to analyze macrophage precursor recruitment to the kidney, we first adoptively transferred bone marrow–derived monocytes expressing a YFP reporter for CD11c, which allows convenient discrimination between DCs and macrophages, into CD45.1-congenic mice with or without UUO. Only a few monocytes reached the kidneys of healthy mice and the contralateral kidney of mice with UUO (Fig. 6A), suggesting that renal phagocytes possess a low turnover rate. In contrast, ligated kidneys contained clearly detectable monocytes 2 d after transfer, and some of them were YFP⁺ (Fig. 6A), indicating differentiation into DCs.

To investigate whether CX₃CR1 is required for the entry of these precursors into the kidney, we transferred monocytes from Cx₃cr1^GFP/GFP mice or Cx₃cr1^GFP/+ mice into CX₃CR1-competent wild-type recipients 1 d after ureter ligation (Fig. 6B, 6C). After an additional 2 d, less CX₃CR1-deficient phagocytes had entered the ligated kidneys than did CX₃CR1-competent phagocytes (Fig. 6C). Notably, no difference was seen in the blood (Supplemental Fig. 2), excluding an effect of CX₃CR1 on the abundance of circulating macrophage precursors. Thus, CX₃CR1 promoted, rather than inhibited, phagocyte precursor recruitment. Furthermore, the levels of the chemokine MIP-1α that can attract macrophage precursors were decreased in ligated and also in contralateral kidneys of CX₃CR1-deficient mice, and there were trends toward reduced levels of RANTES, MCP-1, and MCP-3 (Fig. 6D). This is consistent with reduced macrophage precursor recruitment and may be explained by the reduction in chemokine-producing DCs in CX₃CR1-deficient mice (3).

To study whether CX₃CR1 regulated the differentiation of monocytes into macrophages versus DCs, we determined the ratio between donor-derived DCs and macrophages after entering the ligated kidney. This parameter was not significantly changed in CX₃CR1 deficiency (Fig. 7A). Furthermore, the intrarenal levels of cytokines that affect DC/macrophage differentiation, GM-CSF or M-CSF, were CX₃CR1 independent (Fig. 7B). Taken together, these findings argue against changes in macrophage recruitment or differentiation as an explanation for their increased intrarenal numbers in CX₃CR1-deficient mice.

We next investigated whether CX₃CR1 regulates apoptosis by staining with Annexin V. We found that more Gr1^– and Gr1⁺ macrophages were apoptotic within the ligated kidneys (Fig. 8A, 8B, black line versus gray line). However, lack of CX₃CR1 did not reduce but instead increased the frequency of Annexin V⁺Gr1^– and Gr1⁺ macrophages (Fig. 8A, 8B), excluding apoptosis as the mechanism for the increased number of macrophages in CX3Cr1-deficient animals.

It has been previously reported that tissue macrophage numbers can be regulated by their local proliferation (72, 73). To investigate whether CX₃CR1 decreases intrarenal macrophage numbers by inhibiting their proliferation, we determined expression of the proliferation marker Ki67 on day 3 and uptake of BrdU from day 3 until day 5 after ureteric ligation. We found that Ki67 expression was significantly increased in macrophages of ligated kidneys from CX₃CR1-deficient mice compared with CX₃CR1-competent mice (Fig. 9A), and this was seen in both Gr1^– and Gr1⁺ macrophages (Fig. 9B). Furthermore, more macrophages of CX₃CR1-deficient mice incorporated BrdU (Fig. 9C, 9D), and the uptake per macrophage was higher (Fig. 9E), verifying local CX₃CR1-dependent proliferation of macrophages. Notably, such increased proliferation was not detectable in DCs (Fig. 9D). Thus, CX₃CR1 inhibits local macrophage proliferation in UUO, and this can explain the higher numbers of profibrotic renal macrophages in the absence of CX₃CR1.

The chemokine receptor CX₃CR1 is widely expressed by DCs and macrophages (74), but its functional role is still not fully clear. In this paper, we report a previously undescribed antifibrotic effect of CX₃CR1: it reduced the inflammation-induced proliferation of TGF-β–producing renal macrophages, which are critically important in experimental kidney fibrosis (15–18). This finding was unexpected, given the proinflammatory and profibrotic properties of CX₃CR1 not only in diseases of the kidney (3, 49, 54, 57–59, 65), but also of the lung (75), liver (50), intestine (28), skin (55), arteriosclerotic vessels (76), and the CNS (33). CX₃CR1 has been shown to contribute to the inflammatory recruitment of monocytes from the circulation into all these organs (44). We found that CX₃CR1 had a similar effect in UUO so that fewer monocytes lacking this receptor entered the inflamed kidneys. The lower levels of monocyte-attracting chemokines and of inflammatory mediators that we detected in kidneys of CX₃CR1-deficient mice may be responsible for the reduced kidney recruitment of these cells. It is likely that these cells originated from monocytes, which have been shown to give rise to tissue macrophages under inflammatory conditions (5). The reduction of monocyte-attracting chemokines likely resulted from the paucity of DCs in kidneys of CX₃CR1-deficient animals because DCs are the principal producers of chemokines and inflammatory mediators in the kidney (77–79). However, these findings were hard to reconcile with the accumulation of macrophages in CX₃CR1-deficient mice, implying that other mechanisms must exist that overcompensated for the reduced attraction of macrophage precursors.

One such mechanism might be a negative effect of CX₃CR1 on the survival of tissue macrophages. However, previous studies in disease models of liver, CNS, gut, and arteriosclerotic vessels (74) demonstrated that CX₃CR1 signaling increased rather than reduced the life span of macrophages. In contrast to these findings, one recent study reported increased macrophage numbers and liver fibrosis in CX₃CR1-deficient mice after surgical bile duct ligation, a model that resembles UUO by being induced by mechanic tissue damage (50). The discrepancy between macrophage accumulation and decreased macrophage life span in that study was suggested to have resulted from the release of proinflammatory mediators from CX₃CR1-deficient dying macrophages that attracted further profibrotic macrophages. Although we did observe more dying macrophages after UUO in CX₃CR1-deficient mice, we found neither increased monocyte-attracting chemokines and inflammatory cytokines, nor did we find a stronger influx of inflammatory monocytes, arguing against inflammation resulting from premature macrophage death in the kidney as the underlying reason for macrophage accumulation.

Another hypothetic mechanism is that CX₃CR1 might cause immigrating monocytes to differentiate preferentially into macrophages rather than DCs. DC numbers are severely reduced in CX₃CR1-deficient mice (3), which would be consistent with altered differentiation. DCs do not play a major role in UUO, whereas macrophages are critical (18, 49, 54). This is not unexpected given that DCs primarily regulate adaptive immune responses, explaining why they affected fibrosis in models that involve adaptive immune cells, such as ischemia/reperfusion injury, crescentic glomerulonephritis and hypertensive nephropathy (3, 49, 54, 57). By contrast, a model of mechanic tissue injury like UUO is more likely mediated by cells of the innate immune system, like macrophages. However, adoptive transfer experiments revealed that monocytes expressing or lacking CX₃CR1 gave rise to similar ratios between macrophages and DCs, arguing against an effect of CX₃CR1-signaling on the monocytes' decision to differentiate into either DCs or macrophages. This conclusion was corroborated by unchanged levels of the growth factors that usually govern this decision, M-CSF and GM-CSF.

In vivo multiphoton microscopic analysis revealed morphologic alterations of CX₃CR1-deficient renal phagocytes. Although there were slight changes in the morphology of CX₃CR1⁺ kidney DCs, such as more yet shorter dendrites (our unpublished observations), it is unlikely that these alterations impacted fibrogenesis given that DCs do not affect renal fibrosis (18, 49, 54). However, in addition to resident DCs, two-photon microscopy revealed larger and rounder CX₃CR1⁺ cells that were more motile and maintained their shape over time, which is consistent with the morphology and dynamics of monocytes/macrophages. These cells were more abundant in CX₃CR1-deficient mice, prompting us to hypothesize about a functional link between macrophages and CX₃CR1. Indeed, we noted an inhibitory effect of CX₃CR1 on intrarenal macrophage proliferation. This can explain why these cells accumulated in CX₃CR1-deficient mice, although their precursor recruitment was less efficient and although more of them were apoptotic. As we excluded other potential causes, increased proliferation was most likely the cause for the accumulation of macrophages in the CX₃CR1-deficient kidney. It is possible that CX₃CR1 also prevents macrophage proliferation in DC-dependent models of renal disease (e.g., crescentic glomerulonephritis) but that this anti-inflammatory effect does not become manifest because its proinflammatory effect on the more numerous DCs prevails.

Traditionally, tissue macrophages were thought to arise from circulating monocytes recruited in response to locally produced chemokines (11). Recent findings have revealed that some tissue macrophages are derived from embryonic precursors (22). A sudden demand because of inflammatory conditions is met by the proliferation of such tissue-resident macrophages, in particular under Th2 or M2-associated conditions (72, 73, 80). We found that all macrophages, including those that had just recently been recruited to the kidney, showed stronger proliferation in the absence of CX₃CR1. These findings are consistent with the concept that local proliferation of tissue macrophages regulates their numbers and extend this concept by showing that also recently recruited macrophages are locally regulated in the inflamed kidney. It remains to be seen whether this mechanism also contributes to fibrosis in other organs.

CX₃CR1 not only affected the abundance, but also the functional state of renal macrophages. In its absence, more macrophages produced TGF-β, the key profibrotic mediator in the kidney (69). A previous study showed that CCR2-dependent macrophages produced TGF-β in UUO and that the lack of these cells significantly reduced renal fibrosis (16). Also, we found that both Gr1^– and Gr1⁺ macrophages produced TGF-β and that production was increased in the absence of CX₃CR1. TGFβ production suggested acquisition of a profibrotic functional state change. This state could not be explained by M2 differentiation, nor was it associated with higher renal inflammation. Recently, the classical M1/M2 paradigm was refined by the discovery of an activation-independent functional state that does not fall into the M1 or M2 definition (71). Our findings are consistent with this notion by demonstrating that TGF-β–producing macrophages cannot be categorized as classical M1 and M2 macrophages. Future studies are required to clarify whether these profibrotic macrophages belong to a distinct functional state.

In summary, we report a novel inhibitory effect of CX₃CR1 on local macrophage proliferation and on fibrogenesis in the kidney. In crescentic GN, the previously described proinflammatory influence of CX₃CR1 on DCs prevailed (3) because macrophages are downstream of DCs in the pathomechanism of this disease and because DCs outnumber macrophages in the kidney. However, in DC-independent renal diseases like UUO, the inhibitory effect of CX₃CR1 on macrophages becomes manifest, as we demonstrated in the present study, so that CX₃CR1 inhibition in such diseases aggravates fibrosis. This may apply also to organs in which CX₃CR1 does not regulate DCs, like the liver or the lung (3). In this case, therapeutic CX₃CR1 inhibition in liver and lung disease may turn out profibrotic.

We thank F. Tacke for critically reading the manuscript and Dr. Camden Lo for assistance with in vivo imaging analytic parameters. We acknowledge support by the Flow Cytometry Core Facility and the House for Experimental Therapy of the medical faculty of Bonn University.

The authors have no financial conflicts of interest.