Targeting fibroblast CD248 attenuates CCL17-expressing macrophages and tissue fibrosis

The role of fibroblasts in tissue fibrosis has been extensively studied. Activated fibroblasts, namely myofibroblasts, produce pathological extracellular matrix. CD248, a type I transmembrane glycoprotein, is expressed in fibroblasts after birth. In human chronic kidney disease, upregulated CD248 in myofibroblasts is linked to poor renal survival. In this study, we demonstrated a novel interaction between CD248 and macrophages to be a key step in mediating tissue fibrosis. CD248 was upregulated in myofibroblasts in murine models of renal and peritoneal fibrosis. Cd248 knockout (Cd248–/–) could attenuate both renal and peritoneal fibrosis. By parabiosis of GFP reporter mice and Cd248–/– mice, we showed that attenuation of renal fibrosis was associated with a decrease of macrophage infiltration in Cd248–/– mice. Moreover, decrease of chemokine (C–C motif) ligand 17 and Ccl22 was found in macrophages isolated from the fibrotic kidneys of Cd248–/– mice. Because galectin-3-deficient macrophages showed decreased Ccl17 and Ccl22 in fibrotic kidneys, we further demonstrated that CD248 interacted specifically with galectin-3 of macrophages who then expressed CCL17 to activate collagen production in myofibroblasts. Mice with DNA vaccination targeting CD248 showed decreased fibrosis. We thus propose that CD248 targeting should be studied in the clinical tissue fibrosis setting.


Results
Targeted disruption of the Cd248 gene attenuates tissue fibrosis. CD248 involvement in tissue fibrosis was first examined in a renal fibrosis model induced by unilateral ureteral obstruction (UUO). Cd248 transcript was upregulated in UUO-induced fibrotic kidneys of WT, but not in Cd248 lacZ/lacZ (hereafter Cd248 -/-) mice (Fig. 1a). Loss of CD248 resulted in reduced deposition of Sirius red + collagen (Fig. 1b,c, Supplementary Figure S1), and reduced upregulation of Col1a1 transcript in UUO-induced fibrotic kidneys (Fig. 1d). In transgenic Col1a1-GFP Tg mice who synthesized green fluorescence protein (GFP) under the control of collagen I α1 chain promoter, CD248 expression was confirmed in both Col1a1-GFP + control kidney pericytes and UUO-induced kidney myofibroblasts (Supplementary Figure S2). By fluorescence-activated cell sorting (FACS) from Col1a1-GFP Tg mice, pericytes and myofibroblasts were isolated from the control kidney and UUO fibrotic kidney, respectively. Specifically, Cd248 transcript was detected in the pericytes, but markedly upregulated in myofibroblasts (Fig. 1e). We then crossed Col1a1-GFP Tg mice with Cd248 -/mice, such that the offspring would have GFP-expressing Cd248 -I-UUO-kidney myofibroblasts. FACS-isolated GFP + UUO-kidney myofibroblasts showed that Col1a1 was upregulated in WT mice but attenuated in Cd248 -/mice (Fig. 1f). These data show that collagen expression in renal myofibroblasts was reduced by the loss of their own CD248.
In the second renal fibrosis model induced by unilateral ischemia-reperfusion injury (uIRI), Cd248 disruption resulted in reduced collagen deposition and fibrosis, too (Supplementary Figure S3). CD248 involvement in tissue fibrosis was further examined in a murine model of chemical-induced peritoneal fibrosis by intraperitoneal injection of sodium hypochlorite to mimic severe fibrosis in encapsulating peritoneal sclerosis as previously described 27 . We observed accumulated Cd248-expressing fibroblasts in injured peritoneum (Supplementary Figure S4). Again, Cd248 disruption attenuated peritoneal adhesion and fibrosis (Supplementary Figure S5a-c). Together, these results identified that CD248 plays a significant role in tissue fibrosis and Cd248 disruption attenuates fibrosis.
Cd248 disruption does not affect cell numbers or proliferation of myofibroblasts. To investigate whether reduced fibrosis in UUO kidneys of Cd248 -/mice was due to fewer scar-producing myofibroblasts, we analyzed cell numbers and their proliferation ability. Immunolabeling studies revealed no differences in α-smooth muscle actin (αSMA) + area and Ki-67 + αSMA + cells between Cd248 -/and WT UUO kidneys, indicating that loss of Cd248 did not affect cell numbers and proliferation of myofibroblasts in UUO kidneys (Fig. 2a-c). Upregulation of Acta2 transcripts, which encoded αSMA, in UUO kidneys was not affected by loss of Cd248 (Fig. 2d). We quantified the cell numbers of αSMA + Col1a1-GFP + myofibroblasts and αSMA -Col1a1-GFP + pericytes in the UUO kidneys of Col1a1-GFP Tg mice with or without Cd248 knockout on day 7 after surgery (Supplementary Figure S6). No difference in the cell numbers was found, confirming again that loss of CD248 did not affect the cell proliferation of pericytes/myofibroblasts after UUO injury. Cd248 disruption did not affect the cell numbers of αSMA + myofibroblasts in fibrotic peritoneum, either (Supplementary Figure S5d, e). Platelet-derived growth factor (PDGF)-B-stimulated proliferation did not differ between Cd248 -/and WT myofibroblasts isolated and cultured from UUO kidneys (Fig. 2e,f). Therefore, the reduction of fibrosis in UUO kidneys of Cd248 -/mice could not be attributed to a reduction in myofibroblast numbers or their proliferation potential.
Cd248 affects macrophage infiltration and pro-fibrotic phenotype switching in injured organs. We observed marked accumulation of F4/80 + macrophages in UUO kidneys, but macrophage accumulation was notably reduced in the UUO kidneys of Cd248 -/mice (Fig. 3a,b). Attenuation of macrophage accumulation in injured peritoneum was also found in Cd248 -/mice (Supplementary Figure S5f-h).
In parabiosis experiments designed to investigate monocyte recruitment, each transgenic CAG-EGFP mouse, hereafter referred to as GFP Tg , was surgically joined to a WT or Cd248 -/mouse. FACS showed that > 50% of circulating CD45 + cells expressed GFP in both WT and Cd248 -/parabionts 14 days postoperatively, demonstrating successful cross-circulation (Supplementary Figure S7). Relative to the UUO kidneys of WT parabionts Cd248 disruption attenuated murine obstructive renal fibrosis. (a) Quantitative polymerase chain reaction (qPCR) of renal Cd248 in mice of wild type (WT) and CD248 knockout (Cd248 -/-) before (control), on day 7 (D7) and day 14 (D14) after unilateral ureteral obstruction (UUO) surgery. Cd248 expression was normalized by glyceraldehyde 3-phosphate dehydrogenase (Gapdh) and then compared with that of WT control. n = 10. (b) Representative images of picrosirius red staining in kidneys before (control) and on D14 after UUO surgery. Original magnification, × 100. Scale bar, 100 μm. (c) Quantification of Sirius red + collagen fibrils on low-powered field (LPF) images of kidney sections taken at 100× magnification. n = 10. (d) qPCR of renal Col1a1, which encoded the collagen I α1 chain. n = 6-10. (e) qPCR of Cd248 in pericytes and myofibroblasts isolated from control and D7 UUO kidneys of Col1a1-GFP Tg mice, respectively. n = 6. (f) qPCR of Col1a1 in myofibroblasts isolated from D7 UUO kidneys of WT and Cd248 -/mice. n = 6. Data are expressed as means ± standard errors of the mean. ***P < 0.001 by one-way ANOVA with post hoc Tukey's multiple comparisons test in (a,c,d) and *P < 0.05 by unpaired t-test in (e,f). www.nature.com/scientificreports/ following UUO surgery, immunofluorescence studies showed notably reduced cell numbers of GFP + leukocytes, F4/80 + macrophages, and GFP + F4/80 + macrophages in the UUO kidneys of Cd248 -/parabionts although these cell numbers were similar between control kidneys of WT and Cd248 -/parabionts (Fig. 3c,d), suggesting that loss of CD248 reduced monocyte recruitment to injured kidneys. Transcripts of M2-biased cytokine genes Ccl17 and Ccl22 were downregulated in the macrophages isolated from UUO kidneys of Cd248 -/mice, whereas M1-biased cytokines and enzymes were not affected (Fig. 3e). In the transwell co-cultures of RAW264.7 macrophages with WT UUO-kidney myofibroblasts, the transcripts of Nos2, Arg1 and Ccl17 in the macrophages were increased 2-to 3-folds (Fig. 3f). However, no significant increases were observed for any of the three transcripts in macrophages co-cultured with Cd248 -/-UUO-kidney myofibroblasts (Fig. 3f). Moreover, the effect of CD248 on Ccl17 expression was re-confirmed in bone marrowderived macrophages (BMDMs) co-cultured with WT or Cd248 -/myofibroblasts (Fig. 3g). Supplementation with recombinant CD248 (rCD248) increased Ccl17 expression in BMDMs co-cultured with either WT or Cd248 -/myofibroblasts (Fig. 3g). Moreover, CCL17 was found to increase transwell migration of RAW264.7 macrophages but the effect was blocked by anti-CCL17 antibody (Supplementary Figure S8a,b), suggesting an autocrine stimulatory effect of CCL17 on macrophage migration. To study the Ccl17-expressing cells in the kidneys, we performed in situ hybridization for Ccl17 and Adgre1 (Supplementary Figure S9). Ccl17 and Adgre1 increased expression in the interstitium of UUO kidneys, and the principal Ccl17-expressing cells were Adgre1 + macrophages. These results demonstrate an influence of CD248 on CCL17 expression of macrophages thereby stimulating the migration of macrophages into injured kidneys.

CD248 interacts specifically with galectin-3 and promotes CCL17 expression in macrophages.
Because galectin-3 plays a significant role in the activation of M2 macrophages and UUO-induced renal fibrosis 28,29 , we were intrigued about the interaction between galectin-3 and CD248 in renal fibrosis. The transcripts of Lgals3 increased in the UUO kidneys after surgery (Supplemental Figure S10). We confirmed that UUO-induced renal fibrosis was decreased in Lgals3 -/mice (Supplementary Figure S11). Compared to the expression in the UUO-kidney macrophages of WT mice, UUO-kidney macrophages of Lgals3 -/mice had reduced expression of Ccl17 and Ccl22 while the expression of M1-biased cytokines and enzymes was not different (Fig. 5a), complementing our observations in Cd248 -/mice. In addition to macrophages, the epithelia of renal collecting ducts were positive for galectin-3 immunostaining (Supplementary Figure S12). To study whether galectin-3 was expressed in renal pericytes or myofibroblasts, we performed in situ hybridization for Lgals3 and Pdgfrb (Supplementary Figure S13). Pdgfrb increased expression in interstitium after UUO surgery, but no Pdgfrb-expressing cells were positive for Lgals3. We therefore studied the possible mechanism of myofibroblast CD248 underlying the promotion of CCL17 expression in macrophages subsequently. In the high-powered field image, the proximity of CD248-galectin-3 could be demonstrated (Supplemental Figure S14). We then expressed CD248-DDK and galectin-3 in separate HEK293T cells for use in co-immunoprecipitation cell lysate assays. Anti-DDK antibody pulled down CD248 in association with galectin-3; reciprocally, anti-galectin-3 antibody pulled down galectin-3-associated CD248 (Fig. 5b,c). CD248 extracellular domain-enhanced GFP fusion protein (CD248ECD-EGFP) bound WT BMDMs but not Lgals3 -/-BMDMs (Fig. 5d). Binding specificity was demonstrated by reversing the binding with the addition of lactose, not sucrose to the reaction (Fig. 5e).
The functionality of the CD248 and galectin-3 interaction was analyzed in assays for Ccl17 expression. rCD248 upregulated Ccl17 in WT but not Lgals3 -/-BMDMs ( Fig. 5f). Complementing the finding for pro-fibrotic influence of CCL17 in the UUO model ( Fig. 4), CCL17 stimulated Col1a1 expression of UUO-kidney myofibroblasts who expressed C-C motif chemokine receptor 4 (CCR4) (Fig. 5g,h). Both WT and Cd248 -/myofibroblasts expressed CCR4, and no difference was found in the expression of Col1a1 stimulated by CCL17 (Fig. 5g,h). No effect of CCL17 on Acta2 expression was found (Fig. 5h). Taken together, these data indicate that myofibroblast CD248 induced pro-fibrotic phenotype switching of macrophages through galectin-3, and that macrophage CCL17 upregulated collagen expression of myofibroblasts thereby leading to renal fibrosis. Loss of CD248 reduced CCL17 production by macrophages and then collagen production by myofibroblasts.
Vaccination with Cd248 cDNA fused with the C fragment of tetanus toxoid attenuates renal fibrosis. We hypothesized that vaccination with Cd248 cDNA fused to the C fragment of tetanus toxoid (Cd248-TT) would confer adaptive immunity targeting kidney myofibroblasts and thereby attenuate renal fibrosis. To test this possibility, Cd248-, Cd248-TT-, and Tetc-expressing plasmids were transfected into HEK293T cells and their gene expression was confirmed (Fig. 6a). Anti-CD248 antibodies were detected after intra-muscular plasmid DNA vaccination by electroporation (Fig. 6b,c). Picrosirius red staining showed attenuation of renal fibrosis in the UUO kidneys of mice following Cd248-TT vaccination (Fig. 6d,e). The increased transcripts of Col1a1 in the UUO kidneys were also reduced in mice following Cd248-TT vaccination (Supplementary Figure S15). Moreover, although F4/80 + macrophages accumulated in UUO kidneys, their accumulation was reduced in Cd248-TT vaccinated mice, relative to that in control vector vaccinated mice (Fig. 6f,g). Thus, CD248 vaccination attenuated renal fibrosis effectively.

Discussion
The mechanisms by which macrophages become polarized to a pro-fibrotic phenotype have yet to be fully defined. It may be that a combination of signals from epithelial cells, endothelial cells, myofibroblasts, and perhaps other inflammatory cells trigger phenotype switching in recruited macrophages 14,17,18,30 . The present results are consistent with the notion that signals from myofibroblasts in particular play an important role in enabling fibrosis of tissue tissues. Moreover, we demonstrated, for the first time, that pro-fibrotic phenotype switching of macrophages during tissue fibrosis involves an interaction from CD248 of myofibroblasts. The early influx of pro-inflammatory monocytes/macrophages into injured organs has been shown to promote apoptosis of injured parenchymal cells and remove cellular debris 14,17,18,[20][21][22][23][24] . These pro-inflammatory macrophages switch functionally to a trophic phenotype that suppresses inflammation and promotes cell proliferation and the repair of parenchyma during the resolution phase of acute injury healing 17,20,[30][31][32] . However, when pro-inflammatory macrophages switch to a pro-fibrotic phenotype, they produce growth factors and cytokines that promote myofibroblast proliferation, resulting in tissue fibrosis, especially under conditions of a persistent or recurring injury 14,18,19,[21][22][23][24]30 . Pro-inflammatory macrophages are believed to be recruited to injured tissue in response to chemoattractants produced by resident dendritic cells/macrophages and injured epithelial/endothelial cells 9,14,17,18,[33][34][35] . Previous studies on progressive tissue fibrosis have focused on the pro-fibrotic effects of macrophages on myofibroblast activation and epithelial/endothelial damage [13][14][15]19,30,35 . Consistent with previous reports 14,19,[21][22][23][24]30 , the present results demonstrate myofibroblasts can be activated by CCL17 from macrophages. This finding is in line with previously reported evidence showing that macrophage ablation protects against myofibroblast accumulation and tissue fibrosis, possibly through the amelioration of CCL17-induced activation of myofibroblasts 18,28,36 .
Galectin-3 has been shown crucial for IL-4/IL-13-induced macrophage activation 29 . Selective pharmacological inhibition of extracellular galectin-3 carbohydrate binding has also been shown to block IL-4-induced activation, suggesting that the galectin-3 extracellular domain (ECD) may play a role in the alternative pathway of macrophage activation 29 . Although IL-4/IL-13 is able to switch pro-inflammatory macrophages to arginase-1-, mannose receptor-, and IL-10-expressing macrophages in vitro, similar phenotype switching has been shown to occur independent of IL-4/IL-13 signaling in vivo 17,26 .
In contrast to the indiscernible effect of IL-4/IL-13 signaling on the pro-fibrotic phenotype switching of macrophages in vivo 17,26 , our parabiosis and co-culture experiments showed that myofibroblasts can promote the macrophage recruitment and pro-fibrotic phenotype switching through an interaction of CD248 and galectin-3 during tissue injury and fibrosis. Upon injury, myofibroblasts increased in number and surrounded the recruited pro-inflammatory macrophages. Through the interaction between CD248 and galectin-3, CCL17 was induced in macrophages to activate migration of macrophages and collagen production of myofibroblasts. Conversely, in the absence of CD248, we observed reduced numbers of pro-fibrotic macrophages and reduced expression of CCL17, and these effects were associated with decreased tissue fibrosis. In line with our findings, CCL17, a M2 macrophage-skewed chemokine, shows pro-fibrotic action in tissue fibrosis including lung, skin and peritoneum 18,24,[37][38][39] . Our data showed that more than 80% of pericytes differentiated into αSMA + myofibroblasts on day 14 after UUO surgery and loss of CD248 led to decreased collagen expression rather than cell numbers of myofibroblasts. The proportion of positive αSMA immunostaining in Col1a1-GFP + pericytes on day 14 after UUO surgery was higher than the reported data derived from the dual reporter mouse that expresses red fluorescence protein (RFP) under direction of the Acta2 enhancer/ promoter (αSMA-RFP) in addition to Col1a1-GFP 40 . We thought this discrepancy was a result of different technique used in the experiments 41 . We would be very interested to know the sensitivity and specificity of immunostaining using αSMA antibody in the αSMA-RFP reporter mouse.
We demonstrated that CD248 bound to WT, but not Lgals3 -/macrophages and WT, but not Cd248 -/myofibroblasts induced Ccl17 expression in macrophages. Moreover, our result of CD248ECD-EGFP binding to galectin-3 on macrophage membranes in a manner that could be inhibited by lactose indicates that this interaction between the CD248-ECD and galectin-3 is specific. Our co-culture experiments showing that both soluble and membrane-bound CD248 affect the pro-fibrotic behavior of macrophages preclude a distinction between membrane-bound and soluble forms of CD248 as the major effector. and interferon γ (IFNγ)-primed RAW264.7 macrophages co-cultured with D7 UUOkidney myofibroblasts isolated from WT or Cd248 -/mice in the Transwell system. RAW264.7 macrophages co-cultured with medium were used as control. n = 4. (g) qPCR of Ccl17 in WT bone marrow-derived macrophages (BMDMs, Mϕ) co-cultured with medium only (control) or with WT or Cd248 -/-UUO-kidney myofibroblasts (MF) in the same dish. Recombinant CD248 (rCD248) was included in the culture as indicated. n = 5. Data are expressed as means ± standard errors of the mean. *P < 0.05, **P < 0.01, ***P < 0.001 by one-way ANOVA with post hoc Tukey's multiple comparisons test in (b,g) and unpaired t-test in (d,e,f).  www.nature.com/scientificreports/ Serum levels of CD248 might be a potential evidence to prove the DNA vaccine was working. However, we could not detect consistent serum levels of CD248. We thought that DNA vaccines mainly transfected muscle cells, thereby leading to an enhanced expression in the muscle cells and a concomitant increase in immune responses to the target antigen. Besides, we used CD248-TT DNA plasmid DNA containing full-length Cd248 cDNA and the C fragment of TT which has been used to elicit antigen-specific immune responses. The antigen might not be released to circulation at a detectable level because the target antigen would be a membrane bound CD248. A similar approach did not show the antigen detected in serum after DNA vaccination and the researchers showed the serum antibody titer in the animal model 42 .
The presently demonstrated effect of CD248 on macrophages represents a novel molecular function in profibrotic macrophage switching and myofibroblast activation. CD248 could potentially be targeted clinically to attenuate tissue fibrosis. Because of its restricted expression in normal tissues and upregulation in chronic fibrotic disease, blockade of CD248 in particular should be prioritized for clinical studies.

Methods
Mice. Cd248 lacZ/lacZ mice have been described in detail in our previous study 2 . In brief, the Cd248 gene was disrupted by replacing its single exon with the lacZ reporter gene. The mice were backcrossed to C57BL/6 mice for more than 10 generations. Mice were bred by mating heterozygous (+ /lacZ) parents. Littermates of the wild type (WT, Cd248 +/+ ), and homozygous knockin (Cd248 lacZ/lacZ ) mice were used for the experiments. Because homozygous lacZ knockin led to homozygous Cd248 knockout, Cd248 lacZ/lacZ mice were termed Cd248 -/mice thereafter. Col1a1-GFP Tg mice were generated and validated as described previously on the C57BL/6 background; their kidney pericytes and perivascular fibroblasts expressed enhanced GFP 4,11 . Cd248 -/mice were bred with Col1a1-GFP Tg mice to generate Cd248 -/-;Col1a1-GFP Tg mice. Littermate Cd248 +/+ ;Col1a1-GFP Tg mice were used as controls. Lgals3 -/mice were galectin-3 null; they were developed as described previously and backcrossed to C57BL/6 mice 43 . C57BL/6-Tg(CAG-EGFP)1Osb/J (hereafter referred to as GFP Tg ) mice were generated and validated as previously described on the C57BL/6 background whose cells, with exception of erythrocytes and hair, were green under excitation light 44 . WT C57BL/6 mice were obtained from The Jackson Laboratory (Stock No: 000664, Bar Harbor, ME). All studies were carried out under protocols approved by our Institutional Animal Care and Use Committee.

Models of organ fibrosis. A model of progressive renal fibrosis induced by UUO was conducted with male
adult (8-12-week-old) mice according to a previously described method 4,11 . Briefly, under anesthesia induced by ketamine/xylazine (100/10 mg/kg body weight intraperitoneally), each left ureter was exposed through flank incision with the mouse in the prone position. The ureter was ligated twice using 4-0 nylon surgical sutures at a level referenced to the lower pole of kidney to ensure that obstruction had a comparable effect in every mouse. The contralateral kidney in the same mouse served as a control. In neutralization experiments, UUO mice were injected with 4 μg of anti-mouse CCL17 (AF529; R&D Systems, Minneapolis, MN) or isotype control immunoglobulin G (IgG) (AB-108-C; R&D Systems) intraperitoneally daily until day 10 after UUO induction 37 .
A model of progressive renal fibrosis induced by uIRI was induced in male adult (8-12-week-old) mice according to a previously described method 4 . In brief, the left kidney of each anesthetized mouse was exposed through a surgical flank incision. A non-traumatic micro-aneurysm clip was placed across the renal artery and vein under a homeothermic blanket system (Stoelting Co., Wood Dale, IL), which contained a rectal thermal probe and a heating pad to maintain the core body temperature at 37 °C. The kidney was confirmed to become dusky, and was replaced in the retroperitoneum for 30 min. The clamps were removed and the return of perfusion to the kidneys was confirmed before wound closure.
A model of progressive peritoneal fibrosis induced by injection of sodium hypochlorite was conducted according to a previously described method 27 . Briefly, injury was induced by intraperitoneal injection of 100 ml/kg body weight normal saline with 0.05% sodium hypochlorite. The peritoneal adhesion score was determined by assigning 1 point for each adhesion present between the abdominal wall and intestine, intestine and intestine, intestine and omentum, omentum and kidney, and kidney and liver, according to a previously described method 27 . Parabiosis. Parabiosis was performed in adult (8-12-week-old) mice according to a previously described method 45 . Briefly, anesthetized GFP Tg , WT and Cd248 -/mice were shaved and a unilateral flank skin incision from the elbow to the knee joint was created in each. The skin edge was sutured with 5.0 prolene to generate GFP Tg -WT and GFP Tg -Cd248 -/parabiotic mice. Fourteen days after parabiotic surgery, UUO was performed in GFP Tg , WT and Cd248 -/mice. Mice were sacrificed on day 3 after injury to analyze the recruitment of GFP + cells into injured kidneys of WT and Cd248 -/mice.
Tissues were then counterstained with nuclear fast red (Sigma), dehydrated in alcohols, and mounted with permount. Images were captured and processed using a Leica microscope or Zeiss laser capture confocal microscope 4,11,27 . Specific cells in tissue sections were quantified as described previously 4,11,27 . In brief, sections were co-labeled with DAPI, and GFP + cells were identified by blue and green nuclear co-localization; αSMA + and F4/80 + cells were identified by positive staining in > 75% of the cell areas immediately surrounding nuclei (detected by DAPI) with Cy3 fluorescence indicative of antigen expression; Ki-67 + cells were identified by positive nuclear staining. Specific cells were counted in 10 randomly selected cortical fields between renal capsule and cortico-medullary junction per mouse at 100× magnification. Interstitial fibrosis was quantified in picrosirius red-stained paraffin sections (Picrosirius Red Stain Kit, 24,901; Polysciences, Inc., Warrington, PA). The morphometry of Sirius red + collagen, αSMA + myofibroblasts and F4/80 + macrophages was examined in 10 randomly selected cortical fields between renal capsule and cortico-medullary junction per mouse at 100× magnification using the FoveaPro4 program (Reindeer Graphics, Inc., Asheville, NC). Peritoneal membrane thickness was quantified using Masson's trichrome staining of the peritoneal covering of the liver, as described previously 27 . Pdgfrb, Lgals3, Ccl17 and Adgre1 transcripts were detected on formalin-fixed, paraffin-embedded kidney tissue sections using the RNAscope Multiplex Reagent Kit according to the manufacturer's instructions (Advanced Cell Diagnostics, Newark, CA).
Isolation and culture of pericytes and myofibroblasts from kidneys. Details of pericyte and myofibroblast purification and culture from normal and fibrotic kidneys, respectively, have been provided previously 8,11 . Briefly, the kidneys were decapsulated, diced and then incubated at 37 °C for 30 min (normal kidneys) or 1 h (day 7 UUO-damaged kidneys) with liberase (0.5 mg/ml; Roche Applied Science, Indianapolis, IN) and DNase (100 U/ml; Roche Applied Science) in Hank's balanced salt solution (HBSS; Gibco, Thermo Fisher Scientific). After passage through a 70-µm nylon cell strainer (BD Biosciences, San Jose, CA), an equal volume of HBSS with 10% fetal bovine serum (FBS) was added to the single cell preparation to stop enzyme activity. The mixture was centrifuged at 2000 rpm for 5 min at 4 °C for cell collection. Cells were then resuspended in 5 ml phosphate-buffered saline (PBS) with 1% bovine serum albumin (BSA), and passed through a 40-μm nylon cell strainer (BD Biosciences). Pericytes and myofibroblasts were purified from the single cell suspension by isolating GFP + cells of Col1a1-GFP Tg mice using FACSAria cell sorting. Then, total RNA was isolated using a RNeasy Mini Kit (Qiagen, Valencia, CA) or cells were cultured in Dulbecco's modified Eagle medium (DMEM) supplemented with 20% FBS. The primary cultured cells collected between passages 4 and 6 were used in this study. Cultured myofibroblasts were stimulated with TGF-β1 (5 ng/ml, R&D Systems), CCL17 (100 ng/ml, R&D Systems), or vehicle control for 24 h. Then, the total RNA was extracted for gene expression analysis.

Isolation of renal macrophages by magnetic-activated cell sorting.
Mouse kidneys were decapsulated, diced and then incubated at 37 °C for 1 h with liberase (0.5 mg/ml; Roche Applied Science) and DNase (100 U/ml; Roche Applied Science) in HBSS. After passage through a 70-µm nylon cell strainer (BD Biosciences), an equal volume of HBSS with 10% FBS was added to the single cell preparation to stop enzyme activity. The mixture was centrifuged at 2000 rpm for 5 min at 4 °C for cell collection. Cells were then resuspended in 5 ml PBS with 1% BSA, and passed through a 40-μm nylon cell strainer (BD Biosciences). After centrifugation, the cells were resuspended in 500 μl 1 × BD IMag™ buffer (BD Biosciences). The cells were incubated in a tube with 50 μl BD IMag™ anti-mouse CD11b particles-DM (1:100; BD Biosciences) at 4 °C for 30 min. After mixing with 500 μl 1 × BD IMag™ buffer, the tube was immediately placed on the BD IMagnet™ and incubated at 4 °C for 6-8 min.
With the tube on the BD IMagnet™, the supernatant was carefully aspirated. The tube was then removed from the BD IMagnet™ and 1 × BD IMag™ buffer was added to 1 ml; it was immediately replaced on the BD IMagnet™ and incubated at 4 °C for 4-6 min. This process was repeated twice. After a final wash, the total RNA of CD11b + macrophages were isolated using a RNeasy Mini Kit (Qiagen, Valencia, CA).  www.nature.com/scientificreports/ electroporated using a NEPA21 Super Electroporator (Nepa Gene Co., Chiba, Japan). Two pulses of electroporation at 100 mV for 200 ms were performed. Anti-CD248 antibodies in serum were detected by enzyme-linked immunosorbent assay. Serum samples were 10 × diluted and added to 96 wells containing 5 ug/ml rCD248 (R&D Systems), then incubated at 4 °C overnight. Anti-mouse IgG-HRP and then 3,3′,5,5′-tetramethylbenzidine were used to detect the reaction. Finally, the optical density at 450 nm was read.
Statistical analysis. Data were expressed as means ± standard errors of the mean. Unpaired Student's t-test was used to compare two different groups. One-way ANOVA with post hoc Tukey's multiple comparisons test was used for the comparison between each group. A P value < 0.05 was considered significant. Statistical analyses were carried out using the GraphPad Prism software (GraphPad Software, La Jolla, CA).

Study approval.
All animal studies were carried out under a protocol approved by Institutional Animal Care and Use Committee, National Taiwan University College of Medicine and College of Public Health. The approval numbers were 20150255 and 20190148. All methods were performed in accordance with the relevant guidelines and regulations.