Kidney-resident macrophages promote a proangiogenic environment in the normal and chronically ischemic mouse kidney

Renal artery stenosis (RAS) caused by narrowing of arteries is characterized by microvascular damage. Macrophages are implicated in repair and injury, but the specific populations responsible for these divergent roles have not been identified. Here, we characterized murine kidney F4/80+CD64+ macrophages in three transcriptionally unique populations. Using fate-mapping and parabiosis studies, we demonstrate that CD11b/cint are long-lived kidney-resident (KRM) while CD11chiMϕ, CD11cloMϕ are monocyte-derived macrophages. In a murine model of RAS, KRM self-renewed, while CD11chiMϕ and CD11cloMϕ increased significantly, which was associated with loss of peritubular capillaries. Replacing the native KRM with monocyte-derived KRM using liposomal clodronate and bone marrow transplantation followed by RAS, amplified loss of peritubular capillaries. To further elucidate the nature of interactions between KRM and peritubular endothelial cells, we performed RNA-sequencing on flow-sorted macrophages from Sham and RAS kidneys. KRM showed a prominent activation pattern in RAS with significant enrichment in reparative pathways, like angiogenesis and wound healing. In culture, KRM increased proliferation of renal peritubular endothelial cells implying direct pro-angiogenic properties. Human homologs of KRM identified as CD11bintCD11cintCD68+ increased in post-stenotic kidney biopsies from RAS patients compared to healthy human kidneys, and inversely correlated to kidney function. Thus, KRM may play protective roles in stenotic kidney injury through expansion and upregulation of pro-angiogenic pathways.

contrast to proinflammatory monocyte-derived macrophages, TRMφ may participate in tissue repair, blunting fibrosis and inflammation 9 . To discern myleoid cells subtypes, the Immunological Genome Project has defined mouse dendritic cells (DCs), monocytes, and macrophages based on surface markers 10 . Co-expression of F4/80, CD64, MerTK, and FCRIV, is used to identify macrophages 10 in the kidney 11 , lung, liver, spleen and gut 12 , where they prevent fibrosis by inducing tissue-specific repair programs.
In the mouse kidney F4/80 bright macrophages exhibit features of both DC and macrophages 13,14 . Phenotypic characterization of F4/80 bright macrophages, recently carried by Cao et al. 15 and Stamatiades et al. 11 , suggest expression of MHCII, Cx3cr1, CD11c and FCRIV. Ontogeny studies by Ginoux and colleagues demonstrated that F4/80 bright CD11b int kidney macrophages derive from fetal monocytes that arise from erythromyleoid progenitors generated in the yolk sac 16 . Furthermore, parabiosis studies demonstrate that less than 1% of F4/80 bright cells exchange between parabiont mice and are Ccr2-independent 11 . Recent studies suggest functional role of kidney-resident macrophages (KRM) in recruiting monocytes and neutrophils in the kidney in response to small immune complexes. Furthermore, F4/80 bright KRM and the endothelial cells form a functional unit that monitors the transport of particles 11 , highlighting the physiological function of KRM. However, little is known about their role in RAS, an ischemic kidney disease marked by decrease of renal capillaries. Since KRM is the largest population of macrophages we sought to study its role in RAS.
In this study, based on the expression of CD11b and CD11c, we phenotypically classified renal F4/80 + CD64 + macrophages in three subsets. Using parabiosis and fate-mapping, we identified KRM as F4/80 bright CD11b int CD11c int and monocyte-derived macrophages as CD11c hi Mφ and CD11c lo Mφ. All three macrophage populations expanded in the murine model of RAS, associated with loss of plasmalemma vesicle-associated protein (PLVAP) + CD31 + peritubular capillaries. Using irradiation followed by bone-marrow transplantation, we replaced native KRM with monocyte-derived KRM and studied the effect of RAS on monocyte-derived KRM. However, unlike the native KRM, monocyte-derived KRM did not sustain in RAS kidneys, and their loss was associated with amplified loss of peritubular capillaries. Adminsitration of liposomal clodronate too depleted native KRM which in turn resulted in loss of peri-tubular enodthelial cells. To explore the associated mechanisms we performed transcriptional profiling of all macrophages from Sham and RAS kidneys. We observed that native-KRM in RAS kidneys predominantly upregulate reparative pathways, like angiogenesis and wound healing. Furthermore, in-vitro studies demonstrated that co-incubation with RAS-KRM promote proliferation of peritubular endothelial cells. KRM-like CD11b int CD11c int CD68 + also increased in biopsies from human RAS kidneys compared to healthy subjects, and positively correlated with kidney function. Our findings suggest that KRM may protect the kidney during chronic ischemic injury.

Results
Renal macrophages comprise of long-lived KRM and monocyte-derived CD11c hi and CD11c lo macrophages. Cells were prepared by enzymatically digesting saline-perfused normal C57BL/6 mouse kidneys, followed by lineage depletion and antibody staining for macrophage markers (Figs 1A, S1A). To define the role of KRM in renal ischemia, we first identified F4/80 + CD64 +/lo kidney macrophages by flow cytometry 12,17 .
Furthermore, we performed phenotypic characterization on all three macrophage populations by flow cytometry. We observed that KRM were negative for Ly6c-a marker for blood derived cells, but expressed FCRIV, Cx3cr1, MHC class II and the lowest levels of CD45 (Figs 1B, S1B) as described previously 11,15,18 . Our observations agree with earlier studies suggesting that KRM account for >70% of macrophages 10,19,20 . Thus, KRM were further studied in detail. The CD11c lo Mφ are Ly6c hi FCRIV + and CD45 hi , and therefore could represent classical monocyte-derived macrophages, while CD11c hi Mφ are CD43 + Ly6c int and may be derived from non-classical monocytes 10,11,21 (Table 1) (Figs 1B, S1B).
Taken together, these data indicated that F4/80 bright CD11b int CD11c int are long-lived KRM, while infiltrating monocyte-derived macrophages are CD11c hi Mφ and CD11c lo Mφ. KRM are the largest population of macrophages in the healthy kidney, and we subsequently investigated their behavior pattern in chronic renal injury.
KRM self-renew while monocyte-derived macrophages expand in ischemic kidneys correlating with loss of peri-tubular endothelial cells. RAS caused hypertension and progressive loss of stenotic kidney volume (Figs 2A-C, S5A) 24 . Flow cytometric analysis showed increased numbers of total CD11b + F4/80 + macrophages and complementary DC (cDC1) in the stenotic kidney after 4 weeks of RAS (Fig. S5,A-C). Furthermore, the number of CD11c hi Mφ, CD11c lo Mφ, and KRM steadily rose with duration of ischemia (Fig. 2D). By 4 weeks the number and expression (Fig. 2E-G) of all macrophages increased in stenotic compared to Sham and contralateral kidneys, suggesting that both resident and monocyte-derived macrophages respond to ischemic injury. The CD11c lo Mφ subset was Ly6C hi , while CD11c hi Mφ and KRM were Ly6C lo . The increase in CD11c hi Mφ and CD11c lo Mφ in stenotic kidneys resulted from recruited Ly6C hi monocytes differentiating into Ly6C hi Mφ that contribute to CD11c lo Mφ, or become Ly6C lo macrophages that contribute to CD11c hi Mφ, both of which are CD11b hi , in agreement with previous studies in unilateral ureteral obstruction 25 . CD11b hi cells promote fibrosis and macrophage infiltration in injured kidneys 26,27 . Thus, our data are consistent with the notion that CD11c hi Mφ and CD11c lo Mφ in the stenotic kidney in part differentiate from recruited Ly6C hi monocytes and might cause inflammatory kidney damage.
Using CX3CR1 creER :Rosa26-tdTomato reporter mice, we then fate-mapped and studied KRM kinetics in RAS (Fig. 2H). KRM number increased in RAS (Figs 2I, S6A), and BrdU pulse-labeling demonstrated that KRM and CD11c hi Mφ expanded (Figs 2J, S6B). Interestingly, a small fraction of BrdU + CD11b int Mφ was tdTomato neg , suggesting that in RAS some infiltrating monocytes may contribute to KRM. Thus, all macrophages increase in response to renal ischemia, associated with renal fibrosis (Trichrome and Picrosirius-red staining) (Fig. S5F) and capillary loss (Fig. 2K,L). Peritubular microvascular loss was ascertained using immunofluorescence of CD31 and PLVAP (Fig. 2K), which selectively stains peritubular capillaries 28 , and flow cytometry confirmed reduced PLVAP + CD31 + cells in stenotic kidneys (Fig. S5D), suggesting that RAS induces capillary rarefaction. This was associated with increased expression of pro-inflammatory genes (Fig. S5E) and fibrosis (Fig. S5F) in RAS kidneys. Thus, in renal ischemia monocyte-derived macrophages are recruited from the circulation, while KRM are long-lived and progressively self-renew. The overall increase in macrophages is associated with loss of peritubular capillaries and renal fibrosis.

Donor-derived monocytes repopulate the KRM niche.
To test whether KRM repopulate from BM, we irradiated wild-type C57BL/6 mice (CD45.2), and transplanted bone marrow (BMT) from CD45.1 mice (Fig. 3A). All three renal macrophage populations were replenished by donor-derived BM (Fig. 3B,C). Thus, under stress conditions, monocytes repopulate the TRM niches in an attempt to restore KRM 29 . In the liver, circulating monocytes completely populate empty Kupffer cell niches and eventually form fully functional monocyte-derived Kupffer cells. Similarly, monocyte-derived alveolar macrophages demonstrate similar gene expression profile as embryonic alveolar macrophages 30,31 . Thus, BM cells can repopulate the KRM niche in the irradiated non-stenotic kidneys.

Depletion of KRM amplifies RAS-associated microvascular rarefaction.
To assess the role of KRM in microvascular rarefaction we used two strategies. First, native KRM were replaced by donor-derived KRM in BMT followed by RAS induction in mice (Fig. 3A) and secondly, continuous depletion of native KRM was induced using low-doses of liposomal clodronate in Sham and RAS mice (Fig. 3G).
Previous studies have administered higher doses of clodronate multiple times to achieve complete macrophage depletion [32][33][34] . In our model, liposomal clodronate at a single intraperitoneal dose of 200 ul significantly reduced blood monocytes (Fig. S8A), yet did not completely deplete KRM (Fig. S8B). Also, the depleted macrophages were replenished within 72 hours (Fig. S8B). Infiltrating monocytes and the remaining native KRM may have replenished the macrophage pool. Moreover, depletion of macrophages provoked neutrophil influx. To achieve gradual and continued depletion of KRM, and restrict neutrophil influx, we therefore subsequently used low-doses of clodronate 100 ul (FormuMax, Scientific CA) intraperitoneally every 4 days for 4 weeks.
Administration of liposomal clodronate to RAS mice selectively reduced KRM, but not the CD11c lo Mφ and CD11c hi Mφ (Fig. 3H, Fig. S8D). Reduction of KRM was associated with reduced number of PLVAP + CD31 + cells (Fig.I, J) and increased fibrosis (Fig. S8F) in the stenotic kidneys of the RAS + clodronate group. Similarly, the expression of anti-inflammatory and pro-angiogenic genes such as Arg1, Il4, Il10, Smad7, Angpt1, Igf1, Vcam1, Agtr2, Stat6, Mertk, Icam1 and the transcription factor Hbp1 (Fig. S8G,H) was reduced in KRM from RAS + clodronate group. These findings suggested that monocyte-derived cells could replenish KRM, but in the pathological setting of renal ischemia the monocyte-derived KRM lack the reparative ability of native KRM.
Menezes et al. observed that monocyte-derived Kupffer cells repopulated the liver to give rise to kupffer cells, but were unable to perform native kupffer cell functions 31 . Monocyte-derived alveolar macrophages express significantly higher pro-fibrotic genes than native embryonic-derived alveolar macrophages in bleomycin induced lung injury 35 . Thus, in active disease the monocyte-derived KRM may be less reparative than native KRM.
Wound healing pathways upregulated by RAS-KRM included classical tissue repair genes like Il10, Arg2, Tgfb2, and Tgfbr3. In healthy kidneys, RAS-KRM upregulated Arg2 and Il10 signaling, important pathways promoting wound healing 38 . Our findings corroborate previous reports that KRM are IL10-producing macrophages in the kidney 19 and further support their reparative potential (Fig. 4D,G).
In RAS, monocyte-derived KRM fail to upregulate pro-angiogenic and anti-inflammatory genes expressed by native KRM. To compare the gene expression profile of monocyte-derived with native KRM, we flow-sorted KRM from the stenotic kidneys of RAS and RAS + Clodronate mice, and performed qPCR. We   observed that expression of pro-angiogenic genes (e.g, Angpt1, Vegfa) and anti-inflammatory genes (e.g, Il10, Arg1) was reduced, while the expression of pro-inflammatory genes like Irf5 and Nfkb1 increased significantly (Fig. S8,H,I). Thus, monocyte-derived KRM might be occupying KRM niches, but are unable to mimic native KRM function.

KRM enhance proliferation of peri-tubular endothelial cells.
To study the direct contribution of KRM to these pathways, we performed additional in vitro experiments. The functional effect of KRM on endothelial cell proliferation was studied by co-incubation with PLVAP + CD31 + endothelial-cells, which represent renal peritubular capillaries (Figs 2K, S4G). This population was flow-sorted and then co-incubated with either BM-macrophages, RAS-KRM, or Sham-KRM. Proliferation determined by EdU incorporation was greater when co-incubated with RAS-KRM compared to control and Sham KRMs (Fig. 5A). This was further confirmed using dye-dilution experiments, where co-incubation of peri-tubular endothelial cells with RAS-KRM enhanced their proliferation (Fig. 5B).

KRM inhibit TGF-β-induced Collagen-1α1 expression.
RNA-sequencing showed that KRM upregulated gene involved in extracellular matrix remodeling. Therefore, to determine if KRM directly affect fibrosis, we incubated murine embryonic fibroblasts (MEFs), obtained from mice expressing green fluorescent protein (GFP) under the collagen-α1(I) promoter (Col1-GFP) with TGF-β, and measured GFP expression as readout for collagen synthesis. At 18 h, TGF-β induced a dose-dependent increase in GFP signal intensity, reaching statistical significance at 2 ng/ml (Fig. 5C, Left). The dependence on TGF-β signaling was confirmed using UO126 (MEK pathway inhibitor) and LY2109761 (TGF-β receptor inhibitor) (Fig. 5C, Left). Addition of Sham and RAS KRM to MEF co-incubated with TGF-β reduced the GFP signal (Fig. 5C, Right), suggesting that KRM directly counter TGFβ-mediated pro-fibrotic signaling in MEF, possibly related to their ability to upregulate anti-fibrotic genes in renal ischemia 39 . This was not observed in bone-marrow derived macrophages or CD11c hi/lo Mφ.

CD11c Int CD11b Int CD68 + Macrophages may represent a KRM-like population in human stenotic
kidneys. To assess the potential clinical relevance of our findings we initially identified macrophages by flow cytometry in the unaffected portion of a human kidney removed due to renal cell carcinoma. We used a combination of conventional markers like CD68, HLA-DR, CD11b, CD11c, CD14, CD16, and the additional markers CD64 and MerTK 12,40 . Macrophages were classified as Lineage neg CD45 + HLA-DR + CD68 + and CD11b + CD14 + but CD16 lo-neg , indicating a blood-derived origin 5,33,41 . We then further classified macrophages as CD11b hi CD11c hi , CD11b int CD11c lo-neg , and a small population of CD11b hi (Fig. S7A). Interestingly, unlike in mice, the expression of CD64 and MerTK was higher in the CD14 hi CD11b hi subset (Fig. S6B) 40,42 resembling human dermal CD14 + tissue-resident monocyte-derived macrophages that express CD64 42 . CD11b int CD11c lo-neg Mφ were CD14 lo and thus appear to phenotypically resemble KRM (Fig. S9B).
The number of KRM-like CD11c Int CD11b Int CD68 + increased in stenotic compared to healthy human kidneys (Fig. 6B), correlated directly with stenotic-kidney GFR and volume, and inversely with renal vein oxygen level and fibrosis (Fig. 6C). Hence, KRM-like CD11c Int CD11b Int CD68 + Mφ may play a reparative role in the human kidney.

Discussion
This study shows that F4/80 bright CD64 + CD11c/CD11b int are murine KRM with reparative potential. KRM self-renewed in stenotic kidneys. Irradiation and clodronate-induced depletion of KRM led to their replenishment via donor-derived bone marrow monocytes, but in superimposed RAS, native KRM depletion amplified the loss of peritubular capillaries, suggesting functional deficit of monocyte-derived KRM compared to native KRM. Indeed, native KRM in RAS transcriptionally upregulated expression of pro-angiogenic and wound healing pathways, capable of initiating a reparative transcriptional program to limit kidney damage. In-vitro, RAS-KRM promoted peritubular endothelial cell proliferation and blunted TGF-β-induced collagen-1 production. Furthermore, KRM homologues expand in human stenotic kidneys, and correlate with better function.
Our previous studies in humans and swine have shown that RAS leads to irreversible microvascular rarefaction [43][44][45] . Our murine model of RAS mimics the human RAS and demonstrates renal ischemia, hypertension, reduced glomerular function 46 and fibrosis 47 that in turn leads to renal inflammation and loss of peritubular capillaries and thus is a relevant model for chronic sterile renal injury. Hence, we have observed association of macrophages with RAS in mice, swine, as well as human RAS kidneys.
Renal macrophages have divergent phenotypes, gene expression profiles, and responses to physiologic stimuli 11,19 , which we linked to their origin. Using fate-mapping and parabiosis studies, we identified KRM, classical peri-tubular endothelial cells in representative images of RAS + Vehicle and RAS + Clodronate kidney sections. (J) Quantifying PLVAP + CD31 + cells. Significant loss of peritubular endothelial cells seen after administration of clodronate. n = 6 mice/group; *P ≤ 0.01 vs Sham; § P < 0.05 vs BMT + Sham , RAS + Vehicle; ¥ P < 0.01 vs RAS. Mouse images adopted from Openclipart.org https://openclipart.org/detail/174870/mouse and https:// openclipart.org/detail/28929/kidneyreins.  monocyte-derived CD11c lo Mφ, and non-classical monocyte-derived CD11c hi Mφ. Among these, KRM were the most abundant in the healthy mouse kidney, self-maintained, and progressively self-renewed in the stenotic kidney, whereas CD11c hi Mφ and CD11c lo Mφ were circulation-derived and short-lived. Consistent with previous studies 10,22 , quiescent KRM expressed typical markers of TRM. Thus, our data are congruent with the notion that resident-macrophages are long-lived and capable of self-maintaining 10,21,22,48 .
Irradiation depleted native KRM and unsealed tissue niches that were then occupied by monocytes from donor bone marrow thus giving rise to monocyte-derived KRM 29 . However, while phenotypically comparable to KRM, a superimposed ischemic stress revealed that those cells could not be sustained, and a decline in their numbers in RAS was associated with capillary loss. Continuous administration of low-dose clodronate had a similar effect. Importantly, we observed that monocyte-derived KRM formed in the progressive RAS had greater expression of pro-inflammatory genes than native KRM. Similarly, in injured hearts monocyte-derived macrophages have limited ability to restore TRM, and when restored TRM lack the reparative function of their embryonically-derived counterparts 49 . BM-derived Kupffer cells, despite normal density and location, also have reduced phagocytic activity 31 . Similar phenomenon has been recently demonstrated in lung 35 . Monocyte-derived alveolar macrophages proved to be profibrotic compared to native, alveolar TRM. Thus, although KRM ostensibly repopulate after BMT, the new subset resembles their native counterparts phenotypically by surface markers, rather than functionally. Contrarily, irradiation might have been adversely affected the niche by inducing cell cycle arrest and senescence. Future studies are needed to define whether a longer homing time would allow new cells to acquire the functional attributes of bone-fide KRM 29 .
Our transcriptional profiling agrees with the earlier report that at steady state Fcgr4 is upregulated in KRM compared to CD11c hi Mφ and CD11c lo Mφ 11 . Similarly, KRM differentially expressed microglia marker Tmem119 but not the Kupffer cell marker Clec4f 48,50 . More importantly, we identified Spp1, Mmp12 and Mmp13 as genes that are not expressed in other TRM or other kidney cells, but unique to KRM. We validated Mmp12 and Mmp13 transcripts in kidney macrophages using the mouse cell atlas 51 and kidney single-cell atlas 52 ; however, Spp1 was present in many other kidney cells. Interestingly, while Mmp12 and Mmp13 expression decreased in RAS-KRM, Spp1 increased. Further studies will highlight the role of these genes in the kidney diseases. Unlike Kupffer cells or Microglia, RAS-KRM show genes that are differentially expressed in functional units of kidney such as thick-ascending limb (Umod, Slc12a1 and Cldn10) 53 , intercalated cells of collecting duct (Aqp6, Idh3a, Slc4a1) 53 , distal convoluted tubule (Slc12a3, Tmem52b, Atp1a1), parietal cells (Aqp2) and podocytes (Bcam) 52 . Tissue-resident macrophages at steady state are not known to express these genes, therefore we speculate that in ischemic conditions, KRM may efferocytose apoptotic or senescent kidney cells, or phagocytose the extracellular vesicles released by the damaged cells thereby expressing their genes. Phagocytosing EVs may be unique to KRM because they form an anatomical and functional unit with endothelial cells that monitors the transport of small particles 11 . Further studies are needed to identify genes or transcription factors that promote efferocytosis in KRM.
In RAS kidneys, the overall increase in all macrophage populations was intensified with increased fibrosis and loss of peri-tubular endothelial cells, but only KRM responded by upregulating pro-angiogenic and wound-healing pathways, which are vital for repair by supplying the newly formed tissue with nutrients and oxygen. Top 100 DEGs in the ischemic kidney KRM include the Vegf family of genes that initiate the vessel sprouting. TRM initiate VEGF-dependent vascular anastomosis to form vascular networks 54 . PLVAP, a modulator of VEGF-induced angiogenesis 55 , contributes to peritubular capillary formation 56 . Our studies indicate that in vivo, depletion of native KRM using clodronate was associated with loss was peri-vascular endothelial cells. In culture KRM promoted proliferation of PLVAP + CD31 + peritubular endothelial cells and attenuated an increase in TGFβ-induced Col1a1 expression. The ability of RAS-KRM to promote endothelial cell proliferation and attenuate collagen formation supports the notion that KRM may possess reparative properties. Indeed, resident-macrophages may facilitate resolution of fibrosis in kidney 32,57 and liver injury 57 , possibly via formation of functional physiological units with endothelial cells 11,58 . In BMT + RAS, loss of KRM may have disrupted these units, magnifying loss of peri-tubular capillaries.
We identified human homologues of KRM as CD11b int CD11c int CD68 + . Renal CD11c hi are considered as DC, and CD11b hi CD11c hi CD68 + may be either macrophages or DCs 33,59 . Elevated CD11b int CD11c int CD68 + macrophages in stenotic human kidneys align with our observations in mice. Importantly, their numbers directly correlated with better kidney function and oxygenation, and inversely with fibrosis and atrophy, providing potential clinical support for their functionally consequential reparative role.
Our study may bear possible limitations. Despite a relatively brief period of RAS, stenotic murine kidneys recapitulate many pathological events seen in human and larger animal RAS 60 . Besides, macrophages, smaller populations of monocytes and DCs likely also contribute in RAS. We have studied transcriptomics at 4 weeks of RAS, which may represent tissue repair/reorganization phase; studying earlier time-points may help identify pro-fibrotic macrophage subsets. Stenotic-kidney KRM appear to heterogeneously express both pro-and anti-inflammatory and -fibrotic genes, but a KRM sub-population might have possibly dictated some of these expression patterns. Single-cell RNA-sequencing studies may help elucidate this possibility. We could not fully define the extent to which reciprocal changes in inflammatory macrophages vs. KRM regulate kidney damage.
presented as heat-maps with hierarchical clustering. Mean values per Mφ populations are shown. The z-scorebased color-scale shows gene expression standard deviations below (blue) or above (red) the population mean. While low-dose clodronate selectively deleted KRM amplifying fibrosis, we cannot rule out neutrophil contribution to this effect. Quantification of KRM in human kidney could have thresholding artifacts, which may contribute to some discrepancy between flow cytometry analysis and immunofluorescence.
In summary, our results suggest that KRM are a unique subset of renal macrophages, phenotypically equivalent to fate-mapped renal CD11b int F4/80 hi macrophages that are tissue-resident, self-maintain locally, and replenish slowly 22 . In response to microvascular rarefaction in stenotic mouse kidneys, KRM transcriptionally upregulate proangiogenic and wound healing pathways bearing a potential to repair damaged tissues. Human homologues of KRM identified as CD11b int CD11c int CD68 + increase in stenotic kidneys and correlate with kidney vitality. Further studies to exploit KRM, may open therapeutic avenues for treatment of chronic renal disease.

Methods
The study protocol was approved by Mayo Clinic's Institutional Animal Care and Use Committee (Protocol numbers A00001844-16 and A32415-16) and all experiments were performed in accordance with IACUC guidelines and regulations. Most RAS procedures were performed on n = 30-50; 20-week-old wild-type C57BL/6 J mice (Jackson Laboratories). Syngeneic mice expressing CD45.1 and wild type CD45.2 were used for Parabiosis and BMT studies.

Induction of Renal artery stenosis (RAS).
To induce RAS, mice were anesthetized with 1.75% isoflurane supplemented with O 2 and placed prone on a heating pad at 37 °C. The right kidney was exposed by a flank incision and its renal artery bluntly dissected from the renal vein. A 0.15 mm (ID × 0.5 mm) long polytetrafluoroethylene tube (Braintree Scientific, Braintree, MA) cuff was placed around the right renal artery and tied with 10-0 nylon sutures. Kidneys were then returned to their original positions and the incisions sutured. Blood pressure was measured before RAS or Sham surgery (baseline) and at 2, 4, and 6 weeks after surgery by tail-cuff, using an XBP1000 system (Kent Scientific, Torrington, CT). Sham surgery consisted of isolation of the renal artery without placement of a cuff 61 . At day 3, 7, and 28 days post-RAS 4-6 mice were euthanized to assess expansion of macrophages.
Fate-mapping studies and Tamoxifen dosing. B6.129P2(C)-Cx3cr1tm2.1(cre/ERT2)Jung/J (Jackson Labs #020940) were crossed to B6.Cg-Gt(ROSA)26Sortm14(CAG-tdTomato)Hze/J (Jackson Labs #007914). The F1 offspring were Cx3cr1 CreER+/− Rosa26 +/− . At 6-8 weeks of age mice of both sexes were injected intraperitoneally tamoxifen (Sigma) prepared in warm ethanol and mixed with corn oil. Around 75 mg tamoxifen/kg body weight was injected intraperitoneally to Cx3cr1 CreER+/− Rosa26 +/− induce recombination for 5 consecutive days. Mice were euthanized at 4 weeks and only the resident-macrophages were tdTomato positive. At 20 weeks of age, RAS/ Sham surgeries were performed on n = 20 (n = 10 per group) tamoxifen injected Cx3cr1 CreER+/− Rosa26 +/− mice 22 .  (40 X ) showing DAPI, CD68-AF488 (green), CD11b-AF594 (red), CD11c-AF647 (magenta), and merged (arrows); KRM-like cells were identified as CD11c Int CD11b Int CD68 + in healthy (row 1) and stenotic (rows 2-3) kidneys; G: Glomerulus. (B) CD11c Int CD11b Int CD68 + macrophage numbers are significantly higher in RAS compared to normal human kidneys. (C) The number of CD11c Int CD11b Int CD68 + correlated directly with GFR and kidney volume, and inversely with renal vein oxygen tension and degree of fibrosis measured by Trichrome staining *P < 0.05 vs Normal (n = 5-7 healthy human kidneys, n = 14 stenotic kidneys). Parabiosis. 6-8-week-old C57BL/6 (n = 4) congenic CD45.1 and CD45.2 mice were surgically connected in parabiosis as previously described 20,62,63 . After corresponding lateral skin incisions were made from elbow to knee in each mouse, forelimbs and hindlimbs were tied together using suture and the skin incisions were closed using stainless steel wound clips (Fine Scientific Tools Inc, USA). After surgery, mice were maintained on a diet supplemented with trimethoprim/sulfamethoxazole for prophylaxis of infection. 10 weeks after the parabiosis surgery the mice were euthanized, perfused and kidneys were harvested. Detail methods for tissue digestion, single cell preparation, flow cytometry, RNA-sequencing and validation by gene expression are provided in Supplemental Methods. Patient Protocol. Patients were identified as part of a clinical investigation of tissue oxygenation in human renovascular disease between 2008 and 2012. Informed, written consent was obtained after receiving approval from the Mayo Clinic's Institutional Review Board in adherence with the Declaration of Helsinki. A 3-day inpatient protocol was performed in the Clinical Research Unit of St. Mary's Hospital, Rochester, Minnesota. Fourteen patients underwent transvenous biopsy of the right-sided stenotic kidney via the jugular vein. Inclusion criteria were the presence of unilateral right-sided ARAS >70% obstruction, as previously described 64 (Table S1).
For the healthy group, Implantation biopsies obtained from 15 living kidney donors, selected to have a similar distribution of age and sex, were identified from the Mayo Kidney transplant program as previously described 65 . All research was performed in accordance with Mayo Clinic's IRB regulations. Detailed methods and hemodynamic data for RAS patients is elaborated in Supp Methods.
Immunofluorescence labeling of human kidney biopsies. Non-tumor pieces of kidneys were obtained from patients undergoing nephrectomy for renal cell carcinoma. Informed, written consent was obtained after receiving approval from the Mayo Clinic's Institutional Review Board (IRB#16-009485) in adherence with the Declaration of Helsinki. All research was performed in accordance with Mayo Clinic's IRB regulations. These kidney pieces were enzymatically digested and subjected to flow cytometry to identify macrophage markers (Table S3). Informed, written consent was obtained after receiving approval from the Institutional Review Board of the Mayo Clinic in adherence with the Declaration of Helsinki from all patients.

Statistics.
All statistics and graphs were generated using GraphPad Prism 7.1, and data presented as Mean ± S.E.M. One-sample t-test was used for comparing resolution metric (Rd). For gene expression and percent macrophages of total cells, unpaired Student t-test or Mann-Whitney test was applied, and P < 0.05 considered significant. Multiple groups were tested for significance using ANOVA followed by Dunnett's multiple comparisons test. For RNAseq, pairwise comparisons between macrophage populations (CD11c hi Mφ, CD11c lo Mφ, and KRM), as well as comparisons between sham and RAS for each macrophage population, were conducted by applying Wald test of the negative binomial distribution to the log2 gene counts using the DESeq. 2 statistical package 66 , and genes that showed statistically significant differences were selected (fold-change > 2, P < 0.05).