BASIC RESEARCH www.jasn.org

GM-CSF Promotes Alternative Activation after Renal Ischemia/Reperfusion Injury

† ‡ Sarah C. Huen,* Larry Huynh, Arnaud Marlier,* Yashang Lee,* Gilbert W. Moeckel, and Lloyd G. Cantley*

*Section of Nephrology, Department of Medicine, †Yale Cancer Center, and ‡Department of Pathology, Yale University School of Medicine, New Haven, Connecticut

ABSTRACT After kidney ischemia/reperfusion (I/R) injury, monocytes home to the kidney and differentiate into activated . Whereas proinflammatory macrophages contribute to the initial kidney damage, an alternatively activated phenotype can promote normal renal repair. The microenvironment of the kidney during the repair phase mediates the transition of macrophage activation from a proinflammatory to a reparative phenotype. In this study, we show that macrophages isolated from murine kidneys during the tubular repair phase after I/R exhibit an alternative activation profile that differs from the canonical alternative activation induced by IL-4–stimulated STAT6 signaling. This unique activation profile can be reproduced in vitro by stimulation of bone marrow-derived macrophages with conditioned media from serum-starved mouse proximal tubule cells. Secreted tubular factors were found to activate macro- phage STAT3 and STAT5 but not STAT6, leading to induction of the unique alternative activation pattern. Using STAT3-deficient bone marrow-derived macrophages and pharmacologic inhibition of STAT5, we found that tubular cell-mediated macrophage alternative activation is regulated by STAT5 activation. Both in vitro and after renal I/R, tubular cells expressed GM-CSF, a known STAT5 activator, and this pathway was required for in vitro alternative activation of macrophages by tubular cells. Furthermore, administration of a neutralizing antibody against GM-CSF after renal I/R attenuated kidney macrophage alternative activation and suppressed tubular proliferation. Taken together, these data show that tubular cells can instruct macrophage activation by secreting GM-CSF, leading to a unique macrophage reparative phe- notype that supports tubular proliferation after sterile ischemic injury.

J Am Soc Nephrol 26: 1334–1345, 2015. doi: 10.1681/ASN.2014060612

Macrophages are key participants in regulating the in- subsequently been detected in multiple types of in- flammatory response to infections and sterile injuries. fectious and sterile tissue injuries. M2 macrophages Macrophages are activated in response to pathogens have been typically subcategorized into M2a wound- and endogenous injury stimuli. The inflammation healing macrophages induced by IL-4/IL-13 activation that ensues acts to eliminate pathogens/toxins, restore of the IL-4Ra–STAT6 signaling pathway and regula- tissue homeostasis, andpromote tissue repair.Broadly tory M2b/c macrophages, which are activated by im- defined, macrophage activation has been divided into mune complexes and IL-10 or TGF-b, respectively.1 two major subtypes: classically activated proinflam- matory M1 macrophages and alternatively activated anti-inflammatory M2 macrophages. Classically ac- Received June 25, 2014. Accepted October 9, 2014. tivated M1 macrophages are induced by IFN-g and Published online ahead of print. Publication date available at activation of Toll-like receptors and promote antimi- www.jasn.org. crobial responses downstream of STAT1 and NF-kB Correspondence: Dr. Sarah C. Huen, Yale University School of signaling, respectively. Alternatively activated macro- Medicine, 333 Cedar Street, PO Box 208029, New Haven, CT phages were first defined as IL-4/IL-13–induced mac- 06510. Email: [email protected]. rophages seen in parasitic infections but have Copyright © 2015 by the American Society of Nephrology

1334 ISSN : 1046-6673/2606-1334 JAmSocNephrol26: 1334–1345, 2015 www.jasn.org BASIC RESEARCH

These categories have largely been defined in vitro with individ- RESULTS ual stimuli. In vivo, macrophage differentiation and activation of effector functions during the inflammatory response are dic- Tubular-Mediated Alternative Activation Is IL-4–STAT6 tated by a dynamic and complex microenvironment. Investiga- Independent tion of in vivo macrophages in a wide variety of disease models We have found that proinflammatory macrophages cocultured associated with inflammation, both infectious and sterile, has with MPT cells or primary tubular epithelial cells show increased shown a diverse spectrum of macrophage phenotypes that are expression of the classic alternative activation arginase-1 temporally and functionally dynamic.2 Identification of the in (Arg1)andCd206 (), which is independent of vivo signaling pathways that govern macrophage activation and IL-4Ra–STAT6 signaling10 (Supplemental Figure 1). To deter- regulate macrophage effector functions may, therefore, provide mine if tubular cell-induced alternative activation is distinct therapeutic targets to promote inflammation resolution and tis- from IL-4–induced alternative activation, naïve BMMs were sue repair. stimulated with IL-4 or CM from serum-starved MPT cells. In sterile ischemic kidney injury models, macrophages seem The profiles of IL-4– compared with MPT to play roles in all phases of the injury process, including the CM-treated macrophages show both overlapping (Arg1 and initial injury, subsequent repair, and late fibrosis.3,4 Within 24 Cd206) (Figure 1A) and divergent gene expression. Msr1 is in- hours after ischemia/reperfusion (I/R) injury, monocytes are duced by MPT CM but not IL-4 (Figure 1B), whereas Dectin-1, recruited to the kidney, where they differentiate into macro- Fizz1, Ym1,andIgf-1 are specifictoIL-4–induced macrophage phages. Proinflammatory macrophages predominate during alternative activation (Figure 1C). To determine the relevance of the early injury phase, during which tubular apoptosis is these distinct alternative activation profiles for in vivo macro- prominent. During the tubular repair phase, when tubular phage activation, macrophages were flow-sorted from control cells are proliferating and repopulating the denuded basement kidneys and ischemically injured kidneys on day 5 after reper- membrane, kidney macrophages begin to express markers of fusion (the time at which we detected increased Arg1 and Cd206 alternative activation. Macrophage depletion studies suggest expression10) and analyzed for alternative activation gene ex- that the functional phenotypes of the macrophages correlate pression. Consistent with the tubular cell activation profile, with each phase.5–11 Our previous in vivo studies using fluo- Msr1 expression is upregulated in macrophages at day 5 after rescently tagged ex vivo IFN-g–primed macrophages sugges- I/R injury compared with macrophages from uninjured kidneys, ted that signals within the injured kidney were sufficient to whereas the expression of downstream IL-4 gene targets (Dectin-1, induce this proinflammatory to anti-inflammatory pheno- Fizz1,andYm112) is not induced (Figure 1D). Of note, Igf-1 ex- typic switch in infiltrating macrophages.10 Both in vivo studies pression is upregulated on day 5 after I/R, suggesting that either using IL-4Ra null mice and in vitro coculture studies suggest there is selective IL-4 signaling to induce Igf-1 but not other IL-4 that the primary signal for alternative macrophage activation targets or a third pathway (nontubular and non–IL-4) may be in the injured kidney uses an IL-4Ra–independent pathway.10 responsible for Igf-1 expression. In this study, we show that the pattern of in vivo alternative macrophage activation after I/R injury is distinct from that Tubular Factors That Induce Macrophage Alternative seen after in vitro IL-4 stimulation but consistent with the Activation Are Basolaterally Secreted pattern of activation seen after macrophage exposure to tubular Proximal tubular cells are polarized epithelia, with the apical cell conditioned media (CM). To determine the mechanism by surface facing the tubule lumen exposed to the glomerular which tubular cells can promote macrophage alternative acti- filtrate and the basolateral surface attached to the basement vation, we defined the signaling pathways activated in macro- membrane facing the interstitium. After I/R injury, macro- phages by secreted tubular factors. In vitro studies with primary phages traffic to the interstitium of the kidney and line areas of bone marrow-derived macrophages (BMMs) and CM from tubular damage adjacent to the tubular basement membrane.10 serum-starved mouse proximal tubule (MPT) cells showed To model in vivo polarity and determine whether the polarity that tubular cell-secreted factors activate JAK–STAT (STAT3 of secreted tubular factors would be physiologically relevant in and STAT5) pathways in macrophages. Using BMMs from activating macrophages in the interstitium, MPT cells were fl fl LysM-Cre;Stat3 / mice as well as pharmacologic inhibition grown to confluency on 0.4-mm Transwell inserts, and CM of STAT5, we found that tubule cell-mediated macrophage al- were collected from either the apical- or basolateral-facing ternative activation is regulated by STAT5 activation. GM- chambers (Figure 2A). The basolateral MPT CM induced sig- CSF, a well known STAT5 activator, is upregulated in renal nificantly higher expression of Arg1 and Cd206 than the apical proximal tubule cells after I/R injury. We found that GM- CM (Figure 2, B and C), consistent with the model that tubu- CSF is secreted by MPT cells and required to induce alterna- lar cell-secreted factors can signal to macrophages in vivo. tive activation in BMMs. Functional blockade of GM-CSF with a neutralizing antibody both in vitro and in vivo attenuates Tubular-Secreted Factors Activate JAK–STAT Signaling tubular-mediated macrophage alternative activation, resulting Pathways in decreased tubular cell proliferation during the repair phase The JAK–STAT pathway is a major signaling pathway that after kidney injury. regulates macrophage activation.13–15 To determine

J Am Soc Nephrol 26: 1334–1345, 2015 GM-CSF Activates Kidney Macrophages 1335 BASIC RESEARCH www.jasn.org

Figure 1. Macrophages activated in vitro by secreted tubular factors and in vivo by the microenvironment of the postischemic kidney exhibit an alternative activation gene profile that differs from IL-4–activated macrophages. (A–C) Naïve BMMs were treated with vehicle media (white bars), tubular CM (black bars), or IL-4 (20 ng/ml; gray bars). Cell lysates were collected at 12 hours after treatment for RNA. mRNA expression 2 by quantitative PCR is shown relative to Hprt (n=3). (D) CD45+F4/80+CD11c kidney macrophages were isolated from uninjured kidneys and postischemic kidneys 5 days after I/R injury. mRNA expression by quantitative PCR is shown relative to Hprt (n=3 for day 0 and n=5 for day 5). Data are shown as means6SEMs.Exp,expression;Veh,vehicle.*P,0.05; **P,0.01; ***P,0.001; ****P,0.0001.

tubular cell-mediated macrophage alterna- tive activation. JAK inhibitor I efficiently in- hibited MPT CM-induced STAT3 and STAT5 activation in BMMs (Figure 4, A and B). This JAK–STAT inhibition led to almost complete inhibition of ARG1 expression (Fig- ure 4, A and B) and Arg1, Cd206,andMsr1 Figure 2. Alternative activation is induced by basolaterally secreted tubular cell factors. (A) mRNA expression (Figure 4C). MPT cells (0.3563106/4.2 cm2)wereplatedon0.4-mm Transwell inserts and allowed to grow to confluence (transepithelial resistance of 61.469.2 mOhmzcm2) over a 4-day period. Tubule Cell-Mediated Macrophage The cells were washed, and media were changed to serum-free media with equal volumes Activation Requires STAT5 but Not (2 ml) within the insert (apical) and below within the cell culture well (basolateral) as shown. STAT3 Activation CM were collected 48 hours later. Naïve BMMs were then treated with vehicle media, apical STAT3-dependent induction of alternative CM, or basolateral CM. Cell lysates were collected at (B) 12 hours for mRNA expression activation has been described through IL- (shown relative to Hprt) and (C) 24 hours for protein analysis (densitometry shown normal- –16 ized to b-actin; n=3). Data are shown as means6SEMs. ACTB, b-actin; Basolat, basolateral; 10 and more recently, MyD88-dependent Exp, expression; veh, vehicle. *P,0.05; **P,0.01 (vehicle, apical, and basolateral). autocrine production of IL-6, IL-10, and CSF-3 in mycobacteria-infected macro- phages.17 To determine whether STAT3 is re- whether a JAK–STAT signaling pathway other than IL-4Ra– quired for tubular-mediated alternative activation in STAT6 is involved in tubular cell-mediated macrophage acti- macrophages, a transgenic mouse model with myeloid-specific vation, STATsignaling pathways were analyzed in naïve BMMs knockout of STAT3 was generated. LysM-Cre transgenic mice18 at various time points after exposure to MPT CM. Macro- with myeloid-specific expression of Cre recombinase were mated fl fl phage STAT3 and STAT5 were phosphorylated early (within with Stat3 / mice.19 Macrophage Stat3 deletion was efficient in fl fl 5 [STAT3] to 30 [STAT5] minutes after treatment with tubular BMMs from LysM-Cre;Stat3 / mice, and there was no detectable fl fl CM) (Figure 3A), whereas STAT6 was not activated by CM. phosphorylation of STAT3 in LysM-Cre;Stat3 / BMMs treated Compared with IFN-g–induced macrophages, STAT1 was not with MPT CM (Figure 5, A and B). However, elimination of significantly activated. mRNA expression of the macrophage STAT3 activation did not affect CM-stimulated STAT5 activation alternative activation markers Arg1, Cd206,andMsr1 was in- or induction of ARG1 protein expression (Figure 5, A and B) or duced at 6–12 hours after treatment with tubular CM, whereas Arg1, Cd206,andMsr1 mRNA expression (Figure 5C). the expression of Igf-1 was not induced (Figure 3B). To determine whether STAT5 is required for tubular cell- A pan-JAK inhibitor, JAK inhibitor I, was used to determine mediated macrophage activation, N9-((4-Oxo-4H-chromen-3- whether JAK–STAT signaling is required for the induction of yl)methylene)nicotinohydrazide, a nonpeptidic small molecule

1336 Journal of the American Society of Nephrology J Am Soc Nephrol 26: 1334–1345, 2015 www.jasn.org BASIC RESEARCH

Figure 3. Secreted tubular factors induce JAK–STAT pathways. (A) Naïve BMMs were treated with MPT CM, and cell lysates were harvested for protein at indicated times for Western blot analysis. BMMs stimulated for 30 minutes with IFN-g (100 units/ml) and IL-4 (20 ng/ml) were harvested for protein as M1 and M2 controls, respectively. (B) Naïve BMMs were treated with MPT CM, and cell lysates were harvested at 0 and 30 minutes and 1, 2, 6, 12, 24, and 48 hours followed by RNA harvest and quantitative PCR. mRNA expression is shown relative to Hprt.ACTB,b-actin; Exp, expression.

STAT5 inhibitor that targets the SH2 domain of STAT5,20 was hypothesize that GM-CSF secreted by tubular cells mediates used. The STAT5 inhibitor reduced MPT CM-induced STAT5 STAT5-dependent macrophage alternative activation. Assess- activation by approximately 70% (P=0.02) and significantly de- ment of naïve macrophage activation in response to recombi- creased Cd206, with a concomitant reduction in ARG1 protein nant GM-CSF revealed that expression of Arg1, Cd206,and and Arg1 and Msr1 mRNA expression that did not reach statis- Msr1 was detectably induced at concentrations between 300 tical significance (Figure 6). These data suggest that STAT5 sig- and 500 pg/ml, whereas maximal activation of Arg1 and Msr1 naling rather than STAT3 is required for tubular cell-mediated required much higher concentrations (Figure 8A). Interest- macrophage alternative activation. ingly, the concentration of GM-CSF in basolateral MPT CM was 374654.6 pg/ml, whereas GM-CSF levels in the apical GM-CSF Expression Is Upregulated in the Kidney after MPT CM, which failed to induce alternative activation, were I/R Injury and Precedes STAT5 Activation below the induction threshold (185.4625.4 pg/ml). To deter- GM-CSF (or CSF2) is known to be a regulator of monocyte/ mine if tubular cell-secreted GM-CSF is required for in vitro macrophage differentiation and mediates its actions through the macrophage alternative activation, naïve BMMs were treated activation of JAK2 and STAT5.21,22 We found that GM-CSF ex- with MPT CM with or without the addition of a GM-CSF– pression in the whole kidney is increased within 24 hours after neutralizing antibody. These experiments revealed that the ischemic injury and peaks on day 3, preceding detectable STAT5 neutralizing antibody to GM-CSF prevented STAT5 activation activation and protein expression of the alternative activation after either GM-CSF or MPT CM stimulation (Figure 8, B and markers ARG1 and CD206 seen on days 3–5 after I/R injury C, quantified in Figure 8D). Furthermore, the neutralizing (Figure 7, A and B). To determine whether ischemic injury spe- antibody effectively inhibited Arg1, Cd206,andMsr1 expres- cifically induces tubular cell GM-CSF expression, proximal tu- sion in response to MPT CM, showing that GM-CSF is the bule cells were cell sorted by flow cytometry from kidneys 3 days tubule-derived signal for STAT5 activation and required for after I/R injury. Expression of Csf2 is significantly upregulated in tubule cell-induced alternative activation (Figure 8E). proximal tubule cells from the postischemic kidney compared Comparison of Arg1, Cd206,andMsr1 expression in re- with proximal tubule cells from uninjured kidneys (Figure 7C, sponse to macrophage stimulation with either MPT CM or Supplemental Figure 2 shows sorting specificity for proximal GM-CSF revealed that Arg1 and Msr1 were induced to a tubule cells). Dual immunostaining for phosphorylated STAT5 greater degree by MPT CM than GM-CSF at concentrations and the macrophage marker F4/80 revealed that 11.8%61.6% of comparable with those in MPT CM (compare the columns for F4/80+ cells in the outer medulla were positive for phosphory- 300 and 500 pg/ml in Figure 8A with MPT CM in Figure 8E). lated STAT5 on day 5 after I/R (Figure 7D). In contrast, Cd206 expression seems to be maximally stimu- lated by GM-CSF in this dose range. These findings suggest GM-CSF Secreted by Renal Tubular Cells is Required to that MPT CM contains factors in addition to GM-CSF that Induce Alternative Activation in Macrophages can induce Arg1 and Msr1 expression. Macrophage colony- The observation that GM-CSF expression and macrophage stimulating factor (M-CSF) is known to promote M2 macro- alternative activation are sequential in vivo led us to phage activation, expressed in proximal tubular cells after tubular

J Am Soc Nephrol 26: 1334–1345, 2015 GM-CSF Activates Kidney Macrophages 1337 BASIC RESEARCH www.jasn.org

Blockade of GM-CSF after Renal I/R Injury Attenuates Kidney Macrophage Alternative Activation and Slows Reparative Tubule Proliferation To define the relevance of GM-CSF–induced signaling for macro- phage activation in vivo, mice were subjected to bilateral renal I/R injury followed by administration of the GM-CSF–neutralizing antibody starting at 36 hours after I/R injury, with kidneys harvested for macrophage expression profiling on day 5 after reperfusion (Figure 9A). The timing of the initiation of GM- CSF–neutralizing antibody was chosen such that the initial proinflammatory macrophage influx and tubule injury, both of which peak at 24 hours,10 were unaltered. As expected, the initial tubular injury at 24 hours after reperfusion is identical in the two groups, which is reflected by the creatinine increase to 0.3 on day 1 (Figure 9B). This protocol of one time daily GM- CSF blockade led to a modest decrease in whole-kidney STAT5 activation on day 5 after I/R injury (Figure 9, C and D) but did achieve a significant decrease in the mRNA expression of Cis, a well known GM-CSF–STAT5 downstream gene target,27 by isolated kidney macrophages (Figure 9E). Although the total 2 number of CD45+F4/80+CD11c macrophages recruited to the kidney and the number of CD45+F4/80+CD11c+ cells was not significantly changed by GM-CSF blockade (Figure 9F) (data not shown), the expression of the alternative activa- tion markers Arg1, Cd206, and Msr1 were significantly reduced 2 in F4/80+CD11c macrophages, whereas Igf-1 expression was unchanged (Figure 9G). This partial inhibition of STAT5 acti- vation and macrophage alternative activation significantly re- Figure 4. Tubule cell-mediated alternative macrophage activation duced tubular cell proliferation on day 5 (Figure 9, H–J) but did is dependent on JAK–STAT activation. Naïve BMMs were pre- not affect GFR as estimated by serum creatinine (Figure 9B). treated with 0.5 or 1.0 mM JAK inhibitor I or vehicle (DMSO) for 30 minutes before treatment with MPT CM and vehicle/inhibitor. (A) Cell lysates were harvested at 30 minutes and 24 hours for protein DISCUSSION for Western blot analysis; they are shown relative to vehicle and normalized to b-actin loading control (densitometry in B). (C) At 12 The signals that instruct macrophages to alter their gene hours for RNA for quantitative PCR, expression is shown relative to expression profile and their functional repertoire to promote Hprt (n=3). Data are shown as means6SEMs. ACTB, b-actin; Exp, tubular repair after renal I/R injury are unknown. Our previous expression; Inh, inhibitor. *P,0.05; **P,0.01. studies showed that macrophages transition from a proin- flammatory state during the tubular apoptotic injury phase to a injury, and expressed by MPT cells at mRNA levels 80-fold reparative state during the tubular proliferative phase. We higher than Csf223–26 (Supplemental Figure 3, A and B). To found that tubular cells induce macrophage alternative acti- determine if M-CSF and GM-CSF act in concert to promote vation in vitro through an IL-4Ra–independent pathway. tubule-induced alternative activation, BMMs were treated with Here, we sought to investigate the mechanism by which tubu- GM-CSF with or without M-CSF. These experiments revealed lar cells can promote macrophage alternative activation. that M-CSF alone did not activate STAT5 and was insufficient Because JAK–STAT pathways are critical in regulating mac- to induce Arg1, Cd206, and Msr1 expression, whereas the com- rophage activation, we first determined whether a JAK–STAT bination of GM-CSF and M-CSF induced both Arg1 and Msr1 pathway other than IL-4Ra–STAT6 was involved in tubular to a greater degree than GM-CSF did alone (Supplemental cell-mediated macrophage alternative activation. Using Figure 3, C and D). In contrast, M-CSF did not accentuate mRNA expression profiling of macrophages isolated from in- GM-CSF–stimulated expression of Cd206. These data show jured kidneys, comparisons of IL-4– and tubular CM-induced that GM-CSF is both necessary and sufficient for tubule cell- macrophages, STAT3-deficient BMMs, and STAT5 inhibition, mediated STAT5 phosphorylation and Cd206 expression, we identified GM-CSF as a tubule-derived STAT5 activator whereas maximal expression of other markers of alternative mediating a unique macrophage alternative activation pheno- macrophage activation requires GM-CSF plus additional fac- type that differs from the canonical Th2 IL-4–induced alter- tors, such as M-CSF. native activation state.

1338 Journal of the American Society of Nephrology J Am Soc Nephrol 26: 1334–1345, 2015 www.jasn.org BASIC RESEARCH

Figure 5. Alternative activation of macrophages induced by secreted tubular cell factors is independent of STAT3. Naïve BMMs from wild- Figure 6. Macrophage alternative activation by secreted tubular fl/fl type and LysM-Cre;Stat3 mice were cultured with MPT CM. (A) factors correlates with STAT5 activation. Naïve BMMs were pre- Cell lysates were harvested at 30 minutes and 24 hours for protein. treated for 30 minutes with vehicle, 1.0 mM JAK inhibitor I, or (B) Densitometry of three separate experiments as in A normalized to 100 mM STAT5 inhibitor and then cultured in normal media and b-actin loading control. (C) Cells stimulated as above were harvested vehicle or MPT CM with or without the same inhibitor. (A) Cell 12 hours after stimulation for RNA, and expression is shown relative lysates were harvested at 30 minutes and 24 hours for Western to Hprt (n=3). Data are shown as means6SEMs. ACTB, b-actin; Exp, blot analysis. (B) Densitometry of Western blot analysis from four expression; Veh, vehicle. *P,0.05; **P,0.01; ****P,0.0001. separate experiments as in A shown relative to vehicle and nor- malized to b-actin loading control. (C) mRNA expression 12 hours fi after MPT CM treatment with vehicle or STAT5 inhibitor. Expression GM-CSF was rst described for its ability to generate shown relative to Hprt (n=4). Data are shown as means6SEMs. granulocyte and macrophage populations from bone marrow ACTB, b-actin; Exp, expression; Inh, inhibitor. *P,0.05; **P,0.01. precursor cells.28 It has become evident that GM-CSF also plays important roles in the activation and function of more mature myeloid cells in the setting of infection and inflammation.29 such as rheumatoid arthritis and crescentic GN,33 but also, as Macrophages differentiated in vitro in the presence of GM- an inducer of M2 markers in microglia/macrophages surround- CSF or M-CSF have basal cytokine expression profiles and ing glioblastoma tumor cells.34 Nonhematopoietic cells, such as responsiveness to LPS similar to M1 and M2 macrophages, re- vascular smooth muscle cells and colon epithelial cells, have been spectively.30,31 However, recent characterization of global gene reported to be important sources of GM-CSF, mediating athero- expression profiles of GM-CSF– and M-CSF–derived macro- genic monocyte activation35 and promoting colonic epithelial phages suggests that the M1 and M2 categorizations of these cell proliferation in colitis,36 respectively. In addition, GM-CSF cells may be limited, especially when comparing murine and has been associated to Th2 immune responses in allergic lung human macrophage gene expression.32 Furthermore, because inflammation.37,38 Thus, macrophage activation by GM-CSF the M1/M2 designation of GM-CSF– and M-CSF–differentiated and M-CSF treatment seems to be highly context dependent.39 BMMs is primarily on the basis of in vitro macrophages stimu- In our model of mouse renal I/R injury, an increase in GM- lated with LPS, it is unclear whether this designation holds true CSF expression within the kidney was detectable within 24 hours in the setting of sterile inflammation and in vivo in the presence after I/R and peaked on day 3, at which time proximal tubule cell of additional stimuli. The complexity of GM-CSF biology is expression of GM-CSF significantly increased. STAT5 activation evident, because it is implicated in inflammatory conditions, and expression of alternative activation markers in macrophages

J Am Soc Nephrol 26: 1334–1345, 2015 GM-CSF Activates Kidney Macrophages 1339 BASIC RESEARCH www.jasn.org

Figure 7. Macrophage STAT5 activation follows an increase in whole-kidney GM-CSF expression and correlates with alternative activation. (A and B) Whole-kidney lysates of uninjured kidneys (day 0) and kidneys 1, 3, and 5 days after I/R injury immunoblotted for pSTAT5, GM-CSF, CD206, and ARG1; expression was normalized to loading control b-actin. Densitometry in B is shown relative to baseline expression at day 0(n=4 per group). Data are shown as means6SEMs. *P,0.05; ****P,0.0001 versus day 0. (C) Real-time quantitative PCR for Csf2 using mRNA isolated from proximal tubule cells from kidneys 3 days after I/R. Expression is shown relative to proximal tubule cells from uninjured sham kidneys normalized to Hprt (n=4/group). ***P,0.001. (D) Representative images of F4/80 (red) and pSTAT5 (green) immunofluo- rescence at day 5 after renal I/R. Original magnification, 3400. ACTB, b-actin; DAPI, 49,6-diamidino-2-phenylindole follow during the tubular repair phase (3–5 days after I/R). Gen- GM-CSF–STAT5-independent factors, such as tubule-secreted eral depletion of macrophages during this tubular repair phase M-CSF24–26 and/or macrophage-secreted IGF-1, can provide results in decreased reparative tubular proliferation, suggesting sufficient reparative signals to partially compensate for the loss that the induction of alternative activation genes in macrophages of GM-CSF signaling. Although there was no difference in se- during this time is critical for macrophage-mediated tubular rum creatinine, decreased tubular proliferation during the re- repair.9,10 These studies used in vivo GM-CSF blockade after pair phase can have long-term consequences, such as fibrosis. renal I/R to assess the importance of this pathway in macrophage Serum creatinine can return to baseline levels, even when there activation and tubule repair. One time daily injection of the GM- is significant induction of late tubular fibrosis,40 suggesting that CSF–neutralizing antibody achieved only a partial reduction in hyperfiltration can underlie normalization of creatinine and STAT5 activation in the injured kidney at the one time point mask failure of normal repair. Although this study has iden- assessed (day 5), but the expression of all three alternative acti- tified an important pathway mediating the alternative activa- vation markers shown to require GM-CSF–STAT5 signaling in tion kidney macrophages after I/R injury, the modest ability vitro were significantly reduced in macrophages isolated from of the GM-CSF–neutralizing antibody to prevent STAT5 ac- these kidneys. Furthermore, the proliferation of tubular cells tivation in vivo will likely limit its use for studying these late that is known to underlie effective kidney repair was significantly outcomes. reduced after GM-CSF blockade. Our data show that the fullexpression of alternative activation Although the above observations provide strong support for after tubule injury requires factors other than GM-CSF and are the importance of tubule cell-secreted GM-CSF in promoting consistent with previous reports of M-CSF as an activator of the alternative macrophage activation after kidney injury, it is reparative macrophage phenotype.24–26 GM-CSF and M-CSF interesting to note that GFR, as estimated by serum creatinine, are both secreted by tubular cells and seem to serve as coinducers was not reduced in the mice exposed to the GM-CSF–neutralizing for maximizing macrophage expression of protective factors, antibody. It remains to be determined whether this is because such as arginase-1. In contrast, secreted tubular factors do not we achieved only a partial blockade of GM-CSF signaling with induce macrophage Igf-1 expression, although Igf-1 expression the once a day injection of the neutralizing antibody (compare is upregulated in kidney macrophages at day 5 after I/R (Figure the modest decrease in STAT5 phosphorylation and Cd206 1D). IGF-1 is a well known inducer of tubule cell proliferation,41 and Msr1 expression achieved with the neutralizing anti- and the mechanism behind its macrophage expression after kid- body in vivo with the more robust effect in vitro) or because ney injury clearly deserves additional study. Identification of

1340 Journal of the American Society of Nephrology J Am Soc Nephrol 26: 1334–1345, 2015 www.jasn.org BASIC RESEARCH

process of normal tubule repair as well as maladaptive fibrosis after kidney injury.

CONCISE METHODS

Animals C57BL/6 (National Cancer Institute/National In- 2 2 stitutes of Health), Stat6 / 42 (The Jackson Lab- oratory), LysM-Cre18 (The Jackson Laboratory), fl fl and Stat3 / 19 (gift from Xin-Yuan Fu) mice were used for primary bone marrow-derived macro- phage cultures. All animal protocols were ap- proved by the Yale University Institutional Animal Care and Use Committee.

Reagents and Antibodies JAK inhibitor I (420097) and STAT5 inhibitor (573108) from Calbiochem were purchased from EMD Millipore. Rabbit antibody anti–pSTAT1- Y701 (9171), anti–pSTAT3-Y705 (9145), antitotal STAT3 (9132), anti–pSTAT5-Y694 (9359), and anti–pSTAT6-Y641 (9361) were obtained from Cell Signaling Technology. Rabbit anti-CD206 an- tibody (ab64693) was purchased from Abcam, Inc. Goat antibody anti-ARG1 (sc-18354) and mouse antibody anti–b-actin (sc-69879) are from Santa Cruz Biotechnology. Recombinant mouseGM-CSFandM-CSFwereobtained Figure 8. GM-CSF secreted by tubular cells promotes alternative activation of mac- from R&D Systems. Anti-mouse F4/80 (BM8), rophages. (A) BMMs were treated with recombinant mouse GM-CSF at the indicated anti-mouse GM-CSF antibody for Western blot concentrations for 12 hours. Cell lysates harvested for RNA isolation and mRNA ex- – – pression are shown relative to Hprt.**P,0.01; ***P,0.001 versus 0 pg/ml. (B) Five (clone MP1 22E9, 14 7331), anti-mouse GM- micrograms per milliliter neutralizing (Neut) GM-CSF antibody or isotype (Iso) anti- CSF functional-grade antibody for neutralization body was added to culture media containing 300 pg/ml GM-CSF (GM) and incubated (clone MP1–22E9, 16–7331), and rat isotype con- for 3 hours at 37°C before treating naïve BMMs. Cell lysates of treated BMMs were trol antibody (rat IgG2a, k,16–4321) were ob- harvested at 30 minutes for Western blot analysis of pSTAT5. (C–E) Five micrograms tained from eBioscience. per milliliter Neut GM-CSF antibody or Iso antibody was added to MPT CM and in- cubated for 3 hours at 37°C before treating naïve BMMs. Cell lysates of BMMs har- In Vitro Cell Culture and CM Protocol vested at (C) 30 minutes for Western blot analysis of pSTAT5 (densitometry in D BMMs were isolated and cultured in the presence normalized to loading control b-actin is shown relative to vehicle-treated controls) and of L929 supernatant as per the protocol of (E) 12 hours for RNA. mRNA expression is shown relative to Hprt (n=3 for all experi- Cui et al.43 and as previously described.10 In all ments). Data are shown as means6SEMs. ACTB, b-actin; Exp, expression; Veh, ve- in vitro BMM experiments, after 7 days of differ- hicle. *P,0.05; **P,0.01; ***P,0.001; ****P,0.0001. entiation in the presence of L929 supernatant, macrophages were passaged and plated in MEM-a media containing 10% FBS without ad- these signals within the microenvironment of the postischemic ditional growth factors (i.e., no L929, M-CSF, or GM-CSF) for 24 hours kidney will further enhance our understanding of how repara- before initiation of experimental conditions. For in vitro polarization, tive macrophages are activated. BMMs were stimulated with either 100 units/ml murine recombinant Taken together, our data show that GM-CSF secreted by IFN-g (11276905001; Roche) or 20 ng/ml murine recombinant IL-4 tubular epithelial cells after I/R injury mediates a novel STAT5- (I1020; Sigma-Aldrich). BMMs were treated with recombinant mouse dependent macrophage activation profile that promotes tu- GM-CSF and M-CSF at specified concentrations. BMMs were stimu- bular proliferation during kidney repair. Future studies using lated with MPTCM (described below) for various time points. For MPT conditional knockout of the GM-CSF–STAT5 pathway will be CM, pathogen-free MPT cells (BU-MPT immortalized cells44) were needed to fully address the functional significance of this grown to confluence, washed with PBS three times, and cultured in unique mode of alternative macrophage activation in the serum-free DMEM:F12 media (Gibco). Media supernatant was

J Am Soc Nephrol 26: 1334–1345, 2015 GM-CSF Activates Kidney Macrophages 1341 BASIC RESEARCH www.jasn.org

Figure 9. GM-CSF regulates the expression of a subset of kidney macrophage alternative activation genes and promotes tubular proliferation. (A) Mice were subjected to 24 minutes of bilateral I/R and then injected with isotype antibody or neutralizing GM-CSF antibody daily starting on day 1.5 after I/R. All mice were euthanized on day 5 (2 hours after the final antibody injection). (B) Serum creatinine was measured on day 1 (before initiation of antibody treatment) and day 5 after I/R. (C) Whole-kidney lysates of day 5 postischemic kidneys from three mice receiving isotype control and three mice receiving neutralizing antibody immunoblotted for pSTAT5. (D) Densitometry of pSTAT5 relative to b-actin for all mice as shown in C. (E and G) Real-time quantitative PCR for the indicated 2 genes using mRNA isolated from CD45+F4/80+CD11c kidney macrophages; expression is shown relative to Hprt. (F) Percentage of 2 total gated kidney cells analyzed by FACS that are CD45+F4/80+CD11c .(H–J) Representative images of Ki67 immunostaining of kidney sections 5 days after I/R injury from mice that received either (H) isotype antibody or (I) GM-CSF–neutralizing antibody. (J) Quantification of Ki67-positive tubular cells in outer medulla (n=4 for isotype and n=5 for GM-CSF neutralizing antibody). Data are shown as means6SEMs. ACTB, b-actin; Exp, expression; ISO, isotype. Original magnification, 3400. *P,0.05; ***P,0.001. collected after 48 hours, centrifuged to remove debris, and filtered with a insert for the six-well plate. Separate experiments were performed using 0.2-mm filter. All BMM and MPT CM experiments were performed the 12-mm, 0.4-mm PET Snapwell Insert (9.53104 MPT cells plated with 10% FBS containing MEM-a media and either serum-free on a 1.12-cm2 growth area and grown to confluence in 4 days) for DMEM:F12 media or serum-free MPT CM at a 1:1 ratio. Ussing chamber transepithelial potential measurements. Transepithelial Collection of apical and basolateral CM was performed by plating resistance of the MPT cells at this cell density was 61.469.2 mOhmzcm2, 0.3563106 MPT cells on 0.4-mm transparent PET membrane trans- consistent with polarized proximal tubule cells. well inserts for a six-well plate (353090; Corning). Cells were cultured MPT cells deemed to be free of mycoplasma contamination were for 4 days in DMEM:F12 media with 10% FBS with antibiotics/anti- used to collect CM. Mycoplasma testing was performed with the mycotic (Invitrogen). After 4 days, the cells were washed with PBS LookOut Mycoplasma PCR Detection Kit (Sigma-Aldrich). MPT CM three times, and the medium was changed to serum-free DMEM:F12 were also tested for endotoxin using the ToxinSensor Gel Clot for 48 hours (2 ml in well: basolateral CM; 2 ml in insert: apical CM). Endotoxin Assay Kit (L00351; Genscript). MPT CM endotoxin level Cell density at the end of the 48-hour serum starvation period was was tested to be ,0.25 endotoxin units (EU) /ml (1 EU/ml is approx- approximately 3.53106 MPT cells on a 4.2-cm2 surface of the transwell imately 0.1 ng/ml). Of note, MEM-a media containing 10% FBS used

1342 Journal of the American Society of Nephrology J Am Soc Nephrol 26: 1334–1345, 2015 www.jasn.org BASIC RESEARCH as the vehicle control media tested positive for endotoxin levels.0.25 Western Blot EU/ml (industry standard for FBS is ,10 EU/ml). Lysates of mouse kidney were prepared by homogenization in 0.5 ml RIPA buffer containing SDS (Teknova) with Halt protease and phos- fi Renal I/R Model phataseinhibitorcocktail(ThermoFisherScienti c).Culturedcellswere Male 8- to 10-week-old C57BL/6 mice (National Institutes of Health/ also lysed directly in RIPA buffer containing SDS, supplemented with a National Cancer Institute) were anesthetized with intraperitoneal keta- protease and phosphatase inhibitor cocktail. Equal amounts of protein mine (100 mg/kg) and xylazine (10 mg/kg) and then subjectedtorenalI/R were loaded, and electrophoresis was performed in 10% polyacrylamide using a modified approach to that previously described.10 Briefly, both separating gel/5% stacking gel or precast Mini-Protean TGX 4%–15% the right and left renal pedicles were isolated through a midline abdom- gradient gel (Bio-Rad). were transferred to PVDF membrane inal incision and clamped for 24 minutes using a nontraumatic micro- and blocked with 5% milk in Tris-buffered saline with Tween 20 for 1 aneurysm clip (Fine Science Tools). Mice were kept at 37°C using a hour. Membranes were incubated overnight at 4°C. Blots were then warming pad, and reperfusion of both kidneys was confirmed after washed and incubated with secondary antibody for 1 hour at room clamp release. Mice were given 1 ml normal saline intraperitoneally to temperature. After washing, the second antibody was visualized by prevent dehydration. Mice were euthanized on days 0, 1, 3, and 5 after chemiluminescence reagents (PerkinElmer). reperfusion. Kidney tissue was harvested for histology, immunohisto- chemistry, and FACS sorting for RNA isolation. For in vivo GM-CSF ELISA neutralization experiments, five doses of 300 mg anti-mouse GM-CSF– GM-CSF in MPT CM was measured by sandwich ELISA using anti- neutralizing antibody or rat IgG isotype control were injected intraper- mouse GM-CSF antibody (554404; BD Pharmigen) as a capture itoneally daily starting 36 hours after I/R injury. Mice were euthanized on antibody and biotin anti-mouse GM-CSF antibody (554407; BD day 5; tissue and blood were collected for analysis. Serum creatinine was Pharmigen) as a detection antibody, with recombinant mouse GM- measured by HPLC (Yale Mouse Metabolic Phenotyping Center). CSF (R&D) as the standard.

Isolation of Kidney Proximal Tubule Cells and RNA Isolation, RT-PCR, and Quantitative Real-Time Macrophages. PCR Kidney macrophage isolation at day 5 of I/R injury was performed with RNA was extracted with an RNeasy Mini Kit (Qiagen) and reverse 10 fi kidney digestion as previouslydescribed. After blocking nonspeci cFc transcribed using the iScript cDNA Synthesis Kit (Bio-Rad). Gene binding with anti-mouse CD16/32 (clone 93; eBioscience), kidney sin- expression analysis was determined by quantitative real-time PCR (iTaq – gle-cell suspension was stained with anti CD45-PE (30F11; BD Biosci- Universal Sybr Green Supermix; Bio-Rad) using a CFX96 (Bio-Rad) and – ence), anti-F4/80 FITC (BM8; eBioscience), and anti CD11c-PE-Cy7 normalized to Hprt. Data are expressed using the comparative threshold 2 (HL3; BD Bioscience). Using unstained and isotype controls to identify cycle (dCt) method, and mRNA ratios are given by 2 dCt or fold dif- fl + + fi auto uorescent kidney cell populations, CD45 F4/80 cells were rst ference (ddCt) relative to a control. Primers used include previously 2 + + selected, and then, the CD11c population of the CD45 F4/80 cells published primers for Arg1, Cd206, Hprt, Igf-1, Msr1, Tgfb1,and was gated for isolation on the Sony SY3200. For isolation of kidney Ym110,45 as well as Cis forward: TCGGGAATCTGGGTGGTACT, reverse: proximal tubule cells, kidney single-cell suspensions were stained with GGGTGCTGTCTCGAACTAGG; Csf1 forward: GCTGGATGAT- Fluorescein-labeled Lotus Tetragonolobus Lectin (1:200; Vector Labo- CCTGTTTGCT, reverse: GCTGGAGAGGAGTCTCATGG; Csf2 forward: + + 2 ratories). Total RNAwas extracted from CD45 F4/80 CD11c or Lotus TGGTCTACAGCCTCTCAGCA, reverse: CCGTAGACCCTGCTC- lectin-positive cells and reverse transcribed for real-time PCR analysis of GAATA; Dectin1 forward: GGAATCCTGTGCTTTGTGGT, reverse: gene expression as described below. TAGTTTGGGATGCCTTGGAG; and Fizz1 forward: AGGAAC- TTCTTGCCAATCCA, reverse: CTGGGTTCTCCACCTCTTCA. Immunostaining Kidneys were perfused with PBS, fixed with 10% neutral buffered Statistical Analyses formalin, and processed for histology (hematoxylin and eosin; All results are expressed as means6SEMs. Comparisons were tested paraffin blocks) or immunostaining (5-mm cryostat sections). After with unpaired t tests and one- or two-way ANOVAs with post-test antigen retrieval with BD Retrievagen A, sections were permeabilized Bonferroni or Tukey multiple comparison analyses where appropri- in methanol for 10 minutes at 220°C. The sections were then blocked ate using Prism 6.0 (GraphPad Software, Inc.). Data are presented as at room temperature using PBS containing 0.1% BSA and 10% don- means6SEMs. P#0.05 was considered to be statistically significant. key serum. The following primary antibodies were used: anti-F4/80 (clone BM8; eBioscience and anti-pSTAT5 (9359; Cell Signaling Technology). Ki67 immunohistochemistry staining was performed by Yale Research histology. Quantification of cells expressing the spec- ACKNOWLEDGMENTS ified marker was performed in a blinded fashion by counting positive fl fl cells per total cells (identified as 49,6-diamidino-2-phenylindole+nuclei We thank Dr. Xin-Yuan Fu for providing Stat3 / mice. We also thank for immunofluorescence staining) in 10 randomly chosen 3400 fields Dr. Felix Knauf for technical assistance with Ussing chamber mea- from the outer medulla. Images were taken at 3400 using a Nikon surements of transepithelial membrane potential in our in vitro tu- microscopy system. bular cell culture model.

J Am Soc Nephrol 26: 1334–1345, 2015 GM-CSF Activates Kidney Macrophages 1343 BASIC RESEARCH www.jasn.org

This work was supported by American Heart Association Grant 18. Clausen BE, Burkhardt C, Reith W, Renkawitz R, Förster I: Conditional 13F-TF17070000 (to S.C.H.), National Institutes of Health Grant gene targeting in macrophages and granulocytes using LysMcre mice. – R01-DK93771 (to L.G.C.), and George M. O’Brien Kidney Center at Transgenic Res 8: 265 277, 1999 19. Welte T, Zhang SSM, Wang T, Zhang Z, Hesslein DGT, Yin Z, Kano A, Yale Grants P30-DK079310, T32-DK007276, and U24-DK059635. Iwamoto Y, Li E, Craft JE, Bothwell ALM, Fikrig E, Koni PA, Flavell RA, Fu X-Y: STAT3 deletion during hematopoiesis causes Crohn’s disease-like pathogenesis and lethality: A critical role of STAT3 in innate immunity. Proc Natl Acad Sci U S A 100: 1879–1884, 2003 DISCLOSURES 20. Müller J, Sperl B, Reindl W, Kiessling A, Berg T: Discovery of chromone- None. based inhibitors of the transcription factor STAT5. ChemBioChem 9: 723–727, 2008 21. Mui AL, Wakao H, Harada N, O’Farrell AM, Miyajima A: Interleukin-3, granulocyte-macrophage colony-stimulating factor, and interleukin-5 trans- REFERENCES duce signals through two forms of STAT5. J Leukoc Biol 57: 799–803, 1995 22. Lehtonen A, Matikainen S, Miettinen M, Julkunen I: Granulocyte- 1. Mantovani A, Sica A, Sozzani S, Allavena P, Vecchi A, Locati M: The macrophage colony-stimulating factor (GM-CSF)-induced STAT5 acti- chemokine system in diverse forms of macrophage activation and po- vation and target-gene expression during human monocyte/macrophage larization. Trends Immunol 25: 677–686, 2004 differentiation. JLeukocBiol71: 511–519, 2002 2. Sica A, Mantovani A: Macrophage plasticity and polarization: In vivo 23. Martinez FO, Gordon S, Locati M, Mantovani A: Transcriptional pro- veritas. J Clin Invest 122: 787–795, 2012 filing of the human monocyte-to-macrophage differentiation and po- 3. Williams TM, Little MH, Ricardo SD: Macrophages in renal de- larization: New molecules and patterns of gene expression. JImmunol velopment, injury, and repair. Semin Nephrol 30: 255–267, 2010 177: 7303–7311, 2006 4. Vernon MA, Mylonas KJ, Hughes J: Macrophages and renal fibrosis. 24. Menke J, Iwata Y, Rabacal WA, Basu R, Yeung YG, Humphreys BD, Semin Nephrol 30: 302–317, 2010 Wada T, Schwarting A, Stanley ER, Kelley VR: CSF-1 signals directly to 5. Day Y-J, Huang L, Ye H, Linden J, Okusa MD: Renal ischemia-reperfusion renal tubular epithelial cells to mediate repair in mice. JClinInvest119: injury and adenosine 2A receptor-mediated tissue protection: Role of 2330–2342, 2009 macrophages. Am J Physiol Renal Physiol 288: F722–F731, 2005 25. Alikhan MA, Jones CV, Williams TM, Beckhouse AG, Fletcher AL, Kett 6. Jo S-K, Sung S-A, Cho W-Y, Go K-J, Kim H-K: Macrophages contribute MM, Sakkal S, Samuel CS, Ramsay RG, Deane JA, Wells CA, Little MH, to the initiation of ischaemic acute renal failure in rats. Nephrol Dial Hume DA, Ricardo SD: Colony-stimulating factor-1 promotes kidney Transplant 21: 1231–1239, 2006 growth and repair via alteration of macrophage responses. Am J Pathol 7. Jang H-S, Kim J, Park Y-K, Park KM: Infiltrated macrophages contribute 179: 1243–1256, 2011 to recovery after ischemic injury but not to ischemic preconditioning in 26. Zhang M-Z, Yao B, Yang S, Jiang L, Wang S, Fan X, Yin H, Wong K, kidneys. Transplantation 85: 447–455, 2008 Miyazawa T, Chen J, Chang I, Singh A, Harris RC: CSF-1 signaling medi- 8. Ko GJ, Boo C-S, Jo S-K, Cho WY, Kim HK: Macrophages contribute to ates recovery from acute kidney injury. JClinInvest122: 4519–4532, 2012 the development of renal fibrosis following ischaemia/reperfusion- 27. Matsumoto A, Masuhara M, Mitsui K, Yokouchi M, Ohtsubo M, Misawa induced acute kidney injury. NephrolDialTransplant23: 842–852, H, Miyajima A, Yoshimura A: CIS, a cytokine inducible SH2 protein, is a 2008 target of the JAK-STAT5 pathway and modulates STAT5 activation. 9. Lin S-L, Li B, Rao S, Yeo E-J, Hudson TE, Nowlin BT, Pei H, Chen L, Blood 89: 3148–3154, 1997 Zheng JJ, Carroll TJ, Pollard JW, McMahon AP, Lang RA, Duffield JS: 28. Metcalf D, Johnson GR, Burgess AW: Direct stimulation by purified Macrophage Wnt7b is critical for kidney repair and regeneration. Proc GM-CSF of the proliferation of multipotential and erythroid precursor Natl Acad Sci U S A 107: 4194–4199, 2010 cells. Blood 55: 138–147, 1980 10. Lee S, Huen S, Nishio H, Nishio S, Lee HK, Choi B-S, Ruhrberg C, 29. Hamilton JA, Anderson GP: GM-CSF biology. Growth Factors 22: 225– Cantley LG: Distinct macrophage phenotypes contribute to kidney 231, 2004 injury and repair. JAmSocNephrol22: 317–326, 2011 30. Fleetwood AJ, Dinh H, Cook AD, Hertzog PJ, Hamilton JA: GM-CSF- and 11. Vinuesa E, Hotter G, Jung M, Herrero-Fresneda I, Torras J, Sola A: M-CSF-dependent macrophage phenotypes display differential de- Macrophage involvement in the kidney repair phase after ischaemia/ pendence on type I interferon signaling. JLeukocBiol86: 411–421, 2009 reperfusion injury. J Pathol 214: 104–113, 2008 31. Krausgruber T, Blazek K, Smallie T, Alzabin S, Lockstone H, Sahgal N, 12. Van Dyken SJ, Locksley RM: Interleukin-4- and interleukin-13-mediated Hussell T, Feldmann M, Udalova IA: IRF5 promotes inflammatory alternatively activated macrophages: Roles in homeostasis and dis- macrophage polarization and TH1-TH17 responses. Nat Immunol 12: ease. Annu Rev Immunol 31: 317–343, 2013 231–238, 2011 13. Lawrence T, Natoli G: Transcriptional regulation of macrophage po- 32. Lacey DC, Achuthan A, Fleetwood AJ, Dinh H, Roiniotis J, Scholz GM, larization: Enabling diversity with identity. Nat Rev Immunol 11: 750– Chang MW, Beckman SK, Cook AD, Hamilton JA: Defining GM-CSF- 761, 2011 and macrophage-CSF-dependent macrophage responses by in vitro 14. Rauch I, Müller M, Decker T: The regulation of inflammation by inter- models. JImmunol188: 5752–5765, 2012 ferons and their STATs. JAK-STAT 2: e23820, 2013 33. Hamilton JA: Colony-stimulating factors in inflammation and autoim- 15. Zhou D, Huang C, Lin Z, Zhan S, Kong L, Fang C, Li J: Macrophage munity. Nat Rev Immunol 8: 533–544, 2008 polarization and function with emphasis on the evolving roles of co- 34. Sielska M, Przanowski P, Wylot B, Gabrusiewicz K, Maleszewska M, ordinated regulation of cellular signaling pathways. Cell Signal 26: Kijewska M, Zawadzka M, Kucharska J, Vinnakota K, Kettenmann H, 192–197, 2014 Kotulska K, Grajkowska W, Kaminska B: Distinct roles of CSF family 16. Lang R, Patel D, Morris JJ, Rutschman RL, Murray PJ: Shaping gene cytokines in macrophage infiltration and activation in glioma pro- expression in activated and resting primary macrophages by IL-10. J gression and injury response. J Pathol 230: 310–321, 2013 Immunol 169: 2253–2263, 2002 35. Stojakovic M, Krzesz R, Wagner AH, Hecker M: CD154-stimulated GM- 17. Qualls JE, Neale G, Smith AM, Koo M-S, DeFreitas AA, Zhang H, Kaplan CSF release by vascular smooth muscle cells elicits monocyte activation G, Watowich SS, Murray PJ: Arginine usage in mycobacteria-infected —role in atherogenesis. J Mol Med (Berl) 85: 1229–1238, 2007 macrophages depends on autocrine-paracrine cytokine signaling. Sci 36. Egea L, McAllister CS, Lakhdari O, Minev I, Shenouda S, Kagnoff MF: Signal 3: ra62, 2010 GM-CSF produced by nonhematopoietic cells is required for early

1344 Journal of the American Society of Nephrology J Am Soc Nephrol 26: 1334–1345, 2015 www.jasn.org BASIC RESEARCH

epithelial cell proliferation and repair of injured colonic mucosa. JIm- 42. Takeda K, Tanaka T, Shi W, Matsumoto M, Minami M, Kashiwamura S, munol 190: 1702–1713, 2013 Nakanishi K, Yoshida N, Kishimoto T, Akira S: Essential role of Stat6 in 37. Cates EC, Fattouh R, Wattie J, Inman MD, Goncharova S, Coyle AJ, IL-4 signalling. Nature 380: 627–630, 1996 Gutierrez-Ramos J-C, Jordana M: Intranasal exposure of mice to house 43. Cui W, Ke JZ, Zhang Q, Ke H-Z, Chalouni C, Vignery A: The intracellular dust mite elicits allergic airway inflammation via a GM-CSF-mediated domain of CD44 promotes the fusion of macrophages. Blood 107: 796– mechanism. J Immunol 173: 6384–6392, 2004 805, 2006 38. Willart MAM, Deswarte K, Pouliot P, Braun H, Beyaert R, Lambrecht BN, 44. Sinha D, Wang Z, Price VR, Schwartz JH, Lieberthal W: Chemical anoxia Hammad H: Interleukin-1a controls allergic sensitization to inhaled of tubular cells induces activation of c-Src and its translocation to the house dust mite via the epithelial release of GM-CSF and IL-33. JExp zonula adherens. Am J Physiol Renal Physiol 284: F488–F497, 2003 Med 209: 1505–1517, 2012 45. Huen SC, Moeckel GW, Cantley LG: Macrophage-specific deletion of 39. Hamilton JA, Achuthan A: Colony stimulating factors and myeloid cell transforming growth factor-b1 does not prevent renal fibrosis after biology in health and disease. Trends Immunol 34: 81–89, 2013 severe ischemia-reperfusion or obstructive injury. Am J Physiol Renal 40. Basile DP, Donohoe D, Roethe K, Osborn JL: Renal ischemic injury Physiol 305: F477–F484, 2013 results in permanent damage to peritubular capillaries and influences long-term function. Am J Physiol Renal Physiol 281: F887–F899, 2001 41. Blazer-Yost BL, Watanabe M, Haverty TP, Ziyadeh FN: Role of insulin and IGF1 receptors in proliferation of cultured renal proximal tubule This article contains supplemental material online at http://jasn.asnjournals. cells. Biochim Biophys Acta 1133: 329–335, 1992 org/lookup/suppl/doi:10.1681/ASN.2014060612/-/DCSupplemental.

J Am Soc Nephrol 26: 1334–1345, 2015 GM-CSF Activates Kidney Macrophages 1345 Supplemental Figure 1.

A Arg1 Cd206

105 50 4 10 40 103 30 102 20 101 100 10 Fold Expression Fold Expression 10-1 0 Veh IL-4 Veh IL-4 WT B Stat6-/- Arg1 Cd206

106 60 105 104 40 103 102 20 101 Fold Expression Fold Expression 100 0 M1/veh M1/MPT M1/veh M1/MPT

Supplemental Figure 1. Tubular mediated macrophage alternative activation occurs through a STAT6 independent pathway. (A) BMM from WT and Stat6-/- mice were treated with vehicle or IL-4 (20 ng/mL) for 24 hours. qPCR analysis of macrophage RNA shown relative to naïve BMM, normalized to Hprt. (B) BMM were first primed with IFNγ (100 units/ml) for 24 hours (M1), then cultured with vehicle (M1/veh) or co-cultured with MPT cells (M1/MPT) grown in Transwells. Cells harvested at 48 hours after co-culture. qPCR analysis shown relative to M1, normalized to Hprt. n=3, Data shown as mean ± SEM.

Supplemental Figure 2.

Lrp2 (Megalin) Pck1 (Pepck) Slc12a1 (Nkcc2) Aqp2

4 *** 15 ** 5 ** 0.10 ***

4 0.08 3 10 Hprt Hprt Hprt Hprt 3 0.06 2 2 0.04 5 1 Rel Exp/ Rel Exp/ Rel Exp/ Rel Exp/ 1 0.02

0 0 0 0.00 LTL+ Whole Kidney LTL+ Whole Kidney LTL+ Whole Kidney LTL+ Whole Kidney

Supplemental Figure 2. Specificity of Lotus Tetragonolobus Lectin (LTL) cell sorting for proximal tubule cells. After harvest and digestion with collagenase and DNase, kidney single cell suspensions were stained with fluorescein labeled LTL. Total RNA was isolated from whole kidney and cell-sorted LTL positive kidney cells. Quantitative RT-PCR was performed for proximal tubule segment markers (Lrp2, Pck1), thick ascending loop (Slc12a1), and collecting duct (Aqp2). mRNA expression shown relative Hprt. Data expressed as mean ± SEM. **p<0.01, ***p<0.001. n=4.

Supplemental Figure 3.

A Proximal Tubule Cells B Proximal Tubule Cells Day 3 after IR In vitro

0.06 ** 0.5 0.04 0.4

Hprt 0.02 Uninjured Hprt 0.3 0.006 0.2 IR Day 3 0.010 0.004 *** 0.005

Relative Exp/ 0.002 Relative Exp/

0.000 0.000 Csf1 Csf2 Csf1 Csf2

C pSTAT5 ** 1.5 ** * * Veh CM M GM M+GM 1.0 * * pSTAT5 ACTB 0.5 Arbitrary Units

0.0 Veh CM M GM M+GM

D Arg1 Cd206 Msr1 ** 3 8 **** * ** ** ** 1.0 * 6 ****

Hprt Hprt Hprt *** ** 2 *** * 4 0.5 1 2 Relative Exp/ Relative Exp/ Relative Exp/

0.0 0 0 Veh CM M GM M+GM Veh CM M GM M+GM Veh CM M GM M+GM

Supplemental Figure 3. Additional factors such as M-CSF are required for maximal expression of tubule cell-mediated macrophage alternative activation. (A) Real-time quantitative PCR for Csf1 and Csf2 using mRNA isolated from proximal tubule cells, expression shown relative to Hprt from uninjured sham kidneys and kidneys 3 days after I/R. n=4. (B) mRNA expression levels of Csf1 and Csf2 in MPT cells grown to confluence, mRNA expression shown relative to Hprt. n=3 (C-D) BMM were treated with MPT CM (CM), 25 ng/ml mouse recombinant M-CSF (M), 25 ng/ml mouse recombinant GM-CSF (GM) or both M-CSF and GM- CSF both at 25 ng/ml (M+GM) for (C) 30 minutes for protein, densitometry shown relative to β- actin, and (D) 12 hours for mRNA, mRNA expression shown relative to Hprt. n=3. Data shown as mean ± SEM. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001