Macrophage Wnt7b is critical for kidney repairand regenerationShuei-Liong Lina,1,b, Bing Lia,1,2, Sujata Raoc,d, Eun-Jin Yeoc,d, Thomas E. Hudsona, Brian T. Nowlina, Huaying Peia,Lijun Chene, Jie J. Zhenge, Thomas J. Carrollf, Jeffrey W. Pollardg, Andrew P. McMahonh, Richard A. Langc,d,3,and Jeremy S. Duffielda,3

aLaboratory of Inflammation Research, Renal Division, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA02115; bDepartment of Medicine, National Taiwan University Hospital, Tipei 100, Taiwan; cVisual Systems Group, Divisions of Pediatric Ophthalmology andDevelopmental Biology, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH 45229; dDepartment of Ophthalmology, University of Cincinnati,Cincinnati, OH 45229; eDepartment of Structural Biology, St. Jude Children’s Research Hospital, Memphis, TN 38105; fDepartment of Internal Medicine,Department of Molecular Biology, University of Texas Southwestern Medical Center, Dallas, TX 75390; gDepartment of Molecular and Developmental Biology,Albert Einstein College of Medicine of Yeshiva University, Bronx, NY 10461; and hDepartment of Molecular and Cellular Biology, Harvard University,Cambridge, MA 02138

Edited* by Jeremy Nathans, Johns Hopkins University, Baltimore, MD, and approved January 7, 2010 (received for review October 23, 2009)

Macrophages are required for tissue homeostasis through theirrole in regulation of the immune response and the resolution ofinjury. Here we show, using the kidney as a model, that the Wntpathway ligand Wnt7b is produced by macrophages to stimulaterepair and regeneration. When macrophages are inducibly ablatedfrom the injured kidney, the canonical Wnt pathway response inkidney epithelial cells is reduced. Furthermore, when Wnt7b issomatically deleted in macrophages, repair of injury is greatlydiminished. Finally, injection of the Wnt pathway regulator Dkk2enhances the repair process and suggests a therapeutic option.Because Wnt7b is known to stimulate epithelial responses duringkidney development, these findings suggest that macrophages areable to rapidly invade an injured tissue and reestablish a devel-opmental program that is beneficial for repair and regeneration.

macrophage | repair | regeneration | kidney | canonical Wnt pathway

Prompted by studies that identified a pivotal role for macro-phages in injury repair (1–6) andwork showing thatmacrophage

Wnt ligands establish tissue homeostasis during development (7),we determined whether the canonical Wnt pathway was activatedduring kidney injury and played an active role in repair and regen-eration. Mice subjected to kidney ischemia reperfusion injury (I/R)lose epithelial cells from the proximal tubules in the cortex andouter medulla. Following injury there is a process of repair andregeneration that is well-defined temporally where regeneration ofepithelial tubules occurs (8, 9). The canonical Wnt pathway isknown to regulate scheduled cell proliferationanddeath events, celldifferentiation, and cell-fate decisions during development andunscheduled events in the initiation and perpetuation of neoplasia(10). More recently, evidence that stem cells in bone marrow andskin are regulated byWnt canonical pathway signaling suggested tous that the Wnt pathway may play an important role in tissueregeneration (11, 12).In the following studies, we characterize reactivation of theWnt

pathway particularly in epithelial cells of the kidney followinginjury and show a central role for macrophages as a recruited andimportant source of Wnt ligands in the regenerative process.

Results and DiscussionWnt Pathway Responses Are Induced in the Kidney Following Injury.Kidney injury can be assessed quantitatively using an injury scorethat depends on multiple criteria including the deposition ofnecrotic material within the kidney tubules (Fig. 1F, nec) and theappearance of epithelial cells of flattened morphology (Fig. 1F,arrowheads) that is distinct from the cuboidal morphology of epi-thelial cells in the uninjured kidney (Fig. 1G, arrowheads). Thekidney can be assessed functionally bymeasuring plasmacreatinine,with high levels indicating compromise (8, 9). According to injuryscores (8) and creatinine assays (Fig. S1A), both of which are

improvingby48h, thephaseofkidney repairbeginsonday2.Repairis associated with increased levels of epithelial proliferation,regeneration of epithelial tubules, mild expansion of interstitialmyofibroblasts, and transient deposition of interstitial collagens (8,9). Repair is also temporally associated with recruitment of largenumbers of macrophages (Fig. S1B).We induced kidney injury in BATgal (13) and Axin2-lacz Wnt

pathway reporter mice (14) and noted an injury-induced enhance-ment of the Wnt pathway response. Control, BATgal-negative micedid not show any staining 5 days after injury (Fig. 1A). Uninjuredkidney fromstrain-matched,BATgalmice showedX-gal staining in aproportion of interstitial cells and tubule cells prominent in thepapilla (Fig. 1B), but the cortex and medulla showed almost nostaining. By contrast, 5 days following injury there was marked up-regulation of X-gal staining in both papilla and cortex (Fig. 1C).Similar injury-induced enhancement of the Wnt pathway responsewas noted in Axin2-lacz reporter mice (Fig. S2A) and by detectingphosphorylation of the canonical pathway receptor Lrp6 (Fig. S2B)(15). Immunolabeling of kidney sections from injured BATgalmicerevealed that Lacz was detected in kidney epithelial cells (whichdouble-labeled with lotus lectin; Fig. 1D) but was not detected inmacrophages (Fig. 1E) (SI Materials and Methods). In Axin2+/LacZ

reporter mice, a similar pattern of Wnt pathway responses wasobserved.Wild-type mice (Fig. 1F) and uninjured Axin2+/LacZmice(Fig. 1G) did not show any labeling, whereas Axin2+/LacZ reportermice showed labeling in epithelial cells (Fig. 1H, arrowheads) andinterstitial cells after injury. Double labeling of histological sectionsof injured kidney from Axin2+/LacZ mice for β-galactosidase andfibroblasts (α-SMA; Fig. 1I) or macrophages (F4/80; Fig. 1J) con-firmed that interstitial cells and epithelial cells but not macrophagesshowed a Wnt pathway response. The day 5–7 postinjury enhance-ment of theWnt response suggested that theWnt pathway could bea component of the injury repair mechanism.To determine which Wnt pathway components might be impor-

tant in the injury response, we assessed control and injured whole

Author contributions: S.-L.L., R.A.L., and J.S.D. designed research; S.-L.L., B.L., S.R., E.-J.Y.,T.E.H., B.T.N., H.P., and J.S.D. performed research; S.-L.L., L.C., J.J.Z., T.J.C., J.W.P., A.P.M.,R.A.L., and J.S.D. contributed new reagents/analytic tools; S.-L.L., B.L., R.A.L., and J.S.D.analyzed data; and A.P.M., R.A.L., and J.S.D. wrote the paper.

Conflict of interest statement: R.A.L. and J.S.D. have submitted a patent application forthe use of Dkk2 and other Wnt agonists in regeneration.

*This Direct Submission article had a prearranged editor.1S.-L.L. and B.L. contributed equally to this work.2Present address: Nephrology Department, 2nd Teaching Hospital of Harbin Medical Uni-versity, 345 Xuefu Road, Nangang District, Harbin 150086, China.

3Towhomcorrespondencemaybeaddressed. E-mail: richard.lang@cchmc.orgor jduffield@rics.bwh.harvard.edu.

This article contains supporting information online at www.pnas.org/cgi/content/full/0912228107/DCSupplemental.

4194–4199 | PNAS | March 2, 2010 | vol. 107 | no. 9 www.pnas.org/cgi/doi/10.1073/pnas.0912228107

Dow

nloa

ded

by g

uest

on

Janu

ary

23, 2

020

kidney for expression of mRNAs encoding ligands (Wnts), recep-tors (Fzds), and coreceptors (Lrp5 and 6). This showed that at themidpoint of the repair phase a number of ligands (Wnt2, 2b, 4, 5a,7b, 10a) were up-regulated (Fig. 1K). Although expression levels ofreceptors and coreceptors remained unchanged with injury (Fig.1K),Fzd4,Lrp5, andLrp6were prominently expressed (Fig. 1K). Tofurther define the cell types responsible for expressing these ligandsand receptors, we isolated kidney macrophages and proximal epi-thelial tubule cells (PTECs) using cell-sorting methods (16) and

performed the same expression analysis. Given the paucity ofmacrophages in healthy kidney, macrophages from the injuredkidney were compared with autologous peripheral blood mono-cytes. Because Wnt reporter mice indicated that macrophages didnot respond to the canonical Wnt pathway, we assessed only theexpression of ligand genes. This showed thatWnt4, 7b, 10a, and 10bwere up-regulated (Fig. 1K). Wnt pathway reporter mice alsoindicated that PTECs were responding to the Wnt pathway duringinjury, and so in this purified population we assessed the expressionof receptors.This showed thatFzd3, 4, and 7were expressed at goodlevels in both control and injured kidney (Fig. 1K). Similarly, Lrp5and 6 were expressed, but with little change in level in the injuredtissue (Fig. 1K). These expression data are consistent with reportermouse assessment, suggesting that up-regulation of the canonicalWntpathway is an injury-associated response and thatmacrophagesmay be a source of Wnt ligands to which epithelial cells respond.

Inflammatory Macrophages Promote Kidney Repair and ShowEnhanced Wnt Signaling Activity. To test whether macrophagesfrom the injured kidney showed enhancedWnt signaling activity, weisolated macrophages 5 days after injury and cocultured them withthe canonical Wnt reporter cell line superTOPFLASH (STF).Compared with control autologous monocytes (Fig. 2A, PBM) orbone marrow macrophages (Fig S3A), postinjury macrophagesinduced a 3-fold increase in canonical Wnt signaling via Lrp5 withFzd3, 4, or 7, and a 5- to 6-fold increase via Lrp6 with Fzd3, 4, or 7(Fig. 2B) (SIMaterials andMethods).Despite theexpressionofWnt2,2b, 5a, and 5b in monocytes andWnt2, 3, 5a, and 5b by unactivatedcultured bone marrow (BM) macrophages, no signaling was identi-fied from these leukocytes to STF cells (Fig. S3 A and B), indicatingthat macrophage expression of Wnt2, 2a, 5a, and 5b does not nec-essarily result in paracrine signaling. Addition of the Wnt pathwayinhibitor Dkk1 in recombinant form (Fig. 2C) or cotransfection ofDkk1 in STF reporter cells significantly suppressed kidney macro-phage-stimulated Wnt signaling. The Fzd, Lrp5/6, and Dkk1dependence of the STF response tomacrophages demonstrates thatthis is a canonical Wnt pathway response. Furthermore, these datashow that the Wnt receptors and coreceptors expressed by kidneyepithelium canmediate responses to theWnt activities produced bymacrophages from the injured kidney.Our studies have consistently identifiedWnt7b as amacrophage-

expressed ligand (Fig. 1K and Fig. S3 D and E) (7). To assesswhether this ligandwas capable of inducing a response via Fzd4 and7 with Lrp5 or 6, we stably expressed Wnt7b by retroviral trans-duction in a cell line and also primary, bone-marrow-derived mac-rophages.Weused thesemacrophages for STF cell coculture at day7 after marrow harvest when they do not normally express endog-enous Wnt7b. Wnt7b-expressing BM macrophages induced a Wntpathway response with all receptor/coreceptor combinations, withFzd4/Lrp6 showing the greatest response (Fig. 2D and Fig. S3C). Aswith freshly isolated macrophages from the injured kidney, Dkk1could suppress the signaling (Fig. 2D). These data further supportthe notion that macrophages could be a source of Wnt ligandactivity in the injured kidney. Furthermore, activation of BMmacrophages cultured on glass with (Kdo)2-lipidA, a selective Toll-like receptor 4 agonist, led to robust induction ofWnt7b (Fig. S3Dand E), confirming that Wnt7b is an inducible ligand of theactivated macrophage.As described above, after injury to the kidney there is a defined

phase of repair that begins on day 2 (8, 9, 17). Because the repairphase correlates withmacrophage numbers (Fig. S1), we reasonedthat that macrophageWnt ligands might be involved in repair andtested this by performing macrophage ablation experiments invivo. To test the possibility that kidney macrophages stimulatedWnt pathway responses during injury, we combined the CD11b-DTR allele that allows conditional ablation of macrophages invivo (2) with the Wnt reporter Axin2LacZ and assessed reporterexpression after injury with or without macrophage ablation. We

Fig. 1. Injured proximal tubule epithelial cells are Wnt-pathway-responsive,and inflammatory macrophages are a source of Wnt ligands in the kidneyduring repair following injury. (A–C) Photomicrographs of whole-mount X-gal-stained BATgal kidneys indicating Wnt pathway responses. (D and E)Sections from day 5 injured kidneys from BATgal mice labeled for nuclei(blue), Lacz (green), the proximal epithelial tubule marker LTL (D), or themacrophage marker F4/80 (E). Epithelial cells but not macrophages are Lacz-positive. (F–H) X-gal staining for Wnt signaling activity in Axin2+/LacZ kidneys.In F, arrowheads indicate epithelial cells of flattened morphology typical ofsevere injury that contrasts with cuboidal morphology shown by arrowheadsin uninjured kidney (G). Injured tubules contain necrotic debris (F, nec),whereas uninjured tubules do not (G, asterisks). X-gal staining is prominentin epithelial cells of Axin2+/LacZ kidneys (H and I, arrowheads). (I and J) Sec-tions from day 5 injured kidneys from Axin2+/LacZ mice labeled for nuclei(blue), Lacz (green), the fibroblast marker α-SMA (I), or the macrophagemarker F4/80 (J). In I, Lacz-positive fibroblasts are seen (arrows). (K) Semi-quantitative RT-PCR for transcript levels of Wnt-signaling-pathway ligandsand receptors in whole kidney or purified cell types (C, normal control; I, d5postinjury; P, peripheral blood monocytes). Red arrowheads indicate up-regulated transcripts. All studies were repeated at least three times and gavecomparable results. *P < 0.05. (Scale bars, 50 μm.)

Lin et al. PNAS | March 2, 2010 | vol. 107 | no. 9 | 4195

CELL

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Janu

ary

23, 2

020

injected diphtheria toxin (DT) to ablate macrophages on days 3–6after injury, as this corresponds to the repair phase. We first con-firmed macrophage ablation by histological assessment of mac-rophage numbers usingF4/80 labeling (Fig. 2E–H) and performedmorphometric measurements of the F4/80-positive area inexperimental and control mice (Fig. 2M). This showed, as expec-ted (18), that at least 80% of macrophages in the injured kidneycould be ablated by DT injection (Fig. 2H andM). Morphometricmeasurements showed that with or without DT, uninjured kidneycortex and medulla had limited X-gal-positive area (Fig. 2N,CON) and that, as expected, injury increased the X-gal-positivearea dramatically (Fig. S2; Fig. 2N, I/R, white bar), similar tofindings in BATgalmice (Fig. 1C). Importantly, the application ofDT and the accompanying macrophage ablation resulted in amarked reduction of X-gal staining in all areas of medulla andcortex ofCD11b-DTR;Axin2LacZ kidneys (Fig. 2N, I/R, black bar).These data suggested that macrophages were a major source ofstimulus of Wnt responses in kidney epithelium during injury. Inaddition to reducing expression of the Wnt reporter, macrophageablation resulted in a striking failure of normal regeneration ofkidney tubule epithelium (Fig. 2 I–L and O and Fig. S4), as indi-

cated by an injury score that was approximately doubled in theabsence of macrophages (Fig. 2O). Furthermore, when macro-phages were ablated there was failure of normal functionalrecovery of the kidneys asmeasured byplasma creatinine (Fig. 2P).

Genetic Disruption of Wnt Responses in the Kidney Prevents NormalRepair. Proximal tubule epithelial cells express a limited set of Wntpathway receptors includingFzd4 (Fig. 1).Toconfirm theexpressionpattern ofFzd4, we stained kidney sections fromFzd4+/laczmicewithX-gal (Fig. 3A andB) and with antibodies to F4/80 (Fig. 3C). Thesedata confirmed thatFzd4 expression was restricted to epithelial cellsand not macrophages. To test whether Fzd4 was required for theinjury response, we performed kidney ischemia reperfusion injury inFzd4LacZ/LacZmutantmice andcompared the regeneration responsewith Fzd4+/+ littermate controls. Fzd4LacZ/LacZ mice show reducedgrowth and have abnormalities of capillary development in the eyeand cerebellum (19). However, their kidneys are developmentallynormal. In response to kidney injury, Fzd4LacZ/LacZ mice showed anormal influxofmyofibroblasts (Fig. 3H) andmacrophages (Fig. 3I).Furthermore, therewas no significant difference in the proliferationresponse as determined by the pan cell-cycle marker Ki67 (Fig. 3Jand Fig. S5). By contrast, Fzd4lacZ/lacZ mice showed a modest butstatistically significant persistent injury according to periodic acidSchiff (PAS) staining (Fig. 3 D–G and K and Fig. S5), and this wasaccompanied by increased epithelial cell apoptosis (Fig. 3L and Fig.S5). These data indicate that Fzd4-dependent responses in kidneyepithelial cells are required for repair of injury.With robust Lrp5 and Lrp6 expression in proximal tubule epi-

thelial cells, we reasoned that mutation of these coreceptors might,like Fzd4 mutation, compromise kidney repair and regeneration.Homozygosity for bothLrp5 andLrp6nullmutations results in earlyembryonic lethality, but double heterozygote mice survive toadulthood (20–22). When analyzed after kidney injury, Lrp5+/LacZ;Lrp6+/LacZmice exhibited increased tubule injury early in the repairprocess at day 2 (Fig. 3M–P and S; Fig. S6). Furthermore, there wasa persistence of epithelial injury after 7 days (Fig. 3Q,R, andT; Fig.S6). According to F4/80 labeling, there was no difference in mac-rophage recruitment observed at day 7 (Fig. 3T) and no change inthe total number of kidney cells in the cell cycle (Ki67-positive) atday 2 (Fig. 3S) or 7 (Fig. 3T and Fig. S6). However, we did quantifyan increase in apoptotic tubule cells at day 7 (Fig. 3T and Fig. S6)and an increased number of interstitial myofibroblasts (Fig. 3T).These changes are consistent with those observed when macro-phages are ablated or when Fzd4 is mutated (above). These datastrengthen the case for canonical Wnt-pathway-dependent repairand regeneration in kidney epithelium.

Macrophage Wnt7b Promotes Regeneration by Directing EpithelialCell-Cycle Progression and Basement Membrane Repair. Toassess therole of Wnt7b in macrophages specifically in vivo, we used theWnt7bC3 loxP conditional allele (Fig. 4A) in combination with theleukocyte-specific Csf1R-icre transgene (Fig. 4A) (23), and con-firmed exon 3 deletion specifically in macrophages (Fig. 4B). Thefact that no C3 allele was detected by PCR of genomic DNA frommacrophages indicates that in the experimental mice there wascomplete deletion of Wnt7b. Injury of the kidneys was performedin these mice and control mice and repair was assessed at day 7.Strikingly, whereas repair of the injured epithelium proceeded asexpected in littermate controlWnt7bC3/− andCsf1R-icre;Wnt7bC3/+ mice, repair was substantially retarded in Csf1R-icre;Wnt7bC3/−

experimental mice (Fig. 4C–F and Fig. S7). This was evident fromepithelial injury scores, assessed by blinded morphometry, thatshowed severe persistent epithelial injury in the homozygoussomatic mutant (Fig. 4F). In addition, the PTEC-specific markerexpressed only by injured, dedifferentiated epithelial cells, kidneyinjury molecule-1 (Kim1) (24), was highly expressed in kidneycortex from experimental mice lacking Wnt7b in macrophages 7days after injury, whereas in control mice the level of Kim1 was

Fig. 2. Macrophages are a source of Wnts and mediate Wnt responses inregeneration of the kidney epithelium. (A–D) Relative luciferase activity(RLA) from STF cells expressing Lrp5 or Lrp6 with Fzd3, Fzd4, or Fzd7,induced by coculture with: (A–C) d5 post-I/R kidney macrophages (I/R Mϕ)compared with autologous peripheral blood monocytes (PBM) and inhibited(C) by recombinant Dkk1; or (D) coculture with bone marrow macrophages(Mϕ-CON) or Wnt7b-expressing bone marrow macrophages and inhibited bythe addition of Dkk1-expressing 293T cells. (E–L) PAS-stained kidney sectionsand F4/80 immunofluorescence confocal images of kidney sections frommice with and without conditional macrophage ablation during recoveryfrom injury (asterisks, regenerating tubules; arrowheads, injured flattenedepithelia; nec, necrotic debris). (M–O) Quantification of macrophages, activeWnt signaling (X-gal staining) in kidney cortex and medulla, and tubuleinjury d6 after injury. (P) Kidney function testing (plasma creatinine levels) incohorts of mice (n = 6/group) with or without macrophage ablation from d3to d6. Normal recovery is prevented by ablation. P < 0.05. n = 5 or 6/group.(Scale bars, 50 μm.)

4196 | www.pnas.org/cgi/doi/10.1073/pnas.0912228107 Lin et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

23, 2

020

lower (SI Materials and Methods and Fig. S7). Macrophagerecruitment to the kidneys (Fig. 4G) and the proportion of epi-thelial cells in the cell cycle (Fig. 4H) were not significantlychanged, but there were significant increases in interstitial fibrosisaccording to collagen labeling (Fig. 4I) and in apoptotic epithelialcells (Fig. 4J). These data provide strong evidence that the mac-rophage is a source of Wnt7b that is required for repair andregeneration in the injured kidney.To understand the mechanisms of Wnt7b-mediated repair,

tubule basement membrane (TBM) integrity was scored blindlyfor dissolution (SI Materials and Methods and Fig. 4 K and L).Strikingly Csf1R-icre; Wnt7bC3/− experimental mice had increases

in TBM dissolution, indicating that Wnt7b responses in epithelialcells promote TBM repair. To explore the function of Wnt7b inepithelial cell-cycle progression, G1/S phase in epithelial cells wasidentified by BrdU uptake and G2/M by phosphorylation of His-tone-H3 (pHH3) (SI Materials and Methods and Fig. 4M–O) (25).Although entry into cell cycle, detected by BrdU, was unaffectedby loss of macrophage Wnt7b in Csf1R-icre; Wnt7bC3/C3 exper-imental mice, progression through G2 was suppressed becausethere was a doubling of epithelial cells expressing pHistone-H3 inthe absence of mitosis in experimental mice. This suggests thatWnt7b responses drive epithelial cells through a G2 checkpoint,thereby avoiding apoptotic cell death. Wnt7b responses in epi-thelial cells trigger both basement membrane regeneration andrepopulation of the tubule by overcoming a G2 arrest in the cellcycle. Extracellular matrix genes such as fibronectin and laminin,and cell-cycle progression genes such as n-myc and c-jun, havebeen reported to be directly regulated by the Wnt pathway, sug-gesting that the mechanisms of epithelial repair described heremay be a direct consequence of epithelial Wnt responses (26).

Dickkopf-2 Promotes Kidney Epithelial Repair. The data presentedthus far argue that enhancement of Wnt signaling might providesome therapeutic benefit for damaged kidneys by accelerating theregeneration of tubule epithelium.TheDickkopf family of proteinsare solubleWntmodulators.Dkk2 binds toLrp5 orLrp6 on the cellsurface and can enhance signaling. This enhancement is negativelyregulated by the presence of the transmembrane protein Kremen(27, 28). We cloned, expressed, and purified in quantity the C2cysteine-rich domainofDkk2 that retains the functional activity (SIMaterials and Methods and Fig. 5E). This C2 recombinant form ofDkk2was administered fromday 0 toWTmice that were uninjured(Fig. 5B) or which had kidney injury (Fig. 5D). Kidneys were har-vested onday 2 andday7 after injury and assessedquantitatively forinjury.At both time points, tubule injury scoresweremore severe invehicle-treated mice (Fig. 5 F and G and Fig. S8). Dkk2 admin-istration (3 nmol/g body weight or 210 ng/g body weight) had noeffect on macrophage recruitment at day 7 post-I/R (Fig. 5G), butthere were fewer TUNEL-positive apoptotic tubule epithelial cells(Fig. 5G and Fig. S8) with no significant change in epithelial cells inthe cell cycle (Fig. 5G and Fig. S8). Mice with sham surgeryremainedwell with normal bodyweight and activity, suggesting thatsystemic Dkk2 had no systemic deleterious effects. When systemicdelivery of Dkk2 was delayed until after peak injury at day 1, it wasnevertheless detectable in the kidney (Fig. S8D) and retained thecapacity to improve kidney function as assessed by plasma crea-tinine levels (Fig. 5H). This underscores a primary role for Dkk2 inrepair of the kidney. Dkk2 can either augment or inhibit Wntresponses, depending on the expression of Kremen proteins. Tounderstand whether Dkk2 was functioning to augment the canon-ical pathway, Wnt signaling in kidney cortex of day 5 post-I/Rmicewas assessed by detection of phosphorylated Lrp6 because can-onical signaling requires phosphorylation of this coreceptor (15)(Fig. 5I). pLrp6 was not detected in healthy adult kidney cortex(Fig. S2) butwas robustly detected inkidney cortex following injury.TherapywithDkk2-C2 resulted in enhancedpLrp6detection in thekidney cortex at day 5 post-I/R (Fig. 5I). Therefore,Dkk2 functionsto enhance canonicalWnt responses in regenerating kidney cortex.Because Dkk2 function depends on the expression of Kremenproteins, transcripts of Kremen1 and 2 were assessed and found tobe expressed at very low levels by the Wnt receptive kidney epi-thelial cells compared with macrophages (Fig. 5J). Collectively,these observations are consistent with a major role for Dkk2-C2 asan enhancer of endogenous Wnt responses.In summary, our data indicate that (i) kidney injury results in an

up-regulation of Wnt ligands in macrophages and the canonicalWnt response in epithelial cells, (ii) ex vivo, macrophages from theinjured kidney are a source of increased Wnt activity, (iii) macro-phage ablationduring repair of the injuredkidney results in reduced

Fig. 3. Mutation of Frizzled4 or coreceptors Lrp5 and Lrp6 prevents normalrepair and regeneration of the kidney following ischemia reperfusion injury. (Aand B) Lacz staining in d5 post-IRI kidney showing restriction of Fzd4 expressionto epithelial tubules and (C) low- (top left) and high-power views (split colorpanels) showing Fzd4 receptor expression is not present in F4/80+ macrophagesinpostinjury kidneys. (D–G) PAS-stainedsectionsofnormal (DandF) andd5post-IRI kidneys (E andG) showing persistenceof epithelial injury inmice lacking Fzd4(Fzd4LacZ/LacZ). (H–L) Quantification of inflammation, injury, and repair parame-ters5dpost-kidney-IRI injury inFzd4LacZ/LacZor littermatecontrolmice. (M–R) PAS-stained sections of kidneys. (S and T) Quantification of inflammation, injury, andrepair parameters inWTor Lrp5+/lacz; Lrp6+/lacz kidneys. P< 0.05.n= 5or 6/group.(Scale bars, 50 μm.) Asterisks, regenerating tubules; arrowheads, injured flat-tened epithelia; nec, necrotic debris.

Lin et al. PNAS | March 2, 2010 | vol. 107 | no. 9 | 4197

CELL

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Janu

ary

23, 2

020

canonical Wnt response in kidney epithelial cells, (iv) compromiseof Wnt receptors or conditional deletion of Wnt7b in the macro-phage lineage results in a reduction of the repair response andpersistent injury, and (v) macrophage Wnt7b is required for repairof the kidney tubule basal lamina and relief of a G2 arrest in kidney

epithelial cells. Combined, these outcomes lead to amodel inwhichrepair of damage to kidney tubules is mediated by an influx ofmacrophages that produce Wnt7b and signal locally to remainingkidney epithelial progenitors.These studies show a functional role for canonicalWnt pathway

signaling in any process of solid organ repair following injury.Macrophages are known as critical mediators of the inflammatoryprocess that leads to repair and can produce a number of factorsimplicated in repair including bFGF, IGF, HGF, and IL-10 (29).The involvement ofWnt ligands inmacrophage-mediated repair islogically appealing for several reasons. First, Wnt7b is known tohave an important role in the formation of kidney tubules. Itsmainfunction is to stimulate the polarized cell division that leads toelongation of the tubules as they form (30). This leads directly tothe suggestion that by migrating into the injured kidney and pro-ducing Wnt7b, macrophages are reestablishing a developmentalprogram that is beneficial. Second, it has been established in anumber of systems that the Wnt pathway has a function in pro-moting the renewal of stem or progenitor cells (31), and this is

Fig. 4. Somatic mutation of Wnt7b in macrophages prevents normal repairand regenerationof thekidney following ischemia reperfusion injury. (AandB)Genomic map and PCR products (tail or macrophage genomic DNA) showingthe third exonofWnt7b is deleted inmonocytes/macrophagesofmicewith theLoxP-flanked conditional allele Wnt7bC3 and the transgene Csf1r-icre. WTallele, 153 bp; C3 allele, 200 bp; ΔC3 allele, no product. (C–E) PAS-stained sec-tions of the outer cortex of Csf1R-icre; Wnt7bC3/− kidneys or control kidneys(Csf1R-icre;Wnt7bC3/+ andWnt7bC3/−) 7d following injury. (F–J) Quantificationof inflammation, injury, and repair parameters 7d postinjury of Csf1R-icre;Wnt7bC3/− or control mice (Csf1R-icre; Wnt7bC3/+ and Wnt7bC3/−). (K and L)Immunofluorescence images of kidneys showing dissolution of epithelialbasementmembrane (arrowheads) andquantificationof dissolution in controland experimental (Csf1R-icre; Wnt7bC3/−) mice. (M–O) Immunofluorescenceimages of experimental postinjury kidneys showing the presence ofmarkers ofthe cell cycle in epithelial cells and quantificationof epithelial cells enteringG1/S or inG2Mphase of the cell cycle. P< 0.05.n= 5 or 6/group. (Scale bars, 50 μm.)

Fig. 5. Dickkopf-2 promotes repair in the post-ischemia reperfusion injury kid-ney. (A–D) PAS-stained sections of kidneys. (E) Coomassie-stainedpolyacrylamidegel showingpurifiedDkk2-C2 (D2C2). (F andG) Graphs showingquantified injuryandrepairparameters inkidneys treatedwithvehicleorDkk2. (H)Graphshowingplasma creatinine levels in mice treated with Dkk2 from d1 post-I/R onward. (I)Western blot (upper) for pLrp6 (210 kDa) and loading control (β-actin 45 kDa) inkidney cortex from mice on d5 post-I/R treated with Dkk2-C2 or vehicle. (J) RT-PCRs for Kremen-1 and Kremen-2 in purified d5 post-I/R kidney macrophages(Mϕ) (I) compared with autologous peripheral blood monocytes (P) (upper) andd5post-IRI proximal tubuleepithelial cells (PTEC) (I) or epithelial cells fromcontrolkidneys (C). *P < 0.05. n = 6/group. (Scale bars, 50 μm.)

4198 | www.pnas.org/cgi/doi/10.1073/pnas.0912228107 Lin et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

23, 2

020

consistent with the process of repair where we have shown it islikely that a relatively small population of progenitors expands toprovide replacement tubule epithelial cells (9). The finding of animportant role for macrophages in the resolution of injury isentirely consistent with the Metchnikovian view in which macro-phages were assigned the function of “organismal policemen”withthe role of restoring order from chaos (32). Further, the universalinvolvement of macrophages in repair suggests that Wnt ligandsmay play roles in repair in other organs.Our findings also point to a potential therapy for examples of

repair, like the kidney, that involve theWnt pathway. As describedhere, injection of the Wnt pathway agonist Dkk2-C2 resulted inenhanced repair. The lack of any apparent systemic effects of thistreatment are probably explained by the restriction of Wnt path-way responses to a limited number of normal tissues in the adult.This model therapy deserves further examination for its potentialin the treatment of human disease.

Materials and MethodsMouse Breeding and Genotyping.Wild-typeC57BL/6mice (male,25g,8–12weeksold) were from Charles River Laboratories. Cd11b-DTR mice (FVB/N) were gen-erated andmaintained as previously described (2). Presence of the transgene wasconfirmed by PCR using the following primers: 5′-TTCCACTGGATCTACGGACC-3′,5′-TGTCGGCCATGATATAGACG-3′. BATgal mice were from Jackson Laboratories(13) and genotyping was performed as described. Axin2+/LacZ mice (C57BL6) werefromtheMaxDelbrückCenter forMolecularMedicine,Berlin,andgenotypingwasperformed with the following primer pairs: Ex2As3 (i) 5′-AGTCCATCTT-CATTCCGCCTAGC-3′, NLSBJ1 (ii) 5′-TGGTAATGCTGCAGTGGCTTG-3′, and CKOIN4(iii) 5′-AAGCTGCGTCGGATACTTGAGA-3′. 1+3WTallele,2+3mutantallele.Cd11b-DTRmice (FVB/N)were crossedwithAxin2+/LacZmice and F1progenywas screenedfor the LacZ allele and CD11b-DTR transgene. Frizzled4+/LacZmicewere generatedandmaintained as previously described (19), and genotypingwas performedwiththe following primer pairs: WT allele; 5′-CACACGTGGCAAAAGTGTTG-3′, 5′-CAGTTGAAATCCCACCCAGT-3′; mutant allele: 5′-TGTCTGCTAGATCAGCCTCT-3′,5′-CATCAACATTAAATGTGAGCGAGT-3′. Lrp5+/LacZ and Lrp6+/LacZ mice (C57BL6)

were generated as previously described (7) and genotyped with the followingprimer sets: Lrp5: (i) 5′-GGCTCGGAGGACAGACCTGAG-3′, (ii) 5′-CTGTCAGTGCC-TGTATCTGTCC-3′, (iii) 5′-TCCAAGCGGCTTCGGCCAG-3′. LRP6: (i) 5′-CAGGCATG-TAGCCCTTGGAG-3′, (ii) 5′-ACTACAAGCCCTGCACTGCC-3′, (iii) 5′-GTAGAGTTC-CCAGGAGGAGCC-3′. Transgenic Csf1R-icre mice (BALB/c background) were gen-erated (23) and genotyping was performed using the following primers: 5′-CTAATCGCCATCTTCCAGCAGG-3′, 5′-GCTAAGTGCCTTCTCTACACCT-3′. Thefloxedconditional Wnt7bC3 allele was generated as previously described (33). Mice het-erozygous for the Wnt7bC3 allele (C57BL6) were crossed with germline Cre mice(34) or strain-matched controls, to generate heterozygous mutants (Wnt7b+/−).Genotyping for the presence of the Wnt7bC3 allele was performed with 5′-GTCTCTGTCCTTAGTTGGGTC-3′, 5′-CCAGAGACCAGTACACCTGAG-3′ primers.Mutants were backcrossed withWnt7bC3/C3 mice, and the offspring were crossedwith Csf1R-icre transgenic mice, resulting in Csf1R-icre; Wnt7bC3/− experimentalmice, and Wnt7bC3/− and Csf1R-icre; Wnt7bC3/+ controls. Presence of the Wnt7bconditional allele or WT allele was confirmed using the following primers: 5′-TGACAGAGGATGGGGAGAAG-3′, 5′-GGTCTTTCCAAGGGTGGTCT-3′.

Statistical Analysis. Error bars are standard error of the mean. Comparisonsbetweengroupsweretestedusingthepairedandunpaired t test,orsingle-factorone-way analysis of variance. Survival was analyzed using the Mantel–Cox log-rank test. All tests were carried out using GraphPad Prism (GraphPad Software).

ACKNOWLEDGMENTS.We thankDr.Wei Hsu (University of Rochester) andDr.Walter Birchmeier (Max Delbruck Center, Berlin) for the Axin2+/LacZ mice, Dr.Jeremy Nathans (Johns Hopkins University School of Medicine) for theFrizzled4+/LacZ mice, Dr. Jennifer K. Ondr and Dr. Alfor Lewis (University ofCincinnati) for assistance with initial reporter mouse studies, Dr. Kenneth D.Swanson [Harvard Medical School (HMS)] for assistance with retroviral assays,and Deneen Kozoriz (HMS), Dr. Jayaraj Rajagopal (Harvard), and Huaying Pei(HMS) for assistance. TheDuffield laboratory is supported byNational Institutesof Health (NIH) Grants DK73299, DK84077, and DK87389, the American Societyof Nephrology Gottschalk Award, Genzyme Renal Initiatives Program, and aNational Taiwan Merit Award (to S.-L.L.). The Lang laboratory is supported byNIH RO1s EY16241, EY15766, EY17848, and CA131270 and funds from thePearle Vision Foundation, Research to Prevent Blindness, and the AbrahamsonPediatric Eye Institute Endowment at Children’s Hospital Medical Center ofCincinnati. Work was also supported by NIH Grant DK054364 (to A.P.M.).

1. Mantovani A, Allavena P, Sica A (2004) Tumour-associated macrophages as aprototypic type II polarised phagocyte population: Role in tumour progression. Eur JCancer 40:1660–1667.

2. Duffield JS, et al. (2005) Selective depletion of macrophages reveals distinct, opposingroles during liver injury and repair. J Clin Invest 115:56–65.

3. Nahrendorf M, et al. (2007) The healing myocardium sequentially mobilizes two mono-cyte subsets with divergent and complementary functions. J Exp Med 204:3037–3047.

4. Arnold L, et al. (2007) Inflammatory monocytes recruited after skeletal muscle injuryswitch into antiinflammatory macrophages to support myogenesis. J Exp Med 204:1057–1069.

5. Fallowfield JA, et al. (2007) Scar-associated macrophages are a major source ofhepatic matrix metalloproteinase-13 and facilitate the resolution of murine hepaticfibrosis. J Immunol 178:5288–5295.

6. Castaño AP, et al. (2009) Serum amyloid P inhibits fibrosis through FcγR-dependentmonocyte-macrophage regulation in vivo. Science Transl Med 1:5ra13.

7. Lobov IB, et al. (2005) WNT7b mediates macrophage-induced programmed cell deathin patterning of the vasculature. Nature 437:417–421.

8. Duffield JS, et al. (2005) Restoration of tubular epithelial cells during repair of thepostischemic kidney occurs independently of bone marrow-derived stem cells. J ClinInvest 115:1743–1755.

9. Humphreys BD, et al. (2008) Intrinsic epithelial cells repair the kidney after injury. CellStem Cell 2:284–291.

10. Nusse R (2005) Wnt signaling in disease and in development. Cell Res 15:28–32.11. Greco V, et al. (2009) A two-step mechanism for stem cell activation during hair

regeneration. Cell Stem Cell 4:155–169.12. Ito M, et al. (2007) Wnt-dependent de novo hair follicle regeneration in adult mouse

skin after wounding. Nature 447:316–320.13. Maretto S, et al. (2003) Mapping Wnt/β-catenin signaling during mouse development

and in colorectal tumors. Proc Natl Acad Sci USA 100:3299–3304.14. Yu HM, et al. (2005) The role of Axin2 in calvarial morphogenesis and

craniosynostosis. Development 132:1995–2005.15. Zeng X, et al. (2005) A dual-kinase mechanism for Wnt co-receptor phosphorylation

and activation. Nature 438:873–877.16. Lin SL, Kisseleva T, Brenner DA, Duffield JS (2008) Pericytes and perivascular

fibroblasts are the primary source of collagen-producing cells in obstructive fibrosis ofthe kidney. Am J Pathol 173:1617–1627.

17. Duffield JS, et al. (2006) Resolvin D series and protectin D1 mitigate acute kidneyinjury. J Immunol 177:5902–5911.

18. Duffield JS, et al. (2005) Conditional ablation of macrophages halts progression ofcrescentic glomerulonephritis. Am J Pathol 167:1207–1219.

19. Xu Q, et al. (2004) Vascular development in the retina and inner ear: Control byNorrin and Frizzled-4, a high-affinity ligand-receptor pair. Cell 116:883–895.

20. Glass DA, II, et al. (2005) Canonical Wnt signaling in differentiated osteoblastscontrols osteoclast differentiation. Dev Cell 8:751–764.

21. Holmen SL, et al. (2004) Decreased BMD and limb deformities in mice carryingmutations in both Lrp5 and Lrp6. J Bone Miner Res 19:2033–2040.

22. Kato M, et al. (2002) Cbfa1-independent decrease in osteoblast proliferation,osteopenia, and persistent embryonic eye vascularization in mice deficient in Lrp5, aWnt coreceptor. J Cell Biol 157:303–314.

23. Deng L, et al. (December 30, 2009) A novel mouse model of inflammatory boweldisease links mammalian target of rapamycin-dependent hyperproliferation ofcolonic epithelium to inflammation-associated tumorigenesis. Am J Pathol.

24. Ichimura T, et al. (2008) Kidney injury molecule-1 is a phosphatidylserine receptor thatconfers a phagocytic phenotype on epithelial cells. J Clin Invest 118:1657–1668.

25. Cude K, et al. (2007) Regulation of the G2-M cell cycle progression by the ERK5-NFκBsignaling pathway. J Cell Biol 177:253–264.

26. DickinsonAJ,SiveHL(2009)TheWntantagonistsFrzb-1andCrescentlocallyregulatebasementmembrane dissolution in the developing primarymouth.Development 136:1071–1081.

27. Mao B, et al. (2002) Kremen proteins are Dickkopf receptors that regulate Wnt/β-catenin signalling. Nature 417:664–667.

28. Mao B, Niehrs C (2003) Kremen2 modulates Dickkopf2 activity during Wnt/LRP6signaling. Gene 302:179–183.

29. Inoue T, et al. (2003) Hepatocyte growth factor counteracts transforming growthfactor-β1, through attenuation of connective tissue growth factor induction, andprevents renal fibrogenesis in 5/6 nephrectomized mice. FASEB J 17:268–270.

30. Yu J, et al. (2009) A Wnt7b-dependent pathway regulates the orientation ofepithelial cell division and establishes the cortico-medullary axis of the mammaliankidney. Development 136:161–171.

31. Reya T, et al. (2003) A role for Wnt signalling in self-renewal of haematopoietic stemcells. Nature 423:409–414.

32. Tauber AI (2003) Metchnikoff and the phagocytosis theory. Nat Rev Mol Cell Biol 4:897–901.

33. Rajagopal J, et al. (2008) Wnt7b stimulates embryonic lung growth by coordinatelyincreasing the replication of epithelium and mesenchyme. Development 135:1625–1634.

34. Lakso M, et al. (1992) Targeted oncogene activation by site-specific recombination intransgenic mice. Proc Natl Acad Sci USA 89:6232–6236.

Lin et al. PNAS | March 2, 2010 | vol. 107 | no. 9 | 4199

CELL

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Janu

ary

23, 2

020