Role for compartmentalization in nephronprogenitor differentiationAaron C. Brown, Sree Deepthi Muthukrishnan, Justin A. Guay, Derek C. Adams, Dillon A. Schafer, Jennifer L. Fetting,and Leif Oxburgh1

Center for Molecular Medicine, Maine Medical Center Research Institute, Scarborough, ME 04074

Edited by Frank Costantini, Columbia University, New York, NY, and accepted by the Editorial Board February 7, 2013 (received for review August 27, 2012)

Embryonic nephron progenitor cells are segregated in molecularlydistinct compartments of unknown function. Our study reveals anintegral role for bone morphogenetic protein-SMAD in promotingtransition of progenitors from the primitive Cbp/p300-interactingtransactivator 1 expressing (CITED1+) compartment to the uniquelysine oculis-related homeobox 2 expressing (SIX2-only) compartmentwhere they become inducible by wingless-type mouse mammary tu-mor virus integration site family member (WNT)/β-catenin signaling.Significantly, CITED1+ cells are refractory toWNT/β-catenin induction.We propose a model in which the primitive CITED1+ compartment isrefractory to inductionbyWNT9b/β-catenin, ensuringmaintenanceofundifferentiatedprogenitor cells for future nephrogenesis. Bonemor-phogenetic protein 7-SMAD is then required for transition to adistinctcompartment in which cells become inducible by WNT9b/β-catenin,allowing them to progress toward epithelialization.

cap mesenchyme | niche | pretubular aggregate | nephrogenic zone |kidney development

Metanephric kidney development has been intensely studiedsince the 1950s when Clifford Grobstein used this system

to reveal many of the fundamental principles of epithelial–mesenchymal interaction in organogenesis (1–3). Knockout andtransgenic mouse models have clarified the genetic basis for manyof these principles, and gene expression analyses have uncoveredunanticipated distinctions between cell types in the developingkidney that were not apparent frommorphological studies such asthe arrangement of nephron progenitor cells in a series of mo-lecularly distinct compartments (4, 5). The significance of thesecompartments is unclear, and understanding their functions isimportant for our basic knowledge of kidney development and forongoing efforts to differentiate kidney tissue from pluripotentstem cells.Formation of a sufficient number of nephrons during kidney

development relies on a continuous supply of nephron progenitorcells. As the collecting system branches, progenitor cell aggre-gates at collecting duct tips known as cap mesenchymes are in-duced to form nephrons. The collecting duct provides survival,proliferation, and differentiation signals to the cap mesenchyme,which ultimately gives rise to all epithelial components of thenephron (6–8). In turn, cap mesenchyme provides signals forgrowth and branching of the collecting duct. These events takeplace within the nephrogenic zone, a progenitor cell niche wherebone morphogenetic proteins (BMPs), fibroblast growth factors(FGFs), and wingless-type mouse mammary tumor virus in-tegration site family members (WNTs) maintain the balancebetween renewal and differentiation (9–12). Studies of WNTsignaling in particular have demonstrated bimodal effects on capmesenchyme cells, with WNT9b-activated β-catenin signaling(canonical WNT signaling) causing both renewal of un-differentiated cells and epithelial differentiation (13). The sineoculis-related homeobox 2 (SIX2) transcription factor blocks theinductive activities of WNT signaling, and it has been suggestedthat the level of SIX2 expression in progenitor cells determinesresponsiveness toWNT9b, indicating that contradictory effects ofWnt9b may be explained by differential responsiveness of cells

within cap mesenchyme (13, 14). Using in vitro nephrogenic zoneculture, we have studied the responsiveness of individual capcompartments to inductive signaling and defined signals thatdrive the transition of cells through sequential compartmentstoward nephron differentiation. We find that BMP7-induced tran-sition between compartments is required for cells to acquireresponsiveness to WNT/β-catenin signaling. This unanticipatedfunction of BMP7 is mediated through the SMAD signalingpathway, in contrast to the progenitor renewal effect that we haveshown operates throughmitogen activated protein kinase (MAPK)(15). Based on these findings, we propose a model whereby com-partmentalization is necessary to prevent Cbp/p300-interactingtransactivator 1 (CITED1)-expressing cells from premature in-duction, allowing maintenance of undifferentiated cells within thecap mesenchyme for future rounds of nephrogenesis.

Results and DiscussionBMP7 Promotes Transition of Cap Mesenchyme Cells Between EarlyProgenitor Compartments. The earliest progenitor cells associatedwith the capsular aspect of the collecting duct tip express CITED1(Fig. 1A), whereas the slightly more differentiated populationassociated with the internal aspect of the collecting duct tiploses expression of CITED1 but maintains expression of SIX2(Fig. 1B). Lymphoid enhancer binding factor 1 (LEF1) expressionis limited to the more differentiated pretubular aggregate that isprimed for epithelial conversion (Fig. 1C). Immunofluorescencestaining of CITED1, SIX2, and LEF1 proteins shows that a pop-ulation of “SIX2 only” progenitors resides between the CITED1+

and LEF1+ cells (Fig. 1D). A comparison of green fluorescentprotein (GFP) expression within cap mesenchymes of Cited1-creERT2-Gfp and Six2-creERT2-Gfp strains reveals a similar dis-crepancy, strongly arguing for a genuine difference in expressiondomains rather than distinct half-lives of the RNAs or proteins(7, 8). These results suggest an arrangement of distinct cell states,or compartments, in which CITED1 expression is lost as cellsdifferentiate to a SIX2-only compartment before entering theLEF1+ compartments after which they epithelialize in the renalvesicle (Fig. 1E). Having previously established that CITED1+

nephron progenitors depend on FGF/receptor tyrosine kinasesignaling for maintenance of their phenotype, we sought to un-derstand which signals push CITED1+ cells to the LEF1+ pre-tubular aggregate compartment (16).To screen for pathways that promote CITED1+ progenitor

differentiation, we used the primary nephrogenic zone culturesystem (15, 17). This monolayer culture of nephrogenic zone cells

Author contributions: A.C.B. and L.O. designed research; A.C.B., S.D.M., D.C.A., and D.A.S.performed research; J.A.G. and J.L.F. contributed new reagents/analytic tools; A.C.B. andL.O. analyzed data; and A.C.B. and L.O. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission. F.C. is a guest editor invited by the EditorialBoard.1To whom correspondence should be addressed. E-mail: oxburl@mmc.org.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1213971110/-/DCSupplemental.

4640–4645 | PNAS | March 19, 2013 | vol. 110 | no. 12 www.pnas.org/cgi/doi/10.1073/pnas.1213971110

Dow

nloa

ded

by g

uest

on

July

26,

202

0

(NZCs) essentially lacks inductive cells of the ureteric bud (lessthan 0.04% contamination). We screened signaling pathwayswith known effects in kidney development for their ability to ex-tinguish CITED1+ expression, while maintaining proliferationand growth of the culture (Fig. S1). Surprisingly, stimulatingβ-catenin signaling with the glycogen sythase kinase 3β (GSK3β)inhibitor 6-bromoindirubin-3′-oxime (BIO) caused only a modesteffect (Fig. S1). This finding was unanticipated because β-cateninactivation promotes epithelial differentiation of nephron pro-genitor cells in cultured metanephric mesenchymes (18). How-ever, BMP7 treatment reduced CITED1 protein expressiondrastically compared with all other recombinant proteins as de-termined by immunofluorescence and Western blot (Fig. 1F andFig. S1). BMP7 increased proliferation and significantly expandedthe total number of progenitors while maintaining more than95% viability (Fig. S2). When BMP7 is replaced with vehiclecontrol after 24 h, expression of CITED1 is not reacquired,showing an irreversible exit of cells from the CITED1+ state (Fig.S2). Although BMP7 treatment clearly causes loss of CITED1and maintenance of SIX2, it does not promote LEF1 expression,showing that cells do not differentiate to the pretubular aggregatebut simply transition to the SIX2-only compartment (Fig. 1F). Tounderstand whether this unanticipated effect of BMP7 is a featureof the culture system, or if it occurs in vivo, we analyzed CITED1and SIX2 expression in kidneys from the Bmp7−/− mouse. Atembryonic day (E)14.5, Bmp7−/− kidneys are hypomorphic yet thenephrogenic zone can still be studied. Numbers of CITED1+ cellsare reduced in the mutant kidney, and they are linearly arrangedaround the periphery in contrast to wild type in which they wraparound collecting duct tips (Fig. 1 G–J). However, althoughCITED1+ cells are present in mutant kidneys, loss ofBmp7 causesa substantial decrease of the SIX2-only population (Fig. 1 I–K).These results strongly support a role for BMP7 in promotingtransition of nephron progenitor cells from the CITED1+ com-partment to the SIX2-only compartment in vivo.

SMAD-Dependent Signaling Is Required for Transition of Progenitorsfrom the CITED1+ Compartment. We have shown that BMP7 pro-motes nephron progenitor proliferation through TAK1-MAPK,explaining the reduction of CITED1+ cells seen in the E14.5Bmp7−/− (Fig. 1H) (15). However, we also confirmed that SMAD1/5is phosphorylated in nephron progenitors in response to BMP7treatment (Fig. 2A). Interestingly, tissue staining reveals nuclearphospho-SMAD1/5 (pSMAD1/5) in a subset of progenitors un-derneath the collecting duct tip, at the junction between CITED1+

and CITED1− cells (Fig. 2B). To determine which BMP7-initiatedsignaling branch is responsible for the transition of progenitors fromthe CITED1+ to the SIX2-only compartment, NZCs were treatedwith BMP7 with or without the addition of dorsomorphin, a smallmolecule BMP-SMAD (pSMAD1/5) inhibitor, or TAK1 inhibitor(Fig. 2C) (19, 20). BMP7 treatment with dorsomorphin, but notTAK1 inhibitor, blocked the ability of BMP7 to promote tran-sition out of the CITED1+ compartment, showing that SMADsignaling mediates this effect. Mesenchyme homeobox 1 (Meox1)and D4, zinc and double PHD fingers, family 3 (Dpf3), which arecoexpressed with Cited1 in the cap mesenchyme (5), also showcoregulated expression with Cited1 indicating that BMP7-SMADmediated transition out of the CITED1+ compartment representsa change of cellular state rather than simply a change of Cited1expression (Fig. 2D). Furthermore, tissue staining revealed a loss ofpSMAD1/5 in the distal cap of the Bmp7 null kidney at E14.5,confirming that SMADsare activated in the distal cap specifically byBMP7 in vivo (Fig. 2E and Fig. S3). In summary, BMP7 treatmentpromotes transition of progenitors from the CITED1+ to the SIX2-only compartment by SMAD activation, and the finding thatpSMAD1/5 is regionalized in the distal cap mesenchyme supportsa role for this process in vivo. SMAD4 is required for nucleartranslocation of pSMAD1/5, and its conditional inactivation innephron progenitor cells results in partial recapitulation of theBmp7−/− phenotype with premature cessation of nephrogenesis,but with retention of PAX2 and WT1 expressing nephron pro-genitor cells within the nephrogenic zone (6). This phenotype is

CITED1+SIX2+LEF1-

CITED1- SIX2- LEF1+

CITED1-SIX2+LEF1-

CITED1-SIX2+/-LEF1+

FCITED1

CD

capA

ELEF1

CD

CB

PTA

C

CITED1/SIX2

B

H

G

Veh

icle

BM

P7

CITED1

10X

CITED1/SIX2 LEF1

20X 10X

10X 20X 10X

BM

P7 n

ull

BM

P7 w

t

CITED1 CITED1/SIX2

J

I

0

10

20

30

40

50

60

70

80

Wild type Bmp7 null

Perc

ent

SIX

2 o

nly

clu

ster

s p

er b

ud

tip P<0.00001

CITED1/SIX2/LEF1

D

K

"CITED1+" "SIX2-only" "LEF1+"

Cap mesenchyme Renal vesicleFig. 1. BMP7 promotes transition of CITED1+ pro-genitors to the SIX2-only compartment. (A) CITED1(red) localizes to the capsular aspect of the collect-ing duct (green). (B) Progenitors in the cap loseCITED1 (red) but retain SIX2 (green, arrows). (C)LEF1 (red) is limited to the pretubular aggregateand downstream compartments. (D) A “SIX2 only”progenitor population (green, arrows) resides be-tween the CITED1+ (orange/yellow) and LEF1+

compartments (red). (E) Schematic representationof compartments showing loss of CITED1 as cellsdifferentiate to SIX2-only before entering the LEF1+

pretubular aggregate and renal vesicle compart-ments. (F) BMP7 treatment results in a loss ofCITED1 (red), maintenance of SIX2 (green), anddoes not promote LEF1 (red, absent). E14.5 Bmp7−/−

kidneys are hypomorphic, have a reduced numberof CITED1+ progenitors (G and H), and a substantialdecrease of the SIX2-only population (I and J,arrows). (K) Quantitation of SIX2-only clusters inE14.5 Bmp7−/− (n = 4, 130 total tips) and wild-typekidneys (n = 6, 510 total tips). Error bars representaverage values ± SEM, and the P value is derivedfrom Student t test.

Brown et al. PNAS | March 19, 2013 | vol. 110 | no. 12 | 4641

DEV

ELOPM

ENTA

LBIOLO

GY

Dow

nloa

ded

by g

uest

on

July

26,

202

0

recapitulated by inactivation of Cv2, which acts as an amplifier ofBMP7-induced SMAD signaling in cap mesenchyme (21). Thesegenetic findings strongly support the concept that BMP7/SMADsignaling is essential for progenitor cell differentiation.

SMAD-Dependent BMP Signaling Primes CITED1 Progenitors forInduction by WNT/β-catenin. WNT9b/β-catenin signaling is re-quired for differentiation of cap mesenchyme cells (12, 22).However, in our hands, activation of the WNT pathway directly

in CITED1+ progenitor cells using BIO or recombinant WNT9bhad only a mild effect on differentiation from the CITED1+ tothe LEF1+ compartment (Fig. 3A and Fig. S1). We reasoned thatprogression through sequential cap mesenchyme compartmentsmight be required for acquisition of susceptibility to differenti-ating signals. We therefore tested whether BMP7-induced SIX2-only cells gain the potential for β-catenin–mediated differentia-tion to the LEF1+ stage. We stimulated freshly isolated NZCcultures with BMP7 for 24 h to promote their transition from theCITED1+ to the SIX2-only compartment and added BIO duringthe remaining 16 h (Fig. 3A). Under these conditions, LEF1+

cells became clearly abundant, demonstrating that BMP7-induced transition from the CITED1+ to the SIX2-only com-partment primes cells to respond to WNT/β-catenin induction. Inline with our earlier results, treatment with dorsomorphin com-pletely abrogated the capacity of BMP7 treatment to sensitizethe culture to BIO-mediated differentiation. Similar results wereseen at the transcriptional level (Fig. 3B), and we also demon-strated that BMP7 and BMP4 could function redundantly in thiscapacity (Fig. S4). Interestingly, increased Wnt4 transcriptionaccompanies LEF1 expression upon BMP7 and BIO treatment,indicating that the cells are entering the renal vesicle stage(Fig. 3B). The β-catenin dependence of the response to BIO wasverified by using an inhibitor of WNT/β-catenin (FH535), whichcompletely abrogatedWnt4 andLef1 transcription responses (Fig.3C and Fig. S4). Lithium chloride (LiCl) has been more com-monly used as an inducer of nephrogenesis, and to verify that itfunctions interchangeably with BIOwith respect to differentiationof NZCs, we repeated the experiment with LiCl (Fig. 3C). Indeed,a similar effect was observed with LiCl, but the amplitude ofWnt4activation was lower. These results indicate that SMADmediateddifferentiation from the CITED1+ to the SIX2-only compartmentis required for β-catenin–mediated differentiation. Cells of thecap mesenchyme express Bmp7, and to test whether endoge-nously produced growth factor sensitizes cells to the differenti-ating effect of canonical WNT signaling, we measured Wnt4expression in BIO-treated NZCs after BMP-SMAD inhibition(Fig. 3D). We found that the small increase in Wnt4 expressionseen after BIO treatment can be reduced by dorsomorphin, sug-gesting signaling by endogenously produced BMP7 in this culturesystem. Also, immunoblotting for pSMAD1/5 reveals a modestsignal in untreated NZCs that can be quenched by dorsomorphin(Fig. S4). We hypothesize that the large volume of the monolayersystem dilutes endogenously produced BMP7 to a level that isinadequate to fully prime cells for BIO induction.

Effects of BMP7 on Compartment Transition Are Inherent to CITED1+

Progenitors. To mimic signaling events in the nephrogenic zone,our studies have used primary NZCs. This culture is a mixedpopulation, and there is a possibility that the synergistic effects ofSMAD and β-catenin on CITED1+ cells are not the result ofdirect signaling to these cells. To test whether SMAD andβ-catenin act directly on CITED1+ cells, we developed a methodfor their purification. We have been unable to identify a surfacemarker specific to CITED1+ cells, so we developed an antibody-based protocol to deplete cells other than CITED1+. Majorpopulations identified in the mixed culture are shown in Fig. S5.Using antibody-based magnetic depletion of CD105+, CD140a+,TER-119+, and CD326+ cells, we were able to remove more than99% of these cells (Fig. 4A). Immunostaining demonstrates anenrichment of CITED1+ progenitors from 55 to more than 96%(Fig. 4 B and C). Similar to NZCs, pure CITED1+ cells requireBMP7 priming for a vigorous differentiation response to β-cateninsignaling (Fig. 4 D and E), indicating that niche cells within thenephrogenic zone are neither required for CITED1+ cells tomaintain their refractory state to WNT/β-catenin induction, or fortheir susceptibility to induction after BMP7 treatment.

EVehicleVehicle

BMP7+TAKiBMP7+TAKi

BMP7BMP7

BMP7+DMBMP7+DM

A

D

-5

-4

-3

-2

-1

0

1

2

3

Vehicle BMP7 BMP7+TAKi BMP7+DM

Rel

ativ

e m

RN

A e

xpre

ssio

n

Cited1Meox1Dpf3

qPCR of coexpressed genes within early cap

Wild-type

Bmp7 null

40X (pSmad1/5)

PAX2 / pSMAD1/5PAX2 / pSMAD1/5BB

MP7

Veh

icle

CITED1 / pSMADCITED1 / pSMAD

PTAPTA

PAX2 / pSMAD1/5PAX2 / pSMAD1/5

C

40X (pSmad1/5)

CIT

ED1

Fig. 2. BMP7 promotes SMAD-mediated signaling in nephron progenitors.(A) BMP7 treatment results in pSMAD1/5 (green) activation in PAX2 pro-genitors (red); costaining is yellow. (B) Immunofluorescence of E17.5 kidneysections shows nuclear pSMAD1/5 in the distal cap underneath collectingduct tips (arrows). Inset shows CITED1 (red) and pSMAD1/5 (green) at theCITED1+/CITED1− border. (C) CITED1 staining (red) of NZCs pretreated withvehicle, BMP7, BMP7 + dorsomorphin (DM), or BMP7 + TAK1 inhibitor (TAKi)shows that SMAD1/5 inhibition blocks the ability of BMP7 to reduce CITED1expression. (D) Quantitative PCR analysis shows that inhibition of SMAD-dependent signaling by DM, compared with TAKi, blocks the ability of BMP7to reduce transcription (48 h) of a group of early progenitor markers (Cited1,Meox1, Dpf3). Raw data are normalized to β-actin expression, and foldchanges are relative to the vehicle control. (E) pSMAD1/5 in the distal capmesenchyme is lost in the Bmp7−/−.

4642 | www.pnas.org/cgi/doi/10.1073/pnas.1213971110 Brown et al.

Dow

nloa

ded

by g

uest

on

July

26,

202

0

CITED1+ Progenitors Require BMP and WNT Signals to Form EpithelialTubules. To test the capacity of CITED1+ cells for epithelial in-duction, we modified our culture system for 3D growth, spottingdensely packed cellular aggregates on Nuclepore filters andculturing under serum-free organ culture conditions. E11.5 kid-ney mesenchyme cells undergo epithelialization when inducedwith LiCl, and we therefore tested the differentiation capacityof our culture system by using aggregated single-cell suspensionsof E11.5 mesenchymes (18). Immunostaining for E-cadherinrevealed that LiCl-treated aggregates form epithelia (Fig. 5 Aand B). Surprisingly, aggregates of E17.5 CITED1+ cells do notepithelialize, although LiCl promotes their survival and adher-ence (Fig. 5 C and D). Thus, although epithelia can be differ-entiated from E11.5 mesenchyme cells by using a standard LiCldifferentiation procedure, E17.5 nephron progenitors are re-fractory. LiCl elicits a weaker differentiation response than BIO(Fig. 3C), and we therefore tested the ability of BIO to promoteepithelialization in E17.5 aggregates. We found that 2 μM BIOcauses vigorous epithelial tubule formation in E17.5 CITED1+

aggregates (Fig. 5 E–G). We conclude that our 3D culture sys-tem can be used to study epithelialization of E17.5 nephron

progenitors and that pure CITED1+ cells are competent to dif-ferentiate into epithelial tubules. However, the capacity to inducedifferentiation directly with BIO differs from our monolayersystem, in which BMP7 pretreatment is required to prime cellsfor this response. Aggregate culture differs from monolayer inthat endogenously produced growth factors can be trapped inthe matrix between cells, resulting in high local concentrations.CITED1+ cells produce endogenous BMP (Fig. 3D), and wehypothesized that ligand might be sufficiently locally concen-trated in 3D culture to prime cells for BIO induction, obviatingthe need for BMP7 addition. Loss of expression of the BMPresponse gene Cv2 in cultures treated either with dorsomorphinor with noggin indicates active endogenous signaling, and treat-ment of aggregates with dorsomorphin prevents epithelial in-duction by BIO (Fig. 5 H and I and Fig. S6). We conclude thatBMP-SMAD signaling is required to prime cells for inductionby BIO in 3D culture and that adequate BMP signal intensity isachieved from endogenously produced ligand. In agreement withprevious studies, inhibition of the planar cell polarity and Ca2+

release pathways in CITED1+ progenitors treated with BIO pre-vents tubulogenesis, demonstrating a requirement for noncanonical

LEF1

imm

un

oflu

ore

scen

ce

A B CActivation and inhibition

of Wnt4BMP and WNT synergism

BMP 7 BIO BMP 7+BIO

Cited1Lef1Wnt4

**

*

*

D

BM

P7

BIO

BM

P7+

BIO

BM

P7+

BIO

+FH

535

LiC

l

BM

P7+

LiC

l

BM

P7+

LiC

l+FH

535

Activation of Wnt4by endogenous BMP

BIO

+D

M

BM

P7

BIO

BM

P7+

BIO

0

10

20

30

40

50

60

70

Fold

chan

ge

rela

tive

tove

hic

le

Fold

chan

ge

rela

tive

tove

hic

le-125

-75

-25

25

75

125

175

225

Fold

chan

ge

rela

tive

tove

hic

le

012345

40

45

50

55BMP7 BIO

BMP7+BIO BMP7+BIO+DM

Fig. 3. BMP-SMAD signaling primes nephron progenitors for induction by WNT/β-catenin. (A) LEF1 expression (red) in NZCs treated with the indicatedfactors. Pretreatment with BMP7 is required for BIO-induced transition to the LEF1 compartment and this effect is reversed by dorsomorphin. (B) QuantitativeRT-PCR (RT-qPCR) of NZCs primed with BMP7 before 8-h BIO treatment showing a synergistic induction of Lef1 and Wnt4 and reduction in Cited1. Results arederived from three independent replicates per treatment. Error bars represent average values ± SEM, and P values are derived from the Student t test. *P <0.03, **P < 0.002 when BMP7+ BIO treatment is compared with either BMP7 or BIO alone. (C) RT-qPCR of NZCs showing a synergistic induction of Wnt4 withBMP7 plus BIO or LiCl. Pretreatment with the WNT/β-catenin inhibitor FH535 abrogates the increase inWnt4 expression. Error bars represent average values ±SD. (D) RT-qPCR of NZCs shows a small increase in Wnt4 after BIO treatment that can be quenched by dorsomorphin, indicating a low level of endogenousBMP activity in the monolayer culture system. Error bars represent average values ± SD.

Total NZCs

CITED1 pure

PresumptiveCITED1+

Ter119+

CD105+

CD140a+

PresumptiveCITED1+ >99%

LEF1 immunofluorescenceVehicle BMP7

BIO BMP7+BIO

A

Rela

tive

to v

ehic

le c

on

tro

l

0

2

4

6

8

10

12

14

0

10

20

30

40

50

60

70

80

90

Vehicle BIOBMP7 BMP7+BIO

Lef1

Wnt4

P < 0.005

0

10

20

30

40

50

60

70

80

90

100

Perc

ent

CIT

ED1+

CIT

ED1

enri

ched

Tota

l NZ

Cs

B C D E

Fig. 4. Effects of BMP-SMAD are inherent to CITED1+ progenitors. (A) Flow cytometry showing that CD105, CD140a, TER-119, and CD326 mark specificsubpopulations of NZCs (Upper) and that antibody based magnetic depletion efficiently removed these populations (Lower). PE-high labeled TER-119+ cellsare distinguished from PE-low labeled CD105+ cells as verified by additional FITC staining (Fig. S5B). (B) CITED1 staining of depleted cells demonstrates anenrichment of CITED1+ progenitors (red) from 55 to greater than 96%. (C) Quantified results from B. Error bars represent average values ± SEM, and P valuesare derived from Student’s t test. (D) RT-qPCR showing synergistic induction of Lef1 and Wnt4 in CITED1 pure cultures primed with BMP7 before 8 h of BIOtreatment. (E) LEF1 (red) in CITED1 pure cultures after treatment with the indicated factors as described in Fig. 3A.

Brown et al. PNAS | March 19, 2013 | vol. 110 | no. 12 | 4643

DEV

ELOPM

ENTA

LBIOLO

GY

Dow

nloa

ded

by g

uest

on

July

26,

202

0

WNT signaling (Fig. 5 J and K). Noncanonical WNT signaling islikely mediated through induction of WNT4, because tubulo-genesis with BIO is inhibited with the WNT secretion inhibitorIWP2 (Fig. S6). E11.5 cells are more sensitive to LiCl inductionthan E17.5 cells, and we therefore asked whether there is the samerequirement for BMP-SMAD priming at these two stages. Usingdorsomorphin, we found that BMP-SMAD signaling is dispens-able for the differentiating effect of BIO in E11.5 aggregates (Fig.5 L–N). This observation suggests that the early metanephricmesenchyme may be highly responsive to β-catenin induction,whereas later nephron progenitors require BMP7 priming. Thisfeature could be an adaptation to the thinning of capmesenchymeseen from E11.5–E13.5 and a requirement for maintenance ofundifferentiated CITED1+ cells at the rapidly growing collectingduct tips. Intriguingly, nephron formation occurs in both Bmp7 nulland Bmp7cre;Smad4 conditional mutant kidneys up to approxi-mately E13.5, supporting the idea that BMP-SMAD signaling isrequired for cap differentiation at later stages of nephrogenesis.

CITED1+ Progenitors Must Be Programmed Before Induction by WNT4.Previous studies have shown that WNT4 can directly induce thedifferentiation of E11.5 metanephric mesenchymes (23). WNT4 isknown to elicit a noncanonical response, and there is a possibilitythat this mode of signaling may be exempt from the requirement fortransition to the SIX2-only compartment for induction. Recently,a method for induction of rat metanephric mesenchyme by WNT4

soaked beads was described (24). Using this method, we were un-able to provoke epithelial induction in E17.5 CITED1+ 3D cultures(Fig. S7), indicating that WNT4 is not capable of inducing cells inthe CITED1+ compartment. To confirm this finding, we developedamethod to transfect CITED1+ 3D aggregate cultures (Fig. 5O–T).Transfection of a GFP control construct resulted in protein ex-pression, but aggregates show no evidence of epithelialization up to14 d in culture. As expected, transfection with a construct encodingWNT9b resulted in vigorous epithelialization and expression of theinductionmarkersLef1, jagged 1 (Jag1), andWnt4 after 3 d (Fig. 5Qand R and Fig. S7). Transfection with WNT4, however, did notresult in any noticeable epithelialization and actually decreasedexpression of Lef1, Jag1, and endogenousWnt4 (Fig. 5 S and T andFig. S7). From these studies we find no evidence thatWNT4 has thecapacity to directly induce CITED1+ cells, and we propose thatorderly progression through progenitor compartments is required toacquire susceptibility to WNT4 epithelialization.CITED1+ progenitor maintenance is driven by FGFs 9 and 20

and low-level WNT9b/β-catenin signaling (13, 16, 25). β-catenin–mediated signaling elicited by ureteric bud derived WNT9bdirects progenitors to express WNT4 that, in turn, stimulatesnoncanonical WNT signaling, inducing formation of epithelia fromnephron progenitors in an autocrine manner (12, 24). We there-fore propose a model for epithelial induction of cap mesenchymein which cells of the CITED1+ compartment are refractory to theinductive action of WNT9b until they have undergone BMP7-

BIO+DorsomorphinH. Cv2 expression in CITED1+ aggregate

DM NOG Vehicle0

2

4

6

Fold

ch

ang

e re

lati

ve t

o D

M

CITED1 + vehicle (E17.5)

CITED1 + LiCl (E17.5)

NZCs + LiCl (E11.5)

NZCs vehicle (E11.5)

Vehicle control BIO BIO + dorsomorphin

F vehicle BIO

PAX8+FGF8+

Wnt9bSIX2+

LEF1+

ECAD+

BMP7

WNT4

WNT4

CITED1+/SIX2+ MAPK

pSMAD1/5

β-catenin

β-catenin

WNT4

BIO+SP600125BIO+Cyclosporin A

GFP only Wnt9b Wnt4

A B C D E F G

I J K

P R T

L M N

O Q S

U

GFP Wnt9b Wnt4

BIO - 40X

Wnt4

Wnt9b

B-actin

Wn

t4

Wn

t9b

GFP

construct

PCR

V

Fig. 5. CITED1+ progenitors form tubules directly in response to BMP and WNT signaling. (A and B) E-cadherin staining (green) in single-cell aggregates fromE11.5 mesenchymes epithelialize after 5 d in culture on Nuclepore filters when treated with LiCl (B), but do not survive in the absence of LiCl (A). (C and D)E17.5 CITED1+ progenitors do not undergo epithelialization when treated with LiCl (D). (E–G) E17.5 CITED1+ progenitors undergo expansion and tubulo-genesis when cultured with BIO. DAPI is blue, and lotus lectin is red. G shows 40× enlarged image of epithelialized tubule from boxed area in F. (H) BMPantagonists dorsomorphin and noggin block endogenous BMP activity in E17.5 CITED1+ culture as determined by RT-qPCR of the BMP response gene Cv2. (I)Dorsomorphin blocks tubulogenesis in BIO-treated CITED1+ progenitors. Negative and positive controls shown in E and F. (J and K) Canonical BMP and WNTmediated tubulogenesis of CITED1+ progenitors is extinguished by inhibitors of noncanonical WNT signaling, including Ca2+ release (cyclosporin A) and JNK(SP600125) pathways. (L–N) Dorsomorphin treatment does not block tubulogenesis in single-cell aggregates from E11.5 mesenchymes treated with BIO. Insetsshow lotus lectin staining (red) from the boxed regions. (O–T) E17.5 CITED1+ progenitors transfected with Wnt9b, but not Gfp or Wnt4, undergo tubulo-genesis. Inset in Upper shows GFP expression after 24 h culture. Lower shows lotus lectin staining (red) with DAPI counterstain (blue), after 14 d of culture.One of three independent replicates is shown for each condition. (U) Semiquantitative PCR demonstrating increased Wnt9b and Wnt4 expression in pro-genitors transfected with Wnt9b and Wnt4 constructs, respectively. (V) Model for compartmentalization of the cap mesenchyme.

4644 | www.pnas.org/cgi/doi/10.1073/pnas.1213971110 Brown et al.

Dow

nloa

ded

by g

uest

on

July

26,

202

0

SMAD signaling and transitioned to the SIX2-only compartment(Fig. 5V). An intriguing molecular distinction between early pro-genitor compartments has previously been observed: complexes ofSIX2 and β-catenin can be isolated from SIX2+ but not CITED1+

cells, suggesting significant differences in canonical WNT-inducedtranscriptional responses between SIX2-only and CITED1+

compartments (27). Within the SIX2-only compartment progeni-tors become susceptible to WNT9b/β-catenin induction, leading toexpression of LEF1 and WNT4. WNT4 then promotes epitheli-alization of the LEF1+ pretubular aggregate through autocrinenoncanonical WNT signaling. This orderly progression through aseries of functionally distinct progenitor cell compartments ensuresthat the CITED1+ population is sequestered from induction andretained for future rounds of nephrogenesis. Further studies shouldbe performed to test critical aspects of this model such as how theSMAD versus TAK1-MAPK response to BMP7 is regulated at themolecular level and how the CITED1+ population is carried for-ward to the next collecting duct tip to seed a new nephron.

Materials and MethodsCell Culture. NZCswereextracted from20 to 24embryonic kidneys and culturedin monolayer in keratinocyte serum-free medium (Gibco) supplemented with50 ng/mL FGF2 (R&D Systems) as described (15–17). Cultures were treated with50ng/mLBMP7 (R&DSystems), 500 nMBIO (Calbiochem), 15mMLiCl (Sigma), or10 μM FH535 (Calbiochem). TAK1 inhibitor (AnalytiCon Discovery) concentra-tion was empirically determined as the amount necessary (50 nM) to reverseBMP7-induced proliferation of nephron progenitors. Dorsomorphin (Sigma)concentration (2.5 μM)wasdeterminedas the amount necessary to blockBMP7-inducedphosphorylationof SMAD1/5 innephronprogenitors. Three-dimensionalaggregates containing 250,000 progenitors were spotted from single-cellsuspensions in a 1-μL volume on floating 0.1-μm pore-size VCTP membranefilters (Millipore) and cultured for 5 d in DMEM/F12 supplemented with FGF2(200 ng/mL), 5% (vol/vol) KnockOut Serum Replacement (Invitrogen) andadditional factors as described (26). For CITED1+ progenitor transfectionexperiments, freshly purified cells were incubated at 37 °C in BSA-coated1.5-mL Eppendorf tubes on a nutator for 1.5 h. One hundred microliters ofprepared lipofectamine/DNA reagent (4 μg of DNA to 2 μL of lipofectamine)was added to 1 mL of cells in DMEM/F12 medium. Cells were spun at 300 × gand resuspended in DMEM/F12 for 3D aggregate spotting.

Immunofluorescence and Microscopy. Cells and sections were immunostainedas described (15). Primary antibodies were incubated overnight at 4 °C:CITED1 1:100 (NeoMarkers); LEF1 1:100 (Cell Signaling Technology); PAX2

1:100 (Invitrogen); phistone-H3 1:100; pSMAD1/5 1:100 (both Cell SignalingTechnology); SIX2 1:100 (Santa Cruz Biotechnology); DBA lectin 1:200 (VectorLaboratories); LT lectin 1:200 (Vector Laboratories) and E-cadherin 1:100 (BDTransduction Laboratories). PAX2 and pSMAD1/5 or CITED1 and LEF1 werecostained by converting rabbit pSMAD1/5 or LEF1 antibodies to goat by usinganti-goat FAB fragment according to themanufacturer’s instructions (JacksonImmunoResearch) followed by 5-min fixation with 4% (wt/vol) para-formaldehyde, before incubation with PAX2 or CITED1 antibodies for 24 h at4 °C. Immunofluorescent images were quantified by using ImageJ, and per-centages were normalized to the total number of DAPI+ cells per field.P values were calculated by using a two-tailed homoscedastic Student t test.

Quantitative PCR. RNA purification, cDNA synthesis, and quantitative PCRwere performed as described (16). Average values (±SD) of three technicalreplicates from NZCs of 20–24 pooled embryonic kidneys are shown. P valueswere calculated by using a heteroschedastic two-tailed Student t test withP < 0.05 considered significant.

Magnetic Bead Depletion and Flow Cytometry. Total NZCs were isolated andCITED1+ cells were enriched by negative depletion with magnetic activatedcell sorting (MACS)-phycoerythrin (PE)–conjugated antibodies (Fig. S5A) andanti-PE MicroBeads by following the manufacturer’s protocol (MiltenyiBiotec). Cells were purified by passing NZCs twice through an autoMACSseparator by using the “Depletes” program setting. Total and CITED1enriched cell populations were stained by using PE-, allophycocyanin-, andfluorescein-conjugated antibodies, collected on a FACSCalibur (BD), anddata were analyzed by using FlowJo software.

Mouse Strains. Animal care was performed in accordance with the National Re-searchCouncilGuide for theCareandUseofLaboratoryAnimals.Animalprotocolswere approved by the Institutional Animal Care and Use Committee of MaineMedical Center. Institute for Cancer Research stock was used for all NZC harvests,and Bmp7−/− was maintained on an Institute for Cancer Research background.

ACKNOWLEDGMENTS. We thank Barry Larman for insightful discussionson the work. This work was supported by the National Institutes of Diabetesand Digestive and Kidney Disease (NIDDK) Grant R01DK078161 (to L.O.)and American Recovery and Reinvestment Act-supported supplement GrantR01DK078161 (to L.O.). A.C.B. was supported by a postdoctoral fellowshipfrom the American Heart Association and J.A.G. was supported by a pre-doctoral fellowship from the American Heart Association. J.L.F. was supportedby NIDDK Fellowship F32DK093195. Core facilities support was provided byMaine Medical Center Research Institute core facilities for Molecular Pheno-typing and Stem and Progenitor Cell Analysis [both supported by NationalInstitutes of General Medicine (NIGM) Grant 8P20 GM103465], and Histopa-thology by NIGM Grants 8P20 GM103465 and 8P30 GM103392.

1. Grobstein C (1953) Inductive epitheliomesenchymal interaction in cultured organrudiments of the mouse. Science 118(3054):52–55.

2. Grobstein C (1956) Trans-filter induction of tubules in mouse metanephrogenicmesenchyme. Exp Cell Res 10(2):424–440.

3. Grobstein C (1957) Some transmission characteristics of the tubule-inducing influenceon mouse metanephrogenic mesenchyme. Exp Cell Res 13(3):575–587.

4. McMahon AP, et al.; GUDMAP project (2008) GUDMAP: The genitourinary develop-mental molecular anatomy project. J Am Soc Nephrol 19(4):667–671.

5. Mugford JW, Yu J, Kobayashi A, McMahon AP (2009) High-resolution gene expressionanalysis of the developing mouse kidney defines novel cellular compartments withinthe nephron progenitor population. Dev Biol 333(2):312–323.

6. Oxburgh L, Chu GC, Michael SK, Robertson EJ (2004) TGFbeta superfamily signals arerequired for morphogenesis of the kidney mesenchyme progenitor population. De-velopment 131(18):4593–4605.

7. Kobayashi A, et al. (2008) Six2 defines and regulates a multipotent self-renewingnephron progenitor population throughout mammalian kidney development. CellStem Cell 3(2):169–181.

8. Boyle S, et al. (2008) Fate mapping using Cited1-CreERT2 mice demonstrates that thecap mesenchyme contains self-renewing progenitor cells and gives rise exclusively tonephronic epithelia. Dev Biol 313(1):234–245.

9. Luo G, et al. (1995) BMP-7 is an inducer of nephrogenesis, and is also required for eyedevelopment and skeletal patterning. Genes Dev 9(22):2808–2820.

10. DudleyAT, Lyons KM, Robertson EJ (1995) A requirement for bonemorphogenetic protein-7 during development of the mammalian kidney and eye. Genes Dev 9(22):2795–2807.

11. Poladia DP, et al. (2006) Role of fibroblast growth factor receptors 1 and 2 in themetanephric mesenchyme. Dev Biol 291(2):325–339.

12. Carroll TJ, Park JS, Hayashi S, Majumdar A, McMahon AP (2005) Wnt9b plays a centralrole in the regulation of mesenchymal to epithelial transitions underlying organo-genesis of the mammalian urogenital system. Dev Cell 9(2):283–292.

13. Karner CM, et al. (2011) Canonical Wnt9b signaling balances progenitor cell expan-sion and differentiation during kidney development. Development 138(7):1247–1257.

14. Self M, et al. (2006) Six2 is required for suppression of nephrogenesis and progenitorrenewal in the developing kidney. EMBO J 25(21):5214–5228.

15. Blank U, Brown A, Adams DC, Karolak MJ, Oxburgh L (2009) BMP7 promotes pro-liferation of nephron progenitor cells via a JNK-dependent mechanism. Development136(21):3557–3566.

16. Brown AC, et al. (2011) FGF/EGF signaling regulates the renewal of early nephronprogenitors during embryonic development. Development 138(23):5099–5112.

17. Brown AC, et al. (2011) Isolation and culture of cells from the nephrogenic zone ofthe embryonic mouse kidney. J Vis Exp 2011(50):2555.

18. Davies JA, Garrod DR (1995) Induction of early stages of kidney tubule differentiationby lithium ions. Dev Biol 167(1):50–60.

19. Yu PB, et al. (2008) Dorsomorphin inhibits BMP signals required for embryogenesisand iron metabolism. Nat Chem Biol 4(1):33–41.

20. Ninomiya-Tsuji J, et al. (2003) A resorcylic acid lactone, 5Z-7-oxozeaenol, preventsinflammation by inhibiting the catalytic activity of TAK1 MAPK kinase kinase. J BiolChem 278(20):18485–18490.

21. Ikeya M, et al. (2010) Cv2, functioning as a pro-BMP factor via twisted gastrulation, isrequired for early development of nephron precursors. Dev Biol 337(2):405–414.

22. Park JS, ValeriusMT, McMahon AP (2007) Wnt/beta-catenin signaling regulates nephroninduction during mouse kidney development. Development 134(13):2533–2539.

23. Kispert A, Vainio S, McMahon AP (1998) Wnt-4 is a mesenchymal signal for epithelialtransformation of metanephric mesenchyme in the developing kidney. Development125(21):4225–4234.

24. Tanigawa S, et al. (2011) Wnt4 induces nephronic tubules in metanephric mesen-chyme by a non-canonical mechanism. Dev Biol 352(1):58–69.

25. Barak H, et al. (2012) FGF9 and FGF20 maintain the stemness of nephron progenitorsin mice and man. Dev Cell 22(6):1191–1207.

26. Osafune K, Takasato M, Kispert A, AsashimaM, Nishinakamura R (2006) Identificationof multipotent progenitors in the embryonic mouse kidney by a novel colony-formingassay. Development 133(1):151–161.

27. Park JS, et al. (2012) Six2 and Wnt regulate self-renewal and commitment of nephronprogenitors through shared gene regulatory networks. Dev Cell 23(3):637–651.

Brown et al. PNAS | March 19, 2013 | vol. 110 | no. 12 | 4645

DEV

ELOPM

ENTA

LBIOLO

GY

Dow

nloa

ded

by g

uest

on

July

26,

202

0