The Wilms tumor gene, Wt1, is required for Sox9expression and maintenance of tubular architecturein the developing testisFei Gao*, Sourindra Maiti*, Nargis Alam*, Zhen Zhang†, Jian Min Deng‡, Richard R. Behringer‡, Charlotte Lecureuil§,Florian Guillou§, and Vicki Huff*¶

Departments of *Molecular Genetics�Cancer Genetics, †Experimental Radiation Oncology, and ‡Molecular Genetics, University of Texas M. D. AndersonCancer Center, 1515 Holcombe Boulevard, Houston, TX 77030; and §Unite Mixte de Recherche 6175, Physiologie de la Reproduction, Institut National de laRecherche Agronomique, Centre National de la Recherche Scientifique, Universite de Tours, Haras Nationaux, 37380 Nouzilly, France

Edited by Jordan Kreidberg, Children’s Hospital, Boston, MA, and accepted by the Editorial Board June 16, 2006 (received for review February 9, 2006)

Mutation of the transcription factor and tumor suppressor geneWT1 results in a range of genitourinary anomalies in humans,including 46,XY gonadal dysgenesis, indicating that WT1 plays acritical role in sex determination. However, because knockout ofWt1 in mice results in apoptosis of the genital ridge, it is unknownwhether WT1 is required for testis development after the initialsteps of sex determination. To address this question, we generateda mouse strain carrying a Wt1 conditional knockout allele andablated Wt1 function specifically in Sertoli cells by embryonic day14.5, several days after testis determination. Wt1 knockout re-sulted in disruption of developing seminiferous tubules and sub-sequent progressive loss of Sertoli cells and germ cells such thatpostnatal mutant testes were almost completely devoid of thesecell types and were severely hypoplastic. Thus, Wt1 is essential forthe maintenance of Sertoli cells and seminiferous tubules in thedeveloping testes. Of particular note, expression of the testis-determining gene Sox9 in mutant Sertoli cells was turned off atembryonic day 14.5 after Wt1 ablation, suggesting that WT1regulates Sox9, either directly or indirectly, after Sry expressionceases. Our data, along with previous work demonstrating the roleof Wt1 at early stages of gonadal development, thus indicate thatWt1 is essential at multiple steps in testicular development.

Sertoli cell � testicular cord � Mullerian duct � Amh � Sox8

Mammalian testis determination is initiated by the expressionof SRY. This Y-linked gene encodes a DNA-binding protein

that is thought to up-regulate genes critical for the commitment ofsomatic cells of the genital ridge to become Sertoli cells. The directtargets of SRY remain elusive, although both in vivo and in vitrodata suggest that one may be Sox9 (1–3), which itself plays a criticalrole in testis determination (4–9). Similarly, an important role forWT1 in genital tract development was recognized based on theobservation of sex reversal or genital tract anomalies in XY patientswith heterozygous germ-line WT1 mutations� (10) ranging fromentire gene deletions to missense mutations. Variable expressivity,even among patients with a similar type of mutation, has beenobserved (11), suggesting a model whereby a critical threshold levelof wild-type WT1 protein is required for normal testis determina-tion and development, and stochastic variability in protein expres-sion results in phenotypic variability.

In mice, Wt1 is required for the survival and proliferation of cellsin the genital ridge. Although present at embryonic day (E) 10.5,the genital ridge fails to thicken in Wt1�/� animals, and no gonadsare detectable by E14 (12). Commitment of the murine XY gonadto a testicular phenotype is first detectable at E12.5. Because of thecomplete gonadal agenesis exhibited by Wt1�/� mice and theobservation of no testicular anomalies in Wt1�/� mice (12), assess-ing the role of Wt1 in testis determination and subsequent devel-opment has been difficult. To date, in vivo assessment of this rolehas come primarily from studies of mice expressing only one of thetwo Wt1 ‘‘kts’’ isoforms (13). These studies indicated that Wt1 plays

a role in testis determination, perhaps by up-regulating Sry, but afurther role for Wt1 in testicular development independent of Srycould not be assessed by this model. Consistent with the previousobservation of apoptosis of the genital ridge in Wt1�/� animals,hypoplastic testes with dysgenic tubules containing no Wt1�/�

Sertoli cells are observed in �40% of highly chimeric Wt1�/�7XYadult males (14).

The continued expression of Wt1 in the mature testes suggests arole for Wt1 at later stages of testicular development, as have in vitrostudies implicating Wt1 in the up-regulation of anti-Mullerianhormone, Amh, (also known as Mullerian inhibiting substance,Mis) in the differentiated testes (15). However, in vivo assessmentof the possible role of Wt1 in the XY gonad after testis determi-nation has been problematic because of apoptosis of the Wt1�/�

genital ridge. To circumvent this problem, we generated a mousestrain (Wt1flox) carrying a Wt1 conditional knockout allele. Crossingit with the Wt1� strain (12) and the AMH-Cre transgenic strain,which expresses Cre in Sertoli cells soon after testis commitment(16), we generated Wt1�/flox; AMH-Cre males in which we couldablate Wt1 function in Sertoli cells and determine its role in thecommitted testes.

ResultsGeneration of Wt1 Conditional Knockout (Wt1flox) Mouse Strain. Byhomologous recombination a Wt1 allele in which exons 8 and 9 areflanked by loxP sites (Fig. 1 A–D) was introduced into the mousegerm line. These two exons encode the second and third of the fourDNA-binding zinc-finger domains of Wt1. Missense and truncatingmutations in exons 8 and 9 are observed in patients, and the criticalfunctional importance of the two domains is further underscored byin vitro studies (15, 17, 18). Expression of Cre recombinase resultsin the in-frame deletion of exons 8 and 9 (Fig. 1E) and generatesan allele (Wt1�) encoding a shortened Wt1 protein lacking zincfingers 2 and 3. Germ-line generation of the Wt1� allele wasachieved by using the oocyte-specific Zp3-cre transgene. Consistentwith the expectation that the Wt1� allele acts as a null allele, Wt1�/�

embryos from Wt1�/� � Wt1�/� crosses died during embryogenesisat the same developmental stage (E13–E15) as do Wt1�/� embryos.Additionally, Wt1�/� animals were viable and phenotypically nor-mal, indicating that the Wt1� shortened protein does not act in adominant negative manner (data not shown).

Conflict of interest statement: No conflicts declared.

This paper was submitted directly (Track II) to the PNAS office. J.K. is a guest editor invitedby the Editorial Board.

Abbreviations: IHC, immunohistochemistry; En, embryonic day n; Amh, anti-Mullerianhormone; Pn, postnatal day n; GCNA-1, germ cell nuclear antigen 1.

¶To whom correspondence should be addressed. E-mail: vhuff@mdacc.tmc.edu.

�Huff, V., Villalba, F., Strong, L. C. & Saunder, G. F. (1991) Am. J. Hum. Genet. 49, 44 (Abstr.).

© 2006 by The National Academy of Sciences of the USA

www.pnas.org�cgi�doi�10.1073�pnas.0600994103 PNAS � August 8, 2006 � vol. 103 � no. 32 � 11987–11992

DEV

ELO

PMEN

TAL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Oct

ober

9, 2

020

Generation of Wt1�/flox; AMH-Cre Mice and Validation of ConditionalMutation System. To achieve somatic ablation of Wt1 in testes, weintroduced the Sertoli cell-expressed AMH-Cre transgene and theWt1-null allele into the Wt1flox strain to obtain Wt1�/flox; AMH-Cremales. By immunohistochemistry (IHC), we detected Cre expres-sion specifically in Sertoli cells at E14.5 but not at E13.5 (Fig. 2A).This finding is consistent with the spatio-temporal expressionpattern previously reported for the AMH-Cre strain (16); expressionof the human AMH promoter-driven transgene occurs 1–2 dayslater than the endogenous Amh gene. Cre-mediated excision of Wt1exons 8 and 9 was detectable at E14.5 in testes, but not at E13.5 (Fig.2B), consistent with the Cre expression pattern.

Viability and Gross Morphology of Wt1�/flox; AMH-Cre Males. Wt1�/flox;AMH-Cre animals were fully viable. No gross abnormalities ofexternal genitalia were observed in 7-week-old Wt1�/flox; AMH-Cremales (Fig. 3Aa), but testes size was only �10% of that of controllittermates (Fig. 3Ab). The rest of the reproductive tract was alsohypoplastic in the mutants (Fig. 3 Ac and Ad). Surprisingly, upongross examination of Wt1�/flox; AMH-Cre males, they were found tohave a uterus in addition to a vas deferens, epididymis, and seminalvesicles (Fig. 3Ad). Histologic analysis of the proximal portion of thereproductive tract from mutant males revealed structures with thecoiled morphology and folded epithelium characteristic of oviduct(Fig. 7, which is published as supporting information on the PNASweb site). Thus, the normal regression of the Mullerian duct systemwas deficient in the Wt1�/flox; AMH-Cre males, whereas Wolffianduct differentiation was largely normal.

Aberrant Histology of Wt1�/flox; AMH-Cre Testes. Histologically,Wt1�/flox; AMH-Cre testes from 7-week-old animals bore no re-semblance to age-matched control testes. Mutant testes completelylacked the normal tubular architecture observed in control testes(Fig. 3 Ba and Bb) and consisted primarily of sheets of eosinophiliccells (Fig. 3 Bd and Be) which, by IHC analysis with anti-3�-HSD,were identified as Leydig cells (Fig. 3Bf). Normally, these cells arepresent in small numbers in the testicular interstitium (Fig. 3 Bc),

but, in the Wt1�/flox; AMH-Cre testes, in the absence of tubules, theywere observed in dense clusters interspersed with regions offibroblast-like cells. Thus, Cre-mediated deletion of Wt1 at E14.5in Sertoli cells had a profound effect on testis development,

Fig. 2. Expression of Cre-recombinase and recombination of Wt1flox allele.(A) IHC demonstrating expression of Cre-recombinase specifically in Sertolicells of E14.5 Wt1�/flox; AMH-Cre testes (arrows) but not in mutant testes atE13.5 or in controls (Wt1�/flox). (B) Detection of recombined Wt1� allelespecifically in testes of E14.5 Wt1�/flox; AMH-Cre embryos.

Fig. 1. Generation of the Wt1flox mouse strain. (A)Scheme for generating animals carrying the conditionalknockout allele (Wt1flox) and the recombined allele(Wt1�). (B) Southern blot of EcoRV restricted ES cell DNAprobed with 5� probe yielded a 7.14-kb fragment in re-combinant animals in addition to the wild-type 5.34-kbfragment, confirming insertion of the targeting constructinto the Wt1 locus. (C) PCR genotyping of wild-type andWt1flox allele using primers loxpF and loxpR. (D) PCR geno-typing of wild-type and Wt1flox allele using primers 1.75and 1.55. (E) PCR detection of recombined Wt1� alleleusing primers ckodelF and 1.55.

11988 � www.pnas.org�cgi�doi�10.1073�pnas.0600994103 Gao et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

9, 2

020

resulting in testes greatly reduced in size and composed almostentirely of cells normally found in the interstitium between tubules.Given the dramatically aberrant testicular histology we observed,Wt1�/flox; AMH-Cre animals are predicted to be infertile, althoughthis has not rigorously been assessed.

Wt1 Deletion in Sertoli Cells Results in Testicular Cord Disruption. Todetermine the time course of this aberrant development, weassessed testes from Wt1�/flox; AMH-Cre males between E13.5 andpostnatal day 7 (P7), examining tissues from three to six animals ateach time point. There was a high degree of concordance withrespect to histology between mutant animals at the same age, andrepresentative sections are presented. For controls, we assessedWt1�/flox and Wt1�/flox; AMH-Cre and Wt1�/� littermates at eachtime point. No difference was observed between these three controlgroups, and the control data presented are from mice of any of thesethree genotypes.

No aberrant pathology was discernable in any of the fourWt1�/flox; AMH-Cre animals at E13.5. Nascent development of sexcords could be seen at low magnification (arrowheads, Fig. 4d), and,at higher magnification, well organized tubule structures (Fig. 4e)were clearly present in Wt1�/flox; AMH-Cre testes. The polyclonalWT1 antibody, sc-192, is a WT1-specific rabbit antibody raisedagainst a peptide corresponding to the 9 aa at the carboxyl terminusof WT1 (19). Therefore, it recognizes the mutant protein encodedby the Wt1� in-frame deletion, enabling us to employ this widelyused marker of Sertoli cells to assess Sertoli cell location, number,and morphology in the Wt1�/flox; AMH-Cre testes. By IHC analyses,the location and morphology of Sertoli cells in Wt1�/flox; AMH-Cretestes (Fig. 4f) at E13.5 appeared normal (Fig. 4c), consistent withthe lack of Cre-recombinase expression at this time point (Fig. 2A).

At E14.5, when Cre-recombinase was expressed (Fig. 2A),Wt1�/flox; AMH-Cre testes appeared normal in all five embryosexamined. Developing tubules were observed (arrowheads, Fig. 4j),and there was no salient alteration in architecture of the testes (Fig.4 g and j), although Sertoli cell nuclei in mutant testes were lessrounded than those in control testes (Fig. 4 i and l).

From E15.5 to P7, control testes exhibited progressive tubuledevelopment and organization with Leydig cells (lc) located be-tween tubules (Fig. 5 a–d). Over the same time period, Wt1�/flox;AMH-Cre testes increasingly lost normal tubular architecture (Fig.5 i and j). At E15.5, only a few tubule-like structures were observedat low magnification in Wt1�/flox; AMH-Cre testes (asterisks, Fig. 5i).Most of the testes consisted of disorganized germ cells (gc) andLeydig cells (lc), which could be identified histologically (Fig. 5k).Apparent remnants of tubules consisted primarily of germ cells withfew identifiable Sertoli cells (Fig. 5l). To identify positively Sertolicells and germ cells, anti-WT1 and anti-GCNA-1 (germ cell nuclearantigen 1) antibodies (19, 20) were used for IHC. In control testes,GCNA-1-positive germ cells were localized in the center of tubules(Fig. 5f), surrounded by well organized WT1-positive Sertoli cells(Fig. 5e). However, in mutant testes, most of the Sertoli cells (Fig.5o) and germ cells (Fig. 5s) were scattered and outside of anydiscernible tubular structure.

At P7, mutant testes consisted largely of fibroblast-like stromalcells and clusters of eosinophilic Leydig cells (Fig. 5 j and m), withonly a few aberrant tubule-like structures present (asterisks, Fig. 5j and n). No WT1-positive Sertoli cells were observed except in raretubule remnants (Fig. 5 q and r). These data suggest that, aftertubule disruption, mutant Sertoli cells eventually died or dediffer-entiated into other cell types. Similarly, the few germ cells remain-ing were primarily in tubule remnants (Fig. 5 u and v).

Wt1 Ablation Results in Loss of SOX8, SOX9, and AMH Expression inSertoli Cells. Because the Sox8, Sox9, and Amh genes are expressedin Sertoli cells, we examined how their expression was affected byWt1 inactivation. SOX9 and SOX8 proteins were detected in theSertoli cells of both mutant and control testes at E13.5, and nodifference was observed between three mutant males and threecontrol animals assessed by IHC (Fig. 6 a and b; d and e).

Fig. 3. Phenotype of 7-week-old Wt1�/flox; AMH-Cre males. (A) Normalexternal genitalia of mutant male (Aa), severely reduced size of testes fromthree different Wt1�/flox; AMH-Cre males (Ab), and reproductive tracts fromcontrol male and Wt1�/flox; AMH-Cre male displaying reduced size and thedevelopment of both a vas deferens and uterus in mutant (Ac and Ad). (B)Testis sections stained with H&E or with anti-3�-HSD antibody and showingnormal tubular architecture (asterisk) in control testes (Ba and Bb), lack oftubules in mutant testes (arrowhead) (Bd and Be), and presence of Leydig cells(lc) in both control (Bc) and Wt1�/flox; AMH-Cre (Bf ) testes. Ep, epididymis; t,testis; vd, vas deferens; sv, seminal vesicle; u, uterus.

Fig. 4. Mutant testes at E13.5 and E14.5. (d, e, j, and k) Normal testicularhistology with normal tubules (arrowheads, arrows) observed in mutant testes atlow and high magnification. (c, f, j, and l) WT1 immunostaining of Sertoli cells.

Gao et al. PNAS � August 8, 2006 � vol. 103 � no. 32 � 11989

DEV

ELO

PMEN

TAL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Oct

ober

9, 2

020

Unexpectedly, unlike the robust expression in control testes, SOX9and SOX8 expression was virtually absent in Wt1�/flox; AMH-Cretestes at E14.5 (Fig. 6 j and k), even though Sertoli cells could beidentified in mutant testes both morphologically (arrows, Fig. 6 jand k) and by WT1 IHC (Fig. 4l). This result was consistentlyobserved in the three mutant males assessed by IHC. These datademonstrate that SOX8 and SOX9 expression was down-regulatedin mutant Sertoli cells and suggest that WT1 is required for thecontinued expression of these genes. As expected, given the re-ported role of both SOX9 and WT1 in regulating Amh transcription(21, 22), AMH expression, normal at E13.5, was also absent inmutant testes by E14.5 (Fig. 6l).

DiscussionBecause Wt1 is required for survival of the genital ridge, in vivoexamination of its potential role at later stages of gonad develop-ment has been problematic. With the ability to somatically ablateWt1 using our newly generated Wt1flox strain, we have now been ableto assess the role of Wt1 after commitment of the indifferent gonad.The current study now provides conclusive in vivo evidence that Wt1is essential, even after testis determination, for normal testisstructure and function.

Wt1 Is Required for Maintenance of Testicular Architecture. Althoughmutant and control testes were indistinguishable morphologically atE13.5 before expression of the AMH-Cre transgene, after Wt1ablation, a dramatic and progressive loss of seminiferous tubulararchitecture and Sertoli cells and germ cells was observed

beginning at E15.5. By P7, mutant testes were composed ofLeydig cells and fibroblast-like cells, with only the occasionalaberrant tubule remnant containing a few GCNA1- and WT1-positive cells. We speculate that these rare aberrant tubules arethose in which Cre-mediated Wt1 inactivation did not occur in alarge enough proportion of Sertoli cells within a tubule to triggerits complete disruption.

Given our observation that initially well organized tubules in mu-tant testes progressively lose their normal structure after Cre-mediated Wt1 ablation, we speculate that at least one result of Wt1mutation is the dysregulation of genes that are critical for normalSertoli cell–cell contacts, for Sertoli cell–basal lamina contacts, orfor normal composition of the basal lamina, the components ofwhich are synthesized by both Sertoli cells and the peritubularmyoid cells (reviewed in ref. 23). Altered expression of these genescould be either a direct or indirect effect of losing the transcriptionalregulatory function of WT1. Alternatively, the gene expressiondysregulation resulting from Wt1 mutation may have set intomotion a cellular response that secondarily affected the ability ofSertoli cells to maintain appropriate cell–cell and cell–matrixcontacts.

We detected no increased apoptosis in mutant Sertoli cells (datanot shown), suggesting that tubule disruption is not a result ofSertoli programmed cell death. We did, however, observe a dra-matic decline in the number of Sertoli cells positive for the mitoticcell marker H3P beginning at E14.5 until, by E18.5, no H3P-positiveSertoli cells were observed (Fig. 8, which is published as supportinginformation on the PNAS web site). This cessation of Sertoli cellproliferation may be a result of altered cellular gene expressionbecause of Wt1 ablation or secondarily triggered by the loss ofnormal cellular milieu because of tubule disaggregation. Given theknown importance of cell–cell and cell–matrix interactions forSertoli cell function, we favor this latter scenario but cannot excludethe former.

Loss of Germ Cells After Tubule Disruption in Mutant Testes. GCNA1is a germ cell-specific marker (20). We observed no salient differ-

Fig. 5. Disruption of seminiferous tubule architecture and loss of Sertoli cellsand germ cells in Wt1�/flox; AMH-Cre testes between E15.5 and P7. (a–d)Progressive tubule development and organization in control testes from E15.5to P7 shown by H&E staining. (e–h) Germ cells (GCNA-1-positive) and Sertolicells (WT1-positive) located within tubules of control testes. (i and j) Progres-sive disruption of tubule architecture in Wt1�/flox; AMH-Cre testes (low mag-nification). (k and m) Predominant histology of E15.5 and P7 mutant testes(high magnification). (l and n) Histology of occasional aberrant tubules inE15.5 and P7 mutant testes (high magnification). (o–r) Sertoli cells in mutanttestes identified by WT1 IHC. (s–v) Germ cells in mutant testes identified byGCNA-1 IHC. sc, Sertoli cells; gc, germ cells; lc, Leydig cell.

Fig. 6. Loss of SOX9, SOX8, and AMH expression by E14.5 in Wt1�/flox;AMH-Cre Sertoli cells. Control testes show normal SOX9 (a and g), SOX8 (b andh), and AMH (c and i) expression in Sertoli cells at E13.5 and E14.5. SOX9 (d),SOX8 (e), and AMH ( f) expression was normal in E13.5 Wt1�/flox; AMH-Cretestes but was virtually absent in E14.5 testes, even in those Sertoli cellsassociated with tubules (asterisks, j–l). sc, Sertoli cells.

11990 � www.pnas.org�cgi�doi�10.1073�pnas.0600994103 Gao et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

9, 2

020

ence in the number of GCNA1-positive cells in mutant and controltestes at E15.5, but, as tubule structure disintegrated, there was aprogressive loss of germ cells. This loss paralleled the loss ofidentifiable Sertoli cells and is consistent with the known interde-pendence of Sertoli cells and germ cells. Wt1 expression has beenreported in E12 primordial germ cells (14). However, two lines ofevidence suggest that it is unlikely that this expression played a rolein the loss of germ cells in the Wt1�/flox; AMH-Cre testis. First, byIHC, we did not observe WT1-positive germ cells at E13.5 or later.Second, given the Sertoli cell-specific expression of Cre, we wouldnot expect Wt1 ablation in germ cells; they would not be expectedto display a phenotype. Therefore, we conclude that the loss of germcells in mutant testes is secondary to Sertoli cell loss and�or the lossof tubular architecture.

No Apparent Loss of Leydig Cells in Mutant Testes. IHC analysis usingthe Leydig cell marker, 3�-HSD, demonstrated an abundance ofLeydig cells in embryonic mutant testes (data not shown), consis-tent with the predominant population of Leydig cells in adultmutant testes. We hypothesize that the high proportion of Leydigcells in mutant testes is a function of the absence of germ cells andSertoli cells; with their loss, the testes have effectively collapsed,containing only the normal complement of Leydig and otherinterstitial cells.

Paracrine factors secreted by Sertoli cells are thought to beimportant for the development and function of Leydig cells (re-viewed in ref. 24). Of particular note, testosterone production inmature Leydig cells in vitro has been reported to be enhanced bycoculture with Sertoli cells (25). Therefore, our observations ofcontinued proliferation of Leydig cells in the mutant testes and ofserum testosterone levels within normal ranges in mutant animals(data not presented) were unexpected because they indicated thatLeydig cells in the mutant testes were viable and functional even inthe absence of Sertoli cells. These data suggest that Leydig cells atlater stages of testes development (e.g., approximately E14.5 andonward) do not require Sertoli cells for proliferation and synthesisof testosterone. Thus, unlike Sertoli cells and germ cells, which inthe absence of seminiferous tubules and appropriate cell–cellinteractions are unable to survive, Leydig cells, in their normalextratubular location, can survive and function.

Loss of SOX9 Expression in Wt1�/flox; AMH-Cre testes. Unexpectedly,expression of SOX9 protein in the mutant testes was virtuallysilenced at E14.5 even though Sertoli cells were clearly present inmorphologically intact tubules. Sox9 is normally expressed at lowlevels in the genital ridge of both XX and XY animals, and itsexpression is up-regulated in the male gonad at approximatelyE11.0, soon after Sry expression is detectable (2). Sry is down-regulated at approximately E12.5, but Sox9 expression continues inthe Sertoli cells of the committed testis (1, 2, 26). These data suggestthat Sox9 is up-regulated, directly or indirectly, by SRY, but that itscontinued expression is independent of SRY.

Gonads from mice expressing only the ‘‘�kts’’ isoform of Wt1display reduced Sry expression and complete sex reversal, suggest-ing that WT1 activates Sry (14). The loss of SOX9 expression inWt1�/flox; AMH-Cre testes, however, occurred approximately 2 daysafter Sry is normally down-regulated. Thus, we conclude that thelack of SOX9 expression at E14.5 is independent of WT1’s putativefunction in up-regulating Sry and that WT1 is required for thecontinued expression of Sox9 in the committed testis. The identi-fication of an evolutionarily conserved region 5� of Sox9 thatcontains a putative WT1-binding site supports the notion that WT1is directly involved in the regulation of Sox9 expression (27), butfuture studies will be required to demonstrate a direct role of WT1in Sox9 regulation. Regardless of whether it is a direct or indirecteffect, our data provide an indication that WT1 is required for thecontinued expression of Sox9.

Loss of AMH Expression and Development of Mullerian Ducts inMutant Males. Mullerian duct regression in XY animals is initiatedby AMH and proceeds in a cranial to caudal manner. The Amhgene is normally expressed beginning at approximately E11.5 in themouse (28), but the Mullerian duct is sensitive to its effects onlyduring a narrow window of development (E13.5–E14.5) duringwhich apoptosis of the ductal epithelium, morphologic changes inperiductal mesenchyme, and sexually dimorphic expression ofLhx1, a gene required for Mullerian duct differentiation, begin tobe observed (29–31). Although normal at E13.5, Sertoli cellexpression of AMH expression in mutant males was negligible atE14.5. The presence of uteri and, more proximally, coiled structureswith morphological and histological features of oviducts in mutantmales indicates a failure of both cranial (oviducts) and more caudal(uterus) regression of the Mullerian duct. These data suggest thatthe window of time during which AMH can effect Mullerian ductregression is narrower than previously thought.

In vitro and in vivo studies have identified a functional SOX9-binding site in the proximal promoter of Amh that is required forAmh expression. Steroidogenic factor 1 (NR5A1, also known asSF1) is thought to enhance this interaction, and NR5A1�SF1functional binding sites have been identified within the Amhpromoter (22, 32, 33). In vitro studies have suggested that WT1synergizes with NR5A1�SF1 to activate the Amh promoter, but themechanism for this activation is not clear. WT1 has also beenreported to positively regulate the Nr5a1�Sf1 gene (34), butNR5A1�SF1 expression was not noticeably reduced in the threemutant E15.5 testes we assessed by IHC (Fig. 9, which is publishedas supporting information on the PNAS web site), suggesting thatWT1 is not critical for NR5A1�SF1 expression in Sertoli cells at thisstage of development. Interestingly, missense mutations in Wt1exon 9, one of the exons deleted from the Wt1� allele, abolish theability of WT1 to activate Amh expression (15). Therefore, the lackof AMH expression we observed in mutant testes may be due to lossof SOX9 expression, loss of WT1 function, or both.

Somatic Mutation of Wt1 in the Committed Testis Results in a MoreSevere Phenotype than Somatic Mutation of Sox9. Our data suggesta model whereby Wt1 ablation between E13.5 and E14.5 results inthe loss of SOX9 expression, which in turn results in loss of AMHexpression and lack of Mullerian duct regression. However, inter-estingly, the phenotype of Wt1�/flox; AMH-Cre males was consid-erably more profoundly aberrant than that observed in Sox9flox/flox;Sf1-Cre males in which Sox9 itself was ablated in the male gonad (7),suggesting that WT1 has an additional critical function(s) in main-taining normal testicular structure�function that is independent ofits role in Sox9 regulation.

Another Sox Group E member, SOX8, is expressed in testes andhas also been implicated in the regulation of Amh (35, 36), andanalyses of single and compound Sox8 and Sox9 mutant mice havesuggested a coordinated expression and possible functional redun-dancy of SOX8 and SOX9 in the early XY gonad (7). We observedextinguished SOX8 expression at E14.5 in mutant testes, butwhether this was due to loss of SOX9 expression or a direct orindirect result of Wt1 mutation is not known. The Wt1�/flox;AMH-Cre testes phenotype was considerably more pronounced andpenetrant than that of Sox9flox/flox; Sox8�/�; Sf1-Cre males in whichtestes were observed, further suggesting that Wt1 has a critical rolein maintaining testes structure and function independent of Sox9and Sox8. Analysis of Sox9flox/flox; Sox8�/� males carrying the samerobustly expressing AMH-Cre transgene as used in the studiespresented here will be required to address this further.

In summary, our work expands the role of Wt1 in testiculardevelopment and demonstrates that WT1 is required for SOX9expression in the testis, independently of SRY, and also that it isessential for the maintenance of Sertoli cells and seminiferoustubules in the developing testis. Thus, WT1 plays essential roles atmultiple steps in testis development in the mouse and, likely, given

Gao et al. PNAS � August 8, 2006 � vol. 103 � no. 32 � 11991

DEV

ELO

PMEN

TAL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Oct

ober

9, 2

020

its high degree of sequence conservation (37), in a wide range ofvertebrates.

Materials and MethodsGeneration of a Wt1 Conditional Knockout Mouse Strain. All animalwork was carried out in accordance with institutional animal careand use committee (IACUC) regulations. A targeting vector inwhich Wt1 exons 8 and 9 were flanked by a loxP site and aloxP-neo-loxP cassette (Fig. 1A) was constructed and introducedinto mouse embryonic stem cells (AB1, 129�SvEv) by electropo-ration. Resultant G418-resistant clones were genotyped by South-ern blotting; introduction of the targeting construct into the Wt1locus by means of homologous recombination resulted in thegeneration of a 7.14-kb EcoRV fragment (Fig. 1B) detected withthe 5� probe indicated in Fig. 1A. Exons 7, 8, 9, and 10 and flankingintronic sequence were amplified from several clones, and PCRproducts were sequenced to confirm that intact loxP sites had beenintroduced into the locus and that no other sequence alterations hadoccurred. Clones were then injected into C57BL�6 (B6) blastocysts.Resultant male chimeras were identified by coat color and matedwith wild-type females. Tail biopsies of agouti-pigmented F1 ani-mals were genotyped by using a primer set specific to the neocassette. Animals carrying the flox-neo allele were then mated withCMV-Cre animals (22). Progeny in which the neo cassette had beenexcised by means of Cre-mediated recombination were identified byusing primers (1.75 and 1.55) flanking the 3� loxP site (Fig. 1A).

Generation of Wt1�/flox; AMH-Cre Mice. Wt1�/flox mice were matedwith mice carrying the Wt1-null allele (Wt1�/�) (12) and theAMH-Cre transgene mice (16).

Genotyping. DNA isolated from tail biopsies (postnatal) or fromlimbs (embryos) was used for genotyping. The presence of theWt1f lox allele was determined by PCR amplification usingthe loxpF�loxpR (Fig. 1C) and 1.75�1.55 (Fig. 1D) primer sets. Therecombined Wt1flox allele (Wt1�) was detected by PCR amplifica-tion using the ckodelF and 1.55 primers (Fig. 1E). Genotyping forthe Wt1-null allele and the AMH-Cre transgene was carried out asdescribed (12, 16).

Tissue Collection and Histologic Analysis. Testes and reproductivetracts were dissected immediately after euthanasia. Testes were

dissected from four mutant male embryos at E13.5, five at E14.5,five at E15.5, and six at E16.5. Tissues were collected from threemutant males at each E17.5, E18.5, and P7 time point and fromfive 7-week-old mutant males. Tissues from three to four controllittermates were also collected at each time point. Tissues werefixed in 4% paraformaldehyde for up to 24 h, stored in 70% ethanol,and embedded in paraffin. Four-micrometer-thick sections were cutand mounted on glass slides. After deparaffinization, slides werestained with H&E for histologic analyses. Slides were examinedfrom all mutant embryos�animals collected.

Immunohistochemical Analysis. IHC analysis of tissues from at leastthree mutant males at each time point was performed by using aVectastain ABC (avidin–biotin–peroxidase) kit (Vector Labora-tories, Burlingame, CA) as recommended. Antibodies to WT1(sc-192) and AMH (sc-6886) were purchased from Santa CruzBiotechnology (Santa Cruz, CA). Anti-SOX9 antibody (AB5535)was purchased from Chemicon (Temecula, CA), and antibody toCRE (69050) was obtained from Novagen (La Jolla, CA). Theanti-PCNA-1 antibody was generously given to us by G. Enders(University of Kansas Medical Center, Kansas City, KS) (20); theanti-3�-HSD antibody was a kind gift from C. Richard Parker(University of Alabama at Birmingham) (38), and the anti-SOX8antibody was generously provided by M. Wegner (University ofErlangen, Erlangen, Germany) (39). The IHC procedure was asdescribed (40). Stained slides were examined with a Leica DMREpifluorescence Microscope, and images were captured by aHamamatsu CCD camera.

Note Added in Proof. Rao et al (41) have observed reduced fertilityaccompanied by loss of germ cells and aberrant Sertoli cell morphologywhen Wt1 is down-regulated postnatally.

We thank Allan Bradley (Wellcome Trust Sanger Institute, Cambridge,U.K.) for ABI ES and SNL 76�7 STO cells; George Enders, RichardParker, and Dr. Michael Wegner for antibodies; and Marvin Meistrichfor helpful discussions. This work was supported by National Institutesof Health (NIH) Grants CA78257, CA34936, and HD30284. Blastocystinjection, veterinary resources, and DNA sequencing were partiallysupported by NIH Cancer Center Support (Core) Grant CA16672. Thiswork was also supported in part by the Odyssey Program and theTheodore N. Law Award for Scientific Achievement at M. D. AndersonCancer Center.

1. Kent, J., Wheatley, S. C., Andrews, J. E., Sinclair, A. H. & Koopman, P. (1996) Development(Cambridge, U.K.) 122, 2813–2822.

2. Morais da Silva, S., Hacker, A., Harley, V., Goodfellow, P., Swain, A. & Lovell-Badge, R.(1996) Nat. Genet. 14, 62–68.

3. Sekido, R., Bar, I., Narvaez, V., Penny, G. & Lovell-Badge, R. (2004) Dev. Biol. 274, 271–279.4. Wagner, T., Wirth, J., Meyer, J., Zabel, B., Held, M., Zimmer, J., Pasantes, J., Bricarelli,

F. D., Keutel, J., Hustert, E., et al. (1994) Cell 79, 1111–1120.5. Foster, J. W., Dominguez-Steglich, M. A., Guioli, S., Kowk, G., Weller, P. A., Stevanovic, M.,

Weissenbach, J., Mansour, S., Young, I. D., Goodfellow, P. N., et al. (1994) Nature 372, 525–530.6. Bishop, C. E., Whitworth, D. J., Qin, Y., Agoulnik, A. I., Agoulnik, I. U., Harrison, W. R.,

Behringer, R. R. & Overbeek, P. A. (2000) Nat. Genet. 26, 490–494.7. Chaboissier, M. C., Kobayashi, A., Vidal, V. I., Lutzkendorf, S., van de Kant, H. J., Wegner,

M., de Rooij, D. G., Behringer, R. R. & Schedl, A. (2004) Development (Cambridge, U.K.)131, 1891–1901.

8. Barrionuevo, F., Bagheri-Fam, S., Klattig, J., Kist, R., Taketo, M. M., Englert, C. & Scherer,G. (2006) Biol. Reprod. 74, 195–201.

9. Vidal, V. P., Chaboissier, M. C., de Rooij, D. G. & Schedl, A. (2001) Nat. Genet. 28, 216–217.10. Pelletier, J., Bruening, W., Kashtan, C. E., Mauer, S. M., Manivel, J. C., Striegel, J. E.,

Houghton, D. C., Junien, C., Habib, R., Fouser, L., et al. (1991) Cell 67, 437–447.11. Schumacher, V., Scharer, K., Wuhl, E., Altrogge, H., Bonzel, K.-E., Guschmann, M., Neuhaus,

T. J., Pollastro, R. M., Kuwertz-Broking, E., Bulla, M., et al. (1998) Kidney Int. 53, 1594–1600.12. Kreidberg, J. A., Sariola, H., Loring, J. M., Maeda, M., Pelletier, J., Housman, D. & Jaenisch,

R. (1993) Cell 74, 679–691.13. Hammes, A., Guo, J. K., Lutsch, G., Leheste, J. R., Landrock, D., Ziegler, U., Gubler, M. C.

& Schedl, A. (2001) Cell 106, 319–329.14. Natoli, T. A., Alberta, J. A., Bortvin, A., Taglienti, M. E., Menke, D. B., Loring, J., Jaenisch,

R., Page, D. C., Housman, D. E. & Kreidberg, J. A. (2004) Dev. Biol. 268, 429–440.15. Nachtigal, M. W., Hirokawa, Y., Enyeart-VanHouten, D. L., Flanagan, J. N., Hammer, G. D.

& Ingraham, H. A. (1998) Cell 93, 445–454.16. Lecureuil, C., Fontaine, I., Crepieux, P. & Guillou, F. (2002) Genesis 33, 114–118.17. Royer-Pokora, B., Beier, M., Henzler, M., Alam, R., Schumacher, V., Weirich, A. & Huff,

V. (2004) Am. J. Med. Genet. A 127, 249–257.18. Hossain, A. & Saunders, G. F. (2001) J. Biol. Chem. 276, 16817–16823.19. Silberstein, G. B., Van Horn, K., Strickland, P., Roberts, C. T., Jr., & Daniel, C. W. (1997)

Proc. Natl. Acad. Sci. USA 94, 8132–8137.

20. Enders, G. C. & May, J. J., II. (1994) Dev. Biol. 163, 331–340.21. De Santa Barbara, P., Bonneaud, N., Boizet, B., Desclozeaux, M., Moniot, B., Sudbeck, P.,

Scherer, G., Poulat, F. & Berta, P. (1998) Mol. Cell. Biol. 18, 6653–6665.22. Arango, N. A., Lovell-Badge, R. & Behringer, R. R. (1999) Cell 99, 409–419.23. Skinner, M. K. (2005) in Sertoli Cell Biology, ed. Skinner, M. K. & Griswold, M. D. (Elsevier,

New York), pp. 317–328.24. Skinner, M. K. (1991) Endocr. Rev. 12, 45–77.25. Lejeune, H., Sanchez, P. & Saez, J. M. (1998) Int. J. Androl. 21, 129–140.26. Wilhelm, D., Martinson, F., Bradford, S., Wilson, M. J., Combes, A. N., Beverdam, A.,

Bowles, J., Mizusaki, H. & Koopman, P. (2005) Dev. Biol. 287, 111–124.27. Bagheri-Fam, S., Ferraz, C., Demaille, J., Scherer, G. & Pfeifer, D. (2001) Genomics 78, 73–82.28. Munsterberg, A. & Lovell-Badge, R. (1991) Development (Cambridge, U.K.) 113, 613–624.29. Dyche, W. J. (1979) J. Morphol. 162, 175–209.30. Allard, S., Adin, P., Gouedard, L., di Clemente, N., Joss, N., Orgebin-Crist, M.-C., Picard,

J.-Y. & Xavier, F. (2000) Development (Cambridge, U.K.) 127, 3349–3360.31. Kobayashi, A., Shawlot, W., Kania, A. & Behringer, R. R. (2004) Development (Cambridge,

U.K.) 131, 539–549.32. Shen, W.-H., Moore, C. C., Ikeda, Y., Parker, K. L. & Ingraham, H. A. (1994) Cell 77, 651–661.33. Watanabe, K., Clarke, T. R., Lane, A. H., Wang, X. & Donahoe, P. K. (2000) Proc. Natl.

Acad. Sci. USA 97, 1624–1629.34. Wilhelm, D. & Englert, C. (2002) Genes Dev. 16, 1839–1851.36. Schepers, G. E., Bullejos, M., Hosking, B. M. & Koopman, P. (2000) Nucleic Acids Res. 28,

1473–1480.36. Schepers, G., Wilson, M., Wilhelm, D. & Koopman, P. (2003) J. Biol. Chem. 278, 28101–28108.37. Kent, J., Coriat, A. M., Sharpe, P. T., Hastie, N. D. & van Heyningen, V. (1995) Oncogene

11, 1781–1792.38. Parker, C. R., Jr., Faye-Petersen, O., Stankovic, A. K., Mason, J. I. & Grizzle, W. E. (1995)

Endocr. Res. 21, 69–80.39. Stolt, C. C., Schmitt, S., Lommes, P., Sock, E. & Wegner, M. (2005) Dev. Biol. 281, 309–317.40. Gao, F., Maiti, S., Sun, G., Ordonez, N. G., Udtha, M., Deng, J. M., Behringer, R. R. & Huff,

V. (2004) Mol. Cell. Biol. 24, 9899–9910.41. Rao, M. K., Pham, J., Imam, J. S., MacLean, J. A., Murali, D., Furuta, Y., Sinha-Hikim, A.

P., & Wilkinson, M. F. (2006) Gene Dev. 20, 147–152.

11992 � www.pnas.org�cgi�doi�10.1073�pnas.0600994103 Gao et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

9, 2

020