paracrine-mediated apoptosis in reproductive tract development · programmed cell death during...

a

rdMpreee

i

Developmental Biology 208, 110–122 (1999)Article ID dbio.1998.9190, available online at http://www.idealibrary.com on

Paracrine-Mediated Apoptosis in ReproductiveTract Development

Lori M. Roberts,*,† Yoshifumi Hirokawa,* Mark W. Nachtigal,*nd Holly A. Ingraham*,†

*Department of Physiology and †Graduate Program in Biomedical Sciences, Universityof California, San Francisco, Box 0444, San Francisco, California 94143-0444

In mammalian development, the signaling pathways that couple extracellular death signals with the apoptotic machineryare still poorly understood. We chose to examine Mullerian duct regression in the developing reproductive tract as a possiblemodel of apoptosis during morphogenesis. The TGFb-like hormone, Mullerian inhibiting substance (MIS), initiatesegression of the Mullerian duct or female reproductive tract anlagen; this event is essential for proper male sexualifferentiation and occurs between embryonic days (E) 14 and 17 in the rat. Here, we show that apoptosis occurs duringullerian duct regression in male embryos beginning at E15. Female Mullerian ducts exposed to MIS also exhibited

rominent apoptosis within 13 h, which was blocked by a caspase inhibitor. In both males and females the MIS type-IIeceptor is expressed exclusively in the mesenchymal cell layer surrounding the duct, whereas apoptotic cells localize to thepithelium. In addition, tissue recombination experiments provide evidence that MIS does not act directly on thepithelium to induce apoptosis. Based on these data, we suggest that MIS triggers cell death by altering mesenchymal–pithelial interactions. © 1999 Academic Press

Key Words: Mullerian inhibiting substance; morphogenesis; programmed cell death; epithelial–mesenchymal
nteractions.
INTRODUCTION

During morphogenesis, extracellular cues regulate theshape and size of tissues by regulating programmed celldeath (for review, see Jacobson et al., 1997). Studies of themolecular mechanisms of apoptosis have focused primar-ily on the intracellular death machinery required forapoptosis in all cells. Members of the caspase family ofcysteine proteases (ICE proteases) execute almost all ofthe biochemical and morphological changes observed inapoptotic cells (for review, see Reed, 1997; White, 1996).In turn, members of the Bcl-2 gene family are upstream ofcaspases and are proposed to trigger this proteolyticcascade. The mechanisms by which Bcl-2 proteins inte-grate extracellular death and survival cues remain poorlyunderstood. While much progress has been made inunderstanding the universal cell death machinery, thenature of upstream signaling pathways that regulateapoptosis in development remain largely unknown. Tro-phic factors are known to influence morphogenesis and
programmed cell death. For instance, limiting survivalfactors are proposed to account for the massive cell death
110

observed in the developing nervous system, thus ensur-ing appropriate cell number as well as correct target cellinnervation (Jacobson et al., 1997). Recent studies havedetermined that survival factor signaling may inhibitdeath-promoting bcl-2 family members, and thus sup-press caspase activation. However, only a few signalingpathways that activate the apoptosis program have beenidentified. These include members of the tumor necrosisfactor (TNF) gene family, including Fas ligand (FasL),TNFa, and TRAIL (TNF-related apoptosis ligand, re-viewed in Nagata, 1997). These trimeric ligands bind totheir cognate receptors after metalloproteinase-mediatedproteolytic cleavage generates soluble ligand from amembrane bound form (Gearing et al., 1994). Thesesignaling pathways activate the caspase cascade directlythrough the signaling molecule FADD/MORT1 (Boldin etal., 1996; Muzio et al., 1996). Genetic analyses of natu-rally occurring mutants of FasL (gld) and fas (lpr) in bothhumans and mice underscore the importance of thispathway in maintaining the correct numbers of T lym-
phocytes (Nagata, 1997). Despite the widespread expres-sion of the Fas ligand and receptor, mutant gld and lpr
0012-1606/99 $30.00Copyright © 1999 by Academic Press

All rights of reproduction in any form reserved.

1

eomiiMdwTricaa

rrintespdiwtmlMmmisfe

w

wmHTsz(wdrits

r

i

111Apoptosis in the Reproductive Tract

mice appear morphologically normal (Takahashi et al.,994; Watanabe-Fukunaga et al., 1992), as do mice carry-

ing mutations in downstream effector genes (Yeh et al.,1998). Collectively, these results suggest that vertebratedevelopment utilizes additional tissue- and temporal-specific death factors and receptor systems to mediateprogrammed cell death during development.

We chose to examine Mullerian duct regression in the ratembryo as an experimental model of apoptosis duringmorphogenesis. Prior to initiation of the sex determinationcascade, mammalian embryos possess anlagen for bothmale and female internal reproductive tracts, the Wolffianand Mullerian ducts, respectively. During male sexualdetermination the embryonic testis secretes Mullerian in-hibiting substance (MIS), which induces Mullerian ductregression (Behringer, 1995). Several cellular mechanismshave been proposed to account for the morphological ac-tions of MIS, first noted by Jost and colleagues in the late1940s (reviewed in Josso et al., 1993). First, a migratoryvent could account for regression. This hypothesis is basedn the observation that Mullerian epithelial duct cellsigrate out of the duct to form part of the nephric tubules

n male alligator embryos (Austin, 1995). Second, MIS couldnduce an epithelial to mesenchymal transformation of

ullerian duct epithelium based on observations that theisappearance of the basement membrane concomitantith the loss of cell polarity (Dyche, 1979; Josso, 1992;relstad et al., 1982; Wartenberg, 1985). Third, the Mulle-

ian duct may undergo apoptosis or programmed cell deathn response to MIS as suggested by early electron micros-opy studies showing an increased number of lysosomesnd invasion of macrophages (Djehiche et al., 1994; Price etl., 1977).It is well established that male and female rat genital

idge cultures are competent to regress upon exposure toecombinant MIS ligand (Tsuji et al., 1992). Therefore, its possible to examine the molecular and cellular mecha-isms of MIS-induced regression in defined organ cul-ures of female urogenital ridges, as well as in malembryos in vivo. In this study, we used the organ cultureystem and the TUNEL protocol to address the role ofrogrammed cell death in Mullerian duct regression. Weemonstrate unequivocally that MIS induced apoptosisn female Mullerian ducts, and that apoptosis correlatedith Mullerian duct regression in male embryos. While

he MIS type II receptor was expressed only in theesenchyme surrounding the duct, apoptotic cells

ocalized to the Mullerian epithelium, suggesting thatIS activates the death machinery via a paracrineechanism. In addition, tissue recombination experi-ents showed that MIS induces epithelial apoptosis

ndirectly, perhaps by suppressing mesenchyme-derivedurvival factors. Collectively, our results offer a model
or induction of apoptosis during remodeling of thembryo.
Copyright © 1999 by Academic Press. All right

MATERIALS AND METHODS

Organ culture and regression assays. Male and female raturogenital ridges were dissected from E14.5 or E15.5 embryos withthe plug date defined as E0. Morphological criteria also were usedto stage individual embryos. Dissected organs were either fixedimmediately or cultured on MilliCell-CM (Millipore) culture plateinsert filters floated on 0.3 ml of medium as previously described(Nachtigal and Ingraham, 1996). For female genital ridges, 100 nMof testosterone was added to each culture to preserve Wolffian ductmorphology. Recombinant bioactive (MIS-RARR) and inactive(MIS-RAGA) MIS ligand were obtained from conditioned mediacollected from stably transfected HEK-293S cells expressing the ratMIS cDNA as described (Nachtigal and Ingraham, 1996). Condi-tioned medium was concentrated 93-fold and added to cultures (22ml/ml or ;1–2 mg/ml) at 37°C, 5% CO2 for 6 to 72 h. The amountand integrity of recombinant MIS protein was determined byWestern analysis with purified anti-rat MIS antibody (Nachtigaland Ingraham, 1996). Tissues were fixed overnight in 10% bufferedformalin, pH 7.0, at 4°C, embedded in OTC or paraffin, andsectioned (5–10 mm) or fixed in 4% paraformaldehyde (PFA) for

hole mount analysis.TUNEL and in situ hybridization assays. The TUNEL assay

as performed using the Apoptag FITC kit (Oncor) according toanufacturer’s protocols and counterstained with 1 mg/mloechst-33258 dye to visualize nuclei. For double staining ofUNEL and Pax-2-positive cells, fixed tissue was frozen in OCT,ectioned, and treated with terminal deoxy-transferase (TdT) en-yme and washed. Sections were incubated with rabbit anti-Pax-21:200, BAbCo) for 1 h at 22°C in the Apoptag blocking solution,ashed with PBS, and incubated with two secondary antibodies,onkey anti-TdT-FITC (Oncor) and Cy3-conjugated goat anti-abbit IgG (1:100, Jackson). Double-stained sections were visual-zed using three-dimensional widefield fluorescent microscopy andhe image was deconvoluted by three-dimensional visualizationoftware (Delta Vision).

Whole mount in situ hybridization was performed on PFA-fixedat genital ridges treated with proteinase K (10 mg/ml; 30–45 min at

room temperature) and refixed with 4% PFA. Tissues were incu-bated with digoxigenin (DIG)-labeled RNA probe generated fromthe entire coding region of the rat MISIIR. The MISIIR from P90testes cDNA was isolated by RT-PCR, based on published ratcDNA (Baarends et al., 1994), and subcloned into the pBSKII2

vector by ligating two 935-bp and one 860-bp PCR fragments withthe unique SfiI site; the authenticity of the MISIIR was verified byDNA sequence analysis. Tissue was incubated in hybridizationbuffer (1.33 SSC, pH 5, 50% formamide, 0.2% Tween 20, and 0.5%CHAPS) overnight at 65°C. After extensive washing and blocking(60 min in Boehringer-Mannheim blocking reagent), tissues wereincubated overnight at 4°C with anti-DIG antibody (1:2000,Boehringer-Mannheim). After washing, color development wasperformed using BM Purple AP Substrate for 30–60 min at RT.

For whole mount TUNEL staining of urogenital ridges, tissuewas fixed immediately after dissection, because nonspecific stain-ing was observed when organs were cultured. Genital ridges wereprepared according to the protocol for whole mount in situ hybrid-zation (see above), including storage at 220°C in hybridization

buffer. After several washes in PBS, the TUNEL assay was per-formed according to the Apoptag protocol, with the followingmodifications. Tween 20 was added to the stop/wash buffer and
PBS washes to a final concentration of 0.1%. Washes were carriedout for 15 min with rotation, and the final wash was extended to
s of reproduction in any form reserved.

m1

rdcdRMPgtHiktIrR

oMrdalec

iondpw

112 Roberts et al.

1 h. Genital ridges were blocked for 30 min with Apoptag blockingbuffer before incubation with antibody. Organs were mountedunder coverslips and viewed using standard fluorescence micros-copy.

Epithelial duct cultures. Mullerian duct epithelial and mesen-chymal tissue was isolated as described (Tsuji et al., 1992) with

odifications. E15.5 female urogenital ridges were incubated in% trypsin (Sigma) in HBSS without Ca21/Mg21) for 20 min at 4°C.

Ridges were transferred to HBSS (with Ca21/Mg21) containing 10mg/ml soybean trypsin inhibitor (Sigma). The Mullerian mesen-chyme was stripped from the epithelial duct using fine needles, andpieces of epithelial and mesenchymal tissue were cultured asdescribed for organ culture of genital ridges (see above). Epithelialducts were cultured with Mullerian mesenchyme or heterologousmesenchyme from connective tissue adjacent to the embryonickidney, which was trypsinized alongside genital ridges. Alterna-tively, ducts were cultured in isolation, with or without 10%female FBS, or placed immediately adjacent to the Mullerian ductof an intact female genital ridge (not trypsinized). After 18–24 h ofculture, tissue was prepared for TUNEL analysis (see above).MIS-RARR or MIS-RAGA was added to cultures as described forgenital ridge organ cultures. Observations are reported for threeindependent cultures. Percentage apoptosis was calculated bycounting the number of TUNEL positive cells in 16–24 cross-sections of Mullerian epithelium and the total number of Hoechst-stained nuclei in the epithelium.

RESULTS

Apoptosis Occurs during Mullerian DuctRegression

In the rat, MIS transcripts are detected in testis beginningat E13.5 (Shen et al., 1994) and by E17.5 Mullerian ductegression is complete in males, with only the Wolffianuct visible within the genital ridge (Fig. 1C). Femaleultures normally show no degeneration of the Mullerianucts (Fig. 1C), but are responsive to recombinant MIS-ARR, as evident by the complete degeneration of theullerian duct (Fig. 1D; Nachtigal and Ingraham, 1996).

revious studies have shown that processing of the prore-ion from the mature region of MIS is required for bioac-ivity (Nachtigal and Ingraham, 1996; Wilson et al., 1993).ere, we use a form of recombinant MIS (MIS-RARR) that

s correctly processed and secreted from human embryonicidney 293-S cells, and a mutant form of MIS (MIS-RAGA)hat is not processed (Figs. 1A and 1B, and Nachtigal andngraham, 1996). Regression is not observed when femaleidges are cultured with the inactive form of MIS, MIS-AGA (Fig. 1D).As a first step in defining the cellular and molecular basis

f Mullerian duct regression, we examined whether theullerian duct undergoes programmed cell death during

egression. The TUNEL method uses fluorescent labeling toetect fragmented DNA, which is one of the hallmarks ofpoptosis. TUNEL-positive cells were observed in the Mul-erian ducts of E15.5–17.5 male embryos (Fig. 2), but not at
arlier stages of development (E14, Fig. 2). These dataonfirm the earlier morphological observations of cell death

n the regressing duct. Very few TUNEL-positive cells werebserved in the Mullerian ducts of female littermates (dataot shown). The number of TUNEL-positive cells increasedramatically from E15.5 to E16.5; this increase correlatedrecisely with progression of Mullerian duct degeneration,ith loss of a visible duct observed at E17.5 (Fig. 2).

MIS Induces Apoptosis in the Mullerian Duct

In organ culture, the testes of male urogenital ridges(E14.5–15.5) produce MIS, resulting in complete Mullerianduct regression after three days. One can recapitulate thisprocess by culturing female ridges with recombinant MISfor 72 h (Fig. 1D). To assess whether apoptosis accompaniesMullerian duct regression in female cultures, TUNEL stain-ing was performed on female genital ridges cultured in thepresence of bioactive MIS (MIS-RARR). Following exposureto bioactive MIS, apoptotic cells were clearly visible withinthe Mullerian duct after 13 h of culture (n $ 5, data notshown). The number of TUNEL-positive cells increasedwith longer exposure to MIS (data not shown; 18 h, n 5 2,24 h, n $ 5), and female ridges cultured with MIS for 48 hdisplayed prominent apoptosis in the Mullerian duct (Figs.3C and 3D, MIS-RARR). Untreated female genital ridgesexhibited very few apoptotic cells in the Mullerian duct, asdid female genital ridges cultured with the inactive form ofMIS, MIS-RAGA (Figs. 3A and 3B). This is consistent withthe persistence of the Mullerian duct in organs treated withMIS-RAGA (Fig. 1D). MIS-induced apoptosis occurred spe-cifically in the Mullerian duct, whereas the pattern ofTUNEL staining elsewhere in the genital ridge was similarin untreated or MIS (RARR, RAGA)-treated tissues. To testfurther whether MIS induces apoptosis in the Mullerianduct, we cultured female genital ridges with MIS and abroad-range inhibitor of caspases. After 18 h of culture withMIS and solvent alone, female genital ridges exhibitedTUNEL staining in the Mullerian duct (Fig. 3E). The addi-tion of the caspase inhibitor Boc-D-FMK completely abol-ished TUNEL staining in the genital ridge (Fig. 3F), showingthat MIS activates a classical apoptotic cascade in theMullerian duct.

Apoptosis and MISIIR Expression in UrogenitalRidges

Previous studies demonstrated that degeneration of theduct begins at the cranial end of the duct and progressescaudally in vivo and in organ cultures treated with bioac-tive MIS (Tsuji et al., 1992). As a member of the TGFbsuperfamily, MIS is presumed to signal through a hetero-meric complex of type I and type II serine/threonine kinasereceptors. A cranial-caudal gradient of receptor expressioncould account for the pattern of regression. The cloning andsubsequent genetic proof verifying the authenticity of theMISIIR makes it possible to explore this issue (Baarends et
al., 1994; di Clemente et al., 1994; Mishina et al., 1996).However, whole mount in situ hybridization of embryonic

M(rd

amnt

wss

ligan


male genital ridges revealed uniform expression of the MIStype II receptor (MISIIR) along the length of the duct, withno obvious cranial-caudal gradient of expression (Fig. 4A,E15). This uniform pattern of expression is also observed infemale genital ridges. During the developmental window ofMullerian duct regression, expression of the MISIIR tran-scripts disappeared in male genital ridges along the cranial-caudal axis (Fig. 4A). By contrast, MISIIR transcripts areobserved at all developmental stages in the female genital

FIG. 1. Mullerian duct regression after treatment with bioactive Mith the proregion and mature carboxyl terminus. Two forms of re

hown in the diagram. The authenticity of each MIS product was tetaining. (B) Shows that the mutant MIS-RAGA protein is not cleav

kDa. In contrast, complete cleavage was observed with the MIS-Rmature carboxy-terminus of 12 kDa. (C) E14.5 female and male genfor 72 h (female and male). Two ducts are present in cultured femadjacent duct is the Wolffian duct (WD). The Mullerian duct regresvisible. (D) Female genital ridges incubated with conditioned muntreated female ridges, while ridges cultured with bioactive MIS3).

ridges (Fig. 4B), consistent with persistence of the Mullerianduct in females. Prominent and persistent expression of the

4n


ISIIR is also observed in both embryonic testis and ovaryFigs. 4A and 4B; Teixeira et al., 1996). However, MISIIReceptor expression is extinguished following Mullerianuct regression.Given the cranial-caudal pattern of duct regression, we

sked if apoptosis conforms to this pattern by using wholeount TUNEL staining. In E15.5 male genital ridges, but

ot in female genital ridges, a coincidence between apopto-ic cells and the Mullerian duct was observed (Figs. 4C and

gand. (A) A schematic of the coding region of the rat MIS is showninant MIS used in all studies were MIS-RARR, and MIS-RAGA, asy both Western analysis (data not shown) and by SDS–PAGE silverd migrates under reducing conditions at the full-length size of ;70protein yielding the predicted proregion fragment of 48 kDa and

ridges were cultured with HEK-293S control conditioned mediumenital ridges; the most lateral is the Mullerian duct (MD) and the

cultured male genital ridges, and only the Wolffian duct remainsm containing noncleavable MIS-RAGA (1MIS-RAGA) resembled (1MIS-RARR) undergo Mullerian duct regression after 72 h (n .

IS licombsted bed anARRital

ale gses inediu

D). The spatial distribution of TUNEL-positive cells wasot restricted to the cranial portion of the Mullerian duct,


fM

114 Roberts et al.

FIG. 2. Apoptosis during Mullerian duct regression. Cross-sections of male genital ridges were analyzed for programmed cell death byTUNEL staining. Bright green TUNEL-positive cells are observed in the Mullerian duct region beginning at E15.5 (n 5 2 for E14.5, n 5 3
or E15.5, E16.5). All sections were counterstained with Hoechst dye so that Mullerian ducts (dotted circle) could be visualized.
agnification is 4003 for E14.5 and E17.5 and 6003 for E15.5 and E16.5 sections.Copyright © 1999 by Academic Press. All rights of reproduction in any form reserved.

c

D-FM2


but was observed throughout the length of the duct (Fig.

FIG. 3. MIS induces apoptosis in cultured female genital ridges. (AMIS (1MIS-RAGA) or bioactive MIS (1MIS-RARR) and stained byMIS-RAGA treated cultures (D and B, respectively, n 5 2). A and Cduct (WD) and Mullerian duct (MD) are indicated. After exposure(E and F) The caspase inhibitor Boc-D-FMK prevents MIS-induced(1MIS-RARR) exhibit TUNEL staining in the Mullerian duct. Boc-). Magnification is 4003.

4C). Moreover, TUNEL staining of sections obtained fromrepresentative positions along the length of the genital ridge

cn


onfirmed that both the cranial and caudal ends of the duct

14.5 female genital ridges were cultured for 48 h with nonbioactiveEL. Apoptotic cells are present in the MIS-RARR but not in thethe same sections counterstained with Hoechst dye; the Wolffian

oactive MIS for 48 h, the Mullerian duct appears disorganized (C).osis. Female ridges cultured with MIS for 18 h with solvent aloneK (100 mM) abolishes TUNEL staining (1MIS 1 Boc-D-FMK) (n 5

–D) ETUN

showto biapopt

ontain similar numbers of apoptotic cells (data not shown,5 3). These data, coupled with the uniform presence of


116 Roberts et al.

Copyright © 1999 by Academic Press. All rights of reproduction in any form reserved.

mtdbddanasMltacasteMsi(se

McOnawRsiFmettnwMrroFMsat


MISIIR along the cranial-caudal axis of the duct, imply thatwhile the cranial portion of the Mullerian duct is the first todisappear, MIS signaling occurs along the length of the ductduring regression.

MIS Induces Apoptosis Indirectly

Consistent with previous reports, our nonradioactive insitu hybridization analysis revealed that MISIIR expressionis restricted to the mesenchymal cells surrounding theMullerian duct (Fig. 5A, and Baarends et al., 1994). Further-

ore, MISIIR expression is more prominent in the lateralhan in the medial region of mesenchyme surrounding theuct, consistent with the crescent shape noted previouslyy others (Baarends et al., 1994; Teixeira et al., 1996). Toetermine which cell populations undergo programmed celleath, we performed double staining with TUNEL and annti-Pax-2 antibody. Pax-2 is a marker of epithelial compo-ents of intermediate mesoderm, which includes Mulleriannd Wolffian ducts (Torres et al., 1995). Indeed, this analy-is confirmed that most apoptotic cells localized to theullerian duct epithelium, suggesting that only the epithe-

ial cells undergo cell death (Fig. 5B). However, we notedhat TUNEL-positive cells do not costain with anti-Pax-2ntibody, suggesting that TUNEL stains dead epithelialells that have extinguished Pax-2 expression. These data,nd the absence of MISIIR in the Mullerian epitheliumtrongly suggest that MIS induces apoptosis indirectly,hrough the Mullerian mesenchyme. While MISIIR also isxpressed in the embryonic gonad, we have excluded thatISIIR in the gonad plays a role in Mullerian duct regres-

ion, because removal of the gonad does not prevent MIS-nduced Mullerian duct regression in cultured female ridgesdata not shown, n 5 2). These data demonstrate that MISignals exclusively through MISIIR in the Mullerian mes-nchyme to cause Mullerian duct regression.To test whether Mullerian mesenchyme is required for

FIG. 4. MIS receptor expression and apoptosis are observed alohybridization of the rat MISIIR was carried out on male and femalefemale gonad, but are clearly absent in the adjacent mesonephros, exBarely visible in both the E15 Male and Female panels is the faint ouGross differences in the size of the testis and ovary are observed 1duct is decreased in males at E16 (A, E16), whereas the MISIIR sig(B, E16). In males, nearly complete Mullerian duct regression hasextreme caudal portion of the genital ridge, shown by arrow (A, Elength of the ridge (B, E17). (C and D) Whole mount TUNEL staininin the Mullerian duct (arrows). No positive cells are seen in the Madjacent mesonephric tubules in both males and females. Magnificexperiments.FIG. 5. Mullerian mesenchyme expresses MISIIR, while apoptosisto the rat MISIIR are observed in the periductal region surroundinsignal. MISIIR signal is absent from the Wolffian duct (data not shto the Mullerian duct epithelium (MD) in E16 male genital ridge (n
fluorescence, where as TUNEL-positive cells are bright green. The digitiand is 6003 magnification.

IS-induced apoptosis, we separated the tissue layers andultured Mullerian epithelial ducts without mesenchyme.rgan cultures of isolated epithelial ducts exhibited promi-ent apoptosis, even in the presence of 10% FBS (35%poptosis, Figs. 6A and 6E); no differences were observedhen isolated epithelial ducts were cultured with MIS-ARR or MIS-RAGA (data not shown). Apoptosis wasuppressed when epithelial ducts were placed adjacent tontact female genital ridges (12% apoptosis, respectively,igs. 6B and 6E), suggesting that the epithelium requiresesenchymal survival signals. When MIS was added to

pithelial ducts cultured with heterologous mesenchymehat lacks MISIIR, apoptotic cells were observed in some ofhe epithelial ducts (25%, Figs. 6C and 6E). However, theumber of apoptotic epithelial cells was twofold higherhen MIS was added to epithelial ducts cultured withullerian mesenchyme (53%, Figs. 6D and 6E). These

esults show that MIS does not act directly on the Mulle-ian epithelium to induce apoptosis, consistent with thebservation that the epithelium does not express MISIIR.urthermore, our coculture experiments suggest that theullerian epithelium requires mesenchymal support for

urvival, raising the possibility that MIS induces epithelialpoptosis by repressing mesenchyme-derived survival fac-ors.

DISCUSSION

Programmed cell death is known to play a critical role inthe process of vertebrate morphogenesis. While MIS-induced Mullerian duct regression is a well-studied ex-ample of hormonally regulated morphogenesis, the mecha-nism of regression and the involvement of apoptosis in thisprocess have remained questionable. Although some stud-ies reported sporadic apoptotic cells (;2 TUNEL-positivecells/section) in MIS-treated Mullerian ducts (Catlin et al.,

e length of the Mullerian duct. (A and B) Whole mount in situenital ridges. MISIIR transcripts are observed in both the male andfor the prominent signal observed around the Mullerian duct (MD).of the Wolffian duct, which fails to hybridize to the MISIIR probe.

ater in development. The intensity of the MISIIR in the Mulleriann female genital ridges is equivalent to the signal observed at E15

place at E17, and MISIIR transcripts can be observed only at theIn contrast, MISIIR transcripts persist in females along the entire15.5 male genital ridges (C) shows a band of TUNEL-positive cellsan duct in E15.5 female ridges (D). Positive cells are visible in the

is 1003, representative micrographs are shown for four separate

calized to the Mullerian epithelium. (A) Transcripts correspondinge Mullerian duct (MD), shown by the prominent crescent-shaped. (B) TUNEL-positive cells and cells expressing Pax-2 are confined). Pax-2 is a marker of Mullerian epithelium and is stained by red

ng thurogcepttline

day lnal itaken17).g of Eulleriation

is log th

own)5 2

zed image was acquired as described under Materials and Methods


a

118 Roberts et al.

FIG. 6. Paracrine interactions in the Mullerian duct. Mullerian epithelium requires mesenchyme for survival. (A) TUNEL analysis andcorresponding Hoechst staining of cross sections of isolated epithelial ducts reveals extensive apoptosis in the absence of mesenchyme.Magnification is 6003. (B) TUNEL and Hoechst staining of epithelial ducts (Ep, dotted circle) cultured adjacent to an intact female genitalridge shows reduced apoptosis compared to the isolated epithelium (see A). Magnification is 4003. Epithelial ducts were isolated asdescribed under Materials and Methods; several ducts were cultured together and lost tubular morphology as they flattened in culture.MIS-induced epithelial apoptosis requires Mullerian mesenchyme. (C) TUNEL and Hoechst staining of a cross section of an epithelial ductcultured with heterologous mesenchyme and MIS. (D) TUNEL and Hoechst staining of an epithelial duct cultured with Mullerianmesenchyme and MIS. Apoptosis is increased in the duct (Ep, dotted circle) compared to ducts cultured with MIS and heterologousmesenchyme (see panel C). Magnification is 4003. (E) Quantitation of apoptosis in epithelial–mesenchymal recombination cultures.Epithelium cultured in isolation (E alone) contained threefold more apoptotic cells compared to epithelium cultured adjacent to an intactgenital ridge (E 1 GR). Mullerian epithelium (E, red circle) cultured with Mullerian mesenchyme (M, blue circle) and MIS (E 1 M 1 MIS)contained twofold more apoptotic cells than epithelium combined with heterologous mesenchyme (H, green circle) and MIS (E 1 H 1 MIS).
Error bars represent standard error of the mean for three experiments and the ** above bars indicate statistical significance of P , 0.001s determined by a standard t test.
Copyright © 1999 by Academic Press. All rights of reproduction in any form reserved.

gs(sd

gatttcicmrttpa

MctattresraccCa

al

l(


1997; Price et al., 1977), others have concluded that pro-rammed cell death does not play a decisive role in regres-ion based on ultrastructural analysis in human embryosWartenberg, 1985). Now, using the TUNEL protocol, wehow that programmed cell death correlates with Mullerianuct regression in male embryos in vivo. Importantly,

recombinant MIS induces both regression and apoptosis infemale genital ridges. Furthermore, inhibitors of caspaseactivity prevent MIS-induced apoptosis. Given that expres-sion of the MISIIR is restricted to the mesenchymal layer,we postulate that a paracrine mechanism relays a deathsignal to the underlying epithelial layer.

Cell Death Precedes Mullerian Duct Regression

Our results suggest that there is strong temporal andspatial correlation between apoptosis and Mullerian ductregression. Apoptotic cells are detected as early as 13 h afterMIS is added. Complete regression along the cranial tocaudal axis occurs over 72 h, and MIS must be present forthe first 48 h of culture for the duct to disappear completely(LMR, unpublished data). If MIS is withdrawn earlier, theduct is reduced in size, but remains visible, suggesting thatnot all cells commit to the apoptotic program at once.Degeneration of the Mullerian epithelium is accompaniedby a whorl of condensed mesenchyme in the E17.5 malegenital ridge (Fig. 2). Despite complete regression at thisstage, numerous apoptotic nuclei are observed in the con-densed mesenchyme. Two explanations may account forthis pattern; first, these apoptotic cells may be epithelialcell corpses; alternatively, death of the Mullerian mesen-chyme may follow regression of the epithelium. The latterpossibility is consistent with our finding that mesenchymalexpression of MISIIR transcripts is extinguished as regres-sion proceeds. Taken together, these observations implythat viability of the mesenchyme requires an intact epithe-lium.

Cranial to Caudal Regression Pattern

It has been appreciated for many years that Mullerianduct regression initiates at the cranial end of the ridge andproceeds caudally both in vivo and in vitro. A cranial-caudalradient of MISIIR could produce a higher rate of apoptosist the cranial end of the duct. The uniform appearance ofhe MISIIR transcripts and presence of apoptotic cells alonghe length of Mullerian duct eliminates this explanation forhe pattern of duct regression. However, TUNEL detectsells during a narrow window of the apoptotic program, thats, after DNA fragmentation occurs, and before the cellorpses are degraded completely. Thus, TUNEL stainingay not reflect differences in the onset of apoptosis or the

ate of progression through the apoptotic program. Evenhough TUNEL staining fails to detect a difference betweenhe cranial and caudal ends of the duct, it rules out the
ossibility that a discrete wavefront of cell death occurslong the duct.

What might account for the cranial-caudal pattern ofullerian duct regression? Despite the lack of a cranial-

audal gradient of MISIIR transcript, we have not ruled outhe possibility that the receptor protein is expressed in suchfashion. Another component of the MIS signaling cascade

hat could account for this pattern of cell disappearance ishe Type I receptor for MIS (MISIR). Thus far, Alk2 (R1)emains the best candidate for the MISIR because of itsxpression in the urogenital ridge (based on RT-PCR expres-ion, He et al., 1993) and because MIS ligand-mediatedepression of an aromatase reporter requires both MISIIRnd Alk2 (J. Visser and A. Themmen, personal communi-ation). However, to date we have been unable to detect alear gradient of ALK2 expression (Y.H., unpublished data).learly, future work is needed to determine the identity

nd spatial expression of the MISIR.The spatial pattern of Mullerian duct regression may

rise from regulation of the bioavailability of the MISigand. Components of TGF-b-superfamily signaling that

might regulate the concentration of bioactive MIS areinhibitory binding proteins (such as follistatin) and themetalloproteinases that degrade these binding proteins(such as BMP1, tolloid, or Xolloid). Recent studies inXenopus, Drosophila, and Zebrafish suggest that patterningof the dorsal-ventral axis occurs by interactions betweenthree extracellular components of TGFb signaling, theigand, the binding protein and the metalloproteinaseBlader et al., 1997; Hsu et al., 1998; Marques et al., 1997;Piccolo et al., 1997). Proteolytic degradation of the inacti-vating binding protein by metalloproteinases may help toestablish a gradient of available ligand. Currently, neither aspecific binding protein for MIS nor an extracellular BMP1-like metalloproteinase have been identified for the MISsignaling cascade.

Epithelial–Mesenchymal Interactions in theMullerian Duct

Previous expression studies suggested that the MISIIR islimited to the Mullerian mesenchyme. However, theseexperiments were based on in situ hybridization withradiolabeled probes and lacked the spatial resolution re-quired to detect precisely the boundaries of MISIIR expres-sion. Using high-resolution colorimetric detection, we pro-vide conclusive evidence that the Mullerian epitheliumdoes not express MISIIR. Yet, the epithelium responds toMIS by undergoing apoptosis, suggesting that MIS inducescell death of the Mullerian duct epithelium via a paracrinesignal from the mesenchyme. Indeed, previous studies areconsistent with our hypothesis that MIS regulates paracrineinteractions; Tsuji and colleagues reported that MIS inhib-ited proliferation of dispersed mesenchymal cells but hadno effect on isolated Mullerian epithelial cells (Tsuji et al.,1992). These results are similar to our own, which showthat epithelial ducts do not respond to MIS when they are
cultured with heterologous mesenchyme that lacks MISreceptors. In addition, our coculture experiments suggest

1

a1IsealbrvbEtm

tM

1dide

setBgtmo

otTuat

B

120 Roberts et al.

that the epithelium requires mesenchyme-derived survivalfactors. Tsuji and colleagues did not observe a similarrequirement, perhaps because they cultured epithelial andmesenchymal cells as dispersed monolayers, whereas wecultured intact epithelial ducts. Based on our data, weconclude that MIS induces the Mullerian mesenchyme to“murder” the epithelium. Furthermore, our results suggestthat MIS may repress mesenchyme-derived survival factors,resulting in epithelial apoptosis. Alternatively, MIS mayupregulate a paracrine death signal, which induces apopto-sis even in the presence of mesenchymal survival factors.

Signaling from the epithelium to the mesenchyme mustalso be important for reproductive tract development asrecently reported by Parr and McMahon (1998). Indeed, theyshowed that expression of MISIIR in the Mullerian mesen-chyme requires signaling by Wnt-7a, which is expressed inMullerian epithelium. The loss of Wnt-7a expression fol-lowing death of the Mullerian epithelium may account forthe loss of MISIIR expression in the Mullerian mesenchymeduring regression in males (this study and Teixeira et al.,996).

Paracrine-Mediated Cell Death in Development

Spatial and temporal control of apoptosis represents asignificant mechanism of morphogenesis; to date, the mo-lecular signals that regulate the cell death machinery dur-ing development remain a mystery. The idea that cell–cellinteractions regulate programmed cell death during devel-opment has been proposed for involution of mammarygland (Boudreau et al., 1995), for tail resorption duringmphibian metamorphosis (Brown et al., 1996; Su et al.,997) and for inner ear morphogenesis (Fekete et al., 1997).n the case of mammary involution and tail resorption,ecreted metalloproteinases are proposed to degrade thextracellular matrix (ECM), leading to the death ofnchorage-dependent epithelial cells (Werb, 1997). Simi-arly, competition for attachment to the basement mem-rane leads to apoptosis of embryonic ectodermal cells,esulting in cavitation of the vertebrate embryo (Coucou-anis and Martin, 1995). In our system, dissolution of theasement membrane surrounding the Mullerian duct byCM proteolysis also might induce epithelial apoptosis;herefore, we might predict that MIS regulates matrixetalloproteinase genes.To date, one of the best-characterized apoptotic events is

he interdigital death that occurs during limb development.embers of the TGFb-superfamily, the BMPs, are ex-

pressed in the interdigital mesenchyme that is fated toundergo apoptosis. Apoptosis follows ectopic application ofBMP-2, -4, and -7, suggesting that BMPs induce interdigitalprogrammed cell death (Ganan et al., 1996; Macias et al.,997; Yokouchi et al., 1996). Furthermore, expression ofominant-negative receptors for BMPs (BMPIR) abrogatesnterdigital apoptosis in the chick limb (Zou and Niswan-
er, 1996). MIS-induced apoptosis involves mesenchymal–pithelial interaction; this may not be the case in the limb

ince cultured interdigital mesenchyme cells die whenxposed to BMP (Yokouchi et al., 1996). These data suggesthat interdigital cell death can be activated directly byMPs. However, other studies support a role for the mar-inal ectoderm in digit formation since removal of thisissue prevents interdigital cell death (Ros et al., 1997). Aolecular understanding of apoptosis in the limb and in

ther developmental systems remains to be delineated.Our study provides definitive evidence that apoptosis

ccurs during reproductive tract morphogenesis, and thathis process can be recapitulated in a defined organ culture.hus, our in vitro model system affords an opportunity tonravel the molecular basis of this paracrine death signalnd is likely to reveal effectors shared by other developmen-al programs of cell death.

ACKNOWLEDGMENTS

We are grateful to Evelyn N. Garrett and the morphology core(funded by HD11979) for technical assistance in histology. Weespecially thank Karen Shum, Irina Krylova, and Dr. Yan Chang forpreparation and purification of MIS ligand and for initial contribu-tions to this project. We thank Dr. Pat O’Farrell and Peter Follettefor use and instruction of wide field microscopy and imagingsystem. We thank Drs. Jill Helms and Mike Caterina for commentsand suggestions on this manuscript. Financial support for thesestudies was provided by a predoctoral fellowship from NSF (toL.M.R.), a University of California CRCC fellowship (to Y.H.), anMRC fellowship and NIH training grant (to M.W.N.), and by KO2and RO1 support from NICHD and the March of Dimes BirthDefect Foundation (to H.A.I.).

REFERENCES

Austin, H. B. (1995). DiI analysis of cell migration during mullerianduct regression. Dev. Biol. 169, 29–36.

aarends, W. M., van Helmond, M. J., Post, M., van der Schoot, P. J.,Hoogerbrugge, J. W., de Winter, J. P., Uilenbroek, J. T., Karels, B.,Wilming, L. G., Meijers, J. H., et al. (1994). A novel member ofthe transmembrane serine/threonine kinase receptor family isspecifically expressed in the gonads and in mesenchymal cellsadjacent to the mullerian duct. Development 120, 189–197.

Behringer, R. R. (1995). The mullerian inhibitor and mammaliansexual development. Philos. Trans. R. Soc. London B Biol. Sci.350, 285–288; Discussion, 289.

Blader, P., Rastegar, S., Fischer, N., and Strahle, U. (1997). Cleavageof the BMP-4 antagonist chordin by zebrafish tolloid. Science278, 1937–1940.

Boldin, M. P., Goncharov, T. M., Goltsev, Y. V., and Wallach, D.(1996). Involvement of MACH, a novel MORT1/FADD-interacting protease, in Fas/APO-1- and TNF receptor-inducedcell death. Cell 85, 803–815.

Boudreau, N., Sympson, C. J., Werb, Z., and Bissell, M. J. (1995).Suppression of ICE and apoptosis in mammary epithelial cells byextracellular matrix. Science 267, 891–893.

Brown, D. D., Wang, Z., Furlow, J. D., Kanamori, A., Schwartzman,R. A., Remo, B. F., and Pinder, A. (1996). The thyroid hormone-
induced tail resorption program during Xenopus laevis metamor-phosis. Proc. Natl. Acad. Sci. USA 93, 1924–1929.


Catlin, E. A., Tonnu, V. C., Ebb, R. G., Pacheco, B. A., Manganaro,T. F., Ezzell, R. M., Donahoe, P. K., and Teixeira, J. (1997).Mullerian inhibiting substance inhibits branching morphogen-esis and induces apoptosis in fetal rat lung. Endocrinology 138,790–796.

Coucouvanis, E., and Martin, G. R. (1995). Signals for death andsurvival: A two-step mechanism for cavitation in the vertebrateembryo. Cell 83, 279–287.

di Clemente, N., Wilson, C., Faure, E., Boussin, L., Carmillo, P.,Tizard, R., Picard, J.-Y., Vigier, B., Josso, N., and Cate, R. (1994).Cloning, expression, and alternative splicing of the receptor foranti-Mullerian hormone. Mol. Endocrinol. 8, 1006–1020.

Djehiche, B., Segalen, J., and Chambon, Y. (1994). Ultrastructure ofmullerian and wolffian ducts of fetal rabbit in vivo and in organculture. Tissue Cell 26, 323–332.

Dyche, W. J. (1979). A comparative study of the differentiation andinvolution of the Mullerian duct and Wolffian duct in the maleand female fetal mouse. J. Morphol. 162, 175–209.

Fekete, D. M., Homburger, S. A., Waring, M. T., Riedl, A. E., andGarcia, L. F. (1997). Involvement of programmed cell death inmorphogenesis of the vertebrate inner ear. Development 124,2451–2461.

Ganan, Y., Macias, D., Duterque-Coquillaud, M., Ros, M. A., andHurle, J. M. (1996). Role of TGF beta s and BMPs as signalscontrolling the position of the digits and the areas of interdigitalcell death in the developing chick limb autopod. Development122, 2349–2357.

Gearing, A. J., Beckett, P., Christodoulou, M., Churchill, M.,Clements, J., Davidson, A. H., Drummond, A. H., Galloway,W. A., Gilbert, R., Gordon, J. L., et al. (1994). Processing oftumour necrosis factor-alpha precursor by metalloproteinases.Nature 370, 555–557.

He, W. W., Gustafson, M. L., Hirobe, S., and Donahoe, P. K. (1993).Developmental expression of four novel serine/threonine kinasereceptors homologous to the activin/transforming growth factor-beta type II receptor family. Dev. Dyn. 196, 133–142.

Hsu, D. R., Economides, A. N., Wang, X., Eimon, P. M., andHarland, R. M. (1998). The Xenopus dorsalizing factor Gremlinidentifies a novel family of secreted proteins that antagonizeBMP activities. Mol. Cell. 1, 673–683.

Jacobson, M. D., Weil, M., and Raff, M. C. (1997). Programmed celldeath in animal development. Cell 88, 347–354.

Josso, N. (1992). Hormonal regulation of sexual differentiation.Semin. Perinatol. 16, 279–288.

Josso, N., Cate, R. L., Picard, J. Y., Vigier, B., di Clemente, N.,Wilson, C., Imbeaud, S., Pepinsky, R. B., Guerrier, D., Boussin,L., et al. (1993). Anti-mullerian hormone: The Jost factor. RecentProg. Horm. Res. 48, 1–59.

Macias, D., Ganan, Y., Sampath, T. K., Piedra, M. E., Ros, M. A.,and Hurle, J. M. (1997). Role of BMP-2 and OP-1 (BMP-7) inprogrammed cell death and skeletogenesis during chick limbdevelopment. Development 124, 1109–1117.

Marques, G., Musacchio, M., Shimell, M. J., Wunnenberg-Stapleton, K., Cho, K. W., and O’Connor, M. B. (1997). Produc-tion of a DPP activity gradient in the early Drosophila embryothrough the opposing actions of the SOG and TLD proteins. Cell91, 417–426.

Mishina, Y., Rey, R., Finegold, M. J., Matzuk, M. M., Josso, N.,Cate, R. L., and Behringer, R. R. (1996). Genetic analysis of the
Mullerian-inhibiting substance signal transduction pathway inmammalian sexual differentiation. Genes Dev. 10, 2577–2587.

Muzio, M., Chinnaiyan, A. M., Kischkel, F. C., O’Rourke, K.,Shevchenko, A., Ni, J., Scaffidi, C., Bretz, J. D., Zhang, M., Gentz,R., Mann, M., Krammer, P. H., Peter, M. E., and Dixit, V. M.(1996). FLICE, a novel FADD-homologous ICE/CED-3-like pro-tease, is recruited to the CD95 (Fas/APO-1) death—Inducingsignaling complex. Cell 85, 817–827.

Nachtigal, M. W., and Ingraham, H. A. (1996). Bioactivation ofMullerian inhibiting substance during gonadal development by akex2/subtilisin-like endoprotease. Proc. Natl. Acad. Sci. USA 93,7711–7716.

Nagata, S. (1997). Apoptosis by death factor. Cell 88, 355–365.Parr, B. A., and McMahon, A. P. (1998). Sexually dimorphic

development of the mammalian reproductive tract requires Wnt-7a. Nature 395, 707–710.

Piccolo, S., Agius, E., Lu, B., Goodman, S., Dale, L., and DeRobertis, E. M. (1997). Cleavage of Chordin by Xolloid metallo-protease suggests a role for proteolytic processing in the regula-tion of Spemann organizer activity. Cell 91, 407–416.

Price, J. M., Donahoe, P. K., Ito, Y., and Hendren, W. H. d. (1977).Programmed cell death in the Mullerian duct induced by Mulle-rian inhibiting substance. Am. J. Anat. 149, 353–375.

Reed, J. C. (1997). Double identity for proteins of the Bcl-2 family.Nature 387, 773–776.

Ros, M. A., Piedra, M. E., Fallon, J. F., and Hurle, J. M. (1997).Morphogenetic potential of the chick leg interdigital mesodermwhen diverted from the cell death program. Dev. Dyn. 208,406–419.

Shen, W. H., Moore, C. C., Ikeda, Y., Parker, K. L., and Ingraham,H. A. (1994). Nuclear receptor steroidogenic factor 1 regulatesthe mullerian inhibiting substance gene: A link to the sexdetermination cascade. Cell 77, 651–661.

Su, Y., Shi, Y., Stolow, M. A., and Shi, Y. B. (1997). Thyroidhormone induces apoptosis in primary cell cultures of tadpoleintestine: Cell type specificity and effects of extracellular matrix.J. Cell Biol. 139, 1533–1543.

Takahashi, T., Tanaka, M., Brannan, C. I., Jenkins, N. A., Cope-land, N. G., Suda, T., and Nagata, S. (1994). Generalized lympho-proliferative disease in mice, caused by a point mutation in theFas ligand. Cell 76, 969–976.

Teixeira, J., He, W. W., Shah, P. C., Morikawa, N., Lee, M. M.,Catlin, E. A., Hudson, P. L., Wing, J., Maclaughlin, D. T., andDonahoe, P. K. (1996). Developmental expression of a candidatemullerian inhibiting substance type II receptor. Endocrinology137, 160–165.

Torres, M., Gomez-Pardo, E., Dressler, G. R., and Gruss, P. (1995).Pax-2 controls multiple steps of urogenital development. Devel-opment 121, 4057–4065.

Trelstad, R. L., Hayashi, A., Hayashi, K., and Donahoe, P. K. (1982).The epithelial-mesenchymal interface of the male rate Mullerianduct: Loss of basement membrane integrity and ductal regres-sion. Dev. Biol. 92, 27–40.

Tsuji, M., Shima, H., Yonemura, C. Y., Brody, J., Donahoe, P. K.,and Cunha, G. R. (1992). Effect of human recombinant mullerianinhibiting substance on isolated epithelial and mesenchymalcells during mullerian duct regression in the rat. Endocrinology131, 1481–1488.

Wartenberg, H. (1985). Morphological studies on the role of theperiductal stroma in the regression of the human male Mullerianduct. Anat. Embryol. (Berlin) 171, 311–323.

Watanabe-Fukunaga, R., Brannan, C. I., Copeland, N. G., Jenkins,N. A., and Nagata, S. (1992). Lymphoproliferation disorder in


122 Roberts et al.

mice explained by defects in Fas antigen that mediates apoptosis.Nature 356, 314–317.

Werb, Z. (1997). ECM and cell surface proteolysis: Regulatingcellular ecology. Cell 91, 439–442.

White, E. (1996). Life, death, and the pursuit of apoptosis. Genesand Development 10, 1–15.

Wilson, C. A., di Clemente, N., Ehrenfels, C., Pepinsky, R. B., Josso,N., Vigier, B., and Cate, R. L. (1993). Mullerian inhibitingsubstance requires its N-terminal domain for maintenance ofbiological activity, a novel finding within the transforminggrowth factor-beta superfamily. Mol. Endocrinol. 7, 247–257.

Yeh, W. C., Pompa, J. L., McCurrach, M. E., Shu, H. B., Elia, A. J.,Shahinian, A., Ng, M., Wakeham, A., Khoo, W., Mitchell, K.,


El-Deiry, W. S., Lowe, S. W., Goeddel, D. V., and Mak, T. W.(1998). FADD: Essential for embryo development and signalingfrom some, but not all, inducers of apoptosis. Science 279,1954–1958.

Yokouchi, Y., Sakiyama, J., Kameda, T., Iba, H., Suzuki, A., Ueno,N., and Kuroiwa, A. (1996). BMP-2/-4 mediate programmed celldeath in chicken limb buds. Development 122, 3725–3734.

Zou, H., and Niswander, L. (1996). Requirement for BMP signalingin interdigital apoptosis and scale formation. Science 272, 738–741.

Received for publication November 2, 1998Accepted December 21, 1998


paracrine-mediated apoptosis in reproductive tract development · programmed cell death during...

Documents