Loss of lysine-specific demethylase 1 nonautonomouslycauses stem cell tumors in the Drosophila ovarySusan Eliazer, Nevine A. Shalaby, and Michael Buszczak1

Department of Molecular Biology, University of Texas Southwestern Medical Center, Dallas, TX 75390-9148

Edited by Margaret Tatnall Fuller, Stanford University School of Medicine, Stanford, CA, and approved March 21, 2011 (received for review October 21, 2010)

Specialized microenvironments called niches keep stem cells in anundifferentiated and self-renewing state. Dedicated stromal cellsform niches by producing a variety of factors that act directly onstem cells. The size and signaling output of niches must be finelytuned to ensure proper tissue homeostasis. Although advanceshave been made in identifying factors that promote niche cellfate, the mechanisms that restrict niche cell formation duringdevelopment and limit niche signaling output in adults remainpoorly understood. Here, we show that the histone lysine-specificdemethylase 1 (Lsd1) regulates the size of the germline stem cell(GSC) niche in Drosophila ovaries. GSC maintenance depends onbone morphogenetic protein (BMP) signals produced by a smallcluster of cap cells located at the anterior tip of the germarium.Lsd1 null mutant ovaries carry small germline tumors containingan expanded number of GSC-like cells with round fusomes thatdisplay ectopic BMP signal responsiveness away from the normalniche. Clonal analysis and cell type-specific rescue experimentsdemonstrate that Lsd1 functions within the escort cells (ECs) thatreside immediately adjacent to cap cells and prevents them fromectopically producing niche-specific signals. Temporally restrictedgene knockdown experiments suggest that Lsd1 functions bothduring development, to specify EC fate, and in adulthood, to pre-vent ECs from forming ectopic niches independent of changes incell fate. Further analysis shows that Lsd1 functions to repressdecapentaplegic (dpp) expression in adult germaria. The role ofLsd1 in regulating niche-specific signals may have important impli-cations for understanding how disruption of its mammalian homo-log contributes to cancer and metastasis.

Many adult tissues such as the skin, intestine, and hematopoi-etic system experience constant cell turnover. The homeo-

stasis and function of these organs depend on the self-renewingcapacity of stem cells. Adult stem cells are often maintained inspecialized microenvironments called niches (1). The correct bal-ance between stem cell self-renewal and stem cell daughter dif-ferentiation depends on the exquisite regulation of niche size andsignaling output.The germline stem cells (GSCs) of the Drosophila ovary have

provided many insights into the functional relationships thatexist between stem cells and their niches (2). Ovaries are com-posed of tube-like structures known as ovarioles. Two to threeGSCs reside at the tip of each ovariole in a structure called thegermarium (Fig. 1A). Within each germarium, five to seven so-matic cap cells form the functional GSC niche. These cellsproduce Decapentaplegic (Dpp), a bone morphogenetic protein-like molecule, which initiates a signal transduction cascadewithin GSCs that serves to repress the transcription of the dif-ferentiation factor bag of marbles (bam) (3–5). Escort cells(ECs), also known as inner germarial sheath cells, lie adjacent tothe cap cells and line the anterior region of germaria (6). Thesecells do not normally express niche signals and are thought tosupport the early differentiation of germline cysts (6). Ectopi-cally expressing dpp throughout somatic cells blocks germlinedifferentiation, resulting in the formation of GSC tumors (4).Therefore, limiting the number of cells that produce dpp appearsessential for the normal functional output of the ovary.Recent work has shown that ectopic expression of activated

Notch within somatic cells results in a marked increase in the

number of cap cells (7, 8). The increased number of cap cellssubsequently leads to an expansion of the GSC population. Deltaexpressed by terminal filament cells of the developing gonadactivates Notch in the adjacent somatic cells but not in theremaining somatic cells interspersed among the germ cells (7).Activation of Notch within adult ECs does not cause these cellsto adopt a cap cell fate, whereas overexpression of dpp alone inadult ECs in the absence of expanded Notch signaling supportsGSC maintenance (7). Thus, Notch controls cell fate decisionswithin the developing gonad.Two additional pathways regulate dpp expression within adult

ovaries. Disruption of the Janus kinase/Signal transducer andactivator of transcription (Jak/Stat) pathway results in a GSC lossphenotype, whereas activation of the pathway within ECs leads togermline tumor formation marked by expanded Dpp respon-siveness within germ cells (6, 9, 10). The epidermal growth factor(EGF) pathway also acts to regulate the signaling output of theniche. Stet, an EGF-processing molecule, acts in germline cyststo promote the production of EGF receptor (EGFR) ligands,including Spitz, Keren, and Gurken (11). These molecules acti-vate the RAS-RAF-MEK-MAPK pathway within surroundingsomatic cells, which, in turn, represses the transcription of dally,a factor that regulates Dpp transport and stability (12, 13). Byrepressing dally expression, the EGF pathway serves to restrictDpp signaling to the most anterior region of the germarium (11).This pathway also plays a central role in a feedback loop thatcoordinates somatic cell survival and germ cell proliferation dur-ing development (14).Alterations within local chromatin environments likely underlie

the coordinated specification of cell fate programs within the de-veloping gonad and may help to regulate the homeostatic func-tion of ovarian cells in adulthood. Here, we show that absence oflysine-specific demethylase 1 [Lsd1/Su(var)3-3/CG17149] resultsin GSC tumor formation attributable to an expansion of nichesignaling. Further results indicate that Lsd1 acts to control nichesize both during development and in adulthood.

ResultsWe sought to identify chromatin-associated factors that regulateadult GSC behavior. The histone demethylase Lsd1 emergedas a likely candidate based on its role in various developmentalprocesses. In humans, loss of Lsd1 has been linked to several high-risk cancers (15–17) and Lsd1 has recently been shown to regulatethe transcription of TGF β1, a Dpp homolog, negatively (18).Previous work has shown that Drosophila Lsd1 mutants displaymale and female sterile phenotypes and defects in heterochro-matin formation (19, 20). The earliest steps of germline cystdevelopment appeared severely disrupted in Lsd1ΔN null allelehomozygotes, resulting in the formation of small ovaries (19).

Author contributions: S.E. and M.B. designed research; S.E., N.A.S., and M.B. performedresearch; S.E., N.A.S., and M.B. analyzed data; and M.B. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. E-mail: michael.buszczak@UTSouthwestern.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1015874108/-/DCSupplemental.

7064–7069 | PNAS | April 26, 2011 | vol. 108 | no. 17 www.pnas.org/cgi/doi/10.1073/pnas.1015874108

Dow

nloa

ded

by g

uest

on

Nov

embe

r 27

, 202

0

Lsd1 Mutants Display Small GSC Tumors. To characterize the Lsd1mutant ovarian phenotype further, we stained WT and Lsd1ΔN

mutant ovaries with the germline marker Vasa and the fusomemarker Hts (Fig. 1 B and C). The fusome is a specialized or-ganelle that appears round in GSCs most of the time but becomesbranched as GSC daughters move away from the niche and formmulticellular cysts (21–23). In contrast to controls (average = 2,range: 1–3, n = 30 germaria), Lsd1 mutant ovaries contained anincreased number of single undifferentiated GSC-like cells withround fusomes (average = 26, range: 6–79, n = 79 germaria).These single cells underwent cell division as indicated by phos-pho-histone H3 staining (Fig. 1 D and E). The average overallsize of these Lsd1 mutant tumors did not increase over time,however, because of programmed cell death within the germline(Fig. 1 F and G).

Lsd1 Functions in a Nonautonomous Manner. To begin to evaluatethe function of Lsd1 in the germarium, we generated a polyclonalantibody to the N terminus of the Lsd1 protein. This antibodyrevealed ubiquitous Lsd1 expression throughout the germarium(Fig. 2A and Fig. S1). As expected for a histone demethylase, theprotein predominantly localized to the nuclei of all cells exam-ined. Interestingly, Lsd1 expression appeared highest in the GSCsbut was also clearly present in the ECs and follicle cells of thegermarium.The GSC tumors within Lsd1 mutant ovaries could be caused

by defects in the intrinsic programming of GSCs or by extrinsicdefects in the surrounding somatic cells. We performed clonal

analysis and cell-specific rescue experiments to distinguish be-tween these possibilities. First, we induced negatively marked Lsd1mutant clones in an otherwise heterozygous background in adultsusing FRT/FLP-mediated mitotic recombination (Fig. 2 B–E). In-terestingly, we found that negatively marked Lsd1mutant germlineclones differentiated into morphologically normal egg chamberswithout any apparent block in differentiation, even over long peri-ods of time (Fig. 2B andC). Furthermore,Lsd1ΔNGSC clones weremaintained at levels similar to control clones (Fig. 2D), demon-strating that Lsd1 was not required within the germline for GSCmaintenance. Instead, these results suggest that Lsd1 acts within adifferent cell type (i.e., in a cell-nonautonomous manner) to controlgermline differentiation.We then examined Lsd1 mutant follicle cell clones. These also

appeared normal and were able to envelop germline cysts fullywithout any obvious defects (Fig. 2E). The absence of Lsd1 in thefollicle cells did not result in an abnormal number of GSCs.Furthermore Lsd1ΔN follicle stem cell (FSC) clones were main-tained over long periods of time (Fig. 2D). Together with thegermline clone data, these experiments suggest that Lsd1 func-tions in either ECs or cap cells to limit the number of GSCs in thegermarium. To distinguish between these two possibilities, weknocked down Lsd1 expression using RNAi in a cell-specificmanner (24). Reducing Lsd1 levels in cap cells and terminal fila-ment cells using Lsd1-RNAi in combination with hedgehog (hh)-gal4 (25, 26) (for expression, see Fig. 4C) did not disrupt thenormal morphology of the germarium (Fig. 2F and Fig. S1). Incontrast, Lsd1-RNAi driven by c587-Gal4, which expresses GAL4in most somatic cells in the developing gonad but becomes largelyrestricted to ECs and early follicle cells in adults (27, 28) (forexpression, see Fig. 4G), phenocopied Lsd1mutants (Fig. 2G andFig. S1), resulting in the formation of GSC-like tumors within allexamined germaria. Furthermore, driving Lsd1 WT transgeneswith c587-Gal4 rescued the Lsd1 null mutant phenotype so thatthe normal morphology of mutant germaria and ovarioles wasfully restored in every female tested (Fig. 2H and Fig. S1). Giventhat reducing or eliminating Lsd1 function within follicle cells(Fig. 2 D and E) or cap cells (Fig. 2F) did not result in a pheno-type, the knockdown and rescue experiments using c587-Gal4strongly suggest that Lsd1 functions within ECs to limit thenumber of GSCs. Interestingly, the fully mature eggs produced byLsd1 mutant females expressing rescuing transgenes remainedsterile, suggesting that Lsd1 may also function outside of the ECsbut in a manner unrelated to the expanded GSC phenotype.Perhaps Drosophila Lsd1 has an analogous function to its Cae-norhabditis elegans homolog, which is required for germlinemaintenance over multiple generations (29). Regardless of thisadditional phenotype, the clonal loss-of-function and cell type-specific knockdown and rescue experiments presented hereclearly demonstrate that Lsd1 is required in somatic cells (likelyECs) for nonautonomous control of the number of GSC-like cellswithin the germarium.

Lsd1 Limits Dpp Signaling Within the Germarium. We sought to un-derstand further how Lsd1 regulated the differentiation of GSCdaughters. GSCs express the translational repressor Nanos, a fac-tor essential for GSC maintenance. On displacement away fromthe cap cell niche, GSC daughters express Bam, which repressesthe translation of nanos in differentiating cysts (30). Costainingcontrol and Lsd1 mutant germaria for Nanos and Bam revealedthat less than 1.5% of Lsd1 mutant germaria expressed detectablelevels of Bam (Fig. 3B and Fig. S2). Instead, most germline cellscontinued to express Nanos, indicating that loss of Lsd1 preventsGSC daughters from differentiating into cystoblasts and multi-cellular cysts.To test whether Lsd1 mutant germline cells were capable of

forming multicellular cysts, we expressed bam in Lsd1 mutantovaries using an inducible transgene. Previous studies showedthat bam expression is both necessary and sufficient for germ celldifferentiation (31, 32). Expression of bam in Lsd1 mutantsresulted in the formation of multicellular cysts that contained

Fig. 1. Disruption of Lsd1 results in the formation of GSC-like tumors. (A)Illustration of a WT Drosophila germarium. The cap cells, which form theGSC niche, are located at the anterior tip of the germarium (dark blue). Thefusome (yellow) changes from a predominantly round structure in GSCs toa highly branched structure in developing cysts. The ECs (red) line the an-terior region of the germarium. (B and C) Germaria immunostained for Hts(green) and Vasa (red). (B) WT germaria contain two to three GSCs and fiveto seven cap cells. (C) Lsd1ΔN mutants display an expanded number of GSC-like cells with single round fusomes. WT (D) and Lsd1ΔN (E) homozygousgermaria stained for phosphorylated histone H3 (red). Positive stainingrevealed that Lsd1 mutant GSC-like cells continue to divide. WT (F) andLsd1ΔN (G) mutant germaria stained using an antibody against activatedCaspase 3 (red). Cells undergoing cell death are rarely observed in WT ger-maria, but cell death occurs in Lsd1 mutant samples. In all panels, DNA islabeled with DAPI (blue). (Scale bars, 10 μm.)

Eliazer et al. PNAS | April 26, 2011 | vol. 108 | no. 17 | 7065

DEV

ELOPM

ENTA

LBIOLO

GY

Dow

nloa

ded

by g

uest

on

Nov

embe

r 27

, 202

0

branched fusomes (Fig. 3D), indicating that Lsd1mutant germlinecells can undergo differentiation and form multicellular cysts. Thisresult strongly suggests that the tumorous phenotype exhibited byLsd1mutants is caused by failure to initiate a proper differentiationprogram within GSC daughters.Given the previous findings that Dpp signaling represses Bam

expression in GSCs (3–5), we considered the possibility thatectopic Dpp pathway activity might account for the absence ofBam expression in Lsd1 mutants. To test this idea, we crosseda positive reporter of Dpp signaling, the Dad-LacZ enhancertrap, into the Lsd1ΔN mutant background. Normally, high levelsof Dad-LacZ expression are limited to the two to three GSCsimmediately adjacent to the cap cells (Fig. 3E). Although theoverall levels of Dad-LacZ expression were not as high as incontrol GSCs, we found that the number of Dad-LacZ–positivecells was greatly expanded in Lsd1 mutant germaria (100%, n >100 germaria) (Fig. 3F), suggesting that greater Dpp signalingaccounts for the increased number of GSC-like cells in Lsd1mutant ovaries. Consistent with this idea, two different dppmutations partially suppressed the Lsd1 phenotype (Fig. 3 G andH), resulting in an increased number of germline cysts withbranched fusomes and maturing egg chambers (Fig. S2). Fur-thermore, the expression of two different dpp-RNAi transgenesstrongly suppressed the c587-Gal4 > Lsd1-RNAi-induced GSCtumor phenotype (Fig. 3 I and J and Fig. S2).

Lsd1 Functions During Development to Specify Somatic Cell Fate.Loss of Lsd1 did not appear to result in changes in somatic cellnumbers within developing gonads or adult germaria (Fig. S3).We considered the possibility that Lsd1 functioned in the so-matic precursor cells of the developing gonad when ECs and capcells are being specified. To test whether Lsd1 mutant somaticcells adopt inappropriate fates, we compared the expression ofseveral cap cell-specific markers. Cap cells and terminal filamentcells express high levels of Lamin C and hh (7, 27). CB03410,a previously identified protein trap line (33), is also expressed inadult cap cells and terminal filament cells. In Lsd1 mutants, theexpression of all three markers expanded to most of the somaticcells within the germarium (Fig. 4 A–F), indicating that Lsd1mutant ECs exhibit characteristics of cap cells. To determinewhether Lsd1 mutant ECs completely switch their identity, weexamined c587-Gal4 expression within WT and Lsd1 mutantadult ovaries (Fig. 4 G and H). Unlike WT adult germaria, whichexhibited virtually exclusive expression of c587-Gal4 within ECs

and early follicle cells, Lsd1 mutants also displayed c587-Gal4expression in cap cells. Thus, Lsd1 mutant ECs and cap cells donot differentiate properly and display characteristics of both celltypes in adult germaria.Several signaling pathways have been implicated in the for-

mation and regulation of the GSC niche (7–13). For example,Delta from newly formed terminal filament cells activates Notchsignaling and induces cap cell identity within a small number ofsomatic cells in the developing gonad (7). The EGFR pathwayhas also been implicated in regulating the development of theovary and niche output in adults (11, 14). We did not observegenetic interactions between Lsd1 and Notch (Fig. S3). Fur-thermore a transcriptional reporter of Notch activity, E(spl)mβ-CD2 (34, 35), exhibited a normal pattern of expression in Lsd1pupal gonads, suggesting that loss of Lsd1 does not result ininappropriate derepression of Notch target genes within thedeveloping gonad (Fig. 4 I and J). Likewise, we also found thatactivation of the EGFR-MAPK pathway does not change in Lsd1mutant gonads or in adult ovaries, based on the expression ofphosphorylated extracellular signal-regulated kinase (pERK)(Fig. 4 K and L and Fig. S3). Therefore, Lsd1 does not appear tointeract with or influence the activity of these signaling pathwayswithin developing gonads.

Lsd1 Functions During Development and in Adults to Limit Niche Size.The role of Lsd1 in regulating EC fate does not preclude thepossibility that this histone demethylase continues to restrictGSC number in adult ovaries independent of any developmentaldefects. To determine whether Lsd1 acts only during de-velopment, we took advantage of the temperature sensitivity ofthe gal4 system and performed a series of temperature shiftexperiments. In these experiments, Lsd1 expression was specifi-cally knocked down in somatic cells using Lsd1 RNAi driven byc587-Gal4 (Fig. 5). Quantitation of phenotypes is shown in Fig.S4. First, females were raised at 29 °C during larval and pupaldevelopment and then either kept at 29 °C or shifted down to18 °C for 7 d immediately after eclosion. Consistently, femalesraised and maintained at 29 °C exhibited a pronounced GSCtumor phenotype with no signs of proper egg chamber de-velopment (Fig. 5A and Fig. S4). Females shifted down to 18 °Cfor 7 d displayed signs of germline differentiation, however.Ovaries from these females often contained ovarioles witha number of developing germline cysts that were fully encapsu-lated by follicle cells (Fig. 5B and Fig. S4). These results indicate

Fig. 2. Lsd1 functions in a nonautonomous manner to regulate GSC numbers within the ovary. (A) WT germarium stained for Lsd1 (green), Vasa (red), andDNA (blue). (B and C) Negatively marked germline clones stained for GFP (red) and Hts (green). Lsd1ΔN homozygous germline clones (dotted lines) differentiateto form cysts with branched fusomes and morphologically normal egg chambers. (D) Graph shows the percentage of control and Lsd1 mutant stem cell clonesmaintained after clone induction. The solid red line refers to control germline clones, the dotted red line refers to Lsd1ΔN germline clones, the solid blue linerefers to control follicle cell clones, and the dotted blue line refers to Lsd1ΔN follicle cell clones. (E) Lsd1ΔN homozygous follicle cells (dotted lines) do not exhibita discernible phenotype. Germaria from UAS-Lsd1RNAi/+; hh-gal4/+ (F) and c587-Gal4/+; UAS-Lsd1RNAi/+ (G) females stained for Hts (green), Vasa (red), andDNA (blue). (H) Ovarian cells from c587-Gal4/+; UAS-Lsd1/+; Lsd1ΔN/Lsd1ΔN females stained for Hts (green), Vasa (red), and DNA (blue). (Scale bars, 10 μm.)

7066 | www.pnas.org/cgi/doi/10.1073/pnas.1015874108 Eliazer et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 27

, 202

0

that restoration of Lsd1 function during adulthood can rescueunderlying developmental defects that result from the absenceof Lsd1 during larval and pupal development.

If Lsd1 functions to limit the size of the GSC niche in adults,knocking down Lsd1 expression in females specifically after theyeclose would be predicted to result in a GSC expansion pheno-type. To test this possibility, we raised c587-Gal4; UAS-Lsd1-RNAi females at 18 °C. Ovaries from these females appearednormal and did not exhibit an expanded number of GSCs, evenafter 7 d as adults (Fig. 5C). We observed a striking GSC tumorphenotype when we shifted these females to 29 °C for 7 d aftereclosion, however (Fig. 5D). Lamin C staining showed that thisphenotype was not accompanied by changes in EC identity (Fig.S4). These data demonstrate that Lsd1 functions to limit dppsignaling within germaria during adulthood.

Loss of Lsd1 Results in Increased dpp but Not dally Expression. Al-though our analysis of pERK suggests that loss of Lsd1 does notlead to obvious changes in EGFR signaling (Fig. 4 and Fig. S3),this experiment does not rule out the possibility that Lsd1influences the transcriptional output of the EGFR pathway. Inadults, activation of the EGFR pathway limits Dpp signalingoutside of the niche by repressing the transcription of dally (11),a glypican that facilitates Dpp transport and stability (12, 13). Todetermine whether Lsd1 specifically represses the expression ofgenes involved in promoting Dpp signaling, or perhaps dpp itself,we performed a number of RT-PCR–based experiments. Tocontrol for differences in the developmental state of the samples,we crossed the Lsd1ΔN mutation into a bamΔ86 mutant back-ground. We observed no difference in dally mRNA levels be-tween bamΔ86 and Lsd1ΔN bamΔ86 double-mutant germaria,further suggesting that Lsd1 functions independent of the EGFRpathway (Fig. 6A). In contrast, dppmRNA levels were noticeablyelevated in Lsd1 mutant adult germaria (Fig. 6B). We also foundthat ectopic expression of dpp driven by c587-Gal4 during de-velopment resulted in the expanded expression of the cap cellmarker Lamin C (Fig. S5). These observations suggest that mis-regulation of dpp itself accounts for the phenotypes observed inLsd1 mutants.

DiscussionPrevious work has focused on identifying signaling pathways thatspecify niche cell identity. Equally important in sculpting a fullyfunctional stem cell microenvironment is preventing cells outsidethe normal niche from producing niche-specific signals in aninappropriate manner. The work presented here indicates thatthe conserved histone demethylase Lsd1 performs such a func-tion in the somatic cells of the Drosophila ovary.Lsd1 homologs regulate heterochromatin formation and gene

expression across species (18–20, 29, 36–38). Our results indicatethat Lsd1 acts nonautonomously to limit GSC numbers withinthe Drosophila ovary. These conclusions are based on severallines of experimental evidence. Clonal analyses demonstrate thatLsd1 does not function within the germline or in follicle cells in

Fig. 4. Loss of Lsd1 results in somatic cell fatechanges. WT (A) and Lsd1ΔN homozygous (B) germariastained for Lamin C (green). The arrows indicate ECs.UAS-GFP; hh-gal4 (C) and UAS-GFP; hh-gal4, Lsd1ΔN/Lsd1ΔN (D) germaria stained for the expression of GFP(green). CB03410 (E) and CB03410; Lsd1ΔN/Lsd1ΔN (F)germaria stained for the expression of GFP (green).c587-Gal4; UAS-GFP (G) and c587-Gal4; UAS-GFP;Lsd1ΔN/Lsd1ΔN (H) germaria stained for the expressionof GFP (green) (arrows point to cap cells). Late larvalWT (I and K) and Lsd1ΔN homozygous (J and L) femalegonads stained for the E(spl)mβ-CD2 reporter (I and J)or pERK (K and L). Vasa labels the germline (red in A–J),and Hts stains the fusome (red in K and L). DNA is la-beled with DAPI in all panels (blue). (Scale bars, 10 μm.)

Fig. 3. Loss of Lsd1 results in expanded Dpp signaling. WT (A) and Lsd1ΔN

(B) homozygous mutant germaria immunostained for Nanos (green) andBam (red). Only 1.5% of Lsd1ΔN mutant germaria (n = 244) express Bam. hs-bam/+; Lsd1ΔN/Lsd1ΔN germaria before heat shock (C) and 2 d after heatshock (D) stained for Hts (green) and Vasa (red). (E) High levels of Dad-LacZ(green) expression are normally restricted to GSCs in WT germaria. (F) Lsd1mutants exhibit expanded Dad-LacZ expression in germline cells. dpphr56/+;Lsd1ΔN/Lsd1ΔN (G) and dpphr92/+; Lsd1ΔN/Lsd1ΔN (H) germaria stained for Hts(green) and Vasa (red). Both dpp mutant alleles dominantly suppress theLsd1 phenotype, leading to the appearance of cysts with branched fusomes(arrowheads). Germaria from c587-Gal4/+; UAS-Lsd1-RNAi/+ (I) and c587-Gal4/+; UAS-Lsd1-RNAi/+; UAS-dpp-RNAi/+ (J) females dissected 1 d aftereclosion stained for Hts (green) and Vasa (red). Reduction of dpp expressionby RNAi dramatically suppressed the tumorous phenotype induced by Lsd1-RNAi. DNA is labeled with DAPI (blue in A–D and G–J). (Scale bars, 10 μm.)

Eliazer et al. PNAS | April 26, 2011 | vol. 108 | no. 17 | 7067

DEV

ELOPM

ENTA

LBIOLO

GY

Dow

nloa

ded

by g

uest

on

Nov

embe

r 27

, 202

0

regard to the regulation of GSC number or GSC daughter dif-ferentiation. Furthermore Lsd1 mutant GSCs and FSCs aremaintained for long periods of time, indicating that Lsd1 is notrequired within stem cells for their maintenance in the ovary.Strikingly, RNAi knockdown of Lsd1 within ECs phenocopiesthe expanded GSC phenotype of Lsd1 null mutants. In addition,independent Lsd1 transgenes rescue the Lsd1ΔN tumorous phe-notype when expressed in the somatic cells that line the anteriorregion of the germarium. These data indicate that Lsd1 functionswithin the somatic cells of the germarium to limit the size of thefunctional GSC niche.Previous results show that the Notch pathway helps to specify

cap cells within pupal gonads. Delta expressed by terminal fila-ment cells induces Notch activation within other adjacent somaticcells. Notch activation results in the specification of cap cell fateand the subsequent expression of dpp by these cells. The limitedexpression of Delta within the terminal filament cells of the de-veloping gonad provides a simple mechanism for restricting thenumber of cap cells (7). We find that loss of Lsd1 results in theexpanded expression of cap cell markers throughout the adultgermarium, albeit at lower levels than typically observed in WTcap cells. This observation raises the possibility that additionalfactors help to repress these markers in ECs. We considered thepossibility that Notch target genes become inappropriately dere-pressed far from the Delta-expressing terminal filament cells inthe absence of Lsd1. A Notch transcriptional reporter displaysa normal expression pattern inLsd1mutant gonads, however (Fig.4). Furthermore, temporally restricted knockdown experimentssuggest that Lsd1 limits niche signaling in adults independent ofany role it has in the developing gonad (Fig. 5). Ectopic Notchactivation within adult ECs does not result in expanded nichesignaling (7). Based on these data, we conclude that Lsd1 func-tions independent of the Notch pathway in the Drosophila ovary.Similarly, our data indicate that Lsd1 and the EGFR pathway

do not directly cooperate to regulate niche size. Loss of Lsd1

does not alter pERK expression in the developing gonad or inthe adult germarium. Furthermore, the mRNA levels of theEGFR target gene dally remain unchanged in the absence ofLsd1. Previous studies showed that the Jak/Stat pathway alsoregulates Dpp signaling in adults; however, unlike Lsd1, it doesnot have a developmental role in niche formation (9, 10). De-spite these findings, we cannot rule out the possibility that Lsd1interacts with the Jak/Stat pathway at some level in adult ger-maria. Our results suggest that Lsd1 directs EC cell formationand limits Dpp signaling through a previously unrecognizedmechanism that does not involve the Notch and EGF pathways.Strikingly, loss of Lsd1 results in elevated levels of dpp mRNA

within adult ovaries. This finding is consistent with the observedexpansion of Dad-LacZ expression in Lsd1 mutant germaria.Ectopic expression of dpp within somatic cells during gonad de-velopment results in a greater number of Lamin C-expressingsomatic cells. Based on all the findings presented here, we favora model in which Lsd1 represses dpp transcription outside thenormal niche reiteratively during development and in adult-hood. Within the developing gonad, loss of Lsd1 results in ex-panded dpp expression leading to perturbations in normal capcell and EC development. In other tissues, dpp expression ismaintained through autoregulatory mechanisms (39–41). Per-haps loss of Lsd1 results in low levels of inappropriate dpp ex-pression that become reinforced through similar autoregulatorymechanisms. Restoration of normal Lsd1 activity in adultsappears sufficient to block dpp activity outside of the normalniche caused by defects in EC differentiation, suggesting thatLsd1 function and the genes it targets for repression are likely tobe the same in the developing gonad and in adults.Recent studies show that mammalian Lsd1 directly targets TGF-

β1 for transcriptional repression (18) and has cell-autonomousroles in cancer (15–18). Given the possible links between cancerand stem cells (42) and the observation that Lsd1 has a conservedrole in regulating intercellular signaling molecules, it will be im-portant to determine whether Lsd1 and other chromatin factorshave additional nonautonomous functions that contribute to stemcell maintenance, tumorigenesis, and metastasis.

MethodsDrosophila Stocks. Drosophila stocks were maintained at room temperatureon standard cornmeal-agar medium unless specified otherwise. The follow-ing fly strains were used in this study:w1118 was used as a control; Lsd1ΔN andUAS-Lsd1 (19) were provided by N. Dyson (Massachusetts General HospitalCancer Center, Charlestown, MA); hs-bam (32) and hs FLP; FRT2A histone GFP(43) were provided by D. McKearin (Howard Hughes Medical Institute, ChevyChase, MD); hh-Gal4 (44) was provided by J. Jiang (University of TexasSouthwestern, Dallas, TX); dpphr56(45), c587-Gal4 (28), Dad-LacZ (46), andCB03410 (33) were provided by A. Spradling (Carnegie Institute for Science,Baltimore, MD); and E(spl)mβ-CD2 (7, 47) was provided by D. Drummond–Barbosa (The Johns Hopkins School of Public Health, Baltimore, MD). dpphr92,N55e11, FRT2A, and UAS-GFP as well as UAS-dpp-RNAi (BL-31530 and BL-31531) lines were obtained from the Bloomington Stock Center. UAS-Lsd1-RNAi was obtained from the National Institute of Genetics, Japan.

Immunostaining. Adult ovaries were dissected in Grace’s medium and fixed in4% (vol/vol) paraformaldehyde for 10 min. The ovaries were washed withPBT (PBS, 0.5% BSA, and 0.3% Triton-X 100) and stained with primary an-tibody overnight at 4 °C. The ovaries were washed and incubated in sec-ondary antibody at room temperature for 5 h. Ovaries were then washedagain and mounted in Vectashield containing DAPI (Vector Laboratories).

The following primary antibodies were used: mouse anti-1B1 (Hts) (1:20)(48) and mouse anti-Lamin C LC28.26 (1:20) (49) (Developmental StudiesHybridoma Bank), goat anti-VASA (1:200) (Santa Cruz Biotechnology), rabbitanti-VASA (1:500) (gift from A. Spradling), mouse anti-BamC A7 (1:20) (50)(gift from D. McKearin), rabbit anti-Nanos (gift from A. Nakamura, RIKEN,Kobe, Japan), rabbit anti-Spectrin (1:1,000) (51) (gift from R. Dubreuil, Uni-versity of Illinois at Chicago), mouse anti-β-galactosidase (1:1,000) (Promega),rabbit anti-GFP (1:1,000) (Invitrogen), guinea pig anti-Lsd1 antibody (1:5,000)and mouse anti-CD2 (1:20) (AbD Serotec), rabbit antiphosphorylated ERK1/2(1:100) and rabbit anti-cleaved Caspase-3 (1:250) (Cell Signaling Technology),guinea pig anti-Traffic Jam (1:5,000) (52) (gift from D. Godt, University of Tor-onto, Toronto, ON, Canada), and rabbit anti–phospho-histone H3 (1:250) (Up-

Fig. 6. Lsd1 mutant germaria display elevated levels of dpp mRNA. Ethi-dium bromide-stained gel shows the products of a RT-PCR on RNA isolatedfrom bamΔ86 and Lsd1ΔN bamΔ86 ovaries using dally- (A) or dpp- (B) specificprimers. No difference in the levels of dally expression was observed, al-though dpp mRNA levels were clearly elevated in the absence of Lsd1. Thepresented gels are representative of three biological replicates.

Fig. 5. Lsd1 functions in adulthood to limit niche activity. (A–D) Germariafrom c587-Gal4/+; UAS-Lsd1-RNAi/+ females raised at either 29 °C (highRNAi) or 18 °C (low RNAi) and shifted after eclosion stained for Hts (green),Vasa (red), and DNA (blue). (Scale bars, 10 μm.)

7068 | www.pnas.org/cgi/doi/10.1073/pnas.1015874108 Eliazer et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 27

, 202

0

state Cell Signaling Solutions). Fluorescence-conjugated secondary antibodies(Jackson Laboratories) were used at a dilution of 1:200. The Student t test wasused (two-tail distribution and two-sample unequal variants; Microsoft Excel2008) to compare the number of ovarioles with branched vs. round fusomesbetween genotypes.

Generation of Anti-Lsd1 Antibody. A sequence corresponding to 1–150 aminoacids of Lsd1 protein was cloned into PROEX (Invitrogen) to produce His6-tagged protein. The protein was expressed in Escherichia coli and purifiedwith Ni-NTA agarose (Invitrogen). Polyclonal antisera were generated inguinea pigs (Covance).

Generation of GSC and FSC Clones. FLP/FRT-mediated mitotic recombinationwas used to generate GSC and FSC clones (53). Adult females of the genotype hs-FLP; FRT2A histone GFP/FRT2A, Lsd1ΔN were heat-shocked at 37 °C for 1 h twicea day for 3 d. hs-FLP; FRT2A histone GFP/FRT2A flies were used as controls. Theovaries were dissected on days 7, 14, and 21 after induction of heat shock.

RNA Isolation and RT-PCR. RNA was isolated from bamΔ86 and Lsd1ΔN bamΔ86

mutant ovaries using TRIzol (Invitrogen). The RNA was treated with DNaseand subjected to RT-PCR reaction using the One Step RT-PCR kit (Qiagen).The primers used to amplify dpp and dally mRNA are as follows:

dpp forward: 5′-GGCTTCTACTCCTCGCAGTGdpp reverse: 5′-TGCTTTTGCTAATGCTGTGCdally forward: 5′-TGACTTGCACGAGGACTACdally reverse: 5′-TAATACGACTCACTATAGGGTGAGGAGATGCAGTTTGCAC

ACKNOWLEDGMENTS. We thank N. Dyson, D. McKearin, A. Nakamura,A. Spradling, R. Dubreuil, D. Drummond–Barbosa, and D. Godt for providingreagents. A. Rothenfluh, P. R. Hiesinger, B. Ohlstein, and members of theM.B. laboratory provided comments on the manuscript. This work wasfunded, in part, by National Institutes of Health Grant GM086647 (to M.B.),March of Dimes Grant 5FY0910 (to M.B.), the E. E. and Greer Garson Fogel-son Endowment (University of Texas Southwestern Medical Center), and byNational Institutes of Health Grant T32GM008203 (to S.E.).

1. Ohlstein B, Kai T, Decotto E, Spradling A (2004) The stem cell niche: Theme andvariations. Curr Opin Cell Biol 16:693–699.

2. Kirilly D, Xie T (2007) The Drosophila ovary: An active stem cell community. Cell Res17:15–25.

3. Chen D, McKearin D (2003) Dpp signaling silences bam transcription directly toestablish asymmetric divisions of germline stem cells. Curr Biol 13:1786–1791.

4. Xie T, Spradling AC (1998) decapentaplegic is essential for the maintenance anddivision of germline stem cells in the Drosophila ovary. Cell 94:251–260.

5. Xie T, Spradling AC (2000) A niche maintaining germ line stem cells in the Drosophilaovary. Science 290:328–330.

6. Decotto E, Spradling AC (2005) The Drosophila ovarian and testis stem cell niches:Similar somatic stem cells and signals. Dev Cell 9:501–510.

7. Song X, Call GB, Kirilly D, Xie T (2007) Notch signaling controls germline stem cellniche formation in the Drosophila ovary. Development 134:1071–1080.

8. Ward EJ, et al. (2006) Stem cells signal to the niche through the Notch pathway in theDrosophila ovary. Curr Biol 16:2352–2358.

9. López-Onieva L, Fernández-Miñán A, González-Reyes A (2008) Jak/Stat signalling in nichesupport cells regulates dpp transcription to control germline stem cell maintenance in theDrosophila ovary. Development 135:533–540.

10. Wang L, Li Z, Cai Y (2008) The JAK/STAT pathway positively regulates DPP signaling inthe Drosophila germline stem cell niche. J Cell Biol 180:721–728.

11. Liu M, Lim TM, Cai Y (2010) The Drosophila female germline stem cell lineage acts tospatially restrict DPP function within the niche. Sci Signal 3:ra57.

12. Guo Z, Wang Z (2009) The glypican Dally is required in the niche for the maintenance ofgermline stem cells and short-range BMP signaling in the Drosophila ovary. Development136:3627–3635.

13. Hayashi Y, Kobayashi S, Nakato H (2009) Drosophila glypicans regulate the germlinestem cell niche. J Cell Biol 187:473–480.

14. Gilboa L, Lehmann R (2006) Soma-germline interactions coordinate homeostasis andgrowth in the Drosophila gonad. Nature 443:97–100.

15. Suikki HE, et al. (2010) Genetic alterations and changes in expression of histonedemethylases in prostate cancer. Prostate 70:889–898.

16. Tsai WW, Nguyen TT, Shi Y, BartonMC (2008) p53-targeted LSD1 functions in repressionof chromatin structure and transcription in vivo. Mol Cell Biol 28:5139–5146.

17. Schulte JH, et al. (2009) Lysine-specific demethylase 1 is strongly expressed in poorlydifferentiated neuroblastoma: Implications for therapy. Cancer Res 69:2065–2071.

18. Wang Y, et al. (2009) LSD1 is a subunit of the NuRD complex and targets themetastasis programs in breast cancer. Cell 138:660–672.

19. Di Stefano L, Ji JY, Moon NS, Herr A, Dyson N (2007) Mutation of Drosophila Lsd1disrupts H3-K4 methylation, resulting in tissue-specific defects during development.Curr Biol 17:808–812.

20. Rudolph T, et al. (2007) Heterochromatin formation in Drosophila is initiated throughactive removal of H3K4 methylation by the LSD1 homolog SU(VAR)3-3. Mol Cell 26:103–115.

21. de Cuevas M, Spradling AC (1998) Morphogenesis of the Drosophila fusome and itsimplications for oocyte specification. Development 125:2781–2789.

22. Lin H, Spradling AC (1995) Fusome asymmetry and oocyte determination in Drosophila.Dev Genet 16:6–12.

23. Lin H, Yue L, Spradling AC (1994) The Drosophila fusome, a germline-specificorganelle, contains membrane skeletal proteins and functions in cyst formation.Development 120:947–956.

24. Brand AH, Perrimon N (1993) Targeted gene expression as a means of altering cellfates and generating dominant phenotypes. Development 118:401–415.

25. Forbes AJ, Lin H, Ingham PW, Spradling AC (1996) hedgehog is required for theproliferation and specification of ovarian somatic cells prior to egg chamberformation in Drosophila. Development 122:1125–1135.

26. Pan L, et al. (2007) Stem cell aging is controlled both intrinsically and extrinsically inthe Drosophila ovary. Cell Stem Cell 1:458–469.

27. Song X, et al. (2004) Bmp signals from niche cells directly repress transcription ofa differentiation-promoting gene, bag of marbles, in germline stem cells in theDrosophila ovary. Development 131:1353–1364.

28. Zhu CH, Xie T (2003) Clonal expansion of ovarian germline stem cells during nicheformation in Drosophila. Development 130:2579–2588.

29. Katz DJ, Edwards TM, Reinke V, Kelly WG (2009) A C. elegans LSD1 demethylasecontributes to germline immortality by reprogramming epigenetic memory. Cell 137:308–320.

30. Li Y, Minor NT, Park JK, McKearin DM, Maines JZ (2009) Bam and Bgcn antagonizeNanos-dependent germ-line stem cell maintenance. Proc Natl Acad Sci USA 106:9304–9309.

31. McKearin DM, Spradling AC (1990) bag-of-marbles: A Drosophila gene required toinitiate both male and female gametogenesis. Genes Dev 4(12B):2242–2251.

32. Ohlstein B, McKearin D (1997) Ectopic expression of the Drosophila Bam proteineliminates oogenic germline stem cells. Development 124:3651–3662.

33. Buszczak M, et al. (2007) The carnegie protein trap library: A versatile tool forDrosophila developmental studies. Genetics 175:1505–1531.

34. Cooper MT, Bray SJ (1999) Frizzled regulation of Notch signalling polarizes cell fate inthe Drosophila eye. Nature 397:526–530.

35. de Celis JF, Bray S (1997) Feed-back mechanisms affecting Notch activation at thedorsoventral boundary in the Drosophila wing. Development 124:3241–3251.

36. Amente S, et al. (2010) LSD1-mediated demethylation of histone H3 lysine 4 triggersMyc-induced transcription. Oncogene 29:3691–3702.

37. Chosed R, Dent SY (2007) A two-way street: LSD1 regulates chromatin boundaryformation in S. pombe and Drosophila. Mol Cell 26:160–162.

38. Hu X, et al. (2009) LSD1-mediated epigenetic modification is required for TAL1function and hematopoiesis. Proc Natl Acad Sci USA 106:10141–10146.

39. Hepker J, Blackman RK, Holmgren R (1999) Cubitus interruptus is necessary but notsufficient for direct activation of a wing-specific decapentaplegic enhancer. Development126:3669–3677.

40. Chanut F, Heberlein U (1997) Role of decapentaplegic in initiation and progression ofthe morphogenetic furrow in the developing Drosophila retina. Development 124:559–567.

41. Yu X, Hoppler S, Eresh S, Bienz M (1996) decapentaplegic, a target gene of thewingless signalling pathway in the Drosophila midgut. Development 122:849–858.

42. Feinberg AP, Ohlsson R, Henikoff S (2006) The epigenetic progenitor origin of humancancer. Nat Rev Genet 7:21–33.

43. Maines JZ, Park JK, Williams M, McKearin DM (2007) Stonewalling Drosophila stemcell differentiation by epigenetic controls. Development 134:1471–1479.

44. Tanimoto H, Itoh S, ten Dijke P, Tabata T (2000) Hedgehog creates a gradient of DPPactivity in Drosophila wing imaginal discs. Mol Cell 5:59–71.

45. Irish VF, Gelbart WM (1987) The decapentaplegic gene is required for dorsal-ventralpatterning of the Drosophila embryo. Genes Dev 1:868–879.

46. Tsuneizumi K, et al. (1997) Daughters against dpp modulates dpp organizing activityin Drosophila wing development. Nature 389:627–631.

47. Dobens L, Jaeger A, Peterson JS, Raftery LA (2005) Bunched sets a boundary for Notchsignaling to pattern anterior eggshell structures during Drosophila oogenesis. DevBiol 287:425–437.

48. Ding D, Parkhurst SM, Lipshitz HD (1993) Different genetic requirements for anteriorRNA localization revealed by the distribution of Adducin-like transcripts duringDrosophila oogenesis. Proc Natl Acad Sci USA 90:2512–2516.

49. Riemer D, et al. (1995) Expression of Drosophila lamin C is developmentally regulated:Analogies with vertebrate A-type lamins. J Cell Sci 108:3189–3198.

50. McKearin D, Ohlstein B (1995) A role for the Drosophila bag-of-marbles protein in thedifferentiation of cystoblasts from germline stem cells. Development 121:2937–2947.

51. Lee JK, Coyne RS, Dubreuil RR, Goldstein LS, Branton D (1993) Cell shape andinteraction defects in alpha-spectrin mutants of Drosophila melanogaster. J Cell Biol123:1797–1809.

52. Li MA, Alls JD, Avancini RM, Koo K, Godt D (2003) The large Maf factor Traffic Jamcontrols gonad morphogenesis in Drosophila. Nat Cell Biol 5:994–1000.

53. Xu T, Rubin GM (1993) Analysis of genetic mosaics in developing and adult Drosophilatissues. Development 117:1223–1237.

Eliazer et al. PNAS | April 26, 2011 | vol. 108 | no. 17 | 7069

DEV

ELOPM

ENTA

LBIOLO

GY

Dow

nloa

ded

by g

uest

on

Nov

embe

r 27

, 202

0