Development 114, 89-98 (1992)Printed in Great Britain © The Company of Biologists Limited 1992

89

Probing spermatogenesis in Drosophila with P-element enhancer detectors

PIERRE GONCZY*, SRIDHAR VISWANATHAN* t and STEPHEN DiNARDO

The Rockefeller University, 1230, York Ave, New York City, NY 10021, USA

•The first two authors contributed equally to this workt Present address: Department of Developmental Biology, Stanford University School of Medicine, 94305, USA

Summary

Formation of motile sperm in Drosophila melanogasterrequires the coordination of processes such as stem celldivision, mitotic and meiotic control and structuralreorganization of a cell. Proper execution of spermato-genesis entails the differentiation of cells derived fromtwo distinct embryonic lineages, the germ line and thesomatic mesoderm. Through an analysis of homozygousviable and fertile enhancer detector lines, we haveidentified molecular markers for the different cell typespresent in testes. Some lines label germ cells or somaticcyst cells in a stage-specific manner during then-differentiation program. These expression patternsreveal transient identities for the cyst cells that had notbeen previously recognized by morphological criteria. Amarker line labels early stages of male but not femalegerm cell differentiation and proves useful in the

analysis of germ line sex-determination. Other lineslabel the hub of somatic cells around which germ linestem cells are anchored. By analyzing the fate of thesomatic hub hi an agametic background, we show thatthe germ line plays some role in directing its size and itsposition in the test is. We also describe how marker linesenable us to identify presumptive cells in the embryonicgonadal mesoderm before they give rise to morphologi-cally distinct cell types. Finally, this collection of markerlines will allow the characterization of genes expressedeither in the germ line or in the soma during spermato-genesis.

Key words: spermatogenesis, Drosophila melanogaster, P-element enhancer detectors.

Introduction

Spermatogenesis in Drosophila melanogaster is anelaborate differentiation program amenable to analysisin the context of the whole organism. The sequence ofcellular events occurring' during male gametogenesishas been extensively described at the ultrastructurallevel by electron microscopy (reviewed by Lindsley andTokuyasu, 1980). All morphologically distinct cell typesand stages present in testes have been recognized. It hasbeen shown that differentiation of the germ cells occurswithin a space delimited by two somatic cells, the cystcells. Gametogenesis hence entails close contact anddevelopmental coordination between cells derived fromtwo distinct embryonic lineages, the germ line and thesomatically derived gonadal mesoderm.

The genetic analysis of spermatogenesis began withthe discovery that the Y chromosome, althoughdispensable for viability, is absolutely required for malefertility (Bridges, 1916). Subsequent studies uncoveredloci on other chromosomes that are required forspermatogenesis (reviewed by Lifschytz, 1987). A fewmale-sterile mutants provide valuable informationabout particular aspects of the differentiation program.For instance, mutations in either benign-gonial-cell-

neoplasm (bgcri) or bag-of-marbles (bam) result inderegulated mitotic proliferation of early germ cells(Gateff, 1982; McKearin and Spradling, 1990). Thesegenes could therefore encode proteins that control thestem cell or mitotic division program. Other mutantsare blocked before entry into meiosis and their analysiscould provide clues about the commitment to this keystep in gametogenesis (Lifschytz, 1978; N. Wolf, P.Wilson and M. Fuller, personal communication; D.Castrillon and S. Wasserman, personal communication;P.G., S.V. and S.D., unpublished data). Finally,mutations in the gene encoding /2-tubulin, a structuralprotein specific to spermatogenesis, alter microtubule-based processes during meiosis and in postmeiotic germcells (Kemphues et al., 1982). Second site mutationsthat fail to complement certain /32-tubulin alleles mayidentify genes encoding other components of thecomplex assembly process occurring in haploid germcells (Regan and Fuller, 1988; Green et al., 1990).

Notwithstanding the analysis of these loci, little isknown about the genetic circuitry controlling theproper execution of spermatogenesis. This lack ofknowledge stems from the very large number of locithat can mutate to cause male sterility, as well as fromtheir pleiotropy (reviewed by Lifschytz, 1987). Thus,

90 P. Gonczy, S. Viswanathan and S. DiNardo

the majority of a series of male-sterile mutations provedto be either weak alleles or spermatogenesis-specificalleles of vital genes (Lifschytz, 1978; Lifschytz andYakobovitz, 1978; Geer et al., 1983; Dybas et al., 1983).This could be a reflection of the sensitivity of thedeveloping germ cells to the alteration of basic cellularfunctions. It could also signify that many productsrequired for spermatogenesis are also needed in otherdevelopmental processes. The large number of male-sterile mutants and their pleiotropy renders a thoroughgenetic analysis quite laborious. Ordering gene actionalong a developmental pathway can be difficult withoutknowing the null phenotype for the gene in question. Inthe case of male-sterile mutants, obtaining null allelesmight often yield lethal mutations whose phenotypeduring spermatogenesis could be studied only in mosaicanimals. In addition, choosing particular male-sterilemutants on which to focus has proved difficult becausemost of them do not display a tight arrest phenotype(see Lifschytz, 1987).

The limitations associated with a classical mutantanalysis argue for a complementary approach to studythis developmental pathway. We, therefore, sought toanalyze spermatogenesis in D. melanogaster with P-element enhancer detectors (O'Kane and Gehring,1987; Bellen et al., 1989; Bier et al., 1989). Enhancerdetectors contain a weak promoter fused to the lacZreporter gene. Upon insertion of the P-lacZ element inthe genome, the weak promoter can be brought underthe influence of a neighbouring enhancer, resulting inspatially and temporally restricted ^3-galactosidase ac-tivity. It has been shown that several of the enhancersdetected in this manner control the transcription of aneighbouring gene in a similar pattern (Fasano andKerridge, 1988; Bier et al., 1989; Wilson et al., 1989).

P-element enhancer detectors can be utilized indifferent ways to probe spermatogenesis. First, they canserve to generate male-sterile lines by insertionalmutagenesis. Compared to a classical approach, thisprovides the advantage of indicating where the mutatedgene might be normally expressed. This is especiallyuseful since male-sterile mutants are not easily categor-ized based on their phenotype alone. Second, enhancerdetection is the method of choice to identify genes thatplay a role in spermatogenesis but that are not easilyrecovered in male-sterile screens. Thus, lethal P-insertion lines can be examined as heterozygotes todetermine whether the essential gene in question isexpressed in male gonads. Viable and fertile linesexpressing lacZ during spermatogenesis can serve toidentify genes that are redundant in function or whosemutant phenotype is not male-sterility. Moreover,viable and fertile lines can also identify genes that areessential for male fertility if the location of the insertionhas not led to disruption of gene function. Lastly,enhancer detector lines provide indispensable molecu-lar markers for each of the labeled cell types and stagesduring spermatogenesis. Such marker lines are crucialin following the fate of specific cells in wild-type and indifferent mutant backgrounds.

We have begun to investigate spermatogenesis in D.

melanogaster by looking at patterned lacZ expression inhomozygous viable and fertile enhancer detector lines.We describe marker lines that label germ cells, somaticcyst cells and all other somatic cells present in testes.This series of lines provides an entry point forcharacterizing genes expressed either in the germline orin the soma during male gametogenesis. Particular lineslabel germ cells or cyst cells in a stage-specific mannerduring their differentiation program. These expressionpatterns reveal transient identities for the somatic cystcells that had not been recognized by ultrastructuralcriteria. We also describe how the ability to follow thefate of specific cells is critical in investigating gonadoge-nesis, and in probing interactions between germ lineand soma during spermatogenesis.

Materials and methods

Enhancer detector linesOver 700 viable and fertile autosomal P-element enhancerdetector lines were generously provided for us by thelaboratories of Yuh-Nung and Lily Jan (Bier et al., 1989; 500lines, preselected against those with a staining pattern in theembryonic nervous system), Margaret Fuller and MatthewScott (150 lines) and Judy Kassis (Kassis, 1990; 60 lines). Oneline described in this paper (line 254, Fig. 6A through 6G) ishomozygous lethal. We also included in this report thestaining pattern of one male-sterile line (line ms-985, Fig. 5C).This line has been identified in our laboratory from acollection of enhancer detector lines generously provided forus by John Merriam and the Drosophila laboratories atUCLA.

All lines in this report except line 254, Fig. 6A through 6G,carried the P-lacW construct described in Bier et al. (1989).Upon lacZ expression, these lines gave rise to nuclear X-galstain in premeiotic germ cells (see for instance Fig. 2B) and insomatic cells (see for instance Fig. 4B), as expected from thepresence of the nuclear targeting signal upstream of the lacZreporter gene. /3-galactosidase was present in both the nucleusand the cytoplasm in postmeiotic germ cells, thereby showingup in the elongating sperm tails (see Fig. 2A and 2B,arrowhead). This is probably due to the reshaping of thespermatid nucleus whose diminished volume might not beable to retain the nuclear targeted /3-galactosidase efficiently.

A few lines carried constructs with sequences from theengrailed gene, including a fragment from the engrailedpromoter, fused to the lacZ reporter gene with an AUGcodon (Kassis, 1990). In addition to the engrailed-specificpatterns discussed in Kassis (1990), many of these lines hadadditional expression patterns dependent on the insertion sitein the genome, effectively behaving as enhancer detectors.Owing to the lack of a nuclear targeting signal, linesexpressing lacZ gave rise in this case to cytoplasmic X-galstain (line 254 in this report, Fig. 6A through 6G).

The majority of the enhancer detector lines also expressedlacZ in other cells in the adult fly. However, since this did notinterfere with their use as marker lines to probe spermatogen-esis, the characterization of expression patterns outside thegonads were not pursued further.

We learned how to recognize better the different cell typespresent in testes during the course of this analysis. Subtleexpression patterns may have been disregarded in the initialphase of this study. In addition, the lines examined have beenpreselected as indicated above. Thus, the frequency of

Spermatogenesis in Drosophila 91

occurence for a particular staining pattern could be onlyapproximated, and is indicated in the text when relevant.

X-gal staining of male gonadsTestes from 1-day-old adults were dissected in DrosophilaRinger's solution, transferred to microtiter plates and fixedfor 15 minutes in 1% glutaraldehyde (Fluka); 50 mM sodiumcacodylate. The tissues were rinsed three times with stainingbuffer (7.2 mM Na2HPO4> 2.8 mM NaH2PO4, 1 mM MgCl2,0.15 M NaCl), left at room temperature for 30 minutes andthen incubated at 37°C for 12 to 16 hours in staining bufferplus 5 mM each of potassium ferro- and ferri-cyanide and0.2% X-gal (5-bromo-4-chloro-3-indolyl-)3-D-galactopyrano-side). The tissues were washed three times in PBS plus 1 mMEDTA, dehydrated in ethanol solutions of increasing concen-trations and mounted in 2:1 Canada Balsam:methyl salicy-late.

Reagents for indirect immunofluorescenceThe rabbit anti-/3-galactosidase polyclonal antibody waspurchased from Cappel. The rabbit anti-/B-tubulin polyclonalantibody was a gift from Elizabeth Raff and is described inKimble et al. (1989); the polyclonal antibody was preadsorbedat its final dilution against an equal volume of fixed 0-3hembryos (Kimble et al., 1989). The fasciclin in monoclonalantibody was a gift from Danny Brower and is described asDA.1B6 in Brower et al. (1981). The £-galactosidasemonoclonal antibody was a gift from Alfonso Martinez-Arias.All rhodamine- and fluorescein-conjugated secondary anti-bodies were purchased from Jackson Laboratories. They werepreadsorbed at their final dilution for 2 hours against an equalvolume of fixed embryos from an overnight collection.

Indirect immunofluorescent anti-f$3-tubulin and anti-fi-galactosidase staining of male gonadsMale gonads from third instar larvae (line 600, see results)were dissected in PEM (0.1 M Pipes, 1 mM MgCl2. 1 mMEGTA, pH 6.9) and fixed for 20 minutes in 4% formaldehyde(EM grade; Polysciences) in PEX (PEM plus 0.1% Triton X-100). The tissues were rinsed in PEX, washed twice for 10minutes in PBX (PBS plus 0.1% Triton X-100), and blockedfor 90 minutes in PBX-2 (PBX plus 2% normal goat serum;Vector). The gonads were then incubated overnight at 4°Cwith 1:100 rabbit anti-/33-tubulin and 1:50 mouse anti-/3-galactosidase primary antibodies in PBX-2, and washed threetimes for 30 minutes in PBX at room temperature. The tissueswere then incubated for 60 minutes at room temperature with1:300 rhodamine-conjugated donkey anti-rabbit and 1:300fluorescein-conjugated goat anti-mouse secondary antibodiesin PBX-2, and washed three times for 30 minutes in PBX andfor 30 minutes in PBS. Stained larval gonads were mounted ina drop of Fluoromount-G (Fisher) under a bridged coverslip.

Indirect immunofluorescent anti-fasciclin III and anti-fi-galactosidase staining of male gonadsFasciclin III antigen is sensitive to fixation and best results areobtained when staining is performed on unfixed tissue(Brower et al., 1981). This double-labeling experimenttherefore involved two successive reactions: firstly staining forfasciclin III, performed on live tissue, followed by fixation anda second staining for ^-galactosidase. Testes from 1-day-oldadults (line 254, see results) were dissected in DrosophilaRinger's solution, incubated for 45 minutes at 37°C with 1:500mouse anti-fasciclin III antibody in PBT-2 (PBS plus 0.1%Tween-20 and 2% normal goat serum), and washed four timesfor 5 minutes at 4°C in PBT (PBS plus 0.1% Tween-20). The

tissues were then incubated for 30 minutes at 37°C with 1:300rhodamine-conjugated goat anti-mouse antibody in PBT-2,washed four times for 5 minutes at 4°C in PBT, and fixed in4% formaldehyde in PBT for 30 minutes at 4°C. Theremainder of the experiment was carried out at roomtemperature. The tissues were washed three times for 5minutes in PBT, incubated for 45 minutes in 1:5000 rabbitanti-/3-galactosidase antibody in PBT, and washed three timesfor 20 minutes in PBT. The tissues were then incubated for 45minutes with 1:300 fluorescein-conjugated donkey anti-rabbitantibody in PBT-2, washed three times for 20 minutes in PBT,counterstained for 2 minutes in 1 /ig/ml Hoechst 33258(Sigma) in PBT, rinsed with PBT and washed for 20 minutesin PBS. The apical tip of the testis was dissected and placed ina drop of Fluoromount-G under a coverslip.

ImmunocytochemistryEmbryos (line 254, see results) were collected for 2 hours andaged for 13 hours at 25°C, then permeabilized, fixed anddevitellinized as described (Mitchison and Sedat, 1983). Theembryos were stored in 100% methanol at —20°C for severaldays. Prior to rehydration, they were treated for 15 minuteswith 3% H2O2 in methanol to block endogenous peroxidaseactivity. Testes from 1-day-old adults (line 57, see Results)and agametic gonads (line 254, see results and below) weredissected in Drosophila Ringer's solution, fixed for 20 minutesin 4% formaldehyde in PBX and washed three times for 10minutes in PBX. The immunocytochemistry procedure was amodified version of that described in Kellerman et al. (1990).The samples were blocked for 90 minutes in Blotto (5%powdered milk in PBX) and incubated overnight at 4°C with1:1000 (embryos) or 1:5000 (testes and agametic gonads)rabbit anti-/3-galactosidase antibody in Blotto. The rest of theexperiment was carried out at room temperature. Thesamples were washed three times for 10 minutes in PBX,incubated for 60 minutes with 1:400 biotin-conjugated goatanti-rabbit antibody (Vector) in Blotto, and washed threetimes for 10 minutes in PBX. They were then incubated for 30minutes with 1:500 horseradish peroxidase-conjugated strep-tavidin (Chemicon) in Blotto, washed three times for 10minutes and once for 30 minutes in PBX. Diaminobenzidine(Polysciences) immunocytochemistry was carried out asdescribed in Kellerman et al. (1990). Finally, the sampleswere counterstained for 2 minutes in 1 j/g/ml Hoechst 33258in PBX, rinsed with PBX, washed for 20 minutes in PBS andmounted in a drop of glycerol.

Obtaining agametic males carrying P-lacZHomozygous oskar301 (Lehman and Niisslein-Volhard, 1986)female virgins were collected, aged at 18°C for four days andcrossed to males carrying P-lacZ (heterozygous males fromline 254, see results). At 18°C, osk301 mothers give rise toviable progeny that lack pole cells, resulting in flies withagametic gonads.

Microscopy and photographyThe samples were examined with a Nikon Optiphot micro-scope equipped with Nomarski optics and epifluorescence. X-gal stains and immunocytochemistry were viewed with bright-field optics for low magnifications (see for instance Fig. 2A)and Nomarski optics for higher power views (see for instanceFig. 2B), and photographed with Kodak Ektar 25 film.Indirect immunofluorescence was photographed with KodakTech Pan film set at ASA 800 for rhodamine and ASA 125 forfluorescein and Hoechst. The phase-contrast view of a liveadult testis (Fig. 1A) was photographed with Kodak Pan-Xfilm.

92 P. Gdnczy, S. Vlswanathan and S. DiNardo

Results

Spermatogenesis in Drosophila melanogasterEach adult testis comprises all stages of spermatogen-esis displayed in a developmental gradient, with theearliest cells located at the apical end (Fig. 1A, api) andthe latest at the terminal end (Fig. 1A, ter) of thegonad. The essential stages of this 10 day differentiationprogram (reviewed by Lindsley and Tokuyasu, 1980)will be briefly reviewed below in order to place the lacZexpression patterns into a meaningful context.

Five to nine germ line stem cells and about twice asmany somatic cyst progenitor cells are anchored arounda hub of somatic cells (Fig. IB, hub) at the apical tip ofthe adult testis. A germ line stem cell (Fig. IB, ste) andtwo neighbouring somatic cyst progenitor cells (Fig. IB,cyp), also acting as stem cells, divide asymmetrically(Fig. IB, asy). This yields a primary gonial cell (Fig. IB,spg) and two cyst cells (Fig. IB, eye), respectively. Thetwo cyst cells and the corresponding enclosed germ cellform a cyst, the fundamental unit of this differentiationprogram.

The primary gonial cell then undergoes four mitoticdivisions (Fig.IB, mit), while the cyst cells no longerdivide. This yields a cyst of 16 primary spermatocytes(Fig. IB, spc), interconnected by ring canals as a resultof incomplete cytokinesis. The spermatocytes thenenter an extended growth phase (Fig. IB, gro), afterwhich they undergo the two meiotic divisions (Fig. IB,mei). The 64 haploid spermatids (Fig. IB, spt) theninitiate a program of dramatic morphological changes(Fig. IB, mor) in which virtually all cellular organellesare restructured and the flagellar tail is assembled. Atthe same time, the two cyst cells become structurallydistinct. The head-cyst cell (Fig. IB, cyh) surrounds andinterdigitates with the sperm heads whereas the tail-cystcell (Fig. IB, cyt) elongates along with the growingsperm tails. At the completion of spermatid differen-tiation, the head-cyst cell is entrapped by a specializedepithelial cell (Fig. IB, tec) at the base of the testis. Theprocess of individualization follows, during which singlespermatozoa become invested in their own membrane.Coiling of the sperm bundle ensues, followed by releaseof motile spermatozoa (Fig. IB, spz) in the seminalvesicle.

LacZ expressionWe first examined testes of flies that do not contain anenhancer detector to assess levels of eucaryotic /3-galactosidase. No X-gal staining was observed, exceptin occasional degenerating cysts and in waste bags,which contain superfluous cytoplasmic material re-leased during the final stages of spermatid differen-tiation (see for instance Fig. 6A, solid and outlinedarrowheads). The low level and sporadic nature of thisendogeneous activity could not be confused with thereproducible expression patterns generated by indi-vidual enhancer detector lines. Moreover, eucaryotic /S-galactosidase could easily be distinguished from lacZexpression by using an antibody directed specifically

against the bacterial /J-galactosidase (see for instanceFig. 5F, arrowheads).

We then examined male gonads in homozygousviable and fertile P-element enhancer detector lines(see Materials and methods). The frequencies ofstaining patterns in the germ line and the soma(excluding the epithelial cells from the sheath connect-ing the testis to the seminal vesicle) were determined byclose examination of 120 lines. 55% expressed lacZ inthe germ line only, 5% in the soma only, while 13%labeled both germ cells and somatic cells. 27% did notgive rise to any staining pattern in male gonads.

From over 700 enhancer detector lines examined, wechose about fifty as molecular markers for the differentcell types and stages of spermatogenesis. A descriptionof selected lines follows.

Germ lineMost lines labeling the germline initiated lacZ ex-pression during the spermatocyte growth phase anddisplayed /3-galactosidase activity until after meiosis(Fig. 2A). Germ cells could be easily identified by theircharacteristic morphology and numbers. X-gal-positivecells in the growth phase region were typical ofmaturing spermatocytes, with a large nucleus andprominent nucleolus (Fig. 2B, arrow). X-gal positivecells were organized in groups of 16 large cells up tomeiosis (Fig. 2C, arrow) and 64 smaller cells after that(Fig. 2C, arrowhead). Postmeiotically, X-gal positivenuclei became elongated and eventually rod-shaped(Fig. 2D, arrow), as expected from maturing sperma-tids undergoing nuclear reshaping. The germline natureof the labeled cells was further confirmed by crossingtwo such marker lines to oskar301 (osk301) females(Lehman and Niisslein-Volhard, 1986). Progeny result-ing from these crosses were agametic and carried a copyof the enhancer detector. As expected, lacZ expressionwas absent from these agametic testes (data notshown).

Other lines exhibited a staining pattern in germ cellsthat was restricted to particular stages of differen-tiation. Three expressed lacZ in stem cells and themitotic phases of germ cell differentiation, but not insubsequent stages (Fig. 3A and data not shown). One ofthese labeled germ cells in the early stages of spermato-genesis (Fig. 3A), but not in the corresponding stages ofoogenesis (Fig. 3B, arrow), characterizing it as an earlymarker of male germ cell identity. Another line thatexpressed lacZ in both early and later stages of malegametogenesis also failed to stain early female germcells and was found to map to the same cytologicalposition (data not shown).

Cyst cellsWe identified marker lines that stain the somatic cystcells (Fig. IB, eye). Several expressed lacZ starting inthe growth phase region and in all cyst cells past thatstage (Fig. 4A). The /acZ-expressing cells were ident-ified as cyst cells because each cyst of developing germcells was associated with two X-gal-positive nuclei at itsperiphery (Fig. 4B, arrows), the number and location

I ste

\ sp

g I

hu

b

CyP

cy

c

Fig

. 1.

(A

) Ph

ase

cont

rast

vie

w o

f an

adu

lt t

estis

. T

he

apic

al e

nd (

api)

is l

eft

in a

ll fi

gure

s un

less

oth

erw

ise

note

d;

ter:

ter

min

al e

nd.

Bar

=50

ym

. (B

) Sc

hem

atic

re

pres

enta

tion

of

5 st

ages

of

sper

mat

ogen

esis

. A

rrow

s po

intin

g to

par

t (A

) in

dica

te w

here

app

roxi

mat

ely

in t

he

test

is e

ach

stag

e be

gins

; ce

lls a

re d

ispl

aced

in

an a

pica

l-to

- te

rmin

al d

irec

tion

as t

hey

mat

ure

with

in e

ach

stag

e. G

erm

lin

e st

em c

ells

and

som

atic

cys

t pr

ogen

itor

cel

ls a

re

anch

ored

aro

und

a hu

b of

som

atic

cel

ls (

hub)

at

the

apic

al

tip

of t

he t

estis

. O

nly

one

germ

lin

e st

em c

ell

(ste

) an

d tw

o cy

st p

roge

nito

r ce

lls (

cyp)

are

rep

rese

nted

for

cla

rity

. a*

asym

etri

c di

visi

ons

of a

ger

m l

ine

stem

cel

l an

d tw

o ne

ighb

ouri

ng c

yst

prog

enit

or c

ells

giv

e ri

se t

o on

e pr

imar

y go

nial

cel

l (s

pg)

and

two

cyst

cel

ls (

cyc)

, re

spec

tivel

y. m&

the

sper

mat

ogon

ial

cell

unde

rgoe

s fo

ur m

itotic

div

isio

ns,

whi

le t

he c

yst

cells

no

long

er d

ivid

e. g

ro:

the

resu

lting

16

sper

mat

ocyt

es (

spc)

gro

w d

ram

atic

ally

-mei

: th

e tw

o m

eiot

ic d

ivis

ions

occ

ur,

mar:

the

64 h

apm

d s

perm

atid

s (s

pt)

unde

rgo

dram

atic

mor

phol

ogic

al c

hang

es.

Onl

y 6

elon

gatin

g sp

erm

atid

s ar

e sh

own

for

clar

ity.

Bec

ause

of

the

leng

th o

f th

e sp

erm

tai

l, fu

lly e

long

ated

spe

rmat

ids

have

th

eir

nucl

eus

at t

he t

erm

inal

end

of

the

test

is,

whi

le t

hc t

ail

exte

nds

alm

ost

to t

he a

pica

l en

d. D

urin

g th

is l

ast

stag

e, t

he

two

cyst

cel

ls b

ecom

e st

ruct

ural

ly d

istin

ct,

the

head

-cys

t ce

ll (c

yh)

bein

g as

soci

ated

with

the

spe

rm h

eads

and

the

ta

il-cy

st c

ell

(cyt

) el

onga

ting

with

the

gro

win

g sp

erm

tai

ls.

The

hea

d-cy

st c

ell

then

bec

omes

ent

rapp

ed b

y a

spec

ializ

ed

epith

elia

l ce

ll (t

ec)

loca

ted

in t

he t

erm

inal

par

t of

the

te

stis

. C

oilin

g of

the

spe

rm t

jund

le e

nsue

s, f

ollo

wed

by

rele

ase

of m

otile

spe

rmat

ozoa

(sp

z) i

nto

the

sem

inal

ve

sicl

e. O

nly

one

sper

mat

ozoo

n is

sho

wn

for

clar

ity.

See

te

xt f

or a

ddit

iona

l in

form

atio

n.

e

94 P. Gdnczy, S. Viswanathan and S. DiNardo

expected for cyst cells. The identity of these cells wasfurther ascertained by showing that the /3-galactosidase-positive nuclei (Fig. 4C, arrows) were located withincytoplasm containing /33-tubulin (Fig. 4D), a /J-tubulinisotype specifically expressed in cyst cells in malegonads (Kimble et al., 1989).

Several lines labeled cyst cells in a stage-specificmanner. For instance, some labeled mainly early cystcells (Fig. 5A), including cyst progenitor cells (Fig. 5A,arrows). As expected from the ultrastructural data, cystcell nuclei were smaller and not as spherical asneighbouring germ line stem cells (Fig. 5A, filledarrowhead), and were located further away from theapical cells of the hub (Fig. 5A, outlined arrowhead).Other lines labeled cyst cells associated with lateproliferative and early growth phase germ cells, butnot later stages of differentiation (Fig. 5B, arrows).These transient expression patterns indicate that cystcells adopt distinct identities in the period prior tomeiosis.

Some lines labeled cyst cells only in postmeioticstages of the differentiation program (Fig. 5C, arrows),an observation confirmed by the lack of staining inlarval gonads (data not shown), where only premeioticstages are represented. A few lines specifically labeledtail-cyst cells but not head-cyst cells (Fig. 5D, arrows).The tail-cyst cell nuclei could be easily recognized bytheir invariable location, squeezed against the periph-ery of elongated sperm tail bundles (Fig. 5E). One lineexpressed lacZ in both tail- and head-cyst cells. Thehead-cyst cell nuclei could be identified by theirposition at the base of the testis and their closeassociation with sperm heads (Fig. 5F, arrowheads andcorresponding arrows).

Apical cells of the hubAt the apical tip of the testis, a group of 12 to 16 smallsomatic cells form the hub around which stem cells areanchored (Hardy et al., 1979; Fig. IB, hub). Weidentified marker lines that express lacZ either exclus-ively in these apical cells (Fig. 6A, arrow) or in theseapical cells as well as in early cyst cells (Fig. 5A). Theidentity of the labeled cells was suggested by theirclustering (Fig. 6B, arrow) and by the radial position ofgerm line stem cells (Fig. 6B, arrowhead) around thecluster. We first confirmed that the /acZ-expressing cellsare derived from somatic and not germ line lineage bycrossing these marker lines to oslc^1 females (Lehmanand Niisslein-Volhard, 1986). Progeny resulting fromsuch a cross lack germ cells but retain some somaticcomponents of the gonads, including the apical cells ofthe hub (Aboi'm, 1945). /J-galactosidase remainedpresent in such an agametic gonad (Fig. 6C, arrow). Wealso noted that the hub in the agametic gonad was bothlarger and positioned some distance away from theapical tip as compared to the normal testis (Fig. 6C,arrowhead; see Discussion). Additional evidence thatthe labeled cells indeed constitute the apical hub wasobtained by showing that the cluster of /J-galactosidasepositive nuclei (Fig. 6E, arrow) also expressed fasciclin

Fig. 2. A germ cell marker line initiating expression at thespermatocyte stage (line 817). All figures show X-galstaining unless otherwise noted. (A) Adult testis (B)Apical region, showing perinuclear staining in growingspermatocytes, recognizable by their large nucleus andprominent nucleolus (arrrow); note the strong staining insperm tails (arrowhead). (C) Cyst of 16 germ cells (arrow)in telophase of the first meiotic division (note that thestaining is present in the whole cell) and cyst of 64 youngpostmeiotic spermatids (arrowhead), some of which are outof focus. (D) Two cysts of spermatids at different stages ofpostmeiotic differentiation, as judged by sperm headmorphology; note the more mature rod-shaped spermheads (arrow). Bars=50 /an.Fig. 3. A marker line discriminates between male andfemale germ cells (line 606). (A) Apical region of thetestis, showing labeling restricted to germ line stem cellsand mitotically dividing spermatogonial cells. This line alsolabels tail-cyst cells and somatic cells of the terminalepithelium, not shown here. (B) Lack of staining inovaries; note the germarium (arrow) where theproliferative female germ cells are located. Bars=50 /an.Fig. 4. Marker line labeling cyst cells (line 600). (A) Adulttestis; note that this line also labels the hub (arrow) andthe terminal epithelial cells (arrowhead), as well as thepigment cells of the sheath (not visible in this focal plane).(B) Apical region, showing two labeled nuclei (arrows)nestled at the periphery of each growing cyst.(C,D) Localization of ^-galactosidase signal to cyst cellcytoplasm. (C) Larval gonad labeled with anti-/3-galactosidase antibody visualized by indirectimmunofluorescence. Six positive nuclei are in the plane offocus (two of them highlighted by arrows). Larval gonadscontain the same cell types and stages as those found inadult testes for the period prior to meiosis, including cystcells. Apical tip down. (D) Same gonad as in (C) labeledwith anti-/S3-tubulin antibody visualized by indirectimmunofluorescense. The signal reveals the cyst cellcytoplasm, which surrounds the developing germ cells (notlabeled). Note that /J-galactosidase positive nuclei arelocated within /33-tubulin positive cytoplasm (comparearrows in C and D). Bars=50 /an.

Ill (Fig. 6F, arrow), a marker specific for the apical cellsof the hub in male gonads (Brower et al., 1981).

By morphological criteria, the apical cells can beidentified as early as the first instar larval gonad(Aboim, 1945). We examined one line at this develop-mental stage to determine whether it was a faithfulmarker for these cells. Indeed, we observed strong lacZexpression in the apical cells of the first instar larvalgonad (data not shown). Furthermore, in 13 to 15 hourembryos, a cap of /J-galactosidase positive cells wasobserved at the apical side of the newly formed gonad(Fig. 6G, arrow). Thus, this line represents an earlymolecular marker for the presumptive apical cells in thegonadal mesoderm.

Terminal epitheliumAt the terminal end of the male gonad, the testis sheathincludes a layer of epithelial cells that fuse with theseminal vesicle. We identified marker lines that labelmost epithelial cells of the terminal region (Fig. 7A,

2A

4A

V

Fig. 5. Particular lines label cyst cells in a stage-specific manner. (A) Apical region, showing labeling of early cyst cells,including cyst progenitor cells (arrows). Note a neighbouring unlabeled germ line stem cell (solid arrowhead) and thelabeled apical cells of the hub (outlined arrowhead) (line 842). This line also labels pigment cells, terminal epithelial cellsand head-cyst cells (not shown here). (B) Apical region, showing transient labeling of cyst cells in the late proliferative andearly growth phase regions (arrows point at the youngest and oldest labeled cyst cells, respectively) (line 901). (C) Adulttestis shows labeling of cyst cells postmeiotically only (arrows). Head-cyst cells are indicated by a solid arrowhead (linems-985; this is a male-sterile line, stained here as a heterozygote). This line also labels germ line stem cells (not visible atthis magnification) and terminal epithelial cells (outlined arrowhead). (D, E) Tail-cyst cells (line 498). (D) Adult testis,showing labeling of tail-cyst cells only (arrows). (E) Tail-cyst cell nucleus between two elongated sperm tail bundles.(F) Head-cyst cells. Immunoperoxidase staining of the terminal region of an adult testis with anti-/S-galactosidase antibody.Hoechst counterstain shows the association of rod-shaped sperm head nuclei (arrows) with the /3-galactosidase positivehead-cyst cell nuclei (arrowheads), (line 57; this line displayed unusually high levels of endogenous /J-galactosidase activitywhich necessitated analysis using the antibody directed against bacterial /?-galactosidase). Bars=50 /an.

G i

Fig. 6. Marker line labeling the apical cells of the hub (line254; this line is homozygous lethal and therefore examinedas a heterozygote). (A) Adult testis, showing labeling ofthe hub located at the apical tip of the testis (arrow).There is also non-specific staining in degenerating cysts(solid arrowhead) and waste bags (outlined arrowhead).(B) Apical-most portion, showing labeling of the clusteredsomatic cells of the hub (arrow) and radially disposedunlabeled germ line stem cells (arrowhead) anchoredaround the hub. (C) Immunoperoxidase staining of anagametic testis, demonstrating the somatic nature of the /3-galactosidase positive cells. The hub in the agametic testis(arrow) is larger and positioned further away from theapical tip as compared to the normal testis (arrowhead);

note that this line also labels cells from the seminal vesicle sheath (outlined arrowhead). (D,E,F) Identification of the /S-galactosidase positive cells as the apical cells of the hub. (D) Hoechst counterstain revealing all nuclei present at the apicaltip of the testis, viewed from above. (E) Same apical tip as in D labeled with /3-galactosidase antibody visualized byindirect immunofluorescence; note the clustering of the labeled cells (arrow). (F) Same apical tip as in D and E labeledwith fasciclin III antibody visualized by indirect immunofluorescence; note that the labeled cells (arrow) are the same asthose positive for /3-galactosidase (compare arrows in D and E). (G) Immunoperoxidase staining with anti-^-galactosidaseantibody of a 13-15 hour embryonic gonad. The gonad was dissected away from the embryo to highlight the labelingobserved in the apical portion of the gonad (arrow); note the prominent pole cells (arrowhead). Other cell types also stainduring embryogenesis (M. Whiteley et al., 1992). Bars=50 jxm.

B

Fig. 7. Marker lines label overlapping subsets of cells in the terminal epithelium. (A,B) Most or all terminal epithelial cells(line 34). (A) Adult testis; staining begins in a region where coiling of sperm bundles occurs (arrow) and extends down tothe junction with the seminal vesicle (arrowhead). This line also weakly labels cyst cells in the late proliferative and earlygrowth phase regions (not visible in this focal plane) (B) Terminal region, showing that the labeled cells (arrows) are partof the sheath of the testis rather than the lumen (arrowhead). (C) Terminal region, showing that only a subset of epithelialcells are labeled (solid arrows), probably those that entrap the head-cyst cells (line 429). Note the lack of staining in themore terminal epithelial cells (outlined arrowheads), near the junction with the seminal vesicle (solid arrowhead); note alsothe presence of coiling sperm bundles (outlined arrow) in the area where the labeled cells are located. Bars=50 /an.

Spermatogenesis in Drosophila 95

from the arrow to the arrowhead). The labeled cellswere identified by the fact that the stain reaches theseminal vesicle (Fig. 7A, arrowhead) and on thelocation of the labeled nuclei within the sheath of thetestis (Fig. 7B, arrows), rather than the lumen (Fig. 7B,arrowhead).

One line specifically labeled what appears to be thespecialized epithelial cells that entrap the head-cystcells prior to the coiling of spermatids (Tokuyasu et al.,1972, Fig. IB, tec). We infer this because the labelledcells are a subset of the terminal epithelial cells (Fig.7C, solid arrows, compare with Fig. 7B) and are locatedin an area with many coiled sperm bundles (Fig. 7C,outlined arrow).

Other cells of somatic origin found in testes includepigment and muscle cells from the sheath. We identifiedmarker lines labeling each of these cell types inconjunction with other cell types (data not shown).

Discussion

Little is known about the genes governing the differen-tiation of germ line and somatic cells during spermato-genesis in Drosophila melanogaster. Given the limi-tations of a classical mutant analysis, we sought to studythis developmental pathway using a complementaryapproach. We have examined over 700 homozygousviable and fertile enhancer detector lines for lacZexpression in male gonads. We have described stainingpatterns for the different cell types present in testes,denning in some cases stages that had not beenrecognized previously by morphological criteria. Asummary of the marker lines described in this report isgiven in Table 1.

Expression in the germ line/3-galactosidase activity is found in the male germ line in68% of the lines examined, a figure that is more thandouble that reported in two similar studies conductedwith oogenesis (Fasano and Kerridge, 1988; Grossnik-laus et al., 1989). This could simply be indicative of alarger number of cts-acting elements active during maleversus female gametogenesis. More probably, thiscould be due to the fact that the lines were generated bya transposition event occurring in the male germ line.Indeed, it has been shown that transposition eventsoccurring in the male germ line lead to a higherfrequency of lacZ expression in male than in femalegerm cells, the reverse being the case when thetransposition event occurs in the female germ line(Bownes, 1990).

Most lines that stain germ cells start expressing lacZin the growth phase. This is consistent with uridineincorporation experiments which showed that the bulkof transcription in the germ line occurs during this phase(Olivieri and Olivieri, 1965; Gould-Somero and Hol-land, 1974). Those studies also led to the postulate thattranscription ceases prior to meiosis during Drosophilaspermatogenesis. Accordingly, we have not identifiedany line in which lacZ expression occurs strictlypostmeiotically in the germ line. Such strict postmeiotictranscription units may yet exist, but not be representedin a collection of viable and fertile lines. Alternatively,such loci may be refractory to P-element insertion andbe under-represented among enhancer detector lines.

Male germ cell identityTwo marker lines express lacZ in the proliferative phaseof male but not female gametogenesis, characterizingthem as markers of early male germ cell identity (Fig.3A and 3B and data not shown). Recent work has

Table 1. Summary of the marker lines described in this report. Enhancer detector lines were obtained from thelaboratories ofY.-N. and L. Jan (a), M. Fuller and M. Scott (b), J. Kassis (c), and J. Merriam (d); see

Materials and methods

Line andorigin

Cytogeneticlocation* Staining pattern in testes Figure

34 (a) 62A/B

57 (a)254 (c)429 (a)498 (a)600 (a)

606 (a)

817 (b)842 (b)

901 (b)

ms-985 (d)

82C35C88E24A

21F/22A

28C

(ND)57F

92B

28C

All terminal epithelial cells; cyst cells in late 7A and Bproliferative and early growth regions.

Head- and tail-cyst cells. 5FApical cells of the hub. 6A through GSubset of terminal epithelial cells. 7CTail-cyst cells. 5D and ECyst cells, starting in the growth region; apical 4A through D

cells of the hub; terminal epithelial cells;pigment cells.

Male but not female germ line stem and gonial 3A and Bcells; tail-cyst cells; terminal epithelial cells.

Germ cells, starting in the growth phase. 2A through DEarly cyst cells, including cyst-progenitor cells; 5A

apical cells of the hub; head-cyst cells; terminalepithelial cells; pigment cells.

Cyst cells in late proliferative and early growth 5Bregions.

Cyst cells in postmeiotic aspects; terminal 5Cepithelial cells; germ line stem cells.

*(ND) - not determined.

96 P. Go'nczy, S. Viswanathan and S. DiNardo

begun to unravel the sex-determination pathway in themale and female germ line (Steinman-Zwicky et al.,1989; reviewed by Pauli and Mahowald, 1990). Infemales with mutations that interfere with this pathway,ovaries are filled with small cells that resemble germcells embarked on a male differentiation program. Thesexual identity of such cells can sometimes be ambigu-ous when defined solely by morphological criteria.Markers of early male germ cell identity would,therefore, be useful in ascertaining the sexual nature ofgerm cells in mutants putatively affected in germ linesex-determination. The usefulness of our marker lineswas tested by crossing them to sans-fille (snf), a well-characterized germ line sex-determination mutant(Oliver et al., 1988). In a snf mutant female, XX germcells now express /3-galactosidase (D. Pauli and A.P.Mahowald, personal communication). This confirmsthe transformed sexual identity of XX germ cells in a snfmutant and demonstrates that these two marker linescan indeed serve to probe early male germ cell identity.

Cyst cell differentiationWe have shown that most lines that stain cyst cells labelonly a subset of the cyst cells present during the courseof spermatogenesis. For the period prior to meiosis, wefind at least three distinct types of staining patterns incyst cells. The first type labels early cyst cells, includingcyst progenitor cells and shows little or no expression inlater stages of cyst cell differentiation (Fig. 5A). Thesecond type labels cyst cells exclusively in the lateproliferative and early growth phase regions (Fig. 5B).The third and most common type labels cyst cellsstarting in the growth phase region and most or all thecyst cells in the period after meiosis (Fig. 4A, B). Thesedistinct types of staining patterns suggest that there aredifferent stages in cyst cell differentiation in the periodprior to meiosis. Such distinctions had not beendetected by morphological criteria.

After meiosis occurs in germ cells, the two cyst cellsbecome structurally distinct, one being associated withthe sperm heads and the other with the elongatingsperm tails. Accordingly, we found lines that labeleither only the tail-cyst cells (Fig. 5D, E) or both tail-and head-cyst cells (Fig. 5F). Thus, cyst cell differen-tiation for the period both before and after meiosis canbe broken down into a succession of stages revealed byour marker lines.

Interaction between germ line and somaIn oogenesis, cooperation between germ line and somais crucial for proper development of the egg(Schiipbach, 1987). It is not yet known whether asimilar requirement has to be met for correct spermdevelopment. However, recent experiments in germline sex determination are consistent with some interac-tion taking place in male gametogenesis as well. Whentransplanted into a female host, chromosomally malegerm cells embark on the spermatogenic differentiationprogram but arrest at the spermatocyte stage, possiblybecause they need a compatible soma to differentiatefurther (Steiman-Zwicky et al., 1989).

The marker lines described here can probe the extentto which germ line and soma cooperate during thecourse of spermatogenesis. For example, the lines thatreveal transient cyst cell identities will allow us tomonitor the progression of cyst cell fate in mutantbackgrounds where the differentiation of the germ cellsis blocked.

Agametic gonads provide an experimental situationin which to address the role of the germ line in directingspecific somatic cell fates. We have shown that theapical cells of the hub around which stem cells areanchored are still present in agametic gonads (Fig. 6C).Based on morphological criteria alone, it had beenconcluded that the hub is not altered in agametic gonads(Aboi'm, 1945). However, the hub in an agameticenvironment appears different in two respects. First, itis larger than in normal testes, due to an increase in cellsize, rather than cell number (P.G. and S.D., unpub-lished data). Second, the hub is displaced from the veryapical tip of the agametic testis. We have not yetdetermined the cause of this displacement, nor thenature of the cells occupying the apical-most position.Taken together, our observations confirm that theapical cells can organize into a hub in the absence ofgerm line and suggest that the germ cells plays somerole, direct or indirect, in controlling the size of thesomatic hub and its position in the testis.

GonadogenesisThe gonad is formed at about 12 hours duringembryogenesis, when the pole cells become surroundedby gonadal mesoderm (Sonnenblick, 1941; Aboim,1945; Hay et al., 1988). All morphologically distinctsomatic cells of the adult testis derive from this commonmesodermal primordium. Marker lines that labelspecific somatic cells early in development are instru-mental in addressing when these different cell typesassume their fate. We have shown that a marker linelabeling the apical cells of the hub in the adult testisrepresents an early molecular marker for the presump-tive apical cells in the embryonic gonadal mesoderm(Fig. 6G). Several marker lines that identify the othersomatic cell types in testes also label presumptive cellsin the embryonic gonad (M. Boyle, P.G., SV. andS.D.,unpublished data). Studies using such marker lines canunravel the origin and lineage of the several somaticcomponents of male gonads.

Probing spermatogenesis with enhancer detectorsWe have discussed how marker lines are instrumental ininvestigating gonadogenesis and in revealing the extentof interaction between germ line and soma duringspermatogenesis. In addition, this collection of markerlines should allow the characterization of genes ex-pressed at critical times and places during malegametogenesis. Together with an analysis of enhancerdetector male-sterile mutants, these approaches shouldhelp in defining key regulators of this developmentalpathway.

We thank the laboratories of Yuh-Nung and Lily Jan,

Spermatogenesis in Drosophila 97

Margaret Fuller and Matthew Scott, Judy Kassis, JohnMerriam for generously providing the enhancer detector linesused in this study. We thank Margaret Fuller also for fruitfuldiscussions and constant encouragement. We are grateful toDanny Brower, Alfonso Martinez-Arias and Elizabeth Rafffor their gift of antibodies. We benefited from the efficienthelp of Carrie Bromleigh in determining the cytogeneticlocation of the P-element in the marker lines. We are indebtedto Peter Bokor, Monica Boyle, Claude Desplan, ScottDougan, Laurent Fasano, Marianna Giarrd, Janet Mullenand Elettra Ronchi for careful reading of the manuscript. Allmembers of the DiNardo and Desplan laboratories areacknowledged for providing an exciting scientific environ-ment. We wish to thank Sam Ward for encouraging us toinitiate a molecular genetic analysis of spermatogenesis in D.melanogaster. S.D. would also like to thank Allan Spradlingand members of his laboratory for an early and stimulatingdiscussion about the enhancer detector patterns they wereobtaining in ovaries. S.D. is a Lucille P. Markey Scholar andwork in his lab is supported by the Lucille P. MarkeyCharitable Trust.

NOTE

All marker lines mentioned in this report are availablefrom the authors upon request.

References

Abofan, A. N. (1945). DeVelopement embryonnaire et post-embryonnaire des gonades normales et agam^tiques de Drosophilamelanogaster. Rev. Suisse Zool. 52, 53-154.

BeUen, H. J., O'Kane, C. J., Wilson, C , Grossniklaus, U., Pearson,R. K. and Gehring, W. J. (1989). P-element mediated enhancerdetection: a versatile method to study development in Drosophila.Genes Dev. 3, 1288-1300.

Bier, E., Vaessin, H., Shepherd, S., Lee, K., McCall, K., Barbel, S.,Ackerman, L., Carretto, R., Grell, E., Jan, L. Y. and Jan, Y. N.(1989). Searching for pattern and mutation in the Drosophilagenome with a P-lacZ vector. Genes Dev. 3, 1273-1287.

Bownes, M. (1990). Preferential insertion of P elements into genesexpressed in the germ-line of Drosophila melanogaster. Mol. Gen.Genet. 222, 457-460.

Bridges, C. B. (1916). Non-disjunction as proof of the chromosometheory of heredity. Genetics 1, 1-52 and 107-163.

Brower, D. L., Smith, R. J. and Wllcox, M. (1981). Differentiationwithin the gonads of Drosophila revealed by immunofluorescence.J. Embryol. exp. Morph. 63, 233-242.

Dybas, L. K., Harden, K. D., Machnick, J. L. and Geer, B. W.(1983). Male fertility in Drosophila melanogaster: lesions ofspermatogenesis associated with male-sterile mutations of thevermilion region. J. exp. Zool. 226, 293-302.

Fasano, L., Corl, N. and Kerridge, S. (1988). Expression of areporter gene ressembles that of its neighbour: an insertion in thehairy gene of Drosophila. Roux's Arch. Dev. Biol. 197, 507-512.

Fasano, L. and Kerridge, S. (1988). Monitoring positionalinformation during oogenesis in adult Drosophila. Development104, 245-253.

GatefT, E. (1982). Gonial cell neoplasm of genetic origin affectingboth sexes of Drosophila melanogaster. In EmbryonicDevelopment, Cellular Aspects, pp. 621-632 (eds. M.M. Burger andR. Weber). New York: Alan Liss.

Geer, B. W., Lischwe, T. D. and Murphy, K. G. (1983). Male fertilityin Drosophila melanogaster: genetics in the vermilion region. J.exp. Zool. 225, 107-118.

Gould-Somero, M. and Holland, L. (1974). The timing of RNAsynthesis for spermiogenesis in organ cultures of Drosophila

melanogaster testes. Wilhelm Roux's Arch. Entwicklungsmech.Org. 174, 133-148.

Green, L. L., Wolf, N., McDonald, K. L. and Fuller, M. T. (1990).Two types of genetic interaction implicate the whirligig gene ofDrosophila melanogaster in microtubule organization in theflagellar axoneme. Genetics 126, 961-973.

Grossniklaus, U., Bellen, H. J., Wilson, C. and Gehring, W. J. (1989).P-element-mediated enhancer detection applied to the study ofoogenesis in Drosophila. Development 107, 189-200.

Hardy, R. W., Tokuyasu, K. T., Lindsley, D. L. and Garavito, M.(1979). The germinal proliferation center in the testis of Drosophilamelanogaster. J. Ultrastruct. Res. 69, 180-190.

Hay, B., Ackennan, L., Barbel, S., Jan, L. Y. and Jan, Y. N. (1988).Identification of a component of Drosophila polar granules.Development 103, 625-640.

Kassis, J. A. (1990). Spatial and temporal control elements of theDrosophila engrailed gene. Genes Dev. 4, 433-443.

KeUerman, K. A., Mattson, D. M. and Duncan, I. (1990). Mutationsaffecting the stability of the fushi tarazu protein of Drosophila.Genes Dev. 4, 1936-1950.

Kemphues, K. J., Kaufman, T. C , Raff, R. A. and Raff, E. C. (1982).The testis-specific /3-tubulin subunit in Drosophila melanogaster hasmultiple functions in spermatogenesis. Cell 31, 655-670.

Klmble, M., Incardona, J. P. and Raff, E. C. (1989). A variant £•tubulin isoform of Drosophila melanogaster (yS3) is expressedprimarly in tissues of mesodermal origin in embryos and pupae, andis utilized in populations of transient microtubules. Devi. Biol. 131,415-429.

Lehman, R. and Nttsslein-Volhard, C. (1986). Abdominalsegmentation, pole cell formation, and embryonic polarity requirethe localized activity of oskar, a maternal gene in Drosophila. Cell47, 141-152.

Lifschytz, E. (1978). Uncoupling of gonial and spermatocyte stages bymeans of conditional lethal mutations in Drosophila melanogaster.Devi. Biol. 66, 571-578.

Lifschytz, E. (1987). The developmental program of spermiogenesisin Drosophila: a genetic analysis. In International Review ofCytology, vol. 109, pp. 211-257. London: Academic Press.

Lifschytz, E. and YakoboWtz, N. (1978). The role of X-linked lethaland viable male-sterile mutations in male gametogenesis ofDrosophila melanogaster: genetic analysis. Molec. Gen. Genet.161, 275-284.

Lindsley, D. L. and Tokuyasu, K. T. (1980). Spermatogenesis. In TheGenetics and Biology of Drosophila, vol. 2 (eds M. Ashburner andT.R.F. Wright), pp. 225-294. London:Academic Press.

McKearin, D. M. and Spradling, A. C. (1990). bag-of-marbles: aDrosophila gene required to initiate both male and femalegametogenesis. Genes Dev. 4, 2242-2251.

Mitchison, T. and Sedat, J. (1983). Localization of antigenicdeterminants in whole Drosophila embryos. Devi Biol. 99, 261-264.

O'Kane, C. J. and Gehring, W. J. (1987). Detection in situ of genomicregulatory elements in Drosophila. Proc. Natn. Acad. Sci. USA 84,9123-9127.

Oliver, B., Perrimon, N. and Mahowald, A. P. (1988). Geneticevidence that the sans fille locus is involved in Drosophila sex-determination. Genetics 120, 159-171.

Olivieri, G. and Olivieri, A. (1965). Autoradiographic study of nucleicacid synthesis during spermatogenesis in Drosophila melanogaster.Mut. Res. 2, 366-380.

Pauli, D. and Mahowald, A. P. (1990). Germ-line sex-determinationin Drosophila melanogaster. Trends Genet. 6, 259-264.

Regan, C. L. and Fuller, M. T. (1988). Interacting genes that affectmicrotubule function: the nc2 allele of the haywire locus fails tocomplement mutations in the testis-specific /J-tubulin gene ofDrosophila. Genes Dev. 2, 82-92.

SchOpbach, T. (1987). Germ line and soma cooperate duringoogenesis to establish the dorso-ventral pattern of egg shell andembryo in Drosophila melanogaster. Cell 49, 699-707.

Sonnenblick, B. P. (1941). Germ cell movements and sexdifferentiation of the gonads in the Drosophila embryo. Proc.Natn. Acad. Sci. USA 26, 373-381.

Steinman-Zwicky, M., Schmid, H. and NSthiger, R. (1989). Cell-autonomous and inductive signals can determine the sex of the

98 P. Gdnczy, S. Viswanathan and S. DiNardo

germ line of Drosophila by regulating the gene Sxl. Cell 57, 157-166.

Tokuyasu, K. T., Peacock, W. J. and Hardy, R. W. (1972). Dynamicsof spermiogenesis in Drosophila melanogaster. II. Coiling process.Z. ZeUforsch. 127, 492-525.

Whiteley, M., Noguchi, P. D., Sensabaugh, S. M., Odeuwald, W. F.and Kassis, J. A. (1992). The Drosophila gene escargot encodes azinc finger motif found in snail-related genes. Mech. Dev. (inpress).

Wilson, C , Pearson, R. K., BeUen, H. J., O'Kane, C. J.,Grossniklaus, U. and Gehring, W. J. (1989). P-element-mediatedenhancer detection: an efficient method for isolating andcharacterizing developmentally regulated genes in Drosophila.Genes Dev. 3, 1301-1313.

(Accepted 16 September 1991)