Copyright 0 1988 by the Genetics Society of America
fog-2, a Germ-Line-Specific Sex Determination Gene Required for
Hermaphrodite Spermatogenesis in Caenorhabditis elegans
Tim Schedl and Judith Kimble
Laboratory of Cell and Molecular Biology, Graduate School, and Department of Biochemistry, College of Agnculture and Life Sciences,
University of Wisconsin, Madison, Wisconsin 53 706
Manuscript received November 16, 1987
Accepted January 23, 1988
This paper describes the isolation and characterization of 16 mutations in the germ-line sex
determination gene fog-2 (fog for feminization of the germ line). In the nematode Caenorhabditis
elegans there are normally two sexes, self-fertilizing hermaphrodites ( X X ) and males (XO). Wild-
type XX animals are hermaphrodite in the germ line (spermatogenesis followed by oogenesis), and
female in the soma. fog-2 loss-of-function mutations transform XX animals into females while X 0
animals are unaffected. Thus, wild-type fog-2 is necessary for spermatogenesis in hermaphrodites
but not males. The fern genes and fog-1 are each essential for specification of spermatogenesis in
both XX and X 0 animals. fog-2 acts as a positive regulator of the fern genes and fog-1. The tra-2 and
tra-3 genes act as negative regulators of the fern genes and fog-1 to allow oogenesis. Two models are
discussed for how fog-2 might positively regulate the fern genes and fog-1 to permit spermatogenesis;
fog-2 may act as a negative regulator of tra-2 and tra-3, or fog-2 may act positively on the fern genes
and fog-I rendering them insensitive to the negative action of tra-2 and tra-3.
determination is the ratio of the number of X chro-
mosomes to sets of autosomes (MADL and HERMAN
1979). Diploid XX animals are self-fertilizing her-
maphradites; dipoloid X 0 animals are males. Most
tissues of these two sexes differ morphologically and1
or biochemically. The term “hermaphrodite,” when
applied to C. elegans, describes a self-fertile animal
with a female soma and a germ line that is first male,
producing sperm, and then female, producing ooc-
ytes. Hermaphrodites of C. elegans reproduce either
by self-fertilization or by cross-fertilization after mat-
ing with males.
The C. elegans hermaphrodite soma is essentially
female. It is morphologically indistinguishable from
the female soma of a closely related malelfemale
nematode species Caenorhabditis remanei (SLDHAUS
1974), and is extremely similar to the female soma
of Panagrellus redivivus
198 1, 1982). The hermaphrodite soma is also indis-
tinguishable from the soma of C. elegans females that
arise as a consequence of mutations in sex determin-
ing loci (e.g., fem mutations). In the hermaphrodite
germ line, spermatogenesis occurs first, beginning
during the last larval stage of development (L4) and
ending soon after the molt into adulthood. From this
brief period of spermatogenesis, about 320 sperm
per hermaphrodite are produced from about
primary spermatocytes in each of two gonads. Then,
each gonad switches to oogenesis and oocytes are
N the nematode Caenorhabditis elegans, there are
normally two sexes. The initial signal for sex
Genetics 119: 43-61 (May, 1988).
produced continuously throughout adulthood. The
sexual duality of the XX germ line in a female soma
suggests that hermaphroditism in C. elegans is a prop-
erty of the germ-line tissue. This is to be distinguished
from hermaphroditism in the annelid Lumbricus ter-
restrzj (earthworm) in which each animal has a sepa-
rate ovary and testis (sexual duality in both the germ
line and soma) and reproduction occurs by mating
(HICKMAN, HICKMAN and HICKMAN
In C. elegans, the XIA ratio is transduced by a set
of genes that direct both
dosage compensation (VILLENEUVE and
L. MILLER, J. PLENEFISCH and B.
MEYER, personal communication).
ducer” genes, in turn, regulate both genes that direct
dosage compensation plus genes that
sexual phenotype. Here, we focus on the later group
of sex-determination genes; the transducer genes
and dosage compensation genes are beyond the scope
of this paper.
Seven sex-determining genes have been identified
that specify sexual fate in all tissues of the animal-
both somatic and germ line (HODGKIN and BRENNER
1977; HODCKIN 1980; DONIACH and HODCKIN 1984;
HODGKIN 1986). In addition, one sex-determining
gene has been identified that affects the sexual fate
of a single tissue, the germ line (DONIACH
K. BARTON, personal communication).
and somatic mutant phenotypes of these sex-deter-
mining genes are summarized
genes must act downstream from the
sex determination and
in Table 1. These
T. Schedl and J. Kimble
Summary of sex determination genes used in this study
X 0 phenotype
her-I ( If)b
f e m - 2 ( v
tra-3 ( l f ) ' * '
Male then female
Male then female
Male then female
Male then female
Male then female
Male then female
' y, loss-of-function; for these genes this is the probable null phenotype. gf, gain-of-function. Phenotypes are of homozygotes. For
details of mutant phenotypes, consult references and text.
Homozygous mutant derived from a homozygous mutant mother m( - / - ), z( - /- ).
DONIACH and HODCKIK (1984).
e HODCKIN (1986).
f BARTON, SCHEDL and KIMBLE (1987).
g HODCKIN (1987); T. SCHEDL, unpublished observations. tra-l(lf) is included in this table for comparison with tra-Z(lf) and tra-3(@. The
interaction of tra-1 and fog3 mutations will be discussed elsewhere.
h Gonad abnormal, HODCKIN (1987);
' HODCIN and BRENNER (1977);
J DONIACH (1986a), this paper.
DONIACH (1986b); M. K. BARTON, person communication. It is unclear at this time whether fog-I alleles are If or gf.
T. SCHEDL, unpublished observations.
T. SCHEDL, unpublished observations.
because mutations in them override this initial signal.
Based on the results of a series of experiments in
which the epistasis of mutations in these genes was
analyzed, HODGKIN (1980, 1986) proposes that the
sex-determining genes act in a cascade of negative
regulation to control the state of tru-I, which in turn
specifies somatic sexual phenotype.
The wild-type function of each of the sex-deter-
mining genes has been deduced from the phenotype
of animals homozygous for a
mutation in that gene. Thus, fem-l(Ef),fem-2(lf), and
fem-3(lf) X X and X 0 mutant animals are female
instead of hermaphrodite and
(Table 1); therefore the wild-type fem-1, fem-2, and
fem-3 genes are required for
both the XX germ line and all X 0 tissues (NELSON,
LEW and WARD 1978; DONIACH and HODGKIN
KIMBLE, EDGAR, and HIRSH 1984; HODGKIN 1986).
Similarly, her-1 ( l f ) X 0 animals are hermaphrodite
instead of male; therefore, the wild-type her-1 gene
is required for the development of male somatic
tissues and for continuous spermatogenesis in X 0
animals (HODGKIN 1980; C. TRENT,
W. WOOD, personal communication).
l(lf) mutants are self-fertile (Table l), wild-type her-
1 is not needed for hermaphrodite spermatogenesis.
Finally, tra-l(lf), tra-2(lf), and tra-3(lf) XX animals
are masculinized. The details of masculinization by
loss-of-function ( l f )
male development in
P. SCHEDIN and
mutations in each of the tra genes varies (Table 1).
Both loss-of-function and gain-of-function mutant
phenotypes of tra-2 make this gene stand out
necessary for the switch from spermatogenesis
oogenesis in the hermaphrodite germ line. In con-
trast, the role of tru-1 in the hermaphrodite germ
line is unclear (HODGKIN 1987a; T. SCHEDL, unpub-
lished observations). Therefore, the wild-type tra-1,
tra-2, and tra-3 genes are necessary for female de-
velopment, but their roles in specification of that
development differ (HODGKIN
The production of first sperm and then oocytes by
the C. elegans hermaphrodite raises two major ques-
tions about the regulation
the XX germ line: (1) how is male germ-line devel-
opment initiated within a female' soma? and, (2) how
is the switch from male to female germ-line
opment affected? Since all uncommitted germ cells
of the XX hermaphrodite are XX, whether a precursor
of sperm or oocyte, the XIA ratio is not the primary
signal for sexual choice in the hermaphrodite germ
line. Instead, control of tru-2 and fem-3 activities
appear to be important to germ-line sex determina-
tion in hermaphrodites. In particular, the phenotypes
of gain-of-function (gf) mutations of tra-2 and fem-3
provide some insight into the genetic mechanisms of
control over hermaphrodite germ-line development.
Both tru-2(gf) and fem-?(gf) mutations affect
and BRENNER 1977).
of sex determination in
Hermaphrodite Germ-Line Sex
sexual fate of the XX germ line, but have little or no
effect on the X 0 germ line or the soma of either X X
or X 0 animals. The XX germ line of tru-2(gf) mutants
is feminized: germ cells that would normally differ-
entiate as sperm become oocytes instead, and ooge-
nesis continues throughout adulthood (DONIACH
1986a). Conversely, the XX germ line of fem-3(gf)
mutants is masculinized: sperm are produced contin-
uously, generating a vast excess of sperm, with no
sign of oogenesis. Thus, germ cells that would nor-
mally differentiate as oocytes become sperm instead
(BARTON, SCHEDL and KIMBLE 1987). Feminization
of the XX germ line by tru-2(gf) mutations suggests
that, in wild-type XX animals, tru-2 activity might be
modulated to permit spermatogenesis.
(1986a) suggests that tru-2 is no longer sensitive to
this modulation in tru-2(gf) mutants. Similarly, mas-
culinization of the X X germ line by fern-3(gf) suggests
that, in wild-type XX animals, fem-3 activity is nega-
tively regulated to permit the
BARTON, SCHEDL and
KIMBLE (1987) suggest thatfem-
3(gf) mutants are no longer sensitive to this negative
This paper describes loss-of-function mutations in
a germ-line specific sex-determination gene,
(fog for feminization of the germ line). XX animals
homozygous for ,fog-2 mutations are female, while
X 0 animals are unaffected. Therefore, a homozygous
fog-2 strain can reproduce
but not as a self-fertilizing hermaphroditic strain.
This mutant phenotype indicates that fog-2 is a reg-
ulator of hermaphrodite spermatogenesis. Analysis
of the interaction of fog-2 mutations with mutations
in other sex determining genes provides a framework
for placing fog-2 within the regulatory hierarchy of
sex determination as it applies to the XX germ line.
switch to oogenesis.
as a malelfemale strain,
MATERIALS AND METHODS
General methods for culturing nematodes
described by BRENNER (1974). Experiments
at 20" unless specified
worms were under continuous growth conditions and were
not starved or recovering from the dauer state.
Nomenclature: For certain genes in the C. elegans sex
determination pathway, some loss-of-function
exhibit dominance and some gain-of-function
are recessive. We therefore designate alleles as gain-of-
function with the suffix gf and loss-of-function with the
suffix If instead of abbreviations for dominant and reces-
sive. Numerically designated alleles without a suffix are
assumed to be loss-of-function unless indicated to
contrary; this avoids confusion between "1" and "1" (see
SCHEDL and KIMBLE (1987) for a further descrip-
tion). For experiments where maternal and zygotic geno-
types are important they are indicated by m( ) and z( ),
respectively. For example, m( - I+), I( - / + ) indicates a
heterozygous mutant derived from a heterozygous mutant
mother. All other nomenclature
in the text. For all experiments
Strains: C. elegans var. Bristol isolate N2 is defined as
wild type, and is the strain from which all other stocks are
derived. Most of the mutations used in this study are listed
in HOIKKIN and RII)I)I.F. (1988) and SW.\NSON, EI)(;I.F.\ and
RIIIDIX (1984). The phenotypes of sex determination mu-
tants are shown in Table 1 and are described and referenced
explicitly in the text. The following mutations and chro-
mosomal rearl-angenlents were used [dol (abnormal dauer
larva formation), dp? (dumpy), en16 (embryonic lethal),~/rrtr
(feminization of germ line and soma), /og (fenlinimtion of
germ line), h u (hermaphrotlitization), lott (long), . \ u f i (sup-
pressor), /Ire (transformer), unc (uncoordinatetl)]:
Linkage group (LC;) I : Jog-l(q187).
LG 11: (lfi?-10(~~128),
LC 111: jl'?T/-2(Q210j), dfi\.-19(~1259), 1111C-32((1189), d/lJ-
LC; IV: JPIN-l(81991), 111rc-24(0138),
q95<4, q9hgf). tkq-1 jtrtd11, d~1~-201c12K2),
LC; V : ltwl(~1520, ~156l), l/iw5(o1490), dps-Zl((428).
Plnl~-4(llc60), llnc-5l(P369, PI 189).
LG X : sup-i(.vt5), lon-2(rh78).
Isolation of fog-2 alleles: Four methods were used to
isolatefog-2 alleles: 1. Screen for germ-line feminizing mutants.
L4 hermaphrodites (PO), either N2 or dpy-19 +/+ unc-32
(markers used were for reasons irrelevant to this work),
were mutagenized with 0.05 M ethyl methanesulfonate
(EMS) for 4 hr (BRENNER 1974) and individual Fl self-
progeny were picked to agar-filled Petri plates. Using a
dissecting microscope, the F n self-progeny were screened
for the presence of females (spermless hermaphrodites) at
25". Fourfog-2 alleles were isolated from 12,438 mutagen-
ized haploid genomes; q70 and q71 were from N2 PO while
q154 and q226 were from dpy-19 +/+ unc-32 Po.
2. Screen for mutations that fail to complement fog-Z(q71). unc-
51 hermaphrodites were mutagenized
scribed above and crossed withfog-2(q71) males, either at
15" or 25". Non-Unc FI XX cross-progeny were picked away
from X 0 males at the L4 stage (to ensure virginity) in
groups of 25 to 50. Plates were screened for F1 females
among self-fertile sibs when all animals were adults. Fe-
males arising from the failure of the putative fog-2 allele
[unc-51 fog-2(new)] to complement 971 in trans were crossed
with N2 males and 8 to 12 F4 L4 X X progeny were picked.
For a new fog-:!
allele one expects the F2 to be self-fertile
and about half to segregate Unc-females (unc-51 is tightly
linked tofog-2, see Figure 1) while the remaining FB animals
segregate non-Unc (q71) females. The possibility of reces-
sive lethal events that were induced in cis to unc-51 and
that failed to complementfog-2(q71) was tested by searching
for F2 hermaphrodites that did not
or female animals; none of this class was found. New
dominant FemiFog mutations, e.g., tra-2(&, see below,
caused up to 50% of the F2 cross progeny to be female.
Seven fog-2 alleles were isolated from 23,407 mutagenized
haploid genomes; q86, q123, q124, were obtained at 25",
while q166, q167, ql70, and q177 were obtained at 15". q86
was isolated in cis to unc-51(e369) while all others were in
cW to unc-51(e1189). To ensure independence offog-2 alleles
only one mutant from a given cross was retained.
3. Extragenic suppressors o f fem-jr(q2Ogf). fern-jr(gf) mutants
are self-fertile at 15" and sterile (Mog) at 25". Mog animals
are sterile because they produce a vast excess of sperm and
no oocytes and thus have a ~asculini~ation
phenotype (Table 1; B.\KION, S(,11tm. and Klr,11G 1987).
Mutations i n &-2 suppress frrw3(g/), so the log-2: /ov-
t w -
with EMS as de-
segregate either Unc
gf the germline
T. Schedl and
determined by two- and three-factor crosses (Table 2; see MATE-
RIALS ASD METHODS). The positions of loci shown below the line
are from SWASSOS, EDGLEY and RIDDLE (1984) and HODGKIS and
of the right arm of chromosome V. The
fog-2, emb-4, and Ste(g265) (above the line) were
3(gf) double mutant is self-fertile (see RESULTS).
dpy-20 L4 XX animals were mutagenized with EMS and
allowed to produce self-progeny at 15'. Adult F1 animals
were picked, shifted to 25" and plates were screened for
self-fertile F2 animals. One fog-2 allele, q113, was obtained
by this procedure.
4. Complementation suppression
2(y71)l+. The strongest fem-j(gf) allele is q95, which is
100% Mog as a heterozygote at 25" (B~Rroti, SCHEDL
KIMBLE 1987). It was found that fem-3(q95gf)l+ ; fog-2(q71)/
+ is also Mog at 23". However, when fog-2(y71) is homo-
I\gotIs,~(.,,r-?1qVjS/)i + is no longer;t tlomin;mt sterile ( ; ~ l ) o u t
80-90% o f ' animals are self-trl-tile). .l'hus 21 newly induced
fog-2 allele that fails to complement fog-2(y71) in trans will
suppress the dominant sterility of fem-?(y95gf)l+ ; fog-
2(y71)1+. unc-51 (el 189) L4 XX animals were mutagenized
with EMS and then crossed at 24" with fem-3(y95gf) dpy-
20; him-5 fog-2(y71) males (from a 13' stock). Non-Unc FI
animals were screened for self-fertile hermaphrodites (or
eggs on the plate) among a sea of Mog animals and males.
Any Fl self-fertile animals were picked and new fog-2 alleles
were sought as F2 Unc females. In a number of cases, the
F, self-fertile animal had been mated by a sibling male (as
judged by male progeny in the F2), and as a result, more
than one type of mutagenized
present. If the hermaphrodite had been mated,
Unc F2 L4 XX animals were picked and F B Unc females
were sought. As in the "screen for mutations that
complement fog-2," candidates were examined for sterile
or lethal non-complementing alleles, but none was found.
Four fog-2 alleles (y247, q249, q251, and y263) were ob-
tained from this procedure. In addition, a sterile mutation
that does complement fog-2, and that does not suppress
fem-3(y95gf)l+, Ste(y265), was fortuitously isolated in this
mutagenesis. Ste(q265) is closely linked tofog-2 (Figure I),
and thus useful in balancing unc-51 fog-2 doubles. The
phenotype of Ste(y265) XX animals is an arrest in gonadal
o f fem-j(y95gf)I + ; fog-
unc-51 chromosome was
development, lack of a vulva, and a reduced number
germ cells in which the only gametes to develop are sperm.
All putative fog-2 alleles were out-crossed at
times to N2. Where applicable, fem-j(y20gf or q95gf) dpy-
20 were removed during out-crossing based on loss of the
tightly linked dpy-20 marker and the absence of the Mog
phenotype of fern-j(gf)/ + at 25'. All alleles were then tested
(or retested) for failure
to complement fog-2(q71). Mapping
showed that all alleles are tightly linked to unc-51 on the
right arm of chromosome V (see below). The fog-2 alleles
isolated linked to unc-51 were maintained as heterozygotes
balanced by emb-4(hc60) or by Ste(q265). For construction
of maleifemale strains (see below),
removed by two factor crosses.
Analysis offog-2 mutants: Females can he distingt~ishetl
from hernla~~l~rotlites either using ;I dissecting microscope
or by Nomarski differential interference microscopy. With
a dissecting microscope, adult females are non-eggbearing,
and as such, the ventrally located
appears as a clear patch. Further,
accumulate in females giving
gonads a striped appearance. To identify females unequiv-
ocally, animals were examined by Nomarski optics with a
Zeiss Planapo 63X lens at X 630 magnification. By Komar-
ski, females in L, lethargus or as young adults lack sperm
and primary spermatocytes in both the gonad and the
spermatheca. The gametes that develop most proximally
are oocytes (see RESL-L-I-s and Figure
To determine the penetrance of the Fog phenotype, XX
L4 animals were picked to individual plates and examined
by the dissecting microscope as adults [about 24 hr (20 or
25O) or about 36 hr (1 5")] for a female morphology and
the absence of eggsllarvae. Selected animals were further
examined by Nomarski. In some situations (such as strain
constructions), L4 XX animals were transferred in groups
of 20 to 50 and scored as above.
Males (XO) were examined by the dissecting microscope
for the presence of a male tail, the absence of a vulva and
for mating behavior. The following sexually dimorphic
structures were examined by Nomarski optics for mor-
phology and size to learn if there was any feminization of
males: the germ line, the male gonad consisting of a single
reflexed arm and vas deferens (KLASS, WOLF and HIRSH
1976; KIMBLE and HIRSH 1979), and the
rays (9 pairs), and copulatory
type and position of gametes and germ cells within the
gonad (HIRSH, OPPENHEIM and KLASS 1976) and the pres-
ence of yolk in the pseudocoelom (refractile droplets,
KIMBLE and SHARROCK 1983; DONIACH 1986a) were scored.
For each of the fog-2 alleles, more than 40 X0 animals
the linked unc-51 was
uterus is empty and
the proximal arm of the
2 for further
bursal fan, sensory
spicules of the
1980). Further, the
plane was adjusted to show the two gonad arms. Scale bar = 40 Fm. Line drawings are shown below. (c and d), Higher magnification
photomicrographs of boxed gonad region from (a) and (b), respectively. Scale bar = 10 p,m. Line drawings of gamete type and position
in gonad are shown below. (a and c), Wild-type young adult hermaphrodite. (b and d),
proximal part of the anterior gonad arm of a hermaphrodite. The first, most proximal, germ cells have differentiated as sperm. Subsequent
germ cells have differentiated as oocytes. (d), Gametes in the proximal part of the posterior gonad arm of afog-2 female. The first, most
proximal, germ cells have a transformed sexual fate and have differentiated as oocytes instead of sperm. There is no evidence of sperm
or spermatogenesis. Note that in both the hermaphrodite (c) and the female (d) the spermatheca (striped
arrows-oocytes. Oocytes are very large cells, with a large smooth nucleus, and have a granular cytoplasm. Immature oocytes are smaller,
have a large nucleolus, and also have a granular cytoplasm. Thin arrow-sperm.
Arrow head-primary spermatocyte. Note that the pattern of gametogenesis in the anterior and posterior gonads are
hermaphrodite, and similarly, they are equivalent in the fog-2 female, For further morphological details see HIRSH, OPPESHILSI and
(1976) and KIMBLE and WARD (1988).
andfog-2(971) XX germ-line phenotypes. (a and b), Composite photomicrographs using h'omarski optics. Focal
fog-2 young adult female. (c), Gametes in the
in drawings) is empty. Thick
Sperm are small, with a tiny elevated refractile nucleus.
equivalent in the
Hermaphrodite Germ-Line Sex