SEPA-1 Mediates the Specific Recognition
and Degradation of P Granule
Components by Autophagy in C. elegans
Yuxia Zhang,1,4Libo Yan,2,1,4Zhi Zhou,1,4Peiguo Yang,1,4E Tian,1Kai Zhang,3Yu Zhao,1Zhipeng Li,1Bing Song,1
Jinghua Han,1Long Miao,3and Hong Zhang1,*
1National Institute of Biological Sciences, Beijing 102206, P.R. China
2Graduate Program in Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing 100730, P.R. China
3National Laboratory of Biomacromolecules, Institute of Biophysics, Chinese Academy of Sciences, Beijing 100101, P.R. China
4These authors contributed equally to this work
How autophagy, an evolutionarily conserved intracel-
degrades protein aggregates is poorly understood.
Here, we show that several maternally derived germ
P granule components are selectively eliminated by
genesis. The activity of sepa-1 is required for the
degradation of these P granule components and for
their accumulation into aggregates, termed PGL
granules, in autophagy mutants. SEPA-1 forms pro-
tein aggregates and is also a preferential target of
autophagy. SEPA-1 directly binds to the P granule
component PGL-3 and also to the autophagy protein
LGG-1/Atg8. SEPA-1 aggregates consistently coloc-
alize with PGL granules and with LGG-1 puncta.
Thus, SEPA-1 functions as a bridging molecule in
mediating the specific recognition and degradation
of P granule components by autophagy. Our study
reveals a mechanism for preferential degradation of
protein aggregates by autophagy and emphasizes
the physiological significance of selective autophagy
during animal development.
Autophagy, the primary intracellular catabolic mechanism for
degradation of cytosol or damaged organelles, involves the
formation of a double-membrane structure, termed the autopha-
gosome, which engulfs a portion of the cytoplasm and/or organ-
elles and delivers it to the lysosome for degradation (Klionsky,
2005). In yeast, conserved proteins have been identified that
act collaboratively in distinct steps of autophagosome biogen-
esis (Levine and Klionsky, 2004; Suzuki and Ohsumi, 2007).
These proteins include the Atg1 protein kinase complex that
regulates the induction of autophagy; the Atg6/Vps34 class III
phosphatidylinositol-3 kinase complex that mediates vesicle
nucleation; the two ubiquitin-like conjugation systems that are
required for vesicle expansion; and the retrieval complex that
preautophagosomal structures (PAS) and non-PAS structures
(van der Vaart et al., 2008). In the two ubiquitin-like conjugation
pathways, Atg8 is conjugated to phosphatidylethanolamine
through the sequential action of processing enzyme Atg4,
E1-like activating enzyme Atg7, and E2-like conjugating enzyme
Atg3. The lipidated form of Atg8 associates with the autophago-
somal membrane. The sequential action of Atg7 and E2-like
enzyme Atg10 leads to conjugation of Atg12 to Atg5 (Klionsky,
2005). The autophagic machinery is evolutionarily conserved,
although some functional counterparts have yet to be identified
in other organisms.
Autophagy plays an important role in many physiological
processes in mammalian cells, including response to nutrient
stress, innate and adaptive immunity, and autophagic cell death
(Mizushima et al., 2008). Autophagy also functions as an impor-
tant quality control system in eliminating disease-related mutant
proteins associated with various neurodegenerative disorders,
including mutant variants of huntingtin in Huntington’s disease
and a-synuclein in familial Parkinson’s disease (Rubinsztein,
2006). The basal constitutive level of autophagy also removes
diffuse aggregate-prone proteins, and loss of autophagy activity
leads to intracellular accumulation of polyubiquitinated protein
aggregates, particularly in neurons or hepatocytes (Komatsu
et al., 2006; Hara et al., 2006). The formation of polyubiquitin-
containing protein aggregates requires the activity of p62/
sequestosome-1 (SQSTM1), which possesses a polyubiquitin-
binding activity and is present in ubiquitin-positive protein
aggregates (Komatsu et al., 2007; Nezis et al., 2008). p62 forms
protein aggregates and is a preferential target of autophagy,
LC3/Atg8 (Pankiv et al., 2007; Bjorkoy et al., 2005). Thus, p62
may be involved in linking polyubiquitinated protein aggregates
to the autophagic machinery, facilitating the clearance of such
aggregates. It remains to be determined whether nonubiquiti-
nated protein aggregates could be selectively removed by
autophagy and how such aggregates might be recognized and
308 Cell 136, 308–321, January 23, 2009 ª2009 Elsevier Inc.
A specialized type of protein aggregate, which is suggested to
carry the germ cell determinant, is segregated from oocytes into
germ precursor cells in the formation of germ cells in many
organisms, including C. elegans and Drosophila (Strome and
Lehmann, 2007). In C. elegans, these protein aggregates, called
germ P granules, are maternally contributed and segregated
exclusively into the germline blastomere during early embryonic
divisions (Hird et al., 1996). P granules continue to be expressed
in all of the descendants of the primordial germ cell P4, with the
exception of mature sperm, and are associated with the outer
surface of the nuclear envelope. Restriction of P granules to
germline blastomeres is achieved through a combination of
events, including migration of P granules toward the cytoplasm
inherited by germ blastomeres and degradation and/or disas-
sembly of P granules that remain in the cytoplasm of somatic
daughters during early embryonic divisions (Hird et al., 1996).
However, the mechanism underlying P granule depletion in
somatic cells is not known.
In this study, we show that several maternally derived P
granule components that remain in somatic cells during early
embryonic divisions are selectively degraded by autophagy.
sepa-1 is required both for the formation and degradation of
PGL granules. SEPA-1 directly interacts with the P granule
component PGL-3 and the autophagy protein LGG-1/Atg8,
linking PGL granules to autophagy. SEPA-1 belongs to a family
of C. elegans proteins that are preferential substrates of
autophagy, suggesting that SEPA-1 family members may func-
tion as adaptor proteins in mediating selective autophagic
Formation of PGL Granules in Somatic Cells
in Autophagy Mutants
To understand how germ P granules are exclusively localized in
germ cells, we performed a genome-wide RNAi screen to iden-
tify genes whose loss of function caused ectopic accumulation
of the P-granule-specific reporter, GFP::PGL-1, in somatic cells
(Figures 1A and 1B) (Kawasaki et al., 1998). We found that RNAi
inactivation of lgg-1, encoding the ortholog of yeast autophagy
protein Atg8, resulted in the formation of GFP::PGL-1-positive
granules in somatic cells at embryonic and larval stages (Figures
1C–1E). To determine the distribution of endogenous PGL-1,
lgg-1 mutant animals (without the gfp::pgl-1 reporter) were
stained with the monoclonal anti-PGL-1 antibody K76. One to
three granules were found to be dispersed in the cytoplasm in
lgg-1 somatic cells (Figures 1F–1H). The formation of perinuclear
P granules in germ cells was not affected in lgg-1 mutants
(Figure 1H). Thus, loss of function of lgg-1 leads to the accumu-
lation of PGL-1 granules in somatic cells.
In addition to PGL-1, germ P granules contain other compo-
nents that associate with P granules at all stages of germline
development, including PGL-3 and an unidentified P-granule-
specific epitope recognized by the monoclonal antibody
OIC1D4 (Kawasaki et al., 1998, 2004). Costaining of lgg-1
animals with anti-PGL-1 and anti-PGL-3 antibodies revealed
that PGL-1 and PGL-3 were colocalized in granules in somatic
cells (Figures 1I–1K). The somatic granules stained by OIC1D4
were also colocalized with those labeled by a rabbit anti-
PGL-1 antibody in lgg-1 animals (data not shown). These
somatic PGL-1-positive granules, which contain multiple germ
P granule components, were referred to as PGL granules. PGL
granules, however, lack components that are transiently associ-
ated with germ P granules in germline blastomeres and also do
not contain germline RNA helicases GLH-1 and GLH-4 (Figures
S1 and S2 available online).
When the formation of PGL granules was examined in other
autophagy mutants, we found that loss of function of compo-
nents involved in the two ubiquitin-like pathways, including
atg-4.1, atg-3, lgg-3/atg12, atg-7, and lgg-2/atg8, caused accu-
mulation of PGL granules (Figures 1L–1O and Table S1). PGL
granules were also found in unc-51, atg-18, and vps-16 mutant
embryos (Figure 1P and Table S1), which encode components
of the Atg1 kinase complex, the retrieval complex, and the
complex required for autophagosome/lysosome fusion, respec-
tively. Furthermore, mutant alleles of lgg-1, atg-3, atg-7, atg-10,
and atg-18 were isolated in the genetic screen to identify
mutants that cause formation of PGL granules in somatic cells
(Figure S3). Taken together, these results indicate that autoph-
agy is important for preventing the formation of PGL granules
in somatic cells.
We next determined whether somatic cells in autophagy
mutants have been transformed to a germ cell fate. Reporters for
various types of somatic cells, including neurons, pharyngeal
muscles, seam cells, and intestine cells, are appropriately
expressed in autophagy mutant animals (data not shown). atg-18
null mutants are viable and fertile. Null mutants for lgg-1, atg-3,
and atg-7 develop into morphologically normal-looking larvae.
Thus, despite the presence of germline P granule components,
somatic cells in autophagy mutants still properly differentiate.
Gradual Formation of PGL Granules in Autophagy
The temporal formation of PGL granules in lgg-1 embryos was
examined by immunostaining with K76. P granules were only
detected in the posterior P1 cell at the 2-cell stage (n = 3) and
were exclusively partitioned into the germ blastomere before
the 16-cell stage (n = 3) in lgg-1 embryos (Figures S4A, S4B,
S4K, and S4L). In 16-cell stage lgg-1 embryos, a few PGL gran-
ules were faintly detected in somatic cells, whereas, in 24-cell
stage lgg-1 embryos, many brightly stained PGL granules were
observed (Figures S4C, S4D, S4M, and S4N and data not
shown). PGL granules increased in number as development
proceeded around the comma to 2-fold stage and were present
throughout embryogenesis (Figures S4E–J and S4O–S4T).
These results indicate that PGL granules are progressively
formed in lgg-1 embryos.
PGL Granule Components in Autophagy Mutants
Are Maternally Derived
PGL granule components in autophagy mutants could be
synthesized from zygotically transcribed mRNAs or maternally
contributed. To distinguish between these possibilities, we
determined whether somatic PGL-3-positive PGL granules in
autophagy mutants are dependent on maternal pgl-3. When
the distribution of PGL-3 in embryos born from a cross between
Cell 136, 308–321, January 23, 2009 ª2009 Elsevier Inc. 309
lgg-1 males and lgg-1; pgl-3(0) hermaphrodites was examined,
F1 embryos produced from lgg-1; pgl-3(0) males mated to lgg-1
hermaphrodites contained PGL-3 granules in somatic cells
(Figures 1S and 1T). Similarly, PGL-1-positive PGL granules in
lgg-1 mutants were dependent on maternal pgl-1 (data not
shown). To rule out the possibility that somatic PGL-1 proteins
in autophagy mutants are derived from zygotically transcribed
mRNA, we analyzed the pgl-1 mRNA levels and found no induc-
tion of pgl-1 transcripts in lgg-1 embryos (Figure 1U). Taken
together, our results strongly argue that PGL granule compo-
nents in autophagy mutants are derived from the oocyte. The
maternally loaded pgl-1 and pgl-3 mRNAs, which are present
in somatic blastomeres during early embryonic divisions (Kawa-
saki et al., 2004), could also be translated and accumulated into
PGL granules in autophagy mutants.
Presence of PGL-1 in Membrane-Surrounded
Structures in Somatic Cells
To strengthen the evidence that P granule components in
somatic cells are degraded by autophagy, we performed immu-
noelectron microscopy (immunoEM). Sections of wild-type,
early-stage embryos were incubated with the anti-PGL-1
antibody K76 followed by gold-conjugated secondary antibody
to determine whether PGL-1 was located in membrane-sur-
rounded structures in somatic cells. Consistent with the idea
that germline P granules that remain in somatic cells during early
embryonic divisions are disassembled and quickly removed,
gold-labeled PGL-1 particles were only rarely observed. After
examining 83 EM sections, in the cytoplasm of somatic cells,
we detected nine membrane-surrounded structures containing
PGL-1 particles (Figures 1V and 1W). These membrane-bound
structures are similar in dimension to autophagolysosomes
Figure 1. Formation of PGL Granules in Autophagy Mutants
(A and B) GFP::PGL-1-labeled P granules are restricted to germ precursor cells Z2 and Z3 (arrows in [B]). (A) Nomarski image of the embryo shown in (B).
(C and D) Ectopic GFP::PGL-1-positive granules are present in somatic cells in lgg-1(RNAi) embryos. (C) Nomarski image of the embryo shown in (D).
(E) Formation of ectopic GFP::PGL-1 granules throughout the animal in an lgg-1(RNAi) larva (marked with abar and arrows). Perinuclearlocalization of P granules
in germ cells is highlighted in red. Irregular particular fluorescence in the middle body region is gut autofluorescence.
(F and G) Immunostaining with anti-PGL-1 antibody K76 shows that PGL-1-positive granules are formed in somatic cells in lgg-1 embryos.
(H) Confocal image showing ectopic PGL-1-positive granules in lgg-1 embryos (green). Nuclei are labeled by DAPI (blue).
(I–K) Costaining with a rabbit anti-PGL-1 antibody (I) and a rat anti-PGL-3 antibody (J), showing that the stained granules are colocalized in lgg-1 embryos (K).
(I)–(K) are confocal images.
(L–P) Formation of PGL granules in autophagy mutants. atg-4.1 (L), atg-3 (M), lgg-3 (N), atg-7 (O), and unc-51 (P). PGL granules are shown by gfp::pgl-1 reporter
(L and M) or are detected by K76 (N–P).
(Q and R) No PGL-3-positive granules are formed in somatic cells in embryos born from a cross between lgg-1 males and lgg-1; pgl-3 hermaphrodites. (R) DAPI
image of the embryo shown in (Q).
(S and T) PGL-3-positive P granules are observed in somatic cells in embryos produced from lgg-1; pgl-3 males mated to lgg-1 mothers. (T) DAPI image of the
embryo shown in (S).
(U) The mRNA level of pgl-1, detected by RT-PCR, is the same in wild-type and lgg-1 mutant embryos. The mRNA level of actin serves as a control.
(V–W) ImmunoEM images showing PGL-1 gold particles enclosed by membrane. Arrows indicate gold particles, and arrowheads indicate structures that appear
to be membrane limited.
The number of PGL-1 gold particles in each membrane-bound structure ranged from one to six. Early-stage embryos (before ?100-cell stage) were used for this
310 Cell 136, 308–321, January 23, 2009 ª2009 Elsevier Inc.
that were previously reported in C. elegans (Melendez et al.,
2003), further supporting the notion that P granule components
are removed by autophagy in somatic cells.
sepa-1 Is Essential for Formation of PGL Granules
in Autophagy Mutants
To investigate how PGL granules are formed in autophagy
mutants, we performed genetic screens to identify mutants
that suppressed the formation of PGL granules in lgg-1 mutants
and atg-18 mutants (Figures 2A and 2B). Eleven mutant alleles
of a single genetic locus, termed sepa-1 (suppressor of ectopic
P granule in autophagy mutants), were isolated from ?9500
genomes screened. In sepa-1; lgg-1 and sepa-1; atg-18
mutants, GFP::PGL-1 was homogeneously distributed in the
cytoplasm of somatic cells (Figures 2C and 2D), indicating that
PGL-1 proteins were still present but failed to form granules.
The formation of endogenous PGL granules in lgg-1 and
atg-18 mutants was also suppressed by mutations in sepa-1
(Figures 2E and 2H and data not shown). Mutations in sepa-1
also suppressed the formation of PGL granules in other autoph-
agymutants, including lgg-3andatg-7 animals(datanotshown).
The formation of perinuclear P granules in germline cells,
however, was not affected by sepa-1 mutations (Figures 2F
and 2H). Therefore, sepa-1 is essential for the accumulation of
PGL-1 and PGL-3 into PGL granules in autophagy mutants.
sepa-1 Is Essential for Degradation of PGL Granule
Components in Somatic Cells
We next investigated whether sepa-1 is required for the degra-
dation of P granule components that remain in somatic cells
during early embryonic divisions. In wild-type early embryos,
GFP::PGL-1 is restricted to germ precursor cells. However,
diffuse GFP::PGL-1 signal was observed in somatic cells from
the 2-cell stage onward in sepa-1 mutants (Figure 2I), indicating
that GFP::PGL-1 failed to be removed from somatic cells in
the degradation of P granule components in somatic cells, we
examined the level of endogenous PGL-3 protein. In wild-type
animals, the level of PGL-3 was high in early embryos (before
the 8-cell stage) but decreased dramatically in comma-stage
embryos (Figure 2J). However, high levels of PGL-3 persisted
in sepa-1 mutant comma-stage embryos (Figure 2J). These
results demonstrate that sepa-1 is involved in the clearance of
maternally derived P granule components in somatic cells.
sepa-1 was mapped on the right side of chromosome I, at about
+17.31, and cloned by transformation rescue (Figure 3A). A PCR
fragment containing a single predicted gene, M01E5.6, restored
the formation of PGL granules in sepa-1; atg-18 mutants
(Figure 3A; see Supplemental Experimental Procedures for
details). Furthermore, RNAi knockdown of M01E5.6 disrupted
the formation of PGL granules in atg-18 animals. Thus, suppres-
sion of PGL granule formation in autophagy mutants is caused
by reduced activity of sepa-1.
sepa-1 encodes a protein of 702 amino acids (Figure 3B). The
N terminus of SEPA-1 is rich in helical structures, whereas the
Cterminus contains a proteininteraction KIX domain (Figure3B).
The molecular lesions in 11 sepa-1 mutants were identified by
sequencing the corresponding genomic DNA (Figure 3B).
The presence of helical-rich structures and a KIX domain in
SEPA-1 led us to test whether SEPA-1 self-interacts. In an
in vitro pull-down assay, we detected a strong interaction
preparing a series of SEPA-1 deletion fragments, we mapped
the region required for self-association to amino acids 39 to
160 (Figure 3C and data not shown). Therefore, SEPA-1 is able
His-SEPA-1(Figure 3C). By
SEPA-1 Forms Cytoplasmic Aggregates
To determine the expression pattern of sepa-1, we raised
antibodies against the KIX domain of SEPA-1 (Figure S7). The
specificity of this anti-SEPA-1 antibody is confirmed by the
lack of SEPA-1 staining in sepa-1(bp402) animals (Figure S7),
in which the KIX domain of SEPA-1 is deleted. In wild-type
animals, no obvious expression of sepa-1 could be detected
by anti-SEPA-1 antibody in embryos before the 16-cell stage
(Figures 4A and 4B). A few SEPA-1 aggregates were found
in ?16-cell stage embryos, and the number of SEPA-1 aggre-
gates was dramatically increased as the embryo developed to
the ?100-cell stage (Figures 4C and 4D). As development pro-
ceeded, SEPA-1 aggregates disappeared, and only a few cells
contained SEPA-1 aggregates by the comma stage (Figures
4E and 4F). Immunostaining with anti-SEPA-1 antibody showed
that SEPA-1 was not present in germline P granules at all embry-
onic and larval stages (Figures 4G–4J and S7).
Because of homogenous background fluorescence from the
secondary antibody, we were unable to determine whether
weak diffuse SEPA-1 immunofluorescence signal was present
in somatic cells. Therefore, we constructed a reporter that
contains the entire coding sequence and promoter of sepa-1,
with gfp joined in-frame at the C terminus, to further assess the
expression pattern of sepa-1 in somatic cells. This translational
reporter fully rescued the sepa-1 phenotype. The earliest
appearance of SEPA-1::GFP was in ?16-cell stage embryos,
where it was diffusely distributed in the cytoplasm of almost
all cells, with a few small aggregates (Figures 4K and 4L).
SEPA-1::GFP formed aggregates in a temporal pattern similar
to that shown by the anti-SEPA-1 antibody (Figures 4M–4P).
Diffuse SEPA-1::GFP signal was still evident in most cells at
the comma stage and was greatly diminished by the 2-fold stage
of embryogenesis (Figures 4O and 4P and data not shown). After
hatching, cytoplasmic SEPA-1 aggregates were found in a few
unidentified cells in the head and tail regions and also in the
intestine,especially intheanterior andposteriorpairsofintestine
cells (Figures 4Q and 4R).
SEPA-1 Is Degraded by Autophagy
The gradual disappearance of SEPA-1 during embryogenesis
and its requirement for degradation of PGL granule components
suggested that SEPA-1 could be a preferential target of autoph-
agy. Indeed, compared to the expression pattern of sepa-1 in
wild-type animals, we found that the number of SEPA-1 aggre-
gates was dramatically increased in lgg-1 animals after the
100-cellstage(Figures 4S–4W).Forexample, SEPA-1
Cell 136, 308–321, January 23, 2009 ª2009 Elsevier Inc. 311
Figure 2. Mutations in sepa-1 and pgl-3 Suppress
Formation of PGL Granules in lgg-1 and atg-18
(A) Restrictionof GFP::PGL-1 to germ precursor cells in awild-
(B) GFP::PGL-1-positive granules are observed in somatic
cells in an atg-18 embryo.
(Cand D) Formation of PGLgranules inlgg-1(C) and atg-18 (D)
mutants is suppressed by sepa-1 mutations. GFP::PGL-1
signal is diffusely distributed in somatic cells in a sepa-1;
lgg-1 (C) and a sepa-1; atg-18 embryo (D).
(E–H) Suppression of endogenous PGL granule formation in
atg-18 mutants by sepa-1 mutations. P granules are stained
with K76 (E and F) or anti-PGL-3 antibody (G and H). (G) and
(H) are confocal images.
(I) GFP::PGL-1 is diffusely distributed in somatic cells in
a sepa-1 embryo.
(J) Western blot analysis showing that endogenous PGL-3
8-cell stage) to comma-stage embryos (labeled as ‘‘early’’ and
‘‘late,’’ respectively), whereas high levels of PGL-3 persist in
sepa-1 mutant comma-stage embryos.
(K) Formation of PGL granules in lgg-1 mutants is suppressed
by mutations in pgl-3. GFP::PGL-1 signal is diffusely distrib-
uted in somatic cells in a lgg-1; pgl-3(bp438) embryo. PGL-1
proteins still accumulate into P granules in germline cells in
pgl-3 mutant animals.
(L) GFP::PGL-1 is diffusely distributed in somatic cells in
a pgl-3(bp438) embryo.
Pictures shown in (A)–(D), (I), (K), and (L) were taken using the
same exposure time. Pixel quantification: 1388 3 1040.
312 Cell 136, 308–321, January 23, 2009 ª2009 Elsevier Inc.
aggregates were present throughout the animal body in lgg-1 L1
larvae (Figure 4W). Furthermore, SEPA-1 accumulated into
larger aggregates in lgg-1 animals. Other components of the au-
tophagic machinery, including atg-3, atg-7, and lgg-3, were also
required for the removal of SEPA-1. To rule out the possibility
that the increased expression of sepa-1 in autophagy mutants
was due to upregulation of sepa-1 transcription, we analyzed
sepa-1 mRNA levels and found no obvious alteration in wild-
type and lgg-1 embryos (Figure 4X). These results indicate that
SEPA-1 is removed by autophagy during animal development.
SEPA-1 Aggregates Are Colocalized with PGL Granules
in Autophagy Mutants
nents into granules in autophagy mutants prompted us to
examine whether SEPA-1 aggregates are colocalized with PGL
granules in autophagy mutants. When lgg-1 embryos carrying
the sepa-1 reporter were immunostained with anti-PGL-1 or
anti-PGL-3 antibody, SEPA-1 aggregates were colocalized with
PGL-1-positive or PGL-3-positive granules (Figures 5A–5F).
Costaining with anti-SEPA-1 and anti-PGL-3 antibodies also
revealed the colocalization of endogenous SEPA-1 and PGL-3
component of PGL granules present in autophagy mutants.
SEPA-1 Directly Interacts with PGL-3
To understand how SEPA-1 is involved in the formation of PGL
granules in autophagy mutants, we identified proteins that
interact with SEPA-1 by screening a yeast two-hybrid C. elegans
cDNA library. Eleven positive clones were isolated from about
2.2 3 106clones screened. Nine clones turned out to be PGL-
3 (Figure 5J). Direct interaction between SEPA-1 and PGL-3
was also confirmed by an in vitro GST pull-down assay (data
not shown). By constructing a series of PGL-3 fragments, we
found that amino acids 581 to 614 of PGL-3 interacted with
SEPA-1 (Figures 5J and 5K). The PGL-3 interacting domain in
SEPA-1 was located at amino acids 39 to 160 (Figure 5L).
We next investigated whether SEPA-1 associates with PGL-3
in vivo. Extracts of lgg-1 mutant embryos were immunoprecipi-
tated with anti-PGL-3 antibody, and copurified proteins were
detected by western blotting by using anti-SEPA-1 antibody.
We found that SEPA-1 was specifically coimmunoprecipitated
with anti-PGL-3, but not with the preimmune serum (Figure 5M).
Thus, SEPA-1 associates with PGL-3 both in vitro and in embryo
SEPA-1 Aggregates Are Colocalized
with GFP::LGG-1 Dots
To further investigate the link between SEPA-1 and autophagy,
we examined whether SEPA-1 is colocalized with autophago-
some structures. The GFP::LGG-1 reporter has been widely
used to visualize autophagosomes in C. elegans, as it binds to
autophagic membranes (Melendez et al., 2003). GFP::LGG-1
has diffuse cytoplasmic distribution in embryos with some areas
of punctate staining (Figure 6A). When animals carrying both
sepa-1::rfp and gfp::lgg-1 transgenes were examined, 94% of
SEPA-1 aggregates (n = 34) were colocalized with GFP::LGG-1
dots (Figures 6A–6C). GFP::LGG-1 dots that did not show
SEPA-1::RFP signal may correspond to autophagosomes that
were involved in the degradation of other substrates.
SEPA-1 Directly Interacts with LGG-1
To study how SEPA-1 is recruited to autophagosome structures,
we performed in vivo coimmunoprecipitation experiments to
examine whether SEPA-1 interacts with LGG-1, whose mamma-
lianhomologLC3hasbeen shownto associatewithp62inmedi-
ating its degradation (Pankiv et al., 2007). Extracts from embryos
expressing the gfp::lgg-1 reporter were precipitated with anti-
GFP antibody, and the resulting immunoprecipitates were sub-
jected to western blotting by using anti-SEPA-1 antibody. We
found that SEPA-1 was specifically coimmunoprecipitated by
anti-GFP antibody (Figure 6D).
In vitro pull-down assays were performed to examine whether
SEPA-1 directly interacts with LGG-1. We found that SEPA-1
bound to LGG-1, but not to ATG-18 (Figure 6E). By using a series
of SEPA-1 fragments, we found that the SEPA-1 fragment
containing amino acids 289 to 575, which is distinct from the
PGL-3 interaction domain, strongly bound to LGG-1 (Figure 6E).
Thus,SEPA-1 couldberecruited toautophagosomesthrough its
direct interaction with LGG-1.
PGL-3 Is Required for Accumulation of PGL-1 into PGL
Granules in Autophagy Mutants and for Degradation
of PGL-1 by Autophagy
In addition to sepa-1, we isolated three alleles of the same
genetic locus, bp438, bp439, and bp458, that suppressed the
formation of somatic PGL-1-positive granules in atg-18 and
lgg-1 mutants (Figure 2K and data not shown). Subsequent
genetic and molecular analysis demonstrated that these are
new alleles of pgl-3. In pgl-3 atg-18 and lgg-1; pgl-3 mutants,
PGL-1-positive granules in somatic cells were almost undetect-
able. Concomitantly, GFP::PGL-1 was diffusely distributedin the
cytoplasm (Figure 2K). GFP::PGL-1 also failed to be degraded
and was diffusely localized in the cytoplasm of somatic cells
from the 2-cell stage onward in pgl-3 single-mutant embryos
(Figure 2L). Thus, accumulation of PGL-1 into granules in
somatic cells and degradation of PGL-1 by autophagy require
PGL-3. In contrast, PGL-1 is not required for the accumulation
of PGL-3 into granules in autophagy mutants (data not shown).
SEPA-1 still formed aggregates in pgl-1 and pgl-3 mutants,
and the number of SEPA-1 aggregates was increased in lgg-1;
pgl-1 and lgg-1; pgl-3 mutants (data not shown), indicating
that formation of SEPA-1 aggregates and autophagic degrada-
tion of SEPA-1 are not dependent on PGL-1 and PGL-3.
SEPA-1 Family Proteins Are Selectively Degraded
Bioinformatic analysis revealed that SEPA-1 belongs to
a protein family (the PANTHER family: PTHR21504:SF10), with
11 members in C. elegans (Figure S11). Phylogenetic analysis
grouped the SEPA-1 family members into three major sub-
groups (Figure S12A). Interestingly, the genes of the sepa-1
family are clustered in the genome. Nine genes are located in
an ?67 kb region. The other two members, C35E7.1 and
C35E7.6, are separated from each other by four genes.
Cell 136, 308–321, January 23, 2009 ª2009 Elsevier Inc. 313
Figure 3. Molecular Structure of sepa-1
(A) sepa-1 maps close to the polymorphic marker pkp1133. Fosmid WRM0637ac10 and a PCR product containing gene M01E5.6 both restored formation of
somatic PGL granules in sepa-1; atg-18 animals. Intron-exon boundaries of M01E5.6 were confirmed by cDNA sequencing. LG I, Linkage group I.
(B) The protein sequence of SEPA-1. Mutations identified in sepa-1 mutants are shown in red. The predicted KIX domain is highlighted in yellow.
(C) SEPA-1 self-associates in in vitro pull-down assays, and its self-association domain maps to amino acids 39 to 160. GST-tagged full-length or truncated
SEPA-1 immobilized on glutathione Sepharose beads was incubated with His-SEPA-1, His-SEPA-1 (P71L), or His-SEPA-1 fragment (39–160). The proteins
314 Cell 136, 308–321, January 23, 2009 ª2009 Elsevier Inc.
Translational gfp reporters for other sepa-1 family members
were constructed, and their expression patterns were analyzed
in stable transgenic lines. F44F1.6 exhibited a similar expression
pattern to sepa-1. F44F1.6::GFP was strongly expressed in early
embryonic stages and accumulated into protein aggregates.
After the comma stage, its expression decreased and was
mainly diffusely localized in the cytoplasm (Figures 7A, 7B and
data not shown). Reporters for T04D3.1, T04D3.2, and
ZK1053.4 displayed a similar expression pattern, with both
diffuse and aggregated proteins localized in the cytoplasm at
all embryonic stages (Figures 7E, 7F, and S12). C35E7.6::gfp
and C35E7.1::gfp reporters were weakly expressed, and diffuse
GFP signal was observed in the cytoplasm in embryos (Figures
7I, 7J, and S12). VET-6::GFP was homogenously distributed in
the cytoplasm, and no aggregates were formed during embryo-
genesis (Figures 7M and 7N). ZK1053.3::GFP was restricted to
a few cell types, mainly diffusely localized in intestine cells and
a few head neurons (Figure S12). F44F1.4 was expressed only
in a few cells in the head region. No obvious GFP signal could
be detected for the F44F1.5::gfp reporter. Overall, 9 out of 11
SEPA-1 family members display a dynamic expression pattern
To determine whether other SEPA-1 family members are also
preferential targets of autophagy, we analyzed reporter expres-
sion patterns in lgg-1(RNAi) animals. We found that the expres-
sion levels of F44F1.6, T04D3.1, T04D3.2, ZK1053.4, C35E7.1,
C35E7.6, and ZK1053.3 reporters were dramatically increased
and more protein aggregates were formed in lgg-1(RNAi)
animals than in wild-type animals (Figures 7C, 7D, 7G, 7H, 7K,
7L, and S12). Stronger diffuse VET-6::GFP signal was also
retained after extensive washing were detected by western analysis using anti-His antibody. No interaction could be detected between GST-SEPA-1 and
His-tagged mutant SEPA-1 containing a P71L mutation found in bp409. Twenty percent of proteins used for each pull-down reaction are shown as input.
The lower panels show the amount of GST, GST-SEPA-1, and GST-SEPA-1 (39–160) proteins used in the pull-down assay. WB, western blot. Coom blue,
Figure 4. SEPA-1 Forms Cytoplasmic Aggregates and Is Degraded by Autophagy
(A and B) No obvious expression of sepa-1 could be detected by anti-SEPA-1 antibody in embryos before the 16-cell stage. An ?8-cell-stage embryo is shown.
(A) DAPI image of the embryo shown in (B).
(C and D) The number of SEPA-1 aggregates increases during early stages of embryogenesis (from ?20-cell to ?100-cell stages). (C) DAPI image of the embryo
shown in (D).
(E and F) SEPA-1 aggregates are almost undetectable in a comma-stage embryo. (E) Nomarski image of the embryo shown in (F).
(G–J) SEPA-1 is not present in germline P granules (also see Figure S7). P granules are detected by anti-PGL-3 antibody. (G)–(J) are confocal images.
(K and L) Expression of sepa-1::gfp is first detected in ?16-cell-stage embryos. SEPA-1 distribution is mainly diffuse, with a few aggregates. (K) Nomarski image
of the embryo shown in (L).
(M–P) The number of SEPA-1::GFP aggregates increases during early stages of embryogenesis (before ?100-cell stage) and decreases as development
proceeds. SEPA-1::GFP is still diffusely distributed in a comma-stage embryo (O and P) but is almost undetectable in a 4-fold-stage embryo (data not shown).
(M and O) Nomarski images of the embryos shown in (N) and (P), respectively.
(S–W)Expression ofsepa-1,detectedby anti-SEPA-1 antibody (Sand T)orsepa-1::gfp(U–W),isdramaticallyincreased inlgg-1animals.(Sand T)Comma stage.
(U and V) 4-fold stage. (W) L1 larva. (S and U) Nomarski images of the embryos shown in (T) and (V), respectively.
(X) The mRNA level of sepa-1, detected by RT-PCR, is the same in wild-type and lgg-1 mutants. The mRNA level of actin serves as a control.
Cell 136, 308–321, January 23, 2009 ª2009 Elsevier Inc. 315
A AB BC C
D DE EF F
G GH H
316 Cell 136, 308–321, January 23, 2009 ª2009 Elsevier Inc.
of F44F1.4::gfp and F44F1.5::gfp remained unchanged in
lgg-1(RNAi) animals (data not shown). Unlike sepa-1 mutations,
RNAi inactivation of other SEPA-1 family members affected
neither the formation of PGL granules in autophagy mutants
nor the degradation of PGL granule components by autophagy
(data not shown). In summary, 9 out of 11 SEPA-1 family
members appear to be targets of autophagy, raising the possi-
bility that they could function as adaptor proteins in mediating
selective degradation of distinct substrates by autophagy.
SEPA-1 Functions as an Adaptor Protein in Mediating
Degradation of PGL Granule Components by Autophagy
We showed here that autophagy selectively removes several
P granule components in somatic cells during C. elegans
embryogenesis. The activity of sepa-1 is required for formation
and degradation of PGL granules. In sepa-1 mutants, PGL
granule components are diffusely distributed in the cytoplasm
and fail to be eliminated. SEPA-1 directly interacts with PGL-3,
which, in turn, associates with PGL-1 (Kawasaki et al., 2004).
SEPA-1 also binds to LGG-1. Thus, SEPA-1 acts as an adaptor
protein in linking PGL granule components to the autophagic
machinery. Mutations in sepa-1 have no effect on the formation
of perinuclear germline P granules, which is consistent with the
finding that SEPA-1 is not detected in germline cells. It is also
possible that autophagy activity is inhibited or P granules are
stabilized by other factors in germline cells.
The function of SEPA-1 resembles that of p62 in mammalian
cellsand Atg19inyeast. p62confersa certainlevel ofspecificity,
through its ubiquitin binding activity, in mediating the recognition
of polyubiquitinated protein aggregates by the autophagic
machinery (Komatsu et al., 2007; Nezis et al., 2008). Atg19 is
transports aminopeptidase I (ApeI) and a-mannosidase (Ams1)
precursors from the cytoplasm to the vacuole. Atg19 recognizes
oligomerized cargo proteins and also directly binds to Atg11 and
Atg8, linking the cargo complex to the vesicle formation
machinery (Shintani et al., 2002). Taking these studies together
with our SEPA-1 results, it is possible that selective autophagic
degradation could be mediated by a subset of bridging mole-
cules that simultaneously bind to cargos and the autophagy
protein LGG-1/LC3/Atg8. Binding of LGG-1/LC3/Atg8 to prefer-
ential targets could trigger a cascade of events, leading to the
expansion and elongation of autophagosome structures and
their subsequent degradation by lysosomes.
Diffuse P Granule Components and SEPA-1
Are Degraded by Autophagy
Whether diffuse or aggregated forms of proteins are removed by
autophagy in mammalian cells isnotcompletely understood. We
found that SEPA-1 is diffusely distributed in the cytoplasm in
comma-stage embryos and diffuse SEPA-1 is almost undetect-
able by the 2-fold embryonic stage, indicating that diffuse
SEPA-1 or small intermediate oligomers, which could not be
detected by light microscopy, are recognized by autophagy.
SEPA-1 also forms aggregates, which are first detected in ?16
cell-stage embryos. SEPA-1 aggregates increase during early
embryogenesis (?20- to ?100-cell stage) and then disappear
by the comma stage. SEPA-1 aggregates could be directly
removed by the autophagic machinery. Alternatively, degrada-
tion of diffuse SEPA-1 could result in the movement of SEPA-1
from aggregates into the cytoplasm in a diffuse form.
PGL granules could not be detected during embryogenesis in
wild-type animals, indicating that maternally derived P granule
components that remain in somatic cells are disassembled
in early embryos (before the ?16-cell stage) and the diffuse
P granule components are quickly removed and do not accumu-
ence of diffuse GFP::PGL-1 in early-stage embryos (from the
2-cell stage onward) in sepa-1 mutants argues that sepa-1 is
Figure 5. SEPA-1 Is Colocalized with PGL Granules and Directly Interacts with PGL-3
(A–I) Confocal images of SEPA-1::RFP (A), SEPA-1::GFP (D), endogenous SEPA-1 (G), and PGL granules in lgg-1 animals, showing the colocalization of SEPA-1
aggregates and PGL-1-positive (C) or PGL-3-positive (F and I) granules. PGL granules are labeled by K76 (B) or anti-PGL-3 antibody (E and H). The PGL granules
in lgg-1 animals carrying the sepa-1::rfp or sepa-1::gfp transgene were sometimes larger than those in lgg-1 animals, possibly due to overexpression of
sepa-1::rfp or sepa-1::gfp.
(J)SEPA-1directlyinteractswith thePGL-3 fragments containing amino acids 581–614,asindicated by theblue color, inayeasttwo-hybrid X-gal assay. SEPA-1
does not interact with PGL-3 (614–694) or the C terminus of PGL-1 (701–772). PGL-3 (173–694) and PGL-3 (581–694) were identified from the yeast two-hybrid
reaction is shown as input. The input protein levels of MBP and MBP-fused SEPA-1, detected by Coomassie staining (Coom. Blue), and the GST-fused PGL-3
fragments were also shown. The amount of MBP and MBP-SEPA-1 in all experiments is the same. Asterisks indicate bands with the expected molecular mass.
(L) A SEPA-1 fragment spanning amino acids 39 to 160 interacts with PGL-3 in a pull-down assay, whereas other SEPA-1 fragments do not interact. GST-fused
PGL-3 (amino acids 581 to 614) immobilized on glutathione Sepharose beads was incubated with His-tagged SEPA-1 fragments. The retained proteins were
detected using anti-His antibody. Ten percent of the His fusion protein used for each pull-down reaction is shown as input. The two lower panels show the input
protein levels of GST, GST-PGL-3 (581–614), and His-tagged SEPA-1 fragments. The amount of GST and GST-PGL-3 (581–614) in all experiments is the same.
Asterisks indicate bands with the expected molecular mass.
(M) Endogenous SEPA-1 coimmunoprecipitates with PGL-3. PGL-3 was immunoprecipitated from lgg-1 embryo extracts with anti-PGL-3, and the precipitated
proteincomplex wasanalyzedbywestern blottingbyusinganti-SEPA-1,anti-PGL-3,andanti-actinantibody.Ineachgel,theinputlanecorrespondsto4%ofthe
embryo extract that wasusedinimmunoprecipitation.Preimmune serum wasused forcontrolIP.Actin wasnot coimmunoprecipitated with anti-PGL-3 antibody.
The ‘‘input’’ panels show western blots of the studied proteins in the extracts, indicating that equal amounts of starting material were used for each IP. SEPA-1
could not be immunoprecipitated by anti-PGL-3 antibody using pgl-3 mutant embryo extracts (Figure S10). IP, immunoprecipitation.
Cell 136, 308–321, January 23, 2009 ª2009 Elsevier Inc. 317
expressed before the 16-cell embryonic stage at a level too low
to be detected by our reporters yet still sufficient to mediate the
degradation of P granule components. These observations indi-
cate that diffuse P granule components are degraded by the
basal level of autophagy. In autophagy mutants, P granule
components and SEPA-1 accumulate into PGL granules. This
is consistent with the previous notion that the basal level of
autophagy plays an important role in removing diffuse aggre-
gate-prone proteins to prevent them from accumulating into
aggregates (Komatsu et al., 2006; Hara et al., 2006).
Figure 6. SEPA-1 Aggregates Are Colocalized with GFP::LGG-1 Dots
(A–C) Confocal images that show the colocalization of GFP::LGG-1 and SEPA-1::RFP (C). GFP::LGG-1 is diffusely distributed in the cytoplasm with some punc-
tate staining areas (A), whereas SEPA-1::RFP is mainly localized in aggregates (B).
(D) Coimmunoprecipitation assays reveal that SEPA-1 associates with LGG-1 in vivo. Extracts of embryos expressing lgg-1::gfp were used for immunoprecip-
itation with control or anti-GFP antibody, followed by western blot by using anti-SEPA-1, anti-GFP, and anti-actin antibody. The input lane in each gel
corresponds to 4% of the embryo extract used in immunoprecipitation. Anti-SOP-2 monoclonal antibody was used for control IP. The ‘‘input’’ panels show
western blots of the studied proteins in the extracts, indicating that equal amounts of starting material were used for each IP.
(E) Direct interaction between SEPA-1 and LGG-1 in a GST pull-down assay. GST-tagged full-length or truncated SEPA-1 proteins immobilized on glutathione
Sepharose beadswere incubated withHis-taggedLGG-1 orATG-18proteins.The retained proteins weredetectedby usinganti-His antibody.Tenpercentof the
His fusion protein used for each pull-down reaction is shown as input. The lower panels show the amount of GST-fused full-length and truncated SEPA-1 used in
the pull-down assay.
318 Cell 136, 308–321, January 23, 2009 ª2009 Elsevier Inc.
Selective Autophagic Degradation during Animal
Autophagy has been shown to selectively remove superfluous or
damaged organelles, including ER, peroxisomes, and mitochon-
dria, in certain stress or pathological conditions (van der Vaart
et al., 2008). Our study demonstrates that certain aggregate-
prone proteins are also selectively eliminated by autophagy
during animal development. One physiological function of selec-
tive autophagic degradation in C. elegans is to restrict germline
P granules to germ blastomeres during embryogenesis. Initially,
P granules are dispersed in the cytoplasm in oocytes. After fertil-
ization, the majority of P granules migrate toward the cytoplasm
that is inherited by the germline daughter cells (Hird et al., 1996).
P granules that remain in the cytoplasm destined for the somatic
1996). Proteins that transiently interact with P granules are
removed by the ubiquitin-proteasome system (DeRenzo et al.,
2003), whereas PGL-1 and PGL-3, which have a tendency to
aggregate, are degraded by autophagy. Degradation of these
nutrition for animal development and also prevent them from
forming aggregates, which may be toxic to the animal. Whether
plasm in Xenopus, remains to be investigated.
In addition to acting on maternal proteins stored in oocytes
and aggregate-prone polyubiquitinated proteins, autophagy
has been proposed to degrade cell-surface GABAa receptors
to control the balance of neuronal excitation and inhibition
(Rowland et al., 2006). Other SEPA-1 family members, which
are also preferential targets of autophagy, may function as
adaptor proteins in mediating selective autophagic degradation
in these processes. Therefore, autophagy, previously regarded
as a nonselective bulk degradation process, may have a more
general role in selectively removing different targets during
For RNAi experiments, single-stranded RNA was transcribed from T7- and
SP6-flanked PCR templates. ssRNAs were annealed and injected into
animals. The DNA templates used for RNA synthesis are included in the
Supplemental Experimental Procedures.
Permeabilization of embryos was performed by freeze-cracking methods.
Freeze-cracked slides were fixed, blocked, and incubated with diluted anti-
and incubated with Rhodamine-conjugated or FITC-conjugated secondary
antibody. Slides were viewed using an epifluorescence microscope or
a confocal microscope (Zeiss LSM 510 Meta plus Zeiss Axiovert zoom).
Figure 7. SEPA-1 Protein Family Members Are Preferential Substrates of Autophagy
Expression pattern of F44F1.6::gfp (A–D), T04D3.2::gfp (E–H), C35E7.6::gfp (I–L), and vet-6::gfp (M–P) in wild-type and lgg-1(RNAi) animals. Expression levels of
these reporters are dramatically increased in lgg-1(RNAi) animals (C, G, K, O) compared to wild-type. (B, D, F, H, J, L, N, and P) Nomarski images of the embryos
shown in (A), (C), (E), (G), (I), (K), (M), and (O), respectively.
Cell 136, 308–321, January 23, 2009 ª2009 Elsevier Inc. 319
Isolation of Mutants with PGL Granule Formation in Somatic Cells
gfp::pgl-1 animals were used to identify mutants with somatic PGL granules in
embryos. From ?10,000 genomes screened, 20 mutations were isolated.
Subsequent genetic and molecular analysis revealed that we had isolated
one allele of lgg-1, four alleles of atg-3, six alleles of atg-7, one allele of
atg-10, and one allele of atg-18. Progeny of atg-3(bp406), atg-7(bp413,
bp415, bp426, and bp419), and atg-10(bp421) homozygous mutant animals
were arrested at larval stages. The rest of the identified autophagy mutants
were viable and fertile. PGL granules were not observed in progeny born of
heterozygous mutants and only detected in progeny derived from homozy-
gous mutants. Thus, mutations in these autophagy genes caused maternal
effect formation of PGL granules in somatic cells.
Isolation, Mapping, and Cloning of sepa-1 and pgl-3
atg-18; gfp::pgl-1 and lgg-1; gfp::pgl-1 animals were used to isolate mutations
that suppressed formation of PGL granules. Approximately 7400 genomes
were screened for atg-18; gfp::pgl-1 animals, and six alleles of sepa-1 and
two alleles of pgl-3 were identified. About 2100 genomes were screened for
lgg-1; gfp::pgl-1 animals, and five alleles of sepa-1 and one allele of pgl-3
were obtained. sepa-1 was mapped close to the polymorphic marker
pkp1133. Fosmid WRM0637ac10 located in this region rescued the sepa-1
mutant phenotype. Details of the mapping and cloning of the mutants are in
Supplemental Experimental Procedures.
Reporters for sepa-1 and other sepa-1 family members were constructed by
a PCR fusion-based approach. The fused PCR products were derived from
two overlapping PCR fragments. One contained the promoter region and the
entire ORF. The other contained gfp and the unc-54 30UTR from pPD95.79.
The PCR products were co-injected with pRF4(rol-6[su1006]) into wild-type
animals, and at least two stable transgenic lines were analyzed for each
reporter. Details of the DNA sequences used for constructing sepa-1 family
genes are in the Supplemental Experimental Procedures.
Yeast Two-Hybrid Assay
Yeast two-hybrid experiments were performed with the ProQuest Two-Hybrid
System. Full-length SEPA-1 was fused to the GAL4 DNA binding domain in the
vector pPC97 and then transformed into the yeast host strain mav203 before
screening a C. elegans cDNA library cloned in pPC86. Clones that grew in
-Leu-Trp-His medium supplemented with 25 mM 3-aminotriazole were further
Fragments of PGL-3 (581–614 and 614–694) and PGL-1 (701–772) were
cloned into vector pPC86, and their interactions with SEPA-1 were tested in
a yeast two-hybrid X-gal assay.
In Vitro Pull-Down Assay
Constructs encoding full-length or truncated SEPA-1, LGG-1, PGL-1, PGL-3,
and ATG-18 were made by cloning the corresponding cDNA into pGEX-6p-1
(for GST fusion), pET-28a (for His tagging), or pMal-C2X (for MBP tagging).
GST-fusion or MBP-tagged proteins were incubated with His-tagged proteins
and 10 ml glutathione Sepharose beads (for GST fusion proteins) or Amylose
resin (for MBP-tagged proteins) in binding buffer (25 mM Tris.Cl [pH 7.6],
150 mM NaCl, 1 mM DTT, 0.5% Triton X-100, and 10% Glycerol) for 2 hr
at 4?C. The reactions were washed four times with 1 ml binding buffer. Bound
In Vivo Coimmunoprecipitation Assay
Extracts of embryos expressing gfp::lgg-1 or lgg-1 mutant embryos were
immunoprecipitated with anti-GFP monoclonal antibody (Roche) or anti-
PGL-3 antibody and then incubated with 30 ml protein G Sepharose beads.
After extensive washes, the immunoprecipitates were analyzed by western
blot by using an appropriate antibody.
To make antibodies, fragments of PGL-1 (95–551), PGL-3 (448–540), GLH-1
(137–572), or SEPA-1 (551–702) were cloned into the pET-28a vector,
expressed as His-tagged fusion proteins in E. coli BL21 and purified for use
as an immunogen in rabbits (for PGL-1 and SEPA-1) or rat (for PGL-3 and
GLH-1). The monoclonal antibodies OIC1D4 and K76 were obtained from
the Developmental Studies Hybridoma Bank.
Embryos were fixed in 4% formaldehyde and 0.15% glutaraldehyde in 0.1 M
PBS (pH 7.3). Samples were dehydrated and embedded in Lowicryl resin
HM20. Ultrathin sections (80 nm) were collected on formvar-coated nickel
mesh grids and labeled with anti-PGL-1 antibody K76 followed by anti-mouse
IgG gold conjugate (10 nm). Specimens were examined in a JEM-1230 (JEOL)
operating at 80 KV.
The GenBank accession number for the SEPA-1 sequence reported in this
paper is CAB07643.
Experimental Procedures, 1 table, and 12 figures and can be found with this
article online at: http://www.cell.com/supplemental/S0092-8674(08)01614-0.
We thank Drs. Susan Strome and Xiaodong Wang for their helpful comments
on the manuscript, Dr. David Hall for his advice on immunoEM, and Dr. Isabel
Hanson for editing the manuscript. This work was supported by the National
High Technology Projects 863 (2005AA210910).
Received: May 12, 2008
Revised: September 26, 2008
Accepted: December 9, 2008
Published: January 22, 2009
Bjorkoy, G., Lamark, T., Brech, A., Outzen, H., Perander, M., Overvatn, A.,
Stenmark, H., and Johansen, T. (2005). p62/SQSTM1 forms protein aggre-
gates degraded by autophagy and has a protective effect on huntingtin-
induced cell death. J. Cell Biol. 171, 603–614.
DeRenzo, C., Reese, K.J., and Seydoux, G. (2003). Exclusion of germ plasm
proteins from somatic lineages by cullin-dependent degradation. Nature
Hara, T., Nakamura, K., Matsui, M., Yamamoto, A., Nakahara, Y., Suzuki-
Migishima, R., Yokoyama, M., Mishima, K., Saito, I., Okano, H., et al. (2006).
Suppression of basal autophagy in neural cells causes neurodegenerative
disease in mice. Nature 441, 885–889.
Hird, S.N., Paulsen, J.E., and Strome, S. (1996). Segregation of germ granules
in living Caenorhabditis elegans embryos: Cell-type-specific mechanisms for
cytoplasmic localisation. Development 122, 1303–1312.
Kawasaki, I., Shim, Y.H., Kirchner, J., Kaminker, J., Wood, W.B., and Strome,
S. (1998). PGL-1, a predicted RNA-binding component of germ granules, is
essential for fertility in C. elegans. Cell 94, 635–645.
Kawasaki, I., Amiri, A., Fan, Y., Meyer, N., Dunkelbarger, S., Motohashi, T.,
Karashima, T., Bossinger, O., and Strome, S. (2004). The PGL family proteins
associate with germ granules and function redundantly in Caenorhabditis
elegans germline development. Genetics 167, 645–661.
Klionsky, D.J. (2005). The molecular machinery of autophagy: Unanswered
questions. J. Cell Sci. 118, 7–18.
Komatsu, M., Waguri, S., Chiba, T., Murata, S., Iwata, J., Tanida, I., Ueno, T.,
Koike, M.,Uchiyama, Y., Kominami,E., and Tanaka, K. (2006). Loss of autoph-
agy in the central nervous system causes neurodegeneration in mice. Nature
320 Cell 136, 308–321, January 23, 2009 ª2009 Elsevier Inc.
Komatsu, M., Waguri, S., Koike, M., Sou, Y.S., Ueno, T., Hara, T., Mizushima,
N., Iwata, J., Ezaki, J., Murata, S., et al. (2007). Homeostatic levels of p62
control cytoplasmic inclusion body formation in autophagy-deficient mice.
Cell 131, 1149–1163.
Levine, B., and Klionsky, D.J. (2004). Development by self-digestion: Molec-
ular mechanisms and biological functions of autophagy. Dev. Cell 6, 463–477.
Melendez, A., Talloczy, Z., Seaman, M., Eskelinen, E.L., Hall, D.H., and Levine,
B. (2003). Autophagy genes are essential for dauer development and life-span
extension in C. elegans. Science 301, 1387–1391.
Mizushima, N., Levine, B., Cuervo, A.M., and Klionsky, D.J. (2008). Autophagy
fights disease through cellular self-digestion. Nature 451, 1069–1075.
Nezis, I.P., Simonsen, A., Sagona, A.P., Finley, K., Gaumer, S., Contamine, D.,
Rusten, T.E., Stenmark, H., and Brech, A. (2008). Ref(2)P, the Drosophila mel-
anogaster homologue of mammalian p62, is required for the formation of
protein aggregates in adult brain. J. Cell Biol. 180, 1065–1071.
Pankiv, S., Clausen, T.H., Lamark, T., Brech, A., Bruun, J.A., Outzen, H., Øver-
vatn, A., Bjørkøy, G., and Johansen, T. (2007). p62/SQSTM1 binds directly to
Atg8/LC3 to facilitate degradation of ubiquitinated protein aggregates by
autophagy. J. Biol. Chem. 282, 24131–24145.
Rowland, A.M., Richmond, J.E., Olsen, J.G., Hall, D.H., and Bamber, B.A.
(2006). Presynaptic terminals independently regulate synaptic clustering and
autophagy of GABAA receptors in Caenorhabditis elegans. J. Neurosci. 26,
Rubinsztein, D.C. (2006). The roles of intracellular protein-degradation path-
ways in neurodegeneration. Nature 443, 780–786.
Shintani, T., Huang, W.P., Stromhaug, P.E., and Klionsky, D.J. (2002). Mecha-
nism of cargo selection in the cytoplasm to vacuole targeting pathway. Dev.
Cell 3, 825–837.
Strome, S., and Lehmann, R. (2007). Germ versus soma decisions: Lessons
from flies and worms. Science 316, 392–393.
Suzuki, K., and Ohsumi, Y. (2007). Molecular machinery of autophagosome
formation in yeast, Saccharomyces cerevisiae. FEBS Lett. 581, 2156–2161.
van der Vaart, A., Mari, M., and Reggiori, F. (2008). A picky eater: Exploring the
mechanisms of selective autophagy in human pathologies. Traffic 9, 281–289.
Cell 136, 308–321, January 23, 2009 ª2009 Elsevier Inc. 321
Cell 136, January 23, 2009 ©2009 Elsevier Inc. 207
During embryonic development, cells
are specified to become either nonre-
productive somatic cells or germ cells
that are capable of producing the next
generation. Germline determinants in the
cytoplasm (germplasm) composed of
protein and RNA accumulate in develop-
ing germ cells but are absent in somatic
cells during early animal development
(Strome and Lehmann, 2007) (Figure 1).
This restriction of germplasm to germ
cells is partly accomplished by differ-
ential localization of this material in the
developing animal. In addition, studies
in the nematode Caenorhabditis elegans
have shown that the loss of germplasm
proteins in somatic cells is regulated by
ubiquitin conjugation factors (DeRenzo
et al., 2003), indicating that the ubiq-
uitin-proteasome system is involved in
degrading germplasm in somatic cells.
In this issue, Zhang et al. (2009) provide
evidence that autophagy directed by the
SEPA-1 protein is also required for the
clearance of germplasm proteins from
somatic cells in the developing nema-
Autophagy or “self-eating” is a cata-
bolic process that delivers cytoplasmic
materials, including proteins and organ-
elles, to the lysosomes for degradation.
Whereas the ubiquitin-proteasome sys-
tem regulates the degradation of many
short-lived proteins, autophagy is used to
degrade long-lived cellular structures and
proteins. Three forms of autophagy exist:
autophagy, and micro-autophagy. Of
these processes, macro-autophagy is the
most studied. During macro-autophagy
(hereafter referred to as autophagy), cyto-
plasmic cargo is enclosed in double-mem-
braned vesicles called autophagosomes
that are subsequently delivered to the
lysosome for degradation by hydrolases
(Mizushima, 2007) (Figure 1). Pioneering
studies in yeast identified autophagy (atg)
genes that encode proteins required for
this process. Many of these atg genes
have conserved functions in higher ani-
mals (Mizushima, 2007) where autophagy
has been implicated in many processes,
including the clearance of protein aggre-
gates (Mizushima et al., 2008). Animals
lacking atg gene function accumulate
ubiquitin-positive inclusions in the brain,
have shortened life spans, and develop
2006). Studies of patients with protein
aggregation disorders and animal models
of these disorders reveal that autophagy
can protect cells from these toxic protein
aggregates and can regulate the clear-
ance of abnormal protein aggregates in
the aging nervous system. Now Zhang
and colleagues provide evidence that
autophagy can regulate the clearance of
protein aggregates that are involved in the
specification of germ cell fate during nor-
mal embryonic development. Their results
suggest that autophagy may selectively
target germplasm proteins for degrada-
tion by lysosomes (Figure 1).
In C. elegans, the germplasm contains
P granules, a specialized type of RNA and
protein aggregate that regulates germ cell
fate. To identify genes that are required for
Autophagy SEPArates Germline
and Somatic Cells
Eric H. Baehrecke1,*
1Department of Cancer Biology, University of Massachusetts Medical School, Worcester, MA 01605, USA
Cellular determinants of the germline selectively accumulate in germ cell precursors and influ-
ence cell fate during early development in many organisms. Zhang et al. (2009) now report that
targeted autophagy mediated by the SEPA-1 protein depletes germplasm proteins from somatic
cells during early development of the nematode.
Figure 1. Autophagy Regulates Clearance of Germplasm Proteins
Germplasm proteins (red) are widely distributed in the C. elegans oocyte but become restricted to the germ
cells (green) early during embryogenesis. Germplasm proteins are selectively degraded by autophagy in
somatic cells (blue). The SEPA-1 protein (purple) is required for targeting of germplasm proteins to the
autophagosome for degradation by the autolysosome. Mutant worm embryos defective in autophagy fail to
degrade germplasm proteins and thus retain these proteins in both somatic cells and germ cells.
208 Cell 136, January 23, 2009 ©2009 Elsevier Inc. Download full-text
the selective loss of P granule proteins in
somatic cells of the worm embryo, Zhang
et al. performed a genome-wide RNA
interference (RNAi) screen. Surprisingly,
they find that the autophagy gene lgg-1
(atg8 in yeast) is required for the clearance
of P granule proteins in these somatic
cells. Strikingly, disruption of other worm
orthologs of yeast autophagy genes (atg1,
atg3, atg4, atg7, atg10, atg12, and atg18)
also results in the loss of P granule protein
degradation in somatic cells. The persistent
P granule proteins in somatic cells of lgg-1
worm embryos defective in autophagy
appear to be maternally derived. These
data suggest that maternal P granule pro-
teins are cleared by autophagy as a normal
part of development, and that their accu-
mulation is not a nonphysiological event
caused by a defect in protein clearance.
How does the bulk degradation pro-
cess of autophagy specifically target P
granule components in somatic cells but
not in germ cells? To investigate this ques-
tion, Zhang and colleagues conducted a
genetic screen to identify mutations that
suppress the accumulation of P granule
components in somatic cells of autophagy
mutants. They find that mutations in the
sepa-1 (suppressor of ectopic P granule in
autophagy mutants 1) gene suppress the
formation of P granule protein aggregates
in the somatic cells of autophagy mutants
without affecting P granules in germline
cells. In addition, there is no decrease in
P granule proteins in late-stage sepa-1
mutant worm embryos, indicating that the
protein encoded by sepa-1 is required for
degradation of these factors. The SEPA-1
protein is rich in helical domains and also
contains a KIX protein interaction domain.
Biochemical studies by Zhang et al. indi-
cate that SEPA-1 self-associates. Immu-
nofluorescence staining reveals that
aggregates of SEPA-1 accumulate spe-
cifically in the somatic cells of developing
embryos and that this staining disappears
at late stages of embryogenesis.
These data raise the possibility that
SEPA-1 is associated with and regulates
autophagy. The authors find that clear-
ance of SEPA-1 in somatic cells requires
the autophagy genes lgg-1, atg3, and atg7.
SEPA-1 aggregates colocalize and directly
interact with LGG-1. In support of a role
for SEPA-1 in targeting P granule compo-
nents to autophagosomes, Zhang and col-
leagues observe that SEPA-1 aggregates
also colocalize with P granule proteins and
that the SEPA-1 protein physically interacts
with the P granule protein PGL-3. PGL-3
is required for the accumulation of the P
granule protein PGL-1 in somatic cells of
autophagy-deficient embryos. Interest-
ingly, the formation of SEPA-1 aggregates
does not require either PGL-1 or PGL-3.
Therefore, it appears that SEPA-1 medi-
ates the recruitment of P granule compo-
nents to autophagosomes by aggregating
and physically interacting with both LGG-1
and specific P granule proteins (Figure 1).
The sequestration and degradation of
cytoplasmic components by autophagy
during embryogenesis present interest-
ing questions about the regulation of
autophagy, germplasm protein clearance,
and development. Worms lacking proteins
encoded by atg3, atg7, and atg8 develop
normally; thus, the persistence of certain
P granule components in somatic cells is
clearly not sufficient to transform the fate
of these cells to that of germ cells. These
observations are consistent with the com-
plementary involvement of the ubiquitin-
proteasome system in the degradation of
P granule components (DeRenzo et al.,
2003). Interplay between the ubiquitin-
proteasome system and autophagy has
been observed in the context of protein
aggregation disorders (Pandey et al.,
2007), suggesting that autophagy may
enable the degradation of complexes that
impair the ubiquitin-proteasome system.
However, the degradation of SEPA-1 and
the P granule component PGL-1 in C. ele-
gans is unaffected when the proteasome
is impaired, indicating that in this case,
autophagy degrades P granule compo-
nents in a manner independent of protea-
some function. These data indicate that
specific P granule-associated proteins are
degraded by independent catabolic mech-
anisms. Additional studies are needed to
understand why both autophagy and the
ubiquitin-proteasome system are required
to degrade these factors.
Autophagy is considered to be a bulk
degradation process that is attenuated in
growing and dividing cells (Eskelinen et
al., 2002). The rapid cell division and new
membrane formation that occur during
early C. elegans embryonic development
would likely be in conflict with the nonspe-
cific bulk degradation of cytoplasmic com-
ponents mediated by autophagy. Thus,
autophagy of P granule proteins in the
nematode embryo may require a specific
autophagosome targeting mechanism,
supporting the notion that targeted deg-
radation of specific proteins by autophagy
is a more prevalent phenomenon than
was previously thought (van der Vaart et
al., 2008). Significantly, SEPA-1 appears to
be functionally similar to the mammalian
p62 protein in mediating the recognition
of protein aggregates by the autophagic
machinery (Komatsu et al., 2007). Zhang
and colleagues further describe multiple
SEPA-1-related proteins in C. elegans
that appear to be targets of autophagy,
thus suggesting the possibility that this
protein family plays a broad role in the
recruitment of cargo to autophagosomes.
Notably, mutations in sepa-1 do not affect
other autophagy-associated processes
in C. elegans, including the clearance of
polyglutamine protein aggregates. Future
studies are required to determine the
importance of SEPA-1-related proteins
in the regulation of autophagy in other C.
elegans cell types, and whether the role
for autophagy in eliminating germplasm
determinants from somatic cells is con-
served in other organisms.
I thank the NIH (GM059136 and GM079431) for
DeRenzo, C., Reese, K.J., and Seydoux, G. (2003).
Nature 424, 685–689.
Eskelinen, E.L., Prescott, A.R., Cooper, J., Brach-
mann, S.M., Wang, L., Tang, X., Backer, J.M., and
Lucocq, J.M. (2002). Traffic 3, 878–893.
Komatsu, M., Waguri, S., Koike, M., Sou, Y.S.,
Ueno, T., Hara, T., Mizushima, N., Iwata, J., Ezaki,
J., Murata, S., et al. (2007). Cell 131, 1149–1163.
Mizushima, N. (2007). Genes Dev. 21, 2861–2873.
Mizushima, N., Levine, B., Cuervo, A.M., and
Klionsky, D.J. (2008). Nature 451, 1069–1075.
Pandey, U.B., Nie, Z., Batlevi, Y., McCray, B.A.,
Ritson, G.P., Nedelsky, N.B., Schwartz, S.L.,
DiProspero, N.A., Knight, M.A., Schuldiner, O., et
al. (2007). Nature 447, 859–863.
Rubinsztein, D.C. (2006). Nature 443, 780–786.
Strome, S., and Lehmann, R. (2007). Science 316,
van der Vaart, A., Mari, M., and Reggiori, F. (2008).
Traffic 9, 281–289.
Zhang, Y., Yan, L., Zhou, Z., Yang, P., Tian, E.,
Zhang, K., Zhao, Y., Li, Z., Song, B., Han, J., et al.
(2009). Cell, this issue.