The conserved kinetochore protein
shugoshin protects centromeric
cohesion during meiosis
Tomoya S. Kitajima1
, Shigehiro A. Kawashima1
& Yoshinori Watanabe1,2
1
Department of Biophysics and Biochemistry, Graduate School of Science, University of Tokyo, and 2
SORST, Japan Science and Technology Agency, Hongo,
Tokyo 113-0033, Japan
...........................................................................................................................................................................................................................
Meiosis comprises a pair of specialized nuclear divisions that produce haploid germ cells. To accomplish this, sister chromatids
must segregate together during the first meiotic division (meiosis I), which requires that sister chromatid cohesion persists at
centromeres. The factors that protect centromeric cohesion during meiosis I have remained elusive. Here we identify Sgo1
(shugoshin), a protector of the centromeric cohesin Rec8 in fission yeast. We also identify a homologue of Sgo1 in budding yeast.
We provide evidence that shugoshin is widely conserved among eukaryotes. Moreover, we identify Sgo2, a paralogue of shugoshin
in fission yeast, which is required for faithful mitotic chromosome segregation. Localization of Sgo1 and Sgo2 at centromeres
requires the kinase Bub1, identifying shugoshin as a crucial target for the kinetochore function of Bub1. These findings provide
insights into the evolution of meiosis and kinetochore regulation during mitosis and meiosis.
In eukaryotes, sister chromatid cohesion is established during S
phase and is maintained throughout G2 until the M phase. During
mitosis, this cohesion is destroyed along the entire length of the
chromosome, allowing sister chromatids to segregate to opposite
sides of the cell (equational division), and ensuring that each
daughter cell receives one copy of each chromosome. In contrast,
meiosis consists of two rounds of chromosome segregation following
a single round of DNA replication, leading to the formation of
four haploid gametes from a diploid germ cell. During meiosis I,
homologous chromosomes (homologues) pair up in order to
recombine, forming chiasmata in which one sister chromatid
from one homologue is covalently attached to a sister chromatid
from the other homologue. Hence, for homologues to segregate at
meiosis I, sister chromatid cohesion must be released along the
chromosome arms to resolve chiasmata. However, sister chromatid
cohesion is retained at the centromeres until meiosis II, when sister
chromatids segregate as they do in mitosis, using the residual
centromeric cohesion. Thus, meiotic divisions require sister chromatid
cohesion to be released in two steps, yet the molecular basis
for protection of centromeric cohesion only during meiosis I and
only at the centromeres has remained unknown1
.
There are clues to the molecular nature of sister chromatid
cohesion and the mechanism by which it is released at the onset
of anaphase1–5
. In various eukaryotes, sister chromatid cohesion
depends on a multisubunit cohesin complex including Scc1
(Rad21 in the fission yeast Schizosaccharomyces pombe). Anaphasepromoting
complex (APC)-dependent degradation of the securin
Pds1 (Cut2 in S. pombe) allows release of the Esp1 (Cut1 in S. pombe)
endopeptidase (separase), which in turn cleaves Scc1, releasing
sister chromatid cohesion. During meiosis, the cohesin subunit
Scc1 is replaced by a meiotic counterpart, Rec8 (refs 6–10). As Rec8
complexes reside only at centromeres after meiosis I and depletion
of Rec8 disrupts centromeric cohesion, its presence at centromeres
has been thought to confer the persistence of cohesion throughout
meiosis I (ref. 11). Several lines of evidence12,13
suggest that Rec8
along chromosome arms is cleaved by separase at anaphase I,
whereas centromeric Rec8 is specifically protected until metaphase
II. Budding yeast Spo13 has been implicated in the protection of
centromeric Rec8 (refs 14, 15), but Spo13 is not centromeric and
may function indirectly. Drosophila MEI-S332, a protein that resides
at pericentromeric regions16
and is required for the persistence of
centromeric cohesion during meiosis I (ref. 17), has features of a
candidate protector of meiotic centromeric cohesion, although the
details of such protection have so far not been revealed4
. Despite
the completion of genome sequencing projects in several organisms,
no homologues of Spo13 or MEI-S332 have emerged, preventing
the formulation of a generalized view of protection. Concurrently,
studies in fission yeast18
have illuminated the importance of pericentromeric
heterochromatin for recruiting centromeric Rec8
complexes and ensuring centromeric cohesion during meiosis I.
However, pericentromeric heterochromatin cannot alone confer the
specific protection of Rec8 at meiosis I compared with meiosis II.
We now identify a meiosis-specific protein, Sgo1 (shugoshin,
Japanese for ‘guardian spirit’), that protects centromeric Rec8
from degradation during meiosis I.
Identification of Sgo1 in fission yeast
The replacement of the mitotic cohesin Rad21 with the meiotic
version Rec8 is a prerequisite for protecting centromeric sister
chromatid cohesion through anaphase of meiosis I (refs 19, 20).
However, when we expressed Rec8 ectopically during mitosis, Rec8
localized largely at centromeres but disappeared at anaphase, with
sister chromatids segregating to opposite sides of the cell (Fig. 1c, d).
Moreover, the ectopic expression of non-cleavable Rec8 during
mitosis (note that Rec8 is cleaved by separase Cut1 during meiosis13
)
resulted in an inability of sister chromatids to separate (Supplementary
Fig. 1). Thus, in contrast to the situation during meiosis I,
centromeric Rec8 is cleaved by separase during mitosis, resulting in
sister chromatid separation. These observations prompted us to
postulate a meiosis-I-specific centromeric protector of Rec8. To
identify this factor, we searched for a gene that yields toxicity during
mitotic growth only when co-expressed with Rec8. This screen
identified a novel gene called sgo1þ
(open reading frame (ORF)
SPBP35G2.03C). The hindrance of growth by Sgo1 was indeed
dependent on Rec8, as Sgo1 had little effect on growth when coexpressed
with Rad21 (Fig. 1a). Co-expression of rec8þ
and sgo1þ
frequently led to blocked nuclear division, as centromere-associated
green fluorescent protein markers (cen2–GFP)21
frequently segregated
to the same side of a septated cell (Fig. 1b, c). To test the
possibility that Sgo1 protects Rec8 from degradation at anaphase,
articles
NATURE | VOL 427 | 5 FEBRUARY 2004 | www.nature.com/nature510 ©2004 NaturePublishing Group
we examined the localization of Rec8 in the context of Sgo1
expression. We found that the Rec8 localization at centromeres
persisted through anaphase only when Sgo1 was co-expressed
(Fig. 1d). As Sgo1 is expressed exclusively in meiosis (DNA microarray
data22
; see below), the foregoing results allowed us to postulate
that Sgo1 is a protector of Rec8 during meiosis.
Sgo1 protects centromeric cohesion at meiosis I
To examine whether Sgo1 is indeed required for the protection of
Rec8 during meiosis, we deleted the entire ORF encoding sgo1þ
and
examined the phenotype. sgo1-depleted cells (sgo1D) were viable
and showed normal vegetative growth. To examine meiotic
chromosome segregation, we marked centromere-linked sequences
with GFP (cen2–GFP) on only one of the two homologues in a
zygote and monitored the segregation of the GFP dots during
meiosis. In normal meiosis, monopolar attachment of sister kinetochores
to the spindle is established in metaphase I; therefore, sisters
move together to the same side of the zygote (reductional division)
in the following anaphase I (Fig. 2a). Thanks to the centromeric
cohesion preserved throughout anaphase I, bipolar attachment is
secured at meiosis II, thus leading to faithful disjunction (Fig. 2a).
We found that meiosis I appeared normal in sgo1D cells, as sister
chromatid pairs generally moved together to the same side of each
zygote (Fig. 2b). Hence, monopolar attachment was intact in this
mutant. Moreover, by marking cen2–GFP on both chromosomes,
we determined that homologues underwent faithful disjunction at
meiosis I (data not shown). At meiosis II in sgo1D cells, however,
sister chromatids failed to segregate properly, undergoing nondisjunction
in approximately 50% of cells (Fig. 2b). This value is
consistent with random chromosome segregation at meiosis II.
To examine centromeric cohesion, we monitored cen2–GFP
marked on both homologues in zygotes arrested before meiosis II,
the stage at which centromeric cohesion is normally preserved in
sgo1þ
cells. We found that sgo1D cells frequently displayed precocious
centromeric dissociation as split cen2–GFP signals prevailed
in the dyad nuclei (Fig. 2c). This result may explain why the second
meiotic division is random in sgo1D cells, because cohesion is
required for sister kinetochores to be properly recognized by spindle
microtubules extending from opposite poles. Next we examined
whether protection of Rec8 at centromeres is dependent on Sgo1 by
monitoring Rec8–GFP at late anaphase I and prometaphase II.
Notably, although Rec8 signals were centromeric in wild-type cells,
the Rec8 signals had largely disappeared from the centromeres at
these stages in sgo1D cells (Fig. 2d). All phenotypes of sgo1D cells are
reminiscent of heterochromatin-deficient S. pombe, in which Rec8
localization to the pericentromeric regions is decreased and centromeric
cohesion is lost during meiosis I, leading to random
division at meiosis II (ref. 18). We then examined chromatin
binding by Rec8 in cells arrested before meiosis I using a chromatin
immunoprecipitation (ChIP) assay. In marked contrast to heterochromatin-deficient
cells, Rec8 localization was intact in sgo1D cells
at the pericentromeric regions as well as all other regions tested
(Fig. 2e). As monopolar attachment requires centromeric Rec8
(refs 20, 23), the proper location of Rec8 before meiosis I is
consistent with the fact that monopolar attachment is intact in
sgo1D cells. More importantly, these results indicate that the loss of
centromeric Rec8 after meiosis I is brought about not by an initial
defect in Rec8 localization to centromeres but rather by a defect in
the preservation of centromeric Rec8 during anaphase I. Previous
results suggested that the Cut1 separase becomes active at the onset
of anaphase I and cleaves most chromosomal Rec8, leaving only
centromeric Rec8 intact13
. Our current results advocate that Sgo1
has an essential role in protecting centromeric cohesion at anaphase
I by safeguarding cohesin Rec8 from separase cleavage.
Sgo1 localizes at centromeres during meiosis I
To detect the Sgo1 protein, we raised antibodies specific for Sgo1.
Western blotting indicated that Sgo1 is expressed only around
meiosis I (Fig. 3a). Immunofluorescence microscopy on cells at
various stages of meiosis revealed that Sgo1 appears at late prophase
of meiosis I and is fully localized as several punctate dots until
metaphase I (Fig. 3b). These dots localize closely with the Mis6
kinetochore protein24
, indicating that Sgo1 is associated with
centromeric regions (Fig. 3c). At the onset of anaphase I, Sgo1
signals decrease markedly. We found that Sgo1 remains undegraded
at centromeres in APC-depleted cells arrested at metaphase I, but
undergoes normal degradation in separase-defective cells (Supplementary
Fig. 2), suggesting that Sgo1 degradation at anaphase I
is regulated more directly by the APC rather than through separase.
Although we detected residual Sgo1 signals at the centromeres in
early anaphase I, they disappeared completely by the end of
anaphase I (Fig. 3b). This suggests that a substantial amount of
Sgo1 is required at the onset of anaphase I when separase is fully
activated. As anaphase I progresses, however, smaller and smaller
amounts of Sgo1 would be required. This idea is tenable if the
separase activity is quickly downregulated or prevented from
accessing chromosomes during anaphase I. Sgo1 does not reappear
Figure 1 Co-expression of Sgo1 and Rec8 causes failure of sister chromatid separation
during mitosis. a, The haploid cen2–GFP strains expressing the indicated genes by
exogenous promoters (a constitutive promoter Padh1 for rad21 þ
or rec8 þ
, and a
thiamine-repressible promoter Pnmt1 for sgo1 þ
) were streaked on a thiamine-depleted
plate. b, Examples of Padh1–rec8 þ
Pnmt1–sgo1 þ
cells cultured at 30 8C for 15 h after
thiamine depletion. Note the non-disjunction of cen2–GFP in the septated cells (asterisk).
c, The frequency of non-disjunction was counted among septated cells (n . 100).
d, Padh1–rec8 þ
–GFP strains with or without Pnmt1–sgo1 þ
were cultured as in b, fixed
with formaldehyde and stained with DAPI and antibodies against GFP and tubulin.
Examples of each strain at interphase (left cell) and anaphase (right cell) are shown. Note
that the Rec8 signal persists during anaphase mostly in sgo1 þ
-expressing cells (84%,
n ¼ 129) and only slightly in non-expressing cells (9%, n ¼ 129).
articles
NATURE | VOL 427 | 5 FEBRUARY 2004 | www.nature.com/nature 511©2004 NaturePublishing Group
during meiosis II (Fig. 3b), consistent with the idea that Sgo1 is only
required for the protection of Rec8 during meiosis I.
We have suggested previously that Rec8 localization at pericentromeric
regions is especially important for the persistence of
centromeric cohesion throughout meiosis I (ref. 18). If Sgo1 is a
centromeric protector of Rec8, then it might be expected to localize
there as well. To test this possibility, we delineated Sgo1 localization
more precisely using a ChIP assay. Sgo1 does indeed associate with
pericentromeric heterochromatin regions rather than central core
regions along the centromere sequences (Fig. 3d). As immunoprecipitation
experiments indicated that Sgo1 interacts with Rec8
complexes in vivo (Fig. 3f), the protection is probably carried out
through close interaction. Together, our results indicate that Sgo1
resides at pericentromeric regions and operates to protect Rec8
from cleavage by separase at anaphase I (Fig. 3g).
Sgo2 is a mitotic Sgo1 paralogue in fission yeast
By means of a conventional BLASTsearch of genome databases, we
identified Sgo1-like proteins from Saccharomyces cerevisiae and
Neurospora crassa, suggesting that Sgo1 is a conserved protein (see
below). In the same search, we identified an S. pombe Sgo1
paralogue, which we call Sgo2 (ORF SPAC15A10.15). We disrupted
the sgo2þ
gene and found that sgo2D cells are viable but show
sensitivity to the spindle-destabilizing drug thiabendazole (TBZ)
(Fig. 4a). Consistently, sgo2D cells display an increased incidence of
chromosome mis-segregation at mitosis (Fig. 4c). These mitotic
phenotypes are remarkable, as sgo1D cells never show such a defect
(Fig. 4a). To investigate its cellular distribution, the endogenous
sgo2þ
gene was tagged with GFP. In proliferating cells, Sgo2–GFP
was observed as two or three dots within the nucleus (Fig. 4e);
however, it localized closely with the centromere protein Mis6 at
Figure 2 Sgo1 is required to protect Rec8 and thereby cohesion at centromeres during
anaphase of meiosis I. a, Schematic drawing of the behaviour of homologous
chromosomes (white and grey) in normal meiosis I and II. Expression of cen2–GFP (green
oval) is marked on one of the homologous chromosomes and the location of the cohesin
Rec8 (red oval) is indicated. b, One of the homologues marked with cen2–GFP was
monitored for segregation during meiosis in wild-type (WT) and sgo1D cells (n . 170).
Examples of sgo1D cells are shown (bottom right panel). c, Cells of both homologues
marked with cen2–GFP were arrested before meiosis II by introducing the mes1D
mutation, and were examined for cen2–GFP dots. d, The Rec8–GFP signal was monitored
at late anaphase I (n . 30) and at prometaphase II (n . 100) in the indicated cells, and
the frequency of the cells displaying centromeric Rec8–GFP was counted. The spindles
were visualized by expressing cyan fluorescent protein (CFP)–Atb2 (a2-tubulin)43
.
e, A ChIP assay with anti-GFP antibodies was used to measure Rec8–GFP levels
throughout the indicated chromosome sites in the arrested cells before meiosis I (mei4D
arrest). The bottom panel shows a schematic representation of S. pombe chromosome I
as well as the primers used (cnt, imr, dg, dh, lys1, mes1).
articles
NATURE | VOL 427 | 5 FEBRUARY 2004 | www.nature.com/nature512 ©2004 NaturePublishing Group
metaphase and disappeared during anaphase (Fig. 4d, e). ChIP
assays showed that Sgo2 chromatin association was detectable only
in synchronous populations of mitotic cells, in which Sgo2 localized
to the pericentromeric regions (Fig. 4f). Reinforcing this localization,
sgo2 deletion confers a marked defect in chromosome segregation
if combined with the heterochromatin-deficient swi6D
mutation, which by itself slightly impairs kinetochore function25
(Fig. 4b, c). These results suggest that Sgo2 cooperates with
pericentromeric heterochromatin factors to ensure chromosome
segregation at mitosis. Moreover, we found that Sgo2 persists
throughout meiosis (Supplementary Fig. 4) and that sgo2D cells
have a modest increase in non-disjunction of homologues at
meiosis I (about 15%), suggesting that Sgo2 is also important for
promoting proper meiosis I. Notably, the role of Sgo2 in meiosis
does not overlap with that of Sgo1, as sgo1D neither yields an
apparent defect at meiosis I (Fig. 2b) nor enhances the defect of
sgo2D (data not shown).
Shugoshin location controlled by Bub1
A conserved centromere-associated kinase, Bub1, is thought to have
a function in protecting Rec8 during meiosis, as centromeric Rec8
cannot be detected after meiosis I in fission yeast bub1 mutants26
(Fig. 2d). Although bub1 mutation has pleiotropic effects in meiotic
chromosome segregation26
, we thought that Sgo1 function might
be targeted by Bub1 activity. To address this issue, we examined
Sgo1–GFP signals in bub1D cells undergoing meiosis. Notably,
bub1D cells were almost completely devoid of punctate centromeric
Sgo1–GFP signals, showing instead a diffuse fluorescence within the
nucleus (Fig. 3e). Identical results were obtained using the
Bub1(K762R) point mutation (not shown), which abolishes the
kinase activity27
. As substantial levels of Sgo1 protein were detected
in meiotic bub1D cells by western blot analysis (Fig. 3e), Bub1
primarily regulates localization rather than stability of the Sgo1
protein. Therefore, the observed defects in centromeric protection
in bub1D cells might be explained by an impaired location of Sgo1.
Moreover, these results suggest that Bub1 might regulate the timing
of Sgo1 loading to and the disappearance from centromeres, as
Bub1 localizes at centromeres during a similar period in meiosis I
(ref. 26).
In parallel experiments, we found that mitotic Sgo2 localization
at centromeres was similarly disrupted in bub1 mutants (Fig. 4d),
whereas protein levels of Sgo2 were unchanged (Supplementary
Fig. 3c). It has been suggested that loss of Bub1 function leads
to a weakness in kinetochore function28
. In this regard, the
Bub1(K762R) mutation shows a synthetic defect in growth with
swi6D, a mutation that also impairs slightly kinetochore function
through its role in pericentric heterochromatin formation. We
found that sgo2D similarly showed a synthetic defect in growth
with swi6D (Fig. 4b), exhibiting severe mis-segregation of chromosomes
at mitosis (Fig. 4c). As the sgo2D bub1D double mutant
showed no cumulative defects in growth or in TBZ sensitivity
(Fig. 4a), these genetic analyses confirm that Sgo2 and Bub1
function in tandem to ensure chromosome segregation in mitosis.
Taken together, these results reveal that Sgo1 and Sgo2 localization
at centromeres is a crucial function of Bub1 kinase in meiosis and
mitosis, respectively.
Figure 3 Sgo1 localizes at pericentromeric regions during meiosis I. a, Synchronous
meiosis of diploid pat1-114/pat1-114 cells13
was sampled. Meiotic nuclear division was
monitored by DAPI staining, and the protein level of Sgo1 was detected by western
blotting using anti-Sgo1 antibodies. Circle, one nucleus; square, two nuclei; triangle, 3–4
nuclei. b, Sgo1 (green) was counterstained with tubulin (red) and DAPI (blue) in the meiotic
cell at the indicated stages. c, An sgo1 þ
–GFP cell co-expressing mis6 þ
–CFP was
examined under fluorescence microscopy. Sgo1–GFP (green) and Mis6–CFP (red) are
merged. d, A ChIP assay with anti-GFP antibodies was used to measure Sgo1–GFP levels
throughout the indicated chromosome sites in cells arrested at metaphase I. We used the
same primers as for Fig. 2e together with additional primers at mat (heterochromatin
region at the mating-type locus) and TAS (telomere-associated sequence). e, Sgo1–GFP
(green) was detected at metaphase I in the indicated cells expressing CFP–Atb2 to
visualize spindles (red). Western blot analysis of Sgo1–GFP was carried out on the
indicated strains arrested at metaphase I. The asterisk indicates a nonspecific band.
f, Rec8–HA was expressed with or without Sgo1–Flag in proliferating cells and the
extracts were immunoprecipitated with anti-Flag antibody. g, A model for the action of
shugoshin in meiosis. Shugoshin localizes at centromeres in meiosis I depending on Bub1
kinase, and it protects centromeric Rec8 complexes from cleavage by separase at the
onset of anaphase I, thereby preserving centromeric cohesion until meiosis II. Shugoshin
is degraded depending on APC during anaphase I.
articles
NATURE | VOL 427 | 5 FEBRUARY 2004 | www.nature.com/nature 513©2004 NaturePublishing Group
Characterization of a Sgo1 homologue in budding yeast
We identified a single Sgo1 homologue in S. cerevisiae, ScSgo1 (ORF
YOR073W), which has so far not been analysed. We examined the
cellular localization of ScSgo1 by tagging endogenous ScSGO1 with
GFP, and found that the pattern of ScSgo1 localization closely
resembles that of S. pombe Sgo2 in mitosis and Sgo1 in meiosis
(Supplementary Fig. 5). To examine the function of ScSgo1, we
disrupted the ScSGO1 gene. Scsgo1D cells were viable but grew
slowly and showed sensitivity to the spindle-destabilizing drug
benomyl (Fig. 5a), suggesting that kinetochore function might be
impaired. We used a colony-sectoring assay to compare the rates of
chromosome loss in Scsgo1D cells with those in wild-type cells.
Whereas less than 2% of wild-type colonies contained red sectors
(which indicate chromosome loss), approximately 40% of the
Scsgo1D colonies contained such sectors (Fig. 5b). We conclude
that ScSgo1 has a crucial role at kinetochores for ensuring mitotic
chromosome segregation. Scsgo1D cells showed significant defects
in the initiation of meiosis, as many cells arrested with a single
nucleus in the meiotic condition. Among the leaked tetranucleate
products of meiosis, however, the distribution pattern of cenV–GFP
was consistent with proper segregation at meiosis I but random
segregation at meiosis II (Fig. 5c). We also found that tagging
chromosomal ScSGO1 with a 13-Myc tag at its carboxy terminus,
which by itself yields no detectable defects in mitotic growth or
meiosis I, resulted in impaired segregation at meiosis II (34% nondisjunction
indicating 68% random segregation) (Fig. 5d). Moreover,
the Myc-tagged ScSgo1 cells showed frequent separation of
sister centromeres at late meiotic anaphase I (Fig. 5e), indicating
that centromeric cohesion was not properly protected. Together,
these results support the idea that ScSgo1 has a crucial role in
protecting centromeric cohesion throughout meiosis I, thereby
ensuring normal progression to meiosis II, as does fission yeast
Sgo1.
Conservation of shugoshin among eukaryotes
Our BLAST searches identified only three Sgo1-like proteins, all in
fungi: S. pombe Sgo2, S. cerevisiae ScSgo1 and N. crassa B23G1.060.
As we found two conserved regions among these proteins, we used
the Block Maker and MAST programs29,30
to search for related
proteins under conditions of two-block sequences. This approach
yielded several candidate proteins from various eukaryotes including
fly, nematode, plant, mouse and human (Fig. 6). Notably, the list
included Drosophila MEI-S332, a previously characterized protein
essential for preserving centromeric cohesion in meiosis17
, although
the similarity score is marginal (E-value ¼ 10). All other proteins in
the list show a short stretch of similarity in the C-terminal basic
regions, whereas the primary sequences in the first block are not
conserved except that they all contain a putative coiled-coil. The
space and sequences between these two blocks diverge among the
proteins. As these blocks were previously identified to be important
for MEI-S332 function31
, we investigated the significance of the
conserved regions in Sgo1. We changed several amino acids individually
within these similarity blocks to alanines and investigated the
function of the mutant proteins in vivo (Supplementary Fig. 6). We
found that three conserved amino acids known to be critical for
MEI-S332 function31
were also required for Sgo1 function (N13,
V34 and S384 in MEI-S332; N29, I50 and S294 in Sgo1) (Fig. 6,
marked as arrowheads). Other conserved amino acids within the
second block (P293, R296, K298, L299, R300 in Sgo1) were again
all required for Sgo1 function (Fig. 6, asterisks), whereas nonconserved
residue T297 could be changed to alanine without loss of
function (Fig. 6, circle). These results underscore that the marginal
Figure 4 Sgo2 has a crucial role in mitotic division at the kinetochore. a, Serial dilutions of
the indicated cultures were spotted onto YEA plates containing 0, 5 or 10 mg ml21
TBZ
and incubated at 30 8C for 3 days. b, The indicated strains were streaked on YEA plates
and incubated at 30 8C for 3 days. c, The frequency of non-disjunction of cen2–GFP was
monitored in the indicated proliferating cells. d, Sgo2–GFP (green) was detected at
metaphase in wild-type and bub1D cells expressing CFP–Atb2 for the visualization of
spindles (red). DNA was stained with Hoechst (blue). A wild-type cell at anaphase is also
shown. e, The sgo2 þ
–GFP mis6 þ
–HA cells were fixed and stained with anti-GFP and
anti-HA antibodies. Note that the Sgo2–GFP signal near Mis6–HA is faint or undetectable
in interphase cells but is obvious in metaphase cells. f, A ChIP assay was used to measure
Sgo2–GFP levels throughout the indicated chromosome sites in cells arrested at
prometaphase or in asynchronous cells.
articles
NATURE | VOL 427 | 5 FEBRUARY 2004 | www.nature.com/nature514 ©2004 NaturePublishing Group
structural similarity found between S. pombe Sgo1 and the other
proteins in various eukaryotes is significant. Plants and mammals
carry two shugoshin-like proteins, suggesting that the function of
shugoshin may have diverged to accomplish mitosis and meiosis, as
in fission yeast.
Discussion
Our results provide some mechanistic insight into how shugoshin
protects centromeric cohesion in meiosis. Previously we have
suggested that Rec8 complexes enriched at the pericentromeric
regions are crucial for preserving centromeric cohesion through
meiosis I (ref. 18). Here we show that the protector protein Sgo1
also localizes at pericentromeric regions during meiosis I (Fig. 3d),
consistent with the notion that Sgo1 protects precisely those Rec8
complexes that preserve centromeric cohesion. We found that Rec8
localization is not dependent on Sgo1 and vice versa (Fig. 2e, not
shown). This independency of localization ensures that the mechanism
protects Rec8 only at centromeres and not along chromosome
arm regions. We suggest that shugoshin shields Rec8 physically from
separase action or counteracts it. In this regard, we found that
strong overexpression of Sgo1 moderately interferes with mitotic
growth even in the absence of Rec8 expression (not shown), and that
mild expression of Sgo1 kills a cut1 mutant32
even at the permissive
temperature for the cut1 allele (Supplementary Fig. 7). These results
suggest that Sgo1 itself might have some ability to counteract
separase function in vivo, although further analysis is necessary to
address the precise mechanism by which shugoshin counteracts
separase at the centromeres. We have shown that the co-expression
of Sgo1 and Rec8 leads to the inability of sister chromatids to
separate in mitosis (Fig. 1), suggesting that Sgo1 itself has the ability
to protect Rec8 from degradation. However, in budding yeast cells,
which express only a single shugoshin homologue, expression of
Rec8 does not lead to a block in mitosis unless Spo13 is also
expressed14,15
. This observation suggests that Spo13 is a potential
meiotic activator of budding yeast shugoshin. As S. pombe cells
expressing Rec8 instead of Rad21 exhibit normal mitotic growth
(Fig. 1a), Sgo2 has no obvious activity in protecting Rec8 during
mitosis. Thus, S. pombe Sgo1 appears to be well developed as a
specialized Rec8 protector, obviating the need for meiosis-specific
activators of shugoshin in fission yeast, and potentially explaining
why Spo13 is not conserved.
Although fission yeast Sgo2 has some minor involvement in
meiosis, it appears to function primarily in mitosis, whereas Sgo1 is
dispensable for mitosis. Previous studies in Drosophila revealed that
MEI-S332, presumably the only shugoshin in this organism, localizes
to centromeres in mitosis and may have some role in strengthening
cohesion33
. Here we find that the single budding yeast
shugoshin has an important role in mitotic chromosome segregation,
as the shugoshin mutant shows obvious chromosome
instability during proliferation (Fig. 5b). Fission yeast sgo2D cells
also show a modest defect in mitotic chromosome segregation and it
becomes marked if an HP1 homologue, Swi6, is simultaneously
depleted (Fig. 4b, c). Therefore, Sgo2 has an important role at
mitotic kinetochores in a redundant capacity with centromeric
heterochromatin. As heterochromatin is required to recruit large
amounts of cohesin to centromeres34,35
, our results suggest that
mitotic shugoshin may also have a close functional relationship with
cohesin. One proposal would be that Bub1 senses a lack of
Figure 6 Alignment of the amino-terminal coiled-coil regions and C-terminal basic
regions of shugoshin-like proteins in various organisms. The primary sequences of the
N-terminal regions of Sgo1 are conserved among S. pombe (Sp; Sgo1 and Sgo2),
S. cerevisiae (Sc; ScSgo1) and N. crassa (Nc; B23G1.060), whereas the sequences in
other species, including MEI-S332, are not conserved, although all carry the putative
coiled-coil motif (predicted by COILS program44
). Dm, Drosophila melanogaster; Ce,
Caernorhabditis elegans; At, Arabidopsis thaliana; Mm, Mus musculus; Hs, Homo
sapiens. See text for definition of arrowheads, asterisks and circle.
Figure 5 Analysis of budding yeast shugoshin ScSgo1. a, Serial dilutions of the
indicated cultures were spotted onto YPD plates containing 0 or 15 mg ml21
benomyl.
b, Chromosome loss was analysed in wild-type (WT) and Scsgo1D mutants by a
colony-sectoring assay. The ubr1D mutant was used as a positive control40
. The
frequency of sectoring colonies is shown at the bottom (n . 120). c, Examples of
segregation of cenV–GFP in Scsgo1D tetrads. The segregation patterns in tetrads were
classified mostly as one of the three shown at the right (n ¼ 200). d, ScSGO1–Myc
diploids were induced to synchronous meiosis and examined for the segregation of
cenV–GFP marked on one of two homologues at meiosis I and meiosis II. The cells mostly
underwent a reductional segregation pattern at meiosis I (96%, n ¼ 207), whereas there
was a high incidence of non-disjunction at meiosis II (34%, n ¼ 322). e, Cells marked
with cenV–GFP on both homologues were induced to meiosis and counter-stained with
anti-tubulin antibody and DAPI. Cells at late anaphase I were examined for dots of cenV–
GFP expression. ScSGO1–Myc cells frequently showed split cenV–GFP expression dots at
either pair of sister chromatids (72%, n ¼ 138), whereas control wild-type cells did not
(,2%, n ¼ 106).
articles
NATURE | VOL 427 | 5 FEBRUARY 2004 | www.nature.com/nature 515©2004 NaturePublishing Group
kinetochore–microtubule attachments or tension and thus stabilizes
Sgo2 to maintain cohesion at centromeres. However, we could
not detect any decrease of the cohesin Rad21 from centromeres in
sgo2D cells even when the spindles were destabilized by the nda3
mutation so that all cells were arrested at prometaphase by spindle
checkpoint (T.S.K and Y.W., unpublished observations). The
detailed analysis of shugoshin function at mitotic kinetochores
will be an intriguing subject of future study.
Our studies revealed that the conserved kinetochore kinase Bub1
is required for the centromeric localization of both mitotic and
meiotic shugoshins, suggesting that Bub1 might regulate loading
and/or maintenance of shugoshin at kinetochores. Although it is
unclear whether shugoshin is a direct substrate of Bub1, the
coincident localization of these proteins at metaphase kinetochores
during mitosis and meiosis I suggests a possible close interaction
between these proteins. Notably, sgo1D and sgo2D mutants share
many if not all of the defects of bub1D cells in meiosis and mitosis,
respectively (Figs 2d and 4b, c). Thus, we propose that shugoshin is a
crucial target of Bub1 function at kinetochores. As mutations in
human homologues of Bub1 have been found in subtypes of
colorectal cancer that exhibit chromosome instability36
, our results
suggest that human shugoshin may be a potential oncoprotein. In
this regard, it is intriguing that the human shugoshin-like protein
Q9BVA8 was recently identified as an antigen whose level is elevated
in most breast cancers37
. Chromosome segregation in meiosis is also
clinically important, as failures in this process result in aneuploidy, a
major cause of miscarriage and birth defects in humans38
. In
conclusion, we have identified a novel protein family, shugoshin,
that protects sister chromatid cohesion proteins at centromeres
during meiosis. Moreover, we have also established that shugoshin
has a crucial role at kinetochores in guarding against chromosome
instability during mitosis. This work provides a new model for
considering the evolution of meiosis and an integrated understanding
of eukaryotic chromosome segregation. A
Methods
Screening of the Rec8 protector
We searched for a gene that is toxic only when co-expressed with Rec8 in vegetative cells.
The Rec8 coding sequence fused with GFP was cloned under the thiamine-repressible
nmt1 promoter into pREP82 (ura4þ
marker), to construct pREP82–rec8þ
–GFP. We used
an S. pombe complementary DNA library, which was constructed using messenger RNA
prepared from meiotic cells, and pREP3 vector (nmt1 promoter, LEU2þ
marker)
(Y. Akiyoshi and Y.W., unpublished observations). The leu1 ura4-D18 cells carrying
pREP82– rec8þ
–GFP were transformed with the cDNA library, spread on agar plates
containing thiamine (promoter off) and incubated at 30 8C for 3 days. The colonies were
then replicated onto two thiamine-free plates: one containing 50
-FOA (50
-fluoro-ortoic
acid) and uracil where only cells that drop the plasmid pREP82–rec8þ
–GFP can grow
(thereby expressing a library clone alone), and the other without 50
-FOA and uracil
(allowing co-expression of rec8þ
–GFP and a library clone). We added phloxine B—a drug
that stains dead cells red—to both plates, thereby highlighting dying colonies. After
incubation for two days, the colonies exhibiting only dead cells on the co-expression plate
were picked up, and the library-derived plasmids were recovered and analysed.
Schizosaccharomyces pombe strains
All strains used are listed in Supplementary Table 1. Deletion and GFP- or Flag-tagging of
endogenous sgo1þ
and sgo2þ
were performed by a polymerase chain reaction (PCR)based
gene-targeting method39
. sgo1þ
–Flag–GFP was created by inserting GFP at the
C terminus of the PCR-amplified sgo1þ
–Flag, and integrated at the endogenous sgo1
locus. We further replaced the endogenous promoter of sgo1þ
with the nmt1 promoter to
generate Pnmt–sgo1þ
or Pnmt–sgo1þ
–Flag–GFP by the PCR-based gene targeting
method39
. We abbreviate the tagged protein to Sgo1–GFP or Sgo1–Flag, depending on the
purpose. We used a mei4D mutation to arrest meiotic cells before meiosis I (close to late
prophase in meiosis I) and a mes1D mutation to arrest cells after meiosis I, as described8
.
Chromatin immunoprecipitation assays
We used diploid sgo1þ
–Flag–GFP cells for ChIP assays with Sgo1. To achieve a highly
synchronous culture, we replaced the endogenous slp1þ
(CDC20 homologue) promoter
with the rad21þ
promoter, which is not active during meiosis, to arrest the cells at
metaphase I. The cells were incubated in nitrogen-depleted medium for 12 h at 30 8C with
the result that approximately 60% of the cells were arrested at metaphase I. For ChIP with
Sgo2, nda3-KM311 sgo2þ
–GFP cells were grown at 30 8C, and then shifted to 18 8C. After
incubation for 8 h most of the cells were arrested at prometaphase. The cells were fixed
with 3% paraformaldehyde at 18 8C for 30 min and extracts were prepared. ChIP assays
were carried out as described previously35
. The sequences of primers used have been
described previously20
except for the mat primers 50
-GTATGTGGAACAAGAGAAG-30
and 50
-CTCGCCTGCTTACATTTTAAGG-30
.
Preparation of anti-Sgo1 antibodies
The sgo1þ
ORF was amplified by PCR from an S. pombe cDNA library, and inserted into
plasmids pGEX4T-2 (Pharmacia Biotech) and pET-19b (Novagen) for the production of
recombinant proteins glutathione S-transferase (GST)–Sgo1 and His-tagged Sgo1,
respectively. GST–Sgo1 was used to immunize a rabbit, and the raised antibodies were
purified by His-tagged Sgo1 as previously described13
.
Immunostaining
To stain endogenous Sgo1, wild-type diploid cells cultured for 5 h in MM-N medium were
fixed with 3% formaldehyde and stained by the method described previously13
. Sgo1 was
detected using rabbit anti-Sgo1 antibodies and Alexa-488-conjugated anti-rabbit
antibody (Molecular Probes). Tubulin was detected using the mouse anti-tubulin
antibody TAT-1 (a gift from K. Gull) and Cy3-tagged anti-mouse antibody (Chemicon).
For detecting GFP-tagged proteins, we used mouse anti-GFP antibody (Roche) and
BODIPY FL-conjugated anti-mouse antibody (Molecular Probes). Mis6–haemagglutinin
(HA) was detected with rabbit anti-HA antibody Y-11 (Santa Cruz) and Alexa-488conjugated
anti-rabbit antibody. Cells were counterstained with DAPI (4,6-diamidino-2phenylindole)
to visualize DNA.
Co-immunoprecipitation
Cells of strain Padh–rec8þ
–3HA Pnmt41–sgo1þ
–FLAG–GFP and control Padh–rec8þ
–
3HA cells were cultured without thiamine for 15 h at 30 8C, collected, and extracts were
prepared. To liberate chromatin-bound proteins, we treated the extracts with DNase I.
After clarifying the extracts by centrifugation, the Sgo1–Flag–GFP protein was
immunoprecipitated with anti-Flag antibody M2 (Sigma). Rec8–3HA and Sgo1–Flag–
GFP were detected by anti-HA antibody Y-11 and anti-Flag antibody M2, respectively.
Analysis of budding yeast
All strains used are listed in Supplementary Table 1. The chromosome loss assay was
carried out as described previously40
. The ScSGO1 gene was deleted or epitope-tagged
using PCR-generated cassettes41
. Correct gene targeting was checked by PCR. URA3–GFP
dots marking chromosome V (cenV–GFP) were described previously6
. Sporulation was
induced by incubating cultures of diploid cells at 30 8C as described previously42
. In situ
immunofluorescence was performed as described42
.
Received 24 October; accepted 19 December 2003; doi:10.1038/nature02312.
Published online 18 January 2004.
1. Nasmyth, K. Disseminating the genome: joining, resolving, and separating sister chromatids during
mitosis and meiosis. Annu. Rev. Genet. 35, 673–745 (2001).
2. Koshland, D. E. & Guacci, V. Sister chromatid cohesion: the beginning of a long and beautiful
relationship. Curr. Opin. Cell Biol. 12, 297–301 (2000).
3. Uhlmann, F. Chromosome cohesion and separation: from men and molecules. Curr. Biol. 13,
R104–R114 (2003).
4. Lee, J. Y. & Orr-Weaver, T. L. The molecular basis of sister-chromatid cohesion. Annu. Rev. Cell Dev.
Biol. 17, 753–777 (2001).
5. Hirano, T. The ABCs of SMC proteins: two-armed ATPases for chromosome condensation, cohesion,
and repair. Genes Dev. 16, 399–414 (2002).
6. Klein, F. et al. A central role for cohesins in sister chromatid cohesion, formation of axial elements, and
recombination during yeast meiosis. Cell 98, 91–103 (1999).
7. Parisi, S. et al. Rec8p, a meiotic recombination and sister chromatid cohesion phosphoprotein of the
Rad21p family, conserved from fission yeast to humans. Mol. Cell. Biol. 19, 3515–3528 (1999).
8. Watanabe, Y. & Nurse, P. Cohesin Rec8 is required for reductional chromosome segregation at
meiosis. Nature 400, 461–464 (1999).
9. Pasierbek, P. et al. A Caenorhabditis elegans cohesion protein with functions in meiotic chromosome
pairing and disjunction. Genes Dev. 15, 1349–1360 (2001).
10. Eijpe, M., Offenberg, H., Jessberger, R., Revenkova, E. & Heyting, C. Meiotic cohesin REC8 marks the
axial elements of rat synaptonemal complexes before cohesins SMC1b and SMC3. J. Cell Biol. 160,
657–670 (2003).
11. Stoop-Myer, C. & Amon, A. Meiosis: Rec8 is the reason for cohesion. Nature Cell Biol. 1, E125–E127
(1999).
12. Buonomo, S. B. et al. Disjunction of homologous chromosomes in meiosis I depends on proteolytic
cleavage of the meiotic cohesin Rec8 by separin. Cell 103, 387–398 (2000).
13. Kitajima, T. S., Miyazaki, Y., Yamamoto, M. & Watanabe, Y. Rec8 cleavage by separase is required for
meiotic nuclear divisions in fission yeast. EMBO J. 22, 5643–5653 (2003).
14. Shonn, M. A., McCarroll, R. & Murray, A. W. Spo13 protects meiotic cohesin at centromeres in
meiosis I. Genes Dev. 16, 1659–1671 (2002).
15. Lee, B. H., Amon, A. & Prinz, S. Spo13 regulates cohesin cleavage. Genes Dev. 16, 1672–1681 (2002).
16. Blower, M. D. & Karpen, G. H. The role of Drosophila CID in kinetochore formation, cell-cycle
progression and heterochromatin interactions. Nature Cell Biol. 3, 730–739 (2001).
17. Kerrebrock, A. W., Moore, D. P., Wu, J. S. & Orr-Weaver, T. L. MEI-S332, a Drosophila protein
required for sister-chromatid cohesion, can localize to meiotic centromere regions. Cell 83, 247–256
(1995).
18. Kitajima, T. S., Yokobayashi, S., Yamamoto, M. & Watanabe, Y. Distinct cohesin complexes organize
meiotic chromosome domains. Science 300, 1152–1155 (2003).
19. Toth, A. et al. Functional genomics identifies monopolin: a kinetochore protein required for
segregation of homologs during meiosis I. Cell 103, 1155–1168 (2000).
20. Yokobayashi, S., Yamamoto, M. & Watanabe, Y. Cohesins determine the attachment manner of
kinetochores to spindle microtubules at meiosis I in fission yeast. Mol. Cell. Biol. 23, 3965–3973 (2003).
21. Yamamoto, A. & Hiraoka, Y. Monopolar spindle attachment of sister chromatids is ensured by two
articles
NATURE | VOL 427 | 5 FEBRUARY 2004 | www.nature.com/nature516 ©2004 NaturePublishing Group
distinct mechanisms at the first meiotic division in fission yeast. EMBO J. 22, 2284–2296 (2003).
22. Mata, J., Lyne, R., Burns, G. & Ba¨hler, J. The transcriptional program of meiosis and sporulation in
fission yeast. Nature Genet. 32, 143–147 (2002).
23. Watanabe, Y., Yokobayashi, S., Yamamoto, M. & Nurse, P. Pre-meiotic S phase is linked to reductional
chromosome segregation and recombination. Nature 409, 359–363 (2001).
24. Saitoh, S., Takahashi, K. & Yanagida, M. Mis6, a fission yeast inner centromere protein, acts during
G1/S and forms specialized chromatin required for equal segregation. Cell 90, 131–143 (1997).
25. Ekwall, K. et al. The chromodomain protein Swi6: a key component at fission yeast centromeres.
Science 269, 1429–1431 (1995).
26. Bernard, P., Maure, J. F. & Javerzat, J. P. Fission yeast Bub1 is essential in setting up the meiotic pattern
of chromosome segregation. Nature Cell Biol. 3, 522–526 (2001).
27. Yamaguchi, S., Decottignies, A. & Nurse, P. Function of Cdc2p-dependent Bub1p phosphorylation
and Bub1p kinase activity in the mitotic and meiotic spindle checkpoint. EMBO J. 22, 1075–1087
(2003).
28. Bernard, P., Hardwick, K. & Javerzat, J. P. Fission yeast bub1 is a mitotic centromere protein essential
for the spindle checkpoint and the preservation of correct ploidy through mitosis. J. Cell Biol. 143,
1775–1787 (1998).
29. Henikoff, S., Pietrokovski, S. & Henikoff, J. G. Superior performance in protein homology detection
with the Blocks Database servers. Nucleic Acids Res. 26, 309–312 (1998).
30. Bailey, T. L. & Gribskov, M. Combining evidence using p-values: application to sequence homology
searches. Bioinformatics 14, 48–54 (1998).
31. Tang, T. T., Bickel, S. E., Young, L. M. & Orr-Weaver, T. L. Maintenance of sister-chromatid cohesion
at the centromere by the Drosophila MEI-S322 protein. Genes Dev. 12, 3843–3856 (1998).
32. Kumada, K. et al. Cut1 is loaded onto the spindle by binding to Cut2 and promotes anaphase spindle
movement upon Cut2 proteolysis. Curr. Biol. 8, 633–641 (1998).
33. LeBlanc, H. N., Tang, T. T., Wu, J. S. & Orr-Weaver, T. L. The mitotic centromeric protein MEI-S332
and its role in sister-chromatid cohesion. Chromosoma 108, 401–411 (1999).
34. Bernard, P. et al. Requirement of heterochromatin for cohesion at centromeres. Science 294,
2539–2542 (2001).
35. Nonaka, N. et al. Recruitment of cohesin to heterochromatic regions by Swi6/HP1 in fission yeast.
Nature Cell Biol. 4, 89–93 (2002).
36. Cahill, D. P. et al. Mutations of mitotic checkpoint genes in human cancers. Nature 392, 300–303
(1998).
37. Scanlan, M. J. et al. Humoral immunity to human breast cancer: antigen definition and quantitative
analysis of mRNA expression. Cancer Immun. 1, 4 (2001).
38. Hassold, T. & Hunt, P. To err (meiotically) is human: the genesis of human aneuploidy. Nature Rev.
Genet. 2, 280–291 (2001).
39. Ba¨hler, J. et al. Heterologous modules for efficient and versatile PCR-based gene targeting in
Schizosaccharomyces pombe. Yeast 14, 943–951 (1998).
40. Rao, H., Uhlmann, F., Nasmyth, K. & Varshavsky, A. Degradation of a cohesin subunit by the N-end
rule pathway is essential for chromosome stability. Nature 410, 955–959 (2001).
41. Longtine, M. S. et al. Additional modules for versatile and economical PCR-based gene deletion and
modification in Saccharomyces cerevisiae. Yeast 14, 953–961 (1998).
42. Rabitsch, K. P. et al. Kinetochore recruitment of two nucleolar proteins is required for homolog
segregation in meiosis I. Dev. Cell 4, 535–548 (2003).
43. Glynn, J. M., Lustig, R. J., Berlin, A. & Chang, F. Role of bud6p and tea1p in the interaction between
actin and microtubules for the establishment of cell polarity in fission yeast. Curr. Biol. 11, 836–845
(2001).
44. Lupas, A., Van Dyke, M. & Stock, J. Predicting coiled coils from protein sequences. Science 252,
1162–1164 (1991).
Supplementary Information accompanies the paper on www.nature.com/nature.
Acknowledgements We thank J. P. Cooper for critical reading of the manuscript; S. Hauf,
M. Ohsugi and R. Watanabe for suggestions; J. P. Javerzat, F. Chang, T. Toda, M. Yanagida and
P. Nurse for strains and plasmids of fission yeast; and F. Klein, K. P. Rabitsch, K. Nasmyth,
M. Longtine, A. Shinohara and T. Maeda for strains and methods of budding yeast. We appreciate
the support of M. Yamamoto and all members of his laboratory for their help. This work was
supported in part by grants from the Ministry of Education, Science and Culture of Japan.
Competing interests statement The authors declare that they have no competing financial
interests.
Correspondence and requests for materials should be addressed to Y.W.
(ywatanab@ims.u-tokyo.ac.jp).
articles
NATURE | VOL 427 | 5 FEBRUARY 2004 | www.nature.com/nature 517©2004 NaturePublishing Group