-
Journal of Structural Biology 129, 123–143
(2000)doi:10.1006/jsbi.2000.4255, available online at
http://www.idealibrary.com on
Review: SMCs in the World of Chromosome Biology—From Prokaryotes
to Higher Eukaryotes
Neville Cobbe and Margarete M. S. Heck1
Institute of Cell and Molecular Biology, University of
Edinburgh, Michael Swann Building, King’s Buildings,Mayfield Road,
Edinburgh EH9 3JR, United Kingdom
Received December 16, 1999, and in revised form February 25,
2000
accddamtepssoapstctl
hdiascmlmsb
uaccaaeitpmussk
ifPHi1ad(wmtrleAac(ri
1
The study of higher order chromosome structurend how it is
modified through the course of the cellycle has fascinated
geneticists, biochemists, andell biologists for decades. The
results from manyiverse technical avenues have converged in
theiscovery of a large superfamily of chromosome-ssociated proteins
known as SMCs, for structuralaintenance of chromosomes, which are
predicted
o have ATPase activity. Now found in all eukaryotesxamined, and
numerous prokaryotes as well, SMCslay crucial roles in chromatid
cohesion, chromo-ome condensation, sex chromosome dosage
compen-ation, and DNA recombination repair. In eukary-tes, SMCs
exist in five subfamilies, which appear tossociate with one another
in particular pairs toerform their specific functions. In this
review, weummarize current progress examining the roleshese
proteins, and the complexes they form, play inhromosome metabolism.
We also present a twist inhe SMC story, with the possibility of one
SMC moon-ighting in an unpredicted location. r 2000 Academic
Press
INTRODUCTION
Stretched end-to-end, the DNA in any one cell of auman body
would measure about 2 m. Not onlyoes the cell manage to fit this
huge length of DNAnto its approximately 5-µm-diameter nucleus,
itlso condenses it even further prior to cell division,o that the
length of a single DNA molecule isompacted nearly 10,000-fold in
the metaphase chro-osome. Two mechanistically distinct but
interre-
ated processes are involved in the formation ofitotic
chromosomes. In concert with replication or
hortly thereafter, cohesion must be establishedetween sister
chromatids and properly maintained
1To whom correspondence should be addressed. Fax: 144 (0)
k31 650 7027. E-mail: [email protected].
123
ntil the metaphase to anaphase transition. Addition-lly, the
chromatin must be compacted to yield twoondensed sister chromatids
tightly paired at theentromeric regions and also along the length
of therms. It is critical for this condensation to happen inn
orderly fashion so as to prevent any possiblentanglement or
breakage of sister chromatids dur-ng anaphase which would have dire
consequences tohe cell. This folding of interphase chromatin to
giveaired metaphase chromatids is surely one of theost visually
dramatic events of the cell cycle and
ltimately fundamental for ensuring the faithfulegregation of
genetic information during cell divi-ion (reviewed in Heck, 1997;
Koshland and Strunni-ov, 1996; Murray, 1998).Data suggesting
biochemical differences between
nterphase and mitotic chromatin came initiallyrom studies of the
synchronized nuclear cycles ofhysarum polycephalum, a true slime
mold. Histone1 was found to be extensively hyperphosphorylated
n mitosis (Bradbury et al., 1974; Mueller et al.,985) and
strikingly, histone phosphokinase activitydded exogenously to
segments of Physarum plasmo-ia was able to accelerate the
initiation of mitosisBradbury et al., 1974). Hyperphosphorylation
of H1as also observed in CHO cells, and in addition,itotic-specific
phosphorylation of serine 10 on his-
one H3 was noted (Gurley et al., 1975). Antibodiesecognizing
this highly conserved epitope specificallyabel mitotic chromosomes
in all higher eukaryotesxamined to date (Van Hooser et al., 1998;
Wei andllis, 1998). Mutation of this particular serine tolanine in
Tetrahymena leads to a disruption ofhromosome condensation in
mitosis and meiosisWei et al., 1999). These studies point strongly
to aole for specific histone H1 and H3 phosphorylationn mitotic
chromosome condensation in higher eu-
aryotes.
1047-8477/00 $35.00Copyright r 2000 by Academic Press
All rights of reproduction in any form reserved.
-
tocdts1tt1ptcSG(syedpE
dcnwtanmgrsbtbbSmItswraafbe1dbai
SF
tiwlgsCctNpwfmobamtabep(iadtodtdtdnatAdsBim(rftS
aAa
124 REVIEW: COBBE AND HECK
Not unexpectedly, nonhistone chromosomal pro-eins also play a
role in the dramatic reorganizationf higher order chromosome
structure during the cellycle. Striking electron microscope images
of histone-epleted mitotic chromosomes highlighted the exis-ence of
a proteinaceous substructure constraining aea of DNAloops of 50–100
kb (Paulson and Laemmli,977). This ‘‘scaffold’’ fraction, remaining
after his-one extraction, was remarkably simple in composi-ion: two
major proteins (Sc1 at 170 kDa and Sc2 at35 kDa) and a number of
smaller, less abundantroteins (Lewis and Laemmli, 1982). That this
frac-ion represented more than an artifact of biochemi-al
extraction became clear with the identification ofc1 as
topoisomerase II (Earnshaw et al., 1985;asser et al., 1986) and Sc2
as an SMC protein
Saitoh and Laemmli, 1994). Topoisomerase II washown to be
essential for chromosome segregation ineast (Holm et al., 1985;
Uemura et al., 1987) and, asxpected for a function in chromosome
dynamicsuring cell division, was observed to be a marker
forroliferating, and not quiescent, cells (Heck andarnshaw,
1986).Additional substantive breakthroughs in our un-
erstanding of chromosome behavior during the cellycle came from
the discovery of the SMC family, aovel family of
chromosome-associated ATPaseshich appear to have essential and
specific roles in
he higher order dynamics of chromosome cohesionnd condensation.
The SMC (structural mainte-ance of chromosomes, formerly stability
of minichro-osomes) proteins were initially identified through
enetic studies of chromosome segregation in Saccha-omyces
cerevisiae (Strunnikov et al., 1993). The firstuch molecule, Smc1p,
was originally characterizedy frequent minichromosome
nondisjunction in mu-ants (Larionov et al., 1985) and was later
shown toe essential for viability and maintaining cohesionetween
sister chromatids (Strunnikov et al., 1993).equence comparisons
revealed this molecule to be aember of a highly conserved and
ubiquitous family.
ndeed, today we know of several structurally dis-inct SMC
subgroups playing a key role in chromo-ome dynamics in a host of
eukaryotic organisms asell as archaebacteria and many eubacteria
(Hi-
ano, 1998, 1999; Jessberger et al., 1998; Koshlandnd Strunnikov,
1996; Strunnikov, 1998; Strunnikovnd Jessberger, 1999). Although no
canonical SMCamily members have been found in gram-negativeacteria,
similar phenotypes are displayed by Esch-richia coli mutants
affecting mukB (Niki et al.,992), which encodes an SMC-like protein
(despiteifferences at the termini) with orthologues in
otheracteria. Thus, it appears that SMC proteins haven ancient
origin, reflecting their fundamental role
n chromosome dynamics. The current phylogeny of d
MC subfamilies and their members is displayed inig. 1.A typical
SMC molecule ranges in mass from 115
o 165 kDa and contains five major domains, asnferred from motifs
in the amino acid sequence, inhich the N- and C- termini are
separated by two
ong coiled-coils of 200–450 residues and a central,lobular hinge
region (Jessberger et al., 1998; Peter-on, 1994). The most
characteristic motif is the-terminal ‘‘DA’’ box which was noted to
have aandidate Walker B motif (ATP hydrolysis signa-ure) (Saitoh et
al., 1994; Walker et al., 1982). As the-terminal end of the
molecule also contains autative Walker A motif (ATP binding
domain), itas suggested that a functional ATPase domain may
orm by uniting the DA box with the ATP-bindingotif (Saitoh et
al., 1994). This could occur in either
f two ways: the molecule could bend at the hinge toring the two
termini together or by dimerizing as anntiparallel coiled-coil,
bringing the N-terminal do-ain of one subunit next to the
C-terminal domain of
he other. Indeed, in the case of MukB from E. colind the Smc
protein from Bacillus subtilis, it haseen shown that both
structures are possible (Melbyt al., 1998). When rotary-shadowed
samples of theurified proteins were viewed by electron
microscopyEM), they both showed a striking symmetry, appear-ng as a
flexible hinge connecting two thin, rod-likerms with terminal
globular domains. A range ofifferent conformations were also
observed, in whichhe two arms folded tightly against each other
orpened up to 180° (separating the terminal globularomains by 100
nm). To distinguish which ends ofhe protein corresponded to the
observed globularomains, a modified MukB was created by deletinghe
C-terminal domain and replacing the N-terminalomain with a
rod-shaped 40-kDa fragment of fibro-ectin. When viewed by EM, the
fibronectin domainppeared at both ends, indicating that each half
ofhe V-shaped dimer was an antiparallel coiled-coil.lthough the
structure of other SMC proteins inifferent organisms has yet to be
determined, iteems likely that they share the structure of Smc in.
subtilis as a similar basic head–rod–tail structure
s also conserved in even more distantly relatedolecules such as
the SbcCD nuclease of E. coli
Connelly et al., 1998). Furthermore, the frictionalatio for the
one eukaryotic SMC heterodimer wasound to be similar to that of
MukB, suggesting thathe XCAP-C/XCAP-E heterodimer (and possibly
otherMC molecules) may adopt a similar conformation.Does the
antiparallel dimerization of SMCs gener-
te a functional ATPase? Using the analogue azido-TP, which
covalently bonds proteins after lightctivation (Knight and McEntee,
1985), it has been
emonstrated that only the N-terminal domain of
-
y(d(HbSKSsAlhDarKe(na(bbasaSdshsahtmal
mfatrVaaf11mbasSa
ttathrimas11d1ShpStaYfd1Tspihm‘SSatLSi‘Mip
tdrpms1nSatw
125REVIEW: STRUCTURAL MAINTENANCE OF CHROMOSOMES
east and mammalian SMCs can directly bind ATPAkhmedov et al.,
1998). By contrast, the C-terminalomain appears to be sufficient
for DNA bindingAkhmedov et al., 1998; Graumann et al.,
1998).owever, the presence of double-stranded DNA haseen shown to
stimulate ATP hydrolysis in twoMC-containing complexes (Jessberger
et al., 1996b;imura and Hirano, 1997), whereas the B. subtilismc
homodimer (which binds preferentially to single-tranded DNA) has a
single-strand DNA-stimulatedTPase activity (Hirano and Hirano,
1998). This at
east suggests that ATP hydrolysis might be en-anced by bringing
together the respective ATP andNA binding motifs of the termini.
Conversely,lthough ATP is not required for DNA binding (Hi-ano and
Hirano, 1998; Kimura and Hirano, 1997;imura et al., 1999), it is
clearly required for prefer-ntial binding to positively supercoiled
substratesKimura et al., 1999). Likewise, ATP binding (thoughot
hydrolysis) is also required for the enhancedggregation of the B.
subtilis Smc with ssDNAHirano and Hirano, 1998). Of course, it
remains toe seen if ATP hydrolysis itself is strictly abolishedy
removing the C-terminal domain. However, thebility to form
antiparallel dimers is not necessarilyufficient to generate a
functional ATPase. For ex-mple, the ATP-stimulated activity of a
XenopusMC complex involved in chromosome condensationepends on the
presence of additional non-SMCubunits (Kimura et al., 1998, 1999)
and the SMCeterodimer of the RC-1 recombination compleximilarly
requires other components for full ATPasectivity (Jessberger et
al., 1996b). In any event, ATPydrolysis appears to be required for
the full func-ion of SMC-containing complexes, as shown
byutagenesis of the ATP-binding domain (Chuang et
l., 1994; Verkade et al., 1999) or the use of nonhydro-yzable
ATP analogues (Kimura and Hirano, 1997).
The eukaryotic SMCs may be divided into fiveajor groups,
including the Rad18 subfamily and
our other subfamilies whose members may combines heterodimers in
larger functional complexes. Whilehe Rad18 members are only
essential for DNAepair (Lehmann et al., 1995; Mengiste et al.,
1999;erkade et al., 1999), the other SMC moleculesppear to have
essential nonoverlapping functions,s examples of each type are
known to be requiredor viability (Holt and May, 1996; Michaelis et
al.,997; Saka et al., 1994; Strunnikov et al., 1993,995). The
ability of particular eukaryotic SMColecules to combine as
heterodimers is suggested
y their co-immunoprecipitation in roughly equimolarmounts
(Darwiche et al., 1999; Hirano and Mitchi-on, 1994; Lieb et al.,
1998; Losada et al., 1998;chmiesing et al., 1998; Sutani et al.,
1999) and by
nalogy with the observed homodimerization of bac- (
erial SMCs (Melby et al., 1998). Although the poten-ial for
eukaryotic SMCs to form homodimers haslso been demonstrated, this
nonpreferential associa-ion only occurred when the fusion proteins
wereighly overexpressed, thereby titrating out the natu-al SMC
partner (Strunnikov et al., 1995). Moreover,t seems that
homodimerization is insufficient for
ost eukaryotic SMCs to function as their in vitroctivity depends
on the combined presence of bothubunits (Kimura and Hirano, 1997;
Kimura et al.,999; Schmiesing et al., 1998; Sutani and
Yanagida,997) and mutation of just one SMC partner pro-uces defects
in vivo (Chuang et al., 1994; Lieb et al.,998; Michaelis et al.,
1997; Saka et al., 1994;trunnikov et al., 1993, 1995). On the other
hand,igher levels of oligomerization would appear to berecluded by
considering the overall mass of isolatedMC-containing complexes and
the known mass ofhe other components (Hirano et al., 1997; Hiranond
Mitchison, 1994; Losada et al., 1998; Sutani andanagida, 1997;
Sutani et al., 1999). So far, twoundamental classes of SMC
heterodimer have beenescribed in various organisms (Heck, 1997;
Hirano,999; Hirano et al., 1995; Jessberger et al., 1998).hese
heterodimers may associate with differentets of non-SMC subunits to
yield a range of largerotein complexes with diverse functions, as
shownn Tables I and II and Fig. 2. The SMC2/SMC4eterodimer seems to
have a role in mediatingitotic chromosome condensation, as part of
the
‘condensin complex’’ (Hirano and Mitchison, 1994;utani and
Yanagida, 1997; Sutani et al., 1999). AnMC2 homologue and another
SMC4-like moleculere implicated in sex chromosome dosage
compensa-ion in Caenorhabditis elegans (Chuang et al., 1994;ieb et
al., 1996, 1998). On the other hand, theMC1/SMC3 heterodimer forms
part of a complex
mportant for sister chromatid cohesion, dubbed‘cohesins’’
(Guacci et al., 1997; Losada et al., 1998;
ichaelis et al., 1997; Tóth et al., 1999), and is alsonvolved
in recombination as part of the RC-1 com-lex (Jessberger et al.,
1996a,b).Unlike the SMC proteins of eukaryotes, no cofac-
ors for the B. subtilus Smc have been isolated toate (Sharpe and
Errington, 1999). Nonetheless, aole for the protein in chromosome
structure andartitioning was clearly demonstrated by the abnor-al
nucleoids and accumulation of anucleate cells in
mc mutants (Britton et al., 1998; Moriya et al.,998). Similar
phenotypes were also observed in smcull mutants of Caulobacter
crescentus (Jensen andhapiro, 1999). The B. subtilus smc mutants
werelso characterized by irregular subcellular localiza-ion of
Spo0J [a chromosome partitioning proteinhich binds to sites near
the origin of replication
Lin and Grossman, 1998)]. The role of the Smc
-
stipWPostmcB(aP
126 REVIEW: COBBE AND HECK
FIG. 1. Phylogenetic tree of known SMC proteins. The aboequences
generated by the ClustalW program (Methods Enzymolrees produced by
the ClustalW program were checked by resamplnferred using the
PROTDIST and NEIGHBOR programs of Joe Fhylip.html). The topology of
the branches comprising each indivisconsin Package to construct
separate alignments and then u
AUP program (http://www.lms.si.edu/PAUP/about.html) to find tf
subfamilies were confirmed by analysis of distance matrix
datubfamily was compared with the mean distance of each subfamilree
shown was confirmed by ClustalW alignment of partial dataultiple
substitutions, in addition to comparison with trees base
orrect topology of the more distantly related SMC-like
proteinsestFit program in the Wisconsin Package and then
constructing
length of shorter protein 4 length of alignment) 3 (100% 2 %
idend ClustalW programs were also compared with branch lengths
oHYLIP and ClustalW trees were viewed using Rod Page’s TreeVi
ve tree was constructed based on alignments between SMC protein.
266, 383–402, 1996), correcting for multiple substitutions. The
variousing with 1000 bootstrap trials and compared with
neighbor-joining treeselsenstein’s PHYLIP package
(http://evolution.genetics.washington.edu/idual subfamily was also
confirmed by using the PileUp program in thesing a GCG interface to
the tree-searching options of David Swofford’she optimal topology
by means of parsimony. Moreover, the designationsa, in which the
mean distance between all members of the same SMCy member from all
other proteins in the tree. The overall topology of thesets of
full-length SMC molecules both with and without correction for
d on alignment of the conserved N- and C-terminal domains alone.
Thewas confirmed by conducting all possible optimal alignments
using thea neighbor-joining tree from pairwise distances calculated
as distance 5ntity). The branch lengths in the overall tree
calculated by the PHYLIPf trees containing only members of the same
SMC subfamily. Finally, theew program
(http://taxonomy.zoology.gla.ac.uk/rod/treeview.html).
-
plnmienbntias1scrbs1
dpttrtp
t1ctemmmmom
dflcs
127REVIEW: STRUCTURAL MAINTENANCE OF CHROMOSOMES
rotein in B. subtilus is reflected by its chromosomalocalization
and its presence at the poles of theucleoid as discrete foci
(Britton et al., 1998; Grau-ann et al., 1998). As the N-terminal
region of Smc
s required for the formation of polar foci (Graumannt al.,
1998), it appears to be needed for pairing ofewly replicated
origins (Lin and Grossman, 1998)y mediating Spo0J localization to
the pole of theucleoid and thereby facilitating orderly
segrega-ion. However, due to the small size of bacterial cellst
remains unclear whether the mutant phenotypesre caused primarily by
a defect in chromosome conden-ation, segregation, or both (Sharpe
and Errington,999). Nonetheless, an insight into the possible
conden-ation activity of the B. subtilis Smc homodimer hasome from
the discovery of its ATP-dependent DNAeannealing activity (Hirano
and Hirano, 1998). Asacterial nucleoids contain unconstrained
negativeupercoils that may be easily unpaired (Pettijohn,
FIG. 2. Model of SMCs, from monomer to heterodimer to highomain
near its N-terminus and an ATP hydrolysis motif near iexible hinge
region near the center of the molecule. When heterolose proximity
to the Walker B motif of its partner SMC. SMCubunits, diagrammed as
three ovals, resulting in creation of the c
982), it has been proposed that the energy-depen- a
ent aggregation of single-stranded DNA may com-act bacterial
chromosomes by bringing such regionsogether (Hirano and Hirano,
1998). Furthermore,he increase in twist resulting from
SMC-mediatedestoration of base pairing may compact the DNAhrough
the concomitant generation of compensatoryositive supercoils
(Sutani and Yanagida, 1997).As DNA reannealing activities similar
to that of
he recombination protein recA (Weinstock et al.,979) have also
been observed with the S. pombeut3/cut14 heterodimer (Sutani and
Yanagida, 1997),he bovine bSMC1/bSMC3 heterodimer (Jessbergert al.,
1996b), and even the isolated C-terminal do-ains of Smc1p and Smc2p
from S. cerevisiae (Akh-edov et al., 1998), it has been suggested
that thisay represent an activity characteristic of all SMColecules
(Yanagida, 1998). However, these eukary-
tic proteins differ from the B. subtilis Smc ho-odimer as they
do not require ATP for reannealing
er functional complexes. The SMC monomer has an ATP
bindingrminus. The regions are separated by coiled-coil domains and
azed in an antiparallel fashion, one Walker A motif is brought
intorodimers then participate in complex formation with non-SMCand
condensin complexes.
er ordts C tedimeri
heteohesin
ctivity (Akhmedov et al., 1998; Sutani and Yanagida,
-
1aonw
mtcoa
C
D
C
R
128 REVIEW: COBBE AND HECK
997). As SMCs do not seem to translocate directlylong the DNA
(Kimura and Hirano, 1997), the rolef molecules such as bSMC1 and
bSMC3 in recombi-ation may be related to their role in cohesion,
inhich recombination is facilitated by keeping chro-
TAProtein Complexes Co
Species
SMC subunits
SMC2 type SMC4 type
ondensin complexS. cerevisiae Smc2 (Strunnikov et
al., 1995)Smc4
S. pombe Cut14 Cut3
C. elegans MIX-1 Z69787
D. melanogaster dmSMC2 dmSMC4
X. laevis XCAP-E (Hirano andMitchison, 1994)
XCAP-C (Hirano aMitchison, 1994
G. gallus ScII (Saitoh et al.,1994)
?
H. sapiens hCAP-E (Schmiesinget al., 1998)
hCAP-C (Schmieset al., 1998)
osage compensationC. elegans MIX-1 (Lieb et al.,
1998)DPY-27 (Chuang e
1994, 1996)
TAProtein Complexes Co
Species
SMC subunits
SMC1 type SMC3 type
ohesin complexS. cerevisiae Smc1 (Strunnikov et
al., 1993)Smc3 (Michaelis et al
1997)
S. pombe CAA22432 CAA15722
A. nidulans ? SudA (Holt and May,1996)
C. elegans AAB93638 CAB57898
D. melanogaster dmSMC1 dCAP (Hong andGanetsky, 1996)
X. laevis XSMC1 (Losadaet al., 1998)
XSMC3 (Losada et al1998)
M. musculus SMCB (Darwiche etal., 1999)
SMCD (Darwiche etal., 1999)
R. norvegicus SMC1 Bamacan (Wu andCouchman, 1997)
H. sapiens hSMC1 (Rocques etal., 1995; Schmi-esing et al.,
1998)
hCAP (Shimizu et al.1998)
ecombination com-plex
B. taurus bSMC1 (Jessberger etal., 1996)
bSMC3 (Jessberger eal., 1996)
atids close together. However, this does not explainhe
renaturation activity of SMCs that are active inhromosome
condensation but not cohesion. More-ver, as the isolated C-terminal
domains of Smc1pnd Smc2p are capable of efficient DNA
reannealing
Iing SMC2 and SMC4
Non-SMC subunits
AAB67384 Brrn1 CAB41223
cnd1 (Sutani et al.,1999)
cnd2 (Sutani et al.,1999)
cnd3 (Sutani et al.,1999)
CAA16340 ? ?
EST clot No. 2519 Barren (Bhat et al.,1996)
EST clot No. 2199
XCAP-D2 (Kimura etal., 1998)
XCAP-H (Hirano etal., 1997)
XCAP-G (Hirano etal., 1997)
? ? ?
063880 038553 ?
DPY-28 (Lieb et al.,1998)
DPY-26 (Lieb et al.,1996)
?
IIing SMC1 and SMC3
Non-SMC subunits
1/Scc1 (Guacci et., 1997; Michaelisal., 1997)
Scc2 (Michaelis et al.,1997)
Scc3 (Tóth et al., 1999)
21 (Birkenbihld Subramani,92)
Mis4 (Furuya et al.,1998)
?
? ? ?
? ? ?
09926 Nipped-B (Rollins etal., 1999)
Stromalin (Valdeolmil-los et al., 1998)
D21 (Losada et., 1998)
p155 (Losada et al.,1998)
p95 (Losada et al.,1998)
9 (Darwiche et., 1999)
? Stromal antigen 1 (Car-romolino et al., 1997)
? ? ?
D21 (McKay et., 1996)
? Stromal antigen 1 (Car-romolino et al., 1997)
ligase III (Jess-rger et al., 1993)
DNA Pol e (Jessbergeret al., 1996)
Endonuclease? (Jess-berger et al., 1996)
BLEntain
nd)
ing
t al.,
BLEntain
., Mcdalet
Radan19
AF1
., XRAal
PW2al
, hRAal
t DNAbe
-
otttttbaM(mrpcwti
sa(sitscmctTtittat1ca
sc
staiccwHp(s(ifi(smHpmtcfrcwlepeosbctmtwc
iccdacwcWpsb
trcfncstdfbcmah
129REVIEW: STRUCTURAL MAINTENANCE OF CHROMOSOMES
n their own,2 it seems that the ability of SMCs toether separate
DNA molecules may simply enhancehe reannealing activity by helping
to bring ssDNAogether. Clearly, more detailed analysis of the
reac-ions catalyzed by the terminal domains is requiredo determine
the mechanism of SMC-mediated recom-ination. Lastly, as SMC
molecules have a higherffinity for AT-rich sequences such as SARs
andARs (Scaffold- or Matrix- Associated Regions)
Akhmedov et al., 1998) which may be more easilyelted (Bode et
al., 1992), it is possible that the
enaturing activity of different SMC protein com-lexes in
eukaryotes might be linked to roles inondensation and segregation,
as in B. subtilus. Weill now look more closely at the role of SMCs
in
hese two process, beginning with their involvementn chromosome
condensation.
SMCs AND CHROMOSOME CONDENSATION
The first SMC proteins exhibiting a role in chromo-ome
condensation were found in S. pombe. The cut3nd cut14 mutants
displayed a characteristic ‘‘cut’’cell untimely torn) phenotype, in
which the divisioneptum bisects the nuclear material, due to a
failuren either chromosome condensation or sister chroma-id
segregation (Saka et al., 1994). However, it wasuggested that the
primary defect was a failure inhromosome condensation as high rates
of minichro-osome loss were not observed in cut3 mutants and
entromeric DNA was reported to segregate properlyo the spindle
poles in both cut3 and cut14 mutants.he improper chromosome
disjunction in these mu-ants therefore appeared to be a consequence
ofmpaired chromosome condensation. Similar pheno-ypes were also
observed in S. cerevisiae smc2 mu-ants, encoding a Cut14p
orthologue (Strunnikov etl., 1995). Cut3p was later shown to be
orthologouso Smc4p in S. cerevisiae (Koshland and Strunnikov,996).
In addition, the Sc2 protein of the mitotichromosome scaffold in
chicken cells was identifieds an SMC2 subfamily member, suggesting
a pos-
2Considering that similar molar concentrations were used forhe
reactions with either full-length or partial SMC proteins,
theeannealing activity of isolated SMC terminal domains appears
toonflict with the inability of full-length SMCs to promote
duplexormation unless they can heterodimerize (Hirano, 1999).
Fortu-ately, this discrepancy may be explained by considering
theollision rates of the different proteins with DNA. Presumably,
themall, globular truncated protein has a higher collision rate
thanhe long, rod-shaped full-length molecule because it is able
toiffuse through solution more easily. Likewise, a heterodimerormed
from two different SMCs has a COOH-terminal DNAinding domain at
each end so it is more likely to bind to DNA. Byontrast, as
individual SMC proteins are unlikely to form ho-odimers unless they
are produced in vast excess (Strunnikov et
l., 1995), they will only be able to bind DNA at one end and so
will
Cave a far lower collision rate.
ible structural role for these proteins in mitotichromosome
architecture (Saitoh et al., 1994).Much of our current
understanding of chromo-
ome condensation is based on the in vitro simula-ion of
chromosome condensation when nuclei aredded to mitotic extracts
from Xenopus eggs. Topo-somerase II was shown to be required for
mitotichromosome condensation when either HeLa orhicken erythrocyte
nuclei or demembranated spermere added to Xenopus extracts (Adachi
et al., 1991;irano and Mitchison, 1993). A heterodimeric com-lex
containing XCAP-E (SMC2-type) and XCAP-CSMC4-type) was identified
as a mitotic chromo-omal component in in vitro assembled
chromosomesusing demembranated sperm as substrate),
andmmunofluorescence detection of XCAP-C revealed alamentous
distribution along the chromosome axis
Hirano and Mitchison, 1994), not unlike that ob-erved for
topoisomerase II in ‘‘normal’’ mitotic chro-osomes (Earnshaw et
al., 1985; Earnshaw andeck, 1985). Two different types of condensin
com-lex were later identified by sucrose gradient sedi-entation,
namely an 8S form which proved to be
he XCAP-E/XCAP-C heterodimer and a larger 13Somplex containing
three additional subunits (re-erred to as XCAP-D2, XCAP-H, and
XCAP-G) (Hi-ano et al., 1997). Both the targeting of the
condensinomplex to chromosomes and its in vitro activityere shown
to depend on mitosis-specific phosphory-
ation of these additional non-SMC subunits (Hiranot al., 1997),
with p34cdc2 responsible for the hyper-hosphorylation of XCAP-D2
and XCAP-H (Kimurat al., 1998). XCAP-H was also found to have
homol-gy to Barren, a protein localizing to the chromo-omes of
mitotically active cells in Drosophila em-ryos (Bhat et al., 1996).
Mutants in barren wereharacterized by extensive chromatin bridges
be-ween anaphase chromosomes, in spite of centro-ere separation.
This phenotype was reminiscent of
he cut3 and cut14 mutants in S. pombe, consistentith a role for
Barren in proper mitotic chromosome
ondensation.A larger condensin complex has also been
detected
n S. pombe cell lysates, with a similar subunitomposition to
that of the Xenopus 13S condensinomplex (Sutani et al., 1999). Gene
disruption hasemonstrated that the additional non-SMC subunitsre
essential for viability and the mutants wereharacterized by
hypocondensed chromosomes whichere extended along an elongated
spindle instead of
learly separating, as seen in cut3 and cut14 cells.hereas the
activity of the Xenopus condensin com-
lex is regulated by phosphorylation of its non-SMCubunits, that
of the S. pombe condensin appears toe controlled by
mitosis-specific phosphorylation of
cdc2
ut3p by p34 . This modification was also shown
-
tbttoC
ieroo
cabtdn
130 REVIEW: COBBE AND HECK
o be essential for viability as it restricts the accessi-ility
of a nuclear export signal (NES) in the N-erminus of the protein.
As the intracellular shut-ling of the other condensin subunits
seems to dependn the localization of Cut3p, the phosphorylation
of
FIG. 3. Model for cohesin and condensin deposition and
activhromatin either coincident with or shortly after DNA
replicationnaphase. The condensin complex (blue circles) then
appears to bee thought of as a two-step process resulting first in
shortening ofhe final high degree of chromosome compaction.
Cohesion betwissolved at the transition to anaphase, and the
sisters are segreuclear envelope reformation and cytokinesis, the
chromatin is de
ut3p during mitosis permits entry of the complex M
nto the nucleus while the dephosphorylated formxposes the NES
during interphase and thereforeelegates the complex to the
cytoplasm. This mannerf regulating condensin activity contrasts
with thatf the Xenopus condensin complex (Hirano and
ing the cell cycle. The cohesin complex (red ovals) is loaded
ontohase, thereby ensuring the attachment of sister chromatids
untilduring chromosome condensation in prophase. Condensation
canterloop axis, followed by the introduction of supercoils to
achievee sister chromatids of the final metaphase chromosome is
thenby the microtubule apparatus to the poles of the cell.
Followingsed in preparation for transcription and DNA
synthesis.
ity durin S ploadedthe in
een thgated
conden
itchison, 1994) and the chicken Sc2 protein (Saitoh
-
edctdsatsdbscnh
aanrcabimSracctbaas(osepiH
stecD1acf(i
s(wht‘(tttp1hbicwciw
ftdhotipacptmlsflp(nlDotmtcpwhptpo
131REVIEW: STRUCTURAL MAINTENANCE OF CHROMOSOMES
t al., 1994), which both remain in the nucleusuring interphase
but fail to associate with thehromosomes until mitosis. Until it
can be shownhat the essential activity of the S. pombe
complexepends on the phosphorylation of any of its otherubunits,
one possible explanation for this shuttlingppears to be that the
complex is active throughouthe cell cycle. Consequently, it appears
that chromo-ome condensation during mitosis may be
regulatedifferently in different species although the sameasic
protein complex appears to be involved. Onehould also keep in mind
that S. pombe undergoeslosed nuclear mitosis, in contrast to the
observeduclear envelope breakdown of Xenopus and otherigher
eukaryotes.Purified Cut3p and Cut14p form a stable complex
t a rather low efficiency when mixed in vitro (Sutanind
Yanagida, 1997), so it has been proposed that theon-SMC subunits
Cnd1p and Cnd3p might have aole in linking the SMCs together in the
functionalondensin complex. This is suggested by the report ofweak
homology between Cnd1p and Cnd3p and thesubunit of the AP3 adapter
protein complex, which
s involved in the assembly of rod-like clathrinolecules for
vesicle transport (Sutani et al., 1999).ince both clathrin and SMC
molecules contain twood-like regions linked by a hinge-like region,
Cnd1pnd Cnd3p may enable the rod-like SMC subunits toorrectly
assemble. The intact Xenopus condensinomplex was also shown to
preferentially bind posi-ively supercoiled DNA and DNA with a
distorted orent structure, such as a four-way junction (Kimurand
Hirano, 1997; Kimura et al., 1999). A similarffinity for cruciform
DNA has also been demon-trated with Smc1p and Smc2p from S.
cerevisiaeAkhmedov et al., 1998). As bending and supercoilingf the
DNA may be generated by the SMCs, thisuggests that additional
condensins might bind coop-ratively (Kimura and Hirano, 1997), just
as otherroteins which strongly bind cruciform or bent DNAn vitro
tend to bend it further (Zlatanova and vanolde, 1998).
MECHANISM OF CONDENSIN ACTION
How is the condensin complex involved in chromo-ome
condensation? Although renaturation can con-ribute to supercoiling,
this activity fails to fullyxplain the role of certain SMCs in
chromosomeondensation. After all, the ability to promote duplexNA
is shared by other SMCs (Jessberger et al.,996b) and indeed non-SMC
proteins (Weinstock etl., 1979) which have no obvious direct role
inondensation. As the reannealing reaction is there-ore considered
to be only a part of its activityYanagida, 1998), how does the
condensin complex
nteract with chromatin to induce its mitotic conden- c
ation? Based on the symmetrical structure of BsSMCMelby et al.,
1998) in which each end could interactith both ATP and DNA
(Akhmedov et al., 1998), itas been suggested that the homodimer may
func-ion as an ATP-modulated DNA cross-linker with a‘scissoring’’
action to induce aggregation of DNAHirano, 1999). The possibility
of such conforma-ional changes during SMC activity is supported
byhe finding that sensitivity to proteolytic cleavage ofhe B.
subtilus SMC homodimer depends on theresence of ATP and ssDNA
(Hirano and Hirano,998). By extending this concept to eukaryotic
SMCeterodimers, this scissoring action was proposed toe the key
mechanism underlying all SMC activities,n which the SMC2/SMC4
heterodimers involved inhromosome condensation and dosage
compensationould act as intramolecular DNAcross-linkers which
ompact a single DNA molecule (Hirano, 1999). Whatmpact native
chromatin, in contrast to naked DNA,ould have on this proposed
process is anyone’s guess.An early model for condensin action was
put
orward by Kimura and Hirano when they foundhat the Xenopus 13S
condensin complex can intro-uce positive supercoils into DNA,
fueled by ATPydrolysis (Kimura and Hirano, 1997). The stretchesf
DNA between condensin binding sites could formwisted loops by
compensatory negative supercoil-ng, which are relaxed by treatment
with eitherrokaryotic or eukaryotic type I topoisomerases.
Ingreement with previously suggested mechanisms ofompaction based
on the chromosome scaffold modelroposed originally by Paulson and
Laemmli (1977),he authors proposed that chromosome condensationight
be initiated by the formation of chromatin
oops by condensin-mediated supercoiling at specificites,
followed by shortening of the interloop axis andolding of the
torsionally constrained loops. It wasater shown that condensin
reconfigures DNA in theresence of a type II topoisomerase by
creating knotsKimura et al., 1999). As knotting presumably wouldot
occur if condensin generated supercoils either by
ocally overwinding the DNA or by wrapping theNA around itself,
this implied that the complexperated by generating a global writhe.
When theopology of the knots was determined by electronicroscopy of
RecA-coated DNA, it was reported that
he vast majority were positive, implying that theondensin
complex generated an ordered array ofositive solenoidal supercoils.
As the condensinsere reported to bind to plasmid DNA in vitro at
aigh ratio (Kimura et al., 1998, 1999), it was pro-osed that a high
density of condensins could bindhe full DNA length, touching each
other to form arotein infrastructure capable of nonplanar bendingf
the DNA. This model neatly explains both the
ooperativity of condensin binding and the observed
-
pHthttsi
swHtvs(crsappsmas(nScndptistpci
cpladpmtpptnciF
tbtefcasccsto1anp1cadb
baCoactevrtt(wbswttaaMmtcDtmm
132 REVIEW: COBBE AND HECK
reference for longer DNA fragments (Kimura andirano, 1997) in
terms of cooperative binding. Fur-
hermore, as the two ends of the eukaryotic SMCeterodimer are
similar but not identical (containinghe N- and C-termini of
different SMC molecules),his asymmetry might contribute to the
chirality ofupercoiling, provided that the complex binds DNAn a
fixed orientation.
Although the high concentrations of condensinsupplied in vitro
allowed the complex to bind every-here on the naked plasmid DNA
(Hirano andirano, 1998; Kimura et al., 1999), it is unlikely
that
he same is true of protein-laden chromosomes inivo. Otherwise,
this would conflict with the ob-erved distribution of condensin
SMCs in XenopusHirano and Mitchison, 1994), chicken, and humanells
(Saitoh et al., 1994), in which they appearedestricted to the
chromosome axis with concentratedtaining at the centromeres (Saitoh
et al., 1994). Thebundance of Cut3p in wild-type S. pombe
cellsredicted a density of only one condensin complexer 8 kb of DNA
(Sutani and Yanagida, 1997) and aimilar stoichiometry was estimated
for Xenopusitotic chromosomes assembled in vitro (Kimura et
l., 1999). As SMC proteins were previously ob-erved to
preferentially bind AT-rich sequencesAkhmedov et al., 1998), it
seems possible thatonplanar bending might be initiated at loci such
asARs. By combining these findings, one model forondensation by
condensins would result from theonplanar bending of DNA by higher
order multicon-ensin complexes, generating positive solenoidal
su-ercoils at defined sites along the chromosome andhe simultaneous
generation of negative supercoilsn the intervening regions.
Although these negativeupercoils could be easily removed by the
numerousopoisomerases in a cell, this would presumably berevented
by the binding of additional unidentifiedondensation factors which
could stabilize thesenterwound loops.
Despite the insights provided by these models ofondensation, the
precise mode of action of SMCroteins continues to provoke
discussion. In particu-ar, the interaction between condensin
complexesnd topoisomerase II still remains enigmatic. Evi-ence of
possible genetic interactions was initiallyrovided by analysis of
different cut3 and topoIIutants in S. pombe (Saka et al., 1994),
although
hese results are equally consistent with the tworoteins acting
in a common pathway without directhysical interaction. A functional
interaction be-ween topoisomerase II and Barren was
suggested,onetheless, based on co-immunoprecipitation,
colo-alization on mitotic chromosomes, and interactionn a yeast
two-hybrid assay (Bhat et al., 1996).
urthermore, Barren has been reported to enhance i
he supercoiling activity of topoisomerase II, possi-ly
modulating topoisomerase II-mediated decatena-ion of chromosomal
arms. Finally, Sc2 and topoisom-rase IIa have been reported to
copurify in a complexound in undifferentiated mouse
erythroleukemiaells (Ma et al., 1993) and the two proteins
cofraction-te with and colocalize to the mitotic chromosomecaffold
of chicken cells (Saitoh et al., 1994). Inontrast, the Xenopus
condensins fail to immunopre-ipitate with topoisomerase II (Hirano
and Mitchi-on, 1994) and appear to be independently targetedo
mitotic chromosomes (Hirano et al., 1997). More-ver, unlike
topoisomerase IIa and b (Berrios et al.,985; Meyer et al., 1997;
Petrov et al., 1993; Zini etl., 1992), Sc2 is not a component of
the interphaseuclear matrix, as it readily leaked into the
cyto-lasm during subcellular fractionation (Saitoh et al.,994).
Regardless of whether the members of theondensin complex interact
directly with topoisomer-se II, it is clear that their respective
condensing andecatenating activities contribute synergistically
toring about chromosome condensation.Intriguingly, there may be a
functional similarity
etween mechanisms of chromosome condensationnd the global
regulation of gene expression on the. elegans X chromosome, based
on the involvementf an SMC2/4 heterodimer in sex chromosome dos-ge
compensation. Transcription from each of the Xhromosomes is reduced
in hermaphrodites (XX) ofhis organism to match the level of
X-linked genexpression in males (XO). The discovery that aariant
SMC4 type protein (DPY-27) is an essentialegulator of dosage
compensation through its associa-ion with the X chromosome provided
the first cluehat SMC proteins might be involved in this
processChuang et al., 1994, 1996). Subsequently, MIX-1as identified
as an SMC2-type protein required foroth mitosis and dosage
compensation, the re-tricted localization of which to the X
chromosomeas dependent on DPY-27 (Lieb et al., 1998). Like
he 13S condensin complex, the dosage compensa-ion complex
consists of an SMC2/4 heterodimer andt least two non-SMC subunits,
including DPY-26nd DPY-28 (Hirano, 1999). The mitotic function
ofIX-1 is achieved through its association with aore conventional
SMC4-type protein, suggesting
hat MIX-1 may have been enlisted to the dosageondensation
complex through the evolution ofPY-27 as a highly specialized SMC
protein, altering
he higher order structure of X chromosomes by aechanism perhaps
related to that underlying chro-osome condensation.
SMCs AND SISTER CHROMATID COHESION
Another aspect of mitotic chromosome dynamics
n which the eukaryotic SMC proteins play a funda-
-
mssw1fia(fiqMisarfc(epwteSwaactcsttkRt1
dsmsaeacFcasltfaiD
mmstmsatlaqcP1taA(ldfaitmatXalb1
ptfgets(mtseiltTftr
mt
133REVIEW: STRUCTURAL MAINTENANCE OF CHROMOSOMES
ental role is the establishment and maintenance ofister
chromatid cohesion. The first SMC moleculehown to have a vital role
in maintaining cohesionas Smc1p (Larionov et al., 1985; Strunnikov
et al.,993). Its Smc3p partner was subsequently identi-ed in a
genetic screen for S. cerevisiae mutants withpremature sister
chromatid separation phenotype
Michaelis et al., 1997). The same screen also identi-ed Scc1p
and Scc2p, two non-SMC proteins re-uired for sister chromatid
cohesion. Meanwhile,cd1p (identical to Scc1p) was also identified
in
ndependent screens for high-copy suppressors of anmc1 mutant or
mutants displaying enhanced invi-bility after mitotic arrest
(Guacci et al., 1997). Theole of this protein in chromosome
segregation wasurther verified by the heightened instability
ofircular minichromosomes in an scc1/mcd1 mutantHeo et al., 1998).
The functional association ofither Scc1p/Mcd1p or Scc3p (another
essential com-onent of the cohesin complex) with chromosomesas then
found to depend on each other and also on
he presence of Smc1p, Smc3p, and Scc2p (Michaelist al., 1997;
Tóth et al., 1999). In particular, themc1p, Smc3p, Scc1p/Mcd1p,
and Scc3p proteinsere found to co-immunoprecipitate in roughly
equalmounts, implying that they maintained cohesion ascomplex
(Guacci et al., 1997; Tóth et al., 1999),
oined ‘‘cohesin.’’ By contrast, Scc2p does not appearo be a
stoichiometric component of the cohesinomplex and fails to
colocalize with other cohesinubunits on chromosomes but is
nonetheless essen-ial for the binding of these other subunits to
chroma-in (Tóth et al., 1999). Its orthologue in S. pombe,nown as
Mis4, failed to coprecipitate with eitherad21p (the S. pombe
orthologue of Scc1p/Mcd1p) or
he S. pombe orthologue of Smc3p (Furuya et al.,998).The need to
establish cohesion during S phase was
emonstrated in cells expressing Scc1p/Mcd1p exclu-ively from a
galactose-inducible promoter (Uhl-ann and Nasmyth, 1998).
Ctf7p/Eco1p was later
hown to be another S. cerevisiae protein whosectivity is
essential to establish cohesion along thentire length of the
chromosome during S phase,lthough it was not required for the
maintenance ofohesion (Skibbens et al., 1999; Tóth et al.,
1999).urthermore, Ctf7p/Eco1p does not seem to be aohesin subunit
and its presence has no effect on thessociation of the cohesin
complex with chromo-omes (Tóth et al., 1999). Interestingly, a
syntheticethal interaction was observed between a
tempera-ure-sensitive mutation in ctf7 and the yeast genesor either
PCNA or an RFC-like protein (Skibbens etl., 1999), suggesting that
PCNA might be involvedn loading the cohesin complex onto chromatin
after
NA replication. It has been suggested that PCNA p
ay play a role in the assembly of chromatin (Kel-an, 1997) as
its nuclear distribution in fertilized
tarfish eggs coincided with the chromatin distribu-ion during
the first S phase (Nomura, 1994) andutations in the Drosophila gene
encoding PCNA
uppressed position-effect variegation (Henderson etl., 1994). In
addition, PCNA has been shown to bindhe largest subunit of CAF-1,
demonstrating a directink between replication machinery and
chromatinssembly (Shibahara and Stillman, 1999). Conse-uently, it
was proposed that the loading of cohesinomplexes onto chromatin
might be coupled withCNA-dependent DNA replication (Skibbens et
al.,999). This adds weight to a previous hypothesishat cohesion
might be directly coupled to passage of
replication fork (Uhlmann and Nasmyth, 1998).lthough the murine
orthologue of Scc1p/Mcd1p
known as PW29) and PCNA fail to exhibit similarocalization
patterns (Darwiche et al., 1999), thisoes not necessarily preclude
a role for PCNA inacilitating the loading of cohesins onto
chromatin,s Scc1p/Mcd1p in S. cerevisiae also does not colocal-ze
with Scc2p, even though the latter is essential forhe efficient
binding of Scc1p and other cohesinembers to chromatin (Tóth et
al., 1999). Similarly,
lthough the binding of cohesin subunits to chroma-in occurred
independently of DNA replication inenopus oocyte extracts (Losada
et al., 1998), thislso does not prevent a possible role for PCNA as
aanding pad for SMC proteins as PCNA may alsoind to DNA at times
other than S phase (Nomura,994).The similarities between the S.
cerevisiae and S.
ombe proteins involved in regulating sister chroma-id cohesion,
combined with the conservation ofactors required for chromosome
condensation, sug-ested that the cohesion mechanism might also
bevolutionarily conserved. Indeed, the vertebrate or-hologues of
SMC1, SMC3, and SCC1 have also beenhown to be essential for sister
chromatid cohesionLosada et al., 1998) and proper progression
ofetaphase (Schmiesing et al., 1998), even though
hese proteins appear to dissociate from chromo-omes during
mitosis (Darwiche et al., 1999; Losadat al., 1998; Schmiesing et
al., 1998). Nevertheless,mmunoblotting showed that the murine
ortho-ogues of both SMC1 and SMC3 were expressedhroughout the cell
cycle (Darwiche et al., 1999).his dissociation of the vertebrate
cohesin complex
rom mitotic chromosomes contrasts with the pat-ern observed in
yeast, in which Smc1p and Smc3pemain associated (Michaelis et al.,
1997).Although it is possible that cohesion at this stageight be
supported by other molecules (as yet uniden-
ified), it has been shown that hSMC1 nonetheless
lays a role in the maintenance of chromatid cohe-
-
spsmmewqtsmtoadatrTmltpe((oicc
chmtbAthS(stMbd
cbcbnDr
1bssiasnd1weKqrn1SD1ictts(sDageafarli
cin1mtqtpcstfpsbt
134 REVIEW: COBBE AND HECK
ion as well as its establishment, even though therotein appears
to be excluded from the chromo-omes during mitosis. This was
demonstrated by theitotic arrest of HeLa cells microinjected
duringid/late metaphase with an antibody specific for
ither the middle or C-terminal regions of hSMC1,hereas cells
injected at early anaphase subse-uently went through cytokinesis
normally to yieldwo daughter cells (Schmiesing et al., 1998).
Thisuggests that a residual level of cohesins bound toetaphase
chromosomes may be sufficient to main-
ain cohesion between sister chromatids until thenset of
anaphase, as previously proposed (Losada etl., 1998). Moreover, it
has been proposed that theissociation of most cohesins from the
chromosomest the onset of mitosis in vertebrate cells may loosenhe
linkage between sister chromatids, permittingeorganization of the
chromatin (Losada et al., 1998).his may serve to relieve a steric
barrier whichight otherwise prevent final condensation in such
arge chromosomes, as mediated by replacement ofhe cohesins by
the condensins. This idea is sup-orted by the prevention of
interphase cells fromntering mitosis by overexpression of an
PW29SCC1)–GFP fusion protein in mouse fibroblastsDarwiche et al.,
1999). Transfection with H2B–GFPr the GFP molecule itself produced
no such arrest,mplying that the SCC1 (PW29) protein and itsomplex
with SMC proteins might be involved in theontrol of mitotic cycle
progression.So what function do the SMC proteins fulfill in the
ohesin complex? Assuming that the SMC1/SMC3eterodimers of
cohesin complexes function as ATP-odulated DNA cross-linkers, it
has been suggested
hat these molecules may form intermolecular bridgesetween
separate DNA molecules (Hirano, 1999).lternatively, such bridges
might be produced
hrough the association of two different Smc1/3eterodimers
(possibly mediated by Scc1p/Mcd1p orcc3p), each of which is bound
to a single chromatid
Losada et al., 1998). This latter model seems plau-ible for S.
cerevisiae, as dissolution of sister chroma-id cohesion can be
achieved by cleavage of Scc1p/cd1p (Ciosk et al., 1998; Uhlmann et
al., 1999) and
oth Smc1p and Smc3p persist after Scc1p/Mcd1pissociation (Tanaka
et al., 1999).
LOCALIZATION OF THE COHESIN COMPLEX
Having established the importance of the cohesinomplex for
sister chromatid cohesion, where does itind on the chromosome? The
relative distribution ofohesins along chromosomes was initially
monitoredy modifying existing protocols for chromatin
immu-oprecipitation (ChIP) in S. cerevisiae, in which theNA
immunoprecipitated with cohesin subunits was
adiolabeled for use as a probe (Blat and Kleckner, t
999). These probes were then hybridized to a mem-rane containing
an array of PCR-generated chromo-ome fragments, covering the entire
length of chromo-ome III.Amajority of cohesin binding sites
identifiedn this way were associated with the centromere,lthough
the complex was also shown to bind specificites along the
chromosome arms (Blat and Kleck-er, 1999; Tanaka et al., 1999),
consistent with theiscrete foci seen in chromosome spreads (Tóth
et al.,999). These binding sites were found to correlateith locally
AT-rich sequences, occurring roughlyvery ,15 kb along the
chromosome (Blat andleckner, 1999). This preference for AT-rich
se-uences in centromeric regions was further corrobo-ated by
conventional ChIP analysis, using the immu-oprecipitated DNA as a
PCR template (Megee et al.,999). Although such regions are
reminiscent ofARs, no correlation with the redundant motifs
ofrosophila SARs was observed (Blat and Kleckner,999). However,
this does not preclude the possibil-ty that SARs might indeed be
binding sites forohesins, bearing in mind the differences betweenhe
short, defined centromeres of S. cerevisiae andhe longer regional
centromeres of other eukaryotesuch as S. pombe and Drosophila
melanogasterPluta et al., 1995). Indeed, it was previously ob-erved
that SMC proteins bind preferentially torosophila SARs as well as
sequences containinglternating poly(dA–dT) and yeast centromere
re-ions (Akhmedov et al., 1998). Interestingly, a moreven
distribution of cohesin binding in hydroxyurea-rrested cells
suggested that cohesins bind uni-ormly to chromosomes at the start
of S phase (Blatnd Kleckner, 1999) but relocate to
centromericegions later during the cell cycle so the highestevels
of centromere-bound Scc1p/Mcd1p were seenn cells arrested in M
phase (Megee et al., 1999).
The minimal centromere sequences required forohesin association
were then deduced by artificiallynserting sequences from CEN6 into
a region whichormally has low cohesin affinity (Tanaka et al.,999).
Normally the insertion of additional centro-eres in this manner
would create unstable dicen-
ric chromosomes, so the inserted centromeric se-uences were
conditionally suppressed by placinghem under the control of a
galactose-inducibleromoter. In this way it was deduced that 130
bpontaining CDEI–II–III was sufficient to confer cohe-in binding to
this sequence, whereas cohesin associa-ion with CEN DNA was
abolished by transcriptionrom the GAL promoter. The CDEIII sequence
inarticular was shown to be sufficient for weak cohe-in
association, but could be enhanced by adding 21p of CDEII (Tanaka
et al., 1999). Moreover, associa-ion was abolished by various
CDEIII point muta-
ions, further supporting the importance of this
-
ean(wtqcsciapspwc
asdaaStdtta(idsMotA1cdv
Cigpniocparc
dtwFrcoefwMdreeSosmeMptro
Ssd(dStarXtPnStowmodaIede
135REVIEW: STRUCTURAL MAINTENANCE OF CHROMOSOMES
lement for cohesin binding (Tanaka et al., 1999), ingreement
with previous findings that CDEIII wasecessary but insufficient for
functional cohesionMegee and Koshland, 1999). Despite this,
cohesinsere shown to differ from known centromere pro-
eins by associating strongly with adjacent se-uences as well as
the centromere itself, in someases with even higher affinity for
these flankingequences (Megee et al., 1999). As ectopically
placedentromeres were shown to direct Scc1p/Mcd1p bind-ng to
adjacent sequences which normally had lowffinity for the protein
(Megee et al., 1999), it is thereforeossible that the centromere
primarily contributes toister chromatid cohesion by directing the
cohesin com-lex to AT-rich sequences in the immediate vicinity.
Itill be interesting to see whether similar patterns of
ohesin association are found in other organisms.
THE SEPARATION OF SISTERS
How is the cohesin-mediated cohesion dissolved tollow separation
of sister chromatids during mito-is? Both Scc1p/Mcd1p and Scc3p
were shown toissociate from chromosomes at the metaphase tonaphase
transition (Michaelis et al., 1997; Tóth etl., 1999), leaving
behind the bulk of Smc1p andmc3p cohesin subunits which associate
longer withhe chromosomes (Tanaka et al., 1999). Although
theissociation of Scc1p/Mcd1p was known to depend onhe anaphase
promoting complex (APC), this dissocia-ion could be prevented by
expression of a nondegrad-ble version of Pds1p but not by other APC
inhibitorsCohen-Fix et al., 1996; Michaelis et al., 1997).
Thismplied that ubiquitination by the APC was notirectly
responsible for Scc1p/Mcd1p destruction anduggested that Pds1p
might somehow prevent Scc1p/cd1p dissociation. This was confirmed
by deletion
f the pds1 gene, allowing sister chromatid separa-ion and
Scc1p/Mcd1p dissociation in the absence ofPC function (Ciosk et
al., 1998; Yamamoto et al.,996). However, the premature separation
of sisterhromatids in scc1/mcd1 mutants contrasted with theelayed
separation in pds1 mutants, inferring the in-olvement of other
factors in the control of anaphase.Additional insights were
provided by studies ofut2p, an S. pombe orthologue of the
anaphase
nhibitor Pds1p. This protein was shown to be de-raded by the APC
and copurified with Cut1p, arotein maintained throughout the cell
cycle (Fu-abiki et al., 1996a,b, 1997). Similarly, Pds1p was
mmunoprecipitated with Esp1p (the S. cerevisiaerthologue of
Cut1), a protein required for sisterhromatid separation (Ciosk et
al., 1998). As the tworoteins in the complex had antagonistic
effects onnaphase, the Esp1p and Cut1p were named ‘‘sepa-ins’’
while their inhibitors (Cut2p and Pds1p) were
alled ‘‘securins.’’ Moreover, since Pds1p was not i
egraded by Esp1p, it appeared that securins inhibithe separins
by binding to them but this inhibitionas removed by APC-mediated
proteolysis (Cohen-ix et al., 1996). The activity of Esp1p was
thenevealed when its overexpression permitted sisterhromatid
separation in the presence of Pds1p (Ci-sk et al., 1998). As sister
chromatid separation insp1 mutants was also shown to be prevented
by aailure of Scc1p/Mcd1p degradation, the separinsere therefore
considered to be responsible for Scc1p/cd1p dissociation. This role
was confirmed by
emonstrating that Scc1p/Mcd1p cleavage and itsesultant
dissociation from chromatin occurred inxtracts from cells
overexpressing Esp1p but not insp1 mutant extracts (Uhlmann et al.,
1999). Thecc1p/Mcd1p cleavage sites were then identified
andverexpression of a cleavage-resistant protein washown to prevent
sister chromatid separation (Uhl-ann et al., 1999). In conclusion,
the available
vidence seems to suggest that dissociation of Scc1p/cd1p from
sister chromatids in S. cerevisiae de-
ends on cleavage mediated by Esp1p, which isransported to the
spindle by its inhibitor Pds1p butemains inactive until the APC
triggers proteolysisf Pds1p at the metaphase–anaphase transition.As
the ability of S. cerevisiae cell extracts to cleave
cc1p/Mcd1p correlated with the levels of Esp1p, theimplest
explanation is that Esp1p is the proteaseirectly responsible for
Scc1p/Mcd1p degradationUhlmann et al., 1999). However, until it can
beemonstrated that purified Esp1p is sufficient forcc1p/Mcd1p
cleavage in vitro, one cannot excludehe alternative possibility
that this protein mightctivate another protease instead.
Furthermore, itemains to be seen if cleavage of the human andenopus
Rad21p orthologues can be suppressed by
he recently characterized functional homologue ofds1p in Xenopus
(Zou et al., 1999). Certainly, theeed to remove PW29 (the murine
orthologue ofcc1p/Mcd1p) to allow separation of sister chroma-ids
has been demonstrated by the metaphase arrestf mitotic cells when a
PW29–GFP fusion proteinas overexpressed (Darwiche et al., 1999).
However,urine PW29 protein levels appear constant through-
ut the cell cycle, whereas Scc1p/Mcd1p levels peakuring S phase
and decline thereafter (Darwiche etl., 1999; Guacci et al., 1997;
Michaelis et al., 1997).t therefore seems that anaphase occurs by a
differ-nt mechanism in vertebrates, involving not justegradation of
the cohesin complex but also itsxclusion from the chromosomes.
CONNECTIONS BETWEEN COHESIONAND CONDENSATION
Considering the involvement of complexes contain-
ng SMC proteins in both sister chromatid cohesion
-
atdtcdppttmstcdtpssi(aStctpmB1inc
hei(osdsecachrtoipspep
bt1pocXlMtscli
ipbSscLttdpopcaAettFinttrtS
idrMnir1rh
136 REVIEW: COBBE AND HECK
nd chromosome condensation, is there any struc-ural
interrelationship between these processes? Toate, none of the
non-SMC subunits have been foundo be shared between the cohesin and
condensinomplexes, which would seem to indicate indepen-ent
evolution of these two SMC-containing com-lexes. Clearly
condensation is a prerequisite forroper segregation of sister
chromatids, ensuringhat the entire chromosome is accurately
packagedo avoid such hazards as sister chromatid entangle-ent and
cleavage of trailing chromatin at cytokine-
is. A model depicting the deposition and activity ofhe cohesin
and condensin complexes during the cellycle is given in Fig. 3. In
S. cerevisiae it has beenemonstrated by FISH that chromosome
condensa-ion also depends on the cohesin subunit Scc1p/Mcd1rotein
(Guacci et al., 1997). Similarly, the dispersedtaining of nuclear
material and stretched chromo-omes in S. pombe rad21 mutants may
reflect itsnvolvement in chromosome condensation as wellBirkenbihl
and Subramani, 1995; Tatebayashi etl., 1998). It has been proposed
that placement of thecc1p/Mcd1 protein at the newly replicated
chroma-ids provides an attachment site for recruitment
ofondensation proteins, suggesting a possible explana-ion for the
mirror symmetrical, helically foldedattern often observed in the
condensed sister chro-atids of vertebrate cells (Baumgartner et
al., 1991;oy de la Tour and Laemmli, 1988; Rattner and Lin,985).
Nevertheless, the condensation defects result-ng from mutations
affecting cohesin subunits areot as severe as those affecting
components of theondensin complex (Strunnikov et al., 1995).By
contrast, the vertebrate cohesins do not seem to
ave an effect on chromosome condensation (Losadat al., 1998) and
neither hCAP-E nor hCAP-C aremmunoprecipitated with either hSMC1 or
hSMC3Schmiesing et al., 1998). Furthermore, the bindingf condensins
to chromatin and chromosome conden-ation itself seem unaffected by
cohesin immuno-epletion in Xenopus egg extracts and
cohesinsimilarly bind to chromatin in condensin-depletedxtracts
(Losada et al., 1998). The differences inohesin and condensin
behavior between vertebratesnd yeast may reflect the relatively
small amount ofondensation occurring in yeast compared to that
inigher eukaryotes (Guacci et al., 1994). Thus, theoles of the
cohesin and condensin complexes appearo have become more
specialized in higher eukary-tes so that these complexes associate
and dissociatendependently. Furthermore, no
cell-cycle-specifichosphorylation has been observed for the
cohesinubunits in Xenopus, unlike the mitosis-specific
phos-horylation of condensin complex members (Losadat al., 1998).
On the other hand, the S. pombe Rad21
rotein is initially phosphorylated in G1/S, followed c
y hyperphosphorylation in G2 which is maintainedhroughout
mitosis (Birkenbihl and Subramani,995). Strikingly, the cohesin
complexes from Xeno-us egg extracts resemble condensins as they
alsoccur as two versions with different sedimentationoefficients,
in which the 9S form is a heterodimer ofSMC1 and XSMC3 while the
14S form contains at
east three additional subunits (including the Scc1p/cd1p
orthologue) (Losada et al., 1998). This struc-
ural similarity between cohesins and condensinstrongly suggests
that they may have evolved from aommon ancestor (albeit
independently), particu-arly considering that a single SMC protein
may benvolved in both processes in bacteria.
A further link between cohesion and condensationn budding yeast
is revealed by analysis of the Trf4rotein (topoisomerase I-related
function), whichinds to both Smclp and Smc2p (Castaño et al.,
1996;trunnikov et al., 1993) and is required for chromo-ome
segregation (Castaño et al., 1996) and rDNAhromosome condensation
(Castaño et al., 1996).ikewise, an additional link between the
condensa-
ion and cohesion machinery has been revealed byhe interaction of
either Smc1p or Smc2p withifferent coiled-coil domains of the human
HEC1rotein (highly expressed in cancer) and Tid3p, itsrthologue in
S. cerevisiae (Zheng et al., 1999). Thisrotein is required to
prevent haphazard sisterhromatid segregation in both organisms
(Chen etl., 1997; Zheng et al., 1999) and to repress theTPase
activity of the 26S proteasome subunit (Chent al., 1997),
suggesting possible roles in regulatinghe destruction of
Scc1p/Mcd1p or even controllinghe ATPase activity of SMC-containing
complexes.urthermore, as Tid3p was previously shown to
nteract with a protein required for meiotic recombi-ation and
synaptonemal complex formation (DMC1),his suggests a possible role
for HEC1 in extendinghe activities of SMC proteins to
recombinationepair during meiosis (Dresser et al., 1997).
Clearly,he biochemical effects of HEC1 interaction withMC proteins
demand further study.
COHESIN PROTEINS AND MEIOSIS
The subunits of the cohesin complex also sharemportant links
with proteins required for cohesionuring meiosis, as shown by the
essential meioticoles of Smc3p and Rec8p (a paralogue of
Scc1p/cd1p) (Klein et al., 1999; Parisi et al., 1999; Wata-
abe and Nurse, 1999). The rec8 gene was originallydentified in a
screen for S. pombe mutants witheduced meiotic recombination
(Ponticelli and Smith,989) and the encoded protein was shown to
beequired for sister chromatid cohesion and pairing ofomologous
chromosomes during meiosis I (Kraw-
huk et al., 1999; Krawchuk and Wahls, 1999; Mol-
-
nmrtWmrgtipcsiis1
ib(mcsSw1rscHtas
wmcnWpptbdrcstS1ta(Sp
ttIiCtt(a
Scr(mh(ibhammf
trSaRIeRdab1
sSamarcarRbstfis
137REVIEW: STRUCTURAL MAINTENANCE OF CHROMOSOMES
ar et al., 1995). Synthesis of Rec8p was specific toeiosis
(unlike other cohesins) and deletion of the
ec8 gene resulted in equational rather than reduc-ional
chromosome segregation (Lin et al., 1992;atanabe and Nurse, 1999).
Conversely, the replace-ent of Rad21p by Rec8p during mitosis
produced a
emarkable shift in the pattern of chromosome segre-ation from
equational to reductional. Nonetheless,he ability of the Rad21p
protein to rescue thenviability of rec8 mutant spores showed that
theseroteins share common functions. Moreover, as theohesin
cleavage sites appear to be uniquely con-erved between Scc1p/Mcd1p
in S. cerevisiae, Rad21pn S. pombe, and the Rec8 proteins in both
organ-sms, it seems likely that their cleavage may beimilarly
mediated by separins (Uhlmann et al.,999).The Rec8 protein was
originally believed to be
nvolved in early steps of meiotic recombination,ased on its
early appearance and disappearanceLin et al., 1992). In agreement
with this, rec8utants exhibited defective linear elements
(axial
ore-like structures formed in place of tripartiteynaptonemal
complexes during meiotic prophase in. pombe) which were shorter and
thicker than inild-type cells (Molnar et al., 1995; Parisi et
al.,999). Likewise, both Rec8p and Smc3p (which areequired for
meiotic chromatid cohesion in S. cerevi-iae) are essential for the
formation of synaptonemalomplexes and axial elements (Klein et al.,
1999).owever, neither protein is required for the forma-
ion of double-strand breaks, implying that Rec8pnd Smc3p are
needed to maintain cohesion so thatuch lesions may be repaired
(Klein et al., 1999).The S. pombe Rec8p protein was tightly
associatedith numerous chromosomal foci during prophase ofeiosis I
and was globally distributed around the
entromeric regions, whereas Rad21p was predomi-antly found near
the telomeres (Parisi et al., 1999;atanabe and Nurse, 1999). Just
as Rad21p is
hosphorylated in mitosis, Rec8p also underwenthosphorylation
from prophase onward. Althoughhe level of a Rec8–GFP fusion
detected by Westernlotting declined between the successive
meioticivisions, the protein persisted beyond meiosis I andemained
tightly associated with centromeric hetero-hromatin. Similarly,
Rec8p and Smc3p in S. cerevi-iae colocalized in a continuous line
along the longi-udinal axis of pachytene chromosome cores whilecc1p
was restricted to discrete foci (Klein et al.,999). The Rec8
protein levels were highest at theime of premeiotic DNA replication
but decreasedfter pachytene and disappeared after anaphase IIKlein
et al., 1999). Correspondingly, both Rec8p andmc3p disappeared from
the chromosome arms after
achytene but persisted near the centromeres after s
he separation of homologous chromosomes duringhe first meiotic
division, until anaphase of meiosisI (Klein et al., 1999).
Interestingly, the Rec8p local-zation pattern in both yeasts is
similar to that ofOR1 (a component of the lateral elements of
synap-
onemal complexes in rodent spermatocytes), sugges-ive of a role
in synaptonemal complex formationDobson et al., 1994; Lammers et
al., 1994; Yuan etl., 1998).It will be interesting to see if SMC3
and indeed
MC1 proteins are involved in maintaining meioticohesion in other
organisms, as suggested by a highate of expression in rodent
ovaries and testesShimizu et al., 1998; Stursberg et al., 1999).
Theeiotic function of cohesin proteins seems to be
ighly conserved among eukaryotes as the DIF1determinate,
infertile1) gene of Arabidopsis, encod-ng an orthologue of Rec8p,
was similarly shown toe essential for meiotic chromosome
segregation andence fertility (Bhatt et al., 1999). In addition,
thebility of the human Rec8 protein to partially comple-ent the
reduced spore viability of S. pombe rec8utants suggests at least
some conservation of
unction (Parisi et al., 1999).
SMC PROTEINS AND DNA REPAIR
Anumber of the proteins involved in sister chroma-id cohesion
have also been shown to play criticaloles in recombinational
repair. For example, bothMC1 and SMC3 may have a role during
interphases part of the bovine recombination protein complexC-1, in
which they are complexed with DNA ligase
II, DNA polymerase e, and a DNA structure-specificndonuclease
(Jessberger et al., 1996a,b). Similarly,ad21p of S. pombe was
implicated in the repair ofouble-strand DNA breaks in irradiated
cells inddition to being essential for mitotic growth (Birken-ihl
and Subramani, 1992, 1995; Tatebayashi et al.,998).However, a
specific role in DNA repair is demon-
trated by members of a further subgroup of theMC family, first
identified in S. pombe through thenalysis of rad18 mutants. A
temperature-sensitiveutant was shown to be hypersensitive to both
UV
nd g-irradiation and also exhibited reduced rates ofemoval of UV
photoproducts compared to wild-typeells. However, no significant
difference in endonucle-se activity was observed between extracts
fromad18 cells and wild-type cells, suggesting thatad18p might be
involved in repair of DNA damagey facilitating genetic
recombination. This was sub-equently confirmed by assaying the
ability of cellso repair double-stranded DNA breaks using pulse-eld
gel electrophoresis (Verkade et al., 1999). Corre-pondingly,
severely reduced levels of intrachromo-
omal homologous recombination were demonstrated
-
ir
ipercrstweroiocY
spTirtsipifcdwmeipfitaipmTtdlr
Sota
dW1sstsdStb(gfmtabFwplrim(whcbpiem
rbhbmcrfmiababttsrdm
138 REVIEW: COBBE AND HECK
n an Arabidopsis mim mutant (an orthologue ofad18) (Mengiste et
al., 1999).Immunostaining of cells revealed that the protein
s found in the chromatin compartment of the S.ombe nucleus, as
expected for a protein directlyngaged in DNA repair (Verkade et
al., 1999). Aad18 mutant was completely suppressed by excessopies
of brc1 (encoding a BRCT domain proteinequired for proper
chromosome condensation andegregation) but was synthetically lethal
in combina-ion with mutations in brc1, fin1 (encoding a kinasehich
induces chromatin condensation), or topoisom-rase II, suggesting
that the ability of Rad18p toepair DNA lesions might be related to
a role inrderly chromosome condensation. It will be interest-ng to
see whether Rad18p shares the ability of somether SMC proteins to
directly promote strand ex-hange (Jessberger et al., 1993, 1996a,b;
Sutani andanagida, 1997).The deletion of RAD18 and RHC18 (the S.
cerevi-
iae orthologue) showed that the gene is essential
forroliferation in both species (Lehmann et al., 1995).he mitotic
defect was further characterized by the
solation of an additional temperature-sensitivead18 mutant
(Verkade et al., 1999). After irradia-ion, many cells underwent
cytokinesis in the ab-ence of completed chromosome segregation,
result-ng in cells with nuclei stretched along the divisionlane and
bisected by septa. Such aberrant mitosesn spite of unrepaired DNA
lesions also implied a roleor Rad18p in maintaining the G2 DNA
damageheckpoint. This was verified by the failure of twoifferent
rad18/top2 double mutants to arrest in G2,hereas inhibition of
topoisomerase II activity nor-ally produces such an arrest
(Hartwell and Wein-
rt, 1989). However, Rad18p is not required fornitiation of the
checkpoint, as shown by normalhosphorylation of the Chk1 protein
kinase (thenal element in the signaling cascade activated byhe G2
DNA damage) in rad18 cultures (Verkade etl., 1999). It is therefore
possible that Rad18p activ-ty is induced posttranscriptionally by
this G2 check-oint and the continued activity of this
proteinaintains the arrest until the damage is repaired.hus, the
Rad18 subfamily of SMC proteins appearso have multiple functions in
response to DNAamage, signaling the persistence of unrepairedesions
in DNA and repairing them through a role inecombination-mediated
repair.
MOONLIGHTING IN THE BASEMENT MEMBRANE?
Perhaps the most surprising result concerningMC proteins to date
is a possible additional roleutside the cell. This idea is based on
the identifica-ion of an extracellular, secreted proteoglycan,
known
s bamacan (basement membrane-associated chon- t
roitin sulfate proteoglycan), as an SMC molecule.hen rat bamacan
was cloned (Wu and Couchman,
997), the authors noted that the sequence bore notructural
similarity with any chondroitin/dermatanulfate proteoglycan
reported at that time. However,hey noticed that bamacan and SMC
proteins fromeveral diverse organisms shared a similar five-omain
structure. Unfortunately, as none of theMC proteins used in their
comparison belonged tohe SMC3 subclass, the overall sequence
homology ofamacan to other SMC proteins was found to be loweven
when compared with those of vertebrate ori-in). The human
orthologue of SMC3 was cloned theollowing year and was initially
named HCAP (hu-an chromosome-associated polypeptide). The au-
hors commented that this SMC protein shared 98%mino acid
sequence identity with the published ratamacan protein sequence
(Shimizu et al., 1998).inally, the murine orthologue of the rat
bamacanas cloned and identified as a member of the SMC3rotein
subfamily, as the protein showed the sameevel of homology to the
bovine SMC3 as it did to theat bamacan (Ghiselli et al., 1999).
Meanwhile, anndependent group had succeeded in cloning the
urine homologue of SMC3, known as mSMCDDarwiche et al., 1999).
However, at the time ofriting both groups were seemingly unaware of
justow close bamacan really was to SMC3. When theDNA sequences for
the murine SMC3 and murineamacan are aligned, the corresponding
predictedrotein sequences are 100% identical. In other words,t
appears that the mouse SMC3 is the same mol-cule as a component of
the extracellular basementembrane.As the murine bamacan was cloned
by using the
at bamacan sequence to BLAST the dbEST dataase, the true
significance of the similarity hinges onow reliably the rat bamacan
was cloned. The ratamacan was originally isolated (Wu and Couch-an,
1997) by screening a rat yolk sac carcinoma
DNA expression library with a polyclonal antise-um raised
against a pool of purified proteoglycansrom the murine
Engelbreth–Holm–Swarm tumoratrix (Couchman et al., 1996). To
confirm the
dentity of this clone, rabbit antibodies were raisedgainst two
nonoverlapping fusion proteins encodedy subclones of the bamacan
cDNA and both of thesentibodies were shown to recognize the same
proteiny immunoblotting as the original antiserum. One ofhese
antisera also stained extracellular matrix inissue sections, as did
the original antibody. This istrong evidence that a protein better
known for itsole in chromosome mechanics has a very unpre-icted
extracellular localization. Antibodies to theurine SMC3 were
generated against a peptide in
he C-terminal ATP-binding domain (Darwiche et
-
amdmlcmpr
vtihnscibsTtaNmlmtmcsdpmmdhmctSiSqb
scaseebemN
hpmsodoia
bmtcrfbEtunlsp
pptocnnhViatcdTaDteassGdgdrm
139REVIEW: STRUCTURAL MAINTENANCE OF CHROMOSOMES
l., 1999); these antibodies demonstrated that theurine SMC3
binds to interphase chromatin and
issociates from it at the onset of mitosis, in agree-ent with
previous studies with the Xenopus ortho-
ogues (Losada et al., 1998). It would therefore berucial to see
whether these antibodies used onouse tissue sections independently
reproduce the
attern of staining observed with the antibodies toat
bamacan.Naturally, one might wonder what a protein in-
olved in chromosome dynamics might be doing inhe basement
membrane, outside the cell. Althought is not unheard of for the
same protein molecule toave more than one distinct action (Jeffery,
1999),one of the previously recognized functions of SMC3how any
obvious connection with a role outside theell. However, one clue to
how SMC proteins could benvolved in stabilizing the extracellular
matrix ofasement membranes may be provided by compari-on with the
laminins (Timpl and Brown, 1996).hese molecules also have
coiled-coil domains, inhis case mediating heterotrimerization
between the, b, and g chains. In addition, the globular
laminin-terminal domains mediate Ca21-dependent poly-erization to
yield quasihexagonal networks. These
aminin networks are finally anchored in the base-ent membrane by
integrin and dystroglycan recep-
ors. One may conjecture that SMCs might also formore complex
networks if they are secreted in suffi-
iently high concentration, forming chains as a re-ult of
interactions between the terminal ATPaseomains of adjacent
molecules. Alternatively, it isossible that secreted SMC3/bamacan
may fit into aatrix through chondroitin sulfate side chains, inuch
the same way that perlecan (a heparan sulfate/
ermatan sulfate proteoglycan) interacts with theeparan sulfate
binding site in the C-terminal LGodules of laminin a chains. An
additional question
oncerns how an SMC protein might actually reachhe basement
membrane. Sequence analysis of theMC proteins using available data
for nuclear local-
zation signals (NLS) and NES reveals that eachMC may contain
potential NLS and NES se-uences, of which at least one candidate
NES haseen shown to be functional (Sutani et al., 1999).Indeed,
SMC3 is not the only nuclear protein that
eems to be playing an additional role outside theell. For
example, histone H1 has been shown to acts a binding protein for
thyroglobulin at the cellurface of macrophages, mediating
thyroglobulinndocytosis (Brix et al., 1998), while titin, a
constitu-nt of muscle sarcomeres, has also been proposed toe a
component of Drosophila chromosomes (Machadot al., 1998). Another
recently identified basementembrane-associated proteoglycan with
candidate
ES and NLS motifs, known as leprecan (Wassen- h
ove-McCarthy and McCarthy, 1999), may possiblylay a role in
chromosomal dynamics as it sharesore than 36% identity and 43%
similarity with the
ynaptonemal complex protein SC56 along a stretchf 343 amino
acids. In conclusion, the surprisingiscovery that an SMC protein
may have a roleutside the cell, quite apart from a fundamental
rolen various aspects of chromosomal dynamics, adds togrowing list
of moonlighting proteins.
FUTURE PROSPECTS
The understanding of chromosome structure andehavior has been
greatly enriched by the findingsade over the past few years. It is
already clear that
he SMCs are important for chromosome cohesion,hromosome
condensation, dosage compensation, andecombination repair. The
original eukaryotic sub-amilies of SMC1, SMC2, SMC3, and SMC4
haveeen joined by a fifth branch, the Rad18 subfamily.ven more
exciting is the discovery and analysis of
he single SMC within prokaryotes, leading to thendeniable
conclusion that the SMCs are conservedot only in structure but also
in function. The evo-
ution of the single SMC to a family constituting fiveubfamilies
is certainly a matter of intrigue, as is theossible extracellular
existence of an SMC protein.The Xenopus in vitro extract system,
coupled with
owerful genetics in S. cerevisiae and S. pombe, hasredominantly
contributed to the identification ofhese molecules and their
associated proteins. Studyf the SMC proteins and the complexes and
pro-esses in which they take part has not only illumi-ated the
significant degree to which certain compo-ents and mechanisms are
conserved, but alsoighlighted provocative questions for future
study.ery likely, differences between single-celled organ-
sms and multicellular creatures will be elucidatednd with time
clarified. What is clearly missing fromhe studies published to date
is an analysis of theseomponents in a multicellular organism
amenable toevelopmental, genetic, and cytological approaches.he
identification of the genes for SMCs and associ-ted proteins has
been greatly facilitated by therosophila genome project. This is
currently leading
o the identification of mutations in these genes andxploitation
of the ability to examine these proteinst different times of
development, in different tis-ues, in different types of cell
cycles (e.g., rapid,ynchronized early embryonic cycles lacking G1
and2 phases versus more normal cell cycles), and inifferent types
of chromosomes (e.g., diploid versusiant, banded polytene
chromosomes). There is nooubt that the future years will be as rich
foresearch and progress into understanding the funda-ental
questions of chromosome structure and be-
avior as the past years have been.
-
iiCopBa
A
A
B
B
B
B
B
B
B
B
B
B
B
B
B
C
C
C
C
C
C
C
C
C
C
C
D
D
D
E
E
F
140 REVIEW: COBBE AND HECK
We thank members of the Heck and Earnshaw labs for stimulat-ng
discussions concerning chromosome organization and dynam-cs. We
also acknowledge Soren Steffensen, Paola Coelho, andlaudio Sunkel
at the University of Porto with whom we have anngoing, fruitful
collaboration into the analysis of SMCs andartners in fruit flies.
M.M.S.H. is a Senior Research Fellow in theasic Biomedical
Sciences, funded by the Wellcome Trust. N.C. isPh.D. student
supported by a Darwin Trust Prize Studentship.
REFERENCES
dachi, Y., Luke, M., and Laemmli, U. K. (1991)
Chromosomeassembly in vitro: Topoisomerase II is required for
condensa-tion, Cell 64, 137–148.
khmedov, A. T., Frei, C., Tsai-Pflugfelder, M., Kemper,
B.,Gasser, S. M., and Jessberger, R. (1998) Structural mainte-nance
of chromosomes protein C-terminal domains bind prefer-entially to
DNA with secondary structure, J. Biol. Chem.
273,24088–24094.aumgartner, M., Dutrillaux, B., Lemieux, N.,
Lilienbaum, A.,Paulin, D., and Viegas-Pequignot, E. (1991) Genes
occupy afixed and symmetrical position on sister chromatids, Cell
64,761–766.errios, M., Osheroff, N., and Fisher, P. A. (1985) In
situ localiza-tion of DNA topoisomerase II, a major polypeptide
component ofthe Drosophila nuclear matrix fraction, Proc. Natl.
Acad. Sci.USA 82, 4142–4146.hat, M. A., Philp, A. V., Glover, D.
M., and Bellen, H. J. (1996)Chromatid segregation at anaphase
requires the barren prod-uct, a novel chromosome-associated protein
that interacts withtopoisomerase II, Cell 87, 1103–1114.
hatt, A. M., Lister, C., Page, T., Fransz, P., Findlay, K.,
Jones,G. H., Dickinson, H. G., and Dean, C. (1999) The DIF1 gene
ofArabidopsis is required for meiotic chromosome segregationand
belongs to the REC8/RAD21 cohesin gene family, Plant J.19,
463–472.
irkenbihl, R. P., and Subramani, S. (1992) Cloning and
character-ization of rad21, an essential gene of
Schizosaccharomycespombe involved in DNA double-strand-break
repair, NucleicAcids Res. 20, 6605–6611.irkenbihl, R. P., and
Subramani, S. (1995) The rad21 geneproduct of Schizosaccharomyces
pombe is a nuclear, cell cycle-regulated phosphoprotein, J. Biol.
Chem. 270, 7703–7711.
lat, Y., and Kleckner, N. (1999) Cohesins bind to
preferentialsites along yeast chromosome III, with differential
regulationalong arms versus the centric region, Cell 98,
249–259.
ode, J., Kohwi, Y., Dickinson, L., Joh, T., Klehr, D., Mielke,
C.,and Kohwi-Shigematsu, T. (1992) Biological significance
ofunwinding capability of nuclear matrix-associating DNAs, Sci-ence
255, 195–197.
oy de la Tour, E., and Laemmli, U. K. (1988) The
metaphasescaffold is helically folded: Sister chromatids have
predomi-nantly opposite helical handedness, Cell 55, 937–944.
radbury, E. M., Inglis, R. J., and Matthews, H. R. (1974)
Controlof cell division by very lysine rich histone (F1)
phosphorylation,Nature 247, 257–261.radbury, E. M., Inglis, R. J.,
Matthews, H. R., and Langan, T. A.(1974) Molecular basis of control
of mitotic cell division ineukaryotes, Nature 249, 553–556.
ritton, R.A., Lin, D. C., and Grossman,A. D. (1998)
Characteriza-tion of a prokaryotic SMC protein involved in
chromosomepartitioning, Genes Dev. 12, 1254–1259.
rix, K., Summa, W., Lottspeich, F., and Herzog, V.
(1998)Extracellularly occurring histone H1 mediates the binding
ofthyroglobulin to the cell surface of mouse macrophages, J.
Clin.Invest. 102, 283–293.
arromolino, L., Lee, B., Zaballos, A., Peled, A., Barthelemy,
I., F
Shav-Tal, Y., Prieto, I., Carmi, P., Gothelf, Y., Gonzalez
deBuitrago, G., Aracil, M., Marquez, G., Barbero, J., and Zipori,
D.(1997) SA-1, a nuclear protein encoded by one member of a
novelgene family: Molecular cloning and detection in
hemopoieticorgans. Gene 195, 151–159.astaño, I. B., Brzoska, P.
M., Sadoff, B. U., Chen, H., andChristman, M. F. (1996) Mitotic
chromosome condensation inthe rDNA requires TRF4 and DNA
topoisomerase I in Saccharo-myces cerevisiae, Genes Dev. 10,
2564–2576.astaño, I. B., Heath-Pagliuso, S., Sadoff, B. U.,
Fitzhugh, D. J.,and Christman, M. F. (1996) A novel family of TRF
(DNAtopoisomerase I-related function) genes required for
propernuclear segregation, Nucleic Acids Res. 24, 2404–2410.hen,
Y., Riley, D. J., Chen, P. L., and Lee, W. H. (1997) HEC, anovel
nuclear protein rich in leucine heptad repeats specificallyinvolved
in mitosis, Mol. Cell. Biol. 17, 6049–6056.hen, Y., Sharp, Z. D.,
and Lee, W. H. (1997) HEC binds to theseventh regulatory subunit of
the 26S proteasome and modu-lates the proteolysis of mitotic
cyclins, J. Biol. Chem. 272,24081–24087.huang, P. T., Albertson, D.
G., and Meyer, B. J. (1994) DPY-27: Achromosome condensation
protein homolog that regulates C.elegans dosage compensation
through association with the Xchromosome, Cell 79, 459–474.huang,
P. T., Lieb, J. D., and Meyer, B. J. (1996) Sex-specificassembly of
a dosage compensation complex on the nematode Xchromosome, Science
274, 1736–1739.iosk, R., Zachariae, W., Michaelis, C., Shevchenko,
A., Mann, M.,and Nasmyth, K. (1998) An ESP1/PDS1 complex regulates
lossof sister chromatid cohesion at the metaphase to
anaphasetransition in yeast, Cell 93, 1067–1076.ohen-Fix, O.,
Peters, J. M., Kirschner, M. W., and Koshland, D.(1996) Anaphase
initiation in Saccharomyces cerevisiae is con-trolled by the
APC-dependent degradation of the anaphaseinhibitor Pds1p, Genes
Dev. 10, 3081–3093.onnelly, J. C., Kirkham, L. A., and Leach, D. R.
(1998) TheSbcCD nuclease of Escherichia coli is a structural
maintenanceof chromosomes (SMC) family protein that cleaves
hairpinDNA, Proc. Natl. Acad. Sci. USA 95, 7969–7974.ouchman, J.
R., Kapoor, R., Sthanam, M., and Wu, R. R. (1996)Perlecan and
basement membrane–chondroitin sulfate proteo-glycan (bamacan) are
two basement membrane chondroitin/dermatan sulfate proteoglycans in
the Engelbreth–Holm–Swarm tumor matrix, J. Biol. Chem. 271,
9595–9602.arwiche, N., Freeman, L. A., and Strunnikov, A. (1999)
Charac-terization of the components of the putative mammalian
sisterchromatid cohesion complex, Gene 233, 39–47.obson, M. J.,
Pearlman, R. E., Karaiskakis, A., Spyropoulos, B.,and Moens, P. B.
(1994) Synaptonemal complex proteins: Occur-rence, epitope mapping
and chromosome disjunction, J. CellSci. 107, 2749–2760.resser, M.
E., Ewing, D. J., Conrad, M. N., Dominguez, A. M.,Barstead, R.,
Jiang, H., and Kodadek, T. (1997) DMC1 functionsin a Saccharomyces
cerevisiae meiotic pathway that is largelyindependent of the RAD51
pathway, Genetics 147, 533–544.arnshaw, W. C., Halligan, B., Cooke,
C. A., Heck, M. M. S., andLiu, L. F. (1985) Topoisomerase II is a
structural component ofmitotic chromosome scaffolds, J. Cell Biol.
100, 1706–1715.
arnshaw, W. C., and Heck, M. M. S. (1985) Localization
oftopoisomerase II in mitotic chromosomes, J. Cell Biol.
100,1716–1725.
unabiki, H., Kumada, K., and Yanagida, M. (1996