RESEARCH ARTICLE SUMMARY
◥
CELL BIOLOGY
The proteasome controls ESCRT-III–mediated cell
division in an archaeon
Gabriel Tarrason Risa*, Fredrik Hurtig*, Sian Bray, Anne E. Hafner, Lena Harker-Kirschneck,
Peter Faull, Colin Davis, Dimitra Papatziamou, Delyan R. Mutavchiev, Catherine Fan,
Leticia Meneguello, Andre Arashiro Pulschen, Gautam Dey, Siân Culley, Mairi Kilkenny,
Diorge P. Souza, Luca Pellegrini, Robertus A. M. de Bruin, Ricardo Henriques, Ambrosius P. Snijders,
Anđela Šarić, Ann-Christin Lindås, Nicholas P. Robinson†, Buzz Baum†
INTRODUCTION: Eukaryotes likely arose from a
symbiotic partnership between an archaeal
host and an alpha-proteobacterium, giving
rise to the cell body and the mitochondria, re-
spectively. Because of this, a number of pro-
teins controlling key events in the eukaryotic
cell division cycle have their origins in archaea.
These include ESCRT-III proteins, which cat-
alyze the final step of cytokinesis in many
eukaryotes and in the archaeon Sulfolobus
acidocaldarius. However, to date, no archaeon
has been found that harbors homologs of cell
cycle regulators, like cyclin-dependent kinases
and cyclins, which order events in the cell cycle
across all eukaryotes. Thus, it remains uncer-
tain how key events in the archaeal cell cycle,
including division, are regulated.
RATIONALE: An exception to this is the 20S pro-
teasome, which is conserved between archaea
and eukaryotes and which regulates the eu-
karyotic cell cycle through the degradation of
cyclins. To explore the function of the 20S pro-
teasome in the archaeon S. acidocaldarius, we
determined its structure by crystallography
and carried out in vitro biochemical analyses
of its activity with and without inhibition. The
impact of proteasome inhibition on cell divi-
sion and cell cycle progression was examined
in vivo by flow cytometry and super-resolution
microscopy. Following up with mass spectrom-
etry, we identified proteins degraded by the
proteasome during division. Finally, we used
molecular dynamics simulations to model the
mechanics of this process.
RESULTS: Here, we present a structure of the
20S proteasome of S. acidocaldarius to a reso-
lution of 3.7 Å, which we used to model its sen-
sitivity to the eukaryotic inhibitor bortezomib.
When this inhibitor was added to synchronous
cultures, it was found to arrest cells mid-
division, with a stable ESCRT-III division ring
positioned at the cell center between the two
separated and prereplicative nucleoids. Pro-
teomics was then used to identify a single
archaeal ESCRT-III homolog, CdvB, as a key
target of the proteasome that must be de-
graded to enable division to proceed.
Examining the localization patterns of CdvB
and two other archaeal ESCRT-III homologs,
CdvB1 and CdvB2, by flow cytometry and super-
resolution microscopy revealed the sequence
of events that leads to division. First, a CdvB
ring is assembled. This CdvB ring then templates
the assembly of the contractile ESCRT-III homo-
logs, CdvB1 and CdvB2, to form a composite
division ring. Cell division is then triggered by
proteasome-mediated degradation of CdvB,
which allows the CdvB1:CdvB2 copolymer to
constrict, pulling the membrane with it. During
constriction, the CdvB1:CdvB2 copolymer is
disassembled, thus vacating the membrane
neck to drive abscission, yielding two daugh-
ter cells with diffuse CdvB1 and CdvB2.
CONCLUSION: This study reveals a role for the
proteasome in driving structural changes in a
composite ESCRT-III copolymer, enabling the
stepwise assembly, disassembly, and contrac-
tion of an ESCRT-III–based division ring. Al-
though it is not yet clear how proteasomal
inhibition prevents S. acidocaldarius cells from
resetting the cell cycle to initiate the next S
phase, these data strengthen the case for the
eukaryotic cell cycle regulation having its ori-
gins in archaea.▪
RESEARCH
Tarrason Risa et al., Science 369, 642 (2020)
7 August 2020
1 of 1
The list of author affiliations is available in the full article online.
*These authors contributed equally to this work.
†Corresponding author. Email: b.baum@ucl.ac.uk (B.B.);
n.robinson2@lancaster.ac.uk (N.P.R.)
Cite this article as G. Tarrason Risa et al., Science 369,
eaaz2532 (2020). DOI: 10.1126/science.aaz2532
READ THE FULL ARTICLE AT
https://doi.org/10.1126/science.aaz2532
4  Constriction
of the tensed  
ESCRT-III polymer 
(CdvB1, CdvB2) to
reach its preferred
curvature.
1  Construction 
of a polymeric 
ESCRT-III (CdvB) 
division ring at 
the cell center.
6  Slow degradation 
of free, diffusable
(CdvB1, CdvB2).
2 Templated assembly 
of a composite ESCRT-III 
polymer (CdvB, CdvB1, 
CdvB2).
5  Disassembly of 
the ESCRT-III 
polymer (CdvB1, 
CdvB2) as it 
constricts, leading 
to membrane 
scission. 
3  Selective removal and 
proteasome-mediated 
degradation of one of the 
ESCRT-III subunits (CdvB).
Inhibiting the proteasome prevents the 
degradation of CdvB and arrests cell 
division and S phase initiation.
Model of ESCRT-III–mediated cell division in the archaeon S. acidocaldarius. The model shows the
sequential stages of the division process in S. acidocaldarius (labeled 1 to 6), together with the corresponding
stage-specific images. DNA is in blue, CdvB in purple, and CdvB1 and CdvB2 in green. The broken arrow
represents an extended period where cells progress through G1, S, and G2. Note that Vps4 (not shown) is
likely required for ESCRT-III polymer disassembly.
Downloaded from http://science.sciencemag.org/ on August 6, 2020

----!@#$NewPage!@#$----
RESEARCH ARTICLE
◥
CELL BIOLOGY
The proteasome controls ESCRT-III–mediated cell
division in an archaeon
Gabriel Tarrason Risa1*, Fredrik Hurtig2*, Sian Bray3, Anne E. Hafner1,4,5, Lena Harker-Kirschneck1,4,5,
Peter Faull6, Colin Davis6, Dimitra Papatziamou7, Delyan R. Mutavchiev1, Catherine Fan1,
Leticia Meneguello1, Andre Arashiro Pulschen1, Gautam Dey1, Siân Culley1, Mairi Kilkenny3,
Diorge P. Souza1, Luca Pellegrini3, Robertus A. M. de Bruin1, Ricardo Henriques1,
Ambrosius P. Snijders6, Anđela Šarić1,4,5, Ann-Christin Lindås2, Nicholas P. Robinson7†, Buzz Baum1,4†
Sulfolobus acidocaldarius is the closest experimentally tractable archaeal relative of eukaryotes and,
despite lacking obvious cyclin-dependent kinase and cyclin homologs, has an ordered eukaryote-like cell
cycle with distinct phases of DNA replication and division. Here, in exploring the mechanism of cell
division in S. acidocaldarius, we identify a role for the archaeal proteasome in regulating the transition
from the end of one cell cycle to the beginning of the next. Further, we identify the archaeal ESCRT-III
homolog, CdvB, as a key target of the proteasome and show that its degradation triggers division by
allowing constriction of the CdvB1:CdvB2 ESCRT-III division ring. These findings offer a minimal
mechanism for ESCRT-III–mediated membrane remodeling and point to a conserved role for the
proteasome in eukaryotic and archaeal cell cycle control.
T
he eukaryotic cell cycle is ordered by os-
cillations in the activity of a conserved
set of cyclin-dependent kinases. Although
both cyclins and cyclin-dependent kinases
have yet to be identified outside of eu-
karyotes, Sulfolobus acidocaldarius, a member
of the TACK (Thaumarchaeota, Aigarchaeota,
Crenarchaeota, and Korarchaeota) superphy-
lum of Archaea, possesses an ordered cell cycle
with distinct phases of DNA replication and
division, similar in structure to the cell cycle
observed in many eukaryotes (1). Several of
the enzymes used to drive key events in the
cell cycle are also conserved from archaea to
eukaryotes. For example, archaeal homologs of
eukaryotic Cdc6, Orc, MCM, and GINS proteins
initiate DNA replication at multiple origins in
S. acidocaldarius, just as they do in eukaryotes
(2, 3). Additionally, previous work shows that
homologs of ESCRT-III and the AAA+ adeno-
sine triphosphatase (ATPase) Vps4, proteins
that mediate abscission and other membrane
remodeling processes in eukaryotes, play an
important role in TACK archaeal cell division (4).
These observations beg the question: Are any of
these common events regulated by conserved
elements of the cell cycle control machinery that
are shared across archaea and eukaryotes?
Although archaea lack clear homologs of both
cyclins and cyclin-dependent kinases, archaeal
homologs of the 20S core eukaryotic protea-
some are readily identifiable (5–8). In eukary-
otes, cell division is initiated by activation
of the ubiquitin-E3 ligase, APC, which triggers
proteasome-mediated degradation of cyclin B
and securin and likely a host of other proteins
(9), leading to irreversible mitotic exit (10), chro-
mosome segregation, and cytokinesis (11, 12).
This led us to test whether proteasome-mediated
degradation also plays a role in regulating the
cell cycle reset in TACK archaea (1).
Here, we report that proteasomal activity is
required for S. acidocaldarius cell division, using
inhibitors of the 20S proteasome. Proteomics
analyses identified a single archaeal ESCRT-III
homolog, CdvB (Saci_1373), as a key target for
proteasome-mediated degradation during the
final stages of the cell division cycle. CdvB is part
of the CdvABC operon (Saci_1374, Saci_1373, and
Saci_1372) and has previously been implicated in
S. acidocaldarius cell division (13–15). A combi-
nation of microscopy and flow cytometry data
further suggest that CdvB both templates the
assembly of a contractile ESCRT-III copolymer,
composed ofthe paralogs CdvB1(Saci_0451) and
CdvB2 (Saci_1416), and prevents their constric-
tion. As a consequence, proteasome-mediated
degradation of CdvB triggers cell division.
Results
Bortezomib inhibits the S. acidocaldarius proteasome
To determine a detailed structure of the S.
acidocaldarius proteasome, building on previ-
ous work (16), we coexpressed the a subunit
and N-terminally truncated b subunit of the
proteasome (Saci_0613 and Saci_0662DN). This
enabled us to produce diffracting crystals of
the catalytically inactive, 28-subunit 20S pro-
teasome, allowing us to determine its struc-
ture to a resolution of 3.7 Å [Fig. 1A; figs. S1,
A to C, and S2, A to C; and table S1; Protein
Data Bank (PDB) ID 6Z46]. Structures of the
equivalent complexes from the euryarchaeon
Archaeoglobus fulgidus (PDB ID 1J2Q) and
Saccharomyces cerevisiae (PDB ID 4NNN) were
then used to generate a homology model of
an N-terminally truncated catalytically active b
subunit (Saci_0909DN), which was docked into
the structure of the S. acidocaldarius pro-
teasome in place of one inactive b subunit
(Saci_0662 DN) (see methods) (17). It was then
possible to dock bortezomib (PS-341, Velcade),
an established small-molecule inhibitor of
the proteasomes of euryarchaeota and eukar-
yotes (18–20), into the model’s active site (Fig.
1B and fig. S1, D and E). Here, the boronic
acid moiety of bortezomib forms a bond with
the nucleophilic hydroxyl group of the active
site threonine, indicating that the small mol-
ecule is likely able to inhibit the proteolytic
activity of the S. acidocaldarius 20S proteasome
as it does in eukaryotes and in the euryarchaeon
Haloferax volcanii (20, 21).
To determine whether bortezomib can in-
hibit the S. acidocaldarius proteasome in vitro,
as suggested by this structural analysis, we
examined the ability of the active complex [con-
sisting of a subunits, together with N-terminally
truncated structural and catalytically active b
subunits(Saci_0613/Saci_0662DN/Saci_0909DN)]
to degrade a Urm1-tagged green fluorescent
protein (GFP) substrate, using a biochemical
setup we described previously (see methods) (16).
When this assay was carried out with different
concentrations of inhibitor, the rate of sub-
strate degradation was diminished (Fig. 1C
and fig. S3, A to D). We then mutated a con-
served valine residue (V49) that mediates
critical contacts between the active site and the
inhibitor (22) to a threonine (V49T). As expected,
this led to a reduction in the sensitivity of the
purified S. acidocaldarius 20S proteasome to
bortezomib (fig. S3, E and F), validating the
mode of inhibitor action.
Proteasomal inhibition arrests the cell cycle
of S. acidocaldarius
Having demonstrated that bortezomib in-
hibits the S. acidocaldarius 20S proteasome
activity in vitro, bortezomib was added to
S. acidocaldarius cultures to look for a possi-
ble role for the 20S proteasome in cell cycle
control. For this experiment, cells were syn-
chronized to G2 using acetic acid, released,
and observed by flow cytometry as they reen-
tered the cell cycle and divided (see methods)
(23). Notably, when bortezomib was added to
cultures 80 min after release from the acetic
acid arrest, the cells failed to divide (Fig. 1D and
fig. S4, A and B). Furthermore, this bortezomib-
RESEARCH
Tarrason Risa et al., Science 369, eaaz2532 (2020)
7 August 2020
1 of 8
1MRC-Laboratory for Molecular Cell Biology, University College
London (UCL), London, UK. 2Department of Molecular Biosciences,
The Wenner-Gren Institute, Stockholm University, Stockholm,
Sweden. 3Biochemistry Department, University of Cambridge,
Cambridge, UK. 4Institute for the Physics of Living Systems, UCL,
London, UK. 5Department of Physics and Astronomy, UCL,
London, UK. 6Proteomics Platform, The Francis Crick Institute,
London, UK. 7Faculty of Health and Medicine, Division of
Biomedical and Life Sciences, Lancaster University, Lancaster, UK.
*These authors contributed equally to this work.
†Corresponding author. Email: b.baum@ucl.ac.uk (B.B.);
n.robinson2@lancaster.ac.uk (N.P.R.)
Downloaded from http://science.sciencemag.org/ on August 6, 2020

----!@#$NewPage!@#$----
Tarrason Risa et al., Science 369, eaaz2532 (2020)
7 August 2020
2 of 8
Model of bortezomib bound to the
chymotryptic/active β-subunit (Saci0909ΔN)
ASP168
DMSO
Bortezomib
0
15
10
5
20
25
0
15
10
5
20
25
0
15
10
5
20
25
0
15
10
5
20
25
0
15
10
5
20
25
50
100
150
200
50
100
150
200
50
100
150
200
50
100
150
200
50
100
150
200
Cell count (10 3)
1N
2N
1N
2N
1N
2N
1N
2N
1N
2N
T = 80
T = 120
T = 160
Asynchronous
T = 200
EdU
&
bortezomib
EdU
&
DMSO
DNA
EdU
S-layer
EdU incorporation
T = 140
DNA
S-layer
Separated nucleoids
T = 120
Separated nucleoids
Bortezomib
DMSO
EdU
DNA
120
160
200
240
0
20
40
60
80
100
% EdU(+) cells
Time [minutes]
N = 3
n = 106
EdU & DMSO
EdU & bortezomib
% of cells with separated nuceloids
Inhibition of DNA replication
T = 120
DMSO
T = 120
Bortezomib
**
10
20
30
40
0
N = 3
n = 100
DNA signal
α-ring
(Saci0613)
α-ring
(Saci0613)
β-ring
(Saci0662ΔN)
β-ring
(Saci0662ΔN)
A
B
C
D
Time after release from acetic acid synchronization
Inactive
proteasome
Active
proteasome
Active proteasome
+ bortezomib
Urm1(x3)-GFP signal
N = 3
20S S. acidocaldarius
proteasome crystal structure
In vitro substrate degradation by the proteasome 
with and without bortezomib
E
F
G
H
**
DNA signal
DNA signal
DNA signal
DNA signal
Inactive 
proteasome
Active
proteasome
Bortezomib + active
proteasome
Urm1(x3)-GFP
Saci0909ΔN (β-active)
Saci0613 (α)
Saci0662ΔN (β-inactive)
55 kDa
35 kDa
25 kDa
0
50
100
150
     THR1
     LYS33
SER131
SER171
ASP168
     ASP17
     ASP17
     ASP17
     THR21
     Bortezomib
ALA49
Fig. 1. S. acidocaldarius cell division is arrested after inhibition of the
proteasome. (A) Side view of the S. acidocaldarius 28-subunit Saci_0613/
Saci_0662DN 20S proteasome assembly. Saci_0613 a subunits are indicated
by blue and purple hues; Saci_0662DN b subunits are indicated by red,
pink, and beige hues. (B) Modeled binding of the reversible inhibitor
bortezomib to the active b subunit (Saci_0909DN) chymotryptic site in the
S. acidocaldarius 20S proteasome using the CovDock software. The catalytic
threonine (Thr1) is shown as a red stick covalently bound to the boronic
acid group of the bortezomib molecule. The Asp17, Lys33, Ser129, Asp166, and
Thr169 involved in catalysis are shown as yellow sticks. The two loops (blue)
harboring the conserved residues Ala20, Ser21, and Gly47 to Val49 mediate
binding of the inhibitor. (C) Quantified degradation assay of GFP control or
Urm1(x3)-GFP substrate by the Saci_0613/Saci_0662DN/Saci_0909DN
20S proteasome. Error bars represent means ± SD from N = 3. Data were
analyzed using Welch’s t test; **p = 0.0034. (D) Flow cytometry histograms
showing 1N and 2N DNA signals in cells in an asynchronous and a
synchronized population treated with bortezomib or DMSO 80 min after
release (before division). N = 3, and n = 105. Representative histograms are
shown. T, time after release. (E) Wide-field microscopy images of synchronized
cells treated with DMSO or bortezomib for 40 min, 80 min after release from
acetic acid. Fixed cells were stained with Hoechst to visualize DNA and
concanavalin A to visualize the S-layer. Red arrows indicate cells with separated
nucleoids. Representative images are shown. Scale bar, 1 mm. (F) Quantification
of the percentage of cells in synchronized populations harboring separated
nucleoids after treatment with DMSO or bortezomib. Error bars represent
means ± SD from N = 3 and n = 100. Data were analyzed using a ratio
paired t test; **p = 0.0069. (G) Wide-field microscopy images of synchronized
STK cells treated for 40 min with DMSO or bortezomib, 100 min after release
from acetic acid. Cells were stained with Hoechst to visualize DNA and
concanavalin A to visualize the S-layer, and “click” chemistry was used to
visualize EdU. Red arrows indicate the cells with EdU incorporated into newly
synthesized DNA. Scale bar, 1 mm. (H) Quantification of the percentage of
synchronized cells with incorporated EdU after DMSO or bortezomib treatment.
Error bars represent means ± SD from N = 3 and n = 106.
RESEARCH
|
RESEARCH ARTICLE
Downloaded from http://science.sciencemag.org/ on August 6, 2020

----!@#$NewPage!@#$----
Tarrason Risa et al., Science 369, eaaz2532 (2020)
7 August 2020
3 of 8
1N
2N
DNA signal
CdvB signal [log]
3x103
3x10-3
3x105
3x104
0
CdvB levels
E
D
DNA signal
CdvB1 signal [log]
3x103
3x10-3
3x105
3x104
0
CdvB1 levels
H
DNA signal
CdvB signal [log]
3x103
3x10-3
3x105
3x104
0
CdvB levels with bortezomib
I
DNA signal
CdvB1 signal [log]
3x103
3x10-3
3x105
3x104
0
CdvB1 levels with bortezomib
J
CdvB signal [log]
CdvB1 signal [log]
3x103
3x105
3x104
0
3x103
3x10-3
3x105
3x104
0
CdvB & CdvB1 levels with bortezomib
G
50
100
150
200
0
DNA signal
Cell count (103)
DNA content (in F)
100
200
300
0
K
CdvB
CdvB + bz
CdvB1
CdvB1 + bz
CdvB2
CdvB2 + bz
0
1x104
2x104
3x104
4x104
Intensity (a.u.)
*
ns
ns
1N
2N
Peak CdvB, CdvB1, and CdvB2
Median signal
1N
2N
1N
2N
n=106
n=106
n=106
n=106
n=106
n=106
n=106
F
1
2
β
α
3
CdvB signal [Log]
CdvB1 signal [log]
3x103
3x105
3x104
0
3x103
3x10-3
3x105
3x104
0
CdvB & CdvB1 levels
β
α
3x10-3
3x10-3
2
3
A
C
Fold change [bortezomib arrest] - [wash-out]
Experiment 1
-2
1
-1
0
1
2
Fold change [bortezomib arrest] - [wash-out]
Experiment 2
-1
0
2
3
CdvB/ESCRT-III
CdvA
CdvC/Vps4
CdvB1/ESCRT-III
CdvB2/ESCRT-III
50
100
150
200
DNA Signal
Cell count
Bz arrest
r=0.592
Proteome change before and after bortezomib wash-out
1
5x103
10x103
15x103
20x103
25x103
0
-10
3
10
3
10
4
10
5
Non-induced
Induced
Cell count
n = 10 6
CdvB
Alba
PAN-HA
Alba
PAN-WkB
Non-induced
PAN-WkB
induced
CdvB levels
+/- PAN-WkB
CdvB signal [log]
50
100
150
200
50
100
150
200
Untreated
DMSO
MG132
Bortezomib
CdvB
Alba
B
50
100
150
200
50
100
150
200
CdvB levels
Bz wash-out
Saci_0077
Saci_0054
Saci_1130
Saci_1490
Fig. 2. CdvB is targeted by the S. acidocaldarius proteasome during cell
division. (A) Mass spectrometry scatterplot correlating two independent
replicates of proteome changes in predivision synchronized cells treated with
bortezomib for 40 min versus the population 15 min after bortezomib had
been washed out (after division). The top five hits are in green (apart from
CdvB). See table S6 for the complete data. The inset shows two representative
histograms of the bortezomib-treated and wash-out samples. (B) Western
blots of CdvB and Alba (loading control) for synchronized cells which, 80 min
after release from acetic acid, were left untreated or treated for 40 min with
DMSO, MG132, or bortezomib. (C) (Left) Flow cytometry histogram comparing
the CdvB signal under noninduced and induced conditions for a PAN-WkB-HA
dominant negative overexpression strain and (right) Western blots showing the
change in PAN-WkB-HA and CdvB protein levels. (D and E) Flow cytometry
scatterplots of DNA versus CdvB (D) and CdvB1 (E) protein signals of cells in
an asynchronous culture. The red boxes highlight the “mitotic” cells with high
CdvB and CdvB1 signals in (D) and (E), respectively. Each blue spot represents a
single cell, with the density gradient going from blue to red. Representative
plots are shown; N = 3, and n = 106. (F) Flow cytometry scatterplot of the CdvB
versus CdvB1 signal of an asynchronous culture showing the alternating
accumulation and loss of the proteins. N = 3, and n = 106. (G) Flow cytometry
histogram of the DNA distribution of cells in the boxes shown in (F), showing
the shift from 2N to 1N taking place only after complete CdvB degradation.
(H and I) Flow cytometry scatterplots of DNA versus CdvB and CdvB1 signals for
an asynchronous culture treated with bortezomib. Representative plots are
shown; N = 3, and n = 106. (J) Flow cytometry scatterplot of CdvB versus CdvB1
levels in an asynchronous culture treated with bortezomib showing an
accumulation of cells expressing CdvB and CdvB1 and the selective increase
in CdvB levels. Representative plot is shown; N = 3, and n = 106. (K) Quantification
of median CdvB, CdvB1, and CdvB2 signals in the mitotic populations. Data
were analyzed using a ratio paired two-tailed t test for CdvB with and without
bortezomib (bz); *p = 0.0245. Error bars represent means ± SD from N = 3. ns,
not significant; a.u., arbitrary units.
RESEARCH
|
RESEARCH ARTICLE
Downloaded from http://science.sciencemag.org/ on August 6, 2020

----!@#$NewPage!@#$----
induced cell cycle arrest was accompanied
by a twofold enrichment in the number of cells
harboring compact and separated nucleoids
[relative to those treated with the dimethyl
sulfoxide (DMSO) solvent; Fig. 1, E and F].
Although a similar arrest was observed in cells
treated with MG132, a second proteasome in-
hibitor, this arrest proved more readily revers-
ible, as expected given the inhibitory constant
(Ki) values of bortezomib and MG132 (0.6 and
4 nM, respectively) (fig. S5A).
To test whether predivision cells treated
with bortezomib were inhibited from under-
going DNA replication as well as cell division,
as is the case for a proteasome inhibition–
induced mitotic arrest in eukaryotes, we made
use of a S. acidocaldarius mutant strain ex-
pressing thymidine kinase (STK). STK cells
incorporate the thymidine analog 5-ethynyl-
2′-deoxyuridine (EdU) into newly synthesized
DNA, whichcanbevisualized withclick-itchem-
istry (see methods) (24). Using this assay, the
majority of synchronized STK cells divided and
incorporated EdU into their DNA as they en-
tered S phase [note that EdU prevents the com-
pletion of S phase in S. acidocaldarius (24)].
By contrast, there was no evidence of EdU in-
corporation, and therefore DNA synthesis, in
cells treated with bortezomib before division
(Fig. 1, G and H, and fig. S4, C and D). When
bortezomib was added to synchronous cul-
tures at a later stage to inhibit the proteasome
in newly divided cells, however, these G1 cells
continued unimpeded into S phase and G2 (fig.
S4E). These data indicate that inhibiting protea-
some activity specifically prohibits cells from
dividing and also, directly or indirectly, from
initiating the next round of DNA replication.
CdvB is targeted by the proteasome
during cell division
Having observed cell cycle arrest after pro-
teasome inhibition (with bortezomib and MG132),
we set out to identify proteins that are subject
to proteasome-mediated degradation during cell
division using tandem mass tag (TMT) label-
ing and quantitative mass spectrometry. Hits
were identified by comparing the proteomes
of bortezomib-treated synchronized cultures
enriched for cells arrested before division with
those of cultures 15 min after inhibitor wash-
out, at which point most arrested cells had di-
vided (see methods). Notably, among the large
set of proteins sampled by mass spectrometry,
the protein concentration that decreased the
most after release from the arrest was CdvB,
an ESCRT-III homolog previously identified as
a component of the archaeal division machin-
ery (13–15, 25) (Fig. 2A and table S6). As ex-
pected, CdvB was also identified among the
top hits when the proteomes of MG132-treated
predivision cultures were compared with the
proteomes of DMSO-treated controls (fig. S5B
and table S7). These results were confirmed by
Western blotting using antibodies validated
against cell extracts and purified proteins (Fig.
2B and figs. S6, A and B, and S7B; see methods).
Furthermore, CdvB levels increased markedly
in cells after induced expression of a Walker B
dominant-negativePANAAA+ATPase(Saci_0656),
a protein known to cap the 20S proteasome,
which aids protein degradation by threading
proteins into its core (8) (Fig. 2C). Notably, CdvB
was the only division protein consistently en-
riched in these experiments (Fig. 2A and figs.
S5A and S7B). CdvA (Saci_1374, an archaea-
specific ESCRT-III recruitment protein), CdvC
(Saci_1372, a Vps4 homolog), and CdvB1 and
CdvB2 (Saci_0451 and Saci_1416, ESCRT-III
homologs), were largely unaffected by pro-
teasome inhibition. This selectivity was not
due to differences in Cdv gene transcription,
because all of these components of the cell
division machinery are transcribed as part of
the same predivision wave (13, 14), which we con-
firmed remains active during a bortezomib-
induced predivision arrest (fig. S7A). Taken
together, these observations imply a role for
selective proteasome-mediated degradation
of CdvB at division.
We observed a similar sudden loss of CdvB
protein from cells when we used flow cytom-
etry to assess the levels of different ESCRT-III
proteins in a population of unperturbed cells
as they underwent division (see methods). Again,
CdvB was found accumulating to high levels in
predivision cells but was absent from newly
divided G1 cells (Fig. 2D and fig. S8C). By con-
trast, CdvB1 and CdvB2 were maintained at
high levels throughout the division process and
then partitioned between newly divided daugh-
ter cells (Fig. 2E and fig. S8, A, D, and F). In
examining the dynamics of CdvB and CdvB1
levels during cell division, it also became clear
that CdvB was degraded in cells with 2N DNA
content, i.e., before cell division (Fig. 2, F and G,
and figs. S8E and S10C). As expected, a flow
cytometry analysis of CdvB revealed increased
levelsincellsarrestedbeforedivisionbybortezomib
(Fig. 2H), whereas median levels of CdvB1
and CdvB2 remained unaltered (Fig. 2, I to K,
and fig. S8B). Taken together, these observa-
tions show that CdvB is selectively targeted
for proteasome-mediated degradation part-
way through the division process.
CdvB degradation triggers constriction
of the CdvB1:CdvB2 ring
Previous studies have suggested that CdvB, an
essential gene in S. acidocaldarius, is actively
required for division (13–15, 25, 26). By con-
trast, our data suggest that the protein is de-
graded during division. To reconcile these data
and understand the role of CdvB and its degra-
dation in cell division, we used super-resolution
radial fluctuations (SRRF) super-resolution mi-
croscopy to image the archaeal ESCRT-III
division rings at different stages in the divi-
sion process (see methods) (27, 28). This re-
vealed three distinct categories of ESCRT-III
division rings: (i) CdvB rings that lack CdvB1,
(ii) rings with colocalized CdvB and CdvB1,
and (iii) CdvB1 rings that lack CdvB (Fig. 3A).
Conspicuously, cells harboring a CdvB1 ring,
but lacking CdvB signal, were the only onesseen
in a state of constriction. These data support a
role for the selective proteasome-mediated deg-
radation of CdvB from preassembled CdvB:
CdvB1 rings. This is consistent with the fact
that CdvB rings had a mean diameter of 1.23 mm
(n = 496, SD = 0.15) (Fig. 3B) and CdvB1 rings co-
staining for CdvB had a near-identical mean
diameter of 1.24 mm (n = 285, SD = 0.17), where-
as CdvB1 rings that lacked CdvB tended to be
variable in size and smaller (mean = 0.62 mm,
n = 61, SD = 0.35) (Fig. 3C). Moreover, as ex-
pected for a division ring, the diameter of the
CdvB1 ring was nearly perfectly correlated with
the diameter of the division neck, as measured
by concanavalin A staining of the archaeal
S-layer (fig. S8G). Finally, the diameters of CdvB1
and CdvB2 rings were near perfectly correlated
in all images, suggesting that the two ESCRT-III
homologs work together during the ring con-
striction stage of the division process [N = 399,
correlation coefficient (r) = 0.98, p = 2.2 × 10−16]
(fig. S8H). Taken together, these data suggest
that constriction is triggered by the selective
and rapid degradation of CdvB by the protea-
some. In line with this thinking, we were un-
able to detect CdvB1 rings lacking CdvB (or
CdvB2 rings lacking CdvB) in cells after treat-
ment with the proteasome inhibitor bortezomib.
Moreover, all of the rings in bortezomib-treated
cells had a mean diameter equivalent to the uni-
form width of CdvB rings that span the longest
axis of the cell, highlighting the role of the
proteasome-mediated degradation of CdvB in
triggering cell constriction (mean = 1.23 mm,
N = 346, SD = 0.32) (Fig. 3D).
The flow cytometry profiles of exponentially
growing asynchronous populations co-stained
for CdvB and CdvB1 could also be used to gen-
erate an approximate timeline of cell division
events. This revealed populations of cells in
three distinct phases of the division process:
(i) 2N cells with peak levels of CdvB but low
CdvB1, (ii) 2N cells with peak levels of both
CdvB and CdvB1, and (iii) 2N cells with fall-
ing levels of CdvB that retain peak levels of
CdvB1 (fig. S9A). Because the doubling time
of these S. acidocaldarius cultures was about
2 hours and 45 min, the relative sizes of these
subpopulations could be used to calculate the
time spent by a typical cell in each stage of the
division process. Notably, this requires taking
the exponential age distribution resulting from
binary division into account (fig. S9B) (29). Thus,
the entire process of ESCRT-III–mediated cell
division was estimated to have a duration of
~6.2 min (fig. S9, B and C) (29). The division
process was further resolved on the basis of
Tarrason Risa et al., Science 369, eaaz2532 (2020)
7 August 2020
4 of 8
RESEARCH
|
RESEARCH ARTICLE
Downloaded from http://science.sciencemag.org/ on August 6, 2020

----!@#$NewPage!@#$----
these data to reveal the following sequence of
events: (i) CdvB rings had an average lifetime
of ~2.6 min, before the assembly of CdvB1, (ii)
CdvB:CdvB1 rings then persisted for ~1.8 min,
before CdvB was degraded by the proteasome,
(iii) causing CdvB1 rings to constrict within
a period of ~1.8 min. Although we observed
minor variations in the percentage of cells in
the other two stages of division, the percent-
age of cells in the contractile phase of the
division process (as CdvB is being degraded)
appeared highly reproducible (fig. S9C).
Physical model of ESCRT-III–mediated cell
division in Archaea
To better monitor changes in both the levels
and localization of the CdvB1 ring during the
final stages of the division process, we imaged
fixed cells using linear structured illumination
microscopy (iSIM) [see methods]. This revealed
that constriction of the CdvB1 division ring
was accompanied by both an increase in the
intensity of the fluorescent signal at the neck
and a concomitant increase in the level of the
diffuse cytoplasmic signal (Fig. 4A and fig. S10A).
Throughout division, the sum intensity of CdvB1
remained constant (fig. S10B). As a result, newly
divided cells had high levels of diffuse CdvB1
(something we never observed imaging CdvB),
in line with the flow cytometric data (Fig. 2F
and fig. S8, C and D). The absence of large
polymeric CdvB1:CdvB2 structures in these
early G1 cells implies that CdvB1 and CdvB2
do not form de novo ESCRT-III filaments
without a CdvB template. In keeping with this
conclusion, CdvB2 appeared diffuse when it
was ectopically expressed in cells arrested in
G2 with acetic acid, a treatment that inhibits
the expression of endogenous Cdv proteins
(fig. S11, A to C). Taken together, these observa-
tions suggest that the division machinery fol-
lows a fixed sequence in which CdvB initially
forms an ESCRT-III ring [in a process that
likely depends on CdvA (25, 30)], which then
templates the assembly of CdvB1:CdvB2 rings.
This CdvB1:CdvB2 ring constricts after the
sudden removal and degradation of CdvB
and is then disassembled during the division
process.
To test whether these findings provide a plau-
sible physical mechanism for S. acidocaldarius
cell division, we used these data as the basis for
a coarse-grained molecular dynamics model of
ESCRT-III–mediated cell division, based on a
recently developed physical model of ESCRT-
III–dependent membrane scission (31) (see
methods). In these simulations, a modeled
filament, representing the division ring, was
allowed to associate with the plasma mem-
brane of a cell modeled in three dimensions
Tarrason Risa et al., Science 369, eaaz2532 (2020)
7 August 2020
5 of 8
A
  41% of dividing population   
~ 2.6 minutes
 29% of dividing population
 ~ 1.8 minutes
 29% of dividing population
~ 1.8 minutes
B
x = 1.23 µm
n = 496
x = 1.13µm
n  = 346 
0.0
0.5
1.0
2.0
1.5
Diameter (µm)
CdvB
CdvB1
Division ring diameters
C
D
x = 1.24 µm
n = 285
x = 0.62µm
n  = 61
0.0
0.5
1.0
2.0
1.5
Diameter (µm)
CdvB1 
(with CdvB)
CdvB1
(without CdvB)
CdvB1 division ring diameters
x = 1.21 µm
n = 554
x = 1.23µm
n  = 352
0.0
0.5
1.0
2.0
1.5
Diameter (µm)
CdvB 
with bortezomib
CdvB1
with bortezomib
Division ring diameters 
with bortezomib
CdvBCdvB1
DNA
N = 3
N = 3
N = 3
Images from an asynchronous population of cells
Fig. 3. Targeted degradation of CdvB triggers constriction of CdvB1.
(A) Representative SRRF super-resolution images of ring structures observed in
exponentially growing asynchronous cell cultures. Cells were stained for DNA
with Hoechst (blue), CdvB (purple), and CdvB1 (green); the images below
the composites represent the CdvB (left) and CdvB1 (right) channels. Percentages
refer to the total ring-bearing population. Minutes of the total cell cycle are
adjusted for the exponential (binary fission) age distribution. Scale bar, 0.5 mm.
(B) Quantification of CdvB and CdvB1 ring diameters in an asynchronous culture.
�x, mean ring diameters. (C) Quantification of ring diameters of CdvB and
CdvB1 in the presence and absence of a CdvB ring, showing how the removal of
CdvB leads to constriction of CdvB1. (D) Quantification of ring diameters of
CdvB and CdvB1 in the presence of the proteasomal inhibitor bortezomib. The
data in (B) to (D) all stem from more than six fields of view and three biological
replicates. Red bars represent the mean values.
RESEARCH
|
RESEARCH ARTICLE
Downloaded from http://science.sciencemag.org/ on August 6, 2020

----!@#$NewPage!@#$----
through its membrane-binding interface. To
model the division process, we defined two
different preferred filament conformations:
a CdvB1 (CdvB+) state with low curvature (de-
fined by a large preferred radius R0) and a
CdvB1 (CdvB−) state with high curvature (de-
fined by the small target radius Rtarget) (Fig. 4B).
In this model, the transition between the initial
CdvB1 (CdvB+) state and the CdvB1 (CdvB−)
state represents the rapid loss of CdvB, which
was modeled by an instantaneous shortening
of the bonds connecting neighboring elements
in the filament. Although the constriction of
the CdvB1 ESCRT-III filament was able to
rapidly reduce the cell circumference in these
simulations, it was not sufficient to induce
division because of the steric hindrance
at the neck (Fig. 4C and Movie 1, where
Rtarget/R0 = 5%). This is a generic problem
faced by molecular machines that cut mem-
brane tubes from the inside. Thus, it was only
when CdvB1 filaments were allowed to dis-
assemble as they constricted, as observed in
cells (Fig. 4A), that division occurred (Fig. 4D).
This was modeled as a gradual loss of indi-
vidual subunits from both ends of the helix at
a specified rate, such that the filament length
Tarrason Risa et al., Science 369, eaaz2532 (2020)
7 August 2020
6 of 8
Diameter
Time
R target
R 0
Initial state of CdvB1 
Structure templated by CdvB (not shown)
Target state of CdvB1
State transition triggered 
after the degradation of CdvB
B
Molecular dynamics simulation of cell constriction with disassembly of CdvB1
D
A
Cytoplasmic CdvB1 signal
Cytoplasmic levels of CdvB1 during constriction
Ring diameter (µm) 
0.0
0.5
1.0
1.5
2.0
0.00
0.25
0.50
0.75
1.00
Time
N = 3
n = 58
Molecular dynamics simulation of cell constriction without disassembly of CdvB1
C
Fig. 4. Physical model of ESCRT-III–mediated cell division in Archaea.
(A) Quantification of cytoplasmic CdvB1 signal plotted against the CdvB1 ring
diameter from a linear iSIM microscopy dataset, with linear regression with
95% confidence interval shown in black. The inset shows representative linear
iSIM microscopy images of CdvB1 at various stages of constriction and a cell in
G1 after division. The outline of each cell is highlighted with a white dashed
line. Scale bar, 1 mm. (B) The molecular dynamics simulation of the ESCRT-III
filament is built with three-beaded subunits, which are connected by harmonic
springs to form a helix. Only the green beads are attracted by the modeled cell
membrane. The helix undergoes a geometrical change from a state with low
curvature 1/R0, corresponding to CdvB1 with CdvB, to a state with high curvature
1/Rtarget, corresponding to CdvB1 without CdvB. This change occurs after the
proteasomal degradation of CdvB. (C) Simulation snapshots as a function of time
showing that cell division is not achieved by constriction of the ESCRT-III
filament alone; see Movie 1. (D) Simulation snapshots showing that disassembly
of the ESCRT-III filament is necessary for cell division; see Movie 2.
RESEARCH
|
RESEARCH ARTICLE
Downloaded from http://science.sciencemag.org/ on August 6, 2020

----!@#$NewPage!@#$----
decreases linearly in time (Movie 2, where
Rtarget/R0 = 5%). In sum, these data and sim-
ulations highlight the importance of disas-
sembling the CdvB1:CdvB2 copolymer during
constriction, so that the ESCRT-III filament
does not crowd the neck and prevent scission.
Discussion
These findings demonstrate how proteasome-
mediated degradation of CdvB, a component
of the ESCRT-III ring, plays a key role in or-
dering the changes in ESCRT-III copolymer
structure to drive cell division and suggest an
elegant mechanism by which S. acidocaldarius
cells divide. First, cells assemble a structural
noncontractile CdvB ring at the center of the
cell. This ESCRT-III filament acts as a template
for the assembly of CdvB1 and CdvB2 ESCRT-III
proteins, which have a smaller preferred ra-
dius of curvature. The tension stored in this
CdvB1:CdvB2 polymer is then released once
the CdvB template is removed and degraded
by the proteasome, driving membrane con-
striction. During constriction, this copolymer
is disassembled gradually to enable scission
of the membrane bridge connecting the two
nascent daughter cells.
Although we were able to demonstrate that
the proteasome is able to directly degrade CdvB
in vitro (fig. S12), the process is likely more
complicated in vivo. This is supported by the
observation that a defective PAN proteasome
cap leads to an accumulation of CdvB (Fig. 2C).
Moreover, previous studies have implicated
the ESCRT-III–associated protein Vps4 (CdvC)
in the division process (14, 32). In eukaryotes,
Vps4 plays a role in the stepwise disassembly of
composite ESCRT-III filaments and in mem-
brane deformation (31, 33, 34). Because the role
of Vps4 appears to be conserved, these data
suggest that CdvC might be required to selec-
tively extract CdvB from the composite CdvB,
CdvB1:CdvB2 copolymer. This soluble pool of
CdvB might then be rapidly targeted to the
proteasome via PAN, and possibly other pro-
teins, leading to its degradation before it can
be reincorporated back into the ring. Although
this remains speculation, under this model, the
degradation of CdvB would make this change
in ESCRT-III polymer structure functionally
irreversible. CdvC might then also catalyze
the disassembly of the CdvB1:CdvB2 polymer,
which is required to vacate the division neck,
thus aiding the final process of membrane scis-
sion. In line with this idea, previous studies have
found that truncating the CdvC-interacting
domain in CdvB, CdvB1, or CdvB2 leads to an
arrest at various points during the constriction
process (15), implying that CdvC plays one or
more essential roles in archaeal cell division.
This study reveals that ESCRT-III–mediated
cell division in S. acidocaldarius is regulated by
the selective proteasome-mediated degrada-
tion of CdvB. It also reveals parallels between
cell cycle control in archaea and eukaryotes,
implying that proteasome-mediated regula-
tion predates cyclins and cyclin-dependent
kinases. Further work will be required (i) to
identify all the relevant targets of the protea-
some during this final cell cycle transition in
Archaea, (ii) to determine whether the protea-
some plays an analogous role in driving ESCRT-
III–dependent cell division in eukaryotes, (iii)
to determine whether the degradation of any
of these proteins plays a role in resetting the
archaeal cell cycle in a manner analogous the
role played by cyclin degradation in eukaryotes,
and (iv) to determine if there are other shared
design features governing cell cycle progression
across the archaeal-eukaryotic divide. How-
ever, this study emphasizes the utility of using
archaea as a simple model to explore funda-
mental features of eukaryotic molecular cell
biology that have been conserved over 2 bil-
lion years of evolution.
Materials and methods summary
Cell culturing, synchronization, and drug treatments
S. acidocaldarius strains were grown at 75°C
and pH 3.0 to 3.5 in Brock medium supple-
mented with NZ-amine and sucrose, with the
exception of the STK and MW001 strains that
were also supplemented with uracil. Synchro-
nizationwasachieved with acetic acid treatment
and proteasome inhibition by supplementing
bortezomib or MG132.
Biochemistry and mass spectrometry
Biochemical analyses were carried out on ex-
tracted protein by Western blotting and mass
spectrometry and on RNA by quantitative poly-
merase chain reaction (qPCR). In vitro assays
of proteasome activity were studied in the con-
text of active and inactive proteasome as-
semblies supplemented with a 3xUrm1-GFP
reporter. Proteins of interest were expressed
in Escherichia coli. The antibody specificities
for CdvB, CdvB1, and CdvB2 were tested by
enzyme-linked immunosorbent assay (ELISA)
and whole Western blots.
S. acidocaldarius genetics
Two overexpression strains were generated: (i)
a dominant negative PAN-WalkerB-HA (HA,
hemagglutinin) and (ii) a wild-type CdvB2.
Both proteins were subject to arabinose induc-
tion and maintained on a plasmid comple-
menting the strains’ uracil auxotrophy.
Microscopy and flow cytometry
All microscopy and flow cytometry studies
were done on ethanol-fixed cells, staining for
DNA and the S-layer and with antibodies
against ESCRT-III homologs. For super-resolu-
tion, SRRF and linear iSIM were used.
Crystallography
The inactive proteasome subunits were over-
expressed in and purified from E. coli. Subse-
quently, crystals were grown in CZ buffer (see
methods), and a 3.7-Å dataset was collected
from a single crystal at SOLEIL, France.
Molecular dynamics simulations
A fluid spherical membrane consisting of
48,000 particles was modeled together with
a chiral ESCRT-III filament modeled with
three-beaded subunits (31). Simulations for
the constriction and disassembly of the ESCRT-
III filament were run with the LAMMPS mo-
lecular dynamics package.
REFERENCES AND NOTES
1.
R. Bernander, The cell cycle of Sulfolobus. Mol. Microbiol.
66, 557–562 (2007). doi: 10.1111/j.1365-2958.2007.05917.x;
pmid: 17877709
2.
D. Ausiannikava, T. Allers, Diversity of DNA replication in the
Archaea. Genes 8, 56 (2017). doi: 10.3390/genes8020056;
pmid: 28146124
3.
S. Lang, L. Huang, The Sulfolobus solfataricus GINS complex
stimulates DNA binding and processive DNA unwinding
by minichromosome maintenance helicase. J. Bacteriol.
197, 3409–3420 (2015). doi: 10.1128/JB.00496-15;
pmid: 26283767
4.
Y. Caspi, C. Dekker, Dividing the archaeal way: The ancient Cdv
cell-division machinery. Front. Microbiol. 9, 174 (2018).
doi: 10.3389/fmicb.2018.00174; pmid: 29551994
5.
M. Groll, H. Brandstetter, H. Bartunik, G. Bourenkow, R. Huber,
Investigations on the maturation and regulation of
Tarrason Risa et al., Science 369, eaaz2532 (2020)
7 August 2020
7 of 8
Movie 1. Constriction simulation without fila-
ment disassembly.
Movie 2. Constriction simulation with filament
disassembly.
RESEARCH
|
RESEARCH ARTICLE
Downloaded from http://science.sciencemag.org/ on August 6, 2020

----!@#$NewPage!@#$----
archaebacterial proteasomes. J. Mol. Biol. 327, 75–83 (2003).
doi: 10.1016/S0022-2836(03)00080-9; pmid: 12614609
6.
J. Löwe et al., Crystal structure of the 20S proteasome from
the archaeon T. acidophilum at 3.4 Å resolution. Science
268, 533–539 (1995). doi: 10.1126/science.7725097;
pmid: 7725097
7.
F. Zhang et al., Structural insights into the regulatory particle
of the proteasome from Methanocaldococcus jannaschii.
Mol. Cell 34, 473–484 (2009). doi: 10.1016/j.molcel.2009.04.021;
pmid: 19481527
8.
J. Maupin-Furlow, Proteasomes and protein conjugation across
domains of life. Nat. Rev. Microbiol. 10, 100–111 (2011).
doi: 10.1038/nrmicro2696; pmid: 22183254
9.
J. F. Giménez-Abián, L. A. Díaz-Martínez, K. G. Wirth,
C. De la Torre, D. J. Clarke, Proteasome activity is required for
centromere separation independently of securin degradation in
human cells. Cell Cycle 4, 1558–1560 (2005). doi: 10.4161/
cc.4.11.2145; pmid: 16205121
10. S. López-Avilés, O. Kapuy, B. Novák, F. Uhlmann, Irreversibility
of mitotic exit is the consequence of systems-level feedback.
Nature 459, 592–595 (2009). doi: 10.1038/nature07984;
pmid: 19387440
11. M. Glotzer, A. W. Murray, M. W. Kirschner, Cyclin is degraded
by the ubiquitin pathway. Nature 349, 132–138 (1991).
doi: 10.1038/349132a0; pmid: 1846030
12. K. Wickliffe, A. Williamson, L. Jin, M. Rape, The multiple layers
of ubiquitin-dependent cell cycle control. Chem. Rev. 109,
1537–1548 (2009). doi: 10.1021/cr800414e; pmid: 19146381
13. A.-C. Lindås, E. A. Karlsson, M. T. Lindgren, T. J. G. Ettema,
R. Bernander, A unique cell division machinery in the Archaea.
Proc. Natl. Acad. Sci. U.S.A. 105, 18942–18946 (2008).
doi: 10.1073/pnas.0809467105; pmid: 18987308
14. R. Y. Samson, T. Obita, S. M. Freund, R. L. Williams, S. D. Bell,
A role for the ESCRT system in cell division in Archaea. Science
322, 1710–1713 (2008). doi: 10.1126/science.1165322;
pmid: 19008417
15. J. Liu et al., Functional assignment of multiple ESCRT-III
homologs in cell division and budding in Sulfolobus islandicus.
Mol. Microbiol. 105, 540–553 (2017). doi: 10.1111/mmi.13716;
pmid: 28557139
16. R. S. Anjum et al., Involvement of a eukaryotic-like ubiquitin-
related modifier in the proteasome pathway of the archaeon
Sulfolobus acidocaldarius. Nat. Commun. 6, 8163 (2015).
doi: 10.1038/ncomms9163; pmid: 26348592
17. A. Šali, T. L. Blundell, Comparative protein modelling by
satisfaction of spatial restraints. J. Mol. Biol. 234, 779–815
(1993). doi: 10.1006/jmbi.1993.1626; pmid: 8254673
18. J. Adams et al., Proteasome inhibitors: A novel class of potent
and effective antitumor agents. Cancer Res. 59, 2615–2622
(1999). pmid: 10363983
19. D. Chen, M. Frezza, S. Schmitt, J. Kanwar, Q. P. Dou, Bortezomib
as the first proteasome inhibitor anticancer drug: Current
status and future perspectives. Curr. Cancer Drug Targets 11,
239–253 (2011). doi: 10.2174/156800911794519752;
pmid: 21247388
20. X. Fu et al., Ubiquitin-like proteasome system represents a
eukaryotic-like pathway for targeted proteolysis in Archaea.
mBio 7, e00379–e16 (2016). doi: 10.1128/mBio.00379-16;
pmid: 27190215
21. N. L. Hepowit, J. A. Maupin-Furlow, Rhodanese-like domain
protein UbaC and its role in ubiquitin-like protein modification
and sulfur mobilization in Archaea. J. Bacteriol. 201, e00254-19
(2019). doi: 10.1128/JB.00254-19; pmid: 31085691
22. R. Oerlemans et al., Molecular basis of bortezomib resistance:
Proteasome subunit b5 (PSMB5) gene mutation and
overexpression of PSMB5 protein. Blood 112, 2489–2499
(2008). doi: 10.1182/blood-2007-08-104950; pmid: 18565852
23. M. Lundgren, A. Andersson, L. Chen, P. Nilsson, R. Bernander,
Three replication origins in Sulfolobus species: Synchronous
initiation of chromosome replication and asynchronous
termination. Proc. Natl. Acad. Sci. U.S.A. 101, 7046–7051
(2004). doi: 10.1073/pnas.0400656101; pmid: 15107501
24. T. Gristwood, I. G. Duggin, M. Wagner, S. V. Albers, S. D. Bell,
The sub-cellular localization of Sulfolobus DNA replication.
Nucleic Acids Res. 40, 5487–5496 (2012). doi: 10.1093/nar/
gks217; pmid: 22402489
25. R. Y. Samson et al., Molecular and structural basis of ESCRT-III
recruitment to membranes during archaeal cell division.
Mol. Cell 41, 186–196 (2011). doi: 10.1016/j.molcel.2010.12.018;
pmid: 21255729
26. N. Yang, A. J. M. Driessen, Deletion of cdvB paralogous genes of
Sulfolobus acidocaldarius impairs cell division. Extremophiles 18,
331–339 (2014). doi: 10.1007/s00792-013-0618-5; pmid: 24399085
27. N. Gustafsson et al., Fast live-cell conventional fluorophore
nanoscopy with ImageJ through super-resolution radial
fluctuations. Nat. Commun. 7, 12471 (2016). doi: 10.1038/
ncomms12471; pmid: 27514992
28. R. F. Laine et al., NanoJ: a high-performance open-source
super-resolution microscopy toolbox. bioRxiv 432674
[Preprint]. 6 October 2018; .doi: 10.1101/432674
29. S. Wold, K. Skarstad, H. B. Steen, T. Stokke, E. Boye, The
initiation mass for DNA replication in Escherichia coli K-12 is
dependent on growth rate. EMBO J. 13, 2097–2102 (1994).
doi: 10.1002/j.1460-2075.1994.tb06485.x; pmid: 8187762
30. M. J. Dobro et al., Electron cryotomography of ESCRT
assemblies and dividing Sulfolobus cells suggests that spiraling
filaments are involved in membrane scission. Mol. Biol. Cell 24,
2319–2327 (2013). doi: 10.1091/mbc.e12-11-0785;
pmid: 23761076
31. L. Harker-Kirschneck, B. Baum, A. E. Šarić, Changes in
ESCRT-III filament geometry drive membrane remodelling and
fission in silico. BMC Biol. 17, 82 (2019). doi: 10.1186/
s12915-019-0700-2; pmid: 31640700
32. B. E. Mierzwa et al., Dynamic subunit turnover in ESCRT-III
assemblies is regulated by Vps4 to mediate membrane
remodelling during cytokinesis. Nat. Cell Biol. 19, 787–798
(2017). doi: 10.1038/ncb3559; pmid: 28604678
33. A.-K. Pfitzner, V. Mercier, A. Roux, Vps4 triggers sequential
subunit exchange in ESCRT-III polymers that drives membrane
constriction and fission. bioRxiv 718080 [Preprint]. 29 July 2019;
https://doi.org/10.1101/718080.doi: 10.1101/718080
34. C. Caillat, S. Maity, N. Miguet, W. H. Roos, W. Weissenhorn,
The role of VPS4 in ESCRT-III polymer remodeling.
Biochem. Soc. Trans. 47, 441–448 (2019). doi: 10.1042/
BST20180026; pmid: 30783012
ACKNOWLEDGMENTS
We thank the MRC LMCB at UCL for their support; the flow
cytometry STP at the Francis Crick Institute for assistance, with
special thanks to S. Purewal and D. Davis; C. Bertoli for mentorship
and advice; J. M. Garcia-Arcos for help early on in this project;
the entire Baum lab for their input throughout the project; the Albers
lab for advice and reagents, with special thanks to M. Van Wolferen
and S. Albers; the members of the Wellcome consortium for archaeal
cytoskeleton studies for advice and comments; and J. Löwe,
S. Oliferenko, M. Balasubramanian, and D. Gerlich for discussions
and advice on the manuscript. N.P.R. and S.B. would like to thank
N. Rzechorzek, A. Simon, and S. Anjum for discussion and advice.
Funding: The Baum, Henriques, and de Bruin labs received support
from the MRC (MC_CF12266). G.T.R. was supported by MRC
(1621658). F.H. was supported by Vetenskapsrådet (621-2013-4685)
and A.A.P. by the HFSP (LT001027/2019 fellowship).
A collaborative grant from the Wellcome Trust supported A.A.P.,
D.P.S., G.D., S.C., and the Baum, Lindas, and Henriques labs (203276/
Z/16/Z). D.R.M. was supported by a BBSRC grant awarded to B.B.
(BB/K009001/1). L.M. and R.A.M.d.B. were supported by Cancer
Research UK (C33246/A20147). P.F., C.D., and A.P.S. were
supported by the Francis Crick Institute, which receives its core
funding from Cancer Research UK (FC001999), the UK Medical
Research Council (FC001999), and the Wellcome Trust (FC001999).
L.H.-K. was supported by the Biotechnology and Biological Sciences
Research Council (BB/M009513/1). A.Š. and A.E.H. were supported
by the European Research Council (802960) and the Engineering
and Physical Sciences Research Council (EP/R011818/1). N.P.R.
was supported by the Isaac Newton Trust Research Grant (Trinity
College and Department of Biochemistry, Cambridge) and start-up
funds from the Division of Biomedical and Life Sciences (Lancaster
University) and received a BBSRC Doctoral Training Grant (RG53842)
for S.B. Author contributions: B.B. and G.T.R. conceived the
study. Initial observations were made by F.H. and G.T.R. Structural
work was planned and carried out by S.B. and N.P.R., with
assistance from M.K. and L.P. In vitro proteasome assays were
planned and performed by D.P. and N.P.R. Cell biology methods
and experiments were developed by G.T.R., F.H., G.D., D.R.M., and
S.C. with guidance from B.B., R.H., and A.-C.L. Molecular genetics
were done by F.H., D.R.M., A.A.P., and G.T.R. Biochemical analysis
was carried out by F.H., G.T.R., L.M., A.A.P., and D.P.S. RNA
analysis was carried out by C.F., with guidance from B.B. and
R.A.M.d.B. Mass spectrometry preparation and analysis was
carried out by P.F., C.D., and G.T.R., with guidance from A.P.S.
The physical model was built by A.E.H. and L.H.-K., with guidance
from A.Š. and B.B. The paper was written by G.T.R., F.H., and B.B.,
with input from all other authors. Competing interests: The
authors declare no competing interests. Data and materials
availability: All data are available in the manuscript or the
supplementary materials. The crystal structure of the S. acidocaldarius
20S proteasome is available at PDB, with the ID 6Z46. All
materials are available from B.B. or N.P.R. on request.
SUPPLEMENTARY MATERIALS
science.sciencemag.org/content/369/6504/eaaz2532/suppl/DC1
Materials and Methods
Figs. S1 to S14
Tables S1 to S7
References (35–53)
View/request a protocol for this paper from Bio-protocol.
27 August 2019; resubmitted 30 March 2020
Accepted 11 June 2020
10.1126/science.aaz2532
Tarrason Risa et al., Science 369, eaaz2532 (2020)
7 August 2020
8 of 8
RESEARCH
|
RESEARCH ARTICLE
Downloaded from http://science.sciencemag.org/ on August 6, 2020

----!@#$NewPage!@#$----
−mediated cell division in an archaeon
The proteasome controls ESCRT-III
Saric, Ann-Christin Lindås, Nicholas P. Robinson and Buzz Baum
Mairi Kilkenny, Diorge P. Souza, Luca Pellegrini, Robertus A. M. de Bruin, Ricardo Henriques, Ambrosius P. Snijders, Andela
Papatziamou, Delyan R. Mutavchiev, Catherine Fan, Leticia Meneguello, Andre Arashiro Pulschen, Gautam Dey, Siân Culley, 
Gabriel Tarrason Risa, Fredrik Hurtig, Sian Bray, Anne E. Hafner, Lena Harker-Kirschneck, Peter Faull, Colin Davis, Dimitra
DOI: 10.1126/science.aaz2532
369 (6504), eaaz2532.
Science 
Science, this issue p. eaaz2532
having its origins in Archaea.
of the remaining ESCRT-III−based copolymer. These data strengthen the case for the eukaryotic cell division machinery
−based division ring, where rapid proteasome-mediated degradation of one ESCRT-III subunit triggered the constriction 
archaeal ESCRT-III homolog. Cell division in this archaeon was driven by stepwise remodeling of a composite ESCRT-III
archaeal relative of eukaryotes, Tarrason Risa et al. identified a role for the proteasome in triggering cytokinesis by an 
cytokinesis, a process that culminates in abscission dependent on the protein ESCRT-III. By studying cell division in an 
In eukaryotes, proteasome-mediated degradation of cell cycle factors triggers mitotic exit, DNA segregation, and
Proteasomal control of division in Archaea
ARTICLE TOOLS
http://science.sciencemag.org/content/369/6504/eaaz2532
MATERIALS
SUPPLEMENTARY 
http://science.sciencemag.org/content/suppl/2020/08/05/369.6504.eaaz2532.DC1
REFERENCES
http://science.sciencemag.org/content/369/6504/eaaz2532#BIBL
This article cites 53 articles, 12 of which you can access for free
PERMISSIONS
http://www.sciencemag.org/help/reprints-and-permissions
Terms of Service
Use of this article is subject to the 
Science, 1200 New York Avenue NW, Washington, DC 20005. The title Science is a registered trademark of AAAS.
(print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement of
Science 
Science. No claim to original U.S. Government Works
Copyright © 2020 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of
Downloaded from http://science.sciencemag.org/ on August 6, 2020