polarized myosin produces unequal-size daughters during
TRANSCRIPT
DOI: 10.1126/science.1196112, 677 (2010);330 Science
, et al.Guangshuo OuAsymmetric Cell DivisionPolarized Myosin Produces Unequal-Size Daughters During
This copy is for your personal, non-commercial use only.
clicking here.colleagues, clients, or customers by , you can order high-quality copies for yourIf you wish to distribute this article to others
here.following the guidelines
can be obtained byPermission to republish or repurpose articles or portions of articles
): January 25, 2011 www.sciencemag.org (this infomation is current as of
The following resources related to this article are available online at
http://www.sciencemag.org/content/330/6004/677.full.htmlversion of this article at:
including high-resolution figures, can be found in the onlineUpdated information and services,
http://www.sciencemag.org/content/suppl/2010/09/28/science.1196112.DC1.htmlcan be found at: Supporting Online Material
http://www.sciencemag.org/content/330/6004/677.full.html#relatedfound at:
can berelated to this article A list of selected additional articles on the Science Web sites
http://www.sciencemag.org/content/330/6004/677.full.html#ref-list-1, 8 of which can be accessed free:cites 20 articlesThis article
http://www.sciencemag.org/content/330/6004/677.full.html#related-urls1 articles hosted by HighWire Press; see:cited by This article has been
http://www.sciencemag.org/cgi/collection/cell_biolCell Biology
subject collections:This article appears in the following
registered trademark of AAAS. is aScience2010 by the American Association for the Advancement of Science; all rights reserved. The title
CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theScience
on
Janu
ary
25, 2
011
ww
w.s
cien
cem
ag.o
rgD
ownl
oade
d fr
om
17. T. Adilakshmi, D. L. Bellur, S. A. Woodson, Nature 455,1268 (2008).
18. A. E. Bunner, S. Nord, P. M. Wikström, J. R. Williamson,J. Mol. Biol. 398, 1 (2010).
19. M. R. Sharma et al., Mol. Cell 18, 319 (2005).20. T. Ruiz, M. Radermacher, Methods Mol. Biol. 319,
403 (2006).21. This work was supported by the NIH (grants RR175173,
R37-GM-53757, and GM-52468) and the NSF(graduate research fellowship to A.M.M.). The EM data
collection and analysis were conducted at theNational Resource for Automated Molecular Microscopy,which is supported by the NIH through the NationalCenter for Research Resources P41 program (grantsRR175173 and RR023093). We thank N. R. Vossand members of the Automated Molecular Imagingand Williamson laboratories for critical discussionof experiments. MAP, structure factors, and layer-linedata have been deposited at EMDB with accession codeEMD-1783.
Supporting Online Materialwww.sciencemag.org/cgi/content/full/330/6004/673/DC1Materials and MethodsSOM TextFigs. S1 to S9Table S1References
3 June 2010; accepted 9 September 201010.1126/science.1193220
Polarized Myosin ProducesUnequal-Size Daughters DuringAsymmetric Cell DivisionGuangshuo Ou,*† Nico Stuurman, Michael D’Ambrosio, Ronald D. Vale†
Asymmetric positioning of the mitotic spindle before cytokinesis can produce different-sized daughtercells that have distinct fates. Here, we found an asymmetric division in the Caenorhabditis elegans Qneuroblast lineage that began with a centered spindle but generated different-sized daughters, thesmaller (anterior) of which underwent apoptosis. During this division, more myosin II accumulatedanteriorly, suggesting that asymmetric contractile forces might produce different-sized daughters. Indeed,partial inactivation of anterior myosin by chromophore-assisted laser inactivation created a moresymmetric division and allowed the survival and differentiation of the anterior daughter. Thus, thebalance of myosin activity on the two sides of a dividing cell can govern the size and fate of the daughters.
Most somatic cell divisions equally parti-tion the cytoplasm and produce equiva-lent daughters.Asymmetric cell divisions
are used to generate distinct daughter cells that have
different fates in development and in adult stem celllineages (1, 2). Asymmetric cell division has beenbest studied in the first embryonic cell division inCaenorhabditis elegans (3). Here, unequal dynein-
mediated pulling forces in the anterior-posterioraxis displace the spindle toward the posterior pole(3–5). The cleavage furrow then forms in the mid-dle of the elongating anaphase spindle, but be-cause the spindle is displaced, the cell is dividedinto unequal-size daughters (3, 5). However, inDrosophila neuroblasts, asymmetric cell divisionbegins with the spindle aligned in the middle ofthe cell (6–8). As anaphase progresses, the spindleelongates asymmetrically and the cytokinetic fur-row shifts toward one side of the cell. However,the cellular mechanism responsible for this type ofasymmetric cytokinesis is unknown.
We developed fluorescence markers and liveimaging methodologies to study asymmetric di-
The Howard Hughes Medical Institute and the Department ofCellular and Molecular Pharmacology, University of California,San Francisco, San Francisco, CA 94158, USA.
*Present address: National Laboratory of Biomacromolecules,Institute of Biophysics, Chinese Academy of Sciences, 15 DatunRoad, Chaoyang District, Beijing 100101, China.†To whom correspondence should be addressed. E-mail:[email protected] (G.O.); [email protected] (R.D.V.)
A B
Q
Q.a
Q.ap Q.paQ.aa Q.pp
Q.paa Q.pap
A/PQR
A/PVM SDQL/R
Q.p
QR.a
CentrosomeMembrane Histone
QR.p
A P
QR.a
CentrosomeMembrane Histone
0 sec
40
80
QR.pD
. .Pa PpCa Cp
Sec0
4080
120
0
4080
120
QR.a(n=10)
QR.p(n=10)
Cen
tros
ome
- C
ente
r D
ista
nce
Ant
erio
r/P
oste
rior
0
0.4
0.8
1.2
QR.a QR.p
C
E
Fig. 1. C. elegans Q neuroblast lineage and spindle positioningin asymmetric cell division. (A) Three rounds of asymmetric celldivisions make three different neurons (QL generates PQR, PVM,and SDQL, whereas QR generates AQR, AVM and SDQR) and twoapoptotic cells (in black). (B) Spindle positioning in the second-round division with centrosomes (g-tubulin, TBG-1::GFP) in greenand plasma membrane [mCherry with a myristoylation signal(12)] and histone (his-24::mCherry) in red. The cell anterior istoward the left. Bar, 2.5 mm. Upper right corners show aschematic summary of the results: QR.p spindle is displacedposteriorly but QR.a spindle is centered. (C) Centrosome–to–cell center distance ratio (anterior centrosome distance dividedby posterior centrosome distance); data are shown as the mean TSD (n = 12). Cell center is defined as the midpoint between thetwo cell poles. (D) QR.a (left) and QR.p (right) spindles duringanaphase and cytokinesis. (E) The distance of the centrosome tothe cleavage furrow (Ca, anterior; Cp, posterior) or to the cellpole (Pa, Pp). The distance is quantified from 10 movies and isnormalized compared to the initial distances for each centrosomeat the start of anaphase. Raw data are shown in fig. S1.
www.sciencemag.org SCIENCE VOL 330 29 OCTOBER 2010 677
REPORTS
on
Janu
ary
25, 2
011
ww
w.s
cien
cem
ag.o
rgD
ownl
oade
d fr
om
visions in the C. elegans Q neuroblast lineage. Qneuroblasts undergo three rounds of division tomake three distinct neurons and two apoptoticcells (Fig. 1A). In the second round, asymmetricdivisions give rise to a large cell that continues todivide and differentiate and a small cell that un-dergoes apoptosis (Fig. 1A) (9).
We first examined the spindle positioning ofthe QR.p and QR.a cells at metaphase by measur-ing the distances from each centrosome [markedby g-tubulin–green fluorescent protein (GFP)] tothe center of the cell. In QR.p, the metaphasespindle was displaced toward the posterior; thedistance from the anterior centrosome to the cellcenter was 0.4 times as long as that from the pos-terior centrosome to the cell center (Fig. 1, B andC, and movie S1). At anaphase, the cleavage fur-row formed equidistant between the two centro-somes. As the spindle elongated, the distances ofthe anterior and posterior centrosomes to the fur-row increased in a similar manner (Fig. 1, D andE, fig. S1, and movie S1); as the spindle was dis-placed before anaphase, a large and a small daugh-ter cell were produced when the cytokinetic furrowbisected the spindle. The behavior of the QL.pdivision was similar to that of QR.p (fig. S2, Aand B). Thus, this asymmetric division appears tobe similar to the first cell division in C. elegansembryogenesis (3, 5).
The asymmetric division of the QR.a cell,however, was more similar to that described forDrosophila neuroblasts. In theQR.a cell, anaphasebegan with the two centrosomes positioned equi-distant to the center [imaged with GFP–g-tubulin(Fig. 1, B andC, andmovie S2)] or GFP–a-tubulin(fig. S3 and movie S3). As anaphase progressed,the distance from the anterior centrosome to theingressing cleavage furrow remained constant,whereas the posterior centrosome–to–furrow dis-tance increased (Fig. 1,D andE, fig. S1, andmovieS2). This progressively developing asymmetry thatemerges in a cell with an initially centered spindleraises the possibility that a nonspindle factor maybe driving the asymmetry in daughter cell size.
We next examined the dynamics of GFP-labeled nonmuscle myosin II in the contractile ringduring cytokinesis. In the QR.p cell (posterior-displaced spindle),myosin II (NMY-2 inC. elegans)was equally distributed on the anterior and poste-rior sides of the ingressing furrow throughout cyto-kinesis (Fig. 2, A and B, fig. S4A, and movie S4).Myosin was depleted at both the anterior and pos-terior poles during telophase, as in somatic cell di-vision (10). However, in the QR.a cell, where thecleavage furrow is initiated at the cell center, thedistribution of myosin became asymmetric duringanaphase. More cortical myosin was found at theanterior than at the posterior side of the furrow,particularly at early stages of anaphase, and corticalmyosin was often found at the anterior pole of theQR.a cell, which was rarely seen in the QR.p cell(Fig. 2, A and B, movie S5, and fig. S4). We alsoobserved a similar myosin II asymmetry during theasymmetric cell division in QL.a but not in the QL.p lineage (fig. S2, C and D).
An asymmetric distribution of myosin in theQR.a cells might create a “tense” anterior cortexthat resists deformation and a “relaxed” posteriorcortex that can deform and expand. To explorethis idea, we examined the shape of the plasmamembrane (using anmCherry-taggedplasmamem-brane marker) in the anterior and posterior halvesof the dividing cell. For the QR.p cell, its anteriorpole is the leading edge of cell migration beforeand after cell division (QR.pa cell). The anteriormembrane of the QR.p cell was more dynamicand more ruffled than the posterior portion frommetaphase and throughout cytokinesis (Fig. 2C,movie S6, and fig. S6B). As the spindle elongatedin anaphase, the centrosome–to–cell pole dis-tances decreased equally in the anterior and pos-terior (Fig. 1, D and E, andmovie S1), and the sizeof the anterior half was constantly ~2.2 times aslarge as that of the posterior throughout cyto-kinesis (fig. S5). However, a very different behav-ior was observed for the QR.a cell. Like that of theQR.p cell, the anterior half of theQR.a cell was theleading edge ofmigration before division, andwasmore dynamic and formed more protrusions thanthe posterior portion through metaphase (Fig. 2C,movie S7, and fig. S6A). However, unlike that ofthe QR.p cell, the anterior membrane became lessdynamic and protrusive at anaphase and shrunk
inward, resulting in a gradual decrease of the an-terior pole–to–centrosome distance and also thesize of the anterior daughter (Fig. 1, D and E, figs.S1 and S5, and movies S2 and S7). Furthermore,the membrane in the posterior pole expanded out-ward during anaphase, resulting in an increase inthe posterior pole–to–centrosome distance and theoverall size of the posterior daughter (Fig. 2C, figs.S1 and S5, and movie S7). Thus, higher levels ofcortical tension and contractile forces may drivethe membrane toward the centrosome in the an-terior half and lower cortical tensionmay allow theexpansion of the membrane away from the cen-trosome in the posterior half of the QR.a cell.
To further explorewhethermyosin II asymmetryis involved in generating the QR.a daughter cellsize asymmetry, we examined GFP–myosin II inthe pig-1 (gm344) mutant [PIG-1, a MELK-likekinase, is needed for asymmetric cell division in Qneuroblasts but not in the early embryo (11)]. Inthe pig-1 mutant, GFP–myosin II was distributedsymmetrically on the anterior and posterior sidesof the furrow during QR.a division (Fig. 2, A andB, fig. S4, and movie S8), consistent with a rolefor myosin in the asymmetric cell division of theQR.a cell. However, because the pig-1 mutantalso affects the asymmetric division of the QR.pcell, which occurs by spindle displacement rather
B
QR.a QR.pWT pig-1 WT pig-10
0.5
1
1.5
2
2.5
**
Myo
sin
II R
atio
(A
nter
ior/
Pos
terio
r)
A QR.pQR.a QR.a (pig-1)
*
0 sec
Myosin IIMembrane Histone
80
160
A P
QR.a
C
QR.p 300
QR.a
A P
** **0 Sec 100 300200
* * * *
Fig. 2. Myosin II and membrane dynamics in Q neuroblasts during cytokinesis. (A) Still images of myosinII–GFP dynamics during cytokineses of QR.a in wild-type (WT) or QR.a in pig-1 mutant and QR.p animals.(B) Myosin II–GFP fluorescence intensities ratio between the anterior and posterior parts of QR.a or QR.pcells in WT or pig-1 (gm344) mutant (*P < 0.001, n = 9 to 11); data are shown as the mean T SD. (C)Membrane dynamics of QR.a or QR.p during cytokinesis. The plasma membrane is imaged as in Fig. 1.Green dots are autofluorescent spots in the C. elegans body that provide fiducial marks (original image notshown). Panels on the far right show the alignment of QR.a and QR.p cell peripheries (0 s red, 300 s green)revealing the contraction and expansion of the two sides of the cell. Asters show neighboring Q cells. Moreexamples are shown in fig. S6. Anterior of the cell is toward the left. Bar, 2.5 mm.
29 OCTOBER 2010 VOL 330 SCIENCE www.sciencemag.org678
REPORTS
on
Janu
ary
25, 2
011
ww
w.s
cien
cem
ag.o
rgD
ownl
oade
d fr
om
than asymmetric myosin, this correlation doesnot definitively link the polarization of myosin IIwith asymmetric cell division. To obtain moredirect evidence, we inactivated myosin II activityusing chromophore-assisted laser inactivation[CALI (12)] in the anterior of the QR.a cell atthe onset of cytokinesis and determined whetherthis perturbation changed the sizes of the daugh-ters. CALI uses an intense laser irradiation of afluorophore, including GFP, to damage proteinswithin ~4 nm by generating highly reactive hy-droxyl radicals (13–15). GFP-based CALI of themyosin II regulatory light chain has been shown tospecifically block furrow ingression during cyto-kinesis in Drosophila epithelial cells (16). Sim-ilarly, in the first-round division of the QR cell,we found that CALI illumination of GFP–myosinII applied to one side of the furrow caused theingression on that side to pause while the other,nonilluminated side continued to pinch (fig. S7B,73%, n = 11). Thus, CALI can inhibit the myosinII activity during cytokinesis in the Q cell.
We next applied CALI to the anterior half ofQR.a cell, which had higher GFP–myosin II, andthen measured the sizes of the daughters. AfterCALI, the anterior daughter cell QR.aa was en-larged [QR.aa/QR.ap size ratio of 0.83 T 0.18(mean T SD, n = 39) compared to normal (0.52 T0.08, n = 35) (Fig. 3, A and C)]. Furthermore, theQR.aa cell was sometimes larger than the QR.apcell after CALI treatment, which was not ob-served in thewild type (Fig. 3C). To establish thatthese results were due to inactivation of myosinand not to nonspecific effects of illumination, weperformed a similar CALI treatment of the GFP-tagged membrane protein MIG-2, a small guano-sine triphosphatase involved in Q cell migrationbut not in cytokinesis (17, 18). Performing CALIof MIG-2::GFP in the anterior of the QR.a cellat anaphase did not significantly change theQR.aa/QR.ap cell size ratio (0.56 T 0.07, n = 21)(Fig. 3, E and F). We also performed CALI ofGFP-myosin II in the posterior half of the QR.pcell, whose asymmetric division appears to in-volve spindle displacement rather than asymmetricmyosin (Figs. 1 and 2). In this case, CALI onlymarginally changed the QR.pp/QR.pa size ratio[(0.47 T 0.06, n= 21) versus untreated (0.42 T 0.06,n = 24); Fig. 3, B and D]. Thus, CALI of GFP-tagged myosin II specifically affects the daughtercell sizes of the QR.a division.
We next followed the fate of the QR.aa cellafter CALI by long-term time lapse imaging. Nor-mally, the QR.aa cell becomes engulfed and di-gested by neighboring cells within 150 min (Fig.4, A and B, and movie S9) (9). In contrast, theQR.aa cell derived from the QR.a cell after CALItreatment sometimes escaped apoptosis (21%, n=19) and could differentiate into a neuronal-likecell that extended a long process (11%) (Fig. 4, Aand B, and movie S10). The larger QR.aa cellsthat formed after CALI were the ones that had ahigher chance of survival (fig. S7C). The differ-entiation into a neuronal-like cell appears to be anoutcome associatedwith survival rather thanwith
A P
10 min
99 min
126 min
QR.aaQR.ap
QR.p
QR.aa
QR.pp
QR.pp
MembraneHistone
WT
QR.aa QR.ap
QR.aa
QR.aa
QR.ap
QR.ap
QR.pa
QR.pa
QR.pp
QR.pp
25 min
82 min
189 min
CALI
0
20
40
60
80
100
Apoptosis Survive
Cell Fates in WT (N=21)
Per
cent
age
A B
Cell Fates after CALI (N=19)
C Metaphase Early Cytokinesis Late Cytokinesis Death Survival
Fig. 4. Cell fate after CALI and a proposed model. (A) QR.aa cells normally undergo apoptosis(disappearance at 126 min). After CALI, QR.a cell migrated (82 min, the nonmotile QR.pp provides afiduciary marker) and formed a neurite-like process (189 min). (B) Quantifications of QR.aa cell fates (12)in WT (red) and after CALI (blue) ~150min after birth. Anterior of the cell is toward the left. Bar, 5 mm. (C)Proposed mechanism of QR.aa asymmetric division. Myosin II localizes evenly in QR.a cell at metaphasebut distributes asymmetrically during anaphase. More myosin II at the anterior than at the posterior maygenerate larger cortical tension (long arrows), pushing cellular contents and resulting in a small cell in theanterior that undergoes apoptosis and a large cell in the posterior that survives. Myosin II (green); plasmamembrane and chromosomes (red); centrosome and microtubules (blue).
0.2 0.4 0.6 0.8 1.0 1.20
3
6
9
12
15
18
0.2 0.4 0.6 0.8 1.0 1.20
3
6
9
12
15
18
0
3
6
9
12
15
18
0.2 0.4 0.6 0.8 1.0 1.2
AQR.a Myosin II CALI
A P A P
MembraneHistone Membrane Histone
MembraneHistone
QR.p Myosin II CALI
A P
QR.a MIG-2 CALI
B
C D
E F
Cyt
okin
esis
Eve
nts
Cyt
okin
esis
Eve
nts
Cyt
okin
esis
Eve
nts
Daughter Cell Size Ratio (Anterior/Posterior) Daughter Cell Size Ratio (Anterior/Posterior)
Daughter Cell Size Ratio (Anterior/Posterior)
WT CALIQR.a WT CALIQR.p
QR.aMIG-2::GFP CALI
Myosin II Myosin II
WT
MIG-2
0 Sec 50 Sec 160 Sec
0 Sec 45 Sec 100 Sec0 Sec 75 Sec 155 Sec
Fig. 3. Chromophore-assisted laser inactivation (CALI) of myosin II in Q neuroblasts during cytokinesis.(A) CALI inactivation of myosin II in the anterior of QR.a enlarges the anterior daughter cell size. Upperimages show the chromosome and plasma membrane labeled with mCherry, and lower images show GFP-tagged myosin II. Arrows indicate the region that was treated with CALI at late anaphase (12). (B) CALIinactivation of myosin II in the posterior of QR.p does not change sizes of daughter cells. (C) and (D) arequantifications of the above CALI experiments. (E and F) CALI of MIG-2::GFP does not change the ratio ofQR.aa and QR.ap. Anterior of the cell is toward the left. Bar, 5 mm.
www.sciencemag.org SCIENCE VOL 330 29 OCTOBER 2010 679
REPORTS
on
Janu
ary
25, 2
011
ww
w.s
cien
cem
ag.o
rgD
ownl
oade
d fr
om
CALI per se, because a similar differentiation phe-notype was observed when QR.aa apoptosis wasblocked in a ced-4 mutant (n1162) (fig. S8).
Our results show that asymmetric cortical ten-sion, produced by an unequal distribution of my-osin and perhaps other cortical proteins, producedtwo different-sized QR.a daughter cells. We pro-pose that a stiffer and inward-contracting anteriorpole pushes cytoplasm through the furrow towardthe less contractile posterior pole, which respondsby expanding like a balloon (Fig. 4C). We alsodemonstrate a direct connection between the sizeof a cell and its fate inC. elegans development byshowing that increasing cell size by myosin IICALI can result in a rescue from apoptosis anddifferentiation into a neuronal-like cell.
The downstream “output” of asymmetric cor-tical signaling (3, 5) has been thought to be themitotic spindle, best exemplified by the displace-ment of mitotic spindle in the C. elegans embryo(3, 5). Similarly, asymmetric divisions of Dro-sophila neuroblasts (which start anaphase with asymmetric, centrally positioned spindle) have beenpostulated to arise from the greater elongation ofmicrotubules on one side of the spindle midzonecompared with the other (6–8). We also observedasymmetric spindle elongation in the QR.a cell(Fig. 1, D and E), but propose that this occurssecondarily to myosin polarization, in which thespindle elongates preferentially toward the pole
with lower cortical tension. In support of thisidea, increasing cortical tension at the cell polesprevents spindle elongation and cell extension individingDrosophila S2 cells (19). Consequently,we propose that a polarization of myosin-basedcontractility is the primary driver of the QR.a celldivision asymmetry, although we cannot rule outsome contribution from the spindle as well. Thus,a simple organism like C. elegans [in which 807out of its 949 somatic cell divisions are asymmetric(20)] appears to use at least two physical mecha-nisms (one spindle-based and one cortex-based) togenerate asymmetric cell division.
Note added in proof: Cabernard et al. (21)recently showed an asymmetric distribution ofmyosin II during cytokinesis of Drosophilaneuroblasts.
References and Notes1. K. A. Moore, I. R. Lemischka, Science 311, 1880 (2006).2. R. A. Neumüller, J. A. Knoblich, Genes Dev. 23, 2675 (2009).3. S. Q. Schneider, B. Bowerman, Annu. Rev. Genet. 37,
221 (2003).4. P. Gönczy, S. Pichler, M. Kirkham, A. A. Hyman, J. Cell Biol.
147, 135 (1999).5. P. Gönczy, Nat. Rev. Mol. Cell Biol. 9, 355 (2008).6. K. H. Siller, C. Q. Doe, Nat. Cell Biol. 11, 365 (2009).7. J. A. Kaltschmidt, C. M. Davidson, N. H. Brown, A. H. Brand,
Nat. Cell Biol. 2, 7 (2000).8. Y. Cai, F. Yu, S. Lin, W. Chia, X. Yang, Cell 112, 51 (2003).9. J. E. Sulston, H. R. Horvitz, Dev. Biol. 56, 110 (1977).10. R. D. Vale, J. A. Spudich, E. R. Griffis, J. Cell Biol. 186,
727 (2009).
11. S. Cordes, C. A. Frank, G. Garriga, Development 133,2747 (2006).
12. Materials and methods are available as supportingmaterial on Science Online.
13. T. J. Diefenbach et al., J. Cell Biol. 158, 1207 (2002).14. K. Jacobson, Z. Rajfur, E. Vitriol, K. Hahn, Trends Cell
Biol. 18, 443 (2008).15. F. S. Wang, J. S. Wolenski, R. E. Cheney, M. S. Mooseker,
D. G. Jay, Science 273, 660 (1996).16. B. Monier, A. Pélissier-Monier, A. H. Brand, B. Sanson,
Nat. Cell Biol. 12, 60 (2010).17. G. Ou, R. D. Vale, J. Cell Biol. 185, 77 (2009).18. I. D. Zipkin, R. M. Kindt, C. J. Kenyon, Cell 90, 883 (1997).19. G. R. Hickson, A. Echard, P. H. O’Farrell, Curr. Biol. 16,
359 (2006).20. H. R. Horvitz, I. Herskowitz, Cell 68, 237 (1992).21. C. Cabernard, K. E. Prehoda, C. Q. Doe, Nature 467,
91 (2010).22. We thank L. Cai, D. Sherwood, J. Ziel, P. Zhang,
I. Cheeseman, C. Kenyon, G. Garriga, and CaenorhabditisGenetics Center for reagents, equipments and strains,and thank C. Locke and W. Li for myosin quantifications.This work was supported by the Damon Runyon CancerResearch Foundation (G.O.) and the NIH and HowardHughes Medical Institute (R.D.V.).
Supporting Online Materialwww.sciencemag.org/cgi/content/full/science.1196112/DC1Materials and MethodsFigs. S1 to S8Table S1ReferencesMovies S1 to S10
5 August 2010; accepted 20 September 2010Published online 30 September 2010;10.1126/science.1196112Include this information when citing this paper.
A Size Threshold Limits PrionTransmission and EstablishesPhenotypic DiversityAaron Derdowski,1* Suzanne S. Sindi,1,2*† Courtney L. Klaips,1Susanne DiSalvo,1 Tricia R. Serio1†
According to the prion hypothesis, atypical phenotypes arise when a prion protein adopts analternative conformation and persist when that form assembles into self-replicating aggregates.Amyloid formation in vitro provides a model for this protein-misfolding pathway, but the mechanismby which this process interacts with the cellular environment to produce transmissible phenotypesis poorly understood. Using the yeast prion Sup35/[PSI+], we found that protein conformationdetermined the size distribution of aggregates through its interactions with a molecular chaperone.Shifts in this range created variations in aggregate abundance among cells because of a sizethreshold for transmission, and this heterogeneity, along with aggregate growth and fragmentation,induced age-dependent fluctuations in phenotype. Thus, prion conformations may specifyphenotypes as population averages in a dynamic system.
Prion proteins adopt a spectrum of confor-mations or strains, which create pheno-types of distinct severity and stability in
vivo (1–3). These phenotypes are linked to theassembly of the protein into aggregates that, atdifferent rates, template the conversion of newlymade prion proteins to a similar state and arefragmented (4). But how do these biochemicalevents translate into distinct phenotypes? One pos-sibility is an “abundance-based” model, in whichphenotypes are linked to an equilibrium betweenaggregated and soluble prion protein that deter-mines protein activity and the number of heritableprions (propagons) (5, 6). However, the conversionand fragmentation reactions also create hetero-
geneity in aggregate size, raising the possibilityof a second, “size-based” model in which a sub-population of aggregates establishes and propagatesphenotypes (7).
To distinguish between these models, we fo-cused on the [PSI+]Weak and [PSI+]Strong conforma-tions of the yeast prion protein Sup35, which createphenotypes of different stabilities in vivo (8). Tosustain these phenotypes in a dividing culture,Sup35 protein in the prion conformation must beinherited (7). To test whether conformational dif-ferences affect phenotypic stability by alteringprotein transmissibility,wemonitored Sup35–greenfluorescent protein (GFP) transfer to daughter cells.Through use of fluorescence loss in photobleach-ing (FLIP), a [PSI+]Weak strain transferred half asmuch Sup35-GFP (~15%versus ~30%) (Fig. 1A)and contained ~50% fewer propagons than a[PSI+]Strong strain (Fig. 1B). Thus, conformation-based differences in protein transmission corre-late with phenotypic inheritance.
Following from the models, differences in ag-gregate abundance and/or size may create thisvariation in prion protein transmission (5, 7), andindeed, [PSI+]Weak strains differ from [PSI+]Strong
strains by the accumulation of fewer but larger ag-gregates (9–12). To distinguish between these pos-sibilities, we simulated prion propagation via eachtransmission mechanism (13). For the abundance-based model, we were unable to recapitulate theseverity and stability of [PSI+]Weak and [PSI+]Strong
phenotypes (fig. S5) (13). In contrast with a size-based model, we recapitulated all experimentally
1Department of Molecular Biology, Cell Biology and Biochem-istry, Brown University, 185 Meeting Street, Post Office BoxG-L2, Providence, RI 02912, USA. 2Center for Computa-tional Molecular Biology, Brown University, 115 WatermanStreet, Post Office Box 1910, Providence, RI 02912, USA.
*These authors contributed equally to this work.†To whom correspondence should be addressed. E-mail:[email protected] (S.S.S.); [email protected] (T.R.S.)
29 OCTOBER 2010 VOL 330 SCIENCE www.sciencemag.org680
REPORTS
on
Janu
ary
25, 2
011
ww
w.s
cien
cem
ag.o
rgD
ownl
oade
d fr
om