, 1250 (2009);
et al. Jun Zhang,
as an Effector Protein
Rab35 Controls Actin Bundling by Recruiting Fascin
www.sciencemag.org (this information is current as of September 4, 2009 ):
The following resources related to this article are available online at
version of this article at:
including high-resolution figures, can be found in the online
Updated information and services,
can be found at:
Supporting Online Material
, 12 of which can be accessed for free:
cites 25 articles
This article appears in the following
in whole or in part can be found at:
permission to reproduce
of this article or about obtaining
Information about obtaining
registered trademark of AAAS.
is a Science 2009 by the American Association for the Advancement of Science; all rights reserved. The title
Copyright American 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 September 4, 2009
0.049, P = 0.117 at 0.001 PT and Pearson’s cor-
This suggests that regulatory complexity correlates
with transcript complexity. Single-eQTL SNPs
were also found to influence expression of mul-
tiple genes. At the 0.001 (and 0.01) PT, over 6%
(19%) of eQTL SNPs were associated with the
expression of more than one gene (fig. S10).
the properties of cell type–specific and cell type–
shared genes. We found an overrepresentation of
functions linked to signal transducer activity, cell
communication, development, behavior, cellular
For eQTLs shared in all cell types, we found an
overrepresentation of catalytic activity and transport
properties (Fisher’s exact test, P < 0.05) (table S5).
We have demonstrated that variants affecting
gene regulation act predominantly in a cell type–
specific manner, and even cell types as closely re-
cis eQTLs. We estimate that 69 to 80% of regu-
latory variants are cell-type specific and that reg-
complexity, which implies that there are genotype-
addition, cell type–specific eQTLs have smaller
effects and tend to localize at greater distances
from the TSS, recapitulating enhancer element
distributions. The signal of cell-type specificity
was shown to be primarily due to differential use
of regulatory elements of genes that are expressed
in almost all cell types. As more tissues are inter-
set of informative tissues for the majority of reg-
ulatory variants. Our study highlights the need
differential functional properties.
References and Notes
1. A. S. Dimas et al., PLoS Genet. 4, e1000244 (2008).
2. A. L. Dixon et al., Nat. Genet. 39, 1202 (2007).
3. H. H. Goring et al., Nat. Genet. 39, 1208 (2007).
4. M. Morley et al., Nature 430, 743 (2004).
5. E. E. Schadt et al., Nature 422, 297 (2003).
6. B. E. Stranger et al., PLoS Genet. 1, e78 (2005).
7. B. E. Stranger et al., Nat. Genet. 39, 1217 (2007).
8. M. D. Adams et al., Nature 377, 3 (1995).
9. A. Reymond et al., Nature 420, 582 (2002).
10. A. I. Su et al., Proc. Natl. Acad. Sci. U.S.A. 99, 4465 (2002).
11. V. Emilsson et al., Nature 452, 423 (2008).
12. A. J. Myers et al., Nat. Genet. 39, 1494 (2007).
13. E. E. Schadt et al., PLoS Biol. 6, e107 (2008).
14. D. J. Campbell, S. F. Ziegler, Nat. Rev. Immunol. 7, 305
15. C. J. Cotsapas et al., Mamm. Genome 17, 490 (2006).
16. C. Wu et al., PLoS Genet. 4, e1000070 (2008).
17. International HapMap Consortium et al., Nature 449,
18. B. E. Stranger et al., Science 315, 848 (2007).
19. Materials and methods are available as supporting
material on Science Online.
20. E. L. Heinzen et al., PLoS Biol. 6, e1 (2008).
21. J. D. Storey, R. Tibshirani, Proc. Natl. Acad. Sci. U.S.A.
100, 9440 (2003).
22. H. H. Goring, J. D. Terwilliger, J. Blangero, Am. J. Hum.
Genet. 69, 1357 (2001).
23. J. P. Ioannidis, Epidemiology 19, 640 (2008).
24. K. E. Lohmueller, C. L. Pearce, M. Pike, E. S. Lander,
J. N. Hirschhorn, Nat. Genet. 33, 177 (2003).
25. J. B. Veyrieras et al., PLoS Genet. 4, e1000214 (2008).
26. G. A. McVean et al., Science 304, 581 (2004).
27. E. Birney et al., Nature 447, 799 (2007).
28. M. Ashburner et al., Nat. Genet. 25, 25 (2000).
29. We thank N. Hammond for technical help. We acknowledge
financial support from the Wellcome Trust and NIH to
E.T.D. and Infectigen Foundation, Swiss National Science
Foundation and AnEUploidy EU to S.E.A. Gene expression
data are deposited in NCBI’s Gene Expression Omnibus
under accession number GSE17080.
Supporting Online Material
Materials and Methods
Figs. S1 to S10
Tables S1 to S5
27 March 2009; accepted 14 July 2009
Published online 30 July 2009;
Include this information when citing this paper.
Rab35 Controls Actin Bundling by
Recruiting Fascin as an Effector Protein
Jun Zhang,1Marko Fonovic,2,3Kaye Suyama,1Matthew Bogyo,2Matthew P. Scott1*
Actin filaments are key components of the eukaryotic cytoskeleton that provide mechanical
structure and generate forces during cell shape changes, growth, and migration. Actin filaments
are dynamically assembled into higher-order structures at specified locations to regulate diverse
functions. The Rab family of small guanosine triphosphatases is evolutionarily conserved and
mediates intracellular vesicle trafficking. We found that Rab35 regulates the assembly of actin
filaments during bristle development in Drosophila and filopodia formation in cultured cells. These
effects were mediated by the actin-bundling protein fascin, which directly associated with active
Rab35. Targeting Rab35 to the outer mitochondrial membrane triggered actin recruitment,
demonstrating a role for an intracellular trafficking protein in localized actin assembly.
require polymerization of globular actin mono-
mersintofilaments and bundling ofthe filaments
under the control of actin-binding proteins (ABPs).
Certain ABPs cross-link filamentous actin (F-actin)
dynamic actin network is required for
normal cell morphology, cell locomotion,
and cytokinesis (1, 2). These processes
tural integrity of the cell and are structural compo-
actin at the right times and places during develop-
ment, physiological stresses, injury, and disease.
The importance of F-actin bundling during de-
velopment is readily apparent during bristle forma-
tion in Drosophila. Bristles are mechanosensory
organs found in genetically controlled locations on
chaetae, are formed by a “shaft” cell that extrudes
a cytoplasmic extension. This extension contains
located just beneath the plasma membrane (Fig. 1,
F and G). Bristle morphologies reflect the organi-
zation of actin bundles and can be used to study
the regulation of actin in vivo.
family small guanosine triphosphatases (GTPases).
Rab proteins control formation, motility, and dock-
ing of vesicles in specific trafficking pathways
(8, 9) by recruiting specific effector proteins to
different membrane compartments. Rab proteins
are evolutionarily conserved: Each of the >70 types
ilaRab GTPasessystematicallyfortheir abilities to
influence fly development, with the use of domi-
nant negative (DN) mutant proteins produced in
specific cell types (10). Rab activities are controlled
by a cycle of associations with GTP or guanosine
diphosphate (GDP). The DN mutants contained a
Thr/Ser → Asn mutation that causes the proteins
to bind preferentially to GDP and remain inactive.
acting proteins such as Rab exchange factors in
Only one of the 31 Drosophila Rab proteins
producing DN Rab35 (Rab35DN) in the periph-
eral nervous system (driven by prospero-gal4)
exhibited unique and specific bristle morphology
defects not seen with any other Rab DN gene.
Rab35DN caused the development of adult bristles
that had sharp bends, kinks, and forked ends in the
thorax (Fig. 1, A to E) and other body regions in-
expressing flies, stained with the actin-binding dye
phalloidin at 45 to 47 hours after puparium forma-
1Departments of Developmental Biology, Genetics, and Bio-
engineering and Howard Hughes Medical Institute, Stanford
University School of Medicine, Stanford, CA 94305, USA.
2Department of Pathology and Department of Microbiology
and Immunology, Stanford University School of Medicine,
Stanford, CA 94305, USA.3Department of Biochemistry,
ulica 39, 1000 Ljubljana, Slovenia.
*To whom correspondence should be addressed. E-mail:
4 SEPTEMBER 2009 VOL 325
on September 4, 2009
tion, had a wavy, loose, and disconnected actin
wild-type control (Fig. 1G and fig. S2A). Similar
phenotypes are observed in mutants deficient for
certain ABPs (5–7), which suggests that Rab35
may function as an ABP or through ABP(s). In
the thoracic cuticle, production of Rab11DN or
Rab5DN, which block Rab proteins that regulate
endocytic trafficking, caused extensive defects in
membrane growth and bristle distribution; these
phenotypes are distinct from the Rab35DN effect
scripts and proteins especially abundant in the de-
veloping nervous system (fig. S1, C to H).
To test whether the defective-bristle actin phe-
notype was due to reduced Rab35 function, we
expressed UAS-Rab35 hairpin RNA interference
The RNAi caused the same phenotypes as did
Rab35DN (Fig. 1P), which confirmed that the DN
gene (Fig. 1Q). In an otherwise wild-type genetic
notype as did flyRab35DN(fig.S4C) in the periph-
eral nervous system. Thus, at least some functions
of Rab35 protein are conserved from flies to mam-
mals, an evolutionary span of ~500 million years.
tured cells induced multiple filopodia-like cellular
extensions (Fig. 1J and figs. S3, A, B, C, D, and I,
and S4E). No such effects were seen upon expres-
S4F). Similarly, Rab35 induces peripheral processes
and N1E-115 cells (12). The effect of extra Rab35
on cultured cells might reflect its role in vivo, al-
lowing shaft cells to sprout protrusions during bris-
tle development. Rab35DN, in contrast, stopped
expression driven by tubulin-gal4 caused lethality
Treating Drosophila S2 cells with the actin
polymerization inhibitor latrunculin A, but not
with the microtubule-disrupting agent nocodazole,
efficiently blocked the Rab35-driven morphology
change(Fig.1,Lto N).Thus,Rab35 appearedto
regulate actin assembly.
skeleton,we setout toidentifyeffector proteinsthat
diverse functions in vesicle sorting, motor protein
binding, vesicle trafficking, membrane fusion (13),
and other roles yet to be defined. We used affinity
chromatography to purify proteins that preferentially
bind Rab35-GTP, which has been used to identify
other Rab effectors (14). Purified glutathione S-
transferase (GST)–tagged Rab35WT protein was
loaded with guanosine 5´-O-(3´-thiotriphosphate)
(GTP-g-S) or with GDP and incubated with bovine
to bind Rab35–GTP-g-S specifically (fig. S5A and
table S1). Mass spectrometry revealed a prominent
55-kD polypeptide that bound Rab35–GTP-g-S to
be fascin. Myc-tagged Rab35 and FLAG-tagged
fascin coimmunoprecipitated from cell extracts.
Rab35DN (Fig. 2A). Purified GST-Rab35 fusion
protein bound fascin in vitro (Fig. 2B). Thus,
Rab35 binds fascin directly. Fascin bound more
to Rab35WT preloaded with GDP in immunopre-
in GST pull-downs (fig. S5, B and C), consistent
with the identification of fascin as a Rab35 effector.
protruding (filopodia) and nonprotruding (micro-
Higher than normal fascin levels have been asso-
ciated with cancer cell migration, so the protein has
a therapeutic target (17–19). Fascin is produced in
many tissues and is especially abundant in the ner-
vous system (20). In Drosophila, fascin mutants
(called singed) are female sterile and have ab-
errant mechanosensory bristles (21, 22) due to
Fig. 1. ReducingRab35function causesmorphologicaldefects in bristledevel-
opment. (A to E) Rab35DN expressed in the peripheral nervous system induced
sharp bends, kinks, and forked ends [arrows in (B), (D), and (E)] in bristles
cells (red) assemble and extend 7 to 11 plasma membrane–associated actin
shaft. (G and H) Phalloidin staining of packed actin bundles in developing
caused wavy, loose, thin actin structures (H) relative to controls (G) at 45 to
Rab35WT (J) but not YFP-Rab35DN (K) induced filopodia-like membrane protrusions when expressed in S2 cells. Green, YFP proteins; red, phalloidin (filamentous
cells (L). Treatment with nocodazole did not block the morphological changes (N). Scale bar, 5 mm. (O to Q) Flies expressing Rab35RNAi in the peripheral nervous
system had abnormal bristle morphology (P) relative to controls (O). A mouse Rab35WT transgene reversed this phenotype (Q). Scale bar, 0.2 mm.
VOL 325 4 SEPTEMBER 2009
on September 4, 2009
dysfunctional actin structures. The Drosophila egg
chamber is composed of a germline cyst surrounded
of 15 nurse cells and one oocyte. Cortical cytoskel-
etal structures are required during late oogenesis
when nurse cell cytoplasm is rapidly transferred
to the oocyte. The sterility phenotype of singed
led us to examine the influence of Rab35DN in
nurse cells (driven by tubulin-gal4 at 22°C) or
follicle cells (driven by CY2-gal4). Both caused
female sterility (94%). The interfering Rab35DN
trol flies, and ovary structure was abnormal (fig.
S6, D, F, H, and J).
We tested whether the physical interaction be-
tween Rab35 and fascin was reflected in a genetic
interaction. Rab35RNAi, produced in peripheral
Altered bristle morphology was suppressed when
Increased fascin compensated for reduced Rab35
function, which suggests that fascin is at least one
of the major proteins regulated by Rab35.
Purified GST-fascin and GST-Rab35 were
mixed together or separately with purified non-
muscle F-actin in vitro. Actin-bundling activity
increased with fascin concentration (fig. S7, A
observed, alone or in combination with fascin
(fig. S7, A and C). Thus, Rab35 has no discern-
ible effect on actin bundling in vitro, but its asso-
ciation with fascin may be a means to control
when or where actin is bundled in vivo.
Perhaps activated Rab35 recruits fascin to a
subcellular location where fascin stimulates actin
bundling. To explore this idea, we first examined
the relative locations of Rab35 in different cell
types and its association with fascin in mamma-
Fig. 2. Rab35 directly
interacts with fascin. (A)
HeLa cells were cotrans-
fected with myc-Rab35
and FLAG-tagged fascin.
Cell lysates were sub-
jected to immunoprecipi-
tation with antibody to
myc, followed by immu-
noblotting with monoclonal antibody to FLAG
todetect fascin (top panel).Lower panel:Rab35
was detected with antibodies to myc. At least
6 times as much fascin (6.59 T 0.63) associated
with Rab35WT as with Rab35DN (n = 3, P <
proteins with purified fascin. At least 3 times as
muchfascin(3.27T 0.19)bound toRab35WTas
toRab35DN (n=3,P<0.05).Nobindingwasobservedbetweenfascinand Rab5orRab2.CoomassieBlue–
stained GST proteins were loading controls. (C) Transgenic flies expressing UAS-Rab35RNAi alone, and flies
expressing both UAS-fascin and UAS-Rab35RNAi, were crossed to prospero-gal4, which triggers expression in
flies expressing UAS-Rab35RNAi (arrows in left panel) only. Scale bar, 0.2 mm.
Input (5%) IP myc
1 2 3 4
1 2 3 4 5
Total lysate (10%)Membrane
1 2 3 4 5 6 7 8
Na /K -ATPase
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Fig. 3. Rab35 associates with fascin near the plasma membrane. (A to C) NIH
3T3 cells expressing YFP-Rab35 [(C) and green in (A)] were costained with actin
[(B) and red in (A)] to show the colocalization of the two proteins at plasma
I) NIH 3T3 cells producing either YFP-Rab35 WT [(F) and green in (D)] or YFP-
Rab35DN [(I) and green in (G)] were costained with antibodies to fascin [(E) and
(H) and red in (D) and (G)] to show the colocalization of the Rab35 and fascin
proteins. Scale bars, 5 mm. Insets in (A), (D), and (G) show higher-magnification
views of the edges of cells. (J) Rab35WT enrichment in a membrane fraction of
compared to the fractionation properties of Rab35DN (lane 8, upper panel;
33.4% increase in membrane-associated Rab35 normalized to the total Rab35
lower panel) than with Rab35DN (lane 8, lower panel; 17.7% enrichment of
membrane-associated fascin with Rab35WT normalized to the total fascin level);
10% of total protein lysates were loaded as controls (lanes 1 to 4). (K)
Subcellular fractionation of NIH 3T3 cells. (a) Three markers were used to show
(J) were fractionated. Rab35WT and fascin fractionated mainly with the plasma
but not with the endoplasmic reticulum marker calnexin or the mitochondria
From left to right, 2.5 to 25% iodixanol gradients.
4 SEPTEMBER 2009 VOL 325
on September 4, 2009
and colocalized with fascin at the leading edge of
filopodia and within microspikes in lamellipodia
of cellular structure and accumulated to a greater
extent in the cytosol than along the membrane
(Fig. 3, G to I, and figs. S6H and S8B). In the
presence of Rab35DN, less membrane-associated
of fascin fractionated with the membrane fraction
(Fig. 3J). The association of fascin with the mem-
brane fraction was at least partially dependent on
Rab35, because less Rab35 and less fascin asso-
was expressed. Thus, Rab35 may bring fascin to
structure and initiate filament bundling.
If fascin mediates the effects of Rab35, then
Rab35-induced filopodia formation should be
reduced when fascin function is blocked. Fascin
was depleted in Rab35-expressing cells by in-
troducing fascin small hairpin RNA (16). The
interfering RNA significantly blocked filopodia
formation induced by Rab35 (fig. S9, B and C).
Phosphorylation of fascin at Ser39is important
for its actin-bundling activity and proper local-
ization to filopodia (15, 16). We produced point
mutants that mimic the active dephosphorylated
(Ser39→ Ala, S39A) or inactive phosphorylated
(Ser39→ Asp, S39D) forms of fascin (15, 16).
Expression of tdTomato-tagged S39A or S39D
fascin mutants in NIH 3T3 cells along with
Rab35 had opposite effects on filopodia forma-
tion (figs. S9, D to F, and S10, A to I). The S39A
mutant in combination with Rab35WT caused
more protein to accumulate at the tip of cell
extensions (figs. S9E and S10, D to F) relative to
wild-type fascin plus Rab35WT, but no signifi-
cant increase in the number of filopodia was
observed (fig. S10J). In contrast, S39D in com-
bination with Rab35 reduced the number of
filopodia (fig. S10J). In these cells most of the
consistent with a role for fascin as a downstream
effector of Rab35 in filopodia formation.
We constructed a gene encoding a modified
form of Rab35 targeted to the outer mitochon-
drial membrane, a location that normally has
was used in this experiment, the modified protein
was on the surface of mitochondria (Fig. 4F). No
discernible change in actin structure was ob-
served in the vicinity of mitochondria (Fig. 4H
and fig. S11, C and D). In contrast, when
Rab35WT was brought to the mitochondrial
were consistently decorated with increased actin
meshworks (Fig. 4Land fig.S11, A and B). As a
control, Rab5WT targeted to mitochondria in the
same manner did not cause actin assembly in the
vicinity of mitochondria (Fig. 4, M to P). Fascin
relocation to mitochondria was confirmed by
cell fractionations. Mitochondrial enrichment
of fascin was observed when Rab35WT-mito
was produced (Fig. 3K). Thus, relocation of
Rab35 can drive the location of actin assembly.
Our results show that a Rab35 effector protein,
fascin, is able to stimulate local actin bundling and
thus control bristle and filopodia formation (fig.
S11E). The exact ways in which such a mecha-
nism may be used probably vary among cell and
tissue types. In cultured cells, active Rab35 recruits
cell protrusions. During Drosophila bristle devel-
opment, Rab35 may recruit fascin and induce actin
bundling to initiate the cytoplasmic extension re-
tion leads to bends and kinks in the bristles.
Conflicting results about Rab35 function have
been obtained from different cell types and mod-
in Drosophila S2 cells (23); in HeLa cells, Rab35
also plays a role in cytokinesis (23). No cyto-
kinesis phenotype was observed in mutants of
Caenorhabditis elegans Rab35, but Rab35 trans-
ports yolk receptors in oocytes (24). In HeLa-
complex (MHC) class II is expressed, Rab35 reg-
ulates a recycling pathway in a clathrin-, AP2-,
and dynamin-independent manner (25). In Jurkat
T cells, Rab35-mediated recycling appears to be
clathrin-dependent (11). In PC12 and N1E-115
via a Cdc42-dependent pathway (12). Drosophila
Rab35, like mammalian Rab35, is found near the
plasma membrane, on intracellular vesicles, and in
tion contrasts with more discrete locations of other
suggests that Rab35 may have diverse functions.
Interfering with Rab35 in living flies showed
its function in regulating actin assembly. The
powerful influence of Rab35 on the cytoskeleton
can now be at least partly explained by localization
of fascin and its consequent influence on actin fila-
References and Notes
1. F. A. Barr, U. Gruneberg, Cell 131, 847 (2007).
2. E. S. Chhabra, H. N. Higgs, Nat. Cell Biol. 9, 1110 (2007).
3. M. S. Mooseker, Cell 35, 11 (1983).
Fig. 4. Localized stimulation of actin bundling by Rab35 recruitment of fascin. (A to D) NIH 3T3 cells were
stainedwith MitoTracker(Invitrogen;green) to show mitochondria andactin (red).Box in(C) is shown in (D)at
higher magnification. Arrowheads indicate areas surrounding mitochondria. Scale bar, 5 mm. (E to P) NIH 3T3
cells producing YFP-Rab35DN-mito, YFP-Rab35WT-mito, and YFP-Rab5-mito (all in green) were stained with
phalloidin (red) to detect actin. In 20 to 30% of cells, localization of Rab35WT to mitochondria induced actin
(O) are shown in (D), (H), (L), and (P) at higher magnification. Arrowheads indicate areas surrounding
mitochondria. Detectable accumulation of actin near mitochondria was observed only in (L). Scale bars, 5 mm.
VOL 3254 SEPTEMBER 2009
on September 4, 2009
4. J. Faix, K. Rottner, Curr. Opin. Cell Biol. 18, 18 (2006). Download full-text
5. G. M. Guild, P. S. Connelly, L. Ruggiero, K. A. Vranich,
L. G. Tilney, J. Cell Biol. 162, 1069 (2003).
6. L. G. Tilney, P. Connelly, S. Smith, G. M. Guild, J. Cell
Biol. 135, 1291 (1996).
7. L. G. Tilney, P. S. Connelly, L. Ruggiero, K. A. Vranich,
G. M. Guild, Mol. Biol. Cell 14, 3953 (2003).
8. S. Pfeffer, Biochem. Soc. Trans. 33, 627 (2005).
9. S. Pfeffer, D. Aivazian, Nat. Rev. Mol. Cell Biol. 5, 886
10. J. Zhang et al., Genetics 176, 1307 (2007).
11. G. Patino-Lopez et al., J. Biol. Chem. 283, 18323 (2008).
12. J. Chevallier et al., FEBS Lett. 583, 1096 (2009).
13. B. L. Grosshans, D. Ortiz, P. Novick, Proc. Natl. Acad.
Sci. U.S.A. 103, 11821 (2006).
14. S. Christoforidis, M. Zerial, Methods Enzymol. 329, 120
15. J. C. Adams et al., Mol. Biol. Cell 10, 4177 (1999).
16. D. Vignjevic et al., J. Cell Biol. 174, 863 (2006).
17. N. A. Bakshi, W. G. Finn, B. Schnitzer, R. Valdez,
C. W. Ross, Arch. Pathol. Lab. Med. 131, 742 (2007).
18. H. Zhang et al., J. Clin. Pathol. 59, 958 (2006).
19. E. Kostopoulou et al., Histol. Histopathol. 23, 935 (2008).
20. A. De Arcangelis, E. Georges-Labouesse, J. C. Adams,
Gene Expr. Patterns 4, 637 (2004).
21. L. G. Tilney, M. S. Tilney, G. M. Guild, J. Cell Biol. 130,
22. K.Cant, B. A. Knowles, M. S. Mooseker, L. Cooley,J. Cell Biol.
125, 369 (1994).
23. I. Kouranti, M. Sachse, N. Arouche, B. Goud, A. Echard,
Curr. Biol. 16, 1719 (2006).
24. M. Sato et al., EMBO J. 27, 1183 (2008).
25. E. Walseng, O. Bakke, P. A. Roche, J. Biol. Chem. 283,
26. We thank M. Fish for DNA injections; X. Huang,
E. Bustamante, and C. Gauthier for help with initial
experiments; Scott lab members for valuable discussion
and comments; the Stanford Cell Sciences Imaging Facility
for assistance with scanning electron microscopy studies;
and S. Pfeffer, A. Ghabrial, and R. Rohatgi for critical
reading and comments on the manuscript. Supported by a
Jane Coffin Childs Memorial Fund for Medical Research
fellowship (J.Z.) and by the NIH National Technology
Center for Networks and NIH Pathway grant U54
RR020843 (M.F. and M.B.). The research reported here
was supported by the Howard Hughes Medical Institute.
M.P.S. is an Investigator of the HHMI.
Supporting Online Material
Materials and Methods
Figs. S1 to S12
13 April 2009; accepted 10 July 2009
Regulation of Histone Acetylation in the
Nucleus by Sphingosine-1-Phosphate
Nitai C. Hait,1Jeremy Allegood,1Michael Maceyka,1Graham M. Strub,1
Kuzhuvelil B. Harikumar,1Sandeep K. Singh,1Cheng Luo,2,3Ronen Marmorstein,2
Tomasz Kordula,1Sheldon Milstien,4Sarah Spiegel1*
The pleiotropic lipid mediator sphingosine-1-phosphate (S1P) can act intracellularly
independently of its cell surface receptors through unknown mechanisms. Sphingosine kinase 2
(SphK2), one of the isoenzymes that generates S1P, was associated with histone H3 and produced
S1P that regulated histone acetylation. S1P specifically bound to the histone deacetylases
HDAC1 and HDAC2 and inhibited their enzymatic activity, preventing the removal of acetyl
groups from lysine residues within histone tails. SphK2 associated with HDAC1 and HDAC2 in
repressor complexes and was selectively enriched at the promoters of the genes encoding the
cyclin-dependent kinase inhibitor p21 or the transcriptional regulator c-fos, where it enhanced
local histone H3 acetylation and transcription. Thus, HDACs are direct intracellular targets of
S1P and link nuclear S1P to epigenetic regulation of gene expression.
surface receptors. The recent identification of
nuclear lipid metabolism has highlighted a new
signaling paradigm for phospholipids. The best
characterized of the intranuclear lipids are the
inositol lipids that have critical roles in nuclear
functions, such as pre-mRNA splicing, mRNA
export, transcriptional regulation, and chromatin
remodeling (1). Sphingomyelin has long been
known to be a component of the nuclear matrix
(2), but the possibility that sphingolipids are also
metabolized within the nucleus has only recently
hospholipid and sphingolipid metabolites
have established roles in signal transduc-
ling sphingolipid metabolism, including neutral
sphingomyelinase and ceramidase, are also
present in the nucleus (3).
Sphingosine-1-phosphate (S1P) is a sphin-
golipid metabolite that regulates many cellular
survival, movement, angiogenesis, vascular mat-
uration, immunity, and lymphocyte trafficking
(4–6). Most of its actions are mediated by bind-
ing to a family of five heterotrimeric guanine
nucleotide–binding protein (G protein)-coupled
receptors, designated S1P1-5(5). S1P may also
function inside the cell independently of S1P re-
ceptors (4); however, direct intracellular targets
of S1P have not been identified. Since the dis-
closely related sphingosine kinase isoenzymes,
SphK1 and SphK2, much has been learned about
SphK1 and its functions, yet those of SphK2
remain enigmatic (7).
Because, in many cells, SphK2 is mainly lo-
calized to the nucleus or can shuttle between the
cytosol and the nucleus in accordance with its nu-
breast cancer cells, SphK2 is predominantly lo-
calized to the nucleus (10), where it is enzymati-
(Fig. 1A). The nucleus contained high amounts of
sphingosine (table S1), and SphK2 expression
significantly increased nuclear abundance of S1P
by sixfold and dihydro-S1P, which lacks the trans
Endogenous SphK2 was mainly associated with
isolated chromatin and was not detected in the
nucleoplasm (Fig. 1B). SphK2 was present in
purified mononucleosomes fractionated by su-
Thus, we tested whether SphK2 was associated
with core histone proteins. Immunoprecipitation
of proteins from nuclear extracts prepared from
MCF-7 cells overexpressing SphK2 or SphK1
and subsequent Western blot analysis demon-
strated that histone 3 (H3) was associated with
SphK2 (Fig. 1C), but not with SphK1. H3 was
also coimmunoprecipitated with catalytically
inactive SphK2G212E(in which glycine 212 is
replaced by glutamic acid) (Fig. 1C), which does
not increase nuclear S1P (4.4 T 0.3 and 4.6 T 0.6
these findings suggest that enzymatic activity is
H3 from nuclear extracts also associated with his-
tidine (His)–tagged SphK2 isolated with Ni2+–
nitrilotriacetic acid (NTA)–agarose beads (fig.
S1B). To further confirm the specificity of the
physical interaction of SphK2 with H3, we mea-
sured in vitro association of Ni-NTA-agarose–
bound His-tagged SphK2 or His-SphK1 with
individually purified histones. Only H3, but not
histones H4, H2B, or H2A, bound to SphK2 (fig.
S1C). In contrast, none of the purified histones
interacted with SphK1 (fig. S1C).
acetylation of lysine 9 of H3 (H3-K9), lysine 5 of
(H2B-K12) (Fig. 2A), without affecting acetyla-
In contrast, expression of catalytically inactive
SphK2G212E(Fig. 2A) or SphK1 did not influence
acetylation of any residues examined. Although
catalytically inactive SphK2G212Ealso bound to
1Department of Biochemistry and Molecular Biology and
the Massey Cancer Center, Virginia Commonwealth Uni-
versity School of Medicine, Richmond, VA 23298, USA.
2The Wistar Institute and Department of Chemistry, Uni-
versity of Pennsylvania, Philadelphia, PA 19104, USA.
3State Key Laboratory of Drug Research, Shanghai Institute
of Materia Medica, Chinese Academy of Sciences, Shanghai
201203, P. R. China.4National Institute of Mental Health,
National Institutes of Health, Bethesda, MD 20892, USA.
*To whom correspondence should be addressed. E-mail:
4 SEPTEMBER 2009VOL 325
on September 4, 2009