10.1101/gad.2021411Access the most recent version at doi:
2011 25: 569-580 originally published online February 28, 2011Genes Dev.
Jie Yao, Richard D. Fetter, Ping Hu, et al.
myogenesis
Subnuclear segregation of genes and core promoter factors in
Material
Supplemental http://genesdev.cshlp.org/content/suppl/2011/02/28/gad.2021411.DC1.html
References
http://genesdev.cshlp.org/content/25/6/569.full.html#related-urls
Article cited in:
http://genesdev.cshlp.org/content/25/6/569.full.html#ref-list-1
This article cites 61 articles, 19 of which can be accessed free at:
Open Access Freely available online through the Genes & Development Open Access option.
service
Email alerting
click heretop right corner of the article or
Receive free email alerts when new articles cite this article - sign up in the box at the
http://genesdev.cshlp.org/subscriptions
go to: Genes & DevelopmentTo subscribe to
Copyright © 2011 by Cold Spring Harbor Laboratory Press
Cold Spring Harbor Laboratory Press on March 16, 2012 - Published by genesdev.cshlp.orgDownloaded from
Today
Typewritten Text
干细胞
Subnuclear segregation of genes and core
promoter factors in myogenesis
Jie Yao,1,2 Richard D. Fetter,1 Ping Hu,2 Eric Betzig,1 and Robert Tjian1,2,3
1Janelia Farm Research Campus, The Single Cell Biochemistry Consortium, Howard Hughes Medical Institute, Ashburn,
Virginia 20147, USA; 2Department of Molecular and Cell Biology, Li Kashing Center For Biomedical and Health Sciences, CIRM
Center of Excellence, University of California at Berkeley, Berkeley, California 94720, USA
Recent findings implicate alternate core promoter recognition complexes in regulating cellular differentiation.
Here we report a spatial segregation of the alternative core factor TAF3, but not canonical TFIID subunits, away
from the nuclear periphery, where the key myogenic gene MyoD is preferentially localized in myoblasts. This
segregation is correlated with the differential occupancy of TAF3 versus TFIID at theMyoD promoter. Loss of this
segregation by modulating either the intranuclear location of the MyoD gene or TAF3 protein leads to altered
TAF3 occupancy at the MyoD promoter. Intriguingly, in differentiated myotubes, the MyoD gene is repositioned
to the nuclear interior, where TAF3 resides. The specific high-affinity recognition of H3K4Me3 by the TAF3 PHD
(plant homeodomain) finger appears to be required for the sequestration of TAF3 to the nuclear interior. We
suggest that intranuclear sequestration of core transcription components and their target genes provides an
additional mechanism for promoter selectivity during differentiation.
[Keywords: transcription; nucleus; MyoD; core promoter factors; superresolution microscopy]
Supplemental material is available for this article.
Received December 12, 2010; revised version accepted January 24, 2011.
The regulation of gene transcription plays a seminal role
in the development and differentiation of cell types in
multicellular organisms. Significant progress has been
made in the identification of transcription factors, and
genome-wide mapping of their cognate binding sites has
accelerated with the development of massively parallel
DNA sequencing capabilities (Farnham 2009). Despite this
rapid progress in dissecting the biochemistry of transcrip-
tion, the question of how these gene regulatory factors find
their target promoters in the cell nucleus remains poorly
understood. Genomic DNA in eukaryotic cells is com-
pacted by histone proteins to form chromatin—highly
folded and condensed protein/DNA structures inside
the nucleus (Cremer and Cremer 2001; Spector 2003).
Live-cell imaging analysis suggests that many transcrip-
tion factors rapidly diffuse across the nucleus and tran-
siently bind to their target genes (Darzacq et al. 2009;
Hager et al. 2009). Importantly, it has been recognized
that genes are nonrandomly distributed in the nucleus
and with respect to chromatin territories (Misteli 2007;
Kumaran et al. 2008; Sinclair et al. 2010), and that gene
activation and cellular differentiation may be accompa-
nied by gene repositioning (Moen et al. 2004; Chuang
et al. 2006; Meister et al. 2010). Although the position of
a gene in the nucleus does not obligatorily determine its
activity (Yao et al. 2007; Kumaran et al. 2008), transcrip-
tion factors must be able to navigate the cell nucleus and
access target genes in order to activate transcription. Thus,
an important but challenging question that has largely
escaped analysis is whether access and targeting of tran-
scription factors to specific nuclear subcompartments can
influence and regulate transcription output. From a techni-
cal standpoint, although live-cell imaging provides some
measurement of mobility and kinetics of populations of
transcription factor molecules in the nucleus, individual
transcription factor molecules are not readily visiblewithin
the context of nuclear architecture using conventional
light microscopy. Recent advances in fluorescence micros-
copy with single-molecule resolution provide us an oppor-
tunity to revisit this problem of transcription factor acces-
sibility and selective utilization at target gene promoters.
Another challenge to accurate subnuclear positioning
of regulatory factors is the relative paucity of spatial
landmarks within the nucleus. One readily recognizable
positional element is the nuclear periphery, demarcated
by structures at the inner surface of the nuclear envelope
(Akhtar and Gasser 2007; Lusk et al. 2007). Interactions of
chromatin domains with the nuclear lamina (NL) have
been identified during embryonic stem cell differentiation
3Corresponding author.
E-MAIL jmlim@berkeley.edu; FAX (510) 643-9547.
Article published online ahead of print. Article and publication date are
online at http://www.genesdev.org/cgi/doi/10.1101/gad.2021411. Freely
available online through the Genes & Development Open Access option.
GENES & DEVELOPMENT 25:569–580 � 2011 by Cold Spring Harbor Laboratory Press ISSN 0890-9369/11; www.genesdev.org 569
Cold Spring Harbor Laboratory Press on March 16, 2012 - Published by genesdev.cshlp.orgDownloaded from
(Peric-Hupkes et al. 2010). In yeast, components of the
nuclear periphery have been implicated in both repres-
sion and activation of gene transcription (Andrulis et al.
1998; Schmid et al. 2006). In mammalian cells, experi-
ments that tether reporter genes to the nuclear periph-
ery have resulted in differential expression of some, but
not all, reporters, as well as adjacent endogenous genes
(Finlan et al. 2008; Kumaran and Spector 2008; Reddy
et al. 2008). We speculate that identifying transcription
factors that exhibit differential access to the nuclear
periphery may be informative in exploring the potential
roles of nuclear organizations of genes and proteins as
a mechanism of gene control.
Our recent analysis of skeletal myogenesis suggests
that alternate core promoter recognition factors may play
a hitherto unappreciated role in regulating cell type-
specific transcription (Deato and Tjian 2007; Deato
et al. 2008; D’Alessio et al. 2009; Goodrich and Tjian
2010). During muscle formation, the MyoD gene is
expressed in both myoblasts and later in differentiated
myotubes and acts as a key regulatory factor driving
myogenic differentiation (Tapscott et al. 1988). In con-
trast, theMyogenin gene is turned on only after myocytes
exit the cell cycle and begin to fuse, thus activating genes
at later stages of differentiation (Edmondson and Olson
1989). Another recently uncovered transcriptional event
associated with skeletal myogenesis was the unexpected
loss of the prototypic core promoter recognition complex
TFIID (Deato and Tjian 2007). It had beenwell established
that, in rapidly growing cells, the multisubunit core
transcription complex TFIID is essential for promoter rec-
ognition and potentiating activated transcription from
yeast to humans (Naar et al. 2001). TAF3 is a substoichio-
metric TFIID subunit first identified in Drosophila
(Gangloff et al. 2001), and later shown to interact with
the histone mark H3K4Me3 (Vermeulen et al. 2007).
Remarkably, during myoblast-to-myotube differentiation
of mouse C2C12 cell culture and during muscle develop-
ment in vivo, TFIID is largely eliminated, and, instead,
TAF3 can be detected associated with the core promoter
of the late-expressing Myogenin gene (Deato and Tjian
2007). Curiously, in myoblasts, TFIID and TAF3 are both
present in the same nucleus, but how myogenic genes
differentially use TFIID versus TAF3 in myoblasts posed
an intriguing conundrum.
Here we tracked two key myogenic genes—MyoD and
Myogenin—and alternate core promoter recognition fac-
tors by fluorescence in situ hybridization (FISH), immu-
nofluorescence staining, and dual-color photoactivation
localization microscopy (PALM). By employing ‘‘super-
resolution’’ PALM-based cell imaging approaches, we
more precisely localized individual transcription factor
molecules within distinct nuclear regions. We found that,
in myoblasts, canonical TFIID subunits are present at the
nuclear periphery, where the MyoD gene preferentially
resides, while TAF3 is largely segregated from the nuclear
periphery; this differential subnuclear distribution of
TFIID versus TAF3 is correlated with their selective occu-
pancies at theMyoD promoter inmyoblasts. In contrast, in
myotubes, where TFIID is lost, MyoD becomes reposi-
tioned to the nuclear interior, where TAF3 resides, and this
is accompanied by an increased occupancy of TAF3 at the
MyoD promoter. Furthermore, by ectopically modulating
the locations of theMyoD promoter and/or TAF3 protein,
we show that their spatial segregation is functionally
linked to the selective occupancy of TAF3 at the MyoD
promoter. We also found that specific recognition and
high-affinity binding by the TAF3 plant homeodomain
(PHD) finger to the histone mark H3K4Me3 may be
required for the sequestration of TAF3 to the nuclear
interior. These studies suggest that differential nuclear
compartmentalization of target genes and regulatory fac-
tors may provide an additional mechanism for promoter
selectivity during differentiation of animal cells.
Results
Nuclear locations of key myogenic genes in myoblasts
To begin this study, we determined the positions ofMyoD
by immuno-DNA FISH in mouse C2C12 myoblasts.
Visualization of the nuclear periphery by an antibody
against nuclear Lamin B shows that the MyoD gene is
preferentially localized to the nuclear periphery (Fig. 1A,B;
Figure 1. FISH analysis of MyoD and Myoge-
nin gene loci in C2C12 myoblasts. (A,B) DNA
FISH of MyoD gene (red) in myoblasts. (A) The
nuclear periphery is highlighted by anti-Lamin
B (green). Other micrographs follow the same
color scheme. (B) Frequency histogram versus
the distance of MyoD genes from the NL. B, D,
and F are frequency histograms versus distance
from genes to the NL in the corresponding FISH
experiments shown in A, C, and E, respectively.
(C,D) RNA FISH of MyoD gene in myoblasts.
Kolgomorov-Smirnov (K-S) test of distributions
in B and D: P = 0.50. (E,F) DNA FISH of the
Myogenin gene (red) in myoblasts. K-S test of
distributions in B and F: P < 0.001. Fisher’s
exact tests of MyoD and Myogenin genes that
are located within 0.6-, 0.8-, or 1.0-mm distance
to the NL: P < 0.0001 in all cases. Bars, 5 mm.
Yao et al.
570 GENES & DEVELOPMENT
Cold Spring Harbor Laboratory Press on March 16, 2012 - Published by genesdev.cshlp.orgDownloaded from
Supplemental Fig. S1A), in agreement with a previous
study (Lee et al. 2006). Importantly, nascent MyoD tran-
scripts were visualized through immuno-RNA FISH, and
these transcripts were also located at the nuclear periph-
ery (Fig. 1C,D), confirming that the peripherally localized
MyoD genes are transcriptionally active. In contrast, the
Myogenin gene is inactive in myoblasts, and we found
that this ‘‘later’’ gene is located largely to the nuclear
interior in myoblasts (Fig. 1E,F; Supplemental Fig. S1B).
Chromatin immunoprecipitation (ChIP) assays confirm
that RNA polymerase II (Pol II) occupancy is enhanced at
the MyoD promoter relative to the MyoG promoter in
myoblasts (Supplemental Fig. S1C). These findings reveal
that two key temporally regulated myogenic genes are
differentially localized within the myoblast nucleus,
posing intriguing potential mechanisms for their differ-
ential regulation.
Localizing general transcription factors in myoblasts
The MyoD gene is actively transcribed in myoblasts by
the codependent action of upstream activators and req-
uisite core promoter recognition complexes (Hu et al.
2008). In the case ofMyoD transcription inmyoblasts, the
prototypic core factor TFIID occupies the MyoD pro-
moter. Because theMyoD gene is preferentially localized
at the nuclear periphery in myoblasts (Fig. 1), we set out
to visualize which components of the transcription
apparatus are colocalized at the nuclear periphery. We
investigated the localizations of Pol II and TFIID by
immunofluorescence staining and high-resolution multi-
color confocal microscopy. As expected, Pol II is diffusely
localized throughout the nucleoplasm, including the
zone at the nuclear periphery in myoblasts (Fig. 2A,B).
Furthermore, TAF11, TAF4, and TBP subunits of TFIID
Figure 2. Immunofluorescence staining of
several components of core transcription
machinery in C2C12 myoblasts. (A, panel i)
RNA polymerase II stained with 4H8 anti-
body (green) with anti-Lamin B (red). (Panel
ii) The image in panel i superimposed by
DNA staining with Hoechst 33342 (blue).
(Panel iii) The radial intensity plot inte-
grated over the entire contour of nuclear
lamin from representative images (n = 8).
Error bars are standard deviations. The same
organization follows for the rest of sub-
panels. (B) Ser5-phosphorylated RNA Pol II
(n = 3). (C) TAF11 (n = 3). (D) TAF4 (n = 3). (E)
TBP (n = 3). (F) TAF3 (n = 6). (G) Intensity
plots of antibody staining signals in com-
parison: (Panel i) TAF3, TAF4, and TAF11.
One-way ANOVA test of GFP intensity
values between 0 and 0.4 mm from the
lamina: P < 0.001. (Panel ii) TAF3, Pol II
(4H8), and Pol II (Ser5P): P < 0.04. (Panel iii)
TAF4, Pol II (4H8), and Pol II (Ser5P):
P < 0.04. Bars, 2 mm.
Subnuclear locations of core promoter factors
GENES & DEVELOPMENT 571
Cold Spring Harbor Laboratory Press on March 16, 2012 - Published by genesdev.cshlp.orgDownloaded from
are diffusely localized inside the nucleoplasm of myo-
blasts, and careful examination confirms that these
immunostaining signals are still detectable at the bound-
ary of the nucleus, labeled by an antibody against nuclear
Lamin B (Fig. 2C–E). Because TAF4 is an essential
component of the TFIID complex (Wright et al. 2006), it
is likely that holo-TFIID is not only distributed through-
out the nucleoplasm, but is also present at the nuclear
periphery. Our DNA staining profiles usually show a peak
close to the lamina signal (Supplemental Fig. S2), and this
observation is consistent with earlier studies visualizing
dense DNA structures at the nuclear periphery by elec-
tron microscopy (Davies 1967), and serves as a control to
ascertain appropriate alignment of multiple image chan-
nels in our analysis. It was reported that heterochromatin
regions allow the placement of macromolecules with
molecular weights of ;500 kDa (Bancaud et al. 2009),
which is consistent with our observations that TAF4 and
Pol II are present at the nuclear periphery.
In contrast to Pol II and TFIID, myoblasts stained with
TAF3 antibodies revealed a clearly distinguishable region
immediately adjacent to the NL with measurably lower
immunofluorescence levels (Fig. 2F). The various anti-
bodies against TAF3 used in these nuclear staining
studies were first affinity-purified and extensively char-
acterized (Supplemental Fig. S3A,B). Integrated radial
intensity profiles indicate that the distance of half-max-
imum of TAF3 signals to the center of the lamin signal
is ;400 nm (Fig. 2F, panel iii); this distance is larger than
those measured for TFIID subunits or Pol II (;100–200
nm) (Fig. 2A–E), and is well within our optical resolution
limits. We further found statistically significant differ-
ences among the mean intensity values at the nuclear
periphery between TAF3 versus TAF4/TAF11 subunits
(Fig. 2G, panel i), as well as between TAF3 versus Pol II
(Fig. 2G, panel ii). These cell imaging results are consis-
tent with our previous biochemical observation that
TAF3 is a substoichiometric subunit associated with
the TFIID complex (Liu et al. 2008). Interestingly, the
TAF4 signal appears to be distinct from that of Pol II and
distributes ‘‘closer’’ to the nuclear periphery (Fig. 2G,
panel iii). We also found that H3K9Me3 staining is
detectably enriched at the nuclear periphery in myo-
blasts, while H3K4Me3 staining is slightly shifted to
the nuclear interior (Supplemental Fig. S4A). Taken
together, our imaging analysis demonstrates significantly
differential distributions for core transcription compo-
nents at the nuclear periphery in C2C12 myoblasts, and
we observed that a shell or region directly adjacent to
the nuclear periphery is substantially depleted of TAF3
relative to the nuclear interior.
Extending our studies to living cells, we directly
visualized a distinct layer of reduced GFP-TAF3 fluores-
cence intensity at the nuclear periphery (Supplemental
Fig. S5A, panels iv,v). In contrast, GFP-tagged Rpb9,
TAF11, and human TAF1 each showed fluorescence rela-
tively uniformly distributed throughout the nucleoplasm,
including the region adjacent to or at the nuclear periphery
(Supplemental Fig. S5A, panels i–iii). These live-cell imag-
ing studies are consistent with our immunofluorescence
results, and support the notion that the lower levels of
TAF3 observed at the nuclear periphery likely reflect its
steady-state intranuclear distributions in living myo-
blasts. Immunostaining of TRF3 proteins suggests that
they might also distribute to the nuclear interior (Sup-
plemental Fig. S3F). In primary myoblasts, the MyoD
gene is also largely localized adjacent to the nuclear
periphery and the Myogenin gene is largely localized
adjacent to the nuclear interior, while TAF3 is localized
largely to the nuclear interior and shows no signifi-
cant difference from its distribution in C2C12 cells
(Supplemental Fig. S6). There is also a high enrichment
of H3K9Me3 and a slight decrease of Pol II levels at
the nuclear periphery (Supplemental Fig. S6). Hence,
the spatial distributions of MyoD gene and core tran-
scription components in the in vivo-derived primary
cells are largely consistent with our findings in C2C12
myoblasts.
The observation that Pol II and canonical TFIID sub-
units are present at the nuclear periphery, where TAF3
is underrepresented, supports the notion that the alter-
native core factor TAF3 may be spatially segregated
from the peripherally localized, actively transcribedMyoD
gene. This intriguing finding leads us to speculate whether
the subnuclear segregation of TAF3 from the MyoD gene
may influence or perhaps preclude its association with the
MyoD promoter.
PALM imaging of transcription factor localization
in myoblasts
The nuclear periphery is a challenging subdomain of the
nucleus to target for optical microscopy because its radial
dimensions and lateral microdomains approach the dif-
fraction limit of conventional optical microscopy. With
three-dimensional (3D) structured illumination micros-
copy, nuclear periphery components could be better
resolved in both radial and lateral dimensions compared
with confocal or deconvolution microscopy (Schermelleh
et al. 2008). However, precisely comparing locations of
distinct transcription factors at the nuclear periphery
would best be served by methods that can visualize and
resolve individual molecules within the context of the
nucleus. We adapted dual-color PALM (Betzig et al. 2006;
Bates et al