Crecimiento celular y síntesis proteica

Leading Edge
Minireview
Cell 133, May 16, 2008 ©2008 Elsevier Inc. 577
The rate of cell growth and proliferation is proportional to
the rate of protein synthesis, which is in turn tightly linked to
ribosome biogenesis. The synthesis of rRNA, the rate-limiting
step in ribosome synthesis, is regulated by nutritional
conditions. In this way, the cell strikes a balance between
protein synthesis and the energetically costly investment of
biosynthetic resources in manufacturing ribosomes. To keep
up with the demand of ribosome production, eukaryotes
maintain several hundred genes encoding rRNAs that are
transcribed by RNA polymerase I (Pol I). Pol I transcription
is intricately regulated to be responsive to general metabolism
and specific environmental challenges (for review, see
Grummt, 2003). Transcription of rDNA can be modulated
by varying the transcription rate per gene or by varying the
proportion of active genes. The currently accepted model
for regulation of rDNA transcription posits two overlapping
mechanisms (Figure 1). For short-term regulation in
response to growth factor signaling, nutrients, or stress, the
transcription rate at euchromatic “active” rDNA is altered.
In contrast, when more stable rDNA transcription is needed
(for example, when cell-specific ratios of active versus silent
rDNA copies are established during development), then
rDNA transcription is regulated epigenetically, that is, by
chromatin modifications. These two mechanisms of transcriptional
and epigenetic control have complicated efforts
to identify the major pathways contributing to proliferationdependent
and metabolism-dependent regulation of rDNA
transcription.
A Metabolic Throttle Regulates the
Epigenetic State of rDNA
Ingrid Grummt1,* and Andreas G. Ladurner2,*
1Department of Molecular Biology of the Cell II, German Cancer Research Center, DKFZ-ZMBH Alliance, Im Neuenheimer Feld 581, 69120
Heidelberg, Germany
2Gene Expression Unit, European Molecular Biology Laboratory, Meyerhofstrasse 1, 69117 Heidelberg, Germany
*Correspondence: i.grummt@dkfz-heidelberg.de (I.G.), ladurner@embl.de (A.G.L.)
DOI 10.1016/j.cell.2008.04.026
The synthesis of ribosomal RNA (rRNA) is carefully tuned to match nutritional conditions. In this
issue, Murayama et al. (2008) describe a mechanism that couples the energy status of the cell
to heterochromatin formation and silencing of rRNA genes. They show that an altered NAD+/
NADH ratio in response to glucose starvation regulates the silencing activity of eNoSC, a complex
consisting of the NAD+-dependent histone deacetylase SIRT1, the histone methyltransferase
SUV39H1, and a new protein called nucleomethylin (NML). These results suggest a mechanism
that links cell physiology to rDNA silencing, which in turn is a prerequisite for nucleolar integrity
and cell survival.
Figure 1. Energy-Dependent Regulation of
rDNA Transcription
Genes encoding ribosomal RNAs (rRNAs) exist
in two functionally distinct states. At active
genes, the promoter is unmethylated and associated
with euchromatic histone modifications
such as acetylated histone H3 (H3K9ac).
The TAFI68 subunit of the promoter selectivity
factor SL1 is also acetylated (Ac), facilitating
its binding to the rDNA promoter and the subsequent
assembly of productive transcription
initiation complexes (top left). At epigenetically
silent genes, the chromatin remodeling complex
NoRC, together with associated RNA (purple
line), recruits DNA methyltransferase and histone
deacetylase activities to the rDNA promoter,
leading to deacetylation and methylation
of H3K9, CpG DNA methylation at the promoter
(Me), and transcriptional silencing (top right).
Upon glucose starvation, elevation of the NAD+/
NADH ratio activates SIRT1. SIRT1 deacetylates
the TAFI68 subunit of SL1, impairing DNA
binding and formation of the transcription initiation
complex (bottom left). In addition, eNoSC, comprising SIRT1, SUV39H1, and NML, is activated leading to histone deacetylation, H3K9 methylation,
and transcriptional repression (bottom right). It is not clear whether the chromatin structure of transcriptionally active genes is altered under
low-glucose conditions.
578 Cell 133, May 16, 2008 ©2008 Elsevier Inc.
Turning rRNA Genes on and off
Although rRNA synthesis accounts for more than 50% of cellular
transcriptional activity, a significant fraction of the rDNA repeats
is constitutively silent (for review, see Grummt and Pikaard, 2003).
The technique of chromatin immunoprecipitation (ChIP) has provided
valuable insights into the transcriptional activity status of
rDNA repeats. These studies have revealed that specific histone
modifications distinguish silent or facultative heterochromatin
from transcriptionally permissive euchromatin. Active rRNA genes
exhibit euchromatic features including hypomethylation of the
promoter, acetylation of histone H3 and H4 tails, and methylation
of histone H3 lysine 4 (H3K4). In contrast, silent rDNA repeats are
located within chromatin regions refractory to transcription and
bearing heterochromatic marks such as di- and trimethylation of
histone H3 at lysine 9 (H3K9me2 and H3K9me3, respectively), at
lysine 20 (H3K20me3), and at lysine 27 (H3K27me3) (Santoro et
al., 2002; Yuan et al., 2007). Furthermore, the promoters of silent
rRNA genes are hypermethylated at CpG residues. The transition
of epigenetically active to silent rDNA repeats is mediated
by NoRC (Nucleolar Remodeling Complex), a SNF2h-containing
chromatin remodeling complex that recruits DNA methyltransferase
(DNMT) and histone deacetylase activities to the promoter,
thereby triggering heterochromatin formation and transcriptional
silencing (Santoro and Grummt, 2005). As a consequence of
NoRC’s interaction with DNMTs and with specific corepressors, a
subset of rDNA repeats is silenced and specific epigenetic marks
at the loci are propagated throughout cell division. The current
model for the mechanism of NoRC function suggests that NoRC
serves as a scaffold to coordinate the activities of macromolecular
complexes that modify histones, methylate DNA, and establish
a “closed” heterochromatin structure.
Studies in the budding yeast Saccharomyces cerevisiae and
the fruit fly Drosophila have demonstrated that rDNA silencing
plays an important role in genome stability through the
suppression of nonhomologous recombination pathways.
Loss of silencing at rDNA loci correlates with rDNA instability,
nucleolar disintegration, and cellular senescence (Kobayashi
et al., 2004; Peng and Karpen, 2007). Among the key players
that ensure rDNA stability are the NAD+-dependent histone
deacetylase Sir2 (silent information regulator 2) in S. cerevisiae
and the histone methyltransferase Su(var)3-9 (suppressor
of variegation 3-9) in Drosophila. The mutation of Su(var)3-9
results in decreased levels of H3K9me2, perturbation of nucleolar
structure, and accumulation of extrachromosomal rDNA
circles (Peng and Karpen, 2007). Likewise, mutations in Sir2
lead to increased rDNA instability and shortening of replicative
life span (Sinclair and Guarente, 1997). As Sir2 activity is regulated
by NAD+ levels, the finding that Sir2 plays an important
role in rDNA stability and nucleolar activity links rRNA synthesis
directly to the energy prosperity of the cell.
Linking Glucose to rDNA Transcription
One of the most important environmental variables in the regulation
of rDNA transcription is the availability of nutrients, such
that rRNA synthesis is tightly linked to the metabolic state of a
cell. It has been long recognized that a given nutritional state
gives rise to a cellular equilibrium in which the synthesis of ATP
and GTP is balanced by their use in protein synthesis (Grummt
and Grummt, 1976). Accordingly, rDNA transcription is regulated
by intracellular ATP levels, thus providing a molecular
explanation for the growth rate-dependent control and homeostatic
regulation of ribosome synthesis. Superimposed upon
this regulation is the deacetylation of TAFI68, a subunit of the
Pol I promoter selectivity factor (SL1 in humans and TIF-IB in
mice), by the NAD+-dependent deacetylase SIRT1, which leads
to transcriptional repression (Muth et al., 2001).
In their new study, Murayama and colleagues uncover an
additional relationship between cellular energy status and
rDNA transcription. They show that glucose starvation affects
the epigenetic state of rRNA genes, suggesting a fine-tuned
mechanism by which rDNA silencing may decrease energy
expenditure and protect cells from energy deprivation-induced
apoptosis. The authors identified a new protein complex,
eNoSC (energy-dependent Nucleolar Silencing Complex), that
changes the ratio of active to silent rRNA genes in response to
cellular energy status. Proceeding from the observation that
glucose starvation of cultured HeLa cells induced deacetylation
of H3K9 and elevated the level of H3K9me2 at the rDNA
promoter, the authors identified nucleomethylin (NML), a
nucleolar protein that specifically bound to H3K9me2. NML,
the product of the KIAA0409 gene, exhibits homology to
methyltransferases, is localized to nucleoli, and is associated
with chromatin throughout the rDNA repeats. Overexpression
and knockdown approaches showed that manipulating
the NML protein affected the abundance of rRNA precursors
and the level of H3K9 methylation at rDNA promoters. Specifically,
NML depletion increased pre-rRNA synthesis and
boosted histone H3 acetylation but decreased H3K9 methylation
at rDNA promoters. Conversely, overexpression of NML
decreased H3 acetylation and increased H3K9me2 at rDNA
promoters, resulting in the repression of pre-rRNA synthesis.
These results suggest that NML affects rDNA transcription by
modulating histone H3K9 methylation at the rDNA promoter,
thereby establishing heterochromatic features and promoting
rDNA silencing. Significantly, depletion of NML reduced the
ability of cells to decrease pre-rRNA synthesis and to maintain
ATP levels in response to glucose deprivation, indicating that
NML represses rDNA transcription during limiting metabolic
conditions. By limiting ribosome biogenesis, cellular ATP levels
may be maintained and cells protected from apoptosis due to
energy deprivation.
NML—An Unusual Histone-Binding Protein?
How NML links the cellular energy status with rDNA chromatin
remodeling in the nucleous is not fully clear, but the study
by Murayama and colleagues offers important clues. The
most direct link to metabolism comes from their crystal structure
of the NML C-terminal region. The structure revealed a
methyltransferase domain confirming that NML is structurally
similar to S-adenosyl methionine (SAM)-dependent methyltransferases.
The target of any NML-mediated enzyme activity
remains unknown, but the SAM-binding domain of NML does
show homology to the nucleolar protein Rrp8p of S. cerevisiae.
In budding yeast, Rrp8p interacts with a snoRNP component
and is important for the processing of 35S pre-rRNA
(Bousquet-Antonelli et al., 2000). Whether Rrp8p methylates
Cell 133, May 16, 2008 ©2008 Elsevier Inc. 579
rRNA or catalyzes methylation of nucleolar proteins that are
involved in pre-rRNA processing is not known. What is clear,
however, is that budding yeast lacks repressive H3K9 methylation
and consistent with this, Rrp8p does not have the unusual
H3K9me2-binding domain found in NML (which is different
from known methyl-lysine binding modules). Future studies
will address the molecular details of how this NML domain rich
in both low-complexity sequence and predicted disordered
regions specifically recognizes H3K9me2. Interestingly, Rrp8p
interacts with histone H2A and the chromatin remodeling complex
Isw1, suggesting that the SAM-binding methyltransferase
domain of Rrp8p (and possibly also of NML) is sufficient for
recruitment to chromatin. In support of this, cells expressing a
mutant NML protein defective in SAM binding cannot silence
rDNA and undergo energy deprivation-induced apoptosis.
NML Teams up with SIRT1 and SUV39H1
Pathways that are regulated by feedback loops tend to share
the general feature of having more than one entry point for
regulation. Considering that alteration of NML protein levels
affects histone H3 acetylation and H3K9 methylation at
rDNA repeats, it is perhaps not surprising that Murayama et
al. also found a molecular connection between NML and two
key epigenetic regulators of chromatin structure: the protein
deacetylase SIRT1 and the heterochromatic methyltransferase
SUV39H1. Treatment of cells with nicotinamide, an inhibitor of
NAD+-dependent deacetylases including SIRT1, suppressed
the NML-mediated decrease in pre-rRNA levels. Specific
knockdown of SIRT1, the principal NAD+-dependent deacetylase
for histones H4K16 and H3K9 and other proteins including
p53, TAFI68, MyoD, FOXO3, and PPARγ, prevented NMLdependent
transcriptional repression and increased H3K9
acetylation. Consistent with these observations suggesting a
specific involvement of SIRT1, overexpression of SIRT1 augmented
the repressive effect of NML. Moreover, the authors
found that NML and SIRT1 interacted with SUV39H1, thereby
linking histone H3 deacetylation to H3K9 methylation and rDNA
silencing.
Sir2 and the mammalian homolog SIRT1 have been shown
to deacetylate H3K9 at rDNA repeats (Liou et al., 2005). In S.
cerevisiae, the Sir2-containing RENT complex promotes Pol I
transcription and triggers the silencing of Pol II transcription at
the rDNA locus. However, Sir2 itself does not have any measurable
effect on Pol I transcription or in controlling the number of
active rRNA encoding genes (French et al., 2003), consistent
with the absence of H3K9 methylation in S. cerevisiae (Roguev
et al., 2001). In contrast, deacetylation of H3K9 in higher
eukaryotes facilitates H3K9 methylation, setting the stage for
the action of H3K9-specific histone methyltransferases such
as G9a, SET-DB1, and SUV39H1. Previous studies have shown
that the methyltransferase G9a is a coactivator for Pol I transcription
elongation through active rDNA repeats, whereas
SET-DB1 plays a role in NoRC-dependent heterochromatin
formation and rDNA silencing (Yuan et al., 2007). The results of
Murayama et al. revealed that SUV39H1 contributed to rDNA
silencing, most likely by increasing the number of repressed
rDNA genes, thereby restricting Pol I transcription to a smaller
number of genes. Notably, recent work by the Reinberg laboratory
not only demonstrated that SUV39H1 is acetylated,
but also that SIRT1-dependent deacetylation of lysine 266 in
the catalytic SET domain of SUV39H1 activates its enzymatic
activity to result in increased levels of H3K9me3 (Vaquero
et al., 2007). These findings demonstrate the involvement of
SIRT1-dependent deacetylation in mediating SUV39H1 histone
methyltransferase activity and underscore the functional
link between SUV39H1 and the deacetylase SIRT1. Consistent
with this, Murayama et al. showed that knockdown of either
SIRT1 or SUV39H1 resulted in decreased H3K9 methylation
and impaired NML association with rDNA. Together, these
observations indicate that SIRT1, SUV39H1, and NML, the
three components of eNoSC, cooperate to silence a fraction of
rDNA repeats. Thus, eNoSC joins a growing family of regulatory
protein complexes containing Sir2-related deacetylases.
Integrating Fasting Signals into the Epigenetic Circuitry
Given that caloric restriction decreases the cellular ATP concentration
and increases the NAD+/NADH ratio, the NAD+-
dependent deacetylase activity of Sir2 family members is
enhanced when the intracellular energy supply is limiting
(Guarente and Picard, 2005). Sir2 enzymes cooperate with
other proteins to establish larger domains of silenced chromatin.
In S. cerevisiae, for example, Sir2 binds to other Sir proteins
to form a SIR-repressive complex that is thought to spread in
cis over chromatin to enable silencing. Similarly, eNoSC in
mammals promotes deacetylation and methylation of H3K9 as
well as the spread of H3K9me, which may be aided by binding
of NML’s N-terminal domain to H3K9me2. This mechanism is
reminiscent of the binding of heterochromatin protein 1 (HP1)
to H3K9 methylated by SUV39H1 to facilitate spreading of
silenced chromatin domains. It is possible that NML may carry
out HP1-like functions in the nucleolus.
Further linking metabolism to chromatin structure, the NAD+
metabolite O-acetyl-ADP-ribose (AAR) induces a change in the
structure of the S. cerevisiae SIR complex and promotes the oligomerization
of SIR proteins in vitro (Liou et al., 2005). It is less
clear whether AAR has a similar corepressive role in human cells.
However, in vitro data and structural data do identify a tantalizing
AAR-binding function in the repressive heterochromatic histone
variant macroH2A1.1 (Kustatscher et al., 2005). Therefore, it is
possible that in human cells, Sir2-related enzymes, the cofactor
NAD+, and the metabolite AAR all play a role in chromatin-mediated
transcriptional silencing, including at rDNA repeats.
Further research on NML and eNoSC should address the connections
of this complex to the activities of SIRT7, another nucleolar
member of the sirtuin family. SIRT7 is associated with Pol I and
is a positive regulator of rDNA transcription (Ford et al., 2006). Like
SIRT1, SIRT7 binds to histones, is contained in high molecularweight
protein complexes, and is required for cell viability. Thus,
diet-induced changes in the NAD+/NADH ratio affect both SIRT1
and SIRT7 activity, coupling changing energy levels to rRNA synthesis
and ribosome production. Moreover, it will be important
to study the mechanisms governing the nucleocytoplasmic shuttling
of these enzymes in response to altering metabolic conditions,
as sirtuins and many other proteins show distinct cellular
localizations depending on the cellular energy status. In addition,
while it is clear that SAM binding is a prerequisite for the transcrip580
Cell 133, May 16, 2008 ©2008 Elsevier Inc.
tional corepressor function of NML in the nucleolus, it cannot be
excluded that instead of enzymatically using SAM as a methylgroup
donor, NML may instead act as a conformational sensor
for SAM (or the methyltransferase product S-adenosyl homocysteine).
Clearly, it would be most appealing if NML assumed an
enzymatic role for targeting RNA or protein substrates. A methyltransferase
function for NML would be particularly exciting and
innovative given that the budding yeast homolog Rrp8p seems to
link the synthesis and maturation of rRNA to epigenetic changes
in chromatin structure. Notably, all three subunits of eNoSC bind
to metabolite cofactors and use them to carry out repressive posttranslational
modifications. As cellular levels of ATP, NAD+, and
SAM may vary depending on the physiological state of a cell, it
stands to reason that eNoSC would be a perfect control point for
the regulation of rDNA transcription and silencing through metabolic
feedback loops. This is of particular interest in the context
of recent work on the mutual interdependency between distinct
metabolite-driven posttranslational modifications (Vaquero et al.,
2007). Although this is not a simple “I control you, you control me”
scenario, SIRT1 and SUV39H1 (and many other enzymatically
coupled systems) may profit from this type of interdependency.
Rather than following strict binary logic, the mutual dependence
on distinct metabolites may allow these critical epigenetic regulators
to gain more sophisticated and physiologically responsive
modes of regulation.
How Metabolite Players Balance the Gene Expression
Checkbook
There appears to be a resurgence of interest in the “old” field
of metabolism. Not only do modern mass spectrometry methods
allow us to gather more comprehensive quantitative data
regarding metabolism such that we can begin to model organismal
“metabolomics,” but also many “old” metabolic players are
re-emerging in new guises. This is particularly the case in the
field of gene expression (for review, see Ladurner, 2006). There
will likely be more surprises in store, possibly involving the many
metabolic enzymes that primarily function in the cytoplasm but
can shuttle into the nucleus under certain metabolic conditions.
What new functions might metabolic enzymes have co-opted
to control gene expression? One could borrow an important
concept from the study of metabolic pathways, that is, the well
known model of so-called “futile cycles” in which there is a
very rapid change in metabolic flux upon a small change in a
metabolic parameter. In other words, wasting a small amount of
energy (NAD+, SAM, or acetyl-CoA) in the right place to dynamically
regulate activating and repressive histone modifications
may be the optimal way to keep track of the overall level of cellular
energy expenditure. In the case of rDNA transcription, the
biogenesis of functional ribosomes unquestionably represents a
large commitment in energy expenditure for the cell. Therefore it
may not be so surprising that NML, SIRT1, and SUV39H1 cooperate
with other epigenetic modifiers, such as the methyltransferase
G9a, to tightly regulate the level of rRNA synthesis and
that all of these proteins, directly or indirectly, are under separate
but interconnected forms of metabolic control. Together,
these proteins interact with endogenous metabolites to modify
targets and to establish an activity profile for overall rDNA transcription
that ensures the synthesis of the optimal number of
functional ribosomes. In doing so, these proteins may help to
achieve a kind of homeostatic control, enabling the cell to avoid
precipitous losses in ATP and NADH and hence promoting cell
viability and the long-term survival of the organism.
If we may use a metaphor to highlight one key concept of
the Murayama et al. paper, we need to look no further than
the current woes of the financial markets. Any prudent investor
would skirt “irrational exuberance” (to quote Alan Greenspan,
the former chairman of the United States Federal Reserve) and
instead would invest a good deal of energy in making well-informed
decisions that balance the investment portfolio for the
short-term and the long-term. Similarly, the eNoSC complex,
NoRC, CSB, G9a, and other regulators of rDNA transcription,
remind us that like any other budget or investment manager, a
cell needs to be able to sense and clearly assess its metabolic
state. In this way, a cell can plan its commitment to significant
energy investments, such as the synthesis of ribosomes, and
can decide whether to throttle back on rDNA transcription, the
key regulatory step in ribosome biogenesis.
References
Bousquet-Antonelli, C., Vanrobays, E., Gelugne, J.P., Caizerguess-Ferrer, M.,
and Henry, Y. (2000). RNA 6, 826–843.
French, S.L., Osheim, Y.N., Cioci, F., Nomura, M., and Beyer, A.L. (2003). Mol.
Cell. Biol. 23, 1558–1568.
Ford, E., Voit, R., Liszt, G., Magin, C., Grummt, I., and Guarente, L.P. (2006).
Genes Dev. 20, 1075–1080.
Grummt, I. (2003). Genes Dev. 17, 1691–1702.
Grummt, I., and Grummt, F. (1976). Cell 7, 447–453.
Grummt, I., and Pikaard, C.S. (2003). Nat. Rev. Mol. Cell Biol. 4, 641–649.
Guarente, L., and Picard, F. (2005). Cell 120, 473–482.
Kobayashi, T., Horiuchi, T., Tongaonkar, P., Vu, L., and Nomura, M. (2004). Cell
117, 441–453.
Kustatscher, G., Hothorn, M., Pugieux, C., Scheffzek, K., and Ladurner, A.G.
(2005). Nat. Struct. Mol. Biol. 12, 624–625.
Ladurner, A.G. (2006). Mol. Cell 24, 1–11.
Liou, G., Tanny, J.C., Kruger, R.G., Walz, T., and Moazed, D. (2005). Cell 121,
515–527.
Murayama, A., Ohmori, K., Fujimura, A., Minami, H., Yasuzawa-Tanaka, K.,
Kuroda, T., Oie, H., Okuwaki, M., Nagata, K., Fukamizu, A., et al. (2008). Cell,
this issue.
Muth, V., Nadaud, S., Grummt, I., and Voit, R. (2001). EMBO J. 20, 1353–1362.
Peng, J.C., and Karpen, G.H. (2007). Nat. Cell Biol. 9, 25–35.
Roguev, A., Schaft, D., Shevchenko, A., Pijnappel, W.W., Wilm, M., Aasland, R.,
and Stewart, A.F. (2001). EMBO J. 20, 7137–7148.
Santoro, R., and Grummt, I. (2005). Mol. Cell Biol. 25, 2539–2546.
Santoro, R., Li, J., and Grummt, I. (2002). Nat. Genet. 32, 393–396.
Sinclair, D.A., and Guarente, L. (1997). Cell 91, 1033–1042.
Vaquero, A., Scher, M., Erdjument-Bromage, H., Tempst, P., Serrano, L., and
Reinberg, D. (2007). Nature 450, 440–444.
Yuan, X., Feng, W., Imhof, A., Grummt, I., and Zhou, Y. (2007). Mol. Cell 27,
585–595.