Anda di halaman 1dari 15


discussions, stats, and author profiles for this publication at:

Histone deacetylase inhibitors and the promise

of epigenetic (and more) treatments for cancer

Article in Nature reviews. Cancer February 2006

DOI: 10.1038/nrc1779 Source: PubMed


1,665 282

2 authors:

Saverio Minucci Pier Giuseppe Pelicci

IEO - Istituto Europeo di Oncologia Campus IFOM-IEO


Some of the authors of this publication are also working on these related projects:

Clinical proteomics for mapping of the epigenome of cancer samples, for patient stratification and
novel biomarker discovery View project View project

All content following this page was uploaded by Saverio Minucci on 28 September 2017.

The user has requested enhancement of the downloaded file.


Histone deacetylase inhibitors and

the promise of epigenetic (and more)
treatments for cancer
Saverio Minucci* and Pier Giuseppe Pelicci*
Abstract | Histone deacetylases (HDACs) are considered to be among the most promising
targets in drug development for cancer therapy, and first-generation histone deacetylase
inhibitors (HDACi) are currently being tested in phase I/II clinical trials. A wide-ranging
knowledge of the role of HDACs in tumorigenesis, and of the action of HDACi, has been
achieved. However, several basic aspects are not yet fully understood. Investigating these
aspects in the context of what we now understand about HDACi action both in vitro and
in vivo will further improve the design of optimized clinical protocols.
Histone deacetylases (HDACs) have been intensively An extensive phylogenetic analysis of HDACs has
scrutinized over the past few years for two main rea- been performed2,3. HDACs are members of an ancient
sons. First, they have been linked mechanistically to the enzyme family found in animals, plants, fungi and bac-
pathogenesis of cancer, as well as several other diseases. teria. It is thought that HDACs evolved in the absence
Second, small-molecule HDAC inhibitors (HDACi) exist of histone proteins. Indeed, eukaryotic HDACs can
that have the capacity to interfere with HDAC activity deacetylate non-histone as well as histone substrates, and
and can therefore achieve significant biological effects in some HDACs reside in the cytoplasm (where histones
preclinical models of cancer. These findings justified the are synthesized and acetylated for proper assembly, with-
introduction of HDACi into clinical trials. These initial out the intervention of HDACs) and in mitochondria
clinical trials have just ended and show encouraging (where histones are absent).
results. At this stage it is crucial to evaluate whether our Are HDACs, then, truly HDACs4? We postulate that
current knowledge on the mechanisms of tumorigenesis key HDAC substrates might not be histones, but instead
that are linked to HDACs, and on the mechanisms of belong to the growing list of acetylated non-histone pro-
tumour sensitivity to HDACi, is firm enough to improve teins (FIG. 1b and Supplementary information S1 (table)).
the design of further clinical trials. Recent structural and Additionally, non-protein molecules such as polyamines
chemical data have emerged that will help in the design or metabolic intermediates might also serve as valid sub-
of novel HDACi with more desirable properties than the strates (as found in bacteria).
existing ones. However, key areas of investigation that The search for other substrates, however, should not
might help to further illuminate the design of successful undervalue the fact that histones (as a bulk mass) are by
HDACi-based cancer therapy remain poorly explored. far the most abundant HDAC substrate, and that histone
Novel findings indicate that our understanding of how acetylation surely represents a key target for the action
*Department of Experimental
Oncology, European Institute
HDACi work will probably change significantly, estab- of most HDACs.
of Oncology, Via Ripamanti lishing a new paradigm in the field of intelligent drug
435, 20141, Milan, Italy. design with broad implications for the design of targeted The core of the code

Department of Biomolecular therapies in cancer and possibly other diseases. Histones and DNA constitute the nucleosomes, struc-
sciences and Biotechnology,
tural units of chromatin that are essential in packaging
University of Milan, Milan
20100, Italy. HDACs: enzymes looking for substrate(s) eukaryotic DNA. The tail region of the histones (which

Department of Medicine, Four HDAC classes have been identified (FIG. 1a). One extend away from the nucleosome core) undergo a
University of Milan, Milan of them (class 3 or the so-called sirtuins, from the yeast complex and coordinated series of regulatory modifica-
20100, Italy. protein Sir2) constitutes a structurally unrelated, NAD- tions57. These modifications can also occur within the
Correspondence to S.M.
e-mail: saverio.minucci@
dependent subfamily, and will not be considered here; globular domain of histones that make extensive contacts neither will the class 3-specific HDACi, which are less with DNA (see FIG. 1c for a list of the known acetylation
doi:10.1038/nrc1779 characterized than those for the other classes1. sites in histones).

38 | JANUARY 2006 | VOLUME 6

2006 Nature Publishing Group

At a glance essential for the correct execution of these events. Many

more years of work will be required to further advance
Histone deacetylases (HDACs) and histone acetylases (HATs) are enzymes that are our understanding of the role of histone acetylation in
responsible for deacetylating and acetylating, respectively, the amino-terminal tails
the regulation of nuclear events and in the context of the
of histones. These chromatin changes regulate transcription and many other nuclear
other chromatin modifications.
Non-histone proteins (such as the oncosuppressor p53) and several cytoplasmic
The rest of the acetylome: new roles for (H)DACs
proteins are also regulated by HATs/HDACs.
Our knowledge of the non-histone HDAC substrates
Studies on the molecular pathogenesis of acute myeloid leukaemias have shown that is rapidly increasing, although it remains fragmented.
the aberrant recruitment of HDACs has an important role in leukaemogenesis.
As for histones, we should view acetylation as part of
Leukaemia-associated fusion proteins (such as promyelocytic leukaemia (PML) a complex set of post-translational modifications often
retinoic acid receptor (RAR) and acute myeloid leukaemia 1 (AML1)ETO) recruit
working cooperatively to regulate the function of the
HDACs to repress the transcription of genes involved in differentiation (the fusion
proteins therefore block differentiation) and impair the function of p53.
modified protein(s).
Identifying the entire acetylome deserves a much
Alterations in the expression and/or activity of HATs/HDACs have been also observed
more aggressive and systematic experimental strategy
in solid tumours. Solid tumours show decreased levels of histone acetylation, which
correlates with clinical outcome.
than has been used so far. A potential approach would
be the use of pan-specific antibodies that are able to
HDAC inhibitors (HDACi) have been widely studied and belong to several
recognize acetylated lysines in any protein context for
chemical classes.
proteomic approaches, similar to those used to study
HDACi exert cell-type-specific effects inducing apoptosis, cell-cycle arrest, and
other post-translational modifications14. These antibod-
ies are available and should produce a dynamic map of
In leukaemias, HDACi induce the expression of members of the tumour-necrosis the acetylome and its function15,16.
factor-related apoptosis-inducing ligand (TRAIL) and FAS death receptor pathways.
A growing list of acetylated proteins is currently
This induction is responsible for the pro-apoptotic effects of HDACi.
available (see Supplementary information S1 (table)).
Clinical trials for several HDACi have started, and HDACi-responsive tumours have
As expected from the cellular localization of HATs and
been observed.
HDACs, both cytoplasmic and nuclear proteins can
undergo reversible acetylation. Acetylation of a protein
can have many different effects.
It might be oversimplistic to try to assign specific
regulatory functions to defined post-translational Effects on protein stability. Both acetylation and ubi-
modifications. The widely used, and in some cases mis- quitylation often occur on the same amino-acid residue
used, term histone code emphasizes the fact that some (lysine), and there is a direct cross-talk between these
chromatin functions are specified by multiple and co- two modifications17. Competition between acetylation
ordinated histone modifications, and that the same post- and ubiquitylation influences the stabililty of the sub-
translational modification does not necessarily function strate and, therefore, indirectly regulates its function. So,
in a similar manner in other contexts. However, this HDACs can decrease the half-life of several substrates by
rule has many exceptions, to the extent that it seems exposing the lysine residue for ubiquitylation1821 (FIG. 1b).
more appropriate to consider the various post-synthesis In some cases, the cross-talk is more complex, with both
modifications of histones as an epigenomic alphabet, direct and indirect effects (see below).
in which each modification is a letter, and combined
modifications at a defined genomic region constitute a Effects on proteinprotein interactions. In the case of
word, which might have different functional meanings the interferon- enhanceosome, different HATs acetylate
in different contexts. Recently, a histone alphabet has different lysine residues of the structural protein high
been proposed to unify the nomenclature of histone mobility group protein isoforms I and Y (HMGI/Y,
modifications and is based on similar principles8. also known as HMGA1) in a precise order. This con-
Acetylation (particularly of histone H3 and his- fers either potentiation of transcriptional activity or
tone H4 tails) has almost invariably been linked to a destabilization of the enhanceosome and termination
chromatin state that is poised for transcription or that of the transcriptional response22. In the case of the
corresponds to actively transcribed genomic regions9. transcription factor signal transducer and activa-
Therefore, histone acetylases (HATs) and HDACs (act- tor of transcription 3 (STAT3), which is activated by
ing in the context of multisubunit complexes; BOX 1) cytokine signalling, cytosolic acetylation triggers STAT3
have traditionally been linked to activation and repres- dimerization and subsequent nuclear translocation23,24.
sion mechanisms, respectively. Additionally, acetylation Acetylation of hypoxia-inducible factor 1 (HIF1) by the
of core histones has been correlated with other genome ARD1 HAT apparently leads to increased association
functions, including chromatin assembly, DNA repair with the von HippelLindau (VHL) ubiquitylation
and recombination10,11. A crucial role has also been complex and proteasome-mediated degradation, which
Enhanceosome demonstrated for histone acetylation in imposing repli- has a regulatory role in the cellular response to changes
The assembly of higher-order cation timing of specific genomic regions12. In turn, this in oxygen availability and angiogenesis25,26. Another
three-dimensional transcription
factorenhancer DNA
might lead to large-scale changes in gene expression13. process that is regulated by reversibile acetylation is
complexes that are required to In all of these cases, a complex spatio-temporal inter- the cytosolic association of the mainly nuclear DNA-
activate a gene. play between HATs and HDACs takes place, which is damage-response protein Ku70 (also known as XRCC6)


2006 Nature Publishing Group

A B Inammation,
Class 1 necrosis (monocytes/ j
HDAC1 macrophages)
HDAC2 (86%) Ac Ac
HDAC3 (63%) Apoptosis Importins Nucleus
HDAC8 (43%) c
Stress TF
Class 2 Ac
i Mitochondrion d
HDAC7 (34%) TF:SMAD7, TF
BAX SREBP, RUNX, Ac (p53 unstable, repression)
HDAC6 (49%) ETS Stress (HATs)
HDAC4,5,9 (34%) HDAC Ku
Ac TF p53
Class 4 Ku BAX Ac Ac
HATs b TF (p53 stable, high levels
HDAC11 h Ub
Proteasome of DNA-binding, activation)
Misfolded Aggresome e
C protein
Acetylation TF Nucleosome
Methylation K12 Ub Ac
Ub Degradation
K9 K20 Ub HDAC
K13 K36 K119 K24
K15 K85 Nuclear TF Enhanced
H2A H2B K108 translocation DNA binding
K56 Microtubules
K116 HDAC6i Activation
K27 H3 H4 K120 g
K23 K20 S/Y phosphorylation
K115 K122 K79 K77 K8 Ac Ac Cargo
K18 K16 K12 HAT Cargo
K14 HDACs Cargo
K9 Cytokines STAT3 HSP9O Degradation

Figure 1 | The acetylome. A | A schematic representation of histone deacetylases (HDACs). Class 1, 2 and 4 human
HDACs are represented, with the catalytic domains of the different classes in different colours. The degree of
homology (entire sequences) between HDACs belonging to the same class are in brackets. All HDACs share a
zinc-dependent catalytic domain with a high degree of homology. Less conserved are accessory domains, which seem
to fulfill regulatory functions. In mammals, class 1 HDACs (HDAC13 and HDAC8) are related to the yeast RPD3 HDAC;
class 2 HDACs (HDAC47, HDAC9 and HDAC10) are related to the yeast HDA1 HDAC, and have a more complex
domain organization; a single HDAC, HDAC11, is the unique member of class 4 HDACs. Except for class 4, HDACs are
found in all fully sequenced, free-living eukaryotic organisms. Class 4 proteins are found in all of these organisms
except fungi. The conservation of these proteins indicates that each HDAC class has a non-redundant role in basic cell
biological processes. Class 1 HDACs are widely expressed, whereas classes 2 and 4 show various degrees of tissue
specificity. In addition, class 1 and 4 HDACs are constitutively nuclear proteins, whereas class 2 HDACs shuttle between
the nucleus and the cytoplasm owing to the reversible interaction with 14-3-3 proteins137. B | A schematic view of the
functioning acetylome. A partial list of biological processes that are regulated mechanistically by acetylation is
sketched in aj. Acetylation (Ac) might regulate the association of transcription factors (TF for example, SMAD7,
SRBEP, RUNX and ETS) with the DNA (a). Protein stability is also influenced by histone acetyltransferases (HATs) and
HDACs, as lysines are subject to both acetylation and ubiquitylation (b). The import and export of proteins into the
nucleus is regulated by the acetylation of the importin proteins at the nuclear envelope (c). Multiple, coordinated
functions of p53 are known to be regulated by acetylation (d) and accessibility of the nucleosome is another HAT- and
HDAC-regulated process (see below) (e). Chaperone functions of heat-shock protein 90 (HSP90) have also been shown
to be effected by acetylation. Acetylation of HSP90 in the absence of HDAC6 prevents HSP90 interacting with its
target protein (f). The acetylation of signal transducer and activator of transcription 3 (STAT3) induces protein
14-3-3 proteins
dimerization and translocation to the nucleus (g). The function of the aggresome is regulated by acetylation. HDAC6 is
These proteins are a family of required to properly recruit ubiquitylated proteins to aggresomes, and inhibition of its activity leads to accumulation of
conserved regulatory polyubiquitylated proteins (h). The translocation of BAX to the mitochondria is influenced by acetylation of the DNA-
molecules that are expressed damage associated protein Ku70. When Ku70 is acetylated, BAX is free to localize to the mitochondria (i). Acetylation
in all eukaryotic cells. 14-3-3 also favours nuclear export and cytosolic accumulation of HMG box 1 (HMGB1) before secretion (j). C | The acetylation
proteins have the ability to map of histones. A schematic view of the histone octamer, with the histone tails showing the possible acetylation sites
bind a multitude of functionally (yellow circles) and methylation sites (blue circles) when this occurs on the same residue. Histone tails undergo a
diverse signalling proteins, complex, coordinated series of post-translational modifications (phoshorylation, acetylation, methylation,
including kinases,
ubiquitylation, ADP-ribosylation, deimination, and probably other post-translational modifications) that are required to
phosphatases and
transmembrane receptors.
regulate chromatin function(s)6,7. Two important mechanisms govern the significance of histone modifications. They
Recently, class 2 histone can generate docking sites for interaction with additional proteins; bromo-, chromo- and SANT-domains have been
deacetylases have also been shown to recognize specifically modified histones, and to mediate the recruitment of accessory regulatory factors that
shown to interact with contain those domains. Also, modifications of histones might affect the condensation state and higher order chromatin
members of the 14-3-3 family. structure, owing to their capacity to modulate the exposure of charge patches on the surface of the nucleosomes57.

40 | JANUARY 2006 | VOLUME 6

2006 Nature Publishing Group

Box 1 | Histone deacetylase complexes Other effects. In some cases, the function of HAT/
HDAC substrates is influenced by acetylation at several
A number of histone deacetylase (HDAC)-associated factors and HDAC- levels. The oncosuppressor p53 is regulated by acetyl-
containing complexes have been described for class 1 HDACs. HDAC1 and
ation, which regulates the stability of the protein, its
HDAC2 are mainly found associated in the same molecular complexes with Mi2/
interaction with DNA and its transcriptional activity35.
NURD (nucleosome remodelling and histone deacetylase), SIN3A and Co-REST,
whereas HDAC3 is found in a different complex (the nuclear co-receptor (NCOR)/ Similarly, the DNA-binding affinity of the transcription
silencing mediator for retinoid and thyroid hormone receptors (SMRT) factor nuclear factor B (NF-B) (in its multiple forms)
complex)126133. HDAC6 has been found in cytoplasmic complexes with proteins is regulated by HATs/HDACs, as is its transcriptional
involved in the ubiquitin signalling pathway134. Large multi-enzymatic complexes activation and its association with its regulator, inhibi-
that contain histone acetyltransferase (HAT) or HDAC have also been described, tor of B- (IB)36. Cytoplasmic HDAC6 is crucial
which might represent the biochemical explanation of the dynamic nature of for the formation of aggresomes (organelles required
histone acetylation135. Accessory factors might crucially function by tethering for the efficient clearance of misfolded, toxic, cytoplas-
the complex to defined genomic regions (looking at chromatin as the main mic proteins), although in this case the role of protein
substrate), and to integrate modulatory signals through interactions with other
acetylation remains unclear37. Recently, HDAC6 has
proteins. Apparently, HDACs of different classes can co-exist in the same
been shown to deacetylate the cytoplasmic chaperone
macromolecular complex. It has been suggested that the HDAC3 complex
maintains the enzymatic activity of class 2 HDACs136. These studies have been protein heat-shock protein 90 (HSP90). HSP90 hyper-
performed by biochemical purification of native complexes or by acetylation in HDAC6-deficient cells leads to the loss of
immunopurification of tagged HDACs (or HDAC-associated factors). These chaperone activity38.
studies were performed in cell lines and cannot be easily reproduced in normal Given the multitude of effects that HDACs have,
cells, so it remains to be fully understood whether there are differences in the deciphering the phenotypes of HDAC-deficient model
composition of the complexes according to the cell state. The different results organisms is not always easy. Among class 1 HDACs,
reported by various groups might reflect the heterogeneity of the HDAC- only HDAC1 has been knocked out in mammals so
containing complexes in different cells at different stages of differentiation. far. The phenotype is embryonic lethal at very early
stages of development, mainly owing to an arrest of
cell growth that is associated with upregulation of the
with the pro-apoptotic protein BAX. In its deacetylated cyclin-dependent-kinase (CDK) inhibitors p21 (also
form (maintained by several HDACs), Ku70 keeps BAX known as WAF1) and p27 (also known as KIP1). This
away from the mitochondrion and protects cells from shows a crucial, non-redundant role for HDAC1 in
apoptosis. When the cell is stressed, HATs acetylate regulating cell proliferation39. It is unknown whether
Ku70, leading to dissociation of the complex and trans- the other class 1 HDACs behave similarly. Consistent
location of BAX to the mitochondria, where it activates with their more restricted pattern of expression, class 2
the apoptotic programme27,28. HDAC-knockout mice have demonstrated a crucial role
for distinct class 2 HDACs in modulating the growth
Effects on protein localization. As already described for response of specific tissues such as cartilage and heart
some factors (STAT3 and BAX), reversibile acetylation muscle. In these cases, association of class 2 HDACs
affects the subcellular localization of several proteins. (HDAC5 and HDAC9 in cardiomiocytes and HDAC4 in
In some cases, the nuclear localization signal contains chondrocytes) with specific transcription factors (MEF2
acetylatable lysine residues that favour nuclear retention in heart and RUNX2 in cartilage) leads to repression
when acetylated23. In the case of the multifunctional of genes that are involved in cell growth. Accordingly,
HMG box 1 (HMGB1) protein, acetylation favours the corresponding knockouts show in contrast to
nuclear export and cytosolic accumulation before secre- what is observed in Hdac1-null mice hypertrophic
tion during inflammatory and/or necrotic processes29. growth40,41. The analysis of HDAC-knockout mice has
Proteins involved in nuclear import can themselves be just started, and we do not precisely understand the
regulated by acetylation30. cellular responses to the knockout of specific HDACs
and how they are dependent on changes in acetylation
Effects on DNA binding. Several transcription factors of histones or of other HDAC substrates. In the case of
show increased DNA binding and subsequent tran- Hdac1/ cells, a general increase in histone acetylation
scriptional activity when acetylated, which correlates has been observed, pointing to HDAC1 as a crucial
with the hyper-acetylation of histones in target chro- component in the control of acetylation. The genera-
matin (see references in Supplementary information tion of conditional knockout strains, or the systematic
S1 (table)). Perhaps more surprisingly, in some cases use of RNA interference (RNAi) to knockdown HDAC
acetylation impairs the binding of transcriptional expression in specific tissues after embryonic develop-
activators to the DNA, indicating that HATs and ment, would also be helpful. Unfortunately, the available
HDACs might work in an orchestrated way to achieve HDAC-knockout models have so far not been useful in
the same cellular effect31. Interestingly, general tran- addressing tumour susceptibility.
scription factor 2B (GTF2B, also known as TFIIB) has
been reported to behave as an auto-acetyltransferase, HDACs in cancer: a matter of balance
and acetylation regulates its activity 32. Acetylation We can expect both positive and negative effects of
also impairs the catalytic and DNA-binding activi- HDACs on oncogenic and oncosuppressive mechanisms.
ties of enzymes involved in DNA metabolism and A balance must exist, and shifts in this balance might
repair33,34. have dramatic consequences on the cell phenotype.


2006 Nature Publishing Group

The most informative evidence on how this balance is molecular events lead to the powerful differentiative
altered in cancer cells comes from studies on the patho- burst of promyelocytes that is seen in treated patients.
genesis of leukaemias42. PLZFRAR does not respond to RA, even at pharma-
Acute promyelocytic leukaemia (APL) was the first cological doses, explaining the resistance of patients
model disease in which the involvement of HDACs to RA. However, the combination of RA with HDACi
was demonstrated42. This form of leukaemia is char- is extremely effective in reactivating RA target genes
acterized by an arrest of the leukaemic cells at the and triggers a cellular response in vitro and in vivo in
promyelocytic stage of myeloid differentiation, and PLZFRAR-positive APL patients50.
is caused (as shown in murine models of the disease) The multiple-hit model hypothesizes that the fusion
by fusion proteins (found in 100% of patients with protein has leukaemogenic potential, but needs to co-
APL) of retinoic acid receptor- (RAR) with one of operate with one or more additional genetic mutations to
the following proteins: (in > 95% of cases) promyelo- trigger the development of leukaemia51,52. The search for
cytic leukaemia (PML), ( 5% of cases) promyelocytic these second hits is on, and obvious candidates have been
leukaemia zinc finger (PLZF) or (sporadically) other carefully scrutinized. Among these, TP53 mutations are
genes. Clinically, patients with APL (with the excep- exceedingly rare in patients with APL53, but p53 function
tion of patients expressing PLZFRAR) respond to is compromised through other mechanisms. PMLRAR
treatment with pharmacological doses of the RAR associates with p53 and along with class 1 HDACs causes
ligand, retinoic acid (RA). At the cellular level, the its deacetylation and subsequent degradation through
clinical remission is due to the re-initiation of the dif- the MDM2proteasome pathway in haematopoietic
ferentiation programme of the leukaemic cells, which precursors. These results provide the first link between
go on to full neutrophilic differentiation and then die, alterations in p53 acetylation and tumorigenesis, high-
physiologically, by apoptosis43. lighting the relevance of the study of post-translational
RAR functions as a transcription factor. In the modifications of non-histone proteins in cancer54,55.
absence of RA, RAR is found at DNA response elements Fusion proteins involved in other forms of leukae-
of RA-regulated genes and is associated with HDAC- mia share the capacity to abnormally recruit HDAC-
containing complexes, and this contributes to the tran- containing complexes42. Of course, beyond the apparent
scriptional silencing of these genes. At physiological similarities, differences do exist. In B-cell lymphomas
concentrations of RA, a conformational switch leads to for example, the oncogene BCL6 encodes a transcrip-
the release of the HDAC-containing complexes and to tional repressor that requires HDAC recruitment for its
the association of RAR with transcriptional co-activators oncogenic properties. Interestingly, BCL6 is negatively
(including HATs) and subsequent transcriptional activa- regulated through direct acetylation by p300 HAT, and
tion. RA target genes are involved in key cellular proc- acetylation disrupts its ability to recruit HDACs and to
esses and RAR works in cooperation with several other transform cells56.
transcription factors (for example, cytokine-regulated
factors such as Stats)44 to initate transcription. HDACs and HATs in solid tumours
In APL, all fusion proteins maintain the capacity to In solid tumours, there is patchy but significant evi-
bind to RA-regulated genes, but share an aberrant fea- dence for a disruption in the balance of acetylation and
ture physiological concentrations of RA are unable deacetylation (FIG. 2b). Specific HATs (p300 and CREB
to trigger the switch, and HDACs remain associated (cAMP response element-binding protein)-binding pro-
to RA targets45. To make things worse, RAR fusions tein (CBP)) are targets of viral oncoproteins (adenoviral
oligomerize through self-association domains (present E1A, human papilloma virus E6 and the simian virus 40
in PML, PLZF and the other genes), which leads to (SV40) large T antigen)57. Monoallelic mutations of both
an increased stoichiometric association of HDAC- p300 and CBP are found in patients who are affected by
containing complexes with RA target genes and the congenital RubinsteinTaybi syndrome patients
increased transcriptional silencing46,47. Additionally, with this syndrome show developmental defects and are
recruitment by the fusion protein of other proteins prone to cancer58. Mutations and inactivations of specific
with chromatin modifying activities (DNA methyl- HATs have been observed in a small number of patients
transferases and histone methyl-transferases; FIG. 2a) with cancer. As for leukaemias, comparable mutations
packs the chromatin to an extent that does not allow have so far not been observed for HDACs, hinting at
other physiological stimuli acting on the same genes the selection of a hyperactive HDAC phenotype in
(that is, cytokines that normally trigger differen- cancer cells58. Intriguingly, hyper-expression of HDAC-
tiation) to function properly48,49. Although not fully associated factors occurs relatively often; genes of the
validated experimentally, this model explains most of metastasis-associated protein (Mta) family (in particular,
the observed data. Most importantly, this model fits MTA1), are associated with the metastatic phenotype in
perfectly with the phenomenon of RA sensitivity (and several tumour types59. The mechanistic consequences of
RA resistance in the case of PLZFRAR) of patients overexpressing MTA1 are not well understood. MTA1
with APL. In fact, at pharmacological doses of RA associates with the oestrogen receptor and inhibits its
(10 to 100-fold higher than physiological doses) the function, and in transgenic mice it induces alterations
switch is turned on, HDACs dissociate from RAR and in mammary gland development that lead to tumour
the turnover of the fusion proteins (through protea- formation60. MTA1 and other genes are thought to have
somal degradation) is significantly increased. These a crucial function in integrating the enzymatic activity

42 | JANUARY 2006 | VOLUME 6

2006 Nature Publishing Group

of HDACs with cellular signalling pathways. Therefore, the adenomatosis polyposis coli (APC) tumour suppres-
their deregulated activity might also lead to a change in sor. RNAi-mediated knockdown of HDAC2 in colonic
the activity of the HDACs. However, this has not been cancer cells resulted in cell death, indicating a role for
experimentally observed so far. HDAC2 has been shown HDAC2 in protecting cancer cells against apoptosis61.
to interact functionally with the Wnt pathway, as it is Recently, attempts have been made to correlate
overexpressed in tumours and tissues from mice that lack altered levels and/or activity of HATs or HDACs with
distinct epigenetic states of chromatin. Overexpression
a of MTA1 in oesophageal squamous cell carcinoma
correlates with poor prognosis and lower levels
PML HDAC PML HDAC of acetylated histone H4 in cancer cells 62. A more
Acetyl groups SUV39H1 comprehensive analysis of a panel of normal tissues,
RAR RAR cancer cell lines and primary tumours has shown a
RARE RARE widespread loss of monoacetylated (at K16) and tri-
Histone deacetylation * * *
Histone methylation
methylated (at K20) forms of histone H4 in a large
number of tumours63. This loss occurs in repetitive
DNA sequences and parallels the global loss of DNA
methylation in these areas, which is also associated
DNMT DNMT with tumorigenesis63. Strikingly, there are clinical cor-
relations between distinct histone modification pat-
SUV39H1 terns (modifications examined: histone H3 acetylation
RAR CH3 CH3 RAR CH3 at K9 and K18, dimethylation of K4; and histone H4
RARE RARE acetylation at K12, dimethylation at R3) and risk of
DNA methylation *
tumour recurrence in patients with low-grade pros-
tate cancer64. Low levels of histone acetylation, which
Heterochromatin-like formation by themselves are not sufficient to draw statistically
significant predictions, are correlated with a poorer
clinical outcome. Needless to say, a compelling ques-
Differentiation block tion is whether these changes are restricted to histones
b or are shared by other HDAC substrates.
Besides histone acetylation, alterations in other
X Other chromatin changes? epigenetic pathways (not discussed here) clearly
have an important role in tumorigenesis. Indirectly,
Histone deacetylation the observation that alterations of other chromatin
modifiers can be observed in tumours is an addi-
tional link to deregulation in the balance of histone
acetylation both DNA and histone methylation are
Effects on transcription (silencing) tightly linked to repressive chromatin states and at
Effects on DNA replication Transformed least partially dependent on the concomitant action
Effects on DNA repair phenotype
of HDACs6567. To date, however, the most convincing
Figure 2 | A model for the deregulated action of HDACs on chromatin in APL and evidence that HDACs behave differently in cancer cells
in other cancer cells. a | In acute promyelocytic leukaemia (APL), retinoic acid receptor than in normal cells derives from the pharmacological
(RAR) fusion proteins (represented by promyelocytic leukaemia (PML)RAR in the scheme) manipulation of HDACs through HDACi.
recruit histone deacetylases (HDACs) (through co-regulatory (CoR) proteins such as
nuclear co-receptor (NCOR)/silencing mediator for retinoid and thyroid hormone Inhibiting HDACs
receptors (SMRT)) to the promoters of genes containing retinoic acid-responsive elements A dogmatic view of target identification for drug devel-
(RARE), leading to the deacetylation of histones. PMLRAR will then sequentially recruit
opment would indicate that HDACs are not suitable
other chromatin modifiers (DNA methyltransferases DNMT1 and DNMT3a, and the
histone H3 K9-methyltransferases SUV39H1 and SUV39H2) that will methylate DNA (CH3) targets inhibiting HDACs has too many chances of
and histones (asterisks). The modified chromatin functions as a docking site for accessory interfering with key cellular functions. Luckily, the use
factors that are able to bind methylated DNA (methyl-CpG-binding domain proteins such of HDAC inhibitors pre-dated the discovery of the HDACs
as MBD1) and methylated histones (such as the heterochromatin associated HP1 protein), themselves. In fact, the inhibitors were instrumental in the
thereby leading to the formation of a compact heterochromatin-like structure. Genes purification and cloning of mammalian HDACs.
that are downregulated by PMLRAR are involved in myeloid differentiation, and because A relatively wide range of structures have been identi-
of the action of the fusion protein they become refractory to physiological differentiating fied that are able to inhibit the activity of class 1, class 2
stimuli (for example, cytokines and hormones), resulting in the block of differentiation and and class 4 HDACs68,69. They derive from both natural
prolonged proliferation. b | In non-APL cells, other factors (for example, in leukaemias: sources (FIG. 3a) and from synthetic routes (FIG. 3b). With
other fusion proteins such as acute myeloid leukaemia 1(AML1)ETO; in lymphomas:
a few exceptions, they can all be divided into chemical
BCL6; and in solid tumours: deregulated transcription factors) substitute for PMLRAR in
targeting HDACs to specific genomic regions, possibly leading to the same or similar set classes including hydroxamic acid derivatives, carboxy-
of chromatin alterations observed in APL. The newly formed chromatin structure (directly, lates, benzamides, electrophilic ketones and cyclic pep-
or through recruitment of accessory factors) determines an altered pattern of tides70,71. So far, these inhibitors work equally well against
transcription (for coding areas), or of DNA replication or repair, which cooperate in the class 1, class 2 and class 4 HDACs. Only a few molecules
transformed phenotype. are emerging as preferential inhibitors of class 1 versus


2006 Nature Publishing Group

a HDAC inhibitors from natural sources b Synthetic HDAC inhibitors

NH X = CH, Suberoyl anilide hydroxamic acid (SAHA) AN-9 (Pivanex)
X = N, Pyroxamide
O O Apicidin
Depsipeptide (FK-228) H 3C
( )n H O
H PXD-101

O O R1 H
Trichostatin A (TSA) Sodium butyrate Cyclic hydroxamic acid-containing peptides (CHAPs) H
R = Benzyl, methyl, sec-butyl
O = Benzyl, 4-methyl-benzyl Sulfonamide hydroxamic acid
c Pharmacophore model for HDAC inhibition
( )n O O O
Rim O
R NH HN MS-275
R O R1
S O R1
PCU ( )n ZnC Tubacin
R = Ph; R1 = H, Sodium phenyl butrate
HS R = CH3; R1 = nC3H7, Sodium valporate


Figure 3 | HDAC inhibitors. Representative structures of the main classes of known histone deacteylase (HDAC)
inhibitors (HDACi), either from natural sources (a) or synthetic (b). In HDACs, the active site of the enzyme (containing
the zinc atom) occupies the bottom of a channel delimited by a rim, which corresponds to the substrate pocket
(acetyl-lysine). The schematic view of the HDACi pharmacophore (c) therefore consists of: a metal-binding domain
(ZnC), which chelates zinc and blocks the enzymatic activity; a linker domain, which mimics the substrate and
occupies the enzymatic channel (PCU and HS: polar connecting unit and hydrophobic spacer, respectively); and a
surface domain, which makes contacts with the rim71. The structures of the HDACi chemical groups that carry out
these processes are shown below each of the yellow boxes, as indicated by the arrows.

class 2 HDACs, and even fewer (that is, tubacin for review, we will consider unless specifically mentioned
HDAC6) are able to discriminate efficiently among in the text HDACi as a single entity as they display
HDACs that belong to the same class7275. At present, comparable biological activities. However, they might
this limitation has little relevance to the use of HDACi differ substantially in their pharmacological properties
as potential anti-tumour drugs as there is no definitive (potency, efficacy, stability and toxicity), and will there-
evidence that distinct HDACs have a defined role in fore probably be subject to different degrees of drug
cancer. On the other hand, HDAC-selective inhibitors development in the future.
would constitute (in addition to genetic tools) a useful
experimental tool to address the issue of assigning dis- The cellular effects of HDACi
tinct biological functions to individual HDACs. HDACi induce, to a variable extent, growth arrest,
From the structure of HDAC8 or HDLP (a HDAC- differentiation or apoptosis in vitro and in vivo68,69. In
like protein isolated from Aquifex aeolicus), and the some cases, growth arrest is induced at low doses, and
Pharmacophore structure of these enzymes in complex with a small apoptosis is induced at higher doses; in other cases,
The minimum amount of
chemical groups or
number of HDACi, a few key points have emerged about growth arrest precedes apoptosis. However, cells
functionality in a drug required the active site that support a pharmacophore model for might undergo apoptosis without significant changes
to elicit its action. HDAC inhibition7679 (FIG. 3c). For the purpose of this in their cell-cycle profile. Strikingly, normal cells

44 | JANUARY 2006 | VOLUME 6

2006 Nature Publishing Group

are almost always considerably more resistant than explain why the transcriptional response to HDACi is
tumour cells to HDACi68. so complex). It would be somewhat difficult to ascribe
It is worth stressing that there might be HDACi- a prominent role to hyper-acetylation of chromatin
dependent pathways that work against the desired or transcription factors in the observed response, as
anti-proliferation, pro-apoptotic and pro-differentiation these modifications occur concomitantly. A partial
effects of HDACi. One such example is the E2F family dissection of the molecular mechanisms, however, is
of transcription factors that regulate target genes that sometimes possibile. As an example, p21 (see below)
are crucial for S-phase progression. The oncosuppressor is consistently upregulated by HDACi90. One of the key
retinoblastoma (RB) recruits HDACs to E2F target transcriptional regulators of p21 is p53. Strikingly, both
genes, thereby controlling G1/S cell-cycle progression. hyper-acetylation of chromatin at the CDKN1A (which
HDACi should therefore lead to activation of RB tar- encodes p21) promoter and transcriptional activation
gets, effectively enhancing a key proliferative pathway. of CDKN1A occur at the same level in wild-type and
Despite these reservations, HDAC recruitment seems p53-null cells, implying that hyper-acetylation of p53
crucial only for a subset of RB targets, and cell-cycle is not important for the therapeutic efficacy of HDACi
regulation by RB proceeds largely through HDAC- (although acetylation of other transcription factors
independent mechanisms80. cannot be excluded)91,92. HDACi-mediated acetyla-
Given the known function of histone acetylation in tion of other substrates might lead to transcriptional
transcription, it seems logical to postulate that inhibi- effects only indirectly, or work through distinct, non-
tion of HDACs alone is unlikely to lead to a general- transcriptional mechanisms. In the end, HDACi might
ized increase in the transcription of all known genes. be considered as partially epigenetic drugs.
In fact, acetylation works together with other post- A complex pattern of HDACi targets must therefore
translational modifications, and blocking deacetyla- exist that is cell-type-specific, cell-stage-specific, and
tion might have very different outcomes depending dependent on the normal or pathological state of the
on the previous chromatin state. Up to 20% of all cell. However, some general HDACi-mediated responses
known genes are affected by HDACi8184. Not all of are known, and their relevance to the observed cellular
these genes are necessarily upregulated by treatment effects can be studied. HDACi-induced growth arrest
the ratio of upregulated to downregulated genes is is tightly linked to the induction of p21, and in some
close to 1:1. cases this arrest has been shown to be irreversible and
Looking at chromatin as the target for HDACi action, to resemble the phenomenon of replicative senes-
we should not disregard the possibility that HDACi influ- cence90,91,93. On the other hand, HDACi also induce
ence other nuclear phenomena besides transcription: cell death through caspase-dependent and caspase-
what we have previously discussed indicates effects on independent pathways. In most of the studies, caspase
DNA replication and repair, and both processes can have activation has been reported to occur through the
a key role in modulating the cellular response to HDACi. mitochondria/cytochrome c-mediated apoptotic path-
In fact, HDACi can inhibit DNA repair responses in a way. Nevertheless, HDACi can induce cell death with
few cell lines, which might increase the sensitivity of morphological features of autophagy in the absence of
tumour cells to chemotherapy and radiotherapy by lead- activated caspases, indicating that HDACi might work
ing to increased DNA damage by these treatments85,86. in cells with apoptotic defects94.
At least in part, the failure to activate two cell-cycle In HDACi-induced apoptosis, the mitochondrion
checkpoints that are present in normal cells is respon- and the production of pro-apoptotic reactive oxygen
sible for the tumour-selective action of HDACi87. The species (ROS) seems to be important9496. The precise
first is the G2-phase checkpoint, which when defective mechanisms underlying the accumulation of ROS are
permits cells to enter an aberrant mitosis. The second is under investigation and might help us to understand
the mitotic-spindle checkpoint, which normally detects why normal cells are selectively resistant to HDACi
aberrant mitosis and blocks mitotic exit until the defect treatment. Normal, but not transformed, human
is rectified. The disruption of both checkpoints results fibroblasts accumulate thioredoxin (TXN, also known
in the premature exit of tumour cells from an abortive as TRX), a natural ROS scavenger, after treatment with
mitosis and the subsequent induction of apoptosis88,89. HDACi. Therefore, ROS accumulation might be neu-
One model of action for HDACi derived from these tralized in normal cells by TXN, whereas transformed
studies, and those on the effects of HDACi on DNA cells die because they do not express TXN: indeed,
repair, is that HDACi should lead to an increased accu- knockdown of TXN in normal cells causes accumula-
mulation of DNA damage (by endogenous stress or by tion of ROS and increased cell death96. Interestingly,
drug treatment) in sensitive cells. As the DNA in these Bax/ murine embryonic fibroblasts are also resistant
cells will be unrepaired and these defects will remain to HDACi97. As discussed earlier, hyper-acetylation of
unchecked owing to the altered cell-cycle checkpoints, cytoplasmic Ku70 causes the activation of BAX and
this will lead to catastrophic cell death. apoptosis97. HDACi also cause hyper-acetylation of
Given the pleiotropic effects of HDACs on non- HSP90 and its inactivation, leading to the degrada-
histone substrates, we must expect additional tar- tion of proteins that require the chaperone function of
gets. These targets might be transcription factors HSP90 (including some oncoproteins98,99). Moreover,
themselves, and therefore the ultimate endpoint will selective inhibition of HDAC6 by tubacin leads to
eventually be transcriptional (which might help to cell death through accumulation of ubiquitylated


2006 Nature Publishing Group


subsequent recruitment of CBP. RNAi-based studies
performed in vitro and in vivo show that these genes
are uniquely required for the induction of apoptosis
in leukaemic cells, both from APL mice and freshly
Pre-leukaemic state isolated leukaemic blasts from patients who do not
Normal cell Tumour cell
(by activating oncogenes) have APL. Intriguingly, not all blasts respond to
HDACi by undergoing apoptosis, and these cases do
not show induction of TRAIL or FAS pathways54,102.
The reasons for this lack of response are unknown.
Preliminary evidence indicates that HDACi might
TRAIL, FAS activate components of the death receptor pathways in
various solid tumours. In fact, in several cases HDACi
synergize with exogenously added TRAIL to induce
apoptosis of tumour cell lines103106. However, the effect
on the death receptor pathway is not universal, and
No effect No effect Apoptosis therefore its activation does not necessarily represent
an obligate path for HDACi-mediated cell death.
Figure 4 | Tumour-selective action of HDACi in acute
In conclusion, many areas of investigation on the bio-
promyelocytic leukaemia. Histone deacteylase (HDAC)
inhibitors (HDACi) have little or no effect on short-term logical effects of HDACi show an exciting and complex
cultures of normal haematopoietic progenitors in vitro or network of cellular responses, coupling transcriptional
on haematopoiesis in normal mice. Expression of the fusion and non-transcriptional effects. It is difficult at this stage
oncoprotein promyelocytic leukaemiaretinoic acid to predict the existence of a universal, key cellular target
receptor (PMLRAR) in haematopoietic progenitors leads that is responsible for the HDACi-mediated response.
to deregulation of differentiation and inhibition of p53- To try to address this question, it would be helpful to
dependent stress responses (pre-leukaemia), but does not analyse systematically, and in different model systems,
result in full transformation, which depends on further all of the proposed effectors of HDACi to define their
genomic hits. HDACi are not effective on cells at the pre- relative contributions to the observed biological effects.
leukaemic stage, which behave similarly to normal cells.
In the absence of such studies, we can only speculate
Following transformation, HDACi are able to induce
members of the tumour-necrosis factor-related apoptosis- whether the different mechanisms that have been pro-
inducing ligand (TRAIL) or FAS pathways, leading to posed represent cell-type-specific responses to HDACi,
tumour-specific cell death. or whether the simultaneous activation of several path-
ways (or all of them required for maximal effect) have a
synergistic function.
proteins100. A limitation of these observations is their An encouraging observation is that in several cases
lack of systematic analysis, the use of cell lines as normal cells show a strikingly reduced sensitivity to
model systems which do not necessarily reflect the HDACi treatment, hinting at potentially large differ-
biology of primary tumour cells and the focus on a ences in the acetylome in normal versus tumour cells
limited number of HDACi. that can be exploited clinically.
As a well-defined model system, APL offers an interest-
ing paradigm for studying HDACi action (FIG. 4). HDACi HDACi in clinical development
trigger apoptosis of APL cells both in vitro and in murine The first studies on the clinical use of HDACi have been
models of the disease101,102. An obvious explanation for published recently107109. These studies are phase I/II
this effect would be the reversal of the HDAC-depend- trials, and their endpoint is therefore somewhat limited
ent action of the fusion protein, as discussed earlier in (the main aim being to analyse/exclude unwanted toxi-
this review. Strikingly, however, HDACi function in city and to find optimal doses/schedules), although one
APL through a fusion-protein-independent and p53- of the objectives of phase II studies is to measure the
independent mechanism (see below). But, in appropri- clinical response. It is not our scope to review in detail
ate experimental conditions, such as on the induction single studies on specific HDACi, but rather to draw
of cellular stress, HDACi can revert the leukaemic cell initial and very preliminary conclusions, and to suggest
phenotype through an effect on the RAR fusion proteins55. how such trials need to evolve (TABLE 1).
However, this effect is not necessary for the anticancer These reports generally confirm the initial impres-
effects of HDACi. sion from preclinical studies of low toxicity of HDACi
in patients, if compared with most of the currently used
On the right TRAIL anti-tumour treatments. The toxic effects of HDACi,
HDACi induce hyper-acetylation of the promoters which vary based on the class of HDACi used, might
of several members of the death receptor pathway at least in part be the result of the non-specific side
including TRAIL (tumour-necrosis factor-related effects of each individual HDACi rather than the con-
apoptosis-inducing ligand) and its receptor, death sequences of inhibiting HDAC per se. In fact, some of
receptor 5 (DR5), and FAS ligand (FASL) and FAS, these compounds are only efficacious at high concen-
through a mechanism that seems to involve acetyla- trations in which interference with additional targets
tion of the transcription factors SP1 or SP3, and would probably be responsible for toxicity. As for most

46 | JANUARY 2006 | VOLUME 6

2006 Nature Publishing Group

Table 1 | Summary of the main clinical studies using histone deacetylase inhibitors
Histone Disease Phase Clinical results Side effects References
AN-9 (butyric acid Lung cancer, melanoma IIa/IIb Monotherapy in lung cancer (non- Monotherapy was well 139
prodrug) and leukaemias small-cell carcinomas); patients showed tolerated, whereas a
partial responses in <10% of the cases combination treatment with
docetaxel resulted in severe
toxicity and interruption of the
Sodium Leukaemias and I Haematological improvements with Well tolerated, with 138
phenylbutyrate myelodysplasia no remissions; poor potency and neurological toxicity only at very
pharmacokinetic properties high doses
Valproic acid Leukaemias, I/II In combination with retinoic acid, Neurological toxicity 119,140
myelodysplasia and partial and complete remissions have
cervical cancer been observed in 30% of cases in a pilot
study, whereas an extended phase II
study showed only haematological
FK-228 (cyclic T-cell cutaneous or I/II Monotherapy in T-cell lymphomas Bone marrow toxicity, reversible 114
depsipeptide) peripheral lymphoma, showed partial and complete responses cardiac arrhythmia (which 116,141
leukaemias and solid in up to 57% of patients in a single seems not to be clinically
tumours phase II study; one partial response was relevant), fatigue, nausea and
observed in solid tumour patients vomiting
MS-275 Refractory solid I Long half-life in patients (3980 Toxicity indicated that a once 113
(benzamide) tumours, leukaemias hours); no clinical responses have been every 14 days oral schedule
and lymphomas reported would be best; nausea, vomiting,
anorexia and fatigure were
the dose-limiting toxicities;
cumulative bone marrow
toxicity was also seen
Suberoylanilide Refractory solid II Tested in both intravenous and oral Mild toxicity, dehydration, 111,112
hydroxamic acid tumours, leukaemias formulations; partial and complete fatigue, diarrhoea and
(hydroxamic acid and lymphomas responses have been observed in anorexia; thrombocytopaenia
derivative) haematological and solid tumors and anaemia were the dose-
limiting toxicities; no signs of
LAQ-824 Refractory solid I Partial and complete responses Bone marrow toxicity, fatigue, NA
(hydroxamic acid tumours, leukaemias observed in leukaemias; a close diarrhoea, nausea, and
derivative) and lymphomas analogue (LBH589) has also entered tachycardia (which is only
clinical studies observed following treatment
with LBH589); signs of potential
cardiac toxicity have been
recently reported, which led to
the interruption of the clinical
PXD101 Advanced solid tumours I Unknown clinical effects Apparently well tolerated, with a NA
lack of haematological toxicities
The list is not fully comprehensive due to space limitations, but we have made an effort to include histone deacetylase inhibitors of different structures and to
mention studies that showed clear clinical effects or serious toxic effects. NA, not available.

chemotherapic drugs, bone marrow toxicity (of low directly involved in the pathogenesis. However, in light of
to moderate degreee) is often observed. Strikingly, in the effects of HDACi in murine models of leukaemia, we
some cases the maximal tolerated dose or MTD has not anticipate that HDACi might be useful in other contexts
been reached in clinical trials, implying that specific as they might attack tumour-cell-specific targets that are
HDACi have minimal toxicity and a wide therapeutic not directly involved in tumorigenesis. We also do not
window (TABLE 1). know the key target(s) for HDACi action. Proposed sur-
Unfortunately, the relationship between the toxicity of rogate markers, such as measuring the levels of acetylated
HDACi and their pharmacodynamic/pharmacokinetic histones from peripheral blood cells before and after
properties is still largely unknown. This makes it diffi- treatment16, should serve as indicators of effectiveness,
Maximal tolerated dose cult to optimize HDACi treatment. In particular, we do but these need to be validated clinically and do not
The maximal dosage of a
particular drug that can be
not know which patients are most likely to respond to always seem to correlate strictly with pharmacokinetic
used before patients show HDACi. Naively, we would predict that HDACi might profiles110,111. Alternative strategies could be attempted to
undesired side effects. be useful only in those tumours in which HDACs are identify novel predictive markers; gene profiling studies


2006 Nature Publishing Group

from tumour samples obtained from responders and non- drugs is potentiated by HDACi, but the mechanistic
responders might lead to their identification. Also, we do basis for this synergy has not been fully explored.
not know whether there is a requirement for continuous Given the results previously discussed, one way to
inhibition of HDACs to achieve the maximal effect. increase sensitivity to chemotherapy and/or radio-
Studies in preclinical models (such as leukaemias) therapy would be the suppression of checkpoint and
have shown that an intraperitoneal injection of HDACi repair mechanisms triggered by induced DNA-damage.
induces pulses of histone hyper-acetylation in tumour Unfortunately, synergism in efficacy might be accom-
cells (lasting a few hours), followed by a return to basal panied by adverse effects that are rarely or never seen
levels. This response is owing to the fast degradation with HDACi alone, and caution should be employed
of the HDACi used in those studies, which resemble when considering clinical studies that address HDACi
the kinetics and fast degradation of many HDACi in combination with other therapies. Clinical studies
(particularly those based on hydroxamic acid) in the combining RA treatment with HDACi are in progress
human bloodstream. In mice, this treatment regimen in patients with leukaemia, expecially in the elderly119.
leads to rapid and massive apoptosis of tumour cells, Of particular interest is the combination of HDACi
with little toxicity observed. Are we happy with this with other drugs acting on epigenetic mechanisms.
observation and to use it as a benchmark for a similar DNA-demethylating agents in combination with
response in patients? Although a clinical response HDACi show potent responses in vitro and they are
is observed in the mouse models (and in patients also being tested in clinical trials120. The combination
treated with HDACi see below), animals are not of HDACi with DNA-demethylating agents should be
cured, and as soon as the treatment is stopped, the more carefully analysed in murine models of tumours,
disease kills them in 100% of the cases. It seems plau- with a view to optimizing the design of clinical stud-
sible that the continuous inhibition of HDAC activity ies. HDACi might also work synergistically with HSP90
will have a more positive long-term response, but will inhibitors; this would further inhibit the capacity of
it also be more toxic for normal cells? Two different HSP90 to chaperone oncoproteins that are required
formulations of the same drug (oral and intravenous for the survival of tumour cells121.
suberoylanilide hydroxamic acid (SAHA)) apparently
confirm this trend. Oral SAHA has a more favourable Perspectives
pharmacokinetic profile (half-life of 120 minutes ver- The prevailing conclusion is one of excitement for our
sus 40 minutes for intravenous SAHA) and results in a progress in understanding the function of HDACs
better clinical performance, but it is substantially more in tumour pathogenesis and the tumour response to
toxic111,112. HDACi with substantially longer half-lives HDACi, and hope for the exploitation of this knowledge
(such as MS-275, with a half-life of up to 80 hours) to develop more effective clinical protocols.
show a higher toxicity that precludes daily treatment113. There are areas that we have not covered that might
Valproic acid is being taken at high doses by epileptic become relevant in the future. For instance, the regula-
patients with no side effects, but this HDACi has an tion of HDAC enzymatic activity through post-transla-
extremely low potency and most of the absorbed drug tional modifications (phosphorylation or sumoylation)
(up to 90%) is bound to serum proteins, making it could turn out to be very important in modulating
unlikely to significantly inhibit HDACs at the doses HDAC function. This could potentially be exploited
used in these patients. Notwithstanding these caveats, pharmacologically122. Another issue to be considered
promising results have been observed in terms of clini- is that, so far, there is no conclusive experimental evi-
cal response in the current trials. None of the HDACi dence that points to specific HDACs as being selectively
tested in clinical trials is completely without clinical involved in any form of disease, including cancer. A
effect, confirming that HDACs represent a promis- systematic analysis to try to identify selective HDACi
ing target for further development. The results have that will enable the study of specific HDACs is manda-
been especially good in patients with T-cell cutaneous tory if we are to develop potentially less toxic and more
lymphoma, and in some cases a long-term response effective treatments. There are other areas that we have
has been observed114,115. The molecular basis for this discussed above that also require urgent answers.
response is being explored in vitro116. Tumour regres- A more general consideration also needs to be
sion (in some cases quite dramatic) has also been addressed. It has becoming increasingly clear that tumour
observed in solid tumours111,112. cells are not homogeneous and that cancer stem cells
Acquired resistance to HDACi has been reported exist123,124. It is reasonable to assume that the failures of
anecdotally, implying that it might be a clinically rare current cancer therapy are in part due to the failure to
event117. Depsipeptide has been reported to be a substrate target cancer stem cells. HDACi affect the maintenance of
for P-glycoprotein (Pgp, also known as ATP-binding cas- stem-cell potential in CD34+ human haematopoietic pro-
sette subfamily B, member 1 (ABCB1)), and multidrug genitors in vitro125. So far, no study using HDACi (either
resistance-associated protein 1 (MRP1, also known as preclinical or clinical) has addressed the crucial issue of
ABCC1), and to induce Pgp expression, indicating that tumour stem-cell sensitivity, which should be clarified as
resistance to HDACi might be drug-specific116,118. soon as possible to verify the true potential of HDACi in
Systematic in vitro studies have explored the possi- clinical treatment.
bility that HDACi might synergize with other drugs109. Perhaps, an epigenetic (or partially epigenetic)
The effect of a wide spectrum of chemotherapeutic therapy for cancer (the epi-cure) is getting closer.

48 | JANUARY 2006 | VOLUME 6

2006 Nature Publishing Group

1. Blander, G. & Guarente, L. The Sir2 family of protein 25. Jeong, J. W. et al. Regulation and destabilization of 50. He, L. Z. et al. Histone deacetylase inhibitors induce
deacetylases. Annu. Rev. Biochem. 73, 417435 HIF-1 by ARD1-mediated acetylation. Cell 111, remission in transgenic models of therapy-resistant
(2004). 7097020 (2002). acute promyelocytic leukemia. J. Clin. Invest. 108,
2. Gregoretti, I. V., Lee, Y. M. & Goodson, H. V. Molecular 26. Bilton, R. et al. Arrest-defective-1 protein, an 13211330 (2001).
evolution of the histone deacetylase family: functional acetyltransferase, does not alter stability of hypoxia- 51. Brown, D. et al. A PMLRAR transgene initiates
implications of phylogenetic analysis. J. Mol. Biol. inducible factor (HIF)-1 and is not induced by murine acute promyelocytic leukemia. Proc. Natl
338, 1731 (2004). hypoxia or HIF. J. Biol. Chem. 280, 3113231140 Acad. Sci. USA 94, 25512556 (1997).
An insightful phylogenetic analysis of HDACs. (2005). 52. Grisolano, J. L., Wesselschmidt, R. L., Pelicci, P. G. &
3. Leipe, D. D. & Landsman, D. Histone deacetylases, 27. Cohen HY, et al. Acetylation of the C terminus of Ku70 Ley, T. J. Altered myeloid development and acute
acetoin utilization proteins and acetylpolyamine by CBP and PCAF controls Bax-mediated apoptosis. leukemia in transgenic mice expressing PMLRAR
amidohydrolases are members of an ancient protein Mol. Cell 13, 627638 (2004). under control of cathepsin G regulatory sequences.
superfamily. Nucleic Acids Res. 25, 36933997 Cytoplasmic role for acetylation of Ku70 in Blood 89, 376387 (1997).
(1997). apoptosis. 53. Trecca, D. et al. Analysis of p53 gene mutations in
4. Kouzarides, T. Acetylation: a regulatory modification to 28. Cohen, H. Y. et al. Calorie restriction promotes acute myeloid leukemia. Am. J. Hematol. 46,
rival phosphorylation? EMBO J. 19, 11761179 mammalian cell survival by inducing the SIRT1 304309 (1994).
(2000). deacetylase. Science 305, 390392 (2004). 54. Insinga, A., Pelicci, P. G. & Minucci, S. Leukemia-
5. Richmond, T. J. & Davey, C. A. The structure of DNA 29. Bonaldi, T. et al. Monocytic cells hyperacetylate associated fusion proteins. Multiple mechanisms of
in the nucleosome core. Nature 423, 145150 chromatin protein HMGB1 to redirect it towards action to drive cell transformation. Cell Cycle 4,
(2003). secretion. EMBO J. 22, 55515560 (2003). 6769 (2005).
6. Strahl, B. D. & Allis, C. D. The language of covalent 30. Bannister, A. J., Miska, E. A., Grlich, D. & 55. Insinga, A. et al. Impairment of p53 acetylation,
histone modifications. Nature 403, 4145 (2000). Kouzarides, T. Acetylation of importin- nuclear import stability and function by an oncogenic transcription
7. Turner, B. M. Cellular memory and the histone code. factors by CBP/p300. Curr. Biol. 10, 467470 factor. EMBO J. 23, 11441154 (2004).
Cell 111, 285291 (2002). (2000). Demonstrates that deacetylation of non-histone
References 6 and 7 contain the proposal and one 31. Caillaud, A. et al. Acetylation of interferon regulatory substrates (p53) by HDACs has a role in
of the counter-proposals of the histone code. factor-7 by p300/CREB-binding protein (CBP)- tumorigenesis.
8. Turner, B. M. Reading signals on the nucleosome with associated factor (PCAF) impairs its DNA binding. 56. Bereshchenko, O. R., Gu, W. & Dalla-Favera, R.
a new nomenclature for modified histones. Nature J. Biol. Chem. 277, 4941749421 (2002). Acetylation inactivates the transcriptional repressor
Struct. Mol. Biol. 12, 110112 (2005). 32. Choi, C. H., Hiromura, M. & Usheva, A. Transcription BCL6. Nature Genet. 32, 606613 (2002).
9. Allfrey, V. G., Faulkner, R. & Mirsky, A. E. Acetylation factor IIB acetylates itself to regulate transcription. 57. Eckner, R., Arany, Z., Ewen, M., Sellers, W. &
and methylation of histones and their possible role in Nature 424, 965969 (2003). Livingston, D. M. The adenovirus E1A-associated
the regulation of RNA synthesis. Proc. Natl Acad. Sci. 33. Hasan, S. et al. Acetylation regulates the DNA end- 300-kD protein exhibits properties of a transcriptional
USA 51, 786794 (1964). trimming activity of DNA polymerase . Mol. Cell 10, coactivator and belongs to an evolutionarily conserved
10. Polo, S. E. & Almouzni, G. Histone metabolic pathways 12131222 (2002). family. Cold Spring Harb. Symp. Quant. Biol. 59,
and chromatin assembly factors as proliferation 34. Hasan, S. et al. Regulation of human flap 8595 (1994).
markers. Cancer Lett. 220, 19 (2005). endonuclease-1 activity by acetylation through the 58. Iyer, N. G., Ozdag, H. & Caldas, C. p300/CBP and
11. Vidanes, G. M., Bonilla, C. Y. & Toczyski, D. P. transcriptional coactivator p300. Mol. Cell 7, cancer. Oncogene 23, 42254231 (2004).
Complicated tails: histone modifications and the DNA 12211231 (2001). 59. Kumar, R., Wang, R. A. & Bagheri-Yarmand, R.
damage response. Cell 121, 973976 (2005). 35. Bode, A. M. & Dong, Z. Post-translational modification Emerging roles of MTA family members in human
12. Weinreich, M., Palacios DeBeer, M. A. & Fox, C. A. of p53 in tumorigenesis. Nature Rev. Cancer 4, cancers. Semin. Oncol. 30, 3037 (2003).
The activities of eukaryotic replication origins in 793805 (2004). 60. Bagheri-Yarmand, R., Talukder, A. H., Wang, R. A.,
chromatin. Biochim. Biophys. Acta 1677, 142157 36. Chen, L. F. & Greene, W. C. Shaping the nuclear action Vadlamudi, R. K. & Kumar, R. Metastasis-associated
(2004). of NF-B. Nature Rev. Mol. Cell Biol. 5, 392401 protein 1 deregulation causes inappropriate
13. Zhang, J., Xu, F., Hashimshony, T., Keshet, I. & Cedar, H. (2004). mammary gland development and tumorigenesis.
Establishment of transcriptional competence in early 37. Kawaguchi, Y. et al. The deacetylase HDAC6 regulates Development 131, 34693479 (2004).
and late S phase. Nature 420, 198202 (2002). aggresome formation and cell viability in response to 61. Zhu, P. et al. Induction of HDAC2 expression upon loss
14. Mann, M. & Jensen, O. N. Proteomic analysis of post- misfolded protein stress. Cell 115, 727738 (2003). of APC in colorectal tumorigenesis. Cancer Cell 5,
translational modifications. Nature Biotechnol. 21, 38. Kovacs, J. J. et al. HDAC6 regulates Hsp90 acetylation 455463 (2004).
255261 (2003). and chaperone-dependent activation of glucocorticoid 62. Toh, Y. et al. Expression of the metastasis-associated
15. Qiang, L., Xiao, H., Campos, E. I., Ho, V. C. & Li, G. receptor. Mol. Cell 18, 601607 (2005). MTA1 protein and its relationship to deacetylation of
Development of a PAN-specific, affinity-purified anti- Role of acetylation in the control of the chaperone the histone H4 in esophageal squamous cell
acetylated lysine antibody for detection, identification, function of HSP90. carcinomas. Int. J. Cancer 110, 362367 (2004).
isolation, and intracellular localization of acetylated 39. Lagger, G. et al. Essential function of histone 63. Fraga, M. F. et al. Loss of acetylation at Lys16 and
protein. J. Immunoassay Immunochem. 26, 1323 deacetylase 1 in proliferation control and CDK trimethylation at Lys20 of histone H4 is a common
(2005). inhibitor repression. EMBO J. 21, 26722681 hallmark of human cancer. Nature Genet. 37,
16. Ronzoni, S., Faretta, M., Ballarini, M., Pelicci, P. & (2002). 391400 (2005).
Minucci, S. New method to detect histone acetylation Genetic demonstration that HDAC1 is essential for References 62 and 63 demonstrate that global
levels by flow cytometry. Cytometry A 66A, 5261 proliferation in mammals. histone modifications occur in cancer.
(2005). 40. Zhang, C. L. et al. Class II histone deacetylases act as 64. Seligson, D. B. et al. Global histone modification
17. Caron, C., Boyault, C. & Khochbin, S. Regulatory signal-responsive repressors of cardiac hypertrophy. patterns predict risk of prostate cancer recurrence.
cross-talk between lysine acetylation and Cell 110, 479488 (2002). Nature 435, 12621266 (2005).
ubiquitination: role in the control of protein stability. 41. Vega, R. B. et al. Histone deacetylase 4 controls A coordinated pattern of histone modifications
Bioessays 27, 408415 (2005). chondrocyte hypertrophy during skeletogenesis. Cell predicts clinical outcome in prostate cancer.
18. Giandomenico, V., Simonsson, M., Grnroos, E. & 119, 555566 (2004). 65. Egger, G., Liang, G., Aparicio, A. & Jones, P. A.
Ericsson, J. Coactivator-dependent acetylation 42. Minucci, S., Nervi, C., Lo Coco, F. & Pelicci, P. G. Epigenetics in human disease and prospects for
stabilizes members of the SREBP family of Histone deacetylases: a common molecular target for epigenetic therapy. Nature 429, 457463 (2004).
transcription factors. Mol. Cell. Biol. 23, 25872599 differentiation treatment of acute myeloid leukemias? 66. Feinberg, A. P. & Tycko, B. The history of cancer
(2003). Oncogene 20, 31103115 (2001). epigenetics. Nature Rev. Cancer 4, 143153 (2004).
19. Grnroos, E., Hellman, U., Heldin, C. H. & Ericsson, J. 43. Warrell, R. P. et al. Differentiation therapy of acute 67. Lund, A. H. & van Lohuizen, M. Epigenetics and
Control of Smad7 stability by competition between promyelocytic leukemia with tretinoin (all-trans- cancer. Genes Dev. 18, 23152335 (2004).
acetylation and ubiquitination. Mol. Cell 10, retinoic acid). N. Engl. J. Med. 324, 13851393 68. Johnstone, R. W. Histone-deacetylase inhibitors: novel
483493 (2002). (1991). drugs for the treatment of cancer. Nature Rev. Drug
20. Jin, Y. H. et al. Transforming growth factor- stimulates 44. Minucci, S. & Pelicci, P. G. Retinoid receptors in health Discov. 1, 287299 (2002).
p300-dependent RUNX3 acetylation, which inhibits and disease: co-regulators and the chromatin 69. Marks, P. et al. Histone deacetylases and cancer: causes
ubiquitination-mediated degradation. J. Biol. Chem. connection. Semin. Cell Dev. Biol. 10, 215225 and therapies. Nature Rev. Cancer 1, 194202 (2001).
279, 2940929417 (2004). (1999). 70. Mai, A. et al. Histone deacetylation in epigenetics: an
21. Rausa, F. M., Hughes, D. E. & Costa, R. H. Stability of 45. Lin, R. J., Egan, D. A. & Evans, R. M. Molecular attractive target for anticancer therapy. Med. Res.
the hepatocyte nuclear factor 6 transcription factor genetics of acute promyelocytic leukemia. Trends Rev. 25, 261309 (2005).
requires acetylation by the CREB-binding protein Genet. 15, 179184 (1999). 71. Miller, T. A., Witter, D. J. & Belvedere, S. Histone
coactivator. J. Biol. Chem. 279, 4307043076 46. Lin, R. J. & Evans, R. M. Acquisition of oncogenic deacetylase inhibitors. J. Med. Chem. 46,
(2004). potential by RAR chimeras in acute promyelocytic 50975116 (2003).
22. Munshi, N et al. Coordination of a transcriptional leukemia through formation of homodimers. Mol. Cell 72. Park, J. H. et al. Class I histone deacetylase-selective
switch by HMGI(Y) acetylation. Science 293, 5, 821830 (2000). novel synthetic inhibitors potently inhibit human
11331136 (2001). 47. Minucci, S. et al. Oligomerization of RAR and AML1 tumor proliferation. Clin. Cancer Res. 10, 52715281
23. Yuan, Z. L., Guan, Y. J., Chatterjee, D. & Chin, Y. E. transcription factors as a novel mechanism of (2004).
Stat3 dimerization regulated by reversible acetylation oncogenic activation. Mol. Cell 5, 811820 (2000). 73. Mai, A. et al. Discovery of (aryloxopropenyl)pyrrolyl
of a single lysine residue. Science 307, 269273 48. Di Croce, L. et al. Altered epigenetic signals in human hydroxyamides as selective inhibitors of class IIa
(2005). disease. Cancer Biol. Ther. 3, 831837 (2004). histone deacetylase homologue HD1-A. J. Med. Chem.
24. Wang, R., Cherukuri, P. & Luo, J. Activation of Stat3 49. Di Croce, L. et al. Methyltransferase recruitment and 46, 48264829 (2003).
sequence-specific DNA binding and transcription by DNA hypermethylation of target promoters by an 74. Heltweg, B. et al. Subtype selective substrates for
p300/CREB-binding protein-mediated acetylation. oncogenic transcription factor. Science 295, histone deacetylases. J. Med. Chem. 47, 52355243
J. Biol. Chem. 280, 1152811534 (2005). 10791082 (2002). (2004).


2006 Nature Publishing Group

75. Haggarty, S. J., Koeller, K. M., Wong, J. C., Grozinger, through a process regulated by generation of reactive markers, therapeutic targets, and mechanisms of
C. M. & Schreiber, S. L. Domain-selective small- oxygen species and induction of p21CIP1/WAF1 1. resistance. Blood 103, 46364643 (2004).
molecule inhibitor of histone deacetylase 6 (HDAC6)- Cancer Res. 63, 36373645 (2003). 117. Bandyopadhyay, D., Mishra, A. & Medrano, E. E.
mediated tubulin deacetylation. Proc. Natl Acad. Sci. 96. Ungerstedt, J. S. et al. Role of thioredoxin in the Overexpression of histone deacetylase 1 confers
USA 100, 43894394 (2003). response of normal and transformed cells to histone resistance to sodium butyrate-mediated apoptosis in
76. Finnin, M. S. et al. Structures of a histone deacetylase deacetylase inhibitors. Proc. Natl Acad. Sci. USA 102, melanoma cells through a p53-mediated pathway.
homologue bound to the TSA and SAHA inhibitors. 673678 (2005). Cancer Res. 64, 77067710 (2004).
Nature 401, 188193 (1999). 97. Subramanian, C., Opipari, A. W., Bian, X., Castle, V. P. 118. Xiao, J. J. et al. Chemoresistance to depsipeptide
77. Richon, V. M. et al. Histone deacetylase inhibitors: & Kwok, R. P. Ku70 acetylation mediates FK228 [(E)-(1S,4S,10S,21R)-7-[(Z)-ethylidene]-4,21-
assays to assess effectiveness in vitro and in vivo. neuroblastoma cell death induced by histone diisopropyl-2-oxa-12,13-dithia-5,8,20,23-
Methods Enzymol. 376, 199205 (2004). deacetylase inhibitors. Proc. Natl Acad. Sci. USA 102, tetraazabicyclo[8,7,6]-tricos-16-ene-3,6,9,22-
78. Somoza, J. R. et al. Structural snapshots of human 48424847 (2005). pentanone] is mediated by reversible MDR1 induction
HDAC8 provide insights into the class I histone 98. Atadja, P. et al. Molecular and cellular basis for the in human cancer cell lines. J. Pharmacol. Exp. Ther.
deacetylases. Structure (Camb) 12, 13251334 anti-proliferative effects of the HDAC inhibitor 314, 467475 (2005).
(2004). LAQ824. Novartis Found. Symp. 259, 249266 119. Pilatrino, C. et al. Increase in platelet count in older,
79. Vannini, A. et al. Crystal structure of a eukaryotic zinc- (2004). poor-risk patients with acute myeloid leukemia or
dependent histone deacetylase, human HDAC8, 99. Bali, P. et al. Inhibition of histone deacetylase 6 myelodysplastic syndrome treated with valproic acid
complexed with a hydroxamic acid inhibitor. Proc. Natl acetylates and disrupts the chaperone function of heat and all-trans retinoic acid. Cancer 104, 101109
Acad. Sci. USA 101, 1506415069 (2004). shock protein 90: a novel basis for antileukemia (2005).
References 78 and 79 show the structural analysis activity of histone deacetylase inhibitors. J. Biol. 120. Cameron, E. E., Bachman, K. E., Myhnen, S.,
of mammalian HDAC8 in complex with HDACi. Chem. 280, 2672926734 (2005). Herman, J. G. & Baylin, S. B. Synergy of
80. Siddiqui, H., Solomon, D. A., Gunawardena, R. W., 100. Hideshima, T. et al. Small-molecule inhibition of demethylation and histone deacetylase inhibition in
Wang, Y. & Knudsen, E. S. Histone deacetylation of proteasome and aggresome function induces the re-expression of genes silenced in cancer. Nature
RB-responsive promoters: requisite for specific gene synergistic antitumor activity in multiple myeloma. Genet. 21, 103107 (1999).
repression but dispensable for cell cycle inhibition. Proc. Natl Acad. Sci. USA 102, 85678572 (2005). 121. Rahmani, M. et al. Cotreatment with
Mol. Cell. Biol. 23, 77197731 (2003). 101. Insinga, A. et al. Inhibitors of histone deacetylases suberanoylanilide hydroxamic acid and 17-allylamino
81. Glaser, K. B. et al. Gene expression profiling of induce tumor-selective apoptosis through activation of 17-demethoxygeldanamycin synergistically induces
multiple histone deacetylase (HDAC) inhibitors: the death receptor pathway. Nature Med. 11, 7176 apoptosis in BcrAbl+ cells sensitive and resistant to
defining a common gene set produced by HDAC (2005). STI571 (imatinib mesylate) in association with down-
inhibition in T24 and MDA carcinoma cell lines. Mol. 102. Nebbioso, A. et al. Tumor-selective action of HDAC regulation of BcrAbl, abrogation of signal transducer
Cancer Ther. 2, 151163 (2003). inhibitors involves TRAIL induction in acute myeloid and activator of transcription 5 activity, and Bax
82. Mitsiades, C. S. et al. Transcriptional signature of leukemia cells. Nature Med. 11, 7784 (2005). conformational change. Mol. Pharmacol. 67,
histone deacetylase inhibition in multiple myeloma: References 101 and 102 demonstrate the essential 11661176 (2005).
biological and clinical implications. Proc. Natl Acad. role of induction of the TRAIL pathway for the pro- 122. Girdwood, D. W., Tatham, M. H. & Hay, R. T. SUMO
Sci. USA 101, 540545 (2004). apoptotic response following HDACi treatment in and transcriptional regulation. Semin. Cell Dev. Biol.
83. Peart, M. J. et al. Identification and functional leukaemias. 15, 201210 (2004).
significance of genes regulated by structurally 103. Inoue, H. et al. Histone deacetylase inhibitors 123. Al-Hajj, M., Becker, M. W., Wicha, M., Weissman, I. &
different histone deacetylase inhibitors. Proc. Natl sensitize human colonic adenocarcinoma cell lines to Clarke, M. F. Therapeutic implications of cancer stem
Acad. Sci. USA 102, 36973702 (2005). TNF-related apoptosis inducing ligand-mediated cells. Curr. Opin. Genet. Dev. 14, 4347 (2004).
84. Van Lint, C., Emiliani, S. & Verdin, E. The expression of apoptosis. Int. J. Mol. Med. 9, 521525 (2002). 124. Reya, T., Morrison, S. J., Clarke, M. F. & Weissman, I. L.
a small fraction of cellular genes is changed in 104. Nakata, S. et al. Histone deacetylase inhibitors Stem cells, cancer, and cancer stem cells. Nature 414,
response to histone hyperacetylation. Gene Expr. 5, upregulate death receptor 5/TRAIL-R2 and sensitize 105111 (2001).
245253 (1996). apoptosis induced by TRAIL/APO2-L in human 125. Milhem, M. et al. Modification of hematopoietic stem
References 8184 detail gene-profiling studies of the malignant tumor cells. Oncogene 23, 62616271 cell fate by 5aza 2deoxycytidine and trichostatin A.
cell transcriptional response to HDACi in different (2004). Blood 103, 41024110 (2004).
model systems. The results are not always fully 105. Singh, T. R., Shankar, S. & Srivastava, R. K. HDAC An initial study on the effects of HDACi on stem cells.
consistent among the different systems (the number inhibitors enhance the apoptosis-inducing potential of 126. Guenther, M. G. et al. A core SMRT corepressor
of modulated genes varies from <5% to 20%). TRAIL in breast carcinoma. Oncogene 24, complex containing HDAC3 and TBL1, a WD40-repeat
85. Camphausen, K. et al. Enhanced radiation-induced cell 46094623 (2005). protein linked to deafness. Genes Dev. 14,
killing and prolongation of H2AX foci expression by 106. Watanabe, K., Okamoto, K. & Yonehara, S. 10481057 (2000).
the histone deacetylase inhibitor MS-275. Cancer Sensitization of osteosarcoma cells to death receptor- 127. Huang, E. Y. et al. Nuclear receptor corepressors
Res. 64, 316321 (2004). mediated apoptosis by HDAC inhibitors through partner with class II histone deacetylases in a Sin3-
86. Munshi, A. et al. Histone deacetylase inhibitors downregulation of cellular FLIP. Cell Death Differ. 12, independent repression pathway. Genes Dev. 14,
radiosensitize human melanoma cells by suppressing 1018 (2005). 4554 (2000).
DNA repair activity. Clin. Cancer Res. 11, 49124922 107. Rosato, R. R. & Grant, S. Histone deacetylase 128. Humphrey, G. W. et al. Stable histone deacetylase
(2005). inhibitors in clinical development. Expert Opin. complexes distinguished by the presence of SANT
87. Warrener, R. et al. Tumor cell-selective cytotoxicity by Investig. Drugs 13, 2138 (2004). domain proteins CoREST/kiaa0071 and Mta-L1.
targeting cell cycle checkpoints. FASEB J. 17, 108. Johnstone, R. W. & Licht, J. D. Histone deacetylase J. Biol. Chem. 276, 68176824 (2001).
15501552 (2003). inhibitors in cancer therapy: is transcription the 129. Jones, P. L., Sachs, L. M., Rouse, N., Wade, P. A. & Shi,
88. Beamish, H., Warrener, R. & Gabrielli, B. G. Analysis primary target? Cancer Cell 4, 1318 (2003). Y. B. Multiple N-CoR complexes contain distinct
of checkpoint responses to histone deacetylase 109. Drummond, D. C. et al. Clinical development of histone deacetylases. J. Biol. Chem. 276,
inhibitors. Methods Mol. Biol. 281, 245259 (2004). histone deacetylase inhibitors as anticancer agents. 88078811 (2001).
89. Qiu, L. et al. Histone deacetylase inhibitors trigger a Annu. Rev. Pharmacol. Toxicol. 45, 495528 130. Li, J. et al. Both corepressor proteins SMRT and
G2 checkpoint in normal cells that is defective in (2004). N-CoR exist in large protein complexes containing
tumor cells. Mol. Biol. Cell 11, 20692083 (2000). 110. Chavez-Blanco, A. et al. Histone acetylation and HDAC3. EMBO J. 19, 43424350 (2000).
90. Richon, V. M., Sandhoff, T. W., Rifkind, R. A. & Marks, histone deacetylase activity of magnesium valproate 131. Xue, Y. et al. NURD, a novel complex with both ATP-
P. A. Histone deacetylase inhibitor selectively induces in tumor and peripheral blood of patients with cervical dependent chromatin-remodeling and histone
p21WAF1 expression and gene-associated histone cancer. A phase I study. Mol. Cancer 4, 22 (2005). deacetylase activities. Mol. Cell 2, 851861 (1998).
acetylation. Proc. Natl Acad. Sci. USA 97, 111. Kelly, W. K. et al. Phase I study of an oral histone 132. Yao, Y. L. & Yang, W. M. The metastasis-associated
1001410019 (2000). deacetylase inhibitor, suberoylanilide hydroxamic acid, proteins 1 and 2 form distinct protein complexes with
91. Gui, C. Y., Ngo, L., Xu, W. S, Richon, V. M. & Marks, P. A. in patients with advanced cancer. J. Clin. Oncol. 23, histone deacetylase activity. J. Biol. Chem. 278,
Histone deacetylase (HDAC) inhibitor activation of 39233931 (2005). 4256042568 (2003).
p21WAF1 involves changes in promoter-associated 112. Kelly, W. K. et al. Phase I clinical trial of histone 133. Zhang, Y. et al. Analysis of the NuRD subunits reveals
proteins, including HDAC1. Proc. Natl Acad. Sci. USA deacetylase inhibitor: suberoylanilide hydroxamic acid a histone deacetylase core complex and a connection
101, 12411246 (2004). administered intravenously. Clin. Cancer Res. 9, with DNA methylation. Genes Dev. 13, 19241935
92. Varshochi, R. et al. ICI182,780 induces p21Waf1 35783588 (2003). (1999).
gene transcription through releasing histone 113. Ryan, Q. C. et al. Phase I and pharmacokinetic study 134. Seigneurin-Berny, D. et al. Identification of
deacetylase 1 and estrogen receptor from Sp1 sites of MS-275, a histone deacetylase inhibitor, in patients components of the murine histone deacetylase 6
to induce cell cycle arrest in MCF-7 breast cancer cell with advanced and refractory solid tumors or complex: link between acetylation and ubiquitination
line. J. Biol. Chem. 280, 31853196 (2005). lymphoma. J. Clin. Oncol. 23, 39123922 (2005). signaling pathways. Mol. Cell. Biol. 21, 80358044
93. Archer, S. Y., Meng, S., Shei, A. & Hodin, R. A. 114. Piekarz, R. L. et al. Inhibitor of histone deacetylation, (2001).
p21(WAF1) is required for butyrate-mediated growth depsipeptide (FR901228), in the treatment of 135. Yamagoe, S. et al. Interaction of histone acetylases
inhibition of human colon cancer cells. Proc. Natl peripheral and cutaneous T-cell lymphoma: a case and deacetylases in vivo. Mol. Cell. Biol. 23,
Acad. Sci. USA 95, 67916796 (1998). report. Blood 98, 28652868 (2001). 10251033 (2003).
94. Shao, Y., Gao, Z., Marks, P. A. & Jiang, X. Apoptotic 115. Sandor, V. et al. Phase I trial of the histone 136. Fischle, W. et al. Enzymatic activity associated with
and autophagic cell death induced by histone deacetylase inhibitor, depsipeptide (FR901228, NSC class II HDACs is dependent on a multiprotein
deacetylase inhibitors. Proc. Natl Acad. Sci. USA 101, 630176), in patients with refractory neoplasms. Clin. complex containing HDAC3 and SMRT/N-CoR. Mol.
1803018035 (2004). Cancer Res. 8, 718728 (2002). Cell 9, 4557 (2002).
95. Rosato, R. R., Almenara, J. A., Grant, S. The histone 116. Piekarz, R. L. et al. T-cell lymphoma as a model for the 137. Verdin, E., Dequiedt, F. & Kasler, H. G. Class II histone
deacetylase inhibitor MS-275 promotes use of histone deacetylase inhibitors in cancer deacetylases: versatile regulators. Trends Genet. 19,
differentiation or apoptosis in human leukemia cells therapy: impact of depsipeptide on molecular 286293 (2003).

50 | JANUARY 2006 | VOLUME 6

2006 Nature Publishing Group

138. Gore, S. D. et al. Impact of prolonged infusions of the leukemia and acute myeloid leukemia. Blood 105,
putative differentiating agent sodium phenylbutyrate 959967 (2005). DATABASES
on myelodysplastic syndromes and acute myeloid The following terms in this article are linked online to:
leukemia. Clin. Cancer Res. 8, 963970 (2002). Acknowledgements Entrez Gene:
139. Patnaik, A. et al. A phase I study of pivaloyloxymethyl We acknowledge Bruno Amati, Gordon McVie, Antonello fcgi?db=gene
butyrate, a prodrug of the differentiating agent butyric Mai, Michela Prudenziati, Roberta Carbone Jr, Gabriele BAX | CBP | DR5 | FAS | FASL | HDAC1 | HDAC2 | HDAC4 |
acid, in patients with advanced solid malignancies. Bucci, Oronzina Botrugno, Maurizio Moroni, members of HDAC5 | HDAC6 | HDAC9 | histone H4 | Ku70 | MTA1 | p21 |
Clin. Cancer Res. 8, 21422148 (2002). our laboratories for discussions and critical reading of the p27 | p53 | p300 | PML | PLZF | RAR | RB | STAT3 | TRAIL | TXN
140. Raffoux, E., Chaibi, P., Dombret, H. & Degos, L. manuscript, and the reviewers of this manuscript for their National Cancer Institute:
Valproic acid and all-trans retinoic acid for the insightful comments. Acute promyelocytic leukaemia
treatment of elderly patients with acute myeloid
leukemia. Haematologica 90, 986988 (2005). Competing interests statement
See online article: S1 (table)
141. Byrd, J. C. et al. A phase 1 and pharmacodynamic The authors declare competing financial interests: see web
Access to this links box is available online.
study of depsipeptide (FK228) in chronic lymphocytic version for details.


2006 Nature Publishing Group
View publication stats