For either PMA or TGF-b1 to induce Timp-1 expression, new protein synthesis is required, and the induction of AP-1 factors closely precedes that of Timp-1.. This induction is dependent o
Trang 1phorbol ester- and TGF-b1 induced murine tissue inhibitor
of metalloproteinases-1 gene expression
David A Young*, Olivia Billingham, Clara L Sampieri, Dylan R Edwards and Ian M Clark
School of Biological Sciences, University of East Anglia, Norwich, UK
Remodelling of the extracellular matrix (ECM) is an
essential physiological process in, e.g development,
wound healing and angiogenesis Aberrant ECM
turn-over is also associated with a number of pathological
processes such as joint destruction in the arthritides,
tumour metastasis, and fibrosis [1] Central to the
turnover of ECM is the matrix metalloproteinase
(MMP) family; these number 23 enzymes in man which, between them, have the capability of degrading the majority of ECM proteins [2] Four tissue inhibi-tors of metalloproteinases (TIMPs) safeguard ECM integrity by virtue of their ability to inhibit the MMPs [3] The TIMPs display a high degree of func-tional overlap, but show dramatic differences in their
Keywords
acetylation; c-Jun; TGFb; TIMP; trichostatin A
Correspondence
I M Clark, School of Biological Sciences,
University of East Anglia, Norwich, NR4
7TJ, UK
Tel: +44 1603 592760
Fax: +44 1603 592250
E-mail: i.clark@uea.ac.uk
*Present address
Department of Rheumatology, University of
Newcastle- upon-Tyne, NE2 4HH, UK
(Received 16 November 2004, revised 12
January 2005, accepted 21 February 2005)
doi:10.1111/j.1742-4658.2005.04622.x
Expression of the tissue inhibitor of metalloproteinases-1 (Timp-1) gene can
be induced by either phorbol myristate acetate (PMA) or transforming growth factor b1 (TGF-b1), although the signalling pathways involved are not clearly defined Canonically, histone deacetylase inhibitors (HDACi) such as trichostatin A (TSA) or sodium butyrate (NaB) increase total cellular histone acetylation and activate expression of susceptible genes Remarkably, PMA and TGF-b1 stimulation of Timp-1 show a differential response to TSA or NaB TSA or NaB potentiate PMA-induced Timp-1 expression but repress TGF-b1-induced Timp-1 expression The repression
of TGF-b1-induced Timp-1 by TSA was maximal at 5 ngÆmL)1, while for the superinduction of PMA-induced Timp-1 expression, the maximal dose
is > 500 ngÆmL)1 TSA A further HDACi, valproic acid, did not block TGF-b1-induced Timp-1 expression, demonstrating that different HDACs impact on the induction of Timp-1 For either PMA or TGF-b1 to induce Timp-1 expression, new protein synthesis is required, and the induction of AP-1 factors closely precedes that of Timp-1 The effects of the HDACi can be reiterated in transient transfection using Timp-1 promoter con-structs Mutation or deletion of the AP-1 motif ()59 ⁄ )53) in the Timp-1 promoter diminishes PMA-induction of reporter constructs, however, the further addition of TSA still superinduces the reporter In c-Jun–⁄ – cells, PMA still stimulates Timp-1 expression, but TSA superinduction is lost Transfection of a series of Timp-1 promoter constructs identified three regions through which TSA superinduces PMA-induced Timp-1 and we have demonstrated specific protein binding to two of these regions which contain either an avian erythroblastosis virus E26 (v-ets) oncogene homo-logue (Ets) or Sp1 binding motif
Abbreviations
AP-1, activating protein-1; EMSA, electrophoretic mobility-shift assay; Ets, avian erythroblastosis virus E26 (v-ets) oncogene homologue; HAT, histone acetyltransferase; HDAC, histone deacetylase; HDACi, histone deacetylase inhibitor; MMP, matrix metalloproteinase;
NaB, sodium butyrate; PMA, phorbol myristate acetate; TIMP, tissue inhibitor of metalloproteinases; TGF, transforming growth factor; TSA, trichostatin A; VPA, valproic acid.
Trang 2patterns of expression both during development and in
response to stimuli Timp-1 gene expression can be
induced by a variety of stimuli including phorbol esters
(PMA), serum, transforming growth factor b (TGFb),
retinoids and interleukin-6 family members; where it
has been assessed, induction is at the level of
transcrip-tion [4] The expression of Timp-1 has been implicated
in many disease processes including tumour
progres-sion, fibrosis, cardiovascular disease and arthritis [3]
Modulation of Timp-1 may therefore have therapeutic
potential in several pathologies [5], and an
understand-ing of the mechanisms impactunderstand-ing upon Timp-1 gene
expression is paramount
Dissection of the Timp-1 gene promoter has revealed
several important cis-acting sequences An AP-1 site
located at)59 ⁄ )53 in the murine gene is important in
both basal and inducible Timp-1 gene expression, with
a neighbouring avian erythroblastosis virus E26 (v-ets)
oncogene homologue (Ets)-binding site playing a more
minor role [6–8] Many other regions of the gene
pro-moter and first intron have been shown to be
import-ant, e.g a hypoxic response element has been mapped
to)26 ⁄ )23 [9]; an Sp1 motif confers at least part of the
repressive nature of the first intron of the Timp-1 gene
by binding Sp1 and Sp3, along with an Ets-related
factor [10]
Recently, our laboratories have examined the
induc-tion of Timp-1 gene expression by TGF-b1, and shown
that this is independent of the Smad pathway [11]
This induction is dependent on the promoter proximal
AP-1 site, requires at least c-fos, c-Jun and JunD, and
is sensitive to extracellular signal-regulated kinase
(ERK) and p38 mitogen-acivated protein kinase (p38
MAPK) inhibitors Phorbol esters (e.g phorbol
myri-state acetate, PMA) are also robust inducers of Timp-1
gene expression Compared to the induction of Timp-1
by TGFb, the PMA induction occurs with more rapid
kinetics, is less sensitive to p38 MAPK inhibitors, and
less dependent on c-fos Hence, it appears that whilst
there may be some overlap in the signalling pathways
used by TGFb vs PMA to impact on the Timp-1 gene,
there are also some pathways exclusive to each factor
([11]; D A Young, D R Edwards and I M Clark,
unpublished observation)
The packaging of eukaryotic DNA into chromatin
plays an important role in regulating gene expression
The DNA is wound round a histone octamer
consist-ing of two molecules each of histones H2A, H2B, H3
and H4 to form a nucleosome This unit is repeated at
approximately 200 bp intervals with histone H1
associ-ating with the intervening DNA Nucleosomes are
gen-erally repressive to transcription, hindering access of
the transcriptional apparatus [12] However, two major
mechanisms exist that modulate chromatin structure to allow transcriptional activity: first, ATP-dependent nucleosome remodellers such as the Swi⁄ Snf complex [13,14] and second, the enzymatic modification of histones, via acetylation, methylation and phosphory-lation [15–18]
Acetylation by histone acetyltransferases (HATs) occurs on specific lysine residues on the N-terminal tails
of histone H2A, H2B, H3 and H4 This neutralization
of positive charge leads to a loosening of the his-tone:DNA structure, allowing access of the trans-criptional machinery; furthermore, the acetyl groups may associate with and recruit factors containing bromo-domains [12] Many transcriptional activators or coactivators have (or recruit) HAT activity, giving a mechanism whereby acetylation can be targeted to specific gene promoters [15,16] Conversely, histone deacetylases (HDACs) have also been characterized Hypoacetylation of histones associates with transcrip-tional silence, and several transcriptranscrip-tional repressors and corepressors have been identified which have (or recruit) HDAC activity [17,19] Non-histone substrates of HATs have also been described, e.g p53, E2F, NF-jB, Sp3 and c-Jun; these latter two transcription factors are known to be important in Timp-1 expression [20,21] Trichostatin A (TSA), sodium butyrate (NaB) and valproic acid (VPA) are HDAC inhibitors (HDACi) [22–24] Addition of these reagents to cells should there-fore block histone deacetylation and result in increased acetylation of susceptible genes The prediction would
be that this would lead to an increase in gene expression Here, we demonstrate for the first time that HDACi impact upon Timp-1 gene expression Furthermore, the response of the gene to HDACi is dependent upon the stimulus – either PMA or TGF-b1 – used to induce Timp-1expression Dose–response curves and the use of the more HDAC specific HDACi, VPA, shows that at least two targets of HDACi exist which affect down-stream Timp-1 expression Both TGF-b1 and PMA are known to act via the AP-1 motif to induce the Timp-1 gene We show that for HDACi to superinduce PMA-induced Timp-1, the AP-1 factor c-Jun is essential; how-ever, the HDACi acts through both an Ets and GC-box (Sp factor binding) motif in the Timp-1 promoter itself
Results
The effects of HDAC inhibitors on Timp-1 gene expression
As outlined above, the prediction is that HDAC inhibi-tors should induce expression of susceptible genes Figure 1A,B shows that both TSA and NaB
Trang 3superinduce PMA-induced Timp-1 expression measured
by qRT-PCR in C3H10T1⁄ 2 cells, but potently repress
TGF-b1 induction of the gene These effects appear
spe-cific for Timp-1, as no equivalent alteration in Timp-2
or -3 expression is seen (data not shown) These
unex-pected data suggest a differential involvement of
acetyl-ation in the control of Timp-1 gene expression by PMA
vs TGFb1 Figure 1C shows a comparison between two cell lines, C3H10T1⁄ 2 and Swiss3T3, stimulated with TGF-b1 or PMA, in the presence or absence of TSA to confirm the original observations were not a cell-type dependent phenomenon The fold induction by TSA above the PMA-induced level is higher in C3H10T1⁄ 2 than Swiss 3T3 as the response to PMA alone is greater in the latter cell line The effect of NaB
on induced Timp-1 expression in Swiss 3T3 cells also mirrors that seen in C3H10T1⁄ 2 (data not shown) Using an anti-(acetyl-lysine) Ig, TSA could be seen to cause an increase in acetylation of total histones from C3H10T1⁄ 2 nuclear cell extracts (Fig 1C; lower panel)
PMA- vs TGF-b1 induction of Timp-1 display differential sensitivity to HDACi
Representative dose–response curves of the effect of HDACi’s on either PMA- or TGF-b1-induced Timp-1 gene expression are shown in Fig 2A–C For PMA induction of Timp-1 expression, TSA superinduces this expression with an optimum dose of 250 > 1000 ngÆmL)1 (0.8–3.3 lm, depending upon experiment), whilst for NaB the optimum dose is 5 mm Another known HDACi, VPA, had no effect on PMA-induced Timp-1 expression until a concentration greater than
2 mm was added In contrast to this, TGF-b1-induced Timp-1expression is more sensitive to HDACi, with an optimum dose of TSA being less than 50 ngÆmL)1 (165 nm) (and in a further experiment 5 ngÆmL)1 still potently inhibited TGF-b1 induction of Timp-1) and of NaB less than 1 mm VPA had no effect on TGF-b1-induced Timp-1 expression, but was shown to be functional as, in the same samples, even the lowest concentration of VPA (0.5 mm) repressed TGFb1-induced ADAM12 expression (Fig 2C, inset panel)
A
B
C
Fig 1 Histone deacetylase inhibitors have differential effects on PMA- vs TGF-b1-induced Timp-1 expression (A and B) C3H10T1 ⁄ 2 murine fibroblasts were serum starved for 24 h, then stimulated with 10)7M PMA or 2 ngÆmL)1TGF-b1 for 6 h in the presence or absence of (A) 250 ngÆmL)1TSA or (B) 1 m M NaB Total RNA was isolated and subjected to real-time qRT-PCR using a specific primer set for the Timp-1 gene [49]; data were normalized to the 18S rRNA housekeeping gene Data are plotted as mean + SEM (A)
n ¼ 7 (B) n ¼ 3 (C) C3H10T1 ⁄ 2 and Swiss-3T3 fibroblast cells were serum starved for 24 h, then stimulated with 10)7M PMA
or 4 ngÆmL)1 TGF-b1 for 6 h in the presence or absence of
500 ngÆmL)1 TSA Isolated total RNA subjected to qRT-PCR and normalized as described above Nuclear extracts (10 lg) from the C3H10T1 ⁄ 2 cells were western blotted with an anti-(acetyl–lysine)
Ig to monitor acetylation (histone band assigned by molecular mass and abundance).
Trang 4These data show that different HDACs are involved in
the TGFb vs PMA induction of Timp-1 Two specific
inhibitors of the Sir2 family of deacetylases (class 3
HDACs), sirtinol and nicotinamide, had no effect on
induced or basal Timp-1 expression (data not shown)
showing that class 3 HDACs are not involved
Induction of the Timp-1 gene by PMA or TGF-b1
requires new protein synthesis
In order to assess the possibility that the effects of
HDACi are secondary, acting to modulate the
expression of an intermediate, the requirement for
new protein synthesis in the induction of Timp-1 was
assessed Figure 3 shows that addition of the protein synthesis inhibitor, emetine, completely abrogates both PMA- and TGF-b1-induction of the Timp-1 gene In the presence of emetine, the HDACi have no further effect (data not shown) The action of HDACi could therefore
be either on the Timp-1 gene itself, or on the expression
of a protein(s) required for induction of the Timp-1 gene
Time course of TSA action upon PMA- and TGF-b1-induced Timp-1 expression
The time course of induction of Timp-1 gene expression
by PMA and TGF-b1 measured by qRT-PCR is identi-cal to our previous northern blot data with PMA giving
a more rapid but transient induction and TGF-b1 indu-cing a slower more sustained stimulation of the gene [25] The effect of TSA on both PMA- and TGF-b1-induced Timp-1 expression is evident as early as 3 h after addition of reagents (Fig 4A) This represents the earliest time point that induction of the gene by PMA
or TGF-b1 is measurable by qRT-PCR The magnitude
of TSA superinduction of PMA-induced Timp-1 increa-ses to 12 h, and remains at 24 h, even when the PMA-induced levels have returned to baseline TSA continues
to repress TGF-b1-induced Timp-1 expression for as long as the TGF-b1 induction is measurable (> 24 h)
Induction of c-fos by PMA, TGF-b1 and TSA immediately precedes that of Timp-1
As both PMA and TGF-b1 require new protein synthe-sis to induce Timp-1 expression, we examined the
Fig 2 Histone deacetylase inhibitors display different
dose-responses on PMA- vs TGF-b1-induced Timp-1 expression.
C3H10T1 ⁄ 2 murine fibroblasts were serum starved for 24 h, then
stimulated with 10)7M PMA or 2 ngÆmL)1TGF-b1 for 6 h in the
presence or absence of (A) 0–1000 ngÆmL)1 TSA; (B) 0–10 m M
NaB, or (C) 0–8 m M VPA Total RNA was isolated and subjected to
real-time qRT-PCR for Timp-1; data were normalized to the 18S
rRNA housekeeping gene Data is representative of at least two
independent experiments in all cases.
Fig 3 The induction of Timp-1 by PMA or TGF-b1 is protein synthe-sis dependent C3H10T1 ⁄ 2 murine fibroblasts were serum starved for 24 h, then stimulated with 10)7M PMA or 2 ngÆmL)1TGF-b1 for
6 h in the presence or absence of the protein synthesis inhibitor emetine at 10 lgÆmL)1 Total RNA was isolated and subjected to real-time qRT-PCR for Timp-1; data were normalized to the 18S rRNA housekeeping gene Data is plotted mean + SEM, n ¼ 3.
Trang 5expression of the AP-1 family member c-fos, over the
same time course experiment (Fig 4A) Previously, we
have shown that c-fos overexpression induces Timp-1
expression [11]; c-fos is a known immediate early gene induced by PMA We initially confirmed this using the protein synthesis inhibitor cycloheximide, which as
B
A
C
Fig 4 Time-course of trichostatin A action
on PMA- vs TGF-b1-induced Timp-1 and c-fos expression and AP-1 binding.
C3H10T1 ⁄ 2 murine fibroblasts were serum starved for 24 h, then stimulated with 10)7M
PMA or 2 ngÆmL)1TGF-b1 in the presence or absence of 250 ngÆmL)1TSA (A) Total RNA was isolated at timepoints 1, 3, 6, 12 and
24 h and subjected to real-time qRT-PCR for Timp-1 and c-fos; data were normalized to the 18S rRNA housekeeping gene Data plotted is representative of three independ-ent time course experimindepend-ents (B) Nuclear extracts were isolated at 1 and 3 h after stimulation and subjected to EMSA using the Timp-1 AP-1 motif as probe (C) Isolated nuclear extracts (2 lg) from cells stimulated with PMA or PMA and TSA for 3 h were incubated in the presence or absence of antibodies against either acetyl–lysine and ⁄ or c-fos and subjected to EMSA.
Trang 6expected, did not block the induction of c-fos mRNA by
PMA The kinetics of c-fos induction closely precedes
that of Timp-1 At 1 h, c-fos was dramatically induced
by PMA Further, this induction was superinduced by
TSA; by 3 h the PMA induction was lost, but TSA in
the presence of PMA still superinduced c-fos expression
Compared to PMA, TGF-b1 induction of c-fos was
delayed, only becoming apparent by 3 h; this induction
remained at 6 h but was lost by 12 h Similar to Timp-1
expression, TSA blocked TGF-b1 induced c-fos at all
time points where TGF-b1 alone induced c-fos
expres-sion Interestingly, TSA alone induced c-fos but only
after 12 h of stimulation
The Timp-1 gene contains an AP-1 motif at
)59 ⁄ )53 bp relative to the transcription start site [6]
Nuclear proteins were isolated from cells stimulated
for either 1 or 3 h with TGF-b1 or PMA, with or
without TSA In an electrophoretic mobility-shift assay
(EMSA), AP-1 factors could be seen to bind a 30 bp
sequence encompassing the Timp-1 AP-1 motif and
this binding was induced by PMA (at both 1 and 3 h)
or TGF-b1 stimulation (at 3 h) (Fig 4B) As with the
induction of c-fos mRNA, AP-1 protein binding
activ-ity was greater with PMA than TGF-b1, and the
induction by PMA occurred earlier than that by
TGF-b1 Stimulation with TSA alone or in the presence of
PMA or TGF-b1 appeared to have little effect on the
overall amount of AP-1 protein binding the Timp-1
AP-1 sequence; this was in marked contrast to that
seen for c-fos mRNA levels
The specificity of the AP-1 binding activity was
con-firmed using DNA-binding competition studies The
AP-1 complex could be competed by an excess of
‘wild-type’ DNA, but not by an equivalent DNA
frag-ment containing a mutation in the AP-1 binding
sequence (DAP-1; Fig 4C) The AP-1 complex was not
seen when the DAP-1 DNA was used as the
radio-labelled probe (data not shown) Further, the presence
of c-fos in the AP-1 complex was confirmed by
super-shift analysis using an anti-(c-fos) Ig (Fig 4C) As the
total binding of the AP-1 complex did not significantly
alter on the addition of TSA, and c-Jun, another AP-1
member, is a potential target for acetylation, we
per-formed supershift analysis using an antibody raised
against acetylated lysine When added to nuclear
extracts, a low mobility ‘supershifted’ complex was
evi-dent in all extracts treated with TSA However, the
binding intensity of this complex did not alter upon
various stimulations and the appearance of the
super-shift did not coincide with the loss of another band
(Fig 4C) Further analysis confirmed this complex did
not appear to contain c-Jun or other AP-1 factors [i.e
antibodies to these factors do not alter the supershift
seen with the anti-(acetyl-lysine) Ig, data not shown] and the lack of competition for this band by the excess
of cold AP-1 oligonucleotide (Fig 4C) strongly sug-gests that it is not related to AP-1 factor acetylation
TSA superinduction of PMA-induced Timp-1 requires c-Jun
To establish unequivocally the role of specific AP-1 family members upon Timp-1 expression, qRT-PCR was performed on RNA from c-fos, c-Jun or junD defi-cient cells (–⁄ –) stimulated with PMA or TGF-b1 with
or without TSA; Swiss-3T3 cells were used as a control Surprisingly, Swiss-3T3, c-fos–⁄ – and junD– ⁄ – cells had the same Timp-1 expression profile as each other and as that seen previously for C3H10T1⁄ 2 cells (Fig 5A) The fold induction by PMA or TGF-b1 was remarkably similar between the different cell types and for each of those three cell lines, TSA superinduced PMA-induced Timp-1 expression and repressed TGF-b1-induced Timp-1 expression This shows that neither c-fos nor junD are essential for the observed affects of TSA on Timp-1 expression However, although the expression of Timp-1 in c-Jun–⁄ – cells was induced by either PMA or TGF-b1, TSA was unable to super-induce the PMA-super-induced Timp-1 expression whilst retaining the ability to repress the TGF-b1 induced Timp-1 expression In fact, in c-Jun–⁄ – cells, much like for the TGF-b1 response, TSA repressed PMA-induced Timp-1, indicating that in the absence of c-Jun (or a c-Jun-regulated factor), the default pathway for the effect of TSA on induced Timp-1 expression is repressive
Many AP-1 members are differentially regulated
in response to PMA or TGF-b1 with or without TSA
As even in the absence of c-fos, c-Jun or junD mouse fibroblast cells are able to induce Timp-1 in response
to PMA or TGF-b1, and yet the AP-1 motif in the Timp-1 promoter is important for such a response, the expression profile by qRT-PCR of all the Fos and Jun family members was determined in C3H10T1⁄ 2 cells stimulated for 1 h (Fig 5B) PMA significantly induced fosB, fra-1, fra-2, junB and c-Jun; of these only junB was further induced by TSA, while PMA-induced fra-2 was repressed by TSA TGF-b1 PMA-induced levels of fosB, fra-2 and junB, all of which were then repressed by TSA ATF2 and junD in general showed little regulation by TSA, PMA or TGF-b1 It is there-fore possible that in AP-1 deficient cells, the lack of a specific factor may be compensated for by the presence
Trang 7of an additional, functionally overlapping, family
member However, the function of c-Jun, in mediating
the superinduction of PMA-induced Timp-1
expres-sion, appears unique
The effects of HDAC inhibitors on Timp-1 gene
expression can be reiterated in transient
transfec-tion of Timp-1 promoter-containing plasmids
Figure 6A shows that the effect of TSA on both
PMA- and TGF-b1-induced Timp-1 expression can be
reiterated in a )95 ⁄ +47 Timp-1 promoter-reporter
construct (in the pGL2 vector), transiently transfected
into C3H10T1⁄ 2 cells Deletion mutation to )50 ⁄ +47,
removing the promoter-proximal AP-1 site, shown
previously to be important in induction of the Timp-1 gene, shows that TGF-b1-induction is lost (and hence, TSA can no longer repress this); however, some PMA induction of the deleted construct remains, and this is superinduced by TSA (Fig 6B) These data were con-firmed using point mutation of the AP-1 site in a )223 ⁄ +47 Timp-1 promoter construct (in the pGL3 vector); here, some PMA- and TGF-b1-induction of the mutant construct remains TSA superinduces PMA-induced expression of the wild-type and mutant AP-1 constructs, but TSA no longer represses the residual TGF-b1-induction (Fig 6C,D) This suggests that the effect of TSA on TGF-b1-induced Timp-1 expression is mediated through the promoter proximal AP-1 site, whilst the effect of TSA on PMA-induced
Fig 5 TSA superinduction of PMA-induced Timp-1 requires c-Jun but other AP-1 members may compensate for the loss of c-fos (A) Swiss-3T3, c-fos– ⁄ –, c-Jun– ⁄ – and junD– ⁄ – mouse fibroblast cells were serum starved for 24 h, then stimulated with 10)7M
PMA or 4 ngÆmL)1TGF-b1 in the presence or absence of 500 ngÆmL)1TSA Total RNA was isolated after 6 h and subjected to real-time qRT-PCR for Timp-1 Data were normalized
to the 18S rRNA housekeeping gene Data is representative of two independent experi-ments with each experiment performed in triplicate and data is plotted as mean + SEM (B) C3H10T1 ⁄ 2 murine fibroblasts were serum starved for 24 h, then stimulated with
10)7M PMA or 2 ngÆmL)1TGF-b1 in the presence or absence of 250 ngÆmL)1 trichostatin A (TSA) for 1 h before the isolation of total RNA qRT-PCR for AP-1 members fosB, fra-1, fra-2, ATF2, junB, junD and c-Jun is shown, plotted as fold-control levels for direct comparison C, control; P, PMA (10)7M ) and T, TGF-b1 (4 ngÆmL)1) Data is representative of three independent experiments.
Trang 8Timp-1 expression is independent of the Timp-1 AP-1
motif It should be noted that TSA alone induces
expression from promoter constructs made in pGL3,
and similarly induces the empty pGL3-basic vector
(data not shown) The more robust expression levels
from this pGL3 compared to pGL2 (which is not
induced by TSA) make the data more reliable despite
this difference, and the effects of TSA upon
PMA-and TGF-b1-stimulated promoter-reporter expression
remain clear
TSA superinduction of PMA-induced Timp-1
is lost by mutation of a GC-box or Ets binding
motif within the Timp-1 promoter
In order to establish if promoter elements downstream
of the AP-1 site can mediate the TSA superinduction
of PMA-induced Timp-1, a series of insertion mutants
were prepared in)223 ⁄ +47 In this set of constructs,
the AP-1 site at )59 ⁄ )53 remains intact, but
down-stream of this, blocks of five bases are replaced with
adenosine; these mutants overlap by two bases, giving
a set of 20 mutant constructs Figure 7A shows that only mutants m4, m16 and m20 lose the superinduc-tion of PMA-induced expression by TSA Mutant m4 alters a canonical Ets binding site, shown previously to
be important for basal expression of the Timp-1 gene; mutant 16 alters a canonical Sp1 binding site (GC-box); mutant 20 does not alter any known consensus for transcription factor binding
EMSA was used to determine the protein factors binding to the ‘wild-type’ m4, m16 and m20 or the mutated sequences Specific factors binding to both the
‘wild-type’ m4 and m16 sequences could be seen (Fig 7B,C), however, the binding of these factors did not change upon TSA stimulation (data not shown) The m4 sequence (Fig 7B) bound several complexes, although competition analysis with the ‘wild-type’ and mutant m4 revealed only one complex to be specific The identity of this, presumably Ets family member, remains to be determined The pattern of three bands (a, b and c) bound to the ‘wild-type’ m16 probe in Fig 7C is identical to that described in the literature as deriving from the binding of both Sp1 and Sp3
trans-Fig 6 The impact of TSA on PMA- vs TGF-b1-induced Timp-1 gene expression is reiterated on Timp-1 promoter-reporter constructs and the effect of TSA on PMA-induced Timp-1 is independent of the promoter proximal AP-1 site C3H10T1 ⁄ 2 murine fibroblasts were transi-ently transfected with (A) )95 ⁄ +47, (B) )50 ⁄ +47 Timp-1 promoter constructs in pGL2, (C) )223 ⁄ +47 or (D) )223 ⁄ +47 DAP-1 Timp-1 promo-ter constructs in pGL3 Following serum starvation for 24 h, cells were stimulated with 10)7M PMA or 2 ngÆmL)1TGF-b1 for 6 h in the presence or absence of 250 ngÆmL)1trichostatin A (TSA) prior to harvest for luciferase assay Experiments were performed three times in triplicate Results are plotted as mean + SEM.
Trang 9cription factors [26] The ‘wild-type’ dsDNA (50-fold)
competed for all three Sp bands while the mutant m16
at the same concentration did not An anti-Sp1 Ig
resulted in the loss of band a, and partially blocked b
with the subsequent appearance of a supershifted
com-plex; anti-Sp3 Ig blocks formation of bands b and c
This suggests that band a contains Sp1, band b
con-tains both Sp1 and Sp3, and band c concon-tains Sp3
Discussion
HDACs usually act as transcriptional repressors,
there-fore HDAC inhibitors should induce expression of
susceptible genes, and this is the typical experimental
finding However, in yeast, deletion of the HDAC
Rpd3 down-regulates a subset of genes, and many of these are also repressed by treatment with TSA [27] There are also many individual instances of HDAC inhibitors acting as repressors of gene expression, e.g TSA and NaB cause a reduction in mRNA levels for the cdk1 gene [28]; TSA represses b-casein expression
in mammary epithelial cells [29]; TSA represses cyclin B1 and A [30]; TSA inhibits MMTV transcription [31] One could postulate that these effects are indirect, with TSA leading to the induction of a factor (or factors) involved in the repression of a downstream target; alternatively, a direct effect on either the acetylation of a transcription factor, or recruitment of repressive factors
to acetyl–histones via bromodomain interactions could
be envisaged In support of the latter notion the
down-B
A
C
Fig 7 The impact of TSA on PMA-induced Timp-1 gene expression is mediated via three sites in the proximal promoter (A) C3H10T1 ⁄ 2 murine fibroblasts were transi-ently transfected with a )223 ⁄ +47 Timp-1 promoter construct in pGL3 and 20mutant constructs as shown Following serum starvation for 24 h, cells were stimulated with 10)7M PMA in the presence (solid bars)
or absence (open bars) of 500 ngÆmL)1 trichostatin A (TSA) prior to harvest for luci-ferase assay Data is representative of three independent experiments, each performed in triplicate and results are plotted as mean + SEM On the Timp-1 sequence, AP-1, Ets and Sp1 binding motifs are shown in bold, position of each of the 20 (m1 to m20) d(A) 5
mutation are shown underlined (B) PMA (10)7M ) + TSA (500 ngÆmL)1) stimulated nuclear were incubated with a ‘wild-type’ m4 DNA probe (Table 2) and subjected to EMSA.
A 50-fold excess of self and mutant m4 DNA was used to define binding specificity (C) Nuclear extracts (as in B) were incubated with a ‘wild-type’ m16 DNA probe (Table 2) and subjected to EMSA A 50-fold excess of self and mutant m16 DNA confirmed binding specificity Sp factor binding was confirmed
by incubation of extracts with either 2 lg
of an anti-Sp1 Ig or anti-Sp3 Ig prior to electrophoresis.
Trang 10regulation of cyclin A and B1 upon TSA treatment is
via diminished activity of NF-Y-associated HAT and
is mediated through CCAAT motifs; further, the
diminished HAT activity is mediated by
phosphoryla-tion of hGCN5 [30] Moreover, the inhibiphosphoryla-tion of
MMTV transcription by TSA does not depend on
changes in chromatin remodelling or increased histone
acetylation, but is mediated via the TATA-box region
[31]
Both TGF-b1- and PMA-induction of Timp-1
require new protein synthesis As both events have at
least some dependency on a promoter proximal AP-1
site, this could reflect in part the synthesis of Fos and
Jun family members Immediate early gene induction
correlates with a nucleosomal response whereby
gene-associated nucleosomes are subject to both
phosphory-lation and acetyphosphory-lation on histone H3 and acetyphosphory-lation
on histone H4 [16] Whilst it is reported that TSA does
not activate c-fos or c-Jun expression in C3H10T1⁄ 2
cells [32], TSA can clearly modulate these genes in the
additional presence of TGF-b1 or PMA (Figs 4A and
5B) Moreover, TSA alone did induce c-fos expression
in our experiments after 12 h of stimulation As the
kinetics of c-fos induction so clearly preceded that of
Timp-1, we analysed the expression of Timp-1 in c-fos
deficient mouse fibroblasts (Fig 5A) Surprisingly,
mouse cells lacking c-fos, c-Jun or junD showed no
alteration in Timp-1 induction by PMA or TGF-b1
This could be due to compensation for the lack of the
factor by another AP-1 family member and our data
demonstrate that expression of many AP-1 members
are up-regulated in response to PMA or TGF-b1
However, Timp-1 expression in PMA-stimulated
c-Jun–⁄ – cells was not superinduced in response to
TSA, and was in fact repressed, resembling the
situ-ation with TGF-b1 induction c-Jun is known to be
acetylated at Lys271 by the transcriptional coactivator
p300 upon its interaction with the adenoviral protein
E1A [21] Overexpression of c-Jun in C3H10T1⁄ 2 cells
induced Timp-1 reporter expression by twofold, but
mutation of Lys271fi Arg had no effect on this
induction, even in the presence of PMA and⁄ or TSA
(data not shown)
Phosphorylation of c-Jun by the mitogen-activated
protein kinase (MAPK) JNK on Ser63 and Ser73, as
well as on Thr91 or Thr93, or both, increases its
trans-activating potential and DNA-binding activity by
mediating its dissociation from an inhibitory complex
containing HDAC3, a class 1 HDAC [33–35] HDAC3
associates with class 2 HDACs and c-Jun in repressor
complexes such as those containing the corepressors
N-CoR and SMRT [34,36] Phosphorylation by JNK
causes a reduction in c-Jun ubiquitination and
subsequent protein stabilization [37,38] The proteo-some inhibitor lactacystin inhibited the PMA induction
of Timp-1 and no induction was seen in the additional presence of TSA (data not shown) We propose that lactacystin prevents the degradation of ubiquitinated c-Jun thus leading to its accumulation and preventing its activation by JNK and subsequent downstream acti-vation events that would lead to Timp-1 up-regulation
A possible explanation for the lack of a TSA super-induction of PMA-induced Timp-1 in c-Jun–⁄ – cells is that the expression of one or more HDACs is c-Jun-dependent To test this, we monitored the expression
of HDACs 1–11 in response to TGF-b1 or PMA, with
or without TSA, between C3H10T1⁄ 2, Swiss-3T3 and c-Jun–⁄ – cells by RT-PCR The expression of class 1 HDACs is reported to be ubiquitous while class 2 HDAC appear more tissue-specific [39] All three cell lines expressed the majority of HDACs, with only expression of HDAC9 and )10 being undetectable and HDAC8 was up-regulated in c-Jun–⁄ – cells compared
to C3H10T1⁄ 2 or Swiss-3T3 (data not shown) Only HDAC7 and HDAC11 were regulated differentially by any of the stimuli, and this was identical between all three cell lines HDAC7 was, in all cell lines, repressed
by the presence of TSA while HDAC11 expression was interestingly induced only by the combination of TGF-b1 and TSA (data not shown) Hence, none of the HDACs exhibit c-Jun dependent expression
Although the induction by PMA alone was partially abrogated upon mutation or deletion of the )59 ⁄ )53 AP-1 motif of Timp-1, TSA was still able to superin-duce reporter expression in the presence of PMA (Fig 7) This suggests that the impact of c-Jun on the superinduction of Timp-1 is not directly on the Timp-1 gene, but via a c-Jun-dependent intermediate It should
be noted that whilse transiently transfected plasmid DNA is not integrated into the host cell chromosomes, there is evidence that it can be assembled into a chro-matin-like structure [40,41] If this is true in the current system, then data from transient transfection experi-ments could still be interpreted at the level of histone
or factor acetylation
Using a series of 20 overlapping mutant promoter constructs we demonstrated three mutants, m4, m16 and m20, were no longer able to superinduce reporter expression above PMA alone in the additional presence
of TSA (though variation in absolute levels of induc-tion is seen across the mutant constructs) Further, spe-cific binding of Sp1 and Sp3 to wild-type m16 and a putative Ets factor to wild-type m4 were identified (Fig 7C,D) It is surprising that other mutants that overlap the consensus sequences for these transcription factors do not impact upon the effect of TSA The