THE MECHANISM OF PPARN3 MEDIATED DOWN REGULATION OF SODIUM HYDROGEN EXCHANGER 1 (NHE1) GENE EPXRESSION AND ITS INHIBITION BY ESTROGEN RECEPTOR n1 6

169

4. DISCUSSION
4.1 ESTABLISHING THE RELATIONSHIP BETWEEN PPARγ
ACTIVATION AND NHE1 EXPRESSION
The present study provides evidence supporting a relationship between PPARγ
activation and NHE1 regulation. Firstly, treatment with PPARγ agonists down-
regulated NHE1 protein and mRNA expression in human breast cancer cells
overexpressing PPARγ. Furthermore, a functional PPRE sequence was identified
in the promoter region of NHE1 gene. The repressive effect of PPARγ ligands on
NHE1 was abrogated in the presence of PPARγ antagonist GW9662, further
confirming the inhibitory effect of PPARγ agnists on NHE1 expression is
mediated through PPARγ receptor.
4.1.1 Identification of NHE1 gene as a transcriptional target of PPARγ.
The peroxisome proliferator-activator receptors (PPARs) are transcription factors
that can be ligand-activated and they belong to the nuclear hormone receptor
family. Upon ligand activation, PPARγ heterodimerizes with a retinoid X receptor
(RXR) and subsequently binds to specific peroxisome proliferator response
elements (PPREs) in promoter regions of regulated target genes.
In the past 20 years since the discovery of PPAR, more than 70 PPAR target
genes have been identified to contain functional PPREs. However, majority of the
genes are clustered in two functional pathways of adipocyte differentiation and
fatty acid metabolism. Among the PPAR target genes listed in table 1, Acy-CoA
oxidase, Fatty acid transport protein, HMG-CoA synthase, Malic enzyme and
170

Phosphoenolpyruvate carboxykinase are known to be invoved in fatty acid
metabolism. These genes share common DR1 motif of classical PPRE consensus:
ATGGTCA N AGGTCA, with slight difference in one or two nucleotides.
As the first step to establish NHE1 as a PPARγ target, we located a putative PPRE
situated between nucleotide (nt) –977 to –990 relative to the TATA box in the

human NHE1 promoter (Accession number: L25272). However, the putative
PPRE identified contained DR2 repeats instead of the DR1 motif found in
classical PPRE-regulated genes. Interestingly, evidence from litreature has
demonstrated binding of PPARγ to DR2 element (Fontaine et al., 2003; Kumar et
al., 2004). The PPRE found in NHE1 promoter is located in Alu receptor response
element (AluRRE) (Jurka and Milosavljevic, 1991). Sequence alignment of
AluRRE of the human NHE1 with human myeloperoxidase (MPO) promoter
revealed extensive sequence identity (Figure 1B). Incidentally, PPARγ was
reported to bind to and regulate MPO gene (Kumar et al., 2004). Hence, the
presence of similar PPRE in NHE1 promoter may also allow PPARγ regulation in
the same manner. It is noteworthy that DR2 element can also be recognized by
retinoic acid recptors by functioning as retinoic acid response element (Laperriere
et al., 2007), suggesting the possibility of the same promoter region in NHE1
being regulated by retinoic acids.
To validate the identified PPRE on NHE1 promoter as a functional PPARγ-
binding element, we performed Noshift Transcritption Factor assay as well as
chromatin immunoprecipitation (ChIP). Noshift Transcription Factor assay was
used to assess direct interaction between protein and DNA oligomers in vitro,
171

while the ChIP assay was for detecting in vivo interaction between protein of
interst and DNAs of known sequences. In Noshift Trancription Factor assay, we
confirmed that PPARγ binds to the DR2 of putative PPRE in NHE1 promoter
(Figure 8B). The ability of competitive non-biotinylated oligonucleotides to
dimish the ELISA signal further confirmed the specificity of the protein-DNA
interaction. The assay was also validated using PTEN promoter sequence (Figure
8B), which was reported to be bound by activated PPARγ (Patel et al., 2001). One
limitation of the assay is that it only assesses the binding of protein with naked
DNA that is artificially synthesized and exogenously introduced, thus it does not
reflect the binding between protein and chromatin in their native cellular

environment. Hence, ChIP assay was also performed to confirm the in vivo
binding of activated PPARγ to putative PPRE on NHE1 promoter. Consistent
with data from the Noshift Transcription Factor assay, PPARγ was shown to bind
to the PTEN promoter in vivo (Figure 8C). Interestingly, the binding capacity of
PPARγ to NHE1 promoter appeared to be weaker than that to PTEN promoter in
both assays. As PTEN contains classical DR1 motif of PPRE, we surmise the
binding of PPARγ to DR2 in NHE1 promoter is not optimal compared to DR1.
Furthermore, the binding is depedent on PPARγ activation, as the amount of
PPARγ increased at NHE1 promoter with higher concentration of the ligand
(Figure 8C). Although the binding pattern of auxiliary proteins such as RXR was
not demonstrated, the results from these two assays sufficiently confirmed that
PPARγ can bind to the PPRE identified at the promoter of NHE1.
4.1.2 The mechanism of PPARγ-mediated repression of NHE1 gene.
172

After confirming the prescence of functional PPRE on NHE1 promoter, we next
investigated how PPARγ agonists regulate NHE1 expression in breast cancer cell
lines. To this end, MCF-7, MDA-MB-231 and T47D cells were treated with
different doses of synthetic ligands as well as natural ligand of PPARγ, and the
protein expression and mRNA transcript level of NHE1 were assessed. 15d-PGJ
2

is a metabolite of eicosanoid prostaglandin J
2
, and was reported to be the most
potent natural agonist for PPARγ with K
ds
varying from 325nM to 2.5µM. In all
cell lines tested, 15d-PGJ
2

induced a concentration-dependent down-regulation of
NHE1 expression both at protein and transcriptional level (Figure 2, 3). Synthetic
PPARγ ligands, which belong to the TZD family of drugs, ciglitazone,
troglitazone and rosiglitazone, also induced similar inhibition on NHE1 mRNA in
these cell lines (Figure 2). The extent of inhibition varied from ligand to ligand,
this could be explained by the different binding affinities of these ligands to
PPARγ receptor. Taken together, we surmise that PPARγ ligands induced down-
regulation of NHE1 expression is mediated through PPARγ receptor, and is not a
drug-specific effect.
Although repressive effect was seen in all three cell lines tested, the magnitude of
inhibition varied from cell line to cell line. The most drastic inhibition of NHE1
expression by 15d-PGJ
2
was observed in MDA-MB-231 cells and the least
inhibition was found in T47D (Figure 2, 3). The extent of inhibition correlatively
mirror-imaged the basal PPARγ receptor level present in different cell lines.
T47D which expresses the lowest amount of PPARγ was minimally affected by
treatment with 15d-PGJ
2
. Conversely, MDA-MB-231 with highest level of
173

PPARγ was the most susceptible to 15d-PGJ
2
-induced down-regulation of NHE1
expression. This result provides the first evidence that PPARγ ligands-regulated
NHE1 repression is PPARγ-dependent.
To further confirm the hypothesis, we overexpressed PPARγ in T47D cells, and
checked whether the increased basal PPARγ level would lead to enhanced
repression by PPARγ ligand. As expected, T47D transfected with PPARγ receptor

showed augmented reduction in NHE1 protein and mRNA expression upon
treatment with 15d-PGJ
2
as compared to cells that were transfected with empty
vector (Figure 4). On the other hand, overexpression of PPARγ defective in DNA
binding domain abrogated the ligand-mediated down-regulation of NHE1
expression in MCF-7 cells (Figure 7B, C). We also silenced PPARγ in MDA-MB-
231 cells which express higher basal level of PPARγ and assessed whether the
absence of PPARγ would lead to attenuated down-regualtion of NHE1 by PPARγ
ligand. As expected, silencing of PPARγ produced a reverse effect as compared to
overexpression of PPARγ. The initial 15d-PGJ
2
-mediated down-regualtion of
NHE1 protein was abrogated upon PPARγ silencing (Figure 5). Furthermore,
pharmocoligical PPARγ antagonist, GW9662 produced similar effect as PPARγ
silencing in MDA-MB-231 cells. It was observed that inhibiting PPARγ
activation by GW9662 blocked the down-regulation of NHE1 protein by PPARγ
ligand (Figure 6). Together, these data confirmed our hypothesis that PPARγ
ligands inhibit NHE1 expression through a PPARγ-dependent mechanism.
The inhibition of NHE1 by PPARγ can be due to negative interference with
activating transcription factors such as NF-κB and activator protein-1 (Delerive et
174

al., 1999; Rossi et al., 2000). Alternatively, the inhibitory effect can be caused by
sequestration of limiting coactivator such as CBP (Li et al., 2000). Beside, the
regulatory mechanism can also involve direct recruitment of corepressors such as
NCoR and SMRT by PPARγ in a promoter-specific manner (Guan et al., 2005).
Recently, it was also reported that sumolyated PPARγ inhibits the expression of
inducible nitric oxide synthase gene in a ligand-dependent manner (Pascual et al.,
2005). Given that PPARγ binds to PPRE on the NHE1 promoter and PPARγ

defective in DNA-binding abrogated the PPARγ-mediated NHE1 repression, the
mechanism underlying down-regulation of NHE1 expression by PPARγ ligands
may involve direct binding of sumolyated PPARγ or its recruitment of
corepressors.
4.1.3 Production of ROS/RNS by PPARγ ligands in breast cancer cells.
Cyclopentenone prostaglandins including 15d-PGJ
2
are reported to generate
intracellular ROS/RNS in various cell lines (Li et al., 2001; Bureau et al., 2002;
Lennon et al., 2002). Furthermore, studies have also shown that the TZD class of
synthetic PPARγ ligands are capable of induce ROS/RNS by affecting
mitochondria function (Brunmair et al., 2004; Perez-Ortiz et al., 2007).
Corroborating with these reports, we also confirmed the production of ROS/RNS
in breast cancer cells by 15d-PGJ
2
and ciglitazone. Using redox sensitive DCFDA
staining, we observed an increase in intracellular ROS/RNS induced by PPARγ
ligands (Figure 9).
175

As the DCFDA dyes are sensitive to a variety of ROS/RNS species such as H
2
O
2
,
ONOO
–
, HO• and ROO•, it is imperative to charactarize the specific oxygen
derivatives induced by PPARγ ligands. A study by Perez-Ortiz et al. revealed that
the PPARγ synthetic ligands, glitazones yield peroxynitrite from superoxide anion

and NO in astroglioma cells (Perez-Ortiz et al., 2007). To investigate whether
peroxynitrite is produced in our system, 5,10,15,20-Tetrakis(4-
sulfonatophenyl)prophyrinato iron (III), chloride (FeTPPS) was used. The
porphyrin complex catalytically isomerizes peroxynitrite to nitrate in vivo as well
as in vitro (Misko et al., 1998). As expected, FeTPPS significantly inhibited the
production of ROS/RNS which was detected by DCFDA staining (Figure 11A).
Furthermore, the H
2
O
2
scanvenger catalase powder failed to elicit any change in
the amount of ROS/RNS produced by PPARγ ligand. This strongly suggests that
the oxygen derivative induced by PPARγ ligand is peroxynitrite. It was also
reported that PPARγ ligands 15d-PGJ
2
and ciglitazone induce nitric oxide (NO)
release from endothelial cells (Calnek et al., 2003). Given the fact that
peroxynitrite produced can be a product from combining superoxide with nitric
oxide, we investigated whether NO was produced by PPARγ ligand in our system.
For this, NO spefic probe, DAF-FM dye was used. DAF-FM reacts with NO to
form a fluorescent benzotrozol that can be captured by flow cytometry
(Nakatsubo et al., 1998). The DAF staining showed significant production of NO
in MCF-7 cells upon treatment with 15d-PGJ
2
(Figure 10B). Moreover, the
inhibitor for nitric oxide synthase, L-NG-monomethyl Arginine citrate (L-
176

NMMA) was able to decrease the amount of NO induced by 15d-PGJ
2

(Figure
11B).
Based on the above experimental evidence, it is demonstrated that the ROS/RNS
species generated by 15d-PGJ
2
is peroxynitrite. Given that peroxynitrite can be
formed by NO and superoxide anion at a rapid rate in a nonenzymatic reaction,
the production of NO and superoxide are highly likely in our system. Although
we have convincingly showed the production of NO by 15d-PGJ
2
, the exact
source of production remains to be further elucidated. Study has shown that 15d-
PGJ
2
and ciglitazone stimulate NO release from endothelial cells, through a
transcriptional mechanism (Calnek et al., 2003). However, this mechanism
induced NO release only after 24h of treatment with 15d-PGJ
2
, and may not be
responsible for the rapid and immediate production of NO seen in our system. A
direct activation of cytosolic or mitochondrial NO synthase (NOS) isoforms by
PPARγ ligand can be a plausible mechanism. The other component of
peroxynitrite, suproxide anion, on the other hand can be produced from
mitochondria. It was reported that 15d-PGJ
2
strongly induced ROS/RNS
production by inhibiting the mitochondrial complex I activity in MCF-7 cells, and
the generation of oxidative stress by 15d-PGJ
2
was abolished by complex I

inhibitor, rotenone (Martinez et al., 2005). It was also demonstrated that
ciglitazone resulted in mitochondrial depolarization by opening the mitochondrial
permeability transition pore, thus enhancing ROS/RNS levels through interfering
with the electron transport chain (Masubuchi et al., 2006; Perez-Ortiz et al.,
2007). From these published results, we postulate the source of superoxide anion
177

to be mitochondria. However, further investigation is needed to confirm the
origins of nitric oxide and superoxide in breast cancer cells treated with PPARγ
ligands.
4.1.4 The mechanism of ROS/RNS-mediated repression of NHE1 gene.
It is not novel that intracellular ROS/RNS can regulate transcription of various
target genes. The human cytochrome P450 1A1 (CYP1A1) was shown to be
reduced at the mRNA level by oxidative stress and glutathione depletion in
human HepG2 or rat H4 hepatoma cells (Morel and Barouki, 1998). Our group
has previously demonstrated that NHE1 is a redox-regulated gene, as exposure of
cells to H
2
O
2
inhibited the NHE1 promtoer activity as well as gene expression
(Akram et al., 2006). Recently, the study has been furthered to demonstrate that
the sustained repression of NHE1 by H
2
O
2
is dependent on iron as well as
peroxynitrite. We have previously shown that PPARγ ligands induce production
of peroxynitrite and concurrently repress NHE1 gene expression, hence we
investigated whether the generation of peroxynitrite contributes to the inhibitory

effect of PPARγ ligand on NHE1 expression. To this end, N-acetyl cysteine
(NAC), a general antioxidant, was added to MCF-7 cells before exposure to 15d-
PGJ
2
. NAC functions by reducing the overall oxidation state in the system partly
through formation of glutathione. DCFDA staining revealed that the production of
ROS/RNS was completely blocked in the presence of NAC (Figure 11A). This
result confirms the previous report that treatement with NAC prevents the
production of ROS/RNS in SH-SY5Y cells and human hepatic fibroblasts (Li et
al., 2001). After confirming the effectiveness of NAC in scavenging the
178

ROS/RNS produced by 15d-PGJ
2
in MCF-7 cells, we then assessed its functional
consequence on NHE1 expression at both mRNA and protein level. Surprisingly,
large disparity exists in the extents of resue by NAC between mRNA and protein
expression. While NAC was able to induce an almost complete reversal of NHE1
protein down-regulation by 15d-PGJ
2
, its presence only rescued 20% of the
reduction in NHE1 mRNA by the same PPARγ ligand (Figure 11B, C). This
result suggests that other than regulating the factor controlling NHE1
transcription, NAC has altered the rate of NHE1 protein turnover. Indeed, NAC
was shown to inhibit 26s of proteasome activity and increased the protein level of
those molecules dependent on the ubiquitin/proteasome system for their
degradation in EVC 304 bladder carcinoma cells (Pajonk et al., 2002). Hence, the
observed rescue at protein level can be due to the slowed NHE1 protein
degradation rather than the desuppressed NHE1 transciption. This was confirmed
further by using FeTPPS and L-NMMA, which do not affect the proteosomal

degradation pathway, hence these two drugs only partially rescued the protein
expression in NHE1 (Figure 11D, E). Notably, the rescue of NHE1 expresion by
ROS/RNS scavengers was more prominent at 5µM than 3µM of 15d-PGJ
2
,
suggesting a more important role of ROS/RNS in NHE1 regulation at higher
concentration of PPARγ ligand. The involvement of ROS/RNS in PPARγ ligand-
mediated down-regualtion of NHE1 was further explored by transfecting MCF-7
cells with constructs containing full length NHE1 promtoer or NHE1 promoter
containing only the ROS/RNS response region, and the promoter activity were
assessed. Both constructs contained ROS/RNS response region previously
179

identified by our group (Kumar et al., 2007). At 3µM of 15d-PGJ
2
, construct
bearing only ROS/RNS response region did not show any decrease in promoter
activity, suggesting that the drop in NHE1 transcriptional activity is largely
PPARγ dependent at this concentration. On the other hand, 20% reduction in
NHE1 promoter activity in the construct with only ROS/RNS response region was
observed at 5µM of 15d-PGJ
2
, this reduction was contributed by ROS/RNS
generated by the PPARγ ligand (Figure 12). This agrees with the previous finding
that the ROS scavenger NAC could only rescue 20% of the reduction in NHE1
mRNA brought about by 5µM of 15d-PGJ
2
(Figure 11C). Taken together, we
show that 15d-PGJ
2

-mediated NHE1 repression is largely dependent on PPAR
receptor and minimally through ROS/RNS.
4.2 ANTI-CANCER EFFECTS OF PPARγ LIGANDS
As reviewed in introduction, mounting published results have demonstrated the
anti-tumor effets of PPARγ agonists. In our study, we explored the PPARγ
dependent and PPARγ independent anti-cancer properties of PPARγ liagnds in
breast cancer cells, and established NHE1 as the therapeutic target of PPARγ-
mediated anti-malignancy treatment.
4.2.1 PPARγ-dependent anti-cancer effects of PPARγ agonists.
PPARγ is traditionally regarded as a key regulator of adipogenesis. Its regulatotry
roles include adipogensis, glucose homeostasis and lipid metabolism (Tontonoz et
al., 1994; Lemberger et al., 1996). After the discovery of its synthetic ligands, the
TZDs, PPARγ has been targeted as a nuclear receptor involved in metatablism for
180

treating diabetes and obesity. More recently, it was found to be involved in
various cellular processes, ranging from cell cycle control, carcinogenesis,
inflammation and atherosclerosis (Rosen and Spiegelman, 2001). PPARγ ligands
were shown to be able to induce cell cycle arrest and elicit terminal differentiation
in liposarcoma cells (Tontonoz et al., 1997). Such findings suggested PPARγ-
mediated cellular differentiation and cell cycle arrest. Furthermore, increasing
clinical evidence has demonstrated that breast cancer cells express higher level of
PPARγ as compared to normal breast epithelial cells (Elstner et al., 1998;
Lapillonne et al., 2003). This finding has sparked a new flurry of research to use
PPARγ ligands for differentiation-based cell cycle arrest and proliferation
inhibiton in cancer cells of different tissue origins (Brockman et al., 1998; Elstner
et al., 1998; Kubota et al., 1998; Mueller et al., 1998).
Despite the intensive effort to establish the link between PPARγ expression and
tumor development, the exact role of PPARγ in cancer cells remains unclear. The
first evidence to demonstrate the relevance of PPARγ in breast cancer was a study

by Mueller and his colleagues. They found that overexpression of PPARγ in
human breast cancer cell lines lead to inhibition of cell proliferation, extensive
lipid accumulation and reduction in proliferation and cologenic capacity (Mueller
et al., 1998). One possible mechanism of PPARγ ligand-mediated inhibiton of
breast cancer proliferation is by repressing the expression of a G1-to S-phase
transition regulator, cyclin D1(Wang et al., 2001). Since then, the anti-tumor
effects of PPARγ agonists have been reported in almost all tumor cells on
different tissue origins. However, a parallel line of findings suggested the tumor
181

promoting effects of PPARγ activation and refuted the use of PPARγ ligands as
propective anti-cancer therapeutic agents (Lucarelli et al., 2002; Ishihara et al.,
2004; Saez et al., 2004). For instance, physiological concentrations of TZD and
15d-PGJ
2
were demonstrated to inhibit serum-deprivation induced apoptosis in
epithelial kidney cell line LLC-PK1 (Haraguchi et al., 2002). These findings
contradict the role of PPARγ as a tumor suppressor. Hence, the biological
significance of PPARγ receptor in human carcinoma remains largely
undetermined.
To verify the effect of PPARγ ligand on cell survival in our breast cancer models,
we treated MCF-7 and MDA-MB-231 cells with various doses of 15d-PGJ
2
or
ciglitazone, and observed for any morphogical changes induced by such PPARγ
ligands. After 24h treatment with each ligand, both cells were observed to
decrease in size, round up and detach from culture plates (Figure 13A, B).
Especially at higher dose, majority of the cells ended up floating in medium at the
end of 24h treatment. Cell viability assay using crystal violet showed a
concentration-dependent reduction in the number of viable cells after 24h

exposure to 15d-PGJ
2
in both MDA-MB-231 and MCF-7 cells (Figure 14C). As
cell viability assay only reflects the short-term cytotoxic effects of the drugs
tested, we also performed colony forming assay to assess the long-term effect of
PPARγ ligands on colonogenic capacity of MCF-7 cells. Consistent with result
from cell viability assay, both 15d-PGJ
2
and ciglitazone drastically attenuated the
cells’ ability to form colonies after 16h treatment with the drugs. Other breast
cancer cell lines, MDA-MB-231 and T47D, which express different basal levels
182

of PPARγ receptor, were also tested for their colony-forming capacity after
exposure to 15d-PGJ
2
. While 15d-PGJ
2
induced similar concentration-dependent
loss in cologenic ability in all cell lines, the extent of inhibition varied
significantly from cell line to cell line (Figure 14). The diverse instrinsic PPARγ
level could account for the observed different response to PPARγ ligand in these
cell lines. Among the three spontaneously occurring human breast
adenocarcinomas tested, MDA-MB-231 expresses the highest level of PPARγ and
is a poorly-differentiated cell while T47D is a well-differentiated breast cancer
cell line with the lowest amount of PPARγ receptor. Besides, the cells also have
different estrogen receptor (ER) status. MCF-7 and T47D are ER positive cell
lines whereas MDA-MB-231 does not express estrogen receptor. Interestingly,
MDA-MB-231 cells with the highest PPARγ level showed the greatest sensitivity
to 15d-PGJ

2
, while T47D cells expressing lowest level of PPARγ was the most
resistant to the drug in colony forming assay (Figure 14). This correlative
evidence triggered us to postulate that PPARγ ligand induced anti-tumor effect
may be PPARγ dependent. To further confirm the hypothesis, pharmacological
PPARγ antagonist, GW9662 was used to check its ability to block PPARγ ligand-
mediated loss in colony-forming capacity of MCF-7 cells. It was observed that
pre-incubation with PPARγ antagonist significantly rescued the MCF-7 cells from
15d-PGJ
2
-induced inhibiton in colony forming ability (Figure 15), providing
further evidence that the anti-cancer effects of PPARγ ligands on breast cancer
cells is mediated through PPARγ receptor.
4.2.2 PPARγ-independent anti-cancer effects of PPARγ agonists.
183

Besides their ability to induce PPARγ receptor-dependent cell death, PPARγ
ligands have been reported to elicit PPARγ-independent antitumorigenic effects.
For instance, troglitzone was shown to activate MAPK pathway and induce early
growth response-1 gene (EFR-1), that is linked to apoptosis. The authors
concluded that this antiproliferative mechanism of troglitzone is unique and
independent of PPARγ receptor since other PPARγ ligands fail to induce EGR-1
(Baek et al., 2003). Beside, troglitazone was also reported to induce cell arrest and
cell death in leukemic cells by down-regulating c-myc, c-myb and cyclin D2 in a
PPARγ-independent manner, because these genes lack PPRE in their promoter
regions (Laurora et al., 2003). A more direct evidentce to demonstrate PPARγ-
independent anti-tumorigenic effects was by Palakurthi et al. They showed that
troglitazone equally suppressed turmor growth in PPARγ
+/+
amd PPARγ

-/-
mouse
(Palakurthi et al., 2001).
Based on the evidence from literature, it seems that different PPARγ ligand
induces cell death in its unique PPARγ-independent mechanism that is not related
to its ability to activate PPARγ. However, a more recent study by Perez-Ortiz et al.
showed that the cytotoxic effect of various PPARγ synthetic ligands is mediated
by an increase in mitochondrial-generated ROS/RNS (Perez-Ortiz et al., 2007).
Besides, 15d-PGJ
2
and other cyclopentenone prostaglandins have been reported to
induce ROS/RNS in diverse cell types, which could mediate their effects on cell
death (Li et al., 2001; Bureau et al., 2002; Lennon et al., 2002). This promopted
us to first investigate the prescence of RNS/ROS in PPARγ-ligand treated breast
cancer cells, and its relevance in PPARγ ligand-induced cytotoxicity. As indicated
184

previously, we indeed found potent generation of intracellular ROS/RNS in cells
exposed to ciglitazone and 15d-PGJ
2
(Figure 9A). Furthermore, we showed that
FeTPPS and NAC were able to completely block the ROS/RNS species induced
by 15d-PGJ
2
. NAC was found to inhibit 15d-PGJ
2
-induced cell death in human
neuroblastoma SH-SY5Y cells and human hepatic fibroblasts (Li et al., 2001).
However, other than its ROS/RNS scavenging effect, NAC was reported to have
proteosomal inhibitory effect (Pajonk et al., 2002), hence the observed recue by

NAC might not be entirely resulted from removal of ROS/RNS. We previously
showed that other than NAC, FeTPPS, a peroxynitrite decomposing catalyst was
able to completely block the ROS/RNS production in MCF-7 cells. Hence, we
investigated by inhibiting the ROS/RNS generated by 15d-PGJ
2
, whether FeTPPS
could rescue loss of cell viability as well as colony-forming ability in MCF-7 cells.
In cell viability assay, it was observed that FeTPPS partially reversed the loss in
cell viability induced by 5µM and 10µM of 15d-PGJ
2
(Figure 16A), suggesting
that ROS/RNS contributes more to cell viability loss at higher concentration than
lower concentration of 15d-PGJ
2
. In long-term colony forming assay, FeTPPS
induced an incomplete rescue from 15d-PGJ
2
-mediated loss in colonogenic
capacity of MCF-7 cells (Figure 16B), again providing evidence that the anti-
cancer effect of 15d-PGJ
2
is partially mediated through ROS/RNS. Overall, these
results together with literature evidence demonstrate that PPARγ ligands activate
multiple pathways to exert its PPARγ-dependent and independent cytotoxic
effects in breast cancer cells. Moreover, the production of ROS/RNS by PPARγ
ligand may not be an entirely PPARγ-independent event, as recent reports have
185

shown the up-regulation of human catalase gene by PPARγ ligands in human
primary adipocytes (Okuno et al., 2010). In addition, our lab recently identified

another redox gene MnSOD to be transcriptionally down-regulated by activated
PPARγ in human breast cancer cells, further supporting that the production of
ROS/RNS by PPARγ ligand may be PPARγ-dependent.
The mechanism of peroxynitrite-induced cell death has not been fully elucidated.
However, studies have reported several pathways that can explain its cytotoxic
effect. It was first demonstrated that peroxynitrite induces caspase-independent
apotosis by causing mitochondrial dysfunction via a calcium-dependent process
(Whiteman et al., 2004). Later on, Shacka et al. showed that inactivation of the
phosphatidyl-inositol 3-kinase/ Protein Kinase B (PI3K/Akt) signaling pathway
plays a more important role compared to the mitochrondrial function in
peroxynitrite-induced cell death (Shacka et al., 2006). It was further demonstrated
that peroxynitrite induces apoptosis in endothelial cells through ER stress
(Dickhout et al., 2005). The type of cell death induced by peroxynitrite is a
mixture of necrosis and apoptosis depending on the duration and concentration of
peroxynitite exposure (Bonfoco et al., 1995; Leist et al., 1997; Leist et al., 1997).
Of all the pathways mentioned above, inactivation of PI3K/Akt seems to be the
most plausible signal cascade responsible for cell death caused by PPARγ ligand-
indued peroxynitrie, because transcriptional up-regulation of PTEN, an inhibitor
of PI3K/Akt pathway, was reported to be induced by PPARγ ligand rosiglitazone
(Patel et al., 2001). However, to ascertain the role of peroxynitrite in PPARγ
ligand-mediated cell death in breast cancer cells, and to delink the ROS/RNS
186

effect from pure PPARγ-mediated transcriptional effect in such cell death, further
studies involving silencing PPARγ and using more specific ROS/RNS scavengers
need to be performed.
4.2.3 Repression of NHE1 is involved in anti-tumor effect of PPARγ ligand.
In many cell types, excess intracellular acidification is prevented by NHE1, which
functions as a H
+

pump to remove excess protons from cells (Orlowski and
Grinstein, 2004). Studies have shown that acquiring alkaline pHi is a fundamental
step in transformed cells (Reshkin et al., 2000). The slight elevated pHi in tumor
cells is shown to favour DNA synthesis and hence promotes tumor growth
(Pouyssegur et al., 1985; Lee et al., 2003). On the other hand, celluar acidosis
induces p53-dependent induction of apoptosis in cancer cell lines (Williams et al.,
1999; Gatenby and Gawlinski, 2003). Besides, most tumor cells rely heavily on
glycolysis which results in excessive edogenous acid production. Hence, NHE1,
as the main acid extrusion mechanism, is essential for maintaining a physiological
pH of these cells (Kallinowski et al., 1988; Vaupel et al., 1989; Patel et al., 2001).
Other than regulating the pHi of intracellular milieu, NHE1 was reported to
promote invasion though MT1-MMP and p38 MAPK pathway in metastatic
MDA-MB-231 breast cancer cells (Cardone et al., 2005). Based on the
experimental evidence, a therapeutic approach targeting the acid/base balance in
tumour cells through interfering NHE1 function has been recognized (Neri and
Supuran, 2011; Webb et al., 2011). For this, two strategies have been proposed:
direct pharmacological inhibitors of NHE1 activity or controlling the gene/protein
expression of NHE1.
187

In this study, we have demonstrated that activation of the nuclear receptor PPARγ
down-regualtes NHE1 transcription and thus its protein level. The reduction in
NHE1 expression by PPARγ ligands was shown in breast cancer cell lines.
Incidently, primary tissue derived from breast cancer patients exposed to synthetic
PPARγ ligands for the treatment of diabetes also showed low NHE1 level (Kumar
et al., 2009). Hence, the repressive effect of activated PPARγ on NHE1
expression presents a therapeutic opportunity for treatment of tumor cells
expressing high level of PPARγ. Previous studies from our laboratory showed that
decreased NHE1 level sensitized cancer cells to apoptotic triggers (Akram et al.,
2006). Similarly, decresed NHE1 expression using antisense technology leads to

cell cycle arrest and apotosis in human small cell lung cancer cells (Yan et al.,
2010). To establish the role of NHE1 in breast cancer model, we silenced the
NHE1 using specific siRNA and examined the colony-forming ability of the
transfected MDA-MB-231 cells. As expected, loss of NHE1 significantly reduced
the cells’ ability to form colonies (supplementary Figure 1A of published paper)
(Kumar et al., 2009).
We have thus far shown that PPARγ ligands induce PPAR-mediated anti-cancer
effects in breast cancer cells and PPARγ activation is important for its inhibitory
effect on NHE1 expression. Given that silencing NHE1 renders breast caner cells
incompetent in forming colonies, we then asked whether PPARγ-induced down-
regulation of NHE1 expression plays a role in the cytotoxic effects of PPARγ
ligands. By overexpression of NHE1, we showed that MDA-MB-231 cells
transfected with plasmid encoding functional NHE1 were able to resist the
188

morphologic change as well as loss in cologenic capacity induced by PPARγ
ligand (Figure 2A, B of published paper) (Kumar et al., 2009). This result clearly
demonstrates the involvement of repressed NHE1 in PPARγ-mediated anti-tumor
effects in human breast cancer cells.
Although reports have shown that PPARγ-mediated anti-cancer effects is
dependent on its transciptiontal activity, specific target genes responsible for the
anti-proliferative activity of PPARγ ligands remains unclear. One such target
includes PTEN, which is shown to be up-regulated by rosiglitazone, and its
inhibition on PI3-kinase pathway is thought to be linked to the growth inhibitory
and anti-inflammatory effects of rosiglitazone (Patel et al., 2001). More recently,
transcriptional up-regulation of RhoB by activated PPARγ was implicated in anti-
proliferative effect of PPARγ agonist, RS5444 in anaplastic thyroid carcinoma
(Marlow et al., 2009). Furthermore, PPARγ dependent up-regulation of HIF1α
and BNIP3 were reported to be involved in ligand-induced autophagy in human
brast cancer cell lines (Zhou et al., 2009). Of all the PPARγ-target genes

mentioned above, NHE1 is the only house-keeping gene that is ubiquitously
expressed in all cell types, hence serves as a better therapeutic target for PPARγ-
mediated cancer therapy.
Other than inducing cell death directly, PPARγ ligand was shown to sensitize
cancer cells to apoptotic triggers. For example, synergistic anti-proliferative effect
between rosiglitazone and platinum-based drugs in different cancer cells and
chemically induced tumor models was reported (Girnun et al., 2007; Girnun et al.,
2008). Coincidently, we have also shown previously that decresing NHE1
189

expression is a way to sensitize tumor cells to anti-cancer drugs (Akram et al.,
2006). Hence we surmise that the down-regulation of NHE1 expresssion by
PPARγ ligand may be responsible for the enhanced anti-tumor effects by
platinum-based drugs.
In conclusion, we have shown that PPARγ activation leads to transcriptional
repression of NHE1, which in turn induces loss in cell viability and colonogeic
ability. Together with the previous published data on sensitizing cancer cells to
anti-tumor drugs by reducing NHE1 expression, we propose that, use of PPARγ
ligands alone or in combination with chemotherapeutic drugs for breast cancer
treatment could be a new prominisng therapeutic strategy.
4.3 MECHANISM OF HOW ERα INHIBITS PPARγ-MEDIATED DOWN-
REGULATION OF NHE1
Various reports have shown the inhibitory effects of 17β-estradiol on PPARγ-
mediated metabolic effects as well as anti-cancer effects (Wang and Kilgore, 2002;
Bonofiglio et al., 2005; Jeong and Yoon, 2011). Having established NHE1 gene
as a bona fide PPARγ target gene and an important therapeutic target, we next
examined whether ERα interferes with PPARγ-dependent down-regulation of
NHE1 gene expression.
4.3.1 ERα negatively interferes with PPARγ-mediated down-regulation of
NHE1 gene expression.

The first line of evidence to suggest the possible interference of ERα with PPARγ
on 15d-PGJ
2
-induced NHE1 repression came from the data obtained from
190

charcoal-stripped (CS) serum condition and regular serum condition. It was
observed that PPARγ ligand induced a more potent down-regualtion in cells kept
in media supplemented with CS serum than that in media with regular serum
(Figure 17A). This suggests the presence of components in regular serum that can
possibly interfere with PPARγ-mediated down-regualtion of NHE1. Intrigued by
the observation and previous report that regular serum contains higher level of
17β-estradiol, we tested the basal ERα activation in regular serum condition.
Using luciferse reporter assay and Noshift Transcritpion Factor assay, it was
shown that the basal level of transcriptional activity and the binding to classical
ERE of ERα are considerably higher in regular serum condition compared to CS
serum condition (Figure 17B, C).
To further confirm that the high estrogen content present in regular serum is
involved in the inhibitory effect on PPARγ ligand-induced NHE1 down-
regulation, we re-introduced E2 to cells kept in CS serum condition and assessed
whether the presence of E2 could rescue the reduction in NHE1 expression
brought about by different PPARγ ligands. In all the PPARγ ligands tested,
namely 15d-PGJ
2
and ciglitazone, E2 was able to block their repressive effect on
NHE1 protein (Figure 18A, B). The rescue was also observed at transcriptional
level (Figure 18C), confirming E2’s ability to negatively interfere with PPARγ-
mediated down-regualtion of NHE1 expression.
Having demonstrated that E2 inhibits PPARγ ligand-induced reduction in NHE1
expression, we next investigated whether the effect of E2 is mediated through its

receptor, ERα. Like most of the natural ligands, E2 is reported to exert both ERα–
191

dependent and independent effects in cancer cells (Yager and Davidson, 2006;
Santen et al., 2009). To confirm that the inhibitory effect of E2 on NHE1
repression by PPARγ ligands is mediated through ERα receptor, we used ER
negative breast caner cell line MDA-MB-231 and tested the effect of E2 on its
NHE1 expression. Due to the absence of ERα in these cells, E2 failed to elicit the
inhibitory effect on PPARγ-mediated NHE1 down-regualtion in ER negative
MDA-MB-231 cells (Figure 19A). However, re-expression of ERα by stable
transfection in the same cell line reinstated the effect of E2 on PPARγ-mediated
reduction in NHE1 (Figure 19B). Similar trend was also observed at
transcriptional level (Figure 19C). We next went on to examine whether a reverse
effect would be observed in ER positive MCF-7 cells. By silencing ERα in MCF-
7 cells, we demonstrated that the inhibitory effect of E2 on PPARγ-mediated
NHE1 down-regulation is abrogated in the the absence of ERα receptor both at
protein and mRNA level (Figure 19A, B).
Previously, we provided correlative evidence of masked down-regulation in
NHE1 expression and higher ERα activity in regular serum condition (Figure 17).
To ascertain that the active ERα in regular serum condition is responsible for
blocking the repressive effect of PPARγ ligand on NHE1 expression, we depleted
ERα by ERα siRNA or pharmacological degradation using high doses of
fulvestrant, and observed the extent of NHE1 down-regulation by PPARγ ligand.
In both case, removal of ERα restored the ability of 15d-PGJ
2
to repress the
expression of NHE1 both at protein and mRNA level (Figure 21). In regular
serum condition, re-expression of ERα in ER negative MDA-MB-231 cells also
192

elevated the basal level of NHE1 mRNA, suggesting a constitutive repression of
NHE1 by PPARγ in ER negative MDA-MB-231 cells and its alleviation upon
active ERα interference (Figure 22A). In fact, ER positive MDA-MB-231
responds similarly to MCF-7 cells, showing enhanced repression of NHE1 mRNA
and protein by PPARγ ligand in the presence of ERα antagonist (Figure 22B, C).
Taken together, we rigorously demonstrated that active ERα negatively interferes
with PPARγ-mediated down-regulation of NHE1 expression in breast cancer cells.
This finding corroborated well with previous reports stating that ERα blocks
general transcriptional activity of PPARγ, and inhibits up-regulation of PTEN by
activated PPARγ (Wang and Kilgore, 2002; Bonofiglio et al., 2005). In our case,
we demonstrate that ERα is not only effective in blocking the transcriptional up-
regulation of PPARγ target genes, it is also capable of inhibiting transcriptional
repression of NHE1 gene.
4.3.2 Unravelling the mechanism of how ERα inhibits PPARγ-mediated
down-regulation of NHE1 gene expression.
Previous report has shown that the transcriptional activity of ERα is necessary for
its negative interfering effect with PPARγ-mediated transcription (Wang and
Kilgore, 2002). To testify this finding, we used two ERα antagonists fulvestrant
and raloxifene. Fulvestrant was shown to bind to ERα, redering them incapable of
dimerisation and transcriptioanlly inactive by disabling both AF1 and AF2
domains of ERα (Fawell et al., 1990). Raloxifene acts as an estrogen antagonist in
both breast and uterus (Mitlak and Cohen, 1999) by blocking the assessbility of
193

AF2 domain in ERα receptor (Thiebaud and Secrest, 2001). The antagonistic
properties of the drugs were confirmed first using ERE luciferase reporter assay
and Western blot of ER-response genes. Both drugs were capable of reducing
expression of ER reponse gene, progresterone receptor and c-Myc, as well as
suppressing the ERE-mediated transcriptional activity (Figure 23A, B). By
inhibiting the transcriptional activity of ERα using either antagonist, we observed

abrogated inhibition by ERα on PPARγ-regulated NHE1 protein and mRNA
expression (Figure 23C, D).
To provide more direct evidence on the DNA-binding ability of ERα in its
suppressive effect on PPARγ-mediated NHE1 down-regulation, we overexpressed
a mutant form of ERα, which is defective in DNA-binding. We first ascertained
its ability to repress ERE lucifrease reporter activity, and futher confirmed its
capability to suppress ER-response gene expression in MCF-7 cells (Figure 24A,
B). It was observed that the mutatnt form of ERα restored PPARγ ligand-induced
NHE1 reduction, suggesting that the DNA-binding property of ERα is necessary
for its inhibitory effect on PPARγ-mediated down-regualtion of NHE1 protein
(Figure 24C). This finding also confirms the previous report that exogenously
transfected ERα lacking in DNA-binding domain was unable to inhibit PPRE-
mediated reporter activity in MDA-MB-231 cells (Wang and Kilgore, 2002).
Together, these data implicate the DNA-binding domain of ERα is essential in its
inhibitory effect on PPARγ-mediated down-regulation of NHE1. It was
hyptotheiszed that ERα and PPARγ compete sterically for DNA binding sites in
close proximity, and hence blocking PPARγ from assessing PPRE on NHE1