The role of Smad7 and TRAF6 in Prostate Cancer Cell Invasion, Migration and Survival

(1)

(2)

(3)

To my family

(4)

(5)

List of Papers

This thesis is based on the following papers, which are referred to in the text by their Roman numerals.

I TRAF6 ubiquitinates TGFβ type I receptor to promote its cleavage and nuclear translocation in cancer. Mu Y*, Sundar R*, Thakur N*, Ekman M*, Gudey SK, Yakymovych M, Her- mansson A, Dimitriou H, Bengoechea-Alonso MT, Ericsson J, Heldin CH, Landström M. Nature communication 2011.

II APC and Smad7 link the TGFβ type I receptors to the microtubule system to promote migration. Maria Ekman*, Mu Yabing*, So Young Lee*, Sofia Edlund, Noopur Thakur, Hoanh Tran, Jiang Qian, Joanna Groeden, Carl-Henrik Heldin and Marene Landström. (manuscript)

III Smad7 and APC are required for EGF-induced cell migra- tion in human prostate carcinoma cells. Maria Ekman, Carl- Henrik Heldin, Marené Landström. (manuscript)

IV TGFβ1-induced activation of ATM and p53 mediates apop- tosis in a Smad7-dependent manner. Zhang S*, Ekman M*, Thakur N*, Bu S, Davoodpour P, Grimsby S, Tagami S, Heldin CH, Landström M. Cell cycle, 2006.

* Indicates that these authors contributed equally to the article Reprints were made with permission from the respective publishers.

(6)

(7)

Abbreviations

ADAM a disintegrin and metalloproteinase

APC adenomatous polyposis coli

Arp actin related protein

ARTS apoptosis-related protein in TGFβ signaling pathway

ATM ataxia telangiectasia mutated

BCL B-cell lymphoma

Co-Smad common mediator Smad

EB1 end-binding protein 1

EGF epidermal growth factor

ERK extracellular signal-regulated kinase FADD Fas-associated death domain protein

GSK-3β glycogene synthase kinase 3β

GS-region glycine and serine rich region

GTPase guanosine triphosphatase

HSP-27 heat shock protein 27

ICD intracellular domain

I-Smad inhibitory Smad

IQGAP IQ motif containing GTPase activating protein homologue

JNK c-Jun-N-terminal kinase

MAPK mitogen-activated protein kinase MAPKAPK 2/3 MAPK-activated protein kinase 2/3

MEK MAP kinase-ERK kinase

MH mad homology

MKK mitogen-activated protein kinase kinase

MMP matrix metalloproteinase

MTOC microtubule organizing centre

PAR6 partitioning defective-6

PB Phoex and Bern 1p

PI3K phosphatidylinositol-3 kinase

PIP3 phosphatidylinotiol (3,4,5) tris phosphate β-PIX β-PAK-interacting exchange factor

PKB/Akt protein kinase B

PKC protein kinase C

PTB phosphotyrosine binding

(10)

RING really interesting gene

R-Smad receptor associated Smad

SARA smad anchor for receptor activation

SH2 Src homology 2

TAB1 TAK1-associated binding protein 1

TACE TNFα converting enzyme

TAK1 TGFβ activated kinase 1

TCF/LEF T-cell-dependent factor/ lymphoid enhancer- binding factor

TGFβ transforming growth factor β

TβRI transforming growth factor β receptor I TIEG1 TGFβ-induced early response gene 1

TNF tumor necrosis factors

TRAF6 tumor necrosis factor receptor-associated factor 6

Wnt Wingless/Int

WASP Wilskott-Aldrich syndrome protein

WAVE WASP-family verprolin-homologous protein XIAP X-chromosome-linked inhibitor of apoptosis

protein

(11)

Introduction

Cell signaling is important during embryonic development, but also in the adult for keeping tissue homeostasis and for wound healing. Cells communi- cate with surrounding cells by secretion of cytokines and growth factors, which bind to receptors at the cell surface. This starts intracellular signaling cascades that can regulate for instance cell migration, proliferation and apoptosis. These signaling pathways are carefully regulated, and the cell response is highly dependent of the strength of the signal, as well as of the other signals the cell receive at the same time. During the progression from a normal cell to a cancerogenic cells there are accumulations of mutations. Mutations of proteins required for guarding the DNA replication machinery, accelerate the mutation rate and give rise to genetic instability and tumor progression.

Mutations or changes of expression of proteins in cell signaling pathways can also promote cancer progression. Cancer cells usually have alterations in a number of signaling pathways; changes in TGFβ, Wnt and EGF signaling pathways all have been linked to cancer (Bierie and Moses 2006). The ac- quired mutations make the cancer cells less sensitive to signals which are aimed to induce cell death (apoptosis) or growth arrest. At the same time, the cancer cells become more sensitive to signals which induce cell migration, invasion and proliferation. Cancer cells can acquire the ability to produce growth factors themself or signal to normal cells in the tumor stroma, which can supply the tumor with various growth factors (Hanahan and Weinberg 2011). This makes it possible for the cells to divide more or less uncon- trolled and spread to form metastases at other places in the body. The work in this thesis is mainly performed using a human prostate cancer cell line (PC3U). Upregulation of TGFβ1 in prostate cancer has been found to promote angiogenesis and metastasis and is correlated with a poor patient prognosis (Bierie and Moses 2006). We have therefore studied the molecular mechanisms for TGFβ-induced cellular responses in PC3U cells with a par- ticular focus on the adaptor proteins Smad7 and TRAF6.

(12)

TGFβ signaling

TGFβ in disease

Defect TGFβ signaling has been linked to diseases, such as fibrosis, auto- immune disease and cancer (ten Dijke and Hill 2004). During early tumor development TGFβ1, has been found to act as a tumor suppressor, by induction of growth arrest and apoptosis (Figure 1). The tumor suppressor effect is, however, lost during cancer progression. Many cancers have inactivating or attenuating mutations in the TβRI or TβRII genes or epigenetic silencing of TβRI or TβRII genes, which make cells escape from the growth inhibitory effect of TGFβ. Mutations in Smad genes, or altered Smad expression, as well as increased levels of Smad7 can also make the cells less sensitive to the tumor suppressor activities induced by TGFβ (Liu, Xu et al. 2009).

Figure 1. TGFβ as a tumor suppressor and tumor promoter. Adapted from Dumont and Arteaga 2003.

At later stages of tumor development, TGFβ is actively secreted by tumor cells or stromal cells. TGFβ then enhances tumor progression by inducing epithelial to mesenchymal transition (EMT), which increases the cells migration and invasiveness and ability to form metastases and by affecting the local environment to induce angiogenesis, immune suppression and modifi- cation of the extracellular matrix (ECM) (Kaminska, Wesolowska et al.

2005; Bierie and Moses 2006; Heldin, Landstrom et al. 2009; Bierie and Moses 2010).

(13)

The TGFβ family of ligands and receptors

TGFβ were first discovered in beginning of 1980:s (Roberts, Lamb et al.

1980). The TGFβ family of cytokines contains 33 members in humans but is also represented in other species. The TGFβ family can be divided into two sub families the TGFβ/Activin/Nodal subfamily and the bone morphogenetic protein (BMP)/growth and differentiation factor (GDF)/Muellerian inhibiting substance (MIS) subfamily (Table 1) (Shi and Massague 2003). The diverse set of TGFβ ligands share a set of common sequence and structural features but can give rise to different cellular responses. TGFβ is important during embryonic development but is also of importance in the adult organism where it controls the immune system and stimulates angiogenesis (Heldin, Landstrom et al. 2009). TGFβ is involved in a diverse set of cellular processes, including cell proliferation, differentiation, adhesion, apotosis, migration and specification of developmental fate (Dennler, Goumans et al.

2002; Siegel and Massague 2003). The BMPs, Activins, Nodals, AMH/MIS, and GDFs subfamilies are key regulators of embryonic stem cell differentiation, body axis formation, left-right symmetry and organogenesis (Groppe, Hinck et al. 2008). Some of the TGFβ family members, such as AMH/MIS (Anti-Mullarian hormone or Mullerian inhibiting substance) and GDF8/myostatin, are only present in a few cell types or for limited time during development while others like TGFβ1 and BMP4 are widespread during development, and in the adult tissues (Massague, Blain et al. 2000).

There are different isoforms of TGFβ, i.e. TGFβ1, TGFβ2 and TGFβ3.

TGFβ is synthesized in the endoplasmatic reticulum and secreted from the cell via the Golgi apparatus and the exocytic vesicular system. After secretion, it can be stored in the extracellular matrix in a latent form that requires biochemical activation (Moustakas and Heldin 2008). The active TGFβ ligand consists of two hand-shaped monomers assembled together in the wrist region, stabilized by hydrophobic interactions and a disulfide bond (Greenwald, Groppe et al. 2003; Shi and Massague 2003; Heldin, Landstrom et al. 2009). There are 7 different TGFβ type I receptors also known as acti- vine receptor-like kinases 1 to 7 (ALK1-7), and 5 different type II receptors (ActR-IIA, ActR-IIB, BMPR-II, AMHR-II and TβR-II). All 12 receptors are single-pass transmembrane proteins with a short cysteine-rich extracellular domain, which binds the ligand, an α-helical transmembrane domain and a intracellular serine-threonine kinase domain which also contains phospho- acceptor sites and docking sites for adaptor or signaling proteins (Moustakas and Heldin 2008). The type I and type II receptors are structurally very simi- lar, but the type I receptor has a conserved glycine-serine (GS) domain im- mediately upstream of the kinase domain, which is not present in the type II receptor (Feng and Derynck 2005). Different combinations of the type I and type II receptors attract different ligands from the TGFβ family (Shi and Massague 2003; Massague and Gomis 2006).

(14)

Table 1. TGFβ family of ligands and receptors and the Smads involved in the downstream signaling. Adopted from Massague 2008.

The Smads

There are three different groups of Smad proteins: Receptor-activated Smads (R-Smads; Smad2 and 3 in the TGFβ signaling pathway and 1, 5, 8 in the BMP signaling pathway, Common-mediator Smad (Co-Smad; Smad4) and Inhibitory Smads (I-Smads; Smad6 and 7) (Figure 2). The R-Smads and Smad4 share two highly conserved globular domains, the mad homology 1 (MH1) domain, and mad homology 2 (MH2) domain. The MH2 domain of R-Smads mediates the interaction with the TβRI. The MH2 of both the R- Smads and Smad4 interacts with the phosphorylated SSXS motif in R- Smads. In addition, the MH2 domain is responsible for the interaction with transcription factors, co-activators and co-repressors. The MH1 domains of Smad3 and Smad4 are involved in DNA binding. Smad2 has no DNA binding capacity, due to an insert in the MH1 domain. The MH1 and MH2 domains are connected by a less conserved linker region (ten Dijke and Hill 2004; Schmierer and Hill 2007). The linker region is phosphorylated by kinases such as mitogen activated protein kinases (MAPK), glycogen synthase kinase-3β (GSK-3β) and cyclin dependent kinases (CDK), which might be a way to integrate inputs from other signaling pathways (Schmierer and Hill 2007). The inhibitory Smads contain the conserved MH2 domain, but lacks the SSXS motif present in the R-Smads. The N-terminal domains of the inhibitory Smads are divergent from the MH1 domains and the linker regions present in the R-Smads and the Co-Smads. The N-terminal of Smad6 and Smad7 shows 36% identity to each other. Smad7 is a general antagonist of the TGFβ signaling pathway, while Smad6 is considered to be specific for the BMP signaling pathway. In contrast to the other Smads, the I-Smads have the possibility to interact with DNA via the MH2 domain instead of with the N-terminal part (Yan, Liu et al. 2009).

(15)

TGFβ signaling pathways

Recent studies have shown that majority of the TβRI and TβRII exist as monomers at the cell surface in resting cells (Zhang, Jiang et al. 2009; Huang, David et al. 2011). Upon TGFβ stimulation, the extracellular parts of the type II receptors bind with high affinity to the TGFβ ligand, which enables recruitment of TβRI (ALK5). Each protomer of the TGFβ dimer can recruit one pair of TβRII:TβRI to form a hetero-tetrameric receptor complex (Groppe, Hinck et al. 2008; Massague 2008; Huang, David et al. 2011). The TβRII, which has a constantly active kinase, trans-phosphorylates the TβRI at several serine residues in the glycine-serine-rich (GS)-region. This gives rise to conformational changes of the TβRI-kinase domain and causes its activation. The activated TβRI initiates several intracellular signaling cascades, including the Smad signaling pathway and the TRAF6-TAK1 signaling pathway (Figure 3) (Derynck and Zhang 2003; Schmierer and Hill 2007;

Moustakas and Heldin 2008).

MH1

MH1 linker MH2

MH2 Co-Smads

Smad4 (common)

I-Smads

Smad7 in TGFβ and BMP signaling Smad6 in BMP signaling

R-Smads

Smad2, 3 in TGFβ signaling Smad1, 5, 8 in BMP signaling

linker MH2 - SSXS

MH1 domain DNA binding Interaction with transcription factors

MH2 domain

Receptor interaction (R-Smads, I-Smads) Smad oligomerization

Transcriptional activation Interaction with CBP/p300

Interaction with transcription factors

Interaction with Smurf Interaction with β-catenin

Figure 2. Smad structure

N-terminal domain

(16)

Smad signaling pathway

The Smad signaling pathway is the best known and the most explored of the TGFβ signaling pathways. In the basal state cytosolic Smad2 and 3 bind to Smad anchor for receptor activation (SARA). SARA presents the R-Smads to the activated TGFβ type I receptor (Siegel and Massague 2003). The activated TβRI phorylates the R-Smads on an SSXS motif in the very C- terminal part of the MH2 domain, which gives rise to a conformational change and activation of the R-Smads. The activated R-Smads form hetero- meric complexes with Co-Smads, which translocate and accumulate in the nucleus, where they interact with transcriptional co-activators, such as cyclic AMP response element-binding protein (CBP) and p300, to activate transcription of targets specific genes (Dennler, Goumans et al. 2002; Shi and Massague 2003; Siegel and Massague 2003; Moustakas and Heldin 2008).

p300/CBP function as a transcription co-factor for several nuclear proteins, both oncoproteins and tumor suppressor proteins, and form bridges to bring DNA-binding transcription factors together with each other and the basal transcription machinery. p300/CBP also acetylate histones, which neutralize the positive charge of histones and reduces the tight interaction between histones and DNA, which is necessary for a gene to be actively transcribed (Iyer, Ozdag et al. 2004). One of the early target genes of TGFβ is Smad7, which forms a negative feedback loop to prevent Smad signaling by several different mechanisms. Smad7 has also been found to cause degradation of the TβRI by reqruitment of the E3 ubiqutin ligases SmurfI and SmurfII.

Smad7 binds via the N-terminus to Smurf in the nucleus and translocates to the cytoplasm in response to TGFβ stimulation. In the nucleus Smad7 associates with the acetyltransferase p300, which acetylates Smad7 and thereby protects Smad7 from Smurf mediated ubiquitination and degradation (Gronroos, Hellman et al. 2002). The acetylation, however, is lost when Smad7 leaves the nucleus (Shi and Massague 2003). Binding of the Smad7/Smurf complex to the activated TβRI induces ubiquitination of the receptor. This leads to internalization and degradation of the receptor as well as of Smad7 through proteasomal and lysosomal pathways (Di Guglielmo, Le Roy et al. 2003; Moustakas and Heldin 2008). Smad7 also prevent activation of R-Smads and the downstream signaling by competing with the R- Smads for binding to the activated TβRI. Another mechanism for Smad7 to inhibit Smad signaling is by competing with the R-Smad/Smad4 complex for DNA binding at specific gene promotors (Moustakas and Heldin 2008; Yan, Liu et al. 2009; Yan and Chen 2011).

(17)

Figure 3. Smad signaling pathway and TAK1-TRAF6 signaling pathway. Adapted from Sorrentino, Thakur et al. 2008.

The TAK1-TRAF6 signaling pathway

TRAF6 binds constitutively to a consensus motif in the TβRI. Binding of the TGFβ ligand to the receptor complex causes oligomerization of TRAF6, which in turn initiates auto-ubiquitination and activation of TRAF6, inde- pendently of the kinase domain of the TβRI kinase domain (Sorrentino, Thakur et al. 2008; Yamashita, Fatyol et al. 2008). TRAF6 is a E3 RING (really interesting new gene) domain ubiquitin ligase, which catalyze the synthesis of polyubiquitin chains linked via a lysine at position 63 (Lys63).

Polyubiquitination formed via Lys63 serves as a regulatory signal to provide a scaffold for the assembly of protein kinase complexes to mediate their activation, but can also be important for the localization of the protein; in contrast polyubiquitination via Lys48 is a marker for proteosomal degrada-

(18)

tion (Ikeda, Crosetto et al. 2010). Ubiquitin is a small regulatory protein of 8 kDa, which can be attached to other proteins. The ubiquitination process starts by the E1 enzyme, which activates the ubiquitin in an ATP-dependent manner. The ubiquitin is transferred to a reactive cysteine of the E2 conjugating enzyme, which cooperates with an E3 ligating enzyme that recognizes the substrate, where the E2 conjugating enzyme will attach the ubiquitin (Dikic, Wakatsuki et al. 2009; Nagy and Dikic 2010). There are three different families of E3 ligating enzymes, HECT (homologus to the E6-AP car- boxyl terminus), which has catalytic activity and can accept an ubiqiutin as well as transfer it to the protein target, RING and U-box E3 ligating enzymes, both seem to lack the catalytic activity and works like scaffolds for the E2 conjugating enzymes (Nagy and Dikic 2010). Mdm2 is an E2 conjugating enzyme known to interact with TRAF6. The ubiqutin can be attached as a single moiety (monoubiquitination) or be linked together by E2 or E3 enzymes to form polymeric chains (polyubiquitination) of different lengths.

One of the lysine residues within ubiquitin is linked together with the C- terminal part of another ubiquitin molecule. The structure of the ubiquitin chain is dependent of which lysine residue that is involved in the linkage.

Linkage via Lys48, for instance, give rise to a tighter and less flexible assembly of ubiquitin compared linkage via Lys63 (Figure 4). Ubiquitin signals can be recognized and processed by proteins with specialized ubiquitin binding domains, which can mediate several different responses, such degradation of the protein, receptor trafficking, DNA repair, cell-cycle progression, gene transcription or apoptosis (Ikeda, Crosetto et al. 2010).

Activated TRAF6 causes the activation of TAK1, by Lys63 linked polyubiquitination (Sorrentino, Thakur et al. 2008; Yamashita, Fatyol et al. 2008;

Landstrom 2010). TAK1 is a serine-threonine kinase that belongs to the mitogen-activated protein kinase kinase kinase family (MAPKKK). Activation of TAK1 also requires the adaptor proteins TAK1-associated binding protein-1, -2, -3 (TAB1, 2, 3) (Adhikari, Xu et al. 2007). TAK1 activates MKK3/6 or MKK4/7 by direct phosphorylation. MKK3/6 continues the signaling cascade by phosphorylation of the MAPKase p38, while MKK4/7 induces activation of JNK (Yamashita, Fatyol et al. 2008; Wagner and Nebreda 2009). Smad7 act as a scaffold protein in the TGFβ-induced activation of p38 (Edlund, Bu et al. 2003).

(19)

Figure 4. Attachment process for ubiquitination. Monoubiquitination, Lys48- and Lys63-mediated polyubiquitination. Adapted from Dikic, Wakatsuki et al. 2009.

TGFβ induce activation of TACE

TGFβ has been shown to activate TACE (TNFα converting enzyme) via the MAPK ERK in several different cell lines (Liu, Xu et al. 2009). TGFβ- induced activation of TACE causes its translocation to the cell surface, where it induces cleavage and activation of the precursor forms of the EGF receptor ligands, TGF-α, amphiregulin and heregulin. The EGFR ligands activate autocrine and paracrine signaling to increase migration, proliferation and cell survival (Wang, Xiang et al. 2008). TACE has also been shown to cleave the TβRI, but not TβRII. The cleavage of TβRI was shown to de- crease TGFβ-induced growth inhibition, due to decreased levels of the TβRI at the cell membrane and perturbed Smad signaling (Liu, Xu et al. 2009).

The metalloproteinase TACE also known as ADAM-17, is often overex- pressed in tumors and correlated with a poor disease progression. TACE is a member of the large family of ADAM metalloproteinases, which can have both adhesive and proteolytic properties and can therefore participate in both cell adhesion and cleavage of cell surface molecules. TACE is a multi- domain protein consisting of a prodomain, a metalloenzyme or catalytic domain responsible for shedding the substrate, a disintegrin-like domain, a cysteine rich domain a transmembrane domain and a short cytoplasmic tail.

TACE has rather low sequence similarity to other members of the metalloproteinase family. The pro-domain of TACE inhibits the enzymatic activity of TACE during its translation, and is cleaved of by furin, a pro-protein con- vertase, in the transgolgi network, which causes the activation of TACE (Gooz 2010). Immunohistological staining for TACE suggests that the main part of active TACE is present in the perinuclear regions and smaller

(20)

amounts at the plasma membrane (Schlondorff, Becherer et al. 2000).

TACE activates growth factors such as the EGFR ligand family, by cleavage in the stalk region of the extracellular domain in close proximity to the cell membrane. By increasing the amount of active ligands in the tumor micro- environment it contributes to inflammation, tumor growth and progression.

TACE expression has been observed in 30% of benign prostatic hyperplasias and in all investigated prostatic tumor samples and prostatic tumor cell lines (Gooz 2010). TACE has also been shown to cleave a number of transmembrane receptors, such as Notch, ErbB2 (Codony-Servat, Albanell et al.

1999), ErbB4, CD44 (Nagano, Murakami et al. 2004). After ligand binding to the Notch receptor, TACE mediates cleavage it in the extracellular stalk region. This causes formation of a substrate for the γ-secretase complex, which cleaves the receptor a second time inside of the cell membrane, to release the intracellular domain. The intracellular domain of the Notch receptor translocates to the nucleus where it activates a number of target genes (Guo, Liu et al. ; Watt, Estrach et al. 2008).

TACE can be activated by PKC isoforms (Edwards, Handsley et al. 2008;

Herrlich, Klinman et al. 2008), a family of Ser/Thr kinases that are involved in several signaling pathways. There are three PKC subfamilies: classical or conventional PKCs (α, βI, βII and γ), novel PKCs (δ, θ, ε and η) and atypical PKCs (ζ, ι (human) and λ (mouse))(Corbalan-Garcia and Gomez-Fernandez 2006). All PKC isoforms contain C-terminal kinase domains, which are closely related between the three families (Figure 5), while the regulatory domain differ more between the families. The classical and the novel PKC both have a double C1 domain and a C2 domain. Both these families are activated by diacylglycerol, but only the classical PKCs are responsive to Ca2+. The atypical PKCs lack the C2 domain and contain a single C1 domain, which is not diacylglycerol sensitive. Instead of the C2 domain the atypical PKCs contain a PB1 (Phoex and Bern 1p) domain. This domain binds other PB1 domain-containing proteins such as Par6 or MAPK5. Atypi- cal PKCs are regulated by acidic phospholipids, ceramides and protein- protein interactions (Moscat and Diaz-Meco 2000; Corbalan-Garcia and Gomez-Fernandez 2006; Moscat, Rennert et al. 2006).

(21)

Smad7 is a crosstalk mediator involved in cancer

Smad7 is not just a negative regulator of the TGFβ signaling pathway. It has also been shown to act as an adaptor protein important for activation of the p38 signaling, which is important for TGFβ-induced apoptosis (Edlund, Bu et al. 2003), as well as rearrangements of the actin cytoskeleton (Edlund, Landstrom et al. 2004). In addition to TGFβ and BMP signaling Smad7 can also be induced by EGF, IFN-γ, TNFα, IL-1, ultraviolet irradiation or TPA treatment (Lallemand, Mazars et al. 2001; Yan and Chen 2011). Overexpres- sion of Smad7 has been found to accelerate tumor progression by inhibition of the TGFβ signaling pathway, but also by upregulation of EGF-like growth factors e.g. TGFα, heparin-binding-EGF and amphiregulin. Overexpression of Smad7 in combination with an activated Ras oncogene product made epithelial cells convert from a benign phenotype to a malignant phenotype (Liu, Lee et al. 2003). On the other hand it has been shown that overexpression of Smad7 inhibits formation of metastases in a mouse model for breast cancer, by the regulation of cell-cell adhesion (Azuma, Ehata et al. 2005). It is there for possible that Smad7 can act either as a tumor promoter or a tumor suppressor, depending on the context.

Smad7 associates with components in the Wnt signaling pathway

The Wnt signaling pathway controls differentiation during embryonic development and leads to tumor formation when aberrantly regulated. In the absence of Wnt signaling, free cytoplasmic β-catenin is incorporated in a complex consisting of adenomatous polyposis coli (APC), axin and glycogen synthase-3 (GSK-3) (Figure 6) (Giles, van Es et al. 2003). GSK-3β phosphorylates β-catenin and marks it for proteasomal degradation (Moon, Bowerman et al. 2002). When the Wnt ligand binds to the frizzled receptor a Figure 5. PKC family. Adapted from Corbalan-Garcia and Gomez-Fernandez 2006.

(22)

cascade of events occurs, which involves activation of dishvelled protein and leads to inactivation of GSK-3β and dissociation of the complex (Nelson and Nusse 2004). This causes translocation and accumulation of β-catenin in the nucleus, where it acts as a cofactor and binds to transcriptional regulators T- cell-dependent factor/lymphoid enhancer binding factor (TCF/LEF). This affects the expression of Wnt target genes involved in cell fate, proliferation and apoptosis (Giles, van Es et al. 2003). Smad7 is important for TGFβ- induced inactivation of GSK-3β, which leads to an accumulation of β- catenin. In addition, the N-terminus of Smad7 interacts with β-catenin and its transcription factor TCF/LEF, to regulate gene transcription. This interaction is dependent on p38 and is important for TGFβ-induced apoptosis (Edlund, Lee et al. 2005). Smad7 also stabilizes β-catenin, by binding to axin and thereby making the β-catenin destruction complex dissociate. In breast cancer cells the stabilized β-catenin is not translocated to the nucleus but to the E-cadherin at the cell membrane to form the adherence junction complex, which is important for the cell-cell adhesion. Loss of cell-cell adhesion is a hallmark of carcinogenesis and tumor progression, which in this case can be prevented by Smad7 (Tang, Liu et al. 2008; Ikushima and Miyazono 2011).

(23)

Figure 6. I. Wnt signaling pathway in absence of signal. II. Wnt signaling pathway in presence of Wnt ligand. III. The interactions between components in the TGFβ signaling pathway and Wnt signaling pathway upon TGFβ stimulation. Adapted from Giles, van Es et al. 2003; Edlund, Lee et al. 2005.

(24)

EGF signaling pathway

The EGF receptor (EGFR) belongs to the HER/ErbB family, which consists of 4 closely related tyrosine kinase receptors, EGFR (Erb1), ErbB2, ErbB3 and ErbB4. EGFR and ErbB4 can upon ligand binding form homodimers or heterodimers, which generates intracellular signaling cascades that regulate cell proliferation, migration or differentiation. Erb2 lacks ligand binding capacity and ErbB3 has an impaired kinase domain; despite this both ErbB2 and ErbB3 have important functions by forming heterodimeric complexes with other ErbB receptors (Citri and Yarden 2006; Bublil and Yarden 2007).

Overexpression of the EGFR has been observed in many different cancers including prostate cancer (Lu and Hunter 2004). The EGFR has an extracellular ligand-binding domain, a single transmembrane domain and an intracellular tyrosine kinase containing domain (Sanderson, Dempsey et al.

2006). The majority of the EGFR are located at the cell surface, but the receptors constantly undergo recycling between the plasma membrane and the endosomal compartment. In the absence of ligand, the EGFR is slowly inter- nalized and rapidly recycled to the plasma membrane. Upon ligand binding and activation of the EGFR, the internalization process is accelerated; most of the EGFR signaling occurs within the endosomes. Ligand binding also enhances the lysosomal degradation of the receptor (Wiley 2003). There are 13 different ligands binding to the ErbB receptors, including EGF. The EGF- like factors are synthesized as transmembrane precursors, which needs to undergo ectodomain shedding (cleaved in the proximity to the cell membrane) by different metalloproteases, such as TACE, to release a mature and soluble ligand (Sanderson, Dempsey et al. 2006). The different receptors have different affinity for the ligands. EGF, transforming growth factor α (TGFα) and amphiregulin binds specifically to the EGFR, while betacellu- lin, heparin-binding growth factor (HB-EGF) and epiregulin shows dual specificity and binds both the EGFR and Erb4 (Normanno, De Luca et al.

2006). Ligand binding causes homo- or hetero-dimerisation of the receptors (Burgess, Cho et al. 2003; Normanno, De Luca et al. 2006; Bublil and Yarden 2007), which makes the intracellullar kinase domains to form a head to tail dimer (N-terminal of one kinase domain binds the C-terminal of the other kinase domains) (Figure 7). In this conformation, one of the kinase domains (the activator), cause phosporylation of several tyrosine residues in the C-terminal tail of its partner (Bublil and Yarden 2007; Jura, Zhang et al.

2011). There are six autophosphorylation sites in the C-terminal tail of the

(25)

Figure 7. After EGF has bound to EGFR, the intracellular kinase domains form a head to tail dimer, where one of the kinases activates the other. This leads to phos- phorylations at different sites in the receptor tail and initiates several intracellular signaling cascades. Adapted from Jura, Zhang et al. 2011.

EGFR, i.e. tyrosines 992, 1045, 1068, 1086, 1148 and 1173 (Yamaoka, Frey et al. 2011). Once phosphorylated, these residues serve as docking sites for proteins containing Src homology 2 (SH2) and phosphotyrosine binding (PTB) domains. The recruited proteins get activated and start intracellular signaling cascades (Normanno, De Luca et al. 2006; Yamaoka, Frey et al.

2011). Several signaling pathways downstream the EGFR has been shown important for cell migration in wound healing assays, including ERK MAPK, JNK/SAPK, p38 MAPK, PI3K, Src and PKC (Yamaoka, Frey et al.

2011). EGF has been shown to induce activation of p38, JNK and ERK via Src, independent of PI3K and PLCγ/PKC (Frey, Golovin et al. 2004). Bind- ing of different ligands to the receptors give rise to different phosphorylation patterns. Stimulation with EGF promotes phosphorylation of Tyr 845, 1068, 1086, 1173 and 1148 (Figure 7), while stimulation with for instance betacel- luline only induces phosphorylation of Tyr1068 and Tyr1073. Specific EGFR tyrosines play key roles in determine the cellular response to ligand binding. Phospho-Tyr992 and -Tyr1173 cause recruitment of phospholipase (PL) Cγ, which is important for EGF-induced cell migration. Tyr1045 phosphorylation represents a docking site for c-Cbl, which mediates ubiquitination and thereby causes degradation of the EGFR. Phosphorylation of Tyr1068 and 1086 creates a binding site for the adaptor Grb2, which forms a

(26)

complex with the nuclear exchange molecule Sos, which activates Ras where by the ERK/MAPK signaling pathway is activated. Tyr1068 and Tyr1086 also serve as binding sites for Gab1, which recruits PI3K and activates the Akt signaling pathway. Phosphorylation of Tyr1068 and Tyr1086 has been shown to be essential for EGF-induced wound healing (Boucher, Kehasse et al. 2011). Phosphorylation at Tyr1148 and Tyr1173 serves as docking sites for the adaptor protein Shc, which increases the activity of the ERK/MAPK cascade (Yamaoka, Frey et al. 2011).

(27)

Cell migration and invasion

Cell migration is crucial during the embryonic development, but is also important in the adult organsism; e.g. during tissue repair and renewal. Cell migration also drives disease progression in diseases such as cancer.

When cells are exposed to an extracellular chemoattractant signal, receptors at the cell surface transfer the signal into the cell, leading to a dramatic change of the cell morphology. In a polarized migrating cell the molecular processes and cytoskeletal structures are different in the front compared to the rear of the cell (Figure 8) (Ridley, Schwartz et al. 2003). During cell migration the cell membrane at the leading edge has to be extended forward;

in this process the actin cytoskeleton is thought to provide the driving forces.

This can be accomplished by four different actin structures: filopodia, lamellipodia, blebs and invadopodias. Filopodia consist of actin organized in par- allel bundles and are thought to be important for guidance and exploring of the surrounding. Lamellipodia are thin sheet like structures, at the leading edge of the cell that consists of a dense meshwork of actin filaments. The lamellipodia can also fold backwards to form membrane ruffles. The lamellipodia can extend long distances through the extracellular matrix in vivo and pull the cells through tissues. Membrane blebbing are mainly driving cell migration during development. The invadopodias are protrusions that possess the ability to degrade the extracellular matrix, which also facilitates invasion through tissues, especially when the cells will cross the basement membrane (Ridley 2011). Stress fibres are another kind of actin structures, stretching through the cell body, to contract the cell and pull the rear end forward. The stress fibres anchor the cell to the surface via focal adhesions.

When the cell is moving forward new focal adhesions are established in the front, while the focal adhesions in the rear end are disassembled. The microtubule system is essential for cell division, cell migration, vesicle transport and cell polarisation. The microtubules are highly polarized hollow tubes composed of α- and β-tubulin. The minus ends of the microtubules are organized in the microtubule organizing centre (MTOC) the plus ends of the microtubule are stabilized at the leading edge.

(28)

Figure 8. Cell cytoskeleton, actin (red), microtubule (green). Adapted from Etienne- Manneville 2004.

Signaling in cell polarization

In migrating cells, activation of phosphatidylinositol-3 kinase (PI3K) at the side facing the chemoattractant gradient, is important for the establishment of the leading edge. The PI3K signaling pathway is activated in response to several different growth factors and cytokines (Woodgett 2005). Both EGF and TGFβ are known to activate PI3K (Derynck and Zhang 2003; Bellezza, Bracarda et al. 2006). PI3K generates 3’ phosphatidyl-inositol (3,4,5) tris- phosphat (PIP3) in the plasma membrane in the front of the cell and control the degradation of PIP3 at the sides and the rear end of the cell (Merlot and Firtel 2003). Proteins with PIP3 binding motifs translocates to the membrane, where they interact with other proteins or are activated by PIP3.

Phosphoinositide-dependent kinase 1 (PDK1) and Akt can interact with PIP3. Akt is important for the cell polarization during chemotaxis and has several substrates; it can among other things inactivate GSK-3β by pho- shorylation (Woodgett 2005). PIP3 is also known to activate the small Rho GTPase Rac, which can induce actin polymerization. It is possible that Rac is involved in a positive feedback loop to activate PI3K at the leading edge to create a PIP3 gradient and induce cell polarization (Merlot and Firtel 2003).

(29)

Rho GTPases in actin reorganization

The Rho GTPases are critical signal transducers, important for transmitting the signal from the cell surface receptor to the cytoskeleton to induce actin reorganization. Rho, Rac and Cdc42, are all members in the Rho family of small GTPases. When the Rho GTPases are bound to GTP they are active and can transfer the signal to downstream targets, while binding of GDP makes them inactive (Ridley, Schwartz et al. 2003; Ridley 2011). The exchange of GTP to GDP to activate the Rho GTPases is catalyzed by GEFs.

Many different extracellular stimuli, such as growth factors, cytokines and cell adhesion receptors can induce the formation of the lamellipodia. The Rho GTPases are belived to act in coordination with a lot of other proteins, to activate the actin regulators needed for lamellipodia formation (Ridley 2011). Rac induces formation of the dense actin network, characteristic of lamellopodia by activation of WASP-family verprolin-homologous protein (WAVE), which in turn activates the actin-related protein (Arp) 2/3 complex (Figure 9) (Yamazaki, Kurisu et al. 2005) The Arp2/3 complex is bound to the actin filaments in the leading edge close to the cell membrane and is involved in polymerization of monomeric actin into a branched filament network. Formins extend the actin filaments assembled by Arp2/3. The formins can be activated by Cdc42, Rac and probably also other Rho GTPases (Ridley 2011). Cdc42 can also indirectly activate Rac, via Pak and the Rac guanine nucleotide exchange factor β-PIX to induce actin polymerization and lamellipodia formation at the leading edge (Cau and Hall 2005).

Figure 9. The role of Rho GTPases in actin rearrangement

(30)

Rho has predominantly been implicated in stress fibre formation and focal adhesion assembly and tail retraction, but is also highly active in the membrane ruffles in the lamellipodia (Ridley 2011). Many of the proteins that regulate the formation of the invadopodia are the same as those that take part in the formation of other actin structures such as lamellipodia or filopodia.

However, a major difference is that the invadopodia also have to deliver vesicles containing metalloproteinases to the cell membrane, for degradation of the extracellular matrix. These vesicles are most likely transported via the microtubule system. The actin polymerization in invadopodias is mediated by Cdc42, which activates the Arp2/3 complex via N-WASP/WASP. The actin filament elongation in the invadopodia requires the formin mDia, which not only has actin nucleation capacity but also can bind to the microtubule system (Ridley 2011).

APC in cell migration

Adenomatous polyposis coli (APC) is an important tumor suppressor in the human colon, but is also thought to play a pivotal role in polarized cell migration and adhesion. APC is frequently mutated in colon cancer. The mutated form of APC usually lacks the C-terminal part including the binding sites for β-catenin and Axin, however the Armadillo repeat is usually present. Mutated APC is unable to regulate the levels of β-catenin, which accu- mulates, moves to the nucleus and activates specific genes (Lai, Chien et al.

2009). APC binds directly to microtubules or via EB1 in the microtubule end tips and is crucial for the stabilization and polarization of microtubules (Bienz 2002; Nathke 2005; Watanabe, Noritake et al. 2005). The binding capacity of APC to the microtubule end tips is regulated by GSK-3β. Active GSK-3β phosphorylates APC, which decreases its ability to bind and stabilize microtubules (Zumbrunn, Kinoshita et al. 2001; Etienne-Manneville and Hall 2003). Cdc42 can regulate APCs binding capacity to the microtubule by activation of a Par6-atypical protein kinase C (aPKC) complex (Figure 9), at the leading edge of the cell. Par6-PKCξ interacts directly with GSK-3β and phosphorylates it at Ser9, which causes its inactivation (Etienne-Manneville and Hall 2003), thus preventing GSK-3β to phosphorylate APC, wereby the binding capacity of APC to microtubules remains. The microtubules can thereby be captured at the positive end and be polarized. APC has recently been described to have an actin nucleating activity and acts together with mDia to elongate the actin filaments (Okada, Bartolini et al. 2010; Ridley 2011). mDia are also known to stabilize the microtubule system and it is likely that APC and mDia coordinate the actin cytoskeleton and the microtubule system at the leading edge (Ridley 2011). APC has also been found to interact with IQGAP, an effector molecule of Cdc42 and Rac1, which pro- vides an indirect link between APC and F-actin (Watanabe, Wang et al.

2004; Watanabe, Noritake et al. 2005).

(31)

Figure 10. APC structure and binding partners

MAPKases in cell migration

Both TGFβ and EGF are known to induce activation of the p38 and JNK MAPK, which both are known to play an crucial role in cell migration (Huang, Jacobson et al. 2004). TGFβ can via p38 cause activation of RhoA and Cdc42, which induces actin rearrangement (Edlund, Landstrom et al.

2002). Smad7 localize to membrane ruffles at an early time point after TGFβ stimulation and is required for the TGFβ-induced activation of p38 and Cdc42 (Edlund, Landstrom et al. 2004), as well as for inactivation of GSK- 3β (Edlund, Lee et al. 2005). Active p38 can also activate MAPK-activated protein kinase 2/3 (MAPKAPK 2/3), which in turn phosphorylates heat shock protein 27 (HSP27). Unphosphorylated HSP27 is an actin cap binding protein, while phosphorylation of HSP27 cause detachment of HSP27 from the actin cytoskeleton. This reveals new sites for actin nucleation, leading to enhanced actin polymerization which is important for formation of the actin structures involved in cell migration (Rousseau, Dolado et al. 2006). Acti- vated JNK can induce cell migration by phosphorylation of paxillin, which is involved in cell adhesion or by phosphorylation of spire, another actin nu- cleator factor which cause actin polymerization at unbranched actin filaments (Ridley 2011).

(32)

Cell cycle checkpoints, genetic stability and apoptosis

Cells are constantly exposed to genotoxic agents, such as ionizing or ultraviolet radiation, various chemicals and drugs, and reactive metabolites, which can give rise to DNA damage. Cell cycle checkpoints and DNA repair systems protect the cells from acquiring mutations. The cell cycle checkpoints can either cause cell cycle arrest or initiate pathways leading to apoptosis. In cancers components of the cell checkpoint machinery are frequently mutated, this gives rise to genetic instability and accumulation of mutations.

G1/S arrest can be induced by DNA damage, to prevent replication of damaged DNA (Pearce and Humphrey 2001; Ishikawa, Ishii et al. 2006; Stark and Taylor 2006). G1 arrest can also be induced by cytokines, such as TGFβ, to limit cell proliferation. TGFβ induces cell cycle arrest through Smad- dependent suppression of the oncogene c-Myc, but also by induction of the cell cycle inhibitors p21 and p57 (Heldin, Landstrom et al. 2009). The TGFβ-induced upregulation of p21 is in some cell lines dependent of Smad2/3 interaction with wild type p53. p53 interact with the MH1 domain of Smad2 and Smad3, in a TGFβ-dependent manner (Cordenonsi, Dupont et al. 2003). The G2/M checkpoint prevents segregation of damaged chromosomes during mitosis. G2/M checkpoint can be activated through stress- induced p38-dependent phosphorylation and translocation of Cdc25B (Pearce and Humphrey 2001).The arrest will give the cells time for DNA repair processes or replication to be completed before the cell division (Pearce and Humphrey 2001; Ishikawa, Ishii et al. 2006; Stark and Taylor 2006).

Most of the solid tumors are aneuploid, which means that they have an ab- normal chromosomal content. Aneuploid cells develops due to failure in the mitotic check point machinery (also named spindle assembly check point) (Pearce and Humphrey 2001; Kops, Weaver et al. 2005). The mitotic check point is responsible for correct chromosome segregation and delays mitosis until all chromosomes have successfully made spindle-microtubule attach- ments. Cells that can not satisfy the mitotic checkpoint can die in mitosis by apoptosis or necrosis or exit mitosis as an aneuploid cell. The faith of the aneuploid cell is to die at an G1 tetraploidy checkpoint, due to apoptosis or to be viable but not reproductive. Failure in the G1 tetraploidy check point,

(33)

allows the aneuploid cell to survive and enter a new cell cycle (Pearce and Humphrey 2001; Kops, Weaver et al. 2005). Development of cancer is fa- voured if chromosomes that are lost contain a tumor suppressor gene or if the gained chromosomes contains oncogenes. Cells with instable chromosomes (in tissue culture) seem to escape the mitotic checkpoint in higher frequency compared to normal cells, when treated with microtubule- disrupting agents such as nocodazole (Rajagopalan and Lengauer 2004).

ATM and p53 guardians of DNA

Ataxia telangiectasia mutated (ATM) and p53 are critical guardians against cellular carcinogenesis, due to their role to preserve genetic stability via their regulatory role on cell cycle check points and DNA repair (Kops, Weaver et al. 2005).

ATM is a 370 kDa protein that belongs to PI3K superfamily. If not repaired or if repaired incorrectly the double strand breaks can cause cell death or mutations and chromosomal translocations, which can give rise to cancer. In non-stressed cells, ATM consists as a dimer or multimer with its kinase domain bound to a neighbouring ATM molecule. This interaction makes ATM stable in the cells and also prevents it from phosphorylating other substrates.

Upon DNA damage, ATM is recruited to the free DNA ends (Stucki and Jackson 2006) and an unknown signal makes the kinase domain of one ATM to phosphorylate Ser1981 of the interacting ATM molecule (Bakkenist and Kastan 2003). The phosphorylated ATM molecule dissociates from the complex and phoshorylates target proteins. Activated ATM induces G1-cell cycle arrest, by activation of p53, Mdm2 and Chk2. ATM is also involved in S-phase checkpoint, G2 cell cycle arrests and in DNA repair. At the site for DNA damage, a dramatic accumulation of DNA damage response proteins can be visualized as a distinct but microscopical sub-nuclear structure called DNA damage foci. In the DNA damage foci ATM phosphorylates histone H2AX, which is involved in DNA repair (Bakkenist and Kastan 2003).

p53 is a tumor suppressor, which acts by stopping cell cycle progression or promoting apoptosis as a response to stress stimuli, such as oncogene activation or DNA damage. p53 has a short half life and it is normally expressed in low levels in unstressed mammalian cells. The level of DNA damage regulates the level of p53. Low level of p53 seem to induce transcription of genes involved in cell cycle arrest such as p21 and Mdm2, while a higher level seems to induce transcription of genes involved in apoptosis such as Bax, PUMA and Noxa. This is consistent with the theory that low levels of DNA damage is possible to repair, while if a cell has severe damages in DNA it is better for the organism if it dies to prevent it from developing into a cancer

(34)

(Aylon and Oren 2011). When the cells are exposed to stress stimuli, p53 is stabilized and translocates into the nucleus where it binds DNA and activates or represses specific target genes. There are more than 150 genes regulated by p53, most of them are involved in regulation of cell-cycle arrest, apoptosis and/or DNA repair (Green 1998; Hengartner 2000; Bode and Dong 2004). p53 can also be regulated by posttranslational modifications, such as phosphorylation, ubiquitination, sumoylation and more. Different kinds of modulation can give rise to different gene expressions. Both ATM and p38 are known to phosphorylate p53 on serine 15 as a response to DNA damage or UV-light, this cause activation of p53 and apoptosis (Bode and Dong 2004). Loss of the tumor suppressor p53 is a way for cancer cells to circum- vent apoptosis (Hanahan and Weinberg 2011). p53 is frequently mutated in several different cancers and is correlated with a poor patient prognosis. It has been shown that Smad2 and Smad3 serve as an essential adaptor for formation of a complex between mutated p53 and the p53 family member, p63. Within this complex, mutant-p53 antagonize the function of p63, result- ing in increased TGFβ-induced cell migration and invasion. This shows that mutation of a tumor suppressor not only makes the cells lose the tumor sup- pressive effects but it can also induce tumor promoting effects (Adorno, Cordenonsi et al. 2009).

Apoptosis

Programmed cell death, apoptosis, is a way for the organism to tightly control the cell number and tissue size, but it is also a way to protect from cancer. The apoptotic cascade can be divided in three phases: initiation, integration and execution. In the initiation phase, apoptosis is triggered by stress signals or specific factors (for example TGFβ and Wnt) acting through a subset of receptors. During the integration phase, signals from several signaling pathways are balanced, and the decision is made regarding whether the execution of cell death should start (Schuster, Dunker et al. 2002). Dur- ing the execution phase, a set of cysteine proteases called caspases are activated (Green 1998). There are two main apoptotic pathways in mammalian cells, the death receptor pathway and the mitochondrial pathway (Figure 11).

The death receptor pathway is activated by binding of a death receptor superfamily ligand such as Fas to the Fas receptors, which cause receptor clus- tering. The signal is transferred into the cell via the adaptor molecule FADD (Fas-associated death domain protein), which cause activation of Caspase-8.

The mitochondrial pathway is activated in response to extracellular cues and internal insults, such as DNA damage. p53 activates the mitochondrial pathway via Bax2 (Bode and Dong 2004). Bax and Bid are pro-apoptotic proteins, which form ion channels through the mitochondrial membrane. This allows exit of apoptotic factors, such as cytocrome c, which associates with

(35)

Apaf-1 and procaspase-9 to form the apoptosome. The apoptosome in turn activates caspase-3, which induce activation of the apoptotic program (Green 1998; Hengartner 2000). p53 can also induce apoptosis by inhibition of the anti-apoptotic proteins BCL-2 and BCL-XL (Bode and Dong 2004).

Figur 11. Overview of apoptosis signaling. Adapted from Hengartner 2000.

(36)

TGFβ-induced apoptosis

TGFβ-induced apoptosis involves a crosstalk between Smad and non-Smad signaling pathways (Figure 12). TGFβ upregulates specific pro-apoptotic genes, such as SH2-containing inositol 5-phosphatase (SHIP), death- associated protein kinase (DAPK) and TGFβ-induced early response gene1 (TIEG1), through the Smad signaling pathway. It has been reported Smad3 is important for TGFβ induce expression and activation of the Fas receptor, which in turn activate caspase-8 and the apoptotic program. TGFβ also mediates apoptosis independent of the Smad signaling pathway via the adaptor protein death-associated protein (Daxx) (Valderrama-Carvajal, Cocolakis et al. 2002), which is normally associated with the Fas receptor and mediates the activation of JNK- and Fas-induced apoptosis. Daxx binds directly to the cytoplasmic domain of the TGFβ receptor type II and is required for the activation of the JNK pathway (Schuster, Dunker et al. 2002). Another way for TGFβ to induce apoptosis is by promoting export of the mitochondrial septin family member ARTS (apoptosis-related protein in TGF-b signaling pathway) to the cytoplasm. ARTS inactivates X-chromosome-linked inhibitor of apoptosis protein (XIAP), which in turn leads to caspase-3 activation and apoptosis (Moustakas and Heldin 2005). Both Bax and p53 are key factors in TGFβ-induced apoptosis (Teramoto, Kiss et al. 1998). TGFβ1 is essential for rapid p53-mediated apoptosis or cell cycle block as a result to DNA damage (Ewan, Henshall-Powell et al. 2002). TGFβ can also induce apoptosis via TRAF6-TAK1 mediated activation of p38 (Sorrentino, Thakur et al. 2008). It has been demonstrated that p38 induces apoptosis by activation of caspase 3 (Yu, Hebert et al. 2002). TGFβ and Wnt have been found to synergistically regulate genes involved in apoptosis. As mentioned before, Smad7 is important for TGFβ-induced and p38 dependent inactivation of GSK-3β, which leads to an accumulation of β-catenin (Edlund, Bu et al.

2003). Smad7 interact with β-catenin and its transcription factor TCF/LEF to regulate transcription of c-myc and apoptosis in PC3U and HaCat cells (Edlund, Lee et al. 2005). In addition TGFβ induce cell death by activation of JNK MAPK and by down regulation of the anti-apoptotic ERK MAPK (Schuster, Dunker et al. 2002). Smad7 is important for activation of JNK and induction of apoptosis (Mazars, Lallemand et al. 2001). Smad7 also sensi- tizes different cell lines to apoptosis induced by stimuli distinct from TGFβ such as serum withdrawal and loss of cell adhesion. In epithelial cells, the activity of survival nuclear factor κB (NF-κB) is inhibited by Smad7, leading to potentiated apoptosis (Lallemand, Mazars et al. 2001).

(37)

Figure 12. TGFβ-induced apoptosis. Adapted from Edlund, Lee et al. 2005;

Sorrentino, Thakur et al. 2008.

(38)

Present investigations

Aim

The aim of this thesis was to study the role of the adaptor proteins Smad7 and TRAF6 for TGFβ-induced cellular responses, such as cell migration, invasion and apoptosis in cancer cells. As we observed that Smad7 played an important role for TGFβ-induced cell migration, we also started to investigate the role of Smad7 for EGF-induced cell migration.

TRAF6 ubiquitinates TGFβ type I receptor to promote its cleavage and nuclear translocation in cancer (Paper I)

In Paper I, we investigated the role of TRAF6 in TGFβ-induced responses in prostate carcinoma cells.

Recently, it was shown that activation of the TGFβ receptor complex caused recruitement and kinase-independent activation of TRAF6, which in turn caused Lys63-dependent ubiquitination and activation of TAK1 (Sorrentino, Thakur et al. 2008). We found that TGFβ-induced activation of TRAF6 caused polyubiqitination of TβRI as well as activation of PKCζ, which in turn caused activation of the metalloproteinase TACE. TACE was found to cleave the TβRI between Gly120 and Leu121; this releases the 34 kDa intracellular domain (ICD) of TβRI, which translocates to the nucleus where it interacts with the transcriptional co-regulator p300. The intracellular domain also binds to the Snail promoter and was found to cause upregulation of Snail and MMP2. Both Snail and MMP2 has been linked to tumor invasiveness (Bernardi et al 2008). Our data showed that TGFβ-induced activation of TRAF6, PKCζ and TACE, were important for nuclear accumulation of the TβRI ICD and TGFβ-induced invasion in prostate cancer, breast cancer and lung cancer cells. However, we could not detect the TβRI ICD in the nucleus of primary prostate cancer cells, indicating that this is a mechanism to specifically induce invasion of tumor cells. Interestingly, the nuclear accumulation of the TβRI ICD could also be detected in sections of malignant human tumors.

(39)

APC and Smad7 link the TGFβ type I receptors to the microtubule system to promote cell migration (Paper II)

In Paper II, we investigated the possibility that interactions between proteins in the Wnt signaling pathway (APC, GSK-3β and β-catenin) could be regulated by effectors in the TGFβ signaling pathway (Smad7 and p38). We were also interested in the biological function of APC and Smad7 in this pathway.

Smad7 is believed to work as a scaffold protein in the TGFβ-induced activation of p38 (Edlund, Bu et al. 2003). We observed that Smad7 and active p38 were important for the inactivation of GSK-3β as well as for accumulation of β-catenin. GSK-3β and APC were found to interact with Smad7 and p38 in a TGFβ-dependent manner. We found that Smad7, APC and active p38 were important for TGFβ-induced cell migration and the formation of membrane ruffles. In addition, TβRI, Smad7, APC, p38, GSK-3β and β-catenin were found to localize in membrane ruffles upon 30 minutes of TGFβ stimulation.

APC is known to bind to microtubule end tips and to be important for microtubule polarization and stabilization. We saw that Smad7 and active p38 were important for TGFβ-induced colocalization of APC with the microtubule end tips as well as for binding to β-tubulin, a component of the microtubule system. Smad7 and TβRI were also found to bind β-tubulin in response to TGFβ stimulation; interestingly this interaction was dependent of APC. Knock down of either Smad7 or APC inhibited microtubule polarization.

In summary, we provide evidence that TGFβ induce formation of a complex between p38, Smad7, GSK-3β, APC and β-catenin. This complex forms a bridge between the TβRI at the cell membrane in the leading edge and the microtubule system, and is important for microtubule polarization, ruffle formation and cell migration.

Smad7 and APC are required for EGF-induced cell migration in human prostate epithelial cells (Paper III)

In Paper III, we investigated the role of Smad7 and APC in EGF-induced cell migration.

We observed that EGF-induced cell migration was impaired in cells knocked-down for Smad7 or APC, compared to cells treated with control siRNA. Smad7 and APC were shown to be necessary for EGF-induced activation of p38 and JNK MAPK, which both are known to be involved in regulation of cell migration, but not for phosphorylation of Akt, which indi- cated that some of EGFR signaling capacity still remains unaffected. We

(40)

also noticed that Smad7 and APC are important for EGF-induced phosphoy- lation of the EGFR at Tyr1068, which is known to be important for recruiting Grb2 and activation of JNK (Frey, Golovin et al. 2004). APC was found to be important for both TGFβ- and EGF-induced activation of TAK1. In addition, both Smad7 and APC were found to be important for EGF-induced membrane ruffling.

TGFβ1-induced activation of ATM and p53 mediates apoptosis

in a Smad7-dependent manner (Paper IV)

In Paper IV, we investigated the role of Smad7 in TGFβ-induced apoptosis.

Previous investigations have shown that TGFβ induce apoptosis in prostate cancer (PC3U) cells by activation of p38 MAP kinase in a Smad7-dependent manner (Edlund et. al., 2003). Since p38 is known to phosphorylate p53 upon DNA damage or UV-radiation and cause apoptosis, we wanted to find out if p53 affected p38-mediated and TGFβ-induced apoptosis as well. We found that p38 is important for TGFβ-induced activation of p53 at Ser15, upregulation of the pro-apoptotic p53 target gene PUMA, and induction of p53-mediated apoptosis. p38 was found to interact with both Smad7and p53.

The kinase ATM is known to activate p53 in response to DNA damage (Bakkenist and Kastan, 2003; Bode and Dong, 2004). We found that TGFβ- induced activation of ATM via activation of p38. Smad7 acted as a scaffold protein for the interaction between active p38 and ATM, which was important for the activation of ATM. Inhibition of p38, ATM or p53 was found to inhibit TGFβ-induced apoptosis. Transfection of wt p53 in PC3U cells (which has a mutant non-functional p53) increased the apoptosis frequency both in the absence and presence of TGFβ. ATM is known to induce DNA repair by activation of H2AX in DNA damage foci (Bakkenist and Kastan 2003). Smad7 was found to colocalize with γ-H2AX Ser139 in a TGFβ- and p38-dependent manner. p53 and ATM are both important for genetic stability and cell cycle check points (Kops, Weaver et al. 2005). We saw that cells knocked-down for Smad7 showed mitotic defects, such as multi- and micronuclei.

In this paper, we show that TGFβ induces phosphorylation of ATM Ser1981, which leads to apoptosis in epithelial cells expressing p53. The TGFβ- induced phosphorylation of ATM, as well as of p53, is dependent of Smad7 and p38. Smad7 is also important for TGFβ-induced cell cycle arrest and genetic stability.

(41)

Future perspectives

Paper I

We have seen that the TβRI is cleaved in proximity to the plasmamembrane by TACE. The Notch receptor is also cleaved by TACE, which in this case creates a substrate for a second cleavage inside the cell membrane by γ- secretase, which cause release of the intracellular domain. The intracellular domain translocate to the nucleus where it regulates specific target genes. It would of course be of interest to determine if γ-secretase cleaves the TβRI as well.

Paper II

Smad7, APC and p38 have been shown to be involved in TGFβ-induced cell migration. We have seen that Smad7 is important for TGFβ-induced activation of p38 and inactivation of GSK-3β. GSK-3β is known to inhibit the binding of APC to the microtubule system and to be required for proper cell polarization. By immunoprecipitation studies, we saw that Smad7, APC, p38, GSK-3β and β-catenin interact upon TGFβ stimulation. The complex forms a link between TβRI in the leading edge and the microtubule system, and is important for cell polarization, membrane ruffling and cell migration.

However, we have not knocked-down GSK-3β to determine the relevance of GSK-3β in cell migration, membrane ruffling, cell polarization and for the complex formation. These experiments are ongoing.

Paper III

This is a preliminary study and several of the experiments need to be re- peated. The level of the EGFR has varied a bit in the experiments, but gener- ally it seems to be a little bit less in the presence of Smad7 siRNA or APC siRNA, compared to control siRNA. It would be important to find out if this is due to an effect of Smad7 on the stability of the receptor, or if it is the transcription of the receptor that is affected. Smad7 are known to bind to the TβRI and affect its stability by recruiting Smurf, as well as suppress its signaling by competing with R-Smads for binding. It would be interesting to find out if Smad7 also are able to bind the EGFR, and to find out if Smad7 is required for phosphorylation of Tyr1068. It would be interesting to find out

(42)

if also other tyrosin phosphorylation sites are inhibited, for instance Tyr845, which is important for stabilization of the activation loop or 1082 or 1173, which are also known to be important for Erk MAPK activation and cell migration. We observed that phosphorylation of p38 and JNK are decreased in the presence of Smad7 and APC siRNA. It would be interesting to determine also if phosphorylation of Erk is decreased. In the TGFβ signaling pathway, Smad7 acts as a scaffold protein needed for the activation of TAK1 and p38. In this study, we saw that EGF induce phosphosphorylation of TAK1 and this phosphorylation was dependent on APC. It would be interesting to find out if Smad7 also is of importance in the activation process of TAK1 downstream of the EGFR. The activation mechanisms for p38 and JNK downstream of the EGFR are not known; it would be interesting to investigate if Tyr1068 binding to Grb2 or TAK1 are involved. It would also be interesting to look by immunofluorescence at the EGFR to see if the localization of the receptor is affected in the presence of Smad7 or APC siRNA.

We have found that APC and Smad7 are important for TGFβ- and EGF- induced cell migration. It would be of interest to determine if this actually is a general mechanism or if it is specific.

Paper IV

In Paper IV, we see that Smad7 is important for TGFβ-induced activation of ATM and p53, and induction of p53-mediated apoptosis. We also saw that Smad7 colocalizes with H2AX, a substrate of ATM, in DNA damage foci.

Since H2AX is involved in DNA repair, it would be interesting to find out if Smad7 also is involved in DNA repair. We observed that knock-down of Smad7 increased the number of cells with micronuclei and multinuclei. This indicates that Smad7 is important for genetic stability. We showed in paper II that Smad7 can bind β-tubulin as well as affecting APCs binding to microtubule, APC is also known to be important for chromosome segregation. It would be interesting to find out if Smad7 is important for chromosome segregation as well, or if Smad7 is involved in cell cycle control as ATM and p53.

(43)

Acknowledgements

During the time I spent at Ludwig institute for cancer research I got to know so many nice and interesting people. This work would not have been possible, without collaborations and great support from my colleagues, family and friends.

I would especially like to express my gratitude to my supervisor Marene Landström. Thank you for always being very understanding and supporting.

You are a great scientist and I have really enjoyed sharing your enthusiasm about science. You are always very optimistic, inspiring, friendly and en- couraging. I am happy that you let me join the group and I have really learnt a lot from working with you.

Carl-Henrik Heldin, the director of the institute and my co-supervisor. I am impressed of your great scientific knowledge and for always being so gener- ous with your time. Thank you so much for your support throughout these years I found it invaluable.

In addition I would like to thank previous and present members of the group.

First of all, a great thanks to the ones I have cooperated with: Sofia Edlund, you really took good care of me when I first came to Ludwig as a student. I am so grateful for the time you spent teaching me immunofluorescence.

Noopur Thakur, we have had so much fun together, you always make me laugh. Yabing Mu, I am really grateful that you continued the projects while I was on maternity leave; hope we get our final paper out soon. Reshma Sundar and Shyam Kumar, it was nice to get to know you. So Young Lee and Shouting Zhang. I would also like to thank group members that I have not directly cooperated with but that were always very helpful and that con- tributed to a nice atmosphere in the lab: Anahita Hamidi, Maria and Ihor Yakomowich and Jie Song. And of course I would like to thank some spe- cial people that left the group Anders Marcusson, Susann Grimsby, Ales- sandro Sorrentino and Verena von Bülow, we really had a nice time to- gether!

I would also like to thank a few of the past and present group leaders at the institute: Carina Hedberg and Jonan Lennartsson, thank you for your

(44)

inspiration and advices, for nice discussions both scientifically and on a friendly level and thank you so much for all good laughs we have shared.

Aris Moustakas and Evi Heldin, thank you for all input during our semi- nars and your friendly attitude. Ulf Hellman, I really appreciate that you always lend me a bike when I have to go down town.

A great thanks to all the past and the present PhD students and post Docs, that make each day more fun. Especially you Helena Porsch, your smile always cheer me up. You have helped us so much both privately and scien- tifically. Michael Vanlandewijck, thank you so much for all the help with computer, I don’t know what I would have done without you. Berit, Mar- cus, Erna, Glenda and Peter for nice discussions in the office. Susann Karlsson and Åsa Fransson, for showing that it is also possible to defend with children. You are my inspiration, when I feel tired after sleepless nights. Fatima, for always giving me complements it makes me feel good, even though I know that it is just for making me flatter you back

Past and present members of the institute for making a Ludwig a nice place to be at. Especially Uffe, for all help with the computers and nice friendship, Lasse, Lotti, Aino, Anita, Aive, Ulla, Eva, Ingegärd and Christer, I have really enjoyed your company and I am so grateful for all help and support that I have got from you.

Ett stort tack till alla mina vänner utanför Ludwig institutet som jag tyvärr träffar aldeles för sällan men som alltid ställer upp och som ger perspektiv på vad som är viktigt i livet. Mamma och pappa, ni är underbara, tack för alla gånger ni stöttat mig genom åren, för att ni alltid tror på mig och uppmuntrar mig och alla de resor ni gjort hit till Uppsala för att hjälpa oss med barnen, oavsett om vi varit friska eller sjuka. Ni är ovärderliga! Ett stort tack också till Tobias mamma Lena, som alltid ställer upp och hjälper till med både barn och hushållsarbete, du har gjort en enorm insats! Tobias, jättetack för alla” pep talks” och för att du alltid finns där för mig när jag behöver dig. Jag hade aldrig klarat det här utan dig! Du har dragit ett jättelass på hemmaplan, du är enastående. Älskar dig! Mina underbara barn Hampus, Anton och Lukas, som kanske inte gjort det hela så mycket enklare, men som definitivt gör livet mycket roligare. Det finns inget som går upp emot att se er glada, det ger enormt mycket energi.