**3.1 CD44**

212 Breast Cancer – Focusing Tumor Microenvironment, Stem Cells and Metastasis

aberrations in tooth and follicle development (Weigel and DeAngelis, 2007). It is not fully understood why only HAS2 is absolutely required for organogenesis, but it has been suggested that it produces high molecular weight tissue HA while the other HASs produce the smaller HA sizes (Itano *et al*., 1999). There are differences in the mechanisms by which HAS isoform expression and enzyme activity are regulated that may be relevant to their

BCA cells use several mechanisms to rapidly control the synthesis and release of HA, thereby modifying their ECM, including substrate availability, gene expression, posttranslational control of enzyme activity, and differential response to cytokines and ECM signalling. The availability of UDP sugars can profoundly influence the yield of HAS enzymes (Kakizaki *et al*., 2004). This has been demonstrated by the use of 4- Methylumbelliferone (4-MU), which depletes intracellular levels of UDP-Glc-UA (Kakizaki *et al*., 2004) by serving as a glucuronidation substrate. It blocks HA production and reduces

The genomic plasticity and instability of cancer cells often leads to chromosomal aberrations that can result in both de-regulation of gene expression and allele duplication. Chromatin breakpoint analysis using a BCA line revealed significant chromosomal rearrangements close to the HAS2 gene. These result in de-regulation of HAS2 expression and significantly higher HAS2 mRNA levels in transformed cells compared to normal breast cells (Unger *et al*., 2009). Detailed *in vitro* and *in vivo* studies of BCA lines and xenografts have provided numerous insights into the effects of genetically modifying HAS expression levels on HA concentration within the tumour and peri-tumoural stroma. Antisense inhibition of HAS2 in MDA-MB-231 BCA cells delays proliferation via a transient arrest of the cell cycle (Udabage *et al*., 2005). Knockdown of HAS expression also results in significant alterations in genes associated with HA metabolism. CD44 and HYAL1 expression are both down-regulated in response to antisense inhibition of HAS2. *In vivo*, MDA-MB-231 cells expressing antisense HAS2 do not form tumours in nude mice after 12 weeks, whereas the parental cell line readily establishes both primary and secondary tumours during this time. This clearly implicates tumour cell HA as a significant driver of BCA formation. Elevated HA accumulation within BCA peritumoural stroma is also a prognostic factor and appears to promote a microenvironment suitable for BCA growth. For example, HAS2-/- fibroblasts transplanted with BCA cells into the fat pads of NOD/SCID mice fail to recruit macrophages and promote angiogenesis to the same extent as HAS2+/+ fibroblasts. This defect results in decreased

The expression of all three HASs is controlled by growth factors and cytokines. However, there appear to be subtle differences in the response of each isoform that depend upon the cell type. For example, PDGF and TGFβ induce HAS2 expression in fibroblasts but HAS1 or 3 expression in synoviocytes and keratinocytes, respectively (Karousou *et al*., 2010). H-Ras transformation increases only HAS2 expression in 3Y-1 tumour cells, while transformation with v-src or v-fos increases both HAS1 and HAS2 expression in the same cells (Itano *et al*., 2004). Posttranslational modification of HAS, including phosphorylation by PKC, PKA, and the ERK/ErbB2 MAPK pathways (Goentzel *et al*., 2006, Itano and Kimata, 2008) as well as mono-ubiquitination (Karousou *et al*., 2010) also affects HAS activity. HAS3 serine phosphorylation is enhanced upon treatment with a PKC activator (Goentzel *et al*., 2006). All three HAS isoforms expressed by SKOV3 ovarian cancer cell line are phosphorylated by

functions and essential or non-essential roles in organogenesis (Tammi *et al*., 2008).

BCA tumourigenicty.

tumour volume (Kobayashi *et al*., 2010).

CD44 is a class I transmembrane receptor, which binds to HA via a link domain and is expressed by a variety of cells, including fibroblasts, endothelial and epithelial cells, smooth muscle, and haematopoietic cells. A vital role of CD44 is recruiting cells, including immune cells and fibroblasts, to sites of inflammation through HA-mediated signalling. Under homeostatic conditions, CD44 is in a low HA binding state, but during injury and tumourigenesis its binding affinity is increased and it mediates the inflammatory and tissue repair responses (Thorne *et al*., 2004, Naor *et al.*, 2008). CD44 is expressed as many different isoforms due to extensive splicing in a region proximal to the transmembrane domain (Thorne *et al*., 2004). The smallest CD44 isoform, CD44s (standard form), skips this variable region. The role of CD44s and variants in BCA progression is still controversial. For example, CD44s expression in CD44low MCF-7 human BCA cells results in xenograft metastasis to the liver (Ouhtit *et al*., 2007) while CD44-/- mice develop more lung metastases than wildtype animals in response to polyomavirus middle T (Lopez *et al*., 2005). Importantly, a recent study by Brown *et al*. (2011) demonstrated that CD44s expression is elevated and required for epithelial-mesenchymal transition of immortalized human mammary epithelial cells and for recurrence of HER2/*neu* induced murine mammary tumours (Lopez *et al*., 2005). HA synthesis is elevated in CD44+ BCAs

Hyaluronan Associated Inflammation and Microenvironment

inactivation of BRCA1 (Joukov *et al.,* 2006)

**3.3 TLR2 and TLR4** 

**3.2 RHAMM/HMMR** 

Remodelling Influences Breast Cancer Progression 215

Receptor for HA Mediated Motility (RHAMM/HMMR) belongs to a group of proteins that are found intracellularly as well as extracellularly. RHAMM does not contain a transmembrane domain or classical export signal and is likely exported through an unconventional mechanism that does not involve the Golgi/ER. RHAMM is expressed as multiple isoforms and one of these, an N-terminal truncation that lacks the first 163 aa residues, is transforming in mesenchymal cells (Hall *et al*., 1995). On the cell surface, RHAMM interacts with HA and forms complexes with transmembrane receptors such as CD44, PDGFR, and RON (Maxwell *et al.*, 2008). Interestingly, CD44 surface display is reduced in mesenchymal cells isolated from RHAMM-/- mice, demonstrating functional interplay between these two HA receptors (Tolg *et al*., 2006). RHAMM is elevated in most types of cancer in particular breast, ovarian, and prostate cancer, as well as in MM, AML and CML. In BCA, RHAMM is a tumour marker, novel susceptibility factor and prognostic factor for poor outcome (Maxwell *et al*., 2008). Consistent with these clinical correlations, RHAMM has tumourigenic properties in experimental systems that have been linked to its ability to bind HA. In BCA cells, RHAMM/CD44/HA complexes sustain phosphorylation and activation of the Ras/MAPK (ERK1,2) signalling pathway, leading to BCA progression and constitutively high rates of motility and invasion (Hamilton *et al*., 2007). The relationship between RHAMM and ERK1,2 activation has recently been confirmed in BCA samples where concomitant upregulation of phosphorylated ERK1,2 and RHAMM in tumour samples correlates with a high tumour grade (Ward C., in preparation). Intracellularly, RHAMM binds directly to tubulin and is involved in regulation of microtubule stability and turnover as a result of its association with ERK1,2. In mesenchymal cells, the absence of RHAMM increases microtubule stability resulting in reduced cell migration and aberrant mitotic spindle formation (Tolg *et al*., 2010, Groen *et al.*, 2004). RHAMM interacts directly with ERK1, inferring that RHAMM may act as a scaffolding protein that directs ERK1 to its substrates including microtubule associated proteins that regulate microtubule stability (Tolg *et al*., 2010). Interestingly, RHAMM expression is downregulated by p53, an important tumour suppressor gene, suggesting that RHAMM may be involved in p53 loss-induced tumour progression (Buganim and Rotter, 2008, Godar and Weinberg, 2008, Sohr and Engeland, 2008). RHAMM also acts on the BRCA1, pathway and may play an important role in BCA tumours arising from loss or

Toll like receptors (TLR) are part of a cellular defence mechanism that is based on pattern recognition. TLRs recognize and bind bacterial lipopolysaccharides, DNA, and, in the case of TLR2,4, small HA fragments. In general, HA-TLR2,4 interactions control innate immunity through several mechanisms. For example, TLR 2,4 activation results in cytokine and chemokine release and leads to expression of metalloproteinases (MMPs) in immune cells (Voelcker *et al.*, 2008). Versican, which is associated with poor prognosis and relapse in BCA, interacts with HA polymers to form cord-like structures that link TLR2 on endothelial cells and fibroblasts. This, in turn, causes the secretion of pro-inflammatory cytokines (Theocharis *et al.*, 2010). HA-TLR2,4 interactions also stimulate NFκB signalling and activate TNFα. In BCA cells, TLR 2,4 interact with CD44 and act as co-receptors to stimulate signalling through HA and CD44 regulated pathways which may play a role in breast

compared to CD44- and both CD44+ and HER2+ BCAs are amongst the most aggressive and invasive subtypes of BCA with poor prognosis. Expression of variant exons, in particular exon v6, is associated with increased *in vitro* cell migration and invasion of human BCA cells (Herrera-Gayol and Jothy, 1999). Although CD44v6 expression has been correlated with multiple clinicopathological features (primary tumour size, axillary nodal status, histological grade and pTNM stage) it is not an independent prognostic factor (Ma *et al*., 2005). A study by Rys *et al*. (2003) found a correlation between the expression of CD44 v3 and the presence of BCA metastasis. Additionally, high CD44s expression correlates with increased disease free survival in node negative invasive BCA (Diaz *et al*., 2005). The controversies surrounding CD44 and its role in BCA progression may be caused by a limited number of patient samples in some of these studies, heterogeneity of BCA, and CD44 expression by cancer stem cells. The latter, in particular, has raised much recent interest in CD44 since several groups have identified CD44 as a potential marker for BCA stem cells. This is a highly tumourigenic population of cancer cells that, although only representing a small percentage of cells in the tumour, are thought to be responsible for tumour recurrence, metastasis and treatment failure. Aggressive BCA and BCA tumour progenitor cells have enhanced CD44 expression, associated with an increase in HA synthesis and CD44-HA binding affinity (Heldin *et al*., 2008).

In BCA cells, HA triggers CD44 interactions with a variety of signalling mediators involved in cell proliferation, migration and chemo-resistance. Ankyrin is a membrane-associated component of the cytoskeleton that is involved in regulation of cytoskeleton turnover and IP3 receptor-mediated regulation of intracellular Ca2+. CD44-HA interactions induce CD44 ankyrin coupling and modify receptor-dependent Ca2+ mobilization (Bourguignon *et al*., 2008). CD44 also localizes ankyrin and IP3 receptor to lipid rafts, which are cholesterol and caveolin rich signalling microdomains in the plasma membrane (Fig. 3). The Rho GTPases, RhoA, Rac and CDC42, are key regulators of cell migration and HA stimulates RhoA in BCA cells. RhoA activity is regulated by RhoGEF, a guanine nucleotide exchange factor that forms a complex with CD44 in BCA cells. One of the downstream RhoA targets, ROK, phosphorylates the cytoplasmic domain of CD44 thereby increasing CD44-ankyrin interactions. Other targets of ROK are myosin phosphatase and myosin light chain, two important mediators of actin-myosin dependent membrane ruffling required for cell migration. HA also activates the PI3 kinase/AKT pathway: Gab-1 phosphorylation by ROK stimulates PI3 kinase and AKT activation, leading to increased cell proliferation, invasion and cytokine production (Bourguignon *et al*., 2008). Additionally, ROK phosphorylates and activates NHE1, a Na+-H+ exchanger, causing intracellular and extracellular acidification leading to HYAL-2 driven HA degradation, ECM breakdown and tumour progression. CD44-HA interactions stimulate signalling through Rac1, another RhoGTPase, via the GEF Tiam1. In MDA-MB-231 cells, CD44-HA interactions also activate c-Src kinase resulting in activation and nuclear translocation of the transcription factor Twist, miR-10b expression and down-regulation of the tumour suppressor gene HOXD10 (Bourguignon *et al*., 2010 Toole, 2004). CD44 undergoes sequential proteolytic cleavages resulting in the release of its ectodomain from the cell surface and formation of a CD44 intracellular domain fragment, which is translocated to the nucleus, acting as a transcription co-regulator (Nagano and Saya, 2004). CD44 ectodomain cleavage is mediated by MT1-MMP and is stimulated by multiple factors, including HA fragments and TGF-β (Kuo *et al*., 2009, Sugahara *et al*., 2006) which, contribute to tumour cell migration and invasion (Fig. 3).
