25
6 Discussion
6.1 Function and localization of HAS proteins
It has been challenging to detect hyaluronan synthases at protein levels, because the expression levels of HAS genes are normally relatively low in many cell types (Pienimäki et al. 2001, Recklies et al. 2001). In the first studies the antibodies utilized to detect HAS proteins were non-‐‑commercial (Jacobson et al. 2000, Kanomata et al. 2005, Spicer and McDonald 1998). In later studies also commercially available antibodies against hyaluronan synthases have been used (Miyake et al. 2009, Nykopp et al. 2009). In the study I of this thesis a set of commercial antibodies against different HAS isoenzymes was characterized.
At tissue level the immunostainings were compared to earlier results on mRNA expression levels in mouse embryonic tissues (Tien and Spicer 2005). Subcellular localization of endogenous HASs was detected and compared to earlier results gained with GFP-‐‑tagged recombinant proteins. In essence, the results were in line with earlier HAS mRNA in situ hybridizations and in vitro studies on hyaluronan synthases. The several controls performed to ensure the specificity of antibodies suggest that the antibodies used in this study are specific for each HAS isoenzyme and can be reliably utilized both in mouse and human tissues and cell cultures.
We observed immunostaining of all HASs both in intracellular locations and on the plasma membrane (I). Intracellular staining was found especially in ER and Golgi areas.
These findings support our model for the synthesis and transport of the HAS proteins from ER via Golgi to the plasma membrane, followed by uptake into endosomes and transport to lysosomes for degradation, or recycling to the plasma membrane (Deen et al. 2016). The model was established using GFP-‐‑tagged HASs (Müllegger et al. 2003, Rilla et al. 2005). The use of the GFP-‐‑tagged fusion proteins have shown that the transport to the plasma membrane is essential for the activity of HAS2 and HAS3 (Rilla et al. 2005), and a similar regulation was later shown for HAS1 (Siiskonen et al. 2014). However, the localization of HAS1 is mainly intracellular (Siiskonen et al. 2014), and studies on HAS1 have suggested also intracellular activity (Ghosh et al. 2009). Additionally, HAS2 and HAS3 have different cellular distribution, the relative amount of HAS3 being higher in plasma membrane than HAS2. Different turnover rates or stabilities of the HAS enzymes may affect their relative levels on the plasma membrane. HAS3 has remained active in the plasma membrane preparations in vitro for 8h, HAS2 3h and HAS1 1h (Itano et al. 1999). On the other hand, Deen et al. showed in live cells a rapid endocytosis of HAS3 and recycling back to plasma membrane (Deen et al. 2016). The overall turnover time of the HAS enzymes is dependent on the cell culture conditions (Bansal and Mason 1986, Kitchen and Cysyk 1995). The different lifetimes of HASs may also be explained by their different subcellular localizations.
The immunostainings and the GFP-‐‑tagged fusion proteins used in this study showed a large intracellular pool of HAS2 in the ER/Golgi area (I). This pool is assumed to act as a reservoir, and may be rapidly transported to the plasma membrane and activated as a result of acute response for wounding (Tammi et al. 2005), inflammation (Mack et al. 2012), or other external stimuli (Jacobson et al. 2000). HAS3 had a less prominent accumulation in the Golgi area than HAS2, which suggests that HAS3 may be constitutively active under basal conditions (Rilla et al. 2005). There are potential phosphorylation and O-‐‑
GlcNAcylation sites in HAS2 (Vigetti et al. 2011, Vigetti et al. 2012) and HAS3 (Deen et al.
2016, Goentzel et al. 2006), suggesting that post-‐‑transcriptional modifications may be
26
important regulators of the HAS activity. There is also ubiquitinylation site in HAS2 (Karousou et al. 2010).
Cells overexpressing mouse HAS3 or HAS2 produce long plasma membrane protrusions (Kultti et al. 2006). The present study shows the same phenomenon with human HAS constructs. These microvillus-‐‑like protrusions are dependent on hyaluronan synthase activity. Both HAS2 and HAS3 produce high amount of hyaluronan, but the number of plasma membrane protrusions is higher in HAS3 overexpressing cells (Rilla and Koistinen 2015). HAS1 has lower hyaluronan production activity and is also unable to induce the protrusions, even after induction with cytokines (Siiskonen et al. 2014). This indicates unknown differences in the regulation mechanisms of the different HAS isoforms. Except for acting as hyaluronan producing sites, the functional role of these protrusions has been unknown, but Rilla et al. showed that these protrusions act as a platform for the shedding of extracellular vesicles (Rilla et al. 2013b). Our immunostainings of mouse embryos showed HAS-‐‑positive protrusions also in vivo, findings in line with those of Koistinen et al who investigated the structure and regulation of these protrusions in more detail in vitro (Koistinen et al. 2015), and showed that they exist also in vivo in rat mesothelial tissues (Koistinen et al. 2016).
6.2 Role of hyaluronan in cancer and development 6.2.1 Hyaluronan role in cancer
Hyaluronan accumulation is a typical feature of several malignant tumors such as lung, breast, ovarian and prostate cancers (Tammi et al. 2008). There are several possible mechanisms behind the accumulation of hyaluronan. Induction of HAS transcription, followed by increased HAS protein expression is one possible explanation. Auvinen et al.
showed HAS1-‐‑3 upregulation in breast cancer (Auvinen et al. 2014), and poor patient survival correlated with increased HAS1 and HAS3 protein levels detected by their immunostainings. Fibroblast growth factor receptor (FGFR) activation has been shown to induce HAS2 activation and hyaluronan accumulation in breast cancer (Bohrer et al. 2014).
As discussed above, enhanced hyaluronan synthesis is associated with the formation of long cell surface protrusions (Kultti et al. 2006) and enhanced formation of extracellular vesicles (Rilla et al. 2013b). These protrusions and vesicles might have an important role in cancer progression by creating a favourable microenvironment for tumor progression and facilitating tumor cell invasion. On the other hand, in some cancers, like squamous cell carcinomas (Karvinen et al. 2003a) and melanomas (Siiskonen et al. 2013), hyaluronan content decreases during the progression of the cancer. The content of HAS2 decreases in higher grade melanomas (Siiskonen et al. 2013), and decreased hyaluronan is actually a prognostic factor in mouth squamous cell carcinoma (Kosunen et al. 2004) and melanoma (Karjalainen et al. 2000).
Another factor that regulates hyaluronan content and accumulation in tumor tissues is the rate of its degradation by hyaluronidases, or by non-‐‑enzymatic degradation through free radicals. Hyaluronidase activity can inhibit hyaluronan accumulation and regulate the effects of hyaluronan in cancer progression. In an experimental colon carcinoma model HYAL1 overexpression has been shown to suppress the tumorigenenicity of cells (Jacobson et al. 2002). On the other hand, Bouga et al. showed increased HYAL1 and HYAL2 expression in colorectal cancer (Bouga et al. 2010). HYAL1 and HYAL2 mRNA levels are lower in the endometrial cancer, compared to the normal endometrial tissue (Nykopp et al.
2010). HYAL1 expression correlates with microvessel density and capillary growth in bladder and prostate cancers (Lokeshwar et al. 2005)
27
The balance between the relative activities of the hyaluronan synthases and hyaluronidases might play a key role in the growth and progression of various tumors. In a prostate cancer cell model hyaluronan synthase and hyaluronidase co-‐‑operation is needed for significant tumorigenecity (Simpson 2006). Production of hyaluronan fragments by hyaluronidases may enhance tumor growth and neovascularization due to distinct signals triggered by them in hyaluronan receptors, as compared to the signals induced by high molecular weight hyaluronan (Cyphert et al. 2015). Fragmented hyaluronan might disengage high molecular weight hyaluronan from its receptors, and lead to an inhibition of the normal signaling activity of the receptor.
The present study showed hyaluronan accumulation especially in the cancer cells of mesotheliomas (III). There was a significantly higher amount of hyaluronan in the mesothelial cells as compared to adenocarcinoma cells. The same phenomenon was seen in earlier studies (Afify et al. 2005, Kanomata et al. 2005, Welker et al. 2007). Higher content of hyaluronan in pleural fluid with mesothelioma patients has also been shown (Thylen et al.
1999, Thylen et al. 2001). It has been suggested that hyaluronan content with mesothelioma cells work as marker for mesothelioma (Afify et al. 2005, Azumi et al. 1992), and findings in this study supports this idea. Combined with other mesothelioma marker hyaluronan staining could increase precision of diagnosis. In our study there was no difference in stromal hyaluronan content of these two cancers. Increased hyaluronan synthesis has shown to increase the malignancy of mesothelioma cells (Li and Heldin 2001).
There were no differences in CD44 staining between adenocarcinoma and mesothelioma cells. The stromal cells of mesotheliomas had a tendency for a weak immunostaining of CD44, as compared with the stromal cells of lung adenocarcinoma. Binding of hyaluronan to CD44 can activate signaling for tumor progression (Heldin et al. 2008). On the other hand, CD44 could reduce the effects of hyaluronan by contributing to hyaluronan endocytosis and degradation (Tammi et al. 1998).
6.2.2 Hyaluronan in development
Tien and Spicer have shown the spatial and temporal changes in hyaluronan synthase expression at mRNA level during embryonic development (Tien and Spicer 2005). The present results with HAS immunostainings are mostly in line with those findings. We showed a clear HAS2 staining in the developing heart valves during the E9 and 11 stages, as noted earlier at mRNA level. These results further confirm the key role of HAS2 during the development of the heart valves. Furthermore, the loss of HAS2 function is lethal in an early phase of the embryonic development and the expression of other HASs cannot compensate it (Camenisch et al. 2000). Interestingly, HAS2 expression during the formation of the endocardial cushions in zebrafish is downregulated by a specific microRNA (MIR-‐‑23) (Lagendijk et al. 2011). There is excessive endocardial cushion growth without restricted HAS2 expression and HA production. Brain tissue showed low HAS2 expression both at mRNA (Tien and Spicer 2005) and in protein levels.
However, the mRNA levels of the HAS genes did not always correlate with the HAS proteins, like in the follicles of the vibrissa (Tien and Spicer 2005). Similar contradictions have been detected in human tumors such as endometrial and ovarian carcinomas (Nykopp et al. 2009, Nykopp et al. 2010). Hyaluronan levels and HAS protein immunoreactivity were well correlated in tissues like developing heart and cartilage (I). On the other hand, hyaluronan-‐‑rich tissues like the vitreous body and the stromal compartment of the developing kidney had low levels of HASs. Rapid turnover rate of HAS protein in tissues is a possible explanation for this inconsistency between synthase proteins and hyaluronan (Tammi et al. 2011, Vigetti et al. 2012), or it may arise from temporal changes in hyaluronan synthase activity, for example due to a fluctuation of the precursor sugar levels, or different rates of hyaluronan turnover. Diffusion of hyaluronan from adjacent tissues may also explain these discrepancies. There was an overlapping
28
expression of HASs in many tissues of mesodermal origin, cartilage one of them. However, there were no skeletal malformations in HAS1 KO, HAS3 KO or HAS1/HAS3 douple KO (Bai et al. 2005, Camenisch et al. 2000, Tien and Spicer 2005), suggesting that HAS2 can compensate for the hyaluronan synthesis of missing HAS1 and HAS3 during bone formation. Nevertheless, HAS3 KO mice have chronic inflammation in joints (Chan et al.
2015). Interestingly, mice with HYAL2 deficiency have congenital disorders in frontonasal and vertebral bone formations (Jadin et al. 2008), while HYAL1 deficiency results in mild abnormalities of the articular cartilage (Martin et al. 2008).
6.3 Effects of UV radiation on hyaluronan metabolism
Long-‐‑term and high-‐‑dose UV exposure is the main the risk factor of the epithelial tumors of the skin. UV radiation is also known to influence the hyaluronan content of the skin.
However, little specific information about epidermal hyaluronan metabolism associated with UV-‐‑induced hyperplasia and malignancy has been available. This study showed for the first time the gross changes in the metabolism of epidermal hyaluronan caused by long-‐‑
term UV radiation (II). The results showed an increased content of hyaluronan both in the epidermis and dermis in skin after UV radiation, and with concomitant rise of all HASs and CD44. There was a positive correlation between hyaluronan content and skin hyperplasia, indicating that hyaluronan accumulates in the early stages of skin squamous cell carcinomas. Karvinen et al. described similar observations in human samples (Karvinen et al. 2003a) and Koshiishi et al showed in mice that UV-‐‑dosage associated with epidermal hyperplasia lead to accumulation of hyaluronan in the dermis (Koshiishi et al. 1999).
However, contradictory reports exist. Thus, short term exposure of UVA/UVB was shown to decrease hyaluronan content in mouse skin, being reconstituted within 24 h (Calikoglu et al. 2006), while in human skin it was shown to increase epidermal hyaluronan, but a temporarily decreased dermal hyaluronan content was observed (Werth et al. 2011). Dai and coworkers experiments suggest that long-‐‑term UV irradiation of mouse skin causes loss of hyaluronan and downregulation of the hyaluronan synthases in the dermis, while epidermal hyaluronan does not change (Dai et al. 2007). In line, Werth and coworkers, showed in human skin that repeated UVB irradiation does not alter the content of hyaluronan in the epidermis, and repeated UVA irradiation may even decrease it (Werth et al. 2011). On the other hand, studies on cultured keratinocytes showed an increased hyaluronan content in the culture medium, and increased expression of hyaluronan synthases (Averbeck et al. 2007, Kakizaki et al. 2008, Rauhala et al. 2013). Cultured fibroblasts respond to UVB radiation by elevated HA-‐‑synthesis and HAS levels (Dai et al.
2007). Opposite responses in vivo and in vitro might be due to UVB-‐‑induced signaling from keratinocytes to dermal fibroblasts
There might be several reasons for these contradictory results between the different studies. The spectral types of irradiation used in the different studies are not comparable, being in some cases UVB, UVA or combination of both. Furthermore, the doses and exposure times are also different. Moderate dose of UVB increases hyaluronan, but a higher dose of UV radiation appears to result in a reduced content of hyaluronan (Calikoglu et al.
2006, Kakizaki et al. 2008). The wavelength is apparently important since (Werth et al. 2011) reported that UVA decreases the content of hyaluronan in the epidermis, while UVB, although has no effect on the total epidermal hyaluronan, increases it in the basal layer of the epidermis. A narrower spectrum lamp used in earlier studies, especially those with
29
UVA, had no effect (Kakizaki et al. 2008), or even inhibited hyaluronan synthesis (Calikoglu et al. 2006). The cell culture conditions between the separate studies vary, and likely lead to different responses to the UV radiation. Our study on mouse was a controlled, long-‐‑term experiment which lasted for 10,5 months, using UVA and UVB transmitting lamp (Kumlin et al. 1998), however most of biological effects come from UVB. Dai et al. used UVB transmitting lamp and the mice were exposed three times in a week during 26 weeks (Dai et al. 2007). Biological effects were different between our study and Dai et al (Dai et al.
2007). There was no accumulation of hyaluronan or development of hyperplasia in epidermis according to Dai et al, although in both models squamous cell carcinoma developed in 20 %of exposed animals. Similar data on humans are not available since those by Averbeck et al. (Averbeck et al. 2007) and Werth et al. (Werth et al. 2011) just spanned from hours to days.
Direct comparison of in vivo and in vitro results is difficult. Normal skin has protective elements like melanocytes and a thick stratum corneum. The lack of these elements makes keratinocytes in cell cultures more responsive to the UV radiation than the keratinocytes in an intact skin. The thickness of the stratum corneum differs between species, which may contribute to the differences between the results, and the levels of the epidermal hyaluronan also differ between the species. Normal human skin contains hyaluronan both in the epidermis and dermis. Especially the extracellular space between keratinocytes is rich in hyaluronan in human skin (Tammi et al. 1988, Werth et al. 2011). Both monolayer and organotypic cultures of rat keratinocytes also produce hyaluronan (Rauhala et al. 2013).
However, normal mouse epidermis is almost negative for hyaluronan (II, Figure 1a).
The exact role of hyaluronan in the UV radiated skin is still under investigation. The studies in vivo have shown that hyaluronan treatment may reduce the detrimental effects of UVR in keratinocytes (Hasova et al. 2011). Hyaluronan synthesis induced by HAS2 or HAS3 has been shown to protect against the apoptosis induced by UVR (Wang et al. 2014).
An association between protein kinase p38 function and skin tumor development has been shown (Schindler et al. 2009). Accordingly, acute UVB irradiation triggers p38 activation and Has upregulation in keratinocytes (Rauhala et al. 2013). These findings are in line with earlier studies on humans (Karvinen et al. 2003a, Pirinen et al. 1998) indicating hyaluronan accumulation in hyperplastic epidermis and early phases of the squamous cell carcinoma lesions. Hyaluronan accumulation is thus likely to contribute to the development of the malignant growth in the epidermis during the early stages of the malignancy. Hyaluronan thus seems to have a dual role in the UVradiated skin. It helps to maintain homeostasis through the inflammatory response to irritation by shielding the cells, while a longer duration or a higher dose of the exposure might result in the development of a malignancy.
However, once initiated, the squamocellular cancers proceeds into the next steps that do not need hyaluronan which can actually retard the development of the tumor, and must be reduced, as seen in the high grade squamocellular cancers and melanomas (Karjalainen et al. 2000, Kosunen et al. 2004).
The main hyaluronan receptor CD44 was elevated in the UVR treated epidermis (II).
There was a correlation between hyaluronan and CD44, as seen also in the earlier studies (Karvinen et al. 2003a). On the other hand, a reduction of CD44 has been reported after an acute high dose UVR exposure (Calikoglu et al. 2006, Rauhala et al. 2013). Hyaluronan and CD44 accumulation during epidermal hyperplasia was also seen during wound healing (Tammi et al. 2005) and in psoriasis (Tammi et al. 1994a). Hyaluronan and CD44 accumulation might be required during epidermal activation in hyperplasia caused by an injury and inflammation, and in the early phases of squamous cell carcinomas.
30
6.4 Putative role of hyaluronan and hyaluronan synthases as tools for diagnostics and treatment of cancer
The results of this study support the idea that the detection of hyaluronan and hyaluronan synthases from tissues or tissue fluids are prospective tools for diagnostics. Hyaluronan accumulation and hyaluronan synthase activity have an important role in several pathological conditions, including tumor growth (Sironen et al. 2011) and inflammation (Litwiniuk et al. 2016). Hyaluronan content in the tumor cells or their stroma have a strong potential for prognostic factors (Anttila et al. 2000). The correlation between the levels of hyaluronan synthases and hyaluronan contents with breast cancer aggressiveness and poor patient outcome has been established (Auvinen et al. 2014). Changes in HAS1 and HAS2 were detected in oral lichen planus compared to normal oral mucosa (Siponen et al. 2015).
Hyaluronan and hyaluronan synthases, perhaps combined with other markers, are operable tools for diagnostics of the malignant situations. These can be investigated in tissues from resection or needle biopsy, and from body fluids like ascites, plasma, and urine. The extracellular vesicles positive for hyaluronan and hyaluronan synthases are also promising targets for the detection of malignancies (Rilla et al. 2013b, Rilla et al. 2014).
An early detection of possible changes in the content of hyaluronan during a malignant process would enable blocking the accumulation of hyaluronan, and thus progression of the disease. Hyaluronan content can fluctuate, like in the squamous cell carcinomas (Karvinen et al. 2003a, Pirinen et al. 1998) and melanomas (Siiskonen et al. 2013), where hyaluronan first increases and then decreases in the more advanced disease. Therefore, accurate timing of the interference in hyaluronan metabolism would be needed. There are several possible tools for the inhibition of the accumulation of hyaluronan. HAS gene expression can be inhibited by siRNA or microRNA (Lagendijk et al. 2011, Röck et al. 2015) and by altering DNA methylation (Kohi et al. 2016). One of the key elements in hyaluronan production is the level of the precursor sugars, and more generally the supply of glucose (Rilla et al. 2013a). Mannose inhibits hyaluronan synthesis by reducing cellular pool of UDP-‐‑GlcNAc (Jokela et al. 2008), and 4-‐‑MU affects the availability of UDP-‐‑ GlcUA acid, the hyaluronan precursor sugars, and substrates of the HAS enzymes (Kultti et al. 2009). It is also possible to use hyaluronidase to reduce the content of hyaluronan in cells and tissues.
It has been described recently, that systemically administered pegylated PH20 hyaluronidase depletes tumor hyaluronan (Jiang et al. 2012a, Thompson et al. 2010). There are ongoing phase II clinical trials with with a combination of PEGPH20 and gemcitabine for the treatment of stage IV metastatic pancreatic ductal adenocarcinoma patients
It has been described recently, that systemically administered pegylated PH20 hyaluronidase depletes tumor hyaluronan (Jiang et al. 2012a, Thompson et al. 2010). There are ongoing phase II clinical trials with with a combination of PEGPH20 and gemcitabine for the treatment of stage IV metastatic pancreatic ductal adenocarcinoma patients