In this study, we addressed the question if the level of expression of HAS1-3, HYAL1-5 and HYALP1 has an influence on the survival of ovarian cancer patients. We hypothesized that the expression of HAS2 could have an effect on the success of therapy and sphere formation capability and cohesion of the tumour cells.
Our TNMplot analysis of over 700 ovarian cancer specimens revealed that HAS1, HYAL1 and HYAL4 mRNA expression is significantly upregulated, whereas HAS2, HYAL2 and HYAL3 mRNA expression is significantly downregulated in ovarian cancer tissue compared to controls. These data underscore the clinicopathological significance of the HA biosynthetic and degradative system in ovarian cancer. Of particular relevance, our experiments highlighted that the expression of HAS1 and especially the expression of HAS2 correlates with poorer survival of ovarian cancer patients. Considering the strong association between ovarian cancer aggressiveness and HA deposition (Anttila et al.
2000), we knocked down HAS2 to evaluate if its expression correlates with ovarian cancer cell lines' aggressiveness. HAS2 knockdown has little effect on the expression of HASs and HYALs, with a small increase in HYAL3 expression. Interestingly, HAS2 appears to have an influence on cell cohesion capacity, which was significantly lower with HAS2 deletion, as shown by smaller spheroids in the hanging drop method. Moreover, cell viability was significantly reduced by HAS2 knockdown, but it was only moderately reduced in response to chemotherapy in both controls and HAS2 knockdown cells. All in all HAS2 expression goes along with lower OS and PFS of ovarian cancer patients, and, indeed, our experiments suggest that HAS2 seems to be important for stimulating tumour and cell growth and stability and it encourages cell viability.
Influence of hyaluronan synthases HAS1-3 on ovarian cancer and patient’s survival
First, we compared the survival time of patients with low expression of HA-related enzymes with those with high expression by the use of the Kaplan–Meier Plotter. We found that HAS1 and HAS2 expression correlated with worse survival of ovarian cancer patients. However, in some subgroups, the data must be interpreted with caution, especially when the number of cases was below 50. This applies to HAS1 and HAS2 to data of OS of patients with endometrioid ovarian cancer and of PFS of patients in grade 1. For HAS3 it concerns OS and PFS of patients with endometrioid ovarian cancer and OS and PFS of patients in grading I. Due to the small number of cases, false tendencies could arise.
We analysed a panel of four different ovarian cancer cell lines (i.e., SW 626, SKOV3, Caov-3 and PA-1) and measured the expression levels of several HA-related genes, namely HAS1-3 and HYAL2-3. As SW 626 and SKOV3 had comparable gene expression profiles and the highest relative expression of HAS2, these two cell lines were chosen for the subsequent characterizations.
The influence of HAS1 on the survival of ovarian cancer patients has not yet been studied in detail. However, it has been shown that high HAS1 expression is associated with poor patient survival for ovarian cancer, colon cancer, Waldenström´s macroglobulinemia and multiple myeloma (Siiskonen et al.
2015), and its downregulation correlates with lower growth and development of bladder cancer due to lower hyaluronan production (Golshani et al.
2008). Our results from the Kaplan Meier-Plotter confirm that, indeed, high HAS1 expression leads to lower survival of ovarian cancer patients. One assumption is that high HAS1 expression ends in higher HA production. This could lead to greater tumour growth. This has already been shown for the ability of prostate cancer to metastasise to the bone marrow (Simpson et al.
2002,
2001).
Among all HASs, HAS2 is the most important one implied in both physiological and pathological conditions, including cancer (Camenisch et al.
2000; Passi et al.
2019). Our results report that the higher the HAS2 expression, the lower the OS and PFS for ovarian cancer patients. This finding is in accordance with the literature stating that high HAS2 expression also leads to short OS in pancreatic cancer patients (Yu et al.
2021). Moreover, elevated HAS2 expression is also found in breast cancer cell lines compared to normal breast tissue, and its knockdown leads to decreased proliferation and increased apoptosis (Li et al.
2015). Finally, a correlation between high coexpression of HAS2 and HYAL1 and strong tumour growth and angiogenesis was observed for prostate carcinoma (Simpson
2006).
Our results on the ability of SKOV3 ovarian cancer cell lines to form spheroids confirmed that, effectively, HAS2 could be involved in the reduction of tumour aggressiveness, as we observed a significantly poorer cell cohesion in HAS2 knockdown cells with respect to control cells. A possible explanation could be related to the significant reduction of HA synthesis as a result of HAS2 silencing, which we demonstrated via the particle exclusion assay.
This hypothesis is supported by a study reporting that high expression of HAS and HA correlates with higher metastasization and invasiveness in different tumour types (Jojovic et al.
2002). Furthermore, it has also been shown for ovarian clear cell carcinomas that tumour cell growth is inhibited by low levels of HA (Kato et al.
2016). Concomitantly, we also found significantly reduced cell viability in HAS2 knockdown cells, even without prior chemotherapy treatment. This could be an indication that the cells grow worse due to a lack of HA production. This also fits with the statements of Okuda and colleagues that high HAS2 expression in breast cancer cells correlates with increased growth and metastasis than control cells. Furthermore, it indicates a lower OS time for patients (Okuda et al.
2012). A caveat is associated with our transient transfection approach, as our qPCR analysis revealed that the knockdown was not stable for extended periods. Particularly for the spheroid formation the assay times exceeded the knockdown duration at later timepoints, suggesting that heterogeneity in the cell population may have contributed to the phenotype and acted as a confounder. Nevertheless, this heterogeneity has resulted from HAS2 knockdown, highlighting a causative relation to HAS2.
Interestingly, HAS2 knockdown and control cells both showed a moderate response to the different chemotherapy treatments. The therapeutic effect of chemotherapy for ovarian cancer can be low due to their intrinsic chemotherapy resistance (Ricciardelli et al.
2013). We demonstrated that this is not significantly changed by HAS2 knockdown, but the viability of HAS2 knockdown cells is fundamentally poorer. Indeed, more successful therapy for chemotherapy-resistant serous ovarian cancer cells seems to be possible through the combination of carboplatin and 4-methylumbelliferone (4-MU). This inhibits HA production, cell survival and spheroid formation in these cells. This is therapeutically significant, as increased HAS2 and HAS3 expression was observed in chemotherapy-resistant ovarian cancer cells (Lokman et al.
2019). In addition, Bourguignon et al. showed that chemotherapy resistance in ovarian and breast cancer cells arises via the HA–CD44 interaction by inducing the binding of Ankyrin to MDR1 (Bourguignon et al.
2008, p 44). Ricciardelli et al. also showed that the HA–CD44 signalling pathway could be an important approach for treating the development of resistance to carboplatin in ovarian cancer patients. Indeed, after carboplatin treatment, the expression of HAS2, HAS3, ABCC2 and HA secretion increased. A high HA level in turn correlated with higher survival of CD44 positive ovarian cancer cells. HA thus appears to be a relevant factor in relation to the high survival of tumour cells after carboplatin treatment (Ricciardelli et al.
2013). In order to be able to treat ovarian cancer optimally, further research is needed in this area.
The fact that low expression of HAS2 is correlated with lower tumour cell growth is strengthened by the results of String analysis that HAS1-3 interact close with UGDH. UGDH plays a role in glycosaminoglycan synthesis and therefore is also important in relation to ECM and the synthesis of HA (Egger et al.
2011). CD44 is a non-kinase transmembrane proteoglycan, which mainly ligand is HA. RHAMM also has HA as its main ligand. The binding of HA to CD44 or RHAMM allows intracellular adapter molecules to bind. This promotes cell adhesion, cell migration and cell proliferation (Chen et al.
2018; Savani et al.
2001). Notably, we also found an interaction between HAS1-2 and VCAN, which is an essential proteoglycan supporting growth, survival, angiogenesis, metastasis, migration and invasion of tumour cells (Li et al.
2020; Fujii et al.
2015).
In a study from 2003, it was found that HAS3 is overexpressed in metastatic tissue of colon carcinoma. Furthermore HAS3 knockdown showed inhibition in the growth of both colon cancer and oesophageal squamous cell carcinoma cell lines (Bullard et al.
2003; Twarock et al.
2011). However, we did not find a significant correlation between HAS3 expression and ovarian cancer patient survival. Furthermore, there was no significant connection between HAS2 knockdown and HAS3 expression although the string analysis showed a strong correlation (Fig.
5A). In summary, HAS3 did not appear to play a central role in the survival of ovarian cancer patients in our study.
A deep investigation into the molecular mechanism by which elevated HA deposition drives ovarian cancer aggressiveness is essential to try and develop an efficient targeted therapy aimed at lowering the overall HA amount in the tumor stroma. At the time being, a few molecules have been investigated to target and block HA synthesis and/or signaling. 4-MU is a well known inhibitor of HA synthesis, which is already used in the clinics for the treatment of biliary spasms (Abate et al.
2001). Its potential beneficial effect in the treatment of several cancers such as breast, pancreatic and skin cancers, have been investigated—all studies reports that 4-MU can inhibit the proliferation, migration, and invasion of multiple cancer cells, both in vitro and in vivo (Urakawa et al.
2012; Edward et al.
2010; Hajime et al.
2007; Morohashi et al.
2006). However, the potential long-term consequences of 4-MU administrations are still under debate, as generalised inhibition of HA synthesis could lead to diverse side effect, among which the worsening of atherosclerosis observed in Apo-E deficient mice (Nagy et al.
2010).
A peptide-based aproach targeting CD44 and RHAMM is ongoing (Hauser-Kawaguchi et al.
2019; Weng et al.
2022). Particularly, the A6 eight-aminoacid peptide binds to CD44, enhancing HA binding and the downstream phosphorylation of CD44 signalling components, such as focal adhesion kinase (FAK) and Mitogen-activated protein kinase kinase (MEK). Even though behaving as a CD44 agonist, A6 treatment reduced the migration of cancer cells in vitro and demonstrated increased progression-free survival in patients with ovarian cancer with a positive safety profile (Gold et al.
2012). Preclinical studies using anti-CD44 antibodies to treat cancer have shown promising results, yet failing the clinical trials examining the safety and efficacy of anti-CD44 therapies (Xu et al.
2020).
Overall, our results and already known publications indicate that increased HAS synthesis and consequently increased HA production led to increased tumour cell growth and reduced survival time, respectively. In contrast to this is the observation that high HA production is associated with lower adhesion to the peritoneum in ovarian cancer cells and therefore seems to be protective with respect to metastasis to the peritoneum. (Tamada et al.
2012) Furthermore, HA could also be used in tumour therapy for ovarian cancer patients in the form of cross-linked HA gel. This gel seems to stop further tumour growth by inhibiting the migration and proliferation of cells, as well as reducing the occurrence of adhesions (Pang et al.
2018). With regard to patients with chemotherapy-induced primary ovarian insufficiency, it has been shown in experiments with rats that HA appears to have a preventive effect in these patients due to the promotion of granulosa cells and upregulation of PGRMC1 expression (Zhao et al.
2015). These results show that HA seems to have both, positive and negative effects on ovarian cancer progression and ovarian diseases.
Influence of HYAL1-5 and HYALP1 on ovarian cancer and the patient’s survival
With the use of the Kaplan–Meier-Plotter we could show that patients with high HYAL2 and HYAL3 expression had better survival. HYAL4 had a positive influence on patients in staging III + IV for OS and PH20 for patients with an endometrioid carcinoma for PFS. Referred to HYAL1 and HYALP1 no correlation was found. In agreement with that, it has been reported that HYAL1 is upregulated in clear cell and mucinous ovarian cancer cells, but not in serous and endometrioid ones (Yoffou et al.
2011). Nevertheless, another group found significantly lower levels of HYAL1 in serous ovarian cancer cells. They did not find a changed regulation of HAS1-3 (Nykopp et al.
2009). With regard to the HYALs, we found most significances for HYAL2 and HYAL3. Therefore, only these two HYALs were included in the more detailed laboratory investigation. In particular, HAS2 knockdown leads to a significant upregulation of HYAL3 in SKOV3 and downregulation in SW 626 cell line. HYAL2, instead, did not show a significant correlation. The significance of the upregulation and downregulation of HYAL3 in the two cell lines should be viewed with caution—a fold change of 1.2381 and 0.8384, respectively, represents only a small change in HYAL3 mRNA levels. Whether this is a side effect or a clinically relevant result cannot be said on the basis of the qPCR results. This connection would have to be analysed in more detail to be able to draw conclusions from it. Until then, no significant correlation was found in previous studies. In contrast to our qPCR results, it was reported for breast cancer that HAS2 knockdown in Hs578T cancer cells leads to an upregulation of HAS1, HAS3 and HYAL1. Furthermore produced HA was smaller and the migration of cancer cells was slower (Li et al.
2007). Besides this, it has been reported that HAS2 knockdown in breast cancer cells leads to a downregulation of HYAL2 and CD44 (Udabage et al.
2005). One possibility for an optimized future therapy of ovarian cancer could be the treatment with Irinotecan conjugated to HA, which has been tested in mice. This could make a regionally specified therapy for ovarian cancer cells possible (Montagner et al.
2015).
The ability of HYALs in generating small HA fragments that could have a protumorigenic role makes them an appealing choice for pharmacological targeting in chemotherapy. Interestingly, several clinical trials are ongoing to study the combined use of recombinant HYALs, such as PEGPH20 (PEGylated recombinant human hyaluronidase PH20), to sensitise solid tumours to conventional chemotherapy. These recombinant HYALs have shown to reduce the HA amount in tumour stroma, thus reducing interstitial pressure and allowing the drugs to reach tumour cells and induce cell death (Provenzano et al.
2012). However, several concerns are arising from the potential adverse effects that residual small HA fragments produced by the recombinant HYALs enzymatic activity on surviving tumour cell proliferation, growth and motility.