Emerging Anticancer Potentials of Selenium on Osteosarcoma

Selenium is a trace element essential to humans and forms complexes with proteins, which exert physiological functions in the body. In vitro studies suggested that selenium possesses anticancer effects and may be effective against osteosarcoma. This review aims to summarise current evidence on the anticancer activity of inorganic and organic selenium on osteosarcoma. Cellular studies revealed that inorganic and organic selenium shows cytotoxicity, anti-proliferative and pro-apoptotic effects on various osteosarcoma cell lines. These actions may be mediated by oxidative stress induced by selenium compounds, leading to the activation of p53, proapoptotic proteins and caspases. Inorganic selenium is selective towards cancer cells, but can cause non-selective cell death at a high dose. This condition challenges the controlled release of selenium from biomaterials. Selenium treatment in animals inoculated with osteosarcoma reduced the tumour size, but did not eliminate the incidence of osteosarcoma. Only one study investigated the relationship between selenium and osteosarcoma in humans, but the results were inconclusive. In summary, although selenium may exert anticancer properties on osteosarcoma in experimental model systems, its effects in humans require further investigation.


Introduction
Selenium is an essential trace element in humans and is a non-metal from Group 16 of the periodic table that shares some similar physicochemical properties with sulphur. This element is rarely found in its elementary state but rather in its inorganic or organic forms in natural compounds [1,2]. The examples of inorganic selenium are selenium dioxide (SeO 2 ), selenite (SeO 3 2− ) and selenate (SeO 4 2− ), whereas those for organic selenium are selenide, diselenides, selenol or selenothiol and seleninic acid (selenium-based acid; RSeOH) [2][3][4][5]. Selenium can be obtained mainly from food, such as Brazil nuts, garlic, onion, mushroom, broccoli, meat, egg, seafood and internal organs and in negligible quantity from drinking water [6][7][8]. The main selenium species present in food are selenomethionine (SeMet), selenocysteine (SeC), selenium-methylselenocysteine (Se-MSC), SeO 3 2 and SeO 4 2− [9]. The organic forms of selenium are present in greater amount in food, and their bioavailability is higher because they are absorbed easily compared with inorganic selenium [7,8,10]. The US-recommended dietary allowance and UK-recommended reference nutrient intake of selenium for an adult are 55 and 60 µg/day, respectively [11,12]. The average dietary intake of selenium varies across countries [10,[13][14][15][16]. People from several countries, such as the UK, China, New Zealand and Finland, are traditionally deficient in selenium dietary intake [7,8,10,15,16]. Selenium is vital in various physiological processes and can incorporate into a protein as selenoproteins (SeC-containing protein), selenium-binding proteins (without SeC) and certain proteins rich in cysteine or methionine (such as SeMet), where sulphur is replaced by selenium in a nonspecific

Literature Search
The literature search was performed between 1 st -25 th March 2019 on PubMed and Scopus using the keywords 'Selenium' OR 'Selenite' OR 'Selenate' OR 'Selenide' OR 'Selenol' OR 'Organoselenium' OR 'Selenocysteine' OR 'Selenomethionine' OR 'Selenoprotein' AND 'Osteosarcoma'. We also examined the reference lists of the retrieved articles. Original research articles regarding the anticancer effects of selenium in osteosarcoma published in English were included. A total of 20 relevant studies were included in this review.
Most of the cytotoxicity and anti-proliferation effects of selenium species are time-, concentration-or cell type-dependent [49][50][51]54,55]. For example, MSeA significantly increases the level of necrosis in U-2 OS cells within a longer treatment time period (48 h) [52]. Se-MSC inhibits the growth of MG-63 cells, but enhances that of U-2 OS cells in a concentration-independent manner [54]. This finding indicates that each selenium species may exert different effects on osteosarcoma cell models. At present, the most effective selenium species against osteosarcoma is difficult to determine on the basis of cellular findings. Direct comparison is impossible due to differences in treatment time, concentration and osteosarcoma cell models used. To date, only two studies reported the IC 50 value of selenium (SeO 3 2− and MSeA) in osteosarcoma cells [50,52]. Other selenium species such as SeO 4 2− , selenide, selenol and selenoproteins have not been tested on osteosarcoma.
The knockdown of ATM (by KU55933 inhibitor) and FOXO3a (by hairpin RNA vector transfection) further suppressed MSeA-induced cytotoxicity [53]. Additionally, Werner Syndrome protein (WRN) serves as a potential combinational target, and its downregulation further increased the MSeA-induced U-2 OS cell death [52]. A comprehensive molecular study is required to elucidate the upstream molecular mechanisms of different selenium species in p53/ATM/FOXO3a/WRN signalling pathway axis. SeC also induces early mitochondrial dysfunction via mitochondrial fragmentation (from protonema to punctiform phenotype) and mitochondrial membrane potential loss, a possible result of p53 activation [55]. Additionally, proapoptotic proteins Bax, Bad and pTEN are transcriptionally upregulated by p53 activation [70,71]. In line with this finding, SeO 3 2− [50], Se-MSC [54] and SeC [55] also downregulate the antiapoptotic Bcl-2 and Bcl-XL protein and upregulate the Bax and Bad protein in osteosarcoma cells. SeO 3 2− treatment significantly increases the P53 and PTEN mRNA levels [50].

Role of Selenium in the Tumour Microenvironment
The bone tissue is made up of several cells, including osteoblasts, osteoclasts, chondrocytes, mesenchymal stem cells, blood cells and endothelial cells. The osteoclasts play an important role in the pathogenesis of osteosarcoma, especially in the progression and metastasis of the cancer [74]. Osteoclasts formation and activation are tightly regulated through both receptor-activation of nuclear factor κB (RANK) ligand (RANKL) and macrophage-colony stimulating factor [75]. The naturallyoccurring inhibitor of RANKL, osteoprotegerin (OPG) synthesised by osteoblasts, protects against bone loss by acting as a negative regulator for RANK/RANKL pathway [75]. The RANKL is secreted by osteocytes, osteoblasts and mesenchymal stem cells, as well as osteosarcoma cells [76,77]. RANKL activates its receptor, RANK, and then triggers the bone resorption by osteoclasts, which in turn promotes invasion and metastasis of osteosarcoma [74]. RANKL inhibitors such as denosumab (currently in phase II clinical trial) may be effective against osteosarcoma [78,79]. Thus far, the effects of selenium on the RANK/RANKL/OPG pathway remain inconclusive. SeO3 2− suppresses RANKLinduced osteoclastogenesis through inhibition of ROS-induced signalling pathways in mouse bone marrow-derived monocytes and RAW 264.7 cell line [48,80]. Additionally, selenium (unknown species) was reported to inhibit the transcription activity of RANKL and downregulate the mRNA levels of OPG in human osteosarcoma Saos-2 cells [81] More research is required to confirm the effects of selenium in the RANK/RANKL/OPG pathway.
Additionally, mesenchymal stem cells serve as an important modulator in the pathogenesis of osteosarcoma, wherein it supports the osteosarcoma progression and metastasis via the secretion of cytokines/growth factors [74,82]. Mesenchymal stem cells can be targeted in osteosarcoma treatment

Role of Selenium in the Tumour Microenvironment
The bone tissue is made up of several cells, including osteoblasts, osteoclasts, chondrocytes, mesenchymal stem cells, blood cells and endothelial cells. The osteoclasts play an important role in the pathogenesis of osteosarcoma, especially in the progression and metastasis of the cancer [74]. Osteoclasts formation and activation are tightly regulated through both receptor-activation of nuclear factor κB (RANK) ligand (RANKL) and macrophage-colony stimulating factor [75]. The naturally-occurring inhibitor of RANKL, osteoprotegerin (OPG) synthesised by osteoblasts, protects against bone loss by acting as a negative regulator for RANK/RANKL pathway [75]. The RANKL is secreted by osteocytes, osteoblasts and mesenchymal stem cells, as well as osteosarcoma cells [76,77]. RANKL activates its receptor, RANK, and then triggers the bone resorption by osteoclasts, which in turn promotes invasion and metastasis of osteosarcoma [74]. RANKL inhibitors such as denosumab (currently in phase II clinical trial) may be effective against osteosarcoma [78,79]. Thus far, the effects of selenium on the RANK/RANKL/OPG pathway remain inconclusive. SeO 3 2− suppresses RANKL-induced osteoclastogenesis through inhibition of ROS-induced signalling pathways in mouse bone marrow-derived monocytes and RAW 264.7 cell line [48,80]. Additionally, selenium (unknown species) was reported to inhibit the transcription activity of RANKL and downregulate the mRNA levels of OPG in human osteosarcoma Saos-2 cells [81] More research is required to confirm the effects of selenium in the RANK/RANKL/OPG pathway.
Additionally, mesenchymal stem cells serve as an important modulator in the pathogenesis of osteosarcoma, wherein it supports the osteosarcoma progression and metastasis via the secretion of cytokines/growth factors [74,82]. Mesenchymal stem cells can be targeted in osteosarcoma treatment via the restoration of cytokines/growth factors signalling, or transforming the stem cells into mature osteoblasts [82]. The effects of selenium on mesenchymal stem cells are not conclusive. SeO 3 2− was reported to protect bone marrow stromal cells against hydrogen peroxide-induced inhibition of osteoblastic differentiation through inhibiting oxidative stress and ERK activation [83]. A ruthenium (II) functional selenium nanoparticles and citrate functionalised selenium nanoparticles were reported to promote the proliferation and osteogenic differentiation of human umbilical cord mesenchymal stem cells [84]. Recently, Ahmed et al. demonstrated that 48 h of SeO 2 nanoparticles treatment (2−25 µg/mL) significantly increases the proliferation of rat adipose stem cells and bone marrow stem cells [85]. The induction of proliferation and osteogenic differentiation of mesenchymal stem cells may interfere with metastasis of osteosarcoma. However, further study is required to confirm the role of mesenchymal stem cells and their therapeutic potentials in osteosarcoma.
Regarding anticancer selectivity, selenium selectively exerts anticancer actions on osteosarcoma cells. Both inorganic and organic selenium exhibit no or marginal toxicity on several primary and non-tumourigenic cells including primary rat growth plate chondrocytes [86], primary mouse lung fibroblasts [57], primary human calvarial osteoblasts [87], human bone marrow stem cells (BMSCs) [60,88], human umbilical cord stem cells [84], human lymphocytes [61], mouse preosteoblast MC3T3-E1 cells [57,58,86], mouse 3T3-L1 preadipocytes [89], rat skeletal muscle L6 cells [50], human embryonic kidney 293 cells [50] and human osteoblast hFOB1.19 cells [56,59]. Additionally a low concentration (0.01-1 µM) and 1 h transient treatment of SeO 3 2− exert radioprotective effects on chondrocytes and osteoblasts by protecting them from 20-Gy irradiation-induced cytotoxicity [86]. SeO 3 2− (0.01-1 µM) is also not genotoxic to human blood lymphocytes and does not cause mitotic index change, chromosomal aberration or chromatid break [61]. However, SeO 3 2− (1 µM for 1 h) substantially increases the frequency of dicentric chromosomes (but not deletion and chromatid break) in γ rays-irradiated human blood lymphocytes via an unknown mechanism [61]. Additionally, selenium nanoparticles (500-6000 nm size, 0.005 mg/mL for 1 week) also induce significant cytotoxic effects and apoptosis events on mesenchymal stem cells via morphological observation [90]. These findings require further investigation to ensure the safety of selenium in daily consumption. The molecular mechanisms on the selectivity of selenium have not been studied in detail in osteosarcoma cell models. Generally, cancerous cells are more sensitive to oxidative stress compared with normal cells [91,92]. This characteristic has been exploited by current chemotherapeutic agent, such as doxorubicin, to selectively induce osteosarcoma cell death via oxidative stress [93,94]. Therefore, the selective cytotoxicity of selenium toward osteosarcoma cells may be due to oxidative stress. A high concentration of inorganic or organic selenium induced non-selective cell death on non-tumourigenic cells and osteosarcoma cells [50,52,57,62], probably due to the overwhelming ROS production.
Wang et al. reported that SeO 3 2− -doped HA nanoparticles are directly internalised into the osteosarcoma cells via nonspecific endocytosis [66]. SeO 3 2− in the vesicles is then released into the cytosol upon the degradation of HA nanoparticles during the merger of endosome and acidic lysosome in a pH-dependent manner [66]. The release of SeO 3 2− increases ROS generation, leading to osteosarcoma cell apoptosis in 6% and 10% SeHANs and 2µM SeO 3 2− , but not in HA nanoparticles control and 3% SeHAN groups [66].
SeO 3 2− -doped substrates also showed osteoinductive activities by promoting the growth of non-cancerous human osteoblast hFOB 1.19 cells [56], non-cancerous human preosteoblast MC3T3-E1 cells [58] and primary human calvarial osteoblasts [87]. SeO 3 2− -doped substrates increased the growth of mouse primary lung fibroblasts [57] and BMSCs [60] under a similar treatment. SeO 3 2− -doped titanium substrate [56], selenium (SeO 3 2− )-doped calcium phosphate coating [58] and SeO 3 2− coated-PLLA nanocomposites (SeNP-PLLA) [59] promote bone-forming activities, as evidenced by the increase in alkaline phosphatase activities and extracellular calcium deposition. Furthermore, SeO 3 2− -substituted HA exhibits osteoinductive activity on partially differentiated MC3T3-E1 cells by increasing the mRNA expression of bone γ-carboxyglutamate protein 3 (BGLAP3; osteocalcin-related protein) [57]. The cytotoxicity of SeO 3 2− -doped substrates and nanoparticles relies on its pH-dependent release into the medium and a high concentration of selenium is suggested to induce non-selective cytotoxicity [57,59,62]. A high SeO 3 2− content (3.0 wt%) induces non-cancerous MC3T3-E1 cell death with abnormal morphological changes as early as 24 h of treatment [57]. SeO 3 2− -containing HA (SeHA) 3.0 wt% treatment also decreases BGLAP3 expression due to its cytotoxic effect [57]. SeNP-PLLA reduces the viability of human osteosarcoma MG-63 cells and non-cancerous hFOB cells, though the effects are more selective on former than on the latter [59]. Additionally, the media after overnight incubation with selenium-containing hydroxyapatite/alginate (SeHA/ALG) composite microgranules are cytotoxic to human osteosarcoma Saos-2 cells and non-cancerous hFOB 1.19 cells with almost 90% reduction on viability [62]. According to the authors, this non-selective cytotoxicity of SeHA/ALG microgranules may be due to the rapid release and accumulation of selenium in the culture media [62]. As previously discussed, excessive ROS production is one of the mechanisms of selenium in inducing unspecific cell death [95,96]. The in vitro studies of selenium and its derivatives on osteosarcoma cells are summarised in Table 1.

Evidence from Animal and Human Studies
Several animal studies were conducted to determine the effects of selenium on osteosarcoma [49,51,55,66,97,98]. Bierke and Svedenstal initiated the study on the effects of inorganic selenium in radioactive strontium ( 90 Sr)-induced osteosarcoma mice [98]. Vitamin E (α-tocopherol acetate) with or without SeO 3 2− (10 µg) was administrated intraperitoneally to the mice with osteosarcoma every 2 weeks from day 105 after the 90 Sr exposure until 14 months [98]. The same injection was continued after 14 months but was changed to 30-day intervals for the rest of the life span [98]. Oestrogen (polyestradiol phosphate) was administered in certain groups during 30, 60 and 90 days after the 90 Sr exposure. The results showed no significant difference in osteosarcoma tumour incidence after treatment with vitamin E with or without SeO 3 2− . Post-exposure of antioxidants, including selenium and vitamin E, is not beneficial to prevent the development of 90 Sr-induced osteosarcoma. The combined treatment somehow hastened the onset of osteosarcoma regardless of oestrogen induction [98]. Several studies of inorganic and organic selenium were conducted using osteosarcoma xenograft animal models [49,51,55,66,97]. Some in vitro human osteosarcoma cell lines were used, including human osteosarcoma KOS [49], U-2 OS [51], MG-63 [55] and SOSP-9607 cells [97]. Hiraoka et al. investigated the effect of SeO 2 in BALB/c nude mice implanted with osteosarcoma xenograft [49]. The back of nude mice was subcutaneously inoculated with KOS cells, and the mice were fed with SeO 2− containing drinking water (0.2 and 2 µg/mL) until day 44 after inoculation [49]. SeO 2 dose-dependently decreased the tumour volume to 2.5-fold lower compared with that of control [49]. SeO 2 also induces apoptosis in xenograft tumour tissues without affecting visceral organs. However, this compound does not prevent tumour incidence [49].
For the organic selenium, Wang et al. reported that daily oral administration of Se-Poly isolated from Ziyang green tea (200 and 400 mg/kg) for 28 days significantly reduces tumour volume and weight of U-2 OS xenograft in BALB/c nude mice [51]. Similar to that used by Hiraoka et al., the Se-Poly is non-toxic to nude mice where it does not affect the body weight or cause any lethal incidence [51]. Wang et al. showed that the intravenous injection of SeC (5 and 10 mg/kg; every other day for 2 weeks) significantly and dose-dependently reduces the osteosarcoma MG-63 tumour xenograft volume and weight in nude mice [55]. Mechanistically, SeC induces p53 phosphorylation (Ser 15) and caspase-3 activation in tumour xenograft in a dose-dependent manner [55]. SeC also significantly suppresses cell proliferation and angiogenesis of tumour xenografts as evidenced by the downregulation of Ki-67 and CD-34 biomarkers [55]. Similar to other selenium species, SeC does not affect the body weight, suggesting the lack of systemic toxicity in nude mice [55].
The anticancer effect of SeO 3 2− -doped substrates on osteosarcoma xenograft animal models has also been reported [66,97]. Wang et al. revealed that intratumoural injection of SeHAN for 30 days significantly reduces the tumour volume of intrafemoral human osteosarcoma SOSP-9607 xenograft in nude mice [97]. SeHAN also inhibits osteosarcoma tumour metastasis into the lung and protects other vital organs, such as liver, kidney and cardiac muscles from osteosarcoma-mediated damages [97]. The anticancer effect of SeHAN is mediated by the suppression of tumour invasion but not proliferation, as indicated by the reduction of matrix metallopeptidase-9 (MMP-9; invasion marker) and the lack of change in Ki-67 level (mitotic marker) [97]. Intratumour 10% SeHAN injection (every 3 days for 30 days) significantly reduces the tumour size, weight and volume of osteosarcoma MNNG/HOS tumour xenograft in BALB/c nude mice [66]. SeHAN induces oxidative DNA damage, which will hypothetically trigger the subsequent activation of caspases and the apoptosis in tumour tissues [66]. In parallel with in vitro studies, the anticancer effect of SeHAN is related to the release of SeO 3 2− ions into aqueous solution [97]. SeHAN is completely degraded within tumour tissues with less calcium aggregation and blood vessel vascularization upon histological analysis [66]. Similar to findings in other selenium studies, SeHAN does not cause any significant systemic toxicity in nude mice and has no effect on body weight, lethality, haematological indices and serum biochemical profile, including aspartate aminotransferase, blood urea nitrogen, creatinine and lactate dehydrogenase levels [66,97]. Furthermore, no pathological change has been detected in the liver of nude mice that received SeHAN treatment [66]. One human study was conducted by Huang et al. to identify the relationship between selenium level and osteosarcoma disease [54]. No significant difference was found in the plasma selenium levels between patients with and without osteosarcoma [54]. Selenium levels were significantly higher in osteosarcoma tissues compared with those in normal bone tissues among patients with osteosarcoma [54]. However, further investigation is needed to identify the role of high selenium levels in osteosarcoma tissues. To date, no human study has revealed the beneficial effect of selenium supplementation in preventing osteosarcoma and no clinical trial is being conducted to evaluate the therapeutic effect of selenium in patients with osteosarcoma.
Several epidemiological studies and clinical trials were conducted to determine the relationship between selenium intake and the risk of other solid cancers; however, the findings are heterogeneous [5,34,[99][100][101][102][103][104][105]. The NPC and Linxian Nutritional Intervention Trials reported that selenium intake reduces the risk of lung, colorectal and prostate cancer and mortality related to oesophageal and gastric cancer [33][34][35][36]. The Selenium and Vitamin E Cancer Prevention Trial (SELECT) and Selenium and Celecoxib (Sel/Cel) Trial showed that selenium does not reduce the risk of prostate [33,101] and colorectal cancer [106]. Additionally, a recent systematic review and meta-analysis by Vinceti et al. concluded that selenium supplementation does not reduce the overall cancer incidence or mortality [103,107]. The contradicting findings may be confounded by experimental biases, chemical forms of selenium, basal selenium status, nutritional status and lifestyle factors of the subjects [35,107]. Vinceti et al. emphasised on randomised controlled clinical trials of selenium on various cancers [103]. The osteosarcoma is not involved as currently the relevant clinical trial is not available. Further studies are required to confirm the relationship of selenium and cancer risk, especially on osteosarcoma. Table 2 summarises the effects of selenium in osteosarcoma in vivo.

Conclusions
Selenium and selenium-containing proteins possess potential anticancer activity, as evidenced from cellular and animal studies. Evidence was provided that the underlying mechanism for their anticancer effects involves increased intracellular ROS generation, leading to activation of the p53/ATM/FOXO3a pathway, proapoptotic proteins and caspases. This results in cytotoxicity, antiproliferative and proapoptotic effects of selenium compounds on osteosarcoma. The action of selenium can be selective on osteosarcoma cells without affecting adjacent normal cells, such as chondrocytes and fibroblasts. Moreover, this element is osteogenic for normal bone tissues. Developing SeO 3 2− -doped bone biomaterials, which release this essential element in a controlled manner for therapeutic purposes, remains a challenge because SeO 3 2− at a high concentration exerts non-specific cell death via oxidative stress. Another challenge is identifying the most active selenium form and dose that exert the best anti-osteosarcoma effects in the various cellular models used in previous experiments. In vivo studies showed that selenium can reduce the tumour size in animals with osteosarcoma, but does not diminish the incidence of the tumour. However, the effects of selenium on osteosarcoma have not been validated in a clinical trial. A paucity of epidemiological data revealed a relationship between selenium intake and osteosarcoma. Thus, the role of inorganic and/or organic selenium on osteosarcoma tumour formation remains unknown. These gaps in our knowledge should be filled by researchers to determine the role of selenium in preventing or treating osteosarcoma. Acknowledgments: The authors expressed gratitude to Ministry of Education, Malaysia and Universiti Kebangsaan Malaysia for their generous support.

Conflicts of Interest:
The authors declare no conflict of interest.