Polyoxometalates as Potential Next‐Generation Metallodrugs in the Combat Against Cancer

Abstract Polyoxometalates (POMs) are an emerging class of inorganic metal oxides, which over the last decades demonstrated promising biological activities by the virtue of their great diversity in structures and properties. They possess high potential for the inhibition of various tumor types; however, their unspecific interactions with biomolecules and toxicity impede their clinical usage. The current focus of the field of biologically active POMs lies on organically functionalized and POM‐based nanocomposite structures as these hybrids show enhanced anticancer activity and significantly reduced toxicity towards normal cells in comparison to unmodified POMs. Although the antitumor activity of POMs is well documented, their mechanisms of action are still not well understood. In this Review, an overview is given of the cytotoxic effects of POMs with a special focus on POM‐based hybrid and nanocomposite structures. Furthermore, we aim to provide proposed mode of actions and to identify molecular targets. POMs are expected to develop into the next generation of anticancer drugs that selectively target cancer cells while sparing healthy cells.


Introduction
Cancer is am alignant disease in which abnormal cells divide in an uncontrolled way,l eading to the formation of as olid mass referred to as tumor or to blood cancers. [1] According to the WHO (http://www.who.int/cancer/en/), cancer is one of the leading causes of mortality worldwide and was responsible for 8.8 million deaths in 2015. Them ost prominent cytotoxic drug class is cisplatin (CDDP), which is still one of the most applied chemotherapeutic agents in clinics.B esides CDDP,aseries of other drugs have shown their anticancer potential by temporarily alleviating symptoms,p rolonging the lifespan of patients and in rare cases even curing the cancer. [2] However,a ll of them suffer from major disadvantages such as severe side effects owing to lack of selectivity,l ow efficiencya gainst some cancer types,a nd low bioavailability.T herefore,t here is still the quest for alternative drugs selectively incapacitating cancer cells without severely damaging normal cells.
In this context, polyoxometalates (POMs), which are described as clusters of transition metal (W,M o, V, Nb) and oxygen atoms,h ave in the last decades been found to be promising anticancer drug candidates.P OMs exhibit an overwhelming diversity in size and structure ( Figure 1) with outstanding properties and functions. [3] They have been studied vigorously and are used in aw ide range of applications such as catalysis, [4] nanoscience, [5] macromolecular crystallography, [6][7][8] and medicine. [9,10] Thea nticancer activity of POMs was first mentioned in 1965, when Mukherjee described the in vivo application of amixture named PTMC, acombination of H 3 [PW 12 O 40 ], H 3 [PMo 12 O 40 ]and caffeine,on patients suffering from gastrointestinal cancer. [11] Despite leading to the complete disappearance of tumors in four patients,P TMC was not subjected to further clinical studies. Years later,i n1 974, Jasmin et al. described the inhibitory effect of (NH 4 ) 17 Na[NaSb 9 W 21 O 86 ]a gainst sarcoma virus-induced tumors. [12] Since then, considerable attention was paid to the development of biologically active POMs. [9,13] In this regard, especially Yamase and co-workers performed some important pioneering work by synthesizing [NH 3 Pr i ] 6 [Mo 7 O 24 ]( PM-8) that has been evaluated for its in vitro and in vivo anticancer activities. [14,15] PM-8 was highly efficient in vivo by suppressing the tumor growth in different mice models being partially more active than approved drugs such as 5-fluorouracil (5-FU) and nimustine. [16] In 1991, Fujita et al. tested 50 POMs for their anticancer activity,a mong them different variants of PM-8, and found only four promising compounds,namely PM-8, the reduced form of PM-8 (PM-17), [NH 3 Pr i ] 6 26), and Na 5 [IMo 6 O 24 ]( PM-32). [17] Yamase noticed that the reduced form of PM-8, namely PM-17, which was later identified as [H 2 Mo V 12 O 28 (OH) 12 (Mo VI O 3 ) 4 ] 6À , [18] was highly toxic in comparison to PM-8. Therefore,h ep roposed am echanism for the anticancer activity of PM-8 which involves its reduction to PM-17 and its re-oxidation in which course the tumor cells are reduced and thus killed. [19] The proposed mechanism seemed reasonable as PM-8 can be biologically reduced by FMN,anelectron carrier responsible for the electron transport from NADH to coenzyme Q. [19,20] Polyoxometalates (POMs) are an emerging class of inorganic metal oxides,whichover the last decades demonstrated promising biological activities by the virtue of their great diversity in structures and properties.T hey possess high potential for the inhibition of various tumor types;however,t heir unspecific interactions with biomolecules and toxicity impede their clinical usage.T he current focus of the field of biologically active POMs lies on organically functionalized and POM-based nanocomposite structures as these hybrids show enhanced anticancer activity and significantly reduced toxicity towards normal cells in comparison to unmodified POMs.A lthough the antitumor activity of POMs is well documented, their mechanisms of action are still not well understood. In this Review,anoverview is given of the cytotoxic effects of POMs with aspecial focus on POM-based hybrid and nanocomposite structures.Furthermore,weaim to provide proposed mode of actions and to identify molecular targets.POMs are expected to develop into the next generation of anticancer drugs that selectively target cancer cells while sparing healthy cells.

Summary and Outlook 2996
This process is coupled with the generation of ATPa nd therefore the proposed redox-cyclem echanism is based on the inhibition of ATPf ormation. Despite the success story of PM-8 and others,p urely inorganic POMs mostly suffer from high and long-term toxicity,i mpeding their clinical application. [21] Therefore,t he research focus is switching from inorganic POMs to POMbased organic-inorganic hybrids as the functionalization and/ or encapsulation of POMs with organic moieties not only reduced the toxicity of the POM in most cases but also increased its anticancer activity.Inthis Review,wereport on the development and recent advances in the synthesis of Aleksandar Bijelic is currently ap ostdoctoral research fellow at the Department of Biophysical Chemistry at the University of Vienna (Austria). He received his master's degree in Molecular Life Science in 2012 from the University of Erlangen-Nürnberg (Germany) and aP h.D. in Chemistryf rom the University of Vienna in 2016 (Austria). His research interests include the X-ray structure analysis of diverse metalloenzymes and the investigation of polyoxometalateprotein interactions.
Manuel Aureliano obtained his doctorate at the University of Coimbra. He is an Associate Professor of Biochemistry (Habilitation) and was Director of the Biochemistry degree at the University of Algarve, Faro, Portugal (1998. He has published about 80 peer-reviewed journal articles, reviews and book chapters. He is editor of the book "Vanadium Biochemistry". His research topics include the role of decavanadate in biology,p roteins as POM targets, vanadium and diabetes, and antioxidants: toxic and/or beneficial effects. POMs with proven antiproliferative activity.O ur focus lies especially on organically modified POMs and POM-based nanocomposites and their potential application as chemotherapeutic agents.T his Review provides ac omprehensive overview of proposed molecular targets and mode of actions. Moreover,w ep resent an outlook discussing why POMs, despite their disadvantages,are still potential next-generation metallodrugs in the combat against cancer.

Anticancer Activity of Purely Inorganic Polyoxometalates
After the auspicious results from PM-8 and other polyoxomolybdates (POMos;S upporting Information, Table S1), interest in this type of compounds peaked, leading to av ast number of biologically active POMs (Supporting Information, Tables S1-S3). In 2005, Liu et al. tested the antiproliferative activity of in total 21 POMs against KB cells (HeLa derived human oral carcinoma). Ther esults revealed that the structure has am ajor impact on the antitumoral activity as the polyoxovanadate (POV) K 7 [NiV 13 [22] It is similar with the type of addenda atom as POVs were slightly more active than POMos followed by the least active polyoxotungstates (POTs). Mixed-type POMs showed increased activity when aWatom was substituted by aVor Mo atom. This is in accordance to the proposed redoxbased mechanism of Yamase as the oxidation power follows the sequence POVs > POMos > POTs. [23] Thep romising antitumor activity of POVs is not surprising, as the most prominent representative,d ecavanadate [V 10 O 28 ] 6À (Figure 1e), is known to be highly bioactive, including excellent antitumor activity, [24][25][26] owing to its high affinity towards important enzymes such as kinases, [27] actin, [28] and P-type ATPases. [29,30] 2 ], of which antiproliferative activity against human ovarian (SK-OV-3, IC 50 = 33.3 mgmL À1 ), hepatocellular (SSMC-7721, IC 50 = 20.7 mgmL À1 )a nd liver cancer cells (Hep-G2, IC 50 = 9.3 mgmL À1 )w as significantly higher than that of the parent POM (IC 50 = 203.6, 219.6 and 214.0 mgmL À1 ). [32] Thei ncrease in anticancer activity was explained by the Co II induced changes in polarity,acidity,and redox properties of the POM unit facilitating the cell penetration and target interactions of synergistic Co II -POM systems.
As eries of Krebs-type tungstobismuthates (Figure 1m), (H 2 im) 2 [(W 0. 5 Information,  Table S2). [33] TheN i II containing POM was the most active compound (IC 50 = 25.6 mm), followed by the Mn II and Zn II containing clusters (IC 50 = 30.1 and 32.3 mm), whereas the Co II -POM (IC 50 = 37.3 mm)was the least active one.All tested POMs performed significantly better than CDDP (IC 50 = 66.1 mm)a gainst Hep-G2 cells.H owever, activity tests using human hepatocyte (QSG) cells revealed also as ignificant activity against normal cells (IC 50 = 32.4 mm (Ni II ), 43.2 mm (Mn II ), 49.7 mm (Zn II ), and 38.5 mm (Co II )). Based on this, the Zn II containing Krebs-POM was the most selective and thus clinically most suitable compound. [33] TheM n II [34] and Co II [35] containing Krebs structures were also tested on other cancer cell lines performing partially better than clinically approved drugs and were shown to induce apoptosis (Supporting Information, Table S2).
Another POM inducing evidently apoptosis is the Cu II containing double Keggin POT (Figure 1g)K 7 Na 3 [Cu 4 (-H 2 O) 2 (PW 9 O 34 ) 2 ], which showed inhibitory effects against both human (MG-63) and rat (UMR-106) bone osteosarcoma (IC 50 = 22 and 81 mm). [36] ThePOM increased the intracellular level of reactive oxygen species (ROS) while reducing that of the ROS-scavenger glutathione (GSH) via GSH-POM interactions leading to the dissipation of the mitochondrial membrane potential and finally to apoptosis.M oreover,t he cytotoxic effect of the POT on MG-63 was higher than that of CDDP (IC 50 = 43 mm).
All of the reported inorganic POMs exhibiting antiproliferative activity are summarized in the Supporting Information, Tables S1-S3.

Anticancer Activity of Inorganic-Organic Hybrid Polyoxometalates
Thefunctionalization of POMs with organic groups is the main focus of the field of bioactive POMs,aspurely inorganic POMs generally exhibit toxic side effects and limited cell penetration owing to their surface characteristics.T he introduction of organic moieties into the POM framework can change its surface,c harge,p olarity,a nd redox properties, leading to acompletely new compound with reduced toxicity and increased cell penetration ability.O rganically modified POMs are in general more stable in aqueous solutions and, depending on the attached functionality,their interaction with biological targets are enhanced and more specific.

Organometallo-Substituted Polyoxometalates
Many organometallic molecules possess promising anticancer activities and are thus ag ood choice for the hybridization with POMs to develop hybrids with enhanced biological activity. [37] Aseries of studies have been published describing mainly the in vitro antiproliferative activity of organometallo-substituted POTs containing organotin RSn (R = C 4 H 7 O 2 ,C 5 H 9 O 2 ,a nd NC 3 H 4 )o rm etal-cyclopentadienyl CpM n+ groups (Cp = h 5 -C 5 H 5 ,M= Ti IV ,Z r IV ,V IV ,F e II ) against human cervical (HeLa) and liver (SSMC-7721) cancer cells (Supporting Information, Table S4). [38][39][40][41][42][43][44]  Sn), sandwich-Keggin were superior to Keggin structures,a nd with increasing RSn content the antitumor activity was enhanced. Forv ery closely related compounds the antitumor activity correlates with their redox potential, that is,t he higher the redox potential, the higher the cytotoxicity.Ingeneral, CpM n+ containing POMs exhibited higher antitumor activities than the RSn structures (Supporting Information, Table S4). [45,46] Thei nfluence of the metal in the CpM n+ group on the antitumor activity is cancer cell specific as given cells are more or less susceptible to ac ertain metal. K 6 H[(h 5 -C 5 H 5 Ti)CoW 11 O 39 ]w as applied orally to human liver (SSMC-7721), leukemia (HL-60), and colon cancer (HLC) bearing mice (for 10 d) and significantly decreased the growth of all tumors exhibiting inhibitory rates of 41.9 %( dose = 15 mg kg À1 ), 50.0 %( 100 mg kg À1 )a nd 48.9 %( 100 mg kg À1 ), respectively. [47] ThePOT performed better than the clinically approved drug cyclophosphamide (CP) against SMMC-7721 (inhibitory rate = 37.2 %a t3 6.4 mg kg À1 )b ut was less active than 5-FU against the other cell lines (56-57.4 %a t1 5-30 mg kg À1 ), however, it was by far the least toxic compound. [47] Theo rganotin substituted Keggin K 3 H[{(n-Bu)Sn( OH)} 3 GeW 9 O 34 ]i nhibited the tumor growth in the H22 (murine liver cancer) mice model by 62.5 %after 14 d(dose = 300 mg kg À1 ). [48] Theactivity was significantly lower than that of CP (inhibitory rate = 95.5 %), however, the drug also impaired the growth of mice,which is indicative of toxic side effects,w hereas the POM did not. Theh ybrid showed also promising in vitro activity against aseries of other cancer cells (Supporting Information, Table S4).
Theantitumor activity of other organometallo substituted POMs is listed in the Supporting Information,  O 40 ] 5À led to POT-5-FU hybrids that were more active against the tumor cell lines HeLa, Hep-G2, and SMMC-7721 than 5-FU alone,i ndicating synergistic effects (Supporting Information, Table S5). [49][50][51] Furthermore,t he hybrids were less toxic against normal human embryonic kidney cells (HEK-293) than the free drug. The[PW 12 O 40 ]-5-FU and [SiW 12 O 40 ]-5-FU systems were later extended by the introduction of rare earth metals (Dy,E u, Er, Gd, La, Nd, Pr, Sm, Y), which in most cases did not only lead to an increase in activity (against HeLa and Hep-G2) but also to enhanced selectivity (Supporting Information, Table S5). [51][52][53][54][55] However,POM-rare earth metal complexes (without 5-FU) like K 11 [L(PW 11 O 39 ) 2 ](L= Dy,Er, Gd, La, Y) were remarkably less active than the corresponding 5-FU containing hybrids,indicating no significant synergy between the POM and rare earth metals.

Polyoxometalate-Bisphosphonate Hybrids
TheDolbecq group investigated the antitumor activity of as eries of POM-bisphosphonate complexes. [56][57][58][59] Bisphosphonates (BPs) are drugs used to treat osteoporosis and similar diseases but do also exhibit antitumor activity. [60] BPs have the general formula H 2 O 3 PC(OH)(R)PO 3 H 2 with Rdetermining their drug efficacy,a s, for example,t he primary nitrogen containing alendronate (Ale,R = (CH 2 ) 3 NH 2 )i s1 00-1000 times less active than zoledronate (Zol, R = (H 2 (C 3 H 3 N 2 )), which has ah eterocyclic amine at position R. [61] POM-BP complexes are not classical POM-based structures,a nd the most studied compounds exhibited the general formulas M 6 L 2 and M 4 L 2 X( M = {MoO 6 }, {WO 6 }, {VO 6 }; L = BP and X = Mn II/III ,F e III )( Figure 2). Theh ybrids were active on human non-small cell lung cancer (NCI-H460), glioblastoma (SF-268), and breast cancer (MCF-7) cells,w hereby V-based complexes were the most active (Supporting Information, Table S6). Molybdates and tungstates had only noticeable or good antitumor activity in combination with the most bioactive BP Zol. Thus,t he antitumor activity of the POM-BP systems correlated with the presence of either VorZ ol. Them ost potent complexes,n ot containing V, were Mo 4 Zol 2 Mn II/III ,i ndicating that the introduction of Mn II/III as heteroatom has as ignificant effect on the antitumor activity of these complexes.M o 4 Zol 2 Mn III was the only POM-BP complex that was studied in vivo. [59] Applied to the mouse xenograft model bearing human Ewing sarcoma (SK-ES-1), Mo 4 Zol 2 Mn III (5 mg/mouse for 28 d) decreased the tumor volume by about 85 %inc omparison to the control (buffer). Furthermore,t he complex did not reduce the body weight, indicating the absence of harmful effects on mice.Regarding the mechanism, amino group containing BPs are known to inhibit the prenylation of important proteins like small GTPases,which are involved in tumor growth, and therefore the POM-BP complexes could follow as imilar mechanism (see Section 3.2). [62]

Polyoxometalate-Quinolone Hybrids
Another example for POM-drug complexes with antitumoral properties are POM-quinolone antibiotic structures. [63][64][65][66][67][68] 4À were decorated with different quinolone antibiotics such as PPA, enrofloxacin (enro), norfloxacin (norf), and enoxacin (eno). In most cases aT M( Cu II ,Z n II , Ni II ,C o II )w as additionally incorporated into the structures, leading to different POM-TM-quinolone complexes,w here the POM acts as am ono-or bidentate inorganic ligand and the quinolone as an organic ligand for the TM. In POM-TMquinolone systems,the POM is covalently linked to the TMs and quinolones,w hereas in TM-lacking compounds the quinolones are clustered around the POM via noncovalent interactions.T he complexes exhibited mixed results against as eries of cancer cell lines as only ac ouple of them showed good antitumor activity,whereas the vast majority showed no to moderate activity (Supporting Information, Table S7). Depending on the used constituents,complexes with different coordination modes and POM-quinolone interactions were obtained. In general, hybrids possessing aT Mc enter with af ive-coordinate geometry were less active than those with asix-coordinated TM (Figure 3a), and systems with abidentate POM ligand (one POM binding two TMs,F igure 3b, right) were more active than those having am onodentate POM ligand. Thus,i tw as proposed that certain structural constellations might favor the delocalization of the whole electrons,which increases hybrid-tumor interactions,whereas other constellations lead to rather unsymmetrically polarized POMs with quenched activity. [64] Furthermore,i tw as suggested that structures like [Cu(PPA) 2 ] 2 [PW 12 O 40 ], where the POM unit is surrounded by quinolones,are less active as the interaction of the POM with tumor cells is sterically hindered   in comparison to complexes containing freely accessible interaction sites (Figure 3b).

Anderson Polyoxometalate-Biomolecule Hybrids
Regarding the organic modification of POMs,the Anderson archetype is one of the best studied systems as it can be easily tris-functionalized (tris = tris(hydroxymethyl)aminomethane) leading to an amino group(s) bearing structure, which can be further modified by simple amidation. [69][70][71][72] Thus, the Anderson structure represents an ideal basis for the synthesis of versatile POM-ligand complexes.Y ang et al. synthesized as eries of POM-biomolecule conjugates by grafting different bioactive receptor ligands to the surface of the Anderson POM to improve the selectivity and thus the antitumor activity of the POM. [73] Thet ris-modified Anderson-type molybdate {MnMo 6  CA and DHCA are bile acids that target the farnesoid X receptor (FXR), anuclear receptor,which induces cell death in some breast cancer cell lines upon activation. [74] CHOL is aprecursor for the synthesis of bile acids and building block of the cell membrane,w hich might be helpful for cancer cell targeting. GAL is the diacetal form of galactose,w hich is recognized by some lectins of which expression is strongly associated with tumorigenesis and thus could enhance cancer cell selectivity. [75] AA is an organic dicarboxylic acid, which was used for comparison reasons.T he cytotoxicityo ft hese hybrids was tested against the breast cancer cell lines MCF-7 and MDA-MB-231 and the noncancerous breast cell cline MCF-10A (Supporting Information, Table S8). As expected, the most effective complexes were CA-POM-CA(IC 50 = 55.9, 37.9 and 278.2 mm)a nd DHCA-POM-DHCA( IC 50 = 112.7, 149.0 and > 400 mm)e xerting the highest selectivity and antitumor activity.T he remaining complexes were only weakly or moderately active (IC 50 = 204-400 mm). Thes ignificant synergy between the POMo and CA/DHCA derives from the capability of all constituents to induce apoptosis, whereby the bile acids impart selectivity to the complex by targeting FXR.

Polyoxometalate-Amino Acid Hybrids
Amino acids are promising organic units to functionalize POMs owing to their structural and chemical variety and biocompatibility.Astring of amino acid functionalized POMs was synthesized and some of them showed promising anticancer activity (Supporting Information, Table S9).
The g-isomer of octamolybdate, (ala = alanine,g lygly = glycylglycine,m orph = morpholine, met = methionine) that selectively inhibited the cell growth of human liver and breast cancer cells (Hep-G2 and MCF-7) while showing no significant effect on other cancer cell lines (Supporting Information, Table S9). [76] Thea ntiproliferative activity of these hybrids was not cell-cycle related and thus not induced by apoptosis.Cartuyvels et al. later proposed that the antitumor activity of such hybrids and octamolybdate in general could be related to the hydrolysis of ATPa st hey observed ATPh ydrolysis in the presence of Na 4 [Mo 8 O 26 ( pro) 2 ]( pro = proline) at acidic pH (< 5.8). [77] However, the role of ATPhydrolysis during cancer progression is elusive as ATPe xhibits biphasic actions,t hat is,i th as both tumor promoting and inhibitory effects. [78] Theg lycine-decorated POMo K 2 Na[AsMo 6 O 21 (gly) 3 ] (gly = glycine) showed weak to moderate inhibitory effects against human lung carcinoma cells (A-549, IC 50 = 330.2 mm), which were still superior to that of 5-FU (ca. 40 %inhibition at 1mm). [79] However,t he arsenomolybdate was later found to be highly active on the human leukemia cell lines HL-60 and U937 (IC 50 = 8.6 and 14.5 mm ;S upporting Information, Table S9) being more active than all-trans retinoic acid (ATRA, IC 50 = 20.8 mm vs.H L-60 and 14.9 mm vs.U 937), ac linical anticancer drug, but less active than the antileukemic agent As 2 O 3 (IC 50 = 6.4 mm vs.H L-60 and 8.8 mm vs. U937). [80] However,A s 2 O 3 is also highly cytotoxic towards normal cells such as human umbilical vein endothelial cells (HUVECs) for which it exhibits an alarming IC 50 value of 5.6 mm, [81] while K 2 Na[AsMo 6 O 21 (gly) 3 ]h as no significant activity on HUVECs (IC 50 = 889.2 mm). Thes elective antileukemic activity of the hybrid might arise from As III but its low toxicity towards normal cells in comparison to As 2 O 3 renders it ap romising alternative for the treatment of leukemia.
Other amino acid functionalized POMs with antitumor activity are listed in the Supporting Information, Table S9.   [82] Biz is aw ell-known pharmacophore of which derivatives are active against several cancer types. [83] [Hbiz] 5 [HMo 5 P 2 O 23 ]i sh ighly selective towards neuroblastoma cells as it did not show any significant activity against other cancer or normal cell lines (Supporting Information, Table S10). Thea nti-SHY5Y activity of the hybrid was inferior to that of pure biz (IC 50 = 28.7 mm)b ut biz was also highly toxic on normal cells (IC 50 = 21.5 mm vs.E VC-304), making the hybrid considerably more valuable from ap harmacological point of view.
Another study reporting Strandberg-type hybrids describes the antitumor activity of [Cu(pia) 2 Figure S1). [84] All of the compounds exhibited promising activity against human hepatoma Hep-G2 and SMMC-7721 cells and colorectal carcinoma HCT-116 cells (IC 50  ]s howed clearly distinct cytotoxic behavior emphasizing the importance of the TM type. Furthermore,the coordination mode of the TM is also highly important as the activity of the two Cu II containing hybrids differed remarkably,which is in accordance with the results of the POM-TM-quinolone hybrids (Section 2.2.2.1).
Theanticancer activity of other inorganic-organic hybrids are summarized in the Supporting Information, Table S12.

Anticancer Activity of Polyoxometalate-based Nanocomposites
Thee ncapsulation of drugs is an important area in biomedicine,a st he resulting nanocomposites bring many advantages such as enhanced drug stability,d elivery,a nd activity and the extension of the bioactivity by protecting the drug from premature degradation, which is also associated with minimal side effects. [95] Using this system, the drug is in most cases released slowly and gradually leading to ap rolonged time window within which the therapeutic level of the drug is sustained. In this way,t he pharmacokinetic behavior of aseries of POMs was partially dramatically improved.

Polyoxometalate-Chitosan Nanocomposites
POM-chitosan hybrids are one of the most studied POMbased nanocomposite systems. [26] Chitosan (CT) is al inear polysaccharide obtained by the N-deacetylation of chitin and one of the most widely applied natural polymers in drugdelivery studies.C T-based systems are ideal in terms of extended drug release,asCTisdegraded in the human body by several enzymes,l eading to ac ontrolled release. [96] Biocompatibility studies showed that the nanocomposites {Co 4 (H 2 O) 2 (PW 9 O 34 ) 2 }-CMC and {Eu(SiW 11 O 39 ) 2 }-CMC (CMC = carboxymethyl chitosan) did not show any activity against HeLa cells even at high concentrations (2 mg mL À1 ) and long incubation times (up to 48 h). [97,98] Furthermore, {Eu(SiW 11 O 39 ) 2 }-CMC undergoes fast cellular uptake (within 1h), confirming the toxicity reducing and cellular transport enhancing effect of CT.Comparison between the Keggin-type {CoTiW 11 O 40 }-CMC and {CoTiW 11 O 40 }-TMC (TMC = trimethyl chitosan) hybrids revealed that the carrier properties of the positively charged TMC are superior to that of its negatively charged counterpart CMC. [99] TheT MC matrix had acirca 6times higher POM content, owing to its positive charge and the associated direct electrostatic interactions,and it was more readily uptaken by cells.T he reason for this was attributed to the smaller particle size of the POM-TMC hybrid (61 vs.131 nm), the different morphology and positive charge facilitating cell penetration. Ther elease profile of {CoTiW 11 O 40 }-TMC showed as low but steady POM release (ca. 15 %) over 24 h. Moreover,t he TMC nanocomposite exhibited cytotoxic effects on HeLa cells (inhibitory effect = 50 %a t1 2.5 mgmL À1 ,5h), whereas the parent POM was rather inactive (ca. 10 %a t50mgmL À1 ,24h).
{Gd(W 5 O 18 ) 2 }-CT siRNA nanospheres possess high potential as radiosensitizers for synergistic radiotherapy and gene therapy as they remarkably reduced the radioresistance of human hepatocellular carcinoma cells (BEL-7402) in vitro and in vivo. [100] Radiotherapy is ac linically applied cancer treatment method, which utilizes high-intensity ionizing radiation to inhibit tumor growth by the generation of cytotoxic ROS.H owever,c ells in general possess high amounts of reducing agents such as GSH to combat the production of ROSand, even more important, tumor cells are highly hypoxic minimizing the available amount of activatable oxygen and thus limiting the therapeutic effect of radiotherapy.Hypoxia is lethal to all cells;therefore,t umors develop as et of responses to outstrip their blood supply to adapt to the stressful hypoxic environment. [101] Them aster mediator of this response is the hypoxia-inducible factor 1a (HIF-1a). [102] Therefore,t he authors synthesized aC T-based system consisting of the ROS-level increasing [Gd(W 5 O 18 ) 2 ] 9À and HIF-1a siRNA, which interferes with the expression of HIF-1a,t or educe the radioresistance of cancer cells. Electrostatically driven pH-responsive supramolecular hydrogels consisting of the trilacunary Wells-Dawson [P 2 W 15 O 56 ] 12À ,c hitosan hydrochloride (CTCl), [103] or CMC [104] and poly(methacrylic acid) (PMMA) were antiproliferatively active on breast and cervical cancer cells (MCF-7 and HeLa) with minimal effects on normal Vero cells (Figure 7; Supporting Information, Table S13). Hydrogels are highly hydrated polymeric networks,i ncluding both covalent and non-covalent interactions,a nd are ideal to mimic native tissues owing to their soft consistency, porosity, and stability.A tt he maximum applied concentration of 35 mg mL À1 the hydrogels were slightly less active against both MCF-7 and HeLa (inhibitory effect ca. 72-77 and 68-70 %) than the free POM (ca. 80 and 85 %) and the chemotherapy agent doxorubicin (DOX, ca. 77 %a gainst both cells) but clearly more biocompatible.The incorporation of the pH responsive agent MMA led to ap H-controlled swelling of the gel and thus to as ustained POM release. In vivo experiments using rabbits revealed amaximum tolerable dose of 4000 mg kg À1 and no significant histopathological effects on important organs.
Other POM-CT hybrids with anticancer activity are summarized in the Supporting Information, Table S13.

Polyoxometalate-Starch Nanocomposites
Starch is ap olysaccharide consisting of glucose units linked by a-1,4-gylcosidic bonds that is extensively used in pharmaceutics as adrug carrier owing to its biocompatibility, biodegradation, and stability properties. [105] Thestarch-encap-sulated POM (SEP) {CoTiW 11 O 40 }-SEP showed increased anti-cervical (HeLa) and anti-leukemia (HL-60) activity compared to the free Keggin POM (Supporting Information, Table S14). [106] Thei ncrease in the antitumoral effect was attributed to the enhanced cell penetration ability of the starch complex, which was quantified revealing aH eLa cell penetration efficiencyof83.1 %being more than three times higher than that of the parent POM (25.5 %). However, {CoTiW 11 O 40 }-SEP was stable for only 4h (pH 7.4) before undergoing decomposition. Similar penetration efficiencies were obtained for the dimeric titanotungstosilicate [Si 2 Ti 6 W 18 O 77 ] 14À and its starch encapsulated nanoparticles (24.6 vs.8 7.2 %), confirming the starch-mediated improvement in cell penetration. [107] Thei nvitro release profile revealed an initial burst effect within the first 2-3 h, where weakly bound POMs at the nanoparticle surface are quickly released (ca. 40 %ofPOM), followed by asustained release. Such an initial burst effect, which can lead to local and toxic overdoses of POM, was not observed for CT-based nanoparticles.Invitro antitumor tests showed that the IC 50  Thea nticancer activity of further POM-SEP systems is summarized in the Supporting Information, Table S14.

Polyoxometalate-Liposome Nanoparticles
Liposomes are small spherical vesicles possessing at least one lipid bilayer.O wing to their ability to minimize the systemic toxicity of ahosted drug and protecting it from early degradation, liposomes are ideal to modify the pharmacokinetic behavior of POMs. [105] Theliposome-encapsulated POM (LEP) {Si 2 Ti 6 W 18 O 77 }-LEP was synthesized in two different sizes,n amely 60 and 150 mm, with the smaller nanoparticles having aslightly higher activity against HeLa and HL-60 cells (IC 50 = 3.2 and 4.6 mgmL À1 vs.4 .4 and 5.2 mgmL À1 ), which might be attributed to the facilitated cell penetration of the smaller particles. [108] Thea ntitumor activity of this LEP was comparable to that of its starch encapsulated counterpart  Tables S14,  S15). [109][110][111] Regarding the drug release,{ CoTiW 11 O 40 }-LEP, which has anticancer activity in hepatocellular cancer and leukemia cells,S SMC-7721 and HL-60 (IC 50 = 3.5 and 3.6 mm), showed as low and sustained release (ca. 20 % POM release). [112] In comparison, the same POM encapsulated by TMC showed as imilar release profile but with afaster initial release,whereas the starch hybrid of this POM  Thepenetration efficiencyofLEPs is similar to that of the two biopolymers discussed before as about 81 %o f {SiTiW 11 O 40 }-LEP was found inside HeLa cells and HUVECs,w hereas the penetration efficacy of the POM alone was only about 25 %. [113] Thea nticancer activity of {CoTiW 11 O 40 }-LEP against KB and HeLa cells (IC 50 = 2.2 and 2.3 mgmL À1 )w as higher than that of the free POM (IC 50 = 34.5 and 47.3 mgmL À1 )a nd 5-FU (Supporting Information, Table S15). However,its in vivo inhibitory effect on the tumor growth in HL-60 bearing rats (42 %a t2 6.4 mg kg À1 )w as inferior to that of 5-FU (58.3 %a t2 5mgkg À1 )b ut still remarkably higher than that of the pristine POM (13 %a t 200 mg kg À1 ). TheLD 50 14À with an attached long-chain organoalkoxysilane lipid, exhibited promising antiproliferative activity against human colon adenocarcinoma cells (HT-29). [114] Thel ipid covalently binds to the lacuna of the POM via its organosilicate functionality giving the hybrid an amphiphilic character.T he hybrid spontaneously self-assembles into al iposome with the POM forming the lipid bilayer ( Figure 8). Although the parent [NaP 5 W 30 O 110 ] 14À (Figure 1k)has already astrong anti-HT-29 activity (IC 50 = 3.6 mm), it was even improved by the lipidhybridization (IC 50 = 2.1 mm). Owing to its special structure, the hybrid formed stable complexes with biotinylated sBLM, an atural membrane mimetic.T herefore,a nu ptake mechanism was proposed, where the P 5 W 29 -lipid binds to the membrane and then intercalates into it forming at ransient hybrid-membrane complex, which finally releases the POM within the cytoplasmic space.

Polyoxometalate-Silica Nanocomposites
Silica-based nanoparticles are promising candidates for the drug delivery in cancer therapy because their particle size can be finely tuned to ensure appropriate accommodation of guest molecules,which is especially true for mesoporous silica nanoparticles (MSNs). [115] Karimian et al. reported on acomplex delivery system consisting of at hiolated MSN as drug carrier that covalently binds the N-Boc-cysteine-functionalized Keggin-type POM [GeV 3 W 9 ((CH 2 O) 3 N-Boc-Cys)O 37 ] 4À (Boc = tert-butyloxycarbonyl, Cys = cysteine) via ad isulfide bond. [116] Thea ddition of the fluorescent dye fluorescein isothiocyanate (FITC) to the organic functionality of the POM yields the final POM-MSN-dye complex, where the pores of the MSN are capped by the POM and loaded with the anticancer drug DOX (Figure 9). Ther edox-responsive disulfide bond was chosen for selective POM-release as cancer cells contain elevated levels of GSH, which is able to cleave disulfide bonds,w hereas FITC was attached for cellular tracking reasons.  Table S2). TheP OM-MSN-dye complex loaded with different concentrations of DOX exhibited selective anticancer activity against U-87 cells (inhibitory effect = 70 %a t2mg mL À1 DOX) as it was significantly less active on normal cells (ca. 40 %a t2 mg mL À1 DOX). Compared to the multidrug complex, the antitumor activities of the sole POM and DOX (ca. 50 %a t2mg mL À1 )w ere  clearly inferior.A se xpected the activity of the hybrid increased in aD OX concentration-dependent manner with the DOX release being GSH concentration-dependent.
Thea nticancer activity of other silica-based POMs is summarized in the Supporting Information, Table S16.

Other Polyoxometalate-based Nanocomposites
{Mo 7 O 24 }-gelatin nanoparticles showed promising anticancer activity in vitro and in vivo. [117] Gelatin is an atural polymer that is highly biocompatible and owing to its zwitterionic character it forms astable hydrophobic complex with heptamolybdate (Figure 1i). At high concentrations,the {Mo 7 O 24 }-gelatin nanoparticles exhibited cytotoxicityt o human gastric cancer cells (BGC-823) in vitro,w hich was significantly higher than that of the plain POM (inhibitory rate = 75 vs.20% at 0.5 mg mL À1 ), however, at lower concentrations (< 0.25 mg mL À1 )the hybrid was inactive.The in vivo experiment using ICR mice inoculated with murine liver cancer cells (H22) confirmed the tumor-inhibiting potential of the hybrid as the relative tumor volume (V/V 0 )was 14.5 upon treatment with 20 mg kg À1 of {Mo 7 O 24 }-gelatin for 9d, whereas V/V 0 of the control and the free POM (100 mg kg À1 )w as about 35 and about 27, respectively.A ll tumor-bearing mice died within 31 days in the control group, whereas the hybrid group (100 mg kg À1 )e xhibited as urvival rate of 60 %after 40 d(ca. 20 %inthe heptamolybdate group, 100 mg kg À1 ).
TheP t IV substituted Keggin-type POM [PW 11 O 40 (SiC 3 H 6 NH 2 ) 2 Pt IV (NH 3 ) 2 Cl 2 ] 3À was synthesized as ap harmacological prodrug that has to undergo Pt IV to Pt II reduction to be activated. [118] However,t he cell penetration properties of this organoplatinum substituted POM are limited leading to alow in vitro anticancer activity (inhibitory effect ca. 35 %a t2 0mm vs.H T-29). Therefore,t he prodrug was encapsulated with 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)-2000] (DSPE-PEG 2000 )y ielding the nanocomposite Pt IV -PW 11 -DSPE-PEG 2000 .A se xpected, the hybrid was readily internalized into HT-29 cells,w hich was also reflected in its anti-HT-29 activity (inhibitory effect = 85 %a t2 0 mm)b eing clearly superior to that of CDDP (ca. 45 %). Then anocomposite showed low toxicity towards normal HUVECs (inhibitory effect ca. 30 %a t50mm), whereas CDDP in comparison was highly toxic (ca. 60 %). Upon reduction (Pt IV !Pt II )byGSH, Pt II -PW 11 interacts with DNA, leading to apoptosis.T he hybrid exhibited also promising in vivo activity as it almost completely reduced the tumor size of HT-29-bearing BALB/c mice (dose = 0.8-2 mg kg À1 )p erforming better than CDDP without exhibiting body-weight-reducing effects.A sP tc omplexes are known for their nephrotoxicity,the Pt level in the rat kidney was determined, revealing that the treatment with Pt IV -PW 11 -DSPE-PEG 2000 led to considerably lower Pt levels than in the case of CDDP,r endering the hybrid an effective alternative for cancer therapy.
Theanticancer activity of other POM-based nanocomposites is summarized in the Supporting Information, Table S16.

Cell Penetration of Polyoxometalates
Thea ntiproliferative activity of an anticancer drug is directly associated with its degree of cellular uptake.Itiswell accepted that POMs are able to penetrate cancer cells,a s many studies have experimentally confirmed their location within the cytoplasmic space. [46,99,[109][110][111]119] However,n oc onvincing data has been reported revealing the exact mechanism. Owing to their large size and negative charge,POMs are supposed to be rather unable to penetrate the largely negatively charged cell membranes of mammalian cells. POMs such as the Keggin-type were characterized as super chaotropic agents with asurprisingly high tendency to adsorb on neutral and hydrophilic surfaces,w hereby POMs with lower charge densities were more chaotropic. [120] Owing to this feature,P OMs exhibited destructive activity towards model cell membranes by adsorbing to the vesicle surface, followed by the formation of as table POM-lipid conjugate, which finally desorbs from the membrane ( Figure 10). [121,122] These desorption processes led to leaky membrane structures. Studies with anionic lipids revealed that depending on the charge density,t he POM-lipid interactions can switch from electrostatic to hydrophobic nature,o wing to the charge neutralization by cationic counterions. [123] Nevertheless,i ti ss till generally assumed that POMs are mainly taken up by cells by some form of endocytosis. Different POTswere traced within murine macrophage J774 cells revealing POM containing vacuoles within the cells, wherefore,i tw as suggested that POMs might bind to scavenger receptors and enter the cell by scavenger receptor-mediated endocytosis. [124] Similarly,h uge POMo nanoparticles such as (NH 4 ) 42 30 }] were found to be located within endosomes that were evenly distributed in the cytoplasm of Hep-G2 cells, supporting internalization by an endocytotic pathway. [119] Monitoring of the FITC labeled {Eu(SiW 11 O 39 ) 2 }-CMC within HeLa cells showed that the nanoparticles were preferably located in the perinuclear region in close proximity to the nuclei. [98] Clathrin-mediated endocytosiswas proposed as POM-free chitosan nanoparticles are internalized by this pathway;h owever, this was later excluded as the cellular uptake of POM-CT was not inhibited by the clathrin inhibitor chlorpromazine indicating aclathrin-independent pathway. [99] Thus,m acropinocytosis and caveolae-mediated endocytosis were suggested as internalization pathways for, at least, POM-CT hybrids as chitosan-DNA-poly(g-glutamic acid) complexes are uptaken by these pathways. [125] 3.2. Proposed Modes of Action and Biological Targets of Antitumoral Polyoxometalates Figure 11 shows most of the putative mechanisms of antitumoral POMs in ac omprehensive Scheme.T he first suggested mechanism for the antitumor activity of POMs was that by Yamase,w hich was already mentioned in the introduction. [19] Briefly,r epeated reduction/oxidation cycles between the POM and cell components,most likely members of the electron transport chain, are supposed to interfere with ATPg eneration, which finally leads to apoptosis (Fig-Figure 11. Illustration of most of the proposed modes of action of antitumoral POMs. a) POM induced inhibition of ATPsynthesis by interfering with the electron transfer chain (represented as dark blue entities, the inset shows azoomed view of the chain). b) POM induced increase in ROS-level( for example by oxidizing cell components) and depletion of the GSH pool (by GSH oxidation). c) POM induced enhancement of the expression of pro-apoptotic components (Bax and Bim) and the reduction of the expression of anti-apoptotic components (bcl-2 and NF-kB). d),e) Activation of the p53 and/or p38 pathway by POMs. Please note:t he circles reading p38 and p53 do not represent the respective protein but the pathways. f) Induction of apoptosis by direct DNA damage. g),h) POM-mediated inhibition of angiogenesis via interaction with bFGF and VEGF leading to the disruption of the VEGF/bFGF-receptor interactions (receptors are indicated as yellow and brownish channels).W ithout VEGF/bFGF-receptor binding, the ERK pathway cannot be activated leading to the breakdown of angiogenesis. i) Inhibition of ectonucleotidases by POMs leads to adistortion in the concentrations of nucleotides (NTP, NDP, and NMP) and nucleosides( Ns), which negatively affects the functioningofcancer cells. j) Inhibition of HDAC by POMs leads to the accumulation of acetylated histones, causing fatal changes in the expression of genes. k) Inhibition of P-type ATPases has fatal effects on the cellular ion homeostasis. l) Decavanadate-induced mitochondria membrane depolarization. m) Inhibition of other proteins that affect cell viability (for more details, see text of Section 2.3.1 and 3.2.5, and references therein). n) POM hybrids loaded with siRNA (siRNA is depicted as red RNA structure) downregulate HIF-1, leading to the impairment of angiogenesis and the adaptation of cancer cells to the hypoxic environment. o) Immunostimulating activity of POMs by promoting the expression of antibodies and immune-related components (for example, NK cells). The figure depicts the activation of NK cells via antibody binding enhancing the recognition of tumor cells (antigens are depicted as purple triangles) by NK cells. Dotted lines indicate that the reason of activation/deactivation (for example, enhanced/decreased expression) of certain components is not known. The release of cytochrome c( purple circle) triggers the apoptotic machinery of the cell, which ultimately activate the final executers of apoptosis (caspases). ure 11 a). Thetheory is widely accepted, as aseries of studies revealed ap utative correlation between the cytotoxicitya nd the redox potential of bioactive POMs. [38][39][40][41][42][43][44] This correlation is aweak one and only applies to very closely related structures, as other factors like size,structure,and composition play also as ignificant role.I ng eneral, there is no unambiguous correlation between the anticancer activity of POMs and parameters such as POM size,total net charge,charge density, oxidation power, and archetype.

Activation of Cell Death Pathways
Apoptosis,necrosis,and autophagy are types of cell death, which are generally highly sophisticated processes including complex signaling cascades that are tightly regulated. According to Section 2, anumber of POMs induced apoptosis by the mitochondrial (intrinsic) pathway as no study describes,f or example,t he involvement of ad eath receptor,w hich is ah allmark for extrinsic apoptosis.T he reason for this is that most POMs mainly induced "internal damage" by oxidative stress,w hich is signaled to mitochondria (Figure 11 b). Some POMs like the Wells-Dawson POM are able to induce apoptosis by affecting the expression of cell death regulators, for example,b yi ncreasing the amount of the pro-apoptotic proteins Bax and Bim (Figure 11 c). [126,127] Other POMs,s uch as Krebs-type,d oa lso reduce the expression of the antiapoptotic protein bcl-2 and the transcriptional factor NF-kB (Figure 11 c). [33,34] Active NF-kBisresponsible for the expression of genes that protect the cell from undergoing apoptosis, however, many tumor types have ac onstantly active NF-kB and therefore its inhibition represents ap romising approach in cancer therapy.F urthermore,K rebs-type tungstobismuthates are able to increase the expression of p53 activating apoptosis partially by this pathway (Figure 11 d). [33] The tumor suppressor p53 is at ranscription factor that induces anti-carcinogenesis events such as cell growth arrest and inhibition of angiogenesis,m aking it the guardian of the genome and one of the most promising targets as about 50 % of human cancers are thought to be related to p53 mutations. [128] Some POMs like the Co II containing Krebs-type (H 2 im) 2 [(W(OH) 2 ) 2 (Co(H 2 O) 3 ) 2 (Na 4 (H 2 O) 14 )(BiW 9 O 33 ) 2 ]are able to enhance the activation of caspase-3, the final executioner of apoptosis. [35,127] Na 7 [Cr III Cu II W 11 O 39 ]was observed to induce the upregulation of both cytochrome ca nd activated p38 in human ovarian SK-OV-3c ancer cells (Figure 11 e), despite its very low anticancer activity (IC 50 = 1.87 mm). [127] Cytochrome ci sa ne ssential component of the electron transport chain but plays also adecisive role in apoptosis as its release from the mitochondria into the cytosol upon apoptotic stimuli triggers apoptosis. [129] Thus,P OM-induced overexpression of cytochrome cc an promote cell death. p38 is am itogen-activated protein kinase (MAPK) that has oncogenesis suppressing properties as it is required for dormancy, that is,i nhibition of the cell proliferation upon certain stress stimuli. [130] Furthermore,N a 7 [Cr III Cu II W 11 O 39 ]i sa lso able to trigger autophagy,which means that two cell death pathways can run in parallel. Thes ame was observed for PM-17 as it induced apoptotic and autophagic cell death in vitro and in vivo. [18] Thes andwich POM [H 4 {Cu II 9 As III 6 O 15 ( H 2 O) 6 }(As III W 9 O 33 ) 2 ] 8À exhibited remarkable activity against cancerous K-562 (leukemia) and Hep-G2 (liver) cells (IC 50 = 0.4 mm for both) by affecting lysosomes in vitro,w hich led to the induction of both apoptosis and autophagy. [131] The autophagy inducing property might originate from the As III containing unit as As 2 O 3 is known to induce autophagy in leukemia cells by increasing the level of Beclin-1, ac ritical regulator of autophagy. [132] 3.

DNA Interaction
One of the main stimuli inducing intrinsic apoptosis is DNAdamage.Therefore,the anticancer activity of aseries of POMs and POM-based hybrids was associated with DNA lesions but mainly induced indirectly as the negative charge of both molecules is supposed to prevent direct electrostatic interactions.H owever,s ome POM-based structures were found to directly interact with DNA ( Figure 11 f), for example,[ (CpTi) 3 SiW 9 O 37 ] 7À interacts strongly with DNA, most likely via its CpTim oiety. [43,47] These results were confirmed by the group of Habibi, which investigated the behavior of different CpM-substituted Keggin POTs (M = Zr IV ,T i IV ,F e II )t owards ctDNAa nd suggested ad irect but noncovalent groove or outside stacking binding mode for the POM. [46] [31,[109][110][111] Despite the suggested groove or outside stacking interaction, the exact POM binding mechanism remains elusive.A nother study investigating the interaction between heptamolybdate [Mo 7 O 24 ] 6À and the DNA model bis(p-nitrophenyl)phosphate showed that the POM was able to cleave the phosphodiester bond by ayet unknown mechanism. [133]

Inhibition of Angiogenesis
Angiogenesis describes the formation of new blood vessels,avital process in cell growth. [134] As cancer cells are rapidly dividing, tumors need special blood supply to provide oxygen and other nutrients to continue their abnormal growth. Thus,t hey induce angiogenesis by activating various growth factors such as the vascular endothelial growth factor (VEGF) and the basic fibroblast growth factor (bFGF), which stimulate the formation of blood vessels.T he inhibition of angiogenesis promoting factors is ap otential approach in cancer therapy.Aseries of POMs,i ncluding lacunary and fully saturated Keggin, sandwich Keggin, and Wells-Dawson structures,i nteract strongly with bFGF leading to the inhibition of its proliferation promoting activity in HUVECs. [135,136] Independent of the charge,W ells-Dawson POMs were most efficient in binding bFGF and therefore it was assumed that the POM structure determines the affinity towards bFGF.C ompetition assays with suramin, which is known for its angiosuppressive properties,a nd heparin, an activity enhancer and inhibitor of bFGF,revealed that POMs bind in the vicinity of the heparin binding site,which is rich in positively charged amino acids.According to the dimensions of the proposed binding site,W ells-Dawson structures are supposed to fit better into it than the smaller Keggin ions, which might explain the differences in binding affinity ( Figure 12). POM binding might interfere with the ability of bFGF to interact with its receptors,w hich is required for its angiogenesis promoting function (Figure 11 g) 30 }] exhibited astonishing anticancer activities against some cancer cell lines with high selectivity towards Hep-G2 cells (IC 50 = 9-55 mgmL À1 )mainly by inhibiting angiogenesis (Supporting Information, Table S1, described as Mo-compounds 1-3). [119] TheM o-nanoparticles impaired the formation of blood vessels by inhibiting key processes of angiogenesis,n amely VEGF-induced proliferation, migration, and tube formation of endothelial cells (HUVECs) and neovascularization on CAM( chorio-allantoic membrane). Furthermore,the level of NO,amediator of angiogenesis,was also significantly reduced in the presence of the POMos as was the VEGF-induced phosphorylation of signal-regulated kinase 1a nd 2( ERK1/2) and phosphoinositide 3-kinase (AKT), signaling processes normally associated with the promotion of angiogenesis (Figure 11 h).

Interaction with Proteins
Owing to their negative charge and their tendencytobind to neutral or hydrophilic surfaces,P OMs interact with an array of proteins. [138] Ectonucleotidases are membrane-associated enzymes that hydrolyze extracellular nucleotides to the respective nucleosides and are thus involved in numerous physiological and pathological processes. [139] They can be subdivided into four families,e ctonucleoside triphosphate diphosphohydrolases (NTPDases), ectonucleotide pyrophosphatases (NPPs), alkaline phosphatases (APs), and ecto-5'nucleotidase (eN). POMs are potent inhibitors of ectonucleotidases,w ith some of them being superior to known inhibitors ( Figure 11 i; Supporting Information, Table S17). [140] Crystallographic studies investigating the interaction of decavanadate,m etatungstate,o cta-, and heptamolybdate with rat and bacterial NTPDase1 revealed that decavanadate and heptamolybdate bind either at the periphery (rat NTPDase) or deep inside the active site cleft (bacterial NTPDase) of the enzyme at positively charged patches,w here they interact with substrate binding residues ( Figure 13). [141,142] Therefore,i tw as suggested that enzyme inhibition occurs by blocking the nucleotide-active site interaction. Octamolybdate (Figure 1i)i sl ocated at the entrance of the active site cleft and could also interfere with substrate binding.I nterestingly,o ne of the most potent inhibitors,m etatungstate,b inds distantly from the active site and therefore the inhibitory effect might result from POMmediated destabilization of the entire enzyme structure.
Theinhibition of human NTPDase1, also known as CD39, is of particular pharmacological interest as it converts ATP into adenosine,w hich is an important molecule in the suppression of the antitumor Tc ell response and therefore it is exploited by tumors to overcome immune response. [143] NTPDase1 inhibition by metatungstate led to as ignificant in vivo suppression of murine B16-F10 melanoma (inhibition rate ca. 86.1 %, 5mgkg À1 )a nd to am oderate reduction of murine colon adenocarcinoma (MCA-38, ca. 27.3 %, 5mgkg À1 )i nm ice. [144] Metatungstate was not active in NTPDase1 deficient tumor-bearing mice confirming NTPDase1 inhibition as the mode of action. Further studies investigating the effect of several POMs on human NTPDase1-3, NPP1-3, and rat eN revealed that POMs were most efficient in inhibiting NTPDase1-3 and NPP1, with some of them exhibiting inhibition constants in the nanomolar range (Supporting Information, Table S17). [145,146] [Co 4 (-H 2 O) 2 (PW 9 O 34 ) 2 ] 10À was identified as av ery potent inhibitor of human NTPDase1-3 (K i = 4, 20 and 100 nm), being the most potent inhibitor of human NTPDase1 described to date.
POMs are also potent inhibitors of tissue specific alkaline phosphatase (TSAP,f rom calf intestine) and of tissue nonspecific alkaline phosphatase (TNSAP,f rom pork and human), with some of them being active at nanomolar concentrations (Supporting Information, Table S17). [145,147] Regarding human TNSAP,[ P 6 W 18 O 79 ] 20À (K i = 2.3 mm)i st he most potent inhibitor. [145] APs are highly expressed by osteoblasts to hydrolyze inorganic pyrophosphate for bone mineralization. This osteoblastic activity is remarkably increased in prostate cancer,w hich preferentially metasta- Figure 12. Putative POM-binding site of bFGF. a) Crystal structure of human bFGF (PDB entry 1BFF [137] )with the putative binding site being marked by ared circle. Side chains of amino acids that potentially contribute to heparin binding are shown as sticks. b) Coulombic surface representation of bFGF is illustrated with blue surfaces representingregions exhibiting apositive potential, whereas gray and red surfaces possess neutral and negative potentials, respectively. c) Zoom view and dimensionsofthe putative POM-binding site. d) Polyhedra structures and dimensionso fthe Keggin and Wells-Dawson anions for easier comparison.B lue polyhedra are {MO 6 }, green polyhedra {XO 4 }, green sticks carbon, dark blue sticks nitrogen, red sticks and spheres oxygen. sizes to bone,l eading to dysregulated bone formation. [148] Therefore,the inhibition of APs by POMs could be apromising strategy in the treatment of metastatic prostate cancer.
Aseries of POMs were tested on several kinases showing high specificity for protein kinase CK2. [149] Protein kinases catalyze the phosphorylation of proteins leading to their modification, that is,c hange in enzymatic activity,c ellular location, or interaction behavior to regulate cellular processes such as signal transduction. Protein kinase CK2 is upregulated in many cancer types and thus associated with their increased proliferation rate and ability to suppress apoptosis. [150] Derivates of the Wells-Dawson (IC 50 = 1-70 mm)a nd Preyssler archetype (IC 50 = 1-5 mm), and giant POM anions (IC 50 = 8-70 mm), showed moderate to high inhibition of CK2. In contrast, smaller POMs such as Keggin derivates (IC 50 = 60-1000 mm)w ere less active indicating that the POM structure roughly determines the inhibitory effect. Further analysis revealed that the POM binding site is located at an exposed site outside the ATP/peptide binding pocket and the catalytic cleft from where the POM could however interfere with the activation loop of CK2 ( Supporting Information, Figure S4). [151] Ah igh-throughput screening study assaying 400 POMs was performed to identify potential histone deacetylase (HDAC)i nhibitors revealing [{(n-Bu)Sn(OH)} 3 GeW 9 O 34 ] 4À as the most efficient HDAC-inhibitor (IC 50 = 1.1 mm) ( Figure 11 j). [48] HDAC is an enzyme that is responsible for the removal of the acetyl group from an acetylated lysine making histones to wrap DNAm ore tightly.T his is an essential process as DNAe xpression is regulated by the acetylation and deacetylation of histones.H DACs are involved in cellcycle progression and differentiation and therefore disturbances in HDAC encoding genes are linked to tumor development rendering them promising targets. [152]  ,w hich are known to exhibit antitumor activity,have also been described as potent P-type ATPase inhibitors (Figure 11 k). [153][154][155] Ptype ATPases are al arge group of ion pumps that play ac rucial role in maintaining the ionic balance in cells and therefore have been described as potential molecular targets for the treatment of several diseases.P -type ATPases like Na + /K + -ATPase also act as signal transducer and their activity was found to be significantly changed in tumors making Na + / K + -ATPase inhibitors particularly interesting as anticancer drugs. [156] Therefore,the reported anticancer activity of some POMs,especially that of decavanadate and other POVs,may partially derive from their ability to inhibit P-type ATPases. However,P OM-mediated P-type ATPase inhibition is not tumor-cell-selective.Besides this,decavanadate induces mitochondria membrane depolarization and inhibits mitochondrial oxygen consumption in vivo rendering mitochondria the main toxicological target for decavanadate (Figure 11 l). [157] Them echanism behind the anticancer activity of POMbisphosphonate (POM-BP) complexes,which were discussed in Section 2.2.2, is supposed to be the inhibition of the prenylation of important proteins like the Ras subfamily, which belong to the class of small GTPase.The Ras genes are the most common oncogenes in human cancer, and point mutations within these genes result in the accumulation of activated Ras,which in turn causes apermanent cell growthpromoting signaling,u ltimately leading to cancer. [158] Therefore,t he Ras pathway is ap romising biological target. BPs like Zol are known to inhibit farnesyl diphosphate synthase, an enzyme supplying precursors for the biosynthesis of isoprenoids that are essential for prenylation. In the same way,P OM-BP complexes could inhibit the prenylation of (mutated) Ras,p reventing it from reaching its full functionality,a nd thus impairing tumor growth. This mechanism was supported by the finding that the addition of geranylgeraniol (GGOH), an isoprenol that is also used for protein prenylation, impaired the antitumor activity of Mo-based POM-BP complexes as prenylation was restored. Interestingly,t he Figure 13. POM-binding sites of NTPDase1 from Legionella pneumophila. The structure of bacterial NTPDase1 consists of two domains (cyan and green cartoon)w ith the interface forming the active site cleft. The depicted structure is taken from PDB entry 4BVO [142] as model example. The interacting amino acid residues are shown as sticks. Light blue spheres are tungsten, cyan spheres molybdenum, gray spheres vanadium, cyan/green sticks carbon, dark blue sticks nitrogen, red sticks oxygen. Dashed lines represent POM-protein interactions. Note that octamolybdate forms acovalent bond with aserine (Ser 127). activity of V-containing BP complexes was not affected by GGOH, suggesting that the activity of these complexes is governed by the POVu nit and thus follows aP OV-based mode of action.
Besides the discussed biomacromolecules,POMs interact with an array of other important proteins including phosphatases,kinases,polymerases,proteases,actin, sulfotransferases, and sialyltransferases (Figure 11 m). [28,138,[159][160][161][162] POM-mediated inhibition of certain enzymes has aspecific effect on the cell viability and may thus contribute to pharmacological activities.S ome of these enzymes are membrane-associated proteins of which targeting is facilitated as the POM does not has to enter into the cytoplasmic space for inhibition.

Other Mechanisms
Thea nticancer activity of the {Gd(W 5 O 18 ) 2 }-CT siRNA system is on the one hand based on the radiosensitizing effect of the Gd-POM, leading to the cellular generation of ROSu pon X-ray irradiation and on the other hand by the HIF-1a down-regulating effect of the siRNA ( Figure 11 n). Besides increasing the ROSl evel, the POM oxidizes GSH preventing the cancer cell from counteracting the oxidative stress.T he inhibition of HIF-1a leads to the depletion of downstream proteins,w hich makes the cancer cell more amenable to the lethal hypoxic environment. Theexpression of the angiogenesis promoting factors VEGF and c-Met were remarkably decreased upon HIF-1a inhibition, leading to an increased vulnerability of the cells owing to the detriment of the development of new blood vessels. [100] TheF e III containing sandwich POM [Fe(HPW 7 O 28 ) 2 ] 13À exhibited moderate in vitro cytotoxicityt oas eries of cancer cells (Supporting Information, Table S2) but showed promising in vivo activity by inhibiting the tumor growth in S180 sarcoma bearing mice (inhibitory rate = 45.7 %a t 80 mg kg À1 ). [163] Thei nvivo activity derived mainly from the ability of the POM to enhance the immune response in tumorbearing mice.The activation of the host immune response has been recognized as ap romising approach in the combat against tumors. [164] TheP OM significantly increased the proliferation of splenocytes,t he activity of natural killer (NK) cells and cytotoxic Tl ymphocytes (CTLs), and promoted the production of the Ty pe 1h elper cytokines interleukin-2 (IL-2) and interferon-g (IFN-g). Furthermore, the serum antigen-specific antibody level (IgG2a and IgG2b) was enhanced upon POM treatment. All these results indicate that [Fe(HPW 7 O 28 ) 2 ] 13À has immunomodulatory activity, making this POM ap otential drug for immunotherapy (Figure 11 o).

Summary and Outlook
Aseries of POMs and POM-based hybrid systems possess considerable potential as metallodrugs in the treatment of cancer as evidenced by in vitro and in vivo studies.Owing to the lack of clear correlations between the observed anticancer activities and any structural or chemical POM feature,i ti s quite impossible to anticipate the bioactivity of agiven POM.
In general, the bioactivity of purely inorganic POVs or vanadium-containing POMs was higher than that of POMos and POTs.H owever,t he clinical application of POMs is restricted because of some major drawbacks:P OMs are generally highly toxic [17,18,22] and most POM-archetypes are thermodynamically and kinetically unstable under physiological conditions. [77,106,165] Furthermore,n aked POMs possess rather low cell penetration ability and their promiscuous protein binding leads to low selectivity.D espite these limitations,P OMs are still considered valuable bioactive agents,a sn early every molecular property that affects their biological reactivity can be altered and their surface can be organically modified or encapsulated by biocompatible molecules,e nabling the synthesis of multifunctional and cell-penetrating hybrid POMs that overcome most of the aforementioned drawbacks.G rafting of organic molecules onto the POM core or incorporation of bioactive transition metals into the POM framework had partially immense effects on the bioactivity of the system. However,t he incorporation of POMs into nanocomposites is clearly more effective in terms of increasing the POM stability and decreasing its inherent toxicity.P OM-based hybrids and nanocomposites containing ac ell targeting molecule (for example,areceptor agonist) or afurther bioactive compound were especially efficient in most cases.I ng eneral, the biological and pharmacokinetic properties of POM-based nanocomposites seem to be superior to that of just organically functionalized POMs.
Fundamental questions regarding the mechanisms behind the POM-mediated antiproliferative effects remain largely unanswered. Although as eries of possible mechanisms and potential targets were proposed, the exact mode of action remains unclear. With the existing methods it is almost impossible to pinpoint the POM-induced effect that finally leads to cell death, as cell death is associated with avariety of signaling pathways that crosstalk heavily and involve ap lethora of signaling molecules.D ue to their multifunctional character,P OM-based systems possess al arge repertoire of possibilities to inhibit tumors and therefore their mode of action will probably not be explained by one strict mechanism but rather by multiple interactions interfering with anumber of cellular processes.Owing to the low stability of most POM archetypes,m ore and more researchers propose that monomeric species or unidentified POM fragments are responsible for the observed biological effects. [79,80,92] According to this, the POM might rather act as acarrier of these active species that upon reaching the site of action releases the toxic payload. Therefore,i ti si mportant to develop reliable methods that allow the unambiguous identification of the bioactive species.I nt his way,w hether the activity of POMs correlates with their propensity to dissociate could be investigated. In general, this field requires an immense experimental effort and the development of novel and improved methods to elucidate the mechanism behind the anticancer activity of POMs.Future research will focus on the identification of new targets and the design of novel POM hybrids as they are currently the most efficient (POMcontaining) compounds regarding the biological activity, toxicity,a nd pharmacokinetic properties.T om eet all of these upcoming tasks and to pave the way for POMs as nextgeneration anticancer drugs,i nterdisciplinary cooperation between chemists,b iochemists,c rystallographers,p harmacists,and physicians is required.