Dose dependent effects of cadmium on tumor angiogenesis

Angiogenesis is crucial for tumor growth and metastasis. Cadmium (Cd) exposure is associated with elevated cancer risk and mortality. Such association is, at least in part, attributable to Cd-induced tumor angiogenesis. Nevertheless, the reported effects of Cd on tumor angiogenesis appear to be either stimulatory or inhibitory, depending on the concentrations. Ultra-low concentrations of Cd (<0.5 μM) inhibit endothelial nitric oxide synthase activation, leading to reduced endothelial nitric oxide production and attenuated tumor angiogenesis. In contrast, low-lose Cd (1-10 μM) up-regulates vascular endothelial growth factor (VEGF)-mediated tumor angiogenesis by exerting sub-apoptotic levels of oxidative stress on both tumor cells and endothelial cells (ECs). The consequent activation of protein kinase B/Akt, nuclear factor-κB, and mitogen-activated protein kinase signaling cascades mediate the increased secretion of VEGF by tumor cells and the up-regulated VEGF receptor-2 expression in ECs. Furthermore, Cd in high concentrations (>10 μM) induces EC apoptosis via the activation of caspase-3, resulting in destruction of tumor vasculature. In this review, we summarize the current knowledge concerning the roles of Cd in tumor angiogenesis, with a focus on molecular mechanisms underlying the dose dependent effects of Cd on various EC phenotypes.


INTRODUCTION
Angiogenesis refers to the physiological process in which new blood vessels grow from pre-existing vessels [1,2]. Driven by pro-angiogenic stimuli, pericytes surrounding the normally quiescent endothelial cells (ECs) detach from the basement membrane [3]. Meanwhile, junctions between ECs dilate, enabling the extravasation of plasma proteins which serve as a provisional framework of the extracellular matrix (ECM) [3]. As a result, ECs migrate, proliferate and form tubes which eventually fuse with neighboring vessels [3][4][5]. Angiogenesis is vital for normal physiological processes of fetal development and wound healing [3,6,7]. However, it is also fundamental for solid tumor growth and metastasis [8]. Tumor angiogenesis provides tumors with oxygen (O 2 ) and nutrients, and also aids the disposal of tumor metabolites [7,9]. Tumor sizes are limited to 2mm in diameter without neovascularization [7]. Moreover, tumor angiogenesis often results in hyperpermeable vasculature with a twisted appearance [10]. Aberrant vasculature in turn aggravates regional hypoxia and acidosis, both of which promote tumor progression [10].
Vascular endothelial growth factor (VEGF) is the predominant mediator of tumor angiogenesis [11]. Malignant cells produce VEGF to promote angiogenesis even before the formation of a visible tumor [8]. Nevertheless, due to the exceptional growth rate of cancerous cells, solid tumors still experience hypoxia despite the already increased vessel formation [8]. Oxygen insufficiency stabilizes hypoxia inducible factor-1α (HIF-1α), a constituent of the transcription factor HIF-1 [11][12][13][14][15]. Activated HIF-1 translocates to the cell nucleus where it binds to promoter regions to promote VEGF Review expression [16]. VEGF signaling is mediated by VEGF receptor-2 (VEGFR-2) on ECs [11]. Upon binding with VEGF, VEGFR-2 undergoes tyrosine phosphorylation, triggering multiple down-stream signaling events involving activation of mitogen-activated protein kinases (MAPKs) and protein kinase B (PKB)/Akt, which promote EC survival and proliferation [11,15]. In addition, proinflammatory cytokines released by cancerous cells promote EC migration by increasing vascular permeability [3,17]. Angiogenesis is regulated by a number of stimulators and inhibitors [11]. Inhibitors of angiogenesis have been developed as an attempt to improve cancer treatments [8]. For example, combined therapy with VEGF antagonists and chemotherapy effectively reduced tumor size and invasiveness [10]. On the other hand, environmental chemicals, including toxic metals, may facilitate tumor growth by stimulating angiogenesis [18,19].
Cadmium (Cd) is a naturally occurring element that can be found both in the atmosphere and in soil [20]. Since the 1940s, Cd has been used widely in industrial processes [20]. Epidemiological evidence suggests that Cd poses a significant health threat because it substantially increases the risk of cancer, renal failure, osteoporosis, and developmental abnormality [20][21][22][23][24][25]. For non-occupationally exposed populations, Cd exposure normally results from tobacco consumption or ingestion of contaminated substances [21]. Following absorption by either lung or the intestinal epithelium, Cd enters the systemic circulation [26]. Cd exists as a mixture of free cations and metal compounds in blood [26]. Since Cd has a high affinity to thiol groups, plasma proteins containing thiol groups including albumin, metallothionein (MT), and glutathione (GSH) are considered as major carriers of Cd [26,27]. Both GSH and MT are potent antioxidants, binding of Cd with these proteins neutralizes Cd toxicity [28][29][30]. Nevertheless, increased level of Cd depletes anti-oxidative enzymes, and thus induces oxidative stress which affects the surrounding cells [26,31]. Oxidative stress associated with sub-apoptotic dose of Cd activates pro-survival signaling, leading to enhanced cell proliferation and malignant transformation [9,22,23,[32][33][34]. As a result, Cd has been characterized as a Group 1 human carcinogen by the International Agency for Research on Cancer [21].
As Cd directly alters signaling cascades in both tumor cells and ECs, it has been linked with tumor angiogenesis [1,9,21,22,[35][36][37]. Exposure to Cd increases the production of VEGF by cancerous cells [9,35]. Cd also directly enhances EC survival and proliferation by up-regulating the expression of VEGFR-2 [38]. Therefore, Cd-induced tumor angiogenesis contributes, at least in part, to the association between high Cd intake and increased cancer mortality [7,21]. However, Cd has also been described as an inhibitor of angiogenesis [38][39][40][41][42][43][44][45]. The apparent discrepancy between studies calls for an in-depth review regarding the effect of Cd on tumor angiogenesis. This review will provide a detailed analysis of the interactions between Cd and tumor vasculature, and discuss potential mechanisms underlying the dose dependent effect of Cd.

ULTRA-LOW DOSE CD ATTENUATES ANGIOGENESIS BY INHIBITING ENOS ACTIVITY
Blood Cd concentration serves as a biomarker for Cd exposure level [21]. Data from two Swedish studies indicated that blood Cd concentration in nonoccupationally exposed population may range from just above 0 μM to 0.05 μM [21,46,47]. Nonetheless, the blood Cd concentration of human varies remarkably subject to age, gender, diet, residential area, and smoking status [21,47].
Cd in ultra-low concentrations ( < 0.5 μM) attenuates angiogenesis in both the wound healing assay and the chick choriollantoic membrane (CAM) assay [40]. In addition, ultra-low concentrations of Cd reduce bradykinin (BK), a powerful angiogenic agent, and mediate both tube formation in 3D matrigel matrix and ex vivo angiogenesis in CAM models [39]. Mechanisms behind these observations have not been fully understood, but such anti-angiogenic effects of Cd might be mediated partially by the blockade of eNOS activity [39,40]. eNOS is an enzyme in ECs that catalyzes nitric oxide (NO) production [48][49][50]. Canonically, activation of eNOS is achieved by binding of a calcium/CaM complex to the CaM-binding region of eNOS [50][51][52]. Interaction with heat shock protein 90 (Hsp90), a chaperone protein, causes membrane-associated eNOS to dissociate from caveolin-1 (cav-1) while undergoing phosphorylation [48,50,53]. Phosphorylation of eNOS leads to a flux of electrons through its reductase domain and thus facilitates the oxidative reaction in which L-arginine is transformed to L-citrulline and NO [48,[54][55][56][57].
Upon treatment with ultra-low dose Cd, phosphorylated eNOS in human umbilical vein endothelial cells (HUVECs) is decreased [39,40]. The reduction in activated eNOS is accompanied by a decrease in NO production [40]. Hence, Cd might directly inhibit eNOS phosphorylation, leading to reduced eNOS activation [40]. Meanwhile, when ECs are treated with ultra-low dose of Cd, BK-induced perinuclear translocation of eNOS is abolished [39]. BK is able to initiate eNOS phosphorylation [39]. Soluble BK binds to the membranebound BK2 receptor and activates phospholipase C-γ (PLC-γ), which up-regulates Ca 2+ levels in the cytoplasm [51]. Elevated cytoplasmic Ca 2+ levels facilitate the binding between calcium/CaM complex and eNOS [51]. In addition, calcium/CaM complex activates CaM kinase II (CaMKII) which directly phosphorylates eNOS [51,58]. While membrane association is essential for eNOS activation, restricting eNOS to the caveolae-rich plasmalemma increases the binding between eNOS and cav-1 [48,59]. Cav-1 binding inhibits the enzymatic activity of eNOS [59]. Therefore, ultra-low dose Cd decreases eNOS signaling via the inhibition of eNOS phosphorylation and perinuclear translocation [39] ( Figure  1).
Cd also appears to compete with Ca 2+ for entry into cells [32,48]. This mechanism potentially explains the decrease in intracellular Ca 2+ level in Cd treated ECs [39]. Since Ca 2+ is required for eNOS activation, the competition between Cd 2+ and Ca 2+ for passage through ion channels might be another mechanism underlying Cd-reduced NO production [48] (Figure 1). Furthermore, Cd competes with zinc (Zn) for binding sites on proteins [60]. Since myc-associated zinc-finger protein (MAZ) is a promoter of eNOS, the replacement of Zn by Cd in MAZ might attenuate eNOS activity [48,60]. By suppressing eNOS activation, ultra-low concentrations of Cd reduce NO production by ECs [39,40]. NO is responsible for regulating vascular tone, EC proliferation, and angiogenesis [50]. NO signaling is orchestrated via S-nitrosylation which covalently incorporates NO into a thiol group on the target protein [61]. Under normoxic conditions, S-nitrosylation stabilizes HIF-1α and initiates the transcription of VEGF [16,61,62]. NO also contributes to the accumulation of HIF-1α by inhibiting protein hydroxylase domain containing protein 2 (PHD 2) [63,64]. Hence, decreased NO due to exposure to ultra-low dose Cd reduces VEGF expression. In addition, hypoxia facilitates the binding between cytochrome c oxidase and NO [61,65,66]. Such binding  [39]. Binding between BK and BK2 receptor initiates down-stream signaling of PLC-γ, which involves the up-regulation of intracellular Ca 2+ levels and activation of CaM [51]. Activated calcium/CaM complex binds to eNOS to trigger its canonical activation involving Hsp90 [48,50,53]. In addition, calcium/CaM complex stimulates CaMKII which activates eNOS by direct phosphorylation [51]. Ultra-low dose Cd also impedes eNOS perinuclear translocation [39,40]. Excessive binding of eNOS to the plasmalemma may lead to cav-1-mediated inhibition of eNOS activity [59]. www.impactjournals.com/oncotarget increases intracellular O 2 levels by reducing mitochondrial respiration [66]. Combined with NO insufficiency, PHD is activated and promotes the proteasomal degradation of HIF-1α [65,66]. Therefore, reduced NO level as a result of ultra-low dose Cd exposure leads to decreased VEGF production and impaired angiogenesis [61,65]. Blockage of VEGF signaling and increased Ang-2 expression Inhibitory 41, 42
Low-dose Cd also activated ERK signaling in ECs with an increase in VEGFR-2 expression [38]. Meanwhile, inhibitors of ERK signaling reduces the level of VEGFR-2 [38]. Therefore, Cd-activated ERK up-regulates VEGFR-2 in ECs [38]. The resulting enhanced VEGF signaling contributes, at least in part, to the Cd-induced increase in EC proliferation [80,97]. Specifically, VEGF-A has been found to activate an orphan nuclear receptor transcription factor TR3 in HUVECs [80,98]. TR3 then mediates the expression of cell cycle genes which promote EC proliferation [80,98]. By stimulating EC proliferation, Cdenhanced activation of ERK signaling in ECs facilitates tumor angiogenesis [1].

Figure 5: p38 MAPK mediates high-dose Cd induced EC apoptosis. At high concentrations ( > 10 μM), Cd triggers EC
apoptosis via the activation of p38 MAPK [43]. Activation of p38 MAPK increases Bax expression and cleavage of caspase-9 [105,110,111]. Caspase-9 cleavage activates caspase-3 [112]. Both increased Bax level and caspase-9-activated caspase-3 are prominent mediators of apoptosis [112]. In addition, phosphorylated p38 MAPK inhibits ERK and impairs the pro-survival VEGF signaling [83,94,95,106,113]. www.impactjournals.com/oncotarget JNK signaling JNK is a MAPK that mediates both pro-apoptotic and pro-survival signaling [94,99,100]. JNK is phosphorylated under stresses, such as elevated ROS levels [101]. Binding of c-Jun and c-fos (AP-1 complex), the downstream targets of JNK activation, to the promoter regions of DNA initiates transcriptional production of respective genes [83,84,102]. Upon exposure to low dose Cd, JNK phosphorylation was increased in a dose dependent manner, peaking at 10 μM, and the expression levels of both VEGF and VEGFR-2 were elevated with an increase in cell viability [38]. Inhibition of JNK activation substantially decreased VEGFR-2 expression in HUVECs [38]. JNK also mediates sustained VEGFR-2 phosphorylation, which is essential for the transduction of VEGF signaling [103] (Figure 4). Furthermore, inhibitor of JNK reduces the release of VEGF by human coronary smooth muscle cells [104]. Hence, JNK activation is essential for VEGF production by non-endothelial cells [104]. Therefore, in addition to maintaining VEGFR-2 phosphorylation, JNK activation by low-dose Cd exposure might promote tumor angiogenesis by up-regulating both VEGF and VEGFR-2 [38,103,104].

p38 MAPK signaling
Similar to JNK, p38 MAPK is also a mediator of both pro-survival and pro-apoptotic signaling [105,106]. Meanwhile, it is required for VEGF-induced EC migration [106]. Inhibition of p38 MAPK signaling also resulted in reduced VEGFR-2 expression in HUVECs [38]. Nonetheless, the pro-angiogenic activity of p38 MAPK remains to be comprehensively characterized. A potential explanation for p38 MAPK-mediated increase in VEGFR-2 expression in ECs is the ability of p38 MAPK to activate NF-kB by activating mitogen-and stress-activated kinases (MSK1/2) [38,71,94] (Figure  4). Importantly, it appears that members of the MAPK family, ERK, JNK, and p38 MAPK, produce a plethora  [42,72,138]. VEGF signaling also phosphorylates ERK1/2 via MEK1/2, activating IEX-1, an inhibitor of stress-induced apoptosis [70,113,124,125]. High concentrations of Cd ( > 10 μM) block VEGFR-2 activation, and impaired VEGF signaling leads to EC apoptosis and ultimately attenuated angiogenesis [42]. of down-stream signaling events under the stimulation of low-dose Cd. The overall effect of these is enhanced VEGF signaling and tumor angiogenesis [38]. However, interactions between these signaling events remain to be fully understood.

HIGH-DOSE CD IMPAIRS ANGIOGENESIS BY REDUCING VIABLE ECS
Consistently through the literature, high dose Cd ( > 10 μM) attenuates angiogenesis via the induction of apoptosis [38,[41][42][43]107]. As mentioned previously, both p38 MAPK and JNK mediate stress-induced apoptosis [108]. According to Jung et al., Cd (30 μM) activated all three members of the MAPK family in mouse brain microvascular endothelial cells (bEnd.3) while leading to apoptosis [43]. However, only the inhibition of p38 MAPK results in improved survival, suggesting p38 MAPK is the only active MAPK that mediates high-dose Cd-induced EC apoptosis [43]. Indeed, p38 MAPK activation increases the levels of pro-apoptotic proteins including Bax and Fas [105,109], and is associated with cleavage of caspase-9 and the subsequent activation of caspase-3 [110][111][112]. p38 MAPK is also postulated to inhibit the pro-survival ERK signaling [16,106]. Inhibited ERK activation reduces phosphorylated 4E-BP1, leading to less eIF-4E release and eventually lower rates of protein synthesis, which presumably involves decreased VEGFR-2 expression in ECs [83,113] (Figure 5). Remarkably, with the supposed inhibition of p38 MAPK, ERK activity following Cdexposure remains elevated [38]. The eventual apoptosis might indicate that the induced level of ERK activation is insufficient for cells to survive under stress of such intensity [38]. As described previously, Cd-stimulated JNK activation, triggering EC apoptosis [67]. Moreover, it appears that the well-known inhibition of JNK by NF-κB is conserved in Cd-treated ECs [67,81,82]. Inhibition of JNK activation by NF-κB might explain the failure of JNK inhibitor to preserve cell viability [43]. Future investigation of the role of NF-κB in ECs treated with high-dose Cd may help explain whether phosphorylated JNK is an essential component of Cd-induced apoptosis.
According to Kim et al., phosphorylation of MAPKs in Cd-treated HUVECs were reduced to basal levels when the concentration of Cd exceeded 10 μM but the level of pro-caspase-3 increased with elevation in Cd concentration [38]. Hence, caspase-3 is the major contributor to Cdrelated damage in ECs [38,43]. Although activation of either JNK or p38 seems to be required for apoptosis mediated by caspase-3 [43,114,115], the lack of obvious change in levels of these MAPKs suggests that Cdinduced EC apoptosis might involve other activators of caspase-3 [38,116]. Therefore, regardless of the up-stream signaling cascade involved, high-dose Cd inhibits tumor angiogenesis by inducing caspase-3-mediated apoptosis [38,43].
With similar Cd concentrations and similar exposure time, Kim et al. observed MAPK activation patterns that were inconsistent with the results of Jung et al. [38,43]. The only apparent difference between these studies is the variation in EC types [38,43]. It is well established that ECs with distinct origins exhibit different gene expression patterns, enzymatic activity, and signal transduction [117][118][119]. In particular, differences in VEGF-induced MAPK activation have long been recognized to depend on the origin of ECs [120]. Treating primary cultures of HUVEC, human aortic EC, and human microvascular EC with the same doses of VEGF for the same time period resulted in differential ERK activation [120]. The cerebral vascular EC used by Jung and colleagues possesses distinct protein expression pattern from peripheral vascular ECs, potentially resulting in disparities in protein activation [48,121]. Therefore, endothelial heterogeneity may explain the differences in MAPK phosphorylation pattern across these studies.
In addition, Cd in high concentrations disrupts signaling pathways that are important to vascular maintenance and growth [3,41,42,122]. A high concentration of Cd significantly decreases both VEGF and VEGFR-2 expression, and thus impairs VEGF signaling [42]. In addition to promoting tumor angiogenesis, VEGF signaling protects against Cdinduced apoptosis through a number of mechanisms [70,80]. VEGF up-regulates Bcl-2, which promotes survival by inhibiting caspase activation [42]. VEGF also activates Akt to stimulate production of pro-survival proteins such as survivin [12,72]. Furthermore, VEGF triggers ERK1/2 activation via MEK1/2 [70,113,123]. ERK activation phosphorylates IEX-1 and inhibits stressinduced apoptosis [124,125] (Figure 6). In addition, high-dose Cd inhibits the activation of VEGFR-2 tyrosine kinase (TK) activity [126] (Figure 6). Cd might chelate to ATP and form Cd-ATP which could compete with Mg-ATP for enzyme activation sites on VEGFR-2 TKs [126]. With the assumption of Cd-ATP being a slow substrate for TK activation, the accumulation of Cd-ATP could inhibit VEGFR-2 phosphorylation [126]. Alternatively, high dose Cd competes with Mg for the putative second metalbinding site on VEGFR-2 [126]. Therefore, high dose Cd inhibits the activation of VEGFR-2, preventing the downstream pro-survival signaling transduction [126]. Together, these mechanisms aggravate the cytotoxicity of high-dose Cd while attenuating angiogenesis [42]. Furthermore, high-dose Cd (50 μM) increased the level of angiopoetin-2 (Ang-2) while impairing vascular growth in chick embryos [41]. In the absence of VEGF, high concentrations of Ang-2 inhibit Tie-2 signaling as it displaces Ang-1, the more active ligand, from the receptor [41,127]. Since Tie-2 signaling is required for both vascular maintenance and response to angiogenic stimuli, impairment of Tie-2 signaling by high-dose Cd might attenuate angiogenesis [3,41,122,128]. www.impactjournals.com/oncotarget

MECHANISM UNDERLYING THE DOSE DEPENDENCY OF CD ACTIONS
The effect of Cd has been characterized as dose dependent both in vitro and in vivo [31, [129][130][131]. However, the mechanisms underlying such a property remain elusive. Oxidative stress, characterized by elevated level of ROS, is the primary mediator of Cd toxicity [9,132,133], but damage caused by low levels of oxidative stress can be neutralized by anti-oxidant enzymes [134]. After exposure to low-dose Cd, expression and activity of antioxidant enzymes including MT, catalase, glutathione S-transferase, glutathione peroxidase, and quinone oxidoreductase were substantially increased along with the cellular level of GSH [32, [134][135][136]. Subapoptotic levels of oxidative stress also trigger adaptive responses in affected cells [32]. Protection by antioxidant enzymes together with the activation of pro-survival signaling contribute to enhanced cell proliferation [137]. Excessive oxidative stress, however, overwhelms the cellular defense mechanisms and initiates apoptosis to dispose of the damaged cells [137]. Prolonged exposure to increased concentrations of Cd induces a decrease in intracellular GSH level despite the antioxidant enzyme activity, resulting in a significant reduction in cell viability [134]. Therefore, variation in the level of oxidative stress by exposure to Cd of different concentrations potentially explains the dose dependent effect of Cd.
The variations of the results might also be caused by different experimental settings. Here we propose an investigation using HUVEC as the only cell type of interest. The effects on cultured HUVECs are to be evaluated after exposure to different concentrations of Cd representing ultra-low dose, low dose, and high dose. Oxidative stress characterized by intracellular levels of ROS and GSH could be examined to validate our hypothesis that oxidative stress associated with various concentrations of Cd is responsible for the dose-dependent effects of Cd on tumor angiogenesis. Notably, the majority of studies concerning the effect of Cd on angiogenesis have been carried out only in vitro. More in vivo studies are needed to thoroughly elucidate the effects of different concentrations of Cd on angiogenesis and relevant signaling cascades.
Variations in levels of oxidative stress induced by different concentrations of Cd might explain the dose dependent effect of Cd on tumor angiogenesis. Cells are protected from low levels of oxidative stress owing to antioxidant enzyme activities [134,137]. Subapoptotic levels of oxidative stress trigger adaptive responses, promoting cell survival and proliferation [32]. Excessive oxidative stress induced by high-dose Cd initiates apoptosis of ECs [137]. Mechanisms underlying the effects of Cd on tumor angiogenesis still need to be elucidated. Future investigations using in vivo models are needed to further validate current findings.