According to the difference in phosphorylating amino acids, protein kinases can be primarily classified into two types: tyrosine kinases (TKs) and serine/threonine kinases (S/TKs). TKs contain two large families: transmembrane receptor-linked kinases (RTKs) and cytoplasmic (non-receptor) tyrosine kinases (CTKs). VEGFR, platelet-derived growth factor receptors (PDGFR), fibroblast growth factor receptor (FGFR) and EGFR belong to RTKs while focal adhesion kinase (FAK), janus kinase (JAK) and sarcoma kinase (Src) are CTKs. Most of these kinases are frequently aberrantly activated in cancer and involved in tumor angiogenesis. The most famous S/TKs include protein kinase C (PKC) and cyclin-dependent kinases (CDKs). Of the 43 marine-derived antiangiogenic agents included in this present review, 14 are able to modulate protein kinases, which can be further divided into TK inhibitors (11) and S/TK modulators (3). These agents will be discussed separately in the second section.

Microtubule and actin are the major structural compositions of cytoskeleton that is involved in many important cellular events including cellular shape, mitosis, movement, signaling transduction and substance transportation. Agents targeting microtubule such as taxol and vincristine are also clinically important anticancer drugs, and some of them show apparent antiangiogenic and/or antivascular properties [10]. Six marine-derived cytoskeleton disturbing agents (plinabulin, soblidotin, spongistatin 1, laulimalide, cytochalasin E and dihydromotuporamine C) have been shown to inhibit angiogenesis and/or to disrupt the established tumor vasculature. Among them, plinabulin (Figure 2) and soblidotin (Figure 3) entered phase II clinical trials.

Among the four microtubule inhibitors, three (plinabulin, soblidotin and spongistatin 1) inhibit tubulin polymerization while one (laulimalide) promotes it. These inhibitors have distinct binding sites on microtubule. Soblidotin (TZT-1027), a synthetic derivative of dolastatin-10 isolated from the Indian Ocean sea hare Dolabella auricularia, has two microtubule-binding sites of low and high affinity, partially overlapping the vinblastine-binding site of tubulin [11]. Plinabulin (NPI-2358), a synthetic analogue of phenylahistin (NPI-2350) isolated from marine fungus Aspergillus sp., interacts with the interfacial region of α- and β-tubulin, which partially overlaps with the colchicine binding site [12]. The binding site of spongistatin 1 (Figure 1) is unknown but different from that of vinblastine [13]. In contrast, the tubulin polymerization agent laulimalide (Figure 1) binds to the exterior of the microtubule on β-tubulin, a binding site different from the taxol-binding site [14,15]. Despite these differences, all the 4 inhibitors possess potent cytotoxicity in various tumor cells in vitro and in vivo. They also cause cell cycle arrest in the G₂/M phase and induce apoptosis [16,17,18,19]. And also due to the differences in their binding to tubulin or their unique structural characteristics, some of them such as plinabulin and laulimalide are highly effective even in cancer cells with resistance to taxol or multidrug resistance (MDR) [14,20].

These four microtubule inhibitors show potent angiogenesis inhibitory activity in a variety of angiogenesis experimental models (Table 1). Some inhibitors could prominently inhibit angiogenesis at much lower concentrations/doses than cytotoxic ones [21,22], suggesting the involvement of additional mechanism(s) apart from their microtubule inhibition. Actually, spongistatin 1 could inhibit the enzymatic activity and intracellular translocation of PKCα that is an essential kinase in angiogenesis [21]. Laulimalide could block integrin-associated signaling pathways through suppressing the phosphorylation of FAK and paxillin and the association of VEGFR-2 with FAK and heat shock protein 90 (Hsp90), although it did not inhibit the phosphorylation of VEGFR-2 [22]. In addition to its antiangiogenic activity, soblidotin shows its antivascular effect at a little higher concentrations, which might be independent of its microtubule inhibition [23,24]. Of note, plinabulin demonstrates significant vascular disrupting activity by acting on established tumor blood vessels. This activity has been considered to result from its inhibiting tubulin polymerization, which impairs cytoskeletal function and leads to selective tumor vascular endothelial architectural destabilization and vascular collapse [25].

Both cytochalasin E (Figure 1) and dihydromotuporamine C (Figure 1) target actin. However, cytochalasin E leads to actin depolymerization only at relatively high concentrations. Cytochalasin E at lower concentrations preferentially inhibits endothelial cell proliferation without disrupting actin stress fibers [26]. It also suppresses the phosphorylation of FAK, which may be critical to its antitumor activity [27]. Interestingly, cytochalasin E was reported to activate PKC, leading to the induction of interleukin-8 production and the up-regulation of CD54 [28,29]. Dihydromotuporamine C is one of nine members in the motuporamines family (motuporamine A–I). It could increase the thickness and number of cytoplasmic actin stress fibers, induce the formation of new stress fibers and large focal adhesion complexes, activates the small GTP-binding protein Rho, and stimulate the Rho-dependent sodium-proton exchanger activity. Its ability to activate Rho has been considered to be one important determinant of its anti-invasive activity [30]. As other members in this family, dihydromotuporamine C shows very low cytotoxicity or cell proliferation inhibition. Both cytochalasin E and dihydromotuporamine C displayed antiangiogenesis activity in different in vitro and in vivo models (Table 1) [26,31].

HDACs are the enzymes that remove acetyl groups from target proteins that regulate their activity. Eighteen human HDACs have been found that can be divided into four classes (class I–IV). HDACs are important anticancer targets, and 2 HDAC inhibitors were approved by FDA (vorinostat in 2006 and romidepsin in 2009) for treating cutaneous T-cell lymphoma [32]. Three marine-derived HDAC inhibitors including panobinostat (Figure 4), azumamides A–E (Figure 3) and psammaplin A (Figure 4) have been reported to inhibit angiogenesis. Panobinostat (LBH589), one of the most potent HDAC inhibitors, can inhibit all known HDACs, which leads to G₂/M cell cycle arrest and proliferation inhibition [33]. At noncytotoxic concentrations, panobinostat also inhibits the phosphorylation of protein kinase B (AKT) and extracellular signal-regulated kinase (ERK) 1/2, and the expression of the chemokine receptor CXCR4. Panobinostat exerts antiangiogenesis activity evidenced by its inhibition of endothelial tube formation and Matrigel invasion, and its reduction of angiogenesis in PC-3 xenografts in mice [34]. Azumamides A–E are HDAC inhibitory cyclic tetrapeptides isolated from the marine sponge Mycale izuensis. Azumamide A displayed significant inhibition of vascular formation [35]. The data from an in vitro vascular organization model using mouse induced pluripotent stem (iPS) cells revealed that the antiangiogenic activity of azumamides was closely correlated with their HDAC inhibition [36]. Psammaplin A from marine sponges potently inhibits HDAC enzyme (an IC₅₀ value of 4.2 nM) [37] and relatively weakly suppresses mammalian aminopeptidase N (an IC₅₀ value of 18 μM) [38]. Psammaplin A causes upregulation of the CDK inhibitor p21^WAF1 and downregulation of pRb, cyclins and CDKs [39]. Psammaplin A has potent antiproliferative activity against various tumor cells and significant antiangiogenesis effects in different vascular formation models (Table 1) [38,40].

MetAPs are the essential metalloproteins that remove the initiator NH₂-terminal methionine from nascent proteins. Two types of MetAP enzymes, MetAP-1 and MetAP-2, have been reported. MetAP-2 has been investigated as anticancer and antiangiogenesis targets since the discovery that the marine natural product fumagillin (Figure 4) irreversibly inhibits MetAP-2 (rather than MetAP-1), impairs angiogenesis, and elicits anticancer activity [41,42,43]. Fumagillin covalently binds to His231 of MetAP-2 that is located in the active site of the enzyme and thus prevents the binding of substrates to MetAP-2 [43]. Nevertheless, how MetAP-2 regulates angiogenesis and how fumagillin inhibits it remain to be clarified. TNP-470 (Figure 4) is a semisynthetic analogue of fumagillin with 50-fold increase in inhibiting neovascularization and angiogenesis. TNP-470 potently and selectively acts on endothelial cells. It inhibits endothelial proliferation in vitro at a concentration 3 logs lower than that required to inhibit the proliferation of fibroblasts and tumor cells [44]. TNP-470 was also reported to inhibit FGF1-stimulated angiogenesis both in vitro and in vivo through the binding to the cytoplasmic domain of the FGF1 receptor [45].

Different from fumagillin, the marine natural products bengamides inhibit both MetAP-1 and MetAP-2 with similar potency. Different bengamides can bind to the active center of the enzymes but in differential binding modes [46]. LAF389 (Figure 4) is a synthetic analogue of bengamide B. LAF389 significantly inhibits tumor growth and proliferation, apparently by means of MetAP inhibition [47,48]. LAF389 demonstrates a marked inhibition of tumor vascularization in animal experiments. However, no antiangiogenic activity data was obtained from patients owing to the early discontinuation of its phase I clinical trials [49].

There are additional marine natural products that show antiangiogenic activity in different models (the tube formation assay was used most widely) (Table 1) but via unique or unknown mechanisms. Some of them were reported only in few studies. For example, puupehenone (Figure 2) was revealed to inhibit the tube formation of endothelial cells and the in vivo angiogenesis in the CAM assay [50]; streptochlorin (Figure 2) suppresses the invasion and VEGF-stimulated tube formation of endothelial cells [51]; the lipopeptide somocystinamide A (Figure 3) impairs the proliferation and tube formation of endothelial cells and inhibits the angiogenesis in zebrafish or CAM models [52]; bastadin 6 (Figure 1), pyripyropenes A (Figure 2) and isomarabarican (Figure 2) highly selectively inhibit the proliferation, migration and tube formation of endothelial cells, and bastadin 6 also blocks the VEGF- or bFGF-induced in vivo angiogenesis of mice corneas [53,54,55]; and streptopyrrolidine (Figure 4) significantly blocks the tube formation of endothelial cells at a non-cytotoxic concentration [56]. Several saccharides mainly from marine algae such as SargA from Sargassum stenophyllum [57], sulfated galactan from Codium cylindricum [58] and a series of fucoidans from different brown algae species, also show antiangiogenic activity to different degrees [59,60]. In addition, there are still some other marine-derived agents that can interfere with angiogenesis-related signaling although no direct antiangiogenesis activity has been reported yet. Two examples are laurenditerpenol (Figure 2) and neolamellarin A (Figure 4), both of which can inhibit the activation of hypoxia-inducible factor (HIF)-1, a critical event in tumor neoangiogenesis [61,62].

In contrast, another three antiangiogenesis agents have been investigated in more detail and all of them entered clinical trials. The marine-derived depsipeptide plitidepsin (dehydrodidemnin B, aplidine) (Figure 3), the aminosterol squalamine (Figure 2) and the proteasome inhibitor marizomib (NPI-0052, salinosporamide A) (Figure 4) all show the antiangiogenic effect in multiple in vitro and in vivo angiogenesis models (Table 1). Plitidepsin could block the secretion of VEGF by tumor cells and the production of MMP by endothelial cells [63], and reduce the expression of anigiogenesis-related genes including VEGF and HIF-1 [64]. Its antiangiogenic activity contributed to its antitumor effects [65]. However, the primary target(s) of plitidepsin remain to be discovered. Squalamine alters intracellular pH and impedes the growth factor signaling, possibly by selectively inhibiting the sodium-hydrogen antiporter exchanger of the endothelial cell surface. It exerts the greatest effect on newly emerging vessels with no appreciable effect on unstimulated endothelial cells [66,67,68]. Additionally, squalamine shows potential antitubercular and antimicrobial activities [69,70]. Marizomib inhibits 20S proteasome by covalently modifying its active site threonine residues and exhibits therapeutic activity against a wide variety of tumors [71]. Marizomib significantly interferes with several angiogenesis-associated factors, including VEGF-dependent migration [72], TNF-α-induced VEGF expression [73], and downregulation of HIF-1α and HIF-2α proteins and VEGF secretion [74], inhibition of both constitutive and inducible prosurvival transcription factor NF-κB [73,74].

S/TKs are a large protein kinase family, including PKC, CDKs and Rho-associated, coiled-coil containing protein kinase (ROCK). Several S/TK-targeted marine-derived natural products showed antiangiogenic activity in different models (Table 1). Of them, bryostatin-1 has entered phase II clinical trials.

Numerous TKs such as VEGFR, EGFR, PDGFR and FGFR play critical roles in tumor angiogenesis, which are the major target (s) of most small-molecule angiogenesis inhibitors in the clinic including sorafenib and sunitinib [7]. As yet, however, only 11 TK-targeted marine-derived angiogenesis inhibitors have been reported, not so many as expected. Among them, neovastat (AE-941) entered phase III clinical trials but failed unfortunately, and the others are still in very early investigation stages.

Since 1993, over 80 separate phase I and II clinical trials for cancer therapy using bryostatin-1 alone or in combination with conventional anticancer drugs have been conducted [8]. Different types of human cancers have been investigated in these clinical studies, including chronic lymphocytic leukemia, non-Hodgkin lymphoma, multiple myeloma, melanoma, renal cell carcinoma, colorectal cancer, soft tissue sarcoma, head and neck cancer, ovarian cancer, cervix carcinoma, gastric cancer, gastroesophageal cancer, esophageal cancer and pancreatic carcinoma [79,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144,145,146,147,148,149,150]. From the results of its clinical trials, several important conclusions can be drawn: (1) Myalgia is its main human dose-limiting toxicity with the occurrence of 10%–87.5% found in phase II trials [151]. Although the precise mechanism of bryostatin-1-triggered myalgia remains unknown, it could be manageable by some measures such as exercise, taking analgesics, and dose control. Nevertheless, myalgia is still an important reason for ending clinical studies [151]. (2) Bryostatin-1 alone failed to produce adequate therapeutic effects in almost all examined hematological malignancies and solid tumors. (3) However, its combination with various conventional anticancer drugs seems encouraging. In particular, the combination of bryostatin-1 with paclitaxel, cisplatin or vincristine in phase II clinical studies produced promising therapeutic effects [146,147,148,150]. (4) The important characteristics of bryostatin-1 in these combinations might include its circumvention of tumor MDR and no or limited overlapping toxicities with the conventional anticancer drugs. Therefore, bryostatin-1 is possible to prevent or delay the occurrence of tumor MDR to the combined drug (s) such as paclitaxel or vincristine [148]. On the other hand, the toxicities of the combinations would not increase compared with those of their separate uses [75]. (5) As a PKC activator, bryostatin-1 is possible to act as a tumor promoter to accelerate the growth of some tumors although it has little tumor-promoting activity [75,152]. To reduce this risk, it may need to develop appropriate molecular biomarkers for the selection of patients and the therapeutic surveillance in its clinical studies and potential uses.

In addition, although the antiangiogenic activity of bryostatin-1 was shown in preclinical studies, it was not reported in any clinical trials. So it could not be concluded whether the antiangiogenic activity practically contributes to the clinical anticancer effects of bryostatin-1. According to the present clinical results, the clinical development of bryostatin-1 is faced with several major challenges, including how to reduce its toxicity (mainly myalgia), increase its therapeutic efficacy (mainly through combination with other anticancer drugs), and develop proper molecular biomarkers for the selection of patients and the therapeutic surveillance. Another problem is its very limited yield from natural resources. Moreover, it is also hard to be synthesized [75]. However, because of its unique PKC modulation and anticancer activity, more potent analogues of bryostatin-1 deserve testing in the clinical arena.

The marine-derived angiogenesis inhibitor panobinostat is among the most potent HDAC inhibitors [153]. It is undergoing extensive phase I and phase II clinical trials with 114 clinical studies [154]. Panobinostat is orally or intravenously available [153]. The major toxicities of panobinostat in humans are hematologic (including thrombocytopenia, neutropenia, anemia and leukopenia), fatigue, and gastrointestinal (nausea and diarrhea) but generally manageable [155,156,157,158,159,160,161,162]. Although phase II clinical trials of panobinostat have been conducted in solid (e.g., renal and pancreatic) cancers, the reported positive responses to its single-agent therapy have been predominantly observed in advanced hematologic malignancies including T-cell lymphoma, Hodgkin lymphoma, Waldenström macroglobulinemia, and myelodysplastic syndrome [155,156,157,158,159,160,161,162], showing a similar clinical anticancer spectrum to other HDAC inhibitors. However, the response of multiple myeloma to panobinostat monotherapy was just modest [156,157]. The combination of panobinostat with other anticancer drugs could expand its potential clinical applications. Its combination with paclitaxel and carboplatin reported in a phase I study [163] was well tolerated. In contrast, the combination of panobinostat, bevacizumab and everolimus was not tolerable [164]. The results are interesting because all panobinostat, paclitaxel and carboplatin can inhibit bone marrow but both bevacizumab and everolimus have very limited hematologic adverse effects. The results suggest that the toxicity of these combinations may have additional determinants and careful selection of its combined drugs is very important. The results also suggest that in addition to histone hyperacetylation, other biomarkers more relevant to the clinical therapeutic efficacy and toxicity of panobinostat are urgently needed [165].

Plitidepsin is undergoing phase I and phase II clinical studies and its pivotal randomized phase III trial is also ongoing in patients with relapsed/refractory multiple myeloma [166,167]. Phase I results show that plitidepsin has a unique toxicity profile that is similar in children and adult cancer patients [168]. Its main dose-limiting toxicities are muscular and hepatic toxicities. The most common muscular toxicities are myalgia and asthenia with unknown mechanisms, and co-treatments with l-carnitine can prevent them or accelerate their recovery [168,169,170]. Its hepatic toxicities are transaminase increases but generally reversible. Other toxicities of plitidepsin include fatigue, pyrexia, skin rash, abdominal pain, nausea, vomiting and diarrhea. Of note, plitidepsin does not significantly inhibit bone marrow at pharmacological concentrations [168,170]. Moreover, it does not seem to produce cross-resistance between plitidepsin and other conventional anticancer drugs [171]. These two characteristics are important for they provide a critical basis for its clinical combination designs with the drugs that are myelosuppressive and/or sensitive to drug resistance mechanisms.

Phase II studies with plitidepsin monotherapy further confirmed its favorable safety profile, but did not demonstrate significant clinical anticancer activity in solid tumors such as unresectable advanced medullary thyroid carcinoma [172], refractory advanced malignant melanoma [173], pretreated small cell lung cancer [174] and pretreated non-small cell lung cancer [175]. In contrast, the recently released results of a phase II study with single-agent plitidepsin revealed its clinical activity with an overall response rate of 20.7% (6 of 29; 2 complete responses and 4 partial responses) in patients with non-cutaneous peripheral T-cell lymphoma [176]. However, no responses were observed to other non-Hodgkin’s lymphoma including B-cell lymphoma in the same study. Plitidepsin showed synergistic activity with rituximab in preclinical models of diffuse large cell and Burkitt lymphoma [166]. Its combination with dexamethasone in a phase II clinical trial also displayed therapeutic efficacy though limited [177]. Therefore, future clinical development of plitidepsin might predominantly test for its therapeutic efficacy against hematological malignancies, preferentially through combination with other anticancer drugs.

Preclinical data indicate that marizomib inhibits all three enzymatic activities of the proteasome with preferential inhibition of the chymotrypsin-like and trypsin-like proteasome enzymes. Therefore, marizomib could overcome the resistance to bortezomib that selectively inhibits the chymotrypsin-like activity of the proteasome. In addition, this broader spectrum of enzymatic activity inhibition is also likely to confer a differential anticancer spectrum and a different toxicity profile to marizomib. This unique feature provided the basis for the clinical development of marizomib although bortezomib has been approved for clinical uses [178]. Marizomib is undergoing phase I clinical trials. The results available indicate its safety and preliminary anticancer activity. The most common adverse events of marizomib observed in these clinical studies were fatigue, nausea, diarrhea, vomiting, and dysgeusia/anorexia. Its dose-limiting toxicity could be transient hallucinations (visual imprints when eyes closed) and dizziness/unsteady gait. Notably, marizomib did not cause myelosuppression (neutropenia and thrombocytopenia) or peripheral neuropathy that could be related to bortezomib. In addition, the combination with the HDAC inhibitor vorinostat did not enhance the toxicity of marizomib but could synergistically increase its therapeutic efficacy in patients with melanoma, pancreatic and lung cancer [179,180,181]. With marizomib monotherapy, stable disease was observed in patients with multiple myeloma, melanoma, cervical, colorectal, hepatocellular, adenoid cystic, granulosis cell or ovarian carcinoma [180,182]. Nevertheless, marizomib has a very short half-life of 4–10 min with clearance at 3–15 L/min in patients, indicating that it is necessary to carefully adjust the administration schedule of marizomib so as to provide enough drug exposure to achieve clinical benefit [180,183]. Taken together, marizomib is well tolerated with a unique toxicity profile and has demonstrated preliminary therapeutic effects, which warrants its further clinical development.

The tubulin inhibitor plinabulin is a vascular disrupting agent, two phase I trials of which have been reported. The first one was conducted using single-agent plinabulin in 38 patients with advanced solid tumors (including colorectal, pancreatic, breast, renal thyroid, adrenocortical or prostate carcinoma) or lymphomas (diffuse large cell non-Hodgkin’s lymphoma) [184]. In contrast, the second one tested the combination of plinabulin with docetaxel in 13 patients with advanced non-small cell lung cancer, gastrointestinal stromal tumor, liposarcoma or melanoma [25]. The results showed its favorable safety profile with fatigue, tumor pain, nausea, diarrhea and vomiting as the most common but manageable adverse events. Fever, tumor pain, and transient hypertension were also observed. One dose limiting toxicity of nausea, vomiting, dehydration and neutropenia occurred. Drug-related neurologic toxicity or myelosuppression was not significant. Importantly, the combination of plinabulin and docetaxel did not increase their respective toxicities [25,184]. These phase I trials also demonstrated its antitumor activity. Plinabulin monotherapy achieved stable disease with a rate of 30% after 2 cycles [184]. Its combination with docetaxel led to a partial response in two patients and decreased tumors in four patients among total eight evaluable patients with non-small cell lung cancer [25]. These results support its further clinical development of either monotherapy or combination.