2. Alternative Splicing of VEGF and Angiogenesis
Alternative splicing involves the removal of introns from pre-messenger ribonucleic acid (pre-mRNA), with the remaining exons being connected to each other in different combinations to form mRNAs
[11][12]. When dysregulated, the generation of variants that promote tumorigenesis may occur
[13][14][15][16][17][18][19][20][21]. Moreover, some of the variants formed from the alternative splicing of full-length or total VEGF are linked to impaired angiogenesis and tumor progression
[14][15]. The discovery of several VEGF isoforms with distinct functions has revealed that the physiology of VEGF is more complex than previously thought
[16]. Approximately sixteen isoforms have been identified in humans thus far, and six (VEGF-A
111, VEGF
121, VEGF
145, VEGF
165, VEGF
181, VEGF
206) have been studied extensively in terms of structure and biochemical properties
[16]. In mice, VEGF
120, VEGF
164, and VEGF
188 are abundant
[16][17]. The transcripts are assigned numerical values based on the number of amino acids present
[17]. The alternative splicing of VEGF is stimulated by several factors including pH, hypoxia, and nutrient levels
[18]. Acidic environments (~pH 5.5) have been shown to induce alternative splicing of VEGF, resulting in the formation of mainly VEGF
121a, followed by VEGF
165a. The increase in VEGF
121a is often associated with p38 activation
[18]. Other variants, namely, VEGF
145a and VEGF
189 have merely shown slight increases in an acidic pH. In hypoxic conditions, there is a tilt toward the formation of VEGF
165a and VEGF121a
[18]. This shift has been observed in both cancer cell lines and human tumors, although the pattern of expression of these variants depends on the cancer type
[18].
The bioavailability of VEGF variants is dependent on their chemical structure and properties. For instance, VEGF
111a and VEGF
121a do not bind to matrix glycoproteins and therefore can diffuse easily and are thus readily available. On the other hand, VEGF
145a, VEGF
189a, and VEGF
206a bind to heparin and heparin sulfate proteoglycans on the cell surface and in the extracellular matrix (ECM) with the strongest affinity compared to the other isoforms, and as a result, they have the lowest bioavailability
[19]. It is important to note that VEGF
145a and VEGF
206 appear to be rare compared to other variants
[20]. VEGF
165a binds to heparin proteoglycans with intermediate affinity, exhibits moderate to high bioavailability, and is a more potent inducer of angiogenesis than the other isoforms
[20].
3.1. VEGF111a and VEGF121a
VEGF
111a and VEGF
121a exist as highly soluble molecules and are the most bioavailable VEGF isoforms identified to date. VEGF
111a was identified in 2007 and has since been shown to be a potent EC mitogen and an inducer of angiogenesis in vivo
[21]. It can bind to VEGFR-1 and VEGFR-2, although it is the mechanism involving the latter receptor that has received attention. VEGF
111’s mechanism of action through VEGFR-2 appears to be exerted via the protein kinase C (PKC)-extracellular signal-regulated kinase (ERK) 1/2 pathway. In aortic ECs and human vein umbilical endothelial cells (HUVECs) the isoform has been shown to induce the phosphorylation of VEGFR-2, leading to the downstream activation of ERK 1/2
[22]. Although VEGF
111 binds to VEGFR-2, it does not bind sufficiently to the neuropilin-1 (NRP-1) co-receptor to form the NRP-1/VEGFR-2 complex, thus its angiogenic effects might not be as strong
[22]. VEGF
111a is abundant in the lungs and kidneys
[22].
VEGF
121a can also bind to both VEGFR-1 and VEGFR-2, although there is a paucity of data on its binding to the former receptor. The binding of the ligand to VEGFR-2 activates the PI3k signaling pathway, resulting in endothelial cell survival
[23]. It also promotes lymphatic vessel formation, although there is limited data on its mechanism
[23][24]. As well, VEGF
121a activates mitogen-activated protein kinase kinase (MEK) and ERK 1/2, leading to the formation of EC tubes and their maturation. It is a potent inducer of tumorigenesis in experimental models. In mouse xenografts of renal cell carcinoma (RCC) and non-small cell lung carcinoma (NSCLC), VEGF
121a together with another variant, VEGF
165a, were found to promote angiogenesis
[25]. However, an investigation of the effects of this isoform on vascular physiology is necessary to better understand its contribution to tumor angiogenesis and its possible interaction with other isoforms such as VEGF
165a, which is regarded as the prototype of VEGF.
3.2. VEGF165a
VEGF
165a is a moderately diffusible isoform and approximately 60% of the protein is associated with both the cell surface and the ECM
[20]. It can bind to VEGFR-1 and VEGFR-2, as well as to the co-receptor NRP-1
[20][21]. It induces VEGFR-2 phosphorylation leading to signal transduction mainly via protein kinase B (PKB) and ERK 1/2. The activation of PKB leads to EC survival, while the activation of the ERK 1/2 pathway promotes EC proliferation and regulates vessel diameter
[19][20][21]. In vitro studies conducted using Chinese hamster ovary (CHO) cells have shown that VEGF
165a induces the activation of p38 and mitogen-activated protein kinase (MAPK), resulting in the reorganization of the actin cytoskeleton and ultimately promoting cell migration
[19][20][21]. As well, the downstream activation of focal adhesion kinase stimulates the migration of ECs
[20][21]. In addition, when the VEGFR-2 co-receptor, NRP-1 is overexpressed, it potentiates the effects of VEGF
165a, leading to an increase in the proliferative ability of ECs as well as their invasion
[19][20][21][26]. In vivo, VEGF
165a is overexpressed in several cancers and similar to VEGF
111a and VEGF
121a promotes disease progression
[26][27][28][29].
3.3. VEGF165b
The detection of VEGF
165b was initially described by Bates and colleagues following the observation of a reduced expression of the protein in renal cancer tissue when compared to non-cancerous tissue
[30]. The observations were followed by several reports citing the identification of VEGF
165b in various tissues, including the skin
[31][32][33]. Diverse findings have been reported on the functions of the VEGF
xxxb variants, with some reports indicating that VEGF-A
165b results in a far more reduced angiogenic effect when compared to VEGF
xxxa, while other studies have reported that VEGF
165b inhibits angiogenesis
[31][33][34]. Woolard and colleagues observed that VEGF
165b failed to induce the activation of VEGFR-2 in human microvascular endothelial cells
[35]. The anti-angiogenic effects of VEGF
xxxb seem to emanate from its inhibition of VEGF
xxxa’s interaction with VEGFR-2
[36]. The observations from the different studies on the effects of VEGF
xxxb may not necessarily be contradictory but might be due to the influence of the different tissue environments. With respect to the mechanism of the isoform, researchers have found that the binding of VEGF
165b to VEGFR-2 stimulates ERK 1/2 and PKB phosphorylation in ECs, although the induction of these pathways was considerably weak
[37]. Interestingly, the prototype isoform, VEGF
165a, was shown to stimulate mitogen-activated protein kinase (MAPK), while in the same cell line, VEGF
165b did not activate MAPK
[38]. In addition, there was no hydrolysis of phosphoinositol 4,5-biphosphate (PIP2) observed downstream of VEGF
165b-VEGFR-2
[38]. Moreover, Kawamura et al. noticed that VEGF
165b did not induce tube formation in embryonic stem cells or matrigel plugs and poorly induced VEGFR-2 phosphorylation at the Y1052 site
[39]. Taken together, these observations indicate that the ligand has a markedly low ability to induce angiogenesis. Furthermore, a correlation was found between the binding affinity of VEGF
xxxb for NRP-1 and the inability of the ligand to induce angiogenesis
[39]. Of note is that the VEGF isoforms, including VEGF
165b, are expressed differentially in various cancers, and in some instances, their expression appears to correlate with clinical outcomes
[40][41][42]. In addition, the receptors through which these isoforms communicate can also undergo alternative splicing.
3.4. Alternative Splicing of VEGF Receptors
The VEGF receptor-1 exists in two isoforms that are derived from the alternative splicing of an mRNA sequence transcribed from a single gene
[43]. The two isoforms are the transmembrane-bound protein, VEGFR-1, and a soluble polypeptide, sVEGFR-1
[43]. VEGF binding to the membrane-bound VEGFR-1 induces monocyte migration and is linked to the activation of MMPs
[43][44]. An interesting observation is that the promoter region of the membrane-spanning VEGFR-1 has a HIF-1 consensus, and the receptor is thus responsive to hypoxic conditions
[20][45]. Membrane VEGFR-1 appears to be an important link between tumor angiogenesis and immunity, considering that monocytes are not just involved in mediating immunity, but also secrete factors that promote angiogenesis. Then again, sVEGFR-1 can trap VEGF and lower the levels of the free form of this ligand, thus diminishing its ability to induce angiogenesis
[46]. Furthermore, it is worth noting that the angiogenic effects of VEGFR-1 are weak compared to those induced via VEGFR-2.
Alternative splicing of VEGFR-2 yields a full-length receptor and a soluble form that contains only the extracellular domain, sVEGFR-2
[47][48]. However, the latter variant appears to play a more important role in the regulation of lymphangiogenesis rather than angiogenesis, although it has been detected in human umbilical vein endothelial cells (HUVECs)
[47]. VEGFR-2 variants result from partial intron 13 retention. The translation product of the sVEGFR-2 mRNA is a protein with six (instead of seven) Ig-like domains which differ from the full-length VEGFR-2 in that it has a C-terminal sequence that is not found in the latter
[48]. VEGF binding to the membrane-tethered VEGFR-2 isoform results in the phosphorylation of the receptor, leading to the activation of several signaling molecules, including phosphoinositide phospholipase C (PLCγ), phosphatidylinositol (3,4,5)-triphosphate (PIP3) and Ras
[48][49]. PIP3 activates PKB, resulting in the promotion of cell survival and proliferation. Signaling through PLCγ and NO leads to vaso-permeability. In ECs, VEGFR-2 phosphorylation at Y801 activates the PI3k/PKB and eNOS pathways, while the phosphorylation of Y1059 (pY1059) leads to the flux of calcium which activates the MAPK pathway
[50]. The phosphorylation of Y951 (pY951) is associated with cell motility, whereas pY1175 enables a binding site for PLCγ-l
[48][49][50]. VEGF signaling through this VEGFR-2 isoform also induces iNOS, increasing the levels of this enzyme, and ultimately leading to increased vessel permeability
[49][50]. sVEGFR-1 and sVEGFR-2 have been measured in blood samples of breast cancer patients receiving bevacizumab in combination with chemotherapy and both increased significantly following treatment
[49]. However, the significance of this increase is not yet clear, and studies are needed to unravel the clinical implications of the levels of these soluble proteins. On the contrary, the chemistry and regulation of VEGF isoforms have been studied extensively.
3.5. Regulation of VEGF Splicing
Various factors regulate the generation of VEGF variants. In addition to environmental cues such as hypoxia and low pH, several kinases are involved in regulating VEGF splicing. The splicing of VEGF at the proximal splice site is regulated by serine/arginine-protein kinase 1(SRPK1) through the modulation of serine and arginine-rich splicing factor 1 (SRSF1)
[51][52][53]. SRPK1 activation leads to the nuclear translocation of SRSF1 in a heat shock protein (HSp)90-dependent process
[51]. SRSF1 in turn regulates the alternative splicing of various angiogenesis-promoting genes, namely, RON, TREAD1, and VEGF
[51]. Several splice products formed from the proximal splicing such as VEGF
xxxa, are stimulators of angiogenesis. Moreover, TREAD1 activates total VEGF and thus further contributes to the angiogenic process
[51]. The distal splice site is modulated by the splice kinase CDC-like kinase 1 (Clk1) which regulates the splice factor SRSF6
[51]. The product of distal splicing, VEGF
xxxb, appears to reduce angiogenesis. The SRFs that modulate VEGF splicing could potentially serve as targets for altering the splicing switch and restoring the VEGF
xxxa/xxxb ratio. The restoration of the ratio between VEGF
xxxa and VEGF
xxxb is of importance given the roles of these variants in the clinical outcomes of cancer patients.
3. Clinical Implications of VEGF Splice Products
VEGF
xxxa stimulates tumor angiogenesis, while VEGF
xxxb seems to suppress the process by limiting the binding of VEGF
xxxa to VEGFR-2
[35][39]. As a nascent tumor grows, its nutrient and oxygen demand rise, leading to the increased secretion of total VEGF, which in turn is spliced to various isoforms depending on the pH in the TME and the degree of hypoxia
[17][18]. VEGF
165a represents the predominant form in hypoxic conditions, and after binding to the VEGF receptor-2 on the surface of ECs, results in the activation of these cells and their secretion of various molecules, including proteolytic proteins
[17][18][20]. Proteolysis of the basement membrane and ECM components by MMPs and the plasminogen activator (PA) system promotes the incursion of ECs into the tumor stroma
[54][55]. Tip cells lead to new sprouts and prepare the surrounding area for guidance cues. Stalk cells follow and support tip cells. The adhesion of the tip and stalk cells to the extracellular matrix is facilitated by integrins that are expressed by migrating ECs ()
[17][18][54]. Several signaling pathways including Delta-like ligand 4 (DLL4)-Notch signaling interact to regulate sprout formation. The tip cells anastomose with cells from adjacent sprouts to form vessel loops. The final and stabilizing step consists of the construction of adherent junctions and the basement membrane as well as the recruitment of pericytes
[55]. These steps, which constitute the process of sprouting angiogenesis, lead to an increase in tumor vascularization.
The cleavage of full-length VEGF generates isoforms that are expressed differentially in various tissues and cancers
[56][57]. In non-small cell lung cancer (NSCLC) VEGF
111a is overexpressed and is associated with an increase in the occurrence of metastasis
[24]. The isoform is also highly expressed in breast and ovarian carcinomas, although no correlation has been found between its levels and patient outcome
[58]. Another isoform, VEGF
121a, has been shown to be elevated in prostate cancer when compared to normal prostate tissue
[29]. Important to note is that in cancerous prostate tissue, elevated VEGF
121a levels are associated with cancer cell invasion and metastatic dissemination. The increased expression of the ligand in prostate cancer also correlates with hypoxia, which means oxygen deprivation may be an important driver of VEGF
121a overexpression and possibly angiogenesis in this neoplasm.
In human colorectal cancer, VEGF
121 is highly expressed, and its expression is greater in patients exhibiting extensive infiltration of the lymph nodes by cancer cells
[25]. VEGF
121 is also highly expressed in breast cancer and is associated with increased angiogenesis in this neoplasm
[18][58]. Although it is the predominant isoform in human breast cancer tissue, no association has been found between its expression levels and clinical outcome in breast cancer
[58].
VEGF
165a is expressed in most cancers and is the predominant isoform
[18]. In a previous study, it was detected in 70% of renal cell carcinoma (RCC) patients
[25]. However, since the number of patients was not stipulated, the frequency of expression in RCC cannot be deduced from that study. In colorectal cancer, VEGF
165a is the predominantly expressed variant, and it correlates with lymph node infiltration
[25]. Studies on cervical tissue specimens have revealed that VEGF
165a is overexpressed when compared to non-cancerous tissue isolated from the cervix
[27]. Furthermore, the expression of the ligand correlates with lymph node metastasis. Similarly, in patients with renal squamous cell carcinoma (SCC) the overexpression of VEGF
165a is linked to disease recurrence and low disease-free survival while in esophageal cancer VEGF
165a expression correlates with microvessel density (MVD)
[27][28]. Additionally, patients with various cancer overexpressing VEGF
165a have been reported to exhibit increased MVD and poor overall survival
[28]. Interestingly, no such association has been observed between MVD and total VEGF in cancers such as melanoma and esophageal cancer
[28]. This lack of correlation between total VEGF and MVD in some cancers which is observed with VEGF
xxxa may be due to the distinct and differing effects of VEGF
xxxb variants. It is also worth noting that it is not just the levels of total VEGF that plays a key role in regulating the angiogenic process, but also the balance in the levels of the variants. For instance, an increase in the ratio of VEGF
121a to VEGF
165-189a promotes angiogenesis in prostate cancer
[29]. Studies have further shown that changes in the ratio of VEGF
165a to VEGF
165b contribute to disease progression in some cancers
[31][57].
VEGF
xxxb has been detected in several tumors including melanomas that were growing in both the horizontal and vertical phases
[40]. Of note is that the reduced expression of VEGF
xxxb correlated with the development of metastasis in melanoma patients
[40]. The levels of VEGF
165b expression have been found to be lower in cancer tissue when compared to adjacent non-cancerous tissue
[38][41]. Investigations of circulating levels of VEGF
165b in breast cancer patients revealed that the plasma levels of the ligand were significantly lower in cancer patients compared to healthy individuals
[41]. Additionally, VEGF
165b levels increased following treatment and remained high even two years after chemotherapy
[41]. However, no relationship was found between levels of this ligand and the tendency to relapse. VGEGF
165b levels were found to be elevated in 36% of patients with NSCLC and in 46% of lung adenocarcinoma patients
[42]. Although no clinical significance has been attributed to the levels of the ligand either in lung adenocarcinomas or NSCLC, a shift in the ratio of VEGF
165a to VEGF
165b consistently correlated with lymph node metastasis in these patients
[42]. VEGF
165a is not only the prototype for sprouting angiogenesis but has been shown to play a critical role in non-sprouting angiogenesis. The latter, also known as intussusceptive angiogenesis, involves the creation of vessels through the splitting of existing ones and is an alternative mechanism of vascularization that is used by various tumors
[59]. Intussusception is also a mechanism used by tumors to escape anti-angiogenic therapy
[59]. There is a correlation between intussusceptive angiogenesis and the development of resistance to angiogenesis inhibitors. This form of angiogenesis allows tumors to respond to their metabolic needs and to grow. Vascular bifurcation density analysis revealed that another isoform, VEGF
121, is a potent inducer of intussusceptive angiogenesis
[60]. Interestingly, while the administration of either of these isoforms leads to an increase in non-sprouting angiogenesis, their withdrawal results in a reduction in vessel branches and intussusceptive pruning
[60]. The observations from the various studies highlight the importance of the different variants in tumor angiogenesis and the balance in VEGF
xxxa and VEGF
xxxb, however, more work remains to be undertaken in order to determine the roles of the different variants in the prognosis of various cancers.
Improving VEGF-Targeting Approaches
The observation that VEGF isoforms are associated with different neoplasms underscores a need to actively investigate therapeutic molecules that modulate them. The first anti-VEGF drug to be approved by the Food and Drug Association (FDA), bevacizumab, neutralizes total VEGF including the VEGF
xxxa and VEGF
xxxb isoforms by blocking their kinase domain binding sites
[61]. Other anti-angiogenic drugs, cediranib, vandetanib, pazopanib, and sorafenib were also reported to inhibit the angiogenic effects of VEGF
165a [62]. However, given that these drugs are tyrosine kinase inhibitors (TKI’s), their effects could be due to their actions on VEGF receptors and not necessarily on the ligand. As a result, isoforms that signal through kinases inhibited by these drugs will have a diminished effect. Additionally, VEGF and its variants regulate VEGF receptor function. Full-length VEGF can upregulate its canonical receptors, VEGFR-1 and VEGFR-2
[63]. As well, the administration of VEGF
121a leads to an increase in the expression of VEGFR-1 in capillaries, while the presence of VEGF
165a promotes the co-expression of NRP1 and VEGFR-2
[64][65]. Findings from these studies show that the relationship between VEGF and its receptors is not simple, but rather intricate. The complex interaction between VEGF or its variants with VEGF receptors can potentiate angiogenesis, and as a result, more precise approaches are needed to subvert these interactions in a manner that is tumor specific. A plausible approach is the targeting of regulators associated with VEGF splicing, as well as the VEGF splice products which increase receptor activation and promote tumor vascularization. In experimental models, SRPKI was shown to be involved in the promotion of VEGF splicing, resulting in alterations in the VEGF
165a/VEGF
165b ratio that favored increased VEGF
165a formation. Of interest is that Hulse and colleagues reported that the inhibition of SRPKI leads to a decrease in VEGF
165a levels
[66]. Thus, the blockade of SRPK1 could have therapeutic benefits in cancer treatment as it lowers a variant that promotes both sprouting and intussusceptive angiogenesis, namely, VEGF
165a, without depleting VEGF
xxxb. On the other hand, other studies have focused on VEGF
xxxb, an isoform that halts tumor growth in several pre-clinical models. Rennel et al.
[67] investigated the ability of VEGF
165b transfected and non-transfected cells to induce Ewing sarcoma and renal cell carcinoma. Interestingly, tumorigenesis was suppressed in mice injected with cells overexpressing VEGF
165b. The isoform also reduced the ability of VEGF
165a to induce angiogenesis by blocking its binding to VEGFR-2 and thus inhibiting the phosphorylation of the receptor
[39]. The administration of VEGF
165b in a mouse xenograft of human breast cancer effectively reduced angiogenesis and tumor growth
[53]. Varey et al. showed further that colon cancer cells overexpressing VEGF
165b limited the tumor growth in mouse xenografts
[68]. Similarly, Rennel and colleagues observed reduced growth of prostate cancer due to the presence of VEGF
165b overexpressing cells
[67]. Moreover, metastatic colorectal cancer patients with a low VEGF
165b: total VEGF ratio were found to respond better to bevacizumab than those with a high ratio
[68]. These studies signify a possible therapeutic role for VEGF
165b in antiangiogenic approaches. In another study involving breast cancer patients, it was observed that following adjuvant therapy there was prolonged disease-free survival (DFS) and the VEGF
165b levels remained elevated even after 2 years
[69]. The observations from the various studies have shed light on the effects of interventions employing VEGF
xxxb, nonetheless, additional investigations using larger sample sizes are required. Other regulators of VEGF splicing, pH, and hypoxia can also be modulated to inhibit the formation of certain isoforms. For example, the stabilization of VEGF mRNA was achieved following the use of anisomycin
[26]. This approach can be employed to counter VEGF splicing in response to changes in the milieu of tumor cells, such as decreasing pH or reducing the development of hypoxia. Furthermore, the drug abexinostat which targets molecules that promote HIF expression could be useful in suppressing hypoxia. Recently, the drug showed promising results in relapsed lymphoma
[70]. Panobinostat, which also suppresses hypoxia, may also have potential application as part of a combination strategy with treatments that splice variants or their receptors. The drug was evaluated in Phase II clinical trials for the treatment of B-cell lymphoma and showed promising results
[71]. It is also plausible that abexinostat and panobinostat could alleviate the development of resistance to therapies targeting VEGF splicing or its splice products.