The splicing process provides an opportunity for regulation of both the final protein structure and the level of protein expression. Different combinations of exons can be retained or removed to create a diverse array of mature mRNAs from the single pre-mRNA. In addition, alternative splicing within the non-coding regions of the mRNA controls the final number of regulatory elements such as translation enhancers or RNA stability domains that can significantly affect protein expression levels.

Through alternative splicing , nine isoforms of VEGF -A have been first found which are named according to the total number of amino acids in the mature proteins: 111, 121, 145, 148, 162, 165, 183, 189 and 206. In addition, some of these isoforms exist in VEGF-Axxxa or VEGF-Axxxb versions (Fig. 8.1). The VEGF-A gene contains eight exons. 1–4 are constitutive, being present in all isoforms, with exon 1 encoding an N-terminal signal of 26 hydrophobic amino acids that is typical of secreted proteins. Exons 5–8 are alternatively spliced to produce the different sized isoforms, with exons 6 and 7 containing basic residues that confer an affinity for heparin-binding. Heparin sulphate proteoglycans are present at the cell surface and within the extracellular matrix, thus the presence or absence of exons 6 and 7 affects the ability of that secreted VEGF-A isoform to diffuse away from the cell. In this way the presence or absence of exons 6 and/or 7 controls the spatial distribution and bioavailability of the VEGF-A isoforms (Vempati et al. 2011). As an example, VEGF-A121 lacks exons 6 and 7 thus does not bind heparin and is free to diffuse away after being released from the cell (Fig. 8.1). In contrast the VEGF-A165 and 189 isoforms are able to bind to heparin sulfate on the cell surface and in the extracellular matrix (Houck et al. 1992). These different heparin binding affinities result in the formation of a VEGF-A gradient which is essential for the process of angiogenesis, with soluble isoforms acting at distal sites to promote vascular recruitment and the extracellular membrane-associated isoforms acting locally to promote the expansion of capillary beds (Grunstein et al. 2000). The fact that an identical isoform can have distinct activities at different anatomical sites suggests that the microenvironment of individual tissues dictates VEGF-A function (Guo et al. 2001).

The C-terminus of VEGF -A is encoded by exon 8 and contains an alternative 3′ splice site that gives rise to the VEGF-Axxxa or VEGF-Axxxb versions of the isoform. Even if there is still controversy regarding the existence of the VEGF-Axxxb isoforms (Harris et al. 2012), in vitro and in vivo studies have demonstrated that the VEGF-Axxx (or -xxxa) isoforms are pro-angiogenic and are up-regulated in tumours whereas the VEGF-Axxxb isoforms are anti-angiogenic and are down-regulated in tumours (Bates et al. 2002; Woolard et al. 2004; Pritchard-Jones et al. 2007). To mediate their anti-angiogenic properties, the VEGF-Axxxb isoforms are thought to bind to the VEGF-A receptors but impair their downstream signaling (Woolard et al. 2004; Cebe Suarez et al. 2006), since when the VEGF-Axxxb and -xxx isoforms were co-expressed they acted in a dominant negative way and only partial receptor agonist activity was detected. This also suggests that the inhibitory function of the xxxb isoforms is mediated through competitive binding (Kawamura et al. 2008; Woolard et al. 2009). Further evidence for this has been found using recombinant human VEGF-A165b which possesses similar affinity towards the anti-Vascular Endothelial Growth Factor antibody (bevacizumab) as that of VEGF-A165, supporting the idea of inhibition through competitive binding. Interestingly, this also suggests that the balance between the expression of the anti-angiogenic and pro-angiogenic isoforms could regulate tumour growth and in the same way the anti-angiogenic xxxb isoforms could affect the sensitivity of tumours to bevacizumab through competitive binding (Varey et al. 2008). It was also demonstrated that the balance between the expression of the anti and pro-angiogenic isoforms could regulate follicle development (McFee et al. 2012), spermatogonial stem cell homeostasis in vivo (Caires et al. 2012) but also plays a critical role in the regulation of glomerular permeability (Oltean et al. 2012).

Recombinant human VEGF -A165b shows anti-angiogenic activity in eye hypoxia -driven angiogenesis (Konopatskaya et al. 2006). In addition, recombinant human VEGF-A165b possesses a similar affinity towards the anti-VEGF antibody (bevacizumab) as that of VEGF-A165, supporting the idea of inhibition through competitive binding . Interestingly, this also suggests that the balance between the expression of the anti-angiogenic and pro-angiogenic isoforms could regulate tumour growth and in the same way the anti-angiogenic xxxb isoforms could affect the sensitivity of tumours to bevacizumab through competitive binding (Varey et al. 2008).

The distal splice site used to generate the VEGF -Axxxb isoforms is 66 nucleotides further along the gene from the proximal splice site (Figs. 8.1 and 8.2) and results in a Ser-Leu-Thr-Arg-Lys-Asp C-terminal (Bates et al. 2002). This is very different from the Cys-Asp-Lys-Pro-Arg-Arg C-terminal generated by the proximal splice site in VEGF-Axxx isoforms and undoubtedly results in a distinct tertiary structure of the VEGF-Axxxb isoforms since they contain an acidic residue (Asp) in place of the Cys-160 present in the VEGF-Axxx isoforms, which forms a disulphide bond with the Cys-146 from exon 7. This was clearly proven for VEGF-A165 when the highly-charged C-terminal Pro-Arg-Arg tail of the VEGF-A165 isoform was replaced by the neutral Arg-Lys-Asp of the VEGF-A165b isoform and resulted in a profound alteration of the structure-function relationship of VEGF-A (Cui et al. 2004). VEGF-Axxxb isoforms have been identified for VEGF-A121, VEGF-A183, VEGF-A145, VEGF-165 (Perrin et al. 2005) and VEGF-A189 (Miller-Kasprzak and Jagodzinski 2008). VEGF-A165b was the first xxxb isoform identified and it is the most widely-studied (Bates et al. 2002). Its over-expression inhibits the growth of prostate carcinoma, Ewing’s sarcoma and renal cell carcinoma in xenografted mouse tumor models (Rennel et al. 2008) and inhibits tumor cell-mediated migration and the proliferation of endothelial cells (Rennel et al. 2008). Interestingly, mammary alveolar development during lactation is also inhibited by VEGF-A165b (Qiu et al. 2008) and endogenous VEGF-A165b contributes to survival of trophoblasts exposed to lower oxygen tensions through an autocrine pathway (Bills et al. 2014).

Although many VEGF -A isoforms have now been identified, most VEGF-A-producing cells are thought to preferentially express VEGF-A121, VEGF-A165 and VEGF-A189. These are known as VEGF-A120, 164 and 188 in the mouse since there is one less amino acid in each mouse VEGF-A isoform. To clearly demonstrate the crucial role of alternative splicing in the regulation of VEGF-A activity, and to investigate the specific roles of the individual isoforms, transgenic mice have been generated that express only a single VEGF-A isoform. Mouse embryos expressing only VEGF-A120 (VEGFA ^120/120 ) show impaired post-natal cardiac angiogenesis, resulting in severe myocardial ischemia, early post-natal death and impaired lung vascular development (Mattot et al. 2002). Half of the embryos die in the perinatal period due to congenital birth defects, with the other half perishing within 2 weeks after birth, in part due to myocardial ischemia (Carmeliet et al. 1999). Similar studies with mice expressing only VEGF-A188 (VEGFA ^188/188 ) also show significant effects, with half of all embryos dying between embryonic stage E9.5 and E13.5 (Stalmans et al. 2002). Interestingly, mice expressing only VEGF-A164 (VEGFA ^164/164 ) are healthy (Stalmans et al. 2002).

Together these transgenic mouse studies suggest that different alternatively spliced isoforms are required at different stages in normal development. Other findings support this since it has been shown that different isoforms are expressed in distinct spatio-temporal patterns both during embryonic development and in adult tissues (Ng et al. 2001). However, it is also known that there is some functional redundancy between isoforms, for example during the initial formation of arch arteries (Stalmans et al. 2003).

The importance of alternative splicing of VEGF -A pre-mRNA in terms of VEGF-A regulation is now well-established. Despite this, very little is known about the mechanisms regulating alternative splicing and thus how the cell controls the levels of the different VEGF-A isoforms. Exonic splicing enhancer/silencer sequences are sequences of DNA that promote/reduce the recognition and use of splice sites. An exonic silencer sequence (ESS) has been proposed to be present in exon 6 of VEGF-A, but the regulatory proteins interacting with this sequence remain unknown (Wang et al. 2009). In the alternative splicing process, Serine/Arginine rich proteins (SR proteins) also play a key role in identifying splice sites. Under hypoxic conditions the expression and phosphorylation of the SR proteins ASF/SF2, SRp20 and SRp40 were shown to correlate with increased VEGF-A expression and a shift towards expression of the VEGF-A121 isoform in the endometrial cancer cell line RL95 (Elias and Dias 2008). Increased expression of VEGF-A121 over the longer isoforms is known to be mediated by two splicing factors of the U2AF65 protein family, namely CAPERalpha and CAPERbeta (Dowhan et al. 2005). SR proteins have also been implicated in the alternative splicing of exon 8, with the ASF/SF2 and SRp40 proteins favoring the use of the pro-angiogenic proximal splice-site, and the SRp55 protein promoting the use of the anti-angiogenic distal splice-site (Nowak et al. 2008). However, other factors are also known to play a role in the pro-/anti-angiogenic decision: the E2F1 transcription factor has been shown to favor the expression of anti-angiogenic VEGF-Axxxb isoforms through a mechanism involving the expression of the splicing factor SC35 (Merdzhanova et al. 2010), and now a link between the Wilms’ tumour suppressor gene (WT1) and the regulation of VEGF-A pro-/anti-angiogenic alternative splicing has been established (Amin et al. 2011). WT1 binds to the promoter region of SRPK1 (Serine/Arginine-rich protein-specific kinase 1), a kinase that regulates the activity of SR proteins. This decreases the expression of the kinase, resulting in decreased AFS/SF2 phosphorylation and nuclear localization, and an increase in expression of anti-angiogenic VEGF-A165b. Conversely, mutation of the WT1 tumor suppressor gene results in ASF/SF2 hyperphosphorylation and expression of pro-angiogenic VEGF-A isoforms (Amin et al. 2011). These data suggest that tumors lacking functional WT1 or containing enhanced SRPK1 expression may control angiogenesis through the regulation of alternative splicing of VEGF-A.

There is no doubt that our understanding of the mechanisms regulating alternative splicing is still in its infancy. A recent study proved that the alternative splicing field still has many more doors to open when it described a novel VEGF -A isoform in the lung tissue of a legally aborted female fetus. This alternatively spliced isoform possessed a 20 bases insertion in the third intron, inducing a frame shift mutation that introduced a stop codon in the middle of the fourth exon (Zhou et al. 2012). The mechanism involved in this splicing regulation and whether this alteration impact VEGF-A protein structure and/or function is still unknown but this discovery highlights the discrepancies in our understanding of this important process.

Reinforcing this biological control, a VEGF-A stop-codon readthrough mechanism has been recently discovered that leads to a 22 amino-acids extension generating new antiangiogenic isoforms (Eswarappa et al. 2014).

In cells different mRNAs have very different stabilities. The more stable an mRNA is the longer its lifetime and hence the more protein is produced from it. Short mRNA half-lives allow cells to alter protein synthesis rapidly when necessary (for example in response to environmental changes such as nutrient levels, cytokines, hormones or stress ) and in this way the control of mRNA stability is an important regulator of protein expression (Mitchell and Tollervey 2000; Shim and Karin 2002).

VEGF -A mRNA is highly unstable under normal oxygen and nutrient conditions with a half-life of 15–40 min in vitro (Dibbens et al. 1999; Ikeda et al. 1995; Levy et al. 1996; Shima et al. 1995). Hypoxic conditions induce VEGF-A expression by increasing mRNA stability through a regulatory process that is independent of HIF1A-induced transcription (Ryan et al. 2000). Like most mRNAs with a short half-life, VEGF-A mRNA contains many AU-rich elements (AREs) which are well-known cis-sequences generally composed of different nonameric (AUUUA) or pentameric (AUUUAUUUA) consensus or U-rich sequences. They increase the rate of degradation of the mRNA and so by blocking or promoting access to these sites the cell can control the stability of the mRNA (Barreau et al. 2005; Gingerich et al. 2004). The VEGF-A AREs are clustered in the 3′UTR section of VEGF-A mRNA (Fig. 8.2). In humans, the stability of VEGF-A mRNA under hypoxic conditions is mediated through the 3′UTR (Claffey et al. 1998; Goldberg-Cohen et al. 2002; Levy et al. 1998), therefore research on the control of VEGF-A expression in response to hypoxia has focused on ARE -binding proteins. Interestingly, it has been shown that degradation of mouse VEGF-A mRNA under normoxic conditions requires the independent action of destabilizing elements from across the mRNA (in the 3′UTR, the coding region and the 5′UTR). In contrast to human VEGF-A mRNA, under hypoxic conditions mouse VEGF-A mRNA requires all three sections and these coordinate to stabilize the mRNA (Dibbens et al. 1999). In humans a number of RNA-binding proteins have been shown to interact with the 3′-UTR elements of VEGF-A mRNA and increase its stability including: the CSD/PTB complex (Cold Shock Domain/Polypyrimidine Tract Binding Protein) (Coles et al. 2004); MDM2, a protein translocated from the nucleus to the cytoplasm under hypoxic conditions (Zhou et al. 2011); hnRNPL, which associates with hnRNP complexes to promote mRNA formation, processing and packaging (Shih and Claffey 1999); the double strand RNA binding protein DRBP76/NF90, which under hypoxic conditions facilitates VEGF-A expression by promoting mRNA loading onto polysomes and translation (Vumbaca et al. 2008); HSP70, which was found to bind and stabilize VEGF-A mRNA through a mechanism independent of its chaperone function (Kishor et al. 2012); and HuR (Levy 1998). HuR is a member of the ELAV protein family, which also includes the Hel-N1, HuC and HuD proteins (Fan and Steitz 1998; King 2000; Levine et al. 1993; Peng et al. 1998). ELAV proteins bind to AREs and are thought to regulate the stability of mRNA from splicing through to translation by generating multimeric “ribonucleosomes” due to their nucleation and cooperative binding properties. For example, HuR bound to the 3′UTR has been shown to interact directly with PAIP2 (Poly(A)-Binding Interacting Protein 2) that is bound to a distinct region of the mRNA, and it has been suggested that interaction between these proteins can influence the overall RNA structure and render it inaccessible to endonucleases (Onesto et al. 2004, 2006). Recently it was also proposed that ELAV/Hu proteins can displace miRNAs (which would otherwise silence the mRNA; see Sect. 3.4.3) and thus suppress their destabilizing properties (Simone and Keene 2013). This seems to be the case for mouse VEGF-A mRNA since the HuR and miR-200b binding sites overlap and it was shown that the miR-200b-induced suppression of VEGF-A expression was competitively antagonized by HuR at the 3′UTR (Chang et al. 2013). Finally, HuR has also been implicated in effective VEGF-A mRNA export from the nucleus and loading onto active polysomes under hypoxia. Along with hnRNPL and hnRNPA1, the extra-nuclear shuttling of these mRNA-binding proteins has also been shown to regulate VEGF-A mRNA stability (Vumbaca et al. 2008).

The 3′UTR ARE of VEGF -A mRNA can also be the target of proteins that destabilize mRNA in various mammalian cell types, for example AUF1 and tristetraprolin (TTP ) (Bernstein and Ross 1989; DeMaria and Brewer 1996; Gorgoni and Gray 2004; Stoecklin et al. 2003; Zhang et al. 2002). AUF1 promotes the assembly of other factors necessary for recruiting the mRNA degradation machinery, such as translation initiation factor eIF4G, heat-shock cognate protein hsc70, lactate dehydrogenase and the poly(A)-binding protein. Interestingly, although the poly(A)-binding protein is a stabilizing factor for polyadenylated mRNAs, it has destabilizing effects on VEGF-A mRNA (Gorgoni and Gray 2004; Ma et al. 2006). It was recently shown that AUF1-mediated regulation of VEGF-A expression in mouse macrophage-like RAW-264.7 cells is mediated through its C-terminus domain that contains a region with three arginine-glycine-glycine (RGG) motifs (Fellows et al. 2012). AUF1 itself can be regulated by many signaling pathways and therefore is subjected to numerous post-translational modifications such as phosphorylation, glycosylation, methylation and ubiquitination (for a review, see Gratacos and Brewer 2010). TIS11/Tristetraprolin (TTP) is another ARE-binding protein that regulates the destabilization of VEGF-A mRNA through a signaling pathway involving casein kinase 2 and p38 MAPK (Lee et al. 2011). These examples show the breadth of cellular signaling pathways involved in the regulation of VEGF-A mRNA stability and the depth of their control by the cell.

Polyadenylation is the addition of a poly(A) tail to the 3′UTR of an mRNA and is important for mRNA stability , nuclear export and translation. Before the poly(A) tail is added, the 3′end of the mRNA is cleaved at specific sites, thus if an mRNA contains more than one polyadenylation site this can produce alternative transcripts, as for alternative splicing .

Although the 3′UTR is known to play a pivotal role in the post-transcriptional control of VEGF -A expression (Bartel 2009), our knowledge of the use and regulation of poly(A) signals is very limited. VEGF-A cDNA sequencing has revealed the presence of two major polyadenylation sites: a consensus AAUAAA site and a non-canonical AUUAAA site, located 399 and 1902 nucleotides after the stop codon respectively (Fig. 8.2) (Claffey et al. 1998; Dibbens et al. 2001). The resulting mRNAs from both sites have been characterized (Leung et al. 1989; Keck et al. 1989), but only one study has investigated the use of VEGF-A alternative polyadenylation (Dibbens et al. 2001). Using a mouse model it reported that the majority of VEGF-A transcripts are processed at the distal polyadenylation site (resulting in mRNAs containing the longer form of the 3′UTR), and that the same site is used under both hypoxic and normoxic growth conditions (Dibbens et al. 2001). Given the large section of the regulatory 3′UTR that is missing in transcripts using the proximal polyadenylation site, it is possible that the resulting VEGF-A isoforms play distinct roles. Therefore, much more work is needed to ascertain the importance of this method of VEGF-A regulation.