Class 3 semaphorins (Sema3) are a family of secreted proteins that were originally identified as axon guidance factors mediating their signal transduction by forming complexes with neuropilins and plexins. However, the wide expression pattern of Sema3 suggested additional functions other than those associated with the nervous system, and indeed many studies have now indicated that Sema3 proteins and their receptors play a role in angiogenesis. The present review specifically focuses on recent evidence for this role in both physiological and pathological angiogenesis.
- class 3 semaphorin
- vascular endothelial growth factor (VEGF)
Structure and binding specificity of class 3 semaphorins (Sema3)
There are eight different classes of semaphorins, of which five classes are found in vertebrates (reviewed in ). Sema3 consist of seven secreted glycoproteins of ~100 kDa containing a conserved 500-amino-acid N-terminal region organized into a seven-blade β-propeller structure called the sema domain, an Ig-like domain and a basic C-terminal domain (Figure 1A). The sema domain is essential for signalling and mediates the specificity of binding to the a1a2 domain of neuropilin 1 or 2 (Np1 or Np2), with Sema3A binding specifically to Np1, Sema3F and Sema3G binding with high affinity to Np2 and Sema3B, Sema3C and Sema3D binding to both neuropilins. The basic C-terminal domain of all these semaphorins binds to the b1b2 domain of both neuropilins. In contrast with the rest of the family, Sema3E does not bind to either neuropilin, but binds directly to Plexin D1.
Structure of neuropilins and plexins
Neuropilins (Np1 and Np2) are 130 kDa transmembrane receptors that share a very similar domain structure despite having only 44% sequence homology. Both neuropilins contain large extracellular domains consisting of two complement-binding-like domains, homologous with C1r and C1s (a1 and a2), two coagulation Factor V/VIII homology-like domains (b1 and b2), and the c domain, which is homologous with MAM (meprin, A5 and receptor tyrosine phosphatase μ), a single membrane-spanning region and a short intracellular domain  (Figure 1B). They were originally identified as receptors for Sema3, and deletion analysis revealed that the a1/a2 and b1/b2 domains are essential for Sema3 binding, whereas the c-domain is thought to be involved in receptor dimerization and interaction with plexins. Neuropilins are also receptors for VEGF (vascular endothelial growth factor) which binds to the b1/b2 domain, with binding enhanced by the a1/a2 domain . Sema3 binding to neuropilins inhibits VEGF activity, with the mechanism originally thought to be due to competition between Sema3 and VEGF for binding to the b1/b2 domain of Np. However, recent structural biology analyses demonstrate Sema3 and VEGF binding simultaneously to Np1, suggesting that competition alone may not cause the inhibition .
Plexins consist of nine genes grouped into four subfamilies with different plexins serving as direct binding receptors for different types of semaphorins. Although Sema3E binds directly to Plexin D1, the remaining Sema3 do not bind directly to plexins, but to the neuropilins which associate with type-A plexins (A1–A4) that serve as the signal transducers of the complex. The extracellular domain of plexins consists of a sema domain which is thought to function as an inhibitory domain preventing activation in the absence of ligand, and PSI (plexin/semaphorin/integrin) and GP (glycine/proline)-rich motifs similar to tyrosine kinases belonging to the Met family  (Figure 1C). The intracellular domain contains a split GTPase-activating domain which is highly conserved throughout the plexin family and is activated in the presence of ligand . Moreover, upon Sema3 stimulation, intracellular kinases such as Fes/Fps and Fyn bind to the intracellular domain of plexins, resulting in receptor phosphorylation and signal transduction .
Role of Sema3 and their receptors in physiological angiogenesis
Physiological angiogenesis occurs mainly during embryogenesis and development, during the menstrual cycle, including the formation of the corpus luteum, and during wound healing. Although Sema3 were originally identified as neuronal guidance molecules, a significant body of evidence has suggested a role in angiogenesis. Indeed, our own research has demonstrated that Sema3A and Sema3F cause significant inhibition in tubule formation and migration of HuDMECs (human dermal microvascular endothelial cells) in vitro in both the absence and presence of VEGF . Both Sema3A and Sema3F are expressed by ECs (endothelial cells), suggesting an autocrine function, and indeed they have both been shown to repulse ECs and stimulate apoptosis even in the absence of VEGF . Furthermore, addition of these recombinant semaphorins in vivo results in inhibition of angiogenesis in a variety of different laboratory models . These data are suggestive of Sema3-induced inhibitory signalling pathways and provide further evidence that Sema3 do not simply compete with VEGF for binding to neuropilins. Indeed, Sema3E, which does not bind neuropilins and therefore does not compete with VEGF, also inhibits angiogenic activity in vitro and in vivo . In contrast with the other Sema3, Sema3C is thought to be a pro-angiogenic factor, enhancing endothelial migration and proliferation in renal glomeruli .
Minimal evidence of the expression and role of Sema3 in physiological angiogenesis has been established to date, with the majority of studies investigating developmental angiogenesis using knockout mice. These have demonstrated mixed phenotypes with Sema3A-null mice being viable and fertile, with one report of vascular defects in the head and trunk on a CD1 background . Sema3B- and Sema3G-null mice are also viable and fertile with no overt pathological abnormalities, whereas Sema3F-null mice, although viable and fertile, also demonstrated abnormalities in limbic circuitry [1,14]. In contrast, Sema3C-null mice die perinatally from cardiovascular defects , and Sema3E-null mice demonstrate severe disorganization of intersomitic vessels . Interestingly, although the mouse knockout phenotypes suggest various compensatory mechanisms, it appears that many Sema3 are primarily EC-expressed [13,14]. Indeed, our own research investigating the expression of Sema3 in angiogenic blood vessels during human wound repair, demonstrated Sema3B EC expression . These data suggest that Sema3 control EC functions in an autocrine manner.
The role of neuropilins in angiogenesis was originally elucidated using ECs which were transfected to express only Np1, only VEGFR2 (VEGF receptor 2) or both. When stimulated with VEGF165, ECs only expressing Np1 do not migrate, whereas those expressing both Np1 and VEGFR2 showed enhanced chemotaxis compared with ECs expressing VEGFR2 alone . Knockout studies confirmed the proposed role for neuropilin with Np1-null and Np1 endothelial-specific knockout mice dying in utero with severe vascular defects observed . Similarly, knockdown of Np1 in zebrafish resulted in inhibition of angiogenesis, but not the initial vasculogenesis . In contrast with Np1, Np2-knockout mice are viable with a reduction of small lymphatic vessels, some avascular regions and blood vessel size heterogeneity .
Further research has shown that, in normal healthy adults, endothelial expression of Np1 and Np2 is modulated in response to oestrogen and progesterone in the uterus  and that Np1 expression changes through the menstrual cycle , suggesting a role for these factors in physiological angiogenesis during the menstrual cycle. Moreover, Np1 expression has been shown to increase during follicular development  and in response to oestrogen and progesterone, but not follicle-stimulating hormone in the ovary . Similarly, our studies have demonstrated that both Np1 and Np2 are involved in human wound repair, where expression is elevated early in the wound healing process, i.e. between 2 and 4 weeks following injury, and correlates with MVD (microvessel density) . Moreover, administration of anti-Np1 antibodies in a mouse model of wound healing resulted in decreased angiogenesis and impaired wound healing, suggesting that Np1 has a functional role in tissue repair . Together, these data suggest that Np1 and Np2 are important for physiological angiogenesis.
Plexins A1–A4 and D1
Although Plexins A1–A4 and D1 are receptors/co-receptors for Sema3, known anti-angiogenic proteins, any role of these transmembrane proteins in physiological angiogenesis has not been fully elucidated. Plexin A1-, Plexin A2-, Plexin A3- and Plexin A4-null mice are all viable and fertile with no vascular abnormalities reported, although the emphasis of these studies was the neuronal system [28–31]. Studies in our laboratory have identified weak EC expression of Plexin A1 in human skin and that EC expression of Plexin A3 correlates with MVD in human wound healing . Interestingly, earlier studies investigating the expression patterns of plexins during bone development demonstrated that, after birth, Plexin A3 was strongly expressed in ECs in the ossification front under the growth plate and in capillaries in the metaphysic regions . These data suggest that Plexin A3 may have a direct role in physiological angiogenesis.
Plexin D1 is known to be expressed in the vasculature and, in contrast with type-A plexin-knockout studies, Plexin D1-null mice survive for 2 days after birth, but have severe cardiovascular defects which ultimately result in death . Similar studies in zebrafish reveal that Plexin D1 knockdown results in highly abnormal intersegmental vessels . These data suggest that Plexin D1 acts in conjunction with Sema3E to control EC positioning and patterning of the developing vasculature. Further studies have shown that Sema3E binding to Plexin D1 results in receptor stimulation interfering with R-Ras function, resulting in activation of Arf6 and filopodial retraction in endothelial tip cells, demonstrating a direct role for Plexin D1 in anti-angiogenic mechanisms . However, as yet, there is no direct evidence for a role of Plexin D1 in postnatal physiological angiogenesis.
Role of Sema3 and their receptors in pathological angiogenesis
Angiogenesis is a key factor in over 50 different pathological conditions, including diabetic retinopathy, rheumatoid arthritis, psoriasis and cancer. There are minimal data on the role of Sema3 in most of these diseases, with only Sema3A investigated in non-cancer diseases. Indeed, Sema3A has been demonstrated to contribute to the vascular regression observed in early diabetic retinopathy, with subsequent repulsion of the regenerating vessels away from the ischaemic regions into the vitreous. Silencing Sema3A resulted in regeneration of the vessels within the ischaemic retina, thereby diminishing aberrant neovascularization in the vitreous . These data therefore suggest that inhibiting Sema3A expression may be a potential therapy for the treatment of diabetic retinopathy. Interestingly, treatment with Sema3A has been shown to reduce inflammation and progression of autoimmune arthritis in vivo through interaction with the T-cell population , although it has yet to be demonstrated whether Sema3A also influences angiogenesis in this setting. In comparison, no change in Sema3A expression was reported in psoriatic lesions compared with normal skin .
A solid tumour cannot progress without initiating angiogenesis and therefore it is unsurprising that the majority of research has investigated the expression and regulation of Sema3 in cancer. Indeed, a loss of Sema3A expression in favour of VEGF may be responsible for the angiogenic switch in multiple myeloma and ovarian carcinoma [38,39]. Furthermore, ovarian carcinoma patients with a high VEGF/Sema3A ratio demonstrate reduced survival when compared with those with a low VEGF/Sema3A ratio . Similarly, low Sema3A expression correlates with high MVD and poor prognosis in meningioma  and Sema3A loss is seen in higher-grade prostate and breast cancers [41,42].
Studies investigating a region on human chromosome 3p21.3 which is commonly deleted in cancer identified the genes for Sema3B and Sema3F . These are both down-regulated in highly metastatic lung and ovarian cancers, suggesting tumour-suppressor activity [44,45]. Indeed, Sema3B and Sema3F have both been demonstrated to inhibit tumour cell activity directly via neuropilin binding, resulting in tumour cell apoptosis . Furthermore, recent investigations have shown that acquired promoter methylation is also responsible for loss of Sema3B expression in lung cancer  and that polymorphisms in these genes which render the protein inactive confer an increased prostate cancer risk and poor survival . Our recent studies investigating expression of Sema3 in normal and hyperplastic breast tissue in situ and invasive breast cancer have shown that Sema3B and Sema3F expression decrease with the switch to the invasive phenotype and demonstrate an inverse correlation with VEGF expression . However, no correlation was seen with MVD in this or any other human studies, despite the fact that these Sema3 have been demonstrated to have potential anti-angiogenic effects in vitro and in vivo. Indeed, upon overexpression in various tumour cell lines, Sema3A, Sema3B and Sema3F have been shown to inhibit tumour growth via inhibition of angiogenesis [49–51].
In contrast with the numerous published studies on Sema3A, Sema3B and Sema3F, only one published study has investigated Sema3C in cancer, demonstrating aberrant expression in metastatic lung disease . These data combined with its reported pro-angiogenic activity suggest that, unlike other Sema3, Sema3C may induce tumour progression. However, our recent data in head and neck squamous cell carcinoma show a decrease in Sema3C expression in cancer compared with normal oral mucosa which correlates with an increase in angiogenesis (C.A. Staton and K. Hunter, unpublished work). As yet, there are no reported studies on expression or role of Sema3D or Sema3G in cancer.
Unlike most other Sema3, which seem to demonstrate both tumour suppressor and anti-angiogenic activities, the role of Sema3E appears to be more complex. It has been suggested that Sema3E is anti-angiogenic and has a tumour-suppressor role in the formation of primary breast tumours ; however, it has also been suggested to be pro-metastatic in many cancers. Indeed, Sema3E expression correlates positively with the metastatic progression of melanoma, colon and liver carcinomas, with Sema3E being more highly expressed in metastatic disease than in non-metastatic disease , and Sema3E overexpression is seen in prostate cancer in contrast with normal prostate tissue . Furthermore, knockdown of Sema3E inhibits the metastatic potential of human cancer cell lines, whereas overexpression promotes metastasis despite inhibiting angiogenesis . However, an earlier study demonstrated that overexpression of Sema3E in a model of metastatic melanoma decreases metastasis . As yet, it is not clear which mechanisms are mediating these contrasting effects of Sema3E, although it has been suggested that it may be due to the relative expressions and contributions of Plexin D1 and Np1 to the receptor complex and subsequent signalling .
Neuropilins have been more extensively studied than the Sema3 with Np1 expression elevated in diabetic retinopathy in conjunction with VEGFR2, where it correlates with MVD . Similarly Np1 expression is elevated in rheumatoid arthritis where it correlates with VEGF and MVD , suggesting a role for Np1 in regulating angiogenesis in this disease. More recent data add weight to this suggestion, since an anti-Np1 peptide inhibited both synoviocyte survival and angiogenesis in a model of rheumatoid arthritis . Furthermore, both Np1 and Np2 are elevated in psoriasis compared with normal skin, although this expression has not been directly related to angiogenesis .
In contrast with these relatively limited studies, the expression and role of neuropilins have been more extensively studied in cancer [2,61]. To summarize briefly, the expression of Np1 and Np2 has been found in the majority of tumours studied to date when assessed by either immunohistochemistry or RT (reverse transcription)–PCR, with an up-regulation of Np1 expression during tumorigenesis compared with normal epithelium. In many cases, increased expression of either or both neuropilins has been found to correlate with tumour angiogenesis, tumour aggressiveness, advanced disease stage and poor prognosis. Experiments overexpressing Np1 in cancer cells have resulted in larger more vascular tumours , thought to be due in part to preventing apoptosis of the tumour cells themselves, as well as enhancing angiogenesis in a juxtacrine manner via an association between tumour cell Np1 and VEGFR2 on nearby ECs [2,18]. Indeed, treatment of subcutaneous tumours with anti-Np1 siRNA (short interfering RNA) results in decreased tumour vasculature and tumour volume , and monoclonal antibodies raised to block VEGF binding to Np1 reduce angiogenesis and demonstrate an additive effect with anti-VEGF antibodies in reducing tumour growth in vivo .
Plexins A1–A4 and D1
Comparatively few studies have been carried out to investigate the expression and role of the plexins in pathological angiogenesis. Indeed, the only studies to date have been performed in cancer. Plexin A1 has been shown to be overexpressed in both colorectal carcinoma  and in gastric cancer, where it positively correlated with increased angiogenesis . In contrast, our recent studies in breast cancer demonstrate a reduction in Plexin A1 expression in invasive cancer and an inverse correlation with angiogenesis . These apparently contradictory findings are likely to be due to interaction with different ligands, as Plexin A1 can interact with the pro-tumorigenic and pro-angiogenic Sema6D, as well as the tumour-suppressor and anti-angiogenic Sema3.
Plexin A2 expression has been demonstrated in malignant glioma cells , whereas Plexin A3 expression is decreased in ovarian and breast cancer compared with normal epithelial cells [42,68], and Plexin A4 expression has not yet been investigated in cancer. In contrast, Plexin D1 expression has been demonstrated on activated tumour vasculature and malignant tumour cells in a variety of primary tumours where it correlates with tumour progression [54,69], and knockdown of Plexin D1 reduces the metastatic potential of cancer cell lines . However, to date, there has been no direct demonstration of a relationship between these plexins and angiogenesis in cancer.
Conclusions and future perspectives
There is no doubt from the evidence to date that Sema3 and their receptors are important modulators of tumour progression, although evidence of a role for these proteins in both physiological angiogenesis and pathological angiogenesis in other disease states is limited. Anti-neuropilin strategies have already reached clinical trials with a Phase I clinical trial currently underway, and Sema3A, Sema3F and Sema3B clearly function both as anti-angiogenic and anti-tumour agents and as such may find use in the future as anti-cancer drugs. However, to date, our understanding of Sema3 signalling is far from complete and precise responses are likely to depend upon which of the receptor/ligand combinations are present in the specific disease.
Advances in the Cellular and Molecular Biology of Angiogenesis: A Biochemical Society/Wellcome Trust Focused Meeting held at the University of Birmingham, U.K., 27–29 June 2011. Organized and Edited by Stuart Egginton and Roy Bicknell (Birmingham, U.K.).
Abbreviations: EC, endothelial cell; MAM, meprin, A5 and receptor tyrosine phosphatase μ; MVD, microvessel density; Np, neuropilin; PSI, plexin/semaphorin/integrin; Sema3, class 3 semaphorin; VEGF, vascular endothelial growth factor; VEGFR2, VEGF receptor 2
- © The Authors Journal compilation © 2011 Biochemical Society