Plexin-B1 plays a redundant role during mouse development and in tumour angiogenesis

Background Plexins are a large family of transmembrane receptors for the Semaphorins, known for their role in the assembly of neural circuitry. More recently, Plexins have been implicated in diverse biological functions, including vascular growth, epithelial tissue morphogenesis and tumour development. In particular, PlexinB1, the receptor for Sema4D, has been suggested to play a role in neural development and in tumour angiogenesis, based on in vitro studies. However, the tissue distribution of PlexinB1 has not been extensively studied and the functional relevance of this receptor in vivo still awaits experimental testing. In order to shed light on PlexinB1 function in vivo, we therefore undertook the genomic targeting of the mouse gene to obtain loss of function mutants. Results This study shows that PlexinB1 receptor and its putative ligand, Sema4D, have a selective distribution in nervous and epithelial tissues during development and in the adult. PlexinB1 and Sema4D show largely complementary cell distribution in tissues, consistent with the idea that PlexinB1 acts as the receptor for Sema4D in vivo. Interestingly, PlexinB1 is also expressed in certain tissues in the absence of Sema4D, suggesting Sema4D independent activities. High expression of PlexinB1 was found in lung, kidney, liver and cerebellum. Mutant mice lacking expression of semaphorin receptor PlexinB1 are viable and fertile. Although the axon collapsing activity of Sema4D is impaired in PlexinB1 deficient neurons, we could not detect major defects in development, or in adult histology and basic functional parameters of tissues expressing PlexinB1. Moreover, in the absence of PlexinB1 the angiogenic response induced by orthotopically implanted tumours was not affected, suggesting that the expression of this semaphorin receptor in endothelial cells is redundant. Conclusion Our expression analysis suggests a multifaceted role of PlexinB1 during mouse development and tissue homeostasis in the adult. Nonetheless, the genetic deletion of PlexinB1 does not result in major developmental defects or clear functional abnormalities. We infer that PlexinB1 plays a redundant role in mouse development and it is not strictly required for tumour induced angiogenesis.


Background
Plexins are a highly conserved family of single pass transmembrane receptors which, in mammals, comprises nine genes grouped into four subfamilies (A thru D) based on sequence homology [1]. They are characterized by a conserved sequence, the "sema domain", a structural domain that mediates protein-protein interaction, and phylogenetically links the Plexins to the Semaphorins and the Scatter Factor Receptors [2]. The intracellular domain is highly conserved among the Plexins but does not share striking homology with other known proteins. Although the mechanisms of Plexin-mediated signalling have not been well understood, they are known to impinge on cytoskeletal dynamics and on cell adhesion, e.g. by regulating monomeric GTPases of Rho and Ras families [3].
PlexinB1 mRNA expression has previously been shown in the nervous system [19] and in diverse tissues at E14 of murine development [20]; on the other hand, detailed information on the expression of Sema4D mRNA outside the nervous system is missing. Notably, PlexinB1 and Sema4D expression at protein level by immunohistochemistry has not been reported yet. Furthermore, little is known about the expression of Sema4D and PlexinB1 in the adult tissues.
In spite of extensive studies performed on Plexins of A subfamily, molecular genetic studies to address the functional relevance in vivo of Plexins-B are currently lacking.
Here we report our analysis of PlexinB1 deficient mice, including functional studies to test the relevance of this semaphorin receptor in tumour induced angiogenesis.

PlexinB1 and Sema4D expression during embryo development and in the adult
In order to get insights to their physiological function, we investigated the tissue distributions of PlexinB1 and Sema4D in embryonic and adult tissues. Initially, we explored PlexinB1 mRNA expression by Dot Blot analysis of human tissue ( Figure 1A) and by whole mount In Situ Hybridization of mouse embryos (Additional file 1). Taken together, these data revealed a wide expression of PlexinB1 in various tissues, including nervous system, lung, kidney and liver.
In order to gain further evidence of PlexinB1 expression at the protein level, we generated specific anti-PlexinB1 antibodies (see Additional file 2 for specificity controls), which we used to analyze further embryonic and adult tissues.
The immunohistochemical analysis of PlexinB1 and Sema4D expression revealed an intriguing distribution in developing organs characterized by epithelial branching morphogenesis, including the kidney, the lung and the pancreas. During epithelial branching morphogenesis, tissues are shaped by the outgrowth and branching of tubular epithelial structures, which invade the surrounding mesenchyme under the control of growth factors and Expression of PlexinB1 and Sema4D in in different organs and during branching morphogenesis guidance cues. We found that, at an early stage, PlexinB1 is expressed in the mesenchyme surrounding epithelial tubules in kidney and lung ( Figure 1B, D) and in the pancreas (not shown), while Sema4D is expressed by epithelial cells ( Figure 1C, E).
The developmental expression of this ligand-receptor pair appears to be dynamically regulated. For instance, in the lung at E17, PlexinB1 was expressed in cells of putative smooth muscle lineage recruited to wrap the differentiating airways ( Figure 1F); conversely, in the adult, PlexinB1 expression was mostly localized to bronchial epithelium ( Figure 2A). Notably, while Sema4D expression was clear in developing epithelia ( Figure 1C), it was undetectable in the adult lung (not shown).
In the adult kidney, we found an elevated expression of PlexinB1 in distal tubules and collecting ducts ( Figure  2B), while Sema4D displays a complementary expression in the pars convoluta of proximal tubules ( Figure 2C).
In the mouse liver, we could not clearly detect PlexinB1 expression during early embryonic stages (not shown), though we found a high expression of PlexinB1 in adult hepatocytes ( Figure 2D). Surprisingly, we could not detect Sema4D expression in the liver at any stage (not shown), suggesting Sema4D-independent activities of PlexinB1 in this tissue.
Consistent with previous reports [19], in the nervous system we detected PlexinB1 expression in sensory and motor neurons, but were particularly attracted by the distinctive expression of PlexinB1 and Sema4D during cerebellar development. During postnatal development, Purkinje cell (Pk) dendrites expand into the nascent molecular layer, to form extensive connections with granule cell axons (parallel fibres). In the P10 cerebellum, PlexinB1 is exclusively expressed in Purkinje cells, being distributed in the perikaryon, dendritic tree and axon, up to the terminal branches in the deep cerebellar nuclei (Figure 3A). At the same developmental stage it was shown that Sema4D is expressed by myelinating oligodendrocytes in the folial white matter [21]. In the mature cerebellum PlexinB1 expression remains selectively confined to PK cells. Moreover, labelling intensity of Pk axons now reveals a clear cut rostro-caudal gradient, with higher levels of expression in the posterior cortical lobules (Figures 3B, C). In addition, analysis of frontal sections shows differential expression along parasagittally oriented Pk cell subpopulations ( Figure 3D). At this age, anti-Sema4D antibodies show a diffuse distribution pattern in the molecular layer, suggesting expression in parallel fibres ( Figure 3E).

Generation of PlexinB1 null mutant
To investigate the functional relevance of PlexinB1 in vivo, we undertook the genomic targeting of the murine gene to obtain PlexinB1-deficient mice. To this end, we  In the kidney, Sema4D is mainly expressed in the pars convoluta of the proximal tubules. D) In the adult liver PlexinB1 is diffusely expressed in the liver parenchima.
generated a targeting construct in which we placed exon 22 and 23 of the PlexinB1 gene (encoding the transmembrane domain of the receptor) between two loxP sites ( Figure 4A). Furthermore, we inserted a Neo cassette, flanked by two FRP sites and followed by a LacZ reporter fused to the first exon of the intracellular portion of the PlexinB1 gene; this allowed us, upon FLP recombination, to also obtain knock-in PlexinB1 mutants. The correct targeting of the construct in the genome of ES cells was confirmed by Southern blot analysis ( Figure 4B). Chimeric mice, obtained from a positive ES cell clone, were mated with a CRE deleter strain to achieve gene inactivation, and eventually backcrossed into C57BL/6 genetic background. The actual deletion of exons 22 and 23 in mouse genome, encoding for the transmembrane region of PlexinB1, was confirmed by PCR analysis ( Figure 4C).
By crossing heterozygous mice, we obtained PlexinB1 homozygous mutant mice, which were born and developed normally into adulthood. The genotypic analysis revealed the expected Mendelian ratios in littermates (Table 1). We confirmed the absence of PlexnB1 mRNA in homozygous mutant by RT PCR analysis ( Figure 4D). Moreover, we analyzed tissue extracts of mutant and control mice by Western blotting by means of an antibody directed to the intracellular domain of the receptor, and confirmed that the expression of PlexinB1 was abrogated in homozygous mutant mice at the protein level ( Figure  4E).

PlexinB1 deficient mice do not display major developmental defects
PlexinB1 mutant mice did not show any detectable difference in size and were undistinguishable from their wild type (WT) littermates; in addition, their feeding behaviour and weaning appeared to be regular, they developed normally and did not show defects in reproduction or fertility. Moreover, we measured some of the main blood parameters (i.e. white blood cells counts, red blood cells counts and values, haemoglobin concentration, platelet count) and found them to be normal (data not shown). Thus we undertook a histological analysis of embryonic and adult tissues in PlexinB1 null mice.
Lung development provides a prototypic example of branching morphogenesis and semaphorins and plexins were shown to play a role during this morphogenetic process [22]. However, upon histological examination we failed to reveal any obvious abnormality in the morphology of lung epithelium or in the septation of the airways in PlexinB1 null mice (compare Figure 5A to 5D). Besides, Histology of lung, kidney and liver is normal in PlexinB1 mutant mice    the alveoli were properly inflated and the respiration occurred normally in all the animals analyzed.
In the kidney PlexinB1 is highly expressed; however, the histological analysis of mutant mice did not reveal any gross defect or abnormality in tubules or nephrons (compare Figure 5B to 5E). These specialized structures develop as the result of an intricate dialogue between the epithelial cells and the surrounding mesenchyme, and these results implicate that PlexinB1 is not essential during nephrogenic induction. Furthermore, the number of renal glumeruli were counted and found to be similar in WT and mutant mice ( Figure 5G). Because significant defects in kidney function may not implicate gross histological abnormalities and because PlexinB1 is also expressed in the adult, we performed blood tests on two major indicators of renal function, creatinine and urea levels, which turned out to be comparable in normal and mutant mice (see Tab. 2).
We furthermore examined the liver parenchyma, where PlexinB1 is strongly expressed in WT mice. The haematoxylin/eosin staining revealed no sign of fibrosis, steatosis or other particular abnormality in mutant mice (compare Figure 5C to 5F). We performed blood tests to check two major indicators of liver function: i.e. plasma concentration of fibrinogen and total proteins. In addition, we measured blood levels of hepatic transaminases (glutamic oxalacetic transaminase and glutamic piruvic transaminase), which are elevated in the presence of liver disorders, and we assessed the integrity of biliar tree by measuring the blood levels of gamma-glutamil-transferase and alkaline phosphatase. None of these analyses revealed significant differences between WT and mutant mice ( Table 2).
As mentioned above, we found the highest PlexinB1 protein expression in the central nervous system within the cerebellum. However, in PlexinB1 null mice, the overall cerebellar morphology of the cerebellum appeared normal and we could not reveal any abnormality in the foliation of lobuli or in the number of Pk cells ( Figure 6A, E).
In the cerebellum, myelin formation starts from the first postnatal week, in a period in which Sema4D is expressed by the olygodendrocytes [21]. This prompted us to scrutinize the myelination process of Pk axons by means of confocal analysis. In fact, anti-myelin basic protein (MBP) staining showed a correct wrapping of Pk axons by the glial cells ( Figures 6B, C, D, F, G, and 6H). In addition, Pk axons were immunolabeled with anti-Calbindin (a major Pk cells marker) in the mutant mice, revealing a normal axon patterning. Furthermore, the collateral branching of Pk axons in the infraganglionic plexus, which is normally confined in the most superficial portion of the granular layer [23], did not show any abnormality (see Figure 6B, F). Also, the dendritic arborization of Pk cell in the molecular layer appeared normal in the mutant mice (not shown). These data suggest that, although PlexinB1 is abundantly expressed in Pk cells, its function doesn't seem to be essential for proper development of cerebellar Pk cells.

Axonal growth cones from PlexinB1 mutants fail to collapse in response to Sema4D
We thought to exploit these mutants as a genetic tool to investigate the functional requirement of PlexinB1 in assays previously used to study Sema4D signalling. For instance, it was demonstrated that the axonal growth cones of cultured hippocampal neurons undergo collapse in response to Sema4D in vitro, and it was proposed that this response is mediated by PlexinB1 receptor [9]. We observed that PlexinB1 is expressed by hippocampal neu- rons, although at a lower level compared to Pk cells (not shown); however, the evidence that PlexinB1 is responsible for the transduction of Sema4D repulsive signals in these neurons is still lacking. Thus, to verify whether this functional response is abrogated in PlexinB1 deficient mice, hippocampal neurons derived from WT or mutant mice were cultured for 48 hours ex vivo and then incubated with Sema4D for 1 hour. While the number of collapsed growth cones in wild type neurons was strongly increased upon Sema4D stimulation, PlexinB1 -/-hippocampal neurons did not collapse and underwent normal axonal outgrowth and differentiation, irrespective of Sema4D treatment (Figure 7). This establishes a functional requirement for PlexinB1 in the repelling activity of Sema4D in hippocampal neurons. We then examined the overall histology and the cellular composition of the hippocampus in PlexinB1 knock-out mice. Moreover, we analysed the cytoarchitectonic and the connection pattern of hippocampal neurons (stained for Calbindin marker) and interneurons (immunolabeled with anti-Parvalbumin antibody). However, we did not observe obvious abnormalities of this neural structure in mutant mice (see Additional file 3), consistent with a compensation of Sema4D function in the development of the hippocampus.

PlexinB1 is dispensable for tumour-induced angiogenesis
Sema4D is a potent angiogenetic factor in vivo [16,17] and it is released by many tumour cells [ [18]; and our unpublished data]. Furthermore, it was shown that inhibiting the expression of Sema4D in human head and neck carcinoma cells leads to reduction of tumour burden and vascularity in a mouse xenograft model [18].
In fact, the so called "angiogenic switch", i.e. the ability of tumour cells to recruit new vessels in order to provide blood supply, is a basic step in solid tumour growth and metastatic dissemination. Without overcoming this ratelimiting step, most tumours cannot grow over 2 mm in diameter, nor metastasize. PlexinB1 is known to transduce Sema4D signals in vitro and it is expressed in HUVEC endothelial cells and in the tumour vessels in vivo [16][17][18], however its requirement in tumour angiogenesis has not been clearly established.
We reasoned that, if PlexinB1 would play a major role in this process, its ablation would impact on tumour angiogenesis, leading to reduced growth rate and metastatization of cancer cells. To test this, we exploited the well established B16 melanoma model, which is syngenic with the C57BL6 background of our mutants, and that we found to produce Sema4D (see Additional file 4). Melanoma cells were injected orthotopically in the subcute of WT, heterozygous and homozygous PlexinB1 mutant; tumour growth was followed over time, and tumour explants were eventually weighted at the end of the experiment after 25 days. We did not observe any significant difference between the slope of tumour growth curve or in the final weight of the tumours grown in WT or PlexinB1 null mice ( Figure 8A, B). Moreover, we counted the number of lung metastases formed by tumours in normal and homozygous mutant mice and we did not observe significative differences ( Figure 8C).  To scrutinize further vessel growth within the primary tumours, we stained tissue sections with an antibody against CD31 endothelial marker, and analyzed them by fluorescence microscopy. We could not reveal morphological differences between tumour vessels in mutant and WT mice. In addition, we quantified total vessel area and vessel density (the number of vessels per mm 2 ) in tumour sections. In both cases, we did not measure a significant change in PlexinB1 -/-compared to WT mice.

Discussion
Plexins were originally characterized for their role in the wiring of neural network as receptors for Semaphorin guidance cues. However, accumulating lines of evidence later prompted the investigators to reappraise and revise the biological role of these molecules. First, it was found that many Plexins are widely expressed outside the nervous system. Moreover, as the signal transduction mechanisms start to be uncovered, it is now clear that Plexins can impinge on a range of signaling pathways potentially regulating cell adhesion, cell proliferation and differentiation.
Here we demonstrate, by immunohistochemical analysis, the expression of PlexinB1 receptor and of its ligand Sema4D in diverse epithelial tissues and in the nervous system, suggesting that PlexinB1 might play multiple functional roles in vivo, during development and in the adult stage. In fact, this is the first extensive analysis, in the mouse embryo and in the adult, of PlexinB1 and Sema4D tissue distribution undertaken so far.
Previous expression studies and experiments in vitro had highlighted a potential role for secreted semaphorins and their receptors during tubular morphogenesis in the lung [22] and in the kidney [24], and it was hypothesized that these guidance cues may define permissive and restricted regions for cell migration. Interestingly, based on our expression analysis, Sema4D-PlexinB1 signaling might have a role in this function, for instance at the boundaries between different cell populations. Moreover, PlexinB1 can regulate the activation of Met [13], a tyrosine kinase receptor known to mediate tubular morphogenesis in fetal lung and kidney [25]. In spite of these facts, the genetic deletion of PlexinB1 did not lead to morphological or functional abnormalities, indicating that the function of this semaphorin receptor is redundant in development. Notably, developing kidney and lung also express PlexinB2 [20], another putative Sema4D receptor [26]. Although we observed that PlexinB2 binds to Sema4D with lower affinity than PlexinB1 (J.P. and L.T., unpublished observation), it is conceivable that PlexinB2 can compensate for the lack of PlexinB1 in these organs. In fact, by measuring PlexinB2 mRNA levels in tissues with real time PCR, we found significant expression in liver and kidney, which did not increase in PlexinB1-deficient mice (not shown).
These findings are in line with that found in previously reported mouse mutants of semaphorins or plexins expressed in developing lung (such as Sema3A [27], Sema3F [28], PlexinA1 [29]), which also lack to show obvious abnormalities of these organs [22,30]. Thus, this scenario suggests a high level of redundancy in semaphorin signaling during branching morphogenesis and in the maintenance of tissue homeostasis in lung and kidney. Moreover, since additional guidance cues are known to regulate these developmental processes, it is conceivable that the absence of semaphorin signals may be compensated by functionally related morphogenetic factors such as the Netrins and the Slits (for an extensive review see [31]).
In contrast to the lung and the kidney, PlexinB1 expression in the liver is not consistent with a function as receptor for Sema4D, since we could not detect significant expression of this semaphorin in the mouse liver at any stage. This may suggest that hepatocytes or other liver cells express as yet unidentified alternative ligands for PlexinB1. Furthermore, since it was shown that Plexins can associate in receptor complexes on the cell surface [13,14] or regulate homophilic cell-cell adhesion [32,33], it is also possible that PlexinB1 acts in the maintenance of Tumour growth in PlexinB1 mutant mice  It was previously shown that Sema4D-PlexinB1 signalling can collapse the growth cone of hippocampal neurons in culture [15]. Interestingly, we have observed loss of responsiveness to Sema4D in PlexinB1-deficient neurons derived from our mutants. On the other hand, our histological analysis revealed that the overall morphology, cell composition and neural connections in the hippocampus do not display obvious abnormalities in the absence of PlexinB1, suggesting a compensation of Sema4D function in development.
In the cerebellum, PlexinB1 is highly expressed in Purkinje (Pk) cells during the postnatal development.
Notably, these neurons do not express PlexinB2 or CD72, which are lower affinity receptors for Sema4D [19,21]. Incidentally, it was shown that Pk cells express PlexinB3 [33], but we previously demonstrated that this receptor cannot bind Sema4D [34]. PlexinB1 distribution in the cerebellum appears consistent with the traditional role of Plexins as receptors for neurite outgrowth inhibitory cues. In fact, during the first postnatal days, Purkinje axons show exuberant sprouting of collateral branches that are eventually pruned during cerebellar development [23]. The final shaping of Pk axons is associated with the axonal myelination mediated by the olygodendrocytes that express Sema4D [21]. However, our analysis shows that Pk axons are correctly myelinated and the pruning of excessive branches occurs normally in the PlexinB1 mutant mice. It is noteworthy that PlexinB1 in Pk cells is also present in the perikaryon and dendrites. In addition, the protein shows different levels of expression in Pk cell subsets, corresponding to morphofunctional modules of the cerebellar cortex [35]. Together with the precise distribution of Sema4D in the molecular layer, these observations suggest that Sema4D-PlexinB1 signalling may play a role in regulating the patterning and plasticity of intracortical connectivity, although this may be compensated during cerebellar development by alternative pathways.
The absence of clear developmental defects in PlexinB1 mutants is consistent with that reported for Sema4D-deficient mice. In fact, the latter were only found to have minor defects in the immune response, likely attributed to defective CD72 signaling [36], which is an alternative low-affinity receptor for Sema4D expressed in lymphocytes.
Our expression analysis had revealed that PlexinB1 is importantly expressed in a number of adult tissues, potentially suggesting a physiological role beyond embryo development. To address this point, in addition to study tissue histology, we have performed blood tests to investi-gate major functions in PlexinB1-deficient mice. Our data show that liver and kidney functions were normal. In addition, we found that hematopoiesis and coagulation parameters were not affected (data not shown).
Furthermore, it was proposed that PlexinB1 expressed by endothelial cells may mediate the pro-angiogenic activity of Sema4D released by tumour cells [18]. Tumourinduced angiogenesis is of fundamental importance in cancer progression; in fact, hampering the formation of new vessels is now a popular strategy to restrain tumour growth. In order to test the functional requirement for PlexinB1 in this process, we exploited B16 melanoma transplants, which are commonly used to study tumour growth and angiogenesis in the context of syngenic C57BL/6 mouse models. We found that B16 cells produce Sema4D and tested their ability to grow and induce angiogenesis in PlexinB1-deficient mice. Our results suggest that the angiogenic activity of Sema4D may be mediated not only by PlexinB1, but also via other receptors expressed in endothelial cells (e.g. PlexinB2, see Additional file 4, panel C). Future studies will tell whether other tumour models are equally capable to grow in a PlexinB1-deficient tissue microenvironment.

Conclusion
Collectively, our data concerning PlexinB1 and Sema4D expression suggest that PlexinB1 may carry out multiple functions during development and in the adult. Nonetheless, our analysis of PlexinB1 mutant mice did not show any major impairment in embryo development or differentiated functions in the adult, including the neo-angiogenetic response induced by tumour growth. We conclude that PlexinB1 plays a redundant role during development, in the maintenance of tissue homeostasis, and in tumour angiogenesis. Digital images were processed with Adobe Photoshop 6.0 to adjust contrast.

mRNA expression analysis
The number of glomeruli in the kidney was counted in sagittal sections taken in the middle of the kidney and counterstained with haematoxylin and eosin. Five sections were evaluated per mouse and three wild type mice were compared with three mutant mice. Pictures were taken with a LeicaDM IRB microscope connected to a Leica DC350FX camera and analyzed with Metamorph 6.3 software (Molecular devices, USA).

Generation of PlexinB1 mutant mice
A cosmidic clone containing the murine PlexinB1 gene was obtained by screening a RZPD (Berlin, Germany) genomic library. A targeting vector was designed to flank exons 22 and 23 of PlexinB1 with two loxP sites. A neo gene flanked by two FRT sites and a LacZ reporter gene were placed in frame with exon 24. A 10 kb fragment was used as a 5' homology region, a 1 kb fragment was placed between the two loxP sites, and a 2 kb fragment was used as 3' homology region. Embryonic stem cells (W9.5) were transfected, cultured and selected for resistance against G418. Homologous recombined clones were identified by Southern blot analysis of EcoRI-digested DNA, using an external probe that was amplified by PCR (sense primer GGT CTG ACC CTG GAT ATG GA; antinsense primer CCA CCCTCT TTT ATG CCT GA). Chimeric mice were generated by injection of homologous recombined ES cell clones into blastocystes from C57BL/6 mice. Matings of male chimeras to C57BL/6 females yielded germline transmitted offsprings. Mice carrying the mutated PlexinB1 gene were crossed with C57BL/6 CRE deleter mice in order to achieve the excision of exons 22 and 23. The correct excision was checked by PCR (sense AAC ACC ATG TGT ATG CTG GAG AGG TCA GGG; antisense CAT CGC CTT CTA TCG CCT TCT TGA CG). Mice genotyping was carried out by PCR using a primer pair for the WT (sense AAC ACC ATG TGT ATG CTG GAG AGG TCA GGG; antisense GGG TCA CTG ATT CGT TTC TCA GAA CAC TGA C) and the same primers used to check CRE excision for the mutant allele.
Loss of PlexinB1 expression was confirmed in mutant mice by protein immunopurification and Western blotting using IC-2 antibodies, as previously described [38].

Cultures of hippocampal neurons
Primary hippocampal neurons, isolated from neonatal P1 mice, were cultured as previously described [39]. Briefly, the hippocampi were isolated, trypsinized and cultured at low density in Neurobasal Medium supplemented with 2% B27 and 2% FCS. After 48 hours in culture neurons were stimulated for 1 hour with a purified preparation of Sema4D. Soluble Sema4D was purified from the conditioned media of cells transfected with a cDNA expression construct, generated by recombinant PCR as previously described [40].

Tumourigenesis assays in vivo
Sema4D expression in melanoma cells was detected using clone 30 antibody (BD-Transduction lab), by immunopurification and Western blotting experiments according to standard protocols [38]. Tumourigenesis assays were performed as previously described [41]. Shortly, 75000 B16 melanoma cells were diluted in PBS up to a final volume of 200 μl and injected subcutaneously into the right posterior flanks of 8-week old WT, heterozygous and homozygous PlexinB1 mutant mice. Tumour volume was calculated as described [41]. At the end of the observation period, tumours were weighed. Superficial pulmonary metastases were counted under a stereoscopic microscope taking advantage of the black pigmentation of B16 melanoma cell line.

Blood tests
Blood tests were performed according to ordinary analytical techniques. Briefly, blood samples were collected via intra-cardiac puncture in presence of Na Citrate to avoid blood clotting and analyzed using photometric method (Modular Roche diagnostic).

Abbreviations
Pk Purkinje, WT wild type, MBP myelin basic protein Tumour angiogenesis in PlexinB1 mutant mice