Background
Semaphorins are a group of 20 or more secreted or membrane bound proteins [
1] that act as chemotactic cues for cells expressing their transmembrane receptors plexins [
2]. Semaphorins affect cell behaviour in diverse ways, regulating cell motility [
3], invasive capacity [
4], adhesion [
5] and cell and axon growth cone collapse [
6]. Semaphorins consequently have a function in many physiological processes including angiogenesis [
7], cell migration [
8,
9], immune regulation [
10] and organogenesis, affecting nervous system [
11,
12], lung [
13], kidney [
14] and cardiovascular development [
15,
16] and epithelial-mesenchymal interactions [
14]. The response of a cell to semaphorin stimulation depends on the type of responding cell and particular semaphorins can generate opposite reactions depending on cell type [
17]. The transmembrane semaphorin receptors, plexins, either bind semaphorins directly, or in the case of most class 3 semaphorins, to a complex of neuropilins and plexins [
2,
18]. Semaphorin 4D (Sema4D) binds directly to its receptor, plexinB1 [
2].
Semaphorin/plexin signalling results in activation of receptor tyrosine kinases and modulation of the actin cytoskeleton via regulation of several specific small RhoGTPases. PlexinB1 binds to RacGTP [
19,
20], sequestering it from its downstream effectors, such as Pak1 [
21], and to Rnd [
22] and RhoD [
23]. Rac, Rnd and RhoD all bind to the same region in the cytoplasmic domain of plexinB1, the RhoGTPase binding domain (RBD) [
23]. Rnd binding is required for the binding of R-Ras to plexinB1 and for the R-RasGTPase activating protein (GAP) activity of plexinB1 which inactivates R-Ras resulting in a decrease in integrin and PI3K activation [
24,
25]. Sema4D/plexinB1 signalling activates RhoA through activation of PDZRhoGEF and LARG which bind to the C-terminal of the plexinB1 protein [
26]. PlexinB1 can also mediate the inhibition of RhoA via the recruitment of p190RhoGAP [
27]. PlexinB1 interacts with the receptor tyrosine kinases c-Met [
28] and ErbB2 [
29] via their extracellular domains and Sema4D/plexinB1 signalling results in c-Met and ErbB2 phosphorylation [
28,
29]. Activated ErbB2 phosphorylates tyrosines on plexinB1 creating a binding site for PLCγ. PLCγ binding is required for Sema4D-mediated RhoA activation [
30]. Sema4D can both promote and inhibit migration depending on the plexinB1 co-receptors expressed [
31].
Crystal structures of the RBD [
23] and of the cytoplasmic domain of plexinB1 in complex with Rac1 [
32,
33] have been determined. Bell et al. [
33]., identified a second RhoGTPase binding site in addition to the RBD, adjacent to the Ras site, which stabilises a trimeric structure of plexinB1-Rac complexes.
We have previously found mutations in the plexinB1 gene in 8/9 prostate cancer bone metastases, 7/17 prostate cancer lymph node metastases and 41/89 primary cancers, together with overexpression of the protein [
34]. The mutations in plexinB1 enhance adhesion, migration and invasion
in vitro and inhibit cell collapse [
34]. The finding of functionally significant mutations in plexinB1 in prostate tumours and overexpression of the plexinB1 protein suggests that plexinB1 has a role in prostate cancer and so is a potential target for therapy. However, the mechanism by which plexinB1 contributes to prostate cancer progression is yet to be determined.
Discussion
Mutations in the gene for plexinB1 are frequent in prostate cancer and the plexinB1 protein is overexpressed in prostate tumours, indicating that plexinB1 has a role in prostate cancer. The aim of this study was to investigate the mechanism by which mutation of plexinB1 contributes to prostate cancer progression. The three different cancer-associated plexinB1 mutations that were investigated in this study affect plexinB1-small RhoGTPase signalling in different ways, but have no effect on ErbB2, c-Met binding or RhoA activity.
Surprisingly, in spite of its position in the RBD region of plexinB1, the L1815P change does not impair RhoD binding. Rac has been shown to bind to a second site close to the R-Ras binding site in the cytoplasmic domain of plexinB1, in addition to the RBD site [
33]. It is possible that RhoD also binds to this second site (site B in reference [
33]) so that mutations in the RBD would not prevent RhoD binding. Alternatively the mode of binding of RhoD to the RBD may be different from that of the Rnd1 and Rac1. In an analogous way, the L1815P change inhibits Rnd1 binding to the RBD but not Rnd2 binding [
37]. Binding of RhoDGTP to plexinB1 is also retained by the T1795A and T1697A mutant forms, and binding is enhanced by these mutations. The T1795A amino acid change occurs within the RBD region and the T1697A change occurs outside of this region [
35].
All 3 mutant forms have lost R-RasGAP activity, yet only the L1815P mutation inhibits Rnd binding. Loss of Rnd1 binding would account for the loss of R-RasGAP activity in the L1815P mutant form, since R-RasGAP activity is dependent on Rnd binding [
24]. In contrast, the T1795A and T1697A forms bind Rnd, yet R-RasGAP activity is lost. Interestingly, RhoD has been shown to antagonise Rnd in signalling via plexinA1 [
38]. We have found that RhoDGTP reduced cell collapse when co-expressed with Rnd and plexinB1. If RhoD does antagonise Rnd signalling via plexinB1, the increase in RhoD binding seen in the T1795A and T1697A mutant forms may contribute to the loss of R-Ras GAP activity seen for these forms.
RhoDGTP inhibits stress fibre formation and motility [
39]. Binding of RhoDGTP to WT plexinB1 may sequester RhoD from downstream signalling partners. Expression of the T1697A and T1795A mutant forms of plexinB1 in tumour cells is predicted to have the effect of releasing the cell from the inhibitory effect of RhoD on motility and stress fibre formation, thereby promoting motility.
RacGTP binds and activates the autophosphorylation of Pak1 [
36]. WT plexinB1 binds and sequesters RacGTP and thereby inhibits Pak1 phosphorylation [
21]. Phosphorylation of Pak1 is not however inhibited by co-expression of RacGTP with the L1815P or T1795A mutant forms of plexinB1. The L1815P sequence change inhibits Rac, Rnd and R-Ras binding and the intrinsic R-RasGAP activity of plexinB1 [
34]. The T1795A mutation does not affect Rnd binding but reduces Rac binding and inhibits R-Ras binding and R-RasGAP activity [
34]. The release of inhibition of Pak1 phosphorylation by tumour cells expressing the L1815P and T1795A mutant forms of plexinB1is expected to result in an increase in MAPK signalling and to promote tumour progression.
Activation of Rac results in reorganisation of the actin cytoskeleton to form lamellipodia at the leading edge of migrating cells, a phenotype that is inhibited by WT plexinB1 expression. The L1815P mutation in plexinB1 blocks this inhibition of lamellipodia formation. The release from lamellipodia inhibition is expected to result in an increase in motility, and is consistent with the increase in motility we have observed in cells expressing this mutant form of plexinB1 [
34]. We have previously shown that overexpression of the three characterised mutations in HEK293 cells, without exogenous RacL61 expression, resulted in an increase in cell spreading above that of vector controls [
34]. These results suggest that in addition to the loss of inhibition of Rac function shown here, the mutations confer a gain of function which results in an increase in cell spreading.
Rac has also been shown to act upstream of plexinB1 facilitating the trafficking of plexinB1 to the cell membrane. Cells expressing the L1815P mutated form of plexinB1 show a decrease in cell surface expression of the protein.
Conclusions
Activated plexinB1 functions either as a positive or negative regulator of several signalling pathways that promote cell migration. ErbB2, c-Met and RhoA are activated by plexinB1, enhancing cell migration. In contrast, Rac, Rnd and R-Ras are inhibited by plexinB1, resulting in a decrease in cell motility. The response of a cell to plexinB1 activation depends on a balance between these signaling pathways. The mutations we identified have no effect on ErbB2, c-Met or PDZRhoGEF binding or RhoA activity which enhance migration, but one or more of the mutations inhibit or hinder Rac, Rnd, and R-Ras binding and R-RasGAP [
34] activity. The mutations thus have the net effect of increasing cell motility by the loss of inhibitory pathways. Two of the mutations also promote sequestration of RhoDGTP, an anti-migration factor.
Methods
Plasmid constructs and cell culture
The expression constructs for VSV-plexinB1and Sema4D-AP were kind gifts from Dr. L. Tamagnone. The region on plexinB1 cDNA encoding the intracellular domain (amino acids 1512-2135, accession no. X87904) was amplified by PCR and cloned into pGEX-4 T-3 (Amersham) using SalI and XhoI sites to produce pGEXB1cytoWT. The prostate cancer-associated mutations: L1815P, T1795A and T1697A were introduced into pGEXB1cytoWT and into VSV-plexinB1 by using QuikChange II XL in vitro mutagenesis kit (Stratagene). Other constructs were kindly provided by the following: Dr. KL Guan, pRK5-mycRacL61 and N17; Professor Takeshi Endo (Chiba University, Japan): RhoD pEF-BOS/Myc-RhoD G26V and T31K; Dr. Hitoshi Kikutani (Osaka University, Japan): Sema4D-Fc; Dr. J. Swiercz (Heidelburg): FLAG-PDZRhoGEF, ErbB2, RhoA; Dr M. Driessens: myc-PDZRhoGEF; Prof. J.Chernoff (Fox Chase Cancer Centre): pCMV-myc-hPak1. HEK293 and COS-7 cells were grown in DMEM (10% FCS).
Antibodies
The following antibodies were used: anti-VSV, anti-FLAG, anti-myc (Sigma, V4888, F7425, C956), anti-Human IgG Fc (Jackson ImmunoResearch, 109-005-098), anti-β-actin (Abcam, ab6276), anti-Pak1 (Chemicon, AB3844), anti-P-Pak, anti-phospho-ErbB2 (Cell Signalling), ErbB2 (Millipore), Sema4D and plexinB1 (ECM Biosciences), c-Met (c-28), plexinB1 (H300), (Santa Cruz).
COS-7 cells were transfected with Pak1, constitutively active Rac (RacL61), and plexinB1(WT or mutant) or empty vector using Lipofectamine (Invitrogen). Lysates were analysed for Pak phosphorylation using a phospho-Pak1 antibody.
Recombinant Sema4D
COS-7 cells transfected with Sema4D-Fc or Sema4D-AP or empty vector (control) were grown in serum free medium for 72 h. The conditioned medium was collected and used directly or purified. Sema4D concentration was assessed by western blotting (Figure
3).
Cell spread assays
COS-7 cells transfected with myc-RacL61 and VSV-plexinB1 (WT or mutant) or vector were plated on 10 μg/ml fibronectin. Unattached cells were washed away after 2 h. Attached cells were fixed (4% paraformaldehyde), permeabilized (0.1% Triton-X 100), blocked (2% BSA) and subjected to immunofluorescence staining (anti-myc). At least 200 myc-RacL61 positive cells were scored per slide.
In situ binding of Sema4D-AP with plexinB1
COS-7 cells transfected with plexinB1 (WT or mutant) and RacL61 or vector were washed with HBAH (Hanks balanced salt solution with 20 mM HEPES pH 7.0, 0.5 mg/ml BSA, 0.1% (w/v) NaN3), treated with HBAH containing 1,000 ng/ml Sema4D-AP for 90 min then washed with ice-cold HBAH and fixed (65% (v/v) acetone, 8% (v/v) formalin, 20 mM HEPES pH 7.0). The cells were incubated at 65°C for 100 min, then in BCIP/NBT solution in the dark.
Immunoprecipitation
Lysates of HEK293 transfected cells were incubated with 1 μg of selective antibody for 2 h at 4°C. The antigen-antibody complex was incubated with Protein-G sepharose for 2 h, washed 3 times and analyzed by immunoblotting.
GST-pull down assay
Lysates of COS-7 cells expressing RhoDG26V or RhoDT31K were incubated with GST-fused cyto-plexinB1 (WT or mutant) overnight. Following washing of the matrix, the samples were analysed by SDS-PAGE and immunoblotting.
Collapse assay
COS-7 cells plated on coverslips were co-transfected with plexinB1, Rnd and RhoDG26V-myc or RhoDT31K-myc. 48 h post transfection, the cells were treated with sema4D (4 μg/ml) for 5 mins, then fixed with 4% paraformaldehyde, permeabilised with 0.2% triton, then stained by immunofluorescence using anti-plexinB1 antibody(R&D) (FITC secondary antibody(Southern Biotech)) and anti-myc antibody (Abcam) (TRITC secondary). Co-transfected cells were sized using imageJ and cells with a size of < 500 μm2 with 3 or more processes were scored as 'collapsed'. Cells were scored 'blind'. The experiment was performed 3× in triplicate with a total of 186 or more co-transfected cells counted per condition.
Rho activity assay
HEK293 cells were transfected with RhoA, PDZRhoGEF and plexinB1 (WT or mutant) or vector control. 48 h after transfection, cells were treated with control or Sema4D conditioned medium for 25 mins and the lysate incubated with 50 μl Rhotekin (Upstate) for 1 h, washed 3 times and the pulled down protein analysed by western blot with anti-myc antibody.
Stress fibre assay
COS-7 cells transfected with plexinB1 (WT or mutant) were fixed (4% paraformaldehyde), permeabilized (0.1% Triton-X 100), blocked with 2% BSA then subjected to immunocytochemistry with anti-plexinB1 (SantaCruz) then goat-anti-mouse FITC (SouthernBiotech) supplemented with Phalloidin-TRITC (Sigma). The number of cells possessing more than three straight actin bundles of at least 5 μm long was counted. The identities of the slides counted were blinded to the researcher. At least 50 cells were counted per slide.
HEK293 cells transfected with vector or plexinB1 (WT or mutants) were serum starved over 2 nights then treated with control or sema4D conditioned medium or rhEGF(500 ng/ml, R&D Biosytems) for 20 mins. Lysates were analysed for ErbB2 phosphorylation using an anti phospho-ErbB2 antibody.
Acknowledgements
We thank the Prostate Cancer Research Centre, Smith's Charity, the Rosetrees Trust, the Barcapel Foundation, the McAlpine Foundation and the Orchid Appeal for funding.
Competing interests
A patent has been filed by M.W. and J.R.M. for mutations.
Authors' contributions
CZ and OW carried out the molecular genetic studies. JM participated in its design and coordination. MW conceived of the study, coordinated it and drafted the manuscript. All authors read and approved the final manuscript.