A Novel Splice Variant of Calcium and Integrin-Binding Protein 1 Mediates Protein Kinase D2-Stimulated Tumour Growth by Regulating Angiogenesis

ORIGINAL ARTICLE A novel splice variant of calcium and integrin-binding protein 1 mediates protein kinase D2-stimulated tumour growth by regulating angiogenesis

M Armacki1,2, G Joodi2, SC Nimmagadda2, L de Kimpe3,4, GV Pusapati5, S Vandoninck4, J Van Lint4, A Illing1 and T Seufferlein1,2

Protein kinase D2 (PKD2) is a member of the PKD family of serine/threonine kinases, a subfamily of the CAMK super-family. PKDs have a critical role in cell motility, migration and invasion of cancer cells. Expression of PKD isoforms is deregulated in various tumours and PKDs, in particular PKD2, have been implicated in the regulation of tumour angiogenesis. In order to further elucidate the role of PKD2 in tumours, we investigated the signalling context of this kinase by performing an extensive substrate screen by in vitro expression cloning (IVEC). We identified a novel splice variant of calcium and integrin-binding protein 1, termed CIB1a, as a potential substrate of PKD2. CIB1 is a widely expressed protein that has been implicated in angiogenesis, cell migration and proliferation, all important hallmarks of cancer, and CIB1a was found to be highly expressed in various cancer cell lines. We identify Ser118 as the major PKD2 phosphorylation site in CIB1a and show that PKD2 interacts with CIB1a via its alanine and proline-rich domain. Furthermore, we confirm that CIB1a is indeed a substrate of PKD2 also in intact cells using a phosphorylation-specific antibody against CIB1a-Ser118. Functional analysis of PKD2-mediated CIB1a phosphorylation revealed that on phosphorylation, CIB1a mediates tumour cell invasion, tumour growth and angiogenesis by mediating PKD-induced vascular endothelial growth factor secretion by the tumour cells. Thus, CIB1a is a novel mediator of PKD2-driven carcinogenesis and a potentially interesting therapeutic target.

Oncogene (2014) 33, 1167–1180; doi:10.1038/onc.2013.43; published online 18 March 2013 Keywords: PKD2; CIB1; angiogenesis; carcinogenesis

INTRODUCTION expression cloning (IVEC) to identify potential novel PKD The protein kinase D (PKD) family belongs to the calcium/ substrates that could have an impact on angiogenesis. One calmodulin-dependent protein kinase super-family and comprises potentially novel substrate identified in the screen was a splice PKD1, PKD2 and PKD3.1 PKDs are activated either directly via variant of human calcium and integrin-binding protein 1 (CIB1) phorbol esters or indirectly by various mechanisms including G that we termed CIB1a (GenBank accession number: BankIt1498958 protein-coupled receptors.2 These kinases participate in many key Seq1 JQ246073). signalling pathways in a diverse range of cells and are involved in CIB1 is a 22-kDa EF-hand-containing protein identified originally fundamental biological processes: PKDs have a critical role in as a binding partner of the platelet integrin aIIb, and later found to motility, migration and invasion of cancer cells.3–6 PKD isoforms, inhibit aIIbb3 activation in megakaryocytes.16,17 CIB1 is widely in particular PKD2, are highly expressed in certain tumours7,8 expressed18,19 and has a role in angiogenesis. Apparently, it is not and have been implicated in the regulation of tumour cell required for developmental angiogenesis, as CIB1-KO mice are proliferation and apoptosis.9,10 Recently, we and other viable and female mice are fertile, but is essential for proper EC laboratories have demonstrated that PKD1 and PKD2 have signalling and functions such as migration, proliferation and critical roles in vascular biology and angiogenesis.7,11–15 PKD2 is nascent tubule formation.20,21 Loss of CIB1 in ECs results in a pivotal mediator of vascular endothelial growth factor (VEGF)- attenuated responses to angiogenic growth factors such as VEGF induced endothelial cell (EC) proliferation and migration and and fibroblast growth factor 2 (FGF2), and consequently regulates the expression of VEGF receptor-2 and the production decreased expression of the matrix-degrading proteinase matrix of cytokines in ECs.13 Furthermore, PKD2 has a major role in metalloproteinase 2. CIB1 expression in the host also promotes tumour angiogenesis by mediating hypoxia-induced VEGF tumour growth and tumour-induced angiogenesis by an as yet production and secretion in tumour cells, as well as VEGF- unknown mechanism.22 CIB1 physically associates with proteins induced signalling in the tumour-associated ECs.7 PKD2 such as the transcription factor PAX3,23 the inositol 1, 4, downstream targets responsible for its effect on angiogenesis 5-trisphosphate receptor,24 polo-like kinases,18,25 Rac 3,26 focal are as yet incompletely understood. adhesion kinase (FAK)27 and p21-activated kinase (PAK1).28 For a better understanding of the precise signalling context of Among these binding partners, PAK1 and FAK regulate EC PKD2, in particular with respect to angiogenesis, we used in vitro function and angiogenesis in vitro or in vivo.29–31

1Department of Internal Medicine I, University of Ulm, Ulm, Baden-Wuerttemberg, Germany; 2Department of Internal Medicine I, Martin-Luther-University Halle-Wittenberg, Halle(Saale), Germany; 3Academic Medical Center (AMC), University of Amsterdam, Amsterdam, The Netherlands; 4Department of Cellular and Molecular Medicine, Faculty of Medicine, KU Leuven, Belgium and 5Department of Biochemistry, Stanford University School of Medicine, Stanford, CA, USA. Correspondence: Professor T Seufferlein, Department of Internal Medicine I, Ulm University, Albert-Einstein-Allee 23 D-89081 Ulm-89081, Baden-Wuerttemberg, Germany. E-mail: [email protected] Received 10 January 2012; revised 8 January 2013; accepted 11 January 2013; published online 18 March 2013 CIB1a mediates PKD2-induced tumour growth and angiogenesis M Armacki et al 1168 Here we identify a novel splice variant of CIB1, termed CIB1a, as been implicated in angiogenesis, we decided to further examine a substrate of PKD2 both in vitro and in intact cells, and found this clone, which we termed CIB1a. CIB1a is generated by an Ser118 as the major PKD2 phosphorylation site in CIB1a. Our data intron retaining mode of splicing and has an additional 120 bp also show that CIB1a specifically interacts with the N-terminal located in the coding region of N-terminal part of cDNA. The alanine and proline (AP)-rich domain of PKD2. CIB1a phosphoryla- retained part of the intron encodes amino acids properly, tion by PKD2 results in increased VEGF-A secretion by the containing no stop codon and causing no shift in the reading tumour cells in vitro and stimulates tumour growth and tumour frame (Supplementary Figure S1). angiogenesis on the chicken chorioallantois and enhances tumour To confirm that CIB1a is a direct substrate of PKD2, we cell invasion, making CIB1a a novel downstream mediator of the generated recombinant CIB1a protein from bacteria. Phosphoryla- pro-angiogenic properties of PKD2 in cancer. tion of CIB1a by PKD2 was analysed by in vitro kinase (IVK) assays. Wild-type PKD2 (PKD2-WT) phosphorylated CIB1a in vitro. CIB1a phosphorylation was substantially enhanced when PKD2 was RESULTS activated by phorbol 12-myristate 13-acetate (PMA) or on Identification of potential PKD2 substrates involved in incubation with constitutively active PKD2 (PKD2-2SE). In contrast, angiogenesis CIB1a phosphorylation was virtually undetectable in the presence To identify potential novel PKD2 substrates that could mediate the of catalytically inactive PKD2 (PKD2-DA; Figure 1b). Thus, PKD2 effect of PKD2 in tumour angiogenesis, we performed IVEC.32 phosphorylates CIB1a in vitro. Pools of approximately 100 clones from a complementary DNA CIB1 is widely expressed in normal human tissues.19 (cDNA) library were transcribed and translated in vitro, and the Interestingly, high levels of CIB1 expression were observed in resulting S35-labelled protein pools were incubated with or breast cancer cell lines and tissues18 as well as in bladder, cervix without catalytically active PKD2 (PKD2-2SE) and subjected to and colorectal cancer using microarrays and serial analysis of sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS– gene expression (SAGE) (GeneNote, http://bioinfo2.weizmann.ac.il; PAGE). Phosphorylation of potential substrates was assessed by GeneCards; http://www.genecards.org). To determine the the kinase-dependent alteration in electrophoretic mobility expression of CIB1 and CIB1a in various human cancer cell lines, (Figure 1a, upper panel). Positive cDNA pools were progressively we analysed their expression by reverse transcriptase–PCR using subdivided and single cDNA sub-pools were re-assayed for the primers designed to amplify both a 0.4-kb fragment of CIB1 and a presence of a kinase-induced mobility shift (Figure 1a, lower 0.5-kb fragment of CIB1a (Figure 1c). We detected a single 0.4-kb panel). This procedure yielded specific cDNAs encoding putative band in all cell lines examined (Figure 1c, upper panel), indicating PKD2 substrates. One positive clone with an open reading frame the presence of CIB1-specific mRNA. We speculated that the CIB1a of 697 bp encoding a polypeptide with a predicted mass of 26 kDa isoform might have also been amplified, but this could not be corresponded to a splice variant of the human CIB1. As CIB1 has demonstrated in the same PCR reaction. Therefore, CIB1a was re-

protein pools screen kDa IVK 32 clone 2 26 WB:GFP PKD2 130 PKD2-WT ++---- PKD2-DA --- ++ - PKD2-2SE - --- + + PMA - + - + - +

PKD2-2SE-++ -+-- + single protein screen clone 2 mobility shift Myc-CIB1 + - PKD2-2SE - + Myc-CIB1a - + kDa32 CIB1a (ectopic) 26 CIB1 WCL WB:CIB AGSe HeLa Cos1 MiaPaca1Panc1 HEK293U87 CIB1

HeLa HEK293U87 Panc1 AGSe MiaPaca1 kDa32 CIB1a CIB1a 26 (endo) RT-PCR WCL WB:CIB Figure 1. (a) Identification of PKD2 substrate/s via IVEC. (a, upper panel) 35S-labelled protein pools were generated by in vitro transcription and translation of a plasmid based, human adult brain cDNA library, using Promega’s Gold TnT SP6 Express 96 coupled Transcription and Translation System and incubated with ( þ ) or without ( À ) catalytically active PKD2 (PKD2-2SE (S706/710E)). Potential substrates exhibit a slower relative mobility because of phosphorylation by PKD2-2SE. Alteration in the electrophoretic mobility is indicated with arrows (a, lower panel). Single clones were identified by progressive subdivision of the positive pools. Arrows indicate a single clone (clone 2) mobility shift caused by phosphorylation. (b) In vitro phosphorylation of clone 2 by PKD2. (b, upper panel) For IVK assays HEK293-T cells were transfected with recombinant eGFP-PKD2-WT (wild type), eGFP-PKD2-DA (kinase dead) or eGFP-PKD2-2SE (constitutively active). At 24 h after transfection, cells were incubated for 10 min with or without PMA (400 nM), subsequently lysed and the kinase was immunoprecipitated. (b, lower panel) Lysates were then examined for the equal expression of the recombinant kinase. GST-CIB1a (clone 2) produced in Escherichia coli was incubated with purified recombinant kinase, followed by SDS–PAGE and autoradiography. (c, d) Expression of CIB1 and CIB1a in human cancer cell lines. (c) Reverse transcriptase (RT)–PCR amplification of CIB1 and CIB1a isoforms. (d, upper panel) Western blot analysis of overexpressed CIB1 and CIB1a in HEK293-T cells transfected with Myc-CIB1 or Myc-CIB1a. (d, lower panel) Expression of CIB1 and CIB1a proteins in various human cancer cell lines. AGS -gastric adenocarcinoma cells; HeLa -cervical cancer cells; MiaPaca2/Panc1—pancreatic cancer cell lines; HEK293- T—human embryonic kidney cells; U87—glioblastoma-astrocytoma cells.

Oncogene (2014) 1167 – 1180 & 2014 Macmillan Publishers Limited CIB1a mediates PKD2-induced tumour growth and angiogenesis M Armacki et al 1169 amplified from the same reaction using CIB1a-specific primers position À 3 (arginine (R)) and an aliphatic residue at position À 5 (Figure 1c, lower panel) and could then be detected in all cell lines (isoleucine (I)) with respect to the phosphorylatable serine (S)/ examined. To determine the expression of CIB1a in the respective threonine (T), as indicated by bold alphabets. In addition, we cell lines at the protein level, we performed western blot analysis examined whether the respective serine and threonine residues using an antibody directed against the C-terminus of CIB1. The were positioned at the surface of the protein of interest. To this antibody detected both overexpressed CIB1 and CIB1a (Figure 1d, end, a three-dimensional computer model of the CIB1a protein upper panel). CIB1a could be differentiated from CIB1 because of structure was generated using the DeepView/Swiss-PdbViewer its slower mobility in SDS–PAGE. Interestingly, the predominant based on the known crystal structure of CIB135 (Figure 2a). This endogenous isoform expressed in the tumour cells was CIB1a model revealed that both potential phosphorylation sites, Ser118 (Figure 1d, lower panel). Taken together, these results show that and Thr207, were surface accessible. Next, we generated fusion the splice variant of CIB1, CIB1a, is novel PKD2 substrate and is proteins of CIB1a in which Ser118 or Thr207 were substituted with highly expressed in cancer cells. alanine. These proteins were subjected to IVK assays with PKD2. Exchanging Ser118 to Ala abrogated phosphorylation of CIB1a by PKD2. In contrast, the CIB1a-Thr207Ala mutant showed a similar Identification of the PKD2 phosphorylation site/s in CIB1a level of phosphorylation by PKD2 compared with the wild-type To determine the site in CIB1a that is phosphorylated by PKD2, we protein (Figure 2b). Thus, Ser118, but not Thr207, is the major PKD2 performed an in silico analysis using the Scansite algorithm.33 This phosphorylation site in CIB1a. search revealed two sequences, ERICRVFSTSPAKDS and To determine whether Ser118 is also a major CIB1a phosphor- SDIDRDGTINLFEFQ that resemble PKD substrate phosphorylation ylation site in vivo, we generated a phospho-specific antibody motifs (LXRXX (T*/S*).34 Both sequences have a basic residue at directed against the CIB1a-derived peptide CRVFpSTSPAKDS (pS

PKD minimal motif: **LXRXX(T*/S*) kDa IVK ERICRVFSTSPAKDS SDIDRDGTINLFEFQ 32 CIB1a 26 20 ** ** 12 4 integrated density [a.u] PKD2-2SE + + ++ Serine 118 Threonine 207

CIB1a-WT CIB1a-S118A CIB1a-T207A

2.5 Phospho CIB1a De-phospho CIB1a

1.5 Absorbance

0.5

0 0 1000 2000 4000 8000 1600 32000 6400 128000 256000 Dilution factor

kDa WB:p-Ser118 WB:p-Ser118 32 p-CIB1a 26 WB:GFP WB:GFP 130 PKD2

WB:Myc WB:Myc 32 CIB1a 26 PMA -+-+-+-+ -+-+-+- CIB1a-WT CIB1a-WT CIB1a-WT CIB1a-WT CIB1a-SA CIB1a-SE CIB1a-SA CIB1a-SA PKD2-WT PKD2-DA PKD2-2SE PKD2-3SE PKD2-WT PKD2-WT PKD2-2SE PKD2-3SE Figure 2. (a) Identification of the PKD2 phosphorylation site in CIB1a. Two possible phosphorylation sites within CIB1a were predicted using the Scansite 2.0 library-based algorithm. In a hypothetical three-dimensional model of the CIB1a protein, the surface accessible Ser118 and Thr207 are circled in yellow (DeepView/Swiss-PdnViewer 3.7, GlaxoSmithKlineR&D, Brentford, Middlesex, UK; Swiss Institute of Bioinformatics, Switzerland; http:// www.genebee.msu.su/spdbv/mainpage.htm). (b) PKD2 phosphorylates CIB1a at Ser118 and not at Thr207. IVK assay demonstrates the degree of CIB1a phosphorylation by PKD2 on substitution of Ser118 with Ala or Thr207 with Ala, respectively. Graph represents quantification of integrated density done by AIDA Image Analyser (Raytest Isotopenmegera¨te GmbH, Straubenhardt, Germany). Results shown represent the means±s.e.m. of three independent experiments (**Po0.001). (c) Development of site-specific phospho-CIB1a antibody. Binding of the purified pSer118 antibody to the corresponding phospho- and de-phosphopeptides was analysed via enzyme-linked immunosorbent assay (ELISA) at the indicated dilutions. White bars represent absorbance values (450 nm) for dephospho-peptide recognition, grey bars are absorbance values (450 nm) for phospho-peptide recognition. The ELISA results indicate phosphospecificity of the antibody. (d)P-Ser118-CIB1a antibody only recognized the phosphorylated CIB1a-WT. HEK293-T cells were transfected with wild-type (Myc-CIB1a-WT), phosphorylated (Myc-CIB1a-S118E) or non-phosphorylatable (Myc-CIB1a-S118A) mutants and wild type, kinase dead (eGFP-PKD2-DA), or kinase active (eGFP-PKD2-2SE(S706/710E) and eGFP-PKD2-3SE(S706/710/244E)) PKD2 mutants as indicated. Transfected cells were left untreated (–) or stimulated ( þ )with400nM PMA (10 min). Cell lysates were analysed via western blotting (WB) with anti phospho-Ser118 antibody (p-CIB1a, for CIB1a phosphorylation), anti-Myc antibodies (Myc, for CIB1a expression) or anti-GFP (GFP, for PKD2 expression).

& 2014 Macmillan Publishers Limited Oncogene (2014) 1167 – 1180 CIB1a mediates PKD2-induced tumour growth and angiogenesis M Armacki et al 1170 indicating phosphorylated serine). After sequential purification on However, interaction of PKD2 and CIB1a was lost when the phosphopeptide and dephosphopeptide columns, the antibody N-terminal AP-rich region of PKD2 was missing (Figure 3c). Thus, displayed a very high specificity for phospho-Ser118-CIB1a CIB1a is likely to interact with this N-terminal region of PKD2. (Figure 2c). Additional characterization of the phosphoSer118 Consequently, CIB1a also interacted with PKD1, but failed to interact -CIB1a-specific antibody was done by immune-blot analysis with PKD3 that lacks the N-terminal polar domain (Figure 3d). of phosphorylation of bacterially expressed wild-type CIB1a To further examine whether the interaction between PKD2 and (CIB1a-WT) in the presence of wild type, kinase dead and kinase CIB1a could be biologically meaningful, we examined the active PKD2 (PKD2-WT, PKD2-DA, PKD2-S244/706/710E), with and subcellular localization of CIB1a using GFP-tagged CIB1a expression without stimulation with phorbol esters (Supplementary Figure plasmids. CIB1a was detectable ubiquitously in the cytoplasm and S2A). Furthermore, PMA-induced phosphorylation of endogenous the nucleus of HeLa cells independently of its phosphorylation CIB1a that was abolished by PKD2 knockdown (Supplementary status at Ser118. A more localized CIB1a immunofluorescence was Figure S2B). detectable in the perinuclear compartment where it colocalized Next, cells were transfected with Myc-tagged CIB1a expression with Golgin 97, a Trans-Golgi resident protein (Figure 3e). As we plasmids (wild type, CIB1a-S118A or CIB1a-S118E) together with have shown previously that PKD2 localizes to the nucleus, the various PKD2 constructs, as indicated (Figure 2d). Transfection of cytoplasm and in particular to the trans-Golgi network (TGN) in PKD2-WT induced a low level of Myc-CIB1a phosphorylation as HeLa cells where it colocalizes with Golgin 97,36–38 we conclude determined by the pS118-CIB1a antibody that was enhanced in the that CIB1a and PKD2 occur at similar subcellular localizations in presence of PMA. Similarly, expression of constitutively active tumour cells. Taken together, CIB1a and PKD2 are localized at the PKD2 mutants (PKD2-S706/710E and PKD2-S244/706/710E) induced same subcellular compartments and physically interact with each phosphorylation of Myc-CIB1a at Ser118. There was no detectable other. PKD2 interacts within its N-terminal AP-rich region with CIB1a phosphorylation on expression of catalytically inactive PKD2 CIB1a, and regulates the interaction via its catalytic activity. (PKD2-DA). Furthermore, the antibody did not detect CIB1a when Ser118 was substituted with alanine or glutamic acid, indicating that Ser118 is the only PKD phosphorylation site in CIB1a Expression of phosphomimetic CIB1a-S118E in tumour cells results (Figure 2d). in increased tumour growth and angiogenesis in the Thus, CIB1a is a physiological substrate of PKD2 and Ser118 is the chorioallantois membrane (CAM) model relevant PKD2 phosphorylation site. Both PKD2 and CIB1 have been shown to have a role in tumour growth and angiogenesis.7,22,39,40 Our data show that PKD2 phosphorylates CIB1a at Ser118. To determine whether CIB1a interacts with PKD2 phosphorylation of CIB1a at this site affects tumour formation Having established that CIB1a is a novel PKD2 substrate, we next and angiogenesis in an in vivo setting, we examined the growth of examined whether CIB1a and PKD2 would physically interact. HeLa cells expressing CIB1a-WT, non-phosphorylatable CIB1a Endogenous PKD2 was immunoprecipitated, and the immuno- (CIB1a-S118A) or a phosphomimetic CIB1a mutant (CIB1a-S118E) precipitate was probed for the presence of endogenous CIB1a and on the chicken CAM, a widely used model to study in vivo tumour vice versa. As shown in Figure 3b, both proteins interacted, formation and angiogenesis.41,42 Tumour formation was assessed regardless of the order of immunoprecipitation. This interaction after 4 days of incubation. Tumours derived from cancer cells was confirmed by co-immunoprecipitation assays using ectopi- overexpressing the phosphomimetic CIB1a-S118E mutant were cally expressed enhanced green fluorescent protein (EGFP)-PKD2 significantly larger, compared with tumour cells expressing CIB1a- and Myc-CIB1a in HEK293-T cells: EGFP-PKD2 was detectable in WT or CIB1a-S118A as assessed by measuring the tumour area Myc-CIB1a immunoprecipitates, and Myc-CIB1a was found within the silicon ring and by histological analysis, (Figures 4a and in EGFP-PKD2 immunoprecipitates (Figure 3d). The data shown b). Vice versa, tumours derived from cells expressing the non- in Figure 3b suggested that there might be a slight decrease in the phosphorylatable CIB1a-S118A mutant were significantly smaller interaction between PKD2 and CIB1a on activation of the kinase/ than those expressing CIB1a-WT. Consequently, compared with phosphorylation of the substrate. To further investigate whether CIB1a-WT or CIB1a-S118A-expressing tumours, tumours derived this interaction was indeed dependent on the catalytic activity of from cancer cells expressing CIB1a-S118E exhibited a significantly PKD2, we performed pull-down assays using exogenously increased proliferation index, as evidenced by the number of expressed, GFP-tagged wild type, catalytically active or inactive Ki-67-positive tumour cells (Figures 4d and e). PKD2 mutants. As shown in Figure 3c, CIB1a interacted with both, To confirm that CIB1a is a mediator of PKD2-induced constitutively active PKD2-SE and kinase-dead PKD2-DA suggest- tumour growth in this assay, we performed rescue experiments ing that the interaction between both proteins is constitutive. with CIB1a expression plasmids in Hela cells where PKD2 had been However, a quantitative analysis of the interaction between depleted by short hairpin RNA (shRNA). Selective PKD2 knock- phosphomimetic CIB1a-S118E or non-phosphorylatable CIB1a- down significantly reduced tumour growth on the CAM S118A and PKD2 revealed that phosphorylated CIB1a interacts less (Supplementary Figures S4B–D). However, tumour growth was with PKD2. This suggests that phosphorylation of CIB1a releases at partially restored in tumours developing from cells expressing least to some degree the protein from its interaction with PKD2 CIB1a-S118E. These tumours were significantly larger than tumours (Supplementary Figure 3A). developing from PKD2-depleted HeLa cells expressing CIB1a-WT Next, we were interested which domains in PKD2 would mediate or CIB1a-S118A, respectively (Supplementary Figures S4B–D). the interaction with CIB1a. The three members of the PKD family The effect of CIB1a-S118E on tumour growth could be the result (PKD1, PKD2 and PKD3) share a unique modular structure consisting of a direct stimulation of tumour cell proliferation or could be of two N-terminal cysteine-rich Zn-fingers, a central pleckstrin mediated indirectly. To examine whether CIB1a-S118E directly homology (PH) domain and a C-terminal Ser/Thr protein kinase affects tumour cells proliferation, we performed cell cycle analyses domain.1 The N-terminus of PKD1 and PKD2 starts with a in HeLa cells expressing CIB1a-WT, CIB1a-S118A or CIB1a-S118E, hydrophobic region rich in AP residues. This region is missing in respectively. As shown in Supplementary Figure S4E, there was no PKD3. To identify the site where PKD2 interacts with CIB1a, we increase in the percentage of cells in S phase expressing the incubated various mutants of PKD2 as indicated in Figure 3a CIB1a-S118E mutant. These data suggest that CIB1a-S118E promotes together with recombinant glutathione S-transferase (GST)-tagged tumour growth on the CAM by an indirect mechanism. CIB1a. Interestingly, GST-CIB1a interacted with PKD2 lacking its Both, PKD2 and CIB1a have been implicated in angiogenesis cysteine-rich domain, the PH domain and even the catalytic domain. and the CAM assay is an excellent assay to study tumour

Oncogene (2014) 1167 – 1180 & 2014 Macmillan Publishers Limited CIB1a mediates PKD2-induced tumour growth and angiogenesis M Armacki et al 1171 PMA PKD2 APC1a C2b AC PH kinase IP IgGCIB PKD2 CIB PKD2 WT D695A DA WB:PKD2 S706/710E SE ΔCRD WB:CIB1a ΔKD WCL (10 % input) Δ PH PKD2 ΔAP CIB1a

WT DA SE ΔCRD ΔKD ΔPH ΔAP’ PKD2 WB:GFP / pull down

PKD2 WB:GFP / WCL

IP: GFP Myc IP: GFP Myc IP: GFP Myc

130 PKD2 PKD1 PKD3 WB:GFP WB:GFP WB:GFP 32 CIB1a CIB1a CIB1a 26 WB:Myc WB:Myc WB:Myc

APC1a C2b AC PH kinase PKD1

PKD2

PKD3

GFP-CIB1a-wt Golgin97 merge

GFP-CIB1a-s118a Golgin97 merge

GFP-CIB1a-s118e Golgin97 merge Figure 3. PKD2 physically interacts with CIB1a via its N-terminal polar domain. (a) Schematic representation of PKD2 deletion mutants. (b) Cell lysates (1 mg) of HeLa cells treated with or without PMA (400 nm, 10 min) were subjected to immunoprecipitation with an antibody against PKD2 or CIB1 using Catch and Release Kit (EMD Millipore Corporation, Billerica, MA, USA), followed by CIB1a or PKD2 western blotting. Positions of endogenous PKD2 (105 kDa) and endogenous CIB1a (28 kDa) in WCL are indicated by arrowheads. The Co-immunoprecipitation using isotypic irrelevant antibody represents mock control. (c) Pull-down assay for CIB1a–PKD2 interaction. Equal amount of GST-CIB1a on glutathione sepharose beads was incubated with lysates of HEK293-T cells expressing the eGFP-tagged PKD2 mutants as indicated. Pull-down of PKD2 by CIB1a was analysed by anti-GFP western blotting. Expression levels of the various eGFP-tagged PKD2 plasmids (PKD2 input) was determined by anti-GFP western blotting of cell lysates obtained before incubation with the GST–CIB1a. (d) PKD1 and PKD2 interact with CIB1a but not PKD3. Lysates of HEK293-T cells expressing eGFP-PKD1, eGFP-PKD2 or eGFP-PKD3 together with Myc-CIB1a were subjected to immunoprecipitation with either anti-GFP or anti-Myc antibodies followed by western blot analysis. (d, lower panel) Schematic representation of PKD family members. (e) CIB1a is localized in a perinuclear compartment colocalizing with the trans-Golgi network independently of its phosphorylation status. HeLa cells expressing a wild-type EGFP-CIB1a, EGFP-CIB1a-S118A or EFGP-CIB1a-S118E were ﬁxed followed by anti- Golgin 97/Alexa 568 immunostaining. Arrows indicate localization of the CIB1a mutants. Inlets display localization of CIB1a at TGN. AP, polar domain; Cys, cysteine-rich domain; IP, immunoprecipitation; KD, kinase domain; WCL, whole cell lysate.

& 2014 Macmillan Publishers Limited Oncogene (2014) 1167 – 1180 CIB1a mediates PKD2-induced tumour growth and angiogenesis M Armacki et al 1172 CIB1a-WTCIB1a-SA CIB1a-SE

CIB1a-WT CIB1a-SA CIB1a-SE

250 *** 200 *** kDa WCL 32 CIB1a 26 150 * 40 Actin

100 -SA vector

Tumour area [%] Tumour CIB1a 50 CIB1a-WT CIB1a-SE

CIB1a-WTCIB1a-SACIB1a-SE

CIB1a-WT CIB1a-SA CIB1a-SE 90 * * 70

10 KI67/CK Positive Cells [%] KI67/CK Positive KI67 CK

-SE

CIB1a-WTCIB1a-SACIB1a Figure 4. Expression of a phosphomimetic CIB1a mutant (CIB1a-S118E) in cancer cells augments tumour growth on the chicken CAM. (a, upper panels) Photographs of tumours arising from cells (HeLa) overexpressing IRES-CIB1a-WT-GFP, IRES-CIB1a-S118A-GFP or IRES-CIB1a-S118E-GFP plasmids, as indicated, growing in silicon rings on the CAM. The dashed line indicates the quantified tumour area. Representative photographs of tumours are shown. (a, lower panels) Haematoxylin and eosin-stained cross-sections of paraffin-embedded tumours. (b) Graph represents quantification of the tumour areas within silicon ring. Results shown represent the means± s.e.m. of at least six tumours from three independent experiments (*Po0.05; ***Po0.0001). (c) Aliquots of cells used in the CAM assay were lysed and tested for equal expression of the CIB1a mutants using an antibody directed against GFP. Membranes were reprobed with a b-actin antibody. (d) Paraffin- embedded 4-day-old tumours were subjected to immunohistochemistry with antibodies directed against pan-cytokeratin and Ki-67. Representative images of tumours are shown. (e) Quantification of double positive Ki-67/pan-cytokeratin cells. Bars represent the percentage of double positive pan-CK/Ki-67-positive tumour cells, and error bars represent the standard errors of the means of at least 10 tumours per group (*Po0.05; scale bar: 100 mm).

angiogenesis in vivo. Thus, we examined whether CIB1a-S118E may phosphorylated by PKD2 markedly stimulates tumour angiogen- exert its tumour-promoting effect indirectly by modulating esis, whereas CIB1a-WT and CIB1a-S118A have no effect on tumour tumour angiogenesis. To evaluate the effect of PKD2-induced angiogenesis and angiogenesis could be a potential mechanism CIB1a phosphorylation on tumour angiogenesis, at first we explaining the growth-promoting effect of CIB1a-S118E. performed immunohistochemistry using antibodies directed It has been reported previously that CIB1 protein binds to and against von Willebrand factor and desmin to identify neoplastic regulates the activity of PAK128 and FAK,27 both kinases that are vessels.7 As shown in Figure 5a, there was a significant increase in known contribute to angiogenesis.30,31,43 Also CIB1a interacted von Willebrand factor and desmin immunoreactivity surrounding with both FAK and PAK1, however, the interaction was tumour cells in the CAM when the cells expressed CIB1a-S118E independent of its phosphorylation at Ser118 (Supplementary compared with CIB1a wild-type or non-phosphorylatable Figures S5A and S6A, respectively). Ectopic expression of CIB1a CIB1a-S118A. Figures 5b and c represent the quantification of the increased tyrosine phosphorylation of FAK as a readout immunohistochemistry shown in Figure 5a. Thus, CIB1a for its tyrosine kinase activity compared with vector controls

Oncogene (2014) 1167 – 1180 & 2014 Macmillan Publishers Limited CIB1a mediates PKD2-induced tumour growth and angiogenesis M Armacki et al 1173 CIB1a-WT CIB1a-SA CIB1a-SE Desmin vWF

0.7 ** ** ** *** 0.3 0.5 ** 0.2 0.3

0.1 [%] vWF Area Fraction 0.1 Desmin Area Fraction [%] Desmin Area Fraction

CIB1a-SA CIB1a-WT CIB1a-SA CIB1a-SE CIB1a-WT CIB1a-SE Figure 5. Phosphorylation of CIB1a by PKD2 in cancer cells stimulates angiogenesis on the chicken CAM. (a) Immunohistochemical analysis of tumour growth and angiogenesis on the CAM was performed using specific antibodies directed against desmin, and von Willebrand factor (vWF). Arrows indicate peritumoural ECs and pericytes. (b, c) Quantification of desmin (b) or vWF (c) immunoreactivity in tumours growing on the CAM. The quantification of desmin and vWF immunoreactivity was based on background subtraction and colour deconvolusion using ImageJ software. Results represent the means±s.e.m. of 10 tumours per group (**Po0.001; ***Po0.0001). Representative images of Desmin and vWF-stained tumour cross-sections are shown (scale bar: 100 mm).

(Supplementary Figure S5B). However, both CIB1a-S118A cells.7 As PKD2-induced phosphorylation of CIB1a stimulates and -S118E had no additional effect on FAK tyrosine phosphoryla- angiogenesis, we reasoned that CIB1a, as a substrate of PKD2, tion compared with CIB1a-WT. The catalytic activity of PAK1 was could be an important downstream mediator of VEGF-A not affected on overexpression of CIB1a-WT, CIB1a-S118E or CIB1a- expression under hypoxic conditions. At first, we examined S118A, respectively, as determined by IVK assays (Supplementary whether hypoxia would stimulate phosphorylation of CIB1a at Figure S6B). Thus, we can exclude FAK and PAK1 as potential Ser118. As shown in Figure 7a (top panel), hypoxia as indicated by mediators of the proangiogenic effect of CIB1a-S118E. increased expression of the hypoxia-inducible factor prolyl-4- Along with others, we have previously shown that HDAC5, via hydroxylase, PHD2, increased phosphorylation of endogenous the nuclear orphan receptor Nur77, mediates PKD2-induced CIB1a. Thus, hypoxia that activates PKD2 also induces phosphor- tumour angiogenesis.12 Thus, we examined whether CIB1a has ylation of its substrate CIB1a at Ser118. Next, we analysed VEGF-A an effect on HDAC5 regulated genes such as Nur77.12,44 As shown expression in tumour tissues and the peritumoural area by in Supplementary Figure 7A, CIB1a moderately induced Nur77 immunohistochemistry in tumours expressing CIB1a-WT, CIB1a- transcription, but there was no further change in Nur77 S118E or CIB1a-S118A, respectively. CIB1a-S118E-expressing tumours transcription in the presence of either CIB1a mutant. Thus, CIB1a exhibited significantly increased VEGF-A immunoreactivity in the can regulate Nur77 transcriptional activity, but this effect does tumour cells, the ECs and in particular in the extracellular matrix not contribute to its proangiogenic effect on phosphorylation (ECM) surrounding the tumour cells compared with CIB1a or by PKD2. CIB1a-S118A-expressing tumours (Figure 6). Indeed, VEGF-A As FAK, PAK and HDAC5 signalling pathways were not affected immunoreactivity was barely detectable in the stroma of tumours by PKD2-phosphorylated CIB1a, we examined whether CIB1a was expressing CIB1a-S118A (Figures 6a and c). The high VEGF-A involved in the secretion of pro-angiogenic cytokines that are also immunoreactivity in the stroma of CIB1a-S118E tumours suggests enhanced by active PKD2.7 that CIB1a phosphorylation might regulate VEGF production and/ or secretion in the tumour cells. Therefore, we next examined whether CIB1a would have an 118 CIB1a-S E stimulates VEGF-A secretion by the tumour cells effect of VEGF-A expression in Hela cells. HeLa cells were We have previously shown that PKD2 mediates hypoxia-induced transfected with scrambled or CIB1a-specific small interfering VEGF-A expression in and VEGF-A secretion by epithelial tumour RNA (siRNA) oligonucleotides and subsequently with a VEGF-A-luc

& 2014 Macmillan Publishers Limited Oncogene (2014) 1167 – 1180 CIB1a mediates PKD2-induced tumour growth and angiogenesis M Armacki et al 1174 promoter plasmid and then exposed to a normoxic or hypoxic cells expressed the phosphomimetic CIB1a-S118E mutant atmosphere, respectively. Depletion of CIB1a in the tumour cells (Supplementary Figure S7B). These data confirm previous data decreased the VEGF-A promoter activity by 40% under hypoxic showing that CIB1 can regulate VEGF-A promoter activity, but also conditions (Figure 7a). Hypoxia induced a 70% increase in indicate that PKD2-induced CIB1a phosphorylation does not affect VEGF-A promoter activity in cells expressing CIB1a-WT. VEGF this process. promoter activity was not significantly further enhanced when To examine a potential effect of PKD2-mediated CIB1a phosphorylation on VEGF secretion, we determined the amount of VEGF-A protein in the supernatant of HeLa cells expressing ** CIB1a-WT, CIB1a-S118A or CIB1a-S118E, and in the supernatant of VEGF 0.7 ** cells transfected with scrambled siRNA or siRNA oligonucleotides that specifically target PKD2 either under normoxic or hypoxic 0.5 conditions. The hypoxic environment was verified by upregulation of HIF1alpha as detected by western blotting (Figure 7b). Hypoxia- 0.3 induced VEGF-A secretion by cells expressing the phosphomimetic CIB1a-S118E mutant was significantly increased compared CIB1a-WT with VEGF-A secretion by cells expressing wild-type or non- Peritumoural Area Peritumoural 0.1 phosphorylatable CIB1a-S118A (Figure 7b). Depletion of PKD2 led VEGF Area Fraction [%] VEGF Area Fraction to a decrease in hypoxia-induced VEGF-A secretion, in accordance with our previous data.7 These data indicate that the PKD2-induced CIB1a phosphoryla- CIB1a-WT CIB 1a-SA CIB1a-SE tion is not required for hypoxia-induced VEGF-A expression, but it regulates VEGF-A secretion by the tumour cells downstream of 0.5 CIB1a-SA ** PKD2. ** PKD2 regulates vesicular transport from the TGN to plasma 0.4 membrane. Therefore, we tested whether CIB1a as PKD2 down- 37 0.3 stream target had an effect on vesicular transport using secretory form of the peroxidase (ssHRP) as a marker.45,46 In cells expressing CIB1a-WT, peroxidase (ssHRP) was efficiently

Tumour 0.2 secreted into the supernatant (Figure 7c). The amount of secreted 0.1 peroxidase (ssHRP) was significantly increased in cells expressing 118 CIB1a-SE VEGF Area Fraction [%] VEGF Area Fraction the phosphomimetic CIB1a-S E mutant compared with CIB1a- WT and CIB1a-S118A (Figure 7c). Thus, CIB1a contributes to the transport of cargo from the TGN to the plasma membrane downstream of PKD2. These data are in accordance with our CIB1a-WT CIB 1a-SA CIB1a-SE finding that CIB1a is localized at the TGN in HeLa cells (Figure 3e). 118 Figure 6. Expression of the phosphomimetic CIB1a-S118E mutant in Furthermore, colocalization studies using the pSer -CIB1a anti- cancer cells stimulates VEGF-A expression. (a) Immunohistochemical body revealed that CIB1a phosphorylated at Ser118 almost analysis of VEGF-A expression in the peritumoural area and within exclusively localizes to the TGN further supporting a role of the tumours growing on the CAM was performed using specific phosphorylated CIB1a at the TGN (Figure 7d). The effect of CIB1a antibodies directed against VEGF-A. In tumours derived from cells on protein transport was specific. CIB1a did not affect secretion of expressing the phosphomimetic CIB1a mutant, VEGF-A was another proangiogenic cytokine, FGF2, which has been shown to observed within cytoplasm of epithelial cancer cells (black arrows), be secreted by an ER-Golgi-independent mechanism,47–49 under in the surrounding stroma (blue arrow) and in ECs lining the microvessels (red arrows). (b, c) Quantification of VEGF-A immunor- both normoxic hypoxic condition (Supplementary Figure S7C). eactivity within the tumour (b) and in the peritumoural area (c). Quantification of VEGF-A immunoreactivity was based on back- Phosphorylated CIB1a stimulates tumour cell invasion ground subtraction and colour deconvolusion using ImageJ software. Results represent the means±s.e.m. of 10 tumours per group Tumour invasion is dependent on angiogenesis and both, PKDs 50–52 (**Po0.001). Representative images of VEGF-A-stained tumour and VEGF-A can promote tumour cell invasion. Therefore, we cross-sections are shown (scale bar: 10 mm). examined whether tumours expressing CIB1a-S118E would also

Figure 7. (a, top) Hypoxia increases phosphorylation of endogenous CIB1a at Ser118. HeLa cells cultured under normoxic (N) or hypoxic (H) conditions for 24 h were lysed and endogenous CIB1a was immunoprecipitated using the Catch and Release Kit (500 mg of cell lysate). Phosphorylation of CIB1a in response to hypoxia was detected using the anti-phospho CIB1aSer118 antibody. Hypoxic conditions were verified by examining PHD2 expression. (a, middle) CIB1a mediates hypoxia-induced VEGF-A promoter activation in epithelial tumour cells. HeLa cells transfected with CIB1a-specific (siCIB) or scrambled (siCon) siRNA oligonucleotides were subsequently transfected with pGL2-VEGF-A-Luc and pTK-Renilla for 8 h. Cells were subsequently incubated for 24 h under normoxic or hypoxic conditions and the cell lysates were subjected to luciferase assay. (a, bottom) CIB1a knockdown, hypoxia and equal loading of proteins were verified by western blotting using CIB1a, HIF1a and actin antibodies, respectively. (b) Phosphorylation of CIB1a enhances hypoxia-induced VEGF-A secretion by epithelial tumour cells. HeLa cells were transfected with wild type, non-phosphorylatable CIB1a-S118A, or phosphomimetic CIB1a-S118E mutants, or PKD2-specific (siPKD2) or scrambled siRNA oligonucleotides, respectively. Cells were subsequently incubated under normoxic or hypoxic conditions for 12 h. VEGF-A was determined in supernatant of the tumour cells using a specific ELISA. Hypoxic conditions were verified by western blotting against HIF1a. (c) The PKD2 substrate CIB1a mediates protein transport. ss-HRP and CIB1a-WT, or CIB1a-S118A or CIB1a-S118E were co-transfected in HeLa cells. 24 h after transfection, growth medium was replaced and HRP activity secreted in the medium was measured after 0 h and 2 h by chemiluminescence. Bars represent the means±s.e.m. of three independent experiments of HRP activity in the medium normalized to intracellular ss-HRP expression levels. Protein expression was verified in WCL by immunoblotting with anti-CIB (for CIB1a). Bars are the means±s.e.m. of the three independent experiments (T-test: *Po0.05; **Po0.001). (d) CIB1a phosphorylated at serine 118 is detectable predominantly in the perinuclear compartment. HeLa cells co-expressing a wild-type EGFP-CIB1a and pcDNA3-PKD2-DA or pcDNA-PKD2-SE were fixed followed by anti-Golgin97/Alexa568 and anti-phospho-Ser118/Alexa647 immunostaining. Arrows indicate localization of GFP-CIB1a and phospho-Ser118-CIB1a. Inlet displays localization of phospho-Ser118-CIB1a at TGN. NS, not significant.

Oncogene (2014) 1167 – 1180 & 2014 Macmillan Publishers Limited CIB1a mediates PKD2-induced tumour growth and angiogenesis M Armacki et al 1175 exhibit distinct invasive properties. Indeed, tumours expressing type and was barely detectable in tumours expressing CIB1a-S118A CIB1a-S118E were highly invasive on the CAM as judged by the (Figure 8a). number of invading cytokeratin-positive tumour cells (Figures 4d The migratory properties of tumours cells expressing the and 8a). Invasion of the CIB1a-S118E-expressing tumours was far various CIB1a mutants were assessed in vitro using a transwell more pronounced than that of tumours expressing CIB1a wild matrigel assay.53–56 HeLa cells expressing CIB1a-WT invaded the

N H 60 * normoxia p-CIB1a * hypoxia CIB1a 50 PHD2

l] 40 μ ** * * 30 120

normoxia 20 VEGF-A [pg/ 80 hypoxia 10

40 VEGF-A relative

luciferase activity [%] siCon siPKD2 CIB1a-SA siCon siCIB CIB1a-WT CIB1a-SE normoxia hypoxia normoxia hypoxia normoxia hypoxia endo CIB1a CIB1a PKD HIF1α α HIF1α HIF1 Actin Actin Actin siCon siCIB siCon siCIB WT SA SE WT SA SE siCon siCon siPKD2 siPKD2

* 14.0 0h ** 2h vector CIB1a-WT CIB1a-SA CIB1a-SE

10.0 n.s 123 123123123 CIB1a Actin 6.0 WCL

2.0 Secreted ssHRP Activity [a.u.]

Vector CIB1a-SA CIB1a-WT CIB1a-SE

GFP-CIB1a-wt // PKD2 DA Golgin97 p-Ser-118 merge

GFP-CIB1a-wt // PKD2 SE Golgin97 p-Ser-118 merge

& 2014 Macmillan Publishers Limited Oncogene (2014) 1167 – 1180 CIB1a mediates PKD2-induced tumour growth and angiogenesis M Armacki et al 1176 H&E Trans-well assay Transfection efficiency CIB1a-WT CIB1a-SA CIB1a-SE

200 *** * 160 ***

120

40 Number of Invasive cells Number of Invasive CIB1a-SE CIB1a-SA CIB1a-WT -SA

CIB1a-WT CIB1a CIB1a-SE Figure 8. CIB1a phosphorylated by PKD2 stimulates cancer cell invasion. (a) Immunohistochemical analysis of the peritumoural area in the CAM assay (haematoxylin and eosin (H&E)). Cancer cell overexpressing a phosphomimetic CIB1a mutant (CIB1a-S118E) invade signiﬁcantly more the chorioallantois membrane compared with cells expressing wild-type (CIB1a-WT) or non-phosphorylatable CIB1a mutants (CIB1a- S118A) as indicated by arrows. (b) Trans-well invasion assay. Invasive behaviour of cancer cell expressing phosphomimetic CIB1a mutant was also determined in vitro by trans-well invasion assays (upper panel). The number of invasive cells was quantiﬁed as shown. Bars are the means±s.e.m. of the three independent experiments each performed in duplicates (T-test. *Po0.05; Po0.0001; scale bar: 10 mm).

membrane. Tumour cell invasion was significantly increased proliferation, nascent tubule formation and ischaemia-induced on expression of the phosphomimetic CIB1a-S118E mutant. angiogenesis.20 Moreover, CIB1 can regulate tumour growth and In contrast, there was virtually no invasion of tumour cells intratumoural microvessel density.22 The protein binds to and expressing non-phosphorylatable CIB1a (Figure 8b). We deter- regulates the activity of proteins including PAK128 and FAK27 that mined basal Ser118 phosphorylation of endogenous CIB1a in are known to regulate EC function in vitro, and contribute to various cell lines. The highest baseline level of CIB1a-Ser118 angiogenesis in vivo.30,31,43 These data prompted us to further phosphorylation as well as the highest CIB1a protein level were examine CIB1a as a potential substrate and downstream mediator observed in a highly invasive glioblastoma cell line, U251 of PKD2. (Supplementary Figure S8) also indicating a link between CIB1a CIB1a was widely expressed, and its expression level in the Ser118 phosphorylation and tumour cell invasion.57 tumour cell lines examined (AGS—gastric adenocarcinoma cells, Thus, these data establish CIB1a also as a downstream mediator Panc1 and MiaPaca2—pancreatic cancer cells and U87— of PKD2-induced tumour cell invasion. glioblastoma cells) was even higher than that of CIB1. We identified Ser118 as the PKD2 phosphorylation site in CIB1a. By generating a phospho-specific antibody against Ser118 in CIB1a we show that DISCUSSION CIB1a is phosphorylated by PKD2 in intact cells on stimulation with PKDs are emerging as potentially important mediators of tumour activators of PKDs such as PMA. CIB1a is not only a PKD2 substrate, growth, invasion and tumour angiogenesis, which results in a but also physically interacts with the kinase, and we define the growing interest in the development of strategies to target these N-terminal AP-rich domain in PKD2 as the interaction site. This kinases.39,40,50 We have recently demonstrated that PKD2 is a interaction is constitutive, but regulated by the phosphorylation critical regulator of tumour cell–EC communication by mediating status of CIB1a, with CIB1a being released from the interaction on hypoxia-induced VEGF production and secretion by the tumour phosphorylation by PKD2. As both PKD2 and CIB1 have been cells and VEGF signalling in the ECs.7 In this study, we aimed at implicated in tumour growth and angiogenesis, we were interested further defining this signalling context of PKD2, the PKD isoform whether phosphorylated CIB1a might mediate some of these that is highly expressed in many tumours. We searched for novel functions downstream of PKD2. Tumour cells expressing non- PKD2 substrates that could have a role in angiogenesis using IVEC phosphorylatable CIB1a gave rise to significantly smaller tumours on and identified a novel splice variant of human calcium and the CAM compared with tumour cells expressing CIB1a-WT. integrin-binding protein as a bona fide PKD2 substrate that we Conversely, tumours expressing the phosphomimetic CIB1a mutant termed CIB1a. CIB1 regulates EC function in vitro, and contributes were significantly larger than the tumours expressing the wild-type to angiogenesis in vivo.30,31,43 In particular, it seems to be essential protein and exhibited a significantly higher proliferation index. for proper EC signalling and functions such as migration, Furthermore, phosphomimetic CIB1a partially restored tumour

Oncogene (2014) 1167 – 1180 & 2014 Macmillan Publishers Limited CIB1a mediates PKD2-induced tumour growth and angiogenesis M Armacki et al 1177 growth caused by depletion of PKD2. Thus, we provide evidence specific therapeutic strategy. Our data suggest that CIB1a could be that CIB1a mediates in part the tumour growth-promoting proper- an interesting novel candidate for such a strategy. ties of PKD2 that have been described previously.39,40 Our data also show that PKD2-phosphorylated CIB1a exerts its tumour-promoting effect not by direct stimulation of tumour cell proliferation, but MATERIALS AND METHODS indirectly via inducing tumour angiogenesis. Tumours expressing Cells and transfection the phosphomimetic CIB1a mutant exhibit a significantly increased Cells were cultured in Dulbecco’s modified Eagle’s medium (Invitrogen, tumour angiogenesis as indicated by a higher desmin and von Karlsruhe, Germany) supplemented with 10% fetal calf serum (FCS; PAA, Willebrand factor immunoreactivity. PKD2 is activated by hypoxia Coelbe, Germany) and 1% penicillin/streptomycin, as described previously. and mediates both hypoxia-induced VEGF-A expression and VEGF-A HeLa cells were transfected using TransIT HeLa Monster transfection kit secretion by the tumour cells.7 Importantly, CIB1a phosphorylation according to manufacturer’s instruction (Mirus Bio LLC, Madison, WI, USA). at Ser118 is also induced by hypoxia. VEGF-A expression in the Transfection of HEK293-T was performed with polyethyleneimine (linear, MWB25000, Polysciences Inc., Warrington, PA, USA; stock concentration of tumour cells required CIB1a, but is independent of its 63 phosphorylation status. However, the phosphomimetic CIB1a-S118E 1 mg/ml) with a ratio of 1:7.5 (mg DNA: ml polyethyleneimine). Virus mutant significantly stimulated VEGF-A secretion by the tumour cells supernatants were harvested after 48 h and used for transduction of Panc1 and HeLa cell lines. Cells were subsequently subjected to under hypoxic conditions. We could also identify the underlying 118 pyromycin selection to generate semi-stable cell lines used in assays. mechanism. CIB1a phosphorylated at Ser induces cargo transport ShRNA Constructs: non-target shRNA control (scramble, shc002), from the TGN to the plasma membrane. Thus, CIB1a is a novel and shRNA_PKD2 (Sigma-Aldrich, St Louis, MO, USA, NM_016457.x- downstream effector of PKD2-mediating protein transport from the 1720s1c1 (sh13D4); NM_016457.x-294s1c1 (sh13D5); NM_016457.x- TGN to the plasma membrane. We also provide evidence that 1767s1c1 (sh13D6); NM_016457.x-1335s1c1 (sh13D7)). particularly phosphorylated CIB1a is localized at the TGN where it colocalizes with the TGN resident protein Golgin 97. The effect of CIB1a on the regulation of protein transport was specific. The DNA constructs pEGFP-PKD2-WT, D695A, S706/710E, S244/706/710E, DCRD, DC1a, DC1b, DPH phosphomimetic CIB1a mutant had no effect on secretion of FGF2 36 that is exported by an endoplasmic reticulum/Golgi-independent and DKD have been described previously. Deletion of amino acids 1-137 (PKD2-DAP) was performed in pcDNA3-PKD2-WT36 with a 50 sense primer pathway, the so-called unconventional secretion. In line with our 0 0 0 118 (5 -GCCACCTTCGAGGAATTCCAGAT-3 ) containing an EcoRI and a 3 findings, VEGF secretion on expression of CIB1a-S E was not only antisense primer (50-CATTGCTGGGATCCTGTGAAGAACC-30) containing a detectable in vitro, but also in the tumours growing on the CAM. BamHI site. The fragment was cloned into an EcoRI- and BamHI-digested Indeed, there was a marked increase in VEGF-A immunoreactivity pEGFP-C2 vector (BD Biosciences Clontech, Erembodegem, Belgium). To particularly in the ECM of tumours expressing the phosphomimetic clone the CIB1a, PCR on full-length CIB1a cDNA (cDNA library, Promega, CIB1a mutant, indicative for increased VEGF secretion. There is Madison, WI, USA) was performed using the primers: 50-GAGGCGAGTTG increasing evidence that ECM binding regulates VEGF-dependent GCGGAG-30 and 50-TGTGTCGACAGTGCGGGC-30. PCR product was vascular patterning by controlling VEGF diffusion and gradients.58 subcloned into pCR II-TOPO vector (Invitrogen), used as shuttle vector. Proteolytic release of VEGF from the ECM can occur by cleavage of CIB1a was then subcloned as an EcoRI/SalI fragment into pGEX-4T-1 (Amersham/GE Healthcare, Muenchen, Germany) or pCMV-Tag3B vector matrix-bound VEGF at its C-terminal domain or by cleavage of the (Stratagene, La Jolla, CA, USA). To clone CIB1a into pIRES-AcEGFP (BD ECM, and results in a diffusible VEGF that increases the soluble VEGF Biosciences Clontech), PCR of full-length CIB1a cDNA using pGEX4T-CIB1a concentration, thereby potentiating the angiogenic switch and as template was performed using the primers: 5’-TTGGTGGTGGC 58 leading to neovascularization and tumour growth. Thus, CIB1a GACCATCC-3’ and 50-CTGGGGATCCTGTCACAGGACAA-30. The resulting mediates, at least in part, PKD2-induced VEGF-A secretion by the PCR product was digested with EcoRI and BamHI restriction enzymes, tumour cells in response to hypoxia and increases the VEGF-A and subcloned into the pIRES-AcEGFP vector. Point mutations in the CIB1a content in the ECM, which is likely to contribute to its proangiogenic sequence were generated using Quick-Change Site-Directed Mutagenesis properties. Kit (Stratagene) according to the manufacturer’s instruction using primers: 50-GCAGGGTCTTCGCCACATCCCCAGCC-30;50-GCAGGGTCTTCGAGACATCCC PKD2 and VEGF-A have been shown to promote tumour cell 0 0 0 118 50–52 CAGCC-3 ;5GGATGGAGCCATCAACCTCTTTGAGTTCCAG-3 for CIB1a-S A; invasion and CIB1 positively regulates cell migration and is CIB1a-S118E and CIB1a-T207A, respectively. necessary for the recruitment of FAK to the focal adhesions.59 Our data show that in addition to increased angiogenesis, phosphorylation of CIB1a increases the invasiveness of the The IVEC method tumour cells both in vitro in matrigel and in vivo on the CAM. In Plasmid-based human adult brain cDNA library is expressed using contrast, cells expressing a non-phosphorylatable CIB1a mutant Promega’s GoldTNT Express 96 Transcription/Translation System according exhibit a markedly reduced invasive behaviour even compared to the manufacturer´s instructions. In all, 2.5 ml aliquots of the [35S] with tumour cells expressing the wild-type protein. Thus, CIB1a is methionine-labelled protein pools were mixed with phosphorylation buffer also a downstream mediator of PKD2-induced tumour cell (20 mM Tris-HCl, pH7.5, 10% glycerol, 1.5 mM DTT, 150 nM microcystin-LR, invasion. The effect of phosphorylated CIB1a on tumour cell 8mM NaF, 0.15 mM Na3VO4,10mM MgOAc, 0.5 mM ATP), with or without 0.2 mM PKD2-2SE in a final volume of 15 ml. Samples were incubated at invasion on the CAM is at least in part caused by its proangiogenic 30 1C for 50 min, and then the reactions were terminated by adding effect. The precise molecular mechanisms underlying the effect of 3 Â SDS–PAGE sample buffer. The denatured samples were separated on a CIB1a on tumour cell invasion in vitro still remain to be elucidated 10% SDS polyacrylamide gel (Bio-Rad Laboratories GmbH, Muenchen, in further studies. Taken together, we identify CIB1a as a novel Germany). After electrophoresis, the gels were fixed in a methanol/acetic PKD2 substrate and interactor. On phosphorylation by PKD2, acid/H2O solution (8/2/10, v/v/v) for 15 min, washed with distilled water for CIB1a mediates tumour growth in vivo by promoting 10 min and vacuum dried. Labelled proteins were detected by 2- to 3-day angiogenesis. At the molecular level, we show that PKD- exposure of Hyperfilm ECL film (Amersham/GE Healthcare). Films were mediated phosphorylation of CIB1a mediates protein transport examined for the presence of bands of altered mobility in the presence of from the TGN to the plasma membrane thereby regulating VEGF-A added kinase. Once a pool possessing a candidate substrate was identified, the original cDNA pool was subdivided and retested until the single cDNA secretion by the tumour cells. Thus, CIB1a is a novel downstream encoding the protein of interest was isolated. mediator of PKD2-induced protein transport and secretion and establish a new regulatory role for phosphorylated CIB1a in tumour growth, invasion and angiogenesis. Given the plethora of Semi-quantitative reverse transcriptase–PCR 60–62 effects that are mediated by the PKD kinase family targeting CIB1 and CIB1a were amplified using following specific primers the substrates rather than the kinase itself might be a more 5’-GACGTTCCTGACGAAGCAGGAGA-3’, 5’-CTCAGACGCACTAAGCCGTGTGT-3’

& 2014 Macmillan Publishers Limited Oncogene (2014) 1167 – 1180 CIB1a mediates PKD2-induced tumour growth and angiogenesis M Armacki et al 1178 for CIB1 and 5’-GAGACCGATCGCTGAGAACA-3’, 5’-CTCAGACGCACTAAGC CAM assay CGTGTGT-3 for CIB1a. Shells of fertilized chicken eggs were opened on day 4 and silicon rings (5 mm in diameter) were applied onto the CAM.7 Cancer cell (1 Â 106) Western blotting, immunoprecipitation and IVKs overexpressing CIB1a mutants mixed with Matrigel in 1:1 ratio (BD 64 Matrigel; 10 mg/ml, Becton Dickinson GmbH, Heidelberg, Germany) were Immunoprecipitations were performed as described with anti-GFP transplanted onto CAM. Tumours were photographed and harvested after (Roche Diagnostics, Mannheim, Germany) or anti-CIB (Santa Cruz 4 days. Formalin-fixed tumours were embedded in paraffin using standard Biotechnology, Santa Cruz, CA, USA) antibody and further analysed by histological procedures. The 5 mm sections were processed and stained western blotting using various antibodies as indicated in the figure with antibodies directed against Ki67 (1:800; Dako Deutschland GmbH, legends. IVK assay was performed to determine substrate phosphorylation Hamburg, Germany), pan-cytokeratin (1:400; Dako), Desmin (1:100; Dako); by PKD2 as previously described.36 Immunoprecipitation of endogenous vonWillebrand factor (1:200; Biocare Medical, Concord, CA, USA). proteins was performed using Catch and Release v2.0 Reversible Quantification of staining, shown as percentage of stained area, included Immunoprecipitation System from Millipore (EMD Millipore Corporation, background subtraction and colour deconvolusion. Deconvolusion plugin Billerica, MA, USA) according to the manufacturer’s instructions. for Image J (National Institutes of Health, Bethesda, MD, USA, http:// imagej.nih.gov/ij/), which implements stain separation, is based on Ruifrok Bacterial GST-fusion protein expression and purification and Johnston’s´method.65 Recombinant proteins were produced and purified as described pre- 38 viously. Briefly, Invitrogen’s Escherichia coli (E. Coli) strain TOP 10 (for Trans-well invasion assay PKD2) and NEB Iq Express Competent E. Coli (for CIB1a) were transformed HeLa cells plated in six-well dishes were transfected with 2 mg of the with the pGEX4T-PKD1-S706/710E, pGEX4T-CIB1a-WT, pGEX4T-CIB1a-S118A respective plasmids as indicated in the figure legend. At 24 h after or pGEX4T-CIB1a-S118E expression vectors. Single colonies were inoculated transfection, medium was replaced and cells were starved for 12 h in in a 50 ml of liquid broth medium with appropriate antibiotics and cultured Dulbecco’s modified Eagle’s medium supplemented with 1% penicillin/ overnight at 37 1C. Overnight cultures were inoculated (2% inoculum) streptomycin and 1% FCS. Subsequently, cell were harvested and 50 000 and grown to OD of 0.5 and induced with 1 mM isopropyl 600 cells resuspended in starvation medium were seeded on 12-well format b-d-thiogalactoside for 4 h at room temperature (18 1C for PKD2). Trans-well (Becton Dickinson) coated with 0.5 mg/ml BD Matrigel. Trans- Bacterial cells were pelleted at 4 1C, and pellets were stored at À 80 1C. wells were previously coated with Matrigel for 12 h at 37 1C. Lower Proteins were purified from the bacterial lysates by glutathione-sepharose chamber of the trans-well system was filled with Dulbecco’s modified 4B beads (GE Healthcare) following the manufacturer’s instructions. Eagle’s medium supplemented with 10% FCS and 1% penicillin/ streptomycin. Cell were further incubated for 24 h in humidified atmo- CIB1a pull-down assay sphere with 5% CO2 at 37 1C. Non-invaded cells on the top of the trans-well HEK293-T cells expressing various PKD2 isoforms or mutants were lysed in were scraped off and invaded cells were stained with 4,6-diamidino-2- phenylindole and counted under microscope.66 GST pull-down lysis buffer (TBS, 1% Triton X-100, 2 mM EDTA, 5 mM MgCl2, protease and phosphatase inhibitor cocktail (Roche Diagnostics)), and the extracts were incubated with 50 mg of GST-CIB1a immobilized on Flow cytometric analysis of cellular DNA content glutathione-sepharose beads for 2 h at 4 1C. Beads were washed Hela cells were transfected with EGFP-PKD2-SE, IRES-CIB1a-WT-GFP, IRES- extensively, re-suspended in Laemmli buffer, and analysed by SDS–PAGE 118 118 and western blotting. CIB1a-S A-GFP or IRES-CIB1a-S E-GFP plasmids. At 24 h after transfection, cells were washed with PBS and dissociated into single-cell suspension by incubation with 0.25% trypsin/EDTA. After dissociation, cells were washed Antibody evaluation by enzyme-linked immunosorbent assay once with PBS and dissolved in fluorescence-activated cell sorting (FACS)- Bovine serum albumin-coupled phospho- and dephospho-peptides buffer (PBS þ 2% FCS). The GFP-positive population was sorted with BD (CRVFpSTSPAKDS/CRVFSTSPAKDS; S (serine), pS (phospho-serine)) or FACSAria III Cell Sorter. The GFP-positive cells were gated according to the multiple antigenic peptide conjugates were coated on a microtiter plate negative population in the untransfected control sample. Sorted cell (100 mlof10mg/ml solution in phosphate-buffered saline (PBS) per well) populations were pelleted, followed by fixation with 4% paraformaldehyde overnight at 4 1C. Wells were washed three times with buffer A (PBS, 0.5 M (PFA) in PBS for 15 min at 37 1C. For cell cycle analysis, the fixed cells were NaCl, 0.1% Tween-20, pH 7.2) and incubated with 100 ml of primary pelleted for 10 min at 1800 r.p.m., resuspended in FACS-buffer (PBS þ 2% antibody (dilutions ranging from 1:1000 to 1:256 000 in buffer A containing FCS) and further stained for 30 min at room temperature with propidium 3% PEG 6000) for 2 h at room temperature. Secondary anti-rabbit iodide (10 mg/ml), afterwards washed with FACS-buffer and analysed on the horseradish peroxidase (HRP)-conjugated antibody (diluted in buffer A BD LSRII flow cytometer. Data analysis was performed using BD DIVA containing 3% PEG 6000) was incubated for 1–2 h at room temperature. software (v6) (BD Biosciences, San Jose, CA, USA). Colorimetric detection of antibody was done with 3, 30,5,50-tetramethyl- benzidine as substrate for HRP. The reaction was then stopped by the addition of 100 mlof2M H2SO4 and antibody–antigen recognition was CONFLICT OF INTEREST quantified by reading the absorption at 450 nm in a spectrophotometer. The authors declare no conflict of interest.

Promoter assays HeLa cells plated in six-well dishes were transfected with 1 mg of the ACKNOWLEDGEMENTS respective promoter, 1 mg of the respective expression plasmid and 50 ng This project was funded in part by the German Federal Ministry of Education and pRL-TK Renilla (Promega, Mannheim, Germany) as indicated in the ﬁgure Research (BMBF) grant no. PKB-01GS08209-4 and the German Cancer Aid grant legends using TransIT HeLa Monster reagent according to the manufac- no. 109373 (both to TS) and in part by the Research Programme of the Research turer’s instructions. Luciferase activity was determined using the Dual Foundation-Flanders (FWO) (PhD fellowship to LDK and grant G.0612.07 to JVL). We Luciferase Assay Kit (Promega, Germany). Fireﬂy luciferase relative light thank Holger Reim and Claudia Ruhland for expert technical assistance. units were normalized with Renilla luciferase activity.

Enzyme-linked immunosorbent assay REFERENCES Experiments were performed using the Quantikine human VEGF-A165 or 1 Rykx A, De Kimpe L, Mikhalap S, Vantus T, Seufferlein T, Vandenheede JR et al. FGF Immunoassay Kit (R&D Systems, Minneapolis, MN, USA) according to Protein kinase D: a family affair. FEBS Lett 2003; 546: 81–86. the manufacturer’s instructions. 2 Jamora C, Yamanouye N, Van Lint J, Laudenslager J, Vandenheede JR, Faulkner DJ et al. Gbetagamma-mediated regulation of Golgi organization is through the direct activation of protein kinase D. Cell 1999; 98: 59–68. Hypoxia experiments 3 Bowden ET, Barth M, Thomas D, Glazer RI, Mueller SC. An invasion-related com- A hypoxic environment was generated using Anaerobic Incubation System, plex of cortactin, paxillin and PKCmu associates with invadopodia at sites of Anaerocult (Merck, Darmstadt, Germany). extracellular matrix degradation. Oncogene 1999; 18: 4440–4449.

Oncogene (2014) 1167 – 1180 & 2014 Macmillan Publishers Limited CIB1a mediates PKD2-induced tumour growth and angiogenesis M Armacki et al 1179 4 Eiseler T, Doppler H, Yan IK, Kitatani K, Mizuno K, Storz P. Protein kinase D1 29 Bagheri-Yarmand R, Vadlamudi RK, Wang RA, Mendelsohn J, Kumar R. Vascular regulates cofilin-mediated F-actin reorganization and cell motility through sling- endothelial growth factor up-regulation via p21-activated kinase-1 signaling shot. Nat Cell Biol 2009; 11: 545–556. regulates heregulin-beta1-mediated angiogenesis. J Biol Chem 2000; 275: 5 Eiseler T, Schmid MA, Topbas F, Pfizenmaier K, Hausser A. PKD is recruited to sites 39451–39457. of actin remodelling at the leading edge and negatively regulates cell migration. 30 Kiosses WB, Hood J, Yang S, Gerritsen ME, Cheresh DA, Alderson N et al. A FEBS Lett 2007; 581: 4279–4287. dominant-negative p65 PAK peptide inhibits angiogenesis. Circ Res 2002; 90: 6 Prigozhina NL, Waterman-Storer CM. Protein kinase D-mediated anterograde 697–702. membrane trafficking is required for fibroblast motility. Curr Biol 2004; 14: 88–98. 31 Shen TL, Park AY, Alcaraz A, Peng X, Jang I, Koni P et al. Conditional knockout of 7 Azoitei N, Pusapati GV, Kleger A, Moller P, Kufer R, Genze F et al. Protein kinase D2 focal adhesion kinase in endothelial cells reveals its role in angiogenesis and is a crucial regulator of tumour cell-endothelial cell communication in gastro- vascular development in late embryogenesis. J Cell Biol 2005; 169: 941–952. intestinal tumours. Gut 2010; 59: 1316–1330. 32 Gao ZH, Metherall J, Virshup DM. Identification of casein kinase I substrates by 8 Chen J, Deng F, Singh SV, Wang QJ. Protein kinase D3 (PKD3) contributes to in vitro expression cloning screening. Biochem Biophys Res Commun 2000; 268: prostate cancer cell growth and survival through a PKCepsilon/PKD3 pathway 562–566. downstream of Akt and ERK 1/2. Cancer Res 2008; 68: 3844–3853. 33 Obenauer JC, Yaffe MB. Computational prediction of protein-protein interactions. 9 Mihailovic T, Marx M, Auer A, Van Lint J, Schmid M, Weber C et al. Protein kinase Methods Mol Biol 2004; 261: 445–468. D2 mediates activation of nuclear factor kappaB by Bcr-Abl in Bcr-Abl þ human 34 Nishikawa K, Toker A, Johannes FJ, Songyang Z, Cantley LC. Determination of the myeloid leukemia cells. Cancer Res 2004; 64: 8939–8944. specific substrate sequence motifs of protein kinase C isozymes. J Biol Chem 1997; 10 Peterburs P, Heering J, Link G, Pfizenmaier K, Olayioye MA, Hausser A. Protein 272: 952–960. kinase D regulates cell migration by direct phosphorylation of the cofilin phos- 35 Guex N, Peitsch MC. SWISS-MODEL and the Swiss-PdbViewer: an environment for phatase slingshot 1 like. Cancer Res 2009; 69: 5634–5638. comparative protein modeling. Electrophoresis 1997; 18: 2714–2723. 11 Evans IM, Britton G, Zachary IC. Vascular endothelial growth factor induces heat 36 Auer A, von Blume J, Sturany S, von Wichert G, Van Lint J, Vandenheede J et al. shock protein (HSP) 27 serine 82 phosphorylation and endothelial tubulogenesis Role of the regulatory domain of protein kinase D2 in phorbol ester binding, via protein kinase D and independent of p38 kinase. Cell Signal 2008; 20: catalytic activity, and nucleocytoplasmic shuttling. Mol Biol Cell 2005; 16: 1375–1384. 4375–4385. 12 Ha CH, Wang W, Jhun BS, Wong C, Hausser A, Pfizenmaier K et al. Protein kinase 37 Pusapati GV, Krndija D, Armacki M, von Wichert G, von Blume J, Malhotra V et al. D-dependent phosphorylation and nuclear export of histone deacetylase 5 Role of the second cysteine-rich domain and Pro275 in protein kinase D2 mediates vascular endothelial growth factor-induced gene expression and interaction with ADP-ribosylation factor 1, trans-Golgi network recruitment, and angiogenesis. J Biol Chem 2008; 283: 14590–14599. protein transport. Mol Biol Cell 2010; 21: 1011–1022. 13 Hao Q, Wang L, Zhao ZJ, Tang H. Identification of protein kinase D2 as a pivotal 38 von Blume J, Knippschild U, Dequiedt F, Giamas G, Beck A, Auer A et al. regulator of endothelial cell proliferation, migration, and angiogenesis. J Biol Phosphorylation at Ser244 by CK1 determines nuclear localization and substrate Chem 2009; 284: 799–806. targeting of PKD2. EMBO J 2007; 26: 4619–4633. 14 Qin L, Zeng H, Zhao D. Requirement of protein kinase D tyrosine phosphorylation 39 Azoitei N, Kleger A, Schoo N, Thal DR, Brunner C, Pusapati GV et al. Protein kinase for VEGF-A165-induced angiogenesis through its interaction and regulation of D2 is a novel regulator of glioblastoma growth and tumor formation. Neuro Oncol phospholipase Cgamma phosphorylation. J Biol Chem 2006; 281: 32550–32558. 2011; 13: 710–724. 15 Wong C, Jin ZG. Protein kinase C-dependent protein kinase D activation mod- 40 Guha S, Tanasanvimon S, Sinnett-Smith J, Rozengurt E. Role of protein kinase D ulates ERK signal pathway and endothelial cell proliferation by vascular endo- signaling in pancreatic cancer. Biochem Pharmacol 2010; 80: 1946–1954. thelial growth factor. J Biol Chem 2005; 280: 33262–33269. 41 Norrby K. In vivo models of angiogenesis. J Cell Mol Med 2006; 10: 588–612. 16 Naik UP, Patel PM, Parise LV. Identification of a novel calcium-binding protein that 42 Vogler M, Giagkousiklidis S, Genze F, Gschwend JE, Debatin KM, Fulda S. Inhibition interacts with the integrin alphaIIb cytoplasmic domain. J Biol Chem 1997; 272: of clonogenic tumor growth: a novel function of Smac contributing to its 4651–4654. antitumor activity. Oncogene 2005; 24: 7190–7202. 17 Yuan W, Leisner TM, McFadden AW, Wang Z, Larson MK, Clark S et al. CIB1 is an 43 Hood JD, Frausto R, Kiosses WB, Schwartz MA, Cheresh DA. Differential alphav endogenous inhibitor of agonist-induced integrin alphaIIbbeta3 activation. J Cell integrin-mediated Ras-ERK signaling during two pathways of angiogenesis. JCell Biol 2006; 172: 169–175. Biol 2003; 162: 933–943. 18 Naik MU, Pham NT, Beebe K, Dai W, Naik UP. Calcium-dependent inhibition of 44 Dequiedt F, Kasler H, Fischle W, Kiermer V, Weinstein M, Herndier BG et al. HDAC7, polo-like kinase 3 activity by CIB1 in breast cancer cells. Int J Cancer 2011; 128: a thymus-specific class II histone deacetylase, regulates Nur77 transcription and 587–596. TCR-mediated apoptosis. Immunity 2003; 18: 687–698. 19 Shock DD, Naik UP, Brittain JE, Alahari SK, Sondek J, Parise LV. Calcium-dependent 45 Bard F, Malhotra V. The formation of TGN-to-plasma-membrane transport carriers. properties of CIB binding to the integrin alphaIIb cytoplasmic domain Annu Rev Cell Dev Biol 2006; 22: 439–455. and translocation to the platelet cytoskeleton. Biochem J 1999; 342(Pt 3): 46 Connolly CN, Futter CE, Gibson A, Hopkins CR, Cutler DF. Transport into and out of 729–735. the Golgi complex studied by transfecting cells with cDNAs encoding horseradish 20 Zayed MA, Yuan W, Leisner TM, Chalothorn D, McFadden AW, Schaller MD et al. peroxidase. J Cell Biol 1994; 127: 641–652. CIB1 regulates endothelial cells and ischemia-induced pathological and adaptive 47 Engling A, Backhaus R, Stegmayer C, Zehe C, Seelenmeyer C, Kehlenbach A et al. angiogenesis. Circ Res 2007; 101: 1185–1193. Biosynthetic FGF-2 is targeted to non-lipid raft microdomains following translo- 21 Yuan W, Leisner TM, McFadden AW, Clark S, Hiller S, Maeda N et al. CIB1 is cation to the extracellular surface of CHO cells. J Cell Sci 2002; 115(Pt 18): essential for mouse spermatogenesis. Mol Cell Biol 2006; 26: 8507–8514. 3619–3631. 22 Zayed MA, Yuan W, Chalothorn D, Faber JE, Parise LV. Tumor growth and 48 Florkiewicz RZ, Anchin J, Baird A. The inhibition of fibroblast growth factor-2 angiogenesis is impaired in CIB1 knockout mice. J Angiogenes Res 2010; 2:17. export by cardenolides implies a novel function for the catalytic subunit of 23 Hollenbach AD, McPherson CJ, Lagutina I, Grosveld G. The EF-hand calcium- Na þ ,K þ -ATPase. J Biol Chem 1998; 273: 544–551. binding protein calmyrin inhibits the transcriptional and DNA-binding activity of 49 Florkiewicz RZ, Majack RA, Buechler RD, Florkiewicz E. Quantitative export of Pax3. Biochim Biophys Acta 2002; 1574: 321–328. FGF-2 occurs through an alternative, energy-dependent, non-ER/Golgi pathway. 24 White C, Yang J, Monteiro MJ, Foskett JK. CIB1, a ubiquitously expressed Ca2 þ J Cell Physiol 1995; 162: 388–399. binding protein ligand of the InsP3 receptor Ca2 þ release channel. J Biol Chem 50 Jackson LN, Li J, Chen LA, Townsend CM, Evers BM. Overexpression of wild-type 2006; 281: 20825–20833. PKD2 leads to increased proliferation and invasion of BON endocrine cells. 25 Kauselmann G, Weiler M, Wulff P, Jessberger S, Konietzko U, Scafidi J et al. The Biochem Biophys Res Commun 2006; 348: 945–949. polo-like protein kinases Fnk and Snk associate with a Ca(2 þ )- and integrin- 51 Lacal PM, Ruffini F, Pagani E, D’Atri S. An autocrine loop directed by the vascular binding protein and are regulated dynamically with synaptic plasticity. EMBO J endothelial growth factor promotes invasiveness of human melanoma cells. Int J 1999; 18: 5528–5539. Oncol 2005; 27: 1625–1632. 26 Haataja L, Kaartinen V, Groffen J, Heisterkamp N. The small GTPase Rac3 interacts 52 Wu H, Zhu S, Mo YY. Suppression of cell growth and invasion by miR-205 in breast with the integrin-binding protein CIB and promotes integrin alpha(IIb)beta(3)- cancer. Cell Res 2009; 19: 439–448. mediated adhesion and spreading. J Biol Chem 2002; 277: 8321–8328. 53 Albini A. Tumor and endothelial cell invasion of basement membranes. The 27 Naik MU, Naik UP. Calcium-and integrin-binding protein regulates focal adhesion matrigel chemoinvasion assay as a tool for dissecting molecular mechanisms. kinase activity during platelet spreading on immobilized fibrinogen. Blood 2003; Pathol Oncol Res 1998; 4: 230–241. 102: 3629–3636. 54 Albini A, Benelli R, Noonan DM, Brigati C. The "chemoinvasion assay": a tool to 28 Leisner TM, Liu M, Jaffer ZM, Chernoff J, Parise LV. Essential role of CIB1 in study tumor and endothelial cell invasion of basement membranes. Int J Dev Biol regulating PAK1 activation and cell migration. J Cell Biol 2005; 170: 465–476. 2004; 48: 563–571.

& 2014 Macmillan Publishers Limited Oncogene (2014) 1167 – 1180 CIB1a mediates PKD2-induced tumour growth and angiogenesis M Armacki et al 1180 55 Albini A, Iwamoto Y, Kleinman HK, Martin GR, Aaronson SA, Kozlowski JM et al. A 60 Evans IM, Zachary IC. Protein kinase D in vascular biology and angiogenesis. rapid in vitro assay for quantitating the invasive potential of tumor cells. Cancer IUBMB Life 2011; 63: 258–263. Res 1987; 47: 3239–3245. 61 Malhotra V, Campelo F. PKD regulates membrane fission to generate TGN 56 Bartolome RA, Galvez BG, Longo N, Baleux F, Van Muijen GN, Sanchez-Mateos P to cell surface transport carriers. Cold Spring Harb Perspect Biol 2011; 3: pii: et al. Stromal cell-derived factor-1alpha promotes melanoma cell invasion across a005280. basement membranes involving stimulation of membrane-type 1 matrix metal- 62 Rozengurt E. Protein kinase D signaling: multiple biological functions in health loproteinase and Rho GTPase activities. Cancer Res 2004; 64: 2534–2543. and disease. Physiology (Bethesda) 2011; 26: 23–33. 57 Qazi H, Shi ZD, Tarbell JM. Fluid shear stress regulates the invasive potential of 63 Zhang C, Yadava P, Hughes J. Polyethylenimine strategies for plasmid delivery to glioma cells via modulation of migratory activity and matrix metalloproteinase brain-derived cells. Methods 2004; 33: 144–150. expression. PLoS One 2011; 6: e20348. 64 Sturany S, Van Lint J, Muller F, Wilda M, Hameister H, Hocker M et al. Molecular 58 Vempati P, Mac Gabhann F, Popel AS. Quantifying the proteolytic release of cloning and characterization of the human protein kinase D2. A novel member of extracellular matrix-sequestered VEGF with a computational model. PLoS One the protein kinase D family of serine threonine kinases. J Biol Chem 2001; 276: 2010; 5: e11860. 3310–3318. 59 Naik MU, Naik UP. Contra-regulation of calcium- and integrin-binding protein 65 Ruifrok AC. Quantification of immunohistochemical staining by color translation 1-induced cell migration on fibronectin by PAK1 and MAP kinase signaling. J Cell and automated thresholding. Anal Quant Cytol Histol 1997; 19: 107–113. Biochem 2011; 112: 3289–3299. 66 Celis JE. Cell Biology: A Laboratory Handbook. Academic Press: London, UK, 1998.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene (2014) 1167 – 1180 & 2014 Macmillan Publishers Limited