OUP user menu

KIAA1797/FOCAD encodes a novel focal adhesion protein with tumour suppressor function in gliomas

Antje Brockschmidt, Detlef Trost, Heike Peterziel, Katrin Zimmermann, Marion Ehrler, Henriette Grassmann, Philipp-Niclas Pfenning, Anke Waha, Dirk Wohlleber, Felix F. Brockschmidt, Manfred Jugold, Alexander Hoischen, Claudia Kalla, Andreas Waha, Gerald Seifert, Percy A. Knolle, Eicke Latz, Volkmar H. Hans, Wolfgang Wick, Alexander Pfeifer, Peter Angel, Ruthild G. Weber
DOI: http://dx.doi.org/10.1093/brain/aws045 1027-1041 First published online: 16 March 2012

Summary

In a strategy to identify novel genes involved in glioma pathogenesis by molecular characterization of chromosomal translocation breakpoints, we identified the KIAA1797 gene, encoding a protein with an as yet undefined function, to be disrupted by a 7;9 translocation in a primary glioblastoma culture. Array-based comparative genomic hybridization detected deletions involving KIAA1797 in around half of glioblastoma cell lines and glioblastomas investigated. Quantification of messenger RNA levels in human tissues demonstrated highest KIAA1797 expression in brain, reduced levels in all glioblastoma cell lines and most glioblastomas and similar levels in glial and neuronal cells by analysis of different hippocampal regions from murine brain. Antibodies against KIAA1797 were generated and showed similar protein levels in cortex and subcortical white matter of human brain, while levels were significantly reduced in glioblastomas with KIAA1797 deletion. By immunofluorescence of astrocytoma cells, KIAA1797 co-localized with vinculin in focal adhesions. Physical interaction between KIAA1797 and vinculin was demonstrated via co-immunoprecipitation. Functional in vitro assays demonstrated a significant decrease in colony formation, migration and invasion capacity of LN18 and U87MG glioma cells carrying a homozygous KIAA1797 deletion ectopically expressing KIAA1797 compared with mock-transduced cells. In an in vivo orthotopic xenograft mouse model, U87MG tumour lesions expressing KIAA1797 had a significantly reduced volume compared to tumours not expressing KIAA1797. In summary, the frequently deleted KIAA1797 gene encodes a novel focal adhesion complex protein with tumour suppressor function in gliomas, which we name ‘focadhesin’. Since KIAA1797 genetic variation has been implicated in Alzheimer’s disease, our data are also relevant for neurodegeneration.

  • tumour relevant gene
  • functional characterization
  • focal adhesion complex
  • cell motility
  • glioma

Introduction

Diffuse astrocytic tumours have an invasive phenotype, which increases with malignancy from low-grade (astrocytoma grade II according to the WHO) to high-grade (WHO grade III: anaplastic astrocytoma, and WHO grade IV: glioblastoma) glioma. In glioblastoma, the most frequent and malignant primary brain tumour, invasiveness is one of the crucial features that impedes tumour cell eradication despite multimodal therapy and is thus a main determinate for the poor prognosis of patients with these tumours (Kleihues et al., 2007; Riemenschneider and Reifenberger, 2009; Jansen et al., 2010).

The significance of genes encoding proteins with a role in cell migration or invasion in glioblastoma pathogenesis is underlined by the results of recent comprehensive genetic studies. Among the top 10% of gene sets containing amplified, homozygously deleted or mutated genes most likely to drive glioblastoma tumorigenesis, over half are connected to cell adhesion via cell–matrix interaction, matrix remodelling or integrin priming, or to cytoskeleton remodelling impacting cell motility (Parsons et al., 2008). On the other hand, the complex genetic alterations underlying glioma pathogenesis involve many functional groups and three core pathways influencing cell growth, i.e. receptor tyrosine kinase/RAS/phosphatidylinositol 3-kinase targeting proliferation, survival and translation, p53 with an impact on senescence and apoptosis, and retinoblastoma signalling regulating G1/S-phase progression, are critically involved (The Cancer Genome Atlas Research Network, 2008).

By genome-wide array screens of copy number alterations, methylator phenotypes and expression patterns in gliomas, considerable progress has been made over the last few years in detecting new genetic aberrations and thus defining distinct subgroups and clinically relevant subtypes (Phillips et al., 2006; Parsons et al., 2008; The Cancer Genome Atlas Research Network, 2008; Noushmehr et al., 2010; Verhaak et al., 2010). Here, a different approach was used to identify new genes associated with glioma tumorigenesis. We characterized metaphase chromosomes from 11 glioblastoma primary cultures and cell lines by 24-colour fluorescence in situ hybridization (FISH) analysis to identify chromosomal translocations. From the aberrant chromosomes identified, we selected a derivative chromosome 9 with a translocation in the short arm. Loss on 9p is well established as a frequent event in malignant glioma cells (Riemenschneider and Reifenberger, 2009). The deletions on 9p are thought to preferentially target the genes CDKN2A and CDKN2B encoding proteins that regulate two critical cell cycle regulatory pathways, the p53 and the retinoblastoma pathway. However, molecular characterization of the translocation showed that the cyclin-dependent kinase inhibitor 2A and 2B genes, cyclin-dependent kinase inhibitor 2A (CDKN2A) and cyclin-dependent kinase inhibitor 2B (CDKN2B), were intact on the derivative chromosome 9, whereas the KIAA1797 gene located just 1 Mb distally in 9p21.3 was disrupted. KIAA1797 is a large (genomic size: 337 646 bp, 46 exons), highly conserved gene, which encodes a protein with an as yet undefined function. In this study, we provide evidence that KIAA1797 is a novel component of the focal adhesion complex expressed in glial and neuronal cells with a tumour suppressor function in gliomas, which we name ‘focadhesin’.

Materials and methods

Materials

Glioblastoma samples were collected from 13 female and 19 male patients, and cerebral cortex and subcortical white matter was from six temporal lobe resections for treatment of chronic focal drug-resistant epilepsy. All samples were analysed in an anonymous manner as approved by the ethics committee of the Medical Faculty of the University of Bonn. All tumours were histologically classified as grade IV according to the WHO classification of tumours of the nervous system (Kleihues et al., 2007). A tumour cell content of 80% or more was histologically verified for each tumour sample used for molecular analysis. All grey and white matter samples were free of tumour or inflammatory changes. The tissue was not used for diagnostic purposes and would have been discarded otherwise.

The human primary glioblastoma cultures Tu159 and Tu113, and human glioblastoma cell lines LN229, T98G, LN428, LN308, LN18, U373MG, A172, U138MG, U87MG, A178, U251MG and LN319 were cultured in Ham’s F10-Medium (Biochrom AG) supplemented with 10% BM Condimed (Roche Diagnostics), 7% foetal calf serum (PAA Laboratories), 0.5% l-glutamine (200 mM; Sigma Aldrich), 0.4% hydroxyethyl piperazineethanesulphonic acid buffer (1 M; Gibco, Invitrogen GmbH), 0.4% penicillin/streptomycin (MP Biomedicals) and 0.2% amphotericin B (250 µg/ml; MP Biomedicals), if not stated otherwise. Human astrocytes were obtained from Provitro, and cultured in astrocyte medium (Provitro) supplemented with 2% foetal bovine serum (Provitro), 1% astrocyte growth supplement (Provitro) and 1% penicillin/streptomycin solution (Provitro).

All animal work was approved by the governmental authorities (Karlsruhe, Germany) and supervised by institutional animal protection officials in accordance with the National Institutes of Health ‘Guide for the Care and Use of Laboratory Animals’.

For preparation of murine brain slices, mice (C57Bl6J, post-natal Day 17) were anaesthetized and sacrificed by decapitation. Their brains were dissected, washed and the hemispheres were cut into 200 µm thick slices in frontal orientation using a vibratome (Leica Microsystems VT 1000S). Slice preparation was performed at 6°C in external solution composed of (in mM): NaCl 87, KCl 2.5, MgCl2 7, CaCl2 0.5, d-glucose 25, NaHCO3 25, sucrose 75 (gassed with carbogen). Subsequently, slices were stored in artificial CSF gassed with carbogen at room temperature (pH 7.4).

24-Colour fluorescence in situ hybridization analysis

To screen the karyotype for chromosomal rearrangements, multicolour FISH (Speicher et al., 1996) was performed on metaphase chromosome preparations from glioblastoma primary cultures Tu113 and Tu159, and glioblastoma cell lines LN229, T98G, LN428, LN308, LN18, U373MG, U138MG, U251MG and LN319 using the 24XCyte Human Multicolour FISH (mFISH) Probe Kit (MetaSystems), which contains probes for all human chromosomes labelled with specific fluorochrome combinations, according to the manufacturer’s protocol. Fluorescent images were captured with a Leica DCX epifluorescent microscope coupled to a charge-coupled device camera and equipped with appropriate filter sets to allow specific detection of each fluorochrome used. Multicolour-FISH data were processed using the Leica CW 4000 software.

Characterization of t(7;9) and KIAA1797 copy number determination by fluorescence in situ hybridization

The translocation t(7;9) identified in Tu113 was characterized using multiple large insert clones from the breakpoint regions. Co-hybridization of a whole chromosome paint probe for chromosome 9 (Aquarius, Cytocell) according to the manufacturer’s instructions facilitated the identification of chromosome 9 in the metaphase spread. To investigate KIAA1797 copy number in relation to ploidy-state, bacterial artificial chromosome (BAC) clone RP11-512L9 (CHORI BACPAC Resources Centre) encompassing most of the KIAA1797 gene was co-hybridized with BAC clone RP11-18C9 (CHORI BACPAC Resources Centre) specific for chromosome 2, a chromosome that is usually not involved in aneuploidies in glioblastoma cells, to preparations of metaphase chromosomes and interphase nuclei from 10 glioblastoma primary cultures or cell lines and to tissue sections from the primary glioblastoma of patient Tu113. Labelling of BAC DNA was performed by nick translation, using biotin-deoxycytidine triphosphate (Invitrogen) or digoxigenin-11-dUTP (Roche Diagnostics). Avidin-FITC (fluorescein isothiocyanate) or TRITC (tetramethyl rhodamine isothiocyanate)-labelled antibodies, respectively, were used to detect the probes according to standard protocols.

Array-based comparative genomic hybridization analysis

To investigate copy number variation on a genome wide scale in DNA from glioblastoma primary culture Tu113, DNA microarrays with 6000 large insert clones, including the Sanger Centre 1 Mb clone set covering the genome at an average resolution of ∼1 Mb (Fiegler et al., 2003) and 3000 gene- and region-specific BAC clones (Zielinski et al., 2005), were used for array-based comparative genomic hybridization (array-CGH). Pooled reference DNAs from 10 healthy individuals were sex-matched. Array-CGH data were processed using the ChipYard framework (URL: http://www.dkfz.de/genetics/ChipYard/).

To determine and fine map chromosomal deletions on chromosome 9 encompassing the KIAA1797 gene, DNAs extracted from 10 glioblastoma primary cultures or cell lines (i.e. Tu113, Tu159, LN229, T98G, LN428, LN308, LN18, U373MG, U251MG and LN319) according to standard procedures (Sambrook et al., 1989) and DNAs extracted from 13 glioblastomas (i.e. cases 1, 111, 132, 235, 287, 542, 557, 620, 996, 1061, 1062, 1100 and 1240) by AllPrep® DNA/RNA/Protein Mini Kit (Qiagen) were hybridized to NimbleGen Human CGH 385k Chromosome 9 Tiling Arrays comprising 385 000 50- to 75-mer probes with a median probe spacing of 255 bp (Roche NimbleGen Systems). DNA labelling, array hybridization, post-hybridization washes and scanning were essentially performed according to the manufacturer’s instructions. Arrays were analysed using NimbleScan V2.4 extraction software and visualized in SignalMap V1.9 on a log2 scale (Roche NimbleGen Systems). A window containing 50 consecutive probes, i.e. on an average 13 000 bp, was used as averaging window for breakpoint determination. Alignment of array probes and genes were based on Genome Browser Assembly hg18 (http://www.ucsc.org).

Bisulphite sequencing, combined bisulphite restriction assay and treatment with the demethylating agent 5-aza-2′deoxycytidin

Treatment of DNAs from glioblastoma cell lines LN229, T98G, LN428, U373MG and U251MG, glioblastoma primary cultures Tu113 and Tu159 and primary glioblastoma cases 132, 620 and 1240 with sodium bisulphite using the EpiTect® Bisulfite Kit (Qiagen) was followed by PCR with primers KIAA1797-bf-forward-5′-ttgtagtggtatttttaggt-3′ and KIAA1797-bf-reverse-5′-gatggtgtttgagttgggtt-3′ generating a 345-bp PCR product in a non-repetitive GC rich region within intron 2 of KIAA1797 containing 38 putative CpG methylation sites. PCR products of bisulphite converted DNA were cloned, sequenced and sequence analysis was performed with the BiQ Analyzer software (Bock et al., 2005). After confirmation of sequence identity, bisulphite PCR products of glioblastoma cell lines, primary cultures, tumours and normal control DNA were digested for 5 min at 37°C in a volume of 20 µl containing 2 µl 10× buffer, 1 µl Bsh1236I (both Fermentas), 12 µl H2O and 5 µl PCR product. The PCR product contained six potential restriction sites (CGCG). As a positive control, SssI treated in vitro methylated DNA was included and investigated in the same experiments.

KIAA1797 messenger RNA expression analyses

To quantify the expression of KIAA1797 in neuronal and glial cells, the granule cell layer (containing mostly neuronal cells) and the hilus (polymorphic layer containing a high proportion of glial cells) of the dentate gyrus of the hippocampus was separated in murine brain slices with a small scalpel using a stereo microscope. To isolate total RNA, tissue samples were treated with TRIzol® (Invitrogen) and dissolved in 10 µl diethylpyrocarbonate treated water. Genomic DNA was removed by DNase treatment in a mixture containing PCR buffer, 2.5 mM MgCl2, 10 mM dithiothreitol (all Invitrogen), 20 U DNaseI (Roche) and 40 U RNase inhibitor (Promega) in a final volume of 20 µl and an incubation at 37°C for 30 min. Subsequently, messenger RNA was isolated using oligo(dT)25-linked Dynabeads® (Invitrogen), and the beads with the adherent messenger RNA were suspended in diethylpyrocarbonate treated water (20 µl). For reverse transcription, 10 µl of messenger RNA were incubated in a mixture containing reverse transcriptase (RT) reaction buffer, 4 × 1 mM deoxynucleotidetriphosphates, random hexamer primers, RNase inhibitor and 50 U reverse transcriptase (Applied Biosystems Deutschland GmbH) in a final volume of 20 µl and incubations at 25°C for 10 min, 37°C for 2 h and 85°C for 5 s. Mouse complementary DNA from both areas was subjected to real-time RT-PCR using mouse-specific TaqMan® Gene Expression Assay for mouse BC057079 (NM_001081184, mouse homologue of human KIAA1797) and mouse TaqMan® rodent glyceraldehyde-3-phosphate dehydrogenase control reagent (Applied Biosystems), and comparative Ct quantification (ΔΔCt) was applied. Two experiments were performed in triplicate.

To analyse the expression pattern of KIAA1797 in human tissues, we used the Human Multiple Tissue complementary DNA panels I and II, and the Human Digestive System Multiple Tissue complementary DNA panel for foetal tissues (Clontech-Takara Bio Europe). Relative quantifications in real-time experiments were performed using the inventoried TaqMan® Gene Expression Assay Hs00215057_m1 for the KIAA1797 gene (NM_017794.3, Applied Biosystems). Each sample was normalized to the TaqMan® endogenous control β-2-microglobulin, and ΔΔCt was applied.

To determine KIAA1797 messenger RNA levels in human glioblastoma, RNAs were isolated from 32 primary glioblastomas using the AllPrep® DNA/RNA/Protein Mini Kit (Qiagen) and nine glioblastoma primary cultures and cell lines using the protocol for animal cells of the RNeasy® Mini Kit (Qiagen), and 500 ng of RNA were subjected to reverse transcription using the SuperScript® First Strand Synthesis System for RT-PCR according to manufacturer’s instructions (Invitrogen). Equal amounts of complementary DNA were subjected to real-time RT-PCR analysis using the TaqMan® Gene Expression Assay for the KIAA1797 gene. Each sample was normalized to the TaqMan® endogenous controls β-actin and β-2-microglobulin and ΔΔCt was applied. The expression of the KIAA1797 gene in glioblastoma was calculated relative to the average value in normal whole brain RNA samples purchased from different distributors (Applied Biosystems; Clontech-Takara Bio Europe; BioCat GmbH; Stratagene-Agilent Technologies), which was arbitrarily defined as 1. Two experiments were performed in triplicate.

To analyse the influence of demethylating agents on messenger RNA expression, glioma cell lines LN229, T98G, LN428, LN308, U373MG, A172, U87MG and A178 were grown in the presence or absence of 0.5 or 1 µM 5-aza-2′-deoxycytidine (Sigma) for 3 days prior to RNA isolation. Equal amounts of complementary DNA from untreated cells, those treated with 0.5 µM 5-aza-2′-deoxycytidine, and those treated with 1 µM 5-aza-2′-deoxycytidine were subjected to real-time RT-PCR analysis using the TaqMan® Gene Expression Assay for the KIAA1797 gene. Each sample was normalized to the TaqMan® endogenous control β-2-microglobulin and ΔΔCt was applied.

KIAA1797 expression construct preparation

To generate an expression construct for KIAA1797, the synthetic clone 5′BamHI-NM_017794-SalI3′ was subcloned into the pcDNA3.1plus expression vector [=pcDNA3.1(+)_KIAA1797, imaGenes GmbH]. The construct was verified by sequencing.

To generate a lentivector construct expressing KIAA1797 and GFP, both complementary DNAs were cloned 3′ of a phosphoglycerate kinase (PGK) or cytomegalovirus (CMV) promoter resulting in rrl-cPPT-PGK-KIAA1797-PGK-GFP-WPRE (=PGK-KIAA1797_PGK-GFP) or rrl-cPPT-CMV-KIAA1797-PGK-GFP-WPRE (=CMV-KIAA1797_PGK-GFP). To generate a lentivector construct expressing myc-tagged KIAA1797, the myc-tag containing KIAA1797 complementary DNA was introduced via BamHI/SalI into the lentivector rrl-cPPT-PGK-GFP-WPRE thereby replacing GFP leading to rrl-cPPT-PGK-KIAA1797myc-WPRE (=PGK-KIAA1797myc). The original virus plasmids were kindly provided by Inder Verma, The Salk Institute for Biological Studies, Laboratory of Genetics, La Jolla, CA, USA.

Lentivirus vector preparation

Lentivirus vectors were prepared essentially as described by Pfeifer and Hofmann (2009). Briefly, lentiviral vectors were produced by calcium phosphate-transfection of HEK293T human embryonic kidney cells at ∼50% confluency with the lentivector, the packaging plasmids pMDLg/pRRE, RSV-rev and pMD.G (Dull et al., 1998). Human embryonic kidney cells were cultured in Dulbecco’s Modified Eagle Medium (Invitrogen) supplemented with 10% foetal calf serum (Biochrom AG) and 100 U/ml penicillin G/100 µg/ml streptomycin (Biochrom AG) at 3% CO2 and 37°C. Medium was changed 16–18 h after transfection and cells were cultured at 10% CO2 and 37°C. After 24 and 48 h, cell culture supernatant was collected and lentiviral vectors were concentrated by ultracentrifugation. The two pre-concentrated suspensions were combined, layered on top of a 2 ml 20% (weight/volume) sucrose cushion (Roth), and centrifuged for 2 h at 21 000 rotations per minute and 17°C. The virus pellet was resuspended in Hank’s Buffered Salt Solution (Invitrogen) and vortexed for 45 min at 1400 rpm and 17°C, followed by centrifugation (3 s, 16 000g) to pellet debris. Finally, the opaque supernatant containing the lentivirus vector was aliquoted and stored at −80°C.

Stable transduction of glioblastoma cell lines LN18 and U87MG with the lentivirus vector and flow sorting

Cells were seeded in 3 ml supplemented Dulbecco’s Modified Eagle Medium and incubated at 5% CO2 and 37°C. The cell culture medium was exchanged after 5 h. On the next day, the medium was replaced by 800 µl Dulbecco’s Modified Eagle Medium and transduced with a lentivirus vector containing 200 ng reverse transcriptase as measured with a commercially available colorimetric enzyme-linked immunosorbent assay (Roche Diagnostics). The medium was replaced with 3 ml Dulbecco’s Modified Eagle Medium on the next day, and cells were incubated for another 24 h at 5% CO2 and 37°C, then harvested and stored at −80°C.

Sorting of stably transduced green fluorescent protein (GFP) expressing cells was performed using a DiVa digital high-speed fluorescence activated cell sorter (Becton Dickinson). KIAA1797 expression was verified by Western-blot analysis.

Analysis of KIAA1797 protein levels

KIAA1797 expression was verified and KIAA1797 protein levels were determined by Western-blot analysis. Fresh frozen human cerebral cortex, subcortical white matter and glioblastoma tissue was homogenized and lysates from untransduced and transduced LN18 and U87MG glioma cells were generated in RIPA buffer (50 mM Tris, 150 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulphate, Protease Inhibitor Cocktail), denatured and used in equal amounts with loading buffer containing 100 mM dithiothreitol for sodium dodecyl sulphate-polyacrylamide gel electrophoresis (XCell SureLock™, Invitrogen). Proteins were blotted on a polyvinylidine difluoride membrane (Amersham Hybond-P, GE Healthcare) using a semi-wet transfer unit (XCell II™, Invitrogen). The KIAA1797 protein was detected using primary rabbit antibodies 14-06 or 14-07 (1:100 in 5% bovine serum albumin in Tris-buffered saline-Tween, Eurogentec S.A.) and β-actin using a rabbit anti-β-actin monoclonal antibody (1:16 000, Sigma Aldrich), followed by incubation with horseradish peroxidase-conjugated goat anti-rabbit antibody (1:4000, Santa Cruz Biotechnology Inc.). Blots were developed using a chemiluminescent detection kit (Applichem). KIAA1797 and β-actin Western-blot bands were quantified using the National Institutes of Health ImageJ software. The ratio of both signals was calculated to obtain relative KIAA1797 protein expression.

Migration and invasion assays

Cell migration and invasion assays were performed with flow sorted LN18 cells transduced with the lentivector construct expressing KIAA1797 and GFP (PGK-KIAA1797_PGK-GFP) and flow sorted LN18 cells transduced with the lentivector construct expressing GFP only (PGK-GFP, mock) in non-coated BD Cell Culture Inserts with 8 µm pores (transwell migration assay) or BD BioCoat™ Matrigel™ Invasion chambers with 8 µm pores (matrigel invasion assay; BD Biosciences) according to the manufacturer’s instructions. A total of 15 × 104 cells per insert were seeded; cells on top and bottom side of the membranes were trypsinized after 72 h, resuspended in 100 µl phosphate-buffered saline and quantified by flow cytometry. By fluorescence activated cell sorting-based quantification, migrating or invading GFP-positive cells (bottom side) were calculated relative to GFP-positive cells on top of the membranes.

Cell migration and invasion assays were also performed with U87MG cells. U87MG cells transduced with the lentivector construct expressing KIAA1797 and GFP (PGK-KIAA1797_PGK-GFP) and cells transduced with the lentivector construct expressing GFP only (PGK-GFP, mock) were seeded on both sides of IBIDI cell culture inserts (50 000 cells per side; IBIDI; cat# 80209) and cultured for 24 h to reach confluency. Then, cells were treated with 10 µg/ml mitomycin C (Sigma) for 1 h and washed intensively with phosphate-buffered saline. Subsequently, the inserts were removed (time point 0 h). Nine images per well were taken every 6 h using light microscopy and a digital camera (Olympus). The distance between the migration fronts was measured using the UTHSCSA image tool software (http://ddsdx.uthscsa.edu/dig/itdesc.html). Migration distance was calculated as gap width at 0 h minus gap width at 6, 12, 18 and 24 h divided by two. The cell invasion assays were performed using BD Fluoroblock BioCoat™ Tumor Invasion Systems for 24-wells (BD Biosciences) according to the manufacturer’s instructions with U87MG cells transduced with the lentivector construct expressing KIAA1797 and GFP (CMV-KIAA1797_PGK-GFP) and cells transduced with the lentivector construct expressing GFP only (PGK-GFP, mock). Per insert, 100 000 cells were seeded, and cell invasion was measured after 24 and 48 h. Cells were stained with 4 µg/ml Calcein BD™ AM Fluorescent Dye (BD Biosciences) in phosphate-buffered saline for 1 h at 37°C. The measurement was done using the BMG FLUOstar Optima device (BMG Labtech) with the bottom reading optic. Statistical analysis was performed using an unpaired t-test (http://www.graphpad.com).

Colony formation assay

To analyse cell proliferation in the presence and absence of KIAA1797, 2 × 105 LN18 cells lacking KIAA1797 messenger RNA expression as determined by real-time RT-PCR were seeded per well in a 6-well plate and transiently transfected with pcDNA3.1(+)_KIAA1797 expression vector (imaGenes GmbH) or empty vector (1 µg each) using Attractene transfection reagent (Qiagen). Twenty-four hours after transfection, cells were selected with 400 µg/ml G418 (Applichem) for 15–18 days. Cells were fixed and stained in 20% methanol and crystal violet and pictures were taken using a Fuji s3pro digital camera (FUJIFILM Europe GmbH). Two experiments were done in triplicate. Statistical analysis was performed using an unpaired t-test (http://www.graphpad.com).

Orthotopic xenograft mouse model

A total of 25 × 104 phosphate-buffered saline-suspended flow sorted U87MG cells transduced with the lentivector construct expressing KIAA1797 and GFP (PGK-KIAA1797_PGK-GFP) and flow sorted U87MG cells transduced with the lentivector construct expressing GFP only (PGK-GFP, mock) were stereotactically implanted into the right striatum of CD1 nu/nu mice (Charles River Laboratories). Tumour volume was assessed on Day 28 after tumour cell inoculation by MRI with n = 6 mice in each group. For histological assessment of tumour growth, these same mice were sacrificed at postoperative Day 28, brains were isolated, cryosectioned (8 µm) and stained with haematoxylin and eosin. Microscopic images were taken with a Nikon Eclipse 90i upright automated microscope (Nikon). The proliferation rate was determined using a Ki-67 antibody (1:100, catalogue number 27R5, Cell Marque) on a Ventana BenchMark XT immunostainer (Ventana Medical Systems) applying the manufacturer’s protocol. Approximately 400 nuclei per sample were evaluated and the number of intensely labelled nuclei scored.

Magnetic resonance imaging

MRI was performed using a 1.5-T whole-body scanner (Siemens Symphony) in combination with a custom-made radio-frequency small animal coil for excitation and signal reception. Brain tumours were located on T2-weighted turbo spin echo images (echo delay time, 109 ms; repetition time, 4000 ms; field of view, 40 × 30 mm; matrix, 128; voxel size, 0.3 × 0.3 × 1 mm). T1-weighted imaging was performed using a high resolution spin echo sequence (echo delay time, 600 ms; repetition time, 14 ms; field of view, 40 × 40 mm; matrix, 192; voxel size, 0.2 × 0.2 × 1 mm) 3 min after intraperitoneal administration of gadolinium-diethylenetriaminepentaacetate contrast agent (Magnevist, Bayer Schering Pharma, 0.5 mmol/ml). Statistical significance in tumour volumes was assessed by two-sided Student’s t-test (Excel, Microsoft). Values of P < 0.05 were considered significant.

Antibody generation

To characterize the KIAA1797 protein, a polyclonal antibody was generated using the anti-peptide 28-day Speedy Rabbit Programme (Eurogentec S.A.), which includes immunization of two rabbits with two selected antigens each. Before the first injection of the KIAA1797-specific peptides EP090006 (sequence: H2N-CME SPK EAL SAQ SRD L-COOH, located close to the C terminus) and EP090007 (sequence: H2N-CNG FSE KIH QST NQT P-CONH2, located close to the N terminus) coupled to the carrier protein KLH, sera of both rabbits were collected (pre-immune serum, PPI14 and PPI15 from rabbits SY1714 and SY1715). The antigen injections were repeated on Days 7, 10 and 18. On Day 20, small bleeds (GP14 and GP15) and on Day 28 final bleeds (SAB14 and SAB15) were collected from both rabbits and the sera were separated. After testing the sera, purification of the final bleed sera of both rabbits was performed using both antigens separately, producing four cleared antibody solutions: 14-06 and 14-07 from rabbit SY1714, and 15-06 and 15-07 from rabbit SY1715 (Eurogentec S.A.). Antibodies from rabbit SY1714 were preferentially used because they produced better signals in downstream applications.

Immunofluorescence

Human astrocytes and U87MG glioblastoma cells lacking KIAA1797 expression, which were untransfected or transiently transfected with pcDNA3.1(+)_KIAA1797 expression vector (imaGenes GmbH), were seeded on coverslips before fixation with 4% weight/volume paraformaldehyde, permeabilization with 0.1% Triton X and blocking with 5% fat-free milk powder/0.1% Triton X/1% goat serum. KIAA1797 was detected via the primary rabbit antibody 14-06 (Eurogentec S.A.). Co-localization analyses were performed using mouse anti-vinculin (1:200, Sigma Aldrich) and Alexa Fluor 488 phalloidin (1:40, Invitrogen), which specifically detects F-actin. As secondary antibodies, goat anti-rabbit Alexa Fluor 568 (1:350, Invitrogen) and goat anti-mouse Alexa Fluor 488 (1:500, Invitrogen) were used, and nuclei were visualized with 4′-6-diamidino-2-phenylindole counterstain. Fluorescent images were captured using a Leica TCS SP5 SMD confocal microscope.

Immunoprecipitation of myc-tagged KIAA1797 protein and Western-blot analysis

Protein lysates from LN18 cells stably transduced with lentivector PGK-KIAAmyc and untransduced LN18 cells were generated in 0.05 M hydroxyethyl piperazineethanesulphonic acid buffer containing 1 mM dithiothreitol using a cell scraper. Protein concentrations were measured using the BCA Protein Assay Kit according to the manufacturer’s instructions (Thermo Fisher Scientific). For immunoprecipitation of cell lysates with anti-c-myc conjugated agarose beads (Sigma Aldrich) the ratio of protein lysate (in µg) to anti-c-myc-agarose (in µl) was 1:5. For each cell lysate, a suspension of anti-c-myc conjugated agarose was settled in a microcentrifuge tube by a short spin (45 s at 8000g), the supernatant was removed, and the resin was washed three times with 200 µl phosphate-buffered saline. Subsequently, the respective cell lysate was added, and the final volume was brought to at least 200 µl with phosphate-buffered saline. The suspension of anti-c-myc agarose and cell lysate was incubated for 2.5 h on an orbital shaker at 4°C. Afterwards, the resin was pelleted by centrifugation and was washed three times with 200 µl phosphate-buffered saline. Finally, the supernatant was aspirated. For Western-blot analysis, 0.05 M hydroxyethyl piperazineethanesulphonic acid buffer containing 1 mM dithiothreitol and 2 × sodium dodecyl sulphate sample buffer containing 400 mM dithiothreitol were added in a ratio of 1:1, and the samples were denatured for 5 min at 95°C.

Samples were used in equal amounts for sodium dodecyl sulphate polyacrylamide gel electrophoresis (SE 600 Ruby, GE Healthcare). Proteins were blotted on a polyvinylidine difluoride membrane (Amersham Hybond-P, GE Healthcare) using a semi-dry transfer unit (TE77 ECL, GE Healthcare). The myc-tagged KIAA1797 protein was detected using primary rabbit antibodies 14-06 and 14-07 (1:250 in 5% bovine serum albumin in Tris-buffered saline-Tween, Eurogentec S.A.), and rabbit polyclonal to myc tag (1:1000; Abcam); vinculin using mouse anti-vinculin (1:10 000; Sigma Aldrich); focal adhesion kinase using mouse anti-Fak (B-8) (1:100; Santa Cruz Biotechnology Inc.); vasodilator-stimulated phosphoprotein (VASP) using mouse anti-VASP (A-11) (1:200; Santa Cruz Biotechnology Inc.) each followed by incubation with horseradish peroxidase-conjugated goat anti-mouse or anti-rabbit antibodies (Santa Cruz Biotechnology Inc.). Blots were developed using a chemiluminescent detection kit (Applichem).

Results

Genetic alterations of KIAA1797 in gliomas

Translocation and/or deletion of KIAA1797 in gliomas

24-colour-FISH analyses on metaphase chromosomes from 11 glioblastoma primary cultures and cell lines showed translocation events involving chromosome 9 in three cell lines, i.e. T98G, LN428 and Tu113. According to FISH analysis with locus-specific probes in Tu113, in which a translocation t(7;9) was identified, the breakpoint on the derivative chromosome 9 was located in chromosomal band 9p21.3 disrupting the KIAA1797 gene (genomic position according to NCBI built36/hg18 http://genome.ucsc.edu/cgi-bin/hgGateway: chr9:20 648 309–20 985 954 bp), because clone RP11-4E23 (chr9:20 673 303–20 674 687 bp) comprising the 5′-UTR of KIAA1797 was translocated to the derivative chromosome 7, and clones RP11-43M11 (chr9:20 784 481–20 907 097 bp) and RP11-279F22 (chr9:20 907 098–21 066 398 bp) containing the 3′-region of KIAA1797 were deleted. Clone RP11-149I2 (chr9:21 899 259–22 000 413 bp) containing the CDKN2A and CDKN2B genes was the most distal clone with chromosome 9 material mapping to the derivative chromosome 9 indicating that CDKN2A and CDKN2B were not deleted (Fig. 1A). Array-CGH using a whole genome DNA microarray showed that two other genes from the retinoblastoma pathway, i.e. the cyclin-dependent kinase 6 oncogene CDK6 and the retinoblastoma tumour suppressor gene RB1, had a copy number gain or loss, respectively, in Tu113.

Figure 1

KIAA1797 is frequently disrupted in glioblastomas by translocation or deletion. (A) The KIAA1797 gene located at 9p21.3 is disrupted by translocation and deletion in glioblastoma primary culture Tu113. By 24-colour-FISH, a derivative chromosome 9 [der(9)] was identified. Further FISH analyses mapped the breakpoint between probes RP11-4E23 (9p21.3), translocated to chromosome 7 (data not shown), and RP11-149I2 (9p21.3) indicating that CDKN2A and CDKN2B were intact on der(9). (B) Blow-up of array-CGH profiles (chr9:19 500 000–22 477 000 bp) after hybridization on NimbleGen chromosome 9-specific oligonucleotide microarrays (genomic location of KIAA1797 is shaded in grey). The 6 of 13 glioblastomas analysed are depicted, which show a partial (cases 1 and 132) or complete (cases 1240, 542, 111 and 620) KIAA1797 loss, probably representing a homozygous deletion in four cases (cases 1, 132, 542 and 111). (C) Analysis of primary culture Tu113 confirming the breakpoint within the 5′-UTR of KIAA1797 and a deletion of most of the gene by array-CGH (position chr9:19 500 000–22 477 000 bp is shown), shown to be a heterozygous deletion in ∼40% of cells using RP11-512L9 (red) as a KIAA1797-specific probe and control probe RP11-18C9 [green; 2q31.3 (upper panel, centre)] by FISH, which is already present in ∼50% of glioblastoma cells in tissue section from the primary tumour (lower panel). (D) Glioblastoma cell line LN18 with no structural alterations on chromosome 9 by 24-colour-FISH, a pronounced KIAA1797 loss by array-CGH (position chr9:19 500 000–22 477 000 bp is shown, KIAA1797 is shaded in grey) shown to be a homozygous deletion in ∼94% of cells using RP11-512L9 (red) as a probe for KIAA1797 and RP11-18C9 (green, 2q31.3) as a control probe by FISH to interphase and metaphase chromosomes.

Subsequent analysis of DNA from glioblastoma primary cultures and cell lines by high-resolution array-CGH using a NimbleGen chromosome 9 tiling oligonucleotide array identified microdeletions involving KIAA1797 in 5 of 10 (50%) cases, i.e. in Tu113, Tu159, LN319, LN428 and LN18. Among 13 glioblastomas analysed by array-CGH, six (46%) cases showed complete or partial deletions of the KIAA1797 gene (Fig. 1B).

To detect KIAA1797 microdeletions on a cellular level, FISH-analyses were performed on interphase nuclei of 10 glioblastoma primary cultures and cell lines with clone RP11-512L9 (chr9:20 678 303–20 857 828 bp), encompassing 21 coding exons of the KIAA1797 gene, in comparison with clone RP11-18C9 located on chromosome 2, which usually does not exhibit imbalances in gliomas and therefore represents the ploidy level of the cell. Deletions or relative copy number losses of KIAA1797 in over 20% of investigated cells were found in 9 of 10 (90%) cell lines (Fig. 1C and D; Supplementary Table 1).

To investigate whether the KIAA1797 deletion identified in primary culture Tu113 was already present in the glioblastoma, from which the primary culture was derived, a FISH-analysis with clones RP11-512L9 and RP11-18C9 was performed on a tissue section from the tumour of Patient Tu113. A KIAA1797 deletion or copy number loss was detected in 59% of tumour cells (RP11-18C9:RP11-512L9 ratio of 2:1 in 52/97, 3:2 in 5/97 cells) (Fig. 1C).

No evidence for methylation of KIAA1797 in gliomas

There was no evidence for hypermethylation in seven glioblastoma cell lines or primary cultures, or in three glioblastomas compared to normal brain in the locus tested, a non-repetitive CG-rich region in intron 2 of KIAA1797, as assessed by bisulphite sequencing or combined bisulphite restriction assay (Supplementary Fig. 1). In addition, treatment of DNA from eight glioma cell lines with 0.5 µM or 1 µM of the demethylating agent 5-aza-2′deoxycytidin did not affect KIAA1797 expression as measured by real-time RT-PCR (Supplementary Fig. 2).

Expression of the KIAA1797 gene

BC057079, the murine orthologue of KIAA1797, is expressed by glial and neuronal cells in murine brain

When comparing BC057079 expression by real-time RT-PCR experiments on mouse brain complementary DNA from the granule cell layer (containing mostly neuronal cells) and the hilus (polymorphic layer containing a high proportion of glial cells) of the dentate gyrus dissected from murine brain slices, both specimens showed a similar messenger RNA expression level (two experiments each performed in triplicate) (Fig. 2A).

Figure 2

Messenger RNA and protein expression of the KIAA1797 gene. (A) BC057079, the murine orthologue of KIAA1797, is expressed by glial and neuronal cells in the murine brain. Location of astrocytes and neurons in the murine dentate gyrus as revealed by immunostaining against glial fibrillary acidic protein (GFAP, stains astrocytes, strongest in the hilus) and neuronal nuclei (NeuN, stains neuronal cells, strongest in the stratum granulosum) (upper panel). Cells were microdissected from both regions and analysed separately by real-time RT-PCR showing similar expression levels in the hilus and the stratum granulosum (lower panel, results are means ± SD from two independent experiments). (B) KIAA1797 is most highly expressed in the human foetal and adult brain compared with messenger RNA levels from multiple other human tissues (dark blue bars, adult tissues; light blue bars, foetal tissues). Results are means ± SD from triplicate determination. (C) KIAA1797 messenger RNA expression is substantially reduced in most glioblastomas and all glioblastoma cell lines compared to four different commercially available whole brain messenger RNAs (Brain 1: Clontech-Takara Bio Europe, Brain 2: BioCat GmbH, Brain 3: Stratagene-Agilent Technologies, Brain 4: Applied Biosystems). Results are means ± SD from two experiments performed in triplicate. (D) KIAA1797 protein is expressed at similar levels in subcortical white matter and in cortex of the human brain (mean ± SD from three experiments, n = 6 cerebral cortex samples, n = 6 subcortical white matter samples, P = 0.097). (E) Antibodies generated against KIAA1797 specifically recognize a 200 kDa band in human cerebral cortex, subcortical white matter, astrocytes and glioblastomas without KIAA1797 deletion (GBM 1 and 2 without del). This band is not detected in glioblastoma cases 542 (GBM 1 with homo del) and 111 (GBM 2 with homo del), U87MG and LN18 glioma cells, which all carry homozygous KIAA1797 deletions and/or show no messenger RNA expression. However, in lysates from U87MG cells transduced with lentivector constructs PGK-KIAA1797 or CMV-KIAA1797 and from LN18 cells transfected with pcDNA3.1(+)_KIAA1797, the 200 kDa band is detectable. (F) KIAA1797 protein is expressed at significantly lower levels in glioblastomas with heterozygous or homozygous KIAA1797 deletions than in glioblastomas without KIAA1797 deletions (mean ± SD from three experiments, n = 4 glioblastomas with KIAA1797 deletion, n = 6 glioblastomas without KIAA1797 deletion, P < 0.01). GBM = glioblastoma multiforme.

KIAA1797 is highly expressed in human brain

We quantified the messenger RNA levels of KIAA1797 in various foetal and adult tissues including foetal and adult brain. KIAA1797 was expressed in all analysed tissues. In foetal tissues, KIAA1797 expression was highest in brain followed by muscle, kidney, heart, thymus, lung, liver and spleen. In adult tissues, highest KIAA1797 messenger RNA levels were detected in brain followed by testis, muscle, pancreas, heart, ovary, small intestine, placenta, prostate, thymus, kidney, colon, liver, lung, spleen and leukocytes (Fig. 2B).

KIAA1797 expression is reduced in most glioblastomas and all glioblastoma cell lines

KIAA1797 messenger RNA expression levels were measured in 32 glioblastomas, nine glioblastoma primary cultures and cell lines and four normal total brain RNA samples. Normalization to the housekeeping genes β-2-microglobulin (Fig. 2C) and β-actin (Supplementary Fig. 3) gave similar results showing decreased relative expression levels in all but one primary glioblastoma and all glioblastoma primary cultures and cell lines compared to normal brain. Glioblastoma cell lines LN18 and U87MG showed no measurable KIAA1797 expression (Fig. 2C; Supplementary Fig. 2 and 3).

KIAA1797 protein levels are similar in cerebral cortex and subcortical white matter of human brain

KIAA1797 levels were compared in cerebral cortex and subcortical white matter from six temporal lobe resections for treatment of chronic focal drug-resistant epilepsy. By Western-blot analysis, KIAA1797 expression was detected in both white and grey matter of the human brain. Comparable to the messenger RNA expression in murine brain, KIAA1797 protein levels in human brain samples were on an average slightly higher in white matter than in cortex, but not significantly different (P = 0.097, Fig. 2D and E).

KIAA1797 protein levels are decreased in glioblastomas with a KIAA1797 loss

KIAA1797 levels were determined in four glioblastomas with a KIAA1797 heterozygous or homozygous deletion according to array-CGH, i.e. cases 111, 542, 620, 1240 (Fig. 1B), and in six glioblastomas without a deletion of KIAA1797. Relative protein expression was significantly lower in tumours with heterozygous or homozygous KIAA1797 deletions compared with tumours without KIAA1797 deletion (P < 0.01, Fig. 2E and F).

KIAA1797 functions as a tumour suppressor in human gliomas

Since KIAA1797 is heterozygously or homozygously deleted and messenger RNA expression is reduced in at least half of glioblastoma primary tumours and cell lines, we hypothesized that KIAA1797 might have a tumour suppressor function in malignant gliomas. To identify potential tumour-relevant processes affected by KIAA1797, we performed in vitro and in vivo assays using glioblastoma cell lines LN18 and U87MG determined to have a homozygous KIAA1797 deletion, a complete lack of KIAA1797 messenger RNA expression, and no KIAA1797 protein expression (Fig 2E). Expression constructs containing KIAA1797 were generated and successfully used to reexpress the gene in these cells (Figs 2E and 6D).

Colony formation is reduced when KIAA1797 is introduced into LN18 cells

In the colony formation assay using LN18 cells, an average of 72 colonies formed when cells were transfected with pcDNA3.1(+)_KIAA1797 compared with an average of 289 colonies in the mock transfected cells (Fig. 3A). Thus, ectopic expression of KIAA1797 caused a 75% decrease in colony number (P < 0.001).

Figure 3

KIAA1797 functions as a tumour suppressor in human gliomas: results from in vitro assays using LN18 and U87MG glioma cells with homozygous KIAA1797 deletion. (A) Colony formation is decreased by ectopic expression of KIAA1797 in LN18 cells. The number of colonies formed by LN18 cells is significantly reduced after ectopic expression of KIAA1797 (P < 0.05). Results are means ± SEM from two experiments performed in triplicate. (B and C) Cell migration and invasion are decreased by ectopic expression of KIAA1797 in LN18 cells. Stable transduction of LN18 cells with a lentivector construct expressing KIAA1797 resulted in significantly decreased relative cell migration (B: P < 0.001) and significantly decreased relative cell invasion (C: P < 0.001) as compared with mock transduced LN18 cells. Results are means ± SEM from six (migration) or four (invasion) experiments. (D and E) Cell migration and invasion are also decreased in U87MG cells with ectopic KIAA1797 expression. Stable transduction of U87MG cells with a lentivector construct expressing KIAA1797 resulted in significantly decreased migration distance after 12, 18 and 24 h (D: P < 0.05), and significantly decreased relative cell invasion (E: P < 0.01) as compared with mock transduced U87MG cells. Results are means ± SEM from five (migration) or three (invasion) experiments.

Migration and invasion are reduced when KIAA1797 is introduced into LN18 and U87MG cells

We generated lentivector constructs expressing either both KIAA1797 and GFP (PGK-KIAA1797_PGK-GFP and CMV-KIAA1797_PGK-GFP) or GFP alone (PGK-GFP, mock) and prepared lentivirus vectors to analyse changes in migration or invasion capacity on transduced LN18 and U87MG cells. After transduction of cells with either PGK-KIAA1797_PGK-GFP, CMV-KIAA1797_PGK-GFP or PGK-GFP, GFP-positive cells were selected by flow sorting. We observed a significant decrease in migration of KIAA1797 transduced cells as compared to mock transduced LN18 cells (P < 0.001, Fig. 3B) and U87MG cells (P < 0.05, Fig. 3D). Quantification of GFP-positive LN18 cells revealed a significantly decreased invasion capacity of KIAA1797 transduced cells compared with mock transduced cells (P < 0.001, Fig. 3C). Similarly, invasion capacity was significantly reduced in U87MG cells ectopically expressing KIAA1797 compared with mock transduced cells not expressing KIAA1797 (P < 0.01, Fig. 3E).

Tumour volume is reduced in an orthotopic xenograft mouse model when KIAA1797 is introduced into U87MG cells

U87MG cells transduced with PGK-KIAA1797_PGK-GFP or PGK-GFP (mock) and flow sorted were stereotactically implanted into the right striatum of CD1 nu/nu mice. The volume of U87MG tumour lesions assessed by MRI was significantly lower in the six mice implanted with U87MG cells ectopically expressing KIAA1797 versus the six mice with mock transduced cells not expressing KIAA1797 (P < 0.05, Fig. 4). The average proliferation rate as determined by Ki-67 labelling was lower in four U87MG tumour lesions expressing KIAA1797 (mean ± SD: 40 ± 2.5%) compared with four tumours not expressing KIAA1797 (mean ± SD: 58 ± 3%).

Figure 4

KIAA1797 significantly reduces tumour growth in an in vivo orthotopic xenograft mouse model. (A) Representative cranial T1-based contrast agent enhanced MRI scans and haematoxylin and eosin stained brain sections from CD1 nu/nu mice implanted with U87MG glioma cells, which carry a homozygous KIAA1797 deletion, transduced with mock or KIAA1797 expressing lentiviral constructs. (B) Comparative T1-based MRI volumetry of U87MG tumour lesions not expressing (mock) or ectopically expressing KIAA1797 (mean ± SEM, n = 6 animals per group, P < 0.05).

KIAA1797 is a novel focal adhesion complex protein

To investigate the subcellular localization of KIAA1797, which might provide a clue for how KIAA1797 exerts its tumour suppressor function, antibodies were raised against peptides from the N-terminus (14-07) and the C-terminus (14-06) of the protein. By Western-blot analyses, both antibodies recognized a 200 kDa band, which is in line with the predicted molecular weight of KIAA1797, in lysates from human cerebral cortex, subcortical white matter, astrocytes and glioblastomas without KIAA1797 deletion, whereas this band was not found in lysates from LN18 and U87MG cells, and in glioblastoma cases 111 and 542, which all carry a homozygous KIAA1797 deletion (Fig. 2E). After transduction of U87MG cells with the expression constructs PGK-KIAA1797_PGK-GFP or CMV-KIAA1797_PGK-GFP, and transfection of LN18 cells with the expression construct pcDNA3.1( + )_KIAA1797, KIAA1797 was detectable in lysates as a 200 kDa band (Fig. 2E).

KIAA1797 localizes to the end of actin filaments

By immunofluorescence, the specific signals detected in human astrocytes were located in proximity to the cell membrane and showed a specific pattern reminiscent of focal adhesions, which act as an anchor for the actin cytoskeleton to the cell membrane (antibody 14-06; Fig. 5A). Co-staining with filamentous actin, which normally terminates at focal adhesions, confirmed that the KIAA1797 stain was located at the end of actin filaments (Fig. 5A). Vinculin, a major component of focal adhesions, showed the same specific pattern as KIAA1797 on immunofluorescence of human astrocytes and also localized to the end of filamentous actin (Fig. 5B). These data suggest that KIAA1797 is a novel component of the focal adhesion complex.

Figure 5

KIAA1797, like vinculin, localizes to the end of actin stress fibres, which normally terminate at focal adhesions. In human astrocytes, filamentous actin was visualized using Alexa488-phalloidin, and (A) KIAA1797 and (B) vinculin were stained for. KIAA1797 was visualized by staining with 14-06, a polyclonal anti-peptide antibody generated in this study. Diamidino-2-phenylindole was used as counterstain to delineate the nucleus. Fluorescent images were captured using a Leica TCS SP5 SMD confocal microscope.

KIAA1797 co-localizes with vinculin in human astrocytes and upon ectopic expression in glioblastoma cell line U87MG

The notion that KIAA1797 is part of the focal adhesion complex was further substantiated by the finding that KIAA1797 and vinculin co-localize in human astrocytes (Fig. 6A). There was no signal for KIAA1797 and no co-localization with vinculin in glioblastoma cell line U87MG (Fig. 6B). After transfection of U87MG cells with the expression construct pcDNA3.1(+)_KIAA1797, immunofluorescence showed that KIAA1797 expression and co-localization with vinculin was restored, demonstrating that ectopically expressed KIAA1797 is integrated into focal adhesions (Fig. 6C).

Figure 6

KIAA1797 and vinculin co-localize in focal adhesions and physically interact. (A) Human astrocytes stained with anti-KIAA1797 (14-06) and anti-vinculin antibodies. Staining of KIAA1797 and vinculin overlaps in focal adhesions. (B) Whereas vinculin localizes to focal adhesions, no KIAA1797 staining was detected in glioblastoma cell line U87MG shown to have no protein expression of KIAA1797 by Western-blot analysis (Fig. 2E). (C) Ectopic expression of KIAA1797 in U87MG cells by transfection with a KIAA1797 expression construct reconstitutes co-localization of KIAA1797 and vinculin in focal adhesions. (D) Interaction of KIAA1797 with vinculin is demonstrated by co-immunoprecipitation of vinculin with KIAA1797 in LN18 cells transduced with a lentiviral construct expressing myc-tagged KIAA1797, whereas VASP is not pulled down. The 200 kDa KIAA1797 band is detected by both the anti-myc antibody and 14-07, the anti-KIAA1797 antibody generated here. (E) Scheme showing that KIAA1797 is located at the end of actin filaments and interacts with vinculin in focal adhesions. ECM = extracellular matrix.

KIAA1797 interacts with vinculin

To obtain biochemical evidence for physical interaction between KIAA1797 and the focal adhesion components vinculin and VASP, we transduced LN18 cells with a myc-tagged KIAA1797 lentivector construct (PGK-KIAA1797myc) and performed immunoprecipitation with anti-myc conjugated agarose beads after cell lysis of untransduced and transduced LN18 cells. While input lanes show that vinculin and VASP are present in both transduced and parental LN18 cells, after immunoprecipitation only signals for KIAA1797 and vinculin were detectable in transduced LN18 lysates (Fig. 6D). These data demonstrate co-immunoprecipitation of KIAA1797 with vinculin, as was reported previously in a large scale screen for interaction partners of known focal adhesion proteins (de Hoog et al., 2004), whereas there was no evidence for interaction with VASP. We also show that the anti-myc and 14-07 antibodies both detect a 200 kDa band corresponding to the KIAA1797 protein providing further evidence for the specificity of the anti-KIAA1797 antibodies generated here.

Discussion

In this study, KIAA1797 was identified as a novel tumour suppressor gene in malignant gliomas, because it was either disrupted by translocation or heterozygously or homozygously deleted, and messenger RNA as well as protein expression was reduced in at least half of glioblastoma cell lines or tumours, findings in line with the data for KIAA1797 in 536 glioblastomas contained in the TCGA database (http://cancergenome.nih.gov). Furthermore, colony formation, cell migration and cell invasion were significantly reduced in KIAA1797 expressing versus non-expressing cells in two malignant glioma cell lines using in vitro assays. In an in vivo xenograft model of U87MG glioma cells orthotopically implanted in nude mice, a well-established model for human glioblastoma (Fulda et al., 2002), tumour lesions expressing KIAA1797 had a volume significantly reduced by ∼50% and a proliferation rate decreased by ∼30% compared with tumours not expressing KIAA1797. These data suggest that reduced in vivo growth of U87MG tumours expressing KIAA1797 may be due to both a decrease in proliferation and in spread/infiltration of glioma cells into the brain. However, additional studies are necessary to determine whether KIAA1797 acts predominantly as a suppressor of cell proliferation or migration/invasion capacity.

Our data additionally suggest that KIAA1797 exerts its tumour suppressor activity by functioning as a component of the focal adhesion complex (Fig. 6E). Focal adhesions are specialized attachment and signalling centres, which form at sites of cell-matrix contacts and couple the extracellular matrix through integrins to the actin cytoskeleton (Dubash et al., 2009). Some of the known components of focal adhesions, such as vinculin and paxillin, mainly play a structural role in the maintenance and assembly of the complex (Ziegler et al., 2006), whereas others are signalling molecules, essential for organizing integrin and growth factor signalling cascades. Thus, focal adhesions play a role in cellular processes as diverse as motility, proliferation, differentiation, regulation of gene expression and survival (Dubash et al., 2009).

Focal adhesions influence growth control by transmitting the integrin and tensional signals via cascades, including focal adhesion kinase, Rho GTPases and extracellular signal-regulated kinase, to the cell cycle (Assoian and Klein, 2008). Cells lacking vinculin (vin−/−) have increased survival due to upregulated activity of extracellular signal-regulated kinase, which results from vinculin’s modulation of paxillin-focal adhesion kinase interactions (Subauste et al., 2004). Migration is diminished in cells that are strongly attached with abundant focal adhesions. Accordingly, cell motility increased after knock down of vinculin in mouse embryonic fibroblasts (Rodriguez Fernandez et al., 1993), and vinculin-null mouse F9 embryonal carcinoma cells were less adhesive and moved twice as fast as wild-type F9 cells (Coll et al., 1995). We have shown similar effects in vitro and in vivo for KIAA1797-null glioma cells versus cells ectopically expressing KIAA1797. Most likely, the functional similarities between KIAA1797 and vinculin are due to the fact that both are components of the focal adhesion complex and may be caused by their interaction.

Focal adhesions play a vital role in many organs. Accordingly, vinculin knockout results in heart and brain defects in mouse embryos that do not survive past embryonic Day 10 (Xu et al., 1998), implying that KIAA1797 knockout, which has not been described so far, may also cause a severe phenotype. Interestingly, two recent studies implicate KIAA1797 genetic variation in heart physiology and neurodegeneration (Chapuis et al., 2009; Melton et al., 2010). An intronic single nucleotide polymorphism in KIAA1797 was reported to show the strongest association with heart rate of 2248 single nucleotide polymorphisms tested on 9p21 in American Indians, and KIAA1797 expression levels in lymphocytes were significantly associated with heart rate in Mexican Americans (Melton et al., 2010). In a study of 82 genes differentially expressed in brain tissue of patients with Alzheimer’s disease and controls, KIAA1797 (synonym: FLJ20375) was 1 of 17 genes exhibiting at least one polymorphism associated with risk for Alzheimer’s disease (Chapuis et al., 2009). Our findings shed new light on these data, because we show that KIAA1797 is expressed in the normal foetal and adult human heart and brain. Furthermore, our data suggesting that KIAA1797 encodes a protein associated with the focal adhesion complex implies that KIAA1797 could play a role in cardiomyocyte mechanotransduction, a process that affects the beat-to-beat regulation of cardiac performance, and in Alzheimer’s disease by mediating signal transmission of extracellular matrix components, such as beta-amyloid peptide, to the cell cycle (Samarel, 2005; Caltagarone et al., 2007).

The KIAA1797 gene is located in chromosomal band 9p21.3 ∼1 Mb distally to the tumour suppressor genes CDKN2A and CDKN2B, which code for cell cycle inhibitors regulating G1–S-phase transition and are homozygously deleted or mutated in ∼50% of glioblastomas (Cancer Genome Atlas Research Network, 2008; Parsons et al., 2008). Many deletions identified here include CDKN2A and CDKN2B, in addition to KIAA1797. From our data that KIAA1797 is a component of the focal adhesion complex with an impact on cell motility, it is tempting to speculate that deletions encompassing all three genes will have an additive effect by impacting both the cell cycle and cell invasiveness, generating a tumour cell with accelerated proliferation rates and enhanced migratory activity.

However, there is evidence that KIAA1797 by itself possesses tumour suppression activity. On the derivative chromosome 9 identified in our index primary glioblastoma culture, KIAA1797 was disrupted but both CDKN2A and CDKN2B remained intact. In two glioblastomas, an intragenic deletion of KIAA1797 was detected. In addition, we could show that ectopic KIAA1797 expression suppressed tumour growth in vivo in an orthotopic xenograft mouse model. Further evidence comes from genome-wide single nucleotide polymorphism array studies on glioblastoma DNA identifying homozygous deletions in 9p21.3 including KIAA1797 but excluding CDKN2A and CDKN2B in 3 of 22 cases analysed (Parsons et al., 2008). A recent report even connects the KIAA1797 gene to cancer predisposition. An intragenic deletion encompassing exons 4 to 21 of KIAA1797 and including the hsa-miR-491 microRNA gene was detected in germline DNA from a patient with early-onset colorectal cancer (Venkatachalam et al., 2011). In this report, it was hypothesized that this deletion targets the microRNA gene because microRNAs, which are regulatory non-coding RNAs of 21–24 nt in length, are known to be involved in cancer development (Garzon et al., 2009). Most deletions found here also encompass the hsa-miR-491 gene. However, our findings suggest that the KIAA1797 gene functions as a tumour suppressor implying that the disruption of the KIAA1797 open reading frame may be the genetic defect underlying the cancer predisposition in the patient described.

In summary, we found the previously undefined protein KIAA1797 to be a novel component of the focal adhesion complex expressed in glial and neuronal cells that has a tumour suppressor function in human gliomas. On the basis of our findings, the HUGO Gene Nomenclature Committee has suggested the symbol ‘FOCAD’ for the KIAA1797 gene and the name ‘focadhesin’ for the KIAA1797 protein.

Funding

Wilhelm Sander-Stiftung (2008.014.1) and German Ministry for Education and Research (National Network for Genome Research, NGFN-2, Brain Tumor Network).

Supplementary material

Supplementary material is available at Brain online.

Acknowledgements

The authors thank Sebastian Franken, Volkmar Gieselmann and Martin Theis, University of Bonn; Jochen Hess, Deutsches Krebsforschungszentrum Heidelberg and Dagmar-Christiane Fischer, University of Rostock for helpful scientific discussions; Stefanie Krenzer, Deutsches Krebsforschungszentrum Heidelberg, for technical advice and Christina Ergang, Vera Riehmer and Rabea Wagener, University of Bonn, for technical assistance.

Abbreviations
array-CGH
array-based comparative genomic hybridization
BAC
bacterial artificial chromosome
CDKN2A
cyclin-dependent kinase inhibitor 2A
CDKN2B
cyclin-dependent kinase inhibitor 2B
CMV
cytomegalovirus
FISH
fluorescence in situ hybridization
GFP
green fluorescent protein
PCR
polymerase chain reaction
PGK
phosphoglycerate kinase
RT
reverse transcriptase
VASP
vasodilator-stimulated phosphoprotein

References

View Abstract