Background
Osteosarcomas are the most common primary malignant tumors of bone. They are highly aggressive with poor prognosis[
1,
2] and occur most frequently in children and adolescents[
3]. The efficacy of the current treatments has reached a plateau, and the need of increased biological understanding is crucial to improve treatment options and thus the life of patients.
At the genomic level, osteosarcomas show complex chromosomal rearrangements with multiple gains and losses[
4,
5]. Array comparative genomic hybridization (aCGH) has been used extensively to analyze DNA copy number changes at a higher resolution, identifying recurrent chromosomal alterations[
6‐
11]. We have previously identified a novel, frequent deletion in 3q13.31 in osteosarcomas[
6]. Of the genes located within the deleted region, three have been proposed to be involved in cancer biology: the protein-coding gene limbic system-associated membrane protein (
LSAMP) and the two non-coding RNAs LSAMP RNA antisense 3 (
LSAMP-AS3) (also known as
LOC285194 or
TUSC7) and LSAMP RNA antisense 4 (
LSAMP-AS4) (also known as
BC040587)[
6,
12‐
20].
LSAMP has previously been reported to be a candidate tumor suppressor gene in clear cell renal cell carcinoma and epithelial ovarian cancer[
15‐
17], and subsequently also in osteosarcomas[
6,
12‐
14]. All three of these genes have also been proposed to act in conjunction as tumor suppressors in osteosarcomas[
13].
In this study, the frequency and extent of the deletion and the aberrations of LSAMP were further investigated. To study the potential importance of LSAMP in osteosarcoma biology, we have examined the functional implications of LSAMP- reexpression in an osteosarcoma cell line with a homozygous deletion of the gene.
Discussion
Osteosarcomas are cytogenetically complex malignancies, with a vast number of DNA copy number aberrations. A recurrent deletion within 3q13.31 has been identified[
6,
13,
14,
27,
28]. This deletion has so far only been described in one other cancer type[
20], indicating that it is highly specific to osteosarcomas. This specificity argues that the deletion is not due to a fragile site and strongly suggests involvement of this region in osteosarcoma development. Thus, the deleted region is likely to contain sequences preventing or retarding osteosarcoma development or progression.
The simplest interpretation would be that the region harbors one or more tumor suppressor genes. Concordantly, three genes located within the deleted region have been proposed to have tumor suppressive functions in osteosarcomas:
LSAMP[
6,
12‐
14],
LSAMP-AS3[
13] and
LSAMP-AS4[
13]. However, our results showed either no or low expression levels of
LSAMP-AS3 and
LSAMP-AS4 in all the samples investigated, including the control samples. The control samples originated from osteoblastic cultures, bone, brain and smooth muscle. In addition, we included the Universal Human Reference RNA, which is a pool of RNA from ten different human cell lines representing various cancer types, excluding osteosarcoma. As neither
LSAMP-AS3 nor
LSAMP-AS4 were expressed in any of these control samples, it is reasonable to argue that the low or lack of expression in the osteosarcoma samples is not due to inactivation, but a normal state for both non-cancerous and cancerous cells. The other genes located within the deleted region (only annotated by ENSEMBL, supported by GENCODE/ENCODE) have been annotated only recently, and was thus not investigated in this study. However, stranded total RNA-sequencing data of 10 osteosarcoma cells lines and the non-tumorigenic cell line iMSC#3 (both undifferentiated and with osteogenic differentiation), showed no detectable expression corresponding to any of these genes (Meza-Zepeda
et al., unpublished). Even though we cannot exclude that other genes within the deleted region are important for the development of osteosarcomas, several lines of evidence indicate that
LSAMP is a tumor suppressor gene and the target for the deletion in 3q13.31. DNA methylation of the promoter region has been described, indicating that
LSAMP is epigenetically inactivated in cancer[
6,
17], and absent or low expression of
LSAMP has been reported to be a frequent characteristic of osteosarcomas[
6,
12‐
14]. We have previously shown that expression of
LSAMP is associated with poor survival in a larger panel of osteosarcoma patients[
6], and it has also been shown in patients with epithelial ovarian cancer[
15]. In this study, we saw a trend towards poorer survival in patients with loss of
LSAMP copy number, although not statistically significant, possibly due to the limited sample size (n = 18). Overexpression of
LSAMP was shown to inhibit proliferation in a renal cell carcinoma cell line[
17], whereas depletion of
LSAMP promoted cell proliferation in osteoblasts[
13]. These observations are consistent with the function of
LSAMP to be a tumor suppressor gene. Our results add to these evidences as we showed reduced proliferation rate
in vitro and inhibited tumor growth
in vivo when the expression of
LSAMP was restored.
By western blotting, we investigated whether the
LSAMP transcript was translated, which to our knowledge has not been done in tumor samples. However, we did not find a clear correlation between the mRNA level of
LSAMP and the protein level in our samples. Interestingly, one study has shown that in osteosarcomas with high mRNA levels of
LSAMP, a premature termination codon was detected[
14]. It is tempting to speculate that this is the cause for the apparent lack of translation in some of the samples. However, the premature termination codon was not found in any of 19 osteosarcoma cell lines based on available RNA-sequencing data (Meza-Zepeda
et al., unpublished), indicating that this is not a frequent event of regulation in osteosarcomas. Taking into account that 11/23 clones transfected with
LSAMP ORF failed to translate the transcripts to detectable levels, there could also be post-transcriptional regulation of
LSAMP, even though these transcripts are lacking features of the endogenous mRNA. Also, as only the clones with detectable levels of the LSAMP protein had upregulated
HES1, CTAG2 and
KLF10, our results indicate that the LSAMP protein is indirectly affecting the upregulation of these genes. However, it is also possible that the cells upregulated one or more of these genes to be able to grow in the presence of the
LSAMP protein. On the other hand, there were no differential expression of these genes between the investigated samples with (n = 7) and without (n = 23) detectable levels of the LSAMP protein, based on mRNA expression profiling data ([
21], Kresse
et al., unpublished). This discrepancy could be explained by the relatively small size of the cohort with LSAMP protein. However,
LSAMP is one of four IgLONs[
29], which are cell adhesion molecules that function as dimers, referred to as Diglons[
30]. LSAMP has been shown to only function as heterodimers with either Opioid-binding cell adhesion molecule (OBCAM) or Neurotrimin (NTM)[
30]. If the dimerization partner of LSAMP is not present, it would most likely not function properly.
OBCAM was not expressed in any of the investigated samples (n = 7 with LSAMP protein, n = 23 without LSAMP protein), but
NTM was expressed at different levels. Furthermore, the non-transfected parental cell line IOR/OS14 had among the highest expression levels of
NTM and ranking as number 11 out of the 30 samples (data not shown). Also, the majority of the cohort with LSAMP protein had low expression of
NTM. Thus, it is possible that we do not find an upregulation of
HES1,
CTAG2 and/or
KLF10 in these samples due to the lack of expression of a LSAMP dimerization partner.
HES1, CTAG2 and
KLF10 have all been shown to play a role in cancer biology.
HES1 has been suggested to have both oncogenic and tumor suppressive functions[
31,
32], whereas
KLF10 has been suggested to be a tumor suppressor gene[
33,
34].
CTAG2 is mainly expressed in the reproductive organs, in addition to a variety of cancers[
35‐
37]. To our knowledge, the function of
CTAG2 is not known, but both
HES1 and
KLF10 have been shown to be involved in bone biology. HES1 has been shown to bind to bone-specific promoters together with the runt-related transcription factor 2 (RUNX2) and the retinoblastoma protein (RB)[
38].
KLF10 has been shown to have an impact on the proliferation of osteoblasts, osteoclasts and osteosarcoma cells[
39,
40], which could be the cause for the observed reduced proliferation rate. In a recent study,
LSAMP clustered together with pro-apoptotic genes when gene expression changes of osteosarcoma cell lines were analyzed after induction of apoptosis[
41]. Furthermore, depletion of
LSAMP in osteoblasts had an effect on the expression of the proapoptotic genes
BCL2 and
BimEL[
13]. Although apoptosis could be a possible mechanism for
LSAMP to suppress or delay tumors formation, we did not find evidence of apoptosis in our clones with
LSAMP- reexpression
. This is consistent with another study where
LSAMP was overexpressed in a clear cell renal cell carcinoma cell line without any evidence of apoptotic cells[
17].
The high frequency of the deletion in 3q13.31 and the specificity for osteosarcomas strongly suggest a functional role for this region and that it harbors a tumor suppressor gene important for the development of osteosarcomas. Our results indicate that among the genes investigated in 3q13.31, LSAMP is the target for the deletion. The function of HES1 and KLF10 in bone biology also implies a function for these genes in osteosarcoma development. Furthermore, the reduced proliferation rate in vitro and inhibited tumor growth in vivo is further pointing to a tumor suppressor function of LSAMP.
Materials and methods
Samples
Clinical data for all osteosarcoma and control samples are listed in Additional file
1: Table S1.
Human osteosarcoma clinical samples (n = 39) were analyzed, of which 13 were primary or metastatic samples collected at the Norwegian Radium Hospital, Oslo, Norway and 26 were grown subcutaneously in immunodeficient mice as xenografts (suffix x), obtained either from the Norwegian Radium Hospital (n = 15) or the Department of Pathology, University of Valencia, Spain (n = 11). All tumors were diagnosed according to the current World Health Organization classification[
42]. The informed consent used and the collection of samples were approved by the Ethical Committee of Southern Norway, Project S-06133 or the Institutional Ethical Committee of the University of Valencia. The samples were collected immediately after surgery, snap frozen in liquid nitrogen and stored at -80°C. The establishment and passing of xenografts were in accordance with national and institutional animal care guidelines.
In addition, osteosarcoma cell lines (n = 21) were analyzed. These were: G-292, 143B, HOS, MNNG/HOS, MG-63, OSA (SJSA-1), Saos-2, U-2 OS (n = 9) (American Type Culture Collection; ATCC), HAL, KPD, MHM, OHS (n = 4) (The Norwegian Radium Hospital), IOR/OS9, IOR/OS10, IOR/OS14, IOR/OS15, IOR/OS18, IOR/MOS, IOR/SARG (n = 7) (Istituto Ortopedico Rizzoli, Bologna, Italy), ZK-58 (Heinrich-Heine University, Düsseldorf, Germany) and CAL 72 (University College London, London, UK). The cells were grown in RPMI 1640 or DMEM (both Lonza) supplemented with 10% FBS (Fisher Scientific) and GlutaMAX (Life Technologies), at 37°C in a humidified atmosphere with 5% CO
2. All cell lines were tested for mycoplasma and found negative, and authenticated as previously described[
43].
Control samples (n = 15) were also analyzed. Normal long bone samples were purchased from Capital Biosciences (n = 2) or obtained from amputations of cancer patients at the Norwegian Radium Hospital (n = 4) or University College London (n = 1). The normal bone was collected as distant as possible from the tumor site, and SNP array confirmed normal DNA copy number. Primary osteoblast cultures (n = 2) isolated from human calvaria of different donors were purchased from ScienCell Research Laboratories. The non-tumorigenic cell lines (n = 3) HEPM, hFOB (both ATCC) and iMSC#3 were included; the latter being an immortalized human bone marrow-derived mesenchymal stromal cell line established in our laboratory (Skårn et al., unpublished). Commercial normal RNA samples (n = 3) were used, one from brain tissue (Life Technologies), one from smooth muscle (Clonetech Laboratories) and Universal Human Reference RNA (Agilent Technologies).
DNA copy number and expression level
DNA copy number was determined either by high-resolution aCGH or the real-time PCR based assay TaqMan DNA Copy Number Assay (Life Technologies), as previously described[
21]. aCGH was performed using the Affymetrix Genome-Wide Human SNP Array 6.0 (Affymetrix) and DNA copy number analysis was performed using the Nexus software (BioDiscovery), as previously described[
21]. Expression level was determined using TaqMan Gene Expression Assays (Life Technologies) as previously described[
6], and is hereafter referred to as qRT-PCR (quantitative real-time reverse transcription PCR). The expression level of the clinical samples and cell lines was analyzed as previously described[
6], and the expression level of the clones was analyzed using the 2
-∆∆Ct-method[
44], with TATA box binding protein (
TBP) as an endogenous reference. The assays used and their respective ID number are listed in Additional file
7: Table S2.
Western blotting
Total protein lysate was run on a 4-12% Bis-Tris NuPAGE precast gel (Life Technologies) and transferred onto a PVDF membrane (Millipore). The antibodies used and their respective conditions are listed in Additional file
8: Table S3. The proteins were visualized using SuperSignal West Duration Substrate (Thermo Scientific).
Vector construction and transfection
The LSAMP expression vector was constructed using Gateway Technology (Life Technologies), recombining a vector containing LSAMP open reading frame (ORF) (ID number OCAAo5051A0349D, imaGenes) with pT-REx-DEST30 (Life Technologies). The cell line IOR/OS14 was stably transfected with either the expression vector (named LSAMP ORF) or the backbone vector using Lipofectamine 2000 (Life Technologies). Selection was performed using 450 μg/ml Geneticin (Life Technologies) for 14 days, after which the concentration was reduced to 225 μg/ml.
Immunofluorescence confocal microscopy
The anti-LSAMP antibody was a kind gift from Dr. Aurea F. Pimenta, Vanderbilt University, Nashville, USA. The cells were grown on coverslips and fixed in 10% formalin solution (Sigma-Aldrich), rinsed in PBS and blocked in 5% FBS in PBS for 30 min before incubation with the antibody in 1:100 dilution in 5% FBS for 1 h at RT. After incubation, the cells were washed 3 × 5 min in 5% FBS and incubated in a 1:200 dilution of anti-mouse-IgG/Cy3 (Jackson ImmunoResearch Laboratories) for 30 min at RT. The cells were then washed 3 × 5 min in 5% FBS and rinsed in dH2O. The nuclei were stained using ProLong Gold Antifade Reagent with DAPI (Life Technologies). The fluorescence was visualized using a Zeiss LSM 510 confocal microscope (Zeiss) and pictures were taken of thin single plane sections.
Proliferation rate
Proliferation rate was measured using the CellTiter 96 AQueous One Solution Cell Proliferation Assay (MTS) (Promega). Cells stably transfected with LSAMP ORF or the backbone vector were seeded in quadruplicates in a 96-well plate with 10,000 cells per well in 100 μl medium. The cell viability was measured after 96 h.
In vivo tumorigenicity
Animal experiments were performed according to protocols approved by the National Animal Research Authority in compliance with the European Convention of the Protection of Vertebrates Used for Scientific Purposes (approval ID 1499 and 3275,
http://www.fdu.no/). The experiments were performed as previously described[
45].
Apoptosis and cell cycle distribution
Apoptosis and cell cycle distribution were investigated using flow cytometry. For apoptosis, APO-BRDU (TUNEL) assay (Life Technologies) was performed, according to the manufacturer’s instructions. For investigation of cell cycle distribution, 2*106 cells were harvested and resuspended in 200 μl ice-cold PBS and added to 4 ml ice-cold ethanol and incubated on ice for 45 min. Then, 6 ml of ice-cold staining buffer (SB: 0.5% BSA in PBS) was added, and the cells were centrifuged at 300 × g, at 4°C for 5 min. The pellet was resuspended in 1 ml SB and the centrifugation repeated. The cells were resuspended in 300 μl SB containing 2 μg/ml Hoechst 33342 (Sigma-Aldrich). For both assays, the LSR II UV Laser (BD Bioscience) was used, and the data was analyzed using FlowJo v8.8.7 software (Tree Star).
Migration rate
The migration assay was performed using the IncuCyte system (Essen Bioscience).
Differentiation
Adipogenic differentiation and Oil red O staining were performed as previously described[
46], except that cells were seeded at a density of 3,000 cells/cm
2.
For the osteogenic differentiation, the cells were seeded at a density of 4,500 cells/cm2. Osteogenesis was initiated with osteogenic induction medium containing 10 nM Dexamethasone, 3.5 mM β-Glycerolphosphate and 66.7 μM Ascorbic acid 2-phosphate (all from Sigma-Aldrich). The osteogenic induction medium was replaced every third day. To estimate the degree of differentiation, the wells were washed with PBS, fixed in ice-cold 70% ethanol for 1 h at 4°C, washed with ddH2O and subsequently stained with 0.4% Alizarin Red S solution (w/v, pH 4.2; Sigma-Aldrich) for 10 min at RT. The staining solution was removed by washing the cells 5 × in ddH2O, followed by a 15 min wash in PBS. The cells were dehydrated with 70% ethanol, followed by absolute ethanol and air-dried. This procedure was performed after 0, 14, 21 and 28 days of differentiation, respectively.
mRNA expression profiling
RNA was isolated using the miRNeasy Mini Kit (QIAGEN GmbH). The RNA integrity was evaluated using the Agilent 2100 Bioanalyzer and the RNA nano 6000 kit (Agilent Technologies). For each sample, 500 ng of total RNA was used to make biotin-labeled and amplified cRNA with the Illumina TotalPrep Amplification Kit (Life Technologies). cRNA was hybridized to Illumina’s HumanHT-12 v4 Expression BeadChip as previously described[
47]. Expression values were annotated using the file HumanHT-12_V4_O_R2_15002873_B.bgx (Illumina). The expression data was quantile normalized[
48] in GenomeStudio Gene Expression module v1.9 (Illumina) and log
2-transformed, and a rank product analysis[
49] was performed in J-Express[
50] using a q-value < 0.05 to identify significant changes of gene expression. The dataset has been deposited in the GEO data repository (
http://www.ncbi.nlm.nih.gov/geo/, accession number GSE52089).
Statistical analysis
Statistical analyses were performed using SPSS version 20. A p-value < 0.05 was regarded as statistically significant.
Competing interests
The authors declare that they have no competing interests.
Authors’ contribution
TB carried out the experimental work, performed the statistical analysis and participated in data analysis and experimental design and wrote the manuscript. SHK participated in the statistical and data analysis, experimental design and writing of the manuscript. M. Skårn and M. Stabell participated in the experimental design and data analysis. RC performed the flow experiments and analysis. SL performed the immuofluorescence confocal microscopy and analysis. ALB provided biological material. OM and LAMZ conceived the study, participated in the design, data interpretation and manuscript writing. All authors read and approved the final manuscript.