Identification of Signaling Pathways Associated with Cancer Protection in Laron Syndrome

235

L Lapkina-Gendler et al. Cancer protection in Laron 23:5 399–410 Research syndrome

Identification of signaling pathways associated with cancer protection in Laron syndrome

Lena Lapkina-Gendler1, Itai Rotem1, Metsada Pasmanik-Chor2, David Gurwitz1,3, Rive Sarfstein1, Zvi Laron4 and Haim Werner1,3

1 Department of Human Molecular Genetics and Biochemistry, Sackler School of Medicine, Correspondence Tel Aviv University, Tel Aviv, Israel should be addressed 2 Bioinformatics Unit, George Wise Faculty of Life Sciences, Tel Aviv University, Tel Aviv, Israel to H Werner 3 Yoran Institute for Human Genome Research, Tel Aviv University, Tel Aviv, Israel Email 4 Endocrine and Diabetes Research Unit, Schneider Children’s Medical Center, Petah Tikva, Israel [email protected]

Abstract

The growth hormone (GH)–insulin-like growth factor-1 (IGF1) pathway emerged in recent Key Words years as a critical player in cancer biology. Enhanced expression or activation of specific ff Laron syndrome components of the GH–IGF1 axis, including the IGF1 receptor (IGF1R), is consistently ff growth hormone associated with a transformed phenotype. Recent epidemiological studies have shown receptor that patients with Laron syndrome (LS), the best-characterized entity among the ff insulin-like growth factor-1 congenital IGF1 deficiencies, seem to be protected from cancer development. To identify ff dwarfism IGF1-dependent genes and signaling pathways associated with cancer protection in LS, ff cancer protection

Endocrine-Related Cancer Endocrine-Related we conducted a genome-wide analysis using immortalized lymphoblastoid cells derived from LS patients and healthy controls of the same gender, age range, and ethnic origin. Our analyses identified a collection of genes that are either over- or under-represented in LS-derived lymphoblastoids. Gene differential expression occurs in several gene families, including cell cycle, metabolic control, cytokine–cytokine receptor interaction, Jak-STAT signaling, and PI3K-AKT signaling. Major differences between LS and healthy controls were also noticed in pathways associated with cell cycle distribution, apoptosis, and autophagy. Our results highlight the key role of the GH–IGF1 axis in the initiation and progression of cancer. Furthermore, data are consistent with the concept that homozygous congenital IGF1 deficiency may confer protection against future tumor development. Endocrine-Related Cancer (2016) 23, 399–410

Introduction

The growth hormone (GH)–insulin-like growth factor-1 also produced by extra-hepatic tissues and circulates as a (IGF1) axis has a fundamental role in growth and ternary complex with IGF-binding protein-3 (IGFBP3) and development throughout life (Rosenfeld 2005). As an acid-labile subunit (ALS) (LeRoith et al. 2007). The IGF1 originally postulated by Salmon et al. in the mid-1950s, receptor (IGF1R), which mediates the biological actions anabolic actions of GH are mediated by a liver-produced of IGF1 and IGF2, is structurally and functionally related peptide initially termed somatomedin and, subsequently, to the insulin receptor. IGF1 displays potent proliferative IGF1 (Salmon et al. 1957, LeRoith et al. 2001). IGF1 is and antiapoptotic activities and is regarded as a key player

http://erc.endocrinology-journals.org © 2016 Society for Endocrinology Published by Bioscientifica Ltd. DOI: 10.1530/ERC-16-0054 Printed in Great Britain Downloaded from Bioscientifica.com at 10/02/2021 05:56:59AM via free access

10.1530/ERC-16-0054 Research L Lapkina-Gendler et al. Cancer protection in Laron 23:5 400 syndrome

in malignancy (Baserga et al. 2003, Pollak 2008, Amichay These studies included 538 patients, including 230 et al. 2012, Werner & Bruchim 2012). patients with LS and 308 patients with congenital IGHD Growth retardation in infants is multifactorial, albeit a (cIGHD), GHRH-R defects, and congenital multiple large portion of cases remains idiopathic because no specific pituitary hormone deficiency (cMPHD). The age range of (genetic or other) defect can be identified (Klammt et al. LS patients was 1–75 years. The studies also included 752 2008). Prenatal IGF1 expression is GH-independent and first-degree relatives and further family members. Results becomes GH-dependent before birth. IGF1 biosynthesis of these studies revealed that none of the 230 homozygous remains dependent on GH secretion during postnatal LS patients, including patients who were treated with life. Congenital IGF1 deficiency is characterized by low IGF1 or GH, had a history of cancer. Only one boy with to undetectable serum IGF1 levels, but normal to elevated xeroderma pigmentosum, out of 116 patients with IGHD, GH production. These conditions may result from the presented a basal cell carcinoma. However, 30 patients with following: (1) GH-releasing hormone-receptor (GHRH-R) cancer were reported among 752 first-degree heterozygote gene defect, (2) GH gene deletion (isolated GH deficiency, family members (4%) and 31 malignancies (mainly lung, IGHD), (3) GH receptor (GH-R) gene defects (Laron prostate, breast, and colon tumors) were reported among syndrome, LS), and (4) IGF1 gene deletion. Additional 131 further relatives (23.7%). Despite the small size of this conditions leading to congenital IGF1 deficiency are cohort, differences in cancer incidence between LS patients defects of post-GH-R signaling (e.g. STAT5 defects) and and relatives were regarded as statistically significant. ALS mutations (Woods et al. 1996, Kofoed et al. 2003, Furthermore, this cohort comprises more than half of the Domene et al. 2004, Cohen et al. 2008). estimated total number of LS patients worldwide. This Laron syndrome (OMIM # 262500), or primary GH finding is supported by animal experiments using the insensitivity, is a recessively transmitted genetic form of Gh- r/Gh-Bp KO mouse (‘Laron mouse’) (Laron et al. 2011). dwarfism caused by deletion or mutation of theGH-R gene, Similar to humans, these animals had reduced incidence of leading to congenital IGF1 deficiency (Laron 1999, 2004). prostate and mammary tumors in experimental models of This genetic entity was identified in the late 1950s and cancer (Wang et al. 2005). reported by Laron et al. (1966), and subsequently in families The finding that the rate of cancer incidence in LS with a high incidence of consanguinity (Laron et al. 1966, patients is very low is consistent with the notion that 2011). The typical features of LS are short stature (−4 to homozygous congenital IGF1 deficiency may confer 10 SDS), typical face, obesity, high basal serum GH, and protection against future development of a tumor. The

Endocrine-Related Cancer Endocrine-Related − low IGF1, unresponsive to administration of exogenous present genome-wide analysis was designed to identify GH. The recognition that an inherited defect of the GH-R IGF1-dependent genes and signaling pathways associated gene, including exon deletions and point mutations, with cancer protection in LS. Our results identified a is the etiological factor behind LS that was reported in number of genes that are either over- or under-represented 1989 (Amselem et al. 1989, Godowski et al. 1989). Sixteen in LS-derived lymphoblastoid cells. Gene differential different molecular defects were found in the Israeli cohort, expression occurs in a number of gene families, including whereas the majority of LS patients in the Ecuadorian cohort cell cycle, metabolic control, cytokine–cytokine receptor described below are homozygous for an A to G splice site interaction, Jak-STAT signaling, and PI3K-AKT signaling. mutation at position 180 in exon 6 of the GH-R gene. This Major differences between LS and healthy controls finding reflects the fact that the Israeli cohort originated were also noticed in pathways associated with cell cycle from different areas and ethnic groups (Laron et al. 2011). distribution, apoptosis, and autophagy. Our analyses All of these molecular defects lead to GH insensitivity and, highlight the key role of the GH–IGF1 axis in cancer consequently, impaired IGF1 biosynthesis. development and progression. Epidemiologic studies carried out over the past 20 years have shown a correlation between circulating IGF1 levels and the risk to develop colon, breast, prostate, and other Materials and methods type of cancers (Chan et al. 1998, Hankinson et al. 1998, Lymphoblastoid cultures Renehan et al. 2004). However, the mechanisms by which IGF1 plays a role in cancer initiation and/or progression are Epstein–Barr virus (EBV)-immortalized lymphoblastoid not yet entirely clear. Recent studies have evaluated whether cell lines from LS patients and healthy controls were patients with congenital IGF1 deficiency have a reduced obtained from the National Laboratory for the Genetics risk of cancer (Shevah et al. 2007, Steuerman et al. 2011). of Israeli Populations, Tel Aviv University. At the time

of enrollment, all participants signed written informed methods. Functional analysis was performed using David consent. The research protocol followed the principles of and WebGestalt analysis platforms. Principal component the Declaration of Helsinki and was approved by the local analysis (PCA) was performed using Partek Genomics Suite Ethics Committees. Cells were maintained in RPMI-1640 v. 6.5 using all the genes in the array. medium supplemented with 10% fetal bovine serum, 2 mM glutamine, and 100 units/mL penicillin + 100 μg/mL Real-time quantitative polymerase chain reactions streptomycin. Reagents were purchased from Biological (RQ-PCRs) Industries Ltd (Bet-Haemek, Israel). Cultures were grown upright in tissue culture flasks and maintained under a Total RNA was prepared from lymphoblastoids using the

humidified 5% CO2 atmosphere at 37°C. Trizol Reagent. Single-stranded cDNA was synthesized from total RNA samples using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems). IGF1R, Genomic analyses cyclin A1, UGT2B17, serpinB2, OR5H2, versican, and RNA was isolated from lymphoblastoid cell lines using the AKT3 mRNA levels were measured by RQ-PCR, using Trizol Reagent (Life Technologies). RNA quality was assessed appropriate primers (Table 1). For control purpose, actin by the absorbance ratio at 260/280 nm and gel electrophoresis mRNA levels were measured. before further analyses. Affymetrix GeneChip Human Gene 1.0 ST Arrays (Affymetrix), which offer whole-transcript Western blot analyses coverage, were used. Arrays interrogate 28,869 well- annotated genes and noncoding RNAs with 764,885 probes. Cells were grown to confluence, centrifuged at 300 g Human Gene 1.0 ST Array has greater than 99% coverage of for 10 min, washed twice with phosphate-buffered sequences present in the RefSeq database. saline, and incubated with a lysis buffer for 20 min. The suspension was centrifuged at 16,200 g for 10 min, diluted with sample buffer, and boiled for 5 min. Samples were Bioinformatic analyses electrophoresed through 10% SDS–PAGE, followed by Data analysis was performed on CEL files using Partek blotting of the proteins onto nitrocellulose membranes. Genomics Suite (Partek, St Louis, MO, USA). Data were After blocking, the blots were incubated overnight with normalized and summarized with the robust multiaverage the indicated antibodies, washed, and incubated with a Endocrine-Related Cancer Endocrine-Related method followed by one-way analysis of variance (ANOVA) horseradish peroxidase-conjugated secondary antibody. (Irizarry et al. 2003). Genes of interest that were differentially expressed with P < 0.05 and fold-change difference cutoff Cell cycle analyses >|2| were obtained. Hierarchical cluster analysis was performed using Partek Genomics Suite software with Cells were grown to confluence, after which medium Pearson’s dissimilarity correlation and average linkage was changed to starvation medium and grown for an

Table 1 Sequences of primers employed in RQ-PCR analyses.

Primer Sequence Annealing (°C) Product size SERPINB2-F CACCCAGAACCTCTTCCTCTCC 67.3 134 SERPINB2-R TAACTGCATTGGCTCCCACTTC 67.2 OR5H2-F CACAGGACTTACATATCAGCCAG 60.2 93 OR5H2-R AGACCAAGGTTCCACACAATAGT 62.5 CyclinA1-F TCAGTACCTTAGGGAAGCTGAAA 62.8 103 CyclinA1-R CCAGTCCACCAGAATCGTG 64.4 Versican-F GCACCTGTGTGCCAGGATA 64.9 70 Versican-R CAGGGATTAGAGTGACATTCATCA 63.9 AKT3-F TTTGCAAAGAAGGGATCACA 59 103 AKT3-R ACTGCTCGGCCATAGTCATT 59 UGT2B17-F GCATATGATCTGAAGAAGTGGGA 64.1 99 UGT2B17-R AATGAGCCACATTTCAGCTTTC 64.2 Actin-F CCTGGCACCCAGCACAAT 67.4 144 Actin-R GGGCCGGACTCGTCATACT 66.3

additional 24 h. Cell suspensions were digested with 0.003% trypsin for 10 min. Cells were stained by adding propidium iodide (PI) in a spermine tetrahydrochloride solution. The samples were incubated on ice in the dark for an additional 10 min before analysis. Samples were analyzed using a FACSCalibur system (Becton Dickinson, Franklin Lakes, NJ, USA). Cell cycle analysis on histograms from PI-stained cells was performed using Flowing Software (http://www.flowingsoftware.com).

Apoptosis measurements

Basal levels of apoptosis were determined after cells were grown to confluence. Etoposide-induced level of apoptosis was determined 24 h after treatment. Cells were stained for 15 min with an Annexin V-FITC antibody and PI using an MEBCYTO apoptosis kit (MBL International, Woburn, MA, USA). Flow cytometry was performed with a FACSCalibur system and results were analyzed using the Flowing Software. Necrotic cells were stained with PI as well as Annexin V, whereas apoptotic cells were stained only with Annexin V.

Cell proliferation assays

Proliferation of lymphoblastoid cell lines was examined using an XTT–cell proliferation kit (Biological Industries Ltd, Israel). Cells were counted using a Nexcelom Bioscience Endocrine-Related Cancer Endocrine-Related Cellometer (Lawrence, MA, USA), diluted to a concentration of 250,000 cells/mL, and incubated in 96-well plates in Figure 1 a volume of 100 L for 3 days in RPMI-1640 medium Genome-wide profiling of LS patients. (A) Cluster analysis of differentially μ expressed genes in lymphoblastoids derived from four LS patients and supplemented with 4% of albumin-free serum replacement four controls of the same age range and ethnic origin. The figure depicts (Biogro-2, Biological Industries Ltd, Israel) in triplicate. On a cluster of 39 differentially expressed genes (FC > 2 or < −2 and the third day, activated XTT reagent in a volume of 50 L P-value < 0.05). Upregulated genes are shown in red and downregulated μ genes are shown in blue. FC, fold-change. (B) PCA display of four LS and was added to each well. After 5 h, absorption was measured four control arrays used in the experiment. Blue circles, LS patients; red using an ELISA reader (Bio-Rad microplate reader model circles, controls. PCA revealed a good discrimination between both 680; Bio-Rad Laboratories) at a wavelength of 450 nm and experimental groups. a reference absorbance of 655 nm. Oxidative damage response and autophagy measurements Table 2 Age and ethnic origin of patients and controls analyzed by genome-wide assays. Cells were grown to confluence, after which medium was changed to fresh full medium in the presence of increasing Laron syndrome Controls doses of paraquat dichloride (Sigma-Aldrich). Paraquat Ethnic Chip Ethnic is a quarternary nitrogen herbicide. The mechanism of Chip code Age origin code Age origin paraquat toxicity involves the generation of superoxide LS-4551 37 Iraqi C-1101 37 Iraqi LS-4490 42 Yemeni C-1328 62 Yemen anion, which leads to the formation of toxic reactive LS-5300 47 Iranian C-1419 49 Iranian oxygen species, such as hydrogen peroxide and hydroxyl LS-5859 51 Iranian C-1132 59 Iranian radical, and the oxidation of cellular NADPH. Proliferation Mean age 44.25 (±6.1) 51.75 (±11.3) in response to oxidative damage was measured using an LS, Laron syndrome; C, controls. XTT kit, as described above. Levels of p62 and LC3β as

markers of autophagy were measured by western blot expressed in LS compared with controls were identified analyses. Antibodies were purchased from Cell Signaling (with a P-value of <0.05 and fold-change difference Technology. cutoff >|2|). To summarize multivariate attributes by three dimensions that can be displayed graphically with minimal loss of information, PCA analysis was Results conducted. As shown in Fig. 1B, PCA revealed a good Genome-wide analysis of differentially expressed discrimination between experimental groups. The top 17 genes in LS up- and down-regulated genes (ranked by fold-change) are given in Table 3. In order to identify differentially expressed genes in LS patients in comparison with healthy controls, a Functional enrichment analysis genome-wide analysis was conducted. For this purpose, RNA obtained from EBV-immortalized lymphoblastoids Functional analyses were conducted to find co-expressed derived from four female LS patients and four controls genes sharing the same pathway. Analyses provide of the same age range (LS, 44.25 ± 6.08 years; controls, evidence for a number of shared pathways, including cell 51.75 ± 11.3 years; mean ± sd; P-value = 0.29) and same adhesion, G-protein signaling pathway, cell migration ethnic origin was used. Ages and ethnic origin of patients and motility, immune response, Jak-STAT signaling, and controls are summarized in Table 2. One-way ANOVA apoptosis, and metabolic pathways. For the most part, was performed using Partek Genomics Suite to create a genes involved in the control of cell cycle, motility, list of differentially expressed genes. A cluster analysis growth, and differentiation were downregulated in of differentially expressed genes is depicted in Fig. 1A. LS-derived lymphoblastoid cell lines compared with Thirty-nine annotated genes that were differentially controls (Table 4).

Table 3 Top 17 differentially expressed genes.

RefSeq Gene assignment Gene symbol P-value (LS vs C) Fold-change (LS vs C) Upregulated genes (n = 7) AK289419 UDP glucuronosyltransferase UGT2B15 0.0194396 11.0898 2 family, polypeptide B15

Endocrine-Related Cancer Endocrine-Related ENST00000551239 UDP glucuronosyltransferase UGT2B15 0.0384672 7.13829 2 family, polypeptide B15 U59209 UDP glucuronosyltransferase UGT2B17 0.0158576 7.16028 2 family, polypeptide B17 BC000181 G protein-coupled receptor 160 GPR160 0.0232108 3.7593 ENST00000371532 Zyg-11 homolog A (C. elegans) ZYG11A 0.0219404 3.03514 AK302847 Ribosomal modification protein RIMKLB 0.0378665 3.1161 rimK-like family member B BC152321 // Insulin-like growth factor 1 IGF1 0.044616 1.99144 (somatomedin C) Downregulated genes (n = 10) AK290029 Nephronectin NPNT 0.0147893 −3.12657 AK301897 Cyclin A1 CCNA1 0.0286311 −3.25083 ENST00000343200 Versican VCAN 0.0433405 −3.37884 ENST00000383695 Olfactory receptor, family 5, OR5K3 0.000944527 −4.04477 subfamily K, member 3 BC012609 Serpin peptidase inhibitor, clade SERPINB2 0.0239913 −4.35871 B (ovalbumin), member 2 ENST00000514084 Olfactory receptor, family 5, OR5H7P 0.000637513 −4.40605 subfamily H, member 7 BC136984 Olfactory receptor, family 5, OR5H14 0.0040872 −5.11991 subfamily H, member 14 ENST00000354924 Olfactory receptor, family 5, OR5K4 0.00212643 −5.22628 subfamily K, member 4 BC137523 // Olfactory receptor, family 5, OR5H6 0.00105229 −5.47561 subfamily H, member 6 ENST00000355273 // Olfactory receptor, family 5, OR5H2 0.00184389 −5.79858 subfamily H, member 2

Table 4 Functional analysis. List of 12 biological functions according to David and Webestalt raw data.

Pathway/function Example of genes P-value Cell adhesion UP in LS: DCBLD2, FLOT2, CNTNAP1 P = 0.006 Down in LS: THRA, SELL, ITGA5, NPNT, VCAN, NID1, CD24, SOX9, CCL5 EGF-like domain UP in LS: CNTNAP1 P = 0.043 Down in LS: SELL, NPNT, VCAN, NID1 G-protein-coupled receptor protein UP in LS: GPR160, FFAR2, FZD3, FZD6, CCR10, GPR114 P = 0.015 signaling pathway Down in LS: GNA15, OR5H1, OR5K4, OR5K3, OR5H2, OR5H14, OR5H6, PTGER4, CCL5 Cell migration and motility UP in LS: CCL22, FUT8 P = 0.004 Down in LS: WNT2, ITGA5, IL1B, VCAN, CD24, CCL5 Toll-like receptor signaling pathway UP in LS: P = 0.0002 Down in LS: CCL5, IL1B, TLR1, AKT3 Immune response UP in LS: CCL22, LAX1, IGJ, AICDA P = 0.015 Down in LS: PTGER4, IL7, CD274, TLR1, IL1B, CD24, CCL5 Regulation of cytokine production UP in LS: P = 0.032 Down in LS: IL6ST, TLR1, IL1B, CD24, HDAC7 Jak-STAT signaling pathway UP in LS: P = 0.001 Down in LS: IL7, SPRY4, AKT3, IL6ST Apoptosis UP in LS: CAPN2 P = 0.002 Down in LS: IL1B, AKT3 Metabolism of xenobiotics by UP in LS: UGT2B15, UGT2B17, UGT2B7 P = 0.001 cytochrome P450 Down in LS: Metabolic pathways UP in LS: UGT2B15, UGT2B17, FUT8 , UGT2B7, ALDH3A2, TST, FAH P = 1.69e-05 Down in LS: GBE1, P4HA2, ASS1, ME1, GALNT10, ST3GAL5 Pathways in cancer UP in LS: CCNE2, FZD3 P = 0.0003 Down in LS: AKT3, CCNA1, FZD6, WNT2

Dot-plot and RQ-PCR validation Log2 expression values are shown using dot-plot, which was performed using Partek GS. Dot-plots are shown in To validate the differences in specific gene expression Endocrine-Related Cancer Endocrine-Related Fig. 2 (insets). Levels of UGT2B17 mRNA, a member of the between LS patients and controls identified in gene arrays, UDP-glycosyltransferase (UDPGT) family, were markedly RQ-PCR was performed. Genes that display fold-change elevated in LS-derived cells in comparison with controls. difference higher than three and are potentially relevant The UDPGTs are of major importance in the conjugation in cancer biology were selected. Selected genes included: and subsequent elimination of potentially toxic xenobiotics UGT2B17 (UDP-glycosyltransferase 2B17), VCAN (versican), and endogenous compounds, including steroid hormones. OR5H2 (olfactory receptor 5H2), SERPINB2 (plasminogen However, mRNA levels of VCAN, OR5H2, SERPINB2, CCNA1, activator inhibitor), CCNA1 (cyclin A1), and AKT3. RQ-PCR and AKT3 were significantly lower in LS patients. validations corroborated genomic data for all six genes.

Figure 2 Dot plot and validation analyses of selected genes differentially expressed in LS. Selected genes that display fold-changes higher than three between LS patients and controls were selected for further validation using RQ-PCR. The selected genes include: UGT2B17 (UDP- glycosyltransferase-2B17), VCAN (versican), OR5H2 (olfactory receptor 5H2), SERPINB2 (plasminogen activator inhibitor), CCNA1 (cyclin A1), and AKT3. The insets denote the gene expression values (dot plots) in four individual patients and four controls as measured in gene arrays. Gray bars and dots, LS patients; black bars and dots, controls.

Analysis of transcription factors involved in IGF1R gene regulation

To gain further insight into specific genes whose deregulated expression in LS might impinge upon specific IGF axis components, we measured the basal levels of transcription factors SP1 and pTEN in four LS-derived lymphoblastoid lines and four normal cultures. The rationale for this choice of genes was the fact that transcription factor SP1 was identified as a critical regulator of the IGF1R gene. PTEN was selected because it is a well- characterized target for IGF1 action (Beitner-Johnson et al. 1995, Werner et al. 1996, Werner 2012). Western blot analysis revealed that LS-derived lymphoblastoids express higher levels of tumor suppressor pTEN than controls. However, levels of oncogenic protein SP1 were markedly reduced in LS cells (Fig. 3). Consistent with a reduced rate of cancer incidence, LS-derived cells contain abundant Figure 3 Western blotting analysis of SP1 and pTEN levels in LS-derived quantities of transcription factors associated with both lymphoblastoids. (A) Lymphoblastoid cell lines of four LS patients and IGF1R gene suppression and cell cycle inhibition (i.e. four controls were lysed as described in “Materials and methods” section pTEN). Conversely, LS cells express reduced levels of and extracts were electrophoresed through SDS–PAGE, followed by transfer and incubation with the indicated antibodies. (B) Scanning potentially oncogenic proteins linked to IGF1R gene densitometry analysis of SP1 and pTEN autoradiographs. Bars denote transcription and cell cycle progression (i.e. SP1). mean ± s.e.m. (n = 4). *Significantly different vs controls (P < 0.05). A value of 100% was given to protein values in controls.

Analysis of IGF1R downstream cellular mediators in LS

To examine the expression levels of cellular proteins downstream of IGF1R, we next measured the levels of Akt Endocrine-Related Cancer Endocrine-Related and ERK, two prototypical components of the PI3K-Akt and ras-raf-MAPK pathways, respectively. In accordance with a reduced expression and activation of the IGF1R in LS patients, levels of total and phospho-Akt were reduced in patients compared with controls (Fig. 4). Interestingly, and despite the fact that no major differences in total ERK expression levels were noticed between LS and controls, a significant reduction in ERK phosphorylation was observed in LS cells.

Cell cycle analyses

Following identification of differentially expressed genes involved in cell cycle regulation, we next investigated potential differences in cell cycle dynamics. To evaluate the hypothesis that protection from cancer in LS Figure 4 patients is associated with altered cell cycle dynamics, Western blot analysis of downstream mediators of IGF1 action. lymphoblastoids (n = 3 patients and 3 controls) were Lymphoblastoids of four LS patients and four controls were starved for exposed to etoposide (20 µM) for 24 h, followed by cell 24 h, after which they were lysed and electrophoresed through SDS– PAGE. Membranes were incubated with antibodies against total and cycle profiling (Fig. 5A). Etoposide is an anticancer agent phospho-IGF1R, AKT, and ERK. Tubulin levels were measured as a loading that inhibits DNA topoisomerase II. The majority of LS control.

cells (69%) were in G0–G1 phase, compared with 61% of control cells. The proportion of cells in G2–M phases in control cells was increased by 50% compared with LS (4.4% in controls compared with 2.2% in LS; P = 0.02), while the proportion of cells in S phase was increased by 120% compared with LS (12.3% in controls compared with 5.6% in LS; P = 0.003). Cell cycle analyses are consistent with reduced proliferation of LS-derived cells upon exposure to etoposide in comparison with controls. In agreement with cell cycle distribution data, cyclin A1 (Fig. 2) and cyclin D1 (not shown) levels were markedly reduced in LS in comparison with healthy lymphoblastoids. Notably, no differences in cell cycle distribution between control and LS cells were observed under basal, nonstimulated conditions (not shown).

Apoptosis measurements

To evaluate the sensitivity of LS cells to DNA damage in comparison with control cells, lymphoblastoids were exposed to etoposide (20 µM) for 24 h, followed by apoptosis analysis. Results of flow cytometry indicate that the percentage of apoptotic cells under basal conditions was 40% higher in LS compared with controls (P = 0.0005) (Fig. 5B). The percentage of necrotic cells under basal conditions was increased by 27% in LS compared with controls (P = 0.0002) (Fig. 5B). The percentages of apoptotic and necrotic cells were increased by 20 and Endocrine-Related Cancer Endocrine-Related 17%, respectively, after treatment with etoposide in LS in comparison with controls (not shown).

Proliferation analyses

To explore the effect of IGF1-free medium on proliferation rate in LS cells compared with controls, XTT assays Figure 5 were performed. Cells were grown in RPMI-1640 Cell cycle distribution in LS-derived cells. Lymphoblastoids were exposed to etoposide for 24 h, after which cells were stained with propidium medium supplemented with 4% of albumin-free serum iodide and analyzed with a FACScalibur system to determine the replacement for 3 days, after which proliferation tests were percentage of cells in G1, S, and G2/M phases. (A) The percentage of cells conducted. As shown in Fig. 5C, significant differences within each cell cycle phase (G1, S, and G2/M) is indicated. Results of a representative experiment repeated three times with similar results are in proliferation rates were seen between LS and controls. shown. *Significantly different vs control (P < 0.05). (B) Basal apoptosis Specifically, the proliferation rate of LS cells was reduced was measured by flow cytometry analysis after staining cells with an Annexin-V antibody and propidium iodide. Results of a representative by 50% compared with controls (P = 0.002). experiment repeated three times with similar results are shown. *Significantly different vs control (P < 0.05). (C) Cell proliferation of LS and control cells was measured using an XTT-Cell proliferation kit, as Oxidative damage response described in “Materials and methods” section. *Significantly different vs control (P < 0.05). Gray bars, LS; black bars, controls. To examine the hypothesis that cancer protection in LS might be associated with enhanced resistance to oxidative damage, lymphoblastoids were treated with assessed by XTT assays. Results obtained indicate that increasing doses of the oxidative agent paraquat for LS-derived lymphoblastoids displayed an enhanced different periods of time, after which cell survival was survivability compared with control cells over a broad

Evaluation of autophagy markers

Finally, to investigate the potential involvement of the autophagic machinery in the differential response to oxidative damage, LC3β (Fig. 7A and C) and P62 (Fig. 7B and D) levels were measured by western blots. LC3β and P62 constituted to be validated markers of autophagy, being usually expressed in a diametrically opposite manner (Mizushima et al. 2007, Moscat et al. 2009, Rosenfeldt et al. 2011). Results obtained indicate that, under basal conditions, LC3β levels were significantly higher in control lymphoblastoids than in LS-derived cells. However, basal levels of P62 were three-fold higher in LS cells. Following paraquat treatment for 24 h, P62 levels increased approximately two-fold in LS cells while only a marginal increase was seen in control cells.

Discussion

Laron syndrome is a genetic disorder caused by molecular defects (deletion or mutation) of the GH-R gene, or postreceptor pathways, leading to congenital IGF1 Figure 6 deficiency (Laron 2004). This type of dwarfism was first Response to oxidative damage in LS-derived cells. LS-derived (●) and control (◾) lymphoblastoids were exposed to increasing doses of the identified in 1958 in a Jewish family of Yemenite origin. oxidizing agent paraquat for 24 h, after which cell viability was measured Three family members had dwarfism, whereas other five by XTT assays. The experiment was conducted in serum-containing siblings had a normal stature. Studies reported in 1966 (A) and serum-free (B) media. The graph depicts a pair of LS-derived and control cells, normalized for age and ethnic origin. A value of 100% was established that this condition was associated with high given to the cell number at time 0. Results of a representative experiment serum GH values (Laron et al. 1966). The typical features Endocrine-Related Cancer Endocrine-Related repeated multiple times with similar results are shown. of classic LS are short stature, typical face, obesity, and low serum IGF1 levels, unresponsive to the administration of range of paraquat concentrations (0.01–10 mM) (Fig. 6). exogenous GH. Several molecular defects of the GH-R Of interest, differences in response to oxidative insults gene have been identified, depending on the ethnic group were seen in both the presence (A) and the absence (B) and geographic origin of the patients (Shevah et al. 2006, of serum. Laron et al. 2011).

Figure 7 Western blot analysis of autophagic markers in LS-derived cells. Paraquat-treated or untreated lymphoblastoids (n = 4 LS; n = 4 controls) were lysed and levels of autophagic markers LC3β (A) and P62 (B) were measured by western blots. (C and D) Scanning densitometry analysis of LC3β and P62 autoradiographs. No, no paraquat treatment. Gray bars, LS patients; black bars, controls. *Significantly different vs control (P < 0.01). **Significantly different vs control (P < 0.05).

The role of the GH–IGF1 axis in cancer biology has escape of circulating malignant cells. Immune deficiency been well established (LeRoith et al. 2003, Pollak 2012). has been reported in association with congenital IGF1 IGF1 has been identified as an important progression deficiencies (Green et al. 2011), suggesting that cancer factor during the cell cycle. The IGF1R, which mediates the protection in LS is most probably not related to improved biological actions of IGF1, exhibits potent antiapoptotic, immune surveillance but rather to a drastic reduction in cell survival activities and is usually expressed by most the primary events leading to cancer development. primary tumors and cancer-derived cell lines (Baserga et al. Our study identified a number of genes that are 2003, Werner 2009, Werner & Bruchim 2009). Importantly, differentially expressed in lymphoblastoid cells of LS the IGF1R emerged in recent years as a promising patients compared with healthy controls of the same therapeutic target in oncology, and specific antibodies as gender, age, and ethnic group. This differential expression well as small molecular weight tyrosine kinase inhibitors may, potentially, explain the evasion of LS patients from have been developed targeting tumors with IGF1R cancer. Bioinformatic analyses revealed that the differences overexpression. Controversy emerged, however, regarding in gene expression occur in several pathways, including the role of endocrine IGF1 as a cancer risk factor. While apoptosis, metabolic control, cytokine–cytokine receptor early studies suggested that high levels of circulating IGF1 interaction, Jak-STAT signaling, and PI3K-AKT signaling. Of were associated with an elevated risk of premenopausal relevance, our analyses detected markedly reduced levels of breast and prostate cancers (relative risk > 4), most recent gene transcripts associated with oncogenic transformation epidemiological studies are in agreement that the increased and cell cycle progression. These genes include, among risk conferred by high endocrine IGF1 is only ~1.5–2 others, cyclin A1, cyclin D1, serpin B2, versican, and (Renehan et al. 2004). Furthermore, taller people have zinc finger protein Sp1. Hence, data are consistent with higher plasma IGF1 values and being taller was associated the notion that life-long lack of exposure to endocrine with cancer risk in a cohort of 1,297,124 women followed IGF1 in LS might lead to downregulation of genes with a for a total of 11.7 million person-years (Green et al. 2011). positive impact on proliferation and mitogenesis. While The biochemical and cellular mechanisms associated with the mechanism/s responsible for IGF1 action has/have elevated tumor incidence in an IGF1-rich environment, not been addressed in the present study, it is reasonable however, have not yet been totally dissected. to assume that IGF1 exposure activates epigenetic and The finding that congenital IGF1 deficiencies and, transcription pathways critical for gene expression. Lack in particular, LS are associated with a reduced cancer of exposure to normal IGF1 levels in LS patients abrogates Endocrine-Related Cancer Endocrine-Related incidence offers the opportunity to study previously these signaling pathways, with important consequences unrecognized aspects of IGF1 biology in cancer etiology. in terms of cancer avoidance. In this context, we provide Notably, the epidemiological study by Steuerman and evidence that downregulation of cell cycle genes such as coworkers (2011) included, by most estimates, more cyclin D1 and cyclin A1 was associated with altered cell than 50% of all LS patients worldwide (not including cycle dynamics and apoptosis. Finally, and given the fact the Ecuadorian cohort, see next). None of the 230 LS that lymphoblastoid cultures are maintained in serum- patients included had developed a malignancy, despite containing media, data regarding IGF1 expression are the fact that 66 of them had been treated with IGF1 for sometimes difficult to interpret. Specifically, whileIGF1 several years. The difference between the prevalence of mRNA levels are very low in normal lymphoid cells of LS malignancies in all the first-degree relatives (4%), further patients, our data indicate a 1.99-fold change in LS-derived family members (23.7%), and siblings only (3.3%), with lymphoblastoids compared with healthy cells (P = 0.045) that in patients (0%), was highly significant. Similar (Table 3). We assume that these unexpected results are due protection from cancer was reported in an Ecuadorian to the fact that bovine serum contains high levels of IGF1 cohort of LS patients (Guevara-Aguirre et al. 2011). This or, alternatively, to the immortalized status of these cells. cohort has been followed since 1988 and mortality data Conversely, genomic analyses revealed enhanced were collected for 53 LS subjects who died before 1988. expression of genes associated with protection from toxic Cancer was not a cause of death among the individuals xenobiotic substances and metabolites in LS-derived who died before 1988 and there was no evidence of cancer lymphoblastoid cells. Specifically, theUGT genes, which among 99 LS patients since 1988. Among relatives, cancer code for enzymes that convert xenobiotic and endobiotic prevalence was similar to the general population. The substances into lipophilic compounds, thereby facilitating lack of cancer among LS patients deserves extra attention clearance from the body as part of a liver detoxification considering that tumor spread seems to reflect immune system, were several-fold higher in LS than in healthy

controls. Of special interest is the elevated expression of both crossing led to a significant decrease in prostate epithelial UGT2B15 and UGT2B17 (11- and 7-fold, respectively) in LS cell proliferation and an increase in basal apoptotic indices cells. For UGT2B17, the increased expression was validated (Wang et al. 2005). by RQ-PCR. The UGT enzymes encoded by both genes In summary, our analyses identified genes, pathways, facilitate the catabolism of steroid hormones with key roles and biological processes that are impaired in congenital in certain cancers, in particular breast and prostate tumors. IGF1-deficient patients and are associated with reduced These data points toward an enhanced capacity of LS cells cancer prevalence. Future studies will address the to respond to a variety of cellular insults. Consistent with transcriptional and epigenetic mechanisms that underlie these genomic findings, we demonstrate that the ability of the IGF1-dependent pathways associated with cancer LS cells to respond to oxidative damage was higher than protection at the organism level. that of control cells over a broad range of paraquat doses. Autophagy is a major housekeeping mechanism, crucially involved in the maintenance of normal cellular Declaration of interest homeostasis. The morphological hallmark of autophagy The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported. is the formation of double-membrane vesicles called autophagosomes, which sequester and transport integral regions of the cytoplasm to the lysosomal system. Funding This mechanism enables the clearance of damaged This research was supported by Grant #200937 from the United States– or superfluous proteins and organelles. The role of Israel Binational Science Foundation, Jerusalem, Israel, a grant from the autophagy, however, extends beyond the removal of Tel Aviv University Cancer Biology Research Center, and a grant from the Carol and Leonora Fingerhut Fund, Sackler School of Medicine. This work damaged cell components to many physiological and was performed in partial fulfillment of the requirements for a PhD degree pathological processes such as development, oxidative by Lena Lapkina at the Sackler Faculty of Medicine, Tel Aviv University. stress, and tumorigenesis. Proteins LC3β and P62 are involved in different aspects of autophagosome biology, and are regarded as valid autophagy markers. P62 is Author contribution statement specifically involved in the activation of the mTORC1 L Lapkina, Z Laron, and H Werner conceived of and designed the experiments. The experimental procedures were performed by pathway, which negatively regulates autophagy in L Lapkina, I Rotem, and R Sarfstein, and were analyzed by L Lapkina, response to nutrient availability. The finding that basal R Sarfstein, D Gurwitz, M Pasmanik-Chor, Z Laron, and H Werner. Endocrine-Related Cancer Endocrine-Related LC3β levels were reduced in LS cells while P62 values were L Lapkina and H Werner prepared the manuscript. M Pasmanik-Chor contributed analysis tools. elevated, in comparison with healthy controls, along with a major paraquat-induced increase in P62 levels in patient cells, suggests the existence of a differentially Acknowledgements regulated autophagy machinery in LS. These autophagic The authors wish to thank Leon Bach for critical reading of the manuscript. adaptations correlate with enhanced survival of LS cells in H W is the incumbent of the Lady Davis Chair in Biochemistry. response to oxidative stress. In the context of IGF1R gene regulation, we provide evidence that genes associated with transcriptional References activation (i.e. transcription factor SP1) were markedly Amichay K, Kidron D, Attias-Geva Z, Schayek H, Sarfstein R, Fishman A, reduced in LS cells (Beitner-Johnson et al. 1995). In addition Werner H & Bruchim I 2012 BRCA1 is expressed in uterine serous carcinoma (USC) and controls insulin-like growth factor-I receptor to its cardinal role in IGF1R gene activation, zinc finger (IGF-IR) gene expression in USC cell lines. International Journal of protein SP1 is a major player in cellular transformation and Gynecological Cancer 22 748–754. (doi:10.1097/ oncogenesis. We suggest that reduced levels of SP1 in LS IGC.0b013e318254011f) Amselem S, Duquesnoy P, Attree O, Novelli G, Bousnina S, Postel- cells may provide protection from cancer development via Vinay MC & Goossens M 1989 Laron dwarfism and mutations of the suppression of IGF1R activity. Cancer protection associated growth hormone-receptor gene. New England Journal of Medicine 321 with reduced circulating IGF1 was also noticed in animal 989–995. (doi:10.1056/NEJM198910123211501) Baserga R, Peruzzi F & Reiss K 2003 The IGF-1 receptor in cancer models of IGF1 deficiency. Thus, the ‘Laron mouse’ biology. International Journal of Cancer 107 873–877. (doi:10.1002/ (Gh-r/Gh-Bp KO) was shown to exhibit reduced estrogen- ijc.11487) independent mammary carcinogenesis when crossed with Beitner-Johnson D, Werner H, Roberts CT Jr & LeRoith D 1995 Regulation of insulin-like growth factor I receptor gene expression the C3(1)TAg mouse, which develops estrogen receptor α by Sp1: Physical and functional interactions of Sp1 at GC boxes and negative mammary cancers (Zhang et al. 2007). A similar at a CT element. Molecular Endocrinology 9 1147–1156.

Chan JM, Stampfer MJ, Giovannucci E, Gann PH, Ma J, Wilkinson P, LeRoith D & Yakar S 2007 Mechanisms of disease: metabolic effects of Hennekens CH & Pollak M 1998 Plasma insulin-like growth factor-I growth hormone and insulin-like growth factor-1. Nature Clinical and prostate cancer risk: a prospective study. Science 279 563–566. Practice in Endocrinology and Metabolism 3 302–310. (doi:10.1038/ (doi:10.1126/science.279.5350.563) ncpendmet0427) Cohen P, Rogol AD, Deal CL, Saenger P, Reiter EO, Ross JL, Chernausek SD, Mizushima N & Yoshimori T 2007 How to interpret LC3 Savage MO, Wit JM & Consensus Workshop participants 2008 immunoblotting. Autophagy 3 542–545. (doi:10.4161/auto.4600) Consensus statement on the diagnosis and treatment of children with Moscat J & Diaz-Meco MT 2009 p62 at the crossroads of autophagy, idiopathis short stature: a summary of the Growth Hormone Research apoptosis, and cancer. Cell 137 1001–1004. (doi:10.1016/j. Society, The Lawson Wilkins Pediatric Endocrine Society and the cell.2009.05.023) European Society for Pediatric Endocrinology Workshop. Journal of Pollak M 2008 Insulin and insulin-like growth factor signalling in Clinical Endocrinology and Metabolism 93 4210–4217. neoplasia. Nature Reviews Cancer 8 915–928. (doi:10.1038/nrc2536) Domene H, Bengolea SV, Martinez AS, Ropelato MG, Pennisi P, Scaglia P, Pollak M 2012 The insulin and insulin-like growth factor receptor family Heinrich J & Jasper H 2004 Deficiency of the circulating IGF system in neoplasia: an update. Nature Reviews Cancer 12 159–169. associated with inactivation of the acid-labile subunit gene. New Renehan AG, Zwahlen M, Minder C, O'Dwyer ST, Shalet SM & Egger M England Journal of Medicine 350 570–577. (doi:10.1056/NEJMoa013100) 2004 Insulin-like growth factor-I, IGF binding protein-3, and cancer Godowski PJ, Leung DW, Meacham LR, Galgani JP, Hellmiss R, Keret R, risk: systematic review and meta-regression analysis. Lancet 363 Rotwein PS, Parks JS, Laron Z & Wood WI 1989 Characterization of 1346–1353. the human growth hormone receptor gene and demonstration of a Rosenfeld RG 2005 The molecular basis of idiophatic short stature. partial gene deletion in two patients with Laron-type dwarfism. Growth Hormone and IGF Research 15 (Suppl A) S3–5. PNAS 86 8083–8087. (doi:10.1073/pnas.86.20.8083) Rosenfeldt MT & Ryan KM 2011 The multiple roles of autophagy in Green J, Cairns BJ, Casabonne D, Wright FL, Reeves G, Beral V & cancer. Carcinogenesis 32 955–963. (doi:10.1093/carcin/bgr031) Million women study collaborators 2011 Height and cancer Salmon WD & Daughaday WH 1957 A hormonally controlled serum incidence in the Million Women Study: prospective cohort, and factor which stimulates sulfate incorporation by cartilage in vitro. meta-analysis of prospective studies of height and total cancer risk. Journal of Laboratory and Clinical Medicine 49 825–836. Lancet Oncology 12 785–794. (doi:10.1016/S1470-2045(11)70154-1) Shevah O & Laron Z 2006 Genetic analysis of the pedigrees and molecular Guevara-Aguirre J, Balasubramanian P, Guevara-Aguirre M, Wei M, defects of the GH-receptor gene in the Israeli cohort of patients with Madia F, Cheng CW, Hwang D, Martin-Montalvo A, Saavedra J, Laron syndrome. Pediatric Endocrinology Reviews 3 (Suppl 3) Ingles S, et al. 2011 Growth hormone receptor deficiency is 489–497. associated with a major reduction in pro-aging signaling, cancer, and Shevah O & Laron Z 2007 Patients with congenital deficiency of IGF-I diabetes in humans. Science Translational Medicine 3 70ra13. seem protected from the development of malignancies: a Hankinson SE, Willett WC, Colditz GA, Hunter DJ, Michaud DS, preliminary report. Growth Hormone and IGF Research 17 54–57. Deroo B, Rosner B, Speizer FE & Pollak M 1998 Circulating (doi:10.1016/j.ghir.2006.10.007) concentrations of insulin-like growth factor-I and risk of breast Steuerman R, Shevah O & Laron Z 2011 Congenital IGF1 deficiency cancer. Lancet 351 1393–1396. (doi:10.1016/S0140-6736(97)10384-1) tends to confer protection against post-natal development of Irizarry RA, Hobbs B, Collin F, Beazer-Barclay YD, Antonellis KJ, malignancies. European Journal of Endocrinology 164 485–489. Scherf U & Speed TP 2003 Exploration, normalization, and (doi:10.1530/EJE-10-0859) summaries of high density oligonucleotide array probe level data. Wang Z, Prins GS, Coschigano KT, Kopchick JJ, Green JE, Ray VH, Endocrine-Related Cancer Endocrine-Related Biostatistics 4 249–264. (doi:10.1093/biostatistics/4.2.249) Hedayat S, Christov KT, Unterman TG & Swanson SM 2005 Klammt J, Pfaffle R, Werner H & Kiess W 2008 IGF signaling defects as Disruption of growth hormone signaling retards early stages of causes of growth failure and IUGR. Trends in Endocrinology and prostate carcinogenesis in the C3(1)/T antigen mouse. Endocrinology Metabolism 19 197–205. (doi:10.1016/j.tem.2008.03.003) 146 5188–5196. (doi:10.1210/en.2005-0607) Kofoed EM, Hwa V, Little B, Woods KA, Buckway CK, Tsubaki J, Pratt KL, Werner H 2009 The pathophysiological significance of IGF-I receptor Bezrodni L, Jasper H, Tepper A, et al. 2003 Growth hormone overexpression: new insights. Pediatric Endocrinology Reviews 7 2–5. insensitivity associated with a STAT5b mutation. New England Journal Werner H 2012 Tumor suppressors govern insulin-like growth factor of Medicine 349 1139–1147. (doi:10.1056/NEJMoa022926) signaling pathways: implications in metabolism and cancer. Oncogene Laron Z 1999 Natural history of the classical form of primary growth 31 2703–2714. (doi:10.1038/onc.2011.447) hormone resistance (Laron syndrome). Journal of Pediatric Werner H & Bruchim I 2009 The insulin-like growth factor-I receptor as Endocrinology and Metabolism 12 231–249. an onogene. Archives of Physiology and Biochemistry 115 58–71. Laron Z 2004 Extensive personal experience. Laron syndrome (primary Werner H & Bruchim I 2012 IGF-1 and BRCA1 signalling pathways in growth hormone resistance or insensitivity): the personal experience familial cancer. Lancet Oncology 13 e537–544. 1958–2003. Journal of Clinical Endocrinology and Metabolism 89 Werner H, Karnieli E, Rauscher FJ III & LeRoith D 1996 Wild type and 1031–1044. (doi:10.1210/jc.2003-031033) mutant p53 differentially regulate transcription of the insulin-like Laron Z & Kopchik JJ 2011 Laron Syndrome - From Man to Mouse: Springer- growth factor I receptor gene. PNAS 93 8318–8323. (doi:10.1073/ Verlag. pnas.93.16.8318) Laron Z, Pertzelan A & Mannheimer S 1966 Genetic pituitary dwarfism Woods KA, Camacho-Hubner C, Savage MO & Clark AJL 1996 Intrauterine with high serum concentration of growth hormone-a new inborn growth retardation and postnatal growth failure associated with error of metabolism? Israel Journal of Medical Sciences 2 152–155. deletion of the insulin-like growth factor I gene. New England Journal of LeRoith D, Bondy C, Yakar S, Liu J-L & Butler A 2001 The Somatomedin Medicine 335 1363–1367. (doi:10.1056/NEJM199610313351805) hypothesis: 2001. Endocrine Reviews 22 53–74. (doi:10.1210/ Zhang X, Mehta RG, Lantvit DD, Coschigano KT, Kopchick JJ, Green JE, edrv.22.1.0419) Hedayat S, Christov KT, Ray VH, Unterman TG, et al. 2007 LeRoith D & Roberts CT Jr 2003 The insulin-like growth factor system Inhibition of estrogen-independent mammary carcinogenesis by and cancer. Cancer Letters 195 127–137. (doi:10.1016/S0304- disruption of growth hormone signaling. Carcinogenesis 28 143–150. 3835(03)00159-9) (doi:10.1093/carcin/bgl138)

Received in final form 29 March 2016 Accepted 12 April 2016 Accepted Preprint published online 12 April 2016