J Gastroenterol DOI 10.1007/s00535-016-1181-5

ORIGINAL ARTICLE—ALIMENTARY TRACT

Carcinoma of the colon and rectum with deregulation of insulinlike growth factor 2 signaling: clinical and molecular implications Djeda Belharazem1 • Julia Magdeburg2 • Ann-Kristin Berton1 • Li Beissbarth3 Christian Sauer1 • Carsten Sticht4 • Alexander Marx1 • Ralf Hofheinz5 • Stefan Post2 • Peter Kienle2 • Philipp Stro¨bel3

•

Received: 14 September 2015 / Accepted: 2 February 2016 Ó Japanese Society of Gastroenterology 2016

Abstract Background Loss of imprinting (LOI) of the insulin-like growth factor 2 (IGF2) is an early event in the development of colorectal cancer (CRC). Whether LOI of IGF2 denotes a molecular or clinical cancer subgroup is currently unknown. Methods Tumor biopsies and paired normal mucosa from 399 patients with extensive clinical annotations were analyzed for LOI and IGF2 expression. LOI status in 140 informative cases was correlated with clinicopathologic parameters and outcome. Results LOI was frequent in normal mucosa and tumors and occurred throughout the large intestine. LOI was unrelated to microsatellite instability, KRAS mutation status, stage, and survival. However, CRC with LOI showed increased IGF2 protein levels and activation of AKT1. Gene expression analysis of tumors with and without LOI and knockdown of IGF2 in cell lines revealed that IGF2 induced distinct sets of activated and repressed genes, including Wnt5a, CEACAM6, IGF2BP3, KPN2A, BRCA2, and CDK1.

D. Belharazem, J. Magdeburg contributed equally. & Philipp Stro¨bel [email protected] 1

Institute of Pathology, University Medical Center Mannheim, University of Heidelberg, Heidelberg, Germany

2

Department of Surgery, University Medical Center Mannheim, University of Heidelberg, Heidelberg, Germany

3

Institute of Pathology, University Medical Center Go¨ttingen, Robert-Koch-Str 40, 37075 Go¨ttingen, Germany

4

Medical Research Center (ZMF), University Medical Center Mannheim, University of Heidelberg, Heidelberg, Germany

5

Medical Clinic, University Medical Center Mannheim, University of Heidelberg, Heidelberg, Germany

Inhibition of AKT1 in IGF2-stimulated cells showed that the downstream effects of IGF2 on cell proliferation and gene expression were strictly AKT1-dependent. Conclusions LOI of IGF2 is a frequent and early event in CRC that occurs both in the adenomatous polyposis coli (APC) gene-mutated and serrated route of carcinogenesis. LOI leads to overexpression of IGF2, activates IGF1R and AKT1, and is a powerful driver of cell proliferation. Moreover, our results suggest that IGF2 via AKT1 also contributes to non-canonical wnt signaling. Although LOI had no significant impact on major clinical parameters and outcome, its potential as a target for preventive and therapeutic interventions merits further investigation. Keywords AKT1

Colon cancer  IGF2  Imprinting  Wnt5a 

Introduction Loss of imprinting (LOI) of insulin-like growth factor-2 (IGF2) is an epigenetic change that occurs in a variety of cancers [1]. LOI refers to the aberrant bi-allelic expression of the maternal and paternal allele, when under normal conditions only one allele is expressed. LOI of IGF2 results in the overexpression of IGF2 and in pro-mitogenic and growth promoting signals through binding to the insulinlike growth factor receptor type 1 (IGF1R) [2]. Binding to the IGF1R activates insulin receptor substrates IRS-1 and IRS-2, phosphatidylinositol 3-kinase (PI3 K), and Akt [3]. Akt phosphorylates a wide range of target proteins including glycogen synthase kinase 3b (GSK3b), mouse double minute 2 (MDM2), and the mammalian target of rapamycin (mTOR) [4]. IGF2 overexpression through LOI is an important autocrine loop in cancer [5]. LOI of IGF2

123

123

55 61 F

79 M – 140 R

LOI loss of IGF2 imprinting, ROI retention of IGF2 imprinting, N? cases with regional lymph node metastases, M? cases with distant metastases, LRect tumor located in lower rectum (\6 cm), MRect tumor located in mid rectum (6–12 cm), URect tumor located in upper rectum ([12 cm), DC ? LFL tumor located in descending colon or left flexure, TRC ? RFL tumor located in transverse colon or right flexure, AC tumor located in ascending colon; (N ? cases) number of cases with regional lymph node metastases in a given tumor location

n = 12 (6) n = 17 (5) n = 9 (3) n = 4 (3) n = 27 (8) n = 16 (8) n = 34 (18)

T4: 1 (1) T4: 1 (1) T4: 0 T4: 0 T4: 0 T4: 0 T4: 0 T4: 0

n = 21 (4)

T3: 1 (1) T3: 2 (1) T3: 5 (2) T3: 2 (2) T3: 6 (2) T3: 3 (2) T3: 5 (2) T3: 4 (2)

24

87 Y

T2: 1 (0) T2: 4 (0) T2: 1 (0) T2: 0 T2: 2 (0) T2: 1 (0) T2: 3 (1)

17 19 F

25 M

45–89 (72) 44 ROI

52 N

T1: 0 T1: 1 (0) T1: 0 T1: 0 T1: 0 T1: 0 T1: 0

T2: 1 (0)

T3: 8 (4) T4: 0 T3: 5 (0) T4: 3 (3) T3: 2 (1) T4: 0 T3: 1 (1) T4: 0 T3: 12 (4) T4: 2 (2) T3: 7 (4) T4: 0 T3: 14 (8) T4: 3 (2) T3: 6 (2) T4: 2 (0)

T1: 0

T2: 1 (0) T2: 1 (0) T2: 1 (0) T2: 1 (0) T2: 5 (0) T2: 3 (0) T2: 8 (4)

13

28 Y

16 N

T1: 0 T1: 0 T1: 0 T1: 0 T1: 0 T1: 2 (2) T1: 1 (1)

T2: 6 (0)

T1: 2 (0) 59 Y

36 N

Cecum (N ? cases) AC (N ? cases) TRC ? RFL (N ? cases) DC ? LFL (N ? cases) Sigmoid (N ? cases) URect (N ? cases) MRect (N ? cases) LRect (N ? cases) Survivor (Y:N) M?

11 38 42 F

54 M

43–102 (70) 96

Genomic DNA was extracted from fresh frozen CRC and normal mucosal samples using Purigene Kit (Qiagen, Hilden, Germany) and genotyped for a single-nucleotide

LOI

Analysis of IGF2 Polymorphism and LOI status

N?

A pool of consecutive samples from n = 399 colorectal cancer patients (241 males, 158 females) with clinical follow-up and available fresh frozen material was used for this study. All samples were screened for a single-nucleotide polymorphism (SNP) (820 A/G, refSNP IDrs680) located in IGF2 exon 9. This screen resulted in n = 140 heterozygous (informative) cases that were used for all further analyses. Tumor distribution and stage as well as age and gender of these patients were statistically not different from the total screened cohort (details available upon request). The clinicopathologic details of the 140 patients are summarized in Table 1. The study was performed in accordance with the World Medical Association’s Declaration of Helsinki (1964 and its later amendments) and was approved by the local ethics committee (Medical Faculty Mannheim, University of Heidelberg, Ref. 2006-154 N-MA).

Gender

Patients and tissues samples

Age (median)

Materials and methods

n=

was first identified in embryonal tumors in children, such as Wilms tumor [6, 7] and rhabdomyosarcoma [8], but was later also found in many adult cancers of the ovary, colon [9], lung [10], bladder [11], and in chronic myelogenous leukemia [12]. LOI of IGF2 appears to occur early in carcinogenesis, and is frequently found in morphologically normal tissues in the vicinity of tumors. LOI of IGF2 has also been suggested to be a potential marker of colorectal cancer (CRC) risk [13]. The prognostic impact of LOI of IGF2 in CRC is controversial. Moreover, the consequences of IGF2 LOI on gene expression in human CRC have not been studied in detail. In this study, we analyzed the prognostic and molecular impact of IGF2 LOI in a large cohort of CRC patients with extensive clinical follow-up data. Our results show that LOI leads to increased IGF2 levels of phosphorylated AKT1, resulting in repression and activation of a specific set of genes including Wnt5a, an activator of non-canonical Wnt signaling. LOI tumors were observed throughout the large intestine, and in both microsatellite instable/KRAS wildtype and microsatellite stable/KRAS mutated tumors. Our results indicate that IGF2 is an important mitogenic growth factor in CRC. The utility of LOI as a target for preventive or therapeutic strategies merits further investigation.

Table 1 Clinicopathologic characteristics of 139 colorectal cancer patients heterozygous for IGF2 820 A/G

J Gastroenterol

J Gastroenterol

polymorphism (SNP) (820 A/G, refSNP IDrs680) located in IGF2 exon 9 using an allele specific PCR with primers IGF2rs680F 50 -GAATTGGCTGAGAAACAATTGGC-30 and IGF2rs680Rt 50 -CCACCTGTGATTTCTGGGGT-30 specific for the ‘‘A’’ allele and IGF2rs680Rc 50 CCACCTGTGATTTCTGGGGC-30 specific for the ‘‘G’’ allele. ß-Globin forward 50 -GGTTGGCCAATCTACTCCCAGG-30 and reverse 50 -GGTTGGCCAATCTACTCCCAGG-30 primers were used to control for DNA quality. The PCR was performed at 95 °C for 2 min followed with ten cycles (94 °C for 20 s, 65 °C for 1 min) and 35 cycles (94 °C for 20 s, 62 °C for 1 min and 72 °C for 30 s). For the analysis of IGF2 LOI status, RNA was isolated from the 820 A/G SNP heterozygous samples, treated with DNase (ThermoFisher Scientific, Schwerte, Germany), and reversely transcribed using RT kit (Thermofisher Scientific, Schwerte, Germany). The cDNA was analyzed using the allele-specific PCR described above with the exception that GAPDH RNA specific primers were used to control for RNA quality (GAPDH_F: 50 -CAAGGTCATCCATGA CAACTTTG -30 and GAPDH_R: 50 -GTCCACCACCCTG TTGCTGTAG-30 ).

(Krefeld, Germany) and performed according to the manufacturer’s instructions. Thirty lg proteins extracted from whole tumor and normal tissues were used for the analysis. All the samples were assayed in triplicates. Mean values were used for statistical calculations. Western blot analysis of protein lysates extracted from fresh-frozen tissues was peformed using standard protocols. A 20 lg lysate was loaded onto SDS-PAGE gels and transferred onto nitrocellulose membranes. The membranes were blocked for 1 h in blocking buffer (5 % nonfat dry milk in Tris-buffered saline/Tween 20) and probed using specific primary monoclonal antibodies against insulin-like growth factor 1 receptor (IGF1R), phospho-IGF1R (pIGF1R), and ß-actin (GeneTex, Eching, Germany). The membranes were washed with TBST and re-probed with appropriate secondary antibodies (New England Biolabs, Frankfurt, Germany). Membranes were subjected to enhanced chemiluminescence (Pierce, ThermoFisher Scientific, Schwerte, Germany) and exposed to autoradiography.

Determination of KRAS mutations and MSI status

Then 2 lg RNA was isolated from whole tissues using TRIzol reagent (Invitrogen, ThermoFisher Scientific, Schwerte, Germany). Total RNA was reverse-transcribed with RevertAidTM H Minus Reverse Transcriptase (Fermentas ThermoFisher, Schwerte, Germany) using the manufacturer’s protocol. Real time PCR was performed on the ABI STEP ONE PLUS TaqMan PCR System (Applied Biosystems, Schwerte, Germany) using FAST SYBR Green master mix (Applied Biosystems, Schwerte, Germany). The fold change in expression was calculated using the DDCt method with GAPDH as an internal control. Primer sequences are available upon request.

KRAS (exon 2) and PIK3CA (exons 9 and 20) mutational analyses were performed using 50 ng of genomic DNA and the following primers [14]: KRAS-F: 50 -AGGCCTGCTGAAAATGACTGAATA-30 , KRAS-R: 50 -TGTATCAAAGAATGGTCCTGCAC-30 . The obtained PCR products were then used for direct bidirectional sequencing using the same primers with a life technology BigDye Terminator V1.1 cycle sequencing kit (Applied Biosystems, Inc., Foster City CA, USA) and analyzed on a Genetic Analyzer 3130 (Applied Biosystems, Foster City, CA, USA). Microsatellite instability was assessed using MLH1 immunohistochemical staining as a surrogate marker [15] using standard protocols (Dako, Hamburg, Germany). Gene expression analysis RNA from three samples with retention of imprinting (ROI) and three samples with loss of imprinting (LOI) (KRASwt/ PIK3CAwt) was used for gene expression profiling using Human gene 1.0 ST Affymetrix gene chip arrays (Affymetrix, Regensburg, Germany) according to the manufacturer’s standard protocols (details available on request). Enzyme linked immunosorbent assay (ELISA) and Western Blotting ELISA assays to quantify IGF2 and phospho-AKT (pAkt) in protein lysates were purchased from Beckmann Coulter

Real time PCR analysis

IGF2 knockdown in cell cultures The human colon cancer cell lines HTC116, RKO and HT29 were cultured in suitable media (HT29 in RPMI, HTC116 in McCoy5A and RKO in MEM supplemented with 10 % of fetal bovine serum (FBS), 2 mM of LGlutamine and penicillin/streptomycin at 37 °C with 5 % CO2. Cells were transfected using two IGF2 siRNAs (FlexTube si03092152 and si04949441, Qiagen, Hilden, Germany) and scrambled siRNA. In brief, 105 cells were seeded in 6-well plates. Then 2 lM siRNA was mixed with 100 lL medium without serum or antibiotics and with 12 lL High-Perfect transfection reagent (Qiagen Hilden, Germany). The reaction was incubated at room temperature for 10 min and added to the cells. After 36 h the cells were trypsinized for RNA and protein extractions.

123

J Gastroenterol

Cell treatment and stimulation using IGF2 Cell lines were seeded in 6-well plates and incubated with 200 ng/mL human IGF2 (R&D Systems, Minneapolis, USA) with or without 50 lM of the Akt inhibitor triciribine for 24 h. Untreated cells were used as control. Cell proliferation tests and cell cycle analysis Cell proliferation was measured using the MTT (3-(4, 5-Dimethyl-2-thiazolyl)-2, 5-diphenyl-2H-tetrazolium bromide) assay. Colon cell lines (RKO, HT29, HTC116, and colo320) were collected at a final concentration of 2 9 104 cells/mL in fresh medium with 10 % FBS. Aliquots of 100 ll cell suspension were plated in 96-well tissue culture plates. Cells were treated either with different concentrations (100–500 nM) of different si-IGF2 or scramble siRNA for 48 h or with 50 lM triciribine plus 50–200 ng/mL IGF2. After 48 h incubation, 20lL of a 5 mg/mL MTT solution was added to each well, and the plate was incubated for 4 h, allowing the viable cells to reduce the yellow MTT to darkblue formazan crystals, which were dissolved in 100 lL DMSO. The absorbance in individual wells was determined at 570 nm using a microplate reader (Molecular Devices, Biberach, Germany). Absorbance was used to calculate the number of viable cells for graphic plots. For cell cycle measurements, 2 9 106 RKO, HT29, HTC116, and colo320 cells were stimulated with 200 ng/ mL IGF2 after treatment with 50 lM triciribine for 24 h. Cells were trypsinized and fixed with 70 % ethanol. After overnight incubation at 4 °C, cells were washed with PBS prior to staining with propidium iodide (PI) (25 lg/mL PI; 0.1 % triton X; 50 lg/mL RNase in PBS pH7.2). Cells were analyzed using a guava easyCyteTM Flow Cytometer (Merck Millipore, Darmstadt, Germany). Statistical analyses All statistical tests were performed with GraphPad Prism V6.0 (GraphPad Software, San Diego, CA, USA). A log rank test was used to analyze correlations between single parameters and survival. Cox regression was used for multivariate analysis. A p value \ 0.05 was considered significant. For comparison of differentially expressed genes in ROI and LOI tumors and for the validation of gene responses in the IGF2 knockdown cell lines, two-tailed student’s t-test was used with a \ 0.05 and a confidence level of 95 %. One-way ANOVA test was used in the gene expression analysis of IGF2 and triciribine treatments. A subsequent F test was used to compare variances with a \ 0.05 at a confidence level of 95 % (p \ 0.05 was considered significant). For the analysis of gene expression profiles, Gene Set Enrichment Analysis (GSEA) was used [16]. Genes were

123

ranked based on the correlation between their expression and the class distinction by using any suitable metric.

Results Frequency and clinical correlations of IGF2 820G/A alleles in the DNA from tumor samples and normal mucosa In a first round, DNA samples were screened for IGF2 820G/A heterozygous cases. Of 399 tumors, n = 152 (38.1 %) were heterozygous for IGF2 820G/A, n = 199 (49.9 %) were homozygous for G/G, n = 48 (12.0 %) were homozygous for A/A. Of the 399 paired normal samples, n = 163 (40.9 %) were heterozygous, 190 (47.6 %) were homozygous for G/G, and n = 46 (11.5 %) were homozygous for A/A. The differences between tumor samples and paired normal mucosa were not statistically significant (Chi-square, p [ 0.5). Discordant 820G/A genotype between tumor and paired normal mucosa (i.e., loss of heterozygosity) was observed in 100 of the 399 cases (25.1 %). In contrast to a previous study describing an association between the 820G/A genotype and age at tumor diagnosis in prostate cancer patients [17], no statistically significant association with any tumor- or patient-related parameters such as stage, survival, age, gender, or bodymass index (BMI) was found in colon cancer patients (not shown). LOI of IGF2 has no statistical impact on survival RNA samples from 140 informative IGF2 820G/A heterozygous cases were studied for retention (ROI) or loss of imprinting (LOI). ROI was detected in n = 44 (31 %) and LOI in 96 (69 %) cases. LOI cases were detected throughout the entire large intestine without anatomic predilection sites. Among 96 paired tumor-normal mucosa samples, discrepancy between tumor and normal mucosa was seen in 37 cases (39 %). The majority (n = 24) of these cases showed ROI in normal mucosa and LOI in the tumor samples, but n = 13 cases also showed LOI in the normal mucosa and ROI in tumors. On univariate analysis, the following parameters showed a strong statistical correlation with survival (log rank test): age (p = 0.0002), increased CEA levels (p \ 0.0001), T-stage (p \ 0.0001), N-stage (p \ 0.0001), M-stage (p \ 0.0001), resection status (p \ 0.0001), and tumor grade (p \ 0.0001) (Fig. 1). Among the parameters without statistical significance were gender, KRAS and PIK3CA gene mutation status, microsatellite instability, IGF2 820G/A genotype, IGF2

J Gastroenterol

Fig. 1 Survival curves of patients with colorectal cancer. Survival curves of the 140 patients in this study stratified according to a tumor stage (T1-T4), b lymph node status (N0-N2), c metastasis (M0 vs.

M?), d tumor grade (G2 vs. G3), e KRAS mutation status (wildtype, WT, vs. mutated, mut), and f IGF2 imprinting status (loss of imprinting, LOI, vs. retention of imprinting, ROI)

gene imprinting status, and BMI. On multivariate analysis, only M-status (HR 2.425), N-status (HR 2.103), high CEA (HR 1.906), and age (HR 1.037) had a statistically significant impact on survival (Cox regression, p = 0.0001).

LOI of IGF2 does not correlate with KRAS mutation status and microsatellite instability Results of mutational analyses and MSI testings in cases with and without LOI of IGF2 are summarized in Table 2.

123

J Gastroenterol Table 2 Correlation between IGF2 imprinting status, KRAS mutation and MSI status All cases LOI (n =)

Cases with IGF2 LOI

ROI (n =)

KRASWT

KRASmut

MSS

MSI

KRASWT

KRASmut

Cases with IGF2 ROI MSS

MSI

KRASWT

KRASmut

MSS

MSI

LRect

16

5

5

1

4

1

3

1

2

1

2

0

2

MRect

26

8

15

3

11

1

9

2

6

0

6

1

5

0 1

URect Sigmoid

12 19

4 8

5 10

1 0

5 5

3 1

3 5

1 0

4 4

1 0

2 5

0 0

1 1

2 1

DC ? LFL

2

2

1

1

0

0

1

0

0

0

0

1

0

0 0

TRC ? RFL

3

6

6

0

5

0

1

0

2

0

5

0

3

AC

9

8

8

0

6

0

5

0

3

0

3

0

3

0

Cecum

9

3

10

0

9

1

8

0

7

1

2

0

2

0

96

44

60

6

45

7

35

4

28

3

25

2

17

4

Total

LOI loss of IGF2 imprinting, ROI retention of IGF2 imprinting, KRASWT KRAS wild type status, KRASmut KRAS mutated, MSS microsatellite stable, MSI microsatellite instable, LRect tumor located in lower rectum (\6 cm), MRect tumor located in mid rectum (6–12 cm), URect tumor located in upper rectum ([12 cm), DC ? LFL tumor located in descending colon or left flexure, TRC ? RFL tumor located in transverse colon or right flexure, AC tumor located in ascending colon

Sixty-six cases (39 with LOI, 27 with ROI) of different locations were randomly screened for KRAS mutations and revealed six mutations (details available upon request). There was no statistically significant correlation between IGF2 imprinting status and KRAS mutation status on the Chi-Square test (p = 0.3). Fifty-two cases (31 with LOI, 21 with ROI) of randomly screened tumors at different locations revealed seven cases with MSI. There was no statistically significant correlation between LOI and MSI status on the Chi-Square test (p = 0.5). IGF2, pIGF1R, and pAKT1 are increased in colon carcinoma samples with LOI IGF2 ELISA analysis in 13 LOI and 14 ROI tumor cases revealed significant IGF2 overexpression in tumors with LOI (20649 ± 8463 ng IGF2 in tumors with LOI vs. 11066 ± 4444 ng IGF2 in tumors with ROI, p = 0.0134). In normal tissue samples, IGF2 in LOI cases was only slightly increased (2034 ± 1384 ng IGF2 in LOI vs. 1667 ± 1125 in ROI, p [ 0.05) (Fig. 2b). ELISA in 13 LOI and 14 in ROI tumor cases showed significantly higher activity of pAKT in LOI than in ROI tumors (LOI cases: 36400 U/ng ± 28568 vs. ROI cases: 12145 ± 8052U/ng, p = 0.0078) (Fig. 2c). Western Blot analysis showed stronger phosphorylation of IGF1R in tumors with LOI (Fig. 2d). LOI and ROI tumors differ in gene expression To establish whether tumors with ROI and LOI differ in their gene expression, we compared three tumors with ROI and three tumors with LOI by Affymetrix gene expression arrays. To minimize potential confounders, all selected

123

cases were located in the sigmoid colon and were all UICC stage pT2/3 N0 M0. Fifty-five genes were found significantly differentially expressed (Fig. 3a). Fifteen genes were downregulated, 40 genes were upregulated in LOI vs. ROI tumors. Many of the upregulated genes (e.g., CDK1) in LOI tumors were related to cell cycle progression and proliferation (Table 3), but included also other genes such as WNT5a, CEACAM6, KPN2A, BRCA2, and IGF2BP3. Downregulated genes included RPSAY1, DDX3Y, UTY, and KDM5D, which are mostly responsible for ribosomal proteins or histone demethylation. Another downregulated gene, EGR1, can function both as a tumor suppressor and tumor promoter [18, 19]. To validate a subset of the differentially expressed genes in a larger cohort, quantitative RT-PCR was used in the initial six cases of the screening set and in 11 additional cases with ROI and 12 cases with LOI and confirmed the gene expression results (Fig. 3b–i). Knockdown of IGF2 affects expression of LOIdependent genes and cell proliferation in colon carcinoma cell lines In order to show that IGF2 indeed regulates the differentially expressed genes described above, IGF2 was suppressed using two different siRNAs in the CRC cell lines RKO, HT29, and HTC116 (Fig. 4). Scrambled siRNA was used as a control. IGF2 knockdown resulted in a highly significant downregulation of WNT5A, CEACAM6, and CDK1 in all three cell lines, and a significant downregulation of KPN2A, IGF2BP3, and BRCA2 in two and one cell lines, respectively. The ROI-associated gene EGR1 showed downregulation in RKO cells and upregulation in

J Gastroenterol Fig. 2 LOI of IGF2 leads to increased protein levels of IGF2 and activation of IGF1R and AKT1. Enzyme-linked immunosorbent assay (ELISA) and Western Blot results in tumor samples and paired normal mucosa with (n = 13) and without (n = 14) loss of imprinting of IGF2 (LOI vs. ROI). ELISA showed a significantly increased IGF2 and c phospho-AKT1 levels in tumors with LOI (**p \ 0.01). In contrast, IGF2 was only slightly increased in normal mucosa samples with LOI (b). Western Blot of four samples LOI and four samples with ROI showed increased phosphorylation of IGF1R in tumors with LOI (d)

HT29 and HTC116 cells. Vice versa, stimulation of untreated cell lines with IGF2 resulted in upregulation of the same genes (Fig. 5).

Inhibition of IGF2 and AKT1 abrogates cell proliferation and cell cycle progression in colon cancer cell lines

The downstream effects of IGF2 on LOI-dependent genes are mediated through AKT1 signaling

Knockdown of IGF2 using two different specific siRNAs resulted in a tenfold reduction of cell proliferation in all tested cell lines (mean cell number siRNA treated cells (500 nM) 0.25 9 105 ± 0.04 and 0.16 9 105 ± 0.06 vs. control 2.41 9 105 ± 0.04, p \ 0.05). Very similar effects were observed when cells were treated with the AKT-inhibitor triciribine (mean cell number triciribine treated cells (200 ng/ml) 1.02 9 105 ± 0.16 vs. control 7.2 9 105 ± 1.0, p \ 0.05) (Fig. 6).

IGF2 is known to exert most of its downstream effects via activation of AKT1 [3]. To find out whether all of the identified changes of LOI-dependent genes were due to the increased levels of AKT1 in tumors with LOI, CRC cell lines were stimulated with IGF2 (100 ng/ml) either alone or in combination with the AKT-inhibitor triciribine (50 lM) 24 h prior to IGF2 treatment. Cells were then harvested and expression of WNT5a, CEACAM6, CDK1, KPN2A, and IGF2BP3 mRNAs was measured using real time PCR (Fig. 5). IGF2-mediated upregulation of the studied genes was significantly reduced by inhibition of AKT1 through triciribine. Although not all genes responded uniformly in all cell lines, we conclude that AKT1 plays a major role in mediating the downstream effects of IGF2 on the target genes analyzed here.

Discussion In this study, we analyzed the prognostic impact of loss of imprinting (LOI) of IGF2 in a large cohort of clinically well characterized carcinomas of the colon and rectum (CRC) and found LOI in a high percentage of normal mucosa and paired tumor specimens, confirming the previous notion that

123

J Gastroenterol

Fig. 3 Colon cancer samples with LOI show a specific set of activated and repressed genes. a Gene expression profile of significantly altered genes in three human colon cancer samples with lost (LOI) vs. three samples with retained IGF2 imprinting (ROI). Results

were confirmed by qPCR in 15 tumor samples with LOI and 14 samples with ROI for Wnt5a (b), IGF2BP3 (c), CDK1 (d), CEACAM6 (e), KPNA2 (f), BRCA2 (g), ECT2 (h), and EGR1 (i) (**significant with p \ 0.01, ***significant with p \ 0.0001)

LOI probably represents an early field effect [20], which predisposes to the development of adenomas and carcinomas [21]. Although it is accepted that biallelic expression of IGF2 significantly raises the risk to develop CRC, it is currently unclear whether LOI is a constitutive or even heritable biomarker for colon cancer predisposition [21, 22], or rather an age-related acquired epigenetic event [23]. Depending on the method used, the frequency of LOI varies between 27 and 68 % [5, 20, 24–27]. Considerable interest is linked to the question whether LOI segregates with particular CRC molecular subtypes. Tumors with LOI of IGF2 have been described to occur in the proximal colon [25, 26] and to be particularly frequent among sporadic

microsatellite-instable cancers [20, 28] with an CpG island methylator phenotype (CIMP) that develop along the serrated molecular route [29–32]. In our study, LOI tumors occurred throughout the large intestine without predilection for the right colon, and were unrelated to MSI or KRAS mutation status (as approximative surrogate markers for the serrated vs. non-serrated molecular route of carcinogenesis). Moreover, similar to others [25], we did not observe a statistical correlation between LOI and gender, age, and TNM tumor stage. In contrast to a previous report [33], LOI in our study was not associated with survival, possibly due to technical limitations of our SNP-based approach, which results in a high percentage of uninformative cases.

123

J Gastroenterol Table 3 Differentially regulated pathways in tumors with LOI compared to ROI (GSEA technique)

GENE SET

ES

NES

NP

CENTROSOME

0.5913

2.009

\0.001

ORGANELLE_MEMBRANE

0.4809

2.018

\0.001

RNA_PROCESSING

0.5199

2.0191

\0.001

RIBONUCLEOPROTEIN_COMPLEX

0.5266

2.0254

\0.001

LIGASE_ACTIVITY

0.5454

2.0266

\0.001

ORGANELLE_LOCALIZATION

0.7145

2.0284

\0.001

ENDOMEMBRANE_SYSTEM

0.4997

2.0355

\0.001

DNA_DEPENDENT_ATPASE_ACTIVITY

0.7129

2.0462

\0.001

SPINDLE_POLE

0.7765

2.05

\0.001

CELL_CYCLE_CHECKPOINT_GO_0000075

0.6273

2.0665

\0.001

REGULATION_OF_MITOSIS

0.6504

2.0703

\0.001

PROTEASOME_COMPLEX

0.746

2.0894

\0.001

ORGANELLE_PART

0.4574

2.104

\0.001

DNA_METABOLIC_PROCESS

0.503

2.1051

\0.001

MEMBRANE_ENCLOSED_LUMEN INTRACELLULAR_ORGANELLE_PART

0.4825 0.4607

2.1056 2.1126

\0.001 \0.001

ORGANELLE_LUMEN

0.4825

2.1172

\0.001

MITOTIC_SISTER_CHROMATID_SEGREGATION

0.821

2.1331

\0.001

NUCLEAR_PART

0.4874

2.1615

\0.001

KINETOCHORE

0.74

2.165

\0.001

SISTER_CHROMATID_SEGREGATION

0.8168

2.1785

\0.001

CHROMOSOME_SEGREGATION

0.7178

2.1979

\0.001

DNA_DEPENDENT_DNA_REPLICATION

0.6503

2.212

\0.001

CHROMOSOMEPERICENTRIC_REGION

0.7344

2.2171

\0.001

DNA_REPLICATION

0.5927

2.2175

\0.001

CELL_CYCLE_GO_0007049

0.5276

2.2286

\0.001

SPINDLE

0.7283

2.2975

\0.001

CONDENSED_CHROMOSOME

0.7563

2.3098

\0.001

CHROMOSOMAL_PART

0.6198

2.3106

\0.001

CHROMOSOME

0.6237

2.3782

\0.001

CELL_CYCLE_PHASE CELL_CYCLE_PROCESS

0.602 0.6093

2.3902 2.4664

\0.001 \0.001

M_PHASE

0.6432

2.4712

\0.001

MITOSIS

0.6804

2.4773

\0.001

M_PHASE_OF_MITOTIC_CELL_CYCLE

0.6851

2.4904

\0.001

MITOTIC_CELL_CYCLE

0.6348

2.4922

\0.001

ES enrichment score, NES normalized enrichment scores, NP normalized p value

LOI tumors had significantly higher protein levels of IGF2 and showed activation of IGF1R and AKT1. Of note, LOI is not the only event that leads to increased IGF2 levels in CRC. It was recently shown that another source of IGF2 is amplification of the IGF2 gene locus at chr.11p15.5, together with an intronic microRNA miR-483 in about 7 % of APC mutated, KRAS/TP53/SMAD4 wild type CRC cases [34]. IGF2 and miR-483 are thought to cooperatively promote cell proliferation and the development of dysplasia [35]. Increased IGF2 has been shown to alter the maturation of colonic epithelial cells in mice and

humans [36, 37], and LOI of IGF2 has been shown to upregulate IGF1R and INSR, thereby enhancing IGF2 effects and leading to ‘‘IGF2 addiction’’ [38]. IGF2 has recently been shown to mediate anchorageindependent growth induced by the stem cell factor ZFP57 [39]. In line with previous observations in mice [38], we here show that LOI of IGF2 in human CRC leads to specific alterations in the expression of a defined subset of genes most of which were related to cell cycle and proliferation. Most if not all of the downstream effects of IGF2 on expression of the tested genes were dependent on

123

J Gastroenterol

123

J Gastroenterol b Fig. 4 Effect of IGF2 knockdown on the expression of LOI target

genes. Knockdown of IGF2 with two specific siRNAs [si-IGF2 (1) and si-IGF2 (2)] resulted in downregulation of the genes overexpressed in colon cancer samples with LOI, with some differences between the different cell lines (e.g., downregulation of BRCA2 in RKO, upregulation in HT29 and HTC116 cells). Similarly, EGR1, a gene that was downregulated in tumors with LOI, was strongly repressed in RKO and showed a trend towards upregulation in HT29 and HTC116 cells). (**significant with p \ 0.01, ***significant with p \ 0.0001)

Fig. 5 Effect of AKT1 inhibition on the expression of LOI target genes. CRC cell lines RKO (a), HT29 (b), HTC116 (c), and colo320 (d) were stimulated with IGF2 for 24 h with or without addition of 50 lM of the phospho-AKT inhibitor triciribine. Expression of IGF2 target genes Wnt5a, CEACAM6, CDK1, KPN2A, and IGF2BP3 was

AKT1. Suppression of IGF2 or inhibition of AKT1 in colon cancer cell lines led to a significant reduction in cell proliferation. Among the notable upregulated genes in CRC with LOI of IGF2 were Wnt5a and BRCA2. Wnt5a is particularly interesting, since our results strongly indicate that IGF2 and wnt signaling cooperate in CRC. It has been previously suggested that the promoting effect of IGF2 on the

measured using qPCR. The analysis showed that upregulation of all genes was abrogated by addition of triciribine, indicating that the effects of IGF2 on these genes rely on signaling through AKT1 (**significant with p \ 0.01, ***significant with p \ 0.0001)

123

J Gastroenterol

Fig. 6 Knockdown of IGF2 or inhibition of AKT1 lead to reduced cell proliferation and cell cycle progression. a Inhibition of IGF2 through two siRNAs [solid line: si-IGF2 (1), dotted line: si-IGF2 (2)] and b inhibition of AKT1 through 50 lM triciribine resulted in a significantly reduced proliferation in all tested cell lines (red: RKO,

green: col320, blue: HT29, purple: HCT116; p \ 0.05). c cell cycle analysis (propidium iodide) of cells treated with triciribine showed markedly reduced cell numbers in G0/G1 and G2/M phase and increase of presumably apoptotic cells

development and progression of APC-mutated intestinal adenomas and carcinomas [36, 37, 40] is mediated through interactions with the wnt-signaling pathway [37]. Earlier reports showed IGF2 induced translocation of beta-catenin to the nucleus and activated transcription of beta-catenin/ TCF4 complex target genes [41], although a study comparing LOI vs. ROI intestinal crypts in mice did not find significant differences of WNT signalling pathways [2]. Wnt5a is an activator of the non-canonical wnt pathway

that has previously been shown to drive EMT and metastasis [42], although tumor-supressive functions have also been described [43]. Further studies will be needed to confirm that IGF2 and Wnt5a cooperate in the progression of human CRC. Upregulation of BRCA2 was also noteworthy, since both IGF2 and BRCA belong to a large group of genes that are regulated through CTCF, an important multivalent transcription factor and chromatin insulator [44], that also regulates cell proliferation.

123

J Gastroenterol

In summary, our results indicate that LOI of IGF2 can occur independently of major pathogenetic CRC routes (serrated vs. non-serrated) and may constitute a distinct molecular subgroup that is characterized by increased IGF2 protein levels, activation of AKT1, and a specific set of downstream activated and repressed genes including Wnt5a. Although LOI was not associated with outcome in our study, it is conceivable that activation of the IGF2/AKT1 axis could lead to therapy resistance. LOI has already been successfully used as a target for cancer gene therapy in xenograft mouse models [45]. Direct targeting of LOI in preneoplastic and cancerous lesions could be an attractive means for cancer prevention or therapeutic interventions in CRC patients. Acknowledgments The authors gratefully acknowledge expert help with statistical analyses by Prof. Christel Weiß, Dept. of Medical Statistics, Medical Faculty Mannheim, University of Heidelberg. Compliance with ethical standards Conflict of interest of interest.

The authors declare that they have no conflict

References 1. Pfeifer K. Mechanisms of genomic imprinting. Am J Hum Genet. 2000;67(4):777–87. 2. Kaneda A, Feinberg AP. Loss of imprinting of IGF2: a common epigenetic modifier of intestinal tumor risk. Cancer Res. 2005;65(24):11236–40. 3. Foulstone E, Prince S, Zaccheo O, et al. Insulin-like growth factor ligands, receptors, and binding proteins in cancer. J Pathol. 2005;205(2):145–53. 4. Vivanco I, Sawyers CL. The phosphatidylinositol 3-Kinase AKT pathway in human cancer. Nat Rev Cancer. 2002;2(7):489–501 Epub 2002/07/03. 5. Cui H, Onyango P, Brandenburg S, et al. Loss of imprinting in colorectal cancer linked to hypomethylation of H19 and IGF2. Cancer Res. 2002;62(22):6442–6. 6. Rainier S, Johnson LA, Dobry CJ, et al. Relaxation of imprinted genes in human cancer. Nature. 1993;362(6422):747–9. 7. Ogawa O, Eccles MR, Szeto J, et al. Relaxation of insulin-like growth factor II gene imprinting implicated in Wilms’ tumour. Nature. 1993;362(6422):749–51. 8. Zhan S, Shapiro DN, Helman LJ. Activation of an imprinted allele of the insulin-like growth factor II gene implicated in rhabdomyosarcoma. J Clin Investig. 1994;94(1):445–8. 9. Cui H, Horon IL, Ohlsson R, et al. Loss of imprinting in normal tissue of colorectal cancer patients with microsatellite instability. Nat Med. 1998;4(11):1276–80. 10. Zhan S, Shapiro D, Zhan S, et al. Concordant loss of imprinting of the human insulin-like growth factor II gene promoters in cancer. J Biol Chem. 1995;270(47):27983–6. 11. Elkin M, Shevelev A, Schulze E, et al. The expression of the imprinted H19 and IGF-2 genes in human bladder carcinoma. FEBS Lett. 1995;374(1):57–61. 12. Randhawa GS, Cui H, Barletta JA, et al. Loss of imprinting in disease progression in chronic myelogenous leukemia. Blood. 1998;91(9):3144–7.

13. Cui H, Cruz-Correa M, Giardiello FM, et al. Loss of IGF2 imprinting: a potential marker of colorectal cancer risk. Science. 2003;299(5613):1753–5. 14. Nosho K, Kawasaki T, Ohnishi M, et al. PIK3CA mutation in colorectal cancer: relationship with genetic and epigenetic alterations. Neoplasia. 2008;10(6):534–41. 15. Lindor NM, Burgart LJ, Leontovich O, et al. Immunohistochemistry versus microsatellite instability testing in phenotyping colorectal tumors. J clin oncol. 2002;20(4):1043–8. 16. Subramanian A, Tamayo P, Mootha VK, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci USA. 2005;102(43):15545–50 Epub 2005/10/04. 17. Belharazem D, Kirchner M, Geissler F, et al. Relaxed imprinting of IGF2 in peripheral blood cells of patients with a history of prostate cancer. Endocrine connections. 2012;1(2):87–94 Epub 2013/06/20. 18. Lee SH, Bahn JH, Choi CK, et al. ESE-1/EGR-1 pathway plays a role in tolfenamic acid-induced apoptosis in colorectal cancer cells. Mol Cancer Ther. 2008;7(12):3739–50 Epub 2008/12/17. 19. Moon Y, Bottone FG Jr, McEntee MF, et al. Suppression of tumor cell invasion by cyclooxygenase inhibitors is mediated by thrombospondin-1 via the early growth response gene Egr-1. Mol Cancer Ther. 2005;4(10):1551–8 Epub 2005/10/18. 20. Nakagawa H, Chadwick RB, Peltomaki P, et al. Loss of imprinting of the insulin-like growth factor II gene occurs by biallelic methylation in a core region of H19-associated CTCFbinding sites in colorectal cancer. Proc Natl Acad Sci USA. 2001;98(2):591–6. 21. Cui X, Zhang P, Deng W, et al. Insulin-like growth factor-I inhibits progesterone receptor expression in breast cancer cells via the phosphatidylinositol 3-kinase/Akt/mammalian target of rapamycin pathway: progesterone receptor as a potential indicator of growth factor activity in breast cancer. Mol Endocrinol. 2003;17(4):575–88. 22. Cruz-Correa M, Cui H, Giardiello FM, et al. Loss of imprinting of insulin growth factor II gene: a potential heritable biomarker for colon neoplasia predisposition. Gastroenterology. 2004;126(4):964–70. 23. Ito Y, Koessler T, Ibrahim AE, et al. Somatically acquired hypomethylation of IGF2 in breast and colorectal cancer. Hum Mol Genet. 2008;17(17):2633–43. 24. Cheng YW, Idrees K, Shattock R, et al. Loss of imprinting and marked gene elevation are 2 forms of aberrant IGF2 expression in colorectal cancer. International journal of cancer Journal international du cancer. 2010;127(3):568–77. 25. Ohta M, Sugimoto T, Seto M, et al. Genetic alterations in colorectal cancers with demethylation of insulin-like growth factor II. Hum Pathol. 2008;39(9):1301–8. 26. Sasaki J, Konishi F, Kawamura YJ, et al. Clinicopathological characteristics of colorectal cancers with loss of imprinting of insulin-like growth factor 2. Int J Cancer. 2006;119(1):80–3. 27. Maenaka S, Hikichi T, Imai MA, et al. Loss of imprinting in IGF2 in colorectal carcinoma assessed by microdissection. Oncol Rep. 2006;15(4):791–5. 28. Nishihara S, Hayashida T, Mitsuya K, et al. Multipoint imprinting analysis in sporadic colorectal cancers with and without microsatellite instability. Int J Oncol. 2000;17(2):317–22. 29. Weisenberger DJ, Siegmund KD, Campan M, et al. CpG island methylator phenotype underlies sporadic microsatellite instability and is tightly associated with BRAF mutation in colorectal cancer. Nat Genet. 2006;38(7):787–93. 30. Ogino S, Cantor M, Kawasaki T, et al. CpG island methylator phenotype (CIMP) of colorectal cancer is best characterised by quantitative DNA methylation analysis and prospective cohort studies. Gut. 2006;55(7):1000–6.

123

J Gastroenterol 31. Shiovitz S, Bertagnolli MM, Renfro LA, et al. CpG island methylator phenotype is associated with response to adjuvant irinotecan-based therapy for stage III colon cancer. Gastroenterology. 2014;147(3):637–45. 32. Naito T, Nosho K, Ito M, et al. IGF2 differentially methylated region hypomethylation in relation to pathological and molecular features of serrated lesions. World J Gastroenterol. 2014;20(29):10050–61. 33. Baba Y, Nosho K, Shima K, et al. Hypomethylation of the IGF2 DMR in colorectal tumors, detected by bisulfite pyrosequencing, is associated with poor prognosis. Gastroenterology. 2010;139(6):1855–64. 34. Cancer Genome Atlas. N. Comprehensive molecular characterization of human colon and rectal cancer. Nature. 2012;487(7407):330–7. 35. Li X, Nadauld L, Ootani A, et al. Oncogenic transformation of diverse gastrointestinal tissues in primary organoid culture. Nat Med. 2014;20(7):769–77. 36. Sakatani T, Kaneda A, Iacobuzio-Donahue CA, et al. Loss of imprinting of Igf2 alters intestinal maturation and tumorigenesis in mice. Science. 2005;307(5717):1976–8. 37. Harper J, Burns JL, Foulstone EJ, et al. Soluble IGF2 receptor rescues Apc(Min/?) intestinal adenoma progression induced by Igf2 loss of imprinting. Cancer Res. 2006;66(4):1940–8. 38. Kaneda A, Wang CJ, Cheong R, et al. Enhanced sensitivity to IGF-II signaling links loss of imprinting of IGF2 to increased cell

123

39.

40.

41.

42.

43.

44.

45.

proliferation and tumor risk. Proc Natl Acad Sci USA. 2007;104(52):20926–31. Tada Y, Yamaguchi Y, Kinjo T, et al. The stem cell transcription factor ZFP57 induces IGF2 expression to promote anchorageindependent growth in cancer cells. Oncogene. 2015;34(6):752–60. Hassan AB, Howell JA. Insulin-like growth factor II supply modifies growth of intestinal adenoma in Apc(Min/?) mice. Cancer Res. 2000;60(4):1070–6. Morali OG, Delmas V, Moore R, et al. IGF-II induces rapid betacatenin relocation to the nucleus during epithelium to mesenchyme transition. Oncogene. 2001;20(36):4942–50. Gujral TS, Chan M, Peshkin L, et al. A noncanonical Frizzled2 pathway regulates epithelial-mesenchymal transition and metastasis. Cell. 2014;159(4):844–56. Endo M, Nishita M, Fujii M, et al. Insight into the role of wnt5ainduced signaling in normal and cancer cells. Int Rev Cell Mol Biol. 2015;314:117–48. Klenova EM, Morse HC 3rd, Ohlsson R, et al. The novel BORIS ? CTCF gene family is uniquely involved in the epigenetics of normal biology and cancer. Semin Cancer Biol. 2002;12(5):399–414. Pan Y, He B, Lirong Z, et al. Gene therapy for cancer through adenovirus vectormediated expression of the Ad5 early region gene 1A based on loss of IGF2 imprinting. Oncol Rep. 2013;30(4):1814–22.