Original Paper

Neuroendocrinology 2017;104:302–312 Received: January 12, 2016 DOI: 10.1159/000446917 Accepted after revision: May 15, 2016 Published online: May 25, 2016

Loss of 18 in Neuroendocrine Tumors of the Small Intestine: The Enigma Remains

a a a a b Maike Nieser Tobias Henopp Joachim Brix Laura Stoß Barbara Sitek b c d d Wael Naboulsi Martin Anlauf Anna M. Schlitter Günter Klöppel e f g g Thomas Gress Roland Moll Detlef K. Bartsch Anna E. Heverhagen h i j a Wolfram T. Knoefel Daniel Kaemmerer Johannes Haybaeck Falko Fend a a Jan Sperveslage Bence Sipos a b Institute of Pathology and Neuropathology, University Hospital of Tübingen, Tübingen , Medizinisches c Proteom-Center, Ruhr-Universität Bochum, Bochum , Institute of Pathology and Cytology, St. Vincenz-Krankenhaus, d e Limburg , Department of Pathology, Technische Universität München, Munich , Department of Gastroenterology f and Endocrinology, University Hospital of Marburg, Institute of Pathology, Philipps-Universität Marburg, and g h Department of Visceral, Thorax and Vascular Surgery, University Hospital of Marburg, Marburg , Department of General, Visceral and Pediatric Surgery, Heinrich Heine University and University Hospital Düsseldorf, Düsseldorf , i j and Department of General and Visceral Surgery, Zentralklinik Bad Berka, Bad Berka , Germany; Institute of Pathology, Medical University Graz, Graz , Austria

Key Words ysis of all seven putative Chr18-related tumor suppressors by Neuroendocrine tumors · Small intestine · · quantitative real-time PCR (qRT-PCR), Western blot and im- Tumor suppressors · Quantitative real-time PCR munohistochemistry. Next-generation exome sequencing and SNP array analysis were performed with five SI-NETs with (partial) loss of Chr18. Finally, we analyzed all microRNAs Abstract (miRNAs) located on Chr18 by qRT-PCR, comparing Chr18+/– Background/Aims: Neuroendocrine tumors of the small in- and Chr18+/+ SI-NETs. Results: Only DCC (deleted in colorec- testine (SI-NETs) exhibit an increasing incidence and high tal cancer) revealed loss of/greatly reduced expression in mortality rate. Until now, no fundamental molecular event 6/21 cases (29%). No relevant loss of SMAD2, SMAD4, elon- has been linked to the tumorigenesis and progression of gin A3 and CABLES was detected. PMAIP1 and maspin were these tumors. Only the loss of chromosome 18 (Chr18) has absent at the level. Next-generation sequencing did been shown in up to two thirds of SI-NETs, whereby the sig- not reveal relevant recurrent somatic mutations on Chr18 nificance of this alteration is still not understood. We there- either in an exploratory cohort of five SI-NETs, or in a valida- fore performed the first comprehensive study to identify Chr18-related events at the genetic, epigenetic and / protein expression levels. Methods: We did expression anal- J. Sperveslage and B. Sipos contributed equally to this study.

© 2016 S. Karger AG, Basel Bence Sipos Institute of Pathology, University of Tübingen Liebermeisterstrasse 8 E-Mail [email protected] DE–72076 Tübingen (Germany) www.karger.com/nen E-Mail bence.sipos @ med.uni-tuebingen.de tion cohort (n = 30). SNP array analysis showed no addition- NET, not one of the putative tumor suppressors or onco- al losses. The quantitative analysis of all 27 Chr18-related genic has been studied at the protein level. miRNAs revealed no difference in expression between Other proteins that have been discussed are PMAIP1 Chr18+/– and Chr18+/+ SI-NETs. Conclusion: DCC seems to and CABLES [7, 8]. Single point mutations or deletions/ be the only Chr18-related tumor suppressor affected by the insertions in , especially tumor suppressors of Chr18 monoallelic loss of Chr18 resulting in a loss of DCC protein regions that show an LOH could result in a total loss of expression in one third of SI-NETs. No additional genetic or function. Imprinted genes requiring only one mutational epigenetic alterations were present on Chr18. hit for inactivation are of special interest regarding poten- © 2016 S. Karger AG, Basel tial tumor suppressor genes. TCEB3C , an imprinted gene located on Chr18, functions as a transcription elongation factor and, hence, is an interesting tumor suppressor can- Introduction didate [9] . Furthermore, epigenetic modifications such as promoter methylation of Chr18 genes and/or LOHs may Neuroendocrine tumors (NETs) of the gastrointestinal have an impact on the tumorigenesis of SI-NETs. tract are considered to be rare tumors, representing only In addition to the loss of protein expression of tumor 2% of all gastrointestinal neoplasms. However, NETs of the suppressors, loss of Chr18 may result in a downregulation small intestine (SI-NETs) are the most common tumors of of Chr18-related microRNAs (miRNAs) involved in tu- this part of the gastrointestinal system [1] and have experi- mor progression. Similar effects on miRNA expression by enced a dramatic increase in incidence over the past three chromosome loss have been described for gastrointesti- decades [2]. SI-NETs are tumors with low rates of somatic nal stromal tumors in which the partial loss of chromo- mutations, as shown in 2 recent studies involving the anal- some 14 went hand in hand with the downregulation of ysis of 48 and 180 SI-NETs, respectively, by massive paral- 38 chromosome 14-related miRNAs [10]. Another, albe- lel sequencing, which revealed a mutation frequency (sin- it unlikely, possibility is that the loss of Chr18 represents gle nucleotide variants; SNVs) of approximately 0.1 per 106 a passenger gene alteration with no distinct impact on the nucleotides. PI3K/Akt/mTOR signaling molecules showed tumor progression. copy number changes in 29% of the analyzed cases. Hetero- In order to address these questions, we performed the zygous frameshift mutations of CDKN1B , encoding the tu- first systematic study on Chr18-related alterations at the mor suppressor p27, were found in 8% of the cases, while protein level in SI-NETs, complemented with the assess- 21 out of 48 analyzed SI-NETs showed monoallelic dele- ment of genetic alterations such as SNVs, insertions/de- tions of SMAD2 and SMAD4 [3, 4] . SMAD2 and SMAD4 letions (indels), gain and loss as well as LOH. Our ex- are located in close proximity to each other on chromo- perimental setup is depicted in a flow chart which can some 18 (Chr18), whose loss is the most common event in be found in the online supplementary data 1 (see www. SI-NETs [5] , but the role of this frequent genetic aberration karger.com/doi/10.1159/000446917 for all online suppl. for the tumor progression of SI-NETs has not yet been clar- material). Online supplementary table S1 summarizes all ified. The loss of one Chr18 allele could possibly result in analyses we performed (in detail and in a compact over- the partial loss of tumor suppressor proteins located on view) and the samples which were used. Chr18, such as SMAD2, SMAD4, DCC (deleted in colorec- tal cancer) and maspin. A recent integrated (epi)genomic and transcriptomic study identified 25 Chr18-related genes Material and Methods to be differentially methylated between tumors affected by Chr18 loss of heterozygosity (LOH) and normal small in- Material and methods can be found in the online supplemen- tary data 2. testine. For some of these genes, the aberrant methylation pattern was associated with reduced mRNA expression, e.g. SERPINB5 , coding for maspin. In addition, the authors were able to classify SI-NETs in three distinct subgroups, Results with the Chr18 LOH group being the most favorable and significantly associated with reduced global methylation Chr18 Is Commonly Lost in SI-NETs status [in contrast to no copy number variations (CNVs) Focusing on this characteristic lesion, FISH analysis and multiple CNVs] [6] . Although these recent reports with a Chr18 centromeric enumeration probe was per- gave important new insights into the genetic events in SI- formed on 132 SI-NET samples. The analysis was possible

Chr18-Related Alterations in SI-NETs Neuroendocrinology 2017;104:302–312 303 DOI: 10.1159/000446917 70 SMAD2 Fig. 1. Western blot of the Chr18-associat- 55 ed tumor suppressors SMAD2 and SMAD4. The samples show variable strength of ex- 42 DŽ-Actin pression of SMAD2/4 (SMAD2: 60 kDa, 61 SMAD4 SMAD4: 61 kDa, β-actin: 42 kDa). M = Mosaicism; 1 = 1 × Chr18; 2 = 2 × Chr18; 42 DŽ-Actin P = positive control. Numbers indicate tu- PN12 1 1 1 1 1 1 2 1 M 1 M 1 mor content; red arrow indicates specific 90 50 80 85 90 80 90 80 90 75 90 55 90 65 band.

in 95% (125/132) of all cases. It revealed that 71% of the slightly between samples. One sample was negative for tumor samples (89/125) had a loss of Chr18. 11% showed SMAD2 (sample 6, <20% of β-actin expression). This mosaicism regarding their Chr18 status (14/125); 18% ex- sample depicted loss of Chr18. Half of the samples showed hibited the normal count of two copies of Chr18 (22/125). (slightly) higher SMAD4 expression than expression of In 5% (7/132), it was not possible to determine the Chr18 the housekeeping protein β-actin, whereas the other half status due to tissue degradation. In the three cohorts, 57% of the samples showed similar expression for both pro- of cohort 1, 65% of cohort 2 and 65% of cohort 3 showed teins. a loss of Chr18. A total of 84 out of 87 tumor samples (97%) exhibited The Chr18 status for the 5 samples used for exome strong nuclear expression of SMAD4 in immunohisto- sequencing (Tu1 Met, Tu3, Tu4, Tu5 Met and Tu7 Met) chemical staining ( fig. 2 a). Three samples showed no ex- and one additional sample for SMAD2/4 Western blot pression of this protein, whereas the corresponding epi- (Tu6 Met) was evaluated by SNP array and revealed 4 thelium of the normal mucosa still expressed SMAD4 samples to have lost one copy of Chr18, whereas one ( fig. 2 b). However, compared to the loss of SMAD4 in sample depicted mosaicism (Tu3) and one showed pancreatic ductal adenocarcinomas, the loss in our SI- the normal count of two copies (Tu6 Met). The clinical NET samples is somehow less prominent ( fig. 2 b, small data of all cases can be found in online supplementary picture). table S2. The immunohistochemical staining of SMAD2 re- vealed no difference of expression in the tumor tissues Expression of Chr18-Related Tumor Suppressor Genes compared to the corresponding normal tissues (fig. 2 c). in SI-NETs The protein is strongly expressed in the nuclei of SI- The common loss of Chr18 in SI-NETs suggests that NETs, but to the same degree in the nuclei of the adjacent tumor suppressor genes associated with Chr18 play a role normal epithelium. in the tumorigenesis of this rare type of tumors. Maspin Is Not Expressed by Neuroendocrine Cells of SMAD2 and SMAD4 Are Strongly Expressed in the Ileal Mucosa SI-NETs The immunohistochemical staining for maspin was In order to analyze whether Chr18 loss is accompanied negative in all SI-NET samples; in contrast, the normal with a loss of protein expression of the two tumor sup- mucosa exhibited expression of this protein (fig. 2 d). In pressors SMAD2 and/or SMAD4, Western blot analyses order to explore whether the lacking expression of maspin were performed with protein lysates from 14 SI-NET represents a loss of expression during tumorigenesis or samples (10 with loss of Chr18, confirmed by Chr18 cen- whether, instead, it belongs to the normal phenotype of tromere FISH, 2 without loss of Chr18 and 2 samples with neuroendocrine (NE) cells of the small intestine, double a mosaicism regarding Chr18 status). Both potential immunofluorescence stainings of synaptophysin and tumor suppressor proteins were detected in varying maspin were conducted on 20 normal small intestine tis- amounts in all samples investigated ( fig. 1 , online suppl. sues adjacent to SI-NETs. This experiment revealed that table S3A). Nevertheless, we found that the SI-NET sam- NE cells of the ileal mucosa do not express maspin, in ples showed differences in the strength of expression of contrast to mucin-producing cells of the intestinal mu- SMAD2, whereas the β-actin expression varied only cosa (fig. 2 e).

304 Neuroendocrinology 2017;104:302–312 Nieser et al. DOI: 10.1159/000446917 a b

c d

e

Fig. 2. a Strong expression of SMAD4/DPC4 in an SI-NET sample. SMAD2 in an SI-NET sample and normal epithelium. ×200. d Im- IHC. ×200. b Loss of SMAD4/DPC4 expression in the tumor tissue munohistochemical staining of maspin is only present in normal in contrast to strong expression of the protein in normal mucosa. mucosa. SI-NETs do not express the protein. ×400. e Double im- ×400. The small pictures show loss of SMAD4 expression in pan- munofluorescence staining of synaptophysin (green) and maspin creatic ductal adenocarcinoma in contrast to retained protein ex- (red). NE cells of the ileal mucosa show no expression of maspin. pression in normal tissue. ×400. c Strong nuclear expression of ×1,000. TU = Tumor.

Elongin A3 and CABLES Show Strong Expression in sis was performed, identifying the upper band to be the SI-NETs, whereas PMAIP1 Is Not Expressed specific one (fig. 3 , middle panel; only the first 8 samples Western blot analyses of the two tumor suppressors are shown). Elongin A3 exhibited distinct positive results elongin A3 and CABLES were performed with 21 fresh in 20 of the 21 samples; one sample was negative (not frozen tissues. CABLES was detected in varying amounts shown). in all analyzed samples (fig. 3 , upper panel, only the first Western blot analysis of PMAIP1 revealed negativity 8 samples are shown). Since two different bands were vis- (<20% of β-actin expression) of the first 8 samples ( fig. 3 , ible in the elongin A3 Western blot, a competition analy- lower panel); therefore, no further samples were tested.

Chr18-Related Alterations in SI-NETs Neuroendocrinology 2017;104:302–312 305 DOI: 10.1159/000446917 70 CABLES 55

42 DŽ-Actin

Elongin A3 70

55

70 Elongin A3 + peptide 55

42 DŽ-Actin Fig. 3. Western blot of the Chr18-associat- ed tumor suppressors CABLES, elongin A4 11 PMAIP1 and PMAIP1. CABLES (67 kDa) and elon- gin A3 (60 kDa) are expressed in all sam- 42 DŽ-Actin ples. PMAIP1 (15 kDa) is not expressed. M = Mosaicism; 1 = 1 × Chr18; 2 = 2 × Chr18; M 1 1 1 1 1 1 2 P P = positive control. Numbers indicate tu- 90 90 55 90 80 75 80 90 mor content; red arrows indicate specific bands.

Quantitative Western blot results are shown in online samples 3 and 6 were not analyzable in the second West- supplementary table S3B. ern blot. All samples with lost/reduced expression of DCC depicted only one copy of Chr18. DCC Expression Is Lost/Reduced in Nearly 30% of SI- NET Samples Real-Time PCR Analysis Reveals TCEB3C and Western blot analysis of DCC using the first antibody CABLES to Be Differentially Expressed between was performed with 21 tumor samples. The evaluation of Sample Cohorts with and without Loss of Chr18 the Western blot results revealed total loss or reduced Since only one cryo-sample with the normal count of DCC expression (<20% of β-actin expression) in 29% Chr18 was available for Western blot analyses of TCEB3C, (6/21) of samples (fig. 4 a, online suppl. table S3C). PMAIP1, DCC and CABLES, we decided to additionally With the first 8 tumor samples, a second Western blot perform quantitative real-time PCR (qRT-PCR) to ad- was performed, using another antibody (fig. 4 b). Due to dress the question of the influence of the Chr18 loss on difficulties in distinguishing the bands between 180 and these tumor suppressors. 250 kDa, a competition with the matching peptide was The qRT-PCR of TCEB3C , PMAIP1 , DCC and CA- done. In addition, a label-free identification of DCC-spe- BLES in a great cohort of 69 formalin-fixed paraffin-em- cific peptides in the positive control IMR-32 by mass bedded (FFPE) samples showed no common loss of one spectrometry was performed. The mass spectrometry of the potential tumor suppressor genes with the Cp val- confirmed DCC to be present in the gel bands at ∼ 190 ues ranging between 22 (high expression) and 35 (low kDa. expression). The second Western blot analysis revealed samples 5 Four samples had Cp values >30 in the normalization and 7 to have lost DCC expression. This result was sup- control of β-Actin and were therefore excluded from fur- ported by the first Western blot: sample 7 lost DCC ex- ther analysis. Of the remaining 65 samples, 38 exhibited pression as well and sample 5 exhibited reduced expres- the loss of one Chr18, 13 samples had the normal count sion of DCC. Since the background signal was very high, of two Chr18, and 12 tumors showed mosaicism regard-

306 Neuroendocrinology 2017;104:302–312 Nieser et al. DOI: 10.1159/000446917 ++(+) + (+) +–++ 158 DCC

42 DŽ-Actin

M 1 1 1 1 1 1 2 P 90 90 55 90 80 75 80 90 ++(+) – + + ++

158 DCC

42 DŽ-Actin

1 1 1 1 1 1 1 P 90 85 85 55 55 75 70

(+) + ++ + + +

158 DCC

42 DŽ-Actin

1 1 1 1 1 1 P Fig. 4. a Western blot of the Chr18-associ- a 90 90 80 60 80 100 ated tumor suppressor DCC. DCC is lost in samples 7, 12 and 16. Samples 3, 5, 10, 11 and 17 show greatly reduced expression of DCC (158 kDa). b DCC Western blot of the ++ ? +– ?++ – first 8 tumor samples, using a second anti- 250 180 DCC body. A competition analysis with the matching peptide was carried out to deter- 250 280 DCC + peptide mine the right protein band. DCC is lost in samples 5 and 7 (190 kDa). Due to the 42 DŽ-Actin strong background, samples 3 and 6 are not analyzable. M = Mosaicism; 1 = 1 × Chr18; M 1 1 1 1 1 1 2 P 2 = 2 × Chr18; P = positive control. Num- b 90 90 55 90 80 75 80 90 bers indicate tumor content; red arrows in- dicate specific bands.

ing their Chr18 status. It was not possible to ascertain the detected for DCC (p = 0.2151 for Ex3–4 and p = 0.3134 Chr18 status for 2 samples. for Ex17–18, respectively) and PMAIP1 (p = 0.0652) The comparative analysis of cohorts with/without ( fig. 5 ). loss of Chr18 (the mosaic samples were excluded from this analysis) revealed a significant downregulation of No Additional LOH Detected TCEB3C (p = 0.0101) and CABLES (p = 0.0323) in the SNP array analysis of six SI-NETs revealed 4 samples cohort with the loss of Chr18 (n = 38) compared to the to have a loss of Chr18; one primary tumor exhibited mo- cohort with the normal Chr18 count (n = 13). No signifi- saicism regarding Chr18 status and one sample depicted cant difference in expression between the two cohorts was the normal count of two Chr18. In the samples with a loss

Chr18-Related Alterations in SI-NETs Neuroendocrinology 2017;104:302–312 307 DOI: 10.1159/000446917 20 DCC Ex3–4 20 DCC Ex17–18

15 15

10 10

©&SYDOXHV 5 ©&SYDOXHV 5

0 0 1 × Chr18 2 × Chr18 1 × Chr18 2 × Chr18

10 TCEB3C 15.0 CABLES 8 12.5 6 10.0 4 7.5 2 5.0 ©&SYDOXHV ©&SYDOXHV 0 2.5 –2 0 1 × Chr18 2 × Chr18 1 × Chr18 2 × Chr18

Fig. 5. ΔCp values of the 4 tumor suppres- 15.0 PMAIP1 sor genes DCC (two primer pairs amplify- 12.5 ing different exons of the gene were tested), 10.0 TCEB3C , CABLES and PMAIP1 . The Cp 7.5 values of the genes were normalized to 5.0 β-actin . TCEB3C and CABLES were signif- ©&SYDOXHV icantly downregulated ( * p = 0.0101 and 2.5 p = 0.0323, respectively) in the cohort with 0 1 × Chr18 2 × Chr18 a loss of Chr18 compared to the cohort with a normal count of Chr18.

of Chr18, no further losses (with the analysis settings ex- deleterious effect on protein structure/function. The re- plained in online suppl. data 2: SNP array analysis) on the maining 3 samples showed no SNVs or indels in Chr18- remaining Chr18 could be detected. related genes. All 3 mutations were Sanger validated with DNA ex- No Recurrent Mutations in Chr18-Associated Genes tracted from fresh frozen tissue (tumor as well as corre- Detected sponding normal tissue) confirming a somatic origin. Exome sequencing was performed with 5 different SI- NFATC1 was analyzed in a larger cohort of 30 SI-NETs NET samples (4 with loss of Chr18, 1 with mosaic Chr18 (15 primaries – 8 with matching metastases – 7 non- pattern) and corresponding normal tissue samples. Sta- matching metastases) in which no genetic alterations tistics concerning the mean coverage on target and un- were found. derrepresented bases (total and in percent) are shown in online supplementary table S4. Three somatic mutations Data Comparison with External Data Set in genes located on Chr18 were identified (see online sup- Banck et al. [4] detected 197 somatic alterations in 48 pl. table S6). SNVs in CABYR and NFATC1 were found patients. Concerning Chr18, they identified only one in 1 patient, while a 3-bp deletion in PIEZO2 was present SNV. The gene ANKRD30B exhibited a T > A substitu- in another patient. The SNV in NFATC1 was predicted to tion at position 14796361 (p.V625D, predicted to be be- be deleterious/damaging (Provean/SIFT). In contrast, the nign). The tumor in which the mutation was found SNV in CABYR was predicted to have no damaging effect showed no loss of Chr18. In comparison, in our exome on protein structure/function (neutral/tolerated). The sequencing of 5 patients with SI-NETs, we found two 3- deletion in PIEZO2 was predicted to have a SNVs and one deletion in genes on Chr18 in 2 patients.

308 Neuroendocrinology 2017;104:302–312 Nieser et al. DOI: 10.1159/000446917 The tumor that had mutations in NFATC1 and CABYR Chr18 using RNA or protein-based assays, comparing SI- exhibited a loss of one Chr18; the tumor with a 3-base NETs with and without Chr18 losses. pair deletion in PIEZO2 revealed a mosaic pattern con- SMAD2 and SMAD4 are important signal transduc- cerning Chr18 status. tion molecules in the transforming growth factor-β (TGFβ) pathway [15] and are known to be functionally Chr18-Related miRNAs Are Not Differentially inactivated in different types of cancers. SMAD4 function Expressed is lost in pancreatic adenocarcinomas [16] , metastatic Ten samples with and 10 samples without loss of Chr18 colorectal cancer [17] and small intestine adenocarcino- were analyzed concerning their Chr18-related miRNA ex- mas [18] . SMAD2 has also been described to be altered in pression profiles. In total, 27 miRNAs located on Chr18 a variety of cancers, although to a lesser extent than were analyzed (see online suppl. table S7). Since no ex- SMAD4 [19, 20] . perimental data of expression evidence are available for In the SI-NETs of our series, SMAD2 and SMAD4 ex- miR-5011 and miR-5583, both mature sequences (3p and pression, tested by Western blot and IHC, remained un- 5p) were analyzed. A total of 29 different miRNA assays altered, except for 3 tumors that showed loss of SMAD4 were therefore performed with 11 miRNA assays without expression. These findings correlate with data from Löll- detectable expression of the respective miRNAs in either gen et al. [21], who reported on preserved SMAD4 pro- of the two cohorts of SI-NETs. Four miRNAs (miR-302f, tein expression in 7 SI-NETs. miR-4526, miR-4743-5p and miR-5011-5p) exhibited Maspin has shown to be downregulated in cancers of very low expression with Cp values >35 and were there- the breast, prostate and stomach, and in melanoma, but fore excluded from further analysis. The statistical analy- overexpressed in, e.g., colorectal, pancreatic and thyroid sis of the remaining miRNAs that exhibited moderate to cancers. These contradictory findings may be explained strong expression revealed no significant difference of ex- by the special subcellular localization of the protein in pression between the 2 cohorts. Online supplementary cancer cells, which may be cytoplasmic or nuclear or both table S8 shows significance values and fold changes of 15 [22] . However, intensive studies have shed light on the miRNAs that showed high expression in the RT-PCR tumor suppressor function of maspin by detecting repro- analysis. Only one miRNA was nearly significantly regu- gramming of the tumor proteome via maspin expression, lated (miR-1539; p = 0.06), and therefore the analysis of particularly of protein pathways involved in tumor cell miR-1539 was extended to 40 samples in total (10 addi- extravasation [23] . tional samples with and 10 without Chr18 loss). However, Maspin was not expressed in any of the 87 SI-NETs, no increase in significance was detected (p = 0.26). The but was present in the adjacent normal mucosa. This fold changes between the two cohorts ranged between finding would make maspin an interesting candidate. –2.55 and 1.36. All results concerning the Chr18 associ- However, the 100% loss of a protein during tumorigenesis ated tumor suppressors and miRNAs investigated in our is very unusual. We therefore examined maspin expres- study are summarized in online supplementary table S9. sion in the non-neoplastic single NE cells of the intesti- nal mucosa by immunofluorescent double staining for maspin and synaptophysin. This analysis revealed that Discussion maspin is only present in the mucin-producing cells of the normal ileal mucosa and absent in the NE cells. This Most of SI-NETs are characterized by a loss of (a part) suggests that the lacking maspin expression in neoplastic of Chr18 [11–13]. No other frequent genetic alterations SI-NET cells does not reflect an oncogenic event but the or putative affected pathways have been implicated in the phenotype of the non-neoplastic NE cells of the mucosa tumorigenesis of SI-NETs as yet. In our series of 138 SI- of the small intestine. NETs, we found Chr18 loss in approximately 65% of the In 1990, DCC was identified to be frequently deleted tumors. A number of studies (reviewed in [14] ) investi- in colorectal carcinoma [24] . Subsequently, reduced ex- gated the possible effect of Chr18 loss on the genes that pression of DCC has been described in a variety of can- are located in this chromosomal region. As there are, cers [25, 26] . DCC encodes a netrin-1 receptor, which in- however, no comprehensive data on Chr18-related alter- duces apoptosis in the absence of netrin-1 [27] . Abridged ations at the protein level in SI-NETs, we studied all expression of DCC could result in less apoptosis and thus 7 putative tumor suppressors (i.e. SMAD2, SMAD4, give rise to tumor progression. DCC is a 158.5-kDa trans- maspin, DCC, elongin A3, CABLES and PMAIP1) on membrane protein with at least 18 splice variants (7 being

Chr18-Related Alterations in SI-NETs Neuroendocrinology 2017;104:302–312 309 DOI: 10.1159/000446917 protein coding). Since no reliable anti-DCC antibodies (n = 23) [35], but did not correlate with grade/stage/sur- for IHC exist, we assessed DCC expression using qRT- vival data. Whether the loss of CABLES expression was PCR of different exon boundaries and Western blot as- caused by the loss of Chr18 was not evaluated in this study. says with peptide competition and mass spectrometry. PMAIP1 is a proapoptotic gene whose protein func- The qRT-PCR of DCC revealed no difference in expres- tions in a p53-dependent manner [36] . In 2008, PMAIP1 sion between the two cohorts (38 samples with and 13 was identified as a potential tumor suppressor in pancre- without loss of Chr18). Western blot showed a loss/re- atic cancer by comparative cDNA microarray analysis duced expression of DCC in 29% (6/21) of SI-NETs. The [7] . In our study, the qRT-PCR of PMAIP1 revealed high specificity of the assay was verified with a biological pos- Cp values, suggesting that PMAIP1-mRNA is low abun- itive control (IMR-32 cell line), competition of the reac- dant in SI-NETs irrespective of their Chr18 status. The tion with a corresponding peptide and the mass spec- protein PMAIP1 was not expressed in any of our samples. trometry of the positive control. The mechanism of loss Since only 8 cryo-samples were investigated, the explana- of DCC has yet to be unraveled. A possible explanation is tory power is somehow reduced. We think that PMAIP1, . Remarkably, Francis et al. [3] report- like maspin, is not present in NE cells at all. To confirm ed the highest rate (29%) of intronic alterations in the our hypothesis, we need to do a double immunofluores- DCC gene out of all putative tumor suppressor candidates cence staining (synaptophysin + PMAIP1). However, no on Chr18, rendering this gene even more interesting. suitable antibody for PMAIP1 is available at the moment. TCEB3C encodes for elongin A3, a transcription elon- Our first systematic search for putative tumor suppres- gation factor identified in 2002 [9] . TCEB3C is a mater- sor proteins, which could get lost during tumorigenesis in nally imprinted gene on Chr18 [28] , and hence only one SI-NETs, revealed only DCC as a promising candidate in mutational hit is required to fulfill the Knudson two-hit one third of SI-NETs, although the mechanism of loss re- hypothesis. Edfeldt et al. [29] were able to show that mains unclear. Our study has strengths and limitations. TCEB3C is epigenetically repressed in the human SI-NET The strengths include the comprehensive approach to all cell line CNDT2.5 due to histone and DNA methylation. Chr18-related events and the thorough examinations of Despite this, a high percentage of human SI-NETs in this the putative tumor suppressors at the protein level. The study exhibited intact elongin A3 expression. We found power of the study is limited by the low number of frozen that the qRT-PCR of TCEB3C in 69 FFPE samples revealed tissue samples (14–21); however, most of the tumor sup- significant downregulation in SI-NETs with a loss of Chr18 pressors exhibited an unequivocal expression profile, so compared to the samples with two copies (p = 0.0101). The that it was possible to draw clear conclusions. The second comparison of the mRNA results with the protein expres- limitation concerns the lack of high-value antibodies for sion was somehow limited, due to the fact that only one immunohistochemistry, making impossible to investigate fresh frozen sample with normal Chr18 count was avail- high numbers of FFPE samples and to study co-localiza- able for Western blot analysis. However, no loss of protein tion of tumor suppressors in normal NE cells. The lacking expression was detected in 21 fresh frozen samples. These loss of known tumor suppressors prompted us to proceed results indicate that reduced mRNA expression was not with additional genetic events on the remaining copy of associated with a relevant loss of protein. Chr18 in order to identify novel putative tumor suppres- CABLES is a cell cycle regulatory protein that plays a sor candidates that become inactivated by LOH. In this role in proliferation and differentiation [30] ; the encoding process, we analyzed 5 SI-NETs that showed complete gene was found to be silenced in ovarian, colorectal, en- (n = 4) or partial (n = 1) loss of Chr18 by exome sequenc- dometrial and non-small lung cancer [31–34]. In 2013, the ing. We found no SNVs or indels in known tumor sup- group led by Zukerberg reported that the (partial) loss of pressor genes in the 5 analyzed SI-NETs. To extend this CABLES is also a frequent aberration in SI-NETs (Arna- search, we conducted data mining in a supplementary son et al., 2013, Nature Abstract Laboratory Investiga- data set of the study by Banck et al. [4] and, again, were tion). Our analysis exhibited a slight downregulation of unable to find any relevant additional losses (SNVs in all CABLES (p = 0.0323) in SI-NET samples with Chr18 loss 7 tumor suppressors investigated). compared to tumors with diploid Chr18. However, the Regarding other Chr18-related genes, we found three reduced mRNA expression was not associated with re- somatic mutations by exome sequencing that could be val- duced protein expression; all samples showed strong pro- idated by Sanger sequencing. One patient carried SNVs in tein expression in Western blot analysis. In a cohort of 25 the CABYR and NFATC1 genes, whereas a 3-bp deletion SI-NETs, (partial) loss of CABLES was observed in 92% in PIEZO2 was present in another patient. The 3 remain-

310 Neuroendocrinology 2017;104:302–312 Nieser et al. DOI: 10.1159/000446917 ing patients showed no genetic alterations in Chr18 genes. derrepresented by hemizygous deletions in malignant tu- NFATC1 becomes activated by calcium flux, resulting in mors. In other words, multiple haploinsufficiencies prob- translocation from the cytoplasm to the nucleus, where it ably contribute to the proliferative fitness of cancers. This promotes de novo gene transcription. It seems to play a hypothesis may at least in part explain the significance of role in various cancer types, e.g. by overexpression and/or Chr18 loss in SI-NETs. promotion of tumor angiogenesis [37] . NFATC1 could In conclusion, this first comprehensive study on al- therefore represent an interesting candidate gene for tu- terations of putative tumor suppressor proteins on Chr18 mor progression of SI-NETs, which is why it was further in SI-NETs showed that DCC is the only tumor suppres- analyzed in a set of 30 SI-NET FFPE samples by Sanger sor that is lost in 29% of cases, while SMAD2, SMAD4, sequencing of all functional gene regions. No additional CABLES and elongin A3 retained their expression. NFATC1 mutations were found in this cohort of tumors, PMAIP1 and maspin were not expressed at all. This find- suggesting that NFATC1 mutations are not a common ing should be interpreted as a feature of NE cells in the event in SI-NETs. Since the other two genes, PIEZO2 and small intestine in general rather than as a complete loss of CABYR, are very unlikely to act as tumor suppressors (see these molecules during tumorigenesis. Exome sequenc- online suppl. data 3), we decided against extending the ing of SI-NETs performed by us and others revealed no validation experiments to more samples. These results, in recurrent Chr18 alterations, which could lead to addi- line with recent data from Banck et al. [4] and Delgado tional potential tumor suppressors. Regarding epigenetic Verdugo et al. [38] , indicate clearly that no additional loss- regulation, there is no significant downregulation of all es of heterozygosity occur on Chr18, at least at the DNA known Chr18-related miRNAs. The majority of epithe- level. We continued our comprehensive exploration of lial tumors are triggered by driver genetic events, mostly Chr18-related alterations at the epigenetic level. by mutations. SI-NETs seem to be controlled by different Since the loss of one Chr18 may result in the down- mechanisms. Possibly, multiple monoallelic losses of tu- regulation of Chr18-related miRNAs involved in tumor mor suppressor genes contribute to malignant transfor- progression, we performed a comparative analysis of mation. If this is the case, further studies have to address miRNA expression of 27 miRNAs in 10 samples with and this question using integrative analysis of genetic altera- 10 samples without loss of Chr18. In tumors, miRNAs tions and transcriptome and proteome signatures to shed can either function as oncogenes or tumor suppressors light on the complex and subtle events in SI-NETs. and have been shown to affect tumor progression of var- ious kinds of cancer. In gastrointestinal stromal tumors, a partial loss of chromosome 14 went hand in hand with Acknowledgements downregulation of Chr14-related miRNAs [10] . Two studies confirmed miRNA-133a, a Chr18-associated The authors would like to thank Karen Greif, Christine Be- schorner, Dennis Thiele and Brigitte Frank for their excellent tech- miRNA, to be downregulated during tumor progression nical assistance. [39, 40]. However, the authors did not mention if this downregulation is due to loss of Chr18. In our study, we found no significant downregulation of the 27 Chr18-re- Disclosure Statement lated miRNAs in SI-NETs with loss of Chr18. In summary, the question arises as to how one can ex- The authors declare that there are no conflicts of interest. plain the effect of the Chr18 loss in SI-NETs. One possible mechanism is haploinsufficiency of one of the putative tumor suppressors. Of the 7 putative tumor suppressor genes, it has been shown that SMAD4 haploinsufficiency References 1 Massironi S, Sciola V, Peracchi M, Ciafardini significantly alters TGFβ and BMP (bone morphogenetic C, Spampatti MP, Conte D: Neuroendocrine tumors of the gastro-entero-pancreatic sys- proteins) signaling, and that a SMAD4 dose-dependent tem. World J Gastroenterol 2008; 14: 5377– transcriptional regulation of target genes of the TGFβ and 5384. Wnt signaling pathways in a SMAD4 -mutant mouse 2 Modlin IM, Lye KD, Kidd M: A 5-decade anal- ysis of 13,715 carcinoid tumors. Cancer 2003; model exists [41] . A second explanation may be provided 97: 934–959. by the study of Solimini et al. [42] . They were able to show 3 Francis JM, Kiezun A, Ramos AH, Serra S, et al: Somatic mutation of CDKN1B in small in- that pro-proliferative genes (GO genes) are often en- testine neuroendocrine tumors. Nat Genet riched and anti-proliferative genes (STOP genes) are un- 2013; 45: 1483–1486.

Chr18-Related Alterations in SI-NETs Neuroendocrinology 2017;104:302–312 311 DOI: 10.1159/000446917 4 Banck MS, Kanwar R, Kulkarni AA, Boora GK, Fischer A, Yeo CJ, Hruban RH, Kern SE: c-Abl and facilitates Cdk5 tyrosine phosphory- Metge F, Kipp BR, Zhang LZ, Thorland EC, DPC4, a candidate tumor suppressor gene at lation, kinase upregulation, and neurite out-

Minn KT, Tentu R, Eckloff BW, Wieben ED, human chromosome 18q21.1. Science 1996; growth. Neuron 2000; 26: 633–646.

Wu YH, Cunningham JM, Nagorney DM, Gil- 271: 350–353. 31 Sakamoto H, Friel AM, Wood AW, Guo L, Ilic bert JA, Ames MM, Beutler AS: The genomic 17 Miyaki M, Iijima T, Konishi M, Sakai K, Ishii A, Seiden MV, Chung DC, Lynch MP, Serika- landscape of small intestine neuroendocrine A, Yasuno M, Hishima T, Koike M, Shitara N, wa T, Munro E, Oliva E, Orsulic S, Kirley SD,

tumors. J Clin Invest 2013; 123: 2502–2508. Iwama T, Utsunomiya J, Kuroki T, Mori T: Foster R, Zukerberg LR, Rueda BR: Mecha- 5 Cunningham JL, Diaz de Stahl T, Sjoblom T, Higher frequency of Smad4 gene mutation in nisms of cables 1 gene inactivation in human Westin G, Dumanski JP, Janson ET: Common human colorectal cancer with distant metasta- ovarian cancer development. Cancer Biol Ther

pathogenetic mechanism involving human sis. Oncogene 1999; 18: 3098–3103. 2008; 7: 180–188. chromosome 18 in familial and sporadic ileal 18 Blaker H, von Herbay A, Penzel R, Gross S, 32 Park do Y, Sakamoto H, Kirley SD, Ogino S, carcinoid tumors. Genes Can- Otto HF: Genetics of adenocarcinomas of the Kawasaki T, Kwon E, Mino-Kenudson M,

cer 2011; 50: 82–94. small intestine: frequent deletions at chromo- Lauwers GY, Chung DC, Rueda BR, Zukerberg 6 Karpathakis A, Dibra H, Pipinikas C, Feber A, some 18q and mutations of the SMAD4 gene. LR: The cables gene on chromosome 18q is si-

Morris T, Francis JM, Oukrif D, Mandair D, Oncogene 2002; 21: 158–164. lenced by promoter hypermethylation and al- Pericleous M, Mohmaduvesh M, Serra S, 19 Takagi Y, Koumura H, Futamura M, Aoki S, lelic loss in human colorectal cancer. Am J

Ogunbiyi O, Novelli M, Luong T, Asa SL, Kul- Ymaguchi K, Kida H, Tanemura H, Shimoka- Pathol 2007; 171: 1509–1519. ke MH, Toumpanakis C, Meyer T, Caplin M, wa K, Saji S: Somatic alterations of the SMAD- 33 Zukerberg LR, DeBernardo RL, Kirley SD, Meyerson M, Beck S, Thirlwell C: Prognostic 2 gene in human colorectal cancers. Br J Can- D’Apuzzo M, Lynch MP, Littell RD, Duska LR,

impact of novel molecular subtypes of small in- cer 1998; 78: 1152–1155. Boring L, Rueda BR: Loss of cables, a cyclin- testinal neuroendocrine tumour. Clin Cancer 20 Maliekal TT, Antony ML, Nair A, Paulmuru- dependent kinase regulatory protein, is associ-

Res 2016; 22: 250–258. gan R, Karunagaran D: Loss of expression, and ated with the development of endometrial hy- 7 Ishida M, Sunamura M, Furukawa T, Lefter mutations of Smad 2 and Smad 4 in human perplasia and endometrial cancer. Cancer Res

LP, Morita R, Akada M, Egawa S, Unno M, Ho- cervical cancer. Oncogene 2003; 22: 4889–4897. 2004; 64: 202–208. rii A: The PMAIP1 gene on chromosome 18 is 21 Löllgen RM, Hessman O, Szabo E, Westin G, 34 Tan D, Kirley S, Li Q, Ramnath N, Slocum HK, a candidate tumor suppressor gene in human Akerstrom G: Chromosome 18 deletions are Brooks JS, Wu CL, Zukerberg LR: Loss of ca-

pancreatic cancer. Dig Dis Sci 2008; 53: 2576– common events in classical midgut carcinoid bles protein expression in human non-small

2582. tumors. Int J Cancer 2001; 92: 812–815. cell lung cancer: a tissue microarray study.

8 Wu CL, Kirley SD, Xiao H, Chuang Y, Chung 22 Berardi R, Morgese F, Onofri A, Mazzanti P, Hum Pathol 2003; 34: 143–149. DC, Zukerberg LR: Cables enhances cdk2 tyro- Pistelli M, Ballatore Z, Savini A, De Lisa M, 35 Arnason T, Pino MS, Chung DC, Sapp HL, sine 15 phosphorylation by Wee1, inhibits cell Caramanti M, Rinaldi S, Pagliaretta S, Santoni Rayson D, Deshpande V, Zukerberg LR: Loss growth, and is lost in many human colon and M, Pierantoni C, Cascinu S: Role of maspin in of CABLES expression in neuroendocrine tu-

squamous cancers. Cancer Res 2001; 61: 7325– cancer. Clin Transl Med 2013; 2: 8. mors (NETs) of the gastrointestinal tract, pan-

7332. 23 Chen EI, Yates JR: Maspin and tumor metasta- creas, and lung. Nat Abstracts Lab Invest 2013;

9 Yamazaki K, Guo L, Sugahara K, Zhang C, sis. IUBMB Life 2006; 58: 25–29. 600: 143A–144A. Enzan H, Nakabeppu Y, Kitajima S, Aso T: 24 Fearon ER, Cho KR, Nigro JM, Kern SE, Si- 36 Yakovlev AG, Di Giovanni S, Wang G, Liu W, Identification and biochemical characteriza- mons JW, Ruppert JM, Hamilton SR, Preising- Stoica B, Faden AI: BOK and NOXA are essen- tion of a novel transcription elongation factor, er AC, Thomas G, Kinzler KW, et al: Identifi- tial mediators of p53-dependent apoptosis. J

elongin a3. J Biol Chem 2002; 277: 26444–26451. cation of a chromosome 18q gene that is al- Biol Chem 2004; 279: 28367–28374.

10 Choi HJ, Lee H, Kim H, Kwon JE, Kang HJ, tered in colorectal cancers. Science 1990; 247: 37 Mancini M, Toker A: NFAT proteins: emerg- You KT, Rhee H, Noh SH, Paik YK, Hyung WJ, 49–56. ing roles in cancer progression. Nat Rev Can-

Kim H: MicroRNA expression profile of gas- 25 Miyamoto H, Shuin T, Ikeda I, Hosaka M, cer 2009; 9: 810–820. trointestinal stromal tumors is distinguished Kubota Y: Loss of heterozygosity at the p53, 38 Delgado Verdugo A, Crona J, Maharjan R, by 14q loss and anatomic site. Int J Cancer RB, DCC and APC tumor suppressor gene loci Hellman P, Westin G, Bjorklund P: Exome se-

2010; 126: 1640–1650. in human bladder cancer. J Urol 1996; 155: quencing and CNV analysis on chromosome 11 Cunningham JL, de Stahl TD, Sjoblom T, Wes- 1444–1447. 18 in small intestinal neuroendocrine tumors:

tin G, Dumanski JP, Janson ET: Common 26 Reyes-Mugica M, Rieger-Christ K, Ohgaki H, ruling out a suspect? Horm Metab Res 2015; 47: pathogenetic mechanism involving human Ekstrand BC, Helie M, Kleinman G, Yahanda 452–455. chromosome 18 in familial and sporadic ileal A, Fearon ER, Kleihues P, Reale MA: Loss of 39 Ruebel K, Leontovich AA, Stilling GA, Zhang carcinoid tumors. Genes Chromosomes Can- DCC expression and glioma progression. Can- S, Righi A, Jin L, Lloyd RV: MicroRNA expres-

cer 2011; 50: 82–94. cer Res 1997; 57: 382–386. sion in ileal carcinoid tumors: downregulation 12 Lollgen RM, Hessman O, Szabo E, Westin G, 27 Mehlen P, Rabizadeh S, Snipas SJ, Assa-Munt of microRNA-133a with tumor progression.

Akerstrom G: Chromosome 18 deletions are N, Salvesen GS, Bredesen DE: The DCC gene Mod Pathol 2010; 23: 367–375. common events in classical midgut carcinoid product induces apoptosis by a mechanism re- 40 Li SC, Essaghir A, Martijn C, Lloyd RV, De-

tumors. Int J Cancer 2001; 92: 812–815. quiring receptor proteolysis. Nature 1998; 395: moulin JB, Oberg K, Giandomenico V: Glob- 13 Wang GG, Yao JC, Worah S, White JA, Luna 801–804. al microRNA profiling of well-differentiated R, Wu TT, Hamilton SR, Rashid A: Compari- 28 Li SS, Yu SL, Singh S: Epigenetic states and ex- small intestinal neuroendocrine tumors. Mod

son of genetic alterations in neuroendocrine pression of imprinted genes in human embry- Pathol 2013; 26: 685–696.

tumors: frequent loss of chromosome 18 in il- onic stem cells. World J Stem Cells 2010; 2: 97– 41 Alberici P, Gaspar C, Franken P, Gorski MM,

eal carcinoid tumors. Mod Pathol 2005; 18: 102. de Vries I, Scott RJ, Ristimaki A, Aaltonen LA, 1079–1087. 29 Edfeldt K, Ahmad T, Akerstrom G, Janson ET, Fodde R: Smad4 haploinsufficiency: a matter

14 Nilsson O: Profiling of ileal carcinoids. Neuro- Hellman P, Stalberg P, Bjorklund P, Westin G: of dosage. Pathogenetics 2008; 1: 2.

endocrinology 2013; 97: 7–18. TCEB3C a putative tumor suppressor gene of 42 Solimini NL, Xu Q, Mermel CH, Liang AC, 15 Heldin CH, Moustakas A: Role of Smads in small intestinal neuroendocrine tumors. En- Schlabach MR, Luo J, Burrows AE, Anselmo

TGFβ signaling. Cell Tissue Res 2012; 347: 21– docr Relat Cancer 2014; 21: 275–284. AN, Bredemeyer AL, Li MZ, Beroukhim R, 36. 30 Zukerberg LR, Patrick GN, Nikolic M, Hum- Meyerson M, Elledge SJ: Recurrent hemizy- 16 Hahn SA, Schutte M, Hoque A, Moskaluk CA, bert S, Wu CL, Lanier LM, Gertler FB, Vidal M, gous deletions in cancers may optimize prolif-

daCosta LT, Rozenblum E, Weinstein CL, Van Etten RA, Tsai LH: Cables links Cdk5 and erative potential. Science 2012; 337: 104–109.

312 Neuroendocrinology 2017;104:302–312 Nieser et al. DOI: 10.1159/000446917