Array comparative genomic hybridization
131 fresh-frozen SINET biopsies from 117 patients including 52 biopsies from 43 patients previously published (4) were analyzed with array comparative genomic hybridization (CGH). 109 of the 131 CGH analyzed biopsies were also represented on the tissue microarray used for immunohistochemistry and FISH. The purity of tumor biopsies was assessed by light microscopy using hematoxylin and eosin stained sections, and only biopsies with at least 70% tumor cells were analyzed. Genomic DNA was isolated from the biopsies using the DNeasy® Blood and Tissue kit (Qiagen GmbH, Hilden, Germany). The array CGH experiments, including labelling and hybridization of DNA were performed as previously described and according to the manufacturer’s protocol (SureTag DNA Labeling Kit; Agilent Technologies Inc., Palo Alto, CA, USA) (4). Array CGH analysis was performed using the human genome CGH microarray 4x44K (G4410B/G4426A/G4426B), 2x400K (G4448A), 244K (G4411B) and CGH-SNP microarray 4x180K (G4890A) (Agilent Technologies Inc.). The probes were annotated against genome build UCSC hg19. Hybridized slides were scanned on an Agilent High-Resolution Microarray Scanner followed by data extraction and normalization using Feature Extraction v.4.0.1.21 (protocol CytoCGH_0209_1x or 2x or 4x_Mar14) (Agilent Technologies). Data analysis was carried out using Agilent CytoGenomics 4.0.3.12 (Agilent Technologies Inc.). The ADM-2 (threshold = 6.0) algorithm was used to identify copy number aberrations across the genome. A log2 ratio of ± 0.25 and CNAs ≥ 500 kb was considered as a gain or loss. A log2 ratio of > 2 was designated as high-level amplification and a log2 ratio of < -2 was designated as homozygous loss. Chromosome arm gains/losses were determined using a 70% length threshold. The aberrations were checked manually to confirm the accuracy of the calls. Known recurrent germline copy number variations (Agilent female/male CNV reference) were excluded from the analysis.
Whole exome sequencing
DNA from 9 fresh-frozen tumors and their corresponding normal samples (small intestine, muscle, adipose tissue) were extracted using QIAamp DNA mini kit (Qiagen) and sequencing libraries were prepared using the SureSelect All Exon v3 target enrichment kit (Agilent Technologies, CA). The libraries were sequenced on an Illumina HiSeq 2000 in paired end mode with read length 100bp. After quality trimming using PRINSEQ (12), the reads were aligned to the human reference genome (hg19) using the BWA-MEM algorithm (http://bio-bwa.sourceforge.net/). Marking of duplicates and base recalibration was performed with Picard (https://broadinstitute.github.io/picard/) and GATK (13), respectively. Single nucleotide variants and small insertions and deletions were called for the SMAD4 and CDKN1B genes using HaplotypeCaller (germline variants) and Mutect2 (somatic variants), as implemented in GATK v4.0.8. Only germline variants with a quality normalized by depth above five, and somatic variants annotated as PASS, were considered. Additionally, variants were required to have a population frequency below 0.10 in the 1,000 Genome project.
Tissue microarray (TMA)
The tissue microarray contained 846 tumor biopsies from 412 patients retrieved from patients who underwent surgery for SINET at Sahlgrenska University Hospital in the years 1986 to 2013. Details of the construction of the tissue microarray have previously been described (14). The diagnosis of all tumors was re-validated by staining all tumors for hematoxylin and eosin, synaptophysin and chromogranin A, and reviewed by board-certified pathologist (O.N.).
Immunohistochemistry of TMA and Smad4+/- mice
Immunohistochemistry was performed on SINET TMAs and on gastrointestinal tissue including duodenum, jejunum, ileum from wild-type Smad4+/+ and Smad4+/- mice. The Smad4+/- mice were derived from F5-F8 backcross generations to C57BL/6 of the original F1 129Ola/C57BL/6Jico SMAD4+/E6sad founder (15) and was kindly donated by Professor Riccardo Fodde. Sections (3–4 μm) from formalin-fixed and paraffin-embedded blocks were placed on glass slides and treated in Dako PT-Link using EnVision™ FLEX Target Retrieval Solution (high pH). The following primary antibodies were used: anti-SMAD4 (B-8, Santa Cruz), anti-p27 (DCS-72.F6, Abcam), anti-chromogranin A (PHE5, Chemicon and EP1030Y, Abcam), anti-synaptophysin (SP11, Abcam), anti-5HT (H209, Dako and LS-B7118, LSBio), and anti-Ki67 (MIB1; Dako). Immunohistochemical staining was performed in a Dako Autostainer Link using EnVision™ FLEX according to the manufacturer’s instructions (DakoCytomation). EnVision™ FLEX+ (LINKER) rabbit or mouse was used for all stainings. Positive and negative controls were included in each run. The fraction of Ki67-positive cells was estimated by manually counting 500–2000 tumor cells per sample, using printouts. For relative quantification of SMAD4 and CDKN1B expression, the total global staining intensity of tumor cells was assessed, meaning localized subcellular i.e. nuclear or cytoplasmatic staining intensity was not individually evaluated. A board-certified pathologist (O.N.) examined the TMA tissue samples using a light microscope with a 4× objective until samples could with a high degree of repeatability be subcategorized as either ‘None’. ‘Low’, or ‘High’.
Fluorescence in situ hybridization
Fluorescence in situ hybridization (FISH) was performed on 4 µm paraffin sections from the tissue microarray using dual-color fluorescent probes for SMAD4 and chromosome 18 centromeric region or CDKN1B and chromosome 12 centromeric region (FA0590, FA0305, Abnova). Pre-processing of paraffin sections, hybridization to the probe, post-hybridization washing and fluorescence detection were performed according to manufacturer’s instructions (Abnova). Tumors were examined using an Axioplan 2i epifluorescence microscope (Zeiss, Oberkochen, Germany). Within each section, normal regions/stromal elements served as the internal control to assess quality of hybridization. Cases were scored with using a 100× objective lens, counting at least three distinct areas and at least 30 discrete nuclei per area. Labeled nuclei were imaged using an LSM 780 confocal microscope (Zeiss, Oberkochen, Germany). Brightness/contrast adjustments, scale bar, and image type transformation were performed using ZEN 3.1.
Statistical analysis
Correlation between protein expression and copy number status was estimated with Spearman’s rank correlation coefficient. Comparison of the number of CgA-positive cells in the small intestinal crypts and crypt bases between Smad4+/- and Smad4+/+ mice was performed using unpaired two-sided Student’s t-test. Comparison of SMAD4 expression between tumor sites and patient disease stages was tested with Mann Whitney U-test.