Proteomics Suggests a Role for APC-Survivin in Response to Somatostatin Analog Treatment of Neuroendocrine Tumors.

CONTEXT: Somatostatin analogs are established in the treatment of neuroendocrine tumors (NETs) including small intestinal NET; however, the molecular mechanisms are not well known. Here, we examined the direct effects of lanreotide in NET cell line models. SETTING AND
DESIGN: The cell lines HC45 and H727 were treated with 10nM lanreotide for different time periods and alterations of the proteome were analyzed by in-depth high-resolution isoelectric focusing tandem liquid chromatography-mass spectrometry. We next investigated whether the observed suppression of survivin was mediated by adenomatous polyposis coli (APC) and possible effects on tumor proliferation in vitro. Expression of survivin was assessed by immunohistochemistry in 112 NET cases and compared with patient outcome.
RESULTS: We quantified 6451 and 7801 proteins in HC45 and H727, respectively. After short time lanreotide treatment APC was increased and survivin reduced. Overexpression of APC in H727 cells decreased, and APC knock-down elevated the survivin level. The lanreotide regulation of APC-survivin could be suppressed by small interfering RNA against somatostatin receptor 2. Although lanreotide only gave slight inhibition of proliferation, targeting of survivin with the small molecule YM155 dramatically reduced proliferation. Moderate or high as compared with low or absent total survivin expression was associated with shorter progression-free survival, independent of tumor stage, grade, and localization.
CONCLUSIONS: We report a proteome-wide analysis of changes in response to lanreotide in NET cell lines. This analysis suggests a connection between somatostatin analog, APC, and survivin levels. Survivin is a possible prognostic factor and a new potential therapeutic target in NETs.

Neuroendocrine tumors (NETs) of the gastrointestinal and bronchial tracts, commonly known as carcinoids, show increasing incidence and prevalence in recent years. Small intestinal NET (SI-NET) is the most common type of NET (1).

NETs express somatostatin receptors (SSTRs), and the somatostatin analogs (SSAs) lanreotide and octreotide established in the treatment of NETs effectively relieve hormonal symptoms and suppress tumor growth (2–4). Further, in a multinational study the antiproliferative efficacy of lanreotide was demonstrated for nonfunctioning NETs (4). However, the molecular mechanism behind the antiproliferative effects of SSAs on NETs are not fully elucidated (5).

Both direct and indirect antiproliferative effects of SSAs have been studied in in vitro models. In a pituitary acromegaly model for example, the phosphatidylinositol 3-phosphate kinase (PI3K)/protein kinase B (Akt)/zinc finger regulator of apoptosis and cell cycle arrest pathway was implicated as a direct antiproliferative effect of SSAs (6). By contrast, similar effects were not observed in the NET cell lines BON and H727 applying SSAs at pharmaceutical concentrations (7, 8), indicating that the signaling pathways affected by SSA are cell type dependent (5).

Here, we applied high-resolution isoelectric focusing tandem liquid chromatography-mass spectrometry (HiRIEF LC-MS/MS) (7) to determine the landscape of proteome changes induced by lanreotide at pharmacological concentration in NET cells. Expression of the dysregulated molecules adenomatous polyposis coli (APC) and survivin were further assessed for associations to patient outcome and in vitro effects on tumor phenotype.

Materials and Methods

NET cases

All cases were diagnosed in the routine clinical setting according to the World Health Organization (WHO) Classification criteria (8). Twenty formalin-fixed paraffin-embedded tissue samples with tumor tissue and adjacent normal epithelial tissue from 13 SI-NET patients were retrieved from the biobank at the Karolinska University Hospital and used for immunohistochemistry. Documented patient consent and ethical permission from the local ethical committee were available.

A tissue microarray (TMA) was generated representing 112 NET patients and used for immunohistochemistry. Surgical tissue samples obtained in the period between 1980 and 2012 were retrieved from the archive of the department of pathology, La Paz University Hospital. Tissue samples were fixed in 10% formalin and embedded in paraffin. The clinical details are summarized in Supplemental Table 1 for gender, age, tumor parameters (site, functionality, morphological grade, stage, T classification, and proliferation marker Ki-67), lymph node, and distant metastases and follow-up. All patients underwent surgery as the first treatment. Progression-free survival was defined as the period of time from the surgery until detection of recurrence, clinical progression of the baseline disease or death, and overall survival was defined as the period between the surgery and death or last follow-up. Approval of this study was obtained from the local Ethics Committee. TMA constructions were according to published procedures (9) and are detailed in Supplemental Materials and Methods.

Established human NET cell lines

HC45 and CNDT2 were established from liver metastases of SI-NETs. HC45 was kindly provided by Professor R. V. Lloyd at the Mayo Clinic Rochester (Rochester, MN) (10), and CNDT2 (11) was kindly provided by Professor L. M. Elis at the MD Anderson Cancer Center (Houston, TX). BON1 was derived from a pancreatic neuroendocrine carcinoma, and the neuroendocrine lung cancer cell line NCI-H727 was purchased from ATCC.

Culturing and genotypes are described in Supplemental Materials and Methods (12).

Primary SI-NET culture

A primary SI-NET cell culture was generated from a G1 grade (WHO 2010) lymph node metastasis of a 51-year-old female SI-NET patient with an informed consent and approval from the local Ethics Committee. The tumor tissue was minced with a scalpel in MEM and incubated in 700 μg/mL type II collagenase (Sigma-Aldrich) overnight followed by hyaluronidase (Sigma-Aldrich) for 10 minutes at 37°C. The enzyme was then removed and cells were cultured in DMEM with 10% fetal bovine serum at 37°C, 5% CO2 in a humidified incubator. Cells were passaged up to 10 times.

Real-time quantitative PCR (qRT-PCR)

Gene expression levels were quantified using qRT-PCR and commercially available TaqMan based assays and previously described methodology (12). TaqMan assays were applied for SSTR1 (Hs00265617_s1), SSTR2 (Hs00990356_m1 and Hs00265624_s1), SSTR3 (Hs01066399_m1), SSTR4 (Hs01566620_s1), SSTR5 (Hs00265647_s1), and the endogenous control ACTB (Hs99999903_m1).

Immunocytochemistry and immunohistochemistry

The experimental procedures are detailed in Supplemental Materials and Methods. Immunocytochemistry was performed essentially as previously described (13) on cultured cells using primary antibodies for SSTR2 (Sigma-Aldrich) and SSTR5 (Epitomics). For immunohistochemistry antibodies against survivin (Baculoviral IAP Repeat Containing 5, BIRC5 D-8; Santa Cruz Biotechnology) and APC (EP701Y; Abcam) were used for individual tissue slides and against survivin (Ab469; Abcam) and APC (Ab40778; Abcam) for the TMA according to published methodology (14).

Immunohistochemical patterns detected for survivin include nuclear and cytoplasmic staining, with a diffuse pattern of expression. Nuclear staining intensity was classified as absent (0), weak (1), moderate (2), or high (3); the same approach was used for scoring of cytoplasmic staining intensity. An overall staining intensity score based on the sum of both intensities was implemented (total score 0–6). For APC, only absent, moderate, or high cytoplasmic expression was stratified.

Lanreotide treatments

HC45 and H727 were treated with 10nM lanreotide acetate (Sigma-Aldrich) for up to 48 hours. The medium was replaced by freshly prepared drug every 24 hours. Cells were harvested at 2, 6, or 48 hours after starting the treatment and untreated cells were used as controls. Cells were washed with PBS and trypsinized, and cell pellets were washed in ice cold PBS twice and used for HiRIEF LC-MS/MS and Western blot analysis immediately or kept at −80°C until use.

HiRIEF LC-MS/MS

Protein expression profiles were generated using the HiRIEF LC-MS/MS methodology essentially following previously reported experimental procedures (7). Detailed descriptions of protein extraction and digestion, peptide isoelectric focusing and LC-MS/MS analyses are given in the Supplemental Materials and Methods.

Western blot analysis

Proteins were extracted from NET cell lines, separated in SDS-PAGE, blocked in albumin or milk, and transferred to nitrocellulose membranes (Invitrogen), for incubations with primary antibodies (APC, bone morphogenic protein-binding endothelial regulator protein [BMPER], small integral membrane protein 21 [SMIM21], sperm-associated antigen 16 protein [SPAG16], C14orf14, Proto-oncogene tyrosine-protein kinase FYN, survivin, chromogranin A, and insulinoma-associated protein 1 [INSM1]) and membrane imaging as detailed in Supplemental Materials and Methods.

Proliferation assays

The Bromodeoxyuridine (BrdU) proliferation kit (Roche) and xCELLigence real-time proliferation assays were used to evaluate endpoint and real-time cell proliferation, respectively, following the manufacturer's instructions. Details are given in Supplemental Materials and Methods. Proliferation index was determined by immunocytochemistry using the clone MIB1 antibody (Dako) for detection of the Ki-67 antigen.

In vitro transfection experiments

APC and SSTR2 were knocked down in H727 cells. Lipofectamine 2000 or RNAiMAX was used for transfection of plasmids or small interfering RNAs (siRNAs), respectively, as detailed in Supplemental Materials and Methods.

Statistical analysis

Microsoft Excel or IBM SPSS statistics version 20.0 were used for statistical analyses as detailed in Supplemental Materials and Methods. For statistical analysis of the TMA results, the staining intensity values were dichotomized (absence and weak vs moderate and high) to optimize the survival analysis. Kaplan-Meier curves, log-rank test, and univariate Cox regression were used for the univariate survival analysis (progression-free survival and overall survival analysis), and Cox regression was used for the multivariate analysis. Statistical analyses were performed using IBM SPSS Statistics, version 20.0 (IBM Corp). All statistical tests were 2-sided, and nominal P < .05 was considered statistically significant.

Results

Expression of SSTR1–SSTR5 genes and SSTR2 and SSTR5 proteins in NET cell lines

Expression of all 5 SSTR genes SSTR1–SSTR5 was demonstrated in BON1 and H727 cells by qRT-PCR. SSTR1, SSTR2, SSTR4, and SSTR5 were detected in HC45, whereas CNDT2 only expressed SSTR2 and SSTR5 (Supplemental Figure 1). By immunocytochemistry, HC45, BON1, H727, and the positive control were found to express the SSTR2 and SSTR5 proteins (Supplemental Figure 1).

Protein profiling after treatment with lanreotide using HiRIEF LC-MS/MS

The HC45 and H727 cell lines were treated with lanreotide for 2, 6, and 48 hours, and the proteomes were compared between treated and nontreated cells. Totally, 6836 and 8430 proteins were detected in HC45 and H727 cell lines, respectively (1% False Discovery Rate). By PCA, no obvious outlier samples were observed supporting consistency between duplicates (Supplemental Figure 2). The relative expression of 6451 and 7801 proteins were significantly quantified in lanreotide-treated cells and untreated cells of HC45 and H727, respectively, with 5264 proteins being common between the 2 cell lines (Figure 1A and Supplemental Table 2). Taking all 3 time points into account, 747 proteins were significantly differentially expressed after the treatment in HC45 and 656 proteins in H727, of which 54 were common between the 2 cell lines (Supplemental Table 3). A subset of these proteins was classified as up- or down-regulated based on expression ratios in relation to the cut-offs (0.69 and 1.33 for HC45 and 0.78 and 1.22 for H727) (Supplemental Table 4).

Figure 1.

Summary of proteomics experiments and networks detected in the HC45 and H727 cell lines by MS analysis. A, Venn diagrams showing the total numbers of quantified proteins (left) and significantly differentially expressed proteins at any time point after lanreotide treatment (right). B, DAPPLE gene ontology analysis of differentially expressed proteins in HC45 at 2 and 48 hours of treatment with lanreotide. C, The PI3K/Akt and MAPK signaling pathways identified by Ingenuity in lanreotide-treated HC45 at 2 and 48 hours.

Few significant connections among altered proteins were discovered for H727 and HC45 using Disease Association Protein-Protein Link Evaluator (DAPPLE). In both cell lines, significant associations regarding cell cycle-related proteins were detected after 48 hours of lanreotide treatment, signified especially by overexpression of APC and survivin in HC45 cells (Figure 1B). Several transcription factors found as differentially expressed (Supplemental Table 3) were also detected as associated proteins, suggesting a common mode of dysregulation of master gene regulatory mechanisms. In addition, deregulated proteins in both cell lines displayed significant associations to the cytoskeleton proteins, possibly suggesting that networks of proteins involved in organization of the cytoskeleton are significantly affected by lanreotide treatment. Using Ingenuity, the PI3K/Akt and p38 MAPK signaling pathways were identified as perturbed in lanreotide-treated HC45 cells at 2 and 48 hours (Figure 1C). Furthermore, pathways associated with cell cycle and/or cell growth, proliferation, and interaction were among the top 5 significantly altered at all 3 time points of treatment of both cell lines (Supplemental Figure 2).

Verification of altered proteins by Western blotting

Eight proteins identified as up- or down-regulated after lanreotide treatments were selected for further analyses by Western blotting (Supplemental Table 4), including APC, BMPER, survivin (BIRC5), tyrosine-protein kinase Fyn (FYN), INSM1, SMIM21 (Chromosome 18 open reading frame 62, C18orf62), SPAG16, and uncharacterized protein Chromosome 14 open reading frame 142 (C14orf142). The selection among significantly altered proteins was based on the network/pathway (DAPPLE) analysis (BIRC5, BMPER, and APC), known protein functions such as cell cycle regulation (BIRC5, BMPER, FYN, and APC), proteins coded within frequently altered genomic locations in SI-NETs (SMIM21 and C14orf142), very high levels of fold change observed (SPAG16) and known NET markers (INSM1). Using Western blotting, expression of protein products of expected sizes were verified for all proteins in lanreotide-treated cells and untreated control HC45 and H727 cells (Supplemental Figure 3). In HC45 cells, increased band intensities were observed for BMPER and survivin at 6 and 48 hours and at 2 and 6 hours for FYN (Supplemental Figure 3). C14orf142 was found decreased at 48 hours and SMIM21 was increased at 6 hours (Supplemental Figure 3). Increased expression of both long and short isoforms of APC was detected at 2 and 6 hours in HC45 as well as H727 cells (Supplemental Figure 3). In H727, the expression of INSM1 was reduced at 48 hours, SPAG16 showed attenuation at the 71-kDa band intensities (Supplemental Figure 3).

Effects on proliferation by lanreotide and the survivin inhibitor YM155

The possible antiproliferative effect of lanreotide at different concentrations was evaluated in the HC45, H727, and BON1 cell lines. By time resolution using xCELLigence, a modest inhibition of proliferation was observed in HC45 and BON1 for 100μM lanreotide only (Supplemental Figure 4). Using the end-point BrdU assay, decreased proliferation was observed for BON1 at a concentration of 10μM lanreotide in serum-free condition (Figure 2). No antiproliferative effect of lanreotide was observed in the presence of serum (data not shown). No effects on MIB1 proliferation index was observed after 9 days of treatment of HC45 and H727 cells with 10nM or 1μM lanreotide as compared with nontreated cells (Supplemental Figure 4B).

Figure 2.

Proliferation analysis after treatment with lanreotide and survivin inhibition. Results from BrdU proliferation assay after lanreotide treatment at different concentrations are shown to the left and to the right for combined treatment with lanreotide and the survivin inhibitor YM155 for 48 hours at different concentrations. All optical density values were normalized and compared with untreated controls. Each bar represents the mean of 8 different replicates. The Western blotting below shows expression of survivin after YM155 treatment for 48 hours at different concentrations.

Targeting survivin with small molecule inhibitor YM155 at different concentrations reduced the levels of survivin (Figure 2) and led to a dramatic dose-dependent decrease in proliferation of HC45, H727, and BON1 cells. The antiproliferative effect of YM155 was significant from 5nM in HC45 and H727 and from 100nM in BON1 (Figure 2).

A primary SI-NET culture was analyzed in parallel. Expression of the SSTR2 and SSTR5 proteins (Supplemental Figure 1) and chromogranin A, a marker of neuroendocrine cells (15), was demonstrated in the primary SI-NET culture, thus confirming its neuroendocrine origin. Reduced proliferation of primary SI-NET was noted for YM155 from 5nM (Figure 2).

Regulation of APC-survivin by lanreotide

Increased expression of survivin was observed at the late time points of lanreotide treatment, but not at 2 hours when the APC expression was already augmented. Based on these observations, we hypothesized that lanreotide may increase APC and thereby regulate survivin and that such a regulation could then be circumvented by an adaptive negative feedback of survivin. To test this hypothesis, we observed survivin expression levels in H727 cells after lanreotide treatment and following modulation of APC expression. Treatment with 10nM lanreotide increased APC and reduced survivin at short time points (1–8 h) (Figure 3A). Overexpression of APC resulted in down-regulation of survivin (Figure 3B), and reduction of APC by small hairpin RNA against APC increased survivin expression (Figure 3C). To investigate whether this regulation is specifically connected to SSA-SSTR interaction, we suppressed SSTR2 using siRNA. This resulted in down-regulation of APC and abolishment of the lanreotide induced down-regulation of survivin at 2 hours (Figure 3D).

Figure 3.

In vitro modulation of APC and SSTR2 in H727 cells. A, Western blotting showing induction of APC and inhibition of survivin after lanreotide treatment for different time periods. B, Down-regulation of survivin after APC overexpression. C, Up-regulation of survivin after APC knockdown. D, siRNA suppression of SSTR2 down-regulates APC and blocks the inhibitory effect of lanreotide on survivin. B–D, Western blottings are shown to the left and bar diagrams corresponding to normalizations and quantifications against GAPDH (n = 3) to the right with indication of APC (white bars), survivin (dotted bars), and SSTR2 (filled bars). Each bar represents the mean of 3 different Western blottings. The antibody op44 was used for detection of APC.

Survivin expression is associated with poor survival of NET patients

The expression of survivin and APC was further analyzed by immunohistochemistry. Both proteins were first assessed in a cohort of 20 SI-NET samples from 13 patients. For surviving, both cytoplasmic and nuclear expression was increased in tumor cells as compared with adjacent normal epithelial cells (P < .001 and P = .007, respectively) (Supplemental Table 5). There was a correlation between cytoplasmic and nuclear staining intensities both in tumor and normal intestinal epithelium for survivin (P = .019, r = 0.546; P = .003, r = 0.660). For APC, nuclear and cytoplasmic staining was also detected; however, no difference was observed between normal epithelium and tumor tissue.

The expression of survivin was further evaluated in relation to clinical parameters and patient outcome using a TMA with 112 NET tumor cases (Supplemental Table 1). The result was evaluated separately for nuclear staining and total cytoplasmic and nuclear staining combined. Examples of the scoring are given in Figure 4A. In the univariate analysis, moderate to high staining for survivin was associated with shorter progression-free survival compared with absent or low staining using either nuclear staining or the overall score for total cytoplasmic an nuclear staining (P < .001 and P = .001, respectively) (Figure 4B). Similarly, moderate or high expression for total survivin was associated with shorter overall survival (P = .011) (Figure 4B). After 3 years of follow-up, 72% of cases with moderate or high expression remained disease free as compared with 98% in the group with low or absent expression. There was no association between the Ki-67 categories and total survivin expression (moderate to high vs absent or low; χ2 test P = .113). In multivariate Cox analyses, total survivin expression was shown to be an independent marker for shorter progression-free survival when adjusted to disease stage, morphological grade, Ki-67 proliferation index, and localization (hazardous ratio [HR], 4.1; 95% confidence interval [CI], 1.1–16.2; P = .045) (Table 1).

Figure 4.

Immunohistochemical analyses of survivin expression and survival analysis. A, Examples of survivin expression for SI-NETs included in the TMA. For each case is shown hematoxylin-eosin staining in overview and survivin staining in overview as well as in larger magnification. B, Kaplan-Meier plots showing progression-free survival (PGS) and overall survival (OS) for NET patients in relation to survivin expression determined by immunohistochemistry. Results for scoring of nuclear expression are shown at the top and for combined scoring of total nuclear and cytoplasmic staining below. The nuclear staining was classified as absent (0), weak (1), moderate (2), or high (3). For the total staining, a combined score for nuclear and cytoplasmic staining was applied with classification of absent or weak (0–2) and moderate or high (3–6) intensities. For each case are shown hematoxylin-eosin staining and survivin staining in overview (core biopsy 1 mm in diameter) as well as survivin staining in larger magnification (×400).

Table 1.

Results From Univariate and Multivariate Analysis of Total Survivin Expression by TMA Analyses of NET Cases

Parameter	Univariate Analyses			Multivariate Analyses[a]
	P Value	HR	95% CI	P Value	HR	95% CI
Disease-free survival
Total survivin (score 3–6)	.003 (ss)	6.3	1.9–21.5	.020 (ss)	6.3	1.3–30.2
Grade	<.001 (ss)			.444
Grade 1		1	NA		1	NA
Grade 2	<.001 (ss)	7.3	2.5–20.9	.579	1.4	0.4–5.1
Grade 3	.002 (ss)	6.5	2.0–21.6	.217	2.3	0.6–9.0
Ki-67 (as continuous variable)	.003 (ss)	2.6	1.4–4.9	.204	1.6	0.8–3.0
Stage	<.001 (ss)			.001 (ss)
Stage I		1	NA		1	NA
Stage II	.248	2.9	0.5–17.1	.380	2.3	0.4–14.0
Stage III	.013 (ss)	7.4	1.5–35.6	.029 (ss)	6.4	1.2–34.2
Stage IV	<.001 (ss)	35.7	7.4–171.4	.001 (ss)	21.6	3.7–126.0
Tumor localization
Gastroenteric vs pancreatic	.423	0.7	0.3–1.7	.749	0.8	0.3 − 2.3
Overall survival
Total survivin (score 3–6)	.021 (ss)	4.4	1.3–15.6	.641	1.9	0.3–11.9
Grade	<.001 (ss)			.001 (ss)
Grade 1		1	NA		1	NA
Grade 2	.066			.692	1.5	0.2–13.4
Grade 3	<.001 (ss)	4.8	0.9–25.1	<.001 (ss)	65.2	6.9–616.1
Ki-67 (as continuous variable)	<.001 (ss)	9.4	3.7–24.2	<.001 (ss)	9.8	2.7–34.8
Stage	.001 (ss)			.015
Stage I		1	NA		1	NA
Stage II	.979	1.0	0.1–11.4	.902	0.9	0.1–9.7
Stage III	.051	5.1	1.0–26.5	.654	1.5	0.2–10.5
Stage IV	.001 (ss)	18.0	3.5–91.7	.006 (ss)	19.0	2.3–156.3
Tumor localization
Gastroenteric vs pancreatic	.902	0.9	0.3–2.9	.377	1.9	0.4–8.4

ss, statistically significant; NA, not applicable.

Multivariate analyses adjusted by stage, grade, and tumor localization.

The performance of total survivin expression was also analyzed in NET patient subgroups. Moderate or high total survivin expression was associated with shorter progression-free survival in NET cases with low proliferation index less than or equal to 2% (P = .002; n = 58) as well as in morphological grade 1 NETs (P = .016; n = 68) (Figure 5A). As expected, higher morphological grade, stage, and proliferation index were each associated with shorter survival (Figure 5B).

Figure 5.

Kaplan-Meier plots for survival analyses in NET subgroups. A, Progression-free survival (PGS) and overall survival (OS) according to total surviving expression in NET subgroups with low Ki-67 proliferation index ≤ 2% as well as those with morphological grade 1. Total survivin staining intensities were scored as absent/weak (0–2) or as moderate/high (3–6). B, OS in NET subgroups according to morphological grade 1–3, stage I–IV, and Ki-67 proliferation index ≤2%, 3%–20%, and >20%.

Analyses of APC expression on the same TMA revealed no associations between APC staining and clinical parameters or patient survival (data not shown).

Discussion

In addition to their antisecretory effects, SSAs are also administered to NET patients due to their potential to inhibit tumor growth (3, 4). The antitumoral effects of SSAs have been attributed to direct antiproliferative effects in some tumor types such as pituitary adenomas as well as to indirect suppressive functions of somatostatin on the GH-IGF-1 axis (see references 18–20 below). In contrast to pituitary adenoma, there are controversial reports on the direct antiproliferative effects of SSAs on the NET cells (16). In this study, we assessed changes in the proteome of NET cells during lanreotide treatment. We quantified 6451 and 7801 proteins in HC45 and H727, respectively, and altered expression of 8 candidate proteins was verified by Western blotting, including APC, BMPER, FYN, SMIM21, survivin, and C14orf142 in HC45; and APC, INSM1, and SPAG16 in H727.

Both proteomics and Western blot analysis indicated that APC and specifically its short isoform were overexpressed in lanreotide-treated cells in a time-dependent manner in both studied cell lines. APC is a main component of the β-catenin suppression complex. It maintains β-catenin inactivated, thus preventing its nuclear accumulation and the following transcriptional activation of cell cycle promoting oncoproteins such as cMyc and other cyclin dependent kinases (17). The anticancer effect of APC was recently substantiated in an inducible APC-knockdown mouse model. In this model, restoration of APC led to reduction of tumor lesions in the small intestine and colon and regain of their stem cell function even in the presence of KRAS or TP53 mutations (18).

To our knowledge, this is the first report showing the effects on APC following modulation of SSTR by SSA. The lanreotide-induced survivin overexpression seemed to be halted at the first 2 hours concomitant with an increase in APC expression. We showed that survivin could be inhibited by APC overexpression in vitro in NET cells or induced by APC knockdown. This observation is in accordance with reports in colorectal cancer cell lines where APC-induced inhibition of survivin expression has been attributed to cancer stem cells in intestinal crypt cells (19). Nevertheless, this favorable outcome was reversed by continued treatment, suggesting an adaptive negative feedback imposed by survivin. A similar type of adaptive effect has been suggested to hinder the antiproliferative effect of antiandrogen agents in prostate cancer (20). Survivin-mediated drug resistance is well known (21). Survivin is an oncogene that is overexpressed in many tumor types including NETs (22–24). Survivin expression has been reported in SI-NETs (25), and in pulmonary NETs, its expression was associated with shorter survival (26). In gastroenteropancreatic NETs, nuclear expression of survivin was found to be of prognostic value in well differentiated neuroendocrine carcinomas (WHO, class 2) suggesting that it could be used as a prognostic marker in addition to Ki-67 (26). Previous NET studies based on pheochromocytoma/paraganglioma specimens, indicating survivin as a marker for chromaffin cell-derived tumors. However, no associations between survivin expression and outcome were identified in this tumor entity (27). We showed that inhibition of survivin by the small molecule YM155 reduced survivin levels in NET cell lines and that the antiproliferative effect was especially pronounced in HC45 cells. Similar findings were also obtained in the SI-NET primary cell culture reference, suggesting that the effects observed in established cell lines are also relevant for the short-term cultured primary tumor. Based on these observations a negative feedback loop on the SSA pathway and a prospective for survivin targeting in NETs may be proposed.

In SI-NET tumor samples, survivin showed increased expression in nuclei and cytoplasm of tumor cells as compared with normal adjacent epithelium. Furthermore, survivin expression was found to be associated with shorter progression-free and overall survival, and in multivariate analyses, total survivin remained a marker for progression-free survival independent of disease stage, morphological grade, Ki-67 proliferation index and tumor localization. Upon verification in independent tumors series, survivin expression may be applied as an additional biomarker in NET. Survivin could also be considered as a novel therapeutic target in NET. Indeed, YM155 is presently being evaluated in clinical trials for several cancer types (28, 29), and the current study suggests a rationale for clinical investigation of survivin targeting drugs in NETs.

In HC45, down-regulation of C14orf142 and up-regulation of SMIM21/C18orf62 was observed by HRIEF LC-MS/MS and was confirmed by Western blotting. C14orf142 is an uncharacterized protein encoded by a sequence on chromosomal region 14q, which is recurrently gained in SI-NETs (30). SMIM21 is located in 18q, the SI-NET deletion hotspot (30). BMPER has been proposed to have oncogenic properties (31). We detected higher expression of BMPER in lanreotide-treated HC45 cells by MS and Western blotting. Moreover, we also detected lower expression of INSM1 in lanreotide-treated H727 cells after 48 hours. INSM1 is a biomarker of lung NETs (32), and this demonstrates a potential for SSAs to control tumor characteristics of lung NET cells.

The ontology and pathway analysis indicated, apart from an interesting regulation of APC and survivin levels, alterations in the p38 MAPK and PI3K pathways. p38 is implicated in senescence (33). Senescence could be the case in SSA treatment of NETs, where tumor growth is arrested without a significant decrease in tumor size (2, 3). Besides, cell cycle, cell growth, proliferation, and interaction were always among the top 5 significantly altered functions in response to SSA treatment in NET cells.

In summary, this study provides in-depth quantitative proteomics dataset on cellular response to SSA treatment in NET. The proteomics data and functional validation studies indicates the existence of crosstalk between SSTR and APC-mediated control of survivin. The inhibition of survivin observed in the cell line model as well as the association between survivin expression and adverse outcome in NET patients indicate a role for survivin as a tumor driver. Based on the findings, survivin is a potential prognostic marker and suggested as a candidate new target for combination treatment with SSA in NETs.