Neurotensin Receptor 1 Is Expressed in Gastrointestinal Stromal Tumors but Not in Interstitial Cells of Cajal

Gastrointestinal stromal tumors (GIST) are thought to derive from the interstitial cells of Cajal (ICC) or an ICC precursor. Oncogenic mutations of the KIT or PDGFRA receptor tyrosine kinases are present in the majority of GIST, leading to ligand-independent activation of the intracellular signal transduction pathways. We previously investigated the gene expression profile in the murine Kit K641E GIST model and identified Ntsr1 mRNA, encoding the Neurotensin receptor 1, amongst the upregulated genes. Here we characterized Ntsr1 mRNA and protein expression in the murine Kit K641E GIST model and in tissue microarrays of human GIST. Ntsr1 mRNA upregulation in Kit K641E animals was confirmed by quantitative PCR. Ntsr1 immunoreactivity was not detected in the Kit positive ICC of WT mice, but was present in the Kit positive hyperplasia of Kit K641E mice. In the normal human gut, NTSR1 immunoreactivity was detected in myenteric neurons but not in KIT positive ICC. Two independent tissue microarrays, including a total of 97 GIST, revealed NTSR1 immunoreactivity in all specimens, including the KIT negative GIST with PDGFRA mutation. NTSR1 immunoreactivity exhibited nuclear, cytoplasmic or mixed patterns, which might relate to variable levels of NTSR1 activation. As studies using radio-labeled NTSR1 ligand analogues for whole body tumor imaging and for targeted therapeutic interventions have already been reported, this study opens new perspectives for similar approaches in GIST.


Introduction
Gastrointestinal stromal tumors (GIST) are the most common sarcoma of the gastrointestinal tract. They are thought to derive from the interstitial cells of Cajal (ICC) or an ICC precursor. Approximately 85% of GIST harbor oncogenic KIT mutations and 7% contain oncogenic platelet-derived growth factor receptor alpha (PDGFRA) mutations. KIT and PDGFRA are members of the type III receptor tyrosine kinase family. Oncogenic mutation of KIT or PDGFRA lead to constitutive autophosphorylation, ligand-independent activation of the downstream signal transduction pathways and subsequent deregulation of cell proliferation, survival and migration [1,2].
Since the release of imatinib mesylate (a.k.a. STI571, Gleevec TM ), an inhibitor of KIT and PDGFRA activation, the immunohistochemical demonstration of KIT immunoreactivity (-ir) has been essential for the diagnosis of GIST [3]. However, up to 15% of GIST show weak or negative staining for KIT [4]. Many of these ''KIT-negative'' GIST harbor PDGFRA mutations and may respond to imatinib treatment [5,6], hence need to be classified correctly. Recently, Discovered on GIST-1 (DOG1, a.k.a anoctamin, ANO1), Protein kinase C theta (PKC theta) and Carbonic anhydrase II (CAII) have been suggested as additional diagnostic markers and/or putative therapeutic targets [7][8][9][10]. Noteworthy, these genes are also expressed in normal KIT-ir ICC [7,10,11] and belong to the gene expression profile of KIT-ir ICC in the mouse small intestine [12]. Their presence in GIST thus likely reflects the fact that these tumors arise from the lineage of KIT-ir ICC.
We have recently identified genes expressed in GIST, but not in normal KIT-ir ICC, by comparing gene expression profiles of the antrum of transgenic mice harboring the oncogenic Kit K641E mutation [13] and WT littermates, using cDNA microarray. The G-protein-coupled receptor (GPCR), Neurotensin receptor 1 (Ntsr1) appeared highly upregulated in Kit K641E antrum [14], while it did not belong to the gene expression profile of normal Kit-ir ICC in the mouse small intestine [12]. As studies using radio-labeled NTSR1 ligand analogues have already been reported for whole body tumor imaging and for therapeutic interventions [15], NTSR1 appeared as an interesting prospect. Hence we have characterized Ntsr1 expression in the murine Kit K641E GIST model, in the normal human gut wall and in human GIST. NTSR1-ir was not detected in normal KIT-ir ICC, but was present in all GIST investigated, including KIT negative GIST harboring PDGFRA mutations.

Materials and Methods
The Kit K641E murine model of GIST A knock-in Kit K641E murine model of GIST used in this study has been previously described [13]. Two weeks old (P14) Kit K641E/K641E , Kit WT/K641E and Kit WT/WT littermates, as well as adult (3-6 month old) Kit WT/K641E and Kit WT/WT animals, were used. Mice were bred and experiments performed in accordance with the ethics committee for animal well-being of the Faculty of Medicine, Université Libre de Bruxelles, Brussels, Belgium.
RNA Extraction and PCR. Total RNA was extracted from mouse antrum using RNeasy Mini Kit (Qiagen, Valencia, CA) according to the manufacturer's instructions as described [14]. Genomic DNA was removed with the DNA-free TM kit (Ambion, Inc., Austin, Texas). RNA (1 mg) was reverse transcribed with 200 units of M-MLV Reverse Transcriptase (Invitrogen, Eugene, Oregon) in a reaction containing random primers (Amersham Bioscience, Piscataway, NJ). For reverse transcription PCR (RT-PCR) the cDNA reverse transcription product was amplified with specific primers (Table 1) using 1U Taq DNA polymerase (Qiagen, Valencia, CA) according to the manufacturer's instructions, during 40 cycles of 94uC for 30 sec, 57uC for 45 sec and 72uC for 45 sec. PCR products were separated on 2% agarose gel, stained with ethidium bromide and photographed under UV illumination. Real time quantitative PCR (qPCR) was performed as described [14], using SYBR Green chemistry on a 7500 Real time PCR system (Applied Biosystems, Foster City, CA). Identical conditions, namely 95uC for 10 min; then 40 cycles of 95uC for 15 sec and 60uC for 1 min were used for all primer sets (Table 1). Transcriptional quantification relative to Gapdh and b-actin housekeeping genes was performed using Qbase TM software [16]. Data were reported as the mean of three independent experiments and presented as normalized relative quantities (NRQ). Statistical analysis was performed in Microsoft Office Excel TM 2003 using unpaired Student's t-test. Differences were regarded as statistically significant when p#0.05.
Immunofluorescence staining (IF). Specimens were processed as described [14]. Briefly, slides were brought to RT, rinsed in 10mM Tris (Merck-Belgolabo, Overijse, Belgium) and 0.15M sodium chloride, pH 7.4 TBS, containing 0.1% (v/v) Triton-X 100 (TBS-TX), and incubated for 1 hour in 10% normal horse serum (NHS) (Hormonologie Laboratoire, Marloie, Belgium) and TBS-TX to reduce background staining. The slides were incubated overnight at RT with the primary antibodies  Human GIST tissue micro array (TMA) We examined NTSR1-ir in 2 independent cohorts of GIST using tissue microarray slides of formalin-fixed, paraffin-embedded (FFPE) material. As the clinico-pathological characteristics available for the two cohorts had not been recorded in a uniform way, they were analyzed separately.
For the Cleveland Clinic GIST TMA, a cohort of human GIST samples was retrieved from the anatomic pathology files of the Cleveland Clinic. The collection and analysis of human tissue samples was approved by the Cleveland Clinic Institutional Review Board. GIST samples were used to construct a tissue microarray with 1 mm cores using a TMarrayer TM (P/N 02110016) semi-automated tissue arrayer (Pathology Devices, Westminster, MD, USA) Risk assessment was performed using the criteria of Miettinen and Lasota [17] as recommended by the National Cancer Care Network [18]. Mutations in KIT exons 9, 11, 13, and 17 and PDGFRA exons 12 and 18 were examined as described previously [5]. Each case was present in duplicate on in this array. Clinico-pathological characterization of GIST used for Cleveland Clinic GIST TMA is summarized in Table 2.
SuperBiochips GIST TMA was purchased from Super BioChips laboratories, (#ADD1 , Seoul, South Korea). It contained 48 formalin-fixed and paraffin-embedded human GIST tissue specimens and 9 matched normal gut tissue specimens. According to the manufacturer, specimens (2 mm cores) were prepared from surgical specimens of American patients with GIST under guarantee in US law. Risk assessment was performed according to Fletcher et al. [3]. The clinicopathological characterization of this GIST cohort is summarized in Table 3.  Staining was regarded as positive when specific (i.e. above background) NTSR1-ir signal was present in $10% of tumor cells. The pattern of subcellular distribution of NTSR1-ir was recorded as: nuclear (without distinct cytoplasmic staining), cytoplasmic (without distinct nuclear staining), or mixed. x 2 (chi-squared), Fisher's exact test and Student's t-test, were used to test possible association between Ntsr1-ir and the clinico-pathological features of the tumors.

Ntsr1 expression in murine Kit K641E GIST model
Ntsr2 transcript was detected neither in Kit K641E/K641E homozygous P14 antrum nor in Kit WT/WT littermates. Relative Ntsr3 mRNA expression did not differ in Kit K641E/K641E homozygous P14 antrum compared to Kit WT/WT littermates (p = 0.33) (data not shown).
Ntsr1-ir is not present in WT ICC. Ntsr1-ir was observed in the myenteric plexus and intramuscular nerve fibers in P14 Kit WT/WT (Fig. 1C upper row), as well as in adult Kit WT/WT animals (Fig. S1). Conversely, ICC, identified by Kit-ir, were consistently negative for Ntsr1-ir in WT animals.
Ntsr1-ir is expressed in the hyperplastic layer of Kit-ir cells in KitK641E antrum. In the antrum of P14 Kit K641E/K641E animals, Ntsr1-ir decorated the Kit-ir hyperplastic layer (Fig. 1C bottom row), while in P14 Kit WT/K641E heterozygous antrum, Ntrs1-ir did not differ from P14 WT animals (data not shown). In adult heterozygous Kit WT/K641E animals, Ntsr1-ir was detected also in hyperplastic clusters of Kit-ir cells but not in individual Kit-ir ICC without sign of hyperplasia (Fig. S1). In all genotypes, myenteric plexus and intramuscular nerve fibers consistently exhibited Ntsr1-ir.

NTSR1 expression in human GIST
NTSR1-ir is not present in human KIT-ir ICC. In the normal human colon, NTSR1-ir was detected in MP but not in the adjacent KIT-ir ICC (Fig. 2 upper row). Omission of primary antibody completely wiped out the signal and background (nonspecific) signal in other cell types was minimal.
In a human GIST with KIT K642E mutation, NTSR1-ir was present only in the KIT-ir tumor cells (Fig. 2 bottom row).
NTSR1-ir is present in human GIST irrespective to the mutation status. Immunohistochemistry for NTSR1 was performed on two independent TMA including a total of 97 GIST. 95 samples were considered for analysis, 2 spots were excluded due to poor quality of tissue. Clinico-pathological characteristics of the GIST cases are summarized in Tables 2 and 3.
All GIST stained positively for NTSR1-ir (Fig. 3A left and  Fig. 3B). In adjacent non-tumoral tissue, NTSR1-ir was present in myenteric plexus but not in other cell types (Fig. 3A middle and  right, respectively). Background (non specific) signal in nontumoral tissue was minimal, confirming the specificity of the NTSR1-ir signal observed on TMA.
Noteworthy, 4 KIT negative GIST with PDGFRA mutation were positive for NTSR1 staining (Fig. 3B). All 4 cases were of gastric origin, 3/4 showed epithelioid morphology and belong to the low-moderate risk category (Table 4).
Altogether, no statistically significant correlation could be found between NTSR1-ir and the GIST clinico-pathological features.

NTSR1 -but not Neurotensin (NT) -is expressed in the human GIST882 cell line
In the human GIST882 cell line, NTSR1 expression was confirmed by RT-PCR (Fig. S2A). and WB (Fig. S2B) while transcript of neurotensin/neuromedin N precursor (NT/N) was not detected by RT-PCR. The human astrocytoma cell line 1321N1 served as positive control (Fig. S2A).
NTSR1-ir translocates to the nucleus upon agonist stimulation in GIST882 cells. After overnight starvation in FBS free medium (to avoid any NT that might be present in FBS), GIST882 cells exhibited diffuse perinuclear cytoplasmic, as well as nuclear, NTSR1-ir. Application of the NTSR1 agonist JMV449 (10 mM) for 4 hours markedly enhanced nuclear NTSR1-ir and redistributed cytoplasmic NTSR1-ir into dot-like clumps (Fig. S2C).

Discussion
To the best of our knowledge, the distribution of NTSR1-ir in the muscularis propria of the gut had not been previously reported. We observed NTSR1-ir in the soma of myenteric neurons and in nerve fibers in the muscle layers, in both mouse and human gut, while the KIT-ir ICC and other cell types were not labeled.
Conversely, NTSR1-ir was consistently detected in the hyperplastic Kit-ir cells in the murine Kit K641E GIST model and in all 95 human GIST investigated, irrespective of the presence of KIT-ir or KIT/PDGFRA mutation. The expression of NTSR1-ir in GIST, but not in normal ICC, contrasts with other markers recently reported in human GIST, which are generally also expressed in normal ICC [7,10,11].
We have identified three distinctive patterns of NTSR1-ir (nuclear; cytoplasmic; mixed) in GIST TMA but they could not be correlated with the recorded histopathological characteristic of the tumors. Nuclear localization of NTSR1 has been linked to cell responses evoked by intense and persistent agonist stimuli. [20][21][22][23]. NT, the natural NTSR1 high-affinity ligand, is a gut hormone normally present in serum and surges upon feeding [24,25]. In tissues, NT can also be produced by tumoral cells [26] or by reactive cells in their vicinity [22,[27][28][29][30], creating autocrine or paracrine activating loops, respectively. Hence, the variable proportions of  High affinity NT binding sites have been found in various cancers, including Ewing's sarcoma [31] and NTSR1 expression, driven by the Wnt/b-catenin pathway, is an early event in oncogenic transformation [22,23]. Sustained activation of NTSR1 leads to persistent activation of signaling pathways, including ERK1/2 [20], hence facilitating tumor growth and progression. NTSR1 has been associated with progression in lung, colon, pancreas, prostate and breast epithelial cancers [29,30,32]. According to the Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih. gov/geo/), NTSR1 is also expressed in a variety of sarcoma. Therefore, NTSR1 expression appears to be linked to tumorigenesis in general and cannot be used as a diagnostic tool to distinguish GIST from other tumor types or to predict their risk of malignancy.
Nevertheless, the identification of NTSR1 expression in GIST but not in the KIT expressing ICC cells from which they derive, offers clinical perspectives that deserve further consideration.
Firstly, NTSR1 can be targeted with radio-labeled NT analogues for whole body tumor imaging and possibly even for subsequent therapeutic interventions. Currently, 18F-Fluoro-deoxyglucose (18F-FDG) positron emission tomography (PET) is used for assessing tumor response to tyrosine kinase inhibitors in vivo [33][34][35]. However, 18F-FDG-PET cannot differentiate between GIST and other gastrointestinal malignancies including epithelial cancers and lymphoma and in some cases, 18F-FDG-PET fails to detect the relapse of GIST at an early time point [36]. An analogue based on NT (8-13) NT-XIX, has already been synthesized and characterized. Radioactivity clearance of NT-XIX analogue from healthy organs was faster than from tumors, resulting in improved tumor-totissue ratios and good SPECT/CT imaging [15]. NTSR1 in vivo imaging has already been successfully used in patients with ductal pancreatic adenocarcinoma [37]. Therefore NTSR1 in vivo imaging deserves full consideration in GIST.
NTSR1 may also be considered as a prospective direct therapeutic target. NT antagonists, such as SR48692.21, counteract the stimulatory effect of NT on tumor cells [22]. The influence of SR48692.21 on GIST proliferative characteristics thus deserves consideration.
Finally, NTSR1 could represent the gateway to introduce cytotoxic chemicals or expression vectors targeting the constitutively activated pathways into the cancerous cells [38][39][40].  In conclusion, further studies are needed to substantiate the potential of NTSR1 for clinical interventions in GIST. Figure S1 Ntsr1-ir is expressed in Kit-ir cell clusters in the antrum of adult heterozygous KitWT/K641E mice but not in WT ICC. Representative epifluorescence images of Ntsr1-ir distribution in the adult mouse antrum: Ntsr1-ir (NL559, red) was observed in Kit-ir (Alexa 488, green) cell clusters (insert) in adult heterozygous KitWT/K641E mice, but not in WT littermates. Ntsr1-ir was also consistently found in the myenteric plexus (*) and intramuscular nerve fibers both genotypes. Scale bars: 20 microns.