Overexpression of Membrane Proteins in Primary and Metastatic
Gastrointestinal Neuroendocrine Tumors
Jennifer C. Carr, MD*1, Scott K. Sherman, MD*1, Donghong Wang, MS1, Fadi S. Dahdaleh, MD1,
Andrew M. Bellizzi, MD2, M. Sue O’Dorisio, MD,PhD3, Thomas M. O’Dorisio, MD4,
James R. Howe, MD1
1University of Iowa Carver College of Medicine, Department of Surgery
2University of Iowa Carver College of Medicine, Department of Pathology
3University of Iowa Carver College of Medicine, Department of Pediatrics
4University of Iowa Carver College of Medicine, Department of Internal Medicine
*These authors contributed equally to this work.
Presented at the Endocrine Parallel Session at the 66th Annual Cancer Symposium of the Society
of Surgical Oncology, National Harbor, MD, March 8, 2013
Revised version accepted for publication in Annals of Surgical Oncology, 9/17/2013
James R. Howe, M.D.
Department of Surgery
University of Iowa Hospitals and Clinics
200 Hawkins Drive
Iowa City, IA
Tel. (319) 356-1727
Fax. (319) 353-8940
Quantitative PCR in a large set of primary and metastatic small bowel and pancreatic
neuroendocrine tumors demonstrates that the oxytocin receptor (OXTR) is overexpressed 15-90
fold compared to normal tissue, making it a promising novel therapeutic target.
Background: Small bowel and pancreatic neuroendocrine tumors (SBNETs and PNETs) are
rare tumors whose incidence is increasing. Drugs targeting the somatostatin receptor are
beneficial in these tumors. To identify additional cell-surface targets, we recently found receptors
and membrane proteins with gene expression significantly different from adjacent normal tissues
in a small number of primary SBNETs and PNETs. We set out to validate these expression
differences in a large group of primary NETs, and to determine whether they are present in
corresponding liver and lymph node metastases.
Methods: Primary SBNETs and PNETs, normal tissue, nodal, and liver metastases were
collected and mRNA expression of six target genes was determined by quantitative PCR.
Expression was normalized to GAPDH and POLR2A internal controls, and differences as
compared to normal tissue were assessed by Welch’s t-test.
Results: Gene expression was determined in 45 primary PNETs with 20 nodal and 17 liver
metastases, and 51 SBNETs with 50 nodal and 29 liver metastases. Compared to normal tissue,
the oxytocin receptor (OXTR) showed significant overexpression in both primary and metastatic
SBNETs and PNETs. Significant overexpression was observed for MUC13 and MEP1B in
PNET primaries, and for GPR113 in primary SBNETs and their metastases. SCTR and ADORA1
were significantly underexpressed in PNETs and their metastases. OXTR protein-expression was
confirmed by immunohistochemistry.
Conclusions: OXTR is significantly overexpressed relative to normal tissue in primary SBNETs
and PNETs, and this overexpression is present in their liver and lymph node metastases, making
OXTR a promising target for imaging and therapeutic interventions. ??
Small bowel and pancreatic neuroendocrine tumors (SBNETs and PNETs) are rare
tumors with increasing incidence that present with metastases in over 50% of cases1, 2. While
surgery is the most effective treatment for these tumors, hormonal therapy with somatostatin
analogues (SSAs) can curtail symptoms, and is associated with significantly improved
progression-free survival3, 4. SSAs are synthetic derivatives of the endogenous hormone
somatostatin, and include octreotide, lanreotide, and pasireotide. They bind and activate one or
more of five human somatostatin receptor subtypes5. Although SSAs show efficacy in functional
and non-functional tumors and achieve stable disease in >80% of cases6, most patients eventually
progress and demonstrate increasing SSA resistance over time7. To address late treatment failure,
second-line SSAs have binding affinities broadened from the standard somatostatin receptor
subtype, SSTR2, to those not as well-recognized by first-line drugs, such as SSTR1, 3, and 55.
Yet, in a recent phase II trial, 88% of octreotide-resistant patients failed to improve after
treatment with pasireotide, a drug with expanded SSTR subtype affinity5. The diminishing
returns of new drugs targeting somatostatin receptors demonstrate that further improvement in
NET treatment requires novel cell-surface receptor targets.
An ideal receptor target would display features that underlie the success of somatostatin
receptor-based treatments: high receptor expression in tumor tissue with low expression in
background normal tissue. Such differential expression allows ligands binding the somatostatin
receptor to selectively localize to tumors. Distinct from the anti-proliferative effects achieved by
activating SSTRs, radioisotopes linked to SSAs use SSTRs to selectively accumulate at tumor
tissues, which permits SSTR-based radioimaging and peptide-receptor radionuclide treatment
(PRRT) of neuroendocrine tumors8-11.
Several potential target receptors were recently identified by our group based on early
experiments measuring gene expression in SBNETs and PNETs using g-protein-coupled-
receptor (GPCR) and exon microarrays12. In a limited number of primary tumors (n=26), these
arrays revealed significant overexpression of over 50 genes compared to normal tissues. While
these investigations aimed to identify genes with different expression in tumors of small bowel
versus pancreatic origin, the GPCR arrays’ demonstration of significant upregulation of the
SSTR2 receptor in both SBNETs and PNETs led us to hypothesize that these data could point to
additional receptors useful for NET imaging and therapy. We further hypothesized that due to
variation in expression of individual genes across tumor specimens, it would be necessary to test
expression in a large sample of primary tumors to ensure validity. Finally, for a gene target to be
clinically useful, metastatic tissues should have expression profiles similar to primary tumors.
We therefore set out to determine expression of six target genes identified from our pilot studies
across a much larger group of primary tumor specimens and their associated metastases.
Patients and Tumors
Tumors, adjacent normal tissue, lymph node, and liver metastases were collected at
surgery under an IRB-approved protocol with informed consent. Tissues were preserved in
RNAlater solution (Life Technologies, Grand Island, NY). RNA was recovered and quantitative
PCR (qPCR) performed as described12. Briefly, total RNA was recovered by the Trizol method,
reverse-transcribed into cDNA, and triplicate qPCR was performed on a StepOnePlus Real-Time
PCR or 7900 HT-Fast RT-PCR System using Taqman primers and reagents (Life Technologies).
Target genes were chosen from pilot GPCR and exon expression array experiments as
described12, and included G-protein-coupled-receptor 113 (GPR113; Hs00542378_m1); oxytocin
receptor (OXTR; Hs00168573_m1); secretin receptor (SCTR; Hs01085380_m1); adenosine-A1
receptor (ADORA1; Hs00379752_m1); meprin-A-beta receptor (MEP1B; Hs00195535_m1); and
mucin-13, cell-surface-associated protein (MUC13; Hs00217230_m1). Glyceraldehyde-3-
phosphate dehydrogenase (GAPDH; Hs02758991_g1) and polymerase (RNA) II polypeptide-A
(POLR2A; Hs00172187_m1) served as internal control genes.
Mean threshold cycles (Ct) for each target were normalized to expression of internal
control genes. Over/underexpression was determined by the ddCt method (ddCtGene= primary or
metastatic tumor expression minus normal small bowel or pancreas tissue expression)13. Fold-
changes were calculated as 2-ddCt. Welch’s t-test compared mean ddCts with significance at
p<0.01 due to multiple comparisons (R v.2.15.2, Vienna, Austria).
IHC was performed using goat polyclonal antibody raised against a C-terminus human
OXTR (#sc-8102, Santa Cruz Biotechnology, Santa Cruz, CA). 4-mm sections were
deparaffinized, rehydrated, and subjected to heat-induced epitope retrieval at 125°C for 5
minutes. After incubation with primary antibody for 60 minutes at room temperature, the Dako
Envision Kit (Dako, Carpinteria, CA) was used for detection and slides were lightly
counterstained with hematoxylin. Slides were scored as 0 (no staining) 1+ (faint/barely
perceptible), 2+ (moderate), or 3+ (strong) by our Pathologist (AMB).
Gene Expression in Primary Tumors
Quantitative PCR was performed in 51 primary SBNETs (with 29 liver and 50 nodal
metastases), and 45 primary PNETs (with 17 liver and 20 nodal metastases) and their adjacent
normal tissues. In primary SBNETs, four genes showed significantly different expression in
tumors compared to normal tissues (Table I, p<0.01). SCTR and MEP1B were significantly
underexpressed in primary SBNETs with -2.7-fold and -5.7-fold lower expression, respectively.
Significant overexpression of 9.9-fold and 90.5-fold was found for GPR113 and OXTR. This
dramatic overexpression of GPR113 and OXTR mRNA in primary tumors compared to normal
background tissues revealed these two receptors as promising targets in SBNETs.
In primary PNETs, five genes had significantly different expression levels compared to
normal tissues (Table II, p<0.01). ADORA1 and SCTR showed significant underexpression in
primary PNETs, with -9.5 and -23.1-fold lower expression, respectively. Unlike SBNETs, in
PNET primary tumors, GPR113 showed no significant expression difference compared to
normal tissues (2.2 fold, p=0.012). MEP1B, MUC13, and OXTR were significantly
overexpressed in PNET primary tumors, with 21.9, 6.9, and 15.2-fold increased expression
compared to normal tissues, respectively. From these levels of overexpression we conclude that
MEP1B, MUC13, and OXTR are encouraging gene targets in PNET primary tumors. OXTR’s
high fold-overexpression and overexpression in both SBNET and PNET primary tumors makes it
the most attractive therapeutic target of the group.
Expression in metastases
Expression compared to normal tissues of these six target genes was measured in nodal
and liver metastases associated with these primary tumors. The significant underexpression
observed for ADORA1 and SCTR in PNET primary tumors was also present in PNET nodal and
liver metastases, while underexpression of SCTR and MEP1B in SBNET primary tumors was
found in SBNET nodal metastases, and in liver metastases for MEP1B (Table I and II).
However, for a cell surface molecule to function as a selective marker of tumor tissue,
thereby making it useful for imaging or PRRT, overexpression in metastatic as well as primary
tumor tissues is of greater clinical utility than underexpression. In the genes identified as possible
therapeutic targets, overexpression found in primary tumors was present in metastatic tissues as
well, and was sometimes more pronounced (Tables I and II). Overall, trends of over or
underexpression found in primary tumors were present in metastases for all target genes studied.
In SBNETs, GPR113 was significantly overexpressed in both liver and nodal metastases (31.6
and 109.1-fold, p<0.0001) compared to normal tissues (Fig. 1). In PNETs, MEP1B and MUC13
were significantly overexpressed compared to normal tissues in liver metastases (11.2 and 5.5-
fold, p<0.01). Their 9.4 and 5.5-fold overexpression in nodal metastases did not reach
significance (Fig. 2). OXTR was markedly overexpressed in liver and nodal metastases of both
tumor types, with over 15-fold overexpression in PNET metastases and over 90-fold
overexpression in SBNET metastases (Figs. 1 and 2, p<0.0001). Significant OXTR
overexpression compared to normal tissue in primary tumors, liver, and lymph node metastases
of both SBNETs and PNETs demonstrates its promise as a cell-surface receptor target for novel
therapeutic strategies in these tumors.
Evaluation of a formula to distinguish SBNET and PNET primary tumors
Neuroendocrine tumors often present with liver metastases of unknown primary 2, 12, 14,
and differences in gene expression between SBNETs and PNETs might be useful in diagnosing
the source of the primary. Observing that SBNETs tended to have a greater difference than
PNETs in expression of OXTR and SCTR, our group previously devised the formula 2(Ct SCTR – Ct
OXTR) to distinguish tissue samples’ primary site based on Ct expression levels of these genes12.
Under this formula, a value of >20 indicates an SBNET, while <5 indicates a PNET, with
intervening values called indeterminate. In early experiments, this formula correctly classified
22/26 (84.6%) primary tumors and 8/10 liver metastases12, therefore we sought to validate this
formula with this larger data set.
In 90 primary tumors with complete data, this formula correctly classified 62 tumors
(68.9%) and incorrectly classified 13 (14.4%) with 17 tumors (18.9%) being called
indeterminate. In 45 liver metastases, 32 were correctly classified (71.1%), while 5 (11.1%) were
incorrectly classified with 8 (17.8%) indeterminate. The formula performed slightly better in
PNET liver metastases (13/17, 76.5% correct), than SBNET liver metastases (19/28, 67.9%
correct). From these results, we conclude that these genes can assist in discriminating
neuroendocrine liver metastases of unknown primary source.
To verify expression of OXTR protein in NETs, we performed IHC on 7 primary
SBNETs and 7 PNETs (Figure 3). All tumors demonstrated OXTR staining, with more
pronounced staining in PNETs than in SBNETs (PNETs: four 3+, two 2+, one 1+; SBNETs: six
2+, one 1+). These results confirm that OXTR protein is present in these tumors, as suggested by
In this study we identified target genes with overexpression compared to normal
background tissue in 96 SBNET and PNET primaries, and found that these patterns of gene
expression are maintained in 117 nodal and liver metastases. Of these novel target genes, OXTR
is the most promising due to its high level of overexpression in primary tumors, nodal, and liver
metastases in both SBNET and PNETs. We furthermore found that a formula to determine the
primary site of neuroendocrine liver metastases based on expression of OXTR and SCTR is more
than 70% accurate, and we verified the presence of protein in 14 primary SBNETs and PNETs.
The oxytocin receptor’s overexpression compared to normal tissues suggests that it may be a
useful receptor target for imaging and therapeutic strategies in NETs.
The oxytocin receptor is a 389-amino acid G-protein-coupled-receptor, and is activated
by the hormone oxytocin15. Initially recognized for its role during parturition to stimulate uterine
contraction and lactation15, OXTR has more recently been investigated for its effects on social
behavior, including trust and bonding responses, and in autism-spectrum and anxiety disorders16.
A number of malignant tissues express OXTR, including cancers of the breast, brain,
reproductive system, colon, and lung17, 18.
The nine-residue oxytocin peptide was first synthesized in 1954 and is widely used in
obstetrics to promote labor, while OXTR-antagonists such as atosiban serve as tocolytics15.
Multiple drugs binding OXTR are available, making study of the receptor’s effect on tumor cells
possible15. The effects of OXTR ligands vary by cellular context. Oxytocin promotes
proliferation and migration of OXTR-expressing prostate cancer cell lines PC3 and PC3M19, and
proliferation of small cell lung cancer cell lines DMS79, H146, and H34520. In breast tumor-
derived endothelial cells, treatment with oxytocin increases growth and migration21. In contrast,
oxytocin causes growth inhibition in glial cells and neoplastic nerve tissues22, certain breast
cancers, endometrial tumors, and osteosarcoma cells23. An explanation for these opposite effects
in different cell types rests with OXTR’s ability to couple with multiple g-proteins, leading to
activation of different signal cascades in different settings24.
Consistent with this model, OXTR signaling can modulate multiple downstream
pathways, including phospholipase C, Map-kinase, and the phosphatidylinositol-3-kinase
(PI3K)/Akt pathways18. In the colon cancer cell line Caco2BB, the PI3K/Akt response to
oxytocin treatment is dose- and time-dependent. Whereas low-dose oxytocin treatment causes
increased PI3K signaling and phospho-Akt, with higher concentrations and longer treatments,
phospho-Akt decreases18. These effects depend on the particular g-protein present and on
receptor internalization at high oxytocin concentrations18. Downregulation of mTORC1 and an
inhibitory effect on translation in response to oxytocin stimulation also occurs25. While
oxytocin’s role is not straightforward in these cells, it is notable that the PI3K/Akt and mTOR
pathways are inhibited by OXTR in gut cells, as these pathways are important in NETs. The
PI3K/Akt pathway has been proposed as a pharmacologic target in NETs based on the
observation that blocking the PI3K pathway in pulmonary carcinoid tumors causes reduced
growth26. Similarly, the mTOR inhibitor everolimus has activity against bronchial carcinoids,
and prolongs progression-free survival in PNETs27, 28. Although the pleiotropic effects of OXTR-
stimulation make prediction of response in SBNETs and PNETs to oxytocin difficult, such
evidence linking OXTR ligand binding to PI3K/Akt and mTOR inhibition, coupled with our
finding of its overexpression in NETs, makes OXTR an exciting therapeutic target. Determining
the effects of oxytocin on cultured NET cells will be the next step in evaluating this further.
Detectable mRNA by qPCR does not guarantee a translated protein, evidence from our
study and others suggests that OXTR is present in these NETs. Welch et al. found that tissues in
the rat enteric nervous system produce oxytocin and express OXTR by RT-PCR, which was
confirmed by immunohistochemistry29. Interestingly, OXTRs are widely expressed in intestinal
villi of newborn rats, but by soon after birth, only cells clustering at the crypt-villus junction
have OXTR staining29. These OXTR+ cells reside in the expected location of the
enterochromaffin cells from which NETs originate. Together, these results demonstrate a role for
oxytocin and its receptor in the gut, suggest that OXTR expression may be specific to
neuroendocrine tissues, and support that OXTR protein is present in NET tissues as suggested by
qPCR. Our IHC in 7 SBNETs and 7 PNETs shows positive OXTR immunostaining in all tumors
examined. That SBNETs, which have higher mRNA expression by qPCR, had less staining than
PNETs raises the possibility that additional regulatory mechanisms might modulate OXTR
protein levels in these tumors30, but positive IHC staining in all tumor tissues studied validates
our conclusion from qPCR that OXTR is overexpressed compared to background tissue in these
Expression of OXTR by other tissues and the affinity of oxytocin analogues for
vasopressin receptors present a potential problem for OXTR-directed drugs15, 23. We have
demonstrated OXTR overexpression in NETs, but it is conceivable that OXTR or vasopressin
receptor expression in other tissues could cause false-positive imaging results, or bystander
tissue effects with PRRT. Although we did not measure gene expression in normal GI tissues
other than small bowel and pancreas, Roth et al. studied 45 different tissue types, and found
underexpression of OXTR in normal liver, colon, stomach and spleen, as well as significant
overexpression in brain, bronchus, and female reproductive tissues31. Low OXTR expression in
normal abdominal tissues, with high expression in NET tumors supports the potential of OXTR-
based tumor imaging. Furthermore, Chini et al. successfully imaged OXTR+ breast tumors in
mice with an [111In]-DOTA-oxytocin analogue17. They noted that their oxytocin analogue can
accommodate higher-energy PRRT ligands such as [90Y] and [177Lu], similar to DOTA-
octreotide analogues used in NETs, and demonstrated that OXTRs are internalized after ligand
binding17. Successful tumor imaging with an oxytocin analogue, and internalization of the
receptor, facilitating radiation delivery directly to the tumor, lends further impetus to
development of oxytocin analogues for NET treatment.
At presentation, 50-85% of SBNETs and PNETs have liver metastases and liver
metastases of unknown primary remain a clinical problem2. In one single-institution series, a
primary tumor could not be identified by imaging prior to surgery in 14% of patients with NET
liver metastases14. In this study, we confirmed our earlier observation that differences in SCTR
and OXTR expression can help distinguish between SBNETs and PNETs12, however, the
accuracy for liver metastases of 71% in this validation set was lower than the 80% we reported in
our 10 original metastases12. This again highlights the importance of validating gene expression
findings, and while this accuracy may be insufficient to influence clinical decisions at this time,
we continue to identify additional informative genes, which may improve our model’s
Two strengths of our study are its large sample size and inclusion of metastases. Gene
expression data is susceptible to effects of outlier measurements in small data sets, and low
sample sizes are a limitation of most NET gene expression studies32-35. For this reason, our
earlier study, with only 15 primary PNETs, failed to identify OXTR overexpression as a feature
of PNETs12, despite its 15-fold overexpression in the present study. While investigations with
small numbers of tumors are useful first steps towards identifying interesting gene targets, the
potential of outlying genes to give both false-positive and false-negative results makes
confirmation of such findings in larger data sets essential.
Examining metastases is likewise necessary prior to developing treatments targeting
novel receptors. It is increasingly understood that in most cancers, metastases differ from
primary tumors by mutations in only a few key genes36, yet if these or other “passenger”
mutations altered the expression of our target genes in nodal and liver metastases, then these
targets might not be useful. Our findings confirm that OXTR overexpression in primary tumors is
also present in corresponding metastases, meaning that OXTR-directed therapeutics would also
be expected to be effective at distant sites of disease.
We gratefully acknowledge our patients’ generosity.
Supported by NIH-5T32-CA148062-03 (JCC, SKS).
?? Download full-text
Table I – SBNET gene expression relative to normal tissue. Listed are expression fold-changes
and p-values compared to normal tissue.
*Indicates p<0.01 vs. normal tissue
Table II – PNET gene expression relative to normal tissue. Listed are expression fold-changes
and p-values compared to normal tissue.
*Indicates p<0.01 vs. normal tissue
Primary?? Tumors?? (n=51) Liver?? Metastases?? (n=29)?? Nodal?? Metastases?? (n=50)??
Primary?? Tumors?? (n=45) Liver?? Metastases?? (n=17)?? Nodal?? Metastases?? (n=20)??