RETRACTED ARTICLE: Carboxypeptidase E: Elevated Expression Correlated with Tumor Growth and Metastasis in Pheochromocytomas and Other Cancers

Abstract

Expression of carboxypeptidase E (CPE), a prohormone processing enzyme in different cancer types, was analyzed from data in the GEO profile database (http://www.ncbi.nlm.nih.gov/geo/) and experimentally in pheochromocytomas. Analysis of microarray data demonstrated that significantly elevated levels of CPE mRNA was found in many metastatic non-endocrine cancers: cervical, colon rectal, renal cancers, Ewing sarcomas (bone cancer), and various types of astrocytomas and oligodendrogliomas, whereas expression of CPE mRNA was virtually absent in their respective counterpart normal tissues. Moreover, there was higher CPE mRNA expression in cells from the metastatic tumor compared to those from the primary tumor in colorectal cancer. Elevated CPE mRNA expression was found in neuroendocrine tumors in lung and pituitary adenomas, although the significance is unclear since endocrine and neuroendocrine cells normally express CPE. However, studies of neuroendocrine tumors, pheochromocytomas, revealed expression of not only wild-type CPE, but a variant which was correlated with tumor behavior. Extremely high CPE mRNA copy numbers of the variant were found in very large or invasive tumors, both of which usually indicate poor prognosis. Thus, collectively the data suggest that CPE may play a role in promoting tumor growth and invasion. CPE could potentially serve as a diagnostic and prognostic biomarker for metastasis in different cancer types.

Introduction

Carboxypeptidase E (CPE) is a prohormone processing exopeptidase and it also functions as a prohormone sorting receptor for the regulated secretory pathway (RSP) in endocrine and neuroendocrine cells (Fricker and Snyder 1983; Cool et al. 1997). Precursors of peptide hormones and neuropeptides are synthesized at the rough endoplasmic reticulum (RER) in (neuro)endocrine cells. These precursors are then processed sequentially in the secretory granules, first by prohormone convertases (PC1 and PC2), cleaving either in between or on the carboxyl side of paired-basic residue cleavage sites to yield basic residue-extended peptides (Steiner 1998). A subset of soluble CPE molecules in the secretory granules then cleaves the basic residues from these intermediates to generate mature peptide hormones and neuropeptides (Hook and Loh 1984). Another subset of CPE is the membrane form, which is anchored to cholesterol/sphingolipid-rich microdomains known as lipid rafts or detergent-resistant membranes, at the trans-Golgi network (TGN), and functions as a receptor for sorting prohormones away from other non-RSP proteins at the TGN, and into secretory granules in the RSP of pituitary (Shen and Loh 1997), and pancreatic β-cells (Dhanvantari et al. 2003), as well as in a neuronal cell line (Normant and Loh 1998). Additionally, the C-terminal tail of CPE drives bi-directional transport of BDNF vesicles to maintain vesicle homeostasis and secretion in hippocampal neurons by binding to microtubule motors via dynactin (Park et al. 2008). Thus, CPE plays many cell biological roles in the endocrine and nervous system. Hence, lack of CPE in mice with mutations of the CPE gene, or in CPE-knockout mice lead to many pathophysiological conditions, such as diabetes, infertility, obesity, low bone mineral density, and deficits in learning and memory (Coleman and Eicher 1990; Cawley et al. 2004, 2010; Naggert et al. 1995).

CPE in Non-Endocrine Cancers

While CPE has always been known to be expressed in endocrine cells and peptidergic neurons, it has now been found in epithelia-derived cancer cells as well, for example, in liver cancer cells (but not in normal liver cells) (Ge et al. 2005), although its function in tumor cells is unknown. The first evidence for the expression of CPE in cancerous cells was documented in human hepatoma (Hep G2) cells (Grimwood et al. 1989), although this form of CPE was slightly larger in size than that found in endocrine cells. Tumor cells of non-endocrine tissues such as breast cancer cells have also been shown to express CPE (Du et al. 2001). CPE is normally not expressed in cervical tissues, but during cancerous state it is highly expressed (Table 1a). Given that 70% of the cases of cervical cancer are correlated with human papillomavirus (HPV) infection (Walboomers et al. 1999), it would be interesting to determine if CPE expression is triggered by viral antigens in this pathological condition. CPE expression in colon is minimal when compared to other tissues, but during early onset of colorectal cancer (CRC) and during lymph node metastasis, CPE mRNA expression was substantially elevated (Table 1b, c). Interestingly, CPE expression was also significantly elevated in neoplastic clear cell sarcoma of the kidney compared to the normal fetal kidney cells (Table 1d). Surprisingly, CPE expression was down regulated in Wilms’ tumors, a rare kidney tumor found in children (Table 1e). Additionally, microarray studies have also convincingly shown that CPE mRNA expression was significantly elevated in renal clear cell tumors which have high metastatic potential, compared to the adjacent normal kidney obtained from the same patient (Table 1f). Bone marrow has no detectable CPE expression, but in Ewing sarcomas, malignant round-cell tumors in bone, CPE expression was elevated, however, this was in comparison with rhabdomyosarcomas, tumors derived from skeletal muscle which appear to have identical routine histology as Ewing sarcomas (Table 1o). CPE is expressed in brain, with the highest concentrations found in hippocampus, amygdala, and cortex (Liu et al. 2007). Interestingly, analysis of 12 primary brain glioma biopsies using cDNA microarrays revealed elevation of CPE expression compared to normal brain (Liu et al. 2007). In another study, cDNA microarray data from 50 human gliomas of various histogenesis also revealed elevation of CPE expression compared to normal brain (Table 1g–j).

Table 1.

CPE expression in various cancers

Serial no.	Cancer type	Sample type	State	Number of samples	CPE expression fold changeD	±SEM	P value		Geo profile ID
a	Cervical cancer	Biopsies	Metastatic	33	7.4	0.426	<0.003−22		GDS2416/CPE
b	Colorectal cancer	Biopsies	Metastatic	24	9.5	0.623	<0.0128		GDS2609/CPE
c	Colorectal cancer	Primary cells	Metastatic	6	7	1.020	<0.0007		GDS1780/CPE
dA	Clear cell sarcoma	Biopsies	Metastatic	14	2.49	0.260	<0.0001		GDS1282/CPE
eA	Wilms’ tumor	Biopsies	Metastatic	15	−0.5	0.260	<0.001−4		GDS1282/CPE
fB	Renal clear cell carcinoma	Biopsies	Metastatic	8	16.3	1.894	<0.0007		GDS505/CPE
g	Oligodendroglioma	Biopsies	Benign	8	1.75	0.726	<0.0097		GDS1813/CPE
h	Anaplastic oligodendroglioma	Biopsies	Metastatic	6	1.59	0.698	<0.0112		GDS1813/CPE
i	Glioblastoma	Biopsies	Metastatic	30	1.48	0.425	<0.0029		GDS1813/CPE
j	Astrocytic tumor	Biopsies	Metastatic	5	1.65	0.325	<0.0070		GDS1813/CPE
k	Non-functioning pituitary adenoma	Biopsies	Benign	5	−0.33	Pooled samples			GDS1253/CPE
l	ACTH-secreting pituitary adenoma	Biopsies	Benign	5	−0.5	Pooled samples			GDS1253/CPE
m	GH-secreting pituitary adenoma	Biopsies	Benign	5	1.75	Pooled samples			GDS1253/CPE
n	PRL-secreting pituitary adenoma	Biopsies	Benign	5	1	Pooled samples			GDS1253/CPE
o	Ewing’s sarcomaC	Biopsies	Metastatic	11	8.5	0.650		<0.031	GDS971/CPE

^ACPE expression fold changes compared to normal human fetal kidney

^BCPE expression fold changes compared to adjacent normal human kidney

^CCPE expression fold changes in Ewing sarcomas vs rhabdomyosarcomas

^DCPE expression fold changes in metastatic or begnin tumors versus normal tissues

CPE in Neuroendocrine Cancers

Neuroendocrine tumors are of special interest because of the abundance of CPE expression in these tumors. CPE is required for the biosynthesis of neuropeptides which serve as autocrine growth factors in these tumors (Rozengurt 2002). In a drug-resistant form of small-cell carcinoma of the lung (SCCL) cell line (NCI H82), expression of both 55 and 49 kDa forms of CPE have been reported (North and Du 1998). Quantitative immunostaining of CPE has shown that elevated CPE expression is a statistically significant predictor of good prognosis in pulmonary neuroendocrine tumors such as bronchial carcinoids, small-cell lung cancers, and large-cell neuroendocrine carcinomas (He et al. 2004). However, this is in contrast to recent studies, which suggest that significantly elevated levels of CPE expression in lung cancer patients indicate poor prognosis (Krajnik et al. 2010). In other endocrine tumors, significantly high levels of CPE expression were detected in some pituitary adenomas compared to normal pituitary. An oligonucleotide array analysis carried out on pituitary adenomas showed that CPE was differentially expressed in various subtypes (Table 1k–n). For example, CPE mRNA expression in GH-secreting pituitary adenomas (Table 1m) was significantly higher than normal pituitary, whereas in ACTH-secreting pituitary adenomas it was lower (Table 1l).

Pheochromocytomas/Paragangliomas (PHEO/PGL)

PHEO/PGL are rare typical neuroendocrine tumors. There are different categories of familial PHEO/PGL including mutations of succinate dehydrogenase subunits B, C, and D (SDHB/C/D), multiple endocrine neoplasia type 2 (MEN2), neurofibromatosis type 1 (NF1), and von Hippel-Lindau (VHL) syndrome. These all together account for about 25% of the tumors and the rest are sporadic (Neumann et al. 2002). The incidence of metastatic PHEO/PGL is currently estimated to be 3–36% or even higher, depending on genetic background and primary tumor localization (Pacak et al. 2007). Currently, there are no reliable markers for diagnosing metastatic PHEO/PGL (Lenders et al. 2005). We recently investigated CPE expression in these tumors by quantitative real-time PCR (qRTPCR). Seven sporadic PHEO/PGL resected between 2002 and 2008 were obtained from the National Institutes of Health and analyzed. These patients were followed up at the NIH, post surgery. The patients were chosen based on frozen tumor and pathological data availability, and positive PHEO/PGL-specific imaging studies together with positive biochemistry confirming the diagnosis of PHEO/PGL. During screening for an optimal amplicon of CPE for PCR, we came across a CPE variant smaller than the predicted amplicon, while amplifying at the 5′-UTR of CPE mRNA. We further designed specific primers and analyzed the sporadic PHEO/PGL samples and normal adrenals for this CPE variant. Wild-type CPE was highly expressed in the PHEO/PGL samples and in normal adrenals. To our surprise, the new short CPE variant was not detected in normal adrenals, but was present only in the PHEO/PGL samples. Moreover, the mRNA copy numbers of this variant correlated with the state of the tumor with regard to tumor growth and invasiveness (Table 2). For example, patient S71 with a very large benign tumor (8 cm in diameter) had a ~30 times higher CPE mRNA copy number than other patients (S39, S48, S49, S67) with benign tumors of <5 cm and were disease-free for at least 3 years following surgical resection of the primary tumor. Moreover, patient S76 who had a high CPE mRNA copy number also had a very aggressive tumor showing capsular and vascular invasion, although the tumor was diagnosed as benign based on pathology and biochemistry. Patient S45 who had a tumor with a high CPE mRNA copy number had metastatic papillary thyroid cancer 2 years prior to showing first symptoms of PHEO and was diagnosed with the disease 3 years later. These data suggest that highly elevated CPE variant mRNA levels in the primary PHEO tumor could be an indicator of poor prognosis.

Table 2.

Determination of CPE variant copy number in Sporadic PHEO/PGL tumors

Sample ID	Diagnosisa	Genotype	Copy numbers
S39	Benign	SPORADIC	175,080
S45	Benignb	SPORADIC	6,181,873
S48	Benign	SPORADIC	184,761
S49	Benign	SPORADIC	181,713
S67	Benign	SPORADIC	135,279
S71	Benignc	SPORADIC	6,451,057
S76	Benignd	SPORADIC	7,228,701

The above table shows the expression of CPE variant mRNA copy number in sporadic pheochromocytoma/paraganglioma (PHEO/PGL) samples. Total RNA was extracted from frozen PHEO/PGL after homogenization in TRIzol reagent and 0.2 μg was then converted to cDNA. A standard curve was set up to determine the CPE mRNA copy numbers by qRTPCR using CPE plasmid construct of eight different concentrations. Copy numbers were calculated using a standard formula. QRTPCR was done on PHEO/PGL samples and the crossing point for each sample was plotted on the standard curve to determine the mRNA copy number

^aAt the time of surgery

^bLarge tumor (8 cm)

^cThyroid cancer

^dCapsular and vascular invasion

Conclusions

In this study, we have mined various microarray data from the GEO profile database (http://www.ncbi.nlm.nih.gov/geo/) and the literature to analyze CPE mRNA expression in various non-endocrine and (neuro)endocrine cancers. As well, we have presented our data on the mRNA expression of a CPE variant in PHEO/PGL. Our analyses revealed that many non-endocrine metastatic cancers: cervical, colon, and kidney cancers and Ewing sarcomas show significantly higher levels of expression of CPE mRNA compared to their respective counterpart normal tissues which have substantially lower or no expression of CPE mRNA. Moreover, CPE mRNA levels were even more elevated in metastatic and malignant tumors compared to the primary tumor. For example (Table 1c), CPE mRNA expression in lymph node metastasis (SW620) CRC patients’ isogenic cell lines were much higher compared to those derived from the primary tumor (SW480). However, in brain, anaplastic oligodendrogliomas, glioblastomas, and astrocytic tumors, which are malignant, as well as oligodendrogliomas which are benign all showed higher expression of CPE mRNA to the same extent compared to normal brain. Among the endocrine tumors examined, pituitary adenomas secreting growth hormone exhibited the highest level of CPE expression compared to normal pituitary. This was rather surprising since CPE is not required to process pro-growth hormone. In our studies on PHEO/PGL, we found that expression of a variant CPE mRNA transcript was highly elevated in very large or invasive tumors, although these tumors were diagnosed as benign at time of surgery. The mRNA copy numbers of the CPE variant were more than 10 times greater in the malignant tumors compared to the benign small tumors. While our results are very promising, the patient sample size studied is small and need to be substantiated with larger numbers of well characterized tumor samples. Nevertheless, the data suggest that elevated CPE variant expression in the primary tumor indicates poor prognosis for PHEO/PGL patients. In conclusion, our extensive analysis of microarray data from the GEO profile database (http://www.ncbi.nlm.nih.gov/geo/) and the literature, and our studies on PHEO/PGL have yielded findings that strongly suggest CPE and/or its variant is involved in tumor growth. The data also suggest that CPE may promote tumor invasion and metastasis. Finally, CPE can potentially be a diagnostic and prognostic biomarker for malignancy in different cancer types.