SECRETORY PROTEIN TRAFFICKING AND GRANULE BIOGENESIS IN NEUROENDOCRINE CELLS
, Head, Section on Cellular Neurobiology
Niamh X. Cawley, PhD, Staff Scientist
Hong Lou, MD, Senior Research Assistant
Irina Arnaoutova, PhD, Postdoctoral Fellow
Josh Park, PhD, Postdoctoral Fellow
Andre Phillips, PhD, Postdoctoral Fellow
Tulin Yanik, PhD, Postdoctoral Fellow
Hisatsugu Koshimizu, PhD, Visiting Fellow
Alicja Woronowicz, MD, PhD, Visiting Fellow
Omar Contreras, BS, Postbaccalaureate Fellow (NIH Academy)
Annahita Sarcon, BS, Postbaccalaureate Fellow (Summer)
Vladimir Senatorov, PhD, Medical Student (Summer)
Taeyoon Kim, PhD, Guest Researcher
Mechanism of sorting pro-neuropeptides, neurotrophins, and their processing enzymes to the regulated secretory pathway
The intracellular sorting of pro-neuropeptides and neurotrophins to the regulated secretory pathway (RSP) is essential for the processing, storage, and release of active proteins and peptides in the neuroendocrine cell. We investigated the sorting of pro-opiomelanocortin (POMC; or pro–ACTH/endorphin), proinsulin, and brain-derived neurotrophic factor (BDNF) to the RSP. As a concentration step, pro-proteins undergo homotypic oligomerization as they traverse the cell from the site of synthesis in the endoplasmic reticulum to the trans-Golgi network (TGN), where they are sorted into dense-core granules of the RSP for processing by prohormone convertases (PCs) and carboxypeptidase E (CPE) and for secretion. We identified a consensus sorting motif, consisting of two acidic residues 12 to 15 Å apart, that is exposed on the surface of these molecules; we also found two hydrophobic residues, 5 to7 Å away from the acidic residues, that are also necessary for sorting to the RSP. In addition, we found a sorting motif in BDNF, which is secreted in an activity-dependent manner, but not in constitutively secreted nerve growth factor (NGF), which lacks one amino acid residue to complete the motif. Introduction of the missing residue by mutagenesis (Val20Glu) redirected NGF to the RSP, further confirming the importance of the sorting motif in targeting to the RSP. We identified an RSP sorting receptor that, as the transmembrane form of CPE, was specific for the sorting signal of POMC, proinsulin, and BDNF. The two acidic residues in the prohormone/pro–BDNF sorting motif specifically interact with two basic residues—R255 and K260—of the sorting receptor CPE to bring about sorting to the RSP. Transfection of a dominant negative CPE mutant into INS cells or depletion of CPE in these cells with siRNA caused partial mis-sorting of proinsulin to the constitutive pathway, indicating an in vivo interaction of proinsulin and CPE to effect sorting. Furthermore, using a CPE knockout (KO) mouse model, we showed that BDNF was not sorted to the RSP but was secreted constitutively in cortical and hippocampal neurons of the CPE KO mice. Constitutive secretion of proinsulin from isolated pancreatic islets appeared to be elevated in the mice while plasma levels of proinsulin were significantly higher, indicating a role of CPE in the intracellular trafficking of proinsulin.
In addition, our recent studies of POMC trafficking in primary pituitary cultures from CPE KO mice showed poor production of ACTH and revealed that about 50 percent of newly synthesized POMC was degraded, suggesting that, in the absence of efficient sorting to the granules of the RSP due to the lack of CPE, POMC was targeted for degradation. However, some of the remaining POMC was sorted into the RSP, as evidenced by its ability to be secreted in a stimulated manner. A candidate for a compensatory sorting receptor is Secretogranin III (SgIII), which has been shown to bind to POMC in precipitation assays. Indeed, analysis of pituitary extracts shows that the full-length 65 kDa form of SgIII, which was the form shown to bind to POMC, is elevated in the KO pituitary, suggesting that the increase may facilitate sorting of the residual POMC into the RSP. Our studies provide evidence for a sorting signal–/receptor-mediated mechanism for targeting prohormones, pro-neuropeptides, and the neurotrophin BDNF to the regulated secretory pathway in endocrine cells and neurons.
We demonstrated that the prohormone-processing enzyme CPE and prohormone convertases 1/3 and 2 (PC1 and PC2) are transmembrane secretory-vesicle proteins with an atypical membrane-spanning domain at the C-terminus and a cytoplasmic tail. We have shown that the sorting of these enzymes into granules of the RSP in neuroendocrine cells requires association of their C-terminal domain with cholesterol-glycosphingolipid–rich microdomains, known as lipid rafts, at the TGN. Removal of cholesterol from secretory granule membranes resulted in CPE’s inability to act as a sorting receptor and bind to cargo. Furthermore, cholesterol depletion by treatment of cells with lovastatin eliminated sorting of CPE to the RSP. Recently, we showed that removal of the transmembrane domain of PC1/3 alone (residues 617–638) with the C-terminal domain (residues 639–752) intact led to the mis-sorting of the enzyme to the constitutive pathway. Thus, translocation of the transmembrane domain of PC1/3 through the cholesterol-rich lipid rafts at the TGN appears to be essential for the sorting of PC1/3, and probably of PC2 and CPE, to the regulated secretory pathway.
In collaboration with Bruce Baum, we used our knowledge of the sorting motif of hormones to engineer biologically active mutant hormones that are redirected to the constitutive pathway. We are currently expressing such mutant hormones in salivary glands for systemic secretion and hope to apply the technology to gene therapeutics. Recently, we successfully expressed a mutant form of proinsulin in salivary glands of mice and observed that the proinsulin was secreted constitutively into the blood. Preliminary results indicate that the animals had lower plasma glucose levels than wild-type mice.
Cawley NX, Arnaoutova I, Yanik T, Lou H, Patel N, Loh YP. Techniques in neuropeptides processing, trafficking, and secretion. In: Gozes I, ed. Methods in Neurobiology. Humana Press, 2007; 67-96
Lou H, Kim S-K, Zaitsev E, Snell C, Lu B, Loh YP. Sorting and activity-dependent secretion of BDNF require interaction of motif I16D18I105E106 with the sorting receptor, carboxypeptidase E. Neuron 2005;45:245-55.
Lou H, Smith A, Coates L, Loh YP, Birch N. The transmembrane domain of the prohormone convertase PC3: a key motif for targeting to the regulated secretory pathway. Mol Cell Endocrinol 2007;267:17-25.
Absence of CART linked to obesity and low bone density
We investigated the sorting and processing of a mutant form of cocaine- and amphetamine-regulated transcript (CART) found in a family of obese patients. CART, observed in brain, is an anorectic peptide that has several physiological effects such as inhibiting feeding, regulating energy expenditure, and inducing stress. CART acts downstream of leptin in the obesity-controlling signaling pathway. We found a mutant pro–CART (Leu34Phe) in a 10-year-old Italian boy who had been obese since 2 years of age. This mis-sense Leu34Phe mutation co-segregates in three generations of his maternal relatives along with the phenotype of severe obesity, but his father was not obese. To investigate whether the boy’s family members have mature CART, we used Protein Chip technology (Ciphergen) to analyze serum from the obese boy and affected family members from three generations as well as from an unaffected sibling and five normal controls. All members of the family bearing the Phe34Leu mutation showed only pro– and intermediate CART in their circulation, but no mature CART. In contrast, normal humans and the unaffected sibling showed significant amounts of circulating mature CART. To determine the cellular basis for the lack of mature CART in the obese patients, we examined the trafficking and processing of mutant CART by transfecting wild-type (WT) or mutant (Leu34Phe) CART into AtT-20 cells. Whereas pro–CART was substantially processed to active CART, mutant pro–CART was only minimally processed to yield an intermediate form in these AtT-20 cells. Furthermore, WT CART was secreted in a regulated manner with high potassium stimulation, but mutant pro–CART/CART exhibited high basal release and no significant stimulated secretion. Immunocytochemical studies revealed that immunoreactive WT CART was primarily co-localized in punctate granules with POMC, which is a granule marker, in the processes of AtT-20 cells. However, about half the cells showed no punctate staining of immunoreactive mutant CART co-localized with POMC in cell processes, indicating that mutant pro–CART was partially mis-sorted and secreted via the constitutive pathway. The poor processing of mutant pro–CART is likely attributable to the mis-sorting to the constitutive pathway, which does not have the appropriate processing enzymes; this finding provides a molecular basis for the obese phenotype in this family.
Recently, we investigated the role of CART in bone remodeling. Using Protein Chip technology, we confirmed that, unlike WT mice, the CPE KO mouse, which we previously showed as lacking plasma CART, also lacks mature CART in the hypothalamus. Although obese mice and humans usually have higher bone density, surprisingly, both male and female obese CPE KO mice had lower bone density than WT mice. The mice showed elevated parathyroid hormone (PTH) and plasma calcium levels, significantly reduced osteoblast numbers, and significantly increased osteoclast numbers, all accounting for their poor bone density. Both CART−/− mice and CPE KO mice exhibited a lack of CART and increased osteoclast numbers, indicating a role for CART in bone remodeling. However, the CART−/− mice did not show a decrease in osteoblast number, suggesting that the absence of other neuropeptides or peptide hormones in the CPE KO mouse may play a role in osteoblast formation or survival.
Yanik T, Dominguez G, Kuhar M, Del Guidice EM, Loh YP. The Leu34Phe pro–CART mutation leads to CART deficiency: a possible cause of obesity in humans. Endocrinology 2006;147:39-43.
The CPE knockout mouse exhibits abnormal brain morphology and behavior and deficits in synaptic transmission
We generated the CPE KO mouse by deleting exons 4 and 5 from the CPE gene and then characterized the KO mouse’s phenotype. KO mice became obese by 10 to 12 weeks of age and weighed 60 to 80 g by 40 weeks. The null animals consumed more food overall, were less physically active during the light phase of the light-dark cycle, and burned fewer calories as fat than their WT littermates. Fasting levels of glucose and insulin-like immunoreactivity (IR) in plasma were elevated in KO mice at about 20 weeks. Males recovered from this state by 32 weeks, but the state persisted in females. At this age, the plasma insulin-like IR, which consists primarily of proinsulin, was 50 to 100 times higher than in the WT animals. The KO mice showed impaired glucose clearance and were insulin-resistant. High levels of leptin in plasma were present in the KO mice; however, the animals showed no circulating fully processed CART, a peptide that is responsive to leptin-induced feedback inhibition of feeding. Aside from obesity and diabetes phenotypes, the KO mice were subfertile and showed deficits in GnRH processing in the hypothalamus.
Behavioral analyses revealed that KO animals were less reactive to stimuli and exhibited less muscle strength and coordination, lower visual placing and toe-pinch reflexes than WT mice. Moreover, they showed delayed learning and poor memory consolidation in the water maze and object preference test. Analysis of brain sections at 6 weeks of age and older revealed that the KO mice had an abnormal hippocampal structure, viz. absence of the CA3 region. Electrophysiological studies indicated an inability to generate long-term potentiation when the CA1 region of the hippocampus was stimulated. Furthermore, the mossy fibers terminated prematurely before reaching the CA1 region, indicating that CPE plays an essential role in maintaining hippocampal structure in the adult animal.
Electron-microscopic studies of the hypothalamus of the CPE KO mice demonstrated the absence of presynaptic docked synaptic vesicles in 40 percent of synapses examined. In addition, glutamate release from embryonic hypothalamic neurons and adult hypothalamic synaptosomes was impaired. The mice also exhibited abnormal neurotransmission from the photoreceptors to the inner retina, showing a loss of the b wave in their retinogram. We obtained similar results in mutant mice (Cpefat/fat) bearing a substitution mutation in residue Ser202Pro of CPE. Immunocytochemical studies on these mice revealed a lack of CPE in the outer plexiform layer but an accumulation in the outer nuclear layer where, unlike in the case of normal mice, the cell bodies of the photoreceptor cells are localized. Furthermore, EM studies indicated significantly fewer synaptic vesicles in the spherules (synaptic butons), suggesting a defect in transport of synaptic vesicles carrying CPE. Our recent studies indicate that CPE is present in synaptic vesicles. Thus, CPE appears to play a role in the transport of synaptic vesicles/peptidergic vesicles along the axon and dendrites to the release site at specific synapses in the brain. In addition, CPE may be important for synaptic vesicle docking and exocytosis. The absence of CPE in the KO mice therefore leads to failure in neurotransmission, deficits in learning and memory, and abnormal behavior.
Zhu X, Wu K, Rife L, Cawley NX, Brown B, Adams T, Teofilo K, Lillo C, Williams D, Loh YP, Craft C. Carboxypeptidase E is essential for normal synaptic transmission from photoreceptors to the inner retina. J Neurochem 2005;95:1351-62.
Role of CPE in transport of POMC and BDNF vesicles in endocrine cells and neurons
Post-Golgi transport of hormone and BDNF vesicles for activity-dependent secretion is important in mediating endocrine function and synaptic plasticity. We investigated the role of the CPE cytoplasmic tail in vesicle movement. We showed that overexpression of the CPE cytoplasmic tail diminished localization of endogenous POMC, BDNF, and fluorescence-tagged CPE in the processes of the endocrine cell line AtT-20 and in hippocampal neurons. In these cell types, overexpression of the CPE tail in the cytoplasm inhibited the movement of POMC-/CPE-containing vesicles to the processes. S-tagged CPE tail pulled down endogenous microtubule-based motors, dynactin (p150), dynein, and KIF1A/KIF3A from the cytosol. Finally, overexpression of the CPE tail inhibited the regulated secretion of ACTH from AtT-20 cells. The study thus uncovered a novel mechanism whereby the vesicular CPE cytoplasmic tail plays a mandatory role in anchoring regulated secretory–pathway POMC/ACTH and BDNF vesicles to microtubule-based motors for transport and activity-dependent secretion in endocrine cells and neurons. We recently demonstrated that CPE is associated not only with large dense-core vesicles but also with synaptic vesicles, suggesting that the CPE tail may also be involved in transport of synaptic vesicles.
In recent yeast two-hybrid studies, we showed that the CPE tail also interacted with g-adducin and confirmed the interaction with in vitro precipitation studies. It is known that g-adducin is primarily a component of the cytoskeleton that binds to and stabilizes F-actin and mediates actin binding to fodrin. While the role of adducin has long been relegated to that of a structural protein, we recently made observations suggesting a wider role for the protein. To assess the significance of the CPE tail–g-adducin interaction in a functional system, we overexpressed the last 25 amino acids of the CPE C-terminal domain (CPE25) in the cytoplasm of AtT-20 cells. Immunocytochemical detection of g-adducin indicated a robust distribution of the protein in the cell processes, whereas, in the presence of CPE-25, g-adducin localization shifted to a largely cell-body distribution. Given that overexpression of CPE-25 inhibits interaction of the cytoplasmic tail with dynactin and hence vesicle movement along the microtubules in the processes, the redistribution of g-adducin in the presence of CPE-25 suggests that g-adducin binds to the vesicles via the CPE cytoplasmic tail and moves with these organelles. We hypothesize that g-adducin linked to CPE tail may participate in the tethering of ACTH vesicles to cortical actin just below the plasma membrane in readiness for priming, docking, and release upon stimulation.
Regulation of secretory granule biogenesis by chromogranin A
Formation of large dense-core granules (LDCGs) at the TGN is essential for regulated secretion of hormones and neuropeptides from neuroendocrine cells. Our recent studies uncovered an on/off switch, chromogranin A (CgA), that controls the formation of LDCGs in neuroendocrine cells. Depletion of CgA in rat PC12 cells with antisense technology resulted in the loss of LDCGs and the regulated secretion and degradation of granule proteins, including CgB and synaptotagmin. Overexpression of bovine CgA in these cells rescued the WT phenotype. Transfection of CgA into the mutant endocrine cell line 6T3, which lacks CgA, LDCGs, and regulated hormone secretion, resulted in the rescue of granule biogenesis and the regulated secretory pathway. The importance of CgA in LDCG biogenesis was evident not only in cell lines but also in vivo. In an antisense-mRNA transgenic mouse model deficient in CgA, we observed severe aberrant granule formation both quantitatively and qualitatively in the adrenal medulla and noted a correlation between the amount of depletion of CgA and the reduction in secretory granule biogenesis.
A reduction in granule proteins accompanied the reduction in secretory granule biogenesis in the adrenal medulla of the CgA-deficient transgenic animals as well as in 6T3 cells lacking CgA. Using 6T3 cells lacking secretory granules as a model, we showed that the reduced granule protein levels were attributable to degradation at the Golgi apparatus. Thus, we proposed that regulation of the stability of granule proteins at the Golgi apparatus by CgA is a focal point for control of granule biogenesis in neuroendocrine cells. In support of this view, we found a protease inhibitor, protease nexin-1 (PN-1), in the Golgi that is transcriptionally activated by CgA and upregulated in cells actively forming LDCGs but downregulated in cells minimally expressing CgA and showing low levels of LDCG biogenesis. Moreover, transfection of PN-1 into 6T3 cells lacking CgA prevented LDCG protein degradation and rescued granule biogenesis. Furthermore, downregulation of expression of PN-1 by anti-sense RNA in 6T3 cells transfected with CgA resulted in enhanced degradation of granule proteins and decreased secretory granule formation. Recently, we showed that 6T3 cells incubated with conditioned medium from 6T3-bCgA cells or AtT-20 cells resulted in an increase in PN-1 mRNA and granule biogenesis. Likewise, stimulation of AtT-20 cells with high potassium resulted in an increase in PN-1 mRNA; actinomycin D blocked the increase. We found that an approximately 3kD C-terminal proteolytic fragment of CgA, which we named serpinin, was able to enhance PN-1 transcription and granule biogenesis in 6T3 cells. Using an antibody against serpinin, we detected a serpinin-like peptide in the medium after high K+ stimulation of AtT-20 cells. We hypothesize that serpinin binds to a cognate receptor on the plasma membrane to cause signaling to the nucleus to enhance PN1-mRNA transcription. Thus, we have uncovered a novel mechanism whereby serpinin, a CgA-derived peptide secreted into the medium along with POMC-derived hormones upon stimulation of pituitary corticotrophs, signals replenishment of LDCGs by transcriptionally activating the protease inhibitor PN-1, which then stabilizes granule proteins at the TGN to increase LDCG biogenesis.
Kim T, Gondre-Lewis M, Arnaoutova I, Cawley NX, Loh YP. Neurosecretory protein trafficking and dense-core granule biogenesis in neuroendocrine cells. In: Lajtha A, Banik N, eds. Handbook of Neurochemistry and Molecular Neurobiology, 3rd Ed., Vol. 7, Ch. 3; Neural Proteins: Metabolism and Function. Springer Verlag, 2007;153-68.
Kim T, Gondre-Lewis M, Arnaoutova I, Loh YP. Dense-core secretory granule biogenesis. Physiology 2006;21:124-33.
Kim T, Loh YP. Protease nexin-1 regulates granule protein stability and secretory granule biogenesis at the Golgi. Mol Biol Cell 2006;17:789-98.
Role of aquaporin 1 in hormone secretion and secretory granule biogenesis
Recently, we found that the water channel protein aquaporin 1 (AQP1), which is normally present in the plasma membrane, is also localized in secretory granules of endocrine cells. Moreover, when we transfected 6T3 cells lacking CgA with CgA, which rescues granule biogenesis in these cells, we observed a significant increase in expression of AQP1 mRNA and protein. To investigate the role of AQP1 in hormone sequestration and granule biogenesis, we stably transfected AtT-20 cells with an antisense construct of AQP1 to downregulate expression of the protein. AQP1-deficient AtT-20 cells showed a dramatic reduction of secretory granules. Pulse-chase studies demonstrated a defect in regulated secretion of the endogenous ACTH hormone as well as increased degradation of newly synthesized granule proteins such as POMC and CPE at the Golgi apparatus. However, the deficiency of AQP1 did not affect the transcription and translation of POMC and CPE. AQP1 appears to be critical for the mechanism of secretory granule exocytosis. A defect in secretion leads to feedback downregulation of secretory granule biogenesis. Thus, AQP1 is necessary for maintaining hormone secretion and granule biogenesis in endocrine cells.
CPE regulates growth and metastasis in tumor cell lines
Elucidation of molecules that control cancer cell growth and invasion will greatly facilitate the identification of biomarkers that can predict impending tumor metastasis and provide targets for therapy. Currently, numerous biomarkers, some associated with metastasis, have been reported to be useful in identifying aggressive tumors and for prognosis. However, despite the huge repertoire of biomarkers, there is still a lack of reliable markers for predicting future metastasis of most cancers based on biopsies or resected primary tumors. Recently, we have found that levels of CPE, also known as carboxypeptidase H, are correlated with growth and metastasis in a variety of human cancer cell lines. We showed by Western blot that a human neuroblastoma cell line IMR-32, amplified with MYCN and thus highly metastatic, expressed more CPE than the matched cell line SK-NAS, which has low metastatic potential. Likewise, in a variety of human tumor cell lines from liver, colon, breast, prostate, colon, head, and neck, we observed an elevated expression of CPE in cell lines that were more metastatic than matched tumor lines with low metastatic potential. In addition, we have shown that, in mouse pheochromocytoma cell lines, high metastatic potential (MTT) cells exhibited higher levels of CPE than lines with low metastatic potential (MPC) cells.
We also demonstrated that CPE plays a role in promoting growth and cell invasion in human cancer cells. When si-RNA downregulated CPE expression in highly metastatic human cell lines from breast (MDA-MB-23), prostate (DU145), head and neck (MDA 1986), colon (HT29), and liver (MHCC97M3) by 60 to 85 percent, the si–CPE treated cells showed a 52 to 85 percent inhibition of growth. We used the matrigel invasion assay to assess invasive capacity of the tumor cell lines transduced with CPE si-RNA. In all the tumor lines, suppression of CPE expression led to a 70 to 85 percent inhibition of invasion. We thus concluded that CPE plays a role in promoting growth and invasion in several types of cancer cell lines.
Clinical use of CPE as a biomarker for metastasis and as a therapeutic target
In two retrospective studies, we investigated further whether CPE could serve as a useful biomarker for metastatic cells. Tissue microarray analysis (TMA) of immunoreactive CPE from matched pairs of primary and metastatic colon cancer samples from 31 patients with primary colon cancer that metastasized to the liver revealed that 17 (54.8 percent) patients were negative for CPE and 14 patients had a few CPE-positive cells in the primary tumor. Subsequently, 25 (80 percent) of those patients showed intense CPE staining in many of their liver tumor cells, indicating that CPE is a biomarker for metastasis. Quantitative Western blot analysis of hepatic carcinoma cells (HCCs) from 60 patients with primary resected hepatic carcinoma revealed that, in 20 of 30 patients who remained disease-free, primary tumor CPE levels were equal to or less than twice that of surrounding non-tumor cells; on the other hand, 24 of 30 patients with CPE levels in the primary tumor that were more than twice (average four-fold) that of the surrounding non-tumor cells developed metastasis within six months. Thus, our preliminary results indicate that CPE could be a reliable biomarker for the prediction of impending metastasis for HCCs and possibly other cancers. Suppression of CPE expression might offer a potential therapy for reversing metastasis.
COLLABORATOR
Bruce Baum, DMD, Gene Therapy and Therapeutics Branch, NIDCR, Bethesda, MD
Nigel Birch, PhD, University of Auckland, Auckland, New Zealand
Robert Chow, PhD, University of Southern California, Los Angeles, CA
Cheryl Craft, PhD, University of Southern California, Los Angeles, CA
Emanuele del Guidice, MD, Seconda Università di Napoli, Naples, Italy
Steven Hewitt, MD, PhD, Laboratory of Pathology, NCI, Bethesda, MD
Joanna Hill, PhD, Laboratory of Behavioral Neuroscience, NIMH, Bethesda, MD
Terence Lee, PhD, University of Hong Kong, Hong Kong, China
Joan Marini, MD, PhD, Bone and Extracellular Matrix Branch, NICHD, Bethesda, MD
Karel Pacak, MD, PhD, DSc, Program in Reproductive and Adult Endocrinology, NICHD, Bethesda, MD
Ronnie Poon, MD, PhD, University of Hong Kong, Hong Kong, China
William Wetsel, PhD, Duke University, Durham, NC
For further information, contact lohp@mail.nih.gov.
