GASTROENTEROLOGY 1996;110:1595–1604
`
`g-Aminobutyric Acid Secretion From Pancreatic
`Neuroendocrine Cells
`
`GUDRUN AHNERT–HILGER,* ADELE STADTBA¨UMER,* CARSTEN STRU¨ BING,‡ HANS SCHERU¨ BL,*
`GU¨ NTER SCHULTZ,‡ ERNST–OTTO RIECKEN,* and BERTRAM WIEDENMANN*
`*Medizinische Klinik und Poliklinik, Abteilung Innere Medizin mit Schwerpunkt Gastroenterologie, and ‡Institut fu¨r Pharmakologie,
`Universita¨tsklinikum Benjamin Franklin, Freie Universita¨t Berlin, Berlin, Germany
`
`Recently, it has been suggested that a similar basic
`Background & Aims: Neuroendocrine cells and tumors
`derived therefrom contain hormone-storing large dense
`fusion machinery may be used by all secretory cells.
`core vesicles and neuron-like small synaptic vesicle Membrane fusion is based primarily on the interaction
`analogues with unknown function. The aim of this study
`between soluble fusogens and their receptors. Soluble
`was to characterize the small synaptic vesicle pathway
`fusogens are the N-ethylmaleimide–sensitive factor
`in detail. Methods: In human pancreatic neuroendo-
`(NSF) and a-, b-, and g-soluble NSF attachment proteins
`crine tumors and corresponding mammalian cell lines,
`(SNAPs).14 NSF binds to membranes when one or more
`the expression of key proteins of regulated secretion
`of
`the SNAPs
`link up to their SNAP receptors
`were detected by immunofluorescence microscopy. Us-
`(SNAREs). An integral membrane protein of a transport
`ing 3H–g-aminobutyric acid (GABA), uptake and release
`vesicle (v-SNARE) pairs with one or more integral mem-
`by small synaptic vesicle analogues were studied. Re-
`brane proteins of the target membrane (t-SNARE). Ac-
`sults: Tumor tissues obtained from 14 patients ex-
`cording to this hypothesis, different v-SNAREs and t-
`pressed key proteins of neurosecretion such as synap-
`SNAREs confer specificity to intracellular fusion events,
`tobrevin, syntaxins, and SNAP 25. These proteins were
`whereas NSF and SNAPs are common to all forms.14–16
`also found in the cell lines AR42J, BON, RIN, and INR.
`The cell lines specifically transported GABA by a low-
`In neurons and chromaffin cells, the v-SNAREs for SSVs
`affinity plasma membrane transporter and showed an
`and LDCVs are synaptobrevin, whereas the t-SNAREs
`adenosine triphosphate–sensitive GABA uptake into
`are syntaxins and SNAP 25 (synaptosomal-associated
`an intracellular compartment. Stored GABA was re-
`protein of 25 kilodaltons).14,17–19
`leased upon stimulation by regulated exocytosis. Elec-
`So far, little information exists concerning the secre-
`trophysiological analyses suggested that calcium-de-
`tory pathway of SSV analogues in neuroendocrine cells.
`pendent secretion was mediated by activation of
`The idea that SSV analogues of pancreatic neuroendocrine
`voltage-dependent calcium channels of mainly the L
`cells may store amino acid transmitters has been sup-
`type, but also of the N and probably the T type. Conclu-
`ported by studies on subcellular fractionation of TCb
`sions: Small synaptic vesicle analogues in neuroendo-
`cells,9 by uptake studies in synaptophysin-immunoiso-
`crine cells and tumors can store and secrete GABA and
`lated vesicles,12 and, most importantly, by the regulated
`probably other amino acid transmitters by regulated
`exocytosis comparable with neurons.
`release of g-aminobutyric acid (GABA) from the amphi-
`crine cell line AR42J.11 The present study was under-
`taken to characterize in detail the SSV pathway. For
`pancreatic neuroendocrine tumors and the corresponding
`mammalian cell lines, we show the expression of SNARE
`proteins found to regulate exocytosis from neurons17,19
`
`N
`
`clinical
`irrespective of
`tumors,
`euroendocrine
`symptoms, express proteins characteristic of secre-
`tory vesicles.1–3 In the past, diagnostic and therapeutic
`efforts have focused on large dense core vesicles (LDCVs),
`the storage organelles for hormones and other polypep-
`tides.2 It turned out that secretion by the LDCV pathway
`is operative regardless of whether patients have clinical
`symptoms.4–6 Besides LDCVs, neuroendocrine cells and
`tumors derived therefrom contain another secretory path-
`way that uses vesicles resembling the small synaptic vesi-
`cles (SSVs) of neurons and thus are termed synaptic-like
`microvesicles or SSV analogues.1,2,7–13
`
`Abbreviations used in this paper: DMEM, Dulbecco’s modified Ea-
`gle medium; EGTA, ethylene glycol-bis(b-aminoethyl ether)-N,N,Nⴕ,Nⴕ-
`tetraacetic acid; GABA, g-aminobutyric acid; ICa, whole-cell calcium
`current; KR-HEPES, Krebs’-Ringer-HEPES; LDCV, large dense core
`vesicles; NSF, N-ethylmaleimide–sensitive factor; SNAP, synapto-
`somal-associated protein; SNARE, synaptosomal-associated protein
`receptor; SSV, small synaptic vessel.
`䉷 1996 by the American Gastroenterological Association
`0016-5085/96/$3.00
`
`/ 5e0d$$0033
`
`04-12-96 15:06:48
`
`gasas
`
`WBS-Gastro
`
`Roxane Labs., Inc.
`Exhibit 1036
`Page 001
`
`

`
`1596 AHNERT–HILGER ET AL.
`
`GASTROENTEROLOGY Vol. 110, No. 5
`
`and nonneoplastic neuroendocrine cells.18,20–22 Further-
`more, we show that mammalian pancreatic neuroendo-
`crine cell lines specifically take up GABA, which is stored
`in SSV analogues and released by regulated exocytosis.
`
`Materials and Methods
`Tissues and Cell Cultures
`
`Human tumor tissues were obtained from 14 patients
`by tumor resection, immediately shock frozen, and used ac-
`cording to the standards set by the Ethical Committee of the
`Klinikum Benjamin Franklin, Freie Universita¨t Berlin. PC12
`cells23 were cultivated in Dulbecco’s modified Eagle medium
`(DMEM) supplemented with 10% horse and 5% fetal calf
`serum and AR42J24 in DMEM supplemented with 20% fetal
`calf serum at 10% CO2. RIN 3825 and INR26 were grown in
`RPMI supplemented with 5% fetal calf serum and 5% calf
`newborn serum, BON cells27 in a 1:1 mixture of DMEM, and
`F12K medium containing 10% fetal calf serum and HeLa cells
`in DMEM containing 10% fetal calf serum in an atmosphere
`of 5% CO2. Primary cultures of mouse hypothalamus were
`prepared and grown as previously described.28
`
`Antibodies
`
`Monoclonal antibody against synaptobrevin II and a
`polyclonal antiserum against synaptobrevin I29 were provided
`by Dr. R. Jahn, Yale University, Howard Hughes Medical
`Institute, New Haven, CT. Polyclonal antisera against human
`synaptobrevin II and syntaxin were kindly provided by Dr.
`T. Rapoport, Harvard University, Boston, MA. Monoclonal
`antibody SY 387 and a polyclonal antiserum against synapto-
`physin10 and monoclonal antibody against cytochrome b56113
`were used as previously described. Monoclonal antibodies
`against SNAP 25 and syntaxin (HPC-1 Barnstacle) were ob-
`tained from Sternberger Monoclonals (Baltimore, MD) or
`Sigma Chemical Co. (Deisenhofen, Germany), respectively. An
`anti-rabbit immunoglobulin (Ig) G antiserum from goat cou-
`pled to Texas Red and an anti-mouse IgG antiserum from
`donkey coupled to Texas Red were obtained from Dianova
`(Hamburg, Germany).
`
`Other Chemicals
`
`Streptolysin O, purified30 and provided by Dr. U. Wel-
`ler, Institut fu¨ r Mikrobiologie, Johannes Gutenberg-Universi-
`ta¨t, Mainz, Germany, was used as described earlier.31,32
`(/)Isradipine was a gift from Sandoz AG (Basel, Switzerland).
`The calcium ionophore A23187, aminooxyacetic acid, gaba-
`culin, betain, b-alanin, and diaminobutyric acid were obtained
`from Sigma Chemical Co. a-Latrotoxin and v-conotoxin
`GVIA were purchased from Alamone Laboratories (Jerusalem,
`Israel). GABA (sp act, 60 Ci/mmol) and 125J–protein A (sp act,
`30 mCi/mg) were obtained from Amersham (Braunschweig,
`Germany).
`
`Immunofluorescence Microscopy and
`Immunoreplica Analysis
`
`Immunofluoresecence microscopy from cryosections of
`human tumor tissues was performed as described.7 The various
`cell lines were grown on coverslips and processed for immu-
`nofluoresecence microscopy as previously described33,34 using
`either Texas Red coupled to anti-mouse or anti-rabbit IgG.
`Immunoreplica analysis was performed as previously de-
`scribed34 except that 125J–protein A was used to detect immu-
`noreactive signals.
`
`GABA Uptake and Secretion
`
`GABA uptake into intact cells was performed in the
`presence of 1 mmol/L aminooxyacetic acid and gabaculine in
`Krebs’-Ringer-HEPES buffer containing (in mmol/L): NaCl,
`130; KCl, 4.7; MgSO4, 1.2; CaCl2, 2.5; glucose, 11; and
`HEPES, 10, pH 7.4 (KR-HEPES buffer), for 30 minutes at
`37⬚C in the presence of the various inhibitors. The incubation
`was stopped by diluting the sample with 500 mL of ice-cold
`KR-HEPES buffer followed by rapid centrifugation. The su-
`pernatant was discarded, and the pellet was washed again with
`KR-HEPES buffer and then dissolved in lysis buffer containing
`(in mmol/L): Tris-HCl, 130; CaCl2, 10; NaCl, 75, pH 8,
`supplemented with 0.4% Triton X-100. One part of the cell
`lysate was used to count the amount of radioactivity; from
`the other part the protein content was determined using the
`bicinchoninic acid method. Values are calculated as picomoles
`per milligram of protein and expressed as percent of the uptake
`in the absence of inhibitors.
`Uptake into permeabilized cells was performed using an
`intracellular buffer consisting of (in mmol/L): sucrose, 200;
`KCl, 50; piperazine-N,N ⴕ-bis(2-ethanesulfonic acid), 20; and
`ethylene glycol-bis(b-aminoethyl ether)-N,N,Nⴕ,Nⴕ-tetraace-
`tic acid (EGTA), 4, pH 7.0, with or without 2 mmol/L Mg/
`adenosine triphosphate (ATP) and 3H-GABA. Incubation with
`SLO followed a protocol given elsewhere.31,32 Values are calcu-
`lated as picomoles per milligram of protein. The uptake in the
`absence of ATP was 100%.
`Secretion was performed from preloaded cells as previously
`described.11
`
`Electrophysiology
`
`For electrophysiological experiments, cells were trans-
`ferred into a perfusion chamber (4 mL/min) and whole-cell
`patch-clamp experiments35 were performed at 37⬚C. Patch pi-
`pettes with pipette resistances of 3–6 MV were prepared from
`borosilicate glass capillaries (Jencons, Leight Buzzard, En-
`gland). Currents were recorded using a List LM/EPC 7 patch-
`clamp amplifier (List Electronics, Darmstadt, Germany) using
`the CED (Cambridge Electronic Design, Cambridge, England)
`interface and software. Voltage-dependent calcium currents
`were routinely elicited from a holding potential of 080 mV
`by either 200-millisecond-long voltage pulses to 10 mV or by
`voltage ramps from 0100 to 100 mV (slope, 1 V/s). Calcium
`
`/ 5e0d$$0033
`
`04-12-96 15:06:48
`
`gasas
`
`WBS-Gastro
`
`Roxane Labs., Inc.
`Exhibit 1036
`Page 002
`
`

`
`May 1996
`
`GABA SECRETION FROM NEUROENDOCRINE CELLS 1597
`
`currents were measured as peak inward currents. Leakage cur-
`rents were determined from the hyperpolarizing part of the
`voltage ramps (0100 to 080 mV) and subtracted offline.
`Statistical values are given as means { SEM.
`To isolate inward currents through voltage-dependent cal-
`cium channels, sodium and potassium were omitted from the
`intracellular and extracellular solutions. Barium was used as
`an extracellular charge carrier. The patch pipettes were filled
`with a high-cesium pipette solution containing (in mmol/L):
`CsCl, 120; MgCl2, 1; Mg ATP, 3; HEPES, 10; and EGTA,
`10; adjusted to pH 7.4 with CsOH at 37⬚C. For the measure-
`ments of barium inward currents through calcium channels,
`an external solution was used containing (in mmol/L): D(0)–
`N-methylglucamine, 120; BaCl2, 10.8; MgCl2, 1; CsCl, 5.4;
`glucose, 10; and HEPES, 10; adjusted to pH 7.4 with HCl at
`37⬚C.
`
`Results
`Expression of SNARE Proteins in
`Pancreatic Neuroendocrine Tumors and
`Mammalian Cell Lines
`
`Regulated secretion involves the interaction be-
`tween the vesicular v-SNARE proteins synaptobrevin and
`plasma membrane t-SNARE proteins syntaxins and SNAP
`25. The expression of synaptobrevin, syntaxins, and SNAP
`25 were analyzed in serial sections of various neuroendo-
`crine tumors of the pancreas. As an example, a pancreatic
`gastrinoma from a 33-year-old patient is shown in Figure
`1. Antibodies against synaptobrevin, SNAP 25, and the
`syntaxins immunoreacted only with the tumor cells but
`not with the surrounding connective tissue. Similar results
`were obtained in 14 other pancreatic neuroendocrine tu-
`mors (Table 1) using monoclonal antibodies against
`syntaxins and synaptobrevin II (not shown). No immuno-
`reactivity was obtained with an antiserum against synapto-
`brevin I (data not shown). Table 1 summarizes the expres-
`sion of SNARE proteins in 14 pancreatic tumors. In
`normal human pancreatic tissue analyzed for comparison,
`only the pancreatic islets but not the surrounding exocrine
`tissue immunoreacted with antibodies against SNAP 25,
`syntaxins, and synaptobrevin including synaptobrevin II
`(data not shown). As previously described, marker proteins
`for either LDCVs such as cytochrome b561 and dopamine
`b-hydroxylase or SSVs such as synaptophysin and protein
`SV2 were expressed in these tumors13 and used for compar-
`ison (see Table 1). Similar results were obtained with 16
`neuroendocrine tumors derived from the midgut (data not
`shown).
`Mammalian cell lines derived from pancreatic neuro-
`endocrine tumors but not HeLa cells also highly ex-
`pressed the SNARE proteins necessary for regulated exo-
`cytosis, as could also be detected by immunofluorescence
`
`microscopy (not shown) and in membranes obtained from
`postnuclear supernatants (Figure 2).
`
`Functional Analysis of SSV Analogues in
`Mammalian Neuroendocrine Pancreatic
`Cell Lines
`
`Because primary cultures from human neuroendo-
`crine tumor tissues were not available, functional studies
`of SSV analogues were performed in mammalian neuro-
`endocrine pancreatic tumor cell lines.
`
`GABA Uptake Into Intact and
`Permeabilized Cells
`
`In a first set of experiments, the GABA uptake
`was characterized using various inhibitors specific for
`different plasma membrane GABA transporters. GABA
`uptake in AR42J cells was driven by a low-affinity trans-
`porter with a median inhibitory concentration of 200
`mmol/L for GABA, betain, and b-alanin. In contrast,
`GABA uptake into INR and RIN cells with a median
`inhibitory concentration of 100 mmol/L for GABA was
`sensitive to b-alanin (median inhibitory concentration,
`50 mmol/L for INR and 200 mmol/L for RIN) and almost
`insensitive to betain, in which millimolar concentrations
`were required for inhibition of uptake (Figure 3). Similar
`results were obtained with BON cells (data not shown).
`diaminobutyric acid, a well-known inhibitor of the neu-
`ronal GABA transporter GAT-136 that efficiently inhib-
`ited GABA uptake into mouse hypothalamic neurons
`(data not shown), was effective only in millimolar concen-
`trations in the neuroendocrine cell lines (Figure 3).
`Transmitter uptake into secretory vesicles is an ATP-
`dependent process. To prove that GABA is really stored
`in an intracellular compartment, cells were permeabilized
`by streptolysin O, washed free of endogenous ATP, and
`then incubated with radiolabeled GABA in the presence
`of ATP and various amounts of unlabeled GABA. ATP
`stimulated GABA uptake 2.5–3-fold (Figure 4)
`in
`AR42J and INR cells and twofold in RIN and BON
`cells (not shown). No ATP-dependent uptake was seen
`in permeabilized HeLa cells under these conditions (not
`shown). Adding carbonyl cyanide p-trifluoromethoxy-
`phenylhydrazone (not shown) or increasing the concen-
`tration of unlabeled GABA (Figure 4) inhibited ATP-
`dependent uptake.
`
`Regulated Secretion of GABA
`
`All four cell lines released GABA in a regulated
`fashion (Figure 5). GABA secretion could be elicited only
`by an elevated potassium concentration or by the calcium
`ionophore A23187 (Figure 5). a-Latrotoxin, a potent
`
`/ 5e0d$$0033
`
`04-12-96 15:06:48
`
`gasas
`
`WBS-Gastro
`
`Roxane Labs., Inc.
`Exhibit 1036
`Page 003
`
`

`
`1598 AHNERT–HILGER ET AL.
`
`GASTROENTEROLOGY Vol. 110, No. 5
`
`Figure 1. Expression of SNARE proteins in a pancreatic neuroendocrine tumor (gastrinoma). Serial sections were fixed and immunostained as
`described in Materials and Methods. Immunofluorescence microscopy was performed using a conventional fluorescence microscope. The
`following antibodies were used: (Aand B) polyclonal antiserum against human synaptobrevin II (dilution, 1:100), (Cand D) monoclonal antibody
`against SNAP 25 (1:1000), and (E and F) a polyclonal antiserum against syntaxin (1:100). The antibodies were detected using a Texas Red–
`labeled goat anti-rabbit or goat anti-mouse antiserum. Immunofluorescence micrographs are shown on the right and corresponding phase
`contrast images on the left. Note that only the tumor tissue is immunostained, whereas the surrounding connective tissue shows no immunoreac-
`tivity.
`
`/ 5e0d$$0033
`
`04-12-96 15:06:48
`
`gasas
`
`WBS-Gastro
`
`Roxane Labs., Inc.
`Exhibit 1036
`Page 004
`
`

`
`May 1996
`
`GABA SECRETION FROM NEUROENDOCRINE CELLS 1599
`
`Table 1. Immunocytochemical Detection of SNARE Proteins and Membrane Proteins of LDCVs and SSVs in Pancreatic
`Neuroendocrine Tumor Tissue
`
`LDCVs
`
`SSV
`analogues
`
`SNARE proteins
`
`v-SNAREs
`
`t-SNAREs
`
`Cytochrome b561
`
`Synaptophysin
`
`Synaptobrevins
`
`SNAP 25
`
`Syntaxins
`
`Functional
`Insulinoma (n Å 1)
`Gastrinomas (n Å 5)
`VIPoma (n Å 1)
`Nonfunctional (n Å 7)
`
`//
`//
`5/5
`//
`//
`6/7
`
`///
`///
`5/5
`///
`///
`7/7
`
`//
`//
`5/5
`//
`//
`7/7
`
`///
`///
`5/5
`//
`///
`7/7
`
`/
`/
`4/5
`//
`/
`6/7
`
`NOTE. Besides tumors with hypersecretory symptoms, a second group of pancreatic tumors with neuroendocrine properties lacking these
`symptoms has been identified using antibodies against secretory vesicle proteins. For clinical diagnostic purposes, these tumors may be
`classified in general as nonfunctional.1–3 Immunofluorescence microscopy was performed as described in Materials and Methods. Intensity of
`the reaction was assessed as follows: ///, very strong; //, strong; /, less strong.
`
`secretagogue for exocytosis by SSVs from neurons,37
`which stimulated GABA secretion from mouse hypotha-
`lamic neurons (not shown), had no effect on GABA secre-
`tion from the neuroendocrine cell lines (not shown).
`Other secretagogues that elicit amylase secretion from
`AR42J cells such as carbachol10 did not stimulate GABA
`secretion from the neuroendocrine cell lines. In the ab-
`sence of extracellular calcium, no regulated GABA secre-
`tion was seen (Figure 6). However, for unknown reasons,
`
`Figure 2. SNARE proteins on membranes of mammalian pancreatic
`neuroendocrine cell lines. Postnuclear supernatants (40 mg of total
`protein per slot) of the cell lines indicated were subjected to sodium
`dodecyl sulfate gel electrophoresis, transferred to polyvinylidene
`difluoride membranes, and probed with the antibodies: a polyclonal
`antiserum against synaptobrevin (dilution, 1:1000), a monoclonal an-
`tibody against SNAP 25 (1:2000), and a monoclonal antibody against
`syntaxin (1:1000). Antibody binding was detected by 125J–protein A.
`Note that SNARE proteins were detected in neuroendocrine cell lines
`but not in HeLa cells.
`
`the absence of extracellular calcium increased basal re-
`lease as previously described for AR42J cells.11
`
`Voltage-Dependent Calcium Channels in
`Mammalian Neuroendocrine Pancreatic
`Cell Lines
`
`All neuroendocrine cell lines studied showed volt-
`age-dependent calcium channels. To characterize the
`different calcium channel subtypes underlying the whole-
`cell calcium current (ICa), we used subtype-specific chan-
`nel inhibitors. Extracellular application of the L-type cal-
`cium channel antagonist isradipine (1 mmol/L) inhibited
`ICa in INR, RIN, and AR42J cells by 87% { 4.1%
`(n Å 6), 81.8% { 3.9% (n Å 4), and 71.3% { 7.8%
`(n Å 5), respectively. During washout, the effect of israd-
`ipine reversed completely. Cumulative application of is-
`radipine and the N-type calcium channel blocker v-
`conotoxin (3 mmol/L) almost completely (95%–100%)
`abolished the inward ICa in INR and RIN cells, whereas
`in AR42J a cadmium-sensitive ICa of approximately
`29.5% { 3.6% (n Å 4) remained after application of
`both antagonists (Figure 7). v-Conotoxin alone reduced
`the control ICa by 19.7% { 7.8% (n Å 3), 13.3% {
`3.8% (n Å 3), and 36.2% { 5.2% (n Å 4) in INR, RIN,
`and AR42J, respectively. This indicates the expression of
`dihydropyridine-sensitive L-type calcium channels and
`an v-conotoxin–sensitive ICa component in all three cell
`lines.
`Because the effects of isradipine and v-conotoxin were
`not or only partially additive, we assume v-conotoxin to
`probably work on a subset of L-type channels containing
`the a1D subunit rather than on N-type channels. The
`a1D subunit found in different neuronal and neuroendo-
`
`/ 5e0d$$0033
`
`04-12-96 15:06:48
`
`gasas
`
`WBS-Gastro
`
`Roxane Labs., Inc.
`Exhibit 1036
`Page 005
`
`

`
`1600 AHNERT–HILGER ET AL.
`
`GASTROENTEROLOGY Vol. 110, No. 5
`
`Figure 4. ATP-dependent GABA uptake into permeabilized AR42J and
`INR cells. Cells were permeabilized with SLO in the absence of ATP
`(see Materials and Methods). Excess ATP was removed by a 20-
`minute incubation at 36⬚C in sucrose KCl. The buffer was replaced
`by fresh KS buffer containing either no ATP or 2 mmol/L Mg/ATP and
`3H-GABA supplemented with the given concentrations of unlabeled
`GABA. The incubation was stopped after 30 minutes at 36⬚C by dilu-
`tion of the samples with ice-cold ATP-free KS buffer and centrifugation
`for 30 seconds at 12,000g.The supernatant was discarded. The cells
`were lysed, and the radioactivity and protein content were determined.
`Uptake in the absence of ATP was set at 100%. Values represent the
`means of three samples { SD. In the presence of 250 mmol/L unla-
`beled GABA, the ATP-dependent uptake was almost completely inhib-
`ited.
`
`crine cell types constitutes high voltage–activated cal-
`cium channels sensitive to both dihydropyridines and
`high concentrations of v-conotoxin.38,39
`BON cells predominantly expressed fast inactivating
`T-type calcium currents that were not significantly af-
`fected by isradipine and v-conotoxin (Figure 8). Only a
`small percentage (15%–25%) of cells showed an addi-
`tional high voltage–activated ICa component. This com-
`ponent was variable in amplitude and increased during
`prolonged cultivation.
`
`Discussion
`Neuroendocrine cells such as pancreatic b cells12,40
`and the rat pheochromocytoma cell line PC 1241 possess
`two types of secretory vesicles: LDCVs and SSV ana-
`logues. Regulated exocytosis from neurons by SSVs17,19,29
`from pancreatic b cells21,22 and chromaffin cells18 by
`LDCVs involves the interaction of SNARE proteins.
`
`Figure 3. Pharmacological characterization of the plasma membrane
`transport of GABA in mammalian pancreatic neuroendocrine cell lines.
`The various cell lines were incubated with 3H-GABA in the absence or
`presence of unlabeled GABA (䊊), b-alanin (䉭), betain (䉱), or diamino-
`butyric acid (䊐) in the concentrations given by the abscissa. The 3H-
`GABA uptake in the absence of inhibitors was set 100%. Values are
`the means of three samples. (A) INR, (B) RIN, (C) AR42J.
`
`SNARE Proteins in Neuroendocrine Tumors
`and Cell Lines
`
`Neuroendocrine pancreatic tumors as well as hu-
`man pancreatic islets express the major proteins of the
`SNARE complex, such as SNAP 25, the syntaxins, and
`
`/ 5e0d$$0033
`
`04-12-96 15:06:48
`
`gasas
`
`WBS-Gastro
`
`Roxane Labs., Inc.
`Exhibit 1036
`Page 006
`
`

`
`May 1996
`
`GABA SECRETION FROM NEUROENDOCRINE CELLS 1601
`
`synaptobevin, including synaptobrevin II. We also found
`these proteins in the pancreatic neuroendocrine cell lines
`AR42J, RIN, INR, and BON. So far, the neuronal
`SNARE proteins have been described only in bovine
`chromaffin cells18 in normal rat pancreatic b cells and
`some pancreatic b cell lines.20–22 Thus, neuroendocrine
`cells share similar SNARE proteins with neurons.
`
`GABA Transport
`
`Neurotransmission is terminated by the rapid re-
`uptake of most neurotransmitters. The reuptake of amino
`acid transmitters GABA and glycine is mediated by
`/
`0
`/Cl
`-dependent transporter family.
`members of the Na
`So far, four GABA transporters (GAT-1–4) with differ-
`ent tissue distribution have been cloned.42 Whereas
`GAT-1 and GAT-4 are exclusively expressed in brain,43,44
`GAT-2 and GAT-3 also occur in peripheral tissues.44
`
`GAT-2 also has a strong homology with the renal betain
`transporter BGT-1, which also transports GABA.45 In
`neuroendocrine cell lines, GABA is transported by a low-
`affinity transporter different from the neuronal types. In
`RIN and INR cells, the pharmacological characterization
`of GABA uptake resembles the b-alanine–sensitive
`GAT-3 transporter.44 Interestingly, GABA transport in
`the amphicrine AR42J cells probably is mediated by a
`transporter similar to the BGT-1 transporter because of
`its sensitivity for betain.45 Work is in progress to charac-
`terize further the probable different GABA transporters
`present in the various neuroendocrine tumor cells. How-
`ever, the presence of at least one GABA transporter in
`neuroendocrine tumor cells and tissues makes them likely
`candidates for the detection by positron emission tomog-
`raphy using specific ligands.
`The GABA taken up by neuroendocrine cell lines is
`stored in an intracellular compartment, most likely the
`SSV analogues. After permeabilization of the plasma
`membrane, ATP- and carbonyl cyanide p-trifluorometh-
`oxy-phenylhydrazone–sensitive uptake of GABA is seen.
`These findings are in line with observations from the
`pancreatic b-cell line bTC3, in which ATP-sensitive
`GABA uptake has been described for immunoisolated
`synaptophysin-containing vesicles.12
`
`Mammalian Neuroendocrine Cell Lines as
`Model Systems
`
`The strongest evidence for a neuron-like function
`of SSV analogues in neuroendocrine cells is provided by
`the regulated secretion of GABA. The stored GABA is
`released upon stimulation with either 50 mmol/L KCl or
`the calcium ionophore A23187 depending on millimolar
`
`Figure 5. GABA secretion from (A) BON and (B) AR42J cells. The
`experiment was performed as in Figure 7. BON or AR42J cells were
`stimulated for 20 minutes at 36⬚C with the amount of secretagogues
`indicated by the abscissa. Release in the absence of secretagogue
`was set 100%. Values represent the means of three samples { SD.
`
`Figure 6. Ca2/-dependent secretion of GABA from mammalian pan-
`creatic neuroendocrine cell lines. Cells were prelabeled with 3H-GABA,
`washed, and incubated for 20 minutes at 36⬚C in the presence or
`absence of Ca2/. Stimulation was performed for 20 minutes at 36⬚C
`with either (A) 50 mmol/L KCl or (B) 50mmol/L A23187 in the pres-
`ence or absence of Ca2/. Release under nondepolarizing conditions
`(5 mmol/L KCl) was set 100%. A stimulated release could be ob-
`served only in the presence of Ca2/. Values are the means of three
`determinations { SD.
`
`/ 5e0d$$0033
`
`04-12-96 15:06:48
`
`gasas
`
`WBS-Gastro
`
`Roxane Labs., Inc.
`Exhibit 1036
`Page 007
`
`

`
`1602 AHNERT–HILGER ET AL.
`
`GASTROENTEROLOGY Vol. 110, No. 5
`
`Figure 7. Pharmacological characterization of ICa in (A) RIN, (B) INR,
`and (C) AR42J cells. ICa was measured from the same cells during
`alternating rectangular voltage pulses (left) and voltage ramps (right).
`Control traces (C) and traces after cumulative application of 3 mmol/
`L v-conotoxin (v-CgTx), 1 mmol/L isradipine (Israd), and 100 mmol/
`L cadmium (Cd2/) are shown.
`
`concentrations of extracellular calcium. So far, functional
`studies with SSVs have been restricted to neurons. The
`presence of both SSV analogues and LDCVs together
`with SNARE proteins in pancreatic neuroendocrine tu-
`mor cell lines makes them useful model systems to study
`aspects of regulated secretion by the two pathways.
`Like neurons, the pancreatic neuroendocrine cell lines
`are equipped with a complete set of voltage-dependent
`calcium channels, probably indicating calcium-depen-
`dent secretion. Therefore, calcium channel blockers with
`some cell-type specificity are likely candidates for future
`diagnostic and therapeutic concepts.
`
`Diagnostic and Therapeutic Impact
`
`The GABA released may stimulate GABAA recep-
`tors recently described in normal and neoplastic neuroen-
`docrine cells.46 Activation of these GABAA receptors re-
`sulted in a depolarization of the membrane, followed by
`a calcium influx, which may in turn modulate secretion
`from LDCVs.47 The role altered GABA secretion plays
`
`Figure 8. Properties of fast inactivating ICa in BON cells. (A) The whole-
`cell I–V relation was constucted from the peak ICa elicited by depolariz-
`ing voltage pulses (100 milliseconds) to potentials indicated on the
`x-axis. The holding potential was 080 mV. (B) The effects of 1 mmol/
`L isradipine (Israd), 3 mmol/L v-conotoxin (v-CgTx), and 30 mmol/L
`nickel (Ni2/) on the control ICa (C) is shown. Currents were activated
`by voltage pulses to 010 mV from a holding potential of 080 mV.
`
`in the functional state and tumor progression of neuroen-
`docrine cells remains to be determined.
`Current studies in our clinical department using radio-
`labeled GABAA-receptor ligands for scintigraphic detec-
`tion of GABA receptors show that GABA receptors can
`be used for tumor imaging in patients with neuroendo-
`crine tumors (B. Wiedenmann, S. Faiss, H. Scheru¨ bl,
`and E.-O. Riecken, unpublished observation, 1995).
`Thus, the molecular characterization of this new secretory
`pathway represents the basis for improved diagnosis and
`probably also treatment of neuroendocrine tumor disease.
`References
`
`1. Wiedenmann B, Franke WW, Kuhn C, Moll R, Gould VE. Synapto-
`physin: a marker protein for neuroendocrine cells and neo-
`plasms. Proc Natl Acad Sci USA 1986;83:3500–3504.
`
`/ 5e0d$$0033
`
`04-12-96 15:06:48
`
`gasas
`
`WBS-Gastro
`
`Roxane Labs., Inc.
`Exhibit 1036
`Page 008
`
`

`
`May 1996
`
`GABA SECRETION FROM NEUROENDOCRINE CELLS 1603
`
`2. Wiedenmann B, Huttner WB. Synaptophysin and chromogranins/
`secretogranins—widespread constituents of distinct types of
`neuroendocrine vesicles and new tools in tumor diagnosis. Virch
`Arch B 1989;8:95–121.
`3. Wiedenmann B, Ba¨der HM, Scheru¨bl H, Fett U, Zimmer T Hamm
`B Kopenhagen K, Riecken EO. Gastroenteropancreatic tumor im-
`aging with somatostatin receptor scintigraphy. Semin Oncol
`1994;21:29–32.
`4. O’Connor DT, Deftos U. Secretion of chromogranin A by peptide-
`producing endocrine neoplasms. N Engl J Med 1986;314:1145–
`1151.
`5. Eriksson B. Tumor markers for pancreatic endocrine tumors, in-
`cluding chromogranins, HCG-alpha and HCG-beta. Front Gas-
`trointest Res 1995;23:121–131.
`6. Solcia E, Sessa F, Rindi G, Villani L, Riva C, Buffa R, Capella C.
`Classification and histogenesis of gastroenteropancreatic endo-
`crine tumours. Eur J Clin Invest 1990;20(Suppl 1):72–81.
`7. Wiedenmann B, Franke WW. Identification and localization of syn-
`aptophysin, an integral membrane glycoprotein of Mr 38,000
`characteristic of presynaptic vesicles. Cell 1985;45:1017–
`1028.
`8. Navone F, Jahn R, Di Gioia G, Stukenbrok H, Greengard P, De
`Camilli P. Protein p 38: an integral membrane protein specific
`for small vesicles of neurons and neuroendocrine cells. J Cell
`Biol 1986;103:2511–2527.
`9. Reetz A, Solimena M, Matteoli M, Folli M, Takei K, DeCamilli
`P. GABA and pancreatic b-cells: co-localization of glutamic acid
`decarboxylase (GAD) and GABA with synaptic-like microvesicles
`suggests their role in GABA storage and secretion. EMBO J 1991;
`10:1275–1284.
`10. Rosewicz S, Vogt D, Harth N, Grund C, Franke WW, Ruppert
`S, Schweitzer E, Riecken EO, Wiedenmann B. An amphicrine
`pancreatic cell line: AR42J cells combine exocrine and neuroen-
`docrine properties. Eur J Cell Biol 1992;59:80–91.
`11. Ahnert-Hilger G, Wiedenmann B. The amphicrine pancreatic cell
`line AR42J secretes GABA and amylase by separate regulated
`pathways. FEBS Lett 1992;314:41–44.
`12. Thomas-Reetz A, Hell JW, During MJ, Walch-Solimena C, Jahn R,
`DeCamilli P. A g-aminobutyric acid transporter driven by a proton
`pump is present in synaptic-like microvesicles of pancreatic b
`cells. Proc Natl Acad Sci USA 1993;90:5317–5321.
`13. Ahnert-Hilger G, Grube K, Kvols L, Lee I, Mo¨nch E, Riecken EO,
`Schmitt L, Wiedenmann B. Gastroenteropancreatic neuroendo-
`crine tumours contain a common set of synaptic vesicle proteins
`and amino acid neurotransmitters. Eur J Cancer 1993;9A:1982–
`1984.
`14. So¨llner T, Whiteheart SW, Brunner M, Erdjument-Bromage H, Ger-
`omanos S, Tempst P, Rothman JE. SNAP receptors implicated
`in vesicle targeting and fusion. Nature 1993;362:318–324.
`15. Rothman JE, Orci L. Molecular dissection of the secretory path-
`way. Nature 1992;355:409–415.
`16. Rothman JE. Mechanisms of intracellular protein transport. Na-
`ture 1994;372:55–63.
`17. Su¨dhof TC, DeCamilli P, Niemann H, Jahn R. Membrane fusion
`machinery: insights from synaptic proteins. Cell 1993;75:1–4.
`18. Hodel A, Scha¨fer T, Gerosa D, Burger MM. In chromaffin cells, the
`mammalian sec1p-homologue is a syntaxin 1A-binding protein
`associated with chromaffin granules. J Biol Chem 1994;259:
`8623–8626.
`19. Su¨dhof TC. The synaptic vesicle cycle: a cascade of protein-
`protein interactions. 1995;375:645–653.
`20. Jacobsson G, Bean AJ, Scheller RH, Juntti-Berggren L, Deeney
`JT, Berggren PO, Meister B. Identification of synaptic proteins
`and their isoform mRNAs in compartments of pancreatic endo-
`crine cells. Proc Natl Acad Sci USA 1994;91:12487–12491.
`21. Sadoul K, Lang J, Montecucco C, Weller U, Catsicas