Chemico-Biological Interactions the Food Contaminant Semicarbazide

Chemico-Biological Interactions 183 (2010) 40–48

Contents lists available at ScienceDirect

Chemico-Biological Interactions

journal homepage: www.elsevier.com/locate/chembioint

The food contaminant semicarbazide acts as an endocrine disrupter: Evidence from an integrated in vivo/in vitro approach

Francesca Maranghi a,∗, Roberta Tassinari a, Daniele Marcoccia a, Ilaria Altieri a, Tiziana Catone b, Giovanna De Angelis b, Emanuela Testai b, Sabina Mastrangelo c, Maria Grazia Evandri c, Paola Bolle c, Stefano Lorenzetti a a Department of Food Safety and Veterinary Public Health, Istituto Superiore di Sanità, Viale Regina Elena 299, 00161 Rome, Italy b Department of Environment and Primary Prevention, Istituto Superiore di Sanità, Viale Regina Elena 299, 00161 Rome, Italy c Department of Physiology and Pharmacology “Vittorio Erspamer” Sapienza University of Rome, Piazzale Aldo Moro 5, 00185 Rome, Italy article info abstract

Article history: Semicarbazide (SEM) is a by-product of the blowing agent azodicarbonamide, present in glass jar-sealed Received 3 August 2009 foodstuffs mainly baby foods. The pleiotropic in vivo SEM toxicological effects suggested to explore its Received in revised form possible role as endocrine modulator. Endocrine effects of SEM were assessed in vivo in male and female 18 September 2009 rats after oral administration for 28 days at 0, 40, 75, 140 mg/kg bw pro die during the juvenile period. Accepted 21 September 2009 Vaginal opening and preputial separation were recorded. Concentration of sex steroid in blood, the ex Available online 27 September 2009 vivo hepatic aromatase activity and testosterone catabolism were detected. The in vitro approach to test SEM role as (anti)estrogen or N-methyl-d-aspartate receptors (NMDARs)-(anti)agonist included dif- Keywords: Semicarbazide ferent assays: yeast estrogenicity, MCF-7 proliferation, stimulation of the alkaline phosphatase activity Food contaminant in Ishikawa cells and LNCaP-based NMDAR interference assay. In vivo SEM-treated female rats showed Endocrine disrupters delayed vaginal opening at all tested doses, whereas in males preputial separation was anticipated at SEM Sex steroids 40 and 75 mg/kg and delayed at 140 mg/kg, the latter effect probably due to the signiﬁcantly decreased Estrogen receptor alpha body weight gain seen at the higher dose in both sexes. Serum estrogen levels were dose-dependently N-methyl-D-aspartate receptor reduced in treated females, whereas dehydrotestosterone serum levels were also decreased but a clear dose–response was not evidenced. Testosterone catabolism was altered in a gender-related way, aromatase activity was increased in treated males at 75 and 140 mg/kg and in females in all dose groups. In the three estradiol-competitive assays, SEM showed a weak anti-estrogenic activity, whereas in the LNCaP-based NMDAR interference assay SEM activity resembled MK-801 antagonist effect. SEM appeared to act as an endocrine disrupter showing multiple and gender speciﬁc mechanisms of action(s). A possible cascade-mechanism of SEM on reproductive signalling pathways may be hypothesized. Such in vivo–in vitro approach appeared to be an useful tool to highlight SEM activity on endocrine homeostasis. © 2009 Elsevier Ireland Ltd. All rights reserved.

1. Introduction

Endocrine disrupters (EDs) are a broad group of natural or Abbreviations: DDE, 1,1-dichloro-2,2-bis(p-chlorophenyl)ethylene; E2, 17␤- man-made chemicals which – together with their by-products – estradiol; ALP, alkaline phosphatase; ADC, azodicarbonamide; BW, body weight; can cause adverse health effects in adult organisms, in the off- CNS, central nervous system; DHT, dihydrotestosterone; DMEM, Dulbecco’s Min- imal Essential Medium; EDs, endocrine disrupters; ERs, estrogen receptors; ERE, spring or in susceptible (sub)populations altering the homeostasis estrogen-responsive sequences; EFSA, European Food Safety Authority; FBS, foetal of the endocrine system, mainly involving sex steroid and thy- bovine serum; G6PDH, G6P-dehydrogenase; GABa, gamma-aminobutiric acid; G6P, roid hormones [1]. Experimental studies put into evidence that EDs glucose-6-phosphate; GnRH, gonadotropin-releasing hormone; LH, luteinizing hor- act through different mechanisms and have different targets: the mone; NMDARs, N-methyl-d-aspartate receptors; PND, postnatal day; RLM, rat liver microsomes; SEM, semicarbazide; SSAOs, semicarbazide-sensitive amine oxidases; SD, standard deviation; TST, testosterone; YES, yeast estrogenicity screen; ␤-gal, ␤-galactosidase. ∗ Corresponding author. Tel.: +39 06 49902527; fax: +39 06 49902658. (T. Catone), [email protected] (G. De Angelis), E-mail addresses: [email protected] (F. Maranghi), [email protected] (E. Testai), [email protected] [email protected] (R. Tassinari), [email protected] (S. Mastrangelo), [email protected] (M.G. Evandri), (D. Marcoccia), [email protected] (I. Altieri), [email protected] [email protected] (P. Bolle), [email protected] (S. Lorenzetti).

0009-2797/$ – see front matter © 2009 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.cbi.2009.09.016 F. Maranghi et al. / Chemico-Biological Interactions 183 (2010) 40–48 41 homeostasis of sex steroids, thyroid hormones and their cascade 2. Materials and methods effects on reproduction and fertility being better characterised. The most vulnerable and susceptible phases of life cycle to ED adverse 2.1. Chemicals, reagents and cell culture media effects are pregnancy, childhood and puberty when the endocrine system plays a key role in the development and differentiation of Chemicals and reagents, including SEM hydrochloride (CAS no. the whole organism. In humans, EDs have been mainly associated 563-41-7, purity ≥98–99%), ␤-galactosidase (␤-gal) used in the YES with apparent changes in both male and female reproductive health assay, NMDA (CAS no. 6384-92-5), glycine (CAS no. 54-40-6) and [2]. (+)-MK-801 hydrogen maleate (CAS no. 77086-22-7) used in the Several potential food contaminants have been shown to act NMDAR-interference assay, were purchased from Sigma–Aldrich as EDs. Semicarbazide (SEM) is a by-product of azodicarbonamide (Milan, Italy). All the ingredients for yeast media, unless otherwise (ADC), which is used to foam the plastic gaskets of metal lids. stated, were research grade biochemicals suitable for cell culture In the food-production chain, SEM is released during manufac- and purchased from Sigma–Aldrich (Milan, Italy): yeast minimal ture when packaged foods are heated to ensure a tight seal and medium and yeast growth medium preparations have been previ- it is present as food contaminant in baby foods, fruit juices, jams ously described [17]. Chlorophenol red-␤-d-galactopyranoside was and conserves. Moreover, SEM was also found in different food- obtained from Roche Diagnostics (Monza, Milan, Italy). Cell culture stuffs such as powder and liquid milk, egg and whey powder media, supplements for MCF-7 and Ishikawa cells were purchased [3,4]. from Sigma–Aldrich (Karlsruhe, Germany), whereas RPMI1640 w/o Following the European Food Safety Authority (EFSA) warning, phenol red and foetal bovine serum (FBS) to grow LNCaP cells have the European Commission banned ADC as blowing agent in plastics been provided by Invitrogen (S. Giuliano Milanese, Italy). Estrogen- since August 2005 [5] asking for further data on potential toxicolog- free FBS used in YES, ALP and E-screen assays were obtained from ical effects of SEM. However, ADC is still used in some countries and Hyclone (Logan, Utah, USA) and charcoal-stripped FBS to perform this may explain the presence of SEM in a wide range of breaded NMDAR-interference assay by Bio-Whittaker (Lonza Milano srl, food products [6]. Other sources of SEM include the carrageenan Treviglio, Italy). Cell culture lysis buffer (CyLyse®) and staining extraction process and, in some countries, hypochlorite treatments solution (containing 4,6-diamidino-2-phenylindole/DAPI) were [6–8]. EFSA indicated that risks for consumers appear to be low purchased by Partec GmbH (Münster, Germany). The cell prolif- (SEM levels measured in foods up to 25 ␮g/kg = 25 ppb): never- eration kit “Cell Titer 96 AQueous One Solution Cell Proliferation theless, infants may be a sub-population at increased risk of SEM Assay” to perform the LNCaP-based NMDAR-interference assay exposure [9]. (MTS assay) has been purchased by Promega (Promega Italia, SEM is known to inhibit the following enzymes: (i) Milano, Italy). semicarbazide-sensitive amine oxidases (SSAOs), involved in Detection kits for sex steroid hormones (E2 and dihydrotestos- the oxidative metabolism of dietary and environmental xeno- terone—DHT) have been purchased from Perkin–Elmer (DELFIA® biotics [10–12]; (ii) lysyl oxidase, a key enzyme involved in the Estradiol Kit #1244-056) and Alpha Diagnostic International USA stabilization of extracellular matrix by cross-linking of proteins (DHT ELISA Kit #1940) respectively. such as collagen and elastin [13] and (iii) glutamic acid decarboxy- Roche GmbH (Mannheim, Germany) supplied NADPH, NADP, lase, synthesizing gamma-aminobutiric acid (GABA), the principal glucose-6-phosphate (G6P), and G6P-dehydrogenase (G6PDH). non-peptidal neurotransmitter in the central nervous system Defatted bovine serum albumin was from Serva, Feinbiochemicsa (CNS) [14,15]. (Heidelberg, Germany). Testosterone (TST), corticosterone (as Targets of SEM toxicological effects include skeletal, cardio- internal standard), 4-androsten-3,17-dione (4), and 2␣-, 16␣-, vascular and neurological systems. Experimental data from a 6␤-, 16␤-hydroxy-TST (2␣,16␣,6␤,16␤) were purchased from study conducted in rats exposed during the juvenile period Sigma Chemicals Co. (St. Louis, MO). 2␤-, 6␣-, 7␣-hydroxy-TST (2␤, (the most appropriate window of exposure for children sus- 6␣,7␣) were supplied by Steroids Reference Collection (D.N. Kick, ceptibility [16]) showed that SEM exerted pleiotropic effects Department of Chemistry, Queen Mary College, London, England). on different organs/systems and suggested a possible role as 1.2 ␮Ci/ml [1␤-3H]-4 was supplied from New England Nuclear- an ED [4]. No data are available on potential SEM effects on Perkin–Elmer. All other analytical grade chemicals were obtained endocrine system, in particular on sex steroid metabolism and from commercially available sources. on SEM interaction with estrogen receptors (ERs). Aim of the present study was to evaluate the potential effects of SEM on 2.2. Animals and treatments endocrine homeostasis by means of an in vivo/in vitro integrated approach. All experiments on animals were performed according to the Taking into account that children can represent a sub- European Community Council Directive 86/609/EEC as described population with relatively high exposure to SEM through the in Maranghi et al. [4]. diet, Sprague–Dawley rats were treated with SEM per os during Twenty dams (5 per group) of Sprague–Dawley rats (Harlan, the juvenile period, namely from weaning to sexual maturity. Italy) with offspring were kept under standard laboratory con- Parameters of sexual development in both sexes were evalu- ditions (22 ± 0.5 ◦C room temperature, 50–60% relative humidity, ated, serum sexual steroid levels were measured as well as 12 h dark–light alternation with 12–14 air changes per hour) with hepatic ex vivo testosterone (TST) catabolism and aromatase activ- water and food (4RF25 GLP “Top Certificate” diet purchased from ity. Mucedola, Milan, Italy) available ad libitum. Weaning rats (23-day To evaluate in vitro the ER␣ responsiveness to SEM and SEM old; male weight: 50–60 g; female weight: 40–50 g) were housed competitiveness to 17␤-estradiol (E2)-dependent activities, the in single cages in order to obtain at sacrifice at least 10 ani- following assays were utilized: the yeast estrogenicity screen mals/sex/group. Animals were treated for 28 days, from postnatal (YES), the MCF-7 proliferation (E-screen) and the alkaline phos- day (PND) 23 to sexual maturity (PND50), with 0 mg/kg (control, phatase (ALP) assay in the human endometrial adenocarcinoma vehicle only), 40 mg/kg (SEM40), 75 mg/kg (SEM75) and 140 mg/kg cell line Ishikawa. Furthermore, taking into account the SEM (SEM140) body weight (BW) pro die of SEM dissolved in distilled interference with GABA system, the potential interaction with water and given by oral gavage. Dose levels were selected on the the N-methyl-d-aspartate receptors (NMDARs) was also investi- basis of the available literature data on animal experiments [4,13]. gated. The concentration of SEM solutions was calculated in order to 42 F. Maranghi et al. / Chemico-Biological Interactions 183 (2010) 40–48 supply volumes of 1 ml each 200 g of body weight. Animals were 4):tetrahydrofuran (35:55:10) with a flow-rate of 0.8 ml/min. The checked daily for their health status. Every 2 days, individual BW analytes and the IS were identified by comparing their retention and food consumption were recorded and SEM dose level was times with pure analytical standards; their amounts were quanti- adjusted according to the weight gain for each animal. Starting fied by referring to a calibration straight line (average correlation from PND35 (females) or PND42 (males), the parameters of sex- coefficient R2 = 0.995–0.999) prepared with known amounts of the ual development, vaginal patency and preputial separation, were analytical standards of TST and its metabolites (range 10 ␮Mto observed. At PND62, blood samples were collected by intracardiac 1 mM; 1–75 ␮M, respectively). The coefficient of variation was <8% puncture after anaesthesia with isofluorane, for the evaluations of and the extraction efficiencies were 102 ± 8% [mean ± standard serum DHT and E2 levels. All the animals were sacrificed by asphyx- deviation (SD)] for TST and ≥98% for the different metabolites. iation with CO2: livers were immediately excised, weighted and washed in buffer solution at +4 ◦C; then they were gently wiped 2.5. YES assay and stored at −80 ◦C for microsomal preparation. The Saccharomyces cerevisiae recombinant hER␣- estrogen- 2.3. Sexual steroids in rat serum responsive sequences (ERE)-lacZ strain [17] was kindly supplied by Prof. S. Ottonello (Parma University, Italy) on behalf of Prof. J. Blood samples were allowed to clot at room temperature for Sumpter (Brunel University, Uxbridge, UK). 1 h. Afterwards, they were centrifuged 15 min at 2200 × g and This yeast strain is transfected with both an integrated plasmid the resulting sera were stored at −80 ◦C for further examinations. carrying the DNA sequence of the human ER␣ and an expression Rat DHT and E2 serum levels were determined by ELISA (Alpha plasmid carrying the lacZ reporter gene (encoding the enzyme ␤- Diagnostic International Dihydrotestosterone ELISA Kit #1940) and gal) under the control of ERE. ␤-gal is secreted into the medium DELFIA (DELFIA® Estradiol Kit #1244-056, Perkin–Elmer) methods, where it breaks down the chromogenic substrate CPRG, thus allow- respectively, following manufacturer’s instructions. Measurements ing to monitor indirectly receptor activity by measurement of the have been performed using a Wallac 1420 VICTOR3TM Multilabel absorbance at 540 nm [17]. microplate reader (Perkin–Elmer, Life Sciences, Milan, Italy). SEM was dissolved and serially diluted in yeast minimal medium and tested at the concentration range 0.02–120 mM. Each concen- 2.4. Preparation of rat liver microsomes (RLM) tration (in 10 ␮l) was transferred into 4 wells of 96-well plates. A suspension containing 0.08 × 106 yeast cells and CPRG (1%, w/v) RLM were prepared as previously described in Testai and Vit- was added to each well (final volume 200 ␮l). After 48 h of incuba- tozzi [18]. The microsomes were re-suspended in Tris–Sucrose tion at 32 ◦C, plates were shaken to re-suspend yeast cells and the buffer (pH 7.4) and then stored at −80 ◦C until the enzymatic activ- absorbance was measured (Bio-Rad microplate reader, Hercules, ity assays were performed. The protein content was determined CA, USA). according to Oyama and Eagle [19] using bovine serum albumin as To evaluate synergistic or antagonistic effects, fixed concen- standard. trations of SEM (3, 5, 15, and 30 mM) were added to increasing concentrations of E2 (0.001–60 nM), whereas, to asses the pos- 2.4.1. Aromatase activity assay sible direct interference between SEM and ␤-gal, the latter was Aromatase activity (CYP19) of RLM was measured by a tritiated- solubilised in yeast growth medium at the concentration range of water formation assay as described in You et al. [20]. RLM samples 0.01–0.5 U/ml without or with SEM (3, 5, 15 and 30 mM) and in (0.4 mg/ml) were incubated with 4 mM G6P, 2 U/ml G6PDH, absence of yeast cells. 143.5 ␮M unlabeled androsten-ene-3,17-dione(4) and 1.2 ␮Ci/ml [1␤-3H]-4. The reaction was started by the addition of 0.5 mM 2.6. MCF-7 proliferation assay (E-screen) NADP and carried out for 80 min at 37 ◦C under gentle shaking. Blanks were carried out using standard incubation mixture without The E2-responsive MCF-7 breast cell line was purchased from NADP. The reaction was stopped by the addition of 5 ml chloro- American Type Culture Collection (ATTC, Hanassas, USA). Cells were form and 0.5 ml deionized water. Samples were then extracted cultured in Dulbecco’s Minimal Essential Medium (DMEM) with- and centrifuged for 5 min at 800 × g, the radiolabeled steroids were out phenol red supplemented with 2% penicillin/streptomycin and removed by the addition of a suspension of charcoal–dextran mix- 5% heat inactivated FBS [26]. MCF-7 cells were seeded in 24-well ture. The release of tritiated water was measured using LS6500 plates at a concentration of 0.02 × 106 cells per well 24 h prior to Beckman liquid scintillation counter. incubation with 2 nM to 5 mM SEM, with or without co-incubation with 100 pM E2 in 5% E2-free FBS. Cell proliferation was assessed 2.4.2. TST hydroxylase activity assay after 7 days by cytofluorimetric analysis. Briefly, medium was aspi- The RLM activity of TST hydroxylase was determined by the rated, 150 ␮l lysis buffer (CyLyse®; a detergent-based lysis agent) method of Sanwald et al. [21] with minor modifications, in the were added to each well, and cells were incubated 5 min at +20 ◦C. linear range of dependence on time, substrate and protein con- Nuclei were transferred quantitatively into vials with 750 ␮l stain- centrations. The standard incubation mixture (0.5 ml) contained: ing solution (containing 4,6-diamidino-2-phenylindole/DAPI) and 1 mg/ml microsomal protein, 2 mM G6P, 2 U/ml G6PDH, 1 mM the flow cytometric measurement of the cell number and cellu- NADP in 50 mM Tris buffer, 1 mM EDTA (pH 7.4). The mixtures lar distribution in the cell cycle phases were determined (Partec were kept at 37 ◦C for 3 min before starting the reaction. The PAS cytofluorimeter, Partec GmbH, Münster, Germany; data not enzymatic incubation, at 37 ◦C under gentle shaking, was initiated shown). by the addition of the substrate (1 mM TST, final concentration) and after 10 min was stopped by the addition of 3.5 ml of ice- 2.7. ALP assay in cultured Ishikawa cells cold dichloromethane containing corticosterone (6.25 ␮M final concentration) as internal standard (IS). Blanks were carried out Ishikawa cells were kindly provided by Ken Korach (NIEHS, using standard incubation mixture without NADP. TST, its metabo- USA). Cells were cultured in DMEM with 2 mM glutamine, 2% peni- lites and IS were extracted and then measured by HPLC analysis cillin/streptomycin, and supplemented with 10% FBS. ( = 245 nm). The column (Restek C18 pinnacle ODS ammine) was The ALP assay was performed as reported by Lehmann et al. eluted isocratically with a mobile phase of methanol:water (pH [22]. Briefly, Ishikawa cells were seeded in 96-well plates at a F. Maranghi et al. / Chemico-Biological Interactions 183 (2010) 40–48 43 concentration of 0.02 × 106 cells per well in 5% E2-free FBS 24 h prior to incubation with 2 nM to 2 mM SEM with or without co- incubation with 10 nM E2. After 72 h, cultured cells were washed 3 times with Ca2+- and Mg2+-free PBS and lysed, by freeze and thawing, upon at least 20 min at −80 ◦C and 5 min at 20 ◦C. Then, 50 ␮l of 4-nitrophenylphosphate (5 mM in a buffer containing 1 M diethanolamine and 0.24 mM MgCl2, pH 9.8) were added to each well. After 5 min at room temperature, ALP activity was assessed by photometric measurement of the formation of 4-nitrophenol at 420 nm (Genios, Tecan, Crailsheim, Germany) every 10 min for 1 h. To exclude a cytotoxic effect of SEM on Ishikawa cells, two 24- well plates containing 0.115 × 106 cells per well and 1, 2, and 5 mM SEM were prepared: the first plate was used for cell counting, while the second one was used for ALP activity assessment. Results are expressed as amount of 4-nitrophenol per million cells.

2.8. LNCaP-based NMDAR-interference assay (MTS assay)

The human prostate cancer cell line LNCaP was purchased from European Collection of Cell Cultures (ECACC, Salisbury, UK). Cells were cultured in RPMI1640 w/o phenol red, supplemented with 10% FBS, 2 mM l-glutamine, 1 mM piruvic acid, 100 U/ml penicillin ◦ and streptomycin, at 37 C, 5% CO2, in water-saturated atmo- sphere. In order to assess SEM ability to act as a NMDAR interfer- Fig. 1. Parameters of sexual development. (A) Timing of vaginal opening in female ing compound, NMDA (NMDAR agonist), MK-801 (high affinity, Sprague–Dawley rats treated with 0, 40, 75 and 140 mg/kg bw of SEM. (B) Timing non-competitive NMDAR antagonist) and SEM itself (putative of preputial separation in male Sprague–Dawley rats treated with 0, 40, 75 and 140 mg/kg bw of SEM. Stars indicate the statistical significance of results obtained NMDAR agonist/antagonist) have been tested by MTS assay in treated animals with respect to control group: ***p < 0.001. using the “Cell Titer 96 AQueous One Solution Cell Proliferation Assay” (Promega kit and following the manufacturer’s instructions). 3. Results Briefly, LNCaP cells were plated at a density of 5 × 103 cells per well in 96-well plates in 100 ␮l culture medium per each well. 3.1. In vivo general toxicity Treatments with different concentrations of NMDA (in presence of the glycine cofactor), MK-801 (both dissolved in PBS without Ca2+ Increased mortality (+20% in comparison to control group) was and Mg2+), SEM (dissolved in water) as well as their combinations observed in SEM75- and SEM140-treated rats (both sexes taken were performed in triplicate at the concentrations indicated in fig- together). Statistically significant dose-dependent decrease in body ure’s legends (see Section 3). After 3 days of growth, cell cultures weight gain was observed in male rats at all dose levels; no other were quenched adding 20 ␮l (1/5 cell culture volume) of the “Cell clinical signs of general toxicity were present during the treatment Titer 96 AQueous One Solution Reagent” and further incubated at period (PND23–50). Food consumption was significantly decreased 37 ◦C in the dark. Absorbance was measured at 490 nm using a Wal- at the higher dose level both in male and female rats, whereas was lac 1420 VICTOR3TMMultilabel microplate reader (Perkin–Elmer slightly but significantly increased at SEM75 in females only (data Life Sciences, Milan, Italy). not shown).

3.2. Parameters of sexual development 2.9. Statistical analysis Timing of vaginal opening was significantly delayed at SEM140 Differences among controls and SEM-treated groups were anal- (Fig. 1A). Preputial separation showed significant anticipation at ysed by means of different tests depending on the measured SEM40 and SEM75 whereas at SEM140 was significantly delayed endpoint. Unless otherwise stated, all values are expressed as (Fig. 1B). mean ± SD. In vivo data were analysed by one-way ANOVA followed by the 3.3. Sexual steroids in rat serum appropriate test for multiple comparisons using GraphPad Prism 4 (version 4.01 for Windows) or Sigma-Stat 3.0 (SPSS, Chicago, IL, E2 serum levels were significantly and dose-dependently USA) software. Differences between groups were considered sig- decreased in females of all treatment groups (Fig. 2A). In compar- nificant with p ≤ 0.05. ison to controls, DHT serum levels were significantly decreased at The E2 and DHT serum levels were grouped according to the SEM40 and SEM75 (Fig. 2B). treatment for both hormones and the differences among groups were evaluated using Student’s t-test. Differences among treatment 3.4. Aromatase activity assay and control groups were considered statistically significant if the p value was ≤0.01. At sacrifice (PND62), aromatase activity has been analysed in In vitro data were analysed either by Holm–Sidak test (ER␣ both sexes of the SEM-treated Sprague–Dawley rats. In males, responsiveness) or Student’s t-test (NMDAR responsiveness) and SEM75 and SEM140 treatments induced an increase in aromatase differences among treated and control groups were considered dif- activity in comparison to control animals, although the statistical ferent with a value ≤0.05. significance was attained only in the highest dose group (Fig. 3A). 44 F. Maranghi et al. / Chemico-Biological Interactions 183 (2010) 40–48

Fig. 4. TST hydroxylase activity assay. Testosterone metabolism in hepatic micro- Fig. 2. Sexual steroids in rat serum. (A) 17␤-Estradiol (E2) serum levels in female somal samples from male (A) and female (B) rats treated with 0 mg/kg (open bars), Sprague–Dawley rats treated with 0, 40, 75 and 140 mg/kg bw of SEM. (B) Dihy- 40 mg/kg (squared bars), 75 mg/kg (closed bars) and 140 mg/kg (hatched bars) bw drotestosterone (DHT) serum levels in male Sprague–Dawley rats treated with 0, of SEM, respectively. Each set of bars represent the production of single testosterone 40, 75 and 140 mg/kg bw of SEM. Stars indicate the statistical signiﬁcance of results metabolite (as indicated in the abscissa). Stars indicate the statistical signiﬁcance obtained in treated animals with respect to control group: **p < 0.01. of results obtained in treated animals with respect to control group: *p < 0.05; **p < 0.01; ***p < 0.001.

Similarly, in female rats (Fig. 3B) aromatase activity was increased in all treatment groups, with statistical signiﬁcance evidenced at SEM75 and SEM140.

3.5. TST hydroxylase activity assay

The rate of TST hydroxylation/dehydrogenation analysed at sacrifice (PND62), represents the first step of TST degradation and is catalyzed, mainly in the liver, by different CYP families, at several positions with regio- and stereo-selective reactions: 6␤ (CYP3A2); 16␣,16␤,2␣ and androstenedione (CYP2C11 and 2B1); 6␣ and 7␣ (CYP2A1) [23]. The pattern of TST catabolism is clearly different in control males (Fig. 4A) and females (Fig. 4B) due to the presence of gender-related CYP-dependent activities, mainly associated with CYP2C family. In addition, SEM treatments elicited gender- specific effects. In males, some CYP-dependent activities were significantly increased at the highest dose (SEM140) namely: 7␣ (CYP2A1 marker), 6␤ (CYP3A2 marker), 16␣,2␣ and androstenedione (catalysed by both CYP2C11 and CYP2B1). The last three activities were significantly higher at SEM75 in comparison to control, whereas at SEM40, TST metabolism was not affected (Fig. 4A). No significant changes in the rate of 6␣,16␤ and 2␤ hydroxylation were observed in the liver of male rats (Fig. 4A). In females at both SEM75 and SEM140, the formation of 6␣ (CYP2A1 marker) was dramatically decreased (more than 80%), whereas 16␣ and 16␤ production (attributable mainly to CYP2C family) was significantly increased (Fig. 4B).

3.6. In vitro ER˛ responsiveness

3.6.1. YES assay No cytotoxicity on the yeast was observed as revealed by count- Fig. 3. Aromatase activity assay. Aromatase activity in hepatic microsomes from ing of viable cells [mean cell number per well: (6.0 ± 0.5) × 106 male (A) and female (B) rats treated with 0, 40, 75 and 140 mg/kg bw of SEM. Stars indicate the statistical signiﬁcance of results obtained in treated animals with with 30 mM SEM and (5.4 ± 0.8) × 106 with 60 nM E2]. At the respect to control group: *p < 0.05; ***p < 0.001. used concentrations (0.02–120 mM), SEM was not able to acti- F. Maranghi et al. / Chemico-Biological Interactions 183 (2010) 40–48 45

Fig. 5. Yeast estrogenicity screen (YES) assay. (A) Concentration–response curves of the activation of the human estrogen receptor (ER)-␣ (hER␣). SEM (0.02–30 mM); E2 (0.001–60 nM); (E2 + SEM): E2 (0.001–60 nM) in the presence of ﬁxed SEM concentrations (3, 5, 15, and 30 mM). Points represent the mean ± SD of at least three independent experiments. *p < 0.05 (Holm–Sidak test) and #p < 0.05 (Dunn’s test) vs E2. (B) Activity of ␤-gal (0.01–0.5 ␤-gal) and ␤-gal + SEM (0.01–0.5 ␤-gal in the presence of 3, 5, 15, and 30 mM semicarbazide hydrochloride). Bars represent the mean ± SD of three independent experiments. *p < 0.001 vs ␤-gal alone (Holm–Sidak test).

vate the hER␣ (Fig. 5A). When fixed concentrations of SEM (3, a decrease in the number of cells per well [mean cell num- 5, 15, and 30 mM) were tested in combination with increasing ber per well without SEM: (8.3 ± 0.16) × 105; with 5 mM SEM: E2 concentrations (0.001–60 nM), SEM markedly reduced in a (6.6 ± 0.41) × 105; with 10 nM E2: (8.1 ± 0.42) × 105; with 10 nM dose-dependent and statistically significant way the response of E2+5mMSEM: (7.2 ± 0.07) × 105]. Co-incubation with 10 nM E2 E2, shifting its concentration–response curve downwards (from induced a decrease in the E2 ALP enhancement activity in a dose- −5.5 ± 0.68% with 3 mM SEM to −78.5 ± 5.2% with 30 mM SEM) dependent way (Fig. 6B). (Fig. 5A). Anyway, SEM (3, 5, 15, and 30 mM) showed also a significant, although lower, inhibitory activity on ␤-gal (Fig. 5B). 3.7. In vitro NMDAR responsiveness

3.6.2. E-screen In order to test SEM interference with NMDAR activities, LNCaP At the tested concentrations (2 nM to 5 mM), SEM was not able cells have been treated with the NMDAR agonist NMDA (plus to induce proliferation in MCF-7 breast cancer cell line. When tested glycine) at three different concentrations (100 nm = no effects on in co-incubation to E2 100 pM, SEM (at 1, 2, and 5 mM) showed a LNCaP cell viability and indirect proliferation; 250 nM = almost no statistically signiﬁcant, dose-dependent, anti-proliferative activity effects and 500 nM = about −49% decrease) with either 50 ␮MMK- (Fig. 6A). 801 or 2 mM SEM. Concentrations used for each compound have been previously selected in the same assay, as summarized in 3.6.3. ALP activity Table 1, on the basis of the opposite effects observed singularly When used alone, SEM (2 nM to 2 mM) was not able to enhance with SEM, the antagonist MK-801 and the agonist NMDA. As shown ALP activity in an endometrial-derived Ishikawa cell line, while in Fig. 7, an almost complete rescue of LNCaP proliferation have the highest tested concentration (2 mM) decreased 10 nM E2 been evidenced when either MK-801 or SEM have been added response (data not shown). In the 24-well plate, SEM was slightly together with 500 nM NMDA (about −10% and −14%, respectively), cytotoxic at the highest tested concentration (5 mM) inducing whereas no statistically signiﬁcant differences have been observed 46 F. Maranghi et al. / Chemico-Biological Interactions 183 (2010) 40–48 500 nM NMDA + 50 nM Gly M MK-801 ␮ M MK-801 200 ␮

Fig. 6. (A) MCF-7 proliferation assay (E-screen). Number of MCF-7 cells per well

M MK-801 100 after 7 days treatment with 1, 2, and 5 mM SEM in the presence or not of 100 pM ␮ § E2. Bars are the mean ± SD. p < 0.001 vs DMSO (t-test); *p ≤ 0.001 vs E2 or DMSO 50 alone (Holm–Sidak test). (B) Alkaline phosphatase (ALP) activity in Ishikawa cells. Formation of 4-nitrophenol (4-NP) (pmol/min/million cells) in Ishikawa cells after 72 h treatment with SEM (1, 2, and 5 mM), E2 10 nM and (E2 + SEM) when SEM (1, 2, and 5 mM) was added to 10 nM E2. Bars represent the mean ± SD of at least § three independent experiments. p ≤ 0.001 vs DMSO (t-test); #p < 0.05 vs 10 nM E2 (Holm–Sidak test). M MK-801 ␮ 2 mM SEM 3 mM SEM 4 mM SEM 5 mM SEM A+1nMGly 50nMNMDA+5nMGly 250nMNMDA+25nMGly 25nMNMDA+2.5nMGly 100nMNMDA+10nMGly M MK-801 25 ␮

Fig. 7. LNCaP-based NMDAR-interference by SEM. LNCaP cells have been treated with NMDA at three different concentrations (100, 250 and 500 nM) in presence or absence of 50 ␮M MK-801 or 2 mM SEM. M MK-801 10 ␮ by LNCaP co-treatments with either MK-801 or SEM plus 100 and 250 nM NMDA. This result clearly shows that SEM behaviour on CTRL 1 nM NMDA + 0.1 nM Gly 10 nM NMD 100 119CTRL 0.5 mM SEM 1 mM SEM 128 123 135 95 72 100 102CTRL 1 105LNCaP 99 cell viability/proliferation 100 resembles 102 MK-801 activity and 94 51 that their effects in NMDA LNCaP co-treatments clearly overlap, although SEM potency results to be lower than MK-801.

4. Discussion

In this study, SEM in vivo treatment was performed dur- % proliferation SEM 100 109 114 125 91 85 86 % proliferation NMDAR antagonist % proliferation NMDAR agonist

Table 1 LNCaP-based NMDAR-interference. Selection of individual effective concentrations by cell viability and indirect proliferation (MTS) assay. ing juvenile period, being the phase from weaning to sexual F. Maranghi et al. / Chemico-Biological Interactions 183 (2010) 40–48 47 maturity the appropriate window of exposure to evidence pos- puberty implicate complex neuroendocrine mechanisms that can sible interference with endocrine homeostasis, considering the be qualitatively and quantitatively different in both sexes. They key role played by the endocrine system in such a critical involve ontogenic changes in GnRH and gonadotropin release, but phase of development and maturation [24]. Dose levels were are also connected with modifications in the expression of recep- selected on the basis of the available literature data on animal tors and in the concentration and release of several hypothalamic experiments [13] and of the previous in vivo data on peri- neurotransmitters, such as the excitatory and inhibitory amino acid pubertal SEM oral administration [4], showing pleiotropic effects systems [32]. on several organs/tissues which are also targets of E2 actions It has been demonstrated that E2 alters glutamate transmission [4]. to GnRH neurons and that glutamate, the main NMDAR neu- Since EDs have been demonstrated to exert their action through rotransmitters, plays an important role in the neural control of different mechanisms [25], several hypotheses for SEM activity ovulation [33]. Indeed, ER␣ and NMDAR subunit NR2b have been on endocrine homeostasis have been explored by applying an shown to colocalize in specific areas of the CNS [34]. Moreover, the integrated in vivo/in vitro approach. Parameters of sexual devel- presence of the NMDAR on neuroendocrine secretory vesicles, its opment in both sexes and serum levels of sexual steroid hormones co-expression with GnRH and ER␣ and its regulation by E2 have were evaluated in vivo, in order to understand the possible SEM- been demonstrated [35]. The SEM antagonist effect recorded both induced ED activity, whereas the ex vivo biochemical measurement in NMDAR and ER␣ can account for the gender differences due to and the in vitro activities were carried out in order to clarify sex-specific central and peripheral modulations of NMDARs [31] the possible mechanism(s) of action. Alteration of CYP-dependent as well as the possible role of SEM at hypothalamic level, suggest- TST catabolism and CYP19/aromatase activity would imply that ing its putative involvement in the cascade events which, starting hormone level changes may be due to an unbalanced steroid hor- at the central level, lead to the endocrine control of reproductive mone metabolism. The three different assays used to evaluate function. in vitro the ER␣ responsiveness to SEM are routinely indicated In male treated rats, the effects on aromatase activity (increas- as screening tools for EDs [26] and have also been used to ing the transformation of TST and androstendione in E2) and the test SEM competitiveness to 17␤-estradiol (E2)-dependent activi- enhanced TST catabolism, although limited to 3 specific posi- ties. tions, were consistent with the decreased level of circulating To make the picture complete, the potential interaction with DHT. This result seems to suggest that in males sex steroid the NMDAR was investigated, taking into account the SEM inter- pathways can be altered at the level of androgen-estrogens bal- ference with GABA system. In this respect, a LNCaP-based NMDAR ance. Data in female rats indicated a gender-dependent action interference assay was set up, taking into account that NMDAR of SEM on steroid hepatic metabolism with CYP450 isoforms genes are expressed both in normal and cancer prostate cells involved in TST hepatic metabolism differently affected by the and that they are responsive to both NMDAR agonists (i.e. gluta- treatment, when compared to males, resulting in decreased E2 mate, aspartate, NMDA) and antagonists (i.e. MK-801). NMDAR in serum levels. Similar effects were seen after 1,1-dichloro-2,2-bis(p- prostate cells is necessary to coordinate neural circuits responsi- chlorophenyl)ethylene (DDE, a stable metabolite of the pesticide ble for reproductive male behaviour and it takes part in essential 1,1,1-trichloro-2,2-bis(p-chlorophenyl)ethane) treatment in vivo reproductive functions such as seminal emission and ejaculation to adult male rats, where a large increase in MRL aromatase [27]. activity was recorded, although the difference in serum E2 con- Data from the present study showed that SEM140 was a quite centrations between DDE-treated animals and controls was not high dose to evidence specific effects on the endocrine system, due statistically significant [20]. Moreover, the phytoestrogen genis- to the significant body weight reduction, which effects observed at tein, which is considered a selective estrogen receptor modulator that dose can be associated to. depending its effects on the target tissue, has been shown to At mid and low dose levels, the effects on timing of sexual devel- induce aromatase activity in the hepatic cells HepG2. This sug- opment seem to be in apparent contrast with data on sexual steroid gests that the endogenous production of E2 could be important serum levels: in fact, in females, timing of vaginal opening was in the regulation of aromatase gene expression in hepatic cells unaffected by the treatment whereas E2 serum levels were sig- [36]. nificantly reduced. In males, preputial separation was anticipated In the three E2-competitive assays, SEM showed a weak whereas DHT was reduced. anti-estrogenic activity, since it was able to inhibit the ER- Literature data put into evidence that in vivo administration of mediated: (i) trans-activation of the hER␣ in a yeast-based the NMDAR antagonist MK-801 (0.001 mg/kg) to immature male system; (ii) proliferation of the MCF-7 breast cancer cell line rats resulted in early pubertal development [28]. Moreover, in and (iii) induction of the ALP activity in the Ishikawa endome- female rats, delayed vaginal opening was recorded following treat- trial cancer cell line. Even if these results could also be explained ment with the same NMDAR antagonist during the pubertal period. with an unspecific enzymatic inhibitory effect, they suggest The LNCaP-based NMDAR-interference assay confirmed the pos- the possibility of a specific potential interference on E2 sig- sible interference of SEM with the NMDA signalling system and nalling. its possible involvement in the effects recorded on puberty onset. Overall, SEM administered in vivo to peripubertal rats appeared NMDARs are glutamate-gated ion channels widely expressed in to act as an ED in both sexes resulting in the alteration of the onset the CNS, playing a key role in excitatory synaptic transmission of puberty and sex steroid serum levels by multiple and gender and involved in several neurological disorders [29]. Moreover, specific mechanisms of action(s), not excluding the unbalance of NMDARs are also expressed in reproductive and urogenital tracts steroid metabolism, the interaction with ER and interference with in both sexes, modulating age-dependent connections among CNS function at hypothalamic level. the CNS and peripheral organs [30]. Indeed, NMDAR, can affect Taking into account the in vitro SEM antagonist effects the gonadotropin-releasing hormone (GnRH) neurons controlling on both ER␣ and NMDAR as well as its in vivo activi- reproduction. ties, further investigations on their cross-talk at peripheral In particular, blockade of the NR2b subunit of NMDAR sup- and central levels and possible SEM effects on other nuclear presses pulsatile luteinizing hormone (LH) release in the rat [31], receptors, i.e. androgen receptor, should be performed to thus contributing to the regulation of GnRH/LH release in the better clarify SEM mechanism(s) of action in the juvenile female hypothalamus [30]. Sexual maturation and the onset of period. 48 F. Maranghi et al. / Chemico-Biological Interactions 183 (2010) 40–48

Such in vivo–in vitro approach appeared to be an useful tool to [16] K. Grote, A.J. Andrade, S.W. Grande, S.N. Kuriyama, C.E. Talsness, K.E. Appel, put into evidence SEM activity on endocrine homeostasis. I. Chahoud, Effects of peripubertal exposure to triphenyltin on female sexual development of the rat, Toxicology 1 (2006) 17–24. [17] E.J. Routledge, J.P. Sumpter, Estrogenic activity of surfactants and some of their Conflict of interest statement degradation products assessed using a recombinant yeast screen, Environ. Tox- icol. Chem. 15 (1996) 241–248. [18] E. Testai, L. Vittozzi, Biochemical alterations elicited in rat liver microsomes The authors declare that there are no conflicts of interest. by oxidation and reduction products of chloroform metabolism, Chem. Biol. Interact. 59 (1986) 157–171. [19] V. Oyama, H. Eagle, Measurement of cell growth in tissue culture with a phenol Acknowledgments reagent (folin-ciocalteau), Proc. Soc. Exp. Biol. Med. 91 (1956) 305–307. [20] L. You, M. Sar, E. Bartolucci, S. Ploch, M. Whitt, Induction of hepatic aro- The authors acknowledge Dr. Vincenzo Lagatta and Dr. Valentina matase by p,p -DDE in adult male rats, Mol. Cell. Endocrinol. 178 (2001) 207– 214. Tassinari for their technical assistance during the study. [21] P. Sanwald, E.A. Blankson, B.D. Dulery, J. Schoun, N.D. Huebert, J. Dow, Iso- The present study was supported by: the EU Network of Excel- cratic high-performance liquid chromatographic method for the separation lence “CASCADE” (FOOD-CT-2004-506319); the Italian project of testosterone metabolites, J. Chromatogr. B: Biomed. Appl. 20 (1995) 207– 215. of ISPESL, Ministry of Health and Istituto Superiore di Sanità: [22] L. Lehmann, H.L. Esch, J. Wagner, L. Rohnstock, M. Metzler, Estrogenic and “Endocrine Disrupting Chemicals: Evaluation of Risk and Effects” genotoxic potential of equol and two hydroxylated metabolites of Daidzein (Fasc. Q4C). in cultured human Ishikawa cells, Toxicol. Lett. 28 (2005) 72–86. [23] D.J. Waxman, Interactions of hepatic cytochromes P-450 with steroid hormones: regioselectivity and stereoselectivity of steroid metabolism and References hormonal regulation of rat P-450 enzyme expression, Biochem. Pharmacol. 37 (1988) 71–84. [24] R.L. Brent, Utilization of juvenile animal studies to determine the human effects [1] http://www.iss.it/binary/inte/cont/policy 2004.1127310360.pdf. and risks of environmental toxicants during postnatal developmental stages, [2] A. Mantovani, Hazard identification and risk assessment of endocrine dis- Birth Defects Res. 71 (2004) 303–320. rupting chemicals with regard to developmental effects, Toxicology 27 (2002) [25] D. Crews, E. Willingham, J.K. Skipper, Endocrine disruptors: present issues, 367–370, Review. future directions, Q. Rev. Biol. 75 (2000) 243–260. [3] J.G. Bendall, Semicarbazide is non-specific as a marker metabolite to reveal [26] N.M. Saarinen, C. Bingham, S. Lorenzetti, A. Mortensen, S. Mäkelä, P. Penttinen, nitrofurazone abuse as it can form under Hofmann conditions, Food Addit. I.K. Sørensen, L.M. Valsta, F. Virgili, G. Vollmer, A. Wärri, O. Zierau, Tools to eval- Contam. 12 (2008) 1–10. uate estrogenic potency of dietary phytoestrogens: a consensus paper from the [4] F. Maranghi, R. Tassinari, V. Lagatta, G. Moracci, C. Macrì, A. Eusepi, A. Di Virgilio, EU thematic network “Phytohealth” (QLKI-2002–2453), Genes Nutr. 1 (2006) M.L. Scattoni, G. Calamandrei, Effects of the food contaminant semicarbazide 143–158. following oral administration in juvenile Sprague–Dawley rats, Food Chem. [27] N.F. Gonzalez-Cadavid, I. Ryndin, D. Vernet, T.R. Magee, J. Rajfer, Presence of Toxicol. 47 (2009) 472–479. NMDA receptor subunits in the male lower urogenital tract, J. Androl. 21 (2000) [5] Official Journal of the European Union. Commission Directive 2004/1/EC of 566–578. January 6, 2004, amending Directive 2002/72/EC. [28] J.P. Bourguignon, A. Gérard, M.L. Alvarez Gonzalez, P. Franchimont, Neuroen- [6] P.P. Mulder, B. Beumer, J.A. Van Rhijn, The determination of biurea: a novel docrine mechanism of onset of puberty. Sequential reduction in activity of method to discriminate between nitrofurazone and azodicarbonamide use in inhibitory and facilitatory N-methyl-d-aspartate receptors, J. Clin. Invest. 90 food products, Anal. Chim. Acta 586 (2007) 366–373. (1992) 1736–1744. [7] R. Gatermann, K. Hoenicke, M. Mandix, Formation of semicarbazide (SEM) from [29] P. Paoletti, J. Neyton, NMDA receptor subunits: function and pharmacology, natural compounds in different kind of food products, Czech. J. Food Sci. 22 Curr. Opin. Pharmacol. 7 (2007) 39–47. (2004) 353–354. [30] J.A. Maffucci, M.D. Walker, A. Ikegami, M.J. Wolle, A.C. Gore, The NMDA receptor [8] G.O. Noonan, T.H. Begley, G.W. Diachenko, Semicarbazide formation in flour subunit NR2b: effects on LH release and GnRH gene expression in young and and bread, J. Agric. Food Chem. 26 (2008) 2064–2067. middle-aged female rats, with modulation by estradiol, Neuroendocrinology [9] S. Szilagyi, M.B. de la Calle, Semicarbazide in baby food: a European survey, Eur. 87 (2008) 129–141. Food Res. Technol. 224 (2006) 141–146. [31] A.C Gore, Gonadotropin-releasing hormone neurons, NMDA receptors, and [10] H. Kinemuchi, H. Sugimoto, T. Obata, N. Satoh, S. Ueda, Selective inhibitors of their regulation by steroid hormones across the reproductive life cycle, Brain membrane-bound semicarbazide-sensitive amine oxidase (SSAO) activity in Res. Rev. 37 (2001) 235–248. mammalian tissues, Neurotoxicology 25 (2004) 325–335. [32] J.A. Moguilevsky, W. Wuttke, Changes in the control of gonadotropin secretion [11] B. Gong, P.J. Boor, The role of amine oxidases in xenobiotic metabolism, Expert by neurotransmitters during sexual development in rats, Exp. Clin. Endocrinol. Opin. Drug Metab. Toxicol. 2 (2006) 559–571. Diabetes 109 (2001) 188–195. [12] M. Strolin Benedetti, K.F. Tiptonb, R. Whomsley, Amine oxidases and monooxy- [33] C.A. Christian, J. Pielecka-Fortuna, S.M. Moenter, Estradiol suppresses gluta- genases in the in vivo metabolism of xenobiotic amines in humans: has the matergic transmission to gonadotropin-releasing hormone neurons in a model involvement of amine oxidases been neglected? Fundam. Clin. Pharmacol. 21 of negative feedback in mice, Biol. Reprod. 80 (2009) 1128–1135. (2007) 467–479. [34] T.R. Chakraborty, L. Ng, A.C. Gore, Colocalization and hormone regulation of [13] E.R. Nestmann, B.S. Lynch, K. Musa-Veloso, G.H. Goodfellow, E. Cheng, L.A. estrogen receptor alpha and N-methyl-d-aspartate receptor in the hypothala- Haighton, V.M. Lee-Brotherton, Safety assessment and risk-benefit analysis mus of female rats, Endocrinology 144 (2003) 299–305. of the use of azodicarbonamide in baby food jar closure technology: putting [35] W. Yin, J.M. Mendenhall, S.B. Bratton, T. Oung, W.G. Janssen, J.H. Morri- trace levels of semicarbazide exposure into perspective—a review, Food Addit. son, A.C. Gore, Novel localization of NMDA receptors within neuroendocrine Contam. 22 (2005) 875–891. gonadotropin-releasing hormone terminals, Exp. Biol. Med. (Maywood) 232 [14] J.M. Santos, C.E. Macedo, M.L. Brandão, Gabaergic mechanisms of hypothalamic (2007) 662–673. nuclei in the expression of conditioned fear, Neurobiol. Learn. Mem. 90 (2008) [36] L. Ye, M.Y. Chan, L.K. Leung, The soy isoflavone genistein induces estrogen 560–568. synthesis in an extragonadal pathway, Mol. Cell. Endocrinol. 302 (2009) 73– [15] I.K. Franklin, C.B. Wollheim, GABA in the endocrine pancreas: its putative role as 80. an islet cell paracrine-signalling molecule, J. Gen. Physiol. 123 (2004) 185–190.