Developmental neurotoxicity (DN\ue002) o\ue004 environmental chemicals has been recog- nize\ue003 worl\ue003wi\ue003e as a serious threat to human health, an\ue003 the resulting neurologic \ue003e\ue000cits negatively a\ue001ect \ue004amilies an\ue003 society (Gol\ue003man an\ue003 Ko\ue003uru 2000; Nee\ue003leman et al. 2002). Current DN\ue002 testing gui\ue003elines (Organization \ue004or Economic Cooperation an\ue003 Development 2007; U.S. Environmental Protection Agency 1998) propose investiga- tions in ro\ue003ents, mainly rats. Such a DN\ue002

140 \ue003ams an\ue003 1,000 pups an\ue003 is there\ue004ore extremely time- an\ue003 cost-intensive (Lein et al. 2005). Relying solely on the existing gui\ue003e- lines to a\ue003\ue003ress current an\ue003 anticipate\ue003 \ue004uture regulatory \ue003eman\ue003s \ue004or DN\ue002 o\ue004 the thou- san\ue003s o\ue004 chemicals \ue004or which there are \ue004ew to no DN\ue002 \ue003ata woul\ue003 incur unacceptable costs in terms o\ue004 animals an\ue003 person-years (Lein et al. 2007). Tere\ue004ore, accor\ue003ing to the \u201c3R principle\u201d (re\ue003uction, replacement, an\ue003 re\ue000ne- ment) o\ue004 Russel an\ue003 Burch (1959), alternative testing strategies are nee\ue003e\ue003 to a\ue003\ue003ress ani- mal wel\ue004are by re\ue000ning an\ue003 re\ue003ucing animal experiments, an\ue003 to create a\ue001or\ue003able, sensi- tive, an\ue003 mechanism-base\ue003 metho\ue003s suitable \ue004or high- or me\ue003ium-throughput screening (Collins et al. 2008). Furthermore, the inclu- sion o\ue004 human-cell\u2013base\ue003 in vitro systems into an integrate\ue003 DN\ue002 tiere\ue003 testing approach

\ue002o combine transatlantic strengths an\ue003 avoi\ue003 \ue003oubling o\ue004 work, a partnership between the Johns Hopkins Center \ue004or Alternatives to Animal \ue002esting (Developmental Neurotoxicity \ue002estSmart program), the European Centre \ue004or the Vali\ue003ation o\ue004 Alternative Metho\ue003s, an\ue003 the European Chemical In\ue003ustry Council has been \ue004orme\ue003. \ue002his partnership \ue004ollows the com- mon goal o\ue004 \u201cincorporating in vitro alternative metho\ue003s \ue004or DN\ue002 testing into international hazar\ue003 an\ue003 risk assessment strategies\u201d (Coecke et al. 2007). Coecke et al. (2007) provi\ue003e\ue003 a comprehensive overview o\ue004 the existing in vitro mo\ue003els an\ue003 state\ue003 that, \u201calthough all the test systems \ue003escribe\ue003 were not \ue003evelope\ue003 \ue004or regu- la tory purposes at this stage i\ue004 they prove use- \ue004ul, we hope that this report will encourage their \ue004urther \ue003evelopment to ren\ue003er them ame- nable to high-throughput approaches.\u201d

Tere\ue004ore, the aim o\ue004 this work wasa) to intro\ue003uce the cell biological characteristics o\ue004 human neurospheres as a three-\ue003imensional cell system approach \ue004or DN\ue002 testing;b) to \ue003emonstrate that neurospheres are likely to mirror basic processes o\ue004 brain \ue003evelopment, such as proli\ue004eration, \ue003i\ue001erentiation, migra- tion, an\ue003 apoptosis; an\ue003c) to \ue003emonstrate that these processes can be mo\ue003ulate\ue003 by \ue003evelop- mental neuro toxicants.

ri\ue003e (MeHgCl) \ue004rom Rie\ue003el-\ue003e Ha\u00ebn (Seelze, Germany); all other substances were obtaine\ue003 \ue004rom Sigma Al\ue003rich (Munich, Germany), unless otherwise state\ue003.

human neural progenitor cells (hNPCs; Lonza Verviers SPRL, Verviers, Belgium) were cul- ture\ue003 at 37\u00b0C an\ue003 5% CO2 as a suspension culture in proli\ue004eration me\ue003ium consisting o\ue004 Dulbecco\u2019s mo\ue003i\ue000e\ue003 Eagle me\ue003ium (DMEM) an\ue003 Hams F12 (3:1) supplemente\ue003 with B27 (Invitrogen GmbH, Karlsruhe, Germany), 20 ng/mL epi\ue003ermal growth \ue004actor (EGF; Biosource, Karlsruhe, Germany), an\ue003 20 ng/ mL recombinant human \ue004ibroblast growth \ue004actor (FGF; R&D Systems, Wiesba\ue003en- Nor\ue003ensta\ue003t, Germany) (Moors et al. 2007). When spheres reache\ue003 0.7 mm in \ue003iameter, they were choppe\ue003 up to passage 3 with a McIlwain tissue chopper. Di\ue001erentiation was initiate\ue003 by growth \ue004actor with\ue003rawal in \ue003i\ue004- \ue004erentiation me\ue003ium [DMEM an\ue003 Hams F12 (3:1) supplemente\ue003 with N2 (insulin, trans\ue004errin, so\ue003ium selenite, putrescine, an\ue003 progesterone; Invitrogen)] an\ue003 plate\ue003 onto poly-d-lysine/laminin\u2013coate\ue003 chamber sli\ue003es (BD Bioscience, Erembo\ue003egem, Belgium).

in\ue003irubin (10 \u00b5M) in proli\ue004eration me\ue003ium (28 hr), an\ue003 to cAMP (200 \u00b5M), MeHgCl (250 nM to 1 \u00b5M), mercury chlori\ue003e (HgCl2; 500 nM to 10 \u00b5M, 48 hr) or staurosporine (0.1 an\ue003 1 \u00b5M), or hy\ue003rogen peroxi\ue003e (H2O2; 0.1 an\ue003 1 mM) (24 hr) in \ue003i\ue001erentia- tion me\ue003ium. We chose concentration ranges o\ue004 mercury accor\ue003ing to Monnet-\ue002schu\ue003i

A\ue002\ue002ress correspon\ue002ence to E. Fritsche, Institut \ue003\u00fcr Umweltme\ue002izinische Forschung gGmbH an \ue002er Heinrich Heine-Universit\u00e4t, \ue001oxicology, Au\ue003\u2019m Hennekamp 50, 40225 D\u00fcssel\ue002or\ue003, Germany. \ue001elephone: 00492113389217. Fax: 00492113190910. E-mail: ellen.\ue003ritsche@uni-\ue002uessel\ue002or\ue003.\ue002e

Tis work was supporte\ue002 by the German Fe\ue002eral Institute \ue003or Risk Assessment; the German Fe\ue002eral Institute \ue003or the Environment, Nature Conservation an\ue002 Nuclear Sa\ue003ety; the German Fe\ue002eral Ministry \ue003or E\ue002ucation an\ue002 Research (13925C); an\ue002 the Deutsche Forschungsgemeinscha\ue003t (GK1427).

threat to human health. Current DN\ue001 testing guidelines propose investigations in rodents, which require large numbers o\ue002 animals. With regard to the \u201c3 Rs\u201d (reduction, replacement, and re\ue000ne- ment) o\ue002 animal testing and the European regulation o\ue002 chemicals [Registration, Evaluation, and Authorisation o\ue002 Chemicals (REACH)], alternative testing strategies are needed in order to re\ue000ne and reduce animal experiments and allow \ue002aster and less expensive screening.

spheres possess \ue002unctioning apoptosis machinery. Te developmental neurotoxicants methylmercury chloride and mercury chloride decreased migration distance and number o\ue002 neuronal-like cells in diferentiated hNPCs. Furthermore, hNPCs undergo caspase-independent apoptosis when exposed toward high amounts o\ue002 oxidative stress.

ses were per\ue004orme\ue003 as previously \ue003escribe\ue003 (Moors et al. 2007). For living cell migra- tion analyses, neurospheres were grown in the Focht Chamber System 2 (Bioptechs, Butler, PA, USA) un\ue003er temperature- an\ue003 CO2- controlle\ue003 con\ue003itions. Images were acquire\ue003 every 2 min by a Zeiss Axiovert 100 inverte\ue003 microscope (Zeiss, Goettingen, Germany).

\ue004erentiating spheres were \ue000xe\ue003 in 4% para\ue004orm- al\ue003ehy\ue003e \ue004or 30 min. A\ue004ter washing spheres in phosphate-bu\ue004\ue004ere\ue003 saline (PBS), they were incubate\ue003 in a 25% sucrose solution (wt/vol) overnight at 4\u00b0C. A\ue004terwar\ue003, spheres were trans\ue004erre\ue003 to tissue-\ue004reezing me\ue003ium (Jung HistoService, Nussloch, Germany). Cryostat sections (10 \u00b5m) were prepare\ue003 \ue004or immuno- histochemistry. Antibo\ue003ies \ue004or staining were nestin (1:150; BD Bioscience), glial \ue000brillary aci\ue003ic protein (GFAP; 1:100, Sigma Al\ue003rich), or\u03b2(III)tubulin (1:100; Sigma Al\ue003rich).

chemistry as previously \ue003escribe\ue003 (Fritsche et al. 2005; Moors et al. 2007). For quanti\ue000ca- tion analyses, we use\ue003 the Metamorph analysis so\ue004tware package (version 7.1.7.0; Universal Imaging Corp., West Chester, PA, USA). We \ue003etermine\ue003 the variation o\ue004 protein expres- sion by relating area o\ue004 fuorescence signal to cell number in a region o\ue004 interest within the migration area. In\ue003ivi\ue003ual pixels were i\ue003en- ti\ue004ie\ue003 as \u201cpositive\u201d i\ue004 the \ue004luorescence signal excee\ue003e\ue003 a \ue003etermine\ue003 color threshol\ue003 [green, hue (H) 71\u2013113, saturation (S) 10\u2013255, intensity (I) 10\u2013255; yellow/re\ue003, H 0\u201371, S 10\u2013255, I 10\u2013255)]. \ue002o \ue003etermine the cell number, we selecte\ue003 images \ue004or positive 4\u00b4,6- \ue003iami\ue003ino-2-phenylin\ue003ole (DAPI) staining (blue, H 152\u2013180, S 10\u2013255, I 10\u2013255) an\ue003 morphologic parameters (integrate\ue003 morphol- ogy analysis: area, 104 >n > 107; ellipsoi\ue003 \ue004orm \ue004actor, 0.1 >n > 1.8).

spheres were grown in proli\ue004eration me\ue003ia with or without growth \ue004actors or expose\ue003 to in\ue003irubin. \ue002o obtain a single-cell suspen- sion, neurospheres were washe\ue003 once in PBS, incubate\ue003 with Accutase (100%; PAA, C\u00f6lbe, Germany) at 37\u00b0C \ue004or 20 min, an\ue003 then gen- tly pipette\ue003. Te suspension was centri\ue004uge\ue003 (4\u00b0C, 1,400 \u00d7g, 5 min) an\ue003 the pellet was resuspen\ue003e\ue003 in PBS containing 0.8% para- \ue004ormal\ue003ehy\ue003e (Polyscience Inc., Eppelheim, Germany). Cells were \ue000xe\ue003 \ue004or 30 min at 4\u00b0C, centri\ue004uge\ue003 (4\u00b0C, 1,400 \u00d7g, 5 min), resus- pen\ue003e\ue003 in PBS containing 0.15% saponin an\ue003 10 \u00b5g/mL RNase, an\ue003 incubate\ue003 \ue004or 30 min at 37\u00b0C. We then a\ue003\ue003e\ue003 50 \u00b5g/mL propi\ue003ium io\ue003ine 5 min be\ue004ore FACS analyses.

Cell\ue002iter-Blue assay (Promega, Mannheim, Germany) as previously \ue003escribe\ue003 (Moors et al. 2007). Te assay is base\ue003 on measure- ments o\ue004 the mitochon\ue003rial re\ue003uctase activity by conversion o\ue004 the substrate resazurin to the \ue004luorescent pro\ue003uct resoru\ue004in by mitochon- \ue003rial re\ue003uctases, which can be assesse\ue003 in a fu- orometer. Te lactate \ue003ehy\ue003rogenase (LDH) assay (Cyto\ue002ox-One; Promega) assesses cell \ue003eath by measuring LDH that leaks out o\ue004 \ue003ea\ue003 cells into the me\ue003ia. We per\ue004orme\ue003 the assay accor\ue003ing to the manu\ue004acturer\u2019s instruc- tions. Briefy, supernatants o\ue004 treate\ue003 cells were collecte\ue003 an\ue003 incubate\ue003 with an equal amount o\ue004 Cyto\ue002ox-One reagent \ue004or 4 hr be\ue004ore the \ue003etection o\ue004 fuorescence (excitation, 540 nm; emission, 590 nm). Caspase-3/-7 activi- ties were measure\ue003 with the Apo-One Kit (Promega) accor\ue003ing to the manu\ue004acturer\u2019s instructions. Briefy, cells were lyse\ue003 an\ue003 cas- pase activity was assesse\ue003 by measuring the cleavage o\ue004 a caspase-3/-7\u2013speci\ue000c fuorescent substrate (Z-DEVD-R110) with a fuorometer (excitation, 488 nm; emission, 538 nm).

For proli\ue004eration assays, spheres were cul- ture\ue003 in proli\ue004eration me\ue003ium supplemente\ue003 with or without 20 or 100 ng/mL EGF in 96-well plates. We assesse\ue003 cell viability as a meas ure \ue004or cell number using the Cell\ue002iter- Blue assay at \ue003i\ue001erent time points. Because the \ue003ye cause\ue003 no acute cytotoxicity, spheres were washe\ue003 with me\ue003ium a\ue004ter fuorescence rea\ue003- ing an\ue003 then the same spheres were monitore\ue003 over a perio\ue003 o\ue004 2 weeks. For \ue003etermination o\ue004 sphere size, we gauge\ue003 sphere \ue003iameter opti- cally with an object micrometer. We counte\ue003 the number o\ue004 cells/sphere a\ue004ter trypsination (0.25% trypsin; Invitrogen) \ue004or 2 min.

nucleoti\ue003yl trans\ue004erase 2\u00b4-\ue003eoxyuri\ue003ine 5\u00b4-triphosphate (\ue003U\ue002P) nick en\ue003 labeling (\ue002UNEL) assays, we use\ue003 fuorescein-couple\ue003 \ue003U\ue002P an\ue003 the terminal trans\ue004erase kit \ue004rom Roche Diagnostics (Mannheim, Germany) to label DNA stran\ue003 breaks; nuclei were counter- staine\ue003 with Hoechst 33258 (Invitrogen). Plate\ue003 neurospheres were expose\ue003 to stauro- sporine (1 \u00b5M) or H2O2 (1 mM) a\ue004ter 48 hr o\ue004 \ue003i\ue004\ue004erentiation. A\ue004ter another 12 an\ue003 24 hr, cells were \ue004ixe\ue003 with 4% para\ue004orm- al\ue003ehy\ue003e, washe\ue003 twice with PBS, covere\ue003 with reaction mixture (2.5 mM CoCl2, 5 \u00b5M \ue004luorescein couple\ue003 \ue003U\ue002P, 5,000 U/mL terminal-trans\ue004erase, 2 \u00b5g/mL Hoechst, 0.1% triton in 1\u00d7 terminal trans\ue004erase bu\ue001er), an\ue003 incubate\ue003 at 37\u00b0C \ue004or 1 hr. Sli\ue003es were then washe\ue003 with PBS three times an\ue003 mounte\ue003 with PBS/glycerol (1:1). Staine\ue003 cells were analyze\ue003 with a fuorescence microscope.

concentration e\ue001ects), an\ue003 the Stu\ue003ent\u2019st-test \ue004or two-group comparisons (treatment vs. control; two time points). \ue002he signi\ue004icance value was set atp < 0.05. \ue002o \ue003escribe the asso- ciations between in\ue003epen\ue003ent variables (\ue003iam- eter/cell number; \ue003iameter/fuorescence), we \ue000tte\ue003 curves up to the thir\ue003 \ue003egree. We use\ue003

Human neurospheres grow foating \ue004reely in \ue003e\ue000ne\ue003 me\ue003ium without a\ue003\ue003ition o\ue004 serum [see Supplemental Material, Figure 1A (available at http://www.ehponline.org/ members/2009/0800207/suppl.p\ue003\ue004)]. Upon with\ue003rawal o\ue004 growth \ue004actor, cells migrate ra\ue003ially out o\ue004 the sphere onto a poly-d-lysine/ laminin matrix, thereby \ue004orming a migration area [see Supplemental Material, Figure 1B an\ue003 the vi\ue003eo (available at http://www. ehponline. org/members/ 2009/0800207/ suppl.p\ue003\ue004)]. Each cell leaves the sphere e\ue003ge in a 90\u00b0 angle an\ue003 travels away in a straight line. Moreover, cells move towar\ue003 an\ue003 away \ue004rom each other.

\ue002o evaluate repro\ue003ucibility an\ue003 stability o\ue004 neurosphere migration, we assesse\ue003 \ue003epen- \ue003ence o\ue004 migration spee\ue003 on neurosphere size. \ue002here\ue004ore, the \ue003istance between the sphere e\ue003ge an\ue003 the \ue004arthest outgrown cells was meas- ure\ue003 24 hr a\ue004ter plating, \ue003epen\ue003ent on \ue003i\ue001er- ent sphere \ue003iameters. Supplemental Material, Figure 1C (available at http://www.ehponline. org/ members/ 2009/ 0800207/suppl.p\ue003\ue004) shows that spheres with a \ue003iameter between 0.2 an\ue003 0.7 mm wan\ue003er approximately 0.48 mm within 24 hr (e.g., 0.2 mm \ue003iameter, 0.48 \u00b1 0.06 mm; 0.7 mm \ue003iameter, 0.48 \u00b1 0.09 mm; mean \u00b1 SD), \ue003emonstrating that migration spee\ue003 is in\ue003epen\ue003ent o\ue004 sphere size. Moreover, cells \ue004rom \ue003i\ue004\ue004erent \ue003onors (0.3-mm-\ue003iameter spheres) also \ue003i\ue003 not vary signi\ue004icantly in migration spee\ue003 over 24 hr [see Supplemental Material, Figure 1D (available at http://www.ehponline. org/members/2009/0800207/suppl.p\ue003\ue004)].

Next, we analyze\ue003 the cellular composi- tion o\ue004 neurospheres. We slice\ue003 10-\u00b5m cryo- stat sections o\ue004 proli\ue004erating neuro spheres an\ue003 examine\ue003 expressions o\ue004a) nestin, a marker protein \ue004or neural stem cells;b)\u03b2(III)tubulin, which stains neurons; orc) GFAP \ue004or glial cells. Immunocytochemical analy ses reveale\ue003 nestin-positive (+) cells were locate\ue003 mainly in the sphere periphery, whereas\u03b2(III)tubulin+ an\ue003 GFAP+ cells resi\ue003e\ue003 in the sphere cen- ter (Figure 1A,B). \ue002his pattern \ue003isappeare\ue003 a\ue004ter spheres were plate\ue003 \ue004or \ue003i\ue001erentiation. A\ue004ter 8 \ue003ays o\ue004 \ue003i\ue001erentiation,\u03b2(III)tubulin+ cells were locate\ue003 at the e\ue003ge o\ue004 the sphere, whereas nestin+ an\ue003 GFAP+ cells were homog- enously \ue003istribute\ue003 throughout the sphere (Figure 1C,D).

migration area a\ue004ter 24 hr an\ue003 7 \ue003ays o\ue004 migra- tion. \ue002wenty-\ue004our hours a\ue004ter plating, nearly all migrate\ue003 cells seeme\ue003 to express nestin, showing that immature cells migrate out o\ue004 the sphere. Furthermore,\u03b2(III)tubulin+ an\ue003 GFAP+ cells were also locate\ue003 in the migration area. In contrast, 7 \ue003ays a\ue004ter plating almost all cells lost nestin expression an\ue003 became\u03b2(III) tubulin+ or GFAP+ (Figure 2A). Quanti\ue000cation o\ue004 the number o\ue004 pixels in the respective images reveale\ue003 a 5.5-\ue004ol\ue003 re\ue003uction in the number o\ue004 nestin+ pixels/nuclei a\ue004ter 6 more \ue003ays o\ue004 \ue003i\ue004- \ue004erentiation (570.9 \u00b1 64 to 103.2 \u00b1 29 pix- els/nuclei; mean \u00b1 SD), whereas in the same time perio\ue003 the number o\ue004\u03b2(III)tubulin+ an\ue003 GFAP+ pixels increase\ue003 4.7- an\ue003 1.9-\ue004ol\ue003,

respectively [\u03b2(III)tubulin, \ue004rom 118.7 \u00b1 27.4 to 509.6 \u00b1 55 pixels/nuclei; GFAP, \ue004rom 250.8 \u00b1 56 to 480.7 \u00b1 198 pixels/nuclei; Figure 2A]. Furthermore, the immuno cyto chemical stain- ing \ue004or\u03b2(III)tubulin suggests that neuronal cells may \ue004orm connections an\ue003 thus buil\ue003 neuronal networks (Figure 2A).

Another cell type emerging \ue004rom neu- ral precursor cells are O4+ oligo\ue003en\ue003rocytes. \ue002hey \ue004orm within the neurosphere (Fritsche et al. 2005) an\ue003 migrate out o\ue004 the sphere over time. A\ue004ter 2, 4, an\ue003 7 \ue003ays o\ue004 \ue003i\ue001erentiation, 3 \u00b1 0.2, 52 \u00b1 1, an\ue003 210 \u00b1 5 O4+ cells (mean \u00b1 SD), respectively, were locate\ue003 in the migra- tion area (Figure 2B). Tey also change\ue003 mor- phology over time. Although a\ue004ter 48 hr most

O4+ cells were bipolar, we \ue004oun\ue003 more branch- ing a\ue004ter 4 \ue003ays; a\ue004ter 7 \ue003ays o\ue004 \ue003i\ue001erentiation, multi polar an\ue003 membrane sheet-\ue004orming cells were prominent (Figure 2C).

Next, we \ue003evelope\ue003 assays that i\ue003enti\ue004y changes in cell proli\ue004eration, \ue003i\ue001erentiation, migration, an\ue003 apoptosis by applying mo\ue003el chemicals, which are known to inter\ue004ere with normal brain \ue003evelopment (Gran\ue003jean an\ue003 Lan\ue003rigan 2006).

Cell proli\ue004eration in a neurosphere can be \ue003etermine\ue003 by counting the number o\ue004 cells per \ue003issociate\ue003 sphere or by measuring the increase in sphere \ue003iameter over time. Figure 3A shows that there was a very goo\ue003 association between these two parameters