Analysis of Human Embryonic Stem Cellswith Regulatable Expression of the Cell Adhesion

Molecule L1 in Regeneration after Spinal Cord Injury

Myungsik Yoo,1 Gunho Anthony Lee,1 Christopher Park,1 Rick I. Cohen,2 and Melitta Schachner1,3

Abstract

Cell replacement therapy is one potential avenue for central nervous system (CNS) repair. However, transplanted stem cellsmay not contribute to long-term recovery of the damaged CNS unless they are engineered for functional advantage. To finetune regenerative capabilities, we developed a human neural cell line expressing L1, a regeneration-conducive adhesionmolecule, under the control of a doxycycline regulatable Tet-off promoter. Controlled expression of L1 is desired becauseoverexpression after regenerative events may lead to adverse consequences. The regulated system was tested in several celllines, where doxycycline completely eliminated green fluorescent protein or L1 expression by 3–5 days in vitro. Increasedcolony formation as well as decreased proliferation were observed in H9NSCs without doxycycline (hL1-on). To test therole of L1 in vivo after acute compression spinal cord injury of immunosuppressed mice, quantum dot labeled hL1-on orhL1-off cells were injected at three sites: lesion; proximal; and caudal. Mice transplanted with hL1-on cells showed a betterBasso Mouse Scale score, when compared to those with hL1-off cells. As compared to the hL1-off versus hL1-on celltransplanted mice 6 weeks post-transplantation, expression levels of L1, migration of transplanted cells, and immunore-activity for tyrosine hydroxylase were higher, whereas expression of chondroitin sulfate proteoglycans was lower. Resultsindicate that L1 expression is regulatable in human stem cells by doxycycline in a nonviral engineering approach.Regulatable expression in a prospective nonleaky Tet-off system could hold promise for therapy, based on the multi-functional roles of L1, including neuronal migration and survival, neuritogenesis, myelination, and synaptic plasticity.

Key words: adhesion molecule L1; inducible Tet-off system; regulatable expression; spinal cord injury; stem celltransplantation

Introduction

Embryonic stem cell (ESC) derivatives represent a po-tential approach for cell based therapy as a treatment for ir-

reversible neuronal cell damage.1 Aside from eliminating the riskof tumor/teratoma formation, additional areas of concern need to beaddressed to allow for successful cell therapy. These include, butare not limited to, robust cell survival2–5 and circumvention ofendogenous antiregenerative signals in the acutely or chronicallyinjured host. Based on previous evidence that the regeneration-conducive cell adhesion molecule, L1, enhances recovery in dif-ferent types of mammalian nervous system lesions, we investigatedthe possibility of using L1, in a regulatable fashion, to engineer anoptimized cell therapy vector. We postulated that mirroring thenatural down-regulation of L1 expression in postnatal nervoussystem development by using a regulatable system would be im-

portant to optimize initial regenerative events and avoid compli-cations caused by irreversible overexpression postrepair.

The immunoglobulin superfamily molecule, L1, plays crucial rolesin multiple morphogenetic functions, such as neuronal migration,differentiation, and survival, as well as neuritogenesis, axonal tar-geting, myelination, synapse formation, and synaptic plasticity.6–12 L1is not only crucial during development, but also in regeneration afterinjury of the central and peripheral nervous systems.6,7,13–18 However,constitutively high expression of L1 could be disadvantageous, unlesslimited to sets of functional hot spots, such as generation of inter-neurons in the olfactory bulb or of granule cells in the dentate gyrus,and in altering synaptic efficacy. In a regenerative context after in-jury, although not previously observed in different injury paradigms,overexpression of L1 may induce, for instance, erroneous growth/sprouting axons, such as those of sensory nerve fibers causing allo-dynia and hyperalgesia.19 For therapeutic prospects, L1 expression

1W.M. Keck Center for Collaborative Neuroscience and Department of Cell Biology and Neuroscience, Rutgers University, Piscataway, New Jersey.2Rutgers University, Biomedical Engineering, Piscataway, New Jersey.3Center for Neuroscience, Shantou University Medical College, Shantou, People’s Republic of China.

JOURNAL OF NEUROTRAUMA 31:553–564 (March 15, 2014)ª Mary Ann Liebert, Inc.DOI: 10.1089/neu.2013.2886

553

levels should therefore be controllable in vivo. We have thus chosen anonviral expression system, which may confer advantages, even ifviral transduction would become clinically viable, because virus-mediated cell therapy has the disadvantage that viral DNA sequencesmay be silenced by the host’s cellular protection mechanisms.20 Wehave developed a novel nonviral doxycycline (DOX)-induciblehuman L1 expression system that comprises a single regulatableplasmid with a transrepressor together with a strong promoter, suchas the CAG (chimeric cytomegalovirus and chicken b-actin) pro-moter, and that is efficiently regulatable in glioblastoma and neu-roblastoma cells as well as predifferentiated H9-ESC-derived neuralstem cells (H9NSCs) by DOX in vitro. The stable human cell line,pTet-off-hL1-H9NSC, is also regulatable and functional in vivo incyclosporine-immunosuppressed mice, where locomotor recoveryafter acute compression injury is observed after 5–6 weeks.

Methods

Procedures for H9NSCs in vitro and in vivo

Neural stem cells derived from H9 ESCs (H9ESCs; Fig. 1A)were obtained from the Stem Cell Core Facility at The Stem CellResearch Center (Rutgers University, Piscataway, NJ). Afterengineering the pTet-off-hL1 system (see below), we transfectedthe plasmid system into H9NSCs (Fig 1B, without DOX, and Fig.1B, with DOX). hL1-on and -off cells were expanded and se-lected in the presence of 200 lg/mL of G418 (Fig. 1Cc and 1Dd,respectively). Red q-dot-labeled cells were transplanted intoacutely compression-injured spinal cords of cyclosporine-immunosuppressed mice (see below; Fig. 1Ee) and scored by theBasso Mouse Scale (BMS) every week for 6 weeks without orwith DOX in the drinking water to maintain hL1-on and -off,respectively (Fig. 1Ff ).

FIG. 1. Schematic illustration of experimental procedures for cell lines under in vitro and in vivo conditions. H9 human embryonicstem cells (H9ESCs) (A). Predifferentiated human neural stem cells (H9NSCs) that had been subjected for 7 days to a differentiationprotocol are described in the Methods section (B). Cells were transfected with the pTet-off-hL1 plasmid and maintained under twodifferent conditions: cell line for hL1-on (B, without doxycycline) and cell line for hL1-off (b, with doxycycline). For selection oftransfected cells, cultures were treated with G418 (200 lg/mL) in the culture medium, which was changed every other day for 4 weeksand expanded for storage (C, c, and D, d). Before transplantation into acutely spinal cord injured and cyclosporine-immunosuppressedmice, cells were labeled for quantum dot analysis (E, e). Mice were tested by Basso Mouse Scale every week for 6 weeks withoutdoxycycline (F) and with doxycycline (f ) in the drinking water. DOX, doxycycline; SCI, spinal cord injury.

554 YOO ET AL.

Construction of the vector system

The pTet-off-GFP plasmid is a nonviral single-entity systemcontaining two CAG promoters driving the expression of greenfluorescent protein (GFP) and the transactivator in opposing di-rections (Fig. 2). The vector system was assembled as follows: The

pd2EGFP plasmid (Clontech, Mountain View, CA) was con-structed as the backbone in three steps. First, the pd2EGFP waslinearized using SalI and BglII and ligated with the SalI and BamHIfragment of pCX-EGFP containing the CAG promoter drivingenhanced GFP (eGFP) expression (pCAG-EGFP). Second, anXbaI-digested fragment with seven repeat tetracycline response

FIG. 2. Schematic representation of the pTet-off-GFP and pTet-off-hL1 systems. This system uses a single plasmid doxycyclineTet-off promoter containing seven tetracycline response elements (TREs) located between two oppositely oriented CAG promoters.The CAG promoters drive expression of GFP and the hybrid tetracycline-KRAB repressor. In the absence of doxycycline, the7 · TREs are silenced, allowing activity of the CAG promoters as well as transcription of GFP and the repressor. In this condition, thecells are ‘‘ON’’ for the target genes (A). In the presence of doxycycline, no gene expression is observed, because now the 7 · TREsare bound by the Tet portion of the hybrid repressor, and the CAG motifs are blocked by the KRAB portion. This allows for tightregulation of gene expression and is referred to as ‘‘OFF’’ for the target genes (B). Construction of the nonviral single pTet-off-hL1plasmid, where the GFP gene is replaced by flag tagged for measuring hL1 for hL1-on (C) and hL1-off (D) cells. GFP, greenfluorescent protein; DOX, doxycycline; CAG, chimeric cytomegalovirus and chicken b-actin; TREs, tetracycline response elements;KRAB, Kruppel-associated box.

REGULATED L1 EXPRESSION AND REGENERATION 555

elements (7 · TREs) was cloned using polymerase chain reactionfrom pLVCT-rtTR2SM221 and ligated into SpeI-cut pCAG-EGFP(pCAG-EGFP-TRE). Last, the SpeI fragment from pLVCT-rtTR2SM2 containing the transactivator was ligated together withXbaI/SpeI-cut pCAG-EGFP-TRE (pCAG-Tet-off-GFP, namedpTet-off-GFP). Then, the reverse tetracycline transcriptional re-pressor was fused with the Kruppel-associated box (KRAB) do-main, a transcriptional repressor protein of the eukaryoticubiquitous zinc finger family. Thus, the plasmid system is expectedto enhance repressor functions by the KRAB domain. For L1 ex-pression, the inducible human L1 sequence was exchanged for theGFP sequence by inserting the Klenow-blunted human L1 com-plementary DNA into the EcoRI/blunted site of pCAG-Tet-off-GFP named pTet-off-hL1.

Predifferentiation of H9NSC-ESCs into H9NSCsand immunocytology

We followed a slightly modified adherent monolayer protocol,first published by Smith and coworkers.22,23 The following pro-tocol has been shown to produce the best results: First, the un-differentiated H9ESCs were preconditioned at 80–90%confluence with neural induction medium (NIM), which consistedof a 1:1 ratio of Dulbecco’s modified Eagle’s medium (DMEM)/F12 and neurobasal medium (Life Technologies, Carlsbad, CA).This medium was supplemented for 2 days with B27 supplement(1%, without retinoic acid; Life Technologies) and N2 supple-ment (0.5%; Life Technologies). Preconditioned cells were thenpassaged using Accutase (Life Technologies) and transferred onto10-cm dishes coated with Matrigel (BD Biosciences, San Jose,CA) at a passaging ratio of 1:3. Cells were then maintained for 2more days in NIM. Five days after induction using NIM, themedium was changed to neural precursor media (NPM), whichconsisted of a 1:1 ratio of DMEM/F12 and neurobasal medium,supplemented with B27 supplement (0.5%) and N2 supplement(0.5%), as well as 20 ng/mL of basic fibroblast growth factor(FGF-2; Peprotech, Rocky Hill, NJ). Upon 90–100% confluence,cells were passaged at a ratio of 1:2 (harvested vs. plated cells)and plated onto Matrigel-coated dishes. The culture medium waschanged every other day. After 2–4 days in NPM, cells had as-sumed a flattened, bipolar morphology, typical of neural stemcells (NSCs). To characterize the predifferentiated NSCs andhL1-on and -off cells before transplantation, cells (1.5 · 104) wereplated for indirect immunofluorescence (IF) into four-well glasschamber slides coated with Matrigel (2 h, 37!C). After 2 days,cells were fixed with 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS, pH 7.4) for 15 min at room temperature and,after washing in PBS, for 10 min with PBS containing 0.5% TritonX-100 for permeabilization. After washing in PBS, primary an-tibodies (Abs) were added (diluted in 0.1% Triton X-100, 1%bovine serum albumin [BSA], and 3% nonimmune goat serum)and incubated with cells for 1 h at room temperature. Primary Abswere mouse monoclonal anti-nestin (1:200, catalog no.:MAB3526; Millipore, Temecula, CA), anti-A2B5 (1:300, catalogno.: MAB312; Millipore), and rabbit polyclonal anti-glial fi-brillary acidic protein (GFAP; 1:200, catalog no.: G4546; Sigma-Aldrich, St. Louis, MO), anti-doublecortin (DCX; 1:250, catalogno.: AB18723; Abcam, Cambridge, MA), anti-octamer-bindingtranscription factor 4 (Oct4; 1:500, catalog no.: AB3209; Milli-pore), anti-Ki67 (1:200, catalog no.: AB833; Abcam), anti-betaIII tubulin (1:300, catalog no.: PRB435P; Covance, Emeryville,CA). Secondary Abs were Alexa 555–conjugated goat anti-mouseimmunoglobulin G (IgG) or Alexa 555–conjugated goat anti-rabbit IgG (1:400, catalog no.: 115-001-003; Jackson Im-munoresearch, West Grove, PA) diluted in the buffer used fordilution of primary Abs and incubated for 30 min at room tem-perature. After washing in PBS, 4’,6’-diamidino-2-phenylindole(DAPI; 1 lg/mL) was added for 10 min at room temperature.

Slides were rinsed with PBS and mounted in Aqua Poly/Mountmedium (Polysciences, Warrington, PA), sealed with nail polish,and stored at 4!C. Images were captured with an Axiovert200Fluorescence Live Cell Imaging Workstation (Carl Zeiss AG,Jena, Germany).

Transfection and generation of stable cell lines

Sequence-verified, endotoxin-free pTet-off-GFP and pTet-off-hL1 plasmids were transfected into mouse neuroblastoma N2aand rat C6 glioma cells for measuring regulatable GFP and hL1expression, respectively. After transfection into N2a and C6 cellsusing Fugene HD (Roche Applied Science, Indianapolis, IN) at aratio of 5:2 of Fugene HD versus plasmid DNA, cells were treated24 h later with G418 (200 lg/mL) for selection of stable celllines. One week thereafter, 12 pTet-off-GFP-N2a clonal cell lineswere isolated and expanded in DMEM high-glucose, 1-mMNa-pyruvate, 10% fetal bovine serum (FBS), 1% penicillin/streptomycin with G418 to analyze regulation of GFP expression.To analyze regulatable hL1 expression, stably transfected andselected pTet-off-hL1-C6 cells were maintained with or withoutDOX (1 lg/mL) to generate hL1-off and -on cells in DMEM/F12,GlutaMAX, 10% FBS, 1% penicillin/streptomycin, and thenprobed for inducible expression of hL1 by Western blot analysis.To generate clonal pTet-off-GFP-H9NSC lines, the rat primaryNucleofector kit (Lonza, Allendale, NJ) was used according to themanufacturer’s instructions. In brief, after passaging, 4 · 106 cellsin 100 lL of Nucleofector solution were incubated for 10 min atroom temperature with 2 lg of plasmid DNA. The mixture of cellsand DNA was transferred to 96-well plates and electroporatedusing Lonza software at a setting of ‘‘Neuron Rat High Effi-ciency’’ (Lonza). After transfection, 80 lL of warm culture me-dium was added to each well, and the cell suspension wastransferred to new six-well dishes with FGF-2 (20 ng/mL). G418was added 2 days after transfection at a concentration of 50 lg/mL, being increased to 200 lg/mL after 4 days, when the culturemedium was changed, and maintained for 20 days. The three celllines with the highest percentage (approximately 95% GFP-positive cells with high, middle, and weak fluorescence intensi-ties) of GFP-positive cells were collected as colonies under afluorescence microscope. To generate the stable inducible hL1-expressing cell lines, the same procedure as for the generation ofthe pTet-off-GFP-H9NSC line was used. H9NSCs were trans-fected with plasmid pTet-off-hL1, except that cells weremaintained with or without DOX (1 lg/mL) to generate hL1-offand -on cells, respectively. After five passages, clones wereexpanded and conserved in liquid nitrogen.

Animals and spinal cord injury

Eight-week-old C57BL/6 female mice, purchased from theCharles River Laboratories (Wilmington, MA), were deeply an-esthetized with ketamine-xylazine (ketamine, 160 mg/kg; xyla-zine, 24 mg/kg; Butler Schein Animal Health, Chicago, IL) andsubjected to spinal cord compression injury, as detailed before.24–26

Animals were maintained in the core animal facility at theDivision of Life Science and the W.M. Keck Center for Colla-borative Neuroscience (Rutgers University). After surgery, micewere kept on a warm mat (35!C) for several hours to prevent hy-pothermia, being thereafter singly housed in a temperature- andhumidity-controlled room with water and standard food providedad libitum. Bladders were manually voided once- or twice-daily,depending on the palpability of the bladder. Animals were trans-cardially perfused under anesthesia with 4% PFA in PBS for his-tological and immunohistological analyses, as previouslydescribed.7,27 All experimental procedures were approved by theanimal care and facilities committee of Rutgers, The State Uni-versity of New Jersey.

556 YOO ET AL.

Surgery and cell transplantation into cyclosporine-immunosuppressed mice

Three days before transplantation, mice were injected intraperi-toneally (i.p.) with cyclosporine (10-mg/kg dose) for immunosup-pression, which was continued daily after transplantation. Fortransplantation, mice were anesthetized by an i.p. injection ofketamine/xylazine, and bupivacaine (0.1 mL of 0.125%; Hospira,Lake Forest, IL) was injected around the incision site to providelocal anesthesia. A 3-cm skin incision along the median line on theback of the animals was made and laminectomy was performed withMouse Laminectomy Forceps (Fine Science Tools, Heidelberg,Germany) at the T7–T9 level, followed by a mechanically con-trolled compression injury using a mouse spinal cord compressiondevice.24–26 The spinal cord was compressed for 1 sec for the severecompression injury with a time- strength-controlled electromag-netic device. Both hL1-expressing (hL1-on) cells and hL1-non-expressing (hL1-off ) H9NSCs were labeled using the Qtracker CellLabeling Kit (Life Technologies), according to the manufacturer’sinstructions. Cell transplantation was performed immediately aftercompression injury by inserting a 33-gauge needle connected to a5-lL Hamilton syringe (Hamilton, Reno, NV) using a stereotacticmicromanipulator (Narishige, East Meadow, NY). One microliter ofthe cell suspension (105 cells/lL) was injected 1 mm deep into thecord mid-line of the lesion site and 0.5 mm rostral and caudal to itwith each injection lasting for 7 min. The skin was closed withwound clips. Mice injected with hL1-off H9NSCs were treated withDOX by administration through the drinking water at a concentra-tion of 250 lg/mL in 3% sucrose solution distributed in amberbottles for protection from degradation by light. Mice injected withhL1-on H9NSCs were supplied with 3% sucrose solution withoutDOX. Solutions were changed and measured every other day duringthe course of the experiment. To test for GFP inducibility in vivo, wefollowed the same procedure as the one described above, but wetransplanted GFP-on cells into the injured spinal cord. Mice werethen maintained with or without DOX in the drinking water for 7and 10 days to test for GFP induction in vivo.

Locomotor assessment

We assessed locomotor function by the BMS score25,28,29 1week before and every week after injury. For assessment, micewere allowed to move in an open field, 1 m in diameter, for 5 min.Hindlimb movements were observed and scored according to theBMS scale by two expert and independent observers, blinded to thetreatment.

Immunohistology

Animals were deeply anesthetized with an i.p. injection of keta-mine/xylazine followed by vascular washout with PBS and transcar-dial perfusion with 4% PFA in PBS. Spinal cords were removed andcryoprotected by incubation in 20% sucrose in PBS overnight at 4!C,frozen, and cut into 20-lm-thick serial sections in a sagittal planerostral and caudal to the lesion site. Sections were mounted on mi-croscope slides and saved at - 80!C. Sections of equivalent distance

from the lesion center of each group were thawed to room tempera-ture, washed three times, and blocked with 10% goat serum in PBS for2 h at room temperature. Slides were incubated overnight at 4!C forimmunostaining with mouse monoclonal anti-human L1 Ab (1:400,catalog no.: UJ127; Abcam), anti-chondroitin sulfate Ab (CS56;1:200, catalog no.: C8035; Sigma-Aldrich), and rabbit polyclonal Absto tyrosine hydroxylase (TH; 1:500, catalog no.: AB152; Millipore),and serotonin (5-HT; 1:400, catalog no.: 10385; Abcam). For negativecontrol, nonimmune mouse IgG (1:400, catalog no.: ab37355; Abcam)was used instead of the specific primary Abs. After washing with PBS,slides were incubated with Alexa 555–conjugated goat anti-mouseIgG (1:800, catalog no.: 115-001-003; Jackson Immunoresearch) orAlexa 555–conjugated goat anti-rabbit IgG (1:800; Jackson Im-munoresearch) in PBS for 2 h at room temperature. Some sectionswere incubated with DAPI, rewashed with PBS, mounted with AquaPoly/Mount medium (Polysciences), and tile imaged with an Ax-iovert200 Fluorescence Live Cell Imaging Workstation (Carl Zeiss).

Quantification of immunofluorescence, and Westernblot and cell migration analysis

Fluorescence intensities of spinal cord areas immunolabeled forhL1 were quantified using four serially spaced (400 lm apart) mid-sagittal sections in the rostrocaudal direction from 4 animals. Pho-tographic documentation was performed with the Axiovert200Fluorescence Live Cell Imaging Workstation (Carl Zeiss), AxioVi-sion software (Carl Zeiss), and ImageJ software (National Institutesof Health, Bethesda, MD). Both immunostaining and imaging wereperformed under identical conditions. Staining intensity thresholdsfor Ab were determined after all images were acquired to optimize thesignal-to-noise ratio for a particular Ab. The threshold selected was55 (within the full range of intensities extending from 0 to 255) forq-dot, 85 for hL1, 90 for sections 0.5 mm away from the injury centerto evaluate migration of hL1 immunopositive H9NSCs, 85 for 5HT,75 for TH, and 70 for CS56. Mean fluorescence intensity (MFI) of thearea of immunoreactivity was at 0.8 mm equidistant rostral andcaudal from the center of the injury site for hL1 and CS56. Then,serial sections, 400 lm spaced apart, were evaluated starting from theinjury center up to a rostal and caudal distance of 1.5 mm to analyzethe MFI of hL1. For 5HT and TH, mean immunofluorescence in-tensities were measured 0.5 mm caudal to the injury site, with in-tensities higher than the thresholds stated above. These values werenormalized to the total tissue areas.

Western blot analysis for hL1 has been described previously.27

Briefly, stably transfected pTet-off-hL1 C6 and H9NSC lines weregenerated with or without DOX (1 lg/mL) and saved as pellets at- 80!C until use for Western blot analysis. Cells were thawed on ice,lysed by triturating in radioimmunoprecipitation assay buffer(Sigma-Aldrich), and centrifuged at 1000 · g and 4!C for 20 min toremove insoluble matter. Concentration of extracted proteins wastested by bicinchoninic acid (Pierce Biotechnology, Rockford, IL).A total of 30 lg of protein solution was boiled for 5 min in sodiumdodecyl sulfate (SDS) sample buffer and separated by 4–12% gra-dient SDS/polyacrylamide gel electrophoresis (Life Technologies).Proteins were electroblotted onto polyvinylidene difluoride mem-branes, blocked, and probed with hL1 monoclonal Ab (1:400,

FIG. 3. Nonviral single pTet-off-GFP and pTet-off-hL1 systems efficiently silence gene expression in a doxycycline dose- and time-dependent manner. The pTet-off-GFP plasmid vector was transfected into N2a cells, which were then treated with doxycycline at 0, 0.5,and 1 lg/mL (A, B, and C, respectively). GFP was efficiently silenced by 1 lg/mL of doxycycline in N2a cells within 3 days (G). ThepTet-off-GFP system was transfected into H9NSCs and a clonally selected line was treated with doxycycline at 1 lg/mL for 0, 4, and 8days (D, E, and F, respectively). GFP was silenced 8 days after starting the doxycycline treatment of H9NSCs (H). Western blotanalysis from the selected cell line after pTet-off-hL1 system transfection shows that hL1 expression was silenced by doxycycline in C6cells and H9NSCs (I and J, respectively; n = 3 experiments). Asterisks indicate significant differences between the groups. **p < 0.01, asassessed by one-way ANOVA, followed by Tukey’s post-hoc analysis. Data represent means – standard error of the mean; n = 4 imagesfor each cell line. Scale bar, 100 lm for all panels. GFP, green fluorescent protein; GAPDH, glyceraldehyde 3-phosphate dehydro-genase.

‰

REGULATED L1 EXPRESSION AND REGENERATION 557

558 YOO ET AL.

catalog no.: UJ127; Abcam) or polyclonal flag Ab (1:3000, catalogno.: ab1162; Abcam). Secondary mouse or rabbit Abs conjugated tohorseradish peroxidase with enhanced chemiluminescence intensi-fication (Pierce Biotechnology) were used for detection of hL1.

Statistical analysis

All numerical data are presented as group mean values withstandard error of the mean. The statistical significance of the BMSscore and mean immunoreactivity intensity for each group were es-timated by one-way analysis of variance, followed by Tukey’s post-hoc test. p values < 0.05 were considered statistically significant.

Results

Constructs of the pTet-off-GFP and pTet-off-hL1systems

The novel nonviral single Tet-off plasmid systems, pTet-off-GFPand pTet-off-hL1, were constructed as described in the Methodssection. To increase the efficiency of inducibility, we included sevenrepeats of the TRE between two CAG promoters placed in oppositetranscriptional orientations. These two promoters drive expressionof GFP or hL1 and the tetracycline reverse transactivator (Fig. 2). Inthe absence of DOX (on), the 7 · TRE are dormant, allowing theactivity of the CAG promoters to control the transcription of GFP or

hL1 and of the transrepressor (Fig. 2A,C, respectively). In thepresence of DOX (off ), the TREs are bound by the transrepressor,being fused to a strong KRAB repressor, which leads to repressionof the CAG promoters and silencing of GFP or hL1 and the trans-repressor (Fig. 2B,D, respectively). For construction of the nonviralsingle pTet-off-hL1 plasmid, the GFP insert was replaced by thefull-length hL1 insert with flag tagged in the pTet-off-GFP plasmid.The pTet-off-GFP-transfected clonal cell line, N2a-#12, was main-tained in the presence of DOX at 0, 0.5, and 1 lg/mL (Fig. 3A, B, andC, respectively). We observed that expression of GFP was reduced by1 lg/mL of DOX in the N2a-#12 cell line within 3 days (Fig. 3C).Time-dependent regulation by DOX of GFP in GFP-on and -off cellsin the pTet-off GFP system stably transfected cell line, N2a-#12, wasobserved in GFP-off cells without DOX, with GFP-off cells expres-sing GFP after 9 days (Supplementary Fig. 1) (see online supple-mentary material at http://www.liebertpub.com). A bar graph inFigure 3G shows that in the presence of 0.5 and 1 lg/mL of DOX,GFP was silenced, in comparison to cells maintained in its absence. Aclonal pTet-off-GFP-H9NSC line was clonally selected 4 weeks aftertransfection with the GFP system. The clonal line was treated withDOX at 1 lg/mL for 0, 4, and 8 days (Fig. 3D, E, and F, respectively).The bar graph in Figure 3H illustrates the effect of the differentexposure times to DOX on GFP expression in H9NSCs, which wassignificantly reduced by day 4 and not detectable by day 8. To test

FIG. 4. H9NSCs express stage-specific markers. Immunofluorescence staining of nestin, A2B5, and doublecortin (DCX) (A, B, and C,respectively). A phase-contrast image of the cells demonstrating rosette formation (D). Bar graph showing the percentage of cellspositive for each of the neural stem cell markers (E; n = 3 experiments). More than 96% of all cells are positive for the neural stem cellmarker, nestin, approximately 18% of cells are positive for the neuroglial progenitor marker, A2B5, and 11% of cells are positive for theearly neuronal progenitor marker, DCX. The astrocyte marker, GFAP, and the embryonic stem cell marker, Oct4, were not detected (E).Scale bar, 100 lm for all panels. GFAP, glial fibrillary acidic protein; DAPI, 4’,6-diamidino-2-phenylindole; Oct4, octamer-bindingtranscription factor 4.

REGULATED L1 EXPRESSION AND REGENERATION 559

DOX specificity and toxicity, we constructed a nonregulatable plas-mid pCAG-GFP to generate a clonal nonregulatable pCAG-GFP-H9NSC line, showing that GFP expression and cell viability were notaffected (99% of all cells are GFP positive) by treatment with DOX(1 lg/mL) during 2 weeks (Supplementary Fig. 2) (see online sup-plementary material at http://www.liebertpub.com). Western blotanalysis of the stably hL1-expressing cell line with the pTet-off-hL1system showed that expression was silenced by DOX in C6 cells andin H9NSCs (Fig. 3I and J, respectively).

Characterization of H9NSCs and hL1-H9NSCs

The phenotype of H9NSCs was characterized by indirect IF forexpression of nestin, A2B5, and DCX (Fig. 4). Nestin, a marker forneural stem cells, was strongly positive in H9NSCs. The neuralprogenitor marker, A2B5, and the neuronal progenitor marker,DCX, were only weakly expressed (Fig. 4A–C). Phase-contrastmicroscopy showed rosette formation, characteristic of neural pro-

genitor cells derived from ESCs (Fig. 4D). A bar graph demon-strates the percentage of cells immunoreactive for each marker (Fig.4E), with more than 96% of all cells being positive for nestin, 18%positive for the glial progenitor marker, A2B5, and 11% positive forthe early neuronal progenitor marker, DCX. The astrocyte marker,GFAP, and ESC marker, Oct4, were not detected (Fig. 4E).

The pTet-off-hL1 plasmid was introduced into H9NSCs by elec-troporation in the presence (off ) or absence (on) of DOX (1 lg/mL).Within five passages, the index of each passage for hL1-on cellsshowed more colonies 4 weeks after plating than hL1-off cells(Fig. 5A). hL1-off cells showed a faster doubling time, starting withinthe first passage, than the hL1-on cells (Fig. 5B). After five passagesunder selective pressure of G418 to obtain stable cell lines for hL1-onand -off, cells were characterized for differences in cell type anddevelopmental stage-specific markers using immunocytochemistry.hL1-on cells were reduced by 28% for the proliferation marker, Ki67,and by 8% for the NSC marker, nestin, compared with hL1-off cells(Supplementary Fig. 3A and 3D and 3B and 3E, respectively) (seeonline supplementary material at http://www.liebertpub.com). Ex-pression of the neuronal progenitor marker, DCX, and the matureneuronal marker, beta III tubulin, was not different between cells(Supplementary Fig. 3C and F and G and H, respectively) (see onlinesupplementary material at http://www.liebertpub.com). The bar graphshowed differences for Ki67 and nestin expression by hL1-on versushL1-off cells (Supplementary Fig. 3I, representing the means of threeindependent experiments, with nine images for each experiment)(see online supplementary material at http://www.liebertpub.com).

Evaluation of regulatable expression of hL1 in vivousing pTet-off-GFP-H9NSCs

Before we transplanted hL1-on and -off cells, we testedregulatable expression of GFP using the stable clonal cell line,

FIG. 5. Expression of exogenous hL1 leads to increased colonyformation and lower cell proliferation. H9NSCs were transfectedwith the pTet-off-hL1 plasmid system by electroporation in thepresence (off ) or absence (on) of doxycycline (1 lg/mL). After 4weeks, numbers of colonies were counted (A) and the cells weremaintained for five passages. The doubling index of each passagewas measured (B). Asterisks indicate significant differences be-tween the doxycycline-treated and nontreated groups at the sametime points (*p < 0.05), as assessed by the two-side t-test; n = 3experiments. DOX, doxycycline.

FIG. 6. Increased functional recovery in mice transplanted withhL1-on cells after severe spinal cord compression injury. Onemicroliter containing 105 hL1-on or -off cells were transplanted,immediately after severe compression injury, into three locations:the injury site and 0.5 mm rostral and caudal to the injury site. TheBasso Mouse Scale for analysis of locomotor activity was used toscore functional recovery for 6 weeks after severe spinal cordinjury (SCI). Asterisks indicate significant differences (*p < 0.05)between the transplanted groups at the same time points, beingdetectable at 5 and 6 weeks by one-way ANOVA for repeatedmeasurements, followed by Tukey’s post-hoc analysis. Data rep-resent means – standard error of the mean. Numbers of mice areindicated in brackets. DOX, doxycycline.

560 YOO ET AL.

pTet-off-GFP-H9NSC, for 7 and 10 days using DOX (SupplementaryFig. 4) (see online supplementary material at http://www.liebertpub.com). Mice maintained with without DOX in the drinking watershowed similar levels of GFP and red quantum dot expression after7 days (Supplementary Fig. 4A) (see online supplementary materialat http://www.liebertpub.com). Mice maintained with DOX inthe drinking water showed a decrease of GFP expression, whereasmeasurements for quantum dot showed no significant changes after7 days between treatments with and without DOX (SupplementaryFig. 4B) (see online supplementary material at http://www.liebertpub.com). The bar graph (Supplementary Fig. 4C) (see online sup-plementary material at http://www.liebertpub.com) indicates thecomparison of mice maintained with DOX and mice maintainedwithout DOX, as illustrated by the ratio of red to green. By 7 and 10days after injection of cells, GFP expression was reduced by 61 and56%, respectively.

Mice engrafted with hL1-on cells show betterlocomotor recovery than hL1-off cells

Immunosuppressed mice were injected at three sites into theacutely lesioned spinal cord with 1 lL containing 1 · 105 cells intothe lesion site and 0.5 mm rostral and caudal to the lesion site. DOX

(250 lg/mL) was included in the drinking water to maintainsilencing of hL1 expression. Body weight at 6 weeks after additionof DOX into the drinking water and water consumption was notdifferent between the DOX-treated and nontreated groups(Supplementary Fig. 5A and B, respectively) (see online supple-mentary material at http://www.liebertpub.com). BMS was ana-lyzed weekly to score for locomotor activity over the time period6 weeks after injury. Mice engrafted with hL1-on cells showedbetter recovery than mice engrafted with hL1-off cells, with amarked difference at 5 and 6 weeks after injury (Fig. 6).

Expression of hL1, TH, chondroitin sulfate,and migration of H9NSCs in engrafted mice

Six weeks after injury and injection, hL1-on and -off cells hadsurvived in the host spinal cord and had migrated away from theinjection sites. hL1 immunoreactivity was more intense in miceengrafted with hL1-on cells not treated with DOX than in micetreated with DOX, as quantified by ImageJ software (Fig. 7A vs. B).MFIs of the area at 0.8 mm equidistant rostral and caudal to thelesion center showed more hL1 immunoreactivity with hL1-oncells versus hL1-off cells (Fig. 7C). Immunostaining for hL1 incells labeled in vitro before injection with red quantum dots showed

FIG. 7. hL1-on cells transplanted into severely compression-injured mouse spinal cords express hL1 for at least 6 weeks. Onemicroliter containing 105 hL1-on or -off cells were transplanted immediately after a severe compression injury into three locations: theinjury site and 0.5 mm rostral and caudal to the injury site. After 6 weeks, mice were perfused, and sagittal spinal cord sections wereanalyzed by immunofluorescence using an antibody specific for human L1, not reacting with mouse L1. hL1 immunoreactivity is moreintense in spinal cords of mice that were not given doxycycline in their drinking water (hL1-on) (A), compared with mice givendoxycycline in their drinking water (hL1-off ) (B). Immunoreactivity of the entire image quantified above threshold using ImageJsoftware (National Institutes of Health, Bethesda, MD), 0.8 mm equidistant rostral and caudal to the injury center (C). Arrows indicatethe injury site. Asterisks indicate significant differences between the groups: **p < 0.05, as assessed by two-side t-test. Data representmeans – standard error of the mean (n = 4 mice; in total, 12 slices were analyzed). Scale bar, 200 lm for all panels. DOX, doxycycline.Color image is available online at www.liebertpub.com/neu

REGULATED L1 EXPRESSION AND REGENERATION 561

overlap for hL1 (Fig. 8A,B, purple) with quantum dots (Fig. 8C, D,red), as observed in the merged images (Fig. 7E,F). The q dotlabeling intensity was less pronounced in hL1-on cells than in hL1-off cells. It is likely that higher labeling intensity in aggregatedhL1-off cells resulted from charge transfer between closelyneighboring cells. In mice transplanted with hL1-on cells (Fig.8A,C,E), migration away from the injury site was better than forhL1-off cells (Fig. 8B,D,F). Also, hL1-on cells migrated better upto 1.5 mm away from the injury site in the rostral and caudal di-rection than hL1-off cells (Fig. 8G). We also tested the neuronalmarkers, TH and 5-HT, as well as the glial scar marker, chondroitinsulfate (CSPG). In both groups, immunoreactive 5-HT axons werenot detected in the caudal area 0.5 mm away from the injury center

(Fig. 9A,D). However, TH immunoreactive axons were moreabundant in this area, as indicated by arrowheads in Figure 9B, withhL1-on cells versus hL1-off cells in Fig. 9E. In addition, the volumeof the glial scar was reduced in mice having received the hL1-oncells, as compared with hL1-off cells (Fig. 9C,F). The bar graphillustrates that mice having received hL1-on cells showed a highermean IF intensity of TH and lower CSPG IF intensity, whencompared with hL1-off cells (Fig. 9G).

Discussion

The aim of the present study was to demonstrate that regulatableexpression of the regeneration-conducive adhesion molecule, L1,can be used as a mode to improve cell-based therapy for spinal cord

FIG. 8. H9NSCs expressing hL1 migrate better in injured mousespinal cords than H9NSCs not expressing hL1 after severe spinalcord compression injury. Immediately after spinal cord compres-sion, 1 lL containing 105 hL1-on and -off quantum dot (red) labeledcells were transplanted at 0.5 mm rostral and caudal from the injurysite. After 6 weeks, mice were perfused, and sagittal spinal cordsections were analyzed by immunofluorescence using an antibodyagainst human L1. Mice transplanted with hL1-on cells in the ab-sence of doxycycline (A, C, and E). Mice transplanted with hL1-offcells in the presence of doxycycline (B, D, and F). hL1-on cells hadmigrated better up to 1.5 mm away from the injury site in the rostraland caudal directions, compared with hL1-off cells (G). Mergedimage showing colocalization of hL1 immunofluorescence andquantum dot labeling (E and F). Data represent means – standarderror of the mean (n = 3 mice; in total, nine slices were analyzed).Scale bar, 300 lm for all panels. Color image is available online atwww.liebertpub.com/neu

FIG. 9. hL1-expressing H9NSCs transplanted into the injuredspinal cords express higher levels of TH and show a smaller areaof chondroitin sulfate immunoreactivity. Immediately after spinalcord injury, 1 lL containing 105 hL1-on or -off quantum dot (red)labeled cells were injected 0.5 mm rostral and caudal from theinjury site. After 6 weeks, mice were perfused, and sagittal spi-nal cord sections were analyzed by immunofluorescence withantibodies against 5-HT and TH and the antibody, CS56. Micetransplanted with hL1-on cells in the absence of doxycycline(A, B, and C). Mice transplanted with hL1-off cells in the pres-ence of doxycycline (D, E, and F). Immunoreactivity of an entireimage quantified above threshold using ImageJ software (NationalInstitutes of Health) (G). Mean fluorescence CS56 immunore-activities in the area at 0.8 mm equidistant rostral and caudal to thelesion site for CS56 immunoreactiviy and 0.5 mm caudal to thelesion site for 5-HT and TH were compared between hL1-on and-off cells. Dotted lines in panels indicate the injury site. Rostral inthe panels is left. Asterisks indicate significant differences betweenthe groups: *p < 0.05, as assessed by the two-side t-test. Data rep-resent means – standard error of the mean (n = 3 mice for 5-HTand TH; in total, nine slices were analyzed; n = 4 mice for CS56;in total, 11 slices were analyzed). Scale bar, 200 lm for all panels.5-HT, 5-hydroxytryptamine (serotinin); TH, tyrosine hydroxylase.Color image is available online at www.liebertpub.com/neu

562 YOO ET AL.

injury (SCI). Engineered derivatives of human ESCs comprisingthe L1 sequence under the control of the tightly regulatable Tet-offsystem were phenotypically advantageous in vitro and out-performed their unmodified counterparts in vivo.

This novel Tet regulation system is also functional in other celltypes, such as N2a neuroblastoma and C6 glioma cells and wild-type (WT) H9NSCs. As predicted, a critical increase in L1 ex-pression over nominal basal levels normally found in H9NSCsenhances their survival and migration as well as promotes loco-motor recovery after injury in a mouse model of SCI. Here, weavoided the use of viral expression systems to minimize con-cerns for future use of this system in translational approaches toameliorate the consequences of devastating acute and chronic le-sions to the human nervous system.

The benefits of the novel regulatable system include the fol-lowing: 1) gene expression driven by two independent and strongCAG promoters to regulate a range of gene expression; 2) inclusionof 7 · TREs located between the two promoters to increase sensi-tivity of regulation; 3) generation of a single plasmid systemcombining TRE and transrepressor to reduce leakage of expression;4) driving of promoter activity by the transrepressor is also reg-ulatable by DOX, thus avoiding redundant transrepressor expres-sion; and 5) selection of the transfected cells in vitro by G418independently of regulation by DOX.

Compared with the Tet-on system, the Tet-off system has dis-advantages regarding future therapeutic applications, because ap-plication of DOX to patients is not feasible for prolonged times.The aim of the present study was, however, designed only todemonstrate the possibility to regulate overexpression of L1 inhuman stem cells. We used the Tet-off in this study because it is lessleaky in expression of the regulated molecule than the Tet-onsystem. Newer generations of inducible systems are currently be-coming available and will be considered in the future. Presently,this hybrid Tet-off system can be used with a strong promoter tooverexpress the target gene at crucial times after transplantation ofstem cells, but then be able to control the silencing of target mol-ecule expression. Our aim in the present study was met because wecould demonstrate that L1 expression is regulatable.

The observation that L1-expressing cells tend to form morecolony-forming units (aggregates) in vitro and proliferate less thanthe cells not induced to express L1 is interesting from two points ofview. Whereas L1 is expressed endogenously in H9NSCs, levels ofexpression appear too low to enhance regeneration, in comparisonwith hL1-on cells. Here, we postulate that the increased expressionand density of L1 at the cell surface allows enhanced homo- andheterophilic cis- and transinteractions, which allows beneficialconsequences in vivo. It is noteworthy that, in vivo, cells that arehL1-off proliferated better than hL1-on cells, as observed before forneuronal cells in vitro and in vivo, with L1 expression being up-regulated in postmitotic neurons. Similarly, Schwann cells lackingexpression of L1 proliferate more after a peripheral nerve lesionthan their WT counterparts.30,31 Interestingly, L1 expression bytumor cells of different origin correlates positively with their mi-gratory and metastatic potential. For tumor cells, it has not beendetermined by which molecular mechanisms L1 may contribute toenhanced or reduced proliferation, a question that appears to beeminent in characterizing the functional roles of L1 in tumor bi-ology.32 With the availability of an L1 construct that is capable ofregulating L1 levels and that can be expressed in different celltypes, it appears that this problem can now be tackled experimen-tally in vitro and in vivo. Thus, with the plasmid system thatwe have constructed, it may be feasible not only to engineer a

regeneration-conducive, but also precarious adhesion moleculeafter injury in human stem cells as well as to use this regulatablefeature for gaining insights into the function of this molecule intumor cells and in the developing and adult nervous system, whereproliferation and differentiation of neuronal progenitors and neu-rons are an important aspect of normal and abnormal functions.

Acknowledgments

The authors are very grateful to Dr. Jennifer Moore for providingH9NSCs and generous advice and to the New Jersey Commissionfor Spinal Cord Research for support. M.S. is New Jersey Professorof Spinal Cord Research. R.I.C. is supported, in part, by the SatellFoundation.

Author Disclosure Statement

No competing financial interests exist.

References

1. Abdel-Salam, O.M. (2011). Stem cell therapy for Alzheimer’s disease.CNS Neurol. Disord. Drug Targets 10, 459–485.

2. Francis, K.R., and Wei, L. (2010). Human embryonic stem cell neuraldifferentiation and enhanced cell survival promoted by hypoxic pre-conditioning. Cell Death Dis. 1, e22.

3. Sanchez-Pernaute, R., Studer, L., Ferrari, D., Perrier, A., Lee, H.,Vinuela, A., and Isacson, O. (2005). Long-term survival of dopamineneurons derived from parthenogenetic primate embryonic stem cells(cyno-1) after transplantation. Stem Cells 23, 914–922.

4. Shindo, T., Matsumoto, Y., Wang, Q., Kawai, N., Tamiya, T., andNagao, S. (2006). Differences in the neuronal stem cells survival,neuronal differentiation and neurological improvement after trans-plantation of neural stem cells between mild and severe experimentaltraumatic brain injury. J. Med. Invest. 53, 42–51.

5. Theus, M.H., Wei, L., Cui, L., Francis, K., Hu, X., Keogh, C., andYu, S.P. (2008). In vitro hypoxic preconditioning of embryonic stemcells as a strategy of promoting cell survival and functional benefitsafter transplantation into the ischemic rat brain. Exp. Neurol. 210,656–670.

6. Roonprapunt, C., Huang, W., Grill, R., Friedlander, D., Grumet, M.,Chen, S., Schachner, M., and Young, W. (2003). Soluble cell adhesionmolecule L1-Fc promotes locomotor recovery in rats after spinal cordinjury. J. Neurotrauma 20, 871–882.

7. Chen, J., Bernreuther, C., Dihne, M., and Schachner, M. (2005). Celladhesion molecule L1-transfected embryonic stem cells with en-hanced survival support regrowth of corticospinal tract axons in miceafter spinal cord injury. J. Neurotrauma 22, 896–906.

8. Runyan, S.A., Roy, R., Zhong, H., and Phelps, P.E. (2005). L1 CAMexpression in the superficial dorsal horn is derived from the dorsal rootganglion. J. Comp. Neurol. 485, 267–279.

9. Chen, S., Mantei, N., Dong, L., and Schachner, M. (1999). Preventionof neuronal cell death by neural adhesion molecules L1 and CHL1. J.Neurobiol. 38, 428–439.

10. Kleene, R., Yang, H., Kutsche, M., and Schachner, M. (2001). Theneural recognition molecule L1 is a sialic acid-binding lectin forCD24, which induces promotion and inhibition of neurite outgrowth.J. Biol. Chem. 276, 21656–21663.

11. Brummendorf, T. and Rathjen, F.G. (1994). Cell adhesion molecules.1: immunoglobulin superfamily. Protein Profile 1, 951–1058.

12. Saghatelyan, A., Carleton, A., Lagier, S., de Chevigny, A., and Lledo,P.M. (2003). Local neurons play key roles in the mammalian olfactorybulb. J. Physiol. Paris 97, 517–528.

13. Cui, Y.F., Hargus, G., Xu, J.C., Schmid, J.S., Shen, Y.Q., Glatzel, M.,Schachner, M., and Bernreuther, C. (2010). Embryonic stem cell-derived L1 overexpressing neural aggregates enhance recovery inParkinsonian mice. Brain 133, 189–204.

14. Cui, Y.F., Xu, J.C., Hargus, G., Jakovcevski, I., Schachner, M., andBernreuther, C. (2011). Embryonic stem cell-derived L1 over-expressing neural aggregates enhance recovery after spinal cord injuryin mice. PLoS One 6, e17126.

REGULATED L1 EXPRESSION AND REGENERATION 563

15. Bernreuther, C., Dihne, M., Johann, V., Schiefer, J., Cui, Y., Hargus,G., Schmid, J.S., Xu, J., Kosinski, C.M., and Schachner, M. (2006).Neural cell adhesion molecule L1-transfected embryonic stem cellspromote functional recovery after excitotoxic lesion of the mousestriatum. J. Neurosci. 26, 11532–11539.

16. Becker, C.G., Lieberoth, B.C., Morellini, F., Feldner, J., Becker, T.,and Schachner, M. (2004). L1.1 is involved in spinal cord regenerationin adult zebrafish. J. Neurosci. 24, 7837–7842.

17. Zhang, Y., Bo, X., Schoepfer, R., Holtmaat, A.J., Verhaagen, J.,Emson, P.C., Lieberman, A.R., and Anderson, P.N. (2005). Growth-associated protein GAP-43 and L1 act synergistically to promote re-generative growth of Purkinje cell axons in vivo. Proc. Natl. Acad.Sci. U. S. A. 102, 14883–14888.

18. Xu, G., Nie, D.Y., Wang, W.Z., Zhang, P.H., Shen, J., Ang, B.T., Liu,G.H., Luo, X.G., Chen, N.L., and Xiao, Z.C. (2004). Optic nerveregeneration in polyglycolic acid-chitosan conduits coated with re-combinant L1-Fc. Neuroreport 15, 2167–2172.

19. Yamanaka, H., Obata, K., Kobayashi, K., Dai, Y., Fukuoka, T., andNoguchi, K. (2007). Alteration of the cell adhesion molecule L1 ex-pression in a specific subset of primary afferent neurons contributes toneuropathic pain. Eur. J. Neurosci. 25, 1097–1111.

20. Suzuki, M., Kasai, K., and Saeki, Y. (2006). Plasmid DNA sequencespresent in conventional herpes simplex virus amplicon vectors causerapid transgene silencing by forming inactive chromatin. J. Virol. 80,3293–3300.

21. Szulc, J., Wiznerowicz, M., Sauvain, M.O., Trono, D., and Aebischer,P. (2006). A versatile tool for conditional gene expression andknockdown. Nat. Methods 3, 109–116.

22. Nat, R., Nilbratt, M., Narkilahti, S., Winblad, B., Hovatta, O., andNordberg, A. (2007). Neurogenic neuroepithelial and radial glial cellsgenerated from six human embryonic stem cell lines in serum-freesuspension and adherent cultures. Glia 55, 385–399.

23. Ying, Q.L., Stavridis, M., Griffiths, D., Li, M., and Smith, A. (2003).Conversion of embryonic stem cells into neuroectodermal precursorsin adherent monoculture. Nat. Biotechnol. 21, 183–186.

24. Steward, O., Zheng, B., and Tessier-Lavigne, M. (2003). False res-urrections: distinguishing regenerated from spared axons in the injuredcentral nervous system. J. Comp. Neurol. 459, 1–8.

25. Apostolova, I., Irintchev, A., and Schachner, M. (2006). Tenascin-Rrestricts posttraumatic remodeling of motoneuron innervation and

functional recovery after spinal cord injury in adult mice. J. Neurosci.26, 7849–7859.

26. Curtis, R., Green, D., Lindsay, R.M., and Wilkin, G.P. (1993). Up-regulation of GAP-43 and growth of axons in rat spinal cord aftercompression injury. J. Neurocytol. 22, 51–64.

27. Chen, J., Wu, J., Apostolova, I., Skup, M., Irintchev, A., Kugler, S.,and Schachner, M. (2007). Adeno-associated virus-mediated L1 ex-pression promotes functional recovery after spinal cord injury. Brain130, 954–969.

28. Basso, D.M., Fisher, L.C., Anderson, A.J., Jakeman, L.B., McTigue,D.M., and Popovich, P.G. (2006). Basso Mouse Scale for locomotiondetects differences in recovery after spinal cord injury in five commonmouse strains. J. Neurotrauma 23, 635–659.

29. Engesser-Cesar, C., Anderson, A.J., Basso, D.M., Edgerton, V.R., andCotman, C.W. (2005). Voluntary wheel running improves recoveryfrom a moderate spinal cord injury. J. Neurotrauma 22, 157–171.

30. Guseva, D., Zerwas, M., Xiao, M.F., Jakovcevski, I., Irintchev, A.,and Schachner, M. (2011). Adhesion molecule L1 overexpressedunder the control of the neuronal Thy-1 promoter improves myeli-nation after peripheral nerve injury in adult mice. Exp. Neurol. 229,339–352.

31. Guseva, D., Angelov, D.N., Irintchev, A., and Schachner, M. (2009).Ablation of adhesion molecule L1 in mice favours Schwann cellproliferation and functional recovery after peripheral nerve injury.Brain 132, 2180–2195.

32. Siesser, P.F., and Maness, P.F. (2009). L1 cell adhesion molecules asregulators of tumor cell invasiveness. Cell Adh. Migr. 3, 275–277.

Address correspondence to:Melitta Schachner, PhDCenter for Neuroscience

Shantou University Medical College22 Xin Ling Road

Shantou 515041Guandong Province

People’s Republic of China

E-mail: schachner@stu.edu.cn

564 YOO ET AL.