Combined Immunosuppressive Agents or CD4 AntibodiesProlong Survival of Human Neural Stem Cell Grafts andImprove Disease Outcomes in Amyotrophic Lateral SclerosisTransgenic Mice

JUN YAN,a LEYAN XU,a ANNIE M. WELSH,a DAVID CHEN,a THOMAS HAZEL,e KARL JOHE,e

VASSILIS E. KOLIATSOSa,b,c,d

aDepartment of Pathology, Division of Neuropathology, and Departments of bNeurology, cNeuroscience, anddPsychiatry and Behavioral Sciences, Johns Hopkins Medical Institutions, Baltimore, Maryland, USA; eNeuralStem,

Inc., Gaithersburg, Maryland, USA

Key Words. Differentiation • Motor neuron disease • Motor neurons • Regeneration • Superoxide dismutase

ABSTRACT

Amyotrophic lateral sclerosis (ALS) is a target for cell-replacement therapies, including therapies based on humanneural stem cells (NSCs). These therapies must be firsttested in the appropriate animal models, including trans-genic rodents harboring superoxide dismutase (SOD1) mu-tations linked to familial ALS. However, these rodent sub-jects reject discordant xenografts. In the presentinvestigation, we grafted NSCs from human embryonic spi-nal cord into the ventral lumbar cord of 2-month-old SOD1-G93A transgenic mice. Animals were immunosuppressedwith FK506, FK506 plus rapamycin, FK506 plus rapamycinplus mycophenolate mofetil, or CD4 antibodies. With FK506monotherapy, human NSC grafts were rejected within 1week, whereas combinations of FK506 with one or two of theother agents or CD4 antibodies protected grafts into end-stage illness (i.e., more than 2 months after grafting). The

combination of FK506 with rapamycin appeared to be op-timal with respect to efficacy and simplicity of administra-tion. Graft protection was achieved via the blockade of CD4-and CD8-cell infiltration and attenuation of the microglialphagocytic response from the host. Surviving NSCs differ-entiated extensively into neurons that began to establishnetworks with host nerve cells, including �-motor neurons.Immunosuppressed animals with live cells showed later on-set and a slower progression of motor neuron disease and livedlonger compared with immunosuppressed control animalswith dead NSC grafts. Our findings indicate that combinedimmunosuppression promotes the survival of human NSCsgrafted in the spinal cord of SOD1-G93A mice and, in doing so,allows the differentiation of NSCs into neurons and leads to theimprovement of key parameters of motor neuron disease.STEM CELLS 2006;24:1976–1985

INTRODUCTIONAmyotrophic lateral sclerosis (ALS) is characterized by progres-sive loss of motor neurons in the spinal cord and brain stem, as wellas some pyramidal neurons in motor cortex, leading to muscleatrophy and eventual paralysis and death. Approximately 10%–13% of cases of ALS are familial, and 20% of these cases harbormutations in the Cu/Zn superoxide dismutase (SOD1) gene [1, 2].Transgenic mice expressing mutant SOD1 with a substitution ofalanine for glycine at position 93 (SOD1-G93A) reproduce symp-toms and pathologies of human ALS [3–5]. Despite progress in theelucidation of several cellular and molecular mechanisms operatingin familial ALS, there is presently no disease-modifying strategy

that would prevent the progressive motor neuron death in familialor sporadic forms of the illness [1]. Therefore, neuronal replace-ment therapies with cell grafts are as applicable in ALS as they arein other diseases featured by neuronal cell death [6–8]. Our neu-ronal replacement capabilities have been expanded recently withthe availability of neural stem cells (NSCs) derived ex vivo fromneural tissues or in vitro from embryonic stem cells as alternativesto fetal grafts [7, 8].

Human NSCs (i.e., cells that may eventually be used inclinical applications) are different from their rodent counterpartsin important ways. For example, human NSCs proliferate anddifferentiate more slowly than rodent cells, and this property

Correspondence: Vassilis E. Koliatsos, M.D., The Johns Hopkins University School of Medicine, Neuropathology Division, RossBuilding, Room 558, 720 Rutland Avenue, Baltimore, Maryland 21205, USA. Telephone: 410-502-5191; Fax: 410-955-9777; e-mail:koliat@jhmi.edu Received October 18, 2005; accepted for publication April 10, 2006; first published online in STEM CELLSEXPRESS April 27, 2006. ©AlphaMed Press 1066-5099/2006/$20.00/0 doi: 10.1634/stemcells.2005-0518

TRANSLATIONAL AND CLINICAL RESEARCH

STEM CELLS 2006;24:1976–1985 www.StemCells.com

may alter the extent and/or rate of migration, axonal elongation,and synapse formation [9–11]. Therefore, prior to their consid-eration as potential ALS therapy, human NSCs must be studiedin vivo in the best available models of motor neuron disease(i.e., transgenic [Tg] rodents harboring SOD1 mutations). Asignificant obstacle here is the rejection, by rodent hosts, ofxenografts from species as distant as humans [12]. The use ofcyclosporin A as a single immunosuppressive agent has failed orpartially failed in other laboratories [13, 14] and our own [15,16]. In the present study, we aimed at optimizing human NSCxenograft survival in SOD1-G93A Tg mice with several com-binations of immunosuppressive agents. Our results indicate thatit is possible to achieve good survival and differentiation ofhuman NSCs in the spinal cord of SOD1-G93A Tg mice and, indoing so, to improve key clinical parameters of rodent ALS.

MATERIALS AND METHODS

Experimental Subjects and DesignHeterozygous male B6SJL-TgN (SOD1-G93A) 1Gur mice werepurchased from The Jackson Laboratory (Bar Harbor, ME,http://www.jax.org) and mated with wild-type females. Animalcare and surgical procedures were carried out according toprotocols approved by the Animal Care and Use Committee ofthe Johns Hopkins Medical Institutions (JHMI). Heterozygousoffspring of mixed gender were selected by genotyping at ap-proximately 1 month of age.

Because, in pilot studies, we had failed to achieve accept-able human NSC graft survival with routine cyclosporin immu-nosuppression, the present study was designed with emphasis onalternative immunosuppressive regimens. Based on evidence fora strong T-cell-mediated rejection in pilot studies, our goal herewas to use a potent combination of T-cell activation and prolif-eration blockers, capitalizing on commercially available cal-cineurin-dependent inhibitors (FK506, tacrolimus), non-cal-cineurin-mediated inhibitors (rapamycin, sirolimus), and inosinemonophosphate dehydrogenase inhibitors (mycophenolate es-ters such as mycophenolate mofetil [MMF]). Thus, five exper-imental groups were given different immunosuppressive treat-ments: the first group was treated with FK506 monotherapy(Prograf; Fujisawa Healthcare, Inc., Deerfield, IL, http://www.us.astellas.com) (2 �g/g per day, n � 23); a second group wastreated with a double immunosuppressive regimen combiningFK506 and rapamycin (Calbiochem, San Diego, http://www.embbiosciences.com) (1 �g/g per day each, n � 21); a thirdgroup received triple immunosuppression with FK506, rapamy-cin, and MMF (CellCept; Roche, Nutley, NJ, http://www.rocheusa.com) (1 �g/g per day each for FK506 and rapamycin,100 �g/g per day for MMF, n � 14); a fourth group was treatedwith CD4 antibodies (clone GK1.5, 20 �g/g per day, n � 12);and a fifth group (n � 10) was treated with FK506 plus rapa-mycin exactly as the second group, except that subjects weregrafted with dead cells. Dead-cell grafts used cells that wereexposed to repeated freezing-thawing and served to control forthe potential therapeutic effects of immunosuppressants them-selves on motor neuron disease. The FK506-rapamycin combi-nation was the only treatment given to animals with dead-cellgrafts because it is the simplest immunosuppressant regimenthat protects grafts from rejection (see Results). Males and

females were randomly admixed in the various experimentalgroups to minimize a systemic effect of gender on diseaseprogression and treatment response.

A group of animals was euthanized at 1 week (FK506monotherapy, n � 6; FK506 plus rapamycin treatment, n � 5;FK506 plus rapamycin plus MMF, n � 5; CD4 antibodies, n �4), and another group was euthanized at 1 month (FK506, n �5; FK506 plus rapamycin, n � 5; FK506 plus rapamycin plusMMF, n � 4; CD4 antibodies, n � 3) after grafting in order tomonitor host immune response to grafted cells and graft sur-vival. All others (FK506, n � 12; FK506 plus rapamycin, n �11; FK506 plus rapamycin plus MMF, n � 5; CD4 antibodies,n � 5; FK506 plus rapamycin in animals grafted with dead cells,n � 10) were allowed to survive to end-stage illness (as definedin the “Motor Scoring and Other Tests for Clinical Outcomes”section of Materials and Methods; typically �2 months aftergrafting) in order for us to assess clinical outcomes as well aslong-term graft survival and host responses.

Derivation and Culture of Human NSCsHuman NSCs were prepared from the cervical spinal cord of asingle 8-week human fetus donated by the mother in a mannercompliant with the guidelines of the National Institutes ofHealth and the U.S. Food and Drug Administration and ap-proved by an outside independent review board. All JHMIinstitutional guidelines were followed in obtaining and usingthese cells in our laboratory. The initial culture was expanded asmonolayer in poly-D-lysine and fibronectin-coated dishes usingserum-free medium containing fibroblast growth factor (FGF)-2as described [17]. The resulting cell line, termed “566RSC,” waspassaged 10 to 12 times prior to grafting. Five to 7 days prior tosurgery, one cryopreserved vial of cells was thawed, washed,and cultured again as monolayer. Cultures were seeded so as toreach confluence on the day of surgery. Cells were subsequentlyharvested by brief enzymatic treatment that deactivated FGF-2,washed in buffered saline, and used within 24 hours. As verifiedby trypan blue exclusion, viability of cells on ice was typicallygreater than 80% within this 24-hour period.

Preparation of Monoclonal Antibody AgainstMouse CD4The monoclonal antibody GK1.5, a rat immunoglobulin G (IgG)2b directed against a surface epitope of mouse CD4 cells, hasexcellent in vivo efficacy in multiple models of T-cell-mediatedrejection [12]. The antibody was prepared from GK1.5 hybrid-oma cells (a gift from Dr. William Baldwin, Department ofPathology, JHMI) and purified from cell culture supernatant orascites fluid with a protein G column. For cell culture, cells wereplated in plastic flasks containing hybridoma-SFM (serum-freemedium) medium with 1% ultra-low IgG fetal bovine serum andpenicillin/streptomycin (all from Invitrogen, Carlsbad, CA,http://www.invitrogen.com) in a humidified cell culture incuba-tor with CO2: ambient air � 5:95 (37°C). For ascites prepara-tion, 2-month-old male nude mice with the same strain back-ground as SOD1-G93A (B6.Cg-Foxn 1nu; The JacksonLaboratory) were primed by injecting 0.5 ml of Freund’s in-complete adjuvant 1 week before the injection of 1 � 106

hybridoma cells in 0.5 ml phosphate-buffered saline (PBS) i.p.Ascites fluid was collected 2 weeks later with multiple taps.

1977Yan, Xu, Welsh et al.

www.StemCells.com

Conditioned medium or ascites fluid were centrifuged at2,000g. Undiluted medium supernatant or ascites supernatant di-luted 1:10 in 20 mM phosphate buffer (PB; pH 7.4) were loadedonto a 5-ml Hi-trap protein G column (GE Healthcare, LittleChalfont, Buckinghamshire, U.K., http://www.gehealthcare.com).The column was washed with 20 mM PB (pH 7.4), and boundantibody was eluted with 0.1 M glycine-HCl (pH 2.9) and imme-diately neutralized with 1 M Tris-HCl (pH 9).

Surgical ProceduresSurgeries were carried out under gas anesthesia (enflurane/oxygen/nitrous oxide � 1:33:66) and aseptic conditions, withthe aid of a Zeiss surgical microscope (Carl Zeiss, Jena,Germany, http://www.zeiss.com) and a Kopf spinal stereotaxicunit (David Kopf Instruments, Tujunga, CA, http://www.kopfinstruments.com) fitted with a mouse mouth-and-nose de-vice. NSCs were grafted via a dorsal laminectomy window intothe ventral gray matter of the lower lumbar protuberance of8-week-old SOD1 G93A mice. To achieve a good correspon-dence with L4-L5, injections were targeted to the portion ofspinal cord immediately underneath the T12 vertebra. Cell sus-pensions were delivered under aseptic conditions via four injec-tions aimed at ventral horn (2 � 104 NSCs in 0.5 �l perinjection site, two injection sites on each side of the spinal cord1 mm apart) with pulled, beveled glass micro-pipettes connectedto 10-�l Hamilton microsyringes (Hamilton Company, Reno,NV, http://www.hamiltoncomp.com) via silastic tubing.

Animals were treated with four different immunosuppres-sive regimens as outlined in the “Experimental Subjects andDesign” section of Materials and Methods. Immunosuppressantcompounds were given with daily i.p. injections beginning 1 dayprior to grafting and ending on the day prior to euthanasia. Toprepare FK506 for monotherapy, the commercially availableFK506 solution (5 mg/ml) was diluted to 1 mg/ml with steriledistilled water. When FK506 was combined with rapamycin, 1mg of commercially available rapamycin powder was first dis-solved in 200 �l of dimethyl sulfoxide and then combined with200 �l of the commercial FK506 solution (5 mg/ml); the mix-ture was then diluted to a final 2-ml volume with the addition of1.6 ml of sterile distilled water. For triple immunosuppressionwith FK506, rapamycin, and MMF, rapamycin was mixed withFK506 first, and the dilution step was done with 1.6 ml of steriledistilled water in which 100 mg of MMF had been previouslydissolved. GK1.5 antibodies (diluted 10 mg/ml in sterile PBS)were injected for a total of 9 days beginning 1 day prior tografting and then for 5 consecutive days every 4 weeks untilsacrifice [12].

Motor Scoring and Other Tests forClinical OutcomesAnimals were weighed twice a week, and disease onset wasdefined as the time point when body weight started to decreaseconsecutively [18]. Motor strength testing was used to monitordisease progression and to determine terminal illness. Terminalillness, defined in greater detail below, was used as an endpointfor these experiments, according to the guidelines of the AnimalCare and Use Committee of JHMI. The Basso, Beattie, andBresnahan locomotor rating scale [19] that is commonly used toassess motor strength places emphasis on hindlimbs. In SOD1-G93A mice, motor deficits appear first in hindlimbs but subse-

quently spread to forelimbs. To appropriately address the four-limb involvement and accommodate for the small animal size,we adjusted a 5-point scale originally proposed by Wrathall etal. [20], primarily in order to distinguish between hindlimb andforelimb strength. The modified Wrathall scale scores weredefined as follows: 5, normal gait; 4, mild deficits: hindlimbslose some weight-bearing; 3, moderate deficits: hindlimbs losemost weight-bearing and toe clearance; 2, severe deficits: hind-limbs lose weight-bearing almost completely, and forelimbsbegin to show signs of weakness; and 1, end-stage disease:hindlimbs are completely paralyzed, forelimbs are affected tovarious degrees, and the mouse cannot right itself when laid onits side. All animals were euthanized at stage 1.

Statistical Analysis of Clinical OutcomesClinical outcomes were studied on the same set of data origi-nating from five concurrent treatment groups allowed to surviveto terminal illness. These outcomes included disease onset andlife span as well as disease progression based on consecutivemotor scores (motor score curves) and Kaplan-Meier survival.In one type of analysis, these outcomes were compared amongthe four groups of subjects treated with live cells but withvariable immunosuppression (i.e., FK506 [n � 12], FK506 �rapamycin [n � 11], FK506 � rapamycin � MMF [n � 5], andanti-CD4 antibodies [n � 5]). Variance in disease onset and lifespan was analyzed with one-way analysis of variance(ANOVA), followed by a Fisher least significant difference(LSD) post hoc test. Variance in motor score curves was ana-lyzed by repeated-measures ANOVA, followed by Fisher LSDtesting. Statistical analysis of Kaplan-Meier survival was basedon Log-rank testing.

In a second type of analysis, clinical outcomes were com-pared between two groups of animals, both of which wereimmunosuppressed with FK506 � rapamycin, but one hadreceived live cells whereas the other was treated with dead cells.Disease onset and life span between live- and dead-cell groupswere compared with a Student’s t test. Disease progression wascompared in the two groups by analyzing differences in motorscore and Kaplan-Meier survival curves as described above.

Histology, Immunocytochemistry, and MicroscopyAt end-stage disease and at 1 week and 1 month after grafting,mice were euthanized with an overdose of sodium pentobarbital(5 mg/100 g i.p.) followed by intracardiac perfusion with 4%freshly depolymerized, neutral-buffered paraformaldehyde. Spi-nal cord tissue blocks containing the entire lumbar region weredissected and immersion-fixed in paraformaldehyde for an ad-ditional 4 hours at room temperature after removing the dura.Tissues were then equilibrated in 30% sucrose and sectioned atthe transverse plane (30 �m) on a freezing microtome.

Immunocytochemistry (ICC) studies focused on the differ-entiation of human NSCs, the structural integration of NSCs,and the characterization of type and intensity of immune re-sponse to the graft. Many of these experiments required dual-label immunofluorescence. After permeabilization with 0.1%Triton X-100 and nonspecific site blocking with 5% normalserum from the same species as the secondary antibodies, sec-tions were incubated in primary antibodies in 1 mg/ml bovineserum albumin with 0.1% Triton X-100 (4°C, overnight). Pri-mary antibodies were used to address human (graft) versus

1978 Neural Stem Cells in ALS Mice

mouse (host) cell identity, neuronal, astrocytic, and oligoden-drocytic phenotype specification, and the type and intensity ofhost-versus-graft cellular response (supplemental online Table1). The presence of human cells in mouse tissues can be reliablytraced with antibodies against human nuclear antigen (HNu)[21]. Control sections were stained by replacing the primaryantibodies with pre-immune IgG from the same speciesof origin.

Antigen-antibody binding sites were revealed with Cy2- orCy3-conjugated secondary goat or donkey IgGs directed againstthe species of origin of the corresponding primary antibodies(1:200; Jackson ImmunoResearch Laboratories, Inc., WestGrove, PA, http://www.jacksonimmuno.com). In most cases,Cy3-conjugated goat or donkey anti-mouse IgG was used totrace the human cell marker HNu, and Cy2-conjugated goat ordonkey IgG was used for various cellular markers. Secondaryantibody incubations were performed for 2–4 hours at roomtemperature. All sections were counterstained with the DNA dye4,6-diamidino-2-phenylindole (DAPI) (blue) and then dehy-drated and coverslipped with DPX (a mixture of distyrene,tricresyl phosphate, and xylene). In the few cases in which ourexperimental design required triple labeling, the third secondaryantibody was coupled with the blue fluorescence dye AMCA(7-amino-4-methyl-coumarin-3-acetic acid; 1:200; Jackson Im-munoResearch Laboratories, Inc.) and DAPI counterstain wasomitted. Sections were studied with a Zeiss Axiophot micro-scope equipped for epifluorescence, and images were capturedwith a Spot RT Slider digital camera (Diagnostic Instruments,Inc., Sterling Heights, MI, http://www.diaginc.com). Confocalmicroscopic images were captured and optically resectioned inthe x- and y-axes using a Zeiss LSM 410 unit.

RESULTS

Combined Immunosuppressive Agents or CD4Antibodies Ameliorate Cell-Mediated Rejectionand Improve Human NSC Graft Survival inSOD1-G93A MiceGraft rejection was studied with ICC for HNu (to mark graft-derived cells) and protein epitopes marking blood-borne im-mune cells (i.e., lymphocytes and natural killer [NK] cells) orresident microglia/macrophages. Blood-borne cells were de-tected with antibodies against CD4 and CD8 surface antigens ofT cells (Figs. 1 and 2) and an antibody against a surface epitopepresent in NK cells (DX5; supplemental online Fig. 1) [22].Microglial cells were labeled with an antibody against the Iba-1epitope of a microglia-specific calcium-binding protein ex-pressed by these cells in all functional states, including resting,activated, and phagocytic microglia [23, 24] (Fig. 3). Cytolog-ical features and anatomical relationships with blood vesselsalso helped in the identification of immune cells. For example,blood-borne cells cluster primarily around blood vessels, al-though this pattern is less distinct with advanced rejection.Under all circumstances, blood-borne cells are rare in hosttissues surrounding the grafts.

CD4� and CD8� cell recruitment to grafting sites wassimilar in terms of cytology and perivascular clustering with alltreatments and at all time points, but the CD4 response waspredominant (Figs. 1 and 2). NK cells were not seen frequentlyin our preparations. A moderately intense NK cell response was

seen in subjects treated with FK506 alone, primarily in perivas-cular locations with little parenchymal invasion (supplementalonline Fig. 1A). The cytology of these DX-5� NK cells resem-bled that of CD4� and CD8� cells. In animals treated withcombined immunosuppressive regimens, DX-5 immunoreactiv-ity appeared primarily as debris without obvious cellular local-ization (supplemental online Fig. 1B, inset). Central nervoussystem (CNS) microglia was typically seen to invade the graft

Figure 1. Effects of various immunosuppressive regimens on graftsurvival versus CD8 cell infiltration. Spinal cord sections were takenthrough the graft site and dually labeled with antibodies against humannuclear antigen (HNu) (red) and CD8 (green). Images were capturedunder epifluorescence, except in top inset of (A), which is a confocalimage taken from the same section as the main panel. All insets, exceptthe top one in (A), represent magnifications of the framed areas in mainpanels. Arrows in insets point to representative CD8� T cells. Arrow-heads point to CD8� cellular debris. Cross-sections of some bloodvessels are labeled with asterisks. (A, B): Representative sections ofmice treated with FK506, surviving for 1 week (A) or 1 month (B) aftergrafting. Note many intact CD8� cells in (A) with their typical thincytoplasm. Many CD8� cells are in close proximity to diffuse HNuimmunoreactivity that gives the impression of scaffolding around greenCD8� profiles (arrows in insets in [B]). At 1 month, CD8 immunore-activity is localized as predominantly extracellular debris (arrowheads ininset in [B]). (C, D): A significant increase in graft survival (indicatedby the presence of a dense population of intact HNu � nuclei) and adecrease in CD8� cell infiltration with combined FK506 � rapamycintreatment at 1 week (C) and 1 month (D) after grafting. CD8 immuno-reactivity is present both in intact cells (arrows in insets) and asextracellular debris (arrowheads in insets). CD8� debris is much lessintense than in (B), presumably due to a lower frequency of CD8� Tcells recruited into the graft site. (E, F): Robust graft survival and minimalCD8� cell infiltration 1 month after treatment with combined FK506 �rapamycin � MMF (E) or CD4 antibodies (F). The results resemble thoseof FK506 � rapamycin treatment at 1 month (D). Scale bars � 100 �m(main panels), 20 �m (all insets except confocal in [A]), and 10 �m(confocal in [A] top inset). Abbreviations: Ab, antibody; FK, FK506;MMF, mycophenolate mofetil; mo, month; R, rapamycin; wk, week.

1979Yan, Xu, Welsh et al.

www.StemCells.com

from surrounding tissues replete with microglial cells at variousstages of transformation (Fig. 3). As we established [25], reac-tive microglia is featured by a scant cytoplasm and long, highlyramified processes, whereas phagocytic glia has retracted, thickprocesses and a more substantial cytoplasm volume.

With FK506 treatment alone, HNu staining fails to revealgraft-derived nuclei as early as 1 week after grafting (Figs. 1A,2A, 3A; supplemental online Fig. 1A). Grafting sites are fea-tured by a diffuse, non-nuclear HNu immunoreactivity oftenseen in close proximity to CD8� lymphocytes (Fig. 1A, insets)and, to a lesser degree, to NK cells (supplemental online Fig.1A, insets), but not CD4� cells (Fig. 2A, insets). HNu immu-noreactivity is often localized within the cytoplasm of micro-glial phagocytes (Fig. 3A, insets). One month after grafting,CD8 (Fig. 1B) and CD4 (data not shown) immunoreactivitiesremain strong in the graft region, but they are localized withinsmall round structures without cellular organization that prob-ably represent cellular debris. Very few live CD4 or CD8� cellsare visible at the graft sites (Fig. 1B), a pattern indicating thatactive T-cell rejection is over by that time. Iba-1� microglialcells with phagocytic cytology are numerous at the graft site 1week after grafting (Fig. 3A). At 1 month, microglial reactionremains strong, but many of these Iba-1� cells exhibit cytolo-gies consistent with activated microglia (Fig. 3B).

Combined treatment with FK506 and rapamycin has animpressive effect on graft survival, based on the presence ofnumerous dense, nuclear HNu� profiles in all our experimentalsubjects (Figs. 1C, 1D, 2B, 3C, 3D; supplemental online Fig.1B). CD4 and CD8� cell infiltration is markedly diminished;many such cells appear to have died based on the early presenceof CD4 or CD8� debris. Microglial response is moderate; theoverwhelming majority of Iba-1� cells have the appearance ofactivated, but not phagocytic, microglia (Fig. 3; compare insetsbetween panels C and A). A very low-level of CD4 or CD8�

response is present at 1 month after grafting (Fig. 1D).

Microglial reaction is still strong, and there is an apparentincrease in the presence of macrophages at 1 month (Fig. 3D)and well into end-stage disease (not shown). The addition ofMMF to the regimen of FK506 plus rapamycin had no apparenteffects in graft survival and infiltration by CD4/CD8� cells(Fig. 1E) or by microglia/macrophages (Fig. 3E) compared withthe double immunosuppressive regimen.

The use of CD4 antibodies also significantly promoted theviability of HNu� cells. Invasion of the graft by blood-bornelymphocytes (Fig. 1F) or microglial cells (Fig. 3F) was signif-icantly attenuated. As with combined immunosuppressive

Figure 2. Effects of FK506 (A) and combined FK506 � rapamycintreatment (B) on graft survival versus CD4 cell infiltration 1 week aftergrafting. Spinal cord sections were taken through the graft site anddually stained with antibodies against human nuclear antigen (HNu)(red) and CD4 (green). All images were captured under epifluorescence,and insets represent magnifications of framed areas in main panels,except the confocal image in top inset in (A). Note the sharp differencebetween non-nuclear HNu immunoreactivity in (A) and intense HNuimmunoreactivity of densely packed nuclei in (B). (A): Arrows in insetspoint to representative CD4� cells. In contrast to CD8� cells in Figure1A, most of these cells are not in close proximity to HNu-immunore-active material. Cross-section of a blood vessel is labeled with anasterisk. (B): Arrowheads in inset point to CD4� cellular debris, whichis the predominant CD4-immunoreactive structure in these preparations.Scale bars � 100 �m (main panels), 20 �m (all insets except confocalin [A]), and 10 �m (confocal in [A]). Abbreviations: FK, FK506;R, rapamycin; wk, week.

Figure 3. Effects of various immunosuppressive regimens on graftsurvival versus microglia/macrophage infiltration. Spinal cord sectionswere dually stained with antibodies against human nuclear antigen(HNu) (red) and Iba-1 (green). Images were captured under epifluores-cence, except in top inset of (A), which is a confocal image. All insetsexcept the top one in (A) represent magnifications of framed areas inmain panels. Arrows in insets point to representative microglial cells/macrophages. (A, B): Representative sections of mice treated withFK506 and surviving for 1 week (A) or 1 month (B) after grafting. MostIba-1� cells in (A) show pyramidal shapes with substantial cytoplasmand short processes (i.e., cytological features of macrophages) (arrowsin insets). Colocalization of HNu� material internal to the Iba-1� cellsurface is evident in the confocal inset (arrow, top inset in [A]). Iba-1�

cells in (B) show mixed cytologies with both extensively ramified cellsresembling activated microglia and some macrophage-like profiles. (C,D): A significant increase in graft survival (indicated by the presence ofa dense population of intact HNu � nuclei) with combined FK506 �rapamycin treatment at 1 week (C) or 1 month (D) after grafting. Iba1�

cells are far fewer than in (A) and are composed primarily of activatedmicroglia in (C), whereas in (D) they also include macrophage-likeprofiles. (E, F): Robust graft survival 1 month after treatment withcombined FK506 � rapamycin � MMF (E) or CD4 antibodies (F).There are several macrophage-like Iba-1� cells in (E). In (F), mostIba-1� cells have cytologies consistent with activated microglia. Scalebars � 100 �m (main panels), 20 �m (all insets except confocal in [A]),and 10 �m (confocal in [A]). Abbreviations: Ab, antibody; FK, FK506;MMF, mycophenolate mofetil; mo, month; R, rapamycin; wk, week.

1980 Neural Stem Cells in ALS Mice

treatments, the cytology of most Iba-1� cells resembled that ofactivated, but not phagocytic, microglia (Fig. 3F, inset). Inconcert, the combination of FK506 and rapamycin with orwithout MMF or the use of CD4 antibodies suppressed CD4-and CD8-cell infiltration and also significantly attenuated anddelayed microglia-mediated phagocytosis.

Combined Immunosuppressive Agents and CD4Antibodies Improve Disease Outcomes inSOD1-G93A Mice Grafted with Human NSCsWhen treated with combined immunosuppressive drugs or anti-CD4 antibodies, SOD1-G93A mice grafted with live NSCsshowed delayed disease onset, improved motor scores, andlonger life spans compared with FK506 monotherapy (Fig. 4).Mice treated with FK506 alone had an average disease onset of13.2 � 1.8 weeks. Treatments with the GK1.5 antibody, FK506� rapamycin, and less so with FK506 � rapamycin � MMF alldelayed time to disease onset (15.5 � 1.3, 15.3 � 1.0, and14.5 � 0.9 weeks, respectively) compared with the FK506group (Fig. 4A). Post hoc testing reveals that individual com-parisons between the anti-CD4 or FK506 � rapamycin groupsand FK506 group are significant, but the difference betweenFK506 � rapamycin � MMF and FK506 groups is not. Treat-ment with anti-CD4 antibody, FK506 � rapamycin, and FK506

� rapamycin � MMF significantly extended the life span ofNSC-grafted animals compared with FK506 treatment alone(20.3 � 1.5, 19.7 � 1.9, and 20.4 � 1.7 weeks as comparedwith 17.6 � 2.0 weeks) (Fig. 4B). Motor score testing alsoshowed a delayed loss of muscle strength in the anti-CD4,FK506 � rapamycin, and FK506 � rapamycin � MMF groupscompared with FK506 treatment group (Fig. 4C). Log-ranktesting of Kaplan-Meier survival curves also revealed signifi-cant differences among anti-CD4, FK506 � rapamycin, FK506� rapamycin � MMF, and FK506 groups. The use of Log-ranktesting within pairs of treatment groups revealed significantdifferences between FK506 monotherapy and any of the othertreatment groups (Fig. 4D). Based on the above data, the com-bined FK506 � rapamycin regimen appears to optimize preven-tion of graft rejection, at least for the purpose of generatingclinically meaningful differences in the time framework of mo-tor neuron disease in SOD1-G93A mice.

To control for the potential effects of immunosuppressantson clinical parameters of motor neuron disease, we compareddisease onset, motor score, and life span data between animalsgrafted with live-versus-dead human NSCs, all of which weretreated with FK506 � rapamycin. In animals grafted with livecells, disease onset was delayed by 2.1 weeks (Fig. 5A), and lifespan was extended for 1.7 weeks (Fig. 5B) compared with

Figure 4. Combined immunosuppressive drugs or CD4 antibodies delay disease onset, improve motor scores, and extend life span of SOD1-G93Amice grafted with human neural stem cells and treated with combined immunosuppressants as compared with FK506 monotherapy. (A, B): These twographs indicate that combined immunosuppressive treatments or CD4 antibodies delay disease onset (A) and extend the life span (B) of SOD1 mice.Variance in both measures is significant (analysis of variance [ANOVA]: p � .0029 and p � .068, respectively). Fischer least significant difference(LSD) post hoc testing shows that the significance originates in differences between the combined treatment groups or the CD4 antibody group withthe FK506 monotherapy group. Note that the FK506 � rapamycin � MMF group is not significantly different from the FK506 group with respectto disease onset. (C, D): Variance in the progression of muscle weakness (C) and in survival (D) among treatment groups. Muscle strength was scoredwith open-field testing as explained in Materials and Methods. Repeated-measures ANOVA followed by Fisher LSD post hoc testing of individualdifferences in (C) reveals significant differences comparing FK506 � rapamycin or FK506 � rapamycin � MMF or anti-CD4 groups to FK506 group(p � .007, .008, and .023, respectively). Log-rank testing of Kaplan-Meier survival curves in (D) shows an overall significant variance (p � .011,�2 � 11.11). Using Log-rank testing for comparisons between pairs of groups reveals significant differences between anti-CD4 or FK506 �rapamycin or FK506 � rapamycin � MMF and FK506 monotherapy (p � .023, .016, and .001, respectively). Group data are displayed as mean �SD. (��, p � .01 based on post hoc testing). Abbreviations: FK, FK506; MMF, mycophenolate mofetil; Rapa, rapamycin.

1981Yan, Xu, Welsh et al.

www.StemCells.com

animals that had received dead cells (disease onset: 15.3 � 1 vs.13.1 � 1.9 weeks; life span: 19.7 � 1.9 vs. 18.0 � 1.8 weeks).Motor score testing shows a later onset and slower progressionof weakness in animals treated with live NSCs (p � .02) (Fig.5C), and Kaplan-Meier survival analysis shows a significantdifference between live- and dead-cell groups (p � .001) (Fig.5D). These differences indicate that immunosuppressants influ-ence key clinical parameters of motor neuron disease in SOD1-G93A mice via their protection of NSC survival and not via adirect protective effect on host motor neurons.

Grafted Human NSCs Survive to End-Stage Disease,Differentiate Predominantly into Neurons, andEstablish Synaptic Contacts with Host NeuronsIn end-stage animals (more than 2 months after grafting) treatedwith combinations of immunosuppressive agents or CD4 anti-bodies, the vast majority of human NSCs had differentiated intoTUJ1� neurons (Fig. 6A). Confocal microscopy confirmed thecolocalization of HNu (red) with TUJ1 (green) in the same cells(Fig. 6A�). Although numerous glial fibrillary acidic protein(GFAP)� astrocytic processes were present in the graft site (Fig.6B), the colocalization of GFAP immunoreactivity with HNu�

nuclei within the same cells was a rare phenomenon in themajority of grafts located within the spinal cord parenchyma(Fig. 6B). This profile suggests that most GFAP� processes atthe graft site are of host origin. However, in portions of grafts

located very close to meninges (pia), the frequency of HNu�

cells colocalizing GFAP increased, including the occasionalclustering of double-labeled cells (Fig. 6B�); this pattern showsa degree of plasticity in NSC differentiation determined by localfactors in the spinal cord microenvironment. HNu� nuclearprofiles did not colocalize with O4 or Rip immunoreactivitywithin the same cells, a pattern suggesting that human NSCs didnot differentiate in the oligodendrocyte or Schwann cell lineage

Figure 5. Differences in clinical parameters of motor neuron diseasebetween animals grafted with live or dead human neural stem cells(NSCs), all of which were optimally immunosuppressed with FK506plus rapamycin. (A, B): Live NSC grafts delay disease onset (A) andextend the life span (B) of SOD1 mice compared with dead-cell grafts.On both measures, differences between the two groups are significant(Student’s t test: p � .0025 and p � .012, respectively). (C, D):Differences in the progression of motor weakness (C) and in survival(D) between live- and dead-cell grafted groups. Repeated-measuresanalysis of variance followed by Fisher least significance difference posthoc testing of variance in (C) shows significant differences betweenlive- and dead-cell grafted groups (p � .02). Log-rank testing of Kaplan-Meier survival curves (D) also shows significant differences betweenthe two groups (p � .001, �2 � 11.11). Group data are displayed asmean � SD. (�, p � .05; ��, p � .01). Abbreviations: FK, FK506;Rapa, rapamycin.

Figure 6. Differentiation of human neural stem cells (NSCs) intoneurons and astrocytes in vivo. (A-A�): These two sections were duallystained for HNu (red) and TUJ1 (green) and illustrate the very highfrequency of neurons derived from human NSCs using epifluorescence(A) and confocal microscopy (A�). (A) is a low-power image illustratingthe marked enrichment of TUJ1 immunoreactivity in the graft comparedwith the surrounding host tissue. Confocal images in (A�) illustrate thetypical filamentous cytoplasmic TUJ1 immunoreactivity; image in mainframe was optically resectioned in the x- and y-planes to confirm theintimate apposition of the two immunoreactivities within these denselyclustered neuronal cell bodies. (B-B�): These sections were stained forHNu (red) and GFAP (green) and showcase the sparse astrocytic dif-ferentiation of human NSCs by epifluorescence (B) and confocal mi-croscopy (B�). Note the presence of rare GFAP� cell bodies withenclosed HNu� nuclei in the graft (arrow in [B]) despite the presence ofnumerous GFAP� processes. Confocal microscopy (B�) shows a smallcluster of human NSC-derived GFAP� cells located close to the pialsurface. Confocal sections have been processed as in (A�). (C, C�):These sections were stained for HNu (red) and nestin (green) andillustrate that some human NSCs retain stem cell properties, at least asindicated by their expression of high levels of nestin immunoreactivityby epifluorescence (C) and confocal microscopy (C�). Confocal imageshave been processed as in (A�). Arrows in (C) point to selected nestin�

cells. Scale bars � 50 �m (A), 20 �m (B, C), and 10 �m (A�, B�, C�).Abbreviations: GFAP, glial fibrillary acidic protein; HNu, humannuclear antigen.

1982 Neural Stem Cells in ALS Mice

in our experimental settings (data not shown). A small numberof human NSCs remained undifferentiated as nestin� cells (Fig.6C, 6C�).

When synaptic terminal markers specifying graft originwere combined with generic neuronal markers, large numbers ofhuman synaptophysin� boutons (representing terminals ofgraft-derived neurons) were found apposed to host neurons,especially surviving �-motor neurons (Fig. 7A). Confocal anal-ysis of specimens from 1 month after grafting and end-stageanimals shows anatomical patterns typical of synaptic contacts(Fig. 7A�). Conversely, HNu and TUJ1� cells (i.e., graft-de-rived neurons) were closely apposed to VGLUT1/2� terminalsof mouse origin (Fig. 7B); the synaptic significance of suchappositions was difficult to determine by confocal microscopybecause of the high cell density of grafts (Fig. 7B�). Together,these patterns of reciprocal innervation between host and graft-derived neurons indicate a substantial degree of structural

integration of NSC grafts within the host circuitry. This phe-nomenon is likely to account for the beneficial effects of NSCson clinical outcomes laid out in the previous section.

DISCUSSIONThe goal of this study was to establish effective immunosup-pressive regimens in order to prevent the rejection of humanNSCs grafted in SOD1-G93A mice. Our findings indicate thatcombinations of immunosuppressive drugs have significantlybetter outcomes compared with FK506 monotherapy in prevent-ing graft rejection and, apparently via their promotion of graftsurvival, improving key parameters of motor neuron disease inSOD1-G93A mice. Similar results were obtained with CD4antibodies. The effective suppression of NSC graft rejectionallowed sufficient time for the differentiation of grafted cellsinto neurons and the establishment of networks linking host andgraft neurons. Based on our data, combined immunosuppressiveregimens or CD4 antibodies appeared to protect NSC grafts bysuppressing CD4- and CD8-cell recruitment into the graft areaand attenuating the microglial phagocytic response from sur-rounding spinal cord tissues. NK cells were rarely seen at thegraft sites, and this is consistent with previous findings pointingto the low significance of these cells in xenograft rejection in thebrain [26, 27]. Within the clinical framework of our experimen-tal design, the FK506 � rapamycin combination was optimaldue to its efficacy and relative simplicity.

Xenograft rejection is a serious problem when studying cellgrafts of human or other highly discordant mammal origin inrodent models. For example, human spinal cord-derived progen-itors are rejected 4 weeks after grafting into the adult rat spinalcord despite the use of cyclosporine A [14]. In our hands,FK506, which is 10–100 times more potent than cyclosporine Ain preventing graft rejection [28], has shown good results in ratsgrafted with human NSCs [16] but has fallen short of preventingrejection of the same cells in SOD1-G93A mice as described inthe present study. The degree of species disparity between donorand host appears to be a key factor, and, in this respect, differ-ences between mouse and rat may be significant [27, 29].

Host rejection of xenografts is a complex response that isnot fully understood. Although antibody-mediated mechanismsand participation of the complement have been implicated [30,31], most evidence points to a central role for T-lymphocytesand microglia-derived macrophages [32]. The indefinite sur-vival of human cell grafts in the nervous system of nude athymicrats observed by others [33] and by us [34] strongly supports theidea for a central role of T cells in neural xenograft rejection. Inresponse to foreign antigens presented by the xenograft, CD4 Tcells turn into helper cells that subsequently activate CD8 Tcells into cytotoxic roles, including the lysis of grafted cellsexpressing major histocompatibility complex class-1 epitopeson their membranes. CD4 cells also signal the transformation ofCNS microglia into macrophages. Although CD8 cells are theeffectors that execute the lysis of grafted cells and abound at thegraft site, CD4 lymphocytes are the initiators of obligatory earlysignaling; for example, CD4 antibodies prevent the rejection ofCNS grafts, whereas CD8 antibodies have a very weak effect[12]. Based on our findings, CD4 antibodies are a viable alter-native to immunosuppressive drugs to prevent rejection of hu-man xenografts. However, the efficacy of antibodies is compa-rable with that of combined immunosuppressants, and the cost

Figure 7. Reciprocal innervation between graft-derived and host neu-rons based on triple immunocytochemistry for human Syn, two VGLUTepitopes (1 and 2), and TUJ1. Human synaptophysin immunoreactivityis used as a selective marker for graft-derived synapses. VGLUT1/2 isan excitatory synaptic marker specifying host origin because of theabsence of glutamatergic phenotypes in differentiated neural stem cells(NSCs). TUJ1 is a generic neuronal marker. Color representations ofvarious immunoreactivities are specified on top of panels. (A, A�):These epifluorescent (A) and confocal (A�) images taken through theventral horn of a SOD1-G93A mouse 1 month after grafting show aTUJ1� (green) host �-motor neuron (outlined by the three arrows in[A]) contacted, at both the cell bodies and dendrites, by many humansynaptophysin� terminals deriving from differentiated NSCs (red).DAPI (blue) is used as nuclear counterstain. (B, B�): These epifluores-cent (B) and confocal (B�) photographs are taken through the graft siteand show graft-derived neurons, identified by HNu (red) nuclei and cellbody staining of TUJ1 (blue) in proximity to VGLUT1/2� (green)terminals from host neurons (some are indicated with arrows in [B]).Confocal images have been optically resectioned in the x- and y-planesas explained in Figure 6A�. Scale bars � 20 �m (A, B), 10 �m (A�), and5 �m (B�). Abbreviations: DAPI, 4,6-diamidino-2-phenylindole; HNu,human nuclear antigen; Syn, synaptophysin.

1983Yan, Xu, Welsh et al.

www.StemCells.com

of producing large antibody quantities by inducing ascites innude mice is a significant disadvantage for their routine use inlarge-scale experiments.

The use of combinations of immunosuppressants exploitssome differences in their mechanisms of action without expos-ing animals to the side effects of extremely high doses of singlecompounds [28]. FK506 is a calcineurin-dependent inhibitorthat blocks the production of interleukin 2 (IL-2) and thus theproliferation of T-lymphocytes. It may also have an additionalrole via blocking the glucocorticoid receptor. Rapamycin hasinhibitory effects on IL-2 signaling downstream and indepen-dent of the calcineurin pathway and inhibits the progression ofT cells from the G to S-phase. Thus, FK506 and rapamycininfluence different steps in the IL-2 pathway. MMF reduces theproliferation of T cells by blocking purine synthesis via inhibi-tion of type II isomer of inosine monophosphate dehydrogenase.The principle of mixing these immunosuppressant compoundsin various combinations is routinely used to prevent graft rejec-tion in clinical settings [35]. However, the optimization ofimmunosuppressive treatments to prevent rejection of discor-dant neural xenografts in experimental animals has not beensystematically studied. A potential concern is whether theseimmunosuppressive compounds can cross the blood-brain bar-rier (BBB). FK506 and rapamycin can readily cross the BBB,but there is no published work, nor does the manufacturer haveany data, on the permeability of the BBB to MMF. However,CNS bioavailability is unlikely to influence the efficacy of thesedrugs whose primary mode of action is suppression of T-cellproliferation in sites outside the CNS.

Besides their prevention of xenograft rejection, immunosup-pressive compounds may also exert neuroprotective effects un-der certain conditions [36, 37]. Cyclosporine A, FK506, andrapamycin have all shown variable efficacy in models of trau-matic injury, anoxia-ischemia, and neurodegeneration. With re-spect to SOD1-G93A mice, only cyclosporine A has shownsome efficacy [38], whereas FK506 made no difference [39]. As

far as we know, rapamycin and MMF have not yet been tested.In our hands, cyclosporine A was ineffective in altering thecourse of motor neuron disease in SOD1-G93A mice. In addi-tion, when FK506 and rapamycin were given in combination tomice grafted with dead human NSCs, they had no obviouseffects in clinical outcomes. Therefore, in the context of thepresent grafting experiment, the therapeutic contributions of theimmunosuppressive compounds themselves to Tg motor neurondisease were negligible.

CONCLUSIONTreatments with immunosuppressant compounds or CD4 anti-bodies can be optimized in relatively simple regimens to allowfor long-term survival and differentiation of human NSCs in thespinal cord of SOD1-G93A mice with motor neuron disease.The combination of FK506 with rapamycin is optimal withrespect to efficacy and simplicity of administration. By protect-ing the neuronal differentiation and networking of human cellsin the host spinal cord, immunosuppressive regimens allow thedetection of significant clinical effects of human NSCs whichmerit further characterization in future studies.

ACKNOWLEDGMENTSWe thank Dr. William Baldwin, Department of Pathology,Johns Hopkins Medical Institutions, for advice with the immu-nological aspects of the paper and gift of GK1.5 hybridomacells. This work was supported by grants from the MuscularDystrophy Association (MDA3493), the U.S. Public HealthService (NS45140-03), and the Robert Packard Center for ALSResearch at Johns Hopkins. J.Y. and L.X. contributed equally tothis work.

DISCLOSUREST.H. and K.J. own stock in, have acted as consultants for, haveperformed contract work for, have a financial interest in, andhave served either as an officer or member of the Board ofNeuralStem, Inc.

REFERENCES

1 Bruijn LI, Miller TM, Cleveland DW. Unraveling the mechanisms in-volved in motor neuron degeneration in ALS. Annu Rev Neurosci2004;27:723–749.

2 Price DL, Wong PL, Koliatsos VE et al. Motor neuron disease. In: SiegelGJ, Albers RW, Brady ST et al., eds. Basic Neurochemistry. New York:Elsevier Science, 2005.

3 Gurney ME, Pu H, Chiu AY et al. Motor neuron degeneration in micethat express a human Cu,Zn superoxide dismutase mutation. Science1994;264:1772–1775.

4 Wong PC, Pardo CA, Borchelt DR et al. An adverse property of afamilial ALS-linked SOD1 mutation causes motor neuron disease char-acterized by vacuolar degeneration of mitochondria. Neuron 1995;14:1105–1116.

5 Cleveland DW, Rothstein JD. From Charcot to Lou Gehrig: Deci-phering selective motor neuron death in ALS. Nat Rev Neurosci2001;2:806 – 819.

6 Mallet J, Bjorklund A, Caskey CT et al. Group report: Neuronal replace-ment and functional modification. In: Price DL, Thoenen H, Aguayo AJ,eds. Neurodegenerative Disorders: Mechanisms and Prospects for Ther-apy. New York: John Wiley & Sons, 1991:271–290.

7 Tai YT, Svendsen CN. Stem cells as a potential treatment of neurologicaldisorders. Curr Opin Pharmacol 2004;4:98–104.

8 Silani V, Cova L, Corbo M et al. Stem-cell therapy for amyotrophiclateral sclerosis. Lancet 2004;364:200–202.

9 Ostenfeld T, Caldwell MA, Prowse KR et al. Human neural precursorcells express low levels of telomerase in vitro and show diminishing cellproliferation with extensive axonal outgrowth following transplantation.Exp Neurol 2000;164:215–226.

10 Englund U, Bjorklund A, Wictorin K. Migration patterns and phenotypicdifferentiation of long-term expanded human neural progenitor cells aftertransplantation into the adult rat brain. Brain Res Dev Brain Res 2002;134:123–141.

11 Hurelbrink CB, Barker RA. Migration of cells from primary transplantsof allo- and xenografted foetal striatal tissue in the adult rat brain. EurJ Neurosci 2005;21:1503–1510.

12 Wood MJ, Sloan DJ, Wood KJ et al. Indefinite survival of neuralxenografts induced with anti-CD4 monoclonal antibodies. Neuroscience1996;70:775–789.

13 Klein SM, Behrstock S, McHugh J et al. GDNF delivery using human neuralprogenitor cells in a rat model of ALS. Hum Gene Ther 2005;16:509–521.

14 Li P, Tessler A, Han SS et al. Fate of immortalized human neuronalprogenitor cells transplanted in rat spinal cord. Arch Neurol 2005;62:223–229.

15 Yan J, Xu L, Welsh AM et al. Grafting of human neural stem cells (hNSCs)into the spinal cord of a mouse model of ALS (SOD1–G93A) delays thesymptoms and extends life span [abstract]. Soc Neurosci Abst 2005;131:3.

1984 Neural Stem Cells in ALS Mice

16 Xu L, Yan J, Chen D et al. Human neural stem cell (NSC) grafts in thespinal cord of SOD1- G93A transgenic rats delay motor neuron degen-eration and prolong life span [abstract]. Soc Neurosci Abst 2005;213:3.

17 Johe KK, Hazel TG, Muller T et al. Single factors direct the differenti-ation of stem cells from the fetal and adult central nervous system. GenesDev 1996;10:3129–3140.

18 Weydt P, Hong SY, Kliot M et al. Assessing disease onset andprogression in the SOD1 mouse model of ALS. Neuroreport 2003;14:1051–1054.

19 Basso DM, Beattie MS, Bresnahan JC. A sensitive and reliablelocomotor rating scale for open field testing in rats. J Neurotrauma1995;12:1–21.

20 Wrathall JR, Pettegrew RK, Harvey F. Spinal cord contusion in the rat:Production of graded, reproducible, injury groups. Exp Neurol 1985;88:108–122.

21 Vescovi AL, Parati EA, Gritti A et al. Isolation and cloning of multipo-tential stem cells from the embryonic human CNS and establishment oftransplantable human neural stem cell lines by epigenetic stimulation.Exp Neurol 1999;156:71–83.

22 Ortaldo JR, Winkler-Pickett R, Mason AT et al. The Ly-49 family:Regulation of cytotoxicity and cytokine production in murine CD3�

cells. J Immunol 1998;160:1158–1165.

23 Imai Y, Ibata I, Ito D et al. A novel gene iba1 in the major histocom-patibility complex class III region encoding an EF hand protein ex-pressed in a monocytic lineage. Biochem Biophys Res Commun 1996;224:855–862.

24 Ito D, Imai Y, Ohsawa K et al. Microglia-specific localisation of a novelcalcium binding protein, Iba1. Brain Res Mol Brain Res 1998;57:1–9.

25 Koliatsos VE, Price WL, Pardo CA et al. Ventral root avulsion: Anexperimental model of death of adult motor neurons. J Comp Neurol1994;342:35–44.

26 Brevig T, Holgersson J, Widner H. Xenotransplantation for CNS repair:Immunological barriers and strategies to overcome them. Trends Neu-rosci 2000;23:337–344.

27 Larsson LC, Anderson P, Widner H et al. Enhanced survival of porcineneural xenografts in mice lacking CD1d1, but no effect of NK1.1depletion. Cell Transplant 2001;10:295–304.

28 Masri MA. The mosaic of immunosuppressive drugs. Mol Immunol2003;39:1073–1077.

29 Larsson LC, Frielingsdorf H, Mirza B et al. Porcine neural xenografts inrats and mice: Donor tissue development and characteristics of rejection.Exp Neurol 2001;172:100–114.

30 Larsson LC, Czech KA, Widner H et al. Discordant neural tissue xeno-grafts survive longer in immunoglobulin deficient mice. Transplantation1999;68:1153–1160.

31 Barker RA, Ratcliffe E, McLaughlin M et al. A role for complement inthe rejection of porcine ventral mesencephalic xenografts in a rat modelof Parkinson’s disease. J Neurosci 2000;20:3415–3424.

32 Barker RA, Widner H. Immune problems in central nervous system celltherapy. NeuroRx 2004;1:472–481.

33 Hurelbrink CB, Armstrong RJ, Dunnett SB et al. Neural cells fromprimary human striatal xenografts migrate extensively in the adult ratCNS. Eur J Neurosci 2002;15:1255–1266.

34 Yan J, Welsh AM, Hazel T et al. Grafted human neural stem (NS) cellsdifferentiate into neurons, migrate long distance and project axons inspinal cord and the roots of adult rats. [abstract]. Soc Neurosci Abst2003;29(150.19).

35 Qi S, Xu D, Peng J et al. Effect of tacrolimus (FK506) and sirolimus(rapamycin) mono- and combination therapy in prolongation of renalallograft survival in the monkey. Transplantation 2000;69:1275–1283.

36 Kaminska B, Gaweda-Walerych K, Zawadzka M. Molecular mecha-nisms of neuroprotective action of immunosuppressants-facts and hy-potheses. J Cell Mol Med 2004;8:45–58.

37 Klettner A, Herdegen T. FK506 and its analogs—therapeutic potentialfor neurological disorders. Curr Drug Targets CNS Neurol Disord 2003;2:153–162.

38 Karlsson J, Fong KS, Hansson MJ et al. Life span extension and reducedneuronal death after weekly intraventricular cyclosporin injections in theG93A transgenic mouse model of amyotrophic lateral sclerosis. J Neu-rosurg 2004;101:128–137.

39 Anneser JM, Gmerek A, Gerkrath J et al. Immunosuppressant FK506does not exert beneficial effects in symptomatic G93A superoxide dis-mutase-1 transgenic mice. Neuroreport 2001;12:2663–2665.

See www.StemCells.com for supplemental material available online.

1985Yan, Xu, Welsh et al.

www.StemCells.com