Proc. Natl. Acad. Sci. USAVol. 84, pp. 5957-5961, August 1987Neurobiology

Human fetal cerebellar and cortical tissue transplanted to theanterior eye chamber of athymic rats: Electrophysiologicaland structural studies

(human neuroblast development/xenograft/cerebellum/cortex cerebri/intraocular transplantation)

P. BICKFORD-WIMER*t, A-CH. GRANHOLMt, MARK BYGDEMANt, B. HOFFERt, L. OLSON§, A. SEIGER§,AND I. STROMBERG§*Veterans Administration Medical Center, and tDepartment of Pharmacology, University of Colorado Health Sciences Center, Denver, CO 80262;§Department of Histology, Karolinska Institute, Stockholm, Sweden; and tDepartment of Obstetrics and Gynecology, Karolinska Hospital, Stockholm,Sweden

Communicated by David W. Talmage, April 3, 1987 (received for review February 2, 1987)

ABSTRACT Human fetal tissue fragments from cortexcerebri and cerebellum were grafted to the anterior chamber ofthe eye of adult athymic nude rats. The grafts were obtainedfrom tissue fragments recovered after elective routine abor-tions, performed in weeks 8-11 of gestation. Both cerebellarand cortex cerebri grafts survived and developed in the anteriorchamber of the eye for 14 months. The transplants slowlybecame vascularized from the host iris. The grafts developedblood vessels with laminin-immunoreactive walls and con-tained relatively high amounts of glial fibrillary acidic protein-and neurofilament-immunoreactivity in the neuropil after 4months in oculo. Recordings of extracellular action potentialsfrom the grafts revealed spontaneously active neurons withaction-potential waveforms similar to those observed in imma-ture rodents. Morphologically, the grafts showed no signs ofrejection. Clusters and bands of large neurons resemblingPurkinje cells and dense aggregates of smaller granule-like cellscould be found in the cerebellar grafts. Large neurons were alsoseen in the cortex grafts. Taken together, these data suggestthat the athymic rat may serve as a useful tool for studies ofcentral nervous system tissue from otherwise immunologicallyincompatible species.

Syngeneic grafting of brain tissue has emerged over the lastdecade as a valuable approach to studying the developmentand regeneration of neural connections in the central nervoussystem of mammals (1-3). Recently, interest has focused onthe developmental properties of xenogeneic grafts of centralnervous system tissue (4-6).A major problem with xenogeneic brain tissue grafting has

been poor or variable survivability. This problem has beenovercome in part by daily treatment with immunosuppressiveagents (7-11). The most widely used immunosuppressiveagent, cyclosporin A, is thought to suppress both humoraland cell-mediated immunity (12). However, the preciseaction of cyclosporin A and other immunosuppressive agentsis unknown, as is the extent to which such agents mightinterfere with graft survival and development. In this per-spective, the use of a host/graft system that does not requireany immunosuppressive agents would be a better tool forstudies of development and regeneration of xenogeneicgrafts.The athymic rat has been shown to lack T-cell function (13,

14). Because of its inability to react immunologically toforeign tissue, it has been used extensively as a host tosuccessfully support transplants of various malignant humantumor cell lines without the need for immunosuppressive

therapy (15, 16). Therefore, such a host may be the idealrecipient for xenogeneic brain grafts.The anterior chamber of the eye has been used extensively

for studies of syngeneic brain transplants (see refs. 2 and 3).This technique offers unique advantages over other trans-plantation sites because the survival and growth can bemonitored without invasive procedures (2, 3). Here we reportstudies of human fetal cerebellar and cortical brain tissuegrafted to the anterior chamber of the eye of athymic nuderats. Survival and growth of these grafts was examined byelectrophysiological and morphological techniques.

MATERIALS AND METHODSFetal material to be grafted was obtained after termination offirst-trimester pregnancies. Healthy women with an appar-ently normal pregnancy in weeks 8-11 of gestation andadmitted to the hospital for elective abortion were informedboth orally and in writing about the aim of the study and theprocedure to be used, and they gave their consent. Anonym-ity was strictly maintained. The abortion was performed byusing paracervical blockade following premedication. Afterdilatation of the cervical canal, fetal fragments were removedby forceps after which the abortion was completed byvacuum aspiration. The fetal tissue fragments were collectedand kept in isotonic saline until further processed. The studywas approved by the Regional Ethical Committee of theKarolinska Hospital, and all experiments conformed toguidelines of the Swedish Medical Research Council and theU.S. Public Health Service.Tissues were examined using a stereomicroscope, and

small pieces measuring 1-3 mm3 were prepared for grafting.The medial portion of the cerebellar anlage and small piecesof cerebral cortex were dissected free from pial membraneand inserted into the anterior chamber of the eye of 2-month-old nude rats by methods described earlier for syngeneicrat/rat transplants (see ref. 2). Vascularization and growth ofthe grafts was followed by repeated measurements throughthe translucent cornea.

Spontaneous activity and responses to electrical surfacestimulation were measured by extracellular recordings withsingle-barrel micropipettes. Electrophysiological recordingswere performed on two cortex cerebri grafts and fourcerebellar grafts that were grown for 6-8 weeks in oculo andfrom two cerebellar grafts after 4 months in oculo. Thir-ty-eight neurons from the six cerebellar grafts and 12 neuronsfrom the two cortex grafts were suitable for analysis. Thehost animals were anesthetized (1.25 g of urethane i.p. per kg

Abbreviations: GFA, glial fibrillary acidic protein; NF, neurofila-ment.

5957

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

5958 Neurobiology: Bickford-Wimer et al.

of body weight), and the cornea overlying the graft wasremoved. A Plexiglas perfusion chamber was placed over theeye, and the graft was perfused with Earle's balanced saltsolution at 370C throughout the experiment. Single-unitactivity was recorded as described (17). For each cell, 10-20spikes were averaged by digital computer, and 95% confi-dence limits and mean waveforms were displayed. All elec-trophysiological experiments were initiated in a darkenedroom after a sufficient time had elapsed for recovery fromretinal bleaching during surgery. When recording from cortexcerebri grafts, electrical activity was augmented by additionof 50 mM sodium glutamate to the electrolyte solution in thepipette or 3000 units of sodium penicillin per ml to theperfusion fluid. Electrical stimulation of parallel fibers incerebellar grafts was performed with a bipolar electrode oftwo twisted wires having a tip separation of 0.1 mm placed onthe surface of the graft. Monophasic 0.5- to 1.0-msec square-wave pulses of 1-60 V were utilized.

Grafts were processed for immunohistochemical localiza-tion of glial fibrillary acidic protein (GFA), neurofilament(NF), and laminin by using the indirect immunofluorescencetechnique of Coons (18) as described (11).

RESULTS

Cerebellar and cortical grafts survived well in the anteriorchamber of the eye and became vascularized from the hostiris. Vascularization occurred over the first weeks. Fetalcerebral cortex began to increase in size a few days aftergrafting and grew rapidly between weeks 2 and 6 aftergrafting. Cerebellar grafts initially decreased in size beforeresuming growth after 1 week in oculo. They then grewprogressively but at a slower rate than did cortex cerebri,reaching a size about one-third that of the cortex cerebrigrafts at 6 weeks.

Cresyl violet-stained sections through cerebellar grafts,sampled at 1.5-2 and 4 months after transplantation, revealedimmature brain tissue. Clusters and bands of larger neurons,presumably Purkinje neurons, were often seen. Sometimessmaller and more densely arranged neurons, presumablygranule cells, formed an internal granular layer on one side ofthe Purkinje neurons, with a cell-poor area, probably corre-sponding to the molecular layer, on the other side (Fig. 1 aand b). Dense clusters of small neurons, presumably corre-sponding to remaining patches of the external granular layer,were also found. In addition, pigmented melanocytes mi-grated into graft neuropil.

Laminin immunohistochemistry paralleled the in vivo ob-servation of a relatively slow vascularization of the cerebellargrafts. Thus, at 6 weeks in oculo, the cerebellar graft shownin Fig. id was still not vascularized despite good growth.Eventually all grafts became vascularized, although lamininimmunohistochemistry revealed abnormally thick walls (Fig.le). Cerebellar transplants at 4 months after transplantationcontained a dense plexus of NF-immunoreactive material.This consisted mainly of nerve-fiber bundles, but also in-cluded groups of small- or medium-sized cell bodies (Fig. 1c).However, the larger neurons, presumably corresponding toPurkinje cells, were devoid of NF immunofluorescence.GFA-like immunoreactivity was found in all transplants.

The density of GFA-immunoreactive glial processes wassomewhat higher than in normal adult cerebellar tissue. Inpatches of residual external granular-layer cells, a less denseplexus of GFA-immunoreactive material was seen (Fig. 2).Many areas of the cerebellar transplants contained long,straight GFA-positive fibers running in several differentdirections, suggesting the presence of disorganized Berg-mann glia fibers (Fig. 2b). The amount of GFA in thecerebellar transplants increased with time during the periodstudied (Fig. 2 c and d).

Grafts of cortex cerebri contained large scattered neuronswith one or two processes as well as groups of smaller cells.There were no signs of degenerative processes or rejection ineither the cerebellar or the cortical grafts examined.

Action potentials were recorded from 28 neurons from fourcerebellar grafts at 6-8 weeks in oculo. Initially negative orpositive waveforms were observed with an average durationof 1.9 ± 0.1 msec. The neurons tended to discharge indoublets. The discharge rates were slow, and the pattern ofdischarge was irregular. The interspike-interval histogramsusually consisted of two modal peaks; one at 5-msec, repre-sentative of intradoublet intervals, and a later peak at 150msec that reflected longer pauses between doublets. Atypical cerebellar Purkinje neuron is illustrated in Fig. 3A.Electrical stimulation of the granule-cell parallel fibers withan electrode placed on the surface of the graft was unable toelicit Purkinje-cell discharge in three neurons tested in thesegrafts.'A total of 12 cells were recorded from the two cortex

cerebri grafts. Fig. 3B illustrates the discharge of one of theseneurons. The pattern of firing was intermittent, and neuronstended to discharge when initially recorded and then to stop.Perfusion of penicillin (3000 units/ml) increased dischargerates; under these conditions, cell discharge could be fol-lowed for a period of several minutes. The majority ofaction-potential waveforms were initially negative, but someneurons displayed an initial positivity. The average action-potential duration was 2.5 ± 0.2 msec.Two cerebellar grafts were allowed to develop for 4 months

in oculo. Ten cerebellar neurons were recorded from thesegrafts. These neurons demonstrated firing patterns that weremore mature than those recorded from the four cerebellargrafts grown in oculo for 2 months. Action-potential dura-tions were significantly shorter (1.14 ± 0.1 msec; P < 0.01,two-tailed Student's t test). Sustained discharge was fre-quently seen, and spontaneous firing rates were faster thanthose observed in the younger transplants. Electrical stimu-lation of the cerebellar cortical surface activated parallel-fiber afferents to the Purkinje cells. Surface stimulation wascapable of eliciting evoked discharge in one of three neuronstested (Fig. 4), demonstrating that intrinsic excitatory cir-cuitry was developing in these cerebellar grafts.

DISCUSSIONThe present results demonstrate that human fetal cerebellarand cerebral cortex grafts survive in athymic nude ratrecipients, and continue their development in the anteriorchamber of the eye.The growth of the grafts and their histological appearance

suggests that good viability was maintained in oculo. Neu-rons of various sizes and glial elements could be visualizedreadily, as could vascular elements. The spatial distributionand appearance of the various cellular elements clearlyindicated an immature organization, even in grafts that hadbeen in oculo for 4 months.

Grafts of both brain areas that developed for 6-8 weeks inoculo expressed spontaneous electrical activity, with longduration action potentials and occasional multiple firing,typical for immature cortical and cerebellar neurons inrodents (19). The cerebellar grafts at 2 months in oculo did notrespond to surface electrical stimulation, which suggests thatfunctional intrinsic circuitry had not yet been established.Cerebellar grafts allowed to develop for 4 months in oculohad characteristics of more mature neurons in that thewaveforms were significantly shorter, and interspike-intervalhistograms consisted of mainly shorter intervals (19).Purkinje neurons in the older grafts responded to electricalsurface stimulation, suggesting that intrinsic parallel-fibercircuitry was beginning to develop. These grafts still had

Proc. Natl. Acad. Sci. USA 84 (1987)

Neurobiology: Bickford-Wimer et al. Proc. Natl. Acad. Sci. USA 84 (1987) 5959

Q k ..~~~~~~~~~~~~~~~~~..4~~~~~6kf~~~~~~~~~~~ .

.1~~~~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~~~~~~~~~'~~~~.'.eat~AA

4-. 4L.~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~44* *4

*~~s.¾. *~~~'~'' ,941* vZ. AV~~~~~~~~~~b4*4~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~-

hr.V-'- t. '-.i. % PEZP-%.~-ilb do.~4~. 44

FIG. 1. Cresyl violet-, NF-, and GFA-stained sections of human fetal cerebellar grafts. (a) Cresyl violet-stained overview of the centraltwo-thirds ofthe graft at 4 months in oculo. Toward the bottom is seen the heavily pigmented host iris. Three dense aggregations of nuclei midwaybetween the iris and free surface of the graft may represent residual patches of the external granular layer. Large neurons, presumably Purkinjecells, are seen both scattered throughout the graft neuropil and in layers and groups close to the iris. Note also infiltration of host iris

5960 Neurobiology: Bickford-Wimer et al.

*e

*,Jts a, ^ Er I :

* .*[D'_Jo~~W ...~ 4

4tDf%

9, 1

,~~~ ~ ~ A

*

v*04

¢ A* *~~~~~

....S:AL 9r0 .w41-

Of"~~~~~~~~-Ai__~~~~~~~~ - ;& t2

FIG. 2, GFA immunoreactivity in human cerebellar tissue grafted to hude rats. (a and b) Consecutive sections of the same area of a cerebellartransplant after 4 months in oculo. (a) When the tissue is stained with cresyl violet, the free surface of the transplant is seen at the top witha patch of presumed residual external granule layer cells (G) below. Host iris blood vessels are seen at the bottom. (x260.) (b) When the tissueis labeled with GFA-antibodies, the patch of external granule cells can be shown to contain relatively little GFA immunoactivity. Longer andrelatively straight GFA-positive fibers in the upper half of the transplant may represent disorganized Bergmann glia. (x260.) (c) Overview ofpart of a human cerebellar graft after 6 weeks in oculo. The transplant contains many GFA-immunoreactive cells and processes suggesting amoderate gliosis. The iris is to the left. (x 100.) (d) Cerebellar transplant from a 10-week-old fetus after 4 months in oculo. The amount ofGFA-positive structures has increased as compared with c. In addition, a certain degree of organization of GFA-positive processes can be seen.(X100.)

some characteristics of immaturity; therefore, longer periodsof growth in oculo may be necessary for maturation to takeplace.Recent studies have shown that human fetal tissues can

also survive and develop in cyclosporin A-treated host rats.Thus, human fetal substantia nigra dopamine neuroblasts canfunctionally reinnervate the dopamine-denervated adult ratstriaturn (7, 10) provided that the host animals are givencontinuous daily cyclosporin treatments. Similarly, it hasbeen shown that several different human fetal central nervoussystem areas can survive and develop when grafted to theanterior chamber of the eye ofcyclosporin A-treated host rats(11). However, in spite of rigorous immunosuppression, onlya limited proportion of the grafted material survived

intraocularly (19). It is not known to what extent cyclosporintreatment, in itself, interferes with brain development. Asshown in the present experiments, the athymic nude ratoffers an alternative solution to xenogeneic brain tissuegrafting with several advantages. There is robust growth anddevelopment with no signs of rejection, suggesting that alltransplants that are viable at the time of grafting will survive.

Several examples of the juxtaposition of human and ratcellular elements were observed in the present study. Thehuman transplants became vascularized from the rat host inrs.Moreover, melanocytes with many pigmented processesinvaded the transplants and became integrated with the graftneuropil.We have previously studied syngeneic grafts of cerebellar

melanoeytes in the human brain-tissue neuropil. (x 100.) (b) Close-up of the transplant showing a band of larger neurons, presumably Purkinjecells, surrounded by small cells, presumably granular cells. (X220.) (c) NF-like immunoreactivity in a human cerebellar transplant at 4 monthsin oculo. Many nerve fiber systems express NF immunoreactivity, particularly at the transplant surface. Note also a few NF-positive cell bodies(e.g., at the arrow). In general, most larger neurons, presumably Purkinje neurons, are NF negative. (x 150.) (d and e) Laminin immunoreactivity6 weeks (d) and 4 months (e) after transplantation. At 6 weeks the host iris is rich in laminin immunoreactivity (at the bottom of d), whereasthe transplant neuropil is laminin negative. After 4 months in oculo (e), the transplant neuropil now contains many blood vessels with abnormallythick laminin-immunoreactive walls and thin laminin-immunoreactive sprouts. (d and e, x 100.) (f) Normal rat host cerebellar cortex shown inthe same magnification as d and e, illustrating the thinness of normal brain capillaries visualized by laminin immunohistochemistry. (x100.)

Proc. Natl. Acad. Sci. USA 84 (1987)

;-*

Proc. Natl. Acad. Sci. USA 84 (1987) 5961

A

1 ~iI~I I it III IIIi I I I I

20-

10-

f

t 0.5

20

100 200

B j

100 200

15FIG. 3. Action potentials and discharge patterns of a cerebellar

Purkinje neuron (A) and a cortex cerebri neuron (B) recorded fromhuman fetal grafts after 6-8 weeks in oculo. (A and B Top) Tracerepresenting the mean action potential waveforms from 10-20individual traces that were averaged with a computer. (A and BMiddle) Interspike-interval histogram. (A and B Bottom) Ratemeterrecord illustrating the firing rate over time. In A, the interspike-interval histogram contains an early modal peak representing dou-blet-firing patterns of the Purkinje neuron. Also evident is a secondmodal peak at 150 msec, which represents longer pauses betweendischarges. The neurons discharged intermittently as seen from theratemeter record. In B, a typical neuron recorded from human cortexcerebri grafts in oculo is shown. Action potentials from cortexcerebri grafts were observed with both initially positive or negativewaveforms. The interspike-interval histogram contains a peak withshort intervals representative of an initial firing of the neuron whenit was first encountered. Neurons rarely fired for sustained periodsdespite the presence of 50mM glutamate in the electrode. Calibrationbars in B are for both neurons. For action-potential tracings, thevertical calibration indicates amplitude in millivolts, and the hori-zontal bar represents time in msec. For the interspike-intervalhistograms, the number of events per 1-msec bin is indicated on thevertical axis, and time in milliseconds is represented on the hori-zontal axis. For ratemeter records, time in seconds is represented bythe horizontal bar, and the number of action potentials per second isrepresented by the vertical calibration.

(17) and cerebral (20) cortex in the rat anterior eye chamber.Under optimal conditions, these grafts manifest a matureorganotypic histological and electrophysiological organiza-tion within about 6 weeks after transplantation. Indeed, whenthe fetal age of the donor is taken into account, the matura-tional sequence of histological and electrophysiologicalchanges in cerebellar cortex in oculo (17) differs little fromthat seen in situ (19). In this context, the results of the presentstudies, as well as those where human fetal tissues weregrafted to cyclosporin-treated rat hosts (11), suggest that thetransplanted tissue develops histologically according to ahuman timetable (21) rather than according to the muchshorter host rat timetable. Although no comparative data isavailable, the timetable for the electrophysiological develop-ment is presumably similar to that which occurs in human insitu. Further experiments on xenogeneic grafts that remain inoculo for 1-2 yr will be needed to confirm these speculations.

FIG. 4. Poststimulus-time histogram of a cerebellar neuron froma fetal graft grown in oculo for 4 months. Electrical stimulationoccurred at the time indicated by the arrow. (Inset) Oscilloscopetracing illustrating a single-stimulus trial with the evoked actionpotential following the stimulus with a latency of 5 msec. The verticalaxis represents events per 1-msec bin; the horizontal axis representstime in milliseconds. Calibrations for oscilloscope tracing are inmilliseconds for the horizontal bar and millivolts for the vertical bar.

In conclusion, we have shown that human fetal brain tissuecan survive and grow in the anterior chamber of the eye of theathymic nude rat. This technique uniquely permits study ofhuman brain development, connectivity, and pharmacologi-cal properties in an otherwise immunologically incompatiblespecies.

We thank Dr. Doris Dahl for providing the antibodies used in thisstudy. This work was supported by the U.S. Public Health ServiceGrants AA 03527, AG 04418, and ES 02011; the Swedish MedicalResearch Council (14X-03185 and 04X-2887); the "Expressen"Prenatal Research Foundation; and Magnus Bergvalls Stiftelse.

1. Olson, L., Bjorklund, H., Palmer, M. & Hoffer, B. J. (1984) in Eric K.Fernstrom Symposium on Transplantation in the Mammalian CNS(Elsevier, Holland), pp. 365-388.

2. Olson, L., Seiger, A. & Stromberg, 1. (1983) in Advances in CellularNeurobiology, eds. Federoff, S. & Hertz, L. (Academic, New York),Vol. 4, pp. 407-442.

3. Olson, L., Bjorklund, H. & Hoffer, B. J. (1984) in Neural Transplants,Development and Function, eds. Sladek, J. & Gash, D. (Plenum, NewYork), pp. 125-165.

4. Bjorklund, A., Stenevi, U., Dunnett, S. B. & Gage, F. H. (1982) Nature(London) 298, 652-654.

5. Daniloff, J. K., Low, W. C., Bodony, R. P. & Wells, J. (1985) Exp.Brain Res. 59, 73-82.

6. Low, W. C., Lewis, P. R. & Bunch, S. T. (1983) Brain Res. 262,328-333.

7. Brundin, P., Nilsson, 0. G., Gage, F. H. & Bjorklund, A. (1985) Exp.Brain Res. 60, 204-208.

8. Inoue, H., Kohsaka, S., Yoshida, K., Ohtani, M., Toya, S. & Tsukada,Y. (1985) Neurosci. Lett. 54, 85-90.

9. Zalewski, A. A. & Gulati, A. K. J. (1984) Neurosurgery 60, 828-834.10. Stromberg, I., Bygdeman, M., Goldstein, M., Seiger, A. & Olson, L.

(1986) Neurosci. Lett. 71, 271-276.11. Olson, L., Stromberg, I., Bygdeman, M., Granholm, A.-Ch., Hoffer,

B. J., Freedman, R. & Seiger, A. (1987) Exp. Brain Res., in press.12. Borel, J. F. & Lafferty, K. J. (1983) Transplant. Proc. 15, 1881-1885.13. Festing, M. W., May, D., Connors, T. A., Lovell, D. & Sparrow, S.

(1978) Nature (London) 274, 365-366.14. Vos, J. G., Kreeftenberg, J. G., Kruijt, B. C., Kruizinga, W. &

Steerberg, P. (1980) Clin. Immunol. Immunopathol. 15, 229-440.15. Dawson, P. J., Kluskans, L. F., Colston, J. & Fieldsteel, A. H. (1982)

Cancer 50, 1151-1154.16. Maruo, K., Uevama, Y., Kuwahara, Y., Hoiki, K., Saito, M. &

Tamoaki, N. (1982) Br. J. Cancer 45, 786-789.17. Hoffer, B. J., Seiger, A., Ljundberg, T. & Olson, L. (1974) Brain Res.

79, 165-184.18. Coons, A. H. (1958) in General Cytochemical Methods, ed. Danielli,

J. F. (Academic, New York), pp. 399-422.19. Woodward, D. J., Hoffer, B. J. & Lapham, L. W. (1969) in Neurobiol-

ogy of Cerebellar Evolution and Development, ed. Llinas, R. (AMAEducation and Research Foundation, Chicago), pp. 725-741.

20. Palmer, M. R., Bjorklund, H., Olson, L. & Hoffer, B. J. (1983) Dev.Brain Res. 6, 141-148.

21. Verbitskaya, L. B. (1969) in Neurobiology of Cerebellar Evolution andDevelopment, ed. Llinas, R. (AMA Education and Research Founda-tion, Chicago), pp. 859-874.

Neurobiology: Bickford-Wimer et al.

lilt_ 11 IIIIIIIIIIII list t 6 m6 H i I im