THE JOURNAL OF COMPARATIVE NEUROLOGY 307~311-334 (1991)

Axonal Projections Between Fetal Spinal Cord Transplants and the Adult Rat Spinal Cord: A Neuroanatomical Tracing Study of Local Interactions LYN B. JAKEMAN AND PAUL J. REIER Departments of Neuroscience and Neurosurgery, University of Florida College of Medicine, Gainesville, Florida 32610

ABSTRACT Three neuroanatomical tracers have been employed to map the axonal projections formed between transplants of fetal spinal cord tissue and the surrounding host spinal cord in adult rats. Solid pieces of embryonic day 14 (EJ rat spinal cord were placed into hemisection aspiration cavities in the lumbar spinal cord. Injections of either (1)a mixture of horseradish peroxidase and wheat germ agglutinin- conjugated horseradish peroxidase, (2) Fluoro-Gold, or (3) Phaseolus vulgaris leucoagglutinin (PHA-L) were made into the transplants or the neighboring segments of the host spinal cord at 6 weeks to 14 months post-transplantation. Injections of anterograde and retrograde tracers into the transplants revealed extensive intrinsic projections that often spanned the length of the grafts. Axons arising from the transplants extended into the host spinal cord as far as 5 mm from the host-graft interface, as best revealed by retrograde labeling with Fluoro-Gold. Consistent with these observations, iontophoretic injections of PHA-L into the transplants also produced labeled axonal profiles at comparable distances in the host spinal cord, and in some instances elaborate terminals fields were observed surrounding host neurons. The majority of these efferent fibers labeled with PHA-L, however, were confined to the immediate vicinity of the host-graft boundary, and no fibers were seen traversing cellular partitions between host and transplant tissues. Host afferents to the transplants were also revealed by these tracing methods. For example, the injection of Fluoro-Gold into the grafts resulted in labeling of host neurons within the spinal cord and nearby dorsal root ganglia. In most cases, retrogradely labeled neurons in spinal gray matter were located within 0.5 mm of the graft site, although some were seen as far as 4-6 mm away. The distance and relative density of ingrowth exhibited by host axons into the grafts, however, appeared modest based upon the results of HRP and Fluoro-Gold retrograde labeling. This was further confirmed with the PHA-L anterograde method. Whereas some host fibers were seen extending into the transplants, the majority of PHA-L containing axons formed terminal-like profiles at or within 0.5 mm of the host-graft interface. The comprehensive view of intrinsic connectivity and host-graft projections obtained in these studies indicates that intraspinal grafts of fetal spinal cord tissue can establish a short-range intersegmental circuitry in the injured, adult spinal cord. These observations are consistent with the view that such grafts may contribute to the formation of a functional relay between separated segments of the spinal cord after injury. The fact, however, that the majority of host- and graft-derived axons were concentrated within the vicinity of the host-graft border raises several considerations about the dynamics of axonal growth and cellular interactions. In that regard, the present findings provide a useful baseline for elucidating mechanisms that regulate axonal elongation in this experimental setting. These results also provide a framework

Accepted January 9,1991. Address reprint requests to Dr. Paul J. Reier, Department of Neurological Surgery, Box 5-265, J H M Health Center, University of Florida College of Medicine, Gainesville, FL 32610. L. B. Jakeman is now at the Dept. Developmental Biology, Genentech, Inc. 460 Point San Bruno Blvd., South San Francisco, CA 94080.

o 1991 WILEY-LISS, INC.

312

L.B. JAKEMAN AND P.J. REIER for future tests of surgical, pharmacological, or molecular manipulations that may enhance the degree of axonal interaction between host and graft tissues. Key words: axonal tracers, horseradish peroxidase (HRP), Fluoro-Gold,Phaseolus vulgaris leucoagglutinin(PHA-L),glial fibrillary acidic protein (GFAP)

Over the past several years two transplantation strategies involving fetal neural tissue have been investigated for their potential to promote functional recovery in the injured spinal cord. The first approach focuses on the restoration of supraspinal monoaminergic influences below the level of injury (Nornes et al., '84) and is based upon the role that these descending fibers are believed to play in modulating segmental reflex activity (e.g., Conway et al., '88; Crone et al., '88; Hounsgaard et al., '88) and locomotor (e.g., Grillner, '73, '80; Shik and Orlovsky, '76). Accordingly, dissociated embryonic brainstem cells including noradrenergic or serotoninergic neurons have been introduced into the injured (Nornes et al., '83; Nygren et al., '77; Privat et al., '86, '89) and/or chemically axotomized (Commissiong, '84; Foster et al., '85; Bjorklund et al., '86) spinal cord. Extensive reinnervation of the host gray matter by grafted noradrenergic and serotoninergic neurons has been observed (Nornes et al., '83; Foster et al., '85; Bjorklund et al., '86; Privat et al., '86, '89). Furthermore, physiological and behavioral evidence has been described that shows that such grafts can exert some degree of functional impact related to segmental reflexes (Buchanan and Nornes, '86; Privat et al., '86; '89; Moorman et al., '90). The second intraspinal transplantation approach entails the reestablishment of anatomical and functional continuity at the injury site (Reier et al., '83, '85, '86; Houle and Reier, '88; Jakeman et al., '89). In newborn rats, intraspinal transplants of fetal spinal cord (FSC) tissue can serve as a tissue bridge that is conducive to the elongation of injured, as well as late-developing, supraspinal fiber populations (Bregman, '87). These axonal projections through the lesion site might provide the basis for some functional recovery. For example, recent studies have demonstrated that placement of FSC transplants into hemisection lesions in newborn rats can lead to improved hindlimb function (Kunkel-Bagden and Bregman, '90). Transplantation of embryonic CNS tissue into lesions of the adult rat spinal cord has also been shown to restore anatomical continuity to a considerable degree (Das, '83; Reier et al., '86; Houl6 and Reier, '88). It is not yet known whether fetal grafts in the adult spinal cord can mediate any degree of functional repair. However, based upon initial neuroanatomical findings, it has been suggested that such grafts may establish a neuronal relay that could facilitate synaptic communication between spinal cord segments rostral and caudal to the site of injury (Reier, '85; Reier et al., '86, '88). In view of this hypothesis, the present study was undertaken to explore the patterns of axonal projection established between FSC grafts and adjacent regions of the host spinal cord. As an extension of our earlier wheat germ agglutinin-horseradish peroxidase (WGA-HRP)tracing studies (Reier et al., '86), three complementary neuroanatomical tracers were used to obtain more specific qualitative and quantitative indices of host-graft integration. In addition, we examined the degree to which axonal projections might be influenced by astroglial scarring at the host-graft interface (Reier and Houle, '88; Eng et al., '87; Reier et al., '88).

Portions of this study have been summarized previously (Jakeman and Reier, '88).

MATERIALS AND METHODS Animals and transplantation surgery Female Spraque-Dawley rats (200-300 g) were obtained from Zivic-Miller Laboratories (Allison Park, PA) for use in this study. A total of 81 rats received transplants of FSC tissue according to a modification of previously described methods (Reier et al., '86). The transplant recipients were anesthetized with ketamine (Ketaset, 60-80 mg/kg) and xylazine (Rompun, 10 m a g ) . A laminectomy was performed at the TI, vertebral level (approximately spinal L,-L,) and a cavity of 3-6 mm in length was created in the left half of the spinal cord. The lesions were routinely extended to a full hemisection by removal of both the lateral and ventral columns. By methods detailed in a previous report (Reier et al., '831, the embryonic spinal cord tissue was dissected from fetal rats at day 14 of gestation (EJ. Once hemostasis was achieved in the host, one or two pieces of donor spinal cord were placed into the cavity and a piece of hydrocephalus shunt film (Durafilm; Codman Surlef, Inc.) was positioned above the graft. The dura and superficial wounds were then closed in layers.

Postoperative care All rats received a subcutaneous injection (75,000 U.) of long-acting penicillin (Dual-Pen; Tech America) immediately after surgery. The rats were housed in the University of Florida A.A.L.A.C.- accredited animal resource facility. They were examined daily by a veterinarian or veterinary technician for postoperative complications secondary to spinal cord injury. A small number of animals initiated mild self-mutililation of the left hindlimb and were sacrificed shortly thereafter. This consideration contributed to the range of postgraft survival times shown in Tables 1-5.

Tracer application and tissue processing At postgraft intervals rangingfrom 6 weeks to 14 months, the transplant recipients were reanesthetized with ketamine and xylazine and prepared for tracer application. The region containing the graft and surrounding host spinal cord was exposed by removing new bone growth and extending the original laminectomy rostrocaudally. After the tracer was injected, the surface of the cord was washed with physiological saline and a drop of mineral oil was placed over the spinal cord to minimize diffusion of the tracer into the surrounding tissues. The spinal cord was then covered with a piece of Durafilm and the wound was closed as described above. Horseradishperoxidase (HRP) and wheat germ agglutinin-HRP conjugate (WGA-HRP) Anterograde and retrograde labeling. A combination of HRP (Type VI) and WGA-HRP (both from Sigma Chemical) was used for retrograde labeling of cells and anterograde

INTRASPINAL HOST-GRAFT PROJECTIONS

313

TABLE 1. Summary of Labeling in Host and Graft Tissues Following HRP and WGA-HRP Injections Into FSC Transplants Animal Code (Postgraft Interval) HGP6 (6 wk) HGP7 (5 wk) HGPZ (6 wk) HGP8 (6 wk) HGP3 (6 wk) H G P l l (6 wk) HGPS (9 wki HGPlO ( 6 wk) HG3 ( 2 mol HGPS (6 wk)

Method'

Group'

Intrinsic Labeling

Picospritz Iontophoresis Picospritz Tungsten wire Tungsten wire Tungsten wire Picospritz Hamilton Syringe Tungsten wire Hamilton Syringe

A

+

Labeled Axons In Host (Max. dist. in mmI3

+

A B B B B C

+ +

+ + + + +

C C C

Labeled Cells In Host (#I ( M a x dist. in mmI3

+ (1)(3 mm) + (0.5 mm) + (1.0 mm) + (0.5 mmJ

+ (>30)(1mm)

+ (0.5 mm)

+ (>30)(3mm)

'The tracers were applied using one of five procedures (see Methods). 'Recipients included in analysis were classified according to the extent of tracer diffusion as follows: A. Injection site < 0.5 mm in diameter and confined to graft; B. Injection larger than 0.5 mm and confined to graft; C. Injection site within graft, tracer diffused into or over interface. 3Distances from host-graft interface. +Indicates evidence of projections from labeled profiles present in host or graft tissue. *Reaction product too dense to detect individual axons and cells within graft.

TABLE 2. Summary of Labeling Following HRPIWGA-HRP Injections Into the Host Spinal Cord

Animal code (postgraft int.) H H 3 (14 mo.) HH4 (5.5 mo.) HH6 (2 mo.) HH7 ( 2 mo ) HH8 (2 mo.) HH9 (6.5 mo.i H H l l (6 mo.)

Site of' injection

Dist. from' center of I S . to interface (mm)

Dist. from edge of I S . to interface (mm)

rost. caud. rost. rost. caud. rost. caud. rost. caud. rost. mud. rost. caud.

7.0 10.6 1.9 5.0 5.7 2.9 3.9 2.0 3.7 2.4 2.7 3.0 6.2

2.1 6.8 1.0 3.2 4.3 1.4

Labeled axon5 in graft4

+ +

2.8

-

0.6 1.9 1.2 1.7 0.0 0.0

-

+

'Injection placed either rostral (r0st.J or caudal (caud.) to host-graft interface. 'Measured distance from center of injection site (1,s.) to t h e host-graft interface. 3Measured distance from the edge of HRP reaction product or diffusion to the grafthost interface 'MI axons terminated within 0 5 mm of host-graft interface. +Indicates retrogradely filled cells or anterogradely filled axons present within t h e transplant.

filling of axons (Mesulam, '82), as described in previous studies (Reier et al., '86; Houle and Reier, '88). In the first part of the experiment, mixtures of the two tracers were applied to transplants (n = 16) by using a variety of techniques (Table 1).Different procedures were employed in an attempt to achieve a range of injection sizes and to optimize labeling. Seven rats received pressure injections of a solution of 20% HRP and 1% WGA-HRP through a 1.0 p1 Hamilton syringe or a nitrogen burst apparatus. Two rats received iontophoretic injections of 2% WGA-HRP and one transplant was labeled by placing a pledget of Gelfoam soaked in 20% HRP and 2% WGA-HRP into the exposed graft tissue. HRP and WGA-HRP were applied to the remaining six rats by preparing a thick solution of the tracers in phosphate-buffered saline. One end of a 1-cm piece of tungsten wire was then coated with tracer by repeatedly dipping the wire into the solution until a bolus of approximately 0.5 mm diameter was created. The tungsten wires were stored at -20°C until use. The tracers were then applied by inserting the wire into graft or host tissue until the tracer dissolved. For the reciprocal study (Table 2), HRP and WGA-HRP were applied to the host spinal cord

(n = 11) using this tungsten wire approach. A similarly prepared tungsten wire was placed into one normal rat spinal cord at the T,, vertebral level. HRP histochemistry. After allowing 48-72 hours for transport of the tracer, the recipients of HRP/WGA-HRP injections were anesthetized with sodium pentobarbital and perfused transcardially with 150 ml of heparinized 0.9% NaCl followed by 250 ml of fixative (1.0% paraformaldehyde + 2.5% glutaraldehyde in 0.1 M Sorenson's phosphate buffer). Tissue blocks, including the transplant and 5.010.0 mm of host spinal cord rostral and caudal to the graft, were removed. Vibratome sections (50 pm) were cut in the sagittal or horizontal plane and the sections were reacted immediately according to the tetramethylbenzidine (TMB) protocol of deOlmos ('78). Sections were mounted onto gelatin-coated slides and selected sections were counterstained with 1.0% Neutral Red to reveal the cytoarchitecture of the host and graft tissues. Fluoro-Gold (FG) Retrograde labeling. For the second set of experiments, a 2.0% solution of FG (Fluorochrome, Inc.; Englewood, CO) was made in 0.9% NaCl and the solution was then drawn into glass pipettes (40-50 ym tip diameter). After a dural incision had been made with the beveled end of a 25-gauge needle, the FG solution was injected into the transplants (n = 12) or the host spinal cord (n = 19) by means of a rapid nitrogen burst. The injected volume was estimated by measuring the diameter of the hemispheric droplet ejected onto a parafilm sheet and by using the approximation (v = (2a(d2),)/3). Volumes of 0.1-1.0 y1 were injected into the transplants and 0.5-1.5 pl into the host tissue. One normal rat also received an injection of 0.5 ~1 of FG solution at the T,, vertebral level for comparison. Tissue processing. The FG- containing tissue was processed as described by Schmued and Fallon ('86). Four days following FG- injection, the rats were perfused with saline followed by fixative containing 4.0% paraformaldehyde and 0.25% glutaraldehyde in 0.1 M phosphate buffer. Spinal cord blocks containing the transplant and 4.0 mm of the surrounding rostral and caudal spinal cord were removed and postfixed in the same fixative for 2-18 hours at 4°C. Vibratome sections of 40 ym were cut in the sagittal plane. In addition, every sixth section was saved in phosphatebuffered saline (PBS) for subsequent immunocytochemical detection of glial fibrillary acidic protein (GFAP; see below).

-

-

-

-

-

-

-

-

-

-

-

-

1

2

3

4

30

2

5

0.5-0

'

'

30%

17%

32%

48%

42%

23%

w

4

3

38

5a

1

-

-

0-0.5

-

10

Fusion Index: % o f l i n e a r h o s t - g r a f t i n t e r f a c e which i s deviod o f g l i a l s c a r (average o f 4-10 s e c t i o n s p e r animal).

1.0 mn.

5

1.1-2

1

13

2.1-4

2

27

4.1-6

C e l l s Caudal t o Transplant (Distance i n mn)

0.6-1.0

Scale corresponds t o approx.

34%

37%

47%

49%

21%

25%

1.0%

35%

21%

A 56%

A

(F1.X;)

G r a f t and I n j e c t i o n S i t e Glial (Freehand drawings, t o s c a l e ) sca

33%

35%

(FI.%)l

Glial scar

DISTRIBUTION OF RETROGRADELY LABELED HOST NEURONS FOLLOWING FLUORO-GOLD INJECTIONS I N T O F S C TRANSPLANTS

Nunbers i n each c o l u m represent t h e number o f c e l l s counted i n each region.

FGC-31 (6 wk.)

FGC-27 (14 w . )

(2 w.)

FGC-26

(5.5 mo.)

FGC-H10

( 6 mo.)

FGC-H9

FGC-HS (6.5 m o . )

FGC-H7 ( 2 mo.)

( 2 mo.)

FGC-H6

FGC-2a ( 2 mo.)

1.0-0.6

CelLs R o s t r a l t o Transplant (Distance i n m)

6-4.1 4-2.1 2-1.1

Animal (Post-g I.)

TABLE 3.

R

2

7

1

- -

6

219 2

L

DRG

INTRASPINAL HOST-GRAFT PROJECTIONS

315

TABLE 4. Neuronal Labeling and Glial Reactivity Following Injections of Fluoro-Gold Into the Host Spinal Cord Animal number (Postgraft interval)

Inj. site (in host) (mm from interface)

Number of1 labeled cells

Fusion index' (FI, % k SD)

Glial ratio3 (eraft %/host %)

3.6 mm rostral 4.3 mm rostral 2.8 mm rostral 1.5 mm rostral 4.4mmcaudal 3.6 mmcaudal 1.8 mm caudal

0 38 28 8 697 212 439

47% (213) 19% (k11) 40% (k19) 8% (k13) 51% (220) 40% ( 216) 27% (221)

2 3" 1.2 4.F

FGC-20 (8 mo) FGC-17 (9mu) FGC-21(8 mu) FGC-23 (8 mo) FGC-13 (8 mu) FGC-14 (5.5 mo) FGC-30 (6 wk)

10.0*

3.1* 1.2 1.1

'Total number of retrogradely filled cells within the transplant?Composite Fusion Index (FI):Percentage of the host-graft interface (closest to injection) that is devoid of dense glial scar (average of 6 1 0 section~itransplant).~ Ratio of the percent area occupied by glial elements. *indicates that the glial density within the graft was significantly higher than that in the surroundinghost tissue (p 5 0.01).

TABLE 5. Orthograde Labeling Following Iontophoretic Injections of PHA-L Into FSC Transplants or Adjacent Host Spinal Cord Animal (postgraft int.)

Injection site and location

PHAL-8 (2 mo.)

Transplant and Host ventral tracts

PHAL-10 (10 wk.)

Transplant, dorsal-medial region (rostral)

PHAL-23 (6 mo.)

Transplant, dorsal-medial region (caudal)

PHAL-15 (6 mu.)

Transplant, dorsal-lateral region (caudal) Transplant, caudal region Host, 0.5 mm caudal to transplant Host, 1.5 mm caudal to transplant Host, 1.5 mm rostral to transplant Host, 0.5 mm caudal to transplant *(antem- and retrograde transport) Host, large inj. 2.0 mm rostral

PHAL-32 (10 wk.) P H A L-lI(6 wk.) PHAL-21 (6 mo.) PHAL-33 (10 wk.) PHAL-27 (12 mo.) PHAL-16 (6 mu.)

Axonal labeling Labeled axons extend across interface. Injury filled host axons also evident. Labeled axons in rostral dorsal horn and intermediate gray regions of host. Labeled axons present in caudal dorsal born, dorsal tracts and intermediate gray, and in rostral intermediate gray regions. Labeled axons surround lateral motoneurons. No laheled axons in host. Labeled axons extend s 0.2 mm into graft. Labeled axons extend s 0.3 mm into graft. Labeled axons stop at interface. Labeled cells and axons within graft, most within 1.0 mm of interface. No labeled axons in =aft.

In four of the rats that had received larger injections of FG into the transplants, six additional tissue blocks were also sectioned. These included cross sections of the host spinal cord 4-6 mm rostral and caudal to the transplant, horizontal sections from cervical and thoracic spinal cord, and sections of host brainstem and cortex. The dorsal root ganglia from these recipients were embedded in paraffin and sectioned at 15 pm. The vibratome and paraffin sections were mounted directly onto gelatin-coated slides. Slides from paraffin blocks were heated to 37°C for 12 hours, and deparaffinized. All FG slides were cleared in xylene, coverslipped with Fluoromount (Gurr Bio/medical Specialties; Santa Monica, CA), and viewed on a Zeiss Axiophot microscope with fluorescent UV illumination. Phaseolus vulgaris leucoagglutinin (PHA-L) Anterograde labeling and tissue sectioning. To examine the patterns of axonal elongation into graft and host tissues, anterogradely filled axons were identified by immunocytochemical detection of PHA-L (Vector Laboratories Inc.; Burlingham, CA). The tracer was applied by a modification of the methods of Gerfen and Sawchenko ('84).The PHA-L was dissolved to 2.5% in 10 mM phosphate buffer (pH 8.0) and back-filled into precleaned glass micropipettes (tip diameter of 10-15 pm). After exposing the transplant (n = 13) or host spinal cord (n = 81, the tracer was applied to the appropriate site by iontophoresis for 20 minutes using a 5 pA interrupted positive current (cycles at 7 sec on, 7 sec off).

After allowing 7-17 days for transport of the PHA-L, the recipients were perfused as above with fixative containing 4.0% paraformaldehyde and 0.25% glutaraldehyde. The cord blocks including the transplant and 4-7 mm of the surrounding rostral and caudal spinal cord were removed and postfixed overnight at 4°C. Sagittal sections of these blocks were cut at 40 pm on a Vibratome and stored in 0.02 M potassium phosphate buffered saline (KPBS). Every sixth section was saved for immunocytochemical staining with antibodies to GFAP (see below). Immunocytochemical detection of PHA-L. The remaining free-floating Vibratome sections were processed for the identification of cells and processes containing PHA-L. The sections were first washed in 0.02 M KPBS and incubated for 2-4 hours in a preblocking bath containing 2.0%normal rabbit serum and 0.3% Triton X-100. All the sections were then incubated in goat anti-PHA-L (Vector) diluted 1:5000 in KPBS for 36 hours at 4°C and 2 additional hours at room temperature. The sections were rewashed and then processed with biotinylated rabbit antigoat IgG (1:225) and Vector Avidin-Biotin-peroxidase Complex (ABC). The final peroxidase conjugate was reacted with H,O, in the presence of 0.05%diaminobenzidine (DAB). In some cases, the DAB reaction was enhanced with the addition of 0.125% nickel ammonium sulfate. The nickel-enhanced sections were counterstained with 0.1% Cresyl Violet or 1.0%Neutral Red prior to coverslipping.

Histological analysis of anatomical tracers Mounted serial sections containing the tracer injection sites were examined to determine the location of the injection site and the extent of tracer diffusion with respect to the host-graft interface. Each specimen was then accepted or rejected from the study according to specific transport and diffusion criteria as described in the results section. Retrogradely filled cells were identified and counted in successive 1.0 mm fields at 1 2 5 ~In . FG specimens, cells that displayed nonspecific fluorescence when exposed to rhodamine (510-560 nm) or fluorescein (450-490 nm) microscope filters were not counted. Each field was counted three times and the median value was accepted. All cell counts were corrected according to classical methods (Abercrombie, '46). Total cell number was obtained by assuming an average cell diameter of 40 pm for graft cells and 50 pm for host neurons. The distribution of labeled cells and axons was determined from drawing tube tracings of darkfield or brightfield images (HRP and PHA-L) or from photographic montages of fluorescence micrographs (FG). To determine the distances of anterogradely labeled axonal projections, a digitizing tablet and morphometry software (Zeiss Videoplan) were calibrated for the appropriate magnification. Measure

L.B. JAKEMAN AND P.J. REIER

316

ments were taken from the drawing tube illustrations or photomicrographs. Similar methods were used to document the distances between the injection site and the outermost zone of tracer diffusion as well as the relationship of these regions to the host-graft interface.

Immunocytochemical staining and analysis of GFAP A series of sections (240 km apart) from recipients of FG or PHA-L injections was incubated in rabbit polyclonal antiserum produced against GFAP (gift of Dr. Lawrence F. Eng, VA Medical Center, Palo Alto, CAI. The antiserum was diluted 1:1,200, and sections were incubated overnight at 4°C. Detection of the primary antibody was performed according to the peroxidase anti-peroxidase method as described previously (Reier et al., '86). Tracings of the rostral and caudal interface regions for each section were made using a drawing tube, delineating the regions containing dense GFAP staining between host and graft tissue. For specimens with FG injections, the lengths of the interface and the regions containing glial scar formation were measured with the aid of a digitizing pad and Videoplan software. The composite Fusion Index (FI) for each interface region was defined as the averaged percentage of the interface that was devoid of dense glial scarring (Houle and Reier, '88). The density of glial staining was determined for regions in the transplant and surrounding host tissues in seven graft recipients. A program was developed by using the Zeiss IBAS image analysis system (Kontron, Germany) and a high resolution video camera (DAGE Inc., CCD71). At a viewing magnification of 125x , four pairs of images from each GFAP stained section (each pair including a graft region and host gray matter region) were digitized and converted to binary images. The first field of each pair was segmented interactively by the user to distinguish glial processes from background as described by Bjorklund and Olson ('83). On the assumption that nonspecific staining was consistent within each section, the segmentation settings used for this first image were retained for the second image of the pair. The percent area occupied by glial profiles was calculated for each field, and values were obtained for the average glial density within the graft and host gray matter regions as well as the ratio of grafthost glial density for each section. Statistical comparison of graft and host glial density was performed by using the direct-difference Student's t test for paired samples (Spence et al., '83).

Fig. 1. HRP and WGA-HRP tracing revealed intrinsic interactions and some projections of graft and host axons. (a)Drawing tube tracings of sequential sagittal sections through a graft with a small (Group A) injection. The center of the injection site is shown as solid black and the area containing dense reaction product with no discernable cells and axons is represented by the hatched region. The area outlined by a dotted line represents a high density of labeled cells and axons, and each cell within the graft is indicated by a larger dot. Scale corresponds to 1.0 mm. (b-g) Darkfield photomicrographs from TMB stained sections. (b) Labeled cells and axons were distributed throughout a transplant (t; HGP11) to the host-graft interface (arrowheads) following a larger HRPIWGA-HRP injection into the dorsal region of a graft. Scale corresponds to 200 pm. ( c )Labeled graft axons coursed parallel to this region of the host-graft interface, but do not penetrate the host spinal cord. Scale corresponds to 200 pm. (d-fl Axonal projections formed

RESULTS General transplant characteristics Viable transplants were present in 74/81 of the recipients. Nearly all of the grafts filled the lesion cavity and restored gross anatomical continuity with the host spinal cord surfaces bordering the wound site. The cellular organization of the host-graft interface was similar to that described in previous studies from this laboratory (Reier et al., '86; Houle and Reier, '88). In some cases, the interface was characterized by a well-defined tissue partition consisting of small, densely packed cells resembling glia. In contrast, there were other regions of the interface that exhibited an uninterrupted transition between host and graft neuropil. A similar range of host-graft fusion could be defined in GFAP stained sections.

Neuroanatomical tracing with HRP and WGA-HRP Injections into transplants. Using several injection procedures (see Methods), mixtures of HRP and WGA-HRP were applied to 16 transplants. After histological processing, each section was examined under darkfield illumination, and the extent of the injection site was defined as the area containing a purple-opaque core and the surrounding region of orange TMB reaction product (Fig. la,b). The injection site was restricted solely to the transplant in six cases (Table 1, Groups A, B). The two specimens included in Group A contained the smallest injection sites, with a maximal diameter of less than 0.5 mm. Labeled cells or axons were not observed in the host spinal cord, but the distribution of such profiles within each of these two grafts indicated an extensive network of intrinsic graft projections (Fig. la). The majority of retrogradely labeled cells were located within 0.5 mm from the center of the injection site; however, some tracer-filled neurons were also found at greater distances. In the other four animals (Table 1, Group B), the injection sites were larger than 0.5 mm in diameter but were nevertheless confined to the transplants. An extensive display of intrinsic graft projections was again evident as labeled cells and axons were observed throughout the transplants (Figs. l b , c, e). Three of four specimens in group B provided evidence of some host-graft projections (Fig. 2). One case, (HGP2), contained a single well labeled cell within the host ventral gray matter, approximately 3 mm from the host-graft interface. Anterogradely filled axons could be followed across the interface into the host in

between host and graft tissues. Graft efferent projections were identified by retrograde transport into transplant neurons following injections into the host spinal cord (d), and anterogradely labeled axons (white arrows) extending into the host spinal cord after an injection into the transplant (e). (DExample of short-distance ingrowth of host axons into a transplant following an injection made 1.9 mm rostral to a graft. Scale corresponds to 100 km for (d-fl. (g) Illustration of the potential for axonal interactions following an injection which diffused across the host-graft interface (specimen HH5). Retrogradely filled neurons (*) may represent intrinsic projections labeled by tracer diffusion. However, note the presence of large numbers of labeled fibers (i.e., arrowhead) traversing the interface region between the host (heavy label) and graft (light label) tissues. (h = host; t = graft). Scale = 100 Wm.

Figure 7

318

L.B. JAKEMAN AND P.J. REIER

HG P2

t

HGP3

-

Fig. 2. Tracings of three transplants (t) with HRPmGA-HRP injections (Group B) demonstrating axonal labeling within the host spinal cord (h). The extent of diffusion from the injection site is marked by the shaded area. (*) Indicates the presence of a cyst between host and graft tissues. (I) Refers to extensive labeling of cells and axons intrinsic to the graft (see also Fig. lb,c,e). Three animals exhibited evidence of

limited axonal projections between host and graft tissue by retrograde or anterograde labeling in the host spinal cord (Table 1). Specimen HGP2 (top figure) contained one retrogradely labeled cell 3 mm from the host-graft interface. Specimens HGPl1 and HGP3 each contained anterogradely filled s o n s that extended into the nearby host spinal cord across regions of close host-graft apposition. Scale = 1.0mm.

the remaining two recipients (HGP3, HGP11; see Fig. le). These fibers extended into the host spinal cord and could be followed for up to 1 mm within a single plane of section. Unfortunately, any labeled axons that were oriented perpendicular to the section plane could not be assessed. In these specimens, axons that entered the host spinal cord could be traced through regions where host and graft neuropils were contiguous. In contrast, in areas where a cellular partition was present, labeled axons were aligned with the interface, and none could be seen traversing these scars. Thus, the pattern of axonal elongation visualized in such cases essentially mirrored the contour of the interface (Fig. lc). The four specimens classified in Group C had injection sites that were not confined to the graft. In each case, the outer zone of TMB reaction product extended beyond either the rostral or caudal border interface, whereas the opaque center was fully confined within these grafts. Labeled axons and 50-100 host neurons were found within the host spinal cord (Table 1).Most of these cells were within 1.0 mm of the host-graft interface; additional labeled cells were observed 2-3 mm from the graft in only one case. Znjections into the host spinal cord. In the reciprocal experiment, HRP and WGA-HRP injections were made rostral and caudal to the transplants. The details of the postgraft intervals, injection sites, and distances between the injections and the host-graft interface are summarized

in Table 2. The injections were completely confined to the host spinal cord in seven animals. Evidence for axonal projections between graft and host tissues was obtained from five of the seven recipients (Fig. 3). Each displayed a large injection ( 2 1.5 mm radius) with the edge of the tracer site being within 2.5 mm of the interface region. Four of these five specimens indicated the outgrowth of graft neurons into the host by the presence of retrogradely filled neurons within the transplants (Fig. Id). In three of these grafts (HH3, HH7, HH8), only a modest outgrowth from donor cells could be inferred, as no more than three retrogradely filled neurons were seen. In contrast, a fourth graft (HH11) showed more extensive HRP labeling. In this example, the tracer reaction product extended to the host-graft interface but did not spread into the transplant. Irrespective of the degree of retrograde labeling or localization of the injection sites, nearly all of the labeled graft neurons observed in these four cases were located within 1.0 mm of the host-graft interface. The HRPNGA-HRP applications also resulted in anterograde filling of axons that extended from the injection site into the transplants in three of these five recipients (HH3, HH4, HH8). In each case, the interface between host and graft tissues was readily apparent as a distinct border between the dense axonal labeling in the host and very sparse labeling in the transplant (Fig. lfl.With few exceptions, the labeled axons terminated within 0.5 mm of the

INTRASPINAL HOST-GRAFT PROJECTIONS

319

HH-7

HH-11

. Fig. 3. Tracings of sections from five transplants (t)with HRPWGAHRP injections (shaded areas) restricted to the host spinal cord (see Table 2). (*) Indicates damage to tissue at the injection site. Retrogradely labeled neurons were found in grafts made in animals HH-3,

HH-7, HH-8, and HH-11; each dot represents one labeled neuron. Anterogradely labeled axons were present within three of the recipients (HH-3, HH-4, HH-8). Scale = 1.0 mm.

host-graft interface. This pattern was contrasted in the remaining recipients, where the labeled axons stopped abruptly and uniformly at the interface (not shown). Evidence for more extensive axonal integration within the host-graft interface region was again suggested by one case in which an injection extended over a portion of the interface (HH5). The labeled axons and cell bodies within this graft were excluded from consideration of host-graft projections. However, a region ventral to the area of diffusion was marked by the presence of labeled axons that could be followedbetween host and graft tissues (Fig. lg). Together, the results following HRP and WGA-HRP tracing and histochemistry provided only limited evidence of host-graft interactions. Nevertheless, the positive examples described prompted further investigation of axonal projections with other tracing techniques.

tracer diffusion was confined to the transplant. Seven of these grafts had some labeled cells within the host spinal cord. The numbers ranged from 2-10 labeled cells in four of the specimens to 30-50 neurons in two others. In the seventh case, more than 70 host spinal cord cells and 200 dorsal root ganglion (DRG) neurons were labeled. Table 3 summarizes the location and extent of the graft injections and the distribution of labeled host neurons. Five of these FG injections measured 10.5 mm in hameter (FGC-2a, H6, H7, H8, H10; Fig. 4a). In these specimens, nearly all of the labeled neurons were found within the graft itself. Such neurons were always concentrated in the region near the injection site. However, some retrogradely labeled cells were also found in more distant regions of these transplants (Fig. 4b). Neurons containing FG were also present in the adjacent host spinal cord of two of these recipients. In both cases, the cells containing FG were located within 0.5 mm of the interface zone adjacent to the injection site and were distributed within the medial or lateral intermediate gray (laminae N-VII). The greatest

Fluoro-Gold injections Injections of F G into transplants. In 10 of the 1 2 recipients with FG injections into the grafts, the extent of

L.B. JAKEMAN AND P.J. REIER

320

Figure 4

INTRASPINAL HOST-GRAFT PROJECTIONS

321

number of labeled host neurons were found in the recipient with the injection located within 0.2 mm of the host-graft interface (FGC-H10). In each of the remaining transplants from this group (FGC-H9,25,26,27,31), the injection sites were larger than 0.5 mm in diameter but were still confined to the graft. All of these specimens contained some retrogradely labeled neurons within the host that were most frequently found near the host-graft interface. The number of these neurons, however, was relatively modest in all but two cases in which FG was injected within 0.5 mm of either the rostral or caudal interface. In one of these specimens (FGC-25; Figs. 4c-i), for example, the FG injection spread to less than 0.2 mm from the caudal interface. Retrogradely labeled neurons were found throughout the host spinal cord caudal to the graft, whereas few were found rostral to it. Within the sacral spinal cord (Fig. 4h; i.e., 4-6 mm away), labeled cells were distributed throughout the dorsal and intermediate gray regions, as well as within the ventral horn. In addition, more labeled cells were located ipsilateral than contralatera1 to the graft. In this animal, a large number of ipsilateral DRG cells were also labeled (Fig. 4i). Sections from cervical and thoracic spinal cord, as well as the brainstem and cerebral cortex, were examined in all of the recipients with larger FG injections into the graft. However, no retrogradely labeled cells were found in any of these rostral or supraspinal areas. Injections into the host spinal cord. Fluoro-Gold injections were made into the host spinal cord of 19 recipients at distances ranging from 1.8-5.0 mm from the host-graft interface. In seven of these rats, the following criteria were met: (1)the injections were completely confined to the host spinal cord, (2) there was no evidence of nonspecific fluorescence within the central canal region, and (3)the injections were large enough to label at least a high percentage of host neurons adjacent to the graft. Results from these seven animals are summarized in Table 4. The interface between the host gray matter and graft tissue was characterized by a sharp decrease in the density of labeled neurons between adjoining host and transplanted tissue (Fig. 5a). In all seven grafts, there were regions of fusion between host and graft tissues as identified by the absence of intense GFAP immunoreactivity (see below; Fig. 5b). Six of the grafts contained retrogradely labeled cells. These included three transplants (FGC-17,21,23), each with fewer than 50 retrogradely filled neurons and another three (FGC-13,14,30) with more than 200 labeled cells per graft. In this sample, the differences between the two groups did not correlate with the postgrafting interval or the distance between the injection site and the host-graft interface. Interestingly, more labeled cells were found in the three recipients with FG placed caudal to the graft than

those with injections placed at similar distances rostral to it. FG-labeled neurons that projected into the host spinal cord were distributed throughout the graft tissue (Fig. 5e). Most of these neurons were multipolar and small, measuring 8-40 pm in diameter (Fig. 5c), although larger cells were occasionally labeled (Fig. 5d). Histograms were made for each of these six grafts to show the distribution of the labeled cells as a function of distance from the host-graft interface (Fig. 6). In general, most labeled cells were found within 1 mm of the interface. However, in the three specimens with greater numbers of labeled neurons, these cells were distributed throughout the full length of the grafts, for a distance of 3-7 mm (see legend). Comparison of FG and HRPIWGA-HRP injections in normal rats. To compare the patterns of retrograde cell labeling using HRPIWGA-HRP and FG, each of these tracers was injected into normal rats a t the T,, vertebral level and the tissues were processed in the same manner as the transplant recipients. Histological analyses revealed labeled neurons within the cervical and thoracic spinal cord, brainstem, DRG, and cerebral cortex of both animals. Although total cell numbers were not determined, a rough estimate indicated an approximate fivefold greater retrograde labeling of cells in each of these regions in the rat that had received the pressure injection of FG.

Fig. 4. Retrogradely labeled host neurons following FG injections into transplants. (a,b) Intrinsic labeled neurons following a small FG injection ( < 0.5 mm diameter) into a transplant (t; Specimen FGC-H6). (a) The majority of labeled neurons were located in the immediate vicinity of the injection, which was adjacent to the rostral host-graft interface (arrowheads). Cells were also found at the far end of the graft (b)at a distance of 3 mm from the injection. Scale corresponds to 200 Fm in a: and 100 p,m in b:. (c-i) Distribution of retrogradely labeled cells following a larger injection (dotted line in e) near the caudal border (arrowheads) of specimen FGC-25. (d)Most fluorescent labeled cells (arrowheads) were found immediately adjacent to the host-graft interface. Inset: Verification of the graft border was obtained in each case by

viewing the interface region with darkfield optics, using the location of blood vessels (*) as landmark points. ( e ) GFAP staining of one section from this specimen illustrates a high degree of fusion between host and graft at the interface (arrowheads). (f) A patch of retrogradely labeled neurons found approximately 3 mm caudal to the transplant. (g) Transverse section of the sacral spinal cord contains four faint retrogradely labeled neurons. (h)Composite drawing of 40 sections from the host sacral spinal cord 4-6 mm caudal to the transplant. Left in the figure is ipsilateral to the graft. The photograph in (g) was obtained from the region enclosed in the box. (i)Labeled dorsal root ganglion neurons ipsilateral to the transplant. Scale bars in (c), (d) and (el = 200 Fm; scale bars in (0,(g), and (i) = 100 p,m.

PHA-L injections: Patterns of axonal projections To examine further the axonal trajectories of host and graft neurons, iontophoretic injections of PHA-L were made into either the transplant or the adjacent host spinal cord (Table 5). In each case, the injection was easily defined by the presence of darkly filled perikarya. In some specimens, the injection site contained only a small number of labeled neurons (Fig. 7a,b). Larger injection sites extended up to 0.5 mm in diameter. Projections of transplant neurons. Of 13 recipients with PHA-L injections into the transplants, five had acceptable axonal labeling (Table 5 , top). All of these injections revealed an extensive network of intrinsic axonal projections. Axons near the injection site exhibited abundant branching and the nearby neurons were surrounded by terminal enlargements. Survey electron micrographs of one such graft showed labeled terminal boutons throughout the graft neuropil (Jakeman and Reier, unpublished observations). Whereas the greatest density of fibers was found within the injection region, some labeled axons were also found in more distant regions of these grafts. Three general intrinsic axonal projection patterns were seen. Within graft regions containing densely packed neu-

322

Fig. 5. Identification of graft neurons projecting into the host spinal cord. (a) The interface between host (h) and transplant (t)is characterized by a marked decrease in density of labeled neurons. (b)Anti-GFAPstained section adjacent to that shown in (a). A glial scar is present along the one region of this host-graft interface (bottom half of figure); the host and graft tissues are well fused elsewhere (between arrowheads). Scale in (a) and (b) = 200 pm. ( e ) Higher magnification of

L.B. JAKEMAN AND P.J. REIER

retrogradely labeled cells within this graft. (d)Example of an occasional large graft neuron that projected into the host spinal cord. Scale in (c) and (d) = 100 pm. (e) Photo-montage of a sagittal section from specimen FGC-30 to illustrate the distribution of labeled neurons throughout a transplant following a FG injection into the host spinal cord (h). The host-graft interface is marked by a dashed line. Scale = 200 pm.

cn

5 0

20

2-3.0

(697)

FGC-13

1-2.0

>5.0

> 3.0

0-1.0

0-1.0

2-3.0

1-2.0

2-3.0

FGC-14 (212)

1-2.0

D

3.0

> 3.0

Distance from Host-Graft Interface (mm)

%I 25

-

-

-

0-

5-

10

15

20-

25

30-

0-1.0

0-1.0

Fig. 6. Individual histograms show the distribution of labeled cells within six transplants relative to the host-graft interface closest to the injection site. Each bar represents the number of neurons counted within successive 1.0 mm segments of the graft. Note that the last bar in each graph may represent less than a full millimeter of graft tissue. By discounting this bar, it appears that labeled cells in the bottom three cases are distributed evenly throughout the grafts.

0-1.01-2.02-3.03-4.04-5.0

0-1.0

:lBlL

FGC-17 (38)

DISTRIBUTION OF RETROGRADELY LABELED TRANSPLANT NEURONS FOLLOWING FLUORO-GOLD INJECTIONS INTO THE HOST SPINAL CORD

2-3.0

1-2.0

FGC-30

1-2.0

FGC-23 (8)

D

2.0

> 3.0

w

t3

w

INTRASPINAL HOST-GRAFT PROJECTIONS

325

ronal cell bodies, the labeled axons branched extensively (Figs. 7b,c). In contrast, where few perikarya were found, labeled axons remained mostly unbranched and followed a relatively straight trajectory (Fig. 7b). Finally, at the interface regions, axons often coursed parallel to the host-graft border, and some extended into the host neuropil where no cellular partition was apparent (Fig. 7d). Labeled axons could be followed into the host spinal cord in four of these PHA-L specimens. The pattern of axonal outgrowth was slightly different in each case. It should be noted that in one preparation (PHAL-81, the injection was placed in the ventromedial border of the graft and resulted in injury to fibers in the ventral white matter of the host. The appearance of these injury filled axons was distinct as these fibers were very heavily labeled and often contained large terminal swellings at the cut ends. Similar profiles were not found in any of the other animals in this group. Injections into the dorsal region of two grafts (PHAL-10, 231, yielded a dense network of labeled axons that could be traced into the adjacent dorsal horn (laminae 1-111) or dorsal myelinated fiber tracts of the host (Fig. 8a). In either case, they continued along a primarily longitudinal trajectory and could be followed as far as 4 mm from the interface. In some instances, collaterals from these dorsal fibers innervated deeper regions of the host gray matter (Fig. 8b). In addition to those fibers in the dorsal horn, labeled axons also extended across the interface where the graft tissue was apposed to the intermediate gray regions of the host (Fig. 8c). These axons could be followed for several hundred microns before terminating in the host ventral horn. Evidence for growth into adjacent motoneuron pools was obtained from specimen PHAL-15. A large injection was made near the caudal end of this graft, where numerous labeled axons could be followed across the caudal graft-host border and into the lateral motoneuron pools adjacent to the graft (Fig. 8d). Although labeled fibers were also found at the rostral pole of this graft, no labeled axons were seen within the host spinal cord near the related interface. The last specimen from this group (PHAL-32) had a small injection in the ventral region of the graft. Both rostral and caudal interfaces of this graft were poorly apposed to the host neuropil, and no labeled axons extended into the recipient spinal cord. Projections of local host neurons. Successful iontophoretic injections of PHA-L were made into the host spinal cord at 0.5-2.0 mm from the interface region in five rats (Table 5, bottom). In each case, the host spinal cord contained a dense fiber plexus throughout the gray matter (Fig. 9a,c) and long axonal projections were observed within the dorsal and ventral myelinated tracts. The presence of injured axons and resulting injury-fill labeling patterns were common features of these preparations. Axonal labeling within the graft was observed in four of the five specimens. The pattern of host fiber ingrowth was similar in three cases (PHAL-11,-21 and -33; Fig. 9a,b). In

each, dense axonal labeling was seen in the host spinal cord, with some fibers extending to the interface. From that site, however, only a few fibers ( < 20 per specimen) could be followed into the transplants where they terminated approximately 0.3-0.5 mm from the interface. Nevertheless, at the ends of some of these axons, boutonlike profiles were observed in close approximation to neuronal cell bodies. In specimen P a - 2 7 , a very large injection was made in the host spinal cord 0.5 mm from the host-graft interface. The presence of several lightly filled neuronal perikarya far from the injection site suggested that substantial retrograde transport occurred from this injection (Shu and Peterson, '88). A large number of labeled axons and cells were present within this graft, and some axons were observed coursing through segments of fusion between the host and graft tissues (Fig. 9d). Whereas the greatest concentration of these profiles was found within 1.0 mm of the host-graft junction, a small number of cells and axons were labeled in other regions of the transplant. At the other extreme was the fifth specimen (PHAL-161, which did not contain any labeled axons. In this case, an injection was placed 2.0 mm rostral to this graft, and all labeled fibers stopped abruptly at the host-graft interface (Fig. 9c).

Fig. 7. Intrinsic PHA-L labeled profiles following a small injection into transplant PHAL-8. (a)Sagittal section providing orientation. The injection was placed in the ventro-medial region of the transplant (t). See double arrowheads. Scale bar = to 500 km. (b)Enlargement from a nearby section of the same transplant. Labeled axons within fascicles exhibit little branching (arrows), whereas extensive collateralization is observed around counterstained cell bodies (arrowheads). ( c ) Within

the transplant, but farther from the injection site, labeled s o n s (e.g., arrowheads) exhibit an extensive pattern of projections. (d) Enlargement of the interface region outlined in the box in (a). Labeled axons (arrowheads) within the transplant travel parallel to the host graft interface. One axon then turns abruptly to enter the adjacent host spinal cord (h).Scales in (b), (c),and (d) = 100 pm.

Axonal projections and glial reactivity: FG and PHA-L recipients Fusion of host and graft tissues. The percentage of the host-graft interface length that was devoid of glial scar formation (see Methods) was highly variable across sections in each animal, as well as between animals (cf. Figs. 4b, 6e, 10). In the most dramatic example of this variability, the percentage of fusion at the interface in one animal ranged from 0.0% for a section taken at the medial border of the transplant to 57% at a more parasaggital level. The greatest degree of fusion was usually seen in sections where approximation had occurred between graft and host intermediate or ventral gray matter. When the values from sections of each FG case were averaged, a composite Fusion Index (FI) was obtained for the rostral and caudal interface region of each specimen. These values ranged from 8% to 56% (Tables 3,4). Although these averaged values are reported, it is important to note that they did not differ more than 2% from the values obtained by dividing the total length of host-graft fusion by the total interface length (not shown). No correlation was found between the composite FI and numbers of labeled host neurons following injections into the grafts. Nevertheless, the two recipients with the greatest number of labeled host neurons had composite FI measurements of 48-49% (Table 3). Likewise, in those animals with acceptable FG injections into the host spinal cord (n = 71, no correlation was found between the composite FI of the interface near the injection site and the number of retrogradely labeled cells in the transplants (Table 4). It is interesting that the recipient with the fewest number of labeled cells after an injection into the rostral host spinal

326

L.B. JAKEMAN AND P.J. REIER

327

INTRASPINAL HOST-GRAFT PROJECTIONS cord had only an 8% FI along the rostral interface. However, it was clear from these results that not all of the variations in host neuron ingrowth could be accounted for by the FI. For example, in case FGC-80 with an injection of FG placed into the rostral host spinal cord, the rostral FI was 47%, yet no retrogradely labeled cells were found in the transplant. A comparison of adjacent sections after staining with antibodies to GFAP and PHA-L allowed a qualitative comparison of regional aspects of glial scarring and axonal projections. From these specimens, localized regions of tissue fusion were found to correspond to those sites of axonal projections from transplant neurons into the surrounding host spinal cord (Fig. 10). In contrast, few axons penetrated a glial interface between host and graft tissue or within the transplants. Glial reactivity in adjacent host and graft tissues. In addition to the formation of a glial scar at the interface, varying amounts of astrocytic hypertrophy were observed within the transplants and the surrounding host tissue. The amount of glial reactivity was examined in the recipients with FG injections into the transplants (Table 4). There was a wide range in the density of glial staining within each of the grafts, as well as some variability in the surrounding host gray matter. In order to control for variations in staining, the ratio of grafthost glial staining was determined within each section. As described under Methods, the density was determined as the percentage of stained area within each field. Pairs of microscopic fields were chosen throughout the graft and the host spinal cord 5-10 mm rostral or caudal to the graft. The averaged ratio of graftihost glial density for these animals ranged from 1.1 to 10.0. In four of 10 recipients, the glial reactivity in the graft was significantly higher than that of the surrounding host tissue. However, there was no relationship between the values obtained for the grafthost glial density ratio and the numbers of retrogradely labeled cells in these animals.

issues about the cellular dynamics that govern axonal outgrowth in both developing and mature spinal cord neurons following injury. These issues may need to be resolved for optimal functional integration of host and graft tissues to occur under these grafting conditions.

Patterns of host-graft projections

Consistent with other transplantation paradigms (e.g., Bolam et al., '87; Walker and McAllister, '87; Fonseca et al., '88), the present results demonstrate that cells within FSC grafts establish complex and widespread networks of intrinsic fibers. This circuitry was characterized by a predominance of short distance ( < 1.0 mm) projections as well some connectivity extending the length of the transplants. As FSC grafts have thus far been shown to exhibit just one organotypic feature of the mature spinal cord (Jakeman et al., '89), it is not possible at the present time to determine whether any of these projections are an expression of short-distanceintrinsic circuitries normally seen at segmental levels (Szentagothai, '51; Scheibel and Scheibel, '69). Nevertheless, this possibility deserves consideration in evaluating the overall degree of outgrowth exhibited by the grafted neurons. Attention must also be directed to the possibility that some early developing neurons giving rise to longer propriospinal projections (Altman and Bayer, '84) may not survive transplantation due to axotomy associated with preparation of this tissue for grafting (e.g., Lieberman, '74). Some of the donor neurons were able to extend axons into the adjacent segments of the host spinal cord as seen in 16 of 25 recipients examined. These efferent axons entered the host spinal cord by penetrating regions of the host-graft interface where no intervening glial border was apparent. In the host spinal cord, these fibers were found within dorsal and ventral gray matter, where they terminated near neuronal perikarya. Both the retrograde and anterograde tracing methods indicated that such axons could extend as far as 4-5 mm from the host-graft interface, thereby confirming our previous results from a more limited neuDISCUSSION roanatomical study (Reier et al., '86). Similar distances of Three complementary neuroanatomical tracing tech- axonal outgrowth have been reported from transplants of niques were used in the present study to examine segmen- embryonic brainstem tissue into the chemically-lesioned tal axonal projection patterns between transplants of fetal adult spinal cord (Nornes et al., '83). Limited axonal projections from host to graft were also spinal cord tissue and the host spinal cord. Collectively, our results with retrograde and anterograde methods suggest seen in 14 of 28 cases studied. Injections of FG into either the establishment of an anatomical substratum that could the periphery or the center of the transplants, for example, lend to the development of a functional neuronal relay resulted in some labeled host neurons. As seen in our more across a spinal cord lesion. The relative distance and limited WGA-HRP tracing studies following transplantaamount of axonal elongation into and from these grafts was tion into either acute (Reier et al., '86) or chronic (Houl6 variable, however, and may be due to most fibers being and Reier, '88) spinal lesions, the vast majority of such restricted to the host-graft interface even when a confluent retrogradely labeled cells were located within 0.5 mm of the neuropil was present. Thus the interactions at the bound- host-graft interface. This cellular distribution is similar to ary between donor and host tissue may be even greater than that obtained with peripheral nerve (PNS) grafts to the could be unequivocally demonstrated with any of the adult spinal cord (David and Aguayo, '81; Richardson et al., tracing methods used here. In addition, the apparently '84), as well as other regions of the CNS (Benfey and limited distances of axonal elongation raises fundamental Aguayo, '82). These experiments, involving PNS-to-CNS

Fig. 8. Examples of efferent projections into the host spinal cord following PHA-L injections into transplants. (a) Interface between transplant (t) PHAL-10 and the dorsal horn of the host spinal cord. Labeled axons cross the interface and extend longitudinally within the rostral dorsal horn (hDIJ.(b) Darkfield photomicrograph illustrates labeled axons approximately 4 mm caudal to a transplant with a PHA-L injection. Axon collaterals extend into the deeper layers of the dorsal

horn (arrowheads). ( c ) Example of an axon that extended from a transplant (t) into the host intermediate gray region (h1J and branched around the host neurons. (d) Labeled axons in specimen PHAL-15 project out of the transplant and into the nearby motoneuron pool (hVJ caudal to the graft. Numerous bouton-like profiles are present within the surrounding tissue (arrowheads). Scale = 100 Frn in all figures.

Fig. 9. PHA-L labeled profiles following injections into the host spinal cord. (a) Short-distance ingrowth of host fibers across the interface (dashed line) and into the FSC transplant in specimen PHAI-11. (b) Higher power micrograph of sparse axonal ingrowth in another rat. (c) In specimen PHAL-16, no axons crossed the interface following a large PHA-L injection 2 mm from the host-graft interface. (d) Following a

similar injection in specimen PHAL-27, anterogradely filled axons and retrogradely filled cells (not shown) within the graft suggested a great deal of interaction between the two tissues. Notice the axonal profiles crossing between the two tissues in a region of fusion between host (h) and graft (t). Scale = 100 ym in (a), (c),and (d) and 50 ym in (b).

INTRASPINAL HOST-GRAFT PROJECTIONS

Fig. 10. Correspondence between axonal projections and glial scar formation at the host-graft interface. The left side of the figure contains darkfield micrographs of PHA-L stained fibers following an injection into the dorsal quadrant of transplant PHAL-23 (t). Adjacent sections stained with GFAP are on the right. (a,b)Dense scar separates host from graft in the most lateral regions, and few fibers cross the interface into the host tissue in this region. Note that very few labeled axons cross a smaller glial partition within the transplant itself (*I. (c,d) In

329

this pair, fibers extend from thegraft into the host dorsally, corresponding to a region of fusion between host and graft tissue (arrowheads). Note that few axons cross more ventrally, where a dense glial scar is present. (e,D In the most medial region of the graft, fibers extended into the host spinal cord across the only part of the interface where a break was found in the corresponding glial scar stain. Scale = 200 km in all figures.

330

L.B. JAKEMAN AND P.J. REIER

grafts, suggested that adult neurons axotomized close to the cell body are more likely to extend a process into the PNS tissue; similar principles may influence axonal regeneration or sprouting of host axons into FSC transplants. At this time, we cannot exclude the possibility that some of the cells interpreted as retrogradely labeled host neurons were donor cells that had migrated into the host neuropil. Recent descriptions of intracerebral transplants have illustrated the presence of grafted neurons as far as 0.5 mm away from the site of transplantation in both newborn and adult recipients (McConnell, '85; Finsen and Zimmer, '86). Furthermore, Privat et al. ('86) have reported the migration of transplanted neurons in the adult rat spinal cord as far as 15 mm from brainstem suspension grafts placed below a complete spinal transection. Whether a similar mobility of fetal neurons occurs in the acutely injured adult spinal cord has not yet been determined. This issue, however, should be testable by combining prelabeling of the grafted neurons (Lindsay et al., '87) with a double-labeling strategy in the recipient (Wictorin et al., '88). Following larger FG injections, retrogradely labeled host neurons were also found as far as 6 mm from the host-graft border. In contrast to the ingrowth from neurons next to the transplant, these more distant host cells were labeled only when the tracer injections into the grafts were positioned near the corresponding interface (see Table 3). Together with the anterograde findings, this observation suggested that the distance of ingrowth from host intraspinal neurons located farther than 1mm from the host-graft interface is limited. This is similar to the pattern of axonal projections made be brainstem serotoninergic fibers as visualized by immunocytochemical staining (Reier et al., '86). These short axonal projections ( I0.5 mm) may represent a local sprouting response of injured or uninjured axons. In three recipients with FG injections restricted to the transplants, we also observed labeled dorsal root ganglion cells. The ingrowth of host primary afferent fibers was demonstrated in a previous study in which transected dorsal roots were intentionally placed in direct apposition to developing FSC transplants at the time of grafting (Tessler et al., '88). Since the dorsal roots were not intentionally inserted into the grafts in the present study, the observation of FG- labeled DRG neurons indicates that the ingrowth of host primary afferent fibers has occurred either from juxtaposed dorsal rootlets or from regeneration or sprouting of axons within the host spinal cord (Houle and Reier, '89).

Variability in the degree of labeling Examination of GFAP- and Nissl- stained sections revealed regions of intimate fusion between host and graft tissues in nearly all specimens. Apart from a few exceptional cases, however, the quantity of host-graft interactions seen with the neuroanatomical tracing methods was more limited than that expected based upon the extensive interdigitation of processes seen in this and previous studies with various techniques (Reier et al., '85, '86; Houle and Reier, '88). This type of disparity has also been described in other CNS grafting models (Lund and Harvey, '81; Pritzel et al., '86; Wictorin et al., '88). As noted above, the limited elongation of host and graft axons may reflect a feature of many neurons present within the spinal cord that are normally involved in establishing short-range segmental circuitries. Some limitations of the

neuroanatomical tracers, however, may have also contributed to results. For instance, the possibility of unique axonal transport characteristics of neurons within the host and graft tissues must be considered as the transport of some or all of these tracers may be altered following axotomy, during degeneration (Feringa et al., '88, '89), or after denervation (Berkeley et al., '89). Differences in sensitivity of the tracers may also complicate the quantitation of these results. For example, we observed a greater degree of retrograde labeling with FG than with HRPmGAHRP in normal rats and transplant recipients (see also Cabana and Martin, '84, Castro et al., '89), and we observed more extensive axonal labeling with PHA-L than with the HRP mixtures. Consideration must also be given to the fact that tracing methods do not provide absolute quantitative measures of axonal projections. For example, because the host or donor neurons may establish widely distributed terminal fields, it can be argued that only a limited sample of the total axonal projections formed between host and graft tissues can be obtained even with larger deposits of tracer (see also Harvey and Lund, '81; Pritzel et al., '86). Likewise, conservative definition of an acceptable injection site (Warr et al., '81; Mesulam, '82) may also limit the identification of axonal projections around the interface region. In this study, most injections made near the host-graft interfaces had to be rejected due to the possibility of diffusion and spurious labeling. On the other hand, deposits of tracer that are confined to the center of a transplant were less likely to label the short-range afferents that were demonstrated using complementary anterograde techniques (see also Harvey and Lund, '81; Lund and Harvey, '81; Pritzel et al., '86; Clarke et al., '88; McAllister et al., '89).

Issues related to axonal growth Apart from the above considerations, the degree of axonal growth inferred from these neuroanatomical results may indeed reflect the impact of various intrinsic and extrinsic mechanisms regulating axonal elongation in this setting. In this regard, the outgrowth of s o n s from FSC grafts seemed to occur to a greater extent than the ingrowth of host fibers. Similar observations have been made in other transplantation models in the adult CNS (Oblinger and Das, '82; McLoon and Lund, '83; Raisman and Ebner, '83; Freund et al., '85). The more robust outgrowth from donor neurons may simply reflect the greater metabolic capacity of developing cells. The maturity of the host tissue can also contribute to the extent of axonal interactions, as axonal ingrowth in both intracerebral (Lund and Harvey, '81; McLoon and Lund, '83) and intraspinal grafts (Bregman et al., '89) is more extensive when transplantation is done in newborn, rather than adult, recipients. There is also some recent evidence that different types of mature neuron may exhibit contrasting degrees of axonal elongation into fetal neural transplants (Pritzel et al., '86; Nothias et al., '88; Wictorin et al., '88; Doucet et al., '89). Although the basis for these differences is unknown, it appears from some of the present results and from other observations (Tessler et al., '88; Houle and Reier, '89) that there is a much greater degree of ingrowth exhibited by some populations of primary afferent fibers than by fibers derived from segmental or supraspinal sources. In addition to variability in the growth potential of neurons, immature and mature astrocytes may also influence host-graft integration. For example, astrocytes within

INTRASPINAL HOST-GRAFT PROJECTIONS prelabeled embryonic graft tissue have been reported to migrate into the host CNS (Lindsay and Raisman, '84; Goldberg and Bernstein, '87). These cells may thus serve to guide axons from the grafts into the host spinal cord (e.g., Rakic, '71; Smith et al., '86). Results of the present study, however, suggest that the extent of such axonal guidance may still be limited by other factors. Astroglial elements within the mature CNS undergo a response to injury (i.e., gliosis), which includes the hypertrophy of their processes and an increase in the production of GFAP (Nathaniel and Nathaniel, '81; Eng, '88). The formation of a glial scar at the interface may prevent the development of axonal projections between host and graft tissues (Azmitia and Whitaker, '83; Das, '83; Raisman and Ebner, '83; Kriiger et al., '86; Reier, '86; Reier et al., '88; Reier and Houle, '88). In this study, an influence of gliosis on axonal outgrowth from the grafts was suggested by qualitative observations of PHA-L and HRP labeled fibers. Labeled axons did not appear to travel through interfaces exhibiting dense gliosis, yet they rarely formed club-like endings indicative of abortive axonal elongation (Ramon y Cajal, '28; Sung, '81; Liuzzi and Lasek, '87). Instead, their axonal trajectories paralleled the alignment of the scar, and they crossed the interface at points of discontinuity. The astrocytes may thus provide a favorable substratum for axonal outgrowth, yet the orientation of glial processes within a scar may not be compatible with extension through the host-graft interface (Reier et al., '89). The complex timing of graft development and the histopathological changes following spinal cord injury (e.g., evolving gliosis within degenerating white matter, see Reier, '86) may also regulate the degree of host-graft axonal interactions. Though many components within the embryonic CNS tissue should favor axonal elongation (e.g., Aquino et al., '84; Beasley and Stallcup, '87; Haun and Cunningham, '87; Liesi and Silver, '881, these elements may not be available to all host axons in the injured adult CNS at the appropriate time due t o the rapid maturation of rodent CNS tissue. For example, axons undergoing die-back or other forms of delay prior to regeneration would be unlikely to encounter a donor tissue environment that is still conducive to long-distance elongation. Coincident with developmental loss of growth-promoting conditions, the progressive expression of non-permissive molecules (e.g., Caroni and Schwab, '88; Schwab, '90) could also prevent axonal outgrowth from donor neurons located at a distance from the host spinal cord, as well as the ingrowth of host axons. Finally, the extent of axonal elongation may be governed by the formation of synapses within the graft or the denervated host environment (Bernstein and Bernstein, '71). The latter possibility is especially compelling when considering the potential for extensive neuritic interactions at interfaces where host and graft neuropils are confluent.

Implications for spinal cord reconstruction Many possible mechanisms may underlie observations of behavioral recovery following intracerebral transplantation (Dunnett and Bjorklund, '87; Bjorklund et al., '87; Gage and Buzsziki, '88). It is likely, however, that functional repair in the spinal cord will require the establishment of some type of neuronal circuitry across the lesion. The present results support the hypothesis that intraspinal transplants in adult recipients may allow the formation of a

331 neuronal relay (Reier et al., '85, '88) via the development of local axonal projections between host and graft tissues. The problem of spinal cord repair and regeneration has been traditionally approached in terms of long-range axonal elongation through a lesion site and beyond, as well as the accurate reinnervation of original target sites (Kiernan, '79; Das, '89). At face value, the predominantly short-range hodgraft connections described here do not seem to offer an immediate solution to either of these issues. Nevertheless, such host-graft connectivity closely resembles what has been exhibited by fetal CNS grafts that have also been reported to mediate functional recovery in certain brain sites (e.g., Wictorin et al., '88, '89; Doucet et al., '89). Hence, the possibility exists that some neurophysiological and/or behavioral improvement could emerge in the injured spinal cord via alternative or novel circuitries structured in part by donor neurons. Conceivably, neuritic integration at or near the host-graft interface could lead to functional, polysynaptic (e.g., short propriospinal) pathways for restored processing of descending and ascending, as well as segmental, information. In this regard, Craner et al. ('89) have demonstrated that short-distance axonal outgrowth from retinal transplants in newborn rats can stimulate physiological responses through multisynaptic pathways leading to the visual cortex. Overall, it seems that placement of FSC transplants into lesions near the lumbar or cervical enlargements might offer the greatest potential for facilitating the functional innervation of segmental circuits.

Conclusions These findings provide a useful baseline for future studies in which surgical, molecular, or pharmacological manipulations can be tested for their capacity to enhance neuritic interactions between host and graft. For example, the observations provide a point of reference to which connectivity established following intraspinal transplantation into chronic resection (Houle and Reier, '88, '89) or contusion/ compression lesions (Winialski et al., '87; Reier et al., '88) can be compared. In addition, they establish a conceptual framework for addressing the possibility of graft-mediated functional repair. Based on the identification of some local interactions between host and graft neurons, electrophysiological studies are in progress to examine the impact that transplanted neural tissue may have upon some of the physiological consequences of spinal cord damage (e.g., Thompson et al., '88, '89). Further correlative studies of the anatomical and functional integration of intraspinal transplants will address issues regarding the processes of axonal growth in the injured spinal cord and the degree to which amelioration of deficits is dependent upon reconstruction of the specific circuitries established during development.

ACKNOWLEDGMENTS The authors thank Barbara O'Steen and Minnie Smith for technical assistance, and Daniel Theele, D.V.M., for veterinary consultation and assistance. Our gratitude also goes to Drs. B. Bregman, R. Reep, and L. Ritz for editorial comments and to Dan Williams for assistance with the IBAS image analysis system. Antiserum to GFAP was supplied by Dr. Lawrence Eng. This research was supported by NIH grant NS22316 and the Mark F. Overstreet Fund for Spinal Cord Regeneration Research. L.B.J. received additional support from NIMH training grant MH15737.

L.B. JAKEMAN AND P.J. REIER

332

LITERATURE CITED Abercrombie, M. (1946) Estimation of nuclear populations from microtome sections. Anat. Rec. 94.239-247. Altman, J., and S.A. Bayer (1984) The development of the rat spinal cord. Adv. Anat. Embryol. Cell Biol. 851-164. Aquino, D., R.V. Margolis, and R.K. Margolis (1984) Immunocytochemical localization of chondroitin sulfate proteoglycan in nervous tissue. 11. Studies in developingbrain. J. Cell Biol. 99:113&1139. Azmitia, E.C., and P.M. Whitaker (1983) Formation of a glial scar following microinjection of fetal neurons into the hippocampus or midbrain of the adult rat: An immunocytochemical study. Neurosci. Lett. 38: 145-150. Beasley, L., and W. Stallcup (1987) The nerve growth factor-inducible large external (NILE) glycoprotein and neural cell adhesion molecule (NCAM) have distinct patterns of expression in the developing rat central nervous system. J. Neurosci. Res. 7:708-715. Benfey, M., and A.J. Aguayo (1982) Extensive elongation of axons from rat brain into peripheral nerve grafts. Nature 296:150-152. Berkeley, K.J., C.J. Vierck, J r . (1987) Transient reduction in retrograde labeling of diencephalic projecting neurons in monkey dorsal column nuclei following removal of their dorsal column input. In L. Pubols and S. Sessle (eds): Neurology and Neurobiology Series. Vol. 30: Effects of Injury in Trigeminal and Somatosensory Systems. New York: Alan R. Liss, Inc., pp. 429436. Bernstein, J.J., and M.E. Bernstein (1971) Axonal regeneration and formation of synapses proximal to the site of lesion following hemisection of the rat spinal cord. Exp. Neurol. 30:336-351. Bjorklund, A,, H. Nornes, and F.H. Gage (1986) Cell suspension grafts of noradrenergic locus coeruleus neurons in rat hippocampus and spinal cord: Reinnervation and transmitter turnover. Neuroscience 18:685698. Bjorklund, A., 0. Lindvall, 0. Isacson, P. Brundin, K. Wictorin, R.E. Strecker, D.J. Clarke, and S.B. Dunnett (1987) Mechanisms of action of intracerebral neural implants: Studies on nigral and striatal grafts to the lesioned striatum. TINS 1Ot509-515. Bjorklund, H., and L. Olson (1983) Astrocytic development in intraocular and intracranial cortex cerebri grafts. In P.W. Coates, R.R. Markwald, and A.D. Kenny (eds): Developingand Regenerating Vertebrate Nervous Systems. New York: Alan R. Liss, Inc., pp. 239-246. Bolam, J.P., T.F. Freund, A. Bjorklund, S.B. Dunnett, and A.D. Smith (1987) Synaptic input and local output of dopaminergic neurons in grafts that functionally reinnervate the host neostriatum. Exp. Brain Res. 68:131-146. Bregman, B.S. (1987) Spinal cord transplants permit the growth of serotoninergic axons across the site of neonatal spinal cord transection. Dev. Brain Res. 34265-279. Bregman, B.S., E. Kunkel-Bagden, M. MaAtee, and A. O’Neill (1989) Extension of the critical period for developmental plasticity of the corticospinal pathway. J. Comp. Neurol. 282:355-370. Buchanan, J.T., and H.O. Nornes (1986) Transplants of embryonic brainstem containing the locus coeruleus into spinal cord enhance the hindlimb flexion reflex in adult rats. Brain Res. 381:225-236. Cabana, T., and G.F. Martin (1984) Developmental sequence in the origin of descending spinal pathways. Studies using retrograde transport techniques in the North American opposum (Didelphis virginiana). Dev. Brain Res. 15:247-263. Caroni, P., and M.E. Schwab (1988) Two membrane protein fractions from rat central myelin with inhibitory properties for neurite growth and fibroblast spreading. J. Cell Biol. 106r1281-1288. Castro, A.J., J.C. Sorensen, N. Tonder, L. Bold, and J. Zimmer (1989) Fetal neocortical transplants grafted into cortical lesion cavities made in newborn rats receive multiple host afferents. Rest. Neurol. and Neurosci. 1:13-24. Clarke, D.J., S.B. Dunnett, 0. Isacson, D.J.S. Sirinathsinghji, and A. Bjorklund (1988) Striatal grafts in rats with unilateral neostriatal lesions. I. Ultrastructural evidence of afferent synaptic inputs from the host nigrostrital pathway. Neuroscience 24:791-801. Commissiong, J.W. (1984) Fetal locus coeruleus transplanted into the transected spinal cord of the adult rat: Some observations and implications. Neuroscience 12t839-853. Conway, B.A., H. Hultborn, 0. Kiehn, andI. Mintz (1988)Plateau potentials in a-motoneurones induces by intravenous injection of I-dopa and clonidine in the cat spinal cord. J. Physiol. 405:369-384. Craner, S.L., L.S. Radel, and R.D. Lund (1989)Light-evoked cortical activity

produced by illumination of intracranial retinal transplants: Experimental studies in rats. Exp. Neurol. 104:93-100. Crone, C., H. Hultborn, 0. Kiehn, L. Mazieres, and H. Wigstrom (1988). Maintained changes in motoneuronal excitability by short-lasting synaptic inputs in the decerebrate cat. J. Physiol. 405321-343. Das, G.D. (1983) Neural transplantation in the spinal cord of adult rats. Conditions, survival, cytology and connectivity of the transplants. J. Neurol. Sci. 62:191-210. Das, G.D. (1989) Perspectives in anatomy and pathology of paraplegia in experimental animals. Brain Res. Bull. 22t7-32. David, S., and A.J. Aguayo (1981) Axonal elongation into peripheral nervous system ‘bridges’ after central nervous system injury in adult rats. Science 214:932-933. de Olmos, J., H. Hardy, and L. Heimer (1978)The efferent connections of the main and accessory olfactory bulb formations in the adult rat. An experimental HRP study. J. Comp. Neurol. 181:213-244. Doucet, G., Y. Murata, P. Brundin, 0. Bosler, N. Mons, M. Geffard, C.C. Ouimet, and A. Bjorklund (1989) Host afferents into intrastriatal transplants of fetal ventral mesencephalon. Exp. Neurol. 106;1-19. Dunnett, S.B., and A. Bjorklund (1987) Mechanisms of function of neural grafts in the adult mammalian brain. J. Exp. Biol. 132r265-289. Eng, L.F., P.J. Reier, and J.D. Houle (1987) Astrocyte activation and fibrous gliosis: Glial fibrillary acidic protein immunostaining of astrocytes following intraspinal cord grafting of fetal CNS tissue. Prog. Brain Res. 71r439-455. Eng, L.F. (1988) Glial fibrillary acidic protein: A review of structure, function, and clinical application. In P.J. Marangos, I.C. Campbell, and R.M. Cohen (eds). Neuronal and Glial Proteins: Structure, Function and Clinical Application. New York: Academic Press, pp. 339-358. Feringa, E.R., R.L. McBride, and J.N. Pruitt 11. (1988) Loss of neurons in the red nucleus after spinal cord transection. Exp. Neurol. I00:112-120. Feringa, E.R., R.L. McBride, S. Varon, and M. Manthorpe (1989) Continuous nerve growth factor (NGF) treatment normalizes HRP transport in axotomized rubrospinal neurons. J. Neuropath. Exp. Neurol. 48:362 (Abstract). Finsen, B., and J. Zimmer (1986) Timm staining of hippocampal nerve cell bodies in the Kyoto rat. A cell marker in allo- and xenografting of rat and mouse brain tissue, revealing neuronal migration. Dev. Brain Res. 29.31-59. Fonseca, M., J. DeFelipe, and A. Fairen (1988) Local connections in transplanted and normal cerebral cortex of rats. Exp. Brain Res. 69:387-398. Foster, G.A., M. Schultzberg, F.H. Gage, A. Bjorklund, T. Hokfelt, H. Nornes, A.C. Cuello, A.A. Verhofstad, and T.J. Visser (1985) Transmitter expression and morphological development of embryonic medullary and mesencephalic raphe neurones after transplantation to the adult rat central nervous system. I. Grafts to the spinal cord. Exp. Brain Res. 60:427-444. Freund, T.F., J.P. Bolam, A. Bjrklund, U. Stenevi, S.B. Dunnett, J.F. Powell, and A.D. Smith (1985) Efferent synaptic connections of grafted dopaminergic neurons reinnervating the host neostriatum: A tyrosine hydroxylase immunocytochemical study. J. Neurosci. 5:603-616. Gage, F.H., and G. Buzsaki (1988) CNS grafting: Potential mechanisms of action. In F.J. Seil (ed): Neural Regeneration and Transplantation. New York: Alan R. Liss, Inc. Gerfen, C.R., and P.E. Sawchenko (1984) An anterograde neuroanatomical tracing method that shows the detailed morphology of neurons, their axons and terminals: Immunohistochemical localization of an axonally transported plant lectin, Phaseolus vulgaris leucoagglutinin (PHA-L). Brain Res. 290:219-238. Goldberg, W.J., and J.J. Bernstein (1987) Transplant-derived astrocytes migrate into host lumbar and cervical spinal cord after implantation of E l 4 fetal cerebral cortex into adult thoracic spinal cord. J. Neurosci. Res. 17:391-403. Grillner, S. (1973)Locomotion in the spinal cat. In R.B. Stein, K.G. Pearson, and S.R. Smith (eds.): Control of posture and locomotion. New York: Plenum Press, pp. 515-535. Grillner, S. (1980) Brainstem control of motor pattern and locomotion. Neurosci. Res. Prog. Bull. 18r58-63. Harvey, A.R., and R.D. Lund (1981)Transplantation of tectal tissue in rats. 11. Distribution of host neurons which project to transplants. J. Comp. Neurol. 202:505-520. Haun, F., and T.J. Cunningham (1987) Specific neurotrophic interactions between cortical and subcortical visual structures in developing rat: In vitro studies. J. Comp. Neurol. 256:561-569.

INTRASPINAL HOST-GRAFT PROJECTIONS Houle, J.D. and P.J. Reier (1988) Transplantation of fetal spinal cord tissue into the chronically injured adult rat spinal cord. J. Comp. Neurol. 269t535-547. Houle, J.D., and P.J. Reier (1989) Regrowth of calcitonin gene-related peptide immunoreactive axons from the chronically injured rat spinal cord into fetal spinal cord tissue transplants. Neurosci. Lett. 103,253258. Hounsgaard, J., H. Hultborn, B. Jespersen, and 0. Kiehn (1988) Bistability of a-motoneurons in the decerebrate cat and in the acute spinal cat after intravenous 5-hydroxytryptophan.J. Physiol. 405t345-367. Jakeman, L.B., and P.J. Reier (1988) Axonal projections between fetal spinal cord transplants and the adult rat spinal cord. SOC.Neurosci. Abst. 14t429. Jakeman, L.B., P.J. Reier, B.S. Bregman, E.B. Wade, M. Dailey, R.J. Kastner, B.T. Himes, and A. Tessler (1989) Differentiation of substantia gelatinosa-like regions in intraspinal and intracerebral transplants of embryonic spinal cord tissue in the rat. Exp. Neurol. 103t17-33. Kiernan, J.A. (1979) Hypotheses concerned with axonal regeneration in the mammalian nervous system. Biol. Rev. 54:155-197. Kriiger, S., J. Sievers, C. Hansen, M. Sadler, and M. Berry (1986) Three morphologically distinct types of interface develop between adult host and fetal brain transplants: Implications for scar formation in the adult central nervous system. J. Comp. Neurol. 249t103-116. Kunkel-Bagden, E., and B.S. Bregman (1989) Transplants alter the development of sensorimotor function after neonatal spinal cord damage. Exp. Brain Res. 81:25-34. Lieberman, A.R. (1974) Some factors affecting retrograde neuronal responses to axonal lesions. In R. Bellairs and E.G. Gray (eds): Essays on the Nervous System. Oxford: Clarendon Press, pp. 71-105. Liesi, P., and J. Silver (1988)Is astrocyte laminin involved in axon guidance in the mammalian CNS? Devel. Biol. 13Ot774-785. Lindsay, R.M., and G. Raisman (1984) An autoradiographic study of neuronal development, vascularization and glial cell migration from hippocampal transplants labeled in intermediate explant culture. Neuroscience 12:5 13-530. Lindsay, R.M., G. Raisman, and P.J. Seeley (1987) Intracerebral transplantation of cultured neurons after reaggregation in a plasma clot. Neuroscience 21:685-698. Liuzzi, F.J., and R.J. Lasek (1987) Astrocytes block axonal regeneration in mammals by activating the physiological stop pathway. Science 237t642645. Lund, R.D., and A.R. Harvey (1981) Transplantation of tectal tissue in rats. I. Organization of transplants and pattern of distribution of host afferents within them. J. Comp. Neurol. 201t191-209. McAllister, J.P., S.R. Cober, E.R. Schaible, and P.D. Walker (1989) Minimal connectivity between six month neostriatal transplants and the host substantia nigra. Brain Res. 476t345-350. McConnell, S.K. (1985) Migration and differentiation of cerebral cortical neurons after transplantation into the brains of ferrets. Science229: 126& 1271. McLoon, S.C., and R.D. Lund (1983) Development of fetal retina, tectum, and cortex transplanted to the superior colliculus of adult rats. J. Comp. Neurol. 217t376-389. Mesulam, M.-M. (1982) Tracing Neuronal Connections with Horseradish Peroxidase. New York: John Wiley & Sons. Moorman, S.J., J.R. Whalen, and H.O. Nornes (1990) A neurotransmitter specific functional recovery mediated by fetal implants in the lesioned spinal cord of the rat. Brain Res. 508t194-198. Nathaniel, E.J.H., and D.R. Nathaniel (1981) The reactive astrocyte. Adv. Cell. Neurobiol. 2.249-301. Nornes, H., A. Bjorklund, and U. Stenevi (1983) Reinnervation of the denervated adult spinal cord of rats by intraspinal transplants of embryonic brain stem neurons. Cell Tissue Res. 230:15-35. Nornes, H., A. Bjorklund, and U. Stenevi (1984) Transplantation strategies in spinal cord regeneration. I n J. Sladek and D. Gash (eds): Neural Transplants New York Plenum Press, pp. 407421. Nothias, F., B. Onteniente, M. Geffard, and M. Peschanski (1988) Rapid growth of host aRerents into fetal thalamic transplants. Brain Res. 463t341-345. Nygren, L.-G., L.Olson, and A. Seiger (1977) Monoaminergic reinnervation of the transected spinal cord by homologous fetal brain grafts. Brain Res. 129:221-235. Oblinger, M.M., and G.D. Das (1982) Connectivity of neuronal transplants in

333 adult rats: Analysis of afferents and efferents of neocortical transplants in the cerebeller hemisphere. Brain Res. 249:31-49. Pritzel, M., 0. Isacson, P. Brundin, L. Wicklund, and A. Bjorklund (1986) Afferent and efferent connections of striatal grafts implanted into the ibotenic acid lesioned neostriatum in adult rats. Exp. Brain Res. 65t112-126. Privat, A., H. Mansour, N. Rajaofetra, and M. Geffard (1989) Intraspinal transplants of serotoninergic neurons in the adult rat. Brain Res. Bull. 22t123-129. Privat, A,, H. Mansour, A. Pavy, M. Geffard, and F. Sandillon (1986) Transplantation of dissociated foetal serotonin neurons into the transected spinal cord of adult rats. Neurosci. Lett. 66:61-66. Raisman, G., and F.F. Ebner (1983) Mossy fibre projections into and out of hippocampal transplants. Neuroscience 9: 783-801. Rakic, P. (1971) Neuron-glia relationships during granule cell migration in developing cerebellar cortex. A Golgi and electron microscopic study in Macacus rhesus. J. Comp. Neurol. 141:283-312. Ramon y Cajal, S. (1928) Degeneration and Regeneration of the Nervous System. Translated by R.M. May, trans., 1959. London: Hafner. Reier, P.J. (1985) Neural tissue grafts and repair of the injured spinal cord. Neuropathol. Appl. Neurobiol. 11:81-104. Reier, P.J. (1986) Gliosis Following CNS Injury: The anatomy of astrocytic scars and their influences on axonal regeneration. In S. Federoff (ed): Astrocytes, Vol. 3. New York: Academic Press, pp. 263-324. Reier, P.J., and J.D. Houle (1988) The glial scar: Its bearing on axonal elongation and transplantation approaches to CNS repair. In S.G. Waxman (ed): Physiological Basis for Functional Recovery in Neurological Disease. New York: Raven Press, pp. 87-138. Reier, P.J., M.J. Perlow, and L. Guth (1983) Development of embryonic spinal cord transplants in the rat. Dev. Brain Res. 10:201-219. Reier, P.J., B.S. Bregman, and J.R. Wujek (1985)Intraspinal transplants of embryonic spinal cord tissue in adult and neonatal rats: Evidence for topographical differentiation and axonal interactions with the host CNS. In A. Bjorklund, and U. Stenevi (eds): Neural Grafting in the Mammalian CNS. Amsterdam: Elsevier, pp. 257-263. Reier, P.J., B.S. Bregman, and J.R. Wujek (1986) Intraspinal transplantation of embryonic spinal cord tissue in neonatal and adult rats. J. Comp. Neurol. 247.275-296. Reier, P.J., J.D. Houle, L.B. Jakeman, D. Winialski, and A. Tessler (1988) Transplantation of fetal spinal cord tissue into acute and chronic hemisection and contusion lesions of the adult rat spinal cord. In D.M. Gash, and J.R. Sladek (eds): Progress in Brain Research, Vol. 78. Amsterdam: Elsevier, pp. 173-179. Reier, P.J., L.F. Eng, and L. Jakeman (1989) Reactive astrocyte and axonal outgrowth in the injured CNS: Is gliosis really an impediment to regeneration? In F.J. Seil (ed): Neural Regeneration and Transplantation. New York Alan R. Liss, Inc., pp. 183-210. Richardson, P.M., V.M.K. Issa, and A.J. Aguayo (1984) Regeneration of long spinal axons in the rat. J. Neurocytol. 13tl65-182. Scheibel, M.E., and A.B. Scheibel (1969) A structural analysis of spinal interneurons and renshaw cells. In M.A.B. Brazier (ed): The Interneuron. UCLA Forum Med. Sci. No. 11. Los Angeles: Univ. of California Press, pp. 159-208. Schmued, L.C., and J.H. Fallon (1986) Fluoro-Gold: A new fluorescent retrograde axonal tracer with numerous unique properties. Brain Res. 377: 147-154. Schwab, M.E. (1990) Myelin-associated inhibitors of neurite growth. J. Exp. Neurol. 10952-5. Shik, M.L., and G.M. Orlovsky (1976) Neurophysiology of locomotor automatism. Physiol. Rev. 56t465-501. Shu, S.Y., and G.M. Peterson (1988) Anterograde and retrograde axonal transport of Phaseolus vulgaris leucoagglutinin (PHA-L) from the globus pallidus to the striatum of the rat. J. Neurosci. Meth. 25175-180. Smith, G., R. Miller, and J. Silver (1986) The changing role of the forebrain astrocytes during development, regeneration, and induced regeneration upon transplantation. J. Comp. Neurol. 251:23-43. Spence, J.T., J.W. Cotton, B.J. Underwood, and C.P. Duncan (1983) Elementary Statistics. 4th ed. Englewood Cliffs, NJ: Prentice-Hall, Inc. Sung, J.H. (1981) Tangled masses of regenerated central nerve fibers (non-myelinated central neuromas) in the central nervous system. J. Neuropath. Exp. Neurol. 403345457, Szentagothai, J. (1951) Short propriospinal neurons and intrinsic connections of the spinal gray matter. Acta Morph. Acad. Sci. Hung. 1t81-94. Tessler, A,, B.T. Himes, J.D. Houle, and P.J. Reier (1988) Regeneration of

334 adult dorsal root axons into transplants of embryonic spinal cord. J. Comp. Neurol. 270:537-548. Thompson, F.J., P.J. Reier, G.W. Schrimsher, L.B. Jakeman, D.L. Winialski, and C. Lucas (1988). Reflex physiology and correlated motor perforNeurosci. mance in rats with mild contusion spinal cord injury. SOC. Abst. 14(2):1002. Thompson, F.J., P.J. Reier, G.W. Schrimsher, L.B. Jakeman, D. Winialski, C. Lucas, and L.R. Ray (1989) Early changes in the control of primary afferent excitability following contusion injury of the rodent spinal cord. SOC. Neurosci. Abst. 15(2):1241. Walker, P.D., and J.P. McAllister (1987) Minimal connectivity between neostriatal transplants and the host brain. Brain Res. 4 2 5 3 5 4 4 . Warr, W.B., J.S. de Olmos, and L. Heimer (1981) Horseradish Peroxidase.

L.B. JAKEMAN AND P.J. REIER The Basic Procedure. In L. Heimer and M.J. RoBards feds): Neuroanatomical Tract-Tracing Methods. New York Plenum Press, pp. 207-262. Wictorin, K., 0. Isacson, W. Fischer, F. Nothias, M. Peschanski, and A. Bjorklund (1988) Connectivity of striatal grafts implanted into the ibotenic acid-lesioned striatum. I. Subcortical afferents. Neuroscience 27:547-562. Wictorin, K., R.B. Simerly, 0. Isacson, L.W. Swanson, and A. Bjorklund (1989) Connectivity of striatal grafts implanted into the ibotenic acidlesioned striatum-111. Efferent projecting graft neurons and their relation to host afferents within the grafts. Neuroscience 30:313-330. Winialski, D., J.D. Houl6, L.B. Jakeman, and P.J. Reier (1987) Transplantation of fetal rat spinal cord into longstanding contusion injuries of adult rat spinal cord. SOC.Neurosci. Abst. 13:751.