Selective innervation of embryonic hippocampal transplants by adult host dentate granule cell axons

Selective innervation of embryonic hippocampal transplants by adult host dentate granule cell axons

0306-4522/91 $3.00+0.00 Neuroscience Vol. 41,No. 2/3,pp.713-727,1991 Printed in Great Britain Pergamon Press plc 0 1991 IBRO SELECTIVE INNERVATION ...

12MB Sizes 13 Downloads 52 Views

0306-4522/91 $3.00+0.00

Neuroscience Vol. 41,No. 2/3,pp.713-727,1991 Printed in Great Britain

Pergamon Press plc 0 1991 IBRO

SELECTIVE INNERVATION OF EMBRYONIC HIPPOCAMPAL TRANSPLANTS BY ADULT HOST DENTATE GRANULE CELL AXONS P. M. FIELD,* P. J. SEELEY,* M. FROTSCHER~and G. RAISMAN*$ *Norman and Sadie Lee Research Centre, Laboratory of Neurobiology, National Institute for Medical Research, London NW7 lAA, U.K. TInstitute of Anatomy, University of Freiburg, Albertstr. 17, D-7800, Freiburg,

F.R.G.

Abstract-Fragments containing different cytoarchitectonic fields were dissected out of late embryonic rat hippocampal primordia and transplanted into the hippocampus or septum of adult syngeneic hosts. Field CA3 transplants contained clusters of large, angular (pyramidal) cell bodies surrounded by a radiating corona of dendrites. These cells stained selectively with our monoclonal antibody Py, and a proportion were labelled by [‘Hlthymidine administered on the 15th day of embryonic life. Field CA1 transplants contained smaller, angular, @-negative cells, which formed elongated laminae rather than globular clusters. The ability of the host dentate granule cells to project to the transplants was examined by (1) the Timm stain for mossy fibres, (2) electron microscopy of Golgi-impregnated CA3 pyramidal neurons in the transplants, and (3) quantitative electron microscopic assessment of the proportions of large mossy fibre terminals in the synaptic population of the transplants. The Timm stain showed that CA3 transplants received a projection from host dentate granule cells when the transplants were placed in direct contact with the axons in the host mossy fibre pathway. As in the normal host field CA3, the ingrowing mossy fibres terminated selectively on the juxtacellular regions of the dendritic tree and ignored the major part of the dendrites in the radiating corona. The electron micrographs showed that within this territory the host mossy fibres formed synaptic terminals with all the complex features typical of normal mossy fibres, and were presynaptic to complex spines arising from the juxtacellular region of Golgi-impregnated donor CA3 pyramidal cells. The quantitative electron microscopic study demonstrated that the mossy fibre-innervated juxtacellular regions of the field CA3 transplants had up to 20% of the normal density of mossy fibre synapses found in the stratum lucidum of field CA3 in sifu. CA3 transplants which were placed in the septum, remote from the host mossy fibres, had either trivial numbers of mossy fibre synapses or none. This confirmed that the abundant mossy fibre terminals in the intrahippocampal CA3 transplants were of host origin, and not due to donor dentate granule cells inadvertently included in the grafts. The selectivity of the host dentate projection for field CA3 transplants was demonstrated by the observation that CA1 transplants in the same locations received only slight mossy fibre projections in the Timm stain, and in electron micrographs their synaptic population had only insignificant numbers of large mossy fibre terminals. We have used the dentate granule+CA3 pyramidal cell system of the hippocampus to show that embryonic transplants containing a selected donor cell type can receive a correct and highly differentiated synaptic innervation arising from the appropriate host cell type, and terminating on the correct part of the postsynaptic surface. This specificity is expressed even when the spatial and temporal patterns of the confrontation are completely abnormal, and when the CA3 pyramidal cells have never been previously exposed to dentate granule cell axons.

Transplantation of embryonic neural cells into adult brain provides an experimental approach to study repair processes in the central nervous system. Much work has focussed on the growth of transplant axons into host brain (e.g. Refs 4, 46 and 47) and their formation of synaptic connections with host neurons (e.g. Refs 8 and 15). Until recently, there have been fewer studies of host innervation of transplant cells 6,7,2’,22,30,38,39,4’.45 and they suggest quite a complex situation. In some cases9.‘2,13,24.27 it has been shown that particular afferent systems may have a selective

advantage; in others, 45 the postsynaptic distribution of host-to-transplant projections has been found to be abnormal. The ability of host axons to innervate transplants is also affected by host lesions that alter electrical activity,‘2,‘3 remove host targets,30,42.43 or possibly competing afferents.24 In the normal hippocampus, dentate granule cell axons project selectively to the juxtacellular region of the apical dendrites of the large pyramidal neurons of the adjacent field CA3, but not to the more distant, smaller pyramids of field CA 1. This selective relationship appears to be maintained in host-to-transplant projections. where our previous light microscopic studies3’,32 using the Timm stain provided evidence that adult host dentate granule cells innervate transplanted embryonic CA3 pyramidal neurons. We

$To whom

correspondence should be addressed. El5 etc, the 15th day of embryonic life (the day on which vaginal plugs were found is taken as EO); PBS, phosphate-buffered saline.

Abbreviations:

713

714

P. M. FIELD PI ul.

found that hippocampal transplants containing field CA3 pyramids but devoid of dentate granule cells only received host mossy fibres if they were placed in direct contact with the host mossy fibre system, and not when they lay in other parts of the hippocampus or in the septum. This implied that the projection was of host origin. The mossy fibres terminated as large, Timm-positive structures in the immediate perikaryal region of the transplanted CA3 pyramidal cell clusters, and left the outer corona of radiating dendrites free, thus mirroring the normal juxtacellular mossy fibre distribution. Large Timm-positive granules were not found in areas of the transplants containing CA I pyramids, even when the transpiants were in contact with the host mossy fibre system, and adjacent, CA3containing regions of the transplants received prominent projections. The material, however, had limitations. The Timm stain did not provide evidence for formation 01 differentiated synaptic contacts between host and transplant cells and the interpretation of the field CAI data was complicated by the background granularity. Further, our hippocampal dissections did not select specific cytoarchitectonic fields, so that the cellular constitution of the transplants was variable, making it impossible to exclude that a given

CAScontaining transplant might also have included some of its own dentate granule cells which could have generated intratranspIant mossy fibres, thus giving rise to a positive Timm stain in the absence of host-to-transplant projections. The present study was designed to deal with three questions. (1) Do mossy fibres of host origin innervate transplanted hippocampi? (2) If so, how f&r do such host-to-transplant projections resemble normal synapses? (3) To what extent is the specificity for field CA3 targets maintained? To simplify the analysis of connections, we have used transplants taken from individual hippocampai fields, and have used the comparison of field CA3 and field CA1 transplants to investigate the specificity of formation of mossy fibre projections. We have used electron microscopy to identify mossy fibre synaptic terminals, and the Colgi-electron microscopy technique’“.‘“.ZO to identify the postsynaptic neurons. To allow for the effects of inadvertent inclusion of a small number of donor dentate granuIe cells in the field CA3 transpiants, we have made a quantitative electron microscopic comparison of the numbers of mossy fibre synapses formed in field CA3 transplants placed in the host hippocampus (where the transplant has direct access to host mossy fibres), with the numbers formed in CA3 transplants placed in the

Fig. I, (a) Historesin 2- 3 pm coronal section of an E20 hippocampus to show the lines of dissection of fields CAI, CA3. the dentate gyrus (DGh) and the discarded transitionat zone (TZ). f, fimbria; v, lateral ventricle; arrow, depth of the hippo~ampa~ fissure. Scale bar = IOO~m. (b) Sharpened tungsten needle used for dissection (same magnification).

Innervation of hippocampal grafts septum fibres).

(where

they have no access

EXPERIMENTAL

to host mossy

PROCEDURES

Dissection Inbred PVG strain rats, raised at the National Institute for Medical Research, were used. For dissection of the hippocampus into its component cytoarchitectonic fields, E20 embryos (with EO defined as the day on which vaginal plugs were found) were used. The dissections were carried out under sterile conditions using a stereomicroscope with transmitted illumination. Tissue was immersed in 90% Minimum Essential Medium (Gibco No. 072-01400) with 10% heat-inactivated human placental serum in 60-mmdiameter tissue culture dishes throughout the dissection. The hippocampus was removed over its entire septotemporal extent. It was not separated from cortex along its lateral margin, rather longitudinal scalpel incisions were made so as to include about a 2-mm-wide strip of cortex (used to hold the fragments) adjacent to the hippocampus (Fig. la). Subsequent dissection was carried out using fine tungsten needles electrolytically sharpened in saturated sodium nitrite solution. These needles were extremely fine and sharp (Fig. lb), and allowed good visibility during cutting. Each hippocampus was cut into 7-8 pieces of roughly equal length by incisions transverse to the hippocampal axis. The highly curved fragments from the temporal pole were discarded. The fimbria was removed from each fragment, and the dentate anlage gently separated from field CA1 along the hippocampal fissure. The piece of hippocampus was then “unrolled” along the length of the hippocampal fissure (about 0.35 mm; as far as the arrow in Fig. la) and cuts were made so as to generate pieces of dentate gyrus (which included dentate granule cells together with hilar pyramids and other cells postsynaptic to the dentate granule cells’), and fields CA3 and CAl. The CA3/CAI division (Fig. la) was made by a radial cut starting at the depth of the hippocampal fissure (i.e. somewhat medial to the real cytoarchitectonic junction). This was done deliberately so as to ensure that no CA3 pyramids were included in the CA1 fragment, which was to serve as a control for mossy fibre specificity. However, the CA3 fragment would contain a small proportion of the most lateral CAI pyramids. The region of field CA3 adjacent to the dentate gyrus (“transitional zone”, TZ in Fig. la) was excluded in order to minimize the contamination of CA3 samples by granule cell precursors migrating across this region from the germinal zone of the lateral ventricle to the secondary dentate trigone.28.29.33 Since the individual fragments were comparatively short, they had little curvature and straight cuts with the tungsten needle served to give clean dissection of fields. Transfer was substantially speeded up by dissecting tissue on a 5-mm-thick Perspex plate that was a sliding fit to the culture dish and was drilled with separate wells to accommodate the individual field pieces. With practice, 10 hippocampi could be dissected over 1.5-2 h. Transplantation After 16-19 h incubation at 37”C, pooled fragments from specific hippocampal fields were washed in filtered bovine plasma and transferred to the wells of a Terasaki tissue culture plate (Nunc, Denmark) for incorporation into clots (approximately 8-10 pieces of tissue per clot) by thrombin-mediated coagulation.25 Using a stereotaxically guided glass cannula, the donor tissue was transplanted into the adult host brain (positioned with the incisor bar 5 mm above the ear plugs) into either the hippocampus (4.5 mm lateral, 3.0 mm caudal, and 9.0 mm ventral to the bregma), or the septum (0.5 mm lateral, 1.7 mm rostral, and 5.8 mm ventral to the bregma). The results are based on transplants of pooled fragments of dissected hippocampal fields in 17 host animals with survival times of 20&181 days, and from

715

a previous series of transplants of whole El8 hippocampal primordia in a further 17 host animals with survivals of 23-96 days (Raisman and Ebner, unpublished observations). Histology Light microscopy. For Timm staining,23 animals were perfused transcardially under deep terminal anaesthesia with about 500 ml of a solution of 1% sodium sulphide in sodium phosphate buffer for 10-20min. Frozen sections were cut at 20 pm, mounted and then fixed in 95% ethanol. Adjacent serial sections were stained alternately by the Nissl or Timm methods. Immunohisfochemistry. For immunohistochemistry, tissue was fixed by immersion in acetic alcohol (95% absolute ethanol, 5%.glacial acetic acid), embedded in polyester wax IBDH. Poole. U.K.) and sectioned at 7 urn. The sections _ were mounted and dried on to gelatinized slides. Nonspecific binding sites were blocked with normal goat serum [I : 30 dilution in phosphate-buffered saline (PBS) containing 1% bovine serum albumin] for 30 min prior to incubation in Py antibody solution (1: 500 dilution of ascitic fluid) at 4°C overnight.M After several washes in PBS, the sections were incubated with 1: 200 biotinylated goat-anti-mouse IgM in PBS for 30 min (Vector Laboratories, Peterborough, U.K.) and, after further washing, incubated in avidin-biotin complex (Vector Elite reagent) for 30 min. After additional washes in PBS, peroxidase was developed with diaminobenzidine for 3-5 min. Autoradiography. For autoradiography, timed-pregnant mothers were given intraperitoneal injections of lOpCi/g body weight 6-[3H]thymidine (TRK.61, Amersham International, Amersham, U.K.) in saline, at either 15 or 19 days of gestation (three animals at each age), and whole hippocampi or dissected fields were taken at E20 (as above). The tissue for autoradiography was either transplanted, or fixed in electron microscope fixative (see below), processed, embedded in LKB Historesin, and cut at 2 pm with a glass knife. Animals with transplants were allowed to survive for 68 weeks, and their brains were then removed under deep terminal anaesthesia and fixed by immersion in Carnoy’s solution, processed and embedded in paraffin wax and cut at 7 pm. Both types of section were mounted on gelatinized glass slides and processed for light microscope autoradiography by coating them in Ilford G5 emulsion, exposing for 14-30 days at 4”C, developing in Kodak D19, and lightly counterstaining with aqueous thionin. Semithin sections and electron microscopy. Rats were perfused under deep terminal anaesthesia with a mixture of I % paraformaldehyde and I % glutaraldehyde in 100 mM sodium phosphate at pH 7.2-7.4, postfixed in 1% osmium tetroxide, dehydrated and embedded in epoxy resin. Semithin sections were stained with Methylene Blue and Azur II, ultrathin sections with alcoholic uranyl acetate and lead citrate. For quantitative synaptic analysis, a map of each section was constructed on squared paper. A continuous row of sample squares (grid squares of IOOpm side) were taken across the transplants. Each square was scanned systematically and the synapses counted and classified according to their site of termination (cell somata, dendritic shafts or spines), and whether or not they had the morphology of a mossy fibre terminal. Combined Golgi-electron microscope procedure. Five animals with CA3 transplants grafted to the host hippocampus were processed for combined Golgi-electron microscope analysis. Brains were fixed as for electron microscopy (see above) and each brain was then washed in phosphate buffer. The 40- pm-thick serial Vibratome sections were collected in phosphate buffer and examined in a stereomicroscope for the presence and localization of the transplants. Transplantcontaining sections were then Go&impregnated using a modification’7~‘8~2” of the section-Golgi impregnation procedure.‘”

71h

P. M.

F‘I~LL) <‘I cd.

the Fig. 2. (a) Large field CA3 transplant, with a r lumber of CA3 pyramidal cell clusters.transecting stratum lucidum at the level of the flexure of the adult host hippocampus. Arrowheads indicate cell-free dendritic regions surrounding the perikaryal clustl XS. Semithin section. Scale bar = 200 pm. (b), (c) Details of field-specific transplants of field CA3 (b) and field CA1 (c). Arrows indicate main stem dendrites of field CA3 pyramids. Semit hin sections. Scale bars = 10 pm.

Briefly, the sections were placed on pieces of Parafiim~‘rs’ and piled on top of each other such that each section was sandwiched between two layers of Parafilm. The pile was then covered with 5% agar to re-constitute a “tissue block” of adjacent sections. These blocks were immersed in freshly prepared osmium dichromate solution (I g osmium tetroxide and 12 g potassium dichromate in 500 ml distilled water) for 4 days and, thereafter, for 2 days in 0.75% silver nitrate. After cutting away the covering agar the sections were separated and examined in the light microscope (Fig. 5b, c). Sections containing well-impregnated CA3 pyramidal cells in the transplants were removed from the Parafilm and further processed for gold-toning.‘4 Non-impregnated sections were re-assembled in a pile and impregnated for a second time. After gold-toning. the sections were dehydrated (block-stained with many1 acetate in 70% ethanot) and embedded flat in Araldite between aluminium foil and a transparent plastic foil. Cells were then examined again under the light microscope, photographed and drawn. Selected sections were re-embedded in plastic capsules. Ultrathin serial sections through the gold-toned transplant neurons were mounted on single-slot grids that were coated with formvar film and then stained with lead citrate.

RESULTS

Transplants of ,fieid CA 3. The transplanted CA3 pyramidal cell perikarya were in clusters (Fig, 2a), which were surrounded by cell-free regions in which the dendrites of the centrally-located cells extended (arrowheads in Fig. 2a). These cells had chardcteristitally large (about 20 pm, average of major and minor diameters), somewhat angular cell bodies (Fig. 2b) with abundant cytoplasm that tapered into one or a small number of wide dendrites (arrows in Fig. 2b). Thus, in size and shape they were comparable to field CA3 pyramids in the normal adult hippocampus in situ. To confirm the effectiveness of selection of field CA3 pyramids, we used the monoclonal antibody, Py, raised by Woodhams et a1.44 and shown to be selective in the normal adult hippocampus for field CA3 pyramids, but not field CA1 pyramids. We found that this selectivity was also expressed in

Innervation

of hippocampal

Fig. 3. (a) Large CA3 pyramidal cells rmmunolabelled with donor. survival 12 weeks in adult host hippocampus. Scale heavy nuclear labelling of a cluster of iarge neuronal nuclei administered on EIS. transplant taken at E20. survival 8 bar = 20 fern, transplants: immunohistochemically. the large transplanted CA3 pyramidal neurons (Fig. 3a). like the host CA3 pyramidal cells, had cytoplasmic Py immunoreactivity, extending as fine filamentous material into the dendrites. The smaller pyramids of the field CA I transplants were unstained (see below). Further confirmation of the CA3 identification was provided by the birthdays of the large pyramids of the field CA3 transplants: autoradiographicaily. the

grafts

717

4, m a field CA3 transplant from an E20 bar = 20pm. (b) Autoradiograph showing in a field CA3 transplant. [‘Hj~ymidinc, weeks in adult host htppocampus. Scale

clusters of pyramidal neurons in the field CA3 transplants had appreciable numbers of nuclei (Fig 3b) labelled by ~3~]thymidine injected at EiS (the normal time of generation of CA3 pyramids).3’~3’ Transpfmts of field CA I. These (Fig. 2c) had principal neurons which were significantly smaller (about 16 Ltm, average of major and minor diameters) than those of CA3 transplants, had less abundant cytoplasm. and angular cell bodies that also tapered

71X

P. M. FIELD rl d

into dendrites. Thus the shape and size of the transplanted CA1 pyramids resembled those of field CA1 pyramids in the normal adult hippocampus in situ.* In contrast to the clustered CA3 pyramids, the CA1 transplant neurons tended to be arranged in straight or curving rows that resembled their normal laminar arrangement. CA1 pyramids were devoid of Py immunostaining, although &-positive interneurons were found in these grafts. In the samples available we rarely found CA1 pyramidal cell nuclei labelled with [3H]thymidine administered on E 15. Transplants of dentate gyrus. These were used as a control to show that mossy fibre synapses can form in the septum (see below). The transplants develop small, rounded granule cells with little cytoplasm, in characteristic shell-like c1usters.20.3’.32.36~37.4’ The dissection includes many hilar pyramids and other hilar cells which are normal postsynaptic targets for the dentate granule cell axons.’ A proportion of the dentate granule cell nuclei were labelled by [3H]thymidine administered on E19, but none were found after El5 injections. Transplants with both field CA3 and CA 1. Data from Timm-stained material have been reported previously (Refs 31 and 32; see Introduction), and will not be further described here except for one particularly informative transplant. In the case shown in Fig. 4, the transplant (T in Fig. 4a), derived from an El8 hippocampal primordium, contained one zone with a ring of small CA1 pyramids (Tl in Fig. 4b, c), and an adjacent zone with a cluster of large CA3 pyramids (T3 in Fig. 4b, c). This transplant lay in the host hippocampus with the transplant CA1 zone in direct contact with (and cutting) the host mossy fibre system (arrow in Fig. 4a; H3 in Fig. 4b, c), which is intensely black in the Timm stain (Fig. 4a, c). Serial sections showed that the host mossy fibres give rise to bands of fine Timm-positive fibres (arrowheads in Fig. 4c) which extend through the transplant from the interface with the cut host mossy fibre system, pass along the medial edge of the CAl-containing transplant region (Tl), and terminate as a dense, large granular layer around the cell body clusters in the zone of transplant CA3 pyramids (T3). Whilst it *The sizes of the transplanted CA3 and CA1 pyramids are comparable to those in situ (e.g. Ref. IO) and represent a major volumetric difference-the somata of field CA3 pyramids are double the volume of those of the CAI pyramids.

cannot be asserted that the transplanted CAI pyramids (Tl) were completely devoid of projection, any such projection must have been slight. Electron microscopy Electron microscopy of Golgi-impregnated material. The combined Golgielectron microscopy technique was used to look for ultrastructurally identifiable mossy fibre terminals on the complex juxtacellular dendritic spines of neurons which had previously been positively identified as CA3 pyramids by Golgi impregnation. Figure 5a shows an unstained 40pm Vibratome section just before blocking for Golgi impregnation. In this case a CA3 field transplant has penetrated the host hippocampus, and lies within the host mossy fibre system, which it has totally transected at this level-a situation in which we regularly observed that mossy fibres invaded the transplant (see above). Golgi-stained neurons from this and other comparable field CA3 transplants showed the presence of typical CA3 pyramidal neurons (Figs 5b, c and 6a). The cells tended towards a triangular cell body and had multiple dendritic arbors. Often a main “apical” dendrite could be recognized (Figs 5b and 6a). The proximal segments of the main dendritic shafts bore large spiny excrescences (open arrow in Fig. 6a). The peripheral parts of the dendrites (e.g. arrowheads in Fig. 6a) were densely covered with smaller, simple spines. In some cases these impregnated CA3 pyramidal neurons had a less regular, multipolar appearance (Fig. 5c), thereby resembling the hilar cells. The axons of the transplant cells were rarely impregnated, and could only be followed for short distances from the cell body. We have carefully checked the complete series of sections through the field CA3 transplants for the presence of dentate granule cells. In the present transplants, we were unable to find Golgi-impregnated granule cells, although such cells would have been readily recognizable from their small cell bodies, characteristic unipolar dendritic arbors and lack of basal dendrites.20 Moreover, we never observed the characteristically dense layer of dentate granule cells which is well maintained in transplants of dentate gyrus tissue,20,3'.32,36,37.41

Two well-impregnated, gold-toned, CA3-type transplant pyramidal cells, which were not overlapped by impregnated dendrites of neighbouring

Fig. 4. (a) A transplant (T) lying in the host hippocampus, intersecting the host mossy tibre pathway (arrow). The transplant separates the host field CA3 (H3) ventrally, from the host field CA1 (HI), dorsally. Timm stain. Scale bar = 500 pm. (b) Nissl stain of an enlarged view of the transplant in (a), showing a cluster of large transplant CA3 pyramids (T3) in contact dorsally with the host field CAI (HI), and a ring of smaller transplant CAI pyramids (Tl) in contact ventrally with the host field CA3 (H3). Thionin stain. Scale bar = 100pm. (c) Adjacent section to (b) with Timm stain for mossy fibres. The compact, densely stained host mossy fibre layer in the host field CA3 (H3) makes contact with the part of the transplant containing CA1 pyramids (Tl), to which it makes no significant mossy fibre projection. The upper part of the transplant (T3), containing donor field CA3 pyramids, receives two bands of mossy fibre projections (arrowheads). The host field CAI (HI) remains free. of mossy fibres. Scale bar = 100 pm.

Innervation of hip~mpal

Fig. 4.

grafts

719

720

P. M.

FrELU

et

ul.

Fig. 5. Light microscopy of Golgi-stained transplants. (a) Unstained coronal Vibratome section through the host hippocampus prior to Golgi impregnation. A transplant (T) separates the host CA3 region from the dentate gyrus, and is thus in close apposition to host mossy fibres disconnected from their normal target region. (b), (c) Examples of Golgi-impregnated pyramidal cells in CA3 transplants. H, host hippocampus. Scale bars = 500 pm (a): 100 jtm (b.c).

Fig. 6. CorreIated Iigbt and electron rni~rogra~~ of a CA3 pyramidal ceietlin the transplant shown in Fig. 5a. (a) tight micrograph after Go& impregnation and gold-toning. The open arrow indicates the large spiny excrescences on the proximal dendritic shaft shown in electron micrographs in Fig. 7. Basal dendrites (arrowheads) are densely covered with smaller spines. Asterisks indicate unstained cells which are also seen in the electron micrograph. (b) Electron micrograph of the gold-toned pyramidal cell (fine, intensely black deposits in periphery of cytoplasm), displaying an ovoid nucleus (N) lacking indentations. The cytoplasm is rich in organetles. Scale bars ==20 pm (a); 5 pm (b& NSC41;2:3--N

721

722

P. M.

FIELD rf UL

Pig. 7. Proximal dendritic segment (D, corresponding to the open arrow in Fig. 6a) of the Golgi-impregnated transplant CA3 pyramidal cell (labelled by dense granular gold deposits). Large mossy fibre boutons (mf) lie adjacent to the shaft, and make synaptic contacts (some indicated by small arrows) with numerous, large, gold-labelled spines, one of which (receiving arrowed synapse in b) can be seen to arise from the main dendritic shaft. One mossy fibre terminal (in b) is seen in contact both with spines of the Golgi-impregnated neuron and also with non-~rnp~gnat~ spines (asterisks) of another transplant neuron. Scale bars = I jlrn.

cells, were subjected to an electron microscopic analysis. Figure 6a and b shows light and electron micrographs of the cell body and proximal dendrites of one of the two cells. Several “basal” dendrites and one

main “apical” dendritic shaft emerged from the cell body. The proximal segment of the apical dendrite had large spiny excrescences and the cell body had a smooth, ovoid nucleus which was surrounded by

723

Innervation of hippocampal grafts Table 1. Quantitative analysis of the numbers of mossy fibre synapses in the different groups Group CA3 in situ CA3+sept CA3 +hipp CA1 +hipp DGh+sent

n 1 3 6 4 2

MFsyn 294 7 168 18 113

Totsyn 1571 16,152 14,856 37,847 4844

MF %

Cells

Area

Sp-Sh-Ce

19 0.04 1.1 0.05 2.3

52 221 290 286 64

600 3040 3545 7300 880

77-21-2 70-26-4 79-19-2 82-17-I 80-18-2

Groups: CA3 in situ; CA3 transplanted into the septum (sept), or the hippocampus (hipp); CA1 transplanted into hippocampus; DGh, dentate gyrus (including hilar target cells) transplanted into septum. Parameters: n, number of different animals; MFsyn, total number of mossy fibre synapses; Totsyn, total number of all kinds of synapses; MF %, percentage of all synapses of mossy fibre type; Cells, total number of neuronal perikarya in the sample; Area, total area of the sample in units of lo* pm*; Sp-Sh-Ce, percentages of synapses on dendritic spines, dendritic shafts, and cell bodies. (Standard errors are not shown, as the data are not suitable for parametric analysis.)

varying amounts of cytoplasm rich in endoplasmic reticulum. The fine-structural study focussed on the large spiny excrescences which are normally contacted by mossy fibre boutons. As shown in Fig. 7, the large spines arising (see arrowed synapse in Fig. 7b) from the proximal portion of the apical dendrite of this cell received numerous synaptic contacts from large mossy fibre terminals (mf). These were easily recognized as mossy fibre boutons from their size (up to 8 pm) and their content of numerous clear synaptic vesicles along with a few dense-core vesicles. We observed mossy fibre boutons that contacted both impregnated spines (with dense black-gold granules) of the identified cell, and non-impregnated spines (asterisks in Fig. 7b) of another neuron. Similar large mossy fibre synapses on gold-toned spines were also found on the dendrites of the other Golgiimpregnated transplant CA3 pyramidal cell. Quantitative analysis of synapses. The quantitative electron microscopic study is based on an analysis of the type and distribution of a total of 75,270 synapses in an area of 1.54 x lo6 pm2 from 16 animals in five control and experimental groups (Table 1). The neuropil of all the transplants shows considerable similarity to norma1-e.g. the constancy of the spine/shaft/cell distribution ratios for synapses. The normal adult field CA3 sample was taken from the stratum lucidum, the zone containing the proximal parts of the CA3 apical dendrites, where the mossy fibres selectively terminate, and excluded the distal dendritic layers which do not receive mossy fibres. The sample consisted of 1571 synaptic contacts, of which 294 (i.e. 19%) had sufficient of the characteristic features to be identified as mossy fibre synapses on complex dendritic spines-i.e. large presynaptic elements attached by symmetrical membrane thickenings (puncta adhaerentia) without associated synaptic vesicle clusters to an adjacent large dendritic shaft, and making multiple synaptic contacts (with vesicle clusters and asymmetrical thickenings) on to complex, branching dendritic spines arising from this parent shaft5 Samples from the three animals with field CA3 fragments transplanted into the septal region show

only seven synapses with mossy fibre characteristics in a total of 16,152 synapses-i.e. a negligible proportion (0.04%). This indicates that the field CA3 transplants on their own do not contain a significant source of mossy fibres; only a trivial number of donor dentate granule cells can have been inadvertently included with the CA3 fragments. The two cases with intraseptal transplants of donor dentate gyrus (i.e. dentate granule cells together with their hilar pyramidal and other cells postsynaptic to mossy fibres’) show that the environment of the host septal region is not inimical to the formation of mossy fibre synapses (113 in 4844-i.e. 2.3%). Samples from the six animals with field CA3 transplants into positions in the host hippocampus in contact with the host mossy fibre system have a total of 168 mossy fibre synapses in a total synaptic population of 14,856-i.e. 1.1%. While this is less than the density of mossy fibres in the stratum lucidum of field CA3 in situ, it reflects the fact that the transplant mossy fibre synapses were counted unselectively across the entire transplant. The sample therefore includes areas of distal CA3 dendrites which do not receive mossy fibre synapses, and also a variable proportion of laterally placed CA1 pyramids would have been included (see Fig. la and Experimental Procedures). In the juxtacellular areas of the intrahippocampal field CA3 transplants the density of mossy fibre synapses reached around 20% of that found in the stratum lucidum of the normal field CA3. Taken in relation to the very low numbers of mossy fibre synapses (total of seven) in the three field CA3 transplants in the septal region (Table 1), the total number of 168 mossy fibre synapses in the six intrahippocampal field CA3 transplants, even when averaged out across all the different regions, is highly significant, and indicates that the mossy fibre synapses are of host origin. That the formation of a host-to-transplant mossy fibre projection is specific to field CA3 is shown by the counts from the four animals with intrahippocampal field CA1 transplants in contact with the host mossy fibres. Here we found only 18 mossy fibre synapses

724

in a total proportion

P. M. FIEL.I)et ul. of 37,847 synapses--i.e. of only 0.05%.

again

a trivial

DISCUSSION

There is a considerable literature on transplantto-host projections.“ Two studies from our own group,4h.‘7 using the Thy- I.]/2 allelic marking system in mice, illustrate conclusions which are broadly in agreement with those of most studies. These are that the formation of transplant-to-host projections depends on (1) presence of correctly matched preand postsynaptic cell types, (2) direct contact, and (3) specific deafferentation of the host. Until recently, the formation of host-to-transplant projections has been less documented. This is largely for technical reasons--namely the less organized nature of the transplants, and the fact that here we are dealing with projections into developing tissue. The interpretation is more complex. In the case of transplant-to-host projections the donor neurons have an embryonic axon-growth capacity. but in the converse situation, where adult host axons are to innervate an embryonic graft. it may be necessary for the host fibre systems to be damaged or otherwise primed by axogenic stimuli (e.g. by removal of host targets) before they will grow into the grafts.‘4.3”.J’.4’ A number of studies on host-to-transplant projections indicate that some adult axonal systems react more readily than others, so that some projections are formed in preference to others (e.g. Refs 9. 12, 13. 24 and 27). and Ebner and colleagues showed that the formation of thalamocortical afferents to transplanted somatosensory cortex was dependent on cholinergic deafferentation,14 or activity changes caused by prior peripheral sensory deafferentation.“,” Further, whereas in the adult we can speak of removal of a specific input as deafferentation, the situation for the embryonic transplants is frequently that their neurons have never been exposed to a particular input, because it develops at a later stage than the time when the donor tissue was removed. This is almost certainly the case for the dentate inputs to the CA3 transplants of the present study. In the present study we have addressed the question of the formation of one particular type of host-to-transplant projection---that originating in the dentate granule cells. Our previous results with the Timm stain”‘.j* showed that transplanted embryonic CA3 pyramidal neurons can induct adult host dentate granule cells to form large mossy fibre boutons. This result is specific for field CA3 pyramids, since transplanted CA1 pyramidal neurons do not induce significant numbers of mossy fibre boutons. A very similar result was obtained by Zimmer and Glhwiler4* in a study where slices of dentate gyrus and hippocampal tissue from 7-day-old rats were grown together in slice culture. Dentate mossy fibres grew into slices of field CA3, provided

the slices were in direct contact. However, even if in direct contact, the dentate gyrus only rarely sent a projection to field CA1 slices, and then only a very limited one. In our host-to-transplant studies, a necessary condition for the formation of projections from adult host mossy hbres is that they make direct contact with the graft. However. the mixed CA3/CAl transplant (Fig. 4) demonstrates that the induction does not require the transplanted CA3 pyramids themselves to be at the interface with the host mossy fibre pathway; possibly the initial event may be the induction of “provisional” sprouting by which host dentate granule cell growth cones are able to explore territory for some distance before they encounter the correct CA3 targets. Our electron microscopic and Golgi-electron microscopic observations extend the results with the Timm stain. They confirm that the mossy fibre boutons do indeed represent fully differentiated large presynaptic terminals, which are attached by puncta adhaerentia to the proximal dendritic shafts of identified CA3 pyramidal neurons, and which make multiple synaptic contacts with their complex, branching spines (all precisely as in the normal hippocampal field CA3 in situ). We have no evidence for abnormal patterns of terminal distribution 01 host-to-transplant dentate projections (e.g. as reported in a study of host-to-transplant corticostriate projections4s), but it should be noted that we have studied only this one type of large mossy fibre terminal, and not, for example, the simpler terminals borne on its expansions.’ Moreover, we have concentrated on CA3 and CA1 pyramids, and have not studied the dentate granule cell projections to the hilar cells.’ These are important cautions against over-generalization, especially since the dentate granules are a known example of a type of cell whose normal adult axonal distribution does not reflect the full potential of the axons which, in other experimental situations, can be induced to make additional projections to the dendritic layer of the dentate gyrus” or to field CA1 (see below). Whilst in the previous studies”,” the residual background of the Timm stain made it impossible to exclude the presence of a minor mossy fibre projection to field CA1 transplants. or a small contamination of CA3 transplants by inclusion of some donor dentate granule cells, our present quantitative electron microscopic study helps with these points. Transplants of field CA3 located in the septal nuclei. where they had no possibility of contact with the host mossy fibre system, had only trivial numbers of mossy fibre synapses (0.04%). This confirms that our method of dissection of field CA3 does not include significant numbers of viable dentate granule cells or their precursors. Large numbers of mossy fibre synapses were however, observed in the juxtacellulat regions of intrdhippocampal field CA3 transplants placed in contact with the host mossy fibres; in these

125

Innervation of hippocampal grafts transplants the frequency of mossy fibre synapses reached about 20% of that in the normal stratum lucidum of field CA3 in situ. In the present material, mossy fibre synapses were not formed to any appreciable extent (0.05%) in field CA1 transplants. Thus, under these circumstances, we do not find the sort of mossy fibre projection to field CA1 which has been reported in various other experimental situations, such as selective destruction of field CA3 targets by kainate injections,“,” slice explant culture of neonatal hippocampal slices,48 and Cook and embryo-to-adult transplantation.3’,4 Crutcher” showed that kainate lesions of field CA3 only induced a mossy fibre projection to field CA1 in early neonatal animals; the effect disappears between the 5th and 9th postnatal days, and does not occur in the adult.26 Similarly, the slice culture studies4* and transplant studiesa including our own3’ showing projections from transplanted dentate gyrus to the host field CAl, involved embryonic or very early neonatal dentate gyrus, whereas the present study involves presenting CA3 targets to adult host mossy fibres. Possibly, therefore, the ability to project to CA1 is lost as granule cells mature; the ability to project to CA3, however, is not lost. This question apart, the selective formation of dentatohippocampal host-to-transplant mossy fibre connections to the juxtacellular complex spines of donor field CA3 pyramidal neurons follows the same rules as those governing the normal development of this pathway. The specificity of adult* dentate granule cell axons for the juxtacellular dendritic region of CA3 pyramids is maintained in transplantation situations where both the spatial and *One caveat is appropriate here. The dentate gyms is the one part of the rat brain in which it has been proved that new neurons are continually added during adult life, and that these new neurons form normal axonal projections.jga Whether the ability to form host-to-transplant projections depends on such continuing neurogenesis is an intriguing, and still open, question.

temporal relationships are totally different from those of normal development. CONCLUSIONS

The present observations not only clearly validate the effectiveness of our dissection technique for selection of the CA3 and CA1 field fragments, but they also demonstrate that whatever prior developmental events determine the future characteristic differences between field CA3 and field CA1 pyramids have already taken place by E20, the stage at which the donor tissue was taken. Despite growing up in entirely abnormal situations, and deprived of major external inputs (such as the entorhinal afferents), the late embryonic pyramidal cells develop into mature neurons whose sizes and shapes show exactly comparable field differences to those found in the normal adult hippocampus in situ. Moreover, the Pyimmunoreactive material develops selectively in the transplanted CA3 pyramids, and not in those of field CAl, exactly as in situ.” Since very few dentate granule cells have been formed by E20, and almost certainly none have grown axons to field CA3 at this time,3 our results show that prior exposure to specific afferent axons is not required for the transplanted E20 field CA3 pyramids to express those localized signals in the juxtacellular part of the dendritic membrane which are needed later to make them selectively able to induce specific terminal formation by ingrowing host dentate granule cell axons. Transplanted E20 field CA1 pyramids, on the other hand, do not exhibit this potential. Acknowledgemenls-We thank Melvyn Sherwood and Wendy Jaques-Berg for help with transplantation, Ursula Starega for histology, and Edeltraut Thielen for thinsectioning Golgi-impregnated and gold-toned transplant pyramidal cells. This study was supported by the International Spinal Research Trust and by a grant from the Deutsche Forschungsgemeinschaft (Fr620/1-4).

REFERENCES

1. Amaral D. G. (1978) A Golgi study of cell types in the hilar region of the hippocampus in the rat. J. camp. Neurol. 182, 851-914. 2. Amaral D. G. (1979) Synaptic extensions from the mossy fibers of the fascia dentata. Anat. Embryo/. 155, 241-251. 3. Amaral D. G. and Dent J. A. (1981) Development of the mossy fibers of the dentate gurus: I. A light and electron microscopic study of the mossy fibers and their expansions. J. camp. Neural. 195, 51-86. 4. Bjorklund A. and Stenevi U. (1984) Intracerebral neural implants: neuronal replacement and reconstruction of damaged circuitries. A. Rev. Neurosci. 7. 279-308. 5. Blackstad T. W. and Kjaerheim A. (1961) Special axodendritic synapses in the hippocampal cortex: electron and light microscopic studies on the layer of mossy fibers. J. camp. Neural. 117, 133-159. 6. Castro A. J., Tender N., Sunde N. A. and Zimmer J. (1988) Fetal neocortical transplants grafted to the cerebral cortex of newborn rats receive afferents from the basal forebrain, locus coeruleus and midline raphe. Expl Bruin Res. 69, 613622. I. Clarke D. J., Dunnett S. B., Isacson O., Sirinathsinghji D. J. S. and Bjorklund A. (1988) Striatal grafts in rats with unilateral neostriatal lesions-I. Ultrastructural evidence of afferent synaptic inputs from the host nigrostriatal pathway. Neuroscience 24. 79 I-802. 8. Clarke D. J., Nilsson 0. G., Brundin P. and Bjorklund A. (1990) Synaptic connections formed by grafts of different types of cholinergic neurons in the host hippocampus. Expl Neural. 107, 11-22. -.. --. 9. Clinton K. J. Jr. and Ebner F. F. (1988) Time course of neocortical graft innervation by AChE-positive fibers. J. camp. Neural. 277, 5577577.

726

P. M. FIELD Y/ d.

IO. Cook T. M. and Crutcher

K. A. (198.5) Extensive target cell loss during development results in mossy fibers in the regio formation. Deul Bruin Res. 21, I9 30. I I. Cook T. M. and Crutcher K. A. (1988) Lesion-induced CAI mossy fibers in the rat represent a neoinnervation. E.xpl Brain Res. 70, 433436. 12. Ebner F. F., Erzurumlu R. S. and Lee S. M. (1989) Peripheral nerve damage facilitates functional innervation of brain grafts in adult sensory cortex. Proc. naln. Acad. Sci. U.S.A. 86, 730-734. 13. Erzurumlu R. S. and Ebner F. F. (1988) Peripheral nerve transection induces innervation of embryonic neocortical transplants by specific thalamic fibers in adult mice. J. camp. Neural. 272, 536544. 14. Fair&n A., Peters A. and Saldhana J. (1977) A new procedure for examining impregnated neurons by light and electron microscopy. J. Neuroeytol. 6, 311 337. IS. Freund T. F., Bolam J. P., Bjorklund A., Stenevi U., Dunnett S. B., Powell J. F. and Smith A. D. (1985) Efferent synaptic connections of grafted dopaminergic neurons reinnervating the host neostriatum: a tyrosine hydroxylase immunocytochemical study. J. Neurosci. 5, 603 616. 16. Freund T. F. and Somogyi P. (1983) The section-Golgi impregnation procedure. 1. Description of the method and its combination with histochemistry after intracellular iontophoresis or retrograde transport of horseradish peroxidase. Neuroscience 9, 463-474. 17. Frotscher M., Heimrich B. and Schwegler H. (1990) Plasticity of identified neurons in slice cultures of hippocampus: a combined Golgi/electron microscopic and immunocytochemical study. In Progress in Brain Research (eds StormMathisen J., Zimmer 3. and Ottersen 0. P.), Vol. 83, pp. 323 ~339. Elsevier, Amsterdam. 18. Frotscher M. and Leranth C. (1986) The cholinergic innervation of the rat fascia dentata: identification of target structures on granule cells by combining choline acetyltransferase immunocytochemistry and Golgi impregnation. J. camp. Neural. 243, 58 70. 19. Frotscher M. and Zimmer J. (1983) Lesion-induced mossy hbers to the molecular layer of the rat fascia dentata: identification of postsynaptic granule cells by the Go&EM technique. J. camp. Neural. 215, 299-31 I. 20. Frotscher M. and Zimmer J. (1986) Intracerebral transplants of the rat fascia dentata: a Golgi,‘electron microscope study of dentate granule cells. J. camp. Neural. 246, 181~ 190. 21. Harvey A. R. and Lund R. D. (1981) Transplantation of tectal tissue in rats. II. Distribution of host neurons which project to transplants. J. camp. Neural. 202, 505~~520. 22. Harvey A. R.. Rush R. A. and Keating P. J. (1988) Cultured fetal tectal tissue grafted to the midbrain of the newhorn rat: morphology of grafts and innervation by host retinal and cortical axons. Brain Res. 462, 89-98. 23. Haug F.-M. S. (1973) Heavy metals in the brain. A light microscopic study of the rat with Timm’s sulphide silver method. Methodological considerations and cytological and regional staining patterns. Adt;. Anoi. Ernbr_vo/. CC// Rio/. 47, l-71. 24. Hohmann C. F. and Ebner F. F. (1988) Basal forebrain lesions facilitate adult host fiber mgrowth into neocortical transplants. Brain Res. 448, 53-66. 25. Lindsay R. M., Raisman G. and Seeley P. J. (1987) Intrdcerebral transplantation ofcultured neurons after reaggregation in a plasma clot. Neuroscience 21, 685 ~698. 26. Nadler J. V., Perry B. W. and Cotman C. W. (1980) Selective reinnervation of hippocampal area CA1 and the fascia dentata after destruction of CA3-CA4 aherents with kainic acid. Brain Res. 182, l-9. 27. Nothias F., Onttniente B., Geffard M. and Peschanski M. (1990) Dissimilar responses of adult thalamic monoaminergic and somatosensory afferent fibers to implantation of thalamic fetal cells. Neuroscience 37, 353 366. R. S. and Rakic P. (1979) The mode of migration of neurons to the hippocampus: a Golgi and electron 28. Nowakowski microscopic analysis in fetal rhesus monkey. J. Neuroc:,fol. 8, 697 718. R. S. and Rakic P. (1981) The site of origin and route and rate of migration of neurons to the hippocampal 29. Nowakowski region of the rhesus monkey. J. camp. Neural. 196, I2991 54. of striatal 30. Pritzel M.. Isacson O., Brundin P., Wiklund L. and Bjorklund A. (1987) Afferent and efferent connections grafts implanted into the ibotenic acid lesioned neostriatum in adult rats. Expl Brain Res. 65, 112 126. into and out of hippocampal transplants. Neuro.~oeucr 31. Raisman G. and Ebner F. F. (1983) Mossy fibre projections 9, 783 ~801. transplants demonstrate the ability of the adult brain to receive and 32. Raisman G. and Ebner F. F. (1985) Hippocampal produce mossy libre connections. In Recent Arhiecements in Resroratiue Neurology. I. Upper Motor Neuron Funclion.~ crnd Dy.$uncrions (eds Eccles J. C. and Dimitrijevic M. R.), pp. 280-290. S. Karger, Basel. of radial glial cells during the development of 33. Rickmann M., Amaral D. G. and Cowan W. M. (1987) Organization the rat dentate gyrus. J. camp. Neural. 264, 449 479. study of the time of origin and the 34. Schlessinger A. R., Cowan W. M. and Gottlieb D. I. (1975) An autoradiographic pattern of granule cell migration in the dentate gyrus of the rat. J. camp. Neural. 159, 149-176. 35. Schlessinger A. R., Cowan W. M. and Swanson L. W. (1978) The time of origin of neurons in Ammon’s horn and the associated retrohippocampal fields. Anal. Embryol. 154, I53 173. organization of normal and transplanted rat fascia dentata: 1. A 36. Sorensen T. and Zimmer J. (1988) Ultrastructural qualitative analysis of intracerebral and intraocular grafts. J. camp. Neural. 267, 15-42. organization of normal and transplanted rat fascia dentata. 11. 37. Sorensen T. and Zimmer J. (1988) Ultrastructural A quantitative analysis of the synaptic organization of intracerebral and intraocular grafts. J. romp. Neural. 267, 43 54. R. M. (1987) Reconstruction of the defective cerebellar circuitry in adult Purkinje cell 3x. Sot& C. and Alvarado-Mallart degeneration mutant mice bv Purkinie cell replacement through transplantation of solid embryonic implants. Ne&science 20, l--22. . _ . R. M., Gardette R. and Crepe1 F. (1990) Fate of grafted embryonic Purkinje cells in the 39. Sotelo C., Alvarado-Mallart mutant mouse. I. Development of reciprocal graft host cerebellum of the adult “Purkinje cell degeneration” interactions. J. camp. Neural. 295, l65- 187. . e.. 39a. Stanfield B. B. and Trite J. E. (1988) Evidence that granule cells generated m the dentate gyrus ot adUlt rats extena axonal projections. Expl Brain Res. 72, 399406. 40. Sunde N. and Zimmer J. (1981) Dentate granule cells transplanted to hippocampal field CAI form aberrant mossy fiber projection in rats. Neurosci. Let/. Suppl. 7, S33.

superior (CAI) of the rat hippocampal

Innervation

of hippocampal

grafts

121

41. Sunde N. A. and Zimmer J. (1983) Cellular, histochemical and connective organization of the hippocampus and fascia dentata transplanted to different regions of immature and adult rat brains. Deul Bruin Res. 8, 165-191. 42. Tender N., Sorensen T. and Zimmer J. (1989) Enhanced host perforant path innervation of neonatal dentate tissue after grafting to axon sparing, ibotenic acid lesions in adult rats. Expl Brain Res. 75, 483496. 43. Wictorin K., Clarke D. J., Bolam J. P. and Bjijrklund A. (1989) Host corticostriatal fibres establish synaptic connections with grafted striatal neurons in the ibotenic acid lesioned striatum. Eur. J. Neurosci. 1, 189-195. 44. Woodhams P. L., Webb M., Atkinson D. J. and Seeley P. J. (1989) A monoclonal antibody, Py, distinguishes different classes of hippocampal neurons. J. Neurosci. 9, 2170-2181. 45. Xu Z. C., Wilson C. J. and Emson P. C. (1989) Restoration of the corticostriatal projection in rat neostriatal grafts: electron microscopic analysis. Neuroscience 29, 5399550. 46. Zhou C-F., Li Y. and Raisman G. (1989) Embryonic entorhinal transplants project selectively to the deafferented entorhinal zone of adult mouse hippocampi, as demonstrated by the use of Thy-l allelic immunohistochemistry. Effect of timing of transplantation in relation to deafferentation. Nenroscience 32, 349-362. 47. Zhou C.-F.. Raisman G. and Morris R. J. (1985) Soecific oatterns of fibre outerowth from transolants to host mice hippocampi, shown immunohistochemically‘by the use of allelic forms of Thyj. Neuroscience 16, 819-833. 48. Zimmer J. and Gahwiler B. H. (1987) Growth of hippocampal mossy fibres: a lesion and coculture study of organotypic slice cultures. J. camp. Neural. 264, l-13. (Accepted 31 October 1990)