Entorhinal projections terminate onto principal neurons and interneurons in the subiculum: A quantitative electron microscopical analysis in the rat

Neuroscience 136 (2005) 729 –739

ENTORHINAL PROJECTIONS TERMINATE ONTO PRINCIPAL NEURONS AND INTERNEURONS IN THE SUBICULUM: A QUANTITATIVE ELECTRON MICROSCOPICAL ANALYSIS IN THE RAT L. BAKS-TE BULTE,a F. G. WOUTERLOOD,a M. VINKENOOGa AND M. P. WITTERa,b*

of the entorhinal cortex, either directly or through adjacent cortices of the parahippocampal region (Witter et al., 1989; Burwell et al., 1995; Burwell and Amaral, 1998; Burwell, 2000). The principal neurons in the superficial layers of the entorhinal cortex provide the main cortical input, by way of the perforant pathway, to all four subfields of the hippocampal formation (dentate gyrus, areas CA3 and CA1, and the subiculum) (Witter, 1993; Witter and Amaral, 2004). At the beginning of the era of experimental neuroanatomical tracing, Blackstad (1958) pioneered the study of the organization of the perforant pathway by providing one of the first experimental reports on projections from the entorhinal cortex to the hippocampal formation, including a brief mentioning of perforant pathway fibers in the subiculum. The perforant pathway, which has its origin in layers II and III, shows a differential and selective distribution such that fibers originating from layer II neurons almost exclusively target the dentate gyrus and CA3, whereas layer III neurons project selectively to CA1 and the subiculum (Witter et al., 2000). Aside from the differentiation between the layer II and layer III component, the perforant pathway comprises projections originating in the lateral (LEC) and medial (MEC) entorhinal cortex, commonly known as the lateral and medial perforant pathway, respectively. Fibers belonging to each projection show strikingly different terminal distributions in each of the hippocampal subfields. Since LEC and MEC receive different, and to a large extent complementary sets of cortical inputs, both components of the perforant pathway most likely convey functionally different information to the hippocampus (see for review Burwell and Witter, 2002). In addition to such a functional difference, electrophysiological studies showed considerable differences in the effects of LEC and MEC stimulation on cells of the dentate gyrus as well as on neurons in CA3 (Abraham and McNaughton, 1984; Dahl et al., 1990; Colino and Malenka, 1993; Berzhanskaya et al., 1998; Rush et al., 2001; Do et al., 2002). As yet, no data are available concerning possible differences in the organization and function of the lateral and medial perforant pathway in CA1 and subiculum. Desmond et al. (1994) characterized the types of synapses formed in CA1 of fibers originating from various loci throughout the lateral-to-medial extent of the entorhinal cortex. These authors did not differentiate between the lateral and medial components of the perforant pathway. Similar to the synaptic organization of perforant pathway fibers and terminals in the dentate gyrus and CA3 (for review, see Turner et al., 1998; Witter and Amaral, 2004), Desmond and her colleagues (1994) exclusively

a Graduate School of Neurosciences Amsterdam, Research Institute Neuroscience, Department of Anatomy, MF-G-102C, VU University Medical Center, P.O. Box 7057, 1007 MB Amsterdam, The Netherlands b Centre for the Biology of Memory, Norwegian University of Science and Technology, Olav Kyrres Road 3, NO–7489, Trondheim, Norway

Abstract—The synaptic organization of projections to the subiculum from superficial layers of the lateral and medial entorhinal cortex was analyzed in the rat, using anterograde neuroanatomical tracing followed by electron microscopical quantification. Our aim was to assess the synaptic organization and whether the two projection components (lateral, medial) within the perforant pathway are qualitatively and quantitatively similar with respect to the types of synapses formed and with respect to the postsynaptic targets of these entorhinal projections. The tracer biotinylated dextran amine (BDA) was injected into the lateral and medial entorhinal cortex, respectively, and resulting anterograde labeling in the subiculum was studied. For each of the two projection components, we analyzed in four animals (2ⴛ2) a total of 100 synapses/animal with respect to features of the synapse type, i.e. asymmetrical or symmetrical, as well as regarding their postsynaptic target, i.e. dendritic shaft or spine. No clear differences were observed between the two pathways. The majority of the synapses were of the asymmetrical type, making contact with spines (78%) or with dendritic shafts (14%). A low percentage of symmetrical synapses targeted dendritic shafts (4.2%) or spines (1.3%). About 2.5% of the synapses remained undetermined. The findings indicate that the majority of entorhinal fibers reaching the subiculum exert an excitatory influence primarily onto principal neurons, with a much smaller feed forward inhibitory component. Only a small percentage of entorhinal fibers in the subiculum appears to be inhibitory, largely influencing interneurons. © 2005 Published by Elsevier Ltd on behalf of IBRO. Key words: entorhinal cortex, parahippocampal region, hippocampus, perforant pathway, anatomy.

The hippocampal system plays a critical role in learning and memory. Sensory input to this system mainly converges onto neurons in the superficial layers (layers II–III) *Correspondence to: M. P. Witter, Department of Anatomy, MF-G102C, VU University Medical Center, P.O. Box 7057, 1007 MB Amsterdam, The Netherlands. Tel: ⫹31-20-44-48048; fax: ⫹31-20-4448054. E-mail address: [email protected] (M. P. Witter). Abbreviations: BDA, biotinylated dextran amine; DAB/Ni, nickelenhanced diaminobenzidine; DMSO, dimethyl sulfoxide; LEC, lateral entorhinal cortex; MEC, medial entorhinal cortex; PB, phosphate buffer; TBS-Tx, Tris-buffered saline with Triton. 0306-4522/05$30.00⫹0.00 © 2005 Published by Elsevier Ltd on behalf of IBRO. doi:10.1016/j.neuroscience.2005.03.001

729

730

L. Baks-te Bulte et al. / Neuroscience 136 (2005) 729 –739

found asymmetrical synapses predominantly on spines and to a lesser degree on dendritic shafts. Asymmetrical synapses are commonly associated with excitatory synaptic activity and symmetrical synapses with inhibitory synaptic activity (Uchizono, 1965). Desmond et al.’s (1994) observations are not consistent with the findings of Germroth et al. (1989a,b), who, on the basis of a retrograde tracing study, reported a small population of GABA-ergic sparsely spinous entorhinal layer II and III neurons projecting to the hippocampus. These latter findings are taken to indicate that inhibitory neurons project to hippocampal structures as well. In addition, Behr et al. (1998) reported that in an in vitro situation, stimulation of layer III of MEC produces a combination of short-latency, presumably monosynaptic excitatory and inhibitory responses in the subiculum. These data together indicate that at least part of the entorhinal projection to the subiculum arises from inhibitory neurons. In the present study, we aimed to study whether morphological findings support the presence of both excitatory and inhibitory entorhinal projections to the subiculum. To this end we carried out a quantitative electron microscopical study to establish whether perforant pathway projections from the LEC and MEC form asymmetrical and/or symmetrical synapses with subicular neurons. Second, we wanted to identify whether the neurons postsynaptic to the perforant pathway projections in the subiculum include both principal neurons as well as interneurons. Third, we assessed whether the lateral and medial entorhinal projections to the subiculum differ with respect to their respective synaptic organization.

EXPERIMENTAL PROCEDURES Eight female Wistar rats were used (weight 200 –220 g; Harlan Centraal Proefdierbedrijf, Zeist, The Netherlands). Before surgery, the animals were deeply anesthetized with a 4:3 mixture of ketamine (Ketaset, Aesco, Boxtel, The Netherlands) and xylazine (Rompun, Bayer, Mijdrecht, The Netherlands), total dose 1 ml/kg body weight. They were subsequently mounted in a stereotaxic frame. Glass micropipettes (GC 150F-15, Clark, Reading, UK) with an outer tip diameter of 10 –15 ␮m were filled with a 5% solution of the anterograde tracer biotinylated dextran amine (BDA, MW 10,000; Molecular Probes Inc., Eugene, OR, USA) in 0.1 M sodium phosphate buffer, pH 7.4. The pipettes were stereotaxically lowered into the superficial layers of LEC or MEC. Stereotaxic coordinates were derived from the rat brain atlas of Paxinos and Watson (1998). The tracer was injected by applying a positively pulsed DC current to the micropipette (6 ␮A, 7 s on, 7 s off) for 10 min. The survival time for BDA was taken as 7–9 days. After that, the rats received an overdose of sodium pentobarbital (Nembutal, i.p. 60 mg/kg body weight; Sanofi Sante BV, Maassluis, The Netherlands). They were subsequently transcardially perfused with approximately 50 ml of a Ringer solution (8.5 g NaCl, 0.25 g KCl, 0.2 g NaHCO3 in 1000 ml of distilled water at 40 °C, brought to pH 6.9 with 95% O2/5% CO2) to clear blood cells, followed by perfusion with a solution of 4% freshly depolymerized paraformaldehyde, 0.1% glutaraldehyde and 0.2% picric acid (Merck, Darmstadt, Germany) in 0.125 M phosphate buffer, pH 7.4 (PB). After completion of the perfusion, the brains were removed from the skull, post-fixed for 1 h at room temperature in the same fixative, rinsed in 0.125 M PB and cut immediately after with a vibrating

microtome (model VT1000S, Leica Microsystems, Heidelberg, Germany) into 50 ␮m thick sections. If the tracer had been injected in LEC, the brain was cut in the coronal plane, whereas in case of MEC injection, horizontal sections were prepared. These respective planes of sectioning provide the best way to conserve the overall connectivity such that in a single section, both the injection side as well as the resulting labeling in the hippocampus can be observed (compare Fig. 1A and C). The sections were collected in vials containing 0.125 M PB. All slices were rinsed for 10 min each in an ascending series of 10%, 15% and 20% dimethyl sulfoxide (DMSO) in PB. If not immediately processed, slices were transferred to 20% DMSO/2% glycerin in PB and stored at ⫺20 °C.

Histochemistry Sections were rinsed several times in PB and subsequently in Tris buffered saline with Triton (TBS-Tx; 0.05 M Tris buffer with 0.15 M NaCl and 0.5% Triton X-100, Merck), pH 8.0 and incubated in avidin– biotin–peroxidase complex (ABC-kit; Vectastain PK 4000, Vector, Brunschwig, Amsterdam, The Netherlands) for 48 h. Next, they were rinsed for 3⫻10 min in TBS-Tx, rinsed for 2⫻5 min in Tris/HCl, pH 8.0, and subsequently stained with nickel-enhanced diaminobenzidine (DAB/Ni; 7.5 mg DAB, 220 mg nickel-ammonium sulfate, Merck) in 50 ml 0.05 M Tris/HCl, pH 8.0, to which 10 ␮l of a 30% solution of H2O2 had been added immediately before use. After 5–20 min, when upon inspection in a microscope the projection from the entorhinal cortex appeared to be clearly labeled while the sections contained no, or almost no background staining, the DAB/Ni reaction was stopped with Tris/HCl and the sections were rinsed for 2⫻5 min in Tris/HCl. Brain sections containing the best anterograde labeling in the subiculum were selected and further processed for EM analysis as described below. In case of light microscopical evaluation of sections, they were mounted on glass slides from a 0.2% gelatin solution in 0.05 Tris/HCl, pH 8.0. After overnight drying they were cleared in xylene and coverslipped with Entellan (Merck).

EM preparations Slices with a dense, anterogradely labeled plexus in the subiculum selected from brains that contained optimally positioned tracer injections, were further processed for electron-microscopical examination. In only four animals, the injection site of the tracer was centered in layer III (with additional labeling of layer II) of LEC (n⫽2; cases 2000129 and 2000142) or MEC (n⫽2; cases 2000086 and 2002021), resulting in a densely labeled plexus in the molecular layer of the subiculum. Series of sections taken from these four brains were used. The selected sections were quickly frozen in iso-pentane cooled by solid carbon dioxide and subsequently recovered and thawed (Wouterlood and Jorritsma-Byham, 1993). This freeze/thaw procedure was repeated twice. The incubation of slices was similar as described above, except for the Triton X-100 being left out, the use of pH 7.6 instead of pH 8.0 and the use of DAB (5 mg 3,3=-diaminobenzidine tetrahydrochloride solution in 10 ml of 0.05 M Tris–HCl, to which 3.3 ␮l of a 30% H2O2 was added) instead of DAB/Ni. Sections were subsequently rinsed 2⫻10 min in 0.1 M sodium cacodylate buffer, pH 7.3 (Merck) and postfixed for 1 h in 1% OsO4 (Merck) in 0.1 M sodium cacodylate buffer at 4 °C. After several rinses (3⫻10 min) in the same buffer and 2⫻5 min rinses in distilled water, sections were stained in a 2% solution of uranyl acetate (Merck) in distilled water for 1 h at 4 °C. Next, sections were dehydrated to 100% ethanol through ascending series of ethanol and, after two rinses with propylene oxide, embedded in Epon (Polysciences, Warrington, PA, USA) between polyethylene foil. After curing (24 h at 50 °C) all sections were inspected under the light microscope to determine the location of the labeled plexus in the subiculum. Selected sections were drawn with the use of a light- microscope equipped with a drawing

L. Baks-te Bulte et al. / Neuroscience 136 (2005) 729 –739

731

Fig. 1. Light micrographs of the fiber and terminal labeling in the hippocampal formation after injection of BDA in LEC (A, frontal section at low magnification; detail in B) and MEC (C, horizontal section; detail in D). (A, B) Labeling after injection in LEC. There is dense terminal labeling in the outer one-third of the molecular layer of the dentate gyrus (DG, open arrowheads) and the outer half of the stratum lacunosum-moleculare of CA3 (open arrowheads). In this case, additional labeling occurs throughout stratum lacunosum moleculare in the distal part of CA1 and the outer part of the molecular layer of the adjacent proximal part of the subiculum (Sm-o; arrows in A). (C, D) Following the injection in MEC (B), labeling in DG and CA3 has shifted more toward the cell layer (open arrowheads), whereas terminal labeling in CA1 is now most dense in the proximal part of the stratum lacunosum-moleculare. Dense fiber labeling is also present in the distal part of the outer molecular layer of the subiculum (arrow). CA1-3, subdivisions of the hippocampus proper (cornu ammonis); PP, perforant pathway, RhS, rhinal sulcus; S, subiculum; Sm-i and Sm-o, inner and outer portions of the molecular layer of the subiculum; Sp, pyramidal cell layer.

tube. Small samples containing the terminal plexus in the outer molecular layer of the subiculum were cut out with a dissection knife and transversely embedded in fresh Epon in an embedding mold, and cured at 60 °C. After curing, series of ultrathin sections

were cut with a Reichert OM-U4 ultramicrotome and collected on formvar-coated single slot copper grids. The orientation of the samples was such that sections were cut from distal-to-proximal along the transverse axis of the subiculum. Sections were coun-

732

L. Baks-te Bulte et al. / Neuroscience 136 (2005) 729 –739

terstained with 0.1–2.0% lead citrate (Merck) in distilled water and examined with a Philips CM 100 electron microscope. Areas containing labeled synaptic contacts were digitally imaged with a Kodak Megaplus 1.6i camera. Images were subsequently stored on disk with SIS™ acquisition software (Soft Imaging, Münster, Germany).

Quantitative analysis Synapse morphology was verified by inspecting images of series of sections collected with the electron microscope. Estimates of synapse types and the postsynaptic targets were obtained by randomly sampling 100 consecutive labeled synapses formed by lateral perforant pathway axon terminals and medial perforant pathway axon terminals, respectively. Synapses were sampled in serially cut ultrathin (60 nm) sections, obtained from two 40 ␮m thick sections per animal, spaced at last 180 ␮m apart. These sections were taken through the densest part of the terminal plexus in the molecular layer of the subiculum in order to increase the likelihood of finding labeled synapses. EM-analysis was started in a randomly selected ultrathin section, and in that section all labeled synapses were collected. Sampling was continued in the next section and so on until 100 synapses had been obtained. Spacing between ultrathin sections was at least 120 nm (every third section in a ribbon), in order to avoid double counting of a single labeled synapse in the sample. The redundant sections in between were only used to establish with certainty the categorization as symmetrical synapse and to double-check in case of doubt that no synapses were possibly counted twice. Synapse categories were determined on the basis of their overall morphology, using criteria described in detail elsewhere (van Haeften et al., 1995). We grouped the synapses made by labeled axon terminals into two categories: 1. Asymmetrical: Synapses with postsynaptic densities more than twice as thick as the presynaptic densities and with a wide synaptic cleft. Boutons of asymmetrical synapses contain spherical presynaptic vesicles. 2. Symmetrical: Synapses with pre- and postsynaptic densities of equal thickness and with a translucent and narrow synaptic cleft. Boutons of symmetrical synapses contain pleomorphic vesicles. Note that in case of a presumed symmetrical synapse, all serially collected sections including that bouton were checked in order to make sure that the synaptic densities did not change thickness. In addition to defining the type of synapse type, we also differentiated between the postsynaptic targets of these synapses. Dendritic spines are characterized by the presence of clear cytoplasmic matrix and a distinct spine apparatus, whereas dendritic shafts are recognizable by the presence of microtubules and mitochondria in the cytoplasm (Peters and Palay, 1996).

Blind categorization After categorizing the boutons, all images were presented blindly to an unbiased independent observer who was asked to categorize the boutons according to the above criteria. Boutons that ‘failed the test,’ i.e. being categorized in different classes by the first and second observer, were classified as ‘undetermined.’ All experiments were approved by the local committee on animal welfare and ethics and carried out in accordance with national and European regulations on animal well-being.

RESULTS Light microscopy All injections involving superficial layers II and III of either LEC or MEC resulted in a labeled terminal field in the

dentate gyrus, CA3, CA1, and the subiculum (Fig. 1). In the subiculum, dense anterograde labeling was seen of fibers located in the outer portion of the molecular layer of the subiculum (Fig. 1A, C). In case of LEC injections, labeled fibers were observed in the distal half of CA1 and the adjacent proximal part of the subiculum (Fig. 1A, B). Furthermore, the outer onethird of the molecular layer of the dentate gyrus and the outer half of stratum lacunosum-moleculare of CA3 contained a dense terminal plexus of labeled fibers. Injections in MEC resulted in a labeled plexus in the subiculum, which was confined to the distal part, adjacent to the presubiculum, whereas in CA1, labeled terminal fibers were present in the most proximal portion (Fig. 1C, D). In these latter animals, also the middle one-third of the molecular layer of the dentate gyrus and the inner half of stratum lacunosum-moleculare of CA3 contained labeled fibers. As described above, for further analysis with the electron microscope, four animals were selected on the basis of the observation that they had dense labeling in the molecular layer of the subiculum. In two other animals, the injection site was mainly focused in entorhinal layer II, thus resulting in only weak anterograde labeling in the subiculum. The remaining two animals were not included since the injection had labeled neurons both in LEA and MEA. In view of the known topographical organization of the perforant pathway along the longitudinal axis of the hippocampal formation, which holds true for the layer III to subiculum projection (Witter and Amaral, 2004), we aimed to spread our injections such as to cover at least the central portion of the long axis of the subiculum with respect to terminal labeling. The four selected injections include four different dorsal-to-ventral levels of the entorhinal cortex (see Table 1). In two of the selected four cases, (one LEC-injection and one MEC-injection), we observed retrogradely labeled neurons in CA1 and the subiculum, apparently resulting from uptake of the tracer by axon terminals of these cells in the injected layers of the entorhinal cortex. The tracer, transported retrogradely into the somata, subsequently may have been further transported anterogradely into the axons of these cells, resulting in unintended anterograde labeling of axon-terminals in the subiculum. However, these possible false positive labeled synapses can be Table 1. Stereotaxic coordinates of injection sites in four animals selected for EM analysis # Animal

2000086 2002021 2000129 2000142

(MEC) (MEC) (LEC) (LEC)

Bregma; mm behind

Lateral; mm from midline

Depths from cortical surface

⫺8.4 ⫺7.9 ⫺5.8 ⫺5.8

4.8 5.5 6.3 6.1

3.4 and 3.6 4.0 and 4.2 5.5 5.7

This table provides the stereotaxic coordinates used to inject BDA into the LEC and MEC in four animals. The coordinates are chosen as to label perforant pathway projections that distribute along the central two-thirds of the longitudinal axis of the subiculum.

L. Baks-te Bulte et al. / Neuroscience 136 (2005) 729 –739

differentiated from those of the perforant pathway by their location. The CA1 efferents as well as the local intrinsic subicular projections terminate in the pyramidal layer and deep molecular layer of the subiculum such that they are strictly spatially separated from the perforant pathway fibers that occur in the superficial molecular layer (Harris et al., 2001; Naber et al., 2001). Indeed, labeled fibers were observed in these locations in the subiculum (i.e. outside the sampling areas) in the cases with retrogradely labeled CA1 or subicular neurons while they were absent in the cases with no such retrogradely labeled neurons. This specific laminar organization of subicular connectivity therefore allowed us to carry out our qualitative and quantitative analysis even in the case of such retrograde neuronal labeling. Synaptic types and postsynaptic targets Axon terminals of perforant pathway fibers in the molecular layer of the subiculum could easily be identified in ultrathin sections due to the presence of an opaque deposit of electron dense DAB reaction product in the matrix of the presynaptic bouton (Figs. 2, 3). In some cases, the DAB reaction product was very dense, thus obscuring the presynaptic density of the synaptic junction. These synapses were excluded from further analysis. Only synaptic profiles were taken into account that displayed clear pre- andpostsynaptic densities and a distinct synaptic cleft. We used criteria published previously (van Haeften et al., 1995; Peters and Palay, 1996) to categorize individual synapses into three major groups. Group 1: Boutons with asymmetrical synapses (Fig. 2A, B, D, E): synapses with a postsynaptic density displaying a thickness at least twice that of the presynaptic density. The synaptic cleft is relatively wide, and the synaptic clefts may contain opaque deposits. Boutons of this category contain spherical synaptic vesicles. Group 2: Boutons with symmetrical synapses (Fig. 2C, F): synapses displaying pre- and postsynaptic densities of equal thickness, with relatively narrow and translucent synaptic clefts. Boutons of this category contain pleomorphic synaptic vesicles, which are smaller than those in the terminal boutons belonging to group 1. Group 3: Boutons with undetermined synapses: synapses for which a discrepancy existed between the original observation and judgment by a ‘second opinion’ observer concerning synapse characterization and therefore could not be classified as either the asymmetrical or the symmetrical type. In all four analyzed cases, BDA-labeled axon terminals formed synaptic contacts with dendritic shafts, characterized by the presence of microtubules and mitochondria in the cytoplasm (Fig. 2A, C, D, F). BDAlabeled terminals also contacted large- and intermediatesized dendritic spines, recognizable by the presence of clear dendritic cytoplasm and a distinct spine apparatus (Fig. 2B, E). In only one case a labeled presynaptic element was seen to make an asymmetrical synapse on a soma. No labeling of dendro-dendritic or axo-axonic synapses was observed.

733

Representative cases of an asymmetrical synapse on a dendritic shaft are shown for both the lateral and medial perforant pathway in Fig. 2A and D. The presynaptic bouton is filled with synaptic vesicles, or in some cases contained vesicles aggregated at the presynaptic membrane density. The postsynaptic density is thick compared with the presynaptic membrane. The postsynaptic structures displayed ultrastructural characteristics typical for dendrites: clear matrix, mitochondria and regularly spaced microtubules. Fig. 2B and E show asymmetrical synapses onto a spine formed by axon terminals belonging to either the lateral (2B) or medial (2E) perforant pathway. Again, a clear asymmetry of the thickness of the pre- and postsynaptic membranes can be seen. In some instances, a clear spine apparatus in the postsynaptic structure was seen (indicated in Fig. 2E). The morphological appearance of a synapse in the electron microscope is, among others, to a large degree dependent on the plane of sectioning. If an ultrathin section through a group-1 bouton is positioned immediately adjacent to the postsynaptic membrane specialization, the resulting profile can easily be interpreted as belonging to a group-2 bouton. As a consequence, observations on single ultrathin-sections may lead to false characterization of synapses, with a bias leaning toward categorizing these profiles as symmetrical synapses. In order to assess with certainty that a synaptic specialization indeed was symmetrical, we investigated consecutive serial sections of entire pre- and postsynaptic structures. Synapses with a clear asymmetrical morphology were accepted without further screening. Examples of a symmetrical synapse onto a dendritic shaft in case of the lateral and medial perforant pathway are illustrated in Fig. 2C and F, respectively. The pre- and post-synaptic membranes display thin densities, with a relatively thin, translucent synaptic cleft in between. In consecutive sections this image remained the same, as illustrated in Fig. 3. In all sections through this bouton (only three shown), the synapse appeared to be symmetrical. In addition, the synaptic vesicles in symmetrical boutons had a pleomorphic appearance (Figs. 2F, 3). Quantification of pre- and postsynaptic structures The results of the quantification of synapse types and postsynaptic targets in the molecular layer of the subiculum are summarized in Fig. 4. Analyses of four cases revealed no apparent differences in synaptic organization between the two components of the perforant pathway. Of the two cases with injections in LEC, the far majority of the synapses observed in the subiculum were of the asymmetrical type (Fig. 4; white and gray bars; mean for the two animals⫽91.5%), of which most (mean for two animals⫽79%) were made with large- and intermediate-sized dendritic spines and some (mean for two animals⫽12.5%) on dendritic shafts. A small number (4.0%) of symmetrical synapses were on dendritic shafts. Symmetrical synapses, made with a spine (black bars) were only seldom found in the subiculum (1 and 2% respectively). A few synapses (mean of two animals⫽3%) had intermediate features and

734

L. Baks-te Bulte et al. / Neuroscience 136 (2005) 729 –739

Fig. 2. Electron micrographs of typical examples of the main categories of labeled axon terminals in the subiculum. (A, D) Labeled terminals forming asymmetrical synapses onto the shafts of Den, cut longitudinally (A) or transversally (D). (B, E) Asymmetrical synaptic contact between a labeled terminal and a dendritic Sp. (C, F) Symmetrical synaptic contacts between a labeled AT and a dendritic shaft. Synaptic specializations are indicated with an arrow; microtubuli with arrowheads; mitochondria with asterisks. Scale bar⫽0.5 ␮m in all frames. AT, axon terminal; Den, dendrite; Sp, spine; SpA, spine apparatus.

L. Baks-te Bulte et al. / Neuroscience 136 (2005) 729 –739

735

Fig. 3. Three representative photographs taken from a series of thin sections through a symmetrical synapse between a labeled axon terminal and a dendritic shaft. With a thickness of 60 nm per section the total synapse thickness is estimated to be 300 – 420 nm. Synaptic specializations are indicated with an arrow and microtubuli with arrowheads. Scale bar⫽0.5 ␮m in all frames. Abbreviations as in Fig. 2.

736

L. Baks-te Bulte et al. / Neuroscience 136 (2005) 729 –739

were characterized as ‘undetermined.’ These undetermined synapses terminated on dendritic shafts only. In case of injections in MEC, the majority of the synapses were also of the asymmetrical type (Fig. 4; mean of two animals⫽93%). The majority of these asymmetrical synapses formed synaptic contacts with large- and intermediate-sized dendritic spines (77.5%) and to a lesser degree with dendritic shafts (15.5%). A small fraction of the analyzed synapses were of the symmetrical type and most of them contacted dendritic shafts (4.5%). In only one MEC-injected rat (case 2002021) did we observe a symmetrical synapse with a spine. A few synapses (2.0%) were categorized as ‘undetermined,’ apposing spines (0.5%) and dendritic shafts (1.5%).

DISCUSSION In the present tracing study we analyzed the terminal organization of projections from LEC and MEC to the molecular layer of the subiculum. For the light- and electron microscopic identification of entorhinal-subicular fibers and axon terminals we used the anterograde neuroanatomical tracer BDA (Wouterlood and Jorritsma-Byham, 1993). In particular, we used the morphology of the synapses and the postsynaptic targets of the labeled axon terminals for classification purposes and for the subsequent quantification. For each of the lateral and medial perforant pathways, the qualitative as well as quantitative observations in the two animals analyzed are remarkably consistent. Because of this consistency, as well as the

random sampling method used, we feel confident that our data represent a reliable sample. Our overall light microscopical findings concerning LEC and MEC projections to the subiculum, as well as those to the other subfields of the hippocampus, are in line with previously reported observations (Witter, 1989; Naber et al., 2001). Our results corroborate earlier reports that the proximal part of the subiculum receives input from LEC, while the distal part of the subiculum receives input from MEC. The electron microscopic analysis of the entorhinal terminals in the subiculum revealed a majority of asymmetrical synapses on spines and dendritic shafts, with only a small number of symmetrical synapses, the latter invariably being engaged with dendritic shafts. It is generally accepted that asymmetrical synapses belong to excitatory neurons, while symmetrical synapses belong to inhibitory neurons (Uchizono, 1965; Raviola and Raviola, 1967). It is therefore likely that for both the projections from LEC and MEC, the majority of synapses in the subiculum are excitatory, while the remaining of the synapses should be characterized as inhibitory. These results are in line with the reports on a small proportion of GABA-ergic projection neurons among the overall glutamatergic population of projection neurons in layers II and III of EC (White et al., 1977; Germroth et al., 1989a,b). Likewise, with respect to the projections from layer III to the subiculum in particular, Gloveli et al. (1997) described both pyramidal-shaped spiny, presumably excitatory cells, as well as non-spiny,

Fig. 4. Quantitative representation of synapse types and postsynaptic targets in the subiculum. Data of individual animals for both the LEC and MEC perforant pathway components are represented. White bars: asymmetrical synapses with dendritic spines; gray bars: asymmetrical synapses with dendritic shafts; hatched bars: symmetrical synapses with dendritic shafts; dark gray bars: symmetrical synapses with dendritic spines; dotted bars: unidentified synapses.

L. Baks-te Bulte et al. / Neuroscience 136 (2005) 729 –739

presumably inhibitory cells in layer III of the MEC, both having axons that innervate the deep layers of the entorhinal cortex as well as CA1 and/or the subiculum. Regarding the subicular postsynaptic targets, our data indicate that both principal neurons as well as interneurons receive perforant pathway input. For both the lateral and medial perforant pathway, we observed that a large majority of the asymmetrical synapses (78% out of 92%) had spines as their postsynaptic element. Our observations thus indicate that the entorhinal input largely targets dendrites of spiny subicular neurons. A much smaller number of asymmetrical synapses (14% out of 92%) had the shafts of dendrites as their postsynaptic element. Dendritic shafts were also observed to be the preferred postsynaptic element of the symmetrical synapses. It seems highly unlikely that these dendritic shafts belong to spiny neurons since we never observed in consecutive sections that the identified axon terminals formed synapses onto dendritic shafts that also carried spines. We therefore assume that most if not all of the synapses on dendritic shafts in the subiculum represent contacts between entorhinal fibers and nonspiny neurons. Previous studies showed that subicular principal neurons are commonly spine-bearing pyramidal neurons, and that the non-spiny neurons most likely represent interneurons (Swanson et al., 1987; Greene and Totterdell, 1997; O’Mara et al., 2001; Harris et al., 2001). We therefore conclude that excitatory entorhinal fibers mainly target apical dendrites of principal neurons of the subiculum, whereas in a minority of the cases they target the dendrites of interneurons. The presumed inhibitory perforant pathway fibers mainly target the dendrites of interneurons. Although the sampled portions of the molecular layer contained a low number of cell bodies of presumed interneurons, we did not find symmetrical synapses onto the somata of such neurons. In previous reports concerning entorhinal projections to the dentate gyrus and the Ammon’s horn such axosomatic contacts were reported (Swanson et al., 1987). This difference in observation may be the result of our sampling procedures, since we did not specifically sample the potentially involved interneurons. Our conclusions appear in line with previous in vitro electrophysiological findings that stimulation of layer III of MEC produces in the subiculum a combination of shortlatency strong excitatory and weaker inhibitory responses. The reported short-latency of these subicular responses suggests that both excitation and inhibition result from monosynaptic connections (Behr et al., 1998). Interestingly, Misgeld and Frotscher (1986) reported disinhibition in the guinea-pig subiculum that, according to these authors, is mediated by GABAergic synapses onto GABAergic neurons. Based on the present observations, it is likely that a similar situation holds true for the subiculum of the rat. The present results on entorhinal–subicular projections to a large extent resemble those reported earlier with respect to projections from the entorhinal cortex to CA1 (Desmond et al., 1994), with the exception that no symmetrical synapses were reported in CA1. However, own

737

findings support the existence of symmetrical synapses of perforant pathway fibers in CA1 (Witter et al., 1992). This potential overall similarity between the layer III projections to CA1 and subiculum may not be unexpected in view of data based on intracellular fillings of layer III projection neurons (Gloveli et al., 1997), who reported that at least a proportion of pyramidal cells in layer III of MEC possesses axons that head to both CA1 and the subiculum. We observed no major qualitative and quantitative differences in the types of synapses and postsynaptic targets between the lateral and medial perforant pathway components. This is similar to the situation reported for the dentate gyrus, where also no clear differences have been described regarding overall types of synapses and postsynaptic targets (Witter and Amaral, 2004). Moreover, our present observation that most of the perforant pathway synapses in the subiculum are excitatory, targeting principal cells as well as interneurons is strikingly comparable to findings in the dentate gyrus and CA3 (Nafstad, 1967; Witter et al., 1992; Zipp et al., 1989). This is of interest in view of the fact that the projections to subiculum, and CA1 arise predominantly from layer III cells whereas the projections to the dentate gyrus and CA3 arise largely from layer II cells (Steward and Scoville, 1976), partially as collaterals from the same neuron (Tamamaki and Nojyo, 1993). There is evidence indicating that layer II cells as well as layers V–VI may contribute to a slight extent to the entorhinal-subiculum projection. Although we noted labeling in the dentate gyrus and CA3, indicating that we have not been successful in exclusively labeling the layer III component, it is likely that our findings are representative for the entorhinal projection that originates in layer III. It is well known that the entorhinal projections to all subfields of the hippocampus, including that to the subiculum are topographically organized along the longitudinal or dorsoventral axis. Both intrinsic as well as extrinsic connectivity of hippocampal structures differ along the longitudinal axis (see Witter et al., 1989 for more details). Therefore, one might wonder whether the present data hold true for the entorhinal–subiculum projection along the entire dorsoventral axis. Note that we sampled at four different levels along the dorsoventral axis of the subiculum and that no marked qualitative and quantitative differences were found. We thus feel confident to conclude that the present data are illustrative for the entire entorhinal– subiculum projection. As indicated above, entorhinal fibers may target interneurons in the subiculum, since all layers of the subiculum, including the perforant pathway terminal zone, contain GABA immunoreactive neurons (Köhler et al., 1985; Swanson et al., 1987). These GABAergic neurons are generally taken to exert inhibitory effects onto other subicular neurons. In contrast to the situation in other hippocampal fields, for example CA1, where a quite extensive body of knowledge is available concerning the different types of interneurons with respect to chemical nature, dendritic and axonal arborization (Freund and Buzsaki, 1996), not much is known about this in the subiculum (see Witter and Amaral, 2004 for a recent review). Our own

738

L. Baks-te Bulte et al. / Neuroscience 136 (2005) 729 –739

preliminary data seem to indicate that the situation might actually be comparable to that in CA1. The molecular layer of the subiculum contains interneurons, which are immunopositive for VIP and nNOS and to a lesser extent for calretinin and calbindin. Although nNOS is not a marker for interneurons per se, the majority and highest intensity of nNOS immunopositive neurons in the subiculum involves interneurons (Lin and Totterdell, 1998). Similar types of cells as well as parvalbumin immunopositive neurons are present in the pyramidal cell layer of the subiculum. Of these, both parvalbumin- as well as nNOS-positive neurons have been shown to extend their dendrites into the outer portions of the molecular layer of the subiculum (Lin and Totterdell, 1998), where they may thus form postsynaptic targets to efferent entorhinal fibers. In a parallel study using a confocal approach, we established that among the interneurons targeted by perforant pathway input are indeed parvalbumin positive interneurons (own unpublished observations). Functional summary of the entorhinal–subicular network Based on the results discussed above, we propose that the entorhinal projection to the subiculum mainly comprises excitatory fibers predominantly targeting principal neurons of the subiculum at their apical dendrites. These fibers may also terminate on dendrites of inhibitory neurons, but in a much lower proportion. Although it is unknown whether both groups of synapses originate from the same set of entorhinal projection neurons, this is a likely scenario in view of the fairly restricted number of entorhinal fibers entering the subiculum at any given point of entry and the rather dense terminal plexus in the molecular layer. Moreover, intracellularly filled entorhinal axons on their way to the subiculum have been reported to give rise to multiple collaterals (Lingenhohl and Finch, 1991). The entorhinal projection is likely to also contain a small number of inhibitory fibers, mainly influencing inhibitory neurons in the subiculum. This network organization suggests that entorhinal input to the subiculum will exert a clear excitatory effect onto distal dendrites of pyramidal cells, with addition feed forward inhibition, as well as feed forward disinhibition. Acknowledgments—We are much indebted to Nico Blijleven for his perfect blend of enthusiasm and skilful assistance with the electron microscope hardware and software. Supported by the European Commission (Framework V grant QLG3-CT-199900192).

REFERENCES Abraham WC, McNaughton N (1984) Differences in synaptic transmission between medial and lateral components of the perforant path. Brain Res 303:251–260. Behr J, Gloveli T, Heinemann U (1998) The perforant path projection from the medial entorhinal cortex layer III to the subiculum in the rat combined hippocampal-entorhinal cortex slice. Eur J Neurosci 10: 1011–1018. Berzhanskaya J, Urban NN, Barrionuevo G (1998) Electrophysiological and pharmacological characterization of the direct perforant

path input to hippocampal area CA3. J Neurophysiol 79:2111– 2118. Blackstad TW (1958) On the termination of some afferents to the hippocampus and fascia dentata. An experimental study in the rat. Acta Anat 35:202–214. Burwell RD (2000) The parahippocampal region: corticocortical connectivity. Ann N Y Acad Sci 911:25– 42. Burwell RD, Amaral DG (1998) Perirhinal and postrhinal cortices of the rat: interconnectivity and connections with the entorhinal cortex. J Comp Neurol 391:293–321. Burwell RD, Witter MP (2002) Basic anatomy of the parahippocampal region in rats and monkeys. In: The parahippocampal region, organization and role in cognitive functions (Witter MP, Wouterlood FG, eds), pp 35– 60. Oxford, UK: Oxford University Press. Burwell RD, Witter MP, Amaral DG (1995) Perirhinal and postrhinal cortices of the rat: a review of the neuroanatomical literature and comparison with findings from the monkey brain. Hippocampus 5:390 – 408. Colino A, Malenka RC (1993) Mechanisms underlying induction of long-term potentiation in rat medial and lateral perforant paths in vitro. J Neurophysiol 69:1150 –1159. Dahl D, Burgard EC, Sarvey JM (1990) NMDA receptor antagonists reduce medial, but not lateral, perforant path-evoked EPSPs in dentate gyrus of rat hippocampal slice. Exp Brain Res 83:172–177. Desmond NL, Scott CA, Jane JA, Levy WB (1994) Ultrastructural identification of entorhinal cortical synapses in CA1 stratum lacunosummoleculare of the rat. Hippocampus 4:594–600. Do VH, Martinez CO, Martinez JL Jr, Derrick BE (2002) Long-term potentiation in direct perforant path projections to the hippocampal CA3 region in vivo. J Neurophysiol 87:669 – 678. Freund TF, Buzsaki G (1996) Interneurons of the hippocampus. Hippocampus 6:347– 470. Germroth P, Schwerdtfeger WK, Buhl EH (1989a) GABAergic neurons in the entorhinal cortex project to the hippocampus. Brain Res 494:187–192. Germroth P, Schwerdtfeger WK, Buhl EH (1989b) Morphology of identified entorhinal neurons projecting to the hippocampus. A light microscopical study combining retrograde tracing and intracellular injection. Neuroscience 30:683– 691. Gloveli T, Schmitz D, Empson RM, Dugladze T, Heinemann U (1997) Morphological and electrophysiological characterization of layer III cells of the medial entorhinal cortex of the rat. Neuroscience 77:629 – 648. Greene JR, Totterdell S (1997) Morphology and distribution of electrophysiologically defined classes of pyramidal and nonpyramidal neurons in rat ventral subiculum in vitro. J Comp Neurol 380: 395– 408. Harris E, Witter MP, Weinstein G, Stewart M (2001) Intrinsic connectivity of the rat subiculum: I. Dendritic morphology and patterns of axonal arborization by pyramidal neurons. J Comp Neurol 435: 490 –505. Köhler C, Wu JY, Chan-Palay V (1985) Neurons and terminals in the retrohippocampal region in the rat’s brain identified by anti-gammaaminobutyric acid and anti-glutamic acid decarboxylase immunocytochemistry. Anat Embryol 173:35– 44. Lin H, Totterdell S (1998) Light and electron microscopic study of neuronal nitric oxide synthase-immunoreactive neurons in the rat subiculum. J Comp Neurol 395:195–208. Lingenhohl K, Finch DM (1991) Morphological characterization of rat entorhinal neurons in vivo: soma-dendritic structure and axonal domains. Exp Brain Res 84:57–74. Misgeld U, Frotscher M (1986) Postsynaptic-GABAergic inhibition of non-pyramidal neurons in the guinea-pig hippocampus. Neuroscience 19:193–206. Naber PA, Lopes da Silva FH, Witter MP (2001) Reciprocal connections between the entorhinal cortex and hippocampal fields CA1 and the subiculum are in register with the projections from CA1 to the subiculum. Hippocampus 11:99 –104.

L. Baks-te Bulte et al. / Neuroscience 136 (2005) 729 –739 Nafstad PH (1967) An electron microscope study on the termination of the perforant path fibres in the hippocampus and the fascia dentata. Z Zellforsch Mikrosk Anat 76:532–542. O’Mara SM, Commins S, Anderson M, Gigg J (2001) The subiculum: a review of form, physiology and function. Prog Neurobiol 64: 129 –155. Paxinos G, Watson C (1998) The rat brain in stereotaxic coordinates. Sydney: Academic Press. Peters A, Palay SL (1996) The morphology of synapses. J Neurocytol 25:687–700. Raviola G, Raviola E (1967) Light and electron microscopic observations on the inner plexiform layer of the rabbit retina. Am J Anat 120:403– 425. Rush AM, Rowan MJ, Anwyl R (2001) Application of N-methyl-Daspartate induces long-term potentiation in the medial perforant path and long-term depression in the lateral perforant path of the rat dentate gyrus in vitro. Neurosci Lett 298:175–178. Steward O, Scoville SA (1976) Cells of origin of entorhinal cortical afferents to the hippocampus and fascia dentata of the rat. J Comp Neurol 169:347–370. Swanson LW, Köhler C, Björklund A (1987) The limbic region. I: The septohippocampal system. In: Handbook of chemical neuroanatomy (Björklund A, Hökfelt T, Swanson LW, eds), pp 125–277. Amsterdam: Elsevier. Tamamaki N, Nojyo Y (1993) Projection of the entorhinal layer II neurons in the rat as revealed by intracellular pressure-injection of neurobiotin. Hippocampus 3:471– 480. Turner DA, Buhl EH, Hailer NP, Nitsch R (1998) Morphological features of the entorhinal-hippocampal connection. Prog Neurobiol 55:537–562. Uchizono K (1965) Characteristics of excitatory and inhibitory synapses in the central nervous system of the cat. Nature 207:642–643.

739

van Haeften T, Jorritsma-Byham B, Witter MP (1995) Quantitative morphological analysis of subicular terminals in the rat entorhinal cortex. Hippocampus 5:452– 459. White WF, Nadler JV, Hamberger A, Cotman CW, Cummins JT (1977) Glutamate as transmitter of hippocampal perforant path. Nature 270:356 –357. Witter MP (1989) Connectivity of the rat hippocampus. In: The hippocampus: new vistas (Chan-Palay V, Köhler C, eds), pp 53–69. New York: Allen R. Liss. Witter MP (1993) Organization of the entorhinal-hippocampal system: a review of current anatomical data. Hippocampus 3 Spec No: 33–44. Witter MP, Amaral DG (2004) The hippocampal region. In: The rat brain (Paxinos G, ed), pp 637–703. San Diego, CA: Elsevier. Witter MP, Groenewegen HJ, Lopes da Silva FH, Lohman AH (1989) Functional organization of the extrinsic and intrinsic circuitry of the parahippocampal region. Prog Neurobiol 33:161–253. Witter MP, Jorritsma-Byham B, Wouterlood FG (1992) Perforant pathway projections to the ammons horn and the subiculum in the rat: an electron microscopical PHA-L study. Soc Neurosc Abstr 1: 323. Witter MP, Wouterlood FG, Naber PA, van Haeften T (2000) Anatomical organization of the parahippocampal-hippocampal network. Ann N Y Acad Sci 911:1–24. Wouterlood FG, Jorritsma-Byham B (1993) The anterograde neuroanatomical tracer biotinylated dextran-amine: comparison with the tracer Phaseolus vulgaris-leucoagglutinin in preparations for electron microscopy. J Neurosci Methods 48:75– 87. Zipp F, Nitsch R, Soriano E, Frotscher M (1989) Entorhinal fibers form synaptic contacts on parvalbumin-immunoreactive neurons in the rat fascia dentata. Brain Res 495:161–166.

(Accepted 1 March 2005)