THE JOURNAL OF COMPARATIVE NEUROLOGY 308:381-396 (1991)
Calcium-Binding Protein (Calbindin-D28K) and Parvalbumin Immunocytochemistry in the Normal and Epileptic Human Hippocampus ROBERT S. SLOVITER, ANNE L. SOLLAS, NICHOLAS M. BARBARO, AND KENNETH D. LAXER Neurology Research Center, Helen Hayes Hospital, New York State Department of Health, West Haverstraw, New York 10993 (R.S.S.,A.L.S.), Departments of Pharmacology and Neurology, Columbia University College of Physicians and Surgeons, New York, New York 10032 (R.S.S.), Departments of Neurosurgery (N.M.B.) and Neurology (K.D.L.), University of California San Francisco, San Francisco, California 94143
ABSTRACT The calcium-binding proteins calbindin-D28K (CaBP) and parvalbumin (PV) were localized in the “normal” and “epileptic” human hippocampus to address the possible relationship between the expression of these constitutive cytosolic calcium-binding proteins and the resistance or selective vulnerability of different hippocampal neuron populations in temporal lobe epilepsy. Compared to rodents and a baboon (Papiopupzo), the pattern of CaBP-like immunoreactivity (LI) in the “normal” human hippocampus is unique. CaBP-LI is present in the dentate granule cells, neurons of the “resistant zone” (area CAZ), and presumed interneurons of all regions. Unlike rodent and baboon CA1 pyramidal cells, human CA1 pyramidal cells appear to be devoid of CaBP-LI. Thus, the relatively resistant dentate granule cells and CA2 pyramidal cells are the only human hippocampal principal cells that contain CaBP-LI normally. As in lower mammals, PV-LI is present exclusively in interneurons of all human hippocampal subregions. CaBP- and PV-LI were localized in hippocampi surgically removed in the treatment of intractable temporal lobe epilepsy to determine whether surviving hippocampal cells were those that express these calcium-binding proteins. Hippocampi removed from patients with tumors or arteriovenous malformations that were associated with complex partial seizures arising from this region appeared relatively normal histologically. CaBP- and PV-LI in this patient group appeared similar to that seen in autopsy controls. Conversely, “cryptogenic” epileptics, who exhibit hippocampal sclerosis as the only lesion associated with the epilepsy, exhibited a preferential survival of hippocampal cells that were CaBP- or PV-immunoreactive. In the dentate hilus, which normally contains few CaBP-LI neurons, most of the few surviving hilar neurons were CaBP-immunoreactive. Their number and darkness of staining suggests that CaBP synthesis may be increased in cells that survive. Despite an obvious decrease of PV-LI specificallyin the damaged parts of the sclerotic hippocampi, PV-immunoreactive interneurons were often among the few surviving cells. Nevertheless, large expanses of the surviving granule cell layer appeared to have lost the PV-immunoreactive axosomatic fiber plexus. These results reveal a unique and striking correlation between the human hippocampal cells that normally express these calcium-binding proteins and those that survive in the sclerotic epileptic hippocampus. Key words: temporal lobe epilepsy, dentate gyrus, neuropathology
Accepted January 24, 1991.
D
1991 WILEY-LISS, INC.
R.S. SLOVITER ET AL.
382
The increase in intracellular calcium (Griffiths et al., ’82) that occurs as a result of excitatory amino acid receptor activation (Connor et al., ’88) has been suggested to be the initiating factor in seizure-associated neuronal death (Meldrum, ’81). However, although most or all hippocampal neurons are presumably excited during seizures, some cell populations are more vulnerable than others. It has been hypothesized that differences in the vulnerability of hippocampal neurons to seizure activity may be related to the absence or presence of cytoplasmic calcium-binding proteins (Sloviter, ’89) that buffer free intracellular calcium (Morrissey et al., ’78; Baimbridge and Miller, ’82). In the rat, there is a striking correlation between the cells that survive intense seizure activity in an experimental epilepsy model (Sloviter, ’87) and those that possess either of the calcium binding proteins calbindin-D28K (CaBP) and parvalbumin (Sloviter, ’89).However, the presence of CaBPlike immunoreactivity (CaBP-LI) in rat and baboon CA1 pyramidal cells (Sloviter et al., ’89) is inconsistent with the calcium binding protein hypothesis of selective vulnerability given the known vulnerability of the human hippocampal CA1 region in most patients with intractable temporal lobe epilepsy (Meldrum and Corsellis, ’84; Bruton, ’88). Therefore, this study addresses two questions: 1) which cells of the human hippocampus normally exhibit CaBPand parvalbumin-11 and 2) are the cells that normally express these proteins the least vulnerable in temporal lobe epilepsy?
MATERIALS AND METHODS Autopsy tissue Three inpatients at UCSF who died of non-neurological causes were refrigerated at 4°C within 1 hour of death. Patient 1 (DP) was a 58-year-old woman with anemia, bleeding and liver cirrhosis whose body was autopsied within 4 hours of death. Neuropathological examination revealed multiple white matter petechial hemorrhages. Patient 2 (JD) was a 61-year-old man with chronic myeloid leukemia. The autopsy was performed six hours postmortem and there was no evidence of intracranial hemorrhage. Patient 3 (SN) was a 19-year-old woman with ulcerative colitis. The autopsy was performed 18 hours postmortem and neuropathological examination showed gross normal appearance but microscopic evidence of hyperemia of the grey matter over both superior convexities in the vascular borderzone areas.
Surgical approach The hippocampal specimen was obtained in each case as the final part of an anterior temporal lobectomy. The Abbreutations CaBP g
h hf mf ml P
PV RZ S slm so
sr
calcium-binding protein; calbindin D28K dentate granule cell layer dentate hilus hippocampal fissure mossy fiber pathway dentate molecular layer pyramidal cell layer parvalbumin resistant zone subiculum stratum lacunosum-moleculare stratum oriens stratum radiatum
superior, middle, and inferior temporal and fusiform gyri were removed and the temporal horn of the lateral ventricle was entered. The white matter adjacent to the ventricle and the amygdala were removed, leaving the hippocampus and parahippocampal gyrus. The anterior 5 mm of the hippocampus was resected to identify the medial pial surface. The posterior margin of the hippocampal specimen was chosen (typically 2.0-2.5 cm from its tip) and a transverse section was made to the medial pia. The medial surface of the hippocampus and a portion of the parahippocampal gyms were then dissected from the medial pia. Perforating branches of the anterior choroidal artery were coagulated and divided as the final step of the resection. Thus, the hippocampal blood supply was preserved until just prior to removal of the specimen. The entire hippocampal resection took place over approximately 30 minutes and the interruption of the hippocampal blood supply over the final 10 minutes. The hippocampal tissue block was usually 1.5 cm of hippocampal length and included the most posterior part of the pes hippocampus and the most anterior 5-8 mm of the body of the hippocampus.
Immersion-fixationof human tissue “Normal” hippocampi removed at autopsy (n = 3) or “epileptic” hippocampi removed surgically (n = 38) were immersion fixed (in San Francisco) for 2-4 hours in 400 ml of 2% paraformaldehyde (Sigma Chemical Co., St. Louis, M0)/2% acrolein (Aldrich Chemical Go., Milwaukee, WI) in 0.1 M phosphate buffer, pH 7.4. The tissue was then transferred to 2% paraformaldehyde in 0.1 phosphate buffer, pH 7.4, and shipped overnight to New York. Two to four days after initial immersion, the tissue was embedded in agar and sectioned on avibratome in the plane perpendicular to the long axis of the hippocampus. For Nissl staining, 20-30 pthick sections were mounted directly onto chrom alum and gelatin-coated glass slides and air dried. Slides were dehydrated in graded alcohols and xylene, rehydrated, stained with 1.0% cresyl violet, dehydrated, and coverslipped with Permount. For immunocytochemical staining, 40 or 50 pthick sections were cut and collected in TRIS buffer. For each hippocampus, alternate sections were collected sequentially in different bins so that matching sets of sections from different parts of the hippocampus could be obtained for each of several different antisera.
Perfusion fixation of baboon (Papiopupio) To compare the patterns of localization of calciumbinding proteins in the human and a nonhuman primate, an adult female baboon (Pupiopapio) to be sacrificed for non-neurological health reasons was deeply anesthetized with sodium pentobarbital and perfusion-fixed by the first author and Dr. Brian Meldrum at the Institute of Psychiatry, London, UK. The baboon was perfused €or 3 minutes with 2% paraformaldehyde in 0.1 M sodium acetate buffer, pH 6.5, followed by 2% paraformaldehyde/O. 1%glutaraldehyde in 0.1 M sodium borate buffer, pH 8.5, for 15 minutes. The brain was removed immediately after perfusion and postfixed in 2% paraformaldehyde in 0.1 M sodium borate buffer, pH 8.5, for 3 days before sectioning. It should be noted that this species of primate exhibits photosensitive cortical epilepsy and that this animal had been induced to have seizures. Although hippocampal structure appeared normal and no data suggest temporal lobe involvement in the light-induced seizures, the possibility that this baboon
CALCIUM-BINDING PROTEINS IN HUMAN HIPPOCAMPUS hippocampus was abnormal in some way must be considered.
Tissue processing Immunocytochemical staining was performed as described previously (Sloviter, ’89). Primary rabbit antisera used were raised against the 28,000 MW Vitamin D-dependent CaBP from chick gut (antiserum diluted 15,000; provided by Dr. J.W. Pike) or against purified monkey cerebellar CaBP or rat muscle parvalbumin (both antisera diluted 1:1,000 or 1:2,000 and provided by Dr. K.G. Baimbridge). Controls were of two types; sections were either processed without antiserum or in normal rabbit serum diluted 1:1,000-15,000. Neither treatment resulted in specific staining. Sections were mounted on gelatin and chrom alum-coated slides, air-dried, dehydrated in graded ethanols and xylene, and coverslipped with Permount. Some sections were subsequently counterstained with dilute cresyl violet (0.1%)before coverslipping.
Antiserum specificity and cross reactivity Primary CaBP- or PV-antisera were pre-adsorbed overnight with purified chick gut CaBP or rat muscle PV, respectively, and then incubated with tissue sections as described previously (Sloviter, ’89). Pre-adsorption with as little as 10 ng purified CaBP or PV/ml diluted antiserum completely prevented the specific staining by CaBP or PV antiserum, respectively. Incubation of CaBP antiserum with a 10,000-fold higher concentration of PV (100 p,g purified rat PViml diluted antiserum), or PV antiserum with the same concentration of purified chick gut CaBP, had no noticeable effect on staining (Sloviter, ’89). Adsorption of CaBP and PV antisera with synthetic neuroactive peptides known to be present in hippocampal neurons addressed the cross-reactivity of the antisera. Adsorption with 100 pg peptidelml diluted CaBP or PV antiserum had no effect on specific staining for any of the synthetic peptides tested. These included sulphated cholecystokinin (CCK-8), vasoactive intestinal polypeptide, somatostatin,_,,, human neuropeptide Y, porcine neuropeptide Y, and Substance P (all from Peninsula Labs).
383
and end at the hippocampal fissure (patient 1; Fig. 2). In addition to its presence in granule and CA2 pyramidal cells, diffuse CaBP-LI is present in the stratum lacunosummoleculare of area CA1 and the subiculum (Fig. 1D). The vast majority of “normal” human dentate hilar neurons and virtually all CA3, CA1, and subicular pyramidal cells appear devoid of CaBP-LI (Figs. 1-3). This is in contrast to the baboon, which possesses CaBP-LI in CAI pyramidal cells (Fig. 1).
Parvalbumin-LI in the “normal” human hippocampus The staining for PV-LI was faint in the three autopsy controls compared to the dark PV staining in identically immersion-fixed surgical samples, and in the perfusionfixed baboon. We tentatively conclude that the postmortem delay in fixing the autopsy tissue may have been deleterious to PV-LI. Despite differences in stain quality (i.e., darkness), the general appearance of PV-LI was similar in both the “normal” autopsy and tumor-associated surgical specimens. Thus, photographs that illustrate the qualitative appearance of PV-LI cells in the human hippocampus are of the tumor-associated “epileptic” hippocampi, which appeared relatively normal histologically.
CaBP- and PV-LI in the tumor-associated epileptic hippocampus
CuBP-LZ. The pattern of CaBP-LI described above for the three autopsy patients was also seen in 11 “tumor” or “secondary lesion” patients whose hippocampi appeared relatively normal histologically (Fig. 4). PV-LI. The pattern of PV-LI in the immersion-fixed tumor-associated hippocampi was similar to that seen in normal perfusion-fixed rats (Sloviter, ’89) and the baboon (not shown). PV-LI is present only in presumed interneurons and their processes (Fig. 4C). In the area dentata, immunoreactive cells of different shapes are present in all layers. Figure 5 illustrates the morphological variety of PV-immunoreactive cell types in the dentate gyrus. These include multipolar cells of the hilus that send numerous dendrites into the molecular layer, cells whose processes are limited to the hilus, cells of the granule cell layer with dendrites that extend into both the molecular layer and the hilus, cells of the granule cell layer with the morphology of RESULTS pyramidal-shaped basket cells, and cells of the molecular CaBP-like immunoreactivity (LI) in the layer with dendrites in the molecular and granule cell “normal” human hippocampus layers. The only obvious PV-LI axonal plexus in the dentate CaBP-LI was localized in three brains from patients who gyrus was in the granule cell layer (Fig. 6A,B). In the died of non-neurological causes and the results were quali- tumor-malformation group, this plexus appeared to be tatively similar in all three hippocampi. The patterns of present throughout the granule cell layer (Fig. 6B). This staining produced by the two CaBP antisera were identical. was in contrast to the “cryptogenic” patient group (deAll photographs are of sections stained with the antiserum scribed below), which exhibited expanses of the granule cell raised against monkey cerebellar CaBP. Compared to the layer devoid of the PV-immunoreactive axo-somatic plexus pattern of CaBP immunoreactivity in the baboon (Fig. (Fig. 6D). In the hippocampus “proper,” PV-LI is present in relalA,B), the pattern in the “normal” human hippocampus is unique. The human hippocampus (patient 1;Figure 1C,D) tively small numbers of presumed interneurons. Usually, exhibits CaBP-LI in dentate granule cells, a variety of their apparently spineless but beaded dendrites are oripresumed interneurons of all subregions, and hippocampal ented parallel to those of the pyramidal cells (Fig. 4D). pyramidal cells of the “resistant zone,” which roughly However, the PV-LI cells are a morphologically heterogecorresponds to area CA2 of Lorente de N6 (’34). Labeled neous population and some cells, usually bipolar cells of cells contain CaBP-LI in all of their processes. Thus, dense stratum oriens, have dendrites oriented perpendicular to staining is present in granule cell dendrites of the dentate the pyramidal cell dendrites. The stratum radiatum is molecular layer, in the mossy fibers that innervate area devoid of PV-immunoreactive cell somata although the CA3, and in the dendrites of “resistant zone” neurons that dendrites of stratum pyramidale somata reach down into reach through strata radiatum and lacunosum-moleculare this layer (Fig. 4D). The PV-immunoreactive axon plexus
384
Fig. 1. Calcium-binding protein-like immunoreactivity (CaBP-LI) in the “normal” baboon and human hippocampus. A: Cresyl violetstained section of baboon hippocampus. B: CaBP-LI. No counterstain. Note that CA1 pyramidal cells are CaBP-immunoreactive in the baboon. C: Cresyl violet-stained human hippocampus. D: CaBP-LI
R.S. SLOVITER ET AL.
with cresyl violet counterstain. Note the immunoreactive granule cells ( G )and C A 2 pyramidal cells (area between arrowheads). Human CA1 pyramidal cells appear to be devoid of CaBP-LI. Magnification: X 17 in A and B; x 12.5 in C and D.
CALCIUM-BINDING PROTEINS IN HUMAN HIPPOCAMPUS
Fig. 2. CaBP-LI in the “normal” human hippocampus. A. CaBP-LI without counterstain. CaBP-LI is present primarily in dentate granule cells (G)and the “resistant” zone (area CA2; approximate area between arrowheads). B: CaBP-LI with cresyl violet counterstain showing the
385
detail of area CA2. Note immunoreactive pyramidal cells with stained dendrites. ML, dentate molecular layer; MF, CaBP-immunoreactive granule cell mossy fibers. Magnification: x36 in A x65 in B.
386
R.S. SLOVITER ET AL.
Fig. 3. CaBP-LI in the “normal” human dentate gyrus. Note immunoreactive granule cell somata (G) and dendrites in the molecular layer (ML). CaBP-immunoreactive hilar cells (arrowheads) are relatively rare. Magnification: x 249.
CALCIUM-BINDING PROTEINS IN HUMAN HIPPOCAMPUS
Fig. 4. CaBP- and PV-LI in a tumor-associated epileptic hippocampus. A: Cresyl violet. Note relatively normal hippocampal structure. B: CaBP-LI. Note that relatively few hilar (h) and CA1 neurons are CaBP-immunoreactive, whereas granule cells are immunoreactive. C : PV-LI. Note small number of PV-inimunoreactivecells in dentate hilus and granule cell layer. D: PV-immunoreactive cells in area CA1 are a small subset of the cells of this region. Note that horizontal, often
387
fusiform cells are present in the stratum oriens (so), whereas most stained cells of the pyramidal layer (p) are multipolar and possess beaded dendrites that traverse the stratum radiatum (sr) but do not enter stratum lacunosum-moleculare(slm). Patient KY. Abbreviations: g, granule cell layer; h, hilus; ml, dentate molecular layer. Magnification: x20 in A-C; x50 in D.
388
Fig. 5. PV-LI in the dentate gyrus of patients whose hippocampi were relatively normal histologically. All but one patient had tumors outside the hippocampus. Note that PV-LI is present only in presumed interneurons of different locations and morphology. A: Note in the hilus (h) numerous PV-immunoreactive neurons with dendrites that reach into the molecular layer. Also note cells in the granule cell ( g ) and molecular layers (ml). Patient JSB. B: PV-immunoreactive neuron of the molecular layer. Patient BW. C: immunoreactive cell of the granule
R.S. SLOVITER ET AL.
cell layer. Patient LOW. D: cells of the granule cell and molecular layers. Note PV-immunoreactivecell with dendrites in both the molecular layer and hilus. Patient ERW. E: cells of the hilus and granule cell layer. Patient KY. F-H: PV-immunoreactivity.Patient E M . Note that this patient exhibited relatively normal hippocampal structure and no other identified pathology in the tissues removed. Magnification: x20 in A x57 in B; x 79 in C ; x54 in D-H.
CALCIUM-BINDING PROTEINS IN HUMAN HIPPOCAMPUS
389
Fig. 6. PV-LI in dentate interneurons in tumor-associated and cryptogenic epileptic hippocampi. A Brightfield view of a tumorassociated epileptic hippocampus showing two cells immediately beneath the granule cell layer whose dendrites reach into the molecular layer and a third, horizontally oriented cell of the molecular layer. Patient KY. B: Darkfield view of the same region showing the PVimmunoreactive axonal plexus in the granule cell layer. C and D: Brightfield and darkfield views of a “cryptogenic” epileptic hippocam-
pus showing relatively normal PV-LI cells and axon plexus in an otherwise severely affected hilar region (Nissl-stained section in Fig. 6A). Note, however, that unlike the relatively normal hippocampus shown in A and B, the sclerotic hippocampus exhibits expanses of the granule cell layer devoid of the PV-immunoreactive axo-somatic plexus (asterisks). Patient CIC. Magnification: x61 in A and B; x80 in C and
appears to be restricted to the stratum pyramidale. Stratum lacunosum-moleculare appears to be devoid of both PV-immunoreactive cell bodies and Drocesses.
of surviving hippocampal cells in sclerotic hippocampi were CaBP-immunoreactive. Not surprisingly, the relatively resistant dentate a-anule and CA2 Dvramidd cells that were CaBP-immunoreactive in “normal” hippocampi (Fig. 1) were also CaBP-immunoreactive in “cryptogenic” sclerotic hippocampi. However, whereas relatively few of the many hilar neurons in “normal” and “tumor” epileptic hippocampi were CaBP-immunoreactive (Fig. 41,the majority of the few surviving hilar neurons in sclerotic hippocampi were CaBP-immunoreactive (or PV-immunoreactive as described below) and very darkly stained (Figs. 8-10). This apparent preferential survival of CaBP-immunoreactive cells was also evident in area CA1. This region, although virtually devoid of neurons in most sclerotic hippocampi, usually exhibited surviving CaBP-immunoreactive cells with relatively normal morphology (Figs. 8-10). Conversely, the subiculum, which was consistently spared in otherwise severely sclerotic hippocampi, contained CaBPlike immunoreactivity in presumed subicular interneurons, but not in the subicular pyramidal neurons ofboth “normal” (Fig. 1)and “cryptogenic” epileptic hippocampi (Figs. 8 and
CaBP- and PV-LI in the “cryptogenic” epileptic hippocampus Unlike most epileptic patients with tumors or malformations, patients with temporal lobe epilepsy of unknown cause (“cryptogenic” epilepsy), exhibited hippocampal sclerosis of varying degree. Our population of “cryptogenic” epilepsy patients exhibited patterns of hippocampal damage that ranged from relatively restricted hilar and CA3 cell loss (“endfolium sclerosis”) (Margerison and Corsellis, ’661, to total Ammon’s Horn sclerosis (Bruton, ’88), in which only a small number of granule cells and the “resistant zone” of CA2 pyramidal cells survived (Figs. 7-10). The most common pattern of damage involved severe loss of dentate hilar neurons, CA1 and CA3 pyramidal cells. Most “cryptogenic” patients exhibited survival of the “resistant zone” of pyramidal cells and many granule cells, which often appeared dispersed from their normal, tightly packed appearance (Fig. 9A and Houser, in press). CaBP-LI. The most notable finding in sclerotic hippocampi was that despite extensive cell loss, the vast majority
D.
-
I “
9).
PV-LI. There was an obvious and consistent loss of PV-LI in sclerotic hippocampi compared to hippocampi
390
Fig. 7. Pathology of “endfolium sclerosis” versus “classical” hippocampal sclerosis. A Nissl-stained section showing damage to the dentate hilus (h) and CA3 pyramidal cells with survival of much of area CA1. Patient CIC. B: Classical hippocampal sclerosis. Note loss of hilar
R.S. SLOVITER ET AL.
(h), CA3, and CA1 cells (between arrowheads) with survival of some granule cells, cells of the resistant zone (CA21, and subicular neurons (S).Patient JWF. Magnification: x 16.
CALCIUM-BINDING PROTEINS IN HUMAN HIPPOCAMPUS
Fig. 8. CaBP- and PV-LI in sclerotic “cryptogenic” epileptic hippocampus. A Cresyl violet. Note extensive cell loss in the hilus (H) and area CA1 and the survival of subicular neurons ( S ) .B: CaBP-LI. Note immunoreactivity in many of the remaining cells. C: PV-LI in an
391
alternate section of the same hippocampus showing the loss of PV-LI in area CA1 and the dentate granule cell region (GI. Note the relatively normal staining in the surviving subiculurn ( S ) adjacent to the affected CA1 region. Patient MAB. Magnification: X 16.
R.S. SLOVITER ET AL.
392
Fig. 9. CaBP-LI in severe hippocampal sclerosis: A Cresyl violet. Note extensive loss of granule cells (GI, dispersion of remaining granule cells, and extensive loss of hilar (H) and CA1 neurons. Cells of the resistant zone (RZ) and the subiculum (S) remain. B: CaBP-LI showing that many of the few surviving neurons of the granule cell layer. hilus and resistant zone (asterisks) are CaBP-immunoreactive. Patient MT. Magnification: ~ 1 9 .
from “tumor” epileptics. Both the number of PV-LI neurons and the density of immunoreactive fibers were obviously and consistently reduced (Fig. 8). This was not due to vagaries in the staining since relatively normal PV staining was apparent in the relatively undamaged subiculum of each section. In addition, the decrease in PV-LI was roughly proportional to the degree of cell loss. In a patient with “endfolium sclerosis” (patient CIC), characterized by hilar and some CA3 pyramidal cell loss but survival of many CA1 neurons (Fig. 7A), the loss of PV-LI was restricted to the
damaged regions; the rest of the pyramidal cell layer exhibited relatively normal PV-LI. More severely sclerotic hippocampi routinely exhibited extensive loss of PV-LI in regions of cell loss with relatively normal PV staining in the surviving subicular region of each hippocampus. One consistent difference between the “tumor” patient group and the “cryptogenic” group was the loss of staining of the PV-immunoreactive axo-somatic plexus in the surviving granule cell layer. “Tumor” group patients generally exhibited a plexus throughout the granule cell layer (Fig.
CALCIUM-BINDING PROTEINS IN HUMAN HIPPOCAMPUS
393
Fig. 10. CaBP-LI in a “cryptogenic” epileptic hippocampus. A Higher magnification of section in Figure 9 showing cellular detail. Note CaBP-immunoreactive granule cells (row of arrowheads at left), cells of the resistant zone (RZ), and the abnormally numerous and darkly stained nongranule cells in the granule cell layer (GI, molecular layer (ML), and hilus (H). B: Detail of hilar cells shown in A in the region of the single arrowhead. Note CaBP-immunoreactive cells of
various morphology including a large multipolar neuron (double arrowheads) with dendrites entering the granule cell layer, and small cells (single arrowheads) stained more darkly than granule cells (visible in left portion of panel A). C : detail of resistant zone shown in A. Note survival of presumed pyramidal cells (arrowheads) and presumed interneurons of several locations (asterisks). Patient MT. hf, hippocampal fissure. Magnification: X37 in A, x 124 in B; X 82 in C.
6B), whereas “cryptogenic” patients exhibited expanses of the granule cell layer with no detectable PV-immunoreactive axo-somatic plexus immediately adjacent to expanses with a visible plexus (Fig. 6D). In virtually all specimens, however, some PV-LI neurons survived and appeared similar in appearance to cells in the “tumor” specimens. Thus, despite an obvious decrease in PV-LI in sclerotic hippocampi, PV-immunoreactive cells with relatively normal morphology were usually among the small number of surviving neurons (Fig. 5B,C, Fig. 11).Whether the decreased PV immunostaining represents inhibitory cell death
or simply a reduction in immunocytochemically detectable PV is not resolved by these methods.
DISCUSSION CaBP- and PV-LI in the “normal” hippocampus The main finding of this study is that compared to the patterns of CaBP-like immunoreactivity in the hippocampi of lower mammals, the pattern in the normal human brain
394
R.S. SLOVITER ET AL.
Fig. 11. Survival of dentate hilar CaBP- and PV-LI neurons in sclerotic “cryptogenic” epileptic hippocampus. A Cresyl violet-stained section showing survival of many granule cells (g) but few hilar (h) neurons (arrowheads). B: CaBP-LI in an alternate section of the same hippocampus showing a surviving hilar neuron to be CaBP-immunoreactive. C and D: PV-LI in alternate sections of the same hippo-
campus showing a surviving PV-immunoreactive hilar neuron (C) and presumed dentate basket cells (D). Patient LN. E and F: Surviving PV-immunoreactive interneurons of the granule cell layer (g) and molecular layer (ml) in patients BLT (E) and GRS (F). Note that the light staining of the granule cell layer is nonspecific background staining. Magnification: x80 in A-C and F; x 60 in D and E.
is unique. Rats and mice exhibit CaBP-LI in granule cells, many CA1 and CA2 pyramidal cells, and many interneurons (rat: Sloviter, ‘89; mouse: Sloviter, unpublished). Conversely, rabbits and guinea pigs exhibit CaBP-LI in granule cells and numerous interneurons, but not in any pyramidal cells (Sloviter, unpublished). The baboon (Pupio pupzo) exhibits CaBP-LI in dentate granule cells, interneurons of all regions, and virtually all CA1 and CA2 pyramidal cells. Conversely, the human contains CaBP-LI only in
dentate granule cells, CA2 pyramidal cells, and interneurons. Thus, the human is the only species examined in which the normal distribution of CaBP-LI corresponds to the least vulnerable cell populations. The pattern of CaBP-LI in the “normal” human hippocampus appears similar to the staining by an antibody raised against minced human pheochromocytoma tissue and attributed to its recognition of an adrenal chromaffin cell protein, chromogranin A (Munoz, ’90). The epitope
CALCIUM-BINDING PROTEINS IN HUMAN HIPPOCAMPUS recognized by this antibody has not been identified and it is not known if the antibody recognizes CaBP, which is present in the adrenal gland. Thus, it is not known if the recently reported distribution of chromogranin A immunoreactivity in the human hippocampus represents native chromogranin A or another cellular constituent. In contrast to the unique CaBP-staining pattern seen in the human, the pattern of PV-LI in the human appears very similar to that seen in lower animals. PV-LI is present in a specific subset of hippocampal interneurons that have been identified in animals as GABA neurons (Kosaka et al., ’87). In this study, relatively faint PV staining was obtained in the autopsy tissues. Conversely, excellent staining, similar to that seen in the perfusion-fixed baboon, was obtained in the surgical samples, suggesting some postmortem deterioration of PV-LI in the autopsy controls. Although the PV staining in tumor-associated patient tissue may be similar to what would be obtained in ideal control tissue, it cannot be assumed to be truly “normal” despite the fact that the human hippocampi appeared relatively normal in histological appearance. In fact, the hippocampi of tumor-associated epileptics are not normal. Although they appear relatively normal qualitatively, cell counts have revealed that tumorassociated epileptic patients often exhibit subtle, but statistically significant, cell loss (J. Kim and D.D. Spencer, personal communication).
395
simple matter to determine definitively if the PV-immunoreactive plexus has degenerated or if the GABAimmunoreactive plexus is, or is not, normal. These results in experimental animals (Sloviter, ’91) and epileptic human hippocampi may represent a loss of a subset of GABA neurons not easily evaluated by GABA- or GAD-staining alone (Babb et al., ’89). However, the functional implications, if any, of the loss of PV staining are unclear since, regardless of whether or not PV cell loss has occurred, immunocytochemical staining for PV, GABA, or GAD provides no meaningful information about the state of inhibition in the tissue before its removal. The loss of some dentate granule cells and possible loss of some inhibitory cells that contain supposedly protective substances is paradoxical. Does the survival of many calcium-binding protein-containing cells in sclerotic hippocampi represent the end result of a protective process or does the loss of some of these neurons suggest that CaBP and PV are not protective? Clearly, all cells are ultimately vulnerable since some epileptics exhibit virtually complete loss of hippocampal neurons. It seems most reasonable to conclude that protective mechanisms exist and are competitive in nature in that they can be overcome if the insult is sufficiently intense. What remains to be determined is the identity of the operative protective mechanisms, how they are regulated, and how they can be altered experimentally.
CaBP- and PV-LI in the epileptic hippocampus One of t h e main findings made in t h e sclerotic “cryptogenic” epileptic hippocampus was that a large proportion of the few surviving cells were CaBP- or PVimmunoreactive. In addition, the number and dark staining of surviving hilar neurons in sclerotic hippocampi compared to autopsy controls or hippocampi from the tumorassociated patient group suggests that CaBP synthesis may have been increased in the surviving cells. These results are consistent with the suggestion that the excitation-induced influx of calcium initiates epileptic brain damage (Griffiths et al., ’82) and with the hypothesis that the selective vulnerability of specific cell types may be related to differences between cell populations in their abilities to buffer intracellular calcium (Sloviter, ’89). Although the presence of these proteins in surviving human hippocampal neurons does not prove that calcium-binding proteins play a protective role, it has been demonstrated in rat hippocampal slices that increasing the calcium-binding capacity of identified vulnerable dentate hilar cells is protective against excitation-induced irreversible depolarization (Scharfman and Schwartzkroin, ’89). The second possibly significant finding in the sclerotic hippocampi is that large expanses of the dentate granule cell layer appear to be devoid of their PV-immunoreactive, and presumably inhibitory, axo-somatic plexus. Since PVimmunoreactive cells are a subset of GABA neurons (Kosaka et al., ’87), a loss of PV immunoreactivity may mean that a specific subpopulation of inhibitory neurons may have been lost. This loss of PV staining of the axo-somatic plexus in the granule cell layer is virtually identical to the permanently decreased PV staining seen in an experimental epilepsy model in which hippocampal sclerosis is produced by sustained perforant path stimulation (Sloviter, ’91).However, in the experimental animals, no obvious loss of axo-somatic GABA staining occurred. GABA staining of the human tissue (results to be published separately) was more variable than in perfusion-fixed animals so it is not a
ACKNOWLEDGMENTS We thank Dr. Brian Meldrum for providing the baboon tissue, Edward Ronk and Brian Yarborough for printing the photographs, Dr. J.W. Pike for providing CaBP antiserum, and Dr. K.G. Baimbridge for providing CaBP and PV antisera. This research was supported by NIH grant NS1820 1.
LITERATURE CITED Babb, T.L., J.K. Pretorius, W.R. Kupfer, and P.H. Crandall (1989) Glutamate decarboxylase-immunoreactive neurons are preserved in human epileptic hippocampus. J. Neurosci. 9.9562-2574. Baimbridge, K.G., and J.J.Miller (1982) Immunohistochemical localization of calcium-binding protein in the cerebellum, hippocampal formation and olfactory bulb of the rat. Brain Res. 245.223-229. Bruton, C.J. (1988) The Neuropathology of Temporal Lobe Epilepsy. New York: Oxford University Press. Connor, J.A., W.J. Wadman, P.E. Hockberger, and R.K.S. Wong (1988) Sustained dendritic gradients of Ca2+induced by excitatoly amino acids in CA1 hippocampal neurons. Science 240:649-653. Griffiths, T., M.C. Evans, and B.S. Meldrum (1982) Intracellular sites of early calcium accumulation in the rat hippocampus during status epilepticus. Neurosci. Lett. 30:329-334. Houser, C.R. (1990) Granule cell dispersion in the dentate gyrus of humans with temporal lobe epilepsy. Brain Research, 535: 195-204. Kosaka, T., H. Katsumaru, K. Hama, J.-Y. Wu, and C.W. Heizmann (1987) GABAergic neurons containing the Ca2+-bindingprotein parvalbumin in the rat hippocampus and dentate g y r u s . Brain Res. 419r119-130. Lorente de N6, R. (1934) Studies on the structure of the cerebral cortex. 11. Continuation of the study of the ammonic system. J. Psychol. Neurol. 46:113-117. Margerison, J.H., and J.A.N. Corsellis (1966) Epilepsy and the temporal lobes. Brain 89:499-530. Meldrum, B.S. (1981) Metabolic effects of prolonged epileptic seizures and the causation of epileptic brain damage. In F.C. Rose (ed): Metabolic Disorders of the Nervous System. London: Pitman, pp. 175-187. Meldrum, B.S., and Corsellis, J.A.N. Epilepsy. (1984) In J.A.N. Corsellis and L.W. Duchen (eds): Greenfield’s Neuropathology. New York: John Wiley & Sons, pp. 921-950.
396 Morrissey, R.L., R.N. Empson, D.T. Zolock, D.D. Bikle, andT.J. Bucci (1978) 11.A timed study of Intestinal response to 1,25-dihydroxycholecalciferol. the intracellular localization of calcium-binding protein. Biochim. Biophys. Acta 538:34-41. Munoz, D.G. (1990) Thedistribution of chromogranin A-like immunoreactivity in the human hippocampus coincides with the pattern of resistance to epilepsy-induced neuronal damage. Ann. Neurol. 27:266-275. Scharfman, H.E., and P.A. Schwartzkroin (1989) Protection of dentate hilar mossy cells from prolonged stimulation by intracellular calcium chelation. Science 246r257-260. Sloviter, R.S. (1987) Decreased hippocampal inhibition and a selective loss of interneurons in experimental epilepsy. Science 23573-76.
R.S. SLOVITER ET AL. Sloviter, R.S. (1989) Calcium-binding protein (Calbindin-D28k) and parvalbumin immunocytochemistry: Localization in the rat hippocampus with specific reference to the selective vulnerability of hippocampal neurons to seizure activity. J. Comp. Neurol. 280:183-196. Sloviter, R.S. (1991) Permanently altered hippocampal structure, excitability and inhibition after experimental status epilepticus in the rat; the “dormant basket ceil” hypothesis and its possible relevance to temporal lobe epilepsy. Hippocampus Ir41-66. Sloviter, R.S., N.M. Barbaro, and K.D. Laxer (1989) Calcium-binding protein (calbindin) and pawalbumin-like immunoreactivity (LI) in the “normal” and “epileptic” hippocampus. Epilepsia 30;719.
Calcium-Binding Protein (Calbindin-D28K) and Parvalbumin Immunocytochemistry in the Normal and Epileptic Human Hippocampus ROBERT S. SLOVITER, ANNE L. SOLLAS, NICHOLAS M. BARBARO, AND KENNETH D. LAXER Neurology Research Center, Helen Hayes Hospital, New York State Department of Health, West Haverstraw, New York 10993 (R.S.S.,A.L.S.), Departments of Pharmacology and Neurology, Columbia University College of Physicians and Surgeons, New York, New York 10032 (R.S.S.), Departments of Neurosurgery (N.M.B.) and Neurology (K.D.L.), University of California San Francisco, San Francisco, California 94143
ABSTRACT The calcium-binding proteins calbindin-D28K (CaBP) and parvalbumin (PV) were localized in the “normal” and “epileptic” human hippocampus to address the possible relationship between the expression of these constitutive cytosolic calcium-binding proteins and the resistance or selective vulnerability of different hippocampal neuron populations in temporal lobe epilepsy. Compared to rodents and a baboon (Papiopupzo), the pattern of CaBP-like immunoreactivity (LI) in the “normal” human hippocampus is unique. CaBP-LI is present in the dentate granule cells, neurons of the “resistant zone” (area CAZ), and presumed interneurons of all regions. Unlike rodent and baboon CA1 pyramidal cells, human CA1 pyramidal cells appear to be devoid of CaBP-LI. Thus, the relatively resistant dentate granule cells and CA2 pyramidal cells are the only human hippocampal principal cells that contain CaBP-LI normally. As in lower mammals, PV-LI is present exclusively in interneurons of all human hippocampal subregions. CaBP- and PV-LI were localized in hippocampi surgically removed in the treatment of intractable temporal lobe epilepsy to determine whether surviving hippocampal cells were those that express these calcium-binding proteins. Hippocampi removed from patients with tumors or arteriovenous malformations that were associated with complex partial seizures arising from this region appeared relatively normal histologically. CaBP- and PV-LI in this patient group appeared similar to that seen in autopsy controls. Conversely, “cryptogenic” epileptics, who exhibit hippocampal sclerosis as the only lesion associated with the epilepsy, exhibited a preferential survival of hippocampal cells that were CaBP- or PV-immunoreactive. In the dentate hilus, which normally contains few CaBP-LI neurons, most of the few surviving hilar neurons were CaBP-immunoreactive. Their number and darkness of staining suggests that CaBP synthesis may be increased in cells that survive. Despite an obvious decrease of PV-LI specificallyin the damaged parts of the sclerotic hippocampi, PV-immunoreactive interneurons were often among the few surviving cells. Nevertheless, large expanses of the surviving granule cell layer appeared to have lost the PV-immunoreactive axosomatic fiber plexus. These results reveal a unique and striking correlation between the human hippocampal cells that normally express these calcium-binding proteins and those that survive in the sclerotic epileptic hippocampus. Key words: temporal lobe epilepsy, dentate gyrus, neuropathology
Accepted January 24, 1991.
D
1991 WILEY-LISS, INC.
R.S. SLOVITER ET AL.
382
The increase in intracellular calcium (Griffiths et al., ’82) that occurs as a result of excitatory amino acid receptor activation (Connor et al., ’88) has been suggested to be the initiating factor in seizure-associated neuronal death (Meldrum, ’81). However, although most or all hippocampal neurons are presumably excited during seizures, some cell populations are more vulnerable than others. It has been hypothesized that differences in the vulnerability of hippocampal neurons to seizure activity may be related to the absence or presence of cytoplasmic calcium-binding proteins (Sloviter, ’89) that buffer free intracellular calcium (Morrissey et al., ’78; Baimbridge and Miller, ’82). In the rat, there is a striking correlation between the cells that survive intense seizure activity in an experimental epilepsy model (Sloviter, ’87) and those that possess either of the calcium binding proteins calbindin-D28K (CaBP) and parvalbumin (Sloviter, ’89).However, the presence of CaBPlike immunoreactivity (CaBP-LI) in rat and baboon CA1 pyramidal cells (Sloviter et al., ’89) is inconsistent with the calcium binding protein hypothesis of selective vulnerability given the known vulnerability of the human hippocampal CA1 region in most patients with intractable temporal lobe epilepsy (Meldrum and Corsellis, ’84; Bruton, ’88). Therefore, this study addresses two questions: 1) which cells of the human hippocampus normally exhibit CaBPand parvalbumin-11 and 2) are the cells that normally express these proteins the least vulnerable in temporal lobe epilepsy?
MATERIALS AND METHODS Autopsy tissue Three inpatients at UCSF who died of non-neurological causes were refrigerated at 4°C within 1 hour of death. Patient 1 (DP) was a 58-year-old woman with anemia, bleeding and liver cirrhosis whose body was autopsied within 4 hours of death. Neuropathological examination revealed multiple white matter petechial hemorrhages. Patient 2 (JD) was a 61-year-old man with chronic myeloid leukemia. The autopsy was performed six hours postmortem and there was no evidence of intracranial hemorrhage. Patient 3 (SN) was a 19-year-old woman with ulcerative colitis. The autopsy was performed 18 hours postmortem and neuropathological examination showed gross normal appearance but microscopic evidence of hyperemia of the grey matter over both superior convexities in the vascular borderzone areas.
Surgical approach The hippocampal specimen was obtained in each case as the final part of an anterior temporal lobectomy. The Abbreutations CaBP g
h hf mf ml P
PV RZ S slm so
sr
calcium-binding protein; calbindin D28K dentate granule cell layer dentate hilus hippocampal fissure mossy fiber pathway dentate molecular layer pyramidal cell layer parvalbumin resistant zone subiculum stratum lacunosum-moleculare stratum oriens stratum radiatum
superior, middle, and inferior temporal and fusiform gyri were removed and the temporal horn of the lateral ventricle was entered. The white matter adjacent to the ventricle and the amygdala were removed, leaving the hippocampus and parahippocampal gyrus. The anterior 5 mm of the hippocampus was resected to identify the medial pial surface. The posterior margin of the hippocampal specimen was chosen (typically 2.0-2.5 cm from its tip) and a transverse section was made to the medial pia. The medial surface of the hippocampus and a portion of the parahippocampal gyms were then dissected from the medial pia. Perforating branches of the anterior choroidal artery were coagulated and divided as the final step of the resection. Thus, the hippocampal blood supply was preserved until just prior to removal of the specimen. The entire hippocampal resection took place over approximately 30 minutes and the interruption of the hippocampal blood supply over the final 10 minutes. The hippocampal tissue block was usually 1.5 cm of hippocampal length and included the most posterior part of the pes hippocampus and the most anterior 5-8 mm of the body of the hippocampus.
Immersion-fixationof human tissue “Normal” hippocampi removed at autopsy (n = 3) or “epileptic” hippocampi removed surgically (n = 38) were immersion fixed (in San Francisco) for 2-4 hours in 400 ml of 2% paraformaldehyde (Sigma Chemical Co., St. Louis, M0)/2% acrolein (Aldrich Chemical Go., Milwaukee, WI) in 0.1 M phosphate buffer, pH 7.4. The tissue was then transferred to 2% paraformaldehyde in 0.1 phosphate buffer, pH 7.4, and shipped overnight to New York. Two to four days after initial immersion, the tissue was embedded in agar and sectioned on avibratome in the plane perpendicular to the long axis of the hippocampus. For Nissl staining, 20-30 pthick sections were mounted directly onto chrom alum and gelatin-coated glass slides and air dried. Slides were dehydrated in graded alcohols and xylene, rehydrated, stained with 1.0% cresyl violet, dehydrated, and coverslipped with Permount. For immunocytochemical staining, 40 or 50 pthick sections were cut and collected in TRIS buffer. For each hippocampus, alternate sections were collected sequentially in different bins so that matching sets of sections from different parts of the hippocampus could be obtained for each of several different antisera.
Perfusion fixation of baboon (Papiopupio) To compare the patterns of localization of calciumbinding proteins in the human and a nonhuman primate, an adult female baboon (Pupiopapio) to be sacrificed for non-neurological health reasons was deeply anesthetized with sodium pentobarbital and perfusion-fixed by the first author and Dr. Brian Meldrum at the Institute of Psychiatry, London, UK. The baboon was perfused €or 3 minutes with 2% paraformaldehyde in 0.1 M sodium acetate buffer, pH 6.5, followed by 2% paraformaldehyde/O. 1%glutaraldehyde in 0.1 M sodium borate buffer, pH 8.5, for 15 minutes. The brain was removed immediately after perfusion and postfixed in 2% paraformaldehyde in 0.1 M sodium borate buffer, pH 8.5, for 3 days before sectioning. It should be noted that this species of primate exhibits photosensitive cortical epilepsy and that this animal had been induced to have seizures. Although hippocampal structure appeared normal and no data suggest temporal lobe involvement in the light-induced seizures, the possibility that this baboon
CALCIUM-BINDING PROTEINS IN HUMAN HIPPOCAMPUS hippocampus was abnormal in some way must be considered.
Tissue processing Immunocytochemical staining was performed as described previously (Sloviter, ’89). Primary rabbit antisera used were raised against the 28,000 MW Vitamin D-dependent CaBP from chick gut (antiserum diluted 15,000; provided by Dr. J.W. Pike) or against purified monkey cerebellar CaBP or rat muscle parvalbumin (both antisera diluted 1:1,000 or 1:2,000 and provided by Dr. K.G. Baimbridge). Controls were of two types; sections were either processed without antiserum or in normal rabbit serum diluted 1:1,000-15,000. Neither treatment resulted in specific staining. Sections were mounted on gelatin and chrom alum-coated slides, air-dried, dehydrated in graded ethanols and xylene, and coverslipped with Permount. Some sections were subsequently counterstained with dilute cresyl violet (0.1%)before coverslipping.
Antiserum specificity and cross reactivity Primary CaBP- or PV-antisera were pre-adsorbed overnight with purified chick gut CaBP or rat muscle PV, respectively, and then incubated with tissue sections as described previously (Sloviter, ’89). Pre-adsorption with as little as 10 ng purified CaBP or PV/ml diluted antiserum completely prevented the specific staining by CaBP or PV antiserum, respectively. Incubation of CaBP antiserum with a 10,000-fold higher concentration of PV (100 p,g purified rat PViml diluted antiserum), or PV antiserum with the same concentration of purified chick gut CaBP, had no noticeable effect on staining (Sloviter, ’89). Adsorption of CaBP and PV antisera with synthetic neuroactive peptides known to be present in hippocampal neurons addressed the cross-reactivity of the antisera. Adsorption with 100 pg peptidelml diluted CaBP or PV antiserum had no effect on specific staining for any of the synthetic peptides tested. These included sulphated cholecystokinin (CCK-8), vasoactive intestinal polypeptide, somatostatin,_,,, human neuropeptide Y, porcine neuropeptide Y, and Substance P (all from Peninsula Labs).
383
and end at the hippocampal fissure (patient 1; Fig. 2). In addition to its presence in granule and CA2 pyramidal cells, diffuse CaBP-LI is present in the stratum lacunosummoleculare of area CA1 and the subiculum (Fig. 1D). The vast majority of “normal” human dentate hilar neurons and virtually all CA3, CA1, and subicular pyramidal cells appear devoid of CaBP-LI (Figs. 1-3). This is in contrast to the baboon, which possesses CaBP-LI in CAI pyramidal cells (Fig. 1).
Parvalbumin-LI in the “normal” human hippocampus The staining for PV-LI was faint in the three autopsy controls compared to the dark PV staining in identically immersion-fixed surgical samples, and in the perfusionfixed baboon. We tentatively conclude that the postmortem delay in fixing the autopsy tissue may have been deleterious to PV-LI. Despite differences in stain quality (i.e., darkness), the general appearance of PV-LI was similar in both the “normal” autopsy and tumor-associated surgical specimens. Thus, photographs that illustrate the qualitative appearance of PV-LI cells in the human hippocampus are of the tumor-associated “epileptic” hippocampi, which appeared relatively normal histologically.
CaBP- and PV-LI in the tumor-associated epileptic hippocampus
CuBP-LZ. The pattern of CaBP-LI described above for the three autopsy patients was also seen in 11 “tumor” or “secondary lesion” patients whose hippocampi appeared relatively normal histologically (Fig. 4). PV-LI. The pattern of PV-LI in the immersion-fixed tumor-associated hippocampi was similar to that seen in normal perfusion-fixed rats (Sloviter, ’89) and the baboon (not shown). PV-LI is present only in presumed interneurons and their processes (Fig. 4C). In the area dentata, immunoreactive cells of different shapes are present in all layers. Figure 5 illustrates the morphological variety of PV-immunoreactive cell types in the dentate gyrus. These include multipolar cells of the hilus that send numerous dendrites into the molecular layer, cells whose processes are limited to the hilus, cells of the granule cell layer with dendrites that extend into both the molecular layer and the hilus, cells of the granule cell layer with the morphology of RESULTS pyramidal-shaped basket cells, and cells of the molecular CaBP-like immunoreactivity (LI) in the layer with dendrites in the molecular and granule cell “normal” human hippocampus layers. The only obvious PV-LI axonal plexus in the dentate CaBP-LI was localized in three brains from patients who gyrus was in the granule cell layer (Fig. 6A,B). In the died of non-neurological causes and the results were quali- tumor-malformation group, this plexus appeared to be tatively similar in all three hippocampi. The patterns of present throughout the granule cell layer (Fig. 6B). This staining produced by the two CaBP antisera were identical. was in contrast to the “cryptogenic” patient group (deAll photographs are of sections stained with the antiserum scribed below), which exhibited expanses of the granule cell raised against monkey cerebellar CaBP. Compared to the layer devoid of the PV-immunoreactive axo-somatic plexus pattern of CaBP immunoreactivity in the baboon (Fig. (Fig. 6D). In the hippocampus “proper,” PV-LI is present in relalA,B), the pattern in the “normal” human hippocampus is unique. The human hippocampus (patient 1;Figure 1C,D) tively small numbers of presumed interneurons. Usually, exhibits CaBP-LI in dentate granule cells, a variety of their apparently spineless but beaded dendrites are oripresumed interneurons of all subregions, and hippocampal ented parallel to those of the pyramidal cells (Fig. 4D). pyramidal cells of the “resistant zone,” which roughly However, the PV-LI cells are a morphologically heterogecorresponds to area CA2 of Lorente de N6 (’34). Labeled neous population and some cells, usually bipolar cells of cells contain CaBP-LI in all of their processes. Thus, dense stratum oriens, have dendrites oriented perpendicular to staining is present in granule cell dendrites of the dentate the pyramidal cell dendrites. The stratum radiatum is molecular layer, in the mossy fibers that innervate area devoid of PV-immunoreactive cell somata although the CA3, and in the dendrites of “resistant zone” neurons that dendrites of stratum pyramidale somata reach down into reach through strata radiatum and lacunosum-moleculare this layer (Fig. 4D). The PV-immunoreactive axon plexus
384
Fig. 1. Calcium-binding protein-like immunoreactivity (CaBP-LI) in the “normal” baboon and human hippocampus. A: Cresyl violetstained section of baboon hippocampus. B: CaBP-LI. No counterstain. Note that CA1 pyramidal cells are CaBP-immunoreactive in the baboon. C: Cresyl violet-stained human hippocampus. D: CaBP-LI
R.S. SLOVITER ET AL.
with cresyl violet counterstain. Note the immunoreactive granule cells ( G )and C A 2 pyramidal cells (area between arrowheads). Human CA1 pyramidal cells appear to be devoid of CaBP-LI. Magnification: X 17 in A and B; x 12.5 in C and D.
CALCIUM-BINDING PROTEINS IN HUMAN HIPPOCAMPUS
Fig. 2. CaBP-LI in the “normal” human hippocampus. A. CaBP-LI without counterstain. CaBP-LI is present primarily in dentate granule cells (G)and the “resistant” zone (area CA2; approximate area between arrowheads). B: CaBP-LI with cresyl violet counterstain showing the
385
detail of area CA2. Note immunoreactive pyramidal cells with stained dendrites. ML, dentate molecular layer; MF, CaBP-immunoreactive granule cell mossy fibers. Magnification: x36 in A x65 in B.
386
R.S. SLOVITER ET AL.
Fig. 3. CaBP-LI in the “normal” human dentate gyrus. Note immunoreactive granule cell somata (G) and dendrites in the molecular layer (ML). CaBP-immunoreactive hilar cells (arrowheads) are relatively rare. Magnification: x 249.
CALCIUM-BINDING PROTEINS IN HUMAN HIPPOCAMPUS
Fig. 4. CaBP- and PV-LI in a tumor-associated epileptic hippocampus. A: Cresyl violet. Note relatively normal hippocampal structure. B: CaBP-LI. Note that relatively few hilar (h) and CA1 neurons are CaBP-immunoreactive, whereas granule cells are immunoreactive. C : PV-LI. Note small number of PV-inimunoreactivecells in dentate hilus and granule cell layer. D: PV-immunoreactive cells in area CA1 are a small subset of the cells of this region. Note that horizontal, often
387
fusiform cells are present in the stratum oriens (so), whereas most stained cells of the pyramidal layer (p) are multipolar and possess beaded dendrites that traverse the stratum radiatum (sr) but do not enter stratum lacunosum-moleculare(slm). Patient KY. Abbreviations: g, granule cell layer; h, hilus; ml, dentate molecular layer. Magnification: x20 in A-C; x50 in D.
388
Fig. 5. PV-LI in the dentate gyrus of patients whose hippocampi were relatively normal histologically. All but one patient had tumors outside the hippocampus. Note that PV-LI is present only in presumed interneurons of different locations and morphology. A: Note in the hilus (h) numerous PV-immunoreactive neurons with dendrites that reach into the molecular layer. Also note cells in the granule cell ( g ) and molecular layers (ml). Patient JSB. B: PV-immunoreactive neuron of the molecular layer. Patient BW. C: immunoreactive cell of the granule
R.S. SLOVITER ET AL.
cell layer. Patient LOW. D: cells of the granule cell and molecular layers. Note PV-immunoreactivecell with dendrites in both the molecular layer and hilus. Patient ERW. E: cells of the hilus and granule cell layer. Patient KY. F-H: PV-immunoreactivity.Patient E M . Note that this patient exhibited relatively normal hippocampal structure and no other identified pathology in the tissues removed. Magnification: x20 in A x57 in B; x 79 in C ; x54 in D-H.
CALCIUM-BINDING PROTEINS IN HUMAN HIPPOCAMPUS
389
Fig. 6. PV-LI in dentate interneurons in tumor-associated and cryptogenic epileptic hippocampi. A Brightfield view of a tumorassociated epileptic hippocampus showing two cells immediately beneath the granule cell layer whose dendrites reach into the molecular layer and a third, horizontally oriented cell of the molecular layer. Patient KY. B: Darkfield view of the same region showing the PVimmunoreactive axonal plexus in the granule cell layer. C and D: Brightfield and darkfield views of a “cryptogenic” epileptic hippocam-
pus showing relatively normal PV-LI cells and axon plexus in an otherwise severely affected hilar region (Nissl-stained section in Fig. 6A). Note, however, that unlike the relatively normal hippocampus shown in A and B, the sclerotic hippocampus exhibits expanses of the granule cell layer devoid of the PV-immunoreactive axo-somatic plexus (asterisks). Patient CIC. Magnification: x61 in A and B; x80 in C and
appears to be restricted to the stratum pyramidale. Stratum lacunosum-moleculare appears to be devoid of both PV-immunoreactive cell bodies and Drocesses.
of surviving hippocampal cells in sclerotic hippocampi were CaBP-immunoreactive. Not surprisingly, the relatively resistant dentate a-anule and CA2 Dvramidd cells that were CaBP-immunoreactive in “normal” hippocampi (Fig. 1) were also CaBP-immunoreactive in “cryptogenic” sclerotic hippocampi. However, whereas relatively few of the many hilar neurons in “normal” and “tumor” epileptic hippocampi were CaBP-immunoreactive (Fig. 41,the majority of the few surviving hilar neurons in sclerotic hippocampi were CaBP-immunoreactive (or PV-immunoreactive as described below) and very darkly stained (Figs. 8-10). This apparent preferential survival of CaBP-immunoreactive cells was also evident in area CA1. This region, although virtually devoid of neurons in most sclerotic hippocampi, usually exhibited surviving CaBP-immunoreactive cells with relatively normal morphology (Figs. 8-10). Conversely, the subiculum, which was consistently spared in otherwise severely sclerotic hippocampi, contained CaBPlike immunoreactivity in presumed subicular interneurons, but not in the subicular pyramidal neurons ofboth “normal” (Fig. 1)and “cryptogenic” epileptic hippocampi (Figs. 8 and
CaBP- and PV-LI in the “cryptogenic” epileptic hippocampus Unlike most epileptic patients with tumors or malformations, patients with temporal lobe epilepsy of unknown cause (“cryptogenic” epilepsy), exhibited hippocampal sclerosis of varying degree. Our population of “cryptogenic” epilepsy patients exhibited patterns of hippocampal damage that ranged from relatively restricted hilar and CA3 cell loss (“endfolium sclerosis”) (Margerison and Corsellis, ’661, to total Ammon’s Horn sclerosis (Bruton, ’88), in which only a small number of granule cells and the “resistant zone” of CA2 pyramidal cells survived (Figs. 7-10). The most common pattern of damage involved severe loss of dentate hilar neurons, CA1 and CA3 pyramidal cells. Most “cryptogenic” patients exhibited survival of the “resistant zone” of pyramidal cells and many granule cells, which often appeared dispersed from their normal, tightly packed appearance (Fig. 9A and Houser, in press). CaBP-LI. The most notable finding in sclerotic hippocampi was that despite extensive cell loss, the vast majority
D.
-
I “
9).
PV-LI. There was an obvious and consistent loss of PV-LI in sclerotic hippocampi compared to hippocampi
390
Fig. 7. Pathology of “endfolium sclerosis” versus “classical” hippocampal sclerosis. A Nissl-stained section showing damage to the dentate hilus (h) and CA3 pyramidal cells with survival of much of area CA1. Patient CIC. B: Classical hippocampal sclerosis. Note loss of hilar
R.S. SLOVITER ET AL.
(h), CA3, and CA1 cells (between arrowheads) with survival of some granule cells, cells of the resistant zone (CA21, and subicular neurons (S).Patient JWF. Magnification: x 16.
CALCIUM-BINDING PROTEINS IN HUMAN HIPPOCAMPUS
Fig. 8. CaBP- and PV-LI in sclerotic “cryptogenic” epileptic hippocampus. A Cresyl violet. Note extensive cell loss in the hilus (H) and area CA1 and the survival of subicular neurons ( S ) .B: CaBP-LI. Note immunoreactivity in many of the remaining cells. C: PV-LI in an
391
alternate section of the same hippocampus showing the loss of PV-LI in area CA1 and the dentate granule cell region (GI. Note the relatively normal staining in the surviving subiculurn ( S ) adjacent to the affected CA1 region. Patient MAB. Magnification: X 16.
R.S. SLOVITER ET AL.
392
Fig. 9. CaBP-LI in severe hippocampal sclerosis: A Cresyl violet. Note extensive loss of granule cells (GI, dispersion of remaining granule cells, and extensive loss of hilar (H) and CA1 neurons. Cells of the resistant zone (RZ) and the subiculum (S) remain. B: CaBP-LI showing that many of the few surviving neurons of the granule cell layer. hilus and resistant zone (asterisks) are CaBP-immunoreactive. Patient MT. Magnification: ~ 1 9 .
from “tumor” epileptics. Both the number of PV-LI neurons and the density of immunoreactive fibers were obviously and consistently reduced (Fig. 8). This was not due to vagaries in the staining since relatively normal PV staining was apparent in the relatively undamaged subiculum of each section. In addition, the decrease in PV-LI was roughly proportional to the degree of cell loss. In a patient with “endfolium sclerosis” (patient CIC), characterized by hilar and some CA3 pyramidal cell loss but survival of many CA1 neurons (Fig. 7A), the loss of PV-LI was restricted to the
damaged regions; the rest of the pyramidal cell layer exhibited relatively normal PV-LI. More severely sclerotic hippocampi routinely exhibited extensive loss of PV-LI in regions of cell loss with relatively normal PV staining in the surviving subicular region of each hippocampus. One consistent difference between the “tumor” patient group and the “cryptogenic” group was the loss of staining of the PV-immunoreactive axo-somatic plexus in the surviving granule cell layer. “Tumor” group patients generally exhibited a plexus throughout the granule cell layer (Fig.
CALCIUM-BINDING PROTEINS IN HUMAN HIPPOCAMPUS
393
Fig. 10. CaBP-LI in a “cryptogenic” epileptic hippocampus. A Higher magnification of section in Figure 9 showing cellular detail. Note CaBP-immunoreactive granule cells (row of arrowheads at left), cells of the resistant zone (RZ), and the abnormally numerous and darkly stained nongranule cells in the granule cell layer (GI, molecular layer (ML), and hilus (H). B: Detail of hilar cells shown in A in the region of the single arrowhead. Note CaBP-immunoreactive cells of
various morphology including a large multipolar neuron (double arrowheads) with dendrites entering the granule cell layer, and small cells (single arrowheads) stained more darkly than granule cells (visible in left portion of panel A). C : detail of resistant zone shown in A. Note survival of presumed pyramidal cells (arrowheads) and presumed interneurons of several locations (asterisks). Patient MT. hf, hippocampal fissure. Magnification: X37 in A, x 124 in B; X 82 in C.
6B), whereas “cryptogenic” patients exhibited expanses of the granule cell layer with no detectable PV-immunoreactive axo-somatic plexus immediately adjacent to expanses with a visible plexus (Fig. 6D). In virtually all specimens, however, some PV-LI neurons survived and appeared similar in appearance to cells in the “tumor” specimens. Thus, despite an obvious decrease in PV-LI in sclerotic hippocampi, PV-immunoreactive cells with relatively normal morphology were usually among the small number of surviving neurons (Fig. 5B,C, Fig. 11).Whether the decreased PV immunostaining represents inhibitory cell death
or simply a reduction in immunocytochemically detectable PV is not resolved by these methods.
DISCUSSION CaBP- and PV-LI in the “normal” hippocampus The main finding of this study is that compared to the patterns of CaBP-like immunoreactivity in the hippocampi of lower mammals, the pattern in the normal human brain
394
R.S. SLOVITER ET AL.
Fig. 11. Survival of dentate hilar CaBP- and PV-LI neurons in sclerotic “cryptogenic” epileptic hippocampus. A Cresyl violet-stained section showing survival of many granule cells (g) but few hilar (h) neurons (arrowheads). B: CaBP-LI in an alternate section of the same hippocampus showing a surviving hilar neuron to be CaBP-immunoreactive. C and D: PV-LI in alternate sections of the same hippo-
campus showing a surviving PV-immunoreactive hilar neuron (C) and presumed dentate basket cells (D). Patient LN. E and F: Surviving PV-immunoreactive interneurons of the granule cell layer (g) and molecular layer (ml) in patients BLT (E) and GRS (F). Note that the light staining of the granule cell layer is nonspecific background staining. Magnification: x80 in A-C and F; x 60 in D and E.
is unique. Rats and mice exhibit CaBP-LI in granule cells, many CA1 and CA2 pyramidal cells, and many interneurons (rat: Sloviter, ‘89; mouse: Sloviter, unpublished). Conversely, rabbits and guinea pigs exhibit CaBP-LI in granule cells and numerous interneurons, but not in any pyramidal cells (Sloviter, unpublished). The baboon (Pupio pupzo) exhibits CaBP-LI in dentate granule cells, interneurons of all regions, and virtually all CA1 and CA2 pyramidal cells. Conversely, the human contains CaBP-LI only in
dentate granule cells, CA2 pyramidal cells, and interneurons. Thus, the human is the only species examined in which the normal distribution of CaBP-LI corresponds to the least vulnerable cell populations. The pattern of CaBP-LI in the “normal” human hippocampus appears similar to the staining by an antibody raised against minced human pheochromocytoma tissue and attributed to its recognition of an adrenal chromaffin cell protein, chromogranin A (Munoz, ’90). The epitope
CALCIUM-BINDING PROTEINS IN HUMAN HIPPOCAMPUS recognized by this antibody has not been identified and it is not known if the antibody recognizes CaBP, which is present in the adrenal gland. Thus, it is not known if the recently reported distribution of chromogranin A immunoreactivity in the human hippocampus represents native chromogranin A or another cellular constituent. In contrast to the unique CaBP-staining pattern seen in the human, the pattern of PV-LI in the human appears very similar to that seen in lower animals. PV-LI is present in a specific subset of hippocampal interneurons that have been identified in animals as GABA neurons (Kosaka et al., ’87). In this study, relatively faint PV staining was obtained in the autopsy tissues. Conversely, excellent staining, similar to that seen in the perfusion-fixed baboon, was obtained in the surgical samples, suggesting some postmortem deterioration of PV-LI in the autopsy controls. Although the PV staining in tumor-associated patient tissue may be similar to what would be obtained in ideal control tissue, it cannot be assumed to be truly “normal” despite the fact that the human hippocampi appeared relatively normal in histological appearance. In fact, the hippocampi of tumor-associated epileptics are not normal. Although they appear relatively normal qualitatively, cell counts have revealed that tumorassociated epileptic patients often exhibit subtle, but statistically significant, cell loss (J. Kim and D.D. Spencer, personal communication).
395
simple matter to determine definitively if the PV-immunoreactive plexus has degenerated or if the GABAimmunoreactive plexus is, or is not, normal. These results in experimental animals (Sloviter, ’91) and epileptic human hippocampi may represent a loss of a subset of GABA neurons not easily evaluated by GABA- or GAD-staining alone (Babb et al., ’89). However, the functional implications, if any, of the loss of PV staining are unclear since, regardless of whether or not PV cell loss has occurred, immunocytochemical staining for PV, GABA, or GAD provides no meaningful information about the state of inhibition in the tissue before its removal. The loss of some dentate granule cells and possible loss of some inhibitory cells that contain supposedly protective substances is paradoxical. Does the survival of many calcium-binding protein-containing cells in sclerotic hippocampi represent the end result of a protective process or does the loss of some of these neurons suggest that CaBP and PV are not protective? Clearly, all cells are ultimately vulnerable since some epileptics exhibit virtually complete loss of hippocampal neurons. It seems most reasonable to conclude that protective mechanisms exist and are competitive in nature in that they can be overcome if the insult is sufficiently intense. What remains to be determined is the identity of the operative protective mechanisms, how they are regulated, and how they can be altered experimentally.
CaBP- and PV-LI in the epileptic hippocampus One of t h e main findings made in t h e sclerotic “cryptogenic” epileptic hippocampus was that a large proportion of the few surviving cells were CaBP- or PVimmunoreactive. In addition, the number and dark staining of surviving hilar neurons in sclerotic hippocampi compared to autopsy controls or hippocampi from the tumorassociated patient group suggests that CaBP synthesis may have been increased in the surviving cells. These results are consistent with the suggestion that the excitation-induced influx of calcium initiates epileptic brain damage (Griffiths et al., ’82) and with the hypothesis that the selective vulnerability of specific cell types may be related to differences between cell populations in their abilities to buffer intracellular calcium (Sloviter, ’89). Although the presence of these proteins in surviving human hippocampal neurons does not prove that calcium-binding proteins play a protective role, it has been demonstrated in rat hippocampal slices that increasing the calcium-binding capacity of identified vulnerable dentate hilar cells is protective against excitation-induced irreversible depolarization (Scharfman and Schwartzkroin, ’89). The second possibly significant finding in the sclerotic hippocampi is that large expanses of the dentate granule cell layer appear to be devoid of their PV-immunoreactive, and presumably inhibitory, axo-somatic plexus. Since PVimmunoreactive cells are a subset of GABA neurons (Kosaka et al., ’87), a loss of PV immunoreactivity may mean that a specific subpopulation of inhibitory neurons may have been lost. This loss of PV staining of the axo-somatic plexus in the granule cell layer is virtually identical to the permanently decreased PV staining seen in an experimental epilepsy model in which hippocampal sclerosis is produced by sustained perforant path stimulation (Sloviter, ’91).However, in the experimental animals, no obvious loss of axo-somatic GABA staining occurred. GABA staining of the human tissue (results to be published separately) was more variable than in perfusion-fixed animals so it is not a
ACKNOWLEDGMENTS We thank Dr. Brian Meldrum for providing the baboon tissue, Edward Ronk and Brian Yarborough for printing the photographs, Dr. J.W. Pike for providing CaBP antiserum, and Dr. K.G. Baimbridge for providing CaBP and PV antisera. This research was supported by NIH grant NS1820 1.
LITERATURE CITED Babb, T.L., J.K. Pretorius, W.R. Kupfer, and P.H. Crandall (1989) Glutamate decarboxylase-immunoreactive neurons are preserved in human epileptic hippocampus. J. Neurosci. 9.9562-2574. Baimbridge, K.G., and J.J.Miller (1982) Immunohistochemical localization of calcium-binding protein in the cerebellum, hippocampal formation and olfactory bulb of the rat. Brain Res. 245.223-229. Bruton, C.J. (1988) The Neuropathology of Temporal Lobe Epilepsy. New York: Oxford University Press. Connor, J.A., W.J. Wadman, P.E. Hockberger, and R.K.S. Wong (1988) Sustained dendritic gradients of Ca2+induced by excitatoly amino acids in CA1 hippocampal neurons. Science 240:649-653. Griffiths, T., M.C. Evans, and B.S. Meldrum (1982) Intracellular sites of early calcium accumulation in the rat hippocampus during status epilepticus. Neurosci. Lett. 30:329-334. Houser, C.R. (1990) Granule cell dispersion in the dentate gyrus of humans with temporal lobe epilepsy. Brain Research, 535: 195-204. Kosaka, T., H. Katsumaru, K. Hama, J.-Y. Wu, and C.W. Heizmann (1987) GABAergic neurons containing the Ca2+-bindingprotein parvalbumin in the rat hippocampus and dentate g y r u s . Brain Res. 419r119-130. Lorente de N6, R. (1934) Studies on the structure of the cerebral cortex. 11. Continuation of the study of the ammonic system. J. Psychol. Neurol. 46:113-117. Margerison, J.H., and J.A.N. Corsellis (1966) Epilepsy and the temporal lobes. Brain 89:499-530. Meldrum, B.S. (1981) Metabolic effects of prolonged epileptic seizures and the causation of epileptic brain damage. In F.C. Rose (ed): Metabolic Disorders of the Nervous System. London: Pitman, pp. 175-187. Meldrum, B.S., and Corsellis, J.A.N. Epilepsy. (1984) In J.A.N. Corsellis and L.W. Duchen (eds): Greenfield’s Neuropathology. New York: John Wiley & Sons, pp. 921-950.
396 Morrissey, R.L., R.N. Empson, D.T. Zolock, D.D. Bikle, andT.J. Bucci (1978) 11.A timed study of Intestinal response to 1,25-dihydroxycholecalciferol. the intracellular localization of calcium-binding protein. Biochim. Biophys. Acta 538:34-41. Munoz, D.G. (1990) Thedistribution of chromogranin A-like immunoreactivity in the human hippocampus coincides with the pattern of resistance to epilepsy-induced neuronal damage. Ann. Neurol. 27:266-275. Scharfman, H.E., and P.A. Schwartzkroin (1989) Protection of dentate hilar mossy cells from prolonged stimulation by intracellular calcium chelation. Science 246r257-260. Sloviter, R.S. (1987) Decreased hippocampal inhibition and a selective loss of interneurons in experimental epilepsy. Science 23573-76.
R.S. SLOVITER ET AL. Sloviter, R.S. (1989) Calcium-binding protein (Calbindin-D28k) and parvalbumin immunocytochemistry: Localization in the rat hippocampus with specific reference to the selective vulnerability of hippocampal neurons to seizure activity. J. Comp. Neurol. 280:183-196. Sloviter, R.S. (1991) Permanently altered hippocampal structure, excitability and inhibition after experimental status epilepticus in the rat; the “dormant basket ceil” hypothesis and its possible relevance to temporal lobe epilepsy. Hippocampus Ir41-66. Sloviter, R.S., N.M. Barbaro, and K.D. Laxer (1989) Calcium-binding protein (calbindin) and pawalbumin-like immunoreactivity (LI) in the “normal” and “epileptic” hippocampus. Epilepsia 30;719.
