[1] 16w Alpha-synuclein is not a requisite component of synaptic boutons in the adult human central nervous system

[1] 246w It is increasingly clear that the normal protein a-synuclein is in some manner closely associated with presynaptic components of select neuronal types within the adult human central nervous system (CNS) and, in addition, that in its pathologically altered state a-synuclein aggregates selectively in the form of filamentous inclusion bodies during certain progressive neurodegenerative disorders, such as familial and sporadic Parkinson's disease. By having the antibody AFshp raised specifically to a-synuclein to label Parkinson disease-specific Lewy bodies and Lewy neurites as well as synaptic boutons containing the unaltered protein, an initial attempt is made to map the overall distribution pattern and describe the staining behavior of the immunoreactive punctae in select regions of the prosencephalon. Neocortical immunolabeling is most prominent in the prodigious, but incompletely myelinated, association fields and faintest in the heavily myelinated primary motor and primary sensory fields, with the premotor and first order sensory association areas occupying an intermediate position. Of the thalamic grays evaluated, those containing powerfully myelinated fiber tracts (e.g. centrum medianum, habenular complex) show the weakest immunolabeling, whereas, less sturdily myelinated structures are highly immunoreactive. The fact that the immunostaining spectrum for normal a-synuclein is so broad, together with the fact that some thalamic sites actually are immunonegative leads to the following conclusions (1) a-synuclein, although present in the synaptic boutons of many nerve cells in the adult human CNS, is by no means ubiquitous there, and (2) neuronal types lacking the normal protein cannot generate the Parkinson's disease-specific filamentous pathology.

[1] 171w The protein a-synuclein (a-SN), a small hydrophilic molecule consisting of 140 amino acids, is abundantly expressed in the human central nervous system (CNS) where it usually is localized in synaptic boutons, typically in close proximity to and bound to synaptic vesicles (Maroteaux et al., 1988;Jakes et al., 1994;Iwai et al., 1996;Davidson et al., 1998;Giasson et al., 1999). One possible function attributed to the protein is regulation of the size of the vesicular pool in individual synaptic boutons. Gene mutations of the molecule encountered in familial forms of Parkinson's disease (PD) (Polymeropoulos et al., 1997;Kruger et al., 1998) result in the loss of a-SN's binding capacities to synaptic vesicles (Jensen et al., 1998). The unaltered protein likely plays a role in the maintenance and stabilization of fully mature synapses (Murphy et al., 2000). Related proteins include band g-SN. The former shares with a-SN the almost exclusive expression at presynaptic sites in the CNS, whereas the latter is more prevalent in the peripheral nervous system (Clayton and George, 1998;Lavedan, 1998;Murphy et al., 2000).

[2] 182w In the course of familial and sporadic PD, gradually evolving but abnormal aggregations of a-SN contribute substantially to the formation of characteristic inclusion bodies which appear as Lewy bodies (LBs) in perikarya and as Lewy neurites (LNs) in cellular processes of involved neurons (Spillantini et al., 1997;Wakabayashi et al., 1997;Arima et al., 1998;Baba et al., 1998;Hashimoto et al., 1998;Irizarry et al., 1998;Lippa et al., 1998;Mezey et al., 1998;Spillantini et al., 1998;Takeda et al., 1998;Trojanowski and Lee, 1998;Bayer et al., 1999;Culvenor et al., 1999;Goedert, 1999;Hashimoto and Masliah, 1999). Whereas the pathologically altered and nearly insoluble form of a-SN protein is immunolabeled by both N-terminal directed antibodies as well as those directed against the C-terminus, C-terminal directed antibodies may be more relevant in this regard because N-terminal directed antibodies usually cross react with band, possibly, g-SN. On the other hand, antibodies directed against the C-terminus are better suited to displaying the normal protein in presynaptic boutons (Takeda et al., 1998). band g-SN do not share a-SN's propensity for building filamentous aggregates (Baba et al., 1998;Spillantini et al., 1998;Culvenor et al., 1999;Duda et al., 2000).

[3] 93w Only a small number of the many neuronal types which make up the human CNS are prone to develop the PD-specific LBs and LNs, with a few subtypes of projection neurons having long axons being especially susceptible. Consequently, the synaptic and neuronal damage which ensue in the course of PD is not indiscriminate or diffuse. On the contrary, it results in a distinctive lesional distribution pattern (Braak et al., 1998(Braak et al., , 2000)). The reasons for the pronounced vulnerability of some neuronal types and the decided resistance of others are still unexplained.

[4] 127w Similarly, there is still a considerable gap in our knowledge regarding the distribution of normal a-SN within the adult human CNS, and no clear consensus has emerged based on the results of previous studies, should -for example -the protein be regarded as an omnipresent presynaptic entity which is associated with every synaptic bouton? Or does a-SN occur only in circumscribed regions of the CNS and, then, only within the synaptic boutons of particular nerve cell types? Assessment of such differences with respect to the presence and/or densities of a-SN immunoreactive boutons would have considerable impact on further attempts not only to unravel questions regarding the functional significance of the protein, but also to explain the idiosyncrasies of the selective vulnerability which some neurons typically express during PD.

[1] 78w The antibody used in this study labels both normal and pathologically altered forms of a-SN in nerve cells of the adult human CNS (Figs. 1 and 2). No immunostaining occurs in macroglial cells or non-neuroectodermal cells. Normal a-SN appears solely in the form of modestly immunoreactive punctae which are more or less evenly distributed throughout the neuropil (Fig. 2c) where they are easily distinguishable from the sharply delineated and more intensely a-SN-labeled LNs or LBs (Fig. 2a, arrows).

[2] 85w The punctate pattern of the unaltered a-SN is recognizable in many, but by no means all, of the cortical areas and subcortical nuclei of the human CNS. The small immunoreactive dots probably correspond to synaptic boutons and appear solely in specific nuclear grays in PD cases and non-diseased controls alike (Figs. 1 and 2). Normal a-SN occurs chiefly in presynaptic structures, and neither the cell nucleus, perikaryon, dendrites, nor proximal axon reveal any traces of the protein large enough to be detected by light microscopy.

[3] 104w In the prosencephalon, the density of the a-SN immunoreactive punctae appears to be generally high. Areas rich in such punctae include the neocortex, allocortex, claustrum, amygdala, striatum, and most of the thalamic nuclear grays (Fig. 1a-e). Below this level, a marked reduction occurs. Within the mesencephalon, cerebellum, lower brain stem, and spinal cord, only relatively few areas of gray matter display immunopositive punctae, and the intensity of the immunolabeling usually is weak. Entities which contain such scattered punctae include, among others, the pontine gray (PG), inferior olive (O), and deep cerebellar nuclei as well as granular cell layer of the cerebellum (Fig. 1f -h).

[4] 41w The following description of portions of the neocortex and thalamus represents an initial effort to point out differences between prosencephalic areas which are prominently supplied with a-SN immunoreactive boutons and those lacking them or displaying only a few scattered immunolabeled dots.

[5] 107w The neocortical gray matter appears as a band of immunopositive tissue clearly demarcated against the underlying immunonegative white matter (Fig. 1c). The outlines of small faintly immunoreactive zones correspond in size and shape to those of blood vessels and perikarya of both nerve cells and glial cells (Fig. 2a-c). For the most part, the immunoreactive punctae in the neuropil of the neocortex maintain a certain distance from each other and are not inclined to pool together at any given site. Only those portions of the neuropil immediately surrounding the walls of cortical blood vessels tend to register slight increments in local immunoreactivity (Fig. 2b and c, arrows).

[6] 124w Within different neocortical laminae, the punctae respond with varying degrees of intensity (Braak, 1980). The neocortex appears band-like and displays a threetiered pattern, with a very intensely labeled outer portion corresponding to layers I-IIIa and b. A broad stripe displaying reduced immunoreactivity (layers IIIc -Vb) lies approximately in mid-cortex (Fig. 1c, Fig. 2a and b). The immunoreactivity in this zone as a whole falls off by degrees commencing with layer IIIc, usually reaching its nadir within the reaches of the outer line of Baillarger or line of Gennari respectively in layer IV, and still includes, with gradually increasing labeling, layer Va and the inner line of Baillarger in layer Vb. The innermost tier shows moderately strong immunolabeling and corresponds to the cortical layer VI.

[7] 356w Different areas of the neocortex show varying degrees of a-SN labeling as well. Immunolabeling is least intense in the primary motor field (m) and primary sensory field (s), it increases slightly in the premotor areas and first order sensory association areas, and 1). Series of sections through the thalamus and cut perpendicular to Forel's axis. The immunoreactions in the various thalamic nuclei exhibit varying degrees of intensity: Note the particularly weak immunoreactions in the centrum medianum (in c: CM) and habenular complex (in d: HB). (c) The neocortex appears band-like and displays a three-tiered pattern, with a particularly intensely labeled outer portion corresponding to layers I-IIIab, a broad pale zone in mid-cortex with reduced immunoreactivity corresponding to layers IIIc-Vb, and an innermost tier showing moderately strong immunolabeling and corresponding to layer VI. Note the especially faint staining of the primary motor, somatosensory, and auditory fields and the remarkably vivid immunoreaction in the basal temporal neocortex. The breadth of the pale stripe is widest in the primary areas and narrowest in the basal temporal neocortex. (f -h) Case No. 1 (Table 1). Sections through the brain stem cut perpendicular to Meynert's axis. In comparison to the prosencephalon, the overall density of immunoreactive punctae appears to be low. Labeling to a certain extent is encountered only at a few sites, such as the cerebellar dentate nucleus and inferior olive. (a) A, amygdala; AP, anteroprincipal nucleus; C, caudate nucleus; E, entorhinal cortex; VA, ventroanterior nucleus. (b). AP, anteroprincipal nucleus; E, entorhinal cortex; MD, mediodorsal nucleus; N, substantia nigra; R, red nucleus; VP, ventroposterior nucleus. (c). C, caudate nucleus; CM, centrum medianum; H, hippocampus; MD, mediodorsal nucleus; N, substantia nigra; P, putamen; R, red nucleus; a, primary auditory field; m, primary motor field; s, primary somatosensory field; bt, basal temporal neocortex; ic, internal capsule. (d). CGL, lateral geniculate complex; CGM, medial geniculate complex; E, entorhinal cortex; HB, habenula; LD, laterodorsal nucleus. (e) E, entorhinal cortex; PU, pulvinar. (f) PG, pontine gray. (g). D, dentate nucleus. (h) O, inferior olive; X, dorsal glossopharyngeal-vagal area. Bar in (a) applies also to (b -e). Bar in (f) applies also to (g and h).

[8] 88w reaches peak levels of intensity in the extensive association cortex (Fig. 1c). Within the latter category, it is above all the areas covering the pole region and basal portions of the temporal lobe that show the densest immunolabeling of all of the neocortical areas (Fig. 1c). Correspondingly, the average breadth of the pale stripe is widest in the primary areas and narrowest in the basal temporal association cortex (Fig. 1c: Compare the primary auditory [a], somatosensory [s], and somatomotor [m] areas with the basal temporal [bt] association areas).

[9] 72w The gray matter of the thalamus often resembles immunopositive islets set off sharply from the immediately surrounding immunonegative white matter. The bundles of sturdily myelinated fiber tracts traversing portions of the thalamus likewise are recognizable as light immunonegative patches or stripes. Blood vessels and the perikarya of nerve and glial cells materialize as small pale, almost colorless, areas. In the neuropil, individual immunopositive dots emerge uniformly distributed throughout the spaces between cells.

[10] 337w The overall tint of the immunoreactive neuropil varies among different nuclei of the thalamus (Jones, 1985) (Fig. 1a-e, Fig. 2d and e). Medial portions of nuclei and those located close to the third ventricle are more immunoreactive than nuclei situated more laterally in the vicinity of the internal capsule (ic) (Fig. 1a-e). 1); PEG, 100 mm, without counterstaining. (a) Basal temporal neocortex (fusiform gyrus). Note the strongly immunopositive outer band followed at mid-cortical level by a stripe with reduced immunoreactivity which gradually merges into an inner band with moderately strong labeling. The border towards the white substance is clearly demarcated. The outlines of some pale, almost colorless, areas correspond to blood vessels. Arrows point to intensely immunoreactive LBs. (b) Primary visual field, occipital lobe. Note the conspicuously light-colored mid-cortical band, including the line of Gennari, and slight increments of immunoreactivity around cortical blood vessels (arrow). (c) Punctate pattern of immunoreactive structures displayed at higher magnification in part of the outer pyramidal layer of the first Ammon's horn sector. Cortical blood vessels as well as the somata and apical dendrites of hippocampal projection neurons appear as small non-immunoreactive zones (arrows). (d and e). Parts of Fig. 1c and d showing thalamic nuclei with remarkably low densities of a-SN immunoreactive punctae are detailed at higher magnification (centrum medianum in d and habenular complex in e). CM, centrum medianum; HB, habenula; LI, limitans nucleus; MD, mediodorsal nucleus; PF, parafascicular nucleus; PU, pulvinar; VM, ventromedial nucleus; VPMpc, ventroposterior nucleus, parvocellular subnucleus Immunoreactivity is weakest in the centrum medianum (CM), a nuclear gray which, already upon observation with the naked eye, impresses as a light-colored areal situated below the vividly immunostained mediodorsal nucleus (MD) (Fig. 1c, Fig. 2d). The CM's demarcation lines are sharply drawn in immunoreactions for a-SN, its only blurred boundary being the transition from the CM to the moderately labeled parafascicular nucleus (PF) (Fig. 2d). The parvocellular subnucleus of the ventroposterior nucleus (VPMpc) shows remarkably strong immunoreactions, thereby defining the lowermost limit of the CM (Fig. 1c, Fig. 2d).

[11] 189w Only the nuclear complex of the habenula (HB) is of comparable paleness (Fig. 1d, Fig. 2e). Closer inspection reveals that the small-celled medial nucleus of the habenula exhibits faint signs of immunoreactivity, whereas the lateral nucleus and surrounding areas of the medullary stria remain immunonegative. Moreover, additional and relatively light-colored regions are the terminal fields of the medial lemniscus, i.e. the ventroposterior and ventromedial nuclei (VP, VM) (Fig. 1b, Fig. 2d). The ventrointermediate nucleus and ventroanterior nucleus (VA) (Fig. 1a), by contrast, already show moderately intense staining with gradually increasing punctate densities when proceeding in a posteroanterior and inferosuperior direction. The medial geniculate body (CGM) is more strongly labeled than the lateral one (CGL), and an intermediate degree of immunoreactivity can be detected in the lateral reaches of the pulvinar (PU) (Fig. 1e). The principal nucleus of the anterior nuclear complex (AP), the laterodorsal nucleus (LD), and lateral portions of the mediodorsal nucleus (MD) register somewhat higher degrees of immunoreactivity (Fig. 1a, b and d). Peak labeling intensities are attained in the medial portions of both the mediodorsal nucleus (MD) and medial pulvinar (Fig. 1b and c, Fig. 2d).

[1] 139w The present findings corroborate those of previous studies which demonstrate that unaltered a-SN is localized almost exclusively in synaptic boutons irrespective of which region within the neuropil of the CNS is investigated (Wakabayashi et al., 1997;Clayton and George, 1998;Takeda et al., 1998;Wakabayashi et al., 1998). Nevertheless, up to this point no one has pursued the issue of whether in fact all synaptic boutons in the adult human CNS contain this protein. As such, we examined the overall distribution pattern and intensities of a-SN immunolabeling in serial sections through whole hemispheres. The characteristic punctate pattern rendered visible by the normal protein is difficult to overlook and cannot be confused with the abnormal forms which develop in the course of PD (Maroteaux et al., 1988;Ue ´da et al., 1993;Jakes et al., 1994;Takeda et al., 1998;Wakabayashi et al., 1998;Murphy et al., 2000).

[2] 91w Our findings chiefly point to the fact that a-SN does not occur in all of the nuclear grays of the adult human CNS. Moreover, although all of the nuclear grays evaluated here contain abundant synapses, the strength of the unaltered a-SN immunoreaction varies considerably from one cortical area or subcortical nucleus to another, and -more importantly -some even turn up immunonegative. The most straightforward explanation for this thoroughly unexpected finding is that a-SN does not constitute a sine qua non of every healthy synapse in the CNS of the human adult.

[3] 165w With respect to the degrees of intensity on the part of the immunoreactions, it is primarily the inverse proportional relationship between immunoreactivity and myelination that commands the viewer's attention: Cortical layers containing a powerful plexus of myelinated fibers, e.g. the lines of Baillarger or the line of Gennari, as well as richly myelinated cortical areas, such as the primary fields of the neocortex, exhibit particularly weak immunolabeling (Fig. 1c). Similarly, the most heavily myelinated thalamic nuclei, such as the centrum medianum, habenular complex, and the ventroposterior and ventromedial nuclei, are conspicuously pale in immunostained sections. The reverse is also true, namely, the most sparsely myelinated cortical layers (layers I-IIIb) and cortical areas (e.g. the basal temporal neocortex) exhibit peak immunoreactivity and, analogously, the most poorly myelinated portions of the pulvinar and of the mediodorsal nuclear complex attain the highest intensities of immunolabeling. In other words, increments in average myelin content generally are paralleled by corresponding decrements in the intensity of a-SN immunoreactivity and vice versa.

[4] 100w Considerable inequities in synaptic density may influence the degree of a-SN-immunoreactivity at a given site. The high synapse density in lamina IV of the neocortex as well as in the centrum medianum, however, is antithetical to the assumption that differences in synaptic density chiefly account for the varying degrees of intensity of a-SN immunoreactions. Another explanation for the variations in staining behavior could be that, instead of a-SN, b-SN (or other synuclein forms) are present in those synapses which perhaps by their very nature are devoid of a-SN. Such a scenario is not farfetched and cannot be ruled out completely.

[5] 169w There is now a broadly based consensus that h-SN plays an essential role in the pathogenesis of PD (Spillantini et al., 1997;Wakabayashi et al., 1997;Baba et al., 1998;Irizarry et al., 1998;Lippa et al., 1998;Mezey et al., 1998;Spillantini et al., 1998;Takeda et al., 1998;Trojanowski and Lee, 1998;Wakabayashi et al., 1998;Bayer et al., 1999;Goedert, 1999;Hashimoto and Masliah, 1999;Braak et al., 2000). In this disorder, unaltered a-SN probably leaves the synaptic boutons and, in a still little understood process, becomes transformed into a nearly insoluble material which contributes primarily to the formation of LBs and LNs. The data presented here indicate that a-SN does not constitute in each and every instance an obligatory component of synaptic boutons in the CNS of the adult human. As such, it is conjectured that neuronal types generating synaptic boutons without a-SN are incapable of developing the PD-specific pathological alterations. The results of the present study thus have important implications relevant to the phenomenon of selective neuronal vulnerability which is one of the enigmatic features of PD.

[1] 115w Brains obtained at autopsy from six individuals with clinically diagnosed PD (three females, three males, aged 78.295.3 years, Hoehn and Yahr, 1967 stages IV -V; case nos. 1-6 in Table 1) were studied together with six autopsy brains from patients lacking a history of neurological disorders for purposes of comparison and control (two females, four males, aged 60.89 22.1 years; case nos. 7-12 in Table 1). Additionally, all of the cases were classified according to a procedure permitting differentiation between stages I-VI in the evolution of Alzheimer's disease-related neurofibrillary alterations and stages A-C in the development of bamyloid deposits (Hyman and Trojanowski, 1997, Table 1). All the six control cases were free of PD-specific pathology.

[2] 350w The tissue was fixed by immersion in a 4% buffered solution of formaldehyde. One of the hemispheres of each case was embedded in polyethylene glycol (PEG 1000, Smithson et al., 1983) and sectioned perpendicular to the intercommissural (Forel's) axis into uninterrupted series of 100-mm thick coronal sections. Five sets of free-floating sections were treated with different staining procedures. In each set, sections 1 mm equidistant from each other were prepared as follows; the first collection of the 1st, 11th, 21st, etc. free-floating sections was stained both for lipofuscin pigment (aldehyde-fuchsin) as well as Nissl material (Darrow red) and employed for topographical orientation. The second collection consisting of the 2nd, 12th, 22nd, etc. sections was pre-treated according to a standard protocol designed to inhibit endogenous peroxidase and prevent non-specific binding. Incubation for 48 h in the affinity-purified a-SN antiserum (AFshp) at a dilution of 1:2000 -4000 followed. This antiserum was generated by one of the authors (W.P. Gai) in sheep using a peptide corresponding to the amino acid residues 116 -131 of the human a-SN. The immunoreactions were performed with a peroxidase-linked secondary antibody and diaminobenzidine. Omission of the primary antiserum resulted in non-staining. Under these conditions and in our experience, a-SN immunoreactivity appears relatively stable. There is no difference either in the staining pattern or intensity of a-SN immunoreactivity within the 24-h time span of post-mortem intervals (PMI) in human autopsy material fixed by immersion in 4% formaldehyde solutions. Two further collections consisting of the 3rd, 13th, 23rd, etc. and the 4th, 14th, 24th etc. sections respectively were silver-stained with the Campbell -Switzer silver pyridine method for assessment of LBs, LNs, and b-amyloid deposits (Campbell et al., 1987;Braak andBraak, 1991, 1999;Sandmann-Keil et al., 1999). The fifth collection of sections underwent staining with a modified silver iodide Gallyas technique to detect argyrophilic deposits of abnormal tau protein (Gallyas, 1971;Iqbal et al., 1993;Braak and Braak, 1999). Additional selected sections taken from further collections were immunostained for band g-SN (antisera generated by W.P. Gai) and processed as described above. All of the sections were cleared and mounted in a synthetic resin (Permount, Fischer).

[3] 119w The antibodies to the synuclein isoforms used in the present study were raised against synthetic peptides corresponding to the amino acid sequences which are unique to each isoform of human synuclein, namely, the amino acid sequence 116 -131 of a-SN, the amino sequence 108 -125 of b-SN, and the amino acid sequence 114 -127 of g-SN. The specificity of the antibody to a-SN has been described previously (Gai et al., 1999). The specificities of the antibodies for bor g-SN were confirmed by ELISA and tested in Western blots by using recombinant human band g-SN as standards (data not shown). The b-SN antibody did not cross-react with aor g-SN, nor did the g-SN antibody crossreact with the aor b-SN molecules.