Content uploaded by Philippe Damier
Author content
All content in this area was uploaded by Philippe Damier
Content may be subject to copyright.
Brain (1999), 122, 1421–1436
The substantia nigra of the human brain
I. Nigrosomes and the nigral matrix, a compartmental
organization based on calbindin D
28K
immunohistochemistry
P. Damier,
1,2
E. C. Hirsch,
1
Y. Agid
1
and A. M. Graybiel
2
1
INSERM U289, Ho
ˆ
pital de la Salpe
ˆ
trie
`
re, Paris, France Correspondence to: Dr P. Damier, INSERM U289, Ho
ˆ
pital
and
2
Department of Brain and Cognitive Sciences, de la Salpe
ˆ
trie
`
re, 47, boulevard de l’ho
ˆ
pital, 75013 Paris,
Massachusetts Institute of Technology, Cambridge, France
Massachusetts, USA E-mail: cic.salpetriere@psl.ap-hop-paris.fr
Summary
Parkinson’s disease is characterized by massive
degeneration of dopamine-containing neurons in the
midbrain. However, the vulnerability of these neurons is
heterogeneous both across different midbrain dopamine-
containing cell groups and within the substantia nigra,
the brain structure most affected in this disease. To
determine the exact pattern of cell loss and to map the
cellular distribution of candidate pathogenic molecules,
it is necessary to have landmarks independent of the
degenerative process by which to subdivide the substantia
nigra. We have developed a protocol for this purpose
based on immunostaining for calbindin D
28K
, a protein
present in striatonigral afferent fibres. We used it to
examine post-mortem brain samples from seven subjects
who had had no history of neurological or psychiatric
disease. We found intense immunostaining for calbindin
D
28K
associated with the neuropil of the ventral midbrain.
Keywords: Parkinson’s disease; substantia nigra; dopamine; calbindin; basal ganglia
Abbreviation:TH5tyrosine hydroxylase
Introduction
Dopamine-containing neurons of the substantia nigra are
severely affected by the degenerative process that occurs in
Parkinson’s disease. Evidence suggests that the loss of these
neurons is heterogeneous, so that the ventrolateral part of
the substantia nigra is almost completely destroyed, whereas
its dorsal part is only partly damaged (Hassler, 1938a). Such
lesion selectivity may be critical for understanding the
mechanisms of neuronal death in this disease, but all previous
anatomical subdivisions of the nigral complex in the human
brain have been based on the use of the dopamine-containing
cells themselves as landmarks, and there has been no
consensus in the literature on the major nigral cell groups
because of difficulties in delineating subgroups of these
neurons. Moreover, any such subdivisions based on the
© Oxford University Press 1999
Within the calbindin-positive region, there were
conspicuous calbindin-poor zones. Analysed in serial
sections, many of the calbindin-poor zones seen in
individual sections were continuous with one another,
forming elements of larger, branched three-dimensional
structures. Sixty per cent of all dopamine-containing
neurons in the substantia nigra pars compacta were
located within the calbindin-rich zone, which we named
the nigral matrix, and 40% were packed together within
the calbindin-poor zones, which we named nigrosomes.
We identified five different nigrosomes. This organization
was consistent from one control brain to another. We
propose that subdivision of the human substantia nigra
based on patterns of calbindin immunostaining provides
a key tool for analysing the organization of the substantia
nigra and offers a new approach to analysing molecular
expressionpatterns in thesubstantia nigra andthe specific
patterns of nigral cell degenerationin Parkinson’s disease.
disposition of dopamine-containing neurons are of
questionable utility in analysing brains from patients with
Parkinson’s disease, in which the loss of dopamine-containing
neurons progressively effaces these anatomical reference
marks.
In this study, we developed a method based on
immunohistochemistry for calbindin D
28K
, a protein that is
present in striatonigral afferent fibres and that stains the
neuropil of the substantia nigra, thus providing landmarks
independent of dopamine-containing neurons by which to
subdivide this region in a consistent and reproducible way.
With this method, we were able to determine the numbers
of dopamine-containing neurons in reliably identifiable
subgroups of the substantia nigra of the human brain.
1422 P. Damier et al.
Fig. 1 MRI illustrating planes of sections used in this study.
(A) Rostral (R), intermediate (I) and caudal (C) transverse planes
presented on a sagittal MR image. (B) MRI of the intermediate
transverse plane. Arrows indicate exiting third cranial nerve
fibres.
Material and methods
Brain samples
Specimens were collected post-mortem from the brains of
seven subjects without history of neurological or psychiatric
disease [mean age 6 standard error of the mean (SEM) 5
84 6 3 years]. Pathological examination of one hemisphere
of each brain ruled out generalized degenerative or vascular
disease. Focal abnormalities would not have been detected
at this stage.
In order to ensure that the entire rostrocaudal extent of the
nigral complex could be sampled, brainstems were removed
from the cerebrum at the level of the mammillary bodies,
and the cerebellum was detached. The brainstems were
transected in the sagittal plane to provide one-half of the
substantia nigra and the ventral tegmental area for analysis,
and they were then blocked transversely at the level of the
upper pons. After dissection, the midbrain blocks were fixed
for 72 h in 4% paraformaldehyde containing 15% saturated
picric acid solution, and they were then washed in 0.1 M
dibasic potassium phosphate buffer (pH 5 7.4), cryoprotected
at 4°C in phosphate buffer containing, consecutively, 0, 5,
10, 15 and 20% sucrose (24 h each), frozen in powdered dry
ice, and stored at –80°C until further processing. Serial
Fig. 2 Subdivision of dopamine-containing neurons of the
midbrain into six dopaminergic groups. CGS 5 central grey
substance; CP 5 cerebral peduncle; M 5 medial group; Mv 5
medioventral group; A8 5 dopaminergic group A8; SN 5
substantia nigra; SNpl 5 substantia nigra pars lateralis; RN 5 red
nucleus; ML 5 medial lemniscus. Based on work by Hirsch and
colleagues (Hirsch et al., 1988).
40-µm-thick sections were cut from the tissue blocks on a
sliding microtome. The sections were immersed free-floating
in 0.25 M Tris buffer containing 0.1% sodium azide and
were then stored at 4°C. One midbrain was cut parasagittally
and another horizontally, in a plane perpendicular to the
coronal axis of the brainstem. The five other midbrains were
cut transversely (Fig. 1).
Calbindin D
28K
and tyrosine hydroxylase
immunohistochemistry
Five midbrains (three transversely cut, one sagittally cut
and one horizontally cut) were studied extensively. Every
third section was stained for calbindin D
28K
immunoreactivity, and every ninth section was stained for
tyrosine hydroxylase (TH) immunoreactivity. Each TH-
stained section was immediately adjacent to a calbindin-
stained section. For the two other midbrains, fewer sections
were prepared for analysis; every ninth section was
immunostained for calbindin D
28K
and every 36th for TH.
Immunohistochemistry was performed according to the
double bridge PAP (peroxidase–antiperoxidase) method
described previously (Graybiel et al., 1987). Briefly, primary
incubations were preceded by successive 8-min exposure
to 20% methanol with 3% hydrogen peroxide, 5-min
exposure to 0.2% Triton X-100, and 30-min exposure to
a1:30dilution of normal goat serum in 0.25 M Tris–
HCl, pH 7.4, containing 0.9% NaCl (Tris-buffered saline;
TBS). Primary incubations were carried out at 14°C for
Human substantia nigra 1423
Fig. 3 Subdivision of the ventral midbrain based on
compartmental patterns of calbindin D
28K
immunostaining.
(A) Computerized chart of dopamine-containing neurons in a TH-
stained section through the midbrain. (B) Adjacent section stained
for calbindin immunochemistry. (C) Subdivisions of the substantia
nigra identified in this study (see text for explanation). The dorsal
(D), ventral (V), medial (M) and lateral (L) sides of the sections
are shown in A. Arrows indicate blood vessel used for alignment
of adjacent sections. RN 5 red nucleus; CP 5 cerebral peduncle.
Scale bar 5 1.5 mm.
2 days with mouse anti-calbindin D
28K
antiserum (1 : 500;
Sigma, St Louis, Mo., USA) or rabbit anti-TH (1 : 500;
Eugene Tech, Allendale, NJ, USA). Successive secondary
(goat anti-mouse, 1 : 800 for calbindin, and goat anti-
rabbit, 1: 400 for TH) and tertiary (mouse PAP, 1 : 200
for calbindin, and rabbit PAP, 1 : 100 for TH) incubations
followed (30 min, each at room temperature). All incubation
solutions contained 1% normal goat serum and 1% normal
human serum in TBS. Primary incubation solutions also
contained 0.01% thimerosal. Steps were separated by buffer
washes. Sections were developed in 0.05%
diaminobenzidine.
Fig. 4 Calbindin immunostaining in horizontally cut (A) and
sagittally cut (B) midbrain sections. Rostral (R), caudal (C),
lateral (L) and medial (M) sides of the section in A; rostral (R),
caudal (C), ventral (V) and dorsal (D) sides of the section in B.
RN 5 red nucleus. Scale bar 5 3 mm.
Mapping and quantitative analysis of
dopamine-containing neurons
Dopamine-containing neurons were identified by their TH
content. Many of them contained a visible nucleus, but a
few large neuronal profiles (.10 µm in diameter) in which
the dense immunostaining prevented the nucleus from
being visible were also counted. The locations of the
neurons were plotted with the aid of an image analysis
system (HistoRag, Biocom, Les Ulis, France) that allowed
us to generate and to print precise maps of the distribution
of all TH-positive neurons. Quantitative analysis of
dopamine-containing neurons was performed on the five
transversely cut brains. First, the midbrain was divided
into the six dopaminergic regions previously identified by
Hirsch et al. (1988) and shown in Fig. 2. Next, the
substantia nigra was subdivided by using patterns of
calbindin staining as a template. Sections adjacent to
the TH-immunostained sections, stained for calbindin
immunoreactivity, were used to delineate the calbindin-
poor and calbindin-rich zones observed (see Results), and
the outlines of these zones were projected onto the maps
of TH-positive neurons. Adjacent sections were aligned by
1424 P. Damier et al.
Fig. 5 Reproducibility of calbindin immunostaining at the same rostral (R), intermediate (I) and caudal (C) levels in two control
midbrains (A and B). Symbols indicate different calbindin-poor zones (★, nigrosome 1; r, nigrosome 2; m, nigrosome 3; d, nigrosome
4. RN 5 red nucleus; CP 5 cerebral peduncle; III 5 exiting third cranial nerve fibres. Scale bar 5 3 mm.
reference to anatomical landmarks, especially the cross-
sections of blood vessels. From these drawings, we defined
the contours of different groups of TH-positive neurons
for each section. Finally, the outlines of these groupings
were added to the computer-generated charts of TH-
positive neurons (Fig. 3). The Biocom system calculated
the number of TH-positive neurons within each group
defined by outlines of the zones identified by calbindin
immunostaining. The total number TH-positive neurons in
each subgroup was estimated from the map of TH-positive
neurons. Split-cell counting errors were corrected by the
formula of Abercrombie:
N 5 n[t/(t 1 d)]
where N 5 total number of cells, n 5 number of cells
counted, t 5 section thickness and d 5 cell nucleus
diameter (Abercrombie, 1946). The correction factor was
0.61; no significant differences in cell nucleus size were
found among regions or among brains. The total number
of neurons in each region was calculated from the formula
of Konigsmark:
N
t
5 N
s
3 (S
t
/S
s
)
where N
t
5 total number of neurons, N
s
5 number of
neurons counted, S
t
5 total number of sections through
the region and S
s
5 number of sections in which neurons
were counted (Konigsmark, 1970). To evaluate rostrocaudal
variations, midbrains were analysed in three domains:
anterior to the level of exiting third cranial nerve fibres;
at this level; and posterior to it (Fig. 1).
Results
Calbindin staining patterns in the ventral
midbrain
Calbindin staining was intense across all of the ventral
midbrain and appeared to be associated with the neuropil
rather than with the cell bodies of the substantia nigra.
Calbindin-positive neuropil was present throughout the
substantia nigra pars reticulata and most of the substantia
nigra pars compacta, the main region containing TH-positive
neurons. Some TH-positive neurons appeared above the
Human substantia nigra 1425
Fig. 6A
calbindin-rich region, and for ease of description we termed
this zone the ‘substantia nigra pars dorsalis’ (Fig. 3).
Within this large calbindin-positive zone, there were
smaller but conspicuous calbindin-poor zones. These had
variable shapes, some being long and thin, others being
rounded or branched. Analysed in serial sections, many of
1426 P. Damier et al.
Fig. 6B
Human substantia nigra 1427
Fig. 6 (A–C) Charts of dopamine-containing neurons (right) and photographs of calbindin D
28K
immunostaining in adjacent transverse sections (left). Values indicate distance (in mm) to the level of
rostral exiting third cranial nerve fibres; positive and negative distances denote rostral and caudal
distances, respectively, to this level. Symbols indicate the five different invaginated pockets of low
calbindin staining (nigrosomes) identified within the calbindin-positive neuropil (★, nigrosome 1; r,
nigrosome 2; m, nigrosome 3; d, nigrosome 4; j, nigrosome 5). RN 5 red nucleus; CP 5 cerebral
peduncle; DBC 5 decussation of the brachium conjunctivum; III 5 exiting fibres of third cranial
nerve. Scale bar 5 3 mm.
1428 P. Damier et al.
Fig. 7 Distributions of dopamine-containing neurons (charts on right) and calbindin D
28K
immunostaining (photographs on left) in the horizontal plane. Values indicate distance (in millimetres)
to rostral exiting third cranial nerve fibres. Symbols indicate two different invaginated pockets of low
calbindin staining identified within the calbindin-positive neuropil (★, nigrosome 1; d, nigrosome 4).
RN 5 red nucleus. Scale bar 5 3 mm.
the calbindin-poor zones seen in individual sections were
continuous with one another, forming elements of branched,
three-dimensional subsystems within the region as a whole.
Zones of low calbindin immunoreactivity were evident in
both sagittal and horizontal sections (Fig. 4). We chose for the
quantitative analysis a plane that intersected perpendicularly
most of these zones and that could also be defined by
accessible anatomical landmarks: a transverse plane passing
(i) between the superior and the inferior colliculus and (ii)
through the border between the pons and the midbrain
(Fig. 1). Figure 5 illustrates the patterns of calbindin
immunoreactivity in sections cut in this plane for two brains
at three rostrocaudal levels.
Three-dimensional organization of nigral
compartments delineated by calbindin
immunostaining
To determine the three-dimensional organization of the
calbindin-poor zones distributed within the calbindin-
positive neuropil of the nigral complex, we analysed the
staining patterns in near-serial sections (Figs 6 and 7). We
were able to identify five different calbindin-poor zones,
which we named ‘nigrosomes’ and numbered from 1 to
5. All five nigrosomes appeared in individual sections as
invaginated pockets of low calbindin staining embedded
in a calbindin-rich surround (Figs 6 and 7). Nigrosome 1
(★) was the largest of the five. Its shape was that of a
lens with its concavity dorsal and its main axis parallel
to the rostrocaudal axis of the substantia nigra. It was
present from the caudal part of the calbindin-positive
region to the level of the rostral exit of third cranial nerve
fibres. Nigrosome 1 was situated in the ventral third and
the lateral two-thirds of the calbindin-rich neuropil, at all
but the caudal level, where it had a more dorsal and
lateral position. Nigrosome 2 (r) looked like a cylinder
with its main axis parallel to that of nigrosome 1. It was
centred in the medial third of the calbindin-rich zone of
neuropil, and its rostral border was slightly more caudal
than that of nigrosome 1. Nigrosome 3 (m) corresponded
to a depression in the lateral and caudal part of the
calbindin-positive neuropil. Its rostrocaudal extent was
Human substantia nigra 1429
Fig. 8 Summary of midbrain subdivisions illustrated at three representative transverse levels. CGS 5
central grey substance; M 5 medial group; Mv 5 medioventral group; A8 5 dopaminergic group A8;
SNpd 5 substantia nigra pars dorsalis; SNpl 5 substantia nigra pars lateralis; N 5 nigrosome. RN 5
red nucleus; DBC 5 decussation of the brachium cunjunctivum; CP 5 cerebral peduncle; III 5 exiting
third cranial nerve fibres.
almost the same as that of nigrosome 2, despite some
variations from one brain to another that we attribute to
differences in plane of sectioning. Nigrosome 4 (d) was
well developed in the dorsal and mid-lateral part of the
calbindin-rich neuropil. Its shape was that of a dorsoventrally
flattened cylinder with a main axis parallel to that of
nigrosome 1. Nigrosome 4 was located in the dorsal third
of the calbindin-rich neuropil, and was visible from the
level of the middle extent of the exit zone of third cranial
nerve fibres to levels rostral to the exiting of these fibres.
Nigrosome 5 (j) corresponded to a group of calbindin-
poor pockets located in the rostral part of the midbrain
(anterior to the exiting third cranial nerve fibres) and
situated ventral and medial to nigrosome 4. Schematic
diagrams of these five nigral subdivisions (nigrosomes 1–
5) are shown for three rostrocaudal levels in Fig. 8.
Reproducibility of the calbindin
immunostaining patterns
Analysis of sets of serial sections showed that the patterns
of calbindin immunostaining were reproducible from one
subject to another. Even though variations in the plane
of sectioning, and probable inter-individual differences,
sometimes prevented strictly identical patterns in equivalent
sections from different subjects being obtained, the general
three-dimensional architecture of the nigrosomes was
reproduced in remarkable detail from brain to brain.
Quantitative analysis of dopamine-containing
neurons in relation to calbindin-defined
compartments
Nearly 200 000 dopamine-containing neurons per half-
midbrain were counted for five brains (exact mean 6 SEM 5
199 251 6 18 446). Two-thirds of these neurons were located
in the substantia nigra, defined as being ventral to the medial
lemniscus and lateral to the medial edge of the red nucleus,
or, in the caudal midbrain, the decussation of the brachium
conjunctivum (Fig. 2). Eighty per cent of these nigral
dopamine-containing neurons were within the calbindin-rich
region of the ventral midbrain that we defined as the substantia
nigra pars compacta, and 20% were dorsal to this zone, in
the region that we termed the substantia nigra pars dorsalis
(Figs 8 and 9).
Among dopamine-containing neurons located in the
substantia nigra pars compacta, 60% were sparsely distributed
within the large region of intense calbindin staining, which
we named the nigral ‘matrix’. The other 40% of the dopamine-
containing neurons were included within the different
nigrosomes, with a majority (60%) in nigrosome 1. The intra-
nigrosomal neurons tended to be densely packed together
(Figs 6 and 7). Table 1 shows the mean numbers of
dopamine-containing neurons estimated for the different
subdivisions of the substantia nigra. Table 2 and Fig. 10
indicate how the distribution of dopamine-containing
neurons varied along the rostrocaudal axis of the
substantia nigra.
1430 P. Damier et al.
Fig. 9 Distribution of TH-positive neurons in the substantia nigra,
including the calbindin-rich region (substantia nigra pars
compacta) and the zone above it (substantia nigra pars dorsalis
and pars lateralis) containing TH-positive neurons (d, compared
with the distribution of TH-positive neurons included within the
calbindin-rich zone neuropil (.). Symbols show the mean
numbers of neurons and bars indicate the SEM calculated from
the three midbrains studied in near-serial sections. Values on the
abscissa indicate distance (in mm) to the most rostral level of the
exiting third cranial nerve fibres. Values on the ordinate indicate
the number of TH-positive neurons. The areas under the curves
correspond to the total number of neurons in the regions
indicated; the area of the indicator box corresponds to 5500
TH-positive neurons.
Table 3 summarizes the mean numbers of dopamine-
containing neurons estimated to occur outside the substantia
nigra, in regions of the midbrain containing TH-positive
neurons, together with their rostrocaudal variations (Fig. 11).
The estimated numbers of neurons in each midbrain
group identified were similar (not significantly different)
for the three midbrains studied by near-serial section
analysis and the two midbrains studied on the basis of
less closely spaced sections.
Discussion
Understanding the anatomy of the substantia nigra is crucial
to analysing its pathology. In this study, we developed a
compartmental analysis of the human substantia nigra based
on calbindin D
28K
immunohistochemistry. The compartmental
organization that we describe provides, for the first time, a
reliable and easily reproducible tool with which to divide the
human substantia nigra into major subdivisions. In the second
paper of this series, we demonstrate that this classificatory
scheme, independent of the distribution of dopamine-
containing neurons, makes it possible to study with great
accuracy the pathology of the substantia nigra in
Parkinson’s disease.
The substantia nigra complex defined by
calbindin immunohistochemistry
The A8, A9 and A10 dopaminergic cell groups of the human
midbrain are directly continuous with one another. Thus,
other than on a rough topographic basis, the outlines of the
substantia nigra pars compacta are difficult to assess, as
reflected by the lack of a consensual definition of the nigral
complex despite many attempts to develop one. For some
authors, the medial lemniscus constitutes the dorsal border
of the substantia nigra (Hirsch et al., 1988; Fearnley and
Lees, 1991). Others have suggested defining the substantia
nigra by its dense substance P immunoreactivity (Gibb,
1992; McRitchie et al., 1995). The ventral midbrain region
containing calbindin D
28K
immunoreactivity probably
provides an equivalent definition,given the regional similarity
in distribution of immunostaining for substance P and
calbindin (Gibb, 1992). In the present study, we combined
these criteria. We defined the substantia nigra as the midbrain
region ventral to the medial lemniscus and the substantia
nigra pars compacta as the region containing calbindin-
positive neuropil. As no data are available on the projection
sites of dopamine-containing neurons located dorsal to the
calbindin-immunoreactive region but ventral to the medial
lemniscus, it is difficult to classify these neurons as lying
either in the substantia nigra pars compacta or in the
dopaminergic group A8. Thus, we considered these neurons
as a separate group, and termed the region the ‘substantia
nigra pars dorsalis’.
Neurochemical compartments in the human
substantia nigra pars compacta
The main finding of this study is the compartmentation
of calbindin-positive neuropil in the substantia nigra pars
compacta. We found that six subgroups of dopamine-
containing neurons could be delineated in the substantia nigra
pars compacta according to their locations, within either
the calbindin-rich zone, named the ‘matrix’, or different
calbindin-poor pockets, named ‘nigrosomes’, that are
embedded in the matrix. The organization of the calbindin-
poor zones into five main nigrosomal pockets was reliably
and constantly found in the seven brains analysed in this
study. In each brain, the nigrosomes appeared to be largely
Human substantia nigra 1431
Table 1 Quantitative analysis of dopamine-containing neurons in the substantia nigra
Substantia nigra
SNpl SNpd SNpc
Matrix Nigrosomes
2129 6 239 30 164 6 6744 58 982 6 7894 43 725 6 4046
Total SNpc 5 102 707 6 10 622
Nigrosomes
N1 N2 N3 N4 N5
27 504 6 2451 7891 6 1122 1507 6 430 4719 6 742 2104 6 804
Values are mean 6 standard error of the mean. SNpl 5 substantia nigra pars lateralis; SNpd 5
substantia nigra pars dorsalis; SNpc 5 substantia nigra pars compacta; N 5 nigrosome.
Table 2 Rostrocaudal variation in the number of dopamine-containing neurons in the
substantia nigra
SNpl SNpd Matrix Nigrosomes %
Rostral 68 6 31 3673 6 461 11 415 6 2369 2020 6 587 13
Intermediate 1387 6 223 12 109 6 3203 27 155 6 4774 18 864 6 1812 44
Caudal 674 6 182 14 384 6 5007 20 412 6 2997 22 841 6 3624 43
% 2 22 44 32 100
Values are mean 6 standard error of the mean. SNpl 5 substantia nigra pars lateralis; SNpd 5
substantia nigra pars dorsalis.
separate from one another, but we could not firmly discount
the presence of small connecting zones between one
nigrosome and another. The remarkable consistency of these
features allowed us to number them (nigrosomes 1–5) and
to analyse the distribution of TH-positive neurons from
pocket to pocket and within the surrounding matrix.
Numerous investigators have already attempted to
subdivided the nigral complex anatomically. Hassler (Hassler,
1937) provided probably the most comprehensive study of
the internal substructure of the substantia nigra. However,
the 21 subdivisions that he described are difficult to identify
consistently in every individual. The lack of consensus has
been compounded by the fact that different planes of section
have been used for analysis, from coronal, as in the original
work by Hassler (Hassler,1937,1938),totransverse(Fearnley
and Lees, 1991; Gibb and Lees, 1991) or horizontal
(McRitchie et al., 1995). The most serious difficulty has been
the absence of landmarks independent of the dopamine-
containing neurons themselves. All previous subdivisions of
nigral neurons have been based on a tendency for neurons
in the substantia nigra pars compacta to have a clustered
organization. However, because of the presence of numerous
sparsely distributed neurons in the substantia nigra pars
compacta in addition to the clusters, the borders of the
subdivisions have not been easily defined (Fig. 3). The
calbindin-based method that we describe here avoids this
problem.
Nigrosomes 1–5 in relation to previous
delineations of nigral subgroups
The calbindin-based delineation of the substantia nigra pars
compacta allowed us to recognize several of the main
subgroups already described in this region (Table 4). The
ventrolateral group consistently identified in different studies
(Hassler, 1937; Braak and Braak, 1986; Fearnley and Lees,
1991; van Domburg and ten Donkelaar, 1991; McRitchie
et al., 1995) corresponds to the group of dopamine-containing
neurons included in nigrosome 1.
The key attributes of the calbindin-based classification we
introduce here are that (i) this analysis provides a method
allowing definition in a reproducible way of subdivisions in
the substantia nigra pars compacta, (ii) the calbindin-based
landmarks can be readily recognized in different planes of
section, and (iii) these landmarks are not dependent on the
distribution of the dopamine-containing neurons themselves.
This last point provides anew basis for analysing the midbrain
in brains derived from patients who suffered from Parkinson’s
disease (Damier et al., 1999).
1432 P. Damier et al.
Fig. 10 Distributions of TH-positive neurons in the different groups of the substantia nigra defined on the basis of calbindin D
28K
immunostaining. Dots indicate the mean numbers of neurons and bars indicate the SEM calculated for the three midbrains studied in
near-serial sections. Values on the abscissa indicate distance (in mm) from the rostral exit-point of the third cranial nerve fibres. Values
on the ordinate indicate the number of TH-positive neurons. Areas under the curves correspond to the total number of neurons in the
regions indicated; the area of the indicator box corresponds to 5500 TH-positive neurons. SNpd 5 substantia nigra pars dorsalis;
SNpl 5 substantia nigra pars lateralis; N 5 nigrosome.
Quantitative assessment of dopamine-containing
neurons in nigrosomes 1–5 and in the nigral
matrix
Previous studies have provided counts of the total numbers
of dopamine-containing neurons in the substantia nigra pars
compacta in normal and parkinsonian brains, but to our
knowledge no quantitative information has been available
on dopamine-containing neurons distributed within different
subgroups within the substantia nigra pars compacta. The
calbindin-based method permitted us to carry out such a
quantitative analysis and to estimate the variation in cell
populations within each subgroup (nigrosomes 1–5 and the
matrix). The results reinforced the impression of relative
consistency in these subgroups from brain to brain. The total
number of dopamine-containing neuronsestimated for normal
human midbrain by this method is in good accordance with
values published previously (McGeer et al., 1977; Hirsch
et al., 1988; German et al., 1989; Pakkenberg et al., 1991).
There are more dopamine-containing neurons in the matrix
than in the nigrosomes, even though the nigrosomes include
prominent clusters of cells.
Source of the calbindin-positive fibre template
in the nigral complex
The calbindin-positive neuropil analysed here was formed
by fibres running through the full rostrocaudal extent of the
midbrain in an orientation roughly parallel to that of fibres
in the cerebral peduncle. This pattern is similar to that of the
striatonigral pathway in the monkey (Hedreen and Delong,
1991). Calbindin D
28K
immunostaining has been observed in
medium spiny projection neurons in the human striatum
(Kiyama et al., 1990), and calbindin immunostaining is
greatly decreased in post-mortem samples of the ventral
midbrain from patients who suffered from Huntington’s
disease (Kiyama et al., 1990) and striatonigral degeneration
Human substantia nigra 1433
Fig. 11 Distributions of TH-positive neurons in the different groups of the midbrain. Filled circles indicate the mean numbers of
dopamine-containing neurons and bars indicate the SEM calculated from the three midbrains studied extensively in near-serial sections.
Values on the abscissa indicate distance (in mm) from rostral exit of third cranial nerve fibres. Values on the ordinate indicate the
number of TH-positive neurons. Areas under the curves correspond to the total numbers of neurons in the regions indicated; the area of
the indicator box corresponds to 5500 TH-positive neurons. CGS 5 central grey substance; M 5 medial group; Mv 5 medioventral
group; A8 5 dopaminergic group A8; SN 5 substantia nigra.
Table 3 Quantitative analysis of dopamine-containing neurons in the different midbrain
groups
CGS A8 M Mv SN
Total 5850 6 865 20 651 6 3212 9787 6 1297 27 964 6 1559 135 000 6 16082
Rostral 1052 6 341 287 6 94 944 6 320 4113 6 666 17 176 6 2935
Intermediate 2276 6 504 4158 6 1203 5301 6 1314 13 669 6 1699 59 515 6 9625
Caudal 2522 6 470 16 206 6 2164 3542 6 1152 10 180 6 1996 58 309 6 10381
Values are mean 6 standard error of the mean. CGS 5 central grey substance; A8 5 dopaminergic
group A8; M 5 medial group; Mv 5 medioventral group; SN 5 substantia nigra.
(Ito et al., 1992), both diseases leading to massive
degeneration of these striatal projection neurons. These
findings strongly suggest that the calbindin-positive neuropil
in the nigral complex is mainly formed by GABAergic
striatonigral fibres and terminals.
Interestingly, the intermixing between dopamine-
containing neurons of the substantia nigra pars compacta and
the calbindin-positive neuropil varies markedly in different
species (Fig. 12). In the rat, almost all dopamine-containing
neurons are located dorsal to the calbindin-positive neuropil.
In the monkey, some densely packed dopamine-containing
neurons invaginate the calbindin-rich region, mainly within
calbindin-poor, finger-like zones. In the human, the
penetration of dopamine-containing neuron clusters into the
1434 P. Damier et al.
Fig. 12 Transverse sections immunostained for calbindin D
28K
(A, C) and corresponding charts of TH-
positive neurons (B, D) from rat (A, B) and squirrel monkey (C, D) midbrain. RN 5 red nucleus.
Scale bar 5 0.7 mm in A and B,and1mminCand D.
Table 4 Correspondence between nigral subdivisions defined by calbindin patterns and other classifications
N1 N2 N3 N4 N5 Matrix SNpd SNpl Mv
Hassler, 1937 Spe (d, v, z) Spv (m, i, l) Spcd Spd (v, i, d) Sal Sal Spd Spcd Sam (α–δ)
Sai (m, l, z) Sai (m, l, z) mH
Spzv Spz (v, z)
Olszewski and ααPars βββγPars Pn
Baxter, 1954 lateralis γγ lateralis Pbpg
Braak and Braak, pl pm m ps Di Di m am
1985 al al
ai ai
Fearnley and Lees, vl (m, i ,l) vm (m, i, l) Pars dm (?) dl (?) dl (?) dl Pars
1991 lateralis lateralis
van Domburg pl pm Pars ps al al Pars am
and ten lateralis lateralis pm
Donkelaar, 1991
McRitchie et al., Ventral Ventro- Pars Dorsolateral Dorsal Dorsal Outside Pars Pars
1995 intermediate/ median lateralis intermediate/ intermediate/ SNpc lateralis medialis
ventrolateral dorsolateral dorsolateral
N 5 nigrosome; SNpd 5 substantia nigra pars dorsalis; SNpl 5 substantia nigra pars lateralis; Mv 5 medioventral group. Hassler: for
Spe, Spv, Spcd, Spd, Sal, Spcd, Sam, Sai, Spzv, Spz, S 5 substantia nigra; p 5 posterior; a 5 anterior; e 5 external; m 5 medial; i 5
intermediate; l 5 lateral; v 5 ventral; z 5 central; d 5 dorsal; c 5 caudal; mH 5 medial horn. Olszewski and Baxter: Pn 5
paranigralis; Pbpg 5 parabrachialis pigmentosum. Braak and Braak: pl 5 posterolateral; pm 5 posteromedial; m 5 magnocellular; ps 5
posterosuperior; am 5 anteromedial; al 5 anterolateral; ai 5 antero-intermediate, subnuclei; Di 5 pars diffusa. Fearnley and Lees: for
vl, vm, dm, dl, v 5 ventral; d 5 dorsal; m 5 medial; i 5 intermediate; l 5 lateral. van Domburg and ten Donkelaar: pl 5
posterolateralis; pm 5 posteromedialis; ps 5 posterosuperior; al 5 anterolateralis; am 5 anteromedialis.
Human substantia nigra 1435
calbindin-positive matrix and pockets is much more
pronounced, with deeply buried calbindin-poor invaginations.
The reasons for the evolution of this particular nigrosomal
organization are unclear, but the differences may be important
to the functionsofthenigralcomplex.MostoftheGABAergic
nigral pars reticulata neurons are also probably included
within the calbindin-positive neuropil; these neurons are
scattered among the endings of striatonigral fibres in the
monkey (Francois et al., 1985). It has been found in the rat
that some collaterals of projection neurons in the substantia
nigra pars reticulata directly synapse on, and inhibit,
dopamine-containing neurons of the substantia nigra pars
compacta (Tepper et al., 1995). There are also interactions
between striatonigral afferents and neurons of both the
pars reticulata and pars compacta of the substantia nigra
(Timmerman and Abercrombie, 1996). Such interactions may
be constrained by the nigrosomal organization described here
for the substantia nigra in the human.
Given the relationship between the calbindin
immunostained neuropil and the concentration of TH-positive
neurons in the calbindin-poor pockets, it seems likely that
most striatonigral afferents project more strongly to the
dopamine-containing neurons included in the matrix than to
those belonging to nigrosomes. There is as yet no evidence
to support such differential connectivity. Similarly, the
projection sites of the dopamine-containing neurons of these
different substantia nigra pars compacta compartments are
still not known. One might expect, however, that these are
different, and that the different compartments might project
differentially to the putamen or caudate nucleus, or to the
different histochemical compartments of the striatum, i.e. the
striosomes and matrix (Graybiel and Ragsdale, 1978). Such
differential projection patterns have been suggested in the
cat and monkey (Szabo, 1980; Parent et al., 1983; Jime
´
nez-
Castellanos and Graybiel, 1987; Langer and Graybiel, 1989),
but this issue will not be easy to resolve in the human
brain. What the compartmental patterns of calbindin
immunohistochemistry do provide, for the first time, is a
reliable and easily reproducible method for analysing the
human substantia nigra complex independent of its dopamine-
containing neurons. As we show in the accompanying paper,
this compartmental analysis allows the substantia nigra to be
studied with great accuracy in brains from patients who
suffered from Parkinson’s disease.
Acknowledgements
We wish to thank Mrs Mouatt-Prigent for her helpful
assistance. This study was supported by NIH Javits Award
NS25529, the National Parkinson Foundation, the Fondation
pour la Recherche Me
´
dicale and the French Foreign Office
(programme Lavoisier).
References
Abercrombie M. Estimation of nuclear population from microtome
sections. Anat Rec 1946; 94: 239–47.
Braak H, Braak E. Nuclear configuration and neuronal types of the
nucleus niger in the brain of the human adult. Hum Neurobiol 1986;
5: 71–82.
Damier P, Hirsch EC, Agid Y, Graybiel AM. The substantia nigra
of human brain. II. Patterns of loss of dopamine-containing neurones
in Parkinson’s disease. Brain 1999; 122: 000–000.
Fearnley JM, Lees AJ. Ageing and Parkinson’s disease: substantia
nigra regional selectivity. Brain 1991; 114: 2283–301.
Francois C, Percheron G, Yelnik J, Heyner S. A histological atlas
of the macaque (Macaca mulatta) substantia nigra in ventricular
coordinates. Brain Res Bull 1985; 14: 349–67.
German DC, Manaye K, Smith WK, Woodward DJ, Saper CB.
Midbrain dopaminergic cell loss in Parkinson’s disease: computer
visualization [see comments]. Ann Neurol 1989; 26: 507–14.
Comment in: Ann Neurol 1989; 28: 111–2.
Gibb WR. Melanin, tyrosine hydroxylase, calbindin and substance
P in the human midbrain and substantia nigra in relation to
nigrostriatal projections and differential neuronal susceptibility in
Parkinson’s disease. Brain Res 1992; 581: 283–91.
Gibb WR, Lees AJ. Anatomy, pigmentation, ventral and dorsal
subpopulations of the substantia nigra, and differential cell death in
Parkinson’s disease. J Neurol Neurosurg Psychiatry 1991; 54:
388–96.
Graybiel AM, Ragsdale CW Jr. Histochemically distinct
compartments in the striatum of human, monkey, and cat
demonstrated by acetylthiocholinesterase staining. Proc Natl Acad
Sci USA 1978; 75: 5723–6.
Graybiel AM, Hirsch EC, Agid YA. Differences in tyrosine
hydroxylase-like immunoreactivity characterize the mesostriatal
innervation of striosomes and extrastriosomal matrix at maturity.
Proc Natl Acad Sci USA 1987; 84: 303–7.
Hassler R. Zur Normalanatomie der Substantia nigra. Versuch einer
architektonischen Gliederung. J Psychol Neurol 1937; 48: 1–55.
Hassler R. Zur Pathologie der Paralysis agitans und des
postenzephalitischen Parkinsonismus. J Psychol Neurol 1938; 48:
387–476.
Hedreen JC, DeLong MR. Organization of striatopallidal,
striatonigral, and nigrostriatal projections in the macaque. J Comp
Neurol 1991; 304: 569–95.
Hirsch EC, Graybiel AM, Agid YA. Melanized dopaminergic
neurons are differentially susceptible to degeneration in Parkinson’s
disease. Nature 1988; 334: 345–8.
Ito H, Goto S, Sakamoto S, Hirano A. Calbindin-D28k in the basal
ganglia of patients with Parkinsonism. Ann Neurol 1992; 32: 543–50.
Jime
´
nez-Castellanos J, Graybiel AM. Subdivisions of the dopamine-
containing A8-A9-A10 complex identified by their differential
mesostriatal innervation of striosomes and extrastriosomal matrix.
Neuroscience 1987; 23: 223–42.
Kiyama H, Seto-Ohshima A, Emson PC. Calbindin D28K as a
marker for the degeneration of the striatonigral pathway in
Huntington’s disease. Brain Res 1990; 525: 209–14.
Konigsmark BW. Methods for the counting of neurons. In: Nauta
1436 P. Damier et al.
WJH, Ebbeson SOE, editors. Contemporary research methods in
neuroanatomy. Berlin: Springer; 1970. p. 315–80.
Langer LF, Graybiel AM. Distinct nigrostriatal projection systems
innervate striosomes and matrix in the primate striatum. Brain Res
1989; 498: 344–50.
McGeer PL, McGeer EG, Suzuki JS. Aging and extrapyramidal
function. Arch Neurol 1977; 34: 33–5.
McRitchie DA, Halliday GM, Cartwright H. Quantitative analysis
of the variability of substantia nigra pigmented cell clusters in the
human. Neuroscience 1995; 68: 539–51.
Olszewski J, Baxter D. Cytoarchitecture of the human brainstem.
Philadelphia: Lippincott; 1954. p. 131–2.
Pakkenberg B, Moller A, Gundersen HJ, Mouritzen Dam A,
Pakkenberg H. The absolute number of nerve cells in substantia
nigra in normal subjects and in patients with Parkinson’s disease
estimated with an unbiased stereological method. J Neurol Neurosurg
Psychiatry 1991; 54: 30–3.
Parent A, Mackey A, De Bellefeuille L. The subcortical afferents
to caudate nucleus and putamen in primate: a fluorescence retrograde
double labeling study. Neuroscience 1983; 10: 1137–50.
Szabo J. Organization of the ascending striatal afferents in monkeys.
J Comp Neurol 1980; 189: 307–21.
Tepper JM, Martin LP, Anderson DR. GABA-A receptor-mediated
inhibition of rat substantia nigra dopaminergic neurons by pars
reticulata projection neurons. J Neurosci 1995; 15: 3092–103.
Timmerman W, Abercrombie ED. Amphetamine-induced release of
dendritic dopamine in substantia nigra pars reticulata: D1-mediated
behavioral and electrophysiological effects. Synapse 1996; 23:
280–91.
van Domburg PHMF, ten Donkelaar HJ. The human substantia
nigra and ventral tegmental area. In: Beck F, Hild W, Kriz W, Pauly
JE, Sano Y, Schiebler TH, editors. The human substantia nigra and
ventral tegmental area. Berlin: Springer-Verlag; 1991. p. 32–69.
Received December 10, 1998. Revised March 11, 1999.
Accepted March 15, 1999
