Morphology and distribution of neuropeptide-containing neurons in human cerebral cortex

Neuroscience Vol. 51, No. 2, pp. 363-375, 1992 Printed in Great Britain

0306-4522/92 $5.00 + 0.00 Pergamon Press Ltd © 1992 IBRO

MORPHOLOGY A N D DISTRIBUTION OF NEUROPEPTIDE-CONTAINING NEURONS IN H U M A N CEREBRAL CORTEX J. P. HORNUNG,*t N. DE TRIBOLET,~ and I. TrRK§ tInstitute of Anatomy, University of Lausanne, Rue du Bugnon 9, CH-1005 Lausanne, Switzerland :~Neurosurgical Service, CHUV, CH-1011 Lausanne, Switzerland §School of Anatomy, The New South Wales University, Kensington, Sydney, Australia Abstract--Biopsies of human cerebral cortex were fixed by immersion and immunostained for the detection of neuropeptides in neuronal cell bodies and axons. Four neuropeptides (neuropeptide Y, somatostatin, substance P and cholecystokinin) were visualized in a series of adjacent sections. All populations of immunoreactive neurons had a morphology characteristic of interneurons, with variations in dendritic arborizations and laminar distribution. The cholecystokinin-immunoreactive neurons were most numerous in the supragranular layers, whereas neurons containing the other three peptides occurred mainly in infragranular layers, or even in neurons populating the subcortical white matter. Quantitatively, each population of neuropeptide-containing neurons accounted for 1.4-2.5% of the total neuronal population. The distribution of these neurons varied slightly between cytoarchitectonic divisions, with substance P- and somatostatin-immunoreactive neurons dominating in the temporal lobe and cholecystokinin-immunoreactive neurons in the frontal lobe. Neuropeptide Y-immunoreactive neurons dominated in the gray matter of the frontal half of the hemisphere and in the subcortical white matter of the caudal half of the hemisphere. Furthermore, co-existence of neuropeptide Y or substance P immunoreactivity within somatostatin-immunoreactive neurons could be demonstrated using double labeling immunofluorescence techniques. The axonal plexuses immunoreactive for neuropeptide Y, somatostatin, or substance P were distributed in all layers, with a strong predominance of horizontally oriented fibers in layer I, a moderate plexus of randomly oriented fibers in the supra- and infragranular layers, and a slightly weaker innervation of layer IV. Immunoreactive axons formed, in addition, complex terminal arbors, mostly in older subjects, suggesting that they resulted from an as yet undefined aging process. The present study underlines several aspects of the organization of the neuropeptide-containing neurons of the human cerebral cortex, which are of particular interest in the light of the involvement of these neurons in several neurodegenerative diseases.

Cerebral cortical neurons are two major types characterized morphologically by their dendritic and axonal arborizations: projection neurons (pyramidal neurons) and interneurons (non-pyramidal neurons). 7,45 M a n y interneurons use G A B A as their principal neurotransmitter, ~9'23'49 and some of these also contain one or several neuropeptides [e.g. neuropeptide Y (NPY), somatostatin, substance P (SP), cholecystokinin (CCK)]. 24,30,33,34,39,51 The interneurons can be subdivided into several types, each characterized by specific morphology, connections with other cortical neurons, and biochemical markers. 25,42 They have been implicated in several functional and developmental features of the cerebral cortex. Experimental studies, in cat or monkey, have illustrated the role o f interneurons in modulating response properties of cortical cells. 22,37,5° A subgroup of peptide-containing interneurons is found transiently in the subcortical white matter during late embryonic life, 14 and seems *To whom correspondence should be addressed. Abbreviations: CCK, cholecystokinin; DAB, diaminoben-

zidine; HRP, horseradish peroxidase; -IR, -immunoreactive; NPY, neuropeptide Y; SP, substance P. 363

to contribute to the guidance of thalamocortical afferents to their appropriate target areas, is Neuropeptides have also been demonstrated in the human cerebral cortex with both biochemicaP ,52 and histochemical 1'5'6'1°'16'35'a4'46'47techniques. These neuropeptide-containing neurons have a morphology and a distribution corresponding to that found in other mammalian species. It has also been established that the levels of neuropeptides measured with biochemical techniques and the number of neuronal processes containing those neuropeptides in the cerebral cortex are reduced in several neurological disorders. In particular, the distributions of N P Y and somatostatin are altered in Alzheimer's 9,11,21,36,43 and Parkinson's 36 diseases. It is therefore warranted to have detailed knowledge on the normal distribution and connections of the peptide-containing systems in the human cerebral cortex. Such baseline information is essential if one is to define, after pathological degeneration, which elements of the cortical circuitry were most affected. The use of biopsies of the cerebral cortex offers optimal preservation of the morphology and of the antigens to be demonstated by immunocytochemical analysis. 12 In the present study we

J.P. HORNUNG et al.

364

describe a n d c o m p a r e the morphologies a n d distrib u t i o n s o f four p o p u l a t i o n s o f neuropeptide-containing n e u r o n s of the n o r m a l h u m a n cerebral cortex, with particular a t t e n t i o n to their co-localization in certain neurons.

EXPERIMENTAL PROCEDURES

Fresh biopsies were collected from patients suffering from deep brain tumors. The specimen was trimmed to slices of 5-20 mm thickness, and immersed in a fixative solution (4% paraformaldehyde in 0.1M phosphate) within 1-15 min of excision. These slices were kept in fixative at 4°C for three to six days, then rinsed in phosphate buffer, and cut in 50-#m-thick sections with a Vibratome (Lancer, Oxford Instruments). Free-foating sections were incubated in a series of baths, each containing 0.05 M phosphate, 0.9% NaC1, 0.1-0.3% Triton X-100 at pH 7.35, to which was added: (1) 5% blocking serum for 1 h, (2) primary antibody for 48 h, (3) secondary biotinylated antibody diluted 1 : 200 for 1 h (Vector, Burlingame, CA, U.S.A.), (4) avidim horseradish peroxidase (HRP) complex (Vector), diluted 1 : 100 (or 1 : 50 for ABC-elite from Vector) for 1 h. 29 Each of these steps was followed by three changes of phosphatebuffered saline. The HRP complex was revealed by incubation for 3-5min in 0.05% diaminobenzidine (DAB, Sigma) and 0.003% hydrogen peroxide in Tris buffer, pH 7.0, to which 0.2% nickel ammonium sulfate was added to increase the contrast of the reaction product. Dilutions of primary antibodies were NPY 1:3000 (gift from Dr W. Blessing), SP 1:200 (Sera-Lab, U.K.), CCK 1:800 (Amersham, U.K.); somatostatin (Ortho-Lab, Germany) was used undiluted as supplied. For double labeling experiments, fluorescein-conjugated anti-goat or anti-rat immunoglobulin and rhodamine-conjugated antirabbit immunoglobulin antibodies (Dako, Denmark) were used to visualize two different antigens in the same section. The specificity of each antibody had already been tested by the manufacturer or demonstrated in previous studies. 4'15 Control tests were performed by replacing the primary antibodies with normal serum which resulted in a completely negative immunostaining. In the double-labeling experiments, removing one of the primary antibodies, but leaving the two fluorescent-labeled secondary antibodies resulted in the selective and complete disappearence of the corresponding marker in the histological preparation. Sections adjacent to the immunoreacted ones were Nissl-stained to demonstrate the cytoarchitecture of the cortical area studied and to verify that the sample used was free from major pathological alterations, such as substantial neuronal loss or glial proliferation. Since both poor fixation and pathology of the cerebral cortex could result in a poor or abnormal pattern of immunolabeling, the following two criteria were used to retain the sections for analysis: (1) in Nissl-stained sections, no signs of abnormal cytoarchitecture or gliosis should be observed; (2) in immunoreacted sections, the overall pattern of staining should have an intensity and distribution in accordance with the majority of healthy human samples taken in other experiments from the same region of cerebral cortex and with published data from corresponding regions in primate brains. Twenty-one samples were collected from the frontal lobe, 13 from the temporal lobe, four from the parietal lobe, and one from the occipital lobe. Twenty-seven out of the total of 39 samples were retained for the present study. All patients, with ages ranging from six to 77 years (average: 48.5 years), had no neurological symptoms, except those related to the tumors for which they were operated. The biopsies and the histological and histochemical studies reported here were performed in accordance with the rules and regulations maintained by the Ethical Committee of

the University of Lausanne and by the Committee on Experimental Procedures Involving Human Subjects of the University of New South Wales. For the quantitative analysis, sets of five adjacent sections were considered for each case analysed: one Nissl-stained and four immunostained sections (one for each neuropeptide). Cell counts were restricted to a 300-#m-wide vertical strip extending from the pia mater to 500 pm deep in the white matter. For each sample, at least three counts were made. The number of immunoreactive neurons was expressed as a percentage of the number of neurons counted in the corresponding region of the adjacent Nissl-stained section. Since the purpose of the counting procedure was primarily to establish the relative proportion of certain chemically identified neurons in the total population of neurons within the cortex, no correction for doublecounting or shrinkage was made. RESULTS T h e locations of the 27 samples included in the present study, with reference to the lobe and in some cases the area o f origin, are listed in Table 1. All samples were collected f r o m association (homotypic) cortices, a n d thus the cytoarchitecture of these different areas was basically similar. Moreover, the m o r p h o l o g y of the neuropeptide-containing n e u r o n s a n d the quantitative m e a s u r e m e n t s of n e u r o n a l density did not show systematic differences between these regions. Therefore, d a t a of all samples have often been pooled in the following descriptions.

Morphology of neuropeptide-containing neurons In addition to the cell body, neuropeptidei m m u n o r e a c t i v e (-IR) n e u r o n s h a d their primary dendrites and, for certain antibodies, even m o r e distal dendritic b r a n c h e s stained, allowing c o m p a r i s o n with d a t a o n the various morphologies of cortical i n t e r n e u r o n s published in Golgi studies. 7'38 T h e N P Y - I R n e u r o n s were p r e d o m i n a n t l y multipolar, triangular a n d horizontal (Figs 1 a n d 2). The p r o x i m a l dendrites o f these cells were short, thick a n d s m o o t h , while the distal dendrites became progressively t h i n n e r a n d varicose (Figs 1 a n d 2). The S P - I R n e u r o n s fell into two categories: weakly i m m u n o reactive cells (more numerous, p r e d o m i n a n t l y in the middle layers) a n d strongly immunoreactive cells (less n u m e r o u s , principally located in the deep layers). The m o r p h o l o g y o f the latter p o p u l a t i o n was considered Table 1. Localization of the biopsies Frontal lobe

area 9 (5) area 10 (2) area 46 (4) undefined (3)

Temporal lobe

area 20 (2) area 21 (4) area 22 (1) undefined (3)

Parietal lobe

area 39 (1) area 40 (1)

Occipital lobe

area 19 (1)

(n) = number of samples taken in this area.

Peptidergic human cortical neurons

Fig. 1. (a) Neurons immunoreactive for NPY in layer V of frontal cortex. Note the NPY-IR axonal plexus in the surrounding neuropil with fibers oriented in all directions. (b) NPY-IR neuron in the frontal subcortical white matter, 150 p m below layer VI. Notice the axon leaving a primary dendrite (arrowheads), heading towards layer VI and crossing horizontally oriented immunoreactive axons. (c) Layer V SP-IR neuron in frontal cortex, with its axon leaving the cell body (arrow). Many other SP-IR axons are present in the neuropil (arrowheads). (d) Layer VI somatostatin-IR in temporal cortex with its axon leaving the cell body (arrow). (e) Two layer II CCK-IR neurons (arrowheads) in frontal cortex. In all panels, pial surface is towards the top. Scale bars = 50 #m.

365

366

J.P. HORNUNGet al.

NPY III

:

V

WM

ss

j

ccKI

Via

Fig. 2. Camera lucida drawings of representative neuropeptide-containing neuronal cell types. The laminar localization is given for each neuron, and axonal processes, when identified, are indicated by arrows. SS, somatostatin. Scale bar = 50 #m.

in the present study. They displayed several types of dendritic arborizations, the most frequent being the multipolar type (Figs 1 and 2), but some triangular or even less frequently bipolar neurons were also seen. The somatostatin-IR neurons displayed mainly multipolar and triangular morphologies, although some had horizontally oriented dendrites (Figs 1 and 2). The CCK-IR neurons belonged to several morpho-

logical types (Figs 1 and 2). Many had thin, elongated dendrites, with two main dendritic trunks arising from the upper and lower poles of the soma, providing a bipolar morphology to the neuron. Others had a multipolar morphology, with several short primary dendrites. Finally, a few had a triangular shape, but without a prominent apical dendrite, distinguishing these neurons from typical pyramidal cells.

Peptidergic human cortical neurons

Distribution of neuropeptide-immunoreactive neurons The laminar distribution of neuropeptidecontaining neurons presented two basic patterns (Fig. 3). One, made by CCK-IR neurons, was mostly confined to the supragranular layers, with a few isolated neurons randomly distributed in the infragranular layers and subcortical white matter. The second pattern, made of NPY-, somatostatin-, and dark SP-IR neurons, was mostly confined to the infragranular layers and subcortical white matter, with a few immunoreactive neurons dispersed in the supragranular layers. Each neuropeptide-containing neuronal population accounted for a small fraction of all cortical neurons. CCK-IR neurons were the most numerous (2.55% of all cortical neurons), followed by NPY-IR (2.12%), somatostatin-IR (1.51%) and dark SP-IR (1.45%) neurons. Light SP-IR neurons were 105-2 more frequent than dark SP-IR neurons. Thus, all SP-IR neurons represented 3.5-4% of all cortical neurons. In the white matter bordering the cerebral cortex, 20-45% of all neurons were immunoreactive for each of the four peptides studied. Comparison of the densities of the neuropeptidecontaining neurons between areas of a given lobe did not reveal significant differences. However, comparison of measurements made in the different lobes revealed some trends in these density values. Neuropeptide Y-IR neurons in the gray matter dominated in the frontal and temporal lobes, with densities of 20% and 30% higher than in the parietal and occipital lobes, respectively. However, in the subcortical white matter, the proportion of NPY-IR neurons was slightly higher in the parietal and occipital lobes, as compared to the frontal and temporal lobes. As a result, the overall distribution of NPY-IR neurons across the cerebral cortex was fairly homogeneous. Dark SP-IR neurons were evenly distributed throughout the cerebral cortex with the exception of the temporal lobe, where their density was about 30% higher than in the other lobes, owing to increase of these neurons in layers V and VI. Among all the areas sampled in the temporal lobe, the increase in the density of SP-IR neurons was observed primarily in area 21. The CCK-IR neurons were uniformly distributed, except in the frontal lobe where their density was increased by about 30% in the supragranular layers. A modest increase in the density of somatostatin-IR neurons was observed in the infragranular layers of the areas sampled in the temporal lobe.

Distribution axons

and morphologies of immunoreactive

The different antibodies also revealed, at least in part, an axonal plexus containing the neuropeptides studied. The quality of labeling was, however, different in the case of different neuropeptides. The CCK antibody was the least effective, and the NPY anti-

367

body the most effective in revealing axonal plexuses. Most neuropeptide-IR neurons were thin, with small (about 1 pm in diameter) and round regularly spaced varicosities. The distribution of NPY-IR fibers was rather homogeneous across layers, except for a slightly stronger innervation of layer I. In layer I, axons lay horizontally whereas in the deeper layers they were oriented randomly in all directions (Figs 1a and 4). Of all layers, layer IV contained the lowest density of NPY-IR axons. In the subcortical white matter, NPY-IR axons were less numerous and ran mostly in two directions, parallel or perpendicular to the interface between the cortex and subcortical white matter (Fig. lb). The SP-IR axons belonged to two classes: axons with solid labeling running through all layers, and axons with dotted, discontinuous labeling, restricted to two tiers, one in layer II and another in layer V (Fig. 5). The axons with solid staining formed a dense plexus in the superficial half of layer I. Layer IV was least innervated whereas layers III, V and VI had an intermediate density of SP-IR fibers. In the subcortical white matter, SP-IR axons were distributed like NPY-IR neurons, mainly parallel to the subcortical, unstained fiber bundles. In comparison to the NPY-IR and SP-IR axons, fewer somatostatin-IR axons were demonstrated with the antibody used for this study (Fig. 5). Fibers were found throughout the layers and extended into the subcortical white matter. In layer I, fibers had a predominantly horizontal orientation, but in layers II-IV they were oriented in all directions, and layers V and VI a number of fibers followed a straight and radial trajectory through the cortex and remained for a long distance with the plane of section, perpendicular to the pial surface. Similarly oriented fibers were also revealed with NPY antibodies. In the subcortical white matter, the few immunoreactive fibers were mostly oriented parallel to the boundary of the white matter, eventually turning perpendicular to it as they entered the gray matter. Few axons were stained with the CCK antibody used in the present study. They were found diffusely through all layers of the cortex, but were absent in the subcortical white matter. In addition to the axons with small varicosities, NPY, somatostatin, and SP antibodies revealed complex axonal arborizations with large varicosities clearly different from those of the diffuse axonal plexus (Fig. 6). The complex terminals occupied fields of about 100/~m wide. Such arbors occurred in all layers of the cortex, except layer I and in all samples investigated. There were no indications of any laminar or cellular selectivity in the distribution of these complex axon terminals. In addition, a few CCK-IR axons were found in close association with CCK-IR cell bodies (Fig. 7) and their proximal dendrites. These CCK-IR pericellular arrays were encountered in several areas of the cortex (frontal and temporal association cortices) and exclusively in layer II.

368

J. P. HORNUNG et al.

SS

CCK

1mm

Fig. 3. Distribution of neuropeptide-containing neurons in four adjacent sections of the frontal cerebral cortex. For SP-IR neurons, solid circles represent strongly immunoreactive neurons, open circles weakly stained neurons. SS, somatostatin. Scale bar = 1 mm.

CCK

NPY

I

j,

~\\\\\\\\\\N--

m

i

IV ~

~

IV

v

v

Vl

Vl

WM

~

~

I

0

~

1l0

WM

i

I

510

20 3~0 410 % OF LABELLED NEURONS

0

l

10

210 3~0 4'0 % OF LABELLED NEURONS

SS

SP

I

IV

~\\~"~

iv

v

V VI

5'0

~\~\~\\\'~\\\\~I

--

~\\\\\\\\\\\\\\\\\\\~,---

vI

\\\\\\\\\\\\\\\\'~'--

WM

I

i

i

10

20

30

410

% OF L A B E L L E D NEURONS

5t0

I

r

I

0

10

20

310

' 40

% OF L A B E L L E D NEURONS

Fig. 4. Quantitative analysis of the distribution of neuropeptide-containing neurons. Striped bars represent the percentage of immunoreactive neurons present in each cortical layer relative to the total number of immunoreactive neurons in all layers. Dotted bars represent the percentage of immunoreactive neurons relative to the total number of neurons counted in the same layer. Note the conspicuous absence of NPY-, somatostatin-, and SP-IR neurons in layer I. SS, somatostatin.

i 50

Peptidergic human cortical neurons

NPY

369

SS

SP

II

III

IV

V

VI \"-/,'\'

J

\

~ -

~ iv

\~,~

-

Fig. 5. Camera lucida drawings of neuropeptide-containing axons in corresponding stripes of cortex in three adjacent sections of frontal association cortex. A few immunoreactive neurons are present, mainly in infragranular layers. SS, somatostatin. Scale bar = 50 #m.

Co-existence o f peptides in infragranular neurons In three samples of frontal association cortex and one of temporal cortex, sections were immunoreacted for two different peptides: somatostatin and NPY, or somatostatin and SP. The somatostatin-IR neurons were revealed with fluorescein-conjugated secondary antibodies and N P Y - and SP-IR neurons with rhodamine-conjugated antibodies. In several instances, somatostatin-IR neurons were also found to be N P Y I R or SP-IR (Fig. 8). The double-labeled neurons were primarily located in layer VI and the subcortical white matter. However, many neurons displayed immunoreactivity for a single neuropeptide only.

Since the intensity of labeling with fluorescent markers was weaker than the labeling with H R P conjugated antibodies, the possibility remained that singly-immunoreactive neurons resulted from incomplete detection of peptides. DISCUSSION

In the present study, we have described the morphology and distribution of four neuropeptides in h u m a n cortical neurons. They formed defined subpopulations of interneurons, and had characteristic dendritic and somatic morphologies, axonal branches and laminar distributions of their cell bodies. They

370

J.P. HORNUNGet al.

Fig. 6. Neuropeptide-containing axonal plexuses (arrows) characterized by the clustering of fibers or large varicosities. (A) SP; (B) somatostatin; (C) NPY. Compare these fibers with the axonal arborizations made by the straight fibers with smaller varicosities (arrowheads; a and c). Scale bar = 50/~m. could be divided into two groups, one containing CCK-immunoreactivity located primarily in the supragranular layers, and the other containing NPY, SP, or somatostatin immunoreactivity mainly in the infragranular layers and subcortical white matter. Furthermore, there was direct and indirect evidence for the co-localization of somatostatin with N P Y or SP in the same neurons. The cerebral cortex also

contained dense NPY-, somatostatin-, or SP-IR axonal plexuses innervating mainly layer I and least layer IV. In older cases, neuropeptide-containing complex terminal arbors appeared in the cerebral cortex. S o m e technical considerations

The use of freshly fixed biopsies allowed the detection of neuropeptide-containing neurons and axons

Fig. 7. Two examples of CCK-IR axons with varicosities (arrowheads) aligned along the surface of layer II CCK-IR neuronal cell bodies (arrows), in frontal association cortex. The left cell in A is a CCK-IR neuron without apparent CCK fiber innervation. Scale bar = 50/~m.

Peptidergic human cortical neurons

371

a

!ii

Fig. 8. Pairs of micrographs demonstrating the co-existence of NPY-(a) and somatostatin-(b) IR in the same neurons, or SP-(c) as well as somatostatin-(d) IR in another neuron at curved arrows. Rhodamine labels neurons in a and c and fluorescein in b and d. Neuronal pigments, frequent in adult human neuronal cell bodies, are revealed with fluorescence illumination for fluorescein. Scale bar = 50 #m. with minimal morphological alterations and loss of neuropeptide content. However, during immersion fixation, the fixative penetrated slowly into the block of tissue, resulting in a relatively delayed fixation of its deeper regions. Fortunately, since neuropeptides diffuse and degrade very slowly, their immunostaining of neuropeptides through all layers appeared rather homogeneous. Neuropeptides are present in the nervous tissue at low concentrations, hence often being at the limit of detection sensitivity of immunocytochemistry. Injection of colchicine prior to fixation is sometimes used to enhance the intensity of the immunocytochemical reaction, ~7'26 but this procedure cannot be applied to the material collected in

the human brain for obvious ethical and technical reasons. Therefore, there is a chance of underestimating the population of neurons containing the neuropeptides. Comparison with fairly similar data on the morphology and distribution of neuropeptide-IR neurons in several m a m m a l i a n species 31 shows that these present results have not suffered substantially from the absence of colchicine treatment.

Morphology and distribution of peptide-containing neurons. Although a large variety of cell types were immunoreactive for one or more of the four neuropeptides, they were all non-pyramidal neurons. Even

372

J.P. HORNUNGet al.

if some of them had a triangular cell body, they lacked a prominent apical dendrite, characteristic of pyramidal neurons. Therefore, it remains dubious that pyramidal neurons contain neuropeptides, as suggested in some studies on the mammalian cerebral cortex, 13'43'44 but contradicted by others. 6'1°'33 The dendritic morphology of the immunoreactive neurons ranged from bipolar and bitufted to multipolar, and even horizontal. Considering their laminar position could help, for instance, in differentiating layer II bipolar CCK-IR neurons from deep layer III NPY-IR neurons. On the other hand, there was a striking similarity among layer VI horizontal neurons immunoreactive for NPY, SP, and somatostatin. Many neurons located in the subcortical white matter have been shown, in many mammalian brains, to contain neuropeptidesJ 4'33'34'39Although this has also been shown to be true in the human cortex, 6'1° it is noteworthy that at least 50% of the neurons always failed to show immunoreactivity for any of the four peptides. The possibility also therefore exists that some neurons located in the white matter do not contain any neuropeptide. Our choice of relatively thick sections (50#m) was optimal for analysing the morphology and the distribution of the neuropeptide-containing neurons, but may have reduced the accuracy of total neuronal counts. However, the densities estimated for each neuropeptide-containing population and the relative distribution of them across areas are still in fairly good agreement with comparable data from animal experiments. Two to five per cent of the cortical neurons are thought to contain one or several peptides, 24'31 which is in the range of the density values found in the present study. The regional heterogeneity in the distribution of these populations has seldom been investigated in primate brains. The distibution of somatostatin-IR neurons was reported to vary greatly between the sparsely populated primary visual cortex and the densely populated anterior cingulate cortex. 8 The homologous areas of the human brain were not part of the sample collected for the present study. However, the primate associative auditory area was among the most heavily populated areas, as were the homologous areas in the superior temporal lobe in our sampling. The proportion of light and dark SP-IR neurons described in the macaque cortex 32 is of the same order as that found in our human material. A quantitative study of NPY-IR neurons in the macaque brain 33 concluded that no systematic significant variation in distribution could be demonstrated, due to a large variability of measurements between individuals. Taken together, these reports and our present results converge in many respects towards similar conclusions. Considering that peptide-containing neurons were also shown to contain GABA, 17'51 that there was a certain degree of co-localization among neuropeptide-containing neurons, and that GABAergic neurons form approximately 22% of the total popu-

lation of human cortical neurons,28 one can deduce that the four neuropeptides studied here are found in about 20-30% of GABAergic neurons. 31

Co-localization of neuropeptides The co-localization of neuropeptides, in particular NPY and somatostatin, in cortical neurons has been shown in numerous instances,17'5''56 including in human cerebral cortex. 9'13'55In the present study, two lines of evidence supported this concept. First, labeling of two different neuropeptides in the same section with distinct fluorescent markers demonstrated directly that, in some neurons, SP or NPY co-localized with somatostatin. However, the sensitivity of this technique was lower than that of direct immunoperoxidase staining, and hence did not allow a quantitative evaluation of the degree of co-localization. The second line of evidence was the comparison of the morphology and laminar distribution of the immunoreactive neurons. It was obvious that there were striking morphological similarities between NPY- and somatostatin-IR neurons in layer VI and the subcortical white matter, and up to a certain extent also with SP-IR neurons. The CCK-IR neurons, on the other hand, were clearly distinct. This suggests that, as in non-human mammalian cerebral cortex, 3~ that many cortical neurons in humans contain both NPY- and somatostatin-IR, and that some may also contain SP-IR, while the CCK-IR neurons belong to a separate subpopulation of interneurons. The SP-IR neurons in non-human primates could, as in our own study, be divided into two populations, dark and light immunoreactive neurons. 32 The former population, in the deep layers of the cortex, was shown to also display NPY- and somatostatin-IR, thus expressing neurochemical and laminar characteristics similar to those of the homologous population in our human material.

Origin of neuropeptide-contain&g axonal plexuses The neuropeptide-containing axonal plexuses displayed a laminar distribution with several common features. The fiber density was always highest in layer I, lowest in layer IV, and intermediate in the infraand supragranular layers. Since the cortical neurons containing neuropeptides had the morphology of interneurons, i.e. by definition cells with a local intracortical axonal arborization, these cells were the most likely source for the plexuses of fibers revealed with immunocytochemistry. However, one could not exclude the possibility that some cortical afferents also contain neuropeptides. 1° For example, it is known that in the rat certain serotonergic dorsal raphe neurons also contain SP 26and indirect evidence is now available that SP and serotonin also colocalize in some dorsal raphe neurons in the human? There is an additional substance P-containing cortical afferent that originates in the pontine tegmentum,47'48 but this pathway terminates exclusively in the medial

373

Peptidergic human cortical neurons frontal cortex, which has not been included in the present study. Furthermore, certain noradrenergic neurons of the locus coeruleus accumulate NPY. 4° It is not known, however, whether these neuropeptides are confined to the cell body and dendrites, or whether they are also present at the level of the axon terminals. Local destruction of the infragranular layers of cat visual cortex with excitotoxins almost entirely removes the SP- and N P Y - I R interneurons while, in the remaining cortex, the SP- and N P Y - I R axonal plexuses are drastically reduced but not entirely absent. 27 This suggests that a large proportion of the neuropeptide-containing axonal plexuses have an intrinsic origin, supporting the notion that the plexuses demonstrated in our studies are terminals of the immunoreactive cells identified. Complex axonal terminal arbors

The NPY-, SP-, and somatostatin-IR axons were constantly found to form complex axonal terminal arbors, and to be characterized by numerous larger varicosities clustered in 100-150/~m wide fields. Similar N P Y - I R axons in human cortex 12'13 were considered to be a typical feature of normal human cortical connectivity. These complex terminals were, however, very reminiscent of the aging serotonergic axons described in the adult rat cortex. 54 Similar axonal terminals have also been demonstrated for catecholaminergid ° and serotonergic fibers in human cortex s3 or in aging marmoset cerebral cortex (unpublished observations). There is no evidence that neuropeptide-containing fibers form such complexes in other non-aging adult mammalian cortices. The important question arises whether the complex terminal fields of chemically identified fibers observed in adult human cortex are characteristic features of normal neural connections, or whether they are early signs of aging. In our samples with an average of nearly 50 years of age, most cases presented many complex terminal arbors. However, in the six-yearold and the 24-year-old cases, no such arbors were observed, and only in one of two 30-year-old cases were a few such terminals found. This suggests an association with aging.

CONCLUSIONS The neuropeptide-containing neurons of the human cerebral cortex belong to a heterogeneous population of interneurons. Most of these neurons are likely to also contain G A B A as a neurotransmitter 23m,51 and to be inhibitory in function. The overall laminar distributions of neuropeptide-IR axons show a parallel with that of corticocortical projections and are distinct from that of thalamocortical terminals. There is also evidence for a difference in the density of peptidergic neurons between cortical areas. In the non-human primate, a higher concentration of somatostatin-IR neurons and axons was found in the association (primarily prefrontal) cortex than in primary sensory areas. 34'41 The present study on human cortex also reports a selective distribution of some neuropeptide-containing neurons in different regions of the cortex. The neuropeptide-containing neurons are known to be selectively affected in specific degenerative diseases of human cerebral cortex 2'9'21'41 and the loss of these neurons would result in the alteration of intrinsic cortical connectivity. The specific loss of peptidecontaining interneurons could have a major impact on higher levels of cortical information processing for two reasons. First, the peptide-containing neurons are likely to contain G A B A . 31 Their loss could result in a significant reduction of the fine tuning of neurophysiological responses of cortical units dependent in part on intracortical inhibition. Second, the evidence that larger numbers of peptidergic neurons are present.in certain association cortical areas could be related to the reduction of mnemonic and cognitive functions associated with those areas. Acknowledgements--We are grateful to S. Daldoss and M. Birchen for the illustrations, to C. Blagov and N. Trapp for histology, and to Dr P. G. H. Clarke for linguistic advice. We thank Dr W. W. Blessing for the generous gift of NPY antibody. This work was supported by Swiss NSF grants 3.3230.86 and 3100-009468-3, by the Max Clo&ta Stiftung (Ztirich, Switzerland), and by a grant of NH and MRC of Australia (to I.T.).

REFERENCES

1. Adrian T. E., Allen J. M., Bloom S. R., Ghatei M. A., Rossor M. N., Roberts (3. W., Crow T. J., Tatemoto K. and Polak J. M. (1983) Neuropeptide Y distribution in human brain. Nature 306, 584-586. 2. Allen Y. S., Bloom S. R. and Polak J. M. (1986) The neuropeptide Y-immunoreactive neuronal system: discovery, anatomy and involvement in neurodegenerative disease. Human NeurobioL 5, 227-234. 3. Baker K. G., Halliday G. M., Hornung J. P., Geffen L. B., Cotton R. G. H. and T6rk I. (1991) Distribution, morphology and number of monoaminergic-synthesizing, and substance P-containing neurons in the human dorsal raphe nucleus. Neuroscience 42, 757 775. 4. Blessing W. W., Howe P. R. C., Joh T. H., Oliver J. R. and Willoughby J. O. (1986) Distribution of tyrosine hydroxylase and neuropeptide Y-like immunoreactive neurons in rabbit medulla oblongata, with attention to colocalization studies, presumptive adrenaline-synthesizing perikarya, and vagal preganglionic cells. J. comp. Neurol. 248, 285-300. 5. Bouras C., Magistretti P. J. and Morrison J. H. (1986) An immunohistochemical study of six biologically active peptides in the human brain. Human Neurobiol. 5, 213-226. 6. Braak E., Braak H. and Weindl A. (1985) Somatostatin-like immunoreactivity in non-pyramidal neurons of the human isocortex. Anat. Embryol. 173, 237-246. 7. Cajal S. R. (1911) Histologie du Systkme Nerveux de l'Homme et des Vertdbrbs, Vol. 2. Maloine, Paris.

374

J.P. HORNUNG et al.

8. Campbell M. J., Lewis D. A., Benoit R. and Morrison J. H. (1987) Regional heterogeneity in the distribution of somatostatin-28- and somatostatin-28 1-12-immunoreactive profiles in monkey neocortex. J. Neurosci. 7, 1133-I 144. 9. Chan-Palay V. (1987) Somatostatin immunoreactive neurons in the human hippocampus and cortex shown by immunogold/silver intensification on vibratome sections: coexistence with neuropeptide Y neurons, and effects in Alzheimer-type dementia. J. comp. Neurol. 260, 201-223. 10. Chan-Palay V., Allen Y. S., Land W., Haesler U. and Polak J. M. (1985) Cytology and distribution in normal human cerebral cortex of neurons immunoreactive with antisera against neuropeptide Y. J. camp. Neurol. 238, 382 389. 11. Chan-Palay V., Lang W., Allen Y. S. Haesler U. and Polak J. M. (1985) Cortical neurons immunoreactive with antisera against neuropeptide Y are altered in Alzheimer's-type dementia. J. camp. Neurol. 238, 390-400 (1985). 12. Chan-Palay V. and Yasargil G. (1986) Immunocytochemistry of human brain tissue with a polyclonal antiserum against neuropeptide Y. Anat. Embryol. 174, 27-33 (1986). 13. Chronwall B. M., Chase T. N. and O'Donohue T. L. (1984) Coexistence of neuropeptide Y and somatostatin in rat and human cortical and rat hypothalamic neurons. Neurosci. Lett. 52, 213-217. 14. Chun J. J. M. and Shatz C. J. (1989) The earliest-generated neurons of the cat cerebral cortex: characterization by MAP2 and neurotransmitter immunohistochemistry during fetal life. J. Neurosci. 9, 1648 1667. 15. Cuello A. C., Galfr6 G. and Milstein C. (1979) Detection of substance P in the central nervous system by a monoclonal antibody. Proc. natn. Acad. Sci. U.S.A. 7~ 353~3536. 16. Del Faccio M., Levanti M. C., Dessi M. L. and Zucca G. (1987) The human hippocampal formation and parahippocampal gyrus: localization of substance P-like immunoreactivity in newborn and adult post-mortem tissue. Neuroscience 21, 141-150. 17. Demeulenmeester H., Vandesande F., Orban G. A., Brandon C. and Vanderhaeghen J. J. (1988) Heterogeneity of GABAergic cells in cat visual cortex. J. Neurosci. 8, 988 1000. 18. Friauf E., McConnell S. K. and Shatz C. J. (1990) Functional synaptic circuits in the subplate during fetal and early postnatal development of cat visual cortex. J. Neurosci. 10, 2601 2613. 19. Gabbott P. L. A. and Somogyi P. (1986) Quantitative distribution of GABA-immunoreactive neurons in the visual cortex (area 17) of the cat. Expl Brain Res. 61, 323-331. 20. Gaspar P., Berger B., Febvert A., Vigny A., and Henry J. P. (1989) Catecholamine innervation of the human cerebral cortex as revealed by comparative immunohistochemistry of tyrosine hydroxylase and dopamine-beta-hydroxylase. J. comp Neurol. 279, 249-278. 21. Gaspar P., Duyckaerts C., Febvert A., Benoit R., Beck B. and Berger B. (1989) Subpopulations of somatostatin 28-immunoreactive neurons display different vulnerability in senile dementia of the Alzheimer type. Brain Res. 490, 1-13. 22. Gilbert C. D. (1983) Microcircuitry of the visual cortex. A. Rev. Neurosci. 6, 217-247. 23. Hendry S. H. C. and Jones E. G. (1983) Synaptic organization of immunocytochemically identified GABA neurons in the monkey sensory-motor cortex. J. Neurocytol. 12, 639-660. 24. Hendry S. H. C., Jones E. G., DeFelipe J., Schmechel D., Brandon C. and Emson P. C. (1984) Neuropeptide-containing neurons of the cerebral cortex are also GABAergic. Proc. natn. Acad. Sci. U.S.A. 81, 6526-6530. 25. Hendry S. H. C., Jones E. G., Emson P. C. Lawson D. E. M., Heizmann C. W. and Streit P. (1989) Two classes of cortical GABA neurons defined by differential calcium binding protein immunoreactivities. Expl Brain Res. 76, 467-472. 26. H6kfelt T., Ljungdahl A., Steinbusch H., Verhofstad A., Nilsson G., Brodin E., Pernow B. and Goldstein M. (1978) Immunohistochemical evidence of substance P-like immunoreactivity in some 5-hydroxytryptamine-containing neurons in the rat central nervous system. Neuroscience 3, 517 538. 27. Hornung J. P., Assal F. and Innocenti G. M. (1989) Distribution of diffuse afferents and interneurons in experimentally induced microcortex in cat visual cortex. Eur. J. Neurosci. suppl. 2, 105. 28. Hornung J. P. and De Tribolet N. (1990) Quantitative analysis of the density and laminar distribution of GABA-immunoreactive neurons in the human cerebral cortex. Eur. J. Neurosci. suppl. 3, 21. 29. Hsu S. M., Raine I. and Fanger H. (1981) Use of avidin biotin peroxidase complex (ABC) in immunoperoxidase technique: a comparison between ABC and unlabeled antibody (PAP) procedure. J. Histochem. Cytochem. 29, 577-580. 30. Iritani S., Fujii M. and Satoh K. (1989) The distribution of substance P in the cerebral cortex and hippocampal formation: an immunohistochemical study in the monkey and rat Brain Res. Bull. 22, 295-303. 31. Jones E. G. and Hendry S. H. C. (1986) Co-localization of GABA and neuropeptides in neocortical neurons. Trends Neurosci. 9, 71-76. 32. Jones E. G., DeFelipe J., Hendry S. H. C. and Maggio J. E. (1988) A study of tachykinin-immunoreactive neurons in monkey cerebral cortex. J. Neurosci. 8, 1206-1224. 33. Kuljis R. O. and Rakic P. (1988) Distribution of neuropeptide Y-containing perikarya and axons in various neocortical areas in the macaque monkey. J. comp. Neurol. 280, 383-392. 34. Lewis D. A., Campbell M. J. and Morrison J. H. (1986) An immunohistochemical characterization of somatostatin-28 and somatostatin-28j J2 in monkey prefrontal cortex. J. comp. Neurol. 248, 1-18. 35. Mai J. K., Stephens P. H., HopfA. and Cuello A. C. (1986) Substance P in the human brain. Neuroscience 17, 709-739. 36. Mann D. M. A. and Yates P. O. (1986) Neurotransmitter deficits in Alzheimer's disease and in other dementing disorders. Human Neurobiol. 5, 147-158. 37. Martin K. A. C. (1984) Neuronal circuitry in cat striate cortex. In Cerebral Cortex (eds Peters A. and Jones E. G.), Vol. 2, pp. 241-284. Plenum Press, New York. 38. Meyer G. (1987) Forms and spatial arrangement of neurons in the primary motor cortex of man. J. comp. Neurol. 262, 402-428. 39. Meyer G. and Wahle P. (1988) Early postnatal development of cholecystokinin-immunoreactive structures in the visual cortex of the cat. J. comp. Neurol. 276, 360-386. 40. Moore R. Y. and Gustafson E. L. (1989) The distribution of dopamine-fl-hydroxylase, neuropeptide Y and galanin in locus coeruleus neurons. J. chem. Neuroanat. 2, 95-106. 41. Morrison J. H., Scherr S., Lewis D. A., Campbell M. J. and Bloom F. E. (1986) The laminar and regional distribution of neocortical somatostatin and neuritic plaques: implications for Alzheimer's disease as a global neocortical disconnection syndrome. In The Biological Substrates of,4lzheimer's Disease (ed. Scheibel A.), pp. 115-131. Academic Press, New York.

Peptidergic human cortical neurons

375

42. Naegele J. R. and Barnstable C. J. (1989) Molecular determinants of GABAergic local-circuit neurons in the visual cortex. Trends Neurosci. 12, 28-34. 43. Nakamura S. and Vincent S. R. (1986) Somatostatin- and neuropeptide Y-iummunoreactive neurons in the neocortex of senile dementia of Alzheimer's type. Brain Res. 370, 11-20. 44. Ong. W. Y. and Garey L. J. (1991) Distribution of GABA and neuropeptides in the human cerebral cortex. Anat. Embryol. 183, 397-413. 45. Peters A. and Jones E. G. (1986) Classification of cortical neurons. In Cerebral Cortex (eds Peters A. and Jones E. G.), . pp. 107 121. Plenum Press, New York. 46. Sakamoto N., Takatsuji K., Shiosaka S., Tateishi K., Hashimura E., Miura S., Hamaoka T. and Tohyama M. (1984) Cholecystokinin-8-1ike immunoreactivity in the pre- and post-central gyri of the human cerebral cortex. Brain Res. 307, 77-83. 47. Sakamoto N., Takatsuji K., Shiosaka S. and Tohyama M. (1985) Evidence for the existence of substance P-like immunoreactive neurons in the human cerebral cortex: an immunohistochemical analysis. Brain Res. 325, 32~324. 48. Sakanaka M., Shiosaka S., Takatsuki K. and Tohyama M. (1983) Evidence for the existence of a substance P-containing pathway from the nucleus laterodorsalis tegmenti (Castaldi) to the medial frontal cortex. Brain Res. 259, 123-126. 49. Schiffman S., Campistron G., Tugendhaft P., Brotchi J., Flament-Durand J., Geffard M. and Vanderhaeghen J. J. (1988) Immunocytochemical detection of GABAergic nerve cells in the human temporal cortex using a direct 7-aminobutyric acid antiserum. Brain Res. 442, 270-278. 50. Sillito A. M. (1975) The contribution of inhibitory mechanisms to the receptive field properties of neurones in the striate cortex of the cat. J. Physiol., Lond. 250, 305-329. 51. Somogyi P., Hodgson A. J., Smith A. D., Nunzi M. G., Gorio A. and Wu J. Y. (9184) Different populations of GABAergic neurons in the visual cortex and hippocampus of cat contain somatostatin- or cholecystokinin-immunoreactive material. J. Neurosci. 4, 2590-2603. 52. Taquet H., Javoy-Agid F., Mauborugne A., Benoliel J. J., Agid Y., Legrand J. C., Tramu G., Cesselin F. and Hamon M. (1988) Biochemical mapping of cholecystokinin-, substance P-, (met)enkephalin-, (leu)enkephalin- and dynorphin A(1-8)-like immunoreactivities in the human cerebral cortex. Neuroscience 27, 871-883. 53. T6rk I. and Hornung J. P. (1987) Serotonergic innervation of the human cerebral cortex. Neuroscience suppl. 22, S112. 54. van Luijtelaar M. G. P. A., Steinbusch H. W. M. and Tonnaer J. A. D. M. (1988) Aberrant morphology of serotonergic fibers in the forebrain of the aged rat. Neurosci. Lett. 95, 93-96. 55. Vincent S. R., Johansson O., H6kfelt T., Meyerson B., Sachs C., Elde R. P., Terenius L. and Kimmel J. (1982) Neuropeptide coexistence in human cortical neurones. Nature 298, 65~57. 56. Vincent S. R., Skirboll L., H6kfelt T., Johansson O., Lundberg J. M. Elde R. P., Terenius L. and Kimmel J. (1982) Coexistence of somatostatin- and avian pancreatic polypeptide (APP)-like immunoreactivity in some forebrain neurons. Neuroscience 7, 439-446. (Accepted 24 June 1992)