Nonphosphorylated Neurofilament Protein and Calbindin Immunoreactivity in Layer III Pyramidal Neurons of Human Neocortex
Tamara L. Hayes1 and David A. Lewis1-2
Subpopulations of pyramidal neurons in the neocortex have been shown to contain nonphosphorylated neurofilament protein (NPNFP) and calbindin DUK (Morrison et al., 1987; Campbell and Morrison, 1989; Hof et al., 1990; Kobayashi et al., 1990; Hof and Morrison, 1991; Mesulam and Geula, 1991). However, it is not known what relations, if any, exist between the pyramidal neurons containing each of these proteins. In this study, the expression of NPNFP and calbindin immunoreactivity was compared in six regions of human neocortex. Characteristic laminar patterns of immunoreactivity for each protein were seen in most regions examined, and both NPNFP- and calbindin-labeled pyramidal neurons were found in layer III. However, the pyramidal neurons labeled with NPNFP and calbindin differed in several respects. First the sublaminar distribution of NPNFPlabeled pyramids within layer III differed across regions, ranging from an even distribution throughout the layer in a visual association region (area 18) to a predominance of labeled neurons in the deep half of that layer in a higher association region (area 20). The distribution of calbindin-immunoreactive pyramidal neurons also varied regionally, but in a different manner than that of the NPNFP-labeled neurons. Second, in every region examined, the average size of NPNFP-labeled layer III pyramids was greater than that of calbindin-immunoreactive pyramids. However, there was substantial regional heterogeneity in the extent to which the she distributions of neurons in each of the two populations overlapped. Third, in the regions in which NPNFP- and calbindin-immunoreactive neurons were most similar in size, the amount of colocalization (as identified by double-labeling studies) was also greatest Similarly, in the regions in which there was minimal overlap in the size of the NPNFP- and the calbindin-immunoreactive neurons, there was minimal colocalization. These regional characteristics of NPNFP- and calbindin-immunoreactive layer III pyramidal neurons have implications for the involvement of these neuronal populations in Alzheimer's disease.
Neurons of the cerebral cortex have long been considered to be divisible into two main cell types: pyramidal and nonpyramidal neurons. Pyramidal neurons, the more numerous of these two cell groups, have several distinct morphological features (Feldman, 1984), including a characteristically shaped cell body, dendritic spines, an axon that enters the white matter, and their most uniquely identifying feature, the presence of an apical dendrite. In addition to being morphologically distinct, pyramidal neurons are the major excitatory class of cortical neurons, furnishing projections to both cortical and subcortical regions Despite these common identifying features, pyramidal neurons are by no means a homogeneous population, but rather differ in their size, laminar location, the extent of their dendritic arbor, and the sites of their axon terminations. For example, tract-tracing studies in monkeys have indicated that layer III pyramidal neurons typically furnish corticocortical and callosal projections, whereas layer V pyramids furnish predominantly subcortical projections (see Jones, 1984, for review). In addition, the axon terminals of callosally projecting pyramids may be interdigitated with those of corticocortically projecting neurons (Goldman-Rakic and Schwartz, 1982; Schwartz and Goldman-Rakic, 1984). Morphological features are not the only identifying characteristics of neurons. Immunohistochemical studies have indicated that neurons that have a similar morphology may differ in their chemical phenotypes. The identification of subclasses according to their respective distribution of neurochemicals may prove to be a useful method of differentiating subpopulations of pyramidal neurons, as it has been in the case of nonpyramidal neurons. For example, corticotropinreleasing factor and parvalbumin appear to identify separate populations of a particular morphologically defined subclass of nonpyramidal neurons known as chandelier neurons (DeFelipeetal., 1989a; Lewis and Lund, 1990). In recent years, immunohistochemical studies of several markers of pyramidal neurons have been undertaken. In particular, subpopulations of pyramidal neurons have been shown to contain nonphosphorylated neurofilament protein (NPNFP) (Morrison et al., 1987; Campbell and Morrison, 1989; Hof and Morrison, 1990; Hof et al., 1990) and calbindin (DeFelipe et al., 1989b; Freund et al., 1990; Hof and Morrison,
Departments of ' Behavioral Neuroscience and Psychiatry, University of Pittsburgh, Pittsburgh, Pennsylvania 15213
2
Downloaded from http://cercor.oxfordjournals.org/ at Cornell University Library on November 27, 2012
Cerebral Cortex Jan/Feb 1992,2 56-67; 1047-3211/92/14.00
Tabla 1 Postmortem human brain specimens
Case
(yi)
Se»
HU187
32 19 36 46 22 38
M M M M M F
HU242 HU244 HU245 HU246 HU247
Conical area
PMI
Age
Cause of death
7.5 5.2 4.5 73 3.5 5.4
17
18
20
41
42
Myocarditis Trauma Trauma Trauma Trauma Pidmonary embolism
Materials and Methods
Imnttmobistocbemistry Brain specimens were obtained at autopsy from five males and one female with no known psychiatric or neurologic disorders (Table 1). The mean (±SE) age at death was 32.2 (±4.2) years and the mean postmortem interval (PMI) was 5.6 (±0.6) hr. Tissue blocks 5-7 mm thick were taken from the left hemisphere and placed in cold 4% paraformaldehyde in phosphate buffer (pH 7.4) for 48 hr. Tissue blocks were then washed in a series of cold, graded (12%, 16%, 18%) solutions of sucrose in phosphate-buffered saline and sectioned coronally in a cryostat at 40 nm. Sections processed for immunohistochemistry were first treated with a 15 min incubation in 1% aqueous hydrogen peroxide to remove endogenous peroxidase. Calbindin immunoreactivity was assessed using a rabbit polyclonal antiserum directed against bovine calbindin D M (generously donated by Dr. K. L. Baimbridge, University of British Columbia). Sections processed for calbindin immunoreactivity were incubated for 40 hr at 4°C in phosphate-buffered saline (PBS) containing 0.5% Triton X-100, 2% bovine serum albumin (BSA), 20% normal goat serum (NGS), and a 1:3000 dilution of primary antiserum. These sections were then processed using the extended peroxidaseantiperoxidase (PAP) technique (Sternberger, 1986), which consisted of successive incubations in solu-
tions of Trisbuffered saline (pH 7.4) containing 0.2% Triton X-100, 2% BSA, 20% NGS, 10% normal human serum, and a 1-50 dilution of either goat anti-rabbit IgG (secondary antibody) or rabbit PAP. The incubations consisted of 90 min in secondary, 120 min in PAP, 60 min in fresh secondary, and a final 90 min in the same PAP solution. The procedure was completed with an incubation in 0.05% diaminobenzidine (DAB) and 0.015% aqueous hydrogen peroxide. NPNFP immunoreactivity was examined using SMI-32 (Sternberger-Meyer Immunocytochemicals, Jarretsville, MD), a mouse monoclonal antibody that recognizes a nonphosphorylated epitope on the medium and heavy chains of neurofilament protein in humans (Lee et al., 1988). Sections processed for NPNFP immunoreactivity were incubated overnight at 4°C in PBS containing 0.5% Triton X-100, 2% BSA, 20% NGS, and a 1:10,000 dilution of SMI-32. These sections were then processed by the avidin-biotin method (Hsu et al., 1981), using a Vectastain kit (Vector Laboratories, Burlingame, CA), with a final incubation in 0.05% diaminobenzidine and 0.003% hydrogen peroxide. After processing for either NPNFP or calbindin, immunoreactivity on slide-mounted sections was intensified by serial immersions in 0.005% osmium tetroxide, 0.5% thiocarbohydrizide, and 0.005% osmium tetroxide. Double labeling of sections was done by two methods. In the first method, a streptavidin-conjugated fluorescent label (Texas red, Amersham) was used to visualize the calbindinimmunoreactive structures, and a fluorescein isothiocyanate (FITC) conjugated anti-mouse antiserum (Chemicon) was used to visualize the SMI-32-labeled structures. In the second method, the avidin-biotin method was used to visualize calbindin, and the FITC-conjugated anti-mouse antiserum was again used to visualize SMI-32-labeled structures. In this case, double-labeled neurons had FITC labeled proximal and apical dendrites with DABpositive soma. With both methods, sections were incubated for 40 hr at 4°C in PBS containing 0.5% Triton X-100, 2% BSA, 20% NGS, and both a 1:3000 dilution of anti-calbindin antiserum and a 1:5000 dilution of SMI-32. Sections were then incubated in a solution of PBS containing 0.5% Triton X-100 and 1:200 biotinylated goat anti-rabbit IgG for 1 hr. The third incubation contained 1:200 FITC-conjugated anti-mouse antiserum in PBS with 0.5% Triton X-100, and either 1:200 Texas red or the avidin-biotin complex of the Vectastain kit. Sections for which the ABC method
Cerebral Cortex Jan/Feb 1992, V 2 N 1 57
Downloaded from http://cercor.oxfordjournals.org/ at Cornell University Library on November 27, 2012
1991). However, direct comparisons of the relative distributions of these markers have not been made, and it is not known if they identify unique populations of pyramidal neurons. This information may be particularly important in investigations of the involvement of pyramidal neurons in certain disorders such as Alzheimer's disease. For example, both NPNFPand calbindinimmunoreactive pyramidal neurons have been reported to undergo degeneration in this disorder (McLachlan et al , 1987; Morrison et al., 1987; Hof and Morrison, 1990, 1991; Hof et al., 1990), but it is not apparent whether this vulnerability is associated with two separate populations of pyramidal neurons or is due to an overlap of these two chemical markers. In order to begin to address these types of questions, we used immunohistochemical techniques in this study to compare the expression of NPNFP and calbindin immunoreactivity in layer III pyramidal neurons of human neocortex.
Data Analysis Quantitative studies were performed in three cortical areas (18,20, and 42) of three brains (HU242, HU244, and HU246). Laminar Distribution of Neurons A camera lucida drawing tube was used to plot the sublaminar distributions of all NPNFP- and calbindinimmunoreactive pyramidal neurons in layer III in a 1500-nm-wide cortical traverse. In all cases, pyramidal neurons were distinguished from nonpyramidal neurons by the presence of a characteristically shaped cell body and an apical dendrite. The distance of each pyramidal neuron from the pial surface and its location relative to the layer I I/I II and 111/IV borders were determined, and these data were used to construct histograms of the distributions of both populations of neurons in layer III. These data were also used to determine the percentage of labeled neurons in deep layer III as follows. The plots of layer III were divided into two equal sublaminae, and the number of labeled neurons in each half was determined. The percentage of labeled neurons in deep layer III was calculated as the number of labeled neurons in the deep half of layer III divided by the total number of immunoreactive neurons in layer III. A comparison of these percentages across regions and markers was made using a x2 statistic. Neuronal Areas Measurements of the cross-sectional area of NPNFPand calbindin-immunoreactive cell bodies throughout the full depth of layer III were made using an image analysis system (Southern Micro). Cells were viewed using a 63* objective, and this image was projected through a camera onto a video screen. The outline of the perikarya was then traced, and a computerized calculation of the outlined area was made. Drawings were made of 10 cells per section, four sections per region, three regions per brain. Fields chosen for measurement were cut perpendicular to the pial surface and were not on a sulcal or gyral B8 Layer III Pyramidal Neurons • Hayes and Lewis
curvature. These fields were the same as those in which laminar distribution was determined A two-way analysis of variance was done across brains and regions to determine differences in the average size of neurons labeled for calbindin versus those labeled for NPNFP, and post hoc group comparisons were performed using the Scheffe method (a = 0.05). Double-labeled Neurons Quantification of double-labeled neurons was done using a Zeiss Axioplan microscope equipped with fluorescent optics and separate filter packs for FITC (excitation, band pass 485/20 nm; barrier, band pass 520-560 nm) and Texas red (excitation, band pass 546/12 nm; barrier, long pass 590 nm). Neurons immunoreactive for NPNFP and calbindin were identified and plotted using a low-light camera coupled to the image analysis system. The percentage of neurons double-labeled was then calculated from these plots. Quantification of double labeling that used FITC and DAB was not done because the high intensity of illumination used to visualize the DAB made use of the low-light camera unfeasible. Results Six brains were examined to determine the laminar and regional distributions of NPNFP and calbindin. Although the patterns of immunoreactivity were extremely consistent across all six brains, there were berween-brain differences in the intensity of immunoreactivity. Thus, the three brains with the best immunoreactivity were used in subsequent quantitative analyses Distribution of NPNFP- and Calbindin-immunoreactive Structures The laminar distribution of NPNFP followed a characteristic pattern in most of the cortical regions examined (Fig. 1). Most notably, intensely as well as lightly labeled pyramidal neurons formed distinct bands in both layers III and V, and labeled neurons were also present in layer VI. A plexus of NPNFPimmunoreactive dendrites was seen throughout layers III—VI. Areas 4 and 17 differed from this characteristic pattern in their distribution of NPNFP-immunoreactive structures. In area 4, layer V contained intensely labeled Betz cells in addition to smaller immunoreactive pyramids. In area 17, labeled neurons were present predominantly in layers III, IVb, and VI, and there was a striking absence of immunoreactivity in layer IVc. These findings are consistent with those previously reported by Campbell and Morrison (1989). The laminar distribution of calbindin-immunoreactive structures (Fig. 2), while also consistent across most of the cortical regions examined, was different from that of NPNFP. A distinctive feature of all the regions examined was the presence of numerous, intensely labeled nonpyramidal neurons at the layer 11/ III border, with the predominance of these neurons located in layer III Fewer, more lightly labeled nonpyramidal neurons were present in the deeper cor-
Downloaded from http://cercor.oxfordjournals.org/ at Cornell University Library on November 27, 2012
was used were further incubated in a solution of 0.05% diaminobenzidine and 0.003% hydrogen peroxide. Controls for false positives were done as described previously (Noack and Lewis, 1989). Regions were identified in adjacent Nissl-stained sections according to published cytoarchitectonic criteria (von Economo, 1929) and were named according to Brodmann. The following regions were examined: primary visual cortex (area 17), the adjacent visual association cortex (area 18), higher-order visual association cortex of the inferior temporal gyrus (area 20) immediately rostral to the level of the lateral geniculate nucleus, primary auditory region of Heschl's gyrus (area 41), the adjacent auditory association cortex (area 42), and primary motor cortex (area 4). Thus, the areas examined included both primary and association sensory cortices from more than one sensory modality. Areas 17, 18, and 20 were not available for two of the cases (Table 1).
»'.
*
Downloaded from http://cercor.oxfordjournals.org/ at Cornell University Library on November 27, 2012
R g o r t 1 . Bright-fieM photomicrograph of NPNFP irrmunoreaetiynY m area 42 of HU246. Note the density of m i n i area: w e pyramidal neurons in both layer III and layer V, and the labeled dcndmes present throughout layers III—VI. Scale bar. 200 m -
Cerebral Cortex Jan/Feb 1992, V 2 N 1 69
Downloaded from http://cercor.oxfordjournals.org/ at Cornell University Library on November 27, 2012
WM Figure 2. Biiflhi-fietd phmomtcrograpri of caQrindin immunoreactivnY in area 42 of HU246. Layer I o Cited wnh labeled puma, and there a a pleas of fibers m layers II and superficial III. Distinct radiafly oriented bundles of fibers (fascides) are present m layers Ill-V. Note the abundance of rtonpyramidal neurons in superficial layer III and the numerous pyramidal resorts in deep layer III. Scale bar. 200 fim. Layer III Pyramidal Neurons • Hayes and Lewis
Table 2 Percentage ol labeled layer III pyramidal neurons found BI the deep half of that layer n each canted sea
NPNFP Calbtndn
Area IB
Area 20
Area 42
56.9 ± 2.5 73.1 ± 2.3
76.9 ± 4.2 66.4 ± 1 1
82.5 ± 11 54.8 ± 3.8
Values are means ± Sf of percentages in three brains
Comparison of NPNFP- and Calbindin-containing Layer HI Pyramidal Neurons The distribution of NPNFP- and calbindin-immunoreactive pyramidal neurons in layer III was neither homogeneous nor consistent across regions. Table 2 indicates the percentage of layer III NPNFP- and calbindin-immunoreactive pyramidal neurons that were located in the deep half of that layer in each region. These markers showed significantly different trends in their relative distribution in layer III across regions (x2 = 7.247; P < 0.05). For example, a greater proportion of layer III NPNFP-immunoreactive pyramids were located in the deep half of that layer in area 42 (82.5%) than in area 18 (56.9%). In contrast, a smaller proportion of the calbindin-immunoreactive pyramids were in the deep half of layer III in area 42 (54.8%) than in area 18 (73.1%). However, within each region, there were no differences across brains in the sublaminar distribution of labeled pyramidal neurons (Fig. 3). One of the most striking differences between the NPNFP-immunoreactive and calbindin-immunoreactive pyramids of layer III was the size of the labeled
HU242
HU244
HU246
NPNFP Area 20 Area 18 Area 42
307.2 ± 12.9 352.1 ± 28.0 339.2 ± 261
283.1 ± 4 5 336.0 ± 24.1 3571 ± 38.4
304.3 ± 13.6 252.5 ± 21.7 373.7 ± 28.5
Calbindin Area 20 Area 18 Area 42
208.7 ± 7 7 119.9 ± 3.9 1819 ± 5.6
226.5 ± 7.5 108.4 ± 2.5 172.2 ± 4.4
223.1 ± 9.7 104.7 ± 11 159.4 ± 7.5
neurons. In every region and brain examined, the NPNFP-labeled neurons in deep layer III appeared to be considerably larger than most calbindin-immunoreactive neurons. Measurements of the crosssectional area of NPNFP- and calbindin-immunoreactive neurons from three cortical regions (areas 18, 20, and 42) confirmed these qualitative impressions. Across these three regions, the mean (±SE) crosssectional area of NPNFP-immunoreactive pyramidal neurons (327.6 ± 8.8 Mm2) was significantly larger (F = 4.82; P < 0.01) than that of calbindin-positive pyramidal neurons (166.5 ± 3.4 Mm2). This difference in neuron size was present in every region for each of three cases examined (Table 3). However, the magnitude of the size difference between the NPNFP- and calbindin-immunoreactive layer III pyramidal neurons was region specific. For example, NPNFP-immunoreactive layer III pyramidal neurons differed significantly (F= 4.82; P < 0.01) in size across the three regions examined. Labeled neurons in area 42 (356.7 ± 15.5 Mm2) were significantly larger (Scheffe, P < 0.05) than those in area 20 (298.2 ± 7.3 Mm2), whereas the NPNFP-immunoreactive neurons of area 18 (313.5 ± 19.8 Mm2) were of intermediate size. In contrast, the cross-sectional area of calbindin-immunoreactive layer III pyramids also differed significantly {F = 158.13; P < 0.0001) across regions, but with a different pattern. The mean size of labeled neurons in area 20 (219-4 ±4.8 Mm2) was significantly greater than that in area 42 (171.2 ± 5.0 Mm2), which was in turn significantly greater than that in area 18 (111.00 ± 1 . 8 Mm2). Within each region, the size of neurons labeled with each antibody did not differ significantly across brains, with one exception: the mean size of NPNFP-immunoreactive neurons in area 18 of HU246 (252.5 ± 21.7 Mm2) was significantly smaller (Scheffe, P < 0.05) than that in HU242 (352.1 ± 28.0 Mm2). These regional differences in NPNFP- and calbindin-immunoreactive neuron size were reflected in the amount of overlap in the sizes of the pyramidal neurons labeled with each marker (Fig. 4). The greatest overlap occurred in area 20, where 28% of NPNFPimmunoreactive neurons were larger than the largest calbindin-immunoreactive neurons of that area, and only 3% of calbindin-labeled neurons were smaller than any NPNFP-labeled neuron. In area 42, 43% of Cerebral Cortex Jaji/Feb 1992, V 2 N 1 61
Downloaded from http://cercor.oxfordjournals.org/ at Cornell University Library on November 27, 2012
tical layers. In most areas, calbindin immunoreactivity was also present in pyramidal neurons distributed throughout layer III. However, compared to the other regions examined, primary visual cortex (area 17) contained relatively fewer labeled pyramidal neurons in layer III, and primary motor cortex contained fewer still All regions also contained punctate immunoreactivity in layer I and a plexus of labeled fibers in layers II and superficial III. In most regions examined, fascicles of radially oriented fibers were present in layers III-V; these fascicles were not seen in area 4 or 17. Area 17 also contained a unique pattern of immunoreactivity in layer IV: layer IVa contained irregular patches of punctate immunoreactivity, layer IVb contained only a few immunoreactive neurons and fibers, layer IVca contained numerous labeled nonpyramidal neurons, and layer IVc/S contained only a few labeled fibers and no immunoreactive neurons. In every cortical area examined, the majority of the immunoreactive structures identified by these two markers were clearly different. However, both NPNFPand calbindin-immunoreactive pyramidal neurons were found in layer III. In order to determine the extent of overlap of the subpopulations of layer III pyramidal neurons immunoreactive for NPNFP or calbindin, we compared their distribution within layer III and their relative size in those cortical regions that contained both groups.
Table 3 Mean | ± S £ | cross-sectional area [finf] of NPNFP- and calbtndi neirrons in three regcns of three brains
Calbindin
NPNFP
600" 800i
HU242
10001
1
1200" 1400 "
1
1600 " 1800-
:X J
~-
ce from pial surface
10
20
15
20
400 " 600"
F?
800-
J
HU244 i
1000
I
i
1200 "
J '
y 1400"
10
£
15
20
600
15
20
"—T "
600-
tn
^ 800i ex g 1000-
800-
HU246
1000-
I -
^
1200
1200 -
1400-
1400 -
o
-
10
15
20
percentage of labeled neurons
I
10
15
20
percentage of labeled neurons
R j u r e 3. Histograms of the distributnns of NPNFP- and caflxndin4abeled pyramalal nairons in tayer ID of area 20. TIB i-txe indicates the percentage of the total number of neurons mapped that fall in each ton. The bins {y-wd$ are 75 iim wide, and pial surface is toward the top. Values given are n moons The dotted fines indicate the location of the layer 11/111 border (upperfew)and the lays (II/TV border [bwsr foe). Note that the neurons lie predorranantfy in the deeper pan of the layer and that ths NPNFP- and cafbmdin-immunoreactive narors have simitar distributions across brains.
the NPNFP-labeled neurons were larger than any calbindin-immunoreactive neuron in that region, and over half (53%) of the calbindin-immunoreactive neurons were smaller than any NPNFP-labeled neuron measured in that region. Area 18 had the least amount of overlap: 95% of NPNFP-immunoreactive neurons were larger than any calbindin-labeled neurons measured in that region, and 92% of calbindin62 Layer III Pyramidal Neurons • Hayes and Lewis
immunoreactive neurons were smaller than all NPNFPlabeled neurons. ColocaUzation of NPNFP and Calbindin The overlap in the size and distribution of NPNFPand calbindin-immunoreactive pyramidal neurons suggested that they might be colocalized in some neurons. In order to assess this, sections from areas
Downloaded from http://cercor.oxfordjournals.org/ at Cornell University Library on November 27, 2012
15
area 18. In addition, in quantitative studies (HU244; see Fig. 5C,D), 30% of NPNFPimmunoreactive neurons and 42% of calbindin-immunoreactive neurons in area 20 were doubled labeled, whereas only 1% of NPNFPimmunoreactive neurons and 1% of calbindin-immunoreactive neurons in area 18 were double labeled.
1000
-
HU242
•g o 600
•a
400
200
Area 20
Area 42
Area 18
Area 42
A r e a 18
Area 42
1000
HU244 890
ST
600
400
200
Area 20 ^
1000
S § •|
HU246 890
J6C 600
•a
400
I 200
Area 20
Hgnra 4 . Box plots (Rosre, 1886) indicating the distributors of the cross-sectional areas ol NrW-imrnunoreactrve (open boxes) and caJbindin-imunoreanlve (shaded boxes) pyramidal neurons m three regions of three brains. The center line in each box ndicates the median size. The upper and totw eitem of the boi indicates the see of the IUUIMI falling at the median of the fergest and smallest 50% of the neurons, respectively. The vsrticsl btffs represent the range of values excluding outjyers, vrfich are indicated by the circles outside the b a n Note that there are regnnal differences m the amount of overlap of the NPNFP- and calbindiiwrnrnunoreaaive populaitons, and that these differences are consistent across brains.
18, 20, 41, and 42 were processed for both NPNFP and calbindin immunoreactivity using double-labeling techniques. Both techniques gave similar results in all regions examined. The majority of labeled neurons contained only one marker (Fig. 5, arrowheads). However, some double-labeled neurons were also identified in all regions examined (Fig. 5, arrows). The extent of double labeling was commensurate with that predicted from the overlap in the cross-sectional areas of neurons in the two populations. Qualitative comparisons of the regions indicated that the amount of colocalization was greatest in area 20 and least in
Layer III NPNFP- and Calbtndin-tmmunoreacHve Pyramidal Neurons Although NPNFP and calbindin were both present in layer III pyramidal neurons, the characteristics of the subpopulations of pyramids containing each protein differed in a number of ways. Although both NPNFPand calbindin-immunoreactive pyramidal neurons were always more numerous in the deep than superficial half of layer III, they showed opposite trends across regions in their sublaminar distributions. The proportion of NPNFPimmunoreactive pyramids in deep layer III decreased from area 42 to area 20 to area 18, whereas the percentage of calbindin-positive pyramidal neurons in deep layer III progressively increased across these same regions. In addition to having a different sublaminax distribution than the calbindin-labeled neurons, the NPNFPimmunoreactive pyramidal neurons of layer III were, on average, larger Cerebral Cortex Jan/Feb 1992, V 2 N 1 63
Downloaded from http://cercor.oxfordjournals.org/ at Cornell University Library on November 27, 2012
Area 18
Discussion In this study, we compared the expression of NPNFP and calbindin immunoreactivity in six regions of human neocortex. Although these proteins were present in a number of different structural elements, they were both found in pyramidal neurons. NPNFP-labeled pyramidal neurons were present in layers III, V, and VI of ever)' cortical region examined except primary visual cortex, where they were found in layers III, IVa, and VI. These findings were consistent with those reported previously in human neocortex (Morrison et al., 1987; Campbell and Morrison, 1989; Hof et al., 1990; Mesulam and Geula, 1991). In contrast, calbindin-immunoreactive pyramidal neurons were confined to layer III, and although they were abundant in most of the regions examined, they were present in much lower density in primary visual and especially motor cortices. Calbindin-immunoreactive pyramidal neurons in layer III have also been observed by some investigators in human frontal (Kobayashi et al., 1990; Hof and Morrison, 1991) and temporal (Kobayashi et al., 1990) cortices, but were not seen by others (Ichimiya et al., 1988; Hoffman et al., 1989). Differences in antisera, PMI, and fixation parameters may account for the lack of detectable calbindin immunoreactivity in pyramidal neurons in some studies. It should also be noted that calbindin-positive pyramidal neurons were distinguished from calbindin-labeled nonpyramidal neurons by the presence of a characteristically shaped cell body and an apical dendrite. Although inter-rater reliability was very high using these criteria, it is possible that these characteristics were not obvious for some calbindin-containing pyramidal neurons, and consequently the number of calbindinpositive pyramids may be underreported.
than the calbindinimmunoreactive neurons. The extent to which the distributions of the sizes of the two populations overlapped also varied across regions, ranging from very little overlap of the groups within area 18, to substantial overlap in area 20. These differences in sublaminar distribution and size suggest that NPNFP- and calbindin-labeled layer III pyramidal neurons may play different roles in cortical circuitry. For example, the size of a pyramidal neuron has been correlated with some charaaeristics of its axonal projection. Within the monkey visual system, callosally projecting neurons are larger (on average) than those projecting ipsilaterally, although 89 Layer III Pyramidal Neurons • Hayes and Lewis
layer III pyramidal neurons of all sizes furnish callosal projections (Lund et al., 1981; Van Essen et al., 1982). The calbindinimmunoreactive neurons in area 18 were homogeneous in size and much smaller than any of the NPNFP-immunoreactive neurons. Thus, it may be that the small calbindin-labeled pyramids of area 18 are less likely to furnish callosal projections than the NPNFP-immunoreactive neurons of that region. In addition, the sublaminar distribution of layer III pyramidal neurons may be related to their axonal targets. In a study of connectivity within the visual system, Rockland and Pandya (1979) found that feed-
Downloaded from http://cercor.oxfordjournals.org/ at Cornell University Library on November 27, 2012
Rgoro 6. Ruoretcsni limnmuugitqis of NPNFP- (A C) and caJbirafm- (ft 0) immunoreactive neurons m layer III of the same sections, taken from area 20 of HU247 |A fl) and HU244 [C. 0). Arrows indicate double-labeled neurons: arrowheads indicate smgWabeted nsuroni Scale bar. 100 jim.
Colocaiization of NPNFP and Calbindin Immunoreacttvity Double-labeling experiments showed that NPNFP and calbindin were colocalized in some layer III pyramidal neurons. Regional differences were present in the extent of colocaiization of these proteins, and these differences paralleled the regional patterns in the amount of overlap of the sizes of NPNFP- and calbindin-positive pyramidal neurons. For example, 42% of calbindin-labeled neurons and 30% of NPNFPlabeled neurons in area 20 were double labeled, whereas only 1% of each population in area 18 contained both proteins. The double-labeling experiments provide a lower bound to the extent of colocalization, since both markers are contained in at least those neurons that are immunoreactive for both antibodies. In contrast, the overlap in the cross-sectional areas of the neurons of these two populations provides an upper bound on the percentage of neurons in which these markers can be colocalized, since it is possible that all the neurons of a similar size in the two populations contain both markers. Two caveats apply to these statements. First, the cross-sectional area of neurons measured in unper-
fused tissue is known to be influenced by the PMI. For example, in unperfused monkey cortex, the crosssectional area of neurons immunoreactive for prosomatostatin-derived peptides progressively increased following PMIs of 0.5-12 hr, but did not continue to increase at longer PMIs (Hayes et al., 1991). In that study, the animals were euthanized, and therefore the influence of the PMI was not confounded by agonal state effects. In contrast, in human autopsy material, a prolonged agonal state may exaggerate postmortem effects. In the present study, the PMI for each case was under 8 hr and the agonal state of the cases differed. Thus, direct comparisons across brains of the cross-sectional area of pyramidal neurons must be made with caution. However, for the purposes of comparing neuronal size in different populations within the same section, it seems probable that the changes arising as a result of agonal or postmortem influences would affect both populations equally. The second caveat concerns the quantification in the double-labeling study. Although each fluorescent label was specific to either calbindin or NPNFP immunoreactivity, the immunofluorescent techniques appeared to label fewer neurons than either of the DAB-based methods. Thus, not all NPNFP- or calbindin-containing neurons would have been identified in the dual-label experiments, and counts of immunoreactive neurons might therefore underestimate the true numbers. However, because the intensity of immunoreactivity with each antibody did not appear to differ between single- and double-labeled neurons, it seems reasonable to assume that the quantities of double-labeled neurons and single-labeled neurons were reduced proportionately. Implications for Alzheimer's Disease The findings of this study may have interesting implications concerning the role of NPNFP- and calbindin-immunoreactive neurons in the pathology of Alzheimer's disease. It has been reported that cortical levels of calbindin are reduced in dementia brains (McLachlan et al., 1987), which may reflect a loss of calbindin-containing neurons (Ichimiya et al., 1988; Hofand Morrison, 1991), although some authors have failed to find a loss of calbindin in the neocortex in Alzheimer's disease (Iacopino and Christakos, 1990). NPNFP-immunoreactive pyramidal neurons are also lost in Alzheimer's disease brains (Morrison et al., 1987; Hofand Morrison, 1990; Hof et al., 1990), and both NPNFP- and calbindin-containing layer III pyramidal cells may be part of a group of corticocortical projection neurons that degenerate in Alzheimer's disease (Morrison et al., 1987; Hof and Morrison, 1991). However, it is not clear what relationship exists between these proteins as markers of the vulnerable neurons in this disorder. The findings of the current study suggest several possibilities concerning the relative vulnerability of NPNFP- and calbindin-positive pyramidal neurons in Alzheimer's disease. First, expression of both NPNFP and calbindin in layer III pyramidal neurons may be necessary for those neurons to undergo degeneration in Alzheimer's disCerebral Cortex Jan/Feb 1992, V 2 N 1 86
Downloaded from http://cercor.oxfordjournals.org/ at Cornell University Library on November 27, 2012
forward pyramidal neurons were situated primarily in deep layer III, whereas feedback projections arose primarily from the infragranular layers, but also from superficial layer III. Other authors have reported a similar laminar distribution of feedforward and feedback projection neurons in frontal (Arikuni et al., 1988) and parietal and temporal (Neal et al., 1990) cortices. Thus, the more homogeneous distribution of calbindin-labeled neurons in layer III of area 20 compared to the preponderance of NPNFP-immunoreactive neurons in deep layer III of that region may indicate that calbindin-containing neurons are more likely to be involved in feedback projections from this region than are neurons immunoreactive for NPNFP. Finally, combined tracing and immunohistochemical studies in monkeys have shown that some, but not all, corticocortically projeaing neurons are NPNFP immunoreactive. For example, the majority of layer VI Meynert cells in primary visual cortex that furnish a projection to the middle temporal visual region (MT) are NPNFP positive (McGuire and Siegel, 1990), whereas the pyramidal neurons providing the feedback pathway from MT to area 17 do not contain NPNFP (Kupferschmid et al., 1991). Furthermore, Campbell et al. (1991) found that of layer III neurons retrogradely labeled following tracer injections into monkey prefrontal cortex, about 90% of those in both ipsilateral and contralateral superior temporal cortex were NPNFP immunoreactive, whereas only a quarter of those in an adjacent prefrontal region were NPNFP labeled. Although these data suggest that the projections of subpopulations of pyramidal neurons may be related to their chemical phenorype, further combined tracing and immunohistochemical studies are needed to determine if the presence of NPNFP or calbindin in layer III pyramidal neurons can be correlated to the projection patterns of those neurons.
A second possibility is that calbindin enhances the vulnerability of NPNFP-positive neurons but that its presence is not necessary. This hypothesis is also consistent with the greater colocalization of NPNFP and calbindin in layer III of area 20 as compared to area 18, and with the greater loss of NPNFP-containing layer III pyramids in area 20 than in area 18 in Alzheimer's disease (Hof and Morrison, 1990; Hof et al., 1990). One would predict in this case that in Alzheimer's disease, the loss of NPNFP-immunoreactive pyramidal neurons in layer III would be greater than that in layer V, since only layer IN pyramids also contain calbindin. However, Hof and Morrison (1990) reported that the losses of NPNFP-positive neurons in layers III and V of area 20 were generally comparable and that only a slightly greater loss of NPNFPimmunoreactive pyramids occurred in layer III than in layer V of area 18. Finally, it may be that although the presence of NPNFP is related to pyramidal cell vulnerability in Alzheimer's disease, the presence of calbindin is not. In this case, the loss of cortical calbindin immunoreactivity in this disorder would be due solely to the degeneration of NPNFP-immunoreaaive neurons that also happen to contain calbindin. Although the present study neither supports nor refutes this view, it may be consistent with the apparently greater loss of NPNFP- than of calbindin-immunoreactive layer III 88 Layer III Pyramidal Neurons • Hayes and Lewis
prefrontal neurons in Alzheimer's disease (Hof et al., 1990; Hof and Morrison, 1991). In addition, this hypothesis predicts that a study of NPNFP and calbindin colocalization in Alzheimer's disease would reveal a comparable loss of neurons containing both NPNFP and calbindin, and of neurons containing only NPNFP. In contrast, loss of neurons containing only calbindin would be significantly smaller. Furthermore, since the degree of colocalization of calbindin and NPNFP is greater in area 20 than in area 18, one would predict greater losses of calbindin-containing neurons in area 20. Direct testing of this hypothesis awaits colocalization studies in Alzheimer's disease brains. Conclusion Clearly, the pyramidal neurons of layer III form a heterogeneous population differing in a number of characteristics, including chemical phenotype. The regional differences in the relative size and distributions of NPNFP- and calbindin-immunoreactive layer III pyramidal neurons, and the regional variability in the colocalization of NPNFP and calbindin, suggest that these subpopulations of pyramidal neurons may have distinct roles in the function of these regions. A better understanding of the distinguishing characteristics of these neurons may provide a foundation for investigations of their normal function and of their involvement in neuropsychiatric disorders. Notes We thank Dr. K. Baimbridge for the generous provision of the antibody against calbindin, and M. Brady for excellent photographic assistance. This work was supported by NIMH Research Scientist Development Award MH00519, NIMH Grant MH45156, and the Alzheimer's Disease Research Center Grant AGO5133. Correspondence should be addressed to David A. Lewis, M.D., Department of Psychiatry, University of Pittsburgh, 3811 O'Hara Street, Biomedical Sciences Tower 1652W, Pittsburgh, PA 15213. Roferances Arikuni T, Watanabe K, Kubota K (1988) Connections of area 8 with area 6 in the brain of the macaque monkey. J Comp Neurol 277:21-40. Campbell MJ, Morrison JH (1989) Monoclonal antibody to neurofilament protein (SMI-32) labels a subpopulation of pyramidal neurons in the human and monkey neocortex. J Comp Neurol 282:191-205. Campbell MJ, Hof PR, Morrison JH (1991) A subpopulation of primate corticocortical neurons is distinguished by somatodendritic distribution of neurofilament protein. Brain Res 539:133-136. DeFelipeJ, HendrySHC, Jones EG (1989a) Visualization of chandelier cell axons by parvalbumin immunoreactivity in monkey cerebral cortex Proc Natl Acad Sci USA 86:2093-2097. DeFelipe J, Hendry SHC, Jones EG (1989b) Synapses of double bouquet cells in monkey cerebral cortex visualized by calbindin Immunoreactivity. Brain Res 503:4954. Feldman ML (1984) Morphology of neocortical pyramidal neurons. In: Cerebral cortex, Vol 1 (Peters E, Jones EG, eds), pp 123-200. New York: Plenum. Freund TF, Buzsaki G, Leon A, Baimbridge KG, Somogyi P (1990) Relationship of neuronal vulnerability and calcium binding protein immunoreactivity in Ischemia Exp Brain Res 83:55-66
Downloaded from http://cercor.oxfordjournals.org/ at Cornell University Library on November 27, 2012
ease. From the results of the double-labeling and cell sizing experiments in the present study, 30-72% of the pyramidal neurons immunoreactive for NPNFP in layer III of area 20 also contained calbindin, whereas in area 18, only 1-5% of the NPNFP-immunoreactive neurons also contained calbindin. Thus, if the presence of both NPNFP and calbindin is required for pyramidal neurons to be vulnerable in Alzheimer's disease, then the potential for loss of NPNFP-immunoreactive neurons would be greater in area 20 than in area 18. This is consistent with the reports (Hof and Morrison, 1990; Hof et al, 1990) that in Alzheimer's disease the number of NPNFP-immunoreactive neurons in layer III of area 20 was significantly smaller than in controls, whereas in area 18 the loss of layer III NPNFP-immunoreactive pyramids was much less and was statistically significant only in deep layer III. However, these investigators also reported that larger NPNFP-positive pyramidal neurons were more susceptible to degeneration than smaller ones, and our data indicate that these neurons do not contain calbindin. Moreover, loss of NPNFP-containing pyramidal neurons in Alzheimer's disease is not restricted to layer III. NPNFP-positive layer Vpyramidal neurons in isocortex and the NPNFP-containing layer II neurons of entorhinal cortex are both lost in this disease (Morrison et al., 1987; Hof and Morrison, 1990; Hof et al., 1990), and neither of these groups of neurons contains calbindin (Hof and Morrison, 1991; present study; M. J. Beall and D. A. Lewis, unpublished observations). Therefore, although both calbindin and NPNFP may be present in cortical neurons vulnerable to the pathology of Alzheimer's disease, the presence of both proteins is clearly not necessary.
and temporal lobes of the monkey. Brain Res 524:119132. Noack HJ, Lewis DA (1989) Antibodies directed against tyrosine hydroxylase differentially recognize noradrenergic axons in monkey neocortex. Brain Res 500:313324. Rockland KS, Pandya DN (1979) Laminar origins and terminations of cortical connections of the occipital lobe in the rhesus monkey. Brain Res 1793-20. Rosner B (1986) Fundamentals of biostatistics. Boston: Prindle, Weber and Schmidt. Schwartz ML, Goldman-Rakic PS (1984) Callosal and intrahemispheric connectivity of the prefrontal association cortex in rhesus monkey relation between intraparietal and principal sulcal cortex. J Comp Neurol 226:403-420. Sternberger LA (1986) Immunocytochemistry, 3d ed. New York: Churchill Livingstone Van Essen DC, Newsome WT, BixbyJL (1982) The pattern of interhemispheric connections and its relationship to extrastriate visual areas in the macaque monkey. J Neurosci 2:265-283. von Economo C (1929) The cytoarchitectonics of the human cerebral cortex. London: Oxford UP.
Cerebral Cortex Jan/Feb 1992, V 2 N 1 67
Downloaded from http://cercor.oxfordjournals.org/ at Cornell University Library on November 27, 2012
Goldman-Rakic PS, Schwartz ML (1982) Interdigitation of contralateral and ipsilateral columnar projections to frontal association cortex in primates. Science 216:755-757 Hayes TL, Cameron JL, Fernstrom J, Lewis DA (1991) A comparative analysis of the distribution of somatostatin immunoreactivity in human and monkey neocortex. J Comp Neurol 303:584-599. Hof PR, Morrison JH (1990) Quantitative analysis of a vulnerable subset of pyramidal neurons in Alzheimer's disease: II. Primary and secondary visual cortex. J Comp Neurol 301:55-64. Hof PR, Morrison JH (1991) Neocortical neuronal subpopulations labeled by a monoclonal antibody to calbindin exhibit differential vulnerability in Alzheimer's disease Exp Neurol 111:293-301. Hof PR, Cox K, Morrison JH (1990) Quantitative analysis of a vulnerable subset of pyramidal neurons in Alzheimer's disease. I. Superior frontal and inferior temporal cortex. J Comp Neurol 301:44-54. Hoffman SR, Quigley BJ Jr, Harrington KM, Kowall NW, McKee AC (1989) Calbindin D28K immunoreactive neurons in temporal isocortex resist degeneration in Alzheimer's disease. J Neuropathol Exp Neurol 48:331. Hsu SM, Raine L, Fanger H (1981) Use of avidin-biotinperoxidase complex (ABC) in lmmunoperoxldase techniques: a comparison between ABC and unlabeled antibody (PAP) procedures. J Histochem Cytochem 4:577580. Iacopino AM, Christakos S (1990) Specific reduction of calcium-binding protein (28-kilodalton calbindin-D) gene expression in aging and neurodegenerative diseases. Proc Natl Acad Sci USA 87:4078-4082. Ichimiya Y, Emson PC, Mountjoy CQ, Lawson DEM, Heizmann CW (1988) Loss of calbindin-28k immunoreactive neurones from the cortex in Alzheimer-type dementia. Brain Res 475:156-159. Jones EG (1984) Laminar distribution of cortical efferent cells In: Cerebral cortex, Vol 1 (Peters E, Jones EG, eds), pp 521-553. New York: Plenum. Kobayashi K, Emson PC, Mountjoy CQ, Thornton SN, Lawson DEM, Mann DMA (1990) Cerebral cortical calbindin DUK and parvalbumin neurones in Down's syndrome. Neurosci Lett 113:17-22. Kupferschmid SB, Hof PR, Morrison JH (1991) Corticocortical connections in macaque visual cortex exhibit differential patterns of neurofilament protein distribution. Soc Neurosci Abstr 17:845. Lee VM-Y, Otvos L Jr, Carden MJ, Hollisi M, Dietzschold B, Lazzarinl RA (1988) Identification of the major multiphosphorylation site in mammalian neurofilaments Proc Natl Acad Sci USA 85:1998-2002. Lewis DA, Lund JS (1990) Heterogeneity of chandelier neurons in monkey neocortex- corticotropin-releasing factor and parvalbumin immunoreactive populations. J Comp Neurol 293:599-615. Lund JS, Hendrickson AE, Orgren MP, Tobin EA (1981) Anatomical organization of primate visual cortex area VII. J Comp Neurol 202.19-45. McGuire BA, Siegel RM (1990) Projection patternsof Meynert cells in primate visual cortex identified by immunoreactivity for neurofilaments. Soc Neurosci Abstr 16:5 McLachlan DRC, Wong L, Bergeron C, Baimbridge KG (1987) Calmodulin and calbindin DMK in Alzheimer disease. Alzheimer Dis Assoc Disord 1:171-179. Mesulam M, Geula C (1991) Differential distribution of a neurofilament protein epitope in acetylcholinesterasench neurons of human cerebral neocortex. Brain Res 544: 169-173. Morrison JH, Lewis DA, Campbell MJ, Benson DL, Bouras C (1987) A monoclonal antibody to non-phosphorylated neurofilament protein marks the vulnerable cortical neurons in Alzheimer's disease Brain Res 416:331-336 Neal JW, Pearson RC, Powell TP (1990) The ipsilateral cortico-cortical connections of area 7b, PF, in the parietal
Tamara L. Hayes1 and David A. Lewis1-2
Subpopulations of pyramidal neurons in the neocortex have been shown to contain nonphosphorylated neurofilament protein (NPNFP) and calbindin DUK (Morrison et al., 1987; Campbell and Morrison, 1989; Hof et al., 1990; Kobayashi et al., 1990; Hof and Morrison, 1991; Mesulam and Geula, 1991). However, it is not known what relations, if any, exist between the pyramidal neurons containing each of these proteins. In this study, the expression of NPNFP and calbindin immunoreactivity was compared in six regions of human neocortex. Characteristic laminar patterns of immunoreactivity for each protein were seen in most regions examined, and both NPNFP- and calbindin-labeled pyramidal neurons were found in layer III. However, the pyramidal neurons labeled with NPNFP and calbindin differed in several respects. First the sublaminar distribution of NPNFPlabeled pyramids within layer III differed across regions, ranging from an even distribution throughout the layer in a visual association region (area 18) to a predominance of labeled neurons in the deep half of that layer in a higher association region (area 20). The distribution of calbindin-immunoreactive pyramidal neurons also varied regionally, but in a different manner than that of the NPNFP-labeled neurons. Second, in every region examined, the average size of NPNFP-labeled layer III pyramids was greater than that of calbindin-immunoreactive pyramids. However, there was substantial regional heterogeneity in the extent to which the she distributions of neurons in each of the two populations overlapped. Third, in the regions in which NPNFP- and calbindin-immunoreactive neurons were most similar in size, the amount of colocalization (as identified by double-labeling studies) was also greatest Similarly, in the regions in which there was minimal overlap in the size of the NPNFP- and the calbindin-immunoreactive neurons, there was minimal colocalization. These regional characteristics of NPNFP- and calbindin-immunoreactive layer III pyramidal neurons have implications for the involvement of these neuronal populations in Alzheimer's disease.
Neurons of the cerebral cortex have long been considered to be divisible into two main cell types: pyramidal and nonpyramidal neurons. Pyramidal neurons, the more numerous of these two cell groups, have several distinct morphological features (Feldman, 1984), including a characteristically shaped cell body, dendritic spines, an axon that enters the white matter, and their most uniquely identifying feature, the presence of an apical dendrite. In addition to being morphologically distinct, pyramidal neurons are the major excitatory class of cortical neurons, furnishing projections to both cortical and subcortical regions Despite these common identifying features, pyramidal neurons are by no means a homogeneous population, but rather differ in their size, laminar location, the extent of their dendritic arbor, and the sites of their axon terminations. For example, tract-tracing studies in monkeys have indicated that layer III pyramidal neurons typically furnish corticocortical and callosal projections, whereas layer V pyramids furnish predominantly subcortical projections (see Jones, 1984, for review). In addition, the axon terminals of callosally projecting pyramids may be interdigitated with those of corticocortically projecting neurons (Goldman-Rakic and Schwartz, 1982; Schwartz and Goldman-Rakic, 1984). Morphological features are not the only identifying characteristics of neurons. Immunohistochemical studies have indicated that neurons that have a similar morphology may differ in their chemical phenotypes. The identification of subclasses according to their respective distribution of neurochemicals may prove to be a useful method of differentiating subpopulations of pyramidal neurons, as it has been in the case of nonpyramidal neurons. For example, corticotropinreleasing factor and parvalbumin appear to identify separate populations of a particular morphologically defined subclass of nonpyramidal neurons known as chandelier neurons (DeFelipeetal., 1989a; Lewis and Lund, 1990). In recent years, immunohistochemical studies of several markers of pyramidal neurons have been undertaken. In particular, subpopulations of pyramidal neurons have been shown to contain nonphosphorylated neurofilament protein (NPNFP) (Morrison et al., 1987; Campbell and Morrison, 1989; Hof and Morrison, 1990; Hof et al., 1990) and calbindin (DeFelipe et al., 1989b; Freund et al., 1990; Hof and Morrison,
Departments of ' Behavioral Neuroscience and Psychiatry, University of Pittsburgh, Pittsburgh, Pennsylvania 15213
2
Downloaded from http://cercor.oxfordjournals.org/ at Cornell University Library on November 27, 2012
Cerebral Cortex Jan/Feb 1992,2 56-67; 1047-3211/92/14.00
Tabla 1 Postmortem human brain specimens
Case
(yi)
Se»
HU187
32 19 36 46 22 38
M M M M M F
HU242 HU244 HU245 HU246 HU247
Conical area
PMI
Age
Cause of death
7.5 5.2 4.5 73 3.5 5.4
17
18
20
41
42
Myocarditis Trauma Trauma Trauma Trauma Pidmonary embolism
Materials and Methods
Imnttmobistocbemistry Brain specimens were obtained at autopsy from five males and one female with no known psychiatric or neurologic disorders (Table 1). The mean (±SE) age at death was 32.2 (±4.2) years and the mean postmortem interval (PMI) was 5.6 (±0.6) hr. Tissue blocks 5-7 mm thick were taken from the left hemisphere and placed in cold 4% paraformaldehyde in phosphate buffer (pH 7.4) for 48 hr. Tissue blocks were then washed in a series of cold, graded (12%, 16%, 18%) solutions of sucrose in phosphate-buffered saline and sectioned coronally in a cryostat at 40 nm. Sections processed for immunohistochemistry were first treated with a 15 min incubation in 1% aqueous hydrogen peroxide to remove endogenous peroxidase. Calbindin immunoreactivity was assessed using a rabbit polyclonal antiserum directed against bovine calbindin D M (generously donated by Dr. K. L. Baimbridge, University of British Columbia). Sections processed for calbindin immunoreactivity were incubated for 40 hr at 4°C in phosphate-buffered saline (PBS) containing 0.5% Triton X-100, 2% bovine serum albumin (BSA), 20% normal goat serum (NGS), and a 1:3000 dilution of primary antiserum. These sections were then processed using the extended peroxidaseantiperoxidase (PAP) technique (Sternberger, 1986), which consisted of successive incubations in solu-
tions of Trisbuffered saline (pH 7.4) containing 0.2% Triton X-100, 2% BSA, 20% NGS, 10% normal human serum, and a 1-50 dilution of either goat anti-rabbit IgG (secondary antibody) or rabbit PAP. The incubations consisted of 90 min in secondary, 120 min in PAP, 60 min in fresh secondary, and a final 90 min in the same PAP solution. The procedure was completed with an incubation in 0.05% diaminobenzidine (DAB) and 0.015% aqueous hydrogen peroxide. NPNFP immunoreactivity was examined using SMI-32 (Sternberger-Meyer Immunocytochemicals, Jarretsville, MD), a mouse monoclonal antibody that recognizes a nonphosphorylated epitope on the medium and heavy chains of neurofilament protein in humans (Lee et al., 1988). Sections processed for NPNFP immunoreactivity were incubated overnight at 4°C in PBS containing 0.5% Triton X-100, 2% BSA, 20% NGS, and a 1:10,000 dilution of SMI-32. These sections were then processed by the avidin-biotin method (Hsu et al., 1981), using a Vectastain kit (Vector Laboratories, Burlingame, CA), with a final incubation in 0.05% diaminobenzidine and 0.003% hydrogen peroxide. After processing for either NPNFP or calbindin, immunoreactivity on slide-mounted sections was intensified by serial immersions in 0.005% osmium tetroxide, 0.5% thiocarbohydrizide, and 0.005% osmium tetroxide. Double labeling of sections was done by two methods. In the first method, a streptavidin-conjugated fluorescent label (Texas red, Amersham) was used to visualize the calbindinimmunoreactive structures, and a fluorescein isothiocyanate (FITC) conjugated anti-mouse antiserum (Chemicon) was used to visualize the SMI-32-labeled structures. In the second method, the avidin-biotin method was used to visualize calbindin, and the FITC-conjugated anti-mouse antiserum was again used to visualize SMI-32-labeled structures. In this case, double-labeled neurons had FITC labeled proximal and apical dendrites with DABpositive soma. With both methods, sections were incubated for 40 hr at 4°C in PBS containing 0.5% Triton X-100, 2% BSA, 20% NGS, and both a 1:3000 dilution of anti-calbindin antiserum and a 1:5000 dilution of SMI-32. Sections were then incubated in a solution of PBS containing 0.5% Triton X-100 and 1:200 biotinylated goat anti-rabbit IgG for 1 hr. The third incubation contained 1:200 FITC-conjugated anti-mouse antiserum in PBS with 0.5% Triton X-100, and either 1:200 Texas red or the avidin-biotin complex of the Vectastain kit. Sections for which the ABC method
Cerebral Cortex Jan/Feb 1992, V 2 N 1 57
Downloaded from http://cercor.oxfordjournals.org/ at Cornell University Library on November 27, 2012
1991). However, direct comparisons of the relative distributions of these markers have not been made, and it is not known if they identify unique populations of pyramidal neurons. This information may be particularly important in investigations of the involvement of pyramidal neurons in certain disorders such as Alzheimer's disease. For example, both NPNFPand calbindinimmunoreactive pyramidal neurons have been reported to undergo degeneration in this disorder (McLachlan et al , 1987; Morrison et al., 1987; Hof and Morrison, 1990, 1991; Hof et al., 1990), but it is not apparent whether this vulnerability is associated with two separate populations of pyramidal neurons or is due to an overlap of these two chemical markers. In order to begin to address these types of questions, we used immunohistochemical techniques in this study to compare the expression of NPNFP and calbindin immunoreactivity in layer III pyramidal neurons of human neocortex.
Data Analysis Quantitative studies were performed in three cortical areas (18,20, and 42) of three brains (HU242, HU244, and HU246). Laminar Distribution of Neurons A camera lucida drawing tube was used to plot the sublaminar distributions of all NPNFP- and calbindinimmunoreactive pyramidal neurons in layer III in a 1500-nm-wide cortical traverse. In all cases, pyramidal neurons were distinguished from nonpyramidal neurons by the presence of a characteristically shaped cell body and an apical dendrite. The distance of each pyramidal neuron from the pial surface and its location relative to the layer I I/I II and 111/IV borders were determined, and these data were used to construct histograms of the distributions of both populations of neurons in layer III. These data were also used to determine the percentage of labeled neurons in deep layer III as follows. The plots of layer III were divided into two equal sublaminae, and the number of labeled neurons in each half was determined. The percentage of labeled neurons in deep layer III was calculated as the number of labeled neurons in the deep half of layer III divided by the total number of immunoreactive neurons in layer III. A comparison of these percentages across regions and markers was made using a x2 statistic. Neuronal Areas Measurements of the cross-sectional area of NPNFPand calbindin-immunoreactive cell bodies throughout the full depth of layer III were made using an image analysis system (Southern Micro). Cells were viewed using a 63* objective, and this image was projected through a camera onto a video screen. The outline of the perikarya was then traced, and a computerized calculation of the outlined area was made. Drawings were made of 10 cells per section, four sections per region, three regions per brain. Fields chosen for measurement were cut perpendicular to the pial surface and were not on a sulcal or gyral B8 Layer III Pyramidal Neurons • Hayes and Lewis
curvature. These fields were the same as those in which laminar distribution was determined A two-way analysis of variance was done across brains and regions to determine differences in the average size of neurons labeled for calbindin versus those labeled for NPNFP, and post hoc group comparisons were performed using the Scheffe method (a = 0.05). Double-labeled Neurons Quantification of double-labeled neurons was done using a Zeiss Axioplan microscope equipped with fluorescent optics and separate filter packs for FITC (excitation, band pass 485/20 nm; barrier, band pass 520-560 nm) and Texas red (excitation, band pass 546/12 nm; barrier, long pass 590 nm). Neurons immunoreactive for NPNFP and calbindin were identified and plotted using a low-light camera coupled to the image analysis system. The percentage of neurons double-labeled was then calculated from these plots. Quantification of double labeling that used FITC and DAB was not done because the high intensity of illumination used to visualize the DAB made use of the low-light camera unfeasible. Results Six brains were examined to determine the laminar and regional distributions of NPNFP and calbindin. Although the patterns of immunoreactivity were extremely consistent across all six brains, there were berween-brain differences in the intensity of immunoreactivity. Thus, the three brains with the best immunoreactivity were used in subsequent quantitative analyses Distribution of NPNFP- and Calbindin-immunoreactive Structures The laminar distribution of NPNFP followed a characteristic pattern in most of the cortical regions examined (Fig. 1). Most notably, intensely as well as lightly labeled pyramidal neurons formed distinct bands in both layers III and V, and labeled neurons were also present in layer VI. A plexus of NPNFPimmunoreactive dendrites was seen throughout layers III—VI. Areas 4 and 17 differed from this characteristic pattern in their distribution of NPNFP-immunoreactive structures. In area 4, layer V contained intensely labeled Betz cells in addition to smaller immunoreactive pyramids. In area 17, labeled neurons were present predominantly in layers III, IVb, and VI, and there was a striking absence of immunoreactivity in layer IVc. These findings are consistent with those previously reported by Campbell and Morrison (1989). The laminar distribution of calbindin-immunoreactive structures (Fig. 2), while also consistent across most of the cortical regions examined, was different from that of NPNFP. A distinctive feature of all the regions examined was the presence of numerous, intensely labeled nonpyramidal neurons at the layer 11/ III border, with the predominance of these neurons located in layer III Fewer, more lightly labeled nonpyramidal neurons were present in the deeper cor-
Downloaded from http://cercor.oxfordjournals.org/ at Cornell University Library on November 27, 2012
was used were further incubated in a solution of 0.05% diaminobenzidine and 0.003% hydrogen peroxide. Controls for false positives were done as described previously (Noack and Lewis, 1989). Regions were identified in adjacent Nissl-stained sections according to published cytoarchitectonic criteria (von Economo, 1929) and were named according to Brodmann. The following regions were examined: primary visual cortex (area 17), the adjacent visual association cortex (area 18), higher-order visual association cortex of the inferior temporal gyrus (area 20) immediately rostral to the level of the lateral geniculate nucleus, primary auditory region of Heschl's gyrus (area 41), the adjacent auditory association cortex (area 42), and primary motor cortex (area 4). Thus, the areas examined included both primary and association sensory cortices from more than one sensory modality. Areas 17, 18, and 20 were not available for two of the cases (Table 1).
»'.
*
Downloaded from http://cercor.oxfordjournals.org/ at Cornell University Library on November 27, 2012
R g o r t 1 . Bright-fieM photomicrograph of NPNFP irrmunoreaetiynY m area 42 of HU246. Note the density of m i n i area: w e pyramidal neurons in both layer III and layer V, and the labeled dcndmes present throughout layers III—VI. Scale bar. 200 m -
Cerebral Cortex Jan/Feb 1992, V 2 N 1 69
Downloaded from http://cercor.oxfordjournals.org/ at Cornell University Library on November 27, 2012
WM Figure 2. Biiflhi-fietd phmomtcrograpri of caQrindin immunoreactivnY in area 42 of HU246. Layer I o Cited wnh labeled puma, and there a a pleas of fibers m layers II and superficial III. Distinct radiafly oriented bundles of fibers (fascides) are present m layers Ill-V. Note the abundance of rtonpyramidal neurons in superficial layer III and the numerous pyramidal resorts in deep layer III. Scale bar. 200 fim. Layer III Pyramidal Neurons • Hayes and Lewis
Table 2 Percentage ol labeled layer III pyramidal neurons found BI the deep half of that layer n each canted sea
NPNFP Calbtndn
Area IB
Area 20
Area 42
56.9 ± 2.5 73.1 ± 2.3
76.9 ± 4.2 66.4 ± 1 1
82.5 ± 11 54.8 ± 3.8
Values are means ± Sf of percentages in three brains
Comparison of NPNFP- and Calbindin-containing Layer HI Pyramidal Neurons The distribution of NPNFP- and calbindin-immunoreactive pyramidal neurons in layer III was neither homogeneous nor consistent across regions. Table 2 indicates the percentage of layer III NPNFP- and calbindin-immunoreactive pyramidal neurons that were located in the deep half of that layer in each region. These markers showed significantly different trends in their relative distribution in layer III across regions (x2 = 7.247; P < 0.05). For example, a greater proportion of layer III NPNFP-immunoreactive pyramids were located in the deep half of that layer in area 42 (82.5%) than in area 18 (56.9%). In contrast, a smaller proportion of the calbindin-immunoreactive pyramids were in the deep half of layer III in area 42 (54.8%) than in area 18 (73.1%). However, within each region, there were no differences across brains in the sublaminar distribution of labeled pyramidal neurons (Fig. 3). One of the most striking differences between the NPNFP-immunoreactive and calbindin-immunoreactive pyramids of layer III was the size of the labeled
HU242
HU244
HU246
NPNFP Area 20 Area 18 Area 42
307.2 ± 12.9 352.1 ± 28.0 339.2 ± 261
283.1 ± 4 5 336.0 ± 24.1 3571 ± 38.4
304.3 ± 13.6 252.5 ± 21.7 373.7 ± 28.5
Calbindin Area 20 Area 18 Area 42
208.7 ± 7 7 119.9 ± 3.9 1819 ± 5.6
226.5 ± 7.5 108.4 ± 2.5 172.2 ± 4.4
223.1 ± 9.7 104.7 ± 11 159.4 ± 7.5
neurons. In every region and brain examined, the NPNFP-labeled neurons in deep layer III appeared to be considerably larger than most calbindin-immunoreactive neurons. Measurements of the crosssectional area of NPNFP- and calbindin-immunoreactive neurons from three cortical regions (areas 18, 20, and 42) confirmed these qualitative impressions. Across these three regions, the mean (±SE) crosssectional area of NPNFP-immunoreactive pyramidal neurons (327.6 ± 8.8 Mm2) was significantly larger (F = 4.82; P < 0.01) than that of calbindin-positive pyramidal neurons (166.5 ± 3.4 Mm2). This difference in neuron size was present in every region for each of three cases examined (Table 3). However, the magnitude of the size difference between the NPNFP- and calbindin-immunoreactive layer III pyramidal neurons was region specific. For example, NPNFP-immunoreactive layer III pyramidal neurons differed significantly (F= 4.82; P < 0.01) in size across the three regions examined. Labeled neurons in area 42 (356.7 ± 15.5 Mm2) were significantly larger (Scheffe, P < 0.05) than those in area 20 (298.2 ± 7.3 Mm2), whereas the NPNFP-immunoreactive neurons of area 18 (313.5 ± 19.8 Mm2) were of intermediate size. In contrast, the cross-sectional area of calbindin-immunoreactive layer III pyramids also differed significantly {F = 158.13; P < 0.0001) across regions, but with a different pattern. The mean size of labeled neurons in area 20 (219-4 ±4.8 Mm2) was significantly greater than that in area 42 (171.2 ± 5.0 Mm2), which was in turn significantly greater than that in area 18 (111.00 ± 1 . 8 Mm2). Within each region, the size of neurons labeled with each antibody did not differ significantly across brains, with one exception: the mean size of NPNFP-immunoreactive neurons in area 18 of HU246 (252.5 ± 21.7 Mm2) was significantly smaller (Scheffe, P < 0.05) than that in HU242 (352.1 ± 28.0 Mm2). These regional differences in NPNFP- and calbindin-immunoreactive neuron size were reflected in the amount of overlap in the sizes of the pyramidal neurons labeled with each marker (Fig. 4). The greatest overlap occurred in area 20, where 28% of NPNFPimmunoreactive neurons were larger than the largest calbindin-immunoreactive neurons of that area, and only 3% of calbindin-labeled neurons were smaller than any NPNFP-labeled neuron. In area 42, 43% of Cerebral Cortex Jaji/Feb 1992, V 2 N 1 61
Downloaded from http://cercor.oxfordjournals.org/ at Cornell University Library on November 27, 2012
tical layers. In most areas, calbindin immunoreactivity was also present in pyramidal neurons distributed throughout layer III. However, compared to the other regions examined, primary visual cortex (area 17) contained relatively fewer labeled pyramidal neurons in layer III, and primary motor cortex contained fewer still All regions also contained punctate immunoreactivity in layer I and a plexus of labeled fibers in layers II and superficial III. In most regions examined, fascicles of radially oriented fibers were present in layers III-V; these fascicles were not seen in area 4 or 17. Area 17 also contained a unique pattern of immunoreactivity in layer IV: layer IVa contained irregular patches of punctate immunoreactivity, layer IVb contained only a few immunoreactive neurons and fibers, layer IVca contained numerous labeled nonpyramidal neurons, and layer IVc/S contained only a few labeled fibers and no immunoreactive neurons. In every cortical area examined, the majority of the immunoreactive structures identified by these two markers were clearly different. However, both NPNFPand calbindin-immunoreactive pyramidal neurons were found in layer III. In order to determine the extent of overlap of the subpopulations of layer III pyramidal neurons immunoreactive for NPNFP or calbindin, we compared their distribution within layer III and their relative size in those cortical regions that contained both groups.
Table 3 Mean | ± S £ | cross-sectional area [finf] of NPNFP- and calbtndi neirrons in three regcns of three brains
Calbindin
NPNFP
600" 800i
HU242
10001
1
1200" 1400 "
1
1600 " 1800-
:X J
~-
ce from pial surface
10
20
15
20
400 " 600"
F?
800-
J
HU244 i
1000
I
i
1200 "
J '
y 1400"
10
£
15
20
600
15
20
"—T "
600-
tn
^ 800i ex g 1000-
800-
HU246
1000-
I -
^
1200
1200 -
1400-
1400 -
o
-
10
15
20
percentage of labeled neurons
I
10
15
20
percentage of labeled neurons
R j u r e 3. Histograms of the distributnns of NPNFP- and caflxndin4abeled pyramalal nairons in tayer ID of area 20. TIB i-txe indicates the percentage of the total number of neurons mapped that fall in each ton. The bins {y-wd$ are 75 iim wide, and pial surface is toward the top. Values given are n moons The dotted fines indicate the location of the layer 11/111 border (upperfew)and the lays (II/TV border [bwsr foe). Note that the neurons lie predorranantfy in the deeper pan of the layer and that ths NPNFP- and cafbmdin-immunoreactive narors have simitar distributions across brains.
the NPNFP-labeled neurons were larger than any calbindin-immunoreactive neuron in that region, and over half (53%) of the calbindin-immunoreactive neurons were smaller than any NPNFP-labeled neuron measured in that region. Area 18 had the least amount of overlap: 95% of NPNFP-immunoreactive neurons were larger than any calbindin-labeled neurons measured in that region, and 92% of calbindin62 Layer III Pyramidal Neurons • Hayes and Lewis
immunoreactive neurons were smaller than all NPNFPlabeled neurons. ColocaUzation of NPNFP and Calbindin The overlap in the size and distribution of NPNFPand calbindin-immunoreactive pyramidal neurons suggested that they might be colocalized in some neurons. In order to assess this, sections from areas
Downloaded from http://cercor.oxfordjournals.org/ at Cornell University Library on November 27, 2012
15
area 18. In addition, in quantitative studies (HU244; see Fig. 5C,D), 30% of NPNFPimmunoreactive neurons and 42% of calbindin-immunoreactive neurons in area 20 were doubled labeled, whereas only 1% of NPNFPimmunoreactive neurons and 1% of calbindin-immunoreactive neurons in area 18 were double labeled.
1000
-
HU242
•g o 600
•a
400
200
Area 20
Area 42
Area 18
Area 42
A r e a 18
Area 42
1000
HU244 890
ST
600
400
200
Area 20 ^
1000
S § •|
HU246 890
J6C 600
•a
400
I 200
Area 20
Hgnra 4 . Box plots (Rosre, 1886) indicating the distributors of the cross-sectional areas ol NrW-imrnunoreactrve (open boxes) and caJbindin-imunoreanlve (shaded boxes) pyramidal neurons m three regions of three brains. The center line in each box ndicates the median size. The upper and totw eitem of the boi indicates the see of the IUUIMI falling at the median of the fergest and smallest 50% of the neurons, respectively. The vsrticsl btffs represent the range of values excluding outjyers, vrfich are indicated by the circles outside the b a n Note that there are regnnal differences m the amount of overlap of the NPNFP- and calbindiiwrnrnunoreaaive populaitons, and that these differences are consistent across brains.
18, 20, 41, and 42 were processed for both NPNFP and calbindin immunoreactivity using double-labeling techniques. Both techniques gave similar results in all regions examined. The majority of labeled neurons contained only one marker (Fig. 5, arrowheads). However, some double-labeled neurons were also identified in all regions examined (Fig. 5, arrows). The extent of double labeling was commensurate with that predicted from the overlap in the cross-sectional areas of neurons in the two populations. Qualitative comparisons of the regions indicated that the amount of colocalization was greatest in area 20 and least in
Layer III NPNFP- and Calbtndin-tmmunoreacHve Pyramidal Neurons Although NPNFP and calbindin were both present in layer III pyramidal neurons, the characteristics of the subpopulations of pyramids containing each protein differed in a number of ways. Although both NPNFPand calbindin-immunoreactive pyramidal neurons were always more numerous in the deep than superficial half of layer III, they showed opposite trends across regions in their sublaminar distributions. The proportion of NPNFPimmunoreactive pyramids in deep layer III decreased from area 42 to area 20 to area 18, whereas the percentage of calbindin-positive pyramidal neurons in deep layer III progressively increased across these same regions. In addition to having a different sublaminax distribution than the calbindin-labeled neurons, the NPNFPimmunoreactive pyramidal neurons of layer III were, on average, larger Cerebral Cortex Jan/Feb 1992, V 2 N 1 63
Downloaded from http://cercor.oxfordjournals.org/ at Cornell University Library on November 27, 2012
Area 18
Discussion In this study, we compared the expression of NPNFP and calbindin immunoreactivity in six regions of human neocortex. Although these proteins were present in a number of different structural elements, they were both found in pyramidal neurons. NPNFP-labeled pyramidal neurons were present in layers III, V, and VI of ever)' cortical region examined except primary visual cortex, where they were found in layers III, IVa, and VI. These findings were consistent with those reported previously in human neocortex (Morrison et al., 1987; Campbell and Morrison, 1989; Hof et al., 1990; Mesulam and Geula, 1991). In contrast, calbindin-immunoreactive pyramidal neurons were confined to layer III, and although they were abundant in most of the regions examined, they were present in much lower density in primary visual and especially motor cortices. Calbindin-immunoreactive pyramidal neurons in layer III have also been observed by some investigators in human frontal (Kobayashi et al., 1990; Hof and Morrison, 1991) and temporal (Kobayashi et al., 1990) cortices, but were not seen by others (Ichimiya et al., 1988; Hoffman et al., 1989). Differences in antisera, PMI, and fixation parameters may account for the lack of detectable calbindin immunoreactivity in pyramidal neurons in some studies. It should also be noted that calbindin-positive pyramidal neurons were distinguished from calbindin-labeled nonpyramidal neurons by the presence of a characteristically shaped cell body and an apical dendrite. Although inter-rater reliability was very high using these criteria, it is possible that these characteristics were not obvious for some calbindin-containing pyramidal neurons, and consequently the number of calbindinpositive pyramids may be underreported.
than the calbindinimmunoreactive neurons. The extent to which the distributions of the sizes of the two populations overlapped also varied across regions, ranging from very little overlap of the groups within area 18, to substantial overlap in area 20. These differences in sublaminar distribution and size suggest that NPNFP- and calbindin-labeled layer III pyramidal neurons may play different roles in cortical circuitry. For example, the size of a pyramidal neuron has been correlated with some charaaeristics of its axonal projection. Within the monkey visual system, callosally projecting neurons are larger (on average) than those projecting ipsilaterally, although 89 Layer III Pyramidal Neurons • Hayes and Lewis
layer III pyramidal neurons of all sizes furnish callosal projections (Lund et al., 1981; Van Essen et al., 1982). The calbindinimmunoreactive neurons in area 18 were homogeneous in size and much smaller than any of the NPNFP-immunoreactive neurons. Thus, it may be that the small calbindin-labeled pyramids of area 18 are less likely to furnish callosal projections than the NPNFP-immunoreactive neurons of that region. In addition, the sublaminar distribution of layer III pyramidal neurons may be related to their axonal targets. In a study of connectivity within the visual system, Rockland and Pandya (1979) found that feed-
Downloaded from http://cercor.oxfordjournals.org/ at Cornell University Library on November 27, 2012
Rgoro 6. Ruoretcsni limnmuugitqis of NPNFP- (A C) and caJbirafm- (ft 0) immunoreactive neurons m layer III of the same sections, taken from area 20 of HU247 |A fl) and HU244 [C. 0). Arrows indicate double-labeled neurons: arrowheads indicate smgWabeted nsuroni Scale bar. 100 jim.
Colocaiization of NPNFP and Calbindin Immunoreacttvity Double-labeling experiments showed that NPNFP and calbindin were colocalized in some layer III pyramidal neurons. Regional differences were present in the extent of colocaiization of these proteins, and these differences paralleled the regional patterns in the amount of overlap of the sizes of NPNFP- and calbindin-positive pyramidal neurons. For example, 42% of calbindin-labeled neurons and 30% of NPNFPlabeled neurons in area 20 were double labeled, whereas only 1% of each population in area 18 contained both proteins. The double-labeling experiments provide a lower bound to the extent of colocalization, since both markers are contained in at least those neurons that are immunoreactive for both antibodies. In contrast, the overlap in the cross-sectional areas of the neurons of these two populations provides an upper bound on the percentage of neurons in which these markers can be colocalized, since it is possible that all the neurons of a similar size in the two populations contain both markers. Two caveats apply to these statements. First, the cross-sectional area of neurons measured in unper-
fused tissue is known to be influenced by the PMI. For example, in unperfused monkey cortex, the crosssectional area of neurons immunoreactive for prosomatostatin-derived peptides progressively increased following PMIs of 0.5-12 hr, but did not continue to increase at longer PMIs (Hayes et al., 1991). In that study, the animals were euthanized, and therefore the influence of the PMI was not confounded by agonal state effects. In contrast, in human autopsy material, a prolonged agonal state may exaggerate postmortem effects. In the present study, the PMI for each case was under 8 hr and the agonal state of the cases differed. Thus, direct comparisons across brains of the cross-sectional area of pyramidal neurons must be made with caution. However, for the purposes of comparing neuronal size in different populations within the same section, it seems probable that the changes arising as a result of agonal or postmortem influences would affect both populations equally. The second caveat concerns the quantification in the double-labeling study. Although each fluorescent label was specific to either calbindin or NPNFP immunoreactivity, the immunofluorescent techniques appeared to label fewer neurons than either of the DAB-based methods. Thus, not all NPNFP- or calbindin-containing neurons would have been identified in the dual-label experiments, and counts of immunoreactive neurons might therefore underestimate the true numbers. However, because the intensity of immunoreactivity with each antibody did not appear to differ between single- and double-labeled neurons, it seems reasonable to assume that the quantities of double-labeled neurons and single-labeled neurons were reduced proportionately. Implications for Alzheimer's Disease The findings of this study may have interesting implications concerning the role of NPNFP- and calbindin-immunoreactive neurons in the pathology of Alzheimer's disease. It has been reported that cortical levels of calbindin are reduced in dementia brains (McLachlan et al., 1987), which may reflect a loss of calbindin-containing neurons (Ichimiya et al., 1988; Hofand Morrison, 1991), although some authors have failed to find a loss of calbindin in the neocortex in Alzheimer's disease (Iacopino and Christakos, 1990). NPNFP-immunoreactive pyramidal neurons are also lost in Alzheimer's disease brains (Morrison et al., 1987; Hofand Morrison, 1990; Hof et al., 1990), and both NPNFP- and calbindin-containing layer III pyramidal cells may be part of a group of corticocortical projection neurons that degenerate in Alzheimer's disease (Morrison et al., 1987; Hof and Morrison, 1991). However, it is not clear what relationship exists between these proteins as markers of the vulnerable neurons in this disorder. The findings of the current study suggest several possibilities concerning the relative vulnerability of NPNFP- and calbindin-positive pyramidal neurons in Alzheimer's disease. First, expression of both NPNFP and calbindin in layer III pyramidal neurons may be necessary for those neurons to undergo degeneration in Alzheimer's disCerebral Cortex Jan/Feb 1992, V 2 N 1 86
Downloaded from http://cercor.oxfordjournals.org/ at Cornell University Library on November 27, 2012
forward pyramidal neurons were situated primarily in deep layer III, whereas feedback projections arose primarily from the infragranular layers, but also from superficial layer III. Other authors have reported a similar laminar distribution of feedforward and feedback projection neurons in frontal (Arikuni et al., 1988) and parietal and temporal (Neal et al., 1990) cortices. Thus, the more homogeneous distribution of calbindin-labeled neurons in layer III of area 20 compared to the preponderance of NPNFP-immunoreactive neurons in deep layer III of that region may indicate that calbindin-containing neurons are more likely to be involved in feedback projections from this region than are neurons immunoreactive for NPNFP. Finally, combined tracing and immunohistochemical studies in monkeys have shown that some, but not all, corticocortically projeaing neurons are NPNFP immunoreactive. For example, the majority of layer VI Meynert cells in primary visual cortex that furnish a projection to the middle temporal visual region (MT) are NPNFP positive (McGuire and Siegel, 1990), whereas the pyramidal neurons providing the feedback pathway from MT to area 17 do not contain NPNFP (Kupferschmid et al., 1991). Furthermore, Campbell et al. (1991) found that of layer III neurons retrogradely labeled following tracer injections into monkey prefrontal cortex, about 90% of those in both ipsilateral and contralateral superior temporal cortex were NPNFP immunoreactive, whereas only a quarter of those in an adjacent prefrontal region were NPNFP labeled. Although these data suggest that the projections of subpopulations of pyramidal neurons may be related to their chemical phenorype, further combined tracing and immunohistochemical studies are needed to determine if the presence of NPNFP or calbindin in layer III pyramidal neurons can be correlated to the projection patterns of those neurons.
A second possibility is that calbindin enhances the vulnerability of NPNFP-positive neurons but that its presence is not necessary. This hypothesis is also consistent with the greater colocalization of NPNFP and calbindin in layer III of area 20 as compared to area 18, and with the greater loss of NPNFP-containing layer III pyramids in area 20 than in area 18 in Alzheimer's disease (Hof and Morrison, 1990; Hof et al., 1990). One would predict in this case that in Alzheimer's disease, the loss of NPNFP-immunoreactive pyramidal neurons in layer III would be greater than that in layer V, since only layer IN pyramids also contain calbindin. However, Hof and Morrison (1990) reported that the losses of NPNFP-positive neurons in layers III and V of area 20 were generally comparable and that only a slightly greater loss of NPNFPimmunoreactive pyramids occurred in layer III than in layer V of area 18. Finally, it may be that although the presence of NPNFP is related to pyramidal cell vulnerability in Alzheimer's disease, the presence of calbindin is not. In this case, the loss of cortical calbindin immunoreactivity in this disorder would be due solely to the degeneration of NPNFP-immunoreaaive neurons that also happen to contain calbindin. Although the present study neither supports nor refutes this view, it may be consistent with the apparently greater loss of NPNFP- than of calbindin-immunoreactive layer III 88 Layer III Pyramidal Neurons • Hayes and Lewis
prefrontal neurons in Alzheimer's disease (Hof et al., 1990; Hof and Morrison, 1991). In addition, this hypothesis predicts that a study of NPNFP and calbindin colocalization in Alzheimer's disease would reveal a comparable loss of neurons containing both NPNFP and calbindin, and of neurons containing only NPNFP. In contrast, loss of neurons containing only calbindin would be significantly smaller. Furthermore, since the degree of colocalization of calbindin and NPNFP is greater in area 20 than in area 18, one would predict greater losses of calbindin-containing neurons in area 20. Direct testing of this hypothesis awaits colocalization studies in Alzheimer's disease brains. Conclusion Clearly, the pyramidal neurons of layer III form a heterogeneous population differing in a number of characteristics, including chemical phenotype. The regional differences in the relative size and distributions of NPNFP- and calbindin-immunoreactive layer III pyramidal neurons, and the regional variability in the colocalization of NPNFP and calbindin, suggest that these subpopulations of pyramidal neurons may have distinct roles in the function of these regions. A better understanding of the distinguishing characteristics of these neurons may provide a foundation for investigations of their normal function and of their involvement in neuropsychiatric disorders. Notes We thank Dr. K. Baimbridge for the generous provision of the antibody against calbindin, and M. Brady for excellent photographic assistance. This work was supported by NIMH Research Scientist Development Award MH00519, NIMH Grant MH45156, and the Alzheimer's Disease Research Center Grant AGO5133. Correspondence should be addressed to David A. Lewis, M.D., Department of Psychiatry, University of Pittsburgh, 3811 O'Hara Street, Biomedical Sciences Tower 1652W, Pittsburgh, PA 15213. Roferances Arikuni T, Watanabe K, Kubota K (1988) Connections of area 8 with area 6 in the brain of the macaque monkey. J Comp Neurol 277:21-40. Campbell MJ, Morrison JH (1989) Monoclonal antibody to neurofilament protein (SMI-32) labels a subpopulation of pyramidal neurons in the human and monkey neocortex. J Comp Neurol 282:191-205. Campbell MJ, Hof PR, Morrison JH (1991) A subpopulation of primate corticocortical neurons is distinguished by somatodendritic distribution of neurofilament protein. Brain Res 539:133-136. DeFelipeJ, HendrySHC, Jones EG (1989a) Visualization of chandelier cell axons by parvalbumin immunoreactivity in monkey cerebral cortex Proc Natl Acad Sci USA 86:2093-2097. DeFelipe J, Hendry SHC, Jones EG (1989b) Synapses of double bouquet cells in monkey cerebral cortex visualized by calbindin Immunoreactivity. Brain Res 503:4954. Feldman ML (1984) Morphology of neocortical pyramidal neurons. In: Cerebral cortex, Vol 1 (Peters E, Jones EG, eds), pp 123-200. New York: Plenum. Freund TF, Buzsaki G, Leon A, Baimbridge KG, Somogyi P (1990) Relationship of neuronal vulnerability and calcium binding protein immunoreactivity in Ischemia Exp Brain Res 83:55-66
Downloaded from http://cercor.oxfordjournals.org/ at Cornell University Library on November 27, 2012
ease. From the results of the double-labeling and cell sizing experiments in the present study, 30-72% of the pyramidal neurons immunoreactive for NPNFP in layer III of area 20 also contained calbindin, whereas in area 18, only 1-5% of the NPNFP-immunoreactive neurons also contained calbindin. Thus, if the presence of both NPNFP and calbindin is required for pyramidal neurons to be vulnerable in Alzheimer's disease, then the potential for loss of NPNFP-immunoreactive neurons would be greater in area 20 than in area 18. This is consistent with the reports (Hof and Morrison, 1990; Hof et al, 1990) that in Alzheimer's disease the number of NPNFP-immunoreactive neurons in layer III of area 20 was significantly smaller than in controls, whereas in area 18 the loss of layer III NPNFP-immunoreactive pyramids was much less and was statistically significant only in deep layer III. However, these investigators also reported that larger NPNFP-positive pyramidal neurons were more susceptible to degeneration than smaller ones, and our data indicate that these neurons do not contain calbindin. Moreover, loss of NPNFP-containing pyramidal neurons in Alzheimer's disease is not restricted to layer III. NPNFP-positive layer Vpyramidal neurons in isocortex and the NPNFP-containing layer II neurons of entorhinal cortex are both lost in this disease (Morrison et al., 1987; Hof and Morrison, 1990; Hof et al., 1990), and neither of these groups of neurons contains calbindin (Hof and Morrison, 1991; present study; M. J. Beall and D. A. Lewis, unpublished observations). Therefore, although both calbindin and NPNFP may be present in cortical neurons vulnerable to the pathology of Alzheimer's disease, the presence of both proteins is clearly not necessary.
and temporal lobes of the monkey. Brain Res 524:119132. Noack HJ, Lewis DA (1989) Antibodies directed against tyrosine hydroxylase differentially recognize noradrenergic axons in monkey neocortex. Brain Res 500:313324. Rockland KS, Pandya DN (1979) Laminar origins and terminations of cortical connections of the occipital lobe in the rhesus monkey. Brain Res 1793-20. Rosner B (1986) Fundamentals of biostatistics. Boston: Prindle, Weber and Schmidt. Schwartz ML, Goldman-Rakic PS (1984) Callosal and intrahemispheric connectivity of the prefrontal association cortex in rhesus monkey relation between intraparietal and principal sulcal cortex. J Comp Neurol 226:403-420. Sternberger LA (1986) Immunocytochemistry, 3d ed. New York: Churchill Livingstone Van Essen DC, Newsome WT, BixbyJL (1982) The pattern of interhemispheric connections and its relationship to extrastriate visual areas in the macaque monkey. J Neurosci 2:265-283. von Economo C (1929) The cytoarchitectonics of the human cerebral cortex. London: Oxford UP.
Cerebral Cortex Jan/Feb 1992, V 2 N 1 67
Downloaded from http://cercor.oxfordjournals.org/ at Cornell University Library on November 27, 2012
Goldman-Rakic PS, Schwartz ML (1982) Interdigitation of contralateral and ipsilateral columnar projections to frontal association cortex in primates. Science 216:755-757 Hayes TL, Cameron JL, Fernstrom J, Lewis DA (1991) A comparative analysis of the distribution of somatostatin immunoreactivity in human and monkey neocortex. J Comp Neurol 303:584-599. Hof PR, Morrison JH (1990) Quantitative analysis of a vulnerable subset of pyramidal neurons in Alzheimer's disease: II. Primary and secondary visual cortex. J Comp Neurol 301:55-64. Hof PR, Morrison JH (1991) Neocortical neuronal subpopulations labeled by a monoclonal antibody to calbindin exhibit differential vulnerability in Alzheimer's disease Exp Neurol 111:293-301. Hof PR, Cox K, Morrison JH (1990) Quantitative analysis of a vulnerable subset of pyramidal neurons in Alzheimer's disease. I. Superior frontal and inferior temporal cortex. J Comp Neurol 301:44-54. Hoffman SR, Quigley BJ Jr, Harrington KM, Kowall NW, McKee AC (1989) Calbindin D28K immunoreactive neurons in temporal isocortex resist degeneration in Alzheimer's disease. J Neuropathol Exp Neurol 48:331. Hsu SM, Raine L, Fanger H (1981) Use of avidin-biotinperoxidase complex (ABC) in lmmunoperoxldase techniques: a comparison between ABC and unlabeled antibody (PAP) procedures. J Histochem Cytochem 4:577580. Iacopino AM, Christakos S (1990) Specific reduction of calcium-binding protein (28-kilodalton calbindin-D) gene expression in aging and neurodegenerative diseases. Proc Natl Acad Sci USA 87:4078-4082. Ichimiya Y, Emson PC, Mountjoy CQ, Lawson DEM, Heizmann CW (1988) Loss of calbindin-28k immunoreactive neurones from the cortex in Alzheimer-type dementia. Brain Res 475:156-159. Jones EG (1984) Laminar distribution of cortical efferent cells In: Cerebral cortex, Vol 1 (Peters E, Jones EG, eds), pp 521-553. New York: Plenum. Kobayashi K, Emson PC, Mountjoy CQ, Thornton SN, Lawson DEM, Mann DMA (1990) Cerebral cortical calbindin DUK and parvalbumin neurones in Down's syndrome. Neurosci Lett 113:17-22. Kupferschmid SB, Hof PR, Morrison JH (1991) Corticocortical connections in macaque visual cortex exhibit differential patterns of neurofilament protein distribution. Soc Neurosci Abstr 17:845. Lee VM-Y, Otvos L Jr, Carden MJ, Hollisi M, Dietzschold B, Lazzarinl RA (1988) Identification of the major multiphosphorylation site in mammalian neurofilaments Proc Natl Acad Sci USA 85:1998-2002. Lewis DA, Lund JS (1990) Heterogeneity of chandelier neurons in monkey neocortex- corticotropin-releasing factor and parvalbumin immunoreactive populations. J Comp Neurol 293:599-615. Lund JS, Hendrickson AE, Orgren MP, Tobin EA (1981) Anatomical organization of primate visual cortex area VII. J Comp Neurol 202.19-45. McGuire BA, Siegel RM (1990) Projection patternsof Meynert cells in primate visual cortex identified by immunoreactivity for neurofilaments. Soc Neurosci Abstr 16:5 McLachlan DRC, Wong L, Bergeron C, Baimbridge KG (1987) Calmodulin and calbindin DMK in Alzheimer disease. Alzheimer Dis Assoc Disord 1:171-179. Mesulam M, Geula C (1991) Differential distribution of a neurofilament protein epitope in acetylcholinesterasench neurons of human cerebral neocortex. Brain Res 544: 169-173. Morrison JH, Lewis DA, Campbell MJ, Benson DL, Bouras C (1987) A monoclonal antibody to non-phosphorylated neurofilament protein marks the vulnerable cortical neurons in Alzheimer's disease Brain Res 416:331-336 Neal JW, Pearson RC, Powell TP (1990) The ipsilateral cortico-cortical connections of area 7b, PF, in the parietal
