Abstract
This study was undertaken to estimate the total number of cholinergic cells and the percentage of cholinergic cells that contain estrogen receptor-α (ERα) in the rat basal forebrain. Double immunostaining for choline acetyltransferase (ChAT) and ERα was carried out on 50-μm-thick free-floating sections. Because routine mounting method causes considerable flattening of the sections, we embedded immunostained sections in Durcupan, an epoxy resin known to cause virtually no shrinkage. When this procedure was used the section thickness was well preserved, individual cells could be clearly identified, and subcellular localization of ERα immunoreactivity was easy to verify. Cell counting in these sections revealed that the rat basal forebrain contains 26,390 ± 1097 (mean ± SEM) cholinergic neurons. This comprises 9674 ± 504 in the medial septum-vertical diagonal band of Broca, 9403 ± 484 in the horizontal diagonal band of Broca, and 7312 ± 281 in the nucleus basalis. In these nuclei, 60%, 46%, and 14% of the cholinergic neurons were colocalized with ERα, respectively. We believe that our results are an improvement on existing data because of the better distinction of individual neurons that the Durcupan embedding method brings.
The basal forebrain area contains a number of cholinergic cell groups that are involved in a variety of physiological and behavioral processes. There is evidence that the gonadal estrogens can directly influence on these neurons in females (O'Malley et al. 1987; Lapchak et al. 1990; for review see Gibbs and Aggarwal 1998). This effect may not be restricted to females because it has been shown that in the male brain these cells express estrogen receptors as well (Gibbs 1996). Two structurally related estrogen receptors, α and β, have been identified and characterized (for review see Nilsson et al. 2001). A recent study by Shughrue et al. (2000) showed that whereas many cholinergic cells in the basal forebrain contain estrogen receptor-α (ERα), only few cells contain ERβ. However, in these studies, the total numbers of the cholinergic cells containing or not containing ERs were not reported. Therefore, this study was undertaken to obtain these numeric values, which are required for evaluating the effect of estrogens on the survival of the cholinergic neurons in the basal forebrain in future experiments. The importance of these data is highlighted by the fact that one of the most consistent findings in Alzheimer's disease (AD) is the deterioration of the cholinergic system (Kasa et al. 1997). In addition, recent strategies in the development of therapeutic treatments for this devastating disease are based on estrogens' possible effects on cholinergic neurons (Boissiere et al. 1996; Birge 1997).
Choline acetyltransferase (ChAT), the enzyme responsible for the synthesis of acetylcholine, is located in the cytoplasm, and ERα is expressed primarily in the nucleus (see, e.g., Gibbs 1996). Co-localization of these two proteins could easily be demonstrated on double-immunostained sections at the confocal microscopic level. However, confocal microscope systems can be implemented only for direct counting methods, which are laborious when cell numbers are high in the tissue. More efficient cell counting can be performed using stereological approaches, such as optical fractionator method at the light microscopic level (Gundersen 1986; Gundersen et al. 1988; West et al. 1991; West 1993). For light microscopy, free-floating thick brain sections are traditionally mounted after immunostaining on objective slides, dried, dehydrated, and cleared with different solvents and coverslipped. However, this procedure leads to excessive shrinkage and flattening of the section, especially in the z-dimension. If the area of interest has high number of cells or the cells are clustered, this type of flattening causes two notable problems when cell numbers are counted (Figure 1). One is that in very flat sections one cannot clearly identify individual cells, even with an objective having a high numeric aperture and narrow focal depth. The second problem encountered is that different structure components are very close to each other in the z-dimension and may even merge. For example, in sections double-immunostained for ChAT and ERα, one could not be sure whether a certain ERα-immunoreactive (ir) nucleus belongs to the cell that has a ChAT-ir cytoplasm or whether that particular nucleus has been pushed into the proximity of a ChAT-ir cell that does not have an ERα-ir nucleus. Therefore, for this type of investigation it is very important to prepare the material in a way that avoids these artifacts. Therefore, the major objective would be to prevent section shrinkage and flattening.
In this study we tested and concluded that embedding the samples in an epoxy resin, Durcupan, after immunostaining could be used to maintain section volume during mounting. We illustrate this improvement by presenting the numbers of the cholinergic neurons containing or not containing ERα in different basal forebrain nuclei of the rat. This information provides a baseline in understanding the effects of various experimental manipulations aimed at investigating the role of estrogens in forebrain cholinergic dysfunction that is observed in human diseases, most notably in AD.
Materials and Methods
Male 3-month-old Wistar rats (n = 4) were used. These rats were housed in one cage in a controlled environment (constant temperature 22 ± 1C; humidity 50–60%; lights on 0700–1900 hr) and had free access to food and water. The experiments were approved by the committee for the welfare of laboratory animals of the University of Kuopio.
The rats were deeply anesthetized with a mixture (0.4 ml/100 g, IP) of sodium barbiturate (Synopharm, concentration in the mixture 9.7 mg/ml), chloral hydrate (Merck, Dormstadt, Germany; 10 mg/ml), magnesium sulfate (Merck; 21.2 mg/ml), propylene glycol (Merck; 40%), and absolute ethanol (10%). Then they were perfused transcardially, consecutively with saline (3 min) and then with 300 ml of fixative (30 min). The fixative contained 4% paraformaldehyde (P001; TAAB Laboratories Equipment, Aldermaston, Berks, UK), 0.05% glutaraldehyde (G002; TAAB Laboratories Equipment) and 0.26% picric acid (623; Merck) in 0.1 M phosphate buffer, pH 7.4 (PB). The brains were removed from the skull and 50-μm-thick sections were cut on a vibratome (Leica VT 1000 S, Leica Instruments, Wetzlar, Germany) into six series (Figure 2). Sections were stored in a cryoprotectant solution (30% ethylene glycol and 30% glycerol in PB) at −20C until processed.
To test Durcupan's ability to prevent shrinkage and flattening of the sections, different series of sections were processed as follows: (a) no staining; (b) double immunostaining followed by mounting with Depex; and (c) double immunostaining followed by embedding in Durcupan.
The first series of sections (series a, randomly selected) was washed in PB three times for 10 min. These sections were mounted on slides and immediately covered with a coverslip (to avoid drying). Native section thickness was measured using the set-up described below.
Two series of sections (series b and c) were double-immunostained for ERα and ChAT. Sections were first washed six times for 30 min in PB and then in 0.05 M Tris-buffered saline, pH 7.4 (TBS), twice for 20 min. After this, sections were incubated in 10% normal goat serum (NGS; CS-0922, Colorado Serum Company, Denver, CO) containing 0.5% Triton X-100 in TBS for 40 min and in 1% NGS containing 0.5% Triton X-100 in TBS for 15 min. Incubation in rabbit anti-ERα antiserum (1:10, 000, SC-542; Santa Cruz Technology, Santa Cruz, CA) was carried out at 4C for 48 hr. This was followed by incubation in biotinylated anti-rabbit IgG (1:300, BA-1, 000; Vector Laboratories, Burlingame, CA) overnight at 4C and then in avidin-biotinylated horseradish peroxidase complex (1:500, PK-4, 000; Vector Laboratories) for 3 hr at room temperature (RT). The immunoperoxidase reaction was developed using ammonium nickel sulfate (0.2%, 10029; BDH, Poole, UK)-intensified 3,3′-diaminobenzidine (0.015% DAB, D-5637; Sigma, St Louis, MO) as chromogen, giving a blue-to-black granular reaction end product. Then sections were incubated in rat anti-ChAT antiserum (1:10, 770990; Roche, Basel, Switzerland) 48 hours at 4C, followed by an incubation in rabbit anti-rat IgG (1:50, AB-136; Chemicon, Temecula, CA) for 6 hr at RT and finally in rat peroxidase anti-peroxidase complex (1:300, PAP-20; Chemicon) overnight at 4C. The second immunoperoxidase reaction was developed using 0.05% DAB as a chromogen, which gives a homogeneous brown end product. The dilutions of the antisera and all the washing steps (three times for 30 min) that were carried out between the antibody incubations were done in 0.05 M TBS, pH 7.4, containing 1% NGS and 0.5% Triton X-100.
Schematic diagram showing a side view of a section illustrating the problems arising when sections flatten during histological processing. Different cells and cell compartments (stained nuclei in black, cytoplasm in gray) are readily distinguished in thickness-preserved section
Random systematic sampling strategy and the equation used for estimation of the total number of neurons in different cholinergic nuclei.
After thorough washing in TBS, the series of sections aimed for routine mounting (series b) were mounted on gelatin-coated slides and dried overnight at 37C. Thereafter, these sections were dehydrated in absolute ethanol, cleared in xylene, and covered with Depex and a coverslip. The third series of sections (series c) was washed in TBS, rinsed with distilled water, dehydrated in a series of ethanol (50%, 70%, 90%, 96% for 5 min in each and in absolute ethanol twice for 5 min) and propylene oxide twice for 5 min. The sections were then immersed in Durcupan (AMC; Fluka, Buchs, Switzerland). The sections were freely floating during the dehydration procedure and immersion in Durcupan. After 3 hr at RT in Durcupan, the sections were transferred onto slides and covered with a coverslip. To ensure that the sections were planar between the coverslip and the objective slide and that excess Durcupan was removed, a glass block (weight ∼50 g) was placed on the coverslip to slightly press the coverslip. Durcupan was polymerized at 60C for 24 hr.
Controls for Immunostaining
Control stainings for IHC were carried out by omission of primary antibodies. In addition, omission of only one primary antibody in double immunostaining was performed. Both of these procedures resulted in no staining in those structural components that were found to be positive when the corresponding primary antibody was included. The specificity of both ChAT and ERα antibodies has been previously characterized showing ability to react with these proteins in rat brain (Eckenstein and Thoenen 1982; Pavao and Traish 2001).
Section Thickness Measurement
Section thickness measurements were accomplished with the aid of Stereo Investigator software (MicroBrightField; Colchester, VT). The integrated hardware-software set-up consisted of a PC computer system connected to an ECLIPSE E600 microscope (Nikon; Tokyo, Japan) via 3-chip CCD color video camera (HV-C20; Hitachi, Tokyo, Japan). A motorized stage (three-axis computer-controlled stepping motor system with a 0.1-μm resolution) with a microcator (Heidenhain EXE 610C) attachment (providing a 0.1 μm resolution in the z-axis) was mounted on the microscope. The objective used for the measurements (and later for cell counting) was a CFI Plan Fluor X100 oil immersion objective having the properties NA 1.30 and WD 0.20 mm. For measuring the section thickness, the top of the section was first brought into focus and the depth position (z-axis coordinate) was set to 0. Thereafter the bottom of the section was brought into focus and the z-axis value (i.e., section thickness) was registered. Measurements were made from all sections sampled from each animal used in this study. From each section at least three different sites were measured.
Estimation of Cell Numbers
The number of ChAT-ir cells in the medial septum, vertical and horizontal diagonal band of Broca, and in the nucleus basalis was estimated using the optical fractionator method (Gundersen 1986; West 1993). The analysis was accomplished with the aid of Stereo Investigator software. The fractionator sampling consisted of a section sampling fraction (ssf), an area sampling fraction (asf), and a height sampling fraction (hsf) (Figure 2). For immunostaining, one series out of six series of samples was randomly selected using a random number table. Thus, each starting point had an equal possibility to be sampled and the interval between each section in an individual series was constant. In this random systematic sampling strategy, ssf was 1/6. The nuclear boundaries were first outlined using a CFI Plan Achro X4 objective. Thereafter, the CFI Plan Fluor X100 oil immersion objective was used for the counting. The computer randomly overlaid an 80 μm × 80 μm (A) grid over the drawn contour. This area represents the area associated with each x,y movement and the steps between each counting frame, which had the size of 35 μm × 35 μm (a). The area sampling fraction, asf, is calculated as a/A. The mean thickness of the mounted sections was 40.8 μm (t). The plane for the top of the counting frame in z-dimension was set to 5 μm from the surface of the section (guard zone). Thereafter, the neurons were counted according to the optical disector counting rules (Gundersen 1986) while focusing 30 μm (h, disector height) through the section (Figure 2). The third fraction, hsf, is h/t. The equation
was used to estimate the total number of neurons, where ΣQ- is the number of neurons actually counted in the specimens.
Digital Photography
Low-magnification photomicrographs demonstrating a general view of the distribution of ChAT- and ERα-ir cells in different basal forebrain nuclei (Figure 5) were taken with a Nikon Coolpix 990 digital camera attached to the Optiphot-2 Nikon microscope using a Plan X10 objective (NA 0.30). High-magnification photomicrographs with a CFI Plan Fluor X100 oil immersion objective (NA 1.30) demonstrating optical views of Durcupan-embedded and Depex-mounted sections (Figures 3 and 4) were taken using a 3-chip CCD color video camera (HV-C20; Hitachi), which belongs to the set-up for integrated hardware-software package used for cell counting (described above).
Results
Section Thickness Comparison
Measurements of section thickness revealed that (a) unstained sections were 48.8 ± 1.15 μm (mean ± SEM), (b) double-immunostained sections that were mounted with Depex were 12.8 ± 0.1 μm (see also Figure 4), and (c) double immunostained sections embedded in Durcupan were 40.8 ± 0.35 μm (see also Figure 3). The surfaces of the Durcupan-embedded sections were clear and bright and it was easy to focus on the level at which the field of view on the top of the section turned from unfocused exterior into focused interior of the sample and vice versa at the bottom of the section.
General Observations About Durcupan-embedded Material
Macroscopically, sections mounted in Durcupan were darker than the sections in Depex because of Durcupan's intrinsic brownish color. However, at the microscopic level this did not interfere with identification of the cell bodies or the color contrast between different chromogens in the double-immunostained sections. The major advantage of Durcupan embedding was that individual cells could be clearly identified as their own entities at different focal planes (Figure 3). In contrast, the cells in sections mounted with Depex were flattened and the depth resolution was poor. This made it difficult to identify individual cells, especially those that were close to each other (Figure 4).
Microscopic examination revealed that ERα-ir (dark blue color) had a punctate or granular appearance (Figures 3 and 4). In the majority of cells, ERα-ir was present in the nucleus, where it formed dense aggregates. However, we found that in some cells, ERα-ir occurred in granules scattered throughout the cytoplasm. ChAT-ir (observed as brown) was located exclusively in the cytoplasm (Figures 3 and 4). The texture of ChAT-ir was more homogenous than ERα-ir, filling up the entire cytoplasm of the cell soma and dendrites. The color difference between ERα-ir and ChAT-ir was clear in all sections and was independent of the mounting method used (Figures 3 and 4). However, individual cells and their different subcellular compartments were easier to distinguish from each other in Durcupan-embedded sections because different focal levels were better monitored inside the thicker sections (Figure 3). Counting on Depex-mounted sections was difficult because it was hard to distinguish individual cells. Therefore, cells were counted only from Durcupan-embedded sections.
High-magnification (X100 oil objective) photomicrographs of a series of optical sections through one area in Durcupan-embedded material, demonstrating the counting rules and immunoreactivities for ChAT (brown) and ERα (blue-to-black). Counting frame (35 μm × 35 μm), the focus position meter on the left side of the image (horizontal red or green bar indicates the depth of the focal plane in μm), and the disector height (green vertical bar in the focus position meter) are shown on the images. (
High-magnification (X100 oil objective) photomicrographs demonstrating typical views through a section immunostained for ChAT (brown) and ERα (blue-to-black) and mounted with Depex. The sampling parameters are the same as in Figure 3 except that the green vertical bar in the focus position meter in these images shows the total thickness of the section (i.e., 12 μm), not the disector height. (
Estimation of Cell Numbers
The distribution of the ChAT-ir cells in the basal forebrain was in agreement with previous studies (Mesulam et al. 1983; Schwaber et al. 1987; Kiss et al. 1990; Wainer and Mesulam 1990). Briefly, the ChAT-ir cells in the medial septum (MS) were located in and around the midline of the septum (Figure 5A). The overall pattern of ChAT-ir cells in this area resembled the structure of an onion, with a bulb wrapped inside a couple of layers that contained different densities of ChAT-ir cells. On the midline itself there were a few ChAT-ir cells. Moving laterally, there was a layer in which ChAT-ir cells are rare. This corresponds to the area at which parvalbumin-containing cells have been previously identified (Kiss et al. 1990). More lateral to this ChAT cell-poor region, there was a wider area that contained the highest number of ChAT cells in the MS. Typically, clusters formed by two or three ChAT-ir cells were seen in this area. Ventral to the MS, in the vertical limb of the diagonal band (VDB), ChAT-ir cells were distributed relatively homogeneously. No cell-poor region (as was the case in the MS) was observed. The ChAT-ir cells belonging to the MS and VDB were counted together, and the results are presented as the MSVDB values. In more caudal sections, ventrolaterally to the VDB, the ChAT-ir cells were located in the horizontal diagonal band (HDB), where they formed an oval cell mass having the interior part devoid of ChAT-ir cells (Figure 5B). Typically, in the HDB, almost all ChAT-ir cells had a bipolar morphology and were oriented parallel to the brain ventral surface. The few scattered ChAT-ir cells that belong to the magnocellular preoptic nucleus were excluded from the cell counting.
The ChAT-containing cells referred to as the Ch4 cell group by Mesulam (Mesulam et al. 1983) included the ChAT-ir cells in the region starting caudally after fusion of the anterior commissure. The ChAT-ir cells in the nucleus basalis (NB), which is the major component of the Ch4 group, first appeared as a small cell cluster at more anterior levels. Posteriorly they first spread along ventral surface of the globus pallidus and continued to arise between the internal capsule and globus pallidus (Figure 5C). The scattered ChAT-ir cells of the substantia innominata and ventral pallidum were also included in these cell counts and are collectively called the NB cell group in the present study. The reason why this entire area was included for cell counting was that the borders of it were possible to distinguish from section to section and from case to case in a consistent repetitive manner.
In general, the ERα-ir cells of the basal forebrain levels examined followed the distribution of ChAT-ir cells (Figure 5), except that they were more numerous and were present also in high numbers in the areas surrounding the ChAT-ir regions. Very high densities of ERα-ir cells were observed both in the islands of Calleja and the hypothalamic regions. They were also numerous in the lateral septum and preoptic areas. In addition, some scattered ERα-ir cells were detected in the globus pallidus. No ERα-ir cells were observed in the caudate putamen. Because our primary interest was to estimate the incidence of ERα-ir in the ChAT cells, the neurons containing ERα-ir alone were not counted.
The subcellular distribution of ERα-ir in the ChAT-ir cells was registered while counting the cell numbers. On the basis of the subcellular distribution three types of subclasses of ChAT cells were made: (a) ChAT-ir neurons without ERα-ir; (b) ChAT-ir neurons that had ERα-ir in the nucleus; and (c) ChAT-ir neurons that had ERα-ir in the cytoplasm (Figure 3). In the last group, ERα-ir typically appeared as scattered granules in the cytoplasm, and these types of cells were clearly distinguishable from those ChAT-ir cells having ERα-ir in the nucleus. A few cells contained both nuclear and cytoplasmic ERα staining. The possible ERα content of each ChAT neuron was always confirmed through microscope oculars.
Estimation of the cell numbers in Durcupan sections revealed that the rat basal forebrain (including MS, VDB, HDB, and NB of both hemispheres) contained 26,390 ± 1097 (mean ± SEM) ChAT-ir cells, 42% of which contained ERα-ir (Table 1). In the individual basal forebrain nuclei, the total number of ChAT-ir neurons was 9674 ± 504 (CV = 0.10) in the MSVDB, 9403 ± 484 (CV = 0.10) in the HDB, and 7312 ± 281 (CV = 0.07) in the NB (Table 1). The total number of ChAT-ir neurons containing ERα-ir was 5736 ± 71 in the MSVDB (CV = 0.02), 4304 ± 237 in the HDB (CV = 0.11), and 1044 ± 140 in the NB (CV = 0.26) (Table 1). The number of ChAT-ir neurons containing ERα-ir in the nucleus or the cytoplasm was 3908 ± 200 and 1828 ± 186 in the MSVDB, 2330 ± 55 and 1975 ± 186 in the HDB, and 512 ± 119 and 532 ± 79 in the NB, respectively.
The rat brains were cut at 50 μm into six series (Figure 2). One series from each animal was randomly selected for processing sections for cell counting. In other words, the sections that were sampled for cell counting were 5 times 50 μm apart from each other. With this sampling system, three to five sections contained the MSVDB, five to seven sections contained the HDB, and seven sections contained the NB (Table 1). With the sampling grid 80 μm × 80 μm and counting frame 35 μm × 35 μm, the number of counted ChAT-ir cells (including all three categories of ChAT-ir cells) was in the MSVDB from 203 to 262, in the HDB from 201 to 253, and in the NB from 155 to 184 (Table 1). Because the ChAT-ir cells in the HDB were more evenly dispersed along different rostrocaudal levels than in other nuclei, it would have been unnecessary to count so many cells to be efficient in cell counting. Therefore, we performed an additional counting using a sampling grid size 160 μm × 160 μm. This resulted in the number of counted cells from 48 to 59. The estimate of total number of ChAT-ir (including all three categories of ChAT-ir) cells was then 9151 ± 200 (CV = 0.08). This indicates that for the HDB even such a low number of counted cells would yield a good estimate of the total cell number.
Discussion
In this article we describe a method by which we could prevent shrinkage and flattening of the immunohistochemically stained sections. We found that embedding sections in the epoxy resin Durcupan after immunostaining preserves the three-dimensional structure of the tissue. These sections are excellent for analyses of cell counts for which the optical fractionator method is employed. In addition, the present report is the first to describe estimates of both the total numbers of ChAT-ir neurons and the total numbers of the ChAT-ir neurons that contain ERα-ir in the basal forebrain of male Wistar rats.
Embedding of Immunostained Sections in Durcupan Is Superior to Routine Mounting Methods
The development of modern stereological counting methods, especially the optical fractionator, has led to a great improvement in quantitative studies in many respects. First of all, the shrinkage problem, which cannot be handled unequivocally in biological material is basically avoided (see above, West 1993). By these methods, one yields total numbers per organ, nucleus, or any region that can be clearly defined according to consistent criteria (see West 1993; Hyman et al. 1998). Therefore, knowledge of the actual reference volume or area size from which the cells are counted is not necessary. Furthermore, issues concerning shrinkage or swelling of tissue during the histological processing can be largely disregarded (Gundersen 1986; Braendgaard and Gundersen 1986; West 1993). However, special attention is required to ensure that the samples are processed in an unbiased systematic manner so that each cell has an equal possibility to be sampled. In addition, one must cut the samples equally so that the sections in the series taken for the analyses are a defined distance apart. It is therefore very important that the cutting method gives repetitive results so that section thickness does not vary unexpectedly from section to section. In the present study we cut sections at 50 μm on a vibratome. Our section thickness measurements showed that the native sections were 48.8 ± 1.15 μm thick. This can be considered a reasonably good result. The slight difference compared to aimed value is likely to be due to structure collapse occurring after cutting because of lost connections to longitudinal brain structures that exist in vivo.
Low-magnification photomicrographs demonstrating the distribution of ChAT-ir (brown) and ERα-ir (blue-to-black) (Depex-mounted sections, X10 objective) in the medial septum (
Estimated total numbers of ChAT-ir cells and their ERα-ir content in the medial septum-vertical diagonal band of Broca (MSVDB), horizontal diagonal band of Broca (HDB), and nucleus basalis (NB) of male Wistar rats a
In parentheses, number of cells actually counted during the analysis; n, number of sections containing the region of interest in a series of sections sampled systematically with a random starting point; %, percentage of ChAT cells that contained ERα from the total number of ChAT cells; SEM, standard error of mean; CV, coefficient of variance.
Despite the fact that the optical fractionator method is reported to be unaffected by shrinkage of the tissue or absolute section thickness (Gundersen 1986; West et al. 1991; West 1993), shrinkage can affect analysis itself. We found that, in those regions where cells form clusters, it was difficult to identify individual cells if the sections collapse profoundly during the histological procedure. Therefore, in such cases it is important to try to preserve section thickness to be able to identify distinct cells. In the present study we found a considerable effect of the mounting method used after immunostaining. Section thickness measurement showed that the commonly used dehydration and mounting in Depex results in extensive section shrinkage. Thickness measurements revealed that only 12.8 ± 0.1 remained from the sections cut at 50 μm. This 74% shrinkage from the original agrees with previously published observations (Hornberger et al. 1985; Fischer et al. 1992; see also Dorph-Petersen et al. 2001). In contrast, the sections that were embedded in Durcupan had a section thickness of 40.8 ± 0.35 μm. Therefore, shrinkage has also taken place in these sections. It can be assumed that different chemical procedures (such as Triton X-100 treatment) during immunostaining cause some of this shrinking. However, the shrinkage of the sections embedded in Durcupan is minimal compared to that of the sections mounted with Depex. Interestingly, there was no difference in the x,y-dimensions between these two materials (data not shown). It should be noted that when sections are mounted with Depex they are first mounted on slides, then dried in the air (either at RT or in the oven), after which they are dehydrated and mounted with Depex and a coverslip. We believe that the most critical step in that procedure is the drying of the sections in air, after which section thickness never recovers. In contrast, during the entire Durcupan embedding procedure the sections are freely floating in dehydration solutions and are never exposed to air at any stage.
Resin embedding has also been described earlier (Gerrits and Horobin 1996). However, the principal difference compared to our method is that when, for example, glycolmethacrylate sections are used they can be stained primarily with some classical histological stains that demonstrate nuclei and nucleoli of the cells. However, IHC stainings cannot be thoroughly performed on these types of sections because of the very poor penetration after embedding and mounting, or when antigens of interest are sensitive to dehydration required for embedding (Dorph-Petersen et al. 2001). Our method overcomes this problem because immunostaining is performed before the sections are embedded in resin. On the other hand, similarly to methacrylates, Durcupan embedding also gives excellent support to the structure. Therefore, collapse of watery compartments of the tissue that takes place during routine mounting is avoided. Flattening of compartments with a high content of water or with no structural elements, such as the lumen of blood vessels, is likely to be, for the most part, behind the tissue deformation discussed by Dorph-Petersen and colleagues (2001).
Optically, the Durcupan-embedded sections are very clear, and it is easy to scan on various levels through the section. This facilitates verification of the penetration of immunohistochemicals as well. Because the flatness of Depex-mounted sections it is difficult to determine whether the penetration is uniform throughout the section. If penetration problems go undetected, they will lead to underestimation in the structure counts. Because antibody penetration can be easily verified in Durcupan-embedded sections, this can be considered as a further advantage of the Durcupan embedding method.
Cholinergic Cells in Different Basal Forebrain Nuclei Show Variations in Content of ERα
Our study showed that the total number of ChAT-ir cells in the basal forebrain is 26,390 ± 1097, and this is composed of 9674 ± 504 ChAT-ir neurons in the MSVDB, 9403 ± 484 in the HDB, and 7312 ± 281 in the NB. These numbers differ from the previously published studies. However, we believe we have a more accurate representation of the numbers of ChAT-containing cells because of the ability to distinguish individual neurons associated with our method.
Any direct correlation between our studies and others is difficult because of differences in the age, strain, and sex of the rats, possible operations carried out on the animals (e.g., ovariectomy), the antibodies used, the section mounting method, and the definition of the anatomic boundaries for the different nuclei analyzed. However, it should be noted that we find significantly more ChAT-containing cells in our study. For example, the most recent study by Aggarwal and Gibbs (2000) on young female ovariectomized Sprague-Dawley rats showed about six times fewer cholinergic cells in the basal forebrain compared to our study. Similarly, Fischer et al. (1992) found that at 3 months of age the female Sprague-Dawley rats had 3254 cholinergic neurons in the MSVDB and 2192 in the NB per brain hemisphere. When the values are multiplied by 2, they still represent about one third of the counts we obtained in our study.
We also investigated the proportion of cholinergic cells that contain ERα. In an earlier study, Gibbs (1996) found that in male gonadectomized Sprague-Dawley rats co-localization percentages are 75%, 65%, and 75%, respectively, for the MS, HDB and NB. The corresponding percentages in our study are 60%, 46%, and 14%, respectively. There is a clear discrepancy between these two studies, which can be due to the factors discussed above in respect to ChAT neuron number. In addition, there are also methodological differences that could be important. For example, Gibbs examined only some rostrocaudal levels of the basal forebrain. This is important to note because the co-localization percentages vary rostrocaudally (Shughrue et al. 2000). It is therefore essential to examine sections that represent the different rostrocaudal levels to give an accurate reflection of the true levels of co-localization in these nuclei. It should be also noted that Gibbs used a routine mounting method for the sections. As we demonstrated in the present study, different structure components are very close to each other in the z-dimension in such sections (see Figures 1 and 4). If different focal planes expressing distinct structural components cannot be accurately identified, false interpretation of co-localization can take place. This will lead to overestimation of the co-localization.
Shughrue and colleagues (2000) have also studied co-localization of ChAT and ERα, using another method. Their findings showed significantly lower percentages (i.e., 41% in the MS, 32% in the VDB, 29% in the HDB, and 4% in the NB) compared to our study. The major causes for these differences can be age, sex, ovariectomy, and strain of the rats, as well as the method used (in situ hybridization using radioactively labeled probes for estrogen receptors combined with immunohistochemistry for ChAT). First of all, it is possible that the level of mRNA and the corresponding protein in the neurons is not directly correlated. However, second, the mounting and detection method can also have a strong effect on the results. Using the embedding method described in here we could clearly distinguish individual neurons and study in different focal planes their possible content for ERα. Furthermore, the subcellular distribution could be clearly identified at its natural location while scanning through the section. However, in autoradiography, the signal is detected in the emulsion, which covers the section. Thus, the “co-localization” is examined by matching the immunohistochemically stained ChAT neurons, which are within the section (35 μm thick in Shughrue's paper), with the silver grains in the emulsion that is developed by the radioactive signal originating from the section and is localized on the top of but not in the section. Therefore, these two factors are spatially non-co-localized in strict sense and do not represent proper co-localization.
The antibody against ERα used in the present study is raised against a peptide corresponding to an amino acid sequence at the C-terminus which is believed to be responsible for nuclear localization (Picard et al. 1990; see also Pavao and Traish 2001 for comparison of other ER antibodies). However, we noted two different subclasses of the ChAT-ir neurons according to the subcellular localization of ERα-ir: ERα-ir was located either in the nucleus or in the cytoplasm. At the current stage of knowledge it is difficult to express what is the direct functional importance of this observation. However, we have observed (unpublished observation) in female mice that the receptor location in cholinergic cells changes during the aging, having a decreased preference for nuclear localization at older ages. Because aged females have lower circulating estradiol levels, the cytoplasmically located ERαs may represent receptors that are free of ligand, because ERαs would translocate to the nucleus when estradiol is bound to it. Alternatively, it is also possible that the different distribution reflects the fact that there are functionally different estrogen receptors; nuclear receptors may mediate genomic effects and cytoplasmically located ERαs are involved in the non-genomic activities of estrogens (Aronica et al. 1994; for review see Nilsson et al. 2001). However, different subcellular location of ERαs in distinct ChAT-containing neurons can also suggest about different functional status of distinct cholinergic neurons in respect to receptor trafficking.
In conclusion, we have described here critical caveats that can affect quantification and which should therefore be accounted for before material is processed for cell counting. In addition, the present study shows that even though the number of cholinergic cells does not vary greatly among different basal forebrain nuclei, their contents of ERα differ from each other. This is likely to have fundamental significance to the proposed action of estrogens in these nuclei.
Footnotes
Acknowledgements
Supported by EVO grant (5510) of the University Hospital of Kuopio, the European Commission (QLK6-CT-1999–02112), the National Technological Agency, and Hormos Medical, Ltd. (40414/00).
We thank Prof Asla Pitkänen for her constructive criticism and thoughtful suggestions on the manuscript, Dr Thomas Dunlop for revising the language, and AnnaLisa Gidlund for excellent technical assistance.