All animal procedures were carried out with approval from the Institutional Animal Care and Use Committee at the University of Pittsburgh. Male Sprague-Dawley rats (Hilltop Lab Animals, Inc., Scottdale, PA) weighing 300–400 g were deeply anesthetized with sodium pentobarbital (60 mg/kg i.p.) and pretreated for 15 minutes with sodium diethyldithiocarbamate (DEDTC, Spectrum Chemical Corp., Gardena, CA; 1 mg/kg i.p.) to prevent silver enhancement of endogenous zinc ions in axon terminals (Veznedaroglu and Milner, 1992
). A cohort of paired animals were then perfused using a transcardial approach with 50 mL of heparinized saline (1000 U/mL), followed by 500 mL of 3% paraformaldehyde and 0.15% glutaraldehyde in 0.1 M phosphate buffer, pH 7.4 (PB). The brains were removed and cut in the coronal plane to yield 3–4 mm thick blocks. Tissue was then post-fixed overnight in the same fixative at 4°C. Brains were sectioned through the anterior thalamus and the midbrain to a thickness of 50 µm and collected in PB. Sections were treated with 1% sodium borohydride in PB for 30 minutes and rinsed extensively.
Sections were labeled by immunoperoxidase or immunogold-silver using a polyclonal antibody raised in rabbit against the C-terminus 15 amino acid sequence that is conserved in the human, mouse and rat CHT protein. Several lines of evidence support the specificity of this antibody (Ferguson et al., 2003
). The antiserum labels a band of the predicted molecular weight for CHT by Western blot analysis of mouse whole tissue or brain synaptosomes and from PC12 cells transfected with human CHT. This immunoreactive band is absent from kidney and from PC12 cells transfected with empty vector. Immunohistochemical labeling with the polyclonal rabbit antibody conforms to the expected distribution of cholinergic neurons and fibers throughout the mouse and rat brain and at the neuromuscular junction, and is co-localized with other markers of cholinergic phenotype but not other transmitters. Pre-adsorption with the immunizing peptide abolishes this labeling (Ferguson et al., 2003
). Finally, immunolabeling is absent from CHT −/− mice (Ferguson and Blakely, 2004
Sections for immunoperoxidase visualization of CHT were rinsed in 0.1 M Tris-buffered saline, pH 7.6 (TBS) and incubated for 30 minutes in a blocking solution containing 1% bovine serum albumin (BSA), 5% normal donkey serum (NDS, Jackson Immunoresearch Laboratories, Inc., West Grove, PA), and Triton X-100 (Sigma, St. Louis, MO) at 0.2% or 0.04% for light or electron microscopy, respectively. Sections were then transferred to blocking solution containing the polyclonal rabbit anti-CHT at 1:1000 and incubated overnight at 4°C. After rinsing, sections were incubated for 30 minutes in biotinylated donkey anti-rabbit IgG (Jackson Immunoresearch) diluted at 1:400 in blocking solution. Excess secondary antibody was removed with several rinses in TBS, and sections were incubated in ABC solution (Vectastain kit, Vector Laboratories, Burlingame, CA) for 30 minutes. Peroxidase product was developed by incubation in 0.022% diaminobenzidine (Sigma) and 0.003% hydrogen peroxide for 3 minutes. Sections for light microscopy were rinsed extensively in TBS and 0.01 M phosphate buffered saline prior to mounting on Superfrost microscope slides (Fisher Scientific, Pittsburgh, PA). After drying, slides were dehydrated through a series of increasing ethanol concentrations, defatted in xylene, and coverslipped with Cytoseal-60 mounting medium (Richard Allen Scientific, Kalamazoo, MI). Immunoperoxidase labeled sections for electron microscopy were prepared as detailed below.
Sections for pre-embedding gold-silver immunostaining were treated according to the recommended protocol included with the Aurion RGent SEM kit (Electron Microscopy Sciences, Hatfield, PA). Sections were rinsed in 0.02 M phosphate buffered saline, pH 7.4 (PBS) and permeabilized in 0.05% Triton X-100 for 10 minutes. Non-specific antigenic sites were blocked by a 30 minute incubation in PBS containing 5% BSA, 5% NDS, and 0.1% fish gelatin (GE Healthcare Life Sciences, Waukesha, WI). Sections were then rinsed in an incubation buffer containing 0.2% acetylated BSA (Aurion BSAc, Electron Microscopy Sciences) and incubated overnight at 4°C in incubation buffer containing primary antibody (rabbit anti-CHT, 1:1000). Sections were rinsed extensively in incubation buffer and placed in vials containing secondary antibody (Aurion donkey anti-rabbit 0.8 nm gold conjugate, 1:50, Electron Microscopy Sciences) and 5% NDS in the same buffer. Secondary antibody binding occurred overnight at 4°C. After rinsing in incubation buffer and PBS, antibody complexes were fixed by incubation for 10 minutes in 2% glutaraldehyde in PBS. Following several rinses in PBS, gold particles were silver-enhanced using the Aurion RGENT SEM kit, according to the manufacturer’s instructions and using the optional enhancement conditioning solution (Aurion ECS, Electron Microscopy Sciences). Silver enhancement proceeded from 45–75 minutes and was terminated by several rinses in ECS, followed by 0.1 M PB.
Tissue preparation for light and electron microscopy
Brightfield micrographs were captured on an Olympus BX-51 microscope (Olympus America Inc., Center Valley, PA) equipped with a CCD camera (Hamamatsu, Bridgewater, NJ). Acquired images were imported into Adobe Photoshop (Adobe Systems Incorporated, San Jose, CA) and modified to match brightness and contrast.
Immunoperoxidase and immunogold-silver reacted tissue was processed for electron microscopy by incubation in 2% osmium tetroxide in 0.1 M PB for 1 hour, dehydrated through a series of increasing ethanol concentrations, treated with propylene oxide, and infiltrated with an epoxy resin (EMBed-812, Electron Microscopy Sciences). Sections were then flat-embedded between sheets of commercial plastic (Aclar, Electron Microscopy Sciences). Ultrathin (60–70 nm) sections were cut through the regions of interest on a Leica Ultracut ultramicrotome (Leica Microsystems, Bannockburn, IL) and mounted on either copper mesh grids or carbon coated copper slot grids (Electron Microscopy Sciences). Sections were then counterstained with 5% uranyl acetate and lead citrate and examined on a transmission electron microscope (Morgagni, FEI Company, Hillsboro, Oregon) equipped with a CCD camera (Advanced Microscopy Techniques, Danvers, MA). Digital micrographs of labeled axonal profiles were captured at 14,000 – 28,000x magnification and adjusted for exposure and contrast in Adobe Photoshop.
In cholinergic projections to the cortex, it has been shown that the synaptic incidence extrapolated from single section observations is equal to the rate of synapse formation obtained from the analysis of serial ultrathin sections (Umbriaco et al., 1994
). However, this relationship has not yet been established for the projections of the PPT/LDT cholinergic cells. We therefore evaluated the rate of synapse formation observed in serial section data and compared this to an extrapolated synaptic incidence calculated from data obtained only from single sections.
For each rat, two vibratome sections were examined through each region. For most CHT immunolabeled profiles, we were able to obtain a series of micrographs from adjacent ultrathin sections, and we used these profiles to determine if each displayed a synaptic specialization (n = 637, Data set #1, ). We also used this first data set to determine whether each CHT labeled profile contained at least one dense-cored vesicle or none.
Sampling Scheme for Ultrastructural Analysis
To further analyze the extent of synapse formation, we utilized a second data set of single sections through CHT immunoreactive profiles from 3 of the 6 animals (n = 410, Data set #2, ) and applied a well-accepted method for post hoc size correction to generate an extrapolated synaptic incidence (Beaudet and Sotelo, 1981
). By comparing the values obtained from the serial sections with that extrapolated from single section analysis, we could determine if this extrapolation method can accurately estimate synaptic incidence in these brain regions. More specifically, we recorded the minimum diameter (D), and length of the synaptic junction (d) for each of these profiles. We also measured the thickness of our sections (w) using the minimal fold method (Small, 1968
). We computed the means of these measurements for each region, and applied the following equation:
where P is the probability of observing a synapse in a single section through an axonal profile if all profiles form one synapse (Beaudet and Sotelo, 1981
). In other words, P
*(true synaptic incidence
) = single-section synaptic incidence
. We therefore divided the rate of synapse formation we observed in our single section sample by this probability (P
) to generate the extrapolated synaptic incidence.
For the analysis of CHT immunogold localization, we used a third data set consisting of single sections through each profile from both Data sets #1 and #2 (n = 1047, Data set #3, ). To avoid bias in the selection of a single image through the serial micrographs from Data set #1, we chose the image that was photographed first, regardless of its position within the series. In this case, we assumed that the first encounter with a profile constituted a random event.
Using a commercial image analysis software program (SimplePCi, Hamamatsu), we measured immunoreactive profiles and recorded the number of gold particles in the cytoplasm and associated with the plasma membrane. From these data, we derived the following measurements: (1) profile area, 2) profile perimeter, 3) total gold density of CHT immunogold particles, defined as the total number of gold particles per unit profile area, and 4) membrane density of CHT immunogold particles, defined as the number of membrane associated gold particles per unit profile perimeter. Membrane gold particles were defined as those immediately in contact with the plasmalemma or separated by no more than 20 nm, based on estimates of gold particle and immunoglobulin size (Mathiisen et al., 2005
Silver-enhanced gold particles were observed at high density in axonal profiles that clearly contained synaptic vesicles. The presence of a lower density of gold-silver particles in some dendrites suggested a certain degree of background staining, which varied between animals. Therefore, when there was any concern regarding specific gold labeling in axons, these profiles were followed in a short series of adjacent sections. Profiles that contained fewer than 10 gold particles per square micron were excluded from our sample unless consistent immunolabeling was observed in at least 3 serial micrographs. Non-specific background labeling was similar in the two regions, averaging 1.6 ± 1.3 gold particles per µm2 in the AVN vs. 1.7 ± 1.4 gold particles per µm2 in the VTA. Specific gold-silver labeling in the final data set was determined to be at least 2.5 times above background levels, with the majority of profiles well beyond this minimum level.
Serial micrographs from the first data set () were examined for the presence of synaptic specializations and dense-cored vesicles. The probabilities of synapse presence and dense-cored vesicle presence were modeled using generalized linear mixed models based on the Bernoulli distribution, with the logit as the link function. Brain region was treated as a fixed effect. Animal pair was also treated as a fixed effect to account for the fact that animal tissue was processed in pairs. To account for the correlation among observations within an animal and among observations within each Vibratome section, animal and section nested in animal were treated as independent normally distributed random effects.
The remaining dependent measures were analyzed from the third data set of all labeled profiles (). A linear mixed model was used to model the profile area of CHT-immunoreactive structures, with brain region and animal pair treated as fixed effects. To account for the correlation among observations within an animal and among observations within each Vibratome section, animal and section nested in animal were treated as independent normally distributed random effects.
A generalized linear mixed model, based on a Poisson distribution, was used to model total gold density of CHT immunogold positive profiles, with the log as the link function and the natural logarithm of profile area as an offset variable. Brain region and animal pair were treated as fixed effects. To account for the correlation among observations within an animal and among observations within each Vibratome section, animal and section nested in animal were treated as independent normally distributed random effects.
Membrane density of CHT immunogold positive profiles was modeled in a manner similar to that of total gold density of CHT immunogold positive profiles. The only difference arose from the fact that the model for membrane density used the natural logarithm of profile perimeter as an offset variable and included total gold density and the interaction between brain region and total gold density as fixed effects, in addition to brain region and animal pair.
The analysis of profile area was implemented in SAS PROC MIXED (Version 9.2, SAS Institute Inc., Cary, NC), while all other analyses were implemented in SAS PROC GLIMMIX. The Kenward-Roger degrees of freedom method was used in each analysis. When testing for the significance of fixed effects in a mixed model, the Kenward-Roger method is generally recommended to approximate the denominator degrees of freedom in subsequent F tests (Littell et al., 2006
). This method corrects for the fact that the estimated variability of the fixed effects parameter estimates tends, on average, to be lower than the actual variability of these estimates.
All statistical tests were conducted at the 0.05 significance level.