Supplemental Information
Materials and Methods
All experimental protocols were approved by the Ethics Committee of the Tokyo Metropolitan Institute of Medical Science.
Animals
Wistar rats (Oriental Yeast Co., Ltd., Tokyo, Japan) were maintained in a temperature-controlled room (~23˚C) with a 12 h/12 h light/dark cycle, in accordance with the guidelines for the Animal Use and Care Committee of the Tokyo Metropolitan Institute of Medical science. Experiments were performed during the light period. Eight-week-old rats (male, n=8; female, n=4) and 48-week-old rats (male, n=4) were killed with a lethal dose of anesthesia (pentobarbital sodium), and then perfused with phosphate-buffered saline (PBS) containing 50 IU/ml heparin. The brains were then removed from the skull, and the cerebral cortex was dissected out as described by Herculano-Houzel and Lent.1 The brains were immediately frozen and stored at -80°C until measurements were carried out. The frozen brains were used within a month, except for a group of brains from 8-week-old rats (n=4), which were stored for 385 days in order to investigate the effects of frozen storage days.
For our investigation on the effects of postmortem interval (PMI), rats (8-week-old, male, n=4) were killed by either a lethal dose of anesthesia or by cervical dislocation. The animals were then kept at 4°C for 0, 24, or 48 hours without perfusion. At each PMI, brains were removed from the skull, the cerebral cortex was dissected out as described above, and the tissue was immediately frozen and stored at -80°C until measurements were carried out. These samples differed from the perfused samples in that they contained blood cells.
Postmortem Brains
Frozen brain blocks from the frontopolar (FPC; Brodmann area [BA] 10) and inferior temporal (ITC; BA20) cortices were generously provided by the Stanley Foundation Brain Collection (The Stanley Medical Research Institute, Bethesda, MD). Brain blocks were dissected from the lateral aspect of the frontal pole (BA10) and from the gyral surface of the inferior temporal gyrus (BA20).2 The original collection consisted of 15 normal control and 15 major depressive disorder (MDD) FPCs as well as ITC blocks. Psychiatric diagnoses had been established by two psychiatrists using DSM-IV criteria and were assigned by the Stanley Foundation review committee. We excluded 8 FPC and 7 ITC samples because they lacked entire gray matter layers, and analyzed the remaining samples (FPC, 12 control, 10 MDD; ITC, 12 control, 11 MDD). Demographic data on the brains included in the analysis are provided in Table S2. There were no significant differences between the groups vis-à-vis any of the available confounding factors. None of the MDD subjects had ever exhibited psychotic behavior. Two MDD subjects in the FPC analysis and three MDD subjects in the ITC analysis had been substance abusers at the time of death. Other information about these brains has been described previously.3 The entire gray matter, from the surface to the border between layer VI and the white matter, was carefully dissected out manually from the frozen blocks with a sharp blade. About 20 mg of gray matter was collected from each specimen.
Nuclei isolation
The human and rat brain samples were dissolved and wet weights were measured. A crude suspension of nuclei was obtained through mechanical homogenization of the brain tissue in a 10-fold volume of lysis buffer (V/W) (0.32 M sucrose, 5 mM CaCl2, 3 mM magnesium acetate, 0.1 mM EDTA, 10 mM Tris-HCl [pH 8.0], 0.1% Triton X-100).4 Homogenization was performed with a set of 1.5 ml Eppendorf tubes and a plastic pestle by grinding for about 1 minute. This isolation method dissolved the cytoplasmic membrane and organelles but left the nuclear membrane intact. The homogenized brain tissues were immediately centrifuged (10 min at 1000×g) and the resulting pellets containing nuclei and debris were collected. To detect all nuclei, we used 7-aminoactinomycin D (7-AAD), a fluorescent DNA marker whose emission has a very large stroke shift, with the maximum occurring in the deep red range of 647 nm. In the rat study, the pellets were resuspended in PBS containing 10 μl/ml 7-AAD (BD Pharmingen, Franklin Lakes, NJ) and 50 μg/ml RNase A (Nacalai Tesque, Kyoto, Japan), while in the human study the pellets were resuspended in PBS containing 40 μl/ml 7-AAD and 50 μg/ml RNase A. The final volume of the nuclear suspension was adjusted to 10 mg tissue weight per 1 ml suspension solution.
Nuclei counting
Flow-Count® (Beckman Coulter, Fullerton, CA) was used to count the number of nuclei. Flow-Count® contains control beads of identical bead size that have high fluorescence emissions across all ranges of the flow cytometer; 1 μl of Flow-Count® contains a finite number of beads. An aliquot of 7-AAD-stained nuclei was mixed with an equal amount of Flow-Count® prior to counting the nuclei by FCM (Epix XL, Beckman Coulter, Fullerton, CA). Nuclei counting continued until the flow cytometer had counted 1000 control beads. The exact number of nuclei was determined by averaging the numbers obtained from at least three independent measurements. In preliminary studies, we also counted the number of nuclei using a hemocytometer. The number of 7-AAD(+) nuclei in the rat whole cerebral cortex were confirmed to be almost identical to those previously estimated by Herculano-Houzel and Lent.1
Immunostaining
After the nuclei counts, we centrifuged the nuclear suspension at 1000g for 10 min at room temperature (RT). The resulting pellet was resuspended in 100 μl of blocking solution (3% BSA, 2% skim milk in PBS) and incubated for 30 min at RT. Subsequently, the blocking solution was centrifuged, and the resulting pellet was resuspended in 100 μl of blocking solution containing mouse anti-NeuN IgG (1:1000 in the rat study; 1:500 in the human study; Chemicon, Temecula, CA) and rabbit anti-olig2 IgG (1:2000 in the rat study; 1:1000 in the human study; Chemicon) or isotype IgG controls (anti-mouse IgG from BD Pharmingen; anti-rabbit IgG from Southern Biotech, Birmingham, AL) and incubated overnight at 4oC. We chose NeuN and olig2 as neuronal and oligodendroglial nuclear markers, respectively, partly because we could not find any established and appropriate nuclear markers for other neural cells, such as astrocytes, microglia, and endothelial cells. NeuN is an antibody that recognizes a neuron-specific antigen and selectively stains neuronal nuclei,5 while olig2 is a bHLH transcription factor that regulates key stages of early OL and motor neuron development.6,7 The olig2 marker is frequently used as a pan-oligodendroglial cell-type marker in adult mammalian brains. 8-10
The next day, 1 ml of PBS/0.05% Tween20 was added to the suspension, and the whole was centrifuged. The resulting pellet was resuspended and incubated in blocking solution containing anti-mouse Qdot565 (Molecular Probes, Burlington, Canada) and anti-rabbit Alexa488 (Molecular Probes) for 1 h at RT. Thereafter, stained nuclei were suspended in PBS containing 10 μl/ml 7-AAD and 50 µg/ml RNase A. The results of the FCM analyses are presented as percentages of NeuN(+), olig2(+), and NeuN(-)/olig2(-) nuclei in 20,000 7-AAD(+) nuclei. Each absolute number of nuclei was calculated by multiplying the number of 7-AAD(+) nuclei by the percentage of NeuN(+), olig2(+), and NeuN(-)/olig2(-) nuclei.
Statistical Analysis
Statistical analyses were carried out using an unpaired Student’s t test for two-groups and a one-way analysis of variance (ANOVA) followed by a Dunnett’s post hoc test for more than two groups at the 5% significance level with GraphPad PRISMTM (Graph Pad software, Inc., San Diego, CA) and SPSSTM (SPSS, Inc., Chicago, IL) software. In order to determine whether demographic or descriptive variables, other than diagnosis, contributed significantly to the variances in the FCM measurements, we first performed Pearson’s correlations between the density of nuclei for each cell type and: age, PMI, pH, weight, frozen storage days, age of disease onset, and duration of disease. The nuclei density was then compared between groups with analysis of covariance (ANCOVA) for any descriptive variables that contributed a significant proportion of variance in the FCM measurements. The effects of non-continuous descriptive variables (gender, hemisphere, and severity of alcohol and/or substance abuse) on the nuclei density were assessed using Student’s t-tests for unequal sample sizes assuming unequal variance.
Results
FCM measurement of rat cortical cell nuclei
Using an adult rat cerebral cortex, we first examined whether the isotropic fractionator method established by Herculano-Houzel and Lent1 could be used to measure nuclei obtained from frozen unfixed brain tissue and if counting could be accomplished using FCM instead of by visual counting with a hemocytometer. Although the wet weights of the rat cerebral cortices differed considerably across the two studies (1054 mg vs. 770 mg) — perhaps because the tissue used in this study was unfixed, while the tissue used by Herculano-Houzel and Lent1 was fixed — the number of total (7-AAD(+)) and neuronal (NeuN(+)) nuclei counts in the entire adult rat cerebral cortex proved comparable in both studies (Table S1). The co-efficient of variation was <0.13 for the number of 7-AAD(+) nuclei and <0.11 for that of NeuN(+) nuclei. The number of 7-AAD(+) nuclei was also counted on a hemocytometer (77.04 ± 9.82 million cells in the whole cerebral cortex of adult rat [n = 4]) and was found to be almost equivalent to that counted by FCM (Table S1) and that reported by Herculano-Houzel and Lent.1 These findings demonstrated that using FCM with frozen unfixed brain homogenates is a reliable way to estimate the numbers of cells in the brain.
At the same time, the flow cytometer also counted the numbers of olig2(+) nuclei, which most likely represented oligodendroglial cells (OLs and OPCs) in the adult cerebral cortex. Notably, olig2(+) and NeuN(+) populations were mutually exclusive, whereas olig2(+) populations often exhibited two fluorescence peaks: olig2weak(+) and olig2strong(+) populations (Figure 1a-D). Since olig2 appears to be expressed weakly in mature OLs and strongly in the OPCs of adult mouse and human brains,8, 9 the olig2weak(+) and olig2strong(+) populations roughly represent mature OLs and OPCs, respectively.
Effects of PMI, gender, and frozen storage days on the number of nuclei
We then investigated the effects of PMI, gender, and frozen storage days on the number of nuclei in the rat cerebral cortex. The numbers of nuclei were not affected by confounding factors such as PMI (one-way ANOVA, 7-AAD(+), F(2,9)=1,17, P=0.353; NeuN(+), F(2,9)=1,49, P=0.275; olig2(+), F(2,9)=0.169, P=0.848; NeuN(-)/olig2(-), F(2,9)=0.56, P=0.590), gender (unpaired t test, 7-AAD(+), P=0.175; NeuN(+), P=0.516, olig2(+), P=0.164, NeuN(-)/olig2(-), P=0.542), or frozen storage days (unpaired t test, 7-AAD(+), P=0.989; NeuN(+), P=0.809, olig2(+), P=0.124, NeuN(-)/olig2(-), P=0.311) (Figure 1a-E,F,G).
FCM measurement of human gray matter cell nuclei
As a preliminary test, we then applied this novel cell-counting method to the frozen human postmortem FPCs and ITCs obtained from the Stanley Foundation Brain Collection. We carefully dissected out the gray matter of frozen unfixed brain blocks from patients with MDD (FPC, n = 10; ITC, n = 11) and normal controls (FPC, n = 12; ITC, n = 12). These samples contained the entire gray matter, from the brain surface to the border between layer VI and the white matter. All confounding factors were matched across the diagnostic groups (Table S2). As in the rat study, we obtained four different cell populations from human FPC and ITC gray matter in the FCM measurements (Figure S1).
The mean proportions of NeuN(+) nuclei in the gray matter were 38.1% (control FPC), 40.7% (MDD FPC), 38.8% (control ITC) and 36.3% (MDD ITC), which is consistent with those obtained via alternate methods (e.g., 37.5% in the ACC by Nissl staining;11 44.2% in the FPC by Nissl staining;12 36.0% in the rostral OFC by Nissl staining;13 and 41.6% in whole cortical gray matter by an isotropic fractionator method14). This evidence supports our assertion that our sampling of gray matter was relatively precise and that the inadvertent inclusion of white matter in these samples, if any, was minimal.
The olig2strong(+) and olig2weak(+) populations were also investigated in a subgroup of samples in which two peaks of olig2(+) populations could be clearly measured (FPC, control, n = 9; MDD, n = 8; ITC, control, n = 8, MDD, n = 8). The mean proportions of olig2strong(+) nuclei in the gray matter were 5.3% (control FPC), 5.5% (MDD FPC), 6.4% (control ITC) and 5.5% (MDD ITC), which is comparable to the percentages of human gray matter in the temporal and occipital cortices.9
Effects of descriptive confounding factors
No significant correlation was found in the FPC between any nuclei densities and any descriptive confounding factors, such as age at death, PMI, brain pH, brain weight, frozen storage days, age of onset, or duration of disease (Table S3). There was a significant positive correlation between brain weight and NeuN(+) nuclei density in the ITC (r =0.546, P =0.007). ANCOVA was thus performed with this variable as a covariate. Results for the ITC remained statistically insignificant (P =0.851).
Effects of non-continuous confounding factors
The differences in 7-AAD(+) and olig2(+) nuclear densities in the FPC between the MDD and the control groups were still significant, or maintained strong trends, when covaried for hemisphere, gender, or severity of alcohol and/or substance abuse. The differences became more obvious in the left hemispheres. However, due to the small sample size, left-right differences were not conclusive enough in this preliminary study. Previous neuroimaging studies have demonstrated that certain trait abnormalities exist in the left FPC of MDD15 while brain lesion studies, including those of post-stroke depression,16 have revealed mixed results with respect to lesion location (left or right prefrontal cortex) and depression severity.17 Whether or not abnormalities in the left prefrontal cortex (PFC) are the fundamental pathophysiology of MDD remains a controversial issue with no clear consensus.
References
1. Herculano-Houzel S, Lent R. J Neurosci 2005; 25: 2518-2521.
2. Brodmann K. Vergleichende Lokalisationslehre der Grosshirnrinde in ihren Prinzipien dargestellt auf Grund des Zellenbaues. Barth, Leipzig, German, 1909.
3. Torrey EF, Webster M, Knable M, Johnston N, Yolken RH. Schizophr Res 2000; 44: 151-155.
4. Spalding KL, Bhardwaj RD, Buchholz BA, Druid H, Frisen J. Cell 2005; 122: 133-143.
5. Mullen RJ, Buck CR, Smith AM. Development 1992; 116: 201-211.
6. Lu QR, Sun T, Zhu Z, Ma N, Garcia M, Stiles CD et al. Cell 2002; 109: 75-86.