Neocortical and hippocampal neuron and glial cell numbers in the rhesus monkey.код для вставкиСкачать
THE ANATOMICAL RECORD 290:330–340 (2007) Neocortical and Hippocampal Neuron and Glial Cell Numbers in the Rhesus Monkey JEPPE ROMME CHRISTENSEN,1* KAREN BONDE LARSEN,1 SARAH H. LISANBY,2,3 JASON SCALIA,2,4 VICTORIA ARANGO,2,4 ANDREW J. DWORK,2,4,5 AND BENTE PAKKENBERG1 1 Research Laboratory for Stereology and Neuroscience, Bispebjerg University Hospital, Copenhagen, Denmark 2 Department of Psychiatry, Columbia University, New York, New York 3 Brain Stimulation and Therapeutic Modulation Division, New York State Psychiatric Institute, New York, New York 4 Department of Neuroscience, New York State Psychiatric Institute, New York, New York 5 Department of Pathology, Columbia University, New York, New York ABSTRACT The rhesus monkey is widely used as an experimental animal model in the study of brain function and disease. While previous quantitative studies have provided knowledge of regional numbers, little is known of the total neocortical neuron and glial cell numbers in this species. The aim of this study is to establish quantitative norms. We use the optical fractionator and Cavalieri principle to examine the right hemisphere of eight young rhesus monkeys taken from the control group of an ongoing study. Applying these methods to agar-embedded and vibratome-sectioned tissue, we generate estimates of cell numbers and regional volumes of neocortical and hippocampal regions with coefﬁcients of variance (CV) around 10%. The mean unilateral neocortical neuron number is 1.35 3 109 (CV 6 0.10) and the mean unilateral neocortical glial cell number is 0.78 3 109 (CV 6 0.17). Mean unilateral neocortical volume is found to be 8.5 (CV 6 0.10) cm3 after processing, or 19 cm3 when correcting for shrinkage. The neuron/glia ratio is 1.77. The neurons are distributed with 18% in the frontal cortex, 57% in the temporal and parietal cortices, and 25% in the occipital cortex. In the hippocampal subregions, we found unilateral neuron number of 1.72 3 106 (CV 6 0.13) and glial number of 2.25 3 106 (CV 6 0.17) in CA1, and 0.80 3 106 (CV 6 0.27) neurons and 1.05 3 106 (CV 6 0.26) glial cells in CA2–3. Comparisons with related studies show quantitative variation, but also variations in methods and applications. The results are phylogenetically consistent, apart from the neuron/glia ratio, which is remarkably higher than what is found in other species. Anat Rec, 290:330–340, 2007. Ó 2007 Wiley-Liss, Inc. Key words: stereology; primate brain; cell numbers and volume Grant sponsor: the Copenhagen Hospital Corporation Research Council, Hovedstadens Sygehusfaellesskab; Grant sponsor: National Institute of Mental Health; Grant number: MH60884. *Correspondence to: Jeppe Romme Christensen, Laboratory for Stereology and Neuroscience, Bispebjerg University Hospital, Bispebjerg Bakke 23, DK-2400 Copenhagen NV, Denmark. Fax: 45-35-31-64-34. E-mail: email@example.com Ó 2007 WILEY-LISS, INC. Received 29 June 2006; Accepted 13 November 2006 DOI 10.1002/ar.20504 Published online 15 February 2007 in Wiley InterScience (www. interscience.wiley.com). NEOCORTICAL AND HIPPOCAMPAL NEURON The rhesus monkey (Macaca mulatta) has been the species of choice for many experimental models of aging, brain diseases, and new treatment strategies, since it shares with humans many aspects of neuroanatomy and cognitive function (Gallagher and Rapp, 1997; O’Donnell et al., 1999; Peters, 2002). Until now, only a few studies have estimated standards for neuronal and glial cell numbers in the rhesus monkey neocortex. Stereological studies on the rhesus monkey have been either interventional models or focused on speciﬁc regions (Peters et al., 1998). Total neocortical neuronal and glial cell numbers have been assessed in only one previous study for a control group of four individuals (Lidow and Song, 2001). Ethical and economical considerations make rhesus monkey tissue very valuable; therefore, it is of concern to optimize the study design to employ the tissue as efﬁciently as possible. In an ongoing study modeling clinical interventions, we examined the prefrontal cortex and hippocampus of 24 rhesus monkeys, 8 of which constituted a control group. In the present study, we used an optical fractionator design to examine all cerebral cortical tissue from the control group to establish estimates for the normal neocortical and hippocampal cell numbers and regional volumes. Finally, we point out advantages and disadvantages of different techniques such as the optical fractionator for cell counting and the Cavalieri method for volume estimation in different regions of interest. MATERIALS AND METHODS The sample comprised eight rhesus monkeys forming the control group of an interventional study in which all animals, including the controls, received general anesthesia. Half of the sample was female. Mean weight was 3.7 6 0.6 kg. Mean age was 2.8 6 0.46 years. Interventions and sacriﬁce were conducted at New York State Psychiatric Institute in accordance with an approved Institutional Animal Care and Use Committee protocol. All subjects were pathogen-free and were bred and raised at Covance Laboratories, a National Institutes of Health breeding colony in the United States. The animals were housed in the New York State Psychiatric Institute animal care facility. Following standard 10-week quarantine, they were moved into a colony room and socially housed in groups of three. The light cycle was 12 hr a day of light and 12 hr a day of dark. All monkeys were fed a high protein commercial Monkey Chow diet (LabDiet High Protein Monkey Diet Jumbo; Purina Mills), along with daily supplements of fresh fruit. In preparation for daily anesthesia sessions, subjects were sedated in the home cage with intramuscular injections of ketamine (2.5 mg/kg) and xylazine (0.125 mg/ kg). Following transportation into the treatment room, hair on the head was shaved and an intravenous line was placed in the leg. Physiological monitoring at each treatment session included ECG, scalp EEG, pulse oximetry, end-tidal Pco2, and noninvasive blood pressure. Subjects then received methohexital (0.5 mg/kg intravenous bolus) to induce anesthesia. Each subject received a total of 6 weeks of anesthesia sessions once a day, 5 days/week, for a total of 30 sessions. Once a week, subjects also received intramuscular atropine, 0.4 mg/kg. 331 5-bromo-20 -deoxyuridine (BrdU) was administered to the anesthetized monkey through a series of six injections (100 mg/kg, intravenous) over a period of 8 days to label dividing cells for postmortem analysis. Half of the sample received BrdU injections once daily for 5 days during the ﬁnal intervention week, followed by a ﬁnal injection 2 hr prior to sacriﬁce to examine acute effects on proliferation. The other half of the sample received the same number of injections during the 5th intervention week, 6 weeks prior to sacriﬁce, to examine the survival and differentiation of labeled cells using cell type-speciﬁc markers. The animals were sacriﬁced 72 hr following the last session in the treatment phase. Subjects were sedated in the home cage with intramuscular injections of ketamine (2.5 mg/kg) and xylazine (0.125 mg/kg) and were then transported to the perfusion laboratory. Subjects were anesthetized to a surgical depth with sodium pentobarbital (40 mg/kg, intravenous) and heparinized (15 units/kg). After the induction of deep anesthesia and immediately prior to perfusion, a small burr hole was drilled in the skull, providing access to the left prefrontal cortex. Through the burr hole, a 5 mm diameter core of tissue was taken from prefrontal cortex and frozen for later genetic testing. The right hemispheres had a mean weight of 41 g (SD 6 4). As part of the interventional study, we had examined three subdivisions of the right frontal cortex and two subdivisions of the hippocampus (four rightsided and four left-sided). The remaining tissue was preserved for further examination, with the parietal and temporal cortices dissected out together and sectioned in slabs of either 2 or 3 mm. The left hemisphere underwent neuropathological examination in which no abnormal changes were found (Dwork et al., 2004). Tissue Processing and Stereological Design In this study, we use the optical fractionator design and the Cavalieri principle. The optical fractionator design provides a direct and simple method to estimate the total cell number and is in principle unaffected by tissue shrinkage (West and Gundersen, 1990; West et al., 1991; Howard and Reed, 1998; Dorph-Petersen, 2001). The basic principle is to count every cell in a systematic and uniform random sample (SURS) that constitutes a known fraction of the region of interest. In this study, this known fraction is composed of three fractions, namely, a known fraction of the sections of the region of interest, a known fraction of sectional area, and a known fraction of the section thickness. The optical fractionator design was practically applied in a three-step procedure. In step 1, all brains were perfusion-ﬁxed with sodium sulﬁde (0.37%) followed by phosphate-buffered formalin (10%) or paraformaldehyde (4%). After removal from the skull and dura mater, the brains were stored for variable intervals in 10% phosphate-buffered formalin. After delineation (Fig. 1) of the regions of interest and careful dissection, the selected brain region was embedded in agar and cut in the coronal plane into 2 or 3 mm slabs with a random start point within the slab thickness (Fig. 2). A 100 mm thick section was taken from the top of the slab from every second of the frontal slabs, from every second or third of the 2 or 3 mm thick Fig. 1. Regions of interest delineated on the pial surface. Fig. 2. Exhaustive sectioning of the tissue was done in the coronal plane (A). Resulting tissue slabs from a temporal and parietal lobe are shown in B. From the top of every second slab, a 100 mm thick section was cut on a vibratome. Since the tissue slabs were 2 mm thick, ssf ¼ (1/2) 3 (100/2,000) ¼ 1/40. 333 NEOCORTICAL AND HIPPOCAMPAL NEURON TABLE 1. Stereological parameters 1/ssf Ventral prefrontal Dorsal prefrontal Posterior frontal Temporal and parietal 2 mm. Temporal and parietal 3 mm. Occipital CA1 CA 2-3 40 40 40 40 90 80 20 20 Step-length mm2 R sections Area (counting frame) mm2 R disectors R Qneuron R Qglial 3 3 3 3 3 3 3 3 7 8 9 10 6 7 9 9 1554 1545 1546 1328 1327 1326 1301 1344 82 134 210 68 69 23 146 99 155 288 391 235 221 139 185 160 136 196 364 116 166 84 258 224 6000 6000 6000 6000 4000 4000 500 350 6000 6000 6000 6000 4000 4000 500 350 temporal and parietal tissue slabs, and from every fourth of the occipital neocortical tissue slabs, starting randomly at slab 1–2, 1–3, and 1–4, respectively. Every section thus represents a known ﬁxed fraction of the tissue, the fraction called the section sampling fraction (ssf; Fig. 2), in this case respectively 1/40, 1/90, or 1/80 (see Table 1 for further fractionator parameters). For the hippocampal region, a section was cut from every tissue slab, giving an ssf of 1/20. To monitor the block advance on the vibratome, one tissue block from each animal was chosen at random to be cut exhaustively into 100 micron thick sections, which conﬁrmed that a 2 mm block provided 20 sections. In step 2, all sections were mounted on glass slides (Superfrost Plus) coated with an aqueous solution of gelatin (4.5%) and chromealum (chromium potassium sulfate; 4.0%) and air-dried at room temperature. This procedure was necessary to improve adhesion of the 100 mm thick sections to the slides. Staining was done with a modiﬁed Vogt Cresyl Violet, which provided the best staining of the relevant cells. Subsequently, the stained sections were dehydrated in graded concentrations of alcohol, followed by xylenes, and cover-slipped with Pertex glue. Microscopic examination was done using a high-resolution microscope (1003 oil immersion, NA ¼ 1.4, and ﬁnal magniﬁcation ¼ 3,2003) connected to a computer via a video camera. A motor stepper measured the movements in the x – y directions while a Heidenhain microcator measured the movements in the z-direction with a precision of 0.5 mm. A counting frame was applied to the tissue using CAST software (Fig. 3A). The counting frame constitutes two of the three dimensions of the optical dissector. Its area (Table 1) represents a fraction of the area made up of the step lengths in the x- and y-directions (Fig. 3B). The area sampling fraction (asf) is calculated from: asf ¼ aðframeÞ aðx; y stepÞ Hsf is calculated as the q weighted mean section thickness to compensate for differences in section thickness in and among sections and correlated to the local amount of particles sampled (Dorph-Petersen et al., 2001): h hsf ¼ tQ where tQ P ðti q i Þ i ¼ P qi i where ti is the local section thickness centrally in the ith counting frame with a dissector count of qi. The total cell number estimate (N) is calculated from the equation N¼ X 1 1 1 3 3 3 Q ssf asf hsf where SQ is the sum of all cells counted in all dissectors in a region. Counting Principles For all stereological methods, it is critical that all particles in each sample are counted only once and with the same probability, which is provided by the dissector (Sterio, 1984). With the optical dissector, it is possible to dissect optically the chosen sample using the focal plane, which is moved through the thick section in the z-axis. After having established a constant density within the dissector height, all sampled particles that come into focus inside the counting frame are counted and added to SQ, provided they do not touch the exclusion lines. An upper guard area of 5 mm was used in order to avoid loss of sampled particles due to tissue preparation. Cavalieri Principle where a(frame) is the area of the counting frame and a(x,y-step) is the area represented by the step length in the x–y direction. In step 3, the mean thickness (t) of the sections was determined from measurements made in every fourth dissector measured with the microcator. The height (h) forms the third dimension of the optical dissector and is chosen as a predetermined constant, 15 mm in this study. The height of the dissector relative to the mean section thickness is the height sampling fraction (hsf; Fig. 4). To estimate the formalin-ﬁxed nonprocessed volume of the frontal cortex, a counting grid was laid randomly over the tissue slabs: X Volref ¼ t 3 aðpÞ 3 P where t is the slab thickness, a(p) is the unit area of the x–y grid, and SP is the sum of counted points within the evaluated region. 334 CHRISTENSEN ET AL. Fig. 3. A shows a schematic application of the optical fractionator and Cavalieri method to the cortex of a section. All counting frames were included, but only corner points hitting cortex were included in the calculation of the processed tissue volume. B shows a screen shot with the counting frame applied to the tissue. Red lines are exclusion lines, and green lines are inclusion lines. Each counting frame applied to the tissue constitutes a known fraction of the section. A ﬁxed step length (big squares in A), chosen in accordance with the desired precision of the estimates, determines the number of counting frames (small squares in A) generated from each sample. The area section fraction (asf) is the area of the counting frame divided by the area of the squares made up of the step length in the x- and y-directions. 335 NEOCORTICAL AND HIPPOCAMPAL NEURON Fig. 4. The height sample fraction (hsf) is the fraction of the height of the optical dissector (predetermined 15 mm) related to the mean thickness of the sections. Fig. 5. Coronal section through the hippocampal body showing the regional structures and the bilaminar nature of hippocampus. The picture was taken with a 23 objective, while the actual delineation was done with a 103 objective. The Cavalieri estimate of the processed volume was obtained by counting the upper right corner of all counting frames touching the region of interest. When estimating the ﬁxed but otherwise unprocessed volume, the actual section thickness was substituted with the unprocessed section thickness (100 mm). Volume estimates based on counting points from counting frames are thus given by Volref ¼ t 3 aðx; y stepÞ 3 X 1 3 P ssf 336 CHRISTENSEN ET AL. Delineation The neocortical regions were painted on the pial surface before dissection. Boundaries of the frontal region were deﬁned as the central sulcus and Sylvian ﬁssure, while the dorsal prefrontal and ventral prefrontal regions were deﬁned by the principal and arcuate sulci (Fig. 5) (Barbas and Pandya, 1989; Geyer et al., 2000; Dombrowski et al., 2001). The temporal and parietal cortices were counted together. Their boundaries were deﬁned as the central sulcus, the anterior part of the Sylvian ﬁssure, the parietal-occipital sulcus, and the lunate sulcus. Occipital boundaries were the parietaloccipital and lunate sulci. Archicortical and neocortical transition zones are found in the cingulate gyrus, uncus, and entorhinal cortex. Delineation was done in every examined slice with a 43 objective according to cytoarchitectonic criteria (Amaral et. al., 1987; Vogt et al., 1987; Suzuki and Amaral, 1994, 2003a, 2003b; Gazzaley et al., 1997; Merrill et al., 2000). In the hippocampus, we examined the subregions CA1 and CA2–3 as part of the original study. The deﬁnition of the borders CA1 and CA2–3 was based on criteria from studies on humans (Duvernoy, 1988; West et al., 1988, 1994; Simic et al., 1997; Harding et al., 1998) and rhesus monkeys (Keuker et al., 2003). In most respects, the cytoarchitectonical organization of these ﬁelds are the same within these species (Amaral and Inausti, 1990). Delineation was done under 23 and 103 objectives (Fig. 5). Differentiation of Neurons and Glial Cells While it is usually straightforward to distinguish large- and middle-sized neurons from glial cells, the distinction between small neurons and large glial cells can be challenging, a situation that especially is encountered in the occipital cortex. The following criteria were used as characteristic for neurons: a centrally located nucleolus, a distinctive nucleus, visible cytoplasm, presence of dendritic processes, and larger cell body size. Glial cells were identiﬁed by the following criteria: heterochromatin clumps, sparse cytoplasm, and smaller cell body size (Selemon et. al., 1999; Lidow and Song, 2001; Stark et al., 2004; Jelsing et al., 2006). Statistical Analysis Coefﬁcient of error (CE) for estimates of cell numbers of the individual cortices was calculated from the formula qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ CEðNÞ ¼ CE2 þ CEðtÞ2 where CE is obtained from the formula pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ Varsurs þ Noise P CE ¼ Q where noise is the sum of counted particles and VarSURS is the estimator variance under (SURS) (Gundersen and Jensen, 1987; Gundersen et al., 1999). The VarSURS(N) is obtained from the formula VarSURS ðNÞ ¼ ð3ðA NoiseÞ 4B þ C 240 where the systematic section series of particle count are denoted f1,f2 . . . fn, and A¼ n X f 2i ; B ¼ i¼1 n1 X f i f iþ1 and C ¼ i¼1 n2 X f i f iþ2 i¼1 The CE for the volume estimates is a function of the point-counting noise and the variance of Sarea for a given direction of sectioning under SURS: X X CE P ¼ Noise þ VarSURS area ; where Noise ¼ 0:07247 3 qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ X b pﬃﬃﬃ 3 n3 P; a and (b/Ha) is a constant reﬂecting average proﬁle shape, n is the number of sections, and SP is the sum of points. VarSURS ¼ X area ð3 3 ðPi 3 Pi NoiseÞ 43Pi 3 Piþ1 þ Pi 3 Piþ2 Þ 240 where Pi is the number of points counted in one section and Piþ1 is the number of points counted on the following section and so forth. The interindividual variability is shown as the standard deviation (SD) and the coefﬁcient of variation (CV). In favor of optimizing the efﬁciency and precision of the study, we adjusted the step length in order to get a CE of approximately 10%. On average, we counted 242 neurons and 179 glial cells per cortical region, resulting in coefﬁcients of error for the cell estimates ranging from 6% to 12%, with averages of 8.3% and 10.7% for neurons and glial cells, respectively. A precision in this range is reasonable since the interindividual variation, the CV, ranged from 11% to 29%. The biological variance of the individual volume estimates of the different regions showed variation in the range of CV ¼ 8%–27%, the variation in the frontal subregions being somewhat higher (CV ¼ 21%–29%). RESULTS Unilateral neocortical numbers are presented in Table 2 and unilateral hippocampal numbers in Table 3. Note that processed volumes are listed, which is reﬂected in the densities. The mean frontal cortical volume estimated from point counting on the ﬁxed but unprocessed tissue was 7,196 mm3 (SD 6 878). The volume calculated from the sum of corner points in the processed tissue multiplied by the tissue thickness of 100 mm was 6,522 mm3 (SD 6 508). The difference of about 10% is a product of a comparison of two estimates and some tissue deformation in the x – y axes. Note that this has no inﬂuence on the cell number estimates. 1,77 0,16 0,077 1,73 0,32 0,04 0,113 2,15 0,16 0,02 0,061 1,01 0,08 0,02 0,027 0,032 1,68 1,14 0,09 0,02 0,18 0,10 0,02 0,049 1,21 0,09 0,01 2679 545 0,20 452 97 0,21 908 266 0,29 1319 293 0,22 4875 582 0,12 1061 84 0,08 8499 807 0,10 0,037 Neuron/glia ratio Neuron density (106/mm3) 337 Tissue shrinkage in the z-axis was calculated to be 54% on the basis of mean section thickness and the original section thickness of 100 mm. Accordingly, when correcting for tissue shrinkage, we ﬁnd total neocortical volume of about 18,500 mm3 and a corrected neuronal density of approximately 73,000 neurons/mm3. Total neocortical neuron number was approximately 10% greater in males than in females (Table 4), but this was not statistically signiﬁcant (t ¼ 1.8; df ¼ 6; P ¼ 0.13). DISCUSSION Stereological Design The optical fractionator provides an efﬁcient method for estimating neocortical and hippocampal cell numbers and the Cavalieri principle a method for estimation of regional volumes. The methods were easily applied in an agar and vibratome design, with the purpose of saving tissue for future evaluation by maintaining as much of the tissue integrity as possible. The sampling in each region was optimized according to the precision of the estimate and the required work load. 0,097 0,108 0,105 0,059 0,072 0,091 0,109 0,120 0,122 0,060 0,079 0,093 0,084 0,094 0,085 0,058 0,063 0,089 Total neocortex Occipital Temporal and parietal Posterior frontal Ventral prefrontal Dorsal prefrontal Average St Dev CV Average St Dev CV Average St Dev CV Average St Dev CV Average St Dev CV Average St Dev CV Average St Dev CV 233 31 0,13 41 9 0,23 85 13 0,15 106 19 0,18 782 131 0,17 334 44 0,13 1349 140 0,10 0,071 196 40 0,21 36 6 0,17 52 13 0,24 108 26 0,24 380 94 0,25 200 57 0,29 776 129 0,17 0,078 429 67 0,16 77 14 0,18 138 23 0,17 214 42 0,20 1162 170 0,15 534 94 0,18 2125 202 0,10 0,075 Considerations on Tissue Processing Frontal CE Processed Volume (mm3) CE Total cell (106) CE Glia (106) CE Neurons (106) TABLE 2. Unilateral estimates of cell numbers, volume, density and neuron/glia ratio in neocortical regions NEOCORTICAL AND HIPPOCAMPAL NEURON Generally, ﬁnal section thicknesses of about 25 mm or more are sufﬁcient in optical dissector designs. Due to substantial tissue shrinkage in the z-axis after sectioning, vibratome sections should typically be cut at 70–100 mm. The agar and vibratome design was chosen because it gives wider possibilities for further investigations, e.g., cell volume or immunohistochemical staining, on the unused slabs. Cryostat preparations have similar properties to the agar and vibratome preparation. In the pilot phase, one brain was cut into consecutive parafﬁn sections. This method offers less laboratory work and the opportunity for applying the physical dissector on thin sections. However, parafﬁn preparations exclude the possibilities of future examinations of, e.g., cell size, since it results in substantial shrinkage in all three axes and is only recommendable for cell counting. Finally, we found parafﬁn unsuitable for hippocampal preparations, since its appendage-like morphology combined with the small cracks sometimes seen in parafﬁn sections resulted in artifacts that made several sections unusable. Methacrylate resin preparations result in only limited tissue shrinkage in all three dimensions, but impose limitations on the size of the sections and the use of immunohistochemical staining. Anatomical Considerations The prefrontal regions are heterogeneous, with subregions whose borders are deﬁned cytoarchitectonically, but map descriptions vary considerably (Barbas and Pandya, 1989; Dombrowsky et al., 2001). In the present study, the boundaries were macroscopically deﬁned by the principal sulcus and arcuate sulci, which vary more than the central sulcus and Sylvian ﬁssure delimiting the frontal lobe. Our results reﬂect this circumstance, with higher coefﬁcients of variance for cell numbers in each subregion than in the total frontal region. The volume of the prefrontal cortices varies with coefﬁcients of variance between 20% and 29%, which is consistent with the ﬁndings of O’Donnell et al. (1999). 338 CHRISTENSEN ET AL. TABLE 3. Unilateral estimates of cell numbers, volume, density and neuron/glia ratio in hippocampal subregions CA1 and CA2-3 CA1 CA2-3 Neurons (106) CE Glial cells (106) CE Volume (mm3) CE Density (106/mm3) 1,72 0,23 0,13 0,80 0,22 0,27 0,080 2,25 0,39 0,17 1,05 0,27 0,26 0,072 23,45 3,76 0,16 7,40 2,76 0,37 0,037 0,07 0,033 0,11 Average St Dev CV Average St Dev CV 0,052 0,046 TABLE 4. Neocortical estimates of cell numbers, volume, and neuron/glia ratio In relation to gender Gender Neuron (mill.) Glia (mill.) Total (mill.) Volume (mm3) Neuron/glia ratio F F F F Average M M M M Average 1446 1304 1202 1136 1272 1373 1545 1302 1479 1425 898 620 628 839 746 748 758 958 758 806 2344 1924 1830 1975 2018 2121 2303 2260 2237 2230 – 8270 8458 8448 8392 7548 8854 7862 10154 8604 1,61 2,10 1,91 1,35 1,75 1,84 2,04 1,36 1,95 1,80 The occipital region was also delineated according to macroscopic criteria. In the examination of caudal temporal sections, we occasionally (one or two dissectors in a few brains) encountered microscopically characteristic occipital cortex, reﬂecting that the macroscopic boundaries are not always entirely precise. Since it would contribute only marginally to the occipital cell number, in order to maintain simple criteria, we attributed the cells to the temporal region. Comparisons With Related Studies In an interesting study, Lidow and Song (2001) examined neocortical cell numbers in the rhesus monkey. Here the authors found a mean of 295 3 107 (SD 6 90) neurons and 241 3 107 (SD 6 68) glial cells in the left hemisphere neocortex of the monkeys forming the control group. These numbers are greater than our estimates by a factor of 2 to 3. Neocortical volume was calculated on data from celloidin processed tissue, with an estimated tissue shrinkage of 10%–16%, and was reported to be approximately 21 cm3 (SD 6 2), similar to what we ﬁnd. The neuron/glia ratio was 1.2, compared with a ratio of 1.8 in the present study, possibly reﬂecting differences in application of differentiation criteria or a difference between the populations. Although the Lidow and Song (2001) study used modern stereological principles, the optical dissector and point counting, we ﬁnd differences in the applied methods that might explain some of the discrepancy in the cell numbers. First, counting was done with a counting frame extending the whole depth of the neocortex, while it is unclear how the whole cortical depth can always be identiﬁed in the tissue bars. One would expect, for example, that some of the bars are tangential to the cortical surface in an SURS sample. Second, the cell densities were obtained without guard areas on both sides of the optical dissector, which may result in a small bias. Further, the application of the original formula of CE of Gundersen and Jensen (1987) cannot have been applied appropriately, since counting literally hundred thousands of cells will result in a smaller CE than reported. Finally, the surface area was not calculated from unbiased stereological principles (Howard and Reed, 1998). However, these methodological differences are unlikely to explain a two- to three-fold discrepancy in the cell numbers. The possibility of differences between breeding strains of the monkeys would be an interesting subject for further investigation. Dombrowski et al. (2001) found, in a cryosection design, densities in different prefrontal areas ranging from 39,000 to 59,000 neurons/mm3 and 43,000 to 57,000 glia/mm3 calculated on the basis of ﬁxed unprocessed volume. They found a neuron/glia ratio of about 1. Concordantly, we ﬁnd mean unprocessed densities (by correcting for shrinkage) of 43,000 neurons/mm3 and of 30,000 glia/mm3 in the prefrontal cortex and a neuron/ glia ratio of 1.4. Selemon et al. (1999) examined area 46 of prefrontal cortex in rhesus monkey brains and found densities of 133,000 neurons/mm3 and 92,000 glia/mm3 in an optical dissector design, where the y-axis of the counting frame extended the whole cortical depth. The tissue was celloidin-embedded, and no correction for tissue shrinkage is reported (Selemon et al., 1995). We ﬁnd mean prefrontal densities in processed tissue of approximately 95,000 neurons/mm3 and 70,000 glia/mm3 markedly lower than the ﬁndings of Selemon et al. Dombrowski et al. (2001) report an unprocessed density of about 55,000 neurons/ mm3 in area 46. Although celloidin processing, used by Selemon et al. (1995), is subject to tissue shrinkage, this is limited, and failure to correct for shrinkage is unlikely to account for the whole discrepancy. Biological variation, which might be increased by the use of different NEOCORTICAL AND HIPPOCAMPAL NEURON strains, also may explain some of the discrepancy. The use of different stereological designs can contribute to the discrepancy, as noted above. Sampling the whole cortical depth might give methodological problems that are difﬁcult to evaluate. Finally, the study by Selemon et al. (1995) examines only probes located in gyral crests, which, according to a recent study from Hilgetag and Barbas (2005), contain signiﬁcantly more neurons than sulcal areas. This way of sampling, while potentially systematic and random, is certainly not uniform, and it can result in bias. Keuker et al. (2003) examined the hippocampus of the rhesus monkey using stereological methods for quantifying the neurons, but not the glial cells. The investigators found a mean of 1.18 3 106 (CE 5.5%) in CA1 and 0.6 3 106 (CE 6.2%) in CA2–3. These numbers are lower than the present ﬁndings. The difference may be due to biological variance or to problems in distinguishing small neurons from glia. Comparing the present study with studies on humans (Pakkenberg and Gundersen, 1997; Pakkenberg et al., 2003) suggests that the rhesus monkey has about 8 times fewer neocortical neurons and 25 times fewer glial cells distributed in a neocortical volume 13 times smaller. Analogous to the human studies, our study ﬁnds a tendency toward sex difference in cell numbers and volume. Considering the small population of the present study, this ﬁnding would be interesting to conﬁrm in future studies. From a phylogenetic perspective, the results give rise to some considerations. Frontal neurons constitute 18% of the neurons in the rhesus monkey neocortex, whereas this part is 34% in the human neocortex, an expected ﬁnding since frontal cortex is phylogenetically the youngest part of the neocortex. Human hippocampus has approximately four times as many neurons in CA1 and twice as many in CA2–3. The relatively small difference in neuron number between human and rhesus monkey hippocampus is phylogenetically consistent with hippocampus being part of the archicortex. The neocortical neuron/glia ratio is 1.7 in rhesus monkeys, compared with 0.6 in humans (Pakkenberg et al., 2003), 0.45 in Göttingen minipig (Jelsing et al., 2006), and 0.13 in the minke whale (Eriksen and Pakkenberg, unpublished data). The Göttingen minipig has a similar brain weight to the rhesus monkey (80–100 g), but fewer neocortical neurons by a factor of 8 (Jelsing et al., 2006). In the rat, Mooney and Napper (2005) found a ratio of 2.0 in neocortex. Herculano-Houzel and Lent (2005), using the novel approach of counting nuclei in a homogenate of whole rat brain, obtained a ratio of neurons to other cells of 0.7; since white matter was included, the lower value is not surprising. The animals in this report were exposed, as controls for an intervention study, to repeated episodes of general anesthesia. While this exposure is unlikely to have affected our results, we cannot deﬁnitively rule out that possibility, since the impact of daily anesthesia on neuronal numbers has never been evaluated. However, neuronal loss from anesthesia is unlikely in the present study. Ventilation and monitoring of vital signs and oxygenation were performed continuously during anesthesia, and thorough neuropathological examination, reported previously (Dwork et al., 2004) and extended to include all of the animals in this study, revealed no evidence of 339 neuronal damage or its aftermath. The anesthesia sessions were distributed over the 10 weeks prior to sacriﬁce; had they resulted in signiﬁcant neuronal loss, this would have been accompanied by a correspondingly signiﬁcant cortical gliosis, which was not observed qualitatively by staining many sections from each case for glial ﬁbrillary acidic protein, nor quantitatively in the current study [e.g., by comparison with the glial counts of Lidow and Song (2001)]. In summary, our ﬁndings provide new knowledge about the quantitative structure of the normal young rhesus monkey brain. Though based on a small population, our material sufﬁces to generate estimates of the mean cell numbers and regional volumes with reasonable coefﬁcients of variance, taking the normal biological variation of these quantities into account. A remarkable ﬁnding is a neuron/glia ratio of 1.7, which differs from other species, primates as well as nonprimates. The implication of this ﬁnding remains unclear. Another ﬁnding, though not statistically signiﬁcant, is the tendency toward a difference in neuron number between the sexes. Comparing our results with previous studies, we ﬁnd considerable variations in the results, but also a considerable degree of methodological variation. LITERATURE CITED Amaral DG, Insausti R, Cowan WM. 1987. The entorhinal cortex of the monkey: I, cytoarchitectonic organization. J Comp Neurol 264:326–355. Amaral DG, Inausti R. 1990.Hippocampal formation. In: Paxinos G, editor. The human nervous system. San Diego, CA: Academic Press. p 711–755. Barbas H, Pandya DN. 1989. Architecture and intrinsic connections of the prefrontal cortex in the rhesus monkey. J Comp Neurol 286:353–375. Dombrowski SM, Hilgetag CC, Barbas H. 2001. Quantitative architecture distinguishes prefrontal cortical systems in the rhesus monkey. Cereb Cortex 11:975–988. Dorph-Petersen KA, Nyengaard JR, Gundersen HJ. 2001. Tissue shrinkage and unbiased stereological estimation of particle number and size. J Microsc 204:232–246. Duvernoy HM. 1988. The human hippocampus: an atlas of applied anatomy. München: J.F. Bergmann Verlag. Dwork AJ, Arango V, Underwood M, Ilievski B, Rosoklija G, Sackeim HA, Lisanby SH. 2004. Absence of histological lesions in primate models of ECT and magnetic seizure therapy. Am J Psychiatry 161:576–578. Gallagher M, Rapp PR. 1997. The use of animal models to study the effects of aging on cognition. Annu Rev Psychol 48:339–370. Gazzaley AH, Thakker MM, Hof PR, Morrison JH. 1997. Preserved number of entorhinal cortex layer II neurons in aged macaque monkeys. Neurobiol Aging 18:549–553. Geyer S, Matelli M, Luppino G, Zilles K. 2000. Functional neuroanatomy of the primate isocortical motor system. Anat Embryol 202:443–474. Gundersen HJ, Jensen EB. 1987. The efﬁciency of systematic sampling in stereology and its prediction. J Microsc 147:229–263. Gundersen HJ, Jensen EB, Kieu K, Nielsen J. 1999. The efﬁciency of systematic sampling in stereology: reconsidered. J Microsc 193: 199–211. Harding AJ, Halliday GM, Kril JJ. 1998. Variation in hippocampal neuron number with age and brain volume. Cereb Cortex 8:710– 718. Herculano-Houzel S, Lent R. 2005. Isotropic fractionator: a simple, rapid method for the quantiﬁcation of total cell and neuron numbers in the brain. J Neurosci 25:2518–2521. Hilgetag CC, Barbas H. 2005. Developmental mechanics of the primate cerebral cortex. Anat Embryol 210:411–417. 340 CHRISTENSEN ET AL. Howard V, Reed MG. 1998. Unbiased stereology: three-dimensional measurement in micropscopy. New York: Springer Verlag. Jelsing J, Nielsen R, Olsen AK, Grand N, Hemmingsen R, Pakkenberg B. 2006. The postnatal development of neocortical neurons and glial cells in the Gottingen minipig and the domestic pig brain. J Exp Biol 209:1454–1462. Keuker JIH, Luiten PGM, Fuchs E. 2003. Preservation of hippocampal numbers in aged rhesus monkeys. Neurobiol Aging 24: 157–165. Lidow MS, Song ZM. 2001. Primates exposed to cocaine in utero display reduced density and number of cerebral cortical neurons. J Comp Neurol 435:263–275. Merrill DA, Roberts JA, Tuszynski MH. 2000. Conservation of neuron number and size in entorhinal cortex layers II, III, and V/VI of aged primates. J Comp Neurol 422:396–401. Mooney SM, Napper RM. 2005. Early postnatal exposure to alcohol reduces the number of neurons in the occipital but not the parietal cortex of the rat. Alcohol Clin Exp Res 29:683–691. O’Donnell KA, Rapp PR, Hof PR. 1999. Preservation of prefrontal cortical volume in behaviorally characterized aged macaque monkeys. Exp Neurol 160:300–310. Pakkenberg B, Gundersen HJ. 1997. Neocortical neuron number in humans: effect of sex and age. J Comp Neurol 384:312–320. Pakkenberg B, Pelvig D, Marner L, Bundgaard MJ, Gundersen HJ, Nyengaard JR, Regeur L. 2003. Aging and the human neocortex. Exp Gerontol 38:95–99. Peters A, Morrison JH, Rosene DL, Hyman BT. 1998. Are neurons lost from the primate cerebral cortex during normal aging? Cereb Cortex 8:295–300. Peters A. 2002. Structural changes that occur during normal aging of primate cerebral hemispheres. Neurosci Biobehav Rev 26:733– 741. Selemon LD, Lidow MS, Goldman-Rakic PS. 1999. Increased volume and glial density in primate prefrontal cortex associated with chronic antipsychotic drug exposure. Biol Psychiatry 46:161–172. Simic G, Kostovic I, Winblad B, Bogdanovic N. 1997. Volume and number of neurons of the human hippocampal formation in normal aging and Alzheimer’s disease. J Comp Neurol 379:482–494. Stark AK, Uylings HB, Sanz-Arigita E, Pakkenberg B. 2004. Glial cell loss in the anterior cingulate cortex, a subregion of the prefrontal cortex in subjects with schizophrenia. Am J Psychiatry 161:882–888. Sterio DC. 1984. The unbiased estimation of number and sizes of arbitrary particles using the dissector. J Microsc 134:127–136. Suzuki WA, Amaral DG. 1994. Perirhinal and parahippocampal cortices of the macaque monkey: cortical afferents. J Comp Neurol 350:497–533. Suzuki WA, Amaral DG. 2003a. Perirhinal and parahippocampal cortices of the macaque monkey: cytoarchitectonic and chemoarchitectonic organization. J Comp Neurol 463:67–91. Suzuki WA, Amaral DG. 2003b. Where are the perirhinal and parahippocampal cortices? a historical overview of the nomenclature and boundaries applied to the primate medial temporal lobe. Neuroscience 120:893–906. Vogt BA, Pandya DN, Rosene DL. 1987. Cingulate cortex of the rhesus monkey: 1. Cytoarchitecture and thalamic afferents. J Comp Neurol 262:256–270. West MJ, Coleman PD, Flood DG. 1988. Estimating the number of granule cells in the dentate gyrus with the dissector. Brain Res 448:167–172. West MJ, Gundersen HJG: 1990. Unbiased stereological estimation of the number of neurons in the human hippocampus. J Comp Neurol 296:1–22. West MJ, Slomianka L, Gundersen HJG. 1991. Unbiased stereological estimation of the total number of neurons in the subdivisions of the rat hippocampus using the optical fractionator. Anat Rec 231:482–497. West MJ, Coleman PD, Flood DG, Troncoso JC. 1994. Differences in the pattern of hippocampal neuronal loss in normal ageing and Alzheimer’s disease. Lancet 344:769–772.