Inter-specifically, mammalian species having larger brains built of more numerous neurons have higher cognitive abilities (CA), but at the expense of higher metabolic costs. It is unclear, however, how this pattern emerged, since evolutionary mechanisms act intra-specifically, not inter-specifically. Here we tested the existence of the above pattern at the species level in the hippocampus- the brain structure underlying CA. We used the experimental evolution model system consisting of lines of laboratory mice divergently selected for basal metabolic rate (BMR) - a trait implicated in the evolution of brain size, its metabolic costs and CA. Selection on BMR did not affect hippocampus size. Yet, the high BMR mice had superior CA and manifested increased neuronal density, higher cytochrome C oxidase density (indexing metabolic costs of neuronal activity) and dendritic spine density (indexing connectivity between neurons). Thus, our study calls into question generality of patterns of the evolution of CA apparent interspecifically. At the species level increased CA may arise through re-arrangement of the architecture and function of neurons, without a conspicuous increase of their size.

Animals

We used 3-4-month-old Swiss-Webster female mice from the selection experiment carried out at the Faculty of Biology, University of Bialystok. We divergently select mice for high/low body mass-corrected Basal Metabolic Rate (BMR). In this experiment, we maintain two non-replicated line types: the high-BMR (H-BMR) and low-BMR (L-BMR) line type. Individual mice used in this study were randomly drawn from a stock of animals not qualified for further selection.

The divergence achieved in the primarily selected trait- BMR is sufficiently large to be confidently attributed to the results of the selection, rather than to genetic drift.. Still, however, a straightforward statistical comparison of both selection line types with respect to the traits that are not directly selected, such as those analyzed in our study is questionable, because in their case one cannot rule out that the differences between the L-BMR and H-BMR mice were due to random effects unrelated the applied selection. To remedy this problem, we also used female Swiss-Webster mice from four random-bred lines maintained concurrently with the selection on BMR. Mice of these lines formed the RB line type, serving as a reference to the divergence of traits analyzed therein in animals selected for BMR.

Each mouse used in our study was randomly selected from a separate family belonging to a particular line. For information on the number of animals used in specific analyzes see electronic Supplementary Materials.

Perfusion and tissue preparation

Mice were deeply anesthetized with sodium pentobarbital (133.3 mg/ml), pentobarbital (26.7 mg/ml), and morbital, perfused transcardially with PBS (4°C, 50 ml per animal) and subsequently with 4% PFA (4°C, 70 ml per animal). Brains were dissected, postfixed in 4% PFA (o/n, 4°C), and cryoprotected in 30% sucrose solution (4°C) for at least 3 days. Coronal sections containing the hippocampus (coordinates following Paxinos and Franklin’s atlas, 2001) were cut with the use of a cryostat (-20°C) and kept in PBS (4°C) until further handling.

Hippocampus size, hippocampal pyramidal area, and neuronal density estimated by Nissl staining

 We used 14, 3, and 3 animals from the RB, L-BMR, and H-BMR line types, respectively. Out of 14 animals of the RB line type, 5, 3, 3, and 3 belonged to the respective replicated lines forming this line type.

Transverse (coronal) 40 μm cross-sections of the hippocampus positioned -1.46; -1.82; -2.3 and -2.7 mm from bregma (thereafter Position, coordinates according to Paxinos and Franklin were mounted on gelatin-coated slides and then dried, dehydrated, and stained according to the standard Nissl method: slides were washed in 0.1% cresyl violet solution (cresyl violet 0.1g; distilled water 50 ml; acetate buffer 0.2 M pH 3.6 50 ml) for 3-10 min and then dehydrated through graded alcohols (70, 95, 100, 100%) and xylenes. Next, the slides were cleared in xylol and bonded with coverslip by means of DePeX (Serva Electrophoresis^TM).

The microscope image was captured using a light microscope Leica DM1000 LED connected to a Leica ICC50 camera, at 400x and 40x magnification for neuron counting and hippocampal measurements, respectively. The total cross-sectional surface area (µm²) of the hippocampus (being a proxy of its size), as well as the pyramidal cell layer area (µm², sensu ‘superficial’ cell layer) were traced manually and then measured at each Position by means of CellSens Dimension Desktop (Olympus Corp., Japan).

Neuronal density was estimated in the pyramidal hippocampal cell layer in CA1, CA2, and CA3 regions of the right and left side, in the same cross-sections as used for the estimation of the area occupied by the pyramidal cell layer. Neurons were counted inside rectangle frames of 3000 μm² (counting frame) localized at random in the central part of each of the regions. In sum, four frames (each representing one Position) were analyzed for each hippocampal region, then measurements were averaged within a given region.

To obtain an unbiased estimate of the neuron number within the frame we followed guidelines formulated by Gundersen (44), that is, only neurons crossed by the bottom and right-hand side borders of the frame were taken into account. Conversely, all neurons crossed by the left-hand side and upper borders of the frame were not accounted for.

Neuronal density estimation by DAPI staining

 To validate the estimates of neuronal density in the pyramidal hippocampal layer obtained by means of Nissl staining, we also used DAPI staining. In these analyses, we used a separate set of 3, 3, and 4 animals randomly chosen from the RB, H-BMR, and L-BMR lines types, respectively.

Forty μm coronal sections of hippocampus, positioned -2.06 mm from bregma (coordinates following Paxinos and Franklin’s atlas, 42), were mounted on gelatin-coated slides covered with Vectashield mounting medium containing DAPI.  Next, the sections were photographed by means of a Zeiss LSM 780 confocal microscope, using 63 times magnifying lens for high-resolution imaging (Plan Apochromat 63x/1.4 Oil DIC, Zeiss) at identical capturing parameters for all images.

We assumed that neuronal density was equivalent to the density of neuronal nuclei positively stained with DAPI. Nuclei were counted by means of ImageJ (NIH) software, allowing for the discrimination of neuronal and non-neuronal nuclei by their size and shape. The density was estimated in the pyramidal hippocampal cell layer, in CA1, and CA3 regions of the right and left side of the hippocampus. Due to technical problems with staining, we were unable to collect reliable density estimates for the CA2 region. Eight non-overlapping frames of 2890 μm² (counting frames) were picked at random in the central part of each of the regions. As in the case of the Nissl staining, we obtained unbiased estimates of the neuron following guidelines formulated by Gundersen.

Anti-GFAP immunostaining

We measured hippocampal astrocyte/neuron ratio (estimated as the proportion of the surface area occupied by astrocytes immunostained with Anti-GFAP vs. neurons) in a separate set of 18, 3, and 4 animals coming from the RB, H-BMR and L-BMR lines types, respectively.

Sections from the perfused brain were blocked with NDS (normal donkey serum). After the wash, sections were incubated in 0.1% Triton X-100 in PBS (PBST) for 15 minutes. Then, after preincubation in NDS-buffer (5% NDS in PBST) for 1 hour, the chicken anti-GFAP (Abcam, 1:800, Cambridge, MA, USA,) antibody (Ab) was added for overnight incubation at 4°C. After washing cycles in PBST (3 x 1 min), sections were incubated for 2 hours at room temperature with secondary antibody 647 Alexa Fluor (Jackson ImmunoResearch, West Grove, PA) at 1:400 concentration to detect GFAP. After the last round of washing with PBST (3 x 5 min, RT) slices were mounted (Fluoromount-G, Southern Biotech) on polysine slides. To control for non-specific binding and imaging artifacts ‘no primary antibody’ and ‘no secondary antibody’ immunostainings were carried out. Imaging was performed using an Olympus microscope (Olympus Corporation, Japan) with RGB Camera Hamamatsu ORCA Flash4.0 V2 and at 20x magnification for astrocyte counting. GFAP (astrocyte) immunoreactivity in the hippocampus (ROI) was assessed in CA1-CA3 regions using to the ImageJ (NIH) macro.

Cytochrome oxidase (CCO) activity staining

In this assay, we used 12, 5, and 5 animals from the RB, H-BM,R, and L-BMR line types, respectively. Histochemical reaction for cytochrome oxidase was prepared according to the published protocol (45). Forty μm coronal sections of the hippocampus (located -2.3 mm from bregma, coordinates according to the atlas of Paxinos and Franklin, (42) were incubated ina solution of 0.05 PB (phosphate buffer), 1 g sucrose, 50 mg ammonium nickel, 25 mg DAB, 15 mg cytochrome C, 10 mg catalase and 250 ul imidazole (amount per 100 ml). Slices were incubated for 6 - 8 h in 37°C and then washed (3 x 1 min) in a 0.05 PB.

Dendritic spine density (DSD) analysis

Because of logistic limitations, the DSD analysis was restricted to the H-BMR and L-BMR mice only (5 animals from each line type). Animals were deeply anesthetized with sodium pentobarbital (133.3 mg/ml), pentobarbital (26.7mg/ml), and morbital. Their brains were dissected and fixed in 1.5% PFA (1h, RT) and moved to ice-cold PBS (10 min, 4°C). Next, they were cut into 140 μm sections on a vibratome (Leica VT 1000S, Leica Biosystems Nussloch GmbH, Wetzlar, Germany). Slices were processed for Dil staining. Dendritic spines were visualized using the lipophilic dye Dil (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate, #D282 Life Technologies, Warsaw, Poland). Random dendrite labeling was performed using 1.6 μm tungsten particles (Bio-Rad, Hercules, CA, USA) coated with Dil. Dye was delivered to cells using the Gene Gun (Bio-Rad). After staining, slices were fixed with 0.4% paraformaldehyde in phosphate-buffered saline (PBS; overnight at 4°C) and placed on microscopic slides. Z-stacks of dendrites from the CA1, CA3, and the DG regions of the hippocampus were acquired using a Zeiss LSM 880 confocal microscope with AiryScan on superresolution mode using 63 times magnifying lens for high-resolution imaging (Plan Apochromat 63x/1.4 Oil DIC) (Zeiss, Poznań, Poland). Dil emission was excited using a HeNe 594 nm laser. For each image, the following parameters were applied: 70 nm pixel size, 300 nm Z-intervals, averaging 4. Maximum intensity projections of Z-stacks covering the dendrite length were analyzed using semiautomatic SpineMagick! software. It allowed for marking the dendritic spine head and base manually. Next, the software automatically marked spine edges that were adjusted manually to fully reflect the spine shape.

For each animal, 7-10 single dendrites from each hippocampus area (one dendrite per neuron per image) were analyzed and dendritic spine density was calculated as the number of spines per observation field of 0.05 μm². There were no stained dendrites in the CA2 region, so they were excluded from further analyses.