Long-term changes of spine dynamics and microglia after transient peripheral immune response triggered by LPS in vivo.

Animals

All experimental procedures were carried out in compliance with the institutional guidelines of the University of Tokyo and the government. This study was approved by the animal welfare ethics committee at the University of Tokyo, Faculty of Medicine with the approval ID of P08-016. Every effort was made to minimize the suffering and the number of animals used. For in vivo imaging and histochemical analyses, both male and female C57BL/6 transgenic mice aged two to three months were used. For in vivo spine imaging, transgenic mice expressing green fluorescence protein (GFP) under the control of the Thy1 promoter ( Thy1-GFP M mice) were used [ 28 ]. To visualize microglia, transgenic mice expressing GFP under the control of the Iba1 promoter ( Iba1-GFP mice) were used [ 29 ].

Drug treatment

E.coli lipopolysaccharide (LPS, strain O111:B4, Sigma-Aldrich (catalog number L4391)) was dissolved in a saline solution at a concentration of 0.2 mg/mL and stored at -30°C in small aliquots. Mice were intraperitoneally injected with a single dose of LPS (0.5 mg/kg). The optimal dose of LPS was determined from the morphological changes in microglia two days after LPS injection with doses of 0.1, 0.3, 0.5 or 5.0 mg/kg. We observed morphological changes in microglia at doses higher than 0.5 mg/kg. Injection of LPS at 5.0 mg/kg induced severe behavioral responses, while sickness behavior was less prominent at lower doses. From these pilot experiments, we selected a dose of 0.5 mg/kg for experiments with spine imaging and glial activation.

Surgery

Mice were deeply anesthetized intraperitoneally with ketamine (100 mg/kg body weight) and xylazine (10 mg/kg body weight) diluted in a saline solution. After the disappearance of the pinching response the hair of the scalp was shaved and a midline incision of the scalp was made. Periosteum tissue was removed with a surgical blade and the somatosensory area (-1.5 mm from Bregma and 2.0 mm from the midline) was marked by stereotactic coordinates. A small rectangular metal plate with a round hole was glued on the skull and mice were fixed to the immobilized stage (SR-5M, Narishige) with a heating pad to maintain body temperature. The skull above the imaging area was thinned through the round hole of the metal plate. The thinning was initially done over a small area (a 1.5 × 1.5 mm square) with a high speed micro-drill (KM11, Minimo). When the bone reached ~50 μm in thickness, further thinning was performed manually with micro surgical blades (NORDLAND blade, Salvin Dental) until the skull reached ~15 μm in thickness. We paid particular attention not to push the skull during the thinning process. The final imaging window was a 0.5 × 0.5 mm square. For the repetitive imaging, the brain vasculature pattern was recorded with a CCD camera (GZ-MG70, Victor). In vivo imaging A scanning microscope (FV-300, Olympus) equipped with a pulsed laser (MaiTai HP, Spectra Physics) was used for imaging with a water immersion objective lens (1.05 NA, 25x, Olympus). The wavelength was 920 nm and the average power of the laser after the objective lens was between 10 and 20 mW. The imaging area was 234 μm × 234 μm (low magnification) or 78 μm × 78 μm (high magnification), with an imaging depth 50 μm from the surface of the neocortex (Layer 1) and the step size of the z stack set to 0.75 μm. The pixel sizes of single horizontal images were set to 512 × 512. Low magnification images, together with images of the vasculature pattern taken with a CCD camera, served as the reference maps for repetitive acquisition of higher magnification images from the same cortical area. For repeated imaging, the metal plate attached on the skull was removed and the skin was sutured. The mice were kept on the heating pad until they recovered from the anesthesia and were returned to their home cage. To minimize the damage to the brain tissue due to the re-thinning of the skull, mice were imaged twice for most of the experiments and three times at the maximum.

Fixation of animals and immunohistochemistry

Wild type or Iba1-GFP mice were sampled before or two, seven, 28 days after LPS injection. Mice were deeply anesthetized with pentobarbital and perfused transcardially with PBS followed by 4% paraformaldehyde. Brains were removed and further fixed in 4% paraformaldehyde overnight at 4°C. Slices were made with a vibratome (DTK-1000, Dosaka EM) with a 50 μm thickness. Slices from wild type mice were stained with anti-Iba1 antibody (1/500, Wako Pure Chemicals) or anti-glial fibrillary acidic protein (GFAP) antibody (1/3000, Sigma-Aldrich) followed by the fluorescence conjugated secondary antibodies to visualize the microglia or astrocytes. Mice from various time intervals after LPS injection or control were fixed on the same day and slices were stained at the same time. Images were obtained by using a laser scanning confocal microscope (FV-1000, Olympus) or a wide-field fluorescence microscope (BX-50, Olympus) under the same illumination and collection conditions.

Data analysis

All analyses of spine dynamics, densities, and estimations of sizes were done manually using National Institutes of Health ImageJ software ( http://rsb.info.nih.gov/ij ). We could identify both spines and filopodia [ 30 ] but analyzed only dendritic protrusions classified as spines in this study (Figure 1 ) [ 31 ]. The same dendritic segments (5-50 μm length) were identified from the image stacks at different time points and spines were selected and classified into three groups. New spines were those identified only at the second time point. Eliminated spines were those present only at the first time point but missing at later points. Spines present at both time points were categorized as stable spines. The number of spines in each group was counted and both the formation and elimination rates were calculated as the percentages of eliminated spines and newly formed spines to the total number of spines examined respectively. To ensure the tissue movements and rotations three-dimensional stacks were always used for the analysis. We did not analyze structures that projected mainly along the imaging axis, below or above dendrites. We considered a spine in the second image to be the same spine as in the first image if the second image spine was located within 0.5 μm of the expected location based on its spatial relationship to adjacent spines or landmarks like axonal and dendritic orientation. For the morphological analysis of spines, we selected dendritic segments where eliminated spines could be identified. Binary images were constructed by appropriate thresholding to determine the outline of spines. The length from the neck to the tip of a spine (a) and the maximum width of a head (b) were measured manually and the sum of these two values was taken as a parameter reflecting spine size. For the spine density analysis, we used the same dendritic segments as in the spine turnover analysis. We selected the dendritic segments whose lengths were more than 20 μm for the spine density calculation. Spine density was calculated as (N of spines)/(lengths of dendritic segments). For the analysis of Iba1 expression we took 41 images at the z step of 1 μm and a maximum projection image was generated from the same number of stack images for all the individual slices. Images were obtained with a laser scanning confocal microscope (FV-1000, Olympus) under the same illumination and collection conditions. The total fluorescence intensity of projection image was measured by ImageJ software. The background intensity was subtracted from the total fluorescence intensity. The intensity values were normalized to the mean intensity value of the control. The density of microglia was calculated from the same slices as used for the measurement of intensity. The number of Iba1-positive cell bodies was manually counted three-dimensionally from 41 image stacks. The number of microglia was normalized to the mean number of microglia in the control. Imaris v6.1.3 (Bitplane, Zurich, Switzerland) was used for the three-dimensional analysis of microglial processes. Surface rendering of GFP-positive microglia from z-stacks of confocal microscope images was performed and the lengths of processes were determined by measuring the distance from the centroid of a cell body to the tip along the processes. Structural dynamics of microglia were quantified from time-lapse imaging of soma and protrusions in vivo . Images were obtained every three min for 30 min. We took 51 images at the z step of one μm for each time-interval image. The analysis was performed on maximum-intensity projections of fluorescence image stacks. The same number of stack images was used for each cell analysis. The dynamics of microglia were evaluated by measuring the length of extension and retraction of microglial protrusions. Ten processes of microglia were randomly selected and the length of extension and retraction was measured to compare the overlay of two images at two different time points. All the data are presented as means ± S.D. Multiple comparisons were made by an ANOVA test, followed by a Tukey's test (Figures 2 , 3 , 5 , 6 , 7 ). Statistical significance was evaluated using Student's t-test (Figure 4 ). Differences were considered to be significant if p < 0.05.

Cytokine array analysis

Tissue extracts from control and LPS-treated mice were screened by using a cytokine antibody array kit (Array1, RayBiotech Inc., Atlanta, GA, USA) according to the manufacturer's instructions. After euthanasia of mice with a large dose of pentobarbital, the neocortex and the hippocampus were dissected from the forebrain and homogenized in a lysis buffer. Brain extracts were then centrifuged at 10,000 × g for 10 min at 4°C. Protein concentrations of supernatants were determined and equal amounts of protein were incubated with array membranes. The signals generated by enhanced chemiluminescence were recorded by ChemiDoc XRS (Bio-Rad, Tokyo, Japan). For the analysis of chemiluminescence signals from cytokine spots, total signals were calculated and then background intensity was subtracted. The averaged optical densities of the six positive controls on the membrane were calculated and the optical densities of cytokine spots were divided by these values. These normalized values were compared and the ratio between control and LPS-treated mice was calculated.