Abnormal Microglia and Enhanced Inflammation-Related Gene Transcription in Mice with Conditional Deletion of Ctcf in Camk2a-Cre-Expressing Neurons

Mouse husbandry.

All animal procedures were performed in compliance with the National Institutes of Health's Guide for the Care and Use of Laboratory Animals and approved by the Animal Studies Committee in the Division of Comparative Medicine at Washington University School of Medicine in St. Louis (Protocol #20140044). Mice were housed in a temperature-controlled barrier facility maintained at 24°C. Facility lighting was kept on a fixed 12 h light and 12 h dark cycle. Mice had access to fresh water and rodent chow #5001 (Purina) ad libitum. Mouse health was monitored daily by a staff of licensed veterinarians. Mouse lines used in this experiment have been described previously and include: Ctcf^loxP (Heath et al., 2008), Camk2a-Cre (Tsien et al., 1996; The Jackson Laboratory stock #005359, RRID:IMSR_JAX:005359), Gt(ROSA)26Sor^{tm9(CAG-tdTomato)Hze} (Madisen et al., 2010; The Jackson Laboratory stock #007909, RRID:IMSR_JAX:007909), Thy1-YFP^HJrs/J (Feng et al., 2000; The Jackson Laboratory stock #003782, RRID:IMSR_JAX:003782), Rpl22^tm1.1Psam/J (Sanz et al., 2009; The Jackson Laboratory stock #011029, RRID:IMSR_JAX:011029), and Cx3cr1^tm1Litt/J (Jung et al., 2000; The Jackson Laboratory stock #005582, RRID:IMSR_JAX:005582). We intercrossed Ctcf^loxP and Camk2a-Cre mice to generate Ctcf^loxP/loxP;Camk2a-Cre⁺ (Ctcf CKO) and Ctcf^loxP/loxP;Camk2a-Cre⁻ (control) mice. We intercrossed Ctcf^loxP, Camk2a-Cre, and Thy1-YFP^HJrs/J mice to generate Ctcf^loxP/loxP;Camk2a-Cre⁺;Thy1-YFP^HJrs/J+ (Ctcf CKO;YFP) and Ctcf^loxP/loxP;Camk2a-Cre⁻;Thy1-YFP^HJrs/J+ (control;YFP) mice. Similarly, we intercrossed Ctcf^loxP, Camk2a-Cre, and Rpl22^tm1.1Psam/J mice to generate Ctcf^loxP/loxP;Camk2a-Cre⁺; Rpl22^tm1.1Psam/J+ (Ctcf CKO;RiboTag) and Ctcf^loxP/loxP;Camk2a-Cre⁻; Rpl22^tm1.1Psam/J+ (control;RiboTag) mice. We verified the extent of expression of Cre in our Camk2a-Cre mice by intercrossing them with Gt(ROSA)26Sor^{tm9(CAG-tdTomato)Hze} mice to generate Gt(ROSA)26Sor^{tm9(CAG-tdTomato)Hze};Camk2a-Cre⁺ (tdTomato;Camk2a-Cre) mice. We verified that Cx3cr1 and Camk2a expression patterns were nonoverlapping by intercrossing tdTomato;Camk2a-Cre and Cx3cr1^tm1Litt/J mice to generate animals with all three alleles, Gt(ROSA)26Sor^{tm9(CAG-tdTomato)Hze};Camk2a-Cre⁺;Cx3cr1^tm1Litt/J (Cx3cr1-EGFP; tdTomato;Camk2a-Cre mice). Both male and female animals were used for all experiments.

Mouse behavioral analysis.

We performed behavioral analysis of two cohorts of Ctcf CKO and control mice. Testing was conducted on the first (adult) cohort of 9 Ctcf CKO and 10 control mice, including 11 females and 9 males, at 3–4 months of age. Mice were evaluated on a 1 h locomotor activity test, a battery of sensorimotor measures, the Morris water maze (MWM), and the social approach test, in that order. Testing of the second (adolescent) cohort of 10 Ctcf CKO (5 males, 5 females) and 10 control mice (6 males, 4 females) occurred at 6–8 weeks of age. These mice were evaluated on a 1 h locomotor activity test, a battery of sensorimotor measures, and the MWM, in that order.

The 1 h locomotor activity test was conducted during a 1 h session inside transparent 47.6 × 25.4 × 20.6 cm high polystyrene enclosures as described previously (Wozniak et al., 2004). Computerized photobeam instrumentation was used to measure ambulatory activity and exploration including total ambulations (whole-body movements), number of vertical rearings, and time spent, distance traveled, and entries made into a central 33 × 11 cm zone.

The sensorimotor battery was performed as described previously (Wozniak et al., 2004) and consisted of the ledge and platform tests (to evaluate balance and fine motor coordination); the pole, 60° inclined screen test, and 90° inclined screen test (to assess agility); the inverted screen test (to assess strength); and the walking initiation test (to assess movement initiation).

The MWM was performed as described previously (Wozniak et al., 2007) in a 118-cm-diameter pool of opaque water monitored by a computerized video tracking and recording system (ANY-maze; Stoelting, RRID:SCR_014289) that computed escape path length and latency to reach the target platform and calculated swimming speed, platform crossings, and time spent in each quadrant of the pool. Testing included cued (visible platform marked by a red tennis ball on a pole), place (submerged platform obscured from view), and probe (platform absent) trials. Both cued and place trials occurred in 2 blocks of 2 trials daily (4 trials per day), with a 2 h break between blocks. Each trial lasted a maximum of 60 s and was followed by a 60 s intertrial interval, during which the mouse was allowed to remain on the platform for the first 30 s. Cued trials occurred over 2 consecutive days. Salient spatial cues were not present in the room during the cued trials and the platform location varied from trial to trial to train the mice to navigate to the platform but minimize spatial learning during this phase of testing. Place trials were initiated 3 d later and occurred over 5 consecutive days. Spatial cues were prominently displayed during these trials to encourage spatial (hippocampal-dependent) learning. The probe trial was conducted 1 h after the last place trial on the fifth day. For the probe trial, spatial cues were again prominently displayed, but the escape platform was completely removed from the pool. The mouse was placed in the pool starting from the quadrant diagonal to the last location of the escape platform and allowed to search the pool for 60 s while time spent in each quadrant and the number of crossings over the previous location of the platform was recorded.

The general procedures for conducting the social approach test were similar to our previously published methods (Dougherty et al., 2013), which were adapted from earlier works (Moy et al., 2004; Silverman et al., 2011). The test involved quantifying sociability, or the tendency to initiate social investigation of an unfamiliar conspecific (stimulus mouse) contained in a small withholding cage, compared with the investigation of an empty withholding cage. The apparatus was a rectangular 3-chambered Plexiglas box (each chamber measuring 19.5 cm × 39 cm × 22 cm) containing Plexiglas dividing walls with rectangular openings (5 × 8 cm) covered by sliding Plexiglas doors. A small stainless-steel withholding cage (10 cm h × 10 cm diameter; Galaxy Pencil/Utility Cup; Spectrum Diversified Designs) was used to sequester a stimulus mouse. The withholding cage consisted of vertical bars, which allowed for restricted social interactions between the mice but prevented fighting and sexual contact, and one cage was located in each outer chamber. A digital video camera connected to a PC loaded with a tracking software program (ANY-maze; Stoelting) recorded the movement of the mouse within the apparatus and quantified time spent in each chamber and in investigation zones surrounding the withholding cages, the latter being scored when the head of the mouse intersected the zones. The investigation zones were 12 cm in diameter, encompassing 2 cm around the withholding cages. The test sequence consisted of three consecutive 10 min trials. For the first trial, each mouse was placed in the middle chamber with the doors to the outer chambers shut to become acclimated to the apparatus. During the second trial, the mouse was allowed to investigate and habituate to all three chambers freely, including the empty withholding cages (Page et al., 2009; Naert et al., 2011; Pobbe et al., 2012). The third (test) trial assessed sociability exhibited toward an unfamiliar stimulus mouse versus the familiar, empty withholding cage by placing an unfamiliar, age- and gender-matched stimulus mouse in one withholding cage while the other was left empty. The test mouse was allowed to explore the apparatus freely and investigate the novel mouse in the withholding cage. The locations of the stimuli mice in the outer chambers were counterbalanced within and across groups.

Mouse electroencephalography.

Adult (3–4 months old) control and Ctcf CKO mice (n = 6/genotype) underwent continuous video-EEG monitoring using standard methods for implanting epidural electrodes and performing continuous video-EEG recordings, as described previously (Erbayat-Altay et al., 2007). Briefly, mice were anesthetized with isoflurane and placed in a stereotaxic frame. Five epidural screw electrodes were surgically implanted on the skull and secured using dental cement. Video and EEG data were acquired simultaneously with a stellate video-EEG acquisition system. Continuous 24/7 video-EEG data were obtained for 6 weeks from each mouse and were analyzed for seizures or interictal epileptiform abnormalities. Electrographic seizures were defined as a characteristic pattern of discrete periods of rhythmic spike discharges that evolved in frequency and amplitude lasting at least 10 s.

Western blot.

We harvested hippocampus and cerebral cortex from adult (3–6 months old) Ctcf CKO and control mice (n = 18 animals per genotype for hippocampus; n = 26 animals per genotype for cerebral cortex). Tissues were snap frozen in liquid nitrogen and homogenized in lysis buffer containing 50 mm Tris pH 7.4, 150 mm NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, 1 mm PMSF, 1 mm sodium orthovanadate, and complete protease inhibitor mixture (Sigma-Aldrich). The lysates were cleared by centrifugation at 13,800 × g for 10 min at 4°C and total protein was quantified using the MicroBCA assay kit (Pierce). Proteins were separated by SDS-PAGE and transferred to nitrocellulose membrane (GE Life Sciences). Membranes were blocked for 1 h at room temperature in 5% nonfat dry milk in Tris-buffered saline with 0.05% Tween (TBST) and incubated overnight at 4°C with either 1:1000 anti-CTCF antibody (Cell Signaling Technology catalog#2899, RRID:AB_2086794) in TBST plus 5% bovine serum albumin or 1:1000 anti-GAPDH antibody (Cell Signaling Technology catalog#5174, RRID:AB_10622025) in TBST plus 5% nonfat dry milk. The next day, membranes were washed with TBST and incubated with 1:5000 goat anti-rabbit conjugated to horseradish peroxidase (GE Life Sciences) in TBST for 2 h at room temperature before a final wash. Membranes were developed with Western Bright Quantum HRP substrate (Advansta). Blots were imaged on a G:Box gel documentation system (Syngene) running GeneSys acquisition software (RRID:SCR_015770). Tag image file format-style images were exported to the Fiji build of ImageJ (RRID:SCR_002285) and analyzed using the Gel Analyzer function.

Total mRNA isolation.

Brain regions were harvested from both male and female adult (3–6 months old) Ctcf CKO and control mice after behavioral testing. Tissues were snap frozen in liquid nitrogen and homogenized in TRIzol (Life Technologies) using a bullet blender (Next Advance). The TRIzol procedure was performed according to the manufacturer's instructions. RNA was purified on NucleoSpin RNA columns with on-column DNase digestion (Machery-Nagel) according to the manufacturer's intructions. RNA concentration was determined using a NanoDrop (Thermo Scientific) and RNA integrity was verified using an Agilent Technologies bioanalyzer and either RNA 6000 nano or pico chips, depending on the concentration of the sample. Only high-quality RNA samples, defined as having both a 260/280 nm aborbance ratio of 1.8 to 2.1 on the NanoDrop and RNA integrity number > 8.0 on the Bioanalyzer, were used in downstream microarray and qRT-PCR applications.

Histology, immunofluorescent antigen detection, and microscopy.

Three- to 6-month-old mice were anesthetized with isoflurane and transcardially perfused with either 4% paraformaldehyde in PBS or, for Timm staining, with 0.9% sodium chloride, 0.37% sodium sulfide sulfide solution, and 4% formaldehyde. Brains were dissected and fixed overnight in 4% paraformaldehyde in PBS, cryopreserved in 30% sucrose in PBS, and embedded in Tissue-Tek optimal cutting temperature compound (Sakura Finetek). Then, 40-μm-thick sections were cut on a cryostat and mounted on Fisherbrand Superfrost/Plus microscope slides, which were air-dried and stored at −20C. Nissl staining was performed with cresyl violet (Sigma-Aldrich) as described previously (Paul et al., 2008). Timm staining was performed as described previously (Guo et al., 2013). Light microscope imaging was performed with a NanoZoomer 2.0-HT digital slide scanner (Hamamatsu). For immunofluorescent antigen detection, frozen sections were washed, antigen retrieved for 30 min at 80°C in 10 mm tribasic sodium citrate dihydrate, 0.05% Triton X-100, pH 6, and blocked in TBST with 10% normal horse serum and 0.3% Triton X-100 for 1 h at room temperature. Sections were incubated with primary antibodies diluted in blocking buffer overnight at 4°C. After 3 5 min washes with PBS, sections were incubated with secondary antibodies diluted in blocking buffer for 1 h at room temperature. Immunostained sections were mounted with Vectashield hard-set mounting medium with DAPI (Vector Laboratories, RRID:AB_2336788). Primary antibodies included the following: 1:100 anti-NeuN (EMD Millipore catalog#ABN78A4, RRID:AB_10920751), 1:500 anti-glial fibrillary acidic protein (GFAP) (Millipore catalog #AB5804, RRID:AB_10062746), 1:500 anti-Iba1 (Wako catalog #019-19741, RRID:AB_839504), 1:200 anti-Ki67 (Vector Laboratories catalog#VP-K451, RRID:AB_2314701), 1:150 anti-Caspase-3 D175A (Cell Signaling Technology catalog#9661, RRID:AB_2341188), 1:400 anti-S100-β (S100B; Dako catalog#Z0311, RRID:AB_10013383), 1:200 anti-CD68 (Bio-Rad catalog#MCA1957GA, RRID:AB_324217). Secondary antibodies included 1:500 anti-rabbit Alexa Fluor 488 (Thermo Fisher catalog #A11034, RRID:AB_2576217) and 1:500 anti-rabbit Cy3 (Jackson Immunoresearch catalog#111-165-144, RRID:AB_2338006). Confocal images were captured with a Leica DMI4000 confocal microscope equipped with diode 405 nm and solid-state 488 and 561 nm lasers and 20×, 63×, and 100× oil-immersion objectives. For each stain and antibody described above, at least eight sections from each of three to six brains per genotype were examined. Representative sections are displayed. The LASX software analysis suite (Leica) was used to generate maximum intensity projections of each image. For each antibody, a uniform threshold was applied to all images from both genotypes. Thresholded images were binarized and imported into the Fiji build of ImageJ (RRID:SCR_002285), which was used to calculate the area of the binary image equivalent to the area of the image occupied by a given antibody. This value was divided by the total image area to determine a percentage area. For CD68, we divided by the Iba-1-positive image area to determine the percentage microglial area occupied by CD68. For quantification of dendritic spines, we collected full-thickness confocal z-stacks of apical dendrites using 100× magnification, 512 × 512 resolution, 3 frame average, and 140nm z-step size (n = 10–20 image stacks per animal, 3 animals per genotype). We selected individual dendrites (n = 50 dendrites/animal for hippocampus, n = 28–32 dendrites/animal for cortex) of at least 10 μm length, generated mean intensity projection images using the LASX software analysis suite (Leica), and counted the number of spines manually, which we report as spines per micrometer of dendrite length. Individuals blinded to genotype performed all counting. We evaluated the patterns of enhanced green fluorescent protein (EGFP) and red fluorescent tdTomato protein in the hippocampus and cerebral cortex of Cx3cr1-EGFP; tdTomato;Camk2a-Cre mice using 10 images of each tissue, which were collected from a total of three animals.

RiboTag isolation of Camk2a-Cre expressing cell-specific mRNA: The RiboTag isolation method was performed as described previously (Sanz et al., 2009). Briefly, hippocampus was dissected rapidly from naive adult mice (2 males and 2 females, each at least 3 months old) and homogenized in polysome buffer consisting of 50 mm Tris, pH 7.5, 100 mm KCl, 12 mm MgCl₂, 1% Nonidet P-40, 1 mm DTT, 200 U/ml Rnasin Plus (Promega), 200 U/ml Superasin (Life Technologies), 100 μg/ml cyclohexamide, 1 mg/ml heparin, and 1× complete protease inhibitor tablet mixture (Sigma-Aldrich). Samples were centrifuged at 10,000 × g for 10 min at 4°C. The supernatants were collected and immunoprecipitated with anti-HA antibody (HA.11, BioLegend catalog #901502, RRID:AB_2565007) overnight at 4°C and, the next day, antibody-antigen-RNA complexes were recovered with Dynabeads protein G. The beads were washed 3 times for 5 min each in high-salt buffer consisting of 50 mm Tris, pH 7.5, 300 mm KCl, 12 mm MgCl₂, 1% Nonidet P-40, 1 mm DTT, and 100 μg/ml cyclohexamide. Then, the antigen–antibody–RNA complexes were eluted with RA1 buffer (Machery-Nagel) containing 1:100 β-mercaptoethanol. RNA was recovered from the eluate using NucleoSpin RNA columns (Machery-Nagel) with on-column DNase digestion. RNA concentration was determined using a NanoDrop (Thermo Scientific) and RNA integrity was verified using an Agilent Technologies nioanalyzer and either RNA 6000 nano or pico chips, depending on the concentration of the sample. The remainder of each sample was frozen at −80°C until further use.

qRT-PCR.

We used the qScript cDNA synthesis kit (Quanta Bio) to generate cDNA from 10 ng samples of RiboTag-isolated RNA or 500 μg of TRIzol-isolated RNA. qRT-PCR was performed using PerfeCTa SYBR Green FastMix (Quanta Bio) on an ABI Prism 7900HT Sequence Dectection System (Applied Biosystems). All qRT-PCRs were run using a standard program: 2 min at 50°C, 10 min at 95°C, followed by 40 cycles of 15 s at 95°C, 30 s at 60°C, and 30 s at 72°C. The following primer pairs were used: Gapdh 5′-AATGTGTCCGTCGTGGATCT-3′ forward and 5′-GTTGAAGTCGCAGGAGACAA-3′ reverse; C1ql1 5′-CCAACCTAGGCAACAACTAC-3′ forward and 5′-GTAGTTCTGGTCTGCATCCT-3′ reverse; Ccl17 5′-GATGCCATCGTGTTTCTGAC-3′ forward and 5′-CCAATCTGATGGCCTTCTTC-3′ reverse; Tnfrsf12a 5′-GAGAAAAGTTTACTACCCCCATAGAG-3′ forward and 5′-GGCTGACTCCAGAATGAATGAA-3′ reverse; Grk4 5′-GGTGCATTGAATTCTTGGATG-3′ forward and 5′-GGGACTTCTGACTTCTCTTTG-3′ reverse; Nefh 5′-TGCCGCTTACAGAAAGCTC-3′ forward and 5′-GCGTGGATATGGAGGGAATTT-3′ reverse; Tlr2 5′-ACCTCAGACAAAGCGTCAAA-3′ forward and 5′-TTGCTGAAGAGGACTGTTATGG-3′ reverse; Ptgs2 5′-CCAGAGCAGAGAGATGAAATAC-3′ forward and 5′-TCCTTCTCTCCTGTAAGTTCT-3′ reverse; Ccl2 5′-CTCTCTTCCTCCACCACCAT-3′ forward and 5′-CGTTAACTGCATCTGGCTGAG-3′ reverse; Ccl3 5′-TTCTCTGTACCATGACACTCTGC-3′ forward and 5′-CGTGGAATCTTCCGGCTGTAG-3′ reverse; Ccl9 5′-ATCACACATGCAACAGAGACA-3′ forward and 5′-TGGAACCCCCTCTTGCTGAT-3′ reverse; Cd83 5′-GTTGCTCTTCTCTCTGGTTG-3′ forward and 5′-CTTGTTCCGTACCAGGTTTAG-3′ reverse; Cd14 5′-CCACCGCTGTAAAGGAAAGA-3′ and 5′-CCAGAAGCAACAGCAACAAG-3′ reverse; and Cd52 5′-GGTTGTGATTCAGATACAAACAG-3′ forward and 5′-GAGGTAGAAGAGGCACATTAAG-3′ reverse. We used the standard curve method to quantify gene expression levels, which were normalized to Gapdh expression. For analysis of Ctcf expression, n = 6 samples per genotype per tissue were used. For analysis of Tlr2, Ptgs2, Ccl2, Ccl3, Ccl9, Cd83, Cd14, and Cd52, n = 8 samples per genotype were used. For analysis of C1ql1, Ccl17, Tnfrsf12a, Grk4, and Nefh, n = 12 samples per genotype were used.

Microarray analysis.

For microarray on RNA derived from whole hippocampus tissue, RNA was amplified and reverse transcribed to cDNA using the MessageAmpII kit (Life Technologies) according to the manufacturer's instructions. The amplified cDNA was then fluorescently labeled using the Kreatech ULS kit (Kreatech Diagnostics) according to the manufacturer's instructions. Labeled cDNAs were purified on QIAquick PCR purification columns (Qiagen) and quantified using a NanoDrop spectrophotometer. Then, the labeled cDNAs were hybridized to Agilent Technologies mouse GE 4x44k v2 microarrays according to the manufacturer's instructions. Slides were scanned on an Agilent Technologies SureScan microarray scanner to detect fluorescence. Gridding and analysis of images was performed using Agilent Technologies Feature Extraction software version 11.5 (RRID:SCR_014963) and the resulting feature (probe)-extracted .txt data for each sample was used for downstream bioinformatic analysis as described below. For analysis of whole hippocampus gene expression by microarray, we used three independent biological replicates per genotype (Ctcf CKO and control). Each biological replicate consisted of a pool of paired hippocampi from 3 individual mice 3–6 months of age. One Ctcf CKO biological replicate was hybridized twice to microarrays to generate a single technical replicate. Data from both technical replicates were combined and averaged for downstream analysis. In a separate microarray experiment, RiboTag-derived RNA from hippocampus (n = 4 individual Ctcf CKO;RiboTag and Control;RiboTag animals) was processed as described above, except for the following: (1) RNA was amplified and reverse transcribed to cDNA using the WTA2 Complete Whole Transcriptome Amplification Kit (Sigma-Aldrich); and(2) labeled cDNAs were hybridized to Agilent Technologies SurePrint G3 Mouse GE 8x60k microarrays. Microarray data from all experiments described here was submitted to the ArrayExpress Repository (RRID:SCR_002964; accession number pending; Parkinson et al., 2007).

Bioinformatic analysis of microarray data.

Microarray raw data were normalized using IRON: iterative rank order normalization method (Welsh et al., 2013) and analyzed in the R programming environment using the software packages Bioconductor (Huber et al., 2015; RRID:SCR_006442) and limma (Ritchie et al., 2015; RRID:SCR_010943). Results are expressed in terms of fold changes relative to controls and unadjusted p-values. We used a less stringent cutoff of p < 0.05 because we planned to do extensive wet-lab validation of the genes of interest, keeping in mind that FDR or multiple testing corrections can be too restrictive. Genes with a >±1.5 fold change and p < 0.05 were further analyzed for biological pathway enrichment using EnrichR (Chen et al., 2013; RRID:SCR_001575). We report pathways and biological processes with adjusted p < 0.05. To generate a profile for the non-neural component, we compared the profiles of the whole hippocampus and RiboTag-derived samples and “subtracted” the RiboTag-derived genelist from that of the whole hippocampus.

Flow cytometry.

We used flow cytometry to determine the composition of inflammatory cells in the hippocampus and cortex of Ctcf CKO mice. First, we isolated microglia from these brain regions, which were dissected, dispersed in homogenization medium (RPMI 1640 medium; Thermo Fisher catalog #11875085) plus 2% heat inactivated fetal bovine serum and passed through a disposable 70 μm filter to remove debris. Cells were pelleted by centrifugation at 400 × g for 6 min at 4C and resuspended in 40% Percoll in homogenization media. This was carefully layered upon 80% Percoll in homogenization media and centrifuged at 1000 × g for 20 min at 18C. The immiscible layer (containing the inflammatory infiltrate) was collected, washed in homogenization medium, and centrifuged at 400 × g for 6 min at 4C to pellet the cells. The cells were resuspended in flow cytometry buffer (Hanks' balanced salt solution without calcium, magnesium, or phenol red; Thermo Fisher catalog #14175079, +2% heat inactivated fetal bovine serum) and blocked with 1:100 purified anti-mouse CD16/32 antibody (BioLegend catalog #101302, RRID:AB_312801). The cells were then incubated with 1:100 FITC anti-mouse CD45 antibody (BioLegend catalog #103108, RRID:AB_312973) and 1:100 PE/Cy7 anti-mouse/human CD11b antibody (BioLegend catalog #101216, RRID:AB_312799) for 20 min at 4°C protected from light. Samples were stained with SYTOX blue (Thermo Fisher catalog #S348857) at a final concentration of 1 μm per the manufacturer's instructions. Cells were analyzed on a FACScan 2 flow cytometer (BD Biosciences) using a gating strategy similar to one described previously to isolate microglia, myeloid, and lymphoid cells (Galatro et al., 2017). All gating strategies incorporated size and doublet discrimination based on forward and side scatter parameters. Data analysis was performed using FlowJo (RRID:SCR_008520) version 10.4 for Macintosh.

Sholl analysis of microglial morphology.

We performed Sholl analysis of microglial morphology as described previously (Norris et al., 2014). Briefly, we used confocal microscopy to collect full thickness 63× z-stack images of CA1 hippocampus from 40-μm-thick brain sections that were immunostained for Iba-1 and DAPI (as outlined above; n = 6 mice/genotype, 25 image stacks per animal). Z-stacks were collected at 512 × 512 resolution with 3 frame averages for each color channel and 1 μm z-step size. We used LASX software (Leica) to prepare a maximum intensity projection image of the Iba-1 channel, which we thresholded and exported to the Fiji build of ImageJ (RRID:SCR_002285). For each image, we removed surrounding processes manually in Fiji, thereby isolating a total of 25 microglia per mouse. We used the line segment tool to draw a line from the center of each soma to the tip of its longest process, which provided the maximum process length. We used the Sholl analysis plugin (Ferreira et al., 2014), with the first shell set at 10 μm and subsequent shells set at 5 μm step sizes, to determine intersections at each Sholl radius. This also provided the critical radius (radius value with the highest number of intersections), the process maxiumum (the highest number of intersections regardless of radius value), and the number of primary branches (intersections at the first Sholl radius). We measured the soma size in Fiji and counted branch endpoints using the cell counter plugin. Finally, we counted microglia per 63× image stack manually to determine the mean number of microglia per high-power field.

Experimental design and statistical analysis.

ANOVA models and t tests were used to analyze the behavioral data. Repeated-measures ANOVA (rmANOVA) models containing one between-subjects variable (genotype) and one within-subjects (repeated measures) variable (e.g., blocks of trials, time blocks) were typically used to analyze certain variables within the activity, spatial learning/memory, and social approach datasets. The Greenhouse–Geisser adjustment of α levels was used for all within-subjects effects containing more than two levels to help protect against violations of sphericity/compound symmetry assumptions underlying rmANOVA models. Pairwise comparisons were also conducted following relevant overall ANOVA effects and were considered significant according to Bonferroni-corrected levels. This involved controlling familywise error rates by dividing 0.05 by the exact number of comparisons that were conducted and using that criterion value to determine significance. Planned within-subjects contrasts were also conducted when appropriate. Data from measures within the sensorimotor battery and some variables from the probe trial were analyzed with t tests. We analyzed mouse weight, qRT-PCR, dendritic spine, flow cytometry, and immunostaining data for NeuN, GFAP, and cleaved caspase-3 with two-tailed t tests. Western blot data and immunostaining data for S100B, Ki67, and CD68 did not follow a normal distribution and was analyzed with the Mann–Whitney U test. We analyzed microglia intersections per Sholl radius using an rmANOVA model with Greenhouse–Geisser adjustment, using genotype as the between-subjects variable and radius as the within-subjects (repeated measures) variable. Other Sholl measurements were evaluated with two-tailed t tests. Number, age, and sex of animals used in individual experiments are noted above.