Pex2^flox/flox mice with a floxed exon 7 of Pex2 gene were generated in Unitech (Kanagawa, Japan). Briefly, genomic DNA corresponding to Pex2 locus was isolated from bacterial artificial clone (ID: RP23-184F21) containing C57BL/6 genomic DNA. FRT-Neo-FRT-loxP (FNFL) cassette and a loxP site were engineered into flanking exon 7 of Pex2 gene. The gene targeting vector was as follows: a 2.8-kb short homology arm (11,001–13,800), FNFL cassette, floxed sequence containing exon 7 (13,801–16,607), loxP, and 5.6 kb long homology arm (16,608–22,252) into a vector pBS-DTA carrying the diphtheria toxin A chain (DTA) (Unitech). The replacement vector was linearized by SacII site and electroporated into C57BL/6J embryonic stem (ES) cells that were cultured on G418-resistant mouse embryonic fibroblasts. The resistant colonies were picked up and screened for homologous recombination by Southern blot analysis. Targeted ES cells were injected into C57BL/6 blastocysts to produce chimeric mice. The Neo cassette was then removed in vivo by using flp/FRT recombination by crossing Neo-positive offspring from chimeric breeding with transgenic flp deleter mice. To generate conditional Pex2-knockout mice, Pex2^flox/flox mice were crossed with Cre-ER transgenic mice [B6.Cg-Tg(CAG-cre/Esr1^∗)5Amc/J] (Hayashi and McMahon, 2002) obtained from Jackson Laboratory. Heterozygous floxed Pex2 offspring with Cre-ER (Pex2^flox/flox/Cre-ER^+/–) were backcrossed to Pex2^flox/flox mice. Homozygous floxed Pex2 littermates with Cre-ER (Pex2^flox/flox/Cre-ER^+/–) and homozygous floxed Pex2 littermates lacking Cre-ER (Pex2^flox/flox) were used as conditional Pex2-knockout mice and control mice, respectively. PCR analysis of tail-biopsy genomic DNA was performed using primers 2F (5′-GACTTCTGAGGCACTCATACCTAAC-3′), 2R (5′-ATTGTGAGTTTCAGCATTTCTATGG-3′) to amplify 277 and 435 bp fragments specific for the wild-type allele and targeted allele, respectively. Primers 2F and 2R2 (5′-GACTTCTGAGGCACTCATACCTAAC-3′) were used to amplify 620 bp fragment specific for the disrupted Pex2 allele. CreF (5′-CGCGATTATCTTCTATATCTTCAGG-3′) and CreR (5′-AGGTAGTTATTCGGATCATCAGCTA-3′) were for 500 bp fragment specific for the Cre gene.

Tamoxifen (Sigma) dissolved at 20 mg/ml in corn oil (Sigma) was orally administrated (10 mg/40 g of body weight) for 5 consecutive days (Hayashi and McMahon, 2002). Two weeks after the last injection, these mice were used for behavioral experiments as described below. Four weeks after the last administration, mouse tissues including brain and liver were excised and homogenized for genotyping, protein analysis, and lipid extraction. Wild-type, Pex2^flox/flox/Cre-ER^–/–, and Pex2^flox/+/Cre-ER^+/– mice were used as control mice.

Apparatus for conditioning and context test consisted of an acrylic square chamber (33 cm × 25 cm × 28 cm) with metal grids (0.2 cm diameter, spaced 0.5 cm apart) covered by white acrylic lid (Udo et al., 2008). Grids were wired to a shock generator to deliver an electric foot-shock as the unconditioned stimulus. On training day, male control and male mutant mice were allowed to freely explore the chamber for 4 min to assess motor function, followed by a 0.2 mA foot-shock. On days 1 and 7, the mice were returned to the same conditioning chamber and freezing behavior was scored to measure contextually conditioned fear without foot-shock. Time spent for freezing and distance traveled were monitored using image J software. Conditioned stimuli (CS) were room and light. Equipment and apparatus were cleaned between trials with 70% ethanol.

Pex14^ΔC/ΔC mouse on a C57BL/6 background was previously described (Abe et al., 2018). Heterozygous offspring (Pex14^+/ΔC) were intercrossed to produce homozygous mutant animals. PCR analysis of tail biopsy genomic DNA was undertaken using primers P14F (5′-GTATAAATGTGGGAGTTTCCCTGG-3′) and P14R (5′-GTACTTGTGAACTCTGCTGGTAC-3′) to amplify a 599-bp fragment specific for the wild-type allele and primers P14F and KN52-2 (5′-GTGTTGGGTCGTTTGTTCGG-3′) to amplify a 169 bp fragment specific for the disrupted Pex14 gene.

Mouse monoclonal antibodies to α-tubulin and nestin (Rat-401) were purchased from BD Biosciences (San Jose, CA) and Invitrogen (Carlsbad, CA), respectively. Mouse monoclonal antibody to 2′,3′-cyclic-nucleotide 3′-phosphodiesterase (CNPase) was from Sigma (St Louis, MO). Rabbit antibodies to BDNF (N-20) and TrkB (H-181) were from Santa Cruz Biotechnology (Texas, CA). We used rabbit antisera to PTS1 peptide (Otera et al., 1998), rat catalase (Tsukamoto et al., 1990), mouse alkyldihydroxyacetonephosphate synthase (ADAPS) (Honsho et al., 2008), C-terminal region of rat Pex2 (amino acid residues 226–305) (Harano et al., 1999), 3-ketoacyl-CoA thiolase (Tsukamoto et al., 1990), acyl-CoA oxidase (AOx) (Tsukamoto et al., 1990), and sterol carrier protein x (SCPx) (Otera et al., 2001). Guinea pig anti-Pex14 antiserum (Itoh and Fujiki, 2006) was also used.

Total cellular lipids were extracted by the Bligh and Dyer method (Bligh and Dyer, 1959). Briefly, tissue lysates containing 50 μg of total cellular proteins were dissolved in methanol/chloroform/water at 2:1:0.8 (v/v/v) and then 50 pmol of 1-heptadecanoyl-sn-glycero-3-phosphocholine (LPC, Avanti Polar Lipids, Alabaster, AL), 1, 2-didodecanoyl-sn-glycero-3-phosphocholine (DDPC, Avanti Polar Lipids), and 1, 2-didodecanoyl-sn-glycero-3-phosphoethanolamine (DDPE, Avanti Polar Lipids) were added as internal standards. After incubation for 5 min at room temperature, 1 ml each of water and chloroform was added and the samples were then centrifuged at 2,000 rpm for 5 min in Himac CF-16RX (Hitachi Koki, Tokyo, Japan) to collect the lower organic phase. To re-extract lipids from the water phase, 1 ml chloroform was added. The combined organic phase was evaporated under a nitrogen stream and the extracted lipids were dissolved in methanol.

LC-MS/MS analysis of phospholipids was performed as described (Abe et al., 2014) using a 4000 Q-TRAP quadrupole linear ion trap hybrid mass spectrometer (AB Sciex, Foster City, CA) with an ACQUITY UPLC System (Waters, Milford, MA).

Total RNA was extracted from the tissues using TRIzol reagent (Invitrogen) and first-strand cDNA was synthesized with a PrimeScript RT reagent kit (Takara Bio, Shiga, Japan). Quantitative real-time RT-PCR was performed with SYBR Premix Ex Taq II (Takara Bio) using an Mx3000P QPCR system (Agilent Technologies, Santa Clara, CA). Several sets of primers used are listed in Supplementary Table S1.

Mice were deeply anesthetized with isoflurane and then decapitated. The mouse brains were fixed in 4% paraformaldehyde (PFA) in PBS, pH 7.4, for overnight at 4°C and were transferred to 30% sucrose in PBS for 2 days. The brains were embedded in the Tissue-Tek OCT compound (Sakura Finetek Japan, Tokyo, Japan) and subsequently frozen at −80°C. Cryosections were cut at a thickness of 20 μm using a cryostat Microm HM550-OMP (Thermo Fisher Scientific) and were then mounted on MAS-coated glass slides (Matsunami Glass, Osaka, Japan). The sections were permeabilized by ice-cold methanol, blocked by blocking buffer (10% BSA, 0.3% Triton X-100 in PBS) and were then incubated for overnight at 4°C with primary antibody diluted in blocking buffer. After washing with PBS, the sections were incubated with appropriate secondary antibody conjugated to Alexa 488 or 567; for marker staining of nuclei, the sections were incubated with Hoechst 33,242 in PBS for 2 min at room temperature, and then mounted with PermaFluor. For the staining of nestin, anti-nestin antibody was biotinylated by Biotin Labeling Kit (Dojindo, Kumamoto, Japan) according to the manufacturer’s instructions and visualized by FITC-streptavidin (Vector Laboratories Inc., Burlingame, CA). Images were obtained under a LSM710 with Axio Observer.Z1 (Carl Zeiss, Oberkochen, Germany) or AF 6000LX (LEICA, Wetzlar, Germany). Quantitative analysis was performed by Image J software (National Institutes of Health, Bethesda, MD). Quantification of colocalization in randomly selected cells (40∼50 cells) was performed by calculating Pearson’s correlation coefficient using ZEN 2012 software (Carl Zeiss).

Immunoblotting was performed as described (Honsho et al., 2015). Precision Plus Protein All Blue standards (BioRad, Hercules, CA) were used as molecular size markers. Immunoblots were developed with ECL prime reagent (GE healthcare) and detected by LAS-4000 Mini luminescent image analyzer (Fuji Film, Tokyo, Japan). The immunoreactive band intensities were quantified by Image Gauge software (Fuji Film).

Statistical analysis was performed using R software¹. All Student’s t-tests used were one-tailed. A P-value less than 0.05 was considered statistically significant.

The animal ethics committee of Kyushu University approved all animal experiments.

To investigate the effect of peroxisomal deficiency on the function of CNS in adult mouse, we generated a conditional Pex2-knockout mouse using tamoxifen-inducible Cre/loxP recombination system (Pex2^flox/flox/Cre-ER^+/–; Figure 1). Exon 7 of Pex2 gene containing the coding region of Pex2 protein was floxed with loxP sites (Figure 1A). Administration of tamoxifen gave rise to elimination of exon 7 from Pex2 gene (Figure 1B, lanes 5 and 6). At 2–3 weeks after the administration of tamoxifen, immunoblotting revealed that Pex2 protein level was partly but significantly decreased in conditional Pex2-knockout mouse hippocampus (Figure 2A, lane 2; Figure 2B). We analyzed the mutant mouse at a longer period of time post-tamoxifen administration and did not find any further reduction of Pex2 protein (data not shown). Therefore, we termed it Pex2-knockdown (KD) mouse. To assess the peroxisomal matrix protein import, we analyzed the proteolytic processing/conversion of peroxisomal matrix proteins including ADAPS, 3-ketoacyl-CoA thiolase, AOx, and SCPx. The precursors of these proteins including AOx-A chain were elevated in Pex2-KD mouse (Figure 2A, lane 2; Figure 2C). Apparently concomitant decrease of mature ADAPS, thiolase, and SCPx was discernible, while AOx B-chain level was not altered (Figure 2A, lane 2; Figure 2B), suggesting that peroxisomal matrix protein import was partially impaired. LC-MS/MS analysis revealed that the reduction of PlsEtn (Figure 2D) and accumulation of VLCFA in PC (Figure 2E), suggesting the attenuation of peroxisomal lipid metabolism. We also performed the immunofluorescent staining of hippocampal sections of Pex2-KD mouse using antibodies against peroxisomal matrix proteins. Some neurons showed that catalase was more discernible in the cytosol besides Pex14-positive enlarged peroxisome remnants, so called ghosts (Santos et al., 1988), thereby suggesting that catalase import was attenuated (Figure 3A, asterisks). Moreover, staining with antibodies against PTS1 (Figure 3B) showed defect of matrix PTS1-protein import in hippocampus of Pex2-KD mouse. Pex2-KD mouse also showed the reduced colocalization of a PTS2 protein ADAPS with Pex14 (Figure 3C). In both of the control and Pex2-KD mice, non-specific punctate structures were likely observed around nuclei, as previously reported (Abe et al., 2018). To evaluate the defect of peroxisomal matrix protein import, Pearson’s correlation coefficient was assessed for colocalization of matrix proteins including catalase, PTS1, and ADAPS with Pex14. Pearson’s correlation coefficients in three different stainings were significantly decreased in Pex2-KD mouse (Figures 3D–F). Taken together, in hippocampal neurons of Pex2-KD mouse, Pex2 is only partially depleted (Figure 2A), giving rise to mosaicked defects in peroxisome biogenesis (Figure 3A).

Next, to investigate whether peroxisomal abnormality affects memory in Pex2-KD mice, contextual fear conditioning test was performed (Figure 4A). On the training day, control and Pex2-KD mice were placed in a conditioning box and then received an electric foot-shock as an unconditioned stimulus. On days 1 and 7, Pex2-KD mice were frozen, but significantly less than control mice (Figure 4B). However, there was no difference in the distance traveled (Figure 4C), suggesting no deficit in motor function of Pex2-KD mouse. Therefore, these results suggested that Pex2-KD mice manifest memory disturbance.

Memory disturbance of Pex2-KD mouse suggested the impairment of hippocampal functions. The impairment of cognitive performance in hippocampal-dependent tests, including contextual fear conditioning, is caused by the reduction of neural stem cells (Cameron and Glover, 2015; Gonçalves et al., 2016; Cope and Gould, 2019). We analyzed the neurogenesis in dentate gyrus of adult Pex2-KD mouse using anti-nestin antibody. In adult hippocampus, nestin is expressed in type 1 cells, putative neural stem cells possessing radial processes, and type-2a cells, early neural progenitors (Yu et al., 2008; Gonçalves et al., 2016). Nestin-positive cells were aligned (Figure 5A, arrows) along subgranular zone (SGZ) and some cells protruded radial processes (Figure 5A, arrowheads) into granular cell layer (GCL). Statistical analysis revealed that the number of nestin-positive cells per 1 μm SGZ was significantly decreased in Pex2-KD mouse (Figure 5B). To evaluate the type 1 cells, nestin-positive processes were traced in the dentate gyrus (Figure 5C). The number of nestin-positive processes per 1-μm SGZ was also significantly reduced in Pex2-KD mouse (Figure 5D), suggesting the decrease of neural stem cells. These results suggested that biogenesis and/or functions of peroxisomes were essential for maintaining neural stem cells.

We also assessed the expression level of Bdnf in Pex2-KD mouse. BDNF protein (Figure 6A, lane 2; Figure 6B) and Bdnf mRNA levels (Figure 6C) were significantly elevated in hippocampus of the Pex2-KD mouse. Moreover, protein level (Figure 6A, lane 2; Figure 6B) and mRNA level (Figure 6C) of TrkB-T1 encoding an inactive truncated form of TrkB, but not full-length TrkB-TK+, were likewise elevated in hippocampus of Pex2-KD mouse. Expression of c-fos, a target gene of the BDNF-TrkB signaling pathway (Ip et al., 1993), was significantly reduced (Figure 6C), suggesting the attenuation of BDNF signaling. Immunohistochemical staining and its quantification revealed that BDNF expression was upregulated in hippocampus of the Pex2-KD mouse and well merged with CNPase-positive cells, oligodendrocytes (Figures 7A,B).

Peroxisome-defective Pex14^ΔC/ΔC mouse is a ZS model mouse that manifests neonatal death and CNS abnormality such as neuronal migration defect in cortex and cerebellar malformation (Abe et al., 2018). Pex14^ΔC/ΔC mouse shows the upregulation of BDNF and the TrkB-T1 in the cerebellum. To investigate whether abnormal expression of BDNF and malformation take place in hippocampus of neonatal Pex14^ΔC/ΔC mouse in vivo, we analyzed the BDNF and TrkB expression in the hippocampus region at postnatal day 0.5 (P0.5). Immunoblotting showed that hippocampal BDNF expression was not altered in neonatal Pex14^ΔC/ΔC mouse (Supplementary Figures S1A,B). Likewise, the expression of TrkB, including TrkB-TK⁺ and TrkB-T1, was not affected in Pex14^ΔC/ΔC mouse hippocampus (Supplementary Figures S1A,C). These results suggested that peroxisomes were not involved in the morphogenesis and the regulation of BDNF expression in the hippocampus during fetal development.