We used the previously reported Pex7 hypomorphic mouse model (Braverman et al., 2010) B6;129S6-Pex7 ^tm2.0Brav (named here as Pex7 ^hypo/hypo), to generate an allelic series, (B6;129S6-Pex7 ^tm2.2Brav) referred as Pex7 ^null/null and (B6;129S6-Pex7 ^tm2.3Brav) referred to as Pex7 ^hypo/null. The Pex7 null allele was generated by breeding CMV-driven Cre recombinase deleter mice, B6.C-Tg(CMV-cre)1Cgn/J, with Pex7 ^hypo/hypo mice, in which loxP sites flanked exon 3. Four different primer sets were used to distinguish the Pex7 ^hypo, Pex7 ^null and Pex7 ^WT alleles; unique primer combinations distinguished each genotype (see Supplementary Table S1 for primer list). Pex7 ^WT/null mice were bred to obtain Pex7 ^null/null mice. Pex7 ^WT/null females were bred with Pex7 ^hypo/hypo males to generate the intermediate Pex7 ^hypo/null mouse model. SNP genotyping showed a stable 87% 129S6/SvEvTac and 13% C57BL/6NCrl strain background (MaxBax, Charles River, Wilmington, MA). For PCR genotyping, ear notches were collected in 75 μl of alkaline lysis buffer (25 mM NaOH/0.2 mM Na₂EDTA) at 95°C × 30 min for digestion followed by neutralization with 250 μl of 40 mM Tris-HCl. Gnpat null mice (Gnpat ^tm1Just , MGI: 2670462) (Rodemer et al., 2003) were maintained on an outbred C57BL/6 × CD1 background. Experimental cohorts with Gnpat ^null/null and Gnpat ^WT/WT littermates were obtained by breeding Gnpat ^null/WT animals; genotypes were determined by PCR as described (Rodemer et al., 2003). Mice were housed in the animal facility at the RI-MUHC and the local animal facility of the Medical University of Vienna with 12 h of dark and light cycles. Animals were fed a commercial laboratory rodent diet [Teklad Global 18% Protein, Envigo, Canada (that does not have sources of PA listed)] and had free access to water. For all experiments, we used both males and females at ages 1, 4, and 12 months. Control groups were Pex7 ^WT/WT and Pex7 ^WT/hypo littermates for experiments involving Pex7-deficient mice and Gnpat ^WT/WT littermates for experiments involving Gnpat ^null/null mice. Euthanasia was performed by CO₂ preceded by 5% isoflurane anesthesia. For gene expression, protein studies and biochemical analysis, tissues (cerebral cortex, cerebellum, heart, liver, lung, and kidney) were collected, snap frozen in liquid nitrogen and stored at −80°C until analysis. Blood samples and separation of plasma from erythrocytes was prepared as described (Fallatah et al., 2019). For histology, whole brain tissue was harvested and fixed in 10% neutral buffered formalin at 4°C × 48 h, then transferred to 70% ethanol and processed for paraffin embedding. This study was conducted under a McGill University animal care committee approved protocol (#5538) or a license by the Austrian Federal Ministry of Science and Research (BMBWF-66.009/0174-V/3b/2019).

Total RNA was extracted from ∼20 mg of cerebral cortex or cerebellum using TRIZOL™ (Invitrogen). For cDNA synthesis, we used 1 μg of RNA and OneScript^® Plus cDNA Synthesis SuperMix reverse transcriptase (Abm, G453), and qPCR with EvaGreen 2X qPCR Kit (Abm™ II) and CFX96 Touch Real-Time PCR Detection System (Bio-Rad). Pex7 gene expression was normalized to murine hypoxanthine guanine phosphoribosyltransferase (Hprt) using Bio Rad software (Bio-Rad CFX Maestro). qPCR experiments were performed twice per sample; in each experiment, we used 3 biological and 2 technical replicates per genotype. Primer sets are listed in Supplementary Table S1.

Paraffin embedded brain sections were sagittally sectioned (7 μm) and used for protein detection. We used stainless-steel brain matrices to obtain uniform sagittal sections across different animals. Allen Mouse Brain reference Sagittal Atlas version 1, 2008 (Sunkin et al., 2013) was used as a reference to match levels of cerebral cortex sections and all sections were between levels 15–16 (0.875–1.10 mm lateral to midline). Brain sections were deparaffinized with xylene and rehydrated with decreasing concentrations of ethanol and rinsed in 1X phosphate-buffered saline (PBS). Following antigen retrieval and serum blocking, sections were incubated overnight with primary antibody. Secondary antibodies were applied, and sections processed by fluorescent labeling and mounting using ProLong Gold antifade reagent with DAPI (Invitrogen, Carlsbad, CA) for IF staining, or DAB following a Vectorstain Elite ABC kit (Vector Laboratories) and hematoxylin counterstaining for IHC staining. Images were detected using a Leica DM6000B microscope with DFC345FX camera and LASX software (Richmond Hill, Canada). Primary antibodies and concentrations used were: mouse monoclonal anti-msCalbindin-D-28K (1:200, C9848, Sigma Aldrich), rat monoclonal anti-Myelin Basic Protein antibody “MBP” (1:1000, ab7349, Abcam) and rabbit polyclonal anti-hsPEX7 (1:200, Abcam, ab167036). Secondary antibodies were: goat anti-Mouse IgG (H + L) Alexa Fluor Plus 488 (1:200, A32723, Invitrogen), Goat Anti-Rat IgG H&L HRP conjugated (1:200, ab205720, Abcam).

To calculate the numbers of calbindin-D28K-positive Purkinje cells, mid-sagittal cerebellar sections (0.875–1.10 mm lateral to midline) from 3-4 independent samples per genotype were analyzed. Purkinje cells were counted in three to four cerebellar slices in entirety per mouse genotype at 1, 4, and 12 months of age by two independent researchers using the cell counter plugin from ImageJ software (Schneider et al., 2012).

Cerebral cortex and cerebellar tissues from Pex7 deficient mice were homogenized using TissueLyser II (QIAGEN, cat#:85300) in RIPA buffer as previously described (Argyriou et al., 2019). Primary antibodies used were: rabbit polyclonal anti-beta-Tubulin (1:17,000, Abcam, ab6046), and rabbit monoclonal anti-PEX7 (1:1000 Abcam, ab134962). Visualization of membrane was performed using an Amersham 600 gel imager. Immunoblotting and densitometric analyses were performed as previously described (Argyriou et al., 2019; Fallatah et al., 2020).

Reagents used were authentic Pls standards, tetradeuterated internal standards 26:0-D4 lysoPC (Avanti Polar Lipids, Alabaster, Alabama), 16:0-D4 lyso PAF (Cayman Chemical Company, Ann Arbor, Michigan) and HPLC grade solvents (methanol, acetonitrile, chloroform, water) (Fisher Scientific, Waltham, MA), formic acid (Honeywell Fluka), ammonium formate (Sigma-Aldrich), and PBS (Thermo Fisher Scientific, Waltham, MA). LC-MS/MS analysis was performed in blood and tissues from Pex7 deficient mice as previously described (Fallatah et al., 2019). PlsEtn were detected by multiple reaction monitoring (MRM) transitions representing fragmentation of [M + H] + species to m/z 339, 361, 385, 389, and 313 for compounds with 18:1, 20:4.22:6, 22:4, and 16:0, at the sn-2 position, respectively. Lysophosphatidyl choline (LPC) species were detected by MRM transitions representing fragmentation of [M + H] + species to m/z 104.

Analysis was done in plasma, cerebral cortex and cerebellum from Pex7 and Gnpat deficient mice as well as rodent chow using standard methods in the Peroxisome Disease Laboratory, Kennedy Krieger Institute, Baltimore Maryland (Dacremont et al., 1995; Dacremont and Vincent, 1995; Krauß et al., 2017).

Analysis was performed on mice half cerebrum tissues without cerebellum using High-Performance Liquid Chromatography (HPLC) with electrochemical detection (monoamines) or fluorometric detection (amino acids) as previously reported (Hörtnagl et al., 1991; Peneder et al., 2011; Dorninger et al., 2019b).

Data analysis was done using the GraphPad Prism software package (version 9.0) (GraphPad Software, La Jolla, United States). Statistical analysis was performed using one way or two-way analysis of variance (ANOVA) followed by appropriate correction tests for multiple comparisons. Simple linear regression was performed for correlation analysis between specific variables. Data are shown as the mean ± SD and statistical significance was set at p < 0.05. Since we did not observe any significant differences between male and female animals in biochemical and behavioral results, the data obtained from both genders were combined in the final statistical analysis.

The Pex7 hypomorphic mouse model contains a neo cassette in reverse orientation in intron 2 and lox sites flanking exon 3 (Braverman et al., 2010). The intronic neo cassette reduces full length transcript levels and the lox sites were used to generate the null allele; thus all mice were generated from the same construct (Meyers et al., 1998). To determine Pex7 transcript levels in our Pex7 deficient mouse series, we performed quantitative RT-PCR analysis in cerebral cortex, cerebellum, liver, lung and kidney tissues. The average Pex7 gene expression levels for Pex7 ^hypo/hypo and Pex7 ^hypo/null, were 0.394 ± 0.3% and 0.174 ± 0.1% of wild type Pex7 transcript, respectively. Pex7 transcript was not detected in any tissues from Pex7 ^null/null (Table 1).

To investigate the endogenous amounts of PEX7 protein in Pex7 deficient mice, immunoblot analysis using PEX7 antiserum was performed on brain tissue homogenates. PEX7 protein could not be detected in the brain tissue homogenates from Pex7 ^hypo/null or Pex7 ^null/null mice. However, with higher protein quantity and longer exposure, traces of PEX7 protein were detected in Pex7 ^hypo/hypo mice (Figure 1).

The distribution of PEX7 was determined by IHC. PEX7 expression was markedly localized to cerebellar Purkinje cells. Evaluation of cerebellar tissue sections of Pex7 ^hypo/hypo and Pex7 ^hypo/null at 1 month of age revealed a global reduction in PEX7 expression within Purkinje cells. PEX7 expression was completely undetectable in cerebellar Purkinje cells of the Pex7 ^null/null mice (Figure 2). These observations suggested that PEX7 is highly expressed in Purkinje cells and thus could be critical to their normal physiological function.

Survival data of Pex7 deficient mice showed survival rates of 100% and 91.2% in Pex7 ^hypo/hypo and Pex7 ^hypo/null, respectively, over the 150 days observation period. However, only 22.7% of the Pex7 ^null/null mice survived beyond 21 days. Around 45% of Pex7 ^null/null pups died within the first 3 days of life and 32% died before weaning (Figure 3A). Close observation of the latter showed poor weight gain in general and reduced activity a few days before death; the exact cause of death remains unknown. The Pex7 ^hypo/hypo and Pex7 ^hypo/null animals weighed 70%–80% and 67%–78% of their control littermates respectively. However, the Pex7 ^null/null mice weighed only 47%–53% of littermate controls. Measurements were performed at 1, 4 and 12 months of age (Figure 3B). Although body length was not measured, the mice appeared smaller in size, and their length appeared proportional to the reduction in weight.

Pex7 deficiency results in a complex biochemical phenotype of Pls and DHA deficiencies in addition to elevations in VLCFA and PA (Brites et al., 2003; Brites et al., 2009; Braverman et al., 2010). To determine whether these biochemical phenotypes were specific for RCDP1 due Pex7 deficiency or associated with other RCDP types, we concurrently examined the biochemical profiles of plasma and brain tissues from 4-month-old Gnpat ^null/null mice. We found that Gnpat ^null/null mice were deficient in total PlsEtn and DHA levels. No accumulation of either C26:0-LPC or PA levels was found in plasma or brain tissues from Gnpat ^null/null mice compared to their littermate controls (Supplementary Figure S2).

Hematoxylin and eosin staining of cortical brain sagittal sections did not show any gross anatomical differences between Pex7 deficient mice and their littermate controls at 4 months of age (data not shown). However, a reduction in the number of Purkinje cells on cerebellar sagittal sections was observed in Pex7 ^null/null mice and investigated further.

HPLC analysis was performed to examine the levels of the major monoamine neurotransmitters dopamine (DA), norepinephrine (NE) and serotonin (5-hydroxytryptamine; 5-HT) in addition to the amino acid neurotransmitter gamma aminobutyric acid (GABA) in whole brain tissue homogenates from Pex7 deficient mice. Overall, we found a trend toward reduction in the mean levels of brain neurotransmitters across all genotypes of the Pex7 deficient mouse series. The deficiency in DA and 5-HT levels was robust and significant in all Pex7 deficient mice and the reduction of NE and GABA was significant in Pex7 ^hypo/null and Pex7 ^null/null mice. We also observed a tendency toward gradual decrease in the levels of NE and 5-HT from mild to severe Pex7 deficient mice (Figure 10B).

Additionally, we measured the baseline levels of metabolites derived from monoamine neurotransmitters in whole brain tissue homogenates from Pex7 deficient mice and calculated the ratio of the metabolite to its respective transmitter. This ratio reflects the turnover of monoamine neurotransmitters. No significant differences were observed between Pex7 deficient mice and their littermate controls for the following metabolites: 5-hydroxyindoleacetic acid (5-HIAA), the main metabolite of 5-HT, 3,4-dihydroxyphenylacetic acid (DOPAC) and homovanillic acid (HVA), the primary metabolites for DA. Investigation of the ratio of the metabolites for the monoamine neurotransmitters revealed a significantly higher DOPAC/DA and 5-HIAA/5-HT ratios in Pex7 ^hypo/hypo and Pex7 ^null/null while the HVA/DA ratio was significantly higher only in Pex7 ^hypo/hypo mouse model. This increase in the metabolite/neurotransmitter ratios mainly results from the significant reduction in the levels of neurotransmitters without alteration in their corresponding metabolites (Supplementary Figure S4).

We repeated the open field test to evaluate the baseline activity levels in a subgroup of adult Pex7 deficient mice prior to brain neurotransmitter analysis. Our results confirmed the hyperactive behavior in the Pex7 hypomorphic mouse series compared to littermate controls (Figure 10A). Following the neurotransmitter analysis, a statistical correlation analysis was made to investigate the relation between the neurotransmitter deficits in cerebral cortex and activity levels. We found that the hyperactivity phenotype in Pex7 deficient mice is inversely correlated with DA, NE, 5-HT, and GABA levels in cortical brain tissues (Figure 10C). Additionally, we observed a significant inverse correlation between hyperactivity and PlsEtn levels in plasma and cerebral cortex of Pex7 deficient mice (Figure 10D).

We further investigated the interrelation between brain neurotransmitter measurements and total PlsEtn levels in the cerebral cortex and plasma samples from Pex7 deficient mice. We found that brain neurotransmitters deficits are positively and significantly correlated with the degree of Pls deficiency in cerebral cortex and plasma from Pex7 deficient mice (Figure 11).

Gene expression studies of Pex7 ^hypo/hypo and Pex7 ^hypol/null revealed extremely low levels (<1%) of Pex7 transcript in all tissues evaluated and undetectable Pex7 transcript in Pex7 ^null/null. However, the high cycle quantification (Cq) values associated with Pex7 relative expression (above 30) in the Pex7 hypomorphic models, suggests that Pex7 transcript levels were extremely low and could be beyond the limit of detection in qPCR analysis (Taylor et al., 2010). Consistent with the reduced transcript levels observed, PEX7 protein levels were very low in tissues from the Pex7 ^hypo/hypo mice, and undetectable by immunoblotting in either Pex7 ^hypo/null or Pex7 ^null/null. Although Pex7 hypomorphic mice exhibited very low Pex7 transcript and protein levels, the phenotypic manifestations were dramatically improved in comparison to Pex7 ^null/null. Overall, these data suggest that small increases in Pex7 transcript and/or protein can markedly improve the clinical phenotype, which might be substantially beneficial to accelerate future therapeutic interventions.

Neurobehavioral disorders including autism spectrum disorder (ASD) and attention deficit hyperactivity disorder (ADHD) were reported in patients with milder forms of RCDP (Bams-Mengerink et al., 2013; Yu et al., 2013). In addition, we found that 56% of mild RCDP individuals have behavioral abnormalities (Fallatah et al., 2020). Our results revealed that Pex7 deficient mice displayed a hyperactive behavior, which also matches the previous report of a hyperactivity phenotype in the Gnpat ^null/null (RCDP2) mouse model (Dorninger et al., 2019b). Pls deficiency could be a contributing factor in the development of this hyperactive behavior through the role of PLs in regulating brain neurotransmitter homeostasis.

In this study we detected significant deficiencies in major brain neurotransmitters including DA, 5-HT, NE, and GABA in brain homogenates from Pex7 deficient mice with a significant correlation to activity levels. Also, we found that deficits in NE and 5-HT were strongly correlated with Pls levels and thus the severity of Pex7 genotype. Similar neurotransmitter deficits were reported in Gnpat ^null/null mice and were proposed to be caused by defects in presynaptic neurotransmitter release and an impairment in vesicular uptake (Dorninger et al., 2019b). Brodde et al., (2012) showed an impairment in presynaptic release of glutamate and acetylcholine neurotransmitters in murine nerve terminals (synaptosomes) isolated from Gnpat ^null/null mice with reduction in exocytosis of endogenous acetylcholine and glutamate by 50%—60% of normal control synaptosomes. Additionally, in vitro radiolabeled neurotransmitter release studies of slices from hippocampus and parietal cortex of Gnpat ^null/null mice showed 22%–30% reduction in [³H] norepinephrine release upon chemical and electrical stimuli. Decreased levels of vesicular monoamine transporter type 2 (VMAT2) were observed in striatum from Gnpat ^null/null mice. VMAT2 is responsible for transporting monoamines from the cytosol to presynaptic vesicles and its reduction could affect the transport of neurotransmitters into synaptic vesicles therefore causing the depletion in monoamine neurotransmitters in Pls deficient mice (Dorninger et al., 2019b). Additionally, the hyperactive phenotype was previously reported in non-peroxisomal mouse models with deficiency in brain neurotransmitters including dopamine (Giros et al., 1996), 5-HT (Whitney et al., 2016) and GABA (Chen et al., 2015), which further supports the hypothesis of a direct role of brain neurotransmitter deficits in the manifestation of hyperactivity. For the future, it is possible to investigate RCDP patients for neurotransmitter deficits - urine, plasma or CSF can be analyzed for 5-HT, DA, NE as well as neurotransmitter metabolites (5-HIAA, HVA).

In the nervous system, Pls are considered as a critical source of the polyunsaturated omega-3 essential fatty acid DHA (22:6 n−3) that plays a crucial role in brain development, myelination and function (Hishikawa et al., 2017). In a prior cross-sectional study of 53 school age children with ADHD and age-matched normal controls, a lower proportion of plasma levels of DHA 22:6 n−3 was found in ADHD group (Burgess et al., 2000). Additionally, a deficiency of DHA in rhesus monkeys was associated with a stereotyped behavior (Reisbick et al., 1994). Also, previous studies indicated a deficiency of DHA in brain tissues and erythrocytes from Gnpat ^null/null (Rodemer et al., 2003) and hypomorphic Pex7 mice (Braverman et al., 2010). In the current project, we show a reduction in all Pls in addition to DHA-containing Pls in brain tissues from Pex7 deficient mice using LC-MS/MS. We propose that DHA deficiency might contribute to the development of the hyperactivity phenotype observed in Pex7 deficient mice.

To summarize, we engineered an allelic series of Pex7 deficient mice based on reductions of a normal Pex7 transcript, that mimics the human RCDP1 spectrum. Pex7 deficient mice showed genotype-phenotype correlations in terms of survival, body weight, peroxisome metabolites (Pls, C26:0-LPC, and PA) neurochemical, neuroanatomical and neurobehavioral (neurotransmitter levels, myelination, Purkinje cells numbers, and activity levels) manifestations. This graded genotype-phenotype correlation provided a type of “dose response” that strengthened the association of the phenotypes with Pex7 deficiency. The marked improvement in survival and body weight with only marginal changes in Pex7 transcript and PEX7 protein levels from the null to the hypomorphic model should facilitate the efficacy of any future molecular therapies in RCDP.

Our results of the hyperactivity behavior in Pex7 deficient mice suggest that reduced Pls levels play a significant role in the development of behavioral dysfunction associated with RCDP. This hyperactivity is sensitive to small reductions in Pls levels and could represent the neurobehavioral phenotype observed in mild (nonclassic) RCDP patients. We recently showed that oral supplementation of a mature synthetic Pls improved Pls levels in plasma, erythrocyte, heart, liver, small intestine and skeletal muscle tissues from Pex7 ^hypo/null mice (Fallatah et al., 2019). Although we did not observe an augmentation in Pls levels in the cerebral cortex and cerebellum, mature Pls effectively normalized the hyperactive behavioral phenotype in Pex7 deficient mice.

Our data highlight the significant correlation that was observed in Pex7 deficient mice between the hyperactivity, brain neurotransmitter deficits and Pls deficiency. These key findings emphasize the mechanistic role of Pls deficiency in regulating brain function and behavior. While the underlying mechanisms of PEX7 defects as well as Pls deficiency in complex brain behavior is still unknown, we speculate that Pls deficiency results in altered membrane lipid composition that could directly influence the basic physiological function of cellular membranes in the nervous system. Additionally, Pls deficiency might affect the overall cellular signaling as well as intracellular pathways and synaptic release and transmission, which might lead to impairments in brain function.