Does Seipin Play a Role in Oxidative Stress Protection and Peroxisome Biogenesis? New Insights from Human Brain Autopsies

Seipin is a widely expressed protein but with highest levels found in the brain and testes. Seipin function is not yet completely understood, therefore the aim of this study was to evaluate the expression of BSCL2 transcripts in the central nervous system (CNS) of humans and investigate the effect of their overexpression on a neuron model and their relationship with oxidative stress protection, as well as shed light on the pathogenic mechanisms of Celia's Encephalopathy. We analyzed the expression of BSCL2 transcripts using real-time RT-PCR in samples across the brain regions of subjects who underwent necropsy and from a case with Celia's Encephalopathy. The transcript encoding the long seipin isoform (BSCL2-203, 462 aa) is expressed primarily in the brain and its expression is inversely correlated with age in the temporal lobe, amygdala, and hypothalamus. Strong positive correlations were found between BSCL2 expression and some genes encoding protective enzymes against oxidative stress including SOD1 and SOD2, as well as peroxisome proliferator-activated receptor gamma (PPARG) in the amygdala. These results were experimentally corroborated by overexpressing BSCL2 transcripts in SH-SY5Y cells with lentiviral transduction and assessing their effects on neuron differentiated cells. Confocal microscopy studies showed that both seipin and PEX16 are closely expressed in the hypothalami of healthy human brains, and PEX16 was absent in the same region of the PELD case. We hypothesize that seipin has specific CNS functions and may play a role in peroxisome biogenesis.


INTRODUCTION
The BSCL2 gene encodes seipin, a resident endoplasmic reticulum (ER) protein with two transmembrane domains, a luminal loop, and two amino and carboxy-terminal tails in the cytoplasm.
PELD is an infantile neurodegenerative disease with a fatal prognosis before 9 years previously described by our group (Guillen-Navarro et al., 2013). Neurological regression of Celia's Encephalopathy begins at age 3-4, with early signs of severe myoclonic epilepsy, spastic tetraparesis, and severe encephalopathy leading to death before age 9. This disease is extraordinarily rare, and is a consequence of the BSCL2 gene c.985C > T variant in homozygosis or compound heterozygosis (Alaei et al., 2016;Guillen-Navarro et al., 2013). This variant gives rise to a branch site in exon 7 of the BSCL2 gene, producing the intronization of that exon leading to an aberrant seipin with a different amino acid sequence than wild-type (BSCL2-203) from exon 6 ( Guillen-Navarro et al., 2013). The formation of large aberrant seipin macroaggregates leading to ER stress and nuclear accumulation of this abnormal protein are the pathogenetic mechanisms ultimately responsible for Celia's Encephalopathy (Ruiz-Riquelme et al., 2015).
Seipin is a highly evolutionarily conserved protein functioning primarily to modulate the formation of lipid droplets and identified in yeast, Saccharomyces cerevisiae, Caenorhabditis elegans, zebrafish, fruit fly, and mammals (Cartwright and Goodman, 2012;Salo et al., 2016;Wang et al., 2016). While this gene is mainly expressed in the central nervous system (CNS), pituitary gland, and testis in humans (Guillen-Navarro et al., 2013;Magre et al., 2001), little is known about the function of seipin in the CNS. Recent studies in seipin KO mice have shown a particular action for this protein in the CNS that appears to be mediated by peroxisome proliferator-activated receptor gamma (PPARG) and that influences synaptic transmission primarily through a-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid receptors, and also promotes neuronal differentiation and modulates the behavior and motor skills of these animals (Ebihara et al., 2015;Li et al., 2015;Zhou et al., 2016). Here we evaluate the expression of the various BSCL2 transcripts across human brain regions and extraneural tissues and investigate the putative functions of seipin in the CNS.

EXPERIMENTAL PROCEDURES
This study was approved by the Ethics Review panel of Xunta de Galicia (Spain) and performed in accordance with the ethical guidelines of the Helsinki Declaration.

Tissue samples
Thirteen post-mortem donors were selected, seven men and six women (aged 21-86 years), primarily deceased by suicide, traffic accident, or natural causes (Table 1) in accordance with the current Spanish legislation, and tissue samples were obtained during autopsy conducted within 24 h of death. Brains were removed and sliced coronally into 12 sections. The frontal cortices in the third slice, parietal and temporal cortices with hippocampi in the seventh slice, occipital lobes in the eleventh, insular cortices, amygdalae, thalami, heads of caudate and putamen nuclei, and cerebellum samples were accurately dissected and collected separately from both cerebral and cerebellar hemispheres, along with the right and left sides of the hypothalamus, mesencephalon, protuberance, and medulla oblongata, as well as the dorsal and ventral spinal cord, anterior pituitary, vagus and trigeminal nerves, skeletal muscles (rectus abdominis and gastrocnemius), adipose tissue from Bichat's fat pad, abdominal subcutaneous tissue, visceral areas and lower limbs, renal cortex and medulla, and liver and gonad samples. These tissues were immediately submerged in liquid nitrogen after removal and subsequently stored at À80°C until RNA preparation. These dissections yielded a total of 44 tissues for real-time RT-PCR analysis.

RNA extraction and retrotranscription
Total RNA was isolated with the ReliaPrep TM RNA Tissue Miniprep System (Promega, Madrid, Spain). Frozen tissue samples (20 mg of each tissue sample, except for adipose and muscle tissues which required 100 mg) were weighed and then homogenized in 250 ml lysis buffer (LBA previously supplemented with 2% 1thioglycerol) using a Tissue Ruptor (Qiagen, Hilden, Germany). 750 ml of Trizol reagent (Invitrogen, Madrid, Spain) was added to the homogenized samples, which were then vortexed for 10 s and incubated for 5 min at room temperature to permit complete dissociation of the nucleoprotein complex. Adipose tissue samples required an additional centrifugation step at 12,000g for 10 minutes at 4°C to remove insoluble material, and the fatty layer was removed and discarded while the cleared supernatant was transferred to a new tube. 200 ml of chloroform (Sigma, Madrid, Spain) was then added to the supernatant, which was subsequently shaken vigorously by hand for 15 s and incubated for 2 min at room temperature. This was followed by a centrifugation at 12,000g for 15 min at 4°C, after which the aqueous phase was transferred to a new tube. 35 ml of 2-propanol (Sigma) was added for each 100 ml of supernatant, samples were vortexed for 5 s, and the lysate was transferred to a ReliaPrep column and centrifuged at 14,000g for 30 s at room temperature. Subsequent RNA purification was as follows: 200 ml of wash solution was added followed by a centrifugation step at 14,000g for 15 s at room temperature; 500 ml of RNA wash solution was added and followed by centrifugation at 14,000g for 30 s at room temperature; 300 ml of RNA wash solution was added followed by centrifugation at 20,000g for 2 min at room temperature. Finally, RNA was eluted with 30 ml of nuclease and RNase-free water in a last centrifugation step performed at 14,000g for 1 min at room temperature. The concentration and purity of each sample was determined by spectrophotometer (ND2000; Nanodrop), and RNA samples were stored at À80°C until use. The RNA was reverse transcribed using M-MLV reverse transcriptase (Invitrogen), as previously described (Victoria et al., 2010).

Real-time RT-PCR
Specific primers and probes designed by the Universal ProbeLibrary (Roche Diagnostics, Sant Cugat del Valles, Spain) were used in a Light Cycler 2.0 (Roche Diagnostics) to determine the specific expression of the following genes: CAT, NESTIN, PEX1, PEX11G, PEX16, PPARG, RBFOX3/NeuN, SOD1, SOD2 (Table 2), and three distinct BSCL2 transcripts (Fig. 1). Real-time RT-PCR conditions are available upon request. Results were normalized to the 18S and RNA polymerase II genes using the 2 ÀDD CT method (Livak and Schmittgen, 2001).

Lentiviral transduction
Cells were seeded into 6-well plates (cat. 3516, Corning, Costar) at a density of 5000 cells per cm 2 . The medium was removed twenty-four hours after seeding and the cells were washed with phosphate-buffered saline (PBS). Viral particles (wild-type seipin), and the empty vector as a control, all with functional titers >10 9 transducing units/ml produced by Cyagen Biosciences (Guangzhou, China) were added at a multiplicity of infection (MOI) of 200 viral particles/cell to a serum-free medium in the presence of Lentiblast A/B (1:100/1:1000, Oz Biosciences, Marseille, France). Serum was added 4 hours after initial infection (final infection volume 1 ml), the medium was removed twenty-four hours after infection, and the cells were washed with phosphatebuffered saline (PBS). Cells were then cultured with typical SH-SY5Y medium and puromycin dihydrochloride (2 lg/ml final concentration, cat. P8833, Sigma-Aldrich) was subsequently added to the cell culture medium every 2-3 days until resistant stable cells were formed. Cells were then routinely grown as described above.

Immunostaining
The hypothalamus of the homozygous index case c.985C > T and of two control cases (male and female, 59 and 75 years at death, respectively) were fixed previously with 10% neutral buffered formalin (Bio-Optica, cat. 05-K01022) for 24 h at room temperature and embedded in paraffin. After deparaffinization and rehydration, slides were pre-treated with PT-link (Dako, California, USA) for 20 min at 95°C in Tris/EDTA buffer, pH 9. Slides were then washed four times with PBS and incubated with 1:250 anti PEX16 antibody (Santa Cruz Biotechnology) and 1:125 anti-seipin antibody HPA042394 (Sigma-Aldrich, The Human Protein Atlas) overnight at 4°C. The next day, the slides were washed four times with PBS and incubated with Alexa Fluor 555 (Thermo-Fisher Scientific, cat. A31572) conjugated antimouse secondary antibody, and Alexa Fluor 488 (Thermo-Fisher Scientific, cat. A21202) conjugated antirabbit secondary antibody for 1 h in darkness, washed with double-distilled water. Slides were then treated to quench autofluorescence background with 0.1% Sudan Black B (Sigma, cat. 199664) in 70% ethanol for less than 5 min at room temperature and washed three times, for 5 min each with double-distilled water. Finally, they were mounted in aqueous medium containing 1:1000 DAPI (Sigma-Aldrich, cat. D9542).

Confocal fluorescence microscopy
Immunofluorescence staining was assessed with a Leica TCS SP2 confocal microscope using a HCX PL APO 63Â/1.3 glycerol-immersion objective and Leica Confocal Software (Leica Microsystems Heidelberg GmbH, Mannheim, Germany). Images were obtained by a sequential scan method and three different laser lines to avoid simultaneous excitation and possible overlap. Confocal acquisition of the fluorescence labels was performed as follows: DAPI, color-coded in blue (Blue-Diode, excited at 405 nm and recorded on 425-470 nm), Alexa Fluor 488 nm, color-coded in green (Argon laser, excited at 488 nm and recorded at 500-555 nm), and Alexa Fluor 555 nm, color-coded in red  (DPS diode, excited at 561 nm and recorded at 581-675 nm). Two random visual fields were analyzed for each group.

Statistical analysis
Real-time PCR analyses were done by duplicates, and statistical significance was determined using a nonparametric Kruskal-Wallis test followed by a Mann-Whitney U post-hoc with Bonferroni's correction. Correlations were tested using the Spearman R correlation coefficient and partial correlations were utilized to correlate two variables while adjusting for other one. Data are presented as mean ± SD with statistical significance set at p < 0.05. All statistical analyses were performed using SPSS for Mac (release 22.0; SPSS, Chicago, IL, USA).

Patterns of BSCL2 expression in human tissues
We evaluated the relative expression of seipin transcripts in the CNS, the peripheral nervous system (PNS), and extraneural tissues (ET) in 13 subjects (Fig. 3A, detailed with significance in Table 3). The total expression of BSCL2 was 6.9 times higher in the CNS than in any of the other tissues (p < 0.001): 86.3% in CNS, 1.3% in  Relative expression of the encephalic and non-encephalic post-mortem tissues in female and male subjects, * p < 0.05. All samples were analyzed in duplicate, n = 13. Results were normalized for the 18S gene. PNS, and 12.4% in ET (Fig. 3B), and the pituitary and the gonads had the highest total BSCL2 expression of the non-neural tissues (5.25% and 2.61%, respectively). The predominant transcript is BSCL2-203 (Â6.8), which encodes the long isoform and represents 86.6% of total BSCL2 transcript expression (Fig. 3B). BSCL2-203 was expressed most in the CNS (78.9% of total BSCL2 expression, and 91.2% of BSCL2-203 expression) (Fig. 3B, C) and was higher in the caudate nuclei and in the protuberance (9.2% and 9% of BSCL2-203 expression, respectively). BSCL2-203 expression in the PNS and ET was 0.9% and 7.9%, respectively (Fig. 3C). BSCL2-205/207/210 expression represented only 12.7% of total BSCL2 expression (Fig. 3B) and was expressed the most in ET and the PNS (42.6% and 4.1% of BSCL2-205/207/210 expression, respectively) (Fig. 3C). BSCL2-205/207/210 expression was increased 3.9Â and 2Â respectively in adipose and muscle tissues, compared to BSCL2-203, and BSCL2-201 expression was low in all tissues (0.7% of total BSCL2 expression, Fig. 3B). Within the CNS, the short transcript was primarily expressed in the caudate nucleus, the cerebellum, and the protuberance. Expression levels of BSCL2 transcripts across the primary brain divisions are summarized in Fig. 3C.

Influence of age on BSCL2 expression
A comparison of BSCL2 expression between young and aged individuals (median age 61 years, Table 5) shows that BSCL2 transcripts are generally expressed more highly in younger subjects in both encephalic and nonencephalic tissues. This observation is strongest for BSCL2-203 (Table 5), weaker for BSCL2-201, and weakest for BSCL2-205/207/210. An analysis of the correlations between BSCL2-203 transcript and age

À45
Relative expression of the encephalic and non-encephalic post-mortem tissues in young and old subjects, *: p < 0.05. All samples were analyzed in duplicate, n = 13. Results were normalized for the 18S gene.
In vitro studies. Since the observed correlations do not imply causality, we differentiate SH-SY5Y cells stably transfected with BSCL2 into a neuronal model cell line in order to measure the expression of genes significantly correlated with BSCL2 in human brains. Morphological changes were observed after 3-5 days (Fig. 6), and many SH-SY5Y cells stopped proliferating and expanded an extensive network of long neurites. At the end of differentiation (day 7), cells were evenly distributed and developed a more rounded shape, forming clumps of cell bodies linked by long processes similar to axons (Fig. 6C).
Confocal microscopy studies in human brains. Confocal microscopy was used to localize seipin and PEX16 expression in the hypothalami of healthy human brains  and the PELD case. Image analysis revealed that seipin and PEX16 are expressed in neurons in the hypothalamic nuclei of both controls but not the PELD index case (Fig. 9A-C). Interestingly, we only found cytoplasmic seipin in neurons (and not in astrocytes or oligodendrocytes). The main pattern observed in control brain regions was what we classified as the ''granular profile," in which seipin is scattered throughout the neuronal cytoplasm and PEX16 appears to be closely intertwined in the free spaces (Fig. 9A, B) very close to seipin, with a shape resembling vesicles (Fig. 9D). We also observed intranuclear PEX16 and seipin in control hypothalamus images (Fig. 9B). Finally, the seipin observed in neurons was localized in the cytoplasm in samples from Celia's Encephalopathy case, and the staining was clearly diffused within the neuronal body and the proximal segment of the axons. There were more seipin-containing nuclei in Celia's Encephalopathy case compared to the control, while no PEX16 was found in the hypothalamus of this patient (Fig. 9C).

DISCUSSION
This study describes the differential expression of the BSCL2 in the human brain and other tissues, and identifies BSCL2-203 as the major transcript in the CNS. Remarkably, expression of this transcript appears to be reduced in the more primitive regions of the brain (diencephalon, mesencephalon and rhombencephalon). BSCL2-205/207/210 is primarily expressed in the extraneural tissues, and the transcript responsible for the Celia-seipin (BSCL2-201) is expressed very little but mainly in the caudate, one of the nuclei affected earliest in Celia's Encephalopathy (Araujo-Vilar et al., 2018). These results indicate specific tissue functions for the different BSCL2 transcripts, and these results coupled with the observation that the 64 amino acids corresponding to exon 1 and part of exon 2 of the BSCL2-203 transcript are identical or quasi-identical in hominidae primates (100-98%), very similar in other catarrhine primates (95%), and similar in platyrrhine primates (86-89%), while similarities are reduced in prosimians (75-55%) and in mice (52%) (Fig. 10) suggest that the longer transcript should play some role in the encephalization process.
To the best of our knowledge, this is the first time that BSCL2 transcript expression is mapped in the different regions of human brain. In mice, Garfield et al. (2012) using in situ hybridization to map the location of Bscl2 mRNA-positive cells, found that the regions with strong expression were the basal forebrain, hippocampus, hypothalamus, dorsal and ventral brainstems, with low expression in the cortex, moderate expression in the midbrain and no detectable expression in the caudate putamen. On the other hand, Liu et al. (2016) found a similar expression pattern of seipin in the CNS using immuno-histochemical staining, however with some discrepancies like in the caudate putamen. While in our study, we found highest BSCL2 expression in caudate, putamen, cerebellum and protuberance, and mostly in   recently showed that PEX16 is a PPARG target gene whose expression regulates peroxisome number and lipid metabolism, and demonstrated that it is also required for adipogenesis (Hofer et al., 2017). Studies in rodents have shown the beneficial effects of the pharmacological activation of peroxisome proliferator-activated receptors in Parkinson's disease and dyskinesia models (Barbiero et al., 2014;Grover et al., 2013).
The profiles observed in the confocal images suggest that seipin is involved in peroxisome biogenesis, probably at an early phase. The juxtaposition of seipin and PEX16 observed in some neurons is consistent with the potential close proximity of peroxisomes and lipid droplet biogenesis areas (Binns et al., 2006), suggesting that peroxisomes and seipin are intimately associated. We can further speculate that peroxisomes with vesicle-like shapes may be pre-peroxisomal vesicles emerging from the ER. It should be noted that the structure of the lipid droplets and the peroxisomes are similar, both emerge from the ER, and there is a dialog between both organelles (Shai et al., 2016). In this line, very recently and during the course of the review of this manuscript, Wang et al. 2018 identify in yeasts Pex30 as a factor cooperating with seipin in the biogenesis of both lipid droplets and pre-peroxisomal vesicles from the ER. In our opinion, this recent Wang's paper reinforces our hypothesis about the role of seipin in peroxisomes biogenesis. On the other hand, peroxisomes are very abundant in neurons and play a key role in protection against oxidative stress, and are also critical for the synthesis of phospholipids. Finally, the peroxisome biomarker PEX16 is barely detectable in the brain of the PELD case, where seipin is also scarce. Taken together, these data may suggest that seipin is playing a role in peroxisome biogenesis as a regulator of peroxisomal protein sorting during the first steps of biogenesis from the ER, similarly to what happens in the nascent lipid droplets (Shai et al., 2016;Wang et al. 2018), or could act as an essential factor in the recruitment of other peroxisomal membrane proteins to the ER together with PEX16, which subsequently transports them to the peroxisomes (Kim and Mullen, 2013). Correlation studies of gene expression, lentiviral overexpression experiments, as well as immunohistochemical and localization studies have allowed us to consolidate our hypotheses regarding the pathogenetic mechanisms of Celia's Encephalopathy (Ruiz-Riquelme et al., 2015). Seipin appears to have specific functions in the CNS, and could act as a neuroprotector, modulating the expression of SOD and CAT via PPARG by unknown mechanisms and promoting the proliferation of peroxisomes,   thus decreasing free radicals and consequently protecting against neurodegeneration. The fact that seipin may be playing a role in peroxisome biogenesis makes them an interesting potential therapeutic target.