Age-related changes in regiospecific expression of Lipolysis Stimulated Receptor (LSR) in mice brain

The regulation of cholesterol, an essential brain lipid, ensures proper neuronal development and function, as demonstrated by links between perturbations of cholesterol metabolism and neurodegenerative diseases, including Alzheimer’s disease. The central nervous system (CNS) acquires cholesterol via de novo synthesis, where glial cells provide cholesterol to neurons. Both lipoproteins and lipoprotein receptors are key elements in this intercellular transport, where the latter recognize, bind and endocytose cholesterol containing glia-produced lipoproteins. CNS lipoprotein receptors are like those in the periphery, among which include the ApoB, E binding lipolysis stimulated lipoprotein receptor (LSR). LSR is a multimeric protein complex that has multiple isoforms including α and α’, which are seen as a doublet at 68 kDa, and β at 56 kDa. While complete inactivation of murine lsr gene is embryonic lethal, studies on lsr +/- mice revealed altered brain cholesterol distribution and cognitive functions. In the present study, LSR profiling in different CNS regions revealed regiospecific expression of LSR at both RNA and protein levels. At the RNA level, the hippocampus, hypothalamus, cerebellum, and olfactory bulb, all showed high levels of total lsr compared to whole brain tissues, whereas at the protein level, only the hypothalamus, olfactory bulb, and retina showed the highest levels of total LSR. Interestingly, major regional changes in LSR expression were observed in aged mice which suggests changes in cholesterol homeostasis in specific structures in the aging brain. Immunocytostaining of primary cultures of mature murine neurons and glial cells isolated from different CNS regions showed that LSR is expressed in both neurons and glial cells. However, lsr RNA expression in the cerebellum was predominantly higher in glial cells, which was confirmed by the immunocytostaining profile of cerebellar neurons and glia. Based on this observation, we would propose that LSR in glial cells may play a key role in glia-neuron cross talk, particularly in the feedback control of cholesterol synthesis to avoid cholesterol overload in neurons and to maintain proper functioning of the brain throughout life.


Introduction
Cholesterol is essential for neuronal physiology; a tight regulation of cholesterol homeostasis in the central nervous system (CNS) is essential for proper neuronal development and function. Cholesterol is not only an important structural component for cellular membranes and myelin, it is also required for synapse and dendrite formation [1,2], and axonal guidance [3]. It ensures functional synaptogenesis and is vital for: synaptic vesicles transmission along axonal microtubules via cholesterol-kinesin interactions, exocytotic complex organization in active presynaptic membranes of lipid rafts, neurotransmitter receptors clustering in postsynaptic membranes, extra-synaptic receptors pool recruitment, and pre-and post-synaptic cell-cell adhesion [4]. Cholesterol is delivered to tissues via lipoprotein particles. However, access to the brain of cholesterol and other lipids from the peripheral circulation is limited due to the blood-brain barrier (BBB), which serves as a selective low-permeable multicellular barrier [5]. Cholesterol is therefore supplied by de novo synthesis in the brain, which relies on its own network for synthesizing, internalizing and metabolizing these lipids to provide the necessary components for neuronal cell membrane function [5,6]. Glial cells play a central role towards providing neurons with lipids, particularly cholesterol in the form of lipoproteins. Lipoproteins in the cerebrospinal fluid (CSF) are very different from those of the periphery, and have been characterized as high-density lipoprotein-like (HDL-like) particles containing primarily apolipoprotein (Apo)E and ApoJ [7]. These HDL-like lipoproteins are needed to export cholesterol from astrocytes to neurons, where they bind via ApoE to lipoprotein receptors and are internalized through receptor-mediated endocytosis [4]. A series of lipoprotein receptors expressed in neurons have been identified including the low density lipoprotein receptor (LDL-R) [8], low density lipoprotein receptor-related protein 1 (LRP-1) [8], and lipolysis stimulated lipoprotein receptor (LSR) [9]. LSR is the most recently discovered receptor found to be expressed in the CNS [9]. It is a multimeric protein complex that undergoes conformational changes in the presence of free fatty acids, thereby revealing a binding site that recognizes ApoB or ApoE [10]. There are three different isoforms of LSR that have been clearly identified, LSR α and α' that form a doublet at 68 kDa, and β at 56 kDa [11]. LSR α is the complete protein sequence of LSR, which contains a clathrin binding site and a di-leucine lysosomal targeting signal at the N-terminal, a hydrophobic transmembrane domain, a cysteine-rich region and a group of alternatively negatively and positively charged amino acids at the C-terminal which serve as the lipoprotein binding site [11]. While LSR α' has a similar protein sequence to LSR α, the di-leucine lysosomal signal is deleted, but it can still act as a transmembrane protein. However, the LSR β protein sequence does not contain either the transmembrane domain or the dileucine routing motif responsible for internalization and endocytosis; despite this, LSR β is still able to bind lipoproteins in the presence of fatty acids [11]. LSR is plays an important role in mediating hepatic clearance of triglyceride-rich ApoB, E-containing lipoproteins during the post-prandial phase [12]. In vivo studies have shown that this receptor is necessary for maintaining normal peripheral circulation levels of cholesterol and triglycerides, and in contributing to the regulation of lipid distribution amongst the peripheral tissues [12,13]. Hepatic LSR's role in clearance of lipoproteins was further confirmed by the observation of increased plasma levels of cholesterol and triglycerides following shRNA-mediated knockdown of hepatic LSR expression in mice [13]. Complete inactivation of lsr is associated with in utero lethality at the embryonic stage, most likely due to brain-localized hemorrhages and BBB leakage [14,15]. Complete inactivation of lsr is embryonic lethal, however, in vivo studies conducted on young and aged lsr +/-mice suggest that reduced LSR may be associated with cognitive disturbances related to reactivity to novel environments in aged lsr +/-mice [9]. A significant decrease of lipid droplets, which are lipid-rich cellular organelles that regulate the storage and hydrolysis of neutral lipids, including cholesterol [16], was observed in Purkinje cells of the cerebellum (CB) together with an accumulation of filipin-labeled cholesterol in neuronal membranes of the hippocampus (HIP) in aged lsr +/-mice [9]. Histochemical studies show a neuron-specific strong expression of LSR in HIP, Purkinje cells, at ependymal cells surface between brain parenchyma and CSF, and in the capillary-rich choroid plexus region [9]. Interestingly, Daneman's laboratory identified the presence of lsr gene transcript in endothelial cells (ECs) of the BBB [17] and showed that LSR is a component of paracellular junctions highly enriched in the BBB ECs, but not in ECs in peripheral tissues outside the CNS [15]. They demonstrated that the BBB doesn't seal during embryogenesis in lsr knockout mice. Another study reported the high expression of LSR in tricellular junctions, not only in the BBB, but also in retinal ECs that form the inner blood retinal barrier (BRB) [18]. This indicates that LSR plays a critical role in maintaining the BBB integrity and suggests a potential role of LSR in the transport of lipoproteins between the brain parenchyma and the CSF. Altogether, the evidence point towards a critical role for LSR in cholesterol trafficking in the CNS during the lifespan. Our objective was to establish a detailed profile of LSR RNA and protein expression in the brain of young and old mice, both on whole brain tissue and in specific CNS areas including the hypothalamus (HT), hippocampus (HIP), olfactory bulb (OB), retina (RET), cortex (CX), and cerebellum (CB). We also compared LSR expression between primary cultures of glial and neuronal cells from different CNS regions. This study revealed differential expression of LSR isoforms in different regions, which would help us, in the future, understand the possible role of LSR in the cholesterol crosstalk between glial cells and neurons.

Animals
Three month and eighteen-month-old male and female C57Bl/6JRj mice (Janvier Breeding, Le Genest Saint Isle, France) were used for the study (n = 3 for each group). For primary cell cultures, newborn C57Bl/6JRj mice aged 5-7 days were sacrificed (n = 7-10). The C57Bl/6JRj mice were housed in certified animal facilities (N˚B54-547-24) on a 12-h light/dark cycle with a mean temperature of 21-22˚C and relative humidity of 50 ± 20% and provided rodent chow diet (16.4% protein, 4% fat, ref 2016, Envigo Teklad, Gannat, France) and water ad libitum. Animal care followed French State Council guidelines for the use and handling of animals: all tissues used in the study were collected after sacrifice of animals using isoflurane anesthesia followed by decapitation in order to preserve the nervous structures for further histological analysis. In the frame of a larger behavioral study, this protocol has been approved by the Local Ethical Committee (CELMEA, agreement number APAFIS#12079-201711081110404v2). All personals in contact with animal has been trained and detain an Authorization for Animal Experimentation from the French Authorities.

Immunoblots
Tissues were collected, and different brain regions were isolated: HT, HIP, OB, RET, CX, and CB. Whole cell extracts were prepared using RIPA lysis Buffer (10x RIPA buffer, ref 20

RNA extraction and RT-qPCR
Freshly collected tissues were conserved in RNAlater (ref 76104, Qiagen, Les Ulis, France) as per manufacturer's instructions and stored at -80˚C until use. Different regions of the brain were isolated separately including HT, HIP, OB, RET, CX, and CB. Total RNA was extracted using TRI reagent (ref T9424, Sigma Aldrich), according to the manufacturer's instructions. RNA quantity and purity were estimated by a Nanodrop ND-1000 spectrophotometer (Thermo Scientific; Villebon-sur-Yvette, France), and the samples with a 260/280 nm ratio � 1.7 were used for subsequent analyses. RNA quality was verified by bleach agarose gel electrophoresis [19]. RNA samples showing intact 28S and 18S ribosomal subunits were considered suitable for further cDNA synthesis. Reverse transcription was performed using 1 μg of RNA in a final volume of 20 mL including 0.5 mL of random primers ( . The thermal cycling conditions were: initial 5 min denaturation at 95˚C, followed by 42 cycles of 15 s at 95˚C, 1 min at 60˚C, and a final dissociation step. The primer specificity was determined based on the presence of a single peak in the melting curve. We followed four target mRNA sequences: total lsr, lsr α, lsr α', and lsr β, whose expression levels were compared to those of three reference sequences: hypoxanthine guanine phosphoribosyl transferase (Hprt) [12], phosphoglycerate kinase 1 (Pgk1) [20], and transferrin receptor protein 1 (Tfrc1) [20] (S1 Table). Lsr primer sequences were selected using the Primer-BLAST Genbank based on lsr gene sequence (NM_017405). Quantitation was performed by the 2 -ΔΔCt method [21]. The obtained results were tested for statistical significance (p<0.05) using the Relative Expression Software Tool 2009 (REST Version 2.0.13). Fold changes of mRNA samples of young animals were compared to whole brain levels, and those of old 18-month-old animals were compared to that of 3-month-old ones.

Mixed cell culture
Different brain structures (HT, HIP, OB, RET, CX, and CB) were collected from mice (n = 7-10 mice) of 5-7 days of age immediately after sacrifice. Tissues were collected in D-PBS and were cut into small pieces of 1 mm 3

Immunoisolation and culture of CNS neurons from postnatal mice
Neurons were isolated from HT, HIP, RET, OB and CB of freshly sacrified postnatal C57Bl/6J mice (n = 7-10) of 5-7 days of age with an immunopanning technique, according to a previously published protocol [22]. Mice were sacrificed by decapitation according to institutional guidelines. Different tissues were collected in D-PBS and were cut into small pieces of 1 mm 3 . Afterwards, tissues were digested using papain digestion solution, which contains 33 U/mL papain

Statistical analysis
For immunoblots: the area of bands was calculated using Image J, then the ratio LSR/β-TUB was calculated for each lane, followed by calculation of the mean and standard error. Statistical significance was calculated using t-test ± SEM.
For RT-qPCRs, statistical data in the boxplot were obtained using REST software tool, where (+) represents the mean value, the middle line represents the median, the lower (Q1) and upper (Q3) lines in the bar represent the 25% and 75% quartile, respectively. While the upper and lower lines represent the observations outside the 9-91 percentile range, data falling outside of Q1 and Q3 range are plotted as outliers of the data.

Lsr RNA profiling in mice brain
Variation of lsr mRNA levels in different CNS areas. Expression levels of lsr mRNA were estimated after performing RT-qPCR on total RNA fractions extracted from different brain regions. In young 3-month-old males, we observed a more than 4-fold increase in total lsr expression in both the HT (p = 0.007) and HIP (p = 0.007), while in the OB (p = 0.007) and CB (p = 0.03) there was a 2-2.7 fold increase relative to that measured in whole brain tissues of male mice of the same age (Fig 1A). When comparing different splice variants of lsr, lsr β mRNA expression was significantly higher than that of the whole brain in all of the abovementioned regions (Fig 1B). There was more than 7-fold expression in the HT (p = 0.007), HIP (p = 0.003), OB (p = 0.007), and CX (p = 0.017), as compared to a 2-3 fold increase in RET (p = 0.01), and CB (p = 0.026). Concerning lsr α, there was a 2-3-fold increase in both the HT (p = 0.006), and HIP (p = 0.006). Finally, there was a 2-fold increase of lsr α' in CB (p = 0.013), a 3.5-4.5-fold increase in both the HT (p = 0.013) and HIP (p = 0.009), and a 15-fold increase in the OB (p = 0.013).
Age-related changes in lsr expression. To determine if lsr expression was modified with age, RT-qPCR analysis was also performed on brain regions from older 18-month old mice, and results were expressed compared to 3-month old males (Fig 2). There was a significant decrease in total lsr mRNA expression in both HT (p = 0.032) and HIP (p = 0.039) where levels  were 0.1-0.2-fold as compared to those of 3-month-old mice (Fig 2A), while lsr RNA expression remained relatively unchanged in the other structures. When considering the different variants (Fig 2B), lsr α' mRNA expression was significantly reduced in the HT (0.269-fold, p = 0.032) and HIP (0.27-fold, p = 0.0001). Similarly, lsr β showed a tendency to decrease with age in both HT (0.073-fold, p = 0.14) and HIP (0.138-fold, p = 0.094), but this was not statistically significant (Fig 2B). Although total lsr mRNA levels in the OB tended to increase in the older mice (1.469-fold, p = 0.07, Fig 2A) and lsr α levels were slightly increased (1.251-fold, p = 0.288), lsr α' (0.115-fold, p = 0.03), and lsr β (0.273-fold, p = 0.03) levels were significantly decreased (Fig 2B).
RT-qPCR analysis of total lsr mRNA levels from brain regions of young female mice revealed no statistically significant differences as compared to those of male mice in RET, CB or HIP. However, total lsr mRNA was 4-fold higher (p = 0.03) in the CX region, and 4.5-fold lower (p = 0.03) in HT region of young female as compared to young male mice, while total lsr mRNA was 4.5-fold lower in females compared to young male mice (S1

LSR differential expression in different brain regions and the retina
In order to assess LSR protein levels, immunoblots using anti-LSR antibody (LSR Sigma) to detect LSR were performed on protein extracts from different regions of the brains of 18-month old mice, including the HT, HIP, OB, RET, CX, and CB. Interestingly, comparison of LSR protein levels revealed differences in those of LSR subunits α/α' (68 kDa), and β (56 kDa) in the various CNS regions. Anti-LSR Sigma antibodies (S4 Fig) detected the major two bands corresponding to the α/α' isoforms, seen as a doublet at 68 kDa, while the β isoform was identified as the lower band migrating at 56 kDa. Interestingly, the different isoforms of LSR (α, α', and β) were not equally expressed throughout the CNS. To examine this, the differences in the relative amounts of the three LSR isoforms (Fig 3, immunoblots and table) or total LSR protein normalized to β-TUB in brain regions from old versus young male mice were compared (Fig 3, bar graphs). For example, the LSR β-chain represents only 10% of total LSR in the HIP (Fig 3B), and more than 50% in the CX (Fig 3E).

LSR expression in neurons and glial cells
In view of these results, we performed immunocytostaining of LSR in pure neuronal vs mixed cultures from other CNS regions (Fig 4A). LSR was detected in neurons isolated from HIP, HT, OB, and RET (Fig 4A). LSR staining was observed primarily around cell soma, but some neurites were also weakly stained. On the other hand, immunocytostaining of glial cells from these same regions revealed colocalization of LSR with the glial cell marker, GFAP, as well as the astrocyte marker AQP4 (Fig 4B).

Glial cells highly express LSR in cerebellum
Immunocytostaining of neuronal CB cultures with anti-β-TUB III, a specific neuronal marker and anti-LSR X-25 revealed LSR expression in CB that was relatively weak and mostly located in the soma (Fig 5A). On the other hand, immunocytostaining of glial cell cultures, revealed a strong expression of LSR in this cell type (Fig 5B), with colocalization of GFAP and LSR. RT-qPCR analysis on three different CB cell cultures of neurons and glia were then performed to verify these results (Fig 5B), and confirmed higher lsr expression in glia (0.959-fold, p = 0.35), as compared to that in neurons (0.02-fold, p = 0.0001), when normalized to cerebellar mixed cell cultures. Furthermore, LSR isoforms were differentially expressed in glia, where lsr α was upregulated (1.76, p = 0.017, Fig 5B), while both lsr α' (0.471, p = 0.0001, Fig 5B) and β (0.441, p = 0.019, Fig 5B) were downregulated relative to cerebellar mixed cultures.

Conclusions and discussion
Here we demonstrate that LSR expression in the CNS is regio-specific. Each CNS area has its own expression profile for the different LSR chains, thus allowing for specific combination of subunits forming the LSR multimeric complex. Some CNS regions exhibit a stronger LSR expression at the mRNA and/or protein level. Moreover, we demonstrated age-dependent changes in LSR expression, and a strong glia expression of LSR compared to neurons. As previously reported, LSR may play a role in regulation of cholesterol distribution in the CNS [9]. The presence of the BBB prevents access of circulating peripheral lipoproteins to the CNS. Thus, the brain relies on itself to satisfy neuronal needs of cholesterol [5,6]. Adult neurons depend on astrocytes to fulfill those needs [23]. In adult astrocytes, the newly synthesized cholesterol is loaded into ApoE-containing lipoproteins. These HDL-like lipoproteins are needed to export cholesterol from astrocytes, via ABCA1 and ABCG1 transporters, to neurons where they bind via ApoE as ligand to lipoprotein receptors, such as LSR, and are internalized through receptor-mediated endocytosis [24]. Here, we show that LSR is differentially expressed across the brain at both RNA and protein levels. At the RNA level, the HT, HIP, OB, and CB all show high levels of total lsr RNA expression (Table 1). At the protein level, immunoblots show that the HT, OB, and RET express the highest levels of LSR when normalized to β-TUB, which may reflect a specific need of these regions to tightly regulate cholesterol for proper functioning. It is known that LSR is present in the endothelial cells at tight junctions. Despite the fact that all tissues collected contain blood vessels, the high levels of LSR found in specific brain areas cannot be attributed solely to that found in endothelial cells. Indeed, since these cells are found uniformly throughout the CNS, the differences in LSR expression observed reflect those in CNS cells, and therefore neurons or glial cells. Here, we found that LSR is strongly expressed in glial cells relative to neurons in the CNS, thus suggesting an essential role of this lipoprotein in the cholesterol trafficking between these two cell types. Indeed, while this was clearly demonstrated in the CB, which provided sufficient mRNA to compare lsr levels in glial cells and neurons, immunocytostaining of other structures also demonstrated significant LSR protein levels in GFAP-positive cells. In view of this, and based on LSR's role as lipoprotein receptor, we would hypothesize that the LSR present on glial cells might play a role in the glia-neuron cross talk required for feedback control of cholesterol synthesis, regulation of circulating cholesterol and thus maintenance of proper brain function. Glial LSR may have a possible role in internalizing excess ApoE containing lipoprotein particles excreted from neurons, thus possibly activating a signaling pathway to suppress the synthesis and/or loading of cholesterol onto lipoproteins in glial cells. After internalization, free cholesterol derived from lysosomal processing of ApoE-cholesterol particles [25,26] is transported to membranes. Any excess cholesterol in neurons would then be either uploaded onto ApoE-containing lipoproteins to be exported via ABCG4 to the CSF [24], or esterified into cholesterol esters via acyl-coA cholesterol acyltransferases (ACAT1) to be stored as lipid droplets [27,28], or converted to 24-hydroxycholesterol (24S-OHC) by cholesterol 24-hydroxylase (CYP46A1). Since 24S-OHC can readily cross the BBB, it is transported in the blood to the liver for Summary of total lsr mRNA expression in different brain regions of young male versus the whole brain, young versus old male brain, young females versus old males, excretion [29]. An in vitro study shows that treatment of activated primary microglial cells with ApoE peptide (EP) caused downregulation of ApoE synthesis in culture [30], which suggests that regulation of cholesterol synthesis and/or transport require a strict mechanism of glial retro-control. If LSR is deficient, such control mechanisms might be perturbed in glia cells leading to upregulation of cholesterol synthesis and increased lipoprotein secretion, leading ultimately to a possible saturation of neurons with cholesterol. Excess of cholesterol might then accumulate as lipid droplets and in neuronal membranes, which could in turn disrupt protein and lipid trafficking required for synapse assembly in neurons causing neurodegeneration [31]. With age, lsr RNA expression decreases in both the HT, and HIP. Also, LSR protein levels are clearly downregulated in the HT, and show a tendency to be downregulated in the HIP and OB (Table 2). Those observations are consistent with reported changes in CNS cholesterol during aging and neurodegenerative pathologies. Disturbance of cholesterol homeostasis in the brain is coupled to age-related brain dysfunction including synaptic function, neuronal survival and inflammatory status in the aging brain and may contribute to neurodegenerative pathologies such as Alzheimer disease [32,33]. Inducible glia-specific and neuron-specific conditional knockout of lsr are currently under development to determine if decrease of LSR in the CNS might lead to age-related changes in regiospecific-dependent functions (thermoregulation, sex drive, wake/sleep cycle, or hunger for HT [34]; learning and or memory for HIP [35]; olfactory deficits for OB [36]; and motor control problems for CB [37]).  Ratio of LSR (target) over β-TUB (reference). If ratio is equal to 1: equal expression, less than 1: lower, larger than 1: higher.  3). Anti-LSR Sigma antibody was used to detect LSR in different brain regions, including the HT, HIP, OB, Ret, CX, and CB, as indicated. The β-TUB expression of each region is shown below that of LSR. (TIF) S1 Table. RT-qPCR primers used in the study. Forward and reverse primers ised fpr the three reference genes used Hprt, Pgk1, and Tfrc, and target isoforms of lsr, total (T), α, α', and β.