**Neurodegeneration Caused by S1P-Lyase Deficiency Involves Calcium-Dependent Tau Pathology and Abnormal Histone Acetylation**

#### **Shah Alam 1, Antonia Piazzesi 1, Mariam Abd El Fatah 1, Maren Raucamp <sup>2</sup> and Gerhild van Echten-Deckert 1,\***


Received: 29 June 2020; Accepted: 23 September 2020; Published: 28 September 2020

**Abstract:** We have shown that sphingosine 1-phosphate (S1P) generated by sphingosine kinase 2 (SK2) is toxic in neurons lacking S1P-lyase (SGPL1), the enzyme that catalyzes its irreversible cleavage. Interestingly, patients harboring mutations in the gene encoding this enzyme (*SGPL1*) often present with neurological pathologies. Studies in a mouse model with a developmental neural-specific ablation of SGPL1 (SGPL1fl/fl/Nes) confirmed the importance of S1P metabolism for the presynaptic architecture and neuronal autophagy, known to be essential for brain health. We now investigated in SGPL1-deficient murine brains two other factors involved in neurodegenerative processes, namely tau phosphorylation and histone acetylation. In hippocampal and cortical slices SGPL1 deficiency and hence S1P accumulation are accompanied by hyperphosphorylation of tau and an elevated acetylation of histone3 (H3) and histone4 (H4). Calcium chelation with BAPTA-AM rescued both tau hyperphosphorylation and histone acetylation, designating calcium as an essential mediator of these (patho)physiological functions of S1P in the brain. Studies in primary cultured neurons and astrocytes derived from SGPL1fl/fl/Nes mice revealed hyperphosphorylated tau only in SGPL1-deficient neurons and increased histone acetylation only in SGPL1-deficient astrocytes. Both could be reversed to control values with BAPTA-AM, indicating the close interdependence of S1P metabolism, calcium homeostasis, and brain health.

**Keywords:** Sphingosine 1-phosphate (S1P); S1P-lyase (SGPL1); tau; calcium; histone acetylation; hippocampus; cortex; astrocytes; neurons

#### **1. Introduction**

Sphingosine 1-phosphate (S1P), an evolutionarily conserved catabolic intermediate of sphingolipid metabolism, regulates diverse biological processes in the brain including neural development, differentiation, and survival [1,2]. S1P exerts its functions either as a ligand of five specific G-protein coupled receptors (S1PR1-5) or alternatively as an intracellular second messenger [3,4]. Notably, one of the first described intracellular, receptor-independent effects of S1P is its involvement in calcium homeostasis [5,6].

S1P-lyase (SGPL1) irreversibly cleaves S1P in the final step of sphingolipid catabolism, generating ethanolamine phosphate and a long-chain aldehyde [7]. Of interest, in 2017 different research groups reported patients and their relatives harboring autosomal recessive mutations in *SGPL1* and exhibiting a variety of pathologies including congenital steroid-resistant nephrotic syndrome, primary adrenal insufficiency, and last but not least central and peripheral neurological defects [8].

The essential role of S1P in brain development became clear years ago, when elimination of S1P production was shown to severely disturb neurogenesis including neural tube closure, and angiogenesis leading to embryonic death [9]. Yet, reports on the involvement of S1P in the pathology of neurodegenerative diseases including Alzheimer's disease (AD) are rather conflicting [10]. On the one hand, it is assumed that loss of the neuroprotective factor S1P occurs early in AD pathogenesis [11]. Indeed reduced expression of sphingosine kinase 1, one of the two sphingosine kinases known to generate S1P, and a simultaneous augmented expression of SGPL1 were detected in AD brains [12]. On the other hand, S1P was shown to stimulate neuronal beta-site amyloid precursor protein (APP) cleaving enzyme (BACE1) that catalyzes the rate-limiting step of the formation of amyloid beta peptide (Aβ), the major component of senile plaques in AD [13]. In addition, increased S1P levels were shown to induce death of terminally differentiated post-mitotic neurons [14,15]. Moreover, S1P was found to be increased in cerebrospinal fluid during early stages of AD [16].

In an attempt to clarify the function of S1P in the brain, we generated a mouse model in which SGPL1 was inactivated specifically in neural cells (SGPL1fl/fl/Nes). As expected, SGPL1 ablation leads to S1P accumulation in the brain which was found to affect presynaptic architecture and function [17]. In addition, we demonstrated that SGPL1 deficiency blocks neuronal autophagy at its early stages because of reduced phosphatidylethanolamine (PE) production [18]. Consequently, an accumulation of aggregate-prone proteins such as APP and α-synuclein (SNCA) was detected. All these molecular changes in neurons were accompanied by deficits in motor coordination as well as in spatial and associative learning and memory [17,18].

We have also shown that S1P promotes excessive phosphorylation of tau in neurons generated from SGPL1 systemic knockout mice [14]. Note that tau is the major neuronal microtubule assembly activator protein and there is no doubt regarding its essential involvement in the etiopathogenesis of AD and a family of related neurodegenerative disorders known as tauopathies [19]. Tau neurotoxicity has been linked to heterochromatin relaxation and hence to aberrant gene expression in tauopathies [20]. Studies in primary cultured neurons revealed that nuclear tau directly regulates pericentromericheterochtomatin integrity that appears disrupted in AD neurons [21]. Recently, an epigenome-wide study in which acetylation of lysine 9 of histone3 (H3K9ac) was used as a marker for transcriptionally active open chromatin, also led to the conclusion that in aging and AD brains tau pathology drives chromatin rearrangement [22]. Furthermore, in post-mortem AD brains, increased levels of acetylated H3 and H4 were detected and correlated with the load of hyperphosphorylated tau [23].

Remarkably, in a tumorigenic cell line, S1P generated by sphingosine kinase 2 (SK2) was reported to specifically enhance acetylation of H3 and H4 at K9 and K5, respectively, by directly inhibiting histone deacetylases 1 and 2 (HDACs 1, 2) [24]. In the present study, we demonstrate that accumulation of S1P as a result of SGPL1 deficiency increases tau phosphorylation and histone acetylation also in brain slices. Furthermore, we found that both effects can be rescued in the presence of the calcium chelator BAPTA-AM, indicating that this process is calcium-dependent. Notably, these effects were cell-type specific, with increases in tau phosphorylation and histone acetylation found in neurons and astrocytes, respectively. Taken together, our results further elucidate the extensive and complex interrelation of S1P metabolism and brain health.

#### **2. Materials and Methods**

#### *2.1. List of Abbreviations*

AD: Alzheimer's disease; APP: amyloid precursor protein; ac: acetylated; H3: histone3; H4: histone4; H2B, histone2B; K5, lysine residue 5; K9, lysine residue 9; K12, lysine residue 12; HDAC: histone deacetylase; S1P: sphingosine 1-phosphate; SGPL1: S1P-lyase; SK: sphingosine kinase; S1PR: S1P receptor; S396/404, serine residue 396 and serine residue 404; S262/356, serine residue 262 and serine residue 356.

#### *2.2. Antibodies and Chemicals*

Monoclonal antibody against phosphorylated tau PHF1 (S396/404), 12E8(S262/356) and against total tau (K9JA) was a kind gift from Prof. Dr. Eckhard Mandelkow and Prof. Dr. Eva-Maria Mandelkow (DZNE, University of Bonn, Germany). Acetyl-Histone H3 antibody sampler kit comprising acetylation of K9, K14, K18, K27, and K56, anti-H4K5ac, and anti-H2BK12ac antibody were from Cell Signaling Technology Danvers, MA, USA (9927, 8647 and 5410). Anti-HDAC1, -HDAC2, -HDAC3, and -HDAC6 antibodies were from Cell Signaling Technology (antibody Sampler Kit #9928). Anti-SGPL1 antibody was from abcam (Cambridge, UK; ab56183) and Anti-glial fibrillary acidic protein (GFAP) antibody from Cell Signaling Technology (12389).Secondary antibodies were HRP linked anti-rabbit and anti-mouse IgG (Cell Signaling Technology, 7074 and 7076). 5,5- -Dimethyl-BAPTA-AM was from Sigma- Aldrich, Munich, Germany (16609).

#### *2.3. Animals*

The SGPL1flox/flox lines were generated as recently described [17]. SGPL1flox/flox mice, harboring "floxed" exons 10–12 on both Sgpl1 alleles were crossbred with mice expressing the Nes (nestin) - Cre transgene. Thus SGPL1fl/fl/Nesmice (nKO) in which "floxed" exons are excised by Cre recombinase were obtained. For all the experiments, the floxed mice (SGPL1fl/fl) served as controls. Brain tissue was taken from mice housed in standard conditions at the University of Bonn.

#### *2.4. Ethical Statement*

All animal experiments were conducted in accordance with the guidelines of the Animal Care Committee of the University of Bonn. The experimental protocols were approved by Landesamt für Natur, Umwelt und Verbraucherschutz (LANUV) Nordrhein-Westfalen (NRW) (LANUV NRW, Az. 81–02.05.40.19.013).

#### *2.5. Cell Culture*

Primary neuronal culture: Granular cells were cultured from the cerebella of 6-days old mice as described previously [25]. Briefly, neurons were isolated by mild trypsinization (0.05%, w/v; Sigma-Aldrich, Munich, Germany P6567) and dissociated by passing them repeatedly through a constricted Pasteur pipette in a DNase solution (0.1%, w/v; Roche, Basel, Switzerland 04716728001). The cells were then suspended in Dulbecco's modified Eagle's medium (Thermo Fisher Scientific, Waltham, MA, USA 10566032) containing 10% heat-inactivated horse serum (Thermo Fisher Scientific, 16050130) supplemented with 100 units/mL penicillin and 100 mg/mL streptomycin (Gibco™ Thermo Scientific 15140122) and plated onto precoated poly-L-lysine (Sigma-Aldrich, P6282) 6-well plates, 35 mm in diameter (Sarstedt, Nümbrecht, Germany (83.3920.300). Twenty-four h after plating, 1% cytosine ß-D-arabinofuranoside hydrochloride (Sigma-Aldrich, C6645) was added to the medium to arrest the division of non-neuronal cells. After 10 days in culture, cells were used for experiments as indicated.

Primary astrocyte culture: Mixed cortical cell isolation for astrocyte culture was performed using P1 to P4 mouse pups as described previously [26]. Briefly, cerebral cortices were dissected in Ca2+- and Mg2+-free HBSS (GibcoTM, Thermo Scientific, 14185652) and incubated in 0.125% trypsin for 10 min at 37 ◦C. The resulting cell suspension was diluted in complete media Dulbecco's modified Eagle's medium supplemented 10% fetal bovine serum (PAN biotech, Aidenbach, Germany, P40-47100) and 1% penicillin/ streptomycin. The cell suspension was plated on poly-L-Lysine (P-1399) coated T75 cell culture flasks and kept at 37 ◦C in a humidified 5% CO2 incubator. Medium was renewed every 2 days. After about 21 days, flasks were shaken horizontally, and the medium containing detached microglia and oligodendrocyte precursor cells (OPC) was removed. Later, astrocytes were collected and seeded onto 6-well cell culture dishes (35 mm diameter) and used for experiments after 24 h, as indicated.

#### *2.6. BAPTA- AM Treatment*

Hippocampal and cortical slices of 200 μm thickness were prepared in ice-cold high sucrose solution (220 mM sucrose, 26 mM NaHCO3, 10 mM glucose, 6 mM MgSO4.7H2O, 3 mM KCL solid, 1.25 mM NaH2PO4. H2O, 0.43 mM CaCl2) gassed with carbogen. Then, both hippocampal and cortical slices were incubated in artificial cerebrospinal fluid (119 mMNaCl, 26.2 mM NaHCO3, 2.5 mMKCl, 1 mM NaH2PO4, 1.3 mM MgCl2, 10 mM glucose) with and without 150 μM BAPTA-AM for 2 h and kept at −80 ◦C until use.

#### *2.7. Western Immunoblotting*

Tissue and cell samples were homogenized in RIPA buffer (20 mMTris-HCl, pH 7.5, 150 mMNaCl, 1 mM EDTA, 1 mM EGTA, 1% NP-40 (Thermo Fisher Scientific, Waltham, MA, USA, FNN0021), 1% Nadcap (Sigma-Aldrich, Munich, Germany, D6750), 2.5 mM Na4P2O7, 1 mM b-glycerophosphate, 1 mM Na3VO4, 1 mg/mL leupeptin (Thermo Fisher Scientific, 78435). Samples were kept on ice for 1 h followed by centrifugation at 14,000× rpm at 4 ◦C for 45 min. The protein concentration of the supernatants was determined using the BCA assay (Sigma–Aldrich). Samples were stored at −20 ◦C until use. Laemmli Sample Buffer (Bio-rad Laboratories, Munich, Germany, 1610747) was added to lysates and samples were heated for 10 min at 95 ◦C before loading on SDS-PAGE gel. Proteins were separated by SDS-PAGE in running buffer (25 mM Tris, pH 8.3, 192 mM glycine, 0.1% SDS) at 50 V for 15 min, then 1 h at 150 V. Transfer onto nitrocellulose membranes (Porablot NCL; Macherey-Nagel, Thermo Fisher Scientific, 741290) was performed at 4 ◦C and 400 mA for 2 h in blotting buffer (50 mMTris, pH 9.2, 40 mM glycine, 20% methanol). Membranes were blocked with 5% milk powder (Bio-Rad Laboratories, 1706404) in TBS-Tween 20 (20 mM Tris, pH 7.5, 150 mM NaCl, 0.1% Tween 20, Sigma-Aldrich, P9416) for 1 h, washed 3 times (10 min each) and incubated at 4 ◦C overnight with the primary antibody. Then membranes were washed again and incubated for 1 h at room temperature with an HRP-conjugated secondary antibody. Western BLoT Chemiluminescence HRP Substrate (TAKARA Bio, Saint-Germain-en-Laye, France, T7101B) was used for detection with the VersaDoc 5000 imaging system (Bio-Rad, Hercules, CA, USA). β-actin was used as loading control. Quantification was performed using ImageJ and Prism GraphPad program.

#### *2.8. RNA Isolation and Real-Time PCR*

Up to 1 μg of total RNA (isolated with EXTRAzol from Blirt, 7Bioscience, Hartheim/Rhein, Germany, EM30-200) was used for reverse transcription with the ProtoScript® II First Strand cDNA Synthesis kit (New England Biolabs, Frankfurt/Main, Germany, E6560L). The resulting total cDNA was then applied to real-time PCR (CFX96-real time PCR, Bio-Rad Laboratories, Munich, Germany) using β-actin and 18S RNA as housekeeping genes. The primers for real-time PCR were designed using the online tool from NCBI BLAST primer and obtained from Invitrogen, Carlsbad, CA, USA. They are listed as follows: name: forward primer (for), reverse primer (rev): β-actin, 5- -CTTTGCAGCTCCTTCGTTGC (for) and 5- -CCTTCTGACCCATTCCCACC (rev); 18S RNA, 5- -CCCCTCGATGCTCTTAGCTG (for) and 5- -CTTTCGCTCTGGTCCGTCTT (rev); HDAC1, 5- -AGCTGGGCTTTCCAAGTTACC (for) and 5- -TGGTCCACACCCTTCTCGTA (rev); HDAC2, 5- -CGGCCAAGCCTGACTTAGAT (for) and 5- -TTTTCAGCTGTCCTCGGTGG (rev); HDAC3, 5- -TGCCCCAGATTTCACACTCC (for) and 5- -TGGTCCAGATACTGGCGTGA (rev); HDAC6, 5- - GGCGCAGATTAGAGAGCCTT (for) and 5- -GAAGGGGTGACTGGGGATTG (rev); SGPL1, 5- -TTTCCTCATGGTGTGATGGA (for) and 5- -C CCCAGACAAGCATCCAC3(rev).The reactions were performed at 95 ◦C for 30 s, 95 ◦C for 10 s, and 60 ◦C for 1 min. Relative normalized mRNA expression was obtained from real-time qPCR. Statistical significance of the relative normalized mRNA expression was determined by *t*-test in Prism GraphPad program.

#### *2.9. Immunohistochemistry*

Brains were removed and quick-frozen in liquid nitrogen. Cryo-sectioning was used to produce 10 μm sagittal sections, which were placed on Superfrost Plus positively charged microscope slides. Brain sections were fixed for 5 min in ice-cold 4% (*v*/*v*) paraformaldehyde in phosphate-buffered saline (PBS). Sections were then permeabilized with 0.1% (*v*/*v*) Triton X-100 in PBS for 30 min at room temperature (RT). Tissue sections were blocked in 20% (*v*/*v*) normal goat serum in PBS for 30 min and incubated overnight at 4 ◦C with primary antibody (PHF1 and 12E8). The primary antibodies were diluted 1:200 in PBS containing 0.5% lambda-carrageenan (Sigma Aldrich, Munich, Germany, 22049) and 0.02% sodium azide and applied overnight to the sections at 4 ◦C. Following a washing step, brain sections were incubated with Cy3-conjugated anti-rabbit antibody diluted 1:300 in PBS with the same additions as above for 1 h at RT. Finally, antibody-labeled brain sections were embedded in Fluoromount G medium with DAPI (Electron Microscopy Sciences, Hatfield, PA, USA for microscopic analysis (Zeiss Axioskop 2 epi-fluorescence microscope equipped with a digital Zeiss AxioCamHRc camera, Carl Zeiss Jena, Jena, Germany).

#### *2.10. Immunocytochemistry*

Cover slips with astrocytes were rinsed 3 times with PBS at room temperature (RT) and then fixed in methanol (−20 ◦C, 5 min). Between each incubation step cells were always rinsed 3 times with PBS. Then cells were blocked in 20% (*v*/*v*) normal goat serum in PBS for 30 min. and incubated overnight with anti-GFAP antibody diluted 1:200 with PBS at 4 ◦C and thenwith anti-rabbit Alexa Fluor 488 (1:300)-conjugated secondary antibodies (Invitrogen, Carlsbad, CA, USA) for 50 min at RT. Finally, cells were embedded in Fluoromount G medium with DAPI for microscopic analyses (Zeiss Axioskop 2 epi-fluorescence microscope equipped with a digital Zeiss AxioCamHRc camera, Carl Zeiss Jena, Jena, Germany).

#### *2.11. Statistical Analysis*

The GraphPad Prism 5 software was used for statistical analysis.All values are expressed as means ± SEMobtained from at least 3 independent experiments. The significanceof differences between the experimental groups and controls was assessed by either Student's *t*-test with false discovery rate (FDR) correction or One-Way ANOVA, as appropriate. *p*/*q*< 0.05 was considered statistically significant (\* *p*/*q* < 0.05; \*\* *p*/*q* < 0.01; \*\*\* *p*/*q* < 0.001; compared with the respective control group).

#### **3. Results**

#### *3.1. Elevated Phosphorylation of Tau in SGPL1-Deficient Brains Is Cell Type Specific*

We have previously shown that tau phosphorylation is elevated in primary cultured neurons derived from brains of systemic SGPL1-knockout (KO) mice [14]. Here, we generated a neural-specific Sgpl1 knockout (SGPL1fl/fl/Nes; nKO) mouse and performed our analysis in brain slices, primary neuronal and astrocyte cultures (Supplementary Figure S1). We have found that tau phosphorylation at disease-relevant sites is also significantly increased in hippocampal and cortical slices from SGPL1fl/fl/Nesmice consistent with our previously reported findings in systemic KO mice [14] (Figure 1A,B). Furthermore, this increase in tau phosphorylation at disease-relevant sites was not accompanied by changes in total tau levels (Figure 1B). A more refined analysis in primary cultured neurons and astrocytes from SGPL1fl/fl/Nesmice further revealed that this effect is primarily attributed to neurons, as tau phosphorylation remained unaffected in astrocytes (Figure 1C). Accordingly, phosphorylated tau was increased by about 40% in SGPL1-deficient neurons whereas no changes were detectable in astrocytes lacking SGPL1 when compared with the respective control cells (Figure 1C). This indicates that the increase in tau phosphorylation in hippocampus and cortex is due to hyperphosphorylation of tau in neurons.

**Figure 1.** SGPL1 deficiency results in tau hyperphosphorylation in the brain. (**A**) Representative images of hippocampal and cortical slices stained for phospho-tauS396/404, phospho-tauS262/356, and DAPI from control (**C**) and SGPL1fl/fl/Nes (nKO) mice. Scale bar: 200 μm. (**B**,**C**) Protein quantification of phospho-tauS396/404, phospho-tauS262/356, and total tau (pan-tau) in the hippocampus (Hippo, black), cortex (dark grey), neurons (Neu, light grey) and astrocytes (Astro, white) in control (**C**) and SGPL1fl/fl/Nes (nKO) mice. Bars mean +/<sup>−</sup> S.E.M, Student's *t*-test with false discovery rate (FDR) correction, n = 3–5, \* *q* < 0.05, \*\* *q* < 0.01, ns = not significant.

#### *3.2. Histone Acetylation Levels Vary in Di*ff*erent Cell Types Derived from Brains Lacking SGPL1*

Based on reports that tau stimulates chromatin relaxation [20] and that S1P accumulation elevates histone acetylation in tumorigenic cells [24], we analyzed histone acetylation in hippocampal and cortical slices from SGPL1fl/fl/Nesmice. We found that, upon S1P accumulation in the brain, acetylation of H3 is also significantly increased in both hippocampal and cortical slices by about 36% and 32%, respectively (Figure 2A). An investigation of specific acetylation sites revealed that K9 of H3 was significantly increased by about 45% in the hippocampus, and to a lesser extent (about 35%) in the cortex, as compared to the respective controls (Figure 2A). Interestingly, acetylation of all other lysine residues examined, including K14 and K18 was not affected by SGPL1 deficiency (Supplementary Figure S2). Next, we investigated acetylation of H3 in primary cultured cells derived from SGPL1fl/fl/Nesmice. We found that acetylation of H3 was significantly increased in astrocytes generated from SGPL1fl/fl/Nesmice (Figure 2B), whereas no changes of H3 acetylation were observed in primary cultured neurons of these mice (Figure 2C). Notably, the acetylation of H3 (H3ac) largely resembled that of lysine 9 of H3 (H3K9ac), whereas the total H3 level was only slightly increased (Figure 2B). These results strongly suggest that the increases in histone acetylation observed in the brain

upon SGPL1 deficiency is due to epigenetic changes in astrocytes, rather than neurons. Interestingly, we also found that acetylation of H4 at K5 was significantly increased in astrocytes of SGPL1fl/fl/Nesmice, similar to what was shown previously in breast cancer cells (Figure 2B) [24]. Furthermore, we observed an increase in acetylation of H2B at K12, also consistent with a report of S1P accumulation in breast cancer cells (Figure 2B). Given the differences in histone acetylation observed, and given that S1P can act as an inhibitor of histone deacetylases [24], we next investigated whether the expression of histone deacetylases was affected in SGPL1fl/fl/Nesmice. However, we found that both mRNA and protein levels of HDAC1, 2, 3 and 6 remained unaffected in SGPL1fl/fl/Nesmice, compared to controls (Figure 2D–G). These results show that S1P accumulation in the brain has a cell type-specific effect on protein posttranslational modifications, without affecting the expression levels of the deacetylases responsible.

**Figure 2.** SGPL1 deficiency affects histone acetylation in the brain without affecting histone deacetylases (HDAC) expression. (**A**) Protein quantification of H3 pan-acetylation (H3ac), H3K9 acetylation (H3K9ac), and total H3 in the hippocampus (Hippo, black) and cortex (dark grey) from control (C) and SGPL1fl/fl/Nes (nKO) mice. (**B**) Protein quantification of H3 pan-acetylation (H3ac), H3K9 acetylation (H3K9ac), total H3, H4K5 acetylation (H4K5ac), and H2BK12 acetylation (H2BK12ac) in astrocytes from control (C) and SGPL1fl/fl/Nes (nKO) mice. (**C**) Protein quantification of H3 pan-acetylation (H3ac) and total H3 in primary neuronal culture from control (C) and SGPL1fl/fl/Nes (nKO) mice. (**D**–**E**) qRT-PCR of HDAC1, 2, 3, and 6 in the hippocampus (**D**) and cortex (**E**) of control (C) and SGPL1fl/fl/Nes (nKO) mice. (**F**–**G**) Protein quantification of HDAC1, 2, 3, and 6 in the hippocampus (**F**) and cortex (**G**) of control (C) and SGPL1fl/fl/Nes (nKO) mice. For all: Bars mean +/<sup>−</sup> S.E.M, Student's *<sup>t</sup>*-test with false discovery rate (FDR) correction, n = 3−7, \* *q* < 0.05, \*\* *q* < 0.01, \*\*\* *q* < 0.001, ns = not significant.

#### *Cells* **2020**, *9*, 2189

Together, these results indicate that different cell types can be responsible for interrelated effects detected when studying certain brain regions.

#### *3.3. Calcium Chelation Reverses Both Tau Phosphorylation and Histone Acetylation in the Brain of SGPL1fl*/*fl*/*Nesmice*

We have previously suggested that increased calcium concentrations might account for the neurotoxic effect of S1P in SGPL1-deficient neurons [14]. Calcium measurements in hippocampal slices of SGPL1fl/fl/Nesmice revealed a persistent elevation of basal calcium concentration in pyramidal neurons of the CA1 region amounting about 223 nM, a value that exceeds control concentrations by about 2.5-fold [27]. To find out whether elevated basal calcium concentration is linked to tau phosphorylation, we subjected hippocampal as well as cortical slices to BAPTA-AM treatment. Notably, calcium chelation by BAPTA reversed tau phosphorylation in SGPL1-deficient hippocampal and cortical slices to control values, specifically at the pathological phosphoepitope at serine residue S396/404 (Figure 3A), while phosphorylation of serine residues S262/356 was not affected by BAPTA (Supplementary Figure S3). Furthermore, we found that histone acetylation in the same samples also returned to control levels following BAPTA-AM treatment (Figure 3B).

**Figure 3.** BAPTA-AM treatment reversed tauS396/404hyperphosphorylation and H3K9 acetylation in the brain of SGPL1fl/fl/Nesmice.(**A**) Protein quantification of phospho-tauS396/<sup>404</sup> and total tau (pan-tau) in the hippocampus (Hippo, black) and cortex (dark grey) of control (**C**) and SGPL1fl/fl/Nes(nKO) mice with (+) and without (-) BAPTA-AM treatment. (**B**) Protein quantification of H3K9 acetylation (H3K9ac) and total H3 in the hippocampus (Hippo, black) and cortex (dark grey) of control (**C**) and SGPL1fl/fl/Nes (nKO) mice with (+) and without (-) BAPTA-AM treatment. (**C**) Protein quantification of phospho-tauS396/<sup>404</sup> and H3K9 acetylation (H3K9ac) in neurons (Neu, light grey) and astrocytes (Astro, white) in control (**C**) and SGPL1fl/fl/Nes (nKO) mice with (+) and without (-) BAPTA-AM treatment. For all: Bars mean +/− S.E.M, One-Way ANOVA with Tukey's post-hoc correction, n = 3−7, \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001, \*\*\*\* *p* < 0.0001, ns = not significant.

#### *3.4. Neuronal Tau Pathology and Augmented Histone Acetylation in Astrocytes of SGPL1fl*/*fl*/*Nes Mice Are Calcium Dependent*

As shown above, tau hyperphosphorylation appeared to be neuron specific, whereas the unusual increase of histone acetylation was restricted to astrocytes. The fact that, in hippocampal and cortical slices, BAPTA-AM reversed both parameters to control values prompted us to investigate the effect of calcium chelation in primary cultured neurons and astrocytes, respectively. Treatment of primary cultured neurons from SGPL1fl/fl/Nesmice with BAPTA-AM recapitulated the effect on tau phosphorylation described above in hippocampal and in cortical slices. Thus, the expression of phosphorylated tau at serineS396/404 returned to control values following calcium chelation (Figure 3C). Likewise, the treatment of primary cultured astrocytes derived from SGPL1fl/fl/Nesmice reversed the acetylation of H3 at K9 (H3K9ac) to control amounts (Figure 3C).

#### **4. Discussion**

SGPL1 deficiency in brain has been shown to affect neuronal health and to cause neuroinflammation accompanied by impairment of cognitive and motor skills in mice [17,18,28]. The central aim of the present study was to further unravel the molecular bases of the neurological phenotype of SGPL1fl/fl/Nesmice. We therefore investigated tau expression and phosphorylation in SGPL1-deficient animals. While the amount of total tau remained unaffected in the brains of SGPL1fl/fl/Nesmice, phosphorylation of tau at disease-relevant sites was significantly increased in brain slices and neuronal cultures derived from these mice. Based on reports from the early nineties that altered calcium homeostasis may be a key event leading to altered tau disposition and neuronal degeneration [29], we assumed that increase of phosphorylated tau in SGPL1-deficient brains could be linked to the increase of neuronal calcium in SGPL1fl/fl/Nesmice. Our results confirmed that at least one of the investigated disease-relevant phosphorylation sites can be reversed by BAPTA-AM treatment, indicating that the effect of SPGL1 deficiency on tau phosphorylation is also calcium-dependent. Although several changes of tau expression, mutation and posttranslational modifications were described in diverse pathologies of the central nervous system, hyperphosphorylation appears to be of particular importance for its pathologic function [19,30]. Our results show that SPGL1 deficiency in the brain can lead to a neuronal-specific, calcium-dependent hyperphosphorylation of tau at a site relevant to tauopathies.

We have previously shown that SPGL1 deficiency leads to an accumulation of aggregate-prone proteins in the brain, along with deficits in motor coordination, learning and memory [17,18]. Furthermore, we have shown that SGPL1 deficiency on the one hand triggers the ubiquitin-proteasome system (UPS) in brains of SGPL1fl/fl/Nesmice [17], while on the other hand, it blocks neuronal autophagy flux at an early stage [18]. The implication of both systems in neuronal health and death is well established [31–34]. Notably, both systems are also responsible for tau clearance [35]. While numerous reports suggested that tau is a proteasomal substrate, other studies found that the autophagy/lysosomal pathway is the primary degradation machinery for tau [35]. Our results regarding unchanged amounts of total tau could be explained by these opposing effects of SGPL1 deficiency on the two degradation systems in neurons [17,18]. Further studies into the molecular mechanisms by which SGPL1 deficiency and/or S1P accumulation in the brain affect tau hyperphosphorylation, clearance and pathology could therefore be of great interest to the field of tauopathies.

Although tau is predominantly produced by neurons in the brain, tau pathology is not restricted to neurons [30,36]. However, astrocytes derived from SGPL1-deficient brains did not show any changes in tau expression or phosphorylation. One explanation might be the fact that many studies regarding tau pathology were performed in transgenic animals harboring the wild-type or mutant human tau transgene [30,37]. On the other hand, apart from hyperphosphorylation, microglial activation and thus neuroinflammation appear to be essential for astrocytic tau pathology [30]. We have recently shown that in SGPL1fl/fl/Nesmice, microglial activation and hence release of pro-inflammatory cytokines including interleukin (IL) 6 (IL-6) and tumor necrosis factor alpha (TNF-α), is caused by S1P

released from SGPL1-deficient astrocytes [28]. Whether and how these factors are involved in the tau hyperphosphorylation of SGPL1-deficient astrocytes has to be clarified in future studies.

It is well-established that mobilization of intracellular calcium stores is a universal signaling mechanism that cells employ for responding to a wide range of external stimuli [38]. Notably in SGPL1-deficient neurons, this stimulus is S1P [5]. We also found it rather interesting that both S1P [24] and tau, downstream of its pathologic accumulation [22] have been shown to affect histone acetylation, though in different cellular contexts. Specifically, S1P produced by overexpression of SK2 in the nucleus of breast cancer cells was shown to inhibit HDAC 1 and 2, thus increasing acetylation of H3 at K9, of H4 at K5, and rather weakly that of H2B at K12 [24]. Moreover, a compromised acetylation homeostasis has been suggested to be intimately coupled with neurodegeneration [39]. We therefore decided to investigate whether SGPL1 deficiency in the brain could exert an analogous effect. Intriguingly, we also found calcium-dependent increases in H3K9 and H2BK12 acetylation in brain slices and primary cultured astrocytes, but not in neurons derived from SGPL1fl/fl/Nesmice, without affecting the overall expression levels of HDAC1, 2, 3, or 6. These results indicate that SGPL1 deficiency also plays a role in histone posttranslational modifications in astrocytes, though further studies are needed to elucidate the role of this epigenetic disruption on overall brain health.

In SGPL1-deficient mouse embryonic fibroblasts (MEFs) an increase of acetylated H3K9 was reported, but no changes in the level of H4 and H2B acetylation could be detected [40]. Despite this similarity, the underlying molecular mechanisms appear to diverge in the three cell types. In the SK2-overexpressing cancer cell lines, nuclear S1P was shown to directly interact and thus inhibit HDACs1 and 2 [24] while in SGPL1-deficient MEFs a reduced expression of HDACs1, 2, and 3 was reported [40]. Intriguingly, the decreased expression of HDACs was correlated with an elevated basal calcium level in SGPL1-deficient MEFs [40]. By contrast, in hippocampal and cortical slices derived from SGPL1-deficient murine brains, no changes in the expression of HDACs could be observed. However, calcium chelation by BAPTA-AM restored histone acetylation, suggesting that calcium mediates the effect of S1P on histone acetylation independent of HDAC expression. Given these differences between models, we cannot make assumptions as to the specific molecular mechanism responsible for altered histone acetylation in our system.

Interestingly, epigenetic dysregulation currently attracts much attention as a pivotal player in aging and age-related neurodegenerative disorders, such as AD, Parkinson's disease and Huntington's disease, where it may mediate interactions between genetic and environmental risk factors, or directly interact with disease-specific pathological factors [23,41]. Furthermore, a recent epigenome-wide association study using H3K9ac as a marker for transcriptionally active open chromatin revealed that tau pathology is associated with broad changes in the brain's epigenome [22]. This study was conducted in aged human dorsolateral prefrontal cortices. Also, in post-mortem human brains from AD patients, tau pathology was correlated with augmented H3 and H4 acetylation [23]. Given our findings that SPGL1 deficiency results in both altered tau homeostasis and histone acetylation, further studies into how SPGL1 deficiency and/or S1P accumulation fit into these complex systems are needed. While a closer look in primary cultured neurons and astrocytes uncovered a cell type specificity to the effect of SGPL1 deficiency, this does not exclude their association in the nervous tissue. At present, the association of tau pathology and histone acetylation appears rather conflicting and far from clear. It has been reported that specific inhibition of HDAC3 and consequently an increased acetylation of H3 and H4 was shown to reduce tau phosphorylation at disease- associated sites, including serine 396, and was proposed as a novel neuroprotective mechanism [42]. Similarly, the specific inhibition of HDAC6 caused a significant reduction of tau phosphorylation [43]. HDAC6 inhibition also increased acetylation of Hsp90 which caused ubiquitination of phosphorylated tau thus alleviating abnormal tau accumulation [43]. Despite many open questions, we feel that our results show that SGPL1 deficiency impacts brain health and might help explain the potential molecular mechanisms underlying the diverse neuropathological phenotypes in humans harboring mutations in the *SGPL1* gene [8].

#### **5. Conclusions**

Our results indicate that SGPL1 depletion augments tau phosphorylation in neurons and simultaneously enhances histone acetylation in astrocytes suggesting a negative impact on brain health. On the other hand, immunohistochemical analysis in frontal and entorhinal cortices from 56 human AD brains revealed an augmented SGPL1 expression correlating with amyloid deposits [12]. The same study also reported a decreased expression of sphingosine kinase 1 as well as of S1PR1 suggesting a global deregulation of S1P signaling in human AD brains [12]. Our results rely primarily on SGPL1 deficiency. Hence, although sites relevant to tauopathies including AD are hyperphosphorylated, we feel that, due to the multitude of phosphorylation sites in tau and the complexity of this phenomenon [44], more studies are needed to finally understand the function of SGPL1 in tauopathies. At present, our findings are first and foremost interesting for a better understanding of the phenotype of humans with insufficient SGPL1 activity [8].

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4409/9/10/2189/s1, Figure S1: Neural-specific depletion of SGPL1; Figure S2: SGPL1 deficiency has no effect on acetylaztion of Histone3 at lysine 14 and 18; Figure S3: BAPTA-AM treatment has no effect on tauS262/<sup>356</sup> hyperphosphorylation in the brain of SGPL1fl/fl/Nes mice.

**Author Contributions:** Conceptualization: S.A. and G.v.E.-D.; methodology: S.A., M.A.E.F., and M.R.; Formal analysis and investigation: S.A. and A.P.; supervision and writing, G.v.E.-D. and A.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by a grant EC 118/10-1 to G.v.E.D. from the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG).

**Acknowledgments:** We thank Dieter Swandulla and the LIMES Institute for financial support and Margit Zweyer for excellent technical assistance.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Exogenous Flupirtine as Potential Treatment for CLN3 Disease**

#### **Katia Maalouf 1,**†**, Joelle Makoukji 1,**†**, Sara Saab 1, Nadine J. Makhoul 1, Angelica V. Carmona 2, Nihar Kinarivala 3, Noël Ghanem 4, Paul C. Trippier 2,5,6 and Rose-Mary Boustany 1,7,8,\***


Received: 16 June 2020; Accepted: 6 July 2020; Published: 11 August 2020

**Abstract:** CLN3 disease is a fatal neurodegenerative disorder affecting children. Hallmarks include brain atrophy, accelerated neuronal apoptosis, and ceramide elevation. Treatment regimens are supportive, highlighting the importance of novel, disease-modifying drugs. Flupirtine and its new allyl carbamate derivative (compound 6) confer neuroprotective effects in CLN3-deficient cells. This study lays the groundwork for investigating beneficial effects in *Cln3*Δ*ex7*/*<sup>8</sup>* mice. WT/*Cln3*Δ*ex7*/*<sup>8</sup>* mice received flupirtine/compound 6/vehicle for 14 weeks. Short-term effect of flupirtine or compound 6 was tested using a battery of behavioral testing. For flupirtine, gene expression profiles, astrogliosis, and neuronal cell counts were determined. Flupirtine improved neurobehavioral parameters in open field, pole climbing, and Morris water maze tests in *Cln3*Δ*ex7*/*<sup>8</sup>* mice. Several anti-apoptotic markers and ceramide synthesis/degradation enzymes expression was dysregulated in *Cln3*Δ*ex7*/*<sup>8</sup>* mice. Flupirtine reduced astrogliosis in hippocampus and motor cortex of male and female *Cln3*Δ*ex7*/*<sup>8</sup>* mice. Flupirtine increased neuronal cell counts in male mice. The newly synthesized compound 6 showed promising results in open field and pole climbing. In conclusion, flupirtine improved behavioral, neuropathological and biochemical parameters in *Cln3*Δ*ex7*/*<sup>8</sup>* mice, paving the way for potential therapies for CLN3 disease.

**Keywords:** sphingolipids; neurodegeneration; ceramide; CLN3 disease; *Cln3*Δ*ex7*/*<sup>8</sup>* mice; flupirtine; allyl carbamate derivative; apoptosis

#### **1. Introduction**

The neuronal ceroid lipofuscinoses (NCLs) constitute a family of fatal pediatric neurodegenerative disorders that primarily affect the central nervous system (CNS) [1]. NCLs are atypical lysosomal storage disorders that manifest accumulation of lipopigments in the lysosomes of neurons and other cell types [2]. CLN3 disease arises due to mutations in the CLN3 gene and is the most common variant of the NCL group [3]. This neurological disease manifests at four to eight years of age with progressive visual deterioration, seizures, blindness, motor and cognitive decline, mental retardation, epilepsy and early death during the second or third decade of life [4]. Massive cortical neuronal cell loss due to neuronal apoptosis within the cortex [5], and neuronal loss in hippocampus and microglial activation in this region are documented [6]. Eighty five per cent of patients with CLN3 disease harbor a 1.02 kb deletion eliminating exons 7/8 and creating a truncated CLN3 protein [7,8].

CLN3 protein influences major cellular functions, including apoptosis and cell growth 9. Apoptosis is the mechanism of neuronal and photoreceptor cell loss in human brain from patients with CLN3 disease 3 [9]. Ceramide, a pro-apoptotic lipid second messenger, mediates anti-proliferative events of apoptosis, growth inhibition, cell differentiation, and senescence [10].

Ceramide levels are increased in CLN3-deficient cells and in brain of CLN3 patients [11]. Studies confirm that CLN3 protein expression modulates brain ceramide levels. Levels of lipids ceramide, SM, GalCer, GluCer, and globoside are elevated in human CLN3-deficient fibroblasts. Ceramide levels normalized following restoration of CLN3 function, but not following caspase inhibition by zVAD, a pan-inhibitor of caspases [8,11,12]. Overexpressing CLN3 protein results in a drop in ceramide levels [13]. Increased ceramide levels and neuronal cell loss are evident in brain sections from post-mortem CLN3 disease patients and in brains and sera of *Cln3*Δ*ex7*/*<sup>8</sup>* mice [14]. Treatment regimens for CLN3 disease are largely supportive, not curative, and do not target the underlying causes of the disease.

Flupirtine is a centrally acting non-opioid drug previously widely used in clinics as an analgesic [15,16]. Flupirtine maleate is the salt of this drug, henceforth, referred to as just flupirtine. It is neuroprotective, has muscle relaxant and anticonvulsant properties [17] and suppresses neuronal hyper-excitability [18]. Flupirtine protects photoreceptor and neuronal cells from apoptosis induced by various insults [13]. There is evidence suggesting that flupirtine reduces brain injury, induces remodeling of brain tissue, and diminishes cognitive impairment in in vivo animal models of ischemic stroke [15]. Flupirtine protects lymphoblasts, differentiated human post-mitotic hNT neurons and PC12 neuronal precursor cells from apoptosis induced by etoposide [13,19]. A newly synthesized allyl carbamate derivative of flupirtine (compound 6) has shown potential neuroprotective effects in vitro, as one of nine flupirtine aromatic carbamate derivative [19–21]. Compound 6 imparted a 150% increase in Bcl-2/Bax ratio in vitro which is protective [19]. Flupirtine and its allyl carbamate derivative (compound 6) increased cell viability in human CLN3 patient-derived lymphoblasts and in neuronal precursor PC12 cells transfected with siRNA directed against CLN3, exhibiting significant anti-apoptotic and neuroprotective effects [19].

This study tests, in vivo, oral supplementation of flupirtine for a period of 14 weeks in *Cln3*Δ*ex7*/*<sup>8</sup>* knock-in mice. Outcomes of efficacy include improved behavioral measures, altered gene expression profiles, decreased glial immunoreactivity, and increased neuronal cell numbers in specific brain regions. Supplementation of compound 6 assessed effectiveness in several parameters, as a first step in also developing it as potential treatment for CLN3 disease.

#### **2. Materials and Methods**

#### *2.1. Animals*

This mouse work was conducted in accordance with an approved American University of Beirut (AUB) Institutional Animal Care and Use Committee (IACUC) protocol (IACUC approval #18-08-496). Animal testing was carried out at the AUB Animal Care Facility where animals were housed. C57BL/6J (JAX stock number: 000664) and homozygous *Cln3*Δ*ex7*/*<sup>8</sup>* (JAX stock number: 029471) mice were obtained from the Jackson laboratory, kept in a 12-h light/dark cycle (lights onset at 6:00 am) and supplied with access to food and water ad libitum. Room temperature was maintained between 18–26 ◦C, and relative humidity between 30–70%. Mice were housed in groups of 3–4/cage. All efforts to minimize number of animals and animal suffering were applied. Mice were monitored for weekly weights, basic behavior and general health throughout the study, and were bred in-house.

#### *2.2. Flupirtine and Compound 6 Treatment*

Flupirtine and compound 6 were dissolved in vehicle (0.5% di-methyl sulphoxide (DMSO) in 10% phosphate-buffered saline (PBS)), at a dose of 30 mg/kg body weight for a period of 14 weeks starting at 4 weeks of age. Vehicle treatment consisted of 0.5% DMSO in 10% PBS. Mice were treated 'per os' by drinking water with a consistent supply in a volume of ≈8 mL/day/mouse. Both drugs were synthesized by Dr. Paul Trippier at the department of pharmaceutical sciences in the School of Pharmacy at Texas Tech University Health Sciences Center (Figure 1). Mice were divided into five groups, consisting of 16 mice each (eight males and females) and consisted of C57BL/6J vehicle-treated WT mice, C57BL/6J compound 6-treated WT mice, vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* mice, flupirtine-treated *Cln3*Δ*ex7*/*<sup>8</sup>* mice and compound 6-treated *Cln3*Δ*ex7*/*<sup>8</sup>* mice. Genotype was confirmed by PCR of DNA mouse blood.

**Figure 1.** Chemical structure of flupirtine and its allyl carbamate derivative, compound 6.

#### *2.3. Behavioral Studies*

Mice were held in their cages in the behavioral room for testing, with lights dimmed for 60 min prior to onset of tests, for habituation. All behavioral assays were performed during the light cycle. Each test was performed at *n,* the same time of the day and within the same hour, when possible, to minimize variability between cohorts. Comparison between groups was carried out on males and females separately.

Open field: A mouse was placed on the periphery of a transparent Plexiglas cubic box (dimensions: 50 cm width, 50 cm length and 30 cm in height, with the center 16.4 cm2) so that locomotion would be apparent to the operator and exploratory behavior videotaped. Specific parameters were recorded by a top-mounted video-recorder using EthoVision software (Noldus Information Technology, Wageningen, The Netherlands) for each animal including total distance traveled, average speed, mobility duration, rearing and walling frequencies. Each mouse was allowed to move in the arena freely for 5 min. The box was wiped with 50% ethanol between each mouse to avoid olfactory cuing.

Pole climbing: Mice were habituated to the task in five trials/day for 1 day. On the next test day, five trial measures/mouse were performed by placing the mouse head upward on top of a roughsurfaced pole (1 cm in diameter and 60 cm in height) wrapped with tape to prevent slipping. The time for the mouse to completely turn its head down (tturn), time to reach the middle of the pole (t1/2), time it takes to descend and settle on the floor (ttotal), and time the mouse spent freezing during descent (tstop) were recorded. Each mouse had a maximum time of two minutes to climb down to avoid exhaustion.

Morris water maze (MWZ): The spatial learning abilities of mice were assessed in a MWM task. The apparatus consisted of a circular pool 100 cm diameter, filled up to 50 cm with water made opaque by addition of a small amount of non-toxic white paint and maintained at 21–22 ◦C. The pool is virtually subdivided into four quadrants using the software. A circular escape platform was placed in a fixed south-west (SW) location hidden 0.5 cm below the surface of the water, and five fixed-position geometric visual cues were kept in the brightly lit room throughout the period of testing. A digital camera was positioned above the center of the tank and linked to a tracking system (ANY-Maze

behavioral tracking software, version 6.3, USA) in order to record escape latencies, path distance (m), percentage of thigmotaxis path and swim speed for each trial, together with time spent swimming in each of the four quadrants of the maze.

Mice (eight males and females from each treatment group) were given four consecutive days of acquisition training sessions that consisted of four trials per day. Throughout the course of this acquisition period the position of the hidden platform remained fixed (SW) and the entry point was varied from trial to trial, but the sequence remained fixed for all mice within that day. We used four different entry points (north, south, east, and west). The sequence of starting points was modified from one day to the other. Mice were given 60 s to find the platform, and if the mouse failed to locate the platform within this period, it was guided onto it. All mice were allowed to rest on the platform for a 30-s interval after each trial. At the end of the training block, mice were put in a drying cage and allowed to dry prior to being returned to their experimental cages. The last day of the test, the 'probe trial' was performed with the platform removed from the maze and the rodent released from the North entry point of the pool to find the previous location of the hidden platform.

#### *2.4. Corticosterone Immunoassay Kit*

Blood was collected from the inferior vena cava, left to clot, centrifuged and spun for 15 min at 10,000 rpm at 4 ◦C. Serum corticosterone levels were determined using DetectX® Corticosterone Enzyme Immunoassay Kit (Cat. No K014-H5, Arbor Assays, Ann Arbor, MI, USA), according to manufacturer's instructions. Five μL of standards or mouse serum samples were assayed in duplicate and run altogether on one single plate simultaneously. The absorbance was measured using TriStar2S microplate reader (Berthold Technologies, Bad Wildbad, Germany) at 450 nm.

#### *2.5. RNA Extraction from Brain Tissue*

Mice were deeply anesthetized with a mixture of xylazine and ketamine (10 mg/kg and 100 mg/kg, respectively) and brains were rapidly dissected and "snap" frozen in liquid nitrogen to preserve RNA integrity, and stored at −80 ◦C. A total of 30 mg ground fresh brain tissue (from four males and females in each treatment group) were homogenized using a motorized rotor-stator homogenizer and RNA extracted using RNeasyPlus Mini Kits (Cat No. 74134, Qiagen, Germantown, MD, USA) according to manufacturer's instructions. For assessing RNA quality, A260/A280 and A260/A230 ratios for RNA are analyzed with the ExperionTM Automated Electrophoresis System (BioRad, Hercules, CA, USA). RNA concentrations are determined by absorption at 260 nm wavelength with a ND-1000 spectrometer (Nanodrop Technologies LLC, Wilmington, DE, USA).

#### *2.6. Quantitative Real-Time PCR (qRT-PCR)*

Total RNA extracted from fresh brain tissues was reverse transcribed using RevertAid Reverse Transcriptase (Thermo Fisher Scientific, Waltham, MA, USA) with 2 μg of input RNA and random primers (Thermo Fisher Scientific, USA). qRT-PCR reactions were performed in 384-well plates using specific primers (Tm = 60 ◦C) (TIB MOLBIOL, Berlin, Germany) (Table 1) and iTaq SYBR Green Supermix (BioRad, Hercules, CA, USA) as a fluorescent detection dye, in CFX384TM Real-Time PCR (BioRad, USA), in a final volume of 10 μL. To characterize generated amplicons and to control contamination by unspecific by-products, melt curve analysis was applied. Results were normalized to β-actin or Gapdh mRNA level. All reactions were performed in duplicate, and results were calculated using the ΔΔCt method.


**Table 1.** Mouse cDNA primer sequences.

#### *2.7. Immunohistochemistry*

For morphological and immunohistochemical sectioning, 4 mice/treatment group were deeply anesthetized with a mixture of xylazine and ketamine (10 mg/kg and 100 mg/kg, respectively) and fixed by cardiac puncture with 30 mL of 4% paraformaldehyde (PFA) in PBS. Brains were carefully isolated and fixed with 4% PFA solution (pH of 7.4) for 2 h at 4 ◦C, then cryoprotected in a solution of 20% sucrose overnight. The following day, brains were processed and frozen using embedding medium, Optimal Cutting Temperature (OCT) compound, for later tangential sectioning on glass slides. Brains were cut in coronal sections (20 μm) using a cryostat and sections stored at −20 ◦C for further analysis. Brain coronal cryosections were treated with PBS for 5 min twice, then incubated with PBST (0.1% Triton X-100 in PBS) for 10 min twice. Sections were incubated with blocking solution (PBST 0.1%-FBS 10%) for 1 h at room temperature (RT). They were then incubated with each of the primary antibodies: anti-GFAP (1:500, Abcam, Cambridge, UK, catalogue #ab7260) and anti-NeuN antibody (1/300, Abcam, Cambridge, UK, catalogue #ab104225) in antibody solution (PBST 0.1%-FBS 1%) overnight at 4 ◦C. After washing in PBST, slides were treated with Sudan black for 40 min, then washed with PBS three times. Brain cryosections were then incubated with biotinylated secondary antibody diluted in antibody solution at RT for 1 h. Samples were counterstained with 1:10,000 Hoechst (Sigma, St. Louis, MO, USA) and then mounted in Fluoromount (Sigma, St. Louis, MO, USA).

Signal quantification was assessed using Leica microscope software imaging. For microscopic imaging, three sections/mouse were selected. Primary motor cortex was viewed at 40× magnification with three to four photos/section for motor cortex layers (I–VI). Hippocampus images for GFAP were obtained at 40× magnification. Intensity quantification was depicted by the ratio of integrated density over total area of image using Image J software, version 1.52a. NeuN positive cells were quantified manually using Image J software. Number of NeuN positive cells divided by total area of image.

#### *2.8. Statistical Analysis*

Basic statistical analysis was conducted using GraphPad Prism 6 statistical package (GraphPad Software version 6.04, San Diego, CA, USA). Data was expressed as mean ± standard error of the mean (SEM). For two group comparisons, Student's *t*-test was used with quantitative continuous variables. Comparisons between different groups were statistically tested with either one-way analysis of variants (ANOVA) or two-factorial ANOVA followed by Tukey's post-hoc test for multiple group comparisons. All tests are two sided and a *p*-value < 0.05 is considered as statistically significant.

#### **3. Results**

#### *3.1. Impact of Flupirtine on Motor Behavior of Homozygous Cln3*Δ*ex7*/*<sup>8</sup> Mice*

The open field behavioral test assesses novel environment exploration, general locomotor activity, and provides an initial screen for anxiety-related behavior in rodents. Open field behavioral testing showed that vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* mice exhibit increased spontaneous locomotor activity compared to vehicle-treated WT controls in both genders at 15 weeks of age. Vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* mice were significantly more mobile than their vehicle-treated WT littermates (Figure 2A,B). These results indicate that WT mice display increased anxiety-like behavior when put in the novel test environment compared with *Cln3*Δ*ex7*/*<sup>8</sup>* mice. Normally, rodents display distinct aversion to large, brightly lit, open and unknown environments. They have been phylogenetically conditioned to see these types of environments as dangerous [22]. To confirm whether *Cln3*Δ*ex7*/*<sup>8</sup>* mice have less anxiety than WT mice, we analyzed serum levels of the predominant murine glucocorticoid, corticosterone, in these animals. Male *Cln3*Δ*ex7*/*<sup>8</sup>* mice experienced significantly lower corticosterone levels than those measured in WT animals (Figure 2E). Similarly, female *Cln3*Δ*ex7*/*<sup>8</sup>* mice experienced lower corticosterone levels than WT animals, that was very close to significance (*p*=0.09) (Figure 2F). These results confirm thatWT mice have an enhanced physiologic response to stress, characterized by increased hypothalamic-pituitary-adrenal axis activity. Flupirtine treatment slowed down significantly the locomotor hyperactivity of male and female *Cln3*Δ*ex7*/*<sup>8</sup>* mice (Figure 2A,B).

**Figure 2.** Impact of flupirtine on motor behavior of male and female *Cln3*Δ*ex7*/*8*mice. Open field behavioral parameter (mobility duration) in vehicle-treated WT, vehicle-treated *Cln3*Δ*ex7*/*8*, and flupirtine-treated *Cln3*Δ*ex7*/*8*(**A**) male and (**B**) female mice (*n* = 8 per group). Pole climbing test in vehicle-treated WT, vehicle-treated *Cln3*Δ*ex7*/*8*, and flupirtine-treated *Cln3*Δ*ex7*/*8*(**C**) males and (**D**) female mice (*n* = 8 per group). Time to turn (tturn), time needed to reach center of pole (t1/2), amount of time stopped (tstop), and total time to descend (ttotal) were recorded. All data are expressed as mean <sup>±</sup> SEM. \$ *<sup>p</sup>*: compared to vehicle-treated WT mice; and \* *p*: compared to vehicle-treated *Cln3*Δ*ex7*/*8*mice. \* *p* < 0.05, \*\* *p* < 0.01, \$\$\$ *p* < 0.001, and \$\$\$\$ *p* < 0.0001. Corticosterone levels in vehicle-treated WT and *Cln3*Δ*ex7*/*8*(**E**) male and (**F**) female mice (*n* = 8 per group). Data are expressed as mean ± SEM.

The pole climbing test measures motor coordination, vertical orientation capability, and balance of mice. Vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* male and female mice descended the pole faster than vehicle-treated WT mice (Figure 2C,D), in line with the hyperactivity of *Cln3*Δ*ex7*/*<sup>8</sup>* mice observed in the open field experiment. In males, flupirtine supplementation had a significant impact on increasing and delaying the descent compared to vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* mice (Figure 2C). In females, flupirtine supplementation also trended to increase and delay the descent compared to vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* mice, however it did not reach significance (Figure 2D).

#### *3.2. Impact of Flupirtine on Learning and Memory of Homozygous Cln3*Δ*ex7*/*<sup>8</sup> Mice*

The Morris water maze (MWM) is a test used to assess cognitive function, more specifically, spatial learning and memory. The 'probe trial', in which the platform was removed, was used to assess spatial memory for the previously learned platform location. The time spent in the target quadrant compared to the average time spent in the non-target quadrants is an indicator of the animal's recall of platform location. WT male and female mice spent significantly more time in the target quadrant as compared to the average time spent in the non-target quadrants, suggestive of recall of platform location (Figure 3A–C). Vehicle-treated *Cln3*Δ*ex7*/*8*male and female mice spent significantly less time than their WT male littermates in the target quadrant (Figure 3A–C). Flupirtine treatment improved memory retention in *Cln3*Δ*ex7*/*<sup>8</sup>* male and female mice, with a significant increase in time spent in the target quadrant compared to vehicle-treated *Cln3*Δ*ex7*/*8*mice, suggestive of recall of platform location (Figure 3A–C).

**Figure 3.** Impact of flupirtine on learning and memory of male and female *Cln3*Δ*ex7*/*<sup>8</sup>* mice. Percentage of time spent in target vs. non-target quadrant of (**A**) male and (**B**) female mice in the different groups (vehicle-treated WT, vehicle-treated *Cln3*Δ*ex7*/*8*, and flupirtine-treated *Cln3*Δ*ex7*/*8*) during the probe test (*<sup>n</sup>* <sup>=</sup> 8 per group). All data are expressed as mean <sup>±</sup> SEM. \$ *<sup>p</sup>*: Target quadrant compared to non-target quadrant; and \* *p*: compared to vehicle-treated WT mice. \* *p* < 0.05, \*\* *p* < 0.01, \*\*\*\* *p* < 0.0001, \$\$\$ *p* < 0.001, and \$\$\$\$ *p* < 0.0001. (**C**) Representative heat maps of swimming paths of one mouse from different groups during the probe test in male and female mice. An empty, bolded red circle indicates location of the target platform (o).

#### *3.3. Impact of Flupirtine on Anti-Apoptotic Gene Expression in Male and Female Cln3*Δ*ex7*/*<sup>8</sup> Mice*

The expression of several anti-apoptotic (*Bcl-2, Bcl-xL, Akt, Xiap*) and pro-apoptotic genes (*Fadd, Cytochrome C, Caspase 3, Caspase 6, Caspase 9, Apaf-1, Bad, Bax*) were measured in male and female mice. The anti-apoptotic gene B-cell lymphoma extra-large (*Bcl-xl*) gene expression level was downregulated in vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* mice compared to vehicle-treated WT male mice (Figure 4A). Flupirtine treatment had a significant effect only on *Bcl-xl* expression levels, by increasing mRNA expression level in *Cln3*Δ*ex7*/*<sup>8</sup>* male mice (Figure 4A). In female mice, B-cell lymphoma 2 (*Bcl-2*) gene expression was slightly higher in vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* mice compared to vehicle-treated WT female mice (Figure 4B). Flupirtine, however, significantly increased only the mRNA expression level of *Bcl-2* in *Cln3*Δ*ex7*/*<sup>8</sup>* compared to vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* mice (Figure 4B). The expression level of other anti-apototic and pro-apototic genes remained unchanged (data not shown).

**Figure 4.** Mouse brain gene expression of anti-apoptotic markers. (**A**) *Bcl-xl* gene expression levels in the brain of vehicle-treated WT, vehicle-treated *Cln3*Δ*ex7*/*8*, and flupirtine-treated *Cln3*Δ*ex7*/*<sup>8</sup>* male mice (*<sup>n</sup>* <sup>=</sup> 4 per group). Data are expressed as mean <sup>±</sup> SEM, \* *<sup>p</sup>*: compared to *Cln3*Δ*ex7*/*<sup>8</sup>* vehicle-treated mice. \* *p* < 0.05; (**B**) Bcl-2 gene expression levels in the brain of vehicle-treated WT, vehicle-treated *Cln3*Δ*ex7*/*8*, and flupirtine-treated *Cln3*Δ*ex7*/*<sup>8</sup>* female mice (*<sup>n</sup>* <sup>=</sup> 4 per group). Data are expressed as mean <sup>±</sup> SEM. \$ *p*: compared to vehicle-treated WT mice, and, \* *p*: compared to vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* mice. \*\*\*\* *p* < 0.0001, \$ *p* < 0.05, and \$\$\$\$ *p* < 0.0001.

#### *3.4. Impact of Flupirtine on Gene Expression of Enzymes of Ceramide Metabolism in Cln3*Δ*ex7*/*<sup>8</sup> Mice*

Several ceramide synthesis enzymes (*Sptlc2, Sptlc3, Kdsr, CerS1-6, Degs1, Degs2, Smpd2, Smpd3, Gba, GalC*) and ceramide degradation enzymes (*Asah1, Asah2, Samd8, Ugcg, Ugt8*) were investigated in male and female mice. Serine palmitoyltransferase 3 (*Sptlc3*) levels were significantly upregulated in vehicle-treated *Cln3*Δ*ex7*/*8*mice compared to vehicle-treated WT male mice (Figure 5A). Flupirtine treatment had a significant impact only on the expression level of key ceramide synthesis enzyme, Sptlc3, in the de novo pathway, by reducing mRNA expression levels compared to *Cln3*Δ*ex7*/*8*vehicle-treated male mice (Figure 5A). In female mice, sterile alpha motif domain containing 8 (*Samd8*) gene expression level did not differ between vehicle-treated WT and vehicle-treated *Cln3*Δ*ex7*/*8*female mice. *Samd8* is an endoplasmic reticulum (ER) transferase that converts phosphatidylethanolamine (PE) and ceramide to ceramide phosphoethanolamine (CPE). Flupirtine significantly increased expression levels of only *Samd8* compared to vehicle-treated WT and to *Cln3*Δ*ex7*/*8*female mice (Figure 5B). The expression level of other ceramide synthesis/degradation enzymes remained unchanged (data not shown).

**Figure 5.** Mouse brain gene expression of ceramide synthesis/degradation enzymes. (**A**) Sptlc3 gene expression levels in the brain of vehicle-treated WT, vehicle-treated *Cln3*Δ*ex7*/*8*, and flupirtine-treated *Cln3*Δ*ex7*/*<sup>8</sup>* male mice (*<sup>n</sup>* <sup>=</sup> 4 per group). Data are expressed as mean <sup>±</sup> SEM. \$ *<sup>p</sup>*: compared to WT vehicle-treated mice, and, \* *p*: compared to vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* mice. \* *p* < 0.05, and \$\$ *p* < 0.01; (**B**) Samd8 gene expression levels in the brain of vehicle-treated WT, vehicle-treated *Cln3*Δ*ex7*/*8*, and flupirtine-treated *Cln3*Δ*ex7*/*<sup>8</sup>* female mice (*<sup>n</sup>* <sup>=</sup> 4 per group). Data are expressed as mean <sup>±</sup> SEM. \$ *p*: compared to vehicle-treated WT mice, and, \* *p*: compared to vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* mice. \* *p* < 0.05 and \$ *p* < 0.05.

#### *3.5. E*ff*ect of Flupirtine Supplementation on Astrocytosis in Cln3*Δ*ex7*/*<sup>8</sup> Mouse Brains*

Fluorescence microscopy demonstrated significant enhanced GFAP immunostaining (green) in vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* male mice relative to WT littermates in CA1/2 and CA3 hippocampus, dentate gyrus (DG), and motor cortex (MC) (Figure 6A,B). Treatment with flupirtine decreased glial activation in all brain regions studied in *Cln3*Δ*ex7*/*<sup>8</sup>* male mice (Figure 6A,B). Results were statistically significant only in CA1/2 hippocampus, and motor cortex (MC). Hoechst staining (blue) indicates diminished number of hippocampal neurons in vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* versus WT, and also an increase in neurons in flupirtine-treated male mice versus vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* male mice (Figure 6A).

In females, GFAP immunostaining was significantly enhanced in vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* mice relative to their WT female littermates in CA1/2 and CA3 hippocampus, as well as in DG (Figure 7A,B). Treatment with flupirtine significantly decreased astrogliosis in CA1/2 and CA3 hippocampus, and DG in *Cln3*Δ*ex7*/*<sup>8</sup>* female mice (Figure 7A,B). Also, note an increase in the number of blue, Hoechst-stained neurons in hippocampus and dentate gyrus in both wild type and flupirtine-treated *Cln3*Δ*ex7*/*<sup>8</sup>* female mice compared to vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* female mice (Figure 7A). In motor cortex (MC) of females, no significant difference in GFAP levels was observed (data not shown).

**Figure 6.** Impact of flupirtine supplementation on astrocytosis in hippocampus and motor cortex of male *Cln3*Δ*ex7*/*<sup>8</sup>* mice. (**A**) Representative images of hippocampus regions CA1, CA2, CA3 and dentate gyrus (DG), as well as primary motor cortex from vehicle-treated WT, vehicle-treated *Cln3*Δ*ex7*/*8*, and flupirtine-treated *Cln3*Δ*ex7*/*<sup>8</sup>* male mice stained with GFAP in green and counterstained with Hoechst in blue (*n* = 4; Scale bars = 20 μm).; (**B**) Glial fibrillary acidic protein (GFAP) mean fluorescence over area in vehicle-treated WT, vehicle-treated *Cln3*Δ*ex7*/*8*, and flupirtine-treated *Cln3*Δ*ex7*/*<sup>8</sup>* male mice in hippocampus regions CA1, CA2, CA3 and dentate gyrus (DG), as well as in primary motor cortex (MC). Data are expressed as mean <sup>±</sup> SEM. \$ *<sup>p</sup>* <sup>&</sup>lt; 0.05 compared to vehicle-treated WT, \* *<sup>p</sup>* <sup>&</sup>lt; 0.05 compared to vehicle-treated *Cln3*Δ*ex7*/*8*.

**Figure 7.** Impact of flupirtine supplementation on astrocytosis in hippocampus and motor cortex of female *Cln3*Δ*ex7*/*8*mice. (**A**) Representative images of hippocampus regions CA1, CA2, CA3 and dentate gyrus (DG), from vehicle-treated WT, vehicle-treated *Cln3*Δ*ex7*/*8*, and flupirtine-treated *Cln3*Δ*ex7*/*8*female mice stained with GFAP in green and counterstained with Hoechst in blue (*n* = 4; scale bars = 20 μm).; (**B**) glial fibrillary acidic protein (GFAP) mean fluorescence over area in vehicle-treated WT, vehicle-treated *Cln3*Δ*ex7*/*8*, and flupirtine-treated *Cln3*Δ*ex7*/*8*male mice in hippocampus regions CA1, CA2, CA3 and dentate gyrus (DG), as well as in primary motor cortex (MC). Data are expressed as mean <sup>±</sup> SEM. \$ *p:* compared to vehicle-treated WT, \* *p*: compared to vehicle-treated *Cln3*Δ*ex7*/*8*. \* *p* < 0.05, \*\* *p* < 0.01, \$\$ *p* < 0.01, and \$\$\$ *p* < 0.001.

#### *3.6. Impact of Flupirtine Supplementation on Neuronal Cell Counts in Cln3*Δ*ex7*/*<sup>8</sup> Brains*

The number of NeuN-positive cells decreased significantly in the MC of vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* male mice versus vehicle-treated WT mice (Figure 8A,B). Although not significant, NeuN-stained cells in MC of flupirtine-treated *Cln3*Δ*ex7*/*<sup>8</sup>* male mice compared to vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* mice showed a slight increase, close to the level in WT mice (Figure 8A,B). No significant difference in the neuronal cell counts of female MC was seen (data not shown).

**Figure 8.** Effect of flupirtine treatment on motor cortex neuron numbers in *Cln3*Δ*ex7*/*<sup>8</sup>* mice. (**A**) Representative images of primary motor cortex from vehicle-treated WT, vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* and flupirtine-treated *Cln3*Δ*ex7*/*<sup>8</sup>* male mice stained with NeuN (mature neuronal marker) in green and counterstained with Hoechst in blue (*n* = 4; Scale bars = 50 μm).; (**B**) NeuN-positive cells normalized to area in the primary motor cortex (MC) of vehicle-treated WT, vehicle-treated *Cln3*Δ*ex7*/*8*, and flupirtine-treated *Cln3*Δ*ex7*/*<sup>8</sup>* male mice. Data are expressed as mean <sup>±</sup> SEM. \* *<sup>p</sup>* <sup>&</sup>lt; 0.05 compared to vehicle-treated WT.

#### *3.7. Impact of Compound 6 Supplementation on Behavioral Parameters*

In the open field test, compound 6 significantly slowed down locomotor hyperactivity of male and female *Cln3*Δ*ex7*/*<sup>8</sup>* mice (Figure 9A,B). Similarly, a similar effect was seen in the pole-climbing test, where compound 6 significantly increased time needed by male mice to descend the pole compared to vehicle treated mice (Figure 9C).

**Figure 9.** Impact of compound 6 treatment in *Cln3*Δ*ex7*/*<sup>8</sup>* mice. Open field behavioral parameter (mobility duration) in vehicle-treated WT, vehicle-treated *Cln3*Δ*ex7*/*8*, and compound 6-treated *Cln3*Δ*ex7*/*<sup>8</sup>* (**A**) male and (**B**) female mice (*n* = 8 per group). Data are expressed as mean <sup>±</sup> SEM. \$ *p*: compared to WT vehicle-treated mice, and, \* *p*: compared to vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* mice. \*\*\* *p* < 0.001, \$\$\$ *p* < 0.001, and \$\$\$\$ *p* < 0.0001. (**C**) Pole climbing test in vehicle-treated WT, vehicle-treated *Cln3*Δ*ex7*/*8*, and flupirtine-treated *Cln3*Δ*ex7*/*<sup>8</sup>* male mice (*n* = 8 per group). Time to turn (tturn), time needed to reach center of pole (t1/2), amount of time stopped (tstop), and total time to descend (ttotal) were recorded. Data are expressed as mean ± SEM. \* *p* < 0.05, \*\* *p* < 0.01.

#### **4. Discussion**

This study addresses novel small molecule treatment strategies for CLN3 disease in a *Cln3*Δ*ex7*/*<sup>8</sup>* knock-in mouse model. Hopefully, this will translate into future knowledge to improve the lives of CLN3 patients.

Flupirtineis well known forits significant powerful anti-oxidative, anti-apoptotic, and neuroprotective effects in vitro and in vivo. The effectiveness of the chosen daily dose of flupirtine (30 mg/kg per os) has been demonstrated in testing for anti-nociceptive, anticonvulsant, and anti-apoptotic activity in rodents [23–27]. This study is the first to evaluate the use of flupirtine as potential treatment for CLN3 disease in an animal model, i.e., in vivo. Previous studies proved that flupirtine reaches brain regions, including hippocampus and cortex, and that it rapidly crosses the blood brain barrier and enters other tissues. The liver is the primary organ responsible for its metabolism [15].

Behavioral tests assessed different aspects of *Cln3*Δ*ex7*/*<sup>8</sup>* mouse motor strength, coordination, balance, as well as learning and spatial memory functions before and after flupirtine treatment. Male and female vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* mice exhibit significantly increased mobility with respect to WT controls in the open field test. The hyperactive phenotype prominent in vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* mice is described for the first time here in mice at of 18 weeks of age. This was not previously documented in *Cln3*Δ*ex7*/*<sup>8</sup>* mice at 40 weeks of age [28]. This suggests that, at a young age, *Cln3*Δ*ex7*/*<sup>8</sup>* mice express a distinct behavioral phenotype prior to onset of more severe CLN3 disease symptoms, including motor decline. *Cln3*Δ*ex7*/*<sup>8</sup>* mice also showed increased random and chaotic exploratory activity compared to WT controls in a novel environment. This indicates inattentiveness and diminished executive function. Treatment with flupirtine significantly attenuated this abnormal mobility in male and female *Cln3*Δ*ex7*/*<sup>8</sup>* mice. This phenotype may be consistent with the notion that diminished executive function of vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* mice is driving their inattentive hyperactivity. The vertical pole test evaluates spatial and motor orientation and balance of mice [29]. Reduced latency of vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* mice to descend the pole is explained by their quick and less controlled behavior compared to WT animals, consistent with open field behavioral assessments of excessive mobility. The aforementioned behavioral test reveals prominent hyperactivity and attentional deficits in vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* mice, while treatment with flupirtine lessened this impulsive phenotype. These results are in line with other mouse models of neurodevelopmental psychiatric disorder of attention-deficit hyperactivity disorder (ADHD). Mice recapitulating this disease are characterized by impulsivity, inattentiveness, and hyperactivity [30]. Stress has a direct profound effect on rodent behavior and physical activity [31]. Chronic stress leads to increased corticosterone levels [31], as observed in WT male and female mice with respect to *Cln3*Δ*ex7*/*<sup>8</sup>* mice. Several studies elucidate that animals who do not experience stress show higher exploration, locomotion, and physical activity in an open field test, as a reaction to an unknown environment [32,33]. Therefore, the markedly decreased corticosterone levels in serum of *Cln3*Δ*ex7*/*<sup>8</sup>* mice with respect to WT animals explains their hyperactivity resulting from reduced stress levels in male and female *Cln3*Δ*ex7*/*<sup>8</sup>* mice. Learning and cognitive ability of mice was tested at 16 weeks of age using the Morris water maze (MWM) test. Analysis of the swim paths of mice during the probe trial showed that flupirtine-treated male and female *Cln3*Δ*ex7*/*<sup>8</sup>* mice adopted strategies and maintained spatial preference in the target quadrant, contrary to vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* mice that show coverage of the whole maze. Flupirtine significantly enhanced spatial learning, navigation and memory retention in *Cln3*Δ*ex7*/*<sup>8</sup>* male and female mice.

Apoptosis is a naturally-occurring mechanism of cell death and helps maintain tissue homeostasis [34]. Neuronal cell loss is evident in brain sections from post-mortem CLN3 disease patients [3]. Numerous apoptotic cells are present within cortical brain sections from CLN3 disease patients [3]. CLN3 patient-derived lymphoblasts have decreased growth rate compared to normal lymphoblasts, validating that apoptosis is one of the mechanisms implicated in CLN3 disease pathogenesis [35]. Other studies demonstrate damage and apoptosis of neuronal and glial cells in hippocampus and cortex of CLN3 patients in addition to marked loss of cortical neurons due to apoptotic cell death [6]. CLN3 defects also perturb calcium signaling, leading to a profound defects in neuronal survival [36].

Most neuronal death in CLN2 and CLN3 brains takes place via apoptosis, and the surviving neurons upregulate Bcl-2 [5]. Treatment with flupirtine significantly upregulated expression of anti-apoptotic BCL-2 in CLN3-deficient cells in vitro [20]. This is the case *in vivo* in this current study as flupirtine-treated *Cln3*Δ*ex7*/*<sup>8</sup>* female mice show a remarkable increase in *Bcl-2* expression. In males, another anti-apoptotic protein, *Bcl-xl*, was upregulated following treatment with flupirtine. Different proteins were impacted in male versus female mice, yet the end result was upregulation of anti-apoptotic pathways and hence, reduction of cell death in brains of *Cln3*Δ*ex7*/*<sup>8</sup>* mice given flupirtine. This variation among sexes is not a new observation in this disease. We documented this in another study using exogenous galactosylceramide as potential treatment for CLN3 disease [28].

CLN3 is directly implicated in apoptotic cell death signaling cascades by activating caspase-dependent and caspase-independent pathways [8]. Ceramide is a major sphingolipid second messenger implicated in several cell processes and impacts divergent pathways [37]. Sphingolipids are major bioactive lipids involved in homeostasis, growth, proliferation and cell death [38]. Ceramide mediates anti-proliferative events, such as apoptosis, growth inhibition, cell differentiation, and senescence [10]. This biomolecule possesses complex biophysical properties and acts as a central hub. Regulation of its levels affects catabolism and break-down of various sphingolipid species [10]. Ceramide is generated through several complex interrelated pathways either via the de novo pathway, sphingomyelin, or cerebroside catabolism. Ceramide is synthesized de novo in the endoplasmic reticulum or through breakdown of sphingomyelin in Golgi, plasma membrane, or mitochondrial membrane [39]. Defects in ceramide signaling pathways often result in augmenting programmed cell death in multiple cell types, including neurons [40]. Previous studies show that ceramide levels are increased in CLN3-deficient cells and brain of CLN3 patients [11]. Published reports demonstrate that 17 week-old *Cln3*Δ*ex7*/*8*mice express higher levels of ceramide in brain compared to age-matched WT mice [14]. Sptlc3 catalyzes the initial steps in formation of ceramide via the de novo pathway by condensing serine and palmitoyl Co-A to generate 3-ketoshphinganine (3-KDS) [10]. We documented decreased *Sptlc3* levels in flupirtine-treated *Cln3*Δ*ex7*/*8*mice versus vehicle-treated *Cln3*Δ*ex7*/*8*male mice. Downregulation of *Sptlc3* leads to diminution of ceramide generation via the de novo ceramide pathway. As for the other enzymes of the de novo pathway, the expression of *Sptlc2* and *Degs1* also decreased with flupirtine treatment, but did not reach significance (data not shown). This supports our conclusion that flupirtine treatment in male mice impacts the de novo synthesis pathway. In females, however, a different pathway in ceramide signaling is at play. *Samd8* is an ER transferase that converts phosphatidylethanolamine (PE) and ceramide to ceramide phosphoethanolamine (CPE). *Samd8* levels are increased in flupirtine-treated *Cln3*Δ*ex7*/*8*female mice. *Samd8* operates as a ceramide sensor to control ceramide homeostasis in the ER rather than a converter of ceramides. This implicates that the ceramide salvage pathway is modulated in flupirtine-treated *Cln3*Δ*ex7*/*8*female mice with the end result of diminished ceramide levels. Empirical evidence from our study confirms reduced synthesis (decreased *Sptlc3* expression) in flupirtine-treated male mice, but increased degradation of ceramide (increased *Samd8* expression) in female flupirtine-treated mice. Physiologic gender differences affects drug activity and charactersitics, including pharmacokinetics [41]. Differences in body size result in larger distribution volumes, faster total clearance, and less tissue absorption of some medications in men compared to women [42]. This may explain higher brain absorption in female compared to male mice. Moreover, sex hormones, in females, have a direct effect on drug absorption, distribution, metabolism, elimination and adverse effects [43].This study implies that sex-specific drug dosing regimens may be warranted for treatment of neurological diseases that affect the blood brain barrier, including CLN3 disease in mouse and man.

Activation of the glial cell population contributes to imbalance in CNS function and impacts cognitive function [44]. Hyperactive astrocytes are observed in neurodegenerative disorders and following brain injury documented by GFAP as biomarker. In males, microscopic inspection of vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* mouse brain documents widespread intracellular GFAP staining in hippocampal regions (CA1, CA2, CA3, and the dentate gyrus) and in the MC relative to age-matched, vehicle-treated WT animals. Mice treated with flupirtine had significantly lower levels of GFAP staining in these regions. This data provides evidence that flupirtine attenuates astrogliosis at the level of the hippocampus and MC in *Cln3*Δ*ex7*/*<sup>8</sup>* male mice. In females, flupirtine was able to attenuate GFAP immunostaining in CA1/2, CA3 and DG regions of the hippocampus. The motor cortex did not show any difference in vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* female mice compared to WT. This may explain better performance in *Cln3*Δ*ex7*/*<sup>8</sup>* female mice on the rotarod compared to males as it tests motor skills (data not shown).

The neuronal nuclear protein (NeuN) is a marker not detected in glial cells or other cells in the brain [28]. NeuN assesses neuronal health and loss of this protein is indicative of damage. Neuronal cell loss in CLN3 patients is at the root of CLN3 disease pathogenesis [45]. NeuN-positive cell were significantly ablated in vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* mouse. Although it did not reach significance, flupirtine resulted in an increase in the neuronal population in motor cortex (MC) of male mice. This tsuggests that flupirtine conferred neuroprotection and reduced cell death in brains of *Cln3*Δ*ex7*/*<sup>8</sup>* mice. In females, there was no difference in neuronal counts in motor cortex of vehicle-treated *Cln3*Δ*ex7*/*<sup>8</sup>* female mice compared to WT. In a previous study, affected female mice 44 weeks of age did show a diminution in NeuN positive cells [28], suggesting that unlike males, neuronal loss in females starts at a later age.

The novel, flupirtine-like allyl carbamate derivative, compound 6, developed to possess physicochemical properties desirable for CNS therapeutics had an improved Multiparameter optimization (MPO) score. The latter predicts blood–brain barrier (BBB) penetration, an essential parameter for neuroprotective compounds, and was ≥ 4 more effective [19]. The newly synthesized allyl carbamate derivative of flupirtine, compound 6, showed potential for improved neuroprotection, after screening in vitro nine flupirtine derivatives [20]. Here, we report early promising behavioral in vivo results for compound 6 for treatment of *Cln3*Δ*ex7*/*<sup>8</sup>* mice. Although pharmacokinetics and toxicological safety remain to be established for compound 6, the promising behavioral data obtained in this study are worth reporting. Treatment with compound 6 significantly attenuated the high mobility documented by open field and pole climbing in *Cln3*Δ*ex7*/*<sup>8</sup>* mice, suggesting more work is necessary to determine optimal dosing for this compound.

#### **5. Conclusions**

In conclusion, these findings suggest that flupirtine, and compound 6, improve neurobehavioral measures. Flupirtine impacted ceramide biosynthesis and apoptotic signaling pathways. Flupirtine affected a broad-spectrum of actionable targets, providing insights into the pathobiology of CLN3 disease in humans, particularly uncovering impact on gender-specific signaling pathways. Flupirtine shows promise in males and females, and the allyl carbamate derivative, compound 6, needs further preclinical analyses and development. These findings extend our knowledge of the role of drugs in the treatment of a fatal pediatric neurodegenerative disease implying more work lies ahead for development into clinically applicable therapies for CLN3 patients.

#### **6. Patents**

R.-M.B. has an Application for Method of Treating Batten Disease. Inventor: Rose-Mary Boustany. Duke (File No. 5405-240 PR). US Patent and Trademark No. 10/148,859 (U.S. National Phase); Use Patent issued 11/23/2004 US Patent # 6 821 995, expired 11/23/2014.

N.K., P.C.T. and R.-M.B. are inventors on a patent application detailing the aromatic carbamates described herein: 'Functionalized Pyridine Carbamates with enhanced Neuroprotective Activity' PCT Int. Appl. (2019), WO2019014547 A8.

**Author Contributions:** K.M., J.M., S.S., and N.J.M. performed experiments, analyzed data, and interpreted the results. A.V.C., N.K., and P.C.T. synthesized the drugs. N.G. helped in analyzing data and interpreting results. K.M., J.M., and S.S. prepared the figures, and drafted the main manuscript. R.-M.B. conceived the study, obtained

funding for the study, designed experiments, reviewed data and analyses, and revised and edited the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:**Wewouldlike to thankOpenMinds for financial support (R.-M.B.) for thiswork (# 620229), AUB—Collaborative Research Grant (N.G and R.-M.B.) (#24473), and University Research Board (URB) at AUB (N.G.) and the Lebanese National Council for Scientific Research (N.G.).

**Acknowledgments:** A very particular acknowledgement to my late and dear friend Lina Marie Obeid, who, together with her husband, Yusuf Hannun, introduced me to the importance of sphingolipid biology.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

#### *Article*
