Gemfibrozil and Fenofibrate, Food and Drug Administration-approved Lipid-lowering Drugs, Up-regulate Tripeptidyl-peptidase 1 in Brain Cells via Peroxisome Proliferator-activated Receptor α

Background: Increase in tripeptidyl-peptidase 1 (TPP1) is a possible therapeutic approach for late infantile Batten disease or neuronal ceroid lipofuscinosis (LINCL). Results: Gemfibrozil and fenofibrate, FDA-approved drugs for hyperlipidemia, stimulate TPP1 in brain cells via the PPARα/RXRα pathway. Conclusion: These results delineate a novel TPP1 up-regulating property of gemfibrozil and fenofibrate. Significance: Gemfibrozil and fenofibrate may be of therapeutic benefit in Batten disease. The classical late infantile neuronal ceroid lipofuscinosis (LINCLs) is an autosomal recessive disease, where the defective gene is Cln2, encoding tripeptidyl-peptidase I (TPP1). At the molecular level, LINCL is caused by accumulation of autofluorescent storage materials in neurons and other cell types. Currently, there is no established treatment for this fatal disease. This study reveals a novel use of gemfibrozil and fenofibrate, Food and Drug Administration-approved lipid-lowering drugs, in up-regulating TPP1 in brain cells. Both gemfibrozil and fenofibrate up-regulated mRNA, protein, and enzymatic activity of TPP1 in primary mouse neurons and astrocytes as well as human astrocytes and neuronal cells. Because gemfibrozil and fenofibrate are known to activate peroxisome proliferator-activated receptor-α (PPARα), the role of PPARα in gemfibrozil- and fenofibrate-mediated up-regulation of TPP1 was investigated revealing that both drugs up-regulated TPP1 mRNA, protein, and enzymatic activity both in vitro and in vivo in wild type (WT) and PPARβ−/−, but not PPARα−/−, mice. In an attempt to delineate the mechanism of TPP1 up-regulation, it was found that the effects of the fibrate drugs were abrogated in the absence of retinoid X receptor-α (RXRα), a molecule known to form a heterodimer with PPARα. Accordingly, all-trans-retinoic acid, alone or together with gemfibrozil, up-regulated TPP1. Co-immunoprecipitation and ChIP studies revealed the formation of a PPARα/RXRα heterodimer and binding of the heterodimer to an RXR-binding site on the Cln2 promoter. Together, this study demonstrates a unique mechanism for the up-regulation of TPP1 by fibrate drugs via PPARα/RXRα pathway.

known as ceroid-lipofuscin (5). Currently, there is no established treatment or drugs available for this disease; all approaches are merely supportive or symptomatic, indicating a need for novel therapeutic approaches (7). However, there are different variants of Cln2 mutations, and there have been reports that residual TPP-I activity can be found in patients with LINCL, indicating that there must be a few copies of the normal Cln2 gene remaining in patients affected with LINCL (8,9). Thus, one approach for treatment may be to find ways to enhance the levels and residual activity of the TPP1 protein to ameliorate the disease.
Gemfibrozil and fenofibrate, two of the most significant FDA-approved lipid-lowering drugs, reduce the level of triglycerides in the blood circulation and decrease the risk of hyperlipidemia (10 -12). However, a number of recent studies reveal that apart from its lipid-lowering effects, these drugs, especially gemfibrozil, can also regulate many other signaling pathways responsible for inflammation, such as switching of T-helper cells, cell-to-cell contact, migration, and oxidative stress (13)(14)(15)(16). Here, we describe that both gemfibrozil and fenofibrate are capable of enhancing TPP1 in cultured neurons and glial cells and in vivo in the brain. We also demonstrate that PPAR␣, but not PPAR␤ and PPAR␥, is involved in gemfibrozil-and fenofibrate-mediated up-regulation of TPP1. Furthermore, we also demonstrate that fibrate drugs up-regulate TPP1 via activation of the PPAR␣/RXR␣ heterodimer. Collectively, this study suggests that gemfibrozil and fenofibrate, FDA-approved drugs for hyperlipidemia, may be of therapeutic value in the treatment of LINCL.
Isolation of Mouse Primary Astroglia-Astroglia were isolated from mixed glial cultures as described (17,18) and according to the procedure of Giulian and Baker (19). Briefly, on day 9, the mixed glial cultures were washed three times with Dulbecco's modified Eagle's medium/F-12 and subjected to shaking at 240 rpm for 2 h at 37°C on a rotary shaker to remove microglia. After 2 days, the shaking was repeated for 24 h for the removal of oligodendroglia and to ensure the complete removal of all non-astroglial cells. The attached cells were seeded onto new plates for further studies.
Isolation of Primary Human Astroglia-Primary human astroglia were prepared as described (20,21). All experimental protocols were reviewed and approved by the Institutional Review Board of the Rush University Medical Center. Briefly, 11-17-week-old fetal brains obtained from the Human Embryology Laboratory (University of Washington, Seattle) were dissociated by trituration and trypsinization. On the 9th day, these mixed glial cultures were placed on a rotary shaker at 240 rpm at 37°C for 2 h to remove loosely attached microglia. On the 11th day, the flasks were shaken again at 190 rpm at 37°C for 18 h to remove oligodendroglia. The attached cells remaining were primarily astrocytes. These cells were trypsinized and subcultured in complete media at 37°C with 5% CO 2 in air to yield more viable and healthy cells. By immunofluorescence assay, these cultures homogeneously expressed GFAP, a marker for astrocytes (22).
Isolation of Neurons from Different Brain Regions-Fetal (E18 -E16) mouse neurons were prepared as described previously (23) with modifications. Whole brains were removed, and cortical, hippocampal, striatal, and cerebellar fractions were dissected in serum-free Neurobasal media. The cells were washed by centrifugation three times at 1200 rpm for 10 min; the pellet was dissociated, and the cells were plated at 10% confluence in 8-well chamber slides pretreated for Ͼ2 h with poly-D-lysine (Sigma). After 4 min, the nonadherent cell suspension was aspirated, and 500 ml of complete Neurobasal media (Invitrogen) supplemented with 2% B27 was added to each well. The cells were incubated for 4 days prior to experimentation. Double-label immunofluorescence with ␤-tubulin and either GFAP or CD11b revealed that neurons were more than 98% pure (data not shown). The cells were stimulated with gemfibrozil in Neurobasal media supplemented with 2% B27 minus antioxidants (Invitrogen) for 24 h prior to methanol fixation and immunostaining.
Amplified products were electrophoresed on 2% agarose (Invitrogen) gels and visualized by ethidium bromide (Invitrogen) staining. Glyceraldehyde-3-phosphate dehydrogenase (Gapdh) mRNA was used as a loading control to ascertain that an equivalent amount of cDNA was synthesized from each sample.
Quantitative Real Time PCR-The mRNA quantification was performed using the ABI-Prism7700 sequence detection system (Applied Biosystems, Foster City, CA) using iTaq TM Fast Supermix with ROX (Bio-Rad) and the following 6-FAM/ ZEN/IBFQ-labeled primers for murine genes Cln2 and Gapdh (Integrated DNA Technologies, Coralville, IA). The mRNA expression of the targeted genes was normalized to the level of Gapdh mRNA, and data were processed by the ABI Sequence Detection System 1.6 software.
Immunostaining of Tissue Sections-After 21 days of treatment, mice were sacrificed, and their brains were fixed, embedded, and processed. Sections were made from different brain regions and for immunofluorescence staining on fresh frozen sections, and anti-mouse TPP1 (1:200), goat anti-mouse GFAP (1:100) were used. The samples were mounted and observed under Olympus BX41 fluorescence microscope (26).
Immunoblotting-Western blotting was conducted as described earlier (27,28) with modifications. Briefly, cells were scraped in double-distilled H 2 O and SDS and electrophoresed on NuPAGE Novex 4 -12% BisTris gels (Invitrogen), and proteins were transferred onto a nitrocellulose membrane (Bio-Rad) using the Thermo-Pierce Fast Semi-Dry Blotter. The membrane was then washed for 15 min in TBS plus Tween 20 (TBST) and blocked for 1 h in TBST containing BSA. Next, membranes were incubated overnight at 4°C under shaking conditions with the following 1°antibodies: TPP1 (1:250, Santa Cruz Biotechnology) and ␤-actin (1:800; Abcam, Cambridge, MA). The next day, membranes were washed in TBST for 1 h, incubated in 2°antibodies against 1°antibody hosts (all 1:10,000; Jackson ImmunoResearch) for 1 h at room temperature, washed for 1 more h, and visualized under the Odyssey Infrared Imaging System (Li-COR, Lincoln, NE).
TPP1 Activity Assay-TPP-I activity was assayed in 96-well format plates using the following modification of the method described by Vines and Warburton (6). Briefly, samples and substrate (40 l) were mixed in individual wells of a polystyrene 96-well plate (Nalge Nunc International). The substrate solution consisted of 250 mol/liter Ala-Ala-Phe 7-amido-4-methylcoumarin (catalog no. A3401; Sigma; diluted freshly from a 25 mmol/liter stock solution in dimethyl sulfoxide stored at Ϫ20°C) in 0.15 mol/liter NaCl, 1 g/liter Triton X-100, 0.1 mol/ liter sodium acetate, adjusted to pH 4.0 at 20°C. Plates were centrifuged briefly to dispel bubbles and placed in a 37°C. Plates were mixed for 10 s before each reading. The plates were read from the bottom using 360/20 nm excitation and 460/25 nm emission filters. Prior to the assay, the optimum substrate concentration and total protein in cell extract that can used to get the best results were determined in the same manner described above, using different substrate concentrations and protein concentrations in the cell extract.
Immunoprecipitation from Nuclear Extract-After treatment, cells were washed with PBS, scraped into 1.5-ml tubes, and centrifuged in 4°C for 5 min at 500 rpm. The supernatant was aspirated, and the pellet was resuspended in a membrane lysis buffer consisting of HEPES (pH 8.0), MgCl 2 , KCl, dithiothreitol (DTT), and protease/phosphatase inhibitors, vortexed, and centrifuged at 4°C at 15,000 rpm for 3 min. Again, the supernatant was aspirated, and the pellet was resuspended in a high salt nuclear envelope lysis buffer consisting of HEPES (pH 8.0), MgCl 2 , glycerol, NaCl, EDTA, DTT, and protease/phosphatase inhibitors, rotated vigorously at 4°C for 30 min, and centrifuged at 4°C at 15,000 rpm for 15 min. The resultant nuclear pellet was resuspended in IP buffer, and a fraction was kept separately as lysate. The remaining nuclear extract was then precleared with 25 l of protein A-agarose (50%, v/v). The supernatants were immunoprecipitated with 5 g of anti-RXR␣ or anti-PPAR␣ or normal IgG (Santa Cruz Biotechnology) overnight at 4°C, followed by incubation with protein A-agarose for 4 h at 4°C. Protein A-agarose-antigen-antibody complexes were collected by centrifugation at 12,000 rpm for 60 s at 4°C. The pellets were washed five times with 1 ml of IP buffer (20 mM Tris-HCl (pH 8.0), 137 mM NaCl, 2 mM EDTA, 1% Nonidet P-40, 10% glycerol, 0.1 mM phenylmethylsulfonyl fluoride) for 20 min each time at 4°C. Bound proteins were resolved by SDS-PAGE, followed by Western blotting with the anti-RXR␣ (1:2000, Santa Cruz Biotechnology) and/or anti-PPAR␣ (1:250, Santa Cruz Biotechnology). The lysate was resolved by SDS-PAGE followed by immunoblot for PPAR␣, RXR␣, and H3.
Chromatin Immunoprecipitation Assay-ChIP assays were performed using the method described by Nelson et al. (29), with certain modifications. Briefly, mouse primary astrocytes were stimulated by 10 M gemfibrozil and 0.5 M RA together for 6 h followed by fixing with formaldehyde (1.42% final volume) and quenching with 125 mM glycine. The cells were pelleted and lysed in IP buffer containing 150 mM NaCl, 50 mM Tris-HCl (pH 7.5), 5 mM EDTA, Nonidet P-40 (0.5% v/v), Triton X-100 (1.0% v/v). For 500 ml, add 4.383 g of NaCl, 25 ml of 100 mM EDTA (pH 8.0), 25 ml of 1 M Tris-HCl (pH 7.5), 25 ml of 10% (v/v) Nonidet P-40, and 50 ml of 10% (v/v) Triton X-100 containing the following inhibitors: 10 g/ml leupeptin, 0.5 mM phenylmethylsulfonyl fluoride (PMSF), 30 mM p-nitrophenyl phosphate, 10 mM NaF, 0.1 mM Na 3 VO 4 , 0.1 mM Na 2 MoO 4 , and 10 mM ␤-glycerophosphate. After one wash with 1.0 ml of IP buffer, the pellet was resuspended in 1 ml of IP buffer (containing all inhibitors), and sonicated and sheared chromatin was split into two fractions (one to be used as Input). The remaining fraction was incubated overnight under rotation at 4°C with 5-7 g of anti-PPAR␣ or anti-RXR␣ Abs or normal IgG (Santa Cruz Biotechnology) followed by incubation with protein G-agarose (Santa Cruz Biotechnology) for 2 h at 4°C under rotation. Beads were then washed five times with cold IP buffer, and a total of 100 l of 10% Chelex (10 g/100 ml H 2 O) was added directly to the washed protein G beads and vortexed.
After 10 min boiling, the Chelex/protein G bead suspension was allowed to cool to room temperature. Proteinase K (100 g/ml) was then added, and the beads were incubated for 30 min at 55°C while shaking, followed by another round of boiling for 10 min. The suspension was centrifuged, and the supernatant was collected. The Chelex/protein G beads fraction was vortexed with another 100 l of water, centrifuged again, and the first and the second supernatants combined. Eluate was used directly as a template in PCR. The following primers were used to amplify fragments flanking the RXR-binding element in the mouse Cln2 promoter: Set1, sense 5Ј-CAG CTG CCA TGT CCC CCA GC-3Ј and antisense 5Ј-TGC GCA GCT CTG TGT CAT CCG-3Ј; Set2, sense 5Ј-GCT CCC TCT CCT CAG CTG CCA-3Ј and antisense 5Ј-CAT CCG GAG GCT CCA GGC CA-3Ј. The PCRs were repeated by using varying cycle numbers and different amounts of templates to ensure that results were in the linear range of PCR.
Densitometric Analysis-Protein blots were analyzed using ImageJ (National Institutes of Health, Bethesda), and bands were normalized to their respective ␤-actin loading controls. Data are representative of the average fold change with respect to control for three independent experiments.
Statistics-Values are expressed as means Ϯ S.E. of at least three independent experiments. Statistical analyses for differences were performed via Student's t test. This criterion for statistical significance was p Ͻ 0.05.

Fibrate Drugs Up-regulate TPP1 mRNA and Protein in
Mouse Primary Astrocytes-There have been reports that residual TPP-I activity can be found in patients indicating that a few copies of normal Cln2 gene are left in patients affected with LINCL (8,9,30,31). We examined if FDA-approved lipid-lowering drugs like gemfibrozil and fenofibrate were capable of up-regulating the expression of TPP1 in brain cells. Mouse primary astrocytes were treated in serum-free media with gemfibrozil with different doses and for different time points. Both RT-PCR and real time quantitative PCR (qPCR) analyses clearly indicated that gemfibrozil up-regulated Cln2 mRNA levels in mouse primary astrocytes in a time-and dose-dependent manner with a maximum increase at 24 h of 25 M gemfibrozil treatment ( Fig. 1, A-D). Other lysosomal genes like Cln1 and Cln3, which are responsible for infantile NCL and juvenile NCL, respectively, were also found to increase within 12-24 h (Fig. 1A). The mRNA data were validated by Western blot where a 3-4-fold increase in TPP1 protein level was found with 25 M gemfibrozil and 10 M fenofibrate treatment for 24 h (Fig. 1, E and F). Immunofluorescence of primary mouse astrocytes stimulated with gemfibrozil and fenofibrate also revealed a dose-dependent increase in TPP1 protein (Fig. 1G).
Gemfibrozil and Fenofibrate Up-regulates TPP1 in Neurons from Different Parts of Mouse Brain-Lack of TPP1 enzyme causes accumulation of lipofuscins in neurons leading to loss of neurons in brain causing the disease to progress (5). Hence, we examined the effect of the fibrate drugs in neurons and determined whether the induction of TPP1 occurs throughout the brain. Mouse primary neurons were isolated from different brain regions, viz. cortex, hippocampus, and striatum, and were cultured and treated with gemfibrozil and fenofibrate. The immunofluorescence showed a significant increase in the mouse neurons from all three brain regions (Fig. 1, H, J, and L). Furthermore, the neurons from those brain regions were treated with gemfibrozil for 24 h followed by Western blotting for TPP1, which showed about 2-3-fold increase in TPP1 protein, as determined by densitometric quantification (Fig. 1, I, K, and M).
TPP1 Proteins Up-regulated by the Fibrate Drugs Are Functionally Active-Because the functional activity of the TPP1 protein is of critical importance in the clinical setting for LINCL (6), activity of the enzyme was measured. The cells were homogenized, and the cell extracts were subjected to TPP1 activity assay. Prior to that, the optimal substrate concentration and optimal amount of extract for the assay was determined by using different concentrations of substrate and sample, respectively (supplemental Fig. 1, A and B). TPP1 activity was measured (as described under "Materials and Methods") in mouse primary neurons and mouse primary astrocytes. The product formation increased with increasing doses of treatment indicating an increase in activity of the protein in the cell extracts ( Fig.  1, N and O). This can be attributed to the increase in the levels of proteins in the cells observed in the earlier experiments. Collectively, these data strongly suggest that fibrate drugs can enhance both the mRNA and protein levels resulting in an increased activity of the protein in the cell.
Fibrate Drugs Up-regulate TPP1 mRNA and Protein in Human Brain Cells-We further examined whether a similar increase in Cln2 mRNA and protein was obtained upon treatment of human cells with gemfibrozil and fenofibrate. Human astrocytes were treated in the same way as the mouse cells, and the mRNA levels were quantified. Again, both RT and qPCR data indicated an increase in Cln2 mRNA levels in human astrocytes in a dose-and time-dependent manner with maximum at a dose of 25 M gemfibrozil (ϳ15-fold) and at 12 h (ϳ10-fold) (Fig. 2, A-D). However, fenofibrate was seen to increase the mRNA levels at a relatively lower dose (10 M) but at same time point (12 h) as that of gemfibrozil (Fig. 2, E and F). Once again, the protein levels were assessed in human astrocytes and SH-SY5Y cell lines by immunofluorescence, and in both the cell types, a considerable increase in the level of TPP1 protein was observed. (Fig. 2, G and H).
PPAR␣ Is Involved in Fibrate Drug-mediated Up-regulation of TPP1-Because it is known that PPARs are activated by fibrate drugs, the role of these receptors in mediating up-regulation of TPP1 protein was examined (15). Astrocytes isolated from PPAR␣ Ϫ/Ϫ and PPAR␤ Ϫ/Ϫ and wild type (WT) mice were treated with gemfibrozil and fenofibrate, and Cln2 mRNA levels were measured. The data from semi-quantitative RT-PCR and qRT-PCR showed that WT and PPAR␤ Ϫ/Ϫ cells showed similar patterns of up-regulation, whereas PPAR␣ Ϫ/Ϫ cells showed little or no effect on the up-regulation of Cln2 mRNA expression upon gemfibrozil treatment (Fig. 3, A and B) and fenofibrate treatment (Fig. 3, C and D). When mouse primary astrocytes were treated with GW9662, a PPAR␥ antagonist, followed by gemfibrozil or fenofibrate treatment, there was increased expression of Cln2 mRNA, even in presence of the antagonist (Fig. 3, G and H).
To confirm the mRNA measurements, the WT and PPAR␣ Ϫ/Ϫ and PPAR␤ Ϫ/Ϫ astrocytes were processed for protein analysis. The cells were treated similarly with gemfibrozil and fenofibrate, and immunofluorescence and Western blotting were performed. The immunoblot and densitometric analysis of the blots showed no significant increase in TPP1 levels in PPAR␣ Ϫ/Ϫ cells but about 4 -5-fold increase in WT and PPAR␤ Ϫ/Ϫ cells (Fig. 3, E and F). The data from the Western blot was confirmed by immunofluorescence where a similar effect of the drugs on WT and KO astrocytes was observed, i.e. little or no increase in TPP1 in PPAR␣ Ϫ/Ϫ cells, compared with WT and PPAR␤ Ϫ/Ϫ cells (Fig. 3, I-K).
Furthermore, the presence of active TPP1 enzyme was also confirmed by measurement of TPP1 activity in WT and KO cell types. The enzymatic activity was drastically increased in WT and PPAR␤ Ϫ/Ϫ cells upon treatment with gemfibrozil or fenofibrate (Fig. 3, L and N), whereas PPAR␣ Ϫ/Ϫ cell extracts showed no significant increase in TPP1 enzymatic activity (Fig.  3M). Collectively, these data indicate that PPAR␣, but neither PPAR␤ nor PPAR␥, is involved in the gemfibrozil-and fenofibrate-mediated up-regulation of TPP1.
TPP1 Is Up-regulated by Fibrate Drugs in Vivo in the CNS of WT and PPAR␤ Ϫ/Ϫ , but Not PPAR␣ Ϫ/Ϫ , Mice-Once we confirmed the involvement of PPAR␣ in the fibrate-mediated upregulation of TPP1 protein, we further checked whether the same results could be replicated in in vivo settings. WT, PPAR␣ Ϫ/Ϫ , and PPAR␤ Ϫ/Ϫ mice from same background were treated orally for 21 days with 7.5 mg/kg body weight/day gemfibrozil dissolved in 0.1% methylcellulose, which was also used as vehicle. At the end of the treatment, the mice were killed, and different regions of their brain, viz. substantia nigra pars compacta, cortex, hippocampus, and dentate gyrus were sectioned, and immunofluorescence was performed for the presence of TPP1. Gemfibrozil treatment markedly increased the level of TPP1 both in GFAP-positive cortical astrocytes (Fig. 4, A1-A4) and NeuN-positive cortical neurons (Fig. 4, B1-B4) in WT and PPAR␤ Ϫ/Ϫ , but not PPAR␣ Ϫ/Ϫ , mice. Similarly, gemfibrozil treatment also increased the level of TPP1 in GFAP-positive astrocytes (Fig. 4, C1-C4) and tyrosine hydroxylase-positive neurons (Fig. 4, D1-D4) in the substantia nigra of WT and PPAR␤ Ϫ/Ϫ , but not PPAR␣ Ϫ/Ϫ , mice. Gemfibrozil also increased TPP1 mostly in the non-neuronal cells in the dentate gyrus (Fig. 5, A-D) and CA1 region of the hippocampus (Fig. 5, E-H) of WT and PPAR␤ Ϫ/Ϫ , but not PPAR␣ Ϫ/Ϫ , mice. These data clearly indicate that gemfibrozil increases TPP1 in vivo in the CNS via PPAR␣.  Up-regulation of TPP1 by Fibrate Drugs Involves Both PPAR␣ and RXR␣-Next, we investigated the mechanism of this upregulation. We observed that Cln2 gene promoter lacked the PPAR-binding site but contained an RXR-binding site instead. Because of the facts that RXR␣ is abundant in the brain and astrocytes (32,33) and that PPAR␣ and RXR␣ form a het-erodimer, we thought the mechanism of up-regulation of TPP1 may involve cooperative action of both PPAR␣ and RXR␣ and not PPAR␣ alone. To verify our hypothesis, we performed an array of experiments. First, we checked whether activating RXR by RA, a known activator of RXRs, caused any change in the mRNA or protein levels of TPP1. Interestingly, quantitative real

Up-regulation of Cln2 by Fibrate Drugs
time PCR data showed that even RA alone enhanced the mRNA levels of Cln2 (Fig. 6A). RA at a concentration of 0.5 M caused about a 3.5-fold increase in Cln2 mRNA levels, which is comparable with the effect of 10 M gemfibrozil treatment (ϳ4fold) (Fig. 6A). Moreover, when cells were treated with a low dose of gemfibrozil (10 M) together with RA (at different concentrations), there was a profound increase in the Cln2 levels with an optimum concentration of the combination being at 10 M gemfibrozil and 0.5 M RA (about 12-fold increase) (Fig.  6A). These mRNA data were validated by Western blot performed in mouse astrocytes using similar treatments (Fig. 6B). The densitometry analysis showed a similar pattern of increase in the protein levels of TPP1 as observed from the mRNA data (Fig. 6C). In both real time PCR and immunoblot experiments, the increase of TPP1 expression with the combinatorial treatment (10 M gemfibrozil and 0.5 M RA) was found to be statistically significant when compared with either gemfibrozil (10 M) or RA (0.5 M) treatment alone. This finding clearly indicates the possible involvement of RXR in the up-regulation of the Cln2 gene. Second, to confirm the involvement of RXR␣, we knocked down RXR␣ in astrocytes by RXR␣ siRNA followed by treatment with gemfibrozil and RA. The siRNA was found to specifically knock down RXR␣ but neither RXR␤ nor RXR␥ (Fig. 6D). The effect of gemfibrozil and RA was found to be abrogated in the absence of RXR␣, as observed from the quantitative real time PCR data (Fig. 6E). There was an almost 8 -10fold increase in the Cln2 level in both untransfected cells as well as in cells transfected with scrambled siRNA, whereas cells with RXR␣ knockdown were almost unresponsive to the treatment of gemfibrozil or RA alone as well as the combination (Fig. 6E). Similar results were obtained with the protein analysis. The densitometric analysis for the TPP1 Western blot showed almost no enhancement of TPP1 levels in RXR␣ siRNA-transfected cells (Fig. 6, F and G). Finally, to validate our hypothesis that both PPAR␣ and RXR␣ are involved in the up-regulation process, we checked whether activation of RXR␣ alone (in the absence of PPAR␣) can induce the expression of Cln2. Mouse astrocytes from wild type (WT), PPAR␣ Ϫ/Ϫ , and PPAR␤ Ϫ/Ϫ mice were treated with RA (0.5 M) and the combination of gemfibrozil (10 M) and RA (0.5 M) followed by immunoblot analysis for TPP1. It was observed that neither RA alone nor the combination could induce TPP1 in PPAR␣ Ϫ/Ϫ cells, whereas WT and PPAR␤ Ϫ/Ϫ cells were responsive to the treatment (about 5-6fold induction of TPP1) (Fig. 6, H and I). These data suggest that either PPAR␣ or RXR␣ alone is not sufficient for the up-regulation of TPP1.
Fibrate Drugs Up-regulate TPP1 via Activation of PPAR␣/ RXR␣ Heterodimer-After confirming the involvement of both PPAR␣ and RXR␣, we were interested to find out the actual role of the two factors. First, we examined whether there was any actual physical interaction between PPAR␣ and RXR␣. Mouse astrocytes were treated with gemfibrozil and RA separately as well as in combination, and the nuclear extract was subjected to co-immunoprecipitation for both PPAR␣ and RXR. Immunoprecipitation with PPAR␣ Ab showed increased presence of RXR in the immunoblot for the treated samples compared with control (Fig. 7A (i)). Similarly, increased abundance of PPAR␣ was also observed when the nuclear extracts were immunoprecipitated with RXR Ab (Fig. 7A (ii)). These results demonstrate the presence of the PPAR␣/RXR␣ heterodimer in the nucleus of cells stimulated with gemfibrozil and RA. These results are specific as we did not find any bands with IgG ( Fig. 7A (iii)). Levels of PPAR␣ and RXR␣ and histone 3 (H3) have been shown as loading controls (Fig. 7A (iv)). Next, we performed ChIP studies to show the recruitment of the PPAR␣ and RXR␣ on the RXR-binding site on the Cln2 gene (Fig. 7B). Chromatin fragments from cells treated with gemfibrozil and RA were immunoprecipitated with both PPAR␣ Ab and RXR␣ Ab, and the DNA obtained was amplified by PCR with primers spanning the RXR-binding site on the Cln2 gene promoter. In both cases, we were able to amplify 200-bp fragments flanking the RXRbinding site (Fig. 7C). In contrast, no amplification product was observed in any of the immunoprecipitates obtained with control IgG (Fig. 7C), suggesting the specificity of these interactions. These results suggest that gemfibrozil and RA are capable of recruiting both PPAR␣ and RXR␣ to the RXR-binding site of the Cln2 gene promoter (Fig. 7C).

DISCUSSION
The NCL family of disease can be considered to be one of the most important hereditary neurodegenerative lysosomal storage diseases in children (34). Mutations in the Cln2 gene result in deficiency or loss of function of the TPP1 enzyme (9,30,35). There have been reports of over 68 missense mutations in the Cln2 gene, including 35 single amino acid substitutions. Studies with 14 different naturally occurring disease-associated mutations showed alteration of lysosomal transport, increased halflife of the proenzyme, and improper folding, resulting to loss of function of the enzyme (9). Currently, there is no established drug-mediated therapy for LINCL, a classic subtype of the NCLs. Studies using adeno-associated virus and other viral vectors expressing recombinant TPP1 demonstrate widespread  NOVEMBER 9, 2012 • VOLUME 287 • NUMBER 46 expression of TPP1, and treatment of Cln2-targeted mice with these recombinant vectors show slowing of disease-associated pathology and an increase in survival in mutant mice (36 -38). However, levels of TPPI activity achievable by adeno-associated virus-mediated gene therapy can vary and depend on various critical parameters, and there is considerable doubt whether similar effects can be achieved in humans (36,39).

Up-regulation of Cln2 by Fibrate Drugs
Nevertheless, restoration of activity at even low levels could prove helpful for most lysosomal storage diseases, where resto-ration of even Ͻ10% of normal activity may have therapeutic benefits (40). Studies with hypomorphs of Cln2 mutant mice, expressing different levels of TPP1 enzyme, indicate that even 3% of normal TPP1 activity is capable of delaying the onset of the disease, and 6% of the normal activity attenuates the disease and increases the life span of mice (40). Also, two specific variants of mutated TPP1 were responsive to molecular chaperone treatment, indicating that folding improvement strategies can be used to restore the enzymatic activity (9). Recent studies suggest that some misfolded variants or misprocessed proteins may also be rescued by treatment in permissive temperatures under suitable conditions (9). There have also been reports of some mutations in TPP1 (R447H), which apparently may not have any pathogenic effect (31). Moreover, a sensitive enzyme activity assay detected residual levels of TPP1 activity in various biological samples from patients who were confirmed to have LINCL by genetic analysis (30). This study also showed the presence of enzyme activity in various animals having NCL-like neurodegenerative symptoms rendering them unsuitable for being a model for classical LINCL (30). Furthermore, using a highly sensitive capillary electrophoresis technique, Viglio et al. (8) reported that lymphocytes from patients affected with LINCL exhibited TPP1 activity, although at low levels (in a range between 0.1 and 0.8 milliunits/mg). These findings about the presence of residual enzymatic activity in LINCL patients are very interesting as they indicate the presence of at least a few copies of the functional gene in the system. Therefore, identifying specific drugs and understanding the mechanisms by which these drugs can up-regulate the endogenous normal copies of the gene may be a critical step for LINCL therapy.
Gemfibrozil, marketed as "Lopid," and fenofibrate, known as "Tricor," are FDA-approved drugs prescribed for hyperlipidemia (10,12). Here, we delineate for the first time that these drugs are capable of up-regulating TPP1 in brain cells. This finding was confirmed by both mRNA and protein studies in both mouse and human cells. The increase in protein levels was throughout the brain as neurons isolated from different brain regions of mice showed increased TPP1 expression upon treatment with gemfibrozil. In the case of LINCL, the presence of the functionally active TPP1 enzyme is critical for therapy, as we have to rely on the up-regulation of residual enzyme activity in patients. The TPP1 activity assay, performed in different cell types, clearly showed that there was significant increase in the activity of the enzyme, which is a result of increased levels of the protein. Considering the possibility of treatment by up-regulation of the endogenous Cln2 gene, this finding could be of importance in the therapy of LINCL.
Over the last few years, a number of studies emphasized the role of PPARs in different regulatory and modulatory pathways. It is also well known that PPAR␣ is activated by polyunsaturated fatty acids and oxidized derivatives and by lipid-modify-  NOVEMBER 9, 2012 • VOLUME 287 • NUMBER 46

JOURNAL OF BIOLOGICAL CHEMISTRY 38931
ing drugs of the fibrate family, including fenofibrate and gemfibrozil (41,42). PPAR␣ is present in the cytoplasm as an inactive complex with heat-shock protein 90 (HSP-90) and hepatitis virus B-X-associated protein-2 (XAP-2), which act as an inhibitor of PPAR␣. Fibrate drugs replace the HSP90 repressor complex and help to rescue the transcriptional activity of PPAR␣ (15). Therefore, we investigated the role of the PPAR group of receptors in this phenomenon. We examined all three PPARs, viz. PPAR␣, PPAR␤, and PPAR␥, for their involvement in up-regulation of TPP1. These studies clearly indicate the involvement of PPAR␣, but not PPAR␤ and PPAR␥, in this process. In astrocytes from WT and PPAR␣ Ϫ/Ϫ and PPAR␤ Ϫ/Ϫ mice, both the TPP1 mRNA and protein analysis showed the involvement of only PPAR␣. Involvement of PPAR␥ was ruled out as studies using known antagonist of PPAR␥ revealed no effect. TPP1 enzyme activity in the cell extracts was also increased in WT and PPAR␤ Ϫ/Ϫ , but not PPAR␣ Ϫ/Ϫ , cells. The in vitro studies were further validated by in vivo studies, where we used the knock-out mice for PPAR␣ and PPAR␤. Our in vivo results also supported the cell culture data.
To delineate the mechanism of fibrate drug-mediated upregulation of TPP1, we analyzed the promoter region of the Cln2 gene. Surprisingly, no PPAR-binding site was found in the mouse Cln2 promoter, but further analysis of the promoter revealed an RXR-binding site. It is well known that to bind to DNA and activate transcription, PPAR requires the formation of s heterodimer with the RXR (43). Together, the PPAR/RXR heterodimer regulates the transcription of genes for which products are involved in lipid homeostasis, cell growth, and differentiation (44,45). This led us to think whether the pathway of TPP1 up-regulation requires a cooperative effect of both PPAR and RXR. It was observed that the activation of RXR by low doses of RA alone (0.5 M) was capable of up-regulating TPP1 to a comparable level of that of gemfibrozil (10 M). Also, when cells were treated with both gemfibrozil and RA together, they cooperatively enhanced the expression of TPP1 by almost more than 3-fold compared with the levels achieved by either gemfibrozil or RA alone, which implies that a combinatorial therapy could be more useful than using the compounds separately for treatment. Furthermore, the effects of both gemfibrozil and RA were abrogated in the absence of either RXR␣ or PPAR␣. The co-immunoprecipitation studies performed with the nuclear extracts on astrocytes stimulated with gemfibrozil and RA demonstrated physical interaction between PPAR␣ and RXR. These data clearly suggest that the treatment with gemfibrozil and RA activates both PPAR␣ and RXR␣, which forms a heterodimer in the nucleus. The ChIP data indicated the recruitment of the PPAR␣ and RXR␣ on the RXR-binding site of the Cln2 promoter, hence validating our hypothesis. Collectively, these data outline a unique mechanism where gemfibrozil, a known activator of PPAR␣, and RA, an agonist of RXR␣, together can up-regulate TPP1 in brain cells via the PPAR␣/ RXR␣ heterodimer Fig. 8.
Gemfibrozil and other fibrate drugs are known to reduce superoxide, lipid peroxidation products. It also strengthens the cellular defense by stimulating the activity of anti-oxidant proteins such as paraoxonase and is associated with the free radical scavenging ability as well as metal ion chelation. Therefore, apart from its lipid-lowering effects, these drugs also have antiinflammatory, immunomodulatory, and anti-oxidative properties (14, 27, 46 -48). In the NCL cases, predominantly in LINCL, different brain regions have been shown to be immunoreactive for 4-hydroxynonenal or 8-hydroxydeoxyguanosine, popular markers for evaluation of oxidative stress that may be caused due to accumulation of lipofuscins and elevated cytokine response (1). Therefore, treatment with these drugs will not only lead to the up-regulation of endogenous normal TPP1 leading to clearance of lipofuscins but also can be beneficial for the peripheral immune system by down-regulating the inflammatory pathways generated due to accumulation of lipofuscins.
In summary, we have delineated that gemfibrozil and fenofibrate, FDA-approved lipid-lowering drugs, up-regulate TPP1 in cultured mouse and human brain cells and in vivo in mouse brain via the PPAR␣/RXR␣ pathway. Although the in vitro situation of mouse and human brain cells in culture and its treatment with gemfibrozil and fenofibrate may not truly resemble the in vivo situation of the CNS of patients with LINCLs, our results clearly identify these two drugs as possible therapeutic FIGURE 7. Fibrate drugs up-regulate TPP1 via activation of PPAR␣/RXR␣ heterodimer. A, mouse primary astrocytes were treated with gemfibrozil (10 M) and RA (0. 5 M) alone and in combination in serum-free DMEM/F-12 for 6 h, and the nuclear extract was subjected to the following: (i) immunoprecipitation by PPAR␣ Ab followed by immunoblot for both RXR␣ and PPAR␣; (ii) immunoprecipitation by RXR␣ Ab followed by immunoblot for both PPAR␣ and RXR␣; (iii) immunoprecipitation by control IgG followed by immunoblot for both PPAR␣ and RXR␣, and (iv) nuclear extract was subjected to immunoblot for PPAR␣, RXR␣, and histone 3 (H3). B, schematic diagram for RXR-binding site on the Cln2 promoter with the core sequence and amplicon length. C, mouse astrocytes were treated with the combination of gemfibrozil (10 M) and RA (0.5 M) for 6 h, and recruitment of PPAR␣ and RXR␣ on the RXR-binding site of Cln2 promoter was monitored by ChIP analysis as described under "Materials and Methods." Normal IgG was used as control. All results are representative of at least three independent experiments.