Infantile neuronal ceroid lipofuscinosis (INCL) is a severe neurodegenerative disease caused by deficiency of palmitoyl protein thioesterase 1 (PPT1). INCL results in dramatic loss of thalamocortical neurons, but the disease mechanism has remained elusive. In the present work we describe the first interaction partner of PPT1, the F1-complex of the mitochondrial ATP synthase, by co-purification and in vitro-binding assays. In addition to mitochondria, subunits of F1-complex have been reported to localize in the plasma membrane, and to be capable of acting as receptors for various ligands such as apolipoprotein A-1. We verified here the plasma membrane localization of F1-subunits on mouse primary neurons and fibroblasts by cell surface biotinylation and TIRF-microscopy. To gain further insight into the Ppt1-mediated properties of the F1-complex, we utilized the Ppt1-deficient Ppt1Δex4 mice. While no changes in the mitochondrial function could be detected in the brain of the Ppt1Δex4 mice, the levels of F1-subunits α and β on the plasma membrane were specifically increased in the Ppt1Δex4 neurons. Significant changes were also detected in the apolipoprotein A-I uptake by the Ppt1Δex4 neurons and the serum lipid composition in the Ppt1Δex4 mice. These data indicate neuron-specific changes for F1-complex in the Ppt1-deficient cells and give clues for a possible link between lipid metabolism and neurodegeneration in INCL.
Neuronal ceroid lipofuscinoses (NCLs) constitute the most common group of recessively inherited encephalopathies in children (reviewed in 1). Up to eight different genes have been identified underlying NCLs (2,3), and the disorders share similar clinical and pathological findings including mental and motor deterioration, visual failure, epilepsy, reduced lifespan and autofluorescent lipopigment storage in patients' tissues (4). In most NCL diseases the storage material consists mainly of subunit C of ATP synthase, while in the infantile and congenital NCL the main storage material is sphingolipid activator proteins (SAPs) A and D (5,6). The storage material accumulates in most tissue types, but the hallmark of NCLs is the selective and devastating loss of neurons in the central nervous system.
Palmitoyl protein thioesterase 1, PPT1, enzyme deficient in the infantile form of NCL (INCL) (7), has the structure of an α/β serine hydrolase with three N-linked glycosylation sites and a fatty acid binding groove (8). PPT1 removes palmitic acid from S-acylated proteins in vitro (9), but the exact physiological function and the in vivo substrates have remained unknown. In most cell types PPT1 localizes to lysosomes (10), but in neurons it is found also in the presynaptic areas (11–13). The expression of PPT1 in human and mouse brain is developmentally regulated and coincides with cortical synaptogenesis (14,15). PPT1 has been suggested to be involved in various cellular processes, including apoptosis, endocytosis and synaptic function (16,17). Analyses of PPT1/Ppt1 deficient human and mouse brain have revealed local neuroinflammation and changes in lipid profiles (reviewed in 18,19).
PPT1-associated metabolic pathways in neurons have remained unclear, and the interaction partners of PPT1 have been unknown. Here we report an interaction between PPT1 and the F1-ATP synthase. We studied this interaction and its consequences utilizing Ppt1-deficient Ppt1Δex4 mice. These mice mimic the INCL disease with the typical tissue pathology and early onset neurodegeneration, leading to premature death (20). We show here that Ppt1-deficiency results in alterations in the amount of F1-ATP synthase on the plasma membrane, but not in mitochondria, and that Ppt1-deficiency is also manifested at the level of serum lipids/lipoproteins.
PPT1 interacts with the F1-complex of mitochondrial ATP synthase
Lysosomal enzymes are secreted upon overexpression, and this phenomenon was utilized to purify PPT1 from the conditioned media of CHO cell line stably expressing PPT1 (8,13) by hydrophobic interaction chromatography and size-exclusion chromatography. During purification, a 50 kDa protein was repeatedly co-purified with the recombinant PPT1. MALDI-TOF mass spectrometric peptide mass fingerprint analysis of the 50 kDa protein band identified this protein as the β-subunit of mitochondrial F1-ATP synthase (Supplementary Material), responsible for the final step of oxidative ATP production in mitochondria (21). The potential interaction between PPT1 and β-subunit was investigated by surface plasmon resonance assay (Biacore) using purified PPT1 immobilized on a sensor chip. To test the chip, polyclonal PPT1 antibody was added in the fluid phase and it was found to bind the immobilized PPT1 (data not shown). We then tested the binding of purified β-subunit and the whole F1-complex consisting of five different protein subunits (α3β3γΔϵ) to PPT1. PPT1 showed interaction with the purified F1-complex, but not with the β-subunit alone (Fig. 1A). When the surface plasmon resonance assay was repeated with increasing concentrations of F1-complex, the results confirmed a concentration-dependent interaction between PPT1 and the F1-complex (Fig. 1B and C).
To verify this novel interaction, GST pull-down was utilized. Recombinant full-length PPT1 lacking the signal sequence was expressed as GST-fusion protein, solubilized, purified and then incubated with mouse brain extract. The precipitated proteins were analyzed by western blot with monoclonal antibodies against subunits α and β from the F1-complex. GST-PPT1 bound both of these subunits from mouse brain extract, and neither of them interacted with GST alone (Fig. 2). The interaction between PPT1 and F1-complex but not between PPT1 and β-subunit alone in Biacore-assay allows two possible interpretations of the result: (i) The interaction is dependent on the conformation of the β-subunit and possible only when it is in the F1 -bound form, or (ii) PPT1 interacts with some other subunit in the F1-complex.
Ppt1Δex4 mice brain do not show alterations in mitochondrial function or assembly of mitochondrial respiratory complexes
To determine the possible effect of the Ppt1-deficiency on the assembly or steady-state levels of respiratory complexes, and especially that of the mitochondrial F1-Fo-ATPase, mitochondria were isolated from the cerebrum and cerebellum of 2-month-old control and Ppt1Δex4 animals, and analyzed by blue native (BN) electrophoresis followed by western blot (Fig. 3A) or Coomassie staining (data not shown). The relative amounts of the ATP synthase were compared with complex II, and no reduction in the amount or abnormality in the assembly of the mitochondrial ATP synthase was present in these mice (Fig. 3A). In order to test the mitochondrial function in the Ppt1Δex4 mice brain, the mitochondria were isolated and the oxygen consumption was measured at baseline and in response to the following components or inhibitors of the mitochondrial energy production: (i) pyruvate and malate, (ii) succinate and rotenone, (iii) ascorbate and TMPD, (iv) palmitoyl CoA and (v) α-ketoglutarate. No difference in the oxygen consumption between wild-type and Ppt1Δex4 mice was detected (Fig. 3B).
The α- and β-subunits of the F1-complex localize to the cell surface in neurons and fibroblasts
To explore the tissue distribution of the F1-subunits in mouse brain, immunohistochemical staining patterns of the α- and β-subunits were analyzed. They were similar to that of Ppt1 in the wild-type mouse brain, especially the large pyramidal neurons of cortical layers II–IV were positive for both subunits and Ppt1 (Fig. 4A–C). Ppt1 staining was absent in the Ppt1Δex4 mice, indicating a complete knock-out of the gene (Fig. 4D). Thorough investigation of the brain sections of the Ppt1Δex4 mice demonstrated a slightly more intense α- and β-subunit staining in the large neurons of cortical layers II–IV of the mutant mice compared with wild-type controls (Fig. 4E and F).
Several studies have reported ATP-synthase subunits to localize in lipid rafts on the plasma membrane in addition to mitochondria (reviewed in 22). No phenotype in the mitochondrial ATP synthase was identified in the Ppt1 mutant mice, and thus we examined the putative plasma membrane localization of the α- and β-subunits in the wild-type mouse primary neurons and fibroblasts. In neurons, both the endogenous α- and β-subunit demonstrated a strong signal in the biotinylated protein specimens, indicating their cell surface localization and supporting an extra-mitochondrial role for these subunits in mouse neurons as well (Supplementary Material). To analyze the pattern of these proteins in the plasma membrane, total internal reflection fluorescence (TIRF) microscopy analysis was utilized. TIRF revealed a distinct punctuated staining pattern for both α- and β-subunits on the surface of neurons and fibroblasts. In both cell types, the staining was most intense at the cell periphery and in the neuronal growth cones, while the cell soma showed more scattered staining (Fig. 5).
The amount of the ectopic α- and β-subunits at the plasma membrane is increased in Ppt1Δex4 neurons
We examined the possible role of Ppt1 in the localization of F1-subunits by comparing their concentration at the plasma membrane in murine wild-type and Ppt1Δex4 neurons using TIRF-correlation analysis. Western blot analysis of whole cell lysates from the cultured wild-type and Ppt1-deficient neurons did not show alteration in the total amounts of subunits (data not shown), the majority of that signal coming from mitochondrially located proteins. However, we detected a neuron-specific increase in the amount of both α- and β-subunit on the plasma membrane of the Ppt1Δex4 neurons when compared with wild-type (Fig. 6A), whereas no difference was detected in the fibroblasts (Fig. 6B). As PPT1 is a depalmitoylating enzyme, we next investigated whether the detected changes in the surface location of the F1-subunits are dependent on palmitoylation. We inhibited protein palmitoylation by palmitate analog 2-bromopalmitate (2BP) (23) in wild-type mouse primary neurons and performed the TIRF-correlation analysis. As a control for the 2BP treatment, the location of a palmitoylated neuronal plasma membrane protein PSD-95 was analyzed. The clustering of PSD-95 was lost after blocking palmitoylation with 2BP in accordance with earlier studies (24) (data not shown). However, no clear difference was detected in the amount of F1-subunits on the plasma membrane after 2BP treatment as compared with the untreated wild-type neurons (Fig. 6A). The results in mouse fibroblasts were similar, except for a small decrease in the amount of β-subunit upon 2BP treatment (Fig. 6B).
Apolipoprotein A-I internalization is increased in Ppt1Δex4 neurons and glia
Our recent transcript profiling studies with neuron cultures have demonstrated severe disturbances in lipid metabolic pathways, specifically in cholesterol metabolism, in the Ppt1-deficient neurons (25). The β-subunit of F1-ATP synthase has been linked to lipoprotein metabolism and suggested to act as an apoA-I –sensing receptor in hepatocytes (26), in which its stimulation activates a P2Y13-mediated HDL endocytosis pathway (27). Although the function of the plasma membrane F1-complex is not known in neuronal cells, we examined whether the β-subunit operates as an apoA-I-sensing receptor also in neurons and glial cells. The cells were incubated with an excess of radiolabeled lipid-free apoA-I. In TCA precipitation, the degree of apoA-I degradation was similar between the Ppt1Δex4 and wild-type cells or culture media (data not shown). A significant increase in the levels of internalized apoA-I was detected both in Ppt1-deficient glial cells (118 versus 100%) and neurons (140 versus 100%), suggesting that the elevated amount of the β-subunit on the plasma membrane could mediate enhanced apoA-I uptake (Fig. 7).
Altered serum lipid and lipoprotein profiles in Ppt1Δex4 mice
ApoA-I is the major apolipoprotein in HDL, and therefore it was of interest to investigate possible changes in lipid metabolism in the Ppt1Δex4 mice. In serum samples from 1-month-old mutant and control mice, the levels of apoA-I, apoE, cholesterol and triglycerides as well as the activity of plasma phospholipid transfer protein (PLTP) were measured. In the mutant animals, the cholesterol and apoA-I concentrations as well as the PLTP activity were significantly reduced when compared with the control mice (Fig. 8A). The observed perturbations are logical since decreased HDL cholesterol levels are associated with reduced apoA-I. The decrease in PLTP activity followed the decreased triglyceride levels. PLTP is a factor suggested to maintain serum HDL levels (28). Therefore the observed change in PLTP activity is consistent with the drop of cholesterol and apoA-I concentrations. No significant difference could be detected in serum apoE levels in the Ppt1Δex4 and wild-type mice, but the distribution of apoE between different sized lipoprotein particles was dramatically changed in the Ppt1Δex4 mice. Size-exclusion chromatography analysis demonstrated that high molecular weight apoE-containing lipoprotein particles were almost totally absent from the Ppt1Δex4 mice sera. In contrast, particles of smaller size appeared to be more abundant (Fig. 8B). This observation may indicate functional changes in the reverse cholesterol transport pathway, as the larger apoE-containing HDL particles are more efficient in cholesterol removal from macrophage-foam cells when compared with smaller HDL particles (29). Taken together, these results indicate a change in the systemic lipid homeostasis of Ppt1 deficient mice, and emphasize a novel role for Ppt1/PPT1 in the regulation of lipoprotein metabolism.
INCL and mitochondria
In this study, we have characterized the first interaction partner for PPT1, the F1-complex of the mitochondrial ATP synthase. This finding was intriguing, since ATP synthase has been previously linked with various NCL diseases: a hallmark of disease pathology has been the accumulation of storage material containing the Fo-subunit C of the mitochondrial ATP synthase, and functional and structural abnormalities of mitochondria have been reported in NCL patients and animal models (30–33). INCL patient fibroblasts are known to have reduced basal ATP synthase activity and a defect in the regulation of the enzyme (34). Furthermore, the myocytic and neuronal mitochondria of ppt-1-deficient Caenorhabditis elegans show abnormal morphology, including fewer cristae and whorling of the inner membrane (30). However, previous ultrastructural analysis of the brain of Ppt1Δex4 mice did not indicate any abnormalities in the mitochondrial shape or size, although the storage material of typical GROD ultrastructure was present in the mutant animals (20 and unpublished data).
We studied here the nature of the detected interaction and its importance in INCL pathogenesis, and first explored the mitochondrial function in Ppt1-deficient mouse brain. No significant alterations could be detected in the steady-state levels or the assembly of respiratory complexes in mitochondria, speaking for normal transportation and structure of nuclear-encoded subunits in the organelle. Neither was the mitochondrial respiratory function affected in Ppt1-deficient mouse brain, indicated by the unaltered enzyme activities of the respiratory chain complexes I–IV (data not shown) and mitochondrial oxygen consumption. In the light of the current data, it seems that there are no significant mitochondrial consequences of Ppt1-deficiency in the mouse brain.
F1-ATP synthase on plasma membrane of Ppt1-deficient neurons
Components of ATP synthase have been found to localize on the plasma membrane of many different cell types including hepatocytes, endothelial cells and adipocytes, and they may act as receptors for multiple ligands and participate in processes such as regulation of lipid metabolism and cellular proliferation (26,35,36). A recent proteomic study showed that ATP synthase was present in lipid rafts derived from plasma membrane of different tissues, including the brain (37). We have demonstrated here that the α- and β-subunits are present on the plasma membrane of wild-type mouse neurons and fibroblasts. While Ppt1-deficiency did not seem to cause major alterations in the F1-complex in the mitochondria, we did detect specific and significant changes in the amount of F1-subunits in neuronal plasma membrane. This suggests a role for Ppt1 in regulation of the subcellular distribution of F1-subunits in neurons.
We hypothesized that Ppt1 could directly regulate and reduce the amount of plasma membrane F1-subunits via its depalmitoylating function. However, blocking protein palmitoylation with 2-bromopalmitate did not have any effect on the amount of ectopic subunits in our experimental conditions. Therefore, other processes than palmitoylation have a more prominent role in the regulation of the ectopic localization of this receptor, or the regulation mechanism for palmitoylation differs from that of our positive control PSD-95 (24). It is also possible that PPT1 exhibits alternative functional properties in neurons, as we have recently observed that PPT1 is differently modified in neurons when compared with fibroblasts (38). Furthermore, we showed that in addition to the monomeric form, active PPT1 is present as a high molecular weight complex in PC12 cells (38). Similar findings describing different forms of bovine brain-derived PPT1 have been reported by Camp and Hofmann (9). These data suggest that PPT1 harbors neuron-specific modifications, proposed to affect its intracellular localization and function. These modifications could also have an effect on the molecular interactions of PPT1.
What could be the site for PPT1 and F1-complex interaction? Immunofluoresence analysis of PPT1 and β-subunit in PPT1-overexpressing neurons and fibroblasts did not reveal notable co-localization of the two proteins in either cell type (data not shown). However, we have shown utilizing antibody internalization techniques that the overexpressed PPT1 traffics via plasma membrane in neurons and fibroblasts (38). Furthermore, PPT1 has been shown to localize in lipid rafts and to regulate their lipid content in CHO cells (39). Here, the interaction partner was also co-secreted with overexpressed PPT1 during the purification process, supporting their interaction to occur somewhere along the secretory pathway.
ApoA-I and the altered neuronal cholesterol metabolism in Ppt1-deficient cells
We have previously shown by transcript profiling studies of neuronal cultures that Ppt1-deficient neurons have severe disturbances in cholesterol metabolism pathways (25). Therefore, it was of interest to investigate whether the increased receptor amount had a role in the same pathway. Indeed, we demonstrated that the cellular uptake of the lipid-free apoA-I was increased. In addition to being one of the ligands for the ectopic β-subunit, apoA-I is also a major apolipoprotein in the lipoproteins of cerebrospinal fluid (CSF) (26,40). Outside the CNS, apoA-I promotes atheroprotective functions, including facilitation of cholesterol removal from macrophage-foam cells in a process of reverse cholesterol transport that ultimately leads to cholesterol removal by the liver (reviewed in 41). In the brain, its function seems to be similar, as apoA-I together with the activation of Liver X receptor (LXR) was reported to induce the cholesterol efflux from both neurons and astrocytes (42).
So far, there is no evidence for apoA-I synthesis in the brain, but brain endothelial cells have been shown to synthesize and secrete apoA-I and abundant amounts of apoA-I present in the CNS are transported across the blood brain barrier (43). In vitro studies have suggested that vascular endothelial cells also transcytose apoA-I via the SR-BI or ABCA1 receptors in the brain or in the aorta, respectively (44,45). Interestingly, another NCL-protein cathepsin D was recently found to regulate and promote ABCA1 and apoA-I-mediated cholesterol efflux from macrophages, and cathepsin D inactivation resulted in a lysosomal accumulation of cholesterol (46). Previous data also suggest that when vascular endothelial cells are exposed to cholesterol, the amount of β-subunit in the plasma membrane increases (47). Our recent study demonstrated that while cholesterol biosynthesis was strongly upregulated in the embryonic Ppt1-deficient neurons, no cholesterol accumulation could be detected (25). In future studies it will be of great interest to investigate the relationship of the increased amount of F1-subunits and the increased uptake of apoA-I with the detected alteration of cholesterol biosynthesis in the Ppt1-deficient neurons.
The cross-talk between astrocytes and neurons
ApoE is the major apolipoprotein expressed in the CNS, and the apoE4 protein, encoded by the ϵ4 allele, has been genetically linked to neurodegenerative Alzheimer's disease (48). During their maturation and synaptogenesis, neurons are dependent on the cholesterol provided by astrocytes (reviewed in 49). On the other hand, neurons convert cholesterol to 24(S)-hydroxycholesterol, which induces ABCA1-regulated, apoE-mediated cholesterol efflux specifically from astrocytes (42). Astrocytic activation has been shown to be important for the pathogenesis of INCL (reviewed in 1,18). Although further studies are needed to clarify the neuron-astrocyte interplay during the disease process, it is possible that apoE contributes to INCL pathogenesis even more than apoA-I.
In this study, we could not observe major alterations in apoA-I or apoE immunostaining in the brain of Ppt1Δex4 mice (data not shown) nor in apoE serum levels. However, the reduced amount of large apoE-containing HDL-particles in the sera of Ppt1Δex4 mice suggests a disturbance of lipoprotein homeostasis. This is an interesting observation also in the light of previous studies describing apoE binding to both α- and β-subunits of F1-complex (50,51). Besides decreased levels of cholesterol, apoA-I and abnormal distribution of apoE-containing particles, we also detected decreased PLTP activity in Ppt1Δex4 mouse sera. Significantly, we have recently demonstrated that active PLTP remodels HDL particles to apoE-enriched HDL with larger particle size (52). Moreover, PLTP activity in the CSF was reportedly reduced in various neurological diseases (53). Finally, a direct functional interaction between PLTP and apoE was shown whereby PLTP induced apoE secretion from primary human astrocytes (53).
In this study we have described a novel interaction between PPT1 and F1-ATP synthase. Although mitochondrial abnormalities have been reported in other NCL models, we could not detect any substantial defects in the amount, assembly or function of mitochondrial respiratory complexes in Ppt1-deficient mouse brains. The most striking finding was the increased amount of F1-complex in the plasma membrane of Ppt1-deficient neurons and their increased uptake of apoA-1. An intriguing previous link suggests a role for plasma membrane F1-complex as a receptor for the lipoprotein. Undisturbed lipid homeostasis is vital for the development and maintenance of the normal CNS function (49). Brain cholesterol homeostasis is still poorly understood, but an increasing number of studies link it with neurodegenerative diseases, including the NCL disorders. These monogenic diseases could provide powerful tools for future studies on regulatory pathways that connect neuronal homeostasis and systemic lipid metabolism.
MATERIALS AND METHODS
Expression plasmid for GST-PPT128-306 fusion protein was prepared as follows: the PPT1 coding sequence without the signal sequence was multiplied with PCR using primers 5′CAGGAATTCGACCCGCCGGCGCCGCTGCC and 3′GTCCTCGAGTCCAAGGAATGGTATGATGTG. The resulting PCR fragments were digested with EcoRI and XhoI and ligated to the corresponding sites in pGEX-4T-1 vector. The reading frame was confirmed with ABI3730 Automatic DNA Sequencer using the BigDye™ Terminator Cycle Sequencing Kit v3.1 (Applied Biosystems).
PPT1 purification and protein identification
Recombinant PPT1 was purified according to previous protocol (38). To identify the protein associated with PPT1 during purification, the corresponding protein band was cut out from silver-stained SDS-PAGE gel, reduced with 20 mm DTT, alkylated with 55 mM iodoacetamide and digested with trypsin overnight at +37°C. The resulting peptides were eluted from the gel, desalted with a µC18 ZipTip reversed phase tip (Millipore) and subjected to MALDI-TOF mass spectrometric (Ultraflex TOF/TOF, Bruker Daltonics) peptide mass fingerprint analysis. Protein identification was performed using the MASCOT Peptide Mass Fingerprint program (http://www.matrixscience.com).
Surface plasmon resonance analysis
Binding of PPT1 and F1-complex or β-subunit of ATP synthase was studied with surface plasmon resonance analysis in Biacore 2000 biosensor (Biacore). Saturating amount of purified PPT1 was covalently attached to a flow cell of a biosensor chip using standard amine coupling according to the manufacturer's protocol. Recombinant full-length human β-ATP synthase [LGC Promochem, France, ATCC number 59118 (54)], subcloned in pPROex™HTb (Invitrogen), was transformed into E. coli BL21 for protein expression. After solubilization, the proteins were separated on 3 mm-thickness 10% (w/v) Glycine-SDS–PAGE. One part of the gel was stained by Coomassie brilliant blue R250 and a non-stained band corresponding to the ∼50kDa β-ATP synthase protein was cut off and recovered from the gel by microelectroelution as previously described (55). The eluted protein fraction was then dialyzed in water, quantified and lyophilized. Purity of the β-ATP synthase preparation was confirmed by Coomassie staining and western blot detection with β-subunit monoclonal antibody (Molecular Probes). F1-ATPase sector of the ATP synthase complex was purified from bovine heart mitochondria as previously described (56) and kindly provided by J. Walker (MRC Dunn Human Nutrition Unit, Cambridge, UK). Varying concentrations of purified human β-subunit and bovine F1-ATPase were diluted in Hepes-buffered saline and used as analytes. An empty flow cell was used as a negative control.
GST pull-down assay
E. coli BL21 cells were transformed with GST-vector and GST-fusion construct GST-PPT128-306. The GST fusion protein expression, purification and immobilization to Glutathione-agarose beads were performed according to manufacturer's instructions (Amersham Biosciences). Mouse brain extracts were obtained by homogenizing the brain from adult C57BL mouse in 3 ml extraction buffer (50 mm Hepes pH 7.4, 100 mM NaCl, 5 mM MgCl2, 0.5% Triton X-100 and protease inhibitor cocktail) and centrifuging for 10 min at 4°C at 1000g followed by ultracentrifugation of the supernatant at 100 000g at 4°C for 40 min. Immobilized GST-fusion proteins were incubated with mouse brain extract (2 mg/ml) in extraction buffer for 3 h at 4°C with gentle rotation. Samples were washed, boiled with Laemmli buffer and 70 mM DTT and analyzed by western blot to detect α- and β-subunits (monoclonal antibodies from Molecular Probes).
Animals and cell culture
Ppt1Δex4 mice (20) were maintained on a congenic C57/BL6J strain background. Animals used in this study were a result of ten generations of backcrosses onto C57/BL6J. C57/BL6J mice were used as wild-type controls. The study was approved by the Laboratory Animal Care and Use Committee of the National Public health Institute, Helsinki. The study has been carried out following good practice in laboratory animal handling and the regulations for handling genetically modified organisms. Mouse cortical neurons were prepared from C57BL WT and Ppt1Δex4 mouse embryos (E14–E16) and maintained as previously described (11). The glial cell cultures were prepared from mouse cubs (P2–P5) according to the protocol of Harkke et al. (57).
Brains of three 1-month-old Ppt1Δex4 knock-out mice and three wild-type controls were immersion fixed (4% PFA) and paraffin embedded. The tissue was cut into 5 µm sections and mounted on a Super Frost (Menzel Gläzer) objective glass. Immunostaining with polyclonal PPT1 (1:200) (38) and monoclonal α- and β-subunit antibodies (1:500, Molecular Probes) was performed using a standard protocol (20).
Examination of mitochondrial function
The mitochondrial function was examined according to the protocol of Luiro et al. (32). In brief, two Ppt1Δex4 mice and wild-type controls were sacrificed and the brains were homogenized in SETH buffer (0.25 M sucrose, 2 mm EDTA, 10 mM Tris–HCl, 50 U/ml heparin, pH 7.4) and centrifuged twice at 400 g at 4°C for 10 min. The postnuclear supernatant was further centrifuged twice at 9300 g at 4°C for 10 min to pellet the mitochondria. The pellet was resuspended in Buffer I (0.25 M mannitol, 0.2 M EDTA, 1 mM MgCl2, 10 mM KCl, 10 mM KPi) and the oxygen consumption was measured at baseline and systemically in response to the components of the citric acid cycle and other regulatory components: ADP, (i) pyruvate and malate, (ii) succinate and rotenone, (iii) ascorbate and TMPD (N,N,N’,N’ –tetramethyl-p-phenyleneamine dihydrochloride), (iv) palmitoyl CoA (β-oxidation of fatty acids) and (v) α-ketoglutarate (58). The results were normalized relative to activity of citrate synthase, which has been traditionally used as an indicator of mitochondrial mass.
Mitochondrial isolation and blue native gel electrophoresis
Cerebrum and cerebellum were taken from 2-month-old male mice for mitochondrial isolation. Tissues were homogenized with a zero-clearance Teflon pestle in ice-cold HIM buffer (200 mm mannitol, 70 mM sucrose, 10 mM HEPES, 1 mM EGTA, adjusted to pH 7.5 with KOH) (59) and centrifuged at 600g for 20 min. The supernatant was further centrifuged at 600g for 20 min and at 10 000g for 10 min. The resulting mitochondrial pellet was washed with HIM buffer and centrifuged again at 10 000g for 10 min. The pellet was resuspended in PBS containing protease inhibitor cocktail, and the protein concentration was determined. Native gradient gels (5–12%) were casted and run according to the protocol of Antonicka et al. (60). Equal amounts of protein (10–20 µg) from control and Ppt1Δex4 mouse tissues were analyzed with Coomassie staining and western blot using monoclonal antibodies against β-subunit of complex V (Molecular Probes) and complex II subunit 70 kDa Fp (Mitosciences). They were compared with reveal possible differences caused by alterations in usability of nuclear- and mitochondrial-encoded proteins.
Mouse primary neurons and fibroblasts were seeded on 35 mm glass bottom dishes (MatTek) and grown for 8 and 1 days, respectively. Cells were fixed with 4% PFA and permeabilized with 0.1% Triton X-100 in PBS for 30 min in 4°C. Immunofluorescence staining was performed in room temperature using monoclonal antibodies against α- and β-subunits and secondary Alexa 488-fluorescent antibody (Molecular Probes). Images were taken with TILL imaging system including TILL TIRF condenser, Polychrome IV monochromator, Omnichrome series 43 krypton–argon laser (TILL Photonics) and Olympus NA 1.45 objective on Olympus IX-71 microscope including Andor DV885 EMCCD camera (Andor Technology) with Z488BP filter set (Chroma Technology). The images were further processed with Adobe Photoshop CS and Adobe Illustrator CS software. The correlation analysis was performed with Image Pro Plus software's co-localization tool (Media Cybernetics) by comparing the co-localization of the total fluorescence and the fluorescence from the cell surface. For co-localization analysis a fixed value determined as the intensity on a cell free area in the image has been subtracted. In short, the higher the co-localization correlation value (Pearson's correlation), the more protein is located on the cell surface. Neuronal analysis included total of 20–30 analyzed images per condition from three individual experiments. The correlation was calculated for the whole image for neurons. Fibroblast analysis included total of 13–24 analyzed images per condition from at least two individual experiments. The mitochondria were excluded from the fibroblast analysis to reduce the variation.
Apolipoprotein A–I internalization to glial cells and neurons
The internalization of apolipoprotein A-1 was performed according to the protocol of Siggins et al. (61) with minor modifications. Briefly, purified human apoA-I (kindly provided by Swiss Red Cross, Dr. Peter Lerch) was radiolabeled according to previous protocol (62). Neuronal and glial cell cultures were grown on 58 mm dishes with two coverslips for 8 days. The specificity of the cells was confirmed by immunostaining with neuron-specific mouse anti-tubulin antibody (β-tubulin III, Chemicon) and astrocyte-specific rabbit anti-cow glial fibrillary acidic protein antibody (GFAP, DAKO). Meanwhile, the cells were washed gently with PBS and incubated in serum-free media containing 5 µg/ml of [35S]-apo-A-1 (specific activity ∼20,000 cpm/µg A-I) for 24 h. The cells internalized 1–2% of the available radiolabeled lipid-free apoA-I. Culture media was collected and cells from two dishes were combined and lysed with 1 ml of lysis buffer (0.1 M Tris–HCl pH 7.8, 1% SDS, protease inhibitor cocktail, Roche). Samples from culture media and cell lysates were measured for protein content and precipitated with 5% trichloroacetic acid. The radioactivity of the precipitated (native) and non-precipitated (degraded) samples was measured with liquid scintillation counting, normalized to protein concentration and the radioactivity of culture media and cells were compared.
Mouse serum lipid analysis
One-month-old female mice were fasted for a 4-h period and blood for serum lipid and lipoprotein analysis was collected. Iodoacetic acid (1 mm final) was added to samples to prevent LCAT activity. Wild-type mice (n = 6) and Ppt1Δex4 mice (n = 15) were used for the study. Triglycerides (GPO-PAP 1488872 kit, Roche Diagnostics) and total cholesterol (CHOD-PAP 1489232 kit, Roche Diagnostics) were measured using fully enzymatic methods. PLTP activity was measured using radiometric assay (63). Mouse apoA-I was quantified by a sandwich ELISA (64). Mouse apoE was analyzed by ELISA method using specific antibodies against purified mouse apoE. For the lipoprotein analysis, serum samples were pooled (final pool size: 120–200 µl) and fractionated by fast-performance liquid chromatography (FPLC) using a Superose 6HR 10/30 size-exclusion chromatography column (Amersham Pharmacia Biotech). The column was equilibrated with Tris-buffered saline (10 mM Tris, 150 mM NaCl, pH 7.4) and 0.5 ml fractions were collected at flow rate 0.5 ml/min. Column fractions were analyzed immediately for lipids and apoE.
This study was financially supported by the European Commission 6th Framework Research Grant LSHM-CT-2003-503051, Academy of Finland Centre of Excellence in Complex Disease Genetics Grant 213506 and Centre of Excellence in Mitochondrial Disease and Aging, as well as by the Sigrid Juselius Foundation and Arvo and Lea Ylppö Foundation, which are gratefully acknowledged.
We are grateful to Auli Toivola for excellent technical assistance. Kaija Antila and Tuula Manninen are thanked for the help with the immunohistochemical stainings and mouse serum sample collection, respectively. Gunilla Rönnholm is thanked for the mass spectrometric analyses. Tomas Strandin is thanked for the Biacore-analysis. Jari Metso, Sari Nuutinen and Mikko Muilu are thanked for the help and expertise in serum assays. We thank the Molecular Imaging Unit (MIU, Biomedicum Helsinki), where the image analysis was partly done, for providing the facilities.
Conflict of Interest statement. None declared.