|Home | About | Journals | Submit | Contact Us | Français|
Metabolic reprogramming is a pathological feature of cancer and a driver of tumor cell transformation. N-Acetylaspartate (NAA) is one of the most abundant amino acid derivatives in the brain and serves as a source of metabolic acetate for oligodendrocyte myelination and protein/histone acetylation or a precursor for the synthesis of the neurotransmitter N-acetylaspartylglutamate (NAAG). NAA and NAAG as well as aspartoacylase (ASPA), the enzyme responsible for NAA degradation, are significantly reduced in glioma tumors, suggesting a possible role for decreased acetate metabolism in tumorigenesis. This study sought to examine the effects of NAA and NAAG on primary tumor-derived glioma stem-like cells (GSCs) from oligodendroglioma as well as proneural and mesenchymal glioblastoma, relative to oligodendrocyte progenitor cells (Oli-Neu). Although the NAA dicarboxylate transporter NaDC3 is primarily thought to be expressed by astrocytes, all cell lines expressed NaDC3 and, thus, are capable of NAA up-take. Treatment with NAA or NAAG significantly increased GSC growth and suppressed differentiation of Oli-Neu cells and proneural GSCs. Interestingly, ASPA was expressed in both the cytosol and nuclei of GSCs and exhibited greatest nuclear immunoreactivity in differentiation-resistant GSCs. Both NAA and NAAG elicited the expression of a novel immunoreactive ASPA species in select GSC nuclei, suggesting differential ASPA regulation in response to these metabolites. Therefore, this study highlights a potential role for nuclear ASPA expression in GSC malignancy and suggests that the use of NAA or NAAG is not an appropriate therapeutic approach to increase acetate bioavailability in glioma. Thus, an alternative acetate source is required.
Glioma, the most common adult primary brain cancer, is lethal due to a near inevitable post-surgical tumor recurrence due, in part, to radiation- and chemotherapy-resistant glioma stem-like cells (GSCs).2 Because oligodendrocyte progenitor cells (OPCs) are a glioma cell of origin (1), an understanding of the mechanisms regulating OPC proliferation and differentiation is important toward developing effective therapies against GSCs.
N-Acetylaspartate (NAA) is the second most abundant amino acid derivative in the brain, second only to glutamate, and is the most abundant source of acetate (2). NAA is synthesized in neuronal mitochondria (3) or endoplasmic reticulum (4) via acetylation of aspartate by the enzyme aspartate N-acetyl transferase (Fig. 1). NAA is the primary storage form of metabolic acetate needed for postnatal myelin lipid synthesis but is also hypothesized to contribute to neuronal mitochondrial metabolism, neuronal osmoregulation, and axon-glial signaling (5). Additionally, NAA may be converted to the dipeptide neurotransmitter N-acetylaspartylglutamate (NAAG) by the enzymes NAAG synthetase I and II (6, 7). NAAG selectively activates metabotropic glutamate receptor type 3 (GRM3) and is restored to NAA by glutamate carboxypeptidases (GCPII/III) expressed on the outer surface of astrocytes (8). Accordingly, NAAG serves as a source of perisynaptic NAA and glutamate.
Although NAA is of neuronal origin, its primary site of catabolism is in oligodendrocytes where it is hydrolyzed by the enzyme aspartoacylase (ASPA) to generate free acetate and aspartate (9). NAA-derived acetate needs to be converted to acetyl-coenzyme A (acetyl-CoA) via the cytosolic/nuclear enzyme acetyl-CoA synthetase 1 (AceCS1) for lipid biosynthesis and histone/protein acetylation (10) or mitochondrial acetyl-CoA synthetase 2 for use in the Krebs cycle (11). Liberated aspartate may generate oxaloacetate for utilization in the Krebs cycle or may be used for protein synthesis (5). Although best characterized as a cytosolic lipogenic enzyme (12), ASPA may also serve as an acetate source for the acetylation of protein substrates especially in oligodendrocytes where ASPA is abundant. The nuclear co-localization of ASPA (13) and AceCS1 (14) suggests that ASPA-mediated NAA catalysis may provide acetate for histone acetylation, thereby regulating gene expression necessary for oligodendrocyte differentiation. Thus, ASPA and AceCS1 possess functions distinct from their traditional roles in myelinogenesis (15).
Levels of NAA and NAAG are dysregulated in various disease states associated with aberrant oligodendrocyte differentiation. Missense mutations in ASPA result in Canavan disease (CD), a fatal childhood leukodystrophy characterized by white matter dysmyelination and spongiform degeneration arising from the inability to hydrolyze NAA and liberate acetate for myelin lipid synthesis (16). Rodent models of CD display defects in oligodendrocyte maturation and increased OPC proliferation, providing a possible link between NAA catabolism and OPC cell cycle arrest and/or oligodendrocyte differentiation (17, 18). Whereas increased NAA levels exemplify CD, levels of both NAA and NAAG are decreased in glioma tumors (19, 20) coincident with decreased ASPA expression.3 Inasmuch as ASPA deficiency is associated with increased OPC proliferation in CD, ASPA is decreased in glioma, and OPCs are a cellular source for gliomagenesis, ASPA and NAA/NAAG may be possible targets for therapeutic intervention. This study sought to determine the effects of NAA and NAAG supplementation on OPC and GSC proliferation and differentiation. Collectively, our results show that NAA and NAAG promote GSC growth and inhibit GSC differentiation. Therefore, metabolic changes that increase the abundance of NAA or NAAG may exacerbate glioma tumor pathology by reinforcing a proliferative undifferentiated GSC phenotype.
Oli-Neu cells, derived from mouse OPCs and immortalized by stable constitutive expression of the ErbB2 receptor (21), were grown on poly-l-lysine (PLL; 10 μg/ml)-coated dishes in SATO growth medium (DMEM with 0.1 mg/ml apotransferrin, 0.01 mg/ml insulin, 400 nm triiodothyronine, 2 mm glutamine, 200 nm progesterone, 100 μm putrescine, 220 nm sodium selenite, 500 nm thyroxine, 1% horse serum, and 25 μg/ml G418) (22). Human cerebral cortical astrocytes (HA#1800 ScienCell; Carlsbad, CA) were cultured in basal medium with 2% fetal bovine serum and astrocyte growth supplement (AM#1801 ScienCell). GSCs were maintained in DMEM/F-12 (Corning; Tewksbury MA) supplemented with 2 μm glutamine, 1× B27 (Invitrogen), 20 ng/ml EGF, and 20 ng/ml basic FGF (PeproTech; Rocky Hill, NJ). All media contained 50 units/ml penicillin and 50 μg/ml streptomycin (Invitrogen).
For pharmacological treatments, cells (20,000 cells per well of a 24-well plate or 10,000 cells per cm2 per 6-cm dish) were cultured overnight in growth medium then treated with NAA (100 μm) or NAAG (10 μm) (Sigma). Medium was replenished every 48 h, and cells were harvested after 2, 4, and 6 days in vitro (i.e. 1, 3, 5 days of treatment).
Oli-Neu differentiation was induced using modified SATO medium (DMEM supplemented with 100 μg/ml apotransferrin, 5 μg/ml insulin, 60 nm triiodothyronine, 30 nm sodium selenite, 100 μm putrescine, 1% horse serum, and 25 μg/ml G418) (23) upon plating with the addition of dibutyryl-cAMP (1 mm, Sigma) for up to 5 days or with the ErbB2 antagonist PD174265 (1 μm, sc-204170, Santa Cruz Biotechnology; Santa Cruz, CA) (24) for 48 h. GSC differentiation was induced in differentiation media (DM; DMEM with 10% fetal bovine serum). Medium was replenished every 48 h.
Growth dynamics were assessed using unbiased trypan blue exclusion-based cytometry. Cells were plated (at 10,000 cells per well of a 24-well plate) directly in the absence or presence of NAA (100 μm), NAAG (10 μm), or glutamate (10 μm, 50 nm). After 1, 3, and 5 days of treatment, cells were counted according to the manufacturer's instructions (Countess Automated Cell Counter; Invitrogen).
GSCs were cultured as non-adherent spheres in stem cell medium (SCM), and Oli-Neu cells were cultured on PLL in SATO media at a density of 2 × 105 cells/well of a 6-well plate. After 4 days, total RNA was isolated using STAT-60 (TelTest Inc.; Friendswood, TX), and DNase was treated using the SV Total RNA isolation system (Promega; Madison, WI). RNA (2 μg) was reverse-transcribed using Super Script II reverse transcriptase and random hexamers (Invitrogen). Adult mouse cerebral cortex, human anaplastic oligodendroglioma, and glioblastoma tumors served as positive controls. The cDNA (1 μl) was amplified using a HotStarTaq master mix (Qiagen; Valencia, CA) with the following primers (500 ng/sample): NaDC3 (forward, 5′-GTGGTCATCGCCTTCTTCAC-3′; reverse, 5′-CTTTGACCAGCAAGTGTCCAG-3′, 211 bp), GCPII/III (recognizes both GCPII and GCPIII; forward, 5′-TCAGAGTGGAGCAGCTGTTG-3′; reverse, 5′-CCTCTGCCCACTCAGTAGAAC-3′, 146 bp), mouse GRM2 (forward, 5′-GTTTGCAATGGCCGTGAGG-3′; reverse, 5′-GCTCCAGCCAACTTCCTCCT-3′, 132 bp), human GRM2 (forward, 5′-AAGTATGTTGGGCTCGC-3′; reverse, 5′-TCTGTACCCGGTAGTCACTG-3′, 194 bp), and GRM5 (forward, 5′-AGTGCACAGTCCAGTGAGAG-3′, reverse, 5′-CCACTCTCTGAATGCCATACTG-3′, 154 bp). Three exon-spanning GRM3 primers were used to confirm the absence of GRM3 expression in GSCs. The first span exons 2–3 (forward, 5′-AGCAGTGTTTCCATACAGGTG-3′; reverse, 5′-GCTTTGGCCTGGTAGAAGTC-3′, 149 bp), and the second span exons 5–6 (forward, 5′-CCTGAGTGGCTTTGTGGTCT-3′; reverse, 5′-GATGAGGTGGTGGAGTCGAG-3′, 210 bp). Finally, primers that would give rise to 951-bp or 343-bp products in the absence or presence of exon 2, respectively, were used (forward, 5′-CAAAGCCAGTAAGCTACCTCT-3′; reverse, 5′-ATCCCTGTCTCCCCGTAATC-3′). For loading controls, β-actin was used for murine Oli-Neu cells, whereas GAPDH was used for all human cells: β-actin (forward, 5′-TATTGGCAACGAGCGGTTCC-3′; reverse, 5′-GGCATAGAGGTCTTTACGGATGTC-3′, 139 bp); GAPDH (forward (5′-GAAGGTGAAGGTCGGAGTCA-3′; reverse, 5′-TTGAGGTCAATGAAGGGGTC-3′, 117 bp). After a 15-min 98 °C heat activation step, cycling parameters of 95 °C for 30 s, 58 °C for 30 s, and 72 °C for 30 s were repeated 32 times followed by a 1-min final extension at 72 °C. PCR products (10 μl) were resolved via agarose gel electrophoresis and visualized with ethidium bromide staining using a Chemidoc gel imaging system (Bio-Rad). PCR product specificity was confirmed by sequencing.
Antibodies were as follows: rabbit anti-mouse ASPA (2000× for Western blots) (25), rabbit anti-human ASPA (7500× for Western blots, 500× for immunocytochemistry; GTX13389 GeneTex; Irvine, CA), mouse anti-CD44 (5000× for Western blots, 2000× for immuno, #5640 Cell Signaling; Danvers, MA), mouse anti-porcine glial fibrillary acidic protein (GFAP; 2500× for Western blots, 4000× for immuno G3893; Sigma), neuron-specific βIII tubulin (Tuj1; 200,000× for Western blots; MO15013 Neuromics Antibodies; Edina, MN). Rabbit anti-mouse 2′,3′-cyclic nucleotide 3′-phosphodiesterase (CNPase; 5000× for Western blots, 500× for immuno, sc-30158), mouse anti-human histone H1 (200×, sc-8030), goat anti-human actin (1,000×, sc-1616), and rabbit anti-human GAPDH (5,000× for whole cell lysates, 10,000× for subcellular fractionation, sc-25778) were obtained from Santa Cruz Biotechnology. Rabbit anti-human Ki67 (50×, ab833), mouse anti-human nestin (1000×, ab22035), rabbit anti-human Sox2 (1000×, ab97959), and mouse anti-human Tuj1 (5000× for immuno, ab7751) were from Abcam (Cambridge, MA). Species-specific HRP- (3000×), Cy3- (500×), and Cy2- (100×) conjugated secondary antibodies were obtained from Jackson ImmunoResearch (West Grove, PA).
To determine ASPA spatial localization, cells (2.5 × 105 cells/10 cm dish) were cultured in DM for 4 days, collected by trypsinization (0.025% trypsin/EDTA), centrifuged at 1500 rpm for 5 min, and then washed with Dulbecco's phosphate-buffered saline. Cells (1 × 106) were resuspended in 200 μl of buffer A (10 mm Hepes, pH 7.6, 10 mm KCl, 0.1 mm EDTA, 0.1 mm EGTA, 0.75 mm spermidine, 0.15 mm spermine) with protease inhibitors (2 μm dithiothreitol, 2.5 μm phenylmethanesulfonyl fluoride, 100 μm Na2MoO4, and 5 μg/ml aprotinin, leupeptin, and pepstatin) and incubated on ice for 15 min. Nonidet P-40 (IgepalCA-630, 12.5 μl of 10%) was added dropwise while vortexing for 10 s. After centrifugation at 1300 rpm for 30 s, the supernatant (cytosolic fraction) was removed and stored at −80 °C. Radioimmune precipitation assay buffer (50 mm Tris-HCl, pH 8.0, 150 mm NaCl, 1.0% Nonidet P-40, 0.5% sodium deoxycholate, and 0.1% sodium dodecyl sulfate) plus protease inhibitors (25 μl) were added to the remaining pellet, and samples were heated for 5 min at 95 °C and pulse-sonicated 10 times for 1 s each. These samples were then stored at −80 °C as the nuclear fraction.
SDS-PAGE (25 μg of protein from whole cell lysates), and Western blotting were performed as previously described (26). For antibody blocking, ASPA antibody (0.4 μg, GTX13389 GeneTex) was incubated with either a 5-fold (blocking of novel ASPA species) or 10-fold (blocking of 36-kDa ASPA protein) molar excess of ASPA protein (GTX110699-PRO GeneTex) in Tris-buffered saline (10 mm Tris, pH 7.4, 150 mm NaCl) with 0.4% Tween 20 and 3% bovine serum albumin. To increase protein/antibody interaction, samples were incubated at 37 °C for 1 h (novel ASPA species) or 2 h (36 kDa ASPA) before overnight incubation at 4 °C with shaking. Immune complexes were centrifuged for 15 min at 1300 rpm at 4 °C, and the ASPA-depleted supernatant was incubated with Western blots containing 10 μg of protein (whole cell lysate). Immunocomplexes were visualized by enhanced chemiluminescence (PerkinElmer Life Sciences), and densitometry was performed using Quantity One software (Bio-Rad).
Immunocytochemistry was performed as previously described (27). For antibody blocking, ASPA antibody (0.3 μg) was incubated in a 10-fold molar excess of ASPA protein in blocking buffer (DMEM with 5% FBS, 0.1% glycine, 0.1% lysine, 0.2% sodium azide). Samples were incubated at 37 °C for 1 h and processed as described above for Western blot analysis. Immunoreactivity was visualized with a Nikon epifluorescence microscope (MicroVideo Instruments; Avon, MA), and all digital images were acquired with identical exposure settings using a SPOT RT digital camera (Diagnostic Instruments; Sterling Heights, MI).
Statistical analyses were performed using a minimum of three independently prepared cultures. Data are expressed as the means ± S.E. Significant differences were determined by either one-way or two-way analysis of variance and Bonferroni multiple comparison tests using Prism software (GraphPad; San Diego, CA). p < 0.05 was considered statistically significant.
NAA levels and ASPA expression are up-regulated during oligodendrocyte differentiation (18, 28), and ASPA-mediated NAA catabolism is necessary for proper oligodendrocyte maturation in vivo (17, 18). Conversely, NAAG is proposed to promote survival and expansion of neural progenitor cells in vitro through GRM3 activation (29). However, the effects of NAA and NAAG on OPCs have not been directly investigated. Therefore, the influence of physiological levels of NAA or NAAG (100 and 10 μm, respectively) on ASPA expression and differentiation were examined using a murine OPC line (Oli-Neu) that differentiates to oligodendrocytes with high fidelity. Neither NAA nor NAAG altered Oli-Neu growth (Fig. 2A). Oli-Neu cells expressed the NAA transporter NaDC3 (Fig. 2B); thus, they are competent to up-take NAA. To our knowledge this is the first report of NaDC3 expression in cells of an oligodendroglial lineage. NAAG may be hydrolyzed to NAA and glutamate by GCPII/III or function as a group II metabotropic glutamate receptor agonist, with high selectivity for GRM3 over GRM2 (8). GRM5, a group I metabotropic glutamate receptor, is expressed by neural progenitor cells and OPCs and is implicated in progenitor cell proliferation and survival (30, 31). Although NAAG does not stimulate GRM5 (8), its expression was nonetheless assessed. GRM5, but not GRM3 or GRM2, mRNA was detected in Oli-Neu cells (Fig. 2B). Because Oli-Neu cells did not express detectable GCPII/III mRNA, NAAG may remain as an intact dipeptide and function in the absence of GRM3 or be hydrolyzed to NAA and glutamate via a GCPII/III-independent mechanism. In the CNS, GCPII expression is primarily restricted to astrocytes (32); thus, its absence in Oli-Neu cells is consistent with their OPC phenotype.
Oli-Neu cells demonstrated a temporal increase in ASPA expression with time in culture (Fig. 2C). This temporal increase in ASPA expression is likely not due to spontaneous differentiation induced in confluent cultures as no change in CNPase expression was detected (Fig. 2C). Interestingly, ASPA was expressed as a 36–38-kDa immunoreactive doublet. Whether this doublet reflects post-translational modifications (e.g. phosphorylation, glycosylation) (33) is not known. Although NAA transiently increased ASPA expression, this was not associated with a concordant increase in CNPase expression (Fig. 2C) but was associated with increased process outgrowth (Fig. 2D). In contrast, NAAG treatment abrogated increased ASPA expression (Fig. 2C) and reduced CNPase expression (Fig. 2C) and process outgrowth (Fig. 2D). These results reveal an inhibitory role for NAAG on oligodendrocyte differentiation of Oli-Neu cells.
To determine whether increased ASPA expression failed to promote Oli-Neu CNPase expression and differentiation due to its inability to overcome constitutive ErbB2 expression, cells were induced to differentiate using cAMP (21) or inhibition of ErbB2 signaling using the ErbB2 antagonist PD174265 (24) (Fig. 2, E–G). Treatment with cAMP induced morphological alterations reflective of differentiated oligodendrocytes (e.g. highly arborized processes) (Fig. 2E) and increased CNPase expression; however, cAMP significantly reduced ASPA expression (Fig. 2F). Although NAA alone did not alter ASPA or CNPase expression, NAA abolished the cAMP-induced reduction of ASPA expression and attenuated the cAMP-mediated increase in CNPase (Fig. 2F), with the majority of cells retaining their undifferentiated bipolar morphology (Fig. 2E). Differentiation induced by ErbB2 inhibition did not alter ASPA expression but induced more robust morphological alterations and increased CNPase expression more profoundly than cAMP (Fig. 2G). Collectively, these data support a cAMP-mediated regulation of ASPA rather than a differentiation-mediated regulation. Furthermore, in the context of cAMP signaling, NAA appears to antagonize OPC differentiation.
Because mouse models of oligodendroglioma reveal transformed cells with phenotypic qualities similar to OPCs (34) and NAA and NAAG are decreased in glioma (19, 20), the effects of NAA or NAAG on cell growth were assessed in GSCs derived from oligodendroglioma tumors. NAA and NAAG exerted a growth-promoting effect on the more malignant anaplastic OG35 GSCs but not the grade II OG33 GSCs (Fig. 3A). Although some commercial preparations of NAAG have been reported to contain contaminating glutamate (between 0.1 and 0.5%) (8), neither 50 nm glutamate (equivalent to 0.5% contamination of 10 μm NAAG) nor 10 μm glutamate promoted OG35 GSC growth (Fig. 3B). Similar to Oli-Neu cells, OG33 and OG35 GSCs expressed NaDC3 but not GCPII/III (Fig. 3C) and, thus, are capable of NAA up-take but not GCP-mediated NAAG catabolism. Unlike the murine GRM3 gene, which encodes a single mRNA transcript, the human GRM3 gene codes for multiple mRNA splice variants. GRM3 expression was not detectable in OG GSCs using primers spanning exons 5–6 (Fig. 3C) and exons 2–3 (data not shown). Even using primers that would detect all variants, GRM3 mRNA was not detected in OG33 and OG35 GSCs (data not shown). Although the absence of GCPII/III suggests that NAAG promotes growth by signaling as an intact dipeptide, neither its cognate receptor GRM3 nor the related GRM2 and GRM5 receptors were detected in OG35 GSCs (Fig. 3C). Additionally, the more pronounced growth-enhancing effect of NAAG over NAA occurs independently of glutamate contamination or from glutamate derived by NAAG degradation.
Inasmuch as NAA levels are significantly reduced in glioma tumors relative to normal brain tissue (19), a parsimonious explanation for decreased ASPA expression in glioma in vivo is a compensatory negative feedback mechanism due to reduced bioavailability of NAA, the only known substrate of ASPA (35). Thus, the regulation of ASPA expression was examined in GSCs grown in stem cell medium (SCM) and when induced to differentiate in DM in the absence and presence of NAA and NAAG. ASPA was expressed at comparable levels in SCM and DM (Fig. 3D). NAA and NAAG treatment induced a modest increase in ASPA expression in OG33 GSCs, whereas only NAAG increased ASPA expression in OG35 GSCs (Fig. 3D). Curiously, NAA and NAAG induced the expression of a novel ~26-kDa immunoreactive ASPA species exclusively in OG33 GSCs (Fig. 3D). Consistent with previous reports that ASPA is a cytosolic-nuclear protein (13, 14), subcellular fractionation of GSCs grown in DM with NAAG revealed the putative 36-kDa ASPA in both the cytosol and nucleus (Fig. 3E). Strikingly, the ~26-kDa ASPA protein was expressed exclusively in OG33 nuclei (Fig. 3E). Among the various ASPA mutations detected in CD, a single base change at Tyr-231 occurs with high frequency and creates a premature termination codon and 26.6-kDa inactive protein (36), a protein similar in molecular mass to the novel ASPA isoform. However, sequencing of genomic DNA revealed that OG33 GSCs possess a silent single nucleotide polymorphism at this locus but are homozygous for wild-type ASPA, indicating that the 26-kDa immunoreactive ASPA species does not represent the Y231X ASPA mutation (data not shown). At present, the basis for the 26-kDa ASPA species is not known.
In keeping with Western blot data, immunocytochemical analyses revealed a cytosolic-nuclear pattern of ASPA expression in OG33 and G35 GSCs (Fig. 3F). Both cell lines exhibited the most pronounced nuclear ASPA staining when cultured in DM, suggesting a relationship between ASPA subcellular distribution and growth in differentiation permissive conditions. However, OG33 and OG35 GSCs were differentiation-resistant (i.e. no significantly increased GFAP, CNPase, or Tuj1 expression) in the absence or presence of NAA and NAAG (not shown).
Whereas the Cancer Genome Atlas classifies GBM into four subclasses (i.e. proneural, neural, classical, and mesenchymal) (37), OG33 and OG35 GSCs were profiled immunocytochemically to determine their subclass. Both GSCs were immunoreactive for the mesenchymal marker CD44 but not for proneural markers Sox2 and nestin (Fig. 4). In contrast, a proneural GBM GSC line was highly immunoreactive for Sox2 and nestin but only weakly so for CD44 (Fig. 4). Thus, OG33 and OG35 GSCs best conform to a mesenchymal tumor phenotype and may not be an ideal model to investigate the role of NAA, NAAG, and ASPA in glial differentiation.
We next sought to determine if NAA and NAAG regulated growth or differentiation of proneural GBM GSCs. NAA, but not NAAG, promoted proneural GBM GSC growth in SCM (Fig. 5A). Serum-induced differentiation was associated with reduced growth, which was unaltered by NAA or NAAG (Fig. 5B). The proneural GBM GSCs expressed much lower levels of NaDC3 (Fig. 5C) than Oli-Neu cells (Fig. 2B) or OG GSCs (Fig. 3C). Unlike oligodendroglioma GSCs, proneural GBM GSCs expressed high levels of GCPII/III; hence, NAAG can be hydrolyzed to NAA and glutamate (Fig. 5C). Growth factor depletion of the proneural GBM GSCs was associated with significantly increased expression of the astrocytic marker GFAP and the neuronal marker Tuj1, but there was no change in CNPase expression (Fig. 5D). NAA and NAAG reduced glial differentiation capacity (i.e. attenuated GFAP expression, reduced basal levels of CNPase) without a corresponding increase in neuronal cell fate (i.e. unaltered Tuj1 expression) (Fig. 5D). Collectively, these results suggest that NAA and NAAG suppress the expression of genes associated with glial, but not neuronal, differentiation.
Unlike oligodendroglial GSCs, proneural GBM GSC differentiation was associated with increased ASPA expression, which, similar to Oli-Neu cells, was present as a 36–38-kDa immunoreactive doublet (Fig. 5E). Treatment with NAAG, but not NAA, attenuated increased ASPA expression (Fig. 5E). Immunocytochemical analysis revealed diffuse cytoplasmic and intense nuclear ASPA staining in GSCs (Fig. 5F). Upon growth factor withdrawal, the cells differentiated and retained cytoplasmic and nuclear ASPA expression, which was largely unchanged by NAA or NAAG.
Inasmuch as mesenchymal GBM GSCs exhibit a more glycolytic metabolism (38), the effect of NAA and NAAG on cell growth was examined in mesenchymal GBM GSCs (Fig. 6). Unlike proneural GBM GSCs, which significantly up-regulated GFAP expression in DM, the mesenchymal GBM GSCs did not express GFAP in SCM or DM (Fig. 6A). Conversely, the mesenchymal GBM GSCs expressed abundant CD44 in SCM, which decreased upon differentiation. The proneural GBM GSCs lacked CD44 in SCM but significantly increased expression when induced to differentiate so that CD44 can support process outgrowth. Upon differentiation of proneural GBM GSCs, ASPA was diffusely expressed throughout the cytoplasm where it was co-expressed with GFAP, nestin (Fig. 6B), and Tuj1 (not shown). In contrast, in the mesenchymal GBM GSCs, ASPA expression was enriched within the nucleus, and these cells lacked GFAP, nestin (Fig. 6B), and Tuj1 (not shown) expression but were Ki67-positive and, thus, maintained their proliferative capacity despite growth in differentiation permissive conditions. Because mesenchymal gliomas are associated with a poor prognosis (39), it was not surprising that growth of the mesenchymal GBM GSCs was greater than the proneural GBM GSCs in SCM (Fig. 6C). Furthermore, because the mesenchymal GBM GSCs do not differentiate (Fig. 6B), their proliferation in DM was even greater than in SCM. The mesenchymal GBM GSCS expressed NaDC3, GRM2, and GRM 5 but no GCPII/III or GRM3 (Fig. 6D), similar to the OG GSCs (Fig. 3C). However, NAA and NAAG increased OG35 growth, whereas neither affected growth of the mesenchymal GBM GSCs (Fig. 6C). Thus, although the mesenchymal GBM GSCs share phenotypic similarities with OG33 and OG35 GSCs (i.e. nuclear ASPA expression and limited differentiation potential), NAA and NAAG exerted different growth effects.
Finally, to assess its role in differentiation, nuclear ASPA expression was compared in the proneural and mesenchymal GBM GSCs. Total ASPA expression did not differ in proneural and mesenchymal GSCs but they expressed distinct ASPA isoforms (Fig. 6E). When cultured in DM, the proneural GBM GSCs increased expression of the higher molecular weight component of the ASPA doublet (~38 kDa). In the mesenchymal GBM GSCs, a novel ~24-kDa immunoreactive ASPA isoform was observed. Interestingly, unlike OG33 GSCs, in which the novel ASPA isoform was induced by NAA or NAAG (Fig. 3D), in the mesenchymal GBM GSCs, this isoform was constitutively expressed. Peptide blocking experiments resulted in a loss of reactivity of the 38- and 24-kDa immunoreactive isoforms and diminished reactivity of the 36-kDa ASPA (Fig. 6F). Blocking was also more efficient for ASPA immunolabeling of proneural GSCs than mesenchymal GSCs (Fig. 6F). Subcellular fractionation revealed that ASPA primarily partitioned to the cytosolic fraction in proneural GBM GSCs (Fig. 6G). In contrast, in the mesenchymal GBM GSCs, full-length ASPA (~36 kDa) was expressed exclusively in the cytosol, whereas the novel ~24-kDa ASPA isoform partitioned exclusively to the nuclear fraction. NAA or NAAG did not discernibly alter the subcellular distribution of ASPA (not shown). Collectively, these results identify a unique pattern of ASPA expression that corresponds with increased GSC proliferation and altered differentiation potential.
Although diminished NAA magnetic resonance spectroscopy signal is a well recognized feature of glioma tumors, the physiological consequences of reduced NAA or NAAG bioavailability are a poorly understood area of glioma pathology. This study sought to determine the effects of physiological NAA and NAAG supplementation on OPC and GSC proliferation and differentiation. Our results show that these metabolites promote GSC growth and attenuate glial differentiation, suggesting that increased NAA and NAAG may lead to dysregulated OPC differentiation and exacerbate glioma progression.
Recent metabolomic profiling revealed that NAA and NAAG are significantly reduced in glioma tumors expressing mutant isocitrate dehydrogenase 1 and 2 (IDH1/2) (20). Inasmuch as mutant IDH status is associated with a favorable prognosis in glioma (40), our finding that physiologically relevant doses of NAA and NAAG facilitate GSC growth and differentiation resistance is consistent with the reduction of these metabolites in less aggressive IDH mutant tumors. Despite the prevalence of mutant IDH in oligodendroglioma (41) and proneural GBM (37), the GSCs lines in this study expressed wild-type IDH1 and -2,3 likely owing to the negative selection of cells with IDH mutations in culture (42). Mutant IDH is hypothesized to drive oncogenesis, in part, by promoting epigenetic dysregulation and the induction of a CpG island methylator phenotype (43); however, why IDH mutations are simultaneously associated with reduced glioma malignancy remains elusive. This study raises the possibility that lower levels of NAA and NAAG could be one contributing factor.
The growth-promoting effects of NAA and NAAG were primarily detected after 5 days of treatment, suggesting a metabolic effect rather than activation of a signal transduction mechanism. In light of the established role for NAA as a source of lipogenic acetate during postnatal myelination (12) and that increased fatty acid synthesis is required to sustain anabolic growth of tumors cells (44), it is likely that NAA hydrolysis exacerbates glioma growth by providing lipogenic acetate. It is unlikely that NAAG signals through metabotropic glutamate receptors as NAAG selectively activates GRM3 (8) and NAAG promoted growth of OG35 GSCs that lacked GRM2, GRM3, and GRM5. Although it is possible that NAAG provides acetate via hydrolysis to NAA, OG35 GSC proliferation was increased in the absence of GCPII/III expression. Whether NAAG is converted to NAA by other carboxypeptidases is not known. Our finding that NAA and NAAG produced similar phenotypic effects in Oli-Neu cells and GSCs supports the interpretation that these metabolites function through a common metabolic pathway.
This study is consistent with reports from ASPA-deficient mice insofar as increased NAA levels are associated with increased OPC proliferation and diminished oligodendrocyte gene expression (17, 18). However, unlike ASPA-deficient mice, the OPCs and GSCs express ASPA and are capable of NAA catalysis, suggesting that increased NAA catabolism rather than the accumulation of metabolically inert NAA inhibits OPC differentiation. A major pathological feature of ASPA-deficient mice is the widespread death of immature oligodendrocytes (17). Accordingly, increased OPC proliferation may be a compensatory response to injury arising from defective myelination and diminished oligodendrocyte viability rather than a direct consequence of deficient NAA metabolism.
This study highlights a potential alternative role for NAA in the maintenance of an undifferentiated OPC phenotype. ASPA is expressed by OPCs in mice as early as embryonic day 12.5 (17), and 5–20% of NAA in the postnatal brain is present in proliferating OPCs (45); therefore, OPCs possess the capacity for NAA metabolism independent of myelination. The NAA transporter NaDC3 was expressed in Oli-Neu cells, suggesting that OPCs may take up neuronally derived NAA in vivo. Importantly, we found that NAA inhibited cAMP-mediated Oli-Neu differentiation. Whereas proneural tumors exhibit a gene expression signature that is highly reminiscent of OPCs (37), our observation that NAA and NAAG blunted proneural GBM GSC differentiation supports a role for NAA metabolism in the maintenance of an undifferentiated progenitor-like state.
In the absence of myelinogenesis, NAA-derived acetate/acetyl-CoA may transition from a lipogenic to an epigenetic function. In OPCs, widespread histone deacetylation is required to silence genes that negatively regulate oligodendrocyte differentiation (46). Thus, increased NAA metabolism may reinforce an acetylated histone state conducive to an OPC phenotype and may, likewise, attenuate proneural GSC differentiation. A relationship between NAA metabolism and histone acetylation is congruent with evidence that increased metabolic flux of other key sources of acetyl-CoA such as citrate (47) and acetylcarnitine (48) similarly promote histone acetylation and chromatin remodeling. Moreover, NAA-mediated modulation of acetate bioavailability would add an additional mechanism to regulate histone acetylation underlying OPC differentiation.
An unanticipated finding of this study was the detection of novel ASPA isoforms in OG33 and mesenchymal GBM GSC nuclei. Although ASPA is known to undergo cytosolic-nuclear shuttling (13, 14), its nuclear function is not yet established. Coordinated ASPA-mediated NAA catalysis and AceCS1-mediated acetyl-CoA synthesis may contribute to the nuclear pool of acetyl-CoA required to regulate epigenetic chromatin remodeling during oligodendrocyte development. A caveat of this hypothesis, however, is that nuclear ASPA exhibits diminished catalytic activity toward NAA relative to cytosolic ASPA (13), indicating that, under normal physiological conditions, nuclear ASPA is unlikely to contribute meaningfully to the nuclear acetyl-CoA pool. Alternatively, ASPA may possess a nuclear specific function independent of NAA catalysis. Inasmuch as ASPA-deficient mice show increased histone acetylation, it is possible that ASPA itself functions as a histone deacetylase (18). However, the C-terminal domain of ASPA is so closely juxtaposed with its catalytic site that it limits substrate selectivity to NAA (49). It is, therefore, intriguing that novel lower molecular weight immunoreactive ASPA species were selectively expressed in GSC nuclei. This presents the possibility that post-translational processing or differential mRNA splicing might give rise to ASPA variants with a more accessible catalytic site and confer broader substrate specificity.
Additionally, the mechanisms responsible for ASPA cytosol-nuclear shuttling are not known. It has been suggested that ASPA possesses a non-classical nuclear localization signal near its C terminus (13), but the specific conditions that regulate its nuclear localization are undefined. ASPA did not show overt nuclear regulation in response to NAA and NAAG; however, a lower molecular weight immunoreactive ASPA species was induced by NAA and NAAG in OG33 GSC nuclei, suggesting that its expression is regulated by conditions of heightened acetate load. In contrast, mesenchymal GBM GSCs constitutively expressed a comparable immunoreactive species in the nuclear compartment. Given that OG35 GSCs exhibit a mesenchymal signature similar to OG33 GSCs and the mesenchymal GBM GSCs but did not express a lower molecular weight ASPA isoform, it suggests that an additional regulatory mechanism drives the expression of the novel nuclear ASPA isoform. Thus, the GSC lines used in this study represent a model to better define the nuclear function of ASPA. Furthermore, if nuclear ASPA facilitates GSC malignancy, as is suggested by greater nuclear ASPA immunoreactivity in differentiation-defective mesenchymal GSCs, it would represent a potential target for therapeutic intervention. However, our study demonstrates that the use of NAA or NAAG is not an appropriate therapeutic approach to increase acetate bioavailability; thus, an alternative acetate source is required.
Facilities and equipment, supported by the Neuroscience COBRE (National Institutes of Health (NIH) Center for Research Resources Grant P20 RR016435) and Vermont Cancer Center DNA Analysis facility (NIH Grant P30 CA22435), were instrumental to the completion of the study. We thank Drs. Karen Fortner and Roxana del Rio Guerra (University of Vermont Department of Medicine) for training on flow cytometry use and data analysis, Dr. Felix Eckenstein (University of Vermont Department of Neurological Sciences) for providing Tuj1 antibody, and Dr. Jeffrey Spees (University of Vermont Department of Medicine) for providing Sox2 antibody.
*This work was supported, in whole or in part, by National Institutes of Health (NIH) Grant R01NS045225 (NINDS and Center for Research Resources (NCRR)) and Pilot Project grants from the Vermont Cancer Center/Lake Champlain Cancer Research Organization consortium and Neuroscience COBRE (NIH NCRR P20 RR016435) (to D. M. J.).
3A. R. Tsen and D. M. Jaworski, submitted for publication.
2The abbreviations used are: