Expression of 3-hydroxyisobutyrate dehydrogenase in cultured neural cells


Address correspondence and reprints requests to Dr Radovan Murín, Interfaculty Institute for Biochemistry, University of Tuebingen, Hoppe-Seyler-Str. 4, Tuebingen 72076, Germany.


The branched-chain amino acids (BCAAs) – isoleucine, leucine, and valine – belong to the limited group of substances transported through the blood–brain barrier. One of the functions they are thought to have in brain is to serve as substrates for meeting parenchymal energy demands. Previous studies have shown the ubiquitous expression of a branched-chain alpha-keto acid dehydrogenase among neural cells. This enzyme catalyzes the initial and rate-limiting step in the irreversible degradative pathway for the carbon skeleton of valine and the other two branched-chain amino acids. Unlike the acyl-CoA derivates in the irreversible part of valine catabolism, 3-hydroxyisobutyrate could be expected to be released from cells by transport across the mitochondrial and plasma membranes. This could indeed be demonstrated for cultured astroglial cells. Therefore, to assess the ability of neural cells to make use of this valine-derived carbon skeleton as a metabolic substrate for the generation of energy, we investigated the expression in cultured neural cells of the enzyme processing this hydroxy acid, 3-hydroxyisobutyrate dehydrogenase (HIBDH). To achieve this, HIBDH was purified from bovine liver to serve as antigen for the production of an antiserum. Affinity-purified antibodies against HIBDH specifically recognized the enzyme in liver and brain homogenates. Immunocytochemistry demonstrated the ubiquitous expression of HIBDH among cultured glial (astroglial, oligodendroglial, microglial, and ependymal cells) and neuronal cells. Using an RT-PCR technique, these findings were corroborated by the detection of HIBDH mRNA in these cells. Furthermore, immunofluorescence double-labeling of astroglial cells with antisera against HIBDH and the mitochondrial marker pyruvate dehydrogenase localized HIBDH to mitochondria. The expression of HIBDH in neural cells demonstrates their potential to utilize valine imported into the brain for the generation of energy.

Abbreviations used

astroglia-rich primary culture


branched-chain amino acid


branched-chain α-keto acid dehydrogenase


Dulbecco’s modified Eagle’s medium


fetal calf serum


glial fibrillary acidic protein




3-hydroxyisobutyrate dehydrogenase (3-hydroxy-2-methylpropanoate:NAD oxidoreductase)


horseradish peroxidase


myelin basic protein


amino acid-restricted/deficient minimal culture medium


phosphate-buffered saline


pyruvate dehydrogenase complex


sodium dodecyl sulfate


Tris-buffered saline

Only some of the restricted number of substances that are transported through the blood–brain barrier participate in brain energy metabolism. The amino acids valine, isoleucine, and leucine, which constitute the group of branched-chain amino acids (BCAAs), can enter the brain (Oldendorf 1971a,b) by facilitated transport through the endothelial cells of the blood vessels (Smith et al. 1987). In brain, they can be metabolized (Yudkoff 1997).

In contrast to the metabolism of leucine by brain cells (Murín and Hamprecht 2008) that of valine has not yet been studied intensively. In vitro experiments with slices from cortex showed the capacity of brain cells to oxidize 2-oxoisovalerate, the cognate 2-oxo acid of valine (Brand 1981). Astrocytes, which are considered to serve as ‘fuel processing plants’ for neurons (Hamprecht and Dringen 1995) by supporting them with substrates for energy metabolism (Pellerin 2005), have a high capacity to metabolize BCAAs. The cells of astroglia-rich primary cultures (APC) rapidly remove valine and the other BCAAs from their culture media even in the presence of glucose (Yudkoff et al. 1994; Bixel and Hamprecht 1995). While it has been shown that APC metabolize leucine and are capable of releasing into the culture media some intermediates of leucine catabolism (Bixel and Hamprecht 1995; Bixel et al. 2004), corresponding data for valine are lacking. Also, neurons have been shown to metabolize BCAAs, especially during hypoglycemia (Honegger et al. 2002).

The catabolic pathway of valine (Fig. 1; Sweetman and Williams 2001) shares some similarities with that of the other BCAAs and can be divided into three stages. During the first stage, valine is reversibly transaminated to 2-oxoisovalerate. This may subsequently, in the second stage, enter the irreversible part of valine catabolism and be converted to 3-hydroxyisobutyrate (HIB) by mitochondrial enzymes. In the third stage, HIB is enzymatically converted to a member of the citrate cycle, succinyl-CoA, and therefore is important in energy metabolism. Thus, the catabolism of valine can serve an anaplerotic function for the citrate cycle. This way it provides the opportunity for withdrawing members of the citrate cycle for synthetic purposes such as the biosynthesis of the neurotransmitters glutamate and GABA. In addition, succinyl-CoA may eventually serve as fuel for some neural cells when converted in the citrate cycle, e.g. to malate. This could – under the catalytic actions of malic enzyme and the pyruvate dehydrogenase complex (PDH) – be converted, via pyruvate, to the ultimate fuel, the acetyl residue of acetyl-CoA. Introduced into the citrate cycle, this compound could give rise to malate again. Such ‘pyruvate recycling,’ which implicates the formation of pyruvate from malate by malic enzyme, has been shown to take place in brain (Cerdan et al. 1990; Håberg et al. 1998), astrocytes (Sonnewald et al. 1996), and neurons (Olstad et al. 2007). Of the irreversible part of valine catabolism HIB is the first intermediate that is not linked to CoA and, therefore, can be transported through membranes into the extracellular space and subsequently metabolized by other, neighboring cells. In addition to this, the first enzyme of the irreversible part of valine catabolism, the multienzyme complex branched-chain α-keto acid dehydrogenase (BCKD; EC,, and; Pettit et al. 1978) is ubiquitously expressed among neural cells (Bixel et al. 2001). Thus, all the types of neural cells may convert valine to HIB, which may subsequently be either further degraded in the same cell or released from the cell for final metabolism by neighboring cells. Therefore, to assess the putative capability of neural cells to participate in valine catabolism by making use of HIB, the expression of HIBDH (3-hydroxy-2-methylpropanoate:NAD oxidoreductase; EC, the first enzyme from the third stage of the valine degradative pathway was studied in cultured neural cells by RT-PCR and immunological methods as well as by analyzing the release of HIB from APCs into the culture medium.

Figure 1.

 Scheme of the pathway of valine catabolism [modified version of the pathway presented by Sweetman and Williams (2001)]. The first two enzymes of the pathway, branched-chain amino acid transaminase (A) and branched-chain α-keto acid dehydrogenase (B) are common for all three BCAAs and their cognate branched-chain α-keto acids, respectively. Isobutyryl-CoA is subsequently converted to (S)-3-hydroxyisobutyrate in a sequence of three enzymatic reactions catalyzed by acyl-CoA dehydrogenase (C), enoyl-CoA hydratase (D), and 3-hydroxyisobutyryl-CoA hydrolase (E). Valine catabolism continues with oxidation of (S)-3-hydroxyisobutyrate to (S)-methylmalonic semialdehyde by 3-hydroxyisobutyrate dehydrogenase (F). These two compounds are the only intermediates of the irreversible part of valine catabolism that are not CoA esters. The transformation of (S)-methylmalonic semialdehyde to the member of the citrate cycle, succinyl-CoA, requires the series of enzymatic reactions catalyzed by methylmalonic semialdehyde dehydrogenase (G), propionyl-CoA carboxylase, methylmalonyl-CoA racemase, and methylmalonyl-CoA mutase. (The last three enzymes as well as the intermediates (S)-methylmalonyl-CoA and (R)-methylmalonyl-CoA, which are part of the pathway of propionyl-CoA metabolism, are not depicted in the scheme.) Stages 1, 2, and 3 refer to the three stages of valine catabolism mentioned in the text.

Materials and methods


Acrylamide, N,N′-(1,2-dihydroxyethylene)bisacrylamide, Triton X-100, and Tween 20 were from Carl Roth (Karlsruhe, Germany). Bovine serum albumin was purchased from Roche Diagnostics (Mannheim, Germany). Bovine insulin, 4′,6-diamidino-2-phenylindole dihydrochloride, enhanced chemiluminescent substrate for detection of horseradish peroxidase (HRP), progesterone, and putrescine were from Sigma-Aldrich (Deisenhofen, Germany). A mixture of dNTPs, O’GeneRuler 1 kb DNA ladder, protein ladder, and RevertAid First Strand cDNA Synthesis Kit were from Fermentas (St Leon-Rot, Germany). Dulbecco’s modified Eagle’s medium (DMEM), minimal essential medium, and transferrin were from Gibco-Invitrogen (Karlsruhe, Germany) and GeneRacer kit, oligo(dTTP)18-adapter primer, and sequence specific primers were purchased from Invitrogen (Karlsruhe, Germany). Fetal calf serum (FCS) was from Biochrom (Berlin, Germany). Fibronectin was purified from bovine plasma according to Miekka et al. (1982). HotStarTaq Master Mix Kit, RNeasy kit was from Qiagen (Hilden, Germany). Nitrocellulose transfer membranes were obtained from Millipore Corporation (Eschborn, Germany). Aqueous-based mounting medium Immu-mount was from Thermo Shandon (Pittsburgh, PA, USA). Penicillin G and streptomycin sulfate were from Serva (Heidelberg, Germany). Skimmed milk powder was from Töpfer GmbH (Dietmannsried, Germany). Sodium selenite and sodium dodecyl sulfate (SDS) were from Fluka (Deisenhofen, Germany). Human thrombin was from Aventis Behring (Marburg, Germany). All other chemicals were obtained at analytical grade from E. Merck (Darmstadt, Germany). Sterile plastic material for cell culture was from Nunc (Wiesbaden, Germany) and Greiner (Frickenhausen, Germany).


Mouse monoclonal anti-glial fibrillary acidic protein (GFAP) antibodies were purchased from Sigma-Aldrich. Anti-rat CD11b monoclonal antibodies (OX-42) were from Serotec (Oxford, UK). Alexa Fluor-conjugated anti-goat IgG, anti-rabbit IgG, and anti-mouse IgG were from Molecular Probes Europe (Leiden, The Netherlands). Goat antiserum against myelin basic protein (MBP), anti-rabbit IgG conjugated with HRP, and anti-human-IgG conjugated with tetramethylrhodamine isothiocyanate were from Santa Cruz Biotechnology (Heidelberg, Germany). Human anti-PDH antiserum was a gift from Drs Peter A. Berg and Reinhild Klein (Medical Clinics, University of Tuebingen, Germany; Berg et al. 1982).

Purification of HIBDH, generation of antisera, and affinity purification of anti-HIBDH antibodies

3-Hydroxyisobutyrate dehydrogenase was assayed spectrophotometricaly as described by Rougraff et al. (1988). To purify HIBDH, the method by Rougraff et al. (1988) was used in the modified form detailed as follows. Bovine liver (700 g) was homogenized in 1.4 L of buffer H [50 mM Tris/HCl, 2 mM EDTA, 0.1% (w/v) Triton X-100, 1 mM dithiothreithol, pH 8.0] using a Potter-Elvehjem homogenizer (Fisher Scientific, Schwerte, Germany). After centrifugation (12 000 g, 4°C, 15 min) polyethylene glycol was added to the supernatant to a final concentration of 6.5% (w/v). The supernatant resulting from the subsequent centrifugation (30 000 g, 4°C, 15 min) was applied on a diethylaminoethyl-cellulose column (4.1 × 27 cm) equilibrated with buffer A. The enzyme was eluted from the column with a KCl gradient (0–150 mM) in buffer A. The fractions containing HIBDH activity (50–120 mM KCl) were combined and subsequently subjected to ammonium sulfate fractionation (35–65% saturation). The enzyme was further purified by hydrophobic interaction chromatography on a phenyl-Sepharose column (4.1 × 17 cm) followed by a diethylaminoethyl-cellulose column (2.1 × 15 cm) and again a phenyl-Sepharose column (4.1 × 17 cm). In the final purification step, performed on an Affi-Gel Blue (Bio-Rad, Munich, Germany) column (2.1 × 7 cm), HIBDH was eluted with buffer A supplemented with 150 mM KCl and 0.2 mM NADH.

For immunization, two rabbits (Chinchilla bastards) were initially (day 0) injected subcutaneously with 2.5 mL of an emulsion containing 210 μg of purified HIBDH in complete Freund’s adjuvant. Boosting was carried out on days 33 and 66 using 40 μg of HIBDH in Freund’s incomplete adjuvant. Antibody titers were determined by ELISA. The two rabbits showing titers > 10 000 : 1 after the second booster injection were decapitated on day 71 and sera were prepared. For purification of HIBDH antibodies, rabbit antisera were loaded on an affinity column (1 × 2 cm) of HIBDH covalently linked to CNBr-activated Sepharose. After washing with a solution containing 0.5 M KCl and 10 mM phosphate buffer, pH 7.4, the antibodies were eluted with 15 mL of 4 M MgCl2 in 10 mM phosphate buffer, pH 7.4. The protein containing fractions were collected and dialyzed against phosphate-buffered saline (PBS), the HIBDH antibodies were concentrated by ultrafiltration method by using a Vivaspin 15 ultrafiltation spin column (Vivascience AG, Gottingen, Germany).

Enzymatic determination of HIB in culture medium

In culture dishes (92 mm in diameter) containing 7.5 mL of culture medium (90% DMEM/10% FCS) APC were cultured in an atmosphere of 90% air/10% CO2. After 14 days, the medium was discarded, the cells were washed and subsequently further cultured in either 7.5 mL of DMEM/FCS or in an amino acid-deficient minimal culture medium (MM; 109.5 mM NaCl, 5.4 mM KCl, 1.8 mM CaCl2, 0.8 mM MgSO4, 44 mM NaHCO3, 0.9 mM NaH2PO4, 7 mM d-glucose, and 325 mOsmol/L; Bixel and Hamprecht 1995) or in MM supplemented with 0.8 mM valine (MM-Val). At appropriate times, the media were collected and centrifuged (12 000 g, 4°C, 10 min). After subsequent heat-inactivation of enzymes in the supernatant by using a simmering water bath for 10 min, denatured proteins were removed by centrifugation (12 000 g, 4°C, 10 min). In the supernatants the HIB-concentration was determined by a modification of the method described by Rougraff et al. (1988). Briefly, after an addition of the thus pre-treated medium (200 μL) to 0.8 mL of buffer D (50 mM glycine, 1 mM mercaptoethanol, 3.3 mM MgSO4, 1.7 mM EDTA, and 1 mM NAD) the extinction (E0) was recorded. After an addition of 5 μL of HIBDH (250 μU HIBDH) the mixture was incubated till the extinction reached the plateau value ES. The HIB concentration was calculated from the change in extinction, ES–E0.

Cell cultures

Neuron-rich primary cultures were prepared from the brains of embryonic Wistar rats according to the method described by Löffler et al. (1986). APCs derived from the brains of neonatal Wistar rats were prepared and maintained according to Hamprecht and Löffler (1985). Oligodendroglia-rich secondary cultures derived from APCs in 175-cm2 flasks were prepared according to Hirrlinger et al. (2002). A modification (Hirrlinger et al. 2000) of the method described by Giulian and Baker (1986) was used to prepare microglia-rich secondary cultures from APCs. The method described by Prothmann et al. (2001) was used for the preparation of ependymal cell-rich primary cultures.

Immunoblot analyses

Cell culture homogenates were analyzed by SDS–polyacrylamide gel electrophoresis using a 5% stacking gel and a 12% separating gel. Proteins were transferred onto a nitrocellulose membrane (100 V constant voltage in pre-cooled transfer buffer for 1 h) and residual antibody binding sites on the membrane were saturated by incubation in Tris-buffered saline (TBS), pH 7.5, supplemented with 0.05% (w/v) Tween 20/5% (w/v) skimmed milk powder for 1 h. The membranes were washed three times with washing buffer [TBS/0.05% (w/v) Tween 20] and subsequently incubated with anti-HIBDH serum (diluted 1 : 500 in washing buffer) at 22°C for 4 h. After three washing steps with washing buffer, the membrane was incubated with goat anti-rabbit IgG-HRP conjugate [diluted 1 : 50 000 in TBS/0.05% (w/v) Tween 20] at 22°C for 1 h. The membrane was washed three times with washing buffer and once with PBS. Subsequently, a signal was developed by using an enhanced chemiluminescent detection kit, according the manufacturer’s instructions.


For immunocytochemistry, cells were grown on coverslips (18 × 18 mm) attached to the surface of a culture dish with sterile silicon grease. The coverslips were either untreated to grow astroglial and microglial cells or coated with fibronectin to grow ependymal cells or covered with poly-d-lysine for the cultivation of oligodendroglial or neuronal cells. After removal of the culture media, the cells on coverslips were washed three times with ice-cold PBS and subsequently fixed by immersion in 3.5% (w/v) p-formaldehyde/PBS at 22°C for 10 min. Fixed cells were washed twice in PBS for 5 min and once in PBS containing 0.1% (w/v) glycine for 10 min. To enhance antibody penetration, cells were permeabilized in PBS containing 0.3% (w/v) Triton X-100 for 10 min. To minimize the unspecific signals, the cells were incubated with PBS/0.1% (w/v) Triton X-100/10 mg/mL bovine serum albumin solution at 22°C for 30 min. For immunofluorescence double labeling, the coverslips were treated with diluted primary antibodies (anti-HIBDH, 1 : 200; anti-neurofilament 200, 1 : 200; anti-GFAP, 1 : 400; anti-CD11b, 1 : 40; anti-α-tubulin, 1 : 1000; and anti-PDH, 1 : 1000) in PBS/0.1% (w/v) Triton X-100/1 mg/mL bovine serum albumin and incubated (4°C, overnight) in a humidified chamber. After washing three times with PBS/0.1% (w/v) Triton X-100, the cells were incubated with a mixture of diluted secondary antibodies, i.e. goat anti-rabbit IgG antibodies conjugated with Alexa Fluor 488 and chicken anti-mouse IgG-Alexa Fluor 568. After washing three times with PBS/0.1% (w/v) Triton X-100, the tertiary antibodies, i.e. chicken anti-goat IgG conjugated with Alexa Fluor 468, were applied and coverslips were incubated in the dark at 22°C for 2 h. Working dilutions for secondary and tertiary antibodies conjugated with tetramethylrhodamine isothiocyanate and Alexa Fluor were 1 : 300 and 1 : 1000, respectively. Subsequently, the cells were washed three times in PBS/0.1% (w/v) Triton X-100 at 22°C for 15 min. As the anti-MBP serum originated from goat and the detection of HIBDH involved an anti-goat antibody, double immunofluorescence labeling for MBP and HIBDH was not possible. The anti-MBP serum was applied in a dilution of 1 : 100. To assess the specificity of the staining for HIBDH two different negative controls were applied. The first antibodies were either omitted or applied pre-absorbed with purified antigen. The preparations were viewed by glycerol immersion optics using a Zeiss fluorescence microscope (IM35) with a Plan-Neofluar 253 objective (Zeiss, Jena, Germany).


Using the RNeasy-kit, total RNA from liver, brain and neural cell cultures was isolated according to the instructions supplied by the manufacturer. The First Strand cDNA synthesis Kit was used following the protocol provided by the manufacturer to reverse transcribe total RNA (1 μg) into cDNA in 20 μL of transcription buffer. The reverse transcription reaction was primed with 0.5 μg oligo(dT)18-adapter primer and was carried out with Moloney Murine Leukemia Virus RT (40 U) at 37°C for 1 h.

Amplification of gene specific fragments by PCR was performed in a total volume of 50 μL of PCR buffer (supplied by the manufacturer) containing 1.5 mM MgCl2, 200 μM of each of the four dNTPs, 1.25 U of HotStarTaq DNA polymerase, and 0.5 μM of both forward and reverse primers. As template was used 1 μL of the cDNA solution just produced. The initial heat activation step (95°C, 15 min) was followed by 30 cycles of three step amplification cycling with 30 s at 94°C, 30 s at 58°C, 30 s at 72°C, and followed with a final incubation at 72°C for 10 min. The PCR products were analyzed on 1.2% agarose gels and had the sizes expected for the respective cDNA sequences. The sets of primers for the amplification of the HIBDH and β-actin cDNAs were: HIBDH, forward: 5′-TGTGTTCTAGGTCAATGGCTTCTA-3′ and reverse: 5′-TTCCTCCTCCCGCAGATACT-3′; β-actin, forward: 5′-GGGTCACAAGGACTCCTACG-3′ and reverse: 5′-GGTCTCAAACATGATCTGGG-3′.


The isolation of HIBDH from bovine liver, carried out in a modification of the method by Rougraff et al. (1988), yielded an apparently homogeneous preparation, as concluded from the single band at a molecular mass of 31 kDa appearing after SDS–polyacrylamide gel electrophoresis analysis (Fig. 2, lane 1). The specific activity of HIBDH isolated from bovine liver had a value 12.3 μmol × min−1 ×  (mg protein)−1, a value comparable to that of 10.5 μmol ×min−1 ×  (mg protein)−1, reported for HIBDH isolated from rabbit liver (Rougraff et al. 1988).

Figure 2.

 Analysis of HIBDH purity and determination of the specificity of the affinity purified HIBDH antibodies by immunoblotting. HIBDH was isolated from bovine liver. The homogeneity of the isolated protein was analyzed by SDS–polyacrylamide gel electrophoresis in a 12% polyacrylamide gel (lane 1). After electrophoresis the gel was silver-stained for the presence of proteins. PL, protein ladder (indicated molecular masses are in kDa). Proteins in homogenates derived from rat liver, brain, and astroglial cells (lanes 2–4, respectively) were separated by SDS–polyacrylamide gel electrophoresis, and subjected to immunoblotting by using affinity-purified HIBDH antibodies. After the incubation with the secondary antibodies against rabbit IgG conjugated with peroxidase, the immunosignal was visualized by an enhanced chemiluminescent detection system.

To investigate the expression of HIBDH in neural cells on the protein level, an antiserum against isolated HIBDH was generated. The two rabbits used for immunization developed antibodies against HIBDH. The antisera titers were determined by ELISA (data not shown). HIBDH antibodies were affinity-purified from the antiserum with the highest titer (1 : 12 000). The affinity purified HIBDH-antibodies proved to be monospecific in an immunoblotting analysis of rat liver proteins (Fig. 2, lane 2). A single band with the expected molecular mass of approximately 31 kDa (Rougraff et al. 1988) was detected. The band did not arise if HIBDH-antibodies pre-incubated with purified antigen were used (data not shown). Western blot analysis of homogenates prepared from rat brain and from APC revealed the presence of HIBDH in both homogenates (Fig. 2, lanes 3 and 4). The appearance of a single band in the positive western blots further stresses the specificity of the HIBDH antibodies.

Besides the generation of a HIBDH antiserum, the isolated HIBDH was utilized also for the enzymatic determination of HIB released by astroglial cells into their culture media. APC generate HIB and release it into the culture medium, no matter whether the culture medium was supplemented with valine (MM-Val and DMEM/FCS) or not (MM). The HIB concentration in media MM and MM-Val had reached a plateau at 1 and 3 days of incubation, respectively. In contrast, in DMEM/FCS the concentration of HIB increased continuously during the entire observation period of 7 days (Fig. 3).

Figure 3.

 Time-dependent appearance of HIB in three different media used for carrying out APC. After 14 days of growth in 90% DMEM/10%FCS the culture medium was replaced by either fresh DMEM/FCS (○), or MM (△), or MM supplemented with 0.8 mM valine (bsl00066). Subsequently, the cells were cultured for varying lengths of time before HIB was quantified in the culture medium. The concentrations are mean values ± SD determined in five replica dishes.

If by degrading valine astrocytes can be the donors of the potential fuel molecule HIB, the question arises, which of the other cells in the brain parenchymum could be consumers of this β-hydroxy acid. Cells capable of utilizing HIB need to express HIBDH, the enzyme that reintroduces the compound into metabolism. To obtain an overview of cells that might be able to process HIB, the presence of HIBDH mRNA was investigated in various types of neural cells in culture by an RT-PCR technique. Indeed, the amplified 911 bp long fragment of HIBDH cDNA was detected in all investigated samples, i.e., in brain and in cultured neuronal, astroglial, microglial, oligodendroglial, and ependymal cells, albeit at different intensities (Fig. 4a). A β-actin specific fragment was amplified as a positive control for the RT-PCR method (Fig. 4b).

Figure 4.

 Establishment by RT-PCR of the presence of HIBDH mRNA (a) in rat brain and in neuronal, astroglial, microglial, oligodendroglial, and ependymal cultures (lanes 1–6, respectively). After isolation, mRNA was reverse-transcribed into cDNA and subsequently PCR was performed using primers for the amplification of a fragment 911 bp in length that indicates the presence of HIBDH mRNA. (b) As a positive control for the functioning of the RT-PCR method, a fragment specific for β-actin was amplified. The numbers at the left margins of (a) and (b) represent lengths (bp) of the molecular size markers (M).

The cells in primary neuronal cell cultures were examined by an immunofluorescent double-labeling technique for the expression of HIBDH and the neuron-specific marker, neurofilament protein 200 (Fig. 5). The staining was mainly located in the small rounded somata but partially extended into the thin, long processes. As these cells could be stained also with antibodies against neurofilament protein 200, they were identified as neurons.

Figure 5.

 Immunofluorescence labeling of neuron-rich primary culture with antibodies against HIBDH (a) and neurofilament protein 200 (b). The phase contrast view corresponding to frames (a) and (b) is presented in frame (c). The representative photomicrograph of a negative control for HIBDH staining (d) was obtained after application of HIBDH antibodies pre-absorbed with purified HIBDH as antigen. The scale bar in frame (c) represents 50 μm and applies to all frames.

The vast majority of the cells in APCs were stained with antibodies against HIBDH (Fig. 6a). Because of the expression of GFAP (Fig. 6b), most of the cells can be identified as astrocytes (Fig. 6c). The presence of a punctate signal for HIBDH distributed in the cytoplasm of the cells suggested that the enzyme was located in a cell organelle, most likely the mitochondrion. Therefore, double labeling was performed with antibodies against a mitochondrial marker enzyme, PDH. The merged view of the stainings for HIBDH and PDH shows colocalization of the two enzymes (Fig. 6e–f).

Figure 6.

 Immunofluorescence detection of HIBDH (a and e) in cells of an APC. Astroglial cells are identified by their expression of GFAP (b). The phase contrast view corresponding to frames (a) and (b) is presented in (c). Frame (d) results from merging the views of frames (a) and (b), with nuclei counterstained with 4′,6-diamidino-2-phenylindole dihydrochloride (blue fluorescence). The scale bar in frame (d) represents 50 μm and applies to frames (a–d). The mitochondrial location of HIBDH and pyruvate dehydrogenase is shown in (e and f), respectively. The merged view of frames (e) and (f) is presented in (g); nuclear staining (g) was carried out by using 4′,6-diamidino-2-phenylindole dihydrochloride. The scale bar in (g) represents 20 μm and applies also to (e and f).

The expression of HIBDH in microglial cells was also investigated immunocytochemically. The microglial cells in microglia-rich secondary cultures were identified by immunofluorescence labeling with anti-CD11b antibodies (Fig. 7b). These cells were also positively stained with anti-HIBDH antibodies (Fig. 7a).

Figure 7.

 Immunofluorescence labeling of HIBDH (a) among the cells in microglia-rich secondary cultures. Microglial cells are identified based on the expression of CD11b (b). The phase contrast view presented in frame (c) corresponds to frames (a and b). The representative photomicrograph of a negative control for HIBDH staining (d) was obtained by application of HIBDH antibodies pre-absorbed with isolated HIBDH as antigen. The scale bar in frame (c) represents 50 μm and applies to all four frames.

Oligodendrocytes in oligodendroglia-rich secondary cultures were identified by immunolabeling with goat antiserum against MBP (Fig. 8c). These cells presented under the phase contrast microscope with a bright halo around their small somata and with long, thin, highly branched processes (Fig. 8b and d). Cells of this morphology were positively stained with anti-HIBDH antibodies (Fig. 8a).

Figure 8.

 Immunofluorescence labeling of HIBDH in cells of oligodendroglia-rich secondary culture (a). As double-labeling with antibodies against HIBDH and myelin basic protein was not possible, oligodendroglial cells were identified by similarity in cellular morphology between cells observable by phase contrast microscopy (b and d) and cells expressing the oligodendrocyte-specific marker myelin basic protein (c). The phase contrast views presented in frames (b and d) correspond to frames (a and c), respectively. All micrographs were taken from cells grown in the same culture dish. The scale bars in frames (b and d) represent 50 μm and apply to frames (a and c), respectively.

Immunofluorescence analysis of ependymal cultures for the presence of HIBDH revealed punctate staining of the majority of the cells (Fig. 9a). This staining is colocalized with the hair-like appearance of the immunofluorescence signal for α-tubulin in the cilia on the surface of ependymal cells (Fig. 9b).

Figure 9.

 Immunocytochemical analysis of HIBDH expression in cells of ependyma-rich primary culture (a). Polyciliated ependymal cells are identified by the presence of cilia, which appear as hair-like structures located on the cellular surface if stained with anti-α-tubulin antibodies (b). The phase contrast view corresponding to frames (a and b) is presented in frame (c). The representative photomicrograph of a negative control for HIBDH staining (d) was obtained by application of HIBDH antibodies pre-absorbed with purified HIBDH as antigen. The scale bar in (c) represents 50 μm and applies to all four frames.


The results presented demonstrate (i) the capacity of astroglial cells in cultures to generate and release the valine metabolite HIB into their culture medium; (ii) the ubiquitous expression of HIBDH among neural cells in culture; (iii) the mitochondrial location of HIBDH. Bixel and Hamprecht (1995) have shown that astroglial cells rapidly remove valine from their culture medium. Catabolism of valine, like that of the other BCAAs, starts with the reversible transamination catalyzed by BCAA aminotransferase and resulting in the cognate branched-chain α-keto acid anion, 2-oxoisovalerate (Fig. 1). Part of the brain pool of BCAAs and branched-chain α-keto acids can participate in maintaining the nitrogen balance in the glutamate–glutamine cycle between astrocytes and neurons by serving as a shuttle molecule for amino groups (Yudkoff et al. 1996; Bixel et al. 1997; Murín and Hamprecht 2008). BCKD irreversibly decarboxylates 2-oxoisovalerate (Fig. 1), thereby irrevocably leading the carbon skeleton of valine into final degradation. This part of the pathway comprises three CoA esters and results in the formation of HIB, after the hydrolytic cleavage of a corresponding CoA ester (Fig. 1). It is this removal of the membrane impermeant CoA moiety from the hydroxyisobutyryl residue that enables HIB to be transported through the inner mitochondrial membrane and the plasma membrane. An inter-organ exchange of HIB is well documented; muscle tissue (Lee and Davis 1986), mammary gland (Wohlt et al. 1977), and heart (Letto et al. 1990) release HIB, which is transported by blood for further metabolism by liver or kidney (Letto et al. 1986). Thus, HIB is considered to be an inter-organ metabolite of the valine degradation pathway (Letto et al. 1986). In analogy, the observed release of HIB from APC into the culture medium raises the question of the further fate of HIB in brain. Is the HIB released from astrocytes subsequently metabolized by neighboring cells or is it transported through the blood–brain barrier into the blood stream?

The ubiquitous expression of HIBDH among neural cells indicates the potential of these cells for further metabolism of HIB. Nevertheless, the source of HIB can be either intracellular (catabolism of valine) or extracellular (import). The transport of HIB through cellular membranes is facilitated by monocarboxylate transporters, which have broad substrate specificities (Bröer et al. 1998). Astrocytes and neurons have been shown to express different isoforms of monocarboxylate transporters (Bröer et al. 1997; for review see Pierre and Pellerin 2005). Knowledge on these transporters in glial cells other than astrocytes appears to be lacking. The enzyme HIBDH is product-inhibited by NADH at high concentration (Rougraff et al. 1988). Thus, astroglial cells, which express HIBDH, would release HIB into the extracellular milieu if the concentration of NADH in their mitochondria were high because of low cellular demand of energy. On the other hand, under conditions of high local energy demand in astroglial cells, HIB would be readily oxidized, thereby being channeled into the third stage of valine catabolism (Fig. 1) and, therefore, no longer being available for release. Release from a cell because of lack of local use is also observed with the other standard fuel molecules, lactate (Dringen et al. 1993), ketone bodies, and α-ketoisocaproate (Bixel and Hamprecht 1995).

Besides HIBDH, there are two more key enzymes involved in the conversion of HIB to the citrate cycle member succinyl-CoA (Sweetman and Williams 2001). The first one, methylmalonic semialdehyde dehydrogenase [2-methyl-3-oxopropanoate:NAD+ oxidoreductase (CoA-propanoylating), EC], catalyzes the conversion of methylmalonate semialdehyde to propionyl-CoA (Fig. 1). The second one, propionyl-CoA carboxylase [propanoyl-CoA:carbon-dioxide ligase (ADP-forming), EC], generates methylmalonyl-CoA, which is enzymatically converted to succinyl-CoA (Fig. 1). Both enzymes are present in brain (Kedishvili et al. 1992; Rodriguez-Melendez et al. 2001), however their detailed cell-type specific expression has not yet been elucidated. By yielding the citrate cycle member succinyl-CoA, the catabolism of valine contributes to anaplerosis in brain. Under these circumstances members of the citrate cycle can be utilized as educts for the biosynthesis of compounds important for brain function such as glutamate and GABA. In ‘pyruvate recycling’ pathways, brain (Cerdan et al. 1990), astrocytes (Sonnewald et al. 1996), and neurons (Cerdan et al. 1990; Olstad et al. 2007) have been shown to be capable of also converting certain members of the citrate cycle to acetyl-CoA. Thus, the fate of valine-derived succinyl-CoA would depend on the metabolic requirements of the tissue.

During pathophysiological conditions induced by liver diseases, e.g. cerebral hyperammonemia/hepatic encephalopathy, patients of low protein tolerance may be treated with a dietary supplement of BCAAs (Charlton 2006). Although the mechanism by which BCAAs exert their beneficial effect remains to be fully elucidated, experimental data suggest that catabolism of BCAAs may help to overcome the inhibition by ammonia of the PDH complex (Zwingmann et al. 2003) and the α-ketoglutarate dehydrogenase complex (Lai and Cooper 1986; Cooper and Plum 1987; Hertz et al. 1987). Indeed, the BCAA catabolites acetyl-CoA and succinyl-CoA, respectively, may enter the citrate cycle after the inhibited enzymatic steps, thus promoting continuation of the cycle during hyperammonemia (Ott et al. 2005). In addition, some members of the cycle may participate in lowering the level of ammonia by becoming withdrawn from the cycle in acting as acceptors of ammonia (Ott et al. 2005). This may happen either indirectly (oxaloacetate) via the transamination reaction involving glutamate, or directly (α-ketoglutarate). The ensuing necessity for anaplerosis could be met by the succinyl-CoA generated in the catabolism of the BCAAs valine and isoleucine, thus preventing ammonia-induced toxicity (Zwingmann 2007).

The coexpression of BCKD (Bixel et al. 2001) and HIBDH among neural cells demonstrates the ability of these cells to catabolize valine imported into brain, thus using it for the generation of energy. This coexpression also requires – and therefore makes likely – the simultaneous presence of the downstream operating enzymes of the catabolic pathway of valine. Like the other standard fuel molecules mentioned above, the HIB generated from valine could be used as fuel material in those cells and cell regions, which due to the lack or low abundance of hexokinase would have to rely on oxidative phosphorylation for the production of ATP (Wilkin and Wilson 1977; Kao-Jen and Wilson 1980; Katoh-Semba et al. 1988). It is not possible to asses the contribution of HIB to the pool of substances that can serve in normal brain as fuel molecules in the oxidative generation of energy. Most likely, the quantitatively most important one of these fuel molecules is lactate in comparison to which HIB could be of limited importance. The maximal extent to which the HIB precursor valine can be utilized as a source of fuel material is set by its uptake into normal brain, which is considerably lower than that of leucine (Oldendorf 1971a,b) and glucose (Oldendorf 1971a). But good evidence exist that brain can utilize proteinogenic amino acids for energy production (Geiger 1958). Obviously further work is needed to elucidate (i) the in situ ability of neural cells to utilize extracellular HIB provided by release from valine processing astrocytes and (ii) the cell-type specific expression of HIBDH in the brain.


The authors gratefully acknowledge the generous gift of antibodies by Drs Peter A. Berg and Reinhild Klein and of thrombin for culturing ependymal cells by Dr Mirna Rapp. Also the valuable discussions by Dr Wolfgang Hirschner and the expert technical help of Barbara Birk are gratefully acknowledged. We also like to thank Claudia Heberle for help with typing the manuscript.