An acyl-CoA dehydrogenase microplate activity assay using recombinant porcine electron transfer flavoprotein

Abstract

Acyl-CoA dehydrogenases (ACADs) play key roles in the mitochondrial catabolism of fatty acids and branched-chain amino acids. All nine characterized ACAD enzymes use electron transfer flavoprotein (ETF) as their redox partner. The gold standard for measuring ACAD activity is the anaerobic ETF fluorescence reduction assay, which follows the decrease of pig ETF fluorescence as it accepts electrons from an ACAD in vitro. Although first described 35 years ago, the assay has not been widely used due to the need to maintain an anaerobic assay environment and to purify ETF from pig liver mitochondria. Here, we present a method for expressing recombinant pig ETF in E coli and purifying it to homogeneity. The recombinant protein is virtually pure after one chromatography step, bears higher intrinsic fluorescence than the native enzyme, and provides enhanced activity in the ETF fluorescence reduction assay. Finally, we present a simplified protocol for removing molecular oxygen that allows adaption of the assay to a 96-well plate format. The availability of recombinant pig ETF and the microplate version of the ACAD activity assay will allow wide application of the assay for both basic research and clinical diagnostics.

1. Introduction

There are nine mitochondrial acyl-CoA dehydrogenases (ACADs) that function in the pathways of fatty acid and branched-chain amino acid oxidation, and two more putative ACADs of unknown function dubbed ACAD10 and ACAD11. ACADs play an important role in human health as evidenced by the eight genetic diseases caused by mutations in this gene family (1,2). Despite advances in newborn screening and early detection, these diseases remain a significant cause of morbidity and mortality in many parts of the world. Additionally, changes in the function the ACADs have been linked to the pathophysiology of non-Mendelian diseases such as obesity, diabetes, heart disease, lung disease, and cancer (3–9). Despite the human health significance related to understanding the ACAD enzymes, the currently available assays for measuring ACAD enzymatic activity are either too technically challenging for routine use or are prone to artifacts that limit their application (10). As such, accurately measuring ACAD activities in complex protein mixtures such as cultured cell lysates or tissue extracts remains a challenge.

The ACADs are FAD-containing flavoenzymes that share a common reaction mechanism (11). In the reductive half-reaction, an ACAD dehydrogenates its acyl-CoA substrate, introducing a double bond between the second and third carbons and transferring a hydride to nitrogen atoms on the rings of the FAD cofactor. This produces what is known as the charge-transfer complex. In the oxidative half-reaction, the charge transfer complex is resolved via interaction with electron transfer flavoprotein (ETF). Passing electrons to ETF re-oxidizes the ACAD and thereby completes the conversion of acyl-CoA substrate to enoyl-CoA product.

Three major strategies have been used to assay ACAD activity. In the first and technically simplest, the flow of electrons is followed spectrophotometrically from the ACAD enzyme of interest to chemical electron acceptor dyes such as ferricenium hexafluorophosphate and dichlorophenolindophenol (DCPIP) (10,12,13). Chemical electron acceptors are inexpensive and simple to use, but their utility for assaying activity in complex biological samples is limited due to their promiscuity as electron acceptors. Additionally, changes in an ACAD that disrupt interaction with ETF, such as certain patient mutations or post-translational modifications, will not be detected with chemical electron acceptors.

In the second strategy, the rate of acyl-CoA substrate conversion to enoyl-CoA product is followed by either HPLC or mass spectrometry in the presence of chemical electron acceptors to drive the reaction to completion (14,15). Focusing on conversion of substrate to product rather than on the flow of electrons provides high specificity and overcomes issues related to promiscuity of the chemical electron acceptors. But again, changes in the ACAD that affect ETF binding will not be detected. Other drawbacks include the need for expensive equipment and for synthesizing the enoyl-CA product species for use as HPLC/mass spectrometry standards, as most enoyl-CoAs are not commercially available.

The third strategy relies on ETF, the natural electron acceptor for the ACAD enzymes. While human ETF only weakly fluoresces, pig ETF is highly fluorescent and the fluorescence becomes quenched as the protein accepts electrons from ACAD enzymes (10,16). Following ETF reduction as a measure of ACAD activity leads to high sensitivity and high specificity since ETF will only accept electrons from the nine ACAD enzymes and from two other unrelated dehydrogenases involved in choline degradation (sarcosine oxidase and dimethylglycine dehydrogenase). The specific activity of any of these 11 enzymes can be measured just by use of the appropriate substrate. However, historically there have been substantial barriers to use of the ETF-based assay. The assay must be performed in an anaerobic environment and utilizes a low throughput cuvette-based fluorometer for detection. Above all, native ETF must be purified from pig liver mitochondria, as there is no commercial source for the protein. Here, we overcome these barriers through development of a bacterial expression purification system for pig ETF and the adaptation of the ETF-based assay to a microtiter plate.

2. Methods

2.1. Expression of pig ETF

The cDNAs for pig ETFα (Accession NM_001244403) and ETFβ (Accession NM_001205279) were cloned into the expression vectors pET21b and pET28a, respectively. The mitochondrial leader sequence was removed from ETFα during cloning (residues 2–19). For ETFβ, the entire protein was cloned, as ETFβ does not have a cleavable mitochondrial leader sequence (17). A plasmid containing GroEL/ES, described previously (18), was co-transformed with the plasmids bearing pETFα and pETFβ into bacterial strain BL21(DE3) for expression. A single colony of triple-antibiotic resistant E coli was used to inoculate 10-ml of LB medium containing ampicillin (100 μg/ml), kanamycin (50 μg/ml) and chloramphenicol (34 μg/ml), which was cultured overnight at 37°C with shaking at 220 rpm. Then, one liter of LB medium containing the same antibiotics was inoculated with the 10-ml overnight culture and grown to an OD₆₀₀ of 1.5, at which time ETF expression was induced with 0.5 mM IPTG. Expression continued for 4 hr at 37°C with shaking at 220 rpm; then the cells were pelleted by centrifugation at 4°C. Cells were washed in ice-cold PBS once and pelleted again. The E coli pellet, weighing ~5g, was flash frozen with and stored at −80°C.

2.2. E coli lysis

The frozen cell pellet was thawed on ice in 10 mM NaPO4, pH 7.5 supplemented with 0.5 mM PMSF, 0.05% Tween 20, and 50 μM FAD (10 ml of lysis buffer per gram of bacteria). Lysis was performed by sonication in a glass cooling cell on ice (550 Sonic Dismembrator Fisher Scientific), using five one-minute bursts with a microprobe at power 4, with a one-minute rest on ice between bursts. The lysate was transferred to ice-cold 50-ml tubes; 5% polyethyleneimine (PEI) adjusted to pH 7.9 with HCl was added to a final concentration of 0.1% and immediately mixed by inversion to precipitate bacterial nucleic acids. The mixture was centrifuged at 27,000 × g for 60 minutes at 4°C. The supernatant was transferred to ultracentrifuge tubes pre-cooled on ice and centrifuged at 200,000 × g at 4°C for 60 minutes. The supernatant was immediately subjected to chromatography.

2.3. Fast protein liquid chromatography (FPLC)

A Pharmacia XK26/40 column (~120 ml column volume) containing CM-Sepharose was preequilibrated with 20 mM NaPO₄, pH 7.5 (Buffer A). After sample loading, the column was washed in 400 ml of Buffer A. Then, a 200 ml gradient from Buffer A to 100% Buffer B (20 mM NaPO₄, pH 7.5 with 1 M NaCl) was applied, with pig ETF eluting as a fluorescent green fraction beginning at ~60% Buffer B. Fractions containing pig ETF were pooled and concentrated by centrifugation in an Amicon Ultra-15 (Merck-Millipore Ltd.) and the buffer was exchanged to 50 mM Tris, pH 8.0 with 10% glycerol. The protein was adjusted to 18.2 mg per ml (300 μM) and stored at −80°C. Pig ETF yield was ~40 mg per liter of induced culture. 2L of culture (~10g of bacterial pellet) could be processed in a single chromatographic run without overloading the column, thereby producing ~80 mg of pure pig ETF in a single run.

2.4. Purification of native pig ETF

Mitochondria were isolated from fresh pig livers by differential centrifugation, snap frozen in liquid nitrogen, and stored at −80°C until time of purification (19). Purification was done by FPLC as previously described (19). Briefly, 200–300 g of pig liver mitochondria were lysed by sonication, cleared by centrifugation, and subjected to ammonium sulfate fractionation. The fraction containing ETF was dialyzed overnight at 4°C and then subjected to anion exchange chromatography, cation exchange chromatography, and hydroxyapatite chromatography. Fractions containing ETF were pooled, concentrated, dialyzed against 10 mM Tris, pH 8.0, supplemented with glycerol, and frozen.

2.5. Characterization of pig ETF

Purity was assessed by SDS-PAGE, loading 2.5 μg per lane, followed by staining with coomassie blue. Protein integrity was determined by absorbance scanning from 250 nm to 600 nm in quartz cuvettes on a Jasco V-650 spectrophotometer (Easton, MD). The ratio of the A₂₇₀ peak (total protein) to the A₄₃₆ peak (FAD) for pure recombinant ETF was 6.3. Western blotting for post-translational modifications followed standard, previously published protocols (20,21). Fluorescence scanning was performed in quartz cuvettes on a Jasco FP-6300 spectrofluorometer (Easton, MD).

2.6. Cuvette-based anaerobic ETF fluorescence reduction assay

Recombinant His-tagged human MCAD, LCAD, and VLCAD proteins were purified from E coli as described (21,22). Substrates were purchased from Sigma (palmitoyl-CoA, stearoyl-CoA) or Toronto Research Chemicals (2,6-dimethylheptanoyl-CoA). For each assay, 700 ul of reaction buffer (50 mM Tris-HCl, pH 8.0, 0.5% w/v glucose) was stoppered into a quartz cuvette and degassed by 10 cycles of argon/vacuum applied by a needle inserted through the rubber stopper. Each cycle consisted of 30 seconds of argon gas and 30 seconds of vacuum. Then, glucose oxidase (Sigma G2133, ~20U/mL final concentration) and catalase suspension (Sigma C30, 0.5 μl/ml final) were added with a Hamilton syringe and the cuvettes kept warm in a 32°C water bath. For each assay, a cuvette was transferred from the water bath into a Jasco FP-6300 spectrofluorometer fitted with a heated cuvette holder (32°C) and fluorescence was zeroed (Ex₃₄₀/Em₄₉₀). Then, enzyme sample (250 ng recombinant ACAD, 200 μg human fibroblast lysate, 40 μg liver mitochondrial lysate, or 15 μg mouse heart homogenate) and pig ETF (2 μM final protein concentration) were added and background fluorescence recorded for one minute. Finally, the reaction was initiated by addition of degassed acyl-CoA substrate at a saturating concentration (25 μM final) and fluorescence was recorded for another minute. The slope and Y-intercept was determined using Jasco Spectra Manager software. Activity in milliunits (mU)— where a mU represents the amount of activity needed to reduce 1 nmole of ETF-bound FAD in 1 minute— is calculated from the following equation:

$$=\frac{-\left({\left(\text{Slope}/\text{Y-int}\right)}^{*}{\text{nmodels ETF}}_{\text{FAD}}{}^{*}\left(60\text{sec}/0.91\right)\right)}{\text{mg}\text{sample}}$$

The nmoles of ETF-bound FAD can be accurately calculated from the A₄₃₆ absorbance peak using the published extinction coefficient of 13,400 (23), or when the preparation is virtually pure as with the recombinant ETF, a final ETF protein concentration of 2 μM in the reaction corresponds to 1 μM of FAD (1 FAD per heterodimer), which is equivalent to 0.75 nmoles of FAD when using a 750 μl reaction volume. The time component in the calculation (60 seconds) is adjusted by the factor of 0.91 to reflect the observation that maximum amount of fluorescence reduction that can be achieved is 91% of the starting fluorescence (10).

2.7. Microplate ETF fluorescence reduction assay

The ETF concentration, glucose oxidase/catalase concentrations, substrate concentration, and sample amounts were the same as for the cuvette-based assay. Only the volume of reaction buffer used in the assay was reduced, to yield a final reaction volume of 200 μl rather than 750 μl. Reactions were measured in a BMG Labtech FLUOstar Omega plate reader set to 32°C, using Ex₃₄₀/Em₄₉₀. The FLUOstar Omega can read fluorescence in 8 samples (one plate column) in a 2-second cycle; therefore, 8 samples can be read in a single run, yielding 30 data points for each sample over a 60-second measurement window. In the FLUOstar Omega, doubling the sample number to 16 in a single run gave unsatisfactory results due to greatly increased intra-assay variability. This variable will need to be empirically tested for different brands/models of plate reader. Background was measured for 60 seconds, then the plate reader was opened, substrate added with a multi-channel pipettor, and the plate immediately read for another 60 seconds. Slopes and Y-intercepts were automatically generated by the plate reader software and used in the equation given above to calculate specific activity in mU. For optimal results, we recommend the use of high-resolution, glass-bottomed, black-walled plates (Cellvis 96–1.5H-N), as well as filtration of the reaction buffer with a standard bottle-top vacuum filter. Filtering the buffer and letting it sit in the bottle under benchtop vacuum for 10 minutes reduced the intra-assay CV by half due to a more stable fluorescence signal.

3. Results and Discussion

3.1. Expression and purification of recombinant pig ETF

Mitochondrial ETF is a heterodimer consisting of α and β subunits (32 and 28 kDa, respectively) encoded by separate genes. The holoenzyme contains a single non-covalently attached FAD cofactor per heterodimer, as well as an adenosine monophosphate (AMP) molecule of unknown function (23). We began our optimization of pig ETF expression by arranging the α and β cDNAs in tandem in a single bacterial expression vector. This strategy was previously successful for expressing recombinant human ETF, which unfortunately has very low intrinsic fluorescence and is not suitable for use as an electron acceptor in the ACAD enzyme activity assay (24). However, no expression of the pig enzyme was observed with a variety of different E. coli strains and culture conditions (data not shown). Therefore, cDNAs for the pig α and β genes were cloned into separate plasmids and co-expressed along with the bacterial chaperone GroEL/GroES, which promotes protein folding (18). This three-plasmid system allowed successful expression of recombinant pig ETF. After optimizing conditions for expression and purification (see Methods), ~80 mg of essentially pure pig ETF was purified from a two-liter E coli culture using a single chromatographic step. A typical elution chromatogram from the CM Sepharose column is shown in Fig 1A, and purity of the final product in Fig 1B. In the traditional cuvette-based anaerobic ETF fluorescence reduction assay, with a reaction volume of 0.75 ml, the 80 mg yield would be enough ETF for about 900 ACAD activity assays.

Figure 1. Expression and purification of recombinant pig ETF.

A) Brief outline of purification scheme and representative elution profile of recombinant pig ETF from the CM-Sepharose column. B) SDS page of two samples of recombinant pig ETF demonstrating purity. Each lane contains 2.5 μg of protein. The gel was stained with Coomassie blue.

3.2. Recombinant pig ETF is purer and more fluorescent than native pig ETF

To evaluate the quality of recombinant pig ETF as a laboratory reagent we compared its physical properties to those of native pig ETF. First, we compared the two ETF proteins by SDS-PAGE. The preparation from pig liver contained two prominent contaminant bands of ~35 and ~42 kDa that were absent in the recombinant preparation (Fig 2A). We next immunoblotted the two ETF preparations for lysine post-translational modifications (PTMs). Both ETF subunits have been reported to be acetylated and succinylated on multiple lysine residues (25–28). Because these PTMs eliminate or reverse the charge of lysine residues, and because ETF: ACAD binding depends upon charge-charge interactions mediated at least in part by surface lysine residues (29,30), we postulated that the presence of these modifications on our pig ETF proteins could affect their performance as electron acceptors in the ETF fluorescence reduction assay. Western-blotting with anti-acetyllysine antibody confirmed that native pig ETFα, but not ETFβ, is acetylated (Fig 2B). Recombinant ETFα purified from E coli was also acetylated, although the modification was barely detectable. We also western-blotted the ETF preparations with anti-malonyllsine, anti-succinyllysine, and anti-glutaryllysine antibodies; neither the native or the recombinant ETF proteins contained these lysine modifications (data not shown). We conclude that the native pig ETF is more prone to potential artifacts introduced by lysine acetylation.

Figure 2. Recombinant pig ETF is purer and more fluorescent than native pig ETF.

A) Coomasie-stained SDS -page gel showing purity of 2.5 μg of recombinant pig ETF (R) versus native pig ETF (N). Native ETF has two prominent contaminating bands. B) Anti-acetyllysine western blot of 100 ng each of recombinant (R) and native (N) pig ETF proteins. C) Absorbance scans of recombinant and native pig ETF proteins. Shown are the peaks associated with the FAD cofactor, with the total spectra in the inset. D) Three-dimensional fluorescence scanning of recombinant and pig ETF proteins produces similar spectra, but fluorescence is of higher intensity in recombinant pig ETF. E) The fluorescence maxima for both proteins was at Ex₃₇₅/Em₄₈₅; the intensity of the maxima was 60% higher for the recombinant pig ETF.

Next, the absorbance and fluorescence spectra of the recombinant and native pig ETF proteins were compared. Absorbance scans of 10 μM solutions of each protein produced similar absorbance spectra (Fig 2C). The ratio of the A₂₇₀ peak (total protein) to the A₄₃₆ peak, which is specific for the FAD flavin ring, was 15% lower in the recombinant protein, reflecting either a reduced presence of non-FAD containing contaminants, greater FAD content, or both. To assess the fluorescence properties of the two ETF proteins, three-dimensional fluorescence scanning was performed on 0.2 μM protein solutions. The 3D spectra were overall quite similar between the two proteins. Both showed a fluorescence maximum at Ex₃₇₅/Em₄₈₅, albeit with more intense fluorescence in the recombinant protein (Fig 2D). Further 2D fluorescence emission scanning of more concentrated 2 μM ETF protein solutions after excitation at 375 nm revealed that the 485 nm emission peak is 60% higher in recombinant pig ETF (Fig 2E). While the Ex₃₇₅/Em₄₈₅ peak is the fluorescence maximum, the original protocol for the anaerobic ETF fluorescence reduction assay used Ex₃₄₂/Em₄₉₆ (10), and our group has historically used Ex₃₄₀/Em₄₉₀ with bandwidths of 5 and 10 nm for excitation and emission, respectively. With Ex₃₄₀/Em₄₉₀, the fluorescence intensity of recombinant pig ETF is 42% higher than in the native protein (3.7 fluorescence units versus 2.6). Altogether, these data indicate that recombinant pig ETF is a physically superior assay reagent over the native protein.

3.3. Recombinant pig ETF outperforms native ETF in the anaerobic ETF fluorescence reduction assay

After demonstrating the superior purity and fluorescence of recombinant pig ETF over native enzyme, the performance of the two ETF proteins was compared in the anaerobic fluorescence reduction assay. To account for the lower purity of the native ETF, the amount of ETF used for each assay was calculated from the FAD absorption at A₄₃₆ rather than total protein concentration. Recombinant human long-chain acyl-CoA dehydrogenase (LCAD) and very long-chain acyl-CoA dehydrogenase (VLCAD) were tested with each ETF protein. Recombinant ETF yielded higher enzymatic activity by 37% compared to native ETF for both human LCAD assayed with 2,6 dimethylheptanoyl-CoA and human VLCAD assayed with stearoyl-CoA as substrate (Fig 3).

Figure 3. Recombinant pig ETF outperforms native ETF in the anaerobic ETF fluorescence reduction assay.

Recombinant human LCAD (A) and VLCAD (B) were assayed for enzymatic activity with both ETF proteins. Bar graphs represent means and standard deviations. *P<0.01.

3.4. Simplifying reaction deoxygenation for the ETF fluorescence reduction assay

While the need to isolate pig ETF from native sources was by far the greatest barrier to widespread use of the anaerobic ETF fluorescence reduction assay, there is still the requirement for expensive quartz cuvettes, a stand-alone fluorometer, and the ability to create an anaerobic reaction environment in the assay cuvettes with alternating cycles of vacuum and argon. The ETF fluorescence reduction assay is performed anaerobically due to the propensity for the reduced, semiquinone form of ETF to react with oxygen, thus lowering the apparent activity of the dehydrogenase being measured (31).

As originally described (10), the assay protocol calls for removal of oxygen by both physical and enzymatic means. First, the reaction mixture is subjected to ten rounds of alternating vacuum and argon gas in a sealed cuvette via a needle. Then, glucose oxidase and catalase are added to the cuvette, which contains glucose in the reaction buffer. Glucose oxidase reacts glucose in a 1:1 molar ratio with O₂ to produce H₂O₂. Catalase then splits each mole of H₂O₂ produced by glucose oxidase into 1 mole of water and 0.5 moles of O₂. Therefore, each linked glucose oxidase/catalase reaction will reduce the amount of oxygen in the reaction mixture by half.

To examine the effects of deoxygenation on the ETF fluorescence reduction assay, recombinant LCAD was assayed with palmitoyl-CoA as substrate in triplicate under three conditions: enzymatic + physical deoxygenation (standard protocol), enzymatic deoxygenation alone, or no deoxygenation. Physical + enzymatic deoxygenation and enzymatic deoxygenation alone produced reactions of similar slope that were linear to at least 90 seconds (Fig 4A). The calculated average values of the triplicate reactions for these two conditions were indistinguishable (Fig 4B). Without deoxygenation, the reactions were not linear and could not be used to calculate the enzymatic activity (Fig 4A). Thus, a simple inexpensive mixture of glucose, glucose oxidase, and catalase is sufficient to stabilize the ETF fluorescence reduction reaction.

Figure 4. Comparison of deoxygenation techniques in the cuvette-based ETF fluorescence reduction assay for LCAD activity.

Triplicate reactions were subjected to oxygen removal with either a combination of argon/vacuum (physical) and glucose oxidase/catalase (enzymatic), just the enzymatic deoxygenation alone, or no deoxygenation. A) Representative ETF fluorescence traces, and B) calculated LCAD activities measured with physical + enzymatic deoxygenation versus enzymatic deoxygenation only. Bar graphs represent means and standard deviations.

3.5. The ETF fluorescence reduction assay in a microplate format

Eliminating the need for physical deoxygenation suggested that the ETF fluorescence reduction assay could be adapted to a microplate format. In a proof-of-concept experiment, we used enzymatic deoxygenation to measure recombinant LCAD activity with 2,6-dimethylheptanoyl-CoA, an LCAD-specific substrate. Medium-chain acyl-CoA dehydrogenase (MCAD) was included as a negative control, as we have observed no reactivity of MCAD toward 2,6-dimethylheptanoyl-CoA in the cuvette-based assay (unpublished observations). Recombinant MCAD or LCAD were mixed in a 96-well microtiter plate with recombinant pig ETF and reaction buffer that contained glucose, glucose oxidase, and catalase to remove oxygen. In the absence of acyl-CoA substrate, all reaction wells exhibited a slightly drifting baseline, with ~2% loss of ETF fluorescence per minute. The slope of the drifting baseline was recorded separately for each well and later subtracted from the final data curves. Addition of 2,6-dimethylheptanoyl-CoA led to a clear quenching of ETF fluorescence in the plate wells containing LCAD (Fig 5A). Specific activities were calculated using the slopes and Y-intercepts automatically generated by the plate reader software. LCAD activity measurements showed good reproducibility (Fig 5B). MCAD, activity was calculated to be about 15% that of LCAD with 2,6-dimethylheptanoyl-CoA. The lower signal-to-noise ratio in the plate assay caused difficulty with measuring near-zero activities, leading to an overestimation of MCAD activity toward 2,6-dimethylheptanoyl-CoA as compared to previous observations in the cuvette assay.

Figure 5. The ETF fluorescence reduction assay in a microplate format.

Activities of recombinant LCAD and MCAD were tested in quadruplicate with the LCAD-specifìc substrate 2,6-dimethylheptanoyl-CoA with recombinant pig ETF in a microplate reader; MCAD is expected to have zero or near-zero activity with this substrate. A) Representative trace of the activities of recombinant MCAD and LCAD against 2,6-dimethylheptanoyl-CoA. After addition of substrate, there is a clear loss of ETF fluorescence for recombinant LCAD but not MCAD. B) Means and standard deviations from quadruplicate reactions of LCAD and MCAD with 2,6-dimethylheptanoyl-CoA. C,D) 40 μg of liver mitochondria extract from wild-type and LCAD−/− mice were assayed with 2,6-dimethylheptanoyl-CoA, using either the anaerobic cuvette-based assay (triplicate assays, panel C) or the microplate assay (quadruplicate assays, panel D). Background activity, calculated from the slope prior to substrate addition, was subtracted from each activity value, which in some cases yielded negative numbers.

To be of optimum value, it is desirable for an enzyme activity assay to be valid with complex biological samples. To test the ability of the plate-based assay to discern between wild-type and LCAD−/− mouse liver mitochondria, the activity of mitochondrial extracts from each strain were measured with 2,6-dimethylheptanoyl-CoA as substrate. First, a mitochondrial preparation from each genotype of mouse was assayed in triplicate with the traditional cuvette-based assay. The wild-type mitochondria had robust LCAD activity while LCAD−/− mitochondria exhibited essentially undetectable activity (Fig 5C). The same mitochondria samples were then assayed in quadruplicate with the microplate technique and enzymatic deoxygenation. The intra-assay variability was moderately higher with the plate-based assay but the relative difference between wild-type and LCAD−/− mitochondria was similar to that measured with the cuvette-based assay (Fig 5D). Note that the absolute activity numbers for wild-type mitochondria are approximately 10-fold different between the cuvette and microplate assays for this experiment. This is because the scales of fluorescence intensity differed between the two machines due to differences in detection systems, gain settings, and the effects of plastic plates, which we address below.

3.6. Glass-bottom microplates significantly improve the ETF fluorescence reduction assay

Assays to this point were performed with standard black microplates with clear plastic bottoms (Greiner Bio-One). While the liver mitochondria experiment in Fig 5C–D gave the same conclusion regardless of whether cuvettes or microplates were used, the quality of the data in the microplate was not as high as judged by the correlation coefficients (R²) produced by line-fitting the experimental data. In the cuvette-based assay the R² for the line fit to an activity trace is typically >0.9, while in the microplate assay with plastic-bottom plates it ranged from 0.6 to 0.7. We hypothesized that the low signal: noise ratio in the microplate assay could be improved with plates specifically designed for high-resolution fluorescence assays. To test this, four black-walled plates were compared: Greiner Bio-One #655090 (plastic), Invitrogen # M33089 (plastic), Cellvis #P96–1-N (glass-bottom), and Cellvis #P96–1.5H-N (glass-bottom high performance). Wild-type mouse liver mitochondria isolated from N=4 mice were compared using palmitoyl-CoA as substrate, which represents the combined activities of LCAD, VLCAD, and ACAD9. The Cellvis #P96–1.5H-N high-performance plate improved the signal: noise ratio, with more intense ETF fluorescence and higher calculated specific activity values (Fig 6A). Then, a set of eight mouse mitochondrial lysates were assayed with palmitoyl-CoA on all four microplates. The assay background signal, calculated from the baseline slope of each reaction prior to addition of substrate, was greatly reduced in glass-bottom plates (Fig 6B). The average R² for the Cellvis high-performance plate was 0.9, thereby nearly matching the data quality of the traditional cuvette-based assay (Fig 6C). When the eight mitochondrial samples were assayed in duplicate on the high-performance plate, the intra-assay variability (CV) was 10.7%.

Figure 6. High-performance glass-bottom plates improve the microplate assay.

A) Wild-type mouse liver mitochondria (N=4) were assayed on four different black-walled fluorescence microplates—two plastic (Greiner Bio-one #655090 and Invitrogen #M33089) two glass-bottomed (Cellvis P96–1-N and Cellvis P96–1.5H-N “high performance”). Additional wild-type mouse liver mitochondria (N=8) were then assayed on the four plates; B) the percentage of background subtraction (baseline activity in absence of substrate divided by activity in presence of substrate) and C) the correlation coefficients of each assay were calculated.

3.7. Application of the optimized microplate assay to tissue and patient cell lysates

The use of ETF as an electron acceptor is particularly advantageous when assaying ACAD activity in whole cell or tissue lysates due to its high specificity for the ACAD family. In a final series of experiments, we tested the ETF microplate assay with mouse heart homogenates and with cell extracts prepared from patient fibroblast cultures.

First, we assayed LCAD activity in mouse heart homogenates with a Cellvis high-performance plate and 2,6-dimethylheptanoyl-CoA as substrate. The average background, defined as the change in ETF fluorescence after mixing with homogenate but before addition of substrate, was low, representing only 7.5% of the measured enzymatic activity in wild-type homogenates. Intra-assay CV of four wild-type heart homogenates measured in duplicate was 10.5%. Linear fitting of the reactions by the plate reader software yielded an average R² of 0.85 ± 0.04. A representative activity trace is shown in Fig 7A. Wild-type heart homogenates displayed robust LCAD activity while LCAD−/− heart homogenates had no detectable activity with 2,6-dimethylheptanoyl-CoA demonstrating the high specificity of the assay (Fig 7B).

Figure 7. Application of the microplate assay to mouse tissue and VLCAD patient fibroblast lysates.

Mouse heart homogenates—15 μg of total protein per reaction—from wild-type and LCAD−/− mice (N=4) were assayed with 2,6-dimethylheptanoyl-CoA as substrate. Shown are representative activity traces (A) and calculated activity data from N=4 mice (B). For wild-type heart homogenates the average R²=0.85 ± 0.04. C,D) Two VLCAD-deficient patient fibroblast cell lines (PI, P2) were evaluated for VLCAD antigen (C) and palmitoyl-CoA dehydrogenase activity (D). Activity was measured with both the anaerobic cuvette method and the microplate method. Bars represent data (mean, standard deviation) normalized to activity measured in a normal control fibroblast cell line. Unnormalized mean specific activity values appear in italics near each bar. *P<0.05, t-test of normalized data from microplate versus cuvette method.

Finally, the microplate assay was tested with two VLCAD-deficient patient fibroblast cell lines, and was further validated by comparing the results to those obtained with the cuvette-based assay. Patient 1 (P1) is essentially null for VLCAD, with no detectable VLCAD antigen, while Patient 2 (P2) retains residual VLCAD enzyme (Fig 7C). Two replicate cultures of each patient cell line were tested for VLCAD activity with palmitoyl-CoA as substrate and compared to a normal control fibroblast cell line. P1, null for VLCAD antigen, was found to have 15% residual palmitoyl-CoA dehydrogenase activity with the cuvette-based assay versus only 2% with the microplate assay (Fig 7D). The 15% residual activity is consistent with small contributions of palmitoyl-CoA dehydrogenase activity from MCAD and ACAD9, both of which are expressed in fibroblasts. The microplate assay may thus tend to be somewhat inaccurate when the measured activities are very low or near-zero, as was seen in earlier assays with both recombinant enzymes and isolated mitochondria. Due to the lower resolution of the fluorescence signal, when measuring very low activities the results are more variable, and the rate of background drift in ETF fluorescence in the absence of substrate sometimes exceeds the rate of ETF fluorescence reduction in the presence of substrate (thus resulting in negative numbers upon calculation). P2, which retained residual VLCAD protein expression, was found to have 27% residual activity with the microplate assay which was in close agreement to the result of 29% obtained with the cuvette method (Fig 7D). Overall, these data indicate the suitability of the plate-based assay for discriminating ACAD-deficiencies from normal controls using fibroblast cultures.

4. Conclusion

The present work breaks down the barriers that have prevented the greater research community from using the ETF fluorescence reduction assay for sensitive, accurate determination of ACAD activities. The 11 dehydrogenases known to interact with ETF can be measured with the assay described herein, simply by changing the substrate (Table 1). When assaying recombinant proteins, any of the listed substrates in Table 1 will provide suitable activity measurements. With tissue lysates and patient samples, some a priori knowledge of which ACADs are present in that tissue or cell type is required, such that a substrate can be chosen for the ACAD of interest that has minimal overlap with other enzymes in the family. This particularly applies to the long-chain enzymes LCAD, VLCAD, and ACAD9 which have highly overlapping substrate specificities. In such cases immunoinactivation can be used to remove the unwanted ACAD from the tissue lysate prior to performing the activity assay (32,33). In conclusion, availability of recombinant pig ETF and adaptation of the ETF fluorescence reduction assay to a microplate format is expected to greatly facilitate research into the ACAD enzymes, their respective metabolic pathways, and diseases associated with those pathways.

Table 1.

Substrate Specificity of the 11 Enzymes that Use ETF as a Redox Partner

Acyl-CoA Dehydrogenase Enzyme	Preferred Substrates	Notes on Assaying Cell/Tissue Lysates
Short-chain acyl-CoA dehydrogenase (SCAD)	C4 to C6-CoAs	Substrates are not specific for SCAD (14,32)
Medium-chain acyl-CoA dehydrogenase (MCAD)	C6 to C12-CoAs; 3-phenylpropionyl-CoA	3-phenylpropionyl-CoA is specific for MCAD (34); LCAD has low activity for C8-CoA (35)
Long-chain acyl-CoA dehydrogenase (LCAD)	2,6-dimethylheptanoyl-CoA; C12 to C18-CoAs	2,6-dimethylheptanoyl-CoA is specific for LCAD (36)
Very long-chain acyl-CoA dehydrogenase (VLCAD)	C12 to C20-CoAs	Substrates are not specific for VLCAD (33,37,38)
Acyl-CoA dehydrogenase-9 (ACAD9)	C9 to C11; C15 to C22	Substrates are not specific for ACAD9 (33,39)
Isovaleryl-CoA dehydrogenase (IVD)	Isovaleryl-CoA	Specific for IVD (19)
Isobutyryl-CoA dehydrogenase (IBD)	Isobutyryl-CoA	Specific for IBD in human, not rodents (19)
Short/branched-chain acyl-CoA dehydrogenase (SBCAD)	2-methylbutyryl-CoA	Also used by IBD (40)
Glutaryl-CoA dehydrogenase (GCDH)	Glutaryl-CoA	SCAD, MCAD, LCAD have very low activity with glutaryl-CoA (41)
Dimethylglycine dehydrogenase (DMGDH)	Dimethylglycine	Specific for DMGDH (42)
Sarcosine dehydrogenase (SDH)	Sarcosine	Specific for SDH (43)

Highlights.

• Pig electron transferring flavoprotein (ETF) can be purified in one step from E coli

• Recombinant ETF outperforms native ETF as a reagent for assaying acyl-CoA dehydrogenases

• Recombinant ETF can be used in a microplate assay for acyl-CoA dehydrogenase activity