Mitochondrial NADH Fluorescence is Enhanced by Complex I Binding

Mitochondria isolation

Porcine heart mitochondria were isolated according to methods described (34). Briefly, harvested heart was perfused with ~400 mL isolation buffer (0.28 M sucrose, 10 mM N-(2-hydroxyethyl) piperazine-N'-2-ethanesulfonic acid (HEPES), and 0.2 mM EDTA, pH 7.1, 4°C). The perfused heart was weighed, and the left ventricle was dissected free of fat, large vessels, and the right ventricular free wall. The heart section (~ 100 g) was minced in isolation buffer. Trypsin (0.5 mg per 1 g of tissue) was then added, and the tissue was incubated on ice for 15 min. The digestion was stopped by discarding the supernatant trypsin solution and adding isolation buffer containing 1 mg/mL bovine serum albumin (BSA) and 2.6 mg/mL of tissue of trypsin inhibitor. The supernatant was again discarded, and the remaining tissue was resuspended in isolation buffer. The tissue was homogenized with a loose-fitting Teflon homogenizer (two times) followed by a tight-fitting Teflon pestle (five times). The homogenate was centrifuged at 600g for 10 min at 4°C, and the supernatant was decanted and centrifuged at 8000g for 15 min. The pellet was resuspended in isolation buffer. The wash-and-centrifugation step was repeated twice more. The final pellet was stored at −80°C for no more than two weeks before performing electrophoreses. All procedures that were performed were in accordance with the guidelines described in the Animal Care and Welfare Act (7 USC 2142 §13).

Complex I isolation

Complex I isolation protocol was based on the procedures developed for bovine heart mitochondria in (35) and (36) . The only deviation from the published procedure was omission of the gel filtration step, which was not found to increase the enzyme purity in pig heart mitochondria homogenates. Mitochondria were isolated from pig heart as described in (34). Complex I specific activity, as measured by NADH:ferricyanide reductase assay (37), progressed from 2.6 units/mg in the starting material to 26 units/mg in the final preparation.

Native gel electrophoresis

To study NADH fluorescence enhancement by binding to mitochondrial proteins along with classically purified Complex I, native electrophoresis was used to maintain mitochondrial protein complexes in their intact form (38). Blue Native PAGE (BN-PAGE) relies on the binding of Coomassie Blue to the protein complex to enhance the proteins' differential migration in the PAGE process. Clear Native PAGE (CN-PAGE) is similar to BN-PAGE but does not use the Coomassie Blue to enhance the differential surface charge of the proteins; this results in a poorer resolution of the complexes when compared to BN-PAGE (39). However, CN-PAGE offers advantages in the present study, as Coomassie-dye interferes with further fluorescence analyses on the gels due to its strong absorbance in a wide optical range (450–700 nm).

BN-PAGE was performed according to an Invitrogen protocol for NativePAGE Novex Bis-Tris Gel System, based on (38). 3–12% and 4–16% 1 mm bis-tris gels were used. BN-PAGE was performed at 4°C for 1 hour at 150V, then for 1.2 hour at 250 V. CN-PAGE employed the same gels as BN-PAGE, but no blue Coomassie was added either to the sample or to the cathode running buffer.

To avoid the intense optical absorption of blue Coomassie interference with NADH fluorescence, a new technique was developed, referred to as Ghost Native PAGE (GN-PAGE) in this study. The chemically reduced form of blue Coomassie dye was found to be essentially colorless, and was used in GN-PAGE to obtain a better dispersion of protein complex as compared to CN-PAGE. Coomassie Brilliant Blue G-250 (Sigma, St. Louis, #B0770) was dissolved in methanol (1g of G-250 in 33 ml of methanol) and held under an argon atmosphere. Sodium cyanoborohydride (75 mg per 1 g of G-250) was added to this solution, and the reaction was stirred for 1.5 h. Then, an additional amount of sodium borohydride (80 mg) was added to the reaction mixture, which was stirred for another hour to reduce most of the residual colored impurities. During this time, the color changed from a thick dark blue solution to a thin light green. TLC indicated a completed reaction. Reduced G-250 was purified by flash chromatography using an Analogix Intelliflash Workstation. The sample was evaporated on silica and dry-loaded on a SuperFlash Sepra Si 50 SF40–80NP column with a flow rate of 85 ml/min: 0% MeOH (100% CH₂Cl₂) 2 min; to 20% MeOH over 10 min; 20% 10 min; to 33% MeOH over 10 min; to 50% MeOH over next 20 min. The fraction eluting at 25–33 min was collected and combined, and shown to have an R_F of 0.29 on Silica Gel-60 TLC plates run in 3:1 dichloromethane:MeOH. After evaporation of solvent and subsequent lyophilization, the product was a light green solid (43% yield) from the original impure starting material. The reduction of G-250 was confirmed by 1H-NMR spectroscopy and mass spectrometry. The assignments of the reduction sites are listed below:

• Coomassie G-250

  H NMR (CD3OD, 360 MHz) δ 7.81–7.73 (m, 3H), 7.44–7.21 (m, 8H), 7.01–6.96 (m, 5H), 6.80–6.75 (m, 4H), 4.86–4.82 (m, 4H), 4.04 (q, 2H, J = 7.0 Hz), 3.74–3.67 (m, 4H), 3.34 (m, 2H), 1.83 (s, 6H), 1.39 (t, 3H, J = 7.0 Hz), 1.30 (t, 6H, J = 6.8 Hz). MS (m/z): 876.0 [M+Na-H].Na+.

• Reduced Coomassie G-250

  H NMR (CD3OD, 360 MHz) δ7.79–7.68 (m, 4H), 7.37–7.30 (m, 4H), 7.02–6.98 (m, 2H), 6.84–6.75 (m, 6H), 6.54–6.42 (m, 6H), 5.36 (s, 1H), 4.50 (s, 4H), 3.97 (q, 2H, J = 7.0 Hz), 3.44 (q, 4H, J = 6.9 Hz), 2.03 (s, 6H), 1.36 (t, 3H, J = 7.0 Hz), 1.16 (t, 6H, J = 7.0 Hz). MS (m/z): 900 [M+2Na-2H].Na+; (m/z) 878 [M+Na-H].Na+.

The original Coomassie dye showed 20 protons in the aromatic region (δ7.81–6.75) in four multiplets whereas the reduced dye integrated for 22 protons in approximately the same field (δ7.79–6.75). Significantly two protons seen in the multiplet δ4.86–4.82 in the original Coomassie dye, and assigned as one of the CH2 benzyl groups of the benzyl sulfonic acid moiety, was not present in the reduced version of the dye. The other CH2 benzyl group of the second benzyl sulfonic acid moiety in the original dye appeared as a multiplet (probably due to long-range coupling) at δ3.34. However, this peak was also absent in the reduced version of the dye. Instead, in the reduced version of the dye a new peak was seen as a singlet at δ4.50 corresponding to the four protons of the CH2 benzyl groups of the benzyl sulfonic acid moieties, which were now rendered equivalent in the reduced version due to the creation of a plane of symmetry around the central triphenylmethane core. Finally, another new peak appeared in the spectrum of the reduced dye, at δ5.36 (s, 1H) corresponding to the proton added at the triphenylmethane carbon atom upon reduction of the `quinimine-like' group present in the original Coomassie dye. Mass spectral analysis of the Coomassie dye showed an [M + Na]+ molecular ion at m/z 876, as expected, while analysis of the reduced dye showed m/z 878 [M + Na]+ and m/z 900 [M + 2Na]+. Removal of the positive half of the zwitterionic pair on the quaternary nitrogen present in the Coomassie dye, by reduction, allowed the formation of a second sodium salt of the sulfonic acid groups, as observed by the [M + 2Na]+ m/z ion. We provide our suggested structures for the Reduced Coomassie and optical absorbance spectra in supplemental Figure 1.

Reduced Coomassie was found to be `shelf-stable' for at least several months when stored at −80°C in a sealed container under a nitrogen atmosphere. The reduced and purified preparation was used in GN-PAGE at concentrations identical to the blue Coomassie concentration used in BN-PAGE.

UV gel imaging

In preliminary studies we found that both purified Complex I and mitochondria homogenates rapidly oxidize NADH to NAD⁺ in the presence of oxygen (under ambient conditions). NADH oxidation was accompanied by reactive oxygen species (ROS) production as confirmed by Amplex Red (Invitrogen) oxidation (data are not shown). Thus, all the experiments on NADH fluorescence enhancement in the bound state were conducted in the absence of oxygen to maintain NADH levels. An anaerobic environment was created inside a bench top vinyl chamber (Coylab) 78 L ″× 32″ W × 40″ H filled with a nitrogen:hydrogen mix (20:1). Oxygen was eliminated through a catalytic reaction (oxygen content was less then 0.0005%). The samples were inserted into the chamber through an automatic vacuum airlock.

A custom built UV fluorescence imaging setup was used for the direct imaging of NADH fluorescence in gels of both the purified and native forms of Complex I within a vinyl containment system at room temperature. A Crimescope CS-16–500 (SPEX) equipped with a xenon 500W lamp was used to provide a uniform UV illumination using the manufacturers UV filter system (specification 300–400 nm) supplemented with a 350 (±10) nm band-pass filter (Newport, to narrow the excitation bandwidth. For fluorescence detection, a high sensitivity EMCCD camera (iXon, Andor, South Windsor, CT) was used with an L37c Nikkor UV cut-off filter (50% transmission at 380 nm) for fluorescence detection.

To assay NADH fluorescence enhancement, gels were incubated in a buffer (137mM KCl, 10mM HEPES, 2.5mM MgCl₂, 0.5mM K₂EDTA, pH 7.1) with 50–100μM NADH. Incubation times varied from 5 min to 15 hours; in both anaerobic conditions and air at RT. The gels were directly placed on a black lexan plate. Images were collected at near maximum gain (255), with exposure times of 0.05 s to 0.3 s. The low signal to noise of these images reflects the limited amount of protein present, as well as the UV excitation only penetrating the outer layer of the protein and NADH laden gel. Protein content was limited by the native gel system to reasonably resolve proteins.

To estimate the degree of fluorescent enhancement per mole of Complex 1, we needed to estimate the concentration of Complex 1 in a given band. For most of our loading conditions we used 200 μg of mitochondria protein per lane (note that 100 μg was used for some studies to conserve material see Figure legends). We previously determined that the concentration of cytochrome oxidase (Complex IV) in porcine heart mitochondria was 1 nM/mg protein (40). Using the ratio of Complex I to Complex IV of 6 (41), consistent with our 2D gel analysis, we estimate approximately 0.17 nM Complex 1/mg mitochondria protein. Within a 200 μg mitochondria protein/lane we are adding ~0.03 nM Complex 1. The Complex 1 band in the gel was 5mm × 2mm × 1mm or 10 μL corresponding to a Complex 1 concentration of ~3 μM under ideal conditions. We used 50–100 μM NADH to assure saturation of the NADH binding sites (20) and minimize the free NADH background signal in the ge1; thus, the ratio of NADH to Complex I was roughly 10–30:1.

Region of interest (ROI) amplitude measurements were conducted for fluorescence bands using open source software (J-Image). Identical ROI masks were placed over the fluorescence protein band and the nearest gel region to estimate fluorescence enhancement in the gel.

Mass spectrometric protein identification

Native gel protein complex bands were excised and washed, reduced with TCEP (5 mM) and alkylated with IAA (25 mM). In-gel trypsin digestion (300 ng) was allowed at 37°C for 18 h. Tryptic peptides were extracted three times with 5% formic acid followed by 50% ACN at 37°C. LC-MS/MS analysis was performed using a Micromass Q-TOF Ultima Global (Micromass, Manchester, UK) equipped with a PST SureSpray (75 μm i.d. packed C₁₈ tip) system (Elkton, MD). Data were processed using the MassLynx software package (version 4.1) to generate peak list files, prior to an in-house licensed Mascot search at http://biospec.nih.gov (MatrixScience Ltd., London, UK) through which we could compare our data against the NCBInr database using mammalian as taxonomy.

Coomassie interferes with NADH oxidation in Complex I

BN-PAGE is based on Coomassie Blue binding to protein complexes that enhances surface charge differences to effectively isolate functionally intact mitochondrial complexes using electrophoresis (38). However, Coomassie Blue is known to bind at the dinucleotide binding site in enzymes (42) and could interfere with NADH binding as well as the enzyme activity of dehydrogenases. Similar effects have been observed for other enzyme activities in BN-PAGE gels(43). To test for this interference, the effects of both Blue and reduced Coomassie dye were tested on the NADH ferricyanide oxidoreductase assay used in following the purification process. Both Coomassie dye and reduced Coomassie decreased NADH oxidation in mitochondria homogenates in an identical dose dependent manner. The rate of NADH oxidation decreased from control, 2.4± 0.2 units/mg, to 1.4 ± 0.3 units/mg and 0.8± 0.3 units/mg with 0.01 and 0.025%, respectively, of either dye (n=4). These data suggest that these Coomassie dyes are not inert to the NADH oxidation reaction, and could interfere with the binding assays. Thus, the early focus of these studies for the intact complex was CN-PAGE, which provides gel separation without Coomassie dyes.

NADH fluorescence enhancement occurs in CN-PAGE bands associated with Complex I

Coomassie stained CN-PAGE gel is shown in Fig. 1A. The mass spectrometric protein identifications of CN-PAGE gel bands C1–C5 are given in Table 1. After electrophoresis, the gel was immersed in 100 μM NADH solution and incubated at ambient conditions (25 C° and room air) for 5 min, and then imaged in the dark room (Fig. 1B). This “in-gel” assay permitted the monitoring of NADH oxidation as well as NADH fluorescence enhancement (see below) in the same gel system. The dark regions on the bright NADH fluorescent background correspond to the locations where NADH was oxidized to the non-fluorescent NAD⁺ form faster than diffusion could “refill” the areas with NADH from adjacent regions. NADH oxidation occurs in both Complex I protein regions in C1 and C2 bands, but 50 μM DPI, a specific inhibitor of Complex I activity (44), blocked NADH oxidation in the C2 band (Fig. 1C). These data imply that most of the NADH oxidation in the C1 band is caused by DPI insensitive contaminating dehydrogenases such as dihydrolipoyl dehydrogenase, malate dehydrogenase, methylmalonate-semialdehyde dehydrogenase, isocitrate dehydrogenase and peroxisomal alcohol dehydrogenase, which were found in the C1 band using mass spectroscopy. The concentration of these dehydrogenases was apparently much lower in C2, where NADH oxidation was completely inhibited by DPI. We also found that the FMN fluorescence (determined with a 445 nm excitation and 520 nm emission) specifically increased in the regions where NADH emissions decreased, supporting the notion that these bands represent Complex I (see Fig. 2).

NADH fluorescence enhancement was determined by incubating the gels in NADH in the absence of oxygen, to prevent oxidation. Time courses revealed that NADH equilibrated throughout the gel in less than 4 hours. Any fluorescence enhancement due to protein binding should be reflected by a regional increase NADH fluorescence over background. NADH fluorescence enhancement was only detected in the C2 band (see Fig. 1D and 1E). DPI did not alter the NADH fluorescence enhancement, which is consistent with the notion that DPI blocks NADH oxidation via the iron centers in Complex I rather than the binding of NADH. In addition, another assay for NADH oxidation activity, NBT staining (45), showed that NADH was completely inhibited by DPI (see Fig. 1F, 1G).

Using ROI analysis, the NADH fluorescence intensity was 22% (± 4%: n=3 for 200ug protein loads) higher in the C2 band as compare to the rest of the gel (Fig. 1D, 1E). Assuming that one NADH binds to each Complex I, this intensity corresponds to a ~10 fold NADH fluorescence enhancement in the presence of Complex I. This result is consistent with the intact mitochondria NADH lifetime measurements in this preparation, where ~15% of the matrix NADH had a 10 fold EFL from 0.4 (free NADH) ns to 5.7 ns(20).

Ghost Native PAGE results consistent with NADH fluorescence enhancement by mitochondrial Complex I

GN-PAGE employing a colorless reduced form of Coomassie dye as a charge-shifting agent gives a substantially better resolution as compared to CN-PAGE without the intense blue color of Coomassie, which interferes with the NADH fluorescence measurement. NADH fluorescence enhancement was determined in the hypoxia chamber as described above. Similar to CN-PAGE gels, our results demonstrated NADH fluorescence enhancement in two bands, G1 and G2, even though the protein content for those bands was different from CN-PAGE. Fig. 3A, along with Table 1, provides the Ghost Native band locations and assignments made by mass spectrometric analysis.

Both G1 and G2 bands (and only those) contain Complex I proteins. More than 20 out of 46 total subunits of Complex I were identified in each of those bands by mass spectroscopy. The G1 band is a super complex containing other proteins along with Complex I, including cytochrome bc1 complex, cytochrome c oxidase, and ATP synthase F1 complex. Band G2 is a cleaner version of Complex I, with a little contamination from complex V and no proteins from complexes II, III or IV. It is important to note that in the GN-PAGE bands G1 and G2, no other dehydrogenases were found, in contrast to CN-PAGE. Again, with this separation scheme the NADH fluorescence enhancement observed in G1 and G2 bands (see Fig. 3B, 3C, lanes 2 and 3) was correlated with specific Complex I activity NBT staining (Fig. 3D, lanes 2 and 3), and stayed unchanged in the presence of DPI in the incubation media. Direct detection of NADH oxidation in GN-PAGE gels was not observed, most likely due to the direct inhibition of Complex I NADH oxidation by the reduced Coomassie dye, as discussed above. NBT staining is much more sensitive than the relatively insensitive direct detection of NADH oxidation, and was successful in observing the Complex I oxidation of NADH.

Isolated Complex I enhances NADH fluorescence

Complex I, purified from porcine heart mitochondria, is a multi-subunit membrane-bound assembly. The Complex I preparation contains all of the subunits identified in other preparations of the enzyme, and has NADH:ferricyanide oxidoreductase activity. When loaded on CN-PAGE gels, purified Complex I did not enter the gel, probably due to a different microenvironment and associated proteins that contribute to the inherent surface charge of the proteins critical for CN-PAGE separation. However, in GN-PAGE, it migrated similarly to the G2 band from the total mitochondrial preparation, which was essentially identical to purified Complex I except for some complex V contamination. This purified form of Complex I in GN-PAGE gels enhanced NADH fluorescence under anoxic conditions (see Fig. 3, lane 1). These data are consistent with the notion that the Complex I binding of NADH enhances NADH fluorescence.

Isolated Complex I subunits denatured and separated in SDS-PAGE do not demonstrate NADH fluorescence enhancement

To determine whether any single subunit of purified Complex I can demonstrate the same NADH fluorescence enhancing behavior, we denatured Complex I and loaded it onto a conventional SDS-PAGE gel. After electrophoresis, these gels were treated with NADH and imaged using the same methodology as for the native gels. We did not see NADH fluorescence enhancement in any resolved subunit, most likely because the proteins lost their NADH binding capacity after strong SDS denaturing. Fig. 4 shows a Coomassie stained image and a UV image of the gel.