INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

OPTICAL METHODS are extremely useful in the analysis of mitochondrial biochemical function (4, 7, 14, 16). Mitochondrial swelling and associated light-scattering changes that occur with numerous metabolic transitions represent a significant problem that can affect absorption and fluorescence spectroscopic measurements (3, 25). Light-scattering and absorption [so called "inner filter" (29)] effects in turbid media have primary and secondary components depending on the conditions: the primary effect is defined as scattering or absorption of the excitation light, and the secondary effect is defined as scattering or absorption of the emitted light.

Fluorescence studies in scattering media, like mitochondrial suspensions, require that inner filter effects be minimized or compensated for before a quantitative analysis can be made. These scattering effects can be reduced by working with a light path geometry, in which excitation and sampling occur from the same side of the chamber (surface fluorescence), which minimizes the overall path length (6, 11). The reflected excitation light can be monitored to compensate for primary scattering (30) but cannot correct for secondary effects. Collecting all the light emitted using an integrating sphere or most of the light using wide slits and close-coupled optics can reduce many of the scattering effects (24). Inner filter effects can be corrected for empirically or with direct calculations if the optical characteristics of the filter can be quantitatively evaluated (5, 15). Finally, internal fluorescent standards have been used to correct for inner filter effects in biological tissues (19-21). This latter approach can potentially compensate for primary and secondary effects (20). Rhodamine B is one of the internal reference compounds that has been used; however, this agent can interfere with fat metabolism in rat liver mitochondria (A. P. Koretsky and R. S. Balaban, unpublished observations).

Although a number of recent mitochondrial NADH fluorescence studies were performed using the surface reflectance signal with or without standards to minimize inner filter effects (1, 9, 10, 20, 23), others have presented experiments uncorrected for scattering with an orthogonal excitation-to-emission light path (8, 22, 28, 31) that would predictably maximize the primary and secondary effects.

The purpose of this work was to evaluate the significance of mitochondrial inner filter effects on NADH fluorescence measurements in a commercial fluorometer and develop a method to correct for the primary and secondary components of these effects. Toward this goal, the use of an internal fluorescent standard with appropriate spectral fitting routines was evaluated as an approach to detect and correct for inner filter effects in porcine heart mitochondria. By use of spectral fitting routines, the magnitude of inner filter effects can be followed as well as corrected for in mitochondrial fluorescence studies.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Porcine heart mitochondria were prepared as previously described (26). Mitochondria were isolated from left ventricular tissue and kept at 4°C, except where noted. Approximately 5-g aliquots of ventricle were finely minced using scissors in 10 ml of buffer A (0.28 M sucrose, 10 mM HEPES, 0.2 mM potassium EDTA, pH 7.2). Typically 100 g of tissue were pooled and kept in a total volume of 250 ml of buffer A. This suspension was then treated with trypsin (type III bovine pancreas trypsin, 0.5 mg/g tissue) for 15 min. The supernatant was poured off the settled mixture and replaced volume for volume with buffer A containing 1 mg/ml BSA and trypsin inhibitor (2.6 mg/g tissue). Supernatant was again poured off the settled mixture and replaced with an equal amount of buffer A with 1 mg/ml BSA. The suspension was homogenized with two passes of a 1-mm clearance and five passes of a 0.2-mm clearance Teflon homogenizer (Thomas Scientific, Swedesboro, NJ). The heart homogenate was centrifuged at 600 g for 10 min in 40-ml aliquots, and the pellet was discarded. The mitochondria in the supernatant were pelleted at 8,000 g for 15 min. The pellets were pooled, rinsed, resuspended, and repelleted three times with 80 ml of buffer A (containing 1 mg/ml BSA for the first 2 rinses) at 8,000 g for 10 min each. On each wash the "buffy" coat was carefully removed from the top of the pellet. Mitochondria were finally resuspended in 4 ml of buffer B (137 mM KCl, 10 mM HEPES, 10 µM tetraphenylphosphonium ion, 250 mM P_i, 2.5 mM MgCl₂, 0.5 mM EDTA, pH 7.2). Proteins and enzymes were obtained from Sigma Chemical (St. Louis, MO).

Mitochondrial content was determined from the cytochrome a oxidase concentration (2): 1 nmol of cytochrome from porcine heart mitochondria was determined to be equivalent to 1 mg of protein (26).

To observe the effects of Ca²⁺ on mitochondria, it was necessary to deplete the mitochondria of endogenous Ca²⁺. The mitochondria were depleted of Ca²⁺ by incubation in buffer C (125 mM KCl, 15 mM NaCl, 20 mM HEPES, 1 mM EGTA, 1 mM dipotassium EDTA, 5 mM MgCl₂, 2 mM P_i, 0.1 mM malate, 10 µM tetraphenylphosphonium ion, with 3.4 mM disodium ATP added fresh on the day of the experiment, pH 7.0) in the absence of additional carbon substrates. Experimental procedures were begun after this Ca²⁺ depletion process, when the membrane potential and NADH fluorescence reached a steady state (~8 min). Free Ca²⁺ was calculated in this buffer on the basis of the solution binding constants previously established (13).

Fluorescence standard loading. A fluorescein-containing fluorescent probe was chosen to provide a high fluorescence efficiency with emission and excitation spectral properties close to NADH. 5(6)-Carboxy-2',7'-dichlorofluorescein diacetate, succinimidyl ester (CF; Molecular Probes, Eugene, OR) was used because it was pH insensitive and gave a stable reference signal after ester cleavage. Mitochondria (15 nmol cytochrome a oxidase/ml) were loaded with CF by addition of 1 nmol CF/nmol cytochrome a oxidase and incubated at 25°C for 20 min in 2.5 ml of buffer B. Mitochondria were pelleted (8,000 g for 10 min), and excess CF was washed off by three successive washes with 30 ml of buffer B. The fluorescence of the third supernatant at 530 nm (340-nm excitation) was <0.5% of the CF signal of the loaded mitochondria.

Optical measurements. Fluorescence emission and excitation spectra were collected using two methods. The first used a luminescence spectrometer (model LS50B, Perkin-Elmer, Norwalk, CT) having a standard cuvette holder with a conventional orthogonal excitation-to-emission light path. In the second method, excitation and emission light were transmitted through a single multifiber-optic bundle (Perkin-Elmer). The bundle abutted a single sapphire window embedded in a custom temperature-regulated chamber in a light-tight box. This arrangement detected the surface fluorescence from the mitochondrial suspension. Mixing was carried out with a magnetic stir bar in both orientations. Mitochondria (500 nmol cytochrome a oxidase/ml) were incubated in solution B (buffer B without EDTA) or in buffer C when the Ca²⁺-depletion process was carried out. Excitation was at 340 nm with a 350-nm cutoff filter and 10-nm slit. Emission scans were collected with a 15-nm slit at 200 nm/min from 360 to 660 nm. All experiments were conducted at 37°C.

Data processing. All linear regressions were performed using resident programs in Sigma Plot (version 3.06) on the basis of the Marquardt-Levenberg algorithm with the following conditions: 20 iterations, a step size of 1, and a tolerance 0.001 (Jandel Scientific, San Rafael, CA). Student's t-tests were also run on Sigma Plot with significant differences taken at P  0.05. Where appropriate, values are presented as means ± SD.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Spectral characteristics of mitochondrial scattering effects. The effects of glutamate, Ca²⁺, and the uncoupler FCCP on the absorbance spectrum of mitochondria are shown in Fig. 1A. Uncoupler and, to a lesser extent, Ca²⁺ caused a wavelength-dependent decrease in absorbance consistent with an increase in mitochondrial volume. The absorbance effect increased with decreasing wavelength, which is shown in the difference spectra presented in Fig. 1, B and C, and is consistent with data previously reported for liver (17).

View larger version (17K): [in this window] [in a new window]   View larger version (16K): [in this window] [in a new window]   View larger version (12K): [in this window] [in a new window]   View larger version (15K): [in this window] [in a new window]   Fig. 1.   Frequency-dependent changes in absorbance (A) of porcine heart mitochondria with various additions. A: absorbance spectra of 0.5 nmol cytochrome a oxidase/ml mitochondria from 400 to 600 nm with no substrate (mito), 5 mM glutamate (Glu), 0.67 mM Ca2+ (calculated 0.7 µM free Ca2+) (Ca), and 125 nM uncoupler (FCCP). B and C: difference spectra between one state and previous state showing effect of Ca2+ and uncoupler, respectively, on mitochondrial absorbance. OD, optical density. D: absorbance spectra of 0.087 nmol cytochrome a oxidase/ml mitochondria from 330 to 350 nm with same additions as A.

1

1

Internal reference. CF was chosen as an internal standard for several reasons. The fluorescence of CF is maximal at 530 nm, which is close to the emission frequency of NADH fluorescence (450 nm) but distinct enough to resolve the two emission spectra (see below). Providing a reference frequency measurement as close as possible to the NADH signal is important to minimize the spectral dependence of the inner filter effects. The fluorescence intensity of CF is pH insensitive and varies <5% over a pH range 6.5-7.5 and is also insensitive to other inorganic ions (20). The CF ester is a lipophilic compound that readily loads into the mitochondria, where the ester is cleaved by nonspecific esterases, allowing the CF to bind to structures within the mitochondria. Its net negative charge also retards its release from the matrix, resulting in a stable fluorescence reference in the same macroscopic compartment (i.e., the mitochondrial matrix) as NADH. This provided a stable fluorescence reference for the entire mitochondrial preparation. Loading mitochondria with CF had no significant effect on the absolute maximum rate of ADP/P_i-driven respiration (P > 0.2) or the ratio of state 3 to state 4 respiratory rates compared with mitochondria not treated with CF: 8.4 ± 1.8 for CF mitochondria and 8.3 ± 2.5 for unloaded mitochondria taken through the incubation and extra washes of the CF-loading procedure. In addition, spectral fitting experiments (see below) revealed no effects of CF on NADH fluorescence intensity or spectral characteristics. Thus CF seemed to provide an appropriate inert internal fluorescence standard.

Spectral fitting. As discussed above, CF was used because its emission frequency was close to but distinct from NADH. However, because there is considerable spectral overlap between the NADH and CF emission spectra, as well as other potential sources of light, spectral fitting was required to isolate each element's contribution to the total mitochondrial emission spectrum. This is illustrated in the spectra shown in Fig. 2, where glutamate/malate (5 mM/5 mM) and ADP (0.7 mM) were serially added to previously carbon substrate-depleted mitochondria. There is a significant overlap between NADH and CF emission. The overlap was necessary to keep the emission frequencies close together, providing an adequate correction for secondary inner filter effects. Interestingly, not all the spectral characteristics seen in Fig. 2 can be attributed to NADH or CF. Below ~420 nm there was a gradual increase in light not associated with NADH or CF that varied with the scattering nature of the medium. On the basis of the slit width and scattering media dependence (Figs. 3), this component was ascribed to the excitation source bleed through (EBT). Changing the excitation filter to 390 nm did not eliminate the contribution of EBT and reduced the signal-to-noise ratio. Minimizing the excitation slit width (Fig. 3A) also decreased EBT but again reduced the signal-to-noise ratio. Because EBT was directly dependent on the sample scattering (Fig. 3) and any attempts to eliminate it reduced the signal-to-noise ratio, EBT was left as a component in the spectrum. EBT also gave a second measure of mitochondrial scattering. The resulting three components, EBT, NADH, and CF fluorescence, were used for spectral fitting.

View larger version (16K): [in this window] [in a new window]   Fig. 2.   Emission spectra of porcine mitochondria. Spectra collected from 5(6)-carboxy-2',7'-dichlorofluorescein diacetate, succinimidyl ester (CF)-loaded mitochondria (0.5 nmol cytochrome a oxidase/ml) fully oxidized (mito) and in state 4 with substrate combination 5 mM glutamate-5 mM malate (G/M) and state 3, 1.3 mM ADP. FL, fluorescence intensity (in arbitrary units). All subsequent figures were generated from orthogonal light path geometry data except where noted.

View larger version (15K): [in this window] [in a new window]   View larger version (15K): [in this window] [in a new window]   Fig. 3.   Effects of turbidity on emission spectra in absence of mitochondria. A: excitation bleed through (EBT) with 40 mg/ml Sephadex at 5-, 10-, and 15-nm excitation slits (ex slit). B: EBT with 0.0167 and 0.033% milk in water.

(1)

(2)

(3)

(4)

View larger version (15K): [in this window] [in a new window]   View larger version (15K): [in this window] [in a new window]   Fig. 4.   Model spectra and linear least-squares fit for mitochondrial emission spectrum. A: isolated spectra of 3 components composing every scan: MEBT is modeled by 20 mg/ml Sephadex; MNADH is a difference spectrum between 0.5 nmol cytochrome a oxidase/ml reduced and oxidized mitochondria in absence of CF; MCF is a difference spectrum between 0.5 nmol cytochrome a oxidase/ml reduced mitochondria with CF and a mitochondrial scan matched for quantity of mitochondria without CF. B: spectral fitting results of 3 components (fit) to a scan of CF-loaded mitochondria (data) with residuals (residual).

View larger version (15K): [in this window] [in a new window]   View larger version (17K): [in this window] [in a new window]   Fig. 5.   Effect of detergent on mitochondrial emission spectrum. A: 0.5 nmol cytotochrome a oxidase/ml oxidized mitochondria loaded with CF in absence of carbon substrate (mito), reduced by addition of 5 mM glutamate and 5 mM malate (G/M), and then solubilized with 0.2% polidacanol (polid) in an orthogonal light path geometry. B: same as A, except with use of surface fluorescence.

View larger version (15K): [in this window] [in a new window]   Fig. 6.   Effect of increased turbidity on NADH and CF fluorescence in absence of mitochondria. Spectra were collected of 3 µM NADH, and 1 µM CF with 0.067 or 0.36% milk was added to increase turbidity. Scans were performed with an orthogonal light path geometry.

View larger version (27K): [in this window] [in a new window]   Fig. 7.   Effect of scattering changes on ICF, IEBT, INADH, and INADH/ICF, where I represents contribution of each component to spectrum. Data are from experiment with various amounts of milk (n = 5). NS, not significant. * P  0.05.

Effect of metabolic perturbations. The effects of several metabolic perturbations on EBT and CF (light scattering) and NADH levels in isolated porcine mitochondria using a conventional orthogonal optical path are shown in Fig. 8. The addition of Ca²⁺ to Ca²⁺-depleted mitochondria caused a decrease in I_EBT and an increase in I_CF, consistent with a decrease in light scattering and an increase in mitochondrial volume.

View larger version (14K): [in this window] [in a new window]   Fig. 8.   Effects of glutamate and Ca2+ on mitochondrial emission spectra. Glu, addition of 5 mM glutamate; Ca, addition of 0.67 mM Ca2+ total (calculated: 0.7 µM free); mito, 0.5 nmol cytochrome a oxidase/ml oxidized mitochondria in absence of carbon substrate.

View larger version (9K): [in this window] [in a new window]   View larger version (10K): [in this window] [in a new window]   View larger version (9K): [in this window] [in a new window]   Fig. 9.   Summary of metabolic transition on model parameters. Three components of curve-fitting program: INADH (A), ICF (B), and IEBT (C). G, addition of 5 mM glutamate; M, addition of 5 mM malate; ADP, addition of 1.3 mM ADP; anox, sample at anoxia; see Fig. 8 legend for definition of other abbreviations. All data were normalized to malate condition. * Significantly different from previous condition.

View larger version (36K): [in this window] [in a new window]   Fig. 10.   Summary of methods for determining NADH fluorescence amplitude. Uncorrected 450-nm values (traditional method, raw450) are compared with NADH quantity determined by curve-fitting program (INADH) and corrected NADH value (INADH/ICF). * Significantly from INADH/ICF obtained under same substrate conditions. a n = 6; b n = 3. All data were normalized to malate condition. See legends to Figs. 8 and 9 for concentrations and definition of abbreviations.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

It is well established that light-scattering changes occur within mitochondrial suspensions during metabolic perturbations. In the present study it was confirmed that associated inner filter effects can influence the measurement of mitochondrial NADH fluorescence in porcine heart mitochondria using a conventional spectrofluorometer.

A strategy using a fluorescence standard within the mitochondrial matrix was presented to compensate for the primary and secondary effects of the inner filter in mitochondrial NADH fluorescence measurements. The CF fluorescent probe was used because its emission and excitation spectra are similar to NADH. The CF 530-nm emission, however, was sufficiently different from the NADH 450-nm emission to resolve the two peaks.

The CF emission should be linearly related to the amplitude of the excitation light. Because CF and NADH are in the same compartment, the amplitude of the CF fluorescence should provide a reasonable compensation for the primary inner filter effects. Emitted light from the CF is similar in wavelength to NADH and will experience nearly identical secondary inner filter effects (Fig. 1A). Because of the frequency dependence of the inner filter effects (Fig. 1A), the use of CF is closer to NADH emission frequencies than previously used probes such as rhodamine B (19), which emits in the red. Fluorescence energy transfer from NADH to CF is possible, because the CF excitation spectrum overlaps the NADH emission spectrum and both are trapped in the mitochondria. However, no evidence for fluorescence energy transfer between NADH and CF was found. For example, in Fig. 9 the addition of malate increased I_NADH, whereas I_CF decreased. This occurred with minimal scattering effects (i.e., constant I_EBT). Thus NADH fluorescence does not significantly support CF fluorescence.

By using a linear least-squares fit with appropriate model spectra, the two major spectral components, NADH and CF, of the mitochondrial fluorescence spectrum can be resolved. This approach used the empirical line shapes (18) of NADH and CF in the mitochondria as model components to fit to the mitochondrial fluorescence spectrum. The use of natural line shapes provides an internal correction for instrument and sample modifications of the spectral components. An example of a sample modification was the 5-nm shift in CF observed within the mitochondria. In addition to NADH and CF, the effects of the scattered excitation light bleeding into the fluorescence spectrum (EBT) needed to be corrected for. Most scattering models such as Sephadex or milk were adequate to simulate mitochondria-induced EBT. In the commercial fluorometer used in these studies, EBT could be eliminated only by reducing the slits to which the signal-to-noise ratio became a serious limitation. In addition, no more information was provided using these narrow slit widths (i.e., spectral resolution) with the rather broad emission characteristic of NADH. By use of the increased signal-to-noise ratio of the larger slit widths and addition of an appropriate model component to the spectral fits, EBT effects on NADH signals were eliminated. Because EBT was scattering dependent, its amplitude also helped confirm the direction of scattering changes with different experimental perturbations. EBT alone could not, however, be used to correct for inner filter effects, since the fit of this component was heavily dependent on light below 420 nm, where the inner filter frequency dependence was nonlinear.

NADH fluorescence could be resolved from EBT and CF fluorescence using a linear least-squares fitting routine. Small residuals were obtained with this approach in these well-defined spectra. Simulations varying the NADH component alone over an order of magnitude suggest errors of <0.1% under idealized conditions. With use of this approach, the emission data could be broken down into three elements, the weighted contribution of which (I_NADH, I_EBT, and I_CF) corresponds to the amplitude of the NADH, EBT, and CF signals.

Several results suggest that I_CF tracks changes in sample scatter: 1) Reduction of scattering with detergent resulted in appropriate directional changes in I_CF with orthogonal light path geometry (Fig. 5A) and surface fluorescence (Fig. 5B). 2) I_CF decreased when the turbidity of the medium was increased in solutions and suspensions with the orthogonal light path geometry (Fig. 9B). 3) I_CF paralleled changes in absorbance in the mitochondria (Figs. 1 and 9B). 4) I_CF was inversely related to the scattered excitation light, I_EBT, in the orthogonal geometry (Fig. 9, B and C).

As discussed earlier, we proposed to use the simple I_NADH/I_CF ratio to correct for inner filter effects. The I_NADH/I_CF ratio remained constant with large variations in sample scattering and associated changes in I_NADH, I_CF, and I_EBT (Fig. 7). These results are consistent with I_NADH and I_CF simply tracking the excitation light amplitude as the result of primary light-scattering effects. Any secondary effects were assumed to be the same for I_CF and I_NADH because of the similar emission frequencies.

Errors in NADH determination approached 20% for Ca²⁺ and substrate additions (Fig. 10) without scattering corrections. This large error suggests that correction for inner filter effects is essential for mitochondrial NADH fluorescence assays. Inner filter effects were observed in orthogonal and surface fluorescence light path geometries, suggesting that light path geometry alone may not be adequate to correct this problem.

With regard to primary or secondary effects, the absorbance changes with Ca²⁺ and uncoupler were 10-fold higher at 340 nm (primary) than at 450 nm (secondary; Fig. 1). These results suggest that primary effects dominated the inner filter observed in this study.

There are numerous limitations to the approach presented. It is critical that the model spectra collected for the linear least-squares fit represent only the elements present. Great care was taken to ensure the purity of the model spectra by using internal difference spectra where possible. The R² and residuals provided an objective measure of the quantity of the models and fit. The spectral fitting technique requires the entire fluorescence spectrum to be collected. A significant amount of time and data storage capacity is also needed. Rapid scanning spectrophotometric techniques may be most appropriate under these conditions, inasmuch as this collects the entire spectrum in one acquisition. For the secondary inner filter effects, the frequency differences between CF and NADH were assumed to be insignificant. For the perturbations evaluated in this study, this seems to be a reasonable assumption. However, there are examples in the literature with highly frequency-dependent inner filter effects (24), which would make the choice of a reference compound difficult.

Absolute NADH concentrations would be available from direct analytic assays after extraction (12, 31, 33) or using absorption in intact suspensions with an estimated extinction coefficient (11). The extraction methods suffer from the need to extract this highly active species and make the analysis of time courses difficult. The absorption methods suffer similar scattering problems, low signal-to-noise ratio, and difficulty in determining the extinction coefficient under the rather unique binding conditions in the mitochondria. Absolute NADH concentrations are difficult to calculate using the fluorescence methods and were outside the scope of this study. Quantitation of NADH with fluorescence is complicated by the ill-defined fluorescence enhancement occurring with NADH in the mitochondrial matrix (11). However, NADH/NAD levels can be estimated by determining the I_NADH/I_CF ratio under maximally oxidized and reduced conditions to provide NAD and NADH levels, respectively. This approach of quantitating the NADH/NAD ratio was tested by comparing the -hydroxybutyrate dehydrogenase (HBADH) equilibrium within the mitochondria with literature values. This was accomplished by varying the acetoactate (AA)-to--hydroxybutyrate (HBA) ratio in state 4 mitochondria with minimal respiratory activity to ensure equilibrium of the HBADH reaction. These data are shown in Fig. 11. The HBADH equilibrium

(5)

where [NADH], [H⁺], [AA], [NAD], and [HBA] are NADH, H⁺, AA, NAD, and HBA concentrations, was determined at the estimated midpoint of the NADH fluorescence level between fully oxidized and reduced conditions. The resulting HBADH equilibrium constant was 3.6 × 10⁹, with the assumption of a matrix pH of 7.7. This compares favorably with literature values for mitochondrial HBADH equilibrium constant of 1-4.9 × 10⁹ at 38°C (27, 32). This suggests that using the maximally reduced and oxidized levels provides a reasonable estimate of the NAD and NADH levels for calculating the NADH/NAD ratio.

View larger version (15K): [in this window] [in a new window]   Fig. 11.   Effect of acetoacetate-to--hydroxybutyrate ratio (AA/HBA) on mitochondrial NADH fluorescence. Mitochondria were incubated at 1 nmol cytochrome a oxidase/ml at 37°C. Endogenous substrates were depleted with 200 nmol of ADP; mitochondria were reduced fully with 10 µmol of HBA, and AA was added by titration to determine AA/HBA at NAD/NADH = 1. INADH/ICF values are shown for mitochondria with various AA/HBA. When 50% of NADH was oxidized to NAD, AA/HBA was estimated to be 0.18.

The computational requirements for this approach were minimal and not viewed as a major limitation. Indeed, such routines could be directly programmed into the spectrophotometer.

The use of an internal fluorescent standard to correct for NADH fluorescence in isolated mitochondria was presented. This approach is applicable to other fluorescence probes in mitochondria (e.g., pH and membrane potential) or any fluorescence study in turbid medium. The approach is independent of the light path geometry but requires that accurate model spectra are collected for the components of interest.