1.

Introduction

Fluorescence imaging techniques provide both anatomical and functional information of tissue via intrinsic fluorophores or exogenous tagged proteins.¹ These techniques are widely used to probe tissue redox state and energy homeostasis in organs such as the heart,² brain,³ kidney,^4–6 liver,⁷ skeletal muscle,⁸ cervix,⁹ and colon,¹⁰ as well as to diagnose diseases, such as breast cancer tumor localization and oxygenation.^11,12 Furthermore, these techniques have been shown to have a high sensitivity and specificity for discriminating between diseased and nondiseased tissue.¹³

Tissue metabolic state, which is an indicator of cellular oxygen consumption, can be extracted from fluorescence images of intrinsic fluorophores.^14,15 The mitochondrial metabolic coenzymes nicotinamide adenine dinucleotide (NADH) and flavin adenine dinucleotide (${\mathrm{FADH}}_2$) are the primary electron carriers in oxidative phosphorylation. NADH and ${\mathrm{FADH}}_2$ oxidation via the electron transport chain results in the translocation of protons from complexes I, III, and IV across the inner mitochondrial membrane into the mitochondrial inner membrane region. The resulting proton gradient across the inner mitochondrial membrane, along with adenosine diphosphate (ADP) availability, controls the rate of mitochondrial adenosine triphosphate (ATP) synthesis, which accounts for $\sim 90\%$ of ATP production in lung tissue. Thus, a change in redox ratio (RR) is an index of a change in lung tissue bioenergetics.¹⁶

NADH and FAD (oxidized form of ${\mathrm{FADH}}_2$) are autofluorescent and can be monitored without exogenous labels by noninvasive optical techniques.¹⁷ The fluorescence signals of these intrinsic fluorophores have been used as indicators of tissue metabolism in injuries due to hypoxia, ischemia, and cell death.¹⁸ Ranji et al. demonstrated that the ratio of these fluorophores, termed the mitochondrial redox ratio ($\mathrm{RR}=\mathrm{NADH}/\mathrm{FAD}$), is a marker of the mitochondrial redox and metabolic state of myocardial tissue in intact hearts or in vivo situations.^19–22 There is more than one definition for redox ratio, including the normalized redox ratio $[\mathrm{NADH}/(\mathrm{NADH}+\mathrm{FAD})]$;¹⁸ the definition that is used in this study (NADH/FAD) is chosen since the FAD signal in lung tissue, as compared to the NADH signal, is significantly smaller than in other organ tissues such as the heart. Thus, the normalized redox ratio would be less sensitive to a change in mitochondrial redox state than NADH/FAD.

Rapid freeze trapping of organs in liquid nitrogen temperatures preserves the tissue metabolic state. Subsequent low-temperature fluorescence imaging (cryoimaging) is advantageous since it provides high fluorescence quantum yield of NADH and FAD as compared with room temperature, and 3-D spatial distribution of tissue NADH and FAD fluorescence intensities.^21,23,24

Exposure to elevated ${\mathrm{O}}_2$ (hyperoxia) is a common and necessary therapy for adult and pediatric patients with acute respiratory distress syndrome to restore blood oxygen tension (${\mathrm{PO}}_2$) to a level that sustains vital organ metabolic requirements.^25,26 However, sustained exposure to high ${\mathrm{O}}_2$ ($>50\%$) causes lung ${\mathrm{O}}_2$ toxicity injury. This injury, which is the result of enhanced production of reactive ${\mathrm{O}}_2$ species (ROS), i.e., oxidative stress, may further impair lung function and contribute to the very dysfunction that it is intended to alleviate.^27,28

Ischemia-reperfusion (IR) injury of lung tissues is commonly encountered clinically in conditions such as lung transplantation, necrotizing pneumonias, or crush injury to the chest.²⁹ Approximately 1500 lung transplants are performed each year in the U.S.,³⁰ with many times that number lost due to prohibitive ischemic times. In both the above circumstances, the capacity to distinguish the oxidative state of individual lungs would modify management.

Mitochondrial dysfunction is a cardinal feature of hyperoxic and IR lung injury.^31,32 Although much work has been done in cell cultures and tissue homogenates, studies for probing key tissue mitochondrial functions and the effect of oxidant injury in the intact lungs are limited.²⁵ Thus, the objective of this study is to utilize cryoimaging to evaluate the effects of in vivo exposure of rats to hyperoxia or lung IR on lung tissue mitochondrial redox ratio (NADH/FAD). The results of this study will provide a basis to apply optical fluorescent techniques for evaluating the effects of these and other models of lung injury on lung tissue mitochondrial RR in vivo.

2.

Materials and Methods

2.5.

Cryoimager

Low-temperature fluorescence imaging (cryoimaging) provides three-dimensional (3-D) fluorescence images of cryopreserved intact organs, and a higher quantum yield of fluorescence of NADH and FAD as compared to room temperature.^24,35,36 The cryoimager (Fig. 1) collects 3-D fluorescence images of frozen tissue’s intrinsic or extrinsic fluorophores. The instrument consists of an Aqua Exi charge-coupled device (CCD) camera ($Q$-imaging, Aqua Exi, 14 bit, 6.45 μm pixel), a workstation (Dell computer), a mercury arc lamp (200 W, Oriel), an excitation (EX) filter wheel (which provides up to five excitation wavelengths), an emission (EM) filter wheel with synchronized rotation to EX wheel, and a cryostatic microtome. Fluorescence images are acquired with the digital camera ($1392\times 1040$ pixel array) with a 200 mm Nikkor lens (Nikon, Tokyo, Japan). Two motorized filter wheels containing excitation and emission filters are mounted in front of the light source and camera, respectively. The motor-driven microtome sequentially sections frozen tissue at the desired slice thickness while filtered light from the arc lamp excites fluorophores in the exposed surface of the tissue block for up to five distinct fluorophores. The microtome is housed in a freezer unit that maintains the sample at $-40°\mathrm{C}$ during sample slicing and image acquisition. Computer control of the microtome motor and filter wheels as well as image capture and display is accomplished through LabVIEW (8.6 National Instruments).³⁷

Fig. 1

Schematic of Cryoimager. This device sequentially slices the tissue, imaging the surface between each successive slice, in as many as five channels. The images are then displayed and saved to a computer, where they can be processed to create 3-D renderings of the tissue.³³

The excitation band pass filter used for NADH is 350 nm (80 nm bandwidth, UV Pass Blacklite, HD Dichroic, Los Angeles, CA) and for FAD is 437 nm (20 nm bandwidth, 440QV21, Omega Optical, Brattleboro, VT). The emission filter for NADH is 460 nm (50 nm bandwidth, D460/50M, Chroma, Bellows Falls, VT) and for FAD is 537 nm (50 nm bandwidth, QMAX EM 510-560, Omega Optical, Brattleboro, VT). At each slice, the camera records fluorescence images of the tissue block in pixel dimensions of $22\times 22\mu \mathrm{m}$. The resolution in the $z$ direction of microtome slices can be as small as 10 μm. For this study, a resolution of 25 μm was used in the $z$ direction, which resulted in $\sim 1000$ $z$-slices per lung. Images are acquired with exposure times of 1.5 sec for FAD and 1 sec for NADH. Accounting for the time to rotate the filter wheels, as well as moving and slicing the sample, results in 3 to 4 h (depending on the size of the lung) for imaging the whole rat lung.

3.

Results

Figure 2 shows the 3-D rendering of NADH and FAD fluorescence signals and their ratio ($\mathrm{RR}=\mathrm{NADH}/\mathrm{FAD}$) from representative lungs of each of the four groups (normoxic, $\text{normoxic}+\mathrm{KCN}$, hyperoxic, and IR). As expected, lung treatment with KCN, which inhibits complex IV and hence reduces the chain, increased the NADH signal and decreased the FAD signal and as a result increased RR.

Fig. 2

Representative 3-D reconstructions of lungs from each of four groups (normoxic, $\text{normoxic}+\mathrm{KCN}$, hyperoxic, and IR). From left to right, images shown are NADH, FAD, and mitochondrial redox ratio (RR).

Figure 3 shows histograms of RR for the four sets of lungs in Fig. 2. For the IR lung, the histogram in Fig. 3 is that of both the injured lobe (left lobe) and normal lobe like the other sets. Despite the localized nature of the IR insult, all areas of the lung show a difference when compared to a normal lung. For each histogram, the mean value was calculated as described in Sec. 2. In Fig. 3, the counts of each bin have been normalized to the total number of pixels in the lung. As a result of this normalization, the value of each bin corresponds to the percent of voxels in the lung with intensities falling within the given range. In this manner, the histogram can be thought of as a scaled probability density function of mitochondrial redox ratio intensities for a lung. The mean values of these histograms suggest a more reduced mitochondrial redox state for normoxic lungs treated with KCN, and more oxidized mitochondrial redox state for hyperoxic and IR lungs as compared to normoxic lungs.

Fig. 3

Histogram distribution of Redox ratio for the four lungs displayed in Fig. 2.

Figure 4 shows the $\text{average}\pm \mathrm{SE}$ (standard deviation over the number of samples) of the mean values of the redox ratio histograms for all four groups of rats. Each lung was calibrated to correct for day-to-day variations as described in Sec. 2. In this figure, only the injured lobe of the lung was considered when generating the histograms of the IR lungs. Inhibition of complex IV with KCN increased the mean value of the RR histogram by 30% compared to normoxic lungs. On the other hand, hyperoxia and IR decreased the mean values of the RR histograms by 23% and 28%, respectively, compared to normoxic lungs.

Fig. 4

Bar graph showing the average values and standard errors of the mean value of the histogram of the mitochondrial redox ratio for each of the four groups of lungs. The results show a difference between normoxic (NRM) and $\mathrm{NRM}+\mathrm{KCN}$ (${}^{\mathrm{*}}p<0.005$), NRM and hyperoxic (HYP) (${}^{\mathrm{**}}p<0.01$), and NRM and IR (${}^{\mathrm{***}}p<0.005$). $p$ values were obtained from post-hoc testing of a balanced one-way ANOVA using Tukey’s honestly significant difference criterion.

Figure 5 displays a bar graph plot comparing the contralateral (nonischemic) lobe to the injured lobe of lungs subjected to ischemia-reperfusion. The figure shows that both NADH and FAD signals were decreased for the injured lobe as compared to the nonischemic lobe. However, the mitochondrial redox ratio was only minimally impacted.

Fig. 5

Bar graph plot comparing the average values and standard errors of the mean value of the normal lobe to the IR lobe in IR lungs. The results show a difference between nonischemic (contralateral) and injured ischemic lobe in NADH (${}^{\mathrm{*}}p<0.01$) and FAD (${}^{\mathrm{**}}p<0.05$). $p$ values were obtained from a paired one-tailed student’s t-test.

4.

Discussion and Conclusion

The results of this study demonstrate the utility of cryoimaging for evaluating the redox status of tissue mitochondrial coenzymes NADH and FAD in intact lungs. The redox ratio (RR), NADH/FAD, is an index of lung tissue mitochondrial redox state, which is an important determinant of mitochondrial bioenergetics.

The results revealed that both chronic hyperoxia and IR injury decreased lung RR due to mitochondrial dysfunction caused by oxidative stress. Although these two injury models produce the same mitochondrial dysfunction and caused oxidative stress, it is possible that they might be distinguished due to the heterogeneity of the IR injury. However, the present goal of this study is to quantify oxidative lung injury regardless of cause.

Previously, we demonstrated that rat exposure to 85% ${\mathrm{O}}_2$ for seven days decreased complex I activity by 50% and increased complex III and IV activities by 56% and 90%, respectively, as compared to normoxic lungs.³³ These changes would be expected to lead to a more oxidized chain downstream from complex I and hence could account for the decreased RR measured in this study.

Prediction of hyperoxia-induced lung injury is difficult because the susceptibility of individuals to hyperoxia is quite variable. Serial surface fluorescence measurements that can be obtained from catheter probes inserted through tube thoracostomies could provide critical real-time information regarding the development of oxidative lung injury over days in patients at risk.

The IR model also causes a depression in complex I activity, which would also be expected to oxidize the chain downstream from complex I and hence could account for the decrease in RR.²² Our results show that RR decreased equally for both the ischemic and nonischemic lungs. This suggests that the IR injury was not limited to the ischemic lung. This is consistent with our previous results, which demonstrated an increase in caspase 3 activity and influx of neutrophils in both lungs in this IR model,³¹ clearly demonstrating bilateral injury. The anesthesia “preconditioning” effects on the IR lungs tissue caused by isoflurane are minimized after a 24-h recovery period following the surgery. The left (ischemic and reperfused) and right (nonischemic) lungs would exhibit identical preconditioning effects if there are any.

Because IR is associated with alveolar hemorrhage^31,32 and in vivo estimates of RR would unavoidably be acquired in the presence of blood, we determined the effect of blood on RR as follows. We conducted cryoimaging for one control lung that was not washed free of blood and compared its NADH, FAD, and RR to those of control lungs that were washed of blood. Figure 6 demonstrates the effect of having blood present in the tissue, as would be the case for any in vivo study. The overall fluorescence signal decreased in both channels; however, the RR change is relatively small. This result is consistent with the results from a study by Chance et al. in which they showed that perfusion of isolated organs (e.g., liver) with perfusate containing red blood cells decreases NADH and FAD signals but does not change the redox ratio.²¹ They concluded that the RR in the presence of blood is still a faithful indicator of oxidation-reduction and can compensate for interfering factors such as light scattering and hemoglobin that exist under in vivo conditions.²¹ Several methods exist to correct for these factors, with the most commonly used technique involving normalizing the fluorescence to a reflectance measurement taken at each excitation wavelength.^39,40

Fig. 6

Mean values of NADH, FAD, and RR histograms of two groups of normoxic lungs plus the standard deviation of the mean value for the first group. For the first group (washed lungs) the blood was washed out of the lungs ($n=4$) by perfusing with Krebs bicarbonate solution. For the second (lung containing blood), the lung ($n=1$) was not perfused and was frozen immediately after it was removed from the chest.

Cytosolic NADPH, which has the same fluorescence characteristics as NADH, could be contributing to the signal attributed to NADH in this study. However, Chance et al. demonstrated that the fluorescence signal originates mostly from NADH in the mitochondria and the contribution of NADPH—present in cytosol—is very small.⁴¹ This is consistent with the fact that the quantum yield for NADH is much higher than NADPH (2.5 to 1.25), and its concentration is five times larger than NADPH.^42–44 Thus, NAPDH contribution to the NADH signal and the change in the NADH signal due to hyperoxia or IR in this study is assumed to be small.

Other endogenous fluorophores that may contribute to the NADH signal include collagen and elastin, which are present in the tissue.⁴⁵ However unlike NADH, collagen and elastin contribution would not be expected to change with variations in mitochondrial redox state.^45,46

The current study demonstrates the utility of RR to detect lung mitochondrial oxidant injury under ideal conditions that optimize the quantum yields of NADH and FAD and minimize the effects of confounding factors such as blood. We have not evaluated the effect of temperature on the quantum yields of NADH and FAD. However, Chance et al. addressed this question and demonstrated a high correlation between the RR values in room temperature compared to cryogenic temperatures.²¹ In addition, we have developed a fluorometer to measure lung-surface NADH and FAD signals from an isolated perfused lung at 37°C. Preliminary results using this system in an isolated perfused rat lung revealed that lung treatment with the complex IV inhibitor potassium cyanide (KCN) increased the NADH signal and decreased the FAD signal, and as a result increased the RR in close correlation with that observed in cryoimaging results (Fig. 4). This suggests that the RR is less sensitive to temperature than the quantum yields of NADH and FAD.

The results of this study support the utility of optical imaging of NADH and FAD signals to evaluate lung tissue mitochondrial redox state. While optical determination of RR in tissues such as brain and myocardium is accepted,^19,20,39 the applicability of these methodologies in examination of the redox state in lungs, where the density of mitochondria is much lower, has not been established. The potential clinical importance of real-time optical imaging of lungs in patients with critical illness, patients on high ${\mathrm{O}}_2$, or patients with IR lung injury secondary to lung transplant, chest contusions, is great. Reliable fluorescence determination of RR could be adopted in the same fashion that near-infrared spectroscopy (NIRS) is gaining favor as a noninvasive measure of tissue oxygenation in critically injured patients.⁴⁷ While NIRS is an indirect measure of tissue oxygenation, NADH and FAD data provide information regarding tissue redox and mitochondrial bioenergetics, a truer and more sensitive early measure of organ function. Because NADH and FAD signals can be detected through fiber optic probes placed on the surface of the lung, RR data could be obtained either intraoperatively or through tube thoracostomies (frequently placed for clinical indications in patients with severe lung injury). Our studies support the capacity of fluorescence imaging to detect pulmonary oxidative injury, and set the stage for in vivo studies along with adaptation of the methods (use of reflectance measurements) required to translate this approach to clinical arenas.