Non-linear optical microscopy is well-suited to provide non-invasive high-resolution 3D images of mitochondrial structure and function within live cells, tissues, and animals (27, 28). Through two-photon excited fluorescence (TPEF), the intrinsic fluorescence of nicotinamide and flavin adenine dinucleotides (NADH and FAD, respectively) can be detected without the aid of exogenous dyes (26, 29). Based on the autofluorescence of NADH and FAD, the optical reduction-oxidation (or redox) state of cells can be quantified as FAD/(NADH+FAD). This optical redox ratio (ORR) has been shown to be significantly correlated with mass spectrometry-based measurements of NAD⁺/(NAD⁺ + NADH), and can thus reveal the specific metabolic pathways engaged within a cell (30). Specifically, an increase in ORR has been attributed to increased oxidative phosphorylation because of the oxidation of NADH to non-fluorescent NAD⁺ and FADH₂ to fluorescent FAD. On the other hand, hypoxia-like conditions that drive a buildup of NADH due to the inability to convert to NAD⁺ and increased glucose catabolism has been shown to decrease the ORR (30, 31). Recent work from separate groups has demonstrated that the optical redox ratio is sensitive to dynamic changes in oxygen consumption and can provide metabolic assessments comparable to those of the Seahorse metabolic flux analyzer (32, 33). The optical redox ratio has been used to create metabolic image maps of key organs (34), such as the heart and brain, identify metabolic changes associated with cancer progression (35, 36), determine cellular response to therapy (37–39), and discover a relationship between metastatic potential and cellular metabolism (32, 40, 41).

Alhallak et al. determined the early metabolic alterations in response to radiation in human A549 lung cancer cells and an isogenic radiation-resistant clone (38). This clone was obtained by repeated exposure of parental radiation-sensitive human lung cancer cell line (A549) to ionizing radiation (25 fractions of 2.2 Gy every 3 days). Although there was no significant difference in ORR of radiation-resistant and -sensitive cells prior to administration of radiation, there was a significant decrease in ORR of radiation-resistant cells 24 h after radiation, which was consistent with Seahorse-based quantification of the normalized oxygen consumption rate (n-OCR) (Figure 1). The observed results indicate that the radiation-resistant cancer cells have decreased levels of oxygen consumption both at baseline and post-radiation and resort to increased glucose catabolism after radiation to potentially reduce ROS-induced toxicity. Interestingly, this radiation-induced decrease in the optical redox ratio was also associated with a large increase in the HIF-1 expression in the radiation-resistant A549 clone.

A subsequent by Lee et al. investigated metabolic changes in response to HIF-1 inhibition to determine if the changes in optical redox ratio post-radiation were indeed mediated by HIF-1 and a mechanism to avoid ROS-induced toxicity (39). They used multiphoton microscopy to determine the ORR of A549-RR prior to and post-treatment with YC-1, an established HIF-1 inhibitor. Treatment with YC-1 for 24 h resulted in a significant increase in the ORR compared with baseline, with a concomitant increase in mitochondrial ROS (Figure 2), a decrease in reduced glutathione and a decrease in glucose uptake (39). These results support the conclusion also reached by Furdui and colleagues who found increased glucose uptake that was utilized within the pentose phosphate pathway (PPP) to maintain the NADPH-glutathione buffer. This buffer helps scavenge radiation-induced ROS and hence promote radiation resistance (23). These results demonstrate the enormous potential of autofluorescence microscopy to not only provide clinically translational biomarkers of cellular response to therapy but also create opportunities for investigating radiation biology in live cells and animals at very high resolution.

Fluorescent lifetime imaging microscopy (FLIM) measures the average time that a molecule spends in an excited state prior to emission. One significant advantage of FLIM over measurements of endogenous autofluorescence is that lifetime is independent of the fluorophore concentration. The lifetime of fluorophores, such as NADH and FAD depend on whether they are free or bound to a protein complex. For instance, the lifetime of NADH autofluorescence is shorter (~0.4 ns) when free and longer (~1 ns) when bound to protein complexes, such as malate dehydrogenase and lactate dehydrogenase while the lifetime of FAD autofluorescence is longer when free and shorter when bound to protein complexes, such as alpha-lipoamide dehydrogenase (42–46). By quantifying the ratio of free to protein-bound NADH and their respective lifetimes, FLIM can be used to identify the metabolic state of cells and tissue (42, 47–49). A recent study investigated the application of FLIM in radiation research (50). Campos et al. first treated human cancer cells and normal oral keratinocytes (NOK) with 10 Gy of radiation and recorded the resultant metabolic changes using FLIM. As early as 30 min post treatment, there was a significant decrease in NADH lifetime of cancer cells while there was no change in NADH lifetime of the NOK cells.

In addition to being the powerhouse of the cell, mitochondria are also critical to cell death pathways. The energy demands of a cell are maintained by a delicate balance between the rate of oxidative phosphorylation, tricarboxylic acid (TCA) cycle activity, structural changes to the mitochondrial network, and mitochondrial biogenesis. Mitochondria are continuously changing their organization through fission and fusion allowing for adaptation to different functional demands (51, 52). This dynamic mitochondrial network is sensitive to cell differentiation as well as oxygen and nutrient availability (30, 53–55). Fission is critical for mitochondrial biogenesis, cell division, and mitochondrial autophagy and manifests as numerous mitochondrial fragments. Fusion helps to maintain functionality through the sharing of proteins, genetic material, and metabolites and leads to the generation of interconnected mitochondria (56). Alterations to fusion-fission dynamics and hence the mitochondrial organization have been shown to be associated with several pathological conditions, including hypoxia-reoxygenation injury (57–59). Hypoxia-reoxygenation has been shown to result in a decrease in mitochondrial fusion and subsequent changes in length and shape of mitochondria (60). Targeting the changes in mitochondrial fusion and fission has been shown to protect cells from the effects of hypoxia-reoxygenation injury (61, 62). These studies of changes to mitochondrial structure in response to hypoxia-reoxygenation injury are highly relevant to radiation therapy due to the similarity in mechanisms generating oxidative stress. Radiation therapy leads to reoxygenation of previously hypoxic cells, thereby triggering a large production of mitochondrial ROS. The NADH autofluorescence images, which are used to calculate the optical redox ratio, can also be used to evaluate mitochondrial organization and specifically, fission and fusion. Specifically, Fourier-based power spectral density analysis of NADH autofluorescence images has been used to compute a metric termed mitochondrial clustering to quantify mitochondrial organization (30, 63). An increase in mitochondrial clustering was found during periods of increased glucose catabolism, such as hypoxia, resulting in more fragmented, or fissioned mitochondria. On the other hand, glutaminolysis was found to be associated with a decrease in mitochondrial clustering or more networked mitochondria (fusion). The same method was used to investigate mitochondrial structural dynamics in human skin in vivo (53). A recent study used an improved image processing method in the spatial domain to rapidly quantify the local fractal dimension (FD) within individual cells in response to radiation therapy (64). This analysis found a significant decrease in FD (or an increase in mitochondrial clustering) of radiation-resistant lung cancer cells between 12- and 24-h post-radiation compared with pre-radiation measurements. There were no significant changes in the radiation-sensitive cell population in response to radiation at any time point (Figure 3). The increased mitochondrial clustering observed here is consistent with the decreased optical redox ratio and increased glucose catabolism observed by Alhallak et al. using the same cancer cells (Figure 1) (38, 39).

Diffuse reflectance or elastic scattering spectroscopy is an optical fiber- based technique for non-invasive interrogation of tissue. DRS uses optical fibers to deliver low-power non-ionizing light from a broad-band light source (400–650 nm) to tissue surface. The incident weak light undergoes multiple scattering and absorption events and is remitted back to the tissue surface as diffusely reflected light. Since the collected light has interacted non-destructively with the tissue, it provides a wealth of quantitative information about absorption and scattering, a combination of which is used for tissue pathology. Using models of light-tissue interaction that simulate the travel of photons within a scattering and absorbing medium, it is possible to quantify the diffusely reflected light and extract meaningful information related to tissue scattering as well as prominent tissue absorbers, such as oxygenated and deoxygenated hemoglobin (65–70). By exploiting the differences in light absorption spectra of oxygenated and deoxygenated hemoglobin, we can quantify the vascular oxygen content in tissue and obtain volume-averaged estimates of hemoglobin concentration. Measurements of vascular oxygenation have been shown to be concordant with microelectrode-based determinations of tissue oxygenation (71, 72) and immunohistochemical measurements of tumor hypoxia (73). Cell nuclei, mitochondria, and collagen are among the major contributors to light scattering in tissue and are known to undergo significant changes during disease progression (74). Taking advantage of these non-invasive and quantitative measurements, DRS has been used in several studies, with an eye toward clinical translation, for early cancer detection (75–77), prediction of response to therapy (78–80), and evaluation of tumor surgical margins (81). Given the importance of tumor oxygenation in radiation therapy, DRS can provide a non-invasive approach to quantify the biological response to radiation. Vishwanath et al. used DRS to longitudinally monitor tumor oxygenation and determine whether vascular oxygenation can identify treatment outcome earlier than tumor growth assays in a murine model of head and neck cancer treated with single dose of 39 Gy radiation. As early as 5 days post-radiation, radiation-responsive tumors exhibited faster and greater increase in vascular oxygenation compared with non-responding tumors (82). A more recent study from the same group found similar large increases in vascular oxygenation in both locally controlled and locally recurring tumors when the radiation dosage was split into five daily doses instead of a single dose. Additionally, the study also found that within the locally recurring group of tumors, a faster increase in reoxygenation during therapy was negatively correlated with recurrence time (83). Diaz et al. recently used DRS to study short-term changes in vascular oxygen saturation and hemoglobin concentration in radiation-sensitive and resistant A549 tumors treated with 4 dose fractions of 2 Gy (84), and also found significantly higher reoxygenation in radiation-resistant tumors 24 and 48 h after treatment (Figure 4).

This study was the first to report changes in reoxygenation kinetics measured in tumors which were established from a matched model of radiation-resistance. A matched model of radiation-resistance allows direct comparison of resistance-related features due to similar genetic background. While further studies are necessary to fully understand the mechanism of reoxygenation in the radiation-resistant tumors, results from other studies conducted using the same matched model of radiation resistance hint at the possibility of reduced oxygen consumption as a possible reason for the appearance of increased vascular oxygenation within the radiation-resistant tumors. Although the studies by Hu et al. (83) and Diaz et al. (84) used different cell lines in formation of tumor xenografts and treated them with different doses of radiation, they both showed that radiation-resistant tumors reoxygenate in response to radiation. These results are in agreement with a clinical study by Dietz et al. that used oxygen-sensing microelectrodes to measure pO₂ in the cervical lymph nodes of head and neck cancer patients and found that increased reoxygenation correlated with poor radiation response (85). This suggests that DRS is a sensitive detector of reoxygenation and can provide valuable information about radiation response.

Raman spectroscopy offers the ability to probe biomolecular changes and visualize the complex molecular heterogeneity directly from cells and tissues (86, 87). Spontaneous Raman spectroscopy relies on the inelastic scattering of light, arising from its interactions with the biological specimen, to quantify the unique vibrational modes of molecules within its native context (88, 89). This exquisite chemical specificity of Raman spectroscopy has been exploited primarily within the realm of early detection of cancers of the oral cavity (90, 91), breast (92–98), cervix (99–101), and the brain (88, 102).

Recent studies have shown the presence of radiation-induced alterations in Raman spectral features and biochemical changes in cell lines with varying radiosensitivity (103, 104). The radiation response of single living cells has been studied to demonstrate dose-dependent changes in spectral features using principle component analysis (105, 106). In a series of human cancer cell lines treated with clinically relevant doses of radiation (<10 Gy), Matthews et al. found radiation-induced accumulation of intracellular glycogen in relatively radiation-resistant breast and lung cancer cell lines (107). Recent Raman spectroscopic studies on ex vivo lung and breast tumor xenografts have also identified elevated levels of glycogen in tumors exposed to a single, high radiation dose of 15 Gy (108, 109). These findings are of interest because separate non-imaging studies have identified a critical role for glycogen synthase kinase (GSK-3β) in the development of radiation resistance (110).

Radiation-induced changes in Raman spectra of excised cervical tumors have been shown to differentiate radiation responders from non-responders while pretreatment Raman spectra were incapable of predicting radiation response (111). In a recent study, Paidi et al. investigated whether radiation induced biomolecular changes detected by Raman spectroscopy could differentiate between radiation-resistant and sensitive tumors (112). They treated radiation-resistant and sensitive human head and neck (HN) and lung tumor xenografts with 2 Gy of radiation twice weekly for 2 weeks and conducted chemometric analysis using multivariate curve resolution-alternating least squares (MCR-ALS) to uncover biomolecular changes in the tumor microenvironment. MCR-ALS recovers the pure spectral profiles of the chemical constituents of the tissue specimen without a priori information of the composition of the specimen (113). Paidi et al. found an increase in lipid, collagen, and glycogen (lung only) levels for both sensitive and resistant lung and head neck tumors that were treated with radiation, with a much larger increase in the lipid-rich and collagen-rich signatures in the radiation-sensitive tumors (Figure 5) (112). Comparison of the treated tumors alone (RS-XT vs. RR-XT) pointed to a significantly higher collagen content in the sensitive tumors compared to their resistant counterparts in both lung and HN models, which could be attributed to radiation-induced fibrosis (114, 115). The lipid results are intriguing due to other studies that have found elevated levels of fatty acid synthase (FASN) in radiation-resistant cells (23). These findings demonstrate that clear spectral distinctions exist between radiation-resistant and sensitive tumors, and that these distinctions are consistent with recent work seeking to uncover the molecular mechanisms of radiation resistance.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.