**5. Mitochondrial Function in Previously Frozen Specimens: Overview and Commentary on an Updated Approach to Respirometry**

Several attempts have been made to carry out respirometry on cryopreserved tissue samples. In these approaches, samples were collected and frozen in different solutions trying to preserve the structure and integrity of the mitochondrial membrane. However, the outcome of these attempts was variable and not standardized [143–147]. Mitochondrial complexes and supercomplexes can be isolated from frozen samples and separated in native gels. These isolated complexes and supercomplexes preserve their enzymatic activity and more interestingly, mitochondrial supercomplexes respire [45,46]. These works demonstrate that the electron transport chain is not destroyed by freeze-thawing and that previously frozen sample respirometry is feasible [46]. The challenge is to convert the assay performed in isolated mitochondrial supercomplexes to previously frozen samples, without the need to isolate the ETC components.

We have recently published a new approach to measure respirometry in frozen samples (RIFS) that reconstitutes maximal mitochondrial respiration in previously frozen samples [148,149]. Using RIFS, maximal oxygen consumption of the ETC is measured by providing physiological electron donors and acceptors with oxygen consumption as an integrated readout. It is important to consider that in previously frozen samples, the natural provider of electron equivalents, the tricarboxylic acid (TCA) cycle, is not directly feeding the ETC. For that reason, we need to bypass the components that are lost during the freeze-thaw step by providing electron donors directly to the ETC and correct for variable permeabilization of mitochondrial membranes, when necessary. For example, NADH is used rather than pyruvate + malate since NADH, a reducing equivalent that donates electrons to CI of the ETC, does not have a permeability barrier in frozen tissue. The protocol is straightforward in almost all cases except for tissues that contain a high proportion of fibers and collagen such as skeletal muscle or tissues that are rich in membranes such as the brain. In these tissues, an additional protocol step is required for either enzymatic digestion (skeletal muscle) or additional membrane permeabilization (brain) for the reagents and electron donors, particularly NADH, to be accessible to mitochondria.

The RIFS protocol was optimized and validated using the Seahorse XF96 Extracellular Flux Analyzer, but should be suitable for any respirometry platform. The RIFS assay includes injection of NADH to measure maximal CI respiratory capacity, succinate to measure maximal CII respiratory capacity, as well as TMPD to measure CIV activity. TMPD provides electrons to CIV by directly reducing cytochrome c. Inhibitors of CI, CIII, and CIV are used as controls in each assay to determine sensitivity to respiratory chain complexes. RIFS preserves 90–95% of the maximal respiratory capacity in frozen samples and can be applied to isolated mitochondria, cells, and tissue homogenates with high sensitivity. As an example in cells, Figure 2 compares fresh and frozen respiration in human lymphocytes from two donors. In fresh intact and permeabilized cells, these lymphocyte samples demonstrated low inter-individual variability and the expected response to injected compounds in all assays (Figure 2A,B). However, oxygen consumption rates were lower than expected in permeabilized cells possibly due to incomplete permeabilization. RIFS formulation allows for simultaneous measurements of respiration driven by electron entry through CI and CII while determining individual CIV activity. Additionally, CV, or ATP synthase, activity can be measured via ATP hydrolysis. In lymphocyte cell lysates, respiration driven through CI and CII, as well as CIV and CV activity can be measured with low inter-individual variability. Furthermore, CV is sensitive to dose-dependent inhibition with oligomycin (Figure 2C–E).

**Figure 2.** Extracellular flux analysis in fresh and frozen lymphocytes from two donors. Respirometry was measured in fresh (**A**,**B**) and frozen lymphocytes (**C**–**E**). (**A**) Oxygen consumption was measured in intact cells before and after injection of oligomycin (CV inhibitor), FCCP (chemical uncoupler), and antimycin A (CIII inhibitor) and rotenone (CI inhibitor). (**B**) Respiration in permeabilized cells with CII substrates. A final injection of TMPD was included to measure CIV. (**C**) RIFS protocol with succinate and rotenone to drive respiration through CII. (**D**) CV activity measured by media acidification as a result of ATP hydrolysis. (**E**) Oligomycin-sensitivity in the CV measurement. Abbreviations: AA: Antimycin A; ATP: Adenosine triphosphate; FCCP: Carbonyl cyanide p-trifluoromethoxyphenylhydrazone; oligo: Oligomycin; Rot: Rotenone; Succ: Succinate; TMPD: N, N, N', N'-tetramethyl-p-phenylenediamine.

To measure CV activity, the extracellular acidification (ECAR) is measured to monitor the acidification of the experimental medium in response to hydrolysis of injected, saturating levels of ATP, as previously described in Divakaruni et al. [150]. CV activity can be measured in previously isolated, frozen mitochondria as well as tissue and cell homogenates. In this application, the addition of the CV inhibitor oligomycin acts as an important control to determine that the increased ECAR after injection of ATP is a result of ATP hydrolysis and not non-specific acidification.

This ECAR approach will provide a measure of maximal hydrolytic capacity as a readout of CV enzymatic activity. However, it is important to note that ATP hydrolysis, both through ECAR measurements or with spectrophotometric assays, may underestimate CV deficiencies depending on the specific mutated subunit. For example, mutations in the mtDNA-encoded ATP synthase F<sup>o</sup> subunit 6 that cause severe impairment in ATP synthesis may show little or no defects in ATP hydrolysis [151]. Therefore, while CV activity is a useful readout, it cannot replace measurement of ATP synthesis or respirometry in fresh samples or provide information on physiological levels of ATP hydrolysis.

The simplified sample preparation with RIFS, particularly with homogenates, and the 96-well format of the Seahorse XF96 Analyzer allows using significantly less biological material. RIFS does not require isolation of mitochondria or the use of detergents or plasma membrane permeabilizers, which simplifies the methodology and minimizes changes induced by partial or over permeabilization due to the use of detergents. Acin-Perez et al. provide validation in mouse, zebrafish, and human samples [148]. Another advantage of using total tissue lysates is that it accounts for tissue-specific mitochondrial function, where, for example, soleus showed significantly higher respiration than quadriceps, as previously described [152]. In homogenates and cell lysates, respirometry can also be normalized to

mitochondrial content to provide a measure of respiration per functional unit. In addition to traditional methods to measure mitochondrial mass, we have recently described a high throughput MitoTracker Deep Red protocol that requires minimal biological samples that can be run in parallel to RIFS and used to normalize respiration [148,149]. Since differences in mitochondrial content can account for differences in respiration rates in homogenates, the normalization step can differentiate between changes in mitochondrial content and function when homogenate samples or cell lysates are loaded per total protein. Normalization to mitochondrial content is not necessary with isolated mitochondria since respiration in these samples is already normalized to mitochondrial content when loaded per total protein.

The RIFS technique demonstrated that highly oxidative tissues such as heart, brown adipose tissue, and brain showed higher mitochondrial respiration rates per milligram of tissue whereas white adipose tissue, known for its low mitochondrial content, showed the lowest respiration rates. These results suggest that RIFS reveals physiological differences depending on the energy and metabolic demand of the tissue. When comparing RIFS to traditional methods used in frozen samples such as spectrophotometric enzymatic assays, the latter provides information about the maximal activity of the individual complexes or some of them combined, whereas RIFS allows for an integrative measure of the ETC from CI or CII to the natural electron acceptor that is oxygen. Additionally, RIFS often requires less sample to perform the assay compared to enzymatic assays and can show increased sensitivity to specific inhibition of the ETC complexes, as displayed in Figures 3 and 4 where results were obtained with half as much starting material compared with enzymatic assays. In addition, the assay is straightforward where the number of reagents needed is minimal in comparison to spectrophotometric assays. The simple, standardized protocol of RIFS could limit the inter-laboratory variability that has been reported with enzymatic spectrophotometric assays [153].

Snap freezing samples in liquid nitrogen is the best approach for RIFS. However, due to possible delays between sample collection, freezing, and storage for clinical samples, an initial concern was how quickly samples need to be frozen to preserve mitochondrial function. For that reason, different sampling procedures were tested to determine which approach best preserved maximal respiratory rates when samples could not be immediately flash-frozen in liquid nitrogen. Samples placed on ice for up to 3 h after collection as well as samples immediately stored at −20 ◦C preserved the integrity and function of mitochondria ETC complexes suggesting that this approach could be suitable for samples that require a delay in processing [148,149].

#### *5.1. Applications for Respirometry in Previously Frozen Samples*

There are numerous applications where RIFS can be used in preexisting, stored frozen samples. In adipose samples from pheochromocytoma patients, differences in maximal respiratory capacity that were observed in fresh clinical samples were maintained after long-term storage at −80 ◦C [154]. RIFS generated comparable data to those initially obtained in freshly isolated mitochondria including increased respiration through CI, CII, and CIV and increased mitochondrial content. Furthermore, RIFS made it possible to assay all the samples together compared to fresh samples that had to be run one at a time. This application could be particularly important in age-related studies since it enables a longitudinal design in which bioenergetic profiling can be monitored and compared over years across the same patients. RIFS could also be used as a functional test for patients with mitochondrial disease using less invasive samples. We tested the RIFS protocol in lymphoblastoid cell lines (LCLs), an immortalized B lymphocyte cell line, from patients with the mitochondrial disorder Leber optic atrophy (LHON). These cells were derived from patients with a mtDNA mutation in the NADH dehydrogenase subunit 4 gene. Both RIFS and spectrophotometric assays were able to detect impaired CI activity compared with control cells and revealed similar differences between the LHON lines (Figure 3A). Moreover, we were able to detect impaired NADH-driven respiration with half the amount

of sample or the same amount of sample while simultaneously getting information about CII-driven respiration and CIV activity, both of which were, for the most part, similar to the control cells with the exception of LHON LCL line 1 that showed some decrease in maximal CII respiration (Figure 3B). Comparing a citrate synthase assay to the MTDR assay to measure mitochondrial content, the results from both assays were comparable, but again RIFS required less starting material, fewer reagents, and was less time-sensitive than the citrate synthase assay (Figure 3C).

**Figure 3.** Comparison of RIFS and spectrophotometric approaches in Leber optic atrophy lymphoblastoid cell lines. (**A**) CI activity compared with the RIFS protocol. The spectrophotometric CI assay required 0.5 × 10<sup>6</sup> cells/well compared with 0.2 × 10<sup>6</sup> cells/well with RIFS. The data were also normalized to control cells as a more direct comparison of the two techniques. (**B**) CI, CII, and CIV results were obtained using RIFS. (**C**) Mitochondrial content measurements with citrate synthase activity and the MitoTracker Deep Red plate reader assay. Abbreviations: LCL: Lymphoblastoid cell line; LHON: Leber optic atrophy; RIFS: Respirometry in frozen samples.

**Figure 4.** Comparison of RIFS and spectrophotometric approaches to measure Complex I in human lymphocytes treated with rotenone prior to freezing. (**A**) CI activity compared with the RIFS protocol. The spectrophotometric CI assay required 1 × 10<sup>6</sup> cells/well compared with 0.4 × 10<sup>6</sup> cells/well with RIFS. The data were also normalized to control cells as a more direct comparison of the two techniques. (**B**) Compiled CI, CII, and CIV results from two donors were obtained using RIFS. Abbreviations: CI: Complex I; RIFS: Respirometry in frozen samples; ROT: Rotenone.

An additional potential application for RIFS is in monitoring environmental toxicants to which mitochondria are a key target [155]. For that reason, applications targeted to direct screening of compounds leading to mitochondrial toxicity can benefit from RIFS by testing the functional impact of environmental exposures within a large population [155,156]. We tested rotenone exposure in lymphocytes treated with different concentrations of rotenone before freezing and compared the RIFS protocol to a CI activity spectrophotometric assay. Both approaches were able to reveal impairment in CI, but the RIFS assay was more than twice as sensitive and required half as much starting material (Figure 4A). Additionally, given the high oxygen consumption rates with the RIFS protocol, it is likely that less material is required than was used in these proof-of-concept assays (0.4 × 10<sup>6</sup> cells/well). The requirement of less starting material could be particularly important when the amount of sample is limited. In addition to CI, RIFS also provide information on CII and CIV demonstrating that the rotenone-induced defect was limited to CI (Figure 4B).

#### *5.2. Limitations of RIFS*

While the RIFS protocol was not developed to replace traditional respirometry approaches, it is an alternative method when using fresh samples is not feasible due to sample size and availability, or immediate access to instrumentation is not possible, a problem particularly relevant for testing human samples. Table 5 provides a comparison of the advantages and disadvantages of measuring respiration in fresh and frozen tissue samples. Respirometry in fresh samples will undoubtedly provide the most complete information about mitochondrial function and the bioenergetics of a sample given the coupled nature of fresh samples, that is, respiration is coupled to ATP synthesis. However, RIFS can provide a convenient alternative when respiration in fresh samples is not feasible or as a complementary method to determine maximal respiratory capacity in an integrated, yet uncoupled, system.

**Table 5.** Advantages and disadvantages of respirometry measurements in fresh and frozen tissue samples. Abbreviations: ATP: Adenosine triphosphate; TCA: Tricarboxylic acid cycle.


Since freeze-thaw cycles will disrupt the integrity of the mitochondrial membranes and results in the loss of soluble matrix components, an important limitation is that coupled respiration or ATP synthesis cannot be measured with RIFS. These limitations should be considered when using this approach as it will not be sensitive to changes in substrate transport, substrate oxidation, ATP turnover, or TCA cycle flux. However, it has been demonstrated that disease-related phenotype often correlates with a decrease in maximal respiration [119,157–159] and that defects in coupled respiration are mostly related to some mechanism of uncommon drug toxicity [160]. Therefore, there are multiple applications for RIFS when testing fresh samples is logistically impracticable. While the data presented in this review are preliminary as proof of concepts for use of RIFS in circulating blood cells, the data suggest it could complement the current blood-based bioenergetics approaches. Furthermore, as the field continues to search for a biomarker of mitochondrial function, the RIFS approach provides an additional method for measuring maximal mitochondrial respiration and may be useful for non-invasive samples.

#### **6. Future Outlook: Respirometry in Non-Invasive Samples**

In the last decade, there have been substantial improvements in approaches to measure mitochondrial function in human samples. In particular, measurement of mitochondrial function in circulating platelets and leukocytes has elevated translational mitochondrial research and improved the prospects of clinical monitoring of mitochondrial bioenergetics. However, blood collection still needs to be completed in a medical setting and requires fresh processing. This processing can be intensive, especially when isolating sub-populations of PBMCs, which can present challenges in the ability to scale up for clinical trials. Additionally, blood-based bioenergetics do not always demonstrate changes in mitochondrial function observed in primary tissues [161], which suggests that subpopulation testing is an important consideration as subpopulation bioenergetic correlations may differ between tissues and diseases. More research is needed to understand the bioenergetic differences between subpopulations, population variability, and selection of the appropriate cell type so that this approach can be standardized. With the growing number of mitochondrial drug targets being developed and tested in clinical trials, there is a need for further minimally/non-invasive samples that can be scaled up for clinical testing and large-scale population studies. The RIFS protocol has the potential to be used for these applications if it can be optimized from non-invasive samples such as cells from urine or saliva, or buccal mucosa cells collected from cheek swabs, which can be easily obtained and immediately frozen without processing for later testing. Since RIFS requires minimal starting material, it could be modified for use in non-invasive samples that require

minimal processing. While it is still unknown whether RIFS will be a suitable approach for non-invasive samples, there is some evidence that buccal swabs may be an interesting foundation for the development of non-invasive samples.

Buccal mucosa cells represent a potential non-invasive sample that can be used for mitochondrial bioenergetics profiling. Buccal swab testing of electron transport chain activity is possible [162], but has been reported to show poor sensitivity and specificity for mitochondrial disease [41]. Frederiksen et al. showed that there is a significant correlation in mutation load of the common A3243G mtDNA point mutation in four different tissues: blood leucocytes, buccal cells, skeletal muscle cells, and urine epithelial cells [163]. They suggest that genetic testing of buccal cells and urine epithelial cells may provide an important supplement to the diagnosis of mitochondrial diseases, but also note that skeletal muscle was still the most informative tissue with the highest mutation load [163]. Buccal swabs from a 6-year old patient with suspected myoclonic epilepsy and ragged red fibers (MERRF) presented common mtDNA deletions and decreased ETC activities compared with controls [164]. In a study of 40 patients with suspected mitochondrial disease, when comparing enzymatic activities in buccal mucosa cells to muscle biopsies, there was a 77% correlation for an isolated CI deficiency, 86% for combined CI and CIV defects, and 100% in CIV. Overall, detection of mitochondrial impairment was 82% in buccal swabs compared to muscle biopsies [165]. Alterations in buccal mitochondrial DNA and function have also been reported in lung cancer patients and autism spectrum disorder patients (ASD) [166,167]. Additionally, there is evidence that buccal swab samples can be used to monitor mitochondrial enzymatic activity in response to treatment. In a study of ASD children with or without mitochondrial disease, CI, CIV, and citrate synthase activity in response to fatty acid and folate supplementation was measured in buccal extracts [168]. The results suggest that changes in CI and citrate synthase activities can be observed in buccal swabs in response to common interventions and that increased activity in these enzymes was more discernable in the mitochondrial disease subgroup.

While studies measuring mitochondrial function in buccal swabs are still limited, enough data demonstrate mitochondrial functional changes in buccal swabs to warrant further investigation. It is possible that an integrated readout, such as RIFS, would provide both better sensitivity and specificity for buccal samples. However, a potential limitation to buccal swab bioenergetics measurements could be bacterial contamination that decreases measurement specificity to the human samples [169,170]. Removal of bacteria from these samples will likely be an important step to improve sensitivity in the respirometry assay and ensure that outcomes are accurate. While this approach in buccal mucosa is still hypothetical and would require optimization and validation, it could provide a method to rapidly screen maximal mitochondrial respiratory capacity and provide a complementary approach to blood-based bioenergetic measurements and measurements in primary tissues.

#### **7. Summary**

Impaired mitochondrial function has been shown to contribute to the development and progression of various diseases. There is an unmet need to clinically monitor mitochondrial function. However, until now this has been limited due to the labor-intensive immediate processing of the samples. Blood-based bioenergetics have met some of this need but do not exclude the necessity of immediate processing. With the development of RIFS as a complementary technique to measure mitochondrial function in previously frozen samples, the possibility of clinical monitoring of mitochondrial function in samples collected at remote sites, or retrospectively in samples residing in tissue biobanks becomes more apparent. While these emerging approaches require further optimization, in some cases, and validation of population variability and sensitivity to detect differences between healthy and disease patients, they represent a potential step forward in translation bioenergetic monitoring in human samples.

**Author Contributions:** Conceptualization, L.S. and R.A.-P.; investigation, B.S., L.S. and R.A.-P.; writing—original draft, L.S. and R.A.-P.; writing—review and editing, B.S., C.B., L.S., R.A.-P. and O.S.S.; writing—revised draft, C.B., L.S. and R.A.-P. All authors have read and agreed to the published version of the manuscript.

**Funding:** O.S.S. is funded by American Diabetes Association 1-19-IBS-049.

**Institutional Review Board Statement:** Ethical review and approval were waived for this study due to the use of de-identified/commercially available samples.

**Informed Consent Statement:** Patient consent was waived due to the use of de-identified/commercially available samples.

**Data Availability Statement:** Data available upon request.

**Acknowledgments:** The authors would like to acknowledge Tom Grammatopoulos for providing materials. The authors would like to thank Corey Osto and Shai Rambod, as well as Ilan Benador, Daniel Dagan, Ajit Divakaruni, Marc Liesa-Roig, Marcus Oliveira, Martin Picard, and Amy Wang for scientific discussions and valuable advice.

**Conflicts of Interest:** The following competing interests did not influence the writing of this review: O.S.S. is a cofounder of Enspire Bio; O.S.S. and L.S. are cofounders of MyChon Health.

#### **Abbreviations**

Frequently Used Abbreviations: AA: Antimycin A; ADP: Adenosine diphosphate; Asc: Ascorbate; ATP: Adenosine triphosphate; CI: Complex I; CII: Complex II; CIII: Complex III; CIV: Complex IV; CV: Complex V; DH: Dehydrogenase; DNA: Deoxyribonucleic acid; ECAR: Extracellular acidification rate; ETC: Electron transport chain; FCCP: Carbonyl cyanide p-trifluoromethoxyphenylhydrazone; LCL: Lymphoblastoid cell line; LHON: Leber optic atrophy; NADH: Reduced nicotinamide adenine dinucleotide; nDNA: Nuclear DNA; mtDNA: Mitochondrial DNA; MTDR: MitoTracker Deep Red; OCR: Oxygen consumption rate; Oligo: Oligomycin; OXPHOS: Oxidative phosphorylation; PBMC: Peripheral blood mononuclear cell; RIFS: Respirometry in frozen samples; ROS: Reactive oxygen species; Rot: Rotenone; Succ: Succinate; TCA: Tricarboxylic acid cycle; TMPD: N, N, N',N' tetramethyl-p-phenylenediamine.

#### **References**

