Defining Mitochondrial Cristae Morphology Changes Induced by Aging in Brown Adipose Tissue

Mitochondria are required for energy production and even give brown adipose tissue (BAT) its characteristic color due to their high iron content and abundance. The physiological function and bioenergetic capacity of mitochondria are connected to the structure, folding, and organization of its inner-membrane cristae. During the aging process, mitochondrial dysfunction is observed, and the regulatory balance of mitochondrial dynamics is often disrupted, leading to increased mitochondrial fragmentation in aging cells. Therefore, it is hypothesized that significant morphological changes in BAT mitochondria and cristae will be present with aging. A quantitative 3D electron microscopy approach is developed to map cristae network organization in mouse BAT to test this hypothesis. Using this methodology, the 3D morphology of mitochondrial cristae is investigated in adult (3-month) and aged (2-year) murine BAT tissue via serial block face-scanning electron microscopy (SBF-SEM) and 3D reconstruction software for manual segmentation, analysis, and quantification. Upon investigation, an increase is found in mitochondrial volume, surface area, and complexity and decreased sphericity in aged BAT, alongside significant decreases in cristae volume, area, perimeter, and score. Overall, these data define the nature of the mitochondrial structure in murine BAT across aging.


INTRODUCTION
Mitochondria are complex cellular organelles that serve various physiological roles, including maintenance of Ca 2+ homeostasis, initiation of apoptosis, and cellular energy production [1][2][3] . With such vital cellular roles, mitochondria must be dynamic to meet the fluctuating energy demands of the cell [1,4,8,9] . The ultrastructure and morphology that mitochondria take is tightly associated with their functional capacity [6] . Therefore, it is no surprise that mitochondrial morphology varies considerably between tissue types and the metabolic health of that tissue [4,5] . Specializations within the inner mitochondrial membrane (IMM), known as cristae, substantially increase the surface area for the oxidative phosphorylation machinery to reside, optimizing energetic capacity [6] . Therefore, maintaining cristae's structural integrity and spatial arrangement is essential for bioenergetic homeostasis and proper cellular energy production [7] .
Normal mitochondrial dynamics consist of well-orchestrated, balanced cycles of fusion and fission that allow for content mixing and quality control to maintain a healthy network [9][10][11] . Previously, mitochondrial dynamics were thought to only occur with changes in the cellular environment which altered the overall mitochondrial shape. However, both fusion and fission require coordination of not only the outer mitochondrial membrane (OMM), but also coordination and reorganization of the cristae [ref]. Although key regulators for cristae morphology have been identified, the dynamics of cristae reorganization in physiology and disease remain understudied [3x ref].
Mitochondrial oxidative stress and dysfunction are often associated with the pathophysiology of diseases and the aging process, which can lead to the accumulation of mtDNA mutations as well as reactive oxygen species generation [15][16][17] . The ultrastructural changes that occur with the aging process are only now possible to explore with the development of 3D reconstruction techniques. For example, Faitg et al. (2021) showed in murine brain that hippocampal somatic, dendritic, and axonal mitochondria had differential baseline phenotypes and responses to aging [18] . In murine heart, 3D reconstruction revealed aged mitochondria arranged in a less ordered way than their younger counterparts [19] . These findings establish the applicability of 3D reconstruction to study mitochondrial phenotypes and cellular organization. Mitochondrial number, form, and function are highly tissue-dependent [5,12,20] . Brown adipose tissue (BAT) is rich in mitochondria, which are integral in thermogenesis [21] . Brown adipocytes produce heat primarily through the uncoupling of the cristae's proton gradient, which is facilitated through uncoupling protein 1 (UCP1) [24,26] . In both rodent models and humans, a decrease in BAT thermogenic function has been observed with advanced aging and has been associated with the development of metabolic disorders, including obesity and diabetes. Studies suggest that mitochondrial dysfunction through decreased expression of UCP1 The functional age-related decline in BAT has been associated with mitochondrial dysfunction in several studies. In oxidative-stress-induced models of aging, BAT mitochondria are observed to have decreased expression of UCP1, increased autophagy, and decreased mitochondria size [25] . Yet, to our knowledge, no studies have elucidated changes in the 3D morphology of mitochondria and cristae, as well as their 3D spatial distribution in BAT.
In this study, serial block face-scanning electron microscopy (SBF-SEM) and 3D reconstruction software were utilized to examine the 3D architecture of both mitochondria and cristae in BAT from adult and aged mice. Validated methods of quantification were used to develop a quantitative 3D approach to map mitochondrial networks and cristae. Here, we show that mitochondrial 3D volume, perimeter, 3D area, and complexity index all increase with aging in murine BAT. However, mitochondrial sphericity decreased as age increased. Conversely, mitochondrial cristate showed decreases in 3D volume, perimeter, 3D area, and complexity index while sphericity increased with aging in murine BAT.

Mice Care Procedure
Male C57BL/6J mice were housed at 22 C with a 12-h light, 12-h dark cycle accompanied by free access to water and standard chow following birth. Mice were cared for as in Lam et al with all protocols approved by the University of Iowa Animal Care and Use Committee (IACUC).

SBF-SEM Sample Preparation
Interscapular BAT was excised from 3-month and 2-year aged mice and fixed in 2% glutaraldehyde in 0.1M cacodylate buffer and processed using a heavy metal protocol. BAT samples were then immersed in 3% potassium ferrocyanide, followed by 2% osmium tetroxide for 1 hour each at 4 C following deionized H 2 O (DI H 2 O) washes. Samples were washed again in diH 2 O and immersed in filtered 0.1% thiocarbohydrazide for 20 minutes, followed by diH 2 O washing and subsequent immersion in 2% osmium tetroxide for 30 minutes. Samples were incubated overnight in 1% uranyl acetate at 4 C. The following day, samples were incubated in 0.6% lead aspartate for 30 minutes at 60 C prior to an ethanol graded series dehydration. BAT tissue samples were then infiltrated with epoxy Taab 812 hard resin prior to immersion in fresh resin and polymerization at 60 C for 36-48 hours. The resultant resin blocks were sectioned for transmission electron microscopy (TEM) to identify regions of interest. Samples were then trimmed, glued to an SEM stub, and then placed into a FEI/Thermo Scientific Volumescope 2 SEM. Between 300-400 thin serial sections of 0.09 μ m per sample block were obtained and collected. The resultant micrograph blocks were then aligned and manually segmented and reconstructed in 3D using Thermo Scientific Amira Software (Waltham, MA, USA) [27,29] .

Mitochondrial and Cristae Ultrastructure Calculations and Measurements
Following manual segmentation of mitochondria and cristae in the regions of interest (ROIs), label analyses were performed on each segmented structure using Amira [27] . The SBF-SEM data was acquired from at least three independent experiments to perform blinded-3D structural reconstruction from murine BAT. Manual segmentation of sequential orthoslices was then performed to obtain 300-400 slices of which, 50-100 serial sections were chosen for each 3D reconstruction. Serial sections had approximately equal z-direction intervals and werestacked, aligned, and visualized using Amira software to make videos and quantify volumetric structures. The algorithms for measurements were entered manually (all measurements in the main text) for those not already in the system, and a total of 400 mitochondria from two mice were collected for each quantification.

Structure Quantifications and Statistic Analyses
All data obtained from label analyses and manual measurements were statistically analyzed by Student's t-test, or the non-parametric equivalent where applicable, using GraphPad Prism (San Diego, California, USA). All data considered are biological replicates and dots represent individual data points unless otherwise noted. In some cases, for presentation, certain outliers may not be displayed, but all outliers are considered for statistical analysis. Graphs are shown as means with black bars representing the standard error of mean. Tukey post hoc tests were used for multiple comparisons, and a minimum threshold of p < 0.05 indicated a significant difference. Higher degrees of statistical significance (i.e., **, ***, ****) were defined as p < 0.01, p < 0.001, and p < 0.0001, respectively.

Mitochondrial and Matrix Volumes Significantly Increase with Aging In BAT
BAT biopsies were collected from adult (3-months-old) and aged (2-years-old) mice and imaged using SBF-SEM. With resolutions of 10 µm for the x-and y-planes and 50 µm for the zplane, SBF-SEM allows for 3D reconstruction of organelles, providing a high spatial resolution that is unattainable using 2D techniques. To quantify mitochondrial changes across aging, we examined approximately 400 mitochondria from two separate regions of interest per age group were selected from the male mice (n=2) ( Figure 1A).. Image stacks of ~50 50-µm ortho slices ( Figure 1B) were manually traced at transverse intervals ( Figure 1C). This enabled the the generation of of 3D reconstructions of each mitochondrion, as depicted in the flowchart of Figure 1 ( Figure 1D).
BAT mitochondria are typically abundant, large, and round in young, healthy tissue [22] . Consistent with this, we observed an abundance of mitochondrial that were spherical in structure in the adult (aged 3-months) is believed to be equivalent to approximately 20-years in the human lifespan. In comparison, the 2-year aged samples are believed to be equivalent to approximately 70 years of age in humans. [30]. . When comparing the 3-month to 2-year ages, significant increases in perimeter, 3D area (or surface area), and volume were observed ( Figure 2). When comparing the mitochondrial quantifications from each mouse, they exhibited minimal intergroup heterogeneity but consistent intra-individual variability with increased heterogeneity in aged samples.

BAT Aging Significantly Increases Mitochondrial Complexity
Brown adipocytes are an integral part of BAT and are generally classified as either being high-thermogenic or low-thermogenic. High thermogenic brown adipocytes are characterized by both smaller lipid size, round-shaped mitochondria, and a high basal respiration rate [32] . Conversely, low thermogenic brown adipocytes are characterized by larger lipids, oval-shaped mitochondria, and a low basal respiration rate. Interestingly, a reduced thermogenic capacity is associated with aging and obesity [32] . Thus, characterizing the shape of mitochondria may have important implications for their overall functional state.
Viewing the overall complexity of mitochondria can provide relative insights into the complexity of the mitochondrial networks they form as seen in skeletal and cardiac tissue [12] . The sphericity of mitochondria, or the closeness something is in shape to a perfect sphere, shows a significant decrease across aging ( Figure 3C). Of note, high-thermogenic BAT mitochondria are round whereas low-thermogenic mitochondria are considered oval in shape. The mitochondrial complexity index (MCI) is an index that correlates with the complexity of the mitochondrial shape, such as surface area and branching relative to volume (Surface Area 3 /16 π mitochondrial complexity is changed across mitochondrial volume, we performed a technique known as mito-otyping, which organizes mitochondria based on their volume for each aging point ( Figure 3E). Importantly, using this technique allows for the simultaneous visualization of changes in the complexity and volume across our aging model at each volume point analyzed.

Aged BAT Mitochondria Show Altered Cristae Volume, Area, and Perimeter
Mitochondrial morphology is tightly associated with mitochondrial function, making it an important component that is often unexplored. Alterations in cristae morphology and matrix volume can provide equally important quantitative changes in BAT aging. Although there are morphological restraints, increased cristae volume generally correlates with an increased ATP potential [34] . Notably, inflammation of BAT has been demonstrated to cause loss of cristae structural integrity, possibly related to impaired UCP1 activation also observed [35] . Therefore, we sought to quantify cristae morphological and size changes across aging. Using the workflow of SBF-SEM-based 3D reconstruction for BAT cristae, bwas obtained from mice ( Figure 4A), and 10 µm by 10 µm orthoslices were overlaid for 3D reconstruction ( Figure 4B). Representative orthoslices from the SBF-SEM sectioning were used to generate a detailed 3D reconstruction of cristae ( Figure 4C-D).
Cristae and matrix morphology changes across aging indicate that cristae in the 2-year BAT are more heterogenous in size, but are generally smaller and less complex than those in the 3-month samples. Interestingly, while the mitochondrial parameters measured increased overall, we observe the opposite here, where cristae volume, surface area, and perimeter exhibit a significant decrease. Despite increased heterogeneity and larger outliers, this data indicates cristae density is reduced across aging in BAT.

Aged BAT Mitochondria Show Altered Cristae Complexity
Perturbed cristae morphology has been associated with enlarged and dysfunctional mitochondria [6,36] . Observations of aberrant cristae in senescent cells have also been noted as being partially lost, totally lost, or in circular formation [36] . To understand the morphological changes that occur with cristae across aging, cristae complexity was examined in the 3-month and 2-year aged samples ( Figure 6). Notably, these findings resemble those observed for volume, with reduced cristae complexity observed in aged murine BAT. Mito-otyping was utilized to compare cristae morphology in samples from 3-month and 2-year aged groups across volumes ( Figure 6E).The 2-year aged sample showed the presence of mitochondria with partially lost cristae, along with circular cristae. Notably, we also observed "island" shaped cristae which were smaller and fragmented. The findings confirm that cristae in 2-year samples are not only more heterogeneous in size but also generally smaller and less complex than those in 3-month samples.

DISCUSSION
In this study, we examined the mitochondrial ultrastructure changes that occur in murine BAT with aging using SBF-SEM and 3D reconstruction. With the utilization of these tools, increases in mitochondrial 3D volume, perimeter, 3D area, and complexity index as well as decreases in sphericity were quantified between adult and aged murine BAT. Cristae from adult and aged murine BAT showed decreases in 3D volume, perimeter, 3D area, and complexity index while increased sphericity was observed with aging. (Figures 2, 3, 5). Interestingly, increases in mitochondrial area and perimeter, but not cristae area and perimeter, indicate an increase in mitochondrial matrix volume with aging in murine BAT. (Figure 2D-E). ITheresults presented here also indicate that both mitochondrial and matrix volumes increase with aging, but not cristae volume, in murine BAT. Previous studies have noted the presence of mitochondrial swelling in cases of mitochondrial dysfunction and membrane potential loss [36,39,40] . Under extreme conditions, osmotic swelling of the mitochondrial matrix follows the opening of the inner membrane permeability transition pore (PTP) and can also irreversibly engage in apoptosis [42][43][44] . Notably, the intermembrane proteins released include cytochrome C and apoptosisinducing factor (AIF) which solidify the apoptotic pathway [42][43][44] .
Another interesting observation in our study is the presence of circular cristae in the aged BAT samples that were not present in the young BAT samples. Circular cristae have been described in senescent cells by others and differ from onion cristae [47] given the lack of multiple rings of cristae present [36] . The presence of circular cristae have also been noted in senescent cells [36] , but a lack of 3D studies of cristae morphology makes it difficult to make comparisons in cristae phenotypes. Beyond this, it is evident that the aging process decreases overall mitochondrial cristae density, although the underlying molecular causes have yet to be elucidated ( Figure 5C).
Although the findings here show clear mitochondrial and cristae alterations with aging, it should be noted that murine and human BAT have several key differences. Future studies are needed to understand morphological changes of mitochondria and cristae in human BAT in comparison to murine BAT. Studies have shown that while cristae architecture remains pertinent for energetic capacity across models, aging shows differed functional and structural changes in mice and Drosophila [65] . An important limitation of our study that should be noted is that other studies have shown distinct populations of adipocytes, with higher-thermogenic activity populations, which can dynamically form in exposure to cold [67] . Our study did not account for this heterogeneity in the population, which may account for some of the variation observed in mitochondrial structure. Future studies may consider 3D tissue profiling to see if mitochondrial and cristae structure varies across populations of adipocytes or alterations in gene expression [67] .
In conclusion, to our knowledge, we are the first to elucidate the alterations in 3D mitochondrial and cristae structures across the murine aging process in BAT. Of relevance, this aids to establish standards for the phenotypes presented. Importantly, the phenotypes found here may be related to pre-stress states, alterations in membrane potential, changes in fusion or cristae proteins, sex-dependent differences, or changes in metabolic pathways. Although this study did not delve into the molecular changes that influence the aging phenotypes observed, by establishing these 3D structures, future studies investigating the influence of aging in BAT on all previously mentioned factors can better understand how these factors may relate to specific phenotypes. In the future, this may aid in establishing standards regarding mitochondrial and cristae structures, to aid future research aimed at developing interventions to mitigate dysfunction during aging in BAT.

Declaration of interests
The authors have no Conflicts of Interest to declare.            Complexity Index ✱✱